From c802a03e07b63793c67f629556675bb8298b3d43 Mon Sep 17 00:00:00 2001
From: paboyle <paboyle@ph.ed.ac.uk>
Date: Wed, 4 Mar 2015 04:54:50 +0000
Subject: [PATCH 001/429] Delete INSTALL

---
 INSTALL | 1 -
 1 file changed, 1 deletion(-)
 delete mode 120000 INSTALL

diff --git a/INSTALL b/INSTALL
deleted file mode 120000
index 80a61507..00000000
--- a/INSTALL
+++ /dev/null
@@ -1 +0,0 @@
-/opt/local/share/automake-1.15/INSTALL
\ No newline at end of file

From b0c9282fe9d066a319c2d9771a60751d7453d379 Mon Sep 17 00:00:00 2001
From: paboyle <paboyle@ph.ed.ac.uk>
Date: Wed, 4 Mar 2015 04:55:04 +0000
Subject: [PATCH 002/429] Delete COPYING

---
 COPYING | 1 -
 1 file changed, 1 deletion(-)
 delete mode 120000 COPYING

diff --git a/COPYING b/COPYING
deleted file mode 120000
index af6f8816..00000000
--- a/COPYING
+++ /dev/null
@@ -1 +0,0 @@
-/opt/local/share/automake-1.15/COPYING
\ No newline at end of file

From 1a1474b32339d9873fb46772032256ffe3330b72 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 4 Mar 2015 05:31:44 +0000
Subject: [PATCH 003/429] Better organisation

---
 Grid.h                   | 1122 +-------------------------------------
 Grid_Lattice.h           |   98 +---
 Grid_QCD.h               |   94 ++++
 Grid_aligned_allocator.h |   56 ++
 Grid_config.h            |  101 ++++
 Grid_math_types.h        |  819 ++++++++++++++++++++++++++++
 Grid_signal.cc           |   16 +-
 Grid_simd.h              |   68 ++-
 8 files changed, 1161 insertions(+), 1213 deletions(-)
 create mode 100644 Grid_QCD.h
 create mode 100644 Grid_aligned_allocator.h
 create mode 100644 Grid_config.h
 create mode 100644 Grid_math_types.h

diff --git a/Grid.h b/Grid.h
index b3669ace..042835ca 100644
--- a/Grid.h
+++ b/Grid.h
@@ -40,1123 +40,21 @@
 #endif
 
 
-////////////////////////////////////////////////////////////
-// SIMD Alignment controls
-////////////////////////////////////////////////////////////
-#ifdef HAVE_VAR_ATTRIBUTE_ALIGNED
-#define ALIGN_DIRECTIVE(A) __attribute__ ((aligned(A)))
-#else
-#define ALIGN_DIRECTIVE(A) __declspec(align(A))
-#endif
-
-#ifdef SSE2
-#include <pmmintrin.h>
-#define SIMDalign ALIGN_DIRECTIVE(16)
-#endif
-
-#if defined(AVX1) || defined (AVX2)
-#include <immintrin.h>
-#define SIMDalign ALIGN_DIRECTIVE(32)
-#endif
-
-#ifdef AVX512
-#include <immintrin.h>
-#define SIMDalign ALIGN_DIRECTIVE(64)
-#endif
-
+#include <Grid_simd.h>
+#include <Grid_math_types.h>
+#include <Grid_cartesian.h>
+#include <Grid_aligned_allocator.h>
+#include <Grid_aligned_allocator.h>
+#include <Grid_Lattice.h>
+#include <Grid_QCD.h>
 
 namespace dpo {
 
   void Grid_init(void);
-
-inline double usecond(void)
-{
-    struct timeval tv;
-    gettimeofday(&tv,NULL);
-    return 1.0*tv.tv_usec + 1.0e6*tv.tv_sec;
-}
-    
-    typedef  float  RealF;
-    typedef  double RealD;
-    typedef  RealF  Real;
-    
-    typedef std::complex<RealF> ComplexF;
-    typedef std::complex<RealD> ComplexD;
-    typedef std::complex<Real>  Complex;
-    
-
-    class Zero{};
-    static Zero zero;
-    template<class itype> inline void ZeroIt(itype &arg){ arg=zero;};
-    template<>            inline void ZeroIt(ComplexF &arg){ arg=0; };
-    template<>            inline void ZeroIt(ComplexD &arg){ arg=0; };
-    template<>            inline void ZeroIt(RealF &arg){ arg=0; };
-    template<>            inline void ZeroIt(RealD &arg){ arg=0; };
-
-    // TODO
-    //
-    // Base class to share common code between vRealF, VComplexF etc...
-    //
-    // lattice Broad cast assignment
-    //
-    // where() support
-    // implement with masks, and/or? Type of the mask & boolean support?
-    //
-    // Unary functions
-    // cos,sin, tan, acos, asin, cosh, acosh, tanh, sinh, // Scalar<vReal> only arg
-    // exp, log, sqrt, fabs
-    //
-    // transposeColor, transposeSpin,
-    // adjColor, adjSpin,
-    // traceColor, traceSpin.
-    // peekColor, peekSpin + pokeColor PokeSpin
-    //
-    // copyMask.
-    //
-    // localMaxAbs
-    //
-    // norm2,
-    // sumMulti equivalent.
-    // Fourier transform equivalent.
-    //
-    
-    ////////////////////////////////////////////////////////////////////////////////
-    //Provide support functions for basic real and complex data types required by dpo
-    //Single and double precision versions. Should be able to template this once only.
-    ////////////////////////////////////////////////////////////////////////////////
-    
-    inline void mac (ComplexD * __restrict__ y,const ComplexD * __restrict__ a,const ComplexD *__restrict__ x){ *y = (*a) * (*x)+(*y); };
-    inline void mult(ComplexD * __restrict__ y,const ComplexD * __restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) * (*r);}
-    inline void sub (ComplexD * __restrict__ y,const ComplexD * __restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) - (*r);}
-    inline void add (ComplexD * __restrict__ y,const ComplexD * __restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) + (*r);}
-    inline ComplexD adj(const ComplexD& r){ return(conj(r)); }
-    // conj already supported for complex
-    
-    inline void mac (ComplexF * __restrict__ y,const ComplexF * __restrict__ a,const ComplexF *__restrict__ x){ *y = (*a) * (*x)+(*y); }
-    inline void mult(ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) * (*r); }
-    inline void sub (ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) - (*r); }
-    inline void add (ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) + (*r); }
-    inline Complex  adj(const Complex& r ){ return(conj(r)); }
-    //conj already supported for complex
-    
-    inline void mac (RealD * __restrict__ y,const RealD * __restrict__ a,const RealD *__restrict__ x){  *y = (*a) * (*x)+(*y);}
-    inline void mult(RealD * __restrict__ y,const RealD * __restrict__ l,const RealD *__restrict__ r){ *y = (*l) * (*r);}
-    inline void sub (RealD * __restrict__ y,const RealD * __restrict__ l,const RealD *__restrict__ r){ *y = (*l) - (*r);}
-    inline void add (RealD * __restrict__ y,const RealD * __restrict__ l,const RealD *__restrict__ r){ *y = (*l) + (*r);}
-    inline RealD adj(const RealD & r){ return r; }  // No-op for real
-    inline RealD conj(const RealD & r){ return r; }
-    
-    inline void mac (RealF * __restrict__ y,const RealF * __restrict__ a,const RealF *__restrict__ x){  *y = (*a) * (*x)+(*y); }
-    inline void mult(RealF * __restrict__ y,const RealF * __restrict__ l,const RealF *__restrict__ r){ *y = (*l) * (*r); }
-    inline void sub (RealF * __restrict__ y,const RealF * __restrict__ l,const RealF *__restrict__ r){ *y = (*l) - (*r); }
-    inline void add (RealF * __restrict__ y,const RealF * __restrict__ l,const RealF *__restrict__ r){ *y = (*l) + (*r); }
-    inline RealF adj(const RealF  & r){ return r; }
-    inline RealF conj(const RealF  & r){ return r; }
-    
-    ////////////////////////////////////////////////////////////////////////
-    //  Vector types are arch dependent/////////////////////////////////////
-    ////////////////////////////////////////////////////////////////////////
-#if defined (SSE2)
-    typedef __m128 fvec;
-    typedef __m128d dvec;
-    typedef __m128 cvec;
-    typedef __m128d zvec;
-#endif
-#if defined (AVX1) || defined (AVX2)
-    typedef __m256 fvec;
-    typedef __m256d dvec;
-    typedef __m256 cvec;
-    typedef __m256d zvec;
-#endif
-#if defined (AVX512)
-    typedef __m512  fvec;
-    typedef __m512d dvec;
-    typedef __m512  cvec;
-    typedef __m512d zvec;
-#endif
-#if defined (QPX)
-    typedef float  fvec __attribute__ ((vector_size (16))); // QPX has same SIMD width irrespective of precision
-    typedef float  cvec __attribute__ ((vector_size (16)));
-    
-    typedef vector4double dvec;
-    typedef vector4double zvec;
-#endif
-#if defined (AVX1) || defined (AVX2) || defined (AVX512)
-    inline void v_prefetch0(int size, const char *ptr){
-          for(int i=0;i<size;i+=64){ //  Define L1 linesize above// What about SSE?
-            _mm_prefetch(ptr+i+4096,_MM_HINT_T1);
-            _mm_prefetch(ptr+i+512,_MM_HINT_T0);
-          }
-    }
-#else 
-    inline void v_prefetch0(int size, const char *ptr){};
-#endif
-
-///////////////////////////////////////////////////
-// Scalar, Vector, Matrix objects.
-// These can be composed to form tensor products of internal indices.
-///////////////////////////////////////////////////
-    
-template<class vtype> class iScalar
-{
-public:
-  SIMDalign vtype _internal;
-    iScalar(){};
-    iScalar(Zero &z){ *this = zero; };
-    iScalar<vtype> & operator= (const Zero &hero){
-        zeroit(*this);
-        return *this;
-    }
-    friend void zeroit(iScalar<vtype> &that){
-        zeroit(that._internal);
-    }
-    // Unary negation
-    friend inline iScalar<vtype> operator -(const iScalar<vtype> &r) {
-        iScalar<vtype> ret;
-        ret._internal= -r._internal;
-        return ret;
-    }
-    // *=,+=,-= operators
-    inline iScalar<vtype> &operator *=(const iScalar<vtype> &r) {
-        *this = (*this)*r;
-        return *this;
-    }
-    inline iScalar<vtype> &operator -=(const iScalar<vtype> &r) {
-        *this = (*this)-r;
-        return *this;
-    }
-    inline iScalar<vtype> &operator +=(const iScalar<vtype> &r) {
-        *this = (*this)+r;
-        return *this;
-    }
-    
+  double usecond(void);
+  void Grid_sa_signal_handler(int sig,siginfo_t *si,void * ptr);
+  void Grid_debug_handler_init(void);
 
 };
-    
-template<class vtype,int N> class iVector
-{
-public:
-  SIMDalign vtype _internal[N];
-    iVector(Zero &z){ *this = zero; };
-    iVector() {};
-    iVector<vtype,N> & operator= (Zero &hero){
-        zeroit(*this);
-        return *this;
-    }
-    friend void zeroit(iVector<vtype,N> &that){
-        for(int i=0;i<N;i++){
-            zeroit(that._internal[i]);
-        }
-    }
-    // Unary negation
-    friend inline iVector<vtype,N> operator -(const iVector<vtype,N> &r) {
-        iVector<vtype,N> ret;
-        for(int i=0;i<N;i++) ret._internal[i]= -r._internal[i];
-        return ret;
-    }
-    // *=,+=,-= operators
-    inline iVector<vtype,N> &operator *=(const iScalar<vtype> &r) {
-        *this = (*this)*r;
-        return *this;
-    }
-    inline iVector<vtype,N> &operator -=(const iVector<vtype,N> &r) {
-        *this = (*this)-r;
-        return *this;
-    }
-    inline iVector<vtype,N> &operator +=(const iVector<vtype,N> &r) {
-        *this = (*this)+r;
-        return *this;
-    }
-
-};
-    
-    
-template<class vtype,int N> class iMatrix
-{
-public:
-  SIMDalign    vtype _internal[N][N];
-    iMatrix(Zero &z){ *this = zero; };
-    iMatrix() {};
-    iMatrix<vtype,N> & operator= (Zero &hero){
-        zeroit(*this);
-        return *this;
-    }
-    friend void zeroit(iMatrix<vtype,N> &that){
-        for(int i=0;i<N;i++){
-        for(int j=0;j<N;j++){
-                zeroit(that._internal[i][j]);
-        }}
-    }
-    // Unary negation
-    friend inline iMatrix<vtype,N> operator -(const iMatrix<vtype,N> &r) {
-        iMatrix<vtype,N> ret;
-        for(int i=0;i<N;i++){
-        for(int j=0;j<N;j++){
-            ret._internal[i][j]= -r._internal[i][j];
-        }}
-        return ret;
-    }
-    // *=,+=,-= operators
-    template<class T>
-    inline iMatrix<vtype,N> &operator *=(const T &r) {
-        *this = (*this)*r;
-        return *this;
-    }
-    template<class T>
-    inline iMatrix<vtype,N> &operator -=(const T &r) {
-        *this = (*this)-r;
-        return *this;
-    }
-    template<class T>
-    inline iMatrix<vtype,N> &operator +=(const T &r) {
-        *this = (*this)+r;
-        return *this;
-    }
-
-};
-/*
-    inline vComplexD localInnerProduct(const vComplexD & l, const vComplexD & r) { return conj(l)*r; }
-    inline vComplexF localInnerProduct(const vComplexF & l, const vComplexF & r) { return conj(l)*r; }
-    inline vRealD localInnerProduct(const vRealD & l, const vRealD & r) { return conj(l)*r; }
-    inline vRealF localInnerProduct(const vRealF & l, const vRealF & r) { return conj(l)*r; }
-*/
-    inline ComplexD localInnerProduct(const ComplexD & l, const ComplexD & r) { return conj(l)*r; }
-    inline ComplexF localInnerProduct(const ComplexF & l, const ComplexF & r) { return conj(l)*r; }
-    inline RealD localInnerProduct(const RealD & l, const RealD & r) { return conj(l)*r; }
-    inline RealF localInnerProduct(const RealF & l, const RealF & r) { return conj(l)*r; }
-
-    
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-    /////////////////////////////////////////// ADD         ///////////////////////////////////////////
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-    
-
-// ADD is simple for now; cannot mix types and straightforward template
-// Scalar +/- Scalar
-// Vector +/- Vector
-// Matrix +/- Matrix
-template<class vtype,class ltype,class rtype> inline void add(iScalar<vtype> * __restrict__ ret,
-                                                              const iScalar<ltype> * __restrict__ lhs,
-                                                              const iScalar<rtype> * __restrict__ rhs)
-{
-    add(&ret->_internal,&lhs->_internal,&rhs->_internal);
-}
-template<class vtype,class ltype,class rtype,int N> inline void add(iVector<vtype,N> * __restrict__ ret,
-                                                                    const iVector<ltype,N> * __restrict__ lhs,
-                                                                    const iVector<rtype,N> * __restrict__ rhs)
-{
-    for(int c=0;c<N;c++){
-        ret->_internal[c]=lhs->_internal[c]+rhs->_internal[c];
-    }
-    return;
-}
-template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
-                                                                      const iMatrix<ltype,N> * __restrict__ lhs,
-                                                                      const iMatrix<rtype,N> * __restrict__ rhs)
-{
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        add(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal[c1][c2]);
-    }}
-    return;
-}
-template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
-                                                                      const iScalar<ltype>   * __restrict__ lhs,
-                                                                      const iMatrix<rtype,N> * __restrict__ rhs)
-{
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        add(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
-    }}
-    return;
-}
-template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
-                                                                      const iMatrix<ltype,N> * __restrict__ lhs,
-                                                                      const iScalar<rtype>   * __restrict__ rhs)
-{
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        if ( c1==c2)
-            add(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
-        else
-            ret->_internal[c1][c2]=lhs->_internal[c1][c2];
-    }}
-    return;
-}
-// Need to figure multi-precision.
-template<class Mytype>  Mytype timesI(Mytype &r)
-{
-    iScalar<Complex> i;
-    i._internal = Complex(0,1);
-    return r*i;
-}
-
-                // + operator for scalar, vector, matrix
-template<class ltype,class rtype>
-//inline auto operator + (iScalar<ltype>& lhs,iScalar<rtype>&& rhs) -> iScalar<decltype(lhs._internal + rhs._internal)>
-inline auto operator + (const iScalar<ltype>& lhs,const iScalar<rtype>& rhs) -> iScalar<decltype(lhs._internal + rhs._internal)>
-{
-    typedef iScalar<decltype(lhs._internal+rhs._internal)> ret_t;
-    ret_t ret;
-    add(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator + (const iVector<ltype,N>& lhs,const iVector<rtype,N>& rhs) ->iVector<decltype(lhs._internal[0]+rhs._internal[0]),N>
-{
-    typedef iVector<decltype(lhs._internal[0]+rhs._internal[0]),N> ret_t;
-    ret_t ret;
-    add(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator + (const iMatrix<ltype,N>& lhs,const iMatrix<rtype,N>& rhs) ->iMatrix<decltype(lhs._internal[0][0]+rhs._internal[0][0]),N>
-{
-    typedef iMatrix<decltype(lhs._internal[0][0]+rhs._internal[0][0]),N> ret_t;
-    ret_t ret;
-    add(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator + (const iScalar<ltype>& lhs,const iMatrix<rtype,N>& rhs)->iMatrix<decltype(lhs._internal+rhs._internal[0][0]),N>
-{
-    typedef iMatrix<decltype(lhs._internal+rhs._internal[0][0]),N> ret_t;
-    ret_t ret;
-    add(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator + (const iMatrix<ltype,N>& lhs,const iScalar<rtype>& rhs)->iMatrix<decltype(lhs._internal[0][0]+rhs._internal),N>
-{
-    typedef iMatrix<decltype(lhs._internal[0][0]+rhs._internal),N> ret_t;
-    ret_t ret;
-    add(&ret,&lhs,&rhs);
-    return ret;
-}
-
-
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-    /////////////////////////////////////////// SUB         ///////////////////////////////////////////
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-    
-
-// SUB is simple for now; cannot mix types and straightforward template
-// Scalar +/- Scalar
-// Vector +/- Vector
-// Matrix +/- Matrix
-// Matrix /- scalar
-template<class vtype,class ltype,class rtype> inline void sub(iScalar<vtype> * __restrict__ ret,
-                                                              const iScalar<ltype> * __restrict__ lhs,
-                                                              const iScalar<rtype> * __restrict__ rhs)
-{
-    sub(&ret->_internal,&lhs->_internal,&rhs->_internal);
-}
-
-template<class vtype,class ltype,class rtype,int N> inline void sub(iVector<vtype,N> * __restrict__ ret,
-                                                                    const iVector<ltype,N> * __restrict__ lhs,
-                                                                    const iVector<rtype,N> * __restrict__ rhs)
-{
-    for(int c=0;c<N;c++){
-        ret->_internal[c]=lhs->_internal[c]-rhs->_internal[c];
-    }
-    return;
-}
-template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
-                                                                     const iMatrix<ltype,N> * __restrict__ lhs,
-                                                                     const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        sub(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal[c1][c2]);
-    }}
-    return;
-}
-template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
-                                                                     const iScalar<ltype> * __restrict__ lhs,
-                                                                     const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        if ( c1!=c2) {
-            sub(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
-        } else {
-            // Fails -- need unary minus. Catalogue other unops?
-            ret->_internal[c1][c2]=zero;
-            ret->_internal[c1][c2]=ret->_internal[c1][c2]-rhs->_internal[c1][c2];
-
-        }
-    }}
-    return;
-}
-template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
-                                                                     const iMatrix<ltype,N> * __restrict__ lhs,
-                                                                     const iScalar<rtype> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        if ( c1!=c2)
-            sub(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
-        else
-            ret->_internal[c1][c2]=lhs->_internal[c1][c2];
-    }}
-    return;
-}
-
-template<class v> void vprefetch(const iScalar<v> &vv)
-{
-  vprefetch(vv._internal);
-}
-template<class v,int N> void vprefetch(const iVector<v,N> &vv)
-{
-  for(int i=0;i<N;i++){
-    vprefetch(vv._internal[i]);
-  }
-}
-template<class v,int N> void vprefetch(const iMatrix<v,N> &vv)
-{
-  for(int i=0;i<N;i++){
-  for(int j=0;j<N;j++){
-    vprefetch(vv._internal[i][j]);
-  }}
-}
-
-    // - operator for scalar, vector, matrix
-template<class ltype,class rtype> inline auto
-operator - (const iScalar<ltype>& lhs, const iScalar<rtype>& rhs) -> iScalar<decltype(lhs._internal - rhs._internal)>
-{
-    typedef iScalar<decltype(lhs._internal-rhs._internal)> ret_t;
-    ret_t ret;
-    sub(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator - (const iVector<ltype,N>& lhs,const iVector<rtype,N>& rhs) ->iVector<decltype(lhs._internal[0]-rhs._internal[0]),N>
-{
-    typedef iVector<decltype(lhs._internal[0]-rhs._internal[0]),N> ret_t;
-    ret_t ret;
-    sub(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator - (const iMatrix<ltype,N>& lhs,const iMatrix<rtype,N>& rhs) ->iMatrix<decltype(lhs._internal[0][0]-rhs._internal[0][0]),N>
-{
-    typedef iMatrix<decltype(lhs._internal[0][0]-rhs._internal[0][0]),N> ret_t;
-    ret_t ret;
-    sub(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator - (const iScalar<ltype>& lhs,const iMatrix<rtype,N>& rhs)->iMatrix<decltype(lhs._internal-rhs._internal[0][0]),N>
-{
-    typedef iMatrix<decltype(lhs._internal-rhs._internal[0][0]),N> ret_t;
-    ret_t ret;
-    sub(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator - (const iMatrix<ltype,N>& lhs,const iScalar<rtype>& rhs)->iMatrix<decltype(lhs._internal[0][0]-rhs._internal),N>
-{
-    typedef iMatrix<decltype(lhs._internal[0][0]-rhs._internal),N> ret_t;
-    ret_t ret;
-    sub(&ret,&lhs,&rhs);
-    return ret;
-}
-
-///////////////////////////////////////////////////////////////////////////////////////////////////
-/////////////////////////////////////////// MAC         ///////////////////////////////////////////
-///////////////////////////////////////////////////////////////////////////////////////////////////
-
-    ///////////////////////////
-    // Legal multiplication table
-    ///////////////////////////
-    // scal x scal = scal
-    // mat x  mat  = mat
-    // mat  x scal = mat
-    // scal x mat  = mat
-    // mat  x vec  = vec
-    // vec  x scal = vec
-    // scal x vec  = vec
-    ///////////////////////////
-template<class rtype,class vtype,class mtype>
-inline  void mac(iScalar<rtype> * __restrict__ ret,const iScalar<vtype> * __restrict__ lhs,const iScalar<mtype> * __restrict__ rhs)
-{
-    mac(&ret->_internal,&lhs->_internal,&rhs->_internal);
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-    for(int c3=0;c3<N;c3++){
-        mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c3],&rhs->_internal[c3][c2]);
-    }}}
-    return;
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs){
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
-    }}
-    return;
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iScalar<ltype> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        mac(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
-    }}
-    return;
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iVector<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iVector<rtype,N> * __restrict__ rhs)
-{
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        mac(&ret->_internal[c1],&lhs->_internal[c1][c2],&rhs->_internal[c2]);
-    }}
-    return;
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iVector<rrtype,N> * __restrict__ ret,const iScalar<ltype> * __restrict__ lhs,const iVector<rtype,N> * __restrict__ rhs)
-{
-    for(int c1=0;c1<N;c1++){
-        mac(&ret->_internal[c1],&lhs->_internal,&rhs->_internal[c1]);
-    }
-    return;
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iVector<rrtype,N> * __restrict__ ret,const iVector<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs)
-{
-    for(int c1=0;c1<N;c1++){
-        mac(&ret->_internal[c1],&lhs->_internal[c1],&rhs->_internal);
-    }
-    return;
-}
-
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-    /////////////////////////////////////////// MUL         ///////////////////////////////////////////
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-
-    
-template<class rtype,class vtype,class mtype>
-inline void mult(iScalar<rtype> * __restrict__ ret,const iScalar<mtype> * __restrict__ lhs,const iScalar<vtype> * __restrict__ rhs){
-    mult(&ret->_internal,&lhs->_internal,&rhs->_internal);
-}
-
-template<class rrtype,class ltype,class rtype,int N>
-inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        mult(&ret->_internal[c1][c2],&lhs->_internal[c1][0],&rhs->_internal[0][c2]);
-        for(int c3=1;c3<N;c3++){
-            mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c3],&rhs->_internal[c3][c2]);
-        }
-    }}
-    return;
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        mult(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
-    }}
-    return;
-}
-
-template<class rrtype,class ltype,class rtype, int N>
-inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iScalar<ltype>   * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        mult(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
-    }}
-    return;
-}
-// Matrix left multiplies vector
-template<class rtype,class vtype,class mtype,int N>
-inline void mult(iVector<rtype,N> * __restrict__ ret,const iMatrix<mtype,N> * __restrict__ lhs,const iVector<vtype,N> * __restrict__ rhs)
-{
-    for(int c1=0;c1<N;c1++){
-        mult(&ret->_internal[c1],&lhs->_internal[c1][0],&rhs->_internal[0]);
-        for(int c2=1;c2<N;c2++){
-            mac(&ret->_internal[c1],&lhs->_internal[c1][c2],&rhs->_internal[c2]);
-        }
-    }
-    return;
-}
-template<class rtype,class vtype,class mtype,int N>
-inline void mult(iVector<rtype,N> * __restrict__ ret,
-                 const iScalar<mtype>   * __restrict__ lhs,
-                 const iVector<vtype,N> * __restrict__ rhs){
-    for(int c1=0;c1<N;c1++){
-        mult(&ret->_internal[c1],&lhs->_internal,&rhs->_internal[c1]);
-    }
-}
-template<class rtype,class vtype,class mtype,int N>
-inline void mult(iVector<rtype,N> * __restrict__ ret,
-                 const iVector<vtype,N> * __restrict__ rhs,
-                 const iScalar<mtype> * __restrict__ lhs){
-    mult(ret,lhs,rhs);
-}
-    
-
-
-template<class rtype,class vtype,class mtype,int N> inline
-iVector<rtype,N> operator * (const iMatrix<mtype,N>& lhs,const iVector<vtype,N>& rhs)
-{
-    iVector<rtype,N> ret;
-    mult(&ret,&lhs,&rhs);
-    return ret;
-}
-
-template<class rtype,class vtype,class mtype,int N> inline
-iVector<rtype,N> operator * (const iScalar<mtype>& lhs,const iVector<vtype,N>& rhs)
-{
-    iVector<rtype,N> ret;
-    mult(&ret,&lhs,&rhs);
-    return ret;
-}
-
-template<class rtype,class vtype,class mtype,int N> inline
-iVector<rtype,N> operator * (const iVector<mtype,N>& lhs,const iScalar<vtype>& rhs)
-{
-    iVector<rtype,N> ret;
-    mult(&ret,&lhs,&rhs);
-    return ret;
-}
-    
-    //////////////////////////////////////////////////////////////////
-    // Glue operators to mult routines. Must resolve return type cleverly from typeof(internal)
-    // since nesting matrix<scalar> x matrix<matrix>-> matrix<matrix>
-    // while         matrix<scalar> x matrix<scalar>-> matrix<scalar>
-    // so return type depends on argument types in nasty way.
-    //////////////////////////////////////////////////////////////////
-    // scal x scal = scal
-    // mat x  mat  = mat
-    // mat  x scal = mat
-    // scal x mat  = mat
-    // mat  x vec  = vec
-    // vec  x scal = vec
-    // scal x vec  = vec
-    
-template<class l,class r>
-inline auto operator * (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(lhs._internal * rhs._internal)>
-{
-    typedef iScalar<decltype(lhs._internal*rhs._internal)> ret_t;
-    ret_t ret;
-    mult(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class l,class r,int N> inline
-auto operator * (const iMatrix<l,N>& lhs,const iMatrix<r,N>& rhs) -> iMatrix<decltype(lhs._internal[0][0]*rhs._internal[0][0]),N>
-{
-    typedef decltype(lhs._internal[0][0]*rhs._internal[0][0]) ret_t;
-    iMatrix<ret_t,N> ret;
-    mult(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class l,class r, int N> inline
-auto operator * (const iMatrix<r,N>& lhs,const iScalar<l>& rhs) -> iMatrix<decltype(lhs._internal[0][0]*rhs._internal),N>
-{
-    typedef decltype(lhs._internal[0][0]*rhs._internal) ret_t;
-        
-    iMatrix<ret_t,N> ret;
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        mult(&ret._internal[c1][c2],&lhs._internal[c1][c2],&rhs._internal);
-    }}
-    return ret;
-}
-template<class l,class r,int N> inline
-auto operator * (const iScalar<l>& lhs,const iMatrix<r,N>& rhs) -> iMatrix<decltype(lhs._internal*rhs._internal[0][0]),N>
-{
-    typedef decltype(lhs._internal*rhs._internal[0][0]) ret_t;
-    iMatrix<ret_t,N> ret;
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        mult(&ret._internal[c1][c2],&lhs._internal,&rhs._internal[c1][c2]);
-    }}
-    return ret;
-}
-template<class l,class r,int N> inline
-auto operator * (const iMatrix<l,N>& lhs,const iVector<r,N>& rhs) -> iVector<decltype(lhs._internal[0][0]*rhs._internal[0]),N>
-{
-    typedef decltype(lhs._internal[0][0]*rhs._internal[0]) ret_t;
-    iVector<ret_t,N> ret;
-    for(int c1=0;c1<N;c1++){
-        mult(&ret._internal[c1],&lhs._internal[c1][0],&rhs._internal[0]);
-        for(int c2=1;c2<N;c2++){
-            mac(&ret._internal[c1],&lhs._internal[c1][c2],&rhs._internal[c2]);
-        }
-    }
-    return ret;
-}
-template<class l,class r,int N> inline
-auto operator * (const iScalar<l>& lhs,const iVector<r,N>& rhs) -> iVector<decltype(lhs._internal*rhs._internal[0]),N>
-{
-    typedef decltype(lhs._internal*rhs._internal[0]) ret_t;
-    iVector<ret_t,N> ret;
-    for(int c1=0;c1<N;c1++){
-        mult(&ret._internal[c1],&lhs._internal,&rhs._internal[c1]);
-    }
-    return ret;
-}
-template<class l,class r,int N> inline
-auto operator * (const iVector<l,N>& lhs,const iScalar<r>& rhs) -> iVector<decltype(lhs._internal[0]*rhs._internal),N>
-{
-    typedef decltype(lhs._internal[0]*rhs._internal) ret_t;
-    iVector<ret_t,N> ret;
-    for(int c1=0;c1<N;c1++){
-        mult(&ret._internal[c1],&lhs._internal[c1],&rhs._internal);
-    }
-    return ret;
-}
-    ///////////////////////////////////////////////////////////////////////////////////////
-    // localInnerProduct Scalar x Scalar -> Scalar
-    // localInnerProduct Vector x Vector -> Scalar
-    // localInnerProduct Matrix x Matrix -> Scalar
-    ///////////////////////////////////////////////////////////////////////////////////////
-    template<class l,class r,int N> inline
-    auto localInnerProduct (const iVector<l,N>& lhs,const iVector<r,N>& rhs) -> iScalar<decltype(localInnerProduct(lhs._internal[0],rhs._internal[0]))>
-    {
-        typedef decltype(localInnerProduct(lhs._internal[0],rhs._internal[0])) ret_t;
-        iScalar<ret_t> ret=zero;
-        for(int c1=0;c1<N;c1++){
-            ret._internal += localInnerProduct(lhs._internal[c1],rhs._internal[c1]);
-        }
-        return ret;
-    }
-    template<class l,class r,int N> inline
-    auto localInnerProduct (const iMatrix<l,N>& lhs,const iMatrix<r,N>& rhs) -> iScalar<decltype(localInnerProduct(lhs._internal[0][0],rhs._internal[0][0]))>
-    {
-        typedef decltype(localInnerProduct(lhs._internal[0][0],rhs._internal[0][0])) ret_t;
-        iScalar<ret_t> ret=zero;
-        for(int c1=0;c1<N;c1++){
-        for(int c2=0;c2<N;c2++){
-            ret._internal += localInnerProduct(lhs._internal[c1][c2],rhs._internal[c1][c2]);
-        }}
-        return ret;
-    }
-    template<class l,class r> inline
-    auto localInnerProduct (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(localInnerProduct(lhs._internal,rhs._internal))>
-    {
-        typedef decltype(localInnerProduct(lhs._internal,rhs._internal)) ret_t;
-        iScalar<ret_t> ret;
-        ret._internal = localInnerProduct(lhs._internal,rhs._internal);
-        return ret;
-    }
-
-    ///////////////////////////////////////////////////////////////////////////////////////
-    // outerProduct Scalar x Scalar -> Scalar
-    //              Vector x Vector -> Matrix
-    ///////////////////////////////////////////////////////////////////////////////////////
-
-template<class l,class r,int N> inline
-auto outerProduct (const iVector<l,N>& lhs,const iVector<r,N>& rhs) -> iMatrix<decltype(outerProduct(lhs._internal[0],rhs._internal[0])),N>
-{
-    typedef decltype(outerProduct(lhs._internal[0],rhs._internal[0])) ret_t;
-    iMatrix<ret_t,N> ret;
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        ret._internal[c1][c2] = outerProduct(lhs._internal[c1],rhs._internal[c2]);
-    }}
-    return ret;
-}
-template<class l,class r> inline
-auto outerProduct (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(outerProduct(lhs._internal,rhs._internal))>
-{
-    typedef decltype(outerProduct(lhs._internal,rhs._internal)) ret_t;
-    iScalar<ret_t> ret;
-    ret._internal = outerProduct(lhs._internal,rhs._internal);
-    return ret;
-}
-/*
-    inline vComplexF outerProduct(const vComplexF &l, const vComplexF& r)
-    {
-        return l*r;
-    }
-    inline vComplexD outerProduct(const vComplexD &l, const vComplexD& r)
-    {
-        return l*r;
-    }
-    inline vRealF outerProduct(const vRealF &l, const vRealF& r)
-    {
-        return l*r;
-    }
-    inline vRealD outerProduct(const vRealD &l, const vRealD& r)
-    {
-        return l*r;
-    }
-*/
-    inline ComplexF outerProduct(const ComplexF &l, const ComplexF& r)
-    {
-        return l*r;
-    }
-    inline ComplexD outerProduct(const ComplexD &l, const ComplexD& r)
-    {
-        return l*r;
-    }
-    inline RealF outerProduct(const RealF &l, const RealF& r)
-    {
-        return l*r;
-    }
-    inline RealD outerProduct(const RealD &l, const RealD& r)
-    {
-        return l*r;
-    }
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-    /////////////////////////////////////////// CONJ         ///////////////////////////////////////////
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
- 
-// Conj function for scalar, vector, matrix
-template<class vtype> inline iScalar<vtype> conj(const iScalar<vtype>&r)
-{
-    iScalar<vtype> ret;
-    ret._internal = conj(r._internal);
-    return ret;
-}
-
-// Adj function for scalar, vector, matrix
-template<class vtype> inline iScalar<vtype> adj(const iScalar<vtype>&r)
-{
-    iScalar<vtype> ret;
-    ret._internal = adj(r._internal);
-    return ret;
-}
-template<class vtype,int N> inline iVector<vtype,N> adj(const iVector<vtype,N>&r)
-{
-    iVector<vtype,N> ret;
-    for(int i=0;i<N;i++){
-        ret._internal[i] = adj(r._internal[i]);
-    }
-    return ret;
-}
-template<class vtype,int N> inline iMatrix<vtype,N> adj(const iMatrix<vtype,N> &arg)
-{
-    iMatrix<vtype,N> ret;
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        ret._internal[c1][c2]=adj(arg._internal[c2][c1]);
-    }}
-    return ret;
-}
-
-/////////////////////////////////////////////////////////////////
-// Can only take the real/imag part of scalar objects, since
-// lattice objects of different complexity are non-conformable.
-/////////////////////////////////////////////////////////////////
-template<class itype> inline auto real(const iScalar<itype> &z) -> iScalar<decltype(real(z._internal))>
-{
-    iScalar<decltype(real(z._internal))> ret;
-    ret._internal = real(z._internal);
-    return ret;
-}
-template<class itype,int N> inline auto real(const iMatrix<itype,N> &z) -> iMatrix<decltype(real(z._internal[0][0])),N>
-{
-    iMatrix<decltype(real(z._internal[0][0])),N> ret;
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        ret._internal[c1][c2] = real(z._internal[c1][c2]);
-    }}
-    return ret;
-}
-template<class itype,int N> inline auto real(const iVector<itype,N> &z) -> iVector<decltype(real(z._internal[0])),N>
-{
-    iVector<decltype(real(z._internal[0])),N> ret;
-    for(int c1=0;c1<N;c1++){
-        ret._internal[c1] = real(z._internal[c1]);
-    }
-    return ret;
-}
-    
-template<class itype> inline auto imag(const iScalar<itype> &z) -> iScalar<decltype(imag(z._internal))>
-{
-    iScalar<decltype(imag(z._internal))> ret;
-    ret._internal = imag(z._internal);
-    return ret;
-}
-template<class itype,int N> inline auto imag(const iMatrix<itype,N> &z) -> iMatrix<decltype(imag(z._internal[0][0])),N>
-{
-    iMatrix<decltype(imag(z._internal[0][0])),N> ret;
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        ret._internal[c1][c2] = imag(z._internal[c1][c2]);
-    }}
-    return ret;
-}
-template<class itype,int N> inline auto imag(const iVector<itype,N> &z) -> iVector<decltype(imag(z._internal[0])),N>
-{
-    iVector<decltype(imag(z._internal[0])),N> ret;
-    for(int c1=0;c1<N;c1++){
-        ret._internal[c1] = imag(z._internal[c1]);
-    }
-    return ret;
-}
-
-    /////////////////////////////////
-    // Trace of scalar and matrix
-    /////////////////////////////////
-
-inline Complex trace( const Complex &arg){
-    return arg;
-}
-//inline vComplex trace(const vComplex &arg){
-//    return arg;
-//}
-template<class vtype,int N>
-inline auto trace(const iMatrix<vtype,N> &arg) -> iScalar<decltype(trace(arg._internal[0][0]))>
-{
-    iScalar<decltype( trace(arg._internal[0][0] )) > ret;
-    ZeroIt(ret._internal);
-    for(int i=0;i<N;i++){
-        ret._internal=ret._internal+trace(arg._internal[i][i]);
-    }
-    return ret;
-}
-template<class vtype>
-inline auto trace(const iScalar<vtype> &arg) -> iScalar<decltype(trace(arg._internal))>
-{
-    iScalar<decltype(trace(arg._internal))> ret;
-    ret._internal=trace(arg._internal);
-    return ret;
-}
-    
-/////////////////////////////////////////////////////////////////////////
-// Generic routine to promote object<complex> -> object<vcomplex>
-// Supports the array reordering transformation that gives me SIMD utilisation
-/////////////////////////////////////////////////////////////////////////
-/*
-template<template<class> class object>
-inline object<vComplex> splat(object<Complex >s){
-    object<vComplex> ret;
-    vComplex * v_ptr = (vComplex *)& ret;
-    Complex * s_ptr = (Complex *) &s;
-    for(int i=0;i<sizeof(ret);i+=sizeof(vComplex)){
-        vsplat(*(v_ptr++),*(s_ptr++));
-    }
-    return ret;
-}
-*/
-    
-    /////////////////////////////////////////////////////////////////////////////////////////
-    // Grid Support. Following will go into Grid.h.
-    /////////////////////////////////////////////////////////////////////////////////////////
-    // Cartesian grids
-    // dpo::Grid
-    // dpo::GridCartesian
-    // dpo::GridCartesianRedBlack
-    
-class Grid {
-public:
-    // Give Lattice access
-    template<class object> friend class Lattice;
-
-        
-//protected:
-        
-    // Lattice wide random support. not yet fully implemented. Need seed strategy
-    // and one generator per site.
-    //std::default_random_engine generator;
-  //    static std::mt19937  generator( 9 );
-
-        
-    // Grid information.
-    unsigned long _ndimension;
-    std::vector<int> _layout;     // Which dimensions get relayed out over simd lanes.
-    std::vector<int> _dimensions; // Dimensions of array
-    std::vector<int> _rdimensions;// Reduced dimensions with simd lane images removed
-    std::vector<int> _ostride;    // Outer stride for each dimension
-    std::vector<int> _istride;    // Inner stride i.e. within simd lane
-    int _osites;                  // _isites*_osites = product(dimensions).
-    int _isites;
-        
-    // subslice information
-    std::vector<int> _slice_block;
-    std::vector<int> _slice_stride;
-    std::vector<int> _slice_nblock;
-public:
-    
-    // These routines are key. Subdivide the linearised cartesian index into
-    //      "inner" index identifying which simd lane of object<vFcomplex> is associated with coord
-    //      "outer" index identifying which element of _odata in class "Lattice" is associated with coord.
-    // Compared to, say, Blitz++ we simply need to store BOTH an inner stride and an outer
-    // stride per dimension. The cost of evaluating the indexing information is doubled for an n-dimensional
-    // coordinate. Note, however, for data parallel operations the "inner" indexing cost is not paid and all
-    // lanes are operated upon simultaneously.
-    
-    inline int oIndexReduced(std::vector<int> &rcoor)
-    {
-        int idx=0;
-        for(int d=0;d<_ndimension;d++) idx+=_ostride[d]*rcoor[d];
-        return idx;
-    }
-    virtual int oIndex(std::vector<int> &coor)
-    {
-        int idx=0;
-        for(int d=0;d<_ndimension;d++) idx+=_ostride[d]*(coor[d]%_rdimensions[d]);
-        return idx;
-    }
-    inline int iIndex(std::vector<int> &rcoor)
-    {
-        int idx=0;
-        for(int d=0;d<_ndimension;d++) idx+=_istride[d]*(rcoor[d]/_rdimensions[d]);
-        return idx;
-    }
-        
-    inline int oSites(void) { return _osites; };
-    inline int iSites(void) { return _isites; };
-    virtual int CheckerBoard(std::vector<int> site)=0;
-    virtual int CheckerBoardDestination(int source_cb,int shift)=0;
-    virtual int CheckerBoardShift(int source_cb,int dim,int shift)=0;
-};
-
-////////////////////////////////////////////////////////////////////
-// A lattice of something, but assume the something is SIMDized.
-////////////////////////////////////////////////////////////////////
-template<typename _Tp>
-class myallocator {
-public: 
-  typedef std::size_t     size_type;
-  typedef std::ptrdiff_t  difference_type;
-  typedef _Tp*       pointer;
-  typedef const _Tp* const_pointer;
-  typedef _Tp&       reference;
-  typedef const _Tp& const_reference;
-  typedef _Tp        value_type;
-
-  template<typename _Tp1>  struct rebind { typedef myallocator<_Tp1> other; };
-  myallocator() throw() { }
-  myallocator(const myallocator&) throw() { }
-  template<typename _Tp1> myallocator(const myallocator<_Tp1>&) throw() { }
-  ~myallocator() throw() { }
-  pointer address(reference __x) const { return &__x; }
-  const_pointer address(const_reference __x) const { return &__x; }
-  size_type  max_size() const throw() { return size_t(-1) / sizeof(_Tp); }
-  // Should override allocate and deallocate
-  pointer allocate(size_type __n, const void* = 0)
-  { 
-    //_Tp * ptr = (_Tp *) memalign(sizeof(_Tp),__n*sizeof(_Tp));
-    // _Tp * ptr = (_Tp *) memalign(128,__n*sizeof(_Tp));
-#ifdef AVX512
-    _Tp * ptr = (_Tp *) memalign(128,__n*sizeof(_Tp));
-#else
-    _Tp * ptr = (_Tp *) _mm_malloc(__n*sizeof(_Tp),128);
-#endif
-
-    return ptr;
-  }
-  void deallocate(pointer __p, size_type) { 
-    free(__p); 
-  }
-  void construct(pointer __p, const _Tp& __val) { };
-  void construct(pointer __p) { };
-  void destroy(pointer __p) { };
-};
-
-template<typename _Tp>  inline bool
-operator==(const myallocator<_Tp>&, const myallocator<_Tp>&){ return true; }
-
-template<typename _Tp>  inline bool
-operator!=(const myallocator<_Tp>&, const myallocator<_Tp>&){ return false; }
-
-    
-}; // namespace dpo
-
 
 #endif
diff --git a/Grid_Lattice.h b/Grid_Lattice.h
index f18784fb..47b85b0f 100644
--- a/Grid_Lattice.h
+++ b/Grid_Lattice.h
@@ -1,5 +1,7 @@
+#ifndef GRID_LATTICE_H
+#define GRID_LATTICE_H
+
 #include "Grid.h"
-#include "Grid_vComplexD.h"
 
 namespace dpo {
 
@@ -9,7 +11,7 @@ class Lattice
 public:
     Grid *_grid;
     int checkerboard;
-    std::vector<vobj,myallocator<vobj> > _odata;
+    std::vector<vobj,alignedAllocator<vobj> > _odata;
 
 public:
     
@@ -554,95 +556,5 @@ public:
         return ret;
     }
 
-    
-namespace QCD {
-
-    static const int Nc=3;
-    static const int Ns=4;
-
-    static const int CbRed  =0;
-    static const int CbBlack=1;
-
-    // QCD iMatrix types
-    template<typename vtype> using iSinglet          = iScalar<iScalar<vtype> > ;
-    template<typename vtype> using iSpinMatrix       = iMatrix<iScalar<vtype>, Ns>;
-    template<typename vtype> using iSpinColourMatrix = iMatrix<iMatrix<vtype, Nc>, Ns>;
-
-    template<typename vtype> using iColourMatrix     = iScalar<iMatrix<vtype, Nc>> ;
-
-    template<typename vtype> using iSpinVector       = iVector<iScalar<vtype>, Ns>;
-    template<typename vtype> using iColourVector     = iScalar<iVector<vtype, Nc> >;
-    template<typename vtype> using iSpinColourVector = iVector<iVector<vtype, Nc>, Ns>;
-
-    typedef iSinglet<Complex >          TComplex;    // This is painful. Tensor singlet complex type.
-    typedef iSinglet<vComplex >         vTComplex;
-    typedef iSinglet<Real >          TReal;    // This is painful. Tensor singlet complex type.
-
-    typedef iSpinMatrix<Complex >       SpinMatrix;
-    typedef iColourMatrix<Complex >     ColourMatrix;
-    typedef iSpinColourMatrix<Complex > SpinColourMatrix;
-
-    typedef iSpinVector<Complex >       SpinVector;
-    typedef iColourVector<Complex >     ColourVector;
-    typedef iSpinColourVector<Complex > SpinColourVector;
-
-    
-    typedef iSpinMatrix<vComplex >       vSpinMatrix;
-    typedef iColourMatrix<vComplex >     vColourMatrix;
-    typedef iSpinColourMatrix<vComplex > vSpinColourMatrix;
-    
-    typedef iSpinVector<vComplex >       vSpinVector;
-    typedef iColourVector<vComplex >     vColourVector;
-    typedef iSpinColourVector<vComplex > vSpinColourVector;
-
-    
-    typedef Lattice<vTComplex>         LatticeComplex;
-    
-    typedef Lattice<vColourMatrix>     LatticeColourMatrix;
-    typedef Lattice<vSpinMatrix>       LatticeSpinMatrix;
-    typedef Lattice<vSpinColourMatrix> LatticePropagator;
-    typedef LatticePropagator LatticeSpinColourMatrix;
-
-    typedef Lattice<vSpinColourVector> LatticeFermion;
-    typedef Lattice<vSpinColourVector> LatticeSpinColourVector;
-    typedef Lattice<vSpinVector>       LatticeSpinVector;
-    typedef Lattice<vColourVector>     LatticeColourVector;
-
-    // localNorm2,
-    template<class tt>
-    inline LatticeComplex localNorm2 (const Lattice<tt> &rhs)
-    {
-        LatticeComplex ret(rhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<rhs._grid->oSites(); ss++){
-            ret._odata[ss]=trace(adj(rhs)*rhs);
-        }
-        return ret;
-    }
-    // localInnerProduct
-    template<class tt>
-    inline LatticeComplex localInnerProduct (const Lattice<tt> &lhs,const Lattice<tt> &rhs)
-    {
-        LatticeComplex ret(rhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<rhs._grid->oSites(); ss++){
-            ret._odata[ss]=localInnerProduct(lhs._odata[ss],rhs._odata[ss]);
-        }
-        return ret;
-    }
-    
-    // outerProduct Scalar x Scalar -> Scalar
-    //              Vector x Vector -> Matrix
-    template<class ll,class rr>
-    inline auto outerProduct (const Lattice<ll> &lhs,const Lattice<rr> &rhs) -> Lattice<decltype(outerProduct(lhs._odata[0],rhs._odata[0]))>
-    {
-        Lattice<decltype(outerProduct(lhs._odata[0],rhs._odata[0]))> ret(rhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<rhs._grid->oSites(); ss++){
-            ret._odata[ss]=outerProduct(lhs._odata[ss],rhs._odata[ss]);
-        }
-        return ret;
-     }
-}   //namespace QCD
-
 }
+#endif
diff --git a/Grid_QCD.h b/Grid_QCD.h
new file mode 100644
index 00000000..92a02506
--- /dev/null
+++ b/Grid_QCD.h
@@ -0,0 +1,94 @@
+#ifndef GRID_QCD_H
+#define GRID_QCD_H
+namespace dpo{
+namespace QCD {
+
+    static const int Nc=3;
+    static const int Ns=4;
+
+    static const int CbRed  =0;
+    static const int CbBlack=1;
+
+    // QCD iMatrix types
+    template<typename vtype> using iSinglet          = iScalar<iScalar<vtype> > ;
+    template<typename vtype> using iSpinMatrix       = iMatrix<iScalar<vtype>, Ns>;
+    template<typename vtype> using iSpinColourMatrix = iMatrix<iMatrix<vtype, Nc>, Ns>;
+
+    template<typename vtype> using iColourMatrix     = iScalar<iMatrix<vtype, Nc>> ;
+
+    template<typename vtype> using iSpinVector       = iVector<iScalar<vtype>, Ns>;
+    template<typename vtype> using iColourVector     = iScalar<iVector<vtype, Nc> >;
+    template<typename vtype> using iSpinColourVector = iVector<iVector<vtype, Nc>, Ns>;
+
+    typedef iSinglet<Complex >          TComplex;    // This is painful. Tensor singlet complex type.
+    typedef iSinglet<vComplex >         vTComplex;
+    typedef iSinglet<Real >          TReal;    // This is painful. Tensor singlet complex type.
+
+    typedef iSpinMatrix<Complex >       SpinMatrix;
+    typedef iColourMatrix<Complex >     ColourMatrix;
+    typedef iSpinColourMatrix<Complex > SpinColourMatrix;
+
+    typedef iSpinVector<Complex >       SpinVector;
+    typedef iColourVector<Complex >     ColourVector;
+    typedef iSpinColourVector<Complex > SpinColourVector;
+
+    
+    typedef iSpinMatrix<vComplex >       vSpinMatrix;
+    typedef iColourMatrix<vComplex >     vColourMatrix;
+    typedef iSpinColourMatrix<vComplex > vSpinColourMatrix;
+    
+    typedef iSpinVector<vComplex >       vSpinVector;
+    typedef iColourVector<vComplex >     vColourVector;
+    typedef iSpinColourVector<vComplex > vSpinColourVector;
+
+    
+    typedef Lattice<vTComplex>         LatticeComplex;
+    
+    typedef Lattice<vColourMatrix>     LatticeColourMatrix;
+    typedef Lattice<vSpinMatrix>       LatticeSpinMatrix;
+    typedef Lattice<vSpinColourMatrix> LatticePropagator;
+    typedef LatticePropagator LatticeSpinColourMatrix;
+
+    typedef Lattice<vSpinColourVector> LatticeFermion;
+    typedef Lattice<vSpinColourVector> LatticeSpinColourVector;
+    typedef Lattice<vSpinVector>       LatticeSpinVector;
+    typedef Lattice<vColourVector>     LatticeColourVector;
+
+    // localNorm2,
+    template<class tt>
+    inline LatticeComplex localNorm2 (const Lattice<tt> &rhs)
+    {
+        LatticeComplex ret(rhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<rhs._grid->oSites(); ss++){
+            ret._odata[ss]=trace(adj(rhs)*rhs);
+        }
+        return ret;
+    }
+    // localInnerProduct
+    template<class tt>
+    inline LatticeComplex localInnerProduct (const Lattice<tt> &lhs,const Lattice<tt> &rhs)
+    {
+        LatticeComplex ret(rhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<rhs._grid->oSites(); ss++){
+            ret._odata[ss]=localInnerProduct(lhs._odata[ss],rhs._odata[ss]);
+        }
+        return ret;
+    }
+    
+    // outerProduct Scalar x Scalar -> Scalar
+    //              Vector x Vector -> Matrix
+    template<class ll,class rr>
+    inline auto outerProduct (const Lattice<ll> &lhs,const Lattice<rr> &rhs) -> Lattice<decltype(outerProduct(lhs._odata[0],rhs._odata[0]))>
+    {
+        Lattice<decltype(outerProduct(lhs._odata[0],rhs._odata[0]))> ret(rhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<rhs._grid->oSites(); ss++){
+            ret._odata[ss]=outerProduct(lhs._odata[ss],rhs._odata[ss]);
+        }
+        return ret;
+     }
+}   //namespace QCD
+} // dpo
+#endif
diff --git a/Grid_aligned_allocator.h b/Grid_aligned_allocator.h
new file mode 100644
index 00000000..6040c233
--- /dev/null
+++ b/Grid_aligned_allocator.h
@@ -0,0 +1,56 @@
+#ifndef GRID_ALIGNED_ALLOCATOR_H
+#define GRID_ALIGNED_ALLOCATOR_H
+namespace dpo {
+
+////////////////////////////////////////////////////////////////////
+// A lattice of something, but assume the something is SIMDized.
+////////////////////////////////////////////////////////////////////
+template<typename _Tp>
+class alignedAllocator {
+public: 
+  typedef std::size_t     size_type;
+  typedef std::ptrdiff_t  difference_type;
+  typedef _Tp*       pointer;
+  typedef const _Tp* const_pointer;
+  typedef _Tp&       reference;
+  typedef const _Tp& const_reference;
+  typedef _Tp        value_type;
+
+  template<typename _Tp1>  struct rebind { typedef alignedAllocator<_Tp1> other; };
+  alignedAllocator() throw() { }
+  alignedAllocator(const alignedAllocator&) throw() { }
+  template<typename _Tp1> alignedAllocator(const alignedAllocator<_Tp1>&) throw() { }
+  ~alignedAllocator() throw() { }
+  pointer address(reference __x) const { return &__x; }
+  const_pointer address(const_reference __x) const { return &__x; }
+  size_type  max_size() const throw() { return size_t(-1) / sizeof(_Tp); }
+  // Should override allocate and deallocate
+  pointer allocate(size_type __n, const void* = 0)
+  { 
+    //_Tp * ptr = (_Tp *) memalign(sizeof(_Tp),__n*sizeof(_Tp));
+    // _Tp * ptr = (_Tp *) memalign(128,__n*sizeof(_Tp));
+#ifdef AVX512
+    _Tp * ptr = (_Tp *) memalign(128,__n*sizeof(_Tp));
+#else
+    _Tp * ptr = (_Tp *) _mm_malloc(__n*sizeof(_Tp),128);
+#endif
+
+    return ptr;
+  }
+  void deallocate(pointer __p, size_type) { 
+    free(__p); 
+  }
+  void construct(pointer __p, const _Tp& __val) { };
+  void construct(pointer __p) { };
+  void destroy(pointer __p) { };
+};
+
+template<typename _Tp>  inline bool
+operator==(const alignedAllocator<_Tp>&, const alignedAllocator<_Tp>&){ return true; }
+
+template<typename _Tp>  inline bool
+operator!=(const alignedAllocator<_Tp>&, const alignedAllocator<_Tp>&){ return false; }
+
+    
+}; // namespace dpo
+#endif
diff --git a/Grid_config.h b/Grid_config.h
new file mode 100644
index 00000000..b882917d
--- /dev/null
+++ b/Grid_config.h
@@ -0,0 +1,101 @@
+/* Grid_config.h.  Generated from Grid_config.h.in by configure.  */
+/* Grid_config.h.in.  Generated from configure.ac by autoheader.  */
+
+/* AVX */
+#define AVX1 1
+
+/* AVX2 */
+/* #undef AVX2 */
+
+/* AVX512 */
+/* #undef AVX512 */
+
+/* Define to 1 if you have the `gettimeofday' function. */
+#define HAVE_GETTIMEOFDAY 1
+
+/* Define to 1 if you have the <inttypes.h> header file. */
+#define HAVE_INTTYPES_H 1
+
+/* Define to 1 if you have the <malloc.h> header file. */
+/* #undef HAVE_MALLOC_H */
+
+/* Define to 1 if you have the <malloc/malloc.h> header file. */
+#define HAVE_MALLOC_MALLOC_H 1
+
+/* Define to 1 if you have the <memory.h> header file. */
+#define HAVE_MEMORY_H 1
+
+/* Define to 1 if you have the <stdint.h> header file. */
+#define HAVE_STDINT_H 1
+
+/* Define to 1 if you have the <stdlib.h> header file. */
+#define HAVE_STDLIB_H 1
+
+/* Define to 1 if you have the <strings.h> header file. */
+#define HAVE_STRINGS_H 1
+
+/* Define to 1 if you have the <string.h> header file. */
+#define HAVE_STRING_H 1
+
+/* Define to 1 if you have the <sys/stat.h> header file. */
+#define HAVE_SYS_STAT_H 1
+
+/* Define to 1 if you have the <sys/types.h> header file. */
+#define HAVE_SYS_TYPES_H 1
+
+/* Define to 1 if you have the <unistd.h> header file. */
+#define HAVE_UNISTD_H 1
+
+/* Define to 1 if the system has the `aligned' variable attribute */
+#define HAVE_VAR_ATTRIBUTE_ALIGNED 1
+
+/* Name of package */
+#define PACKAGE "grid"
+
+/* Define to the address where bug reports for this package should be sent. */
+#define PACKAGE_BUGREPORT "paboyle@ph.ed.ac.uk"
+
+/* Define to the full name of this package. */
+#define PACKAGE_NAME "Grid"
+
+/* Define to the full name and version of this package. */
+#define PACKAGE_STRING "Grid 1.0"
+
+/* Define to the one symbol short name of this package. */
+#define PACKAGE_TARNAME "grid"
+
+/* Define to the home page for this package. */
+#define PACKAGE_URL ""
+
+/* Define to the version of this package. */
+#define PACKAGE_VERSION "1.0"
+
+/* SSE2 */
+/* #undef SSE2 */
+
+/* Define to 1 if you have the ANSI C header files. */
+#define STDC_HEADERS 1
+
+/* Version number of package */
+#define VERSION "1.0"
+
+/* Define for Solaris 2.5.1 so the uint32_t typedef from <sys/synch.h>,
+   <pthread.h>, or <semaphore.h> is not used. If the typedef were allowed, the
+   #define below would cause a syntax error. */
+/* #undef _UINT32_T */
+
+/* Define for Solaris 2.5.1 so the uint64_t typedef from <sys/synch.h>,
+   <pthread.h>, or <semaphore.h> is not used. If the typedef were allowed, the
+   #define below would cause a syntax error. */
+/* #undef _UINT64_T */
+
+/* Define to `unsigned int' if <sys/types.h> does not define. */
+/* #undef size_t */
+
+/* Define to the type of an unsigned integer type of width exactly 32 bits if
+   such a type exists and the standard includes do not define it. */
+/* #undef uint32_t */
+
+/* Define to the type of an unsigned integer type of width exactly 64 bits if
+   such a type exists and the standard includes do not define it. */
+/* #undef uint64_t */
diff --git a/Grid_math_types.h b/Grid_math_types.h
new file mode 100644
index 00000000..405d5c9a
--- /dev/null
+++ b/Grid_math_types.h
@@ -0,0 +1,819 @@
+#ifndef GRID_MATH_TYPES_H
+#define GRID_MATH_TYPES_H
+namespace dpo {
+///////////////////////////////////////////////////
+// Scalar, Vector, Matrix objects.
+// These can be composed to form tensor products of internal indices.
+///////////////////////////////////////////////////
+    
+template<class vtype> class iScalar
+{
+public:
+  SIMDalign vtype _internal;
+    iScalar(){};
+    iScalar(Zero &z){ *this = zero; };
+    iScalar<vtype> & operator= (const Zero &hero){
+        zeroit(*this);
+        return *this;
+    }
+    friend void zeroit(iScalar<vtype> &that){
+        zeroit(that._internal);
+    }
+    // Unary negation
+    friend inline iScalar<vtype> operator -(const iScalar<vtype> &r) {
+        iScalar<vtype> ret;
+        ret._internal= -r._internal;
+        return ret;
+    }
+    // *=,+=,-= operators
+    inline iScalar<vtype> &operator *=(const iScalar<vtype> &r) {
+        *this = (*this)*r;
+        return *this;
+    }
+    inline iScalar<vtype> &operator -=(const iScalar<vtype> &r) {
+        *this = (*this)-r;
+        return *this;
+    }
+    inline iScalar<vtype> &operator +=(const iScalar<vtype> &r) {
+        *this = (*this)+r;
+        return *this;
+    }
+    
+
+};
+    
+template<class vtype,int N> class iVector
+{
+public:
+  SIMDalign vtype _internal[N];
+    iVector(Zero &z){ *this = zero; };
+    iVector() {};
+    iVector<vtype,N> & operator= (Zero &hero){
+        zeroit(*this);
+        return *this;
+    }
+    friend void zeroit(iVector<vtype,N> &that){
+        for(int i=0;i<N;i++){
+            zeroit(that._internal[i]);
+        }
+    }
+    // Unary negation
+    friend inline iVector<vtype,N> operator -(const iVector<vtype,N> &r) {
+        iVector<vtype,N> ret;
+        for(int i=0;i<N;i++) ret._internal[i]= -r._internal[i];
+        return ret;
+    }
+    // *=,+=,-= operators
+    inline iVector<vtype,N> &operator *=(const iScalar<vtype> &r) {
+        *this = (*this)*r;
+        return *this;
+    }
+    inline iVector<vtype,N> &operator -=(const iVector<vtype,N> &r) {
+        *this = (*this)-r;
+        return *this;
+    }
+    inline iVector<vtype,N> &operator +=(const iVector<vtype,N> &r) {
+        *this = (*this)+r;
+        return *this;
+    }
+
+};
+    
+    
+template<class vtype,int N> class iMatrix
+{
+public:
+  SIMDalign    vtype _internal[N][N];
+    iMatrix(Zero &z){ *this = zero; };
+    iMatrix() {};
+    iMatrix<vtype,N> & operator= (Zero &hero){
+        zeroit(*this);
+        return *this;
+    }
+    friend void zeroit(iMatrix<vtype,N> &that){
+        for(int i=0;i<N;i++){
+        for(int j=0;j<N;j++){
+                zeroit(that._internal[i][j]);
+        }}
+    }
+    // Unary negation
+    friend inline iMatrix<vtype,N> operator -(const iMatrix<vtype,N> &r) {
+        iMatrix<vtype,N> ret;
+        for(int i=0;i<N;i++){
+        for(int j=0;j<N;j++){
+            ret._internal[i][j]= -r._internal[i][j];
+        }}
+        return ret;
+    }
+    // *=,+=,-= operators
+    template<class T>
+    inline iMatrix<vtype,N> &operator *=(const T &r) {
+        *this = (*this)*r;
+        return *this;
+    }
+    template<class T>
+    inline iMatrix<vtype,N> &operator -=(const T &r) {
+        *this = (*this)-r;
+        return *this;
+    }
+    template<class T>
+    inline iMatrix<vtype,N> &operator +=(const T &r) {
+        *this = (*this)+r;
+        return *this;
+    }
+
+};
+
+    
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    /////////////////////////////////////////// ADD         ///////////////////////////////////////////
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    
+
+// ADD is simple for now; cannot mix types and straightforward template
+// Scalar +/- Scalar
+// Vector +/- Vector
+// Matrix +/- Matrix
+template<class vtype,class ltype,class rtype> inline void add(iScalar<vtype> * __restrict__ ret,
+                                                              const iScalar<ltype> * __restrict__ lhs,
+                                                              const iScalar<rtype> * __restrict__ rhs)
+{
+    add(&ret->_internal,&lhs->_internal,&rhs->_internal);
+}
+template<class vtype,class ltype,class rtype,int N> inline void add(iVector<vtype,N> * __restrict__ ret,
+                                                                    const iVector<ltype,N> * __restrict__ lhs,
+                                                                    const iVector<rtype,N> * __restrict__ rhs)
+{
+    for(int c=0;c<N;c++){
+        ret->_internal[c]=lhs->_internal[c]+rhs->_internal[c];
+    }
+    return;
+}
+template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
+                                                                      const iMatrix<ltype,N> * __restrict__ lhs,
+                                                                      const iMatrix<rtype,N> * __restrict__ rhs)
+{
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        add(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal[c1][c2]);
+    }}
+    return;
+}
+template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
+                                                                      const iScalar<ltype>   * __restrict__ lhs,
+                                                                      const iMatrix<rtype,N> * __restrict__ rhs)
+{
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        add(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
+    }}
+    return;
+}
+template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
+                                                                      const iMatrix<ltype,N> * __restrict__ lhs,
+                                                                      const iScalar<rtype>   * __restrict__ rhs)
+{
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        if ( c1==c2)
+            add(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
+        else
+            ret->_internal[c1][c2]=lhs->_internal[c1][c2];
+    }}
+    return;
+}
+// Need to figure multi-precision.
+template<class Mytype>  Mytype timesI(Mytype &r)
+{
+    iScalar<Complex> i;
+    i._internal = Complex(0,1);
+    return r*i;
+}
+
+                // + operator for scalar, vector, matrix
+template<class ltype,class rtype>
+//inline auto operator + (iScalar<ltype>& lhs,iScalar<rtype>&& rhs) -> iScalar<decltype(lhs._internal + rhs._internal)>
+inline auto operator + (const iScalar<ltype>& lhs,const iScalar<rtype>& rhs) -> iScalar<decltype(lhs._internal + rhs._internal)>
+{
+    typedef iScalar<decltype(lhs._internal+rhs._internal)> ret_t;
+    ret_t ret;
+    add(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class ltype,class rtype,int N>
+inline auto operator + (const iVector<ltype,N>& lhs,const iVector<rtype,N>& rhs) ->iVector<decltype(lhs._internal[0]+rhs._internal[0]),N>
+{
+    typedef iVector<decltype(lhs._internal[0]+rhs._internal[0]),N> ret_t;
+    ret_t ret;
+    add(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class ltype,class rtype,int N>
+inline auto operator + (const iMatrix<ltype,N>& lhs,const iMatrix<rtype,N>& rhs) ->iMatrix<decltype(lhs._internal[0][0]+rhs._internal[0][0]),N>
+{
+    typedef iMatrix<decltype(lhs._internal[0][0]+rhs._internal[0][0]),N> ret_t;
+    ret_t ret;
+    add(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class ltype,class rtype,int N>
+inline auto operator + (const iScalar<ltype>& lhs,const iMatrix<rtype,N>& rhs)->iMatrix<decltype(lhs._internal+rhs._internal[0][0]),N>
+{
+    typedef iMatrix<decltype(lhs._internal+rhs._internal[0][0]),N> ret_t;
+    ret_t ret;
+    add(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class ltype,class rtype,int N>
+inline auto operator + (const iMatrix<ltype,N>& lhs,const iScalar<rtype>& rhs)->iMatrix<decltype(lhs._internal[0][0]+rhs._internal),N>
+{
+    typedef iMatrix<decltype(lhs._internal[0][0]+rhs._internal),N> ret_t;
+    ret_t ret;
+    add(&ret,&lhs,&rhs);
+    return ret;
+}
+
+
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    /////////////////////////////////////////// SUB         ///////////////////////////////////////////
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    
+
+// SUB is simple for now; cannot mix types and straightforward template
+// Scalar +/- Scalar
+// Vector +/- Vector
+// Matrix +/- Matrix
+// Matrix /- scalar
+template<class vtype,class ltype,class rtype> inline void sub(iScalar<vtype> * __restrict__ ret,
+                                                              const iScalar<ltype> * __restrict__ lhs,
+                                                              const iScalar<rtype> * __restrict__ rhs)
+{
+    sub(&ret->_internal,&lhs->_internal,&rhs->_internal);
+}
+
+template<class vtype,class ltype,class rtype,int N> inline void sub(iVector<vtype,N> * __restrict__ ret,
+                                                                    const iVector<ltype,N> * __restrict__ lhs,
+                                                                    const iVector<rtype,N> * __restrict__ rhs)
+{
+    for(int c=0;c<N;c++){
+        ret->_internal[c]=lhs->_internal[c]-rhs->_internal[c];
+    }
+    return;
+}
+template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
+                                                                     const iMatrix<ltype,N> * __restrict__ lhs,
+                                                                     const iMatrix<rtype,N> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        sub(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal[c1][c2]);
+    }}
+    return;
+}
+template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
+                                                                     const iScalar<ltype> * __restrict__ lhs,
+                                                                     const iMatrix<rtype,N> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        if ( c1!=c2) {
+            sub(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
+        } else {
+            // Fails -- need unary minus. Catalogue other unops?
+            ret->_internal[c1][c2]=zero;
+            ret->_internal[c1][c2]=ret->_internal[c1][c2]-rhs->_internal[c1][c2];
+
+        }
+    }}
+    return;
+}
+template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
+                                                                     const iMatrix<ltype,N> * __restrict__ lhs,
+                                                                     const iScalar<rtype> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        if ( c1!=c2)
+            sub(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
+        else
+            ret->_internal[c1][c2]=lhs->_internal[c1][c2];
+    }}
+    return;
+}
+
+template<class v> void vprefetch(const iScalar<v> &vv)
+{
+  vprefetch(vv._internal);
+}
+template<class v,int N> void vprefetch(const iVector<v,N> &vv)
+{
+  for(int i=0;i<N;i++){
+    vprefetch(vv._internal[i]);
+  }
+}
+template<class v,int N> void vprefetch(const iMatrix<v,N> &vv)
+{
+  for(int i=0;i<N;i++){
+  for(int j=0;j<N;j++){
+    vprefetch(vv._internal[i][j]);
+  }}
+}
+
+    // - operator for scalar, vector, matrix
+template<class ltype,class rtype> inline auto
+operator - (const iScalar<ltype>& lhs, const iScalar<rtype>& rhs) -> iScalar<decltype(lhs._internal - rhs._internal)>
+{
+    typedef iScalar<decltype(lhs._internal-rhs._internal)> ret_t;
+    ret_t ret;
+    sub(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class ltype,class rtype,int N>
+inline auto operator - (const iVector<ltype,N>& lhs,const iVector<rtype,N>& rhs) ->iVector<decltype(lhs._internal[0]-rhs._internal[0]),N>
+{
+    typedef iVector<decltype(lhs._internal[0]-rhs._internal[0]),N> ret_t;
+    ret_t ret;
+    sub(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class ltype,class rtype,int N>
+inline auto operator - (const iMatrix<ltype,N>& lhs,const iMatrix<rtype,N>& rhs) ->iMatrix<decltype(lhs._internal[0][0]-rhs._internal[0][0]),N>
+{
+    typedef iMatrix<decltype(lhs._internal[0][0]-rhs._internal[0][0]),N> ret_t;
+    ret_t ret;
+    sub(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class ltype,class rtype,int N>
+inline auto operator - (const iScalar<ltype>& lhs,const iMatrix<rtype,N>& rhs)->iMatrix<decltype(lhs._internal-rhs._internal[0][0]),N>
+{
+    typedef iMatrix<decltype(lhs._internal-rhs._internal[0][0]),N> ret_t;
+    ret_t ret;
+    sub(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class ltype,class rtype,int N>
+inline auto operator - (const iMatrix<ltype,N>& lhs,const iScalar<rtype>& rhs)->iMatrix<decltype(lhs._internal[0][0]-rhs._internal),N>
+{
+    typedef iMatrix<decltype(lhs._internal[0][0]-rhs._internal),N> ret_t;
+    ret_t ret;
+    sub(&ret,&lhs,&rhs);
+    return ret;
+}
+
+///////////////////////////////////////////////////////////////////////////////////////////////////
+/////////////////////////////////////////// MAC         ///////////////////////////////////////////
+///////////////////////////////////////////////////////////////////////////////////////////////////
+
+    ///////////////////////////
+    // Legal multiplication table
+    ///////////////////////////
+    // scal x scal = scal
+    // mat x  mat  = mat
+    // mat  x scal = mat
+    // scal x mat  = mat
+    // mat  x vec  = vec
+    // vec  x scal = vec
+    // scal x vec  = vec
+    ///////////////////////////
+template<class rtype,class vtype,class mtype>
+inline  void mac(iScalar<rtype> * __restrict__ ret,const iScalar<vtype> * __restrict__ lhs,const iScalar<mtype> * __restrict__ rhs)
+{
+    mac(&ret->_internal,&lhs->_internal,&rhs->_internal);
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+    for(int c3=0;c3<N;c3++){
+        mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c3],&rhs->_internal[c3][c2]);
+    }}}
+    return;
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs){
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
+    }}
+    return;
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iScalar<ltype> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        mac(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
+    }}
+    return;
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mac(iVector<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iVector<rtype,N> * __restrict__ rhs)
+{
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        mac(&ret->_internal[c1],&lhs->_internal[c1][c2],&rhs->_internal[c2]);
+    }}
+    return;
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mac(iVector<rrtype,N> * __restrict__ ret,const iScalar<ltype> * __restrict__ lhs,const iVector<rtype,N> * __restrict__ rhs)
+{
+    for(int c1=0;c1<N;c1++){
+        mac(&ret->_internal[c1],&lhs->_internal,&rhs->_internal[c1]);
+    }
+    return;
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mac(iVector<rrtype,N> * __restrict__ ret,const iVector<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs)
+{
+    for(int c1=0;c1<N;c1++){
+        mac(&ret->_internal[c1],&lhs->_internal[c1],&rhs->_internal);
+    }
+    return;
+}
+
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    /////////////////////////////////////////// MUL         ///////////////////////////////////////////
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+
+    
+template<class rtype,class vtype,class mtype>
+inline void mult(iScalar<rtype> * __restrict__ ret,const iScalar<mtype> * __restrict__ lhs,const iScalar<vtype> * __restrict__ rhs){
+    mult(&ret->_internal,&lhs->_internal,&rhs->_internal);
+}
+
+template<class rrtype,class ltype,class rtype,int N>
+inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        mult(&ret->_internal[c1][c2],&lhs->_internal[c1][0],&rhs->_internal[0][c2]);
+        for(int c3=1;c3<N;c3++){
+            mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c3],&rhs->_internal[c3][c2]);
+        }
+    }}
+    return;
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        mult(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
+    }}
+    return;
+}
+
+template<class rrtype,class ltype,class rtype, int N>
+inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iScalar<ltype>   * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        mult(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
+    }}
+    return;
+}
+// Matrix left multiplies vector
+template<class rtype,class vtype,class mtype,int N>
+inline void mult(iVector<rtype,N> * __restrict__ ret,const iMatrix<mtype,N> * __restrict__ lhs,const iVector<vtype,N> * __restrict__ rhs)
+{
+    for(int c1=0;c1<N;c1++){
+        mult(&ret->_internal[c1],&lhs->_internal[c1][0],&rhs->_internal[0]);
+        for(int c2=1;c2<N;c2++){
+            mac(&ret->_internal[c1],&lhs->_internal[c1][c2],&rhs->_internal[c2]);
+        }
+    }
+    return;
+}
+template<class rtype,class vtype,class mtype,int N>
+inline void mult(iVector<rtype,N> * __restrict__ ret,
+                 const iScalar<mtype>   * __restrict__ lhs,
+                 const iVector<vtype,N> * __restrict__ rhs){
+    for(int c1=0;c1<N;c1++){
+        mult(&ret->_internal[c1],&lhs->_internal,&rhs->_internal[c1]);
+    }
+}
+template<class rtype,class vtype,class mtype,int N>
+inline void mult(iVector<rtype,N> * __restrict__ ret,
+                 const iVector<vtype,N> * __restrict__ rhs,
+                 const iScalar<mtype> * __restrict__ lhs){
+    mult(ret,lhs,rhs);
+}
+    
+
+
+template<class rtype,class vtype,class mtype,int N> inline
+iVector<rtype,N> operator * (const iMatrix<mtype,N>& lhs,const iVector<vtype,N>& rhs)
+{
+    iVector<rtype,N> ret;
+    mult(&ret,&lhs,&rhs);
+    return ret;
+}
+
+template<class rtype,class vtype,class mtype,int N> inline
+iVector<rtype,N> operator * (const iScalar<mtype>& lhs,const iVector<vtype,N>& rhs)
+{
+    iVector<rtype,N> ret;
+    mult(&ret,&lhs,&rhs);
+    return ret;
+}
+
+template<class rtype,class vtype,class mtype,int N> inline
+iVector<rtype,N> operator * (const iVector<mtype,N>& lhs,const iScalar<vtype>& rhs)
+{
+    iVector<rtype,N> ret;
+    mult(&ret,&lhs,&rhs);
+    return ret;
+}
+    
+    //////////////////////////////////////////////////////////////////
+    // Glue operators to mult routines. Must resolve return type cleverly from typeof(internal)
+    // since nesting matrix<scalar> x matrix<matrix>-> matrix<matrix>
+    // while         matrix<scalar> x matrix<scalar>-> matrix<scalar>
+    // so return type depends on argument types in nasty way.
+    //////////////////////////////////////////////////////////////////
+    // scal x scal = scal
+    // mat x  mat  = mat
+    // mat  x scal = mat
+    // scal x mat  = mat
+    // mat  x vec  = vec
+    // vec  x scal = vec
+    // scal x vec  = vec
+    
+template<class l,class r>
+inline auto operator * (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(lhs._internal * rhs._internal)>
+{
+    typedef iScalar<decltype(lhs._internal*rhs._internal)> ret_t;
+    ret_t ret;
+    mult(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class l,class r,int N> inline
+auto operator * (const iMatrix<l,N>& lhs,const iMatrix<r,N>& rhs) -> iMatrix<decltype(lhs._internal[0][0]*rhs._internal[0][0]),N>
+{
+    typedef decltype(lhs._internal[0][0]*rhs._internal[0][0]) ret_t;
+    iMatrix<ret_t,N> ret;
+    mult(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class l,class r, int N> inline
+auto operator * (const iMatrix<r,N>& lhs,const iScalar<l>& rhs) -> iMatrix<decltype(lhs._internal[0][0]*rhs._internal),N>
+{
+    typedef decltype(lhs._internal[0][0]*rhs._internal) ret_t;
+        
+    iMatrix<ret_t,N> ret;
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        mult(&ret._internal[c1][c2],&lhs._internal[c1][c2],&rhs._internal);
+    }}
+    return ret;
+}
+template<class l,class r,int N> inline
+auto operator * (const iScalar<l>& lhs,const iMatrix<r,N>& rhs) -> iMatrix<decltype(lhs._internal*rhs._internal[0][0]),N>
+{
+    typedef decltype(lhs._internal*rhs._internal[0][0]) ret_t;
+    iMatrix<ret_t,N> ret;
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        mult(&ret._internal[c1][c2],&lhs._internal,&rhs._internal[c1][c2]);
+    }}
+    return ret;
+}
+template<class l,class r,int N> inline
+auto operator * (const iMatrix<l,N>& lhs,const iVector<r,N>& rhs) -> iVector<decltype(lhs._internal[0][0]*rhs._internal[0]),N>
+{
+    typedef decltype(lhs._internal[0][0]*rhs._internal[0]) ret_t;
+    iVector<ret_t,N> ret;
+    for(int c1=0;c1<N;c1++){
+        mult(&ret._internal[c1],&lhs._internal[c1][0],&rhs._internal[0]);
+        for(int c2=1;c2<N;c2++){
+            mac(&ret._internal[c1],&lhs._internal[c1][c2],&rhs._internal[c2]);
+        }
+    }
+    return ret;
+}
+template<class l,class r,int N> inline
+auto operator * (const iScalar<l>& lhs,const iVector<r,N>& rhs) -> iVector<decltype(lhs._internal*rhs._internal[0]),N>
+{
+    typedef decltype(lhs._internal*rhs._internal[0]) ret_t;
+    iVector<ret_t,N> ret;
+    for(int c1=0;c1<N;c1++){
+        mult(&ret._internal[c1],&lhs._internal,&rhs._internal[c1]);
+    }
+    return ret;
+}
+template<class l,class r,int N> inline
+auto operator * (const iVector<l,N>& lhs,const iScalar<r>& rhs) -> iVector<decltype(lhs._internal[0]*rhs._internal),N>
+{
+    typedef decltype(lhs._internal[0]*rhs._internal) ret_t;
+    iVector<ret_t,N> ret;
+    for(int c1=0;c1<N;c1++){
+        mult(&ret._internal[c1],&lhs._internal[c1],&rhs._internal);
+    }
+    return ret;
+}
+    ///////////////////////////////////////////////////////////////////////////////////////
+    // localInnerProduct Scalar x Scalar -> Scalar
+    // localInnerProduct Vector x Vector -> Scalar
+    // localInnerProduct Matrix x Matrix -> Scalar
+    ///////////////////////////////////////////////////////////////////////////////////////
+    template<class l,class r,int N> inline
+    auto localInnerProduct (const iVector<l,N>& lhs,const iVector<r,N>& rhs) -> iScalar<decltype(localInnerProduct(lhs._internal[0],rhs._internal[0]))>
+    {
+        typedef decltype(localInnerProduct(lhs._internal[0],rhs._internal[0])) ret_t;
+        iScalar<ret_t> ret=zero;
+        for(int c1=0;c1<N;c1++){
+            ret._internal += localInnerProduct(lhs._internal[c1],rhs._internal[c1]);
+        }
+        return ret;
+    }
+    template<class l,class r,int N> inline
+    auto localInnerProduct (const iMatrix<l,N>& lhs,const iMatrix<r,N>& rhs) -> iScalar<decltype(localInnerProduct(lhs._internal[0][0],rhs._internal[0][0]))>
+    {
+        typedef decltype(localInnerProduct(lhs._internal[0][0],rhs._internal[0][0])) ret_t;
+        iScalar<ret_t> ret=zero;
+        for(int c1=0;c1<N;c1++){
+        for(int c2=0;c2<N;c2++){
+            ret._internal += localInnerProduct(lhs._internal[c1][c2],rhs._internal[c1][c2]);
+        }}
+        return ret;
+    }
+    template<class l,class r> inline
+    auto localInnerProduct (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(localInnerProduct(lhs._internal,rhs._internal))>
+    {
+        typedef decltype(localInnerProduct(lhs._internal,rhs._internal)) ret_t;
+        iScalar<ret_t> ret;
+        ret._internal = localInnerProduct(lhs._internal,rhs._internal);
+        return ret;
+    }
+
+    ///////////////////////////////////////////////////////////////////////////////////////
+    // outerProduct Scalar x Scalar -> Scalar
+    //              Vector x Vector -> Matrix
+    ///////////////////////////////////////////////////////////////////////////////////////
+
+template<class l,class r,int N> inline
+auto outerProduct (const iVector<l,N>& lhs,const iVector<r,N>& rhs) -> iMatrix<decltype(outerProduct(lhs._internal[0],rhs._internal[0])),N>
+{
+    typedef decltype(outerProduct(lhs._internal[0],rhs._internal[0])) ret_t;
+    iMatrix<ret_t,N> ret;
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        ret._internal[c1][c2] = outerProduct(lhs._internal[c1],rhs._internal[c2]);
+    }}
+    return ret;
+}
+template<class l,class r> inline
+auto outerProduct (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(outerProduct(lhs._internal,rhs._internal))>
+{
+    typedef decltype(outerProduct(lhs._internal,rhs._internal)) ret_t;
+    iScalar<ret_t> ret;
+    ret._internal = outerProduct(lhs._internal,rhs._internal);
+    return ret;
+}
+
+inline ComplexF outerProduct(const ComplexF &l, const ComplexF& r)
+{
+  return l*r;
+}
+inline ComplexD outerProduct(const ComplexD &l, const ComplexD& r)
+{
+  return l*r;
+}
+inline RealF outerProduct(const RealF &l, const RealF& r)
+{
+  return l*r;
+}
+inline RealD outerProduct(const RealD &l, const RealD& r)
+{
+  return l*r;
+}
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    /////////////////////////////////////////// CONJ         ///////////////////////////////////////////
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+ 
+// Conj function for scalar, vector, matrix
+template<class vtype> inline iScalar<vtype> conj(const iScalar<vtype>&r)
+{
+    iScalar<vtype> ret;
+    ret._internal = conj(r._internal);
+    return ret;
+}
+
+// Adj function for scalar, vector, matrix
+template<class vtype> inline iScalar<vtype> adj(const iScalar<vtype>&r)
+{
+    iScalar<vtype> ret;
+    ret._internal = adj(r._internal);
+    return ret;
+}
+template<class vtype,int N> inline iVector<vtype,N> adj(const iVector<vtype,N>&r)
+{
+    iVector<vtype,N> ret;
+    for(int i=0;i<N;i++){
+        ret._internal[i] = adj(r._internal[i]);
+    }
+    return ret;
+}
+template<class vtype,int N> inline iMatrix<vtype,N> adj(const iMatrix<vtype,N> &arg)
+{
+    iMatrix<vtype,N> ret;
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        ret._internal[c1][c2]=adj(arg._internal[c2][c1]);
+    }}
+    return ret;
+}
+
+/////////////////////////////////////////////////////////////////
+// Can only take the real/imag part of scalar objects, since
+// lattice objects of different complexity are non-conformable.
+/////////////////////////////////////////////////////////////////
+template<class itype> inline auto real(const iScalar<itype> &z) -> iScalar<decltype(real(z._internal))>
+{
+    iScalar<decltype(real(z._internal))> ret;
+    ret._internal = real(z._internal);
+    return ret;
+}
+template<class itype,int N> inline auto real(const iMatrix<itype,N> &z) -> iMatrix<decltype(real(z._internal[0][0])),N>
+{
+    iMatrix<decltype(real(z._internal[0][0])),N> ret;
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        ret._internal[c1][c2] = real(z._internal[c1][c2]);
+    }}
+    return ret;
+}
+template<class itype,int N> inline auto real(const iVector<itype,N> &z) -> iVector<decltype(real(z._internal[0])),N>
+{
+    iVector<decltype(real(z._internal[0])),N> ret;
+    for(int c1=0;c1<N;c1++){
+        ret._internal[c1] = real(z._internal[c1]);
+    }
+    return ret;
+}
+    
+template<class itype> inline auto imag(const iScalar<itype> &z) -> iScalar<decltype(imag(z._internal))>
+{
+    iScalar<decltype(imag(z._internal))> ret;
+    ret._internal = imag(z._internal);
+    return ret;
+}
+template<class itype,int N> inline auto imag(const iMatrix<itype,N> &z) -> iMatrix<decltype(imag(z._internal[0][0])),N>
+{
+    iMatrix<decltype(imag(z._internal[0][0])),N> ret;
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        ret._internal[c1][c2] = imag(z._internal[c1][c2]);
+    }}
+    return ret;
+}
+template<class itype,int N> inline auto imag(const iVector<itype,N> &z) -> iVector<decltype(imag(z._internal[0])),N>
+{
+    iVector<decltype(imag(z._internal[0])),N> ret;
+    for(int c1=0;c1<N;c1++){
+        ret._internal[c1] = imag(z._internal[c1]);
+    }
+    return ret;
+}
+
+    /////////////////////////////////
+    // Trace of scalar and matrix
+    /////////////////////////////////
+
+inline Complex trace( const Complex &arg){
+    return arg;
+}
+//inline vComplex trace(const vComplex &arg){
+//    return arg;
+//}
+template<class vtype,int N>
+inline auto trace(const iMatrix<vtype,N> &arg) -> iScalar<decltype(trace(arg._internal[0][0]))>
+{
+    iScalar<decltype( trace(arg._internal[0][0] )) > ret;
+    ZeroIt(ret._internal);
+    for(int i=0;i<N;i++){
+        ret._internal=ret._internal+trace(arg._internal[i][i]);
+    }
+    return ret;
+}
+template<class vtype>
+inline auto trace(const iScalar<vtype> &arg) -> iScalar<decltype(trace(arg._internal))>
+{
+    iScalar<decltype(trace(arg._internal))> ret;
+    ret._internal=trace(arg._internal);
+    return ret;
+}
+};
+/////////////////////////////////////////////////////////////////////////
+// Generic routine to promote object<complex> -> object<vcomplex>
+// Supports the array reordering transformation that gives me SIMD utilisation
+/////////////////////////////////////////////////////////////////////////
+/*
+template<template<class> class object>
+inline object<vComplex> splat(object<Complex >s){
+    object<vComplex> ret;
+    vComplex * v_ptr = (vComplex *)& ret;
+    Complex * s_ptr = (Complex *) &s;
+    for(int i=0;i<sizeof(ret);i+=sizeof(vComplex)){
+        vsplat(*(v_ptr++),*(s_ptr++));
+    }
+    return ret;
+}
+*/
+    
+#endif
diff --git a/Grid_signal.cc b/Grid_signal.cc
index 39676d30..e943c3b0 100755
--- a/Grid_signal.cc
+++ b/Grid_signal.cc
@@ -16,13 +16,15 @@
 #undef __X86_64
 namespace dpo {
 
-  void Grid_sa_signal_handler(int sig,siginfo_t *si,void * ptr);
-  void Grid_debug_handler_init(void);
-
-  void Grid_init(void)
-  {
-    Grid_debug_handler_init();
-  }
+void Grid_init(void)
+{
+  Grid_debug_handler_init();
+}
+double usecond(void) {
+  struct timeval tv;
+  gettimeofday(&tv,NULL);
+  return 1.0*tv.tv_usec + 1.0e6*tv.tv_sec;
+}
 
 void Grid_sa_signal_handler(int sig,siginfo_t *si,void * ptr)
 {
diff --git a/Grid_simd.h b/Grid_simd.h
index aa16286b..8c7f3a1c 100644
--- a/Grid_simd.h
+++ b/Grid_simd.h
@@ -10,6 +10,33 @@
 //
 // Vector types are arch dependent
 ////////////////////////////////////////////////////////////////////////
+    // TODO
+    //
+    // Base class to share common code between vRealF, VComplexF etc...
+    //
+    // lattice Broad cast assignment
+    //
+    // where() support
+    // implement with masks, and/or? Type of the mask & boolean support?
+    //
+    // Unary functions
+    // cos,sin, tan, acos, asin, cosh, acosh, tanh, sinh, // Scalar<vReal> only arg
+    // exp, log, sqrt, fabs
+    //
+    // transposeColor, transposeSpin,
+    // adjColor, adjSpin,
+    // traceColor, traceSpin.
+    // peekColor, peekSpin + pokeColor PokeSpin
+    //
+    // copyMask.
+    //
+    // localMaxAbs
+    //
+    // norm2,
+    // sumMulti equivalent.
+    // Fourier transform equivalent.
+    //
+    
 
 namespace dpo {
 
@@ -21,7 +48,46 @@ namespace dpo {
   typedef std::complex<RealD> ComplexD;
   typedef std::complex<Real>  Complex;
   
-  
+
+  inline RealF adj(const RealF  & r){ return r; }
+  inline RealF conj(const RealF  & r){ return r; }
+  inline ComplexD localInnerProduct(const ComplexD & l, const ComplexD & r) { return conj(l)*r; }
+  inline ComplexF localInnerProduct(const ComplexF & l, const ComplexF & r) { return conj(l)*r; }
+  inline RealD localInnerProduct(const RealD & l, const RealD & r) { return l*r; }
+  inline RealF localInnerProduct(const RealF & l, const RealF & r) { return l*r; }
+
+    ////////////////////////////////////////////////////////////////////////////////
+    //Provide support functions for basic real and complex data types required by dpo
+    //Single and double precision versions. Should be able to template this once only.
+    ////////////////////////////////////////////////////////////////////////////////
+    inline void mac (ComplexD * __restrict__ y,const ComplexD * __restrict__ a,const ComplexD *__restrict__ x){ *y = (*a) * (*x)+(*y); };
+    inline void mult(ComplexD * __restrict__ y,const ComplexD * __restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) * (*r);}
+    inline void sub (ComplexD * __restrict__ y,const ComplexD * __restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) - (*r);}
+    inline void add (ComplexD * __restrict__ y,const ComplexD * __restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) + (*r);}
+    inline ComplexD adj(const ComplexD& r){ return(conj(r)); }
+    // conj already supported for complex
+    
+    inline void mac (ComplexF * __restrict__ y,const ComplexF * __restrict__ a,const ComplexF *__restrict__ x){ *y = (*a) * (*x)+(*y); }
+    inline void mult(ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) * (*r); }
+    inline void sub (ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) - (*r); }
+    inline void add (ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) + (*r); }
+    inline Complex  adj(const Complex& r ){ return(conj(r)); }
+    //conj already supported for complex
+    
+    inline void mac (RealD * __restrict__ y,const RealD * __restrict__ a,const RealD *__restrict__ x){  *y = (*a) * (*x)+(*y);}
+    inline void mult(RealD * __restrict__ y,const RealD * __restrict__ l,const RealD *__restrict__ r){ *y = (*l) * (*r);}
+    inline void sub (RealD * __restrict__ y,const RealD * __restrict__ l,const RealD *__restrict__ r){ *y = (*l) - (*r);}
+    inline void add (RealD * __restrict__ y,const RealD * __restrict__ l,const RealD *__restrict__ r){ *y = (*l) + (*r);}
+    inline RealD adj(const RealD & r){ return r; }  // No-op for real
+    inline RealD conj(const RealD & r){ return r; }
+    
+    inline void mac (RealF * __restrict__ y,const RealF * __restrict__ a,const RealF *__restrict__ x){  *y = (*a) * (*x)+(*y); }
+    inline void mult(RealF * __restrict__ y,const RealF * __restrict__ l,const RealF *__restrict__ r){ *y = (*l) * (*r); }
+    inline void sub (RealF * __restrict__ y,const RealF * __restrict__ l,const RealF *__restrict__ r){ *y = (*l) - (*r); }
+    inline void add (RealF * __restrict__ y,const RealF * __restrict__ l,const RealF *__restrict__ r){ *y = (*l) + (*r); }
+    
+
+
   class Zero{};
   static Zero zero;
   template<class itype> inline void ZeroIt(itype &arg){ arg=zero;};

From b1cb3f255d552febe57befa6bb1e6304ad426def Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 4 Mar 2015 05:33:26 +0000
Subject: [PATCH 004/429] Update organisation

---
 Grid_Cartesian.h | 79 +++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 78 insertions(+), 1 deletion(-)

diff --git a/Grid_Cartesian.h b/Grid_Cartesian.h
index 779cf44f..03044bdd 100644
--- a/Grid_Cartesian.h
+++ b/Grid_Cartesian.h
@@ -1,7 +1,83 @@
+#ifndef GRID_CARTESIAN_H
+#define GRID_CARTESIAN_H
+
 #include "Grid.h"
-#include "Grid_vComplexD.h"
 
 namespace dpo{
+
+    
+/////////////////////////////////////////////////////////////////////////////////////////
+// Grid Support. Following will go into Grid.h.
+/////////////////////////////////////////////////////////////////////////////////////////
+    // Cartesian grids
+    // dpo::Grid
+    // dpo::GridCartesian
+    // dpo::GridCartesianRedBlack
+    
+class Grid {
+public:
+    // Give Lattice access
+    template<class object> friend class Lattice;
+
+        
+//protected:
+        
+    // Lattice wide random support. not yet fully implemented. Need seed strategy
+    // and one generator per site.
+    //std::default_random_engine generator;
+  //    static std::mt19937  generator( 9 );
+
+        
+    // Grid information.
+    unsigned long _ndimension;
+    std::vector<int> _layout;     // Which dimensions get relayed out over simd lanes.
+    std::vector<int> _dimensions; // Dimensions of array
+    std::vector<int> _rdimensions;// Reduced dimensions with simd lane images removed
+    std::vector<int> _ostride;    // Outer stride for each dimension
+    std::vector<int> _istride;    // Inner stride i.e. within simd lane
+    int _osites;                  // _isites*_osites = product(dimensions).
+    int _isites;
+        
+    // subslice information
+    std::vector<int> _slice_block;
+    std::vector<int> _slice_stride;
+    std::vector<int> _slice_nblock;
+public:
+    
+    // These routines are key. Subdivide the linearised cartesian index into
+    //      "inner" index identifying which simd lane of object<vFcomplex> is associated with coord
+    //      "outer" index identifying which element of _odata in class "Lattice" is associated with coord.
+    // Compared to, say, Blitz++ we simply need to store BOTH an inner stride and an outer
+    // stride per dimension. The cost of evaluating the indexing information is doubled for an n-dimensional
+    // coordinate. Note, however, for data parallel operations the "inner" indexing cost is not paid and all
+    // lanes are operated upon simultaneously.
+    
+    inline int oIndexReduced(std::vector<int> &rcoor)
+    {
+        int idx=0;
+        for(int d=0;d<_ndimension;d++) idx+=_ostride[d]*rcoor[d];
+        return idx;
+    }
+    virtual int oIndex(std::vector<int> &coor)
+    {
+        int idx=0;
+        for(int d=0;d<_ndimension;d++) idx+=_ostride[d]*(coor[d]%_rdimensions[d]);
+        return idx;
+    }
+    inline int iIndex(std::vector<int> &rcoor)
+    {
+        int idx=0;
+        for(int d=0;d<_ndimension;d++) idx+=_istride[d]*(rcoor[d]/_rdimensions[d]);
+        return idx;
+    }
+        
+    inline int oSites(void) { return _osites; };
+    inline int iSites(void) { return _isites; };
+    virtual int CheckerBoard(std::vector<int> site)=0;
+    virtual int CheckerBoardDestination(int source_cb,int shift)=0;
+    virtual int CheckerBoardShift(int source_cb,int dim,int shift)=0;
+};
+
 class GridCartesian: public Grid {
 public:
     virtual int CheckerBoard(std::vector<int> site){
@@ -176,3 +252,4 @@ protected:
 };
 
 }
+#endif

From 5aefa56e5c10dba004977864fce11890077912e3 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 4 Mar 2015 05:34:15 +0000
Subject: [PATCH 005/429] Better organisation

---
 Grid.h           |  2 +-
 Grid_config.h.in | 11 +++++++----
 2 files changed, 8 insertions(+), 5 deletions(-)

diff --git a/Grid.h b/Grid.h
index 042835ca..0cbe583b 100644
--- a/Grid.h
+++ b/Grid.h
@@ -42,7 +42,7 @@
 
 #include <Grid_simd.h>
 #include <Grid_math_types.h>
-#include <Grid_cartesian.h>
+#include <Grid_Cartesian.h>
 #include <Grid_aligned_allocator.h>
 #include <Grid_aligned_allocator.h>
 #include <Grid_Lattice.h>
diff --git a/Grid_config.h.in b/Grid_config.h.in
index 3fc06db0..88003589 100644
--- a/Grid_config.h.in
+++ b/Grid_config.h.in
@@ -1,7 +1,7 @@
 /* Grid_config.h.in.  Generated from configure.ac by autoheader.  */
 
 /* AVX */
-#undef AVX
+#undef AVX1
 
 /* AVX2 */
 #undef AVX2
@@ -48,6 +48,9 @@
 /* Define to 1 if the system has the `aligned' variable attribute */
 #undef HAVE_VAR_ATTRIBUTE_ALIGNED
 
+/* Name of package */
+#undef PACKAGE
+
 /* Define to the address where bug reports for this package should be sent. */
 #undef PACKAGE_BUGREPORT
 
@@ -66,15 +69,15 @@
 /* Define to the version of this package. */
 #undef PACKAGE_VERSION
 
-/* SSE */
-#undef SSE
-
 /* SSE2 */
 #undef SSE2
 
 /* Define to 1 if you have the ANSI C header files. */
 #undef STDC_HEADERS
 
+/* Version number of package */
+#undef VERSION
+
 /* Define for Solaris 2.5.1 so the uint32_t typedef from <sys/synch.h>,
    <pthread.h>, or <semaphore.h> is not used. If the typedef were allowed, the
    #define below would cause a syntax error. */

From eb15a8dacbb892efeab4842e5158624aaaba1f19 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@c024008.wlan.net.ed.ac.uk>
Date: Wed, 4 Mar 2015 11:38:10 +0000
Subject: [PATCH 006/429] AVX2 fix

---
 Grid_Cartesian.h | 3 +--
 Grid_config.h    | 4 ++--
 Grid_main.cc     | 9 ++++++++-
 3 files changed, 11 insertions(+), 5 deletions(-)

diff --git a/Grid_Cartesian.h b/Grid_Cartesian.h
index 03044bdd..488d47e9 100644
--- a/Grid_Cartesian.h
+++ b/Grid_Cartesian.h
@@ -108,9 +108,8 @@ public:
         for(int d=0;d<_ndimension;d++){
             _dimensions[d] = dimensions[d];
             _layout[d]     = layout[d];
-                
             // Use a reduced simd grid
-            _rdimensions[d]= _dimensions[d]/_layout[d];
+            _rdimensions[d]= _dimensions[d]/_layout[d]; //<-- _layout[d] is zero
             _osites *= _rdimensions[d];
             _isites *= _layout[d];
                 
diff --git a/Grid_config.h b/Grid_config.h
index b882917d..b3833326 100644
--- a/Grid_config.h
+++ b/Grid_config.h
@@ -2,10 +2,10 @@
 /* Grid_config.h.in.  Generated from configure.ac by autoheader.  */
 
 /* AVX */
-#define AVX1 1
+/* #undef AVX1 */
 
 /* AVX2 */
-/* #undef AVX2 */
+#define AVX2 1
 
 /* AVX512 */
 /* #undef AVX512 */
diff --git a/Grid_main.cc b/Grid_main.cc
index a3ef32c8..33dfc37b 100644
--- a/Grid_main.cc
+++ b/Grid_main.cc
@@ -46,12 +46,19 @@ int main (int argc, char ** argv)
     simd_layout[2] = 2;
     simd_layout[3] = 2;
 #endif
-#ifdef AVX1
+#if defined (AVX1)|| defined (AVX2)
     simd_layout[0] = 1;
     simd_layout[1] = 1;
     simd_layout[2] = 2;
     simd_layout[3] = 2;
 #endif
+#if defined (SSE2)
+    simd_layout[0] = 1;
+    simd_layout[1] = 1;
+    simd_layout[2] = 1;
+    simd_layout[3] = 2;
+#endif
+
     
     GridCartesian Fine(latt_size,simd_layout);
     GridRedBlackCartesian rbFine(latt_size,simd_layout);

From 09582509e50079caa2b746c165e4b80b704a9d47 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 4 Mar 2015 11:44:05 +0000
Subject: [PATCH 007/429] Improving benchmark to include Cshift

---
 Grid_main.cc | 18 ++++++++----------
 1 file changed, 8 insertions(+), 10 deletions(-)

diff --git a/Grid_main.cc b/Grid_main.cc
index a3ef32c8..4727470b 100644
--- a/Grid_main.cc
+++ b/Grid_main.cc
@@ -182,21 +182,19 @@ int main (int argc, char ** argv)
     printf("mult NumThread %d , Lattice size %d , %f Mflop/s\n",omp,lat,flops/(t1-t0));
     printf("mult NumThread %d , Lattice size %d , %f MB/s\n",omp,lat,bytes/(t1-t0));
 
-/*
-        mult(FooBar,Foo,Bar);
-	//        FooBar = Foo * Bar;
+    mult(FooBar,Foo,Bar);
+    FooBar = Foo * Bar;
     t0=usecond();
     for(int i=0;i<ncall;i++){
-      //      mult(FooBar,Foo,Cshift(Bar,1,-2));
-      mult(FooBar,Foo,Bar);
-	//        FooBar = Foo * Bar; // this is bad
+      mult(FooBar,Foo,Cshift(Bar,1,-2));
+      //mult(FooBar,Foo,Bar);
+      //FooBar = Foo * Bar; // this is bad
     }
     t1=usecond();
     
-    printf("A: NumThread %d , Lattice size %d , %f us per call\n",omp,lat,(t1-t0)/ncall);
-    printf("A: NumThread %d , Lattice size %d , %f Mflop/s\n",omp,lat,flops/(t1-t0));
-    printf("A: NumThread %d , Lattice size %d , %f MB/s\n",omp,lat,bytes/(t1-t0));
-*/
+    printf("Cshift Mult: NumThread %d , Lattice size %d , %f us per call\n",omp,lat,(t1-t0)/ncall);
+    printf("Cshift Mult: NumThread %d , Lattice size %d , %f Mflop/s\n",omp,lat,flops/(t1-t0));
+    printf("Cshift Mult: NumThread %d , Lattice size %d , %f MB/s\n",omp,lat,bytes/(t1-t0));
 
     pickCheckerboard(0,rFoo,FooBar);
     pickCheckerboard(1,bFoo,FooBar);

From d9cff588c565102b74f4dc0996c08fa33e45a7e0 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 4 Mar 2015 11:50:59 +0000
Subject: [PATCH 008/429] Better openMP for Cshift

---
 Grid_Lattice.h | 91 +++++++-------------------------------------------
 Grid_config.h  |  4 +--
 2 files changed, 14 insertions(+), 81 deletions(-)

diff --git a/Grid_Lattice.h b/Grid_Lattice.h
index 47b85b0f..d0fb9465 100644
--- a/Grid_Lattice.h
+++ b/Grid_Lattice.h
@@ -68,7 +68,8 @@ public:
         }
         return ret;
     };
-#if 0
+
+
     friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
     {
         Lattice<vobj> ret(rhs._grid);
@@ -131,20 +132,24 @@ public:
             }
 
             if ( permute_slice ) {
-                
-                int internal=sizeof(vobj)/sizeof(vComplex);
+
+	        int internal=sizeof(vobj)/sizeof(vComplex);
                 int num =rhs._grid->_slice_block[dimension]*internal;
-                
+
+#pragma omp parallel for collapse(2)
                 for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-                    vComplex *optr = (vComplex *)&ret._odata[o];
-                    vComplex *iptr = (vComplex *)&rhs._odata[so];
                     for(int b=0;b<num;b++){
-                        permute(optr[b],iptr[b],permute_type);
+		      vComplex *optr = (vComplex *)&ret._odata[o];
+		      vComplex *iptr = (vComplex *)&rhs._odata[so];
+                      permute(optr[b],iptr[b],permute_type);
                     }
                     o+=rhs._grid->_slice_stride[dimension];
                     so+=rhs._grid->_slice_stride[dimension];
                 }
+
             } else {
+
+#pragma omp parallel for collapse(2)
                 for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
                     for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
                         ret._odata[o+i]=rhs._odata[so+i];
@@ -157,78 +162,6 @@ public:
         }
         return ret;
     }
-#else
-    friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
-    {
-        Lattice<vobj> ret(rhs._grid);
-	int sx,so,o;
-        
-        ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
-        shift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift);
-        int rd = rhs._grid->_rdimensions[dimension];
-        int ld = rhs._grid->_dimensions[dimension];
-        
-        // Map to always positive shift.
-        shift = (shift+ld)%ld;
-        
-        // Work out whether to permute and the permute type
-        // ABCDEFGH ->   AE BF CG DH       permute
-        // Shift 0       AE BF CG DH       0 0 0 0    ABCDEFGH
-        // Shift 1       DH AE BF CG       1 0 0 0    HABCDEFG
-        // Shift 2       CG DH AE BF       1 1 0 0    GHABCDEF
-        // Shift 3       BF CG DH AE       1 1 1 0    FGHACBDE
-        // Shift 4       AE BF CG DH       1 1 1 1    EFGHABCD
-        // Shift 5       DH AE BF CG       0 1 1 1    DEFGHABC
-        // Shift 6       CG DH AE BF       0 0 1 1    CDEFGHAB
-        // Shift 7       BF CG DH AE       0 0 0 1    BCDEFGHA
-        int permute_dim =rhs._grid->_layout[dimension]>1 ;
-        int permute_type=0;
-        for(int d=0;d<dimension;d++)
-            if (rhs._grid->_layout[d]>1 ) permute_type++;
-        
-        
-        // loop over all work
-        int work =rd*rhs._grid->_slice_nblock[dimension]*rhs._grid->_slice_block[dimension];
-//#pragma omp parallel for
-        for(int ww=0;ww<work;ww++){
-
-            
-            // can optimise this if know w moves congtiguously for a given thread.
-            // b=(b+1);
-            // if (b==_slice_block) {b=0; n=n+1;}
-            // if (n==_slice_nblock) { n=0; x=x+1}
-            //
-            // Perhaps a five cycle iterator, or so.
-            int w=ww;
-            int b = w%rhs._grid->_slice_block[dimension] ; w=w/rhs._grid->_slice_block[dimension];
-            int n = w%rhs._grid->_slice_nblock[dimension]; w=w/rhs._grid->_slice_nblock[dimension];
-            int x = w;
-
-            sx = (x-shift+ld)%rd;
-            o  = x*rhs._grid->_ostride[dimension]+n*rhs._grid->_slice_stride[dimension]; // common sub expression alert.
-            so =sx*rhs._grid->_ostride[dimension]+n*rhs._grid->_slice_stride[dimension];
-            
-            int permute_slice=0;
-            if ( permute_dim ) {
-                permute_slice = shift/rd;
-                if ( x<shift%rd ) permute_slice = 1-permute_slice;
-            }
-            
-            if ( permute_slice ) {
-                
-                int internal=sizeof(vobj)/sizeof(vComplex);
-                vComplex *optr = (vComplex *)&ret._odata[o+b];
-                vComplex *iptr = (vComplex *)&rhs._odata[so+b];
-                for(int i=0;i<internal;i++){
-                    permute(optr[i],iptr[i],permute_type);
-                }
-            } else {
-                ret._odata[o+b]=rhs._odata[so+b];
-            }
-        }
-        return ret;
-    }
-#endif
 
     template<class sobj>
     inline Lattice<vobj> & operator = (const sobj & r){
diff --git a/Grid_config.h b/Grid_config.h
index b3833326..b882917d 100644
--- a/Grid_config.h
+++ b/Grid_config.h
@@ -2,10 +2,10 @@
 /* Grid_config.h.in.  Generated from configure.ac by autoheader.  */
 
 /* AVX */
-/* #undef AVX1 */
+#define AVX1 1
 
 /* AVX2 */
-#define AVX2 1
+/* #undef AVX2 */
 
 /* AVX512 */
 /* #undef AVX512 */

From 523abad40fc743fafc3e1161401e1987095e2b05 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 4 Mar 2015 11:53:59 +0000
Subject: [PATCH 009/429] Place them in to avoid forced autoreconf on user

---
 Makefile.in |  885 +++++++
 configure   | 6545 +++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 7430 insertions(+)
 create mode 100644 Makefile.in
 create mode 100755 configure

diff --git a/Makefile.in b/Makefile.in
new file mode 100644
index 00000000..c1d09a38
--- /dev/null
+++ b/Makefile.in
@@ -0,0 +1,885 @@
+# Makefile.in generated by automake 1.15 from Makefile.am.
+# @configure_input@
+
+# Copyright (C) 1994-2014 Free Software Foundation, Inc.
+
+# This Makefile.in is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY, to the extent permitted by law; without
+# even the implied warranty of MERCHANTABILITY or FITNESS FOR A
+# PARTICULAR PURPOSE.
+
+@SET_MAKE@
+
+
+
+VPATH = @srcdir@
+am__is_gnu_make = { \
+  if test -z '$(MAKELEVEL)'; then \
+    false; \
+  elif test -n '$(MAKE_HOST)'; then \
+    true; \
+  elif test -n '$(MAKE_VERSION)' && test -n '$(CURDIR)'; then \
+    true; \
+  else \
+    false; \
+  fi; \
+}
+am__make_running_with_option = \
+  case $${target_option-} in \
+      ?) ;; \
+      *) echo "am__make_running_with_option: internal error: invalid" \
+              "target option '$${target_option-}' specified" >&2; \
+         exit 1;; \
+  esac; \
+  has_opt=no; \
+  sane_makeflags=$$MAKEFLAGS; \
+  if $(am__is_gnu_make); then \
+    sane_makeflags=$$MFLAGS; \
+  else \
+    case $$MAKEFLAGS in \
+      *\\[\ \	]*) \
+        bs=\\; \
+        sane_makeflags=`printf '%s\n' "$$MAKEFLAGS" \
+          | sed "s/$$bs$$bs[$$bs $$bs	]*//g"`;; \
+    esac; \
+  fi; \
+  skip_next=no; \
+  strip_trailopt () \
+  { \
+    flg=`printf '%s\n' "$$flg" | sed "s/$$1.*$$//"`; \
+  }; \
+  for flg in $$sane_makeflags; do \
+    test $$skip_next = yes && { skip_next=no; continue; }; \
+    case $$flg in \
+      *=*|--*) continue;; \
+        -*I) strip_trailopt 'I'; skip_next=yes;; \
+      -*I?*) strip_trailopt 'I';; \
+        -*O) strip_trailopt 'O'; skip_next=yes;; \
+      -*O?*) strip_trailopt 'O';; \
+        -*l) strip_trailopt 'l'; skip_next=yes;; \
+      -*l?*) strip_trailopt 'l';; \
+      -[dEDm]) skip_next=yes;; \
+      -[JT]) skip_next=yes;; \
+    esac; \
+    case $$flg in \
+      *$$target_option*) has_opt=yes; break;; \
+    esac; \
+  done; \
+  test $$has_opt = yes
+am__make_dryrun = (target_option=n; $(am__make_running_with_option))
+am__make_keepgoing = (target_option=k; $(am__make_running_with_option))
+pkgdatadir = $(datadir)/@PACKAGE@
+pkgincludedir = $(includedir)/@PACKAGE@
+pkglibdir = $(libdir)/@PACKAGE@
+pkglibexecdir = $(libexecdir)/@PACKAGE@
+am__cd = CDPATH="$${ZSH_VERSION+.}$(PATH_SEPARATOR)" && cd
+install_sh_DATA = $(install_sh) -c -m 644
+install_sh_PROGRAM = $(install_sh) -c
+install_sh_SCRIPT = $(install_sh) -c
+INSTALL_HEADER = $(INSTALL_DATA)
+transform = $(program_transform_name)
+NORMAL_INSTALL = :
+PRE_INSTALL = :
+POST_INSTALL = :
+NORMAL_UNINSTALL = :
+PRE_UNINSTALL = :
+POST_UNINSTALL = :
+bin_PROGRAMS = Grid_main$(EXEEXT)
+subdir = .
+ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
+am__aclocal_m4_deps = $(top_srcdir)/m4/ax_gcc_var_attribute.m4 \
+	$(top_srcdir)/configure.ac
+am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
+	$(ACLOCAL_M4)
+DIST_COMMON = $(srcdir)/Makefile.am $(top_srcdir)/configure \
+	$(am__configure_deps) $(include_HEADERS) $(am__DIST_COMMON)
+am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \
+ configure.lineno config.status.lineno
+mkinstalldirs = $(install_sh) -d
+CONFIG_HEADER = Grid_config.h
+CONFIG_CLEAN_FILES =
+CONFIG_CLEAN_VPATH_FILES =
+am__vpath_adj_setup = srcdirstrip=`echo "$(srcdir)" | sed 's|.|.|g'`;
+am__vpath_adj = case $$p in \
+    $(srcdir)/*) f=`echo "$$p" | sed "s|^$$srcdirstrip/||"`;; \
+    *) f=$$p;; \
+  esac;
+am__strip_dir = f=`echo $$p | sed -e 's|^.*/||'`;
+am__install_max = 40
+am__nobase_strip_setup = \
+  srcdirstrip=`echo "$(srcdir)" | sed 's/[].[^$$\\*|]/\\\\&/g'`
+am__nobase_strip = \
+  for p in $$list; do echo "$$p"; done | sed -e "s|$$srcdirstrip/||"
+am__nobase_list = $(am__nobase_strip_setup); \
+  for p in $$list; do echo "$$p $$p"; done | \
+  sed "s| $$srcdirstrip/| |;"' / .*\//!s/ .*/ ./; s,\( .*\)/[^/]*$$,\1,' | \
+  $(AWK) 'BEGIN { files["."] = "" } { files[$$2] = files[$$2] " " $$1; \
+    if (++n[$$2] == $(am__install_max)) \
+      { print $$2, files[$$2]; n[$$2] = 0; files[$$2] = "" } } \
+    END { for (dir in files) print dir, files[dir] }'
+am__base_list = \
+  sed '$$!N;$$!N;$$!N;$$!N;$$!N;$$!N;$$!N;s/\n/ /g' | \
+  sed '$$!N;$$!N;$$!N;$$!N;s/\n/ /g'
+am__uninstall_files_from_dir = { \
+  test -z "$$files" \
+    || { test ! -d "$$dir" && test ! -f "$$dir" && test ! -r "$$dir"; } \
+    || { echo " ( cd '$$dir' && rm -f" $$files ")"; \
+         $(am__cd) "$$dir" && rm -f $$files; }; \
+  }
+am__installdirs = "$(DESTDIR)$(libdir)" "$(DESTDIR)$(bindir)" \
+	"$(DESTDIR)$(includedir)"
+LIBRARIES = $(lib_LIBRARIES)
+AR = ar
+ARFLAGS = cru
+AM_V_AR = $(am__v_AR_@AM_V@)
+am__v_AR_ = $(am__v_AR_@AM_DEFAULT_V@)
+am__v_AR_0 = @echo "  AR      " $@;
+am__v_AR_1 = 
+libGrid_a_AR = $(AR) $(ARFLAGS)
+libGrid_a_LIBADD =
+am_libGrid_a_OBJECTS = Grid_signal.$(OBJEXT)
+libGrid_a_OBJECTS = $(am_libGrid_a_OBJECTS)
+PROGRAMS = $(bin_PROGRAMS)
+am_Grid_main_OBJECTS = Grid_main.$(OBJEXT)
+Grid_main_OBJECTS = $(am_Grid_main_OBJECTS)
+Grid_main_DEPENDENCIES = libGrid.a
+AM_V_P = $(am__v_P_@AM_V@)
+am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
+am__v_P_0 = false
+am__v_P_1 = :
+AM_V_GEN = $(am__v_GEN_@AM_V@)
+am__v_GEN_ = $(am__v_GEN_@AM_DEFAULT_V@)
+am__v_GEN_0 = @echo "  GEN     " $@;
+am__v_GEN_1 = 
+AM_V_at = $(am__v_at_@AM_V@)
+am__v_at_ = $(am__v_at_@AM_DEFAULT_V@)
+am__v_at_0 = @
+am__v_at_1 = 
+DEFAULT_INCLUDES = -I.@am__isrc@
+depcomp = $(SHELL) $(top_srcdir)/depcomp
+am__depfiles_maybe = depfiles
+am__mv = mv -f
+CXXCOMPILE = $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) \
+	$(AM_CPPFLAGS) $(CPPFLAGS) $(AM_CXXFLAGS) $(CXXFLAGS)
+AM_V_CXX = $(am__v_CXX_@AM_V@)
+am__v_CXX_ = $(am__v_CXX_@AM_DEFAULT_V@)
+am__v_CXX_0 = @echo "  CXX     " $@;
+am__v_CXX_1 = 
+CXXLD = $(CXX)
+CXXLINK = $(CXXLD) $(AM_CXXFLAGS) $(CXXFLAGS) $(AM_LDFLAGS) $(LDFLAGS) \
+	-o $@
+AM_V_CXXLD = $(am__v_CXXLD_@AM_V@)
+am__v_CXXLD_ = $(am__v_CXXLD_@AM_DEFAULT_V@)
+am__v_CXXLD_0 = @echo "  CXXLD   " $@;
+am__v_CXXLD_1 = 
+SOURCES = $(libGrid_a_SOURCES) $(Grid_main_SOURCES)
+DIST_SOURCES = $(libGrid_a_SOURCES) $(Grid_main_SOURCES)
+am__can_run_installinfo = \
+  case $$AM_UPDATE_INFO_DIR in \
+    n|no|NO) false;; \
+    *) (install-info --version) >/dev/null 2>&1;; \
+  esac
+HEADERS = $(include_HEADERS)
+am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) \
+	$(LISP)Grid_config.h.in
+# Read a list of newline-separated strings from the standard input,
+# and print each of them once, without duplicates.  Input order is
+# *not* preserved.
+am__uniquify_input = $(AWK) '\
+  BEGIN { nonempty = 0; } \
+  { items[$$0] = 1; nonempty = 1; } \
+  END { if (nonempty) { for (i in items) print i; }; } \
+'
+# Make sure the list of sources is unique.  This is necessary because,
+# e.g., the same source file might be shared among _SOURCES variables
+# for different programs/libraries.
+am__define_uniq_tagged_files = \
+  list='$(am__tagged_files)'; \
+  unique=`for i in $$list; do \
+    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+  done | $(am__uniquify_input)`
+ETAGS = etags
+CTAGS = ctags
+CSCOPE = cscope
+AM_RECURSIVE_TARGETS = cscope
+am__DIST_COMMON = $(srcdir)/Grid_config.h.in $(srcdir)/Makefile.in \
+	AUTHORS COPYING ChangeLog INSTALL NEWS README compile depcomp \
+	install-sh missing
+DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
+distdir = $(PACKAGE)-$(VERSION)
+top_distdir = $(distdir)
+am__remove_distdir = \
+  if test -d "$(distdir)"; then \
+    find "$(distdir)" -type d ! -perm -200 -exec chmod u+w {} ';' \
+      && rm -rf "$(distdir)" \
+      || { sleep 5 && rm -rf "$(distdir)"; }; \
+  else :; fi
+am__post_remove_distdir = $(am__remove_distdir)
+DIST_ARCHIVES = $(distdir).tar.gz
+GZIP_ENV = --best
+DIST_TARGETS = dist-gzip
+distuninstallcheck_listfiles = find . -type f -print
+am__distuninstallcheck_listfiles = $(distuninstallcheck_listfiles) \
+  | sed 's|^\./|$(prefix)/|' | grep -v '$(infodir)/dir$$'
+distcleancheck_listfiles = find . -type f -print
+ACLOCAL = @ACLOCAL@
+AMTAR = @AMTAR@
+AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
+AUTOCONF = @AUTOCONF@
+AUTOHEADER = @AUTOHEADER@
+AUTOMAKE = @AUTOMAKE@
+AWK = @AWK@
+CC = @CC@
+CCDEPMODE = @CCDEPMODE@
+CFLAGS = @CFLAGS@
+CPP = @CPP@
+CPPFLAGS = @CPPFLAGS@
+CXX = @CXX@
+CXXDEPMODE = @CXXDEPMODE@
+CXXFLAGS = @CXXFLAGS@
+CYGPATH_W = @CYGPATH_W@
+DEFS = @DEFS@
+DEPDIR = @DEPDIR@
+ECHO_C = @ECHO_C@
+ECHO_N = @ECHO_N@
+ECHO_T = @ECHO_T@
+EGREP = @EGREP@
+EXEEXT = @EXEEXT@
+GREP = @GREP@
+INSTALL = @INSTALL@
+INSTALL_DATA = @INSTALL_DATA@
+INSTALL_PROGRAM = @INSTALL_PROGRAM@
+INSTALL_SCRIPT = @INSTALL_SCRIPT@
+INSTALL_STRIP_PROGRAM = @INSTALL_STRIP_PROGRAM@
+LDFLAGS = @LDFLAGS@
+LIBOBJS = @LIBOBJS@
+LIBS = @LIBS@
+LTLIBOBJS = @LTLIBOBJS@
+MAKEINFO = @MAKEINFO@
+MKDIR_P = @MKDIR_P@
+OBJEXT = @OBJEXT@
+OPENMP_CFLAGS = @OPENMP_CFLAGS@
+PACKAGE = @PACKAGE@
+PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
+PACKAGE_NAME = @PACKAGE_NAME@
+PACKAGE_STRING = @PACKAGE_STRING@
+PACKAGE_TARNAME = @PACKAGE_TARNAME@
+PACKAGE_URL = @PACKAGE_URL@
+PACKAGE_VERSION = @PACKAGE_VERSION@
+PATH_SEPARATOR = @PATH_SEPARATOR@
+RANLIB = @RANLIB@
+SET_MAKE = @SET_MAKE@
+SHELL = @SHELL@
+STRIP = @STRIP@
+VERSION = @VERSION@
+abs_builddir = @abs_builddir@
+abs_srcdir = @abs_srcdir@
+abs_top_builddir = @abs_top_builddir@
+abs_top_srcdir = @abs_top_srcdir@
+ac_ct_CC = @ac_ct_CC@
+ac_ct_CXX = @ac_ct_CXX@
+am__include = @am__include@
+am__leading_dot = @am__leading_dot@
+am__quote = @am__quote@
+am__tar = @am__tar@
+am__untar = @am__untar@
+bindir = @bindir@
+build_alias = @build_alias@
+builddir = @builddir@
+datadir = @datadir@
+datarootdir = @datarootdir@
+docdir = @docdir@
+dvidir = @dvidir@
+exec_prefix = @exec_prefix@
+host_alias = @host_alias@
+htmldir = @htmldir@
+includedir = @includedir@
+infodir = @infodir@
+install_sh = @install_sh@
+libdir = @libdir@
+libexecdir = @libexecdir@
+localedir = @localedir@
+localstatedir = @localstatedir@
+mandir = @mandir@
+mkdir_p = @mkdir_p@
+oldincludedir = @oldincludedir@
+pdfdir = @pdfdir@
+prefix = @prefix@
+program_transform_name = @program_transform_name@
+psdir = @psdir@
+sbindir = @sbindir@
+sharedstatedir = @sharedstatedir@
+srcdir = @srcdir@
+sysconfdir = @sysconfdir@
+target_alias = @target_alias@
+top_build_prefix = @top_build_prefix@
+top_builddir = @top_builddir@
+top_srcdir = @top_srcdir@
+
+# additional include paths necessary to compile the C++ library
+AM_CXXFLAGS = -I$(top_srcdir)/
+
+#
+# Libraries
+#
+lib_LIBRARIES = libGrid.a
+libGrid_a_SOURCES = Grid_signal.cc
+
+#
+# Include files
+#
+include_HEADERS = Grid.h\
+	Grid_vComplexD.h\
+	Grid_vComplexF.h\
+	Grid_vRealD.h\
+	Grid_vRealF.h\
+	Grid_Cartesian.h\
+	Grid_Lattice.h\
+	Grid_config.h
+
+Grid_main_SOURCES = Grid_main.cc
+Grid_main_LDADD = libGrid.a
+all: Grid_config.h
+	$(MAKE) $(AM_MAKEFLAGS) all-am
+
+.SUFFIXES:
+.SUFFIXES: .cc .o .obj
+am--refresh: Makefile
+	@:
+$(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
+	@for dep in $?; do \
+	  case '$(am__configure_deps)' in \
+	    *$$dep*) \
+	      echo ' cd $(srcdir) && $(AUTOMAKE) --gnu'; \
+	      $(am__cd) $(srcdir) && $(AUTOMAKE) --gnu \
+		&& exit 0; \
+	      exit 1;; \
+	  esac; \
+	done; \
+	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu Makefile'; \
+	$(am__cd) $(top_srcdir) && \
+	  $(AUTOMAKE) --gnu Makefile
+Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
+	@case '$?' in \
+	  *config.status*) \
+	    echo ' $(SHELL) ./config.status'; \
+	    $(SHELL) ./config.status;; \
+	  *) \
+	    echo ' cd $(top_builddir) && $(SHELL) ./config.status $@ $(am__depfiles_maybe)'; \
+	    cd $(top_builddir) && $(SHELL) ./config.status $@ $(am__depfiles_maybe);; \
+	esac;
+
+$(top_builddir)/config.status: $(top_srcdir)/configure $(CONFIG_STATUS_DEPENDENCIES)
+	$(SHELL) ./config.status --recheck
+
+$(top_srcdir)/configure:  $(am__configure_deps)
+	$(am__cd) $(srcdir) && $(AUTOCONF)
+$(ACLOCAL_M4):  $(am__aclocal_m4_deps)
+	$(am__cd) $(srcdir) && $(ACLOCAL) $(ACLOCAL_AMFLAGS)
+$(am__aclocal_m4_deps):
+
+Grid_config.h: stamp-h1
+	@test -f $@ || rm -f stamp-h1
+	@test -f $@ || $(MAKE) $(AM_MAKEFLAGS) stamp-h1
+
+stamp-h1: $(srcdir)/Grid_config.h.in $(top_builddir)/config.status
+	@rm -f stamp-h1
+	cd $(top_builddir) && $(SHELL) ./config.status Grid_config.h
+$(srcdir)/Grid_config.h.in:  $(am__configure_deps) 
+	($(am__cd) $(top_srcdir) && $(AUTOHEADER))
+	rm -f stamp-h1
+	touch $@
+
+distclean-hdr:
+	-rm -f Grid_config.h stamp-h1
+install-libLIBRARIES: $(lib_LIBRARIES)
+	@$(NORMAL_INSTALL)
+	@list='$(lib_LIBRARIES)'; test -n "$(libdir)" || list=; \
+	list2=; for p in $$list; do \
+	  if test -f $$p; then \
+	    list2="$$list2 $$p"; \
+	  else :; fi; \
+	done; \
+	test -z "$$list2" || { \
+	  echo " $(MKDIR_P) '$(DESTDIR)$(libdir)'"; \
+	  $(MKDIR_P) "$(DESTDIR)$(libdir)" || exit 1; \
+	  echo " $(INSTALL_DATA) $$list2 '$(DESTDIR)$(libdir)'"; \
+	  $(INSTALL_DATA) $$list2 "$(DESTDIR)$(libdir)" || exit $$?; }
+	@$(POST_INSTALL)
+	@list='$(lib_LIBRARIES)'; test -n "$(libdir)" || list=; \
+	for p in $$list; do \
+	  if test -f $$p; then \
+	    $(am__strip_dir) \
+	    echo " ( cd '$(DESTDIR)$(libdir)' && $(RANLIB) $$f )"; \
+	    ( cd "$(DESTDIR)$(libdir)" && $(RANLIB) $$f ) || exit $$?; \
+	  else :; fi; \
+	done
+
+uninstall-libLIBRARIES:
+	@$(NORMAL_UNINSTALL)
+	@list='$(lib_LIBRARIES)'; test -n "$(libdir)" || list=; \
+	files=`for p in $$list; do echo $$p; done | sed -e 's|^.*/||'`; \
+	dir='$(DESTDIR)$(libdir)'; $(am__uninstall_files_from_dir)
+
+clean-libLIBRARIES:
+	-test -z "$(lib_LIBRARIES)" || rm -f $(lib_LIBRARIES)
+
+libGrid.a: $(libGrid_a_OBJECTS) $(libGrid_a_DEPENDENCIES) $(EXTRA_libGrid_a_DEPENDENCIES) 
+	$(AM_V_at)-rm -f libGrid.a
+	$(AM_V_AR)$(libGrid_a_AR) libGrid.a $(libGrid_a_OBJECTS) $(libGrid_a_LIBADD)
+	$(AM_V_at)$(RANLIB) libGrid.a
+install-binPROGRAMS: $(bin_PROGRAMS)
+	@$(NORMAL_INSTALL)
+	@list='$(bin_PROGRAMS)'; test -n "$(bindir)" || list=; \
+	if test -n "$$list"; then \
+	  echo " $(MKDIR_P) '$(DESTDIR)$(bindir)'"; \
+	  $(MKDIR_P) "$(DESTDIR)$(bindir)" || exit 1; \
+	fi; \
+	for p in $$list; do echo "$$p $$p"; done | \
+	sed 's/$(EXEEXT)$$//' | \
+	while read p p1; do if test -f $$p \
+	  ; then echo "$$p"; echo "$$p"; else :; fi; \
+	done | \
+	sed -e 'p;s,.*/,,;n;h' \
+	    -e 's|.*|.|' \
+	    -e 'p;x;s,.*/,,;s/$(EXEEXT)$$//;$(transform);s/$$/$(EXEEXT)/' | \
+	sed 'N;N;N;s,\n, ,g' | \
+	$(AWK) 'BEGIN { files["."] = ""; dirs["."] = 1 } \
+	  { d=$$3; if (dirs[d] != 1) { print "d", d; dirs[d] = 1 } \
+	    if ($$2 == $$4) files[d] = files[d] " " $$1; \
+	    else { print "f", $$3 "/" $$4, $$1; } } \
+	  END { for (d in files) print "f", d, files[d] }' | \
+	while read type dir files; do \
+	    if test "$$dir" = .; then dir=; else dir=/$$dir; fi; \
+	    test -z "$$files" || { \
+	      echo " $(INSTALL_PROGRAM_ENV) $(INSTALL_PROGRAM) $$files '$(DESTDIR)$(bindir)$$dir'"; \
+	      $(INSTALL_PROGRAM_ENV) $(INSTALL_PROGRAM) $$files "$(DESTDIR)$(bindir)$$dir" || exit $$?; \
+	    } \
+	; done
+
+uninstall-binPROGRAMS:
+	@$(NORMAL_UNINSTALL)
+	@list='$(bin_PROGRAMS)'; test -n "$(bindir)" || list=; \
+	files=`for p in $$list; do echo "$$p"; done | \
+	  sed -e 'h;s,^.*/,,;s/$(EXEEXT)$$//;$(transform)' \
+	      -e 's/$$/$(EXEEXT)/' \
+	`; \
+	test -n "$$list" || exit 0; \
+	echo " ( cd '$(DESTDIR)$(bindir)' && rm -f" $$files ")"; \
+	cd "$(DESTDIR)$(bindir)" && rm -f $$files
+
+clean-binPROGRAMS:
+	-test -z "$(bin_PROGRAMS)" || rm -f $(bin_PROGRAMS)
+
+Grid_main$(EXEEXT): $(Grid_main_OBJECTS) $(Grid_main_DEPENDENCIES) $(EXTRA_Grid_main_DEPENDENCIES) 
+	@rm -f Grid_main$(EXEEXT)
+	$(AM_V_CXXLD)$(CXXLINK) $(Grid_main_OBJECTS) $(Grid_main_LDADD) $(LIBS)
+
+mostlyclean-compile:
+	-rm -f *.$(OBJEXT)
+
+distclean-compile:
+	-rm -f *.tab.c
+
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_main.Po@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_signal.Po@am__quote@
+
+.cc.o:
+@am__fastdepCXX_TRUE@	$(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
+@am__fastdepCXX_TRUE@	$(AM_V_at)$(am__mv) $(DEPDIR)/$*.Tpo $(DEPDIR)/$*.Po
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ $<
+
+.cc.obj:
+@am__fastdepCXX_TRUE@	$(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ `$(CYGPATH_W) '$<'`
+@am__fastdepCXX_TRUE@	$(AM_V_at)$(am__mv) $(DEPDIR)/$*.Tpo $(DEPDIR)/$*.Po
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ `$(CYGPATH_W) '$<'`
+install-includeHEADERS: $(include_HEADERS)
+	@$(NORMAL_INSTALL)
+	@list='$(include_HEADERS)'; test -n "$(includedir)" || list=; \
+	if test -n "$$list"; then \
+	  echo " $(MKDIR_P) '$(DESTDIR)$(includedir)'"; \
+	  $(MKDIR_P) "$(DESTDIR)$(includedir)" || exit 1; \
+	fi; \
+	for p in $$list; do \
+	  if test -f "$$p"; then d=; else d="$(srcdir)/"; fi; \
+	  echo "$$d$$p"; \
+	done | $(am__base_list) | \
+	while read files; do \
+	  echo " $(INSTALL_HEADER) $$files '$(DESTDIR)$(includedir)'"; \
+	  $(INSTALL_HEADER) $$files "$(DESTDIR)$(includedir)" || exit $$?; \
+	done
+
+uninstall-includeHEADERS:
+	@$(NORMAL_UNINSTALL)
+	@list='$(include_HEADERS)'; test -n "$(includedir)" || list=; \
+	files=`for p in $$list; do echo $$p; done | sed -e 's|^.*/||'`; \
+	dir='$(DESTDIR)$(includedir)'; $(am__uninstall_files_from_dir)
+
+ID: $(am__tagged_files)
+	$(am__define_uniq_tagged_files); mkid -fID $$unique
+tags: tags-am
+TAGS: tags
+
+tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
+	set x; \
+	here=`pwd`; \
+	$(am__define_uniq_tagged_files); \
+	shift; \
+	if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
+	  test -n "$$unique" || unique=$$empty_fix; \
+	  if test $$# -gt 0; then \
+	    $(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
+	      "$$@" $$unique; \
+	  else \
+	    $(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
+	      $$unique; \
+	  fi; \
+	fi
+ctags: ctags-am
+
+CTAGS: ctags
+ctags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
+	$(am__define_uniq_tagged_files); \
+	test -z "$(CTAGS_ARGS)$$unique" \
+	  || $(CTAGS) $(CTAGSFLAGS) $(AM_CTAGSFLAGS) $(CTAGS_ARGS) \
+	     $$unique
+
+GTAGS:
+	here=`$(am__cd) $(top_builddir) && pwd` \
+	  && $(am__cd) $(top_srcdir) \
+	  && gtags -i $(GTAGS_ARGS) "$$here"
+cscope: cscope.files
+	test ! -s cscope.files \
+	  || $(CSCOPE) -b -q $(AM_CSCOPEFLAGS) $(CSCOPEFLAGS) -i cscope.files $(CSCOPE_ARGS)
+clean-cscope:
+	-rm -f cscope.files
+cscope.files: clean-cscope cscopelist
+cscopelist: cscopelist-am
+
+cscopelist-am: $(am__tagged_files)
+	list='$(am__tagged_files)'; \
+	case "$(srcdir)" in \
+	  [\\/]* | ?:[\\/]*) sdir="$(srcdir)" ;; \
+	  *) sdir=$(subdir)/$(srcdir) ;; \
+	esac; \
+	for i in $$list; do \
+	  if test -f "$$i"; then \
+	    echo "$(subdir)/$$i"; \
+	  else \
+	    echo "$$sdir/$$i"; \
+	  fi; \
+	done >> $(top_builddir)/cscope.files
+
+distclean-tags:
+	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
+	-rm -f cscope.out cscope.in.out cscope.po.out cscope.files
+
+distdir: $(DISTFILES)
+	$(am__remove_distdir)
+	test -d "$(distdir)" || mkdir "$(distdir)"
+	@srcdirstrip=`echo "$(srcdir)" | sed 's/[].[^$$\\*]/\\\\&/g'`; \
+	topsrcdirstrip=`echo "$(top_srcdir)" | sed 's/[].[^$$\\*]/\\\\&/g'`; \
+	list='$(DISTFILES)'; \
+	  dist_files=`for file in $$list; do echo $$file; done | \
+	  sed -e "s|^$$srcdirstrip/||;t" \
+	      -e "s|^$$topsrcdirstrip/|$(top_builddir)/|;t"`; \
+	case $$dist_files in \
+	  */*) $(MKDIR_P) `echo "$$dist_files" | \
+			   sed '/\//!d;s|^|$(distdir)/|;s,/[^/]*$$,,' | \
+			   sort -u` ;; \
+	esac; \
+	for file in $$dist_files; do \
+	  if test -f $$file || test -d $$file; then d=.; else d=$(srcdir); fi; \
+	  if test -d $$d/$$file; then \
+	    dir=`echo "/$$file" | sed -e 's,/[^/]*$$,,'`; \
+	    if test -d "$(distdir)/$$file"; then \
+	      find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
+	    fi; \
+	    if test -d $(srcdir)/$$file && test $$d != $(srcdir); then \
+	      cp -fpR $(srcdir)/$$file "$(distdir)$$dir" || exit 1; \
+	      find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
+	    fi; \
+	    cp -fpR $$d/$$file "$(distdir)$$dir" || exit 1; \
+	  else \
+	    test -f "$(distdir)/$$file" \
+	    || cp -p $$d/$$file "$(distdir)/$$file" \
+	    || exit 1; \
+	  fi; \
+	done
+	-test -n "$(am__skip_mode_fix)" \
+	|| find "$(distdir)" -type d ! -perm -755 \
+		-exec chmod u+rwx,go+rx {} \; -o \
+	  ! -type d ! -perm -444 -links 1 -exec chmod a+r {} \; -o \
+	  ! -type d ! -perm -400 -exec chmod a+r {} \; -o \
+	  ! -type d ! -perm -444 -exec $(install_sh) -c -m a+r {} {} \; \
+	|| chmod -R a+r "$(distdir)"
+dist-gzip: distdir
+	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
+	$(am__post_remove_distdir)
+
+dist-bzip2: distdir
+	tardir=$(distdir) && $(am__tar) | BZIP2=$${BZIP2--9} bzip2 -c >$(distdir).tar.bz2
+	$(am__post_remove_distdir)
+
+dist-lzip: distdir
+	tardir=$(distdir) && $(am__tar) | lzip -c $${LZIP_OPT--9} >$(distdir).tar.lz
+	$(am__post_remove_distdir)
+
+dist-xz: distdir
+	tardir=$(distdir) && $(am__tar) | XZ_OPT=$${XZ_OPT--e} xz -c >$(distdir).tar.xz
+	$(am__post_remove_distdir)
+
+dist-tarZ: distdir
+	@echo WARNING: "Support for distribution archives compressed with" \
+		       "legacy program 'compress' is deprecated." >&2
+	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
+	tardir=$(distdir) && $(am__tar) | compress -c >$(distdir).tar.Z
+	$(am__post_remove_distdir)
+
+dist-shar: distdir
+	@echo WARNING: "Support for shar distribution archives is" \
+	               "deprecated." >&2
+	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
+	shar $(distdir) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).shar.gz
+	$(am__post_remove_distdir)
+
+dist-zip: distdir
+	-rm -f $(distdir).zip
+	zip -rq $(distdir).zip $(distdir)
+	$(am__post_remove_distdir)
+
+dist dist-all:
+	$(MAKE) $(AM_MAKEFLAGS) $(DIST_TARGETS) am__post_remove_distdir='@:'
+	$(am__post_remove_distdir)
+
+# This target untars the dist file and tries a VPATH configuration.  Then
+# it guarantees that the distribution is self-contained by making another
+# tarfile.
+distcheck: dist
+	case '$(DIST_ARCHIVES)' in \
+	*.tar.gz*) \
+	  GZIP=$(GZIP_ENV) gzip -dc $(distdir).tar.gz | $(am__untar) ;;\
+	*.tar.bz2*) \
+	  bzip2 -dc $(distdir).tar.bz2 | $(am__untar) ;;\
+	*.tar.lz*) \
+	  lzip -dc $(distdir).tar.lz | $(am__untar) ;;\
+	*.tar.xz*) \
+	  xz -dc $(distdir).tar.xz | $(am__untar) ;;\
+	*.tar.Z*) \
+	  uncompress -c $(distdir).tar.Z | $(am__untar) ;;\
+	*.shar.gz*) \
+	  GZIP=$(GZIP_ENV) gzip -dc $(distdir).shar.gz | unshar ;;\
+	*.zip*) \
+	  unzip $(distdir).zip ;;\
+	esac
+	chmod -R a-w $(distdir)
+	chmod u+w $(distdir)
+	mkdir $(distdir)/_build $(distdir)/_build/sub $(distdir)/_inst
+	chmod a-w $(distdir)
+	test -d $(distdir)/_build || exit 0; \
+	dc_install_base=`$(am__cd) $(distdir)/_inst && pwd | sed -e 's,^[^:\\/]:[\\/],/,'` \
+	  && dc_destdir="$${TMPDIR-/tmp}/am-dc-$$$$/" \
+	  && am__cwd=`pwd` \
+	  && $(am__cd) $(distdir)/_build/sub \
+	  && ../../configure \
+	    $(AM_DISTCHECK_CONFIGURE_FLAGS) \
+	    $(DISTCHECK_CONFIGURE_FLAGS) \
+	    --srcdir=../.. --prefix="$$dc_install_base" \
+	  && $(MAKE) $(AM_MAKEFLAGS) \
+	  && $(MAKE) $(AM_MAKEFLAGS) dvi \
+	  && $(MAKE) $(AM_MAKEFLAGS) check \
+	  && $(MAKE) $(AM_MAKEFLAGS) install \
+	  && $(MAKE) $(AM_MAKEFLAGS) installcheck \
+	  && $(MAKE) $(AM_MAKEFLAGS) uninstall \
+	  && $(MAKE) $(AM_MAKEFLAGS) distuninstallcheck_dir="$$dc_install_base" \
+	        distuninstallcheck \
+	  && chmod -R a-w "$$dc_install_base" \
+	  && ({ \
+	       (cd ../.. && umask 077 && mkdir "$$dc_destdir") \
+	       && $(MAKE) $(AM_MAKEFLAGS) DESTDIR="$$dc_destdir" install \
+	       && $(MAKE) $(AM_MAKEFLAGS) DESTDIR="$$dc_destdir" uninstall \
+	       && $(MAKE) $(AM_MAKEFLAGS) DESTDIR="$$dc_destdir" \
+	            distuninstallcheck_dir="$$dc_destdir" distuninstallcheck; \
+	      } || { rm -rf "$$dc_destdir"; exit 1; }) \
+	  && rm -rf "$$dc_destdir" \
+	  && $(MAKE) $(AM_MAKEFLAGS) dist \
+	  && rm -rf $(DIST_ARCHIVES) \
+	  && $(MAKE) $(AM_MAKEFLAGS) distcleancheck \
+	  && cd "$$am__cwd" \
+	  || exit 1
+	$(am__post_remove_distdir)
+	@(echo "$(distdir) archives ready for distribution: "; \
+	  list='$(DIST_ARCHIVES)'; for i in $$list; do echo $$i; done) | \
+	  sed -e 1h -e 1s/./=/g -e 1p -e 1x -e '$$p' -e '$$x'
+distuninstallcheck:
+	@test -n '$(distuninstallcheck_dir)' || { \
+	  echo 'ERROR: trying to run $@ with an empty' \
+	       '$$(distuninstallcheck_dir)' >&2; \
+	  exit 1; \
+	}; \
+	$(am__cd) '$(distuninstallcheck_dir)' || { \
+	  echo 'ERROR: cannot chdir into $(distuninstallcheck_dir)' >&2; \
+	  exit 1; \
+	}; \
+	test `$(am__distuninstallcheck_listfiles) | wc -l` -eq 0 \
+	   || { echo "ERROR: files left after uninstall:" ; \
+	        if test -n "$(DESTDIR)"; then \
+	          echo "  (check DESTDIR support)"; \
+	        fi ; \
+	        $(distuninstallcheck_listfiles) ; \
+	        exit 1; } >&2
+distcleancheck: distclean
+	@if test '$(srcdir)' = . ; then \
+	  echo "ERROR: distcleancheck can only run from a VPATH build" ; \
+	  exit 1 ; \
+	fi
+	@test `$(distcleancheck_listfiles) | wc -l` -eq 0 \
+	  || { echo "ERROR: files left in build directory after distclean:" ; \
+	       $(distcleancheck_listfiles) ; \
+	       exit 1; } >&2
+check-am: all-am
+check: check-am
+all-am: Makefile $(LIBRARIES) $(PROGRAMS) $(HEADERS) Grid_config.h
+installdirs:
+	for dir in "$(DESTDIR)$(libdir)" "$(DESTDIR)$(bindir)" "$(DESTDIR)$(includedir)"; do \
+	  test -z "$$dir" || $(MKDIR_P) "$$dir"; \
+	done
+install: install-am
+install-exec: install-exec-am
+install-data: install-data-am
+uninstall: uninstall-am
+
+install-am: all-am
+	@$(MAKE) $(AM_MAKEFLAGS) install-exec-am install-data-am
+
+installcheck: installcheck-am
+install-strip:
+	if test -z '$(STRIP)'; then \
+	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
+	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
+	      install; \
+	else \
+	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
+	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
+	    "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'" install; \
+	fi
+mostlyclean-generic:
+
+clean-generic:
+
+distclean-generic:
+	-test -z "$(CONFIG_CLEAN_FILES)" || rm -f $(CONFIG_CLEAN_FILES)
+	-test . = "$(srcdir)" || test -z "$(CONFIG_CLEAN_VPATH_FILES)" || rm -f $(CONFIG_CLEAN_VPATH_FILES)
+
+maintainer-clean-generic:
+	@echo "This command is intended for maintainers to use"
+	@echo "it deletes files that may require special tools to rebuild."
+clean: clean-am
+
+clean-am: clean-binPROGRAMS clean-generic clean-libLIBRARIES \
+	mostlyclean-am
+
+distclean: distclean-am
+	-rm -f $(am__CONFIG_DISTCLEAN_FILES)
+	-rm -rf ./$(DEPDIR)
+	-rm -f Makefile
+distclean-am: clean-am distclean-compile distclean-generic \
+	distclean-hdr distclean-tags
+
+dvi: dvi-am
+
+dvi-am:
+
+html: html-am
+
+html-am:
+
+info: info-am
+
+info-am:
+
+install-data-am: install-includeHEADERS
+
+install-dvi: install-dvi-am
+
+install-dvi-am:
+
+install-exec-am: install-binPROGRAMS install-libLIBRARIES
+
+install-html: install-html-am
+
+install-html-am:
+
+install-info: install-info-am
+
+install-info-am:
+
+install-man:
+
+install-pdf: install-pdf-am
+
+install-pdf-am:
+
+install-ps: install-ps-am
+
+install-ps-am:
+
+installcheck-am:
+
+maintainer-clean: maintainer-clean-am
+	-rm -f $(am__CONFIG_DISTCLEAN_FILES)
+	-rm -rf $(top_srcdir)/autom4te.cache
+	-rm -rf ./$(DEPDIR)
+	-rm -f Makefile
+maintainer-clean-am: distclean-am maintainer-clean-generic
+
+mostlyclean: mostlyclean-am
+
+mostlyclean-am: mostlyclean-compile mostlyclean-generic
+
+pdf: pdf-am
+
+pdf-am:
+
+ps: ps-am
+
+ps-am:
+
+uninstall-am: uninstall-binPROGRAMS uninstall-includeHEADERS \
+	uninstall-libLIBRARIES
+
+.MAKE: all install-am install-strip
+
+.PHONY: CTAGS GTAGS TAGS all all-am am--refresh check check-am clean \
+	clean-binPROGRAMS clean-cscope clean-generic \
+	clean-libLIBRARIES cscope cscopelist-am ctags ctags-am dist \
+	dist-all dist-bzip2 dist-gzip dist-lzip dist-shar dist-tarZ \
+	dist-xz dist-zip distcheck distclean distclean-compile \
+	distclean-generic distclean-hdr distclean-tags distcleancheck \
+	distdir distuninstallcheck dvi dvi-am html html-am info \
+	info-am install install-am install-binPROGRAMS install-data \
+	install-data-am install-dvi install-dvi-am install-exec \
+	install-exec-am install-html install-html-am \
+	install-includeHEADERS install-info install-info-am \
+	install-libLIBRARIES install-man install-pdf install-pdf-am \
+	install-ps install-ps-am install-strip installcheck \
+	installcheck-am installdirs maintainer-clean \
+	maintainer-clean-generic mostlyclean mostlyclean-compile \
+	mostlyclean-generic pdf pdf-am ps ps-am tags tags-am uninstall \
+	uninstall-am uninstall-binPROGRAMS uninstall-includeHEADERS \
+	uninstall-libLIBRARIES
+
+.PRECIOUS: Makefile
+
+
+# Tell versions [3.59,3.63) of GNU make to not export all variables.
+# Otherwise a system limit (for SysV at least) may be exceeded.
+.NOEXPORT:
diff --git a/configure b/configure
new file mode 100755
index 00000000..ea75c846
--- /dev/null
+++ b/configure
@@ -0,0 +1,6545 @@
+#! /bin/sh
+# Guess values for system-dependent variables and create Makefiles.
+# Generated by GNU Autoconf 2.69 for Grid 1.0.
+#
+# Report bugs to <paboyle@ph.ed.ac.uk>.
+#
+#
+# Copyright (C) 1992-1996, 1998-2012 Free Software Foundation, Inc.
+#
+#
+# This configure script is free software; the Free Software Foundation
+# gives unlimited permission to copy, distribute and modify it.
+## -------------------- ##
+## M4sh Initialization. ##
+## -------------------- ##
+
+# Be more Bourne compatible
+DUALCASE=1; export DUALCASE # for MKS sh
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
+  emulate sh
+  NULLCMD=:
+  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '${1+"$@"}'='"$@"'
+  setopt NO_GLOB_SUBST
+else
+  case `(set -o) 2>/dev/null` in #(
+  *posix*) :
+    set -o posix ;; #(
+  *) :
+     ;;
+esac
+fi
+
+
+as_nl='
+'
+export as_nl
+# Printing a long string crashes Solaris 7 /usr/bin/printf.
+as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
+as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
+as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
+# Prefer a ksh shell builtin over an external printf program on Solaris,
+# but without wasting forks for bash or zsh.
+if test -z "$BASH_VERSION$ZSH_VERSION" \
+    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
+  as_echo='print -r --'
+  as_echo_n='print -rn --'
+elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+  as_echo='printf %s\n'
+  as_echo_n='printf %s'
+else
+  if test "X`(/usr/ucb/echo -n -n $as_echo) 2>/dev/null`" = "X-n $as_echo"; then
+    as_echo_body='eval /usr/ucb/echo -n "$1$as_nl"'
+    as_echo_n='/usr/ucb/echo -n'
+  else
+    as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
+    as_echo_n_body='eval
+      arg=$1;
+      case $arg in #(
+      *"$as_nl"*)
+	expr "X$arg" : "X\\(.*\\)$as_nl";
+	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
+      esac;
+      expr "X$arg" : "X\\(.*\\)" | tr -d "$as_nl"
+    '
+    export as_echo_n_body
+    as_echo_n='sh -c $as_echo_n_body as_echo'
+  fi
+  export as_echo_body
+  as_echo='sh -c $as_echo_body as_echo'
+fi
+
+# The user is always right.
+if test "${PATH_SEPARATOR+set}" != set; then
+  PATH_SEPARATOR=:
+  (PATH='/bin;/bin'; FPATH=$PATH; sh -c :) >/dev/null 2>&1 && {
+    (PATH='/bin:/bin'; FPATH=$PATH; sh -c :) >/dev/null 2>&1 ||
+      PATH_SEPARATOR=';'
+  }
+fi
+
+
+# IFS
+# We need space, tab and new line, in precisely that order.  Quoting is
+# there to prevent editors from complaining about space-tab.
+# (If _AS_PATH_WALK were called with IFS unset, it would disable word
+# splitting by setting IFS to empty value.)
+IFS=" ""	$as_nl"
+
+# Find who we are.  Look in the path if we contain no directory separator.
+as_myself=
+case $0 in #((
+  *[\\/]* ) as_myself=$0 ;;
+  *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+  done
+IFS=$as_save_IFS
+
+     ;;
+esac
+# We did not find ourselves, most probably we were run as `sh COMMAND'
+# in which case we are not to be found in the path.
+if test "x$as_myself" = x; then
+  as_myself=$0
+fi
+if test ! -f "$as_myself"; then
+  $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
+  exit 1
+fi
+
+# Unset variables that we do not need and which cause bugs (e.g. in
+# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
+# suppresses any "Segmentation fault" message there.  '((' could
+# trigger a bug in pdksh 5.2.14.
+for as_var in BASH_ENV ENV MAIL MAILPATH
+do eval test x\${$as_var+set} = xset \
+  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
+done
+PS1='$ '
+PS2='> '
+PS4='+ '
+
+# NLS nuisances.
+LC_ALL=C
+export LC_ALL
+LANGUAGE=C
+export LANGUAGE
+
+# CDPATH.
+(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
+
+# Use a proper internal environment variable to ensure we don't fall
+  # into an infinite loop, continuously re-executing ourselves.
+  if test x"${_as_can_reexec}" != xno && test "x$CONFIG_SHELL" != x; then
+    _as_can_reexec=no; export _as_can_reexec;
+    # We cannot yet assume a decent shell, so we have to provide a
+# neutralization value for shells without unset; and this also
+# works around shells that cannot unset nonexistent variables.
+# Preserve -v and -x to the replacement shell.
+BASH_ENV=/dev/null
+ENV=/dev/null
+(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
+case $- in # ((((
+  *v*x* | *x*v* ) as_opts=-vx ;;
+  *v* ) as_opts=-v ;;
+  *x* ) as_opts=-x ;;
+  * ) as_opts= ;;
+esac
+exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
+# Admittedly, this is quite paranoid, since all the known shells bail
+# out after a failed `exec'.
+$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
+as_fn_exit 255
+  fi
+  # We don't want this to propagate to other subprocesses.
+          { _as_can_reexec=; unset _as_can_reexec;}
+if test "x$CONFIG_SHELL" = x; then
+  as_bourne_compatible="if test -n \"\${ZSH_VERSION+set}\" && (emulate sh) >/dev/null 2>&1; then :
+  emulate sh
+  NULLCMD=:
+  # Pre-4.2 versions of Zsh do word splitting on \${1+\"\$@\"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '\${1+\"\$@\"}'='\"\$@\"'
+  setopt NO_GLOB_SUBST
+else
+  case \`(set -o) 2>/dev/null\` in #(
+  *posix*) :
+    set -o posix ;; #(
+  *) :
+     ;;
+esac
+fi
+"
+  as_required="as_fn_return () { (exit \$1); }
+as_fn_success () { as_fn_return 0; }
+as_fn_failure () { as_fn_return 1; }
+as_fn_ret_success () { return 0; }
+as_fn_ret_failure () { return 1; }
+
+exitcode=0
+as_fn_success || { exitcode=1; echo as_fn_success failed.; }
+as_fn_failure && { exitcode=1; echo as_fn_failure succeeded.; }
+as_fn_ret_success || { exitcode=1; echo as_fn_ret_success failed.; }
+as_fn_ret_failure && { exitcode=1; echo as_fn_ret_failure succeeded.; }
+if ( set x; as_fn_ret_success y && test x = \"\$1\" ); then :
+
+else
+  exitcode=1; echo positional parameters were not saved.
+fi
+test x\$exitcode = x0 || exit 1
+test -x / || exit 1"
+  as_suggested="  as_lineno_1=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_1a=\$LINENO
+  as_lineno_2=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_2a=\$LINENO
+  eval 'test \"x\$as_lineno_1'\$as_run'\" != \"x\$as_lineno_2'\$as_run'\" &&
+  test \"x\`expr \$as_lineno_1'\$as_run' + 1\`\" = \"x\$as_lineno_2'\$as_run'\"' || exit 1
+test \$(( 1 + 1 )) = 2 || exit 1"
+  if (eval "$as_required") 2>/dev/null; then :
+  as_have_required=yes
+else
+  as_have_required=no
+fi
+  if test x$as_have_required = xyes && (eval "$as_suggested") 2>/dev/null; then :
+
+else
+  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+as_found=false
+for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+  as_found=:
+  case $as_dir in #(
+	 /*)
+	   for as_base in sh bash ksh sh5; do
+	     # Try only shells that exist, to save several forks.
+	     as_shell=$as_dir/$as_base
+	     if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
+		    { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$as_shell"; } 2>/dev/null; then :
+  CONFIG_SHELL=$as_shell as_have_required=yes
+		   if { $as_echo "$as_bourne_compatible""$as_suggested" | as_run=a "$as_shell"; } 2>/dev/null; then :
+  break 2
+fi
+fi
+	   done;;
+       esac
+  as_found=false
+done
+$as_found || { if { test -f "$SHELL" || test -f "$SHELL.exe"; } &&
+	      { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$SHELL"; } 2>/dev/null; then :
+  CONFIG_SHELL=$SHELL as_have_required=yes
+fi; }
+IFS=$as_save_IFS
+
+
+      if test "x$CONFIG_SHELL" != x; then :
+  export CONFIG_SHELL
+             # We cannot yet assume a decent shell, so we have to provide a
+# neutralization value for shells without unset; and this also
+# works around shells that cannot unset nonexistent variables.
+# Preserve -v and -x to the replacement shell.
+BASH_ENV=/dev/null
+ENV=/dev/null
+(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
+case $- in # ((((
+  *v*x* | *x*v* ) as_opts=-vx ;;
+  *v* ) as_opts=-v ;;
+  *x* ) as_opts=-x ;;
+  * ) as_opts= ;;
+esac
+exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
+# Admittedly, this is quite paranoid, since all the known shells bail
+# out after a failed `exec'.
+$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
+exit 255
+fi
+
+    if test x$as_have_required = xno; then :
+  $as_echo "$0: This script requires a shell more modern than all"
+  $as_echo "$0: the shells that I found on your system."
+  if test x${ZSH_VERSION+set} = xset ; then
+    $as_echo "$0: In particular, zsh $ZSH_VERSION has bugs and should"
+    $as_echo "$0: be upgraded to zsh 4.3.4 or later."
+  else
+    $as_echo "$0: Please tell bug-autoconf@gnu.org and
+$0: paboyle@ph.ed.ac.uk about your system, including any
+$0: error possibly output before this message. Then install
+$0: a modern shell, or manually run the script under such a
+$0: shell if you do have one."
+  fi
+  exit 1
+fi
+fi
+fi
+SHELL=${CONFIG_SHELL-/bin/sh}
+export SHELL
+# Unset more variables known to interfere with behavior of common tools.
+CLICOLOR_FORCE= GREP_OPTIONS=
+unset CLICOLOR_FORCE GREP_OPTIONS
+
+## --------------------- ##
+## M4sh Shell Functions. ##
+## --------------------- ##
+# as_fn_unset VAR
+# ---------------
+# Portably unset VAR.
+as_fn_unset ()
+{
+  { eval $1=; unset $1;}
+}
+as_unset=as_fn_unset
+
+# as_fn_set_status STATUS
+# -----------------------
+# Set $? to STATUS, without forking.
+as_fn_set_status ()
+{
+  return $1
+} # as_fn_set_status
+
+# as_fn_exit STATUS
+# -----------------
+# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
+as_fn_exit ()
+{
+  set +e
+  as_fn_set_status $1
+  exit $1
+} # as_fn_exit
+
+# as_fn_mkdir_p
+# -------------
+# Create "$as_dir" as a directory, including parents if necessary.
+as_fn_mkdir_p ()
+{
+
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || eval $as_mkdir_p || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
+
+
+} # as_fn_mkdir_p
+
+# as_fn_executable_p FILE
+# -----------------------
+# Test if FILE is an executable regular file.
+as_fn_executable_p ()
+{
+  test -f "$1" && test -x "$1"
+} # as_fn_executable_p
+# as_fn_append VAR VALUE
+# ----------------------
+# Append the text in VALUE to the end of the definition contained in VAR. Take
+# advantage of any shell optimizations that allow amortized linear growth over
+# repeated appends, instead of the typical quadratic growth present in naive
+# implementations.
+if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
+  eval 'as_fn_append ()
+  {
+    eval $1+=\$2
+  }'
+else
+  as_fn_append ()
+  {
+    eval $1=\$$1\$2
+  }
+fi # as_fn_append
+
+# as_fn_arith ARG...
+# ------------------
+# Perform arithmetic evaluation on the ARGs, and store the result in the
+# global $as_val. Take advantage of shells that can avoid forks. The arguments
+# must be portable across $(()) and expr.
+if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
+  eval 'as_fn_arith ()
+  {
+    as_val=$(( $* ))
+  }'
+else
+  as_fn_arith ()
+  {
+    as_val=`expr "$@" || test $? -eq 1`
+  }
+fi # as_fn_arith
+
+
+# as_fn_error STATUS ERROR [LINENO LOG_FD]
+# ----------------------------------------
+# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
+# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
+# script with STATUS, using 1 if that was 0.
+as_fn_error ()
+{
+  as_status=$1; test $as_status -eq 0 && as_status=1
+  if test "$4"; then
+    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
+  fi
+  $as_echo "$as_me: error: $2" >&2
+  as_fn_exit $as_status
+} # as_fn_error
+
+if expr a : '\(a\)' >/dev/null 2>&1 &&
+   test "X`expr 00001 : '.*\(...\)'`" = X001; then
+  as_expr=expr
+else
+  as_expr=false
+fi
+
+if (basename -- /) >/dev/null 2>&1 && test "X`basename -- / 2>&1`" = "X/"; then
+  as_basename=basename
+else
+  as_basename=false
+fi
+
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
+
+as_me=`$as_basename -- "$0" ||
+$as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
+	 X"$0" : 'X\(//\)$' \| \
+	 X"$0" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X/"$0" |
+    sed '/^.*\/\([^/][^/]*\)\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\/\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\/\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
+
+
+  as_lineno_1=$LINENO as_lineno_1a=$LINENO
+  as_lineno_2=$LINENO as_lineno_2a=$LINENO
+  eval 'test "x$as_lineno_1'$as_run'" != "x$as_lineno_2'$as_run'" &&
+  test "x`expr $as_lineno_1'$as_run' + 1`" = "x$as_lineno_2'$as_run'"' || {
+  # Blame Lee E. McMahon (1931-1989) for sed's syntax.  :-)
+  sed -n '
+    p
+    /[$]LINENO/=
+  ' <$as_myself |
+    sed '
+      s/[$]LINENO.*/&-/
+      t lineno
+      b
+      :lineno
+      N
+      :loop
+      s/[$]LINENO\([^'$as_cr_alnum'_].*\n\)\(.*\)/\2\1\2/
+      t loop
+      s/-\n.*//
+    ' >$as_me.lineno &&
+  chmod +x "$as_me.lineno" ||
+    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2; as_fn_exit 1; }
+
+  # If we had to re-execute with $CONFIG_SHELL, we're ensured to have
+  # already done that, so ensure we don't try to do so again and fall
+  # in an infinite loop.  This has already happened in practice.
+  _as_can_reexec=no; export _as_can_reexec
+  # Don't try to exec as it changes $[0], causing all sort of problems
+  # (the dirname of $[0] is not the place where we might find the
+  # original and so on.  Autoconf is especially sensitive to this).
+  . "./$as_me.lineno"
+  # Exit status is that of the last command.
+  exit
+}
+
+ECHO_C= ECHO_N= ECHO_T=
+case `echo -n x` in #(((((
+-n*)
+  case `echo 'xy\c'` in
+  *c*) ECHO_T='	';;	# ECHO_T is single tab character.
+  xy)  ECHO_C='\c';;
+  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
+       ECHO_T='	';;
+  esac;;
+*)
+  ECHO_N='-n';;
+esac
+
+rm -f conf$$ conf$$.exe conf$$.file
+if test -d conf$$.dir; then
+  rm -f conf$$.dir/conf$$.file
+else
+  rm -f conf$$.dir
+  mkdir conf$$.dir 2>/dev/null
+fi
+if (echo >conf$$.file) 2>/dev/null; then
+  if ln -s conf$$.file conf$$ 2>/dev/null; then
+    as_ln_s='ln -s'
+    # ... but there are two gotchas:
+    # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
+    # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
+    # In both cases, we have to default to `cp -pR'.
+    ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
+      as_ln_s='cp -pR'
+  elif ln conf$$.file conf$$ 2>/dev/null; then
+    as_ln_s=ln
+  else
+    as_ln_s='cp -pR'
+  fi
+else
+  as_ln_s='cp -pR'
+fi
+rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
+rmdir conf$$.dir 2>/dev/null
+
+if mkdir -p . 2>/dev/null; then
+  as_mkdir_p='mkdir -p "$as_dir"'
+else
+  test -d ./-p && rmdir ./-p
+  as_mkdir_p=false
+fi
+
+as_test_x='test -x'
+as_executable_p=as_fn_executable_p
+
+# Sed expression to map a string onto a valid CPP name.
+as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
+
+# Sed expression to map a string onto a valid variable name.
+as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
+
+
+test -n "$DJDIR" || exec 7<&0 </dev/null
+exec 6>&1
+
+# Name of the host.
+# hostname on some systems (SVR3.2, old GNU/Linux) returns a bogus exit status,
+# so uname gets run too.
+ac_hostname=`(hostname || uname -n) 2>/dev/null | sed 1q`
+
+#
+# Initializations.
+#
+ac_default_prefix=/usr/local
+ac_clean_files=
+ac_config_libobj_dir=.
+LIBOBJS=
+cross_compiling=no
+subdirs=
+MFLAGS=
+MAKEFLAGS=
+
+# Identity of this package.
+PACKAGE_NAME='Grid'
+PACKAGE_TARNAME='grid'
+PACKAGE_VERSION='1.0'
+PACKAGE_STRING='Grid 1.0'
+PACKAGE_BUGREPORT='paboyle@ph.ed.ac.uk'
+PACKAGE_URL=''
+
+ac_unique_file="Grid.h"
+# Factoring default headers for most tests.
+ac_includes_default="\
+#include <stdio.h>
+#ifdef HAVE_SYS_TYPES_H
+# include <sys/types.h>
+#endif
+#ifdef HAVE_SYS_STAT_H
+# include <sys/stat.h>
+#endif
+#ifdef STDC_HEADERS
+# include <stdlib.h>
+# include <stddef.h>
+#else
+# ifdef HAVE_STDLIB_H
+#  include <stdlib.h>
+# endif
+#endif
+#ifdef HAVE_STRING_H
+# if !defined STDC_HEADERS && defined HAVE_MEMORY_H
+#  include <memory.h>
+# endif
+# include <string.h>
+#endif
+#ifdef HAVE_STRINGS_H
+# include <strings.h>
+#endif
+#ifdef HAVE_INTTYPES_H
+# include <inttypes.h>
+#endif
+#ifdef HAVE_STDINT_H
+# include <stdint.h>
+#endif
+#ifdef HAVE_UNISTD_H
+# include <unistd.h>
+#endif"
+
+ac_subst_vars='am__EXEEXT_FALSE
+am__EXEEXT_TRUE
+LTLIBOBJS
+LIBOBJS
+EGREP
+GREP
+CPP
+RANLIB
+OPENMP_CFLAGS
+am__fastdepCC_FALSE
+am__fastdepCC_TRUE
+CCDEPMODE
+ac_ct_CC
+CFLAGS
+CC
+am__fastdepCXX_FALSE
+am__fastdepCXX_TRUE
+CXXDEPMODE
+am__nodep
+AMDEPBACKSLASH
+AMDEP_FALSE
+AMDEP_TRUE
+am__quote
+am__include
+DEPDIR
+OBJEXT
+EXEEXT
+ac_ct_CXX
+CPPFLAGS
+LDFLAGS
+CXXFLAGS
+CXX
+AM_BACKSLASH
+AM_DEFAULT_VERBOSITY
+AM_DEFAULT_V
+AM_V
+am__untar
+am__tar
+AMTAR
+am__leading_dot
+SET_MAKE
+AWK
+mkdir_p
+MKDIR_P
+INSTALL_STRIP_PROGRAM
+STRIP
+install_sh
+MAKEINFO
+AUTOHEADER
+AUTOMAKE
+AUTOCONF
+ACLOCAL
+VERSION
+PACKAGE
+CYGPATH_W
+am__isrc
+INSTALL_DATA
+INSTALL_SCRIPT
+INSTALL_PROGRAM
+target_alias
+host_alias
+build_alias
+LIBS
+ECHO_T
+ECHO_N
+ECHO_C
+DEFS
+mandir
+localedir
+libdir
+psdir
+pdfdir
+dvidir
+htmldir
+infodir
+docdir
+oldincludedir
+includedir
+localstatedir
+sharedstatedir
+sysconfdir
+datadir
+datarootdir
+libexecdir
+sbindir
+bindir
+program_transform_name
+prefix
+exec_prefix
+PACKAGE_URL
+PACKAGE_BUGREPORT
+PACKAGE_STRING
+PACKAGE_VERSION
+PACKAGE_TARNAME
+PACKAGE_NAME
+PATH_SEPARATOR
+SHELL'
+ac_subst_files=''
+ac_user_opts='
+enable_option_checking
+enable_silent_rules
+enable_dependency_tracking
+enable_openmp
+enable_simd
+'
+      ac_precious_vars='build_alias
+host_alias
+target_alias
+CXX
+CXXFLAGS
+LDFLAGS
+LIBS
+CPPFLAGS
+CCC
+CC
+CFLAGS
+CPP'
+
+
+# Initialize some variables set by options.
+ac_init_help=
+ac_init_version=false
+ac_unrecognized_opts=
+ac_unrecognized_sep=
+# The variables have the same names as the options, with
+# dashes changed to underlines.
+cache_file=/dev/null
+exec_prefix=NONE
+no_create=
+no_recursion=
+prefix=NONE
+program_prefix=NONE
+program_suffix=NONE
+program_transform_name=s,x,x,
+silent=
+site=
+srcdir=
+verbose=
+x_includes=NONE
+x_libraries=NONE
+
+# Installation directory options.
+# These are left unexpanded so users can "make install exec_prefix=/foo"
+# and all the variables that are supposed to be based on exec_prefix
+# by default will actually change.
+# Use braces instead of parens because sh, perl, etc. also accept them.
+# (The list follows the same order as the GNU Coding Standards.)
+bindir='${exec_prefix}/bin'
+sbindir='${exec_prefix}/sbin'
+libexecdir='${exec_prefix}/libexec'
+datarootdir='${prefix}/share'
+datadir='${datarootdir}'
+sysconfdir='${prefix}/etc'
+sharedstatedir='${prefix}/com'
+localstatedir='${prefix}/var'
+includedir='${prefix}/include'
+oldincludedir='/usr/include'
+docdir='${datarootdir}/doc/${PACKAGE_TARNAME}'
+infodir='${datarootdir}/info'
+htmldir='${docdir}'
+dvidir='${docdir}'
+pdfdir='${docdir}'
+psdir='${docdir}'
+libdir='${exec_prefix}/lib'
+localedir='${datarootdir}/locale'
+mandir='${datarootdir}/man'
+
+ac_prev=
+ac_dashdash=
+for ac_option
+do
+  # If the previous option needs an argument, assign it.
+  if test -n "$ac_prev"; then
+    eval $ac_prev=\$ac_option
+    ac_prev=
+    continue
+  fi
+
+  case $ac_option in
+  *=?*) ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
+  *=)   ac_optarg= ;;
+  *)    ac_optarg=yes ;;
+  esac
+
+  # Accept the important Cygnus configure options, so we can diagnose typos.
+
+  case $ac_dashdash$ac_option in
+  --)
+    ac_dashdash=yes ;;
+
+  -bindir | --bindir | --bindi | --bind | --bin | --bi)
+    ac_prev=bindir ;;
+  -bindir=* | --bindir=* | --bindi=* | --bind=* | --bin=* | --bi=*)
+    bindir=$ac_optarg ;;
+
+  -build | --build | --buil | --bui | --bu)
+    ac_prev=build_alias ;;
+  -build=* | --build=* | --buil=* | --bui=* | --bu=*)
+    build_alias=$ac_optarg ;;
+
+  -cache-file | --cache-file | --cache-fil | --cache-fi \
+  | --cache-f | --cache- | --cache | --cach | --cac | --ca | --c)
+    ac_prev=cache_file ;;
+  -cache-file=* | --cache-file=* | --cache-fil=* | --cache-fi=* \
+  | --cache-f=* | --cache-=* | --cache=* | --cach=* | --cac=* | --ca=* | --c=*)
+    cache_file=$ac_optarg ;;
+
+  --config-cache | -C)
+    cache_file=config.cache ;;
+
+  -datadir | --datadir | --datadi | --datad)
+    ac_prev=datadir ;;
+  -datadir=* | --datadir=* | --datadi=* | --datad=*)
+    datadir=$ac_optarg ;;
+
+  -datarootdir | --datarootdir | --datarootdi | --datarootd | --dataroot \
+  | --dataroo | --dataro | --datar)
+    ac_prev=datarootdir ;;
+  -datarootdir=* | --datarootdir=* | --datarootdi=* | --datarootd=* \
+  | --dataroot=* | --dataroo=* | --dataro=* | --datar=*)
+    datarootdir=$ac_optarg ;;
+
+  -disable-* | --disable-*)
+    ac_useropt=`expr "x$ac_option" : 'x-*disable-\(.*\)'`
+    # Reject names that are not valid shell variable names.
+    expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
+      as_fn_error $? "invalid feature name: $ac_useropt"
+    ac_useropt_orig=$ac_useropt
+    ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
+    case $ac_user_opts in
+      *"
+"enable_$ac_useropt"
+"*) ;;
+      *) ac_unrecognized_opts="$ac_unrecognized_opts$ac_unrecognized_sep--disable-$ac_useropt_orig"
+	 ac_unrecognized_sep=', ';;
+    esac
+    eval enable_$ac_useropt=no ;;
+
+  -docdir | --docdir | --docdi | --doc | --do)
+    ac_prev=docdir ;;
+  -docdir=* | --docdir=* | --docdi=* | --doc=* | --do=*)
+    docdir=$ac_optarg ;;
+
+  -dvidir | --dvidir | --dvidi | --dvid | --dvi | --dv)
+    ac_prev=dvidir ;;
+  -dvidir=* | --dvidir=* | --dvidi=* | --dvid=* | --dvi=* | --dv=*)
+    dvidir=$ac_optarg ;;
+
+  -enable-* | --enable-*)
+    ac_useropt=`expr "x$ac_option" : 'x-*enable-\([^=]*\)'`
+    # Reject names that are not valid shell variable names.
+    expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
+      as_fn_error $? "invalid feature name: $ac_useropt"
+    ac_useropt_orig=$ac_useropt
+    ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
+    case $ac_user_opts in
+      *"
+"enable_$ac_useropt"
+"*) ;;
+      *) ac_unrecognized_opts="$ac_unrecognized_opts$ac_unrecognized_sep--enable-$ac_useropt_orig"
+	 ac_unrecognized_sep=', ';;
+    esac
+    eval enable_$ac_useropt=\$ac_optarg ;;
+
+  -exec-prefix | --exec_prefix | --exec-prefix | --exec-prefi \
+  | --exec-pref | --exec-pre | --exec-pr | --exec-p | --exec- \
+  | --exec | --exe | --ex)
+    ac_prev=exec_prefix ;;
+  -exec-prefix=* | --exec_prefix=* | --exec-prefix=* | --exec-prefi=* \
+  | --exec-pref=* | --exec-pre=* | --exec-pr=* | --exec-p=* | --exec-=* \
+  | --exec=* | --exe=* | --ex=*)
+    exec_prefix=$ac_optarg ;;
+
+  -gas | --gas | --ga | --g)
+    # Obsolete; use --with-gas.
+    with_gas=yes ;;
+
+  -help | --help | --hel | --he | -h)
+    ac_init_help=long ;;
+  -help=r* | --help=r* | --hel=r* | --he=r* | -hr*)
+    ac_init_help=recursive ;;
+  -help=s* | --help=s* | --hel=s* | --he=s* | -hs*)
+    ac_init_help=short ;;
+
+  -host | --host | --hos | --ho)
+    ac_prev=host_alias ;;
+  -host=* | --host=* | --hos=* | --ho=*)
+    host_alias=$ac_optarg ;;
+
+  -htmldir | --htmldir | --htmldi | --htmld | --html | --htm | --ht)
+    ac_prev=htmldir ;;
+  -htmldir=* | --htmldir=* | --htmldi=* | --htmld=* | --html=* | --htm=* \
+  | --ht=*)
+    htmldir=$ac_optarg ;;
+
+  -includedir | --includedir | --includedi | --included | --include \
+  | --includ | --inclu | --incl | --inc)
+    ac_prev=includedir ;;
+  -includedir=* | --includedir=* | --includedi=* | --included=* | --include=* \
+  | --includ=* | --inclu=* | --incl=* | --inc=*)
+    includedir=$ac_optarg ;;
+
+  -infodir | --infodir | --infodi | --infod | --info | --inf)
+    ac_prev=infodir ;;
+  -infodir=* | --infodir=* | --infodi=* | --infod=* | --info=* | --inf=*)
+    infodir=$ac_optarg ;;
+
+  -libdir | --libdir | --libdi | --libd)
+    ac_prev=libdir ;;
+  -libdir=* | --libdir=* | --libdi=* | --libd=*)
+    libdir=$ac_optarg ;;
+
+  -libexecdir | --libexecdir | --libexecdi | --libexecd | --libexec \
+  | --libexe | --libex | --libe)
+    ac_prev=libexecdir ;;
+  -libexecdir=* | --libexecdir=* | --libexecdi=* | --libexecd=* | --libexec=* \
+  | --libexe=* | --libex=* | --libe=*)
+    libexecdir=$ac_optarg ;;
+
+  -localedir | --localedir | --localedi | --localed | --locale)
+    ac_prev=localedir ;;
+  -localedir=* | --localedir=* | --localedi=* | --localed=* | --locale=*)
+    localedir=$ac_optarg ;;
+
+  -localstatedir | --localstatedir | --localstatedi | --localstated \
+  | --localstate | --localstat | --localsta | --localst | --locals)
+    ac_prev=localstatedir ;;
+  -localstatedir=* | --localstatedir=* | --localstatedi=* | --localstated=* \
+  | --localstate=* | --localstat=* | --localsta=* | --localst=* | --locals=*)
+    localstatedir=$ac_optarg ;;
+
+  -mandir | --mandir | --mandi | --mand | --man | --ma | --m)
+    ac_prev=mandir ;;
+  -mandir=* | --mandir=* | --mandi=* | --mand=* | --man=* | --ma=* | --m=*)
+    mandir=$ac_optarg ;;
+
+  -nfp | --nfp | --nf)
+    # Obsolete; use --without-fp.
+    with_fp=no ;;
+
+  -no-create | --no-create | --no-creat | --no-crea | --no-cre \
+  | --no-cr | --no-c | -n)
+    no_create=yes ;;
+
+  -no-recursion | --no-recursion | --no-recursio | --no-recursi \
+  | --no-recurs | --no-recur | --no-recu | --no-rec | --no-re | --no-r)
+    no_recursion=yes ;;
+
+  -oldincludedir | --oldincludedir | --oldincludedi | --oldincluded \
+  | --oldinclude | --oldinclud | --oldinclu | --oldincl | --oldinc \
+  | --oldin | --oldi | --old | --ol | --o)
+    ac_prev=oldincludedir ;;
+  -oldincludedir=* | --oldincludedir=* | --oldincludedi=* | --oldincluded=* \
+  | --oldinclude=* | --oldinclud=* | --oldinclu=* | --oldincl=* | --oldinc=* \
+  | --oldin=* | --oldi=* | --old=* | --ol=* | --o=*)
+    oldincludedir=$ac_optarg ;;
+
+  -prefix | --prefix | --prefi | --pref | --pre | --pr | --p)
+    ac_prev=prefix ;;
+  -prefix=* | --prefix=* | --prefi=* | --pref=* | --pre=* | --pr=* | --p=*)
+    prefix=$ac_optarg ;;
+
+  -program-prefix | --program-prefix | --program-prefi | --program-pref \
+  | --program-pre | --program-pr | --program-p)
+    ac_prev=program_prefix ;;
+  -program-prefix=* | --program-prefix=* | --program-prefi=* \
+  | --program-pref=* | --program-pre=* | --program-pr=* | --program-p=*)
+    program_prefix=$ac_optarg ;;
+
+  -program-suffix | --program-suffix | --program-suffi | --program-suff \
+  | --program-suf | --program-su | --program-s)
+    ac_prev=program_suffix ;;
+  -program-suffix=* | --program-suffix=* | --program-suffi=* \
+  | --program-suff=* | --program-suf=* | --program-su=* | --program-s=*)
+    program_suffix=$ac_optarg ;;
+
+  -program-transform-name | --program-transform-name \
+  | --program-transform-nam | --program-transform-na \
+  | --program-transform-n | --program-transform- \
+  | --program-transform | --program-transfor \
+  | --program-transfo | --program-transf \
+  | --program-trans | --program-tran \
+  | --progr-tra | --program-tr | --program-t)
+    ac_prev=program_transform_name ;;
+  -program-transform-name=* | --program-transform-name=* \
+  | --program-transform-nam=* | --program-transform-na=* \
+  | --program-transform-n=* | --program-transform-=* \
+  | --program-transform=* | --program-transfor=* \
+  | --program-transfo=* | --program-transf=* \
+  | --program-trans=* | --program-tran=* \
+  | --progr-tra=* | --program-tr=* | --program-t=*)
+    program_transform_name=$ac_optarg ;;
+
+  -pdfdir | --pdfdir | --pdfdi | --pdfd | --pdf | --pd)
+    ac_prev=pdfdir ;;
+  -pdfdir=* | --pdfdir=* | --pdfdi=* | --pdfd=* | --pdf=* | --pd=*)
+    pdfdir=$ac_optarg ;;
+
+  -psdir | --psdir | --psdi | --psd | --ps)
+    ac_prev=psdir ;;
+  -psdir=* | --psdir=* | --psdi=* | --psd=* | --ps=*)
+    psdir=$ac_optarg ;;
+
+  -q | -quiet | --quiet | --quie | --qui | --qu | --q \
+  | -silent | --silent | --silen | --sile | --sil)
+    silent=yes ;;
+
+  -sbindir | --sbindir | --sbindi | --sbind | --sbin | --sbi | --sb)
+    ac_prev=sbindir ;;
+  -sbindir=* | --sbindir=* | --sbindi=* | --sbind=* | --sbin=* \
+  | --sbi=* | --sb=*)
+    sbindir=$ac_optarg ;;
+
+  -sharedstatedir | --sharedstatedir | --sharedstatedi \
+  | --sharedstated | --sharedstate | --sharedstat | --sharedsta \
+  | --sharedst | --shareds | --shared | --share | --shar \
+  | --sha | --sh)
+    ac_prev=sharedstatedir ;;
+  -sharedstatedir=* | --sharedstatedir=* | --sharedstatedi=* \
+  | --sharedstated=* | --sharedstate=* | --sharedstat=* | --sharedsta=* \
+  | --sharedst=* | --shareds=* | --shared=* | --share=* | --shar=* \
+  | --sha=* | --sh=*)
+    sharedstatedir=$ac_optarg ;;
+
+  -site | --site | --sit)
+    ac_prev=site ;;
+  -site=* | --site=* | --sit=*)
+    site=$ac_optarg ;;
+
+  -srcdir | --srcdir | --srcdi | --srcd | --src | --sr)
+    ac_prev=srcdir ;;
+  -srcdir=* | --srcdir=* | --srcdi=* | --srcd=* | --src=* | --sr=*)
+    srcdir=$ac_optarg ;;
+
+  -sysconfdir | --sysconfdir | --sysconfdi | --sysconfd | --sysconf \
+  | --syscon | --sysco | --sysc | --sys | --sy)
+    ac_prev=sysconfdir ;;
+  -sysconfdir=* | --sysconfdir=* | --sysconfdi=* | --sysconfd=* | --sysconf=* \
+  | --syscon=* | --sysco=* | --sysc=* | --sys=* | --sy=*)
+    sysconfdir=$ac_optarg ;;
+
+  -target | --target | --targe | --targ | --tar | --ta | --t)
+    ac_prev=target_alias ;;
+  -target=* | --target=* | --targe=* | --targ=* | --tar=* | --ta=* | --t=*)
+    target_alias=$ac_optarg ;;
+
+  -v | -verbose | --verbose | --verbos | --verbo | --verb)
+    verbose=yes ;;
+
+  -version | --version | --versio | --versi | --vers | -V)
+    ac_init_version=: ;;
+
+  -with-* | --with-*)
+    ac_useropt=`expr "x$ac_option" : 'x-*with-\([^=]*\)'`
+    # Reject names that are not valid shell variable names.
+    expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
+      as_fn_error $? "invalid package name: $ac_useropt"
+    ac_useropt_orig=$ac_useropt
+    ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
+    case $ac_user_opts in
+      *"
+"with_$ac_useropt"
+"*) ;;
+      *) ac_unrecognized_opts="$ac_unrecognized_opts$ac_unrecognized_sep--with-$ac_useropt_orig"
+	 ac_unrecognized_sep=', ';;
+    esac
+    eval with_$ac_useropt=\$ac_optarg ;;
+
+  -without-* | --without-*)
+    ac_useropt=`expr "x$ac_option" : 'x-*without-\(.*\)'`
+    # Reject names that are not valid shell variable names.
+    expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
+      as_fn_error $? "invalid package name: $ac_useropt"
+    ac_useropt_orig=$ac_useropt
+    ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
+    case $ac_user_opts in
+      *"
+"with_$ac_useropt"
+"*) ;;
+      *) ac_unrecognized_opts="$ac_unrecognized_opts$ac_unrecognized_sep--without-$ac_useropt_orig"
+	 ac_unrecognized_sep=', ';;
+    esac
+    eval with_$ac_useropt=no ;;
+
+  --x)
+    # Obsolete; use --with-x.
+    with_x=yes ;;
+
+  -x-includes | --x-includes | --x-include | --x-includ | --x-inclu \
+  | --x-incl | --x-inc | --x-in | --x-i)
+    ac_prev=x_includes ;;
+  -x-includes=* | --x-includes=* | --x-include=* | --x-includ=* | --x-inclu=* \
+  | --x-incl=* | --x-inc=* | --x-in=* | --x-i=*)
+    x_includes=$ac_optarg ;;
+
+  -x-libraries | --x-libraries | --x-librarie | --x-librari \
+  | --x-librar | --x-libra | --x-libr | --x-lib | --x-li | --x-l)
+    ac_prev=x_libraries ;;
+  -x-libraries=* | --x-libraries=* | --x-librarie=* | --x-librari=* \
+  | --x-librar=* | --x-libra=* | --x-libr=* | --x-lib=* | --x-li=* | --x-l=*)
+    x_libraries=$ac_optarg ;;
+
+  -*) as_fn_error $? "unrecognized option: \`$ac_option'
+Try \`$0 --help' for more information"
+    ;;
+
+  *=*)
+    ac_envvar=`expr "x$ac_option" : 'x\([^=]*\)='`
+    # Reject names that are not valid shell variable names.
+    case $ac_envvar in #(
+      '' | [0-9]* | *[!_$as_cr_alnum]* )
+      as_fn_error $? "invalid variable name: \`$ac_envvar'" ;;
+    esac
+    eval $ac_envvar=\$ac_optarg
+    export $ac_envvar ;;
+
+  *)
+    # FIXME: should be removed in autoconf 3.0.
+    $as_echo "$as_me: WARNING: you should use --build, --host, --target" >&2
+    expr "x$ac_option" : ".*[^-._$as_cr_alnum]" >/dev/null &&
+      $as_echo "$as_me: WARNING: invalid host type: $ac_option" >&2
+    : "${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}"
+    ;;
+
+  esac
+done
+
+if test -n "$ac_prev"; then
+  ac_option=--`echo $ac_prev | sed 's/_/-/g'`
+  as_fn_error $? "missing argument to $ac_option"
+fi
+
+if test -n "$ac_unrecognized_opts"; then
+  case $enable_option_checking in
+    no) ;;
+    fatal) as_fn_error $? "unrecognized options: $ac_unrecognized_opts" ;;
+    *)     $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2 ;;
+  esac
+fi
+
+# Check all directory arguments for consistency.
+for ac_var in	exec_prefix prefix bindir sbindir libexecdir datarootdir \
+		datadir sysconfdir sharedstatedir localstatedir includedir \
+		oldincludedir docdir infodir htmldir dvidir pdfdir psdir \
+		libdir localedir mandir
+do
+  eval ac_val=\$$ac_var
+  # Remove trailing slashes.
+  case $ac_val in
+    */ )
+      ac_val=`expr "X$ac_val" : 'X\(.*[^/]\)' \| "X$ac_val" : 'X\(.*\)'`
+      eval $ac_var=\$ac_val;;
+  esac
+  # Be sure to have absolute directory names.
+  case $ac_val in
+    [\\/$]* | ?:[\\/]* )  continue;;
+    NONE | '' ) case $ac_var in *prefix ) continue;; esac;;
+  esac
+  as_fn_error $? "expected an absolute directory name for --$ac_var: $ac_val"
+done
+
+# There might be people who depend on the old broken behavior: `$host'
+# used to hold the argument of --host etc.
+# FIXME: To remove some day.
+build=$build_alias
+host=$host_alias
+target=$target_alias
+
+# FIXME: To remove some day.
+if test "x$host_alias" != x; then
+  if test "x$build_alias" = x; then
+    cross_compiling=maybe
+  elif test "x$build_alias" != "x$host_alias"; then
+    cross_compiling=yes
+  fi
+fi
+
+ac_tool_prefix=
+test -n "$host_alias" && ac_tool_prefix=$host_alias-
+
+test "$silent" = yes && exec 6>/dev/null
+
+
+ac_pwd=`pwd` && test -n "$ac_pwd" &&
+ac_ls_di=`ls -di .` &&
+ac_pwd_ls_di=`cd "$ac_pwd" && ls -di .` ||
+  as_fn_error $? "working directory cannot be determined"
+test "X$ac_ls_di" = "X$ac_pwd_ls_di" ||
+  as_fn_error $? "pwd does not report name of working directory"
+
+
+# Find the source files, if location was not specified.
+if test -z "$srcdir"; then
+  ac_srcdir_defaulted=yes
+  # Try the directory containing this script, then the parent directory.
+  ac_confdir=`$as_dirname -- "$as_myself" ||
+$as_expr X"$as_myself" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_myself" : 'X\(//\)[^/]' \| \
+	 X"$as_myself" : 'X\(//\)$' \| \
+	 X"$as_myself" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_myself" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+  srcdir=$ac_confdir
+  if test ! -r "$srcdir/$ac_unique_file"; then
+    srcdir=..
+  fi
+else
+  ac_srcdir_defaulted=no
+fi
+if test ! -r "$srcdir/$ac_unique_file"; then
+  test "$ac_srcdir_defaulted" = yes && srcdir="$ac_confdir or .."
+  as_fn_error $? "cannot find sources ($ac_unique_file) in $srcdir"
+fi
+ac_msg="sources are in $srcdir, but \`cd $srcdir' does not work"
+ac_abs_confdir=`(
+	cd "$srcdir" && test -r "./$ac_unique_file" || as_fn_error $? "$ac_msg"
+	pwd)`
+# When building in place, set srcdir=.
+if test "$ac_abs_confdir" = "$ac_pwd"; then
+  srcdir=.
+fi
+# Remove unnecessary trailing slashes from srcdir.
+# Double slashes in file names in object file debugging info
+# mess up M-x gdb in Emacs.
+case $srcdir in
+*/) srcdir=`expr "X$srcdir" : 'X\(.*[^/]\)' \| "X$srcdir" : 'X\(.*\)'`;;
+esac
+for ac_var in $ac_precious_vars; do
+  eval ac_env_${ac_var}_set=\${${ac_var}+set}
+  eval ac_env_${ac_var}_value=\$${ac_var}
+  eval ac_cv_env_${ac_var}_set=\${${ac_var}+set}
+  eval ac_cv_env_${ac_var}_value=\$${ac_var}
+done
+
+#
+# Report the --help message.
+#
+if test "$ac_init_help" = "long"; then
+  # Omit some internal or obsolete options to make the list less imposing.
+  # This message is too long to be a string in the A/UX 3.1 sh.
+  cat <<_ACEOF
+\`configure' configures Grid 1.0 to adapt to many kinds of systems.
+
+Usage: $0 [OPTION]... [VAR=VALUE]...
+
+To assign environment variables (e.g., CC, CFLAGS...), specify them as
+VAR=VALUE.  See below for descriptions of some of the useful variables.
+
+Defaults for the options are specified in brackets.
+
+Configuration:
+  -h, --help              display this help and exit
+      --help=short        display options specific to this package
+      --help=recursive    display the short help of all the included packages
+  -V, --version           display version information and exit
+  -q, --quiet, --silent   do not print \`checking ...' messages
+      --cache-file=FILE   cache test results in FILE [disabled]
+  -C, --config-cache      alias for \`--cache-file=config.cache'
+  -n, --no-create         do not create output files
+      --srcdir=DIR        find the sources in DIR [configure dir or \`..']
+
+Installation directories:
+  --prefix=PREFIX         install architecture-independent files in PREFIX
+                          [$ac_default_prefix]
+  --exec-prefix=EPREFIX   install architecture-dependent files in EPREFIX
+                          [PREFIX]
+
+By default, \`make install' will install all the files in
+\`$ac_default_prefix/bin', \`$ac_default_prefix/lib' etc.  You can specify
+an installation prefix other than \`$ac_default_prefix' using \`--prefix',
+for instance \`--prefix=\$HOME'.
+
+For better control, use the options below.
+
+Fine tuning of the installation directories:
+  --bindir=DIR            user executables [EPREFIX/bin]
+  --sbindir=DIR           system admin executables [EPREFIX/sbin]
+  --libexecdir=DIR        program executables [EPREFIX/libexec]
+  --sysconfdir=DIR        read-only single-machine data [PREFIX/etc]
+  --sharedstatedir=DIR    modifiable architecture-independent data [PREFIX/com]
+  --localstatedir=DIR     modifiable single-machine data [PREFIX/var]
+  --libdir=DIR            object code libraries [EPREFIX/lib]
+  --includedir=DIR        C header files [PREFIX/include]
+  --oldincludedir=DIR     C header files for non-gcc [/usr/include]
+  --datarootdir=DIR       read-only arch.-independent data root [PREFIX/share]
+  --datadir=DIR           read-only architecture-independent data [DATAROOTDIR]
+  --infodir=DIR           info documentation [DATAROOTDIR/info]
+  --localedir=DIR         locale-dependent data [DATAROOTDIR/locale]
+  --mandir=DIR            man documentation [DATAROOTDIR/man]
+  --docdir=DIR            documentation root [DATAROOTDIR/doc/grid]
+  --htmldir=DIR           html documentation [DOCDIR]
+  --dvidir=DIR            dvi documentation [DOCDIR]
+  --pdfdir=DIR            pdf documentation [DOCDIR]
+  --psdir=DIR             ps documentation [DOCDIR]
+_ACEOF
+
+  cat <<\_ACEOF
+
+Program names:
+  --program-prefix=PREFIX            prepend PREFIX to installed program names
+  --program-suffix=SUFFIX            append SUFFIX to installed program names
+  --program-transform-name=PROGRAM   run sed PROGRAM on installed program names
+_ACEOF
+fi
+
+if test -n "$ac_init_help"; then
+  case $ac_init_help in
+     short | recursive ) echo "Configuration of Grid 1.0:";;
+   esac
+  cat <<\_ACEOF
+
+Optional Features:
+  --disable-option-checking  ignore unrecognized --enable/--with options
+  --disable-FEATURE       do not include FEATURE (same as --enable-FEATURE=no)
+  --enable-FEATURE[=ARG]  include FEATURE [ARG=yes]
+  --enable-silent-rules   less verbose build output (undo: "make V=1")
+  --disable-silent-rules  verbose build output (undo: "make V=0")
+  --enable-dependency-tracking
+                          do not reject slow dependency extractors
+  --disable-dependency-tracking
+                          speeds up one-time build
+  --disable-openmp        do not use OpenMP
+  --enable-simd=SSE|AVX|AVX2|AVX512
+                          Select instructions
+
+Some influential environment variables:
+  CXX         C++ compiler command
+  CXXFLAGS    C++ compiler flags
+  LDFLAGS     linker flags, e.g. -L<lib dir> if you have libraries in a
+              nonstandard directory <lib dir>
+  LIBS        libraries to pass to the linker, e.g. -l<library>
+  CPPFLAGS    (Objective) C/C++ preprocessor flags, e.g. -I<include dir> if
+              you have headers in a nonstandard directory <include dir>
+  CC          C compiler command
+  CFLAGS      C compiler flags
+  CPP         C preprocessor
+
+Use these variables to override the choices made by `configure' or to help
+it to find libraries and programs with nonstandard names/locations.
+
+Report bugs to <paboyle@ph.ed.ac.uk>.
+_ACEOF
+ac_status=$?
+fi
+
+if test "$ac_init_help" = "recursive"; then
+  # If there are subdirs, report their specific --help.
+  for ac_dir in : $ac_subdirs_all; do test "x$ac_dir" = x: && continue
+    test -d "$ac_dir" ||
+      { cd "$srcdir" && ac_pwd=`pwd` && srcdir=. && test -d "$ac_dir"; } ||
+      continue
+    ac_builddir=.
+
+case "$ac_dir" in
+.) ac_dir_suffix= ac_top_builddir_sub=. ac_top_build_prefix= ;;
+*)
+  ac_dir_suffix=/`$as_echo "$ac_dir" | sed 's|^\.[\\/]||'`
+  # A ".." for each directory in $ac_dir_suffix.
+  ac_top_builddir_sub=`$as_echo "$ac_dir_suffix" | sed 's|/[^\\/]*|/..|g;s|/||'`
+  case $ac_top_builddir_sub in
+  "") ac_top_builddir_sub=. ac_top_build_prefix= ;;
+  *)  ac_top_build_prefix=$ac_top_builddir_sub/ ;;
+  esac ;;
+esac
+ac_abs_top_builddir=$ac_pwd
+ac_abs_builddir=$ac_pwd$ac_dir_suffix
+# for backward compatibility:
+ac_top_builddir=$ac_top_build_prefix
+
+case $srcdir in
+  .)  # We are building in place.
+    ac_srcdir=.
+    ac_top_srcdir=$ac_top_builddir_sub
+    ac_abs_top_srcdir=$ac_pwd ;;
+  [\\/]* | ?:[\\/]* )  # Absolute name.
+    ac_srcdir=$srcdir$ac_dir_suffix;
+    ac_top_srcdir=$srcdir
+    ac_abs_top_srcdir=$srcdir ;;
+  *) # Relative name.
+    ac_srcdir=$ac_top_build_prefix$srcdir$ac_dir_suffix
+    ac_top_srcdir=$ac_top_build_prefix$srcdir
+    ac_abs_top_srcdir=$ac_pwd/$srcdir ;;
+esac
+ac_abs_srcdir=$ac_abs_top_srcdir$ac_dir_suffix
+
+    cd "$ac_dir" || { ac_status=$?; continue; }
+    # Check for guested configure.
+    if test -f "$ac_srcdir/configure.gnu"; then
+      echo &&
+      $SHELL "$ac_srcdir/configure.gnu" --help=recursive
+    elif test -f "$ac_srcdir/configure"; then
+      echo &&
+      $SHELL "$ac_srcdir/configure" --help=recursive
+    else
+      $as_echo "$as_me: WARNING: no configuration information is in $ac_dir" >&2
+    fi || ac_status=$?
+    cd "$ac_pwd" || { ac_status=$?; break; }
+  done
+fi
+
+test -n "$ac_init_help" && exit $ac_status
+if $ac_init_version; then
+  cat <<\_ACEOF
+Grid configure 1.0
+generated by GNU Autoconf 2.69
+
+Copyright (C) 2012 Free Software Foundation, Inc.
+This configure script is free software; the Free Software Foundation
+gives unlimited permission to copy, distribute and modify it.
+_ACEOF
+  exit
+fi
+
+## ------------------------ ##
+## Autoconf initialization. ##
+## ------------------------ ##
+
+# ac_fn_cxx_try_compile LINENO
+# ----------------------------
+# Try to compile conftest.$ac_ext, and return whether this succeeded.
+ac_fn_cxx_try_compile ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  rm -f conftest.$ac_objext
+  if { { ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compile") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_retval=1
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_cxx_try_compile
+
+# ac_fn_c_try_compile LINENO
+# --------------------------
+# Try to compile conftest.$ac_ext, and return whether this succeeded.
+ac_fn_c_try_compile ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  rm -f conftest.$ac_objext
+  if { { ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compile") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_retval=1
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_compile
+
+# ac_fn_c_try_link LINENO
+# -----------------------
+# Try to link conftest.$ac_ext, and return whether this succeeded.
+ac_fn_c_try_link ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  rm -f conftest.$ac_objext conftest$ac_exeext
+  if { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 test -x conftest$ac_exeext
+       }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_retval=1
+fi
+  # Delete the IPA/IPO (Inter Procedural Analysis/Optimization) information
+  # created by the PGI compiler (conftest_ipa8_conftest.oo), as it would
+  # interfere with the next link command; also delete a directory that is
+  # left behind by Apple's compiler.  We do this before executing the actions.
+  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_link
+
+# ac_fn_c_try_cpp LINENO
+# ----------------------
+# Try to preprocess conftest.$ac_ext, and return whether this succeeded.
+ac_fn_c_try_cpp ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if { { ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } > conftest.i && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+    ac_retval=1
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_cpp
+
+# ac_fn_c_check_header_mongrel LINENO HEADER VAR INCLUDES
+# -------------------------------------------------------
+# Tests whether HEADER exists, giving a warning if it cannot be compiled using
+# the include files in INCLUDES and setting the cache variable VAR
+# accordingly.
+ac_fn_c_check_header_mongrel ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if eval \${$3+:} false; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 usability" >&5
+$as_echo_n "checking $2 usability... " >&6; }
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+#include <$2>
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_header_compiler=yes
+else
+  ac_header_compiler=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 presence" >&5
+$as_echo_n "checking $2 presence... " >&6; }
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <$2>
+_ACEOF
+if ac_fn_c_try_cpp "$LINENO"; then :
+  ac_header_preproc=yes
+else
+  ac_header_preproc=no
+fi
+rm -f conftest.err conftest.i conftest.$ac_ext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in #((
+  yes:no: )
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $2: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $2:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $2: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
+( $as_echo "## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##"
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  eval "$3=\$ac_header_compiler"
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_header_mongrel
+
+# ac_fn_c_try_run LINENO
+# ----------------------
+# Try to link conftest.$ac_ext, and return whether this succeeded. Assumes
+# that executables *can* be run.
+ac_fn_c_try_run ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && { ac_try='./conftest$ac_exeext'
+  { { case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: program exited with status $ac_status" >&5
+       $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+       ac_retval=$ac_status
+fi
+  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_run
+
+# ac_fn_c_check_header_compile LINENO HEADER VAR INCLUDES
+# -------------------------------------------------------
+# Tests whether HEADER exists and can be compiled using the include files in
+# INCLUDES, setting the cache variable VAR accordingly.
+ac_fn_c_check_header_compile ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+#include <$2>
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  eval "$3=yes"
+else
+  eval "$3=no"
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_header_compile
+
+# ac_fn_c_check_type LINENO TYPE VAR INCLUDES
+# -------------------------------------------
+# Tests whether TYPE exists after having included INCLUDES, setting cache
+# variable VAR accordingly.
+ac_fn_c_check_type ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  eval "$3=no"
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+if (sizeof ($2))
+	 return 0;
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+if (sizeof (($2)))
+	    return 0;
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+
+else
+  eval "$3=yes"
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_type
+
+# ac_fn_c_find_uintX_t LINENO BITS VAR
+# ------------------------------------
+# Finds an unsigned integer type with width BITS, setting cache variable VAR
+# accordingly.
+ac_fn_c_find_uintX_t ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for uint$2_t" >&5
+$as_echo_n "checking for uint$2_t... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  eval "$3=no"
+     # Order is important - never check a type that is potentially smaller
+     # than half of the expected target width.
+     for ac_type in uint$2_t 'unsigned int' 'unsigned long int' \
+	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
+       cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+static int test_array [1 - 2 * !((($ac_type) -1 >> ($2 / 2 - 1)) >> ($2 / 2 - 1) == 3)];
+test_array [0] = 0;
+return test_array [0];
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  case $ac_type in #(
+  uint$2_t) :
+    eval "$3=yes" ;; #(
+  *) :
+    eval "$3=\$ac_type" ;;
+esac
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+       if eval test \"x\$"$3"\" = x"no"; then :
+
+else
+  break
+fi
+     done
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_find_uintX_t
+
+# ac_fn_c_check_func LINENO FUNC VAR
+# ----------------------------------
+# Tests whether FUNC exists, setting the cache variable VAR accordingly
+ac_fn_c_check_func ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+/* Define $2 to an innocuous variant, in case <limits.h> declares $2.
+   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
+#define $2 innocuous_$2
+
+/* System header to define __stub macros and hopefully few prototypes,
+    which can conflict with char $2 (); below.
+    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
+    <limits.h> exists even on freestanding compilers.  */
+
+#ifdef __STDC__
+# include <limits.h>
+#else
+# include <assert.h>
+#endif
+
+#undef $2
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char $2 ();
+/* The GNU C library defines this for functions which it implements
+    to always fail with ENOSYS.  Some functions are actually named
+    something starting with __ and the normal name is an alias.  */
+#if defined __stub_$2 || defined __stub___$2
+choke me
+#endif
+
+int
+main ()
+{
+return $2 ();
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  eval "$3=yes"
+else
+  eval "$3=no"
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_func
+cat >config.log <<_ACEOF
+This file contains any messages produced by compilers while
+running configure, to aid debugging if configure makes a mistake.
+
+It was created by Grid $as_me 1.0, which was
+generated by GNU Autoconf 2.69.  Invocation command line was
+
+  $ $0 $@
+
+_ACEOF
+exec 5>>config.log
+{
+cat <<_ASUNAME
+## --------- ##
+## Platform. ##
+## --------- ##
+
+hostname = `(hostname || uname -n) 2>/dev/null | sed 1q`
+uname -m = `(uname -m) 2>/dev/null || echo unknown`
+uname -r = `(uname -r) 2>/dev/null || echo unknown`
+uname -s = `(uname -s) 2>/dev/null || echo unknown`
+uname -v = `(uname -v) 2>/dev/null || echo unknown`
+
+/usr/bin/uname -p = `(/usr/bin/uname -p) 2>/dev/null || echo unknown`
+/bin/uname -X     = `(/bin/uname -X) 2>/dev/null     || echo unknown`
+
+/bin/arch              = `(/bin/arch) 2>/dev/null              || echo unknown`
+/usr/bin/arch -k       = `(/usr/bin/arch -k) 2>/dev/null       || echo unknown`
+/usr/convex/getsysinfo = `(/usr/convex/getsysinfo) 2>/dev/null || echo unknown`
+/usr/bin/hostinfo      = `(/usr/bin/hostinfo) 2>/dev/null      || echo unknown`
+/bin/machine           = `(/bin/machine) 2>/dev/null           || echo unknown`
+/usr/bin/oslevel       = `(/usr/bin/oslevel) 2>/dev/null       || echo unknown`
+/bin/universe          = `(/bin/universe) 2>/dev/null          || echo unknown`
+
+_ASUNAME
+
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    $as_echo "PATH: $as_dir"
+  done
+IFS=$as_save_IFS
+
+} >&5
+
+cat >&5 <<_ACEOF
+
+
+## ----------- ##
+## Core tests. ##
+## ----------- ##
+
+_ACEOF
+
+
+# Keep a trace of the command line.
+# Strip out --no-create and --no-recursion so they do not pile up.
+# Strip out --silent because we don't want to record it for future runs.
+# Also quote any args containing shell meta-characters.
+# Make two passes to allow for proper duplicate-argument suppression.
+ac_configure_args=
+ac_configure_args0=
+ac_configure_args1=
+ac_must_keep_next=false
+for ac_pass in 1 2
+do
+  for ac_arg
+  do
+    case $ac_arg in
+    -no-create | --no-c* | -n | -no-recursion | --no-r*) continue ;;
+    -q | -quiet | --quiet | --quie | --qui | --qu | --q \
+    | -silent | --silent | --silen | --sile | --sil)
+      continue ;;
+    *\'*)
+      ac_arg=`$as_echo "$ac_arg" | sed "s/'/'\\\\\\\\''/g"` ;;
+    esac
+    case $ac_pass in
+    1) as_fn_append ac_configure_args0 " '$ac_arg'" ;;
+    2)
+      as_fn_append ac_configure_args1 " '$ac_arg'"
+      if test $ac_must_keep_next = true; then
+	ac_must_keep_next=false # Got value, back to normal.
+      else
+	case $ac_arg in
+	  *=* | --config-cache | -C | -disable-* | --disable-* \
+	  | -enable-* | --enable-* | -gas | --g* | -nfp | --nf* \
+	  | -q | -quiet | --q* | -silent | --sil* | -v | -verb* \
+	  | -with-* | --with-* | -without-* | --without-* | --x)
+	    case "$ac_configure_args0 " in
+	      "$ac_configure_args1"*" '$ac_arg' "* ) continue ;;
+	    esac
+	    ;;
+	  -* ) ac_must_keep_next=true ;;
+	esac
+      fi
+      as_fn_append ac_configure_args " '$ac_arg'"
+      ;;
+    esac
+  done
+done
+{ ac_configure_args0=; unset ac_configure_args0;}
+{ ac_configure_args1=; unset ac_configure_args1;}
+
+# When interrupted or exit'd, cleanup temporary files, and complete
+# config.log.  We remove comments because anyway the quotes in there
+# would cause problems or look ugly.
+# WARNING: Use '\'' to represent an apostrophe within the trap.
+# WARNING: Do not start the trap code with a newline, due to a FreeBSD 4.0 bug.
+trap 'exit_status=$?
+  # Save into config.log some information that might help in debugging.
+  {
+    echo
+
+    $as_echo "## ---------------- ##
+## Cache variables. ##
+## ---------------- ##"
+    echo
+    # The following way of writing the cache mishandles newlines in values,
+(
+  for ac_var in `(set) 2>&1 | sed -n '\''s/^\([a-zA-Z_][a-zA-Z0-9_]*\)=.*/\1/p'\''`; do
+    eval ac_val=\$$ac_var
+    case $ac_val in #(
+    *${as_nl}*)
+      case $ac_var in #(
+      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
+$as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
+      esac
+      case $ac_var in #(
+      _ | IFS | as_nl) ;; #(
+      BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
+      *) { eval $ac_var=; unset $ac_var;} ;;
+      esac ;;
+    esac
+  done
+  (set) 2>&1 |
+    case $as_nl`(ac_space='\'' '\''; set) 2>&1` in #(
+    *${as_nl}ac_space=\ *)
+      sed -n \
+	"s/'\''/'\''\\\\'\'''\''/g;
+	  s/^\\([_$as_cr_alnum]*_cv_[_$as_cr_alnum]*\\)=\\(.*\\)/\\1='\''\\2'\''/p"
+      ;; #(
+    *)
+      sed -n "/^[_$as_cr_alnum]*_cv_[_$as_cr_alnum]*=/p"
+      ;;
+    esac |
+    sort
+)
+    echo
+
+    $as_echo "## ----------------- ##
+## Output variables. ##
+## ----------------- ##"
+    echo
+    for ac_var in $ac_subst_vars
+    do
+      eval ac_val=\$$ac_var
+      case $ac_val in
+      *\'\''*) ac_val=`$as_echo "$ac_val" | sed "s/'\''/'\''\\\\\\\\'\'''\''/g"`;;
+      esac
+      $as_echo "$ac_var='\''$ac_val'\''"
+    done | sort
+    echo
+
+    if test -n "$ac_subst_files"; then
+      $as_echo "## ------------------- ##
+## File substitutions. ##
+## ------------------- ##"
+      echo
+      for ac_var in $ac_subst_files
+      do
+	eval ac_val=\$$ac_var
+	case $ac_val in
+	*\'\''*) ac_val=`$as_echo "$ac_val" | sed "s/'\''/'\''\\\\\\\\'\'''\''/g"`;;
+	esac
+	$as_echo "$ac_var='\''$ac_val'\''"
+      done | sort
+      echo
+    fi
+
+    if test -s confdefs.h; then
+      $as_echo "## ----------- ##
+## confdefs.h. ##
+## ----------- ##"
+      echo
+      cat confdefs.h
+      echo
+    fi
+    test "$ac_signal" != 0 &&
+      $as_echo "$as_me: caught signal $ac_signal"
+    $as_echo "$as_me: exit $exit_status"
+  } >&5
+  rm -f core *.core core.conftest.* &&
+    rm -f -r conftest* confdefs* conf$$* $ac_clean_files &&
+    exit $exit_status
+' 0
+for ac_signal in 1 2 13 15; do
+  trap 'ac_signal='$ac_signal'; as_fn_exit 1' $ac_signal
+done
+ac_signal=0
+
+# confdefs.h avoids OS command line length limits that DEFS can exceed.
+rm -f -r conftest* confdefs.h
+
+$as_echo "/* confdefs.h */" > confdefs.h
+
+# Predefined preprocessor variables.
+
+cat >>confdefs.h <<_ACEOF
+#define PACKAGE_NAME "$PACKAGE_NAME"
+_ACEOF
+
+cat >>confdefs.h <<_ACEOF
+#define PACKAGE_TARNAME "$PACKAGE_TARNAME"
+_ACEOF
+
+cat >>confdefs.h <<_ACEOF
+#define PACKAGE_VERSION "$PACKAGE_VERSION"
+_ACEOF
+
+cat >>confdefs.h <<_ACEOF
+#define PACKAGE_STRING "$PACKAGE_STRING"
+_ACEOF
+
+cat >>confdefs.h <<_ACEOF
+#define PACKAGE_BUGREPORT "$PACKAGE_BUGREPORT"
+_ACEOF
+
+cat >>confdefs.h <<_ACEOF
+#define PACKAGE_URL "$PACKAGE_URL"
+_ACEOF
+
+
+# Let the site file select an alternate cache file if it wants to.
+# Prefer an explicitly selected file to automatically selected ones.
+ac_site_file1=NONE
+ac_site_file2=NONE
+if test -n "$CONFIG_SITE"; then
+  # We do not want a PATH search for config.site.
+  case $CONFIG_SITE in #((
+    -*)  ac_site_file1=./$CONFIG_SITE;;
+    */*) ac_site_file1=$CONFIG_SITE;;
+    *)   ac_site_file1=./$CONFIG_SITE;;
+  esac
+elif test "x$prefix" != xNONE; then
+  ac_site_file1=$prefix/share/config.site
+  ac_site_file2=$prefix/etc/config.site
+else
+  ac_site_file1=$ac_default_prefix/share/config.site
+  ac_site_file2=$ac_default_prefix/etc/config.site
+fi
+for ac_site_file in "$ac_site_file1" "$ac_site_file2"
+do
+  test "x$ac_site_file" = xNONE && continue
+  if test /dev/null != "$ac_site_file" && test -r "$ac_site_file"; then
+    { $as_echo "$as_me:${as_lineno-$LINENO}: loading site script $ac_site_file" >&5
+$as_echo "$as_me: loading site script $ac_site_file" >&6;}
+    sed 's/^/| /' "$ac_site_file" >&5
+    . "$ac_site_file" \
+      || { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "failed to load site script $ac_site_file
+See \`config.log' for more details" "$LINENO" 5; }
+  fi
+done
+
+if test -r "$cache_file"; then
+  # Some versions of bash will fail to source /dev/null (special files
+  # actually), so we avoid doing that.  DJGPP emulates it as a regular file.
+  if test /dev/null != "$cache_file" && test -f "$cache_file"; then
+    { $as_echo "$as_me:${as_lineno-$LINENO}: loading cache $cache_file" >&5
+$as_echo "$as_me: loading cache $cache_file" >&6;}
+    case $cache_file in
+      [\\/]* | ?:[\\/]* ) . "$cache_file";;
+      *)                      . "./$cache_file";;
+    esac
+  fi
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: creating cache $cache_file" >&5
+$as_echo "$as_me: creating cache $cache_file" >&6;}
+  >$cache_file
+fi
+
+# Check that the precious variables saved in the cache have kept the same
+# value.
+ac_cache_corrupted=false
+for ac_var in $ac_precious_vars; do
+  eval ac_old_set=\$ac_cv_env_${ac_var}_set
+  eval ac_new_set=\$ac_env_${ac_var}_set
+  eval ac_old_val=\$ac_cv_env_${ac_var}_value
+  eval ac_new_val=\$ac_env_${ac_var}_value
+  case $ac_old_set,$ac_new_set in
+    set,)
+      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
+$as_echo "$as_me: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&2;}
+      ac_cache_corrupted=: ;;
+    ,set)
+      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was not set in the previous run" >&5
+$as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
+      ac_cache_corrupted=: ;;
+    ,);;
+    *)
+      if test "x$ac_old_val" != "x$ac_new_val"; then
+	# differences in whitespace do not lead to failure.
+	ac_old_val_w=`echo x $ac_old_val`
+	ac_new_val_w=`echo x $ac_new_val`
+	if test "$ac_old_val_w" != "$ac_new_val_w"; then
+	  { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' has changed since the previous run:" >&5
+$as_echo "$as_me: error: \`$ac_var' has changed since the previous run:" >&2;}
+	  ac_cache_corrupted=:
+	else
+	  { $as_echo "$as_me:${as_lineno-$LINENO}: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
+$as_echo "$as_me: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&2;}
+	  eval $ac_var=\$ac_old_val
+	fi
+	{ $as_echo "$as_me:${as_lineno-$LINENO}:   former value:  \`$ac_old_val'" >&5
+$as_echo "$as_me:   former value:  \`$ac_old_val'" >&2;}
+	{ $as_echo "$as_me:${as_lineno-$LINENO}:   current value: \`$ac_new_val'" >&5
+$as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
+      fi;;
+  esac
+  # Pass precious variables to config.status.
+  if test "$ac_new_set" = set; then
+    case $ac_new_val in
+    *\'*) ac_arg=$ac_var=`$as_echo "$ac_new_val" | sed "s/'/'\\\\\\\\''/g"` ;;
+    *) ac_arg=$ac_var=$ac_new_val ;;
+    esac
+    case " $ac_configure_args " in
+      *" '$ac_arg' "*) ;; # Avoid dups.  Use of quotes ensures accuracy.
+      *) as_fn_append ac_configure_args " '$ac_arg'" ;;
+    esac
+  fi
+done
+if $ac_cache_corrupted; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+  { $as_echo "$as_me:${as_lineno-$LINENO}: error: changes in the environment can compromise the build" >&5
+$as_echo "$as_me: error: changes in the environment can compromise the build" >&2;}
+  as_fn_error $? "run \`make distclean' and/or \`rm $cache_file' and start over" "$LINENO" 5
+fi
+## -------------------- ##
+## Main body of script. ##
+## -------------------- ##
+
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+
+
+am__api_version='1.15'
+
+ac_aux_dir=
+for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
+  if test -f "$ac_dir/install-sh"; then
+    ac_aux_dir=$ac_dir
+    ac_install_sh="$ac_aux_dir/install-sh -c"
+    break
+  elif test -f "$ac_dir/install.sh"; then
+    ac_aux_dir=$ac_dir
+    ac_install_sh="$ac_aux_dir/install.sh -c"
+    break
+  elif test -f "$ac_dir/shtool"; then
+    ac_aux_dir=$ac_dir
+    ac_install_sh="$ac_aux_dir/shtool install -c"
+    break
+  fi
+done
+if test -z "$ac_aux_dir"; then
+  as_fn_error $? "cannot find install-sh, install.sh, or shtool in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" "$LINENO" 5
+fi
+
+# These three variables are undocumented and unsupported,
+# and are intended to be withdrawn in a future Autoconf release.
+# They can cause serious problems if a builder's source tree is in a directory
+# whose full name contains unusual characters.
+ac_config_guess="$SHELL $ac_aux_dir/config.guess"  # Please don't use this var.
+ac_config_sub="$SHELL $ac_aux_dir/config.sub"  # Please don't use this var.
+ac_configure="$SHELL $ac_aux_dir/configure"  # Please don't use this var.
+
+
+# Find a good install program.  We prefer a C program (faster),
+# so one script is as good as another.  But avoid the broken or
+# incompatible versions:
+# SysV /etc/install, /usr/sbin/install
+# SunOS /usr/etc/install
+# IRIX /sbin/install
+# AIX /bin/install
+# AmigaOS /C/install, which installs bootblocks on floppy discs
+# AIX 4 /usr/bin/installbsd, which doesn't work without a -g flag
+# AFS /usr/afsws/bin/install, which mishandles nonexistent args
+# SVR4 /usr/ucb/install, which tries to use the nonexistent group "staff"
+# OS/2's system install, which has a completely different semantic
+# ./install, which can be erroneously created by make from ./install.sh.
+# Reject install programs that cannot install multiple files.
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a BSD-compatible install" >&5
+$as_echo_n "checking for a BSD-compatible install... " >&6; }
+if test -z "$INSTALL"; then
+if ${ac_cv_path_install+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    # Account for people who put trailing slashes in PATH elements.
+case $as_dir/ in #((
+  ./ | .// | /[cC]/* | \
+  /etc/* | /usr/sbin/* | /usr/etc/* | /sbin/* | /usr/afsws/bin/* | \
+  ?:[\\/]os2[\\/]install[\\/]* | ?:[\\/]OS2[\\/]INSTALL[\\/]* | \
+  /usr/ucb/* ) ;;
+  *)
+    # OSF1 and SCO ODT 3.0 have their own names for install.
+    # Don't use installbsd from OSF since it installs stuff as root
+    # by default.
+    for ac_prog in ginstall scoinst install; do
+      for ac_exec_ext in '' $ac_executable_extensions; do
+	if as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext"; then
+	  if test $ac_prog = install &&
+	    grep dspmsg "$as_dir/$ac_prog$ac_exec_ext" >/dev/null 2>&1; then
+	    # AIX install.  It has an incompatible calling convention.
+	    :
+	  elif test $ac_prog = install &&
+	    grep pwplus "$as_dir/$ac_prog$ac_exec_ext" >/dev/null 2>&1; then
+	    # program-specific install script used by HP pwplus--don't use.
+	    :
+	  else
+	    rm -rf conftest.one conftest.two conftest.dir
+	    echo one > conftest.one
+	    echo two > conftest.two
+	    mkdir conftest.dir
+	    if "$as_dir/$ac_prog$ac_exec_ext" -c conftest.one conftest.two "`pwd`/conftest.dir" &&
+	      test -s conftest.one && test -s conftest.two &&
+	      test -s conftest.dir/conftest.one &&
+	      test -s conftest.dir/conftest.two
+	    then
+	      ac_cv_path_install="$as_dir/$ac_prog$ac_exec_ext -c"
+	      break 3
+	    fi
+	  fi
+	fi
+      done
+    done
+    ;;
+esac
+
+  done
+IFS=$as_save_IFS
+
+rm -rf conftest.one conftest.two conftest.dir
+
+fi
+  if test "${ac_cv_path_install+set}" = set; then
+    INSTALL=$ac_cv_path_install
+  else
+    # As a last resort, use the slow shell script.  Don't cache a
+    # value for INSTALL within a source directory, because that will
+    # break other packages using the cache if that directory is
+    # removed, or if the value is a relative name.
+    INSTALL=$ac_install_sh
+  fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $INSTALL" >&5
+$as_echo "$INSTALL" >&6; }
+
+# Use test -z because SunOS4 sh mishandles braces in ${var-val}.
+# It thinks the first close brace ends the variable substitution.
+test -z "$INSTALL_PROGRAM" && INSTALL_PROGRAM='${INSTALL}'
+
+test -z "$INSTALL_SCRIPT" && INSTALL_SCRIPT='${INSTALL}'
+
+test -z "$INSTALL_DATA" && INSTALL_DATA='${INSTALL} -m 644'
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether build environment is sane" >&5
+$as_echo_n "checking whether build environment is sane... " >&6; }
+# Reject unsafe characters in $srcdir or the absolute working directory
+# name.  Accept space and tab only in the latter.
+am_lf='
+'
+case `pwd` in
+  *[\\\"\#\$\&\'\`$am_lf]*)
+    as_fn_error $? "unsafe absolute working directory name" "$LINENO" 5;;
+esac
+case $srcdir in
+  *[\\\"\#\$\&\'\`$am_lf\ \	]*)
+    as_fn_error $? "unsafe srcdir value: '$srcdir'" "$LINENO" 5;;
+esac
+
+# Do 'set' in a subshell so we don't clobber the current shell's
+# arguments.  Must try -L first in case configure is actually a
+# symlink; some systems play weird games with the mod time of symlinks
+# (eg FreeBSD returns the mod time of the symlink's containing
+# directory).
+if (
+   am_has_slept=no
+   for am_try in 1 2; do
+     echo "timestamp, slept: $am_has_slept" > conftest.file
+     set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
+     if test "$*" = "X"; then
+	# -L didn't work.
+	set X `ls -t "$srcdir/configure" conftest.file`
+     fi
+     if test "$*" != "X $srcdir/configure conftest.file" \
+	&& test "$*" != "X conftest.file $srcdir/configure"; then
+
+	# If neither matched, then we have a broken ls.  This can happen
+	# if, for instance, CONFIG_SHELL is bash and it inherits a
+	# broken ls alias from the environment.  This has actually
+	# happened.  Such a system could not be considered "sane".
+	as_fn_error $? "ls -t appears to fail.  Make sure there is not a broken
+  alias in your environment" "$LINENO" 5
+     fi
+     if test "$2" = conftest.file || test $am_try -eq 2; then
+       break
+     fi
+     # Just in case.
+     sleep 1
+     am_has_slept=yes
+   done
+   test "$2" = conftest.file
+   )
+then
+   # Ok.
+   :
+else
+   as_fn_error $? "newly created file is older than distributed files!
+Check your system clock" "$LINENO" 5
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+# If we didn't sleep, we still need to ensure time stamps of config.status and
+# generated files are strictly newer.
+am_sleep_pid=
+if grep 'slept: no' conftest.file >/dev/null 2>&1; then
+  ( sleep 1 ) &
+  am_sleep_pid=$!
+fi
+
+rm -f conftest.file
+
+test "$program_prefix" != NONE &&
+  program_transform_name="s&^&$program_prefix&;$program_transform_name"
+# Use a double $ so make ignores it.
+test "$program_suffix" != NONE &&
+  program_transform_name="s&\$&$program_suffix&;$program_transform_name"
+# Double any \ or $.
+# By default was `s,x,x', remove it if useless.
+ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
+program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
+
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
+
+if test x"${MISSING+set}" != xset; then
+  case $am_aux_dir in
+  *\ * | *\	*)
+    MISSING="\${SHELL} \"$am_aux_dir/missing\"" ;;
+  *)
+    MISSING="\${SHELL} $am_aux_dir/missing" ;;
+  esac
+fi
+# Use eval to expand $SHELL
+if eval "$MISSING --is-lightweight"; then
+  am_missing_run="$MISSING "
+else
+  am_missing_run=
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: 'missing' script is too old or missing" >&5
+$as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
+fi
+
+if test x"${install_sh+set}" != xset; then
+  case $am_aux_dir in
+  *\ * | *\	*)
+    install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
+  *)
+    install_sh="\${SHELL} $am_aux_dir/install-sh"
+  esac
+fi
+
+# Installed binaries are usually stripped using 'strip' when the user
+# run "make install-strip".  However 'strip' might not be the right
+# tool to use in cross-compilation environments, therefore Automake
+# will honor the 'STRIP' environment variable to overrule this program.
+if test "$cross_compiling" != no; then
+  if test -n "$ac_tool_prefix"; then
+  # Extract the first word of "${ac_tool_prefix}strip", so it can be a program name with args.
+set dummy ${ac_tool_prefix}strip; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_STRIP+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$STRIP"; then
+  ac_cv_prog_STRIP="$STRIP" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_STRIP="${ac_tool_prefix}strip"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+STRIP=$ac_cv_prog_STRIP
+if test -n "$STRIP"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $STRIP" >&5
+$as_echo "$STRIP" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+fi
+if test -z "$ac_cv_prog_STRIP"; then
+  ac_ct_STRIP=$STRIP
+  # Extract the first word of "strip", so it can be a program name with args.
+set dummy strip; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_ac_ct_STRIP+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$ac_ct_STRIP"; then
+  ac_cv_prog_ac_ct_STRIP="$ac_ct_STRIP" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_ac_ct_STRIP="strip"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+ac_ct_STRIP=$ac_cv_prog_ac_ct_STRIP
+if test -n "$ac_ct_STRIP"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_STRIP" >&5
+$as_echo "$ac_ct_STRIP" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+  if test "x$ac_ct_STRIP" = x; then
+    STRIP=":"
+  else
+    case $cross_compiling:$ac_tool_warned in
+yes:)
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+$as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
+ac_tool_warned=yes ;;
+esac
+    STRIP=$ac_ct_STRIP
+  fi
+else
+  STRIP="$ac_cv_prog_STRIP"
+fi
+
+fi
+INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a thread-safe mkdir -p" >&5
+$as_echo_n "checking for a thread-safe mkdir -p... " >&6; }
+if test -z "$MKDIR_P"; then
+  if ${ac_cv_path_mkdir+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH$PATH_SEPARATOR/opt/sfw/bin
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_prog in mkdir gmkdir; do
+	 for ac_exec_ext in '' $ac_executable_extensions; do
+	   as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext" || continue
+	   case `"$as_dir/$ac_prog$ac_exec_ext" --version 2>&1` in #(
+	     'mkdir (GNU coreutils) '* | \
+	     'mkdir (coreutils) '* | \
+	     'mkdir (fileutils) '4.1*)
+	       ac_cv_path_mkdir=$as_dir/$ac_prog$ac_exec_ext
+	       break 3;;
+	   esac
+	 done
+       done
+  done
+IFS=$as_save_IFS
+
+fi
+
+  test -d ./--version && rmdir ./--version
+  if test "${ac_cv_path_mkdir+set}" = set; then
+    MKDIR_P="$ac_cv_path_mkdir -p"
+  else
+    # As a last resort, use the slow shell script.  Don't cache a
+    # value for MKDIR_P within a source directory, because that will
+    # break other packages using the cache if that directory is
+    # removed, or if the value is a relative name.
+    MKDIR_P="$ac_install_sh -d"
+  fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $MKDIR_P" >&5
+$as_echo "$MKDIR_P" >&6; }
+
+for ac_prog in gawk mawk nawk awk
+do
+  # Extract the first word of "$ac_prog", so it can be a program name with args.
+set dummy $ac_prog; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_AWK+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$AWK"; then
+  ac_cv_prog_AWK="$AWK" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_AWK="$ac_prog"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+AWK=$ac_cv_prog_AWK
+if test -n "$AWK"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $AWK" >&5
+$as_echo "$AWK" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+  test -n "$AWK" && break
+done
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether ${MAKE-make} sets \$(MAKE)" >&5
+$as_echo_n "checking whether ${MAKE-make} sets \$(MAKE)... " >&6; }
+set x ${MAKE-make}
+ac_make=`$as_echo "$2" | sed 's/+/p/g; s/[^a-zA-Z0-9_]/_/g'`
+if eval \${ac_cv_prog_make_${ac_make}_set+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat >conftest.make <<\_ACEOF
+SHELL = /bin/sh
+all:
+	@echo '@@@%%%=$(MAKE)=@@@%%%'
+_ACEOF
+# GNU make sometimes prints "make[1]: Entering ...", which would confuse us.
+case `${MAKE-make} -f conftest.make 2>/dev/null` in
+  *@@@%%%=?*=@@@%%%*)
+    eval ac_cv_prog_make_${ac_make}_set=yes;;
+  *)
+    eval ac_cv_prog_make_${ac_make}_set=no;;
+esac
+rm -f conftest.make
+fi
+if eval test \$ac_cv_prog_make_${ac_make}_set = yes; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+  SET_MAKE=
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+  SET_MAKE="MAKE=${MAKE-make}"
+fi
+
+rm -rf .tst 2>/dev/null
+mkdir .tst 2>/dev/null
+if test -d .tst; then
+  am__leading_dot=.
+else
+  am__leading_dot=_
+fi
+rmdir .tst 2>/dev/null
+
+# Check whether --enable-silent-rules was given.
+if test "${enable_silent_rules+set}" = set; then :
+  enableval=$enable_silent_rules;
+fi
+
+case $enable_silent_rules in # (((
+  yes) AM_DEFAULT_VERBOSITY=0;;
+   no) AM_DEFAULT_VERBOSITY=1;;
+    *) AM_DEFAULT_VERBOSITY=1;;
+esac
+am_make=${MAKE-make}
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $am_make supports nested variables" >&5
+$as_echo_n "checking whether $am_make supports nested variables... " >&6; }
+if ${am_cv_make_support_nested_variables+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if $as_echo 'TRUE=$(BAR$(V))
+BAR0=false
+BAR1=true
+V=1
+am__doit:
+	@$(TRUE)
+.PHONY: am__doit' | $am_make -f - >/dev/null 2>&1; then
+  am_cv_make_support_nested_variables=yes
+else
+  am_cv_make_support_nested_variables=no
+fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_make_support_nested_variables" >&5
+$as_echo "$am_cv_make_support_nested_variables" >&6; }
+if test $am_cv_make_support_nested_variables = yes; then
+    AM_V='$(V)'
+  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
+else
+  AM_V=$AM_DEFAULT_VERBOSITY
+  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
+fi
+AM_BACKSLASH='\'
+
+if test "`cd $srcdir && pwd`" != "`pwd`"; then
+  # Use -I$(srcdir) only when $(srcdir) != ., so that make's output
+  # is not polluted with repeated "-I."
+  am__isrc=' -I$(srcdir)'
+  # test to see if srcdir already configured
+  if test -f $srcdir/config.status; then
+    as_fn_error $? "source directory already configured; run \"make distclean\" there first" "$LINENO" 5
+  fi
+fi
+
+# test whether we have cygpath
+if test -z "$CYGPATH_W"; then
+  if (cygpath --version) >/dev/null 2>/dev/null; then
+    CYGPATH_W='cygpath -w'
+  else
+    CYGPATH_W=echo
+  fi
+fi
+
+
+# Define the identity of the package.
+ PACKAGE='grid'
+ VERSION='1.0'
+
+
+cat >>confdefs.h <<_ACEOF
+#define PACKAGE "$PACKAGE"
+_ACEOF
+
+
+cat >>confdefs.h <<_ACEOF
+#define VERSION "$VERSION"
+_ACEOF
+
+# Some tools Automake needs.
+
+ACLOCAL=${ACLOCAL-"${am_missing_run}aclocal-${am__api_version}"}
+
+
+AUTOCONF=${AUTOCONF-"${am_missing_run}autoconf"}
+
+
+AUTOMAKE=${AUTOMAKE-"${am_missing_run}automake-${am__api_version}"}
+
+
+AUTOHEADER=${AUTOHEADER-"${am_missing_run}autoheader"}
+
+
+MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
+
+# For better backward compatibility.  To be removed once Automake 1.9.x
+# dies out for good.  For more background, see:
+# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
+# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
+mkdir_p='$(MKDIR_P)'
+
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
+# Always define AMTAR for backward compatibility.  Yes, it's still used
+# in the wild :-(  We should find a proper way to deprecate it ...
+AMTAR='$${TAR-tar}'
+
+
+# We'll loop over all known methods to create a tar archive until one works.
+_am_tools='gnutar  pax cpio none'
+
+am__tar='$${TAR-tar} chof - "$$tardir"' am__untar='$${TAR-tar} xf -'
+
+
+
+
+
+
+# POSIX will say in a future version that running "rm -f" with no argument
+# is OK; and we want to be able to make that assumption in our Makefile
+# recipes.  So use an aggressive probe to check that the usage we want is
+# actually supported "in the wild" to an acceptable degree.
+# See automake bug#10828.
+# To make any issue more visible, cause the running configure to be aborted
+# by default if the 'rm' program in use doesn't match our expectations; the
+# user can still override this though.
+if rm -f && rm -fr && rm -rf; then : OK; else
+  cat >&2 <<'END'
+Oops!
+
+Your 'rm' program seems unable to run without file operands specified
+on the command line, even when the '-f' option is present.  This is contrary
+to the behaviour of most rm programs out there, and not conforming with
+the upcoming POSIX standard: <http://austingroupbugs.net/view.php?id=542>
+
+Please tell bug-automake@gnu.org about your system, including the value
+of your $PATH and any error possibly output before this message.  This
+can help us improve future automake versions.
+
+END
+  if test x"$ACCEPT_INFERIOR_RM_PROGRAM" = x"yes"; then
+    echo 'Configuration will proceed anyway, since you have set the' >&2
+    echo 'ACCEPT_INFERIOR_RM_PROGRAM variable to "yes"' >&2
+    echo >&2
+  else
+    cat >&2 <<'END'
+Aborting the configuration process, to ensure you take notice of the issue.
+
+You can download and install GNU coreutils to get an 'rm' implementation
+that behaves properly: <http://www.gnu.org/software/coreutils/>.
+
+If you want to complete the configuration process using your problematic
+'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
+to "yes", and re-run configure.
+
+END
+    as_fn_error $? "Your 'rm' program is bad, sorry." "$LINENO" 5
+  fi
+fi
+
+
+
+ac_config_headers="$ac_config_headers Grid_config.h"
+
+
+# Checks for programs.
+ac_ext=cpp
+ac_cpp='$CXXCPP $CPPFLAGS'
+ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
+if test -z "$CXX"; then
+  if test -n "$CCC"; then
+    CXX=$CCC
+  else
+    if test -n "$ac_tool_prefix"; then
+  for ac_prog in g++ c++ gpp aCC CC cxx cc++ cl.exe FCC KCC RCC xlC_r xlC
+  do
+    # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
+set dummy $ac_tool_prefix$ac_prog; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_CXX+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$CXX"; then
+  ac_cv_prog_CXX="$CXX" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_CXX="$ac_tool_prefix$ac_prog"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+CXX=$ac_cv_prog_CXX
+if test -n "$CXX"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CXX" >&5
+$as_echo "$CXX" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+    test -n "$CXX" && break
+  done
+fi
+if test -z "$CXX"; then
+  ac_ct_CXX=$CXX
+  for ac_prog in g++ c++ gpp aCC CC cxx cc++ cl.exe FCC KCC RCC xlC_r xlC
+do
+  # Extract the first word of "$ac_prog", so it can be a program name with args.
+set dummy $ac_prog; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_ac_ct_CXX+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$ac_ct_CXX"; then
+  ac_cv_prog_ac_ct_CXX="$ac_ct_CXX" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_ac_ct_CXX="$ac_prog"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+ac_ct_CXX=$ac_cv_prog_ac_ct_CXX
+if test -n "$ac_ct_CXX"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CXX" >&5
+$as_echo "$ac_ct_CXX" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+  test -n "$ac_ct_CXX" && break
+done
+
+  if test "x$ac_ct_CXX" = x; then
+    CXX="g++"
+  else
+    case $cross_compiling:$ac_tool_warned in
+yes:)
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+$as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
+ac_tool_warned=yes ;;
+esac
+    CXX=$ac_ct_CXX
+  fi
+fi
+
+  fi
+fi
+# Provide some information about the compiler.
+$as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler version" >&5
+set X $ac_compile
+ac_compiler=$2
+for ac_option in --version -v -V -qversion; do
+  { { ac_try="$ac_compiler $ac_option >&5"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    sed '10a\
+... rest of stderr output deleted ...
+         10q' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+  fi
+  rm -f conftest.er1 conftest.err
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }
+done
+
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+ac_clean_files_save=$ac_clean_files
+ac_clean_files="$ac_clean_files a.out a.out.dSYM a.exe b.out"
+# Try to create an executable without -o first, disregard a.out.
+# It will help us diagnose broken compilers, and finding out an intuition
+# of exeext.
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether the C++ compiler works" >&5
+$as_echo_n "checking whether the C++ compiler works... " >&6; }
+ac_link_default=`$as_echo "$ac_link" | sed 's/ -o *conftest[^ ]*//'`
+
+# The possible output files:
+ac_files="a.out conftest.exe conftest a.exe a_out.exe b.out conftest.*"
+
+ac_rmfiles=
+for ac_file in $ac_files
+do
+  case $ac_file in
+    *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | *.map | *.inf | *.dSYM | *.o | *.obj ) ;;
+    * ) ac_rmfiles="$ac_rmfiles $ac_file";;
+  esac
+done
+rm -f $ac_rmfiles
+
+if { { ac_try="$ac_link_default"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link_default") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then :
+  # Autoconf-2.13 could set the ac_cv_exeext variable to `no'.
+# So ignore a value of `no', otherwise this would lead to `EXEEXT = no'
+# in a Makefile.  We should not override ac_cv_exeext if it was cached,
+# so that the user can short-circuit this test for compilers unknown to
+# Autoconf.
+for ac_file in $ac_files ''
+do
+  test -f "$ac_file" || continue
+  case $ac_file in
+    *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | *.map | *.inf | *.dSYM | *.o | *.obj )
+	;;
+    [ab].out )
+	# We found the default executable, but exeext='' is most
+	# certainly right.
+	break;;
+    *.* )
+	if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
+	then :; else
+	   ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
+	fi
+	# We set ac_cv_exeext here because the later test for it is not
+	# safe: cross compilers may not add the suffix if given an `-o'
+	# argument, so we may need to know it at that point already.
+	# Even if this section looks crufty: it has the advantage of
+	# actually working.
+	break;;
+    * )
+	break;;
+  esac
+done
+test "$ac_cv_exeext" = no && ac_cv_exeext=
+
+else
+  ac_file=''
+fi
+if test -z "$ac_file"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+$as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error 77 "C++ compiler cannot create executables
+See \`config.log' for more details" "$LINENO" 5; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler default output file name" >&5
+$as_echo_n "checking for C++ compiler default output file name... " >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_file" >&5
+$as_echo "$ac_file" >&6; }
+ac_exeext=$ac_cv_exeext
+
+rm -f -r a.out a.out.dSYM a.exe conftest$ac_cv_exeext b.out
+ac_clean_files=$ac_clean_files_save
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of executables" >&5
+$as_echo_n "checking for suffix of executables... " >&6; }
+if { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then :
+  # If both `conftest.exe' and `conftest' are `present' (well, observable)
+# catch `conftest.exe'.  For instance with Cygwin, `ls conftest' will
+# work properly (i.e., refer to `conftest.exe'), while it won't with
+# `rm'.
+for ac_file in conftest.exe conftest conftest.*; do
+  test -f "$ac_file" || continue
+  case $ac_file in
+    *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | *.map | *.inf | *.dSYM | *.o | *.obj ) ;;
+    *.* ) ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
+	  break;;
+    * ) break;;
+  esac
+done
+else
+  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "cannot compute suffix of executables: cannot compile and link
+See \`config.log' for more details" "$LINENO" 5; }
+fi
+rm -f conftest conftest$ac_cv_exeext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_exeext" >&5
+$as_echo "$ac_cv_exeext" >&6; }
+
+rm -f conftest.$ac_ext
+EXEEXT=$ac_cv_exeext
+ac_exeext=$EXEEXT
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <stdio.h>
+int
+main ()
+{
+FILE *f = fopen ("conftest.out", "w");
+ return ferror (f) || fclose (f) != 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+ac_clean_files="$ac_clean_files conftest.out"
+# Check that the compiler produces executables we can run.  If not, either
+# the compiler is broken, or we cross compile.
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are cross compiling" >&5
+$as_echo_n "checking whether we are cross compiling... " >&6; }
+if test "$cross_compiling" != yes; then
+  { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }
+  if { ac_try='./conftest$ac_cv_exeext'
+  { { case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; }; then
+    cross_compiling=no
+  else
+    if test "$cross_compiling" = maybe; then
+	cross_compiling=yes
+    else
+	{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "cannot run C++ compiled programs.
+If you meant to cross compile, use \`--host'.
+See \`config.log' for more details" "$LINENO" 5; }
+    fi
+  fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $cross_compiling" >&5
+$as_echo "$cross_compiling" >&6; }
+
+rm -f conftest.$ac_ext conftest$ac_cv_exeext conftest.out
+ac_clean_files=$ac_clean_files_save
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of object files" >&5
+$as_echo_n "checking for suffix of object files... " >&6; }
+if ${ac_cv_objext+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.o conftest.obj
+if { { ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compile") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then :
+  for ac_file in conftest.o conftest.obj conftest.*; do
+  test -f "$ac_file" || continue;
+  case $ac_file in
+    *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | *.map | *.inf | *.dSYM ) ;;
+    *) ac_cv_objext=`expr "$ac_file" : '.*\.\(.*\)'`
+       break;;
+  esac
+done
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "cannot compute suffix of object files: cannot compile
+See \`config.log' for more details" "$LINENO" 5; }
+fi
+rm -f conftest.$ac_cv_objext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_objext" >&5
+$as_echo "$ac_cv_objext" >&6; }
+OBJEXT=$ac_cv_objext
+ac_objext=$OBJEXT
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C++ compiler" >&5
+$as_echo_n "checking whether we are using the GNU C++ compiler... " >&6; }
+if ${ac_cv_cxx_compiler_gnu+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+#ifndef __GNUC__
+       choke me
+#endif
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  ac_compiler_gnu=yes
+else
+  ac_compiler_gnu=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+ac_cv_cxx_compiler_gnu=$ac_compiler_gnu
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_cxx_compiler_gnu" >&5
+$as_echo "$ac_cv_cxx_compiler_gnu" >&6; }
+if test $ac_compiler_gnu = yes; then
+  GXX=yes
+else
+  GXX=
+fi
+ac_test_CXXFLAGS=${CXXFLAGS+set}
+ac_save_CXXFLAGS=$CXXFLAGS
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX accepts -g" >&5
+$as_echo_n "checking whether $CXX accepts -g... " >&6; }
+if ${ac_cv_prog_cxx_g+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_save_cxx_werror_flag=$ac_cxx_werror_flag
+   ac_cxx_werror_flag=yes
+   ac_cv_prog_cxx_g=no
+   CXXFLAGS="-g"
+   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  ac_cv_prog_cxx_g=yes
+else
+  CXXFLAGS=""
+      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+
+else
+  ac_cxx_werror_flag=$ac_save_cxx_werror_flag
+	 CXXFLAGS="-g"
+	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  ac_cv_prog_cxx_g=yes
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+   ac_cxx_werror_flag=$ac_save_cxx_werror_flag
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cxx_g" >&5
+$as_echo "$ac_cv_prog_cxx_g" >&6; }
+if test "$ac_test_CXXFLAGS" = set; then
+  CXXFLAGS=$ac_save_CXXFLAGS
+elif test $ac_cv_prog_cxx_g = yes; then
+  if test "$GXX" = yes; then
+    CXXFLAGS="-g -O2"
+  else
+    CXXFLAGS="-g"
+  fi
+else
+  if test "$GXX" = yes; then
+    CXXFLAGS="-O2"
+  else
+    CXXFLAGS=
+  fi
+fi
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+DEPDIR="${am__leading_dot}deps"
+
+ac_config_commands="$ac_config_commands depfiles"
+
+
+am_make=${MAKE-make}
+cat > confinc << 'END'
+am__doit:
+	@echo this is the am__doit target
+.PHONY: am__doit
+END
+# If we don't find an include directive, just comment out the code.
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for style of include used by $am_make" >&5
+$as_echo_n "checking for style of include used by $am_make... " >&6; }
+am__include="#"
+am__quote=
+_am_result=none
+# First try GNU make style include.
+echo "include confinc" > confmf
+# Ignore all kinds of additional output from 'make'.
+case `$am_make -s -f confmf 2> /dev/null` in #(
+*the\ am__doit\ target*)
+  am__include=include
+  am__quote=
+  _am_result=GNU
+  ;;
+esac
+# Now try BSD make style include.
+if test "$am__include" = "#"; then
+   echo '.include "confinc"' > confmf
+   case `$am_make -s -f confmf 2> /dev/null` in #(
+   *the\ am__doit\ target*)
+     am__include=.include
+     am__quote="\""
+     _am_result=BSD
+     ;;
+   esac
+fi
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $_am_result" >&5
+$as_echo "$_am_result" >&6; }
+rm -f confinc confmf
+
+# Check whether --enable-dependency-tracking was given.
+if test "${enable_dependency_tracking+set}" = set; then :
+  enableval=$enable_dependency_tracking;
+fi
+
+if test "x$enable_dependency_tracking" != xno; then
+  am_depcomp="$ac_aux_dir/depcomp"
+  AMDEPBACKSLASH='\'
+  am__nodep='_no'
+fi
+ if test "x$enable_dependency_tracking" != xno; then
+  AMDEP_TRUE=
+  AMDEP_FALSE='#'
+else
+  AMDEP_TRUE='#'
+  AMDEP_FALSE=
+fi
+
+
+
+depcc="$CXX"  am_compiler_list=
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
+$as_echo_n "checking dependency style of $depcc... " >&6; }
+if ${am_cv_CXX_dependencies_compiler_type+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
+  # We make a subdir and do the tests there.  Otherwise we can end up
+  # making bogus files that we don't know about and never remove.  For
+  # instance it was reported that on HP-UX the gcc test will end up
+  # making a dummy file named 'D' -- because '-MD' means "put the output
+  # in D".
+  rm -rf conftest.dir
+  mkdir conftest.dir
+  # Copy depcomp to subdir because otherwise we won't find it if we're
+  # using a relative directory.
+  cp "$am_depcomp" conftest.dir
+  cd conftest.dir
+  # We will build objects and dependencies in a subdirectory because
+  # it helps to detect inapplicable dependency modes.  For instance
+  # both Tru64's cc and ICC support -MD to output dependencies as a
+  # side effect of compilation, but ICC will put the dependencies in
+  # the current directory while Tru64 will put them in the object
+  # directory.
+  mkdir sub
+
+  am_cv_CXX_dependencies_compiler_type=none
+  if test "$am_compiler_list" = ""; then
+     am_compiler_list=`sed -n 's/^#*\([a-zA-Z0-9]*\))$/\1/p' < ./depcomp`
+  fi
+  am__universal=false
+  case " $depcc " in #(
+     *\ -arch\ *\ -arch\ *) am__universal=true ;;
+     esac
+
+  for depmode in $am_compiler_list; do
+    # Setup a source with many dependencies, because some compilers
+    # like to wrap large dependency lists on column 80 (with \), and
+    # we should not choose a depcomp mode which is confused by this.
+    #
+    # We need to recreate these files for each test, as the compiler may
+    # overwrite some of them when testing with obscure command lines.
+    # This happens at least with the AIX C compiler.
+    : > sub/conftest.c
+    for i in 1 2 3 4 5 6; do
+      echo '#include "conftst'$i'.h"' >> sub/conftest.c
+      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
+      # Solaris 10 /bin/sh.
+      echo '/* dummy */' > sub/conftst$i.h
+    done
+    echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
+
+    # We check with '-c' and '-o' for the sake of the "dashmstdout"
+    # mode.  It turns out that the SunPro C++ compiler does not properly
+    # handle '-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs.
+    am__obj=sub/conftest.${OBJEXT-o}
+    am__minus_obj="-o $am__obj"
+    case $depmode in
+    gcc)
+      # This depmode causes a compiler race in universal mode.
+      test "$am__universal" = false || continue
+      ;;
+    nosideeffect)
+      # After this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested.
+      if test "x$enable_dependency_tracking" = xyes; then
+	continue
+      else
+	break
+      fi
+      ;;
+    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
+      # This compiler won't grok '-c -o', but also, the minuso test has
+      # not run yet.  These depmodes are late enough in the game, and
+      # so weak that their functioning should not be impacted.
+      am__obj=conftest.${OBJEXT-o}
+      am__minus_obj=
+      ;;
+    none) break ;;
+    esac
+    if depmode=$depmode \
+       source=sub/conftest.c object=$am__obj \
+       depfile=sub/conftest.Po tmpdepfile=sub/conftest.TPo \
+       $SHELL ./depcomp $depcc -c $am__minus_obj sub/conftest.c \
+         >/dev/null 2>conftest.err &&
+       grep sub/conftst1.h sub/conftest.Po > /dev/null 2>&1 &&
+       grep sub/conftst6.h sub/conftest.Po > /dev/null 2>&1 &&
+       grep $am__obj sub/conftest.Po > /dev/null 2>&1 &&
+       ${MAKE-make} -s -f confmf > /dev/null 2>&1; then
+      # icc doesn't choke on unknown options, it will just issue warnings
+      # or remarks (even with -Werror).  So we grep stderr for any message
+      # that says an option was ignored or not supported.
+      # When given -MP, icc 7.0 and 7.1 complain thusly:
+      #   icc: Command line warning: ignoring option '-M'; no argument required
+      # The diagnosis changed in icc 8.0:
+      #   icc: Command line remark: option '-MP' not supported
+      if (grep 'ignoring option' conftest.err ||
+          grep 'not supported' conftest.err) >/dev/null 2>&1; then :; else
+        am_cv_CXX_dependencies_compiler_type=$depmode
+        break
+      fi
+    fi
+  done
+
+  cd ..
+  rm -rf conftest.dir
+else
+  am_cv_CXX_dependencies_compiler_type=none
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CXX_dependencies_compiler_type" >&5
+$as_echo "$am_cv_CXX_dependencies_compiler_type" >&6; }
+CXXDEPMODE=depmode=$am_cv_CXX_dependencies_compiler_type
+
+ if
+  test "x$enable_dependency_tracking" != xno \
+  && test "$am_cv_CXX_dependencies_compiler_type" = gcc3; then
+  am__fastdepCXX_TRUE=
+  am__fastdepCXX_FALSE='#'
+else
+  am__fastdepCXX_TRUE='#'
+  am__fastdepCXX_FALSE=
+fi
+
+
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+if test -n "$ac_tool_prefix"; then
+  # Extract the first word of "${ac_tool_prefix}gcc", so it can be a program name with args.
+set dummy ${ac_tool_prefix}gcc; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_CC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$CC"; then
+  ac_cv_prog_CC="$CC" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_CC="${ac_tool_prefix}gcc"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+CC=$ac_cv_prog_CC
+if test -n "$CC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+$as_echo "$CC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+fi
+if test -z "$ac_cv_prog_CC"; then
+  ac_ct_CC=$CC
+  # Extract the first word of "gcc", so it can be a program name with args.
+set dummy gcc; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_ac_ct_CC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$ac_ct_CC"; then
+  ac_cv_prog_ac_ct_CC="$ac_ct_CC" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_ac_ct_CC="gcc"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+ac_ct_CC=$ac_cv_prog_ac_ct_CC
+if test -n "$ac_ct_CC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
+$as_echo "$ac_ct_CC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+  if test "x$ac_ct_CC" = x; then
+    CC=""
+  else
+    case $cross_compiling:$ac_tool_warned in
+yes:)
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+$as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
+ac_tool_warned=yes ;;
+esac
+    CC=$ac_ct_CC
+  fi
+else
+  CC="$ac_cv_prog_CC"
+fi
+
+if test -z "$CC"; then
+          if test -n "$ac_tool_prefix"; then
+    # Extract the first word of "${ac_tool_prefix}cc", so it can be a program name with args.
+set dummy ${ac_tool_prefix}cc; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_CC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$CC"; then
+  ac_cv_prog_CC="$CC" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_CC="${ac_tool_prefix}cc"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+CC=$ac_cv_prog_CC
+if test -n "$CC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+$as_echo "$CC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+  fi
+fi
+if test -z "$CC"; then
+  # Extract the first word of "cc", so it can be a program name with args.
+set dummy cc; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_CC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$CC"; then
+  ac_cv_prog_CC="$CC" # Let the user override the test.
+else
+  ac_prog_rejected=no
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    if test "$as_dir/$ac_word$ac_exec_ext" = "/usr/ucb/cc"; then
+       ac_prog_rejected=yes
+       continue
+     fi
+    ac_cv_prog_CC="cc"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+if test $ac_prog_rejected = yes; then
+  # We found a bogon in the path, so make sure we never use it.
+  set dummy $ac_cv_prog_CC
+  shift
+  if test $# != 0; then
+    # We chose a different compiler from the bogus one.
+    # However, it has the same basename, so the bogon will be chosen
+    # first if we set CC to just the basename; use the full file name.
+    shift
+    ac_cv_prog_CC="$as_dir/$ac_word${1+' '}$@"
+  fi
+fi
+fi
+fi
+CC=$ac_cv_prog_CC
+if test -n "$CC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+$as_echo "$CC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+fi
+if test -z "$CC"; then
+  if test -n "$ac_tool_prefix"; then
+  for ac_prog in cl.exe
+  do
+    # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
+set dummy $ac_tool_prefix$ac_prog; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_CC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$CC"; then
+  ac_cv_prog_CC="$CC" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_CC="$ac_tool_prefix$ac_prog"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+CC=$ac_cv_prog_CC
+if test -n "$CC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+$as_echo "$CC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+    test -n "$CC" && break
+  done
+fi
+if test -z "$CC"; then
+  ac_ct_CC=$CC
+  for ac_prog in cl.exe
+do
+  # Extract the first word of "$ac_prog", so it can be a program name with args.
+set dummy $ac_prog; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_ac_ct_CC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$ac_ct_CC"; then
+  ac_cv_prog_ac_ct_CC="$ac_ct_CC" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_ac_ct_CC="$ac_prog"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+ac_ct_CC=$ac_cv_prog_ac_ct_CC
+if test -n "$ac_ct_CC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
+$as_echo "$ac_ct_CC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+  test -n "$ac_ct_CC" && break
+done
+
+  if test "x$ac_ct_CC" = x; then
+    CC=""
+  else
+    case $cross_compiling:$ac_tool_warned in
+yes:)
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+$as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
+ac_tool_warned=yes ;;
+esac
+    CC=$ac_ct_CC
+  fi
+fi
+
+fi
+
+
+test -z "$CC" && { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "no acceptable C compiler found in \$PATH
+See \`config.log' for more details" "$LINENO" 5; }
+
+# Provide some information about the compiler.
+$as_echo "$as_me:${as_lineno-$LINENO}: checking for C compiler version" >&5
+set X $ac_compile
+ac_compiler=$2
+for ac_option in --version -v -V -qversion; do
+  { { ac_try="$ac_compiler $ac_option >&5"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    sed '10a\
+... rest of stderr output deleted ...
+         10q' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+  fi
+  rm -f conftest.er1 conftest.err
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }
+done
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C compiler" >&5
+$as_echo_n "checking whether we are using the GNU C compiler... " >&6; }
+if ${ac_cv_c_compiler_gnu+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+#ifndef __GNUC__
+       choke me
+#endif
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_compiler_gnu=yes
+else
+  ac_compiler_gnu=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+ac_cv_c_compiler_gnu=$ac_compiler_gnu
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_c_compiler_gnu" >&5
+$as_echo "$ac_cv_c_compiler_gnu" >&6; }
+if test $ac_compiler_gnu = yes; then
+  GCC=yes
+else
+  GCC=
+fi
+ac_test_CFLAGS=${CFLAGS+set}
+ac_save_CFLAGS=$CFLAGS
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC accepts -g" >&5
+$as_echo_n "checking whether $CC accepts -g... " >&6; }
+if ${ac_cv_prog_cc_g+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_save_c_werror_flag=$ac_c_werror_flag
+   ac_c_werror_flag=yes
+   ac_cv_prog_cc_g=no
+   CFLAGS="-g"
+   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_cv_prog_cc_g=yes
+else
+  CFLAGS=""
+      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+
+else
+  ac_c_werror_flag=$ac_save_c_werror_flag
+	 CFLAGS="-g"
+	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_cv_prog_cc_g=yes
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+   ac_c_werror_flag=$ac_save_c_werror_flag
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_g" >&5
+$as_echo "$ac_cv_prog_cc_g" >&6; }
+if test "$ac_test_CFLAGS" = set; then
+  CFLAGS=$ac_save_CFLAGS
+elif test $ac_cv_prog_cc_g = yes; then
+  if test "$GCC" = yes; then
+    CFLAGS="-g -O2"
+  else
+    CFLAGS="-g"
+  fi
+else
+  if test "$GCC" = yes; then
+    CFLAGS="-O2"
+  else
+    CFLAGS=
+  fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to accept ISO C89" >&5
+$as_echo_n "checking for $CC option to accept ISO C89... " >&6; }
+if ${ac_cv_prog_cc_c89+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_cv_prog_cc_c89=no
+ac_save_CC=$CC
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <stdarg.h>
+#include <stdio.h>
+struct stat;
+/* Most of the following tests are stolen from RCS 5.7's src/conf.sh.  */
+struct buf { int x; };
+FILE * (*rcsopen) (struct buf *, struct stat *, int);
+static char *e (p, i)
+     char **p;
+     int i;
+{
+  return p[i];
+}
+static char *f (char * (*g) (char **, int), char **p, ...)
+{
+  char *s;
+  va_list v;
+  va_start (v,p);
+  s = g (p, va_arg (v,int));
+  va_end (v);
+  return s;
+}
+
+/* OSF 4.0 Compaq cc is some sort of almost-ANSI by default.  It has
+   function prototypes and stuff, but not '\xHH' hex character constants.
+   These don't provoke an error unfortunately, instead are silently treated
+   as 'x'.  The following induces an error, until -std is added to get
+   proper ANSI mode.  Curiously '\x00'!='x' always comes out true, for an
+   array size at least.  It's necessary to write '\x00'==0 to get something
+   that's true only with -std.  */
+int osf4_cc_array ['\x00' == 0 ? 1 : -1];
+
+/* IBM C 6 for AIX is almost-ANSI by default, but it replaces macro parameters
+   inside strings and character constants.  */
+#define FOO(x) 'x'
+int xlc6_cc_array[FOO(a) == 'x' ? 1 : -1];
+
+int test (int i, double x);
+struct s1 {int (*f) (int a);};
+struct s2 {int (*f) (double a);};
+int pairnames (int, char **, FILE *(*)(struct buf *, struct stat *, int), int, int);
+int argc;
+char **argv;
+int
+main ()
+{
+return f (e, argv, 0) != argv[0]  ||  f (e, argv, 1) != argv[1];
+  ;
+  return 0;
+}
+_ACEOF
+for ac_arg in '' -qlanglvl=extc89 -qlanglvl=ansi -std \
+	-Ae "-Aa -D_HPUX_SOURCE" "-Xc -D__EXTENSIONS__"
+do
+  CC="$ac_save_CC $ac_arg"
+  if ac_fn_c_try_compile "$LINENO"; then :
+  ac_cv_prog_cc_c89=$ac_arg
+fi
+rm -f core conftest.err conftest.$ac_objext
+  test "x$ac_cv_prog_cc_c89" != "xno" && break
+done
+rm -f conftest.$ac_ext
+CC=$ac_save_CC
+
+fi
+# AC_CACHE_VAL
+case "x$ac_cv_prog_cc_c89" in
+  x)
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: none needed" >&5
+$as_echo "none needed" >&6; } ;;
+  xno)
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: unsupported" >&5
+$as_echo "unsupported" >&6; } ;;
+  *)
+    CC="$CC $ac_cv_prog_cc_c89"
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_c89" >&5
+$as_echo "$ac_cv_prog_cc_c89" >&6; } ;;
+esac
+if test "x$ac_cv_prog_cc_c89" != xno; then :
+
+fi
+
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC understands -c and -o together" >&5
+$as_echo_n "checking whether $CC understands -c and -o together... " >&6; }
+if ${am_cv_prog_cc_c_o+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+  # Make sure it works both with $CC and with simple cc.
+  # Following AC_PROG_CC_C_O, we do the test twice because some
+  # compilers refuse to overwrite an existing .o file with -o,
+  # though they will create one.
+  am_cv_prog_cc_c_o=yes
+  for am_i in 1 2; do
+    if { echo "$as_me:$LINENO: $CC -c conftest.$ac_ext -o conftest2.$ac_objext" >&5
+   ($CC -c conftest.$ac_ext -o conftest2.$ac_objext) >&5 2>&5
+   ac_status=$?
+   echo "$as_me:$LINENO: \$? = $ac_status" >&5
+   (exit $ac_status); } \
+         && test -f conftest2.$ac_objext; then
+      : OK
+    else
+      am_cv_prog_cc_c_o=no
+      break
+    fi
+  done
+  rm -f core conftest*
+  unset am_i
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_prog_cc_c_o" >&5
+$as_echo "$am_cv_prog_cc_c_o" >&6; }
+if test "$am_cv_prog_cc_c_o" != yes; then
+   # Losing compiler, so override with the script.
+   # FIXME: It is wrong to rewrite CC.
+   # But if we don't then we get into trouble of one sort or another.
+   # A longer-term fix would be to have automake use am__CC in this case,
+   # and then we could set am__CC="\$(top_srcdir)/compile \$(CC)"
+   CC="$am_aux_dir/compile $CC"
+fi
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+
+
+depcc="$CC"   am_compiler_list=
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
+$as_echo_n "checking dependency style of $depcc... " >&6; }
+if ${am_cv_CC_dependencies_compiler_type+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
+  # We make a subdir and do the tests there.  Otherwise we can end up
+  # making bogus files that we don't know about and never remove.  For
+  # instance it was reported that on HP-UX the gcc test will end up
+  # making a dummy file named 'D' -- because '-MD' means "put the output
+  # in D".
+  rm -rf conftest.dir
+  mkdir conftest.dir
+  # Copy depcomp to subdir because otherwise we won't find it if we're
+  # using a relative directory.
+  cp "$am_depcomp" conftest.dir
+  cd conftest.dir
+  # We will build objects and dependencies in a subdirectory because
+  # it helps to detect inapplicable dependency modes.  For instance
+  # both Tru64's cc and ICC support -MD to output dependencies as a
+  # side effect of compilation, but ICC will put the dependencies in
+  # the current directory while Tru64 will put them in the object
+  # directory.
+  mkdir sub
+
+  am_cv_CC_dependencies_compiler_type=none
+  if test "$am_compiler_list" = ""; then
+     am_compiler_list=`sed -n 's/^#*\([a-zA-Z0-9]*\))$/\1/p' < ./depcomp`
+  fi
+  am__universal=false
+  case " $depcc " in #(
+     *\ -arch\ *\ -arch\ *) am__universal=true ;;
+     esac
+
+  for depmode in $am_compiler_list; do
+    # Setup a source with many dependencies, because some compilers
+    # like to wrap large dependency lists on column 80 (with \), and
+    # we should not choose a depcomp mode which is confused by this.
+    #
+    # We need to recreate these files for each test, as the compiler may
+    # overwrite some of them when testing with obscure command lines.
+    # This happens at least with the AIX C compiler.
+    : > sub/conftest.c
+    for i in 1 2 3 4 5 6; do
+      echo '#include "conftst'$i'.h"' >> sub/conftest.c
+      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
+      # Solaris 10 /bin/sh.
+      echo '/* dummy */' > sub/conftst$i.h
+    done
+    echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
+
+    # We check with '-c' and '-o' for the sake of the "dashmstdout"
+    # mode.  It turns out that the SunPro C++ compiler does not properly
+    # handle '-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs.
+    am__obj=sub/conftest.${OBJEXT-o}
+    am__minus_obj="-o $am__obj"
+    case $depmode in
+    gcc)
+      # This depmode causes a compiler race in universal mode.
+      test "$am__universal" = false || continue
+      ;;
+    nosideeffect)
+      # After this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested.
+      if test "x$enable_dependency_tracking" = xyes; then
+	continue
+      else
+	break
+      fi
+      ;;
+    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
+      # This compiler won't grok '-c -o', but also, the minuso test has
+      # not run yet.  These depmodes are late enough in the game, and
+      # so weak that their functioning should not be impacted.
+      am__obj=conftest.${OBJEXT-o}
+      am__minus_obj=
+      ;;
+    none) break ;;
+    esac
+    if depmode=$depmode \
+       source=sub/conftest.c object=$am__obj \
+       depfile=sub/conftest.Po tmpdepfile=sub/conftest.TPo \
+       $SHELL ./depcomp $depcc -c $am__minus_obj sub/conftest.c \
+         >/dev/null 2>conftest.err &&
+       grep sub/conftst1.h sub/conftest.Po > /dev/null 2>&1 &&
+       grep sub/conftst6.h sub/conftest.Po > /dev/null 2>&1 &&
+       grep $am__obj sub/conftest.Po > /dev/null 2>&1 &&
+       ${MAKE-make} -s -f confmf > /dev/null 2>&1; then
+      # icc doesn't choke on unknown options, it will just issue warnings
+      # or remarks (even with -Werror).  So we grep stderr for any message
+      # that says an option was ignored or not supported.
+      # When given -MP, icc 7.0 and 7.1 complain thusly:
+      #   icc: Command line warning: ignoring option '-M'; no argument required
+      # The diagnosis changed in icc 8.0:
+      #   icc: Command line remark: option '-MP' not supported
+      if (grep 'ignoring option' conftest.err ||
+          grep 'not supported' conftest.err) >/dev/null 2>&1; then :; else
+        am_cv_CC_dependencies_compiler_type=$depmode
+        break
+      fi
+    fi
+  done
+
+  cd ..
+  rm -rf conftest.dir
+else
+  am_cv_CC_dependencies_compiler_type=none
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CC_dependencies_compiler_type" >&5
+$as_echo "$am_cv_CC_dependencies_compiler_type" >&6; }
+CCDEPMODE=depmode=$am_cv_CC_dependencies_compiler_type
+
+ if
+  test "x$enable_dependency_tracking" != xno \
+  && test "$am_cv_CC_dependencies_compiler_type" = gcc3; then
+  am__fastdepCC_TRUE=
+  am__fastdepCC_FALSE='#'
+else
+  am__fastdepCC_TRUE='#'
+  am__fastdepCC_FALSE=
+fi
+
+
+
+
+  OPENMP_CFLAGS=
+  # Check whether --enable-openmp was given.
+if test "${enable_openmp+set}" = set; then :
+  enableval=$enable_openmp;
+fi
+
+  if test "$enable_openmp" != no; then
+    { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to support OpenMP" >&5
+$as_echo_n "checking for $CC option to support OpenMP... " >&6; }
+if ${ac_cv_prog_c_openmp+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+#ifndef _OPENMP
+ choke me
+#endif
+#include <omp.h>
+int main () { return omp_get_num_threads (); }
+
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  ac_cv_prog_c_openmp='none needed'
+else
+  ac_cv_prog_c_openmp='unsupported'
+	  	  	  	  	  	  	                                	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp -homp \
+                           -Popenmp --openmp; do
+	    ac_save_CFLAGS=$CFLAGS
+	    CFLAGS="$CFLAGS $ac_option"
+	    cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+#ifndef _OPENMP
+ choke me
+#endif
+#include <omp.h>
+int main () { return omp_get_num_threads (); }
+
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  ac_cv_prog_c_openmp=$ac_option
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+	    CFLAGS=$ac_save_CFLAGS
+	    if test "$ac_cv_prog_c_openmp" != unsupported; then
+	      break
+	    fi
+	  done
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_c_openmp" >&5
+$as_echo "$ac_cv_prog_c_openmp" >&6; }
+    case $ac_cv_prog_c_openmp in #(
+      "none needed" | unsupported)
+	;; #(
+      *)
+	OPENMP_CFLAGS=$ac_cv_prog_c_openmp ;;
+    esac
+  fi
+
+
+if test -n "$ac_tool_prefix"; then
+  # Extract the first word of "${ac_tool_prefix}ranlib", so it can be a program name with args.
+set dummy ${ac_tool_prefix}ranlib; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_RANLIB+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$RANLIB"; then
+  ac_cv_prog_RANLIB="$RANLIB" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_RANLIB="${ac_tool_prefix}ranlib"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+RANLIB=$ac_cv_prog_RANLIB
+if test -n "$RANLIB"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $RANLIB" >&5
+$as_echo "$RANLIB" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+fi
+if test -z "$ac_cv_prog_RANLIB"; then
+  ac_ct_RANLIB=$RANLIB
+  # Extract the first word of "ranlib", so it can be a program name with args.
+set dummy ranlib; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_ac_ct_RANLIB+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$ac_ct_RANLIB"; then
+  ac_cv_prog_ac_ct_RANLIB="$ac_ct_RANLIB" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_ac_ct_RANLIB="ranlib"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+ac_ct_RANLIB=$ac_cv_prog_ac_ct_RANLIB
+if test -n "$ac_ct_RANLIB"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_RANLIB" >&5
+$as_echo "$ac_ct_RANLIB" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+  if test "x$ac_ct_RANLIB" = x; then
+    RANLIB=":"
+  else
+    case $cross_compiling:$ac_tool_warned in
+yes:)
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+$as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
+ac_tool_warned=yes ;;
+esac
+    RANLIB=$ac_ct_RANLIB
+  fi
+else
+  RANLIB="$ac_cv_prog_RANLIB"
+fi
+
+
+# Checks for libraries.
+
+
+
+    { $as_echo "$as_me:${as_lineno-$LINENO}: checking for __attribute__((aligned))" >&5
+$as_echo_n "checking for __attribute__((aligned))... " >&6; }
+if ${ax_cv_have_var_attribute_aligned+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+        cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+
+                    int foo __attribute__((aligned(32)));
+
+int
+main ()
+{
+
+
+
+  ;
+  return 0;
+}
+
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+                                      if test -s conftest.err; then :
+  ax_cv_have_var_attribute_aligned=no
+else
+  ax_cv_have_var_attribute_aligned=yes
+fi
+else
+  ax_cv_have_var_attribute_aligned=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_var_attribute_aligned" >&5
+$as_echo "$ax_cv_have_var_attribute_aligned" >&6; }
+
+    if test yes = $ax_cv_have_var_attribute_aligned; then :
+
+cat >>confdefs.h <<_ACEOF
+#define HAVE_VAR_ATTRIBUTE_ALIGNED 1
+_ACEOF
+
+fi
+
+
+
+
+# Checks for header files.
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking how to run the C preprocessor" >&5
+$as_echo_n "checking how to run the C preprocessor... " >&6; }
+# On Suns, sometimes $CPP names a directory.
+if test -n "$CPP" && test -d "$CPP"; then
+  CPP=
+fi
+if test -z "$CPP"; then
+  if ${ac_cv_prog_CPP+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+      # Double quotes because CPP needs to be expanded
+    for CPP in "$CC -E" "$CC -E -traditional-cpp" "/lib/cpp"
+    do
+      ac_preproc_ok=false
+for ac_c_preproc_warn_flag in '' yes
+do
+  # Use a header file that comes with gcc, so configuring glibc
+  # with a fresh cross-compiler works.
+  # Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
+  # <limits.h> exists even on freestanding compilers.
+  # On the NeXT, cc -E runs the code through the compiler's parser,
+  # not just through cpp. "Syntax error" is here to catch this case.
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#ifdef __STDC__
+# include <limits.h>
+#else
+# include <assert.h>
+#endif
+		     Syntax error
+_ACEOF
+if ac_fn_c_try_cpp "$LINENO"; then :
+
+else
+  # Broken: fails on valid input.
+continue
+fi
+rm -f conftest.err conftest.i conftest.$ac_ext
+
+  # OK, works on sane cases.  Now check whether nonexistent headers
+  # can be detected and how.
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <ac_nonexistent.h>
+_ACEOF
+if ac_fn_c_try_cpp "$LINENO"; then :
+  # Broken: success on invalid input.
+continue
+else
+  # Passes both tests.
+ac_preproc_ok=:
+break
+fi
+rm -f conftest.err conftest.i conftest.$ac_ext
+
+done
+# Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
+rm -f conftest.i conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then :
+  break
+fi
+
+    done
+    ac_cv_prog_CPP=$CPP
+
+fi
+  CPP=$ac_cv_prog_CPP
+else
+  ac_cv_prog_CPP=$CPP
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $CPP" >&5
+$as_echo "$CPP" >&6; }
+ac_preproc_ok=false
+for ac_c_preproc_warn_flag in '' yes
+do
+  # Use a header file that comes with gcc, so configuring glibc
+  # with a fresh cross-compiler works.
+  # Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
+  # <limits.h> exists even on freestanding compilers.
+  # On the NeXT, cc -E runs the code through the compiler's parser,
+  # not just through cpp. "Syntax error" is here to catch this case.
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#ifdef __STDC__
+# include <limits.h>
+#else
+# include <assert.h>
+#endif
+		     Syntax error
+_ACEOF
+if ac_fn_c_try_cpp "$LINENO"; then :
+
+else
+  # Broken: fails on valid input.
+continue
+fi
+rm -f conftest.err conftest.i conftest.$ac_ext
+
+  # OK, works on sane cases.  Now check whether nonexistent headers
+  # can be detected and how.
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <ac_nonexistent.h>
+_ACEOF
+if ac_fn_c_try_cpp "$LINENO"; then :
+  # Broken: success on invalid input.
+continue
+else
+  # Passes both tests.
+ac_preproc_ok=:
+break
+fi
+rm -f conftest.err conftest.i conftest.$ac_ext
+
+done
+# Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
+rm -f conftest.i conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then :
+
+else
+  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "C preprocessor \"$CPP\" fails sanity check
+See \`config.log' for more details" "$LINENO" 5; }
+fi
+
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for grep that handles long lines and -e" >&5
+$as_echo_n "checking for grep that handles long lines and -e... " >&6; }
+if ${ac_cv_path_GREP+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -z "$GREP"; then
+  ac_path_GREP_found=false
+  # Loop through the user's path and test for each of PROGNAME-LIST
+  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_prog in grep ggrep; do
+    for ac_exec_ext in '' $ac_executable_extensions; do
+      ac_path_GREP="$as_dir/$ac_prog$ac_exec_ext"
+      as_fn_executable_p "$ac_path_GREP" || continue
+# Check for GNU ac_path_GREP and select it if it is found.
+  # Check for GNU $ac_path_GREP
+case `"$ac_path_GREP" --version 2>&1` in
+*GNU*)
+  ac_cv_path_GREP="$ac_path_GREP" ac_path_GREP_found=:;;
+*)
+  ac_count=0
+  $as_echo_n 0123456789 >"conftest.in"
+  while :
+  do
+    cat "conftest.in" "conftest.in" >"conftest.tmp"
+    mv "conftest.tmp" "conftest.in"
+    cp "conftest.in" "conftest.nl"
+    $as_echo 'GREP' >> "conftest.nl"
+    "$ac_path_GREP" -e 'GREP$' -e '-(cannot match)-' < "conftest.nl" >"conftest.out" 2>/dev/null || break
+    diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
+    as_fn_arith $ac_count + 1 && ac_count=$as_val
+    if test $ac_count -gt ${ac_path_GREP_max-0}; then
+      # Best one so far, save it but keep looking for a better one
+      ac_cv_path_GREP="$ac_path_GREP"
+      ac_path_GREP_max=$ac_count
+    fi
+    # 10*(2^10) chars as input seems more than enough
+    test $ac_count -gt 10 && break
+  done
+  rm -f conftest.in conftest.tmp conftest.nl conftest.out;;
+esac
+
+      $ac_path_GREP_found && break 3
+    done
+  done
+  done
+IFS=$as_save_IFS
+  if test -z "$ac_cv_path_GREP"; then
+    as_fn_error $? "no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
+  fi
+else
+  ac_cv_path_GREP=$GREP
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_GREP" >&5
+$as_echo "$ac_cv_path_GREP" >&6; }
+ GREP="$ac_cv_path_GREP"
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for egrep" >&5
+$as_echo_n "checking for egrep... " >&6; }
+if ${ac_cv_path_EGREP+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if echo a | $GREP -E '(a|b)' >/dev/null 2>&1
+   then ac_cv_path_EGREP="$GREP -E"
+   else
+     if test -z "$EGREP"; then
+  ac_path_EGREP_found=false
+  # Loop through the user's path and test for each of PROGNAME-LIST
+  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_prog in egrep; do
+    for ac_exec_ext in '' $ac_executable_extensions; do
+      ac_path_EGREP="$as_dir/$ac_prog$ac_exec_ext"
+      as_fn_executable_p "$ac_path_EGREP" || continue
+# Check for GNU ac_path_EGREP and select it if it is found.
+  # Check for GNU $ac_path_EGREP
+case `"$ac_path_EGREP" --version 2>&1` in
+*GNU*)
+  ac_cv_path_EGREP="$ac_path_EGREP" ac_path_EGREP_found=:;;
+*)
+  ac_count=0
+  $as_echo_n 0123456789 >"conftest.in"
+  while :
+  do
+    cat "conftest.in" "conftest.in" >"conftest.tmp"
+    mv "conftest.tmp" "conftest.in"
+    cp "conftest.in" "conftest.nl"
+    $as_echo 'EGREP' >> "conftest.nl"
+    "$ac_path_EGREP" 'EGREP$' < "conftest.nl" >"conftest.out" 2>/dev/null || break
+    diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
+    as_fn_arith $ac_count + 1 && ac_count=$as_val
+    if test $ac_count -gt ${ac_path_EGREP_max-0}; then
+      # Best one so far, save it but keep looking for a better one
+      ac_cv_path_EGREP="$ac_path_EGREP"
+      ac_path_EGREP_max=$ac_count
+    fi
+    # 10*(2^10) chars as input seems more than enough
+    test $ac_count -gt 10 && break
+  done
+  rm -f conftest.in conftest.tmp conftest.nl conftest.out;;
+esac
+
+      $ac_path_EGREP_found && break 3
+    done
+  done
+  done
+IFS=$as_save_IFS
+  if test -z "$ac_cv_path_EGREP"; then
+    as_fn_error $? "no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
+  fi
+else
+  ac_cv_path_EGREP=$EGREP
+fi
+
+   fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_EGREP" >&5
+$as_echo "$ac_cv_path_EGREP" >&6; }
+ EGREP="$ac_cv_path_EGREP"
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for ANSI C header files" >&5
+$as_echo_n "checking for ANSI C header files... " >&6; }
+if ${ac_cv_header_stdc+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <stdlib.h>
+#include <stdarg.h>
+#include <string.h>
+#include <float.h>
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_cv_header_stdc=yes
+else
+  ac_cv_header_stdc=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+
+if test $ac_cv_header_stdc = yes; then
+  # SunOS 4.x string.h does not declare mem*, contrary to ANSI.
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <string.h>
+
+_ACEOF
+if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
+  $EGREP "memchr" >/dev/null 2>&1; then :
+
+else
+  ac_cv_header_stdc=no
+fi
+rm -f conftest*
+
+fi
+
+if test $ac_cv_header_stdc = yes; then
+  # ISC 2.0.2 stdlib.h does not declare free, contrary to ANSI.
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <stdlib.h>
+
+_ACEOF
+if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
+  $EGREP "free" >/dev/null 2>&1; then :
+
+else
+  ac_cv_header_stdc=no
+fi
+rm -f conftest*
+
+fi
+
+if test $ac_cv_header_stdc = yes; then
+  # /bin/cc in Irix-4.0.5 gets non-ANSI ctype macros unless using -ansi.
+  if test "$cross_compiling" = yes; then :
+  :
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <ctype.h>
+#include <stdlib.h>
+#if ((' ' & 0x0FF) == 0x020)
+# define ISLOWER(c) ('a' <= (c) && (c) <= 'z')
+# define TOUPPER(c) (ISLOWER(c) ? 'A' + ((c) - 'a') : (c))
+#else
+# define ISLOWER(c) \
+		   (('a' <= (c) && (c) <= 'i') \
+		     || ('j' <= (c) && (c) <= 'r') \
+		     || ('s' <= (c) && (c) <= 'z'))
+# define TOUPPER(c) (ISLOWER(c) ? ((c) | 0x40) : (c))
+#endif
+
+#define XOR(e, f) (((e) && !(f)) || (!(e) && (f)))
+int
+main ()
+{
+  int i;
+  for (i = 0; i < 256; i++)
+    if (XOR (islower (i), ISLOWER (i))
+	|| toupper (i) != TOUPPER (i))
+      return 2;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_run "$LINENO"; then :
+
+else
+  ac_cv_header_stdc=no
+fi
+rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
+  conftest.$ac_objext conftest.beam conftest.$ac_ext
+fi
+
+fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_header_stdc" >&5
+$as_echo "$ac_cv_header_stdc" >&6; }
+if test $ac_cv_header_stdc = yes; then
+
+$as_echo "#define STDC_HEADERS 1" >>confdefs.h
+
+fi
+
+# On IRIX 5.3, sys/types and inttypes.h are conflicting.
+for ac_header in sys/types.h sys/stat.h stdlib.h string.h memory.h strings.h \
+		  inttypes.h stdint.h unistd.h
+do :
+  as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+ac_fn_c_check_header_compile "$LINENO" "$ac_header" "$as_ac_Header" "$ac_includes_default
+"
+if eval test \"x\$"$as_ac_Header"\" = x"yes"; then :
+  cat >>confdefs.h <<_ACEOF
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+_ACEOF
+
+fi
+
+done
+
+
+for ac_header in stdint.h
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "stdint.h" "ac_cv_header_stdint_h" "$ac_includes_default"
+if test "x$ac_cv_header_stdint_h" = xyes; then :
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_STDINT_H 1
+_ACEOF
+
+fi
+
+done
+
+for ac_header in malloc/malloc.h
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "malloc/malloc.h" "ac_cv_header_malloc_malloc_h" "$ac_includes_default"
+if test "x$ac_cv_header_malloc_malloc_h" = xyes; then :
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_MALLOC_MALLOC_H 1
+_ACEOF
+
+fi
+
+done
+
+for ac_header in malloc.h
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "malloc.h" "ac_cv_header_malloc_h" "$ac_includes_default"
+if test "x$ac_cv_header_malloc_h" = xyes; then :
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_MALLOC_H 1
+_ACEOF
+
+fi
+
+done
+
+
+# Checks for typedefs, structures, and compiler characteristics.
+ac_fn_c_check_type "$LINENO" "size_t" "ac_cv_type_size_t" "$ac_includes_default"
+if test "x$ac_cv_type_size_t" = xyes; then :
+
+else
+
+cat >>confdefs.h <<_ACEOF
+#define size_t unsigned int
+_ACEOF
+
+fi
+
+ac_fn_c_find_uintX_t "$LINENO" "32" "ac_cv_c_uint32_t"
+case $ac_cv_c_uint32_t in #(
+  no|yes) ;; #(
+  *)
+
+$as_echo "#define _UINT32_T 1" >>confdefs.h
+
+
+cat >>confdefs.h <<_ACEOF
+#define uint32_t $ac_cv_c_uint32_t
+_ACEOF
+;;
+  esac
+
+ac_fn_c_find_uintX_t "$LINENO" "64" "ac_cv_c_uint64_t"
+case $ac_cv_c_uint64_t in #(
+  no|yes) ;; #(
+  *)
+
+$as_echo "#define _UINT64_T 1" >>confdefs.h
+
+
+cat >>confdefs.h <<_ACEOF
+#define uint64_t $ac_cv_c_uint64_t
+_ACEOF
+;;
+  esac
+
+
+# Checks for library functions.
+for ac_func in gettimeofday
+do :
+  ac_fn_c_check_func "$LINENO" "gettimeofday" "ac_cv_func_gettimeofday"
+if test "x$ac_cv_func_gettimeofday" = xyes; then :
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_GETTIMEOFDAY 1
+_ACEOF
+
+fi
+done
+
+
+# Check whether --enable-simd was given.
+if test "${enable_simd+set}" = set; then :
+  enableval=$enable_simd; ac_SIMD=${enable_simd}
+else
+  ac_SIMD=AVX2
+fi
+
+
+case ${ac_SIMD} in
+     SSE2)
+       echo Configuring for SSE2
+
+$as_echo "#define SSE2 1" >>confdefs.h
+
+     ;;
+     AVX)
+       echo Configuring for AVX
+
+$as_echo "#define AVX1 1" >>confdefs.h
+
+     ;;
+     AVX2)
+       echo Configuring for AVX2
+
+$as_echo "#define AVX2 1" >>confdefs.h
+
+     ;;
+     AVX512)
+       echo Configuring for AVX512
+
+$as_echo "#define AVX512 1" >>confdefs.h
+
+     ;;
+     *)
+     as_fn_error $? "${ac_SIMD} unsupported --enable-simd option" "$LINENO" 5;
+     ;;
+esac
+
+ac_config_files="$ac_config_files Makefile"
+
+cat >confcache <<\_ACEOF
+# This file is a shell script that caches the results of configure
+# tests run on this system so they can be shared between configure
+# scripts and configure runs, see configure's option --config-cache.
+# It is not useful on other systems.  If it contains results you don't
+# want to keep, you may remove or edit it.
+#
+# config.status only pays attention to the cache file if you give it
+# the --recheck option to rerun configure.
+#
+# `ac_cv_env_foo' variables (set or unset) will be overridden when
+# loading this file, other *unset* `ac_cv_foo' will be assigned the
+# following values.
+
+_ACEOF
+
+# The following way of writing the cache mishandles newlines in values,
+# but we know of no workaround that is simple, portable, and efficient.
+# So, we kill variables containing newlines.
+# Ultrix sh set writes to stderr and can't be redirected directly,
+# and sets the high bit in the cache file unless we assign to the vars.
+(
+  for ac_var in `(set) 2>&1 | sed -n 's/^\([a-zA-Z_][a-zA-Z0-9_]*\)=.*/\1/p'`; do
+    eval ac_val=\$$ac_var
+    case $ac_val in #(
+    *${as_nl}*)
+      case $ac_var in #(
+      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
+$as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
+      esac
+      case $ac_var in #(
+      _ | IFS | as_nl) ;; #(
+      BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
+      *) { eval $ac_var=; unset $ac_var;} ;;
+      esac ;;
+    esac
+  done
+
+  (set) 2>&1 |
+    case $as_nl`(ac_space=' '; set) 2>&1` in #(
+    *${as_nl}ac_space=\ *)
+      # `set' does not quote correctly, so add quotes: double-quote
+      # substitution turns \\\\ into \\, and sed turns \\ into \.
+      sed -n \
+	"s/'/'\\\\''/g;
+	  s/^\\([_$as_cr_alnum]*_cv_[_$as_cr_alnum]*\\)=\\(.*\\)/\\1='\\2'/p"
+      ;; #(
+    *)
+      # `set' quotes correctly as required by POSIX, so do not add quotes.
+      sed -n "/^[_$as_cr_alnum]*_cv_[_$as_cr_alnum]*=/p"
+      ;;
+    esac |
+    sort
+) |
+  sed '
+     /^ac_cv_env_/b end
+     t clear
+     :clear
+     s/^\([^=]*\)=\(.*[{}].*\)$/test "${\1+set}" = set || &/
+     t end
+     s/^\([^=]*\)=\(.*\)$/\1=${\1=\2}/
+     :end' >>confcache
+if diff "$cache_file" confcache >/dev/null 2>&1; then :; else
+  if test -w "$cache_file"; then
+    if test "x$cache_file" != "x/dev/null"; then
+      { $as_echo "$as_me:${as_lineno-$LINENO}: updating cache $cache_file" >&5
+$as_echo "$as_me: updating cache $cache_file" >&6;}
+      if test ! -f "$cache_file" || test -h "$cache_file"; then
+	cat confcache >"$cache_file"
+      else
+        case $cache_file in #(
+        */* | ?:*)
+	  mv -f confcache "$cache_file"$$ &&
+	  mv -f "$cache_file"$$ "$cache_file" ;; #(
+        *)
+	  mv -f confcache "$cache_file" ;;
+	esac
+      fi
+    fi
+  else
+    { $as_echo "$as_me:${as_lineno-$LINENO}: not updating unwritable cache $cache_file" >&5
+$as_echo "$as_me: not updating unwritable cache $cache_file" >&6;}
+  fi
+fi
+rm -f confcache
+
+test "x$prefix" = xNONE && prefix=$ac_default_prefix
+# Let make expand exec_prefix.
+test "x$exec_prefix" = xNONE && exec_prefix='${prefix}'
+
+DEFS=-DHAVE_CONFIG_H
+
+ac_libobjs=
+ac_ltlibobjs=
+U=
+for ac_i in : $LIBOBJS; do test "x$ac_i" = x: && continue
+  # 1. Remove the extension, and $U if already installed.
+  ac_script='s/\$U\././;s/\.o$//;s/\.obj$//'
+  ac_i=`$as_echo "$ac_i" | sed "$ac_script"`
+  # 2. Prepend LIBOBJDIR.  When used with automake>=1.10 LIBOBJDIR
+  #    will be set to the directory where LIBOBJS objects are built.
+  as_fn_append ac_libobjs " \${LIBOBJDIR}$ac_i\$U.$ac_objext"
+  as_fn_append ac_ltlibobjs " \${LIBOBJDIR}$ac_i"'$U.lo'
+done
+LIBOBJS=$ac_libobjs
+
+LTLIBOBJS=$ac_ltlibobjs
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking that generated files are newer than configure" >&5
+$as_echo_n "checking that generated files are newer than configure... " >&6; }
+   if test -n "$am_sleep_pid"; then
+     # Hide warnings about reused PIDs.
+     wait $am_sleep_pid 2>/dev/null
+   fi
+   { $as_echo "$as_me:${as_lineno-$LINENO}: result: done" >&5
+$as_echo "done" >&6; }
+ if test -n "$EXEEXT"; then
+  am__EXEEXT_TRUE=
+  am__EXEEXT_FALSE='#'
+else
+  am__EXEEXT_TRUE='#'
+  am__EXEEXT_FALSE=
+fi
+
+if test -z "${AMDEP_TRUE}" && test -z "${AMDEP_FALSE}"; then
+  as_fn_error $? "conditional \"AMDEP\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
+fi
+if test -z "${am__fastdepCXX_TRUE}" && test -z "${am__fastdepCXX_FALSE}"; then
+  as_fn_error $? "conditional \"am__fastdepCXX\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
+fi
+if test -z "${am__fastdepCC_TRUE}" && test -z "${am__fastdepCC_FALSE}"; then
+  as_fn_error $? "conditional \"am__fastdepCC\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
+fi
+
+: "${CONFIG_STATUS=./config.status}"
+ac_write_fail=0
+ac_clean_files_save=$ac_clean_files
+ac_clean_files="$ac_clean_files $CONFIG_STATUS"
+{ $as_echo "$as_me:${as_lineno-$LINENO}: creating $CONFIG_STATUS" >&5
+$as_echo "$as_me: creating $CONFIG_STATUS" >&6;}
+as_write_fail=0
+cat >$CONFIG_STATUS <<_ASEOF || as_write_fail=1
+#! $SHELL
+# Generated by $as_me.
+# Run this file to recreate the current configuration.
+# Compiler output produced by configure, useful for debugging
+# configure, is in config.log if it exists.
+
+debug=false
+ac_cs_recheck=false
+ac_cs_silent=false
+
+SHELL=\${CONFIG_SHELL-$SHELL}
+export SHELL
+_ASEOF
+cat >>$CONFIG_STATUS <<\_ASEOF || as_write_fail=1
+## -------------------- ##
+## M4sh Initialization. ##
+## -------------------- ##
+
+# Be more Bourne compatible
+DUALCASE=1; export DUALCASE # for MKS sh
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
+  emulate sh
+  NULLCMD=:
+  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '${1+"$@"}'='"$@"'
+  setopt NO_GLOB_SUBST
+else
+  case `(set -o) 2>/dev/null` in #(
+  *posix*) :
+    set -o posix ;; #(
+  *) :
+     ;;
+esac
+fi
+
+
+as_nl='
+'
+export as_nl
+# Printing a long string crashes Solaris 7 /usr/bin/printf.
+as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
+as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
+as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
+# Prefer a ksh shell builtin over an external printf program on Solaris,
+# but without wasting forks for bash or zsh.
+if test -z "$BASH_VERSION$ZSH_VERSION" \
+    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
+  as_echo='print -r --'
+  as_echo_n='print -rn --'
+elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+  as_echo='printf %s\n'
+  as_echo_n='printf %s'
+else
+  if test "X`(/usr/ucb/echo -n -n $as_echo) 2>/dev/null`" = "X-n $as_echo"; then
+    as_echo_body='eval /usr/ucb/echo -n "$1$as_nl"'
+    as_echo_n='/usr/ucb/echo -n'
+  else
+    as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
+    as_echo_n_body='eval
+      arg=$1;
+      case $arg in #(
+      *"$as_nl"*)
+	expr "X$arg" : "X\\(.*\\)$as_nl";
+	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
+      esac;
+      expr "X$arg" : "X\\(.*\\)" | tr -d "$as_nl"
+    '
+    export as_echo_n_body
+    as_echo_n='sh -c $as_echo_n_body as_echo'
+  fi
+  export as_echo_body
+  as_echo='sh -c $as_echo_body as_echo'
+fi
+
+# The user is always right.
+if test "${PATH_SEPARATOR+set}" != set; then
+  PATH_SEPARATOR=:
+  (PATH='/bin;/bin'; FPATH=$PATH; sh -c :) >/dev/null 2>&1 && {
+    (PATH='/bin:/bin'; FPATH=$PATH; sh -c :) >/dev/null 2>&1 ||
+      PATH_SEPARATOR=';'
+  }
+fi
+
+
+# IFS
+# We need space, tab and new line, in precisely that order.  Quoting is
+# there to prevent editors from complaining about space-tab.
+# (If _AS_PATH_WALK were called with IFS unset, it would disable word
+# splitting by setting IFS to empty value.)
+IFS=" ""	$as_nl"
+
+# Find who we are.  Look in the path if we contain no directory separator.
+as_myself=
+case $0 in #((
+  *[\\/]* ) as_myself=$0 ;;
+  *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+  done
+IFS=$as_save_IFS
+
+     ;;
+esac
+# We did not find ourselves, most probably we were run as `sh COMMAND'
+# in which case we are not to be found in the path.
+if test "x$as_myself" = x; then
+  as_myself=$0
+fi
+if test ! -f "$as_myself"; then
+  $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
+  exit 1
+fi
+
+# Unset variables that we do not need and which cause bugs (e.g. in
+# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
+# suppresses any "Segmentation fault" message there.  '((' could
+# trigger a bug in pdksh 5.2.14.
+for as_var in BASH_ENV ENV MAIL MAILPATH
+do eval test x\${$as_var+set} = xset \
+  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
+done
+PS1='$ '
+PS2='> '
+PS4='+ '
+
+# NLS nuisances.
+LC_ALL=C
+export LC_ALL
+LANGUAGE=C
+export LANGUAGE
+
+# CDPATH.
+(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
+
+
+# as_fn_error STATUS ERROR [LINENO LOG_FD]
+# ----------------------------------------
+# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
+# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
+# script with STATUS, using 1 if that was 0.
+as_fn_error ()
+{
+  as_status=$1; test $as_status -eq 0 && as_status=1
+  if test "$4"; then
+    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
+  fi
+  $as_echo "$as_me: error: $2" >&2
+  as_fn_exit $as_status
+} # as_fn_error
+
+
+# as_fn_set_status STATUS
+# -----------------------
+# Set $? to STATUS, without forking.
+as_fn_set_status ()
+{
+  return $1
+} # as_fn_set_status
+
+# as_fn_exit STATUS
+# -----------------
+# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
+as_fn_exit ()
+{
+  set +e
+  as_fn_set_status $1
+  exit $1
+} # as_fn_exit
+
+# as_fn_unset VAR
+# ---------------
+# Portably unset VAR.
+as_fn_unset ()
+{
+  { eval $1=; unset $1;}
+}
+as_unset=as_fn_unset
+# as_fn_append VAR VALUE
+# ----------------------
+# Append the text in VALUE to the end of the definition contained in VAR. Take
+# advantage of any shell optimizations that allow amortized linear growth over
+# repeated appends, instead of the typical quadratic growth present in naive
+# implementations.
+if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
+  eval 'as_fn_append ()
+  {
+    eval $1+=\$2
+  }'
+else
+  as_fn_append ()
+  {
+    eval $1=\$$1\$2
+  }
+fi # as_fn_append
+
+# as_fn_arith ARG...
+# ------------------
+# Perform arithmetic evaluation on the ARGs, and store the result in the
+# global $as_val. Take advantage of shells that can avoid forks. The arguments
+# must be portable across $(()) and expr.
+if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
+  eval 'as_fn_arith ()
+  {
+    as_val=$(( $* ))
+  }'
+else
+  as_fn_arith ()
+  {
+    as_val=`expr "$@" || test $? -eq 1`
+  }
+fi # as_fn_arith
+
+
+if expr a : '\(a\)' >/dev/null 2>&1 &&
+   test "X`expr 00001 : '.*\(...\)'`" = X001; then
+  as_expr=expr
+else
+  as_expr=false
+fi
+
+if (basename -- /) >/dev/null 2>&1 && test "X`basename -- / 2>&1`" = "X/"; then
+  as_basename=basename
+else
+  as_basename=false
+fi
+
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
+
+as_me=`$as_basename -- "$0" ||
+$as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
+	 X"$0" : 'X\(//\)$' \| \
+	 X"$0" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X/"$0" |
+    sed '/^.*\/\([^/][^/]*\)\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\/\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\/\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
+
+ECHO_C= ECHO_N= ECHO_T=
+case `echo -n x` in #(((((
+-n*)
+  case `echo 'xy\c'` in
+  *c*) ECHO_T='	';;	# ECHO_T is single tab character.
+  xy)  ECHO_C='\c';;
+  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
+       ECHO_T='	';;
+  esac;;
+*)
+  ECHO_N='-n';;
+esac
+
+rm -f conf$$ conf$$.exe conf$$.file
+if test -d conf$$.dir; then
+  rm -f conf$$.dir/conf$$.file
+else
+  rm -f conf$$.dir
+  mkdir conf$$.dir 2>/dev/null
+fi
+if (echo >conf$$.file) 2>/dev/null; then
+  if ln -s conf$$.file conf$$ 2>/dev/null; then
+    as_ln_s='ln -s'
+    # ... but there are two gotchas:
+    # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
+    # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
+    # In both cases, we have to default to `cp -pR'.
+    ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
+      as_ln_s='cp -pR'
+  elif ln conf$$.file conf$$ 2>/dev/null; then
+    as_ln_s=ln
+  else
+    as_ln_s='cp -pR'
+  fi
+else
+  as_ln_s='cp -pR'
+fi
+rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
+rmdir conf$$.dir 2>/dev/null
+
+
+# as_fn_mkdir_p
+# -------------
+# Create "$as_dir" as a directory, including parents if necessary.
+as_fn_mkdir_p ()
+{
+
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || eval $as_mkdir_p || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
+
+
+} # as_fn_mkdir_p
+if mkdir -p . 2>/dev/null; then
+  as_mkdir_p='mkdir -p "$as_dir"'
+else
+  test -d ./-p && rmdir ./-p
+  as_mkdir_p=false
+fi
+
+
+# as_fn_executable_p FILE
+# -----------------------
+# Test if FILE is an executable regular file.
+as_fn_executable_p ()
+{
+  test -f "$1" && test -x "$1"
+} # as_fn_executable_p
+as_test_x='test -x'
+as_executable_p=as_fn_executable_p
+
+# Sed expression to map a string onto a valid CPP name.
+as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
+
+# Sed expression to map a string onto a valid variable name.
+as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
+
+
+exec 6>&1
+## ----------------------------------- ##
+## Main body of $CONFIG_STATUS script. ##
+## ----------------------------------- ##
+_ASEOF
+test $as_write_fail = 0 && chmod +x $CONFIG_STATUS || ac_write_fail=1
+
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+# Save the log message, to keep $0 and so on meaningful, and to
+# report actual input values of CONFIG_FILES etc. instead of their
+# values after options handling.
+ac_log="
+This file was extended by Grid $as_me 1.0, which was
+generated by GNU Autoconf 2.69.  Invocation command line was
+
+  CONFIG_FILES    = $CONFIG_FILES
+  CONFIG_HEADERS  = $CONFIG_HEADERS
+  CONFIG_LINKS    = $CONFIG_LINKS
+  CONFIG_COMMANDS = $CONFIG_COMMANDS
+  $ $0 $@
+
+on `(hostname || uname -n) 2>/dev/null | sed 1q`
+"
+
+_ACEOF
+
+case $ac_config_files in *"
+"*) set x $ac_config_files; shift; ac_config_files=$*;;
+esac
+
+case $ac_config_headers in *"
+"*) set x $ac_config_headers; shift; ac_config_headers=$*;;
+esac
+
+
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+# Files that config.status was made for.
+config_files="$ac_config_files"
+config_headers="$ac_config_headers"
+config_commands="$ac_config_commands"
+
+_ACEOF
+
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+ac_cs_usage="\
+\`$as_me' instantiates files and other configuration actions
+from templates according to the current configuration.  Unless the files
+and actions are specified as TAGs, all are instantiated by default.
+
+Usage: $0 [OPTION]... [TAG]...
+
+  -h, --help       print this help, then exit
+  -V, --version    print version number and configuration settings, then exit
+      --config     print configuration, then exit
+  -q, --quiet, --silent
+                   do not print progress messages
+  -d, --debug      don't remove temporary files
+      --recheck    update $as_me by reconfiguring in the same conditions
+      --file=FILE[:TEMPLATE]
+                   instantiate the configuration file FILE
+      --header=FILE[:TEMPLATE]
+                   instantiate the configuration header FILE
+
+Configuration files:
+$config_files
+
+Configuration headers:
+$config_headers
+
+Configuration commands:
+$config_commands
+
+Report bugs to <paboyle@ph.ed.ac.uk>."
+
+_ACEOF
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+ac_cs_config="`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`"
+ac_cs_version="\\
+Grid config.status 1.0
+configured by $0, generated by GNU Autoconf 2.69,
+  with options \\"\$ac_cs_config\\"
+
+Copyright (C) 2012 Free Software Foundation, Inc.
+This config.status script is free software; the Free Software Foundation
+gives unlimited permission to copy, distribute and modify it."
+
+ac_pwd='$ac_pwd'
+srcdir='$srcdir'
+INSTALL='$INSTALL'
+MKDIR_P='$MKDIR_P'
+AWK='$AWK'
+test -n "\$AWK" || AWK=awk
+_ACEOF
+
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+# The default lists apply if the user does not specify any file.
+ac_need_defaults=:
+while test $# != 0
+do
+  case $1 in
+  --*=?*)
+    ac_option=`expr "X$1" : 'X\([^=]*\)='`
+    ac_optarg=`expr "X$1" : 'X[^=]*=\(.*\)'`
+    ac_shift=:
+    ;;
+  --*=)
+    ac_option=`expr "X$1" : 'X\([^=]*\)='`
+    ac_optarg=
+    ac_shift=:
+    ;;
+  *)
+    ac_option=$1
+    ac_optarg=$2
+    ac_shift=shift
+    ;;
+  esac
+
+  case $ac_option in
+  # Handling of the options.
+  -recheck | --recheck | --rechec | --reche | --rech | --rec | --re | --r)
+    ac_cs_recheck=: ;;
+  --version | --versio | --versi | --vers | --ver | --ve | --v | -V )
+    $as_echo "$ac_cs_version"; exit ;;
+  --config | --confi | --conf | --con | --co | --c )
+    $as_echo "$ac_cs_config"; exit ;;
+  --debug | --debu | --deb | --de | --d | -d )
+    debug=: ;;
+  --file | --fil | --fi | --f )
+    $ac_shift
+    case $ac_optarg in
+    *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
+    '') as_fn_error $? "missing file argument" ;;
+    esac
+    as_fn_append CONFIG_FILES " '$ac_optarg'"
+    ac_need_defaults=false;;
+  --header | --heade | --head | --hea )
+    $ac_shift
+    case $ac_optarg in
+    *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
+    esac
+    as_fn_append CONFIG_HEADERS " '$ac_optarg'"
+    ac_need_defaults=false;;
+  --he | --h)
+    # Conflict between --help and --header
+    as_fn_error $? "ambiguous option: \`$1'
+Try \`$0 --help' for more information.";;
+  --help | --hel | -h )
+    $as_echo "$ac_cs_usage"; exit ;;
+  -q | -quiet | --quiet | --quie | --qui | --qu | --q \
+  | -silent | --silent | --silen | --sile | --sil | --si | --s)
+    ac_cs_silent=: ;;
+
+  # This is an error.
+  -*) as_fn_error $? "unrecognized option: \`$1'
+Try \`$0 --help' for more information." ;;
+
+  *) as_fn_append ac_config_targets " $1"
+     ac_need_defaults=false ;;
+
+  esac
+  shift
+done
+
+ac_configure_extra_args=
+
+if $ac_cs_silent; then
+  exec 6>/dev/null
+  ac_configure_extra_args="$ac_configure_extra_args --silent"
+fi
+
+_ACEOF
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+if \$ac_cs_recheck; then
+  set X $SHELL '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
+  shift
+  \$as_echo "running CONFIG_SHELL=$SHELL \$*" >&6
+  CONFIG_SHELL='$SHELL'
+  export CONFIG_SHELL
+  exec "\$@"
+fi
+
+_ACEOF
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+exec 5>>config.log
+{
+  echo
+  sed 'h;s/./-/g;s/^.../## /;s/...$/ ##/;p;x;p;x' <<_ASBOX
+## Running $as_me. ##
+_ASBOX
+  $as_echo "$ac_log"
+} >&5
+
+_ACEOF
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+#
+# INIT-COMMANDS
+#
+AMDEP_TRUE="$AMDEP_TRUE" ac_aux_dir="$ac_aux_dir"
+
+_ACEOF
+
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+
+# Handling of arguments.
+for ac_config_target in $ac_config_targets
+do
+  case $ac_config_target in
+    "Grid_config.h") CONFIG_HEADERS="$CONFIG_HEADERS Grid_config.h" ;;
+    "depfiles") CONFIG_COMMANDS="$CONFIG_COMMANDS depfiles" ;;
+    "Makefile") CONFIG_FILES="$CONFIG_FILES Makefile" ;;
+
+  *) as_fn_error $? "invalid argument: \`$ac_config_target'" "$LINENO" 5;;
+  esac
+done
+
+
+# If the user did not use the arguments to specify the items to instantiate,
+# then the envvar interface is used.  Set only those that are not.
+# We use the long form for the default assignment because of an extremely
+# bizarre bug on SunOS 4.1.3.
+if $ac_need_defaults; then
+  test "${CONFIG_FILES+set}" = set || CONFIG_FILES=$config_files
+  test "${CONFIG_HEADERS+set}" = set || CONFIG_HEADERS=$config_headers
+  test "${CONFIG_COMMANDS+set}" = set || CONFIG_COMMANDS=$config_commands
+fi
+
+# Have a temporary directory for convenience.  Make it in the build tree
+# simply because there is no reason against having it here, and in addition,
+# creating and moving files from /tmp can sometimes cause problems.
+# Hook for its removal unless debugging.
+# Note that there is a small window in which the directory will not be cleaned:
+# after its creation but before its name has been assigned to `$tmp'.
+$debug ||
+{
+  tmp= ac_tmp=
+  trap 'exit_status=$?
+  : "${ac_tmp:=$tmp}"
+  { test ! -d "$ac_tmp" || rm -fr "$ac_tmp"; } && exit $exit_status
+' 0
+  trap 'as_fn_exit 1' 1 2 13 15
+}
+# Create a (secure) tmp directory for tmp files.
+
+{
+  tmp=`(umask 077 && mktemp -d "./confXXXXXX") 2>/dev/null` &&
+  test -d "$tmp"
+}  ||
+{
+  tmp=./conf$$-$RANDOM
+  (umask 077 && mkdir "$tmp")
+} || as_fn_error $? "cannot create a temporary directory in ." "$LINENO" 5
+ac_tmp=$tmp
+
+# Set up the scripts for CONFIG_FILES section.
+# No need to generate them if there are no CONFIG_FILES.
+# This happens for instance with `./config.status config.h'.
+if test -n "$CONFIG_FILES"; then
+
+
+ac_cr=`echo X | tr X '\015'`
+# On cygwin, bash can eat \r inside `` if the user requested igncr.
+# But we know of no other shell where ac_cr would be empty at this
+# point, so we can use a bashism as a fallback.
+if test "x$ac_cr" = x; then
+  eval ac_cr=\$\'\\r\'
+fi
+ac_cs_awk_cr=`$AWK 'BEGIN { print "a\rb" }' </dev/null 2>/dev/null`
+if test "$ac_cs_awk_cr" = "a${ac_cr}b"; then
+  ac_cs_awk_cr='\\r'
+else
+  ac_cs_awk_cr=$ac_cr
+fi
+
+echo 'BEGIN {' >"$ac_tmp/subs1.awk" &&
+_ACEOF
+
+
+{
+  echo "cat >conf$$subs.awk <<_ACEOF" &&
+  echo "$ac_subst_vars" | sed 's/.*/&!$&$ac_delim/' &&
+  echo "_ACEOF"
+} >conf$$subs.sh ||
+  as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
+ac_delim_num=`echo "$ac_subst_vars" | grep -c '^'`
+ac_delim='%!_!# '
+for ac_last_try in false false false false false :; do
+  . ./conf$$subs.sh ||
+    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
+
+  ac_delim_n=`sed -n "s/.*$ac_delim\$/X/p" conf$$subs.awk | grep -c X`
+  if test $ac_delim_n = $ac_delim_num; then
+    break
+  elif $ac_last_try; then
+    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
+  else
+    ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
+  fi
+done
+rm -f conf$$subs.sh
+
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+cat >>"\$ac_tmp/subs1.awk" <<\\_ACAWK &&
+_ACEOF
+sed -n '
+h
+s/^/S["/; s/!.*/"]=/
+p
+g
+s/^[^!]*!//
+:repl
+t repl
+s/'"$ac_delim"'$//
+t delim
+:nl
+h
+s/\(.\{148\}\)..*/\1/
+t more1
+s/["\\]/\\&/g; s/^/"/; s/$/\\n"\\/
+p
+n
+b repl
+:more1
+s/["\\]/\\&/g; s/^/"/; s/$/"\\/
+p
+g
+s/.\{148\}//
+t nl
+:delim
+h
+s/\(.\{148\}\)..*/\1/
+t more2
+s/["\\]/\\&/g; s/^/"/; s/$/"/
+p
+b
+:more2
+s/["\\]/\\&/g; s/^/"/; s/$/"\\/
+p
+g
+s/.\{148\}//
+t delim
+' <conf$$subs.awk | sed '
+/^[^""]/{
+  N
+  s/\n//
+}
+' >>$CONFIG_STATUS || ac_write_fail=1
+rm -f conf$$subs.awk
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+_ACAWK
+cat >>"\$ac_tmp/subs1.awk" <<_ACAWK &&
+  for (key in S) S_is_set[key] = 1
+  FS = ""
+
+}
+{
+  line = $ 0
+  nfields = split(line, field, "@")
+  substed = 0
+  len = length(field[1])
+  for (i = 2; i < nfields; i++) {
+    key = field[i]
+    keylen = length(key)
+    if (S_is_set[key]) {
+      value = S[key]
+      line = substr(line, 1, len) "" value "" substr(line, len + keylen + 3)
+      len += length(value) + length(field[++i])
+      substed = 1
+    } else
+      len += 1 + keylen
+  }
+
+  print line
+}
+
+_ACAWK
+_ACEOF
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+if sed "s/$ac_cr//" < /dev/null > /dev/null 2>&1; then
+  sed "s/$ac_cr\$//; s/$ac_cr/$ac_cs_awk_cr/g"
+else
+  cat
+fi < "$ac_tmp/subs1.awk" > "$ac_tmp/subs.awk" \
+  || as_fn_error $? "could not setup config files machinery" "$LINENO" 5
+_ACEOF
+
+# VPATH may cause trouble with some makes, so we remove sole $(srcdir),
+# ${srcdir} and @srcdir@ entries from VPATH if srcdir is ".", strip leading and
+# trailing colons and then remove the whole line if VPATH becomes empty
+# (actually we leave an empty line to preserve line numbers).
+if test "x$srcdir" = x.; then
+  ac_vpsub='/^[	 ]*VPATH[	 ]*=[	 ]*/{
+h
+s///
+s/^/:/
+s/[	 ]*$/:/
+s/:\$(srcdir):/:/g
+s/:\${srcdir}:/:/g
+s/:@srcdir@:/:/g
+s/^:*//
+s/:*$//
+x
+s/\(=[	 ]*\).*/\1/
+G
+s/\n//
+s/^[^=]*=[	 ]*$//
+}'
+fi
+
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+fi # test -n "$CONFIG_FILES"
+
+# Set up the scripts for CONFIG_HEADERS section.
+# No need to generate them if there are no CONFIG_HEADERS.
+# This happens for instance with `./config.status Makefile'.
+if test -n "$CONFIG_HEADERS"; then
+cat >"$ac_tmp/defines.awk" <<\_ACAWK ||
+BEGIN {
+_ACEOF
+
+# Transform confdefs.h into an awk script `defines.awk', embedded as
+# here-document in config.status, that substitutes the proper values into
+# config.h.in to produce config.h.
+
+# Create a delimiter string that does not exist in confdefs.h, to ease
+# handling of long lines.
+ac_delim='%!_!# '
+for ac_last_try in false false :; do
+  ac_tt=`sed -n "/$ac_delim/p" confdefs.h`
+  if test -z "$ac_tt"; then
+    break
+  elif $ac_last_try; then
+    as_fn_error $? "could not make $CONFIG_HEADERS" "$LINENO" 5
+  else
+    ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
+  fi
+done
+
+# For the awk script, D is an array of macro values keyed by name,
+# likewise P contains macro parameters if any.  Preserve backslash
+# newline sequences.
+
+ac_word_re=[_$as_cr_Letters][_$as_cr_alnum]*
+sed -n '
+s/.\{148\}/&'"$ac_delim"'/g
+t rset
+:rset
+s/^[	 ]*#[	 ]*define[	 ][	 ]*/ /
+t def
+d
+:def
+s/\\$//
+t bsnl
+s/["\\]/\\&/g
+s/^ \('"$ac_word_re"'\)\(([^()]*)\)[	 ]*\(.*\)/P["\1"]="\2"\
+D["\1"]=" \3"/p
+s/^ \('"$ac_word_re"'\)[	 ]*\(.*\)/D["\1"]=" \2"/p
+d
+:bsnl
+s/["\\]/\\&/g
+s/^ \('"$ac_word_re"'\)\(([^()]*)\)[	 ]*\(.*\)/P["\1"]="\2"\
+D["\1"]=" \3\\\\\\n"\\/p
+t cont
+s/^ \('"$ac_word_re"'\)[	 ]*\(.*\)/D["\1"]=" \2\\\\\\n"\\/p
+t cont
+d
+:cont
+n
+s/.\{148\}/&'"$ac_delim"'/g
+t clear
+:clear
+s/\\$//
+t bsnlc
+s/["\\]/\\&/g; s/^/"/; s/$/"/p
+d
+:bsnlc
+s/["\\]/\\&/g; s/^/"/; s/$/\\\\\\n"\\/p
+b cont
+' <confdefs.h | sed '
+s/'"$ac_delim"'/"\\\
+"/g' >>$CONFIG_STATUS || ac_write_fail=1
+
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+  for (key in D) D_is_set[key] = 1
+  FS = ""
+}
+/^[\t ]*#[\t ]*(define|undef)[\t ]+$ac_word_re([\t (]|\$)/ {
+  line = \$ 0
+  split(line, arg, " ")
+  if (arg[1] == "#") {
+    defundef = arg[2]
+    mac1 = arg[3]
+  } else {
+    defundef = substr(arg[1], 2)
+    mac1 = arg[2]
+  }
+  split(mac1, mac2, "(") #)
+  macro = mac2[1]
+  prefix = substr(line, 1, index(line, defundef) - 1)
+  if (D_is_set[macro]) {
+    # Preserve the white space surrounding the "#".
+    print prefix "define", macro P[macro] D[macro]
+    next
+  } else {
+    # Replace #undef with comments.  This is necessary, for example,
+    # in the case of _POSIX_SOURCE, which is predefined and required
+    # on some systems where configure will not decide to define it.
+    if (defundef == "undef") {
+      print "/*", prefix defundef, macro, "*/"
+      next
+    }
+  }
+}
+{ print }
+_ACAWK
+_ACEOF
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+  as_fn_error $? "could not setup config headers machinery" "$LINENO" 5
+fi # test -n "$CONFIG_HEADERS"
+
+
+eval set X "  :F $CONFIG_FILES  :H $CONFIG_HEADERS    :C $CONFIG_COMMANDS"
+shift
+for ac_tag
+do
+  case $ac_tag in
+  :[FHLC]) ac_mode=$ac_tag; continue;;
+  esac
+  case $ac_mode$ac_tag in
+  :[FHL]*:*);;
+  :L* | :C*:*) as_fn_error $? "invalid tag \`$ac_tag'" "$LINENO" 5;;
+  :[FH]-) ac_tag=-:-;;
+  :[FH]*) ac_tag=$ac_tag:$ac_tag.in;;
+  esac
+  ac_save_IFS=$IFS
+  IFS=:
+  set x $ac_tag
+  IFS=$ac_save_IFS
+  shift
+  ac_file=$1
+  shift
+
+  case $ac_mode in
+  :L) ac_source=$1;;
+  :[FH])
+    ac_file_inputs=
+    for ac_f
+    do
+      case $ac_f in
+      -) ac_f="$ac_tmp/stdin";;
+      *) # Look for the file first in the build tree, then in the source tree
+	 # (if the path is not absolute).  The absolute path cannot be DOS-style,
+	 # because $ac_f cannot contain `:'.
+	 test -f "$ac_f" ||
+	   case $ac_f in
+	   [\\/$]*) false;;
+	   *) test -f "$srcdir/$ac_f" && ac_f="$srcdir/$ac_f";;
+	   esac ||
+	   as_fn_error 1 "cannot find input file: \`$ac_f'" "$LINENO" 5;;
+      esac
+      case $ac_f in *\'*) ac_f=`$as_echo "$ac_f" | sed "s/'/'\\\\\\\\''/g"`;; esac
+      as_fn_append ac_file_inputs " '$ac_f'"
+    done
+
+    # Let's still pretend it is `configure' which instantiates (i.e., don't
+    # use $as_me), people would be surprised to read:
+    #    /* config.h.  Generated by config.status.  */
+    configure_input='Generated from '`
+	  $as_echo "$*" | sed 's|^[^:]*/||;s|:[^:]*/|, |g'
+	`' by configure.'
+    if test x"$ac_file" != x-; then
+      configure_input="$ac_file.  $configure_input"
+      { $as_echo "$as_me:${as_lineno-$LINENO}: creating $ac_file" >&5
+$as_echo "$as_me: creating $ac_file" >&6;}
+    fi
+    # Neutralize special characters interpreted by sed in replacement strings.
+    case $configure_input in #(
+    *\&* | *\|* | *\\* )
+       ac_sed_conf_input=`$as_echo "$configure_input" |
+       sed 's/[\\\\&|]/\\\\&/g'`;; #(
+    *) ac_sed_conf_input=$configure_input;;
+    esac
+
+    case $ac_tag in
+    *:-:* | *:-) cat >"$ac_tmp/stdin" \
+      || as_fn_error $? "could not create $ac_file" "$LINENO" 5 ;;
+    esac
+    ;;
+  esac
+
+  ac_dir=`$as_dirname -- "$ac_file" ||
+$as_expr X"$ac_file" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$ac_file" : 'X\(//\)[^/]' \| \
+	 X"$ac_file" : 'X\(//\)$' \| \
+	 X"$ac_file" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$ac_file" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+  as_dir="$ac_dir"; as_fn_mkdir_p
+  ac_builddir=.
+
+case "$ac_dir" in
+.) ac_dir_suffix= ac_top_builddir_sub=. ac_top_build_prefix= ;;
+*)
+  ac_dir_suffix=/`$as_echo "$ac_dir" | sed 's|^\.[\\/]||'`
+  # A ".." for each directory in $ac_dir_suffix.
+  ac_top_builddir_sub=`$as_echo "$ac_dir_suffix" | sed 's|/[^\\/]*|/..|g;s|/||'`
+  case $ac_top_builddir_sub in
+  "") ac_top_builddir_sub=. ac_top_build_prefix= ;;
+  *)  ac_top_build_prefix=$ac_top_builddir_sub/ ;;
+  esac ;;
+esac
+ac_abs_top_builddir=$ac_pwd
+ac_abs_builddir=$ac_pwd$ac_dir_suffix
+# for backward compatibility:
+ac_top_builddir=$ac_top_build_prefix
+
+case $srcdir in
+  .)  # We are building in place.
+    ac_srcdir=.
+    ac_top_srcdir=$ac_top_builddir_sub
+    ac_abs_top_srcdir=$ac_pwd ;;
+  [\\/]* | ?:[\\/]* )  # Absolute name.
+    ac_srcdir=$srcdir$ac_dir_suffix;
+    ac_top_srcdir=$srcdir
+    ac_abs_top_srcdir=$srcdir ;;
+  *) # Relative name.
+    ac_srcdir=$ac_top_build_prefix$srcdir$ac_dir_suffix
+    ac_top_srcdir=$ac_top_build_prefix$srcdir
+    ac_abs_top_srcdir=$ac_pwd/$srcdir ;;
+esac
+ac_abs_srcdir=$ac_abs_top_srcdir$ac_dir_suffix
+
+
+  case $ac_mode in
+  :F)
+  #
+  # CONFIG_FILE
+  #
+
+  case $INSTALL in
+  [\\/$]* | ?:[\\/]* ) ac_INSTALL=$INSTALL ;;
+  *) ac_INSTALL=$ac_top_build_prefix$INSTALL ;;
+  esac
+  ac_MKDIR_P=$MKDIR_P
+  case $MKDIR_P in
+  [\\/$]* | ?:[\\/]* ) ;;
+  */*) ac_MKDIR_P=$ac_top_build_prefix$MKDIR_P ;;
+  esac
+_ACEOF
+
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+# If the template does not know about datarootdir, expand it.
+# FIXME: This hack should be removed a few years after 2.60.
+ac_datarootdir_hack=; ac_datarootdir_seen=
+ac_sed_dataroot='
+/datarootdir/ {
+  p
+  q
+}
+/@datadir@/p
+/@docdir@/p
+/@infodir@/p
+/@localedir@/p
+/@mandir@/p'
+case `eval "sed -n \"\$ac_sed_dataroot\" $ac_file_inputs"` in
+*datarootdir*) ac_datarootdir_seen=yes;;
+*@datadir@*|*@docdir@*|*@infodir@*|*@localedir@*|*@mandir@*)
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
+$as_echo "$as_me: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&2;}
+_ACEOF
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+  ac_datarootdir_hack='
+  s&@datadir@&$datadir&g
+  s&@docdir@&$docdir&g
+  s&@infodir@&$infodir&g
+  s&@localedir@&$localedir&g
+  s&@mandir@&$mandir&g
+  s&\\\${datarootdir}&$datarootdir&g' ;;
+esac
+_ACEOF
+
+# Neutralize VPATH when `$srcdir' = `.'.
+# Shell code in configure.ac might set extrasub.
+# FIXME: do we really want to maintain this feature?
+cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+ac_sed_extra="$ac_vpsub
+$extrasub
+_ACEOF
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+:t
+/@[a-zA-Z_][a-zA-Z_0-9]*@/!b
+s|@configure_input@|$ac_sed_conf_input|;t t
+s&@top_builddir@&$ac_top_builddir_sub&;t t
+s&@top_build_prefix@&$ac_top_build_prefix&;t t
+s&@srcdir@&$ac_srcdir&;t t
+s&@abs_srcdir@&$ac_abs_srcdir&;t t
+s&@top_srcdir@&$ac_top_srcdir&;t t
+s&@abs_top_srcdir@&$ac_abs_top_srcdir&;t t
+s&@builddir@&$ac_builddir&;t t
+s&@abs_builddir@&$ac_abs_builddir&;t t
+s&@abs_top_builddir@&$ac_abs_top_builddir&;t t
+s&@INSTALL@&$ac_INSTALL&;t t
+s&@MKDIR_P@&$ac_MKDIR_P&;t t
+$ac_datarootdir_hack
+"
+eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$ac_tmp/subs.awk" \
+  >$ac_tmp/out || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+
+test -z "$ac_datarootdir_hack$ac_datarootdir_seen" &&
+  { ac_out=`sed -n '/\${datarootdir}/p' "$ac_tmp/out"`; test -n "$ac_out"; } &&
+  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' \
+      "$ac_tmp/out"`; test -z "$ac_out"; } &&
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file contains a reference to the variable \`datarootdir'
+which seems to be undefined.  Please make sure it is defined" >&5
+$as_echo "$as_me: WARNING: $ac_file contains a reference to the variable \`datarootdir'
+which seems to be undefined.  Please make sure it is defined" >&2;}
+
+  rm -f "$ac_tmp/stdin"
+  case $ac_file in
+  -) cat "$ac_tmp/out" && rm -f "$ac_tmp/out";;
+  *) rm -f "$ac_file" && mv "$ac_tmp/out" "$ac_file";;
+  esac \
+  || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+ ;;
+  :H)
+  #
+  # CONFIG_HEADER
+  #
+  if test x"$ac_file" != x-; then
+    {
+      $as_echo "/* $configure_input  */" \
+      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs"
+    } >"$ac_tmp/config.h" \
+      || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+    if diff "$ac_file" "$ac_tmp/config.h" >/dev/null 2>&1; then
+      { $as_echo "$as_me:${as_lineno-$LINENO}: $ac_file is unchanged" >&5
+$as_echo "$as_me: $ac_file is unchanged" >&6;}
+    else
+      rm -f "$ac_file"
+      mv "$ac_tmp/config.h" "$ac_file" \
+	|| as_fn_error $? "could not create $ac_file" "$LINENO" 5
+    fi
+  else
+    $as_echo "/* $configure_input  */" \
+      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs" \
+      || as_fn_error $? "could not create -" "$LINENO" 5
+  fi
+# Compute "$ac_file"'s index in $config_headers.
+_am_arg="$ac_file"
+_am_stamp_count=1
+for _am_header in $config_headers :; do
+  case $_am_header in
+    $_am_arg | $_am_arg:* )
+      break ;;
+    * )
+      _am_stamp_count=`expr $_am_stamp_count + 1` ;;
+  esac
+done
+echo "timestamp for $_am_arg" >`$as_dirname -- "$_am_arg" ||
+$as_expr X"$_am_arg" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$_am_arg" : 'X\(//\)[^/]' \| \
+	 X"$_am_arg" : 'X\(//\)$' \| \
+	 X"$_am_arg" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$_am_arg" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`/stamp-h$_am_stamp_count
+ ;;
+
+  :C)  { $as_echo "$as_me:${as_lineno-$LINENO}: executing $ac_file commands" >&5
+$as_echo "$as_me: executing $ac_file commands" >&6;}
+ ;;
+  esac
+
+
+  case $ac_file$ac_mode in
+    "depfiles":C) test x"$AMDEP_TRUE" != x"" || {
+  # Older Autoconf quotes --file arguments for eval, but not when files
+  # are listed without --file.  Let's play safe and only enable the eval
+  # if we detect the quoting.
+  case $CONFIG_FILES in
+  *\'*) eval set x "$CONFIG_FILES" ;;
+  *)   set x $CONFIG_FILES ;;
+  esac
+  shift
+  for mf
+  do
+    # Strip MF so we end up with the name of the file.
+    mf=`echo "$mf" | sed -e 's/:.*$//'`
+    # Check whether this is an Automake generated Makefile or not.
+    # We used to match only the files named 'Makefile.in', but
+    # some people rename them; so instead we look at the file content.
+    # Grep'ing the first line is not enough: some people post-process
+    # each Makefile.in and add a new line on top of each file to say so.
+    # Grep'ing the whole file is not good either: AIX grep has a line
+    # limit of 2048, but all sed's we know have understand at least 4000.
+    if sed -n 's,^#.*generated by automake.*,X,p' "$mf" | grep X >/dev/null 2>&1; then
+      dirpart=`$as_dirname -- "$mf" ||
+$as_expr X"$mf" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$mf" : 'X\(//\)[^/]' \| \
+	 X"$mf" : 'X\(//\)$' \| \
+	 X"$mf" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$mf" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+    else
+      continue
+    fi
+    # Extract the definition of DEPDIR, am__include, and am__quote
+    # from the Makefile without running 'make'.
+    DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
+    test -z "$DEPDIR" && continue
+    am__include=`sed -n 's/^am__include = //p' < "$mf"`
+    test -z "$am__include" && continue
+    am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
+    # Find all dependency output files, they are included files with
+    # $(DEPDIR) in their names.  We invoke sed twice because it is the
+    # simplest approach to changing $(DEPDIR) to its actual value in the
+    # expansion.
+    for file in `sed -n "
+      s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
+	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
+      # Make sure the directory exists.
+      test -f "$dirpart/$file" && continue
+      fdir=`$as_dirname -- "$file" ||
+$as_expr X"$file" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$file" : 'X\(//\)[^/]' \| \
+	 X"$file" : 'X\(//\)$' \| \
+	 X"$file" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$file" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      as_dir=$dirpart/$fdir; as_fn_mkdir_p
+      # echo "creating $dirpart/$file"
+      echo '# dummy' > "$dirpart/$file"
+    done
+  done
+}
+ ;;
+
+  esac
+done # for ac_tag
+
+
+as_fn_exit 0
+_ACEOF
+ac_clean_files=$ac_clean_files_save
+
+test $ac_write_fail = 0 ||
+  as_fn_error $? "write failure creating $CONFIG_STATUS" "$LINENO" 5
+
+
+# configure is writing to config.log, and then calls config.status.
+# config.status does its own redirection, appending to config.log.
+# Unfortunately, on DOS this fails, as config.log is still kept open
+# by configure, so config.status won't be able to write to it; its
+# output is simply discarded.  So we exec the FD to /dev/null,
+# effectively closing config.log, so it can be properly (re)opened and
+# appended to by config.status.  When coming back to configure, we
+# need to make the FD available again.
+if test "$no_create" != yes; then
+  ac_cs_success=:
+  ac_config_status_args=
+  test "$silent" = yes &&
+    ac_config_status_args="$ac_config_status_args --quiet"
+  exec 5>/dev/null
+  $SHELL $CONFIG_STATUS $ac_config_status_args || ac_cs_success=false
+  exec 5>>config.log
+  # Use ||, not &&, to avoid exiting from the if with $? = 1, which
+  # would make configure fail if this is the last instruction.
+  $ac_cs_success || as_fn_exit 1
+fi
+if test -n "$ac_unrecognized_opts" && test "$enable_option_checking" != no; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
+$as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2;}
+fi
+

From 110ddd19003e3f08e25070d07a1be3f73e8310bb Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 4 Mar 2015 11:54:46 +0000
Subject: [PATCH 010/429] install-sh distro

---
 install-sh | 1 +
 1 file changed, 1 insertion(+)
 create mode 120000 install-sh

diff --git a/install-sh b/install-sh
new file mode 120000
index 00000000..545f1114
--- /dev/null
+++ b/install-sh
@@ -0,0 +1 @@
+/opt/local/share/automake-1.15/install-sh
\ No newline at end of file

From 63c7eb262eeacd66c935de3f395ca370edec8f57 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 4 Mar 2015 11:55:44 +0000
Subject: [PATCH 011/429] file

---
 install-sh | 502 ++++++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 501 insertions(+), 1 deletion(-)
 mode change 120000 => 100755 install-sh

diff --git a/install-sh b/install-sh
deleted file mode 120000
index 545f1114..00000000
--- a/install-sh
+++ /dev/null
@@ -1 +0,0 @@
-/opt/local/share/automake-1.15/install-sh
\ No newline at end of file
diff --git a/install-sh b/install-sh
new file mode 100755
index 00000000..0b0fdcbb
--- /dev/null
+++ b/install-sh
@@ -0,0 +1,501 @@
+#!/bin/sh
+# install - install a program, script, or datafile
+
+scriptversion=2013-12-25.23; # UTC
+
+# This originates from X11R5 (mit/util/scripts/install.sh), which was
+# later released in X11R6 (xc/config/util/install.sh) with the
+# following copyright and license.
+#
+# Copyright (C) 1994 X Consortium
+#
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to
+# deal in the Software without restriction, including without limitation the
+# rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+# sell copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL THE
+# X CONSORTIUM BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+# AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNEC-
+# TION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+#
+# Except as contained in this notice, the name of the X Consortium shall not
+# be used in advertising or otherwise to promote the sale, use or other deal-
+# ings in this Software without prior written authorization from the X Consor-
+# tium.
+#
+#
+# FSF changes to this file are in the public domain.
+#
+# Calling this script install-sh is preferred over install.sh, to prevent
+# 'make' implicit rules from creating a file called install from it
+# when there is no Makefile.
+#
+# This script is compatible with the BSD install script, but was written
+# from scratch.
+
+tab='	'
+nl='
+'
+IFS=" $tab$nl"
+
+# Set DOITPROG to "echo" to test this script.
+
+doit=${DOITPROG-}
+doit_exec=${doit:-exec}
+
+# Put in absolute file names if you don't have them in your path;
+# or use environment vars.
+
+chgrpprog=${CHGRPPROG-chgrp}
+chmodprog=${CHMODPROG-chmod}
+chownprog=${CHOWNPROG-chown}
+cmpprog=${CMPPROG-cmp}
+cpprog=${CPPROG-cp}
+mkdirprog=${MKDIRPROG-mkdir}
+mvprog=${MVPROG-mv}
+rmprog=${RMPROG-rm}
+stripprog=${STRIPPROG-strip}
+
+posix_mkdir=
+
+# Desired mode of installed file.
+mode=0755
+
+chgrpcmd=
+chmodcmd=$chmodprog
+chowncmd=
+mvcmd=$mvprog
+rmcmd="$rmprog -f"
+stripcmd=
+
+src=
+dst=
+dir_arg=
+dst_arg=
+
+copy_on_change=false
+is_target_a_directory=possibly
+
+usage="\
+Usage: $0 [OPTION]... [-T] SRCFILE DSTFILE
+   or: $0 [OPTION]... SRCFILES... DIRECTORY
+   or: $0 [OPTION]... -t DIRECTORY SRCFILES...
+   or: $0 [OPTION]... -d DIRECTORIES...
+
+In the 1st form, copy SRCFILE to DSTFILE.
+In the 2nd and 3rd, copy all SRCFILES to DIRECTORY.
+In the 4th, create DIRECTORIES.
+
+Options:
+     --help     display this help and exit.
+     --version  display version info and exit.
+
+  -c            (ignored)
+  -C            install only if different (preserve the last data modification time)
+  -d            create directories instead of installing files.
+  -g GROUP      $chgrpprog installed files to GROUP.
+  -m MODE       $chmodprog installed files to MODE.
+  -o USER       $chownprog installed files to USER.
+  -s            $stripprog installed files.
+  -t DIRECTORY  install into DIRECTORY.
+  -T            report an error if DSTFILE is a directory.
+
+Environment variables override the default commands:
+  CHGRPPROG CHMODPROG CHOWNPROG CMPPROG CPPROG MKDIRPROG MVPROG
+  RMPROG STRIPPROG
+"
+
+while test $# -ne 0; do
+  case $1 in
+    -c) ;;
+
+    -C) copy_on_change=true;;
+
+    -d) dir_arg=true;;
+
+    -g) chgrpcmd="$chgrpprog $2"
+        shift;;
+
+    --help) echo "$usage"; exit $?;;
+
+    -m) mode=$2
+        case $mode in
+          *' '* | *"$tab"* | *"$nl"* | *'*'* | *'?'* | *'['*)
+            echo "$0: invalid mode: $mode" >&2
+            exit 1;;
+        esac
+        shift;;
+
+    -o) chowncmd="$chownprog $2"
+        shift;;
+
+    -s) stripcmd=$stripprog;;
+
+    -t)
+        is_target_a_directory=always
+        dst_arg=$2
+        # Protect names problematic for 'test' and other utilities.
+        case $dst_arg in
+          -* | [=\(\)!]) dst_arg=./$dst_arg;;
+        esac
+        shift;;
+
+    -T) is_target_a_directory=never;;
+
+    --version) echo "$0 $scriptversion"; exit $?;;
+
+    --) shift
+        break;;
+
+    -*) echo "$0: invalid option: $1" >&2
+        exit 1;;
+
+    *)  break;;
+  esac
+  shift
+done
+
+# We allow the use of options -d and -T together, by making -d
+# take the precedence; this is for compatibility with GNU install.
+
+if test -n "$dir_arg"; then
+  if test -n "$dst_arg"; then
+    echo "$0: target directory not allowed when installing a directory." >&2
+    exit 1
+  fi
+fi
+
+if test $# -ne 0 && test -z "$dir_arg$dst_arg"; then
+  # When -d is used, all remaining arguments are directories to create.
+  # When -t is used, the destination is already specified.
+  # Otherwise, the last argument is the destination.  Remove it from $@.
+  for arg
+  do
+    if test -n "$dst_arg"; then
+      # $@ is not empty: it contains at least $arg.
+      set fnord "$@" "$dst_arg"
+      shift # fnord
+    fi
+    shift # arg
+    dst_arg=$arg
+    # Protect names problematic for 'test' and other utilities.
+    case $dst_arg in
+      -* | [=\(\)!]) dst_arg=./$dst_arg;;
+    esac
+  done
+fi
+
+if test $# -eq 0; then
+  if test -z "$dir_arg"; then
+    echo "$0: no input file specified." >&2
+    exit 1
+  fi
+  # It's OK to call 'install-sh -d' without argument.
+  # This can happen when creating conditional directories.
+  exit 0
+fi
+
+if test -z "$dir_arg"; then
+  if test $# -gt 1 || test "$is_target_a_directory" = always; then
+    if test ! -d "$dst_arg"; then
+      echo "$0: $dst_arg: Is not a directory." >&2
+      exit 1
+    fi
+  fi
+fi
+
+if test -z "$dir_arg"; then
+  do_exit='(exit $ret); exit $ret'
+  trap "ret=129; $do_exit" 1
+  trap "ret=130; $do_exit" 2
+  trap "ret=141; $do_exit" 13
+  trap "ret=143; $do_exit" 15
+
+  # Set umask so as not to create temps with too-generous modes.
+  # However, 'strip' requires both read and write access to temps.
+  case $mode in
+    # Optimize common cases.
+    *644) cp_umask=133;;
+    *755) cp_umask=22;;
+
+    *[0-7])
+      if test -z "$stripcmd"; then
+        u_plus_rw=
+      else
+        u_plus_rw='% 200'
+      fi
+      cp_umask=`expr '(' 777 - $mode % 1000 ')' $u_plus_rw`;;
+    *)
+      if test -z "$stripcmd"; then
+        u_plus_rw=
+      else
+        u_plus_rw=,u+rw
+      fi
+      cp_umask=$mode$u_plus_rw;;
+  esac
+fi
+
+for src
+do
+  # Protect names problematic for 'test' and other utilities.
+  case $src in
+    -* | [=\(\)!]) src=./$src;;
+  esac
+
+  if test -n "$dir_arg"; then
+    dst=$src
+    dstdir=$dst
+    test -d "$dstdir"
+    dstdir_status=$?
+  else
+
+    # Waiting for this to be detected by the "$cpprog $src $dsttmp" command
+    # might cause directories to be created, which would be especially bad
+    # if $src (and thus $dsttmp) contains '*'.
+    if test ! -f "$src" && test ! -d "$src"; then
+      echo "$0: $src does not exist." >&2
+      exit 1
+    fi
+
+    if test -z "$dst_arg"; then
+      echo "$0: no destination specified." >&2
+      exit 1
+    fi
+    dst=$dst_arg
+
+    # If destination is a directory, append the input filename; won't work
+    # if double slashes aren't ignored.
+    if test -d "$dst"; then
+      if test "$is_target_a_directory" = never; then
+        echo "$0: $dst_arg: Is a directory" >&2
+        exit 1
+      fi
+      dstdir=$dst
+      dst=$dstdir/`basename "$src"`
+      dstdir_status=0
+    else
+      dstdir=`dirname "$dst"`
+      test -d "$dstdir"
+      dstdir_status=$?
+    fi
+  fi
+
+  obsolete_mkdir_used=false
+
+  if test $dstdir_status != 0; then
+    case $posix_mkdir in
+      '')
+        # Create intermediate dirs using mode 755 as modified by the umask.
+        # This is like FreeBSD 'install' as of 1997-10-28.
+        umask=`umask`
+        case $stripcmd.$umask in
+          # Optimize common cases.
+          *[2367][2367]) mkdir_umask=$umask;;
+          .*0[02][02] | .[02][02] | .[02]) mkdir_umask=22;;
+
+          *[0-7])
+            mkdir_umask=`expr $umask + 22 \
+              - $umask % 100 % 40 + $umask % 20 \
+              - $umask % 10 % 4 + $umask % 2
+            `;;
+          *) mkdir_umask=$umask,go-w;;
+        esac
+
+        # With -d, create the new directory with the user-specified mode.
+        # Otherwise, rely on $mkdir_umask.
+        if test -n "$dir_arg"; then
+          mkdir_mode=-m$mode
+        else
+          mkdir_mode=
+        fi
+
+        posix_mkdir=false
+        case $umask in
+          *[123567][0-7][0-7])
+            # POSIX mkdir -p sets u+wx bits regardless of umask, which
+            # is incompatible with FreeBSD 'install' when (umask & 300) != 0.
+            ;;
+          *)
+            tmpdir=${TMPDIR-/tmp}/ins$RANDOM-$$
+            trap 'ret=$?; rmdir "$tmpdir/d" "$tmpdir" 2>/dev/null; exit $ret' 0
+
+            if (umask $mkdir_umask &&
+                exec $mkdirprog $mkdir_mode -p -- "$tmpdir/d") >/dev/null 2>&1
+            then
+              if test -z "$dir_arg" || {
+                   # Check for POSIX incompatibilities with -m.
+                   # HP-UX 11.23 and IRIX 6.5 mkdir -m -p sets group- or
+                   # other-writable bit of parent directory when it shouldn't.
+                   # FreeBSD 6.1 mkdir -m -p sets mode of existing directory.
+                   ls_ld_tmpdir=`ls -ld "$tmpdir"`
+                   case $ls_ld_tmpdir in
+                     d????-?r-*) different_mode=700;;
+                     d????-?--*) different_mode=755;;
+                     *) false;;
+                   esac &&
+                   $mkdirprog -m$different_mode -p -- "$tmpdir" && {
+                     ls_ld_tmpdir_1=`ls -ld "$tmpdir"`
+                     test "$ls_ld_tmpdir" = "$ls_ld_tmpdir_1"
+                   }
+                 }
+              then posix_mkdir=:
+              fi
+              rmdir "$tmpdir/d" "$tmpdir"
+            else
+              # Remove any dirs left behind by ancient mkdir implementations.
+              rmdir ./$mkdir_mode ./-p ./-- 2>/dev/null
+            fi
+            trap '' 0;;
+        esac;;
+    esac
+
+    if
+      $posix_mkdir && (
+        umask $mkdir_umask &&
+        $doit_exec $mkdirprog $mkdir_mode -p -- "$dstdir"
+      )
+    then :
+    else
+
+      # The umask is ridiculous, or mkdir does not conform to POSIX,
+      # or it failed possibly due to a race condition.  Create the
+      # directory the slow way, step by step, checking for races as we go.
+
+      case $dstdir in
+        /*) prefix='/';;
+        [-=\(\)!]*) prefix='./';;
+        *)  prefix='';;
+      esac
+
+      oIFS=$IFS
+      IFS=/
+      set -f
+      set fnord $dstdir
+      shift
+      set +f
+      IFS=$oIFS
+
+      prefixes=
+
+      for d
+      do
+        test X"$d" = X && continue
+
+        prefix=$prefix$d
+        if test -d "$prefix"; then
+          prefixes=
+        else
+          if $posix_mkdir; then
+            (umask=$mkdir_umask &&
+             $doit_exec $mkdirprog $mkdir_mode -p -- "$dstdir") && break
+            # Don't fail if two instances are running concurrently.
+            test -d "$prefix" || exit 1
+          else
+            case $prefix in
+              *\'*) qprefix=`echo "$prefix" | sed "s/'/'\\\\\\\\''/g"`;;
+              *) qprefix=$prefix;;
+            esac
+            prefixes="$prefixes '$qprefix'"
+          fi
+        fi
+        prefix=$prefix/
+      done
+
+      if test -n "$prefixes"; then
+        # Don't fail if two instances are running concurrently.
+        (umask $mkdir_umask &&
+         eval "\$doit_exec \$mkdirprog $prefixes") ||
+          test -d "$dstdir" || exit 1
+        obsolete_mkdir_used=true
+      fi
+    fi
+  fi
+
+  if test -n "$dir_arg"; then
+    { test -z "$chowncmd" || $doit $chowncmd "$dst"; } &&
+    { test -z "$chgrpcmd" || $doit $chgrpcmd "$dst"; } &&
+    { test "$obsolete_mkdir_used$chowncmd$chgrpcmd" = false ||
+      test -z "$chmodcmd" || $doit $chmodcmd $mode "$dst"; } || exit 1
+  else
+
+    # Make a couple of temp file names in the proper directory.
+    dsttmp=$dstdir/_inst.$$_
+    rmtmp=$dstdir/_rm.$$_
+
+    # Trap to clean up those temp files at exit.
+    trap 'ret=$?; rm -f "$dsttmp" "$rmtmp" && exit $ret' 0
+
+    # Copy the file name to the temp name.
+    (umask $cp_umask && $doit_exec $cpprog "$src" "$dsttmp") &&
+
+    # and set any options; do chmod last to preserve setuid bits.
+    #
+    # If any of these fail, we abort the whole thing.  If we want to
+    # ignore errors from any of these, just make sure not to ignore
+    # errors from the above "$doit $cpprog $src $dsttmp" command.
+    #
+    { test -z "$chowncmd" || $doit $chowncmd "$dsttmp"; } &&
+    { test -z "$chgrpcmd" || $doit $chgrpcmd "$dsttmp"; } &&
+    { test -z "$stripcmd" || $doit $stripcmd "$dsttmp"; } &&
+    { test -z "$chmodcmd" || $doit $chmodcmd $mode "$dsttmp"; } &&
+
+    # If -C, don't bother to copy if it wouldn't change the file.
+    if $copy_on_change &&
+       old=`LC_ALL=C ls -dlL "$dst"     2>/dev/null` &&
+       new=`LC_ALL=C ls -dlL "$dsttmp"  2>/dev/null` &&
+       set -f &&
+       set X $old && old=:$2:$4:$5:$6 &&
+       set X $new && new=:$2:$4:$5:$6 &&
+       set +f &&
+       test "$old" = "$new" &&
+       $cmpprog "$dst" "$dsttmp" >/dev/null 2>&1
+    then
+      rm -f "$dsttmp"
+    else
+      # Rename the file to the real destination.
+      $doit $mvcmd -f "$dsttmp" "$dst" 2>/dev/null ||
+
+      # The rename failed, perhaps because mv can't rename something else
+      # to itself, or perhaps because mv is so ancient that it does not
+      # support -f.
+      {
+        # Now remove or move aside any old file at destination location.
+        # We try this two ways since rm can't unlink itself on some
+        # systems and the destination file might be busy for other
+        # reasons.  In this case, the final cleanup might fail but the new
+        # file should still install successfully.
+        {
+          test ! -f "$dst" ||
+          $doit $rmcmd -f "$dst" 2>/dev/null ||
+          { $doit $mvcmd -f "$dst" "$rmtmp" 2>/dev/null &&
+            { $doit $rmcmd -f "$rmtmp" 2>/dev/null; :; }
+          } ||
+          { echo "$0: cannot unlink or rename $dst" >&2
+            (exit 1); exit 1
+          }
+        } &&
+
+        # Now rename the file to the real destination.
+        $doit $mvcmd "$dsttmp" "$dst"
+      }
+    fi || exit 1
+
+    trap '' 0
+  fi
+done
+
+# Local variables:
+# eval: (add-hook 'write-file-hooks 'time-stamp)
+# time-stamp-start: "scriptversion="
+# time-stamp-format: "%:y-%02m-%02d.%02H"
+# time-stamp-time-zone: "UTC"
+# time-stamp-end: "; # UTC"
+# End:

From 96260b60b07679cf5fce4ab457f50c613a1014f5 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 4 Mar 2015 11:57:14 +0000
Subject: [PATCH 012/429] files

---
 compile | 347 +++++++++++++++++++++++++
 depcomp | 791 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 missing | 215 +++++++++++++++
 3 files changed, 1353 insertions(+)
 create mode 100755 compile
 create mode 100755 depcomp
 create mode 100755 missing

diff --git a/compile b/compile
new file mode 100755
index 00000000..a85b723c
--- /dev/null
+++ b/compile
@@ -0,0 +1,347 @@
+#! /bin/sh
+# Wrapper for compilers which do not understand '-c -o'.
+
+scriptversion=2012-10-14.11; # UTC
+
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
+# Written by Tom Tromey <tromey@cygnus.com>.
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2, or (at your option)
+# any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+# As a special exception to the GNU General Public License, if you
+# distribute this file as part of a program that contains a
+# configuration script generated by Autoconf, you may include it under
+# the same distribution terms that you use for the rest of that program.
+
+# This file is maintained in Automake, please report
+# bugs to <bug-automake@gnu.org> or send patches to
+# <automake-patches@gnu.org>.
+
+nl='
+'
+
+# We need space, tab and new line, in precisely that order.  Quoting is
+# there to prevent tools from complaining about whitespace usage.
+IFS=" ""	$nl"
+
+file_conv=
+
+# func_file_conv build_file lazy
+# Convert a $build file to $host form and store it in $file
+# Currently only supports Windows hosts. If the determined conversion
+# type is listed in (the comma separated) LAZY, no conversion will
+# take place.
+func_file_conv ()
+{
+  file=$1
+  case $file in
+    / | /[!/]*) # absolute file, and not a UNC file
+      if test -z "$file_conv"; then
+	# lazily determine how to convert abs files
+	case `uname -s` in
+	  MINGW*)
+	    file_conv=mingw
+	    ;;
+	  CYGWIN*)
+	    file_conv=cygwin
+	    ;;
+	  *)
+	    file_conv=wine
+	    ;;
+	esac
+      fi
+      case $file_conv/,$2, in
+	*,$file_conv,*)
+	  ;;
+	mingw/*)
+	  file=`cmd //C echo "$file " | sed -e 's/"\(.*\) " *$/\1/'`
+	  ;;
+	cygwin/*)
+	  file=`cygpath -m "$file" || echo "$file"`
+	  ;;
+	wine/*)
+	  file=`winepath -w "$file" || echo "$file"`
+	  ;;
+      esac
+      ;;
+  esac
+}
+
+# func_cl_dashL linkdir
+# Make cl look for libraries in LINKDIR
+func_cl_dashL ()
+{
+  func_file_conv "$1"
+  if test -z "$lib_path"; then
+    lib_path=$file
+  else
+    lib_path="$lib_path;$file"
+  fi
+  linker_opts="$linker_opts -LIBPATH:$file"
+}
+
+# func_cl_dashl library
+# Do a library search-path lookup for cl
+func_cl_dashl ()
+{
+  lib=$1
+  found=no
+  save_IFS=$IFS
+  IFS=';'
+  for dir in $lib_path $LIB
+  do
+    IFS=$save_IFS
+    if $shared && test -f "$dir/$lib.dll.lib"; then
+      found=yes
+      lib=$dir/$lib.dll.lib
+      break
+    fi
+    if test -f "$dir/$lib.lib"; then
+      found=yes
+      lib=$dir/$lib.lib
+      break
+    fi
+    if test -f "$dir/lib$lib.a"; then
+      found=yes
+      lib=$dir/lib$lib.a
+      break
+    fi
+  done
+  IFS=$save_IFS
+
+  if test "$found" != yes; then
+    lib=$lib.lib
+  fi
+}
+
+# func_cl_wrapper cl arg...
+# Adjust compile command to suit cl
+func_cl_wrapper ()
+{
+  # Assume a capable shell
+  lib_path=
+  shared=:
+  linker_opts=
+  for arg
+  do
+    if test -n "$eat"; then
+      eat=
+    else
+      case $1 in
+	-o)
+	  # configure might choose to run compile as 'compile cc -o foo foo.c'.
+	  eat=1
+	  case $2 in
+	    *.o | *.[oO][bB][jJ])
+	      func_file_conv "$2"
+	      set x "$@" -Fo"$file"
+	      shift
+	      ;;
+	    *)
+	      func_file_conv "$2"
+	      set x "$@" -Fe"$file"
+	      shift
+	      ;;
+	  esac
+	  ;;
+	-I)
+	  eat=1
+	  func_file_conv "$2" mingw
+	  set x "$@" -I"$file"
+	  shift
+	  ;;
+	-I*)
+	  func_file_conv "${1#-I}" mingw
+	  set x "$@" -I"$file"
+	  shift
+	  ;;
+	-l)
+	  eat=1
+	  func_cl_dashl "$2"
+	  set x "$@" "$lib"
+	  shift
+	  ;;
+	-l*)
+	  func_cl_dashl "${1#-l}"
+	  set x "$@" "$lib"
+	  shift
+	  ;;
+	-L)
+	  eat=1
+	  func_cl_dashL "$2"
+	  ;;
+	-L*)
+	  func_cl_dashL "${1#-L}"
+	  ;;
+	-static)
+	  shared=false
+	  ;;
+	-Wl,*)
+	  arg=${1#-Wl,}
+	  save_ifs="$IFS"; IFS=','
+	  for flag in $arg; do
+	    IFS="$save_ifs"
+	    linker_opts="$linker_opts $flag"
+	  done
+	  IFS="$save_ifs"
+	  ;;
+	-Xlinker)
+	  eat=1
+	  linker_opts="$linker_opts $2"
+	  ;;
+	-*)
+	  set x "$@" "$1"
+	  shift
+	  ;;
+	*.cc | *.CC | *.cxx | *.CXX | *.[cC]++)
+	  func_file_conv "$1"
+	  set x "$@" -Tp"$file"
+	  shift
+	  ;;
+	*.c | *.cpp | *.CPP | *.lib | *.LIB | *.Lib | *.OBJ | *.obj | *.[oO])
+	  func_file_conv "$1" mingw
+	  set x "$@" "$file"
+	  shift
+	  ;;
+	*)
+	  set x "$@" "$1"
+	  shift
+	  ;;
+      esac
+    fi
+    shift
+  done
+  if test -n "$linker_opts"; then
+    linker_opts="-link$linker_opts"
+  fi
+  exec "$@" $linker_opts
+  exit 1
+}
+
+eat=
+
+case $1 in
+  '')
+     echo "$0: No command.  Try '$0 --help' for more information." 1>&2
+     exit 1;
+     ;;
+  -h | --h*)
+    cat <<\EOF
+Usage: compile [--help] [--version] PROGRAM [ARGS]
+
+Wrapper for compilers which do not understand '-c -o'.
+Remove '-o dest.o' from ARGS, run PROGRAM with the remaining
+arguments, and rename the output as expected.
+
+If you are trying to build a whole package this is not the
+right script to run: please start by reading the file 'INSTALL'.
+
+Report bugs to <bug-automake@gnu.org>.
+EOF
+    exit $?
+    ;;
+  -v | --v*)
+    echo "compile $scriptversion"
+    exit $?
+    ;;
+  cl | *[/\\]cl | cl.exe | *[/\\]cl.exe )
+    func_cl_wrapper "$@"      # Doesn't return...
+    ;;
+esac
+
+ofile=
+cfile=
+
+for arg
+do
+  if test -n "$eat"; then
+    eat=
+  else
+    case $1 in
+      -o)
+	# configure might choose to run compile as 'compile cc -o foo foo.c'.
+	# So we strip '-o arg' only if arg is an object.
+	eat=1
+	case $2 in
+	  *.o | *.obj)
+	    ofile=$2
+	    ;;
+	  *)
+	    set x "$@" -o "$2"
+	    shift
+	    ;;
+	esac
+	;;
+      *.c)
+	cfile=$1
+	set x "$@" "$1"
+	shift
+	;;
+      *)
+	set x "$@" "$1"
+	shift
+	;;
+    esac
+  fi
+  shift
+done
+
+if test -z "$ofile" || test -z "$cfile"; then
+  # If no '-o' option was seen then we might have been invoked from a
+  # pattern rule where we don't need one.  That is ok -- this is a
+  # normal compilation that the losing compiler can handle.  If no
+  # '.c' file was seen then we are probably linking.  That is also
+  # ok.
+  exec "$@"
+fi
+
+# Name of file we expect compiler to create.
+cofile=`echo "$cfile" | sed 's|^.*[\\/]||; s|^[a-zA-Z]:||; s/\.c$/.o/'`
+
+# Create the lock directory.
+# Note: use '[/\\:.-]' here to ensure that we don't use the same name
+# that we are using for the .o file.  Also, base the name on the expected
+# object file name, since that is what matters with a parallel build.
+lockdir=`echo "$cofile" | sed -e 's|[/\\:.-]|_|g'`.d
+while true; do
+  if mkdir "$lockdir" >/dev/null 2>&1; then
+    break
+  fi
+  sleep 1
+done
+# FIXME: race condition here if user kills between mkdir and trap.
+trap "rmdir '$lockdir'; exit 1" 1 2 15
+
+# Run the compile.
+"$@"
+ret=$?
+
+if test -f "$cofile"; then
+  test "$cofile" = "$ofile" || mv "$cofile" "$ofile"
+elif test -f "${cofile}bj"; then
+  test "${cofile}bj" = "$ofile" || mv "${cofile}bj" "$ofile"
+fi
+
+rmdir "$lockdir"
+exit $ret
+
+# Local Variables:
+# mode: shell-script
+# sh-indentation: 2
+# eval: (add-hook 'write-file-hooks 'time-stamp)
+# time-stamp-start: "scriptversion="
+# time-stamp-format: "%:y-%02m-%02d.%02H"
+# time-stamp-time-zone: "UTC"
+# time-stamp-end: "; # UTC"
+# End:
diff --git a/depcomp b/depcomp
new file mode 100755
index 00000000..fc98710e
--- /dev/null
+++ b/depcomp
@@ -0,0 +1,791 @@
+#! /bin/sh
+# depcomp - compile a program generating dependencies as side-effects
+
+scriptversion=2013-05-30.07; # UTC
+
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2, or (at your option)
+# any later version.
+
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+# As a special exception to the GNU General Public License, if you
+# distribute this file as part of a program that contains a
+# configuration script generated by Autoconf, you may include it under
+# the same distribution terms that you use for the rest of that program.
+
+# Originally written by Alexandre Oliva <oliva@dcc.unicamp.br>.
+
+case $1 in
+  '')
+    echo "$0: No command.  Try '$0 --help' for more information." 1>&2
+    exit 1;
+    ;;
+  -h | --h*)
+    cat <<\EOF
+Usage: depcomp [--help] [--version] PROGRAM [ARGS]
+
+Run PROGRAMS ARGS to compile a file, generating dependencies
+as side-effects.
+
+Environment variables:
+  depmode     Dependency tracking mode.
+  source      Source file read by 'PROGRAMS ARGS'.
+  object      Object file output by 'PROGRAMS ARGS'.
+  DEPDIR      directory where to store dependencies.
+  depfile     Dependency file to output.
+  tmpdepfile  Temporary file to use when outputting dependencies.
+  libtool     Whether libtool is used (yes/no).
+
+Report bugs to <bug-automake@gnu.org>.
+EOF
+    exit $?
+    ;;
+  -v | --v*)
+    echo "depcomp $scriptversion"
+    exit $?
+    ;;
+esac
+
+# Get the directory component of the given path, and save it in the
+# global variables '$dir'.  Note that this directory component will
+# be either empty or ending with a '/' character.  This is deliberate.
+set_dir_from ()
+{
+  case $1 in
+    */*) dir=`echo "$1" | sed -e 's|/[^/]*$|/|'`;;
+      *) dir=;;
+  esac
+}
+
+# Get the suffix-stripped basename of the given path, and save it the
+# global variable '$base'.
+set_base_from ()
+{
+  base=`echo "$1" | sed -e 's|^.*/||' -e 's/\.[^.]*$//'`
+}
+
+# If no dependency file was actually created by the compiler invocation,
+# we still have to create a dummy depfile, to avoid errors with the
+# Makefile "include basename.Plo" scheme.
+make_dummy_depfile ()
+{
+  echo "#dummy" > "$depfile"
+}
+
+# Factor out some common post-processing of the generated depfile.
+# Requires the auxiliary global variable '$tmpdepfile' to be set.
+aix_post_process_depfile ()
+{
+  # If the compiler actually managed to produce a dependency file,
+  # post-process it.
+  if test -f "$tmpdepfile"; then
+    # Each line is of the form 'foo.o: dependency.h'.
+    # Do two passes, one to just change these to
+    #   $object: dependency.h
+    # and one to simply output
+    #   dependency.h:
+    # which is needed to avoid the deleted-header problem.
+    { sed -e "s,^.*\.[$lower]*:,$object:," < "$tmpdepfile"
+      sed -e "s,^.*\.[$lower]*:[$tab ]*,," -e 's,$,:,' < "$tmpdepfile"
+    } > "$depfile"
+    rm -f "$tmpdepfile"
+  else
+    make_dummy_depfile
+  fi
+}
+
+# A tabulation character.
+tab='	'
+# A newline character.
+nl='
+'
+# Character ranges might be problematic outside the C locale.
+# These definitions help.
+upper=ABCDEFGHIJKLMNOPQRSTUVWXYZ
+lower=abcdefghijklmnopqrstuvwxyz
+digits=0123456789
+alpha=${upper}${lower}
+
+if test -z "$depmode" || test -z "$source" || test -z "$object"; then
+  echo "depcomp: Variables source, object and depmode must be set" 1>&2
+  exit 1
+fi
+
+# Dependencies for sub/bar.o or sub/bar.obj go into sub/.deps/bar.Po.
+depfile=${depfile-`echo "$object" |
+  sed 's|[^\\/]*$|'${DEPDIR-.deps}'/&|;s|\.\([^.]*\)$|.P\1|;s|Pobj$|Po|'`}
+tmpdepfile=${tmpdepfile-`echo "$depfile" | sed 's/\.\([^.]*\)$/.T\1/'`}
+
+rm -f "$tmpdepfile"
+
+# Avoid interferences from the environment.
+gccflag= dashmflag=
+
+# Some modes work just like other modes, but use different flags.  We
+# parameterize here, but still list the modes in the big case below,
+# to make depend.m4 easier to write.  Note that we *cannot* use a case
+# here, because this file can only contain one case statement.
+if test "$depmode" = hp; then
+  # HP compiler uses -M and no extra arg.
+  gccflag=-M
+  depmode=gcc
+fi
+
+if test "$depmode" = dashXmstdout; then
+  # This is just like dashmstdout with a different argument.
+  dashmflag=-xM
+  depmode=dashmstdout
+fi
+
+cygpath_u="cygpath -u -f -"
+if test "$depmode" = msvcmsys; then
+  # This is just like msvisualcpp but w/o cygpath translation.
+  # Just convert the backslash-escaped backslashes to single forward
+  # slashes to satisfy depend.m4
+  cygpath_u='sed s,\\\\,/,g'
+  depmode=msvisualcpp
+fi
+
+if test "$depmode" = msvc7msys; then
+  # This is just like msvc7 but w/o cygpath translation.
+  # Just convert the backslash-escaped backslashes to single forward
+  # slashes to satisfy depend.m4
+  cygpath_u='sed s,\\\\,/,g'
+  depmode=msvc7
+fi
+
+if test "$depmode" = xlc; then
+  # IBM C/C++ Compilers xlc/xlC can output gcc-like dependency information.
+  gccflag=-qmakedep=gcc,-MF
+  depmode=gcc
+fi
+
+case "$depmode" in
+gcc3)
+## gcc 3 implements dependency tracking that does exactly what
+## we want.  Yay!  Note: for some reason libtool 1.4 doesn't like
+## it if -MD -MP comes after the -MF stuff.  Hmm.
+## Unfortunately, FreeBSD c89 acceptance of flags depends upon
+## the command line argument order; so add the flags where they
+## appear in depend2.am.  Note that the slowdown incurred here
+## affects only configure: in makefiles, %FASTDEP% shortcuts this.
+  for arg
+  do
+    case $arg in
+    -c) set fnord "$@" -MT "$object" -MD -MP -MF "$tmpdepfile" "$arg" ;;
+    *)  set fnord "$@" "$arg" ;;
+    esac
+    shift # fnord
+    shift # $arg
+  done
+  "$@"
+  stat=$?
+  if test $stat -ne 0; then
+    rm -f "$tmpdepfile"
+    exit $stat
+  fi
+  mv "$tmpdepfile" "$depfile"
+  ;;
+
+gcc)
+## Note that this doesn't just cater to obsosete pre-3.x GCC compilers.
+## but also to in-use compilers like IMB xlc/xlC and the HP C compiler.
+## (see the conditional assignment to $gccflag above).
+## There are various ways to get dependency output from gcc.  Here's
+## why we pick this rather obscure method:
+## - Don't want to use -MD because we'd like the dependencies to end
+##   up in a subdir.  Having to rename by hand is ugly.
+##   (We might end up doing this anyway to support other compilers.)
+## - The DEPENDENCIES_OUTPUT environment variable makes gcc act like
+##   -MM, not -M (despite what the docs say).  Also, it might not be
+##   supported by the other compilers which use the 'gcc' depmode.
+## - Using -M directly means running the compiler twice (even worse
+##   than renaming).
+  if test -z "$gccflag"; then
+    gccflag=-MD,
+  fi
+  "$@" -Wp,"$gccflag$tmpdepfile"
+  stat=$?
+  if test $stat -ne 0; then
+    rm -f "$tmpdepfile"
+    exit $stat
+  fi
+  rm -f "$depfile"
+  echo "$object : \\" > "$depfile"
+  # The second -e expression handles DOS-style file names with drive
+  # letters.
+  sed -e 's/^[^:]*: / /' \
+      -e 's/^['$alpha']:\/[^:]*: / /' < "$tmpdepfile" >> "$depfile"
+## This next piece of magic avoids the "deleted header file" problem.
+## The problem is that when a header file which appears in a .P file
+## is deleted, the dependency causes make to die (because there is
+## typically no way to rebuild the header).  We avoid this by adding
+## dummy dependencies for each header file.  Too bad gcc doesn't do
+## this for us directly.
+## Some versions of gcc put a space before the ':'.  On the theory
+## that the space means something, we add a space to the output as
+## well.  hp depmode also adds that space, but also prefixes the VPATH
+## to the object.  Take care to not repeat it in the output.
+## Some versions of the HPUX 10.20 sed can't process this invocation
+## correctly.  Breaking it into two sed invocations is a workaround.
+  tr ' ' "$nl" < "$tmpdepfile" \
+    | sed -e 's/^\\$//' -e '/^$/d' -e "s|.*$object$||" -e '/:$/d' \
+    | sed -e 's/$/ :/' >> "$depfile"
+  rm -f "$tmpdepfile"
+  ;;
+
+hp)
+  # This case exists only to let depend.m4 do its work.  It works by
+  # looking at the text of this script.  This case will never be run,
+  # since it is checked for above.
+  exit 1
+  ;;
+
+sgi)
+  if test "$libtool" = yes; then
+    "$@" "-Wp,-MDupdate,$tmpdepfile"
+  else
+    "$@" -MDupdate "$tmpdepfile"
+  fi
+  stat=$?
+  if test $stat -ne 0; then
+    rm -f "$tmpdepfile"
+    exit $stat
+  fi
+  rm -f "$depfile"
+
+  if test -f "$tmpdepfile"; then  # yes, the sourcefile depend on other files
+    echo "$object : \\" > "$depfile"
+    # Clip off the initial element (the dependent).  Don't try to be
+    # clever and replace this with sed code, as IRIX sed won't handle
+    # lines with more than a fixed number of characters (4096 in
+    # IRIX 6.2 sed, 8192 in IRIX 6.5).  We also remove comment lines;
+    # the IRIX cc adds comments like '#:fec' to the end of the
+    # dependency line.
+    tr ' ' "$nl" < "$tmpdepfile" \
+      | sed -e 's/^.*\.o://' -e 's/#.*$//' -e '/^$/ d' \
+      | tr "$nl" ' ' >> "$depfile"
+    echo >> "$depfile"
+    # The second pass generates a dummy entry for each header file.
+    tr ' ' "$nl" < "$tmpdepfile" \
+      | sed -e 's/^.*\.o://' -e 's/#.*$//' -e '/^$/ d' -e 's/$/:/' \
+      >> "$depfile"
+  else
+    make_dummy_depfile
+  fi
+  rm -f "$tmpdepfile"
+  ;;
+
+xlc)
+  # This case exists only to let depend.m4 do its work.  It works by
+  # looking at the text of this script.  This case will never be run,
+  # since it is checked for above.
+  exit 1
+  ;;
+
+aix)
+  # The C for AIX Compiler uses -M and outputs the dependencies
+  # in a .u file.  In older versions, this file always lives in the
+  # current directory.  Also, the AIX compiler puts '$object:' at the
+  # start of each line; $object doesn't have directory information.
+  # Version 6 uses the directory in both cases.
+  set_dir_from "$object"
+  set_base_from "$object"
+  if test "$libtool" = yes; then
+    tmpdepfile1=$dir$base.u
+    tmpdepfile2=$base.u
+    tmpdepfile3=$dir.libs/$base.u
+    "$@" -Wc,-M
+  else
+    tmpdepfile1=$dir$base.u
+    tmpdepfile2=$dir$base.u
+    tmpdepfile3=$dir$base.u
+    "$@" -M
+  fi
+  stat=$?
+  if test $stat -ne 0; then
+    rm -f "$tmpdepfile1" "$tmpdepfile2" "$tmpdepfile3"
+    exit $stat
+  fi
+
+  for tmpdepfile in "$tmpdepfile1" "$tmpdepfile2" "$tmpdepfile3"
+  do
+    test -f "$tmpdepfile" && break
+  done
+  aix_post_process_depfile
+  ;;
+
+tcc)
+  # tcc (Tiny C Compiler) understand '-MD -MF file' since version 0.9.26
+  # FIXME: That version still under development at the moment of writing.
+  #        Make that this statement remains true also for stable, released
+  #        versions.
+  # It will wrap lines (doesn't matter whether long or short) with a
+  # trailing '\', as in:
+  #
+  #   foo.o : \
+  #    foo.c \
+  #    foo.h \
+  #
+  # It will put a trailing '\' even on the last line, and will use leading
+  # spaces rather than leading tabs (at least since its commit 0394caf7
+  # "Emit spaces for -MD").
+  "$@" -MD -MF "$tmpdepfile"
+  stat=$?
+  if test $stat -ne 0; then
+    rm -f "$tmpdepfile"
+    exit $stat
+  fi
+  rm -f "$depfile"
+  # Each non-empty line is of the form 'foo.o : \' or ' dep.h \'.
+  # We have to change lines of the first kind to '$object: \'.
+  sed -e "s|.*:|$object :|" < "$tmpdepfile" > "$depfile"
+  # And for each line of the second kind, we have to emit a 'dep.h:'
+  # dummy dependency, to avoid the deleted-header problem.
+  sed -n -e 's|^  *\(.*\) *\\$|\1:|p' < "$tmpdepfile" >> "$depfile"
+  rm -f "$tmpdepfile"
+  ;;
+
+## The order of this option in the case statement is important, since the
+## shell code in configure will try each of these formats in the order
+## listed in this file.  A plain '-MD' option would be understood by many
+## compilers, so we must ensure this comes after the gcc and icc options.
+pgcc)
+  # Portland's C compiler understands '-MD'.
+  # Will always output deps to 'file.d' where file is the root name of the
+  # source file under compilation, even if file resides in a subdirectory.
+  # The object file name does not affect the name of the '.d' file.
+  # pgcc 10.2 will output
+  #    foo.o: sub/foo.c sub/foo.h
+  # and will wrap long lines using '\' :
+  #    foo.o: sub/foo.c ... \
+  #     sub/foo.h ... \
+  #     ...
+  set_dir_from "$object"
+  # Use the source, not the object, to determine the base name, since
+  # that's sadly what pgcc will do too.
+  set_base_from "$source"
+  tmpdepfile=$base.d
+
+  # For projects that build the same source file twice into different object
+  # files, the pgcc approach of using the *source* file root name can cause
+  # problems in parallel builds.  Use a locking strategy to avoid stomping on
+  # the same $tmpdepfile.
+  lockdir=$base.d-lock
+  trap "
+    echo '$0: caught signal, cleaning up...' >&2
+    rmdir '$lockdir'
+    exit 1
+  " 1 2 13 15
+  numtries=100
+  i=$numtries
+  while test $i -gt 0; do
+    # mkdir is a portable test-and-set.
+    if mkdir "$lockdir" 2>/dev/null; then
+      # This process acquired the lock.
+      "$@" -MD
+      stat=$?
+      # Release the lock.
+      rmdir "$lockdir"
+      break
+    else
+      # If the lock is being held by a different process, wait
+      # until the winning process is done or we timeout.
+      while test -d "$lockdir" && test $i -gt 0; do
+        sleep 1
+        i=`expr $i - 1`
+      done
+    fi
+    i=`expr $i - 1`
+  done
+  trap - 1 2 13 15
+  if test $i -le 0; then
+    echo "$0: failed to acquire lock after $numtries attempts" >&2
+    echo "$0: check lockdir '$lockdir'" >&2
+    exit 1
+  fi
+
+  if test $stat -ne 0; then
+    rm -f "$tmpdepfile"
+    exit $stat
+  fi
+  rm -f "$depfile"
+  # Each line is of the form `foo.o: dependent.h',
+  # or `foo.o: dep1.h dep2.h \', or ` dep3.h dep4.h \'.
+  # Do two passes, one to just change these to
+  # `$object: dependent.h' and one to simply `dependent.h:'.
+  sed "s,^[^:]*:,$object :," < "$tmpdepfile" > "$depfile"
+  # Some versions of the HPUX 10.20 sed can't process this invocation
+  # correctly.  Breaking it into two sed invocations is a workaround.
+  sed 's,^[^:]*: \(.*\)$,\1,;s/^\\$//;/^$/d;/:$/d' < "$tmpdepfile" \
+    | sed -e 's/$/ :/' >> "$depfile"
+  rm -f "$tmpdepfile"
+  ;;
+
+hp2)
+  # The "hp" stanza above does not work with aCC (C++) and HP's ia64
+  # compilers, which have integrated preprocessors.  The correct option
+  # to use with these is +Maked; it writes dependencies to a file named
+  # 'foo.d', which lands next to the object file, wherever that
+  # happens to be.
+  # Much of this is similar to the tru64 case; see comments there.
+  set_dir_from  "$object"
+  set_base_from "$object"
+  if test "$libtool" = yes; then
+    tmpdepfile1=$dir$base.d
+    tmpdepfile2=$dir.libs/$base.d
+    "$@" -Wc,+Maked
+  else
+    tmpdepfile1=$dir$base.d
+    tmpdepfile2=$dir$base.d
+    "$@" +Maked
+  fi
+  stat=$?
+  if test $stat -ne 0; then
+     rm -f "$tmpdepfile1" "$tmpdepfile2"
+     exit $stat
+  fi
+
+  for tmpdepfile in "$tmpdepfile1" "$tmpdepfile2"
+  do
+    test -f "$tmpdepfile" && break
+  done
+  if test -f "$tmpdepfile"; then
+    sed -e "s,^.*\.[$lower]*:,$object:," "$tmpdepfile" > "$depfile"
+    # Add 'dependent.h:' lines.
+    sed -ne '2,${
+               s/^ *//
+               s/ \\*$//
+               s/$/:/
+               p
+             }' "$tmpdepfile" >> "$depfile"
+  else
+    make_dummy_depfile
+  fi
+  rm -f "$tmpdepfile" "$tmpdepfile2"
+  ;;
+
+tru64)
+  # The Tru64 compiler uses -MD to generate dependencies as a side
+  # effect.  'cc -MD -o foo.o ...' puts the dependencies into 'foo.o.d'.
+  # At least on Alpha/Redhat 6.1, Compaq CCC V6.2-504 seems to put
+  # dependencies in 'foo.d' instead, so we check for that too.
+  # Subdirectories are respected.
+  set_dir_from  "$object"
+  set_base_from "$object"
+
+  if test "$libtool" = yes; then
+    # Libtool generates 2 separate objects for the 2 libraries.  These
+    # two compilations output dependencies in $dir.libs/$base.o.d and
+    # in $dir$base.o.d.  We have to check for both files, because
+    # one of the two compilations can be disabled.  We should prefer
+    # $dir$base.o.d over $dir.libs/$base.o.d because the latter is
+    # automatically cleaned when .libs/ is deleted, while ignoring
+    # the former would cause a distcleancheck panic.
+    tmpdepfile1=$dir$base.o.d          # libtool 1.5
+    tmpdepfile2=$dir.libs/$base.o.d    # Likewise.
+    tmpdepfile3=$dir.libs/$base.d      # Compaq CCC V6.2-504
+    "$@" -Wc,-MD
+  else
+    tmpdepfile1=$dir$base.d
+    tmpdepfile2=$dir$base.d
+    tmpdepfile3=$dir$base.d
+    "$@" -MD
+  fi
+
+  stat=$?
+  if test $stat -ne 0; then
+    rm -f "$tmpdepfile1" "$tmpdepfile2" "$tmpdepfile3"
+    exit $stat
+  fi
+
+  for tmpdepfile in "$tmpdepfile1" "$tmpdepfile2" "$tmpdepfile3"
+  do
+    test -f "$tmpdepfile" && break
+  done
+  # Same post-processing that is required for AIX mode.
+  aix_post_process_depfile
+  ;;
+
+msvc7)
+  if test "$libtool" = yes; then
+    showIncludes=-Wc,-showIncludes
+  else
+    showIncludes=-showIncludes
+  fi
+  "$@" $showIncludes > "$tmpdepfile"
+  stat=$?
+  grep -v '^Note: including file: ' "$tmpdepfile"
+  if test $stat -ne 0; then
+    rm -f "$tmpdepfile"
+    exit $stat
+  fi
+  rm -f "$depfile"
+  echo "$object : \\" > "$depfile"
+  # The first sed program below extracts the file names and escapes
+  # backslashes for cygpath.  The second sed program outputs the file
+  # name when reading, but also accumulates all include files in the
+  # hold buffer in order to output them again at the end.  This only
+  # works with sed implementations that can handle large buffers.
+  sed < "$tmpdepfile" -n '
+/^Note: including file:  *\(.*\)/ {
+  s//\1/
+  s/\\/\\\\/g
+  p
+}' | $cygpath_u | sort -u | sed -n '
+s/ /\\ /g
+s/\(.*\)/'"$tab"'\1 \\/p
+s/.\(.*\) \\/\1:/
+H
+$ {
+  s/.*/'"$tab"'/
+  G
+  p
+}' >> "$depfile"
+  echo >> "$depfile" # make sure the fragment doesn't end with a backslash
+  rm -f "$tmpdepfile"
+  ;;
+
+msvc7msys)
+  # This case exists only to let depend.m4 do its work.  It works by
+  # looking at the text of this script.  This case will never be run,
+  # since it is checked for above.
+  exit 1
+  ;;
+
+#nosideeffect)
+  # This comment above is used by automake to tell side-effect
+  # dependency tracking mechanisms from slower ones.
+
+dashmstdout)
+  # Important note: in order to support this mode, a compiler *must*
+  # always write the preprocessed file to stdout, regardless of -o.
+  "$@" || exit $?
+
+  # Remove the call to Libtool.
+  if test "$libtool" = yes; then
+    while test "X$1" != 'X--mode=compile'; do
+      shift
+    done
+    shift
+  fi
+
+  # Remove '-o $object'.
+  IFS=" "
+  for arg
+  do
+    case $arg in
+    -o)
+      shift
+      ;;
+    $object)
+      shift
+      ;;
+    *)
+      set fnord "$@" "$arg"
+      shift # fnord
+      shift # $arg
+      ;;
+    esac
+  done
+
+  test -z "$dashmflag" && dashmflag=-M
+  # Require at least two characters before searching for ':'
+  # in the target name.  This is to cope with DOS-style filenames:
+  # a dependency such as 'c:/foo/bar' could be seen as target 'c' otherwise.
+  "$@" $dashmflag |
+    sed "s|^[$tab ]*[^:$tab ][^:][^:]*:[$tab ]*|$object: |" > "$tmpdepfile"
+  rm -f "$depfile"
+  cat < "$tmpdepfile" > "$depfile"
+  # Some versions of the HPUX 10.20 sed can't process this sed invocation
+  # correctly.  Breaking it into two sed invocations is a workaround.
+  tr ' ' "$nl" < "$tmpdepfile" \
+    | sed -e 's/^\\$//' -e '/^$/d' -e '/:$/d' \
+    | sed -e 's/$/ :/' >> "$depfile"
+  rm -f "$tmpdepfile"
+  ;;
+
+dashXmstdout)
+  # This case only exists to satisfy depend.m4.  It is never actually
+  # run, as this mode is specially recognized in the preamble.
+  exit 1
+  ;;
+
+makedepend)
+  "$@" || exit $?
+  # Remove any Libtool call
+  if test "$libtool" = yes; then
+    while test "X$1" != 'X--mode=compile'; do
+      shift
+    done
+    shift
+  fi
+  # X makedepend
+  shift
+  cleared=no eat=no
+  for arg
+  do
+    case $cleared in
+    no)
+      set ""; shift
+      cleared=yes ;;
+    esac
+    if test $eat = yes; then
+      eat=no
+      continue
+    fi
+    case "$arg" in
+    -D*|-I*)
+      set fnord "$@" "$arg"; shift ;;
+    # Strip any option that makedepend may not understand.  Remove
+    # the object too, otherwise makedepend will parse it as a source file.
+    -arch)
+      eat=yes ;;
+    -*|$object)
+      ;;
+    *)
+      set fnord "$@" "$arg"; shift ;;
+    esac
+  done
+  obj_suffix=`echo "$object" | sed 's/^.*\././'`
+  touch "$tmpdepfile"
+  ${MAKEDEPEND-makedepend} -o"$obj_suffix" -f"$tmpdepfile" "$@"
+  rm -f "$depfile"
+  # makedepend may prepend the VPATH from the source file name to the object.
+  # No need to regex-escape $object, excess matching of '.' is harmless.
+  sed "s|^.*\($object *:\)|\1|" "$tmpdepfile" > "$depfile"
+  # Some versions of the HPUX 10.20 sed can't process the last invocation
+  # correctly.  Breaking it into two sed invocations is a workaround.
+  sed '1,2d' "$tmpdepfile" \
+    | tr ' ' "$nl" \
+    | sed -e 's/^\\$//' -e '/^$/d' -e '/:$/d' \
+    | sed -e 's/$/ :/' >> "$depfile"
+  rm -f "$tmpdepfile" "$tmpdepfile".bak
+  ;;
+
+cpp)
+  # Important note: in order to support this mode, a compiler *must*
+  # always write the preprocessed file to stdout.
+  "$@" || exit $?
+
+  # Remove the call to Libtool.
+  if test "$libtool" = yes; then
+    while test "X$1" != 'X--mode=compile'; do
+      shift
+    done
+    shift
+  fi
+
+  # Remove '-o $object'.
+  IFS=" "
+  for arg
+  do
+    case $arg in
+    -o)
+      shift
+      ;;
+    $object)
+      shift
+      ;;
+    *)
+      set fnord "$@" "$arg"
+      shift # fnord
+      shift # $arg
+      ;;
+    esac
+  done
+
+  "$@" -E \
+    | sed -n -e '/^# [0-9][0-9]* "\([^"]*\)".*/ s:: \1 \\:p' \
+             -e '/^#line [0-9][0-9]* "\([^"]*\)".*/ s:: \1 \\:p' \
+    | sed '$ s: \\$::' > "$tmpdepfile"
+  rm -f "$depfile"
+  echo "$object : \\" > "$depfile"
+  cat < "$tmpdepfile" >> "$depfile"
+  sed < "$tmpdepfile" '/^$/d;s/^ //;s/ \\$//;s/$/ :/' >> "$depfile"
+  rm -f "$tmpdepfile"
+  ;;
+
+msvisualcpp)
+  # Important note: in order to support this mode, a compiler *must*
+  # always write the preprocessed file to stdout.
+  "$@" || exit $?
+
+  # Remove the call to Libtool.
+  if test "$libtool" = yes; then
+    while test "X$1" != 'X--mode=compile'; do
+      shift
+    done
+    shift
+  fi
+
+  IFS=" "
+  for arg
+  do
+    case "$arg" in
+    -o)
+      shift
+      ;;
+    $object)
+      shift
+      ;;
+    "-Gm"|"/Gm"|"-Gi"|"/Gi"|"-ZI"|"/ZI")
+        set fnord "$@"
+        shift
+        shift
+        ;;
+    *)
+        set fnord "$@" "$arg"
+        shift
+        shift
+        ;;
+    esac
+  done
+  "$@" -E 2>/dev/null |
+  sed -n '/^#line [0-9][0-9]* "\([^"]*\)"/ s::\1:p' | $cygpath_u | sort -u > "$tmpdepfile"
+  rm -f "$depfile"
+  echo "$object : \\" > "$depfile"
+  sed < "$tmpdepfile" -n -e 's% %\\ %g' -e '/^\(.*\)$/ s::'"$tab"'\1 \\:p' >> "$depfile"
+  echo "$tab" >> "$depfile"
+  sed < "$tmpdepfile" -n -e 's% %\\ %g' -e '/^\(.*\)$/ s::\1\::p' >> "$depfile"
+  rm -f "$tmpdepfile"
+  ;;
+
+msvcmsys)
+  # This case exists only to let depend.m4 do its work.  It works by
+  # looking at the text of this script.  This case will never be run,
+  # since it is checked for above.
+  exit 1
+  ;;
+
+none)
+  exec "$@"
+  ;;
+
+*)
+  echo "Unknown depmode $depmode" 1>&2
+  exit 1
+  ;;
+esac
+
+exit 0
+
+# Local Variables:
+# mode: shell-script
+# sh-indentation: 2
+# eval: (add-hook 'write-file-hooks 'time-stamp)
+# time-stamp-start: "scriptversion="
+# time-stamp-format: "%:y-%02m-%02d.%02H"
+# time-stamp-time-zone: "UTC"
+# time-stamp-end: "; # UTC"
+# End:
diff --git a/missing b/missing
new file mode 100755
index 00000000..f62bbae3
--- /dev/null
+++ b/missing
@@ -0,0 +1,215 @@
+#! /bin/sh
+# Common wrapper for a few potentially missing GNU programs.
+
+scriptversion=2013-10-28.13; # UTC
+
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+# Originally written by Fran,cois Pinard <pinard@iro.umontreal.ca>, 1996.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2, or (at your option)
+# any later version.
+
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+# As a special exception to the GNU General Public License, if you
+# distribute this file as part of a program that contains a
+# configuration script generated by Autoconf, you may include it under
+# the same distribution terms that you use for the rest of that program.
+
+if test $# -eq 0; then
+  echo 1>&2 "Try '$0 --help' for more information"
+  exit 1
+fi
+
+case $1 in
+
+  --is-lightweight)
+    # Used by our autoconf macros to check whether the available missing
+    # script is modern enough.
+    exit 0
+    ;;
+
+  --run)
+    # Back-compat with the calling convention used by older automake.
+    shift
+    ;;
+
+  -h|--h|--he|--hel|--help)
+    echo "\
+$0 [OPTION]... PROGRAM [ARGUMENT]...
+
+Run 'PROGRAM [ARGUMENT]...', returning a proper advice when this fails due
+to PROGRAM being missing or too old.
+
+Options:
+  -h, --help      display this help and exit
+  -v, --version   output version information and exit
+
+Supported PROGRAM values:
+  aclocal   autoconf  autoheader   autom4te  automake  makeinfo
+  bison     yacc      flex         lex       help2man
+
+Version suffixes to PROGRAM as well as the prefixes 'gnu-', 'gnu', and
+'g' are ignored when checking the name.
+
+Send bug reports to <bug-automake@gnu.org>."
+    exit $?
+    ;;
+
+  -v|--v|--ve|--ver|--vers|--versi|--versio|--version)
+    echo "missing $scriptversion (GNU Automake)"
+    exit $?
+    ;;
+
+  -*)
+    echo 1>&2 "$0: unknown '$1' option"
+    echo 1>&2 "Try '$0 --help' for more information"
+    exit 1
+    ;;
+
+esac
+
+# Run the given program, remember its exit status.
+"$@"; st=$?
+
+# If it succeeded, we are done.
+test $st -eq 0 && exit 0
+
+# Also exit now if we it failed (or wasn't found), and '--version' was
+# passed; such an option is passed most likely to detect whether the
+# program is present and works.
+case $2 in --version|--help) exit $st;; esac
+
+# Exit code 63 means version mismatch.  This often happens when the user
+# tries to use an ancient version of a tool on a file that requires a
+# minimum version.
+if test $st -eq 63; then
+  msg="probably too old"
+elif test $st -eq 127; then
+  # Program was missing.
+  msg="missing on your system"
+else
+  # Program was found and executed, but failed.  Give up.
+  exit $st
+fi
+
+perl_URL=http://www.perl.org/
+flex_URL=http://flex.sourceforge.net/
+gnu_software_URL=http://www.gnu.org/software
+
+program_details ()
+{
+  case $1 in
+    aclocal|automake)
+      echo "The '$1' program is part of the GNU Automake package:"
+      echo "<$gnu_software_URL/automake>"
+      echo "It also requires GNU Autoconf, GNU m4 and Perl in order to run:"
+      echo "<$gnu_software_URL/autoconf>"
+      echo "<$gnu_software_URL/m4/>"
+      echo "<$perl_URL>"
+      ;;
+    autoconf|autom4te|autoheader)
+      echo "The '$1' program is part of the GNU Autoconf package:"
+      echo "<$gnu_software_URL/autoconf/>"
+      echo "It also requires GNU m4 and Perl in order to run:"
+      echo "<$gnu_software_URL/m4/>"
+      echo "<$perl_URL>"
+      ;;
+  esac
+}
+
+give_advice ()
+{
+  # Normalize program name to check for.
+  normalized_program=`echo "$1" | sed '
+    s/^gnu-//; t
+    s/^gnu//; t
+    s/^g//; t'`
+
+  printf '%s\n' "'$1' is $msg."
+
+  configure_deps="'configure.ac' or m4 files included by 'configure.ac'"
+  case $normalized_program in
+    autoconf*)
+      echo "You should only need it if you modified 'configure.ac',"
+      echo "or m4 files included by it."
+      program_details 'autoconf'
+      ;;
+    autoheader*)
+      echo "You should only need it if you modified 'acconfig.h' or"
+      echo "$configure_deps."
+      program_details 'autoheader'
+      ;;
+    automake*)
+      echo "You should only need it if you modified 'Makefile.am' or"
+      echo "$configure_deps."
+      program_details 'automake'
+      ;;
+    aclocal*)
+      echo "You should only need it if you modified 'acinclude.m4' or"
+      echo "$configure_deps."
+      program_details 'aclocal'
+      ;;
+   autom4te*)
+      echo "You might have modified some maintainer files that require"
+      echo "the 'autom4te' program to be rebuilt."
+      program_details 'autom4te'
+      ;;
+    bison*|yacc*)
+      echo "You should only need it if you modified a '.y' file."
+      echo "You may want to install the GNU Bison package:"
+      echo "<$gnu_software_URL/bison/>"
+      ;;
+    lex*|flex*)
+      echo "You should only need it if you modified a '.l' file."
+      echo "You may want to install the Fast Lexical Analyzer package:"
+      echo "<$flex_URL>"
+      ;;
+    help2man*)
+      echo "You should only need it if you modified a dependency" \
+           "of a man page."
+      echo "You may want to install the GNU Help2man package:"
+      echo "<$gnu_software_URL/help2man/>"
+    ;;
+    makeinfo*)
+      echo "You should only need it if you modified a '.texi' file, or"
+      echo "any other file indirectly affecting the aspect of the manual."
+      echo "You might want to install the Texinfo package:"
+      echo "<$gnu_software_URL/texinfo/>"
+      echo "The spurious makeinfo call might also be the consequence of"
+      echo "using a buggy 'make' (AIX, DU, IRIX), in which case you might"
+      echo "want to install GNU make:"
+      echo "<$gnu_software_URL/make/>"
+      ;;
+    *)
+      echo "You might have modified some files without having the proper"
+      echo "tools for further handling them.  Check the 'README' file, it"
+      echo "often tells you about the needed prerequisites for installing"
+      echo "this package.  You may also peek at any GNU archive site, in"
+      echo "case some other package contains this missing '$1' program."
+      ;;
+  esac
+}
+
+give_advice "$1" | sed -e '1s/^/WARNING: /' \
+                       -e '2,$s/^/         /' >&2
+
+# Propagate the correct exit status (expected to be 127 for a program
+# not found, 63 for a program that failed due to version mismatch).
+exit $st
+
+# Local variables:
+# eval: (add-hook 'write-file-hooks 'time-stamp)
+# time-stamp-start: "scriptversion="
+# time-stamp-format: "%:y-%02m-%02d.%02H"
+# time-stamp-time-zone: "UTC"
+# time-stamp-end: "; # UTC"
+# End:

From d4a93ec7b4e18a45be42401aa538b48d40d66994 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 4 Mar 2015 11:58:45 +0000
Subject: [PATCH 013/429] missing

---
 aclocal.m4 | 1153 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 1153 insertions(+)
 create mode 100644 aclocal.m4

diff --git a/aclocal.m4 b/aclocal.m4
new file mode 100644
index 00000000..34cdc48d
--- /dev/null
+++ b/aclocal.m4
@@ -0,0 +1,1153 @@
+# generated automatically by aclocal 1.15 -*- Autoconf -*-
+
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY, to the extent permitted by law; without
+# even the implied warranty of MERCHANTABILITY or FITNESS FOR A
+# PARTICULAR PURPOSE.
+
+m4_ifndef([AC_CONFIG_MACRO_DIRS], [m4_defun([_AM_CONFIG_MACRO_DIRS], [])m4_defun([AC_CONFIG_MACRO_DIRS], [_AM_CONFIG_MACRO_DIRS($@)])])
+m4_ifndef([AC_AUTOCONF_VERSION],
+  [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
+m4_if(m4_defn([AC_AUTOCONF_VERSION]), [2.69],,
+[m4_warning([this file was generated for autoconf 2.69.
+You have another version of autoconf.  It may work, but is not guaranteed to.
+If you have problems, you may need to regenerate the build system entirely.
+To do so, use the procedure documented by the package, typically 'autoreconf'.])])
+
+# Copyright (C) 2002-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# AM_AUTOMAKE_VERSION(VERSION)
+# ----------------------------
+# Automake X.Y traces this macro to ensure aclocal.m4 has been
+# generated from the m4 files accompanying Automake X.Y.
+# (This private macro should not be called outside this file.)
+AC_DEFUN([AM_AUTOMAKE_VERSION],
+[am__api_version='1.15'
+dnl Some users find AM_AUTOMAKE_VERSION and mistake it for a way to
+dnl require some minimum version.  Point them to the right macro.
+m4_if([$1], [1.15], [],
+      [AC_FATAL([Do not call $0, use AM_INIT_AUTOMAKE([$1]).])])dnl
+])
+
+# _AM_AUTOCONF_VERSION(VERSION)
+# -----------------------------
+# aclocal traces this macro to find the Autoconf version.
+# This is a private macro too.  Using m4_define simplifies
+# the logic in aclocal, which can simply ignore this definition.
+m4_define([_AM_AUTOCONF_VERSION], [])
+
+# AM_SET_CURRENT_AUTOMAKE_VERSION
+# -------------------------------
+# Call AM_AUTOMAKE_VERSION and AM_AUTOMAKE_VERSION so they can be traced.
+# This function is AC_REQUIREd by AM_INIT_AUTOMAKE.
+AC_DEFUN([AM_SET_CURRENT_AUTOMAKE_VERSION],
+[AM_AUTOMAKE_VERSION([1.15])dnl
+m4_ifndef([AC_AUTOCONF_VERSION],
+  [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
+_AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
+
+# AM_AUX_DIR_EXPAND                                         -*- Autoconf -*-
+
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# For projects using AC_CONFIG_AUX_DIR([foo]), Autoconf sets
+# $ac_aux_dir to '$srcdir/foo'.  In other projects, it is set to
+# '$srcdir', '$srcdir/..', or '$srcdir/../..'.
+#
+# Of course, Automake must honor this variable whenever it calls a
+# tool from the auxiliary directory.  The problem is that $srcdir (and
+# therefore $ac_aux_dir as well) can be either absolute or relative,
+# depending on how configure is run.  This is pretty annoying, since
+# it makes $ac_aux_dir quite unusable in subdirectories: in the top
+# source directory, any form will work fine, but in subdirectories a
+# relative path needs to be adjusted first.
+#
+# $ac_aux_dir/missing
+#    fails when called from a subdirectory if $ac_aux_dir is relative
+# $top_srcdir/$ac_aux_dir/missing
+#    fails if $ac_aux_dir is absolute,
+#    fails when called from a subdirectory in a VPATH build with
+#          a relative $ac_aux_dir
+#
+# The reason of the latter failure is that $top_srcdir and $ac_aux_dir
+# are both prefixed by $srcdir.  In an in-source build this is usually
+# harmless because $srcdir is '.', but things will broke when you
+# start a VPATH build or use an absolute $srcdir.
+#
+# So we could use something similar to $top_srcdir/$ac_aux_dir/missing,
+# iff we strip the leading $srcdir from $ac_aux_dir.  That would be:
+#   am_aux_dir='\$(top_srcdir)/'`expr "$ac_aux_dir" : "$srcdir//*\(.*\)"`
+# and then we would define $MISSING as
+#   MISSING="\${SHELL} $am_aux_dir/missing"
+# This will work as long as MISSING is not called from configure, because
+# unfortunately $(top_srcdir) has no meaning in configure.
+# However there are other variables, like CC, which are often used in
+# configure, and could therefore not use this "fixed" $ac_aux_dir.
+#
+# Another solution, used here, is to always expand $ac_aux_dir to an
+# absolute PATH.  The drawback is that using absolute paths prevent a
+# configured tree to be moved without reconfiguration.
+
+AC_DEFUN([AM_AUX_DIR_EXPAND],
+[AC_REQUIRE([AC_CONFIG_AUX_DIR_DEFAULT])dnl
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
+])
+
+# AM_CONDITIONAL                                            -*- Autoconf -*-
+
+# Copyright (C) 1997-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# AM_CONDITIONAL(NAME, SHELL-CONDITION)
+# -------------------------------------
+# Define a conditional.
+AC_DEFUN([AM_CONDITIONAL],
+[AC_PREREQ([2.52])dnl
+ m4_if([$1], [TRUE],  [AC_FATAL([$0: invalid condition: $1])],
+       [$1], [FALSE], [AC_FATAL([$0: invalid condition: $1])])dnl
+AC_SUBST([$1_TRUE])dnl
+AC_SUBST([$1_FALSE])dnl
+_AM_SUBST_NOTMAKE([$1_TRUE])dnl
+_AM_SUBST_NOTMAKE([$1_FALSE])dnl
+m4_define([_AM_COND_VALUE_$1], [$2])dnl
+if $2; then
+  $1_TRUE=
+  $1_FALSE='#'
+else
+  $1_TRUE='#'
+  $1_FALSE=
+fi
+AC_CONFIG_COMMANDS_PRE(
+[if test -z "${$1_TRUE}" && test -z "${$1_FALSE}"; then
+  AC_MSG_ERROR([[conditional "$1" was never defined.
+Usually this means the macro was only invoked conditionally.]])
+fi])])
+
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+
+# There are a few dirty hacks below to avoid letting 'AC_PROG_CC' be
+# written in clear, in which case automake, when reading aclocal.m4,
+# will think it sees a *use*, and therefore will trigger all it's
+# C support machinery.  Also note that it means that autoscan, seeing
+# CC etc. in the Makefile, will ask for an AC_PROG_CC use...
+
+
+# _AM_DEPENDENCIES(NAME)
+# ----------------------
+# See how the compiler implements dependency checking.
+# NAME is "CC", "CXX", "OBJC", "OBJCXX", "UPC", or "GJC".
+# We try a few techniques and use that to set a single cache variable.
+#
+# We don't AC_REQUIRE the corresponding AC_PROG_CC since the latter was
+# modified to invoke _AM_DEPENDENCIES(CC); we would have a circular
+# dependency, and given that the user is not expected to run this macro,
+# just rely on AC_PROG_CC.
+AC_DEFUN([_AM_DEPENDENCIES],
+[AC_REQUIRE([AM_SET_DEPDIR])dnl
+AC_REQUIRE([AM_OUTPUT_DEPENDENCY_COMMANDS])dnl
+AC_REQUIRE([AM_MAKE_INCLUDE])dnl
+AC_REQUIRE([AM_DEP_TRACK])dnl
+
+m4_if([$1], [CC],   [depcc="$CC"   am_compiler_list=],
+      [$1], [CXX],  [depcc="$CXX"  am_compiler_list=],
+      [$1], [OBJC], [depcc="$OBJC" am_compiler_list='gcc3 gcc'],
+      [$1], [OBJCXX], [depcc="$OBJCXX" am_compiler_list='gcc3 gcc'],
+      [$1], [UPC],  [depcc="$UPC"  am_compiler_list=],
+      [$1], [GCJ],  [depcc="$GCJ"  am_compiler_list='gcc3 gcc'],
+                    [depcc="$$1"   am_compiler_list=])
+
+AC_CACHE_CHECK([dependency style of $depcc],
+               [am_cv_$1_dependencies_compiler_type],
+[if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
+  # We make a subdir and do the tests there.  Otherwise we can end up
+  # making bogus files that we don't know about and never remove.  For
+  # instance it was reported that on HP-UX the gcc test will end up
+  # making a dummy file named 'D' -- because '-MD' means "put the output
+  # in D".
+  rm -rf conftest.dir
+  mkdir conftest.dir
+  # Copy depcomp to subdir because otherwise we won't find it if we're
+  # using a relative directory.
+  cp "$am_depcomp" conftest.dir
+  cd conftest.dir
+  # We will build objects and dependencies in a subdirectory because
+  # it helps to detect inapplicable dependency modes.  For instance
+  # both Tru64's cc and ICC support -MD to output dependencies as a
+  # side effect of compilation, but ICC will put the dependencies in
+  # the current directory while Tru64 will put them in the object
+  # directory.
+  mkdir sub
+
+  am_cv_$1_dependencies_compiler_type=none
+  if test "$am_compiler_list" = ""; then
+     am_compiler_list=`sed -n ['s/^#*\([a-zA-Z0-9]*\))$/\1/p'] < ./depcomp`
+  fi
+  am__universal=false
+  m4_case([$1], [CC],
+    [case " $depcc " in #(
+     *\ -arch\ *\ -arch\ *) am__universal=true ;;
+     esac],
+    [CXX],
+    [case " $depcc " in #(
+     *\ -arch\ *\ -arch\ *) am__universal=true ;;
+     esac])
+
+  for depmode in $am_compiler_list; do
+    # Setup a source with many dependencies, because some compilers
+    # like to wrap large dependency lists on column 80 (with \), and
+    # we should not choose a depcomp mode which is confused by this.
+    #
+    # We need to recreate these files for each test, as the compiler may
+    # overwrite some of them when testing with obscure command lines.
+    # This happens at least with the AIX C compiler.
+    : > sub/conftest.c
+    for i in 1 2 3 4 5 6; do
+      echo '#include "conftst'$i'.h"' >> sub/conftest.c
+      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
+      # Solaris 10 /bin/sh.
+      echo '/* dummy */' > sub/conftst$i.h
+    done
+    echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
+
+    # We check with '-c' and '-o' for the sake of the "dashmstdout"
+    # mode.  It turns out that the SunPro C++ compiler does not properly
+    # handle '-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs.
+    am__obj=sub/conftest.${OBJEXT-o}
+    am__minus_obj="-o $am__obj"
+    case $depmode in
+    gcc)
+      # This depmode causes a compiler race in universal mode.
+      test "$am__universal" = false || continue
+      ;;
+    nosideeffect)
+      # After this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested.
+      if test "x$enable_dependency_tracking" = xyes; then
+	continue
+      else
+	break
+      fi
+      ;;
+    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
+      # This compiler won't grok '-c -o', but also, the minuso test has
+      # not run yet.  These depmodes are late enough in the game, and
+      # so weak that their functioning should not be impacted.
+      am__obj=conftest.${OBJEXT-o}
+      am__minus_obj=
+      ;;
+    none) break ;;
+    esac
+    if depmode=$depmode \
+       source=sub/conftest.c object=$am__obj \
+       depfile=sub/conftest.Po tmpdepfile=sub/conftest.TPo \
+       $SHELL ./depcomp $depcc -c $am__minus_obj sub/conftest.c \
+         >/dev/null 2>conftest.err &&
+       grep sub/conftst1.h sub/conftest.Po > /dev/null 2>&1 &&
+       grep sub/conftst6.h sub/conftest.Po > /dev/null 2>&1 &&
+       grep $am__obj sub/conftest.Po > /dev/null 2>&1 &&
+       ${MAKE-make} -s -f confmf > /dev/null 2>&1; then
+      # icc doesn't choke on unknown options, it will just issue warnings
+      # or remarks (even with -Werror).  So we grep stderr for any message
+      # that says an option was ignored or not supported.
+      # When given -MP, icc 7.0 and 7.1 complain thusly:
+      #   icc: Command line warning: ignoring option '-M'; no argument required
+      # The diagnosis changed in icc 8.0:
+      #   icc: Command line remark: option '-MP' not supported
+      if (grep 'ignoring option' conftest.err ||
+          grep 'not supported' conftest.err) >/dev/null 2>&1; then :; else
+        am_cv_$1_dependencies_compiler_type=$depmode
+        break
+      fi
+    fi
+  done
+
+  cd ..
+  rm -rf conftest.dir
+else
+  am_cv_$1_dependencies_compiler_type=none
+fi
+])
+AC_SUBST([$1DEPMODE], [depmode=$am_cv_$1_dependencies_compiler_type])
+AM_CONDITIONAL([am__fastdep$1], [
+  test "x$enable_dependency_tracking" != xno \
+  && test "$am_cv_$1_dependencies_compiler_type" = gcc3])
+])
+
+
+# AM_SET_DEPDIR
+# -------------
+# Choose a directory name for dependency files.
+# This macro is AC_REQUIREd in _AM_DEPENDENCIES.
+AC_DEFUN([AM_SET_DEPDIR],
+[AC_REQUIRE([AM_SET_LEADING_DOT])dnl
+AC_SUBST([DEPDIR], ["${am__leading_dot}deps"])dnl
+])
+
+
+# AM_DEP_TRACK
+# ------------
+AC_DEFUN([AM_DEP_TRACK],
+[AC_ARG_ENABLE([dependency-tracking], [dnl
+AS_HELP_STRING(
+  [--enable-dependency-tracking],
+  [do not reject slow dependency extractors])
+AS_HELP_STRING(
+  [--disable-dependency-tracking],
+  [speeds up one-time build])])
+if test "x$enable_dependency_tracking" != xno; then
+  am_depcomp="$ac_aux_dir/depcomp"
+  AMDEPBACKSLASH='\'
+  am__nodep='_no'
+fi
+AM_CONDITIONAL([AMDEP], [test "x$enable_dependency_tracking" != xno])
+AC_SUBST([AMDEPBACKSLASH])dnl
+_AM_SUBST_NOTMAKE([AMDEPBACKSLASH])dnl
+AC_SUBST([am__nodep])dnl
+_AM_SUBST_NOTMAKE([am__nodep])dnl
+])
+
+# Generate code to set up dependency tracking.              -*- Autoconf -*-
+
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+
+# _AM_OUTPUT_DEPENDENCY_COMMANDS
+# ------------------------------
+AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
+[{
+  # Older Autoconf quotes --file arguments for eval, but not when files
+  # are listed without --file.  Let's play safe and only enable the eval
+  # if we detect the quoting.
+  case $CONFIG_FILES in
+  *\'*) eval set x "$CONFIG_FILES" ;;
+  *)   set x $CONFIG_FILES ;;
+  esac
+  shift
+  for mf
+  do
+    # Strip MF so we end up with the name of the file.
+    mf=`echo "$mf" | sed -e 's/:.*$//'`
+    # Check whether this is an Automake generated Makefile or not.
+    # We used to match only the files named 'Makefile.in', but
+    # some people rename them; so instead we look at the file content.
+    # Grep'ing the first line is not enough: some people post-process
+    # each Makefile.in and add a new line on top of each file to say so.
+    # Grep'ing the whole file is not good either: AIX grep has a line
+    # limit of 2048, but all sed's we know have understand at least 4000.
+    if sed -n 's,^#.*generated by automake.*,X,p' "$mf" | grep X >/dev/null 2>&1; then
+      dirpart=`AS_DIRNAME("$mf")`
+    else
+      continue
+    fi
+    # Extract the definition of DEPDIR, am__include, and am__quote
+    # from the Makefile without running 'make'.
+    DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
+    test -z "$DEPDIR" && continue
+    am__include=`sed -n 's/^am__include = //p' < "$mf"`
+    test -z "$am__include" && continue
+    am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
+    # Find all dependency output files, they are included files with
+    # $(DEPDIR) in their names.  We invoke sed twice because it is the
+    # simplest approach to changing $(DEPDIR) to its actual value in the
+    # expansion.
+    for file in `sed -n "
+      s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
+	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
+      # Make sure the directory exists.
+      test -f "$dirpart/$file" && continue
+      fdir=`AS_DIRNAME(["$file"])`
+      AS_MKDIR_P([$dirpart/$fdir])
+      # echo "creating $dirpart/$file"
+      echo '# dummy' > "$dirpart/$file"
+    done
+  done
+}
+])# _AM_OUTPUT_DEPENDENCY_COMMANDS
+
+
+# AM_OUTPUT_DEPENDENCY_COMMANDS
+# -----------------------------
+# This macro should only be invoked once -- use via AC_REQUIRE.
+#
+# This code is only required when automatic dependency tracking
+# is enabled.  FIXME.  This creates each '.P' file that we will
+# need in order to bootstrap the dependency handling code.
+AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
+[AC_CONFIG_COMMANDS([depfiles],
+     [test x"$AMDEP_TRUE" != x"" || _AM_OUTPUT_DEPENDENCY_COMMANDS],
+     [AMDEP_TRUE="$AMDEP_TRUE" ac_aux_dir="$ac_aux_dir"])
+])
+
+# Do all the work for Automake.                             -*- Autoconf -*-
+
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# This macro actually does too much.  Some checks are only needed if
+# your package does certain things.  But this isn't really a big deal.
+
+dnl Redefine AC_PROG_CC to automatically invoke _AM_PROG_CC_C_O.
+m4_define([AC_PROG_CC],
+m4_defn([AC_PROG_CC])
+[_AM_PROG_CC_C_O
+])
+
+# AM_INIT_AUTOMAKE(PACKAGE, VERSION, [NO-DEFINE])
+# AM_INIT_AUTOMAKE([OPTIONS])
+# -----------------------------------------------
+# The call with PACKAGE and VERSION arguments is the old style
+# call (pre autoconf-2.50), which is being phased out.  PACKAGE
+# and VERSION should now be passed to AC_INIT and removed from
+# the call to AM_INIT_AUTOMAKE.
+# We support both call styles for the transition.  After
+# the next Automake release, Autoconf can make the AC_INIT
+# arguments mandatory, and then we can depend on a new Autoconf
+# release and drop the old call support.
+AC_DEFUN([AM_INIT_AUTOMAKE],
+[AC_PREREQ([2.65])dnl
+dnl Autoconf wants to disallow AM_ names.  We explicitly allow
+dnl the ones we care about.
+m4_pattern_allow([^AM_[A-Z]+FLAGS$])dnl
+AC_REQUIRE([AM_SET_CURRENT_AUTOMAKE_VERSION])dnl
+AC_REQUIRE([AC_PROG_INSTALL])dnl
+if test "`cd $srcdir && pwd`" != "`pwd`"; then
+  # Use -I$(srcdir) only when $(srcdir) != ., so that make's output
+  # is not polluted with repeated "-I."
+  AC_SUBST([am__isrc], [' -I$(srcdir)'])_AM_SUBST_NOTMAKE([am__isrc])dnl
+  # test to see if srcdir already configured
+  if test -f $srcdir/config.status; then
+    AC_MSG_ERROR([source directory already configured; run "make distclean" there first])
+  fi
+fi
+
+# test whether we have cygpath
+if test -z "$CYGPATH_W"; then
+  if (cygpath --version) >/dev/null 2>/dev/null; then
+    CYGPATH_W='cygpath -w'
+  else
+    CYGPATH_W=echo
+  fi
+fi
+AC_SUBST([CYGPATH_W])
+
+# Define the identity of the package.
+dnl Distinguish between old-style and new-style calls.
+m4_ifval([$2],
+[AC_DIAGNOSE([obsolete],
+             [$0: two- and three-arguments forms are deprecated.])
+m4_ifval([$3], [_AM_SET_OPTION([no-define])])dnl
+ AC_SUBST([PACKAGE], [$1])dnl
+ AC_SUBST([VERSION], [$2])],
+[_AM_SET_OPTIONS([$1])dnl
+dnl Diagnose old-style AC_INIT with new-style AM_AUTOMAKE_INIT.
+m4_if(
+  m4_ifdef([AC_PACKAGE_NAME], [ok]):m4_ifdef([AC_PACKAGE_VERSION], [ok]),
+  [ok:ok],,
+  [m4_fatal([AC_INIT should be called with package and version arguments])])dnl
+ AC_SUBST([PACKAGE], ['AC_PACKAGE_TARNAME'])dnl
+ AC_SUBST([VERSION], ['AC_PACKAGE_VERSION'])])dnl
+
+_AM_IF_OPTION([no-define],,
+[AC_DEFINE_UNQUOTED([PACKAGE], ["$PACKAGE"], [Name of package])
+ AC_DEFINE_UNQUOTED([VERSION], ["$VERSION"], [Version number of package])])dnl
+
+# Some tools Automake needs.
+AC_REQUIRE([AM_SANITY_CHECK])dnl
+AC_REQUIRE([AC_ARG_PROGRAM])dnl
+AM_MISSING_PROG([ACLOCAL], [aclocal-${am__api_version}])
+AM_MISSING_PROG([AUTOCONF], [autoconf])
+AM_MISSING_PROG([AUTOMAKE], [automake-${am__api_version}])
+AM_MISSING_PROG([AUTOHEADER], [autoheader])
+AM_MISSING_PROG([MAKEINFO], [makeinfo])
+AC_REQUIRE([AM_PROG_INSTALL_SH])dnl
+AC_REQUIRE([AM_PROG_INSTALL_STRIP])dnl
+AC_REQUIRE([AC_PROG_MKDIR_P])dnl
+# For better backward compatibility.  To be removed once Automake 1.9.x
+# dies out for good.  For more background, see:
+# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
+# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
+AC_SUBST([mkdir_p], ['$(MKDIR_P)'])
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
+AC_REQUIRE([AC_PROG_AWK])dnl
+AC_REQUIRE([AC_PROG_MAKE_SET])dnl
+AC_REQUIRE([AM_SET_LEADING_DOT])dnl
+_AM_IF_OPTION([tar-ustar], [_AM_PROG_TAR([ustar])],
+	      [_AM_IF_OPTION([tar-pax], [_AM_PROG_TAR([pax])],
+			     [_AM_PROG_TAR([v7])])])
+_AM_IF_OPTION([no-dependencies],,
+[AC_PROVIDE_IFELSE([AC_PROG_CC],
+		  [_AM_DEPENDENCIES([CC])],
+		  [m4_define([AC_PROG_CC],
+			     m4_defn([AC_PROG_CC])[_AM_DEPENDENCIES([CC])])])dnl
+AC_PROVIDE_IFELSE([AC_PROG_CXX],
+		  [_AM_DEPENDENCIES([CXX])],
+		  [m4_define([AC_PROG_CXX],
+			     m4_defn([AC_PROG_CXX])[_AM_DEPENDENCIES([CXX])])])dnl
+AC_PROVIDE_IFELSE([AC_PROG_OBJC],
+		  [_AM_DEPENDENCIES([OBJC])],
+		  [m4_define([AC_PROG_OBJC],
+			     m4_defn([AC_PROG_OBJC])[_AM_DEPENDENCIES([OBJC])])])dnl
+AC_PROVIDE_IFELSE([AC_PROG_OBJCXX],
+		  [_AM_DEPENDENCIES([OBJCXX])],
+		  [m4_define([AC_PROG_OBJCXX],
+			     m4_defn([AC_PROG_OBJCXX])[_AM_DEPENDENCIES([OBJCXX])])])dnl
+])
+AC_REQUIRE([AM_SILENT_RULES])dnl
+dnl The testsuite driver may need to know about EXEEXT, so add the
+dnl 'am__EXEEXT' conditional if _AM_COMPILER_EXEEXT was seen.  This
+dnl macro is hooked onto _AC_COMPILER_EXEEXT early, see below.
+AC_CONFIG_COMMANDS_PRE(dnl
+[m4_provide_if([_AM_COMPILER_EXEEXT],
+  [AM_CONDITIONAL([am__EXEEXT], [test -n "$EXEEXT"])])])dnl
+
+# POSIX will say in a future version that running "rm -f" with no argument
+# is OK; and we want to be able to make that assumption in our Makefile
+# recipes.  So use an aggressive probe to check that the usage we want is
+# actually supported "in the wild" to an acceptable degree.
+# See automake bug#10828.
+# To make any issue more visible, cause the running configure to be aborted
+# by default if the 'rm' program in use doesn't match our expectations; the
+# user can still override this though.
+if rm -f && rm -fr && rm -rf; then : OK; else
+  cat >&2 <<'END'
+Oops!
+
+Your 'rm' program seems unable to run without file operands specified
+on the command line, even when the '-f' option is present.  This is contrary
+to the behaviour of most rm programs out there, and not conforming with
+the upcoming POSIX standard: <http://austingroupbugs.net/view.php?id=542>
+
+Please tell bug-automake@gnu.org about your system, including the value
+of your $PATH and any error possibly output before this message.  This
+can help us improve future automake versions.
+
+END
+  if test x"$ACCEPT_INFERIOR_RM_PROGRAM" = x"yes"; then
+    echo 'Configuration will proceed anyway, since you have set the' >&2
+    echo 'ACCEPT_INFERIOR_RM_PROGRAM variable to "yes"' >&2
+    echo >&2
+  else
+    cat >&2 <<'END'
+Aborting the configuration process, to ensure you take notice of the issue.
+
+You can download and install GNU coreutils to get an 'rm' implementation
+that behaves properly: <http://www.gnu.org/software/coreutils/>.
+
+If you want to complete the configuration process using your problematic
+'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
+to "yes", and re-run configure.
+
+END
+    AC_MSG_ERROR([Your 'rm' program is bad, sorry.])
+  fi
+fi
+dnl The trailing newline in this macro's definition is deliberate, for
+dnl backward compatibility and to allow trailing 'dnl'-style comments
+dnl after the AM_INIT_AUTOMAKE invocation. See automake bug#16841.
+])
+
+dnl Hook into '_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
+dnl add the conditional right here, as _AC_COMPILER_EXEEXT may be further
+dnl mangled by Autoconf and run in a shell conditional statement.
+m4_define([_AC_COMPILER_EXEEXT],
+m4_defn([_AC_COMPILER_EXEEXT])[m4_provide([_AM_COMPILER_EXEEXT])])
+
+# When config.status generates a header, we must update the stamp-h file.
+# This file resides in the same directory as the config header
+# that is generated.  The stamp files are numbered to have different names.
+
+# Autoconf calls _AC_AM_CONFIG_HEADER_HOOK (when defined) in the
+# loop where config.status creates the headers, so we can generate
+# our stamp files there.
+AC_DEFUN([_AC_AM_CONFIG_HEADER_HOOK],
+[# Compute $1's index in $config_headers.
+_am_arg=$1
+_am_stamp_count=1
+for _am_header in $config_headers :; do
+  case $_am_header in
+    $_am_arg | $_am_arg:* )
+      break ;;
+    * )
+      _am_stamp_count=`expr $_am_stamp_count + 1` ;;
+  esac
+done
+echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_count])
+
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# AM_PROG_INSTALL_SH
+# ------------------
+# Define $install_sh.
+AC_DEFUN([AM_PROG_INSTALL_SH],
+[AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
+if test x"${install_sh+set}" != xset; then
+  case $am_aux_dir in
+  *\ * | *\	*)
+    install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
+  *)
+    install_sh="\${SHELL} $am_aux_dir/install-sh"
+  esac
+fi
+AC_SUBST([install_sh])])
+
+# Copyright (C) 2003-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# Check whether the underlying file-system supports filenames
+# with a leading dot.  For instance MS-DOS doesn't.
+AC_DEFUN([AM_SET_LEADING_DOT],
+[rm -rf .tst 2>/dev/null
+mkdir .tst 2>/dev/null
+if test -d .tst; then
+  am__leading_dot=.
+else
+  am__leading_dot=_
+fi
+rmdir .tst 2>/dev/null
+AC_SUBST([am__leading_dot])])
+
+# Check to see how 'make' treats includes.	            -*- Autoconf -*-
+
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# AM_MAKE_INCLUDE()
+# -----------------
+# Check to see how make treats includes.
+AC_DEFUN([AM_MAKE_INCLUDE],
+[am_make=${MAKE-make}
+cat > confinc << 'END'
+am__doit:
+	@echo this is the am__doit target
+.PHONY: am__doit
+END
+# If we don't find an include directive, just comment out the code.
+AC_MSG_CHECKING([for style of include used by $am_make])
+am__include="#"
+am__quote=
+_am_result=none
+# First try GNU make style include.
+echo "include confinc" > confmf
+# Ignore all kinds of additional output from 'make'.
+case `$am_make -s -f confmf 2> /dev/null` in #(
+*the\ am__doit\ target*)
+  am__include=include
+  am__quote=
+  _am_result=GNU
+  ;;
+esac
+# Now try BSD make style include.
+if test "$am__include" = "#"; then
+   echo '.include "confinc"' > confmf
+   case `$am_make -s -f confmf 2> /dev/null` in #(
+   *the\ am__doit\ target*)
+     am__include=.include
+     am__quote="\""
+     _am_result=BSD
+     ;;
+   esac
+fi
+AC_SUBST([am__include])
+AC_SUBST([am__quote])
+AC_MSG_RESULT([$_am_result])
+rm -f confinc confmf
+])
+
+# Fake the existence of programs that GNU maintainers use.  -*- Autoconf -*-
+
+# Copyright (C) 1997-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# AM_MISSING_PROG(NAME, PROGRAM)
+# ------------------------------
+AC_DEFUN([AM_MISSING_PROG],
+[AC_REQUIRE([AM_MISSING_HAS_RUN])
+$1=${$1-"${am_missing_run}$2"}
+AC_SUBST($1)])
+
+# AM_MISSING_HAS_RUN
+# ------------------
+# Define MISSING if not defined so far and test if it is modern enough.
+# If it is, set am_missing_run to use it, otherwise, to nothing.
+AC_DEFUN([AM_MISSING_HAS_RUN],
+[AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
+AC_REQUIRE_AUX_FILE([missing])dnl
+if test x"${MISSING+set}" != xset; then
+  case $am_aux_dir in
+  *\ * | *\	*)
+    MISSING="\${SHELL} \"$am_aux_dir/missing\"" ;;
+  *)
+    MISSING="\${SHELL} $am_aux_dir/missing" ;;
+  esac
+fi
+# Use eval to expand $SHELL
+if eval "$MISSING --is-lightweight"; then
+  am_missing_run="$MISSING "
+else
+  am_missing_run=
+  AC_MSG_WARN(['missing' script is too old or missing])
+fi
+])
+
+# Helper functions for option handling.                     -*- Autoconf -*-
+
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# _AM_MANGLE_OPTION(NAME)
+# -----------------------
+AC_DEFUN([_AM_MANGLE_OPTION],
+[[_AM_OPTION_]m4_bpatsubst($1, [[^a-zA-Z0-9_]], [_])])
+
+# _AM_SET_OPTION(NAME)
+# --------------------
+# Set option NAME.  Presently that only means defining a flag for this option.
+AC_DEFUN([_AM_SET_OPTION],
+[m4_define(_AM_MANGLE_OPTION([$1]), [1])])
+
+# _AM_SET_OPTIONS(OPTIONS)
+# ------------------------
+# OPTIONS is a space-separated list of Automake options.
+AC_DEFUN([_AM_SET_OPTIONS],
+[m4_foreach_w([_AM_Option], [$1], [_AM_SET_OPTION(_AM_Option)])])
+
+# _AM_IF_OPTION(OPTION, IF-SET, [IF-NOT-SET])
+# -------------------------------------------
+# Execute IF-SET if OPTION is set, IF-NOT-SET otherwise.
+AC_DEFUN([_AM_IF_OPTION],
+[m4_ifset(_AM_MANGLE_OPTION([$1]), [$2], [$3])])
+
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# _AM_PROG_CC_C_O
+# ---------------
+# Like AC_PROG_CC_C_O, but changed for automake.  We rewrite AC_PROG_CC
+# to automatically call this.
+AC_DEFUN([_AM_PROG_CC_C_O],
+[AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
+AC_REQUIRE_AUX_FILE([compile])dnl
+AC_LANG_PUSH([C])dnl
+AC_CACHE_CHECK(
+  [whether $CC understands -c and -o together],
+  [am_cv_prog_cc_c_o],
+  [AC_LANG_CONFTEST([AC_LANG_PROGRAM([])])
+  # Make sure it works both with $CC and with simple cc.
+  # Following AC_PROG_CC_C_O, we do the test twice because some
+  # compilers refuse to overwrite an existing .o file with -o,
+  # though they will create one.
+  am_cv_prog_cc_c_o=yes
+  for am_i in 1 2; do
+    if AM_RUN_LOG([$CC -c conftest.$ac_ext -o conftest2.$ac_objext]) \
+         && test -f conftest2.$ac_objext; then
+      : OK
+    else
+      am_cv_prog_cc_c_o=no
+      break
+    fi
+  done
+  rm -f core conftest*
+  unset am_i])
+if test "$am_cv_prog_cc_c_o" != yes; then
+   # Losing compiler, so override with the script.
+   # FIXME: It is wrong to rewrite CC.
+   # But if we don't then we get into trouble of one sort or another.
+   # A longer-term fix would be to have automake use am__CC in this case,
+   # and then we could set am__CC="\$(top_srcdir)/compile \$(CC)"
+   CC="$am_aux_dir/compile $CC"
+fi
+AC_LANG_POP([C])])
+
+# For backward compatibility.
+AC_DEFUN_ONCE([AM_PROG_CC_C_O], [AC_REQUIRE([AC_PROG_CC])])
+
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# AM_RUN_LOG(COMMAND)
+# -------------------
+# Run COMMAND, save the exit status in ac_status, and log it.
+# (This has been adapted from Autoconf's _AC_RUN_LOG macro.)
+AC_DEFUN([AM_RUN_LOG],
+[{ echo "$as_me:$LINENO: $1" >&AS_MESSAGE_LOG_FD
+   ($1) >&AS_MESSAGE_LOG_FD 2>&AS_MESSAGE_LOG_FD
+   ac_status=$?
+   echo "$as_me:$LINENO: \$? = $ac_status" >&AS_MESSAGE_LOG_FD
+   (exit $ac_status); }])
+
+# Check to make sure that the build environment is sane.    -*- Autoconf -*-
+
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# AM_SANITY_CHECK
+# ---------------
+AC_DEFUN([AM_SANITY_CHECK],
+[AC_MSG_CHECKING([whether build environment is sane])
+# Reject unsafe characters in $srcdir or the absolute working directory
+# name.  Accept space and tab only in the latter.
+am_lf='
+'
+case `pwd` in
+  *[[\\\"\#\$\&\'\`$am_lf]]*)
+    AC_MSG_ERROR([unsafe absolute working directory name]);;
+esac
+case $srcdir in
+  *[[\\\"\#\$\&\'\`$am_lf\ \	]]*)
+    AC_MSG_ERROR([unsafe srcdir value: '$srcdir']);;
+esac
+
+# Do 'set' in a subshell so we don't clobber the current shell's
+# arguments.  Must try -L first in case configure is actually a
+# symlink; some systems play weird games with the mod time of symlinks
+# (eg FreeBSD returns the mod time of the symlink's containing
+# directory).
+if (
+   am_has_slept=no
+   for am_try in 1 2; do
+     echo "timestamp, slept: $am_has_slept" > conftest.file
+     set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
+     if test "$[*]" = "X"; then
+	# -L didn't work.
+	set X `ls -t "$srcdir/configure" conftest.file`
+     fi
+     if test "$[*]" != "X $srcdir/configure conftest.file" \
+	&& test "$[*]" != "X conftest.file $srcdir/configure"; then
+
+	# If neither matched, then we have a broken ls.  This can happen
+	# if, for instance, CONFIG_SHELL is bash and it inherits a
+	# broken ls alias from the environment.  This has actually
+	# happened.  Such a system could not be considered "sane".
+	AC_MSG_ERROR([ls -t appears to fail.  Make sure there is not a broken
+  alias in your environment])
+     fi
+     if test "$[2]" = conftest.file || test $am_try -eq 2; then
+       break
+     fi
+     # Just in case.
+     sleep 1
+     am_has_slept=yes
+   done
+   test "$[2]" = conftest.file
+   )
+then
+   # Ok.
+   :
+else
+   AC_MSG_ERROR([newly created file is older than distributed files!
+Check your system clock])
+fi
+AC_MSG_RESULT([yes])
+# If we didn't sleep, we still need to ensure time stamps of config.status and
+# generated files are strictly newer.
+am_sleep_pid=
+if grep 'slept: no' conftest.file >/dev/null 2>&1; then
+  ( sleep 1 ) &
+  am_sleep_pid=$!
+fi
+AC_CONFIG_COMMANDS_PRE(
+  [AC_MSG_CHECKING([that generated files are newer than configure])
+   if test -n "$am_sleep_pid"; then
+     # Hide warnings about reused PIDs.
+     wait $am_sleep_pid 2>/dev/null
+   fi
+   AC_MSG_RESULT([done])])
+rm -f conftest.file
+])
+
+# Copyright (C) 2009-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# AM_SILENT_RULES([DEFAULT])
+# --------------------------
+# Enable less verbose build rules; with the default set to DEFAULT
+# ("yes" being less verbose, "no" or empty being verbose).
+AC_DEFUN([AM_SILENT_RULES],
+[AC_ARG_ENABLE([silent-rules], [dnl
+AS_HELP_STRING(
+  [--enable-silent-rules],
+  [less verbose build output (undo: "make V=1")])
+AS_HELP_STRING(
+  [--disable-silent-rules],
+  [verbose build output (undo: "make V=0")])dnl
+])
+case $enable_silent_rules in @%:@ (((
+  yes) AM_DEFAULT_VERBOSITY=0;;
+   no) AM_DEFAULT_VERBOSITY=1;;
+    *) AM_DEFAULT_VERBOSITY=m4_if([$1], [yes], [0], [1]);;
+esac
+dnl
+dnl A few 'make' implementations (e.g., NonStop OS and NextStep)
+dnl do not support nested variable expansions.
+dnl See automake bug#9928 and bug#10237.
+am_make=${MAKE-make}
+AC_CACHE_CHECK([whether $am_make supports nested variables],
+   [am_cv_make_support_nested_variables],
+   [if AS_ECHO([['TRUE=$(BAR$(V))
+BAR0=false
+BAR1=true
+V=1
+am__doit:
+	@$(TRUE)
+.PHONY: am__doit']]) | $am_make -f - >/dev/null 2>&1; then
+  am_cv_make_support_nested_variables=yes
+else
+  am_cv_make_support_nested_variables=no
+fi])
+if test $am_cv_make_support_nested_variables = yes; then
+  dnl Using '$V' instead of '$(V)' breaks IRIX make.
+  AM_V='$(V)'
+  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
+else
+  AM_V=$AM_DEFAULT_VERBOSITY
+  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
+fi
+AC_SUBST([AM_V])dnl
+AM_SUBST_NOTMAKE([AM_V])dnl
+AC_SUBST([AM_DEFAULT_V])dnl
+AM_SUBST_NOTMAKE([AM_DEFAULT_V])dnl
+AC_SUBST([AM_DEFAULT_VERBOSITY])dnl
+AM_BACKSLASH='\'
+AC_SUBST([AM_BACKSLASH])dnl
+_AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
+])
+
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# AM_PROG_INSTALL_STRIP
+# ---------------------
+# One issue with vendor 'install' (even GNU) is that you can't
+# specify the program used to strip binaries.  This is especially
+# annoying in cross-compiling environments, where the build's strip
+# is unlikely to handle the host's binaries.
+# Fortunately install-sh will honor a STRIPPROG variable, so we
+# always use install-sh in "make install-strip", and initialize
+# STRIPPROG with the value of the STRIP variable (set by the user).
+AC_DEFUN([AM_PROG_INSTALL_STRIP],
+[AC_REQUIRE([AM_PROG_INSTALL_SH])dnl
+# Installed binaries are usually stripped using 'strip' when the user
+# run "make install-strip".  However 'strip' might not be the right
+# tool to use in cross-compilation environments, therefore Automake
+# will honor the 'STRIP' environment variable to overrule this program.
+dnl Don't test for $cross_compiling = yes, because it might be 'maybe'.
+if test "$cross_compiling" != no; then
+  AC_CHECK_TOOL([STRIP], [strip], :)
+fi
+INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
+AC_SUBST([INSTALL_STRIP_PROGRAM])])
+
+# Copyright (C) 2006-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# _AM_SUBST_NOTMAKE(VARIABLE)
+# ---------------------------
+# Prevent Automake from outputting VARIABLE = @VARIABLE@ in Makefile.in.
+# This macro is traced by Automake.
+AC_DEFUN([_AM_SUBST_NOTMAKE])
+
+# AM_SUBST_NOTMAKE(VARIABLE)
+# --------------------------
+# Public sister of _AM_SUBST_NOTMAKE.
+AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
+
+# Check how to create a tarball.                            -*- Autoconf -*-
+
+# Copyright (C) 2004-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# _AM_PROG_TAR(FORMAT)
+# --------------------
+# Check how to create a tarball in format FORMAT.
+# FORMAT should be one of 'v7', 'ustar', or 'pax'.
+#
+# Substitute a variable $(am__tar) that is a command
+# writing to stdout a FORMAT-tarball containing the directory
+# $tardir.
+#     tardir=directory && $(am__tar) > result.tar
+#
+# Substitute a variable $(am__untar) that extract such
+# a tarball read from stdin.
+#     $(am__untar) < result.tar
+#
+AC_DEFUN([_AM_PROG_TAR],
+[# Always define AMTAR for backward compatibility.  Yes, it's still used
+# in the wild :-(  We should find a proper way to deprecate it ...
+AC_SUBST([AMTAR], ['$${TAR-tar}'])
+
+# We'll loop over all known methods to create a tar archive until one works.
+_am_tools='gnutar m4_if([$1], [ustar], [plaintar]) pax cpio none'
+
+m4_if([$1], [v7],
+  [am__tar='$${TAR-tar} chof - "$$tardir"' am__untar='$${TAR-tar} xf -'],
+
+  [m4_case([$1],
+    [ustar],
+     [# The POSIX 1988 'ustar' format is defined with fixed-size fields.
+      # There is notably a 21 bits limit for the UID and the GID.  In fact,
+      # the 'pax' utility can hang on bigger UID/GID (see automake bug#8343
+      # and bug#13588).
+      am_max_uid=2097151 # 2^21 - 1
+      am_max_gid=$am_max_uid
+      # The $UID and $GID variables are not portable, so we need to resort
+      # to the POSIX-mandated id(1) utility.  Errors in the 'id' calls
+      # below are definitely unexpected, so allow the users to see them
+      # (that is, avoid stderr redirection).
+      am_uid=`id -u || echo unknown`
+      am_gid=`id -g || echo unknown`
+      AC_MSG_CHECKING([whether UID '$am_uid' is supported by ustar format])
+      if test $am_uid -le $am_max_uid; then
+         AC_MSG_RESULT([yes])
+      else
+         AC_MSG_RESULT([no])
+         _am_tools=none
+      fi
+      AC_MSG_CHECKING([whether GID '$am_gid' is supported by ustar format])
+      if test $am_gid -le $am_max_gid; then
+         AC_MSG_RESULT([yes])
+      else
+        AC_MSG_RESULT([no])
+        _am_tools=none
+      fi],
+
+  [pax],
+    [],
+
+  [m4_fatal([Unknown tar format])])
+
+  AC_MSG_CHECKING([how to create a $1 tar archive])
+
+  # Go ahead even if we have the value already cached.  We do so because we
+  # need to set the values for the 'am__tar' and 'am__untar' variables.
+  _am_tools=${am_cv_prog_tar_$1-$_am_tools}
+
+  for _am_tool in $_am_tools; do
+    case $_am_tool in
+    gnutar)
+      for _am_tar in tar gnutar gtar; do
+        AM_RUN_LOG([$_am_tar --version]) && break
+      done
+      am__tar="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$$tardir"'
+      am__tar_="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$tardir"'
+      am__untar="$_am_tar -xf -"
+      ;;
+    plaintar)
+      # Must skip GNU tar: if it does not support --format= it doesn't create
+      # ustar tarball either.
+      (tar --version) >/dev/null 2>&1 && continue
+      am__tar='tar chf - "$$tardir"'
+      am__tar_='tar chf - "$tardir"'
+      am__untar='tar xf -'
+      ;;
+    pax)
+      am__tar='pax -L -x $1 -w "$$tardir"'
+      am__tar_='pax -L -x $1 -w "$tardir"'
+      am__untar='pax -r'
+      ;;
+    cpio)
+      am__tar='find "$$tardir" -print | cpio -o -H $1 -L'
+      am__tar_='find "$tardir" -print | cpio -o -H $1 -L'
+      am__untar='cpio -i -H $1 -d'
+      ;;
+    none)
+      am__tar=false
+      am__tar_=false
+      am__untar=false
+      ;;
+    esac
+
+    # If the value was cached, stop now.  We just wanted to have am__tar
+    # and am__untar set.
+    test -n "${am_cv_prog_tar_$1}" && break
+
+    # tar/untar a dummy directory, and stop if the command works.
+    rm -rf conftest.dir
+    mkdir conftest.dir
+    echo GrepMe > conftest.dir/file
+    AM_RUN_LOG([tardir=conftest.dir && eval $am__tar_ >conftest.tar])
+    rm -rf conftest.dir
+    if test -s conftest.tar; then
+      AM_RUN_LOG([$am__untar <conftest.tar])
+      AM_RUN_LOG([cat conftest.dir/file])
+      grep GrepMe conftest.dir/file >/dev/null 2>&1 && break
+    fi
+  done
+  rm -rf conftest.dir
+
+  AC_CACHE_VAL([am_cv_prog_tar_$1], [am_cv_prog_tar_$1=$_am_tool])
+  AC_MSG_RESULT([$am_cv_prog_tar_$1])])
+
+AC_SUBST([am__tar])
+AC_SUBST([am__untar])
+]) # _AM_PROG_TAR
+
+m4_include([m4/ax_gcc_var_attribute.m4])

From 59eff71fc597b57a4840442ef99e6dfa5f7fef94 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 4 Mar 2015 12:03:07 +0000
Subject: [PATCH 014/429] Extra files

---
 COPYING | 674 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 INSTALL | 370 +++++++++++++++++++++++++++++++
 2 files changed, 1044 insertions(+)
 create mode 100644 COPYING
 create mode 100644 INSTALL

diff --git a/COPYING b/COPYING
new file mode 100644
index 00000000..94a9ed02
--- /dev/null
+++ b/COPYING
@@ -0,0 +1,674 @@
+                    GNU GENERAL PUBLIC LICENSE
+                       Version 3, 29 June 2007
+
+ Copyright (C) 2007 Free Software Foundation, Inc. <http://fsf.org/>
+ Everyone is permitted to copy and distribute verbatim copies
+ of this license document, but changing it is not allowed.
+
+                            Preamble
+
+  The GNU General Public License is a free, copyleft license for
+software and other kinds of works.
+
+  The licenses for most software and other practical works are designed
+to take away your freedom to share and change the works.  By contrast,
+the GNU General Public License is intended to guarantee your freedom to
+share and change all versions of a program--to make sure it remains free
+software for all its users.  We, the Free Software Foundation, use the
+GNU General Public License for most of our software; it applies also to
+any other work released this way by its authors.  You can apply it to
+your programs, too.
+
+  When we speak of free software, we are referring to freedom, not
+price.  Our General Public Licenses are designed to make sure that you
+have the freedom to distribute copies of free software (and charge for
+them if you wish), that you receive source code or can get it if you
+want it, that you can change the software or use pieces of it in new
+free programs, and that you know you can do these things.
+
+  To protect your rights, we need to prevent others from denying you
+these rights or asking you to surrender the rights.  Therefore, you have
+certain responsibilities if you distribute copies of the software, or if
+you modify it: responsibilities to respect the freedom of others.
+
+  For example, if you distribute copies of such a program, whether
+gratis or for a fee, you must pass on to the recipients the same
+freedoms that you received.  You must make sure that they, too, receive
+or can get the source code.  And you must show them these terms so they
+know their rights.
+
+  Developers that use the GNU GPL protect your rights with two steps:
+(1) assert copyright on the software, and (2) offer you this License
+giving you legal permission to copy, distribute and/or modify it.
+
+  For the developers' and authors' protection, the GPL clearly explains
+that there is no warranty for this free software.  For both users' and
+authors' sake, the GPL requires that modified versions be marked as
+changed, so that their problems will not be attributed erroneously to
+authors of previous versions.
+
+  Some devices are designed to deny users access to install or run
+modified versions of the software inside them, although the manufacturer
+can do so.  This is fundamentally incompatible with the aim of
+protecting users' freedom to change the software.  The systematic
+pattern of such abuse occurs in the area of products for individuals to
+use, which is precisely where it is most unacceptable.  Therefore, we
+have designed this version of the GPL to prohibit the practice for those
+products.  If such problems arise substantially in other domains, we
+stand ready to extend this provision to those domains in future versions
+of the GPL, as needed to protect the freedom of users.
+
+  Finally, every program is threatened constantly by software patents.
+States should not allow patents to restrict development and use of
+software on general-purpose computers, but in those that do, we wish to
+avoid the special danger that patents applied to a free program could
+make it effectively proprietary.  To prevent this, the GPL assures that
+patents cannot be used to render the program non-free.
+
+  The precise terms and conditions for copying, distribution and
+modification follow.
+
+                       TERMS AND CONDITIONS
+
+  0. Definitions.
+
+  "This License" refers to version 3 of the GNU General Public License.
+
+  "Copyright" also means copyright-like laws that apply to other kinds of
+works, such as semiconductor masks.
+
+  "The Program" refers to any copyrightable work licensed under this
+License.  Each licensee is addressed as "you".  "Licensees" and
+"recipients" may be individuals or organizations.
+
+  To "modify" a work means to copy from or adapt all or part of the work
+in a fashion requiring copyright permission, other than the making of an
+exact copy.  The resulting work is called a "modified version" of the
+earlier work or a work "based on" the earlier work.
+
+  A "covered work" means either the unmodified Program or a work based
+on the Program.
+
+  To "propagate" a work means to do anything with it that, without
+permission, would make you directly or secondarily liable for
+infringement under applicable copyright law, except executing it on a
+computer or modifying a private copy.  Propagation includes copying,
+distribution (with or without modification), making available to the
+public, and in some countries other activities as well.
+
+  To "convey" a work means any kind of propagation that enables other
+parties to make or receive copies.  Mere interaction with a user through
+a computer network, with no transfer of a copy, is not conveying.
+
+  An interactive user interface displays "Appropriate Legal Notices"
+to the extent that it includes a convenient and prominently visible
+feature that (1) displays an appropriate copyright notice, and (2)
+tells the user that there is no warranty for the work (except to the
+extent that warranties are provided), that licensees may convey the
+work under this License, and how to view a copy of this License.  If
+the interface presents a list of user commands or options, such as a
+menu, a prominent item in the list meets this criterion.
+
+  1. Source Code.
+
+  The "source code" for a work means the preferred form of the work
+for making modifications to it.  "Object code" means any non-source
+form of a work.
+
+  A "Standard Interface" means an interface that either is an official
+standard defined by a recognized standards body, or, in the case of
+interfaces specified for a particular programming language, one that
+is widely used among developers working in that language.
+
+  The "System Libraries" of an executable work include anything, other
+than the work as a whole, that (a) is included in the normal form of
+packaging a Major Component, but which is not part of that Major
+Component, and (b) serves only to enable use of the work with that
+Major Component, or to implement a Standard Interface for which an
+implementation is available to the public in source code form.  A
+"Major Component", in this context, means a major essential component
+(kernel, window system, and so on) of the specific operating system
+(if any) on which the executable work runs, or a compiler used to
+produce the work, or an object code interpreter used to run it.
+
+  The "Corresponding Source" for a work in object code form means all
+the source code needed to generate, install, and (for an executable
+work) run the object code and to modify the work, including scripts to
+control those activities.  However, it does not include the work's
+System Libraries, or general-purpose tools or generally available free
+programs which are used unmodified in performing those activities but
+which are not part of the work.  For example, Corresponding Source
+includes interface definition files associated with source files for
+the work, and the source code for shared libraries and dynamically
+linked subprograms that the work is specifically designed to require,
+such as by intimate data communication or control flow between those
+subprograms and other parts of the work.
+
+  The Corresponding Source need not include anything that users
+can regenerate automatically from other parts of the Corresponding
+Source.
+
+  The Corresponding Source for a work in source code form is that
+same work.
+
+  2. Basic Permissions.
+
+  All rights granted under this License are granted for the term of
+copyright on the Program, and are irrevocable provided the stated
+conditions are met.  This License explicitly affirms your unlimited
+permission to run the unmodified Program.  The output from running a
+covered work is covered by this License only if the output, given its
+content, constitutes a covered work.  This License acknowledges your
+rights of fair use or other equivalent, as provided by copyright law.
+
+  You may make, run and propagate covered works that you do not
+convey, without conditions so long as your license otherwise remains
+in force.  You may convey covered works to others for the sole purpose
+of having them make modifications exclusively for you, or provide you
+with facilities for running those works, provided that you comply with
+the terms of this License in conveying all material for which you do
+not control copyright.  Those thus making or running the covered works
+for you must do so exclusively on your behalf, under your direction
+and control, on terms that prohibit them from making any copies of
+your copyrighted material outside their relationship with you.
+
+  Conveying under any other circumstances is permitted solely under
+the conditions stated below.  Sublicensing is not allowed; section 10
+makes it unnecessary.
+
+  3. Protecting Users' Legal Rights From Anti-Circumvention Law.
+
+  No covered work shall be deemed part of an effective technological
+measure under any applicable law fulfilling obligations under article
+11 of the WIPO copyright treaty adopted on 20 December 1996, or
+similar laws prohibiting or restricting circumvention of such
+measures.
+
+  When you convey a covered work, you waive any legal power to forbid
+circumvention of technological measures to the extent such circumvention
+is effected by exercising rights under this License with respect to
+the covered work, and you disclaim any intention to limit operation or
+modification of the work as a means of enforcing, against the work's
+users, your or third parties' legal rights to forbid circumvention of
+technological measures.
+
+  4. Conveying Verbatim Copies.
+
+  You may convey verbatim copies of the Program's source code as you
+receive it, in any medium, provided that you conspicuously and
+appropriately publish on each copy an appropriate copyright notice;
+keep intact all notices stating that this License and any
+non-permissive terms added in accord with section 7 apply to the code;
+keep intact all notices of the absence of any warranty; and give all
+recipients a copy of this License along with the Program.
+
+  You may charge any price or no price for each copy that you convey,
+and you may offer support or warranty protection for a fee.
+
+  5. Conveying Modified Source Versions.
+
+  You may convey a work based on the Program, or the modifications to
+produce it from the Program, in the form of source code under the
+terms of section 4, provided that you also meet all of these conditions:
+
+    a) The work must carry prominent notices stating that you modified
+    it, and giving a relevant date.
+
+    b) The work must carry prominent notices stating that it is
+    released under this License and any conditions added under section
+    7.  This requirement modifies the requirement in section 4 to
+    "keep intact all notices".
+
+    c) You must license the entire work, as a whole, under this
+    License to anyone who comes into possession of a copy.  This
+    License will therefore apply, along with any applicable section 7
+    additional terms, to the whole of the work, and all its parts,
+    regardless of how they are packaged.  This License gives no
+    permission to license the work in any other way, but it does not
+    invalidate such permission if you have separately received it.
+
+    d) If the work has interactive user interfaces, each must display
+    Appropriate Legal Notices; however, if the Program has interactive
+    interfaces that do not display Appropriate Legal Notices, your
+    work need not make them do so.
+
+  A compilation of a covered work with other separate and independent
+works, which are not by their nature extensions of the covered work,
+and which are not combined with it such as to form a larger program,
+in or on a volume of a storage or distribution medium, is called an
+"aggregate" if the compilation and its resulting copyright are not
+used to limit the access or legal rights of the compilation's users
+beyond what the individual works permit.  Inclusion of a covered work
+in an aggregate does not cause this License to apply to the other
+parts of the aggregate.
+
+  6. Conveying Non-Source Forms.
+
+  You may convey a covered work in object code form under the terms
+of sections 4 and 5, provided that you also convey the
+machine-readable Corresponding Source under the terms of this License,
+in one of these ways:
+
+    a) Convey the object code in, or embodied in, a physical product
+    (including a physical distribution medium), accompanied by the
+    Corresponding Source fixed on a durable physical medium
+    customarily used for software interchange.
+
+    b) Convey the object code in, or embodied in, a physical product
+    (including a physical distribution medium), accompanied by a
+    written offer, valid for at least three years and valid for as
+    long as you offer spare parts or customer support for that product
+    model, to give anyone who possesses the object code either (1) a
+    copy of the Corresponding Source for all the software in the
+    product that is covered by this License, on a durable physical
+    medium customarily used for software interchange, for a price no
+    more than your reasonable cost of physically performing this
+    conveying of source, or (2) access to copy the
+    Corresponding Source from a network server at no charge.
+
+    c) Convey individual copies of the object code with a copy of the
+    written offer to provide the Corresponding Source.  This
+    alternative is allowed only occasionally and noncommercially, and
+    only if you received the object code with such an offer, in accord
+    with subsection 6b.
+
+    d) Convey the object code by offering access from a designated
+    place (gratis or for a charge), and offer equivalent access to the
+    Corresponding Source in the same way through the same place at no
+    further charge.  You need not require recipients to copy the
+    Corresponding Source along with the object code.  If the place to
+    copy the object code is a network server, the Corresponding Source
+    may be on a different server (operated by you or a third party)
+    that supports equivalent copying facilities, provided you maintain
+    clear directions next to the object code saying where to find the
+    Corresponding Source.  Regardless of what server hosts the
+    Corresponding Source, you remain obligated to ensure that it is
+    available for as long as needed to satisfy these requirements.
+
+    e) Convey the object code using peer-to-peer transmission, provided
+    you inform other peers where the object code and Corresponding
+    Source of the work are being offered to the general public at no
+    charge under subsection 6d.
+
+  A separable portion of the object code, whose source code is excluded
+from the Corresponding Source as a System Library, need not be
+included in conveying the object code work.
+
+  A "User Product" is either (1) a "consumer product", which means any
+tangible personal property which is normally used for personal, family,
+or household purposes, or (2) anything designed or sold for incorporation
+into a dwelling.  In determining whether a product is a consumer product,
+doubtful cases shall be resolved in favor of coverage.  For a particular
+product received by a particular user, "normally used" refers to a
+typical or common use of that class of product, regardless of the status
+of the particular user or of the way in which the particular user
+actually uses, or expects or is expected to use, the product.  A product
+is a consumer product regardless of whether the product has substantial
+commercial, industrial or non-consumer uses, unless such uses represent
+the only significant mode of use of the product.
+
+  "Installation Information" for a User Product means any methods,
+procedures, authorization keys, or other information required to install
+and execute modified versions of a covered work in that User Product from
+a modified version of its Corresponding Source.  The information must
+suffice to ensure that the continued functioning of the modified object
+code is in no case prevented or interfered with solely because
+modification has been made.
+
+  If you convey an object code work under this section in, or with, or
+specifically for use in, a User Product, and the conveying occurs as
+part of a transaction in which the right of possession and use of the
+User Product is transferred to the recipient in perpetuity or for a
+fixed term (regardless of how the transaction is characterized), the
+Corresponding Source conveyed under this section must be accompanied
+by the Installation Information.  But this requirement does not apply
+if neither you nor any third party retains the ability to install
+modified object code on the User Product (for example, the work has
+been installed in ROM).
+
+  The requirement to provide Installation Information does not include a
+requirement to continue to provide support service, warranty, or updates
+for a work that has been modified or installed by the recipient, or for
+the User Product in which it has been modified or installed.  Access to a
+network may be denied when the modification itself materially and
+adversely affects the operation of the network or violates the rules and
+protocols for communication across the network.
+
+  Corresponding Source conveyed, and Installation Information provided,
+in accord with this section must be in a format that is publicly
+documented (and with an implementation available to the public in
+source code form), and must require no special password or key for
+unpacking, reading or copying.
+
+  7. Additional Terms.
+
+  "Additional permissions" are terms that supplement the terms of this
+License by making exceptions from one or more of its conditions.
+Additional permissions that are applicable to the entire Program shall
+be treated as though they were included in this License, to the extent
+that they are valid under applicable law.  If additional permissions
+apply only to part of the Program, that part may be used separately
+under those permissions, but the entire Program remains governed by
+this License without regard to the additional permissions.
+
+  When you convey a copy of a covered work, you may at your option
+remove any additional permissions from that copy, or from any part of
+it.  (Additional permissions may be written to require their own
+removal in certain cases when you modify the work.)  You may place
+additional permissions on material, added by you to a covered work,
+for which you have or can give appropriate copyright permission.
+
+  Notwithstanding any other provision of this License, for material you
+add to a covered work, you may (if authorized by the copyright holders of
+that material) supplement the terms of this License with terms:
+
+    a) Disclaiming warranty or limiting liability differently from the
+    terms of sections 15 and 16 of this License; or
+
+    b) Requiring preservation of specified reasonable legal notices or
+    author attributions in that material or in the Appropriate Legal
+    Notices displayed by works containing it; or
+
+    c) Prohibiting misrepresentation of the origin of that material, or
+    requiring that modified versions of such material be marked in
+    reasonable ways as different from the original version; or
+
+    d) Limiting the use for publicity purposes of names of licensors or
+    authors of the material; or
+
+    e) Declining to grant rights under trademark law for use of some
+    trade names, trademarks, or service marks; or
+
+    f) Requiring indemnification of licensors and authors of that
+    material by anyone who conveys the material (or modified versions of
+    it) with contractual assumptions of liability to the recipient, for
+    any liability that these contractual assumptions directly impose on
+    those licensors and authors.
+
+  All other non-permissive additional terms are considered "further
+restrictions" within the meaning of section 10.  If the Program as you
+received it, or any part of it, contains a notice stating that it is
+governed by this License along with a term that is a further
+restriction, you may remove that term.  If a license document contains
+a further restriction but permits relicensing or conveying under this
+License, you may add to a covered work material governed by the terms
+of that license document, provided that the further restriction does
+not survive such relicensing or conveying.
+
+  If you add terms to a covered work in accord with this section, you
+must place, in the relevant source files, a statement of the
+additional terms that apply to those files, or a notice indicating
+where to find the applicable terms.
+
+  Additional terms, permissive or non-permissive, may be stated in the
+form of a separately written license, or stated as exceptions;
+the above requirements apply either way.
+
+  8. Termination.
+
+  You may not propagate or modify a covered work except as expressly
+provided under this License.  Any attempt otherwise to propagate or
+modify it is void, and will automatically terminate your rights under
+this License (including any patent licenses granted under the third
+paragraph of section 11).
+
+  However, if you cease all violation of this License, then your
+license from a particular copyright holder is reinstated (a)
+provisionally, unless and until the copyright holder explicitly and
+finally terminates your license, and (b) permanently, if the copyright
+holder fails to notify you of the violation by some reasonable means
+prior to 60 days after the cessation.
+
+  Moreover, your license from a particular copyright holder is
+reinstated permanently if the copyright holder notifies you of the
+violation by some reasonable means, this is the first time you have
+received notice of violation of this License (for any work) from that
+copyright holder, and you cure the violation prior to 30 days after
+your receipt of the notice.
+
+  Termination of your rights under this section does not terminate the
+licenses of parties who have received copies or rights from you under
+this License.  If your rights have been terminated and not permanently
+reinstated, you do not qualify to receive new licenses for the same
+material under section 10.
+
+  9. Acceptance Not Required for Having Copies.
+
+  You are not required to accept this License in order to receive or
+run a copy of the Program.  Ancillary propagation of a covered work
+occurring solely as a consequence of using peer-to-peer transmission
+to receive a copy likewise does not require acceptance.  However,
+nothing other than this License grants you permission to propagate or
+modify any covered work.  These actions infringe copyright if you do
+not accept this License.  Therefore, by modifying or propagating a
+covered work, you indicate your acceptance of this License to do so.
+
+  10. Automatic Licensing of Downstream Recipients.
+
+  Each time you convey a covered work, the recipient automatically
+receives a license from the original licensors, to run, modify and
+propagate that work, subject to this License.  You are not responsible
+for enforcing compliance by third parties with this License.
+
+  An "entity transaction" is a transaction transferring control of an
+organization, or substantially all assets of one, or subdividing an
+organization, or merging organizations.  If propagation of a covered
+work results from an entity transaction, each party to that
+transaction who receives a copy of the work also receives whatever
+licenses to the work the party's predecessor in interest had or could
+give under the previous paragraph, plus a right to possession of the
+Corresponding Source of the work from the predecessor in interest, if
+the predecessor has it or can get it with reasonable efforts.
+
+  You may not impose any further restrictions on the exercise of the
+rights granted or affirmed under this License.  For example, you may
+not impose a license fee, royalty, or other charge for exercise of
+rights granted under this License, and you may not initiate litigation
+(including a cross-claim or counterclaim in a lawsuit) alleging that
+any patent claim is infringed by making, using, selling, offering for
+sale, or importing the Program or any portion of it.
+
+  11. Patents.
+
+  A "contributor" is a copyright holder who authorizes use under this
+License of the Program or a work on which the Program is based.  The
+work thus licensed is called the contributor's "contributor version".
+
+  A contributor's "essential patent claims" are all patent claims
+owned or controlled by the contributor, whether already acquired or
+hereafter acquired, that would be infringed by some manner, permitted
+by this License, of making, using, or selling its contributor version,
+but do not include claims that would be infringed only as a
+consequence of further modification of the contributor version.  For
+purposes of this definition, "control" includes the right to grant
+patent sublicenses in a manner consistent with the requirements of
+this License.
+
+  Each contributor grants you a non-exclusive, worldwide, royalty-free
+patent license under the contributor's essential patent claims, to
+make, use, sell, offer for sale, import and otherwise run, modify and
+propagate the contents of its contributor version.
+
+  In the following three paragraphs, a "patent license" is any express
+agreement or commitment, however denominated, not to enforce a patent
+(such as an express permission to practice a patent or covenant not to
+sue for patent infringement).  To "grant" such a patent license to a
+party means to make such an agreement or commitment not to enforce a
+patent against the party.
+
+  If you convey a covered work, knowingly relying on a patent license,
+and the Corresponding Source of the work is not available for anyone
+to copy, free of charge and under the terms of this License, through a
+publicly available network server or other readily accessible means,
+then you must either (1) cause the Corresponding Source to be so
+available, or (2) arrange to deprive yourself of the benefit of the
+patent license for this particular work, or (3) arrange, in a manner
+consistent with the requirements of this License, to extend the patent
+license to downstream recipients.  "Knowingly relying" means you have
+actual knowledge that, but for the patent license, your conveying the
+covered work in a country, or your recipient's use of the covered work
+in a country, would infringe one or more identifiable patents in that
+country that you have reason to believe are valid.
+
+  If, pursuant to or in connection with a single transaction or
+arrangement, you convey, or propagate by procuring conveyance of, a
+covered work, and grant a patent license to some of the parties
+receiving the covered work authorizing them to use, propagate, modify
+or convey a specific copy of the covered work, then the patent license
+you grant is automatically extended to all recipients of the covered
+work and works based on it.
+
+  A patent license is "discriminatory" if it does not include within
+the scope of its coverage, prohibits the exercise of, or is
+conditioned on the non-exercise of one or more of the rights that are
+specifically granted under this License.  You may not convey a covered
+work if you are a party to an arrangement with a third party that is
+in the business of distributing software, under which you make payment
+to the third party based on the extent of your activity of conveying
+the work, and under which the third party grants, to any of the
+parties who would receive the covered work from you, a discriminatory
+patent license (a) in connection with copies of the covered work
+conveyed by you (or copies made from those copies), or (b) primarily
+for and in connection with specific products or compilations that
+contain the covered work, unless you entered into that arrangement,
+or that patent license was granted, prior to 28 March 2007.
+
+  Nothing in this License shall be construed as excluding or limiting
+any implied license or other defenses to infringement that may
+otherwise be available to you under applicable patent law.
+
+  12. No Surrender of Others' Freedom.
+
+  If conditions are imposed on you (whether by court order, agreement or
+otherwise) that contradict the conditions of this License, they do not
+excuse you from the conditions of this License.  If you cannot convey a
+covered work so as to satisfy simultaneously your obligations under this
+License and any other pertinent obligations, then as a consequence you may
+not convey it at all.  For example, if you agree to terms that obligate you
+to collect a royalty for further conveying from those to whom you convey
+the Program, the only way you could satisfy both those terms and this
+License would be to refrain entirely from conveying the Program.
+
+  13. Use with the GNU Affero General Public License.
+
+  Notwithstanding any other provision of this License, you have
+permission to link or combine any covered work with a work licensed
+under version 3 of the GNU Affero General Public License into a single
+combined work, and to convey the resulting work.  The terms of this
+License will continue to apply to the part which is the covered work,
+but the special requirements of the GNU Affero General Public License,
+section 13, concerning interaction through a network will apply to the
+combination as such.
+
+  14. Revised Versions of this License.
+
+  The Free Software Foundation may publish revised and/or new versions of
+the GNU General Public License from time to time.  Such new versions will
+be similar in spirit to the present version, but may differ in detail to
+address new problems or concerns.
+
+  Each version is given a distinguishing version number.  If the
+Program specifies that a certain numbered version of the GNU General
+Public License "or any later version" applies to it, you have the
+option of following the terms and conditions either of that numbered
+version or of any later version published by the Free Software
+Foundation.  If the Program does not specify a version number of the
+GNU General Public License, you may choose any version ever published
+by the Free Software Foundation.
+
+  If the Program specifies that a proxy can decide which future
+versions of the GNU General Public License can be used, that proxy's
+public statement of acceptance of a version permanently authorizes you
+to choose that version for the Program.
+
+  Later license versions may give you additional or different
+permissions.  However, no additional obligations are imposed on any
+author or copyright holder as a result of your choosing to follow a
+later version.
+
+  15. Disclaimer of Warranty.
+
+  THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
+APPLICABLE LAW.  EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT
+HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY
+OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO,
+THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+PURPOSE.  THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM
+IS WITH YOU.  SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF
+ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
+
+  16. Limitation of Liability.
+
+  IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
+WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS
+THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY
+GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE
+USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF
+DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD
+PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS),
+EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
+SUCH DAMAGES.
+
+  17. Interpretation of Sections 15 and 16.
+
+  If the disclaimer of warranty and limitation of liability provided
+above cannot be given local legal effect according to their terms,
+reviewing courts shall apply local law that most closely approximates
+an absolute waiver of all civil liability in connection with the
+Program, unless a warranty or assumption of liability accompanies a
+copy of the Program in return for a fee.
+
+                     END OF TERMS AND CONDITIONS
+
+            How to Apply These Terms to Your New Programs
+
+  If you develop a new program, and you want it to be of the greatest
+possible use to the public, the best way to achieve this is to make it
+free software which everyone can redistribute and change under these terms.
+
+  To do so, attach the following notices to the program.  It is safest
+to attach them to the start of each source file to most effectively
+state the exclusion of warranty; and each file should have at least
+the "copyright" line and a pointer to where the full notice is found.
+
+    <one line to give the program's name and a brief idea of what it does.>
+    Copyright (C) <year>  <name of author>
+
+    This program is free software: you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation, either version 3 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+Also add information on how to contact you by electronic and paper mail.
+
+  If the program does terminal interaction, make it output a short
+notice like this when it starts in an interactive mode:
+
+    <program>  Copyright (C) <year>  <name of author>
+    This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
+    This is free software, and you are welcome to redistribute it
+    under certain conditions; type `show c' for details.
+
+The hypothetical commands `show w' and `show c' should show the appropriate
+parts of the General Public License.  Of course, your program's commands
+might be different; for a GUI interface, you would use an "about box".
+
+  You should also get your employer (if you work as a programmer) or school,
+if any, to sign a "copyright disclaimer" for the program, if necessary.
+For more information on this, and how to apply and follow the GNU GPL, see
+<http://www.gnu.org/licenses/>.
+
+  The GNU General Public License does not permit incorporating your program
+into proprietary programs.  If your program is a subroutine library, you
+may consider it more useful to permit linking proprietary applications with
+the library.  If this is what you want to do, use the GNU Lesser General
+Public License instead of this License.  But first, please read
+<http://www.gnu.org/philosophy/why-not-lgpl.html>.
diff --git a/INSTALL b/INSTALL
new file mode 100644
index 00000000..20998407
--- /dev/null
+++ b/INSTALL
@@ -0,0 +1,370 @@
+Installation Instructions
+*************************
+
+Copyright (C) 1994-1996, 1999-2002, 2004-2013 Free Software Foundation,
+Inc.
+
+   Copying and distribution of this file, with or without modification,
+are permitted in any medium without royalty provided the copyright
+notice and this notice are preserved.  This file is offered as-is,
+without warranty of any kind.
+
+Basic Installation
+==================
+
+   Briefly, the shell command `./configure && make && make install'
+should configure, build, and install this package.  The following
+more-detailed instructions are generic; see the `README' file for
+instructions specific to this package.  Some packages provide this
+`INSTALL' file but do not implement all of the features documented
+below.  The lack of an optional feature in a given package is not
+necessarily a bug.  More recommendations for GNU packages can be found
+in *note Makefile Conventions: (standards)Makefile Conventions.
+
+   The `configure' shell script attempts to guess correct values for
+various system-dependent variables used during compilation.  It uses
+those values to create a `Makefile' in each directory of the package.
+It may also create one or more `.h' files containing system-dependent
+definitions.  Finally, it creates a shell script `config.status' that
+you can run in the future to recreate the current configuration, and a
+file `config.log' containing compiler output (useful mainly for
+debugging `configure').
+
+   It can also use an optional file (typically called `config.cache'
+and enabled with `--cache-file=config.cache' or simply `-C') that saves
+the results of its tests to speed up reconfiguring.  Caching is
+disabled by default to prevent problems with accidental use of stale
+cache files.
+
+   If you need to do unusual things to compile the package, please try
+to figure out how `configure' could check whether to do them, and mail
+diffs or instructions to the address given in the `README' so they can
+be considered for the next release.  If you are using the cache, and at
+some point `config.cache' contains results you don't want to keep, you
+may remove or edit it.
+
+   The file `configure.ac' (or `configure.in') is used to create
+`configure' by a program called `autoconf'.  You need `configure.ac' if
+you want to change it or regenerate `configure' using a newer version
+of `autoconf'.
+
+   The simplest way to compile this package is:
+
+  1. `cd' to the directory containing the package's source code and type
+     `./configure' to configure the package for your system.
+
+     Running `configure' might take a while.  While running, it prints
+     some messages telling which features it is checking for.
+
+  2. Type `make' to compile the package.
+
+  3. Optionally, type `make check' to run any self-tests that come with
+     the package, generally using the just-built uninstalled binaries.
+
+  4. Type `make install' to install the programs and any data files and
+     documentation.  When installing into a prefix owned by root, it is
+     recommended that the package be configured and built as a regular
+     user, and only the `make install' phase executed with root
+     privileges.
+
+  5. Optionally, type `make installcheck' to repeat any self-tests, but
+     this time using the binaries in their final installed location.
+     This target does not install anything.  Running this target as a
+     regular user, particularly if the prior `make install' required
+     root privileges, verifies that the installation completed
+     correctly.
+
+  6. You can remove the program binaries and object files from the
+     source code directory by typing `make clean'.  To also remove the
+     files that `configure' created (so you can compile the package for
+     a different kind of computer), type `make distclean'.  There is
+     also a `make maintainer-clean' target, but that is intended mainly
+     for the package's developers.  If you use it, you may have to get
+     all sorts of other programs in order to regenerate files that came
+     with the distribution.
+
+  7. Often, you can also type `make uninstall' to remove the installed
+     files again.  In practice, not all packages have tested that
+     uninstallation works correctly, even though it is required by the
+     GNU Coding Standards.
+
+  8. Some packages, particularly those that use Automake, provide `make
+     distcheck', which can by used by developers to test that all other
+     targets like `make install' and `make uninstall' work correctly.
+     This target is generally not run by end users.
+
+Compilers and Options
+=====================
+
+   Some systems require unusual options for compilation or linking that
+the `configure' script does not know about.  Run `./configure --help'
+for details on some of the pertinent environment variables.
+
+   You can give `configure' initial values for configuration parameters
+by setting variables in the command line or in the environment.  Here
+is an example:
+
+     ./configure CC=c99 CFLAGS=-g LIBS=-lposix
+
+   *Note Defining Variables::, for more details.
+
+Compiling For Multiple Architectures
+====================================
+
+   You can compile the package for more than one kind of computer at the
+same time, by placing the object files for each architecture in their
+own directory.  To do this, you can use GNU `make'.  `cd' to the
+directory where you want the object files and executables to go and run
+the `configure' script.  `configure' automatically checks for the
+source code in the directory that `configure' is in and in `..'.  This
+is known as a "VPATH" build.
+
+   With a non-GNU `make', it is safer to compile the package for one
+architecture at a time in the source code directory.  After you have
+installed the package for one architecture, use `make distclean' before
+reconfiguring for another architecture.
+
+   On MacOS X 10.5 and later systems, you can create libraries and
+executables that work on multiple system types--known as "fat" or
+"universal" binaries--by specifying multiple `-arch' options to the
+compiler but only a single `-arch' option to the preprocessor.  Like
+this:
+
+     ./configure CC="gcc -arch i386 -arch x86_64 -arch ppc -arch ppc64" \
+                 CXX="g++ -arch i386 -arch x86_64 -arch ppc -arch ppc64" \
+                 CPP="gcc -E" CXXCPP="g++ -E"
+
+   This is not guaranteed to produce working output in all cases, you
+may have to build one architecture at a time and combine the results
+using the `lipo' tool if you have problems.
+
+Installation Names
+==================
+
+   By default, `make install' installs the package's commands under
+`/usr/local/bin', include files under `/usr/local/include', etc.  You
+can specify an installation prefix other than `/usr/local' by giving
+`configure' the option `--prefix=PREFIX', where PREFIX must be an
+absolute file name.
+
+   You can specify separate installation prefixes for
+architecture-specific files and architecture-independent files.  If you
+pass the option `--exec-prefix=PREFIX' to `configure', the package uses
+PREFIX as the prefix for installing programs and libraries.
+Documentation and other data files still use the regular prefix.
+
+   In addition, if you use an unusual directory layout you can give
+options like `--bindir=DIR' to specify different values for particular
+kinds of files.  Run `configure --help' for a list of the directories
+you can set and what kinds of files go in them.  In general, the
+default for these options is expressed in terms of `${prefix}', so that
+specifying just `--prefix' will affect all of the other directory
+specifications that were not explicitly provided.
+
+   The most portable way to affect installation locations is to pass the
+correct locations to `configure'; however, many packages provide one or
+both of the following shortcuts of passing variable assignments to the
+`make install' command line to change installation locations without
+having to reconfigure or recompile.
+
+   The first method involves providing an override variable for each
+affected directory.  For example, `make install
+prefix=/alternate/directory' will choose an alternate location for all
+directory configuration variables that were expressed in terms of
+`${prefix}'.  Any directories that were specified during `configure',
+but not in terms of `${prefix}', must each be overridden at install
+time for the entire installation to be relocated.  The approach of
+makefile variable overrides for each directory variable is required by
+the GNU Coding Standards, and ideally causes no recompilation.
+However, some platforms have known limitations with the semantics of
+shared libraries that end up requiring recompilation when using this
+method, particularly noticeable in packages that use GNU Libtool.
+
+   The second method involves providing the `DESTDIR' variable.  For
+example, `make install DESTDIR=/alternate/directory' will prepend
+`/alternate/directory' before all installation names.  The approach of
+`DESTDIR' overrides is not required by the GNU Coding Standards, and
+does not work on platforms that have drive letters.  On the other hand,
+it does better at avoiding recompilation issues, and works well even
+when some directory options were not specified in terms of `${prefix}'
+at `configure' time.
+
+Optional Features
+=================
+
+   If the package supports it, you can cause programs to be installed
+with an extra prefix or suffix on their names by giving `configure' the
+option `--program-prefix=PREFIX' or `--program-suffix=SUFFIX'.
+
+   Some packages pay attention to `--enable-FEATURE' options to
+`configure', where FEATURE indicates an optional part of the package.
+They may also pay attention to `--with-PACKAGE' options, where PACKAGE
+is something like `gnu-as' or `x' (for the X Window System).  The
+`README' should mention any `--enable-' and `--with-' options that the
+package recognizes.
+
+   For packages that use the X Window System, `configure' can usually
+find the X include and library files automatically, but if it doesn't,
+you can use the `configure' options `--x-includes=DIR' and
+`--x-libraries=DIR' to specify their locations.
+
+   Some packages offer the ability to configure how verbose the
+execution of `make' will be.  For these packages, running `./configure
+--enable-silent-rules' sets the default to minimal output, which can be
+overridden with `make V=1'; while running `./configure
+--disable-silent-rules' sets the default to verbose, which can be
+overridden with `make V=0'.
+
+Particular systems
+==================
+
+   On HP-UX, the default C compiler is not ANSI C compatible.  If GNU
+CC is not installed, it is recommended to use the following options in
+order to use an ANSI C compiler:
+
+     ./configure CC="cc -Ae -D_XOPEN_SOURCE=500"
+
+and if that doesn't work, install pre-built binaries of GCC for HP-UX.
+
+   HP-UX `make' updates targets which have the same time stamps as
+their prerequisites, which makes it generally unusable when shipped
+generated files such as `configure' are involved.  Use GNU `make'
+instead.
+
+   On OSF/1 a.k.a. Tru64, some versions of the default C compiler cannot
+parse its `<wchar.h>' header file.  The option `-nodtk' can be used as
+a workaround.  If GNU CC is not installed, it is therefore recommended
+to try
+
+     ./configure CC="cc"
+
+and if that doesn't work, try
+
+     ./configure CC="cc -nodtk"
+
+   On Solaris, don't put `/usr/ucb' early in your `PATH'.  This
+directory contains several dysfunctional programs; working variants of
+these programs are available in `/usr/bin'.  So, if you need `/usr/ucb'
+in your `PATH', put it _after_ `/usr/bin'.
+
+   On Haiku, software installed for all users goes in `/boot/common',
+not `/usr/local'.  It is recommended to use the following options:
+
+     ./configure --prefix=/boot/common
+
+Specifying the System Type
+==========================
+
+   There may be some features `configure' cannot figure out
+automatically, but needs to determine by the type of machine the package
+will run on.  Usually, assuming the package is built to be run on the
+_same_ architectures, `configure' can figure that out, but if it prints
+a message saying it cannot guess the machine type, give it the
+`--build=TYPE' option.  TYPE can either be a short name for the system
+type, such as `sun4', or a canonical name which has the form:
+
+     CPU-COMPANY-SYSTEM
+
+where SYSTEM can have one of these forms:
+
+     OS
+     KERNEL-OS
+
+   See the file `config.sub' for the possible values of each field.  If
+`config.sub' isn't included in this package, then this package doesn't
+need to know the machine type.
+
+   If you are _building_ compiler tools for cross-compiling, you should
+use the option `--target=TYPE' to select the type of system they will
+produce code for.
+
+   If you want to _use_ a cross compiler, that generates code for a
+platform different from the build platform, you should specify the
+"host" platform (i.e., that on which the generated programs will
+eventually be run) with `--host=TYPE'.
+
+Sharing Defaults
+================
+
+   If you want to set default values for `configure' scripts to share,
+you can create a site shell script called `config.site' that gives
+default values for variables like `CC', `cache_file', and `prefix'.
+`configure' looks for `PREFIX/share/config.site' if it exists, then
+`PREFIX/etc/config.site' if it exists.  Or, you can set the
+`CONFIG_SITE' environment variable to the location of the site script.
+A warning: not all `configure' scripts look for a site script.
+
+Defining Variables
+==================
+
+   Variables not defined in a site shell script can be set in the
+environment passed to `configure'.  However, some packages may run
+configure again during the build, and the customized values of these
+variables may be lost.  In order to avoid this problem, you should set
+them in the `configure' command line, using `VAR=value'.  For example:
+
+     ./configure CC=/usr/local2/bin/gcc
+
+causes the specified `gcc' to be used as the C compiler (unless it is
+overridden in the site shell script).
+
+Unfortunately, this technique does not work for `CONFIG_SHELL' due to
+an Autoconf limitation.  Until the limitation is lifted, you can use
+this workaround:
+
+     CONFIG_SHELL=/bin/bash ./configure CONFIG_SHELL=/bin/bash
+
+`configure' Invocation
+======================
+
+   `configure' recognizes the following options to control how it
+operates.
+
+`--help'
+`-h'
+     Print a summary of all of the options to `configure', and exit.
+
+`--help=short'
+`--help=recursive'
+     Print a summary of the options unique to this package's
+     `configure', and exit.  The `short' variant lists options used
+     only in the top level, while the `recursive' variant lists options
+     also present in any nested packages.
+
+`--version'
+`-V'
+     Print the version of Autoconf used to generate the `configure'
+     script, and exit.
+
+`--cache-file=FILE'
+     Enable the cache: use and save the results of the tests in FILE,
+     traditionally `config.cache'.  FILE defaults to `/dev/null' to
+     disable caching.
+
+`--config-cache'
+`-C'
+     Alias for `--cache-file=config.cache'.
+
+`--quiet'
+`--silent'
+`-q'
+     Do not print messages saying which checks are being made.  To
+     suppress all normal output, redirect it to `/dev/null' (any error
+     messages will still be shown).
+
+`--srcdir=DIR'
+     Look for the package's source code in directory DIR.  Usually
+     `configure' can determine that directory automatically.
+
+`--prefix=DIR'
+     Use DIR as the installation prefix.  *note Installation Names::
+     for more details, including other options available for fine-tuning
+     the installation locations.
+
+`--no-create'
+`-n'
+     Run the configure checks, but stop before creating any output
+     files.
+
+`configure' also accepts some other, not widely useful, options.  Run
+`configure --help' for more details.

From 4e8b9c6928301a9b90bf8935d30b0b3e5c6c4a38 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@frontend05.ecdf.ed.ac.uk>
Date: Wed, 4 Mar 2015 13:25:23 +0000
Subject: [PATCH 015/429] Changes for MIC

---
 Grid.h           |    2 +
 Grid_Lattice.h   |  134 +-
 Grid_config.h    |   14 +-
 Grid_config.h.in |    6 -
 Grid_main.cc     |   12 +-
 Makefile.in      |  360 +---
 aclocal.m4       |  708 +++----
 configure        | 4701 ++++++++++++++++++++++++++--------------------
 configure.ac     |    3 +-
 9 files changed, 3153 insertions(+), 2787 deletions(-)

diff --git a/Grid.h b/Grid.h
index 0cbe583b..f73d52ba 100644
--- a/Grid.h
+++ b/Grid.h
@@ -20,12 +20,14 @@
 #include <stdlib.h>
 #include <sys/time.h>
 #include <stdio.h>
+#include <signal.h>
 
 #include <Grid_config.h>
 
 ////////////////////////////////////////////////////////////
 // Tunable header includes
 ////////////////////////////////////////////////////////////
+#define HAVE_OPENMP
 #ifdef HAVE_OPENMP
 #define OMP
 #include <omp.h>
diff --git a/Grid_Lattice.h b/Grid_Lattice.h
index d0fb9465..223ad2cd 100644
--- a/Grid_Lattice.h
+++ b/Grid_Lattice.h
@@ -69,7 +69,8 @@ public:
         return ret;
     };
 
-
+#if 0 
+    // Collapse doesn't appear to work the way I think it should in icpc
     friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
     {
         Lattice<vobj> ret(rhs._grid);
@@ -131,25 +132,23 @@ public:
                 if ( x<shift%rd ) permute_slice = 1-permute_slice;
             }
 
+
+#if 0
             if ( permute_slice ) {
-
-	        int internal=sizeof(vobj)/sizeof(vComplex);
+                
+                int internal=sizeof(vobj)/sizeof(vComplex);
                 int num =rhs._grid->_slice_block[dimension]*internal;
-
-#pragma omp parallel for collapse(2)
+                
                 for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+                    vComplex *optr = (vComplex *)&ret._odata[o];
+                    vComplex *iptr = (vComplex *)&rhs._odata[so];
                     for(int b=0;b<num;b++){
-		      vComplex *optr = (vComplex *)&ret._odata[o];
-		      vComplex *iptr = (vComplex *)&rhs._odata[so];
-                      permute(optr[b],iptr[b],permute_type);
+                        permute(optr[b],iptr[b],permute_type);
                     }
                     o+=rhs._grid->_slice_stride[dimension];
                     so+=rhs._grid->_slice_stride[dimension];
                 }
-
             } else {
-
-#pragma omp parallel for collapse(2)
                 for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
                     for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
                         ret._odata[o+i]=rhs._odata[so+i];
@@ -159,9 +158,122 @@ public:
                 }
             
             }
+#else
+
+            if ( permute_slice ) {
+
+	        int internal=sizeof(vobj)/sizeof(vComplex);
+                int num =rhs._grid->_slice_block[dimension]*internal;
+
+
+#pragma omp parallel for collapse(2)
+                for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+                    for(int b=0;b<num;b++){
+		      vComplex *optr = (vComplex *)&ret._odata[o +n*rhs._grid->_slice_stride[dimension]];
+		      vComplex *iptr = (vComplex *)&rhs._odata[so+n*rhs._grid->_slice_stride[dimension]];
+                      permute(optr[b],iptr[b],permute_type);
+                    }
+                }
+
+            } else {
+
+
+#pragma omp parallel for collapse(2)
+                for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+                    for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
+		      int oo = o+ n*rhs._grid->_slice_stride[dimension];
+		      int soo=so+ n*rhs._grid->_slice_stride[dimension];
+		      ret._odata[oo+i]=rhs._odata[soo+i];
+                    }
+                }
+            
+            }
+#endif
         }
         return ret;
     }
+#else
+    friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
+    {
+        Lattice<vobj> ret(rhs._grid);
+        
+        ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
+        shift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift);
+        int rd = rhs._grid->_rdimensions[dimension];
+        int ld = rhs._grid->_dimensions[dimension];
+        
+        // Map to always positive shift.
+        shift = (shift+ld)%ld;
+        
+        // Work out whether to permute and the permute type
+        // ABCDEFGH ->   AE BF CG DH       permute
+        // Shift 0       AE BF CG DH       0 0 0 0    ABCDEFGH
+        // Shift 1       DH AE BF CG       1 0 0 0    HABCDEFG
+        // Shift 2       CG DH AE BF       1 1 0 0    GHABCDEF
+        // Shift 3       BF CG DH AE       1 1 1 0    FGHACBDE
+        // Shift 4       AE BF CG DH       1 1 1 1    EFGHABCD
+        // Shift 5       DH AE BF CG       0 1 1 1    DEFGHABC
+        // Shift 6       CG DH AE BF       0 0 1 1    CDEFGHAB
+        // Shift 7       BF CG DH AE       0 0 0 1    BCDEFGHA
+        int permute_dim =rhs._grid->_layout[dimension]>1 ;
+        int permute_type=0;
+        for(int d=0;d<dimension;d++)
+            if (rhs._grid->_layout[d]>1 ) permute_type++;
+        
+        
+        // loop over all work
+        int work =rd*rhs._grid->_slice_nblock[dimension]*rhs._grid->_slice_block[dimension];
+
+#pragma omp parallel for
+        for(int ww=0;ww<work;ww++){
+
+            
+            // can optimise this if know w moves congtiguously for a given thread.
+            // b=(b+1);
+            // if (b==_slice_block) {b=0; n=n+1;}
+            // if (n==_slice_nblock) { n=0; x=x+1}
+            //
+            // Perhaps a five cycle iterator, or so.
+            int w=ww;
+            int b = w%rhs._grid->_slice_block[dimension] ; w=w/rhs._grid->_slice_block[dimension];
+            int n = w%rhs._grid->_slice_nblock[dimension]; w=w/rhs._grid->_slice_nblock[dimension];
+            int x = w;
+
+	    int sx,so,o;
+            sx = (x-shift+ld)%rd;
+            o  = x*rhs._grid->_ostride[dimension]+n*rhs._grid->_slice_stride[dimension]; // common sub expression alert.
+            so =sx*rhs._grid->_ostride[dimension]+n*rhs._grid->_slice_stride[dimension];
+            
+            int permute_slice=0;
+            if ( permute_dim ) {
+                permute_slice = shift/rd;
+                if ( x<shift%rd ) permute_slice = 1-permute_slice;
+            }
+            
+            if ( permute_slice ) {
+                
+                int internal=sizeof(vobj)/sizeof(vComplex);
+                vComplex *optr = (vComplex *)&ret._odata[o+b];
+                vComplex *iptr = (vComplex *)&rhs._odata[so+b];
+		const char *pf = (const char *)iptr;
+		for(int i=0;i<sizeof(vobj);i+=64){
+		  _mm_prefetch(pf+i,_MM_HINT_T0);
+		}
+
+                for(int i=0;i<internal;i++){
+                    permute(optr[i],iptr[i],permute_type);
+                }
+            } else {
+	        const char *pf = (const char *) &rhs._odata[so+b];
+		for(int i=0;i<sizeof(vobj);i+=64){
+		  _mm_prefetch(pf+i,_MM_HINT_T0);
+		}
+                ret._odata[o+b]=rhs._odata[so+b];
+            }
+        }
+        return ret;
+    }
+#endif
 
     template<class sobj>
     inline Lattice<vobj> & operator = (const sobj & r){
diff --git a/Grid_config.h b/Grid_config.h
index b882917d..12eeb8ac 100644
--- a/Grid_config.h
+++ b/Grid_config.h
@@ -2,13 +2,13 @@
 /* Grid_config.h.in.  Generated from configure.ac by autoheader.  */
 
 /* AVX */
-#define AVX1 1
+/* #undef AVX1 */
 
 /* AVX2 */
 /* #undef AVX2 */
 
 /* AVX512 */
-/* #undef AVX512 */
+#define AVX512 1
 
 /* Define to 1 if you have the `gettimeofday' function. */
 #define HAVE_GETTIMEOFDAY 1
@@ -17,10 +17,10 @@
 #define HAVE_INTTYPES_H 1
 
 /* Define to 1 if you have the <malloc.h> header file. */
-/* #undef HAVE_MALLOC_H */
+#define HAVE_MALLOC_H 1
 
 /* Define to 1 if you have the <malloc/malloc.h> header file. */
-#define HAVE_MALLOC_MALLOC_H 1
+/* #undef HAVE_MALLOC_MALLOC_H */
 
 /* Define to 1 if you have the <memory.h> header file. */
 #define HAVE_MEMORY_H 1
@@ -46,9 +46,6 @@
 /* Define to 1 if you have the <unistd.h> header file. */
 #define HAVE_UNISTD_H 1
 
-/* Define to 1 if the system has the `aligned' variable attribute */
-#define HAVE_VAR_ATTRIBUTE_ALIGNED 1
-
 /* Name of package */
 #define PACKAGE "grid"
 
@@ -64,9 +61,6 @@
 /* Define to the one symbol short name of this package. */
 #define PACKAGE_TARNAME "grid"
 
-/* Define to the home page for this package. */
-#define PACKAGE_URL ""
-
 /* Define to the version of this package. */
 #define PACKAGE_VERSION "1.0"
 
diff --git a/Grid_config.h.in b/Grid_config.h.in
index 88003589..5c57e4cf 100644
--- a/Grid_config.h.in
+++ b/Grid_config.h.in
@@ -45,9 +45,6 @@
 /* Define to 1 if you have the <unistd.h> header file. */
 #undef HAVE_UNISTD_H
 
-/* Define to 1 if the system has the `aligned' variable attribute */
-#undef HAVE_VAR_ATTRIBUTE_ALIGNED
-
 /* Name of package */
 #undef PACKAGE
 
@@ -63,9 +60,6 @@
 /* Define to the one symbol short name of this package. */
 #undef PACKAGE_TARNAME
 
-/* Define to the home page for this package. */
-#undef PACKAGE_URL
-
 /* Define to the version of this package. */
 #undef PACKAGE_VERSION
 
diff --git a/Grid_main.cc b/Grid_main.cc
index 45dbaccb..0b0c3e99 100644
--- a/Grid_main.cc
+++ b/Grid_main.cc
@@ -27,13 +27,17 @@ int main (int argc, char ** argv)
     std::vector<int> latt_size(4);
     std::vector<int> simd_layout(4);
 
- for(int omp=1;omp<2;omp*=2){
+#ifdef AVX512
+ for(int omp=128;omp<236;omp+=16){
+#else
+ for(int omp=1;omp<8;omp*=2){
+#endif
 
 #ifdef OMP
    omp_set_num_threads(omp);
 #endif 
 
-  for(int lat=8;lat<=12;lat+=2){
+  for(int lat=16;lat<=20;lat+=4){
     latt_size[0] = lat;
     latt_size[1] = lat;
     latt_size[2] = lat;
@@ -191,6 +195,8 @@ int main (int argc, char ** argv)
 
     mult(FooBar,Foo,Bar);
     FooBar = Foo * Bar;
+
+    bytes = ncall*1.0*volume*Nc*Nc    *2*5*sizeof(dpo::Real);
     t0=usecond();
     for(int i=0;i<ncall;i++){
       mult(FooBar,Foo,Cshift(Bar,1,-2));
@@ -198,6 +204,8 @@ int main (int argc, char ** argv)
       //FooBar = Foo * Bar; // this is bad
     }
     t1=usecond();
+
+    FooBar = Foo * Bar;
     
     printf("Cshift Mult: NumThread %d , Lattice size %d , %f us per call\n",omp,lat,(t1-t0)/ncall);
     printf("Cshift Mult: NumThread %d , Lattice size %d , %f Mflop/s\n",omp,lat,flops/(t1-t0));
diff --git a/Makefile.in b/Makefile.in
index c1d09a38..b841357d 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -1,8 +1,9 @@
-# Makefile.in generated by automake 1.15 from Makefile.am.
+# Makefile.in generated by automake 1.11.1 from Makefile.am.
 # @configure_input@
 
-# Copyright (C) 1994-2014 Free Software Foundation, Inc.
-
+# Copyright (C) 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002,
+# 2003, 2004, 2005, 2006, 2007, 2008, 2009  Free Software Foundation,
+# Inc.
 # This Makefile.in is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
@@ -17,61 +18,6 @@
 
 
 VPATH = @srcdir@
-am__is_gnu_make = { \
-  if test -z '$(MAKELEVEL)'; then \
-    false; \
-  elif test -n '$(MAKE_HOST)'; then \
-    true; \
-  elif test -n '$(MAKE_VERSION)' && test -n '$(CURDIR)'; then \
-    true; \
-  else \
-    false; \
-  fi; \
-}
-am__make_running_with_option = \
-  case $${target_option-} in \
-      ?) ;; \
-      *) echo "am__make_running_with_option: internal error: invalid" \
-              "target option '$${target_option-}' specified" >&2; \
-         exit 1;; \
-  esac; \
-  has_opt=no; \
-  sane_makeflags=$$MAKEFLAGS; \
-  if $(am__is_gnu_make); then \
-    sane_makeflags=$$MFLAGS; \
-  else \
-    case $$MAKEFLAGS in \
-      *\\[\ \	]*) \
-        bs=\\; \
-        sane_makeflags=`printf '%s\n' "$$MAKEFLAGS" \
-          | sed "s/$$bs$$bs[$$bs $$bs	]*//g"`;; \
-    esac; \
-  fi; \
-  skip_next=no; \
-  strip_trailopt () \
-  { \
-    flg=`printf '%s\n' "$$flg" | sed "s/$$1.*$$//"`; \
-  }; \
-  for flg in $$sane_makeflags; do \
-    test $$skip_next = yes && { skip_next=no; continue; }; \
-    case $$flg in \
-      *=*|--*) continue;; \
-        -*I) strip_trailopt 'I'; skip_next=yes;; \
-      -*I?*) strip_trailopt 'I';; \
-        -*O) strip_trailopt 'O'; skip_next=yes;; \
-      -*O?*) strip_trailopt 'O';; \
-        -*l) strip_trailopt 'l'; skip_next=yes;; \
-      -*l?*) strip_trailopt 'l';; \
-      -[dEDm]) skip_next=yes;; \
-      -[JT]) skip_next=yes;; \
-    esac; \
-    case $$flg in \
-      *$$target_option*) has_opt=yes; break;; \
-    esac; \
-  done; \
-  test $$has_opt = yes
-am__make_dryrun = (target_option=n; $(am__make_running_with_option))
-am__make_keepgoing = (target_option=k; $(am__make_running_with_option))
 pkgdatadir = $(datadir)/@PACKAGE@
 pkgincludedir = $(includedir)/@PACKAGE@
 pkglibdir = $(libdir)/@PACKAGE@
@@ -90,13 +36,14 @@ PRE_UNINSTALL = :
 POST_UNINSTALL = :
 bin_PROGRAMS = Grid_main$(EXEEXT)
 subdir = .
+DIST_COMMON = README $(am__configure_deps) $(include_HEADERS) \
+	$(srcdir)/Grid_config.h.in $(srcdir)/Makefile.am \
+	$(srcdir)/Makefile.in $(top_srcdir)/configure AUTHORS COPYING \
+	ChangeLog INSTALL NEWS compile depcomp install-sh missing
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
-am__aclocal_m4_deps = $(top_srcdir)/m4/ax_gcc_var_attribute.m4 \
-	$(top_srcdir)/configure.ac
+am__aclocal_m4_deps = $(top_srcdir)/configure.ac
 am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
 	$(ACLOCAL_M4)
-DIST_COMMON = $(srcdir)/Makefile.am $(top_srcdir)/configure \
-	$(am__configure_deps) $(include_HEADERS) $(am__DIST_COMMON)
 am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \
  configure.lineno config.status.lineno
 mkinstalldirs = $(install_sh) -d
@@ -124,21 +71,11 @@ am__nobase_list = $(am__nobase_strip_setup); \
 am__base_list = \
   sed '$$!N;$$!N;$$!N;$$!N;$$!N;$$!N;$$!N;s/\n/ /g' | \
   sed '$$!N;$$!N;$$!N;$$!N;s/\n/ /g'
-am__uninstall_files_from_dir = { \
-  test -z "$$files" \
-    || { test ! -d "$$dir" && test ! -f "$$dir" && test ! -r "$$dir"; } \
-    || { echo " ( cd '$$dir' && rm -f" $$files ")"; \
-         $(am__cd) "$$dir" && rm -f $$files; }; \
-  }
 am__installdirs = "$(DESTDIR)$(libdir)" "$(DESTDIR)$(bindir)" \
 	"$(DESTDIR)$(includedir)"
 LIBRARIES = $(lib_LIBRARIES)
 AR = ar
 ARFLAGS = cru
-AM_V_AR = $(am__v_AR_@AM_V@)
-am__v_AR_ = $(am__v_AR_@AM_DEFAULT_V@)
-am__v_AR_0 = @echo "  AR      " $@;
-am__v_AR_1 = 
 libGrid_a_AR = $(AR) $(ARFLAGS)
 libGrid_a_LIBADD =
 am_libGrid_a_OBJECTS = Grid_signal.$(OBJEXT)
@@ -147,88 +84,33 @@ PROGRAMS = $(bin_PROGRAMS)
 am_Grid_main_OBJECTS = Grid_main.$(OBJEXT)
 Grid_main_OBJECTS = $(am_Grid_main_OBJECTS)
 Grid_main_DEPENDENCIES = libGrid.a
-AM_V_P = $(am__v_P_@AM_V@)
-am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
-am__v_P_0 = false
-am__v_P_1 = :
-AM_V_GEN = $(am__v_GEN_@AM_V@)
-am__v_GEN_ = $(am__v_GEN_@AM_DEFAULT_V@)
-am__v_GEN_0 = @echo "  GEN     " $@;
-am__v_GEN_1 = 
-AM_V_at = $(am__v_at_@AM_V@)
-am__v_at_ = $(am__v_at_@AM_DEFAULT_V@)
-am__v_at_0 = @
-am__v_at_1 = 
 DEFAULT_INCLUDES = -I.@am__isrc@
 depcomp = $(SHELL) $(top_srcdir)/depcomp
 am__depfiles_maybe = depfiles
 am__mv = mv -f
 CXXCOMPILE = $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) \
 	$(AM_CPPFLAGS) $(CPPFLAGS) $(AM_CXXFLAGS) $(CXXFLAGS)
-AM_V_CXX = $(am__v_CXX_@AM_V@)
-am__v_CXX_ = $(am__v_CXX_@AM_DEFAULT_V@)
-am__v_CXX_0 = @echo "  CXX     " $@;
-am__v_CXX_1 = 
 CXXLD = $(CXX)
 CXXLINK = $(CXXLD) $(AM_CXXFLAGS) $(CXXFLAGS) $(AM_LDFLAGS) $(LDFLAGS) \
 	-o $@
-AM_V_CXXLD = $(am__v_CXXLD_@AM_V@)
-am__v_CXXLD_ = $(am__v_CXXLD_@AM_DEFAULT_V@)
-am__v_CXXLD_0 = @echo "  CXXLD   " $@;
-am__v_CXXLD_1 = 
 SOURCES = $(libGrid_a_SOURCES) $(Grid_main_SOURCES)
 DIST_SOURCES = $(libGrid_a_SOURCES) $(Grid_main_SOURCES)
-am__can_run_installinfo = \
-  case $$AM_UPDATE_INFO_DIR in \
-    n|no|NO) false;; \
-    *) (install-info --version) >/dev/null 2>&1;; \
-  esac
 HEADERS = $(include_HEADERS)
-am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) \
-	$(LISP)Grid_config.h.in
-# Read a list of newline-separated strings from the standard input,
-# and print each of them once, without duplicates.  Input order is
-# *not* preserved.
-am__uniquify_input = $(AWK) '\
-  BEGIN { nonempty = 0; } \
-  { items[$$0] = 1; nonempty = 1; } \
-  END { if (nonempty) { for (i in items) print i; }; } \
-'
-# Make sure the list of sources is unique.  This is necessary because,
-# e.g., the same source file might be shared among _SOURCES variables
-# for different programs/libraries.
-am__define_uniq_tagged_files = \
-  list='$(am__tagged_files)'; \
-  unique=`for i in $$list; do \
-    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
-  done | $(am__uniquify_input)`
 ETAGS = etags
 CTAGS = ctags
-CSCOPE = cscope
-AM_RECURSIVE_TARGETS = cscope
-am__DIST_COMMON = $(srcdir)/Grid_config.h.in $(srcdir)/Makefile.in \
-	AUTHORS COPYING ChangeLog INSTALL NEWS README compile depcomp \
-	install-sh missing
 DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
 distdir = $(PACKAGE)-$(VERSION)
 top_distdir = $(distdir)
 am__remove_distdir = \
-  if test -d "$(distdir)"; then \
-    find "$(distdir)" -type d ! -perm -200 -exec chmod u+w {} ';' \
-      && rm -rf "$(distdir)" \
-      || { sleep 5 && rm -rf "$(distdir)"; }; \
-  else :; fi
-am__post_remove_distdir = $(am__remove_distdir)
+  { test ! -d "$(distdir)" \
+    || { find "$(distdir)" -type d ! -perm -200 -exec chmod u+w {} ';' \
+         && rm -fr "$(distdir)"; }; }
 DIST_ARCHIVES = $(distdir).tar.gz
 GZIP_ENV = --best
-DIST_TARGETS = dist-gzip
 distuninstallcheck_listfiles = find . -type f -print
-am__distuninstallcheck_listfiles = $(distuninstallcheck_listfiles) \
-  | sed 's|^\./|$(prefix)/|' | grep -v '$(infodir)/dir$$'
 distcleancheck_listfiles = find . -type f -print
 ACLOCAL = @ACLOCAL@
 AMTAR = @AMTAR@
-AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
 AUTOCONF = @AUTOCONF@
 AUTOHEADER = @AUTOHEADER@
 AUTOMAKE = @AUTOMAKE@
@@ -268,7 +150,6 @@ PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
 PACKAGE_NAME = @PACKAGE_NAME@
 PACKAGE_STRING = @PACKAGE_STRING@
 PACKAGE_TARNAME = @PACKAGE_TARNAME@
-PACKAGE_URL = @PACKAGE_URL@
 PACKAGE_VERSION = @PACKAGE_VERSION@
 PATH_SEPARATOR = @PATH_SEPARATOR@
 RANLIB = @RANLIB@
@@ -348,7 +229,7 @@ all: Grid_config.h
 
 .SUFFIXES:
 .SUFFIXES: .cc .o .obj
-am--refresh: Makefile
+am--refresh:
 	@:
 $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	@for dep in $?; do \
@@ -363,6 +244,7 @@ $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu Makefile'; \
 	$(am__cd) $(top_srcdir) && \
 	  $(AUTOMAKE) --gnu Makefile
+.PRECIOUS: Makefile
 Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
 	@case '$?' in \
 	  *config.status*) \
@@ -383,8 +265,10 @@ $(ACLOCAL_M4):  $(am__aclocal_m4_deps)
 $(am__aclocal_m4_deps):
 
 Grid_config.h: stamp-h1
-	@test -f $@ || rm -f stamp-h1
-	@test -f $@ || $(MAKE) $(AM_MAKEFLAGS) stamp-h1
+	@if test ! -f $@; then \
+	  rm -f stamp-h1; \
+	  $(MAKE) $(AM_MAKEFLAGS) stamp-h1; \
+	else :; fi
 
 stamp-h1: $(srcdir)/Grid_config.h.in $(top_builddir)/config.status
 	@rm -f stamp-h1
@@ -398,6 +282,7 @@ distclean-hdr:
 	-rm -f Grid_config.h stamp-h1
 install-libLIBRARIES: $(lib_LIBRARIES)
 	@$(NORMAL_INSTALL)
+	test -z "$(libdir)" || $(MKDIR_P) "$(DESTDIR)$(libdir)"
 	@list='$(lib_LIBRARIES)'; test -n "$(libdir)" || list=; \
 	list2=; for p in $$list; do \
 	  if test -f $$p; then \
@@ -405,8 +290,6 @@ install-libLIBRARIES: $(lib_LIBRARIES)
 	  else :; fi; \
 	done; \
 	test -z "$$list2" || { \
-	  echo " $(MKDIR_P) '$(DESTDIR)$(libdir)'"; \
-	  $(MKDIR_P) "$(DESTDIR)$(libdir)" || exit 1; \
 	  echo " $(INSTALL_DATA) $$list2 '$(DESTDIR)$(libdir)'"; \
 	  $(INSTALL_DATA) $$list2 "$(DESTDIR)$(libdir)" || exit $$?; }
 	@$(POST_INSTALL)
@@ -423,29 +306,26 @@ uninstall-libLIBRARIES:
 	@$(NORMAL_UNINSTALL)
 	@list='$(lib_LIBRARIES)'; test -n "$(libdir)" || list=; \
 	files=`for p in $$list; do echo $$p; done | sed -e 's|^.*/||'`; \
-	dir='$(DESTDIR)$(libdir)'; $(am__uninstall_files_from_dir)
+	test -n "$$files" || exit 0; \
+	echo " ( cd '$(DESTDIR)$(libdir)' && rm -f "$$files" )"; \
+	cd "$(DESTDIR)$(libdir)" && rm -f $$files
 
 clean-libLIBRARIES:
 	-test -z "$(lib_LIBRARIES)" || rm -f $(lib_LIBRARIES)
-
-libGrid.a: $(libGrid_a_OBJECTS) $(libGrid_a_DEPENDENCIES) $(EXTRA_libGrid_a_DEPENDENCIES) 
-	$(AM_V_at)-rm -f libGrid.a
-	$(AM_V_AR)$(libGrid_a_AR) libGrid.a $(libGrid_a_OBJECTS) $(libGrid_a_LIBADD)
-	$(AM_V_at)$(RANLIB) libGrid.a
+libGrid.a: $(libGrid_a_OBJECTS) $(libGrid_a_DEPENDENCIES) 
+	-rm -f libGrid.a
+	$(libGrid_a_AR) libGrid.a $(libGrid_a_OBJECTS) $(libGrid_a_LIBADD)
+	$(RANLIB) libGrid.a
 install-binPROGRAMS: $(bin_PROGRAMS)
 	@$(NORMAL_INSTALL)
+	test -z "$(bindir)" || $(MKDIR_P) "$(DESTDIR)$(bindir)"
 	@list='$(bin_PROGRAMS)'; test -n "$(bindir)" || list=; \
-	if test -n "$$list"; then \
-	  echo " $(MKDIR_P) '$(DESTDIR)$(bindir)'"; \
-	  $(MKDIR_P) "$(DESTDIR)$(bindir)" || exit 1; \
-	fi; \
 	for p in $$list; do echo "$$p $$p"; done | \
 	sed 's/$(EXEEXT)$$//' | \
-	while read p p1; do if test -f $$p \
-	  ; then echo "$$p"; echo "$$p"; else :; fi; \
+	while read p p1; do if test -f $$p; \
+	  then echo "$$p"; echo "$$p"; else :; fi; \
 	done | \
-	sed -e 'p;s,.*/,,;n;h' \
-	    -e 's|.*|.|' \
+	sed -e 'p;s,.*/,,;n;h' -e 's|.*|.|' \
 	    -e 'p;x;s,.*/,,;s/$(EXEEXT)$$//;$(transform);s/$$/$(EXEEXT)/' | \
 	sed 'N;N;N;s,\n, ,g' | \
 	$(AWK) 'BEGIN { files["."] = ""; dirs["."] = 1 } \
@@ -466,18 +346,16 @@ uninstall-binPROGRAMS:
 	@list='$(bin_PROGRAMS)'; test -n "$(bindir)" || list=; \
 	files=`for p in $$list; do echo "$$p"; done | \
 	  sed -e 'h;s,^.*/,,;s/$(EXEEXT)$$//;$(transform)' \
-	      -e 's/$$/$(EXEEXT)/' \
-	`; \
+	      -e 's/$$/$(EXEEXT)/' `; \
 	test -n "$$list" || exit 0; \
 	echo " ( cd '$(DESTDIR)$(bindir)' && rm -f" $$files ")"; \
 	cd "$(DESTDIR)$(bindir)" && rm -f $$files
 
 clean-binPROGRAMS:
 	-test -z "$(bin_PROGRAMS)" || rm -f $(bin_PROGRAMS)
-
-Grid_main$(EXEEXT): $(Grid_main_OBJECTS) $(Grid_main_DEPENDENCIES) $(EXTRA_Grid_main_DEPENDENCIES) 
+Grid_main$(EXEEXT): $(Grid_main_OBJECTS) $(Grid_main_DEPENDENCIES) 
 	@rm -f Grid_main$(EXEEXT)
-	$(AM_V_CXXLD)$(CXXLINK) $(Grid_main_OBJECTS) $(Grid_main_LDADD) $(LIBS)
+	$(CXXLINK) $(Grid_main_OBJECTS) $(Grid_main_LDADD) $(LIBS)
 
 mostlyclean-compile:
 	-rm -f *.$(OBJEXT)
@@ -489,25 +367,22 @@ distclean-compile:
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_signal.Po@am__quote@
 
 .cc.o:
-@am__fastdepCXX_TRUE@	$(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
-@am__fastdepCXX_TRUE@	$(AM_V_at)$(am__mv) $(DEPDIR)/$*.Tpo $(DEPDIR)/$*.Po
-@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
+@am__fastdepCXX_TRUE@	$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
+@am__fastdepCXX_TRUE@	$(am__mv) $(DEPDIR)/$*.Tpo $(DEPDIR)/$*.Po
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
 @AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
-@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ $<
+@am__fastdepCXX_FALSE@	$(CXXCOMPILE) -c -o $@ $<
 
 .cc.obj:
-@am__fastdepCXX_TRUE@	$(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ `$(CYGPATH_W) '$<'`
-@am__fastdepCXX_TRUE@	$(AM_V_at)$(am__mv) $(DEPDIR)/$*.Tpo $(DEPDIR)/$*.Po
-@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
+@am__fastdepCXX_TRUE@	$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ `$(CYGPATH_W) '$<'`
+@am__fastdepCXX_TRUE@	$(am__mv) $(DEPDIR)/$*.Tpo $(DEPDIR)/$*.Po
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
 @AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
-@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ `$(CYGPATH_W) '$<'`
+@am__fastdepCXX_FALSE@	$(CXXCOMPILE) -c -o $@ `$(CYGPATH_W) '$<'`
 install-includeHEADERS: $(include_HEADERS)
 	@$(NORMAL_INSTALL)
+	test -z "$(includedir)" || $(MKDIR_P) "$(DESTDIR)$(includedir)"
 	@list='$(include_HEADERS)'; test -n "$(includedir)" || list=; \
-	if test -n "$$list"; then \
-	  echo " $(MKDIR_P) '$(DESTDIR)$(includedir)'"; \
-	  $(MKDIR_P) "$(DESTDIR)$(includedir)" || exit 1; \
-	fi; \
 	for p in $$list; do \
 	  if test -f "$$p"; then d=; else d="$(srcdir)/"; fi; \
 	  echo "$$d$$p"; \
@@ -521,17 +396,30 @@ uninstall-includeHEADERS:
 	@$(NORMAL_UNINSTALL)
 	@list='$(include_HEADERS)'; test -n "$(includedir)" || list=; \
 	files=`for p in $$list; do echo $$p; done | sed -e 's|^.*/||'`; \
-	dir='$(DESTDIR)$(includedir)'; $(am__uninstall_files_from_dir)
+	test -n "$$files" || exit 0; \
+	echo " ( cd '$(DESTDIR)$(includedir)' && rm -f" $$files ")"; \
+	cd "$(DESTDIR)$(includedir)" && rm -f $$files
 
-ID: $(am__tagged_files)
-	$(am__define_uniq_tagged_files); mkid -fID $$unique
-tags: tags-am
-TAGS: tags
+ID: $(HEADERS) $(SOURCES) $(LISP) $(TAGS_FILES)
+	list='$(SOURCES) $(HEADERS) $(LISP) $(TAGS_FILES)'; \
+	unique=`for i in $$list; do \
+	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+	  done | \
+	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
+	      END { if (nonempty) { for (i in files) print i; }; }'`; \
+	mkid -fID $$unique
+tags: TAGS
 
-tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
+TAGS:  $(HEADERS) $(SOURCES) Grid_config.h.in $(TAGS_DEPENDENCIES) \
+		$(TAGS_FILES) $(LISP)
 	set x; \
 	here=`pwd`; \
-	$(am__define_uniq_tagged_files); \
+	list='$(SOURCES) $(HEADERS) Grid_config.h.in $(LISP) $(TAGS_FILES)'; \
+	unique=`for i in $$list; do \
+	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+	  done | \
+	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
+	      END { if (nonempty) { for (i in files) print i; }; }'`; \
 	shift; \
 	if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
 	  test -n "$$unique" || unique=$$empty_fix; \
@@ -543,11 +431,15 @@ tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
 	      $$unique; \
 	  fi; \
 	fi
-ctags: ctags-am
-
-CTAGS: ctags
-ctags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
-	$(am__define_uniq_tagged_files); \
+ctags: CTAGS
+CTAGS:  $(HEADERS) $(SOURCES) Grid_config.h.in $(TAGS_DEPENDENCIES) \
+		$(TAGS_FILES) $(LISP)
+	list='$(SOURCES) $(HEADERS) Grid_config.h.in $(LISP) $(TAGS_FILES)'; \
+	unique=`for i in $$list; do \
+	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+	  done | \
+	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
+	      END { if (nonempty) { for (i in files) print i; }; }'`; \
 	test -z "$(CTAGS_ARGS)$$unique" \
 	  || $(CTAGS) $(CTAGSFLAGS) $(AM_CTAGSFLAGS) $(CTAGS_ARGS) \
 	     $$unique
@@ -556,31 +448,9 @@ GTAGS:
 	here=`$(am__cd) $(top_builddir) && pwd` \
 	  && $(am__cd) $(top_srcdir) \
 	  && gtags -i $(GTAGS_ARGS) "$$here"
-cscope: cscope.files
-	test ! -s cscope.files \
-	  || $(CSCOPE) -b -q $(AM_CSCOPEFLAGS) $(CSCOPEFLAGS) -i cscope.files $(CSCOPE_ARGS)
-clean-cscope:
-	-rm -f cscope.files
-cscope.files: clean-cscope cscopelist
-cscopelist: cscopelist-am
-
-cscopelist-am: $(am__tagged_files)
-	list='$(am__tagged_files)'; \
-	case "$(srcdir)" in \
-	  [\\/]* | ?:[\\/]*) sdir="$(srcdir)" ;; \
-	  *) sdir=$(subdir)/$(srcdir) ;; \
-	esac; \
-	for i in $$list; do \
-	  if test -f "$$i"; then \
-	    echo "$(subdir)/$$i"; \
-	  else \
-	    echo "$$sdir/$$i"; \
-	  fi; \
-	done >> $(top_builddir)/cscope.files
 
 distclean-tags:
 	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
-	-rm -f cscope.out cscope.in.out cscope.po.out cscope.files
 
 distdir: $(DISTFILES)
 	$(am__remove_distdir)
@@ -623,42 +493,36 @@ distdir: $(DISTFILES)
 	|| chmod -R a+r "$(distdir)"
 dist-gzip: distdir
 	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 
 dist-bzip2: distdir
-	tardir=$(distdir) && $(am__tar) | BZIP2=$${BZIP2--9} bzip2 -c >$(distdir).tar.bz2
-	$(am__post_remove_distdir)
+	tardir=$(distdir) && $(am__tar) | bzip2 -9 -c >$(distdir).tar.bz2
+	$(am__remove_distdir)
 
-dist-lzip: distdir
-	tardir=$(distdir) && $(am__tar) | lzip -c $${LZIP_OPT--9} >$(distdir).tar.lz
-	$(am__post_remove_distdir)
+dist-lzma: distdir
+	tardir=$(distdir) && $(am__tar) | lzma -9 -c >$(distdir).tar.lzma
+	$(am__remove_distdir)
 
 dist-xz: distdir
-	tardir=$(distdir) && $(am__tar) | XZ_OPT=$${XZ_OPT--e} xz -c >$(distdir).tar.xz
-	$(am__post_remove_distdir)
+	tardir=$(distdir) && $(am__tar) | xz -c >$(distdir).tar.xz
+	$(am__remove_distdir)
 
 dist-tarZ: distdir
-	@echo WARNING: "Support for distribution archives compressed with" \
-		       "legacy program 'compress' is deprecated." >&2
-	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	tardir=$(distdir) && $(am__tar) | compress -c >$(distdir).tar.Z
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 
 dist-shar: distdir
-	@echo WARNING: "Support for shar distribution archives is" \
-	               "deprecated." >&2
-	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	shar $(distdir) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).shar.gz
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 
 dist-zip: distdir
 	-rm -f $(distdir).zip
 	zip -rq $(distdir).zip $(distdir)
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 
-dist dist-all:
-	$(MAKE) $(AM_MAKEFLAGS) $(DIST_TARGETS) am__post_remove_distdir='@:'
-	$(am__post_remove_distdir)
+dist dist-all: distdir
+	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
+	$(am__remove_distdir)
 
 # This target untars the dist file and tries a VPATH configuration.  Then
 # it guarantees that the distribution is self-contained by making another
@@ -669,8 +533,8 @@ distcheck: dist
 	  GZIP=$(GZIP_ENV) gzip -dc $(distdir).tar.gz | $(am__untar) ;;\
 	*.tar.bz2*) \
 	  bzip2 -dc $(distdir).tar.bz2 | $(am__untar) ;;\
-	*.tar.lz*) \
-	  lzip -dc $(distdir).tar.lz | $(am__untar) ;;\
+	*.tar.lzma*) \
+	  lzma -dc $(distdir).tar.lzma | $(am__untar) ;;\
 	*.tar.xz*) \
 	  xz -dc $(distdir).tar.xz | $(am__untar) ;;\
 	*.tar.Z*) \
@@ -680,19 +544,17 @@ distcheck: dist
 	*.zip*) \
 	  unzip $(distdir).zip ;;\
 	esac
-	chmod -R a-w $(distdir)
-	chmod u+w $(distdir)
-	mkdir $(distdir)/_build $(distdir)/_build/sub $(distdir)/_inst
+	chmod -R a-w $(distdir); chmod u+w $(distdir)
+	mkdir $(distdir)/_build
+	mkdir $(distdir)/_inst
 	chmod a-w $(distdir)
 	test -d $(distdir)/_build || exit 0; \
 	dc_install_base=`$(am__cd) $(distdir)/_inst && pwd | sed -e 's,^[^:\\/]:[\\/],/,'` \
 	  && dc_destdir="$${TMPDIR-/tmp}/am-dc-$$$$/" \
 	  && am__cwd=`pwd` \
-	  && $(am__cd) $(distdir)/_build/sub \
-	  && ../../configure \
-	    $(AM_DISTCHECK_CONFIGURE_FLAGS) \
+	  && $(am__cd) $(distdir)/_build \
+	  && ../configure --srcdir=.. --prefix="$$dc_install_base" \
 	    $(DISTCHECK_CONFIGURE_FLAGS) \
-	    --srcdir=../.. --prefix="$$dc_install_base" \
 	  && $(MAKE) $(AM_MAKEFLAGS) \
 	  && $(MAKE) $(AM_MAKEFLAGS) dvi \
 	  && $(MAKE) $(AM_MAKEFLAGS) check \
@@ -715,21 +577,13 @@ distcheck: dist
 	  && $(MAKE) $(AM_MAKEFLAGS) distcleancheck \
 	  && cd "$$am__cwd" \
 	  || exit 1
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 	@(echo "$(distdir) archives ready for distribution: "; \
 	  list='$(DIST_ARCHIVES)'; for i in $$list; do echo $$i; done) | \
 	  sed -e 1h -e 1s/./=/g -e 1p -e 1x -e '$$p' -e '$$x'
 distuninstallcheck:
-	@test -n '$(distuninstallcheck_dir)' || { \
-	  echo 'ERROR: trying to run $@ with an empty' \
-	       '$$(distuninstallcheck_dir)' >&2; \
-	  exit 1; \
-	}; \
-	$(am__cd) '$(distuninstallcheck_dir)' || { \
-	  echo 'ERROR: cannot chdir into $(distuninstallcheck_dir)' >&2; \
-	  exit 1; \
-	}; \
-	test `$(am__distuninstallcheck_listfiles) | wc -l` -eq 0 \
+	@$(am__cd) '$(distuninstallcheck_dir)' \
+	&& test `$(distuninstallcheck_listfiles) | wc -l` -le 1 \
 	   || { echo "ERROR: files left after uninstall:" ; \
 	        if test -n "$(DESTDIR)"; then \
 	          echo "  (check DESTDIR support)"; \
@@ -762,15 +616,10 @@ install-am: all-am
 
 installcheck: installcheck-am
 install-strip:
-	if test -z '$(STRIP)'; then \
-	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
-	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
-	      install; \
-	else \
-	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
-	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
-	    "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'" install; \
-	fi
+	$(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
+	  install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
+	  `test -z '$(STRIP)' || \
+	    echo "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'"` install
 mostlyclean-generic:
 
 clean-generic:
@@ -858,10 +707,9 @@ uninstall-am: uninstall-binPROGRAMS uninstall-includeHEADERS \
 
 .MAKE: all install-am install-strip
 
-.PHONY: CTAGS GTAGS TAGS all all-am am--refresh check check-am clean \
-	clean-binPROGRAMS clean-cscope clean-generic \
-	clean-libLIBRARIES cscope cscopelist-am ctags ctags-am dist \
-	dist-all dist-bzip2 dist-gzip dist-lzip dist-shar dist-tarZ \
+.PHONY: CTAGS GTAGS all all-am am--refresh check check-am clean \
+	clean-binPROGRAMS clean-generic clean-libLIBRARIES ctags dist \
+	dist-all dist-bzip2 dist-gzip dist-lzma dist-shar dist-tarZ \
 	dist-xz dist-zip distcheck distclean distclean-compile \
 	distclean-generic distclean-hdr distclean-tags distcleancheck \
 	distdir distuninstallcheck dvi dvi-am html html-am info \
@@ -873,12 +721,10 @@ uninstall-am: uninstall-binPROGRAMS uninstall-includeHEADERS \
 	install-ps install-ps-am install-strip installcheck \
 	installcheck-am installdirs maintainer-clean \
 	maintainer-clean-generic mostlyclean mostlyclean-compile \
-	mostlyclean-generic pdf pdf-am ps ps-am tags tags-am uninstall \
+	mostlyclean-generic pdf pdf-am ps ps-am tags uninstall \
 	uninstall-am uninstall-binPROGRAMS uninstall-includeHEADERS \
 	uninstall-libLIBRARIES
 
-.PRECIOUS: Makefile
-
 
 # Tell versions [3.59,3.63) of GNU make to not export all variables.
 # Otherwise a system limit (for SysV at least) may be exceeded.
diff --git a/aclocal.m4 b/aclocal.m4
index 34cdc48d..bcc213b4 100644
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -1,7 +1,7 @@
-# generated automatically by aclocal 1.15 -*- Autoconf -*-
-
-# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+# generated automatically by aclocal 1.11.1 -*- Autoconf -*-
 
+# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
+# 2005, 2006, 2007, 2008, 2009  Free Software Foundation, Inc.
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
@@ -11,16 +11,15 @@
 # even the implied warranty of MERCHANTABILITY or FITNESS FOR A
 # PARTICULAR PURPOSE.
 
-m4_ifndef([AC_CONFIG_MACRO_DIRS], [m4_defun([_AM_CONFIG_MACRO_DIRS], [])m4_defun([AC_CONFIG_MACRO_DIRS], [_AM_CONFIG_MACRO_DIRS($@)])])
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
-m4_if(m4_defn([AC_AUTOCONF_VERSION]), [2.69],,
-[m4_warning([this file was generated for autoconf 2.69.
+m4_if(m4_defn([AC_AUTOCONF_VERSION]), [2.63],,
+[m4_warning([this file was generated for autoconf 2.63.
 You have another version of autoconf.  It may work, but is not guaranteed to.
 If you have problems, you may need to regenerate the build system entirely.
-To do so, use the procedure documented by the package, typically 'autoreconf'.])])
+To do so, use the procedure documented by the package, typically `autoreconf'.])])
 
-# Copyright (C) 2002-2014 Free Software Foundation, Inc.
+# Copyright (C) 2002, 2003, 2005, 2006, 2007, 2008  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -32,10 +31,10 @@ To do so, use the procedure documented by the package, typically 'autoreconf'.])
 # generated from the m4 files accompanying Automake X.Y.
 # (This private macro should not be called outside this file.)
 AC_DEFUN([AM_AUTOMAKE_VERSION],
-[am__api_version='1.15'
+[am__api_version='1.11'
 dnl Some users find AM_AUTOMAKE_VERSION and mistake it for a way to
 dnl require some minimum version.  Point them to the right macro.
-m4_if([$1], [1.15], [],
+m4_if([$1], [1.11.1], [],
       [AC_FATAL([Do not call $0, use AM_INIT_AUTOMAKE([$1]).])])dnl
 ])
 
@@ -51,22 +50,22 @@ m4_define([_AM_AUTOCONF_VERSION], [])
 # Call AM_AUTOMAKE_VERSION and AM_AUTOMAKE_VERSION so they can be traced.
 # This function is AC_REQUIREd by AM_INIT_AUTOMAKE.
 AC_DEFUN([AM_SET_CURRENT_AUTOMAKE_VERSION],
-[AM_AUTOMAKE_VERSION([1.15])dnl
+[AM_AUTOMAKE_VERSION([1.11.1])dnl
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
 _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 
 # AM_AUX_DIR_EXPAND                                         -*- Autoconf -*-
 
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2001, 2003, 2005  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
 # For projects using AC_CONFIG_AUX_DIR([foo]), Autoconf sets
-# $ac_aux_dir to '$srcdir/foo'.  In other projects, it is set to
-# '$srcdir', '$srcdir/..', or '$srcdir/../..'.
+# $ac_aux_dir to `$srcdir/foo'.  In other projects, it is set to
+# `$srcdir', `$srcdir/..', or `$srcdir/../..'.
 #
 # Of course, Automake must honor this variable whenever it calls a
 # tool from the auxiliary directory.  The problem is that $srcdir (and
@@ -85,7 +84,7 @@ _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 #
 # The reason of the latter failure is that $top_srcdir and $ac_aux_dir
 # are both prefixed by $srcdir.  In an in-source build this is usually
-# harmless because $srcdir is '.', but things will broke when you
+# harmless because $srcdir is `.', but things will broke when you
 # start a VPATH build or use an absolute $srcdir.
 #
 # So we could use something similar to $top_srcdir/$ac_aux_dir/missing,
@@ -103,26 +102,30 @@ _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 # configured tree to be moved without reconfiguration.
 
 AC_DEFUN([AM_AUX_DIR_EXPAND],
-[AC_REQUIRE([AC_CONFIG_AUX_DIR_DEFAULT])dnl
-# Expand $ac_aux_dir to an absolute path.
-am_aux_dir=`cd "$ac_aux_dir" && pwd`
+[dnl Rely on autoconf to set up CDPATH properly.
+AC_PREREQ([2.50])dnl
+# expand $ac_aux_dir to an absolute path
+am_aux_dir=`cd $ac_aux_dir && pwd`
 ])
 
 # AM_CONDITIONAL                                            -*- Autoconf -*-
 
-# Copyright (C) 1997-2014 Free Software Foundation, Inc.
+# Copyright (C) 1997, 2000, 2001, 2003, 2004, 2005, 2006, 2008
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 9
+
 # AM_CONDITIONAL(NAME, SHELL-CONDITION)
 # -------------------------------------
 # Define a conditional.
 AC_DEFUN([AM_CONDITIONAL],
-[AC_PREREQ([2.52])dnl
- m4_if([$1], [TRUE],  [AC_FATAL([$0: invalid condition: $1])],
-       [$1], [FALSE], [AC_FATAL([$0: invalid condition: $1])])dnl
+[AC_PREREQ(2.52)dnl
+ ifelse([$1], [TRUE],  [AC_FATAL([$0: invalid condition: $1])],
+	[$1], [FALSE], [AC_FATAL([$0: invalid condition: $1])])dnl
 AC_SUBST([$1_TRUE])dnl
 AC_SUBST([$1_FALSE])dnl
 _AM_SUBST_NOTMAKE([$1_TRUE])dnl
@@ -141,14 +144,16 @@ AC_CONFIG_COMMANDS_PRE(
 Usually this means the macro was only invoked conditionally.]])
 fi])])
 
-# Copyright (C) 1999-2014 Free Software Foundation, Inc.
+# Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2009
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 10
 
-# There are a few dirty hacks below to avoid letting 'AC_PROG_CC' be
+# There are a few dirty hacks below to avoid letting `AC_PROG_CC' be
 # written in clear, in which case automake, when reading aclocal.m4,
 # will think it sees a *use*, and therefore will trigger all it's
 # C support machinery.  Also note that it means that autoscan, seeing
@@ -158,7 +163,7 @@ fi])])
 # _AM_DEPENDENCIES(NAME)
 # ----------------------
 # See how the compiler implements dependency checking.
-# NAME is "CC", "CXX", "OBJC", "OBJCXX", "UPC", or "GJC".
+# NAME is "CC", "CXX", "GCJ", or "OBJC".
 # We try a few techniques and use that to set a single cache variable.
 #
 # We don't AC_REQUIRE the corresponding AC_PROG_CC since the latter was
@@ -171,13 +176,12 @@ AC_REQUIRE([AM_OUTPUT_DEPENDENCY_COMMANDS])dnl
 AC_REQUIRE([AM_MAKE_INCLUDE])dnl
 AC_REQUIRE([AM_DEP_TRACK])dnl
 
-m4_if([$1], [CC],   [depcc="$CC"   am_compiler_list=],
-      [$1], [CXX],  [depcc="$CXX"  am_compiler_list=],
-      [$1], [OBJC], [depcc="$OBJC" am_compiler_list='gcc3 gcc'],
-      [$1], [OBJCXX], [depcc="$OBJCXX" am_compiler_list='gcc3 gcc'],
-      [$1], [UPC],  [depcc="$UPC"  am_compiler_list=],
-      [$1], [GCJ],  [depcc="$GCJ"  am_compiler_list='gcc3 gcc'],
-                    [depcc="$$1"   am_compiler_list=])
+ifelse([$1], CC,   [depcc="$CC"   am_compiler_list=],
+       [$1], CXX,  [depcc="$CXX"  am_compiler_list=],
+       [$1], OBJC, [depcc="$OBJC" am_compiler_list='gcc3 gcc'],
+       [$1], UPC,  [depcc="$UPC"  am_compiler_list=],
+       [$1], GCJ,  [depcc="$GCJ"  am_compiler_list='gcc3 gcc'],
+                   [depcc="$$1"   am_compiler_list=])
 
 AC_CACHE_CHECK([dependency style of $depcc],
                [am_cv_$1_dependencies_compiler_type],
@@ -185,9 +189,8 @@ AC_CACHE_CHECK([dependency style of $depcc],
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named 'D' -- because '-MD' means "put the output
-  # in D".
-  rm -rf conftest.dir
+  # making a dummy file named `D' -- because `-MD' means `put the output
+  # in D'.
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -226,16 +229,16 @@ AC_CACHE_CHECK([dependency style of $depcc],
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
-      # Solaris 10 /bin/sh.
-      echo '/* dummy */' > sub/conftst$i.h
+      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
+      # Solaris 8's {/usr,}/bin/sh.
+      touch sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with '-c' and '-o' for the sake of the "dashmstdout"
+    # We check with `-c' and `-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle '-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs.
+    # handle `-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -244,16 +247,16 @@ AC_CACHE_CHECK([dependency style of $depcc],
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # After this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested.
+      # after this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
-      # This compiler won't grok '-c -o', but also, the minuso test has
+    msvisualcpp | msvcmsys)
+      # This compiler won't grok `-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -301,7 +304,7 @@ AM_CONDITIONAL([am__fastdep$1], [
 # AM_SET_DEPDIR
 # -------------
 # Choose a directory name for dependency files.
-# This macro is AC_REQUIREd in _AM_DEPENDENCIES.
+# This macro is AC_REQUIREd in _AM_DEPENDENCIES
 AC_DEFUN([AM_SET_DEPDIR],
 [AC_REQUIRE([AM_SET_LEADING_DOT])dnl
 AC_SUBST([DEPDIR], ["${am__leading_dot}deps"])dnl
@@ -311,39 +314,34 @@ AC_SUBST([DEPDIR], ["${am__leading_dot}deps"])dnl
 # AM_DEP_TRACK
 # ------------
 AC_DEFUN([AM_DEP_TRACK],
-[AC_ARG_ENABLE([dependency-tracking], [dnl
-AS_HELP_STRING(
-  [--enable-dependency-tracking],
-  [do not reject slow dependency extractors])
-AS_HELP_STRING(
-  [--disable-dependency-tracking],
-  [speeds up one-time build])])
+[AC_ARG_ENABLE(dependency-tracking,
+[  --disable-dependency-tracking  speeds up one-time build
+  --enable-dependency-tracking   do not reject slow dependency extractors])
 if test "x$enable_dependency_tracking" != xno; then
   am_depcomp="$ac_aux_dir/depcomp"
   AMDEPBACKSLASH='\'
-  am__nodep='_no'
 fi
 AM_CONDITIONAL([AMDEP], [test "x$enable_dependency_tracking" != xno])
 AC_SUBST([AMDEPBACKSLASH])dnl
 _AM_SUBST_NOTMAKE([AMDEPBACKSLASH])dnl
-AC_SUBST([am__nodep])dnl
-_AM_SUBST_NOTMAKE([am__nodep])dnl
 ])
 
 # Generate code to set up dependency tracking.              -*- Autoconf -*-
 
-# Copyright (C) 1999-2014 Free Software Foundation, Inc.
+# Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2008
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+#serial 5
 
 # _AM_OUTPUT_DEPENDENCY_COMMANDS
 # ------------------------------
 AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
 [{
-  # Older Autoconf quotes --file arguments for eval, but not when files
+  # Autoconf 2.62 quotes --file arguments for eval, but not when files
   # are listed without --file.  Let's play safe and only enable the eval
   # if we detect the quoting.
   case $CONFIG_FILES in
@@ -356,7 +354,7 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
     # Strip MF so we end up with the name of the file.
     mf=`echo "$mf" | sed -e 's/:.*$//'`
     # Check whether this is an Automake generated Makefile or not.
-    # We used to match only the files named 'Makefile.in', but
+    # We used to match only the files named `Makefile.in', but
     # some people rename them; so instead we look at the file content.
     # Grep'ing the first line is not enough: some people post-process
     # each Makefile.in and add a new line on top of each file to say so.
@@ -368,19 +366,21 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
       continue
     fi
     # Extract the definition of DEPDIR, am__include, and am__quote
-    # from the Makefile without running 'make'.
+    # from the Makefile without running `make'.
     DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
     test -z "$DEPDIR" && continue
     am__include=`sed -n 's/^am__include = //p' < "$mf"`
-    test -z "$am__include" && continue
+    test -z "am__include" && continue
     am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
+    # When using ansi2knr, U may be empty or an underscore; expand it
+    U=`sed -n 's/^U = //p' < "$mf"`
     # Find all dependency output files, they are included files with
     # $(DEPDIR) in their names.  We invoke sed twice because it is the
     # simplest approach to changing $(DEPDIR) to its actual value in the
     # expansion.
     for file in `sed -n "
       s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
-	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
+	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g' -e 's/\$U/'"$U"'/g'`; do
       # Make sure the directory exists.
       test -f "$dirpart/$file" && continue
       fdir=`AS_DIRNAME(["$file"])`
@@ -398,7 +398,7 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
 # This macro should only be invoked once -- use via AC_REQUIRE.
 #
 # This code is only required when automatic dependency tracking
-# is enabled.  FIXME.  This creates each '.P' file that we will
+# is enabled.  FIXME.  This creates each `.P' file that we will
 # need in order to bootstrap the dependency handling code.
 AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 [AC_CONFIG_COMMANDS([depfiles],
@@ -408,21 +408,18 @@ AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 
 # Do all the work for Automake.                             -*- Autoconf -*-
 
-# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
+# 2005, 2006, 2008, 2009 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 16
+
 # This macro actually does too much.  Some checks are only needed if
 # your package does certain things.  But this isn't really a big deal.
 
-dnl Redefine AC_PROG_CC to automatically invoke _AM_PROG_CC_C_O.
-m4_define([AC_PROG_CC],
-m4_defn([AC_PROG_CC])
-[_AM_PROG_CC_C_O
-])
-
 # AM_INIT_AUTOMAKE(PACKAGE, VERSION, [NO-DEFINE])
 # AM_INIT_AUTOMAKE([OPTIONS])
 # -----------------------------------------------
@@ -435,7 +432,7 @@ m4_defn([AC_PROG_CC])
 # arguments mandatory, and then we can depend on a new Autoconf
 # release and drop the old call support.
 AC_DEFUN([AM_INIT_AUTOMAKE],
-[AC_PREREQ([2.65])dnl
+[AC_PREREQ([2.62])dnl
 dnl Autoconf wants to disallow AM_ names.  We explicitly allow
 dnl the ones we care about.
 m4_pattern_allow([^AM_[A-Z]+FLAGS$])dnl
@@ -464,42 +461,33 @@ AC_SUBST([CYGPATH_W])
 # Define the identity of the package.
 dnl Distinguish between old-style and new-style calls.
 m4_ifval([$2],
-[AC_DIAGNOSE([obsolete],
-             [$0: two- and three-arguments forms are deprecated.])
-m4_ifval([$3], [_AM_SET_OPTION([no-define])])dnl
+[m4_ifval([$3], [_AM_SET_OPTION([no-define])])dnl
  AC_SUBST([PACKAGE], [$1])dnl
  AC_SUBST([VERSION], [$2])],
 [_AM_SET_OPTIONS([$1])dnl
 dnl Diagnose old-style AC_INIT with new-style AM_AUTOMAKE_INIT.
-m4_if(
-  m4_ifdef([AC_PACKAGE_NAME], [ok]):m4_ifdef([AC_PACKAGE_VERSION], [ok]),
-  [ok:ok],,
+m4_if(m4_ifdef([AC_PACKAGE_NAME], 1)m4_ifdef([AC_PACKAGE_VERSION], 1), 11,,
   [m4_fatal([AC_INIT should be called with package and version arguments])])dnl
  AC_SUBST([PACKAGE], ['AC_PACKAGE_TARNAME'])dnl
  AC_SUBST([VERSION], ['AC_PACKAGE_VERSION'])])dnl
 
 _AM_IF_OPTION([no-define],,
-[AC_DEFINE_UNQUOTED([PACKAGE], ["$PACKAGE"], [Name of package])
- AC_DEFINE_UNQUOTED([VERSION], ["$VERSION"], [Version number of package])])dnl
+[AC_DEFINE_UNQUOTED(PACKAGE, "$PACKAGE", [Name of package])
+ AC_DEFINE_UNQUOTED(VERSION, "$VERSION", [Version number of package])])dnl
 
 # Some tools Automake needs.
 AC_REQUIRE([AM_SANITY_CHECK])dnl
 AC_REQUIRE([AC_ARG_PROGRAM])dnl
-AM_MISSING_PROG([ACLOCAL], [aclocal-${am__api_version}])
-AM_MISSING_PROG([AUTOCONF], [autoconf])
-AM_MISSING_PROG([AUTOMAKE], [automake-${am__api_version}])
-AM_MISSING_PROG([AUTOHEADER], [autoheader])
-AM_MISSING_PROG([MAKEINFO], [makeinfo])
+AM_MISSING_PROG(ACLOCAL, aclocal-${am__api_version})
+AM_MISSING_PROG(AUTOCONF, autoconf)
+AM_MISSING_PROG(AUTOMAKE, automake-${am__api_version})
+AM_MISSING_PROG(AUTOHEADER, autoheader)
+AM_MISSING_PROG(MAKEINFO, makeinfo)
 AC_REQUIRE([AM_PROG_INSTALL_SH])dnl
 AC_REQUIRE([AM_PROG_INSTALL_STRIP])dnl
-AC_REQUIRE([AC_PROG_MKDIR_P])dnl
-# For better backward compatibility.  To be removed once Automake 1.9.x
-# dies out for good.  For more background, see:
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
-AC_SUBST([mkdir_p], ['$(MKDIR_P)'])
-# We need awk for the "check" target (and possibly the TAP driver).  The
-# system "awk" is bad on some platforms.
+AC_REQUIRE([AM_PROG_MKDIR_P])dnl
+# We need awk for the "check" target.  The system "awk" is bad on
+# some platforms.
 AC_REQUIRE([AC_PROG_AWK])dnl
 AC_REQUIRE([AC_PROG_MAKE_SET])dnl
 AC_REQUIRE([AM_SET_LEADING_DOT])dnl
@@ -508,82 +496,34 @@ _AM_IF_OPTION([tar-ustar], [_AM_PROG_TAR([ustar])],
 			     [_AM_PROG_TAR([v7])])])
 _AM_IF_OPTION([no-dependencies],,
 [AC_PROVIDE_IFELSE([AC_PROG_CC],
-		  [_AM_DEPENDENCIES([CC])],
-		  [m4_define([AC_PROG_CC],
-			     m4_defn([AC_PROG_CC])[_AM_DEPENDENCIES([CC])])])dnl
+		  [_AM_DEPENDENCIES(CC)],
+		  [define([AC_PROG_CC],
+			  defn([AC_PROG_CC])[_AM_DEPENDENCIES(CC)])])dnl
 AC_PROVIDE_IFELSE([AC_PROG_CXX],
-		  [_AM_DEPENDENCIES([CXX])],
-		  [m4_define([AC_PROG_CXX],
-			     m4_defn([AC_PROG_CXX])[_AM_DEPENDENCIES([CXX])])])dnl
+		  [_AM_DEPENDENCIES(CXX)],
+		  [define([AC_PROG_CXX],
+			  defn([AC_PROG_CXX])[_AM_DEPENDENCIES(CXX)])])dnl
 AC_PROVIDE_IFELSE([AC_PROG_OBJC],
-		  [_AM_DEPENDENCIES([OBJC])],
-		  [m4_define([AC_PROG_OBJC],
-			     m4_defn([AC_PROG_OBJC])[_AM_DEPENDENCIES([OBJC])])])dnl
-AC_PROVIDE_IFELSE([AC_PROG_OBJCXX],
-		  [_AM_DEPENDENCIES([OBJCXX])],
-		  [m4_define([AC_PROG_OBJCXX],
-			     m4_defn([AC_PROG_OBJCXX])[_AM_DEPENDENCIES([OBJCXX])])])dnl
+		  [_AM_DEPENDENCIES(OBJC)],
+		  [define([AC_PROG_OBJC],
+			  defn([AC_PROG_OBJC])[_AM_DEPENDENCIES(OBJC)])])dnl
 ])
-AC_REQUIRE([AM_SILENT_RULES])dnl
-dnl The testsuite driver may need to know about EXEEXT, so add the
-dnl 'am__EXEEXT' conditional if _AM_COMPILER_EXEEXT was seen.  This
-dnl macro is hooked onto _AC_COMPILER_EXEEXT early, see below.
+_AM_IF_OPTION([silent-rules], [AC_REQUIRE([AM_SILENT_RULES])])dnl
+dnl The `parallel-tests' driver may need to know about EXEEXT, so add the
+dnl `am__EXEEXT' conditional if _AM_COMPILER_EXEEXT was seen.  This macro
+dnl is hooked onto _AC_COMPILER_EXEEXT early, see below.
 AC_CONFIG_COMMANDS_PRE(dnl
 [m4_provide_if([_AM_COMPILER_EXEEXT],
   [AM_CONDITIONAL([am__EXEEXT], [test -n "$EXEEXT"])])])dnl
-
-# POSIX will say in a future version that running "rm -f" with no argument
-# is OK; and we want to be able to make that assumption in our Makefile
-# recipes.  So use an aggressive probe to check that the usage we want is
-# actually supported "in the wild" to an acceptable degree.
-# See automake bug#10828.
-# To make any issue more visible, cause the running configure to be aborted
-# by default if the 'rm' program in use doesn't match our expectations; the
-# user can still override this though.
-if rm -f && rm -fr && rm -rf; then : OK; else
-  cat >&2 <<'END'
-Oops!
-
-Your 'rm' program seems unable to run without file operands specified
-on the command line, even when the '-f' option is present.  This is contrary
-to the behaviour of most rm programs out there, and not conforming with
-the upcoming POSIX standard: <http://austingroupbugs.net/view.php?id=542>
-
-Please tell bug-automake@gnu.org about your system, including the value
-of your $PATH and any error possibly output before this message.  This
-can help us improve future automake versions.
-
-END
-  if test x"$ACCEPT_INFERIOR_RM_PROGRAM" = x"yes"; then
-    echo 'Configuration will proceed anyway, since you have set the' >&2
-    echo 'ACCEPT_INFERIOR_RM_PROGRAM variable to "yes"' >&2
-    echo >&2
-  else
-    cat >&2 <<'END'
-Aborting the configuration process, to ensure you take notice of the issue.
-
-You can download and install GNU coreutils to get an 'rm' implementation
-that behaves properly: <http://www.gnu.org/software/coreutils/>.
-
-If you want to complete the configuration process using your problematic
-'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
-to "yes", and re-run configure.
-
-END
-    AC_MSG_ERROR([Your 'rm' program is bad, sorry.])
-  fi
-fi
-dnl The trailing newline in this macro's definition is deliberate, for
-dnl backward compatibility and to allow trailing 'dnl'-style comments
-dnl after the AM_INIT_AUTOMAKE invocation. See automake bug#16841.
 ])
 
-dnl Hook into '_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
+dnl Hook into `_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
 dnl add the conditional right here, as _AC_COMPILER_EXEEXT may be further
 dnl mangled by Autoconf and run in a shell conditional statement.
 m4_define([_AC_COMPILER_EXEEXT],
 m4_defn([_AC_COMPILER_EXEEXT])[m4_provide([_AM_COMPILER_EXEEXT])])
 
+
 # When config.status generates a header, we must update the stamp-h file.
 # This file resides in the same directory as the config header
 # that is generated.  The stamp files are numbered to have different names.
@@ -605,7 +545,7 @@ for _am_header in $config_headers :; do
 done
 echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_count])
 
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2001, 2003, 2005, 2008  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -616,7 +556,7 @@ echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_co
 # Define $install_sh.
 AC_DEFUN([AM_PROG_INSTALL_SH],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
-if test x"${install_sh+set}" != xset; then
+if test x"${install_sh}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -624,14 +564,16 @@ if test x"${install_sh+set}" != xset; then
     install_sh="\${SHELL} $am_aux_dir/install-sh"
   esac
 fi
-AC_SUBST([install_sh])])
+AC_SUBST(install_sh)])
 
-# Copyright (C) 2003-2014 Free Software Foundation, Inc.
+# Copyright (C) 2003, 2005  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 2
+
 # Check whether the underlying file-system supports filenames
 # with a leading dot.  For instance MS-DOS doesn't.
 AC_DEFUN([AM_SET_LEADING_DOT],
@@ -647,12 +589,14 @@ AC_SUBST([am__leading_dot])])
 
 # Check to see how 'make' treats includes.	            -*- Autoconf -*-
 
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2001, 2002, 2003, 2005, 2009  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 4
+
 # AM_MAKE_INCLUDE()
 # -----------------
 # Check to see how make treats includes.
@@ -670,7 +614,7 @@ am__quote=
 _am_result=none
 # First try GNU make style include.
 echo "include confinc" > confmf
-# Ignore all kinds of additional output from 'make'.
+# Ignore all kinds of additional output from `make'.
 case `$am_make -s -f confmf 2> /dev/null` in #(
 *the\ am__doit\ target*)
   am__include=include
@@ -697,12 +641,15 @@ rm -f confinc confmf
 
 # Fake the existence of programs that GNU maintainers use.  -*- Autoconf -*-
 
-# Copyright (C) 1997-2014 Free Software Foundation, Inc.
+# Copyright (C) 1997, 1999, 2000, 2001, 2003, 2004, 2005, 2008
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 6
+
 # AM_MISSING_PROG(NAME, PROGRAM)
 # ------------------------------
 AC_DEFUN([AM_MISSING_PROG],
@@ -710,10 +657,11 @@ AC_DEFUN([AM_MISSING_PROG],
 $1=${$1-"${am_missing_run}$2"}
 AC_SUBST($1)])
 
+
 # AM_MISSING_HAS_RUN
 # ------------------
-# Define MISSING if not defined so far and test if it is modern enough.
-# If it is, set am_missing_run to use it, otherwise, to nothing.
+# Define MISSING if not defined so far and test if it supports --run.
+# If it does, set am_missing_run to use it, otherwise, to nothing.
 AC_DEFUN([AM_MISSING_HAS_RUN],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
 AC_REQUIRE_AUX_FILE([missing])dnl
@@ -726,35 +674,63 @@ if test x"${MISSING+set}" != xset; then
   esac
 fi
 # Use eval to expand $SHELL
-if eval "$MISSING --is-lightweight"; then
-  am_missing_run="$MISSING "
+if eval "$MISSING --run true"; then
+  am_missing_run="$MISSING --run "
 else
   am_missing_run=
-  AC_MSG_WARN(['missing' script is too old or missing])
+  AC_MSG_WARN([`missing' script is too old or missing])
 fi
 ])
 
-# Helper functions for option handling.                     -*- Autoconf -*-
-
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2003, 2004, 2005, 2006  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# AM_PROG_MKDIR_P
+# ---------------
+# Check for `mkdir -p'.
+AC_DEFUN([AM_PROG_MKDIR_P],
+[AC_PREREQ([2.60])dnl
+AC_REQUIRE([AC_PROG_MKDIR_P])dnl
+dnl Automake 1.8 to 1.9.6 used to define mkdir_p.  We now use MKDIR_P,
+dnl while keeping a definition of mkdir_p for backward compatibility.
+dnl @MKDIR_P@ is magic: AC_OUTPUT adjusts its value for each Makefile.
+dnl However we cannot define mkdir_p as $(MKDIR_P) for the sake of
+dnl Makefile.ins that do not define MKDIR_P, so we do our own
+dnl adjustment using top_builddir (which is defined more often than
+dnl MKDIR_P).
+AC_SUBST([mkdir_p], ["$MKDIR_P"])dnl
+case $mkdir_p in
+  [[\\/$]]* | ?:[[\\/]]*) ;;
+  */*) mkdir_p="\$(top_builddir)/$mkdir_p" ;;
+esac
+])
+
+# Helper functions for option handling.                     -*- Autoconf -*-
+
+# Copyright (C) 2001, 2002, 2003, 2005, 2008  Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# serial 4
+
 # _AM_MANGLE_OPTION(NAME)
 # -----------------------
 AC_DEFUN([_AM_MANGLE_OPTION],
 [[_AM_OPTION_]m4_bpatsubst($1, [[^a-zA-Z0-9_]], [_])])
 
 # _AM_SET_OPTION(NAME)
-# --------------------
+# ------------------------------
 # Set option NAME.  Presently that only means defining a flag for this option.
 AC_DEFUN([_AM_SET_OPTION],
-[m4_define(_AM_MANGLE_OPTION([$1]), [1])])
+[m4_define(_AM_MANGLE_OPTION([$1]), 1)])
 
 # _AM_SET_OPTIONS(OPTIONS)
-# ------------------------
+# ----------------------------------
 # OPTIONS is a space-separated list of Automake options.
 AC_DEFUN([_AM_SET_OPTIONS],
 [m4_foreach_w([_AM_Option], [$1], [_AM_SET_OPTION(_AM_Option)])])
@@ -765,82 +741,24 @@ AC_DEFUN([_AM_SET_OPTIONS],
 AC_DEFUN([_AM_IF_OPTION],
 [m4_ifset(_AM_MANGLE_OPTION([$1]), [$2], [$3])])
 
-# Copyright (C) 1999-2014 Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# _AM_PROG_CC_C_O
-# ---------------
-# Like AC_PROG_CC_C_O, but changed for automake.  We rewrite AC_PROG_CC
-# to automatically call this.
-AC_DEFUN([_AM_PROG_CC_C_O],
-[AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
-AC_REQUIRE_AUX_FILE([compile])dnl
-AC_LANG_PUSH([C])dnl
-AC_CACHE_CHECK(
-  [whether $CC understands -c and -o together],
-  [am_cv_prog_cc_c_o],
-  [AC_LANG_CONFTEST([AC_LANG_PROGRAM([])])
-  # Make sure it works both with $CC and with simple cc.
-  # Following AC_PROG_CC_C_O, we do the test twice because some
-  # compilers refuse to overwrite an existing .o file with -o,
-  # though they will create one.
-  am_cv_prog_cc_c_o=yes
-  for am_i in 1 2; do
-    if AM_RUN_LOG([$CC -c conftest.$ac_ext -o conftest2.$ac_objext]) \
-         && test -f conftest2.$ac_objext; then
-      : OK
-    else
-      am_cv_prog_cc_c_o=no
-      break
-    fi
-  done
-  rm -f core conftest*
-  unset am_i])
-if test "$am_cv_prog_cc_c_o" != yes; then
-   # Losing compiler, so override with the script.
-   # FIXME: It is wrong to rewrite CC.
-   # But if we don't then we get into trouble of one sort or another.
-   # A longer-term fix would be to have automake use am__CC in this case,
-   # and then we could set am__CC="\$(top_srcdir)/compile \$(CC)"
-   CC="$am_aux_dir/compile $CC"
-fi
-AC_LANG_POP([C])])
-
-# For backward compatibility.
-AC_DEFUN_ONCE([AM_PROG_CC_C_O], [AC_REQUIRE([AC_PROG_CC])])
-
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# AM_RUN_LOG(COMMAND)
-# -------------------
-# Run COMMAND, save the exit status in ac_status, and log it.
-# (This has been adapted from Autoconf's _AC_RUN_LOG macro.)
-AC_DEFUN([AM_RUN_LOG],
-[{ echo "$as_me:$LINENO: $1" >&AS_MESSAGE_LOG_FD
-   ($1) >&AS_MESSAGE_LOG_FD 2>&AS_MESSAGE_LOG_FD
-   ac_status=$?
-   echo "$as_me:$LINENO: \$? = $ac_status" >&AS_MESSAGE_LOG_FD
-   (exit $ac_status); }])
-
 # Check to make sure that the build environment is sane.    -*- Autoconf -*-
 
-# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+# Copyright (C) 1996, 1997, 2000, 2001, 2003, 2005, 2008
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 5
+
 # AM_SANITY_CHECK
 # ---------------
 AC_DEFUN([AM_SANITY_CHECK],
 [AC_MSG_CHECKING([whether build environment is sane])
+# Just in case
+sleep 1
+echo timestamp > conftest.file
 # Reject unsafe characters in $srcdir or the absolute working directory
 # name.  Accept space and tab only in the latter.
 am_lf='
@@ -851,40 +769,32 @@ case `pwd` in
 esac
 case $srcdir in
   *[[\\\"\#\$\&\'\`$am_lf\ \	]]*)
-    AC_MSG_ERROR([unsafe srcdir value: '$srcdir']);;
+    AC_MSG_ERROR([unsafe srcdir value: `$srcdir']);;
 esac
 
-# Do 'set' in a subshell so we don't clobber the current shell's
+# Do `set' in a subshell so we don't clobber the current shell's
 # arguments.  Must try -L first in case configure is actually a
 # symlink; some systems play weird games with the mod time of symlinks
 # (eg FreeBSD returns the mod time of the symlink's containing
 # directory).
 if (
-   am_has_slept=no
-   for am_try in 1 2; do
-     echo "timestamp, slept: $am_has_slept" > conftest.file
-     set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
-     if test "$[*]" = "X"; then
-	# -L didn't work.
-	set X `ls -t "$srcdir/configure" conftest.file`
-     fi
-     if test "$[*]" != "X $srcdir/configure conftest.file" \
-	&& test "$[*]" != "X conftest.file $srcdir/configure"; then
+   set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
+   if test "$[*]" = "X"; then
+      # -L didn't work.
+      set X `ls -t "$srcdir/configure" conftest.file`
+   fi
+   rm -f conftest.file
+   if test "$[*]" != "X $srcdir/configure conftest.file" \
+      && test "$[*]" != "X conftest.file $srcdir/configure"; then
+
+      # If neither matched, then we have a broken ls.  This can happen
+      # if, for instance, CONFIG_SHELL is bash and it inherits a
+      # broken ls alias from the environment.  This has actually
+      # happened.  Such a system could not be considered "sane".
+      AC_MSG_ERROR([ls -t appears to fail.  Make sure there is not a broken
+alias in your environment])
+   fi
 
-	# If neither matched, then we have a broken ls.  This can happen
-	# if, for instance, CONFIG_SHELL is bash and it inherits a
-	# broken ls alias from the environment.  This has actually
-	# happened.  Such a system could not be considered "sane".
-	AC_MSG_ERROR([ls -t appears to fail.  Make sure there is not a broken
-  alias in your environment])
-     fi
-     if test "$[2]" = conftest.file || test $am_try -eq 2; then
-       break
-     fi
-     # Just in case.
-     sleep 1
-     am_has_slept=yes
-   done
    test "$[2]" = conftest.file
    )
 then
@@ -894,85 +804,9 @@ else
    AC_MSG_ERROR([newly created file is older than distributed files!
 Check your system clock])
 fi
-AC_MSG_RESULT([yes])
-# If we didn't sleep, we still need to ensure time stamps of config.status and
-# generated files are strictly newer.
-am_sleep_pid=
-if grep 'slept: no' conftest.file >/dev/null 2>&1; then
-  ( sleep 1 ) &
-  am_sleep_pid=$!
-fi
-AC_CONFIG_COMMANDS_PRE(
-  [AC_MSG_CHECKING([that generated files are newer than configure])
-   if test -n "$am_sleep_pid"; then
-     # Hide warnings about reused PIDs.
-     wait $am_sleep_pid 2>/dev/null
-   fi
-   AC_MSG_RESULT([done])])
-rm -f conftest.file
-])
+AC_MSG_RESULT(yes)])
 
-# Copyright (C) 2009-2014 Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# AM_SILENT_RULES([DEFAULT])
-# --------------------------
-# Enable less verbose build rules; with the default set to DEFAULT
-# ("yes" being less verbose, "no" or empty being verbose).
-AC_DEFUN([AM_SILENT_RULES],
-[AC_ARG_ENABLE([silent-rules], [dnl
-AS_HELP_STRING(
-  [--enable-silent-rules],
-  [less verbose build output (undo: "make V=1")])
-AS_HELP_STRING(
-  [--disable-silent-rules],
-  [verbose build output (undo: "make V=0")])dnl
-])
-case $enable_silent_rules in @%:@ (((
-  yes) AM_DEFAULT_VERBOSITY=0;;
-   no) AM_DEFAULT_VERBOSITY=1;;
-    *) AM_DEFAULT_VERBOSITY=m4_if([$1], [yes], [0], [1]);;
-esac
-dnl
-dnl A few 'make' implementations (e.g., NonStop OS and NextStep)
-dnl do not support nested variable expansions.
-dnl See automake bug#9928 and bug#10237.
-am_make=${MAKE-make}
-AC_CACHE_CHECK([whether $am_make supports nested variables],
-   [am_cv_make_support_nested_variables],
-   [if AS_ECHO([['TRUE=$(BAR$(V))
-BAR0=false
-BAR1=true
-V=1
-am__doit:
-	@$(TRUE)
-.PHONY: am__doit']]) | $am_make -f - >/dev/null 2>&1; then
-  am_cv_make_support_nested_variables=yes
-else
-  am_cv_make_support_nested_variables=no
-fi])
-if test $am_cv_make_support_nested_variables = yes; then
-  dnl Using '$V' instead of '$(V)' breaks IRIX make.
-  AM_V='$(V)'
-  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
-else
-  AM_V=$AM_DEFAULT_VERBOSITY
-  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
-fi
-AC_SUBST([AM_V])dnl
-AM_SUBST_NOTMAKE([AM_V])dnl
-AC_SUBST([AM_DEFAULT_V])dnl
-AM_SUBST_NOTMAKE([AM_DEFAULT_V])dnl
-AC_SUBST([AM_DEFAULT_VERBOSITY])dnl
-AM_BACKSLASH='\'
-AC_SUBST([AM_BACKSLASH])dnl
-_AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
-])
-
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2001, 2003, 2005  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -980,32 +814,34 @@ _AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
 
 # AM_PROG_INSTALL_STRIP
 # ---------------------
-# One issue with vendor 'install' (even GNU) is that you can't
+# One issue with vendor `install' (even GNU) is that you can't
 # specify the program used to strip binaries.  This is especially
 # annoying in cross-compiling environments, where the build's strip
 # is unlikely to handle the host's binaries.
 # Fortunately install-sh will honor a STRIPPROG variable, so we
-# always use install-sh in "make install-strip", and initialize
+# always use install-sh in `make install-strip', and initialize
 # STRIPPROG with the value of the STRIP variable (set by the user).
 AC_DEFUN([AM_PROG_INSTALL_STRIP],
 [AC_REQUIRE([AM_PROG_INSTALL_SH])dnl
-# Installed binaries are usually stripped using 'strip' when the user
-# run "make install-strip".  However 'strip' might not be the right
+# Installed binaries are usually stripped using `strip' when the user
+# run `make install-strip'.  However `strip' might not be the right
 # tool to use in cross-compilation environments, therefore Automake
-# will honor the 'STRIP' environment variable to overrule this program.
-dnl Don't test for $cross_compiling = yes, because it might be 'maybe'.
+# will honor the `STRIP' environment variable to overrule this program.
+dnl Don't test for $cross_compiling = yes, because it might be `maybe'.
 if test "$cross_compiling" != no; then
   AC_CHECK_TOOL([STRIP], [strip], :)
 fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 AC_SUBST([INSTALL_STRIP_PROGRAM])])
 
-# Copyright (C) 2006-2014 Free Software Foundation, Inc.
+# Copyright (C) 2006, 2008  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 2
+
 # _AM_SUBST_NOTMAKE(VARIABLE)
 # ---------------------------
 # Prevent Automake from outputting VARIABLE = @VARIABLE@ in Makefile.in.
@@ -1013,22 +849,24 @@ AC_SUBST([INSTALL_STRIP_PROGRAM])])
 AC_DEFUN([_AM_SUBST_NOTMAKE])
 
 # AM_SUBST_NOTMAKE(VARIABLE)
-# --------------------------
+# ---------------------------
 # Public sister of _AM_SUBST_NOTMAKE.
 AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 
 # Check how to create a tarball.                            -*- Autoconf -*-
 
-# Copyright (C) 2004-2014 Free Software Foundation, Inc.
+# Copyright (C) 2004, 2005  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 2
+
 # _AM_PROG_TAR(FORMAT)
 # --------------------
 # Check how to create a tarball in format FORMAT.
-# FORMAT should be one of 'v7', 'ustar', or 'pax'.
+# FORMAT should be one of `v7', `ustar', or `pax'.
 #
 # Substitute a variable $(am__tar) that is a command
 # writing to stdout a FORMAT-tarball containing the directory
@@ -1038,116 +876,76 @@ AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 # Substitute a variable $(am__untar) that extract such
 # a tarball read from stdin.
 #     $(am__untar) < result.tar
-#
 AC_DEFUN([_AM_PROG_TAR],
-[# Always define AMTAR for backward compatibility.  Yes, it's still used
-# in the wild :-(  We should find a proper way to deprecate it ...
-AC_SUBST([AMTAR], ['$${TAR-tar}'])
-
-# We'll loop over all known methods to create a tar archive until one works.
-_am_tools='gnutar m4_if([$1], [ustar], [plaintar]) pax cpio none'
-
+[# Always define AMTAR for backward compatibility.
+AM_MISSING_PROG([AMTAR], [tar])
 m4_if([$1], [v7],
-  [am__tar='$${TAR-tar} chof - "$$tardir"' am__untar='$${TAR-tar} xf -'],
+     [am__tar='${AMTAR} chof - "$$tardir"'; am__untar='${AMTAR} xf -'],
+     [m4_case([$1], [ustar],, [pax],,
+              [m4_fatal([Unknown tar format])])
+AC_MSG_CHECKING([how to create a $1 tar archive])
+# Loop over all known methods to create a tar archive until one works.
+_am_tools='gnutar m4_if([$1], [ustar], [plaintar]) pax cpio none'
+_am_tools=${am_cv_prog_tar_$1-$_am_tools}
+# Do not fold the above two line into one, because Tru64 sh and
+# Solaris sh will not grok spaces in the rhs of `-'.
+for _am_tool in $_am_tools
+do
+  case $_am_tool in
+  gnutar)
+    for _am_tar in tar gnutar gtar;
+    do
+      AM_RUN_LOG([$_am_tar --version]) && break
+    done
+    am__tar="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$$tardir"'
+    am__tar_="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$tardir"'
+    am__untar="$_am_tar -xf -"
+    ;;
+  plaintar)
+    # Must skip GNU tar: if it does not support --format= it doesn't create
+    # ustar tarball either.
+    (tar --version) >/dev/null 2>&1 && continue
+    am__tar='tar chf - "$$tardir"'
+    am__tar_='tar chf - "$tardir"'
+    am__untar='tar xf -'
+    ;;
+  pax)
+    am__tar='pax -L -x $1 -w "$$tardir"'
+    am__tar_='pax -L -x $1 -w "$tardir"'
+    am__untar='pax -r'
+    ;;
+  cpio)
+    am__tar='find "$$tardir" -print | cpio -o -H $1 -L'
+    am__tar_='find "$tardir" -print | cpio -o -H $1 -L'
+    am__untar='cpio -i -H $1 -d'
+    ;;
+  none)
+    am__tar=false
+    am__tar_=false
+    am__untar=false
+    ;;
+  esac
 
-  [m4_case([$1],
-    [ustar],
-     [# The POSIX 1988 'ustar' format is defined with fixed-size fields.
-      # There is notably a 21 bits limit for the UID and the GID.  In fact,
-      # the 'pax' utility can hang on bigger UID/GID (see automake bug#8343
-      # and bug#13588).
-      am_max_uid=2097151 # 2^21 - 1
-      am_max_gid=$am_max_uid
-      # The $UID and $GID variables are not portable, so we need to resort
-      # to the POSIX-mandated id(1) utility.  Errors in the 'id' calls
-      # below are definitely unexpected, so allow the users to see them
-      # (that is, avoid stderr redirection).
-      am_uid=`id -u || echo unknown`
-      am_gid=`id -g || echo unknown`
-      AC_MSG_CHECKING([whether UID '$am_uid' is supported by ustar format])
-      if test $am_uid -le $am_max_uid; then
-         AC_MSG_RESULT([yes])
-      else
-         AC_MSG_RESULT([no])
-         _am_tools=none
-      fi
-      AC_MSG_CHECKING([whether GID '$am_gid' is supported by ustar format])
-      if test $am_gid -le $am_max_gid; then
-         AC_MSG_RESULT([yes])
-      else
-        AC_MSG_RESULT([no])
-        _am_tools=none
-      fi],
+  # If the value was cached, stop now.  We just wanted to have am__tar
+  # and am__untar set.
+  test -n "${am_cv_prog_tar_$1}" && break
 
-  [pax],
-    [],
-
-  [m4_fatal([Unknown tar format])])
-
-  AC_MSG_CHECKING([how to create a $1 tar archive])
-
-  # Go ahead even if we have the value already cached.  We do so because we
-  # need to set the values for the 'am__tar' and 'am__untar' variables.
-  _am_tools=${am_cv_prog_tar_$1-$_am_tools}
-
-  for _am_tool in $_am_tools; do
-    case $_am_tool in
-    gnutar)
-      for _am_tar in tar gnutar gtar; do
-        AM_RUN_LOG([$_am_tar --version]) && break
-      done
-      am__tar="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$$tardir"'
-      am__tar_="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$tardir"'
-      am__untar="$_am_tar -xf -"
-      ;;
-    plaintar)
-      # Must skip GNU tar: if it does not support --format= it doesn't create
-      # ustar tarball either.
-      (tar --version) >/dev/null 2>&1 && continue
-      am__tar='tar chf - "$$tardir"'
-      am__tar_='tar chf - "$tardir"'
-      am__untar='tar xf -'
-      ;;
-    pax)
-      am__tar='pax -L -x $1 -w "$$tardir"'
-      am__tar_='pax -L -x $1 -w "$tardir"'
-      am__untar='pax -r'
-      ;;
-    cpio)
-      am__tar='find "$$tardir" -print | cpio -o -H $1 -L'
-      am__tar_='find "$tardir" -print | cpio -o -H $1 -L'
-      am__untar='cpio -i -H $1 -d'
-      ;;
-    none)
-      am__tar=false
-      am__tar_=false
-      am__untar=false
-      ;;
-    esac
-
-    # If the value was cached, stop now.  We just wanted to have am__tar
-    # and am__untar set.
-    test -n "${am_cv_prog_tar_$1}" && break
-
-    # tar/untar a dummy directory, and stop if the command works.
-    rm -rf conftest.dir
-    mkdir conftest.dir
-    echo GrepMe > conftest.dir/file
-    AM_RUN_LOG([tardir=conftest.dir && eval $am__tar_ >conftest.tar])
-    rm -rf conftest.dir
-    if test -s conftest.tar; then
-      AM_RUN_LOG([$am__untar <conftest.tar])
-      AM_RUN_LOG([cat conftest.dir/file])
-      grep GrepMe conftest.dir/file >/dev/null 2>&1 && break
-    fi
-  done
+  # tar/untar a dummy directory, and stop if the command works
   rm -rf conftest.dir
+  mkdir conftest.dir
+  echo GrepMe > conftest.dir/file
+  AM_RUN_LOG([tardir=conftest.dir && eval $am__tar_ >conftest.tar])
+  rm -rf conftest.dir
+  if test -s conftest.tar; then
+    AM_RUN_LOG([$am__untar <conftest.tar])
+    grep GrepMe conftest.dir/file >/dev/null 2>&1 && break
+  fi
+done
+rm -rf conftest.dir
 
-  AC_CACHE_VAL([am_cv_prog_tar_$1], [am_cv_prog_tar_$1=$_am_tool])
-  AC_MSG_RESULT([$am_cv_prog_tar_$1])])
-
+AC_CACHE_VAL([am_cv_prog_tar_$1], [am_cv_prog_tar_$1=$_am_tool])
+AC_MSG_RESULT([$am_cv_prog_tar_$1])])
 AC_SUBST([am__tar])
 AC_SUBST([am__untar])
 ]) # _AM_PROG_TAR
 
-m4_include([m4/ax_gcc_var_attribute.m4])
diff --git a/configure b/configure
index ea75c846..f63f8031 100755
--- a/configure
+++ b/configure
@@ -1,22 +1,20 @@
 #! /bin/sh
 # Guess values for system-dependent variables and create Makefiles.
-# Generated by GNU Autoconf 2.69 for Grid 1.0.
+# Generated by GNU Autoconf 2.63 for Grid 1.0.
 #
 # Report bugs to <paboyle@ph.ed.ac.uk>.
 #
-#
-# Copyright (C) 1992-1996, 1998-2012 Free Software Foundation, Inc.
-#
-#
+# Copyright (C) 1992, 1993, 1994, 1995, 1996, 1998, 1999, 2000, 2001,
+# 2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
 # This configure script is free software; the Free Software Foundation
 # gives unlimited permission to copy, distribute and modify it.
-## -------------------- ##
-## M4sh Initialization. ##
-## -------------------- ##
+## --------------------- ##
+## M4sh Initialization.  ##
+## --------------------- ##
 
 # Be more Bourne compatible
 DUALCASE=1; export DUALCASE # for MKS sh
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
   emulate sh
   NULLCMD=:
   # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
@@ -24,15 +22,23 @@ if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
   alias -g '${1+"$@"}'='"$@"'
   setopt NO_GLOB_SUBST
 else
-  case `(set -o) 2>/dev/null` in #(
-  *posix*) :
-    set -o posix ;; #(
-  *) :
-     ;;
+  case `(set -o) 2>/dev/null` in
+  *posix*) set -o posix ;;
 esac
+
 fi
 
 
+
+
+# PATH needs CR
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
+
 as_nl='
 '
 export as_nl
@@ -40,13 +46,7 @@ export as_nl
 as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
-# Prefer a ksh shell builtin over an external printf program on Solaris,
-# but without wasting forks for bash or zsh.
-if test -z "$BASH_VERSION$ZSH_VERSION" \
-    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
-  as_echo='print -r --'
-  as_echo_n='print -rn --'
-elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+if (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
   as_echo='printf %s\n'
   as_echo_n='printf %s'
 else
@@ -57,7 +57,7 @@ else
     as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
     as_echo_n_body='eval
       arg=$1;
-      case $arg in #(
+      case $arg in
       *"$as_nl"*)
 	expr "X$arg" : "X\\(.*\\)$as_nl";
 	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
@@ -80,6 +80,13 @@ if test "${PATH_SEPARATOR+set}" != set; then
   }
 fi
 
+# Support unset when possible.
+if ( (MAIL=60; unset MAIL) || exit) >/dev/null 2>&1; then
+  as_unset=unset
+else
+  as_unset=false
+fi
+
 
 # IFS
 # We need space, tab and new line, in precisely that order.  Quoting is
@@ -89,16 +96,15 @@ fi
 IFS=" ""	$as_nl"
 
 # Find who we are.  Look in the path if we contain no directory separator.
-as_myself=
-case $0 in #((
+case $0 in
   *[\\/]* ) as_myself=$0 ;;
   *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
 for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
-  done
+  test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+done
 IFS=$as_save_IFS
 
      ;;
@@ -110,16 +116,12 @@ if test "x$as_myself" = x; then
 fi
 if test ! -f "$as_myself"; then
   $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
-  exit 1
+  { (exit 1); exit 1; }
 fi
 
-# Unset variables that we do not need and which cause bugs (e.g. in
-# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
-# suppresses any "Segmentation fault" message there.  '((' could
-# trigger a bug in pdksh 5.2.14.
-for as_var in BASH_ENV ENV MAIL MAILPATH
-do eval test x\${$as_var+set} = xset \
-  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
+# Work around bugs in pre-3.0 UWIN ksh.
+for as_var in ENV MAIL MAILPATH
+do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
 done
 PS1='$ '
 PS2='> '
@@ -131,294 +133,7 @@ export LC_ALL
 LANGUAGE=C
 export LANGUAGE
 
-# CDPATH.
-(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
-
-# Use a proper internal environment variable to ensure we don't fall
-  # into an infinite loop, continuously re-executing ourselves.
-  if test x"${_as_can_reexec}" != xno && test "x$CONFIG_SHELL" != x; then
-    _as_can_reexec=no; export _as_can_reexec;
-    # We cannot yet assume a decent shell, so we have to provide a
-# neutralization value for shells without unset; and this also
-# works around shells that cannot unset nonexistent variables.
-# Preserve -v and -x to the replacement shell.
-BASH_ENV=/dev/null
-ENV=/dev/null
-(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
-case $- in # ((((
-  *v*x* | *x*v* ) as_opts=-vx ;;
-  *v* ) as_opts=-v ;;
-  *x* ) as_opts=-x ;;
-  * ) as_opts= ;;
-esac
-exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
-# Admittedly, this is quite paranoid, since all the known shells bail
-# out after a failed `exec'.
-$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
-as_fn_exit 255
-  fi
-  # We don't want this to propagate to other subprocesses.
-          { _as_can_reexec=; unset _as_can_reexec;}
-if test "x$CONFIG_SHELL" = x; then
-  as_bourne_compatible="if test -n \"\${ZSH_VERSION+set}\" && (emulate sh) >/dev/null 2>&1; then :
-  emulate sh
-  NULLCMD=:
-  # Pre-4.2 versions of Zsh do word splitting on \${1+\"\$@\"}, which
-  # is contrary to our usage.  Disable this feature.
-  alias -g '\${1+\"\$@\"}'='\"\$@\"'
-  setopt NO_GLOB_SUBST
-else
-  case \`(set -o) 2>/dev/null\` in #(
-  *posix*) :
-    set -o posix ;; #(
-  *) :
-     ;;
-esac
-fi
-"
-  as_required="as_fn_return () { (exit \$1); }
-as_fn_success () { as_fn_return 0; }
-as_fn_failure () { as_fn_return 1; }
-as_fn_ret_success () { return 0; }
-as_fn_ret_failure () { return 1; }
-
-exitcode=0
-as_fn_success || { exitcode=1; echo as_fn_success failed.; }
-as_fn_failure && { exitcode=1; echo as_fn_failure succeeded.; }
-as_fn_ret_success || { exitcode=1; echo as_fn_ret_success failed.; }
-as_fn_ret_failure && { exitcode=1; echo as_fn_ret_failure succeeded.; }
-if ( set x; as_fn_ret_success y && test x = \"\$1\" ); then :
-
-else
-  exitcode=1; echo positional parameters were not saved.
-fi
-test x\$exitcode = x0 || exit 1
-test -x / || exit 1"
-  as_suggested="  as_lineno_1=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_1a=\$LINENO
-  as_lineno_2=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_2a=\$LINENO
-  eval 'test \"x\$as_lineno_1'\$as_run'\" != \"x\$as_lineno_2'\$as_run'\" &&
-  test \"x\`expr \$as_lineno_1'\$as_run' + 1\`\" = \"x\$as_lineno_2'\$as_run'\"' || exit 1
-test \$(( 1 + 1 )) = 2 || exit 1"
-  if (eval "$as_required") 2>/dev/null; then :
-  as_have_required=yes
-else
-  as_have_required=no
-fi
-  if test x$as_have_required = xyes && (eval "$as_suggested") 2>/dev/null; then :
-
-else
-  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-as_found=false
-for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-  as_found=:
-  case $as_dir in #(
-	 /*)
-	   for as_base in sh bash ksh sh5; do
-	     # Try only shells that exist, to save several forks.
-	     as_shell=$as_dir/$as_base
-	     if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
-		    { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$as_shell"; } 2>/dev/null; then :
-  CONFIG_SHELL=$as_shell as_have_required=yes
-		   if { $as_echo "$as_bourne_compatible""$as_suggested" | as_run=a "$as_shell"; } 2>/dev/null; then :
-  break 2
-fi
-fi
-	   done;;
-       esac
-  as_found=false
-done
-$as_found || { if { test -f "$SHELL" || test -f "$SHELL.exe"; } &&
-	      { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$SHELL"; } 2>/dev/null; then :
-  CONFIG_SHELL=$SHELL as_have_required=yes
-fi; }
-IFS=$as_save_IFS
-
-
-      if test "x$CONFIG_SHELL" != x; then :
-  export CONFIG_SHELL
-             # We cannot yet assume a decent shell, so we have to provide a
-# neutralization value for shells without unset; and this also
-# works around shells that cannot unset nonexistent variables.
-# Preserve -v and -x to the replacement shell.
-BASH_ENV=/dev/null
-ENV=/dev/null
-(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
-case $- in # ((((
-  *v*x* | *x*v* ) as_opts=-vx ;;
-  *v* ) as_opts=-v ;;
-  *x* ) as_opts=-x ;;
-  * ) as_opts= ;;
-esac
-exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
-# Admittedly, this is quite paranoid, since all the known shells bail
-# out after a failed `exec'.
-$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
-exit 255
-fi
-
-    if test x$as_have_required = xno; then :
-  $as_echo "$0: This script requires a shell more modern than all"
-  $as_echo "$0: the shells that I found on your system."
-  if test x${ZSH_VERSION+set} = xset ; then
-    $as_echo "$0: In particular, zsh $ZSH_VERSION has bugs and should"
-    $as_echo "$0: be upgraded to zsh 4.3.4 or later."
-  else
-    $as_echo "$0: Please tell bug-autoconf@gnu.org and
-$0: paboyle@ph.ed.ac.uk about your system, including any
-$0: error possibly output before this message. Then install
-$0: a modern shell, or manually run the script under such a
-$0: shell if you do have one."
-  fi
-  exit 1
-fi
-fi
-fi
-SHELL=${CONFIG_SHELL-/bin/sh}
-export SHELL
-# Unset more variables known to interfere with behavior of common tools.
-CLICOLOR_FORCE= GREP_OPTIONS=
-unset CLICOLOR_FORCE GREP_OPTIONS
-
-## --------------------- ##
-## M4sh Shell Functions. ##
-## --------------------- ##
-# as_fn_unset VAR
-# ---------------
-# Portably unset VAR.
-as_fn_unset ()
-{
-  { eval $1=; unset $1;}
-}
-as_unset=as_fn_unset
-
-# as_fn_set_status STATUS
-# -----------------------
-# Set $? to STATUS, without forking.
-as_fn_set_status ()
-{
-  return $1
-} # as_fn_set_status
-
-# as_fn_exit STATUS
-# -----------------
-# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
-as_fn_exit ()
-{
-  set +e
-  as_fn_set_status $1
-  exit $1
-} # as_fn_exit
-
-# as_fn_mkdir_p
-# -------------
-# Create "$as_dir" as a directory, including parents if necessary.
-as_fn_mkdir_p ()
-{
-
-  case $as_dir in #(
-  -*) as_dir=./$as_dir;;
-  esac
-  test -d "$as_dir" || eval $as_mkdir_p || {
-    as_dirs=
-    while :; do
-      case $as_dir in #(
-      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
-      *) as_qdir=$as_dir;;
-      esac
-      as_dirs="'$as_qdir' $as_dirs"
-      as_dir=`$as_dirname -- "$as_dir" ||
-$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_dir" : 'X\(//\)[^/]' \| \
-	 X"$as_dir" : 'X\(//\)$' \| \
-	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_dir" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      test -d "$as_dir" && break
-    done
-    test -z "$as_dirs" || eval "mkdir $as_dirs"
-  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
-
-
-} # as_fn_mkdir_p
-
-# as_fn_executable_p FILE
-# -----------------------
-# Test if FILE is an executable regular file.
-as_fn_executable_p ()
-{
-  test -f "$1" && test -x "$1"
-} # as_fn_executable_p
-# as_fn_append VAR VALUE
-# ----------------------
-# Append the text in VALUE to the end of the definition contained in VAR. Take
-# advantage of any shell optimizations that allow amortized linear growth over
-# repeated appends, instead of the typical quadratic growth present in naive
-# implementations.
-if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
-  eval 'as_fn_append ()
-  {
-    eval $1+=\$2
-  }'
-else
-  as_fn_append ()
-  {
-    eval $1=\$$1\$2
-  }
-fi # as_fn_append
-
-# as_fn_arith ARG...
-# ------------------
-# Perform arithmetic evaluation on the ARGs, and store the result in the
-# global $as_val. Take advantage of shells that can avoid forks. The arguments
-# must be portable across $(()) and expr.
-if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
-  eval 'as_fn_arith ()
-  {
-    as_val=$(( $* ))
-  }'
-else
-  as_fn_arith ()
-  {
-    as_val=`expr "$@" || test $? -eq 1`
-  }
-fi # as_fn_arith
-
-
-# as_fn_error STATUS ERROR [LINENO LOG_FD]
-# ----------------------------------------
-# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
-# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
-# script with STATUS, using 1 if that was 0.
-as_fn_error ()
-{
-  as_status=$1; test $as_status -eq 0 && as_status=1
-  if test "$4"; then
-    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
-  fi
-  $as_echo "$as_me: error: $2" >&2
-  as_fn_exit $as_status
-} # as_fn_error
-
+# Required to use basename.
 if expr a : '\(a\)' >/dev/null 2>&1 &&
    test "X`expr 00001 : '.*\(...\)'`" = X001; then
   as_expr=expr
@@ -432,12 +147,8 @@ else
   as_basename=false
 fi
 
-if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
-  as_dirname=dirname
-else
-  as_dirname=false
-fi
 
+# Name of the executable.
 as_me=`$as_basename -- "$0" ||
 $as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
 	 X"$0" : 'X\(//\)$' \| \
@@ -457,19 +168,295 @@ $as_echo X/"$0" |
 	  }
 	  s/.*/./; q'`
 
-# Avoid depending upon Character Ranges.
-as_cr_letters='abcdefghijklmnopqrstuvwxyz'
-as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
-as_cr_Letters=$as_cr_letters$as_cr_LETTERS
-as_cr_digits='0123456789'
-as_cr_alnum=$as_cr_Letters$as_cr_digits
+# CDPATH.
+$as_unset CDPATH
 
 
-  as_lineno_1=$LINENO as_lineno_1a=$LINENO
-  as_lineno_2=$LINENO as_lineno_2a=$LINENO
-  eval 'test "x$as_lineno_1'$as_run'" != "x$as_lineno_2'$as_run'" &&
-  test "x`expr $as_lineno_1'$as_run' + 1`" = "x$as_lineno_2'$as_run'"' || {
-  # Blame Lee E. McMahon (1931-1989) for sed's syntax.  :-)
+if test "x$CONFIG_SHELL" = x; then
+  if (eval ":") 2>/dev/null; then
+  as_have_required=yes
+else
+  as_have_required=no
+fi
+
+  if test $as_have_required = yes &&	 (eval ":
+(as_func_return () {
+  (exit \$1)
+}
+as_func_success () {
+  as_func_return 0
+}
+as_func_failure () {
+  as_func_return 1
+}
+as_func_ret_success () {
+  return 0
+}
+as_func_ret_failure () {
+  return 1
+}
+
+exitcode=0
+if as_func_success; then
+  :
+else
+  exitcode=1
+  echo as_func_success failed.
+fi
+
+if as_func_failure; then
+  exitcode=1
+  echo as_func_failure succeeded.
+fi
+
+if as_func_ret_success; then
+  :
+else
+  exitcode=1
+  echo as_func_ret_success failed.
+fi
+
+if as_func_ret_failure; then
+  exitcode=1
+  echo as_func_ret_failure succeeded.
+fi
+
+if ( set x; as_func_ret_success y && test x = \"\$1\" ); then
+  :
+else
+  exitcode=1
+  echo positional parameters were not saved.
+fi
+
+test \$exitcode = 0) || { (exit 1); exit 1; }
+
+(
+  as_lineno_1=\$LINENO
+  as_lineno_2=\$LINENO
+  test \"x\$as_lineno_1\" != \"x\$as_lineno_2\" &&
+  test \"x\`expr \$as_lineno_1 + 1\`\" = \"x\$as_lineno_2\") || { (exit 1); exit 1; }
+") 2> /dev/null; then
+  :
+else
+  as_candidate_shells=
+    as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+  case $as_dir in
+	 /*)
+	   for as_base in sh bash ksh sh5; do
+	     as_candidate_shells="$as_candidate_shells $as_dir/$as_base"
+	   done;;
+       esac
+done
+IFS=$as_save_IFS
+
+
+      for as_shell in $as_candidate_shells $SHELL; do
+	 # Try only shells that exist, to save several forks.
+	 if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
+		{ ("$as_shell") 2> /dev/null <<\_ASEOF
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
+  emulate sh
+  NULLCMD=:
+  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '${1+"$@"}'='"$@"'
+  setopt NO_GLOB_SUBST
+else
+  case `(set -o) 2>/dev/null` in
+  *posix*) set -o posix ;;
+esac
+
+fi
+
+
+:
+_ASEOF
+}; then
+  CONFIG_SHELL=$as_shell
+	       as_have_required=yes
+	       if { "$as_shell" 2> /dev/null <<\_ASEOF
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
+  emulate sh
+  NULLCMD=:
+  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '${1+"$@"}'='"$@"'
+  setopt NO_GLOB_SUBST
+else
+  case `(set -o) 2>/dev/null` in
+  *posix*) set -o posix ;;
+esac
+
+fi
+
+
+:
+(as_func_return () {
+  (exit $1)
+}
+as_func_success () {
+  as_func_return 0
+}
+as_func_failure () {
+  as_func_return 1
+}
+as_func_ret_success () {
+  return 0
+}
+as_func_ret_failure () {
+  return 1
+}
+
+exitcode=0
+if as_func_success; then
+  :
+else
+  exitcode=1
+  echo as_func_success failed.
+fi
+
+if as_func_failure; then
+  exitcode=1
+  echo as_func_failure succeeded.
+fi
+
+if as_func_ret_success; then
+  :
+else
+  exitcode=1
+  echo as_func_ret_success failed.
+fi
+
+if as_func_ret_failure; then
+  exitcode=1
+  echo as_func_ret_failure succeeded.
+fi
+
+if ( set x; as_func_ret_success y && test x = "$1" ); then
+  :
+else
+  exitcode=1
+  echo positional parameters were not saved.
+fi
+
+test $exitcode = 0) || { (exit 1); exit 1; }
+
+(
+  as_lineno_1=$LINENO
+  as_lineno_2=$LINENO
+  test "x$as_lineno_1" != "x$as_lineno_2" &&
+  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2") || { (exit 1); exit 1; }
+
+_ASEOF
+}; then
+  break
+fi
+
+fi
+
+      done
+
+      if test "x$CONFIG_SHELL" != x; then
+  for as_var in BASH_ENV ENV
+	do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
+	done
+	export CONFIG_SHELL
+	exec "$CONFIG_SHELL" "$as_myself" ${1+"$@"}
+fi
+
+
+    if test $as_have_required = no; then
+  echo This script requires a shell more modern than all the
+      echo shells that I found on your system.  Please install a
+      echo modern shell, or manually run the script under such a
+      echo shell if you do have one.
+      { (exit 1); exit 1; }
+fi
+
+
+fi
+
+fi
+
+
+
+(eval "as_func_return () {
+  (exit \$1)
+}
+as_func_success () {
+  as_func_return 0
+}
+as_func_failure () {
+  as_func_return 1
+}
+as_func_ret_success () {
+  return 0
+}
+as_func_ret_failure () {
+  return 1
+}
+
+exitcode=0
+if as_func_success; then
+  :
+else
+  exitcode=1
+  echo as_func_success failed.
+fi
+
+if as_func_failure; then
+  exitcode=1
+  echo as_func_failure succeeded.
+fi
+
+if as_func_ret_success; then
+  :
+else
+  exitcode=1
+  echo as_func_ret_success failed.
+fi
+
+if as_func_ret_failure; then
+  exitcode=1
+  echo as_func_ret_failure succeeded.
+fi
+
+if ( set x; as_func_ret_success y && test x = \"\$1\" ); then
+  :
+else
+  exitcode=1
+  echo positional parameters were not saved.
+fi
+
+test \$exitcode = 0") || {
+  echo No shell found that supports shell functions.
+  echo Please tell bug-autoconf@gnu.org about your system,
+  echo including any error possibly output before this message.
+  echo This can help us improve future autoconf versions.
+  echo Configuration will now proceed without shell functions.
+}
+
+
+
+  as_lineno_1=$LINENO
+  as_lineno_2=$LINENO
+  test "x$as_lineno_1" != "x$as_lineno_2" &&
+  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2" || {
+
+  # Create $as_me.lineno as a copy of $as_myself, but with $LINENO
+  # uniformly replaced by the line number.  The first 'sed' inserts a
+  # line-number line after each line using $LINENO; the second 'sed'
+  # does the real work.  The second script uses 'N' to pair each
+  # line-number line with the line containing $LINENO, and appends
+  # trailing '-' during substitution so that $LINENO is not a special
+  # case at line end.
+  # (Raja R Harinath suggested sed '=', and Paul Eggert wrote the
+  # scripts with optimization help from Paolo Bonzini.  Blame Lee
+  # E. McMahon (1931-1989) for sed's syntax.  :-)
   sed -n '
     p
     /[$]LINENO/=
@@ -486,12 +473,9 @@ as_cr_alnum=$as_cr_Letters$as_cr_digits
       s/-\n.*//
     ' >$as_me.lineno &&
   chmod +x "$as_me.lineno" ||
-    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2; as_fn_exit 1; }
+    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2
+   { (exit 1); exit 1; }; }
 
-  # If we had to re-execute with $CONFIG_SHELL, we're ensured to have
-  # already done that, so ensure we don't try to do so again and fall
-  # in an infinite loop.  This has already happened in practice.
-  _as_can_reexec=no; export _as_can_reexec
   # Don't try to exec as it changes $[0], causing all sort of problems
   # (the dirname of $[0] is not the place where we might find the
   # original and so on.  Autoconf is especially sensitive to this).
@@ -500,18 +484,29 @@ as_cr_alnum=$as_cr_Letters$as_cr_digits
   exit
 }
 
+
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
+
 ECHO_C= ECHO_N= ECHO_T=
-case `echo -n x` in #(((((
+case `echo -n x` in
 -n*)
-  case `echo 'xy\c'` in
+  case `echo 'x\c'` in
   *c*) ECHO_T='	';;	# ECHO_T is single tab character.
-  xy)  ECHO_C='\c';;
-  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
-       ECHO_T='	';;
+  *)   ECHO_C='\c';;
   esac;;
 *)
   ECHO_N='-n';;
 esac
+if expr a : '\(a\)' >/dev/null 2>&1 &&
+   test "X`expr 00001 : '.*\(...\)'`" = X001; then
+  as_expr=expr
+else
+  as_expr=false
+fi
 
 rm -f conf$$ conf$$.exe conf$$.file
 if test -d conf$$.dir; then
@@ -526,29 +521,49 @@ if (echo >conf$$.file) 2>/dev/null; then
     # ... but there are two gotchas:
     # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
     # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
-    # In both cases, we have to default to `cp -pR'.
+    # In both cases, we have to default to `cp -p'.
     ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
-      as_ln_s='cp -pR'
+      as_ln_s='cp -p'
   elif ln conf$$.file conf$$ 2>/dev/null; then
     as_ln_s=ln
   else
-    as_ln_s='cp -pR'
+    as_ln_s='cp -p'
   fi
 else
-  as_ln_s='cp -pR'
+  as_ln_s='cp -p'
 fi
 rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
 rmdir conf$$.dir 2>/dev/null
 
 if mkdir -p . 2>/dev/null; then
-  as_mkdir_p='mkdir -p "$as_dir"'
+  as_mkdir_p=:
 else
   test -d ./-p && rmdir ./-p
   as_mkdir_p=false
 fi
 
-as_test_x='test -x'
-as_executable_p=as_fn_executable_p
+if test -x / >/dev/null 2>&1; then
+  as_test_x='test -x'
+else
+  if ls -dL / >/dev/null 2>&1; then
+    as_ls_L_option=L
+  else
+    as_ls_L_option=
+  fi
+  as_test_x='
+    eval sh -c '\''
+      if test -d "$1"; then
+	test -d "$1/.";
+      else
+	case $1 in
+	-*)set "./$1";;
+	esac;
+	case `ls -ld'$as_ls_L_option' "$1" 2>/dev/null` in
+	???[sx]*):;;*)false;;esac;fi
+    '\'' sh
+  '
+fi
+as_executable_p=$as_test_x
 
 # Sed expression to map a string onto a valid CPP name.
 as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
@@ -557,11 +572,11 @@ as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
 as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
 
 
-test -n "$DJDIR" || exec 7<&0 </dev/null
-exec 6>&1
+
+exec 7<&0 </dev/null 6>&1
 
 # Name of the host.
-# hostname on some systems (SVR3.2, old GNU/Linux) returns a bogus exit status,
+# hostname on some systems (SVR3.2, Linux) returns a bogus exit status,
 # so uname gets run too.
 ac_hostname=`(hostname || uname -n) 2>/dev/null | sed 1q`
 
@@ -576,6 +591,7 @@ cross_compiling=no
 subdirs=
 MFLAGS=
 MAKEFLAGS=
+SHELL=${CONFIG_SHELL-/bin/sh}
 
 # Identity of this package.
 PACKAGE_NAME='Grid'
@@ -583,7 +599,6 @@ PACKAGE_TARNAME='grid'
 PACKAGE_VERSION='1.0'
 PACKAGE_STRING='Grid 1.0'
 PACKAGE_BUGREPORT='paboyle@ph.ed.ac.uk'
-PACKAGE_URL=''
 
 ac_unique_file="Grid.h"
 # Factoring default headers for most tests.
@@ -640,7 +655,6 @@ CC
 am__fastdepCXX_FALSE
 am__fastdepCXX_TRUE
 CXXDEPMODE
-am__nodep
 AMDEPBACKSLASH
 AMDEP_FALSE
 AMDEP_TRUE
@@ -654,10 +668,6 @@ CPPFLAGS
 LDFLAGS
 CXXFLAGS
 CXX
-AM_BACKSLASH
-AM_DEFAULT_VERBOSITY
-AM_DEFAULT_V
-AM_V
 am__untar
 am__tar
 AMTAR
@@ -711,7 +721,6 @@ bindir
 program_transform_name
 prefix
 exec_prefix
-PACKAGE_URL
 PACKAGE_BUGREPORT
 PACKAGE_STRING
 PACKAGE_VERSION
@@ -722,7 +731,6 @@ SHELL'
 ac_subst_files=''
 ac_user_opts='
 enable_option_checking
-enable_silent_rules
 enable_dependency_tracking
 enable_openmp
 enable_simd
@@ -801,9 +809,8 @@ do
   fi
 
   case $ac_option in
-  *=?*) ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
-  *=)   ac_optarg= ;;
-  *)    ac_optarg=yes ;;
+  *=*)	ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
+  *)	ac_optarg=yes ;;
   esac
 
   # Accept the important Cygnus configure options, so we can diagnose typos.
@@ -848,7 +855,8 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*disable-\(.*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid feature name: $ac_useropt"
+      { $as_echo "$as_me: error: invalid feature name: $ac_useropt" >&2
+   { (exit 1); exit 1; }; }
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -874,7 +882,8 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*enable-\([^=]*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid feature name: $ac_useropt"
+      { $as_echo "$as_me: error: invalid feature name: $ac_useropt" >&2
+   { (exit 1); exit 1; }; }
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1078,7 +1087,8 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*with-\([^=]*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid package name: $ac_useropt"
+      { $as_echo "$as_me: error: invalid package name: $ac_useropt" >&2
+   { (exit 1); exit 1; }; }
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1094,7 +1104,8 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*without-\(.*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid package name: $ac_useropt"
+      { $as_echo "$as_me: error: invalid package name: $ac_useropt" >&2
+   { (exit 1); exit 1; }; }
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1124,17 +1135,17 @@ do
   | --x-librar=* | --x-libra=* | --x-libr=* | --x-lib=* | --x-li=* | --x-l=*)
     x_libraries=$ac_optarg ;;
 
-  -*) as_fn_error $? "unrecognized option: \`$ac_option'
-Try \`$0 --help' for more information"
+  -*) { $as_echo "$as_me: error: unrecognized option: $ac_option
+Try \`$0 --help' for more information." >&2
+   { (exit 1); exit 1; }; }
     ;;
 
   *=*)
     ac_envvar=`expr "x$ac_option" : 'x\([^=]*\)='`
     # Reject names that are not valid shell variable names.
-    case $ac_envvar in #(
-      '' | [0-9]* | *[!_$as_cr_alnum]* )
-      as_fn_error $? "invalid variable name: \`$ac_envvar'" ;;
-    esac
+    expr "x$ac_envvar" : ".*[^_$as_cr_alnum]" >/dev/null &&
+      { $as_echo "$as_me: error: invalid variable name: $ac_envvar" >&2
+   { (exit 1); exit 1; }; }
     eval $ac_envvar=\$ac_optarg
     export $ac_envvar ;;
 
@@ -1143,7 +1154,7 @@ Try \`$0 --help' for more information"
     $as_echo "$as_me: WARNING: you should use --build, --host, --target" >&2
     expr "x$ac_option" : ".*[^-._$as_cr_alnum]" >/dev/null &&
       $as_echo "$as_me: WARNING: invalid host type: $ac_option" >&2
-    : "${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}"
+    : ${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}
     ;;
 
   esac
@@ -1151,13 +1162,15 @@ done
 
 if test -n "$ac_prev"; then
   ac_option=--`echo $ac_prev | sed 's/_/-/g'`
-  as_fn_error $? "missing argument to $ac_option"
+  { $as_echo "$as_me: error: missing argument to $ac_option" >&2
+   { (exit 1); exit 1; }; }
 fi
 
 if test -n "$ac_unrecognized_opts"; then
   case $enable_option_checking in
     no) ;;
-    fatal) as_fn_error $? "unrecognized options: $ac_unrecognized_opts" ;;
+    fatal) { $as_echo "$as_me: error: unrecognized options: $ac_unrecognized_opts" >&2
+   { (exit 1); exit 1; }; } ;;
     *)     $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2 ;;
   esac
 fi
@@ -1180,7 +1193,8 @@ do
     [\\/$]* | ?:[\\/]* )  continue;;
     NONE | '' ) case $ac_var in *prefix ) continue;; esac;;
   esac
-  as_fn_error $? "expected an absolute directory name for --$ac_var: $ac_val"
+  { $as_echo "$as_me: error: expected an absolute directory name for --$ac_var: $ac_val" >&2
+   { (exit 1); exit 1; }; }
 done
 
 # There might be people who depend on the old broken behavior: `$host'
@@ -1194,6 +1208,8 @@ target=$target_alias
 if test "x$host_alias" != x; then
   if test "x$build_alias" = x; then
     cross_compiling=maybe
+    $as_echo "$as_me: WARNING: If you wanted to set the --build type, don't use --host.
+    If a cross compiler is detected then cross compile mode will be used." >&2
   elif test "x$build_alias" != "x$host_alias"; then
     cross_compiling=yes
   fi
@@ -1208,9 +1224,11 @@ test "$silent" = yes && exec 6>/dev/null
 ac_pwd=`pwd` && test -n "$ac_pwd" &&
 ac_ls_di=`ls -di .` &&
 ac_pwd_ls_di=`cd "$ac_pwd" && ls -di .` ||
-  as_fn_error $? "working directory cannot be determined"
+  { $as_echo "$as_me: error: working directory cannot be determined" >&2
+   { (exit 1); exit 1; }; }
 test "X$ac_ls_di" = "X$ac_pwd_ls_di" ||
-  as_fn_error $? "pwd does not report name of working directory"
+  { $as_echo "$as_me: error: pwd does not report name of working directory" >&2
+   { (exit 1); exit 1; }; }
 
 
 # Find the source files, if location was not specified.
@@ -1249,11 +1267,13 @@ else
 fi
 if test ! -r "$srcdir/$ac_unique_file"; then
   test "$ac_srcdir_defaulted" = yes && srcdir="$ac_confdir or .."
-  as_fn_error $? "cannot find sources ($ac_unique_file) in $srcdir"
+  { $as_echo "$as_me: error: cannot find sources ($ac_unique_file) in $srcdir" >&2
+   { (exit 1); exit 1; }; }
 fi
 ac_msg="sources are in $srcdir, but \`cd $srcdir' does not work"
 ac_abs_confdir=`(
-	cd "$srcdir" && test -r "./$ac_unique_file" || as_fn_error $? "$ac_msg"
+	cd "$srcdir" && test -r "./$ac_unique_file" || { $as_echo "$as_me: error: $ac_msg" >&2
+   { (exit 1); exit 1; }; }
 	pwd)`
 # When building in place, set srcdir=.
 if test "$ac_abs_confdir" = "$ac_pwd"; then
@@ -1293,7 +1313,7 @@ Configuration:
       --help=short        display options specific to this package
       --help=recursive    display the short help of all the included packages
   -V, --version           display version information and exit
-  -q, --quiet, --silent   do not print \`checking ...' messages
+  -q, --quiet, --silent   do not print \`checking...' messages
       --cache-file=FILE   cache test results in FILE [disabled]
   -C, --config-cache      alias for \`--cache-file=config.cache'
   -n, --no-create         do not create output files
@@ -1353,12 +1373,8 @@ Optional Features:
   --disable-option-checking  ignore unrecognized --enable/--with options
   --disable-FEATURE       do not include FEATURE (same as --enable-FEATURE=no)
   --enable-FEATURE[=ARG]  include FEATURE [ARG=yes]
-  --enable-silent-rules   less verbose build output (undo: "make V=1")
-  --disable-silent-rules  verbose build output (undo: "make V=0")
-  --enable-dependency-tracking
-                          do not reject slow dependency extractors
-  --disable-dependency-tracking
-                          speeds up one-time build
+  --disable-dependency-tracking  speeds up one-time build
+  --enable-dependency-tracking   do not reject slow dependency extractors
   --disable-openmp        do not use OpenMP
   --enable-simd=SSE|AVX|AVX2|AVX512
                           Select instructions
@@ -1369,7 +1385,7 @@ Some influential environment variables:
   LDFLAGS     linker flags, e.g. -L<lib dir> if you have libraries in a
               nonstandard directory <lib dir>
   LIBS        libraries to pass to the linker, e.g. -l<library>
-  CPPFLAGS    (Objective) C/C++ preprocessor flags, e.g. -I<include dir> if
+  CPPFLAGS    C/C++/Objective C preprocessor flags, e.g. -I<include dir> if
               you have headers in a nonstandard directory <include dir>
   CC          C compiler command
   CFLAGS      C compiler flags
@@ -1442,522 +1458,21 @@ test -n "$ac_init_help" && exit $ac_status
 if $ac_init_version; then
   cat <<\_ACEOF
 Grid configure 1.0
-generated by GNU Autoconf 2.69
+generated by GNU Autoconf 2.63
 
-Copyright (C) 2012 Free Software Foundation, Inc.
+Copyright (C) 1992, 1993, 1994, 1995, 1996, 1998, 1999, 2000, 2001,
+2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
 This configure script is free software; the Free Software Foundation
 gives unlimited permission to copy, distribute and modify it.
 _ACEOF
   exit
 fi
-
-## ------------------------ ##
-## Autoconf initialization. ##
-## ------------------------ ##
-
-# ac_fn_cxx_try_compile LINENO
-# ----------------------------
-# Try to compile conftest.$ac_ext, and return whether this succeeded.
-ac_fn_cxx_try_compile ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  rm -f conftest.$ac_objext
-  if { { ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compile") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_retval=1
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_cxx_try_compile
-
-# ac_fn_c_try_compile LINENO
-# --------------------------
-# Try to compile conftest.$ac_ext, and return whether this succeeded.
-ac_fn_c_try_compile ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  rm -f conftest.$ac_objext
-  if { { ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compile") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_retval=1
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_c_try_compile
-
-# ac_fn_c_try_link LINENO
-# -----------------------
-# Try to link conftest.$ac_ext, and return whether this succeeded.
-ac_fn_c_try_link ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  rm -f conftest.$ac_objext conftest$ac_exeext
-  if { { ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_link") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 test -x conftest$ac_exeext
-       }; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_retval=1
-fi
-  # Delete the IPA/IPO (Inter Procedural Analysis/Optimization) information
-  # created by the PGI compiler (conftest_ipa8_conftest.oo), as it would
-  # interfere with the next link command; also delete a directory that is
-  # left behind by Apple's compiler.  We do this before executing the actions.
-  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_c_try_link
-
-# ac_fn_c_try_cpp LINENO
-# ----------------------
-# Try to preprocess conftest.$ac_ext, and return whether this succeeded.
-ac_fn_c_try_cpp ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  if { { ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } > conftest.i && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-    ac_retval=1
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_c_try_cpp
-
-# ac_fn_c_check_header_mongrel LINENO HEADER VAR INCLUDES
-# -------------------------------------------------------
-# Tests whether HEADER exists, giving a warning if it cannot be compiled using
-# the include files in INCLUDES and setting the cache variable VAR
-# accordingly.
-ac_fn_c_check_header_mongrel ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  if eval \${$3+:} false; then :
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 usability" >&5
-$as_echo_n "checking $2 usability... " >&6; }
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-#include <$2>
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  ac_header_compiler=yes
-else
-  ac_header_compiler=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 presence" >&5
-$as_echo_n "checking $2 presence... " >&6; }
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <$2>
-_ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
-  ac_header_preproc=yes
-else
-  ac_header_preproc=no
-fi
-rm -f conftest.err conftest.i conftest.$ac_ext
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in #((
-  yes:no: )
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $2: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $2:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $2: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
-( $as_echo "## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##"
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  eval "$3=\$ac_header_compiler"
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_check_header_mongrel
-
-# ac_fn_c_try_run LINENO
-# ----------------------
-# Try to link conftest.$ac_ext, and return whether this succeeded. Assumes
-# that executables *can* be run.
-ac_fn_c_try_run ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  if { { ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_link") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && { ac_try='./conftest$ac_exeext'
-  { { case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; }; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: program exited with status $ac_status" >&5
-       $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-       ac_retval=$ac_status
-fi
-  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_c_try_run
-
-# ac_fn_c_check_header_compile LINENO HEADER VAR INCLUDES
-# -------------------------------------------------------
-# Tests whether HEADER exists and can be compiled using the include files in
-# INCLUDES, setting the cache variable VAR accordingly.
-ac_fn_c_check_header_compile ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-#include <$2>
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  eval "$3=yes"
-else
-  eval "$3=no"
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_check_header_compile
-
-# ac_fn_c_check_type LINENO TYPE VAR INCLUDES
-# -------------------------------------------
-# Tests whether TYPE exists after having included INCLUDES, setting cache
-# variable VAR accordingly.
-ac_fn_c_check_type ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  eval "$3=no"
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-int
-main ()
-{
-if (sizeof ($2))
-	 return 0;
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-int
-main ()
-{
-if (sizeof (($2)))
-	    return 0;
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-
-else
-  eval "$3=yes"
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_check_type
-
-# ac_fn_c_find_uintX_t LINENO BITS VAR
-# ------------------------------------
-# Finds an unsigned integer type with width BITS, setting cache variable VAR
-# accordingly.
-ac_fn_c_find_uintX_t ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for uint$2_t" >&5
-$as_echo_n "checking for uint$2_t... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  eval "$3=no"
-     # Order is important - never check a type that is potentially smaller
-     # than half of the expected target width.
-     for ac_type in uint$2_t 'unsigned int' 'unsigned long int' \
-	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
-       cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-static int test_array [1 - 2 * !((($ac_type) -1 >> ($2 / 2 - 1)) >> ($2 / 2 - 1) == 3)];
-test_array [0] = 0;
-return test_array [0];
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  case $ac_type in #(
-  uint$2_t) :
-    eval "$3=yes" ;; #(
-  *) :
-    eval "$3=\$ac_type" ;;
-esac
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-       if eval test \"x\$"$3"\" = x"no"; then :
-
-else
-  break
-fi
-     done
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_find_uintX_t
-
-# ac_fn_c_check_func LINENO FUNC VAR
-# ----------------------------------
-# Tests whether FUNC exists, setting the cache variable VAR accordingly
-ac_fn_c_check_func ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-/* Define $2 to an innocuous variant, in case <limits.h> declares $2.
-   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
-#define $2 innocuous_$2
-
-/* System header to define __stub macros and hopefully few prototypes,
-    which can conflict with char $2 (); below.
-    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
-    <limits.h> exists even on freestanding compilers.  */
-
-#ifdef __STDC__
-# include <limits.h>
-#else
-# include <assert.h>
-#endif
-
-#undef $2
-
-/* Override any GCC internal prototype to avoid an error.
-   Use char because int might match the return type of a GCC
-   builtin and then its argument prototype would still apply.  */
-#ifdef __cplusplus
-extern "C"
-#endif
-char $2 ();
-/* The GNU C library defines this for functions which it implements
-    to always fail with ENOSYS.  Some functions are actually named
-    something starting with __ and the normal name is an alias.  */
-#if defined __stub_$2 || defined __stub___$2
-choke me
-#endif
-
-int
-main ()
-{
-return $2 ();
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_link "$LINENO"; then :
-  eval "$3=yes"
-else
-  eval "$3=no"
-fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_check_func
 cat >config.log <<_ACEOF
 This file contains any messages produced by compilers while
 running configure, to aid debugging if configure makes a mistake.
 
 It was created by Grid $as_me 1.0, which was
-generated by GNU Autoconf 2.69.  Invocation command line was
+generated by GNU Autoconf 2.63.  Invocation command line was
 
   $ $0 $@
 
@@ -1993,8 +1508,8 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    $as_echo "PATH: $as_dir"
-  done
+  $as_echo "PATH: $as_dir"
+done
 IFS=$as_save_IFS
 
 } >&5
@@ -2031,9 +1546,9 @@ do
       ac_arg=`$as_echo "$ac_arg" | sed "s/'/'\\\\\\\\''/g"` ;;
     esac
     case $ac_pass in
-    1) as_fn_append ac_configure_args0 " '$ac_arg'" ;;
+    1) ac_configure_args0="$ac_configure_args0 '$ac_arg'" ;;
     2)
-      as_fn_append ac_configure_args1 " '$ac_arg'"
+      ac_configure_args1="$ac_configure_args1 '$ac_arg'"
       if test $ac_must_keep_next = true; then
 	ac_must_keep_next=false # Got value, back to normal.
       else
@@ -2049,13 +1564,13 @@ do
 	  -* ) ac_must_keep_next=true ;;
 	esac
       fi
-      as_fn_append ac_configure_args " '$ac_arg'"
+      ac_configure_args="$ac_configure_args '$ac_arg'"
       ;;
     esac
   done
 done
-{ ac_configure_args0=; unset ac_configure_args0;}
-{ ac_configure_args1=; unset ac_configure_args1;}
+$as_unset ac_configure_args0 || test "${ac_configure_args0+set}" != set || { ac_configure_args0=; export ac_configure_args0; }
+$as_unset ac_configure_args1 || test "${ac_configure_args1+set}" != set || { ac_configure_args1=; export ac_configure_args1; }
 
 # When interrupted or exit'd, cleanup temporary files, and complete
 # config.log.  We remove comments because anyway the quotes in there
@@ -2067,9 +1582,11 @@ trap 'exit_status=$?
   {
     echo
 
-    $as_echo "## ---------------- ##
+    cat <<\_ASBOX
+## ---------------- ##
 ## Cache variables. ##
-## ---------------- ##"
+## ---------------- ##
+_ASBOX
     echo
     # The following way of writing the cache mishandles newlines in values,
 (
@@ -2078,13 +1595,13 @@ trap 'exit_status=$?
     case $ac_val in #(
     *${as_nl}*)
       case $ac_var in #(
-      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
+      *_cv_*) { $as_echo "$as_me:$LINENO: WARNING: cache variable $ac_var contains a newline" >&5
 $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
       esac
       case $ac_var in #(
       _ | IFS | as_nl) ;; #(
       BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
-      *) { eval $ac_var=; unset $ac_var;} ;;
+      *) $as_unset $ac_var ;;
       esac ;;
     esac
   done
@@ -2103,9 +1620,11 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
 )
     echo
 
-    $as_echo "## ----------------- ##
+    cat <<\_ASBOX
+## ----------------- ##
 ## Output variables. ##
-## ----------------- ##"
+## ----------------- ##
+_ASBOX
     echo
     for ac_var in $ac_subst_vars
     do
@@ -2118,9 +1637,11 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
     echo
 
     if test -n "$ac_subst_files"; then
-      $as_echo "## ------------------- ##
+      cat <<\_ASBOX
+## ------------------- ##
 ## File substitutions. ##
-## ------------------- ##"
+## ------------------- ##
+_ASBOX
       echo
       for ac_var in $ac_subst_files
       do
@@ -2134,9 +1655,11 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
     fi
 
     if test -s confdefs.h; then
-      $as_echo "## ----------- ##
+      cat <<\_ASBOX
+## ----------- ##
 ## confdefs.h. ##
-## ----------- ##"
+## ----------- ##
+_ASBOX
       echo
       cat confdefs.h
       echo
@@ -2150,39 +1673,37 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
     exit $exit_status
 ' 0
 for ac_signal in 1 2 13 15; do
-  trap 'ac_signal='$ac_signal'; as_fn_exit 1' $ac_signal
+  trap 'ac_signal='$ac_signal'; { (exit 1); exit 1; }' $ac_signal
 done
 ac_signal=0
 
 # confdefs.h avoids OS command line length limits that DEFS can exceed.
 rm -f -r conftest* confdefs.h
 
-$as_echo "/* confdefs.h */" > confdefs.h
-
 # Predefined preprocessor variables.
 
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_NAME "$PACKAGE_NAME"
 _ACEOF
 
+
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_TARNAME "$PACKAGE_TARNAME"
 _ACEOF
 
+
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_VERSION "$PACKAGE_VERSION"
 _ACEOF
 
+
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_STRING "$PACKAGE_STRING"
 _ACEOF
 
-cat >>confdefs.h <<_ACEOF
-#define PACKAGE_BUGREPORT "$PACKAGE_BUGREPORT"
-_ACEOF
 
 cat >>confdefs.h <<_ACEOF
-#define PACKAGE_URL "$PACKAGE_URL"
+#define PACKAGE_BUGREPORT "$PACKAGE_BUGREPORT"
 _ACEOF
 
 
@@ -2191,12 +1712,7 @@ _ACEOF
 ac_site_file1=NONE
 ac_site_file2=NONE
 if test -n "$CONFIG_SITE"; then
-  # We do not want a PATH search for config.site.
-  case $CONFIG_SITE in #((
-    -*)  ac_site_file1=./$CONFIG_SITE;;
-    */*) ac_site_file1=$CONFIG_SITE;;
-    *)   ac_site_file1=./$CONFIG_SITE;;
-  esac
+  ac_site_file1=$CONFIG_SITE
 elif test "x$prefix" != xNONE; then
   ac_site_file1=$prefix/share/config.site
   ac_site_file2=$prefix/etc/config.site
@@ -2207,23 +1723,19 @@ fi
 for ac_site_file in "$ac_site_file1" "$ac_site_file2"
 do
   test "x$ac_site_file" = xNONE && continue
-  if test /dev/null != "$ac_site_file" && test -r "$ac_site_file"; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: loading site script $ac_site_file" >&5
+  if test -r "$ac_site_file"; then
+    { $as_echo "$as_me:$LINENO: loading site script $ac_site_file" >&5
 $as_echo "$as_me: loading site script $ac_site_file" >&6;}
     sed 's/^/| /' "$ac_site_file" >&5
-    . "$ac_site_file" \
-      || { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "failed to load site script $ac_site_file
-See \`config.log' for more details" "$LINENO" 5; }
+    . "$ac_site_file"
   fi
 done
 
 if test -r "$cache_file"; then
-  # Some versions of bash will fail to source /dev/null (special files
-  # actually), so we avoid doing that.  DJGPP emulates it as a regular file.
-  if test /dev/null != "$cache_file" && test -f "$cache_file"; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: loading cache $cache_file" >&5
+  # Some versions of bash will fail to source /dev/null (special
+  # files actually), so we avoid doing that.
+  if test -f "$cache_file"; then
+    { $as_echo "$as_me:$LINENO: loading cache $cache_file" >&5
 $as_echo "$as_me: loading cache $cache_file" >&6;}
     case $cache_file in
       [\\/]* | ?:[\\/]* ) . "$cache_file";;
@@ -2231,7 +1743,7 @@ $as_echo "$as_me: loading cache $cache_file" >&6;}
     esac
   fi
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: creating cache $cache_file" >&5
+  { $as_echo "$as_me:$LINENO: creating cache $cache_file" >&5
 $as_echo "$as_me: creating cache $cache_file" >&6;}
   >$cache_file
 fi
@@ -2246,11 +1758,11 @@ for ac_var in $ac_precious_vars; do
   eval ac_new_val=\$ac_env_${ac_var}_value
   case $ac_old_set,$ac_new_set in
     set,)
-      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
+      { $as_echo "$as_me:$LINENO: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
 $as_echo "$as_me: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&2;}
       ac_cache_corrupted=: ;;
     ,set)
-      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was not set in the previous run" >&5
+      { $as_echo "$as_me:$LINENO: error: \`$ac_var' was not set in the previous run" >&5
 $as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
       ac_cache_corrupted=: ;;
     ,);;
@@ -2260,17 +1772,17 @@ $as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
 	ac_old_val_w=`echo x $ac_old_val`
 	ac_new_val_w=`echo x $ac_new_val`
 	if test "$ac_old_val_w" != "$ac_new_val_w"; then
-	  { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' has changed since the previous run:" >&5
+	  { $as_echo "$as_me:$LINENO: error: \`$ac_var' has changed since the previous run:" >&5
 $as_echo "$as_me: error: \`$ac_var' has changed since the previous run:" >&2;}
 	  ac_cache_corrupted=:
 	else
-	  { $as_echo "$as_me:${as_lineno-$LINENO}: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
+	  { $as_echo "$as_me:$LINENO: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
 $as_echo "$as_me: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&2;}
 	  eval $ac_var=\$ac_old_val
 	fi
-	{ $as_echo "$as_me:${as_lineno-$LINENO}:   former value:  \`$ac_old_val'" >&5
+	{ $as_echo "$as_me:$LINENO:   former value:  \`$ac_old_val'" >&5
 $as_echo "$as_me:   former value:  \`$ac_old_val'" >&2;}
-	{ $as_echo "$as_me:${as_lineno-$LINENO}:   current value: \`$ac_new_val'" >&5
+	{ $as_echo "$as_me:$LINENO:   current value: \`$ac_new_val'" >&5
 $as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
       fi;;
   esac
@@ -2282,20 +1794,43 @@ $as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
     esac
     case " $ac_configure_args " in
       *" '$ac_arg' "*) ;; # Avoid dups.  Use of quotes ensures accuracy.
-      *) as_fn_append ac_configure_args " '$ac_arg'" ;;
+      *) ac_configure_args="$ac_configure_args '$ac_arg'" ;;
     esac
   fi
 done
 if $ac_cache_corrupted; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+  { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-  { $as_echo "$as_me:${as_lineno-$LINENO}: error: changes in the environment can compromise the build" >&5
+  { $as_echo "$as_me:$LINENO: error: changes in the environment can compromise the build" >&5
 $as_echo "$as_me: error: changes in the environment can compromise the build" >&2;}
-  as_fn_error $? "run \`make distclean' and/or \`rm $cache_file' and start over" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: run \`make distclean' and/or \`rm $cache_file' and start over" >&5
+$as_echo "$as_me: error: run \`make distclean' and/or \`rm $cache_file' and start over" >&2;}
+   { (exit 1); exit 1; }; }
 fi
-## -------------------- ##
-## Main body of script. ##
-## -------------------- ##
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
 
 ac_ext=c
 ac_cpp='$CPP $CPPFLAGS'
@@ -2304,7 +1839,7 @@ ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
 
-am__api_version='1.15'
+am__api_version='1.11'
 
 ac_aux_dir=
 for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
@@ -2323,7 +1858,9 @@ for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
   fi
 done
 if test -z "$ac_aux_dir"; then
-  as_fn_error $? "cannot find install-sh, install.sh, or shtool in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: cannot find install-sh or install.sh in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" >&5
+$as_echo "$as_me: error: cannot find install-sh or install.sh in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" >&2;}
+   { (exit 1); exit 1; }; }
 fi
 
 # These three variables are undocumented and unsupported,
@@ -2349,10 +1886,10 @@ ac_configure="$SHELL $ac_aux_dir/configure"  # Please don't use this var.
 # OS/2's system install, which has a completely different semantic
 # ./install, which can be erroneously created by make from ./install.sh.
 # Reject install programs that cannot install multiple files.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a BSD-compatible install" >&5
+{ $as_echo "$as_me:$LINENO: checking for a BSD-compatible install" >&5
 $as_echo_n "checking for a BSD-compatible install... " >&6; }
 if test -z "$INSTALL"; then
-if ${ac_cv_path_install+:} false; then :
+if test "${ac_cv_path_install+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
@@ -2360,11 +1897,11 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    # Account for people who put trailing slashes in PATH elements.
-case $as_dir/ in #((
-  ./ | .// | /[cC]/* | \
+  # Account for people who put trailing slashes in PATH elements.
+case $as_dir/ in
+  ./ | .// | /cC/* | \
   /etc/* | /usr/sbin/* | /usr/etc/* | /sbin/* | /usr/afsws/bin/* | \
-  ?:[\\/]os2[\\/]install[\\/]* | ?:[\\/]OS2[\\/]INSTALL[\\/]* | \
+  ?:\\/os2\\/install\\/* | ?:\\/OS2\\/INSTALL\\/* | \
   /usr/ucb/* ) ;;
   *)
     # OSF1 and SCO ODT 3.0 have their own names for install.
@@ -2372,7 +1909,7 @@ case $as_dir/ in #((
     # by default.
     for ac_prog in ginstall scoinst install; do
       for ac_exec_ext in '' $ac_executable_extensions; do
-	if as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext"; then
+	if { test -f "$as_dir/$ac_prog$ac_exec_ext" && $as_test_x "$as_dir/$ac_prog$ac_exec_ext"; }; then
 	  if test $ac_prog = install &&
 	    grep dspmsg "$as_dir/$ac_prog$ac_exec_ext" >/dev/null 2>&1; then
 	    # AIX install.  It has an incompatible calling convention.
@@ -2401,7 +1938,7 @@ case $as_dir/ in #((
     ;;
 esac
 
-  done
+done
 IFS=$as_save_IFS
 
 rm -rf conftest.one conftest.two conftest.dir
@@ -2417,7 +1954,7 @@ fi
     INSTALL=$ac_install_sh
   fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $INSTALL" >&5
+{ $as_echo "$as_me:$LINENO: result: $INSTALL" >&5
 $as_echo "$INSTALL" >&6; }
 
 # Use test -z because SunOS4 sh mishandles braces in ${var-val}.
@@ -2428,73 +1965,68 @@ test -z "$INSTALL_SCRIPT" && INSTALL_SCRIPT='${INSTALL}'
 
 test -z "$INSTALL_DATA" && INSTALL_DATA='${INSTALL} -m 644'
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether build environment is sane" >&5
+{ $as_echo "$as_me:$LINENO: checking whether build environment is sane" >&5
 $as_echo_n "checking whether build environment is sane... " >&6; }
+# Just in case
+sleep 1
+echo timestamp > conftest.file
 # Reject unsafe characters in $srcdir or the absolute working directory
 # name.  Accept space and tab only in the latter.
 am_lf='
 '
 case `pwd` in
   *[\\\"\#\$\&\'\`$am_lf]*)
-    as_fn_error $? "unsafe absolute working directory name" "$LINENO" 5;;
+    { { $as_echo "$as_me:$LINENO: error: unsafe absolute working directory name" >&5
+$as_echo "$as_me: error: unsafe absolute working directory name" >&2;}
+   { (exit 1); exit 1; }; };;
 esac
 case $srcdir in
   *[\\\"\#\$\&\'\`$am_lf\ \	]*)
-    as_fn_error $? "unsafe srcdir value: '$srcdir'" "$LINENO" 5;;
+    { { $as_echo "$as_me:$LINENO: error: unsafe srcdir value: \`$srcdir'" >&5
+$as_echo "$as_me: error: unsafe srcdir value: \`$srcdir'" >&2;}
+   { (exit 1); exit 1; }; };;
 esac
 
-# Do 'set' in a subshell so we don't clobber the current shell's
+# Do `set' in a subshell so we don't clobber the current shell's
 # arguments.  Must try -L first in case configure is actually a
 # symlink; some systems play weird games with the mod time of symlinks
 # (eg FreeBSD returns the mod time of the symlink's containing
 # directory).
 if (
-   am_has_slept=no
-   for am_try in 1 2; do
-     echo "timestamp, slept: $am_has_slept" > conftest.file
-     set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
-     if test "$*" = "X"; then
-	# -L didn't work.
-	set X `ls -t "$srcdir/configure" conftest.file`
-     fi
-     if test "$*" != "X $srcdir/configure conftest.file" \
-	&& test "$*" != "X conftest.file $srcdir/configure"; then
+   set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
+   if test "$*" = "X"; then
+      # -L didn't work.
+      set X `ls -t "$srcdir/configure" conftest.file`
+   fi
+   rm -f conftest.file
+   if test "$*" != "X $srcdir/configure conftest.file" \
+      && test "$*" != "X conftest.file $srcdir/configure"; then
+
+      # If neither matched, then we have a broken ls.  This can happen
+      # if, for instance, CONFIG_SHELL is bash and it inherits a
+      # broken ls alias from the environment.  This has actually
+      # happened.  Such a system could not be considered "sane".
+      { { $as_echo "$as_me:$LINENO: error: ls -t appears to fail.  Make sure there is not a broken
+alias in your environment" >&5
+$as_echo "$as_me: error: ls -t appears to fail.  Make sure there is not a broken
+alias in your environment" >&2;}
+   { (exit 1); exit 1; }; }
+   fi
 
-	# If neither matched, then we have a broken ls.  This can happen
-	# if, for instance, CONFIG_SHELL is bash and it inherits a
-	# broken ls alias from the environment.  This has actually
-	# happened.  Such a system could not be considered "sane".
-	as_fn_error $? "ls -t appears to fail.  Make sure there is not a broken
-  alias in your environment" "$LINENO" 5
-     fi
-     if test "$2" = conftest.file || test $am_try -eq 2; then
-       break
-     fi
-     # Just in case.
-     sleep 1
-     am_has_slept=yes
-   done
    test "$2" = conftest.file
    )
 then
    # Ok.
    :
 else
-   as_fn_error $? "newly created file is older than distributed files!
-Check your system clock" "$LINENO" 5
+   { { $as_echo "$as_me:$LINENO: error: newly created file is older than distributed files!
+Check your system clock" >&5
+$as_echo "$as_me: error: newly created file is older than distributed files!
+Check your system clock" >&2;}
+   { (exit 1); exit 1; }; }
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+{ $as_echo "$as_me:$LINENO: result: yes" >&5
 $as_echo "yes" >&6; }
-# If we didn't sleep, we still need to ensure time stamps of config.status and
-# generated files are strictly newer.
-am_sleep_pid=
-if grep 'slept: no' conftest.file >/dev/null 2>&1; then
-  ( sleep 1 ) &
-  am_sleep_pid=$!
-fi
-
-rm -f conftest.file
-
 test "$program_prefix" != NONE &&
   program_transform_name="s&^&$program_prefix&;$program_transform_name"
 # Use a double $ so make ignores it.
@@ -2505,8 +2037,8 @@ test "$program_suffix" != NONE &&
 ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
 program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
 
-# Expand $ac_aux_dir to an absolute path.
-am_aux_dir=`cd "$ac_aux_dir" && pwd`
+# expand $ac_aux_dir to an absolute path
+am_aux_dir=`cd $ac_aux_dir && pwd`
 
 if test x"${MISSING+set}" != xset; then
   case $am_aux_dir in
@@ -2517,15 +2049,15 @@ if test x"${MISSING+set}" != xset; then
   esac
 fi
 # Use eval to expand $SHELL
-if eval "$MISSING --is-lightweight"; then
-  am_missing_run="$MISSING "
+if eval "$MISSING --run true"; then
+  am_missing_run="$MISSING --run "
 else
   am_missing_run=
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: 'missing' script is too old or missing" >&5
-$as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
+  { $as_echo "$as_me:$LINENO: WARNING: \`missing' script is too old or missing" >&5
+$as_echo "$as_me: WARNING: \`missing' script is too old or missing" >&2;}
 fi
 
-if test x"${install_sh+set}" != xset; then
+if test x"${install_sh}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -2534,17 +2066,17 @@ if test x"${install_sh+set}" != xset; then
   esac
 fi
 
-# Installed binaries are usually stripped using 'strip' when the user
-# run "make install-strip".  However 'strip' might not be the right
+# Installed binaries are usually stripped using `strip' when the user
+# run `make install-strip'.  However `strip' might not be the right
 # tool to use in cross-compilation environments, therefore Automake
-# will honor the 'STRIP' environment variable to overrule this program.
+# will honor the `STRIP' environment variable to overrule this program.
 if test "$cross_compiling" != no; then
   if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}strip", so it can be a program name with args.
 set dummy ${ac_tool_prefix}strip; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_STRIP+:} false; then :
+if test "${ac_cv_prog_STRIP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$STRIP"; then
@@ -2555,24 +2087,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_STRIP="${ac_tool_prefix}strip"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 STRIP=$ac_cv_prog_STRIP
 if test -n "$STRIP"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $STRIP" >&5
+  { $as_echo "$as_me:$LINENO: result: $STRIP" >&5
 $as_echo "$STRIP" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2582,9 +2114,9 @@ if test -z "$ac_cv_prog_STRIP"; then
   ac_ct_STRIP=$STRIP
   # Extract the first word of "strip", so it can be a program name with args.
 set dummy strip; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_STRIP+:} false; then :
+if test "${ac_cv_prog_ac_ct_STRIP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_STRIP"; then
@@ -2595,24 +2127,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_ac_ct_STRIP="strip"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_STRIP=$ac_cv_prog_ac_ct_STRIP
 if test -n "$ac_ct_STRIP"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_STRIP" >&5
+  { $as_echo "$as_me:$LINENO: result: $ac_ct_STRIP" >&5
 $as_echo "$ac_ct_STRIP" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2621,7 +2153,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -2634,10 +2166,10 @@ fi
 fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a thread-safe mkdir -p" >&5
+{ $as_echo "$as_me:$LINENO: checking for a thread-safe mkdir -p" >&5
 $as_echo_n "checking for a thread-safe mkdir -p... " >&6; }
 if test -z "$MKDIR_P"; then
-  if ${ac_cv_path_mkdir+:} false; then :
+  if test "${ac_cv_path_mkdir+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
@@ -2645,9 +2177,9 @@ for as_dir in $PATH$PATH_SEPARATOR/opt/sfw/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_prog in mkdir gmkdir; do
+  for ac_prog in mkdir gmkdir; do
 	 for ac_exec_ext in '' $ac_executable_extensions; do
-	   as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext" || continue
+	   { test -f "$as_dir/$ac_prog$ac_exec_ext" && $as_test_x "$as_dir/$ac_prog$ac_exec_ext"; } || continue
 	   case `"$as_dir/$ac_prog$ac_exec_ext" --version 2>&1` in #(
 	     'mkdir (GNU coreutils) '* | \
 	     'mkdir (coreutils) '* | \
@@ -2657,12 +2189,11 @@ do
 	   esac
 	 done
        done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 
-  test -d ./--version && rmdir ./--version
   if test "${ac_cv_path_mkdir+set}" = set; then
     MKDIR_P="$ac_cv_path_mkdir -p"
   else
@@ -2670,19 +2201,26 @@ fi
     # value for MKDIR_P within a source directory, because that will
     # break other packages using the cache if that directory is
     # removed, or if the value is a relative name.
+    test -d ./--version && rmdir ./--version
     MKDIR_P="$ac_install_sh -d"
   fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $MKDIR_P" >&5
+{ $as_echo "$as_me:$LINENO: result: $MKDIR_P" >&5
 $as_echo "$MKDIR_P" >&6; }
 
+mkdir_p="$MKDIR_P"
+case $mkdir_p in
+  [\\/$]* | ?:[\\/]*) ;;
+  */*) mkdir_p="\$(top_builddir)/$mkdir_p" ;;
+esac
+
 for ac_prog in gawk mawk nawk awk
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_AWK+:} false; then :
+if test "${ac_cv_prog_AWK+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$AWK"; then
@@ -2693,24 +2231,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_AWK="$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 AWK=$ac_cv_prog_AWK
 if test -n "$AWK"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $AWK" >&5
+  { $as_echo "$as_me:$LINENO: result: $AWK" >&5
 $as_echo "$AWK" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2718,11 +2256,11 @@ fi
   test -n "$AWK" && break
 done
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether ${MAKE-make} sets \$(MAKE)" >&5
+{ $as_echo "$as_me:$LINENO: checking whether ${MAKE-make} sets \$(MAKE)" >&5
 $as_echo_n "checking whether ${MAKE-make} sets \$(MAKE)... " >&6; }
 set x ${MAKE-make}
 ac_make=`$as_echo "$2" | sed 's/+/p/g; s/[^a-zA-Z0-9_]/_/g'`
-if eval \${ac_cv_prog_make_${ac_make}_set+:} false; then :
+if { as_var=ac_cv_prog_make_${ac_make}_set; eval "test \"\${$as_var+set}\" = set"; }; then
   $as_echo_n "(cached) " >&6
 else
   cat >conftest.make <<\_ACEOF
@@ -2730,7 +2268,7 @@ SHELL = /bin/sh
 all:
 	@echo '@@@%%%=$(MAKE)=@@@%%%'
 _ACEOF
-# GNU make sometimes prints "make[1]: Entering ...", which would confuse us.
+# GNU make sometimes prints "make[1]: Entering...", which would confuse us.
 case `${MAKE-make} -f conftest.make 2>/dev/null` in
   *@@@%%%=?*=@@@%%%*)
     eval ac_cv_prog_make_${ac_make}_set=yes;;
@@ -2740,11 +2278,11 @@ esac
 rm -f conftest.make
 fi
 if eval test \$ac_cv_prog_make_${ac_make}_set = yes; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+  { $as_echo "$as_me:$LINENO: result: yes" >&5
 $as_echo "yes" >&6; }
   SET_MAKE=
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
   SET_MAKE="MAKE=${MAKE-make}"
 fi
@@ -2758,52 +2296,15 @@ else
 fi
 rmdir .tst 2>/dev/null
 
-# Check whether --enable-silent-rules was given.
-if test "${enable_silent_rules+set}" = set; then :
-  enableval=$enable_silent_rules;
-fi
-
-case $enable_silent_rules in # (((
-  yes) AM_DEFAULT_VERBOSITY=0;;
-   no) AM_DEFAULT_VERBOSITY=1;;
-    *) AM_DEFAULT_VERBOSITY=1;;
-esac
-am_make=${MAKE-make}
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $am_make supports nested variables" >&5
-$as_echo_n "checking whether $am_make supports nested variables... " >&6; }
-if ${am_cv_make_support_nested_variables+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if $as_echo 'TRUE=$(BAR$(V))
-BAR0=false
-BAR1=true
-V=1
-am__doit:
-	@$(TRUE)
-.PHONY: am__doit' | $am_make -f - >/dev/null 2>&1; then
-  am_cv_make_support_nested_variables=yes
-else
-  am_cv_make_support_nested_variables=no
-fi
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_make_support_nested_variables" >&5
-$as_echo "$am_cv_make_support_nested_variables" >&6; }
-if test $am_cv_make_support_nested_variables = yes; then
-    AM_V='$(V)'
-  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
-else
-  AM_V=$AM_DEFAULT_VERBOSITY
-  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
-fi
-AM_BACKSLASH='\'
-
 if test "`cd $srcdir && pwd`" != "`pwd`"; then
   # Use -I$(srcdir) only when $(srcdir) != ., so that make's output
   # is not polluted with repeated "-I."
   am__isrc=' -I$(srcdir)'
   # test to see if srcdir already configured
   if test -f $srcdir/config.status; then
-    as_fn_error $? "source directory already configured; run \"make distclean\" there first" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: source directory already configured; run \"make distclean\" there first" >&5
+$as_echo "$as_me: error: source directory already configured; run \"make distclean\" there first" >&2;}
+   { (exit 1); exit 1; }; }
   fi
 fi
 
@@ -2847,72 +2348,19 @@ AUTOHEADER=${AUTOHEADER-"${am_missing_run}autoheader"}
 
 MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 
-# For better backward compatibility.  To be removed once Automake 1.9.x
-# dies out for good.  For more background, see:
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
-mkdir_p='$(MKDIR_P)'
+# We need awk for the "check" target.  The system "awk" is bad on
+# some platforms.
+# Always define AMTAR for backward compatibility.
 
-# We need awk for the "check" target (and possibly the TAP driver).  The
-# system "awk" is bad on some platforms.
-# Always define AMTAR for backward compatibility.  Yes, it's still used
-# in the wild :-(  We should find a proper way to deprecate it ...
-AMTAR='$${TAR-tar}'
+AMTAR=${AMTAR-"${am_missing_run}tar"}
 
-
-# We'll loop over all known methods to create a tar archive until one works.
-_am_tools='gnutar  pax cpio none'
-
-am__tar='$${TAR-tar} chof - "$$tardir"' am__untar='$${TAR-tar} xf -'
+am__tar='${AMTAR} chof - "$$tardir"'; am__untar='${AMTAR} xf -'
 
 
 
 
 
 
-# POSIX will say in a future version that running "rm -f" with no argument
-# is OK; and we want to be able to make that assumption in our Makefile
-# recipes.  So use an aggressive probe to check that the usage we want is
-# actually supported "in the wild" to an acceptable degree.
-# See automake bug#10828.
-# To make any issue more visible, cause the running configure to be aborted
-# by default if the 'rm' program in use doesn't match our expectations; the
-# user can still override this though.
-if rm -f && rm -fr && rm -rf; then : OK; else
-  cat >&2 <<'END'
-Oops!
-
-Your 'rm' program seems unable to run without file operands specified
-on the command line, even when the '-f' option is present.  This is contrary
-to the behaviour of most rm programs out there, and not conforming with
-the upcoming POSIX standard: <http://austingroupbugs.net/view.php?id=542>
-
-Please tell bug-automake@gnu.org about your system, including the value
-of your $PATH and any error possibly output before this message.  This
-can help us improve future automake versions.
-
-END
-  if test x"$ACCEPT_INFERIOR_RM_PROGRAM" = x"yes"; then
-    echo 'Configuration will proceed anyway, since you have set the' >&2
-    echo 'ACCEPT_INFERIOR_RM_PROGRAM variable to "yes"' >&2
-    echo >&2
-  else
-    cat >&2 <<'END'
-Aborting the configuration process, to ensure you take notice of the issue.
-
-You can download and install GNU coreutils to get an 'rm' implementation
-that behaves properly: <http://www.gnu.org/software/coreutils/>.
-
-If you want to complete the configuration process using your problematic
-'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
-to "yes", and re-run configure.
-
-END
-    as_fn_error $? "Your 'rm' program is bad, sorry." "$LINENO" 5
-  fi
-fi
-
-
 
 ac_config_headers="$ac_config_headers Grid_config.h"
 
@@ -2932,9 +2380,9 @@ if test -z "$CXX"; then
   do
     # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
 set dummy $ac_tool_prefix$ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CXX+:} false; then :
+if test "${ac_cv_prog_CXX+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CXX"; then
@@ -2945,24 +2393,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_CXX="$ac_tool_prefix$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 CXX=$ac_cv_prog_CXX
 if test -n "$CXX"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CXX" >&5
+  { $as_echo "$as_me:$LINENO: result: $CXX" >&5
 $as_echo "$CXX" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2976,9 +2424,9 @@ if test -z "$CXX"; then
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_CXX+:} false; then :
+if test "${ac_cv_prog_ac_ct_CXX+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CXX"; then
@@ -2989,24 +2437,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_ac_ct_CXX="$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CXX=$ac_cv_prog_ac_ct_CXX
 if test -n "$ac_ct_CXX"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CXX" >&5
+  { $as_echo "$as_me:$LINENO: result: $ac_ct_CXX" >&5
 $as_echo "$ac_ct_CXX" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3019,7 +2467,7 @@ done
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -3030,31 +2478,48 @@ fi
   fi
 fi
 # Provide some information about the compiler.
-$as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler version" >&5
+$as_echo "$as_me:$LINENO: checking for C++ compiler version" >&5
 set X $ac_compile
 ac_compiler=$2
-for ac_option in --version -v -V -qversion; do
-  { { ac_try="$ac_compiler $ac_option >&5"
+{ (ac_try="$ac_compiler --version >&5"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler --version >&5") 2>&5
   ac_status=$?
-  if test -s conftest.err; then
-    sed '10a\
-... rest of stderr output deleted ...
-         10q' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-  fi
-  rm -f conftest.er1 conftest.err
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }
-done
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
+{ (ac_try="$ac_compiler -v >&5"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler -v >&5") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
+{ (ac_try="$ac_compiler -V >&5"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler -V >&5") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
 
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3070,8 +2535,8 @@ ac_clean_files="$ac_clean_files a.out a.out.dSYM a.exe b.out"
 # Try to create an executable without -o first, disregard a.out.
 # It will help us diagnose broken compilers, and finding out an intuition
 # of exeext.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether the C++ compiler works" >&5
-$as_echo_n "checking whether the C++ compiler works... " >&6; }
+{ $as_echo "$as_me:$LINENO: checking for C++ compiler default output file name" >&5
+$as_echo_n "checking for C++ compiler default output file name... " >&6; }
 ac_link_default=`$as_echo "$ac_link" | sed 's/ -o *conftest[^ ]*//'`
 
 # The possible output files:
@@ -3087,17 +2552,17 @@ do
 done
 rm -f $ac_rmfiles
 
-if { { ac_try="$ac_link_default"
+if { (ac_try="$ac_link_default"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
   (eval "$ac_link_default") 2>&5
   ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; then :
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; then
   # Autoconf-2.13 could set the ac_cv_exeext variable to `no'.
 # So ignore a value of `no', otherwise this would lead to `EXEEXT = no'
 # in a Makefile.  We should not override ac_cv_exeext if it was cached,
@@ -3114,7 +2579,7 @@ do
 	# certainly right.
 	break;;
     *.* )
-	if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
+        if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
 	then :; else
 	   ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
 	fi
@@ -3133,41 +2598,84 @@ test "$ac_cv_exeext" = no && ac_cv_exeext=
 else
   ac_file=''
 fi
-if test -z "$ac_file"; then :
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-$as_echo "$as_me: failed program was:" >&5
+
+{ $as_echo "$as_me:$LINENO: result: $ac_file" >&5
+$as_echo "$ac_file" >&6; }
+if test -z "$ac_file"; then
+  $as_echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
 
-{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error 77 "C++ compiler cannot create executables
-See \`config.log' for more details" "$LINENO" 5; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
-$as_echo "yes" >&6; }
+{ { $as_echo "$as_me:$LINENO: error: C++ compiler cannot create executables
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: C++ compiler cannot create executables
+See \`config.log' for more details." >&2;}
+   { (exit 77); exit 77; }; }; }
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler default output file name" >&5
-$as_echo_n "checking for C++ compiler default output file name... " >&6; }
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_file" >&5
-$as_echo "$ac_file" >&6; }
+
 ac_exeext=$ac_cv_exeext
 
+# Check that the compiler produces executables we can run.  If not, either
+# the compiler is broken, or we cross compile.
+{ $as_echo "$as_me:$LINENO: checking whether the C++ compiler works" >&5
+$as_echo_n "checking whether the C++ compiler works... " >&6; }
+# FIXME: These cross compiler hacks should be removed for Autoconf 3.0
+# If not cross compiling, check that we can run a simple program.
+if test "$cross_compiling" != yes; then
+  if { ac_try='./$ac_file'
+  { (case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; }; then
+    cross_compiling=no
+  else
+    if test "$cross_compiling" = maybe; then
+	cross_compiling=yes
+    else
+	{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+{ { $as_echo "$as_me:$LINENO: error: cannot run C++ compiled programs.
+If you meant to cross compile, use \`--host'.
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: cannot run C++ compiled programs.
+If you meant to cross compile, use \`--host'.
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
+    fi
+  fi
+fi
+{ $as_echo "$as_me:$LINENO: result: yes" >&5
+$as_echo "yes" >&6; }
+
 rm -f -r a.out a.out.dSYM a.exe conftest$ac_cv_exeext b.out
 ac_clean_files=$ac_clean_files_save
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of executables" >&5
+# Check that the compiler produces executables we can run.  If not, either
+# the compiler is broken, or we cross compile.
+{ $as_echo "$as_me:$LINENO: checking whether we are cross compiling" >&5
+$as_echo_n "checking whether we are cross compiling... " >&6; }
+{ $as_echo "$as_me:$LINENO: result: $cross_compiling" >&5
+$as_echo "$cross_compiling" >&6; }
+
+{ $as_echo "$as_me:$LINENO: checking for suffix of executables" >&5
 $as_echo_n "checking for suffix of executables... " >&6; }
-if { { ac_try="$ac_link"
+if { (ac_try="$ac_link"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
   (eval "$ac_link") 2>&5
   ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; then :
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; then
   # If both `conftest.exe' and `conftest' are `present' (well, observable)
 # catch `conftest.exe'.  For instance with Cygwin, `ls conftest' will
 # work properly (i.e., refer to `conftest.exe'), while it won't with
@@ -3182,83 +2690,32 @@ for ac_file in conftest.exe conftest conftest.*; do
   esac
 done
 else
-  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+  { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "cannot compute suffix of executables: cannot compile and link
-See \`config.log' for more details" "$LINENO" 5; }
+{ { $as_echo "$as_me:$LINENO: error: cannot compute suffix of executables: cannot compile and link
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: cannot compute suffix of executables: cannot compile and link
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
 fi
-rm -f conftest conftest$ac_cv_exeext
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_exeext" >&5
+
+rm -f conftest$ac_cv_exeext
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_exeext" >&5
 $as_echo "$ac_cv_exeext" >&6; }
 
 rm -f conftest.$ac_ext
 EXEEXT=$ac_cv_exeext
 ac_exeext=$EXEEXT
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <stdio.h>
-int
-main ()
-{
-FILE *f = fopen ("conftest.out", "w");
- return ferror (f) || fclose (f) != 0;
-
-  ;
-  return 0;
-}
-_ACEOF
-ac_clean_files="$ac_clean_files conftest.out"
-# Check that the compiler produces executables we can run.  If not, either
-# the compiler is broken, or we cross compile.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are cross compiling" >&5
-$as_echo_n "checking whether we are cross compiling... " >&6; }
-if test "$cross_compiling" != yes; then
-  { { ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_link") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }
-  if { ac_try='./conftest$ac_cv_exeext'
-  { { case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; }; then
-    cross_compiling=no
-  else
-    if test "$cross_compiling" = maybe; then
-	cross_compiling=yes
-    else
-	{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "cannot run C++ compiled programs.
-If you meant to cross compile, use \`--host'.
-See \`config.log' for more details" "$LINENO" 5; }
-    fi
-  fi
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $cross_compiling" >&5
-$as_echo "$cross_compiling" >&6; }
-
-rm -f conftest.$ac_ext conftest$ac_cv_exeext conftest.out
-ac_clean_files=$ac_clean_files_save
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of object files" >&5
+{ $as_echo "$as_me:$LINENO: checking for suffix of object files" >&5
 $as_echo_n "checking for suffix of object files... " >&6; }
-if ${ac_cv_objext+:} false; then :
+if test "${ac_cv_objext+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3270,17 +2727,17 @@ main ()
 }
 _ACEOF
 rm -f conftest.o conftest.obj
-if { { ac_try="$ac_compile"
+if { (ac_try="$ac_compile"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
   (eval "$ac_compile") 2>&5
   ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; then :
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; then
   for ac_file in conftest.o conftest.obj conftest.*; do
   test -f "$ac_file" || continue;
   case $ac_file in
@@ -3293,23 +2750,31 @@ else
   $as_echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
 
-{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "cannot compute suffix of object files: cannot compile
-See \`config.log' for more details" "$LINENO" 5; }
+{ { $as_echo "$as_me:$LINENO: error: cannot compute suffix of object files: cannot compile
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: cannot compute suffix of object files: cannot compile
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
 fi
+
 rm -f conftest.$ac_cv_objext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_objext" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_objext" >&5
 $as_echo "$ac_cv_objext" >&6; }
 OBJEXT=$ac_cv_objext
 ac_objext=$OBJEXT
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C++ compiler" >&5
+{ $as_echo "$as_me:$LINENO: checking whether we are using the GNU C++ compiler" >&5
 $as_echo_n "checking whether we are using the GNU C++ compiler... " >&6; }
-if ${ac_cv_cxx_compiler_gnu+:} false; then :
+if test "${ac_cv_cxx_compiler_gnu+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3323,16 +2788,37 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_compiler_gnu=yes
 else
-  ac_compiler_gnu=no
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_compiler_gnu=no
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 ac_cv_cxx_compiler_gnu=$ac_compiler_gnu
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_cxx_compiler_gnu" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_cxx_compiler_gnu" >&5
 $as_echo "$ac_cv_cxx_compiler_gnu" >&6; }
 if test $ac_compiler_gnu = yes; then
   GXX=yes
@@ -3341,16 +2827,20 @@ else
 fi
 ac_test_CXXFLAGS=${CXXFLAGS+set}
 ac_save_CXXFLAGS=$CXXFLAGS
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX accepts -g" >&5
+{ $as_echo "$as_me:$LINENO: checking whether $CXX accepts -g" >&5
 $as_echo_n "checking whether $CXX accepts -g... " >&6; }
-if ${ac_cv_prog_cxx_g+:} false; then :
+if test "${ac_cv_prog_cxx_g+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   ac_save_cxx_werror_flag=$ac_cxx_werror_flag
    ac_cxx_werror_flag=yes
    ac_cv_prog_cxx_g=no
    CXXFLAGS="-g"
-   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+   cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3361,11 +2851,35 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_prog_cxx_g=yes
 else
-  CXXFLAGS=""
-      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	CXXFLAGS=""
+      cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3376,12 +2890,36 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  :
 else
-  ac_cxx_werror_flag=$ac_save_cxx_werror_flag
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cxx_werror_flag=$ac_save_cxx_werror_flag
 	 CXXFLAGS="-g"
-	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+	 cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3392,17 +2930,42 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_prog_cxx_g=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
    ac_cxx_werror_flag=$ac_save_cxx_werror_flag
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cxx_g" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cxx_g" >&5
 $as_echo "$ac_cv_prog_cxx_g" >&6; }
 if test "$ac_test_CXXFLAGS" = set; then
   CXXFLAGS=$ac_save_CXXFLAGS
@@ -3436,14 +2999,14 @@ am__doit:
 .PHONY: am__doit
 END
 # If we don't find an include directive, just comment out the code.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for style of include used by $am_make" >&5
+{ $as_echo "$as_me:$LINENO: checking for style of include used by $am_make" >&5
 $as_echo_n "checking for style of include used by $am_make... " >&6; }
 am__include="#"
 am__quote=
 _am_result=none
 # First try GNU make style include.
 echo "include confinc" > confmf
-# Ignore all kinds of additional output from 'make'.
+# Ignore all kinds of additional output from `make'.
 case `$am_make -s -f confmf 2> /dev/null` in #(
 *the\ am__doit\ target*)
   am__include=include
@@ -3464,19 +3027,18 @@ if test "$am__include" = "#"; then
 fi
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $_am_result" >&5
+{ $as_echo "$as_me:$LINENO: result: $_am_result" >&5
 $as_echo "$_am_result" >&6; }
 rm -f confinc confmf
 
 # Check whether --enable-dependency-tracking was given.
-if test "${enable_dependency_tracking+set}" = set; then :
+if test "${enable_dependency_tracking+set}" = set; then
   enableval=$enable_dependency_tracking;
 fi
 
 if test "x$enable_dependency_tracking" != xno; then
   am_depcomp="$ac_aux_dir/depcomp"
   AMDEPBACKSLASH='\'
-  am__nodep='_no'
 fi
  if test "x$enable_dependency_tracking" != xno; then
   AMDEP_TRUE=
@@ -3490,18 +3052,17 @@ fi
 
 depcc="$CXX"  am_compiler_list=
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
+{ $as_echo "$as_me:$LINENO: checking dependency style of $depcc" >&5
 $as_echo_n "checking dependency style of $depcc... " >&6; }
-if ${am_cv_CXX_dependencies_compiler_type+:} false; then :
+if test "${am_cv_CXX_dependencies_compiler_type+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named 'D' -- because '-MD' means "put the output
-  # in D".
-  rm -rf conftest.dir
+  # making a dummy file named `D' -- because `-MD' means `put the output
+  # in D'.
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -3535,16 +3096,16 @@ else
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
-      # Solaris 10 /bin/sh.
-      echo '/* dummy */' > sub/conftst$i.h
+      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
+      # Solaris 8's {/usr,}/bin/sh.
+      touch sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with '-c' and '-o' for the sake of the "dashmstdout"
+    # We check with `-c' and `-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle '-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs.
+    # handle `-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -3553,16 +3114,16 @@ else
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # After this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested.
+      # after this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
-      # This compiler won't grok '-c -o', but also, the minuso test has
+    msvisualcpp | msvcmsys)
+      # This compiler won't grok `-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -3601,7 +3162,7 @@ else
 fi
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CXX_dependencies_compiler_type" >&5
+{ $as_echo "$as_me:$LINENO: result: $am_cv_CXX_dependencies_compiler_type" >&5
 $as_echo "$am_cv_CXX_dependencies_compiler_type" >&6; }
 CXXDEPMODE=depmode=$am_cv_CXX_dependencies_compiler_type
 
@@ -3624,9 +3185,9 @@ ac_compiler_gnu=$ac_cv_c_compiler_gnu
 if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}gcc", so it can be a program name with args.
 set dummy ${ac_tool_prefix}gcc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
+if test "${ac_cv_prog_CC+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3637,24 +3198,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_CC="${ac_tool_prefix}gcc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+  { $as_echo "$as_me:$LINENO: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3664,9 +3225,9 @@ if test -z "$ac_cv_prog_CC"; then
   ac_ct_CC=$CC
   # Extract the first word of "gcc", so it can be a program name with args.
 set dummy gcc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_CC+:} false; then :
+if test "${ac_cv_prog_ac_ct_CC+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CC"; then
@@ -3677,24 +3238,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_ac_ct_CC="gcc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CC=$ac_cv_prog_ac_ct_CC
 if test -n "$ac_ct_CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
+  { $as_echo "$as_me:$LINENO: result: $ac_ct_CC" >&5
 $as_echo "$ac_ct_CC" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3703,7 +3264,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -3717,9 +3278,9 @@ if test -z "$CC"; then
           if test -n "$ac_tool_prefix"; then
     # Extract the first word of "${ac_tool_prefix}cc", so it can be a program name with args.
 set dummy ${ac_tool_prefix}cc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
+if test "${ac_cv_prog_CC+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3730,24 +3291,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_CC="${ac_tool_prefix}cc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+  { $as_echo "$as_me:$LINENO: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3757,9 +3318,9 @@ fi
 if test -z "$CC"; then
   # Extract the first word of "cc", so it can be a program name with args.
 set dummy cc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
+if test "${ac_cv_prog_CC+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3771,18 +3332,18 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     if test "$as_dir/$ac_word$ac_exec_ext" = "/usr/ucb/cc"; then
        ac_prog_rejected=yes
        continue
      fi
     ac_cv_prog_CC="cc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 if test $ac_prog_rejected = yes; then
@@ -3801,10 +3362,10 @@ fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+  { $as_echo "$as_me:$LINENO: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3816,9 +3377,9 @@ if test -z "$CC"; then
   do
     # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
 set dummy $ac_tool_prefix$ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
+if test "${ac_cv_prog_CC+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3829,24 +3390,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_CC="$ac_tool_prefix$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+  { $as_echo "$as_me:$LINENO: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3860,9 +3421,9 @@ if test -z "$CC"; then
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_CC+:} false; then :
+if test "${ac_cv_prog_ac_ct_CC+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CC"; then
@@ -3873,24 +3434,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_ac_ct_CC="$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CC=$ac_cv_prog_ac_ct_CC
 if test -n "$ac_ct_CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
+  { $as_echo "$as_me:$LINENO: result: $ac_ct_CC" >&5
 $as_echo "$ac_ct_CC" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3903,7 +3464,7 @@ done
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -3914,42 +3475,62 @@ fi
 fi
 
 
-test -z "$CC" && { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+test -z "$CC" && { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "no acceptable C compiler found in \$PATH
-See \`config.log' for more details" "$LINENO" 5; }
+{ { $as_echo "$as_me:$LINENO: error: no acceptable C compiler found in \$PATH
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: no acceptable C compiler found in \$PATH
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
 
 # Provide some information about the compiler.
-$as_echo "$as_me:${as_lineno-$LINENO}: checking for C compiler version" >&5
+$as_echo "$as_me:$LINENO: checking for C compiler version" >&5
 set X $ac_compile
 ac_compiler=$2
-for ac_option in --version -v -V -qversion; do
-  { { ac_try="$ac_compiler $ac_option >&5"
+{ (ac_try="$ac_compiler --version >&5"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler --version >&5") 2>&5
   ac_status=$?
-  if test -s conftest.err; then
-    sed '10a\
-... rest of stderr output deleted ...
-         10q' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-  fi
-  rm -f conftest.er1 conftest.err
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }
-done
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
+{ (ac_try="$ac_compiler -v >&5"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler -v >&5") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
+{ (ac_try="$ac_compiler -V >&5"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler -V >&5") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C compiler" >&5
+{ $as_echo "$as_me:$LINENO: checking whether we are using the GNU C compiler" >&5
 $as_echo_n "checking whether we are using the GNU C compiler... " >&6; }
-if ${ac_cv_c_compiler_gnu+:} false; then :
+if test "${ac_cv_c_compiler_gnu+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3963,16 +3544,37 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_compiler_gnu=yes
 else
-  ac_compiler_gnu=no
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_compiler_gnu=no
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 ac_cv_c_compiler_gnu=$ac_compiler_gnu
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_c_compiler_gnu" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_compiler_gnu" >&5
 $as_echo "$ac_cv_c_compiler_gnu" >&6; }
 if test $ac_compiler_gnu = yes; then
   GCC=yes
@@ -3981,16 +3583,20 @@ else
 fi
 ac_test_CFLAGS=${CFLAGS+set}
 ac_save_CFLAGS=$CFLAGS
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC accepts -g" >&5
+{ $as_echo "$as_me:$LINENO: checking whether $CC accepts -g" >&5
 $as_echo_n "checking whether $CC accepts -g... " >&6; }
-if ${ac_cv_prog_cc_g+:} false; then :
+if test "${ac_cv_prog_cc_g+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   ac_save_c_werror_flag=$ac_c_werror_flag
    ac_c_werror_flag=yes
    ac_cv_prog_cc_g=no
    CFLAGS="-g"
-   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+   cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -4001,11 +3607,35 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_prog_cc_g=yes
 else
-  CFLAGS=""
-      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	CFLAGS=""
+      cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -4016,12 +3646,36 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  :
 else
-  ac_c_werror_flag=$ac_save_c_werror_flag
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_c_werror_flag=$ac_save_c_werror_flag
 	 CFLAGS="-g"
-	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+	 cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -4032,17 +3686,42 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_prog_cc_g=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
    ac_c_werror_flag=$ac_save_c_werror_flag
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_g" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cc_g" >&5
 $as_echo "$ac_cv_prog_cc_g" >&6; }
 if test "$ac_test_CFLAGS" = set; then
   CFLAGS=$ac_save_CFLAGS
@@ -4059,18 +3738,23 @@ else
     CFLAGS=
   fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to accept ISO C89" >&5
+{ $as_echo "$as_me:$LINENO: checking for $CC option to accept ISO C89" >&5
 $as_echo_n "checking for $CC option to accept ISO C89... " >&6; }
-if ${ac_cv_prog_cc_c89+:} false; then :
+if test "${ac_cv_prog_cc_c89+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   ac_cv_prog_cc_c89=no
 ac_save_CC=$CC
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <stdarg.h>
 #include <stdio.h>
-struct stat;
+#include <sys/types.h>
+#include <sys/stat.h>
 /* Most of the following tests are stolen from RCS 5.7's src/conf.sh.  */
 struct buf { int x; };
 FILE * (*rcsopen) (struct buf *, struct stat *, int);
@@ -4122,9 +3806,32 @@ for ac_arg in '' -qlanglvl=extc89 -qlanglvl=ansi -std \
 	-Ae "-Aa -D_HPUX_SOURCE" "-Xc -D__EXTENSIONS__"
 do
   CC="$ac_save_CC $ac_arg"
-  if ac_fn_c_try_compile "$LINENO"; then :
+  rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_prog_cc_c89=$ac_arg
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
 fi
+
 rm -f core conftest.err conftest.$ac_objext
   test "x$ac_cv_prog_cc_c89" != "xno" && break
 done
@@ -4135,19 +3842,17 @@ fi
 # AC_CACHE_VAL
 case "x$ac_cv_prog_cc_c89" in
   x)
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: none needed" >&5
+    { $as_echo "$as_me:$LINENO: result: none needed" >&5
 $as_echo "none needed" >&6; } ;;
   xno)
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: unsupported" >&5
+    { $as_echo "$as_me:$LINENO: result: unsupported" >&5
 $as_echo "unsupported" >&6; } ;;
   *)
     CC="$CC $ac_cv_prog_cc_c89"
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_c89" >&5
+    { $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cc_c89" >&5
 $as_echo "$ac_cv_prog_cc_c89" >&6; } ;;
 esac
-if test "x$ac_cv_prog_cc_c89" != xno; then :
 
-fi
 
 ac_ext=c
 ac_cpp='$CPP $CPPFLAGS'
@@ -4155,79 +3860,19 @@ ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
 ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC understands -c and -o together" >&5
-$as_echo_n "checking whether $CC understands -c and -o together... " >&6; }
-if ${am_cv_prog_cc_c_o+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-  # Make sure it works both with $CC and with simple cc.
-  # Following AC_PROG_CC_C_O, we do the test twice because some
-  # compilers refuse to overwrite an existing .o file with -o,
-  # though they will create one.
-  am_cv_prog_cc_c_o=yes
-  for am_i in 1 2; do
-    if { echo "$as_me:$LINENO: $CC -c conftest.$ac_ext -o conftest2.$ac_objext" >&5
-   ($CC -c conftest.$ac_ext -o conftest2.$ac_objext) >&5 2>&5
-   ac_status=$?
-   echo "$as_me:$LINENO: \$? = $ac_status" >&5
-   (exit $ac_status); } \
-         && test -f conftest2.$ac_objext; then
-      : OK
-    else
-      am_cv_prog_cc_c_o=no
-      break
-    fi
-  done
-  rm -f core conftest*
-  unset am_i
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_prog_cc_c_o" >&5
-$as_echo "$am_cv_prog_cc_c_o" >&6; }
-if test "$am_cv_prog_cc_c_o" != yes; then
-   # Losing compiler, so override with the script.
-   # FIXME: It is wrong to rewrite CC.
-   # But if we don't then we get into trouble of one sort or another.
-   # A longer-term fix would be to have automake use am__CC in this case,
-   # and then we could set am__CC="\$(top_srcdir)/compile \$(CC)"
-   CC="$am_aux_dir/compile $CC"
-fi
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-
-
 depcc="$CC"   am_compiler_list=
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
+{ $as_echo "$as_me:$LINENO: checking dependency style of $depcc" >&5
 $as_echo_n "checking dependency style of $depcc... " >&6; }
-if ${am_cv_CC_dependencies_compiler_type+:} false; then :
+if test "${am_cv_CC_dependencies_compiler_type+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named 'D' -- because '-MD' means "put the output
-  # in D".
-  rm -rf conftest.dir
+  # making a dummy file named `D' -- because `-MD' means `put the output
+  # in D'.
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -4261,16 +3906,16 @@ else
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
-      # Solaris 10 /bin/sh.
-      echo '/* dummy */' > sub/conftst$i.h
+      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
+      # Solaris 8's {/usr,}/bin/sh.
+      touch sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with '-c' and '-o' for the sake of the "dashmstdout"
+    # We check with `-c' and `-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle '-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs.
+    # handle `-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -4279,16 +3924,16 @@ else
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # After this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested.
+      # after this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
-      # This compiler won't grok '-c -o', but also, the minuso test has
+    msvisualcpp | msvcmsys)
+      # This compiler won't grok `-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -4327,7 +3972,7 @@ else
 fi
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CC_dependencies_compiler_type" >&5
+{ $as_echo "$as_me:$LINENO: result: $am_cv_CC_dependencies_compiler_type" >&5
 $as_echo "$am_cv_CC_dependencies_compiler_type" >&6; }
 CCDEPMODE=depmode=$am_cv_CC_dependencies_compiler_type
 
@@ -4346,18 +3991,17 @@ fi
 
   OPENMP_CFLAGS=
   # Check whether --enable-openmp was given.
-if test "${enable_openmp+set}" = set; then :
+if test "${enable_openmp+set}" = set; then
   enableval=$enable_openmp;
 fi
 
   if test "$enable_openmp" != no; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to support OpenMP" >&5
+    { $as_echo "$as_me:$LINENO: checking for $CC option to support OpenMP" >&5
 $as_echo_n "checking for $CC option to support OpenMP... " >&6; }
-if ${ac_cv_prog_c_openmp+:} false; then :
+if test "${ac_cv_prog_c_openmp+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
+  cat >conftest.$ac_ext <<_ACEOF
 
 #ifndef _OPENMP
  choke me
@@ -4366,16 +4010,37 @@ else
 int main () { return omp_get_num_threads (); }
 
 _ACEOF
-if ac_fn_c_try_link "$LINENO"; then :
+rm -f conftest.$ac_objext conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_link") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 $as_test_x conftest$ac_exeext
+       }; then
   ac_cv_prog_c_openmp='none needed'
 else
-  ac_cv_prog_c_openmp='unsupported'
-	  	  	  	  	  	  	                                	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp -homp \
-                           -Popenmp --openmp; do
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cv_prog_c_openmp='unsupported'
+	  	  	  	  	  	  	  	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp; do
 	    ac_save_CFLAGS=$CFLAGS
 	    CFLAGS="$CFLAGS $ac_option"
-	    cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
+	    cat >conftest.$ac_ext <<_ACEOF
 
 #ifndef _OPENMP
  choke me
@@ -4384,27 +4049,56 @@ else
 int main () { return omp_get_num_threads (); }
 
 _ACEOF
-if ac_fn_c_try_link "$LINENO"; then :
+rm -f conftest.$ac_objext conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_link") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 $as_test_x conftest$ac_exeext
+       }; then
   ac_cv_prog_c_openmp=$ac_option
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
 fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
+
+rm -rf conftest.dSYM
+rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
+      conftest$ac_exeext conftest.$ac_ext
 	    CFLAGS=$ac_save_CFLAGS
 	    if test "$ac_cv_prog_c_openmp" != unsupported; then
 	      break
 	    fi
 	  done
 fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
+
+rm -rf conftest.dSYM
+rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
+      conftest$ac_exeext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_c_openmp" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_c_openmp" >&5
 $as_echo "$ac_cv_prog_c_openmp" >&6; }
     case $ac_cv_prog_c_openmp in #(
       "none needed" | unsupported)
-	;; #(
+        ;; #(
       *)
-	OPENMP_CFLAGS=$ac_cv_prog_c_openmp ;;
+        OPENMP_CFLAGS=$ac_cv_prog_c_openmp ;;
     esac
   fi
 
@@ -4412,9 +4106,9 @@ $as_echo "$ac_cv_prog_c_openmp" >&6; }
 if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}ranlib", so it can be a program name with args.
 set dummy ${ac_tool_prefix}ranlib; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_RANLIB+:} false; then :
+if test "${ac_cv_prog_RANLIB+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$RANLIB"; then
@@ -4425,24 +4119,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_RANLIB="${ac_tool_prefix}ranlib"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 RANLIB=$ac_cv_prog_RANLIB
 if test -n "$RANLIB"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $RANLIB" >&5
+  { $as_echo "$as_me:$LINENO: result: $RANLIB" >&5
 $as_echo "$RANLIB" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -4452,9 +4146,9 @@ if test -z "$ac_cv_prog_RANLIB"; then
   ac_ct_RANLIB=$RANLIB
   # Extract the first word of "ranlib", so it can be a program name with args.
 set dummy ranlib; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_RANLIB+:} false; then :
+if test "${ac_cv_prog_ac_ct_RANLIB+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_RANLIB"; then
@@ -4465,24 +4159,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_ac_ct_RANLIB="ranlib"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_RANLIB=$ac_cv_prog_ac_ct_RANLIB
 if test -n "$ac_ct_RANLIB"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_RANLIB" >&5
+  { $as_echo "$as_me:$LINENO: result: $ac_ct_RANLIB" >&5
 $as_echo "$ac_ct_RANLIB" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -4491,7 +4185,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -4503,58 +4197,7 @@ fi
 
 
 # Checks for libraries.
-
-
-
-    { $as_echo "$as_me:${as_lineno-$LINENO}: checking for __attribute__((aligned))" >&5
-$as_echo_n "checking for __attribute__((aligned))... " >&6; }
-if ${ax_cv_have_var_attribute_aligned+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-        cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-
-                    int foo __attribute__((aligned(32)));
-
-int
-main ()
-{
-
-
-
-  ;
-  return 0;
-}
-
-_ACEOF
-if ac_fn_c_try_link "$LINENO"; then :
-                                      if test -s conftest.err; then :
-  ax_cv_have_var_attribute_aligned=no
-else
-  ax_cv_have_var_attribute_aligned=yes
-fi
-else
-  ax_cv_have_var_attribute_aligned=no
-fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_var_attribute_aligned" >&5
-$as_echo "$ax_cv_have_var_attribute_aligned" >&6; }
-
-    if test yes = $ax_cv_have_var_attribute_aligned; then :
-
-cat >>confdefs.h <<_ACEOF
-#define HAVE_VAR_ATTRIBUTE_ALIGNED 1
-_ACEOF
-
-fi
-
-
-
+#AX_GCC_VAR_ATTRIBUTE(aligned)
 
 # Checks for header files.
 ac_ext=c
@@ -4562,14 +4205,14 @@ ac_cpp='$CPP $CPPFLAGS'
 ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
 ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking how to run the C preprocessor" >&5
+{ $as_echo "$as_me:$LINENO: checking how to run the C preprocessor" >&5
 $as_echo_n "checking how to run the C preprocessor... " >&6; }
 # On Suns, sometimes $CPP names a directory.
 if test -n "$CPP" && test -d "$CPP"; then
   CPP=
 fi
 if test -z "$CPP"; then
-  if ${ac_cv_prog_CPP+:} false; then :
+  if test "${ac_cv_prog_CPP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
       # Double quotes because CPP needs to be expanded
@@ -4584,7 +4227,11 @@ do
   # <limits.h> exists even on freestanding compilers.
   # On the NeXT, cc -E runs the code through the compiler's parser,
   # not just through cpp. "Syntax error" is here to catch this case.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #ifdef __STDC__
 # include <limits.h>
@@ -4593,34 +4240,78 @@ do
 #endif
 		     Syntax error
 _ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
-
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  :
 else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
   # Broken: fails on valid input.
 continue
 fi
-rm -f conftest.err conftest.i conftest.$ac_ext
+
+rm -f conftest.err conftest.$ac_ext
 
   # OK, works on sane cases.  Now check whether nonexistent headers
   # can be detected and how.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <ac_nonexistent.h>
 _ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
   # Broken: success on invalid input.
 continue
 else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
   # Passes both tests.
 ac_preproc_ok=:
 break
 fi
-rm -f conftest.err conftest.i conftest.$ac_ext
+
+rm -f conftest.err conftest.$ac_ext
 
 done
 # Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
-rm -f conftest.i conftest.err conftest.$ac_ext
-if $ac_preproc_ok; then :
+rm -f conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then
   break
 fi
 
@@ -4632,7 +4323,7 @@ fi
 else
   ac_cv_prog_CPP=$CPP
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $CPP" >&5
+{ $as_echo "$as_me:$LINENO: result: $CPP" >&5
 $as_echo "$CPP" >&6; }
 ac_preproc_ok=false
 for ac_c_preproc_warn_flag in '' yes
@@ -4643,7 +4334,11 @@ do
   # <limits.h> exists even on freestanding compilers.
   # On the NeXT, cc -E runs the code through the compiler's parser,
   # not just through cpp. "Syntax error" is here to catch this case.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #ifdef __STDC__
 # include <limits.h>
@@ -4652,40 +4347,87 @@ do
 #endif
 		     Syntax error
 _ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
-
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  :
 else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
   # Broken: fails on valid input.
 continue
 fi
-rm -f conftest.err conftest.i conftest.$ac_ext
+
+rm -f conftest.err conftest.$ac_ext
 
   # OK, works on sane cases.  Now check whether nonexistent headers
   # can be detected and how.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <ac_nonexistent.h>
 _ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
   # Broken: success on invalid input.
 continue
 else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
   # Passes both tests.
 ac_preproc_ok=:
 break
 fi
-rm -f conftest.err conftest.i conftest.$ac_ext
+
+rm -f conftest.err conftest.$ac_ext
 
 done
 # Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
-rm -f conftest.i conftest.err conftest.$ac_ext
-if $ac_preproc_ok; then :
-
+rm -f conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then
+  :
 else
-  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+  { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "C preprocessor \"$CPP\" fails sanity check
-See \`config.log' for more details" "$LINENO" 5; }
+{ { $as_echo "$as_me:$LINENO: error: C preprocessor \"$CPP\" fails sanity check
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: C preprocessor \"$CPP\" fails sanity check
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
 fi
 
 ac_ext=c
@@ -4695,9 +4437,9 @@ ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for grep that handles long lines and -e" >&5
+{ $as_echo "$as_me:$LINENO: checking for grep that handles long lines and -e" >&5
 $as_echo_n "checking for grep that handles long lines and -e... " >&6; }
-if ${ac_cv_path_GREP+:} false; then :
+if test "${ac_cv_path_GREP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -z "$GREP"; then
@@ -4708,10 +4450,10 @@ for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_prog in grep ggrep; do
+  for ac_prog in grep ggrep; do
     for ac_exec_ext in '' $ac_executable_extensions; do
       ac_path_GREP="$as_dir/$ac_prog$ac_exec_ext"
-      as_fn_executable_p "$ac_path_GREP" || continue
+      { test -f "$ac_path_GREP" && $as_test_x "$ac_path_GREP"; } || continue
 # Check for GNU ac_path_GREP and select it if it is found.
   # Check for GNU $ac_path_GREP
 case `"$ac_path_GREP" --version 2>&1` in
@@ -4728,7 +4470,7 @@ case `"$ac_path_GREP" --version 2>&1` in
     $as_echo 'GREP' >> "conftest.nl"
     "$ac_path_GREP" -e 'GREP$' -e '-(cannot match)-' < "conftest.nl" >"conftest.out" 2>/dev/null || break
     diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
-    as_fn_arith $ac_count + 1 && ac_count=$as_val
+    ac_count=`expr $ac_count + 1`
     if test $ac_count -gt ${ac_path_GREP_max-0}; then
       # Best one so far, save it but keep looking for a better one
       ac_cv_path_GREP="$ac_path_GREP"
@@ -4743,24 +4485,26 @@ esac
       $ac_path_GREP_found && break 3
     done
   done
-  done
+done
 IFS=$as_save_IFS
   if test -z "$ac_cv_path_GREP"; then
-    as_fn_error $? "no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&5
+$as_echo "$as_me: error: no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&2;}
+   { (exit 1); exit 1; }; }
   fi
 else
   ac_cv_path_GREP=$GREP
 fi
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_GREP" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_path_GREP" >&5
 $as_echo "$ac_cv_path_GREP" >&6; }
  GREP="$ac_cv_path_GREP"
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for egrep" >&5
+{ $as_echo "$as_me:$LINENO: checking for egrep" >&5
 $as_echo_n "checking for egrep... " >&6; }
-if ${ac_cv_path_EGREP+:} false; then :
+if test "${ac_cv_path_EGREP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if echo a | $GREP -E '(a|b)' >/dev/null 2>&1
@@ -4774,10 +4518,10 @@ for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_prog in egrep; do
+  for ac_prog in egrep; do
     for ac_exec_ext in '' $ac_executable_extensions; do
       ac_path_EGREP="$as_dir/$ac_prog$ac_exec_ext"
-      as_fn_executable_p "$ac_path_EGREP" || continue
+      { test -f "$ac_path_EGREP" && $as_test_x "$ac_path_EGREP"; } || continue
 # Check for GNU ac_path_EGREP and select it if it is found.
   # Check for GNU $ac_path_EGREP
 case `"$ac_path_EGREP" --version 2>&1` in
@@ -4794,7 +4538,7 @@ case `"$ac_path_EGREP" --version 2>&1` in
     $as_echo 'EGREP' >> "conftest.nl"
     "$ac_path_EGREP" 'EGREP$' < "conftest.nl" >"conftest.out" 2>/dev/null || break
     diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
-    as_fn_arith $ac_count + 1 && ac_count=$as_val
+    ac_count=`expr $ac_count + 1`
     if test $ac_count -gt ${ac_path_EGREP_max-0}; then
       # Best one so far, save it but keep looking for a better one
       ac_cv_path_EGREP="$ac_path_EGREP"
@@ -4809,10 +4553,12 @@ esac
       $ac_path_EGREP_found && break 3
     done
   done
-  done
+done
 IFS=$as_save_IFS
   if test -z "$ac_cv_path_EGREP"; then
-    as_fn_error $? "no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&5
+$as_echo "$as_me: error: no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&2;}
+   { (exit 1); exit 1; }; }
   fi
 else
   ac_cv_path_EGREP=$EGREP
@@ -4820,17 +4566,21 @@ fi
 
    fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_EGREP" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_path_EGREP" >&5
 $as_echo "$ac_cv_path_EGREP" >&6; }
  EGREP="$ac_cv_path_EGREP"
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for ANSI C header files" >&5
+{ $as_echo "$as_me:$LINENO: checking for ANSI C header files" >&5
 $as_echo_n "checking for ANSI C header files... " >&6; }
-if ${ac_cv_header_stdc+:} false; then :
+if test "${ac_cv_header_stdc+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <stdlib.h>
 #include <stdarg.h>
@@ -4845,23 +4595,48 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_header_stdc=yes
 else
-  ac_cv_header_stdc=no
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cv_header_stdc=no
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 
 if test $ac_cv_header_stdc = yes; then
   # SunOS 4.x string.h does not declare mem*, contrary to ANSI.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <string.h>
 
 _ACEOF
 if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
-  $EGREP "memchr" >/dev/null 2>&1; then :
-
+  $EGREP "memchr" >/dev/null 2>&1; then
+  :
 else
   ac_cv_header_stdc=no
 fi
@@ -4871,14 +4646,18 @@ fi
 
 if test $ac_cv_header_stdc = yes; then
   # ISC 2.0.2 stdlib.h does not declare free, contrary to ANSI.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <stdlib.h>
 
 _ACEOF
 if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
-  $EGREP "free" >/dev/null 2>&1; then :
-
+  $EGREP "free" >/dev/null 2>&1; then
+  :
 else
   ac_cv_header_stdc=no
 fi
@@ -4888,10 +4667,14 @@ fi
 
 if test $ac_cv_header_stdc = yes; then
   # /bin/cc in Irix-4.0.5 gets non-ANSI ctype macros unless using -ansi.
-  if test "$cross_compiling" = yes; then :
+  if test "$cross_compiling" = yes; then
   :
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <ctype.h>
 #include <stdlib.h>
@@ -4918,33 +4701,118 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_run "$LINENO"; then :
-
+rm -f conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && { ac_try='./conftest$ac_exeext'
+  { (case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; }; then
+  :
 else
-  ac_cv_header_stdc=no
+  $as_echo "$as_me: program exited with status $ac_status" >&5
+$as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+( exit $ac_status )
+ac_cv_header_stdc=no
 fi
-rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
-  conftest.$ac_objext conftest.beam conftest.$ac_ext
+rm -rf conftest.dSYM
+rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext conftest.$ac_objext conftest.$ac_ext
 fi
 
+
 fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_header_stdc" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_header_stdc" >&5
 $as_echo "$ac_cv_header_stdc" >&6; }
 if test $ac_cv_header_stdc = yes; then
 
-$as_echo "#define STDC_HEADERS 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define STDC_HEADERS 1
+_ACEOF
 
 fi
 
 # On IRIX 5.3, sys/types and inttypes.h are conflicting.
+
+
+
+
+
+
+
+
+
 for ac_header in sys/types.h sys/stat.h stdlib.h string.h memory.h strings.h \
 		  inttypes.h stdint.h unistd.h
-do :
-  as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-ac_fn_c_check_header_compile "$LINENO" "$ac_header" "$as_ac_Header" "$ac_includes_default
-"
-if eval test \"x\$"$as_ac_Header"\" = x"yes"; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  eval "$as_ac_Header=yes"
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	eval "$as_ac_Header=no"
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
 #define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
@@ -4954,36 +4822,453 @@ fi
 done
 
 
+
 for ac_header in stdint.h
-do :
-  ac_fn_c_check_header_mongrel "$LINENO" "stdint.h" "ac_cv_header_stdint_h" "$ac_includes_default"
-if test "x$ac_cv_header_stdint_h" = xyes; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
+$as_echo_n "checking $ac_header usability... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_header_compiler=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_header_compiler=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
+$as_echo_n "checking $ac_header presence... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <$ac_header>
+_ACEOF
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  ac_header_preproc=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_header_preproc=no
+fi
+
+rm -f conftest.err conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
+  yes:no: )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
+    ac_header_preproc=yes
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
+$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
+    ( cat <<\_ASBOX
+## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##
+_ASBOX
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  eval "$as_ac_Header=\$ac_header_preproc"
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+
+fi
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_STDINT_H 1
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
 
 fi
 
 done
 
+
 for ac_header in malloc/malloc.h
-do :
-  ac_fn_c_check_header_mongrel "$LINENO" "malloc/malloc.h" "ac_cv_header_malloc_malloc_h" "$ac_includes_default"
-if test "x$ac_cv_header_malloc_malloc_h" = xyes; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
+$as_echo_n "checking $ac_header usability... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_header_compiler=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_header_compiler=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
+$as_echo_n "checking $ac_header presence... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <$ac_header>
+_ACEOF
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  ac_header_preproc=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_header_preproc=no
+fi
+
+rm -f conftest.err conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
+  yes:no: )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
+    ac_header_preproc=yes
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
+$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
+    ( cat <<\_ASBOX
+## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##
+_ASBOX
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  eval "$as_ac_Header=\$ac_header_preproc"
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+
+fi
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_MALLOC_MALLOC_H 1
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
 
 fi
 
 done
 
+
 for ac_header in malloc.h
-do :
-  ac_fn_c_check_header_mongrel "$LINENO" "malloc.h" "ac_cv_header_malloc_h" "$ac_includes_default"
-if test "x$ac_cv_header_malloc_h" = xyes; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
+$as_echo_n "checking $ac_header usability... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_header_compiler=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_header_compiler=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
+$as_echo_n "checking $ac_header presence... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <$ac_header>
+_ACEOF
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  ac_header_preproc=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_header_preproc=no
+fi
+
+rm -f conftest.err conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
+  yes:no: )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
+    ac_header_preproc=yes
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
+$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
+    ( cat <<\_ASBOX
+## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##
+_ASBOX
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  eval "$as_ac_Header=\$ac_header_preproc"
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+
+fi
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_MALLOC_H 1
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
 
 fi
@@ -4992,9 +5277,102 @@ done
 
 
 # Checks for typedefs, structures, and compiler characteristics.
-ac_fn_c_check_type "$LINENO" "size_t" "ac_cv_type_size_t" "$ac_includes_default"
-if test "x$ac_cv_type_size_t" = xyes; then :
+{ $as_echo "$as_me:$LINENO: checking for size_t" >&5
+$as_echo_n "checking for size_t... " >&6; }
+if test "${ac_cv_type_size_t+set}" = set; then
+  $as_echo_n "(cached) " >&6
+else
+  ac_cv_type_size_t=no
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+if (sizeof (size_t))
+       return 0;
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+if (sizeof ((size_t)))
+	  return 0;
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  :
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
 
+	ac_cv_type_size_t=yes
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_type_size_t" >&5
+$as_echo "$ac_cv_type_size_t" >&6; }
+if test "x$ac_cv_type_size_t" = x""yes; then
+  :
 else
 
 cat >>confdefs.h <<_ACEOF
@@ -5003,12 +5381,75 @@ _ACEOF
 
 fi
 
-ac_fn_c_find_uintX_t "$LINENO" "32" "ac_cv_c_uint32_t"
-case $ac_cv_c_uint32_t in #(
+
+  { $as_echo "$as_me:$LINENO: checking for uint32_t" >&5
+$as_echo_n "checking for uint32_t... " >&6; }
+if test "${ac_cv_c_uint32_t+set}" = set; then
+  $as_echo_n "(cached) " >&6
+else
+  ac_cv_c_uint32_t=no
+     for ac_type in 'uint32_t' 'unsigned int' 'unsigned long int' \
+	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
+       cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+static int test_array [1 - 2 * !(($ac_type) -1 >> (32 - 1) == 1)];
+test_array [0] = 0
+
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  case $ac_type in
+  uint32_t) ac_cv_c_uint32_t=yes ;;
+  *) ac_cv_c_uint32_t=$ac_type ;;
+esac
+
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+       test "$ac_cv_c_uint32_t" != no && break
+     done
+fi
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_uint32_t" >&5
+$as_echo "$ac_cv_c_uint32_t" >&6; }
+  case $ac_cv_c_uint32_t in #(
   no|yes) ;; #(
   *)
 
-$as_echo "#define _UINT32_T 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define _UINT32_T 1
+_ACEOF
 
 
 cat >>confdefs.h <<_ACEOF
@@ -5017,12 +5458,75 @@ _ACEOF
 ;;
   esac
 
-ac_fn_c_find_uintX_t "$LINENO" "64" "ac_cv_c_uint64_t"
-case $ac_cv_c_uint64_t in #(
+
+  { $as_echo "$as_me:$LINENO: checking for uint64_t" >&5
+$as_echo_n "checking for uint64_t... " >&6; }
+if test "${ac_cv_c_uint64_t+set}" = set; then
+  $as_echo_n "(cached) " >&6
+else
+  ac_cv_c_uint64_t=no
+     for ac_type in 'uint64_t' 'unsigned int' 'unsigned long int' \
+	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
+       cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+static int test_array [1 - 2 * !(($ac_type) -1 >> (64 - 1) == 1)];
+test_array [0] = 0
+
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  case $ac_type in
+  uint64_t) ac_cv_c_uint64_t=yes ;;
+  *) ac_cv_c_uint64_t=$ac_type ;;
+esac
+
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+       test "$ac_cv_c_uint64_t" != no && break
+     done
+fi
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_uint64_t" >&5
+$as_echo "$ac_cv_c_uint64_t" >&6; }
+  case $ac_cv_c_uint64_t in #(
   no|yes) ;; #(
   *)
 
-$as_echo "#define _UINT64_T 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define _UINT64_T 1
+_ACEOF
 
 
 cat >>confdefs.h <<_ACEOF
@@ -5033,12 +5537,102 @@ _ACEOF
 
 
 # Checks for library functions.
+
 for ac_func in gettimeofday
-do :
-  ac_fn_c_check_func "$LINENO" "gettimeofday" "ac_cv_func_gettimeofday"
-if test "x$ac_cv_func_gettimeofday" = xyes; then :
+do
+as_ac_var=`$as_echo "ac_cv_func_$ac_func" | $as_tr_sh`
+{ $as_echo "$as_me:$LINENO: checking for $ac_func" >&5
+$as_echo_n "checking for $ac_func... " >&6; }
+if { as_var=$as_ac_var; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+/* Define $ac_func to an innocuous variant, in case <limits.h> declares $ac_func.
+   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
+#define $ac_func innocuous_$ac_func
+
+/* System header to define __stub macros and hopefully few prototypes,
+    which can conflict with char $ac_func (); below.
+    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
+    <limits.h> exists even on freestanding compilers.  */
+
+#ifdef __STDC__
+# include <limits.h>
+#else
+# include <assert.h>
+#endif
+
+#undef $ac_func
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char $ac_func ();
+/* The GNU C library defines this for functions which it implements
+    to always fail with ENOSYS.  Some functions are actually named
+    something starting with __ and the normal name is an alias.  */
+#if defined __stub_$ac_func || defined __stub___$ac_func
+choke me
+#endif
+
+int
+main ()
+{
+return $ac_func ();
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_link") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 $as_test_x conftest$ac_exeext
+       }; then
+  eval "$as_ac_var=yes"
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	eval "$as_ac_var=no"
+fi
+
+rm -rf conftest.dSYM
+rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
+      conftest$ac_exeext conftest.$ac_ext
+fi
+ac_res=`eval 'as_val=${'$as_ac_var'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+as_val=`eval 'as_val=${'$as_ac_var'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_GETTIMEOFDAY 1
+#define `$as_echo "HAVE_$ac_func" | $as_tr_cpp` 1
 _ACEOF
 
 fi
@@ -5046,7 +5640,7 @@ done
 
 
 # Check whether --enable-simd was given.
-if test "${enable_simd+set}" = set; then :
+if test "${enable_simd+set}" = set; then
   enableval=$enable_simd; ac_SIMD=${enable_simd}
 else
   ac_SIMD=AVX2
@@ -5057,29 +5651,39 @@ case ${ac_SIMD} in
      SSE2)
        echo Configuring for SSE2
 
-$as_echo "#define SSE2 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define SSE2 1
+_ACEOF
 
      ;;
      AVX)
        echo Configuring for AVX
 
-$as_echo "#define AVX1 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define AVX1 1
+_ACEOF
 
      ;;
      AVX2)
        echo Configuring for AVX2
 
-$as_echo "#define AVX2 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define AVX2 1
+_ACEOF
 
      ;;
      AVX512)
        echo Configuring for AVX512
 
-$as_echo "#define AVX512 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define AVX512 1
+_ACEOF
 
      ;;
      *)
-     as_fn_error $? "${ac_SIMD} unsupported --enable-simd option" "$LINENO" 5;
+     { { $as_echo "$as_me:$LINENO: error: ${ac_SIMD} unsupported --enable-simd option" >&5
+$as_echo "$as_me: error: ${ac_SIMD} unsupported --enable-simd option" >&2;}
+   { (exit 1); exit 1; }; };
      ;;
 esac
 
@@ -5112,13 +5716,13 @@ _ACEOF
     case $ac_val in #(
     *${as_nl}*)
       case $ac_var in #(
-      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
+      *_cv_*) { $as_echo "$as_me:$LINENO: WARNING: cache variable $ac_var contains a newline" >&5
 $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
       esac
       case $ac_var in #(
       _ | IFS | as_nl) ;; #(
       BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
-      *) { eval $ac_var=; unset $ac_var;} ;;
+      *) $as_unset $ac_var ;;
       esac ;;
     esac
   done
@@ -5126,8 +5730,8 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
   (set) 2>&1 |
     case $as_nl`(ac_space=' '; set) 2>&1` in #(
     *${as_nl}ac_space=\ *)
-      # `set' does not quote correctly, so add quotes: double-quote
-      # substitution turns \\\\ into \\, and sed turns \\ into \.
+      # `set' does not quote correctly, so add quotes (double-quote
+      # substitution turns \\\\ into \\, and sed turns \\ into \).
       sed -n \
 	"s/'/'\\\\''/g;
 	  s/^\\([_$as_cr_alnum]*_cv_[_$as_cr_alnum]*\\)=\\(.*\\)/\\1='\\2'/p"
@@ -5149,23 +5753,12 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
      :end' >>confcache
 if diff "$cache_file" confcache >/dev/null 2>&1; then :; else
   if test -w "$cache_file"; then
-    if test "x$cache_file" != "x/dev/null"; then
-      { $as_echo "$as_me:${as_lineno-$LINENO}: updating cache $cache_file" >&5
+    test "x$cache_file" != "x/dev/null" &&
+      { $as_echo "$as_me:$LINENO: updating cache $cache_file" >&5
 $as_echo "$as_me: updating cache $cache_file" >&6;}
-      if test ! -f "$cache_file" || test -h "$cache_file"; then
-	cat confcache >"$cache_file"
-      else
-        case $cache_file in #(
-        */* | ?:*)
-	  mv -f confcache "$cache_file"$$ &&
-	  mv -f "$cache_file"$$ "$cache_file" ;; #(
-        *)
-	  mv -f confcache "$cache_file" ;;
-	esac
-      fi
-    fi
+    cat confcache >$cache_file
   else
-    { $as_echo "$as_me:${as_lineno-$LINENO}: not updating unwritable cache $cache_file" >&5
+    { $as_echo "$as_me:$LINENO: not updating unwritable cache $cache_file" >&5
 $as_echo "$as_me: not updating unwritable cache $cache_file" >&6;}
   fi
 fi
@@ -5179,29 +5772,20 @@ DEFS=-DHAVE_CONFIG_H
 
 ac_libobjs=
 ac_ltlibobjs=
-U=
 for ac_i in : $LIBOBJS; do test "x$ac_i" = x: && continue
   # 1. Remove the extension, and $U if already installed.
   ac_script='s/\$U\././;s/\.o$//;s/\.obj$//'
   ac_i=`$as_echo "$ac_i" | sed "$ac_script"`
   # 2. Prepend LIBOBJDIR.  When used with automake>=1.10 LIBOBJDIR
   #    will be set to the directory where LIBOBJS objects are built.
-  as_fn_append ac_libobjs " \${LIBOBJDIR}$ac_i\$U.$ac_objext"
-  as_fn_append ac_ltlibobjs " \${LIBOBJDIR}$ac_i"'$U.lo'
+  ac_libobjs="$ac_libobjs \${LIBOBJDIR}$ac_i\$U.$ac_objext"
+  ac_ltlibobjs="$ac_ltlibobjs \${LIBOBJDIR}$ac_i"'$U.lo'
 done
 LIBOBJS=$ac_libobjs
 
 LTLIBOBJS=$ac_ltlibobjs
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking that generated files are newer than configure" >&5
-$as_echo_n "checking that generated files are newer than configure... " >&6; }
-   if test -n "$am_sleep_pid"; then
-     # Hide warnings about reused PIDs.
-     wait $am_sleep_pid 2>/dev/null
-   fi
-   { $as_echo "$as_me:${as_lineno-$LINENO}: result: done" >&5
-$as_echo "done" >&6; }
  if test -n "$EXEEXT"; then
   am__EXEEXT_TRUE=
   am__EXEEXT_FALSE='#'
@@ -5211,26 +5795,34 @@ else
 fi
 
 if test -z "${AMDEP_TRUE}" && test -z "${AMDEP_FALSE}"; then
-  as_fn_error $? "conditional \"AMDEP\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: conditional \"AMDEP\" was never defined.
+Usually this means the macro was only invoked conditionally." >&5
+$as_echo "$as_me: error: conditional \"AMDEP\" was never defined.
+Usually this means the macro was only invoked conditionally." >&2;}
+   { (exit 1); exit 1; }; }
 fi
 if test -z "${am__fastdepCXX_TRUE}" && test -z "${am__fastdepCXX_FALSE}"; then
-  as_fn_error $? "conditional \"am__fastdepCXX\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: conditional \"am__fastdepCXX\" was never defined.
+Usually this means the macro was only invoked conditionally." >&5
+$as_echo "$as_me: error: conditional \"am__fastdepCXX\" was never defined.
+Usually this means the macro was only invoked conditionally." >&2;}
+   { (exit 1); exit 1; }; }
 fi
 if test -z "${am__fastdepCC_TRUE}" && test -z "${am__fastdepCC_FALSE}"; then
-  as_fn_error $? "conditional \"am__fastdepCC\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: conditional \"am__fastdepCC\" was never defined.
+Usually this means the macro was only invoked conditionally." >&5
+$as_echo "$as_me: error: conditional \"am__fastdepCC\" was never defined.
+Usually this means the macro was only invoked conditionally." >&2;}
+   { (exit 1); exit 1; }; }
 fi
 
-: "${CONFIG_STATUS=./config.status}"
+: ${CONFIG_STATUS=./config.status}
 ac_write_fail=0
 ac_clean_files_save=$ac_clean_files
 ac_clean_files="$ac_clean_files $CONFIG_STATUS"
-{ $as_echo "$as_me:${as_lineno-$LINENO}: creating $CONFIG_STATUS" >&5
+{ $as_echo "$as_me:$LINENO: creating $CONFIG_STATUS" >&5
 $as_echo "$as_me: creating $CONFIG_STATUS" >&6;}
-as_write_fail=0
-cat >$CONFIG_STATUS <<_ASEOF || as_write_fail=1
+cat >$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 #! $SHELL
 # Generated by $as_me.
 # Run this file to recreate the current configuration.
@@ -5240,18 +5832,17 @@ cat >$CONFIG_STATUS <<_ASEOF || as_write_fail=1
 debug=false
 ac_cs_recheck=false
 ac_cs_silent=false
-
 SHELL=\${CONFIG_SHELL-$SHELL}
-export SHELL
-_ASEOF
-cat >>$CONFIG_STATUS <<\_ASEOF || as_write_fail=1
-## -------------------- ##
-## M4sh Initialization. ##
-## -------------------- ##
+_ACEOF
+
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+## --------------------- ##
+## M4sh Initialization.  ##
+## --------------------- ##
 
 # Be more Bourne compatible
 DUALCASE=1; export DUALCASE # for MKS sh
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
   emulate sh
   NULLCMD=:
   # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
@@ -5259,15 +5850,23 @@ if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
   alias -g '${1+"$@"}'='"$@"'
   setopt NO_GLOB_SUBST
 else
-  case `(set -o) 2>/dev/null` in #(
-  *posix*) :
-    set -o posix ;; #(
-  *) :
-     ;;
+  case `(set -o) 2>/dev/null` in
+  *posix*) set -o posix ;;
 esac
+
 fi
 
 
+
+
+# PATH needs CR
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
+
 as_nl='
 '
 export as_nl
@@ -5275,13 +5874,7 @@ export as_nl
 as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
-# Prefer a ksh shell builtin over an external printf program on Solaris,
-# but without wasting forks for bash or zsh.
-if test -z "$BASH_VERSION$ZSH_VERSION" \
-    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
-  as_echo='print -r --'
-  as_echo_n='print -rn --'
-elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+if (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
   as_echo='printf %s\n'
   as_echo_n='printf %s'
 else
@@ -5292,7 +5885,7 @@ else
     as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
     as_echo_n_body='eval
       arg=$1;
-      case $arg in #(
+      case $arg in
       *"$as_nl"*)
 	expr "X$arg" : "X\\(.*\\)$as_nl";
 	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
@@ -5315,6 +5908,13 @@ if test "${PATH_SEPARATOR+set}" != set; then
   }
 fi
 
+# Support unset when possible.
+if ( (MAIL=60; unset MAIL) || exit) >/dev/null 2>&1; then
+  as_unset=unset
+else
+  as_unset=false
+fi
+
 
 # IFS
 # We need space, tab and new line, in precisely that order.  Quoting is
@@ -5324,16 +5924,15 @@ fi
 IFS=" ""	$as_nl"
 
 # Find who we are.  Look in the path if we contain no directory separator.
-as_myself=
-case $0 in #((
+case $0 in
   *[\\/]* ) as_myself=$0 ;;
   *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
 for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
-  done
+  test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+done
 IFS=$as_save_IFS
 
      ;;
@@ -5345,16 +5944,12 @@ if test "x$as_myself" = x; then
 fi
 if test ! -f "$as_myself"; then
   $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
-  exit 1
+  { (exit 1); exit 1; }
 fi
 
-# Unset variables that we do not need and which cause bugs (e.g. in
-# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
-# suppresses any "Segmentation fault" message there.  '((' could
-# trigger a bug in pdksh 5.2.14.
-for as_var in BASH_ENV ENV MAIL MAILPATH
-do eval test x\${$as_var+set} = xset \
-  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
+# Work around bugs in pre-3.0 UWIN ksh.
+for as_var in ENV MAIL MAILPATH
+do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
 done
 PS1='$ '
 PS2='> '
@@ -5366,89 +5961,7 @@ export LC_ALL
 LANGUAGE=C
 export LANGUAGE
 
-# CDPATH.
-(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
-
-
-# as_fn_error STATUS ERROR [LINENO LOG_FD]
-# ----------------------------------------
-# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
-# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
-# script with STATUS, using 1 if that was 0.
-as_fn_error ()
-{
-  as_status=$1; test $as_status -eq 0 && as_status=1
-  if test "$4"; then
-    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
-  fi
-  $as_echo "$as_me: error: $2" >&2
-  as_fn_exit $as_status
-} # as_fn_error
-
-
-# as_fn_set_status STATUS
-# -----------------------
-# Set $? to STATUS, without forking.
-as_fn_set_status ()
-{
-  return $1
-} # as_fn_set_status
-
-# as_fn_exit STATUS
-# -----------------
-# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
-as_fn_exit ()
-{
-  set +e
-  as_fn_set_status $1
-  exit $1
-} # as_fn_exit
-
-# as_fn_unset VAR
-# ---------------
-# Portably unset VAR.
-as_fn_unset ()
-{
-  { eval $1=; unset $1;}
-}
-as_unset=as_fn_unset
-# as_fn_append VAR VALUE
-# ----------------------
-# Append the text in VALUE to the end of the definition contained in VAR. Take
-# advantage of any shell optimizations that allow amortized linear growth over
-# repeated appends, instead of the typical quadratic growth present in naive
-# implementations.
-if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
-  eval 'as_fn_append ()
-  {
-    eval $1+=\$2
-  }'
-else
-  as_fn_append ()
-  {
-    eval $1=\$$1\$2
-  }
-fi # as_fn_append
-
-# as_fn_arith ARG...
-# ------------------
-# Perform arithmetic evaluation on the ARGs, and store the result in the
-# global $as_val. Take advantage of shells that can avoid forks. The arguments
-# must be portable across $(()) and expr.
-if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
-  eval 'as_fn_arith ()
-  {
-    as_val=$(( $* ))
-  }'
-else
-  as_fn_arith ()
-  {
-    as_val=`expr "$@" || test $? -eq 1`
-  }
-fi # as_fn_arith
-
-
+# Required to use basename.
 if expr a : '\(a\)' >/dev/null 2>&1 &&
    test "X`expr 00001 : '.*\(...\)'`" = X001; then
   as_expr=expr
@@ -5462,12 +5975,8 @@ else
   as_basename=false
 fi
 
-if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
-  as_dirname=dirname
-else
-  as_dirname=false
-fi
 
+# Name of the executable.
 as_me=`$as_basename -- "$0" ||
 $as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
 	 X"$0" : 'X\(//\)$' \| \
@@ -5487,25 +5996,76 @@ $as_echo X/"$0" |
 	  }
 	  s/.*/./; q'`
 
-# Avoid depending upon Character Ranges.
-as_cr_letters='abcdefghijklmnopqrstuvwxyz'
-as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
-as_cr_Letters=$as_cr_letters$as_cr_LETTERS
-as_cr_digits='0123456789'
-as_cr_alnum=$as_cr_Letters$as_cr_digits
+# CDPATH.
+$as_unset CDPATH
+
+
+
+  as_lineno_1=$LINENO
+  as_lineno_2=$LINENO
+  test "x$as_lineno_1" != "x$as_lineno_2" &&
+  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2" || {
+
+  # Create $as_me.lineno as a copy of $as_myself, but with $LINENO
+  # uniformly replaced by the line number.  The first 'sed' inserts a
+  # line-number line after each line using $LINENO; the second 'sed'
+  # does the real work.  The second script uses 'N' to pair each
+  # line-number line with the line containing $LINENO, and appends
+  # trailing '-' during substitution so that $LINENO is not a special
+  # case at line end.
+  # (Raja R Harinath suggested sed '=', and Paul Eggert wrote the
+  # scripts with optimization help from Paolo Bonzini.  Blame Lee
+  # E. McMahon (1931-1989) for sed's syntax.  :-)
+  sed -n '
+    p
+    /[$]LINENO/=
+  ' <$as_myself |
+    sed '
+      s/[$]LINENO.*/&-/
+      t lineno
+      b
+      :lineno
+      N
+      :loop
+      s/[$]LINENO\([^'$as_cr_alnum'_].*\n\)\(.*\)/\2\1\2/
+      t loop
+      s/-\n.*//
+    ' >$as_me.lineno &&
+  chmod +x "$as_me.lineno" ||
+    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2
+   { (exit 1); exit 1; }; }
+
+  # Don't try to exec as it changes $[0], causing all sort of problems
+  # (the dirname of $[0] is not the place where we might find the
+  # original and so on.  Autoconf is especially sensitive to this).
+  . "./$as_me.lineno"
+  # Exit status is that of the last command.
+  exit
+}
+
+
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
 
 ECHO_C= ECHO_N= ECHO_T=
-case `echo -n x` in #(((((
+case `echo -n x` in
 -n*)
-  case `echo 'xy\c'` in
+  case `echo 'x\c'` in
   *c*) ECHO_T='	';;	# ECHO_T is single tab character.
-  xy)  ECHO_C='\c';;
-  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
-       ECHO_T='	';;
+  *)   ECHO_C='\c';;
   esac;;
 *)
   ECHO_N='-n';;
 esac
+if expr a : '\(a\)' >/dev/null 2>&1 &&
+   test "X`expr 00001 : '.*\(...\)'`" = X001; then
+  as_expr=expr
+else
+  as_expr=false
+fi
 
 rm -f conf$$ conf$$.exe conf$$.file
 if test -d conf$$.dir; then
@@ -5520,85 +6080,49 @@ if (echo >conf$$.file) 2>/dev/null; then
     # ... but there are two gotchas:
     # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
     # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
-    # In both cases, we have to default to `cp -pR'.
+    # In both cases, we have to default to `cp -p'.
     ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
-      as_ln_s='cp -pR'
+      as_ln_s='cp -p'
   elif ln conf$$.file conf$$ 2>/dev/null; then
     as_ln_s=ln
   else
-    as_ln_s='cp -pR'
+    as_ln_s='cp -p'
   fi
 else
-  as_ln_s='cp -pR'
+  as_ln_s='cp -p'
 fi
 rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
 rmdir conf$$.dir 2>/dev/null
 
-
-# as_fn_mkdir_p
-# -------------
-# Create "$as_dir" as a directory, including parents if necessary.
-as_fn_mkdir_p ()
-{
-
-  case $as_dir in #(
-  -*) as_dir=./$as_dir;;
-  esac
-  test -d "$as_dir" || eval $as_mkdir_p || {
-    as_dirs=
-    while :; do
-      case $as_dir in #(
-      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
-      *) as_qdir=$as_dir;;
-      esac
-      as_dirs="'$as_qdir' $as_dirs"
-      as_dir=`$as_dirname -- "$as_dir" ||
-$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_dir" : 'X\(//\)[^/]' \| \
-	 X"$as_dir" : 'X\(//\)$' \| \
-	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_dir" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      test -d "$as_dir" && break
-    done
-    test -z "$as_dirs" || eval "mkdir $as_dirs"
-  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
-
-
-} # as_fn_mkdir_p
 if mkdir -p . 2>/dev/null; then
-  as_mkdir_p='mkdir -p "$as_dir"'
+  as_mkdir_p=:
 else
   test -d ./-p && rmdir ./-p
   as_mkdir_p=false
 fi
 
-
-# as_fn_executable_p FILE
-# -----------------------
-# Test if FILE is an executable regular file.
-as_fn_executable_p ()
-{
-  test -f "$1" && test -x "$1"
-} # as_fn_executable_p
-as_test_x='test -x'
-as_executable_p=as_fn_executable_p
+if test -x / >/dev/null 2>&1; then
+  as_test_x='test -x'
+else
+  if ls -dL / >/dev/null 2>&1; then
+    as_ls_L_option=L
+  else
+    as_ls_L_option=
+  fi
+  as_test_x='
+    eval sh -c '\''
+      if test -d "$1"; then
+	test -d "$1/.";
+      else
+	case $1 in
+	-*)set "./$1";;
+	esac;
+	case `ls -ld'$as_ls_L_option' "$1" 2>/dev/null` in
+	???[sx]*):;;*)false;;esac;fi
+    '\'' sh
+  '
+fi
+as_executable_p=$as_test_x
 
 # Sed expression to map a string onto a valid CPP name.
 as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
@@ -5608,19 +6132,13 @@ as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
 
 
 exec 6>&1
-## ----------------------------------- ##
-## Main body of $CONFIG_STATUS script. ##
-## ----------------------------------- ##
-_ASEOF
-test $as_write_fail = 0 && chmod +x $CONFIG_STATUS || ac_write_fail=1
 
-cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-# Save the log message, to keep $0 and so on meaningful, and to
+# Save the log message, to keep $[0] and so on meaningful, and to
 # report actual input values of CONFIG_FILES etc. instead of their
 # values after options handling.
 ac_log="
 This file was extended by Grid $as_me 1.0, which was
-generated by GNU Autoconf 2.69.  Invocation command line was
+generated by GNU Autoconf 2.63.  Invocation command line was
 
   CONFIG_FILES    = $CONFIG_FILES
   CONFIG_HEADERS  = $CONFIG_HEADERS
@@ -5652,15 +6170,13 @@ _ACEOF
 
 cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 ac_cs_usage="\
-\`$as_me' instantiates files and other configuration actions
-from templates according to the current configuration.  Unless the files
-and actions are specified as TAGs, all are instantiated by default.
+\`$as_me' instantiates files from templates according to the
+current configuration.
 
-Usage: $0 [OPTION]... [TAG]...
+Usage: $0 [OPTION]... [FILE]...
 
   -h, --help       print this help, then exit
   -V, --version    print version number and configuration settings, then exit
-      --config     print configuration, then exit
   -q, --quiet, --silent
                    do not print progress messages
   -d, --debug      don't remove temporary files
@@ -5679,17 +6195,16 @@ $config_headers
 Configuration commands:
 $config_commands
 
-Report bugs to <paboyle@ph.ed.ac.uk>."
+Report bugs to <bug-autoconf@gnu.org>."
 
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-ac_cs_config="`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`"
 ac_cs_version="\\
 Grid config.status 1.0
-configured by $0, generated by GNU Autoconf 2.69,
-  with options \\"\$ac_cs_config\\"
+configured by $0, generated by GNU Autoconf 2.63,
+  with options \\"`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`\\"
 
-Copyright (C) 2012 Free Software Foundation, Inc.
+Copyright (C) 2008 Free Software Foundation, Inc.
 This config.status script is free software; the Free Software Foundation
 gives unlimited permission to copy, distribute and modify it."
 
@@ -5707,16 +6222,11 @@ ac_need_defaults=:
 while test $# != 0
 do
   case $1 in
-  --*=?*)
+  --*=*)
     ac_option=`expr "X$1" : 'X\([^=]*\)='`
     ac_optarg=`expr "X$1" : 'X[^=]*=\(.*\)'`
     ac_shift=:
     ;;
-  --*=)
-    ac_option=`expr "X$1" : 'X\([^=]*\)='`
-    ac_optarg=
-    ac_shift=:
-    ;;
   *)
     ac_option=$1
     ac_optarg=$2
@@ -5730,29 +6240,27 @@ do
     ac_cs_recheck=: ;;
   --version | --versio | --versi | --vers | --ver | --ve | --v | -V )
     $as_echo "$ac_cs_version"; exit ;;
-  --config | --confi | --conf | --con | --co | --c )
-    $as_echo "$ac_cs_config"; exit ;;
   --debug | --debu | --deb | --de | --d | -d )
     debug=: ;;
   --file | --fil | --fi | --f )
     $ac_shift
     case $ac_optarg in
     *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
-    '') as_fn_error $? "missing file argument" ;;
     esac
-    as_fn_append CONFIG_FILES " '$ac_optarg'"
+    CONFIG_FILES="$CONFIG_FILES '$ac_optarg'"
     ac_need_defaults=false;;
   --header | --heade | --head | --hea )
     $ac_shift
     case $ac_optarg in
     *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
     esac
-    as_fn_append CONFIG_HEADERS " '$ac_optarg'"
+    CONFIG_HEADERS="$CONFIG_HEADERS '$ac_optarg'"
     ac_need_defaults=false;;
   --he | --h)
     # Conflict between --help and --header
-    as_fn_error $? "ambiguous option: \`$1'
-Try \`$0 --help' for more information.";;
+    { $as_echo "$as_me: error: ambiguous option: $1
+Try \`$0 --help' for more information." >&2
+   { (exit 1); exit 1; }; };;
   --help | --hel | -h )
     $as_echo "$ac_cs_usage"; exit ;;
   -q | -quiet | --quiet | --quie | --qui | --qu | --q \
@@ -5760,10 +6268,11 @@ Try \`$0 --help' for more information.";;
     ac_cs_silent=: ;;
 
   # This is an error.
-  -*) as_fn_error $? "unrecognized option: \`$1'
-Try \`$0 --help' for more information." ;;
+  -*) { $as_echo "$as_me: error: unrecognized option: $1
+Try \`$0 --help' for more information." >&2
+   { (exit 1); exit 1; }; } ;;
 
-  *) as_fn_append ac_config_targets " $1"
+  *) ac_config_targets="$ac_config_targets $1"
      ac_need_defaults=false ;;
 
   esac
@@ -5780,7 +6289,7 @@ fi
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 if \$ac_cs_recheck; then
-  set X $SHELL '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
+  set X '$SHELL' '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
   shift
   \$as_echo "running CONFIG_SHELL=$SHELL \$*" >&6
   CONFIG_SHELL='$SHELL'
@@ -5818,7 +6327,9 @@ do
     "depfiles") CONFIG_COMMANDS="$CONFIG_COMMANDS depfiles" ;;
     "Makefile") CONFIG_FILES="$CONFIG_FILES Makefile" ;;
 
-  *) as_fn_error $? "invalid argument: \`$ac_config_target'" "$LINENO" 5;;
+  *) { { $as_echo "$as_me:$LINENO: error: invalid argument: $ac_config_target" >&5
+$as_echo "$as_me: error: invalid argument: $ac_config_target" >&2;}
+   { (exit 1); exit 1; }; };;
   esac
 done
 
@@ -5841,24 +6352,26 @@ fi
 # after its creation but before its name has been assigned to `$tmp'.
 $debug ||
 {
-  tmp= ac_tmp=
+  tmp=
   trap 'exit_status=$?
-  : "${ac_tmp:=$tmp}"
-  { test ! -d "$ac_tmp" || rm -fr "$ac_tmp"; } && exit $exit_status
+  { test -z "$tmp" || test ! -d "$tmp" || rm -fr "$tmp"; } && exit $exit_status
 ' 0
-  trap 'as_fn_exit 1' 1 2 13 15
+  trap '{ (exit 1); exit 1; }' 1 2 13 15
 }
 # Create a (secure) tmp directory for tmp files.
 
 {
   tmp=`(umask 077 && mktemp -d "./confXXXXXX") 2>/dev/null` &&
-  test -d "$tmp"
+  test -n "$tmp" && test -d "$tmp"
 }  ||
 {
   tmp=./conf$$-$RANDOM
   (umask 077 && mkdir "$tmp")
-} || as_fn_error $? "cannot create a temporary directory in ." "$LINENO" 5
-ac_tmp=$tmp
+} ||
+{
+   $as_echo "$as_me: cannot create a temporary directory in ." >&2
+   { (exit 1); exit 1; }
+}
 
 # Set up the scripts for CONFIG_FILES section.
 # No need to generate them if there are no CONFIG_FILES.
@@ -5866,13 +6379,7 @@ ac_tmp=$tmp
 if test -n "$CONFIG_FILES"; then
 
 
-ac_cr=`echo X | tr X '\015'`
-# On cygwin, bash can eat \r inside `` if the user requested igncr.
-# But we know of no other shell where ac_cr would be empty at this
-# point, so we can use a bashism as a fallback.
-if test "x$ac_cr" = x; then
-  eval ac_cr=\$\'\\r\'
-fi
+ac_cr='
'
 ac_cs_awk_cr=`$AWK 'BEGIN { print "a\rb" }' </dev/null 2>/dev/null`
 if test "$ac_cs_awk_cr" = "a${ac_cr}b"; then
   ac_cs_awk_cr='\\r'
@@ -5880,7 +6387,7 @@ else
   ac_cs_awk_cr=$ac_cr
 fi
 
-echo 'BEGIN {' >"$ac_tmp/subs1.awk" &&
+echo 'BEGIN {' >"$tmp/subs1.awk" &&
 _ACEOF
 
 
@@ -5889,18 +6396,24 @@ _ACEOF
   echo "$ac_subst_vars" | sed 's/.*/&!$&$ac_delim/' &&
   echo "_ACEOF"
 } >conf$$subs.sh ||
-  as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
-ac_delim_num=`echo "$ac_subst_vars" | grep -c '^'`
+  { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
+$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
+   { (exit 1); exit 1; }; }
+ac_delim_num=`echo "$ac_subst_vars" | grep -c '$'`
 ac_delim='%!_!# '
 for ac_last_try in false false false false false :; do
   . ./conf$$subs.sh ||
-    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
+$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
+   { (exit 1); exit 1; }; }
 
   ac_delim_n=`sed -n "s/.*$ac_delim\$/X/p" conf$$subs.awk | grep -c X`
   if test $ac_delim_n = $ac_delim_num; then
     break
   elif $ac_last_try; then
-    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
+$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
+   { (exit 1); exit 1; }; }
   else
     ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
   fi
@@ -5908,7 +6421,7 @@ done
 rm -f conf$$subs.sh
 
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-cat >>"\$ac_tmp/subs1.awk" <<\\_ACAWK &&
+cat >>"\$tmp/subs1.awk" <<\\_ACAWK &&
 _ACEOF
 sed -n '
 h
@@ -5922,7 +6435,7 @@ s/'"$ac_delim"'$//
 t delim
 :nl
 h
-s/\(.\{148\}\)..*/\1/
+s/\(.\{148\}\).*/\1/
 t more1
 s/["\\]/\\&/g; s/^/"/; s/$/\\n"\\/
 p
@@ -5936,7 +6449,7 @@ s/.\{148\}//
 t nl
 :delim
 h
-s/\(.\{148\}\)..*/\1/
+s/\(.\{148\}\).*/\1/
 t more2
 s/["\\]/\\&/g; s/^/"/; s/$/"/
 p
@@ -5956,7 +6469,7 @@ t delim
 rm -f conf$$subs.awk
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 _ACAWK
-cat >>"\$ac_tmp/subs1.awk" <<_ACAWK &&
+cat >>"\$tmp/subs1.awk" <<_ACAWK &&
   for (key in S) S_is_set[key] = 1
   FS = ""
 
@@ -5988,29 +6501,23 @@ if sed "s/$ac_cr//" < /dev/null > /dev/null 2>&1; then
   sed "s/$ac_cr\$//; s/$ac_cr/$ac_cs_awk_cr/g"
 else
   cat
-fi < "$ac_tmp/subs1.awk" > "$ac_tmp/subs.awk" \
-  || as_fn_error $? "could not setup config files machinery" "$LINENO" 5
+fi < "$tmp/subs1.awk" > "$tmp/subs.awk" \
+  || { { $as_echo "$as_me:$LINENO: error: could not setup config files machinery" >&5
+$as_echo "$as_me: error: could not setup config files machinery" >&2;}
+   { (exit 1); exit 1; }; }
 _ACEOF
 
-# VPATH may cause trouble with some makes, so we remove sole $(srcdir),
-# ${srcdir} and @srcdir@ entries from VPATH if srcdir is ".", strip leading and
+# VPATH may cause trouble with some makes, so we remove $(srcdir),
+# ${srcdir} and @srcdir@ from VPATH if srcdir is ".", strip leading and
 # trailing colons and then remove the whole line if VPATH becomes empty
 # (actually we leave an empty line to preserve line numbers).
 if test "x$srcdir" = x.; then
-  ac_vpsub='/^[	 ]*VPATH[	 ]*=[	 ]*/{
-h
-s///
-s/^/:/
-s/[	 ]*$/:/
-s/:\$(srcdir):/:/g
-s/:\${srcdir}:/:/g
-s/:@srcdir@:/:/g
-s/^:*//
+  ac_vpsub='/^[	 ]*VPATH[	 ]*=/{
+s/:*\$(srcdir):*/:/
+s/:*\${srcdir}:*/:/
+s/:*@srcdir@:*/:/
+s/^\([^=]*=[	 ]*\):*/\1/
 s/:*$//
-x
-s/\(=[	 ]*\).*/\1/
-G
-s/\n//
 s/^[^=]*=[	 ]*$//
 }'
 fi
@@ -6022,7 +6529,7 @@ fi # test -n "$CONFIG_FILES"
 # No need to generate them if there are no CONFIG_HEADERS.
 # This happens for instance with `./config.status Makefile'.
 if test -n "$CONFIG_HEADERS"; then
-cat >"$ac_tmp/defines.awk" <<\_ACAWK ||
+cat >"$tmp/defines.awk" <<\_ACAWK ||
 BEGIN {
 _ACEOF
 
@@ -6034,11 +6541,13 @@ _ACEOF
 # handling of long lines.
 ac_delim='%!_!# '
 for ac_last_try in false false :; do
-  ac_tt=`sed -n "/$ac_delim/p" confdefs.h`
-  if test -z "$ac_tt"; then
+  ac_t=`sed -n "/$ac_delim/p" confdefs.h`
+  if test -z "$ac_t"; then
     break
   elif $ac_last_try; then
-    as_fn_error $? "could not make $CONFIG_HEADERS" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_HEADERS" >&5
+$as_echo "$as_me: error: could not make $CONFIG_HEADERS" >&2;}
+   { (exit 1); exit 1; }; }
   else
     ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
   fi
@@ -6123,7 +6632,9 @@ cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 _ACAWK
 _ACEOF
 cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-  as_fn_error $? "could not setup config headers machinery" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: could not setup config headers machinery" >&5
+$as_echo "$as_me: error: could not setup config headers machinery" >&2;}
+   { (exit 1); exit 1; }; }
 fi # test -n "$CONFIG_HEADERS"
 
 
@@ -6136,7 +6647,9 @@ do
   esac
   case $ac_mode$ac_tag in
   :[FHL]*:*);;
-  :L* | :C*:*) as_fn_error $? "invalid tag \`$ac_tag'" "$LINENO" 5;;
+  :L* | :C*:*) { { $as_echo "$as_me:$LINENO: error: invalid tag $ac_tag" >&5
+$as_echo "$as_me: error: invalid tag $ac_tag" >&2;}
+   { (exit 1); exit 1; }; };;
   :[FH]-) ac_tag=-:-;;
   :[FH]*) ac_tag=$ac_tag:$ac_tag.in;;
   esac
@@ -6155,7 +6668,7 @@ do
     for ac_f
     do
       case $ac_f in
-      -) ac_f="$ac_tmp/stdin";;
+      -) ac_f="$tmp/stdin";;
       *) # Look for the file first in the build tree, then in the source tree
 	 # (if the path is not absolute).  The absolute path cannot be DOS-style,
 	 # because $ac_f cannot contain `:'.
@@ -6164,10 +6677,12 @@ do
 	   [\\/$]*) false;;
 	   *) test -f "$srcdir/$ac_f" && ac_f="$srcdir/$ac_f";;
 	   esac ||
-	   as_fn_error 1 "cannot find input file: \`$ac_f'" "$LINENO" 5;;
+	   { { $as_echo "$as_me:$LINENO: error: cannot find input file: $ac_f" >&5
+$as_echo "$as_me: error: cannot find input file: $ac_f" >&2;}
+   { (exit 1); exit 1; }; };;
       esac
       case $ac_f in *\'*) ac_f=`$as_echo "$ac_f" | sed "s/'/'\\\\\\\\''/g"`;; esac
-      as_fn_append ac_file_inputs " '$ac_f'"
+      ac_file_inputs="$ac_file_inputs '$ac_f'"
     done
 
     # Let's still pretend it is `configure' which instantiates (i.e., don't
@@ -6178,7 +6693,7 @@ do
 	`' by configure.'
     if test x"$ac_file" != x-; then
       configure_input="$ac_file.  $configure_input"
-      { $as_echo "$as_me:${as_lineno-$LINENO}: creating $ac_file" >&5
+      { $as_echo "$as_me:$LINENO: creating $ac_file" >&5
 $as_echo "$as_me: creating $ac_file" >&6;}
     fi
     # Neutralize special characters interpreted by sed in replacement strings.
@@ -6190,8 +6705,10 @@ $as_echo "$as_me: creating $ac_file" >&6;}
     esac
 
     case $ac_tag in
-    *:-:* | *:-) cat >"$ac_tmp/stdin" \
-      || as_fn_error $? "could not create $ac_file" "$LINENO" 5 ;;
+    *:-:* | *:-) cat >"$tmp/stdin" \
+      || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; } ;;
     esac
     ;;
   esac
@@ -6219,7 +6736,47 @@ $as_echo X"$ac_file" |
 	    q
 	  }
 	  s/.*/./; q'`
-  as_dir="$ac_dir"; as_fn_mkdir_p
+  { as_dir="$ac_dir"
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || { $as_mkdir_p && mkdir -p "$as_dir"; } || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || { { $as_echo "$as_me:$LINENO: error: cannot create directory $as_dir" >&5
+$as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
+   { (exit 1); exit 1; }; }; }
   ac_builddir=.
 
 case "$ac_dir" in
@@ -6276,6 +6833,7 @@ cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 # If the template does not know about datarootdir, expand it.
 # FIXME: This hack should be removed a few years after 2.60.
 ac_datarootdir_hack=; ac_datarootdir_seen=
+
 ac_sed_dataroot='
 /datarootdir/ {
   p
@@ -6285,11 +6843,12 @@ ac_sed_dataroot='
 /@docdir@/p
 /@infodir@/p
 /@localedir@/p
-/@mandir@/p'
+/@mandir@/p
+'
 case `eval "sed -n \"\$ac_sed_dataroot\" $ac_file_inputs"` in
 *datarootdir*) ac_datarootdir_seen=yes;;
 *@datadir@*|*@docdir@*|*@infodir@*|*@localedir@*|*@mandir@*)
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
+  { $as_echo "$as_me:$LINENO: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
 $as_echo "$as_me: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&2;}
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
@@ -6299,7 +6858,7 @@ cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
   s&@infodir@&$infodir&g
   s&@localedir@&$localedir&g
   s&@mandir@&$mandir&g
-  s&\\\${datarootdir}&$datarootdir&g' ;;
+    s&\\\${datarootdir}&$datarootdir&g' ;;
 esac
 _ACEOF
 
@@ -6327,24 +6886,27 @@ s&@INSTALL@&$ac_INSTALL&;t t
 s&@MKDIR_P@&$ac_MKDIR_P&;t t
 $ac_datarootdir_hack
 "
-eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$ac_tmp/subs.awk" \
-  >$ac_tmp/out || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$tmp/subs.awk" >$tmp/out \
+  || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; }
 
 test -z "$ac_datarootdir_hack$ac_datarootdir_seen" &&
-  { ac_out=`sed -n '/\${datarootdir}/p' "$ac_tmp/out"`; test -n "$ac_out"; } &&
-  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' \
-      "$ac_tmp/out"`; test -z "$ac_out"; } &&
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file contains a reference to the variable \`datarootdir'
-which seems to be undefined.  Please make sure it is defined" >&5
+  { ac_out=`sed -n '/\${datarootdir}/p' "$tmp/out"`; test -n "$ac_out"; } &&
+  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' "$tmp/out"`; test -z "$ac_out"; } &&
+  { $as_echo "$as_me:$LINENO: WARNING: $ac_file contains a reference to the variable \`datarootdir'
+which seems to be undefined.  Please make sure it is defined." >&5
 $as_echo "$as_me: WARNING: $ac_file contains a reference to the variable \`datarootdir'
-which seems to be undefined.  Please make sure it is defined" >&2;}
+which seems to be undefined.  Please make sure it is defined." >&2;}
 
-  rm -f "$ac_tmp/stdin"
+  rm -f "$tmp/stdin"
   case $ac_file in
-  -) cat "$ac_tmp/out" && rm -f "$ac_tmp/out";;
-  *) rm -f "$ac_file" && mv "$ac_tmp/out" "$ac_file";;
+  -) cat "$tmp/out" && rm -f "$tmp/out";;
+  *) rm -f "$ac_file" && mv "$tmp/out" "$ac_file";;
   esac \
-  || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+  || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; }
  ;;
   :H)
   #
@@ -6353,21 +6915,27 @@ which seems to be undefined.  Please make sure it is defined" >&2;}
   if test x"$ac_file" != x-; then
     {
       $as_echo "/* $configure_input  */" \
-      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs"
-    } >"$ac_tmp/config.h" \
-      || as_fn_error $? "could not create $ac_file" "$LINENO" 5
-    if diff "$ac_file" "$ac_tmp/config.h" >/dev/null 2>&1; then
-      { $as_echo "$as_me:${as_lineno-$LINENO}: $ac_file is unchanged" >&5
+      && eval '$AWK -f "$tmp/defines.awk"' "$ac_file_inputs"
+    } >"$tmp/config.h" \
+      || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; }
+    if diff "$ac_file" "$tmp/config.h" >/dev/null 2>&1; then
+      { $as_echo "$as_me:$LINENO: $ac_file is unchanged" >&5
 $as_echo "$as_me: $ac_file is unchanged" >&6;}
     else
       rm -f "$ac_file"
-      mv "$ac_tmp/config.h" "$ac_file" \
-	|| as_fn_error $? "could not create $ac_file" "$LINENO" 5
+      mv "$tmp/config.h" "$ac_file" \
+	|| { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; }
     fi
   else
     $as_echo "/* $configure_input  */" \
-      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs" \
-      || as_fn_error $? "could not create -" "$LINENO" 5
+      && eval '$AWK -f "$tmp/defines.awk"' "$ac_file_inputs" \
+      || { { $as_echo "$as_me:$LINENO: error: could not create -" >&5
+$as_echo "$as_me: error: could not create -" >&2;}
+   { (exit 1); exit 1; }; }
   fi
 # Compute "$ac_file"'s index in $config_headers.
 _am_arg="$ac_file"
@@ -6405,7 +6973,7 @@ $as_echo X"$_am_arg" |
 	  s/.*/./; q'`/stamp-h$_am_stamp_count
  ;;
 
-  :C)  { $as_echo "$as_me:${as_lineno-$LINENO}: executing $ac_file commands" >&5
+  :C)  { $as_echo "$as_me:$LINENO: executing $ac_file commands" >&5
 $as_echo "$as_me: executing $ac_file commands" >&6;}
  ;;
   esac
@@ -6413,7 +6981,7 @@ $as_echo "$as_me: executing $ac_file commands" >&6;}
 
   case $ac_file$ac_mode in
     "depfiles":C) test x"$AMDEP_TRUE" != x"" || {
-  # Older Autoconf quotes --file arguments for eval, but not when files
+  # Autoconf 2.62 quotes --file arguments for eval, but not when files
   # are listed without --file.  Let's play safe and only enable the eval
   # if we detect the quoting.
   case $CONFIG_FILES in
@@ -6426,7 +6994,7 @@ $as_echo "$as_me: executing $ac_file commands" >&6;}
     # Strip MF so we end up with the name of the file.
     mf=`echo "$mf" | sed -e 's/:.*$//'`
     # Check whether this is an Automake generated Makefile or not.
-    # We used to match only the files named 'Makefile.in', but
+    # We used to match only the files named `Makefile.in', but
     # some people rename them; so instead we look at the file content.
     # Grep'ing the first line is not enough: some people post-process
     # each Makefile.in and add a new line on top of each file to say so.
@@ -6460,19 +7028,21 @@ $as_echo X"$mf" |
       continue
     fi
     # Extract the definition of DEPDIR, am__include, and am__quote
-    # from the Makefile without running 'make'.
+    # from the Makefile without running `make'.
     DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
     test -z "$DEPDIR" && continue
     am__include=`sed -n 's/^am__include = //p' < "$mf"`
-    test -z "$am__include" && continue
+    test -z "am__include" && continue
     am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
+    # When using ansi2knr, U may be empty or an underscore; expand it
+    U=`sed -n 's/^U = //p' < "$mf"`
     # Find all dependency output files, they are included files with
     # $(DEPDIR) in their names.  We invoke sed twice because it is the
     # simplest approach to changing $(DEPDIR) to its actual value in the
     # expansion.
     for file in `sed -n "
       s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
-	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
+	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g' -e 's/\$U/'"$U"'/g'`; do
       # Make sure the directory exists.
       test -f "$dirpart/$file" && continue
       fdir=`$as_dirname -- "$file" ||
@@ -6498,7 +7068,47 @@ $as_echo X"$file" |
 	    q
 	  }
 	  s/.*/./; q'`
-      as_dir=$dirpart/$fdir; as_fn_mkdir_p
+      { as_dir=$dirpart/$fdir
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || { $as_mkdir_p && mkdir -p "$as_dir"; } || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || { { $as_echo "$as_me:$LINENO: error: cannot create directory $as_dir" >&5
+$as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
+   { (exit 1); exit 1; }; }; }
       # echo "creating $dirpart/$file"
       echo '# dummy' > "$dirpart/$file"
     done
@@ -6510,12 +7120,15 @@ $as_echo X"$file" |
 done # for ac_tag
 
 
-as_fn_exit 0
+{ (exit 0); exit 0; }
 _ACEOF
+chmod +x $CONFIG_STATUS
 ac_clean_files=$ac_clean_files_save
 
 test $ac_write_fail = 0 ||
-  as_fn_error $? "write failure creating $CONFIG_STATUS" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: write failure creating $CONFIG_STATUS" >&5
+$as_echo "$as_me: error: write failure creating $CONFIG_STATUS" >&2;}
+   { (exit 1); exit 1; }; }
 
 
 # configure is writing to config.log, and then calls config.status.
@@ -6536,10 +7149,10 @@ if test "$no_create" != yes; then
   exec 5>>config.log
   # Use ||, not &&, to avoid exiting from the if with $? = 1, which
   # would make configure fail if this is the last instruction.
-  $ac_cs_success || as_fn_exit 1
+  $ac_cs_success || { (exit 1); exit 1; }
 fi
 if test -n "$ac_unrecognized_opts" && test "$enable_option_checking" != no; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
+  { $as_echo "$as_me:$LINENO: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
 $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2;}
 fi
 
diff --git a/configure.ac b/configure.ac
index d007df0e..becb98dd 100644
--- a/configure.ac
+++ b/configure.ac
@@ -1,5 +1,4 @@
 # Process this file with autoconf to produce a configure script.
-AC_PREREQ([2.69])
 AC_INIT([Grid], [1.0], [paboyle@ph.ed.ac.uk])
 AM_INIT_AUTOMAKE
 AC_CONFIG_MACRO_DIR([m4])
@@ -12,7 +11,7 @@ AC_OPENMP
 AC_PROG_RANLIB
 
 # Checks for libraries.
-AX_GCC_VAR_ATTRIBUTE(aligned)
+#AX_GCC_VAR_ATTRIBUTE(aligned)
 
 # Checks for header files.
 AC_CHECK_HEADERS(stdint.h)

From a8238d0b25af98b623ec5f16f949e74e96eb4ced Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 4 Mar 2015 13:43:19 +0000
Subject: [PATCH 016/429] Move to better name

---
 Grid_init.cc | 82 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 82 insertions(+)
 create mode 100755 Grid_init.cc

diff --git a/Grid_init.cc b/Grid_init.cc
new file mode 100755
index 00000000..e943c3b0
--- /dev/null
+++ b/Grid_init.cc
@@ -0,0 +1,82 @@
+/****************************************************************************/
+/* PAB: Signal magic. Processor state dump is x86-64 specific               */
+/****************************************************************************/
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <stdint.h>
+#include <unistd.h>
+#include <sys/mman.h>
+#include <sys/stat.h> 
+#include <sys/time.h>
+#include <signal.h>
+
+#include "Grid.h"
+
+#undef __X86_64
+namespace dpo {
+
+void Grid_init(void)
+{
+  Grid_debug_handler_init();
+}
+double usecond(void) {
+  struct timeval tv;
+  gettimeofday(&tv,NULL);
+  return 1.0*tv.tv_usec + 1.0e6*tv.tv_sec;
+}
+
+void Grid_sa_signal_handler(int sig,siginfo_t *si,void * ptr)
+{
+         ucontext_t * uc= (ucontext_t *)ptr;
+
+  printf("Caught signal %d\n",si->si_signo);
+  printf("  mem address %llx\n",(uint64_t)si->si_addr);
+  printf("         code %d\n",si->si_code);
+
+#ifdef __X86_64
+  struct sigcontext *sc = (struct sigcontext *)&uc->uc_mcontext;
+  printf("  instruction %llx\n",(uint64_t)sc->rip);
+
+#define REG(A)  printf("  %s %lx\n",#A, sc-> A);
+  REG(rdi);
+  REG(rsi);
+  REG(rbp);
+  REG(rbx);
+  REG(rdx);
+  REG(rax);
+  REG(rcx);
+  REG(rsp);
+  REG(rip);
+
+
+  REG(r8);
+  REG(r9);
+  REG(r10);
+  REG(r11);
+  REG(r12);
+  REG(r13);
+  REG(r14);
+  REG(r15);
+#endif
+
+  fflush(stdout);
+
+  if ( si->si_signo == SIGSEGV ) {
+    printf("Grid_sa_signal_handler: Oops... this was a sigsegv you naughty naughty programmer. Goodbye\n");
+    fflush(stdout);
+    exit(-1);
+  }
+  return;
+};
+
+void Grid_debug_handler_init(void)
+{
+  struct sigaction sa,osa;
+  sigemptyset (&sa.sa_mask);
+  sa.sa_sigaction= Grid_sa_signal_handler;
+  sa.sa_flags    = SA_SIGINFO;
+  sigaction(SIGSEGV,&sa,NULL);
+  sigaction(SIGTRAP,&sa,NULL);
+}
+}

From d9054f67fde53bef0088f259757eddca894de659 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 4 Mar 2015 13:43:48 +0000
Subject: [PATCH 017/429] remove

---
 Grid_signal.cc | 82 --------------------------------------------------
 1 file changed, 82 deletions(-)
 delete mode 100755 Grid_signal.cc

diff --git a/Grid_signal.cc b/Grid_signal.cc
deleted file mode 100755
index e943c3b0..00000000
--- a/Grid_signal.cc
+++ /dev/null
@@ -1,82 +0,0 @@
-/****************************************************************************/
-/* PAB: Signal magic. Processor state dump is x86-64 specific               */
-/****************************************************************************/
-
-#include <stdlib.h>
-#include <stdio.h>
-#include <stdint.h>
-#include <unistd.h>
-#include <sys/mman.h>
-#include <sys/stat.h> 
-#include <sys/time.h>
-#include <signal.h>
-
-#include "Grid.h"
-
-#undef __X86_64
-namespace dpo {
-
-void Grid_init(void)
-{
-  Grid_debug_handler_init();
-}
-double usecond(void) {
-  struct timeval tv;
-  gettimeofday(&tv,NULL);
-  return 1.0*tv.tv_usec + 1.0e6*tv.tv_sec;
-}
-
-void Grid_sa_signal_handler(int sig,siginfo_t *si,void * ptr)
-{
-         ucontext_t * uc= (ucontext_t *)ptr;
-
-  printf("Caught signal %d\n",si->si_signo);
-  printf("  mem address %llx\n",(uint64_t)si->si_addr);
-  printf("         code %d\n",si->si_code);
-
-#ifdef __X86_64
-  struct sigcontext *sc = (struct sigcontext *)&uc->uc_mcontext;
-  printf("  instruction %llx\n",(uint64_t)sc->rip);
-
-#define REG(A)  printf("  %s %lx\n",#A, sc-> A);
-  REG(rdi);
-  REG(rsi);
-  REG(rbp);
-  REG(rbx);
-  REG(rdx);
-  REG(rax);
-  REG(rcx);
-  REG(rsp);
-  REG(rip);
-
-
-  REG(r8);
-  REG(r9);
-  REG(r10);
-  REG(r11);
-  REG(r12);
-  REG(r13);
-  REG(r14);
-  REG(r15);
-#endif
-
-  fflush(stdout);
-
-  if ( si->si_signo == SIGSEGV ) {
-    printf("Grid_sa_signal_handler: Oops... this was a sigsegv you naughty naughty programmer. Goodbye\n");
-    fflush(stdout);
-    exit(-1);
-  }
-  return;
-};
-
-void Grid_debug_handler_init(void)
-{
-  struct sigaction sa,osa;
-  sigemptyset (&sa.sa_mask);
-  sa.sa_sigaction= Grid_sa_signal_handler;
-  sa.sa_flags    = SA_SIGINFO;
-  sigaction(SIGSEGV,&sa,NULL);
-  sigaction(SIGTRAP,&sa,NULL);
-}
-}

From 82eeb9d07e58ce5b91a14481b59b9198dd32aa76 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 4 Mar 2015 13:44:33 +0000
Subject: [PATCH 018/429] Improving

---
 Grid.h        | 2 +-
 Grid_config.h | 8 ++++----
 Makefile.am   | 2 +-
 3 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/Grid.h b/Grid.h
index f73d52ba..5a3571e4 100644
--- a/Grid.h
+++ b/Grid.h
@@ -27,7 +27,7 @@
 ////////////////////////////////////////////////////////////
 // Tunable header includes
 ////////////////////////////////////////////////////////////
-#define HAVE_OPENMP
+
 #ifdef HAVE_OPENMP
 #define OMP
 #include <omp.h>
diff --git a/Grid_config.h b/Grid_config.h
index 12eeb8ac..64003419 100644
--- a/Grid_config.h
+++ b/Grid_config.h
@@ -2,13 +2,13 @@
 /* Grid_config.h.in.  Generated from configure.ac by autoheader.  */
 
 /* AVX */
-/* #undef AVX1 */
+#define AVX1 1
 
 /* AVX2 */
 /* #undef AVX2 */
 
 /* AVX512 */
-#define AVX512 1
+/* #undef AVX512 */
 
 /* Define to 1 if you have the `gettimeofday' function. */
 #define HAVE_GETTIMEOFDAY 1
@@ -17,10 +17,10 @@
 #define HAVE_INTTYPES_H 1
 
 /* Define to 1 if you have the <malloc.h> header file. */
-#define HAVE_MALLOC_H 1
+/* #undef HAVE_MALLOC_H */
 
 /* Define to 1 if you have the <malloc/malloc.h> header file. */
-/* #undef HAVE_MALLOC_MALLOC_H */
+#define HAVE_MALLOC_MALLOC_H 1
 
 /* Define to 1 if you have the <memory.h> header file. */
 #define HAVE_MEMORY_H 1
diff --git a/Makefile.am b/Makefile.am
index b6c2ef16..13c2484b 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -5,7 +5,7 @@ AM_CXXFLAGS = -I$(top_srcdir)/
 # Libraries
 #
 lib_LIBRARIES = libGrid.a
-libGrid_a_SOURCES = Grid_signal.cc
+libGrid_a_SOURCES = Grid_init.cc
 
 #
 # Include files

From 483cab34ab3982635b17c5600975df58e95b5db3 Mon Sep 17 00:00:00 2001
From: paboyle <paboyle@ph.ed.ac.uk>
Date: Sat, 7 Mar 2015 07:00:39 +0000
Subject: [PATCH 019/429] Update AUTHORS

---
 AUTHORS | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/AUTHORS b/AUTHORS
index e69de29b..4ec04f53 100644
--- a/AUTHORS
+++ b/AUTHORS
@@ -0,0 +1,5 @@
+Peter Boyle
+Azusa Yamaguchi
+Intel Parallel Computing Centre @ Higgs Centre for Theoretical Physics
+University of Edinburgh
+Scotland, UK

From 1e08841fa5b9f49a91e0bdd558b5680436ab96a8 Mon Sep 17 00:00:00 2001
From: paboyle <paboyle@ph.ed.ac.uk>
Date: Sat, 7 Mar 2015 07:09:09 +0000
Subject: [PATCH 020/429] Update INSTALL

---
 INSTALL | 86 ++++++---------------------------------------------------
 1 file changed, 8 insertions(+), 78 deletions(-)

diff --git a/INSTALL b/INSTALL
index 20998407..9996b52c 100644
--- a/INSTALL
+++ b/INSTALL
@@ -102,9 +102,13 @@ for details on some of the pertinent environment variables.
 
    You can give `configure' initial values for configuration parameters
 by setting variables in the command line or in the environment.  Here
-is an example:
+is are examples:
 
-     ./configure CC=c99 CFLAGS=-g LIBS=-lposix
+     ./configure CXX=clang++ CXXFLAGS="-std=c++11 -O3 -mavx" --enable-simd=AVX1
+
+     ./configure CXX=clang++ CXXFLAGS="-std=c++11 -O3 -mavx2" --enable-simd=AVX2
+
+     ./configure CXX=icpc CXXFLAGS="-std=c++11 -O3 -mmic" --enable-simd=AVX512 --host=none
 
    *Note Defining Variables::, for more details.
 
@@ -124,19 +128,6 @@ architecture at a time in the source code directory.  After you have
 installed the package for one architecture, use `make distclean' before
 reconfiguring for another architecture.
 
-   On MacOS X 10.5 and later systems, you can create libraries and
-executables that work on multiple system types--known as "fat" or
-"universal" binaries--by specifying multiple `-arch' options to the
-compiler but only a single `-arch' option to the preprocessor.  Like
-this:
-
-     ./configure CC="gcc -arch i386 -arch x86_64 -arch ppc -arch ppc64" \
-                 CXX="g++ -arch i386 -arch x86_64 -arch ppc -arch ppc64" \
-                 CPP="gcc -E" CXXCPP="g++ -E"
-
-   This is not guaranteed to produce working output in all cases, you
-may have to build one architecture at a time and combine the results
-using the `lipo' tool if you have problems.
 
 Installation Names
 ==================
@@ -218,70 +209,9 @@ overridden with `make V=0'.
 Particular systems
 ==================
 
-   On HP-UX, the default C compiler is not ANSI C compatible.  If GNU
-CC is not installed, it is recommended to use the following options in
-order to use an ANSI C compiler:
+Cross compiling for native execution on XeonPhi requires 
 
-     ./configure CC="cc -Ae -D_XOPEN_SOURCE=500"
-
-and if that doesn't work, install pre-built binaries of GCC for HP-UX.
-
-   HP-UX `make' updates targets which have the same time stamps as
-their prerequisites, which makes it generally unusable when shipped
-generated files such as `configure' are involved.  Use GNU `make'
-instead.
-
-   On OSF/1 a.k.a. Tru64, some versions of the default C compiler cannot
-parse its `<wchar.h>' header file.  The option `-nodtk' can be used as
-a workaround.  If GNU CC is not installed, it is therefore recommended
-to try
-
-     ./configure CC="cc"
-
-and if that doesn't work, try
-
-     ./configure CC="cc -nodtk"
-
-   On Solaris, don't put `/usr/ucb' early in your `PATH'.  This
-directory contains several dysfunctional programs; working variants of
-these programs are available in `/usr/bin'.  So, if you need `/usr/ucb'
-in your `PATH', put it _after_ `/usr/bin'.
-
-   On Haiku, software installed for all users goes in `/boot/common',
-not `/usr/local'.  It is recommended to use the following options:
-
-     ./configure --prefix=/boot/common
-
-Specifying the System Type
-==========================
-
-   There may be some features `configure' cannot figure out
-automatically, but needs to determine by the type of machine the package
-will run on.  Usually, assuming the package is built to be run on the
-_same_ architectures, `configure' can figure that out, but if it prints
-a message saying it cannot guess the machine type, give it the
-`--build=TYPE' option.  TYPE can either be a short name for the system
-type, such as `sun4', or a canonical name which has the form:
-
-     CPU-COMPANY-SYSTEM
-
-where SYSTEM can have one of these forms:
-
-     OS
-     KERNEL-OS
-
-   See the file `config.sub' for the possible values of each field.  If
-`config.sub' isn't included in this package, then this package doesn't
-need to know the machine type.
-
-   If you are _building_ compiler tools for cross-compiling, you should
-use the option `--target=TYPE' to select the type of system they will
-produce code for.
-
-   If you want to _use_ a cross compiler, that generates code for a
-platform different from the build platform, you should specify the
-"host" platform (i.e., that on which the generated programs will
-eventually be run) with `--host=TYPE'.
+   --build=none
 
 Sharing Defaults
 ================

From f04b6312c8bba06d33bce0234b3a9f4f8c9d370f Mon Sep 17 00:00:00 2001
From: paboyle <paboyle@ph.ed.ac.uk>
Date: Sat, 7 Mar 2015 07:19:01 +0000
Subject: [PATCH 021/429] Update README

---
 README | 42 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 42 insertions(+)

diff --git a/README b/README
index e69de29b..41b66b6b 100644
--- a/README
+++ b/README
@@ -0,0 +1,42 @@
+This library provides data parallel C++ container classes with internal memory layout
+that is transformed to map efficiently to SIMD architectures. CSHIFT facilities
+are provided, similar to HPF and cmfortran, and user control is given over the mapping of
+array indices to both MPI tasks and SIMD processing elements.
+
+* Identically shaped arrays then be processed with perfect data parallelisation.
+* Such identically shapped arrays are called conformable arrays.
+
+The transformation is based on the observation that Cartesian array processing involves
+identical processing to be performed on different regions of the Cartesian array.
+
+The library will (eventually) both geometrically decompose into MPI tasks and across SIMD lanes.
+
+Data parallel array operations can then be specified with a SINGLE data parallel paradigm, but
+optimally use MPI, OpenMP and SIMD parallelism under the hood. This is a significant simplification
+for most programmers.
+
+The layout transformations are parametrised by the SIMD vector length. This adapts according to the architecture.
+Presently SSE2 (128 bit) AVX, AVX2 (256 bit) and IMCI and AVX512 (512 bit) targets are supported.
+
+These are presented as 
+
+  vRealF, vRealD, vComplexF, vComplexD 
+
+internal vector data types. These may be useful in themselves for other programmers.
+The corresponding scalar types are named
+
+  RealF, RealD, ComplexF, ComplexD
+
+MPI parallelism is UNIMPLEMENTED and for now only OpenMP and SIMD parallelism is present in the library.
+
+   You can give `configure' initial values for configuration parameters
+by setting variables in the command line or in the environment.  Here
+is are examples:
+
+     ./configure CXX=clang++ CXXFLAGS="-std=c++11 -O3 -mavx" --enable-simd=AVX1
+
+     ./configure CXX=clang++ CXXFLAGS="-std=c++11 -O3 -mavx2" --enable-simd=AVX2
+
+     ./configure CXX=icpc CXXFLAGS="-std=c++11 -O3 -mmic" --enable-simd=AVX512 --host=none
+     
+     

From 886df6ed1e8afc76bb7fab10d096e7bfbc7f7a30 Mon Sep 17 00:00:00 2001
From: paboyle <paboyle@ph.ed.ac.uk>
Date: Sat, 7 Mar 2015 07:20:12 +0000
Subject: [PATCH 022/429] Update README.md

---
 README.md | 43 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 43 insertions(+)

diff --git a/README.md b/README.md
index 9f2e51a6..42537417 100644
--- a/README.md
+++ b/README.md
@@ -1,2 +1,45 @@
 # Grid
 Data parallel C++ mathematical object library
+
+This library provides data parallel C++ container classes with internal memory layout
+that is transformed to map efficiently to SIMD architectures. CSHIFT facilities
+are provided, similar to HPF and cmfortran, and user control is given over the mapping of
+array indices to both MPI tasks and SIMD processing elements.
+
+* Identically shaped arrays then be processed with perfect data parallelisation.
+* Such identically shapped arrays are called conformable arrays.
+
+The transformation is based on the observation that Cartesian array processing involves
+identical processing to be performed on different regions of the Cartesian array.
+
+The library will (eventually) both geometrically decompose into MPI tasks and across SIMD lanes.
+
+Data parallel array operations can then be specified with a SINGLE data parallel paradigm, but
+optimally use MPI, OpenMP and SIMD parallelism under the hood. This is a significant simplification
+for most programmers.
+
+The layout transformations are parametrised by the SIMD vector length. This adapts according to the architecture.
+Presently SSE2 (128 bit) AVX, AVX2 (256 bit) and IMCI and AVX512 (512 bit) targets are supported.
+
+These are presented as 
+
+  vRealF, vRealD, vComplexF, vComplexD 
+
+internal vector data types. These may be useful in themselves for other programmers.
+The corresponding scalar types are named
+
+  RealF, RealD, ComplexF, ComplexD
+
+MPI parallelism is UNIMPLEMENTED and for now only OpenMP and SIMD parallelism is present in the library.
+
+   You can give `configure' initial values for configuration parameters
+by setting variables in the command line or in the environment.  Here
+is are examples:
+
+     ./configure CXX=clang++ CXXFLAGS="-std=c++11 -O3 -mavx" --enable-simd=AVX1
+
+     ./configure CXX=clang++ CXXFLAGS="-std=c++11 -O3 -mavx2" --enable-simd=AVX2
+
+     ./configure CXX=icpc CXXFLAGS="-std=c++11 -O3 -mmic" --enable-simd=AVX512 --host=none
+     
+     

From 196fd203e246546749db358c4582b33ee40c6cf7 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sun, 29 Mar 2015 20:35:37 +0100
Subject: [PATCH 023/429] Fixing the Checkerboarding cshift. Implemented "fake"
 communications in preparation for the leap to MPI.

---
 Grid/Grid.xcodeproj/project.pbxproj           | 239 +++++++++++
 .../contents.xcworkspacedata                  |   7 +
 .../UserInterfaceState.xcuserstate            | Bin 0 -> 4124 bytes
 .../pab.xcuserdatad/xcschemes/Grid.xcscheme   |  43 ++
 .../xcschemes/xcschememanagement.plist        |  22 +
 Grid_Cartesian.h                              | 184 ++++++---
 Grid_Communicator.h                           |  39 ++
 Grid_Lattice.h                                | 299 ++++----------
 Grid_QCD.h                                    |   1 -
 Grid_config.h                                 |  12 +
 Grid_fake.cc                                  |  34 ++
 Grid_fake_cshift.h                            | 246 +++++++++++
 Grid_init.cc                                  |   4 +-
 Grid_main.cc                                  | 124 ++++--
 Grid_math_types.h                             | 113 +++++-
 Grid_mpi.cc                                   |  85 ++++
 Grid_mpi_cshift.h                             | 384 ++++++++++++++++++
 Grid_none_cshift.h                            | 313 ++++++++++++++
 Grid_simd.h                                   |   4 +-
 Grid_vComplexD.h                              |  46 ++-
 Grid_vComplexF.h                              |  47 ++-
 Grid_vRealD.h                                 |  35 ++
 Grid_vRealF.h                                 |  38 +-
 23 files changed, 1976 insertions(+), 343 deletions(-)
 create mode 100644 Grid/Grid.xcodeproj/project.pbxproj
 create mode 100644 Grid/Grid.xcodeproj/project.xcworkspace/contents.xcworkspacedata
 create mode 100644 Grid/Grid.xcodeproj/project.xcworkspace/xcuserdata/pab.xcuserdatad/UserInterfaceState.xcuserstate
 create mode 100644 Grid/Grid.xcodeproj/xcuserdata/pab.xcuserdatad/xcschemes/Grid.xcscheme
 create mode 100644 Grid/Grid.xcodeproj/xcuserdata/pab.xcuserdatad/xcschemes/xcschememanagement.plist
 create mode 100644 Grid_Communicator.h
 create mode 100644 Grid_fake.cc
 create mode 100644 Grid_fake_cshift.h
 create mode 100644 Grid_mpi.cc
 create mode 100644 Grid_mpi_cshift.h
 create mode 100644 Grid_none_cshift.h

diff --git a/Grid/Grid.xcodeproj/project.pbxproj b/Grid/Grid.xcodeproj/project.pbxproj
new file mode 100644
index 00000000..20cd1d1a
--- /dev/null
+++ b/Grid/Grid.xcodeproj/project.pbxproj
@@ -0,0 +1,239 @@
+// !$*UTF8*$!
+{
+	archiveVersion = 1;
+	classes = {
+	};
+	objectVersion = 46;
+	objects = {
+
+/* Begin PBXBuildFile section */
+		0BF5DA461AC370840045EA34 /* main.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 0BF5DA451AC370840045EA34 /* main.cpp */; };
+/* End PBXBuildFile section */
+
+/* Begin PBXCopyFilesBuildPhase section */
+		0BF5DA401AC370840045EA34 /* CopyFiles */ = {
+			isa = PBXCopyFilesBuildPhase;
+			buildActionMask = 2147483647;
+			dstPath = /usr/share/man/man1/;
+			dstSubfolderSpec = 0;
+			files = (
+			);
+			runOnlyForDeploymentPostprocessing = 1;
+		};
+/* End PBXCopyFilesBuildPhase section */
+
+/* Begin PBXFileReference section */
+		0BF5DA421AC370840045EA34 /* Grid */ = {isa = PBXFileReference; explicitFileType = "compiled.mach-o.executable"; includeInIndex = 0; path = Grid; sourceTree = BUILT_PRODUCTS_DIR; };
+		0BF5DA451AC370840045EA34 /* main.cpp */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.cpp.cpp; path = main.cpp; sourceTree = "<group>"; };
+/* End PBXFileReference section */
+
+/* Begin PBXFrameworksBuildPhase section */
+		0BF5DA3F1AC370840045EA34 /* Frameworks */ = {
+			isa = PBXFrameworksBuildPhase;
+			buildActionMask = 2147483647;
+			files = (
+			);
+			runOnlyForDeploymentPostprocessing = 0;
+		};
+/* End PBXFrameworksBuildPhase section */
+
+/* Begin PBXGroup section */
+		0BF5DA391AC370840045EA34 = {
+			isa = PBXGroup;
+			children = (
+				0BF5DA441AC370840045EA34 /* Grid */,
+				0BF5DA431AC370840045EA34 /* Products */,
+			);
+			sourceTree = "<group>";
+		};
+		0BF5DA431AC370840045EA34 /* Products */ = {
+			isa = PBXGroup;
+			children = (
+				0BF5DA421AC370840045EA34 /* Grid */,
+			);
+			name = Products;
+			sourceTree = "<group>";
+		};
+		0BF5DA441AC370840045EA34 /* Grid */ = {
+			isa = PBXGroup;
+			children = (
+				0BF5DA451AC370840045EA34 /* main.cpp */,
+			);
+			path = Grid;
+			sourceTree = "<group>";
+		};
+/* End PBXGroup section */
+
+/* Begin PBXNativeTarget section */
+		0BF5DA411AC370840045EA34 /* Grid */ = {
+			isa = PBXNativeTarget;
+			buildConfigurationList = 0BF5DA491AC370840045EA34 /* Build configuration list for PBXNativeTarget "Grid" */;
+			buildPhases = (
+				0BF5DA3E1AC370840045EA34 /* Sources */,
+				0BF5DA3F1AC370840045EA34 /* Frameworks */,
+				0BF5DA401AC370840045EA34 /* CopyFiles */,
+			);
+			buildRules = (
+			);
+			dependencies = (
+			);
+			name = Grid;
+			productName = Grid;
+			productReference = 0BF5DA421AC370840045EA34 /* Grid */;
+			productType = "com.apple.product-type.tool";
+		};
+/* End PBXNativeTarget section */
+
+/* Begin PBXProject section */
+		0BF5DA3A1AC370840045EA34 /* Project object */ = {
+			isa = PBXProject;
+			attributes = {
+				LastUpgradeCheck = 0610;
+				ORGANIZATIONNAME = "University of Edinburgh";
+				TargetAttributes = {
+					0BF5DA411AC370840045EA34 = {
+						CreatedOnToolsVersion = 6.1.1;
+					};
+				};
+			};
+			buildConfigurationList = 0BF5DA3D1AC370840045EA34 /* Build configuration list for PBXProject "Grid" */;
+			compatibilityVersion = "Xcode 3.2";
+			developmentRegion = English;
+			hasScannedForEncodings = 0;
+			knownRegions = (
+				en,
+			);
+			mainGroup = 0BF5DA391AC370840045EA34;
+			productRefGroup = 0BF5DA431AC370840045EA34 /* Products */;
+			projectDirPath = "";
+			projectRoot = "";
+			targets = (
+				0BF5DA411AC370840045EA34 /* Grid */,
+			);
+		};
+/* End PBXProject section */
+
+/* Begin PBXSourcesBuildPhase section */
+		0BF5DA3E1AC370840045EA34 /* Sources */ = {
+			isa = PBXSourcesBuildPhase;
+			buildActionMask = 2147483647;
+			files = (
+				0BF5DA461AC370840045EA34 /* main.cpp in Sources */,
+			);
+			runOnlyForDeploymentPostprocessing = 0;
+		};
+/* End PBXSourcesBuildPhase section */
+
+/* Begin XCBuildConfiguration section */
+		0BF5DA471AC370840045EA34 /* Debug */ = {
+			isa = XCBuildConfiguration;
+			buildSettings = {
+				ALWAYS_SEARCH_USER_PATHS = NO;
+				CLANG_CXX_LANGUAGE_STANDARD = "gnu++0x";
+				CLANG_CXX_LIBRARY = "libc++";
+				CLANG_ENABLE_MODULES = YES;
+				CLANG_ENABLE_OBJC_ARC = YES;
+				CLANG_WARN_BOOL_CONVERSION = YES;
+				CLANG_WARN_CONSTANT_CONVERSION = YES;
+				CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
+				CLANG_WARN_EMPTY_BODY = YES;
+				CLANG_WARN_ENUM_CONVERSION = YES;
+				CLANG_WARN_INT_CONVERSION = YES;
+				CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
+				CLANG_WARN_UNREACHABLE_CODE = YES;
+				CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
+				COPY_PHASE_STRIP = NO;
+				ENABLE_STRICT_OBJC_MSGSEND = YES;
+				GCC_C_LANGUAGE_STANDARD = gnu99;
+				GCC_DYNAMIC_NO_PIC = NO;
+				GCC_OPTIMIZATION_LEVEL = 0;
+				GCC_PREPROCESSOR_DEFINITIONS = (
+					"DEBUG=1",
+					"$(inherited)",
+				);
+				GCC_SYMBOLS_PRIVATE_EXTERN = NO;
+				GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
+				GCC_WARN_ABOUT_RETURN_TYPE = YES_ERROR;
+				GCC_WARN_UNDECLARED_SELECTOR = YES;
+				GCC_WARN_UNINITIALIZED_AUTOS = YES_AGGRESSIVE;
+				GCC_WARN_UNUSED_FUNCTION = YES;
+				GCC_WARN_UNUSED_VARIABLE = YES;
+				MACOSX_DEPLOYMENT_TARGET = 10.10;
+				MTL_ENABLE_DEBUG_INFO = YES;
+				ONLY_ACTIVE_ARCH = YES;
+				SDKROOT = macosx;
+			};
+			name = Debug;
+		};
+		0BF5DA481AC370840045EA34 /* Release */ = {
+			isa = XCBuildConfiguration;
+			buildSettings = {
+				ALWAYS_SEARCH_USER_PATHS = NO;
+				CLANG_CXX_LANGUAGE_STANDARD = "gnu++0x";
+				CLANG_CXX_LIBRARY = "libc++";
+				CLANG_ENABLE_MODULES = YES;
+				CLANG_ENABLE_OBJC_ARC = YES;
+				CLANG_WARN_BOOL_CONVERSION = YES;
+				CLANG_WARN_CONSTANT_CONVERSION = YES;
+				CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
+				CLANG_WARN_EMPTY_BODY = YES;
+				CLANG_WARN_ENUM_CONVERSION = YES;
+				CLANG_WARN_INT_CONVERSION = YES;
+				CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
+				CLANG_WARN_UNREACHABLE_CODE = YES;
+				CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
+				COPY_PHASE_STRIP = YES;
+				DEBUG_INFORMATION_FORMAT = "dwarf-with-dsym";
+				ENABLE_NS_ASSERTIONS = NO;
+				ENABLE_STRICT_OBJC_MSGSEND = YES;
+				GCC_C_LANGUAGE_STANDARD = gnu99;
+				GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
+				GCC_WARN_ABOUT_RETURN_TYPE = YES_ERROR;
+				GCC_WARN_UNDECLARED_SELECTOR = YES;
+				GCC_WARN_UNINITIALIZED_AUTOS = YES_AGGRESSIVE;
+				GCC_WARN_UNUSED_FUNCTION = YES;
+				GCC_WARN_UNUSED_VARIABLE = YES;
+				MACOSX_DEPLOYMENT_TARGET = 10.10;
+				MTL_ENABLE_DEBUG_INFO = NO;
+				SDKROOT = macosx;
+			};
+			name = Release;
+		};
+		0BF5DA4A1AC370840045EA34 /* Debug */ = {
+			isa = XCBuildConfiguration;
+			buildSettings = {
+				PRODUCT_NAME = "$(TARGET_NAME)";
+			};
+			name = Debug;
+		};
+		0BF5DA4B1AC370840045EA34 /* Release */ = {
+			isa = XCBuildConfiguration;
+			buildSettings = {
+				PRODUCT_NAME = "$(TARGET_NAME)";
+			};
+			name = Release;
+		};
+/* End XCBuildConfiguration section */
+
+/* Begin XCConfigurationList section */
+		0BF5DA3D1AC370840045EA34 /* Build configuration list for PBXProject "Grid" */ = {
+			isa = XCConfigurationList;
+			buildConfigurations = (
+				0BF5DA471AC370840045EA34 /* Debug */,
+				0BF5DA481AC370840045EA34 /* Release */,
+			);
+			defaultConfigurationIsVisible = 0;
+			defaultConfigurationName = Release;
+		};
+		0BF5DA491AC370840045EA34 /* Build configuration list for PBXNativeTarget "Grid" */ = {
+			isa = XCConfigurationList;
+			buildConfigurations = (
+				0BF5DA4A1AC370840045EA34 /* Debug */,
+				0BF5DA4B1AC370840045EA34 /* Release */,
+			);
+			defaultConfigurationIsVisible = 0;
+		};
+/* End XCConfigurationList section */
+	};
+	rootObject = 0BF5DA3A1AC370840045EA34 /* Project object */;
+}
diff --git a/Grid/Grid.xcodeproj/project.xcworkspace/contents.xcworkspacedata b/Grid/Grid.xcodeproj/project.xcworkspace/contents.xcworkspacedata
new file mode 100644
index 00000000..ce76f4f5
--- /dev/null
+++ b/Grid/Grid.xcodeproj/project.xcworkspace/contents.xcworkspacedata
@@ -0,0 +1,7 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<Workspace
+   version = "1.0">
+   <FileRef
+      location = "self:Grid.xcodeproj">
+   </FileRef>
+</Workspace>
diff --git a/Grid/Grid.xcodeproj/project.xcworkspace/xcuserdata/pab.xcuserdatad/UserInterfaceState.xcuserstate b/Grid/Grid.xcodeproj/project.xcworkspace/xcuserdata/pab.xcuserdatad/UserInterfaceState.xcuserstate
new file mode 100644
index 0000000000000000000000000000000000000000..638ce019b327e05f2249ae344a55de843e067d13
GIT binary patch
literal 4124
zcma)933O9s7QQzxdH;KVB2CLy*0!`35n2O;E}*69rUhD{X-g^4@R~l_2Wb*t5=vR*
zN6<McIOBr599nQdXB<XHMa6Y=Mjgi^FqL5#bvPck!>A~Wh@#B@@|r@_Imh;#yyd_D
zUB3I>d;ixO4rr06%hd-677)M+HV~2FhNhgAsvglo!KR#0>vGi}jV#Pjbbp7&U)SeE
zL*X>Q8@KOqt^fiO42B^v6vo44Faa)yi7*NBARn%PsW2U8z)UEID!3XJ!D3hfE#QNt
zpg=45p$$4<Ijn_sa2>1%1{>gd*a$bl&G2iu8}5O7;Xb$@9)ySBxA1#-6#f8Dz_ajY
zcm>{sze69q1Bc)k{1-lh&*2L=4qw7oZ~}gS^9YD&!9kdUPE5r#9EziGG>*Y+ybN79
z8H;f`mS6=|<9w_`FD}BxxCB)UVH8*4+J>B9ED-ny$Uwmk4oHEt1-0J%6>4{+nP2-M
zY0U8llt`o(E`dSJ!h~LMLMpSeA<X4-4ymrFToBS%M8b++tqA#JooX=ZbLJGf3Oton
zg;jaeii(T!++~&SJkQhucV303sNCc6lzR%x%M*r&!C1)M1;Zg7M!-lI1*2gMvoVno
zCNY^&X5R%FkO^5Z4zeK!a+rf%!cv%%rLn>Me-L-0Ojnf^;gA-LMm+wgwo-|zZ9eD7
z3U!$h3q&j1v}j28M)})FRY-5Bn-lRlhgYbrvG#T?*uFpuwuQRN)DC5(7Sesrltdur
z0+&s%Q~f+JQ%_>GUyUamuHu15s)IFJP#aLnol1Hx4+SHkfGQ`VcEM!G-M*a-ItXr<
z1_e;qYyurrI@KogR^_G#*LrJWQKdDYR%kr3P*BmkcX06{C|+Faz3{H7);l-OVJ94n
z=krP^>ERg@SgL8E6g({hx;(nBa5be+2E{z*@keu81(a-OsZzqtYhXU)?uKfZ1+!re
z)IcrFg?Vf!8^(sSbT)#GWTV*V-8?~FsOQNU#Fv}mDH_AZ^5h77Inzwi2!ktesOv+a
zK&zrxM;f(=#v^MPSgF!^^gd@6e=%2YQ+2g%fL--UE5|UZhXMg!v_9trb=AQt&f|z*
zSJhxmsEzNRbTKVWn-}&r1hs2pYIPe=sJ2W~dBQWec$_xs&s+03OD<NJ<cV)`Rarq%
zd2vBrrMs{)&po-QAg`?0U7lC&DyS$gce`C4*Hq3RLxCB*(pq*=?3{>e))crWWfm8%
zo0Pd`vfJgFl*x-`-8wTLDzx{&T#jOfnZ`2EdX20l60`y$M)d@s6M_(eFkA~d%VOhL
zHp^j`vRpQP7q6TQSP5OQieFdr%5kvE*i=@)uQQG68Rm&Zc!N|D1dq&7x<fHu#!l|E
zw<FY5%L}7j;U(mA=KWtF8lqZ2i*|c>C3<<ojW%j(m#Jc4yRr78<RoEPp+>a!;Lm6<
zoG|rkYGes_!K_3>GJ(S;xDj%D;089K2Y$saH{wdIXsjP_r8=s1ModU=fs$U>3|rVl
zHkqX+6#NG6fV5t?4Q^+XSY9vO30qk{yMnova5WdR#mS$I;A)iipcaj_sq?Z+H>QK}
zXQrqK9_TgdGog7K^!!Nk4z76$b7hx?GUNLalDptxBZ%G1-2;!XY4ISuYJg8BwQT@J
zpEI+`&uN~|ahR<|^9L64y-_{pkH&O0p>Gd7HUP@wtk7sSMyr1kN_N0guos?YMQl2|
z@&X9^;DsbtL+}!(QL*7_xHQStSK+lJNgdu`<Jk;D@)njP`4;R?k~{zh*;TB>7#{|~
z2J&iT^CxEh2}h0oa~S>wN8sP^F1!ct!+%&Q^RO~j&MH_Xt72F0=A9>l(>x14f{)=7
z_>^}X2Pb+8yM`@hOPFdP=`>hm_GeW$j2T2S-kD_L4ZG1i4fKzTSZgOox!HU&`^sRx
zqkpEv6&YH1I{IJDGqy*fysy-$VO@>z8^!3@yji6t><;j;IS?<hdfv1W**Xc|K<-XB
z1*hR_IK!&hEH<0X*$HRiTlfynu^Q%O4Xlw3YMP@ii`Un}I<37Ue%EZ)tm)=ZBd|mb
z4mjFx1_hM38MLAeMI@}2&1Li0{GA-aEVM%ghc1QHarii7qZ$=mGiDF()M}T3RisYo
zYE%L-V>FFSUhU_p*B=V=7Mz*17~dY!qrAfXx)$cGgzq1gBofzN-yK%t9L2%VzMW+x
zd>)4BoX9ww)%QRa_s^*A{)t+p1){2M1n$Iy_*l$L@E)_+f_~m(4ljJX6mwY<YaZY{
zPQXb?{5*yE9Q%a^lh<s&n8`Q=rzJ_A#zLN>MRCb_Ns=@0N^`m=JerA~0guX9OTwc{
z!=oy^n)z7k1&?N7&Bb=G7U#01Oku8tJh3JpvrDz5NfzsILDHTkT*&;a&8V`JIBSi=
zj>-G%(#0>u_x526`fw@OQNdR9BR?{w>7h{6<dQ+n+W0sf-mV6<Sj4n!3{rJf)Lcmq
z>slv2k^E$dwKHJL*z!ZT4BNSZ4%BctH=<xCKccq6^d^4H4z%fNuraLg-{>O0p*I*c
z7{J-n)_<1t#E;_!1(R!PQApPk^~Ei<GmSyAtuVLN3DXT*tb>bi0h3uVK2g!noiisY
zcH?TFb3DiB=ZAkX7`?a#ihsFyd?4UDyw2y${sk6#alN7amx%JT)_P~F-D;Z&k*XiS
z>v3a4{FKWOBz~A4$G=jh@voDan2uv{B2MLB1C=-n=Ws@Ou>lw0LJZ^GxD)?`PvTyD
z2KV9f_#(cH`|%+5;UPSXNAO*IACKZ0%MeSxrPQ*>60>Z!Y_Z&Dx!>}DWt(M}<#EeC
z%K^(_%SV<kEMHpATFzO15I~THbRk0+FSvxs!c@U6c!WBkUT74Wgl1u}&>}1qT7`%Z
z6S{<MVU4g(STAf4HVQWgJ;EP_KH*DihSg)e-uj?*uk|JCE7reQU$gGFzGMBs`jz#h
z^|bYj^{n+f>-X04wm~+hEzLH>Hq4f88)+MD8*3}H&9bewZLvLPd)xM&?U?Oz+i}|o
z+bP@EA{7UV>EcLnv^Z9rD7wTcqFXE$XNXscGsPNlsi=s{#fZ2{yg|HK+$`QI-XU%k
z?-uVBd&NEC<Kh$IUhx@mpZK<TLOf3_#7YQ}iJc4~qsSPNL9$3Tagk}HkQ9>=QcB85
z1*sv;#77k3Cmm!t36LP^BAduf<QB4p+)nN!caeL@@5mG6Wpa@8kwfG#IYQng?~|kC
z19Fm_CTGZ5@*Vk}oR?4%B%72j<w{pc*GMhW25GCbU3ye{N_tv)R(ei)L3&MkLwZws
zOL|*6ARUoDl#a_p9wLvCGvq8eTfS7zmtFD{*)12yMRKDYk?)fC$WP19$ou5i<u~QG
z<o)tNxljI7J|>@$zm?C)=P6QwI%p~#Oo!6pG>1;06KNiuLfy207EuqarC!=V7tqDD
zg)XJ7bOl{SyXk7Wmfl1+(_86n^bWd}ZlgQsLv$B?m_9}K(dX%l^i}#-`a1m^?W6C}
zWAr%vik_t3&~NEE`h(qRA7Q`TKGVL?9<{Huud=VUueEQrZ?ivSf6U%zKWhKL{-OP2
z`=<_c2o9TrIAlkfW2j@eV}xUzV~%6KV~t~z;}ORm$2*Qs9G^K(#QPjrEanwAnb*Xx
H<JA8EDt*1p

literal 0
HcmV?d00001

diff --git a/Grid/Grid.xcodeproj/xcuserdata/pab.xcuserdatad/xcschemes/Grid.xcscheme b/Grid/Grid.xcodeproj/xcuserdata/pab.xcuserdatad/xcschemes/Grid.xcscheme
new file mode 100644
index 00000000..b59f83df
--- /dev/null
+++ b/Grid/Grid.xcodeproj/xcuserdata/pab.xcuserdatad/xcschemes/Grid.xcscheme
@@ -0,0 +1,43 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<Scheme
+   LastUpgradeVersion = "0610"
+   version = "1.3">
+   <BuildAction
+      parallelizeBuildables = "YES"
+      buildImplicitDependencies = "YES">
+   </BuildAction>
+   <TestAction
+      selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
+      selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
+      shouldUseLaunchSchemeArgsEnv = "YES"
+      buildConfiguration = "Debug">
+      <Testables>
+      </Testables>
+   </TestAction>
+   <LaunchAction
+      selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
+      selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
+      launchStyle = "0"
+      useCustomWorkingDirectory = "NO"
+      buildConfiguration = "Debug"
+      ignoresPersistentStateOnLaunch = "NO"
+      debugDocumentVersioning = "YES"
+      allowLocationSimulation = "YES">
+      <AdditionalOptions>
+      </AdditionalOptions>
+   </LaunchAction>
+   <ProfileAction
+      shouldUseLaunchSchemeArgsEnv = "YES"
+      savedToolIdentifier = ""
+      useCustomWorkingDirectory = "NO"
+      buildConfiguration = "Release"
+      debugDocumentVersioning = "YES">
+   </ProfileAction>
+   <AnalyzeAction
+      buildConfiguration = "Debug">
+   </AnalyzeAction>
+   <ArchiveAction
+      buildConfiguration = "Release"
+      revealArchiveInOrganizer = "YES">
+   </ArchiveAction>
+</Scheme>
diff --git a/Grid/Grid.xcodeproj/xcuserdata/pab.xcuserdatad/xcschemes/xcschememanagement.plist b/Grid/Grid.xcodeproj/xcuserdata/pab.xcuserdatad/xcschemes/xcschememanagement.plist
new file mode 100644
index 00000000..1d4ba5a6
--- /dev/null
+++ b/Grid/Grid.xcodeproj/xcuserdata/pab.xcuserdatad/xcschemes/xcschememanagement.plist
@@ -0,0 +1,22 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
+<plist version="1.0">
+<dict>
+	<key>SchemeUserState</key>
+	<dict>
+		<key>Grid.xcscheme</key>
+		<dict>
+			<key>orderHint</key>
+			<integer>0</integer>
+		</dict>
+	</dict>
+	<key>SuppressBuildableAutocreation</key>
+	<dict>
+		<key>0BF5DA411AC370840045EA34</key>
+		<dict>
+			<key>primary</key>
+			<true/>
+		</dict>
+	</dict>
+</dict>
+</plist>
diff --git a/Grid_Cartesian.h b/Grid_Cartesian.h
index 488d47e9..bfe301a9 100644
--- a/Grid_Cartesian.h
+++ b/Grid_Cartesian.h
@@ -1,10 +1,9 @@
 #ifndef GRID_CARTESIAN_H
 #define GRID_CARTESIAN_H
 
-#include "Grid.h"
-
+#include <Grid.h>
+#include <Grid_Communicator.h>
 namespace dpo{
-
     
 /////////////////////////////////////////////////////////////////////////////////////////
 // Grid Support. Following will go into Grid.h.
@@ -13,28 +12,47 @@ namespace dpo{
     // dpo::Grid
     // dpo::GridCartesian
     // dpo::GridCartesianRedBlack
-    
-class Grid {
+
+
+class SimdGrid : public CartesianCommunicator {
 public:
+
+  
+ SimdGrid(std::vector<int> & processor_grid) : CartesianCommunicator(processor_grid) {};
+
     // Give Lattice access
     template<class object> friend class Lattice;
-
         
 //protected:
         
-    // Lattice wide random support. not yet fully implemented. Need seed strategy
-    // and one generator per site.
-    //std::default_random_engine generator;
+  // Lattice wide random support. not yet fully implemented. Need seed strategy
+  // and one generator per site.
+  //std::default_random_engine generator;
   //    static std::mt19937  generator( 9 );
-
         
     // Grid information.
-    unsigned long _ndimension;
-    std::vector<int> _layout;     // Which dimensions get relayed out over simd lanes.
-    std::vector<int> _dimensions; // Dimensions of array
-    std::vector<int> _rdimensions;// Reduced dimensions with simd lane images removed
+
+    // Commicator provides
+    //    unsigned long _ndimension;
+    //    std::vector<int> _processors; // processor grid
+    //    int              _processor;  // linear processor rank
+    //    std::vector<int> _processor_coor;  // linear processor rank
+
+    std::vector<int> _simd_layout;     // Which dimensions get relayed out over simd lanes.
+
+
+    std::vector<int> _fdimensions;// Global dimensions of array with cb removed
+    std::vector<int> _gdimensions;// Global dimensions of array
+    std::vector<int> _ldimensions;// local dimensions of array with processor images removed
+    std::vector<int> _rdimensions;// Reduced local dimensions with simd lane images and processor images removed 
+
+    //    std::vector<int> _lstart;     // local start of array in gcoors. _processor_coor[d]*_ldimensions[d]
+    //    std::vector<int> _lend;       // local end of array in gcoors    _processor_coor[d]*_ldimensions[d]+_ldimensions_[d]-1
+
+ 
     std::vector<int> _ostride;    // Outer stride for each dimension
     std::vector<int> _istride;    // Inner stride i.e. within simd lane
+
     int _osites;                  // _isites*_osites = product(dimensions).
     int _isites;
         
@@ -70,15 +88,33 @@ public:
         for(int d=0;d<_ndimension;d++) idx+=_istride[d]*(rcoor[d]/_rdimensions[d]);
         return idx;
     }
-        
+    
     inline int oSites(void) { return _osites; };
     inline int iSites(void) { return _isites; };
+
+    inline int CheckerboardFromOsite (int Osite){
+      std::vector<int> ocoor;
+      CoordFromOsite(ocoor,Osite);
+      int ss=0;
+      for(int d=0;d<_ndimension;d++){
+	ss=ss+ocoor[d];
+      }      
+      return ss&0x1;
+    }
+    inline void CoordFromOsite (std::vector<int>& coor,int Osite){
+      coor.resize(_ndimension);
+      for(int d=0;d<_ndimension;d++){
+	coor[d] = Osite % _rdimensions[d];
+	Osite   = Osite / _rdimensions[d];
+      }
+    }
+
     virtual int CheckerBoard(std::vector<int> site)=0;
     virtual int CheckerBoardDestination(int source_cb,int shift)=0;
-    virtual int CheckerBoardShift(int source_cb,int dim,int shift)=0;
+    virtual int CheckerBoardShift(int source_cb,int dim,int shift,int osite)=0;
 };
 
-class GridCartesian: public Grid {
+class GridCartesian: public SimdGrid {
 public:
     virtual int CheckerBoard(std::vector<int> site){
         return 0;
@@ -86,19 +122,24 @@ public:
     virtual int CheckerBoardDestination(int cb,int shift){
         return 0;
     }
-    virtual int CheckerBoardShift(int source_cb,int dim,int shift){
+    virtual int CheckerBoardShift(int source_cb,int dim,int shift, int osite){
         return shift;
     }
-    GridCartesian(std::vector<int> &dimensions,std::vector<int> layout)
+    GridCartesian(std::vector<int> &dimensions,
+		  std::vector<int> &simd_layout,
+		  std::vector<int> &processor_grid
+		  ) : SimdGrid(processor_grid)
     {
         ///////////////////////
         // Grid information
         ///////////////////////
         _ndimension = dimensions.size();
             
-        _dimensions.resize(_ndimension);
+        _fdimensions.resize(_ndimension);
+        _gdimensions.resize(_ndimension);
+        _ldimensions.resize(_ndimension);
         _rdimensions.resize(_ndimension);
-        _layout.resize(_ndimension);
+        _simd_layout.resize(_ndimension);
             
         _ostride.resize(_ndimension);
         _istride.resize(_ndimension);
@@ -106,21 +147,26 @@ public:
         _osites = 1;
         _isites = 1;
         for(int d=0;d<_ndimension;d++){
-            _dimensions[d] = dimensions[d];
-            _layout[d]     = layout[d];
-            // Use a reduced simd grid
-            _rdimensions[d]= _dimensions[d]/_layout[d]; //<-- _layout[d] is zero
-            _osites *= _rdimensions[d];
-            _isites *= _layout[d];
+	  _fdimensions[d] = dimensions[d]; // Global dimensions
+	  _gdimensions[d] = _fdimensions[d]; // Global dimensions
+	  _simd_layout[d] = simd_layout[d];
+
+	  //FIXME check for exact division
+
+	  // Use a reduced simd grid
+	  _ldimensions[d]= _gdimensions[d]/_processors[d];  //local dimensions
+	  _rdimensions[d]= _ldimensions[d]/_simd_layout[d]; //overdecomposition
+	  _osites *= _rdimensions[d];
+	  _isites *= _simd_layout[d];
                 
-            // Addressing support
-            if ( d==0 ) {
-                _ostride[d] = 1;
-                _istride[d] = 1;
-            } else {
-                _ostride[d] = _ostride[d-1]*_rdimensions[d-1];
-                _istride[d] = _istride[d-1]*_layout[d-1];
-            }
+	  // Addressing support
+	  if ( d==0 ) {
+	    _ostride[d] = 1;
+	    _istride[d] = 1;
+	  } else {
+	    _ostride[d] = _ostride[d-1]*_rdimensions[d-1];
+	    _istride[d] = _istride[d-1]*_simd_layout[d-1];
+	  }
         }
         
         ///////////////////////
@@ -148,46 +194,57 @@ public:
         }
     };
 };
-
+ 
 // Specialise this for red black grids storing half the data like a chess board.
-class GridRedBlackCartesian : public Grid
+class GridRedBlackCartesian : public SimdGrid
 {
 public:
     virtual int CheckerBoard(std::vector<int> site){
-        return site[0]&0x1;
+      return (site[0]+site[1]+site[2]+site[3])&0x1;
     }
-    virtual int CheckerBoardShift(int source_cb,int dim,int shift){
-        if ( dim == 0 ){
-            int fulldim =2*_dimensions[0];
-            shift = (shift+fulldim)%fulldim;
-            if ( source_cb ) {
-                // Shift 0,1 -> 0
-                return (shift+1)/2;
-            } else {
-                // Shift 0->0, 1 -> 1, 2->1
-                return (shift)/2;
-            }
-        } else {
-            return shift;
-        }
+
+    // Depending on the cb of site, we toggle source cb.
+    // for block #b, element #e = (b, e)
+    // we need 
+    virtual int CheckerBoardShift(int source_cb,int dim,int shift,int osite){
+
+      if(dim != 0) return shift;
+
+      int fulldim =_fdimensions[0];
+      shift = (shift+fulldim)%fulldim;
+
+      // Probably faster with table lookup;
+      // or by looping over x,y,z and multiply rather than computing checkerboard.
+      int ocb=CheckerboardFromOsite(osite);
+	  
+      if ( (source_cb+ocb)&1 ) {
+	return (shift)/2;
+      } else {
+	return (shift+1)/2;
+      }
     }
+
     virtual int CheckerBoardDestination(int source_cb,int shift){
-        if ((shift+2*_dimensions[0])&0x1) {
+        if ((shift+_fdimensions[0])&0x1) {
             return 1-source_cb;
         } else {
             return source_cb;
         }
     };
-    GridRedBlackCartesian(std::vector<int> &dimensions,std::vector<int> layout)
+    GridRedBlackCartesian(std::vector<int> &dimensions,
+			  std::vector<int> &simd_layout,
+			  std::vector<int> &processor_grid) : SimdGrid(processor_grid)
     {
     ///////////////////////
     // Grid information
     ///////////////////////
         _ndimension = dimensions.size();
         
-        _dimensions.resize(_ndimension);
+        _fdimensions.resize(_ndimension);
+        _gdimensions.resize(_ndimension);
+        _ldimensions.resize(_ndimension);
         _rdimensions.resize(_ndimension);
-        _layout.resize(_ndimension);
+        _simd_layout.resize(_ndimension);
         
         _ostride.resize(_ndimension);
         _istride.resize(_ndimension);
@@ -195,14 +252,17 @@ public:
         _osites = 1;
         _isites = 1;
         for(int d=0;d<_ndimension;d++){
-            _dimensions[d] = dimensions[d];
-            _dimensions[0] = dimensions[0]/2;
-            _layout[d]     = layout[d];
+            _fdimensions[d] = dimensions[d];
+            _gdimensions[d] = _fdimensions[d];
+            if (d==0) _gdimensions[0] = _gdimensions[0]/2; // Remove a checkerboard
+            _ldimensions[d] = _gdimensions[d]/_processors[d];
                 
             // Use a reduced simd grid
-            _rdimensions[d]= _dimensions[d]/_layout[d];
+            _simd_layout[d] = simd_layout[d];
+            _rdimensions[d]= _ldimensions[d]/_simd_layout[d];
+
             _osites *= _rdimensions[d];
-            _isites *= _layout[d];
+            _isites *= _simd_layout[d];
                 
             // Addressing support
             if ( d==0 ) {
@@ -210,7 +270,7 @@ public:
                 _istride[d] = 1;
             } else {
                 _ostride[d] = _ostride[d-1]*_rdimensions[d-1];
-                _istride[d] = _istride[d-1]*_layout[d-1];
+                _istride[d] = _istride[d-1]*_simd_layout[d-1];
             }
         }
             
diff --git a/Grid_Communicator.h b/Grid_Communicator.h
new file mode 100644
index 00000000..5be6f26f
--- /dev/null
+++ b/Grid_Communicator.h
@@ -0,0 +1,39 @@
+#ifndef GRID_COMMUNICATOR_H
+#define GRID_COMMUNICATOR_H
+///////////////////////////////////
+// Processor layout information
+///////////////////////////////////
+namespace dpo {
+class CartesianCommunicator {
+  public:    
+
+  // Communicator should know nothing of the physics grid, only processor grid.
+
+    std::vector<int> _processors;      // Which dimensions get relayed out over processors lanes.
+    int              _processor;       // linear processor rank
+    std::vector<int> _processor_coor;  // linear processor coordinate
+    unsigned long _ndimension;
+
+#ifdef GRID_COMMS_MPI
+    MPI_Comm communicator;
+#endif
+
+    CartesianCommunicator(std::vector<int> &pdimensions_in);
+
+    int  Rank(std::vector<int> coor);
+    void GlobalSumF(float &);
+    void GlobalSumFVector(float *,int N);
+
+    void GlobalSumF(double &);
+    void GlobalSumFVector(double *,int N);
+
+    void SendToRecvFrom(void *xmit,
+			std::vector<int> to_coordinate,
+			void *recv,
+			std::vector<int> from_coordinate,
+			int bytes);
+    
+}; 
+}
+
+#endif
diff --git a/Grid_Lattice.h b/Grid_Lattice.h
index 223ad2cd..8c8dc386 100644
--- a/Grid_Lattice.h
+++ b/Grid_Lattice.h
@@ -3,19 +3,31 @@
 
 #include "Grid.h"
 
+
+
 namespace dpo {
 
+// Permute the pointers 32bitx16 = 512
+static int permute_map[4][16] = { 
+  { 1,0,3,2,5,4,7,6,9,8,11,10,13,12,15,14},
+  { 2,3,0,1,6,7,4,5,10,11,8,9,14,15,12,13},
+  { 4,5,6,7,0,1,2,3,12,13,14,15,8,9,10,11},
+  { 9,10,11,12,13,14,15,0,1,2,3,4,5,6,7,8}
+};
+
+
 template<class vobj>
 class Lattice
 {
 public:
-    Grid *_grid;
+    SimdGrid *_grid;
     int checkerboard;
     std::vector<vobj,alignedAllocator<vobj> > _odata;
 
 public:
-    
-    Lattice(Grid *grid) : _grid(grid) {
+
+
+    Lattice(SimdGrid *grid) : _grid(grid) {
         _odata.reserve(_grid->oSites());
         if ( ((uint64_t)&_odata[0])&0xF) {
             exit(-1);
@@ -23,6 +35,19 @@ public:
         checkerboard=0;
     }
     
+
+#ifdef GRID_COMMS_NONE
+#include <Grid_none_cshift.h>
+#endif
+
+#ifdef GRID_COMMS_FAKE
+#include <Grid_fake_cshift.h>
+#endif
+
+#ifdef GRID_COMMS_MPI
+#include <Grid_mpi_cshift.h>
+#endif 
+
     
     // overloading dpo::conformable but no conformable in dpo ...?:w
     template<class obj1,class obj2>
@@ -69,212 +94,6 @@ public:
         return ret;
     };
 
-#if 0 
-    // Collapse doesn't appear to work the way I think it should in icpc
-    friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
-    {
-        Lattice<vobj> ret(rhs._grid);
-        
-        ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
-        shift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift);
-        int sx,so,o;
-        int rd = rhs._grid->_rdimensions[dimension];
-        int ld = rhs._grid->_dimensions[dimension];
-
-        // Map to always positive shift.
-        shift = (shift+ld)%ld;
-
-        // Work out whether to permute and the permute type
-        // ABCDEFGH ->   AE BF CG DH       permute
-        // Shift 0       AE BF CG DH       0 0 0 0    ABCDEFGH
-        // Shift 1       DH AE BF CG       1 0 0 0    HABCDEFG
-        // Shift 2       CG DH AE BF       1 1 0 0    GHABCDEF
-        // Shift 3       BF CG DH AE       1 1 1 0    FGHACBDE
-        // Shift 4       AE BF CG DH       1 1 1 1    EFGHABCD
-        // Shift 5       DH AE BF CG       0 1 1 1    DEFGHABC
-        // Shift 6       CG DH AE BF       0 0 1 1    CDEFGHAB
-        // Shift 7       BF CG DH AE       0 0 0 1    BCDEFGHA
-        int permute_dim =rhs._grid->_layout[dimension]>1 ;
-        int permute_type=0;
-        for(int d=0;d<dimension;d++)
-            if (rhs._grid->_layout[d]>1 ) permute_type++;
-
-        
-        // loop over perp slices.
-        // Threading considerations:
-        //   Need to map thread_num to
-        //
-        //               x_min,x_max for Loop-A
-        //               n_min,n_max for Loop-B
-        //               b_min,b_max for Loop-C
-        //  In a way that maximally load balances.
-        //
-        //  Optimal:
-        //      There are rd*n_block*block items of work.
-        //      These serialise as item "w"
-        //      b=w%block; w=w/block
-        //      n=w%nblock; x=w/nblock. Perhaps 20 cycles?
-        //
-        //  Logic:
-        //      x_chunk = (rd+thread)/nthreads simply divide work across nodes.
-        //
-        //      rd=5 , threads = 8;
-        //      0 1 2 3 4 5 6 7
-        //      0 0 0 1 1 1 1 1
-        for(int x=0;x<rd;x++){         // Loop A
-            sx = (x-shift+ld)%rd;
-            o  = x*rhs._grid->_ostride[dimension];
-            so =sx*rhs._grid->_ostride[dimension];
-
-            int permute_slice=0;
-            if ( permute_dim ) {
-                permute_slice = shift/rd;
-                if ( x<shift%rd ) permute_slice = 1-permute_slice;
-            }
-
-
-#if 0
-            if ( permute_slice ) {
-                
-                int internal=sizeof(vobj)/sizeof(vComplex);
-                int num =rhs._grid->_slice_block[dimension]*internal;
-                
-                for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-                    vComplex *optr = (vComplex *)&ret._odata[o];
-                    vComplex *iptr = (vComplex *)&rhs._odata[so];
-                    for(int b=0;b<num;b++){
-                        permute(optr[b],iptr[b],permute_type);
-                    }
-                    o+=rhs._grid->_slice_stride[dimension];
-                    so+=rhs._grid->_slice_stride[dimension];
-                }
-            } else {
-                for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-                    for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
-                        ret._odata[o+i]=rhs._odata[so+i];
-                    }
-                    o+=rhs._grid->_slice_stride[dimension];
-                    so+=rhs._grid->_slice_stride[dimension];
-                }
-            
-            }
-#else
-
-            if ( permute_slice ) {
-
-	        int internal=sizeof(vobj)/sizeof(vComplex);
-                int num =rhs._grid->_slice_block[dimension]*internal;
-
-
-#pragma omp parallel for collapse(2)
-                for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-                    for(int b=0;b<num;b++){
-		      vComplex *optr = (vComplex *)&ret._odata[o +n*rhs._grid->_slice_stride[dimension]];
-		      vComplex *iptr = (vComplex *)&rhs._odata[so+n*rhs._grid->_slice_stride[dimension]];
-                      permute(optr[b],iptr[b],permute_type);
-                    }
-                }
-
-            } else {
-
-
-#pragma omp parallel for collapse(2)
-                for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-                    for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
-		      int oo = o+ n*rhs._grid->_slice_stride[dimension];
-		      int soo=so+ n*rhs._grid->_slice_stride[dimension];
-		      ret._odata[oo+i]=rhs._odata[soo+i];
-                    }
-                }
-            
-            }
-#endif
-        }
-        return ret;
-    }
-#else
-    friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
-    {
-        Lattice<vobj> ret(rhs._grid);
-        
-        ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
-        shift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift);
-        int rd = rhs._grid->_rdimensions[dimension];
-        int ld = rhs._grid->_dimensions[dimension];
-        
-        // Map to always positive shift.
-        shift = (shift+ld)%ld;
-        
-        // Work out whether to permute and the permute type
-        // ABCDEFGH ->   AE BF CG DH       permute
-        // Shift 0       AE BF CG DH       0 0 0 0    ABCDEFGH
-        // Shift 1       DH AE BF CG       1 0 0 0    HABCDEFG
-        // Shift 2       CG DH AE BF       1 1 0 0    GHABCDEF
-        // Shift 3       BF CG DH AE       1 1 1 0    FGHACBDE
-        // Shift 4       AE BF CG DH       1 1 1 1    EFGHABCD
-        // Shift 5       DH AE BF CG       0 1 1 1    DEFGHABC
-        // Shift 6       CG DH AE BF       0 0 1 1    CDEFGHAB
-        // Shift 7       BF CG DH AE       0 0 0 1    BCDEFGHA
-        int permute_dim =rhs._grid->_layout[dimension]>1 ;
-        int permute_type=0;
-        for(int d=0;d<dimension;d++)
-            if (rhs._grid->_layout[d]>1 ) permute_type++;
-        
-        
-        // loop over all work
-        int work =rd*rhs._grid->_slice_nblock[dimension]*rhs._grid->_slice_block[dimension];
-
-#pragma omp parallel for
-        for(int ww=0;ww<work;ww++){
-
-            
-            // can optimise this if know w moves congtiguously for a given thread.
-            // b=(b+1);
-            // if (b==_slice_block) {b=0; n=n+1;}
-            // if (n==_slice_nblock) { n=0; x=x+1}
-            //
-            // Perhaps a five cycle iterator, or so.
-            int w=ww;
-            int b = w%rhs._grid->_slice_block[dimension] ; w=w/rhs._grid->_slice_block[dimension];
-            int n = w%rhs._grid->_slice_nblock[dimension]; w=w/rhs._grid->_slice_nblock[dimension];
-            int x = w;
-
-	    int sx,so,o;
-            sx = (x-shift+ld)%rd;
-            o  = x*rhs._grid->_ostride[dimension]+n*rhs._grid->_slice_stride[dimension]; // common sub expression alert.
-            so =sx*rhs._grid->_ostride[dimension]+n*rhs._grid->_slice_stride[dimension];
-            
-            int permute_slice=0;
-            if ( permute_dim ) {
-                permute_slice = shift/rd;
-                if ( x<shift%rd ) permute_slice = 1-permute_slice;
-            }
-            
-            if ( permute_slice ) {
-                
-                int internal=sizeof(vobj)/sizeof(vComplex);
-                vComplex *optr = (vComplex *)&ret._odata[o+b];
-                vComplex *iptr = (vComplex *)&rhs._odata[so+b];
-		const char *pf = (const char *)iptr;
-		for(int i=0;i<sizeof(vobj);i+=64){
-		  _mm_prefetch(pf+i,_MM_HINT_T0);
-		}
-
-                for(int i=0;i<internal;i++){
-                    permute(optr[i],iptr[i],permute_type);
-                }
-            } else {
-	        const char *pf = (const char *) &rhs._odata[so+b];
-		for(int i=0;i<sizeof(vobj);i+=64){
-		  _mm_prefetch(pf+i,_MM_HINT_T0);
-		}
-                ret._odata[o+b]=rhs._odata[so+b];
-            }
-        }
-        return ret;
-    }
-#endif
-
     template<class sobj>
     inline Lattice<vobj> & operator = (const sobj & r){
 #pragma omp parallel for
@@ -288,14 +107,15 @@ public:
     template<class sobj>
     friend void pokeSite(const sobj &s,Lattice<vobj> &l,std::vector<int> &site){
 
-        if ( l._grid.checkerboard != l._grid->Checkerboard(site)){
-            printf("Poking wrong checkerboard\n");
-            exit(EXIT_FAILURE);
+        if ( l.checkerboard != l._grid->CheckerBoard(site)){
+	  printf("Poking wrong checkerboard\n");
+	  exit(EXIT_FAILURE);
         }
 
         int o_index = l._grid->oIndex(site);
         int i_index = l._grid->iIndex(site);
         
+	// BUGGY. This assumes complex real
         Real *v_ptr = (Real *)&l._odata[o_index];
         Real *s_ptr = (Real *)&s;
         v_ptr = v_ptr + 2*i_index;
@@ -348,6 +168,22 @@ public:
         }
     };
 
+    friend void lex_sites(Lattice<vobj> &l){
+
+        Real *v_ptr = (Real *)&l._odata[0];
+	size_t o_len = l._grid->oSites();
+        size_t v_len = sizeof(vobj)/sizeof(vRealF);
+	size_t vec_len = vRealF::Nsimd();
+
+        for(int i=0;i<o_len;i++){
+        for(int j=0;j<v_len;j++){
+	  for(int vv=0;vv<vec_len;vv+=2){
+            v_ptr[i*v_len*vec_len+j*vec_len+vv  ]= i+vv*500;
+            v_ptr[i*v_len*vec_len+j*vec_len+vv+1]= i+vv*500;
+	  }
+        }}
+    };
+
     friend void gaussian(Lattice<vobj> &l){
         // Zero mean, unit variance.
         std::normal_distribution<double> distribution(0.0,1.0);
@@ -427,22 +263,41 @@ public:
         return ret;
     };
 
-
-
     // remove and insert a half checkerboard
     friend void pickCheckerboard(int cb,Lattice<vobj> &half,const Lattice<vobj> &full){
+      half.checkerboard = cb;
+      int ssh=0;
 #pragma omp parallel for
-        for(int ss=0;ss<half._grid->oSites();ss++){
-            half._odata[ss] = full._odata[ss*2+cb];
-        }
-        half.checkerboard = cb;
+      for(int ss=0;ss<full._grid->oSites();ss++){
+	std::vector<int> coor;
+	int cbos;
+	
+	full._grid->CoordFromOsite(coor,ss);
+	cbos=half._grid->CheckerBoard(coor);
+	
+	if (cbos==cb) {
+
+	  half._odata[ssh] = full._odata[ss];
+	  ssh++;
+	}
+      }
     }
     friend void setCheckerboard(Lattice<vobj> &full,const Lattice<vobj> &half){
-        int cb = half.checkerboard;
+      int cb = half.checkerboard;
+      int ssh=0;
 #pragma omp parallel for
-        for(int ss=0;ss<half._grid->oSites();ss++){
-            full._odata[ss*2+cb]=half._odata[ss];
-        }
+      for(int ss=0;ss<full._grid->oSites();ss++){
+	std::vector<int> coor;
+	int cbos;
+	
+	full._grid->CoordFromOsite(coor,ss);
+	cbos=half._grid->CheckerBoard(coor);
+
+	if (cbos==cb) {
+	  full._odata[ss]=half._odata[ssh];
+	  ssh++;
+	}
+      }
     }
 }; // class Lattice
 
diff --git a/Grid_QCD.h b/Grid_QCD.h
index 92a02506..e51e2641 100644
--- a/Grid_QCD.h
+++ b/Grid_QCD.h
@@ -5,7 +5,6 @@ namespace QCD {
 
     static const int Nc=3;
     static const int Ns=4;
-
     static const int CbRed  =0;
     static const int CbBlack=1;
 
diff --git a/Grid_config.h b/Grid_config.h
index 64003419..2a442ea5 100644
--- a/Grid_config.h
+++ b/Grid_config.h
@@ -10,6 +10,15 @@
 /* AVX512 */
 /* #undef AVX512 */
 
+/* GRID_COMMS_FAKE */
+#define GRID_COMMS_FAKE 1
+
+/* GRID_COMMS_MPI */
+/* #undef GRID_COMMS_MPI */
+
+/* GRID_COMMS_NONE */
+/* #undef GRID_COMMS_NONE */
+
 /* Define to 1 if you have the `gettimeofday' function. */
 #define HAVE_GETTIMEOFDAY 1
 
@@ -61,6 +70,9 @@
 /* Define to the one symbol short name of this package. */
 #define PACKAGE_TARNAME "grid"
 
+/* Define to the home page for this package. */
+#define PACKAGE_URL ""
+
 /* Define to the version of this package. */
 #define PACKAGE_VERSION "1.0"
 
diff --git a/Grid_fake.cc b/Grid_fake.cc
new file mode 100644
index 00000000..7e664f33
--- /dev/null
+++ b/Grid_fake.cc
@@ -0,0 +1,34 @@
+#include "Grid.h"
+namespace dpo {
+
+CartesianCommunicator::CartesianCommunicator(std::vector<int> &processors)
+{
+  _ndimension = _processors.size();
+  _processor_coor.resize(_ndimension);
+  _processors = processors;
+  
+  // Require 1^N processor grid for fake
+  for(int d=0;d<_ndimension;d++) if(_processors[d]!=1) exit(-1);
+
+  _processor = 0;// I am the one. The only one..
+  for(int d=0;d<_ndimension;d++) _processor_coor[d] = 0;
+}
+
+void CartesianCommunicator::GlobalSumF(float &){}
+void CartesianCommunicator::GlobalSumFVector(float *,int N){}
+void CartesianCommunicator::GlobalSumF(double &){}
+void CartesianCommunicator::GlobalSumFVector(double *,int N){}
+
+// Basic Halo comms primitive
+void CartesianCommunicator::SendToRecvFrom(void *xmit,
+		    std::vector<int> to_coordinate,
+		    void *recv,
+		    std::vector<int> from_coordinate,
+		    int bytes)
+{
+  exit(-1);
+  bcopy(xmit,recv,bytes);
+}
+
+}
+
diff --git a/Grid_fake_cshift.h b/Grid_fake_cshift.h
new file mode 100644
index 00000000..0f306b1b
--- /dev/null
+++ b/Grid_fake_cshift.h
@@ -0,0 +1,246 @@
+#ifndef _GRID_FAKE_H_
+#define _GRID_FAKE_H_
+
+      
+
+friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
+{
+  typedef typename vobj::vector_type vector_type;
+  typedef typename vobj::scalar_type scalar_type;
+  const int Nsimd = vector_type::Nsimd();
+
+  Lattice<vobj> ret(rhs._grid);
+  
+  int fd = rhs._grid->_fdimensions[dimension];
+  int rd = rhs._grid->_rdimensions[dimension];
+  //int ld = rhs._grid->_ldimensions[dimension];
+  //int gd = rhs._grid->_gdimensions[dimension];
+  
+
+  // Map to always positive shift modulo global full dimension.
+  shift = (shift+fd)%fd;
+
+  ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
+        
+  // the permute type
+
+  int permute_dim =rhs._grid->_simd_layout[dimension]>1 ;
+  int permute_type=0;
+  for(int d=0;d<dimension;d++){
+    if (rhs._grid->_simd_layout[d]>1 ) permute_type++;
+  }
+
+  ///////////////////////////////////////////////
+  // Move via a fake comms buffer
+  // Simd direction uses an extract/merge pair
+  ///////////////////////////////////////////////
+  int buffer_size = rhs._grid->_slice_nblock[dimension]*rhs._grid->_slice_block[dimension];
+  int words = sizeof(vobj)/sizeof(vector_type);
+
+  std::vector<vobj,alignedAllocator<vobj> > comm_buf(buffer_size);
+  std::vector<std::vector<scalar_type> > comm_buf_extract(Nsimd,std::vector<scalar_type>(buffer_size*words) );
+  std::vector<scalar_type *> pointers(Nsimd);
+
+  for(int x=0;x<rd;x++){       
+
+    for(int i=0;i<vobj::vector_type::Nsimd();i++){
+      pointers[i] = (scalar_type *)&comm_buf_extract[i][0];
+    }
+
+    int ro  = x*rhs._grid->_ostride[dimension]; // base offset for result
+
+    if ( permute_dim ) {
+
+      int o   = 0;                                // relative offset to base
+      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+	  
+	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
+	  
+	  int sx = (x+sshift)%rd;
+	  
+	  // base offset for source
+	  int so = sx*rhs._grid->_ostride[dimension];
+
+	  int permute_slice=0;
+	  int wrap = sshift/rd;
+	  int  num = sshift%rd;
+	    
+	  if ( x< rd-num ) permute_slice=wrap;
+	  else permute_slice = 1-wrap;
+
+	  if ( permute_slice ) {
+	    extract(rhs._odata[so+o+b],pointers);
+	  } else {
+	    ret._odata[ro+o+b]=rhs._odata[so+o+b];
+	  }
+
+	}
+	o +=rhs._grid->_slice_stride[dimension];
+      }
+
+      for(int i=0;i<vobj::vector_type::Nsimd();i++){
+	pointers[i] = (scalar_type *)&comm_buf_extract[permute_map[permute_type][i]][0];
+      }
+
+      o   = 0;                                // relative offset to base
+      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+	  
+	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
+	  int sx = (x+sshift)%rd;
+	  
+	  // base offset for source
+	  int so = sx*rhs._grid->_ostride[dimension];
+
+	  int permute_slice=0;
+	  int wrap = sshift/rd;
+	  int  num = sshift%rd;
+	    
+	  if ( x< rd-num ) permute_slice=wrap;
+	  else permute_slice = 1-wrap;
+
+	  if ( permute_slice ) {
+	    merge(ret._odata[ro+o+b],pointers);
+	  }
+	}
+	o +=rhs._grid->_slice_stride[dimension];
+      }
+
+    } else {
+
+      int co; // comm offset
+      int o;
+
+      co=0; o=0;
+      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+
+	  // This call in inner loop is annoying but necessary for dimension=0
+	  // in the case of RedBlack grids. Could optimise away with 
+	  // alternate code paths for all other cases.
+	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
+	  int sx = (x+sshift)%rd;
+	  int so = sx*rhs._grid->_ostride[dimension];
+
+	  comm_buf[co++]=rhs._odata[so+o+b];
+
+	}
+	o +=rhs._grid->_slice_stride[dimension];
+      }
+
+      // Step through a copy into a comms buffer and pull back in.
+      // Genuine fake implementation could calculate if loops back
+      co=0; o=0;
+      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+	  ret._odata[ro+o+b]=comm_buf[co++];
+	}
+	o +=rhs._grid->_slice_stride[dimension];
+      }
+    }
+  }
+  return ret;
+}
+
+/*
+    friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
+    {
+      Lattice<vobj> ret(rhs._grid);
+        
+        int rd = rhs._grid->_rdimensions[dimension];
+        int ld = rhs._grid->_ldimensions[dimension];
+        int gd = rhs._grid->_gdimensions[dimension];
+        
+        // Map to always positive shift.
+        shift = (shift+gd)%gd;
+
+        ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
+        shift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift);
+        
+        // Work out whether to permute and the permute type
+        // ABCDEFGH ->   AE BF CG DH       permute
+        // Shift 0       AE BF CG DH       0 0 0 0    ABCDEFGH
+        // Shift 1       BF CG DH AE       0 0 0 1    BCDEFGHA
+        // Shift 2       CG DH AE BF       0 0 1 1    CDEFGHAB
+        // Shift 3       DH AE BF CG       0 1 1 1    DEFGHABC
+        // Shift 4       AE BF CG DH       1 1 1 1    EFGHABCD
+        // Shift 5       BF CG DH AE       1 1 1 0    FGHACBDE
+        // Shift 6       CG DH AE BF       1 1 0 0    GHABCDEF
+        // Shift 7       DH AE BF CG       1 0 0 0    HABCDEFG
+
+        int permute_dim =rhs._grid->_simd_layout[dimension]>1 ;
+        int permute_type=0;
+        for(int d=0;d<dimension;d++)
+            if (rhs._grid->_simd_layout[d]>1 ) permute_type++;
+        
+        
+        // loop over all work
+        int work =rd*rhs._grid->_slice_nblock[dimension]*rhs._grid->_slice_block[dimension];
+
+	// Packed gather sequence is clean
+	int buffer_size = rhs._grid->_slice_nblock[dimension]*rhs._grid->_slice_block[dimension];
+	
+	typedef typename vobj::scalar_type scalar_t;
+	typedef typename vobj::vector_type vector_t;
+	const int ns=sizeof(vobj)/sizeof(scalar_t);
+	const int nv=sizeof(vobj)/sizeof(vector_t);
+	std::vector<vobj,alignedAllocator<vobj> > comm_buf(buffer_size);
+
+        for(int x=0;x<rd;x++){       
+	  
+	  int sx = (x+shift)%rd;
+	  int o  = x*rhs._grid->_ostride[dimension];
+	  int so =sx*rhs._grid->_ostride[dimension];
+
+	
+	  int permute_slice=0;
+	  if ( permute_dim ) {
+	    permute_slice = shift/rd;
+	    if ( x<shift%rd ) permute_slice = 1-permute_slice;
+	  }
+
+	  if ( permute_slice ) {
+	    exit(0);
+	    // For fake communication ALWAYS extract and either merge one way or other
+	    scalar_t * bptr = (scalar_t *) &comm_buf[0];
+
+	    int bo=0;
+	    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	    
+	      vector_t *optr = (vector_t *)&ret._odata[o];
+	      vector_t *iptr = (vector_t *)&rhs._odata[so];
+	      int skew    = buffer_size*ns/2;
+
+	      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){ 
+		for(int n=0;n<nv;n++){// number of simd vecsscalars in a vector
+		  extract(iptr[b*nv+n],&bptr[n],skew,permute_type);
+		}
+	      }
+	      o+=rhs._grid->_slice_stride[dimension];
+	      //	      bo+=rhs._grid->_slice_stride[dimension]*ns/2;
+
+	    }
+
+	  } else {
+	    int bo=0;
+	    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	      for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
+		comm_buf[bo++] =rhs._odata[so+i];
+	      }
+	      so+=rhs._grid->_slice_stride[dimension];
+	    }
+	    bo=0;
+	    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	      for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
+		ret._odata[o+i]=comm_buf[bo++];
+	      }
+	      o+=rhs._grid->_slice_stride[dimension];
+	    }
+	  }
+	}
+        return ret;
+    };
+*/
+
+#endif
diff --git a/Grid_init.cc b/Grid_init.cc
index e943c3b0..bb22ce29 100755
--- a/Grid_init.cc
+++ b/Grid_init.cc
@@ -28,13 +28,13 @@ double usecond(void) {
 
 void Grid_sa_signal_handler(int sig,siginfo_t *si,void * ptr)
 {
-         ucontext_t * uc= (ucontext_t *)ptr;
-
   printf("Caught signal %d\n",si->si_signo);
   printf("  mem address %llx\n",(uint64_t)si->si_addr);
   printf("         code %d\n",si->si_code);
 
 #ifdef __X86_64
+    ucontext_t * uc= (ucontext_t *)ptr;
+
   struct sigcontext *sc = (struct sigcontext *)&uc->uc_mcontext;
   printf("  instruction %llx\n",(uint64_t)sc->rip);
 
diff --git a/Grid_main.cc b/Grid_main.cc
index 0b0c3e99..4079e62a 100644
--- a/Grid_main.cc
+++ b/Grid_main.cc
@@ -14,7 +14,7 @@ using namespace dpo::QCD;
 int main (int argc, char ** argv)
 {
 
-#ifdef KNL
+#if 0
   struct sigaction sa,osa;
   sigemptyset (&sa.sa_mask);
   sa.sa_sigaction= sa_action;
@@ -27,6 +27,9 @@ int main (int argc, char ** argv)
     std::vector<int> latt_size(4);
     std::vector<int> simd_layout(4);
 
+    std::vector<int> mpi_layout(4);
+    for(int d=0;d<4;d++) mpi_layout[d]=1;
+
 #ifdef AVX512
  for(int omp=128;omp<236;omp+=16){
 #else
@@ -37,7 +40,7 @@ int main (int argc, char ** argv)
    omp_set_num_threads(omp);
 #endif 
 
-  for(int lat=16;lat<=20;lat+=4){
+  for(int lat=4;lat<=16;lat+=40){
     latt_size[0] = lat;
     latt_size[1] = lat;
     latt_size[2] = lat;
@@ -62,11 +65,9 @@ int main (int argc, char ** argv)
     simd_layout[2] = 1;
     simd_layout[3] = 2;
 #endif
-
     
-    GridCartesian Fine(latt_size,simd_layout);
-    GridRedBlackCartesian rbFine(latt_size,simd_layout);
-
+    GridCartesian Fine(latt_size,simd_layout,mpi_layout);
+    GridRedBlackCartesian rbFine(latt_size,simd_layout,mpi_layout);
     
     LatticeColourMatrix Foo(&Fine);
     LatticeColourMatrix Bar(&Fine);
@@ -152,19 +153,23 @@ int main (int argc, char ** argv)
     trscMat = trace(scMat); // Trace
     
     FooBar = Bar;
-    int shift=1;
-    int dir=0;
  
+    /*
+    { 
+      std::vector<int> coor(4);
+      for(int d=0;d<4;d++) coor[d] = 0;
+      peekSite(cmat,Foo,coor);
+      Foo = zero;
+      pokeSite(cmat,Foo,coor);
+    }
     random(Foo);
-    pickCheckerboard(CbRed,rFoo,Foo);    // Pick out red or black checkerboards
-    pickCheckerboard(CbBlack,bFoo,Foo);
+    */
+    lex_sites(Foo);
+
+
+    //setCheckerboard(ShiftedCheck,rFoo); 
+    //setCheckerboard(ShiftedCheck,bFoo); 
 
-    Shifted  = Cshift(Foo,dir,shift);    // Shift everything
-    bShifted = Cshift(rFoo,dir,shift);   // Shift red
-    rShifted = Cshift(bFoo,dir,shift);   // Shift black
-    
-    setCheckerboard(ShiftedCheck,bShifted); // Put them all together
-    setCheckerboard(ShiftedCheck,rShifted); // and check the results (later)
 
     // Lattice SU(3) x SU(3)
     FooBar = Foo * Bar;
@@ -211,25 +216,69 @@ int main (int argc, char ** argv)
     printf("Cshift Mult: NumThread %d , Lattice size %d , %f Mflop/s\n",omp,lat,flops/(t1-t0));
     printf("Cshift Mult: NumThread %d , Lattice size %d , %f MB/s\n",omp,lat,bytes/(t1-t0));
 
-    pickCheckerboard(0,rFoo,FooBar);
-    pickCheckerboard(1,bFoo,FooBar);
-    
-    setCheckerboard(FooBar,rFoo);
-    setCheckerboard(FooBar,bFoo);
+    //    pickCheckerboard(0,rFoo,FooBar);
+    //    pickCheckerboard(1,bFoo,FooBar);
+    //    setCheckerboard(FooBar,rFoo);
+    //    setCheckerboard(FooBar,bFoo);
     
     double nrm=0;
+
+    for(int dir=0;dir<4;dir++){
+      for(int shift=0;shift<latt_size[dir];shift++){
+
+
+
+	pickCheckerboard(0,rFoo,Foo);    // Pick out red or black checkerboards
+	pickCheckerboard(1,bFoo,Foo);
     
-    std::vector<int> coor(4);
-    for(coor[0]=0;coor[0]<latt_size[0];coor[0]++){
-    for(coor[1]=0;coor[1]<latt_size[1];coor[1]++){
-    for(coor[2]=0;coor[2]<latt_size[2];coor[2]++){
-    for(coor[3]=0;coor[3]<latt_size[3];coor[3]++){
+	std::cout << "Shifting both parities by "<< shift <<" direction "<< dir <<std::endl;
+	Shifted  = Cshift(Foo,dir,shift);    // Shift everything
+
+	std::cout << "Shifting even parities"<<std::endl;
+	bShifted = Cshift(rFoo,dir,shift);   // Shift red->black
+	std::cout << "Shifting odd parities"<<std::endl;
+	rShifted = Cshift(bFoo,dir,shift);   // Shift black->red
+    
+	ShiftedCheck=zero;
+	setCheckerboard(ShiftedCheck,bShifted); // Put them all together
+	setCheckerboard(ShiftedCheck,rShifted); // and check the results (later)
+    
+	// Check results
+	std::vector<int> coor(4);
+	for(coor[3]=0;coor[3]<latt_size[3];coor[3]++){
+	for(coor[2]=0;coor[2]<latt_size[2];coor[2]++){
+	for(coor[1]=0;coor[1]<latt_size[1];coor[1]++){
+	for(coor[0]=0;coor[0]<latt_size[0];coor[0]++){
  
         std::complex<dpo::Real> diff;
                     
         std::vector<int> shiftcoor = coor;
-        shiftcoor[dir]=(shiftcoor[dir]-shift+latt_size[dir])%latt_size[dir];
-                    
+        shiftcoor[dir]=(shiftcoor[dir]+shift+latt_size[dir])%latt_size[dir];
+
+	std::vector<int> rl(4);
+	for(int dd=0;dd<4;dd++){
+	  rl[dd] = latt_size[dd]/simd_layout[dd];
+	}
+	int lex =  coor[0]%rl[0]
+	  + (coor[1]%rl[1])*rl[0]
+	  + (coor[2]%rl[2])*rl[0]*rl[1]
+	  + (coor[3]%rl[3])*rl[0]*rl[1]*rl[2];
+	lex += 
+	  +1000*(coor[0]/rl[0])
+	  +1000*(coor[1]/rl[1])*simd_layout[0]
+	  +1000*(coor[2]/rl[2])*simd_layout[0]*simd_layout[1]
+	  +1000*(coor[3]/rl[3])*simd_layout[0]*simd_layout[1]*simd_layout[2];
+
+	int lex_coor = shiftcoor[0]%rl[0]
+	  + (shiftcoor[1]%rl[1])*rl[0]
+	  + (shiftcoor[2]%rl[2])*rl[0]*rl[1]
+	  + (shiftcoor[3]%rl[3])*rl[0]*rl[1]*rl[2];
+	lex_coor += 
+	  +1000*(shiftcoor[0]/rl[0])
+	  +1000*(shiftcoor[1]/rl[1])*simd_layout[0]
+	  +1000*(shiftcoor[2]/rl[2])*simd_layout[0]*simd_layout[1]
+	  +1000*(shiftcoor[3]/rl[3])*simd_layout[0]*simd_layout[1]*simd_layout[2];
+
         ColourMatrix foo;
         ColourMatrix bar;
         ColourMatrix shifted1;
@@ -242,7 +291,6 @@ int main (int argc, char ** argv)
         peekSite(shifted1,Shifted,coor);
         peekSite(shifted2,Foo,shiftcoor);
         peekSite(shifted3,ShiftedCheck,coor);
-       
         peekSite(foo,Foo,coor);
         
         mdiff = shifted1-shifted2;
@@ -258,13 +306,13 @@ int main (int argc, char ** argv)
             diff =shifted1._internal._internal[r][c]-shifted2._internal._internal[r][c];
             nn=real(conj(diff)*diff);
             if ( nn > 0 )
-                cout<<"Shift fail "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "
+                cout<<"Shift fail (shifted1/shifted2-ref) "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "
                     <<shifted1._internal._internal[r][c]<<" "<<shifted2._internal._internal[r][c]
-                    << " "<< foo._internal._internal[r][c]<<endl;
+                    << " "<< foo._internal._internal[r][c]<< " lex expect " << lex_coor << " lex "<<lex<<endl;
             else if(0)
-                cout<<"Shift pass "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "
+                cout<<"Shift pass 1vs2 "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "
                     <<shifted1._internal._internal[r][c]<<" "<<shifted2._internal._internal[r][c]
-                    << " "<< foo._internal._internal[r][c]<<endl;
+                    << " "<< foo._internal._internal[r][c]<< " lex expect " << lex_coor << " lex "<<lex<<endl;
         }}
         
         for(int r=0;r<3;r++){
@@ -272,13 +320,13 @@ int main (int argc, char ** argv)
             diff =shifted3._internal._internal[r][c]-shifted2._internal._internal[r][c];
             nn=real(conj(diff)*diff);
             if ( nn > 0 )
-                cout<<"Shift rb fail "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "
+                cout<<"Shift rb fail (shifted3/shifted2-ref) "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "
                 <<shifted3._internal._internal[r][c]<<" "<<shifted2._internal._internal[r][c]
-                << " "<< foo._internal._internal[r][c]<<endl;
+                << " "<< foo._internal._internal[r][c]<< " lex expect " << lex_coor << " lex "<<lex<<endl;
             else if(0)
-                cout<<"Shift rb pass "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "
+                cout<<"Shift rb pass 3vs2 "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "
                 <<shifted3._internal._internal[r][c]<<" "<<shifted2._internal._internal[r][c]
-                << " "<< foo._internal._internal[r][c]<<endl;
+                << " "<< foo._internal._internal[r][c]<< " lex expect " << lex_coor << " lex "<<lex<<endl;
         }}
         peekSite(bar,Bar,coor);
                     
@@ -290,7 +338,7 @@ int main (int argc, char ** argv)
             nrm = nrm + real(conj(diff)*diff);
         }}
     }}}}
-
+      }}
     std::cout << "LatticeColorMatrix * LatticeColorMatrix nrm diff = "<<nrm<<std::endl;
 
    } // loop for lat
diff --git a/Grid_math_types.h b/Grid_math_types.h
index 405d5c9a..e5387f12 100644
--- a/Grid_math_types.h
+++ b/Grid_math_types.h
@@ -1,6 +1,63 @@
 #ifndef GRID_MATH_TYPES_H
 #define GRID_MATH_TYPES_H
 namespace dpo {
+
+
+
+//////////////////////////////////////////////////////////////////////////////////
+// Want to recurse: GridTypeMapper<Matrix<vComplexD> >::scalar_type == ComplexD.
+//////////////////////////////////////////////////////////////////////////////////
+
+  template <class T> class GridTypeMapper {
+  public:
+    typedef typename T::scalar_type scalar_type;
+    typedef typename T::vector_type vector_type;
+  };
+
+  template<> class GridTypeMapper<RealF> {
+  public:
+    typedef RealF scalar_type;
+    typedef RealF vector_type;
+  };
+  template<> class GridTypeMapper<RealD> {
+  public:
+    typedef RealD scalar_type;
+    typedef RealD vector_type;
+  };
+  template<> class GridTypeMapper<ComplexF> {
+  public:
+    typedef ComplexF scalar_type;
+    typedef ComplexF vector_type;
+  };
+  template<> class GridTypeMapper<ComplexD> {
+  public:
+    typedef ComplexD scalar_type;
+    typedef ComplexD vector_type;
+  };
+
+  template<> class GridTypeMapper<vRealF> {
+  public:
+    typedef RealF  scalar_type;
+    typedef vRealF vector_type;
+  };
+  template<> class GridTypeMapper<vRealD> {
+  public:
+    typedef RealD  scalar_type;
+    typedef vRealD vector_type;
+  };
+  template<> class GridTypeMapper<vComplexF> {
+  public:
+    typedef ComplexF  scalar_type;
+    typedef vComplexF vector_type;
+  };
+  template<> class GridTypeMapper<vComplexD> {
+  public:
+    typedef ComplexD  scalar_type;
+    typedef vComplexD vector_type;
+  };
+
+
+
 ///////////////////////////////////////////////////
 // Scalar, Vector, Matrix objects.
 // These can be composed to form tensor products of internal indices.
@@ -10,6 +67,10 @@ template<class vtype> class iScalar
 {
 public:
   SIMDalign vtype _internal;
+
+  typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
+  typedef typename GridTypeMapper<vtype>::vector_type vector_type;
+
     iScalar(){};
     iScalar(Zero &z){ *this = zero; };
     iScalar<vtype> & operator= (const Zero &hero){
@@ -19,6 +80,15 @@ public:
     friend void zeroit(iScalar<vtype> &that){
         zeroit(that._internal);
     }
+    friend void permute(iScalar<vtype> &out,iScalar<vtype> &in,int permutetype){
+      permute(out._internal,in._internal,permutetype);
+    }
+    friend void extract(iScalar<vtype> &in,std::vector<scalar_type *> &out){
+      extract(in._internal,out); // extract advances the pointers in out
+    }
+    friend void merge(iScalar<vtype> &in,std::vector<scalar_type *> &out){
+      merge(in._internal,out); // extract advances the pointers in out
+    }
     // Unary negation
     friend inline iScalar<vtype> operator -(const iScalar<vtype> &r) {
         iScalar<vtype> ret;
@@ -38,14 +108,16 @@ public:
         *this = (*this)+r;
         return *this;
     }
-    
-
 };
     
 template<class vtype,int N> class iVector
 {
 public:
   SIMDalign vtype _internal[N];
+
+  typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
+  typedef typename GridTypeMapper<vtype>::vector_type vector_type;
+
     iVector(Zero &z){ *this = zero; };
     iVector() {};
     iVector<vtype,N> & operator= (Zero &hero){
@@ -57,6 +129,21 @@ public:
             zeroit(that._internal[i]);
         }
     }
+    friend void permute(iVector<vtype,N> &out,iVector<vtype,N> &in,int permutetype){
+      for(int i=0;i<N;i++){
+	permute(out._internal[i],in._internal[i],permutetype);
+      }
+    }
+    friend void extract(iVector<vtype,N> &in,std::vector<scalar_type *> &out){
+      for(int i=0;i<N;i++){
+	extract(in._internal[i],out);// extract advances pointers in out
+      }
+    }
+    friend void merge(iVector<vtype,N> &in,std::vector<scalar_type *> &out){
+      for(int i=0;i<N;i++){
+	merge(in._internal[i],out);// extract advances pointers in out
+      }
+    }
     // Unary negation
     friend inline iVector<vtype,N> operator -(const iVector<vtype,N> &r) {
         iVector<vtype,N> ret;
@@ -84,6 +171,10 @@ template<class vtype,int N> class iMatrix
 {
 public:
   SIMDalign    vtype _internal[N][N];
+
+  typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
+  typedef typename GridTypeMapper<vtype>::vector_type vector_type;
+
     iMatrix(Zero &z){ *this = zero; };
     iMatrix() {};
     iMatrix<vtype,N> & operator= (Zero &hero){
@@ -96,6 +187,24 @@ public:
                 zeroit(that._internal[i][j]);
         }}
     }
+    friend void permute(iMatrix<vtype,N> &out,iMatrix<vtype,N> &in,int permutetype){
+      for(int i=0;i<N;i++){
+      for(int j=0;j<N;j++){
+	permute(out._internal[i][j],in._internal[i][j],permutetype);
+      }}
+    }
+    friend void extract(iMatrix<vtype,N> &in,std::vector<scalar_type *> &out){
+      for(int i=0;i<N;i++){
+      for(int j=0;j<N;j++){
+	extract(in._internal[i][j],out);// extract advances pointers in out
+      }}
+    }
+    friend void merge(iMatrix<vtype,N> &in,std::vector<scalar_type *> &out){
+      for(int i=0;i<N;i++){
+      for(int j=0;j<N;j++){
+	merge(in._internal[i][j],out);// extract advances pointers in out
+      }}
+    }
     // Unary negation
     friend inline iMatrix<vtype,N> operator -(const iMatrix<vtype,N> &r) {
         iMatrix<vtype,N> ret;
diff --git a/Grid_mpi.cc b/Grid_mpi.cc
new file mode 100644
index 00000000..08d49312
--- /dev/null
+++ b/Grid_mpi.cc
@@ -0,0 +1,85 @@
+#include "Grid.h"
+#include <mpi.h>
+
+namespace dpo {
+
+  // Should error check all MPI calls.
+
+CartesianCommunicator::CartesianCommunicator(std::vector<int> &processors)
+{
+  _ndimension = _processors.size();
+  std::vector<int> periodic(_ndimension,1);
+
+  _processors = processors;
+  _processor_coords.resize(_ndimension);
+
+  MPI_Cart_create(MPI_COMM_WORLD _ndimension,&_processors[0],&periodic[0],1,&communicator);
+  MPI_Comm_rank(communicator,&_processor);
+  MPI_Cart_coords(communicator,_processor,_ndimension,&_processor_coords[0]);
+
+}
+
+void CartesianCommunicator::GlobalSumF(float &f){
+  MPI_Allreduce(&f,&f,1,MPI_FLOAT,MPI_SUM,communicator);
+}
+void CartesianCommunicator::GlobalSumFVector(float *f,int N)
+{
+  MPI_Allreduce(f,f,N,MPI_FLOAT,MPI_SUM,communicator);
+}
+void CartesianCommunicator::GlobalSumF(double &d)
+{
+  MPI_Allreduce(&d,&d,1,MPI_DOUBLE,MPI_SUM,communicator);
+}
+void CartesianCommunicator::GlobalSumFVector(double *d,int N)
+{
+  MPI_Allreduce(d,d,N,MPI_DOUBLE,MPI_SUM,communicator);
+}
+
+int CartesianCommunicator::ShiftedRank(int dim,int shift)
+{
+  int rank;
+  MPI_Cart_shift(communicator,dim,shift,&_processor,&rank);
+  return rank;
+}
+int CartesianCommunicator::Rank(std::vector<int> coor)
+{
+  int rank;
+  MPI_Cart_rank  (communicator, &coor[0], &rank);
+  return rank;
+}
+
+// Basic Halo comms primitive
+void CartesianCommunicator::SendToRecvFrom(void *xmit,
+					   std::vector<int> &to_coordinate,
+					   void *recv,
+					   std::vector<int> &from_coordinate,
+					   int bytes)
+{
+  int dest = Rank(to_coordinate);
+  int from = Rank(from_coordinate);
+
+  MPI_Request x_req;
+  MPI_Request r_req;
+  MPI_Status OkeyDokey;
+
+  MPI_Isend(xmit, bytes, MPI_CHAR,dest,dest,communicator,&x_req);
+  MPI_Irecv(recv, bytes, MPI_CHAR,from,from,communicator,&r_req);
+  MPI_Wait(&x_req,&OkeyDokey);
+  MPI_Wait(&r_req,&OkeyDokey);
+  MPI_Barrier();
+}
+
+}
+
+#if 0
+
+// Could possibly do a direct block strided send?
+  int MPI_Type_vector(
+		      int count,
+		      int blocklength,
+		      int stride,
+		      MPI_Datatype old_type,
+  MPI_Datatype *newtype_p
+		      );
+
+#endif
diff --git a/Grid_mpi_cshift.h b/Grid_mpi_cshift.h
new file mode 100644
index 00000000..9031170b
--- /dev/null
+++ b/Grid_mpi_cshift.h
@@ -0,0 +1,384 @@
+#ifndef _GRID_MPI_CSHIFT_H_
+#define _GRID_MPI_CSHIFT_H_
+
+      
+//////////////////////////////////////////////
+// Q. Split this into seperate sub functions?
+//////////////////////////////////////////////
+
+// CshiftCB_comms_splice
+// CshiftCB_comms
+// CshiftCB_local
+// CshiftCB_local_permute
+
+// Cshift_comms_splice
+// Cshift_comms
+// Cshift_local
+// Cshift_local_permute
+
+// Broadly I remain annoyed that the iteration is so painful
+// for red black data layout, when simple block strided descriptors suffice for non-cb. 
+//
+// The other option is to do it table driven, or perhaps store the CB of each site in a table.
+//
+// Must not lose sight that goal is to be able to construct really efficient
+// gather to a point stencil code. CSHIFT is not the best way, so probably need
+// additional stencil support.
+//
+// Could still do a templated syntax tree and make CSHIFT return lattice vector.
+//
+// Stencil based code could pre-exchange haloes and use a table lookup for neighbours
+//
+// Lattice <foo> could also allocate haloes which get used for stencil code.
+//
+// Grid could create a neighbour index table for a given stencil.
+// Could also implement CovariantCshift.
+
+//////////////////////////////////////////////////////
+//Non checkerboarded support functions
+//////////////////////////////////////////////////////
+
+friend void Gather_plane        (Lattice<vobj> &rhs,std::vector<vobj> &buffer,             int dimension,int plane)
+{
+  const int Nsimd = vector_type::Nsimd();
+  int rd = rhs._grid->_rdimensions[dimension];
+
+  int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+  int o   = 0;                                    // relative offset to base within plane
+#pragma omp parallel for collapse(2)
+  for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+    for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+	  
+	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
+	  int sx = (x+sshift)%rd;
+	  int so = sx*rhs._grid->_ostride[dimension];
+	  int permute_slice=0;
+	  int wrap = sshift/rd;
+	  int  num = sshift%rd;
+	  
+	  if ( x< rd-num ) permute_slice=wrap;
+	  else permute_slice = 1-wrap;
+	  
+	  if ( permute_slice ) {
+	    permute(ret._odata[ro+o+b],rhs._odata[so+o+b],permute_type);
+	  } else {
+	    ret._odata[ro+o+b]=rhs._odata[so+o+b];
+	  }
+	}
+	o +=rhs._grid->_slice_stride[dimension];
+      }
+  }
+  
+}
+//friend void Gather_plane_extract(Lattice<vobj> &rhs,std::vector<scalar_type *> pointers,int dimension,int plane);
+//
+//friend void Scatter_plane       (Lattice<vobj> &rhs,std::vector<vobj> face,             int dimension,int plane);
+//friend void Scatter_plane_merge (Lattice<vobj> &rhs,std::vector<scalar_type *> pointers,int dimension,int plane);
+//
+//template<int permute_type> friend void Copy_plane_permute(Lattice<vobj> &rhs,std::vector<vobj> face, int dimension,int plane);
+//                           friend void Copy_plane(Lattice<vobj> &rhs,std::vector<vobj> face, int dimension,int plane);
+//
+
+//////////////////////////////////////////////////////
+//Checkerboarded support functions
+//////////////////////////////////////////////////////
+
+//friend void GatherCB_plane        (Lattice<vobj> &rhs,std::vector<vobj> face,             int dimension,int plane);
+//friend void GatherCB_plane_extract(Lattice<vobj> &rhs,std::vector<scalar_type *> pointers,int dimension,int plane);
+//
+//friend void ScatterCB_plane       (Lattice<vobj> &rhs,std::vector<vobj> face,             int dimension,int plane);
+//friend void ScatterCB_plane_merge (Lattice<vobj> &rhs,std::vector<scalar_type *> pointers,int dimension,int plane);
+//
+//template<int permute_type> friend void CopyCB_plane_permute(Lattice<vobj> &rhs,std::vector<vobj> face, int dimension,int plane);
+//                           friend void Copy_plane(Lattice<vobj> &rhs,std::vector<vobj> face, int dimension,int plane);
+
+
+friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
+{
+  typedef typename vobj::vector_type vector_type;
+  typedef typename vobj::scalar_type scalar_type;
+  const int Nsimd = vector_type::Nsimd();
+
+  Lattice<vobj> ret(rhs._grid);
+  
+  int fd = rhs._grid->_fdimensions[dimension];
+  int rd = rhs._grid->_rdimensions[dimension];
+  //int ld = rhs._grid->_ldimensions[dimension];
+  //int gd = rhs._grid->_gdimensions[dimension];
+  
+
+  // Map to always positive shift modulo global full dimension.
+  shift = (shift+fd)%fd;
+
+  ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
+        
+  // the permute type
+  int simd_layout     = rhs._grid->_simd_layout[dimension];
+  int comm_dim        = rhs._grid->_processors[dimension] >1 ;
+  int permute_dim     = rhs._grid->_simd_layout[dimension]>1 && (!comm_dim);
+  int splice_dim      = rhs._grid->_simd_layout[dimension]>1 && (comm_dim);
+
+  int permute_type=0;
+  for(int d=0;d<dimension;d++){
+    if (rhs._grid->_simd_layout[d]>1 ) permute_type++;
+  }
+
+  // Logic for non-distributed dimension	  
+  std::vector<int> comm_offnode(simd_layout);
+  std::vector<int> comm_to     (simd_layout);
+  std::vector<int> comm_from   (simd_layout);
+  std::vector<int> comm_rx     (simd_layout);  // reduced coordinate of neighbour plane
+  std::vector<int> comm_simd_lane(simd_layout);// simd lane of neigbour plane
+
+  ///////////////////////////////////////////////
+  // Move via a fake comms buffer
+  // Simd direction uses an extract/merge pair
+  ///////////////////////////////////////////////
+  int buffer_size = rhs._grid->_slice_nblock[dimension]*rhs._grid->_slice_block[dimension];
+  int words = sizeof(vobj)/sizeof(vector_type);
+
+  std::vector<vobj,alignedAllocator<vobj> > comm_buf(buffer_size);
+  std::vector<std::vector<scalar_type> > comm_buf_extract(Nsimd,std::vector<scalar_type>(buffer_size*words) );
+  std::vector<scalar_type *> pointers(Nsimd);
+
+
+
+  if ( permute_dim ) {
+
+    for(int x=0;x<rd;x++){       
+      int ro  = x*rhs._grid->_ostride[dimension]; // base offset for result
+      int o   = 0;                                // relative offset to base
+      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+	  
+	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
+	  int sx = (x+sshift)%rd;
+	  int so = sx*rhs._grid->_ostride[dimension];
+	  int permute_slice=0;
+	  int wrap = sshift/rd;
+	  int  num = sshift%rd;
+	  
+	  if ( x< rd-num ) permute_slice=wrap;
+	  else permute_slice = 1-wrap;
+	  
+	  if ( permute_slice ) {
+	    permute(ret._odata[ro+o+b],rhs._odata[so+o+b],permute_type);
+	  } else {
+	    ret._odata[ro+o+b]=rhs._odata[so+o+b];
+	  }
+	}
+	o +=rhs._grid->_slice_stride[dimension];
+      }
+    }
+
+  } else if ( splice_dim ) {
+
+    if ( rhs._grid->_simd_layout[dimension] > 2 ) exit(-1); // use Cassert. Audit code for exit and replace
+    if ( rhs._grid->_simd_layout[dimension] < 1 ) exit(-1);
+	    
+
+    for(int i=0;i<vobj::vector_type::Nsimd();i++){
+      pointers[i] = (scalar_type *)&comm_buf_extract[i][0];
+    }
+
+    
+    for(int x=0;x<rd;x++){       
+
+
+      ///////////////////////////////////////////
+      // Extract one orthogonal slice at a time
+      ///////////////////////////////////////////
+      int ro  = x*rhs._grid->_ostride[dimension]; // base offset for result
+
+      o   = 0;                                    // relative offset to base      
+      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+	  
+	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
+	  int sx = (x+sshift)%rd;
+	  
+	  // base offset for source
+	  int so = sx*rhs._grid->_ostride[dimension];
+
+	  int permute_slice=0;
+	  int wrap = sshift/rd;
+	  int  num = sshift%rd;
+	    
+	  if ( x< rd-num ) permute_slice=wrap;
+	  else permute_slice = 1-wrap;
+
+	  if ( permute_slice ) {
+	    extract(rhs._odata[so+o+b],pointers);
+	  }
+	}
+	o +=rhs._grid->_slice_stride[dimension];
+      }
+
+    
+      ///////////////////////////////////////////
+      // Work out what to send where
+      ///////////////////////////////////////////
+
+      for(int s=0;s<simd_layout;s++) {
+
+
+	// shift to "neighbour" takes us off node
+	// coordinates (rx, simd_lane) of neighbour
+	// how many nodes away is this shift
+	// where we should send to
+	// where we should receive from
+	int shifted_x = x+s*rd+shift;
+	comm_offnode[s]           = shifted_x > ld; 
+	comm_send_rx[s]           = shifted_x%rd;     // which slice geton the other node
+	comm_send_simd_lane [s]   = shifted_x/rd;     // which slice on the other node
+	comm_from[s]    = shifted_x/ld;     
+	comm_to  [s] = (2*_processors[dimension]-comm_from[s]) % _processors[dimension];
+	comm_from[s] = (comm_from[s]+_processors[dimension])   % _processors[dimension];
+	      
+      }
+
+      ////////////////////////////////////////////////
+      // Insert communication phase
+      ////////////////////////////////////////////////
+#if 0
+	  } else if (comm_dim ) {
+	    
+	    // Packed gather sequence is clean
+	    int buffer_size = rhs._grid->_slice_nblock[dimension]*rhs._grid->_slice_nblock[dimension];
+	    std::vector<vobj,alignedAllocator<vobj> > send_buf(buffer_size);
+	    std::vector<vobj,alignedAllocator<vobj> > recv_buf(buffer_size);
+
+	    // off node; communcate slice (ld==rd)
+	    if ( x+shift > rd ) { 
+	      int sb=0;
+	      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+		for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
+		  send_buf[sb++]=rhs._odata[so+i];
+		}
+		so+=rhs._grid->_slice_stride[dimension];
+	      }	      
+
+	      // Make a comm_fake them mimics comms in periodic case.
+	      
+	      // scatter face
+	      int rb=0;
+	      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+		for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
+		  ret._odata[so+i]=recv_buf[rb++];
+		}
+		so+=rhs._grid->_slice_stride[dimension];
+	      }	      
+
+	    } else { 
+
+	      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+		for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
+		  ret._odata[o+i]=rhs._odata[so+i];
+		}
+		o+=rhs._grid->_slice_stride[dimension];
+		so+=rhs._grid->_slice_stride[dimension];
+	      }
+	    }
+
+#endif	
+    
+      ////////////////////////////////////////////////
+      // Pull receive buffers and permuted buffers in
+      ////////////////////////////////////////////////
+      for(int i=0;i<vobj::vector_type::Nsimd();i++){
+	pointers[i] = (scalar_type *)&comm_buf_extract[permute_map[permute_type][i]][0];
+      }
+
+      o   = 0;                                    // relative offset to base
+      int ro  = x*rhs._grid->_ostride[dimension]; // base offset for result
+      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+	  
+	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
+	  int sx = (x+sshift)%rd;
+	  
+	  // base offset for source
+	  int so = sx*rhs._grid->_ostride[dimension];
+	  
+	  int permute_slice=0;
+	  int wrap = sshift/rd;
+	  int  num = sshift%rd;
+	  
+	  if ( x< rd-num ) permute_slice=wrap;
+	  else permute_slice = 1-wrap;
+	  
+	  if ( permute_slice ) {
+	    merge(ret._odata[ro+o+b],pointers);
+	  }
+	}
+	o +=rhs._grid->_slice_stride[dimension];
+      }
+    }
+  
+
+  } else if ( comm_dim ) { 
+	
+
+    int co; // comm offset
+    int o;
+    
+    co=0;
+    for(int x=0;x<rd;x++){       
+      o=0;
+      int ro  = x*rhs._grid->_ostride[dimension]; // base offset for result
+      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+
+	  // This call in inner loop is annoying but necessary for dimension=0
+	  // in the case of RedBlack grids. Could optimise away with 
+	  // alternate code paths for all other cases.
+	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
+	  int sx = (x+sshift)%rd;
+	  int so = sx*rhs._grid->_ostride[dimension];
+
+	  comm_buf[co++]=rhs._odata[so+o+b];
+
+	}
+	o +=rhs._grid->_slice_stride[dimension];
+      }
+
+      // Step through a copy into a comms buffer and pull back in.
+      // Genuine fake implementation could calculate if loops back
+      co=0; o=0;
+      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+	  ret._odata[ro+o+b]=comm_buf[co++];
+	}
+	o +=rhs._grid->_slice_stride[dimension];
+      }
+    }
+
+  } else { // Local dimension, no permute required
+
+    for(int x=0;x<rd;x++){       
+      int o=0;
+      int ro  = x*rhs._grid->_ostride[dimension]; // base offset for result
+      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+
+	  // This call in inner loop is annoying but necessary for dimension=0
+	  // in the case of RedBlack grids. Could optimise away with 
+	  // alternate code paths for all other cases.
+	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
+	  
+	  int sx = (x+sshift)%rd;
+	  int so = sx*rhs._grid->_ostride[dimension];
+	  ret._odata[bo+o+b]=rhs._odata[so+o+b];
+
+	}
+	o +=rhs._grid->_slice_stride[dimension];
+      }
+    }
+
+  }
+
+  return ret;
+}
+
+
+#endif
diff --git a/Grid_none_cshift.h b/Grid_none_cshift.h
new file mode 100644
index 00000000..87ae6a68
--- /dev/null
+++ b/Grid_none_cshift.h
@@ -0,0 +1,313 @@
+#ifndef _GRID_NONE_CSHIFT_H_
+#define _GRID_NONE_CSHIFT_H_
+  // Work out whether to permute 
+  // ABCDEFGH ->   AE BF CG DH       permute              wrap num
+  //
+  // Shift 0       AE BF CG DH       0 0 0 0    ABCDEFGH   0   0
+  // Shift 1       BF CG DH AE       0 0 0 1    BCDEFGHA   0   1
+  // Shift 2       CG DH AE BF       0 0 1 1    CDEFGHAB   0   2
+  // Shift 3       DH AE BF CG       0 1 1 1    DEFGHABC   0   3
+  // Shift 4       AE BF CG DH       1 1 1 1    EFGHABCD   1   0 
+  // Shift 5       BF CG DH AE       1 1 1 0    FGHACBDE   1   1 
+  // Shift 6       CG DH AE BF       1 1 0 0    GHABCDEF   1   2
+  // Shift 7       DH AE BF CG       1 0 0 0    HABCDEFG   1   3
+
+  // Suppose 4way simd in one dim.
+  // ABCDEFGH ->   AECG BFDH      permute              wrap num
+
+  // Shift 0       AECG BFDH      0,00 0,00 ABCDEFGH         0     0
+  // Shift 1       BFDH CGEA      0,00 1,01 BCDEFGHA         0     1
+  // Shift 2       CGEA DHFB      1,01 1,01 CDEFGHAB         1     0
+  // Shift 3       DHFB EAGC      1,01 1,11 DEFGHABC         1     1
+  // Shift 4       EAGC FBHD      1,11 1,11 EFGHABCD         2     0 
+  // Shift 5       FBHD GCAE      1,11 1,10 FGHABCDE         2     1
+  // Shift 6       GCAE HDBF      1,10 1,10 GHABCDEF         3     0
+  // Shift 7       HDBF AECG      1,10 0,00 HABCDEFG         3     1
+
+  // Generalisation to 8 way simd, 16 way simd required.
+  //
+  // Need log2 Nway masks. consisting of 
+  //	    1 bit  256 bit granule
+  //	    2 bit  128 bit granule
+  //        4 bits 64  bit granule
+  //        8 bits 32  bit granules
+  //
+  //        15 bits....
+  
+// For optimisation:
+//
+// split into Cshift_none_rb_permute
+// split into Cshift_none_rb_simple
+//
+// split into Cshift_none_permute
+// split into Cshift_none_simple
+
+friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
+{
+  Lattice<vobj> ret(rhs._grid);
+  
+  int fd = rhs._grid->_fdimensions[dimension];
+  int rd = rhs._grid->_rdimensions[dimension];
+  int ld = rhs._grid->_ldimensions[dimension];
+  int gd = rhs._grid->_gdimensions[dimension];
+  
+
+  // Map to always positive shift modulo global full dimension.
+  shift = (shift+fd)%fd;
+
+  ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
+        
+  // the permute type
+
+  int permute_dim =rhs._grid->_simd_layout[dimension]>1 ;
+  int permute_type=0;
+  for(int d=0;d<dimension;d++){
+    if (rhs._grid->_simd_layout[d]>1 ) permute_type++;
+  }
+
+  for(int x=0;x<rd;x++){       
+
+    int bo  = x*rhs._grid->_ostride[dimension];
+    int o   = 0;
+
+    if ( permute_dim ) {
+
+      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+	  
+	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
+	  
+	  int sx = (x+sshift)%rd;
+	  int so = sx*rhs._grid->_ostride[dimension];
+
+	  int permute_slice=0;
+	  int wrap = sshift/rd;
+	  int  num = sshift%rd;
+	    
+	  if ( x< rd-num ) permute_slice=wrap;
+	  else permute_slice = 1-wrap;
+	  
+	  if ( permute_slice ) {
+	    permute(ret._odata[bo+o+b],rhs._odata[so+o+b],permute_type);
+	  } else {
+	    ret._odata[bo+o+b]=rhs._odata[so+o+b];
+	  }
+
+	}
+	o +=rhs._grid->_slice_stride[dimension];
+
+      }
+    } else {
+      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+
+	  // This call in inner loop is annoying but necessary for dimension=0
+	  // in the case of RedBlack grids. Could optimise away with 
+	  // alternate code paths for all other cases.
+	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
+	  
+	  int sx = (x+sshift)%rd;
+	  int so = sx*rhs._grid->_ostride[dimension];
+	  ret._odata[bo+o+b]=rhs._odata[so+o+b];
+
+	}
+	o +=rhs._grid->_slice_stride[dimension];
+      }
+    }
+  }
+  return ret;
+}
+        
+#if 0 
+// Collapse doesn't appear to work the way I think it should in icpc
+friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
+{
+  Lattice<vobj> ret(rhs._grid);
+        
+  ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
+  shift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift);
+  int sx,so,o;
+  int rd = rhs._grid->_rdimensions[dimension];
+  int ld = rhs._grid->_dimensions[dimension];
+  // Map to always positive shift.
+  shift = (shift+ld)%ld;
+  // Work out whether to permute and the permute type
+  // ABCDEFGH ->   AE BF CG DH       permute
+  // Shift 0       AE BF CG DH       0 0 0 0    ABCDEFGH
+  // Shift 1       DH AE BF CG       1 0 0 0    HABCDEFG
+  // Shift 2       CG DH AE BF       1 1 0 0    GHABCDEF
+  // Shift 3       BF CG DH AE       1 1 1 0    FGHACBDE
+  // Shift 4       AE BF CG DH       1 1 1 1    EFGHABCD
+  // Shift 5       DH AE BF CG       0 1 1 1    DEFGHABC
+  // Shift 6       CG DH AE BF       0 0 1 1    CDEFGHAB
+  // Shift 7       BF CG DH AE       0 0 0 1    BCDEFGHA
+  int permute_dim =rhs._grid->_layout[dimension]>1 ;
+  int permute_type=0;
+  for(int d=0;d<dimension;d++)
+    if (rhs._grid->_layout[d]>1 ) permute_type++;
+        
+  // loop over perp slices.
+  // Threading considerations:
+  //   Need to map thread_num to
+  //
+  //               x_min,x_max for Loop-A
+  //               n_min,n_max for Loop-B
+  //               b_min,b_max for Loop-C
+  //  In a way that maximally load balances.
+  //
+  //  Optimal:
+  //      There are rd*n_block*block items of work.
+  //      These serialise as item "w"
+  //      b=w%block; w=w/block
+  //      n=w%nblock; x=w/nblock. Perhaps 20 cycles?
+  //
+  //  Logic:
+  //      x_chunk = (rd+thread)/nthreads simply divide work across nodes.
+  //
+  //      rd=5 , threads = 8;
+  //      0 1 2 3 4 5 6 7
+  //      0 0 0 1 1 1 1 1
+  for(int x=0;x<rd;x++){         // Loop A
+    sx = (x-shift+ld)%rd;
+    o  = x*rhs._grid->_ostride[dimension];
+    so =sx*rhs._grid->_ostride[dimension];
+    int permute_slice=0;
+    if ( permute_dim ) {
+      permute_slice = shift/rd;
+      if ( x<shift%rd ) permute_slice = 1-permute_slice;
+    }
+#if 0
+    if ( permute_slice ) {
+                
+      int internal=sizeof(vobj)/sizeof(vComplex);
+      int num =rhs._grid->_slice_block[dimension]*internal;
+                
+      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	vComplex *optr = (vComplex *)&ret._odata[o];
+	vComplex *iptr = (vComplex *)&rhs._odata[so];
+	for(int b=0;b<num;b++){
+	  permute(optr[b],iptr[b],permute_type);
+	}
+	o+=rhs._grid->_slice_stride[dimension];
+	so+=rhs._grid->_slice_stride[dimension];
+      }
+    } else {
+      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
+	  ret._odata[o+i]=rhs._odata[so+i];
+	}
+	o+=rhs._grid->_slice_stride[dimension];
+	so+=rhs._grid->_slice_stride[dimension];
+      }
+            
+    }
+#else
+    if ( permute_slice ) {
+      int internal=sizeof(vobj)/sizeof(vComplex);
+      int num =rhs._grid->_slice_block[dimension]*internal;
+#pragma omp parallel for collapse(2)
+      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	for(int b=0;b<num;b++){
+	  vComplex *optr = (vComplex *)&ret._odata[o +n*rhs._grid->_slice_stride[dimension]];
+	  vComplex *iptr = (vComplex *)&rhs._odata[so+n*rhs._grid->_slice_stride[dimension]];
+	  permute(optr[b],iptr[b],permute_type);
+	}
+      }
+    } else {
+#pragma omp parallel for collapse(2)
+      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+	for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
+	  int oo = o+ n*rhs._grid->_slice_stride[dimension];
+	  int soo=so+ n*rhs._grid->_slice_stride[dimension];
+	  ret._odata[oo+i]=rhs._odata[soo+i];
+	}
+      }
+            
+    }
+#endif
+  }
+  return ret;
+}
+friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
+{
+  Lattice<vobj> ret(rhs._grid);
+        
+  ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
+  shift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift);
+  int rd = rhs._grid->_rdimensions[dimension];
+  int ld = rhs._grid->_dimensions[dimension];
+        
+  // Map to always positive shift.
+  shift = (shift+ld)%ld;
+        
+  // Work out whether to permute and the permute type
+  // ABCDEFGH ->   AE BF CG DH       permute
+  // Shift 0       AE BF CG DH       0 0 0 0    ABCDEFGH
+  // Shift 1       DH AE BF CG       1 0 0 0    HABCDEFG
+  // Shift 2       CG DH AE BF       1 1 0 0    GHABCDEF
+  // Shift 3       BF CG DH AE       1 1 1 0    FGHACBDE
+  // Shift 4       AE BF CG DH       1 1 1 1    EFGHABCD
+  // Shift 5       DH AE BF CG       0 1 1 1    DEFGHABC
+  // Shift 6       CG DH AE BF       0 0 1 1    CDEFGHAB
+  // Shift 7       BF CG DH AE       0 0 0 1    BCDEFGHA
+  int permute_dim =rhs._grid->_layout[dimension]>1 ;
+  int permute_type=0;
+  for(int d=0;d<dimension;d++)
+    if (rhs._grid->_layout[d]>1 ) permute_type++;
+        
+        
+  // loop over all work
+  int work =rd*rhs._grid->_slice_nblock[dimension]*rhs._grid->_slice_block[dimension];
+
+#pragma omp parallel for
+  for(int ww=0;ww<work;ww++){
+
+            
+    // can optimise this if know w moves congtiguously for a given thread.
+    // b=(b+1);
+    // if (b==_slice_block) {b=0; n=n+1;}
+    // if (n==_slice_nblock) { n=0; x=x+1}
+    //
+    // Perhaps a five cycle iterator, or so.
+    int w=ww;
+    int b = w%rhs._grid->_slice_block[dimension] ; w=w/rhs._grid->_slice_block[dimension];
+    int n = w%rhs._grid->_slice_nblock[dimension]; w=w/rhs._grid->_slice_nblock[dimension];
+    int x = w;
+
+    int sx,so,o;
+    sx = (x-shift+ld)%rd;
+    o  = x*rhs._grid->_ostride[dimension]+n*rhs._grid->_slice_stride[dimension]; // common sub expression alert.
+    so =sx*rhs._grid->_ostride[dimension]+n*rhs._grid->_slice_stride[dimension];
+            
+    int permute_slice=0;
+    if ( permute_dim ) {
+      permute_slice = shift/rd;
+      if ( x<shift%rd ) permute_slice = 1-permute_slice;
+    }
+            
+    if ( permute_slice ) {
+                
+      int internal=sizeof(vobj)/sizeof(vComplex);
+      vComplex *optr = (vComplex *)&ret._odata[o+b];
+      vComplex *iptr = (vComplex *)&rhs._odata[so+b];
+      const char *pf = (const char *)iptr;
+      for(int i=0;i<sizeof(vobj);i+=64){
+	_mm_prefetch(pf+i,_MM_HINT_T0);
+      }
+
+      for(int i=0;i<internal;i++){
+	permute(optr[i],iptr[i],permute_type);
+      }
+    } else {
+      const char *pf = (const char *) &rhs._odata[so+b];
+      for(int i=0;i<sizeof(vobj);i+=64){
+	_mm_prefetch(pf+i,_MM_HINT_T0);
+      }
+      ret._odata[o+b]=rhs._odata[so+b];
+    }
+  }
+  return ret;
+}
+#endif
+#endif
+
diff --git a/Grid_simd.h b/Grid_simd.h
index 8c7f3a1c..bef21bcf 100644
--- a/Grid_simd.h
+++ b/Grid_simd.h
@@ -47,7 +47,7 @@ namespace dpo {
   typedef std::complex<RealF> ComplexF;
   typedef std::complex<RealD> ComplexD;
   typedef std::complex<Real>  Complex;
-  
+
 
   inline RealF adj(const RealF  & r){ return r; }
   inline RealF conj(const RealF  & r){ return r; }
@@ -130,7 +130,7 @@ namespace dpo {
 #if defined (AVX1) || defined (AVX2)
     typedef __m256 fvec;
     typedef __m256d dvec;
-    typedef __m256 cvec;
+    typedef __m256  cvec;
     typedef __m256d zvec;
 #endif
 #if defined (AVX512)
diff --git a/Grid_vComplexD.h b/Grid_vComplexD.h
index 71fa45ef..15dc00c6 100644
--- a/Grid_vComplexD.h
+++ b/Grid_vComplexD.h
@@ -8,6 +8,9 @@ namespace dpo {
     protected:
         zvec v;
     public:
+	typedef zvec     vector_type;
+	typedef ComplexD scalar_type;
+
         vComplexD & operator = ( Zero & z){
             vzero(*this);
             return (*this);
@@ -145,14 +148,43 @@ namespace dpo {
             ymm0 =  _mm512_swizzle_pd(b.v, _MM_SWIZ_REG_CDAB); // OK
             ret.v=  _mm512_fmadd_pd(ymm0,imag,ymm1);
              /* Imag OK */
-            
-
 #endif
 #ifdef QPX
             ret.v = vec_mul(a.v,b.v);
 #endif
             return ret;
         };
+
+        /////////////////////////////////////////////////////////////////
+        // Extract
+        /////////////////////////////////////////////////////////////////
+        friend inline void extract(vComplexD &y,std::vector<ComplexD *> &extracted){
+	  // Bounce off stack is painful
+	  // temporary hack while I figure out the right interface
+	  const int Nsimd = vComplexD::Nsimd();
+	  std::vector<ComplexD> buf(Nsimd); 
+
+	  vstore(y,&buf[0]);
+
+	  for(int i=0;i<Nsimd;i++){
+	    *extracted[i] = buf[i];
+	    extracted[i]++;
+	  }
+        };
+
+        friend inline void merge(vComplexD &y,std::vector<ComplexD *> &extracted){
+	  // Bounce off stack is painful
+	  // temporary hack while I figure out the right interface
+	  const int Nsimd = vComplexD::Nsimd();
+	  std::vector<ComplexD> buf(Nsimd); 
+
+	  for(int i=0;i<Nsimd;i++){
+	    buf[i]=*extracted[i];
+	    extracted[i]++;
+	  }
+	  vset(y,&buf[0]); 
+        };
+        
         
         /////////////////////////////////////////////////////////////////
         // Permute
@@ -165,7 +197,7 @@ namespace dpo {
                 // AB => BA i.e. ab cd =>cd ab
 #endif
 #ifdef SSE2
-                    //No cases here
+		  break;
 #endif
 #ifdef AVX512
                     // 4 complex=>2 permute
@@ -205,7 +237,7 @@ namespace dpo {
 #endif
         }
 
-        friend inline void vset(vComplexD &ret, std::complex<double> *a){
+        friend inline void vset(vComplexD &ret,ComplexD *a){
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_set_pd(a[1].imag(),a[1].real(),a[0].imag(),a[0].real());
 #endif
@@ -221,7 +253,7 @@ namespace dpo {
 #endif
         }
 
-friend inline void vstore(vComplexD &ret, std::complex<double> *a){
+friend inline void vstore(vComplexD &ret, ComplexD *a){
 #if defined (AVX1)|| defined (AVX2)
        _mm256_store_pd((double *)a,ret.v);
 #endif
@@ -273,10 +305,10 @@ friend inline void vstore(vComplexD &ret, std::complex<double> *a){
 	 //            return std::complex<double>(_mm256_mask_reduce_add_pd(0x55, in.v),_mm256_mask_reduce_add_pd(0xAA, in.v));
 	 __attribute__ ((aligned(32))) double c_[4];
          _mm256_store_pd(c_,in.v);
-	 return std::complex<double>(c_[0]+c_[2],c_[1]+c_[3]);
+	 return ComplexD(c_[0]+c_[2],c_[1]+c_[3]);
 #endif 
 #ifdef AVX512
-            return std::complex<double>(_mm512_mask_reduce_add_pd(0x5555, in.v),_mm512_mask_reduce_add_pd(0xAAAA, in.v));
+            return ComplexD(_mm512_mask_reduce_add_pd(0x5555, in.v),_mm512_mask_reduce_add_pd(0xAAAA, in.v));
 #endif 
 #ifdef QPX
 #endif
diff --git a/Grid_vComplexF.h b/Grid_vComplexF.h
index c5c6e58f..0b4f5e61 100644
--- a/Grid_vComplexF.h
+++ b/Grid_vComplexF.h
@@ -8,6 +8,11 @@ namespace dpo {
         cvec v;
         
     public:
+        static inline int Nsimd(void) { return sizeof(cvec)/sizeof(float)/2;}
+    public:
+	typedef cvec     vector_type;
+	typedef ComplexF scalar_type;
+
         vComplexF & operator = ( Zero & z){
             vzero(*this);
             return (*this);
@@ -125,6 +130,39 @@ namespace dpo {
             return ret;
         };
 
+        /////////////////////////////////////////////////////////////////
+        // Extract
+        /////////////////////////////////////////////////////////////////
+        friend inline void extract(vComplexF &y,std::vector<ComplexF *> &extracted){
+	  // Bounce off heap is painful
+	  // temporary hack while I figure out the right interface
+            vComplexF vbuf;
+            ComplexF *buf = (ComplexF *)&vbuf;
+            
+	  vstore(y,&buf[0]);
+	  for(int i=0;i<vComplexF::Nsimd();i++){
+	    *extracted[i] = buf[i];
+	    extracted[i]++;
+	  }
+
+        };
+
+        friend inline void merge(vComplexF &y,std::vector<ComplexF *> &extracted){
+	  // Bounce off stack is painful
+	  // temporary hack while I figure out the right interface
+	  const int Nsimd = vComplexF::Nsimd();
+            vComplexF vbuf;
+            ComplexF *buf = (ComplexF *)&vbuf;
+
+	  for(int i=0;i<Nsimd;i++){
+	    buf[i]=*extracted[i];
+	    extracted[i]++;
+	  }
+	  vset(y,&buf[0]); 
+        };
+        
+
+
         /////////////////////////////////////////////////////////////////
         // Permute
         /////////////////////////////////////////////////////////////////
@@ -185,7 +223,7 @@ namespace dpo {
             vsplat(ret,a,b);
         }
 
-friend inline void vstore(vComplexF &ret, std::complex<float> *a){
+friend inline void vstore(vComplexF &ret, ComplexF *a){
 #if defined (AVX1)|| defined (AVX2)
         _mm256_store_ps((float *)a,ret.v);
 #endif
@@ -225,11 +263,11 @@ exit(-1);
 #if defined (AVX1) || defined(AVX2)
          __attribute__ ((aligned(32))) float c_[8];
          _mm256_store_ps(c_,in.v);
-         return std::complex<float>(c_[0]+c_[2]+c_[4]+c_[6],c_[1]+c_[3]+c_[5]+c_[7]);
+         return ComplexF(c_[0]+c_[2]+c_[4]+c_[6],c_[1]+c_[3]+c_[5]+c_[7]);
 
 #endif
 #ifdef AVX512
-            return std::complex<float>(_mm512_mask_reduce_add_ps(0x5555, in.v),_mm512_mask_reduce_add_ps(0xAAAA, in.v));
+            return ComplexF(_mm512_mask_reduce_add_ps(0x5555, in.v),_mm512_mask_reduce_add_ps(0xAAAA, in.v));
 #endif
 #ifdef QPX
 #endif
@@ -321,9 +359,6 @@ exit(-1);
             return *this;
         }
 
-        
-    public:
-        static int Nsimd(void) { return sizeof(cvec)/sizeof(float)/2;}
     };
 
     inline vComplexF localInnerProduct(const vComplexF & l, const vComplexF & r) { return conj(l)*r; }
diff --git a/Grid_vRealD.h b/Grid_vRealD.h
index d092d509..e5d3d4c6 100644
--- a/Grid_vRealD.h
+++ b/Grid_vRealD.h
@@ -7,7 +7,11 @@ namespace dpo{
     class vRealD  {
     protected:
         dvec v; // dvec is double precision vector
+
     public:
+	typedef dvec  vector_type;
+	typedef RealD scalar_type;
+
         vRealD(){};
 
         friend inline void mult(vRealD * __restrict__ y,const vRealD * __restrict__ l,const vRealD *__restrict__ r) {*y = (*l) * (*r);}
@@ -94,6 +98,37 @@ namespace dpo{
 #endif
             return ret;
         };
+
+        /////////////////////////////////////////////////////////////////
+        // Extract
+        /////////////////////////////////////////////////////////////////
+        friend inline void extract(vRealD &y,std::vector<RealD *> &extracted){
+	  // Bounce off stack is painful
+	  // temporary hack while I figure out the right interface
+	  const int Nsimd = vRealD::Nsimd();
+	  RealD buf[Nsimd]; 
+
+	  vstore(y,buf);
+
+	  for(int i=0;i<Nsimd;i++){
+	    *extracted[i] = buf[i];
+	    extracted[i]++;
+	  }
+        };
+
+        friend inline void merge(vRealD &y,std::vector<RealD *> &extracted){
+	  // Bounce off stack is painful
+	  // temporary hack while I figure out the right interface
+	  const int Nsimd = vRealD::Nsimd();
+	  RealD buf[Nsimd]; 
+
+	  for(int i=0;i<Nsimd;i++){
+	    buf[i]=*extracted[i];
+	    extracted[i]++;
+	  }
+	  vset(y,buf); 
+        };
+
         
         // Permute plans
         // Permute 0 every ABCDEFGH -> BA DC FE HG
diff --git a/Grid_vRealF.h b/Grid_vRealF.h
index 72f7d9e6..db74b005 100644
--- a/Grid_vRealF.h
+++ b/Grid_vRealF.h
@@ -7,7 +7,12 @@ namespace dpo {
     class vRealF  {
     protected:
         fvec v;
+
     public:
+
+	typedef fvec  vector_type;
+	typedef RealF scalar_type;
+
         vRealF(){};
         ////////////////////////////////////
         // Arithmetic operator overloads +,-,*
@@ -113,6 +118,37 @@ namespace dpo {
         //////////////////////////////////
         friend inline void vone (vRealF &ret){vsplat(ret,1.0);}
         friend inline void vzero(vRealF &ret){vsplat(ret,0.0);}
+
+
+        /////////////////////////////////////////////////////////////////
+        // Extract
+        /////////////////////////////////////////////////////////////////
+        friend inline void extract(vRealF &y,std::vector<RealF *> &extracted){
+	  // Bounce off stack is painful
+	  // temporary hack while I figure out the right interface
+	  const int Nsimd = vRealF::Nsimd();
+	  RealF buf[Nsimd]; 
+
+	  vstore(y,buf);
+
+	  for(int i=0;i<Nsimd;i++){
+	    *extracted[i] = buf[i];
+	    extracted[i]++;
+	  }
+        };
+
+        friend inline void merge(vRealF &y,std::vector<RealF *> &extracted){
+	  // Bounce off stack is painful
+	  // temporary hack while I figure out the right interface
+	  const int Nsimd = vRealF::Nsimd();
+	  RealF buf[Nsimd]; 
+
+	  for(int i=0;i<Nsimd;i++){
+	    buf[i]=*extracted[i];
+	    extracted[i]++;
+	  }
+	  vset(y,buf); 
+        };
         
         //////////////////////////////////////////////////////////
         // Permute
@@ -256,7 +292,7 @@ friend inline void vstore(vRealF &ret, float *a){
             return *this;
         }
     public:
-        static int Nsimd(void) { return sizeof(fvec)/sizeof(float);}
+        static inline int Nsimd(void) { return sizeof(fvec)/sizeof(float);}
     };
     inline vRealF localInnerProduct(const vRealF & l, const vRealF & r) { return conj(l)*r; }
     inline void  zeroit(vRealF &z){ vzero(z);}

From 9bea4e25eede59e5f294a1e95bacead05a8a7515 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sun, 29 Mar 2015 21:44:22 +0100
Subject: [PATCH 024/429] Make file and configure

---
 Makefile.am  |  5 ++++-
 configure.ac | 22 ++++++++++++++++++++++
 2 files changed, 26 insertions(+), 1 deletion(-)

diff --git a/Makefile.am b/Makefile.am
index 13c2484b..b9763db1 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -23,5 +23,8 @@ include_HEADERS = Grid.h\
 # Test code
 #
 bin_PROGRAMS = Grid_main
-Grid_main_SOURCES = Grid_main.cc
+Grid_main_SOURCES = \
+	Grid_main.cc\
+	Grid_fake.cc
+
 Grid_main_LDADD = libGrid.a
diff --git a/configure.ac b/configure.ac
index becb98dd..147ce21f 100644
--- a/configure.ac
+++ b/configure.ac
@@ -50,5 +50,27 @@ case ${ac_SIMD} in
      ;;
 esac
 
+
+AC_ARG_ENABLE([comms],[AC_HELP_STRING([--enable-comms=none|fake|mpi],[Select communications])],[ac_COMMS=${enable_comms}],[ac_COMMS=none])
+
+case ${ac_COMMS} in
+     fake)
+       echo Configuring for FAKE communications
+       AC_DEFINE([GRID_COMMS_FAKE],[1],[GRID_COMMS_FAKE] )
+     ;;
+     none)
+       echo Configuring for NO communications
+       AC_DEFINE([GRID_COMMS_NONE],[1],[GRID_COMMS_NONE] )
+     ;;
+     mpi)
+       echo Configuring for MPI communications
+       AC_DEFINE([GRID_COMMS_MPI],[1],[GRID_COMMS_MPI] )
+     ;;
+     *)
+     AC_MSG_ERROR([${ac_COMMS} unsupported --enable-comms option]); 
+     ;;
+esac
+
+
 AC_CONFIG_FILES(Makefile)
 AC_OUTPUT

From 4f57ccc66cc6820f12b4d94a5126ec39d3870591 Mon Sep 17 00:00:00 2001
From: azusayamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 29 Mar 2015 21:50:20 +0100
Subject: [PATCH 025/429] No compile fix

---
 Grid_fake.cc | 1 -
 1 file changed, 1 deletion(-)

diff --git a/Grid_fake.cc b/Grid_fake.cc
index 7e664f33..abc54c8e 100644
--- a/Grid_fake.cc
+++ b/Grid_fake.cc
@@ -27,7 +27,6 @@ void CartesianCommunicator::SendToRecvFrom(void *xmit,
 		    int bytes)
 {
   exit(-1);
-  bcopy(xmit,recv,bytes);
 }
 
 }

From e0af0e658d4bb7c7cd3d05663adfb95240bc7978 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sun, 29 Mar 2015 22:04:49 +0100
Subject: [PATCH 026/429] Commit

---
 Makefile.in |  371 ++--
 configure   | 4687 ++++++++++++++++++++++-----------------------------
 2 files changed, 2294 insertions(+), 2764 deletions(-)

diff --git a/Makefile.in b/Makefile.in
index b841357d..9a79034e 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -1,9 +1,8 @@
-# Makefile.in generated by automake 1.11.1 from Makefile.am.
+# Makefile.in generated by automake 1.15 from Makefile.am.
 # @configure_input@
 
-# Copyright (C) 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002,
-# 2003, 2004, 2005, 2006, 2007, 2008, 2009  Free Software Foundation,
-# Inc.
+# Copyright (C) 1994-2014 Free Software Foundation, Inc.
+
 # This Makefile.in is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
@@ -18,6 +17,61 @@
 
 
 VPATH = @srcdir@
+am__is_gnu_make = { \
+  if test -z '$(MAKELEVEL)'; then \
+    false; \
+  elif test -n '$(MAKE_HOST)'; then \
+    true; \
+  elif test -n '$(MAKE_VERSION)' && test -n '$(CURDIR)'; then \
+    true; \
+  else \
+    false; \
+  fi; \
+}
+am__make_running_with_option = \
+  case $${target_option-} in \
+      ?) ;; \
+      *) echo "am__make_running_with_option: internal error: invalid" \
+              "target option '$${target_option-}' specified" >&2; \
+         exit 1;; \
+  esac; \
+  has_opt=no; \
+  sane_makeflags=$$MAKEFLAGS; \
+  if $(am__is_gnu_make); then \
+    sane_makeflags=$$MFLAGS; \
+  else \
+    case $$MAKEFLAGS in \
+      *\\[\ \	]*) \
+        bs=\\; \
+        sane_makeflags=`printf '%s\n' "$$MAKEFLAGS" \
+          | sed "s/$$bs$$bs[$$bs $$bs	]*//g"`;; \
+    esac; \
+  fi; \
+  skip_next=no; \
+  strip_trailopt () \
+  { \
+    flg=`printf '%s\n' "$$flg" | sed "s/$$1.*$$//"`; \
+  }; \
+  for flg in $$sane_makeflags; do \
+    test $$skip_next = yes && { skip_next=no; continue; }; \
+    case $$flg in \
+      *=*|--*) continue;; \
+        -*I) strip_trailopt 'I'; skip_next=yes;; \
+      -*I?*) strip_trailopt 'I';; \
+        -*O) strip_trailopt 'O'; skip_next=yes;; \
+      -*O?*) strip_trailopt 'O';; \
+        -*l) strip_trailopt 'l'; skip_next=yes;; \
+      -*l?*) strip_trailopt 'l';; \
+      -[dEDm]) skip_next=yes;; \
+      -[JT]) skip_next=yes;; \
+    esac; \
+    case $$flg in \
+      *$$target_option*) has_opt=yes; break;; \
+    esac; \
+  done; \
+  test $$has_opt = yes
+am__make_dryrun = (target_option=n; $(am__make_running_with_option))
+am__make_keepgoing = (target_option=k; $(am__make_running_with_option))
 pkgdatadir = $(datadir)/@PACKAGE@
 pkgincludedir = $(includedir)/@PACKAGE@
 pkglibdir = $(libdir)/@PACKAGE@
@@ -36,14 +90,12 @@ PRE_UNINSTALL = :
 POST_UNINSTALL = :
 bin_PROGRAMS = Grid_main$(EXEEXT)
 subdir = .
-DIST_COMMON = README $(am__configure_deps) $(include_HEADERS) \
-	$(srcdir)/Grid_config.h.in $(srcdir)/Makefile.am \
-	$(srcdir)/Makefile.in $(top_srcdir)/configure AUTHORS COPYING \
-	ChangeLog INSTALL NEWS compile depcomp install-sh missing
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/configure.ac
 am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
 	$(ACLOCAL_M4)
+DIST_COMMON = $(srcdir)/Makefile.am $(top_srcdir)/configure \
+	$(am__configure_deps) $(include_HEADERS) $(am__DIST_COMMON)
 am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \
  configure.lineno config.status.lineno
 mkinstalldirs = $(install_sh) -d
@@ -71,46 +123,111 @@ am__nobase_list = $(am__nobase_strip_setup); \
 am__base_list = \
   sed '$$!N;$$!N;$$!N;$$!N;$$!N;$$!N;$$!N;s/\n/ /g' | \
   sed '$$!N;$$!N;$$!N;$$!N;s/\n/ /g'
+am__uninstall_files_from_dir = { \
+  test -z "$$files" \
+    || { test ! -d "$$dir" && test ! -f "$$dir" && test ! -r "$$dir"; } \
+    || { echo " ( cd '$$dir' && rm -f" $$files ")"; \
+         $(am__cd) "$$dir" && rm -f $$files; }; \
+  }
 am__installdirs = "$(DESTDIR)$(libdir)" "$(DESTDIR)$(bindir)" \
 	"$(DESTDIR)$(includedir)"
 LIBRARIES = $(lib_LIBRARIES)
 AR = ar
 ARFLAGS = cru
+AM_V_AR = $(am__v_AR_@AM_V@)
+am__v_AR_ = $(am__v_AR_@AM_DEFAULT_V@)
+am__v_AR_0 = @echo "  AR      " $@;
+am__v_AR_1 = 
 libGrid_a_AR = $(AR) $(ARFLAGS)
 libGrid_a_LIBADD =
-am_libGrid_a_OBJECTS = Grid_signal.$(OBJEXT)
+am_libGrid_a_OBJECTS = Grid_init.$(OBJEXT)
 libGrid_a_OBJECTS = $(am_libGrid_a_OBJECTS)
 PROGRAMS = $(bin_PROGRAMS)
-am_Grid_main_OBJECTS = Grid_main.$(OBJEXT)
+am_Grid_main_OBJECTS = Grid_main.$(OBJEXT) Grid_fake.$(OBJEXT)
 Grid_main_OBJECTS = $(am_Grid_main_OBJECTS)
 Grid_main_DEPENDENCIES = libGrid.a
+AM_V_P = $(am__v_P_@AM_V@)
+am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
+am__v_P_0 = false
+am__v_P_1 = :
+AM_V_GEN = $(am__v_GEN_@AM_V@)
+am__v_GEN_ = $(am__v_GEN_@AM_DEFAULT_V@)
+am__v_GEN_0 = @echo "  GEN     " $@;
+am__v_GEN_1 = 
+AM_V_at = $(am__v_at_@AM_V@)
+am__v_at_ = $(am__v_at_@AM_DEFAULT_V@)
+am__v_at_0 = @
+am__v_at_1 = 
 DEFAULT_INCLUDES = -I.@am__isrc@
 depcomp = $(SHELL) $(top_srcdir)/depcomp
 am__depfiles_maybe = depfiles
 am__mv = mv -f
 CXXCOMPILE = $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) \
 	$(AM_CPPFLAGS) $(CPPFLAGS) $(AM_CXXFLAGS) $(CXXFLAGS)
+AM_V_CXX = $(am__v_CXX_@AM_V@)
+am__v_CXX_ = $(am__v_CXX_@AM_DEFAULT_V@)
+am__v_CXX_0 = @echo "  CXX     " $@;
+am__v_CXX_1 = 
 CXXLD = $(CXX)
 CXXLINK = $(CXXLD) $(AM_CXXFLAGS) $(CXXFLAGS) $(AM_LDFLAGS) $(LDFLAGS) \
 	-o $@
+AM_V_CXXLD = $(am__v_CXXLD_@AM_V@)
+am__v_CXXLD_ = $(am__v_CXXLD_@AM_DEFAULT_V@)
+am__v_CXXLD_0 = @echo "  CXXLD   " $@;
+am__v_CXXLD_1 = 
 SOURCES = $(libGrid_a_SOURCES) $(Grid_main_SOURCES)
 DIST_SOURCES = $(libGrid_a_SOURCES) $(Grid_main_SOURCES)
+am__can_run_installinfo = \
+  case $$AM_UPDATE_INFO_DIR in \
+    n|no|NO) false;; \
+    *) (install-info --version) >/dev/null 2>&1;; \
+  esac
 HEADERS = $(include_HEADERS)
+am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) \
+	$(LISP)Grid_config.h.in
+# Read a list of newline-separated strings from the standard input,
+# and print each of them once, without duplicates.  Input order is
+# *not* preserved.
+am__uniquify_input = $(AWK) '\
+  BEGIN { nonempty = 0; } \
+  { items[$$0] = 1; nonempty = 1; } \
+  END { if (nonempty) { for (i in items) print i; }; } \
+'
+# Make sure the list of sources is unique.  This is necessary because,
+# e.g., the same source file might be shared among _SOURCES variables
+# for different programs/libraries.
+am__define_uniq_tagged_files = \
+  list='$(am__tagged_files)'; \
+  unique=`for i in $$list; do \
+    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+  done | $(am__uniquify_input)`
 ETAGS = etags
 CTAGS = ctags
+CSCOPE = cscope
+AM_RECURSIVE_TARGETS = cscope
+am__DIST_COMMON = $(srcdir)/Grid_config.h.in $(srcdir)/Makefile.in \
+	AUTHORS COPYING ChangeLog INSTALL NEWS README compile depcomp \
+	install-sh missing
 DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
 distdir = $(PACKAGE)-$(VERSION)
 top_distdir = $(distdir)
 am__remove_distdir = \
-  { test ! -d "$(distdir)" \
-    || { find "$(distdir)" -type d ! -perm -200 -exec chmod u+w {} ';' \
-         && rm -fr "$(distdir)"; }; }
+  if test -d "$(distdir)"; then \
+    find "$(distdir)" -type d ! -perm -200 -exec chmod u+w {} ';' \
+      && rm -rf "$(distdir)" \
+      || { sleep 5 && rm -rf "$(distdir)"; }; \
+  else :; fi
+am__post_remove_distdir = $(am__remove_distdir)
 DIST_ARCHIVES = $(distdir).tar.gz
 GZIP_ENV = --best
+DIST_TARGETS = dist-gzip
 distuninstallcheck_listfiles = find . -type f -print
+am__distuninstallcheck_listfiles = $(distuninstallcheck_listfiles) \
+  | sed 's|^\./|$(prefix)/|' | grep -v '$(infodir)/dir$$'
 distcleancheck_listfiles = find . -type f -print
 ACLOCAL = @ACLOCAL@
 AMTAR = @AMTAR@
+AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
 AUTOCONF = @AUTOCONF@
 AUTOHEADER = @AUTOHEADER@
 AUTOMAKE = @AUTOMAKE@
@@ -150,6 +267,7 @@ PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
 PACKAGE_NAME = @PACKAGE_NAME@
 PACKAGE_STRING = @PACKAGE_STRING@
 PACKAGE_TARNAME = @PACKAGE_TARNAME@
+PACKAGE_URL = @PACKAGE_URL@
 PACKAGE_VERSION = @PACKAGE_VERSION@
 PATH_SEPARATOR = @PATH_SEPARATOR@
 RANLIB = @RANLIB@
@@ -208,7 +326,7 @@ AM_CXXFLAGS = -I$(top_srcdir)/
 # Libraries
 #
 lib_LIBRARIES = libGrid.a
-libGrid_a_SOURCES = Grid_signal.cc
+libGrid_a_SOURCES = Grid_init.cc
 
 #
 # Include files
@@ -222,14 +340,17 @@ include_HEADERS = Grid.h\
 	Grid_Lattice.h\
 	Grid_config.h
 
-Grid_main_SOURCES = Grid_main.cc
+Grid_main_SOURCES = \
+	Grid_main.cc\
+	Grid_fake.cc
+
 Grid_main_LDADD = libGrid.a
 all: Grid_config.h
 	$(MAKE) $(AM_MAKEFLAGS) all-am
 
 .SUFFIXES:
 .SUFFIXES: .cc .o .obj
-am--refresh:
+am--refresh: Makefile
 	@:
 $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	@for dep in $?; do \
@@ -244,7 +365,6 @@ $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu Makefile'; \
 	$(am__cd) $(top_srcdir) && \
 	  $(AUTOMAKE) --gnu Makefile
-.PRECIOUS: Makefile
 Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
 	@case '$?' in \
 	  *config.status*) \
@@ -265,10 +385,8 @@ $(ACLOCAL_M4):  $(am__aclocal_m4_deps)
 $(am__aclocal_m4_deps):
 
 Grid_config.h: stamp-h1
-	@if test ! -f $@; then \
-	  rm -f stamp-h1; \
-	  $(MAKE) $(AM_MAKEFLAGS) stamp-h1; \
-	else :; fi
+	@test -f $@ || rm -f stamp-h1
+	@test -f $@ || $(MAKE) $(AM_MAKEFLAGS) stamp-h1
 
 stamp-h1: $(srcdir)/Grid_config.h.in $(top_builddir)/config.status
 	@rm -f stamp-h1
@@ -282,7 +400,6 @@ distclean-hdr:
 	-rm -f Grid_config.h stamp-h1
 install-libLIBRARIES: $(lib_LIBRARIES)
 	@$(NORMAL_INSTALL)
-	test -z "$(libdir)" || $(MKDIR_P) "$(DESTDIR)$(libdir)"
 	@list='$(lib_LIBRARIES)'; test -n "$(libdir)" || list=; \
 	list2=; for p in $$list; do \
 	  if test -f $$p; then \
@@ -290,6 +407,8 @@ install-libLIBRARIES: $(lib_LIBRARIES)
 	  else :; fi; \
 	done; \
 	test -z "$$list2" || { \
+	  echo " $(MKDIR_P) '$(DESTDIR)$(libdir)'"; \
+	  $(MKDIR_P) "$(DESTDIR)$(libdir)" || exit 1; \
 	  echo " $(INSTALL_DATA) $$list2 '$(DESTDIR)$(libdir)'"; \
 	  $(INSTALL_DATA) $$list2 "$(DESTDIR)$(libdir)" || exit $$?; }
 	@$(POST_INSTALL)
@@ -306,26 +425,29 @@ uninstall-libLIBRARIES:
 	@$(NORMAL_UNINSTALL)
 	@list='$(lib_LIBRARIES)'; test -n "$(libdir)" || list=; \
 	files=`for p in $$list; do echo $$p; done | sed -e 's|^.*/||'`; \
-	test -n "$$files" || exit 0; \
-	echo " ( cd '$(DESTDIR)$(libdir)' && rm -f "$$files" )"; \
-	cd "$(DESTDIR)$(libdir)" && rm -f $$files
+	dir='$(DESTDIR)$(libdir)'; $(am__uninstall_files_from_dir)
 
 clean-libLIBRARIES:
 	-test -z "$(lib_LIBRARIES)" || rm -f $(lib_LIBRARIES)
-libGrid.a: $(libGrid_a_OBJECTS) $(libGrid_a_DEPENDENCIES) 
-	-rm -f libGrid.a
-	$(libGrid_a_AR) libGrid.a $(libGrid_a_OBJECTS) $(libGrid_a_LIBADD)
-	$(RANLIB) libGrid.a
+
+libGrid.a: $(libGrid_a_OBJECTS) $(libGrid_a_DEPENDENCIES) $(EXTRA_libGrid_a_DEPENDENCIES) 
+	$(AM_V_at)-rm -f libGrid.a
+	$(AM_V_AR)$(libGrid_a_AR) libGrid.a $(libGrid_a_OBJECTS) $(libGrid_a_LIBADD)
+	$(AM_V_at)$(RANLIB) libGrid.a
 install-binPROGRAMS: $(bin_PROGRAMS)
 	@$(NORMAL_INSTALL)
-	test -z "$(bindir)" || $(MKDIR_P) "$(DESTDIR)$(bindir)"
 	@list='$(bin_PROGRAMS)'; test -n "$(bindir)" || list=; \
+	if test -n "$$list"; then \
+	  echo " $(MKDIR_P) '$(DESTDIR)$(bindir)'"; \
+	  $(MKDIR_P) "$(DESTDIR)$(bindir)" || exit 1; \
+	fi; \
 	for p in $$list; do echo "$$p $$p"; done | \
 	sed 's/$(EXEEXT)$$//' | \
-	while read p p1; do if test -f $$p; \
-	  then echo "$$p"; echo "$$p"; else :; fi; \
+	while read p p1; do if test -f $$p \
+	  ; then echo "$$p"; echo "$$p"; else :; fi; \
 	done | \
-	sed -e 'p;s,.*/,,;n;h' -e 's|.*|.|' \
+	sed -e 'p;s,.*/,,;n;h' \
+	    -e 's|.*|.|' \
 	    -e 'p;x;s,.*/,,;s/$(EXEEXT)$$//;$(transform);s/$$/$(EXEEXT)/' | \
 	sed 'N;N;N;s,\n, ,g' | \
 	$(AWK) 'BEGIN { files["."] = ""; dirs["."] = 1 } \
@@ -346,16 +468,18 @@ uninstall-binPROGRAMS:
 	@list='$(bin_PROGRAMS)'; test -n "$(bindir)" || list=; \
 	files=`for p in $$list; do echo "$$p"; done | \
 	  sed -e 'h;s,^.*/,,;s/$(EXEEXT)$$//;$(transform)' \
-	      -e 's/$$/$(EXEEXT)/' `; \
+	      -e 's/$$/$(EXEEXT)/' \
+	`; \
 	test -n "$$list" || exit 0; \
 	echo " ( cd '$(DESTDIR)$(bindir)' && rm -f" $$files ")"; \
 	cd "$(DESTDIR)$(bindir)" && rm -f $$files
 
 clean-binPROGRAMS:
 	-test -z "$(bin_PROGRAMS)" || rm -f $(bin_PROGRAMS)
-Grid_main$(EXEEXT): $(Grid_main_OBJECTS) $(Grid_main_DEPENDENCIES) 
+
+Grid_main$(EXEEXT): $(Grid_main_OBJECTS) $(Grid_main_DEPENDENCIES) $(EXTRA_Grid_main_DEPENDENCIES) 
 	@rm -f Grid_main$(EXEEXT)
-	$(CXXLINK) $(Grid_main_OBJECTS) $(Grid_main_LDADD) $(LIBS)
+	$(AM_V_CXXLD)$(CXXLINK) $(Grid_main_OBJECTS) $(Grid_main_LDADD) $(LIBS)
 
 mostlyclean-compile:
 	-rm -f *.$(OBJEXT)
@@ -363,26 +487,30 @@ mostlyclean-compile:
 distclean-compile:
 	-rm -f *.tab.c
 
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_fake.Po@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_init.Po@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_main.Po@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_signal.Po@am__quote@
 
 .cc.o:
-@am__fastdepCXX_TRUE@	$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
-@am__fastdepCXX_TRUE@	$(am__mv) $(DEPDIR)/$*.Tpo $(DEPDIR)/$*.Po
-@AMDEP_TRUE@@am__fastdepCXX_FALSE@	source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
+@am__fastdepCXX_TRUE@	$(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
+@am__fastdepCXX_TRUE@	$(AM_V_at)$(am__mv) $(DEPDIR)/$*.Tpo $(DEPDIR)/$*.Po
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
 @AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
-@am__fastdepCXX_FALSE@	$(CXXCOMPILE) -c -o $@ $<
+@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ $<
 
 .cc.obj:
-@am__fastdepCXX_TRUE@	$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ `$(CYGPATH_W) '$<'`
-@am__fastdepCXX_TRUE@	$(am__mv) $(DEPDIR)/$*.Tpo $(DEPDIR)/$*.Po
-@AMDEP_TRUE@@am__fastdepCXX_FALSE@	source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
+@am__fastdepCXX_TRUE@	$(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ `$(CYGPATH_W) '$<'`
+@am__fastdepCXX_TRUE@	$(AM_V_at)$(am__mv) $(DEPDIR)/$*.Tpo $(DEPDIR)/$*.Po
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
 @AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
-@am__fastdepCXX_FALSE@	$(CXXCOMPILE) -c -o $@ `$(CYGPATH_W) '$<'`
+@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ `$(CYGPATH_W) '$<'`
 install-includeHEADERS: $(include_HEADERS)
 	@$(NORMAL_INSTALL)
-	test -z "$(includedir)" || $(MKDIR_P) "$(DESTDIR)$(includedir)"
 	@list='$(include_HEADERS)'; test -n "$(includedir)" || list=; \
+	if test -n "$$list"; then \
+	  echo " $(MKDIR_P) '$(DESTDIR)$(includedir)'"; \
+	  $(MKDIR_P) "$(DESTDIR)$(includedir)" || exit 1; \
+	fi; \
 	for p in $$list; do \
 	  if test -f "$$p"; then d=; else d="$(srcdir)/"; fi; \
 	  echo "$$d$$p"; \
@@ -396,30 +524,17 @@ uninstall-includeHEADERS:
 	@$(NORMAL_UNINSTALL)
 	@list='$(include_HEADERS)'; test -n "$(includedir)" || list=; \
 	files=`for p in $$list; do echo $$p; done | sed -e 's|^.*/||'`; \
-	test -n "$$files" || exit 0; \
-	echo " ( cd '$(DESTDIR)$(includedir)' && rm -f" $$files ")"; \
-	cd "$(DESTDIR)$(includedir)" && rm -f $$files
+	dir='$(DESTDIR)$(includedir)'; $(am__uninstall_files_from_dir)
 
-ID: $(HEADERS) $(SOURCES) $(LISP) $(TAGS_FILES)
-	list='$(SOURCES) $(HEADERS) $(LISP) $(TAGS_FILES)'; \
-	unique=`for i in $$list; do \
-	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
-	  done | \
-	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
-	      END { if (nonempty) { for (i in files) print i; }; }'`; \
-	mkid -fID $$unique
-tags: TAGS
+ID: $(am__tagged_files)
+	$(am__define_uniq_tagged_files); mkid -fID $$unique
+tags: tags-am
+TAGS: tags
 
-TAGS:  $(HEADERS) $(SOURCES) Grid_config.h.in $(TAGS_DEPENDENCIES) \
-		$(TAGS_FILES) $(LISP)
+tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
 	set x; \
 	here=`pwd`; \
-	list='$(SOURCES) $(HEADERS) Grid_config.h.in $(LISP) $(TAGS_FILES)'; \
-	unique=`for i in $$list; do \
-	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
-	  done | \
-	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
-	      END { if (nonempty) { for (i in files) print i; }; }'`; \
+	$(am__define_uniq_tagged_files); \
 	shift; \
 	if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
 	  test -n "$$unique" || unique=$$empty_fix; \
@@ -431,15 +546,11 @@ TAGS:  $(HEADERS) $(SOURCES) Grid_config.h.in $(TAGS_DEPENDENCIES) \
 	      $$unique; \
 	  fi; \
 	fi
-ctags: CTAGS
-CTAGS:  $(HEADERS) $(SOURCES) Grid_config.h.in $(TAGS_DEPENDENCIES) \
-		$(TAGS_FILES) $(LISP)
-	list='$(SOURCES) $(HEADERS) Grid_config.h.in $(LISP) $(TAGS_FILES)'; \
-	unique=`for i in $$list; do \
-	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
-	  done | \
-	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
-	      END { if (nonempty) { for (i in files) print i; }; }'`; \
+ctags: ctags-am
+
+CTAGS: ctags
+ctags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
+	$(am__define_uniq_tagged_files); \
 	test -z "$(CTAGS_ARGS)$$unique" \
 	  || $(CTAGS) $(CTAGSFLAGS) $(AM_CTAGSFLAGS) $(CTAGS_ARGS) \
 	     $$unique
@@ -448,9 +559,31 @@ GTAGS:
 	here=`$(am__cd) $(top_builddir) && pwd` \
 	  && $(am__cd) $(top_srcdir) \
 	  && gtags -i $(GTAGS_ARGS) "$$here"
+cscope: cscope.files
+	test ! -s cscope.files \
+	  || $(CSCOPE) -b -q $(AM_CSCOPEFLAGS) $(CSCOPEFLAGS) -i cscope.files $(CSCOPE_ARGS)
+clean-cscope:
+	-rm -f cscope.files
+cscope.files: clean-cscope cscopelist
+cscopelist: cscopelist-am
+
+cscopelist-am: $(am__tagged_files)
+	list='$(am__tagged_files)'; \
+	case "$(srcdir)" in \
+	  [\\/]* | ?:[\\/]*) sdir="$(srcdir)" ;; \
+	  *) sdir=$(subdir)/$(srcdir) ;; \
+	esac; \
+	for i in $$list; do \
+	  if test -f "$$i"; then \
+	    echo "$(subdir)/$$i"; \
+	  else \
+	    echo "$$sdir/$$i"; \
+	  fi; \
+	done >> $(top_builddir)/cscope.files
 
 distclean-tags:
 	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
+	-rm -f cscope.out cscope.in.out cscope.po.out cscope.files
 
 distdir: $(DISTFILES)
 	$(am__remove_distdir)
@@ -493,36 +626,42 @@ distdir: $(DISTFILES)
 	|| chmod -R a+r "$(distdir)"
 dist-gzip: distdir
 	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
-	$(am__remove_distdir)
+	$(am__post_remove_distdir)
 
 dist-bzip2: distdir
-	tardir=$(distdir) && $(am__tar) | bzip2 -9 -c >$(distdir).tar.bz2
-	$(am__remove_distdir)
+	tardir=$(distdir) && $(am__tar) | BZIP2=$${BZIP2--9} bzip2 -c >$(distdir).tar.bz2
+	$(am__post_remove_distdir)
 
-dist-lzma: distdir
-	tardir=$(distdir) && $(am__tar) | lzma -9 -c >$(distdir).tar.lzma
-	$(am__remove_distdir)
+dist-lzip: distdir
+	tardir=$(distdir) && $(am__tar) | lzip -c $${LZIP_OPT--9} >$(distdir).tar.lz
+	$(am__post_remove_distdir)
 
 dist-xz: distdir
-	tardir=$(distdir) && $(am__tar) | xz -c >$(distdir).tar.xz
-	$(am__remove_distdir)
+	tardir=$(distdir) && $(am__tar) | XZ_OPT=$${XZ_OPT--e} xz -c >$(distdir).tar.xz
+	$(am__post_remove_distdir)
 
 dist-tarZ: distdir
+	@echo WARNING: "Support for distribution archives compressed with" \
+		       "legacy program 'compress' is deprecated." >&2
+	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	tardir=$(distdir) && $(am__tar) | compress -c >$(distdir).tar.Z
-	$(am__remove_distdir)
+	$(am__post_remove_distdir)
 
 dist-shar: distdir
+	@echo WARNING: "Support for shar distribution archives is" \
+	               "deprecated." >&2
+	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	shar $(distdir) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).shar.gz
-	$(am__remove_distdir)
+	$(am__post_remove_distdir)
 
 dist-zip: distdir
 	-rm -f $(distdir).zip
 	zip -rq $(distdir).zip $(distdir)
-	$(am__remove_distdir)
+	$(am__post_remove_distdir)
 
-dist dist-all: distdir
-	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
-	$(am__remove_distdir)
+dist dist-all:
+	$(MAKE) $(AM_MAKEFLAGS) $(DIST_TARGETS) am__post_remove_distdir='@:'
+	$(am__post_remove_distdir)
 
 # This target untars the dist file and tries a VPATH configuration.  Then
 # it guarantees that the distribution is self-contained by making another
@@ -533,8 +672,8 @@ distcheck: dist
 	  GZIP=$(GZIP_ENV) gzip -dc $(distdir).tar.gz | $(am__untar) ;;\
 	*.tar.bz2*) \
 	  bzip2 -dc $(distdir).tar.bz2 | $(am__untar) ;;\
-	*.tar.lzma*) \
-	  lzma -dc $(distdir).tar.lzma | $(am__untar) ;;\
+	*.tar.lz*) \
+	  lzip -dc $(distdir).tar.lz | $(am__untar) ;;\
 	*.tar.xz*) \
 	  xz -dc $(distdir).tar.xz | $(am__untar) ;;\
 	*.tar.Z*) \
@@ -544,17 +683,19 @@ distcheck: dist
 	*.zip*) \
 	  unzip $(distdir).zip ;;\
 	esac
-	chmod -R a-w $(distdir); chmod u+w $(distdir)
-	mkdir $(distdir)/_build
-	mkdir $(distdir)/_inst
+	chmod -R a-w $(distdir)
+	chmod u+w $(distdir)
+	mkdir $(distdir)/_build $(distdir)/_build/sub $(distdir)/_inst
 	chmod a-w $(distdir)
 	test -d $(distdir)/_build || exit 0; \
 	dc_install_base=`$(am__cd) $(distdir)/_inst && pwd | sed -e 's,^[^:\\/]:[\\/],/,'` \
 	  && dc_destdir="$${TMPDIR-/tmp}/am-dc-$$$$/" \
 	  && am__cwd=`pwd` \
-	  && $(am__cd) $(distdir)/_build \
-	  && ../configure --srcdir=.. --prefix="$$dc_install_base" \
+	  && $(am__cd) $(distdir)/_build/sub \
+	  && ../../configure \
+	    $(AM_DISTCHECK_CONFIGURE_FLAGS) \
 	    $(DISTCHECK_CONFIGURE_FLAGS) \
+	    --srcdir=../.. --prefix="$$dc_install_base" \
 	  && $(MAKE) $(AM_MAKEFLAGS) \
 	  && $(MAKE) $(AM_MAKEFLAGS) dvi \
 	  && $(MAKE) $(AM_MAKEFLAGS) check \
@@ -577,13 +718,21 @@ distcheck: dist
 	  && $(MAKE) $(AM_MAKEFLAGS) distcleancheck \
 	  && cd "$$am__cwd" \
 	  || exit 1
-	$(am__remove_distdir)
+	$(am__post_remove_distdir)
 	@(echo "$(distdir) archives ready for distribution: "; \
 	  list='$(DIST_ARCHIVES)'; for i in $$list; do echo $$i; done) | \
 	  sed -e 1h -e 1s/./=/g -e 1p -e 1x -e '$$p' -e '$$x'
 distuninstallcheck:
-	@$(am__cd) '$(distuninstallcheck_dir)' \
-	&& test `$(distuninstallcheck_listfiles) | wc -l` -le 1 \
+	@test -n '$(distuninstallcheck_dir)' || { \
+	  echo 'ERROR: trying to run $@ with an empty' \
+	       '$$(distuninstallcheck_dir)' >&2; \
+	  exit 1; \
+	}; \
+	$(am__cd) '$(distuninstallcheck_dir)' || { \
+	  echo 'ERROR: cannot chdir into $(distuninstallcheck_dir)' >&2; \
+	  exit 1; \
+	}; \
+	test `$(am__distuninstallcheck_listfiles) | wc -l` -eq 0 \
 	   || { echo "ERROR: files left after uninstall:" ; \
 	        if test -n "$(DESTDIR)"; then \
 	          echo "  (check DESTDIR support)"; \
@@ -616,10 +765,15 @@ install-am: all-am
 
 installcheck: installcheck-am
 install-strip:
-	$(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
-	  install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
-	  `test -z '$(STRIP)' || \
-	    echo "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'"` install
+	if test -z '$(STRIP)'; then \
+	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
+	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
+	      install; \
+	else \
+	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
+	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
+	    "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'" install; \
+	fi
 mostlyclean-generic:
 
 clean-generic:
@@ -707,9 +861,10 @@ uninstall-am: uninstall-binPROGRAMS uninstall-includeHEADERS \
 
 .MAKE: all install-am install-strip
 
-.PHONY: CTAGS GTAGS all all-am am--refresh check check-am clean \
-	clean-binPROGRAMS clean-generic clean-libLIBRARIES ctags dist \
-	dist-all dist-bzip2 dist-gzip dist-lzma dist-shar dist-tarZ \
+.PHONY: CTAGS GTAGS TAGS all all-am am--refresh check check-am clean \
+	clean-binPROGRAMS clean-cscope clean-generic \
+	clean-libLIBRARIES cscope cscopelist-am ctags ctags-am dist \
+	dist-all dist-bzip2 dist-gzip dist-lzip dist-shar dist-tarZ \
 	dist-xz dist-zip distcheck distclean distclean-compile \
 	distclean-generic distclean-hdr distclean-tags distcleancheck \
 	distdir distuninstallcheck dvi dvi-am html html-am info \
@@ -721,10 +876,12 @@ uninstall-am: uninstall-binPROGRAMS uninstall-includeHEADERS \
 	install-ps install-ps-am install-strip installcheck \
 	installcheck-am installdirs maintainer-clean \
 	maintainer-clean-generic mostlyclean mostlyclean-compile \
-	mostlyclean-generic pdf pdf-am ps ps-am tags uninstall \
+	mostlyclean-generic pdf pdf-am ps ps-am tags tags-am uninstall \
 	uninstall-am uninstall-binPROGRAMS uninstall-includeHEADERS \
 	uninstall-libLIBRARIES
 
+.PRECIOUS: Makefile
+
 
 # Tell versions [3.59,3.63) of GNU make to not export all variables.
 # Otherwise a system limit (for SysV at least) may be exceeded.
diff --git a/configure b/configure
index f63f8031..e048369e 100755
--- a/configure
+++ b/configure
@@ -1,20 +1,22 @@
 #! /bin/sh
 # Guess values for system-dependent variables and create Makefiles.
-# Generated by GNU Autoconf 2.63 for Grid 1.0.
+# Generated by GNU Autoconf 2.69 for Grid 1.0.
 #
 # Report bugs to <paboyle@ph.ed.ac.uk>.
 #
-# Copyright (C) 1992, 1993, 1994, 1995, 1996, 1998, 1999, 2000, 2001,
-# 2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
+#
+# Copyright (C) 1992-1996, 1998-2012 Free Software Foundation, Inc.
+#
+#
 # This configure script is free software; the Free Software Foundation
 # gives unlimited permission to copy, distribute and modify it.
-## --------------------- ##
-## M4sh Initialization.  ##
-## --------------------- ##
+## -------------------- ##
+## M4sh Initialization. ##
+## -------------------- ##
 
 # Be more Bourne compatible
 DUALCASE=1; export DUALCASE # for MKS sh
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
   emulate sh
   NULLCMD=:
   # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
@@ -22,23 +24,15 @@ if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
   alias -g '${1+"$@"}'='"$@"'
   setopt NO_GLOB_SUBST
 else
-  case `(set -o) 2>/dev/null` in
-  *posix*) set -o posix ;;
+  case `(set -o) 2>/dev/null` in #(
+  *posix*) :
+    set -o posix ;; #(
+  *) :
+     ;;
 esac
-
 fi
 
 
-
-
-# PATH needs CR
-# Avoid depending upon Character Ranges.
-as_cr_letters='abcdefghijklmnopqrstuvwxyz'
-as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
-as_cr_Letters=$as_cr_letters$as_cr_LETTERS
-as_cr_digits='0123456789'
-as_cr_alnum=$as_cr_Letters$as_cr_digits
-
 as_nl='
 '
 export as_nl
@@ -46,7 +40,13 @@ export as_nl
 as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
-if (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+# Prefer a ksh shell builtin over an external printf program on Solaris,
+# but without wasting forks for bash or zsh.
+if test -z "$BASH_VERSION$ZSH_VERSION" \
+    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
+  as_echo='print -r --'
+  as_echo_n='print -rn --'
+elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
   as_echo='printf %s\n'
   as_echo_n='printf %s'
 else
@@ -57,7 +57,7 @@ else
     as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
     as_echo_n_body='eval
       arg=$1;
-      case $arg in
+      case $arg in #(
       *"$as_nl"*)
 	expr "X$arg" : "X\\(.*\\)$as_nl";
 	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
@@ -80,13 +80,6 @@ if test "${PATH_SEPARATOR+set}" != set; then
   }
 fi
 
-# Support unset when possible.
-if ( (MAIL=60; unset MAIL) || exit) >/dev/null 2>&1; then
-  as_unset=unset
-else
-  as_unset=false
-fi
-
 
 # IFS
 # We need space, tab and new line, in precisely that order.  Quoting is
@@ -96,15 +89,16 @@ fi
 IFS=" ""	$as_nl"
 
 # Find who we are.  Look in the path if we contain no directory separator.
-case $0 in
+as_myself=
+case $0 in #((
   *[\\/]* ) as_myself=$0 ;;
   *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
 for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
-done
+    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+  done
 IFS=$as_save_IFS
 
      ;;
@@ -116,12 +110,16 @@ if test "x$as_myself" = x; then
 fi
 if test ! -f "$as_myself"; then
   $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
-  { (exit 1); exit 1; }
+  exit 1
 fi
 
-# Work around bugs in pre-3.0 UWIN ksh.
-for as_var in ENV MAIL MAILPATH
-do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
+# Unset variables that we do not need and which cause bugs (e.g. in
+# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
+# suppresses any "Segmentation fault" message there.  '((' could
+# trigger a bug in pdksh 5.2.14.
+for as_var in BASH_ENV ENV MAIL MAILPATH
+do eval test x\${$as_var+set} = xset \
+  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
 done
 PS1='$ '
 PS2='> '
@@ -133,7 +131,294 @@ export LC_ALL
 LANGUAGE=C
 export LANGUAGE
 
-# Required to use basename.
+# CDPATH.
+(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
+
+# Use a proper internal environment variable to ensure we don't fall
+  # into an infinite loop, continuously re-executing ourselves.
+  if test x"${_as_can_reexec}" != xno && test "x$CONFIG_SHELL" != x; then
+    _as_can_reexec=no; export _as_can_reexec;
+    # We cannot yet assume a decent shell, so we have to provide a
+# neutralization value for shells without unset; and this also
+# works around shells that cannot unset nonexistent variables.
+# Preserve -v and -x to the replacement shell.
+BASH_ENV=/dev/null
+ENV=/dev/null
+(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
+case $- in # ((((
+  *v*x* | *x*v* ) as_opts=-vx ;;
+  *v* ) as_opts=-v ;;
+  *x* ) as_opts=-x ;;
+  * ) as_opts= ;;
+esac
+exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
+# Admittedly, this is quite paranoid, since all the known shells bail
+# out after a failed `exec'.
+$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
+as_fn_exit 255
+  fi
+  # We don't want this to propagate to other subprocesses.
+          { _as_can_reexec=; unset _as_can_reexec;}
+if test "x$CONFIG_SHELL" = x; then
+  as_bourne_compatible="if test -n \"\${ZSH_VERSION+set}\" && (emulate sh) >/dev/null 2>&1; then :
+  emulate sh
+  NULLCMD=:
+  # Pre-4.2 versions of Zsh do word splitting on \${1+\"\$@\"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '\${1+\"\$@\"}'='\"\$@\"'
+  setopt NO_GLOB_SUBST
+else
+  case \`(set -o) 2>/dev/null\` in #(
+  *posix*) :
+    set -o posix ;; #(
+  *) :
+     ;;
+esac
+fi
+"
+  as_required="as_fn_return () { (exit \$1); }
+as_fn_success () { as_fn_return 0; }
+as_fn_failure () { as_fn_return 1; }
+as_fn_ret_success () { return 0; }
+as_fn_ret_failure () { return 1; }
+
+exitcode=0
+as_fn_success || { exitcode=1; echo as_fn_success failed.; }
+as_fn_failure && { exitcode=1; echo as_fn_failure succeeded.; }
+as_fn_ret_success || { exitcode=1; echo as_fn_ret_success failed.; }
+as_fn_ret_failure && { exitcode=1; echo as_fn_ret_failure succeeded.; }
+if ( set x; as_fn_ret_success y && test x = \"\$1\" ); then :
+
+else
+  exitcode=1; echo positional parameters were not saved.
+fi
+test x\$exitcode = x0 || exit 1
+test -x / || exit 1"
+  as_suggested="  as_lineno_1=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_1a=\$LINENO
+  as_lineno_2=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_2a=\$LINENO
+  eval 'test \"x\$as_lineno_1'\$as_run'\" != \"x\$as_lineno_2'\$as_run'\" &&
+  test \"x\`expr \$as_lineno_1'\$as_run' + 1\`\" = \"x\$as_lineno_2'\$as_run'\"' || exit 1
+test \$(( 1 + 1 )) = 2 || exit 1"
+  if (eval "$as_required") 2>/dev/null; then :
+  as_have_required=yes
+else
+  as_have_required=no
+fi
+  if test x$as_have_required = xyes && (eval "$as_suggested") 2>/dev/null; then :
+
+else
+  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+as_found=false
+for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+  as_found=:
+  case $as_dir in #(
+	 /*)
+	   for as_base in sh bash ksh sh5; do
+	     # Try only shells that exist, to save several forks.
+	     as_shell=$as_dir/$as_base
+	     if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
+		    { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$as_shell"; } 2>/dev/null; then :
+  CONFIG_SHELL=$as_shell as_have_required=yes
+		   if { $as_echo "$as_bourne_compatible""$as_suggested" | as_run=a "$as_shell"; } 2>/dev/null; then :
+  break 2
+fi
+fi
+	   done;;
+       esac
+  as_found=false
+done
+$as_found || { if { test -f "$SHELL" || test -f "$SHELL.exe"; } &&
+	      { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$SHELL"; } 2>/dev/null; then :
+  CONFIG_SHELL=$SHELL as_have_required=yes
+fi; }
+IFS=$as_save_IFS
+
+
+      if test "x$CONFIG_SHELL" != x; then :
+  export CONFIG_SHELL
+             # We cannot yet assume a decent shell, so we have to provide a
+# neutralization value for shells without unset; and this also
+# works around shells that cannot unset nonexistent variables.
+# Preserve -v and -x to the replacement shell.
+BASH_ENV=/dev/null
+ENV=/dev/null
+(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
+case $- in # ((((
+  *v*x* | *x*v* ) as_opts=-vx ;;
+  *v* ) as_opts=-v ;;
+  *x* ) as_opts=-x ;;
+  * ) as_opts= ;;
+esac
+exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
+# Admittedly, this is quite paranoid, since all the known shells bail
+# out after a failed `exec'.
+$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
+exit 255
+fi
+
+    if test x$as_have_required = xno; then :
+  $as_echo "$0: This script requires a shell more modern than all"
+  $as_echo "$0: the shells that I found on your system."
+  if test x${ZSH_VERSION+set} = xset ; then
+    $as_echo "$0: In particular, zsh $ZSH_VERSION has bugs and should"
+    $as_echo "$0: be upgraded to zsh 4.3.4 or later."
+  else
+    $as_echo "$0: Please tell bug-autoconf@gnu.org and
+$0: paboyle@ph.ed.ac.uk about your system, including any
+$0: error possibly output before this message. Then install
+$0: a modern shell, or manually run the script under such a
+$0: shell if you do have one."
+  fi
+  exit 1
+fi
+fi
+fi
+SHELL=${CONFIG_SHELL-/bin/sh}
+export SHELL
+# Unset more variables known to interfere with behavior of common tools.
+CLICOLOR_FORCE= GREP_OPTIONS=
+unset CLICOLOR_FORCE GREP_OPTIONS
+
+## --------------------- ##
+## M4sh Shell Functions. ##
+## --------------------- ##
+# as_fn_unset VAR
+# ---------------
+# Portably unset VAR.
+as_fn_unset ()
+{
+  { eval $1=; unset $1;}
+}
+as_unset=as_fn_unset
+
+# as_fn_set_status STATUS
+# -----------------------
+# Set $? to STATUS, without forking.
+as_fn_set_status ()
+{
+  return $1
+} # as_fn_set_status
+
+# as_fn_exit STATUS
+# -----------------
+# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
+as_fn_exit ()
+{
+  set +e
+  as_fn_set_status $1
+  exit $1
+} # as_fn_exit
+
+# as_fn_mkdir_p
+# -------------
+# Create "$as_dir" as a directory, including parents if necessary.
+as_fn_mkdir_p ()
+{
+
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || eval $as_mkdir_p || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
+
+
+} # as_fn_mkdir_p
+
+# as_fn_executable_p FILE
+# -----------------------
+# Test if FILE is an executable regular file.
+as_fn_executable_p ()
+{
+  test -f "$1" && test -x "$1"
+} # as_fn_executable_p
+# as_fn_append VAR VALUE
+# ----------------------
+# Append the text in VALUE to the end of the definition contained in VAR. Take
+# advantage of any shell optimizations that allow amortized linear growth over
+# repeated appends, instead of the typical quadratic growth present in naive
+# implementations.
+if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
+  eval 'as_fn_append ()
+  {
+    eval $1+=\$2
+  }'
+else
+  as_fn_append ()
+  {
+    eval $1=\$$1\$2
+  }
+fi # as_fn_append
+
+# as_fn_arith ARG...
+# ------------------
+# Perform arithmetic evaluation on the ARGs, and store the result in the
+# global $as_val. Take advantage of shells that can avoid forks. The arguments
+# must be portable across $(()) and expr.
+if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
+  eval 'as_fn_arith ()
+  {
+    as_val=$(( $* ))
+  }'
+else
+  as_fn_arith ()
+  {
+    as_val=`expr "$@" || test $? -eq 1`
+  }
+fi # as_fn_arith
+
+
+# as_fn_error STATUS ERROR [LINENO LOG_FD]
+# ----------------------------------------
+# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
+# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
+# script with STATUS, using 1 if that was 0.
+as_fn_error ()
+{
+  as_status=$1; test $as_status -eq 0 && as_status=1
+  if test "$4"; then
+    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
+  fi
+  $as_echo "$as_me: error: $2" >&2
+  as_fn_exit $as_status
+} # as_fn_error
+
 if expr a : '\(a\)' >/dev/null 2>&1 &&
    test "X`expr 00001 : '.*\(...\)'`" = X001; then
   as_expr=expr
@@ -147,8 +432,12 @@ else
   as_basename=false
 fi
 
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
 
-# Name of the executable.
 as_me=`$as_basename -- "$0" ||
 $as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
 	 X"$0" : 'X\(//\)$' \| \
@@ -168,295 +457,19 @@ $as_echo X/"$0" |
 	  }
 	  s/.*/./; q'`
 
-# CDPATH.
-$as_unset CDPATH
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
 
 
-if test "x$CONFIG_SHELL" = x; then
-  if (eval ":") 2>/dev/null; then
-  as_have_required=yes
-else
-  as_have_required=no
-fi
-
-  if test $as_have_required = yes &&	 (eval ":
-(as_func_return () {
-  (exit \$1)
-}
-as_func_success () {
-  as_func_return 0
-}
-as_func_failure () {
-  as_func_return 1
-}
-as_func_ret_success () {
-  return 0
-}
-as_func_ret_failure () {
-  return 1
-}
-
-exitcode=0
-if as_func_success; then
-  :
-else
-  exitcode=1
-  echo as_func_success failed.
-fi
-
-if as_func_failure; then
-  exitcode=1
-  echo as_func_failure succeeded.
-fi
-
-if as_func_ret_success; then
-  :
-else
-  exitcode=1
-  echo as_func_ret_success failed.
-fi
-
-if as_func_ret_failure; then
-  exitcode=1
-  echo as_func_ret_failure succeeded.
-fi
-
-if ( set x; as_func_ret_success y && test x = \"\$1\" ); then
-  :
-else
-  exitcode=1
-  echo positional parameters were not saved.
-fi
-
-test \$exitcode = 0) || { (exit 1); exit 1; }
-
-(
-  as_lineno_1=\$LINENO
-  as_lineno_2=\$LINENO
-  test \"x\$as_lineno_1\" != \"x\$as_lineno_2\" &&
-  test \"x\`expr \$as_lineno_1 + 1\`\" = \"x\$as_lineno_2\") || { (exit 1); exit 1; }
-") 2> /dev/null; then
-  :
-else
-  as_candidate_shells=
-    as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-  case $as_dir in
-	 /*)
-	   for as_base in sh bash ksh sh5; do
-	     as_candidate_shells="$as_candidate_shells $as_dir/$as_base"
-	   done;;
-       esac
-done
-IFS=$as_save_IFS
-
-
-      for as_shell in $as_candidate_shells $SHELL; do
-	 # Try only shells that exist, to save several forks.
-	 if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
-		{ ("$as_shell") 2> /dev/null <<\_ASEOF
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
-  emulate sh
-  NULLCMD=:
-  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
-  # is contrary to our usage.  Disable this feature.
-  alias -g '${1+"$@"}'='"$@"'
-  setopt NO_GLOB_SUBST
-else
-  case `(set -o) 2>/dev/null` in
-  *posix*) set -o posix ;;
-esac
-
-fi
-
-
-:
-_ASEOF
-}; then
-  CONFIG_SHELL=$as_shell
-	       as_have_required=yes
-	       if { "$as_shell" 2> /dev/null <<\_ASEOF
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
-  emulate sh
-  NULLCMD=:
-  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
-  # is contrary to our usage.  Disable this feature.
-  alias -g '${1+"$@"}'='"$@"'
-  setopt NO_GLOB_SUBST
-else
-  case `(set -o) 2>/dev/null` in
-  *posix*) set -o posix ;;
-esac
-
-fi
-
-
-:
-(as_func_return () {
-  (exit $1)
-}
-as_func_success () {
-  as_func_return 0
-}
-as_func_failure () {
-  as_func_return 1
-}
-as_func_ret_success () {
-  return 0
-}
-as_func_ret_failure () {
-  return 1
-}
-
-exitcode=0
-if as_func_success; then
-  :
-else
-  exitcode=1
-  echo as_func_success failed.
-fi
-
-if as_func_failure; then
-  exitcode=1
-  echo as_func_failure succeeded.
-fi
-
-if as_func_ret_success; then
-  :
-else
-  exitcode=1
-  echo as_func_ret_success failed.
-fi
-
-if as_func_ret_failure; then
-  exitcode=1
-  echo as_func_ret_failure succeeded.
-fi
-
-if ( set x; as_func_ret_success y && test x = "$1" ); then
-  :
-else
-  exitcode=1
-  echo positional parameters were not saved.
-fi
-
-test $exitcode = 0) || { (exit 1); exit 1; }
-
-(
-  as_lineno_1=$LINENO
-  as_lineno_2=$LINENO
-  test "x$as_lineno_1" != "x$as_lineno_2" &&
-  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2") || { (exit 1); exit 1; }
-
-_ASEOF
-}; then
-  break
-fi
-
-fi
-
-      done
-
-      if test "x$CONFIG_SHELL" != x; then
-  for as_var in BASH_ENV ENV
-	do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
-	done
-	export CONFIG_SHELL
-	exec "$CONFIG_SHELL" "$as_myself" ${1+"$@"}
-fi
-
-
-    if test $as_have_required = no; then
-  echo This script requires a shell more modern than all the
-      echo shells that I found on your system.  Please install a
-      echo modern shell, or manually run the script under such a
-      echo shell if you do have one.
-      { (exit 1); exit 1; }
-fi
-
-
-fi
-
-fi
-
-
-
-(eval "as_func_return () {
-  (exit \$1)
-}
-as_func_success () {
-  as_func_return 0
-}
-as_func_failure () {
-  as_func_return 1
-}
-as_func_ret_success () {
-  return 0
-}
-as_func_ret_failure () {
-  return 1
-}
-
-exitcode=0
-if as_func_success; then
-  :
-else
-  exitcode=1
-  echo as_func_success failed.
-fi
-
-if as_func_failure; then
-  exitcode=1
-  echo as_func_failure succeeded.
-fi
-
-if as_func_ret_success; then
-  :
-else
-  exitcode=1
-  echo as_func_ret_success failed.
-fi
-
-if as_func_ret_failure; then
-  exitcode=1
-  echo as_func_ret_failure succeeded.
-fi
-
-if ( set x; as_func_ret_success y && test x = \"\$1\" ); then
-  :
-else
-  exitcode=1
-  echo positional parameters were not saved.
-fi
-
-test \$exitcode = 0") || {
-  echo No shell found that supports shell functions.
-  echo Please tell bug-autoconf@gnu.org about your system,
-  echo including any error possibly output before this message.
-  echo This can help us improve future autoconf versions.
-  echo Configuration will now proceed without shell functions.
-}
-
-
-
-  as_lineno_1=$LINENO
-  as_lineno_2=$LINENO
-  test "x$as_lineno_1" != "x$as_lineno_2" &&
-  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2" || {
-
-  # Create $as_me.lineno as a copy of $as_myself, but with $LINENO
-  # uniformly replaced by the line number.  The first 'sed' inserts a
-  # line-number line after each line using $LINENO; the second 'sed'
-  # does the real work.  The second script uses 'N' to pair each
-  # line-number line with the line containing $LINENO, and appends
-  # trailing '-' during substitution so that $LINENO is not a special
-  # case at line end.
-  # (Raja R Harinath suggested sed '=', and Paul Eggert wrote the
-  # scripts with optimization help from Paolo Bonzini.  Blame Lee
-  # E. McMahon (1931-1989) for sed's syntax.  :-)
+  as_lineno_1=$LINENO as_lineno_1a=$LINENO
+  as_lineno_2=$LINENO as_lineno_2a=$LINENO
+  eval 'test "x$as_lineno_1'$as_run'" != "x$as_lineno_2'$as_run'" &&
+  test "x`expr $as_lineno_1'$as_run' + 1`" = "x$as_lineno_2'$as_run'"' || {
+  # Blame Lee E. McMahon (1931-1989) for sed's syntax.  :-)
   sed -n '
     p
     /[$]LINENO/=
@@ -473,9 +486,12 @@ test \$exitcode = 0") || {
       s/-\n.*//
     ' >$as_me.lineno &&
   chmod +x "$as_me.lineno" ||
-    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2
-   { (exit 1); exit 1; }; }
+    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2; as_fn_exit 1; }
 
+  # If we had to re-execute with $CONFIG_SHELL, we're ensured to have
+  # already done that, so ensure we don't try to do so again and fall
+  # in an infinite loop.  This has already happened in practice.
+  _as_can_reexec=no; export _as_can_reexec
   # Don't try to exec as it changes $[0], causing all sort of problems
   # (the dirname of $[0] is not the place where we might find the
   # original and so on.  Autoconf is especially sensitive to this).
@@ -484,29 +500,18 @@ test \$exitcode = 0") || {
   exit
 }
 
-
-if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
-  as_dirname=dirname
-else
-  as_dirname=false
-fi
-
 ECHO_C= ECHO_N= ECHO_T=
-case `echo -n x` in
+case `echo -n x` in #(((((
 -n*)
-  case `echo 'x\c'` in
+  case `echo 'xy\c'` in
   *c*) ECHO_T='	';;	# ECHO_T is single tab character.
-  *)   ECHO_C='\c';;
+  xy)  ECHO_C='\c';;
+  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
+       ECHO_T='	';;
   esac;;
 *)
   ECHO_N='-n';;
 esac
-if expr a : '\(a\)' >/dev/null 2>&1 &&
-   test "X`expr 00001 : '.*\(...\)'`" = X001; then
-  as_expr=expr
-else
-  as_expr=false
-fi
 
 rm -f conf$$ conf$$.exe conf$$.file
 if test -d conf$$.dir; then
@@ -521,49 +526,29 @@ if (echo >conf$$.file) 2>/dev/null; then
     # ... but there are two gotchas:
     # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
     # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
-    # In both cases, we have to default to `cp -p'.
+    # In both cases, we have to default to `cp -pR'.
     ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
-      as_ln_s='cp -p'
+      as_ln_s='cp -pR'
   elif ln conf$$.file conf$$ 2>/dev/null; then
     as_ln_s=ln
   else
-    as_ln_s='cp -p'
+    as_ln_s='cp -pR'
   fi
 else
-  as_ln_s='cp -p'
+  as_ln_s='cp -pR'
 fi
 rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
 rmdir conf$$.dir 2>/dev/null
 
 if mkdir -p . 2>/dev/null; then
-  as_mkdir_p=:
+  as_mkdir_p='mkdir -p "$as_dir"'
 else
   test -d ./-p && rmdir ./-p
   as_mkdir_p=false
 fi
 
-if test -x / >/dev/null 2>&1; then
-  as_test_x='test -x'
-else
-  if ls -dL / >/dev/null 2>&1; then
-    as_ls_L_option=L
-  else
-    as_ls_L_option=
-  fi
-  as_test_x='
-    eval sh -c '\''
-      if test -d "$1"; then
-	test -d "$1/.";
-      else
-	case $1 in
-	-*)set "./$1";;
-	esac;
-	case `ls -ld'$as_ls_L_option' "$1" 2>/dev/null` in
-	???[sx]*):;;*)false;;esac;fi
-    '\'' sh
-  '
-fi
-as_executable_p=$as_test_x
+as_test_x='test -x'
+as_executable_p=as_fn_executable_p
 
 # Sed expression to map a string onto a valid CPP name.
 as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
@@ -572,11 +557,11 @@ as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
 as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
 
 
-
-exec 7<&0 </dev/null 6>&1
+test -n "$DJDIR" || exec 7<&0 </dev/null
+exec 6>&1
 
 # Name of the host.
-# hostname on some systems (SVR3.2, Linux) returns a bogus exit status,
+# hostname on some systems (SVR3.2, old GNU/Linux) returns a bogus exit status,
 # so uname gets run too.
 ac_hostname=`(hostname || uname -n) 2>/dev/null | sed 1q`
 
@@ -591,7 +576,6 @@ cross_compiling=no
 subdirs=
 MFLAGS=
 MAKEFLAGS=
-SHELL=${CONFIG_SHELL-/bin/sh}
 
 # Identity of this package.
 PACKAGE_NAME='Grid'
@@ -599,6 +583,7 @@ PACKAGE_TARNAME='grid'
 PACKAGE_VERSION='1.0'
 PACKAGE_STRING='Grid 1.0'
 PACKAGE_BUGREPORT='paboyle@ph.ed.ac.uk'
+PACKAGE_URL=''
 
 ac_unique_file="Grid.h"
 # Factoring default headers for most tests.
@@ -655,6 +640,7 @@ CC
 am__fastdepCXX_FALSE
 am__fastdepCXX_TRUE
 CXXDEPMODE
+am__nodep
 AMDEPBACKSLASH
 AMDEP_FALSE
 AMDEP_TRUE
@@ -668,6 +654,10 @@ CPPFLAGS
 LDFLAGS
 CXXFLAGS
 CXX
+AM_BACKSLASH
+AM_DEFAULT_VERBOSITY
+AM_DEFAULT_V
+AM_V
 am__untar
 am__tar
 AMTAR
@@ -721,6 +711,7 @@ bindir
 program_transform_name
 prefix
 exec_prefix
+PACKAGE_URL
 PACKAGE_BUGREPORT
 PACKAGE_STRING
 PACKAGE_VERSION
@@ -731,9 +722,11 @@ SHELL'
 ac_subst_files=''
 ac_user_opts='
 enable_option_checking
+enable_silent_rules
 enable_dependency_tracking
 enable_openmp
 enable_simd
+enable_comms
 '
       ac_precious_vars='build_alias
 host_alias
@@ -809,8 +802,9 @@ do
   fi
 
   case $ac_option in
-  *=*)	ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
-  *)	ac_optarg=yes ;;
+  *=?*) ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
+  *=)   ac_optarg= ;;
+  *)    ac_optarg=yes ;;
   esac
 
   # Accept the important Cygnus configure options, so we can diagnose typos.
@@ -855,8 +849,7 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*disable-\(.*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid feature name: $ac_useropt" >&2
-   { (exit 1); exit 1; }; }
+      as_fn_error $? "invalid feature name: $ac_useropt"
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -882,8 +875,7 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*enable-\([^=]*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid feature name: $ac_useropt" >&2
-   { (exit 1); exit 1; }; }
+      as_fn_error $? "invalid feature name: $ac_useropt"
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1087,8 +1079,7 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*with-\([^=]*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid package name: $ac_useropt" >&2
-   { (exit 1); exit 1; }; }
+      as_fn_error $? "invalid package name: $ac_useropt"
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1104,8 +1095,7 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*without-\(.*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid package name: $ac_useropt" >&2
-   { (exit 1); exit 1; }; }
+      as_fn_error $? "invalid package name: $ac_useropt"
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1135,17 +1125,17 @@ do
   | --x-librar=* | --x-libra=* | --x-libr=* | --x-lib=* | --x-li=* | --x-l=*)
     x_libraries=$ac_optarg ;;
 
-  -*) { $as_echo "$as_me: error: unrecognized option: $ac_option
-Try \`$0 --help' for more information." >&2
-   { (exit 1); exit 1; }; }
+  -*) as_fn_error $? "unrecognized option: \`$ac_option'
+Try \`$0 --help' for more information"
     ;;
 
   *=*)
     ac_envvar=`expr "x$ac_option" : 'x\([^=]*\)='`
     # Reject names that are not valid shell variable names.
-    expr "x$ac_envvar" : ".*[^_$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid variable name: $ac_envvar" >&2
-   { (exit 1); exit 1; }; }
+    case $ac_envvar in #(
+      '' | [0-9]* | *[!_$as_cr_alnum]* )
+      as_fn_error $? "invalid variable name: \`$ac_envvar'" ;;
+    esac
     eval $ac_envvar=\$ac_optarg
     export $ac_envvar ;;
 
@@ -1154,7 +1144,7 @@ Try \`$0 --help' for more information." >&2
     $as_echo "$as_me: WARNING: you should use --build, --host, --target" >&2
     expr "x$ac_option" : ".*[^-._$as_cr_alnum]" >/dev/null &&
       $as_echo "$as_me: WARNING: invalid host type: $ac_option" >&2
-    : ${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}
+    : "${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}"
     ;;
 
   esac
@@ -1162,15 +1152,13 @@ done
 
 if test -n "$ac_prev"; then
   ac_option=--`echo $ac_prev | sed 's/_/-/g'`
-  { $as_echo "$as_me: error: missing argument to $ac_option" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "missing argument to $ac_option"
 fi
 
 if test -n "$ac_unrecognized_opts"; then
   case $enable_option_checking in
     no) ;;
-    fatal) { $as_echo "$as_me: error: unrecognized options: $ac_unrecognized_opts" >&2
-   { (exit 1); exit 1; }; } ;;
+    fatal) as_fn_error $? "unrecognized options: $ac_unrecognized_opts" ;;
     *)     $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2 ;;
   esac
 fi
@@ -1193,8 +1181,7 @@ do
     [\\/$]* | ?:[\\/]* )  continue;;
     NONE | '' ) case $ac_var in *prefix ) continue;; esac;;
   esac
-  { $as_echo "$as_me: error: expected an absolute directory name for --$ac_var: $ac_val" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "expected an absolute directory name for --$ac_var: $ac_val"
 done
 
 # There might be people who depend on the old broken behavior: `$host'
@@ -1208,8 +1195,6 @@ target=$target_alias
 if test "x$host_alias" != x; then
   if test "x$build_alias" = x; then
     cross_compiling=maybe
-    $as_echo "$as_me: WARNING: If you wanted to set the --build type, don't use --host.
-    If a cross compiler is detected then cross compile mode will be used." >&2
   elif test "x$build_alias" != "x$host_alias"; then
     cross_compiling=yes
   fi
@@ -1224,11 +1209,9 @@ test "$silent" = yes && exec 6>/dev/null
 ac_pwd=`pwd` && test -n "$ac_pwd" &&
 ac_ls_di=`ls -di .` &&
 ac_pwd_ls_di=`cd "$ac_pwd" && ls -di .` ||
-  { $as_echo "$as_me: error: working directory cannot be determined" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "working directory cannot be determined"
 test "X$ac_ls_di" = "X$ac_pwd_ls_di" ||
-  { $as_echo "$as_me: error: pwd does not report name of working directory" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "pwd does not report name of working directory"
 
 
 # Find the source files, if location was not specified.
@@ -1267,13 +1250,11 @@ else
 fi
 if test ! -r "$srcdir/$ac_unique_file"; then
   test "$ac_srcdir_defaulted" = yes && srcdir="$ac_confdir or .."
-  { $as_echo "$as_me: error: cannot find sources ($ac_unique_file) in $srcdir" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "cannot find sources ($ac_unique_file) in $srcdir"
 fi
 ac_msg="sources are in $srcdir, but \`cd $srcdir' does not work"
 ac_abs_confdir=`(
-	cd "$srcdir" && test -r "./$ac_unique_file" || { $as_echo "$as_me: error: $ac_msg" >&2
-   { (exit 1); exit 1; }; }
+	cd "$srcdir" && test -r "./$ac_unique_file" || as_fn_error $? "$ac_msg"
 	pwd)`
 # When building in place, set srcdir=.
 if test "$ac_abs_confdir" = "$ac_pwd"; then
@@ -1313,7 +1294,7 @@ Configuration:
       --help=short        display options specific to this package
       --help=recursive    display the short help of all the included packages
   -V, --version           display version information and exit
-  -q, --quiet, --silent   do not print \`checking...' messages
+  -q, --quiet, --silent   do not print \`checking ...' messages
       --cache-file=FILE   cache test results in FILE [disabled]
   -C, --config-cache      alias for \`--cache-file=config.cache'
   -n, --no-create         do not create output files
@@ -1373,11 +1354,17 @@ Optional Features:
   --disable-option-checking  ignore unrecognized --enable/--with options
   --disable-FEATURE       do not include FEATURE (same as --enable-FEATURE=no)
   --enable-FEATURE[=ARG]  include FEATURE [ARG=yes]
-  --disable-dependency-tracking  speeds up one-time build
-  --enable-dependency-tracking   do not reject slow dependency extractors
+  --enable-silent-rules   less verbose build output (undo: "make V=1")
+  --disable-silent-rules  verbose build output (undo: "make V=0")
+  --enable-dependency-tracking
+                          do not reject slow dependency extractors
+  --disable-dependency-tracking
+                          speeds up one-time build
   --disable-openmp        do not use OpenMP
   --enable-simd=SSE|AVX|AVX2|AVX512
                           Select instructions
+  --enable-comms=none|fake|mpi
+                          Select communications
 
 Some influential environment variables:
   CXX         C++ compiler command
@@ -1385,7 +1372,7 @@ Some influential environment variables:
   LDFLAGS     linker flags, e.g. -L<lib dir> if you have libraries in a
               nonstandard directory <lib dir>
   LIBS        libraries to pass to the linker, e.g. -l<library>
-  CPPFLAGS    C/C++/Objective C preprocessor flags, e.g. -I<include dir> if
+  CPPFLAGS    (Objective) C/C++ preprocessor flags, e.g. -I<include dir> if
               you have headers in a nonstandard directory <include dir>
   CC          C compiler command
   CFLAGS      C compiler flags
@@ -1458,21 +1445,522 @@ test -n "$ac_init_help" && exit $ac_status
 if $ac_init_version; then
   cat <<\_ACEOF
 Grid configure 1.0
-generated by GNU Autoconf 2.63
+generated by GNU Autoconf 2.69
 
-Copyright (C) 1992, 1993, 1994, 1995, 1996, 1998, 1999, 2000, 2001,
-2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
+Copyright (C) 2012 Free Software Foundation, Inc.
 This configure script is free software; the Free Software Foundation
 gives unlimited permission to copy, distribute and modify it.
 _ACEOF
   exit
 fi
+
+## ------------------------ ##
+## Autoconf initialization. ##
+## ------------------------ ##
+
+# ac_fn_cxx_try_compile LINENO
+# ----------------------------
+# Try to compile conftest.$ac_ext, and return whether this succeeded.
+ac_fn_cxx_try_compile ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  rm -f conftest.$ac_objext
+  if { { ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compile") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_retval=1
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_cxx_try_compile
+
+# ac_fn_c_try_compile LINENO
+# --------------------------
+# Try to compile conftest.$ac_ext, and return whether this succeeded.
+ac_fn_c_try_compile ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  rm -f conftest.$ac_objext
+  if { { ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compile") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_retval=1
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_compile
+
+# ac_fn_c_try_link LINENO
+# -----------------------
+# Try to link conftest.$ac_ext, and return whether this succeeded.
+ac_fn_c_try_link ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  rm -f conftest.$ac_objext conftest$ac_exeext
+  if { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 test -x conftest$ac_exeext
+       }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_retval=1
+fi
+  # Delete the IPA/IPO (Inter Procedural Analysis/Optimization) information
+  # created by the PGI compiler (conftest_ipa8_conftest.oo), as it would
+  # interfere with the next link command; also delete a directory that is
+  # left behind by Apple's compiler.  We do this before executing the actions.
+  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_link
+
+# ac_fn_c_try_cpp LINENO
+# ----------------------
+# Try to preprocess conftest.$ac_ext, and return whether this succeeded.
+ac_fn_c_try_cpp ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if { { ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } > conftest.i && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+    ac_retval=1
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_cpp
+
+# ac_fn_c_check_header_mongrel LINENO HEADER VAR INCLUDES
+# -------------------------------------------------------
+# Tests whether HEADER exists, giving a warning if it cannot be compiled using
+# the include files in INCLUDES and setting the cache variable VAR
+# accordingly.
+ac_fn_c_check_header_mongrel ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if eval \${$3+:} false; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 usability" >&5
+$as_echo_n "checking $2 usability... " >&6; }
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+#include <$2>
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_header_compiler=yes
+else
+  ac_header_compiler=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 presence" >&5
+$as_echo_n "checking $2 presence... " >&6; }
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <$2>
+_ACEOF
+if ac_fn_c_try_cpp "$LINENO"; then :
+  ac_header_preproc=yes
+else
+  ac_header_preproc=no
+fi
+rm -f conftest.err conftest.i conftest.$ac_ext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in #((
+  yes:no: )
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $2: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $2:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $2: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
+( $as_echo "## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##"
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  eval "$3=\$ac_header_compiler"
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_header_mongrel
+
+# ac_fn_c_try_run LINENO
+# ----------------------
+# Try to link conftest.$ac_ext, and return whether this succeeded. Assumes
+# that executables *can* be run.
+ac_fn_c_try_run ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && { ac_try='./conftest$ac_exeext'
+  { { case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: program exited with status $ac_status" >&5
+       $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+       ac_retval=$ac_status
+fi
+  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_run
+
+# ac_fn_c_check_header_compile LINENO HEADER VAR INCLUDES
+# -------------------------------------------------------
+# Tests whether HEADER exists and can be compiled using the include files in
+# INCLUDES, setting the cache variable VAR accordingly.
+ac_fn_c_check_header_compile ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+#include <$2>
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  eval "$3=yes"
+else
+  eval "$3=no"
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_header_compile
+
+# ac_fn_c_check_type LINENO TYPE VAR INCLUDES
+# -------------------------------------------
+# Tests whether TYPE exists after having included INCLUDES, setting cache
+# variable VAR accordingly.
+ac_fn_c_check_type ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  eval "$3=no"
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+if (sizeof ($2))
+	 return 0;
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+if (sizeof (($2)))
+	    return 0;
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+
+else
+  eval "$3=yes"
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_type
+
+# ac_fn_c_find_uintX_t LINENO BITS VAR
+# ------------------------------------
+# Finds an unsigned integer type with width BITS, setting cache variable VAR
+# accordingly.
+ac_fn_c_find_uintX_t ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for uint$2_t" >&5
+$as_echo_n "checking for uint$2_t... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  eval "$3=no"
+     # Order is important - never check a type that is potentially smaller
+     # than half of the expected target width.
+     for ac_type in uint$2_t 'unsigned int' 'unsigned long int' \
+	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
+       cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+static int test_array [1 - 2 * !((($ac_type) -1 >> ($2 / 2 - 1)) >> ($2 / 2 - 1) == 3)];
+test_array [0] = 0;
+return test_array [0];
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  case $ac_type in #(
+  uint$2_t) :
+    eval "$3=yes" ;; #(
+  *) :
+    eval "$3=\$ac_type" ;;
+esac
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+       if eval test \"x\$"$3"\" = x"no"; then :
+
+else
+  break
+fi
+     done
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_find_uintX_t
+
+# ac_fn_c_check_func LINENO FUNC VAR
+# ----------------------------------
+# Tests whether FUNC exists, setting the cache variable VAR accordingly
+ac_fn_c_check_func ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+/* Define $2 to an innocuous variant, in case <limits.h> declares $2.
+   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
+#define $2 innocuous_$2
+
+/* System header to define __stub macros and hopefully few prototypes,
+    which can conflict with char $2 (); below.
+    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
+    <limits.h> exists even on freestanding compilers.  */
+
+#ifdef __STDC__
+# include <limits.h>
+#else
+# include <assert.h>
+#endif
+
+#undef $2
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char $2 ();
+/* The GNU C library defines this for functions which it implements
+    to always fail with ENOSYS.  Some functions are actually named
+    something starting with __ and the normal name is an alias.  */
+#if defined __stub_$2 || defined __stub___$2
+choke me
+#endif
+
+int
+main ()
+{
+return $2 ();
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  eval "$3=yes"
+else
+  eval "$3=no"
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_func
 cat >config.log <<_ACEOF
 This file contains any messages produced by compilers while
 running configure, to aid debugging if configure makes a mistake.
 
 It was created by Grid $as_me 1.0, which was
-generated by GNU Autoconf 2.63.  Invocation command line was
+generated by GNU Autoconf 2.69.  Invocation command line was
 
   $ $0 $@
 
@@ -1508,8 +1996,8 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  $as_echo "PATH: $as_dir"
-done
+    $as_echo "PATH: $as_dir"
+  done
 IFS=$as_save_IFS
 
 } >&5
@@ -1546,9 +2034,9 @@ do
       ac_arg=`$as_echo "$ac_arg" | sed "s/'/'\\\\\\\\''/g"` ;;
     esac
     case $ac_pass in
-    1) ac_configure_args0="$ac_configure_args0 '$ac_arg'" ;;
+    1) as_fn_append ac_configure_args0 " '$ac_arg'" ;;
     2)
-      ac_configure_args1="$ac_configure_args1 '$ac_arg'"
+      as_fn_append ac_configure_args1 " '$ac_arg'"
       if test $ac_must_keep_next = true; then
 	ac_must_keep_next=false # Got value, back to normal.
       else
@@ -1564,13 +2052,13 @@ do
 	  -* ) ac_must_keep_next=true ;;
 	esac
       fi
-      ac_configure_args="$ac_configure_args '$ac_arg'"
+      as_fn_append ac_configure_args " '$ac_arg'"
       ;;
     esac
   done
 done
-$as_unset ac_configure_args0 || test "${ac_configure_args0+set}" != set || { ac_configure_args0=; export ac_configure_args0; }
-$as_unset ac_configure_args1 || test "${ac_configure_args1+set}" != set || { ac_configure_args1=; export ac_configure_args1; }
+{ ac_configure_args0=; unset ac_configure_args0;}
+{ ac_configure_args1=; unset ac_configure_args1;}
 
 # When interrupted or exit'd, cleanup temporary files, and complete
 # config.log.  We remove comments because anyway the quotes in there
@@ -1582,11 +2070,9 @@ trap 'exit_status=$?
   {
     echo
 
-    cat <<\_ASBOX
-## ---------------- ##
+    $as_echo "## ---------------- ##
 ## Cache variables. ##
-## ---------------- ##
-_ASBOX
+## ---------------- ##"
     echo
     # The following way of writing the cache mishandles newlines in values,
 (
@@ -1595,13 +2081,13 @@ _ASBOX
     case $ac_val in #(
     *${as_nl}*)
       case $ac_var in #(
-      *_cv_*) { $as_echo "$as_me:$LINENO: WARNING: cache variable $ac_var contains a newline" >&5
+      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
 $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
       esac
       case $ac_var in #(
       _ | IFS | as_nl) ;; #(
       BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
-      *) $as_unset $ac_var ;;
+      *) { eval $ac_var=; unset $ac_var;} ;;
       esac ;;
     esac
   done
@@ -1620,11 +2106,9 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
 )
     echo
 
-    cat <<\_ASBOX
-## ----------------- ##
+    $as_echo "## ----------------- ##
 ## Output variables. ##
-## ----------------- ##
-_ASBOX
+## ----------------- ##"
     echo
     for ac_var in $ac_subst_vars
     do
@@ -1637,11 +2121,9 @@ _ASBOX
     echo
 
     if test -n "$ac_subst_files"; then
-      cat <<\_ASBOX
-## ------------------- ##
+      $as_echo "## ------------------- ##
 ## File substitutions. ##
-## ------------------- ##
-_ASBOX
+## ------------------- ##"
       echo
       for ac_var in $ac_subst_files
       do
@@ -1655,11 +2137,9 @@ _ASBOX
     fi
 
     if test -s confdefs.h; then
-      cat <<\_ASBOX
-## ----------- ##
+      $as_echo "## ----------- ##
 ## confdefs.h. ##
-## ----------- ##
-_ASBOX
+## ----------- ##"
       echo
       cat confdefs.h
       echo
@@ -1673,46 +2153,53 @@ _ASBOX
     exit $exit_status
 ' 0
 for ac_signal in 1 2 13 15; do
-  trap 'ac_signal='$ac_signal'; { (exit 1); exit 1; }' $ac_signal
+  trap 'ac_signal='$ac_signal'; as_fn_exit 1' $ac_signal
 done
 ac_signal=0
 
 # confdefs.h avoids OS command line length limits that DEFS can exceed.
 rm -f -r conftest* confdefs.h
 
+$as_echo "/* confdefs.h */" > confdefs.h
+
 # Predefined preprocessor variables.
 
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_NAME "$PACKAGE_NAME"
 _ACEOF
 
-
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_TARNAME "$PACKAGE_TARNAME"
 _ACEOF
 
-
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_VERSION "$PACKAGE_VERSION"
 _ACEOF
 
-
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_STRING "$PACKAGE_STRING"
 _ACEOF
 
-
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_BUGREPORT "$PACKAGE_BUGREPORT"
 _ACEOF
 
+cat >>confdefs.h <<_ACEOF
+#define PACKAGE_URL "$PACKAGE_URL"
+_ACEOF
+
 
 # Let the site file select an alternate cache file if it wants to.
 # Prefer an explicitly selected file to automatically selected ones.
 ac_site_file1=NONE
 ac_site_file2=NONE
 if test -n "$CONFIG_SITE"; then
-  ac_site_file1=$CONFIG_SITE
+  # We do not want a PATH search for config.site.
+  case $CONFIG_SITE in #((
+    -*)  ac_site_file1=./$CONFIG_SITE;;
+    */*) ac_site_file1=$CONFIG_SITE;;
+    *)   ac_site_file1=./$CONFIG_SITE;;
+  esac
 elif test "x$prefix" != xNONE; then
   ac_site_file1=$prefix/share/config.site
   ac_site_file2=$prefix/etc/config.site
@@ -1723,19 +2210,23 @@ fi
 for ac_site_file in "$ac_site_file1" "$ac_site_file2"
 do
   test "x$ac_site_file" = xNONE && continue
-  if test -r "$ac_site_file"; then
-    { $as_echo "$as_me:$LINENO: loading site script $ac_site_file" >&5
+  if test /dev/null != "$ac_site_file" && test -r "$ac_site_file"; then
+    { $as_echo "$as_me:${as_lineno-$LINENO}: loading site script $ac_site_file" >&5
 $as_echo "$as_me: loading site script $ac_site_file" >&6;}
     sed 's/^/| /' "$ac_site_file" >&5
-    . "$ac_site_file"
+    . "$ac_site_file" \
+      || { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "failed to load site script $ac_site_file
+See \`config.log' for more details" "$LINENO" 5; }
   fi
 done
 
 if test -r "$cache_file"; then
-  # Some versions of bash will fail to source /dev/null (special
-  # files actually), so we avoid doing that.
-  if test -f "$cache_file"; then
-    { $as_echo "$as_me:$LINENO: loading cache $cache_file" >&5
+  # Some versions of bash will fail to source /dev/null (special files
+  # actually), so we avoid doing that.  DJGPP emulates it as a regular file.
+  if test /dev/null != "$cache_file" && test -f "$cache_file"; then
+    { $as_echo "$as_me:${as_lineno-$LINENO}: loading cache $cache_file" >&5
 $as_echo "$as_me: loading cache $cache_file" >&6;}
     case $cache_file in
       [\\/]* | ?:[\\/]* ) . "$cache_file";;
@@ -1743,7 +2234,7 @@ $as_echo "$as_me: loading cache $cache_file" >&6;}
     esac
   fi
 else
-  { $as_echo "$as_me:$LINENO: creating cache $cache_file" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: creating cache $cache_file" >&5
 $as_echo "$as_me: creating cache $cache_file" >&6;}
   >$cache_file
 fi
@@ -1758,11 +2249,11 @@ for ac_var in $ac_precious_vars; do
   eval ac_new_val=\$ac_env_${ac_var}_value
   case $ac_old_set,$ac_new_set in
     set,)
-      { $as_echo "$as_me:$LINENO: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
+      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
 $as_echo "$as_me: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&2;}
       ac_cache_corrupted=: ;;
     ,set)
-      { $as_echo "$as_me:$LINENO: error: \`$ac_var' was not set in the previous run" >&5
+      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was not set in the previous run" >&5
 $as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
       ac_cache_corrupted=: ;;
     ,);;
@@ -1772,17 +2263,17 @@ $as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
 	ac_old_val_w=`echo x $ac_old_val`
 	ac_new_val_w=`echo x $ac_new_val`
 	if test "$ac_old_val_w" != "$ac_new_val_w"; then
-	  { $as_echo "$as_me:$LINENO: error: \`$ac_var' has changed since the previous run:" >&5
+	  { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' has changed since the previous run:" >&5
 $as_echo "$as_me: error: \`$ac_var' has changed since the previous run:" >&2;}
 	  ac_cache_corrupted=:
 	else
-	  { $as_echo "$as_me:$LINENO: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
+	  { $as_echo "$as_me:${as_lineno-$LINENO}: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
 $as_echo "$as_me: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&2;}
 	  eval $ac_var=\$ac_old_val
 	fi
-	{ $as_echo "$as_me:$LINENO:   former value:  \`$ac_old_val'" >&5
+	{ $as_echo "$as_me:${as_lineno-$LINENO}:   former value:  \`$ac_old_val'" >&5
 $as_echo "$as_me:   former value:  \`$ac_old_val'" >&2;}
-	{ $as_echo "$as_me:$LINENO:   current value: \`$ac_new_val'" >&5
+	{ $as_echo "$as_me:${as_lineno-$LINENO}:   current value: \`$ac_new_val'" >&5
 $as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
       fi;;
   esac
@@ -1794,43 +2285,20 @@ $as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
     esac
     case " $ac_configure_args " in
       *" '$ac_arg' "*) ;; # Avoid dups.  Use of quotes ensures accuracy.
-      *) ac_configure_args="$ac_configure_args '$ac_arg'" ;;
+      *) as_fn_append ac_configure_args " '$ac_arg'" ;;
     esac
   fi
 done
 if $ac_cache_corrupted; then
-  { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-  { $as_echo "$as_me:$LINENO: error: changes in the environment can compromise the build" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: error: changes in the environment can compromise the build" >&5
 $as_echo "$as_me: error: changes in the environment can compromise the build" >&2;}
-  { { $as_echo "$as_me:$LINENO: error: run \`make distclean' and/or \`rm $cache_file' and start over" >&5
-$as_echo "$as_me: error: run \`make distclean' and/or \`rm $cache_file' and start over" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "run \`make distclean' and/or \`rm $cache_file' and start over" "$LINENO" 5
 fi
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+## -------------------- ##
+## Main body of script. ##
+## -------------------- ##
 
 ac_ext=c
 ac_cpp='$CPP $CPPFLAGS'
@@ -1839,7 +2307,7 @@ ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
 
-am__api_version='1.11'
+am__api_version='1.15'
 
 ac_aux_dir=
 for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
@@ -1858,9 +2326,7 @@ for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
   fi
 done
 if test -z "$ac_aux_dir"; then
-  { { $as_echo "$as_me:$LINENO: error: cannot find install-sh or install.sh in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" >&5
-$as_echo "$as_me: error: cannot find install-sh or install.sh in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "cannot find install-sh, install.sh, or shtool in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" "$LINENO" 5
 fi
 
 # These three variables are undocumented and unsupported,
@@ -1886,10 +2352,10 @@ ac_configure="$SHELL $ac_aux_dir/configure"  # Please don't use this var.
 # OS/2's system install, which has a completely different semantic
 # ./install, which can be erroneously created by make from ./install.sh.
 # Reject install programs that cannot install multiple files.
-{ $as_echo "$as_me:$LINENO: checking for a BSD-compatible install" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a BSD-compatible install" >&5
 $as_echo_n "checking for a BSD-compatible install... " >&6; }
 if test -z "$INSTALL"; then
-if test "${ac_cv_path_install+set}" = set; then
+if ${ac_cv_path_install+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
@@ -1897,11 +2363,11 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  # Account for people who put trailing slashes in PATH elements.
-case $as_dir/ in
-  ./ | .// | /cC/* | \
+    # Account for people who put trailing slashes in PATH elements.
+case $as_dir/ in #((
+  ./ | .// | /[cC]/* | \
   /etc/* | /usr/sbin/* | /usr/etc/* | /sbin/* | /usr/afsws/bin/* | \
-  ?:\\/os2\\/install\\/* | ?:\\/OS2\\/INSTALL\\/* | \
+  ?:[\\/]os2[\\/]install[\\/]* | ?:[\\/]OS2[\\/]INSTALL[\\/]* | \
   /usr/ucb/* ) ;;
   *)
     # OSF1 and SCO ODT 3.0 have their own names for install.
@@ -1909,7 +2375,7 @@ case $as_dir/ in
     # by default.
     for ac_prog in ginstall scoinst install; do
       for ac_exec_ext in '' $ac_executable_extensions; do
-	if { test -f "$as_dir/$ac_prog$ac_exec_ext" && $as_test_x "$as_dir/$ac_prog$ac_exec_ext"; }; then
+	if as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext"; then
 	  if test $ac_prog = install &&
 	    grep dspmsg "$as_dir/$ac_prog$ac_exec_ext" >/dev/null 2>&1; then
 	    # AIX install.  It has an incompatible calling convention.
@@ -1938,7 +2404,7 @@ case $as_dir/ in
     ;;
 esac
 
-done
+  done
 IFS=$as_save_IFS
 
 rm -rf conftest.one conftest.two conftest.dir
@@ -1954,7 +2420,7 @@ fi
     INSTALL=$ac_install_sh
   fi
 fi
-{ $as_echo "$as_me:$LINENO: result: $INSTALL" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $INSTALL" >&5
 $as_echo "$INSTALL" >&6; }
 
 # Use test -z because SunOS4 sh mishandles braces in ${var-val}.
@@ -1965,68 +2431,73 @@ test -z "$INSTALL_SCRIPT" && INSTALL_SCRIPT='${INSTALL}'
 
 test -z "$INSTALL_DATA" && INSTALL_DATA='${INSTALL} -m 644'
 
-{ $as_echo "$as_me:$LINENO: checking whether build environment is sane" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether build environment is sane" >&5
 $as_echo_n "checking whether build environment is sane... " >&6; }
-# Just in case
-sleep 1
-echo timestamp > conftest.file
 # Reject unsafe characters in $srcdir or the absolute working directory
 # name.  Accept space and tab only in the latter.
 am_lf='
 '
 case `pwd` in
   *[\\\"\#\$\&\'\`$am_lf]*)
-    { { $as_echo "$as_me:$LINENO: error: unsafe absolute working directory name" >&5
-$as_echo "$as_me: error: unsafe absolute working directory name" >&2;}
-   { (exit 1); exit 1; }; };;
+    as_fn_error $? "unsafe absolute working directory name" "$LINENO" 5;;
 esac
 case $srcdir in
   *[\\\"\#\$\&\'\`$am_lf\ \	]*)
-    { { $as_echo "$as_me:$LINENO: error: unsafe srcdir value: \`$srcdir'" >&5
-$as_echo "$as_me: error: unsafe srcdir value: \`$srcdir'" >&2;}
-   { (exit 1); exit 1; }; };;
+    as_fn_error $? "unsafe srcdir value: '$srcdir'" "$LINENO" 5;;
 esac
 
-# Do `set' in a subshell so we don't clobber the current shell's
+# Do 'set' in a subshell so we don't clobber the current shell's
 # arguments.  Must try -L first in case configure is actually a
 # symlink; some systems play weird games with the mod time of symlinks
 # (eg FreeBSD returns the mod time of the symlink's containing
 # directory).
 if (
-   set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
-   if test "$*" = "X"; then
-      # -L didn't work.
-      set X `ls -t "$srcdir/configure" conftest.file`
-   fi
-   rm -f conftest.file
-   if test "$*" != "X $srcdir/configure conftest.file" \
-      && test "$*" != "X conftest.file $srcdir/configure"; then
-
-      # If neither matched, then we have a broken ls.  This can happen
-      # if, for instance, CONFIG_SHELL is bash and it inherits a
-      # broken ls alias from the environment.  This has actually
-      # happened.  Such a system could not be considered "sane".
-      { { $as_echo "$as_me:$LINENO: error: ls -t appears to fail.  Make sure there is not a broken
-alias in your environment" >&5
-$as_echo "$as_me: error: ls -t appears to fail.  Make sure there is not a broken
-alias in your environment" >&2;}
-   { (exit 1); exit 1; }; }
-   fi
+   am_has_slept=no
+   for am_try in 1 2; do
+     echo "timestamp, slept: $am_has_slept" > conftest.file
+     set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
+     if test "$*" = "X"; then
+	# -L didn't work.
+	set X `ls -t "$srcdir/configure" conftest.file`
+     fi
+     if test "$*" != "X $srcdir/configure conftest.file" \
+	&& test "$*" != "X conftest.file $srcdir/configure"; then
 
+	# If neither matched, then we have a broken ls.  This can happen
+	# if, for instance, CONFIG_SHELL is bash and it inherits a
+	# broken ls alias from the environment.  This has actually
+	# happened.  Such a system could not be considered "sane".
+	as_fn_error $? "ls -t appears to fail.  Make sure there is not a broken
+  alias in your environment" "$LINENO" 5
+     fi
+     if test "$2" = conftest.file || test $am_try -eq 2; then
+       break
+     fi
+     # Just in case.
+     sleep 1
+     am_has_slept=yes
+   done
    test "$2" = conftest.file
    )
 then
    # Ok.
    :
 else
-   { { $as_echo "$as_me:$LINENO: error: newly created file is older than distributed files!
-Check your system clock" >&5
-$as_echo "$as_me: error: newly created file is older than distributed files!
-Check your system clock" >&2;}
-   { (exit 1); exit 1; }; }
+   as_fn_error $? "newly created file is older than distributed files!
+Check your system clock" "$LINENO" 5
 fi
-{ $as_echo "$as_me:$LINENO: result: yes" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
 $as_echo "yes" >&6; }
+# If we didn't sleep, we still need to ensure time stamps of config.status and
+# generated files are strictly newer.
+am_sleep_pid=
+if grep 'slept: no' conftest.file >/dev/null 2>&1; then
+  ( sleep 1 ) &
+  am_sleep_pid=$!
+fi
+
+rm -f conftest.file
+
 test "$program_prefix" != NONE &&
   program_transform_name="s&^&$program_prefix&;$program_transform_name"
 # Use a double $ so make ignores it.
@@ -2037,8 +2508,8 @@ test "$program_suffix" != NONE &&
 ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
 program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
 
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 
 if test x"${MISSING+set}" != xset; then
   case $am_aux_dir in
@@ -2049,15 +2520,15 @@ if test x"${MISSING+set}" != xset; then
   esac
 fi
 # Use eval to expand $SHELL
-if eval "$MISSING --run true"; then
-  am_missing_run="$MISSING --run "
+if eval "$MISSING --is-lightweight"; then
+  am_missing_run="$MISSING "
 else
   am_missing_run=
-  { $as_echo "$as_me:$LINENO: WARNING: \`missing' script is too old or missing" >&5
-$as_echo "$as_me: WARNING: \`missing' script is too old or missing" >&2;}
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: 'missing' script is too old or missing" >&5
+$as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
 fi
 
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -2066,17 +2537,17 @@ if test x"${install_sh}" != xset; then
   esac
 fi
 
-# Installed binaries are usually stripped using `strip' when the user
-# run `make install-strip'.  However `strip' might not be the right
+# Installed binaries are usually stripped using 'strip' when the user
+# run "make install-strip".  However 'strip' might not be the right
 # tool to use in cross-compilation environments, therefore Automake
-# will honor the `STRIP' environment variable to overrule this program.
+# will honor the 'STRIP' environment variable to overrule this program.
 if test "$cross_compiling" != no; then
   if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}strip", so it can be a program name with args.
 set dummy ${ac_tool_prefix}strip; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_STRIP+set}" = set; then
+if ${ac_cv_prog_STRIP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$STRIP"; then
@@ -2087,24 +2558,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_STRIP="${ac_tool_prefix}strip"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 STRIP=$ac_cv_prog_STRIP
 if test -n "$STRIP"; then
-  { $as_echo "$as_me:$LINENO: result: $STRIP" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $STRIP" >&5
 $as_echo "$STRIP" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2114,9 +2585,9 @@ if test -z "$ac_cv_prog_STRIP"; then
   ac_ct_STRIP=$STRIP
   # Extract the first word of "strip", so it can be a program name with args.
 set dummy strip; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_ac_ct_STRIP+set}" = set; then
+if ${ac_cv_prog_ac_ct_STRIP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_STRIP"; then
@@ -2127,24 +2598,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_ac_ct_STRIP="strip"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_STRIP=$ac_cv_prog_ac_ct_STRIP
 if test -n "$ac_ct_STRIP"; then
-  { $as_echo "$as_me:$LINENO: result: $ac_ct_STRIP" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_STRIP" >&5
 $as_echo "$ac_ct_STRIP" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2153,7 +2624,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -2166,10 +2637,10 @@ fi
 fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 
-{ $as_echo "$as_me:$LINENO: checking for a thread-safe mkdir -p" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a thread-safe mkdir -p" >&5
 $as_echo_n "checking for a thread-safe mkdir -p... " >&6; }
 if test -z "$MKDIR_P"; then
-  if test "${ac_cv_path_mkdir+set}" = set; then
+  if ${ac_cv_path_mkdir+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
@@ -2177,9 +2648,9 @@ for as_dir in $PATH$PATH_SEPARATOR/opt/sfw/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_prog in mkdir gmkdir; do
+    for ac_prog in mkdir gmkdir; do
 	 for ac_exec_ext in '' $ac_executable_extensions; do
-	   { test -f "$as_dir/$ac_prog$ac_exec_ext" && $as_test_x "$as_dir/$ac_prog$ac_exec_ext"; } || continue
+	   as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext" || continue
 	   case `"$as_dir/$ac_prog$ac_exec_ext" --version 2>&1` in #(
 	     'mkdir (GNU coreutils) '* | \
 	     'mkdir (coreutils) '* | \
@@ -2189,11 +2660,12 @@ do
 	   esac
 	 done
        done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 
+  test -d ./--version && rmdir ./--version
   if test "${ac_cv_path_mkdir+set}" = set; then
     MKDIR_P="$ac_cv_path_mkdir -p"
   else
@@ -2201,26 +2673,19 @@ fi
     # value for MKDIR_P within a source directory, because that will
     # break other packages using the cache if that directory is
     # removed, or if the value is a relative name.
-    test -d ./--version && rmdir ./--version
     MKDIR_P="$ac_install_sh -d"
   fi
 fi
-{ $as_echo "$as_me:$LINENO: result: $MKDIR_P" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $MKDIR_P" >&5
 $as_echo "$MKDIR_P" >&6; }
 
-mkdir_p="$MKDIR_P"
-case $mkdir_p in
-  [\\/$]* | ?:[\\/]*) ;;
-  */*) mkdir_p="\$(top_builddir)/$mkdir_p" ;;
-esac
-
 for ac_prog in gawk mawk nawk awk
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_AWK+set}" = set; then
+if ${ac_cv_prog_AWK+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$AWK"; then
@@ -2231,24 +2696,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_AWK="$ac_prog"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 AWK=$ac_cv_prog_AWK
 if test -n "$AWK"; then
-  { $as_echo "$as_me:$LINENO: result: $AWK" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $AWK" >&5
 $as_echo "$AWK" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2256,11 +2721,11 @@ fi
   test -n "$AWK" && break
 done
 
-{ $as_echo "$as_me:$LINENO: checking whether ${MAKE-make} sets \$(MAKE)" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether ${MAKE-make} sets \$(MAKE)" >&5
 $as_echo_n "checking whether ${MAKE-make} sets \$(MAKE)... " >&6; }
 set x ${MAKE-make}
 ac_make=`$as_echo "$2" | sed 's/+/p/g; s/[^a-zA-Z0-9_]/_/g'`
-if { as_var=ac_cv_prog_make_${ac_make}_set; eval "test \"\${$as_var+set}\" = set"; }; then
+if eval \${ac_cv_prog_make_${ac_make}_set+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   cat >conftest.make <<\_ACEOF
@@ -2268,7 +2733,7 @@ SHELL = /bin/sh
 all:
 	@echo '@@@%%%=$(MAKE)=@@@%%%'
 _ACEOF
-# GNU make sometimes prints "make[1]: Entering...", which would confuse us.
+# GNU make sometimes prints "make[1]: Entering ...", which would confuse us.
 case `${MAKE-make} -f conftest.make 2>/dev/null` in
   *@@@%%%=?*=@@@%%%*)
     eval ac_cv_prog_make_${ac_make}_set=yes;;
@@ -2278,11 +2743,11 @@ esac
 rm -f conftest.make
 fi
 if eval test \$ac_cv_prog_make_${ac_make}_set = yes; then
-  { $as_echo "$as_me:$LINENO: result: yes" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
 $as_echo "yes" >&6; }
   SET_MAKE=
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
   SET_MAKE="MAKE=${MAKE-make}"
 fi
@@ -2296,15 +2761,52 @@ else
 fi
 rmdir .tst 2>/dev/null
 
+# Check whether --enable-silent-rules was given.
+if test "${enable_silent_rules+set}" = set; then :
+  enableval=$enable_silent_rules;
+fi
+
+case $enable_silent_rules in # (((
+  yes) AM_DEFAULT_VERBOSITY=0;;
+   no) AM_DEFAULT_VERBOSITY=1;;
+    *) AM_DEFAULT_VERBOSITY=1;;
+esac
+am_make=${MAKE-make}
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $am_make supports nested variables" >&5
+$as_echo_n "checking whether $am_make supports nested variables... " >&6; }
+if ${am_cv_make_support_nested_variables+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if $as_echo 'TRUE=$(BAR$(V))
+BAR0=false
+BAR1=true
+V=1
+am__doit:
+	@$(TRUE)
+.PHONY: am__doit' | $am_make -f - >/dev/null 2>&1; then
+  am_cv_make_support_nested_variables=yes
+else
+  am_cv_make_support_nested_variables=no
+fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_make_support_nested_variables" >&5
+$as_echo "$am_cv_make_support_nested_variables" >&6; }
+if test $am_cv_make_support_nested_variables = yes; then
+    AM_V='$(V)'
+  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
+else
+  AM_V=$AM_DEFAULT_VERBOSITY
+  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
+fi
+AM_BACKSLASH='\'
+
 if test "`cd $srcdir && pwd`" != "`pwd`"; then
   # Use -I$(srcdir) only when $(srcdir) != ., so that make's output
   # is not polluted with repeated "-I."
   am__isrc=' -I$(srcdir)'
   # test to see if srcdir already configured
   if test -f $srcdir/config.status; then
-    { { $as_echo "$as_me:$LINENO: error: source directory already configured; run \"make distclean\" there first" >&5
-$as_echo "$as_me: error: source directory already configured; run \"make distclean\" there first" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "source directory already configured; run \"make distclean\" there first" "$LINENO" 5
   fi
 fi
 
@@ -2348,19 +2850,72 @@ AUTOHEADER=${AUTOHEADER-"${am_missing_run}autoheader"}
 
 MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
-# Always define AMTAR for backward compatibility.
+# For better backward compatibility.  To be removed once Automake 1.9.x
+# dies out for good.  For more background, see:
+# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
+# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
+mkdir_p='$(MKDIR_P)'
 
-AMTAR=${AMTAR-"${am_missing_run}tar"}
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
+# Always define AMTAR for backward compatibility.  Yes, it's still used
+# in the wild :-(  We should find a proper way to deprecate it ...
+AMTAR='$${TAR-tar}'
 
-am__tar='${AMTAR} chof - "$$tardir"'; am__untar='${AMTAR} xf -'
+
+# We'll loop over all known methods to create a tar archive until one works.
+_am_tools='gnutar  pax cpio none'
+
+am__tar='$${TAR-tar} chof - "$$tardir"' am__untar='$${TAR-tar} xf -'
 
 
 
 
 
 
+# POSIX will say in a future version that running "rm -f" with no argument
+# is OK; and we want to be able to make that assumption in our Makefile
+# recipes.  So use an aggressive probe to check that the usage we want is
+# actually supported "in the wild" to an acceptable degree.
+# See automake bug#10828.
+# To make any issue more visible, cause the running configure to be aborted
+# by default if the 'rm' program in use doesn't match our expectations; the
+# user can still override this though.
+if rm -f && rm -fr && rm -rf; then : OK; else
+  cat >&2 <<'END'
+Oops!
+
+Your 'rm' program seems unable to run without file operands specified
+on the command line, even when the '-f' option is present.  This is contrary
+to the behaviour of most rm programs out there, and not conforming with
+the upcoming POSIX standard: <http://austingroupbugs.net/view.php?id=542>
+
+Please tell bug-automake@gnu.org about your system, including the value
+of your $PATH and any error possibly output before this message.  This
+can help us improve future automake versions.
+
+END
+  if test x"$ACCEPT_INFERIOR_RM_PROGRAM" = x"yes"; then
+    echo 'Configuration will proceed anyway, since you have set the' >&2
+    echo 'ACCEPT_INFERIOR_RM_PROGRAM variable to "yes"' >&2
+    echo >&2
+  else
+    cat >&2 <<'END'
+Aborting the configuration process, to ensure you take notice of the issue.
+
+You can download and install GNU coreutils to get an 'rm' implementation
+that behaves properly: <http://www.gnu.org/software/coreutils/>.
+
+If you want to complete the configuration process using your problematic
+'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
+to "yes", and re-run configure.
+
+END
+    as_fn_error $? "Your 'rm' program is bad, sorry." "$LINENO" 5
+  fi
+fi
+
+
 
 ac_config_headers="$ac_config_headers Grid_config.h"
 
@@ -2380,9 +2935,9 @@ if test -z "$CXX"; then
   do
     # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
 set dummy $ac_tool_prefix$ac_prog; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_CXX+set}" = set; then
+if ${ac_cv_prog_CXX+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CXX"; then
@@ -2393,24 +2948,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_CXX="$ac_tool_prefix$ac_prog"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 CXX=$ac_cv_prog_CXX
 if test -n "$CXX"; then
-  { $as_echo "$as_me:$LINENO: result: $CXX" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CXX" >&5
 $as_echo "$CXX" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2424,9 +2979,9 @@ if test -z "$CXX"; then
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_ac_ct_CXX+set}" = set; then
+if ${ac_cv_prog_ac_ct_CXX+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CXX"; then
@@ -2437,24 +2992,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_ac_ct_CXX="$ac_prog"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CXX=$ac_cv_prog_ac_ct_CXX
 if test -n "$ac_ct_CXX"; then
-  { $as_echo "$as_me:$LINENO: result: $ac_ct_CXX" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CXX" >&5
 $as_echo "$ac_ct_CXX" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2467,7 +3022,7 @@ done
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -2478,48 +3033,31 @@ fi
   fi
 fi
 # Provide some information about the compiler.
-$as_echo "$as_me:$LINENO: checking for C++ compiler version" >&5
+$as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler version" >&5
 set X $ac_compile
 ac_compiler=$2
-{ (ac_try="$ac_compiler --version >&5"
+for ac_option in --version -v -V -qversion; do
+  { { ac_try="$ac_compiler $ac_option >&5"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler --version >&5") 2>&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
-{ (ac_try="$ac_compiler -v >&5"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler -v >&5") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
-{ (ac_try="$ac_compiler -V >&5"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler -V >&5") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
+  if test -s conftest.err; then
+    sed '10a\
+... rest of stderr output deleted ...
+         10q' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+  fi
+  rm -f conftest.er1 conftest.err
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }
+done
 
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2535,8 +3073,8 @@ ac_clean_files="$ac_clean_files a.out a.out.dSYM a.exe b.out"
 # Try to create an executable without -o first, disregard a.out.
 # It will help us diagnose broken compilers, and finding out an intuition
 # of exeext.
-{ $as_echo "$as_me:$LINENO: checking for C++ compiler default output file name" >&5
-$as_echo_n "checking for C++ compiler default output file name... " >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether the C++ compiler works" >&5
+$as_echo_n "checking whether the C++ compiler works... " >&6; }
 ac_link_default=`$as_echo "$ac_link" | sed 's/ -o *conftest[^ ]*//'`
 
 # The possible output files:
@@ -2552,17 +3090,17 @@ do
 done
 rm -f $ac_rmfiles
 
-if { (ac_try="$ac_link_default"
+if { { ac_try="$ac_link_default"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
   (eval "$ac_link_default") 2>&5
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; then
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then :
   # Autoconf-2.13 could set the ac_cv_exeext variable to `no'.
 # So ignore a value of `no', otherwise this would lead to `EXEEXT = no'
 # in a Makefile.  We should not override ac_cv_exeext if it was cached,
@@ -2579,7 +3117,7 @@ do
 	# certainly right.
 	break;;
     *.* )
-        if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
+	if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
 	then :; else
 	   ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
 	fi
@@ -2598,84 +3136,41 @@ test "$ac_cv_exeext" = no && ac_cv_exeext=
 else
   ac_file=''
 fi
-
-{ $as_echo "$as_me:$LINENO: result: $ac_file" >&5
-$as_echo "$ac_file" >&6; }
-if test -z "$ac_file"; then
-  $as_echo "$as_me: failed program was:" >&5
+if test -z "$ac_file"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+$as_echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
 
-{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: C++ compiler cannot create executables
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: C++ compiler cannot create executables
-See \`config.log' for more details." >&2;}
-   { (exit 77); exit 77; }; }; }
-fi
-
-ac_exeext=$ac_cv_exeext
-
-# Check that the compiler produces executables we can run.  If not, either
-# the compiler is broken, or we cross compile.
-{ $as_echo "$as_me:$LINENO: checking whether the C++ compiler works" >&5
-$as_echo_n "checking whether the C++ compiler works... " >&6; }
-# FIXME: These cross compiler hacks should be removed for Autoconf 3.0
-# If not cross compiling, check that we can run a simple program.
-if test "$cross_compiling" != yes; then
-  if { ac_try='./$ac_file'
-  { (case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; }; then
-    cross_compiling=no
-  else
-    if test "$cross_compiling" = maybe; then
-	cross_compiling=yes
-    else
-	{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: cannot run C++ compiled programs.
-If you meant to cross compile, use \`--host'.
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: cannot run C++ compiled programs.
-If you meant to cross compile, use \`--host'.
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
-    fi
-  fi
-fi
-{ $as_echo "$as_me:$LINENO: result: yes" >&5
+as_fn_error 77 "C++ compiler cannot create executables
+See \`config.log' for more details" "$LINENO" 5; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
 $as_echo "yes" >&6; }
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler default output file name" >&5
+$as_echo_n "checking for C++ compiler default output file name... " >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_file" >&5
+$as_echo "$ac_file" >&6; }
+ac_exeext=$ac_cv_exeext
 
 rm -f -r a.out a.out.dSYM a.exe conftest$ac_cv_exeext b.out
 ac_clean_files=$ac_clean_files_save
-# Check that the compiler produces executables we can run.  If not, either
-# the compiler is broken, or we cross compile.
-{ $as_echo "$as_me:$LINENO: checking whether we are cross compiling" >&5
-$as_echo_n "checking whether we are cross compiling... " >&6; }
-{ $as_echo "$as_me:$LINENO: result: $cross_compiling" >&5
-$as_echo "$cross_compiling" >&6; }
-
-{ $as_echo "$as_me:$LINENO: checking for suffix of executables" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of executables" >&5
 $as_echo_n "checking for suffix of executables... " >&6; }
-if { (ac_try="$ac_link"
+if { { ac_try="$ac_link"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
   (eval "$ac_link") 2>&5
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; then
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then :
   # If both `conftest.exe' and `conftest' are `present' (well, observable)
 # catch `conftest.exe'.  For instance with Cygwin, `ls conftest' will
 # work properly (i.e., refer to `conftest.exe'), while it won't with
@@ -2690,32 +3185,83 @@ for ac_file in conftest.exe conftest conftest.*; do
   esac
 done
 else
-  { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: cannot compute suffix of executables: cannot compile and link
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: cannot compute suffix of executables: cannot compile and link
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
+as_fn_error $? "cannot compute suffix of executables: cannot compile and link
+See \`config.log' for more details" "$LINENO" 5; }
 fi
-
-rm -f conftest$ac_cv_exeext
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_exeext" >&5
+rm -f conftest conftest$ac_cv_exeext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_exeext" >&5
 $as_echo "$ac_cv_exeext" >&6; }
 
 rm -f conftest.$ac_ext
 EXEEXT=$ac_cv_exeext
 ac_exeext=$EXEEXT
-{ $as_echo "$as_me:$LINENO: checking for suffix of object files" >&5
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <stdio.h>
+int
+main ()
+{
+FILE *f = fopen ("conftest.out", "w");
+ return ferror (f) || fclose (f) != 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+ac_clean_files="$ac_clean_files conftest.out"
+# Check that the compiler produces executables we can run.  If not, either
+# the compiler is broken, or we cross compile.
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are cross compiling" >&5
+$as_echo_n "checking whether we are cross compiling... " >&6; }
+if test "$cross_compiling" != yes; then
+  { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }
+  if { ac_try='./conftest$ac_cv_exeext'
+  { { case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; }; then
+    cross_compiling=no
+  else
+    if test "$cross_compiling" = maybe; then
+	cross_compiling=yes
+    else
+	{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "cannot run C++ compiled programs.
+If you meant to cross compile, use \`--host'.
+See \`config.log' for more details" "$LINENO" 5; }
+    fi
+  fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $cross_compiling" >&5
+$as_echo "$cross_compiling" >&6; }
+
+rm -f conftest.$ac_ext conftest$ac_cv_exeext conftest.out
+ac_clean_files=$ac_clean_files_save
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of object files" >&5
 $as_echo_n "checking for suffix of object files... " >&6; }
-if test "${ac_cv_objext+set}" = set; then
+if ${ac_cv_objext+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2727,17 +3273,17 @@ main ()
 }
 _ACEOF
 rm -f conftest.o conftest.obj
-if { (ac_try="$ac_compile"
+if { { ac_try="$ac_compile"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
   (eval "$ac_compile") 2>&5
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; then
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then :
   for ac_file in conftest.o conftest.obj conftest.*; do
   test -f "$ac_file" || continue;
   case $ac_file in
@@ -2750,31 +3296,23 @@ else
   $as_echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
 
-{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: cannot compute suffix of object files: cannot compile
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: cannot compute suffix of object files: cannot compile
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
+as_fn_error $? "cannot compute suffix of object files: cannot compile
+See \`config.log' for more details" "$LINENO" 5; }
 fi
-
 rm -f conftest.$ac_cv_objext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_objext" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_objext" >&5
 $as_echo "$ac_cv_objext" >&6; }
 OBJEXT=$ac_cv_objext
 ac_objext=$OBJEXT
-{ $as_echo "$as_me:$LINENO: checking whether we are using the GNU C++ compiler" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C++ compiler" >&5
 $as_echo_n "checking whether we are using the GNU C++ compiler... " >&6; }
-if test "${ac_cv_cxx_compiler_gnu+set}" = set; then
+if ${ac_cv_cxx_compiler_gnu+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2788,37 +3326,16 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_cxx_try_compile "$LINENO"; then :
   ac_compiler_gnu=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_compiler_gnu=no
+  ac_compiler_gnu=no
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 ac_cv_cxx_compiler_gnu=$ac_compiler_gnu
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_cxx_compiler_gnu" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_cxx_compiler_gnu" >&5
 $as_echo "$ac_cv_cxx_compiler_gnu" >&6; }
 if test $ac_compiler_gnu = yes; then
   GXX=yes
@@ -2827,20 +3344,16 @@ else
 fi
 ac_test_CXXFLAGS=${CXXFLAGS+set}
 ac_save_CXXFLAGS=$CXXFLAGS
-{ $as_echo "$as_me:$LINENO: checking whether $CXX accepts -g" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX accepts -g" >&5
 $as_echo_n "checking whether $CXX accepts -g... " >&6; }
-if test "${ac_cv_prog_cxx_g+set}" = set; then
+if ${ac_cv_prog_cxx_g+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   ac_save_cxx_werror_flag=$ac_cxx_werror_flag
    ac_cxx_werror_flag=yes
    ac_cv_prog_cxx_g=no
    CXXFLAGS="-g"
-   cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2851,35 +3364,11 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_cxx_try_compile "$LINENO"; then :
   ac_cv_prog_cxx_g=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	CXXFLAGS=""
-      cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  CXXFLAGS=""
+      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2890,36 +3379,12 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+if ac_fn_cxx_try_compile "$LINENO"; then :
 
-	ac_cxx_werror_flag=$ac_save_cxx_werror_flag
+else
+  ac_cxx_werror_flag=$ac_save_cxx_werror_flag
 	 CXXFLAGS="-g"
-	 cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2930,42 +3395,17 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_cxx_try_compile "$LINENO"; then :
   ac_cv_prog_cxx_g=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
    ac_cxx_werror_flag=$ac_save_cxx_werror_flag
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cxx_g" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cxx_g" >&5
 $as_echo "$ac_cv_prog_cxx_g" >&6; }
 if test "$ac_test_CXXFLAGS" = set; then
   CXXFLAGS=$ac_save_CXXFLAGS
@@ -2999,14 +3439,14 @@ am__doit:
 .PHONY: am__doit
 END
 # If we don't find an include directive, just comment out the code.
-{ $as_echo "$as_me:$LINENO: checking for style of include used by $am_make" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for style of include used by $am_make" >&5
 $as_echo_n "checking for style of include used by $am_make... " >&6; }
 am__include="#"
 am__quote=
 _am_result=none
 # First try GNU make style include.
 echo "include confinc" > confmf
-# Ignore all kinds of additional output from `make'.
+# Ignore all kinds of additional output from 'make'.
 case `$am_make -s -f confmf 2> /dev/null` in #(
 *the\ am__doit\ target*)
   am__include=include
@@ -3027,18 +3467,19 @@ if test "$am__include" = "#"; then
 fi
 
 
-{ $as_echo "$as_me:$LINENO: result: $_am_result" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $_am_result" >&5
 $as_echo "$_am_result" >&6; }
 rm -f confinc confmf
 
 # Check whether --enable-dependency-tracking was given.
-if test "${enable_dependency_tracking+set}" = set; then
+if test "${enable_dependency_tracking+set}" = set; then :
   enableval=$enable_dependency_tracking;
 fi
 
 if test "x$enable_dependency_tracking" != xno; then
   am_depcomp="$ac_aux_dir/depcomp"
   AMDEPBACKSLASH='\'
+  am__nodep='_no'
 fi
  if test "x$enable_dependency_tracking" != xno; then
   AMDEP_TRUE=
@@ -3052,17 +3493,18 @@ fi
 
 depcc="$CXX"  am_compiler_list=
 
-{ $as_echo "$as_me:$LINENO: checking dependency style of $depcc" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
 $as_echo_n "checking dependency style of $depcc... " >&6; }
-if test "${am_cv_CXX_dependencies_compiler_type+set}" = set; then
+if ${am_cv_CXX_dependencies_compiler_type+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named `D' -- because `-MD' means `put the output
-  # in D'.
+  # making a dummy file named 'D' -- because '-MD' means "put the output
+  # in D".
+  rm -rf conftest.dir
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -3096,16 +3538,16 @@ else
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
-      # Solaris 8's {/usr,}/bin/sh.
-      touch sub/conftst$i.h
+      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
+      # Solaris 10 /bin/sh.
+      echo '/* dummy */' > sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with `-c' and `-o' for the sake of the "dashmstdout"
+    # We check with '-c' and '-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle `-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs
+    # handle '-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs.
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -3114,16 +3556,16 @@ else
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # after this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested
+      # After this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested.
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvisualcpp | msvcmsys)
-      # This compiler won't grok `-c -o', but also, the minuso test has
+    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
+      # This compiler won't grok '-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -3162,7 +3604,7 @@ else
 fi
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $am_cv_CXX_dependencies_compiler_type" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CXX_dependencies_compiler_type" >&5
 $as_echo "$am_cv_CXX_dependencies_compiler_type" >&6; }
 CXXDEPMODE=depmode=$am_cv_CXX_dependencies_compiler_type
 
@@ -3185,9 +3627,9 @@ ac_compiler_gnu=$ac_cv_c_compiler_gnu
 if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}gcc", so it can be a program name with args.
 set dummy ${ac_tool_prefix}gcc; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_CC+set}" = set; then
+if ${ac_cv_prog_CC+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3198,24 +3640,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_CC="${ac_tool_prefix}gcc"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:$LINENO: result: $CC" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3225,9 +3667,9 @@ if test -z "$ac_cv_prog_CC"; then
   ac_ct_CC=$CC
   # Extract the first word of "gcc", so it can be a program name with args.
 set dummy gcc; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_ac_ct_CC+set}" = set; then
+if ${ac_cv_prog_ac_ct_CC+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CC"; then
@@ -3238,24 +3680,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_ac_ct_CC="gcc"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CC=$ac_cv_prog_ac_ct_CC
 if test -n "$ac_ct_CC"; then
-  { $as_echo "$as_me:$LINENO: result: $ac_ct_CC" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
 $as_echo "$ac_ct_CC" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3264,7 +3706,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -3278,9 +3720,9 @@ if test -z "$CC"; then
           if test -n "$ac_tool_prefix"; then
     # Extract the first word of "${ac_tool_prefix}cc", so it can be a program name with args.
 set dummy ${ac_tool_prefix}cc; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_CC+set}" = set; then
+if ${ac_cv_prog_CC+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3291,24 +3733,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_CC="${ac_tool_prefix}cc"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:$LINENO: result: $CC" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3318,9 +3760,9 @@ fi
 if test -z "$CC"; then
   # Extract the first word of "cc", so it can be a program name with args.
 set dummy cc; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_CC+set}" = set; then
+if ${ac_cv_prog_CC+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3332,18 +3774,18 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     if test "$as_dir/$ac_word$ac_exec_ext" = "/usr/ucb/cc"; then
        ac_prog_rejected=yes
        continue
      fi
     ac_cv_prog_CC="cc"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 if test $ac_prog_rejected = yes; then
@@ -3362,10 +3804,10 @@ fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:$LINENO: result: $CC" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3377,9 +3819,9 @@ if test -z "$CC"; then
   do
     # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
 set dummy $ac_tool_prefix$ac_prog; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_CC+set}" = set; then
+if ${ac_cv_prog_CC+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3390,24 +3832,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_CC="$ac_tool_prefix$ac_prog"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:$LINENO: result: $CC" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3421,9 +3863,9 @@ if test -z "$CC"; then
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_ac_ct_CC+set}" = set; then
+if ${ac_cv_prog_ac_ct_CC+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CC"; then
@@ -3434,24 +3876,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_ac_ct_CC="$ac_prog"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CC=$ac_cv_prog_ac_ct_CC
 if test -n "$ac_ct_CC"; then
-  { $as_echo "$as_me:$LINENO: result: $ac_ct_CC" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
 $as_echo "$ac_ct_CC" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3464,7 +3906,7 @@ done
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -3475,62 +3917,42 @@ fi
 fi
 
 
-test -z "$CC" && { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+test -z "$CC" && { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: no acceptable C compiler found in \$PATH
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: no acceptable C compiler found in \$PATH
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
+as_fn_error $? "no acceptable C compiler found in \$PATH
+See \`config.log' for more details" "$LINENO" 5; }
 
 # Provide some information about the compiler.
-$as_echo "$as_me:$LINENO: checking for C compiler version" >&5
+$as_echo "$as_me:${as_lineno-$LINENO}: checking for C compiler version" >&5
 set X $ac_compile
 ac_compiler=$2
-{ (ac_try="$ac_compiler --version >&5"
+for ac_option in --version -v -V -qversion; do
+  { { ac_try="$ac_compiler $ac_option >&5"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler --version >&5") 2>&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
-{ (ac_try="$ac_compiler -v >&5"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler -v >&5") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
-{ (ac_try="$ac_compiler -V >&5"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler -V >&5") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
+  if test -s conftest.err; then
+    sed '10a\
+... rest of stderr output deleted ...
+         10q' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+  fi
+  rm -f conftest.er1 conftest.err
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }
+done
 
-{ $as_echo "$as_me:$LINENO: checking whether we are using the GNU C compiler" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C compiler" >&5
 $as_echo_n "checking whether we are using the GNU C compiler... " >&6; }
-if test "${ac_cv_c_compiler_gnu+set}" = set; then
+if ${ac_cv_c_compiler_gnu+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -3544,37 +3966,16 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_c_try_compile "$LINENO"; then :
   ac_compiler_gnu=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_compiler_gnu=no
+  ac_compiler_gnu=no
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 ac_cv_c_compiler_gnu=$ac_compiler_gnu
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_compiler_gnu" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_c_compiler_gnu" >&5
 $as_echo "$ac_cv_c_compiler_gnu" >&6; }
 if test $ac_compiler_gnu = yes; then
   GCC=yes
@@ -3583,20 +3984,16 @@ else
 fi
 ac_test_CFLAGS=${CFLAGS+set}
 ac_save_CFLAGS=$CFLAGS
-{ $as_echo "$as_me:$LINENO: checking whether $CC accepts -g" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC accepts -g" >&5
 $as_echo_n "checking whether $CC accepts -g... " >&6; }
-if test "${ac_cv_prog_cc_g+set}" = set; then
+if ${ac_cv_prog_cc_g+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   ac_save_c_werror_flag=$ac_c_werror_flag
    ac_c_werror_flag=yes
    ac_cv_prog_cc_g=no
    CFLAGS="-g"
-   cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -3607,35 +4004,11 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_c_try_compile "$LINENO"; then :
   ac_cv_prog_cc_g=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	CFLAGS=""
-      cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  CFLAGS=""
+      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -3646,36 +4019,12 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+if ac_fn_c_try_compile "$LINENO"; then :
 
-	ac_c_werror_flag=$ac_save_c_werror_flag
+else
+  ac_c_werror_flag=$ac_save_c_werror_flag
 	 CFLAGS="-g"
-	 cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -3686,42 +4035,17 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_c_try_compile "$LINENO"; then :
   ac_cv_prog_cc_g=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
    ac_c_werror_flag=$ac_save_c_werror_flag
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cc_g" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_g" >&5
 $as_echo "$ac_cv_prog_cc_g" >&6; }
 if test "$ac_test_CFLAGS" = set; then
   CFLAGS=$ac_save_CFLAGS
@@ -3738,23 +4062,18 @@ else
     CFLAGS=
   fi
 fi
-{ $as_echo "$as_me:$LINENO: checking for $CC option to accept ISO C89" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to accept ISO C89" >&5
 $as_echo_n "checking for $CC option to accept ISO C89... " >&6; }
-if test "${ac_cv_prog_cc_c89+set}" = set; then
+if ${ac_cv_prog_cc_c89+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   ac_cv_prog_cc_c89=no
 ac_save_CC=$CC
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <stdarg.h>
 #include <stdio.h>
-#include <sys/types.h>
-#include <sys/stat.h>
+struct stat;
 /* Most of the following tests are stolen from RCS 5.7's src/conf.sh.  */
 struct buf { int x; };
 FILE * (*rcsopen) (struct buf *, struct stat *, int);
@@ -3806,32 +4125,9 @@ for ac_arg in '' -qlanglvl=extc89 -qlanglvl=ansi -std \
 	-Ae "-Aa -D_HPUX_SOURCE" "-Xc -D__EXTENSIONS__"
 do
   CC="$ac_save_CC $ac_arg"
-  rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+  if ac_fn_c_try_compile "$LINENO"; then :
   ac_cv_prog_cc_c89=$ac_arg
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
 fi
-
 rm -f core conftest.err conftest.$ac_objext
   test "x$ac_cv_prog_cc_c89" != "xno" && break
 done
@@ -3842,17 +4138,19 @@ fi
 # AC_CACHE_VAL
 case "x$ac_cv_prog_cc_c89" in
   x)
-    { $as_echo "$as_me:$LINENO: result: none needed" >&5
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: none needed" >&5
 $as_echo "none needed" >&6; } ;;
   xno)
-    { $as_echo "$as_me:$LINENO: result: unsupported" >&5
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: unsupported" >&5
 $as_echo "unsupported" >&6; } ;;
   *)
     CC="$CC $ac_cv_prog_cc_c89"
-    { $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cc_c89" >&5
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_c89" >&5
 $as_echo "$ac_cv_prog_cc_c89" >&6; } ;;
 esac
+if test "x$ac_cv_prog_cc_c89" != xno; then :
 
+fi
 
 ac_ext=c
 ac_cpp='$CPP $CPPFLAGS'
@@ -3860,19 +4158,79 @@ ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
 ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC understands -c and -o together" >&5
+$as_echo_n "checking whether $CC understands -c and -o together... " >&6; }
+if ${am_cv_prog_cc_c_o+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+  # Make sure it works both with $CC and with simple cc.
+  # Following AC_PROG_CC_C_O, we do the test twice because some
+  # compilers refuse to overwrite an existing .o file with -o,
+  # though they will create one.
+  am_cv_prog_cc_c_o=yes
+  for am_i in 1 2; do
+    if { echo "$as_me:$LINENO: $CC -c conftest.$ac_ext -o conftest2.$ac_objext" >&5
+   ($CC -c conftest.$ac_ext -o conftest2.$ac_objext) >&5 2>&5
+   ac_status=$?
+   echo "$as_me:$LINENO: \$? = $ac_status" >&5
+   (exit $ac_status); } \
+         && test -f conftest2.$ac_objext; then
+      : OK
+    else
+      am_cv_prog_cc_c_o=no
+      break
+    fi
+  done
+  rm -f core conftest*
+  unset am_i
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_prog_cc_c_o" >&5
+$as_echo "$am_cv_prog_cc_c_o" >&6; }
+if test "$am_cv_prog_cc_c_o" != yes; then
+   # Losing compiler, so override with the script.
+   # FIXME: It is wrong to rewrite CC.
+   # But if we don't then we get into trouble of one sort or another.
+   # A longer-term fix would be to have automake use am__CC in this case,
+   # and then we could set am__CC="\$(top_srcdir)/compile \$(CC)"
+   CC="$am_aux_dir/compile $CC"
+fi
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+
+
 depcc="$CC"   am_compiler_list=
 
-{ $as_echo "$as_me:$LINENO: checking dependency style of $depcc" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
 $as_echo_n "checking dependency style of $depcc... " >&6; }
-if test "${am_cv_CC_dependencies_compiler_type+set}" = set; then
+if ${am_cv_CC_dependencies_compiler_type+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named `D' -- because `-MD' means `put the output
-  # in D'.
+  # making a dummy file named 'D' -- because '-MD' means "put the output
+  # in D".
+  rm -rf conftest.dir
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -3906,16 +4264,16 @@ else
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
-      # Solaris 8's {/usr,}/bin/sh.
-      touch sub/conftst$i.h
+      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
+      # Solaris 10 /bin/sh.
+      echo '/* dummy */' > sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with `-c' and `-o' for the sake of the "dashmstdout"
+    # We check with '-c' and '-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle `-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs
+    # handle '-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs.
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -3924,16 +4282,16 @@ else
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # after this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested
+      # After this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested.
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvisualcpp | msvcmsys)
-      # This compiler won't grok `-c -o', but also, the minuso test has
+    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
+      # This compiler won't grok '-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -3972,7 +4330,7 @@ else
 fi
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $am_cv_CC_dependencies_compiler_type" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CC_dependencies_compiler_type" >&5
 $as_echo "$am_cv_CC_dependencies_compiler_type" >&6; }
 CCDEPMODE=depmode=$am_cv_CC_dependencies_compiler_type
 
@@ -3991,17 +4349,18 @@ fi
 
   OPENMP_CFLAGS=
   # Check whether --enable-openmp was given.
-if test "${enable_openmp+set}" = set; then
+if test "${enable_openmp+set}" = set; then :
   enableval=$enable_openmp;
 fi
 
   if test "$enable_openmp" != no; then
-    { $as_echo "$as_me:$LINENO: checking for $CC option to support OpenMP" >&5
+    { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to support OpenMP" >&5
 $as_echo_n "checking for $CC option to support OpenMP... " >&6; }
-if test "${ac_cv_prog_c_openmp+set}" = set; then
+if ${ac_cv_prog_c_openmp+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
 
 #ifndef _OPENMP
  choke me
@@ -4010,37 +4369,16 @@ else
 int main () { return omp_get_num_threads (); }
 
 _ACEOF
-rm -f conftest.$ac_objext conftest$ac_exeext
-if { (ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_link") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 $as_test_x conftest$ac_exeext
-       }; then
+if ac_fn_c_try_link "$LINENO"; then :
   ac_cv_prog_c_openmp='none needed'
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_cv_prog_c_openmp='unsupported'
-	  	  	  	  	  	  	  	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp; do
+  ac_cv_prog_c_openmp='unsupported'
+	  	  	  	  	  	  	                                	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp -homp \
+                           -Popenmp --openmp; do
 	    ac_save_CFLAGS=$CFLAGS
 	    CFLAGS="$CFLAGS $ac_option"
-	    cat >conftest.$ac_ext <<_ACEOF
+	    cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
 
 #ifndef _OPENMP
  choke me
@@ -4049,56 +4387,27 @@ sed 's/^/| /' conftest.$ac_ext >&5
 int main () { return omp_get_num_threads (); }
 
 _ACEOF
-rm -f conftest.$ac_objext conftest$ac_exeext
-if { (ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_link") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 $as_test_x conftest$ac_exeext
-       }; then
+if ac_fn_c_try_link "$LINENO"; then :
   ac_cv_prog_c_openmp=$ac_option
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
 fi
-
-rm -rf conftest.dSYM
-rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
-      conftest$ac_exeext conftest.$ac_ext
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
 	    CFLAGS=$ac_save_CFLAGS
 	    if test "$ac_cv_prog_c_openmp" != unsupported; then
 	      break
 	    fi
 	  done
 fi
-
-rm -rf conftest.dSYM
-rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
-      conftest$ac_exeext conftest.$ac_ext
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_c_openmp" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_c_openmp" >&5
 $as_echo "$ac_cv_prog_c_openmp" >&6; }
     case $ac_cv_prog_c_openmp in #(
       "none needed" | unsupported)
-        ;; #(
+	;; #(
       *)
-        OPENMP_CFLAGS=$ac_cv_prog_c_openmp ;;
+	OPENMP_CFLAGS=$ac_cv_prog_c_openmp ;;
     esac
   fi
 
@@ -4106,9 +4415,9 @@ $as_echo "$ac_cv_prog_c_openmp" >&6; }
 if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}ranlib", so it can be a program name with args.
 set dummy ${ac_tool_prefix}ranlib; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_RANLIB+set}" = set; then
+if ${ac_cv_prog_RANLIB+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$RANLIB"; then
@@ -4119,24 +4428,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_RANLIB="${ac_tool_prefix}ranlib"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 RANLIB=$ac_cv_prog_RANLIB
 if test -n "$RANLIB"; then
-  { $as_echo "$as_me:$LINENO: result: $RANLIB" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $RANLIB" >&5
 $as_echo "$RANLIB" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -4146,9 +4455,9 @@ if test -z "$ac_cv_prog_RANLIB"; then
   ac_ct_RANLIB=$RANLIB
   # Extract the first word of "ranlib", so it can be a program name with args.
 set dummy ranlib; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_ac_ct_RANLIB+set}" = set; then
+if ${ac_cv_prog_ac_ct_RANLIB+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_RANLIB"; then
@@ -4159,24 +4468,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_ac_ct_RANLIB="ranlib"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_RANLIB=$ac_cv_prog_ac_ct_RANLIB
 if test -n "$ac_ct_RANLIB"; then
-  { $as_echo "$as_me:$LINENO: result: $ac_ct_RANLIB" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_RANLIB" >&5
 $as_echo "$ac_ct_RANLIB" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -4185,7 +4494,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -4205,14 +4514,14 @@ ac_cpp='$CPP $CPPFLAGS'
 ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
 ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
-{ $as_echo "$as_me:$LINENO: checking how to run the C preprocessor" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking how to run the C preprocessor" >&5
 $as_echo_n "checking how to run the C preprocessor... " >&6; }
 # On Suns, sometimes $CPP names a directory.
 if test -n "$CPP" && test -d "$CPP"; then
   CPP=
 fi
 if test -z "$CPP"; then
-  if test "${ac_cv_prog_CPP+set}" = set; then
+  if ${ac_cv_prog_CPP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
       # Double quotes because CPP needs to be expanded
@@ -4227,11 +4536,7 @@ do
   # <limits.h> exists even on freestanding compilers.
   # On the NeXT, cc -E runs the code through the compiler's parser,
   # not just through cpp. "Syntax error" is here to catch this case.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #ifdef __STDC__
 # include <limits.h>
@@ -4240,78 +4545,34 @@ cat >>conftest.$ac_ext <<_ACEOF
 #endif
 		     Syntax error
 _ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+if ac_fn_c_try_cpp "$LINENO"; then :
 
+else
   # Broken: fails on valid input.
 continue
 fi
-
-rm -f conftest.err conftest.$ac_ext
+rm -f conftest.err conftest.i conftest.$ac_ext
 
   # OK, works on sane cases.  Now check whether nonexistent headers
   # can be detected and how.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <ac_nonexistent.h>
 _ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
+if ac_fn_c_try_cpp "$LINENO"; then :
   # Broken: success on invalid input.
 continue
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
   # Passes both tests.
 ac_preproc_ok=:
 break
 fi
-
-rm -f conftest.err conftest.$ac_ext
+rm -f conftest.err conftest.i conftest.$ac_ext
 
 done
 # Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
-rm -f conftest.err conftest.$ac_ext
-if $ac_preproc_ok; then
+rm -f conftest.i conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then :
   break
 fi
 
@@ -4323,7 +4584,7 @@ fi
 else
   ac_cv_prog_CPP=$CPP
 fi
-{ $as_echo "$as_me:$LINENO: result: $CPP" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $CPP" >&5
 $as_echo "$CPP" >&6; }
 ac_preproc_ok=false
 for ac_c_preproc_warn_flag in '' yes
@@ -4334,11 +4595,7 @@ do
   # <limits.h> exists even on freestanding compilers.
   # On the NeXT, cc -E runs the code through the compiler's parser,
   # not just through cpp. "Syntax error" is here to catch this case.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #ifdef __STDC__
 # include <limits.h>
@@ -4347,87 +4604,40 @@ cat >>conftest.$ac_ext <<_ACEOF
 #endif
 		     Syntax error
 _ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+if ac_fn_c_try_cpp "$LINENO"; then :
 
+else
   # Broken: fails on valid input.
 continue
 fi
-
-rm -f conftest.err conftest.$ac_ext
+rm -f conftest.err conftest.i conftest.$ac_ext
 
   # OK, works on sane cases.  Now check whether nonexistent headers
   # can be detected and how.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <ac_nonexistent.h>
 _ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
+if ac_fn_c_try_cpp "$LINENO"; then :
   # Broken: success on invalid input.
 continue
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
   # Passes both tests.
 ac_preproc_ok=:
 break
 fi
-
-rm -f conftest.err conftest.$ac_ext
+rm -f conftest.err conftest.i conftest.$ac_ext
 
 done
 # Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
-rm -f conftest.err conftest.$ac_ext
-if $ac_preproc_ok; then
-  :
+rm -f conftest.i conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then :
+
 else
-  { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: C preprocessor \"$CPP\" fails sanity check
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: C preprocessor \"$CPP\" fails sanity check
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
+as_fn_error $? "C preprocessor \"$CPP\" fails sanity check
+See \`config.log' for more details" "$LINENO" 5; }
 fi
 
 ac_ext=c
@@ -4437,9 +4647,9 @@ ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
 
-{ $as_echo "$as_me:$LINENO: checking for grep that handles long lines and -e" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for grep that handles long lines and -e" >&5
 $as_echo_n "checking for grep that handles long lines and -e... " >&6; }
-if test "${ac_cv_path_GREP+set}" = set; then
+if ${ac_cv_path_GREP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -z "$GREP"; then
@@ -4450,10 +4660,10 @@ for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_prog in grep ggrep; do
+    for ac_prog in grep ggrep; do
     for ac_exec_ext in '' $ac_executable_extensions; do
       ac_path_GREP="$as_dir/$ac_prog$ac_exec_ext"
-      { test -f "$ac_path_GREP" && $as_test_x "$ac_path_GREP"; } || continue
+      as_fn_executable_p "$ac_path_GREP" || continue
 # Check for GNU ac_path_GREP and select it if it is found.
   # Check for GNU $ac_path_GREP
 case `"$ac_path_GREP" --version 2>&1` in
@@ -4470,7 +4680,7 @@ case `"$ac_path_GREP" --version 2>&1` in
     $as_echo 'GREP' >> "conftest.nl"
     "$ac_path_GREP" -e 'GREP$' -e '-(cannot match)-' < "conftest.nl" >"conftest.out" 2>/dev/null || break
     diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
-    ac_count=`expr $ac_count + 1`
+    as_fn_arith $ac_count + 1 && ac_count=$as_val
     if test $ac_count -gt ${ac_path_GREP_max-0}; then
       # Best one so far, save it but keep looking for a better one
       ac_cv_path_GREP="$ac_path_GREP"
@@ -4485,26 +4695,24 @@ esac
       $ac_path_GREP_found && break 3
     done
   done
-done
+  done
 IFS=$as_save_IFS
   if test -z "$ac_cv_path_GREP"; then
-    { { $as_echo "$as_me:$LINENO: error: no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&5
-$as_echo "$as_me: error: no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
   fi
 else
   ac_cv_path_GREP=$GREP
 fi
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_path_GREP" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_GREP" >&5
 $as_echo "$ac_cv_path_GREP" >&6; }
  GREP="$ac_cv_path_GREP"
 
 
-{ $as_echo "$as_me:$LINENO: checking for egrep" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for egrep" >&5
 $as_echo_n "checking for egrep... " >&6; }
-if test "${ac_cv_path_EGREP+set}" = set; then
+if ${ac_cv_path_EGREP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if echo a | $GREP -E '(a|b)' >/dev/null 2>&1
@@ -4518,10 +4726,10 @@ for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_prog in egrep; do
+    for ac_prog in egrep; do
     for ac_exec_ext in '' $ac_executable_extensions; do
       ac_path_EGREP="$as_dir/$ac_prog$ac_exec_ext"
-      { test -f "$ac_path_EGREP" && $as_test_x "$ac_path_EGREP"; } || continue
+      as_fn_executable_p "$ac_path_EGREP" || continue
 # Check for GNU ac_path_EGREP and select it if it is found.
   # Check for GNU $ac_path_EGREP
 case `"$ac_path_EGREP" --version 2>&1` in
@@ -4538,7 +4746,7 @@ case `"$ac_path_EGREP" --version 2>&1` in
     $as_echo 'EGREP' >> "conftest.nl"
     "$ac_path_EGREP" 'EGREP$' < "conftest.nl" >"conftest.out" 2>/dev/null || break
     diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
-    ac_count=`expr $ac_count + 1`
+    as_fn_arith $ac_count + 1 && ac_count=$as_val
     if test $ac_count -gt ${ac_path_EGREP_max-0}; then
       # Best one so far, save it but keep looking for a better one
       ac_cv_path_EGREP="$ac_path_EGREP"
@@ -4553,12 +4761,10 @@ esac
       $ac_path_EGREP_found && break 3
     done
   done
-done
+  done
 IFS=$as_save_IFS
   if test -z "$ac_cv_path_EGREP"; then
-    { { $as_echo "$as_me:$LINENO: error: no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&5
-$as_echo "$as_me: error: no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
   fi
 else
   ac_cv_path_EGREP=$EGREP
@@ -4566,21 +4772,17 @@ fi
 
    fi
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_path_EGREP" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_EGREP" >&5
 $as_echo "$ac_cv_path_EGREP" >&6; }
  EGREP="$ac_cv_path_EGREP"
 
 
-{ $as_echo "$as_me:$LINENO: checking for ANSI C header files" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for ANSI C header files" >&5
 $as_echo_n "checking for ANSI C header files... " >&6; }
-if test "${ac_cv_header_stdc+set}" = set; then
+if ${ac_cv_header_stdc+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <stdlib.h>
 #include <stdarg.h>
@@ -4595,48 +4797,23 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_c_try_compile "$LINENO"; then :
   ac_cv_header_stdc=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_cv_header_stdc=no
+  ac_cv_header_stdc=no
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 
 if test $ac_cv_header_stdc = yes; then
   # SunOS 4.x string.h does not declare mem*, contrary to ANSI.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <string.h>
 
 _ACEOF
 if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
-  $EGREP "memchr" >/dev/null 2>&1; then
-  :
+  $EGREP "memchr" >/dev/null 2>&1; then :
+
 else
   ac_cv_header_stdc=no
 fi
@@ -4646,18 +4823,14 @@ fi
 
 if test $ac_cv_header_stdc = yes; then
   # ISC 2.0.2 stdlib.h does not declare free, contrary to ANSI.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <stdlib.h>
 
 _ACEOF
 if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
-  $EGREP "free" >/dev/null 2>&1; then
-  :
+  $EGREP "free" >/dev/null 2>&1; then :
+
 else
   ac_cv_header_stdc=no
 fi
@@ -4667,14 +4840,10 @@ fi
 
 if test $ac_cv_header_stdc = yes; then
   # /bin/cc in Irix-4.0.5 gets non-ANSI ctype macros unless using -ansi.
-  if test "$cross_compiling" = yes; then
+  if test "$cross_compiling" = yes; then :
   :
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <ctype.h>
 #include <stdlib.h>
@@ -4701,118 +4870,33 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest$ac_exeext
-if { (ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_link") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && { ac_try='./conftest$ac_exeext'
-  { (case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; }; then
-  :
+if ac_fn_c_try_run "$LINENO"; then :
+
 else
-  $as_echo "$as_me: program exited with status $ac_status" >&5
-$as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-( exit $ac_status )
-ac_cv_header_stdc=no
+  ac_cv_header_stdc=no
 fi
-rm -rf conftest.dSYM
-rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext conftest.$ac_objext conftest.$ac_ext
+rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
+  conftest.$ac_objext conftest.beam conftest.$ac_ext
 fi
 
-
 fi
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_header_stdc" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_header_stdc" >&5
 $as_echo "$ac_cv_header_stdc" >&6; }
 if test $ac_cv_header_stdc = yes; then
 
-cat >>confdefs.h <<\_ACEOF
-#define STDC_HEADERS 1
-_ACEOF
+$as_echo "#define STDC_HEADERS 1" >>confdefs.h
 
 fi
 
 # On IRIX 5.3, sys/types and inttypes.h are conflicting.
-
-
-
-
-
-
-
-
-
 for ac_header in sys/types.h sys/stat.h stdlib.h string.h memory.h strings.h \
 		  inttypes.h stdint.h unistd.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  eval "$as_ac_Header=yes"
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	eval "$as_ac_Header=no"
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+ac_fn_c_check_header_compile "$LINENO" "$ac_header" "$as_ac_Header" "$ac_includes_default
+"
+if eval test \"x\$"$as_ac_Header"\" = x"yes"; then :
   cat >>confdefs.h <<_ACEOF
 #define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
@@ -4822,453 +4906,36 @@ fi
 done
 
 
-
 for ac_header in stdint.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
-$as_echo_n "checking $ac_header usability... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_header_compiler=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_header_compiler=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
-$as_echo_n "checking $ac_header presence... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <$ac_header>
-_ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  ac_header_preproc=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-  ac_header_preproc=no
-fi
-
-rm -f conftest.err conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
-  yes:no: )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
-    ac_header_preproc=yes
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
-$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
-    ( cat <<\_ASBOX
-## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##
-_ASBOX
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  eval "$as_ac_Header=\$ac_header_preproc"
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-
-fi
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "stdint.h" "ac_cv_header_stdint_h" "$ac_includes_default"
+if test "x$ac_cv_header_stdint_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+#define HAVE_STDINT_H 1
 _ACEOF
 
 fi
 
 done
 
-
 for ac_header in malloc/malloc.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
-$as_echo_n "checking $ac_header usability... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_header_compiler=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_header_compiler=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
-$as_echo_n "checking $ac_header presence... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <$ac_header>
-_ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  ac_header_preproc=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-  ac_header_preproc=no
-fi
-
-rm -f conftest.err conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
-  yes:no: )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
-    ac_header_preproc=yes
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
-$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
-    ( cat <<\_ASBOX
-## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##
-_ASBOX
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  eval "$as_ac_Header=\$ac_header_preproc"
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-
-fi
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "malloc/malloc.h" "ac_cv_header_malloc_malloc_h" "$ac_includes_default"
+if test "x$ac_cv_header_malloc_malloc_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+#define HAVE_MALLOC_MALLOC_H 1
 _ACEOF
 
 fi
 
 done
 
-
 for ac_header in malloc.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
-$as_echo_n "checking $ac_header usability... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_header_compiler=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_header_compiler=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
-$as_echo_n "checking $ac_header presence... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <$ac_header>
-_ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  ac_header_preproc=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-  ac_header_preproc=no
-fi
-
-rm -f conftest.err conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
-  yes:no: )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
-    ac_header_preproc=yes
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
-$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
-    ( cat <<\_ASBOX
-## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##
-_ASBOX
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  eval "$as_ac_Header=\$ac_header_preproc"
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-
-fi
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "malloc.h" "ac_cv_header_malloc_h" "$ac_includes_default"
+if test "x$ac_cv_header_malloc_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+#define HAVE_MALLOC_H 1
 _ACEOF
 
 fi
@@ -5277,102 +4944,9 @@ done
 
 
 # Checks for typedefs, structures, and compiler characteristics.
-{ $as_echo "$as_me:$LINENO: checking for size_t" >&5
-$as_echo_n "checking for size_t... " >&6; }
-if test "${ac_cv_type_size_t+set}" = set; then
-  $as_echo_n "(cached) " >&6
-else
-  ac_cv_type_size_t=no
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-if (sizeof (size_t))
-       return 0;
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-if (sizeof ((size_t)))
-	  return 0;
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+ac_fn_c_check_type "$LINENO" "size_t" "ac_cv_type_size_t" "$ac_includes_default"
+if test "x$ac_cv_type_size_t" = xyes; then :
 
-	ac_cv_type_size_t=yes
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_type_size_t" >&5
-$as_echo "$ac_cv_type_size_t" >&6; }
-if test "x$ac_cv_type_size_t" = x""yes; then
-  :
 else
 
 cat >>confdefs.h <<_ACEOF
@@ -5381,75 +4955,12 @@ _ACEOF
 
 fi
 
-
-  { $as_echo "$as_me:$LINENO: checking for uint32_t" >&5
-$as_echo_n "checking for uint32_t... " >&6; }
-if test "${ac_cv_c_uint32_t+set}" = set; then
-  $as_echo_n "(cached) " >&6
-else
-  ac_cv_c_uint32_t=no
-     for ac_type in 'uint32_t' 'unsigned int' 'unsigned long int' \
-	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
-       cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-static int test_array [1 - 2 * !(($ac_type) -1 >> (32 - 1) == 1)];
-test_array [0] = 0
-
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  case $ac_type in
-  uint32_t) ac_cv_c_uint32_t=yes ;;
-  *) ac_cv_c_uint32_t=$ac_type ;;
-esac
-
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-       test "$ac_cv_c_uint32_t" != no && break
-     done
-fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_uint32_t" >&5
-$as_echo "$ac_cv_c_uint32_t" >&6; }
-  case $ac_cv_c_uint32_t in #(
+ac_fn_c_find_uintX_t "$LINENO" "32" "ac_cv_c_uint32_t"
+case $ac_cv_c_uint32_t in #(
   no|yes) ;; #(
   *)
 
-cat >>confdefs.h <<\_ACEOF
-#define _UINT32_T 1
-_ACEOF
+$as_echo "#define _UINT32_T 1" >>confdefs.h
 
 
 cat >>confdefs.h <<_ACEOF
@@ -5458,75 +4969,12 @@ _ACEOF
 ;;
   esac
 
-
-  { $as_echo "$as_me:$LINENO: checking for uint64_t" >&5
-$as_echo_n "checking for uint64_t... " >&6; }
-if test "${ac_cv_c_uint64_t+set}" = set; then
-  $as_echo_n "(cached) " >&6
-else
-  ac_cv_c_uint64_t=no
-     for ac_type in 'uint64_t' 'unsigned int' 'unsigned long int' \
-	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
-       cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-static int test_array [1 - 2 * !(($ac_type) -1 >> (64 - 1) == 1)];
-test_array [0] = 0
-
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  case $ac_type in
-  uint64_t) ac_cv_c_uint64_t=yes ;;
-  *) ac_cv_c_uint64_t=$ac_type ;;
-esac
-
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-       test "$ac_cv_c_uint64_t" != no && break
-     done
-fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_uint64_t" >&5
-$as_echo "$ac_cv_c_uint64_t" >&6; }
-  case $ac_cv_c_uint64_t in #(
+ac_fn_c_find_uintX_t "$LINENO" "64" "ac_cv_c_uint64_t"
+case $ac_cv_c_uint64_t in #(
   no|yes) ;; #(
   *)
 
-cat >>confdefs.h <<\_ACEOF
-#define _UINT64_T 1
-_ACEOF
+$as_echo "#define _UINT64_T 1" >>confdefs.h
 
 
 cat >>confdefs.h <<_ACEOF
@@ -5537,102 +4985,12 @@ _ACEOF
 
 
 # Checks for library functions.
-
 for ac_func in gettimeofday
-do
-as_ac_var=`$as_echo "ac_cv_func_$ac_func" | $as_tr_sh`
-{ $as_echo "$as_me:$LINENO: checking for $ac_func" >&5
-$as_echo_n "checking for $ac_func... " >&6; }
-if { as_var=$as_ac_var; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-/* Define $ac_func to an innocuous variant, in case <limits.h> declares $ac_func.
-   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
-#define $ac_func innocuous_$ac_func
-
-/* System header to define __stub macros and hopefully few prototypes,
-    which can conflict with char $ac_func (); below.
-    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
-    <limits.h> exists even on freestanding compilers.  */
-
-#ifdef __STDC__
-# include <limits.h>
-#else
-# include <assert.h>
-#endif
-
-#undef $ac_func
-
-/* Override any GCC internal prototype to avoid an error.
-   Use char because int might match the return type of a GCC
-   builtin and then its argument prototype would still apply.  */
-#ifdef __cplusplus
-extern "C"
-#endif
-char $ac_func ();
-/* The GNU C library defines this for functions which it implements
-    to always fail with ENOSYS.  Some functions are actually named
-    something starting with __ and the normal name is an alias.  */
-#if defined __stub_$ac_func || defined __stub___$ac_func
-choke me
-#endif
-
-int
-main ()
-{
-return $ac_func ();
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext conftest$ac_exeext
-if { (ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_link") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 $as_test_x conftest$ac_exeext
-       }; then
-  eval "$as_ac_var=yes"
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	eval "$as_ac_var=no"
-fi
-
-rm -rf conftest.dSYM
-rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
-      conftest$ac_exeext conftest.$ac_ext
-fi
-ac_res=`eval 'as_val=${'$as_ac_var'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-as_val=`eval 'as_val=${'$as_ac_var'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_c_check_func "$LINENO" "gettimeofday" "ac_cv_func_gettimeofday"
+if test "x$ac_cv_func_gettimeofday" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_func" | $as_tr_cpp` 1
+#define HAVE_GETTIMEOFDAY 1
 _ACEOF
 
 fi
@@ -5640,7 +4998,7 @@ done
 
 
 # Check whether --enable-simd was given.
-if test "${enable_simd+set}" = set; then
+if test "${enable_simd+set}" = set; then :
   enableval=$enable_simd; ac_SIMD=${enable_simd}
 else
   ac_SIMD=AVX2
@@ -5651,42 +5009,66 @@ case ${ac_SIMD} in
      SSE2)
        echo Configuring for SSE2
 
-cat >>confdefs.h <<\_ACEOF
-#define SSE2 1
-_ACEOF
+$as_echo "#define SSE2 1" >>confdefs.h
 
      ;;
      AVX)
        echo Configuring for AVX
 
-cat >>confdefs.h <<\_ACEOF
-#define AVX1 1
-_ACEOF
+$as_echo "#define AVX1 1" >>confdefs.h
 
      ;;
      AVX2)
        echo Configuring for AVX2
 
-cat >>confdefs.h <<\_ACEOF
-#define AVX2 1
-_ACEOF
+$as_echo "#define AVX2 1" >>confdefs.h
 
      ;;
      AVX512)
        echo Configuring for AVX512
 
-cat >>confdefs.h <<\_ACEOF
-#define AVX512 1
-_ACEOF
+$as_echo "#define AVX512 1" >>confdefs.h
 
      ;;
      *)
-     { { $as_echo "$as_me:$LINENO: error: ${ac_SIMD} unsupported --enable-simd option" >&5
-$as_echo "$as_me: error: ${ac_SIMD} unsupported --enable-simd option" >&2;}
-   { (exit 1); exit 1; }; };
+     as_fn_error $? "${ac_SIMD} unsupported --enable-simd option" "$LINENO" 5;
      ;;
 esac
 
+
+# Check whether --enable-comms was given.
+if test "${enable_comms+set}" = set; then :
+  enableval=$enable_comms; ac_COMMS=${enable_comms}
+else
+  ac_COMMS=none
+fi
+
+
+case ${ac_COMMS} in
+     fake)
+       echo Configuring for FAKE communications
+
+$as_echo "#define GRID_COMMS_FAKE 1" >>confdefs.h
+
+     ;;
+     none)
+       echo Configuring for NO communications
+
+$as_echo "#define GRID_COMMS_NONE 1" >>confdefs.h
+
+     ;;
+     mpi)
+       echo Configuring for MPI communications
+
+$as_echo "#define GRID_COMMS_MPI 1" >>confdefs.h
+
+     ;;
+     *)
+     as_fn_error $? "${ac_COMMS} unsupported --enable-comms option" "$LINENO" 5;
+     ;;
+esac
+
+
 ac_config_files="$ac_config_files Makefile"
 
 cat >confcache <<\_ACEOF
@@ -5716,13 +5098,13 @@ _ACEOF
     case $ac_val in #(
     *${as_nl}*)
       case $ac_var in #(
-      *_cv_*) { $as_echo "$as_me:$LINENO: WARNING: cache variable $ac_var contains a newline" >&5
+      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
 $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
       esac
       case $ac_var in #(
       _ | IFS | as_nl) ;; #(
       BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
-      *) $as_unset $ac_var ;;
+      *) { eval $ac_var=; unset $ac_var;} ;;
       esac ;;
     esac
   done
@@ -5730,8 +5112,8 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
   (set) 2>&1 |
     case $as_nl`(ac_space=' '; set) 2>&1` in #(
     *${as_nl}ac_space=\ *)
-      # `set' does not quote correctly, so add quotes (double-quote
-      # substitution turns \\\\ into \\, and sed turns \\ into \).
+      # `set' does not quote correctly, so add quotes: double-quote
+      # substitution turns \\\\ into \\, and sed turns \\ into \.
       sed -n \
 	"s/'/'\\\\''/g;
 	  s/^\\([_$as_cr_alnum]*_cv_[_$as_cr_alnum]*\\)=\\(.*\\)/\\1='\\2'/p"
@@ -5753,12 +5135,23 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
      :end' >>confcache
 if diff "$cache_file" confcache >/dev/null 2>&1; then :; else
   if test -w "$cache_file"; then
-    test "x$cache_file" != "x/dev/null" &&
-      { $as_echo "$as_me:$LINENO: updating cache $cache_file" >&5
+    if test "x$cache_file" != "x/dev/null"; then
+      { $as_echo "$as_me:${as_lineno-$LINENO}: updating cache $cache_file" >&5
 $as_echo "$as_me: updating cache $cache_file" >&6;}
-    cat confcache >$cache_file
+      if test ! -f "$cache_file" || test -h "$cache_file"; then
+	cat confcache >"$cache_file"
+      else
+        case $cache_file in #(
+        */* | ?:*)
+	  mv -f confcache "$cache_file"$$ &&
+	  mv -f "$cache_file"$$ "$cache_file" ;; #(
+        *)
+	  mv -f confcache "$cache_file" ;;
+	esac
+      fi
+    fi
   else
-    { $as_echo "$as_me:$LINENO: not updating unwritable cache $cache_file" >&5
+    { $as_echo "$as_me:${as_lineno-$LINENO}: not updating unwritable cache $cache_file" >&5
 $as_echo "$as_me: not updating unwritable cache $cache_file" >&6;}
   fi
 fi
@@ -5772,20 +5165,29 @@ DEFS=-DHAVE_CONFIG_H
 
 ac_libobjs=
 ac_ltlibobjs=
+U=
 for ac_i in : $LIBOBJS; do test "x$ac_i" = x: && continue
   # 1. Remove the extension, and $U if already installed.
   ac_script='s/\$U\././;s/\.o$//;s/\.obj$//'
   ac_i=`$as_echo "$ac_i" | sed "$ac_script"`
   # 2. Prepend LIBOBJDIR.  When used with automake>=1.10 LIBOBJDIR
   #    will be set to the directory where LIBOBJS objects are built.
-  ac_libobjs="$ac_libobjs \${LIBOBJDIR}$ac_i\$U.$ac_objext"
-  ac_ltlibobjs="$ac_ltlibobjs \${LIBOBJDIR}$ac_i"'$U.lo'
+  as_fn_append ac_libobjs " \${LIBOBJDIR}$ac_i\$U.$ac_objext"
+  as_fn_append ac_ltlibobjs " \${LIBOBJDIR}$ac_i"'$U.lo'
 done
 LIBOBJS=$ac_libobjs
 
 LTLIBOBJS=$ac_ltlibobjs
 
 
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking that generated files are newer than configure" >&5
+$as_echo_n "checking that generated files are newer than configure... " >&6; }
+   if test -n "$am_sleep_pid"; then
+     # Hide warnings about reused PIDs.
+     wait $am_sleep_pid 2>/dev/null
+   fi
+   { $as_echo "$as_me:${as_lineno-$LINENO}: result: done" >&5
+$as_echo "done" >&6; }
  if test -n "$EXEEXT"; then
   am__EXEEXT_TRUE=
   am__EXEEXT_FALSE='#'
@@ -5795,34 +5197,26 @@ else
 fi
 
 if test -z "${AMDEP_TRUE}" && test -z "${AMDEP_FALSE}"; then
-  { { $as_echo "$as_me:$LINENO: error: conditional \"AMDEP\" was never defined.
-Usually this means the macro was only invoked conditionally." >&5
-$as_echo "$as_me: error: conditional \"AMDEP\" was never defined.
-Usually this means the macro was only invoked conditionally." >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "conditional \"AMDEP\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 if test -z "${am__fastdepCXX_TRUE}" && test -z "${am__fastdepCXX_FALSE}"; then
-  { { $as_echo "$as_me:$LINENO: error: conditional \"am__fastdepCXX\" was never defined.
-Usually this means the macro was only invoked conditionally." >&5
-$as_echo "$as_me: error: conditional \"am__fastdepCXX\" was never defined.
-Usually this means the macro was only invoked conditionally." >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "conditional \"am__fastdepCXX\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 if test -z "${am__fastdepCC_TRUE}" && test -z "${am__fastdepCC_FALSE}"; then
-  { { $as_echo "$as_me:$LINENO: error: conditional \"am__fastdepCC\" was never defined.
-Usually this means the macro was only invoked conditionally." >&5
-$as_echo "$as_me: error: conditional \"am__fastdepCC\" was never defined.
-Usually this means the macro was only invoked conditionally." >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "conditional \"am__fastdepCC\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 
-: ${CONFIG_STATUS=./config.status}
+: "${CONFIG_STATUS=./config.status}"
 ac_write_fail=0
 ac_clean_files_save=$ac_clean_files
 ac_clean_files="$ac_clean_files $CONFIG_STATUS"
-{ $as_echo "$as_me:$LINENO: creating $CONFIG_STATUS" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: creating $CONFIG_STATUS" >&5
 $as_echo "$as_me: creating $CONFIG_STATUS" >&6;}
-cat >$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+as_write_fail=0
+cat >$CONFIG_STATUS <<_ASEOF || as_write_fail=1
 #! $SHELL
 # Generated by $as_me.
 # Run this file to recreate the current configuration.
@@ -5832,17 +5226,18 @@ cat >$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 debug=false
 ac_cs_recheck=false
 ac_cs_silent=false
-SHELL=\${CONFIG_SHELL-$SHELL}
-_ACEOF
 
-cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-## --------------------- ##
-## M4sh Initialization.  ##
-## --------------------- ##
+SHELL=\${CONFIG_SHELL-$SHELL}
+export SHELL
+_ASEOF
+cat >>$CONFIG_STATUS <<\_ASEOF || as_write_fail=1
+## -------------------- ##
+## M4sh Initialization. ##
+## -------------------- ##
 
 # Be more Bourne compatible
 DUALCASE=1; export DUALCASE # for MKS sh
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
   emulate sh
   NULLCMD=:
   # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
@@ -5850,23 +5245,15 @@ if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
   alias -g '${1+"$@"}'='"$@"'
   setopt NO_GLOB_SUBST
 else
-  case `(set -o) 2>/dev/null` in
-  *posix*) set -o posix ;;
+  case `(set -o) 2>/dev/null` in #(
+  *posix*) :
+    set -o posix ;; #(
+  *) :
+     ;;
 esac
-
 fi
 
 
-
-
-# PATH needs CR
-# Avoid depending upon Character Ranges.
-as_cr_letters='abcdefghijklmnopqrstuvwxyz'
-as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
-as_cr_Letters=$as_cr_letters$as_cr_LETTERS
-as_cr_digits='0123456789'
-as_cr_alnum=$as_cr_Letters$as_cr_digits
-
 as_nl='
 '
 export as_nl
@@ -5874,7 +5261,13 @@ export as_nl
 as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
-if (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+# Prefer a ksh shell builtin over an external printf program on Solaris,
+# but without wasting forks for bash or zsh.
+if test -z "$BASH_VERSION$ZSH_VERSION" \
+    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
+  as_echo='print -r --'
+  as_echo_n='print -rn --'
+elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
   as_echo='printf %s\n'
   as_echo_n='printf %s'
 else
@@ -5885,7 +5278,7 @@ else
     as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
     as_echo_n_body='eval
       arg=$1;
-      case $arg in
+      case $arg in #(
       *"$as_nl"*)
 	expr "X$arg" : "X\\(.*\\)$as_nl";
 	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
@@ -5908,13 +5301,6 @@ if test "${PATH_SEPARATOR+set}" != set; then
   }
 fi
 
-# Support unset when possible.
-if ( (MAIL=60; unset MAIL) || exit) >/dev/null 2>&1; then
-  as_unset=unset
-else
-  as_unset=false
-fi
-
 
 # IFS
 # We need space, tab and new line, in precisely that order.  Quoting is
@@ -5924,15 +5310,16 @@ fi
 IFS=" ""	$as_nl"
 
 # Find who we are.  Look in the path if we contain no directory separator.
-case $0 in
+as_myself=
+case $0 in #((
   *[\\/]* ) as_myself=$0 ;;
   *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
 for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
-done
+    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+  done
 IFS=$as_save_IFS
 
      ;;
@@ -5944,12 +5331,16 @@ if test "x$as_myself" = x; then
 fi
 if test ! -f "$as_myself"; then
   $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
-  { (exit 1); exit 1; }
+  exit 1
 fi
 
-# Work around bugs in pre-3.0 UWIN ksh.
-for as_var in ENV MAIL MAILPATH
-do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
+# Unset variables that we do not need and which cause bugs (e.g. in
+# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
+# suppresses any "Segmentation fault" message there.  '((' could
+# trigger a bug in pdksh 5.2.14.
+for as_var in BASH_ENV ENV MAIL MAILPATH
+do eval test x\${$as_var+set} = xset \
+  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
 done
 PS1='$ '
 PS2='> '
@@ -5961,7 +5352,89 @@ export LC_ALL
 LANGUAGE=C
 export LANGUAGE
 
-# Required to use basename.
+# CDPATH.
+(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
+
+
+# as_fn_error STATUS ERROR [LINENO LOG_FD]
+# ----------------------------------------
+# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
+# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
+# script with STATUS, using 1 if that was 0.
+as_fn_error ()
+{
+  as_status=$1; test $as_status -eq 0 && as_status=1
+  if test "$4"; then
+    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
+  fi
+  $as_echo "$as_me: error: $2" >&2
+  as_fn_exit $as_status
+} # as_fn_error
+
+
+# as_fn_set_status STATUS
+# -----------------------
+# Set $? to STATUS, without forking.
+as_fn_set_status ()
+{
+  return $1
+} # as_fn_set_status
+
+# as_fn_exit STATUS
+# -----------------
+# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
+as_fn_exit ()
+{
+  set +e
+  as_fn_set_status $1
+  exit $1
+} # as_fn_exit
+
+# as_fn_unset VAR
+# ---------------
+# Portably unset VAR.
+as_fn_unset ()
+{
+  { eval $1=; unset $1;}
+}
+as_unset=as_fn_unset
+# as_fn_append VAR VALUE
+# ----------------------
+# Append the text in VALUE to the end of the definition contained in VAR. Take
+# advantage of any shell optimizations that allow amortized linear growth over
+# repeated appends, instead of the typical quadratic growth present in naive
+# implementations.
+if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
+  eval 'as_fn_append ()
+  {
+    eval $1+=\$2
+  }'
+else
+  as_fn_append ()
+  {
+    eval $1=\$$1\$2
+  }
+fi # as_fn_append
+
+# as_fn_arith ARG...
+# ------------------
+# Perform arithmetic evaluation on the ARGs, and store the result in the
+# global $as_val. Take advantage of shells that can avoid forks. The arguments
+# must be portable across $(()) and expr.
+if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
+  eval 'as_fn_arith ()
+  {
+    as_val=$(( $* ))
+  }'
+else
+  as_fn_arith ()
+  {
+    as_val=`expr "$@" || test $? -eq 1`
+  }
+fi # as_fn_arith
+
+
 if expr a : '\(a\)' >/dev/null 2>&1 &&
    test "X`expr 00001 : '.*\(...\)'`" = X001; then
   as_expr=expr
@@ -5975,8 +5448,12 @@ else
   as_basename=false
 fi
 
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
 
-# Name of the executable.
 as_me=`$as_basename -- "$0" ||
 $as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
 	 X"$0" : 'X\(//\)$' \| \
@@ -5996,76 +5473,25 @@ $as_echo X/"$0" |
 	  }
 	  s/.*/./; q'`
 
-# CDPATH.
-$as_unset CDPATH
-
-
-
-  as_lineno_1=$LINENO
-  as_lineno_2=$LINENO
-  test "x$as_lineno_1" != "x$as_lineno_2" &&
-  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2" || {
-
-  # Create $as_me.lineno as a copy of $as_myself, but with $LINENO
-  # uniformly replaced by the line number.  The first 'sed' inserts a
-  # line-number line after each line using $LINENO; the second 'sed'
-  # does the real work.  The second script uses 'N' to pair each
-  # line-number line with the line containing $LINENO, and appends
-  # trailing '-' during substitution so that $LINENO is not a special
-  # case at line end.
-  # (Raja R Harinath suggested sed '=', and Paul Eggert wrote the
-  # scripts with optimization help from Paolo Bonzini.  Blame Lee
-  # E. McMahon (1931-1989) for sed's syntax.  :-)
-  sed -n '
-    p
-    /[$]LINENO/=
-  ' <$as_myself |
-    sed '
-      s/[$]LINENO.*/&-/
-      t lineno
-      b
-      :lineno
-      N
-      :loop
-      s/[$]LINENO\([^'$as_cr_alnum'_].*\n\)\(.*\)/\2\1\2/
-      t loop
-      s/-\n.*//
-    ' >$as_me.lineno &&
-  chmod +x "$as_me.lineno" ||
-    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2
-   { (exit 1); exit 1; }; }
-
-  # Don't try to exec as it changes $[0], causing all sort of problems
-  # (the dirname of $[0] is not the place where we might find the
-  # original and so on.  Autoconf is especially sensitive to this).
-  . "./$as_me.lineno"
-  # Exit status is that of the last command.
-  exit
-}
-
-
-if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
-  as_dirname=dirname
-else
-  as_dirname=false
-fi
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
 
 ECHO_C= ECHO_N= ECHO_T=
-case `echo -n x` in
+case `echo -n x` in #(((((
 -n*)
-  case `echo 'x\c'` in
+  case `echo 'xy\c'` in
   *c*) ECHO_T='	';;	# ECHO_T is single tab character.
-  *)   ECHO_C='\c';;
+  xy)  ECHO_C='\c';;
+  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
+       ECHO_T='	';;
   esac;;
 *)
   ECHO_N='-n';;
 esac
-if expr a : '\(a\)' >/dev/null 2>&1 &&
-   test "X`expr 00001 : '.*\(...\)'`" = X001; then
-  as_expr=expr
-else
-  as_expr=false
-fi
 
 rm -f conf$$ conf$$.exe conf$$.file
 if test -d conf$$.dir; then
@@ -6080,49 +5506,85 @@ if (echo >conf$$.file) 2>/dev/null; then
     # ... but there are two gotchas:
     # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
     # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
-    # In both cases, we have to default to `cp -p'.
+    # In both cases, we have to default to `cp -pR'.
     ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
-      as_ln_s='cp -p'
+      as_ln_s='cp -pR'
   elif ln conf$$.file conf$$ 2>/dev/null; then
     as_ln_s=ln
   else
-    as_ln_s='cp -p'
+    as_ln_s='cp -pR'
   fi
 else
-  as_ln_s='cp -p'
+  as_ln_s='cp -pR'
 fi
 rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
 rmdir conf$$.dir 2>/dev/null
 
+
+# as_fn_mkdir_p
+# -------------
+# Create "$as_dir" as a directory, including parents if necessary.
+as_fn_mkdir_p ()
+{
+
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || eval $as_mkdir_p || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
+
+
+} # as_fn_mkdir_p
 if mkdir -p . 2>/dev/null; then
-  as_mkdir_p=:
+  as_mkdir_p='mkdir -p "$as_dir"'
 else
   test -d ./-p && rmdir ./-p
   as_mkdir_p=false
 fi
 
-if test -x / >/dev/null 2>&1; then
-  as_test_x='test -x'
-else
-  if ls -dL / >/dev/null 2>&1; then
-    as_ls_L_option=L
-  else
-    as_ls_L_option=
-  fi
-  as_test_x='
-    eval sh -c '\''
-      if test -d "$1"; then
-	test -d "$1/.";
-      else
-	case $1 in
-	-*)set "./$1";;
-	esac;
-	case `ls -ld'$as_ls_L_option' "$1" 2>/dev/null` in
-	???[sx]*):;;*)false;;esac;fi
-    '\'' sh
-  '
-fi
-as_executable_p=$as_test_x
+
+# as_fn_executable_p FILE
+# -----------------------
+# Test if FILE is an executable regular file.
+as_fn_executable_p ()
+{
+  test -f "$1" && test -x "$1"
+} # as_fn_executable_p
+as_test_x='test -x'
+as_executable_p=as_fn_executable_p
 
 # Sed expression to map a string onto a valid CPP name.
 as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
@@ -6132,13 +5594,19 @@ as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
 
 
 exec 6>&1
+## ----------------------------------- ##
+## Main body of $CONFIG_STATUS script. ##
+## ----------------------------------- ##
+_ASEOF
+test $as_write_fail = 0 && chmod +x $CONFIG_STATUS || ac_write_fail=1
 
-# Save the log message, to keep $[0] and so on meaningful, and to
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+# Save the log message, to keep $0 and so on meaningful, and to
 # report actual input values of CONFIG_FILES etc. instead of their
 # values after options handling.
 ac_log="
 This file was extended by Grid $as_me 1.0, which was
-generated by GNU Autoconf 2.63.  Invocation command line was
+generated by GNU Autoconf 2.69.  Invocation command line was
 
   CONFIG_FILES    = $CONFIG_FILES
   CONFIG_HEADERS  = $CONFIG_HEADERS
@@ -6170,13 +5638,15 @@ _ACEOF
 
 cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 ac_cs_usage="\
-\`$as_me' instantiates files from templates according to the
-current configuration.
+\`$as_me' instantiates files and other configuration actions
+from templates according to the current configuration.  Unless the files
+and actions are specified as TAGs, all are instantiated by default.
 
-Usage: $0 [OPTION]... [FILE]...
+Usage: $0 [OPTION]... [TAG]...
 
   -h, --help       print this help, then exit
   -V, --version    print version number and configuration settings, then exit
+      --config     print configuration, then exit
   -q, --quiet, --silent
                    do not print progress messages
   -d, --debug      don't remove temporary files
@@ -6195,16 +5665,17 @@ $config_headers
 Configuration commands:
 $config_commands
 
-Report bugs to <bug-autoconf@gnu.org>."
+Report bugs to <paboyle@ph.ed.ac.uk>."
 
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+ac_cs_config="`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`"
 ac_cs_version="\\
 Grid config.status 1.0
-configured by $0, generated by GNU Autoconf 2.63,
-  with options \\"`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`\\"
+configured by $0, generated by GNU Autoconf 2.69,
+  with options \\"\$ac_cs_config\\"
 
-Copyright (C) 2008 Free Software Foundation, Inc.
+Copyright (C) 2012 Free Software Foundation, Inc.
 This config.status script is free software; the Free Software Foundation
 gives unlimited permission to copy, distribute and modify it."
 
@@ -6222,11 +5693,16 @@ ac_need_defaults=:
 while test $# != 0
 do
   case $1 in
-  --*=*)
+  --*=?*)
     ac_option=`expr "X$1" : 'X\([^=]*\)='`
     ac_optarg=`expr "X$1" : 'X[^=]*=\(.*\)'`
     ac_shift=:
     ;;
+  --*=)
+    ac_option=`expr "X$1" : 'X\([^=]*\)='`
+    ac_optarg=
+    ac_shift=:
+    ;;
   *)
     ac_option=$1
     ac_optarg=$2
@@ -6240,27 +5716,29 @@ do
     ac_cs_recheck=: ;;
   --version | --versio | --versi | --vers | --ver | --ve | --v | -V )
     $as_echo "$ac_cs_version"; exit ;;
+  --config | --confi | --conf | --con | --co | --c )
+    $as_echo "$ac_cs_config"; exit ;;
   --debug | --debu | --deb | --de | --d | -d )
     debug=: ;;
   --file | --fil | --fi | --f )
     $ac_shift
     case $ac_optarg in
     *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
+    '') as_fn_error $? "missing file argument" ;;
     esac
-    CONFIG_FILES="$CONFIG_FILES '$ac_optarg'"
+    as_fn_append CONFIG_FILES " '$ac_optarg'"
     ac_need_defaults=false;;
   --header | --heade | --head | --hea )
     $ac_shift
     case $ac_optarg in
     *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
     esac
-    CONFIG_HEADERS="$CONFIG_HEADERS '$ac_optarg'"
+    as_fn_append CONFIG_HEADERS " '$ac_optarg'"
     ac_need_defaults=false;;
   --he | --h)
     # Conflict between --help and --header
-    { $as_echo "$as_me: error: ambiguous option: $1
-Try \`$0 --help' for more information." >&2
-   { (exit 1); exit 1; }; };;
+    as_fn_error $? "ambiguous option: \`$1'
+Try \`$0 --help' for more information.";;
   --help | --hel | -h )
     $as_echo "$ac_cs_usage"; exit ;;
   -q | -quiet | --quiet | --quie | --qui | --qu | --q \
@@ -6268,11 +5746,10 @@ Try \`$0 --help' for more information." >&2
     ac_cs_silent=: ;;
 
   # This is an error.
-  -*) { $as_echo "$as_me: error: unrecognized option: $1
-Try \`$0 --help' for more information." >&2
-   { (exit 1); exit 1; }; } ;;
+  -*) as_fn_error $? "unrecognized option: \`$1'
+Try \`$0 --help' for more information." ;;
 
-  *) ac_config_targets="$ac_config_targets $1"
+  *) as_fn_append ac_config_targets " $1"
      ac_need_defaults=false ;;
 
   esac
@@ -6289,7 +5766,7 @@ fi
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 if \$ac_cs_recheck; then
-  set X '$SHELL' '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
+  set X $SHELL '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
   shift
   \$as_echo "running CONFIG_SHELL=$SHELL \$*" >&6
   CONFIG_SHELL='$SHELL'
@@ -6327,9 +5804,7 @@ do
     "depfiles") CONFIG_COMMANDS="$CONFIG_COMMANDS depfiles" ;;
     "Makefile") CONFIG_FILES="$CONFIG_FILES Makefile" ;;
 
-  *) { { $as_echo "$as_me:$LINENO: error: invalid argument: $ac_config_target" >&5
-$as_echo "$as_me: error: invalid argument: $ac_config_target" >&2;}
-   { (exit 1); exit 1; }; };;
+  *) as_fn_error $? "invalid argument: \`$ac_config_target'" "$LINENO" 5;;
   esac
 done
 
@@ -6352,26 +5827,24 @@ fi
 # after its creation but before its name has been assigned to `$tmp'.
 $debug ||
 {
-  tmp=
+  tmp= ac_tmp=
   trap 'exit_status=$?
-  { test -z "$tmp" || test ! -d "$tmp" || rm -fr "$tmp"; } && exit $exit_status
+  : "${ac_tmp:=$tmp}"
+  { test ! -d "$ac_tmp" || rm -fr "$ac_tmp"; } && exit $exit_status
 ' 0
-  trap '{ (exit 1); exit 1; }' 1 2 13 15
+  trap 'as_fn_exit 1' 1 2 13 15
 }
 # Create a (secure) tmp directory for tmp files.
 
 {
   tmp=`(umask 077 && mktemp -d "./confXXXXXX") 2>/dev/null` &&
-  test -n "$tmp" && test -d "$tmp"
+  test -d "$tmp"
 }  ||
 {
   tmp=./conf$$-$RANDOM
   (umask 077 && mkdir "$tmp")
-} ||
-{
-   $as_echo "$as_me: cannot create a temporary directory in ." >&2
-   { (exit 1); exit 1; }
-}
+} || as_fn_error $? "cannot create a temporary directory in ." "$LINENO" 5
+ac_tmp=$tmp
 
 # Set up the scripts for CONFIG_FILES section.
 # No need to generate them if there are no CONFIG_FILES.
@@ -6379,7 +5852,13 @@ $debug ||
 if test -n "$CONFIG_FILES"; then
 
 
-ac_cr='
'
+ac_cr=`echo X | tr X '\015'`
+# On cygwin, bash can eat \r inside `` if the user requested igncr.
+# But we know of no other shell where ac_cr would be empty at this
+# point, so we can use a bashism as a fallback.
+if test "x$ac_cr" = x; then
+  eval ac_cr=\$\'\\r\'
+fi
 ac_cs_awk_cr=`$AWK 'BEGIN { print "a\rb" }' </dev/null 2>/dev/null`
 if test "$ac_cs_awk_cr" = "a${ac_cr}b"; then
   ac_cs_awk_cr='\\r'
@@ -6387,7 +5866,7 @@ else
   ac_cs_awk_cr=$ac_cr
 fi
 
-echo 'BEGIN {' >"$tmp/subs1.awk" &&
+echo 'BEGIN {' >"$ac_tmp/subs1.awk" &&
 _ACEOF
 
 
@@ -6396,24 +5875,18 @@ _ACEOF
   echo "$ac_subst_vars" | sed 's/.*/&!$&$ac_delim/' &&
   echo "_ACEOF"
 } >conf$$subs.sh ||
-  { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
-$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
-   { (exit 1); exit 1; }; }
-ac_delim_num=`echo "$ac_subst_vars" | grep -c '$'`
+  as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
+ac_delim_num=`echo "$ac_subst_vars" | grep -c '^'`
 ac_delim='%!_!# '
 for ac_last_try in false false false false false :; do
   . ./conf$$subs.sh ||
-    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
-$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
 
   ac_delim_n=`sed -n "s/.*$ac_delim\$/X/p" conf$$subs.awk | grep -c X`
   if test $ac_delim_n = $ac_delim_num; then
     break
   elif $ac_last_try; then
-    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
-$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
   else
     ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
   fi
@@ -6421,7 +5894,7 @@ done
 rm -f conf$$subs.sh
 
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-cat >>"\$tmp/subs1.awk" <<\\_ACAWK &&
+cat >>"\$ac_tmp/subs1.awk" <<\\_ACAWK &&
 _ACEOF
 sed -n '
 h
@@ -6435,7 +5908,7 @@ s/'"$ac_delim"'$//
 t delim
 :nl
 h
-s/\(.\{148\}\).*/\1/
+s/\(.\{148\}\)..*/\1/
 t more1
 s/["\\]/\\&/g; s/^/"/; s/$/\\n"\\/
 p
@@ -6449,7 +5922,7 @@ s/.\{148\}//
 t nl
 :delim
 h
-s/\(.\{148\}\).*/\1/
+s/\(.\{148\}\)..*/\1/
 t more2
 s/["\\]/\\&/g; s/^/"/; s/$/"/
 p
@@ -6469,7 +5942,7 @@ t delim
 rm -f conf$$subs.awk
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 _ACAWK
-cat >>"\$tmp/subs1.awk" <<_ACAWK &&
+cat >>"\$ac_tmp/subs1.awk" <<_ACAWK &&
   for (key in S) S_is_set[key] = 1
   FS = ""
 
@@ -6501,23 +5974,29 @@ if sed "s/$ac_cr//" < /dev/null > /dev/null 2>&1; then
   sed "s/$ac_cr\$//; s/$ac_cr/$ac_cs_awk_cr/g"
 else
   cat
-fi < "$tmp/subs1.awk" > "$tmp/subs.awk" \
-  || { { $as_echo "$as_me:$LINENO: error: could not setup config files machinery" >&5
-$as_echo "$as_me: error: could not setup config files machinery" >&2;}
-   { (exit 1); exit 1; }; }
+fi < "$ac_tmp/subs1.awk" > "$ac_tmp/subs.awk" \
+  || as_fn_error $? "could not setup config files machinery" "$LINENO" 5
 _ACEOF
 
-# VPATH may cause trouble with some makes, so we remove $(srcdir),
-# ${srcdir} and @srcdir@ from VPATH if srcdir is ".", strip leading and
+# VPATH may cause trouble with some makes, so we remove sole $(srcdir),
+# ${srcdir} and @srcdir@ entries from VPATH if srcdir is ".", strip leading and
 # trailing colons and then remove the whole line if VPATH becomes empty
 # (actually we leave an empty line to preserve line numbers).
 if test "x$srcdir" = x.; then
-  ac_vpsub='/^[	 ]*VPATH[	 ]*=/{
-s/:*\$(srcdir):*/:/
-s/:*\${srcdir}:*/:/
-s/:*@srcdir@:*/:/
-s/^\([^=]*=[	 ]*\):*/\1/
+  ac_vpsub='/^[	 ]*VPATH[	 ]*=[	 ]*/{
+h
+s///
+s/^/:/
+s/[	 ]*$/:/
+s/:\$(srcdir):/:/g
+s/:\${srcdir}:/:/g
+s/:@srcdir@:/:/g
+s/^:*//
 s/:*$//
+x
+s/\(=[	 ]*\).*/\1/
+G
+s/\n//
 s/^[^=]*=[	 ]*$//
 }'
 fi
@@ -6529,7 +6008,7 @@ fi # test -n "$CONFIG_FILES"
 # No need to generate them if there are no CONFIG_HEADERS.
 # This happens for instance with `./config.status Makefile'.
 if test -n "$CONFIG_HEADERS"; then
-cat >"$tmp/defines.awk" <<\_ACAWK ||
+cat >"$ac_tmp/defines.awk" <<\_ACAWK ||
 BEGIN {
 _ACEOF
 
@@ -6541,13 +6020,11 @@ _ACEOF
 # handling of long lines.
 ac_delim='%!_!# '
 for ac_last_try in false false :; do
-  ac_t=`sed -n "/$ac_delim/p" confdefs.h`
-  if test -z "$ac_t"; then
+  ac_tt=`sed -n "/$ac_delim/p" confdefs.h`
+  if test -z "$ac_tt"; then
     break
   elif $ac_last_try; then
-    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_HEADERS" >&5
-$as_echo "$as_me: error: could not make $CONFIG_HEADERS" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "could not make $CONFIG_HEADERS" "$LINENO" 5
   else
     ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
   fi
@@ -6632,9 +6109,7 @@ cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 _ACAWK
 _ACEOF
 cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-  { { $as_echo "$as_me:$LINENO: error: could not setup config headers machinery" >&5
-$as_echo "$as_me: error: could not setup config headers machinery" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "could not setup config headers machinery" "$LINENO" 5
 fi # test -n "$CONFIG_HEADERS"
 
 
@@ -6647,9 +6122,7 @@ do
   esac
   case $ac_mode$ac_tag in
   :[FHL]*:*);;
-  :L* | :C*:*) { { $as_echo "$as_me:$LINENO: error: invalid tag $ac_tag" >&5
-$as_echo "$as_me: error: invalid tag $ac_tag" >&2;}
-   { (exit 1); exit 1; }; };;
+  :L* | :C*:*) as_fn_error $? "invalid tag \`$ac_tag'" "$LINENO" 5;;
   :[FH]-) ac_tag=-:-;;
   :[FH]*) ac_tag=$ac_tag:$ac_tag.in;;
   esac
@@ -6668,7 +6141,7 @@ $as_echo "$as_me: error: invalid tag $ac_tag" >&2;}
     for ac_f
     do
       case $ac_f in
-      -) ac_f="$tmp/stdin";;
+      -) ac_f="$ac_tmp/stdin";;
       *) # Look for the file first in the build tree, then in the source tree
 	 # (if the path is not absolute).  The absolute path cannot be DOS-style,
 	 # because $ac_f cannot contain `:'.
@@ -6677,12 +6150,10 @@ $as_echo "$as_me: error: invalid tag $ac_tag" >&2;}
 	   [\\/$]*) false;;
 	   *) test -f "$srcdir/$ac_f" && ac_f="$srcdir/$ac_f";;
 	   esac ||
-	   { { $as_echo "$as_me:$LINENO: error: cannot find input file: $ac_f" >&5
-$as_echo "$as_me: error: cannot find input file: $ac_f" >&2;}
-   { (exit 1); exit 1; }; };;
+	   as_fn_error 1 "cannot find input file: \`$ac_f'" "$LINENO" 5;;
       esac
       case $ac_f in *\'*) ac_f=`$as_echo "$ac_f" | sed "s/'/'\\\\\\\\''/g"`;; esac
-      ac_file_inputs="$ac_file_inputs '$ac_f'"
+      as_fn_append ac_file_inputs " '$ac_f'"
     done
 
     # Let's still pretend it is `configure' which instantiates (i.e., don't
@@ -6693,7 +6164,7 @@ $as_echo "$as_me: error: cannot find input file: $ac_f" >&2;}
 	`' by configure.'
     if test x"$ac_file" != x-; then
       configure_input="$ac_file.  $configure_input"
-      { $as_echo "$as_me:$LINENO: creating $ac_file" >&5
+      { $as_echo "$as_me:${as_lineno-$LINENO}: creating $ac_file" >&5
 $as_echo "$as_me: creating $ac_file" >&6;}
     fi
     # Neutralize special characters interpreted by sed in replacement strings.
@@ -6705,10 +6176,8 @@ $as_echo "$as_me: creating $ac_file" >&6;}
     esac
 
     case $ac_tag in
-    *:-:* | *:-) cat >"$tmp/stdin" \
-      || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; } ;;
+    *:-:* | *:-) cat >"$ac_tmp/stdin" \
+      || as_fn_error $? "could not create $ac_file" "$LINENO" 5 ;;
     esac
     ;;
   esac
@@ -6736,47 +6205,7 @@ $as_echo X"$ac_file" |
 	    q
 	  }
 	  s/.*/./; q'`
-  { as_dir="$ac_dir"
-  case $as_dir in #(
-  -*) as_dir=./$as_dir;;
-  esac
-  test -d "$as_dir" || { $as_mkdir_p && mkdir -p "$as_dir"; } || {
-    as_dirs=
-    while :; do
-      case $as_dir in #(
-      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
-      *) as_qdir=$as_dir;;
-      esac
-      as_dirs="'$as_qdir' $as_dirs"
-      as_dir=`$as_dirname -- "$as_dir" ||
-$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_dir" : 'X\(//\)[^/]' \| \
-	 X"$as_dir" : 'X\(//\)$' \| \
-	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_dir" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      test -d "$as_dir" && break
-    done
-    test -z "$as_dirs" || eval "mkdir $as_dirs"
-  } || test -d "$as_dir" || { { $as_echo "$as_me:$LINENO: error: cannot create directory $as_dir" >&5
-$as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
-   { (exit 1); exit 1; }; }; }
+  as_dir="$ac_dir"; as_fn_mkdir_p
   ac_builddir=.
 
 case "$ac_dir" in
@@ -6833,7 +6262,6 @@ cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 # If the template does not know about datarootdir, expand it.
 # FIXME: This hack should be removed a few years after 2.60.
 ac_datarootdir_hack=; ac_datarootdir_seen=
-
 ac_sed_dataroot='
 /datarootdir/ {
   p
@@ -6843,12 +6271,11 @@ ac_sed_dataroot='
 /@docdir@/p
 /@infodir@/p
 /@localedir@/p
-/@mandir@/p
-'
+/@mandir@/p'
 case `eval "sed -n \"\$ac_sed_dataroot\" $ac_file_inputs"` in
 *datarootdir*) ac_datarootdir_seen=yes;;
 *@datadir@*|*@docdir@*|*@infodir@*|*@localedir@*|*@mandir@*)
-  { $as_echo "$as_me:$LINENO: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
 $as_echo "$as_me: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&2;}
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
@@ -6858,7 +6285,7 @@ cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
   s&@infodir@&$infodir&g
   s&@localedir@&$localedir&g
   s&@mandir@&$mandir&g
-    s&\\\${datarootdir}&$datarootdir&g' ;;
+  s&\\\${datarootdir}&$datarootdir&g' ;;
 esac
 _ACEOF
 
@@ -6886,27 +6313,24 @@ s&@INSTALL@&$ac_INSTALL&;t t
 s&@MKDIR_P@&$ac_MKDIR_P&;t t
 $ac_datarootdir_hack
 "
-eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$tmp/subs.awk" >$tmp/out \
-  || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; }
+eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$ac_tmp/subs.awk" \
+  >$ac_tmp/out || as_fn_error $? "could not create $ac_file" "$LINENO" 5
 
 test -z "$ac_datarootdir_hack$ac_datarootdir_seen" &&
-  { ac_out=`sed -n '/\${datarootdir}/p' "$tmp/out"`; test -n "$ac_out"; } &&
-  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' "$tmp/out"`; test -z "$ac_out"; } &&
-  { $as_echo "$as_me:$LINENO: WARNING: $ac_file contains a reference to the variable \`datarootdir'
-which seems to be undefined.  Please make sure it is defined." >&5
+  { ac_out=`sed -n '/\${datarootdir}/p' "$ac_tmp/out"`; test -n "$ac_out"; } &&
+  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' \
+      "$ac_tmp/out"`; test -z "$ac_out"; } &&
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file contains a reference to the variable \`datarootdir'
+which seems to be undefined.  Please make sure it is defined" >&5
 $as_echo "$as_me: WARNING: $ac_file contains a reference to the variable \`datarootdir'
-which seems to be undefined.  Please make sure it is defined." >&2;}
+which seems to be undefined.  Please make sure it is defined" >&2;}
 
-  rm -f "$tmp/stdin"
+  rm -f "$ac_tmp/stdin"
   case $ac_file in
-  -) cat "$tmp/out" && rm -f "$tmp/out";;
-  *) rm -f "$ac_file" && mv "$tmp/out" "$ac_file";;
+  -) cat "$ac_tmp/out" && rm -f "$ac_tmp/out";;
+  *) rm -f "$ac_file" && mv "$ac_tmp/out" "$ac_file";;
   esac \
-  || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; }
+  || as_fn_error $? "could not create $ac_file" "$LINENO" 5
  ;;
   :H)
   #
@@ -6915,27 +6339,21 @@ $as_echo "$as_me: error: could not create $ac_file" >&2;}
   if test x"$ac_file" != x-; then
     {
       $as_echo "/* $configure_input  */" \
-      && eval '$AWK -f "$tmp/defines.awk"' "$ac_file_inputs"
-    } >"$tmp/config.h" \
-      || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; }
-    if diff "$ac_file" "$tmp/config.h" >/dev/null 2>&1; then
-      { $as_echo "$as_me:$LINENO: $ac_file is unchanged" >&5
+      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs"
+    } >"$ac_tmp/config.h" \
+      || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+    if diff "$ac_file" "$ac_tmp/config.h" >/dev/null 2>&1; then
+      { $as_echo "$as_me:${as_lineno-$LINENO}: $ac_file is unchanged" >&5
 $as_echo "$as_me: $ac_file is unchanged" >&6;}
     else
       rm -f "$ac_file"
-      mv "$tmp/config.h" "$ac_file" \
-	|| { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; }
+      mv "$ac_tmp/config.h" "$ac_file" \
+	|| as_fn_error $? "could not create $ac_file" "$LINENO" 5
     fi
   else
     $as_echo "/* $configure_input  */" \
-      && eval '$AWK -f "$tmp/defines.awk"' "$ac_file_inputs" \
-      || { { $as_echo "$as_me:$LINENO: error: could not create -" >&5
-$as_echo "$as_me: error: could not create -" >&2;}
-   { (exit 1); exit 1; }; }
+      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs" \
+      || as_fn_error $? "could not create -" "$LINENO" 5
   fi
 # Compute "$ac_file"'s index in $config_headers.
 _am_arg="$ac_file"
@@ -6973,7 +6391,7 @@ $as_echo X"$_am_arg" |
 	  s/.*/./; q'`/stamp-h$_am_stamp_count
  ;;
 
-  :C)  { $as_echo "$as_me:$LINENO: executing $ac_file commands" >&5
+  :C)  { $as_echo "$as_me:${as_lineno-$LINENO}: executing $ac_file commands" >&5
 $as_echo "$as_me: executing $ac_file commands" >&6;}
  ;;
   esac
@@ -6981,7 +6399,7 @@ $as_echo "$as_me: executing $ac_file commands" >&6;}
 
   case $ac_file$ac_mode in
     "depfiles":C) test x"$AMDEP_TRUE" != x"" || {
-  # Autoconf 2.62 quotes --file arguments for eval, but not when files
+  # Older Autoconf quotes --file arguments for eval, but not when files
   # are listed without --file.  Let's play safe and only enable the eval
   # if we detect the quoting.
   case $CONFIG_FILES in
@@ -6994,7 +6412,7 @@ $as_echo "$as_me: executing $ac_file commands" >&6;}
     # Strip MF so we end up with the name of the file.
     mf=`echo "$mf" | sed -e 's/:.*$//'`
     # Check whether this is an Automake generated Makefile or not.
-    # We used to match only the files named `Makefile.in', but
+    # We used to match only the files named 'Makefile.in', but
     # some people rename them; so instead we look at the file content.
     # Grep'ing the first line is not enough: some people post-process
     # each Makefile.in and add a new line on top of each file to say so.
@@ -7028,21 +6446,19 @@ $as_echo X"$mf" |
       continue
     fi
     # Extract the definition of DEPDIR, am__include, and am__quote
-    # from the Makefile without running `make'.
+    # from the Makefile without running 'make'.
     DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
     test -z "$DEPDIR" && continue
     am__include=`sed -n 's/^am__include = //p' < "$mf"`
-    test -z "am__include" && continue
+    test -z "$am__include" && continue
     am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
-    # When using ansi2knr, U may be empty or an underscore; expand it
-    U=`sed -n 's/^U = //p' < "$mf"`
     # Find all dependency output files, they are included files with
     # $(DEPDIR) in their names.  We invoke sed twice because it is the
     # simplest approach to changing $(DEPDIR) to its actual value in the
     # expansion.
     for file in `sed -n "
       s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
-	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g' -e 's/\$U/'"$U"'/g'`; do
+	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
       # Make sure the directory exists.
       test -f "$dirpart/$file" && continue
       fdir=`$as_dirname -- "$file" ||
@@ -7068,47 +6484,7 @@ $as_echo X"$file" |
 	    q
 	  }
 	  s/.*/./; q'`
-      { as_dir=$dirpart/$fdir
-  case $as_dir in #(
-  -*) as_dir=./$as_dir;;
-  esac
-  test -d "$as_dir" || { $as_mkdir_p && mkdir -p "$as_dir"; } || {
-    as_dirs=
-    while :; do
-      case $as_dir in #(
-      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
-      *) as_qdir=$as_dir;;
-      esac
-      as_dirs="'$as_qdir' $as_dirs"
-      as_dir=`$as_dirname -- "$as_dir" ||
-$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_dir" : 'X\(//\)[^/]' \| \
-	 X"$as_dir" : 'X\(//\)$' \| \
-	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_dir" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      test -d "$as_dir" && break
-    done
-    test -z "$as_dirs" || eval "mkdir $as_dirs"
-  } || test -d "$as_dir" || { { $as_echo "$as_me:$LINENO: error: cannot create directory $as_dir" >&5
-$as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
-   { (exit 1); exit 1; }; }; }
+      as_dir=$dirpart/$fdir; as_fn_mkdir_p
       # echo "creating $dirpart/$file"
       echo '# dummy' > "$dirpart/$file"
     done
@@ -7120,15 +6496,12 @@ $as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
 done # for ac_tag
 
 
-{ (exit 0); exit 0; }
+as_fn_exit 0
 _ACEOF
-chmod +x $CONFIG_STATUS
 ac_clean_files=$ac_clean_files_save
 
 test $ac_write_fail = 0 ||
-  { { $as_echo "$as_me:$LINENO: error: write failure creating $CONFIG_STATUS" >&5
-$as_echo "$as_me: error: write failure creating $CONFIG_STATUS" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "write failure creating $CONFIG_STATUS" "$LINENO" 5
 
 
 # configure is writing to config.log, and then calls config.status.
@@ -7149,10 +6522,10 @@ if test "$no_create" != yes; then
   exec 5>>config.log
   # Use ||, not &&, to avoid exiting from the if with $? = 1, which
   # would make configure fail if this is the last instruction.
-  $ac_cs_success || { (exit 1); exit 1; }
+  $ac_cs_success || as_fn_exit 1
 fi
 if test -n "$ac_unrecognized_opts" && test "$enable_option_checking" != no; then
-  { $as_echo "$as_me:$LINENO: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
 $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2;}
 fi
 

From 19372bf5ca986660744b30c84afd0a8357275560 Mon Sep 17 00:00:00 2001
From: azusayamaguchi <ayamaguc@YAMAKAZE.local>
Date: Sun, 29 Mar 2015 22:16:13 +0100
Subject: [PATCH 027/429] COnfig file

---
 Grid_config.h    |  4 ++--
 Grid_config.h.in | 12 ++++++++++++
 2 files changed, 14 insertions(+), 2 deletions(-)

diff --git a/Grid_config.h b/Grid_config.h
index 2a442ea5..94ef0ea2 100644
--- a/Grid_config.h
+++ b/Grid_config.h
@@ -11,13 +11,13 @@
 /* #undef AVX512 */
 
 /* GRID_COMMS_FAKE */
-#define GRID_COMMS_FAKE 1
+/* #undef GRID_COMMS_FAKE */
 
 /* GRID_COMMS_MPI */
 /* #undef GRID_COMMS_MPI */
 
 /* GRID_COMMS_NONE */
-/* #undef GRID_COMMS_NONE */
+#define GRID_COMMS_NONE 1
 
 /* Define to 1 if you have the `gettimeofday' function. */
 #define HAVE_GETTIMEOFDAY 1
diff --git a/Grid_config.h.in b/Grid_config.h.in
index 5c57e4cf..ff15a834 100644
--- a/Grid_config.h.in
+++ b/Grid_config.h.in
@@ -9,6 +9,15 @@
 /* AVX512 */
 #undef AVX512
 
+/* GRID_COMMS_FAKE */
+#undef GRID_COMMS_FAKE
+
+/* GRID_COMMS_MPI */
+#undef GRID_COMMS_MPI
+
+/* GRID_COMMS_NONE */
+#undef GRID_COMMS_NONE
+
 /* Define to 1 if you have the `gettimeofday' function. */
 #undef HAVE_GETTIMEOFDAY
 
@@ -60,6 +69,9 @@
 /* Define to the one symbol short name of this package. */
 #undef PACKAGE_TARNAME
 
+/* Define to the home page for this package. */
+#undef PACKAGE_URL
+
 /* Define to the version of this package. */
 #undef PACKAGE_VERSION
 

From 7b97e50b7ba66d1a231aff3e52edbc937459aa7b Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 3 Apr 2015 04:52:53 +0100
Subject: [PATCH 028/429] MPI is now working and passing basic tests. Will
 start to construct a more sensible test suite shortly since testing
 requirements now go beyond what a single Grid_main.cc can do.

Will need a more organised src tree for this and will require substantial reorg of build system.
---
 Grid.h               |   2 +-
 Grid_Cartesian.h     |  27 +-
 Grid_Communicator.h  |  10 +-
 Grid_Lattice.h       |   5 +-
 Grid_config.h        |   4 +-
 Grid_config.h.in     |  12 +
 Grid_cshift_common.h | 384 +++++++++++++++++++++++
 Grid_init.cc         |   5 +-
 Grid_main.cc         |  46 ++-
 Grid_mpi.cc          |  46 +--
 Grid_mpi_cshift.h    | 625 ++++++++++++++++++--------------------
 Grid_none_cshift.h   | 304 +------------------
 Makefile.am          |  14 +-
 Makefile.in          |  16 +-
 aclocal.m4           | 705 +++++++++++++++++++++++++++----------------
 configure            |  43 +++
 configure.ac         |   4 +
 17 files changed, 1298 insertions(+), 954 deletions(-)
 create mode 100644 Grid_cshift_common.h

diff --git a/Grid.h b/Grid.h
index 5a3571e4..cc8b36c6 100644
--- a/Grid.h
+++ b/Grid.h
@@ -52,7 +52,7 @@
 
 namespace dpo {
 
-  void Grid_init(void);
+  void Grid_init(int *argc,char ***argv);
   double usecond(void);
   void Grid_sa_signal_handler(int sig,siginfo_t *si,void * ptr);
   void Grid_debug_handler_init(void);
diff --git a/Grid_Cartesian.h b/Grid_Cartesian.h
index bfe301a9..908d08a8 100644
--- a/Grid_Cartesian.h
+++ b/Grid_Cartesian.h
@@ -41,8 +41,8 @@ public:
     std::vector<int> _simd_layout;     // Which dimensions get relayed out over simd lanes.
 
 
-    std::vector<int> _fdimensions;// Global dimensions of array with cb removed
-    std::vector<int> _gdimensions;// Global dimensions of array
+    std::vector<int> _fdimensions;// Global dimensions of array prior to cb removal
+    std::vector<int> _gdimensions;// Global dimensions of array after cb removal
     std::vector<int> _ldimensions;// local dimensions of array with processor images removed
     std::vector<int> _rdimensions;// Reduced local dimensions with simd lane images and processor images removed 
 
@@ -88,11 +88,20 @@ public:
         for(int d=0;d<_ndimension;d++) idx+=_istride[d]*(rcoor[d]/_rdimensions[d]);
         return idx;
     }
+    inline int iCoordFromIsite(int lane,int mu)
+    {
+      std::vector<int> coor(_ndimension);
+      for(int d=0;d<_ndimension;d++){
+	coor[d] = lane % _simd_layout[d];
+	lane    = lane / _simd_layout[d];
+      }
+      return coor[mu];
+    }
     
     inline int oSites(void) { return _osites; };
     inline int iSites(void) { return _isites; };
 
-    inline int CheckerboardFromOsite (int Osite){
+    inline int CheckerBoardFromOsite (int Osite){
       std::vector<int> ocoor;
       CoordFromOsite(ocoor,Osite);
       int ss=0;
@@ -109,6 +118,7 @@ public:
       }
     }
 
+    virtual int CheckerBoarded(int dim)=0;
     virtual int CheckerBoard(std::vector<int> site)=0;
     virtual int CheckerBoardDestination(int source_cb,int shift)=0;
     virtual int CheckerBoardShift(int source_cb,int dim,int shift,int osite)=0;
@@ -116,6 +126,9 @@ public:
 
 class GridCartesian: public SimdGrid {
 public:
+    virtual int CheckerBoarded(int dim){
+      return 0;
+    }
     virtual int CheckerBoard(std::vector<int> site){
         return 0;
     }
@@ -199,6 +212,10 @@ public:
 class GridRedBlackCartesian : public SimdGrid
 {
 public:
+    virtual int CheckerBoarded(int dim){
+      if( dim==0) return 1;
+      else return 0;
+    }
     virtual int CheckerBoard(std::vector<int> site){
       return (site[0]+site[1]+site[2]+site[3])&0x1;
     }
@@ -215,11 +232,13 @@ public:
 
       // Probably faster with table lookup;
       // or by looping over x,y,z and multiply rather than computing checkerboard.
-      int ocb=CheckerboardFromOsite(osite);
+      int ocb=CheckerBoardFromOsite(osite);
 	  
       if ( (source_cb+ocb)&1 ) {
+	printf("Checkerboard shift %d\n",(shift)/2);
 	return (shift)/2;
       } else {
+	printf("Checkerboard shift %d\n",(shift+1)/2);
 	return (shift+1)/2;
       }
     }
diff --git a/Grid_Communicator.h b/Grid_Communicator.h
index 5be6f26f..19e25167 100644
--- a/Grid_Communicator.h
+++ b/Grid_Communicator.h
@@ -3,12 +3,16 @@
 ///////////////////////////////////
 // Processor layout information
 ///////////////////////////////////
+#ifdef GRID_COMMS_MPI
+#include <mpi.h>
+#endif
 namespace dpo {
 class CartesianCommunicator {
   public:    
 
   // Communicator should know nothing of the physics grid, only processor grid.
 
+    int              _Nprocessors;     // How many in all
     std::vector<int> _processors;      // Which dimensions get relayed out over processors lanes.
     int              _processor;       // linear processor rank
     std::vector<int> _processor_coor;  // linear processor coordinate
@@ -20,7 +24,9 @@ class CartesianCommunicator {
 
     CartesianCommunicator(std::vector<int> &pdimensions_in);
 
+    void ShiftedRanks(int dim,int shift,int & source, int & dest);
     int  Rank(std::vector<int> coor);
+    int  MyRank(void);
     void GlobalSumF(float &);
     void GlobalSumFVector(float *,int N);
 
@@ -28,9 +34,9 @@ class CartesianCommunicator {
     void GlobalSumFVector(double *,int N);
 
     void SendToRecvFrom(void *xmit,
-			std::vector<int> to_coordinate,
+			int xmit_to_rank,
 			void *recv,
-			std::vector<int> from_coordinate,
+			int recv_from_rank,
 			int bytes);
     
 }; 
diff --git a/Grid_Lattice.h b/Grid_Lattice.h
index 8c8dc386..fcff36eb 100644
--- a/Grid_Lattice.h
+++ b/Grid_Lattice.h
@@ -23,7 +23,8 @@ public:
     SimdGrid *_grid;
     int checkerboard;
     std::vector<vobj,alignedAllocator<vobj> > _odata;
-
+    typedef typename vobj::scalar_type scalar_type;
+    typedef typename vobj::vector_type vector_type;
 public:
 
 
@@ -36,6 +37,8 @@ public:
     }
     
 
+#include <Grid_cshift_common.h>
+
 #ifdef GRID_COMMS_NONE
 #include <Grid_none_cshift.h>
 #endif
diff --git a/Grid_config.h b/Grid_config.h
index 2a442ea5..8a0fdc26 100644
--- a/Grid_config.h
+++ b/Grid_config.h
@@ -11,10 +11,10 @@
 /* #undef AVX512 */
 
 /* GRID_COMMS_FAKE */
-#define GRID_COMMS_FAKE 1
+/* #undef GRID_COMMS_FAKE */
 
 /* GRID_COMMS_MPI */
-/* #undef GRID_COMMS_MPI */
+#define GRID_COMMS_MPI 1
 
 /* GRID_COMMS_NONE */
 /* #undef GRID_COMMS_NONE */
diff --git a/Grid_config.h.in b/Grid_config.h.in
index 5c57e4cf..ff15a834 100644
--- a/Grid_config.h.in
+++ b/Grid_config.h.in
@@ -9,6 +9,15 @@
 /* AVX512 */
 #undef AVX512
 
+/* GRID_COMMS_FAKE */
+#undef GRID_COMMS_FAKE
+
+/* GRID_COMMS_MPI */
+#undef GRID_COMMS_MPI
+
+/* GRID_COMMS_NONE */
+#undef GRID_COMMS_NONE
+
 /* Define to 1 if you have the `gettimeofday' function. */
 #undef HAVE_GETTIMEOFDAY
 
@@ -60,6 +69,9 @@
 /* Define to the one symbol short name of this package. */
 #undef PACKAGE_TARNAME
 
+/* Define to the home page for this package. */
+#undef PACKAGE_URL
+
 /* Define to the version of this package. */
 #undef PACKAGE_VERSION
 
diff --git a/Grid_cshift_common.h b/Grid_cshift_common.h
new file mode 100644
index 00000000..b8624826
--- /dev/null
+++ b/Grid_cshift_common.h
@@ -0,0 +1,384 @@
+#ifndef _GRID_CSHIFT_COMMON_H_
+#define _GRID_CSHIFT_COMMON_H_
+
+//////////////////////////////////////////////////////
+// Gather for when there is no need to SIMD split
+//////////////////////////////////////////////////////
+friend void Gather_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllocator<vobj> > &buffer,             int dimension,int plane,int cbmask)
+{
+  int rd = rhs._grid->_rdimensions[dimension];
+
+  //  printf("Gather plane _simple mask %d\n",cbmask);
+
+  if ( !rhs._grid->CheckerBoarded(dimension) ) {
+
+    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+    int o   = 0;                                    // relative offset to base within plane
+    int bo  = 0;                                    // offset in buffer
+
+    // Simple block stride gather of SIMD objects
+#pragma omp parallel for collapse(2)
+    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+	buffer[bo++]=rhs._odata[so+o+b];
+      }
+      o +=rhs._grid->_slice_stride[dimension];
+    }
+
+  } else { 
+
+    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+    int o   = 0;                                      // relative offset to base within plane
+    int bo  = 0;                                      // offset in buffer
+
+    //    int jjj=0;
+#pragma omp parallel for collapse(2)
+    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+
+	int ocb=1<<rhs._grid->CheckerBoardFromOsite(o+b);// Could easily be a table lookup
+	if ( ocb &cbmask ) {
+	  buffer[bo]=rhs._odata[so+o+b];
+	  //	  float * ptr = (float *)& rhs._odata[so+o+b];
+	  //	  if( (cbmask!=3)&&(jjj<8)){
+	  //	    printf("Gather_plane_simple %d %le bo %d\n",so+o+b,*ptr,bo);
+	  //	    jjj++;
+	  //	  }
+	  bo++;
+	}
+
+      }
+      o +=rhs._grid->_slice_stride[dimension];
+    }
+  }
+}
+
+
+//////////////////////////////////////////////////////
+// Gather for when there *is* need to SIMD split
+//////////////////////////////////////////////////////
+friend void Gather_plane_extract(Lattice<vobj> &rhs,std::vector<scalar_type *> pointers,int dimension,int plane,int cbmask)
+{
+  int rd = rhs._grid->_rdimensions[dimension];
+
+  if ( !rhs._grid->CheckerBoarded(dimension) ) {
+
+    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+    int o   = 0;                                    // relative offset to base within plane
+    int bo  = 0;                                    // offset in buffer
+
+    // Simple block stride gather of SIMD objects
+#pragma omp parallel for collapse(2)
+    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+	extract(rhs._odata[so+o+b],pointers);
+      }
+      o +=rhs._grid->_slice_stride[dimension];
+    }
+
+  } else { 
+
+    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+    int o   = 0;                                      // relative offset to base within plane
+    int bo  = 0;                                      // offset in buffer
+    
+#pragma omp parallel for collapse(2)
+    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+
+	int ocb=1<<rhs._grid->CheckerBoardFromOsite(o+b);
+	if ( ocb & cbmask ) {
+	  extract(rhs._odata[so+o+b],pointers);
+	}
+
+      }
+      o +=rhs._grid->_slice_stride[dimension];
+    }
+  }
+}
+
+
+
+//////////////////////////////////////////////////////
+// Scatter for when there is no need to SIMD split
+//////////////////////////////////////////////////////
+friend void Scatter_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllocator<vobj> > &buffer,             int dimension,int plane,int cbmask)
+{
+  int rd = rhs._grid->_rdimensions[dimension];
+
+  if ( !rhs._grid->CheckerBoarded(dimension) ) {
+
+    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+    int o   = 0;                                    // relative offset to base within plane
+    int bo  = 0;                                    // offset in buffer
+
+    // Simple block stride gather of SIMD objects
+#pragma omp parallel for collapse(2)
+    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+	rhs._odata[so+o+b]=buffer[bo++];
+      }
+      o +=rhs._grid->_slice_stride[dimension];
+    }
+
+  } else { 
+
+    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+    int o   = 0;                                      // relative offset to base within plane
+    int bo  = 0;                                      // offset in buffer
+    
+#pragma omp parallel for collapse(2)
+    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+
+	int ocb=1<<rhs._grid->CheckerBoardFromOsite(o+b);// Could easily be a table lookup
+	if ( ocb & cbmask ) {
+	  rhs._odata[so+o+b]=buffer[bo++];
+	}
+
+      }
+      o +=rhs._grid->_slice_stride[dimension];
+    }
+  }
+}
+
+
+//////////////////////////////////////////////////////
+// Scatter for when there *is* need to SIMD split
+//////////////////////////////////////////////////////
+friend void Scatter_plane_merge(Lattice<vobj> &rhs,std::vector<scalar_type *> pointers,int dimension,int plane,int cbmask)
+{
+  int rd = rhs._grid->_rdimensions[dimension];
+
+  if ( !rhs._grid->CheckerBoarded(dimension) ) {
+
+    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+    int o   = 0;                                    // relative offset to base within plane
+    int bo  = 0;                                    // offset in buffer
+
+    // Simple block stride gather of SIMD objects
+#pragma omp parallel for collapse(2)
+    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+	merge(rhs._odata[so+o+b],pointers);
+      }
+      o +=rhs._grid->_slice_stride[dimension];
+    }
+
+  } else { 
+
+    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+    int o   = 0;                                      // relative offset to base within plane
+    int bo  = 0;                                      // offset in buffer
+    
+#pragma omp parallel for collapse(2)
+    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+
+	int ocb=1<<rhs._grid->CheckerBoardFromOsite(o+b);
+	if ( ocb&cbmask ) {
+	  merge(rhs._odata[so+o+b],pointers);
+	}
+
+      }
+      o +=rhs._grid->_slice_stride[dimension];
+    }
+  }
+}
+
+
+//////////////////////////////////////////////////////
+// local to node block strided copies
+//////////////////////////////////////////////////////
+// if lhs is odd, rhs even??
+friend void Copy_plane(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int dimension,int lplane,int rplane,int cbmask)
+{
+  int rd = rhs._grid->_rdimensions[dimension];
+
+  if ( !rhs._grid->CheckerBoarded(dimension) ) {
+
+    int o   = 0;                                     // relative offset to base within plane
+    int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+    int lo  = lplane*lhs._grid->_ostride[dimension]; // offset in buffer
+
+  // Simple block stride gather of SIMD objects
+#pragma omp parallel for collapse(2)
+    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+	lhs._odata[lo+o+b]=rhs._odata[ro+o+b];
+      }
+      o +=rhs._grid->_slice_stride[dimension];
+    }
+
+  } else {
+
+    int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+    int lo  = lplane*lhs._grid->_ostride[dimension]; // base offset for start of plane 
+    int o   = 0;                                     // relative offset to base within plane
+    //    int jjj=0;    
+#pragma omp parallel for collapse(2)
+    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+
+	int ocb=1<<lhs._grid->CheckerBoardFromOsite(o+b);
+
+	if ( ocb&cbmask ) {
+	  lhs._odata[lo+o+b]=rhs._odata[ro+o+b];
+	  //	  float *ptr =(float *) &rhs._odata[ro+o+b];
+	  //	  if((cbmask!=0x3)&&jjj<8) {
+	    //	    printf("Copy_plane %d %le n,b=%d,%d mask %d ocb %d\n",ro+o+b,*ptr,n,b,cbmask,ocb);
+	    //	    jjj++;
+	    //	  }
+	}
+
+      }
+      o +=rhs._grid->_slice_stride[dimension];
+    }
+
+  }
+}
+
+friend void Copy_plane_permute(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int dimension,int lplane,int rplane,int cbmask,int permute_type)
+{
+  int rd = rhs._grid->_rdimensions[dimension];
+
+
+  if ( !rhs._grid->CheckerBoarded(dimension) ) {
+
+    int o   = 0;                                     // relative offset to base within plane
+    int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+    int lo  = lplane*rhs._grid->_ostride[dimension]; // offset in buffer
+
+  // Simple block stride gather of SIMD objects
+#pragma omp parallel for collapse(2)
+    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+	permute(lhs._odata[lo+o+b],rhs._odata[ro+o+b],permute_type);
+      }
+      o +=rhs._grid->_slice_stride[dimension];
+    }
+
+  } else {
+
+    int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+    int lo  = lplane*lhs._grid->_ostride[dimension]; // base offset for start of plane 
+    int o   = 0;                                     // relative offset to base within plane
+    
+#pragma omp parallel for collapse(2)
+    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+
+	int ocb=1<<lhs._grid->CheckerBoardFromOsite(o+b);
+
+	if ( ocb&cbmask ) {
+	  permute(lhs._odata[lo+o+b],rhs._odata[ro+o+b],permute_type);
+	}
+
+      }
+      o +=rhs._grid->_slice_stride[dimension];
+    }
+
+  }
+}
+
+//////////////////////////////////////////////////////
+// Local to node Cshift
+//////////////////////////////////////////////////////
+
+  // Work out whether to permute 
+  // ABCDEFGH ->   AE BF CG DH       permute              wrap num
+  //
+  // Shift 0       AE BF CG DH       0 0 0 0    ABCDEFGH   0   0
+  // Shift 1       BF CG DH AE       0 0 0 1    BCDEFGHA   0   1
+  // Shift 2       CG DH AE BF       0 0 1 1    CDEFGHAB   0   2
+  // Shift 3       DH AE BF CG       0 1 1 1    DEFGHABC   0   3
+  // Shift 4       AE BF CG DH       1 1 1 1    EFGHABCD   1   0 
+  // Shift 5       BF CG DH AE       1 1 1 0    FGHACBDE   1   1 
+  // Shift 6       CG DH AE BF       1 1 0 0    GHABCDEF   1   2
+  // Shift 7       DH AE BF CG       1 0 0 0    HABCDEFG   1   3
+
+  // Suppose 4way simd in one dim.
+  // ABCDEFGH ->   AECG BFDH      permute              wrap num
+
+  // Shift 0       AECG BFDH      0,00 0,00 ABCDEFGH         0     0
+  // Shift 1       BFDH CGEA      0,00 1,01 BCDEFGHA         0     1
+  // Shift 2       CGEA DHFB      1,01 1,01 CDEFGHAB         1     0
+  // Shift 3       DHFB EAGC      1,01 1,11 DEFGHABC         1     1
+  // Shift 4       EAGC FBHD      1,11 1,11 EFGHABCD         2     0 
+  // Shift 5       FBHD GCAE      1,11 1,10 FGHABCDE         2     1
+  // Shift 6       GCAE HDBF      1,10 1,10 GHABCDEF         3     0
+  // Shift 7       HDBF AECG      1,10 0,00 HABCDEFG         3     1
+
+  // Generalisation to 8 way simd, 16 way simd required.
+  //
+  // Need log2 Nway masks. consisting of 
+  //	    1 bit  256 bit granule
+  //	    2 bit  128 bit granule
+  //        4 bits 64  bit granule
+  //        8 bits 32  bit granules
+  //
+  //        15 bits....
+
+friend void Cshift_local(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimension,int shift)
+{
+  int sshift[2];
+
+  sshift[0] = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,0);
+  sshift[1] = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,1);
+
+  if ( sshift[0] == sshift[1] ) {
+    Cshift_local(ret,rhs,dimension,shift,0x3);
+  } else {
+    Cshift_local(ret,rhs,dimension,shift,0x1);// if checkerboard is unfavourable take two passes
+    Cshift_local(ret,rhs,dimension,shift,0x2);// both with block stride loop iteration
+  }
+}
+
+
+friend Lattice<vobj> Cshift_local(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int shift,int cbmask)
+{
+  int fd = rhs._grid->_fdimensions[dimension];
+  int rd = rhs._grid->_rdimensions[dimension];
+  int ld = rhs._grid->_ldimensions[dimension];
+  int gd = rhs._grid->_gdimensions[dimension];
+  
+
+  // Map to always positive shift modulo global full dimension.
+  shift = (shift+fd)%fd;
+
+  ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
+        
+  // the permute type
+  int permute_dim =rhs._grid->_simd_layout[dimension]>1 ;
+  int permute_type=0;
+  for(int d=0;d<dimension;d++){
+    if (rhs._grid->_simd_layout[d]>1 ) permute_type++;
+  }
+
+  for(int x=0;x<rd;x++){       
+
+    int o   = 0;
+    int bo  = x * rhs._grid->_ostride[dimension];
+    
+    int cb= (cbmask==0x2)? 1 : 0;
+
+    int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,cb);
+    int sx     = (x+sshift)%rd;
+	
+    int permute_slice=0;
+    if(permute_dim){
+      int wrap = sshift/rd;
+      int  num = sshift%rd;
+      if ( x< rd-num ) permute_slice=wrap;
+      else permute_slice = 1-wrap;
+    }
+
+    if ( permute_slice ) Copy_plane_permute(ret,rhs,dimension,x,sx,cbmask,permute_type);
+    else                 Copy_plane(ret,rhs,dimension,x,sx,cbmask); 
+
+  
+  }
+  return ret;
+}
+
+#endif
diff --git a/Grid_init.cc b/Grid_init.cc
index bb22ce29..4f7d1849 100755
--- a/Grid_init.cc
+++ b/Grid_init.cc
@@ -16,8 +16,11 @@
 #undef __X86_64
 namespace dpo {
 
-void Grid_init(void)
+void Grid_init(int *argc,char ***argv)
 {
+#ifdef GRID_COMMS_MPI
+  MPI_Init(argc,argv);
+#endif
   Grid_debug_handler_init();
 }
 double usecond(void) {
diff --git a/Grid_main.cc b/Grid_main.cc
index 4079e62a..795ea395 100644
--- a/Grid_main.cc
+++ b/Grid_main.cc
@@ -13,34 +13,28 @@ using namespace dpo::QCD;
 
 int main (int argc, char ** argv)
 {
+  Grid_init(&argc,&argv);
 
-#if 0
-  struct sigaction sa,osa;
-  sigemptyset (&sa.sa_mask);
-  sa.sa_sigaction= sa_action;
-  sa.sa_flags    = SA_ONSTACK|SA_SIGINFO;
-
-  sigaction(SIGSEGV,&sa,NULL);
-  sigaction(SIGTRAP,&sa,NULL);
-#endif
-
-    std::vector<int> latt_size(4);
-    std::vector<int> simd_layout(4);
-
-    std::vector<int> mpi_layout(4);
-    for(int d=0;d<4;d++) mpi_layout[d]=1;
+  std::vector<int> latt_size(4);
+  std::vector<int> simd_layout(4);
+  
+  std::vector<int> mpi_layout(4);
+  mpi_layout[0]=4;
+  mpi_layout[1]=2;
+  mpi_layout[2]=4;
+  mpi_layout[3]=2;
 
 #ifdef AVX512
  for(int omp=128;omp<236;omp+=16){
 #else
- for(int omp=1;omp<8;omp*=2){
+ for(int omp=1;omp<8;omp*=20){
 #endif
 
 #ifdef OMP
    omp_set_num_threads(omp);
 #endif 
 
-  for(int lat=4;lat<=16;lat+=40){
+  for(int lat=16;lat<=16;lat+=40){
     latt_size[0] = lat;
     latt_size[1] = lat;
     latt_size[2] = lat;
@@ -234,9 +228,9 @@ int main (int argc, char ** argv)
 	std::cout << "Shifting both parities by "<< shift <<" direction "<< dir <<std::endl;
 	Shifted  = Cshift(Foo,dir,shift);    // Shift everything
 
-	std::cout << "Shifting even parities"<<std::endl;
+	std::cout << "Shifting even source parities to odd result"<<std::endl;
 	bShifted = Cshift(rFoo,dir,shift);   // Shift red->black
-	std::cout << "Shifting odd parities"<<std::endl;
+	std::cout << "Shifting odd parities to even result"<<std::endl;
 	rShifted = Cshift(bFoo,dir,shift);   // Shift black->red
     
 	ShiftedCheck=zero;
@@ -245,19 +239,19 @@ int main (int argc, char ** argv)
     
 	// Check results
 	std::vector<int> coor(4);
-	for(coor[3]=0;coor[3]<latt_size[3];coor[3]++){
-	for(coor[2]=0;coor[2]<latt_size[2];coor[2]++){
-	for(coor[1]=0;coor[1]<latt_size[1];coor[1]++){
-	for(coor[0]=0;coor[0]<latt_size[0];coor[0]++){
+	for(coor[3]=0;coor[3]<latt_size[3]/mpi_layout[3];coor[3]++){
+	for(coor[2]=0;coor[2]<latt_size[2]/mpi_layout[2];coor[2]++){
+	for(coor[1]=0;coor[1]<latt_size[1]/mpi_layout[1];coor[1]++){
+	for(coor[0]=0;coor[0]<latt_size[0]/mpi_layout[0];coor[0]++){
  
         std::complex<dpo::Real> diff;
                     
         std::vector<int> shiftcoor = coor;
-        shiftcoor[dir]=(shiftcoor[dir]+shift+latt_size[dir])%latt_size[dir];
+        shiftcoor[dir]=(shiftcoor[dir]+shift+latt_size[dir])%(latt_size[dir]/mpi_layout[dir]);
 
 	std::vector<int> rl(4);
 	for(int dd=0;dd<4;dd++){
-	  rl[dd] = latt_size[dd]/simd_layout[dd];
+	  rl[dd] = latt_size[dd]/simd_layout[dd]/mpi_layout[dd];
 	}
 	int lex =  coor[0]%rl[0]
 	  + (coor[1]%rl[1])*rl[0]
@@ -344,4 +338,6 @@ int main (int argc, char ** argv)
    } // loop for lat
  } // loop for omp
 
+ MPI_Finalize();
+
 }
diff --git a/Grid_mpi.cc b/Grid_mpi.cc
index 08d49312..6cda0f1b 100644
--- a/Grid_mpi.cc
+++ b/Grid_mpi.cc
@@ -7,16 +7,23 @@ namespace dpo {
 
 CartesianCommunicator::CartesianCommunicator(std::vector<int> &processors)
 {
-  _ndimension = _processors.size();
+  _ndimension = processors.size();
   std::vector<int> periodic(_ndimension,1);
 
+  _Nprocessors=1;
   _processors = processors;
-  _processor_coords.resize(_ndimension);
-
-  MPI_Cart_create(MPI_COMM_WORLD _ndimension,&_processors[0],&periodic[0],1,&communicator);
+  _processor_coor.resize(_ndimension);
+  
+  MPI_Cart_create(MPI_COMM_WORLD, _ndimension,&_processors[0],&periodic[0],1,&communicator);
   MPI_Comm_rank(communicator,&_processor);
-  MPI_Cart_coords(communicator,_processor,_ndimension,&_processor_coords[0]);
-
+  MPI_Cart_coords(communicator,_processor,_ndimension,&_processor_coor[0]);
+  printf("Hello world from processor [");
+  for(int i=0;i<_ndimension;i++){
+    printf("%d ",_processor_coor[i]);
+    _Nprocessors*=_processors[i];
+  }
+  printf("]\n");
+  fflush(stdout);
 }
 
 void CartesianCommunicator::GlobalSumF(float &f){
@@ -35,11 +42,9 @@ void CartesianCommunicator::GlobalSumFVector(double *d,int N)
   MPI_Allreduce(d,d,N,MPI_DOUBLE,MPI_SUM,communicator);
 }
 
-int CartesianCommunicator::ShiftedRank(int dim,int shift)
+void CartesianCommunicator::ShiftedRanks(int dim,int shift,int &source,int &dest)
 {
-  int rank;
-  MPI_Cart_shift(communicator,dim,shift,&_processor,&rank);
-  return rank;
+  MPI_Cart_shift(communicator,dim,shift,&source,&dest);
 }
 int CartesianCommunicator::Rank(std::vector<int> coor)
 {
@@ -50,23 +55,18 @@ int CartesianCommunicator::Rank(std::vector<int> coor)
 
 // Basic Halo comms primitive
 void CartesianCommunicator::SendToRecvFrom(void *xmit,
-					   std::vector<int> &to_coordinate,
+					   int dest,
 					   void *recv,
-					   std::vector<int> &from_coordinate,
+					   int from,
 					   int bytes)
 {
-  int dest = Rank(to_coordinate);
-  int from = Rank(from_coordinate);
+  MPI_Request reqs[2];
+  MPI_Status OkeyDokey[2];
+  int rank = _processor;
+  MPI_Isend(xmit, bytes, MPI_CHAR,dest,_processor,communicator,&reqs[0]);
+  MPI_Irecv(recv, bytes, MPI_CHAR,from,from,communicator,&reqs[1]);
+  MPI_Waitall(2,reqs,OkeyDokey);
 
-  MPI_Request x_req;
-  MPI_Request r_req;
-  MPI_Status OkeyDokey;
-
-  MPI_Isend(xmit, bytes, MPI_CHAR,dest,dest,communicator,&x_req);
-  MPI_Irecv(recv, bytes, MPI_CHAR,from,from,communicator,&r_req);
-  MPI_Wait(&x_req,&OkeyDokey);
-  MPI_Wait(&r_req,&OkeyDokey);
-  MPI_Barrier();
 }
 
 }
diff --git a/Grid_mpi_cshift.h b/Grid_mpi_cshift.h
index 9031170b..4029f152 100644
--- a/Grid_mpi_cshift.h
+++ b/Grid_mpi_cshift.h
@@ -1,26 +1,9 @@
 #ifndef _GRID_MPI_CSHIFT_H_
 #define _GRID_MPI_CSHIFT_H_
 
-      
-//////////////////////////////////////////////
-// Q. Split this into seperate sub functions?
-//////////////////////////////////////////////
-
-// CshiftCB_comms_splice
-// CshiftCB_comms
-// CshiftCB_local
-// CshiftCB_local_permute
-
-// Cshift_comms_splice
-// Cshift_comms
-// Cshift_local
-// Cshift_local_permute
-
-// Broadly I remain annoyed that the iteration is so painful
-// for red black data layout, when simple block strided descriptors suffice for non-cb. 
-//
-// The other option is to do it table driven, or perhaps store the CB of each site in a table.
-//
+#define MAX(x,y) ((x)>(y)?(x):(y))
+#define MIN(x,y) ((x)>(y)?(y):(x))
+//////////////////////////////////////////////////////////////////////////////////////////
 // Must not lose sight that goal is to be able to construct really efficient
 // gather to a point stencil code. CSHIFT is not the best way, so probably need
 // additional stencil support.
@@ -33,79 +16,31 @@
 //
 // Grid could create a neighbour index table for a given stencil.
 // Could also implement CovariantCshift.
+//////////////////////////////////////////////////////////////////////////////////////////
 
-//////////////////////////////////////////////////////
-//Non checkerboarded support functions
-//////////////////////////////////////////////////////
+      
+/////////////////////////////////////////////////////////////
+// Q. Further split this into separate sub functions?
+/////////////////////////////////////////////////////////////
 
-friend void Gather_plane        (Lattice<vobj> &rhs,std::vector<vobj> &buffer,             int dimension,int plane)
-{
-  const int Nsimd = vector_type::Nsimd();
-  int rd = rhs._grid->_rdimensions[dimension];
+// CshiftCB_local
+// CshiftCB_local_permute
 
-  int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-  int o   = 0;                                    // relative offset to base within plane
-#pragma omp parallel for collapse(2)
-  for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-    for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-	  
-	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
-	  int sx = (x+sshift)%rd;
-	  int so = sx*rhs._grid->_ostride[dimension];
-	  int permute_slice=0;
-	  int wrap = sshift/rd;
-	  int  num = sshift%rd;
-	  
-	  if ( x< rd-num ) permute_slice=wrap;
-	  else permute_slice = 1-wrap;
-	  
-	  if ( permute_slice ) {
-	    permute(ret._odata[ro+o+b],rhs._odata[so+o+b],permute_type);
-	  } else {
-	    ret._odata[ro+o+b]=rhs._odata[so+o+b];
-	  }
-	}
-	o +=rhs._grid->_slice_stride[dimension];
-      }
-  }
-  
-}
-//friend void Gather_plane_extract(Lattice<vobj> &rhs,std::vector<scalar_type *> pointers,int dimension,int plane);
-//
-//friend void Scatter_plane       (Lattice<vobj> &rhs,std::vector<vobj> face,             int dimension,int plane);
-//friend void Scatter_plane_merge (Lattice<vobj> &rhs,std::vector<scalar_type *> pointers,int dimension,int plane);
-//
-//template<int permute_type> friend void Copy_plane_permute(Lattice<vobj> &rhs,std::vector<vobj> face, int dimension,int plane);
-//                           friend void Copy_plane(Lattice<vobj> &rhs,std::vector<vobj> face, int dimension,int plane);
-//
-
-//////////////////////////////////////////////////////
-//Checkerboarded support functions
-//////////////////////////////////////////////////////
-
-//friend void GatherCB_plane        (Lattice<vobj> &rhs,std::vector<vobj> face,             int dimension,int plane);
-//friend void GatherCB_plane_extract(Lattice<vobj> &rhs,std::vector<scalar_type *> pointers,int dimension,int plane);
-//
-//friend void ScatterCB_plane       (Lattice<vobj> &rhs,std::vector<vobj> face,             int dimension,int plane);
-//friend void ScatterCB_plane_merge (Lattice<vobj> &rhs,std::vector<scalar_type *> pointers,int dimension,int plane);
-//
-//template<int permute_type> friend void CopyCB_plane_permute(Lattice<vobj> &rhs,std::vector<vobj> face, int dimension,int plane);
-//                           friend void Copy_plane(Lattice<vobj> &rhs,std::vector<vobj> face, int dimension,int plane);
+// Cshift_comms_splice
+// Cshift_comms
+// Cshift_local
+// Cshift_local_permute
 
 
 friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
 {
   typedef typename vobj::vector_type vector_type;
   typedef typename vobj::scalar_type scalar_type;
-  const int Nsimd = vector_type::Nsimd();
 
   Lattice<vobj> ret(rhs._grid);
   
   int fd = rhs._grid->_fdimensions[dimension];
   int rd = rhs._grid->_rdimensions[dimension];
-  //int ld = rhs._grid->_ldimensions[dimension];
-  //int gd = rhs._grid->_gdimensions[dimension];
-  
 
   // Map to always positive shift modulo global full dimension.
   shift = (shift+fd)%fd;
@@ -115,270 +50,288 @@ friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
   // the permute type
   int simd_layout     = rhs._grid->_simd_layout[dimension];
   int comm_dim        = rhs._grid->_processors[dimension] >1 ;
-  int permute_dim     = rhs._grid->_simd_layout[dimension]>1 && (!comm_dim);
   int splice_dim      = rhs._grid->_simd_layout[dimension]>1 && (comm_dim);
 
-  int permute_type=0;
-  for(int d=0;d<dimension;d++){
-    if (rhs._grid->_simd_layout[d]>1 ) permute_type++;
-  }
-
-  // Logic for non-distributed dimension	  
-  std::vector<int> comm_offnode(simd_layout);
-  std::vector<int> comm_to     (simd_layout);
-  std::vector<int> comm_from   (simd_layout);
-  std::vector<int> comm_rx     (simd_layout);  // reduced coordinate of neighbour plane
-  std::vector<int> comm_simd_lane(simd_layout);// simd lane of neigbour plane
-
-  ///////////////////////////////////////////////
-  // Move via a fake comms buffer
-  // Simd direction uses an extract/merge pair
-  ///////////////////////////////////////////////
-  int buffer_size = rhs._grid->_slice_nblock[dimension]*rhs._grid->_slice_block[dimension];
-  int words = sizeof(vobj)/sizeof(vector_type);
-
-  std::vector<vobj,alignedAllocator<vobj> > comm_buf(buffer_size);
-  std::vector<std::vector<scalar_type> > comm_buf_extract(Nsimd,std::vector<scalar_type>(buffer_size*words) );
-  std::vector<scalar_type *> pointers(Nsimd);
-
-
-
-  if ( permute_dim ) {
-
-    for(int x=0;x<rd;x++){       
-      int ro  = x*rhs._grid->_ostride[dimension]; // base offset for result
-      int o   = 0;                                // relative offset to base
-      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-	  
-	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
-	  int sx = (x+sshift)%rd;
-	  int so = sx*rhs._grid->_ostride[dimension];
-	  int permute_slice=0;
-	  int wrap = sshift/rd;
-	  int  num = sshift%rd;
-	  
-	  if ( x< rd-num ) permute_slice=wrap;
-	  else permute_slice = 1-wrap;
-	  
-	  if ( permute_slice ) {
-	    permute(ret._odata[ro+o+b],rhs._odata[so+o+b],permute_type);
-	  } else {
-	    ret._odata[ro+o+b]=rhs._odata[so+o+b];
-	  }
-	}
-	o +=rhs._grid->_slice_stride[dimension];
-      }
-    }
 
+  if ( !comm_dim ) {
+    Cshift_local(ret,rhs,dimension,shift); // Handles checkerboarding
   } else if ( splice_dim ) {
-
-    if ( rhs._grid->_simd_layout[dimension] > 2 ) exit(-1); // use Cassert. Audit code for exit and replace
-    if ( rhs._grid->_simd_layout[dimension] < 1 ) exit(-1);
-	    
-
-    for(int i=0;i<vobj::vector_type::Nsimd();i++){
-      pointers[i] = (scalar_type *)&comm_buf_extract[i][0];
-    }
-
-    
-    for(int x=0;x<rd;x++){       
-
-
-      ///////////////////////////////////////////
-      // Extract one orthogonal slice at a time
-      ///////////////////////////////////////////
-      int ro  = x*rhs._grid->_ostride[dimension]; // base offset for result
-
-      o   = 0;                                    // relative offset to base      
-      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-	  
-	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
-	  int sx = (x+sshift)%rd;
-	  
-	  // base offset for source
-	  int so = sx*rhs._grid->_ostride[dimension];
-
-	  int permute_slice=0;
-	  int wrap = sshift/rd;
-	  int  num = sshift%rd;
-	    
-	  if ( x< rd-num ) permute_slice=wrap;
-	  else permute_slice = 1-wrap;
-
-	  if ( permute_slice ) {
-	    extract(rhs._odata[so+o+b],pointers);
-	  }
-	}
-	o +=rhs._grid->_slice_stride[dimension];
-      }
-
-    
-      ///////////////////////////////////////////
-      // Work out what to send where
-      ///////////////////////////////////////////
-
-      for(int s=0;s<simd_layout;s++) {
-
-
-	// shift to "neighbour" takes us off node
-	// coordinates (rx, simd_lane) of neighbour
-	// how many nodes away is this shift
-	// where we should send to
-	// where we should receive from
-	int shifted_x = x+s*rd+shift;
-	comm_offnode[s]           = shifted_x > ld; 
-	comm_send_rx[s]           = shifted_x%rd;     // which slice geton the other node
-	comm_send_simd_lane [s]   = shifted_x/rd;     // which slice on the other node
-	comm_from[s]    = shifted_x/ld;     
-	comm_to  [s] = (2*_processors[dimension]-comm_from[s]) % _processors[dimension];
-	comm_from[s] = (comm_from[s]+_processors[dimension])   % _processors[dimension];
-	      
-      }
-
-      ////////////////////////////////////////////////
-      // Insert communication phase
-      ////////////////////////////////////////////////
-#if 0
-	  } else if (comm_dim ) {
-	    
-	    // Packed gather sequence is clean
-	    int buffer_size = rhs._grid->_slice_nblock[dimension]*rhs._grid->_slice_nblock[dimension];
-	    std::vector<vobj,alignedAllocator<vobj> > send_buf(buffer_size);
-	    std::vector<vobj,alignedAllocator<vobj> > recv_buf(buffer_size);
-
-	    // off node; communcate slice (ld==rd)
-	    if ( x+shift > rd ) { 
-	      int sb=0;
-	      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-		for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
-		  send_buf[sb++]=rhs._odata[so+i];
-		}
-		so+=rhs._grid->_slice_stride[dimension];
-	      }	      
-
-	      // Make a comm_fake them mimics comms in periodic case.
-	      
-	      // scatter face
-	      int rb=0;
-	      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-		for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
-		  ret._odata[so+i]=recv_buf[rb++];
-		}
-		so+=rhs._grid->_slice_stride[dimension];
-	      }	      
-
-	    } else { 
-
-	      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-		for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
-		  ret._odata[o+i]=rhs._odata[so+i];
-		}
-		o+=rhs._grid->_slice_stride[dimension];
-		so+=rhs._grid->_slice_stride[dimension];
-	      }
-	    }
-
-#endif	
-    
-      ////////////////////////////////////////////////
-      // Pull receive buffers and permuted buffers in
-      ////////////////////////////////////////////////
-      for(int i=0;i<vobj::vector_type::Nsimd();i++){
-	pointers[i] = (scalar_type *)&comm_buf_extract[permute_map[permute_type][i]][0];
-      }
-
-      o   = 0;                                    // relative offset to base
-      int ro  = x*rhs._grid->_ostride[dimension]; // base offset for result
-      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-	  
-	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
-	  int sx = (x+sshift)%rd;
-	  
-	  // base offset for source
-	  int so = sx*rhs._grid->_ostride[dimension];
-	  
-	  int permute_slice=0;
-	  int wrap = sshift/rd;
-	  int  num = sshift%rd;
-	  
-	  if ( x< rd-num ) permute_slice=wrap;
-	  else permute_slice = 1-wrap;
-	  
-	  if ( permute_slice ) {
-	    merge(ret._odata[ro+o+b],pointers);
-	  }
-	}
-	o +=rhs._grid->_slice_stride[dimension];
-      }
-    }
-  
-
-  } else if ( comm_dim ) { 
-	
-
-    int co; // comm offset
-    int o;
-    
-    co=0;
-    for(int x=0;x<rd;x++){       
-      o=0;
-      int ro  = x*rhs._grid->_ostride[dimension]; // base offset for result
-      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-
-	  // This call in inner loop is annoying but necessary for dimension=0
-	  // in the case of RedBlack grids. Could optimise away with 
-	  // alternate code paths for all other cases.
-	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
-	  int sx = (x+sshift)%rd;
-	  int so = sx*rhs._grid->_ostride[dimension];
-
-	  comm_buf[co++]=rhs._odata[so+o+b];
-
-	}
-	o +=rhs._grid->_slice_stride[dimension];
-      }
-
-      // Step through a copy into a comms buffer and pull back in.
-      // Genuine fake implementation could calculate if loops back
-      co=0; o=0;
-      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-	  ret._odata[ro+o+b]=comm_buf[co++];
-	}
-	o +=rhs._grid->_slice_stride[dimension];
-      }
-    }
-
-  } else { // Local dimension, no permute required
-
-    for(int x=0;x<rd;x++){       
-      int o=0;
-      int ro  = x*rhs._grid->_ostride[dimension]; // base offset for result
-      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-
-	  // This call in inner loop is annoying but necessary for dimension=0
-	  // in the case of RedBlack grids. Could optimise away with 
-	  // alternate code paths for all other cases.
-	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
-	  
-	  int sx = (x+sshift)%rd;
-	  int so = sx*rhs._grid->_ostride[dimension];
-	  ret._odata[bo+o+b]=rhs._odata[so+o+b];
-
-	}
-	o +=rhs._grid->_slice_stride[dimension];
-      }
-    }
-
+    Cshift_comms_simd(ret,rhs,dimension,shift);
+  } else {
+    Cshift_comms(ret,rhs,dimension,shift);
   }
-
   return ret;
 }
 
+friend void Cshift_comms(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimension,int shift)
+{
+  int sshift[2];
+
+  sshift[0] = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,0);
+  sshift[1] = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,1);
+
+  if ( sshift[0] == sshift[1] ) {
+    //    printf("Cshift_comms : single pass\n");
+    Cshift_comms(ret,rhs,dimension,shift,0x3);
+  } else {
+    //    printf("Cshift_comms : two pass\n");
+    //    printf("call1\n");
+    Cshift_comms(ret,rhs,dimension,shift,0x1);// if checkerboard is unfavourable take two passes
+    //    printf("call2\n");
+    Cshift_comms(ret,rhs,dimension,shift,0x2);// both with block stride loop iteration
+    //    printf("done\n");
+
+  }
+}
+
+friend void Cshift_comms_simd(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimension,int shift)
+{
+  int sshift[2];
+
+  sshift[0] = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,0);
+  sshift[1] = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,1);
+
+  if ( sshift[0] == sshift[1] ) {
+    Cshift_comms_simd(ret,rhs,dimension,shift,0x3);
+  } else {
+    //    printf("call1 0x1 cb=even\n");
+    Cshift_comms_simd(ret,rhs,dimension,shift,0x1);// if checkerboard is unfavourable take two passes
+    //    printf("call2 0x2 cb=odd\n");
+    Cshift_comms_simd(ret,rhs,dimension,shift,0x2);// both with block stride loop iteration
+    //    printf("done\n");
+  }
+}
+
+
+friend void Cshift_comms(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int shift,int cbmask)
+{
+  typedef typename vobj::vector_type vector_type;
+  typedef typename vobj::scalar_type scalar_type;
+
+  SimdGrid *grid=rhs._grid;
+  Lattice<vobj> temp(rhs._grid);
+
+  int fd              = rhs._grid->_fdimensions[dimension];
+  int rd              = rhs._grid->_rdimensions[dimension];
+  int simd_layout     = rhs._grid->_simd_layout[dimension];
+  int comm_dim        = rhs._grid->_processors[dimension] >1 ;
+  assert(simd_layout==1);
+  assert(comm_dim==1);
+  assert(shift>=0);
+  assert(shift<fd);
+  
+  // Packed gather sequence is clean
+  int buffer_size = rhs._grid->_slice_nblock[dimension]*rhs._grid->_slice_block[dimension];
+  std::vector<vobj,alignedAllocator<vobj> > send_buf(buffer_size);
+  std::vector<vobj,alignedAllocator<vobj> > recv_buf(buffer_size);
+
+  // This code could be simplified by multiple calls to single routine with extra params to
+  // encapsulate the difference in the code paths.
+  int cb= (cbmask==0x2)? 1 : 0;
+  int sshift= rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,cb);
+
+  for(int x=0;x<rd;x++){       
+
+    int offnode = ( x+sshift >= rd );
+    int sx        = (x+sshift)%rd;
+    int comm_proc = (x+sshift)/rd;
+    
+    if (!offnode) {
+      //      printf("local x %d sshift %d offnode %d rd %d cb %d\n",x,sshift,offnode,rd,cb);
+      Copy_plane(ret,rhs,dimension,x,sx,cbmask); 
+    } else {
+
+      int words = send_buf.size();
+      if (cbmask != 0x3) words=words>>1;
+
+      int bytes = words * sizeof(vobj);
+
+      //      printf("nonlocal x %d sx %d sshift %d offnode %d rd %d cb %d cbmask %d rhscb %d comm_proc %d\n",
+      //	     x,sx,sshift,offnode,rd,cb,cbmask,rhs.checkerboard,comm_proc);
+      //      Copy_plane(temp,rhs,dimension,x,sx,cbmask); 
+
+      // Bug found; cbmask may differ between sx plan and rx plane.
+      Gather_plane_simple (rhs,send_buf,dimension,sx,cbmask);
+      //      for(int i=0;i<MIN(words,8);i++){
+      //	float *ptr = (float *)&send_buf[i];
+      //	printf("send buf shift %d cbmask %d i %d %le\n",sshift,cbmask,i,*ptr);
+      //      }
+      //      Gather_plane_simple (rhs,send_buf,dimension,sx,cbmask^0x3);
+      //      for(int i=0;i<MIN(words,8);i++){
+      //	float *ptr = (float *)&send_buf[i];
+      //	printf("send buf shift %d cbmask %d i %d %le\n",sshift,cbmask,i,*ptr);
+      //      }
+      //      recv_buf=send_buf;
+
+      int rank           = grid->_processor;
+      int recv_from_rank;
+      int xmit_to_rank;
+      grid->ShiftedRanks(dimension,comm_proc,xmit_to_rank,recv_from_rank);
+
+      //      printf("bytes %d node %d sending to %d receiving from %d\n",bytes,rank,xmit_to_rank,recv_from_rank );
+      grid->SendToRecvFrom((void *)&send_buf[0],
+			   xmit_to_rank,
+			   (void *)&recv_buf[0],
+			   recv_from_rank,
+			   bytes);
+
+      Scatter_plane_simple (ret,recv_buf,dimension,x,cbmask);
+    }
+  }
+}
+
+
+friend void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int shift,int cbmask)
+{
+  const int Nsimd = vector_type::Nsimd();
+  SimdGrid *grid=rhs._grid;
+  typedef typename vobj::vector_type vector_type;
+  typedef typename vobj::scalar_type scalar_type;
+   
+  int fd = grid->_fdimensions[dimension];
+  int rd = grid->_rdimensions[dimension];
+  int ld = grid->_ldimensions[dimension];
+  int simd_layout     = grid->_simd_layout[dimension];
+  int comm_dim        = grid->_processors[dimension] >1 ;
+
+  assert(comm_dim==1);
+  assert(simd_layout==2);
+  assert(shift>=0);
+  assert(shift<fd);
+
+  int permute_type=0;
+  for(int d=0;d<dimension;d++){
+    if (grid->_simd_layout[d]>1 ) permute_type++;
+  }
+
+  ///////////////////////////////////////////////
+  // Simd direction uses an extract/merge pair
+  ///////////////////////////////////////////////
+  int buffer_size = grid->_slice_nblock[dimension]*grid->_slice_block[dimension];
+  int words = sizeof(vobj)/sizeof(vector_type);
+
+  std::vector<std::vector<scalar_type> > send_buf_extract(Nsimd,std::vector<scalar_type>(buffer_size*words) );
+  std::vector<std::vector<scalar_type> > recv_buf_extract(Nsimd,std::vector<scalar_type>(buffer_size*words) );
+  int bytes = buffer_size*words*sizeof(scalar_type);
+
+  std::vector<scalar_type *> pointers(Nsimd);  // 
+  std::vector<scalar_type *> rpointers(Nsimd); // received pointers
+
+  ///////////////////////////////////////////
+  // Work out what to send where
+  ///////////////////////////////////////////
+
+  int cb    = (cbmask==0x2)? 1 : 0;
+  int sshift= grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,cb);
+  
+  //  printf("cshift-comms-simd: shift = %d ; sshift = %d ; cbmask %d ; simd_layout %d\n",shift,sshift,cbmask,simd_layout);
+  std::vector<int> comm_offnode(simd_layout);
+  std::vector<int> comm_proc   (simd_layout);  //relative processor coord in dim=dimension
+
+  // Strategy
+  //
+  //*  Loop over source planes
+  //*    if any communication needed extract and send
+  //*    if communication needed extract and send
+
+  for(int x=0;x<rd;x++){       
+
+    int comm_any = 0;
+    for(int s=0;s<simd_layout;s++) {
+      // does shift to "neighbour" takes us off node?
+      // coordinates (reduce plane, simd_lane) of neighbour?
+      // how many nodes away is this shift?
+      // where we should send to?
+      // where we should receive from?
+      int shifted_x   = x+s*rd+sshift;
+      comm_offnode[s] = shifted_x >= ld; 
+      comm_any        = comm_any | comm_offnode[s];
+      comm_proc[s]    = shifted_x/ld;     
+      //      printf("rd %d x %d shifted %d s=%d comm_any %d\n",rd, x,shifted_x,s,comm_any);
+    }
+    
+    int o    = 0;
+    int bo   = x*grid->_ostride[dimension];
+    int sx   = (x+sshift)%rd;
+
+    // Need Convenience function in _grid. Move this in
+    if ( comm_any ) {
+
+      for(int i=0;i<Nsimd;i++){
+	pointers[i] = (scalar_type *)&send_buf_extract[i][0];
+      }
+      Gather_plane_extract(rhs,pointers,dimension,sx,cbmask);
+      //      for(int i=0;i<Nsimd;i++){
+      //	printf("extracted %d %le\n",i,real(send_buf_extract[i][0]));
+      //      }
+
+      for(int i=0;i<Nsimd;i++){
+
+	int s = grid->iCoordFromIsite(i,dimension);
+
+	if(comm_offnode[s]){
+
+	  int rank           = grid->_processor;
+	  int recv_from_rank;
+	  int xmit_to_rank;
+	  grid->ShiftedRanks(dimension,comm_proc[s],xmit_to_rank,recv_from_rank);
+	  
+
+	  grid->SendToRecvFrom((void *)&send_buf_extract[i][0],
+			    xmit_to_rank,
+			    (void *)&recv_buf_extract[i][0],
+			    recv_from_rank,
+			    bytes);
+
+	  //	  printf("Cshift_simd comms %d %le %le\n",i,real(recv_buf_extract[i][0]),real(send_buf_extract[i][0]));
+
+	  rpointers[i] = (scalar_type *)&recv_buf_extract[i][0];
+
+	} else { 
+
+	  rpointers[i] = (scalar_type *)&send_buf_extract[i][0];
+	  //	  printf("Cshift_simd local %d %le \n",i,real(send_buf_extract[i][0]));
+
+	}
+
+      }
+
+      // Permute by swizzling pointers in merge
+      int permute_slice=0;
+      int lshift=sshift%ld;
+      int wrap  =lshift/rd;
+      int  num  =lshift%rd;
+
+      if ( x< rd-num ) permute_slice=wrap;
+      else permute_slice = 1-wrap;
+
+      for(int i=0;i<vobj::vector_type::Nsimd();i++){
+	if ( permute_slice ) {
+	  pointers[i] = rpointers[permute_map[permute_type][i]];
+	} else {
+	  pointers[i] = rpointers[i];
+	}
+	//	printf("Cshift_simd perm %d num %d wrap %d swiz %d %le unswiz %le\n",permute_slice,num,wrap,i,real(pointers[i][0]),real(rpointers[i][0]));
+      }
+
+      Scatter_plane_merge(ret,pointers,dimension,x,cbmask);
+
+    } else { 
+
+      int permute_slice=0;
+      int wrap = sshift/rd;
+      int  num = sshift%rd;
+      if ( x< rd-num ) permute_slice=wrap;
+      else permute_slice = 1-wrap;
+
+      if ( permute_slice ) Copy_plane_permute(ret,rhs,dimension,x,sx,cbmask,permute_type);
+      else                 Copy_plane(ret,rhs,dimension,x,sx,cbmask); 
+
+    }
+  }
+}
+
+
+
 
 #endif
diff --git a/Grid_none_cshift.h b/Grid_none_cshift.h
index 87ae6a68..f16b403b 100644
--- a/Grid_none_cshift.h
+++ b/Grid_none_cshift.h
@@ -1,313 +1,11 @@
 #ifndef _GRID_NONE_CSHIFT_H_
 #define _GRID_NONE_CSHIFT_H_
-  // Work out whether to permute 
-  // ABCDEFGH ->   AE BF CG DH       permute              wrap num
-  //
-  // Shift 0       AE BF CG DH       0 0 0 0    ABCDEFGH   0   0
-  // Shift 1       BF CG DH AE       0 0 0 1    BCDEFGHA   0   1
-  // Shift 2       CG DH AE BF       0 0 1 1    CDEFGHAB   0   2
-  // Shift 3       DH AE BF CG       0 1 1 1    DEFGHABC   0   3
-  // Shift 4       AE BF CG DH       1 1 1 1    EFGHABCD   1   0 
-  // Shift 5       BF CG DH AE       1 1 1 0    FGHACBDE   1   1 
-  // Shift 6       CG DH AE BF       1 1 0 0    GHABCDEF   1   2
-  // Shift 7       DH AE BF CG       1 0 0 0    HABCDEFG   1   3
-
-  // Suppose 4way simd in one dim.
-  // ABCDEFGH ->   AECG BFDH      permute              wrap num
-
-  // Shift 0       AECG BFDH      0,00 0,00 ABCDEFGH         0     0
-  // Shift 1       BFDH CGEA      0,00 1,01 BCDEFGHA         0     1
-  // Shift 2       CGEA DHFB      1,01 1,01 CDEFGHAB         1     0
-  // Shift 3       DHFB EAGC      1,01 1,11 DEFGHABC         1     1
-  // Shift 4       EAGC FBHD      1,11 1,11 EFGHABCD         2     0 
-  // Shift 5       FBHD GCAE      1,11 1,10 FGHABCDE         2     1
-  // Shift 6       GCAE HDBF      1,10 1,10 GHABCDEF         3     0
-  // Shift 7       HDBF AECG      1,10 0,00 HABCDEFG         3     1
-
-  // Generalisation to 8 way simd, 16 way simd required.
-  //
-  // Need log2 Nway masks. consisting of 
-  //	    1 bit  256 bit granule
-  //	    2 bit  128 bit granule
-  //        4 bits 64  bit granule
-  //        8 bits 32  bit granules
-  //
-  //        15 bits....
-  
-// For optimisation:
-//
-// split into Cshift_none_rb_permute
-// split into Cshift_none_rb_simple
-//
-// split into Cshift_none_permute
-// split into Cshift_none_simple
 
 friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
 {
   Lattice<vobj> ret(rhs._grid);
-  
-  int fd = rhs._grid->_fdimensions[dimension];
-  int rd = rhs._grid->_rdimensions[dimension];
-  int ld = rhs._grid->_ldimensions[dimension];
-  int gd = rhs._grid->_gdimensions[dimension];
-  
-
-  // Map to always positive shift modulo global full dimension.
-  shift = (shift+fd)%fd;
-
   ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
-        
-  // the permute type
-
-  int permute_dim =rhs._grid->_simd_layout[dimension]>1 ;
-  int permute_type=0;
-  for(int d=0;d<dimension;d++){
-    if (rhs._grid->_simd_layout[d]>1 ) permute_type++;
-  }
-
-  for(int x=0;x<rd;x++){       
-
-    int bo  = x*rhs._grid->_ostride[dimension];
-    int o   = 0;
-
-    if ( permute_dim ) {
-
-      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-	  
-	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
-	  
-	  int sx = (x+sshift)%rd;
-	  int so = sx*rhs._grid->_ostride[dimension];
-
-	  int permute_slice=0;
-	  int wrap = sshift/rd;
-	  int  num = sshift%rd;
-	    
-	  if ( x< rd-num ) permute_slice=wrap;
-	  else permute_slice = 1-wrap;
-	  
-	  if ( permute_slice ) {
-	    permute(ret._odata[bo+o+b],rhs._odata[so+o+b],permute_type);
-	  } else {
-	    ret._odata[bo+o+b]=rhs._odata[so+o+b];
-	  }
-
-	}
-	o +=rhs._grid->_slice_stride[dimension];
-
-      }
-    } else {
-      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-
-	  // This call in inner loop is annoying but necessary for dimension=0
-	  // in the case of RedBlack grids. Could optimise away with 
-	  // alternate code paths for all other cases.
-	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
-	  
-	  int sx = (x+sshift)%rd;
-	  int so = sx*rhs._grid->_ostride[dimension];
-	  ret._odata[bo+o+b]=rhs._odata[so+o+b];
-
-	}
-	o +=rhs._grid->_slice_stride[dimension];
-      }
-    }
-  }
-  return ret;
+  Cshift_local(ret,rhs,dimension,shift);
 }
         
-#if 0 
-// Collapse doesn't appear to work the way I think it should in icpc
-friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
-{
-  Lattice<vobj> ret(rhs._grid);
-        
-  ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
-  shift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift);
-  int sx,so,o;
-  int rd = rhs._grid->_rdimensions[dimension];
-  int ld = rhs._grid->_dimensions[dimension];
-  // Map to always positive shift.
-  shift = (shift+ld)%ld;
-  // Work out whether to permute and the permute type
-  // ABCDEFGH ->   AE BF CG DH       permute
-  // Shift 0       AE BF CG DH       0 0 0 0    ABCDEFGH
-  // Shift 1       DH AE BF CG       1 0 0 0    HABCDEFG
-  // Shift 2       CG DH AE BF       1 1 0 0    GHABCDEF
-  // Shift 3       BF CG DH AE       1 1 1 0    FGHACBDE
-  // Shift 4       AE BF CG DH       1 1 1 1    EFGHABCD
-  // Shift 5       DH AE BF CG       0 1 1 1    DEFGHABC
-  // Shift 6       CG DH AE BF       0 0 1 1    CDEFGHAB
-  // Shift 7       BF CG DH AE       0 0 0 1    BCDEFGHA
-  int permute_dim =rhs._grid->_layout[dimension]>1 ;
-  int permute_type=0;
-  for(int d=0;d<dimension;d++)
-    if (rhs._grid->_layout[d]>1 ) permute_type++;
-        
-  // loop over perp slices.
-  // Threading considerations:
-  //   Need to map thread_num to
-  //
-  //               x_min,x_max for Loop-A
-  //               n_min,n_max for Loop-B
-  //               b_min,b_max for Loop-C
-  //  In a way that maximally load balances.
-  //
-  //  Optimal:
-  //      There are rd*n_block*block items of work.
-  //      These serialise as item "w"
-  //      b=w%block; w=w/block
-  //      n=w%nblock; x=w/nblock. Perhaps 20 cycles?
-  //
-  //  Logic:
-  //      x_chunk = (rd+thread)/nthreads simply divide work across nodes.
-  //
-  //      rd=5 , threads = 8;
-  //      0 1 2 3 4 5 6 7
-  //      0 0 0 1 1 1 1 1
-  for(int x=0;x<rd;x++){         // Loop A
-    sx = (x-shift+ld)%rd;
-    o  = x*rhs._grid->_ostride[dimension];
-    so =sx*rhs._grid->_ostride[dimension];
-    int permute_slice=0;
-    if ( permute_dim ) {
-      permute_slice = shift/rd;
-      if ( x<shift%rd ) permute_slice = 1-permute_slice;
-    }
-#if 0
-    if ( permute_slice ) {
-                
-      int internal=sizeof(vobj)/sizeof(vComplex);
-      int num =rhs._grid->_slice_block[dimension]*internal;
-                
-      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	vComplex *optr = (vComplex *)&ret._odata[o];
-	vComplex *iptr = (vComplex *)&rhs._odata[so];
-	for(int b=0;b<num;b++){
-	  permute(optr[b],iptr[b],permute_type);
-	}
-	o+=rhs._grid->_slice_stride[dimension];
-	so+=rhs._grid->_slice_stride[dimension];
-      }
-    } else {
-      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
-	  ret._odata[o+i]=rhs._odata[so+i];
-	}
-	o+=rhs._grid->_slice_stride[dimension];
-	so+=rhs._grid->_slice_stride[dimension];
-      }
-            
-    }
-#else
-    if ( permute_slice ) {
-      int internal=sizeof(vobj)/sizeof(vComplex);
-      int num =rhs._grid->_slice_block[dimension]*internal;
-#pragma omp parallel for collapse(2)
-      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	for(int b=0;b<num;b++){
-	  vComplex *optr = (vComplex *)&ret._odata[o +n*rhs._grid->_slice_stride[dimension]];
-	  vComplex *iptr = (vComplex *)&rhs._odata[so+n*rhs._grid->_slice_stride[dimension]];
-	  permute(optr[b],iptr[b],permute_type);
-	}
-      }
-    } else {
-#pragma omp parallel for collapse(2)
-      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
-	  int oo = o+ n*rhs._grid->_slice_stride[dimension];
-	  int soo=so+ n*rhs._grid->_slice_stride[dimension];
-	  ret._odata[oo+i]=rhs._odata[soo+i];
-	}
-      }
-            
-    }
-#endif
-  }
-  return ret;
-}
-friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
-{
-  Lattice<vobj> ret(rhs._grid);
-        
-  ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
-  shift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift);
-  int rd = rhs._grid->_rdimensions[dimension];
-  int ld = rhs._grid->_dimensions[dimension];
-        
-  // Map to always positive shift.
-  shift = (shift+ld)%ld;
-        
-  // Work out whether to permute and the permute type
-  // ABCDEFGH ->   AE BF CG DH       permute
-  // Shift 0       AE BF CG DH       0 0 0 0    ABCDEFGH
-  // Shift 1       DH AE BF CG       1 0 0 0    HABCDEFG
-  // Shift 2       CG DH AE BF       1 1 0 0    GHABCDEF
-  // Shift 3       BF CG DH AE       1 1 1 0    FGHACBDE
-  // Shift 4       AE BF CG DH       1 1 1 1    EFGHABCD
-  // Shift 5       DH AE BF CG       0 1 1 1    DEFGHABC
-  // Shift 6       CG DH AE BF       0 0 1 1    CDEFGHAB
-  // Shift 7       BF CG DH AE       0 0 0 1    BCDEFGHA
-  int permute_dim =rhs._grid->_layout[dimension]>1 ;
-  int permute_type=0;
-  for(int d=0;d<dimension;d++)
-    if (rhs._grid->_layout[d]>1 ) permute_type++;
-        
-        
-  // loop over all work
-  int work =rd*rhs._grid->_slice_nblock[dimension]*rhs._grid->_slice_block[dimension];
-
-#pragma omp parallel for
-  for(int ww=0;ww<work;ww++){
-
-            
-    // can optimise this if know w moves congtiguously for a given thread.
-    // b=(b+1);
-    // if (b==_slice_block) {b=0; n=n+1;}
-    // if (n==_slice_nblock) { n=0; x=x+1}
-    //
-    // Perhaps a five cycle iterator, or so.
-    int w=ww;
-    int b = w%rhs._grid->_slice_block[dimension] ; w=w/rhs._grid->_slice_block[dimension];
-    int n = w%rhs._grid->_slice_nblock[dimension]; w=w/rhs._grid->_slice_nblock[dimension];
-    int x = w;
-
-    int sx,so,o;
-    sx = (x-shift+ld)%rd;
-    o  = x*rhs._grid->_ostride[dimension]+n*rhs._grid->_slice_stride[dimension]; // common sub expression alert.
-    so =sx*rhs._grid->_ostride[dimension]+n*rhs._grid->_slice_stride[dimension];
-            
-    int permute_slice=0;
-    if ( permute_dim ) {
-      permute_slice = shift/rd;
-      if ( x<shift%rd ) permute_slice = 1-permute_slice;
-    }
-            
-    if ( permute_slice ) {
-                
-      int internal=sizeof(vobj)/sizeof(vComplex);
-      vComplex *optr = (vComplex *)&ret._odata[o+b];
-      vComplex *iptr = (vComplex *)&rhs._odata[so+b];
-      const char *pf = (const char *)iptr;
-      for(int i=0;i<sizeof(vobj);i+=64){
-	_mm_prefetch(pf+i,_MM_HINT_T0);
-      }
-
-      for(int i=0;i<internal;i++){
-	permute(optr[i],iptr[i],permute_type);
-      }
-    } else {
-      const char *pf = (const char *) &rhs._odata[so+b];
-      for(int i=0;i<sizeof(vobj);i+=64){
-	_mm_prefetch(pf+i,_MM_HINT_T0);
-      }
-      ret._odata[o+b]=rhs._odata[so+b];
-    }
-  }
-  return ret;
-}
-#endif
-#endif
 
diff --git a/Makefile.am b/Makefile.am
index b9763db1..b681e0c5 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -23,8 +23,20 @@ include_HEADERS = Grid.h\
 # Test code
 #
 bin_PROGRAMS = Grid_main
+
+extra_sources=
+if BUILD_COMMS_MPI
+  extra_sources+=Grid_mpi.cc
+endif
+if BUILD_COMMS_FAKE
+  extra_sources+=Grid_fake.cc
+endif
+if BUILD_COMMS_NONE
+  extra_sources+=Grid_fake.cc
+endif
+
 Grid_main_SOURCES = \
 	Grid_main.cc\
-	Grid_fake.cc
+	$(extra_sources)
 
 Grid_main_LDADD = libGrid.a
diff --git a/Makefile.in b/Makefile.in
index 9a79034e..43ea6616 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -89,6 +89,9 @@ NORMAL_UNINSTALL = :
 PRE_UNINSTALL = :
 POST_UNINSTALL = :
 bin_PROGRAMS = Grid_main$(EXEEXT)
+@BUILD_COMMS_MPI_TRUE@am__append_1 = Grid_mpi.cc
+@BUILD_COMMS_FAKE_TRUE@am__append_2 = Grid_fake.cc
+@BUILD_COMMS_NONE_TRUE@am__append_3 = Grid_fake.cc
 subdir = .
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/configure.ac
@@ -143,7 +146,12 @@ libGrid_a_LIBADD =
 am_libGrid_a_OBJECTS = Grid_init.$(OBJEXT)
 libGrid_a_OBJECTS = $(am_libGrid_a_OBJECTS)
 PROGRAMS = $(bin_PROGRAMS)
-am_Grid_main_OBJECTS = Grid_main.$(OBJEXT) Grid_fake.$(OBJEXT)
+am__Grid_main_SOURCES_DIST = Grid_main.cc Grid_mpi.cc Grid_fake.cc
+@BUILD_COMMS_MPI_TRUE@am__objects_1 = Grid_mpi.$(OBJEXT)
+@BUILD_COMMS_FAKE_TRUE@am__objects_2 = Grid_fake.$(OBJEXT)
+@BUILD_COMMS_NONE_TRUE@am__objects_3 = Grid_fake.$(OBJEXT)
+am__objects_4 = $(am__objects_1) $(am__objects_2) $(am__objects_3)
+am_Grid_main_OBJECTS = Grid_main.$(OBJEXT) $(am__objects_4)
 Grid_main_OBJECTS = $(am_Grid_main_OBJECTS)
 Grid_main_DEPENDENCIES = libGrid.a
 AM_V_P = $(am__v_P_@AM_V@)
@@ -176,7 +184,7 @@ am__v_CXXLD_ = $(am__v_CXXLD_@AM_DEFAULT_V@)
 am__v_CXXLD_0 = @echo "  CXXLD   " $@;
 am__v_CXXLD_1 = 
 SOURCES = $(libGrid_a_SOURCES) $(Grid_main_SOURCES)
-DIST_SOURCES = $(libGrid_a_SOURCES) $(Grid_main_SOURCES)
+DIST_SOURCES = $(libGrid_a_SOURCES) $(am__Grid_main_SOURCES_DIST)
 am__can_run_installinfo = \
   case $$AM_UPDATE_INFO_DIR in \
     n|no|NO) false;; \
@@ -340,9 +348,10 @@ include_HEADERS = Grid.h\
 	Grid_Lattice.h\
 	Grid_config.h
 
+extra_sources = $(am__append_1) $(am__append_2) $(am__append_3)
 Grid_main_SOURCES = \
 	Grid_main.cc\
-	Grid_fake.cc
+	$(extra_sources)
 
 Grid_main_LDADD = libGrid.a
 all: Grid_config.h
@@ -490,6 +499,7 @@ distclean-compile:
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_fake.Po@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_init.Po@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_main.Po@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_mpi.Po@am__quote@
 
 .cc.o:
 @am__fastdepCXX_TRUE@	$(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
diff --git a/aclocal.m4 b/aclocal.m4
index bcc213b4..f3018f6c 100644
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -1,7 +1,7 @@
-# generated automatically by aclocal 1.11.1 -*- Autoconf -*-
+# generated automatically by aclocal 1.15 -*- Autoconf -*-
+
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 
-# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
-# 2005, 2006, 2007, 2008, 2009  Free Software Foundation, Inc.
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
@@ -11,15 +11,16 @@
 # even the implied warranty of MERCHANTABILITY or FITNESS FOR A
 # PARTICULAR PURPOSE.
 
+m4_ifndef([AC_CONFIG_MACRO_DIRS], [m4_defun([_AM_CONFIG_MACRO_DIRS], [])m4_defun([AC_CONFIG_MACRO_DIRS], [_AM_CONFIG_MACRO_DIRS($@)])])
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
-m4_if(m4_defn([AC_AUTOCONF_VERSION]), [2.63],,
-[m4_warning([this file was generated for autoconf 2.63.
+m4_if(m4_defn([AC_AUTOCONF_VERSION]), [2.69],,
+[m4_warning([this file was generated for autoconf 2.69.
 You have another version of autoconf.  It may work, but is not guaranteed to.
 If you have problems, you may need to regenerate the build system entirely.
-To do so, use the procedure documented by the package, typically `autoreconf'.])])
+To do so, use the procedure documented by the package, typically 'autoreconf'.])])
 
-# Copyright (C) 2002, 2003, 2005, 2006, 2007, 2008  Free Software Foundation, Inc.
+# Copyright (C) 2002-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -31,10 +32,10 @@ To do so, use the procedure documented by the package, typically `autoreconf'.])
 # generated from the m4 files accompanying Automake X.Y.
 # (This private macro should not be called outside this file.)
 AC_DEFUN([AM_AUTOMAKE_VERSION],
-[am__api_version='1.11'
+[am__api_version='1.15'
 dnl Some users find AM_AUTOMAKE_VERSION and mistake it for a way to
 dnl require some minimum version.  Point them to the right macro.
-m4_if([$1], [1.11.1], [],
+m4_if([$1], [1.15], [],
       [AC_FATAL([Do not call $0, use AM_INIT_AUTOMAKE([$1]).])])dnl
 ])
 
@@ -50,22 +51,22 @@ m4_define([_AM_AUTOCONF_VERSION], [])
 # Call AM_AUTOMAKE_VERSION and AM_AUTOMAKE_VERSION so they can be traced.
 # This function is AC_REQUIREd by AM_INIT_AUTOMAKE.
 AC_DEFUN([AM_SET_CURRENT_AUTOMAKE_VERSION],
-[AM_AUTOMAKE_VERSION([1.11.1])dnl
+[AM_AUTOMAKE_VERSION([1.15])dnl
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
 _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 
 # AM_AUX_DIR_EXPAND                                         -*- Autoconf -*-
 
-# Copyright (C) 2001, 2003, 2005  Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
 # For projects using AC_CONFIG_AUX_DIR([foo]), Autoconf sets
-# $ac_aux_dir to `$srcdir/foo'.  In other projects, it is set to
-# `$srcdir', `$srcdir/..', or `$srcdir/../..'.
+# $ac_aux_dir to '$srcdir/foo'.  In other projects, it is set to
+# '$srcdir', '$srcdir/..', or '$srcdir/../..'.
 #
 # Of course, Automake must honor this variable whenever it calls a
 # tool from the auxiliary directory.  The problem is that $srcdir (and
@@ -84,7 +85,7 @@ _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 #
 # The reason of the latter failure is that $top_srcdir and $ac_aux_dir
 # are both prefixed by $srcdir.  In an in-source build this is usually
-# harmless because $srcdir is `.', but things will broke when you
+# harmless because $srcdir is '.', but things will broke when you
 # start a VPATH build or use an absolute $srcdir.
 #
 # So we could use something similar to $top_srcdir/$ac_aux_dir/missing,
@@ -102,30 +103,26 @@ _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 # configured tree to be moved without reconfiguration.
 
 AC_DEFUN([AM_AUX_DIR_EXPAND],
-[dnl Rely on autoconf to set up CDPATH properly.
-AC_PREREQ([2.50])dnl
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+[AC_REQUIRE([AC_CONFIG_AUX_DIR_DEFAULT])dnl
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 ])
 
 # AM_CONDITIONAL                                            -*- Autoconf -*-
 
-# Copyright (C) 1997, 2000, 2001, 2003, 2004, 2005, 2006, 2008
-# Free Software Foundation, Inc.
+# Copyright (C) 1997-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 9
-
 # AM_CONDITIONAL(NAME, SHELL-CONDITION)
 # -------------------------------------
 # Define a conditional.
 AC_DEFUN([AM_CONDITIONAL],
-[AC_PREREQ(2.52)dnl
- ifelse([$1], [TRUE],  [AC_FATAL([$0: invalid condition: $1])],
-	[$1], [FALSE], [AC_FATAL([$0: invalid condition: $1])])dnl
+[AC_PREREQ([2.52])dnl
+ m4_if([$1], [TRUE],  [AC_FATAL([$0: invalid condition: $1])],
+       [$1], [FALSE], [AC_FATAL([$0: invalid condition: $1])])dnl
 AC_SUBST([$1_TRUE])dnl
 AC_SUBST([$1_FALSE])dnl
 _AM_SUBST_NOTMAKE([$1_TRUE])dnl
@@ -144,16 +141,14 @@ AC_CONFIG_COMMANDS_PRE(
 Usually this means the macro was only invoked conditionally.]])
 fi])])
 
-# Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2009
-# Free Software Foundation, Inc.
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 10
 
-# There are a few dirty hacks below to avoid letting `AC_PROG_CC' be
+# There are a few dirty hacks below to avoid letting 'AC_PROG_CC' be
 # written in clear, in which case automake, when reading aclocal.m4,
 # will think it sees a *use*, and therefore will trigger all it's
 # C support machinery.  Also note that it means that autoscan, seeing
@@ -163,7 +158,7 @@ fi])])
 # _AM_DEPENDENCIES(NAME)
 # ----------------------
 # See how the compiler implements dependency checking.
-# NAME is "CC", "CXX", "GCJ", or "OBJC".
+# NAME is "CC", "CXX", "OBJC", "OBJCXX", "UPC", or "GJC".
 # We try a few techniques and use that to set a single cache variable.
 #
 # We don't AC_REQUIRE the corresponding AC_PROG_CC since the latter was
@@ -176,12 +171,13 @@ AC_REQUIRE([AM_OUTPUT_DEPENDENCY_COMMANDS])dnl
 AC_REQUIRE([AM_MAKE_INCLUDE])dnl
 AC_REQUIRE([AM_DEP_TRACK])dnl
 
-ifelse([$1], CC,   [depcc="$CC"   am_compiler_list=],
-       [$1], CXX,  [depcc="$CXX"  am_compiler_list=],
-       [$1], OBJC, [depcc="$OBJC" am_compiler_list='gcc3 gcc'],
-       [$1], UPC,  [depcc="$UPC"  am_compiler_list=],
-       [$1], GCJ,  [depcc="$GCJ"  am_compiler_list='gcc3 gcc'],
-                   [depcc="$$1"   am_compiler_list=])
+m4_if([$1], [CC],   [depcc="$CC"   am_compiler_list=],
+      [$1], [CXX],  [depcc="$CXX"  am_compiler_list=],
+      [$1], [OBJC], [depcc="$OBJC" am_compiler_list='gcc3 gcc'],
+      [$1], [OBJCXX], [depcc="$OBJCXX" am_compiler_list='gcc3 gcc'],
+      [$1], [UPC],  [depcc="$UPC"  am_compiler_list=],
+      [$1], [GCJ],  [depcc="$GCJ"  am_compiler_list='gcc3 gcc'],
+                    [depcc="$$1"   am_compiler_list=])
 
 AC_CACHE_CHECK([dependency style of $depcc],
                [am_cv_$1_dependencies_compiler_type],
@@ -189,8 +185,9 @@ AC_CACHE_CHECK([dependency style of $depcc],
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named `D' -- because `-MD' means `put the output
-  # in D'.
+  # making a dummy file named 'D' -- because '-MD' means "put the output
+  # in D".
+  rm -rf conftest.dir
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -229,16 +226,16 @@ AC_CACHE_CHECK([dependency style of $depcc],
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
-      # Solaris 8's {/usr,}/bin/sh.
-      touch sub/conftst$i.h
+      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
+      # Solaris 10 /bin/sh.
+      echo '/* dummy */' > sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with `-c' and `-o' for the sake of the "dashmstdout"
+    # We check with '-c' and '-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle `-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs
+    # handle '-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs.
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -247,16 +244,16 @@ AC_CACHE_CHECK([dependency style of $depcc],
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # after this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested
+      # After this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested.
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvisualcpp | msvcmsys)
-      # This compiler won't grok `-c -o', but also, the minuso test has
+    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
+      # This compiler won't grok '-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -304,7 +301,7 @@ AM_CONDITIONAL([am__fastdep$1], [
 # AM_SET_DEPDIR
 # -------------
 # Choose a directory name for dependency files.
-# This macro is AC_REQUIREd in _AM_DEPENDENCIES
+# This macro is AC_REQUIREd in _AM_DEPENDENCIES.
 AC_DEFUN([AM_SET_DEPDIR],
 [AC_REQUIRE([AM_SET_LEADING_DOT])dnl
 AC_SUBST([DEPDIR], ["${am__leading_dot}deps"])dnl
@@ -314,34 +311,39 @@ AC_SUBST([DEPDIR], ["${am__leading_dot}deps"])dnl
 # AM_DEP_TRACK
 # ------------
 AC_DEFUN([AM_DEP_TRACK],
-[AC_ARG_ENABLE(dependency-tracking,
-[  --disable-dependency-tracking  speeds up one-time build
-  --enable-dependency-tracking   do not reject slow dependency extractors])
+[AC_ARG_ENABLE([dependency-tracking], [dnl
+AS_HELP_STRING(
+  [--enable-dependency-tracking],
+  [do not reject slow dependency extractors])
+AS_HELP_STRING(
+  [--disable-dependency-tracking],
+  [speeds up one-time build])])
 if test "x$enable_dependency_tracking" != xno; then
   am_depcomp="$ac_aux_dir/depcomp"
   AMDEPBACKSLASH='\'
+  am__nodep='_no'
 fi
 AM_CONDITIONAL([AMDEP], [test "x$enable_dependency_tracking" != xno])
 AC_SUBST([AMDEPBACKSLASH])dnl
 _AM_SUBST_NOTMAKE([AMDEPBACKSLASH])dnl
+AC_SUBST([am__nodep])dnl
+_AM_SUBST_NOTMAKE([am__nodep])dnl
 ])
 
 # Generate code to set up dependency tracking.              -*- Autoconf -*-
 
-# Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2008
-# Free Software Foundation, Inc.
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-#serial 5
 
 # _AM_OUTPUT_DEPENDENCY_COMMANDS
 # ------------------------------
 AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
 [{
-  # Autoconf 2.62 quotes --file arguments for eval, but not when files
+  # Older Autoconf quotes --file arguments for eval, but not when files
   # are listed without --file.  Let's play safe and only enable the eval
   # if we detect the quoting.
   case $CONFIG_FILES in
@@ -354,7 +356,7 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
     # Strip MF so we end up with the name of the file.
     mf=`echo "$mf" | sed -e 's/:.*$//'`
     # Check whether this is an Automake generated Makefile or not.
-    # We used to match only the files named `Makefile.in', but
+    # We used to match only the files named 'Makefile.in', but
     # some people rename them; so instead we look at the file content.
     # Grep'ing the first line is not enough: some people post-process
     # each Makefile.in and add a new line on top of each file to say so.
@@ -366,21 +368,19 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
       continue
     fi
     # Extract the definition of DEPDIR, am__include, and am__quote
-    # from the Makefile without running `make'.
+    # from the Makefile without running 'make'.
     DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
     test -z "$DEPDIR" && continue
     am__include=`sed -n 's/^am__include = //p' < "$mf"`
-    test -z "am__include" && continue
+    test -z "$am__include" && continue
     am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
-    # When using ansi2knr, U may be empty or an underscore; expand it
-    U=`sed -n 's/^U = //p' < "$mf"`
     # Find all dependency output files, they are included files with
     # $(DEPDIR) in their names.  We invoke sed twice because it is the
     # simplest approach to changing $(DEPDIR) to its actual value in the
     # expansion.
     for file in `sed -n "
       s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
-	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g' -e 's/\$U/'"$U"'/g'`; do
+	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
       # Make sure the directory exists.
       test -f "$dirpart/$file" && continue
       fdir=`AS_DIRNAME(["$file"])`
@@ -398,7 +398,7 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
 # This macro should only be invoked once -- use via AC_REQUIRE.
 #
 # This code is only required when automatic dependency tracking
-# is enabled.  FIXME.  This creates each `.P' file that we will
+# is enabled.  FIXME.  This creates each '.P' file that we will
 # need in order to bootstrap the dependency handling code.
 AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 [AC_CONFIG_COMMANDS([depfiles],
@@ -408,18 +408,21 @@ AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 
 # Do all the work for Automake.                             -*- Autoconf -*-
 
-# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
-# 2005, 2006, 2008, 2009 Free Software Foundation, Inc.
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 16
-
 # This macro actually does too much.  Some checks are only needed if
 # your package does certain things.  But this isn't really a big deal.
 
+dnl Redefine AC_PROG_CC to automatically invoke _AM_PROG_CC_C_O.
+m4_define([AC_PROG_CC],
+m4_defn([AC_PROG_CC])
+[_AM_PROG_CC_C_O
+])
+
 # AM_INIT_AUTOMAKE(PACKAGE, VERSION, [NO-DEFINE])
 # AM_INIT_AUTOMAKE([OPTIONS])
 # -----------------------------------------------
@@ -432,7 +435,7 @@ AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 # arguments mandatory, and then we can depend on a new Autoconf
 # release and drop the old call support.
 AC_DEFUN([AM_INIT_AUTOMAKE],
-[AC_PREREQ([2.62])dnl
+[AC_PREREQ([2.65])dnl
 dnl Autoconf wants to disallow AM_ names.  We explicitly allow
 dnl the ones we care about.
 m4_pattern_allow([^AM_[A-Z]+FLAGS$])dnl
@@ -461,33 +464,42 @@ AC_SUBST([CYGPATH_W])
 # Define the identity of the package.
 dnl Distinguish between old-style and new-style calls.
 m4_ifval([$2],
-[m4_ifval([$3], [_AM_SET_OPTION([no-define])])dnl
+[AC_DIAGNOSE([obsolete],
+             [$0: two- and three-arguments forms are deprecated.])
+m4_ifval([$3], [_AM_SET_OPTION([no-define])])dnl
  AC_SUBST([PACKAGE], [$1])dnl
  AC_SUBST([VERSION], [$2])],
 [_AM_SET_OPTIONS([$1])dnl
 dnl Diagnose old-style AC_INIT with new-style AM_AUTOMAKE_INIT.
-m4_if(m4_ifdef([AC_PACKAGE_NAME], 1)m4_ifdef([AC_PACKAGE_VERSION], 1), 11,,
+m4_if(
+  m4_ifdef([AC_PACKAGE_NAME], [ok]):m4_ifdef([AC_PACKAGE_VERSION], [ok]),
+  [ok:ok],,
   [m4_fatal([AC_INIT should be called with package and version arguments])])dnl
  AC_SUBST([PACKAGE], ['AC_PACKAGE_TARNAME'])dnl
  AC_SUBST([VERSION], ['AC_PACKAGE_VERSION'])])dnl
 
 _AM_IF_OPTION([no-define],,
-[AC_DEFINE_UNQUOTED(PACKAGE, "$PACKAGE", [Name of package])
- AC_DEFINE_UNQUOTED(VERSION, "$VERSION", [Version number of package])])dnl
+[AC_DEFINE_UNQUOTED([PACKAGE], ["$PACKAGE"], [Name of package])
+ AC_DEFINE_UNQUOTED([VERSION], ["$VERSION"], [Version number of package])])dnl
 
 # Some tools Automake needs.
 AC_REQUIRE([AM_SANITY_CHECK])dnl
 AC_REQUIRE([AC_ARG_PROGRAM])dnl
-AM_MISSING_PROG(ACLOCAL, aclocal-${am__api_version})
-AM_MISSING_PROG(AUTOCONF, autoconf)
-AM_MISSING_PROG(AUTOMAKE, automake-${am__api_version})
-AM_MISSING_PROG(AUTOHEADER, autoheader)
-AM_MISSING_PROG(MAKEINFO, makeinfo)
+AM_MISSING_PROG([ACLOCAL], [aclocal-${am__api_version}])
+AM_MISSING_PROG([AUTOCONF], [autoconf])
+AM_MISSING_PROG([AUTOMAKE], [automake-${am__api_version}])
+AM_MISSING_PROG([AUTOHEADER], [autoheader])
+AM_MISSING_PROG([MAKEINFO], [makeinfo])
 AC_REQUIRE([AM_PROG_INSTALL_SH])dnl
 AC_REQUIRE([AM_PROG_INSTALL_STRIP])dnl
-AC_REQUIRE([AM_PROG_MKDIR_P])dnl
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
+AC_REQUIRE([AC_PROG_MKDIR_P])dnl
+# For better backward compatibility.  To be removed once Automake 1.9.x
+# dies out for good.  For more background, see:
+# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
+# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
+AC_SUBST([mkdir_p], ['$(MKDIR_P)'])
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
 AC_REQUIRE([AC_PROG_AWK])dnl
 AC_REQUIRE([AC_PROG_MAKE_SET])dnl
 AC_REQUIRE([AM_SET_LEADING_DOT])dnl
@@ -496,34 +508,82 @@ _AM_IF_OPTION([tar-ustar], [_AM_PROG_TAR([ustar])],
 			     [_AM_PROG_TAR([v7])])])
 _AM_IF_OPTION([no-dependencies],,
 [AC_PROVIDE_IFELSE([AC_PROG_CC],
-		  [_AM_DEPENDENCIES(CC)],
-		  [define([AC_PROG_CC],
-			  defn([AC_PROG_CC])[_AM_DEPENDENCIES(CC)])])dnl
+		  [_AM_DEPENDENCIES([CC])],
+		  [m4_define([AC_PROG_CC],
+			     m4_defn([AC_PROG_CC])[_AM_DEPENDENCIES([CC])])])dnl
 AC_PROVIDE_IFELSE([AC_PROG_CXX],
-		  [_AM_DEPENDENCIES(CXX)],
-		  [define([AC_PROG_CXX],
-			  defn([AC_PROG_CXX])[_AM_DEPENDENCIES(CXX)])])dnl
+		  [_AM_DEPENDENCIES([CXX])],
+		  [m4_define([AC_PROG_CXX],
+			     m4_defn([AC_PROG_CXX])[_AM_DEPENDENCIES([CXX])])])dnl
 AC_PROVIDE_IFELSE([AC_PROG_OBJC],
-		  [_AM_DEPENDENCIES(OBJC)],
-		  [define([AC_PROG_OBJC],
-			  defn([AC_PROG_OBJC])[_AM_DEPENDENCIES(OBJC)])])dnl
+		  [_AM_DEPENDENCIES([OBJC])],
+		  [m4_define([AC_PROG_OBJC],
+			     m4_defn([AC_PROG_OBJC])[_AM_DEPENDENCIES([OBJC])])])dnl
+AC_PROVIDE_IFELSE([AC_PROG_OBJCXX],
+		  [_AM_DEPENDENCIES([OBJCXX])],
+		  [m4_define([AC_PROG_OBJCXX],
+			     m4_defn([AC_PROG_OBJCXX])[_AM_DEPENDENCIES([OBJCXX])])])dnl
 ])
-_AM_IF_OPTION([silent-rules], [AC_REQUIRE([AM_SILENT_RULES])])dnl
-dnl The `parallel-tests' driver may need to know about EXEEXT, so add the
-dnl `am__EXEEXT' conditional if _AM_COMPILER_EXEEXT was seen.  This macro
-dnl is hooked onto _AC_COMPILER_EXEEXT early, see below.
+AC_REQUIRE([AM_SILENT_RULES])dnl
+dnl The testsuite driver may need to know about EXEEXT, so add the
+dnl 'am__EXEEXT' conditional if _AM_COMPILER_EXEEXT was seen.  This
+dnl macro is hooked onto _AC_COMPILER_EXEEXT early, see below.
 AC_CONFIG_COMMANDS_PRE(dnl
 [m4_provide_if([_AM_COMPILER_EXEEXT],
   [AM_CONDITIONAL([am__EXEEXT], [test -n "$EXEEXT"])])])dnl
+
+# POSIX will say in a future version that running "rm -f" with no argument
+# is OK; and we want to be able to make that assumption in our Makefile
+# recipes.  So use an aggressive probe to check that the usage we want is
+# actually supported "in the wild" to an acceptable degree.
+# See automake bug#10828.
+# To make any issue more visible, cause the running configure to be aborted
+# by default if the 'rm' program in use doesn't match our expectations; the
+# user can still override this though.
+if rm -f && rm -fr && rm -rf; then : OK; else
+  cat >&2 <<'END'
+Oops!
+
+Your 'rm' program seems unable to run without file operands specified
+on the command line, even when the '-f' option is present.  This is contrary
+to the behaviour of most rm programs out there, and not conforming with
+the upcoming POSIX standard: <http://austingroupbugs.net/view.php?id=542>
+
+Please tell bug-automake@gnu.org about your system, including the value
+of your $PATH and any error possibly output before this message.  This
+can help us improve future automake versions.
+
+END
+  if test x"$ACCEPT_INFERIOR_RM_PROGRAM" = x"yes"; then
+    echo 'Configuration will proceed anyway, since you have set the' >&2
+    echo 'ACCEPT_INFERIOR_RM_PROGRAM variable to "yes"' >&2
+    echo >&2
+  else
+    cat >&2 <<'END'
+Aborting the configuration process, to ensure you take notice of the issue.
+
+You can download and install GNU coreutils to get an 'rm' implementation
+that behaves properly: <http://www.gnu.org/software/coreutils/>.
+
+If you want to complete the configuration process using your problematic
+'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
+to "yes", and re-run configure.
+
+END
+    AC_MSG_ERROR([Your 'rm' program is bad, sorry.])
+  fi
+fi
+dnl The trailing newline in this macro's definition is deliberate, for
+dnl backward compatibility and to allow trailing 'dnl'-style comments
+dnl after the AM_INIT_AUTOMAKE invocation. See automake bug#16841.
 ])
 
-dnl Hook into `_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
+dnl Hook into '_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
 dnl add the conditional right here, as _AC_COMPILER_EXEEXT may be further
 dnl mangled by Autoconf and run in a shell conditional statement.
 m4_define([_AC_COMPILER_EXEEXT],
 m4_defn([_AC_COMPILER_EXEEXT])[m4_provide([_AM_COMPILER_EXEEXT])])
 
-
 # When config.status generates a header, we must update the stamp-h file.
 # This file resides in the same directory as the config header
 # that is generated.  The stamp files are numbered to have different names.
@@ -545,7 +605,7 @@ for _am_header in $config_headers :; do
 done
 echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_count])
 
-# Copyright (C) 2001, 2003, 2005, 2008  Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -556,7 +616,7 @@ echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_co
 # Define $install_sh.
 AC_DEFUN([AM_PROG_INSTALL_SH],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -564,16 +624,14 @@ if test x"${install_sh}" != xset; then
     install_sh="\${SHELL} $am_aux_dir/install-sh"
   esac
 fi
-AC_SUBST(install_sh)])
+AC_SUBST([install_sh])])
 
-# Copyright (C) 2003, 2005  Free Software Foundation, Inc.
+# Copyright (C) 2003-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 2
-
 # Check whether the underlying file-system supports filenames
 # with a leading dot.  For instance MS-DOS doesn't.
 AC_DEFUN([AM_SET_LEADING_DOT],
@@ -589,14 +647,12 @@ AC_SUBST([am__leading_dot])])
 
 # Check to see how 'make' treats includes.	            -*- Autoconf -*-
 
-# Copyright (C) 2001, 2002, 2003, 2005, 2009  Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 4
-
 # AM_MAKE_INCLUDE()
 # -----------------
 # Check to see how make treats includes.
@@ -614,7 +670,7 @@ am__quote=
 _am_result=none
 # First try GNU make style include.
 echo "include confinc" > confmf
-# Ignore all kinds of additional output from `make'.
+# Ignore all kinds of additional output from 'make'.
 case `$am_make -s -f confmf 2> /dev/null` in #(
 *the\ am__doit\ target*)
   am__include=include
@@ -641,15 +697,12 @@ rm -f confinc confmf
 
 # Fake the existence of programs that GNU maintainers use.  -*- Autoconf -*-
 
-# Copyright (C) 1997, 1999, 2000, 2001, 2003, 2004, 2005, 2008
-# Free Software Foundation, Inc.
+# Copyright (C) 1997-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 6
-
 # AM_MISSING_PROG(NAME, PROGRAM)
 # ------------------------------
 AC_DEFUN([AM_MISSING_PROG],
@@ -657,11 +710,10 @@ AC_DEFUN([AM_MISSING_PROG],
 $1=${$1-"${am_missing_run}$2"}
 AC_SUBST($1)])
 
-
 # AM_MISSING_HAS_RUN
 # ------------------
-# Define MISSING if not defined so far and test if it supports --run.
-# If it does, set am_missing_run to use it, otherwise, to nothing.
+# Define MISSING if not defined so far and test if it is modern enough.
+# If it is, set am_missing_run to use it, otherwise, to nothing.
 AC_DEFUN([AM_MISSING_HAS_RUN],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
 AC_REQUIRE_AUX_FILE([missing])dnl
@@ -674,63 +726,35 @@ if test x"${MISSING+set}" != xset; then
   esac
 fi
 # Use eval to expand $SHELL
-if eval "$MISSING --run true"; then
-  am_missing_run="$MISSING --run "
+if eval "$MISSING --is-lightweight"; then
+  am_missing_run="$MISSING "
 else
   am_missing_run=
-  AC_MSG_WARN([`missing' script is too old or missing])
+  AC_MSG_WARN(['missing' script is too old or missing])
 fi
 ])
 
-# Copyright (C) 2003, 2004, 2005, 2006  Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# AM_PROG_MKDIR_P
-# ---------------
-# Check for `mkdir -p'.
-AC_DEFUN([AM_PROG_MKDIR_P],
-[AC_PREREQ([2.60])dnl
-AC_REQUIRE([AC_PROG_MKDIR_P])dnl
-dnl Automake 1.8 to 1.9.6 used to define mkdir_p.  We now use MKDIR_P,
-dnl while keeping a definition of mkdir_p for backward compatibility.
-dnl @MKDIR_P@ is magic: AC_OUTPUT adjusts its value for each Makefile.
-dnl However we cannot define mkdir_p as $(MKDIR_P) for the sake of
-dnl Makefile.ins that do not define MKDIR_P, so we do our own
-dnl adjustment using top_builddir (which is defined more often than
-dnl MKDIR_P).
-AC_SUBST([mkdir_p], ["$MKDIR_P"])dnl
-case $mkdir_p in
-  [[\\/$]]* | ?:[[\\/]]*) ;;
-  */*) mkdir_p="\$(top_builddir)/$mkdir_p" ;;
-esac
-])
-
 # Helper functions for option handling.                     -*- Autoconf -*-
 
-# Copyright (C) 2001, 2002, 2003, 2005, 2008  Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 4
-
 # _AM_MANGLE_OPTION(NAME)
 # -----------------------
 AC_DEFUN([_AM_MANGLE_OPTION],
 [[_AM_OPTION_]m4_bpatsubst($1, [[^a-zA-Z0-9_]], [_])])
 
 # _AM_SET_OPTION(NAME)
-# ------------------------------
+# --------------------
 # Set option NAME.  Presently that only means defining a flag for this option.
 AC_DEFUN([_AM_SET_OPTION],
-[m4_define(_AM_MANGLE_OPTION([$1]), 1)])
+[m4_define(_AM_MANGLE_OPTION([$1]), [1])])
 
 # _AM_SET_OPTIONS(OPTIONS)
-# ----------------------------------
+# ------------------------
 # OPTIONS is a space-separated list of Automake options.
 AC_DEFUN([_AM_SET_OPTIONS],
 [m4_foreach_w([_AM_Option], [$1], [_AM_SET_OPTION(_AM_Option)])])
@@ -741,24 +765,82 @@ AC_DEFUN([_AM_SET_OPTIONS],
 AC_DEFUN([_AM_IF_OPTION],
 [m4_ifset(_AM_MANGLE_OPTION([$1]), [$2], [$3])])
 
-# Check to make sure that the build environment is sane.    -*- Autoconf -*-
-
-# Copyright (C) 1996, 1997, 2000, 2001, 2003, 2005, 2008
-# Free Software Foundation, Inc.
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 5
+# _AM_PROG_CC_C_O
+# ---------------
+# Like AC_PROG_CC_C_O, but changed for automake.  We rewrite AC_PROG_CC
+# to automatically call this.
+AC_DEFUN([_AM_PROG_CC_C_O],
+[AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
+AC_REQUIRE_AUX_FILE([compile])dnl
+AC_LANG_PUSH([C])dnl
+AC_CACHE_CHECK(
+  [whether $CC understands -c and -o together],
+  [am_cv_prog_cc_c_o],
+  [AC_LANG_CONFTEST([AC_LANG_PROGRAM([])])
+  # Make sure it works both with $CC and with simple cc.
+  # Following AC_PROG_CC_C_O, we do the test twice because some
+  # compilers refuse to overwrite an existing .o file with -o,
+  # though they will create one.
+  am_cv_prog_cc_c_o=yes
+  for am_i in 1 2; do
+    if AM_RUN_LOG([$CC -c conftest.$ac_ext -o conftest2.$ac_objext]) \
+         && test -f conftest2.$ac_objext; then
+      : OK
+    else
+      am_cv_prog_cc_c_o=no
+      break
+    fi
+  done
+  rm -f core conftest*
+  unset am_i])
+if test "$am_cv_prog_cc_c_o" != yes; then
+   # Losing compiler, so override with the script.
+   # FIXME: It is wrong to rewrite CC.
+   # But if we don't then we get into trouble of one sort or another.
+   # A longer-term fix would be to have automake use am__CC in this case,
+   # and then we could set am__CC="\$(top_srcdir)/compile \$(CC)"
+   CC="$am_aux_dir/compile $CC"
+fi
+AC_LANG_POP([C])])
+
+# For backward compatibility.
+AC_DEFUN_ONCE([AM_PROG_CC_C_O], [AC_REQUIRE([AC_PROG_CC])])
+
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# AM_RUN_LOG(COMMAND)
+# -------------------
+# Run COMMAND, save the exit status in ac_status, and log it.
+# (This has been adapted from Autoconf's _AC_RUN_LOG macro.)
+AC_DEFUN([AM_RUN_LOG],
+[{ echo "$as_me:$LINENO: $1" >&AS_MESSAGE_LOG_FD
+   ($1) >&AS_MESSAGE_LOG_FD 2>&AS_MESSAGE_LOG_FD
+   ac_status=$?
+   echo "$as_me:$LINENO: \$? = $ac_status" >&AS_MESSAGE_LOG_FD
+   (exit $ac_status); }])
+
+# Check to make sure that the build environment is sane.    -*- Autoconf -*-
+
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
 
 # AM_SANITY_CHECK
 # ---------------
 AC_DEFUN([AM_SANITY_CHECK],
 [AC_MSG_CHECKING([whether build environment is sane])
-# Just in case
-sleep 1
-echo timestamp > conftest.file
 # Reject unsafe characters in $srcdir or the absolute working directory
 # name.  Accept space and tab only in the latter.
 am_lf='
@@ -769,32 +851,40 @@ case `pwd` in
 esac
 case $srcdir in
   *[[\\\"\#\$\&\'\`$am_lf\ \	]]*)
-    AC_MSG_ERROR([unsafe srcdir value: `$srcdir']);;
+    AC_MSG_ERROR([unsafe srcdir value: '$srcdir']);;
 esac
 
-# Do `set' in a subshell so we don't clobber the current shell's
+# Do 'set' in a subshell so we don't clobber the current shell's
 # arguments.  Must try -L first in case configure is actually a
 # symlink; some systems play weird games with the mod time of symlinks
 # (eg FreeBSD returns the mod time of the symlink's containing
 # directory).
 if (
-   set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
-   if test "$[*]" = "X"; then
-      # -L didn't work.
-      set X `ls -t "$srcdir/configure" conftest.file`
-   fi
-   rm -f conftest.file
-   if test "$[*]" != "X $srcdir/configure conftest.file" \
-      && test "$[*]" != "X conftest.file $srcdir/configure"; then
-
-      # If neither matched, then we have a broken ls.  This can happen
-      # if, for instance, CONFIG_SHELL is bash and it inherits a
-      # broken ls alias from the environment.  This has actually
-      # happened.  Such a system could not be considered "sane".
-      AC_MSG_ERROR([ls -t appears to fail.  Make sure there is not a broken
-alias in your environment])
-   fi
+   am_has_slept=no
+   for am_try in 1 2; do
+     echo "timestamp, slept: $am_has_slept" > conftest.file
+     set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
+     if test "$[*]" = "X"; then
+	# -L didn't work.
+	set X `ls -t "$srcdir/configure" conftest.file`
+     fi
+     if test "$[*]" != "X $srcdir/configure conftest.file" \
+	&& test "$[*]" != "X conftest.file $srcdir/configure"; then
 
+	# If neither matched, then we have a broken ls.  This can happen
+	# if, for instance, CONFIG_SHELL is bash and it inherits a
+	# broken ls alias from the environment.  This has actually
+	# happened.  Such a system could not be considered "sane".
+	AC_MSG_ERROR([ls -t appears to fail.  Make sure there is not a broken
+  alias in your environment])
+     fi
+     if test "$[2]" = conftest.file || test $am_try -eq 2; then
+       break
+     fi
+     # Just in case.
+     sleep 1
+     am_has_slept=yes
+   done
    test "$[2]" = conftest.file
    )
 then
@@ -804,9 +894,85 @@ else
    AC_MSG_ERROR([newly created file is older than distributed files!
 Check your system clock])
 fi
-AC_MSG_RESULT(yes)])
+AC_MSG_RESULT([yes])
+# If we didn't sleep, we still need to ensure time stamps of config.status and
+# generated files are strictly newer.
+am_sleep_pid=
+if grep 'slept: no' conftest.file >/dev/null 2>&1; then
+  ( sleep 1 ) &
+  am_sleep_pid=$!
+fi
+AC_CONFIG_COMMANDS_PRE(
+  [AC_MSG_CHECKING([that generated files are newer than configure])
+   if test -n "$am_sleep_pid"; then
+     # Hide warnings about reused PIDs.
+     wait $am_sleep_pid 2>/dev/null
+   fi
+   AC_MSG_RESULT([done])])
+rm -f conftest.file
+])
 
-# Copyright (C) 2001, 2003, 2005  Free Software Foundation, Inc.
+# Copyright (C) 2009-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# AM_SILENT_RULES([DEFAULT])
+# --------------------------
+# Enable less verbose build rules; with the default set to DEFAULT
+# ("yes" being less verbose, "no" or empty being verbose).
+AC_DEFUN([AM_SILENT_RULES],
+[AC_ARG_ENABLE([silent-rules], [dnl
+AS_HELP_STRING(
+  [--enable-silent-rules],
+  [less verbose build output (undo: "make V=1")])
+AS_HELP_STRING(
+  [--disable-silent-rules],
+  [verbose build output (undo: "make V=0")])dnl
+])
+case $enable_silent_rules in @%:@ (((
+  yes) AM_DEFAULT_VERBOSITY=0;;
+   no) AM_DEFAULT_VERBOSITY=1;;
+    *) AM_DEFAULT_VERBOSITY=m4_if([$1], [yes], [0], [1]);;
+esac
+dnl
+dnl A few 'make' implementations (e.g., NonStop OS and NextStep)
+dnl do not support nested variable expansions.
+dnl See automake bug#9928 and bug#10237.
+am_make=${MAKE-make}
+AC_CACHE_CHECK([whether $am_make supports nested variables],
+   [am_cv_make_support_nested_variables],
+   [if AS_ECHO([['TRUE=$(BAR$(V))
+BAR0=false
+BAR1=true
+V=1
+am__doit:
+	@$(TRUE)
+.PHONY: am__doit']]) | $am_make -f - >/dev/null 2>&1; then
+  am_cv_make_support_nested_variables=yes
+else
+  am_cv_make_support_nested_variables=no
+fi])
+if test $am_cv_make_support_nested_variables = yes; then
+  dnl Using '$V' instead of '$(V)' breaks IRIX make.
+  AM_V='$(V)'
+  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
+else
+  AM_V=$AM_DEFAULT_VERBOSITY
+  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
+fi
+AC_SUBST([AM_V])dnl
+AM_SUBST_NOTMAKE([AM_V])dnl
+AC_SUBST([AM_DEFAULT_V])dnl
+AM_SUBST_NOTMAKE([AM_DEFAULT_V])dnl
+AC_SUBST([AM_DEFAULT_VERBOSITY])dnl
+AM_BACKSLASH='\'
+AC_SUBST([AM_BACKSLASH])dnl
+_AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
+])
+
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -814,34 +980,32 @@ AC_MSG_RESULT(yes)])
 
 # AM_PROG_INSTALL_STRIP
 # ---------------------
-# One issue with vendor `install' (even GNU) is that you can't
+# One issue with vendor 'install' (even GNU) is that you can't
 # specify the program used to strip binaries.  This is especially
 # annoying in cross-compiling environments, where the build's strip
 # is unlikely to handle the host's binaries.
 # Fortunately install-sh will honor a STRIPPROG variable, so we
-# always use install-sh in `make install-strip', and initialize
+# always use install-sh in "make install-strip", and initialize
 # STRIPPROG with the value of the STRIP variable (set by the user).
 AC_DEFUN([AM_PROG_INSTALL_STRIP],
 [AC_REQUIRE([AM_PROG_INSTALL_SH])dnl
-# Installed binaries are usually stripped using `strip' when the user
-# run `make install-strip'.  However `strip' might not be the right
+# Installed binaries are usually stripped using 'strip' when the user
+# run "make install-strip".  However 'strip' might not be the right
 # tool to use in cross-compilation environments, therefore Automake
-# will honor the `STRIP' environment variable to overrule this program.
-dnl Don't test for $cross_compiling = yes, because it might be `maybe'.
+# will honor the 'STRIP' environment variable to overrule this program.
+dnl Don't test for $cross_compiling = yes, because it might be 'maybe'.
 if test "$cross_compiling" != no; then
   AC_CHECK_TOOL([STRIP], [strip], :)
 fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 AC_SUBST([INSTALL_STRIP_PROGRAM])])
 
-# Copyright (C) 2006, 2008  Free Software Foundation, Inc.
+# Copyright (C) 2006-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 2
-
 # _AM_SUBST_NOTMAKE(VARIABLE)
 # ---------------------------
 # Prevent Automake from outputting VARIABLE = @VARIABLE@ in Makefile.in.
@@ -849,24 +1013,22 @@ AC_SUBST([INSTALL_STRIP_PROGRAM])])
 AC_DEFUN([_AM_SUBST_NOTMAKE])
 
 # AM_SUBST_NOTMAKE(VARIABLE)
-# ---------------------------
+# --------------------------
 # Public sister of _AM_SUBST_NOTMAKE.
 AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 
 # Check how to create a tarball.                            -*- Autoconf -*-
 
-# Copyright (C) 2004, 2005  Free Software Foundation, Inc.
+# Copyright (C) 2004-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 2
-
 # _AM_PROG_TAR(FORMAT)
 # --------------------
 # Check how to create a tarball in format FORMAT.
-# FORMAT should be one of `v7', `ustar', or `pax'.
+# FORMAT should be one of 'v7', 'ustar', or 'pax'.
 #
 # Substitute a variable $(am__tar) that is a command
 # writing to stdout a FORMAT-tarball containing the directory
@@ -876,75 +1038,114 @@ AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 # Substitute a variable $(am__untar) that extract such
 # a tarball read from stdin.
 #     $(am__untar) < result.tar
+#
 AC_DEFUN([_AM_PROG_TAR],
-[# Always define AMTAR for backward compatibility.
-AM_MISSING_PROG([AMTAR], [tar])
-m4_if([$1], [v7],
-     [am__tar='${AMTAR} chof - "$$tardir"'; am__untar='${AMTAR} xf -'],
-     [m4_case([$1], [ustar],, [pax],,
-              [m4_fatal([Unknown tar format])])
-AC_MSG_CHECKING([how to create a $1 tar archive])
-# Loop over all known methods to create a tar archive until one works.
+[# Always define AMTAR for backward compatibility.  Yes, it's still used
+# in the wild :-(  We should find a proper way to deprecate it ...
+AC_SUBST([AMTAR], ['$${TAR-tar}'])
+
+# We'll loop over all known methods to create a tar archive until one works.
 _am_tools='gnutar m4_if([$1], [ustar], [plaintar]) pax cpio none'
-_am_tools=${am_cv_prog_tar_$1-$_am_tools}
-# Do not fold the above two line into one, because Tru64 sh and
-# Solaris sh will not grok spaces in the rhs of `-'.
-for _am_tool in $_am_tools
-do
-  case $_am_tool in
-  gnutar)
-    for _am_tar in tar gnutar gtar;
-    do
-      AM_RUN_LOG([$_am_tar --version]) && break
-    done
-    am__tar="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$$tardir"'
-    am__tar_="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$tardir"'
-    am__untar="$_am_tar -xf -"
-    ;;
-  plaintar)
-    # Must skip GNU tar: if it does not support --format= it doesn't create
-    # ustar tarball either.
-    (tar --version) >/dev/null 2>&1 && continue
-    am__tar='tar chf - "$$tardir"'
-    am__tar_='tar chf - "$tardir"'
-    am__untar='tar xf -'
-    ;;
-  pax)
-    am__tar='pax -L -x $1 -w "$$tardir"'
-    am__tar_='pax -L -x $1 -w "$tardir"'
-    am__untar='pax -r'
-    ;;
-  cpio)
-    am__tar='find "$$tardir" -print | cpio -o -H $1 -L'
-    am__tar_='find "$tardir" -print | cpio -o -H $1 -L'
-    am__untar='cpio -i -H $1 -d'
-    ;;
-  none)
-    am__tar=false
-    am__tar_=false
-    am__untar=false
-    ;;
-  esac
 
-  # If the value was cached, stop now.  We just wanted to have am__tar
-  # and am__untar set.
-  test -n "${am_cv_prog_tar_$1}" && break
+m4_if([$1], [v7],
+  [am__tar='$${TAR-tar} chof - "$$tardir"' am__untar='$${TAR-tar} xf -'],
 
-  # tar/untar a dummy directory, and stop if the command works
+  [m4_case([$1],
+    [ustar],
+     [# The POSIX 1988 'ustar' format is defined with fixed-size fields.
+      # There is notably a 21 bits limit for the UID and the GID.  In fact,
+      # the 'pax' utility can hang on bigger UID/GID (see automake bug#8343
+      # and bug#13588).
+      am_max_uid=2097151 # 2^21 - 1
+      am_max_gid=$am_max_uid
+      # The $UID and $GID variables are not portable, so we need to resort
+      # to the POSIX-mandated id(1) utility.  Errors in the 'id' calls
+      # below are definitely unexpected, so allow the users to see them
+      # (that is, avoid stderr redirection).
+      am_uid=`id -u || echo unknown`
+      am_gid=`id -g || echo unknown`
+      AC_MSG_CHECKING([whether UID '$am_uid' is supported by ustar format])
+      if test $am_uid -le $am_max_uid; then
+         AC_MSG_RESULT([yes])
+      else
+         AC_MSG_RESULT([no])
+         _am_tools=none
+      fi
+      AC_MSG_CHECKING([whether GID '$am_gid' is supported by ustar format])
+      if test $am_gid -le $am_max_gid; then
+         AC_MSG_RESULT([yes])
+      else
+        AC_MSG_RESULT([no])
+        _am_tools=none
+      fi],
+
+  [pax],
+    [],
+
+  [m4_fatal([Unknown tar format])])
+
+  AC_MSG_CHECKING([how to create a $1 tar archive])
+
+  # Go ahead even if we have the value already cached.  We do so because we
+  # need to set the values for the 'am__tar' and 'am__untar' variables.
+  _am_tools=${am_cv_prog_tar_$1-$_am_tools}
+
+  for _am_tool in $_am_tools; do
+    case $_am_tool in
+    gnutar)
+      for _am_tar in tar gnutar gtar; do
+        AM_RUN_LOG([$_am_tar --version]) && break
+      done
+      am__tar="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$$tardir"'
+      am__tar_="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$tardir"'
+      am__untar="$_am_tar -xf -"
+      ;;
+    plaintar)
+      # Must skip GNU tar: if it does not support --format= it doesn't create
+      # ustar tarball either.
+      (tar --version) >/dev/null 2>&1 && continue
+      am__tar='tar chf - "$$tardir"'
+      am__tar_='tar chf - "$tardir"'
+      am__untar='tar xf -'
+      ;;
+    pax)
+      am__tar='pax -L -x $1 -w "$$tardir"'
+      am__tar_='pax -L -x $1 -w "$tardir"'
+      am__untar='pax -r'
+      ;;
+    cpio)
+      am__tar='find "$$tardir" -print | cpio -o -H $1 -L'
+      am__tar_='find "$tardir" -print | cpio -o -H $1 -L'
+      am__untar='cpio -i -H $1 -d'
+      ;;
+    none)
+      am__tar=false
+      am__tar_=false
+      am__untar=false
+      ;;
+    esac
+
+    # If the value was cached, stop now.  We just wanted to have am__tar
+    # and am__untar set.
+    test -n "${am_cv_prog_tar_$1}" && break
+
+    # tar/untar a dummy directory, and stop if the command works.
+    rm -rf conftest.dir
+    mkdir conftest.dir
+    echo GrepMe > conftest.dir/file
+    AM_RUN_LOG([tardir=conftest.dir && eval $am__tar_ >conftest.tar])
+    rm -rf conftest.dir
+    if test -s conftest.tar; then
+      AM_RUN_LOG([$am__untar <conftest.tar])
+      AM_RUN_LOG([cat conftest.dir/file])
+      grep GrepMe conftest.dir/file >/dev/null 2>&1 && break
+    fi
+  done
   rm -rf conftest.dir
-  mkdir conftest.dir
-  echo GrepMe > conftest.dir/file
-  AM_RUN_LOG([tardir=conftest.dir && eval $am__tar_ >conftest.tar])
-  rm -rf conftest.dir
-  if test -s conftest.tar; then
-    AM_RUN_LOG([$am__untar <conftest.tar])
-    grep GrepMe conftest.dir/file >/dev/null 2>&1 && break
-  fi
-done
-rm -rf conftest.dir
 
-AC_CACHE_VAL([am_cv_prog_tar_$1], [am_cv_prog_tar_$1=$_am_tool])
-AC_MSG_RESULT([$am_cv_prog_tar_$1])])
+  AC_CACHE_VAL([am_cv_prog_tar_$1], [am_cv_prog_tar_$1=$_am_tool])
+  AC_MSG_RESULT([$am_cv_prog_tar_$1])])
+
 AC_SUBST([am__tar])
 AC_SUBST([am__untar])
 ]) # _AM_PROG_TAR
diff --git a/configure b/configure
index e048369e..cb6bab86 100755
--- a/configure
+++ b/configure
@@ -626,6 +626,12 @@ ac_subst_vars='am__EXEEXT_FALSE
 am__EXEEXT_TRUE
 LTLIBOBJS
 LIBOBJS
+BUILD_COMMS_NONE_FALSE
+BUILD_COMMS_NONE_TRUE
+BUILD_COMMS_FAKE_FALSE
+BUILD_COMMS_FAKE_TRUE
+BUILD_COMMS_MPI_FALSE
+BUILD_COMMS_MPI_TRUE
 EGREP
 GREP
 CPP
@@ -5068,6 +5074,31 @@ $as_echo "#define GRID_COMMS_MPI 1" >>confdefs.h
      ;;
 esac
 
+ if  test "X${ac_COMMS}X" == "XmpiX" ; then
+  BUILD_COMMS_MPI_TRUE=
+  BUILD_COMMS_MPI_FALSE='#'
+else
+  BUILD_COMMS_MPI_TRUE='#'
+  BUILD_COMMS_MPI_FALSE=
+fi
+
+ if  test "X${ac_COMMS}X" == "XfakeX" ; then
+  BUILD_COMMS_FAKE_TRUE=
+  BUILD_COMMS_FAKE_FALSE='#'
+else
+  BUILD_COMMS_FAKE_TRUE='#'
+  BUILD_COMMS_FAKE_FALSE=
+fi
+
+ if  test "X${ac_COMMS}X" == "XnoneX" ; then
+  BUILD_COMMS_NONE_TRUE=
+  BUILD_COMMS_NONE_FALSE='#'
+else
+  BUILD_COMMS_NONE_TRUE='#'
+  BUILD_COMMS_NONE_FALSE=
+fi
+
+
 
 ac_config_files="$ac_config_files Makefile"
 
@@ -5208,6 +5239,18 @@ if test -z "${am__fastdepCC_TRUE}" && test -z "${am__fastdepCC_FALSE}"; then
   as_fn_error $? "conditional \"am__fastdepCC\" was never defined.
 Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
+if test -z "${BUILD_COMMS_MPI_TRUE}" && test -z "${BUILD_COMMS_MPI_FALSE}"; then
+  as_fn_error $? "conditional \"BUILD_COMMS_MPI\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
+fi
+if test -z "${BUILD_COMMS_FAKE_TRUE}" && test -z "${BUILD_COMMS_FAKE_FALSE}"; then
+  as_fn_error $? "conditional \"BUILD_COMMS_FAKE\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
+fi
+if test -z "${BUILD_COMMS_NONE_TRUE}" && test -z "${BUILD_COMMS_NONE_FALSE}"; then
+  as_fn_error $? "conditional \"BUILD_COMMS_NONE\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
+fi
 
 : "${CONFIG_STATUS=./config.status}"
 ac_write_fail=0
diff --git a/configure.ac b/configure.ac
index 147ce21f..22d0365e 100644
--- a/configure.ac
+++ b/configure.ac
@@ -71,6 +71,10 @@ case ${ac_COMMS} in
      ;;
 esac
 
+AM_CONDITIONAL(BUILD_COMMS_MPI,[ test "X${ac_COMMS}X" == "XmpiX" ])
+AM_CONDITIONAL(BUILD_COMMS_FAKE,[ test "X${ac_COMMS}X" == "XfakeX" ])
+AM_CONDITIONAL(BUILD_COMMS_NONE,[ test "X${ac_COMMS}X" == "XnoneX" ])
+
 
 AC_CONFIG_FILES(Makefile)
 AC_OUTPUT

From 06843d4574ea15414f13604883122331fadc98ee Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 3 Apr 2015 04:58:03 +0100
Subject: [PATCH 029/429] Rename some files to make naming consistent

---
 Grid_cshift.h                            | 16 ++++++++++++++++
 Grid_fake_cshift.h => Grid_cshift_fake.h |  0
 Grid_mpi_cshift.h => Grid_cshift_mpi.h   |  0
 Grid_none_cshift.h => Grid_cshift_none.h |  0
 4 files changed, 16 insertions(+)
 create mode 100644 Grid_cshift.h
 rename Grid_fake_cshift.h => Grid_cshift_fake.h (100%)
 rename Grid_mpi_cshift.h => Grid_cshift_mpi.h (100%)
 rename Grid_none_cshift.h => Grid_cshift_none.h (100%)

diff --git a/Grid_cshift.h b/Grid_cshift.h
new file mode 100644
index 00000000..2df2370e
--- /dev/null
+++ b/Grid_cshift.h
@@ -0,0 +1,16 @@
+#ifndef _GRID_CSHIFT_H_
+#define _GRID_CSHIFT_H_
+#include <Grid_cshift_common.h>
+
+#ifdef GRID_COMMS_NONE
+#include <Grid_cshift_none.h>
+#endif
+
+#ifdef GRID_COMMS_FAKE
+#include <Grid_cshift_fake.h>
+#endif
+
+#ifdef GRID_COMMS_MPI
+#include <Grid_cshift_mpi.h>
+#endif 
+#endif
diff --git a/Grid_fake_cshift.h b/Grid_cshift_fake.h
similarity index 100%
rename from Grid_fake_cshift.h
rename to Grid_cshift_fake.h
diff --git a/Grid_mpi_cshift.h b/Grid_cshift_mpi.h
similarity index 100%
rename from Grid_mpi_cshift.h
rename to Grid_cshift_mpi.h
diff --git a/Grid_none_cshift.h b/Grid_cshift_none.h
similarity index 100%
rename from Grid_none_cshift.h
rename to Grid_cshift_none.h

From 0c5c974e6d5a24e68424ca8121bea337e400b8f7 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 3 Apr 2015 05:29:54 +0100
Subject: [PATCH 030/429] Renamed the namespace to Grid

---
 Grid.h                   |  2 +-
 Grid_Cartesian.h         |  8 ++++----
 Grid_Communicator.h      |  2 +-
 Grid_Lattice.h           | 19 +++----------------
 Grid_QCD.h               |  4 ++--
 Grid_aligned_allocator.h |  4 ++--
 Grid_fake.cc             |  2 +-
 Grid_init.cc             |  2 +-
 Grid_main.cc             | 12 ++++++------
 Grid_math_types.h        |  2 +-
 Grid_mpi.cc              |  2 +-
 Grid_simd.h              |  4 ++--
 Grid_vComplexD.h         |  2 +-
 Grid_vComplexF.h         |  2 +-
 Grid_vRealD.h            |  2 +-
 Grid_vRealF.h            |  2 +-
 16 files changed, 29 insertions(+), 42 deletions(-)

diff --git a/Grid.h b/Grid.h
index cc8b36c6..184e7e36 100644
--- a/Grid.h
+++ b/Grid.h
@@ -50,7 +50,7 @@
 #include <Grid_Lattice.h>
 #include <Grid_QCD.h>
 
-namespace dpo {
+namespace Grid {
 
   void Grid_init(int *argc,char ***argv);
   double usecond(void);
diff --git a/Grid_Cartesian.h b/Grid_Cartesian.h
index 908d08a8..ced06825 100644
--- a/Grid_Cartesian.h
+++ b/Grid_Cartesian.h
@@ -3,15 +3,15 @@
 
 #include <Grid.h>
 #include <Grid_Communicator.h>
-namespace dpo{
+namespace Grid{
     
 /////////////////////////////////////////////////////////////////////////////////////////
 // Grid Support. Following will go into Grid.h.
 /////////////////////////////////////////////////////////////////////////////////////////
     // Cartesian grids
-    // dpo::Grid
-    // dpo::GridCartesian
-    // dpo::GridCartesianRedBlack
+    // Grid::Grid
+    // Grid::GridCartesian
+    // Grid::GridCartesianRedBlack
 
 
 class SimdGrid : public CartesianCommunicator {
diff --git a/Grid_Communicator.h b/Grid_Communicator.h
index 19e25167..4f83da34 100644
--- a/Grid_Communicator.h
+++ b/Grid_Communicator.h
@@ -6,7 +6,7 @@
 #ifdef GRID_COMMS_MPI
 #include <mpi.h>
 #endif
-namespace dpo {
+namespace Grid {
 class CartesianCommunicator {
   public:    
 
diff --git a/Grid_Lattice.h b/Grid_Lattice.h
index fcff36eb..6219235a 100644
--- a/Grid_Lattice.h
+++ b/Grid_Lattice.h
@@ -5,7 +5,7 @@
 
 
 
-namespace dpo {
+namespace Grid {
 
 // Permute the pointers 32bitx16 = 512
 static int permute_map[4][16] = { 
@@ -37,22 +37,9 @@ public:
     }
     
 
-#include <Grid_cshift_common.h>
-
-#ifdef GRID_COMMS_NONE
-#include <Grid_none_cshift.h>
-#endif
-
-#ifdef GRID_COMMS_FAKE
-#include <Grid_fake_cshift.h>
-#endif
-
-#ifdef GRID_COMMS_MPI
-#include <Grid_mpi_cshift.h>
-#endif 
-
+#include <Grid_cshift.h>
     
-    // overloading dpo::conformable but no conformable in dpo ...?:w
+    // overloading Grid::conformable but no conformable in Grid ...?:w
     template<class obj1,class obj2>
     friend void conformable(const Lattice<obj1> &lhs,const Lattice<obj2> &rhs);
 
diff --git a/Grid_QCD.h b/Grid_QCD.h
index e51e2641..f6cc6087 100644
--- a/Grid_QCD.h
+++ b/Grid_QCD.h
@@ -1,6 +1,6 @@
 #ifndef GRID_QCD_H
 #define GRID_QCD_H
-namespace dpo{
+namespace Grid{
 namespace QCD {
 
     static const int Nc=3;
@@ -89,5 +89,5 @@ namespace QCD {
         return ret;
      }
 }   //namespace QCD
-} // dpo
+} // Grid
 #endif
diff --git a/Grid_aligned_allocator.h b/Grid_aligned_allocator.h
index 6040c233..dbaa0ba3 100644
--- a/Grid_aligned_allocator.h
+++ b/Grid_aligned_allocator.h
@@ -1,6 +1,6 @@
 #ifndef GRID_ALIGNED_ALLOCATOR_H
 #define GRID_ALIGNED_ALLOCATOR_H
-namespace dpo {
+namespace Grid {
 
 ////////////////////////////////////////////////////////////////////
 // A lattice of something, but assume the something is SIMDized.
@@ -52,5 +52,5 @@ template<typename _Tp>  inline bool
 operator!=(const alignedAllocator<_Tp>&, const alignedAllocator<_Tp>&){ return false; }
 
     
-}; // namespace dpo
+}; // namespace Grid
 #endif
diff --git a/Grid_fake.cc b/Grid_fake.cc
index abc54c8e..83d5475e 100644
--- a/Grid_fake.cc
+++ b/Grid_fake.cc
@@ -1,5 +1,5 @@
 #include "Grid.h"
-namespace dpo {
+namespace Grid {
 
 CartesianCommunicator::CartesianCommunicator(std::vector<int> &processors)
 {
diff --git a/Grid_init.cc b/Grid_init.cc
index 4f7d1849..3a601209 100755
--- a/Grid_init.cc
+++ b/Grid_init.cc
@@ -14,7 +14,7 @@
 #include "Grid.h"
 
 #undef __X86_64
-namespace dpo {
+namespace Grid {
 
 void Grid_init(int *argc,char ***argv)
 {
diff --git a/Grid_main.cc b/Grid_main.cc
index 795ea395..504f7778 100644
--- a/Grid_main.cc
+++ b/Grid_main.cc
@@ -7,8 +7,8 @@
 #include "Grid_Lattice.h"
 
 using namespace std;
-using namespace dpo;
-using namespace dpo::QCD;
+using namespace Grid;
+using namespace Grid::QCD;
 
 
 int main (int argc, char ** argv)
@@ -175,10 +175,10 @@ int main (int argc, char ** argv)
     double t0,t1,flops;
     double bytes;
     int ncall=100;
-    int Nc = dpo::QCD::Nc;
+    int Nc = Grid::QCD::Nc;
 
     flops = ncall*1.0*volume*(8*Nc*Nc*Nc);
-    bytes = ncall*1.0*volume*Nc*Nc    *2*3*sizeof(dpo::Real);
+    bytes = ncall*1.0*volume*Nc*Nc    *2*3*sizeof(Grid::Real);
     printf("%f flop and %f bytes\n",flops,bytes/ncall);
         FooBar = Foo * Bar;
     t0=usecond();
@@ -195,7 +195,7 @@ int main (int argc, char ** argv)
     mult(FooBar,Foo,Bar);
     FooBar = Foo * Bar;
 
-    bytes = ncall*1.0*volume*Nc*Nc    *2*5*sizeof(dpo::Real);
+    bytes = ncall*1.0*volume*Nc*Nc    *2*5*sizeof(Grid::Real);
     t0=usecond();
     for(int i=0;i<ncall;i++){
       mult(FooBar,Foo,Cshift(Bar,1,-2));
@@ -244,7 +244,7 @@ int main (int argc, char ** argv)
 	for(coor[1]=0;coor[1]<latt_size[1]/mpi_layout[1];coor[1]++){
 	for(coor[0]=0;coor[0]<latt_size[0]/mpi_layout[0];coor[0]++){
  
-        std::complex<dpo::Real> diff;
+        std::complex<Grid::Real> diff;
                     
         std::vector<int> shiftcoor = coor;
         shiftcoor[dir]=(shiftcoor[dir]+shift+latt_size[dir])%(latt_size[dir]/mpi_layout[dir]);
diff --git a/Grid_math_types.h b/Grid_math_types.h
index e5387f12..533ce424 100644
--- a/Grid_math_types.h
+++ b/Grid_math_types.h
@@ -1,6 +1,6 @@
 #ifndef GRID_MATH_TYPES_H
 #define GRID_MATH_TYPES_H
-namespace dpo {
+namespace Grid {
 
 
 
diff --git a/Grid_mpi.cc b/Grid_mpi.cc
index 6cda0f1b..e00e3dce 100644
--- a/Grid_mpi.cc
+++ b/Grid_mpi.cc
@@ -1,7 +1,7 @@
 #include "Grid.h"
 #include <mpi.h>
 
-namespace dpo {
+namespace Grid {
 
   // Should error check all MPI calls.
 
diff --git a/Grid_simd.h b/Grid_simd.h
index bef21bcf..b5d34d4a 100644
--- a/Grid_simd.h
+++ b/Grid_simd.h
@@ -38,7 +38,7 @@
     //
     
 
-namespace dpo {
+namespace Grid {
 
   typedef  float  RealF;
   typedef  double RealD;
@@ -57,7 +57,7 @@ namespace dpo {
   inline RealF localInnerProduct(const RealF & l, const RealF & r) { return l*r; }
 
     ////////////////////////////////////////////////////////////////////////////////
-    //Provide support functions for basic real and complex data types required by dpo
+    //Provide support functions for basic real and complex data types required by Grid
     //Single and double precision versions. Should be able to template this once only.
     ////////////////////////////////////////////////////////////////////////////////
     inline void mac (ComplexD * __restrict__ y,const ComplexD * __restrict__ a,const ComplexD *__restrict__ x){ *y = (*a) * (*x)+(*y); };
diff --git a/Grid_vComplexD.h b/Grid_vComplexD.h
index 15dc00c6..f575f3ad 100644
--- a/Grid_vComplexD.h
+++ b/Grid_vComplexD.h
@@ -3,7 +3,7 @@
 #include "Grid.h"
 #include "Grid_vComplexF.h"
 
-namespace dpo {
+namespace Grid {
     class vComplexD {
     protected:
         zvec v;
diff --git a/Grid_vComplexF.h b/Grid_vComplexF.h
index 0b4f5e61..6c3e9eac 100644
--- a/Grid_vComplexF.h
+++ b/Grid_vComplexF.h
@@ -2,7 +2,7 @@
 #define VCOMPLEXF
 #include "Grid.h"
 
-namespace dpo {
+namespace Grid {
     class vComplexF {
     protected:
         cvec v;
diff --git a/Grid_vRealD.h b/Grid_vRealD.h
index e5d3d4c6..34150e4b 100644
--- a/Grid_vRealD.h
+++ b/Grid_vRealD.h
@@ -3,7 +3,7 @@
 
 #include "Grid.h"
 
-namespace dpo{
+namespace Grid {
     class vRealD  {
     protected:
         dvec v; // dvec is double precision vector
diff --git a/Grid_vRealF.h b/Grid_vRealF.h
index db74b005..94875f7e 100644
--- a/Grid_vRealF.h
+++ b/Grid_vRealF.h
@@ -3,7 +3,7 @@
 
 #include "Grid.h"
 
-namespace dpo {
+namespace Grid {
     class vRealF  {
     protected:
         fvec v;

From 154a8700f4206c8de7a5f032d6f4f7660f72c9a3 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 3 Apr 2015 05:30:58 +0100
Subject: [PATCH 031/429] Removing the Xcode project

---
 Grid/Grid.xcodeproj/project.pbxproj           | 239 ------------------
 .../contents.xcworkspacedata                  |   7 -
 .../UserInterfaceState.xcuserstate            | Bin 4124 -> 0 bytes
 .../pab.xcuserdatad/xcschemes/Grid.xcscheme   |  43 ----
 .../xcschemes/xcschememanagement.plist        |  22 --
 5 files changed, 311 deletions(-)
 delete mode 100644 Grid/Grid.xcodeproj/project.pbxproj
 delete mode 100644 Grid/Grid.xcodeproj/project.xcworkspace/contents.xcworkspacedata
 delete mode 100644 Grid/Grid.xcodeproj/project.xcworkspace/xcuserdata/pab.xcuserdatad/UserInterfaceState.xcuserstate
 delete mode 100644 Grid/Grid.xcodeproj/xcuserdata/pab.xcuserdatad/xcschemes/Grid.xcscheme
 delete mode 100644 Grid/Grid.xcodeproj/xcuserdata/pab.xcuserdatad/xcschemes/xcschememanagement.plist

diff --git a/Grid/Grid.xcodeproj/project.pbxproj b/Grid/Grid.xcodeproj/project.pbxproj
deleted file mode 100644
index 20cd1d1a..00000000
--- a/Grid/Grid.xcodeproj/project.pbxproj
+++ /dev/null
@@ -1,239 +0,0 @@
-// !$*UTF8*$!
-{
-	archiveVersion = 1;
-	classes = {
-	};
-	objectVersion = 46;
-	objects = {
-
-/* Begin PBXBuildFile section */
-		0BF5DA461AC370840045EA34 /* main.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 0BF5DA451AC370840045EA34 /* main.cpp */; };
-/* End PBXBuildFile section */
-
-/* Begin PBXCopyFilesBuildPhase section */
-		0BF5DA401AC370840045EA34 /* CopyFiles */ = {
-			isa = PBXCopyFilesBuildPhase;
-			buildActionMask = 2147483647;
-			dstPath = /usr/share/man/man1/;
-			dstSubfolderSpec = 0;
-			files = (
-			);
-			runOnlyForDeploymentPostprocessing = 1;
-		};
-/* End PBXCopyFilesBuildPhase section */
-
-/* Begin PBXFileReference section */
-		0BF5DA421AC370840045EA34 /* Grid */ = {isa = PBXFileReference; explicitFileType = "compiled.mach-o.executable"; includeInIndex = 0; path = Grid; sourceTree = BUILT_PRODUCTS_DIR; };
-		0BF5DA451AC370840045EA34 /* main.cpp */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.cpp.cpp; path = main.cpp; sourceTree = "<group>"; };
-/* End PBXFileReference section */
-
-/* Begin PBXFrameworksBuildPhase section */
-		0BF5DA3F1AC370840045EA34 /* Frameworks */ = {
-			isa = PBXFrameworksBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-/* End PBXFrameworksBuildPhase section */
-
-/* Begin PBXGroup section */
-		0BF5DA391AC370840045EA34 = {
-			isa = PBXGroup;
-			children = (
-				0BF5DA441AC370840045EA34 /* Grid */,
-				0BF5DA431AC370840045EA34 /* Products */,
-			);
-			sourceTree = "<group>";
-		};
-		0BF5DA431AC370840045EA34 /* Products */ = {
-			isa = PBXGroup;
-			children = (
-				0BF5DA421AC370840045EA34 /* Grid */,
-			);
-			name = Products;
-			sourceTree = "<group>";
-		};
-		0BF5DA441AC370840045EA34 /* Grid */ = {
-			isa = PBXGroup;
-			children = (
-				0BF5DA451AC370840045EA34 /* main.cpp */,
-			);
-			path = Grid;
-			sourceTree = "<group>";
-		};
-/* End PBXGroup section */
-
-/* Begin PBXNativeTarget section */
-		0BF5DA411AC370840045EA34 /* Grid */ = {
-			isa = PBXNativeTarget;
-			buildConfigurationList = 0BF5DA491AC370840045EA34 /* Build configuration list for PBXNativeTarget "Grid" */;
-			buildPhases = (
-				0BF5DA3E1AC370840045EA34 /* Sources */,
-				0BF5DA3F1AC370840045EA34 /* Frameworks */,
-				0BF5DA401AC370840045EA34 /* CopyFiles */,
-			);
-			buildRules = (
-			);
-			dependencies = (
-			);
-			name = Grid;
-			productName = Grid;
-			productReference = 0BF5DA421AC370840045EA34 /* Grid */;
-			productType = "com.apple.product-type.tool";
-		};
-/* End PBXNativeTarget section */
-
-/* Begin PBXProject section */
-		0BF5DA3A1AC370840045EA34 /* Project object */ = {
-			isa = PBXProject;
-			attributes = {
-				LastUpgradeCheck = 0610;
-				ORGANIZATIONNAME = "University of Edinburgh";
-				TargetAttributes = {
-					0BF5DA411AC370840045EA34 = {
-						CreatedOnToolsVersion = 6.1.1;
-					};
-				};
-			};
-			buildConfigurationList = 0BF5DA3D1AC370840045EA34 /* Build configuration list for PBXProject "Grid" */;
-			compatibilityVersion = "Xcode 3.2";
-			developmentRegion = English;
-			hasScannedForEncodings = 0;
-			knownRegions = (
-				en,
-			);
-			mainGroup = 0BF5DA391AC370840045EA34;
-			productRefGroup = 0BF5DA431AC370840045EA34 /* Products */;
-			projectDirPath = "";
-			projectRoot = "";
-			targets = (
-				0BF5DA411AC370840045EA34 /* Grid */,
-			);
-		};
-/* End PBXProject section */
-
-/* Begin PBXSourcesBuildPhase section */
-		0BF5DA3E1AC370840045EA34 /* Sources */ = {
-			isa = PBXSourcesBuildPhase;
-			buildActionMask = 2147483647;
-			files = (
-				0BF5DA461AC370840045EA34 /* main.cpp in Sources */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-/* End PBXSourcesBuildPhase section */
-
-/* Begin XCBuildConfiguration section */
-		0BF5DA471AC370840045EA34 /* Debug */ = {
-			isa = XCBuildConfiguration;
-			buildSettings = {
-				ALWAYS_SEARCH_USER_PATHS = NO;
-				CLANG_CXX_LANGUAGE_STANDARD = "gnu++0x";
-				CLANG_CXX_LIBRARY = "libc++";
-				CLANG_ENABLE_MODULES = YES;
-				CLANG_ENABLE_OBJC_ARC = YES;
-				CLANG_WARN_BOOL_CONVERSION = YES;
-				CLANG_WARN_CONSTANT_CONVERSION = YES;
-				CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
-				CLANG_WARN_EMPTY_BODY = YES;
-				CLANG_WARN_ENUM_CONVERSION = YES;
-				CLANG_WARN_INT_CONVERSION = YES;
-				CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
-				CLANG_WARN_UNREACHABLE_CODE = YES;
-				CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
-				COPY_PHASE_STRIP = NO;
-				ENABLE_STRICT_OBJC_MSGSEND = YES;
-				GCC_C_LANGUAGE_STANDARD = gnu99;
-				GCC_DYNAMIC_NO_PIC = NO;
-				GCC_OPTIMIZATION_LEVEL = 0;
-				GCC_PREPROCESSOR_DEFINITIONS = (
-					"DEBUG=1",
-					"$(inherited)",
-				);
-				GCC_SYMBOLS_PRIVATE_EXTERN = NO;
-				GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
-				GCC_WARN_ABOUT_RETURN_TYPE = YES_ERROR;
-				GCC_WARN_UNDECLARED_SELECTOR = YES;
-				GCC_WARN_UNINITIALIZED_AUTOS = YES_AGGRESSIVE;
-				GCC_WARN_UNUSED_FUNCTION = YES;
-				GCC_WARN_UNUSED_VARIABLE = YES;
-				MACOSX_DEPLOYMENT_TARGET = 10.10;
-				MTL_ENABLE_DEBUG_INFO = YES;
-				ONLY_ACTIVE_ARCH = YES;
-				SDKROOT = macosx;
-			};
-			name = Debug;
-		};
-		0BF5DA481AC370840045EA34 /* Release */ = {
-			isa = XCBuildConfiguration;
-			buildSettings = {
-				ALWAYS_SEARCH_USER_PATHS = NO;
-				CLANG_CXX_LANGUAGE_STANDARD = "gnu++0x";
-				CLANG_CXX_LIBRARY = "libc++";
-				CLANG_ENABLE_MODULES = YES;
-				CLANG_ENABLE_OBJC_ARC = YES;
-				CLANG_WARN_BOOL_CONVERSION = YES;
-				CLANG_WARN_CONSTANT_CONVERSION = YES;
-				CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
-				CLANG_WARN_EMPTY_BODY = YES;
-				CLANG_WARN_ENUM_CONVERSION = YES;
-				CLANG_WARN_INT_CONVERSION = YES;
-				CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
-				CLANG_WARN_UNREACHABLE_CODE = YES;
-				CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
-				COPY_PHASE_STRIP = YES;
-				DEBUG_INFORMATION_FORMAT = "dwarf-with-dsym";
-				ENABLE_NS_ASSERTIONS = NO;
-				ENABLE_STRICT_OBJC_MSGSEND = YES;
-				GCC_C_LANGUAGE_STANDARD = gnu99;
-				GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
-				GCC_WARN_ABOUT_RETURN_TYPE = YES_ERROR;
-				GCC_WARN_UNDECLARED_SELECTOR = YES;
-				GCC_WARN_UNINITIALIZED_AUTOS = YES_AGGRESSIVE;
-				GCC_WARN_UNUSED_FUNCTION = YES;
-				GCC_WARN_UNUSED_VARIABLE = YES;
-				MACOSX_DEPLOYMENT_TARGET = 10.10;
-				MTL_ENABLE_DEBUG_INFO = NO;
-				SDKROOT = macosx;
-			};
-			name = Release;
-		};
-		0BF5DA4A1AC370840045EA34 /* Debug */ = {
-			isa = XCBuildConfiguration;
-			buildSettings = {
-				PRODUCT_NAME = "$(TARGET_NAME)";
-			};
-			name = Debug;
-		};
-		0BF5DA4B1AC370840045EA34 /* Release */ = {
-			isa = XCBuildConfiguration;
-			buildSettings = {
-				PRODUCT_NAME = "$(TARGET_NAME)";
-			};
-			name = Release;
-		};
-/* End XCBuildConfiguration section */
-
-/* Begin XCConfigurationList section */
-		0BF5DA3D1AC370840045EA34 /* Build configuration list for PBXProject "Grid" */ = {
-			isa = XCConfigurationList;
-			buildConfigurations = (
-				0BF5DA471AC370840045EA34 /* Debug */,
-				0BF5DA481AC370840045EA34 /* Release */,
-			);
-			defaultConfigurationIsVisible = 0;
-			defaultConfigurationName = Release;
-		};
-		0BF5DA491AC370840045EA34 /* Build configuration list for PBXNativeTarget "Grid" */ = {
-			isa = XCConfigurationList;
-			buildConfigurations = (
-				0BF5DA4A1AC370840045EA34 /* Debug */,
-				0BF5DA4B1AC370840045EA34 /* Release */,
-			);
-			defaultConfigurationIsVisible = 0;
-		};
-/* End XCConfigurationList section */
-	};
-	rootObject = 0BF5DA3A1AC370840045EA34 /* Project object */;
-}
diff --git a/Grid/Grid.xcodeproj/project.xcworkspace/contents.xcworkspacedata b/Grid/Grid.xcodeproj/project.xcworkspace/contents.xcworkspacedata
deleted file mode 100644
index ce76f4f5..00000000
--- a/Grid/Grid.xcodeproj/project.xcworkspace/contents.xcworkspacedata
+++ /dev/null
@@ -1,7 +0,0 @@
-<?xml version="1.0" encoding="UTF-8"?>
-<Workspace
-   version = "1.0">
-   <FileRef
-      location = "self:Grid.xcodeproj">
-   </FileRef>
-</Workspace>
diff --git a/Grid/Grid.xcodeproj/project.xcworkspace/xcuserdata/pab.xcuserdatad/UserInterfaceState.xcuserstate b/Grid/Grid.xcodeproj/project.xcworkspace/xcuserdata/pab.xcuserdatad/UserInterfaceState.xcuserstate
deleted file mode 100644
index 638ce019b327e05f2249ae344a55de843e067d13..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 4124
zcma)933O9s7QQzxdH;KVB2CLy*0!`35n2O;E}*69rUhD{X-g^4@R~l_2Wb*t5=vR*
zN6<McIOBr599nQdXB<XHMa6Y=Mjgi^FqL5#bvPck!>A~Wh@#B@@|r@_Imh;#yyd_D
zUB3I>d;ixO4rr06%hd-677)M+HV~2FhNhgAsvglo!KR#0>vGi}jV#Pjbbp7&U)SeE
zL*X>Q8@KOqt^fiO42B^v6vo44Faa)yi7*NBARn%PsW2U8z)UEID!3XJ!D3hfE#QNt
zpg=45p$$4<Ijn_sa2>1%1{>gd*a$bl&G2iu8}5O7;Xb$@9)ySBxA1#-6#f8Dz_ajY
zcm>{sze69q1Bc)k{1-lh&*2L=4qw7oZ~}gS^9YD&!9kdUPE5r#9EziGG>*Y+ybN79
z8H;f`mS6=|<9w_`FD}BxxCB)UVH8*4+J>B9ED-ny$Uwmk4oHEt1-0J%6>4{+nP2-M
zY0U8llt`o(E`dSJ!h~LMLMpSeA<X4-4ymrFToBS%M8b++tqA#JooX=ZbLJGf3Oton
zg;jaeii(T!++~&SJkQhucV303sNCc6lzR%x%M*r&!C1)M1;Zg7M!-lI1*2gMvoVno
zCNY^&X5R%FkO^5Z4zeK!a+rf%!cv%%rLn>Me-L-0Ojnf^;gA-LMm+wgwo-|zZ9eD7
z3U!$h3q&j1v}j28M)})FRY-5Bn-lRlhgYbrvG#T?*uFpuwuQRN)DC5(7Sesrltdur
z0+&s%Q~f+JQ%_>GUyUamuHu15s)IFJP#aLnol1Hx4+SHkfGQ`VcEM!G-M*a-ItXr<
z1_e;qYyurrI@KogR^_G#*LrJWQKdDYR%kr3P*BmkcX06{C|+Faz3{H7);l-OVJ94n
z=krP^>ERg@SgL8E6g({hx;(nBa5be+2E{z*@keu81(a-OsZzqtYhXU)?uKfZ1+!re
z)IcrFg?Vf!8^(sSbT)#GWTV*V-8?~FsOQNU#Fv}mDH_AZ^5h77Inzwi2!ktesOv+a
zK&zrxM;f(=#v^MPSgF!^^gd@6e=%2YQ+2g%fL--UE5|UZhXMg!v_9trb=AQt&f|z*
zSJhxmsEzNRbTKVWn-}&r1hs2pYIPe=sJ2W~dBQWec$_xs&s+03OD<NJ<cV)`Rarq%
zd2vBrrMs{)&po-QAg`?0U7lC&DyS$gce`C4*Hq3RLxCB*(pq*=?3{>e))crWWfm8%
zo0Pd`vfJgFl*x-`-8wTLDzx{&T#jOfnZ`2EdX20l60`y$M)d@s6M_(eFkA~d%VOhL
zHp^j`vRpQP7q6TQSP5OQieFdr%5kvE*i=@)uQQG68Rm&Zc!N|D1dq&7x<fHu#!l|E
zw<FY5%L}7j;U(mA=KWtF8lqZ2i*|c>C3<<ojW%j(m#Jc4yRr78<RoEPp+>a!;Lm6<
zoG|rkYGes_!K_3>GJ(S;xDj%D;089K2Y$saH{wdIXsjP_r8=s1ModU=fs$U>3|rVl
zHkqX+6#NG6fV5t?4Q^+XSY9vO30qk{yMnova5WdR#mS$I;A)iipcaj_sq?Z+H>QK}
zXQrqK9_TgdGog7K^!!Nk4z76$b7hx?GUNLalDptxBZ%G1-2;!XY4ISuYJg8BwQT@J
zpEI+`&uN~|ahR<|^9L64y-_{pkH&O0p>Gd7HUP@wtk7sSMyr1kN_N0guos?YMQl2|
z@&X9^;DsbtL+}!(QL*7_xHQStSK+lJNgdu`<Jk;D@)njP`4;R?k~{zh*;TB>7#{|~
z2J&iT^CxEh2}h0oa~S>wN8sP^F1!ct!+%&Q^RO~j&MH_Xt72F0=A9>l(>x14f{)=7
z_>^}X2Pb+8yM`@hOPFdP=`>hm_GeW$j2T2S-kD_L4ZG1i4fKzTSZgOox!HU&`^sRx
zqkpEv6&YH1I{IJDGqy*fysy-$VO@>z8^!3@yji6t><;j;IS?<hdfv1W**Xc|K<-XB
z1*hR_IK!&hEH<0X*$HRiTlfynu^Q%O4Xlw3YMP@ii`Un}I<37Ue%EZ)tm)=ZBd|mb
z4mjFx1_hM38MLAeMI@}2&1Li0{GA-aEVM%ghc1QHarii7qZ$=mGiDF()M}T3RisYo
zYE%L-V>FFSUhU_p*B=V=7Mz*17~dY!qrAfXx)$cGgzq1gBofzN-yK%t9L2%VzMW+x
zd>)4BoX9ww)%QRa_s^*A{)t+p1){2M1n$Iy_*l$L@E)_+f_~m(4ljJX6mwY<YaZY{
zPQXb?{5*yE9Q%a^lh<s&n8`Q=rzJ_A#zLN>MRCb_Ns=@0N^`m=JerA~0guX9OTwc{
z!=oy^n)z7k1&?N7&Bb=G7U#01Oku8tJh3JpvrDz5NfzsILDHTkT*&;a&8V`JIBSi=
zj>-G%(#0>u_x526`fw@OQNdR9BR?{w>7h{6<dQ+n+W0sf-mV6<Sj4n!3{rJf)Lcmq
z>slv2k^E$dwKHJL*z!ZT4BNSZ4%BctH=<xCKccq6^d^4H4z%fNuraLg-{>O0p*I*c
z7{J-n)_<1t#E;_!1(R!PQApPk^~Ei<GmSyAtuVLN3DXT*tb>bi0h3uVK2g!noiisY
zcH?TFb3DiB=ZAkX7`?a#ihsFyd?4UDyw2y${sk6#alN7amx%JT)_P~F-D;Z&k*XiS
z>v3a4{FKWOBz~A4$G=jh@voDan2uv{B2MLB1C=-n=Ws@Ou>lw0LJZ^GxD)?`PvTyD
z2KV9f_#(cH`|%+5;UPSXNAO*IACKZ0%MeSxrPQ*>60>Z!Y_Z&Dx!>}DWt(M}<#EeC
z%K^(_%SV<kEMHpATFzO15I~THbRk0+FSvxs!c@U6c!WBkUT74Wgl1u}&>}1qT7`%Z
z6S{<MVU4g(STAf4HVQWgJ;EP_KH*DihSg)e-uj?*uk|JCE7reQU$gGFzGMBs`jz#h
z^|bYj^{n+f>-X04wm~+hEzLH>Hq4f88)+MD8*3}H&9bewZLvLPd)xM&?U?Oz+i}|o
z+bP@EA{7UV>EcLnv^Z9rD7wTcqFXE$XNXscGsPNlsi=s{#fZ2{yg|HK+$`QI-XU%k
z?-uVBd&NEC<Kh$IUhx@mpZK<TLOf3_#7YQ}iJc4~qsSPNL9$3Tagk}HkQ9>=QcB85
z1*sv;#77k3Cmm!t36LP^BAduf<QB4p+)nN!caeL@@5mG6Wpa@8kwfG#IYQng?~|kC
z19Fm_CTGZ5@*Vk}oR?4%B%72j<w{pc*GMhW25GCbU3ye{N_tv)R(ei)L3&MkLwZws
zOL|*6ARUoDl#a_p9wLvCGvq8eTfS7zmtFD{*)12yMRKDYk?)fC$WP19$ou5i<u~QG
z<o)tNxljI7J|>@$zm?C)=P6QwI%p~#Oo!6pG>1;06KNiuLfy207EuqarC!=V7tqDD
zg)XJ7bOl{SyXk7Wmfl1+(_86n^bWd}ZlgQsLv$B?m_9}K(dX%l^i}#-`a1m^?W6C}
zWAr%vik_t3&~NEE`h(qRA7Q`TKGVL?9<{Huud=VUueEQrZ?ivSf6U%zKWhKL{-OP2
z`=<_c2o9TrIAlkfW2j@eV}xUzV~%6KV~t~z;}ORm$2*Qs9G^K(#QPjrEanwAnb*Xx
H<JA8EDt*1p

diff --git a/Grid/Grid.xcodeproj/xcuserdata/pab.xcuserdatad/xcschemes/Grid.xcscheme b/Grid/Grid.xcodeproj/xcuserdata/pab.xcuserdatad/xcschemes/Grid.xcscheme
deleted file mode 100644
index b59f83df..00000000
--- a/Grid/Grid.xcodeproj/xcuserdata/pab.xcuserdatad/xcschemes/Grid.xcscheme
+++ /dev/null
@@ -1,43 +0,0 @@
-<?xml version="1.0" encoding="UTF-8"?>
-<Scheme
-   LastUpgradeVersion = "0610"
-   version = "1.3">
-   <BuildAction
-      parallelizeBuildables = "YES"
-      buildImplicitDependencies = "YES">
-   </BuildAction>
-   <TestAction
-      selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
-      selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
-      shouldUseLaunchSchemeArgsEnv = "YES"
-      buildConfiguration = "Debug">
-      <Testables>
-      </Testables>
-   </TestAction>
-   <LaunchAction
-      selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
-      selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
-      launchStyle = "0"
-      useCustomWorkingDirectory = "NO"
-      buildConfiguration = "Debug"
-      ignoresPersistentStateOnLaunch = "NO"
-      debugDocumentVersioning = "YES"
-      allowLocationSimulation = "YES">
-      <AdditionalOptions>
-      </AdditionalOptions>
-   </LaunchAction>
-   <ProfileAction
-      shouldUseLaunchSchemeArgsEnv = "YES"
-      savedToolIdentifier = ""
-      useCustomWorkingDirectory = "NO"
-      buildConfiguration = "Release"
-      debugDocumentVersioning = "YES">
-   </ProfileAction>
-   <AnalyzeAction
-      buildConfiguration = "Debug">
-   </AnalyzeAction>
-   <ArchiveAction
-      buildConfiguration = "Release"
-      revealArchiveInOrganizer = "YES">
-   </ArchiveAction>
-</Scheme>
diff --git a/Grid/Grid.xcodeproj/xcuserdata/pab.xcuserdatad/xcschemes/xcschememanagement.plist b/Grid/Grid.xcodeproj/xcuserdata/pab.xcuserdatad/xcschemes/xcschememanagement.plist
deleted file mode 100644
index 1d4ba5a6..00000000
--- a/Grid/Grid.xcodeproj/xcuserdata/pab.xcuserdatad/xcschemes/xcschememanagement.plist
+++ /dev/null
@@ -1,22 +0,0 @@
-<?xml version="1.0" encoding="UTF-8"?>
-<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
-<plist version="1.0">
-<dict>
-	<key>SchemeUserState</key>
-	<dict>
-		<key>Grid.xcscheme</key>
-		<dict>
-			<key>orderHint</key>
-			<integer>0</integer>
-		</dict>
-	</dict>
-	<key>SuppressBuildableAutocreation</key>
-	<dict>
-		<key>0BF5DA411AC370840045EA34</key>
-		<dict>
-			<key>primary</key>
-			<true/>
-		</dict>
-	</dict>
-</dict>
-</plist>

From d29733069fe7c4a886f931baa553f92ffa336f5c Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 3 Apr 2015 05:33:13 +0100
Subject: [PATCH 032/429] Patch

---
 Grid_cshift_none.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/Grid_cshift_none.h b/Grid_cshift_none.h
index f16b403b..37d20fa3 100644
--- a/Grid_cshift_none.h
+++ b/Grid_cshift_none.h
@@ -8,4 +8,4 @@ friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
   Cshift_local(ret,rhs,dimension,shift);
 }
         
-
+#endif

From 18b26a7a7003824f86da0e42932659b4c73c69dd Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 3 Apr 2015 05:34:51 +0100
Subject: [PATCH 033/429] MPI added

---
 Grid_config.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/Grid_config.h b/Grid_config.h
index 599afe2e..8a0fdc26 100644
--- a/Grid_config.h
+++ b/Grid_config.h
@@ -17,7 +17,7 @@
 #define GRID_COMMS_MPI 1
 
 /* GRID_COMMS_NONE */
-#define GRID_COMMS_NONE 1
+/* #undef GRID_COMMS_NONE */
 
 /* Define to 1 if you have the `gettimeofday' function. */
 #define HAVE_GETTIMEOFDAY 1

From 0c1f8b70e9caa42f8b13c8527dbc034dd2568a78 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 3 Apr 2015 05:51:05 +0100
Subject: [PATCH 034/429] Reorg

---
 Makefile.am | 15 +++++++++++++--
 Makefile.in | 14 ++++++++++++--
 2 files changed, 25 insertions(+), 4 deletions(-)

diff --git a/Makefile.am b/Makefile.am
index b681e0c5..d3a7b329 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -10,14 +10,25 @@ libGrid_a_SOURCES = Grid_init.cc
 #
 # Include files
 #
-include_HEADERS = Grid.h\
+include_HEADERS = Grid_config.h\
+	Grid.h\
+	Grid_simd.h\
 	Grid_vComplexD.h\
 	Grid_vComplexF.h\
 	Grid_vRealD.h\
 	Grid_vRealF.h\
 	Grid_Cartesian.h\
 	Grid_Lattice.h\
-	Grid_config.h
+	Grid_Communicator.h\
+	Grid_QCD.h\
+	Grid_aligned_allocator.h\
+	Grid_cshift.h\
+	Grid_cshift_common.h\
+	Grid_cshift_fake.h\
+	Grid_cshift_mpi.h\
+	Grid_cshift_none.h\
+	Grid_math_types.h
+
 
 #
 # Test code
diff --git a/Makefile.in b/Makefile.in
index 43ea6616..6983b96a 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -339,14 +339,24 @@ libGrid_a_SOURCES = Grid_init.cc
 #
 # Include files
 #
-include_HEADERS = Grid.h\
+include_HEADERS = Grid_config.h\
+	Grid.h\
+	Grid_simd.h\
 	Grid_vComplexD.h\
 	Grid_vComplexF.h\
 	Grid_vRealD.h\
 	Grid_vRealF.h\
 	Grid_Cartesian.h\
 	Grid_Lattice.h\
-	Grid_config.h
+	Grid_Communicator.h\
+	Grid_QCD.h\
+	Grid_aligned_allocator.h\
+	Grid_cshift.h\
+	Grid_cshift_common.h\
+	Grid_cshift_fake.h\
+	Grid_cshift_mpi.h\
+	Grid_cshift_none.h\
+	Grid_math_types.h
 
 extra_sources = $(am__append_1) $(am__append_2) $(am__append_3)
 Grid_main_SOURCES = \

From 15dda435e61614a7627f1174968312ef8535c1ec Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 3 Apr 2015 09:28:58 +0100
Subject: [PATCH 035/429] TODO list for preparing this for real use and QDP++
 replacement.

---
 TODO | 28 ++++++++++++++++++++++++++++
 1 file changed, 28 insertions(+)
 create mode 100644 TODO

diff --git a/TODO b/TODO
new file mode 100644
index 00000000..8108fbc6
--- /dev/null
+++ b/TODO
@@ -0,0 +1,28 @@
+* Make peekSite, pokeSite work in an MPI context.
+
+* Conditional execution Subset, where etc...
+* Coordinate information, integers etc...
+* Integer type padding/union to vector.
+
+* Optimise the extract/merge SIMD routines
+
+* Broadcast, reduction tests.
+
+* QDP++ regression suite and comparative benchmark
+
+* NERSC Lattice loading, plaquette test
+
+* Conformable test in Cshift routines.
+
+* Gamma/Dirac structures
+* Fourspin, two spin project
+
+* Stencil operator support
+
+* Check for missing functionality
+
+* I/O support
+  - MPI IO?
+  - BinaryWriter, TextWriter etc...
+  - protocol buffers?
+  - 

From ad31cd0c23ce5128b2745bd524f48d59a0efe3af Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 3 Apr 2015 22:54:13 +0100
Subject: [PATCH 036/429] Clean up but no major changes

---
 Grid_Cartesian.h     |  13 ++----
 Grid_Lattice.h       | 108 +++++++++++++++----------------------------
 Grid_cshift_common.h |  27 ++++++-----
 Grid_cshift_mpi.h    |  82 ++------------------------------
 Grid_math_types.h    |   2 +-
 Grid_mpi.cc          |  22 +++------
 Grid_vComplexD.h     |   8 ++--
 Grid_vComplexF.h     |   7 ++-
 Grid_vRealD.h        |   5 +-
 Grid_vRealF.h        |   5 +-
 10 files changed, 74 insertions(+), 205 deletions(-)

diff --git a/Grid_Cartesian.h b/Grid_Cartesian.h
index ced06825..a983639c 100644
--- a/Grid_Cartesian.h
+++ b/Grid_Cartesian.h
@@ -201,10 +201,8 @@ public:
             block = block*_rdimensions[d];
         }
             
-        if ( _isites != vComplex::Nsimd()) {
-            printf("bad layout for grid isites %d Nsimd %d\n",_isites,vComplex::Nsimd());
-            exit(0);
-        }
+        assert( _isites == vComplex::Nsimd());
+
     };
 };
  
@@ -235,10 +233,8 @@ public:
       int ocb=CheckerBoardFromOsite(osite);
 	  
       if ( (source_cb+ocb)&1 ) {
-	printf("Checkerboard shift %d\n",(shift)/2);
 	return (shift)/2;
       } else {
-	printf("Checkerboard shift %d\n",(shift+1)/2);
 	return (shift+1)/2;
       }
     }
@@ -314,10 +310,7 @@ public:
             block = block*_rdimensions[d];
         }
             
-        if ( _isites != vComplex::Nsimd()) {
-            printf("bad layout for grid isites %d Nsimd %d\n",_isites,vComplex::Nsimd());
-            exit(0);
-        }
+        assert ( _isites == vComplex::Nsimd());
     };
 protected:
     virtual int oIndex(std::vector<int> &coor)
diff --git a/Grid_Lattice.h b/Grid_Lattice.h
index 6219235a..aab4fec9 100644
--- a/Grid_Lattice.h
+++ b/Grid_Lattice.h
@@ -30,9 +30,7 @@ public:
 
     Lattice(SimdGrid *grid) : _grid(grid) {
         _odata.reserve(_grid->oSites());
-        if ( ((uint64_t)&_odata[0])&0xF) {
-            exit(-1);
-        }
+        assert((((uint64_t)&_odata[0])&0xF) ==0);
         checkerboard=0;
     }
     
@@ -97,26 +95,25 @@ public:
     template<class sobj>
     friend void pokeSite(const sobj &s,Lattice<vobj> &l,std::vector<int> &site){
 
-        if ( l.checkerboard != l._grid->CheckerBoard(site)){
-	  printf("Poking wrong checkerboard\n");
-	  exit(EXIT_FAILURE);
-        }
+      typedef typename vobj::scalar_type stype;
+      typedef typename vobj::vector_type vtype;
 
-        int o_index = l._grid->oIndex(site);
-        int i_index = l._grid->iIndex(site);
+      assert( l.checkerboard == l._grid->CheckerBoard(site));
+
+      int o_index = l._grid->oIndex(site);
+      int i_index = l._grid->iIndex(site);
+      
+      stype *v_ptr = (stype *)&l._odata[o_index];
+      stype *s_ptr = (stype *)&s;
+      v_ptr = v_ptr + 2*i_index;
         
-	// BUGGY. This assumes complex real
-        Real *v_ptr = (Real *)&l._odata[o_index];
-        Real *s_ptr = (Real *)&s;
-        v_ptr = v_ptr + 2*i_index;
-        
-        for(int i=0;i<sizeof(sobj);i+=2*sizeof(Real)){
-            v_ptr[0] = s_ptr[0];
-            v_ptr[1] = s_ptr[1];
-            v_ptr+=2*vComplex::Nsimd();
-            s_ptr+=2;
-        }
-        return;
+      for(int i=0;i<sizeof(sobj);i+=2*sizeof(stype)){
+	v_ptr[0] = s_ptr[0];
+	v_ptr[1] = s_ptr[1];
+	v_ptr+=2*vtype::Nsimd();
+	s_ptr+=2;
+      }
+      return;
     };
     
     
@@ -124,25 +121,25 @@ public:
     template<class sobj>
     friend void peekSite(sobj &s,const Lattice<vobj> &l,std::vector<int> &site){
         
-        // FIXME : define exceptions set and throw up.
-        if ( l.checkerboard != l._grid->CheckerBoard(site)){
-            printf("Peeking wrong checkerboard\n");
-            exit(EXIT_FAILURE);
-        }
-        int o_index = l._grid->oIndex(site);
-        int i_index = l._grid->iIndex(site);
-        
-        Real *v_ptr = (Real *)&l._odata[o_index];
-        Real *s_ptr = (Real *)&s;
-        v_ptr = v_ptr + 2*i_index;
-        
-        for(int i=0;i<sizeof(sobj);i+=2*sizeof(Real)){
-            s_ptr[0] = v_ptr[0];
-            s_ptr[1] = v_ptr[1];
-            v_ptr+=2*vComplex::Nsimd();
-            s_ptr+=2;
-        }
-        return;
+      typedef typename vobj::scalar_type stype;
+      typedef typename vobj::vector_type vtype;
+
+      assert( l.checkerboard== l._grid->CheckerBoard(site));
+
+      int o_index = l._grid->oIndex(site);
+      int i_index = l._grid->iIndex(site);
+      
+      stype *v_ptr = (stype *)&l._odata[o_index];
+      stype *s_ptr = (stype *)&s;
+      v_ptr = v_ptr + 2*i_index;
+      
+      for(int i=0;i<sizeof(sobj);i+=2*sizeof(stype)){
+	s_ptr[0] = v_ptr[0];
+	s_ptr[1] = v_ptr[1];
+	v_ptr+=2*vtype::Nsimd();
+	s_ptr+=2;
+      }
+      return;
     };
     
     // Randomise
@@ -291,15 +288,6 @@ public:
     }
 }; // class Lattice
 
-    /* Need to implement the multiplication return type matching S S -> S, S M -> M, M S -> M through
-     all nested possibilities.
-     template<template<class> class lhs,template<class> class rhs>
-     class MultTypeSelector {
-     template<typename vtype> using ltype = lhs
-     typedef lhs type;
-     };
-     */
-    
     template<class obj1,class obj2>
     void conformable(const Lattice<obj1> &lhs,const Lattice<obj2> &rhs)
     {
@@ -313,28 +301,6 @@ public:
 	uint32_t vec_len = lhs._grid->oSites();
 #pragma omp parallel for
         for(int ss=0;ss<vec_len;ss++){
-
-	  const char * ptr =(const char*)&lhs._odata[ss];
-#ifdef PREFETCH
-          v_prefetch0(sizeof(obj2), ptr);
-#endif
-
-	  for(int i=0;i<sizeof(obj2);i+=64){
-	    _mm_prefetch(ptr+i+4096,_MM_HINT_T1);
-	    _mm_prefetch(ptr+i+256,_MM_HINT_T0);
-	  }
-
-	  ptr =(const char*)&rhs._odata[ss];
-#ifdef PREFETCH
-          v_prefetch0(sizeof(obj3), ptr);
-#endif
-
-	  for(int i=0;i<sizeof(obj3);i+=64){
-	    _mm_prefetch(ptr+i+4096,_MM_HINT_T1);
-	    _mm_prefetch(ptr+i+256,_MM_HINT_T0);
-	  }
-
-
 	  mult(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
         }
     }
diff --git a/Grid_cshift_common.h b/Grid_cshift_common.h
index b8624826..203bf2f5 100644
--- a/Grid_cshift_common.h
+++ b/Grid_cshift_common.h
@@ -1,5 +1,17 @@
 #ifndef _GRID_CSHIFT_COMMON_H_
 #define _GRID_CSHIFT_COMMON_H_
+//////////////////////////////////////////////////////////////////////////////////////////
+// Must not lose sight that goal is to be able to construct really efficient
+// gather to a point stencil code. CSHIFT is not the best way, so probably need
+// additional stencil support.
+//
+// Stencil based code could pre-exchange haloes and use a table lookup for neighbours
+//
+// Lattice <foo> could also allocate haloes which get used for stencil code.
+//
+// Grid could create a neighbour index table for a given stencil.
+// Could also implement CovariantCshift.
+//////////////////////////////////////////////////////////////////////////////////////////
 
 //////////////////////////////////////////////////////
 // Gather for when there is no need to SIMD split
@@ -8,8 +20,6 @@ friend void Gather_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllo
 {
   int rd = rhs._grid->_rdimensions[dimension];
 
-  //  printf("Gather plane _simple mask %d\n",cbmask);
-
   if ( !rhs._grid->CheckerBoarded(dimension) ) {
 
     int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
@@ -31,7 +41,6 @@ friend void Gather_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllo
     int o   = 0;                                      // relative offset to base within plane
     int bo  = 0;                                      // offset in buffer
 
-    //    int jjj=0;
 #pragma omp parallel for collapse(2)
     for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
       for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
@@ -39,11 +48,6 @@ friend void Gather_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllo
 	int ocb=1<<rhs._grid->CheckerBoardFromOsite(o+b);// Could easily be a table lookup
 	if ( ocb &cbmask ) {
 	  buffer[bo]=rhs._odata[so+o+b];
-	  //	  float * ptr = (float *)& rhs._odata[so+o+b];
-	  //	  if( (cbmask!=3)&&(jjj<8)){
-	  //	    printf("Gather_plane_simple %d %le bo %d\n",so+o+b,*ptr,bo);
-	  //	    jjj++;
-	  //	  }
 	  bo++;
 	}
 
@@ -215,7 +219,7 @@ friend void Copy_plane(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int dimension,int
     int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
     int lo  = lplane*lhs._grid->_ostride[dimension]; // base offset for start of plane 
     int o   = 0;                                     // relative offset to base within plane
-    //    int jjj=0;    
+
 #pragma omp parallel for collapse(2)
     for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
       for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
@@ -224,11 +228,6 @@ friend void Copy_plane(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int dimension,int
 
 	if ( ocb&cbmask ) {
 	  lhs._odata[lo+o+b]=rhs._odata[ro+o+b];
-	  //	  float *ptr =(float *) &rhs._odata[ro+o+b];
-	  //	  if((cbmask!=0x3)&&jjj<8) {
-	    //	    printf("Copy_plane %d %le n,b=%d,%d mask %d ocb %d\n",ro+o+b,*ptr,n,b,cbmask,ocb);
-	    //	    jjj++;
-	    //	  }
 	}
 
       }
diff --git a/Grid_cshift_mpi.h b/Grid_cshift_mpi.h
index 4029f152..17aff2f4 100644
--- a/Grid_cshift_mpi.h
+++ b/Grid_cshift_mpi.h
@@ -1,36 +1,10 @@
 #ifndef _GRID_MPI_CSHIFT_H_
 #define _GRID_MPI_CSHIFT_H_
 
+#ifndef MAX
 #define MAX(x,y) ((x)>(y)?(x):(y))
 #define MIN(x,y) ((x)>(y)?(y):(x))
-//////////////////////////////////////////////////////////////////////////////////////////
-// Must not lose sight that goal is to be able to construct really efficient
-// gather to a point stencil code. CSHIFT is not the best way, so probably need
-// additional stencil support.
-//
-// Could still do a templated syntax tree and make CSHIFT return lattice vector.
-//
-// Stencil based code could pre-exchange haloes and use a table lookup for neighbours
-//
-// Lattice <foo> could also allocate haloes which get used for stencil code.
-//
-// Grid could create a neighbour index table for a given stencil.
-// Could also implement CovariantCshift.
-//////////////////////////////////////////////////////////////////////////////////////////
-
-      
-/////////////////////////////////////////////////////////////
-// Q. Further split this into separate sub functions?
-/////////////////////////////////////////////////////////////
-
-// CshiftCB_local
-// CshiftCB_local_permute
-
-// Cshift_comms_splice
-// Cshift_comms
-// Cshift_local
-// Cshift_local_permute
-
+#endif
 
 friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
 {
@@ -71,16 +45,10 @@ friend void Cshift_comms(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimension,int
   sshift[1] = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,1);
 
   if ( sshift[0] == sshift[1] ) {
-    //    printf("Cshift_comms : single pass\n");
     Cshift_comms(ret,rhs,dimension,shift,0x3);
   } else {
-    //    printf("Cshift_comms : two pass\n");
-    //    printf("call1\n");
     Cshift_comms(ret,rhs,dimension,shift,0x1);// if checkerboard is unfavourable take two passes
-    //    printf("call2\n");
     Cshift_comms(ret,rhs,dimension,shift,0x2);// both with block stride loop iteration
-    //    printf("done\n");
-
   }
 }
 
@@ -94,11 +62,8 @@ friend void Cshift_comms_simd(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimensio
   if ( sshift[0] == sshift[1] ) {
     Cshift_comms_simd(ret,rhs,dimension,shift,0x3);
   } else {
-    //    printf("call1 0x1 cb=even\n");
     Cshift_comms_simd(ret,rhs,dimension,shift,0x1);// if checkerboard is unfavourable take two passes
-    //    printf("call2 0x2 cb=odd\n");
     Cshift_comms_simd(ret,rhs,dimension,shift,0x2);// both with block stride loop iteration
-    //    printf("done\n");
   }
 }
 
@@ -120,13 +85,10 @@ friend void Cshift_comms(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int
   assert(shift>=0);
   assert(shift<fd);
   
-  // Packed gather sequence is clean
   int buffer_size = rhs._grid->_slice_nblock[dimension]*rhs._grid->_slice_block[dimension];
   std::vector<vobj,alignedAllocator<vobj> > send_buf(buffer_size);
   std::vector<vobj,alignedAllocator<vobj> > recv_buf(buffer_size);
 
-  // This code could be simplified by multiple calls to single routine with extra params to
-  // encapsulate the difference in the code paths.
   int cb= (cbmask==0x2)? 1 : 0;
   int sshift= rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,cb);
 
@@ -137,8 +99,9 @@ friend void Cshift_comms(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int
     int comm_proc = (x+sshift)/rd;
     
     if (!offnode) {
-      //      printf("local x %d sshift %d offnode %d rd %d cb %d\n",x,sshift,offnode,rd,cb);
+
       Copy_plane(ret,rhs,dimension,x,sx,cbmask); 
+
     } else {
 
       int words = send_buf.size();
@@ -146,29 +109,13 @@ friend void Cshift_comms(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int
 
       int bytes = words * sizeof(vobj);
 
-      //      printf("nonlocal x %d sx %d sshift %d offnode %d rd %d cb %d cbmask %d rhscb %d comm_proc %d\n",
-      //	     x,sx,sshift,offnode,rd,cb,cbmask,rhs.checkerboard,comm_proc);
-      //      Copy_plane(temp,rhs,dimension,x,sx,cbmask); 
-
-      // Bug found; cbmask may differ between sx plan and rx plane.
       Gather_plane_simple (rhs,send_buf,dimension,sx,cbmask);
-      //      for(int i=0;i<MIN(words,8);i++){
-      //	float *ptr = (float *)&send_buf[i];
-      //	printf("send buf shift %d cbmask %d i %d %le\n",sshift,cbmask,i,*ptr);
-      //      }
-      //      Gather_plane_simple (rhs,send_buf,dimension,sx,cbmask^0x3);
-      //      for(int i=0;i<MIN(words,8);i++){
-      //	float *ptr = (float *)&send_buf[i];
-      //	printf("send buf shift %d cbmask %d i %d %le\n",sshift,cbmask,i,*ptr);
-      //      }
-      //      recv_buf=send_buf;
 
       int rank           = grid->_processor;
       int recv_from_rank;
       int xmit_to_rank;
       grid->ShiftedRanks(dimension,comm_proc,xmit_to_rank,recv_from_rank);
 
-      //      printf("bytes %d node %d sending to %d receiving from %d\n",bytes,rank,xmit_to_rank,recv_from_rank );
       grid->SendToRecvFrom((void *)&send_buf[0],
 			   xmit_to_rank,
 			   (void *)&recv_buf[0],
@@ -224,46 +171,29 @@ friend void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimensi
   int cb    = (cbmask==0x2)? 1 : 0;
   int sshift= grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,cb);
   
-  //  printf("cshift-comms-simd: shift = %d ; sshift = %d ; cbmask %d ; simd_layout %d\n",shift,sshift,cbmask,simd_layout);
   std::vector<int> comm_offnode(simd_layout);
   std::vector<int> comm_proc   (simd_layout);  //relative processor coord in dim=dimension
 
-  // Strategy
-  //
-  //*  Loop over source planes
-  //*    if any communication needed extract and send
-  //*    if communication needed extract and send
-
   for(int x=0;x<rd;x++){       
 
     int comm_any = 0;
     for(int s=0;s<simd_layout;s++) {
-      // does shift to "neighbour" takes us off node?
-      // coordinates (reduce plane, simd_lane) of neighbour?
-      // how many nodes away is this shift?
-      // where we should send to?
-      // where we should receive from?
       int shifted_x   = x+s*rd+sshift;
       comm_offnode[s] = shifted_x >= ld; 
       comm_any        = comm_any | comm_offnode[s];
       comm_proc[s]    = shifted_x/ld;     
-      //      printf("rd %d x %d shifted %d s=%d comm_any %d\n",rd, x,shifted_x,s,comm_any);
     }
     
     int o    = 0;
     int bo   = x*grid->_ostride[dimension];
     int sx   = (x+sshift)%rd;
 
-    // Need Convenience function in _grid. Move this in
     if ( comm_any ) {
 
       for(int i=0;i<Nsimd;i++){
 	pointers[i] = (scalar_type *)&send_buf_extract[i][0];
       }
       Gather_plane_extract(rhs,pointers,dimension,sx,cbmask);
-      //      for(int i=0;i<Nsimd;i++){
-      //	printf("extracted %d %le\n",i,real(send_buf_extract[i][0]));
-      //      }
 
       for(int i=0;i<Nsimd;i++){
 
@@ -283,14 +213,11 @@ friend void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimensi
 			    recv_from_rank,
 			    bytes);
 
-	  //	  printf("Cshift_simd comms %d %le %le\n",i,real(recv_buf_extract[i][0]),real(send_buf_extract[i][0]));
-
 	  rpointers[i] = (scalar_type *)&recv_buf_extract[i][0];
 
 	} else { 
 
 	  rpointers[i] = (scalar_type *)&send_buf_extract[i][0];
-	  //	  printf("Cshift_simd local %d %le \n",i,real(send_buf_extract[i][0]));
 
 	}
 
@@ -311,7 +238,6 @@ friend void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimensi
 	} else {
 	  pointers[i] = rpointers[i];
 	}
-	//	printf("Cshift_simd perm %d num %d wrap %d swiz %d %le unswiz %le\n",permute_slice,num,wrap,i,real(pointers[i][0]),real(rpointers[i][0]));
       }
 
       Scatter_plane_merge(ret,pointers,dimension,x,cbmask);
diff --git a/Grid_math_types.h b/Grid_math_types.h
index 533ce424..3f7714d5 100644
--- a/Grid_math_types.h
+++ b/Grid_math_types.h
@@ -830,7 +830,7 @@ template<class vtype,int N> inline iMatrix<vtype,N> adj(const iMatrix<vtype,N> &
 
 /////////////////////////////////////////////////////////////////
 // Can only take the real/imag part of scalar objects, since
-// lattice objects of different complexity are non-conformable.
+// lattice objects of different complex nature are non-conformable.
 /////////////////////////////////////////////////////////////////
 template<class itype> inline auto real(const iScalar<itype> &z) -> iScalar<decltype(real(z._internal))>
 {
diff --git a/Grid_mpi.cc b/Grid_mpi.cc
index e00e3dce..c9b91210 100644
--- a/Grid_mpi.cc
+++ b/Grid_mpi.cc
@@ -17,13 +17,15 @@ CartesianCommunicator::CartesianCommunicator(std::vector<int> &processors)
   MPI_Cart_create(MPI_COMM_WORLD, _ndimension,&_processors[0],&periodic[0],1,&communicator);
   MPI_Comm_rank(communicator,&_processor);
   MPI_Cart_coords(communicator,_processor,_ndimension,&_processor_coor[0]);
-  printf("Hello world from processor [");
+
   for(int i=0;i<_ndimension;i++){
-    printf("%d ",_processor_coor[i]);
     _Nprocessors*=_processors[i];
   }
-  printf("]\n");
-  fflush(stdout);
+  
+  int Size; 
+  MPI_Comm_size(communicator,&Size);
+  
+  assert(Size==_Nprocessors);
 }
 
 void CartesianCommunicator::GlobalSumF(float &f){
@@ -71,15 +73,3 @@ void CartesianCommunicator::SendToRecvFrom(void *xmit,
 
 }
 
-#if 0
-
-// Could possibly do a direct block strided send?
-  int MPI_Type_vector(
-		      int count,
-		      int blocklength,
-		      int stride,
-		      MPI_Datatype old_type,
-  MPI_Datatype *newtype_p
-		      );
-
-#endif
diff --git a/Grid_vComplexD.h b/Grid_vComplexD.h
index f575f3ad..d1caefb1 100644
--- a/Grid_vComplexD.h
+++ b/Grid_vComplexD.h
@@ -48,7 +48,6 @@ namespace Grid {
 #endif
 #ifdef AVX512
             ret.v = _mm512_add_pd(a.v,b.v);
-            //printf("%s %f\n",__func__,_mm512_reduce_mul_pd(ret.v));
 #endif
 #ifdef QPX
             ret.v = vec_add(a.v,b.v);
@@ -210,7 +209,7 @@ namespace Grid {
 #ifdef QPX
 #error // Not implemented yet
 #endif
-                default: exit(EXIT_FAILURE); break;
+ 	        default: assert(0); break;
             }
         };
         void vload(zvec& a){
@@ -265,8 +264,7 @@ friend inline void vstore(vComplexD &ret, ComplexD *a){
    //Note v has a3 a2 a1 a0
 #endif
 #ifdef QPX
-	printf("%s Not implemented\n",__func__);
-	exit(-1);
+	assert(0);
 #endif
         }
       friend inline void vprefetch(const vComplexD &v)
@@ -294,7 +292,7 @@ friend inline void vstore(vComplexD &ret, ComplexD *a){
              
 #endif
 #ifdef QPX
-            exit(0); // not implemented
+	     assert(0);
 #endif
             return ret;
         }
diff --git a/Grid_vComplexF.h b/Grid_vComplexF.h
index 6c3e9eac..578228a3 100644
--- a/Grid_vComplexF.h
+++ b/Grid_vComplexF.h
@@ -193,7 +193,7 @@ namespace Grid {
 #ifdef QPX
 #error
 #endif
-                default: exit(EXIT_FAILURE); break;
+	        default: assert(0); break;
             }
         };
         
@@ -235,8 +235,7 @@ friend inline void vstore(vComplexF &ret, ComplexF *a){
 //Note v has a3 a2 a1 a0
 #endif
 #ifdef QPX
-	printf("%s Not implemented\n",__func__);
-exit(-1);
+	assert(0);
 #endif
 }
       friend inline void vprefetch(const vComplexF &v)
@@ -333,7 +332,7 @@ exit(-1);
             ret.v = _mm512_mask_sub_ps(in.v,0xaaaa,ret.v,in.v); // Zero out 0+real 0-imag 
 #endif
 #ifdef QPX
-            exit(0); // not implemented
+            assert(0);
 #endif
             return ret;
         }
diff --git a/Grid_vRealD.h b/Grid_vRealD.h
index 34150e4b..1abc0804 100644
--- a/Grid_vRealD.h
+++ b/Grid_vRealD.h
@@ -161,7 +161,7 @@ namespace Grid {
 #ifdef QPX
 #error
 #endif
-                default: exit(EXIT_FAILURE); break;
+	    default: assert(0);break;
             }
         };
 // gona be bye bye
@@ -214,8 +214,7 @@ namespace Grid {
             // Note v has a7 a6 a5ba4 a3 a2 a1 a0
 #endif
 #ifdef QPX
-          printf("%s Not implemented\n",__func__); 
-          exit(-1);
+	    assert(0);
 #endif
 	}
         friend inline void vprefetch(const vRealD &v)
diff --git a/Grid_vRealF.h b/Grid_vRealF.h
index 94875f7e..22809b83 100644
--- a/Grid_vRealF.h
+++ b/Grid_vRealF.h
@@ -185,7 +185,7 @@ namespace Grid {
 #ifdef QPX
 #error not implemented
 #endif
-                default: exit(EXIT_FAILURE); break;
+	    default: assert(0); break;
             }
         };
         
@@ -236,8 +236,7 @@ friend inline void vstore(vRealF &ret, float *a){
 	// Note v has a7 a6 a5ba4 a3 a2 a1 a0
 #endif
 #ifdef QPX
-	printf("%s Not implemented\n",__func__);
-	exit(-1);
+	assert(0);
 #endif
         }
 

From 02262b00195f36ad5120eb67e1786f7c42fb2fd8 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 6 Apr 2015 06:30:48 +0100
Subject: [PATCH 037/429] Bringing in LatticeInteger with the idea of
 implemented predicated assignment, subsets etc. c.f the QDP++ "where" syntax

---
 Grid_Cartesian.h     | 283 ++++++++++++++++++++++++++-------------
 Grid_Communicator.h  |  54 +++++++-
 Grid_Lattice.h       |  93 +++++++------
 Grid_QCD.h           |  14 +-
 Grid_cshift_common.h |  12 +-
 Grid_cshift_fake.h   |   8 +-
 Grid_cshift_mpi.h    |  14 +-
 Grid_main.cc         |  51 ++++---
 Grid_math_types.h    |  22 +---
 Grid_mpi.cc          |  29 +++-
 Grid_simd.h          | 119 +++++++++++++----
 Grid_vInteger.h      | 307 +++++++++++++++++++++++++++++++++++++++++++
 TODO                 |   3 +-
 13 files changed, 788 insertions(+), 221 deletions(-)
 create mode 100644 Grid_vInteger.h

diff --git a/Grid_Cartesian.h b/Grid_Cartesian.h
index a983639c..004c2fa5 100644
--- a/Grid_Cartesian.h
+++ b/Grid_Cartesian.h
@@ -6,126 +6,236 @@
 namespace Grid{
     
 /////////////////////////////////////////////////////////////////////////////////////////
-// Grid Support. Following will go into Grid.h.
+// Grid Support.
 /////////////////////////////////////////////////////////////////////////////////////////
-    // Cartesian grids
-    // Grid::Grid
-    // Grid::GridCartesian
-    // Grid::GridCartesianRedBlack
+//
+// Cartesian grid inheritance
+//            Grid::GridBase
+//                     |
+//           __________|___________
+//          |                      |
+// Grid::GridCartesian   Grid::GridCartesianRedBlack
+//
+// TODO: document the following as an API guaranteed public interface
 
+    /* 
+     *       Rough map of functionality against QDP++ Layout
+     *
+     *       Param     |     Grid                     |     QDP++             
+     *       -----------------------------------------
+     *                 |                              |
+     *        void     |     oSites, iSites, lSites   |  sitesOnNode 
+     *        void     |     gSites                   |  vol
+     *                 |                              |
+     *        gcoor    |     oIndex, iIndex           |  linearSiteIndex // no virtual node in QDP
+     *        lcoor    |                              |
+     * 
+     *        void     |     CheckerBoarded           |  -        // No checkerboarded in QDP
+     *        void     |     FullDimensions           |  lattSize
+     *        void     |     GlobalDimensions         |  lattSize // No checkerboarded in QDP
+     *        void     |     LocalDimensions          |  subgridLattSize
+     *        void     |     VirtualLocalDimensions   |  subgridLattSize // no virtual node in QDP
+     *                 |                              |
+     *       int x 3   |     oiSiteRankToGlobal       |  siteCoords
+     *                 |     ProcessorCoorLocalCoorToGlobalCoor | 
+     *                 |                              |
+     *     vector<int> |     GlobalCoorToRankIndex   |  nodeNumber(coord)
+     *     vector<int> |     GlobalCoorToProcessorCoorLocalCoor|  nodeCoord(coord)
+     *                 |                              |
+     *     void        |     Processors               |  logicalSize    // returns cart array shape
+     *     void        |     ThisRank        |  nodeNumber();  // returns this node rank
+     *     void        |     ThisProcessorCoor        |    // returns this node coor
+     *     void        |     isBoss(void)             |  primaryNode();
+     *                 |                              |
+     *                 |     RankFromProcessorCoor    |  getLogicalCoorFrom(node)
+     *                 |     ProcessorCoorFromRank    |  getNodeNumberFrom(logical_coord)
+     */
 
-class SimdGrid : public CartesianCommunicator {
+class GridBase : public CartesianCommunicator {
 public:
 
+ // Give Lattice access
+ template<class object> friend class Lattice;
   
- SimdGrid(std::vector<int> & processor_grid) : CartesianCommunicator(processor_grid) {};
+ GridBase(std::vector<int> & processor_grid) : CartesianCommunicator(processor_grid) {};
 
-    // Give Lattice access
-    template<class object> friend class Lattice;
         
-//protected:
-        
-  // Lattice wide random support. not yet fully implemented. Need seed strategy
-  // and one generator per site.
-  //std::default_random_engine generator;
-  //    static std::mt19937  generator( 9 );
-        
-    // Grid information.
-
-    // Commicator provides
+ //protected:
+ // Lattice wide random support. not yet fully implemented. Need seed strategy
+ // and one generator per site.
+ // std::default_random_engine generator;
+ //    static std::mt19937  generator( 9 );
+ 
+    //////////////////////////////////////////////////////////////////////
+    // Commicator provides information on the processor grid
+    //////////////////////////////////////////////////////////////////////
     //    unsigned long _ndimension;
     //    std::vector<int> _processors; // processor grid
     //    int              _processor;  // linear processor rank
     //    std::vector<int> _processor_coor;  // linear processor rank
+    //////////////////////////////////////////////////////////////////////
 
+    // Physics Grid information.
     std::vector<int> _simd_layout;     // Which dimensions get relayed out over simd lanes.
-
-
     std::vector<int> _fdimensions;// Global dimensions of array prior to cb removal
     std::vector<int> _gdimensions;// Global dimensions of array after cb removal
     std::vector<int> _ldimensions;// local dimensions of array with processor images removed
     std::vector<int> _rdimensions;// Reduced local dimensions with simd lane images and processor images removed 
+    std::vector<int> _ostride;    // Outer stride for each dimension
+    std::vector<int> _istride;    // Inner stride i.e. within simd lane
+    int _osites;                  // _isites*_osites = product(dimensions).
+    int _isites;
+    std::vector<int> _slice_block;   // subslice information
+    std::vector<int> _slice_stride;
+    std::vector<int> _slice_nblock;
 
+    // Might need these at some point
     //    std::vector<int> _lstart;     // local start of array in gcoors. _processor_coor[d]*_ldimensions[d]
     //    std::vector<int> _lend;       // local end of array in gcoors    _processor_coor[d]*_ldimensions[d]+_ldimensions_[d]-1
 
- 
-    std::vector<int> _ostride;    // Outer stride for each dimension
-    std::vector<int> _istride;    // Inner stride i.e. within simd lane
-
-    int _osites;                  // _isites*_osites = product(dimensions).
-    int _isites;
-        
-    // subslice information
-    std::vector<int> _slice_block;
-    std::vector<int> _slice_stride;
-    std::vector<int> _slice_nblock;
 public:
-    
-    // These routines are key. Subdivide the linearised cartesian index into
-    //      "inner" index identifying which simd lane of object<vFcomplex> is associated with coord
-    //      "outer" index identifying which element of _odata in class "Lattice" is associated with coord.
-    // Compared to, say, Blitz++ we simply need to store BOTH an inner stride and an outer
-    // stride per dimension. The cost of evaluating the indexing information is doubled for an n-dimensional
-    // coordinate. Note, however, for data parallel operations the "inner" indexing cost is not paid and all
-    // lanes are operated upon simultaneously.
-    
-    inline int oIndexReduced(std::vector<int> &rcoor)
-    {
-        int idx=0;
-        for(int d=0;d<_ndimension;d++) idx+=_ostride[d]*rcoor[d];
-        return idx;
-    }
-    virtual int oIndex(std::vector<int> &coor)
-    {
-        int idx=0;
-        for(int d=0;d<_ndimension;d++) idx+=_ostride[d]*(coor[d]%_rdimensions[d]);
-        return idx;
-    }
-    inline int iIndex(std::vector<int> &rcoor)
-    {
-        int idx=0;
-        for(int d=0;d<_ndimension;d++) idx+=_istride[d]*(rcoor[d]/_rdimensions[d]);
-        return idx;
-    }
-    inline int iCoordFromIsite(int lane,int mu)
-    {
-      std::vector<int> coor(_ndimension);
-      for(int d=0;d<_ndimension;d++){
-	coor[d] = lane % _simd_layout[d];
-	lane    = lane / _simd_layout[d];
-      }
-      return coor[mu];
-    }
-    
-    inline int oSites(void) { return _osites; };
-    inline int iSites(void) { return _isites; };
 
-    inline int CheckerBoardFromOsite (int Osite){
+    ////////////////////////////////////////////////////////////////
+    // Checkerboarding interface is virtual and overridden by 
+    // GridCartesian / GridRedBlackCartesian
+    ////////////////////////////////////////////////////////////////
+    virtual int CheckerBoarded(int dim)=0;
+    virtual int CheckerBoard(std::vector<int> site)=0;
+    virtual int CheckerBoardDestination(int source_cb,int shift)=0;
+    virtual int CheckerBoardShift(int source_cb,int dim,int shift,int osite)=0;
+    inline int  CheckerBoardFromOindex (int Oindex){
       std::vector<int> ocoor;
-      CoordFromOsite(ocoor,Osite);
+      oCoorFromOindex(ocoor,Oindex); 
       int ss=0;
       for(int d=0;d<_ndimension;d++){
 	ss=ss+ocoor[d];
       }      
       return ss&0x1;
     }
-    inline void CoordFromOsite (std::vector<int>& coor,int Osite){
+
+    //////////////////////////////////////////////////////////////////////////////////////////////
+    // Local layout calculations
+    //////////////////////////////////////////////////////////////////////////////////////////////
+    // These routines are key. Subdivide the linearised cartesian index into
+    //      "inner" index identifying which simd lane of object<vFcomplex> is associated with coord
+    //      "outer" index identifying which element of _odata in class "Lattice" is associated with coord.
+    //
+    // Compared to, say, Blitz++ we simply need to store BOTH an inner stride and an outer
+    // stride per dimension. The cost of evaluating the indexing information is doubled for an n-dimensional
+    // coordinate. Note, however, for data parallel operations the "inner" indexing cost is not paid and all
+    // lanes are operated upon simultaneously.
+  
+    virtual int oIndex(std::vector<int> &coor)
+    {
+        int idx=0;
+	// Works with either global or local coordinates
+        for(int d=0;d<_ndimension;d++) idx+=_ostride[d]*(coor[d]%_rdimensions[d]);
+        return idx;
+    }
+    inline int oIndexReduced(std::vector<int> &ocoor)
+    {
+      int idx=0; 
+      // ocoor is already reduced so can eliminate the modulo operation
+      // for fast indexing and inline the routine
+      for(int d=0;d<_ndimension;d++) idx+=_ostride[d]*ocoor[d];
+      return idx;
+    }
+    inline void oCoorFromOindex (std::vector<int>& coor,int Oindex){
       coor.resize(_ndimension);
       for(int d=0;d<_ndimension;d++){
-	coor[d] = Osite % _rdimensions[d];
-	Osite   = Osite / _rdimensions[d];
+	coor[d] = Oindex % _rdimensions[d];
+	Oindex  = Oindex / _rdimensions[d];
       }
     }
 
-    virtual int CheckerBoarded(int dim)=0;
-    virtual int CheckerBoard(std::vector<int> site)=0;
-    virtual int CheckerBoardDestination(int source_cb,int shift)=0;
-    virtual int CheckerBoardShift(int source_cb,int dim,int shift,int osite)=0;
+    //////////////////////////////////////////////////////////
+    // SIMD lane addressing
+    //////////////////////////////////////////////////////////
+    inline int iIndex(std::vector<int> &lcoor)
+    {
+        int idx=0;
+        for(int d=0;d<_ndimension;d++) idx+=_istride[d]*(lcoor[d]/_rdimensions[d]);
+        return idx;
+    }
+    inline void iCoorFromIindex(std::vector<int> &coor,int lane)
+    {
+      coor.resize(_ndimension);
+      for(int d=0;d<_ndimension;d++){
+	coor[d] = lane % _simd_layout[d];
+	lane    = lane / _simd_layout[d];
+      }
+    }
+    
+    ////////////////////////////////////////////////////////////////
+    // Array sizing queries
+    ////////////////////////////////////////////////////////////////
+
+    inline int iSites(void) { return _isites; };
+    inline int Nsimd(void)  { return _isites; };// Synonymous with iSites
+    inline int oSites(void) { return _osites; };
+    inline int lSites(void) { return _isites*_osites; }; 
+    inline int gSites(void) { return _isites*_osites*_Nprocessors; }; 
+    inline int Nd    (void) { return _ndimension;};
+    inline const std::vector<int> &FullDimensions(void)         { return _fdimensions;};
+    inline const std::vector<int> &GlobalDimensions(void)       { return _gdimensions;};
+    inline const std::vector<int> &LocalDimensions(void)        { return _ldimensions;};
+    inline const std::vector<int> &VirtualLocalDimensions(void) { return _ldimensions;};
+
+    ////////////////////////////////////////////////////////////////
+    // Global addressing
+    ////////////////////////////////////////////////////////////////
+    void RankIndexToGlobalCoor(int rank, int o_idx, int i_idx , std::vector<int> &gcoor)
+    {
+      gcoor.resize(_ndimension);
+      std::vector<int> coor(_ndimension);
+
+      ProcessorCoorFromRank(rank,coor);
+      for(int mu=0;mu<_ndimension;mu++) gcoor[mu] = _ldimensions[mu]&coor[mu];
+
+      iCoorFromIindex(coor,i_idx);
+      for(int mu=0;mu<_ndimension;mu++) gcoor[mu] += _rdimensions[mu]&coor[mu];
+
+      oCoorFromOindex (coor,o_idx);
+      for(int mu=0;mu<_ndimension;mu++) gcoor[mu] += coor[mu];
+      
+    }
+    void RankIndexCbToFullGlobalCoor(int rank, int o_idx, int i_idx, int cb,std::vector<int> &fcoor)
+    {
+      RankIndexToGlobalCoor(rank,o_idx,i_idx ,fcoor);
+      if(CheckerBoarded(0)){
+	fcoor[0] = fcoor[0]*2+cb;
+      }
+    }
+    void ProcessorCoorLocalCoorToGlobalCoor(std::vector<int> &Pcoor,std::vector<int> &Lcoor,std::vector<int> &gcoor)
+    {
+      gcoor.resize(_ndimension);
+      for(int mu=0;mu<_ndimension;mu++) gcoor[mu] = Pcoor[mu]*_ldimensions[mu]+Lcoor[mu];
+    }
+    void GlobalCoorToProcessorCoorLocalCoor(std::vector<int> &pcoor,std::vector<int> &lcoor,const std::vector<int> &gcoor)
+    {
+      pcoor.resize(_ndimension);
+      lcoor.resize(_ndimension);
+      for(int mu=0;mu<_ndimension;mu++){
+	pcoor[mu] = gcoor[mu]/_ldimensions[mu];
+	lcoor[mu] = gcoor[mu]%_ldimensions[mu];
+      }
+    }
+    void GlobalCoorToRankIndex(int &rank, int &o_idx, int &i_idx ,const std::vector<int> &gcoor)
+    {
+      std::vector<int> pcoor;
+      std::vector<int> lcoor;
+      GlobalCoorToProcessorCoorLocalCoor(pcoor,lcoor,gcoor);
+      rank = RankFromProcessorCoor(pcoor);
+      i_idx= iIndex(lcoor);
+      o_idx= oIndex(lcoor);
+    }
+
 };
 
-class GridCartesian: public SimdGrid {
+class GridCartesian: public GridBase {
+
 public:
+
     virtual int CheckerBoarded(int dim){
       return 0;
     }
@@ -141,7 +251,7 @@ public:
     GridCartesian(std::vector<int> &dimensions,
 		  std::vector<int> &simd_layout,
 		  std::vector<int> &processor_grid
-		  ) : SimdGrid(processor_grid)
+		  ) : GridBase(processor_grid)
     {
         ///////////////////////
         // Grid information
@@ -200,14 +310,12 @@ public:
             _slice_nblock[d]=nblock;
             block = block*_rdimensions[d];
         }
-            
-        assert( _isites == vComplex::Nsimd());
 
     };
 };
  
 // Specialise this for red black grids storing half the data like a chess board.
-class GridRedBlackCartesian : public SimdGrid
+class GridRedBlackCartesian : public GridBase
 {
 public:
     virtual int CheckerBoarded(int dim){
@@ -230,7 +338,7 @@ public:
 
       // Probably faster with table lookup;
       // or by looping over x,y,z and multiply rather than computing checkerboard.
-      int ocb=CheckerBoardFromOsite(osite);
+      int ocb=CheckerBoardFromOindex(osite);
 	  
       if ( (source_cb+ocb)&1 ) {
 	return (shift)/2;
@@ -248,7 +356,7 @@ public:
     };
     GridRedBlackCartesian(std::vector<int> &dimensions,
 			  std::vector<int> &simd_layout,
-			  std::vector<int> &processor_grid) : SimdGrid(processor_grid)
+			  std::vector<int> &processor_grid) : GridBase(processor_grid)
     {
     ///////////////////////
     // Grid information
@@ -310,7 +418,6 @@ public:
             block = block*_rdimensions[d];
         }
             
-        assert ( _isites == vComplex::Nsimd());
     };
 protected:
     virtual int oIndex(std::vector<int> &coor)
diff --git a/Grid_Communicator.h b/Grid_Communicator.h
index 4f83da34..c7e71f12 100644
--- a/Grid_Communicator.h
+++ b/Grid_Communicator.h
@@ -22,23 +22,63 @@ class CartesianCommunicator {
     MPI_Comm communicator;
 #endif
 
+    // Constructor
     CartesianCommunicator(std::vector<int> &pdimensions_in);
 
+    // Wraps MPI_Cart routines
     void ShiftedRanks(int dim,int shift,int & source, int & dest);
-    int  Rank(std::vector<int> coor);
-    int  MyRank(void);
-    void GlobalSumF(float &);
-    void GlobalSumFVector(float *,int N);
+    int  RankFromProcessorCoor(std::vector<int> &coor);
+    void ProcessorCoorFromRank(int rank,std::vector<int> &coor);
 
-    void GlobalSumF(double &);
-    void GlobalSumFVector(double *,int N);
+    /////////////////////////////////
+    // Grid information queries
+    /////////////////////////////////
+    int                      IsBoss(void)            { return _processor==0; };
+    int                      ThisRank(void)          { return _processor; };
+    const std::vector<int> & ThisProcessorCoor(void) { return _processor_coor; };
+    const std::vector<int> & ProcessorGrid(void)     { return _processors; };
+    int                      ProcessorCount(void)    { return _Nprocessors; };
 
+    ////////////////////////////////////////////////////////////
+    // Reduction
+    ////////////////////////////////////////////////////////////
+    void GlobalSum(float &);
+    void GlobalSumVector(float *,int N);
+
+    void GlobalSum(double &);
+    void GlobalSumVector(double *,int N);
+    
+    template<class obj> void GlobalSumObj(obj &o){
+
+      typedef typename obj::scalar_type scalar_type;
+      int words = sizeof(obj)/sizeof(scalar_type);
+
+      scalar_type * ptr = (scalar_type *)& o;
+      GlobalSum(ptr,words);
+    }
+    ////////////////////////////////////////////////////////////
+    // Face exchange
+    ////////////////////////////////////////////////////////////
     void SendToRecvFrom(void *xmit,
 			int xmit_to_rank,
 			void *recv,
 			int recv_from_rank,
 			int bytes);
-    
+
+    ////////////////////////////////////////////////////////////
+    // Barrier
+    ////////////////////////////////////////////////////////////
+    void Barrier(void);
+
+    ////////////////////////////////////////////////////////////
+    // Broadcast a buffer and composite larger
+    ////////////////////////////////////////////////////////////
+    void Broadcast(int root,void* data, int bytes);
+    template<class obj> void Broadcast(int root,obj &data)
+    {
+      Broadcast(root,(void *)&data,sizeof(data));
+    };
+
 }; 
 }
 
diff --git a/Grid_Lattice.h b/Grid_Lattice.h
index aab4fec9..2aa46b97 100644
--- a/Grid_Lattice.h
+++ b/Grid_Lattice.h
@@ -20,7 +20,7 @@ template<class vobj>
 class Lattice
 {
 public:
-    SimdGrid *_grid;
+    GridBase *_grid;
     int checkerboard;
     std::vector<vobj,alignedAllocator<vobj> > _odata;
     typedef typename vobj::scalar_type scalar_type;
@@ -28,7 +28,7 @@ public:
 public:
 
 
-    Lattice(SimdGrid *grid) : _grid(grid) {
+    Lattice(GridBase *grid) : _grid(grid) {
         _odata.reserve(_grid->oSites());
         assert((((uint64_t)&_odata[0])&0xF) ==0);
         checkerboard=0;
@@ -95,56 +95,67 @@ public:
     template<class sobj>
     friend void pokeSite(const sobj &s,Lattice<vobj> &l,std::vector<int> &site){
 
-      typedef typename vobj::scalar_type stype;
-      typedef typename vobj::vector_type vtype;
+      GridBase *grid=l._grid;
 
-      assert( l.checkerboard == l._grid->CheckerBoard(site));
+      typedef typename vobj::scalar_type scalar_type;
+      typedef typename vobj::vector_type vector_type;
 
-      int o_index = l._grid->oIndex(site);
-      int i_index = l._grid->iIndex(site);
+      int Nsimd = grid->Nsimd();
+
+      assert( l.checkerboard== l._grid->CheckerBoard(site));
+      assert( sizeof(sobj)*Nsimd == sizeof(vobj));
+
+      int rank,odx,idx;
+      grid->GlobalCoorToRankIndex(rank,odx,idx,site);
+
+      // Optional to broadcast from node 0.
+      grid->Broadcast(0,s);
+
+      std::vector<sobj> buf(Nsimd);
+      std::vector<scalar_type *> pointers(Nsimd);  
+      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
+
+      // extract-modify-merge cycle is easiest way and this is not perf critical
+      extract(l._odata[odx],pointers);
       
-      stype *v_ptr = (stype *)&l._odata[o_index];
-      stype *s_ptr = (stype *)&s;
-      v_ptr = v_ptr + 2*i_index;
-        
-      for(int i=0;i<sizeof(sobj);i+=2*sizeof(stype)){
-	v_ptr[0] = s_ptr[0];
-	v_ptr[1] = s_ptr[1];
-	v_ptr+=2*vtype::Nsimd();
-	s_ptr+=2;
-      }
+      buf[idx] = s;
+
+      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
+      merge(l._odata[odx],pointers);
+
       return;
     };
     
-    
     // Peek a scalar object from the SIMD array
     template<class sobj>
-    friend void peekSite(sobj &s,const Lattice<vobj> &l,std::vector<int> &site){
+    friend void peekSite(sobj &s,Lattice<vobj> &l,std::vector<int> &site){
         
-      typedef typename vobj::scalar_type stype;
-      typedef typename vobj::vector_type vtype;
+      GridBase *grid=l._grid;
+
+      typedef typename vobj::scalar_type scalar_type;
+      typedef typename vobj::vector_type vector_type;
+
+      int Nsimd = grid->Nsimd();
 
       assert( l.checkerboard== l._grid->CheckerBoard(site));
+      assert( sizeof(sobj)*Nsimd == sizeof(vobj));
 
-      int o_index = l._grid->oIndex(site);
-      int i_index = l._grid->iIndex(site);
+      int rank,odx,idx;
+      grid->GlobalCoorToRankIndex(rank,odx,idx,site);
+      std::vector<sobj> buf(Nsimd);
+      std::vector<scalar_type *> pointers(Nsimd);  
+      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
+
+      extract(l._odata[odx],pointers);
       
-      stype *v_ptr = (stype *)&l._odata[o_index];
-      stype *s_ptr = (stype *)&s;
-      v_ptr = v_ptr + 2*i_index;
-      
-      for(int i=0;i<sizeof(sobj);i+=2*sizeof(stype)){
-	s_ptr[0] = v_ptr[0];
-	s_ptr[1] = v_ptr[1];
-	v_ptr+=2*vtype::Nsimd();
-	s_ptr+=2;
-      }
+      s = buf[idx];
+      grid->Broadcast(rank,s);
+
       return;
     };
     
-    // Randomise
+    // FIXME Randomise; deprecate this
     friend void random(Lattice<vobj> &l){
-
         Real *v_ptr = (Real *)&l._odata[0];
         size_t v_len = l._grid->oSites()*sizeof(vobj);
         size_t d_len = v_len/sizeof(Real);
@@ -154,9 +165,9 @@ public:
             v_ptr[i]=drand48();
         }
     };
-
+    
+    // FIXME for debug; deprecate this
     friend void lex_sites(Lattice<vobj> &l){
-
         Real *v_ptr = (Real *)&l._odata[0];
 	size_t o_len = l._grid->oSites();
         size_t v_len = sizeof(vobj)/sizeof(vRealF);
@@ -183,7 +194,6 @@ public:
         for(int i=0;i<d_len;i++){
 	  v_ptr[i]= drand48();
         }
-        
     };
 
 
@@ -241,6 +251,7 @@ public:
         }
         return ret;
     };
+
     inline friend Lattice<vobj> conj(const Lattice<vobj> &lhs){
         Lattice<vobj> ret(lhs._grid);
 #pragma omp parallel for
@@ -259,7 +270,7 @@ public:
 	std::vector<int> coor;
 	int cbos;
 	
-	full._grid->CoordFromOsite(coor,ss);
+	full._grid->oCoorFromOindex(coor,ss);
 	cbos=half._grid->CheckerBoard(coor);
 	
 	if (cbos==cb) {
@@ -277,7 +288,7 @@ public:
 	std::vector<int> coor;
 	int cbos;
 	
-	full._grid->CoordFromOsite(coor,ss);
+	full._grid->oCoorFromOindex(coor,ss);
 	cbos=half._grid->CheckerBoard(coor);
 
 	if (cbos==cb) {
@@ -351,7 +362,6 @@ public:
     template<class left,class right>
     inline auto operator * (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs*rhs._odata[0])>
     {
-	std::cerr <<"Oscalar * Lattice calling mult"<<std::endl;
         Lattice<decltype(lhs*rhs._odata[0])> ret(rhs._grid);
 
 #pragma omp parallel for
@@ -383,7 +393,6 @@ public:
     template<class left,class right>
     inline auto operator * (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]*rhs)>
     {
-	std::cerr <<"Lattice * Oscalar calling mult"<<std::endl;
         Lattice<decltype(lhs._odata[0]*rhs)> ret(lhs._grid);
 #pragma omp parallel for
         for(int ss=0;ss<rhs._grid->oSites(); ss++){
diff --git a/Grid_QCD.h b/Grid_QCD.h
index f6cc6087..a0f7801a 100644
--- a/Grid_QCD.h
+++ b/Grid_QCD.h
@@ -21,7 +21,13 @@ namespace QCD {
 
     typedef iSinglet<Complex >          TComplex;    // This is painful. Tensor singlet complex type.
     typedef iSinglet<vComplex >         vTComplex;
-    typedef iSinglet<Real >          TReal;    // This is painful. Tensor singlet complex type.
+    typedef iSinglet<Real >             TReal;    // This is painful. Tensor singlet complex type.
+
+
+    typedef iSinglet<vIntegerF >         vTIntegerF;
+    typedef iSinglet<vIntegerD >         vTIntegerD;
+    typedef iSinglet<vIntegerC >         vTIntegerC;
+    typedef iSinglet<vIntegerZ >         vTIntegerZ;
 
     typedef iSpinMatrix<Complex >       SpinMatrix;
     typedef iColourMatrix<Complex >     ColourMatrix;
@@ -39,9 +45,13 @@ namespace QCD {
     typedef iSpinVector<vComplex >       vSpinVector;
     typedef iColourVector<vComplex >     vColourVector;
     typedef iSpinColourVector<vComplex > vSpinColourVector;
-
     
     typedef Lattice<vTComplex>         LatticeComplex;
+
+    typedef Lattice<vTIntegerF>            LatticeIntegerF; // Predicates for "where"
+    typedef Lattice<vTIntegerD>            LatticeIntegerD; 
+    typedef Lattice<vTIntegerC>            LatticeIntegerC;
+    typedef Lattice<vTIntegerZ>            LatticeIntegerZ;
     
     typedef Lattice<vColourMatrix>     LatticeColourMatrix;
     typedef Lattice<vSpinMatrix>       LatticeSpinMatrix;
diff --git a/Grid_cshift_common.h b/Grid_cshift_common.h
index 203bf2f5..72518d7d 100644
--- a/Grid_cshift_common.h
+++ b/Grid_cshift_common.h
@@ -45,7 +45,7 @@ friend void Gather_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllo
     for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
       for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
 
-	int ocb=1<<rhs._grid->CheckerBoardFromOsite(o+b);// Could easily be a table lookup
+	int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);// Could easily be a table lookup
 	if ( ocb &cbmask ) {
 	  buffer[bo]=rhs._odata[so+o+b];
 	  bo++;
@@ -90,7 +90,7 @@ friend void Gather_plane_extract(Lattice<vobj> &rhs,std::vector<scalar_type *> p
     for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
       for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
 
-	int ocb=1<<rhs._grid->CheckerBoardFromOsite(o+b);
+	int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);
 	if ( ocb & cbmask ) {
 	  extract(rhs._odata[so+o+b],pointers);
 	}
@@ -135,7 +135,7 @@ friend void Scatter_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAll
     for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
       for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
 
-	int ocb=1<<rhs._grid->CheckerBoardFromOsite(o+b);// Could easily be a table lookup
+	int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);// Could easily be a table lookup
 	if ( ocb & cbmask ) {
 	  rhs._odata[so+o+b]=buffer[bo++];
 	}
@@ -179,7 +179,7 @@ friend void Scatter_plane_merge(Lattice<vobj> &rhs,std::vector<scalar_type *> po
     for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
       for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
 
-	int ocb=1<<rhs._grid->CheckerBoardFromOsite(o+b);
+	int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);
 	if ( ocb&cbmask ) {
 	  merge(rhs._odata[so+o+b],pointers);
 	}
@@ -224,7 +224,7 @@ friend void Copy_plane(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int dimension,int
     for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
       for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
 
-	int ocb=1<<lhs._grid->CheckerBoardFromOsite(o+b);
+	int ocb=1<<lhs._grid->CheckerBoardFromOindex(o+b);
 
 	if ( ocb&cbmask ) {
 	  lhs._odata[lo+o+b]=rhs._odata[ro+o+b];
@@ -267,7 +267,7 @@ friend void Copy_plane_permute(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int dimens
     for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
       for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
 
-	int ocb=1<<lhs._grid->CheckerBoardFromOsite(o+b);
+	int ocb=1<<lhs._grid->CheckerBoardFromOindex(o+b);
 
 	if ( ocb&cbmask ) {
 	  permute(lhs._odata[lo+o+b],rhs._odata[ro+o+b],permute_type);
diff --git a/Grid_cshift_fake.h b/Grid_cshift_fake.h
index 0f306b1b..8b6c293b 100644
--- a/Grid_cshift_fake.h
+++ b/Grid_cshift_fake.h
@@ -7,9 +7,11 @@ friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
 {
   typedef typename vobj::vector_type vector_type;
   typedef typename vobj::scalar_type scalar_type;
-  const int Nsimd = vector_type::Nsimd();
 
   Lattice<vobj> ret(rhs._grid);
+
+  GridBase *grid=rhs._grid;
+  const int Nsimd = grid->Nsimd();
   
   int fd = rhs._grid->_fdimensions[dimension];
   int rd = rhs._grid->_rdimensions[dimension];
@@ -43,7 +45,7 @@ friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
 
   for(int x=0;x<rd;x++){       
 
-    for(int i=0;i<vobj::vector_type::Nsimd();i++){
+    for(int i=0;i<Nsimd;i++){
       pointers[i] = (scalar_type *)&comm_buf_extract[i][0];
     }
 
@@ -79,7 +81,7 @@ friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
 	o +=rhs._grid->_slice_stride[dimension];
       }
 
-      for(int i=0;i<vobj::vector_type::Nsimd();i++){
+      for(int i=0;i<Nsimd;i++){
 	pointers[i] = (scalar_type *)&comm_buf_extract[permute_map[permute_type][i]][0];
       }
 
diff --git a/Grid_cshift_mpi.h b/Grid_cshift_mpi.h
index 17aff2f4..75a4dc96 100644
--- a/Grid_cshift_mpi.h
+++ b/Grid_cshift_mpi.h
@@ -73,7 +73,7 @@ friend void Cshift_comms(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int
   typedef typename vobj::vector_type vector_type;
   typedef typename vobj::scalar_type scalar_type;
 
-  SimdGrid *grid=rhs._grid;
+  GridBase *grid=rhs._grid;
   Lattice<vobj> temp(rhs._grid);
 
   int fd              = rhs._grid->_fdimensions[dimension];
@@ -130,8 +130,8 @@ friend void Cshift_comms(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int
 
 friend void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int shift,int cbmask)
 {
-  const int Nsimd = vector_type::Nsimd();
-  SimdGrid *grid=rhs._grid;
+  GridBase *grid=rhs._grid;
+  const int Nsimd = grid->Nsimd();
   typedef typename vobj::vector_type vector_type;
   typedef typename vobj::scalar_type scalar_type;
    
@@ -173,6 +173,7 @@ friend void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimensi
   
   std::vector<int> comm_offnode(simd_layout);
   std::vector<int> comm_proc   (simd_layout);  //relative processor coord in dim=dimension
+  std::vector<int> icoor(grid->Nd());
 
   for(int x=0;x<rd;x++){       
 
@@ -197,7 +198,10 @@ friend void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimensi
 
       for(int i=0;i<Nsimd;i++){
 
-	int s = grid->iCoordFromIsite(i,dimension);
+
+	int s;
+	grid->iCoorFromIindex(icoor,i);
+	s = icoor[dimension];
 
 	if(comm_offnode[s]){
 
@@ -232,7 +236,7 @@ friend void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimensi
       if ( x< rd-num ) permute_slice=wrap;
       else permute_slice = 1-wrap;
 
-      for(int i=0;i<vobj::vector_type::Nsimd();i++){
+      for(int i=0;i<Nsimd;i++){
 	if ( permute_slice ) {
 	  pointers[i] = rpointers[permute_map[permute_type][i]];
 	} else {
diff --git a/Grid_main.cc b/Grid_main.cc
index 504f7778..cb45fd85 100644
--- a/Grid_main.cc
+++ b/Grid_main.cc
@@ -19,9 +19,9 @@ int main (int argc, char ** argv)
   std::vector<int> simd_layout(4);
   
   std::vector<int> mpi_layout(4);
-  mpi_layout[0]=4;
-  mpi_layout[1]=2;
-  mpi_layout[2]=4;
+  mpi_layout[0]=2;
+  mpi_layout[1]=1;
+  mpi_layout[2]=1;
   mpi_layout[3]=2;
 
 #ifdef AVX512
@@ -34,7 +34,7 @@ int main (int argc, char ** argv)
    omp_set_num_threads(omp);
 #endif 
 
-  for(int lat=16;lat<=16;lat+=40){
+  for(int lat=8;lat<=16;lat+=40){
     latt_size[0] = lat;
     latt_size[1] = lat;
     latt_size[2] = lat;
@@ -166,6 +166,7 @@ int main (int argc, char ** argv)
 
 
     // Lattice SU(3) x SU(3)
+    Fine.Barrier();
     FooBar = Foo * Bar;
     
     // Lattice 12x12 GEMM
@@ -179,37 +180,47 @@ int main (int argc, char ** argv)
 
     flops = ncall*1.0*volume*(8*Nc*Nc*Nc);
     bytes = ncall*1.0*volume*Nc*Nc    *2*3*sizeof(Grid::Real);
-    printf("%f flop and %f bytes\n",flops,bytes/ncall);
+    if ( Fine.IsBoss() ) {
+      printf("%f flop and %f bytes\n",flops,bytes/ncall);
+    }
         FooBar = Foo * Bar;
+    Fine.Barrier();
     t0=usecond();
     for(int i=0;i<ncall;i++){
-        mult(FooBar,Foo,Bar); // this is better
+      Fine.Barrier();
+      mult(FooBar,Foo,Bar); // this is better
     }
     t1=usecond();
+    Fine.Barrier();
+    if ( Fine.IsBoss() ) {
 #ifdef OMP
-    printf("mult NumThread %d , Lattice size %d , %f us per call\n",omp_get_max_threads(),lat,(t1-t0)/ncall);
+      printf("mult NumThread %d , Lattice size %d , %f us per call\n",omp_get_max_threads(),lat,(t1-t0)/ncall);
 #endif
-    printf("mult NumThread %d , Lattice size %d , %f Mflop/s\n",omp,lat,flops/(t1-t0));
-    printf("mult NumThread %d , Lattice size %d , %f MB/s\n",omp,lat,bytes/(t1-t0));
-
+      printf("mult NumThread %d , Lattice size %d , %f Mflop/s\n",omp,lat,flops/(t1-t0));
+      printf("mult NumThread %d , Lattice size %d , %f MB/s\n",omp,lat,bytes/(t1-t0));
+    }
     mult(FooBar,Foo,Bar);
     FooBar = Foo * Bar;
 
     bytes = ncall*1.0*volume*Nc*Nc    *2*5*sizeof(Grid::Real);
+    Fine.Barrier();
     t0=usecond();
     for(int i=0;i<ncall;i++){
-      mult(FooBar,Foo,Cshift(Bar,1,-2));
+      Fine.Barrier();
+      mult(FooBar,Foo,Cshift(Bar,1,-1));
       //mult(FooBar,Foo,Bar);
       //FooBar = Foo * Bar; // this is bad
     }
     t1=usecond();
+    Fine.Barrier();
 
     FooBar = Foo * Bar;
     
-    printf("Cshift Mult: NumThread %d , Lattice size %d , %f us per call\n",omp,lat,(t1-t0)/ncall);
-    printf("Cshift Mult: NumThread %d , Lattice size %d , %f Mflop/s\n",omp,lat,flops/(t1-t0));
-    printf("Cshift Mult: NumThread %d , Lattice size %d , %f MB/s\n",omp,lat,bytes/(t1-t0));
-
+    if ( Fine.IsBoss() ) {
+      printf("Cshift Mult: NumThread %d , Lattice size %d , %f us per call\n",omp,lat,(t1-t0)/ncall);
+      printf("Cshift Mult: NumThread %d , Lattice size %d , %f Mflop/s\n",omp,lat,flops/(t1-t0));
+      printf("Cshift Mult: NumThread %d , Lattice size %d , %f MB/s\n",omp,lat,bytes/(t1-t0));
+    }
     //    pickCheckerboard(0,rFoo,FooBar);
     //    pickCheckerboard(1,bFoo,FooBar);
     //    setCheckerboard(FooBar,rFoo);
@@ -225,12 +236,12 @@ int main (int argc, char ** argv)
 	pickCheckerboard(0,rFoo,Foo);    // Pick out red or black checkerboards
 	pickCheckerboard(1,bFoo,Foo);
     
-	std::cout << "Shifting both parities by "<< shift <<" direction "<< dir <<std::endl;
+	if ( Fine.IsBoss() ) {
+	  std::cout << "Shifting both parities by "<< shift <<" direction "<< dir <<std::endl;
+	}
 	Shifted  = Cshift(Foo,dir,shift);    // Shift everything
 
-	std::cout << "Shifting even source parities to odd result"<<std::endl;
 	bShifted = Cshift(rFoo,dir,shift);   // Shift red->black
-	std::cout << "Shifting odd parities to even result"<<std::endl;
 	rShifted = Cshift(bFoo,dir,shift);   // Shift black->red
     
 	ShiftedCheck=zero;
@@ -332,8 +343,10 @@ int main (int argc, char ** argv)
             nrm = nrm + real(conj(diff)*diff);
         }}
     }}}}
+	if( Fine.IsBoss() ){
+	  std::cout << "LatticeColorMatrix * LatticeColorMatrix nrm diff = "<<nrm<<std::endl;
+	}
       }}
-    std::cout << "LatticeColorMatrix * LatticeColorMatrix nrm diff = "<<nrm<<std::endl;
 
    } // loop for lat
  } // loop for omp
diff --git a/Grid_math_types.h b/Grid_math_types.h
index 3f7714d5..25524966 100644
--- a/Grid_math_types.h
+++ b/Grid_math_types.h
@@ -66,7 +66,7 @@ namespace Grid {
 template<class vtype> class iScalar
 {
 public:
-  SIMDalign vtype _internal;
+  vtype _internal;
 
   typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
   typedef typename GridTypeMapper<vtype>::vector_type vector_type;
@@ -113,7 +113,7 @@ public:
 template<class vtype,int N> class iVector
 {
 public:
-  SIMDalign vtype _internal[N];
+  vtype _internal[N];
 
   typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
   typedef typename GridTypeMapper<vtype>::vector_type vector_type;
@@ -170,7 +170,7 @@ public:
 template<class vtype,int N> class iMatrix
 {
 public:
-  SIMDalign    vtype _internal[N][N];
+  vtype _internal[N][N];
 
   typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
   typedef typename GridTypeMapper<vtype>::vector_type vector_type;
@@ -908,21 +908,5 @@ inline auto trace(const iScalar<vtype> &arg) -> iScalar<decltype(trace(arg._inte
     return ret;
 }
 };
-/////////////////////////////////////////////////////////////////////////
-// Generic routine to promote object<complex> -> object<vcomplex>
-// Supports the array reordering transformation that gives me SIMD utilisation
-/////////////////////////////////////////////////////////////////////////
-/*
-template<template<class> class object>
-inline object<vComplex> splat(object<Complex >s){
-    object<vComplex> ret;
-    vComplex * v_ptr = (vComplex *)& ret;
-    Complex * s_ptr = (Complex *) &s;
-    for(int i=0;i<sizeof(ret);i+=sizeof(vComplex)){
-        vsplat(*(v_ptr++),*(s_ptr++));
-    }
-    return ret;
-}
-*/
     
 #endif
diff --git a/Grid_mpi.cc b/Grid_mpi.cc
index c9b91210..a4943478 100644
--- a/Grid_mpi.cc
+++ b/Grid_mpi.cc
@@ -28,18 +28,18 @@ CartesianCommunicator::CartesianCommunicator(std::vector<int> &processors)
   assert(Size==_Nprocessors);
 }
 
-void CartesianCommunicator::GlobalSumF(float &f){
+void CartesianCommunicator::GlobalSum(float &f){
   MPI_Allreduce(&f,&f,1,MPI_FLOAT,MPI_SUM,communicator);
 }
-void CartesianCommunicator::GlobalSumFVector(float *f,int N)
+void CartesianCommunicator::GlobalSumVector(float *f,int N)
 {
   MPI_Allreduce(f,f,N,MPI_FLOAT,MPI_SUM,communicator);
 }
-void CartesianCommunicator::GlobalSumF(double &d)
+void CartesianCommunicator::GlobalSum(double &d)
 {
   MPI_Allreduce(&d,&d,1,MPI_DOUBLE,MPI_SUM,communicator);
 }
-void CartesianCommunicator::GlobalSumFVector(double *d,int N)
+void CartesianCommunicator::GlobalSumVector(double *d,int N)
 {
   MPI_Allreduce(d,d,N,MPI_DOUBLE,MPI_SUM,communicator);
 }
@@ -48,12 +48,17 @@ void CartesianCommunicator::ShiftedRanks(int dim,int shift,int &source,int &dest
 {
   MPI_Cart_shift(communicator,dim,shift,&source,&dest);
 }
-int CartesianCommunicator::Rank(std::vector<int> coor)
+int CartesianCommunicator::RankFromProcessorCoor(std::vector<int> &coor)
 {
   int rank;
   MPI_Cart_rank  (communicator, &coor[0], &rank);
   return rank;
 }
+void  CartesianCommunicator::ProcessorCoorFromRank(int rank, std::vector<int> &coor)
+{
+  coor.resize(_ndimension);
+  MPI_Cart_coords  (communicator, rank, _ndimension,&coor[0]);
+}
 
 // Basic Halo comms primitive
 void CartesianCommunicator::SendToRecvFrom(void *xmit,
@@ -71,5 +76,19 @@ void CartesianCommunicator::SendToRecvFrom(void *xmit,
 
 }
 
+void CartesianCommunicator::Barrier(void)
+{
+  MPI_Barrier(communicator);
+}
+
+void CartesianCommunicator::Broadcast(int root,void* data, int bytes)
+{
+  MPI_Bcast(data,
+	    bytes,
+	    MPI_BYTE,
+	    root,
+	    communicator);
+}
+
 }
 
diff --git a/Grid_simd.h b/Grid_simd.h
index b5d34d4a..ffcbff15 100644
--- a/Grid_simd.h
+++ b/Grid_simd.h
@@ -37,6 +37,29 @@
     // Fourier transform equivalent.
     //
     
+  ////////////////////////////////////////////////////////////
+  // SIMD Alignment controls
+  ////////////////////////////////////////////////////////////
+#ifdef HAVE_VAR_ATTRIBUTE_ALIGNED
+#define ALIGN_DIRECTIVE(A) __attribute__ ((aligned(A)))
+#else
+#define ALIGN_DIRECTIVE(A) __declspec(align(A))
+#endif
+
+#ifdef SSE2
+#include <pmmintrin.h>
+#define SIMDalign ALIGN_DIRECTIVE(16)
+#endif
+
+#if defined(AVX1) || defined (AVX2)
+#include <immintrin.h>
+#define SIMDalign ALIGN_DIRECTIVE(32)
+#endif
+
+#ifdef AVX512
+#include <immintrin.h>
+#define SIMDalign ALIGN_DIRECTIVE(64)
+#endif
 
 namespace Grid {
 
@@ -49,6 +72,8 @@ namespace Grid {
   typedef std::complex<Real>  Complex;
 
 
+
+
   inline RealF adj(const RealF  & r){ return r; }
   inline RealF conj(const RealF  & r){ return r; }
   inline ComplexD localInnerProduct(const ComplexD & l, const ComplexD & r) { return conj(l)*r; }
@@ -97,47 +122,27 @@ namespace Grid {
   template<>            inline void ZeroIt(RealD &arg){ arg=0; };
 
 
-  ////////////////////////////////////////////////////////////
-  // SIMD Alignment controls
-  ////////////////////////////////////////////////////////////
-#ifdef HAVE_VAR_ATTRIBUTE_ALIGNED
-#define ALIGN_DIRECTIVE(A) __attribute__ ((aligned(A)))
-#else
-#define ALIGN_DIRECTIVE(A) __declspec(align(A))
-#endif
-
-#ifdef SSE2
-#include <pmmintrin.h>
-#define SIMDalign ALIGN_DIRECTIVE(16)
-#endif
-
-#if defined(AVX1) || defined (AVX2)
-#include <immintrin.h>
-#define SIMDalign ALIGN_DIRECTIVE(32)
-#endif
-
-#ifdef AVX512
-#include <immintrin.h>
-#define SIMDalign ALIGN_DIRECTIVE(64)
-#endif
 
 #if defined (SSE2)
     typedef __m128 fvec;
     typedef __m128d dvec;
     typedef __m128 cvec;
     typedef __m128d zvec;
+    typedef __m128i ivec;
 #endif
 #if defined (AVX1) || defined (AVX2)
     typedef __m256 fvec;
     typedef __m256d dvec;
     typedef __m256  cvec;
     typedef __m256d zvec;
+    typedef __m256i ivec;
 #endif
 #if defined (AVX512)
     typedef __m512  fvec;
     typedef __m512d dvec;
     typedef __m512  cvec;
     typedef __m512d zvec;
+    typedef __m512i ivec;
 #endif
 #if defined (QPX)
     typedef float  fvec __attribute__ ((vector_size (16))); // QPX has same SIMD width irrespective of precision
@@ -159,10 +164,76 @@ namespace Grid {
 
 };
 
+
+/////////////////////////////////////////////////////////////////
+// Generic extract/merge/permute
+/////////////////////////////////////////////////////////////////
+template<class vsimd,class scalar,int Nsimd>
+inline void Gextract(vsimd &y,std::vector<scalar *> &extracted){
+  // Bounce off stack is painful
+  // temporary hack while I figure out the right interface
+  scalar buf[Nsimd]; 
+  vstore(y,buf);
+  for(int i=0;i<Nsimd;i++){
+    *extracted[i] = buf[i];
+    extracted[i]++;
+  }
+};
+template<class vsimd,class scalar,int Nsimd>
+inline void Gmerge(vsimd &y,std::vector<scalar *> &extracted){
+  scalar buf[Nsimd]; 
+  for(int i=0;i<Nsimd;i++){
+    buf[i]=*extracted[i];
+    extracted[i]++;
+  }
+  vset(y,buf); 
+};
+
+//////////////////////////////////////////////////////////
+// Permute
+// Permute 0 every ABCDEFGH -> BA DC FE HG
+// Permute 1 every ABCDEFGH -> CD AB GH EF
+// Permute 2 every ABCDEFGH -> EFGH ABCD
+// Permute 3 possible on longer iVector lengths (512bit = 8 double = 16 single)
+// Permute 4 possible on half precision @512bit vectors.
+//////////////////////////////////////////////////////////
+// Should be able to make the permute/extract/merge independent of the
+// vector subtype and reduce the volume of code.
+template<class vsimd>
+inline void Gpermute(vsimd &y,vsimd b,int perm){
+      switch (perm){
+#if defined(AVX1)||defined(AVX2)
+      // 8x32 bits=>3 permutes
+      case 2: y.v = _mm256_shuffle_ps(b.v,b.v,_MM_SHUFFLE(2,3,0,1)); break;
+      case 1: y.v = _mm256_shuffle_ps(b.v,b.v,_MM_SHUFFLE(1,0,3,2)); break;
+      case 0: y.v = _mm256_permute2f128_ps(b.v,b.v,0x01); break;
+#endif
+#ifdef SSE2
+      case 1: y.v = _mm_shuffle_ps(b.v,b.v,_MM_SHUFFLE(2,3,0,1)); break;
+      case 0: y.v = _mm_shuffle_ps(b.v,b.v,_MM_SHUFFLE(1,0,3,2));break;
+#endif
+#ifdef AVX512
+	// 16 floats=> permutes
+        // Permute 0 every abcd efgh ijkl mnop -> badc fehg jilk nmpo 
+        // Permute 1 every abcd efgh ijkl mnop -> cdab ghef jkij opmn 
+        // Permute 2 every abcd efgh ijkl mnop -> efgh abcd mnop ijkl
+        // Permute 3 every abcd efgh ijkl mnop -> ijkl mnop abcd efgh
+      case 3: y.v = _mm512_swizzle_ps(b.v,_MM_SWIZ_REG_CDAB); break;
+      case 2: y.v = _mm512_swizzle_ps(b.v,_MM_SWIZ_REG_BADC); break;
+      case 1: y.v = _mm512_permute4f128_ps(b.v,(_MM_PERM_ENUM)_MM_SHUFFLE(2,3,0,1)); break;
+      case 0: y.v = _mm512_permute4f128_ps(b.v,(_MM_PERM_ENUM)_MM_SHUFFLE(1,0,3,2)); break;
+#endif
+#ifdef QPX
+#error not implemented
+#endif
+      default: assert(0); break;
+      }
+    };
+
 #include <Grid_vRealF.h>
 #include <Grid_vRealD.h>
 #include <Grid_vComplexF.h>
 #include <Grid_vComplexD.h>
-
+#include <Grid_vInteger.h>
 
 #endif
diff --git a/Grid_vInteger.h b/Grid_vInteger.h
new file mode 100644
index 00000000..7a0c1c4f
--- /dev/null
+++ b/Grid_vInteger.h
@@ -0,0 +1,307 @@
+#ifndef VINTEGER_H
+#define VINTEGER_H
+
+#include "Grid.h"
+
+namespace Grid {
+
+#define _mm256_set_m128i(hi,lo) _mm256_insertf128_si256(_mm256_castsi128_si256(lo),(hi),1)
+// _mm256_set_m128i(hi,lo); // not defined in all versions of immintrin.h
+
+  typedef uint32_t Integer;
+
+    class vInteger {
+    protected:
+
+    public:
+
+      ivec v;
+
+	typedef ivec     vector_type;
+	typedef Integer scalar_type;
+
+        vInteger(){};
+        ////////////////////////////////////
+        // Arithmetic operator overloads +,-,*
+        ////////////////////////////////////
+        friend inline vInteger operator + ( vInteger a,  vInteger b)
+        {
+	  vInteger ret;
+#if defined (AVX1) 
+	  __m128i a0,a1;
+	  __m128i b0,b1;
+	  a0 = _mm256_extractf128_si256(a.v,0);
+	  b0 = _mm256_extractf128_si256(b.v,0);
+	  a1 = _mm256_extractf128_si256(a.v,1);
+	  b1 = _mm256_extractf128_si256(b.v,1);
+	  a0 = _mm_add_epi32(a0,b0);
+	  a1 = _mm_add_epi32(a1,b1);
+	  ret.v = _mm256_set_m128i(a1,a0);
+#endif
+#if defined (AVX2)
+            ret.v = _mm256_add_epi32(a.v,b.v);
+#endif
+#ifdef SSE2
+            ret.v = _mm_add_epi32(a.v,b.v);
+#endif
+#ifdef AVX512
+            ret.v = _mm512_add_epi32(a.v,b.v);
+#endif
+#ifdef QPX
+            // Implement as array of ints is only option
+#error
+#endif
+            return ret;
+        };
+        
+        friend inline vInteger operator - ( vInteger a, vInteger b)
+        {
+            vInteger ret;
+#if defined (AVX1) 
+	  __m128i a0,a1;
+	  __m128i b0,b1;
+	  a0 = _mm256_extractf128_si256(a.v,0);
+	  b0 = _mm256_extractf128_si256(b.v,0);
+	  a1 = _mm256_extractf128_si256(a.v,1);
+	  b1 = _mm256_extractf128_si256(b.v,1);
+	  a0 = _mm_sub_epi32(a0,b0);
+	  a1 = _mm_sub_epi32(a1,b1);
+	  ret.v = _mm256_set_m128i(a1,a0);
+#endif
+#if defined (AVX2)
+            ret.v = _mm256_sub_epi32(a.v,b.v);
+#endif
+#ifdef SSE2
+            ret.v = _mm_sub_epi32(a.v,b.v);
+#endif
+#ifdef AVX512
+            ret.v = _mm512_sub_epi32(a.v,b.v);
+#endif
+#ifdef QPX
+            // Implement as array of ints is only option
+#error
+#endif
+            return ret;
+        };
+
+        friend inline vInteger operator * ( vInteger a, vInteger b)
+        {
+            vInteger ret;
+#if defined (AVX1) 
+	  __m128i a0,a1;
+	  __m128i b0,b1;
+	  a0 = _mm256_extractf128_si256(a.v,0);
+	  b0 = _mm256_extractf128_si256(b.v,0);
+	  a1 = _mm256_extractf128_si256(a.v,1);
+	  b1 = _mm256_extractf128_si256(b.v,1);
+	  a0 = _mm_mul_epi32(a0,b0);
+	  a1 = _mm_mul_epi32(a1,b1);
+	  ret.v = _mm256_set_m128i(a1,a0);
+#endif
+#if defined (AVX2)
+            ret.v = _mm256_mul_epi32(a.v,b.v);
+#endif
+#ifdef SSE2
+            ret.v = _mm_mul_epi32(a.v,b.v);
+#endif
+#ifdef AVX512
+            ret.v = _mm512_mul_epi32(a.v,b.v);
+#endif
+#ifdef QPX
+            // Implement as array of ints is only option
+#error
+#endif
+            return ret;
+        };
+        
+        ///////////////////////////////////////////////
+        // mult, sub, add, adj,conj, mac functions
+        ///////////////////////////////////////////////
+        friend inline void mult(vInteger * __restrict__ y,const vInteger * __restrict__ l,const vInteger *__restrict__ r) {*y = (*l) * (*r);}
+        friend inline void sub (vInteger * __restrict__ y,const vInteger * __restrict__ l,const vInteger *__restrict__ r) {*y = (*l) - (*r);}
+        friend inline void add (vInteger * __restrict__ y,const vInteger * __restrict__ l,const vInteger *__restrict__ r) {*y = (*l) + (*r);}
+        friend inline void mac (vInteger &y,const vInteger a,const vInteger x){
+            y = a*x+y;
+	}
+        
+        //////////////////////////////////
+        // Initialise to 1,0,i
+        //////////////////////////////////
+        friend inline void vone (vInteger &ret){vsplat(ret,1);}
+        friend inline void vzero(vInteger &ret){vsplat(ret,0);}
+        friend inline void vtrue (vInteger &ret){vsplat(ret,0xFFFFFFFF);}
+        friend inline void vfalse(vInteger &ret){vsplat(ret,0);}
+
+        
+        /////////////////////////////////////////////////////
+        // Broadcast a value across Nsimd copies.
+        /////////////////////////////////////////////////////
+        friend inline void vsplat(vInteger &ret,scalar_type a){
+#if defined (AVX1)|| defined (AVX2)
+            ret.v = _mm256_set1_epi32(a);
+#endif
+#ifdef SSE2
+            ret.v = _mm_set1_epi32(a);
+#endif
+#ifdef AVX512
+            ret.v = _mm512_set1_epi32(a);
+#endif
+#ifdef QPX
+#error
+#endif
+        }
+        friend inline void vset(vInteger &ret,scalar_type *a){
+#if defined (AVX1)|| defined (AVX2)
+	  ret.v = _mm256_set_epi32(a[7],a[6],a[5],a[4],a[3],a[2],a[1],a[0]);
+#endif
+#ifdef SSE2
+	  ret.v = _mm_set_epi32(a[0],a[1],a[2],a[3]);
+#endif
+#ifdef AVX512
+	  ret.v = _mm512_set_epi32( a[15],a[14],a[13],a[12],a[11],a[10],a[9],a[8],
+                                   a[7],a[6],a[5],a[4],a[3],a[2],a[1],a[0]);
+#endif
+#ifdef QPX
+#error
+#endif
+	}
+
+friend inline void vstore(vInteger &ret, Integer *a){
+#if defined (AVX1)|| defined (AVX2)
+        _mm256_store_si256((__m256i*)a,ret.v);
+#endif
+#ifdef SSE2
+	_mm_store_si128(a,ret.v);
+#endif
+#ifdef AVX512
+	_mm512_store_si512(a,ret.v);
+#endif
+#ifdef QPX
+	assert(0);
+#endif
+        }
+
+        friend inline void vprefetch(const vInteger &v)
+        {
+            _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
+        }
+        // Unary negation
+        friend inline vInteger operator -(const vInteger &r) {
+            vInteger ret;
+            vzero(ret);
+            ret = ret - r;
+            return ret;
+        }
+       friend inline Integer Reduce(const vInteger & in)
+       {
+	 // unimplemented
+	 assert(0);
+       }
+        // *=,+=,-= operators
+        inline vInteger &operator *=(const vInteger &r) {
+            *this = (*this)*r;
+            return *this;
+        }
+        inline vInteger &operator +=(const vInteger &r) {
+            *this = *this+r;
+            return *this;
+        }
+        inline vInteger &operator -=(const vInteger &r) {
+            *this = *this-r;
+            return *this;
+        }
+    public:
+        static inline int Nsimd(void) { return sizeof(fvec)/sizeof(float);}
+    };
+
+    inline vInteger localInnerProduct(const vInteger & l, const vInteger & r) { return l*r; }
+
+    inline void  zeroit(vInteger &z){ vzero(z);}
+
+    inline vInteger outerProduct(const vInteger &l, const vInteger& r)
+    {
+        return l*r;
+    }
+ 
+
+    class vIntegerF : public vInteger
+    {
+    public:
+      static inline int Nsimd(void) { return sizeof(ivec)/sizeof(float);}
+      
+      friend inline void permute(vIntegerF &y,vIntegerF b,int perm)
+      {
+	Gpermute<vIntegerF>(y,b,perm);
+      }
+      friend inline void merge(vIntegerF &y,std::vector<Integer *> &extracted)
+      {
+	Gmerge<vIntegerF,Integer,sizeof(ivec)/sizeof(float) >(y,extracted);
+      }
+      friend inline void extract(vIntegerF &y,std::vector<Integer *> &extracted)
+      {
+	Gextract<vIntegerF,Integer,sizeof(ivec)/sizeof(float) >(y,extracted);
+      }
+    };
+
+
+    class vIntegerD : public vInteger
+    {
+    public:
+      static inline int Nsimd(void) { return sizeof(ivec)/sizeof(double);}
+      
+      friend inline void permute(vIntegerD &y,vIntegerD b,int perm)
+      {
+	Gpermute<vIntegerD>(y,b,perm);
+      }
+      friend inline void merge(vIntegerD &y,std::vector<Integer *> &extracted)
+      {
+	Gmerge<vIntegerD,Integer,sizeof(ivec)/sizeof(double) >(y,extracted);
+      }
+      friend inline void extract(vIntegerD &y,std::vector<Integer *> &extracted)
+      {
+	Gextract<vIntegerD,Integer,sizeof(ivec)/sizeof(double) >(y,extracted);
+      }
+    };
+
+
+    class vIntegerC : public vInteger
+    {
+    public:
+      static inline int Nsimd(void) { return sizeof(ivec)/sizeof(ComplexF);}
+      
+      friend inline void permute(vIntegerC &y,vIntegerC b,int perm)
+      {
+	Gpermute<vIntegerC>(y,b,perm);
+      }
+      friend inline void merge(vIntegerC &y,std::vector<Integer *> &extracted)
+      {
+	Gmerge<vIntegerC,Integer,sizeof(ivec)/sizeof(ComplexF) >(y,extracted);
+      }
+      friend inline void extract(vIntegerC &y,std::vector<Integer *> &extracted)
+      {
+	Gextract<vIntegerC,Integer,sizeof(ivec)/sizeof(ComplexF) >(y,extracted);
+      }
+    };
+
+    class vIntegerZ : public vInteger
+    {
+    public:
+      static inline int Nsimd(void) { return sizeof(ivec)/sizeof(ComplexD);}
+      
+      friend inline void permute(vIntegerZ &y,vIntegerZ b,int perm)
+      {
+	Gpermute<vIntegerZ>(y,b,perm);
+      }
+      friend inline void merge(vIntegerZ &y,std::vector<Integer *> &extracted)
+      {
+	Gmerge<vIntegerZ,Integer,sizeof(ivec)/sizeof(ComplexD) >(y,extracted);
+      }
+      friend inline void extract(vIntegerZ &y,std::vector<Integer *> &extracted)
+      {
+	Gextract<vIntegerZ,Integer,sizeof(ivec)/sizeof(ComplexD) >(y,extracted);
+      }
+    };
+
+}
+
+#endif
diff --git a/TODO b/TODO
index 8108fbc6..9febd6c3 100644
--- a/TODO
+++ b/TODO
@@ -1,8 +1,9 @@
-* Make peekSite, pokeSite work in an MPI context.
+* FIXME audit
 
 * Conditional execution Subset, where etc...
 * Coordinate information, integers etc...
 * Integer type padding/union to vector.
+* LatticeCoordinate[mu]
 
 * Optimise the extract/merge SIMD routines
 

From 9e597ac50a77caf5c5e8c06cd4ffbf39f9001768 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 6 Apr 2015 09:27:17 +0100
Subject: [PATCH 038/429] Removing older file

---
 Grid_cshift_fake.h | 248 ---------------------------------------------
 1 file changed, 248 deletions(-)
 delete mode 100644 Grid_cshift_fake.h

diff --git a/Grid_cshift_fake.h b/Grid_cshift_fake.h
deleted file mode 100644
index 8b6c293b..00000000
--- a/Grid_cshift_fake.h
+++ /dev/null
@@ -1,248 +0,0 @@
-#ifndef _GRID_FAKE_H_
-#define _GRID_FAKE_H_
-
-      
-
-friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
-{
-  typedef typename vobj::vector_type vector_type;
-  typedef typename vobj::scalar_type scalar_type;
-
-  Lattice<vobj> ret(rhs._grid);
-
-  GridBase *grid=rhs._grid;
-  const int Nsimd = grid->Nsimd();
-  
-  int fd = rhs._grid->_fdimensions[dimension];
-  int rd = rhs._grid->_rdimensions[dimension];
-  //int ld = rhs._grid->_ldimensions[dimension];
-  //int gd = rhs._grid->_gdimensions[dimension];
-  
-
-  // Map to always positive shift modulo global full dimension.
-  shift = (shift+fd)%fd;
-
-  ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
-        
-  // the permute type
-
-  int permute_dim =rhs._grid->_simd_layout[dimension]>1 ;
-  int permute_type=0;
-  for(int d=0;d<dimension;d++){
-    if (rhs._grid->_simd_layout[d]>1 ) permute_type++;
-  }
-
-  ///////////////////////////////////////////////
-  // Move via a fake comms buffer
-  // Simd direction uses an extract/merge pair
-  ///////////////////////////////////////////////
-  int buffer_size = rhs._grid->_slice_nblock[dimension]*rhs._grid->_slice_block[dimension];
-  int words = sizeof(vobj)/sizeof(vector_type);
-
-  std::vector<vobj,alignedAllocator<vobj> > comm_buf(buffer_size);
-  std::vector<std::vector<scalar_type> > comm_buf_extract(Nsimd,std::vector<scalar_type>(buffer_size*words) );
-  std::vector<scalar_type *> pointers(Nsimd);
-
-  for(int x=0;x<rd;x++){       
-
-    for(int i=0;i<Nsimd;i++){
-      pointers[i] = (scalar_type *)&comm_buf_extract[i][0];
-    }
-
-    int ro  = x*rhs._grid->_ostride[dimension]; // base offset for result
-
-    if ( permute_dim ) {
-
-      int o   = 0;                                // relative offset to base
-      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-	  
-	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
-	  
-	  int sx = (x+sshift)%rd;
-	  
-	  // base offset for source
-	  int so = sx*rhs._grid->_ostride[dimension];
-
-	  int permute_slice=0;
-	  int wrap = sshift/rd;
-	  int  num = sshift%rd;
-	    
-	  if ( x< rd-num ) permute_slice=wrap;
-	  else permute_slice = 1-wrap;
-
-	  if ( permute_slice ) {
-	    extract(rhs._odata[so+o+b],pointers);
-	  } else {
-	    ret._odata[ro+o+b]=rhs._odata[so+o+b];
-	  }
-
-	}
-	o +=rhs._grid->_slice_stride[dimension];
-      }
-
-      for(int i=0;i<Nsimd;i++){
-	pointers[i] = (scalar_type *)&comm_buf_extract[permute_map[permute_type][i]][0];
-      }
-
-      o   = 0;                                // relative offset to base
-      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-	  
-	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
-	  int sx = (x+sshift)%rd;
-	  
-	  // base offset for source
-	  int so = sx*rhs._grid->_ostride[dimension];
-
-	  int permute_slice=0;
-	  int wrap = sshift/rd;
-	  int  num = sshift%rd;
-	    
-	  if ( x< rd-num ) permute_slice=wrap;
-	  else permute_slice = 1-wrap;
-
-	  if ( permute_slice ) {
-	    merge(ret._odata[ro+o+b],pointers);
-	  }
-	}
-	o +=rhs._grid->_slice_stride[dimension];
-      }
-
-    } else {
-
-      int co; // comm offset
-      int o;
-
-      co=0; o=0;
-      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-
-	  // This call in inner loop is annoying but necessary for dimension=0
-	  // in the case of RedBlack grids. Could optimise away with 
-	  // alternate code paths for all other cases.
-	  int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,o+b);
-	  int sx = (x+sshift)%rd;
-	  int so = sx*rhs._grid->_ostride[dimension];
-
-	  comm_buf[co++]=rhs._odata[so+o+b];
-
-	}
-	o +=rhs._grid->_slice_stride[dimension];
-      }
-
-      // Step through a copy into a comms buffer and pull back in.
-      // Genuine fake implementation could calculate if loops back
-      co=0; o=0;
-      for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-	  ret._odata[ro+o+b]=comm_buf[co++];
-	}
-	o +=rhs._grid->_slice_stride[dimension];
-      }
-    }
-  }
-  return ret;
-}
-
-/*
-    friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
-    {
-      Lattice<vobj> ret(rhs._grid);
-        
-        int rd = rhs._grid->_rdimensions[dimension];
-        int ld = rhs._grid->_ldimensions[dimension];
-        int gd = rhs._grid->_gdimensions[dimension];
-        
-        // Map to always positive shift.
-        shift = (shift+gd)%gd;
-
-        ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
-        shift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift);
-        
-        // Work out whether to permute and the permute type
-        // ABCDEFGH ->   AE BF CG DH       permute
-        // Shift 0       AE BF CG DH       0 0 0 0    ABCDEFGH
-        // Shift 1       BF CG DH AE       0 0 0 1    BCDEFGHA
-        // Shift 2       CG DH AE BF       0 0 1 1    CDEFGHAB
-        // Shift 3       DH AE BF CG       0 1 1 1    DEFGHABC
-        // Shift 4       AE BF CG DH       1 1 1 1    EFGHABCD
-        // Shift 5       BF CG DH AE       1 1 1 0    FGHACBDE
-        // Shift 6       CG DH AE BF       1 1 0 0    GHABCDEF
-        // Shift 7       DH AE BF CG       1 0 0 0    HABCDEFG
-
-        int permute_dim =rhs._grid->_simd_layout[dimension]>1 ;
-        int permute_type=0;
-        for(int d=0;d<dimension;d++)
-            if (rhs._grid->_simd_layout[d]>1 ) permute_type++;
-        
-        
-        // loop over all work
-        int work =rd*rhs._grid->_slice_nblock[dimension]*rhs._grid->_slice_block[dimension];
-
-	// Packed gather sequence is clean
-	int buffer_size = rhs._grid->_slice_nblock[dimension]*rhs._grid->_slice_block[dimension];
-	
-	typedef typename vobj::scalar_type scalar_t;
-	typedef typename vobj::vector_type vector_t;
-	const int ns=sizeof(vobj)/sizeof(scalar_t);
-	const int nv=sizeof(vobj)/sizeof(vector_t);
-	std::vector<vobj,alignedAllocator<vobj> > comm_buf(buffer_size);
-
-        for(int x=0;x<rd;x++){       
-	  
-	  int sx = (x+shift)%rd;
-	  int o  = x*rhs._grid->_ostride[dimension];
-	  int so =sx*rhs._grid->_ostride[dimension];
-
-	
-	  int permute_slice=0;
-	  if ( permute_dim ) {
-	    permute_slice = shift/rd;
-	    if ( x<shift%rd ) permute_slice = 1-permute_slice;
-	  }
-
-	  if ( permute_slice ) {
-	    exit(0);
-	    // For fake communication ALWAYS extract and either merge one way or other
-	    scalar_t * bptr = (scalar_t *) &comm_buf[0];
-
-	    int bo=0;
-	    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	    
-	      vector_t *optr = (vector_t *)&ret._odata[o];
-	      vector_t *iptr = (vector_t *)&rhs._odata[so];
-	      int skew    = buffer_size*ns/2;
-
-	      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){ 
-		for(int n=0;n<nv;n++){// number of simd vecsscalars in a vector
-		  extract(iptr[b*nv+n],&bptr[n],skew,permute_type);
-		}
-	      }
-	      o+=rhs._grid->_slice_stride[dimension];
-	      //	      bo+=rhs._grid->_slice_stride[dimension]*ns/2;
-
-	    }
-
-	  } else {
-	    int bo=0;
-	    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	      for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
-		comm_buf[bo++] =rhs._odata[so+i];
-	      }
-	      so+=rhs._grid->_slice_stride[dimension];
-	    }
-	    bo=0;
-	    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-	      for(int i=0;i<rhs._grid->_slice_block[dimension];i++){
-		ret._odata[o+i]=comm_buf[bo++];
-	      }
-	      o+=rhs._grid->_slice_stride[dimension];
-	    }
-	  }
-	}
-        return ret;
-    };
-*/
-
-#endif

From 982274e5a0e9f4433dd05be420e31fb98fbc4357 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 6 Apr 2015 11:26:24 +0100
Subject: [PATCH 039/429] Major rework of extract/merge/permute processing
 debugged and working.

---
 Grid.h                                    |   3 +-
 Grid_Cartesian.h                          |  58 +++---------
 Grid_Lattice.h                            |  20 +----
 Grid_QCD.h                                |  12 +--
 Grid_aligned_allocator.h                  |   3 +
 Grid_fake.cc => Grid_communicator_fake.cc |   0
 Grid_mpi.cc => Grid_communicator_mpi.cc   |   0
 Grid_config.h                             |   3 -
 Grid_config.h.in                          |   3 -
 Grid_cshift_common.h                      |  76 +++-------------
 Grid_cshift_mpi.h                         |  14 ++-
 Grid_main.cc                              |   4 +-
 Grid_simd.h                               |  74 ++++++----------
 Grid_stencil.h                            |  12 +++
 Grid_vComplexD.h                          |  81 +++++------------
 Grid_vComplexF.h                          |  92 +++++--------------
 Grid_vInteger.h                           |  63 +------------
 Grid_vRealD.h                             |  87 +++++-------------
 Grid_vRealF.h                             |  90 +++++--------------
 Makefile.am                               |   8 +-
 Makefile.in                               |  29 +++---
 TODO                                      | 103 ++++++++++++++++++++++
 configure                                 |  23 +----
 configure.ac                              |   7 +-
 24 files changed, 291 insertions(+), 574 deletions(-)
 rename Grid_fake.cc => Grid_communicator_fake.cc (100%)
 rename Grid_mpi.cc => Grid_communicator_mpi.cc (100%)
 create mode 100644 Grid_stencil.h

diff --git a/Grid.h b/Grid.h
index 184e7e36..0a4024bf 100644
--- a/Grid.h
+++ b/Grid.h
@@ -42,11 +42,10 @@
 #endif
 
 
+#include <Grid_aligned_allocator.h>
 #include <Grid_simd.h>
 #include <Grid_math_types.h>
 #include <Grid_Cartesian.h>
-#include <Grid_aligned_allocator.h>
-#include <Grid_aligned_allocator.h>
 #include <Grid_Lattice.h>
 #include <Grid_QCD.h>
 
diff --git a/Grid_Cartesian.h b/Grid_Cartesian.h
index 004c2fa5..c891edcc 100644
--- a/Grid_Cartesian.h
+++ b/Grid_Cartesian.h
@@ -8,48 +8,6 @@ namespace Grid{
 /////////////////////////////////////////////////////////////////////////////////////////
 // Grid Support.
 /////////////////////////////////////////////////////////////////////////////////////////
-//
-// Cartesian grid inheritance
-//            Grid::GridBase
-//                     |
-//           __________|___________
-//          |                      |
-// Grid::GridCartesian   Grid::GridCartesianRedBlack
-//
-// TODO: document the following as an API guaranteed public interface
-
-    /* 
-     *       Rough map of functionality against QDP++ Layout
-     *
-     *       Param     |     Grid                     |     QDP++             
-     *       -----------------------------------------
-     *                 |                              |
-     *        void     |     oSites, iSites, lSites   |  sitesOnNode 
-     *        void     |     gSites                   |  vol
-     *                 |                              |
-     *        gcoor    |     oIndex, iIndex           |  linearSiteIndex // no virtual node in QDP
-     *        lcoor    |                              |
-     * 
-     *        void     |     CheckerBoarded           |  -        // No checkerboarded in QDP
-     *        void     |     FullDimensions           |  lattSize
-     *        void     |     GlobalDimensions         |  lattSize // No checkerboarded in QDP
-     *        void     |     LocalDimensions          |  subgridLattSize
-     *        void     |     VirtualLocalDimensions   |  subgridLattSize // no virtual node in QDP
-     *                 |                              |
-     *       int x 3   |     oiSiteRankToGlobal       |  siteCoords
-     *                 |     ProcessorCoorLocalCoorToGlobalCoor | 
-     *                 |                              |
-     *     vector<int> |     GlobalCoorToRankIndex   |  nodeNumber(coord)
-     *     vector<int> |     GlobalCoorToProcessorCoorLocalCoor|  nodeCoord(coord)
-     *                 |                              |
-     *     void        |     Processors               |  logicalSize    // returns cart array shape
-     *     void        |     ThisRank        |  nodeNumber();  // returns this node rank
-     *     void        |     ThisProcessorCoor        |    // returns this node coor
-     *     void        |     isBoss(void)             |  primaryNode();
-     *                 |                              |
-     *                 |     RankFromProcessorCoor    |  getLogicalCoorFrom(node)
-     *                 |     ProcessorCoorFromRank    |  getNodeNumberFrom(logical_coord)
-     */
 
 class GridBase : public CartesianCommunicator {
 public:
@@ -60,7 +18,8 @@ public:
  GridBase(std::vector<int> & processor_grid) : CartesianCommunicator(processor_grid) {};
 
         
- //protected:
+ //FIXME 
+ // protected:
  // Lattice wide random support. not yet fully implemented. Need seed strategy
  // and one generator per site.
  // std::default_random_engine generator;
@@ -165,7 +124,16 @@ public:
 	lane    = lane / _simd_layout[d];
       }
     }
-    
+    inline int PermuteDim(int dimension){
+      return _simd_layout[dimension]>1;
+    }
+    inline int PermuteType(int dimension){
+      int permute_type=0;
+      for(int d=_ndimension-1;d>dimension;d--){
+	if (_simd_layout[d]>1 ) permute_type++;
+      }
+      return permute_type;
+    }
     ////////////////////////////////////////////////////////////////
     // Array sizing queries
     ////////////////////////////////////////////////////////////////
@@ -399,8 +367,6 @@ public:
             
         ////////////////////////////////////////////////////////////////////////////////////////////
         // subplane information
-        // It may be worth the investment of generating a more general subplane "iterator",
-        // and providing support for threads grabbing a unit of allocation.
         ////////////////////////////////////////////////////////////////////////////////////////////
         _slice_block.resize(_ndimension);
         _slice_stride.resize(_ndimension);
diff --git a/Grid_Lattice.h b/Grid_Lattice.h
index 2aa46b97..ebf68560 100644
--- a/Grid_Lattice.h
+++ b/Grid_Lattice.h
@@ -4,17 +4,9 @@
 #include "Grid.h"
 
 
-
 namespace Grid {
 
-// Permute the pointers 32bitx16 = 512
-static int permute_map[4][16] = { 
-  { 1,0,3,2,5,4,7,6,9,8,11,10,13,12,15,14},
-  { 2,3,0,1,6,7,4,5,10,11,8,9,14,15,12,13},
-  { 4,5,6,7,0,1,2,3,12,13,14,15,8,9,10,11},
-  { 9,10,11,12,13,14,15,0,1,2,3,4,5,6,7,8}
-};
-
+  extern int GridCshiftPermuteMap[4][16];
 
 template<class vobj>
 class Lattice
@@ -37,11 +29,10 @@ public:
 
 #include <Grid_cshift.h>
     
-    // overloading Grid::conformable but no conformable in Grid ...?:w
     template<class obj1,class obj2>
     friend void conformable(const Lattice<obj1> &lhs,const Lattice<obj2> &rhs);
 
-    // Performance difference between operator * and mult is troubling.
+    // FIXME Performance difference between operator * and mult is troubling.
     // Auto move constructor seems to lose surprisingly much.
 
     // Site wise binary operations
@@ -182,23 +173,20 @@ public:
         }}
     };
 
+    // FIXME Implement a consistent seed management strategy
     friend void gaussian(Lattice<vobj> &l){
         // Zero mean, unit variance.
         std::normal_distribution<double> distribution(0.0,1.0);
         Real *v_ptr = (Real *)&l._odata[0];
         size_t v_len = l._grid->oSites()*sizeof(vobj);
         size_t d_len = v_len/sizeof(Real);
-        
-        // Not a parallel RNG. Could make up some seed per 4d site, seed
-        // per hypercube type scheme.
+
         for(int i=0;i<d_len;i++){
 	  v_ptr[i]= drand48();
         }
     };
 
-
     // Unary functions and Unops
-    // Unary negation
     friend inline Lattice<vobj> operator -(const Lattice<vobj> &r) {
         Lattice<vobj> ret(r._grid);
 #pragma omp parallel for
diff --git a/Grid_QCD.h b/Grid_QCD.h
index a0f7801a..3e852633 100644
--- a/Grid_QCD.h
+++ b/Grid_QCD.h
@@ -24,10 +24,7 @@ namespace QCD {
     typedef iSinglet<Real >             TReal;    // This is painful. Tensor singlet complex type.
 
 
-    typedef iSinglet<vIntegerF >         vTIntegerF;
-    typedef iSinglet<vIntegerD >         vTIntegerD;
-    typedef iSinglet<vIntegerC >         vTIntegerC;
-    typedef iSinglet<vIntegerZ >         vTIntegerZ;
+    typedef iSinglet<vInteger >         vTInteger;
 
     typedef iSpinMatrix<Complex >       SpinMatrix;
     typedef iColourMatrix<Complex >     ColourMatrix;
@@ -46,12 +43,9 @@ namespace QCD {
     typedef iColourVector<vComplex >     vColourVector;
     typedef iSpinColourVector<vComplex > vSpinColourVector;
     
-    typedef Lattice<vTComplex>         LatticeComplex;
+    typedef Lattice<vTComplex>            LatticeComplex;
 
-    typedef Lattice<vTIntegerF>            LatticeIntegerF; // Predicates for "where"
-    typedef Lattice<vTIntegerD>            LatticeIntegerD; 
-    typedef Lattice<vTIntegerC>            LatticeIntegerC;
-    typedef Lattice<vTIntegerZ>            LatticeIntegerZ;
+    typedef Lattice<vTInteger>            LatticeInteger; // Predicates for "where"
     
     typedef Lattice<vColourMatrix>     LatticeColourMatrix;
     typedef Lattice<vSpinMatrix>       LatticeSpinMatrix;
diff --git a/Grid_aligned_allocator.h b/Grid_aligned_allocator.h
index dbaa0ba3..9008105a 100644
--- a/Grid_aligned_allocator.h
+++ b/Grid_aligned_allocator.h
@@ -1,5 +1,8 @@
 #ifndef GRID_ALIGNED_ALLOCATOR_H
 #define GRID_ALIGNED_ALLOCATOR_H
+
+#include <immintrin.h>
+
 namespace Grid {
 
 ////////////////////////////////////////////////////////////////////
diff --git a/Grid_fake.cc b/Grid_communicator_fake.cc
similarity index 100%
rename from Grid_fake.cc
rename to Grid_communicator_fake.cc
diff --git a/Grid_mpi.cc b/Grid_communicator_mpi.cc
similarity index 100%
rename from Grid_mpi.cc
rename to Grid_communicator_mpi.cc
diff --git a/Grid_config.h b/Grid_config.h
index 8a0fdc26..bb885708 100644
--- a/Grid_config.h
+++ b/Grid_config.h
@@ -10,9 +10,6 @@
 /* AVX512 */
 /* #undef AVX512 */
 
-/* GRID_COMMS_FAKE */
-/* #undef GRID_COMMS_FAKE */
-
 /* GRID_COMMS_MPI */
 #define GRID_COMMS_MPI 1
 
diff --git a/Grid_config.h.in b/Grid_config.h.in
index ff15a834..91948fa2 100644
--- a/Grid_config.h.in
+++ b/Grid_config.h.in
@@ -9,9 +9,6 @@
 /* AVX512 */
 #undef AVX512
 
-/* GRID_COMMS_FAKE */
-#undef GRID_COMMS_FAKE
-
 /* GRID_COMMS_MPI */
 #undef GRID_COMMS_MPI
 
diff --git a/Grid_cshift_common.h b/Grid_cshift_common.h
index 72518d7d..08e75cff 100644
--- a/Grid_cshift_common.h
+++ b/Grid_cshift_common.h
@@ -1,17 +1,5 @@
 #ifndef _GRID_CSHIFT_COMMON_H_
 #define _GRID_CSHIFT_COMMON_H_
-//////////////////////////////////////////////////////////////////////////////////////////
-// Must not lose sight that goal is to be able to construct really efficient
-// gather to a point stencil code. CSHIFT is not the best way, so probably need
-// additional stencil support.
-//
-// Stencil based code could pre-exchange haloes and use a table lookup for neighbours
-//
-// Lattice <foo> could also allocate haloes which get used for stencil code.
-//
-// Grid could create a neighbour index table for a given stencil.
-// Could also implement CovariantCshift.
-//////////////////////////////////////////////////////////////////////////////////////////
 
 //////////////////////////////////////////////////////
 // Gather for when there is no need to SIMD split
@@ -57,7 +45,6 @@ friend void Gather_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllo
   }
 }
 
-
 //////////////////////////////////////////////////////
 // Gather for when there *is* need to SIMD split
 //////////////////////////////////////////////////////
@@ -101,8 +88,6 @@ friend void Gather_plane_extract(Lattice<vobj> &rhs,std::vector<scalar_type *> p
   }
 }
 
-
-
 //////////////////////////////////////////////////////
 // Scatter for when there is no need to SIMD split
 //////////////////////////////////////////////////////
@@ -146,7 +131,6 @@ friend void Scatter_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAll
   }
 }
 
-
 //////////////////////////////////////////////////////
 // Scatter for when there *is* need to SIMD split
 //////////////////////////////////////////////////////
@@ -190,11 +174,9 @@ friend void Scatter_plane_merge(Lattice<vobj> &rhs,std::vector<scalar_type *> po
   }
 }
 
-
 //////////////////////////////////////////////////////
 // local to node block strided copies
 //////////////////////////////////////////////////////
-// if lhs is odd, rhs even??
 friend void Copy_plane(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int dimension,int lplane,int rplane,int cbmask)
 {
   int rd = rhs._grid->_rdimensions[dimension];
@@ -284,40 +266,6 @@ friend void Copy_plane_permute(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int dimens
 // Local to node Cshift
 //////////////////////////////////////////////////////
 
-  // Work out whether to permute 
-  // ABCDEFGH ->   AE BF CG DH       permute              wrap num
-  //
-  // Shift 0       AE BF CG DH       0 0 0 0    ABCDEFGH   0   0
-  // Shift 1       BF CG DH AE       0 0 0 1    BCDEFGHA   0   1
-  // Shift 2       CG DH AE BF       0 0 1 1    CDEFGHAB   0   2
-  // Shift 3       DH AE BF CG       0 1 1 1    DEFGHABC   0   3
-  // Shift 4       AE BF CG DH       1 1 1 1    EFGHABCD   1   0 
-  // Shift 5       BF CG DH AE       1 1 1 0    FGHACBDE   1   1 
-  // Shift 6       CG DH AE BF       1 1 0 0    GHABCDEF   1   2
-  // Shift 7       DH AE BF CG       1 0 0 0    HABCDEFG   1   3
-
-  // Suppose 4way simd in one dim.
-  // ABCDEFGH ->   AECG BFDH      permute              wrap num
-
-  // Shift 0       AECG BFDH      0,00 0,00 ABCDEFGH         0     0
-  // Shift 1       BFDH CGEA      0,00 1,01 BCDEFGHA         0     1
-  // Shift 2       CGEA DHFB      1,01 1,01 CDEFGHAB         1     0
-  // Shift 3       DHFB EAGC      1,01 1,11 DEFGHABC         1     1
-  // Shift 4       EAGC FBHD      1,11 1,11 EFGHABCD         2     0 
-  // Shift 5       FBHD GCAE      1,11 1,10 FGHABCDE         2     1
-  // Shift 6       GCAE HDBF      1,10 1,10 GHABCDEF         3     0
-  // Shift 7       HDBF AECG      1,10 0,00 HABCDEFG         3     1
-
-  // Generalisation to 8 way simd, 16 way simd required.
-  //
-  // Need log2 Nway masks. consisting of 
-  //	    1 bit  256 bit granule
-  //	    2 bit  128 bit granule
-  //        4 bits 64  bit granule
-  //        8 bits 32  bit granules
-  //
-  //        15 bits....
-
 friend void Cshift_local(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimension,int shift)
 {
   int sshift[2];
@@ -333,35 +281,31 @@ friend void Cshift_local(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimension,int
   }
 }
 
-
 friend Lattice<vobj> Cshift_local(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int shift,int cbmask)
 {
-  int fd = rhs._grid->_fdimensions[dimension];
-  int rd = rhs._grid->_rdimensions[dimension];
-  int ld = rhs._grid->_ldimensions[dimension];
-  int gd = rhs._grid->_gdimensions[dimension];
-  
+  GridBase *grid = rhs._grid;
+  int fd = grid->_fdimensions[dimension];
+  int rd = grid->_rdimensions[dimension];
+  int ld = grid->_ldimensions[dimension];
+  int gd = grid->_gdimensions[dimension];
 
   // Map to always positive shift modulo global full dimension.
   shift = (shift+fd)%fd;
 
-  ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
+  ret.checkerboard = grid->CheckerBoardDestination(rhs.checkerboard,shift);
         
   // the permute type
-  int permute_dim =rhs._grid->_simd_layout[dimension]>1 ;
-  int permute_type=0;
-  for(int d=0;d<dimension;d++){
-    if (rhs._grid->_simd_layout[d]>1 ) permute_type++;
-  }
+  int permute_dim =grid->PermuteDim(dimension);
+  int permute_type=grid->PermuteType(dimension);
 
   for(int x=0;x<rd;x++){       
 
     int o   = 0;
-    int bo  = x * rhs._grid->_ostride[dimension];
+    int bo  = x * grid->_ostride[dimension];
     
     int cb= (cbmask==0x2)? 1 : 0;
 
-    int sshift = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,cb);
+    int sshift = grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,cb);
     int sx     = (x+sshift)%rd;
 	
     int permute_slice=0;
diff --git a/Grid_cshift_mpi.h b/Grid_cshift_mpi.h
index 75a4dc96..4ca8f6b5 100644
--- a/Grid_cshift_mpi.h
+++ b/Grid_cshift_mpi.h
@@ -146,10 +146,7 @@ friend void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimensi
   assert(shift>=0);
   assert(shift<fd);
 
-  int permute_type=0;
-  for(int d=0;d<dimension;d++){
-    if (grid->_simd_layout[d]>1 ) permute_type++;
-  }
+  int permute_type=grid->PermuteType(dimension);
 
   ///////////////////////////////////////////////
   // Simd direction uses an extract/merge pair
@@ -236,9 +233,12 @@ friend void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimensi
       if ( x< rd-num ) permute_slice=wrap;
       else permute_slice = 1-wrap;
 
+      int toggle_bit = (Nsimd>>(permute_type+1));
+      int PermuteMap;
       for(int i=0;i<Nsimd;i++){
 	if ( permute_slice ) {
-	  pointers[i] = rpointers[permute_map[permute_type][i]];
+	  PermuteMap=i^toggle_bit;
+	  pointers[i] = rpointers[PermuteMap];
 	} else {
 	  pointers[i] = rpointers[i];
 	}
@@ -260,8 +260,4 @@ friend void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimensi
     }
   }
 }
-
-
-
-
 #endif
diff --git a/Grid_main.cc b/Grid_main.cc
index cb45fd85..62356436 100644
--- a/Grid_main.cc
+++ b/Grid_main.cc
@@ -20,8 +20,8 @@ int main (int argc, char ** argv)
   
   std::vector<int> mpi_layout(4);
   mpi_layout[0]=2;
-  mpi_layout[1]=1;
-  mpi_layout[2]=1;
+  mpi_layout[1]=2;
+  mpi_layout[2]=2;
   mpi_layout[3]=2;
 
 #ifdef AVX512
diff --git a/Grid_simd.h b/Grid_simd.h
index ffcbff15..53cc8909 100644
--- a/Grid_simd.h
+++ b/Grid_simd.h
@@ -10,32 +10,6 @@
 //
 // Vector types are arch dependent
 ////////////////////////////////////////////////////////////////////////
-    // TODO
-    //
-    // Base class to share common code between vRealF, VComplexF etc...
-    //
-    // lattice Broad cast assignment
-    //
-    // where() support
-    // implement with masks, and/or? Type of the mask & boolean support?
-    //
-    // Unary functions
-    // cos,sin, tan, acos, asin, cosh, acosh, tanh, sinh, // Scalar<vReal> only arg
-    // exp, log, sqrt, fabs
-    //
-    // transposeColor, transposeSpin,
-    // adjColor, adjSpin,
-    // traceColor, traceSpin.
-    // peekColor, peekSpin + pokeColor PokeSpin
-    //
-    // copyMask.
-    //
-    // localMaxAbs
-    //
-    // norm2,
-    // sumMulti equivalent.
-    // Fourier transform equivalent.
-    //
     
   ////////////////////////////////////////////////////////////
   // SIMD Alignment controls
@@ -71,9 +45,6 @@ namespace Grid {
   typedef std::complex<RealD> ComplexD;
   typedef std::complex<Real>  Complex;
 
-
-
-
   inline RealF adj(const RealF  & r){ return r; }
   inline RealF conj(const RealF  & r){ return r; }
   inline ComplexD localInnerProduct(const ComplexD & l, const ComplexD & r) { return conj(l)*r; }
@@ -122,7 +93,6 @@ namespace Grid {
   template<>            inline void ZeroIt(RealD &arg){ arg=0; };
 
 
-
 #if defined (SSE2)
     typedef __m128 fvec;
     typedef __m128d dvec;
@@ -162,31 +132,46 @@ namespace Grid {
     inline void v_prefetch0(int size, const char *ptr){};
 #endif
 
-};
-
 
 /////////////////////////////////////////////////////////////////
 // Generic extract/merge/permute
 /////////////////////////////////////////////////////////////////
-template<class vsimd,class scalar,int Nsimd>
+template<class vsimd,class scalar>
 inline void Gextract(vsimd &y,std::vector<scalar *> &extracted){
-  // Bounce off stack is painful
-  // temporary hack while I figure out the right interface
-  scalar buf[Nsimd]; 
-  vstore(y,buf);
-  for(int i=0;i<Nsimd;i++){
-    *extracted[i] = buf[i];
-    extracted[i]++;
+#if 1
+  // FIXME: bounce off stack is painful
+  // temporary hack while I figure out better way.
+  // There are intrinsics to do this work without the storage.
+  int Nsimd = extracted.size();
+  {
+    std::vector<scalar,alignedAllocator<scalar> > buf(Nsimd); 
+    vstore(y,&buf[0]);
+    for(int i=0;i<Nsimd;i++){
+      *extracted[i] = buf[i];
+      extracted[i]++;
+    }
   }
+#else 
+  int NSo   = extracted.size();
+  int NSv   = vsimd::Nsimd();
+  int sparse= NSv/NSo;
+  for(int i=0;i<NSv;i+=sparse){
+    
+  }
+#endif
 };
-template<class vsimd,class scalar,int Nsimd>
+template<class vsimd,class scalar>
 inline void Gmerge(vsimd &y,std::vector<scalar *> &extracted){
-  scalar buf[Nsimd]; 
+#if 1
+  int Nsimd = extracted.size();
+  std::vector<scalar> buf(Nsimd); 
   for(int i=0;i<Nsimd;i++){
     buf[i]=*extracted[i];
     extracted[i]++;
   }
-  vset(y,buf); 
+  vset(y,&buf[0]); 
+#else
+#endif
 };
 
 //////////////////////////////////////////////////////////
@@ -197,8 +182,6 @@ inline void Gmerge(vsimd &y,std::vector<scalar *> &extracted){
 // Permute 3 possible on longer iVector lengths (512bit = 8 double = 16 single)
 // Permute 4 possible on half precision @512bit vectors.
 //////////////////////////////////////////////////////////
-// Should be able to make the permute/extract/merge independent of the
-// vector subtype and reduce the volume of code.
 template<class vsimd>
 inline void Gpermute(vsimd &y,vsimd b,int perm){
       switch (perm){
@@ -229,6 +212,7 @@ inline void Gpermute(vsimd &y,vsimd b,int perm){
       default: assert(0); break;
       }
     };
+};
 
 #include <Grid_vRealF.h>
 #include <Grid_vRealD.h>
diff --git a/Grid_stencil.h b/Grid_stencil.h
new file mode 100644
index 00000000..d93204fc
--- /dev/null
+++ b/Grid_stencil.h
@@ -0,0 +1,12 @@
+//////////////////////////////////////////////////////////////////////////////////////////
+// Must not lose sight that goal is to be able to construct really efficient
+// gather to a point stencil code. CSHIFT is not the best way, so probably need
+// additional stencil support.
+//
+// Stencil based code could pre-exchange haloes and use a table lookup for neighbours
+//
+// Lattice <foo> could also allocate haloes which get used for stencil code.
+//
+// Grid could create a neighbour index table for a given stencil.
+// Could also implement CovariantCshift.
+//////////////////////////////////////////////////////////////////////////////////////////
diff --git a/Grid_vComplexD.h b/Grid_vComplexD.h
index d1caefb1..330de26e 100644
--- a/Grid_vComplexD.h
+++ b/Grid_vComplexD.h
@@ -5,7 +5,7 @@
 
 namespace Grid {
     class vComplexD {
-    protected:
+    public:
         zvec v;
     public:
 	typedef zvec     vector_type;
@@ -154,64 +154,27 @@ namespace Grid {
             return ret;
         };
 
-        /////////////////////////////////////////////////////////////////
-        // Extract
-        /////////////////////////////////////////////////////////////////
-        friend inline void extract(vComplexD &y,std::vector<ComplexD *> &extracted){
-	  // Bounce off stack is painful
-	  // temporary hack while I figure out the right interface
-	  const int Nsimd = vComplexD::Nsimd();
-	  std::vector<ComplexD> buf(Nsimd); 
+	////////////////////////////////////////////////////////////////////
+	// General permute; assumes vector length is same across 
+	// all subtypes; may not be a good assumption, but could
+	// add the vector width as a template param for BG/Q for example
+	////////////////////////////////////////////////////////////////////
+	friend inline void permute(vComplexD &y,vComplexD b,int perm)
+	{
+	  Gpermute<vComplexD>(y,b,perm);
+	}
+	friend inline void merge(vComplexD &y,std::vector<ComplexD *> &extracted)
+	{
+	  Gmerge<vComplexD,ComplexD >(y,extracted);
+	}
+	friend inline void extract(vComplexD &y,std::vector<ComplexD *> &extracted)
+	{
+	  Gextract<vComplexD,ComplexD>(y,extracted);
+	}
 
-	  vstore(y,&buf[0]);
-
-	  for(int i=0;i<Nsimd;i++){
-	    *extracted[i] = buf[i];
-	    extracted[i]++;
-	  }
-        };
-
-        friend inline void merge(vComplexD &y,std::vector<ComplexD *> &extracted){
-	  // Bounce off stack is painful
-	  // temporary hack while I figure out the right interface
-	  const int Nsimd = vComplexD::Nsimd();
-	  std::vector<ComplexD> buf(Nsimd); 
-
-	  for(int i=0;i<Nsimd;i++){
-	    buf[i]=*extracted[i];
-	    extracted[i]++;
-	  }
-	  vset(y,&buf[0]); 
-        };
-        
-        
-        /////////////////////////////////////////////////////////////////
-        // Permute
-        /////////////////////////////////////////////////////////////////
-        friend inline void permute(vComplexD &y,vComplexD b,int perm){
-            switch (perm){
-                    // 2 complex=>1 permute
-#if defined(AVX1)||defined(AVX2)
-                case 0: y.v = _mm256_permute2f128_pd(b.v,b.v,0x01); break;
-                // AB => BA i.e. ab cd =>cd ab
-#endif
-#ifdef SSE2
-		  break;
-#endif
-#ifdef AVX512
-                    // 4 complex=>2 permute
-                // ABCD => BADC i.e. abcd efgh => cdab ghef
-                // ABCD => CDAB i.e. abcd efgh => efgh abcd
-                case 0: y.v = _mm512_swizzle_pd(b.v,_MM_SWIZ_REG_BADC); break;
-                case 1: y.v = _mm512_permute4f128_ps(b.v,(_MM_PERM_ENUM)_MM_SHUFFLE(1,0,3,2)); // permute for double is not implemented 
-
-#endif
-#ifdef QPX
-#error // Not implemented yet
-#endif
- 	        default: assert(0); break;
-            }
-        };
+	////////////////////////////////////////////////////////////////////////
+	// FIXME:  gonna remove these load/store, get, set, prefetch
+	////////////////////////////////////////////////////////////////////////
         void vload(zvec& a){
           this->v = a;
         }
@@ -296,7 +259,7 @@ friend inline void vstore(vComplexD &ret, ComplexD *a){
 #endif
             return ret;
         }
-// REDUCE
+// REDUCE FIXME must be a cleaner implementation
        friend inline ComplexD Reduce(const vComplexD & in)
        { 
 #if defined (AVX1) || defined(AVX2)
diff --git a/Grid_vComplexF.h b/Grid_vComplexF.h
index 578228a3..b7fb3d6a 100644
--- a/Grid_vComplexF.h
+++ b/Grid_vComplexF.h
@@ -4,7 +4,9 @@
 
 namespace Grid {
     class vComplexF {
-    protected:
+      //    protected:
+
+    public:
         cvec v;
         
     public:
@@ -129,75 +131,11 @@ namespace Grid {
 #endif
             return ret;
         };
+      
 
-        /////////////////////////////////////////////////////////////////
-        // Extract
-        /////////////////////////////////////////////////////////////////
-        friend inline void extract(vComplexF &y,std::vector<ComplexF *> &extracted){
-	  // Bounce off heap is painful
-	  // temporary hack while I figure out the right interface
-            vComplexF vbuf;
-            ComplexF *buf = (ComplexF *)&vbuf;
-            
-	  vstore(y,&buf[0]);
-	  for(int i=0;i<vComplexF::Nsimd();i++){
-	    *extracted[i] = buf[i];
-	    extracted[i]++;
-	  }
-
-        };
-
-        friend inline void merge(vComplexF &y,std::vector<ComplexF *> &extracted){
-	  // Bounce off stack is painful
-	  // temporary hack while I figure out the right interface
-	  const int Nsimd = vComplexF::Nsimd();
-            vComplexF vbuf;
-            ComplexF *buf = (ComplexF *)&vbuf;
-
-	  for(int i=0;i<Nsimd;i++){
-	    buf[i]=*extracted[i];
-	    extracted[i]++;
-	  }
-	  vset(y,&buf[0]); 
-        };
-        
-
-
-        /////////////////////////////////////////////////////////////////
-        // Permute
-        /////////////////////////////////////////////////////////////////
-        friend inline void permute(vComplexF &y,vComplexF b,int perm){
-            switch (perm){
-#if defined(AVX1)||defined(AVX2)
-//HERE
-                    // 4 complex=>2 permutes
-                    // case 0 ABCD->BADC
-                    // case 1 ABCD->CDAB
-                case 0: y.v = _mm256_shuffle_ps(b.v,b.v,_MM_SHUFFLE(1,0,3,2)); break;
-                case 1: y.v = _mm256_permute2f128_ps(b.v,b.v,0x01); break;
-#endif
-#ifdef SSE2
-                case 0: y.v = _mm_shuffle_ps(b.v,b.v,_MM_SHUFFLE(1,0,3,2));break;
-#endif
-#ifdef AVX512
-//#error should permute for  512
-                    // 8 complex=>3 permutes
-                    // case 0 ABCD EFGH -> BADC FEHG
-                    // case 1 ABCD EFGH -> CDAB GHEF
-                    // case 2 ABCD EFGH -> EFGH ABCD
-                case 0: y.v = _mm512_swizzle_ps(b.v,_MM_SWIZ_REG_CDAB); break; // OK
-                case 1: y.v = _mm512_swizzle_ps(b.v,_MM_SWIZ_REG_BADC); break; // OK
-                case 2: y.v = _mm512_permute4f128_ps(b.v, (_MM_PERM_ENUM)_MM_SHUFFLE(2,3,0,1)); break; // OK
-
-#endif
-#ifdef QPX
-#error
-#endif
-	        default: assert(0); break;
-            }
-        };
-        
-
+	////////////////////////////////////////////////////////////////////////
+	// FIXME:  gonna remove these load/store, get, set, prefetch
+	////////////////////////////////////////////////////////////////////////
         friend inline void vset(vComplexF &ret, Complex *a){
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_set_ps(a[3].imag(),a[3].real(),a[2].imag(),a[2].real(),a[1].imag(),a[1].real(),a[0].imag(),a[0].real());
@@ -358,6 +296,20 @@ friend inline void vstore(vComplexF &ret, ComplexF *a){
             return *this;
         }
 
+      friend inline void permute(vComplexF &y,vComplexF b,int perm)
+      {
+	Gpermute<vComplexF>(y,b,perm);
+      }
+      friend inline void merge(vComplexF &y,std::vector<ComplexF *> &extracted)
+      {
+	Gmerge<vComplexF,ComplexF >(y,extracted);
+      }
+      friend inline void extract(vComplexF &y,std::vector<ComplexF *> &extracted)
+      {
+	Gextract<vComplexF,ComplexF>(y,extracted);
+      }
+
+
     };
 
     inline vComplexF localInnerProduct(const vComplexF & l, const vComplexF & r) { return conj(l)*r; }
@@ -371,7 +323,5 @@ friend inline void vstore(vComplexF &ret, ComplexF *a){
         return l*r;
     }
 
-
-
 }
 #endif
diff --git a/Grid_vInteger.h b/Grid_vInteger.h
index 7a0c1c4f..6ddce191 100644
--- a/Grid_vInteger.h
+++ b/Grid_vInteger.h
@@ -235,70 +235,11 @@ friend inline void vstore(vInteger &ret, Integer *a){
       }
       friend inline void merge(vIntegerF &y,std::vector<Integer *> &extracted)
       {
-	Gmerge<vIntegerF,Integer,sizeof(ivec)/sizeof(float) >(y,extracted);
+	Gmerge<vIntegerF,Integer>(y,extracted);
       }
       friend inline void extract(vIntegerF &y,std::vector<Integer *> &extracted)
       {
-	Gextract<vIntegerF,Integer,sizeof(ivec)/sizeof(float) >(y,extracted);
-      }
-    };
-
-
-    class vIntegerD : public vInteger
-    {
-    public:
-      static inline int Nsimd(void) { return sizeof(ivec)/sizeof(double);}
-      
-      friend inline void permute(vIntegerD &y,vIntegerD b,int perm)
-      {
-	Gpermute<vIntegerD>(y,b,perm);
-      }
-      friend inline void merge(vIntegerD &y,std::vector<Integer *> &extracted)
-      {
-	Gmerge<vIntegerD,Integer,sizeof(ivec)/sizeof(double) >(y,extracted);
-      }
-      friend inline void extract(vIntegerD &y,std::vector<Integer *> &extracted)
-      {
-	Gextract<vIntegerD,Integer,sizeof(ivec)/sizeof(double) >(y,extracted);
-      }
-    };
-
-
-    class vIntegerC : public vInteger
-    {
-    public:
-      static inline int Nsimd(void) { return sizeof(ivec)/sizeof(ComplexF);}
-      
-      friend inline void permute(vIntegerC &y,vIntegerC b,int perm)
-      {
-	Gpermute<vIntegerC>(y,b,perm);
-      }
-      friend inline void merge(vIntegerC &y,std::vector<Integer *> &extracted)
-      {
-	Gmerge<vIntegerC,Integer,sizeof(ivec)/sizeof(ComplexF) >(y,extracted);
-      }
-      friend inline void extract(vIntegerC &y,std::vector<Integer *> &extracted)
-      {
-	Gextract<vIntegerC,Integer,sizeof(ivec)/sizeof(ComplexF) >(y,extracted);
-      }
-    };
-
-    class vIntegerZ : public vInteger
-    {
-    public:
-      static inline int Nsimd(void) { return sizeof(ivec)/sizeof(ComplexD);}
-      
-      friend inline void permute(vIntegerZ &y,vIntegerZ b,int perm)
-      {
-	Gpermute<vIntegerZ>(y,b,perm);
-      }
-      friend inline void merge(vIntegerZ &y,std::vector<Integer *> &extracted)
-      {
-	Gmerge<vIntegerZ,Integer,sizeof(ivec)/sizeof(ComplexD) >(y,extracted);
-      }
-      friend inline void extract(vIntegerZ &y,std::vector<Integer *> &extracted)
-      {
-	Gextract<vIntegerZ,Integer,sizeof(ivec)/sizeof(ComplexD) >(y,extracted);
+	Gextract<vIntegerF,Integer>(y,extracted);
       }
     };
 
diff --git a/Grid_vRealD.h b/Grid_vRealD.h
index 1abc0804..13ceedbe 100644
--- a/Grid_vRealD.h
+++ b/Grid_vRealD.h
@@ -5,7 +5,7 @@
 
 namespace Grid {
     class vRealD  {
-    protected:
+    public:
         dvec v; // dvec is double precision vector
 
     public:
@@ -99,72 +99,27 @@ namespace Grid {
             return ret;
         };
 
-        /////////////////////////////////////////////////////////////////
-        // Extract
-        /////////////////////////////////////////////////////////////////
-        friend inline void extract(vRealD &y,std::vector<RealD *> &extracted){
-	  // Bounce off stack is painful
-	  // temporary hack while I figure out the right interface
-	  const int Nsimd = vRealD::Nsimd();
-	  RealD buf[Nsimd]; 
+	////////////////////////////////////////////////////////////////////
+	// General permute; assumes vector length is same across 
+	// all subtypes; may not be a good assumption, but could
+	// add the vector width as a template param for BG/Q for example
+	////////////////////////////////////////////////////////////////////
+	friend inline void permute(vRealD &y,vRealD b,int perm)
+	{
+	  Gpermute<vRealD>(y,b,perm);
+	}
+	friend inline void merge(vRealD &y,std::vector<RealD *> &extracted)
+	{
+	  Gmerge<vRealD,RealD >(y,extracted);
+	}
+	friend inline void extract(vRealD &y,std::vector<RealD *> &extracted)
+	{
+	  Gextract<vRealD,RealD>(y,extracted);
+	}
 
-	  vstore(y,buf);
-
-	  for(int i=0;i<Nsimd;i++){
-	    *extracted[i] = buf[i];
-	    extracted[i]++;
-	  }
-        };
-
-        friend inline void merge(vRealD &y,std::vector<RealD *> &extracted){
-	  // Bounce off stack is painful
-	  // temporary hack while I figure out the right interface
-	  const int Nsimd = vRealD::Nsimd();
-	  RealD buf[Nsimd]; 
-
-	  for(int i=0;i<Nsimd;i++){
-	    buf[i]=*extracted[i];
-	    extracted[i]++;
-	  }
-	  vset(y,buf); 
-        };
-
-        
-        // Permute plans
-        // Permute 0 every ABCDEFGH -> BA DC FE HG
-        // Permute 1 every ABCDEFGH -> CD AB GH EF
-        // Permute 2 every ABCDEFGH -> EFGH ABCD
-        // Permute 3 possible on longer iVector lengths (512bit = 8 double = 16 single)
-        // Permute 4 possible on half precision @512bit vectors.
-        friend inline void permute(vRealD &y,vRealD b,int perm){
-            switch (perm){
-                    // 4 doubles=>2 permutes
-#if defined(AVX1)||defined(AVX2)
-                case 0: y.v = _mm256_shuffle_pd(b.v,b.v,0x5); break;
-                case 1: y.v = _mm256_permute2f128_pd(b.v,b.v,0x01); break;
-#endif
-#ifdef SSE2
-                case 0: y.v = _mm_shuffle_pd(b.v,b.v,0x1); break;
-#endif
-#ifdef AVX512
-                    // 8 double => 3 permutes
-        // Permute 0 every abcd efgh -> badc fehg 
-        // Permute 1 every abcd efgh -> cdab ghef 
-        // Permute 2 every abcd efgh -> efgh abcd 
-        // NOTE: mm_512_permutex_pd not implemented
-        // NOTE: ignore warning
-                case 0: y.v = _mm512_swizzle_pd(b.v,_MM_SWIZ_REG_CDAB); break;
-                case 1: y.v = _mm512_swizzle_pd(b.v,_MM_SWIZ_REG_BADC); break;
-                case 2: y.v = _mm512_permute4f128_ps(b.v,(_MM_PERM_ENUM)_MM_SHUFFLE(1,0,3,2)); break;
-      
-#endif
-#ifdef QPX
-#error
-#endif
-	    default: assert(0);break;
-            }
-        };
-// gona be bye bye
+	////////////////////////////////////////////////////////////////////////
+	// FIXME:  gonna remove these load/store, get, set, prefetch
+	////////////////////////////////////////////////////////////////////////
         void vload(dvec& a){
           this->v = a;
         }
diff --git a/Grid_vRealF.h b/Grid_vRealF.h
index 22809b83..185f4da8 100644
--- a/Grid_vRealF.h
+++ b/Grid_vRealF.h
@@ -5,7 +5,7 @@
 
 namespace Grid {
     class vRealF  {
-    protected:
+    public:
         fvec v;
 
     public:
@@ -120,74 +120,25 @@ namespace Grid {
         friend inline void vzero(vRealF &ret){vsplat(ret,0.0);}
 
 
-        /////////////////////////////////////////////////////////////////
-        // Extract
-        /////////////////////////////////////////////////////////////////
-        friend inline void extract(vRealF &y,std::vector<RealF *> &extracted){
-	  // Bounce off stack is painful
-	  // temporary hack while I figure out the right interface
-	  const int Nsimd = vRealF::Nsimd();
-	  RealF buf[Nsimd]; 
+	////////////////////////////////////////////////////////////////////
+	// General permute; assumes vector length is same across 
+	// all subtypes; may not be a good assumption, but could
+	// add the vector width as a template param for BG/Q for example
+	////////////////////////////////////////////////////////////////////
+	friend inline void permute(vRealF &y,vRealF b,int perm)
+	{
+	  Gpermute<vRealF>(y,b,perm);
+	}
+	friend inline void merge(vRealF &y,std::vector<RealF *> &extracted)
+	{
+	  Gmerge<vRealF,RealF >(y,extracted);
+	}
+	friend inline void extract(vRealF &y,std::vector<RealF *> &extracted)
+	{
+	  Gextract<vRealF,RealF>(y,extracted);
+	}
 
-	  vstore(y,buf);
 
-	  for(int i=0;i<Nsimd;i++){
-	    *extracted[i] = buf[i];
-	    extracted[i]++;
-	  }
-        };
-
-        friend inline void merge(vRealF &y,std::vector<RealF *> &extracted){
-	  // Bounce off stack is painful
-	  // temporary hack while I figure out the right interface
-	  const int Nsimd = vRealF::Nsimd();
-	  RealF buf[Nsimd]; 
-
-	  for(int i=0;i<Nsimd;i++){
-	    buf[i]=*extracted[i];
-	    extracted[i]++;
-	  }
-	  vset(y,buf); 
-        };
-        
-        //////////////////////////////////////////////////////////
-        // Permute
-        // Permute 0 every ABCDEFGH -> BA DC FE HG
-        // Permute 1 every ABCDEFGH -> CD AB GH EF
-        // Permute 2 every ABCDEFGH -> EFGH ABCD
-        // Permute 3 possible on longer iVector lengths (512bit = 8 double = 16 single)
-        // Permute 4 possible on half precision @512bit vectors.
-        //////////////////////////////////////////////////////////
-        friend inline void permute(vRealF &y,vRealF b,int perm){
-            switch (perm){
-                    // 8 floats=>3 permutes
-#if defined(AVX1)||defined(AVX2)
-                case 0: y.v = _mm256_shuffle_ps(b.v,b.v,_MM_SHUFFLE(2,3,0,1)); break;
-                case 1: y.v = _mm256_shuffle_ps(b.v,b.v,_MM_SHUFFLE(1,0,3,2)); break;
-                case 2: y.v = _mm256_permute2f128_ps(b.v,b.v,0x01); break;
-#endif
-#ifdef SSE2
-                case 0: y.v = _mm_shuffle_ps(b.v,b.v,_MM_SHUFFLE(2,3,0,1)); break;
-                case 1: y.v = _mm_shuffle_ps(b.v,b.v,_MM_SHUFFLE(1,0,3,2));break;
-#endif
-#ifdef AVX512
-                    // 16 floats=> permutes
-        // Permute 0 every abcd efgh ijkl mnop -> badc fehg jilk nmpo 
-        // Permute 1 every abcd efgh ijkl mnop -> cdab ghef jkij opmn 
-        // Permute 2 every abcd efgh ijkl mnop -> efgh abcd mnop ijkl
-        // Permute 3 every abcd efgh ijkl mnop -> ijkl mnop abcd efgh
-//#error not implemented should do something
-                case 0: y.v = _mm512_swizzle_ps(b.v,_MM_SWIZ_REG_CDAB); break;
-                case 1: y.v = _mm512_swizzle_ps(b.v,_MM_SWIZ_REG_BADC); break;
-                case 2: y.v = _mm512_permute4f128_ps(b.v,(_MM_PERM_ENUM)_MM_SHUFFLE(2,3,0,1)); break;
-                case 3: y.v = _mm512_permute4f128_ps(b.v,(_MM_PERM_ENUM)_MM_SHUFFLE(1,0,3,2)); break;
-#endif
-#ifdef QPX
-#error not implemented
-#endif
-	    default: assert(0); break;
-            }
-        };
         
         /////////////////////////////////////////////////////
         // Broadcast a value across Nsimd copies.
@@ -207,6 +158,8 @@ namespace Grid {
             ret.v = {a,a,a,a};
 #endif
         }
+
+
         friend inline void vset(vRealF &ret, float *a){
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_set_ps(a[7],a[6],a[5],a[4],a[3],a[2],a[1],a[0]);
@@ -224,6 +177,9 @@ namespace Grid {
 #endif
 	}
 
+	////////////////////////////////////////////////////////////////////////
+	// FIXME:  gonna remove these load/store, get, set, prefetch
+	////////////////////////////////////////////////////////////////////////
 friend inline void vstore(vRealF &ret, float *a){
 #if defined (AVX1)|| defined (AVX2)
 	_mm256_store_ps(a,ret.v);
diff --git a/Makefile.am b/Makefile.am
index d3a7b329..d11f28d5 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -24,7 +24,6 @@ include_HEADERS = Grid_config.h\
 	Grid_aligned_allocator.h\
 	Grid_cshift.h\
 	Grid_cshift_common.h\
-	Grid_cshift_fake.h\
 	Grid_cshift_mpi.h\
 	Grid_cshift_none.h\
 	Grid_math_types.h
@@ -37,13 +36,10 @@ bin_PROGRAMS = Grid_main
 
 extra_sources=
 if BUILD_COMMS_MPI
-  extra_sources+=Grid_mpi.cc
-endif
-if BUILD_COMMS_FAKE
-  extra_sources+=Grid_fake.cc
+  extra_sources+=Grid_communicator_mpi.cc
 endif
 if BUILD_COMMS_NONE
-  extra_sources+=Grid_fake.cc
+  extra_sources+=Grid_communicator_fake.cc
 endif
 
 Grid_main_SOURCES = \
diff --git a/Makefile.in b/Makefile.in
index 6983b96a..16981e52 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -89,9 +89,8 @@ NORMAL_UNINSTALL = :
 PRE_UNINSTALL = :
 POST_UNINSTALL = :
 bin_PROGRAMS = Grid_main$(EXEEXT)
-@BUILD_COMMS_MPI_TRUE@am__append_1 = Grid_mpi.cc
-@BUILD_COMMS_FAKE_TRUE@am__append_2 = Grid_fake.cc
-@BUILD_COMMS_NONE_TRUE@am__append_3 = Grid_fake.cc
+@BUILD_COMMS_MPI_TRUE@am__append_1 = Grid_communicator_mpi.cc
+@BUILD_COMMS_NONE_TRUE@am__append_2 = Grid_communicator_fake.cc
 subdir = .
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/configure.ac
@@ -146,12 +145,13 @@ libGrid_a_LIBADD =
 am_libGrid_a_OBJECTS = Grid_init.$(OBJEXT)
 libGrid_a_OBJECTS = $(am_libGrid_a_OBJECTS)
 PROGRAMS = $(bin_PROGRAMS)
-am__Grid_main_SOURCES_DIST = Grid_main.cc Grid_mpi.cc Grid_fake.cc
-@BUILD_COMMS_MPI_TRUE@am__objects_1 = Grid_mpi.$(OBJEXT)
-@BUILD_COMMS_FAKE_TRUE@am__objects_2 = Grid_fake.$(OBJEXT)
-@BUILD_COMMS_NONE_TRUE@am__objects_3 = Grid_fake.$(OBJEXT)
-am__objects_4 = $(am__objects_1) $(am__objects_2) $(am__objects_3)
-am_Grid_main_OBJECTS = Grid_main.$(OBJEXT) $(am__objects_4)
+am__Grid_main_SOURCES_DIST = Grid_main.cc Grid_communicator_mpi.cc \
+	Grid_communicator_fake.cc
+@BUILD_COMMS_MPI_TRUE@am__objects_1 = Grid_communicator_mpi.$(OBJEXT)
+@BUILD_COMMS_NONE_TRUE@am__objects_2 =  \
+@BUILD_COMMS_NONE_TRUE@	Grid_communicator_fake.$(OBJEXT)
+am__objects_3 = $(am__objects_1) $(am__objects_2)
+am_Grid_main_OBJECTS = Grid_main.$(OBJEXT) $(am__objects_3)
 Grid_main_OBJECTS = $(am_Grid_main_OBJECTS)
 Grid_main_DEPENDENCIES = libGrid.a
 AM_V_P = $(am__v_P_@AM_V@)
@@ -214,8 +214,8 @@ CTAGS = ctags
 CSCOPE = cscope
 AM_RECURSIVE_TARGETS = cscope
 am__DIST_COMMON = $(srcdir)/Grid_config.h.in $(srcdir)/Makefile.in \
-	AUTHORS COPYING ChangeLog INSTALL NEWS README compile depcomp \
-	install-sh missing
+	AUTHORS COPYING ChangeLog INSTALL NEWS README TODO compile \
+	depcomp install-sh missing
 DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
 distdir = $(PACKAGE)-$(VERSION)
 top_distdir = $(distdir)
@@ -353,12 +353,11 @@ include_HEADERS = Grid_config.h\
 	Grid_aligned_allocator.h\
 	Grid_cshift.h\
 	Grid_cshift_common.h\
-	Grid_cshift_fake.h\
 	Grid_cshift_mpi.h\
 	Grid_cshift_none.h\
 	Grid_math_types.h
 
-extra_sources = $(am__append_1) $(am__append_2) $(am__append_3)
+extra_sources = $(am__append_1) $(am__append_2)
 Grid_main_SOURCES = \
 	Grid_main.cc\
 	$(extra_sources)
@@ -506,10 +505,10 @@ mostlyclean-compile:
 distclean-compile:
 	-rm -f *.tab.c
 
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_fake.Po@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_communicator_fake.Po@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_communicator_mpi.Po@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_init.Po@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_main.Po@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_mpi.Po@am__quote@
 
 .cc.o:
 @am__fastdepCXX_TRUE@	$(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
diff --git a/TODO b/TODO
index 9febd6c3..d2373ba1 100644
--- a/TODO
+++ b/TODO
@@ -1,4 +1,7 @@
 * FIXME audit
+* Remove vload/store etc..
+* Replace vset with a call to merge.
+* Replace vset with a call to merge.
 
 * Conditional execution Subset, where etc...
 * Coordinate information, integers etc...
@@ -27,3 +30,103 @@
   - BinaryWriter, TextWriter etc...
   - protocol buffers?
   - 
+// Cartesian grid inheritance
+//            Grid::GridBase
+//                     |
+//           __________|___________
+//          |                      |
+// Grid::GridCartesian   Grid::GridCartesianRedBlack
+//
+// TODO: document the following as an API guaranteed public interface
+
+    /* 
+     *       Rough map of functionality against QDP++ Layout
+     *
+     *       Param     |     Grid                     |     QDP++             
+     *       -----------------------------------------
+     *                 |                              |
+     *        void     |     oSites, iSites, lSites   |  sitesOnNode 
+     *        void     |     gSites                   |  vol
+     *                 |                              |
+     *        gcoor    |     oIndex, iIndex           |  linearSiteIndex // no virtual node in QDP
+     *        lcoor    |                              |
+     * 
+     *        void     |     CheckerBoarded           |  -        // No checkerboarded in QDP
+     *        void     |     FullDimensions           |  lattSize
+     *        void     |     GlobalDimensions         |  lattSize // No checkerboarded in QDP
+     *        void     |     LocalDimensions          |  subgridLattSize
+     *        void     |     VirtualLocalDimensions   |  subgridLattSize // no virtual node in QDP
+     *                 |                              |
+     *       int x 3   |     oiSiteRankToGlobal       |  siteCoords
+     *                 |     ProcessorCoorLocalCoorToGlobalCoor | 
+     *                 |                              |
+     *     vector<int> |     GlobalCoorToRankIndex   |  nodeNumber(coord)
+     *     vector<int> |     GlobalCoorToProcessorCoorLocalCoor|  nodeCoord(coord)
+     *                 |                              |
+     *     void        |     Processors               |  logicalSize    // returns cart array shape
+     *     void        |     ThisRank        |  nodeNumber();  // returns this node rank
+     *     void        |     ThisProcessorCoor        |    // returns this node coor
+     *     void        |     isBoss(void)             |  primaryNode();
+     *                 |                              |
+     *                 |     RankFromProcessorCoor    |  getLogicalCoorFrom(node)
+     *                 |     ProcessorCoorFromRank    |  getNodeNumberFrom(logical_coord)
+     */
+  // Work out whether to permute 
+  // ABCDEFGH ->   AE BF CG DH       permute              wrap num
+  //
+  // Shift 0       AE BF CG DH       0 0 0 0    ABCDEFGH   0   0
+  // Shift 1       BF CG DH AE       0 0 0 1    BCDEFGHA   0   1
+  // Shift 2       CG DH AE BF       0 0 1 1    CDEFGHAB   0   2
+  // Shift 3       DH AE BF CG       0 1 1 1    DEFGHABC   0   3
+  // Shift 4       AE BF CG DH       1 1 1 1    EFGHABCD   1   0 
+  // Shift 5       BF CG DH AE       1 1 1 0    FGHACBDE   1   1 
+  // Shift 6       CG DH AE BF       1 1 0 0    GHABCDEF   1   2
+  // Shift 7       DH AE BF CG       1 0 0 0    HABCDEFG   1   3
+
+  // Suppose 4way simd in one dim.
+  // ABCDEFGH ->   AECG BFDH      permute              wrap num
+
+  // Shift 0       AECG BFDH      0,00 0,00 ABCDEFGH         0     0
+  // Shift 1       BFDH CGEA      0,00 1,01 BCDEFGHA         0     1
+  // Shift 2       CGEA DHFB      1,01 1,01 CDEFGHAB         1     0
+  // Shift 3       DHFB EAGC      1,01 1,11 DEFGHABC         1     1
+  // Shift 4       EAGC FBHD      1,11 1,11 EFGHABCD         2     0 
+  // Shift 5       FBHD GCAE      1,11 1,10 FGHABCDE         2     1
+  // Shift 6       GCAE HDBF      1,10 1,10 GHABCDEF         3     0
+  // Shift 7       HDBF AECG      1,10 0,00 HABCDEFG         3     1
+
+  // Generalisation to 8 way simd, 16 way simd required.
+  //
+  // Need log2 Nway masks. consisting of 
+  //	    1 bit  256 bit granule
+  //	    2 bit  128 bit granule
+  //        4 bits 64  bit granule
+  //        8 bits 32  bit granules
+  //
+  //        15 bits....
+    // TODO
+    //
+    // Base class to share common code between vRealF, VComplexF etc...
+    //
+    // lattice Broad cast assignment
+    //
+    // where() support
+    // implement with masks, and/or? Type of the mask & boolean support?
+    //
+    // Unary functions
+    // cos,sin, tan, acos, asin, cosh, acosh, tanh, sinh, // Scalar<vReal> only arg
+    // exp, log, sqrt, fabs
+    //
+    // transposeColor, transposeSpin,
+    // adjColor, adjSpin,
+    // traceColor, traceSpin.
+    // peekColor, peekSpin + pokeColor PokeSpin
+    //
+    // copyMask.
+    //
+    // localMaxAbs
+    //
+    // norm2,
+    // sumMulti equivalent.
+    // Fourier transform equivalent.
+    //
diff --git a/configure b/configure
index cb6bab86..c6f93abc 100755
--- a/configure
+++ b/configure
@@ -628,8 +628,6 @@ LTLIBOBJS
 LIBOBJS
 BUILD_COMMS_NONE_FALSE
 BUILD_COMMS_NONE_TRUE
-BUILD_COMMS_FAKE_FALSE
-BUILD_COMMS_FAKE_TRUE
 BUILD_COMMS_MPI_FALSE
 BUILD_COMMS_MPI_TRUE
 EGREP
@@ -1369,8 +1367,7 @@ Optional Features:
   --disable-openmp        do not use OpenMP
   --enable-simd=SSE|AVX|AVX2|AVX512
                           Select instructions
-  --enable-comms=none|fake|mpi
-                          Select communications
+  --enable-comms=none|mpi Select communications
 
 Some influential environment variables:
   CXX         C++ compiler command
@@ -5051,12 +5048,6 @@ fi
 
 
 case ${ac_COMMS} in
-     fake)
-       echo Configuring for FAKE communications
-
-$as_echo "#define GRID_COMMS_FAKE 1" >>confdefs.h
-
-     ;;
      none)
        echo Configuring for NO communications
 
@@ -5082,14 +5073,6 @@ else
   BUILD_COMMS_MPI_FALSE=
 fi
 
- if  test "X${ac_COMMS}X" == "XfakeX" ; then
-  BUILD_COMMS_FAKE_TRUE=
-  BUILD_COMMS_FAKE_FALSE='#'
-else
-  BUILD_COMMS_FAKE_TRUE='#'
-  BUILD_COMMS_FAKE_FALSE=
-fi
-
  if  test "X${ac_COMMS}X" == "XnoneX" ; then
   BUILD_COMMS_NONE_TRUE=
   BUILD_COMMS_NONE_FALSE='#'
@@ -5243,10 +5226,6 @@ if test -z "${BUILD_COMMS_MPI_TRUE}" && test -z "${BUILD_COMMS_MPI_FALSE}"; then
   as_fn_error $? "conditional \"BUILD_COMMS_MPI\" was never defined.
 Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
-if test -z "${BUILD_COMMS_FAKE_TRUE}" && test -z "${BUILD_COMMS_FAKE_FALSE}"; then
-  as_fn_error $? "conditional \"BUILD_COMMS_FAKE\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
-fi
 if test -z "${BUILD_COMMS_NONE_TRUE}" && test -z "${BUILD_COMMS_NONE_FALSE}"; then
   as_fn_error $? "conditional \"BUILD_COMMS_NONE\" was never defined.
 Usually this means the macro was only invoked conditionally." "$LINENO" 5
diff --git a/configure.ac b/configure.ac
index 22d0365e..bfe745e5 100644
--- a/configure.ac
+++ b/configure.ac
@@ -51,13 +51,9 @@ case ${ac_SIMD} in
 esac
 
 
-AC_ARG_ENABLE([comms],[AC_HELP_STRING([--enable-comms=none|fake|mpi],[Select communications])],[ac_COMMS=${enable_comms}],[ac_COMMS=none])
+AC_ARG_ENABLE([comms],[AC_HELP_STRING([--enable-comms=none|mpi],[Select communications])],[ac_COMMS=${enable_comms}],[ac_COMMS=none])
 
 case ${ac_COMMS} in
-     fake)
-       echo Configuring for FAKE communications
-       AC_DEFINE([GRID_COMMS_FAKE],[1],[GRID_COMMS_FAKE] )
-     ;;
      none)
        echo Configuring for NO communications
        AC_DEFINE([GRID_COMMS_NONE],[1],[GRID_COMMS_NONE] )
@@ -72,7 +68,6 @@ case ${ac_COMMS} in
 esac
 
 AM_CONDITIONAL(BUILD_COMMS_MPI,[ test "X${ac_COMMS}X" == "XmpiX" ])
-AM_CONDITIONAL(BUILD_COMMS_FAKE,[ test "X${ac_COMMS}X" == "XfakeX" ])
 AM_CONDITIONAL(BUILD_COMMS_NONE,[ test "X${ac_COMMS}X" == "XnoneX" ])
 
 

From ce6a3a8ed4ec5fcc1e5bd8af37866dee786b0f4f Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 6 Apr 2015 11:28:00 +0100
Subject: [PATCH 040/429] Some popular configure commands

---
 configure-commands | 4 ++++
 1 file changed, 4 insertions(+)
 create mode 100644 configure-commands

diff --git a/configure-commands b/configure-commands
new file mode 100644
index 00000000..d22e9fdd
--- /dev/null
+++ b/configure-commands
@@ -0,0 +1,4 @@
+CXX=clang++ ./configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx"  --enable-comms=mpi
+CXX=clang++ ./configure --enable-simd=AVX CXXFLAGS="-mavx -g -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx"  --enable-comms=mpi
+CXX=clang++ ./configure --enable-simd=AVX CXXFLAGS="-mavx -g  -std=c++11" LDFLAGS= LIBS=-lmpi  --enable-comms=fake
+CXX=clang++ ./configure --enable-simd=AVX CXXFLAGS="-mavx -g  -std=c++11" LDFLAGS= LIBS=-lmpi  --enable-comms=none

From 81d5eabf6c5b31bbbe3d13e8cbf0891a89da8a73 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 6 Apr 2015 11:29:55 +0100
Subject: [PATCH 041/429] Remove stub files

---
 .gitignore | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/.gitignore b/.gitignore
index 0c02bc65..ac77b76b 100644
--- a/.gitignore
+++ b/.gitignore
@@ -31,6 +31,10 @@ errs
 # http://www.gnu.org/software/automake
 
 Makefile.in
+Makefile
+config.log
+config.status
+.deps
 
 # http://www.gnu.org/software/autoconf
 

From 31f4f4f1e11075a170efa51cb6c58045cf88530b Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 9 Apr 2015 08:06:03 +0200
Subject: [PATCH 042/429] "where" and integer comparisons logic implemented for
 conditional assignment. LatticeCoordinate helper to get global (reduced)
 coordinate.

Some more work of similar type perhaps needed, but the bulk of the required
structure for masked array assignment is now in place.
---
 Grid.h                  |   2 +-
 Grid_Lattice.h          |  36 ++--
 Grid_QCD.h              |  27 ++-
 Grid_comparison.h       | 264 +++++++++++++++++++++++++
 Grid_cshift_common.h    |   1 -
 Grid_main.cc            |  73 ++++++-
 Grid_math_type_mapper.h |  83 ++++++++
 Grid_math_types.h       | 414 +++++++++++++++++++++++++++++-----------
 Grid_predicated.h       |  62 ++++++
 Grid_simd.h             |  90 +++++----
 Grid_stencil.h          | 352 +++++++++++++++++++++++++++++++++-
 Grid_vComplexD.h        |  23 ++-
 Grid_vComplexF.h        |  67 ++++++-
 Grid_vInteger.h         |  69 ++++---
 Grid_vRealD.h           |  15 +-
 Grid_vRealF.h           |  15 +-
 TODO                    |  67 +++++--
 17 files changed, 1420 insertions(+), 240 deletions(-)
 create mode 100644 Grid_comparison.h
 create mode 100644 Grid_math_type_mapper.h
 create mode 100644 Grid_predicated.h

diff --git a/Grid.h b/Grid.h
index 0a4024bf..caa4ed5e 100644
--- a/Grid.h
+++ b/Grid.h
@@ -41,12 +41,12 @@
 #include <malloc.h>
 #endif
 
-
 #include <Grid_aligned_allocator.h>
 #include <Grid_simd.h>
 #include <Grid_math_types.h>
 #include <Grid_Cartesian.h>
 #include <Grid_Lattice.h>
+#include <Grid_comparison.h>
 #include <Grid_QCD.h>
 
 namespace Grid {
diff --git a/Grid_Lattice.h b/Grid_Lattice.h
index ebf68560..01b95755 100644
--- a/Grid_Lattice.h
+++ b/Grid_Lattice.h
@@ -19,16 +19,14 @@ public:
     typedef typename vobj::vector_type vector_type;
 public:
 
-
     Lattice(GridBase *grid) : _grid(grid) {
         _odata.reserve(_grid->oSites());
         assert((((uint64_t)&_odata[0])&0xF) ==0);
         checkerboard=0;
     }
-    
 
 #include <Grid_cshift.h>
-    
+   
     template<class obj1,class obj2>
     friend void conformable(const Lattice<obj1> &lhs,const Lattice<obj2> &rhs);
 
@@ -156,23 +154,23 @@ public:
             v_ptr[i]=drand48();
         }
     };
-    
+
     // FIXME for debug; deprecate this
     friend void lex_sites(Lattice<vobj> &l){
-        Real *v_ptr = (Real *)&l._odata[0];
-	size_t o_len = l._grid->oSites();
-        size_t v_len = sizeof(vobj)/sizeof(vRealF);
-	size_t vec_len = vRealF::Nsimd();
+      Real *v_ptr = (Real *)&l._odata[0];
+      size_t o_len = l._grid->oSites();
+      size_t v_len = sizeof(vobj)/sizeof(vRealF);
+      size_t vec_len = vRealF::Nsimd();
 
-        for(int i=0;i<o_len;i++){
-        for(int j=0;j<v_len;j++){
-	  for(int vv=0;vv<vec_len;vv+=2){
-            v_ptr[i*v_len*vec_len+j*vec_len+vv  ]= i+vv*500;
-            v_ptr[i*v_len*vec_len+j*vec_len+vv+1]= i+vv*500;
+      for(int i=0;i<o_len;i++){
+	for(int j=0;j<v_len;j++){
+          for(int vv=0;vv<vec_len;vv+=2){
+	    v_ptr[i*v_len*vec_len+j*vec_len+vv  ]= i+vv*500;
+	    v_ptr[i*v_len*vec_len+j*vec_len+vv+1]= i+vv*500;
 	  }
-        }}
-    };
-
+	}}
+    }
+    
     // FIXME Implement a consistent seed management strategy
     friend void gaussian(Lattice<vobj> &l){
         // Zero mean, unit variance.
@@ -195,7 +193,7 @@ public:
         }
         return ret;
     }
-    // *=,+=,-= operators
+    // *=,+=,-= operators inherit behvour from correspond */+/- operation
     template<class T>
     inline Lattice<vobj> &operator *=(const T &r) {
         *this = (*this)*r;
@@ -351,7 +349,6 @@ public:
     inline auto operator * (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs*rhs._odata[0])>
     {
         Lattice<decltype(lhs*rhs._odata[0])> ret(rhs._grid);
-
 #pragma omp parallel for
         for(int ss=0;ss<rhs._grid->oSites(); ss++){
             ret._odata[ss]=lhs*rhs._odata[ss];
@@ -383,7 +380,7 @@ public:
     {
         Lattice<decltype(lhs._odata[0]*rhs)> ret(lhs._grid);
 #pragma omp parallel for
-        for(int ss=0;ss<rhs._grid->oSites(); ss++){
+        for(int ss=0;ss<lhs._grid->oSites(); ss++){
             ret._odata[ss]=lhs._odata[ss]*rhs;
         }
         return ret;
@@ -409,5 +406,6 @@ public:
         return ret;
     }
 
+
 }
 #endif
diff --git a/Grid_QCD.h b/Grid_QCD.h
index 3e852633..1d52f911 100644
--- a/Grid_QCD.h
+++ b/Grid_QCD.h
@@ -45,7 +45,7 @@ namespace QCD {
     
     typedef Lattice<vTComplex>            LatticeComplex;
 
-    typedef Lattice<vTInteger>            LatticeInteger; // Predicates for "where"
+    typedef Lattice<vInteger>            LatticeInteger; // Predicates for "where"
     
     typedef Lattice<vColourMatrix>     LatticeColourMatrix;
     typedef Lattice<vSpinMatrix>       LatticeSpinMatrix;
@@ -92,6 +92,31 @@ namespace QCD {
         }
         return ret;
      }
+
+    // FIXME for debug; deprecate this
+   inline void LatticeCoordinate(LatticeInteger &l,int mu){
+      GridBase *grid = l._grid;
+      int Nsimd = grid->iSites();
+      std::vector<int> gcoor;
+      std::vector<Integer> mergebuf(Nsimd);
+      std::vector<Integer *> mergeptr(Nsimd);
+      for(int o=0;o<grid->oSites();o++){
+	for(int i=0;i<grid->iSites();i++){
+	  //	  RankIndexToGlobalCoor(grid->ThisRank(),o,i,gcoor);
+	  grid->RankIndexToGlobalCoor(0,o,i,gcoor);
+	  mergebuf[i]=gcoor[mu];
+	  mergeptr[i]=&mergebuf[i];
+	}
+	merge(l._odata[o],mergeptr);
+      }
+    };
+
+#include <Grid_predicated.h>
+
+#if 0
+
+#endif
+
 }   //namespace QCD
 } // Grid
 #endif
diff --git a/Grid_comparison.h b/Grid_comparison.h
new file mode 100644
index 00000000..c890d342
--- /dev/null
+++ b/Grid_comparison.h
@@ -0,0 +1,264 @@
+#ifndef GRID_COMPARISON_H
+#define GRID_COMPARISON_H
+namespace Grid {
+
+    // Generic list of functors
+    template<class lobj,class robj> class veq {
+    public:
+      vInteger operator()(const lobj &lhs, const robj &rhs)
+	{ 
+	  return lhs == rhs;
+	}
+    };
+    template<class lobj,class robj> class vne {
+    public:
+      vInteger operator()(const lobj &lhs, const robj &rhs)
+	{ 
+	  return lhs != rhs;
+	}
+    };
+    template<class lobj,class robj> class vlt {
+    public:
+      vInteger operator()(const lobj &lhs, const robj &rhs)
+	{ 
+	  return lhs < rhs;
+	}
+    };
+    template<class lobj,class robj> class vle {
+    public:
+      vInteger operator()(const lobj &lhs, const robj &rhs)
+	{ 
+	  return lhs <= rhs;
+	}
+    };
+    template<class lobj,class robj> class vgt {
+    public:
+      vInteger operator()(const lobj &lhs, const robj &rhs)
+	{ 
+	  return lhs > rhs;
+	}
+    };
+    template<class lobj,class robj> class vge {
+    public:
+      vInteger operator()(const lobj &lhs, const robj &rhs)
+	{ 
+	  return lhs >= rhs;
+	}
+    };
+
+    // Generic list of functors
+    template<class lobj,class robj> class seq {
+    public:
+      Integer operator()(const lobj &lhs, const robj &rhs)
+	{ 
+	  return lhs == rhs;
+	}
+    };
+    template<class lobj,class robj> class sne {
+    public:
+      Integer operator()(const lobj &lhs, const robj &rhs)
+	{ 
+	  return lhs != rhs;
+	}
+    };
+    template<class lobj,class robj> class slt {
+    public:
+      Integer operator()(const lobj &lhs, const robj &rhs)
+	{ 
+	  return lhs < rhs;
+	}
+    };
+    template<class lobj,class robj> class sle {
+    public:
+      Integer operator()(const lobj &lhs, const robj &rhs)
+	{ 
+	  return lhs <= rhs;
+	}
+    };
+    template<class lobj,class robj> class sgt {
+    public:
+      Integer operator()(const lobj &lhs, const robj &rhs)
+	{ 
+	  return lhs > rhs;
+	}
+    };
+    template<class lobj,class robj> class sge {
+    public:
+      Integer operator()(const lobj &lhs, const robj &rhs)
+	{ 
+	  return lhs >= rhs;
+	}
+    };
+
+
+    //////////////////////////////////////////////////////////////////////////////////////////////////////
+    // Integer gets extra relational functions. Could also implement these for RealF, RealD etc..
+    //////////////////////////////////////////////////////////////////////////////////////////////////////
+    template<class sfunctor> 
+    inline vInteger Comparison(sfunctor sop,const vInteger & lhs, const vInteger & rhs)
+    {
+      std::vector<Integer> vlhs(vInteger::Nsimd());   // Use functors to reduce this to single implementation
+      std::vector<Integer> vrhs(vInteger::Nsimd());
+      vInteger ret;
+      extract(lhs,vlhs);
+      extract(rhs,vrhs);
+      for(int s=0;s<vInteger::Nsimd();s++){
+	vlhs[s] = sop(vlhs[s],vrhs[s]);
+      }
+      merge(ret,vlhs);
+      return ret;
+    }
+    inline vInteger operator < (const vInteger & lhs, const vInteger & rhs)
+    {
+      return Comparison(slt<Integer,Integer>(),lhs,rhs);
+    }
+    inline vInteger operator <= (const vInteger & lhs, const vInteger & rhs)
+    {
+      return Comparison(sle<Integer,Integer>(),lhs,rhs);
+    }
+    inline vInteger operator > (const vInteger & lhs, const vInteger & rhs)
+    {
+      return Comparison(sgt<Integer,Integer>(),lhs,rhs);
+    }
+    inline vInteger operator >= (const vInteger & lhs, const vInteger & rhs)
+    {
+      return Comparison(sge<Integer,Integer>(),lhs,rhs);
+    }
+    inline vInteger operator == (const vInteger & lhs, const vInteger & rhs)
+    {
+      return Comparison(seq<Integer,Integer>(),lhs,rhs);
+    }
+    inline vInteger operator != (const vInteger & lhs, const vInteger & rhs)
+    {
+      return Comparison(sne<Integer,Integer>(),lhs,rhs);
+    }
+
+    //////////////////////////////////////////////////////////////////////////
+    // relational operators
+    // 
+    // Support <,>,<=,>=,==,!=
+    //
+    //Query supporting bitwise &, |, ^, !
+    //Query supporting logical &&, ||, 
+    //////////////////////////////////////////////////////////////////////////
+    template<class vfunctor,class lobj,class robj> 
+    inline Lattice<vInteger> LLComparison(vfunctor op,const Lattice<lobj> &lhs,const Lattice<robj> &rhs)
+    {
+      Lattice<vInteger> ret(rhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<rhs._grid->oSites(); ss++){
+	  ret._odata[ss]=op(lhs._odata[ss],rhs._odata[ss]);
+        }
+        return ret;
+    }
+    template<class vfunctor,class lobj,class robj> 
+    inline Lattice<vInteger> LSComparison(vfunctor op,const Lattice<lobj> &lhs,const robj &rhs)
+    {
+      Lattice<vInteger> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites(); ss++){
+	  ret._odata[ss]=op(lhs._odata[ss],rhs);
+        }
+        return ret;
+    }
+    template<class vfunctor,class lobj,class robj> 
+    inline Lattice<vInteger> SLComparison(vfunctor op,const lobj &lhs,const Lattice<robj> &rhs)
+    {
+      Lattice<vInteger> ret(rhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<rhs._grid->oSites(); ss++){
+	  ret._odata[ss]=op(lhs._odata[ss],rhs);
+        }
+        return ret;
+    }
+
+    // Less than
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator < (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
+     return LLComparison(vlt<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator < (const Lattice<lobj> & lhs, const robj & rhs) {
+     return LSComparison(vlt<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator < (const lobj & lhs, const Lattice<robj> & rhs) {
+     return SLComparison(vlt<lobj,robj>(),lhs,rhs);
+   }
+
+   // Less than equal
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator <= (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
+     return LLComparison(vle<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator <= (const Lattice<lobj> & lhs, const robj & rhs) {
+     return LSComparison(vle<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator <= (const lobj & lhs, const Lattice<robj> & rhs) {
+     return SLComparison(vle<lobj,robj>(),lhs,rhs);
+   }
+
+   // Greater than 
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator > (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
+     return LLComparison(vgt<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator > (const Lattice<lobj> & lhs, const robj & rhs) {
+     return LSComparison(vgt<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator > (const lobj & lhs, const Lattice<robj> & rhs) {
+     return SLComparison(vgt<lobj,robj>(),lhs,rhs);
+   }
+
+
+   // Greater than equal
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator >= (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
+     return LLComparison(vge<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator >= (const Lattice<lobj> & lhs, const robj & rhs) {
+     return LSComparison(vge<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator >= (const lobj & lhs, const Lattice<robj> & rhs) {
+     return SLComparison(vge<lobj,robj>(),lhs,rhs);
+   }
+
+
+   // equal
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator == (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
+     return LLComparison(veq<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator == (const Lattice<lobj> & lhs, const robj & rhs) {
+     return LSComparison(veq<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator == (const lobj & lhs, const Lattice<robj> & rhs) {
+     return SLComparison(veq<lobj,robj>(),lhs,rhs);
+   }
+
+
+   // not equal
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator != (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
+     return LLComparison(vne<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator != (const Lattice<lobj> & lhs, const robj & rhs) {
+     return LSComparison(vne<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator != (const lobj & lhs, const Lattice<robj> & rhs) {
+     return SLComparison(vne<lobj,robj>(),lhs,rhs);
+   }
+
+
+}
+#endif
diff --git a/Grid_cshift_common.h b/Grid_cshift_common.h
index 08e75cff..2910151c 100644
--- a/Grid_cshift_common.h
+++ b/Grid_cshift_common.h
@@ -265,7 +265,6 @@ friend void Copy_plane_permute(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int dimens
 //////////////////////////////////////////////////////
 // Local to node Cshift
 //////////////////////////////////////////////////////
-
 friend void Cshift_local(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimension,int shift)
 {
   int sshift[2];
diff --git a/Grid_main.cc b/Grid_main.cc
index 62356436..1676a07b 100644
--- a/Grid_main.cc
+++ b/Grid_main.cc
@@ -19,10 +19,10 @@ int main (int argc, char ** argv)
   std::vector<int> simd_layout(4);
   
   std::vector<int> mpi_layout(4);
-  mpi_layout[0]=2;
-  mpi_layout[1]=2;
-  mpi_layout[2]=2;
-  mpi_layout[3]=2;
+  mpi_layout[0]=1;
+  mpi_layout[1]=1;
+  mpi_layout[2]=1;
+  mpi_layout[3]=1;
 
 #ifdef AVX512
  for(int omp=128;omp<236;omp+=16){
@@ -121,12 +121,34 @@ int main (int argc, char ** argv)
     // Non-lattice (const objects) * Lattice
     ColourMatrix cm;
     SpinColourMatrix scm;
-    
+    vSpinColourMatrix vscm;
+    Complex cplx(1.0);
+    Integer myint=1;
+    double mydouble=1.0;
+
+    //    vSpinColourMatrix vscm;
     scMat = cMat*scMat;
     scm = cm * scm;         // SpinColourMatrix  = ColourMatrix     * SpinColourMatrix
     scm = scm *cm;          // SpinColourMatrix  = SpinColourMartix * ColourMatrix
     scm = GammaFive * scm ; // SpinColourMatrix  = SpinMatrix       * SpinColourMatrix
     scm = scm* GammaFive  ; // SpinColourMatrix  = SpinColourMatrix * SpinMatrix
+
+    scm = scm*cplx;
+    vscm = vscm*cplx;
+    scMat = scMat*cplx;
+
+    scm = cplx*scm;
+    vscm = cplx*vscm;
+    scMat = cplx*scMat;
+    scm = myint*scm;
+    vscm = myint*vscm;
+    scMat = scMat*myint;
+    
+    scm = scm*mydouble;
+    vscm = vscm*mydouble;
+    scMat = scMat*mydouble;
+    scMat = mydouble*scMat;
+    cMat = mydouble*cMat;
     
     sMat = adj(sMat);       // LatticeSpinMatrix adjoint
     sMat = iGammaFive*sMat; // SpinMatrix * LatticeSpinMatrix
@@ -160,7 +182,35 @@ int main (int argc, char ** argv)
     */
     lex_sites(Foo);
 
+    Integer mm[4];
+    mm[0]=1;
+    mm[1]=Fine._rdimensions[0];
+    mm[2]=Fine._ldimensions[0]*Fine._ldimensions[1];
+    mm[3]=Fine._ldimensions[0]*Fine._ldimensions[1]*Fine._ldimensions[2];
 
+    LatticeInteger lex(&Fine);
+    lex=zero;
+    for(int d=0;d<4;d++){
+      LatticeInteger coor(&Fine);
+      LatticeCoordinate(coor,d);
+      lex = lex + coor*mm[d];
+    }
+    Bar = zero;
+    Bar = where(lex<10,Foo,Bar);
+    {
+      std::vector<int> coor(4);
+      for(coor[3]=0;coor[3]<latt_size[3]/mpi_layout[3];coor[3]++){
+      for(coor[2]=0;coor[2]<latt_size[2]/mpi_layout[2];coor[2]++){
+      for(coor[1]=0;coor[1]<latt_size[1]/mpi_layout[1];coor[1]++){
+      for(coor[0]=0;coor[0]<latt_size[0]/mpi_layout[0];coor[0]++){
+        ColourMatrix bar;
+        peekSite(bar,Bar,coor);
+        for(int r=0;r<3;r++){
+        for(int c=0;c<3;c++){
+	  cout<<"bar "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "<<bar._internal._internal[r][c]<<std::endl;
+	}}
+      }}}}
+    }    
     //setCheckerboard(ShiftedCheck,rFoo); 
     //setCheckerboard(ShiftedCheck,bFoo); 
 
@@ -228,11 +278,15 @@ int main (int argc, char ** argv)
     
     double nrm=0;
 
+
+    LatticeColourMatrix deriv(&Fine);
+    double half=0.5;
+    deriv = 0.5*Cshift(Foo,0,1) - 0.5*Cshift(Foo,0,-1);
+
+
     for(int dir=0;dir<4;dir++){
       for(int shift=0;shift<latt_size[dir];shift++){
 
-
-
 	pickCheckerboard(0,rFoo,Foo);    // Pick out red or black checkerboards
 	pickCheckerboard(1,bFoo,Foo);
     
@@ -254,6 +308,8 @@ int main (int argc, char ** argv)
 	for(coor[2]=0;coor[2]<latt_size[2]/mpi_layout[2];coor[2]++){
 	for(coor[1]=0;coor[1]<latt_size[1]/mpi_layout[1];coor[1]++){
 	for(coor[0]=0;coor[0]<latt_size[0]/mpi_layout[0];coor[0]++){
+
+	 
  
         std::complex<Grid::Real> diff;
                     
@@ -305,7 +361,8 @@ int main (int argc, char ** argv)
         double nn=Ttr._internal._internal;
         if ( nn > 0 )
             cout<<"Shift real trace fail "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<endl;
-        
+     
+
         for(int r=0;r<3;r++){
         for(int c=0;c<3;c++){
             diff =shifted1._internal._internal[r][c]-shifted2._internal._internal[r][c];
diff --git a/Grid_math_type_mapper.h b/Grid_math_type_mapper.h
new file mode 100644
index 00000000..856f2c6b
--- /dev/null
+++ b/Grid_math_type_mapper.h
@@ -0,0 +1,83 @@
+#ifndef GRID_MATH_TYPE_MAPPER_H
+#define GRID_MATH_TYPE_MAPPER_H
+namespace Grid {
+//////////////////////////////////////////////////////////////////////////////////
+// Want to recurse: GridTypeMapper<Matrix<vComplexD> >::scalar_type == ComplexD.
+//////////////////////////////////////////////////////////////////////////////////
+  
+  template <class T> class GridTypeMapper {
+  public:
+    typedef typename T::scalar_type scalar_type;
+    typedef typename T::vector_type vector_type;
+    typedef typename T::tensor_reduced tensor_reduced;
+  };
+
+//////////////////////////////////////////////////////////////////////////////////
+// Recursion stops with these template specialisations
+//////////////////////////////////////////////////////////////////////////////////
+  template<> class GridTypeMapper<RealF> {
+  public:
+    typedef RealF scalar_type;
+    typedef RealF vector_type;
+    typedef RealF tensor_reduced ;
+  };
+  template<> class GridTypeMapper<RealD> {
+  public:
+    typedef RealD scalar_type;
+    typedef RealD vector_type;
+    typedef RealD tensor_reduced;
+  };
+  template<> class GridTypeMapper<ComplexF> {
+  public:
+    typedef ComplexF scalar_type;
+    typedef ComplexF vector_type;
+    typedef ComplexF tensor_reduced;
+  };
+  template<> class GridTypeMapper<ComplexD> {
+  public:
+    typedef ComplexD scalar_type;
+    typedef ComplexD vector_type;
+    typedef ComplexD tensor_reduced;
+  };
+
+  template<> class GridTypeMapper<vRealF> {
+  public:
+    typedef RealF  scalar_type;
+    typedef vRealF vector_type;
+    typedef vRealF tensor_reduced;
+  };
+  template<> class GridTypeMapper<vRealD> {
+  public:
+    typedef RealD  scalar_type;
+    typedef vRealD vector_type;
+    typedef vRealD tensor_reduced;
+  };
+  template<> class GridTypeMapper<vComplexF> {
+  public:
+    typedef ComplexF  scalar_type;
+    typedef vComplexF vector_type;
+    typedef vComplexF tensor_reduced;
+  };
+  template<> class GridTypeMapper<vComplexD> {
+  public:
+    typedef ComplexD  scalar_type;
+    typedef vComplexD vector_type;
+    typedef vComplexD tensor_reduced;
+  };
+  template<> class GridTypeMapper<vInteger> {
+  public:
+    typedef Integer  scalar_type;
+    typedef vInteger vector_type;
+    typedef vInteger tensor_reduced;
+  };
+
+  // Again terminate the recursion.
+  inline vRealD    TensorRemove(vRealD    arg){ return arg;}
+  inline vRealF    TensorRemove(vRealF    arg){ return arg;}
+  inline vComplexF TensorRemove(vComplexF arg){ return arg;}
+  inline vComplexD TensorRemove(vComplexD arg){ return arg;}
+  inline vInteger  TensorRemove(vInteger  arg){ return arg;}
+
+}
+
+#endif
diff --git a/Grid_math_types.h b/Grid_math_types.h
index 25524966..a2a3f97c 100644
--- a/Grid_math_types.h
+++ b/Grid_math_types.h
@@ -1,63 +1,11 @@
 #ifndef GRID_MATH_TYPES_H
 #define GRID_MATH_TYPES_H
+
+#include <Grid_math_type_mapper.h>
+
 namespace Grid {
 
 
-
-//////////////////////////////////////////////////////////////////////////////////
-// Want to recurse: GridTypeMapper<Matrix<vComplexD> >::scalar_type == ComplexD.
-//////////////////////////////////////////////////////////////////////////////////
-
-  template <class T> class GridTypeMapper {
-  public:
-    typedef typename T::scalar_type scalar_type;
-    typedef typename T::vector_type vector_type;
-  };
-
-  template<> class GridTypeMapper<RealF> {
-  public:
-    typedef RealF scalar_type;
-    typedef RealF vector_type;
-  };
-  template<> class GridTypeMapper<RealD> {
-  public:
-    typedef RealD scalar_type;
-    typedef RealD vector_type;
-  };
-  template<> class GridTypeMapper<ComplexF> {
-  public:
-    typedef ComplexF scalar_type;
-    typedef ComplexF vector_type;
-  };
-  template<> class GridTypeMapper<ComplexD> {
-  public:
-    typedef ComplexD scalar_type;
-    typedef ComplexD vector_type;
-  };
-
-  template<> class GridTypeMapper<vRealF> {
-  public:
-    typedef RealF  scalar_type;
-    typedef vRealF vector_type;
-  };
-  template<> class GridTypeMapper<vRealD> {
-  public:
-    typedef RealD  scalar_type;
-    typedef vRealD vector_type;
-  };
-  template<> class GridTypeMapper<vComplexF> {
-  public:
-    typedef ComplexF  scalar_type;
-    typedef vComplexF vector_type;
-  };
-  template<> class GridTypeMapper<vComplexD> {
-  public:
-    typedef ComplexD  scalar_type;
-    typedef vComplexD vector_type;
-  };
-
-
-
 ///////////////////////////////////////////////////
 // Scalar, Vector, Matrix objects.
 // These can be composed to form tensor products of internal indices.
@@ -70,9 +18,16 @@ public:
 
   typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
   typedef typename GridTypeMapper<vtype>::vector_type vector_type;
+  typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
+  typedef iScalar<tensor_reduced_v> tensor_reduced;
+
 
     iScalar(){};
+    
+    iScalar(scalar_type s) : _internal(s) {};// recurse down and hit the constructor for vector_type
+
     iScalar(Zero &z){ *this = zero; };
+
     iScalar<vtype> & operator= (const Zero &hero){
         zeroit(*this);
         return *this;
@@ -80,22 +35,27 @@ public:
     friend void zeroit(iScalar<vtype> &that){
         zeroit(that._internal);
     }
-    friend void permute(iScalar<vtype> &out,iScalar<vtype> &in,int permutetype){
+    friend void permute(iScalar<vtype> &out,const iScalar<vtype> &in,int permutetype){
       permute(out._internal,in._internal,permutetype);
     }
-    friend void extract(iScalar<vtype> &in,std::vector<scalar_type *> &out){
+    friend void extract(const iScalar<vtype> &in,std::vector<scalar_type *> &out){
       extract(in._internal,out); // extract advances the pointers in out
     }
     friend void merge(iScalar<vtype> &in,std::vector<scalar_type *> &out){
       merge(in._internal,out); // extract advances the pointers in out
     }
+    friend inline iScalar<vtype>::vector_type TensorRemove(iScalar<vtype> arg)
+    {
+      return TensorRemove(arg._internal);
+    }
+
     // Unary negation
     friend inline iScalar<vtype> operator -(const iScalar<vtype> &r) {
         iScalar<vtype> ret;
         ret._internal= -r._internal;
         return ret;
     }
-    // *=,+=,-= operators
+    // *=,+=,-= operators inherit from corresponding "*,-,+" behaviour
     inline iScalar<vtype> &operator *=(const iScalar<vtype> &r) {
         *this = (*this)*r;
         return *this;
@@ -117,6 +77,9 @@ public:
 
   typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
   typedef typename GridTypeMapper<vtype>::vector_type vector_type;
+  typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
+  typedef iScalar<tensor_reduced_v> tensor_reduced;
+
 
     iVector(Zero &z){ *this = zero; };
     iVector() {};
@@ -129,12 +92,12 @@ public:
             zeroit(that._internal[i]);
         }
     }
-    friend void permute(iVector<vtype,N> &out,iVector<vtype,N> &in,int permutetype){
+    friend void permute(iVector<vtype,N> &out,const iVector<vtype,N> &in,int permutetype){
       for(int i=0;i<N;i++){
 	permute(out._internal[i],in._internal[i],permutetype);
       }
     }
-    friend void extract(iVector<vtype,N> &in,std::vector<scalar_type *> &out){
+    friend void extract(const iVector<vtype,N> &in,std::vector<scalar_type *> &out){
       for(int i=0;i<N;i++){
 	extract(in._internal[i],out);// extract advances pointers in out
       }
@@ -150,7 +113,7 @@ public:
         for(int i=0;i<N;i++) ret._internal[i]= -r._internal[i];
         return ret;
     }
-    // *=,+=,-= operators
+    // *=,+=,-= operators inherit from corresponding "*,-,+" behaviour
     inline iVector<vtype,N> &operator *=(const iScalar<vtype> &r) {
         *this = (*this)*r;
         return *this;
@@ -163,10 +126,8 @@ public:
         *this = (*this)+r;
         return *this;
     }
-
 };
     
-    
 template<class vtype,int N> class iMatrix
 {
 public:
@@ -174,62 +135,64 @@ public:
 
   typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
   typedef typename GridTypeMapper<vtype>::vector_type vector_type;
+  typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
+  typedef iScalar<tensor_reduced_v> tensor_reduced;
 
-    iMatrix(Zero &z){ *this = zero; };
-    iMatrix() {};
-    iMatrix<vtype,N> & operator= (Zero &hero){
-        zeroit(*this);
-        return *this;
-    }
-    friend void zeroit(iMatrix<vtype,N> &that){
-        for(int i=0;i<N;i++){
-        for(int j=0;j<N;j++){
-                zeroit(that._internal[i][j]);
-        }}
-    }
-    friend void permute(iMatrix<vtype,N> &out,iMatrix<vtype,N> &in,int permutetype){
-      for(int i=0;i<N;i++){
+  iMatrix(Zero &z){ *this = zero; };
+  iMatrix() {};
+  iMatrix<vtype,N> & operator= (Zero &hero){
+    zeroit(*this);
+    return *this;
+  }
+  friend void zeroit(iMatrix<vtype,N> &that){
+    for(int i=0;i<N;i++){
+      for(int j=0;j<N;j++){
+	zeroit(that._internal[i][j]);
+    }}
+  }
+  friend void permute(iMatrix<vtype,N> &out,const iMatrix<vtype,N> &in,int permutetype){
+    for(int i=0;i<N;i++){
       for(int j=0;j<N;j++){
 	permute(out._internal[i][j],in._internal[i][j],permutetype);
-      }}
-    }
-    friend void extract(iMatrix<vtype,N> &in,std::vector<scalar_type *> &out){
-      for(int i=0;i<N;i++){
+    }}
+  }
+  friend void extract(const iMatrix<vtype,N> &in,std::vector<scalar_type *> &out){
+    for(int i=0;i<N;i++){
       for(int j=0;j<N;j++){
 	extract(in._internal[i][j],out);// extract advances pointers in out
-      }}
-    }
-    friend void merge(iMatrix<vtype,N> &in,std::vector<scalar_type *> &out){
-      for(int i=0;i<N;i++){
+    }}
+  }
+  friend void merge(iMatrix<vtype,N> &in,std::vector<scalar_type *> &out){
+    for(int i=0;i<N;i++){
       for(int j=0;j<N;j++){
 	merge(in._internal[i][j],out);// extract advances pointers in out
-      }}
-    }
-    // Unary negation
-    friend inline iMatrix<vtype,N> operator -(const iMatrix<vtype,N> &r) {
-        iMatrix<vtype,N> ret;
-        for(int i=0;i<N;i++){
-        for(int j=0;j<N;j++){
-            ret._internal[i][j]= -r._internal[i][j];
-        }}
-        return ret;
-    }
-    // *=,+=,-= operators
-    template<class T>
-    inline iMatrix<vtype,N> &operator *=(const T &r) {
-        *this = (*this)*r;
-        return *this;
-    }
-    template<class T>
-    inline iMatrix<vtype,N> &operator -=(const T &r) {
-        *this = (*this)-r;
-        return *this;
-    }
-    template<class T>
-    inline iMatrix<vtype,N> &operator +=(const T &r) {
-        *this = (*this)+r;
-        return *this;
-    }
+    }}
+  }
+  // Unary negation
+  friend inline iMatrix<vtype,N> operator -(const iMatrix<vtype,N> &r) {
+    iMatrix<vtype,N> ret;
+    for(int i=0;i<N;i++){
+      for(int j=0;j<N;j++){
+	ret._internal[i][j]= -r._internal[i][j];
+    }}
+    return ret;
+  }
+  // *=,+=,-= operators inherit from corresponding "*,-,+" behaviour
+  template<class T>
+  inline iMatrix<vtype,N> &operator *=(const T &r) {
+    *this = (*this)*r;
+    return *this;
+  }
+  template<class T>
+  inline iMatrix<vtype,N> &operator -=(const T &r) {
+    *this = (*this)-r;
+    return *this;
+  }
+  template<class T>
+  inline iMatrix<vtype,N> &operator +=(const T &r) {
+    *this = (*this)+r;
+    return *this;
+  }
 
 };
 
@@ -642,7 +605,8 @@ iVector<rtype,N> operator * (const iVector<mtype,N>& lhs,const iScalar<vtype>& r
     // mat  x vec  = vec
     // vec  x scal = vec
     // scal x vec  = vec
-    
+    //
+    // We can special case scalar_type ??
 template<class l,class r>
 inline auto operator * (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(lhs._internal * rhs._internal)>
 {
@@ -715,6 +679,229 @@ auto operator * (const iVector<l,N>& lhs,const iScalar<r>& rhs) -> iVector<declt
     }
     return ret;
 }
+//////////////////////////////////////////////////////////////////////////////////////////
+// Must support native C++ types Integer, Complex, Real
+//////////////////////////////////////////////////////////////////////////////////////////
+
+// multiplication by fundamental scalar type
+template<class l,int N> inline iScalar<l> operator * (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
+{
+  typename iScalar<l>::tensor_reduced srhs(rhs);
+  return lhs*srhs;
+}
+template<class l,int N> inline iScalar<l> operator * (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
+{
+  typename iVector<l,N>::tensor_reduced srhs(rhs);
+  return lhs*srhs;
+}
+template<class l,int N> inline iVector<l,N> operator * (const typename iScalar<l>::scalar_type lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type &rhs) 
+{
+  typename iMatrix<l,N>::tensor_reduced srhs(rhs);
+  return lhs*srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator * (const typename iScalar<l>::scalar_type & lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
+
+////////////////////////////////////////////////////////////////////
+// Double support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l> inline iScalar<l> operator * (double lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l,int N> inline iVector<l,N> operator * (double lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator * (double lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
+
+////////////////////////////////////////////////////////////////////
+// Integer support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l> inline iScalar<l> operator * (Integer lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l,int N> inline iVector<l,N> operator * (Integer lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator * (Integer lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
+
+
+
+///////////////////////////////////////////////////////////////////////////////////////////////
+// addition by fundamental scalar type applies to matrix(down diag) and scalar
+///////////////////////////////////////////////////////////////////////////////////////////////
+template<class l,int N> inline iScalar<l> operator + (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
+{
+  typename iScalar<l>::tensor_reduced srhs(rhs);
+  return lhs+srhs;
+}
+template<class l,int N> inline iScalar<l> operator + (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
+{
+  typename iMatrix<l,N>::tensor_reduced srhs(rhs);
+  return lhs+srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator + (const typename iScalar<l>::scalar_type lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
+
+////////////////////////////////////////////////////////////////////
+// Double support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator + (const iScalar<l>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs+srhs;
+}
+template<class l> inline iScalar<l> operator + (double lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs+srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator + (double lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
+
+////////////////////////////////////////////////////////////////////
+// Integer support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator + (const iScalar<l>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs+srhs;
+}
+template<class l> inline iScalar<l> operator + (Integer lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs+srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator + (Integer lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
+
+
+///////////////////////////////////////////////////////////////////////////////////////////////
+// subtraction of fundamental scalar type applies to matrix(down diag) and scalar
+///////////////////////////////////////////////////////////////////////////////////////////////
+template<class l,int N> inline iScalar<l> operator - (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
+{
+  typename iScalar<l>::tensor_reduced srhs(rhs);
+  return lhs-srhs;
+}
+template<class l,int N> inline iScalar<l> operator - (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) 
+{
+  typename iScalar<l>::tensor_reduced slhs(lhs);
+  return slhs-rhs;
+}
+
+template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
+{
+  typename iScalar<l>::tensor_reduced srhs(rhs);
+  return lhs-srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator - (const typename iScalar<l>::scalar_type lhs,const iMatrix<l,N>& rhs) 
+{
+  typename iScalar<l>::tensor_reduced slhs(lhs);
+  return slhs-rhs;
+}
+
+////////////////////////////////////////////////////////////////////
+// Double support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator - (const iScalar<l>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs-srhs;
+}
+template<class l> inline iScalar<l> operator - (double lhs,const iScalar<l>& rhs) 
+{
+  typename iScalar<l>::scalar_type t(lhs);
+  typename iScalar<l>::tensor_reduced slhs(t);
+  return slhs-rhs;
+}
+
+template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs-srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator - (double lhs,const iMatrix<l,N>& rhs) 
+{
+  typename iScalar<l>::scalar_type t(lhs);
+  typename iScalar<l>::tensor_reduced slhs(t);
+  return slhs-rhs;
+}
+
+////////////////////////////////////////////////////////////////////
+// Integer support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator - (const iScalar<l>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs-srhs;
+}
+template<class l> inline iScalar<l> operator - (Integer lhs,const iScalar<l>& rhs) 
+{
+  typename iScalar<l>::scalar_type t(lhs);
+  typename iScalar<l>::tensor_reduced slhs(t);
+  return slhs-rhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs-srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator - (Integer lhs,const iMatrix<l,N>& rhs) 
+{
+  typename iScalar<l>::scalar_type t(lhs);
+  typename iScalar<l>::tensor_reduced slhs(t);
+  return slhs-rhs;
+}
+
+
+
+
     ///////////////////////////////////////////////////////////////////////////////////////
     // localInnerProduct Scalar x Scalar -> Scalar
     // localInnerProduct Vector x Vector -> Scalar
@@ -907,6 +1094,7 @@ inline auto trace(const iScalar<vtype> &arg) -> iScalar<decltype(trace(arg._inte
     ret._internal=trace(arg._internal);
     return ret;
 }
+
 };
     
 #endif
diff --git a/Grid_predicated.h b/Grid_predicated.h
new file mode 100644
index 00000000..c6f29204
--- /dev/null
+++ b/Grid_predicated.h
@@ -0,0 +1,62 @@
+#ifndef GRID_PREDICATED_H
+#define GRID_PREDICATED_H
+
+// Must implement the predicate gating the 
+// Must be able to reduce the predicate down to a single vInteger per site.
+// Must be able to require the type be iScalar x iScalar x ....
+//                              give a GetVtype method in iScalar
+//                              and blow away the tensor structures.
+//
+template<class vobj>
+inline void where(Lattice<vobj> &ret,const LatticeInteger &predicate,Lattice<vobj> &iftrue,Lattice<vobj> &iffalse)
+{
+  conformable(iftrue,iffalse);
+  conformable(iftrue,predicate);
+  conformable(iftrue,ret);
+
+  GridBase *grid=iftrue._grid;
+  typedef typename vobj::scalar_type scalar_type;
+  typedef typename vobj::vector_type vector_type;
+
+  const int Nsimd = grid->Nsimd();
+  const int words = sizeof(vobj)/sizeof(vector_type);
+
+  std::vector<Integer> mask(Nsimd);
+  std::vector<std::vector<scalar_type> > truevals (Nsimd,std::vector<scalar_type>(words) );
+  std::vector<std::vector<scalar_type> > falsevals(Nsimd,std::vector<scalar_type>(words) );
+  std::vector<scalar_type *> pointers(Nsimd);
+
+#pragma omp parallel for
+  for(int ss=0;ss<iftrue._grid->oSites(); ss++){
+
+    for(int s=0;s<Nsimd;s++) pointers[s] = & truevals[s][0];
+    extract(iftrue._odata[ss]   ,pointers);
+
+    for(int s=0;s<Nsimd;s++) pointers[s] = & falsevals[s][0];
+    extract(iffalse._odata[ss]  ,pointers);
+
+    extract(predicate._odata[ss],mask);
+
+    for(int s=0;s<Nsimd;s++){
+      if (mask[s]) pointers[s]=&truevals[s][0];
+      else         pointers[s]=&falsevals[s][0];
+    }
+
+    merge(ret._odata[ss],pointers);
+  }
+}
+
+template<class vobj>
+inline Lattice<vobj> where(const LatticeInteger &predicate,Lattice<vobj> &iftrue,Lattice<vobj> &iffalse)
+{
+  conformable(iftrue,iffalse);
+  conformable(iftrue,predicate);
+
+  Lattice<vobj> ret(iftrue._grid);
+
+  where(ret,predicate,iftrue,iffalse);
+
+  return ret;
+}
+
+#endif
diff --git a/Grid_simd.h b/Grid_simd.h
index 53cc8909..1229d328 100644
--- a/Grid_simd.h
+++ b/Grid_simd.h
@@ -11,28 +11,15 @@
 // Vector types are arch dependent
 ////////////////////////////////////////////////////////////////////////
     
-  ////////////////////////////////////////////////////////////
-  // SIMD Alignment controls
-  ////////////////////////////////////////////////////////////
-#ifdef HAVE_VAR_ATTRIBUTE_ALIGNED
-#define ALIGN_DIRECTIVE(A) __attribute__ ((aligned(A)))
-#else
-#define ALIGN_DIRECTIVE(A) __declspec(align(A))
-#endif
 
 #ifdef SSE2
 #include <pmmintrin.h>
-#define SIMDalign ALIGN_DIRECTIVE(16)
 #endif
-
 #if defined(AVX1) || defined (AVX2)
 #include <immintrin.h>
-#define SIMDalign ALIGN_DIRECTIVE(32)
 #endif
-
 #ifdef AVX512
 #include <immintrin.h>
-#define SIMDalign ALIGN_DIRECTIVE(64)
 #endif
 
 namespace Grid {
@@ -137,41 +124,66 @@ namespace Grid {
 // Generic extract/merge/permute
 /////////////////////////////////////////////////////////////////
 template<class vsimd,class scalar>
-inline void Gextract(vsimd &y,std::vector<scalar *> &extracted){
-#if 1
+inline void Gextract(const vsimd &y,std::vector<scalar *> &extracted){
   // FIXME: bounce off stack is painful
   // temporary hack while I figure out better way.
   // There are intrinsics to do this work without the storage.
-  int Nsimd = extracted.size();
-  {
-    std::vector<scalar,alignedAllocator<scalar> > buf(Nsimd); 
-    vstore(y,&buf[0]);
-    for(int i=0;i<Nsimd;i++){
-      *extracted[i] = buf[i];
-      extracted[i]++;
-    }
+  int Nextr=extracted.size();
+  int Nsimd=vsimd::Nsimd();
+  int s=Nsimd/Nextr;
+
+  std::vector<scalar,alignedAllocator<scalar> > buf(Nsimd); 
+  vstore(y,&buf[0]);
+  for(int i=0;i<Nextr;i++){
+    *extracted[i] = buf[i*s];
+    extracted[i]++;
   }
-#else 
-  int NSo   = extracted.size();
-  int NSv   = vsimd::Nsimd();
-  int sparse= NSv/NSo;
-  for(int i=0;i<NSv;i+=sparse){
-    
-  }
-#endif
 };
 template<class vsimd,class scalar>
 inline void Gmerge(vsimd &y,std::vector<scalar *> &extracted){
-#if 1
-  int Nsimd = extracted.size();
+  int Nextr=extracted.size();
+  int Nsimd=vsimd::Nsimd();
+  int s=Nsimd/Nextr;
+
   std::vector<scalar> buf(Nsimd); 
-  for(int i=0;i<Nsimd;i++){
-    buf[i]=*extracted[i];
+  for(int i=0;i<Nextr;i++){
+    for(int ii=0;ii<s;ii++){
+      buf[i*s+ii]=*extracted[i];
+    }
     extracted[i]++;
   }
   vset(y,&buf[0]); 
-#else
-#endif
+};
+template<class vsimd,class scalar>
+inline void Gextract(const vsimd &y,std::vector<scalar> &extracted){
+  // FIXME: bounce off stack is painful
+  // temporary hack while I figure out better way.
+  // There are intrinsics to do this work without the storage.
+  int Nextr=extracted.size();
+  int Nsimd=vsimd::Nsimd();
+  int s=Nsimd/Nextr;
+
+  std::vector<scalar,alignedAllocator<scalar> > buf(Nsimd); 
+
+  vstore(y,&buf[0]);
+
+  for(int i=0;i<Nextr;i++){
+    extracted[i] = buf[i*s];
+  }
+};
+template<class vsimd,class scalar>
+inline void Gmerge(vsimd &y,std::vector<scalar> &extracted){
+  int Nextr=extracted.size();
+  int Nsimd=vsimd::Nsimd();
+  int s=Nsimd/Nextr;
+
+  std::vector<scalar> buf(Nsimd); 
+  for(int i=0;i<Nextr;i++){
+    for(int ii=0;ii<s;ii++){
+      buf[i*s+ii]=extracted[i];
+    }
+  }
+  vset(y,&buf[0]); 
 };
 
 //////////////////////////////////////////////////////////
@@ -183,7 +195,7 @@ inline void Gmerge(vsimd &y,std::vector<scalar *> &extracted){
 // Permute 4 possible on half precision @512bit vectors.
 //////////////////////////////////////////////////////////
 template<class vsimd>
-inline void Gpermute(vsimd &y,vsimd b,int perm){
+inline void Gpermute(vsimd &y,const vsimd &b,int perm){
       switch (perm){
 #if defined(AVX1)||defined(AVX2)
       // 8x32 bits=>3 permutes
@@ -214,10 +226,10 @@ inline void Gpermute(vsimd &y,vsimd b,int perm){
     };
 };
 
+#include <Grid_vInteger.h>
 #include <Grid_vRealF.h>
 #include <Grid_vRealD.h>
 #include <Grid_vComplexF.h>
 #include <Grid_vComplexD.h>
-#include <Grid_vInteger.h>
 
 #endif
diff --git a/Grid_stencil.h b/Grid_stencil.h
index d93204fc..4d73f436 100644
--- a/Grid_stencil.h
+++ b/Grid_stencil.h
@@ -8,5 +8,355 @@
 // Lattice <foo> could also allocate haloes which get used for stencil code.
 //
 // Grid could create a neighbour index table for a given stencil.
-// Could also implement CovariantCshift.
+//
+// Could also implement CovariantCshift, to fuse the loops and enhance performance.
+//
+//
+// General stencil computation:
+//
+// Generic services
+// 0) Prebuild neighbour tables
+// 1) Compute sizes of all haloes/comms buffers; allocate them.
+//
+// 2) Gather all faces, and communicate.
+// 3) Loop over result sites, giving nbr index/offnode info for each
+// 
+// Could take a 
+// SpinProjectFaces 
+// start comms
+// complete comms 
+// Reconstruct Umu
+//
+// Approach.
+//
 //////////////////////////////////////////////////////////////////////////////////////////
+
+namespace Grid {
+
+    class Stencil {
+  public:
+
+      Stencil(GridBase *grid,
+	      int npoints,
+	      int checkerboard,
+	      std::vector<int> directions,
+	      std::vector<int> distances);
+
+      void Stencil_local     (int dimension,int shift,int cbmask);
+      void Stencil_comms     (int dimension,int shift,int cbmask);
+      void Stencil_comms_simd(int dimension,int shift,int cbmask);
+      // Will need to implement actions for
+      //
+      Copy_plane;
+      Copy_plane_permute;
+      Gather_plane;
+
+
+
+      // The offsets to all neibours in stencil in each direction
+      int                               _checkerboard;
+      int                               _npoints; // Move to template param?
+      GridBase *                        _grid;
+
+      // Store these as SIMD Integer needed
+      //
+      // std::vector< iVector<Integer, Npoint> > _offsets;
+      // std::vector< iVector<Integer, Npoint> > _local;
+      // std::vector< iVector<Integer, Npoint> > _comm_buf_size;
+      // std::vector< iVector<Integer, Npoint> > _permute;
+
+      std::vector<std::vector<int>    > _offsets;
+      std::vector<std::vector<int>    > _local;
+      std::vector<int>                  _comm_buf_size;
+      std::vector<int>                  _permute;
+
+    };
+
+    Stencil::Stencil(GridBase *grid,
+		     int npoints,
+		     int checkerboard,
+		     std::vector<int> directions,
+		     std::vector<int> distances){
+      
+      _npoints = npoints;
+      _grid    = grid;
+      
+      for(int i=0;i<npoints;i++){
+
+	int dimension    = directions[i];
+	int displacement = distances[i];
+
+	int fd = _grid->_fdimensions[dimension];
+	int rd = _grid->_rdimensions[dimension];
+
+	_checkerboard = checkerboard;
+
+	// the permute type
+	int simd_layout     = _grid->_simd_layout[dimension];
+	int comm_dim        = _grid->_processors[dimension] >1 ;
+	int splice_dim      = _grid->_simd_layout[dimension]>1 && (comm_dim);
+
+	int sshift[2];
+
+	if ( !comm_dim ) {
+	  sshift[0] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,0);
+	  sshift[1] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,1);
+
+	  if ( sshift[0] == sshift[1] ) {
+	    Stencil_local(dimension,shift,0x3);
+	  } else {
+	    Stencil_local(dimension,shift,0x1);// if checkerboard is unfavourable take two passes
+	    Stencil_local(dimension,shift,0x2);// both with block stride loop iteration
+	  }
+	} else if ( splice_dim ) {
+	  sshift[0] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,0);
+	  sshift[1] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,1);
+	  
+	  if ( sshift[0] == sshift[1] ) {
+	    Stencil_comms_simd(dimension,shift,0x3);
+	  } else {
+	    Stencil_comms_simd(dimension,shift,0x1);// if checkerboard is unfavourable take two passes
+	    Stencil_comms_simd(dimension,shift,0x2);// both with block stride loop iteration
+	  }
+	} else {
+	  //	  Cshift_comms(ret,rhs,dimension,shift);
+	  sshift[0] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,0);
+	  sshift[1] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,1);
+	  if ( sshift[0] == sshift[1] ) {
+	    Stencil_comms(dimension,shift,0x3);
+	  } else {
+	    Stencil_comms(dimension,shift,0x1);// if checkerboard is unfavourable take two passes
+	    Stencil_comms(dimension,shift,0x2);// both with block stride loop iteration
+	  }
+	}
+      }
+    }
+
+
+      void Stencil::Stencil_local     (int dimension,int shift,int cbmask)
+      {
+	int fd = _grid->_fdimensions[dimension];
+	int rd = _grid->_rdimensions[dimension];
+	int ld = _grid->_ldimensions[dimension];
+	int gd = _grid->_gdimensions[dimension];
+
+	// Map to always positive shift modulo global full dimension.
+	shift = (shift+fd)%fd;
+	
+	// the permute type
+	int permute_dim =_grid->PermuteDim(dimension);
+	int permute_type=_grid->PermuteType(dimension);
+	
+	for(int x=0;x<rd;x++){       
+	  
+	  int o   = 0;
+	  int bo  = x * _grid->_ostride[dimension];
+	  
+	  int cb= (cbmask==0x2)? 1 : 0;
+	  
+	  int sshift = _grid->CheckerBoardShift(_checkerboard,dimension,shift,cb);
+	  int sx     = (x+sshift)%rd;
+	  
+	  int permute_slice=0;
+	  if(permute_dim){
+	    int wrap = sshift/rd;
+	    int  num = sshift%rd;
+	    if ( x< rd-num ) permute_slice=wrap;
+	    else permute_slice = 1-wrap;
+	  }
+
+	  if ( permute_slice ) Copy_plane_permute(dimension,x,sx,cbmask,permute_type);
+	  else                 Copy_plane        (dimension,x,sx,cbmask); 
+  
+	}
+      }
+
+      void Stencil::Stencil_comms     (int dimension,int shift,int cbmask)
+      {
+	typedef typename vobj::vector_type vector_type;
+	typedef typename vobj::scalar_type scalar_type;
+	
+	GridBase *grid=_grid;
+
+	int fd              = _grid->_fdimensions[dimension];
+	int rd              = _grid->_rdimensions[dimension];
+	int simd_layout     = _grid->_simd_layout[dimension];
+	int comm_dim        = _grid->_processors[dimension] >1 ;
+
+	assert(simd_layout==1);
+	assert(comm_dim==1);
+	assert(shift>=0);
+	assert(shift<fd);
+	
+	int buffer_size = _grid->_slice_nblock[dimension]*rhs._grid->_slice_block[dimension];
+	// FIXME: Do something with buffer_size??
+
+	int cb= (cbmask==0x2)? 1 : 0;
+	int sshift= _grid->CheckerBoardShift(_checkerboard,dimension,shift,cb);
+	
+	for(int x=0;x<rd;x++){       
+	  
+	  int offnode = ( x+sshift >= rd );
+	  int sx        = (x+sshift)%rd;
+	  int comm_proc = (x+sshift)/rd;
+	  
+	  if (!offnode) {
+	    
+	    Copy_plane(dimension,x,sx,cbmask); 
+	    
+	  } else {
+	    
+	    int words = send_buf.size();
+	    if (cbmask != 0x3) words=words>>1;
+	    
+	    int bytes = words * sizeof(vobj);
+	    
+	    Gather_plane_simple (dimension,sx,cbmask);
+	    
+	    int rank           = grid->_processor;
+	    int recv_from_rank;
+	    int xmit_to_rank;
+	    grid->ShiftedRanks(dimension,comm_proc,xmit_to_rank,recv_from_rank);
+	    /*	    
+	    grid->SendToRecvFrom((void *)&send_buf[0],
+				 xmit_to_rank,
+				 (void *)&recv_buf[0],
+				 recv_from_rank,
+				 bytes);
+	    */
+	    Scatter_plane_simple (dimension,x,cbmask);
+	  }
+	}
+      }
+
+      void Stencil::Stencil_comms_simd(int dimension,int shift,int cbmask)
+      {
+	GridBase *grid=_grid;
+	const int Nsimd = _grid->Nsimd();
+	typedef typename vobj::vector_type vector_type;
+	typedef typename vobj::scalar_type scalar_type;
+	
+	int fd = _grid->_fdimensions[dimension];
+	int rd = _grid->_rdimensions[dimension];
+	int ld = _grid->_ldimensions[dimension];
+	int simd_layout     = _grid->_simd_layout[dimension];
+	int comm_dim        = _grid->_processors[dimension] >1 ;
+	
+	assert(comm_dim==1);
+	assert(simd_layout==2);
+	assert(shift>=0);
+	assert(shift<fd);
+	
+	int permute_type=_grid->PermuteType(dimension);
+	
+	///////////////////////////////////////////////
+	// Simd direction uses an extract/merge pair
+	///////////////////////////////////////////////
+	int buffer_size = _grid->_slice_nblock[dimension]*grid->_slice_block[dimension];
+	// FIXME do something with buffer size
+	
+	std::vector<scalar_type *> pointers(Nsimd);  // 
+	std::vector<scalar_type *> rpointers(Nsimd); // received pointers
+	
+	///////////////////////////////////////////
+	// Work out what to send where
+	///////////////////////////////////////////
+	
+	int cb    = (cbmask==0x2)? 1 : 0;
+	int sshift= _grid->CheckerBoardShift(_checkerboard,dimension,shift,cb);
+	
+	std::vector<int> comm_offnode(simd_layout);
+	std::vector<int> comm_proc   (simd_layout);  //relative processor coord in dim=dimension
+	std::vector<int> icoor(grid->Nd());
+	
+	for(int x=0;x<rd;x++){       
+	  
+	  int comm_any = 0;
+	  for(int s=0;s<simd_layout;s++) {
+	    int shifted_x   = x+s*rd+sshift;
+	    comm_offnode[s] = shifted_x >= ld; 
+	    comm_any        = comm_any | comm_offnode[s];
+	    comm_proc[s]    = shifted_x/ld;     
+	  }
+	  
+	  int o    = 0;
+	  int bo   = x*grid->_ostride[dimension];
+	  int sx   = (x+sshift)%rd;
+	  
+	  if ( comm_any ) {
+	    
+	    for(int i=0;i<Nsimd;i++){
+	      pointers[i] = (scalar_type *)&send_buf_extract[i][0];
+	    }
+	    Gather_plane_extract(rhs,pointers,dimension,sx,cbmask);
+	    
+	    for(int i=0;i<Nsimd;i++){
+	      
+	      int s;
+	      grid->iCoorFromIindex(icoor,i);
+	      s = icoor[dimension];
+	      
+	      if(comm_offnode[s]){
+		
+		int rank           = grid->_processor;
+		int recv_from_rank;
+		int xmit_to_rank;
+		grid->ShiftedRanks(dimension,comm_proc[s],xmit_to_rank,recv_from_rank);
+		
+		/*		
+		grid->SendToRecvFrom((void *)&send_buf_extract[i][0],
+				     xmit_to_rank,
+				     (void *)&recv_buf_extract[i][0],
+				     recv_from_rank,
+				     bytes);
+		*/
+
+		rpointers[i] = (scalar_type *)&recv_buf_extract[i][0];
+		
+	      } else { 
+		
+		rpointers[i] = (scalar_type *)&send_buf_extract[i][0];
+
+	      }
+	      
+	    }
+	    
+	    // Permute by swizzling pointers in merge
+	    int permute_slice=0;
+	    int lshift=sshift%ld;
+	    int wrap  =lshift/rd;
+	    int  num  =lshift%rd;
+
+	    if ( x< rd-num ) permute_slice=wrap;
+	    else permute_slice = 1-wrap;
+	    
+	    int toggle_bit = (Nsimd>>(permute_type+1));
+	    int PermuteMap;
+	    for(int i=0;i<Nsimd;i++){
+	      if ( permute_slice ) {
+		PermuteMap=i^toggle_bit;
+		pointers[i] = rpointers[PermuteMap];
+	      } else {
+		pointers[i] = rpointers[i];
+	      }
+	    }
+
+	    Scatter_plane_merge(pointers,dimension,x,cbmask);
+	    
+	  } else { 
+
+	    int permute_slice=0;
+	    int wrap = sshift/rd;
+	    int  num = sshift%rd;
+	    if ( x< rd-num ) permute_slice=wrap;
+	    else permute_slice = 1-wrap;
+
+	    if ( permute_slice ) Copy_plane_permute(ret,rhs,dimension,x,sx,cbmask,permute_type);
+	    else                 Copy_plane(ret,rhs,dimension,x,sx,cbmask); 
+
+	  }
+	}
+      }
+
+
+};
diff --git a/Grid_vComplexD.h b/Grid_vComplexD.h
index 330de26e..88d69ecb 100644
--- a/Grid_vComplexD.h
+++ b/Grid_vComplexD.h
@@ -16,6 +16,12 @@ namespace Grid {
             return (*this);
         }
         vComplexD(){};
+        vComplexD(ComplexD a){
+	  vsplat(*this,a);
+	};
+        vComplexD(double a){
+	  vsplat(*this,ComplexD(a));
+	};
  
         ///////////////////////////////////////////////
         // mac, mult, sub, add, adj
@@ -167,7 +173,15 @@ namespace Grid {
 	{
 	  Gmerge<vComplexD,ComplexD >(y,extracted);
 	}
-	friend inline void extract(vComplexD &y,std::vector<ComplexD *> &extracted)
+	friend inline void extract(const vComplexD &y,std::vector<ComplexD *> &extracted)
+	{
+	  Gextract<vComplexD,ComplexD>(y,extracted);
+	}
+	friend inline void merge(vComplexD &y,std::vector<ComplexD > &extracted)
+	{
+	  Gmerge<vComplexD,ComplexD >(y,extracted);
+	}
+	friend inline void extract(const vComplexD &y,std::vector<ComplexD > &extracted)
 	{
 	  Gextract<vComplexD,ComplexD>(y,extracted);
 	}
@@ -184,6 +198,11 @@ namespace Grid {
         ///////////////////////
         // Splat
         ///////////////////////
+        friend inline void vsplat(vComplexD &ret,ComplexD c){
+            float a= real(c);
+            float b= imag(c);
+            vsplat(ret,a,b);
+        }
         friend inline void vsplat(vComplexD &ret,double rl,double ig){
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_set_pd(ig,rl,ig,rl);
@@ -215,7 +234,7 @@ namespace Grid {
 #endif
         }
 
-friend inline void vstore(vComplexD &ret, ComplexD *a){
+friend inline void vstore(const vComplexD &ret, ComplexD *a){
 #if defined (AVX1)|| defined (AVX2)
        _mm256_store_pd((double *)a,ret.v);
 #endif
diff --git a/Grid_vComplexF.h b/Grid_vComplexF.h
index b7fb3d6a..b6b7ebe9 100644
--- a/Grid_vComplexF.h
+++ b/Grid_vComplexF.h
@@ -20,6 +20,12 @@ namespace Grid {
             return (*this);
         }
         vComplexF(){};
+        vComplexF(ComplexF a){
+	  vsplat(*this,a);
+	};
+        vComplexF(double a){
+	  vsplat(*this,ComplexF(a));
+	};
        
         ///////////////////////////////////////////////
         // mac, mult, sub, add, adj
@@ -161,7 +167,7 @@ namespace Grid {
             vsplat(ret,a,b);
         }
 
-friend inline void vstore(vComplexF &ret, ComplexF *a){
+friend inline void vstore(const vComplexF &ret, ComplexF *a){
 #if defined (AVX1)|| defined (AVX2)
         _mm256_store_ps((float *)a,ret.v);
 #endif
@@ -210,27 +216,47 @@ friend inline void vstore(vComplexF &ret, ComplexF *a){
 #endif
         }
 
-
         friend inline vComplexF operator * (const Complex &a, vComplexF b){
             vComplexF va;
             vsplat(va,a);
             return va*b;
         }
         friend inline vComplexF operator * (vComplexF b,const Complex &a){
+	  return a*b;
+        }
+
+       /*
+	template<class real>
+        friend inline vComplexF operator * (vComplexF b,const real &a){
             vComplexF va;
-            vsplat(va,a);
+	    Complex ca(a,0);
+            vsplat(va,ca);
             return va*b;
         }
+	template<class real>
+	friend inline vComplexF operator * (const real &a,vComplexF b){
+	  return a*b;
+	}
+
         friend inline vComplexF operator + (const Complex &a, vComplexF b){
             vComplexF va;
             vsplat(va,a);
             return va+b;
         }
         friend inline vComplexF operator + (vComplexF b,const Complex &a){
-            vComplexF va;
-            vsplat(va,a);
-            return b+va;
+            return a+b;
         }
+	template<class real>
+        friend inline vComplexF operator + (vComplexF b,const real &a){
+            vComplexF va;
+	    Complex ca(a,0);
+            vsplat(va,ca);
+            return va+b;
+        }
+	template<class real>
+	friend inline vComplexF operator + (const real &a,vComplexF b){
+	  return a+b;
+	}
         friend inline vComplexF operator - (const Complex &a, vComplexF b){
             vComplexF va;
             vsplat(va,a);
@@ -241,7 +267,24 @@ friend inline void vstore(vComplexF &ret, ComplexF *a){
             vsplat(va,a);
             return b-va;
         }
-        // NB: Template the following on "type Complex" and then implement *,+,- for ComplexF, ComplexD, RealF, RealD above to
+	template<class real>
+        friend inline vComplexF operator - (vComplexF b,const real &a){
+            vComplexF va;
+	    Complex ca(a,0);
+            vsplat(va,ca);
+            return b-va;
+        }
+	template<class real>
+	friend inline vComplexF operator - (const real &a,vComplexF b){
+            vComplexF va;
+	    Complex ca(a,0);
+            vsplat(va,ca);
+            return va-b;
+	}
+       */
+       
+        // NB: Template the following on "type Complex" and then implement *,+,- for 
+	// ComplexF, ComplexD, RealF, RealD above to
         // get full generality of binops with scalars.
         friend inline void mac (vComplexF *__restrict__ y,const Complex *__restrict__ a,const vComplexF *__restrict__ x){ *y = (*a)*(*x)+(*y); };
         friend inline void mult(vComplexF *__restrict__ y,const Complex *__restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) * (*r); }
@@ -304,7 +347,15 @@ friend inline void vstore(vComplexF &ret, ComplexF *a){
       {
 	Gmerge<vComplexF,ComplexF >(y,extracted);
       }
-      friend inline void extract(vComplexF &y,std::vector<ComplexF *> &extracted)
+      friend inline void extract(const vComplexF &y,std::vector<ComplexF *> &extracted)
+      {
+	Gextract<vComplexF,ComplexF>(y,extracted);
+      }
+      friend inline void merge(vComplexF &y,std::vector<ComplexF > &extracted)
+      {
+	Gmerge<vComplexF,ComplexF >(y,extracted);
+      }
+      friend inline void extract(const vComplexF &y,std::vector<ComplexF > &extracted)
       {
 	Gextract<vComplexF,ComplexF>(y,extracted);
       }
diff --git a/Grid_vInteger.h b/Grid_vInteger.h
index 6ddce191..82adbd8e 100644
--- a/Grid_vInteger.h
+++ b/Grid_vInteger.h
@@ -10,7 +10,7 @@ namespace Grid {
 
   typedef uint32_t Integer;
 
-    class vInteger {
+  class vInteger {
     protected:
 
     public:
@@ -21,6 +21,13 @@ namespace Grid {
 	typedef Integer scalar_type;
 
         vInteger(){};
+        vInteger & operator = (const Zero & z){
+            vzero(*this);
+            return (*this);
+        }
+        vInteger(Integer a){
+	  vsplat(*this,a);
+	};
         ////////////////////////////////////
         // Arithmetic operator overloads +,-,*
         ////////////////////////////////////
@@ -166,18 +173,18 @@ namespace Grid {
 #endif
 	}
 
-friend inline void vstore(vInteger &ret, Integer *a){
+	friend inline void vstore(const vInteger &ret, Integer *a){
 #if defined (AVX1)|| defined (AVX2)
-        _mm256_store_si256((__m256i*)a,ret.v);
+	  _mm256_store_si256((__m256i*)a,ret.v);
 #endif
 #ifdef SSE2
-	_mm_store_si128(a,ret.v);
+	  _mm_store_si128(a,ret.v);
 #endif
 #ifdef AVX512
-	_mm512_store_si512(a,ret.v);
+	  _mm512_store_si512(a,ret.v);
 #endif
 #ifdef QPX
-	assert(0);
+	  assert(0);
 #endif
         }
 
@@ -185,6 +192,7 @@ friend inline void vstore(vInteger &ret, Integer *a){
         {
             _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
         }
+
         // Unary negation
         friend inline vInteger operator -(const vInteger &r) {
             vInteger ret;
@@ -210,9 +218,32 @@ friend inline void vstore(vInteger &ret, Integer *a){
             *this = *this-r;
             return *this;
         }
+
+      friend inline void permute(vInteger &y,const vInteger b,int perm)
+      {
+	Gpermute<vInteger>(y,b,perm);
+      }
+      friend inline void merge(vInteger &y,std::vector<Integer *> &extracted)
+      {
+	Gmerge<vInteger,Integer>(y,extracted);
+      }
+      friend inline void extract(const vInteger &y,std::vector<Integer *> &extracted)
+      {
+	Gextract<vInteger,Integer>(y,extracted);
+      }
+      friend inline void merge(vInteger &y,std::vector<Integer> &extracted)
+      {
+	Gmerge<vInteger,Integer>(y,extracted);
+      }
+      friend inline void extract(const vInteger &y,std::vector<Integer> &extracted)
+      {
+	Gextract<vInteger,Integer>(y,extracted);
+      }
+
+
     public:
-        static inline int Nsimd(void) { return sizeof(fvec)/sizeof(float);}
-    };
+        static inline int Nsimd(void) { return sizeof(ivec)/sizeof(Integer);}
+  };
 
     inline vInteger localInnerProduct(const vInteger & l, const vInteger & r) { return l*r; }
 
@@ -222,27 +253,7 @@ friend inline void vstore(vInteger &ret, Integer *a){
     {
         return l*r;
     }
- 
-
-    class vIntegerF : public vInteger
-    {
-    public:
-      static inline int Nsimd(void) { return sizeof(ivec)/sizeof(float);}
-      
-      friend inline void permute(vIntegerF &y,vIntegerF b,int perm)
-      {
-	Gpermute<vIntegerF>(y,b,perm);
-      }
-      friend inline void merge(vIntegerF &y,std::vector<Integer *> &extracted)
-      {
-	Gmerge<vIntegerF,Integer>(y,extracted);
-      }
-      friend inline void extract(vIntegerF &y,std::vector<Integer *> &extracted)
-      {
-	Gextract<vIntegerF,Integer>(y,extracted);
-      }
-    };
-
+    
 }
 
 #endif
diff --git a/Grid_vRealD.h b/Grid_vRealD.h
index 13ceedbe..a5b59be3 100644
--- a/Grid_vRealD.h
+++ b/Grid_vRealD.h
@@ -13,6 +13,9 @@ namespace Grid {
 	typedef RealD scalar_type;
 
         vRealD(){};
+        vRealD(RealD a){
+	  vsplat(*this,a);
+	};
 
         friend inline void mult(vRealD * __restrict__ y,const vRealD * __restrict__ l,const vRealD *__restrict__ r) {*y = (*l) * (*r);}
         friend inline void sub (vRealD * __restrict__ y,const vRealD * __restrict__ l,const vRealD *__restrict__ r) {*y = (*l) - (*r);}
@@ -112,7 +115,15 @@ namespace Grid {
 	{
 	  Gmerge<vRealD,RealD >(y,extracted);
 	}
-	friend inline void extract(vRealD &y,std::vector<RealD *> &extracted)
+	friend inline void extract(const vRealD &y,std::vector<RealD *> &extracted)
+	{
+	  Gextract<vRealD,RealD>(y,extracted);
+	}
+	friend inline void merge(vRealD &y,std::vector<RealD > &extracted)
+	{
+	  Gmerge<vRealD,RealD >(y,extracted);
+	}
+	friend inline void extract(const vRealD &y,std::vector<RealD > &extracted)
 	{
 	  Gextract<vRealD,RealD>(y,extracted);
 	}
@@ -157,7 +168,7 @@ namespace Grid {
 #endif
 	}
 
-	friend inline void vstore(vRealD &ret, double *a){
+	friend inline void vstore(const vRealD &ret, double *a){
 #if defined (AVX1)|| defined (AVX2)
             _mm256_store_pd(a,ret.v);
 #endif
diff --git a/Grid_vRealF.h b/Grid_vRealF.h
index 185f4da8..0fe68f43 100644
--- a/Grid_vRealF.h
+++ b/Grid_vRealF.h
@@ -14,6 +14,9 @@ namespace Grid {
 	typedef RealF scalar_type;
 
         vRealF(){};
+        vRealF(RealF a){
+	  vsplat(*this,a);
+	};
         ////////////////////////////////////
         // Arithmetic operator overloads +,-,*
         ////////////////////////////////////
@@ -133,7 +136,15 @@ namespace Grid {
 	{
 	  Gmerge<vRealF,RealF >(y,extracted);
 	}
-	friend inline void extract(vRealF &y,std::vector<RealF *> &extracted)
+	friend inline void extract(const vRealF &y,std::vector<RealF *> &extracted)
+	{
+	  Gextract<vRealF,RealF>(y,extracted);
+	}
+	friend inline void merge(vRealF &y,std::vector<RealF> &extracted)
+	{
+	  Gmerge<vRealF,RealF >(y,extracted);
+	}
+	friend inline void extract(const vRealF &y,std::vector<RealF> &extracted)
 	{
 	  Gextract<vRealF,RealF>(y,extracted);
 	}
@@ -180,7 +191,7 @@ namespace Grid {
 	////////////////////////////////////////////////////////////////////////
 	// FIXME:  gonna remove these load/store, get, set, prefetch
 	////////////////////////////////////////////////////////////////////////
-friend inline void vstore(vRealF &ret, float *a){
+friend inline void vstore(const vRealF &ret, float *a){
 #if defined (AVX1)|| defined (AVX2)
 	_mm256_store_ps(a,ret.v);
 #endif
diff --git a/TODO b/TODO
index d2373ba1..fb7d177a 100644
--- a/TODO
+++ b/TODO
@@ -1,29 +1,68 @@
+
 * FIXME audit
 * Remove vload/store etc..
 * Replace vset with a call to merge.
 * Replace vset with a call to merge.
+* Const audit
+* extract / merge extra implementation removal
 
-* Conditional execution Subset, where etc...
-* Coordinate information, integers etc...
-* Integer type padding/union to vector.
-* LatticeCoordinate[mu]
+* Conditional execution, where etc...         -----DONE, simple test
+* Integer relational support                         -----DONE
+* Coordinate information, integers etc...            -----DONE
+* Integer type padding/union to vector.              -----DONE 
+* LatticeCoordinate[mu]                              -----DONE
 
-* Optimise the extract/merge SIMD routines
 
-* Broadcast, reduction tests.
+* Stencil operator support                           -----Initial thoughts
+
+* Subset support, slice sums etc...                  -----Only need slice sum?
+                                                     -----Generic cartesian subslicing?
+                                                     -----Array ranges / boost extents?
+                                                     -----Multigrid grid transferral.
+                                                     -----Suggests generalised cartesian subblocking
+                                                          sums, returning modified grid.
+
+    Two classes of subset;
+i) red black parit subsetting.
+   (pick checkerboard).
+
+ii) Need to be able to project one Grid to another Grid.
+    Generic concept is to subdivide (based on RD so applies to red/black or full).
+    Return a type on SUB-grid from CellSum TOP-grid
+    SUB-grid need not distribute but be replicated in some dims if that is how the
+    cartesian communicator works.
+
+iii) No general permutation map.
+
+
+* Consider switch std::vector to boost arrays.
+  boost::multi_array<type, 3> A()...    to replace multi1d, multi2d etc..
+
+*? Cell definition <-> sliceSum.
+ ? Replicated arrays.
+
+
+
+* Check for missing functionality                    - partially audited against QDP++ layout
+
+* Optimise the extract/merge SIMD routines; Azusa??
+
+ -- I have collated into single location at least.
+
+ -- Need to use _mm_*insert/extract routines.
+
+* Conformable test in Cshift routines.
+
+* Gamma/Dirac structures
+
+* Fourspin, two spin project
+
+* Broadcast, reduction tests. innerProduct, localInnerProduct
 
 * QDP++ regression suite and comparative benchmark
 
 * NERSC Lattice loading, plaquette test
 
-* Conformable test in Cshift routines.
-
-* Gamma/Dirac structures
-* Fourspin, two spin project
-
-* Stencil operator support
-
-* Check for missing functionality
 
 * I/O support
   - MPI IO?

From 927c62d8a3181be9ddcce8146b8c8cf28e61e10b Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 10 Apr 2015 05:21:48 +0200
Subject: [PATCH 043/429] Patch for comms none nocompile

---
 Grid_communicator_fake.cc | 39 +++++++++++++++++++++++++++++++--------
 1 file changed, 31 insertions(+), 8 deletions(-)

diff --git a/Grid_communicator_fake.cc b/Grid_communicator_fake.cc
index 83d5475e..2b94ea4b 100644
--- a/Grid_communicator_fake.cc
+++ b/Grid_communicator_fake.cc
@@ -14,20 +14,43 @@ CartesianCommunicator::CartesianCommunicator(std::vector<int> &processors)
   for(int d=0;d<_ndimension;d++) _processor_coor[d] = 0;
 }
 
-void CartesianCommunicator::GlobalSumF(float &){}
-void CartesianCommunicator::GlobalSumFVector(float *,int N){}
-void CartesianCommunicator::GlobalSumF(double &){}
-void CartesianCommunicator::GlobalSumFVector(double *,int N){}
+void CartesianCommunicator::GlobalSum(float &){}
+void CartesianCommunicator::GlobalSumVector(float *,int N){}
+void CartesianCommunicator::GlobalSum(double &){}
+void CartesianCommunicator::GlobalSumVector(double *,int N){}
 
 // Basic Halo comms primitive
 void CartesianCommunicator::SendToRecvFrom(void *xmit,
-		    std::vector<int> to_coordinate,
-		    void *recv,
-		    std::vector<int> from_coordinate,
-		    int bytes)
+					   int dest,
+					   void *recv,
+					   int from,
+					   int bytes)
 {
   exit(-1);
 }
 
+void CartesianCommunicator::Barrier(void)
+{
+}
+
+void CartesianCommunicator::Broadcast(int root,void* data, int bytes)
+{
+}
+
+
+void CartesianCommunicator::ShiftedRanks(int dim,int shift,int &source,int &dest)
+{
+  source =1;
+  dest=1;
+}
+int CartesianCommunicator::RankFromProcessorCoor(std::vector<int> &coor)
+{
+  return 1;
+}
+void  CartesianCommunicator::ProcessorCoorFromRank(int rank, std::vector<int> &coor)
+{
+}
+
+
 }
 

From 993419d9fbd857ab921944056d29832c6a501d4e Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 10 Apr 2015 05:22:36 +0200
Subject: [PATCH 044/429] MPI exposed incorrectly in main

---
 Grid.h       | 1 +
 Grid_init.cc | 6 ++++++
 2 files changed, 7 insertions(+)

diff --git a/Grid.h b/Grid.h
index caa4ed5e..a0989e0b 100644
--- a/Grid.h
+++ b/Grid.h
@@ -52,6 +52,7 @@
 namespace Grid {
 
   void Grid_init(int *argc,char ***argv);
+  void Grid_finalize(void);
   double usecond(void);
   void Grid_sa_signal_handler(int sig,siginfo_t *si,void * ptr);
   void Grid_debug_handler_init(void);
diff --git a/Grid_init.cc b/Grid_init.cc
index 3a601209..9e8b8b96 100755
--- a/Grid_init.cc
+++ b/Grid_init.cc
@@ -23,6 +23,12 @@ void Grid_init(int *argc,char ***argv)
 #endif
   Grid_debug_handler_init();
 }
+void Grid_finalize(void)
+{
+#ifdef GRID_COMMS_MPI
+  MPI_Finalize();
+#endif
+}
 double usecond(void) {
   struct timeval tv;
   gettimeofday(&tv,NULL);

From 6e90038bf63492c268b0c50d757a0fdcab6ab1ed Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 10 Apr 2015 05:24:01 +0200
Subject: [PATCH 045/429] Fixing nocompile

---
 Grid_config.h    |  4 ++--
 Grid_main.cc     |  4 +---
 Grid_stencil.h   |  7 ++++---
 Grid_vComplexD.h |  9 ---------
 Grid_vRealD.h    |  9 ---------
 TODO             | 37 ++++++++++++++++++++++++-------------
 6 files changed, 31 insertions(+), 39 deletions(-)

diff --git a/Grid_config.h b/Grid_config.h
index bb885708..deb6c74c 100644
--- a/Grid_config.h
+++ b/Grid_config.h
@@ -11,10 +11,10 @@
 /* #undef AVX512 */
 
 /* GRID_COMMS_MPI */
-#define GRID_COMMS_MPI 1
+/* #undef GRID_COMMS_MPI */
 
 /* GRID_COMMS_NONE */
-/* #undef GRID_COMMS_NONE */
+#define GRID_COMMS_NONE 1
 
 /* Define to 1 if you have the `gettimeofday' function. */
 #define HAVE_GETTIMEOFDAY 1
diff --git a/Grid_main.cc b/Grid_main.cc
index 1676a07b..770de9a3 100644
--- a/Grid_main.cc
+++ b/Grid_main.cc
@@ -407,7 +407,5 @@ int main (int argc, char ** argv)
 
    } // loop for lat
  } // loop for omp
-
- MPI_Finalize();
-
+ Grid_finalize();
 }
diff --git a/Grid_stencil.h b/Grid_stencil.h
index 4d73f436..e10cabaa 100644
--- a/Grid_stencil.h
+++ b/Grid_stencil.h
@@ -1,3 +1,5 @@
+#ifndef GRID_STENCIL_H
+#define GRID_STENCIL_H
 //////////////////////////////////////////////////////////////////////////////////////////
 // Must not lose sight that goal is to be able to construct really efficient
 // gather to a point stencil code. CSHIFT is not the best way, so probably need
@@ -45,14 +47,12 @@ namespace Grid {
       void Stencil_local     (int dimension,int shift,int cbmask);
       void Stencil_comms     (int dimension,int shift,int cbmask);
       void Stencil_comms_simd(int dimension,int shift,int cbmask);
+
       // Will need to implement actions for
-      //
       Copy_plane;
       Copy_plane_permute;
       Gather_plane;
 
-
-
       // The offsets to all neibours in stencil in each direction
       int                               _checkerboard;
       int                               _npoints; // Move to template param?
@@ -360,3 +360,4 @@ namespace Grid {
 
 
 };
+#endif
diff --git a/Grid_vComplexD.h b/Grid_vComplexD.h
index 88d69ecb..3278b15e 100644
--- a/Grid_vComplexD.h
+++ b/Grid_vComplexD.h
@@ -186,15 +186,6 @@ namespace Grid {
 	  Gextract<vComplexD,ComplexD>(y,extracted);
 	}
 
-	////////////////////////////////////////////////////////////////////////
-	// FIXME:  gonna remove these load/store, get, set, prefetch
-	////////////////////////////////////////////////////////////////////////
-        void vload(zvec& a){
-          this->v = a;
-        }
-        zvec vget(){
-          return this->v ;
-        }
         ///////////////////////
         // Splat
         ///////////////////////
diff --git a/Grid_vRealD.h b/Grid_vRealD.h
index a5b59be3..c51e41cf 100644
--- a/Grid_vRealD.h
+++ b/Grid_vRealD.h
@@ -128,15 +128,6 @@ namespace Grid {
 	  Gextract<vRealD,RealD>(y,extracted);
 	}
 
-	////////////////////////////////////////////////////////////////////////
-	// FIXME:  gonna remove these load/store, get, set, prefetch
-	////////////////////////////////////////////////////////////////////////
-        void vload(dvec& a){
-          this->v = a;
-        }
-        dvec vget(){
-          return this->v ;
-        }
         
         friend inline void vsplat(vRealD &ret,double a){
 #if defined (AVX1)|| defined (AVX2)
diff --git a/TODO b/TODO
index fb7d177a..5297e213 100644
--- a/TODO
+++ b/TODO
@@ -1,12 +1,12 @@
 
 * FIXME audit
-* Remove vload/store etc..
-* Replace vset with a call to merge.
-* Replace vset with a call to merge.
+* Replace vset with a call to merge.; 
+* care in Gmerge,Gextract over set.
 * Const audit
 * extract / merge extra implementation removal
 
-* Conditional execution, where etc...         -----DONE, simple test
+
+* Conditional execution, where etc...                -----DONE, simple test
 * Integer relational support                         -----DONE
 * Coordinate information, integers etc...            -----DONE
 * Integer type padding/union to vector.              -----DONE 
@@ -22,18 +22,32 @@
                                                      -----Suggests generalised cartesian subblocking
                                                           sums, returning modified grid.
 
-    Two classes of subset;
-i) red black parit subsetting.
-   (pick checkerboard).
+i)  Two classes of subset;   red black parity subsetting (pick checkerboard).
+                             cartesian sub-block subsetting
+
 
 ii) Need to be able to project one Grid to another Grid.
+
+Interface: (?)
+
+Lattice<vobj> coarse_data SubBlockSum (GridBase *CoarseGrid, Lattice<vobj> &fine_data)
+
+Operation ensure either:
+ rd[dim] divide rd[dim] fine_data
+
+This will give a distributed array over mpi ranks in a given dim IF coarse gd != 1 and _processors[d]>1
+Dimension can be *replicated* on all ranks in dimension. Need a "replicated" option on GridCartesian etc..
+
+This will give "slice" summation and fourier projection assistance.
+
     Generic concept is to subdivide (based on RD so applies to red/black or full).
     Return a type on SUB-grid from CellSum TOP-grid
     SUB-grid need not distribute but be replicated in some dims if that is how the
     cartesian communicator works.
 
-iii) No general permutation map.
+Instead of subsetting 
 
+iii) No general permutation map.
 
 * Consider switch std::vector to boost arrays.
   boost::multi_array<type, 3> A()...    to replace multi1d, multi2d etc..
@@ -41,15 +55,12 @@ iii) No general permutation map.
 *? Cell definition <-> sliceSum.
  ? Replicated arrays.
 
-
-
 * Check for missing functionality                    - partially audited against QDP++ layout
 
 * Optimise the extract/merge SIMD routines; Azusa??
 
- -- I have collated into single location at least.
-
- -- Need to use _mm_*insert/extract routines.
+ - I have collated into single location at least.
+ - Need to use _mm_*insert/extract routines.
 
 * Conformable test in Cshift routines.
 

From 526765874844641e3c0061ddc7145a8eeba0b44a Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 10 Apr 2015 05:53:09 +0200
Subject: [PATCH 046/429] Fixing the comms=none compile

---
 Grid_Lattice.h     | 3 ++-
 Grid_cshift_none.h | 1 +
 Grid_main.cc       | 4 +++-
 3 files changed, 6 insertions(+), 2 deletions(-)

diff --git a/Grid_Lattice.h b/Grid_Lattice.h
index 01b95755..119d6181 100644
--- a/Grid_Lattice.h
+++ b/Grid_Lattice.h
@@ -15,9 +15,10 @@ public:
     GridBase *_grid;
     int checkerboard;
     std::vector<vobj,alignedAllocator<vobj> > _odata;
+public:
+
     typedef typename vobj::scalar_type scalar_type;
     typedef typename vobj::vector_type vector_type;
-public:
 
     Lattice(GridBase *grid) : _grid(grid) {
         _odata.reserve(_grid->oSites());
diff --git a/Grid_cshift_none.h b/Grid_cshift_none.h
index 37d20fa3..402e52b4 100644
--- a/Grid_cshift_none.h
+++ b/Grid_cshift_none.h
@@ -6,6 +6,7 @@ friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
   Lattice<vobj> ret(rhs._grid);
   ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
   Cshift_local(ret,rhs,dimension,shift);
+  return ret;
 }
         
 #endif
diff --git a/Grid_main.cc b/Grid_main.cc
index 770de9a3..c51aff6d 100644
--- a/Grid_main.cc
+++ b/Grid_main.cc
@@ -195,6 +195,7 @@ int main (int argc, char ** argv)
       LatticeCoordinate(coor,d);
       lex = lex + coor*mm[d];
     }
+    /*
     Bar = zero;
     Bar = where(lex<10,Foo,Bar);
     {
@@ -210,7 +211,8 @@ int main (int argc, char ** argv)
 	  cout<<"bar "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "<<bar._internal._internal[r][c]<<std::endl;
 	}}
       }}}}
-    }    
+    } 
+    */   
     //setCheckerboard(ShiftedCheck,rFoo); 
     //setCheckerboard(ShiftedCheck,bFoo); 
 

From 977c7721d52bb933824dede7abe6d188912c7a6e Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 10 Apr 2015 05:54:02 +0200
Subject: [PATCH 047/429] where switched back on

---
 Grid_main.cc | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/Grid_main.cc b/Grid_main.cc
index c51aff6d..8cf0dae3 100644
--- a/Grid_main.cc
+++ b/Grid_main.cc
@@ -195,7 +195,7 @@ int main (int argc, char ** argv)
       LatticeCoordinate(coor,d);
       lex = lex + coor*mm[d];
     }
-    /*
+
     Bar = zero;
     Bar = where(lex<10,Foo,Bar);
     {
@@ -212,7 +212,7 @@ int main (int argc, char ** argv)
 	}}
       }}}}
     } 
-    */   
+
     //setCheckerboard(ShiftedCheck,rFoo); 
     //setCheckerboard(ShiftedCheck,bFoo); 
 

From 2d54ef2a521aa6bd137ce163989de74f41f67cbb Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 14 Apr 2015 20:22:04 +0100
Subject: [PATCH 048/429] Stencil code pretty much shaken out. Beginning of
 inner product and norm2.

---
 Grid.h                 |   1 +
 Grid_Lattice.h         |  12 +
 Grid_config.h          |   4 +-
 Grid_cshift_common.h   |   2 +
 Grid_main.cc           |  15 +-
 Grid_math_types.h      |   3 +-
 Grid_simd.h            |  53 +++-
 Grid_stencil.h         | 572 ++++++++++++++++++++---------------------
 Grid_stencil_common.cc | 258 +++++++++++++++++++
 Grid_vComplexD.h       |   7 +-
 Grid_vComplexF.h       |  38 +--
 Makefile.am            |  33 ++-
 Makefile.in            |  53 ++--
 TODO                   |   2 +-
 test_Grid_jacobi.cc    | 206 +++++++++++++++
 test_Grid_stencil.cc   | 154 +++++++++++
 16 files changed, 1050 insertions(+), 363 deletions(-)
 create mode 100644 Grid_stencil_common.cc
 create mode 100644 test_Grid_jacobi.cc
 create mode 100644 test_Grid_stencil.cc

diff --git a/Grid.h b/Grid.h
index a0989e0b..2a1072ea 100644
--- a/Grid.h
+++ b/Grid.h
@@ -47,6 +47,7 @@
 #include <Grid_Cartesian.h>
 #include <Grid_Lattice.h>
 #include <Grid_comparison.h>
+#include <Grid_stencil.h>
 #include <Grid_QCD.h>
 
 namespace Grid {
diff --git a/Grid_Lattice.h b/Grid_Lattice.h
index 119d6181..1e2c820e 100644
--- a/Grid_Lattice.h
+++ b/Grid_Lattice.h
@@ -407,6 +407,18 @@ public:
         return ret;
     }
 
+    template<class vobj>
+    inline RealD norm2(const Lattice<vobj> &arg){
 
+      typedef typename vobj::scalar_type scalar;
+      decltype(localInnerProduct(arg._odata[0],arg._odata[0])) vnrm=zero;
+
+      scalar nrm;
+      for(int ss=0;ss<arg._grid->oSites(); ss++){
+	vnrm = vnrm + localInnerProduct(arg._odata[ss],arg._odata[ss]);
+      }
+      nrm = Reduce(TensorRemove(vnrm));
+      return real(nrm);
+    }
 }
 #endif
diff --git a/Grid_config.h b/Grid_config.h
index deb6c74c..bb885708 100644
--- a/Grid_config.h
+++ b/Grid_config.h
@@ -11,10 +11,10 @@
 /* #undef AVX512 */
 
 /* GRID_COMMS_MPI */
-/* #undef GRID_COMMS_MPI */
+#define GRID_COMMS_MPI 1
 
 /* GRID_COMMS_NONE */
-#define GRID_COMMS_NONE 1
+/* #undef GRID_COMMS_NONE */
 
 /* Define to 1 if you have the `gettimeofday' function. */
 #define HAVE_GETTIMEOFDAY 1
diff --git a/Grid_cshift_common.h b/Grid_cshift_common.h
index 2910151c..e962f6ea 100644
--- a/Grid_cshift_common.h
+++ b/Grid_cshift_common.h
@@ -8,6 +8,7 @@ friend void Gather_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllo
 {
   int rd = rhs._grid->_rdimensions[dimension];
 
+  printf("Gather_plane_simple buf size %d rhs size %d \n",buffer.size(),rhs._odata.size());fflush(stdout);
   if ( !rhs._grid->CheckerBoarded(dimension) ) {
 
     int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
@@ -18,6 +19,7 @@ friend void Gather_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllo
 #pragma omp parallel for collapse(2)
     for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
       for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+	printf("Gather_plane_simple %d ; %d %d %d\n",bo,so,o,b);fflush(stdout);
 	buffer[bo++]=rhs._odata[so+o+b];
       }
       o +=rhs._grid->_slice_stride[dimension];
diff --git a/Grid_main.cc b/Grid_main.cc
index 8cf0dae3..13c3d9ee 100644
--- a/Grid_main.cc
+++ b/Grid_main.cc
@@ -1,10 +1,4 @@
 #include "Grid.h"
-#include "Grid_vRealD.h"
-#include "Grid_vRealF.h"
-#include "Grid_vComplexD.h"
-#include "Grid_vComplexF.h"
-#include "Grid_Cartesian.h"
-#include "Grid_Lattice.h"
 
 using namespace std;
 using namespace Grid;
@@ -16,13 +10,14 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> latt_size(4);
+
   std::vector<int> simd_layout(4);
   
   std::vector<int> mpi_layout(4);
-  mpi_layout[0]=1;
-  mpi_layout[1]=1;
-  mpi_layout[2]=1;
-  mpi_layout[3]=1;
+  mpi_layout[0]=2;
+  mpi_layout[1]=2;
+  mpi_layout[2]=2;
+  mpi_layout[3]=2;
 
 #ifdef AVX512
  for(int omp=128;omp<236;omp+=16){
diff --git a/Grid_math_types.h b/Grid_math_types.h
index a2a3f97c..e3432645 100644
--- a/Grid_math_types.h
+++ b/Grid_math_types.h
@@ -922,9 +922,10 @@ template<class l,int N> inline iMatrix<l,N> operator - (Integer lhs,const iMatri
     {
         typedef decltype(localInnerProduct(lhs._internal[0][0],rhs._internal[0][0])) ret_t;
         iScalar<ret_t> ret=zero;
+        iScalar<ret_t> tmp;
         for(int c1=0;c1<N;c1++){
         for(int c2=0;c2<N;c2++){
-            ret._internal += localInnerProduct(lhs._internal[c1][c2],rhs._internal[c1][c2]);
+	  ret._internal+=localInnerProduct(lhs._internal[c1][c2],rhs._internal[c1][c2]);
         }}
         return ret;
     }
diff --git a/Grid_simd.h b/Grid_simd.h
index 1229d328..23061632 100644
--- a/Grid_simd.h
+++ b/Grid_simd.h
@@ -26,11 +26,10 @@ namespace Grid {
 
   typedef  float  RealF;
   typedef  double RealD;
-  typedef  RealF  Real;
   
   typedef std::complex<RealF> ComplexF;
   typedef std::complex<RealD> ComplexD;
-  typedef std::complex<Real>  Complex;
+
 
   inline RealF adj(const RealF  & r){ return r; }
   inline RealF conj(const RealF  & r){ return r; }
@@ -54,7 +53,7 @@ namespace Grid {
     inline void mult(ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) * (*r); }
     inline void sub (ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) - (*r); }
     inline void add (ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) + (*r); }
-    inline Complex  adj(const Complex& r ){ return(conj(r)); }
+    inline ComplexF  adj(const ComplexF& r ){ return(conj(r)); }
     //conj already supported for complex
     
     inline void mac (RealD * __restrict__ y,const RealD * __restrict__ a,const RealD *__restrict__ x){  *y = (*a) * (*x)+(*y);}
@@ -232,4 +231,52 @@ inline void Gpermute(vsimd &y,const vsimd &b,int perm){
 #include <Grid_vComplexF.h>
 #include <Grid_vComplexD.h>
 
+namespace Grid {
+
+  // NB: Template the following on "type Complex" and then implement *,+,- for 
+  // ComplexF, ComplexD, RealF, RealD above to
+  // get full generality of binops with scalars.
+   inline void mac (vComplexF *__restrict__ y,const ComplexF *__restrict__ a,const vComplexF *__restrict__ x){ *y = (*a)*(*x)+(*y); };
+   inline void mult(vComplexF *__restrict__ y,const ComplexF *__restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) * (*r); }
+   inline void sub (vComplexF *__restrict__ y,const ComplexF *__restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) - (*r); }
+   inline void add (vComplexF *__restrict__ y,const ComplexF *__restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) + (*r); }
+   inline void mac (vComplexF *__restrict__ y,const vComplexF *__restrict__ a,const ComplexF *__restrict__ x){ *y = (*a)*(*x)+(*y); };
+   inline void mult(vComplexF *__restrict__ y,const vComplexF *__restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) * (*r); }
+   inline void sub (vComplexF *__restrict__ y,const vComplexF *__restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) - (*r); }
+   inline void add (vComplexF *__restrict__ y,const vComplexF *__restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) + (*r); }
+
+   inline void mac (vComplexD *__restrict__ y,const ComplexD *__restrict__ a,const vComplexD *__restrict__ x){ *y = (*a)*(*x)+(*y); };
+   inline void mult(vComplexD *__restrict__ y,const ComplexD *__restrict__ l,const vComplexD *__restrict__ r){ *y = (*l) * (*r); }
+   inline void sub (vComplexD *__restrict__ y,const ComplexD *__restrict__ l,const vComplexD *__restrict__ r){ *y = (*l) - (*r); }
+   inline void add (vComplexD *__restrict__ y,const ComplexD *__restrict__ l,const vComplexD *__restrict__ r){ *y = (*l) + (*r); }
+   inline void mac (vComplexD *__restrict__ y,const vComplexD *__restrict__ a,const ComplexD *__restrict__ x){ *y = (*a)*(*x)+(*y); };
+   inline void mult(vComplexD *__restrict__ y,const vComplexD *__restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) * (*r); }
+   inline void sub (vComplexD *__restrict__ y,const vComplexD *__restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) - (*r); }
+   inline void add (vComplexD *__restrict__ y,const vComplexD *__restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) + (*r); }
+
+   inline void mac (vRealF *__restrict__ y,const RealF *__restrict__ a,const vRealF *__restrict__ x){ *y = (*a)*(*x)+(*y); };
+   inline void mult(vRealF *__restrict__ y,const RealF *__restrict__ l,const vRealF *__restrict__ r){ *y = (*l) * (*r); }
+   inline void sub (vRealF *__restrict__ y,const RealF *__restrict__ l,const vRealF *__restrict__ r){ *y = (*l) - (*r); }
+   inline void add (vRealF *__restrict__ y,const RealF *__restrict__ l,const vRealF *__restrict__ r){ *y = (*l) + (*r); }
+   inline void mac (vRealF *__restrict__ y,const vRealF *__restrict__ a,const RealF *__restrict__ x){ *y = (*a)*(*x)+(*y); };
+   inline void mult(vRealF *__restrict__ y,const vRealF *__restrict__ l,const RealF *__restrict__ r){ *y = (*l) * (*r); }
+   inline void sub (vRealF *__restrict__ y,const vRealF *__restrict__ l,const RealF *__restrict__ r){ *y = (*l) - (*r); }
+   inline void add (vRealF *__restrict__ y,const vRealF *__restrict__ l,const RealF *__restrict__ r){ *y = (*l) + (*r); }
+
+   inline void mac (vRealD *__restrict__ y,const RealD *__restrict__ a,const vRealD *__restrict__ x){ *y = (*a)*(*x)+(*y); };
+   inline void mult(vRealD *__restrict__ y,const RealD *__restrict__ l,const vRealD *__restrict__ r){ *y = (*l) * (*r); }
+   inline void sub (vRealD *__restrict__ y,const RealD *__restrict__ l,const vRealD *__restrict__ r){ *y = (*l) - (*r); }
+   inline void add (vRealD *__restrict__ y,const RealD *__restrict__ l,const vRealD *__restrict__ r){ *y = (*l) + (*r); }
+   inline void mac (vRealD *__restrict__ y,const vRealD *__restrict__ a,const RealD *__restrict__ x){ *y = (*a)*(*x)+(*y); };
+   inline void mult(vRealD *__restrict__ y,const vRealD *__restrict__ l,const RealD *__restrict__ r){ *y = (*l) * (*r); }
+   inline void sub (vRealD *__restrict__ y,const vRealD *__restrict__ l,const RealD *__restrict__ r){ *y = (*l) - (*r); }
+   inline void add (vRealD *__restrict__ y,const vRealD *__restrict__ l,const RealD *__restrict__ r){ *y = (*l) + (*r); }
+
+  // Default precision
+  typedef RealD  Real;
+  typedef std::complex<Real>  Complex;
+
+  typedef vRealD vReal;
+  typedef vComplexD vComplex;
+}
 #endif
diff --git a/Grid_stencil.h b/Grid_stencil.h
index e10cabaa..d8debac0 100644
--- a/Grid_stencil.h
+++ b/Grid_stencil.h
@@ -1,11 +1,15 @@
 #ifndef GRID_STENCIL_H
 #define GRID_STENCIL_H
+
 //////////////////////////////////////////////////////////////////////////////////////////
 // Must not lose sight that goal is to be able to construct really efficient
-// gather to a point stencil code. CSHIFT is not the best way, so probably need
+// gather to a point stencil code. CSHIFT is not the best way, so need
 // additional stencil support.
 //
-// Stencil based code could pre-exchange haloes and use a table lookup for neighbours
+// Stencil based code will pre-exchange haloes and use a table lookup for neighbours.
+// This will be done with generality to allow easier efficient implementations.
+// Overlap of comms and compute could be semi-automated by tabulating off-node connected,
+// and 
 //
 // Lattice <foo> could also allocate haloes which get used for stencil code.
 //
@@ -35,329 +39,313 @@
 
 namespace Grid {
 
-    class Stencil {
+  struct CommsRequest { 
+    int words;
+    int unified_buffer_offset;
+    int tag;
+    int to_rank;
+    int from_rank;
+  } ;
+
+  class CartesianStencil { // Stencil runs along coordinate axes only; NO diagonal fill in.
   public:
 
-      Stencil(GridBase *grid,
-	      int npoints,
-	      int checkerboard,
-	      std::vector<int> directions,
-	      std::vector<int> distances);
-
-      void Stencil_local     (int dimension,int shift,int cbmask);
-      void Stencil_comms     (int dimension,int shift,int cbmask);
-      void Stencil_comms_simd(int dimension,int shift,int cbmask);
-
-      // Will need to implement actions for
-      Copy_plane;
-      Copy_plane_permute;
-      Gather_plane;
-
-      // The offsets to all neibours in stencil in each direction
       int                               _checkerboard;
       int                               _npoints; // Move to template param?
       GridBase *                        _grid;
-
-      // Store these as SIMD Integer needed
-      //
-      // std::vector< iVector<Integer, Npoint> > _offsets;
-      // std::vector< iVector<Integer, Npoint> > _local;
-      // std::vector< iVector<Integer, Npoint> > _comm_buf_size;
-      // std::vector< iVector<Integer, Npoint> > _permute;
-
-      std::vector<std::vector<int>    > _offsets;
-      std::vector<std::vector<int>    > _local;
+      
+      // npoints of these
+      std::vector<int>                  _directions;
+      std::vector<int>                  _distances;
       std::vector<int>                  _comm_buf_size;
-      std::vector<int>                  _permute;
+      std::vector<int>                  _permute_type;
 
-    };
+      // npoints x Osites() of these
+      std::vector<std::vector<int>    > _offsets;
+      std::vector<std::vector<int>    > _is_local;
+      std::vector<std::vector<int> >    _permute;
 
-    Stencil::Stencil(GridBase *grid,
-		     int npoints,
-		     int checkerboard,
-		     std::vector<int> directions,
-		     std::vector<int> distances){
-      
-      _npoints = npoints;
-      _grid    = grid;
-      
-      for(int i=0;i<npoints;i++){
+      int _unified_buffer_size;
+      int _request_count;
 
-	int dimension    = directions[i];
-	int displacement = distances[i];
+      std::vector<CommsRequest>  CommsRequests;
 
-	int fd = _grid->_fdimensions[dimension];
-	int rd = _grid->_rdimensions[dimension];
+      CartesianStencil(GridBase *grid,
+		       int npoints,
+		       int checkerboard,
+		       const std::vector<int> &directions,
+		       const std::vector<int> &distances);
 
-	_checkerboard = checkerboard;
 
-	// the permute type
-	int simd_layout     = _grid->_simd_layout[dimension];
-	int comm_dim        = _grid->_processors[dimension] >1 ;
-	int splice_dim      = _grid->_simd_layout[dimension]>1 && (comm_dim);
+      // Add to tables for various cases;  is this mistaken. only local if 1 proc in dim
+      // Can this be avoided with simpler coding of comms?
+      void Local     (int point, int dimension,int shift,int cbmask);
+      void Comms     (int point, int dimension,int shift,int cbmask);
+      void CopyPlane(int point, int dimension,int lplane,int rplane,int cbmask,int permute);
+      void ScatterPlane (int point,int dimension,int plane,int cbmask,int offset);
 
-	int sshift[2];
+      // Could allow a functional munging of the halo to another type during the comms.
+      // this could implement the 16bit/32bit/64bit compression.
+      template<class vobj> void HaloExchange(Lattice<vobj> &source,
+					     std::vector<vobj,alignedAllocator<vobj> > &u_comm_buf)
+      {
+	// conformable(source._grid,_grid);
+	assert(source._grid==_grid);
+	if (u_comm_buf.size() != _unified_buffer_size ) u_comm_buf.resize(_unified_buffer_size);
+	int u_comm_offset=0;
 
-	if ( !comm_dim ) {
-	  sshift[0] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,0);
-	  sshift[1] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,1);
+	// Gather all comms buffers
+	typedef typename vobj::vector_type vector_type;
+	typedef typename vobj::scalar_type scalar_type;
 
-	  if ( sshift[0] == sshift[1] ) {
-	    Stencil_local(dimension,shift,0x3);
-	  } else {
-	    Stencil_local(dimension,shift,0x1);// if checkerboard is unfavourable take two passes
-	    Stencil_local(dimension,shift,0x2);// both with block stride loop iteration
-	  }
-	} else if ( splice_dim ) {
-	  sshift[0] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,0);
-	  sshift[1] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,1);
+	for(int point = 0 ; point < _npoints; point++) {
+
+	  printf("Point %d \n",point);fflush(stdout);
+	  int dimension    = _directions[point];
+	  int displacement = _distances[point];
 	  
-	  if ( sshift[0] == sshift[1] ) {
-	    Stencil_comms_simd(dimension,shift,0x3);
-	  } else {
-	    Stencil_comms_simd(dimension,shift,0x1);// if checkerboard is unfavourable take two passes
-	    Stencil_comms_simd(dimension,shift,0x2);// both with block stride loop iteration
-	  }
-	} else {
-	  //	  Cshift_comms(ret,rhs,dimension,shift);
-	  sshift[0] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,0);
-	  sshift[1] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,1);
-	  if ( sshift[0] == sshift[1] ) {
-	    Stencil_comms(dimension,shift,0x3);
-	  } else {
-	    Stencil_comms(dimension,shift,0x1);// if checkerboard is unfavourable take two passes
-	    Stencil_comms(dimension,shift,0x2);// both with block stride loop iteration
+	  int fd = _grid->_fdimensions[dimension];
+	  int rd = _grid->_rdimensions[dimension];
+
+
+	  // Map to always positive shift modulo global full dimension.
+	  int shift = (displacement+fd)%fd;
+	  
+     	  int checkerboard = _grid->CheckerBoardDestination(source.checkerboard,shift);
+	  assert (checkerboard== _checkerboard);
+
+	  // the permute type
+	  int simd_layout     = _grid->_simd_layout[dimension];
+	  int comm_dim        = _grid->_processors[dimension] >1 ;
+	  int splice_dim      = _grid->_simd_layout[dimension]>1 && (comm_dim);
+
+	  // Gather phase
+	  int sshift [2];
+	  if ( comm_dim ) {
+	    sshift[0] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,0);
+	    sshift[1] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,1);
+	    if ( sshift[0] == sshift[1] ) {
+	      if (splice_dim) {
+		printf("splice 0x3 \n");fflush(stdout);
+		GatherStartCommsSimd(source,dimension,shift,0x3,u_comm_buf,u_comm_offset);
+	      } else { 
+		printf("NO splice 0x3 \n");fflush(stdout);
+		GatherStartComms(source,dimension,shift,0x3,u_comm_buf,u_comm_offset);
+	      }
+	    } else {
+	      if(splice_dim){
+		printf("splice 0x1,2 \n");fflush(stdout);
+		GatherStartCommsSimd(source,dimension,shift,0x1,u_comm_buf,u_comm_offset);// if checkerboard is unfavourable take two passes
+		GatherStartCommsSimd(source,dimension,shift,0x2,u_comm_buf,u_comm_offset);// both with block stride loop iteration
+	      } else {
+		printf("NO splice 0x1,2 \n");fflush(stdout);
+		GatherStartComms(source,dimension,shift,0x1,u_comm_buf,u_comm_offset);
+		GatherStartComms(source,dimension,shift,0x2,u_comm_buf,u_comm_offset);
+	      }
+	    }
 	  }
 	}
       }
-    }
 
-
-      void Stencil::Stencil_local     (int dimension,int shift,int cbmask)
-      {
-	int fd = _grid->_fdimensions[dimension];
-	int rd = _grid->_rdimensions[dimension];
-	int ld = _grid->_ldimensions[dimension];
-	int gd = _grid->_gdimensions[dimension];
-
-	// Map to always positive shift modulo global full dimension.
-	shift = (shift+fd)%fd;
-	
-	// the permute type
-	int permute_dim =_grid->PermuteDim(dimension);
-	int permute_type=_grid->PermuteType(dimension);
-	
-	for(int x=0;x<rd;x++){       
+      template<class vobj> void GatherStartComms(Lattice<vobj> &rhs,int dimension,int shift,int cbmask,
+						 std::vector<vobj,alignedAllocator<vobj> > &u_comm_buf,
+						 int &u_comm_offset)
+	{
+	  typedef typename vobj::vector_type vector_type;
+	  typedef typename vobj::scalar_type scalar_type;
 	  
-	  int o   = 0;
-	  int bo  = x * _grid->_ostride[dimension];
+	  GridBase *grid=_grid;
+	  assert(rhs._grid==_grid);
+	  //	  conformable(_grid,rhs._grid);
+
+	  int fd              = _grid->_fdimensions[dimension];
+	  int rd              = _grid->_rdimensions[dimension];
+	  int simd_layout     = _grid->_simd_layout[dimension];
+	  int comm_dim        = _grid->_processors[dimension] >1 ;
+	  assert(simd_layout==1);
+	  assert(comm_dim==1);
+	  assert(shift>=0);
+	  assert(shift<fd);
+	  
+	  int buffer_size = _grid->_slice_nblock[dimension]*_grid->_slice_block[dimension];
+	  
+	  std::vector<vobj,alignedAllocator<vobj> > send_buf(buffer_size); // hmm...
+	  std::vector<vobj,alignedAllocator<vobj> > recv_buf(buffer_size);
 	  
 	  int cb= (cbmask==0x2)? 1 : 0;
+	  int sshift= _grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,cb);
 	  
-	  int sshift = _grid->CheckerBoardShift(_checkerboard,dimension,shift,cb);
-	  int sx     = (x+sshift)%rd;
-	  
-	  int permute_slice=0;
-	  if(permute_dim){
-	    int wrap = sshift/rd;
-	    int  num = sshift%rd;
-	    if ( x< rd-num ) permute_slice=wrap;
-	    else permute_slice = 1-wrap;
-	  }
-
-	  if ( permute_slice ) Copy_plane_permute(dimension,x,sx,cbmask,permute_type);
-	  else                 Copy_plane        (dimension,x,sx,cbmask); 
-  
-	}
-      }
-
-      void Stencil::Stencil_comms     (int dimension,int shift,int cbmask)
-      {
-	typedef typename vobj::vector_type vector_type;
-	typedef typename vobj::scalar_type scalar_type;
-	
-	GridBase *grid=_grid;
-
-	int fd              = _grid->_fdimensions[dimension];
-	int rd              = _grid->_rdimensions[dimension];
-	int simd_layout     = _grid->_simd_layout[dimension];
-	int comm_dim        = _grid->_processors[dimension] >1 ;
-
-	assert(simd_layout==1);
-	assert(comm_dim==1);
-	assert(shift>=0);
-	assert(shift<fd);
-	
-	int buffer_size = _grid->_slice_nblock[dimension]*rhs._grid->_slice_block[dimension];
-	// FIXME: Do something with buffer_size??
-
-	int cb= (cbmask==0x2)? 1 : 0;
-	int sshift= _grid->CheckerBoardShift(_checkerboard,dimension,shift,cb);
-	
-	for(int x=0;x<rd;x++){       
-	  
-	  int offnode = ( x+sshift >= rd );
-	  int sx        = (x+sshift)%rd;
-	  int comm_proc = (x+sshift)/rd;
-	  
-	  if (!offnode) {
+	  for(int x=0;x<rd;x++){       
 	    
-	    Copy_plane(dimension,x,sx,cbmask); 
+	    printf("GatherStartComms x %d/%d\n",x,rd);fflush(stdout);
+	    int offnode = ( x+sshift >= rd );
+	    int sx        = (x+sshift)%rd;
+	    int comm_proc = (x+sshift)/rd;
 	    
-	  } else {
-	    
-	    int words = send_buf.size();
-	    if (cbmask != 0x3) words=words>>1;
-	    
-	    int bytes = words * sizeof(vobj);
-	    
-	    Gather_plane_simple (dimension,sx,cbmask);
-	    
-	    int rank           = grid->_processor;
-	    int recv_from_rank;
-	    int xmit_to_rank;
-	    grid->ShiftedRanks(dimension,comm_proc,xmit_to_rank,recv_from_rank);
-	    /*	    
-	    grid->SendToRecvFrom((void *)&send_buf[0],
-				 xmit_to_rank,
-				 (void *)&recv_buf[0],
-				 recv_from_rank,
-				 bytes);
-	    */
-	    Scatter_plane_simple (dimension,x,cbmask);
-	  }
-	}
-      }
-
-      void Stencil::Stencil_comms_simd(int dimension,int shift,int cbmask)
-      {
-	GridBase *grid=_grid;
-	const int Nsimd = _grid->Nsimd();
-	typedef typename vobj::vector_type vector_type;
-	typedef typename vobj::scalar_type scalar_type;
-	
-	int fd = _grid->_fdimensions[dimension];
-	int rd = _grid->_rdimensions[dimension];
-	int ld = _grid->_ldimensions[dimension];
-	int simd_layout     = _grid->_simd_layout[dimension];
-	int comm_dim        = _grid->_processors[dimension] >1 ;
-	
-	assert(comm_dim==1);
-	assert(simd_layout==2);
-	assert(shift>=0);
-	assert(shift<fd);
-	
-	int permute_type=_grid->PermuteType(dimension);
-	
-	///////////////////////////////////////////////
-	// Simd direction uses an extract/merge pair
-	///////////////////////////////////////////////
-	int buffer_size = _grid->_slice_nblock[dimension]*grid->_slice_block[dimension];
-	// FIXME do something with buffer size
-	
-	std::vector<scalar_type *> pointers(Nsimd);  // 
-	std::vector<scalar_type *> rpointers(Nsimd); // received pointers
-	
-	///////////////////////////////////////////
-	// Work out what to send where
-	///////////////////////////////////////////
-	
-	int cb    = (cbmask==0x2)? 1 : 0;
-	int sshift= _grid->CheckerBoardShift(_checkerboard,dimension,shift,cb);
-	
-	std::vector<int> comm_offnode(simd_layout);
-	std::vector<int> comm_proc   (simd_layout);  //relative processor coord in dim=dimension
-	std::vector<int> icoor(grid->Nd());
-	
-	for(int x=0;x<rd;x++){       
-	  
-	  int comm_any = 0;
-	  for(int s=0;s<simd_layout;s++) {
-	    int shifted_x   = x+s*rd+sshift;
-	    comm_offnode[s] = shifted_x >= ld; 
-	    comm_any        = comm_any | comm_offnode[s];
-	    comm_proc[s]    = shifted_x/ld;     
-	  }
-	  
-	  int o    = 0;
-	  int bo   = x*grid->_ostride[dimension];
-	  int sx   = (x+sshift)%rd;
-	  
-	  if ( comm_any ) {
-	    
-	    for(int i=0;i<Nsimd;i++){
-	      pointers[i] = (scalar_type *)&send_buf_extract[i][0];
-	    }
-	    Gather_plane_extract(rhs,pointers,dimension,sx,cbmask);
-	    
-	    for(int i=0;i<Nsimd;i++){
+	    if (offnode) {
 	      
-	      int s;
-	      grid->iCoorFromIindex(icoor,i);
-	      s = icoor[dimension];
+	      printf("GatherStartComms offnode x %d\n",x);fflush(stdout);
+	      int words = send_buf.size();
+	      if (cbmask != 0x3) words=words>>1;
+	    
+	      int bytes = words * sizeof(vobj);
+
+	      printf("Gather_plane_simple dimension %d sx %d cbmask %d\n",dimension,sx,cbmask);fflush(stdout);
+	      Gather_plane_simple (rhs,send_buf,dimension,sx,cbmask);
+
+	      printf("GatherStartComms gathered offnode x %d\n",x);fflush(stdout);
+
+	      int rank           = _grid->_processor;
+	      int recv_from_rank;
+	      int xmit_to_rank;
+	      _grid->ShiftedRanks(dimension,comm_proc,xmit_to_rank,recv_from_rank);
 	      
-	      if(comm_offnode[s]){
-		
-		int rank           = grid->_processor;
-		int recv_from_rank;
-		int xmit_to_rank;
-		grid->ShiftedRanks(dimension,comm_proc[s],xmit_to_rank,recv_from_rank);
-		
-		/*		
-		grid->SendToRecvFrom((void *)&send_buf_extract[i][0],
-				     xmit_to_rank,
-				     (void *)&recv_buf_extract[i][0],
-				     recv_from_rank,
-				     bytes);
-		*/
-
-		rpointers[i] = (scalar_type *)&recv_buf_extract[i][0];
-		
-	      } else { 
-		
-		rpointers[i] = (scalar_type *)&send_buf_extract[i][0];
+	      //      FIXME Implement asynchronous send & also avoid buffer copy
+	      _grid->SendToRecvFrom((void *)&send_buf[0],
+				   xmit_to_rank,
+				   (void *)&recv_buf[0],
+				   recv_from_rank,
+				   bytes);
+	      printf("GatherStartComms communicated offnode x %d\n",x);fflush(stdout);
 
+	      printf("GatherStartComms inserting %d buf size %d\n",u_comm_offset,buffer_size);fflush(stdout);
+	      for(int i=0;i<buffer_size;i++){
+		u_comm_buf[u_comm_offset+i]=recv_buf[i];
 	      }
-	      
+	      u_comm_offset+=buffer_size;
+	      printf("GatherStartComms inserted x %d\n",x);fflush(stdout);
 	    }
-	    
-	    // Permute by swizzling pointers in merge
-	    int permute_slice=0;
-	    int lshift=sshift%ld;
-	    int wrap  =lshift/rd;
-	    int  num  =lshift%rd;
-
-	    if ( x< rd-num ) permute_slice=wrap;
-	    else permute_slice = 1-wrap;
-	    
-	    int toggle_bit = (Nsimd>>(permute_type+1));
-	    int PermuteMap;
-	    for(int i=0;i<Nsimd;i++){
-	      if ( permute_slice ) {
-		PermuteMap=i^toggle_bit;
-		pointers[i] = rpointers[PermuteMap];
-	      } else {
-		pointers[i] = rpointers[i];
-	      }
-	    }
-
-	    Scatter_plane_merge(pointers,dimension,x,cbmask);
-	    
-	  } else { 
-
-	    int permute_slice=0;
-	    int wrap = sshift/rd;
-	    int  num = sshift%rd;
-	    if ( x< rd-num ) permute_slice=wrap;
-	    else permute_slice = 1-wrap;
-
-	    if ( permute_slice ) Copy_plane_permute(ret,rhs,dimension,x,sx,cbmask,permute_type);
-	    else                 Copy_plane(ret,rhs,dimension,x,sx,cbmask); 
-
 	  }
 	}
-      }
 
 
-};
+      template<class vobj>
+	void  GatherStartCommsSimd(Lattice<vobj> &rhs,int dimension,int shift,int cbmask,
+				   std::vector<vobj,alignedAllocator<vobj> > &u_comm_buf,
+				   int &u_comm_offset)
+	{
+	  const int Nsimd = _grid->Nsimd();
+	  typedef typename vobj::vector_type vector_type;
+	  typedef typename vobj::scalar_type scalar_type;
+	  
+	  int fd = _grid->_fdimensions[dimension];
+	  int rd = _grid->_rdimensions[dimension];
+	  int ld = _grid->_ldimensions[dimension];
+	  int simd_layout     = _grid->_simd_layout[dimension];
+	  int comm_dim        = _grid->_processors[dimension] >1 ;
+
+	  assert(comm_dim==1);
+	  assert(simd_layout==2);
+	  assert(shift>=0);
+	  assert(shift<fd);
+
+	  int permute_type=_grid->PermuteType(dimension);
+
+	  ///////////////////////////////////////////////
+	  // Simd direction uses an extract/merge pair
+	  ///////////////////////////////////////////////
+	  int buffer_size = _grid->_slice_nblock[dimension]*_grid->_slice_block[dimension];
+	  int words = sizeof(vobj)/sizeof(vector_type);
+
+	  /*   FIXME ALTERNATE BUFFER DETERMINATION */
+	  std::vector<std::vector<scalar_type> > send_buf_extract(Nsimd,std::vector<scalar_type>(buffer_size*words) ); 
+	  std::vector<std::vector<scalar_type> > recv_buf_extract(Nsimd,std::vector<scalar_type>(buffer_size*words) );
+	  int bytes = buffer_size*words*sizeof(scalar_type);
+
+	  std::vector<scalar_type *> pointers(Nsimd);  //
+	  std::vector<scalar_type *> rpointers(Nsimd); // received pointers
+	  
+	  ///////////////////////////////////////////
+	  // Work out what to send where
+	  ///////////////////////////////////////////
+
+	  int cb    = (cbmask==0x2)? 1 : 0;
+	  int sshift= _grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,cb);
+	  
+	  std::vector<int> comm_offnode(simd_layout);
+	  std::vector<int> comm_proc   (simd_layout);  //relative processor coord in dim=dimension
+	  std::vector<int> icoor(_grid->Nd());
+	  
+	  for(int x=0;x<rd;x++){
+	    
+	    int comm_any = 0;
+	    for(int s=0;s<simd_layout;s++) {
+	      int shifted_x   = x+s*rd+sshift;
+	      comm_offnode[s] = shifted_x >= ld;
+	      comm_any        = comm_any | comm_offnode[s];
+	      comm_proc[s]    = shifted_x/ld;
+	    }
+    
+	    int o    = 0;
+	    int bo   = x*_grid->_ostride[dimension];
+	    int sx   = (x+sshift)%rd;
+	    
+	    if ( comm_any ) {
+
+	      for(int i=0;i<Nsimd;i++){
+		pointers[i] = (scalar_type *)&send_buf_extract[i][0];
+	      }
+	      Gather_plane_extract(rhs,pointers,dimension,sx,cbmask);
+	      
+	      for(int i=0;i<Nsimd;i++){
+		
+		int s;
+		_grid->iCoorFromIindex(icoor,i);
+		s = icoor[dimension];
+		
+		if(comm_offnode[s]){
+		  
+		  int rank           = _grid->_processor;
+		  int recv_from_rank;
+		  int xmit_to_rank;
+
+		  _grid->ShiftedRanks(dimension,comm_proc[s],xmit_to_rank,recv_from_rank);
+	  
+
+		  _grid->SendToRecvFrom((void *)&send_buf_extract[i][0],
+					xmit_to_rank,
+					(void *)&recv_buf_extract[i][0],
+					recv_from_rank,
+					bytes);
+
+		  rpointers[i] = (scalar_type *)&recv_buf_extract[i][0];
+		  
+		} else { 
+		  
+		  rpointers[i] = (scalar_type *)&send_buf_extract[i][0];
+
+		}
+		
+	      }
+
+	      // Permute by swizzling pointers in merge
+	      int permute_slice=0;
+	      int lshift=sshift%ld;
+	      int wrap  =lshift/rd;
+	      int  num  =lshift%rd;
+
+	      if ( x< rd-num ) permute_slice=wrap;
+	      else permute_slice = 1-wrap;
+
+	      int toggle_bit = (Nsimd>>(permute_type+1));
+	      int PermuteMap;
+	      for(int i=0;i<Nsimd;i++){
+		if ( permute_slice ) {
+		  PermuteMap=i^toggle_bit;
+		  pointers[i] = rpointers[PermuteMap];
+		} else {
+		  pointers[i] = rpointers[i];
+		}
+	      }
+
+	      // Here we don't want to scatter, just place into a buffer.
+	      for(int i=0;i<buffer_size;i++){
+		merge(u_comm_buf[u_comm_offset+i],pointers);
+	      }
+
+	    }
+	  }
+	}
+  };
+}
 #endif
diff --git a/Grid_stencil_common.cc b/Grid_stencil_common.cc
new file mode 100644
index 00000000..01bdce5d
--- /dev/null
+++ b/Grid_stencil_common.cc
@@ -0,0 +1,258 @@
+#include "Grid.h"
+
+namespace Grid {
+
+  CartesianStencil::CartesianStencil(GridBase *grid,
+				     int npoints,
+				     int checkerboard,
+				     const std::vector<int> &directions,
+				     const std::vector<int> &distances) 
+    :   _offsets(npoints), 
+	_is_local(npoints), 
+	_comm_buf_size(npoints), 
+	_permute_type(npoints),
+	_permute(npoints)
+    {
+      _npoints = npoints;
+      _grid    = grid;
+      _directions = directions;
+      _distances  = distances;
+      _unified_buffer_size=0;
+      _request_count =0;
+      CommsRequests.resize(0);
+
+      int osites  = _grid->oSites();
+
+      for(int i=0;i<npoints;i++){
+
+	int point = i;
+
+	_offsets[i].resize( osites);
+	_is_local[i].resize(osites);
+	_permute[i].resize( osites);
+
+	int dimension    = directions[i];
+	int displacement = distances[i];
+	int shift = displacement;
+	
+	int fd = _grid->_fdimensions[dimension];
+	int rd = _grid->_rdimensions[dimension];
+	_permute_type[point]=_grid->PermuteType(dimension);
+
+	_checkerboard = checkerboard;
+
+	// the permute type
+	int simd_layout     = _grid->_simd_layout[dimension];
+	int comm_dim        = _grid->_processors[dimension] >1 ;
+	int splice_dim      = _grid->_simd_layout[dimension]>1 && (comm_dim);
+
+	int sshift[2];
+	
+	// Underlying approach. For each local site build
+	// up a table containing the npoint "neighbours" and whether they 
+	// live in lattice or a comms buffer.
+	if ( !comm_dim ) {
+	  sshift[0] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,0);
+	  sshift[1] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,1);
+
+	  if ( sshift[0] == sshift[1] ) {
+	    Local(point,dimension,shift,0x3);
+	  } else {
+	    Local(point,dimension,shift,0x1);// if checkerboard is unfavourable take two passes
+	    Local(point,dimension,shift,0x2);// both with block stride loop iteration
+	  }
+	} else { // All permute extract done in comms phase prior to Stencil application
+	  //        So tables are the same whether comm_dim or splice_dim
+	  sshift[0] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,0);
+	  sshift[1] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,1);
+	  if ( sshift[0] == sshift[1] ) {
+	    Comms(point,dimension,shift,0x3);
+	  } else {
+	    Comms(point,dimension,shift,0x1);// if checkerboard is unfavourable take two passes
+	    Comms(point,dimension,shift,0x2);// both with block stride loop iteration
+	  }
+	}
+      }
+    }
+
+
+    void CartesianStencil::Local     (int point, int dimension,int shift,int cbmask)
+    {
+      int fd = _grid->_fdimensions[dimension];
+      int rd = _grid->_rdimensions[dimension];
+      int ld = _grid->_ldimensions[dimension];
+      int gd = _grid->_gdimensions[dimension];
+      
+      // Map to always positive shift modulo global full dimension.
+      shift = (shift+fd)%fd;
+      
+      // the permute type
+      int permute_dim =_grid->PermuteDim(dimension);
+      
+      for(int x=0;x<rd;x++){       
+	
+	int o   = 0;
+	int bo  = x * _grid->_ostride[dimension];
+	
+	int cb= (cbmask==0x2)? 1 : 0;
+	  
+	int sshift = _grid->CheckerBoardShift(_checkerboard,dimension,shift,cb);
+	int sx     = (x+sshift)%rd;
+	  
+	int permute_slice=0;
+	if(permute_dim){
+	  int wrap = sshift/rd;
+	  int  num = sshift%rd;
+	  if ( x< rd-num ) permute_slice=wrap;
+	  else permute_slice = 1-wrap;
+	}
+
+  	CopyPlane(point,dimension,x,sx,cbmask,permute_slice);
+  
+      }
+    }
+
+    void CartesianStencil::Comms     (int point,int dimension,int shift,int cbmask)
+    {
+      GridBase *grid=_grid;
+      
+      int fd              = _grid->_fdimensions[dimension];
+      int rd              = _grid->_rdimensions[dimension];
+      int simd_layout     = _grid->_simd_layout[dimension];
+      int comm_dim        = _grid->_processors[dimension] >1 ;
+      
+      assert(simd_layout==1);
+      assert(comm_dim==1);
+      assert(shift>=0);
+      assert(shift<fd);
+      
+      int buffer_size = _grid->_slice_nblock[dimension]*_grid->_slice_block[dimension];
+      _comm_buf_size[point] = buffer_size; // Size of _one_ plane. Multiple planes may be gathered and
+                                           // send to one or more remote nodes.
+
+      int cb= (cbmask==0x2)? 1 : 0;
+      int sshift= _grid->CheckerBoardShift(_checkerboard,dimension,shift,cb);
+      
+      for(int x=0;x<rd;x++){       
+	
+	int offnode = ( x+sshift >= rd );
+	int sx        = (x+sshift)%rd;
+	int comm_proc = (x+sshift)/rd;
+	  
+	if (!offnode) {
+	  
+	  int permute_slice=0;
+	  CopyPlane(point,dimension,x,sx,cbmask,permute_slice); 
+	  
+	} else {
+	  
+	  int words = buffer_size;
+	  if (cbmask != 0x3) words=words>>1;
+	  
+	  //	  GatherPlaneSimple (point,dimension,sx,cbmask);
+	  
+	  int rank           = grid->_processor;
+	  int recv_from_rank;
+	  int xmit_to_rank;
+
+	  CommsRequest cr;
+
+	  cr.tag                   = _request_count++;
+	  cr.words                 = words;
+	  cr.unified_buffer_offset = _unified_buffer_size;
+	  _unified_buffer_size    += words;
+	  grid->ShiftedRanks(dimension,comm_proc,cr.to_rank,cr.from_rank);
+
+	  CommsRequests.push_back(cr);
+
+	  ScatterPlane(point,dimension,x,cbmask,cr.unified_buffer_offset); // permute/extract/merge is done in comms phase
+	  
+	}
+      }
+    }
+  // Routine builds up integer table for each site in _offsets, _is_local, _permute
+  void CartesianStencil::CopyPlane(int point, int dimension,int lplane,int rplane,int cbmask,int permute)
+    {
+      int rd = _grid->_rdimensions[dimension];
+      
+      if ( !_grid->CheckerBoarded(dimension) ) {
+	
+	int o   = 0;                                     // relative offset to base within plane
+	int ro  = rplane*_grid->_ostride[dimension]; // base offset for start of plane 
+	int lo  = lplane*_grid->_ostride[dimension]; // offset in buffer
+	
+	// Simple block stride gather of SIMD objects
+	for(int n=0;n<_grid->_slice_nblock[dimension];n++){
+	  for(int b=0;b<_grid->_slice_block[dimension];b++){
+	    _offsets [point][lo+o+b]=ro+o+b;
+	    _is_local[point][lo+o+b]=1;
+	    _permute [point][lo+o+b]=permute;
+	  }
+	  o +=_grid->_slice_stride[dimension];
+	}
+	
+      } else {
+	
+	int ro  = rplane*_grid->_ostride[dimension]; // base offset for start of plane 
+	int lo  = lplane*_grid->_ostride[dimension]; // base offset for start of plane 
+	int o   = 0;                                     // relative offset to base within plane
+	
+	for(int n=0;n<_grid->_slice_nblock[dimension];n++){
+	  for(int b=0;b<_grid->_slice_block[dimension];b++){
+	    
+	    int ocb=1<<_grid->CheckerBoardFromOindex(o+b);
+	    
+	    if ( ocb&cbmask ) {
+	      _offsets [point][lo+o+b]=ro+o+b;
+	      _is_local[point][lo+o+b]=1;
+	      _permute [point][lo+o+b]=permute;
+	    }
+	    
+	    }
+	  o +=_grid->_slice_stride[dimension];
+	}
+	
+      }
+    }
+  // Routine builds up integer table for each site in _offsets, _is_local, _permute
+    void CartesianStencil::ScatterPlane (int point,int dimension,int plane,int cbmask,int offset)
+    {
+      int rd = _grid->_rdimensions[dimension];
+      
+      if ( !_grid->CheckerBoarded(dimension) ) {
+	
+	int so  = plane*_grid->_ostride[dimension]; // base offset for start of plane 
+	int o   = 0;                                    // relative offset to base within plane
+	int bo  = 0;                                    // offset in buffer
+
+	// Simple block stride gather of SIMD objects
+	for(int n=0;n<_grid->_slice_nblock[dimension];n++){
+	  for(int b=0;b<_grid->_slice_block[dimension];b++){
+	    _offsets [point][so+o+b]=offset+(bo++);
+	    _is_local[point][so+o+b]=0;
+	    _permute [point][so+o+b]=0;
+	  }
+	  o +=_grid->_slice_stride[dimension];
+	}
+
+      } else { 
+	
+	int so  = plane*_grid->_ostride[dimension]; // base offset for start of plane 
+	int o   = 0;                                      // relative offset to base within plane
+	int bo  = 0;                                      // offset in buffer
+    
+	for(int n=0;n<_grid->_slice_nblock[dimension];n++){
+	  for(int b=0;b<_grid->_slice_block[dimension];b++){
+
+	    int ocb=1<<_grid->CheckerBoardFromOindex(o+b);// Could easily be a table lookup
+	    if ( ocb & cbmask ) {
+	      _offsets [point][so+o+b]=offset+(bo++);
+	      _is_local[point][so+o+b]=0;
+	      _permute [point][so+o+b]=0;
+	    }
+	  }
+	  o +=_grid->_slice_stride[dimension];
+	}
+      }
+    }
+}
diff --git a/Grid_vComplexD.h b/Grid_vComplexD.h
index 3278b15e..f1c335e8 100644
--- a/Grid_vComplexD.h
+++ b/Grid_vComplexD.h
@@ -194,6 +194,8 @@ namespace Grid {
             float b= imag(c);
             vsplat(ret,a,b);
         }
+
+
         friend inline void vsplat(vComplexD &ret,double rl,double ig){
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_set_pd(ig,rl,ig,rl);
@@ -321,13 +323,14 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
     {
         return l*r;
     }
-    inline vComplex trace(const vComplex &arg){
+    inline vComplexD trace(const vComplexD &arg){
         return arg;
     }
 /////////////////////////////////////////////////////////////////////////
 //// Generic routine to promote object<complex> -> object<vcomplex>
 //// Supports the array reordering transformation that gives me SIMD utilisation
 ///////////////////////////////////////////////////////////////////////////
+/*
 template<template<class> class object>
 inline object<vComplex> splat(object<Complex >s){
       object<vComplex> ret;
@@ -338,6 +341,6 @@ inline object<vComplex> splat(object<Complex >s){
       }
       return ret;
     }
-
+*/
 }
 #endif
diff --git a/Grid_vComplexF.h b/Grid_vComplexF.h
index b6b7ebe9..18cf1749 100644
--- a/Grid_vComplexF.h
+++ b/Grid_vComplexF.h
@@ -3,6 +3,16 @@
 #include "Grid.h"
 
 namespace Grid {
+
+  /*
+  inline void Print(const char *A,cvec c) { 
+	  float *fp=(float *)&c; 
+	  printf(A); 
+	  printf(" %le %le %le %le %le %le %le %le\n",
+		 fp[0],fp[1],fp[2],fp[3],fp[4],fp[5],fp[6],fp[7]);
+	}
+  */
+
     class vComplexF {
       //    protected:
 
@@ -108,7 +118,7 @@ namespace Grid {
             ymm1 = _mm256_shuffle_ps(b.v,b.v,_MM_SHUFFLE(2,3,0,1)); // ymm1 <- br,bi
             ymm2 = _mm256_shuffle_ps(a.v,a.v,_MM_SHUFFLE(3,3,1,1)); // ymm2 <- ai,ai
             ymm1 = _mm256_mul_ps(ymm1,ymm2);       // ymm1 <- br ai, ai bi
-            ret.v= _mm256_addsub_ps(ymm0,ymm1);    // FIXME -- AVX2 could MAC
+            ret.v= _mm256_addsub_ps(ymm0,ymm1);    
 #endif
 #ifdef SSE2
             cvec ymm0,ymm1,ymm2;
@@ -142,7 +152,7 @@ namespace Grid {
 	////////////////////////////////////////////////////////////////////////
 	// FIXME:  gonna remove these load/store, get, set, prefetch
 	////////////////////////////////////////////////////////////////////////
-        friend inline void vset(vComplexF &ret, Complex *a){
+        friend inline void vset(vComplexF &ret, ComplexF *a){
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_set_ps(a[3].imag(),a[3].real(),a[2].imag(),a[2].real(),a[1].imag(),a[1].real(),a[0].imag(),a[0].real());
 #endif
@@ -216,12 +226,12 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
 #endif
         }
 
-        friend inline vComplexF operator * (const Complex &a, vComplexF b){
+        friend inline vComplexF operator * (const ComplexF &a, vComplexF b){
             vComplexF va;
             vsplat(va,a);
             return va*b;
         }
-        friend inline vComplexF operator * (vComplexF b,const Complex &a){
+        friend inline vComplexF operator * (vComplexF b,const ComplexF &a){
 	  return a*b;
         }
 
@@ -283,22 +293,11 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
 	}
        */
        
-        // NB: Template the following on "type Complex" and then implement *,+,- for 
-	// ComplexF, ComplexD, RealF, RealD above to
-        // get full generality of binops with scalars.
-        friend inline void mac (vComplexF *__restrict__ y,const Complex *__restrict__ a,const vComplexF *__restrict__ x){ *y = (*a)*(*x)+(*y); };
-        friend inline void mult(vComplexF *__restrict__ y,const Complex *__restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) * (*r); }
-        friend inline void sub (vComplexF *__restrict__ y,const Complex *__restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) - (*r); }
-        friend inline void add (vComplexF *__restrict__ y,const Complex *__restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) + (*r); }
-        friend inline void mac (vComplexF *__restrict__ y,const vComplexF *__restrict__ a,const Complex *__restrict__ x){ *y = (*a)*(*x)+(*y); };
-        friend inline void mult(vComplexF *__restrict__ y,const vComplexF *__restrict__ l,const Complex *__restrict__ r){ *y = (*l) * (*r); }
-        friend inline void sub (vComplexF *__restrict__ y,const vComplexF *__restrict__ l,const Complex *__restrict__ r){ *y = (*l) - (*r); }
-        friend inline void add (vComplexF *__restrict__ y,const vComplexF *__restrict__ l,const Complex *__restrict__ r){ *y = (*l) + (*r); }
-
 
         ///////////////////////
         // Conjugate
         ///////////////////////
+								     
         friend inline vComplexF conj(const vComplexF &in){
             vComplexF ret ; vzero(ret);
 #if defined (AVX1)|| defined (AVX2)
@@ -363,10 +362,11 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
 
     };
 
-    inline vComplexF localInnerProduct(const vComplexF & l, const vComplexF & r) { return conj(l)*r; }
+    inline vComplexF localInnerProduct(const vComplexF & l, const vComplexF & r) 
+    {
+      return conj(l)*r; 
+    }
 
-    typedef  vComplexF vFComplex;
-    typedef  vComplexF vComplex;
     inline void zeroit(vComplexF &z){ vzero(z);}
 
     inline vComplexF outerProduct(const vComplexF &l, const vComplexF& r)
diff --git a/Makefile.am b/Makefile.am
index d11f28d5..e306eb56 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -1,11 +1,22 @@
 # additional include paths necessary to compile the C++ library
 AM_CXXFLAGS = -I$(top_srcdir)/
 
+
+extra_sources=
+if BUILD_COMMS_MPI
+  extra_sources+=Grid_communicator_mpi.cc
+  extra_sources+=Grid_stencil_common.cc
+endif
+if BUILD_COMMS_NONE
+  extra_sources+=Grid_communicator_fake.cc
+  extra_sources+=Grid_stencil_common.cc
+endif
+
 #
 # Libraries
 #
 lib_LIBRARIES = libGrid.a
-libGrid_a_SOURCES = Grid_init.cc
+libGrid_a_SOURCES = Grid_init.cc $(extra_sources)
 
 #
 # Include files
@@ -26,24 +37,18 @@ include_HEADERS = Grid_config.h\
 	Grid_cshift_common.h\
 	Grid_cshift_mpi.h\
 	Grid_cshift_none.h\
+	Grid_stencil.h\
 	Grid_math_types.h
 
 
 #
 # Test code
 #
-bin_PROGRAMS = Grid_main
-
-extra_sources=
-if BUILD_COMMS_MPI
-  extra_sources+=Grid_communicator_mpi.cc
-endif
-if BUILD_COMMS_NONE
-  extra_sources+=Grid_communicator_fake.cc
-endif
-
-Grid_main_SOURCES = \
-	Grid_main.cc\
-	$(extra_sources)
+bin_PROGRAMS = Grid_main test_Grid_stencil
 
+Grid_main_SOURCES = Grid_main.cc
 Grid_main_LDADD = libGrid.a
+
+test_Grid_stencil_SOURCES = test_Grid_Stencil.cc
+test_Grid_stencil_LDADD = libGrid.a
+
diff --git a/Makefile.in b/Makefile.in
index 16981e52..11b86688 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -88,9 +88,11 @@ POST_INSTALL = :
 NORMAL_UNINSTALL = :
 PRE_UNINSTALL = :
 POST_UNINSTALL = :
-bin_PROGRAMS = Grid_main$(EXEEXT)
-@BUILD_COMMS_MPI_TRUE@am__append_1 = Grid_communicator_mpi.cc
-@BUILD_COMMS_NONE_TRUE@am__append_2 = Grid_communicator_fake.cc
+@BUILD_COMMS_MPI_TRUE@am__append_1 = Grid_communicator_mpi.cc \
+@BUILD_COMMS_MPI_TRUE@	Grid_stencil_common.cc
+@BUILD_COMMS_NONE_TRUE@am__append_2 = Grid_communicator_fake.cc \
+@BUILD_COMMS_NONE_TRUE@	Grid_stencil_common.cc
+bin_PROGRAMS = Grid_main$(EXEEXT) test_Grid_stencil$(EXEEXT)
 subdir = .
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/configure.ac
@@ -142,18 +144,23 @@ am__v_AR_0 = @echo "  AR      " $@;
 am__v_AR_1 = 
 libGrid_a_AR = $(AR) $(ARFLAGS)
 libGrid_a_LIBADD =
-am_libGrid_a_OBJECTS = Grid_init.$(OBJEXT)
+am__libGrid_a_SOURCES_DIST = Grid_init.cc Grid_communicator_mpi.cc \
+	Grid_stencil_common.cc Grid_communicator_fake.cc
+@BUILD_COMMS_MPI_TRUE@am__objects_1 = Grid_communicator_mpi.$(OBJEXT) \
+@BUILD_COMMS_MPI_TRUE@	Grid_stencil_common.$(OBJEXT)
+@BUILD_COMMS_NONE_TRUE@am__objects_2 =  \
+@BUILD_COMMS_NONE_TRUE@	Grid_communicator_fake.$(OBJEXT) \
+@BUILD_COMMS_NONE_TRUE@	Grid_stencil_common.$(OBJEXT)
+am__objects_3 = $(am__objects_1) $(am__objects_2)
+am_libGrid_a_OBJECTS = Grid_init.$(OBJEXT) $(am__objects_3)
 libGrid_a_OBJECTS = $(am_libGrid_a_OBJECTS)
 PROGRAMS = $(bin_PROGRAMS)
-am__Grid_main_SOURCES_DIST = Grid_main.cc Grid_communicator_mpi.cc \
-	Grid_communicator_fake.cc
-@BUILD_COMMS_MPI_TRUE@am__objects_1 = Grid_communicator_mpi.$(OBJEXT)
-@BUILD_COMMS_NONE_TRUE@am__objects_2 =  \
-@BUILD_COMMS_NONE_TRUE@	Grid_communicator_fake.$(OBJEXT)
-am__objects_3 = $(am__objects_1) $(am__objects_2)
-am_Grid_main_OBJECTS = Grid_main.$(OBJEXT) $(am__objects_3)
+am_Grid_main_OBJECTS = Grid_main.$(OBJEXT)
 Grid_main_OBJECTS = $(am_Grid_main_OBJECTS)
 Grid_main_DEPENDENCIES = libGrid.a
+am_test_Grid_stencil_OBJECTS = test_Grid_Stencil.$(OBJEXT)
+test_Grid_stencil_OBJECTS = $(am_test_Grid_stencil_OBJECTS)
+test_Grid_stencil_DEPENDENCIES = libGrid.a
 AM_V_P = $(am__v_P_@AM_V@)
 am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
 am__v_P_0 = false
@@ -183,8 +190,10 @@ AM_V_CXXLD = $(am__v_CXXLD_@AM_V@)
 am__v_CXXLD_ = $(am__v_CXXLD_@AM_DEFAULT_V@)
 am__v_CXXLD_0 = @echo "  CXXLD   " $@;
 am__v_CXXLD_1 = 
-SOURCES = $(libGrid_a_SOURCES) $(Grid_main_SOURCES)
-DIST_SOURCES = $(libGrid_a_SOURCES) $(am__Grid_main_SOURCES_DIST)
+SOURCES = $(libGrid_a_SOURCES) $(Grid_main_SOURCES) \
+	$(test_Grid_stencil_SOURCES)
+DIST_SOURCES = $(am__libGrid_a_SOURCES_DIST) $(Grid_main_SOURCES) \
+	$(test_Grid_stencil_SOURCES)
 am__can_run_installinfo = \
   case $$AM_UPDATE_INFO_DIR in \
     n|no|NO) false;; \
@@ -329,12 +338,13 @@ top_srcdir = @top_srcdir@
 
 # additional include paths necessary to compile the C++ library
 AM_CXXFLAGS = -I$(top_srcdir)/
+extra_sources = $(am__append_1) $(am__append_2)
 
 #
 # Libraries
 #
 lib_LIBRARIES = libGrid.a
-libGrid_a_SOURCES = Grid_init.cc
+libGrid_a_SOURCES = Grid_init.cc $(extra_sources)
 
 #
 # Include files
@@ -355,14 +365,13 @@ include_HEADERS = Grid_config.h\
 	Grid_cshift_common.h\
 	Grid_cshift_mpi.h\
 	Grid_cshift_none.h\
+	Grid_stencil.h\
 	Grid_math_types.h
 
-extra_sources = $(am__append_1) $(am__append_2)
-Grid_main_SOURCES = \
-	Grid_main.cc\
-	$(extra_sources)
-
+Grid_main_SOURCES = Grid_main.cc
 Grid_main_LDADD = libGrid.a
+test_Grid_stencil_SOURCES = test_Grid_Stencil.cc
+test_Grid_stencil_LDADD = libGrid.a
 all: Grid_config.h
 	$(MAKE) $(AM_MAKEFLAGS) all-am
 
@@ -499,6 +508,10 @@ Grid_main$(EXEEXT): $(Grid_main_OBJECTS) $(Grid_main_DEPENDENCIES) $(EXTRA_Grid_
 	@rm -f Grid_main$(EXEEXT)
 	$(AM_V_CXXLD)$(CXXLINK) $(Grid_main_OBJECTS) $(Grid_main_LDADD) $(LIBS)
 
+test_Grid_stencil$(EXEEXT): $(test_Grid_stencil_OBJECTS) $(test_Grid_stencil_DEPENDENCIES) $(EXTRA_test_Grid_stencil_DEPENDENCIES) 
+	@rm -f test_Grid_stencil$(EXEEXT)
+	$(AM_V_CXXLD)$(CXXLINK) $(test_Grid_stencil_OBJECTS) $(test_Grid_stencil_LDADD) $(LIBS)
+
 mostlyclean-compile:
 	-rm -f *.$(OBJEXT)
 
@@ -509,6 +522,8 @@ distclean-compile:
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_communicator_mpi.Po@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_init.Po@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_main.Po@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_stencil_common.Po@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/test_Grid_Stencil.Po@am__quote@
 
 .cc.o:
 @am__fastdepCXX_TRUE@	$(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
diff --git a/TODO b/TODO
index 5297e213..665e3a55 100644
--- a/TODO
+++ b/TODO
@@ -3,7 +3,7 @@
 * Replace vset with a call to merge.; 
 * care in Gmerge,Gextract over set.
 * Const audit
-* extract / merge extra implementation removal
+* extract / merge extra implementation removal      
 
 
 * Conditional execution, where etc...                -----DONE, simple test
diff --git a/test_Grid_jacobi.cc b/test_Grid_jacobi.cc
new file mode 100644
index 00000000..95e808f1
--- /dev/null
+++ b/test_Grid_jacobi.cc
@@ -0,0 +1,206 @@
+#include "Grid.h"
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class vobj>
+class LinearOperator {
+public:
+  operator () (const Lattice<vobj>&src,Lattice<vobj> &result) {};
+};
+
+template<class vobj>
+class LinearOperatorJacobi : public LinearOperator<vobj>
+{
+  CartesianStencil *Stencil;
+  GridBase *_grid;
+  std::vector<vobj,alignedAllocator<vobj> >  comm_buf;
+  LinearOperatorJacobi(GridCartesian *grid)    
+  {
+    _grid = grid;
+    int npoint=9;
+    std::vector<int> directions(npoint);
+    std::vector<int> displacements(npoint);
+    for(int mu=0;mu<4;mu++){
+      for(int mp=0;mp<2;mp++){
+	int dir = 2*mu+mp;
+	directions[dir]   = mu;
+	displacements[dir]= -1+2*mp;
+      }
+    }
+    directions[8] = 0;
+    displacements[8] = 0;
+
+    Stencil = new CartesianStencil(grid,npoint,0,directions,displacements);  
+
+    comm_buf.resize(Stencil->_unified_buffer_size);
+
+  }
+  operator () (const Lattice<vobj>&src,Lattice<vobj> &result)
+  {
+    const int npoint=9;
+    printf("calling halo exchange\n");fflush(stdout);
+    myStencil.HaloExchange(Foo,comm_buf);
+    vobj tmp;
+    vobj
+    for(int i=0;i<_grid->oSites();i++){
+      for(int p=0;p<npoint;p++){
+
+	int offset = Stencil->_offsets [p][i];
+	int  local = Stencil->_is_local[p][i];
+	int ptype  = Stencil->_permute_type[p];
+	int perm   = Stencil->_permute[0][i];
+	
+	vobj *nbr;
+	if ( local && perm ){
+	  permute(tmp,src._odata[offset],ptype);
+	  nbr = &tmp;
+	} else if (local) {
+	  nbr = &src._odata[offset];
+	} else  {
+	  nbr = &comm_buf[offset];
+	}
+       	result[i] = result[i]+*nbr;
+      }
+    }	
+  }
+  ~LinearOperatorJacobi()
+  {
+    delete Stencil;
+  }
+}
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> latt_size  (4);
+  std::vector<int> simd_layout(4);
+  std::vector<int> mpi_layout (4);
+
+  int omp=1;
+  int lat=8;
+
+  mpi_layout[0]=1;
+  mpi_layout[1]=2;
+  mpi_layout[2]=1;
+  mpi_layout[3]=1;
+
+    latt_size[0] = lat;
+    latt_size[1] = lat;
+    latt_size[2] = lat;
+    latt_size[3] = lat;
+    double volume = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
+    
+#ifdef AVX512
+    simd_layout[0] = 1;
+    simd_layout[1] = 2;
+    simd_layout[2] = 2;
+    simd_layout[3] = 2;
+#endif
+#if defined (AVX1)|| defined (AVX2)
+    simd_layout[0] = 1;
+    simd_layout[1] = 1;
+    simd_layout[2] = 2;
+    simd_layout[3] = 2;
+#endif
+#if defined (SSE2)
+    simd_layout[0] = 1;
+    simd_layout[1] = 1;
+    simd_layout[2] = 1;
+    simd_layout[3] = 2;
+#endif
+    
+    GridCartesian Fine(latt_size,simd_layout,mpi_layout);
+    GridRedBlackCartesian rbFine(latt_size,simd_layout,mpi_layout);
+    
+    LatticeColourMatrix Foo(&Fine);
+    LatticeColourMatrix Bar(&Fine);
+    LatticeColourMatrix Check(&Fine);
+    LatticeColourMatrix Diff(&Fine);
+    
+    random(Foo);
+    gaussian(Bar);
+
+
+    for(int dir=0;dir<4;dir++){
+      for(int disp=0;disp<Fine._rdimensions[dir];disp++){
+
+	// start to test the Cartesian npoint stencil infrastructure
+	int npoint=;
+	std::vector<int> directions(npoint,dir);
+	std::vector<int> displacements(npoint,disp);
+
+	CartesianStencil myStencil(&Fine,npoint,0,directions,displacements);
+
+	printf("STENCIL: osites %d %d dir %d disp %d\n",Fine.oSites(),(int)myStencil._offsets[0].size(),dir,disp);
+	std::vector<int> ocoor(4);
+	for(int o=0;o<Fine.oSites();o++){
+	  Fine.oCoorFromOindex(ocoor,o);
+	  ocoor[dir]=(ocoor[dir]+disp)%Fine._rdimensions[dir];
+	  int nbr = Fine.oIndexReduced(ocoor);
+	  int stcl= myStencil._offsets[0][o];
+	  if(nbr!=stcl){
+	    printf("STENCIL: nbr %d stencil._offset %d\n",nbr,stcl);
+	  }
+	}
+	
+	printf("allocating %d buffers\n",myStencil._unified_buffer_size);
+	fflush(stdout);
+	std::vector<vColourMatrix,alignedAllocator<vColourMatrix> >  comm_buf(myStencil._unified_buffer_size);
+	printf("calling halo exchange\n");fflush(stdout);
+	myStencil.HaloExchange(Foo,comm_buf);
+
+	Bar = Cshift(Foo,dir,disp);
+
+	// Implement a stencil code that should agree with cshift!
+	for(int i=0;i<Check._grid->oSites();i++){
+
+	  int offset = myStencil._offsets [0][i];
+	  int  local = myStencil._is_local[0][i];
+	  int permute_type = myStencil._permute_type[0];
+	  int perm =myStencil._permute[0][i];
+	  if ( local && perm )
+	    permute(Check._odata[i],Foo._odata[offset],permute_type);
+	  else if (local)
+	    Check._odata[i] = Foo._odata[offset];
+	  else 
+	    Check._odata[i] = comm_buf[offset];
+	  
+
+	}
+
+
+	std::vector<int> coor(4);
+	for(coor[3]=0;coor[3]<latt_size[3]/mpi_layout[3];coor[3]++){
+	for(coor[2]=0;coor[2]<latt_size[2]/mpi_layout[2];coor[2]++){
+	for(coor[1]=0;coor[1]<latt_size[1]/mpi_layout[1];coor[1]++){
+	for(coor[0]=0;coor[0]<latt_size[0]/mpi_layout[0];coor[0]++){
+
+	  Complex diff;
+	  ColourMatrix check,bar;
+	  peekSite(check,Check,coor);
+	  peekSite(bar,Bar,coor);
+
+	  for(int r=0;r<3;r++){
+	  for(int c=0;c<3;c++){
+            diff =check._internal._internal[r][c]-bar._internal._internal[r][c];
+            double nn=real(conj(diff)*diff);
+            if ( nn > 0 ){
+	      printf("Coor (%d %d %d %d) \t rc %d%d \t %le %le %le\n",
+		     coor[0],coor[1],coor[2],coor[3],r,c,
+		     nn,
+		     real(check._internal._internal[r][c]),
+		     real(bar._internal._internal[r][c])
+		     );
+	    }
+	  }}
+	 
+	}}}}
+
+      }
+    }
+
+ Grid_finalize();
+}
diff --git a/test_Grid_stencil.cc b/test_Grid_stencil.cc
new file mode 100644
index 00000000..8077562f
--- /dev/null
+++ b/test_Grid_stencil.cc
@@ -0,0 +1,154 @@
+#include "Grid.h"
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> latt_size  (4);
+  std::vector<int> simd_layout(4);
+  std::vector<int> mpi_layout (4);
+
+  int omp=1;
+  int lat=8;
+
+  mpi_layout[0]=1;
+  mpi_layout[1]=1;
+  mpi_layout[2]=1;
+  mpi_layout[3]=1;
+
+    latt_size[0] = lat;
+    latt_size[1] = lat;
+    latt_size[2] = lat;
+    latt_size[3] = lat;
+    double volume = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
+    
+#ifdef AVX512
+    simd_layout[0] = 1;
+    simd_layout[1] = 1;
+    simd_layout[2] = 2;
+    simd_layout[3] = 2;
+#endif
+#if defined (AVX1)|| defined (AVX2)
+    simd_layout[0] = 1;
+    simd_layout[1] = 1;
+    simd_layout[2] = 1;
+    simd_layout[3] = 2;
+#endif
+#if defined (SSE2)
+    simd_layout[0] = 1;
+    simd_layout[1] = 1;
+    simd_layout[2] = 1;
+    simd_layout[3] = 2;
+#endif
+    
+    GridCartesian Fine(latt_size,simd_layout,mpi_layout);
+    GridRedBlackCartesian rbFine(latt_size,simd_layout,mpi_layout);
+    
+    LatticeColourMatrix Foo(&Fine);
+    LatticeColourMatrix Bar(&Fine);
+    LatticeColourMatrix Check(&Fine);
+    LatticeColourMatrix Diff(&Fine);
+    
+    random(Foo);
+    gaussian(Bar);
+
+
+    for(int dir=0;dir<4;dir++){
+      for(int disp=0;disp<Fine._rdimensions[dir];disp++){
+
+	// start to test the Cartesian npoint stencil infrastructure
+	int npoint=1;
+	std::vector<int> directions(npoint,dir);
+	std::vector<int> displacements(npoint,disp);
+
+	CartesianStencil myStencil(&Fine,npoint,0,directions,displacements);
+
+	printf("STENCIL: osites %d %d dir %d disp %d\n",Fine.oSites(),(int)myStencil._offsets[0].size(),dir,disp);
+	std::vector<int> ocoor(4);
+	for(int o=0;o<Fine.oSites();o++){
+	  Fine.oCoorFromOindex(ocoor,o);
+	  ocoor[dir]=(ocoor[dir]+disp)%Fine._rdimensions[dir];
+	  int nbr = Fine.oIndexReduced(ocoor);
+	  int stcl= myStencil._offsets[0][o];
+	  if(nbr!=stcl){
+	    printf("STENCIL: nbr %d stencil._offset %d\n",nbr,stcl);
+	  }
+	}
+	
+	printf("allocating %d buffers\n",myStencil._unified_buffer_size);
+	fflush(stdout);
+	std::vector<vColourMatrix,alignedAllocator<vColourMatrix> >  comm_buf(myStencil._unified_buffer_size);
+	printf("calling halo exchange\n");fflush(stdout);
+	myStencil.HaloExchange(Foo,comm_buf);
+
+	Bar = Cshift(Foo,dir,disp);
+
+	// Implement a stencil code that should agree with cshift!
+	for(int i=0;i<Check._grid->oSites();i++){
+
+	  int offset = myStencil._offsets [0][i];
+	  int  local = myStencil._is_local[0][i];
+	  int permute_type = myStencil._permute_type[0];
+	  int perm =myStencil._permute[0][i];
+	  if ( local && perm )
+	    permute(Check._odata[i],Foo._odata[offset],permute_type);
+	  else if (local)
+	    Check._odata[i] = Foo._odata[offset];
+	  else 
+	    Check._odata[i] = comm_buf[offset];
+	  
+
+	}
+
+	Real nrmC = norm2(Check);
+	Real nrmB = norm2(Bar);
+	Real nrm = norm2(Check-Bar);
+	printf("N2diff = %le (%le, %le) \n",nrm,nrmC,nrmB);fflush(stdout);
+
+	Real snrmC =0;
+	Real snrmB =0;
+	Real snrm  =0;
+
+	std::vector<int> coor(4);
+	for(coor[3]=0;coor[3]<latt_size[3]/mpi_layout[3];coor[3]++){
+	for(coor[2]=0;coor[2]<latt_size[2]/mpi_layout[2];coor[2]++){
+	for(coor[1]=0;coor[1]<latt_size[1]/mpi_layout[1];coor[1]++){
+	for(coor[0]=0;coor[0]<latt_size[0]/mpi_layout[0];coor[0]++){
+
+	  Complex diff;
+	  ColourMatrix check,bar;
+	  peekSite(check,Check,coor);
+	  peekSite(bar,Bar,coor);
+
+	  for(int r=0;r<3;r++){
+	  for(int c=0;c<3;c++){
+            diff =check._internal._internal[r][c]-bar._internal._internal[r][c];
+            double nn=real(conj(diff)*diff);
+            if ( nn > 0){
+	      printf("Coor (%d %d %d %d) \t rc %d%d \t %le %le %le\n",
+		     coor[0],coor[1],coor[2],coor[3],r,c,
+		     nn,
+		     real(check._internal._internal[r][c]),
+		     real(bar._internal._internal[r][c])
+		     );
+	    }
+	    snrmC=snrmC+real(conj(check._internal._internal[r][c])*check._internal._internal[r][c]);
+	    snrmB=snrmB+real(conj(bar._internal._internal[r][c])*bar._internal._internal[r][c]);
+	    snrm=snrm+nn;
+	  }}
+	 
+	}}}}
+
+	printf("scalar N2diff = %le (%le, %le) \n",snrm,snrmC,snrmB);fflush(stdout);
+
+
+      }
+    }
+
+ Grid_finalize();
+}

From f1876b7e959b0a1477e113b63b933d23f13a7c4d Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 14 Apr 2015 20:25:51 +0100
Subject: [PATCH 049/429] Modified

---
 TODO | 22 +++++++++++++++-------
 1 file changed, 15 insertions(+), 7 deletions(-)

diff --git a/TODO b/TODO
index 665e3a55..2c8fdcf8 100644
--- a/TODO
+++ b/TODO
@@ -1,26 +1,35 @@
 
 * FIXME audit
+
 * Replace vset with a call to merge.; 
-* care in Gmerge,Gextract over set.
+
+* care in Gmerge,Gextract over vset .
+
 * Const audit
+
 * extract / merge extra implementation removal      
 
 
 * Conditional execution, where etc...                -----DONE, simple test
+
 * Integer relational support                         -----DONE
+
 * Coordinate information, integers etc...            -----DONE
+
 * Integer type padding/union to vector.              -----DONE 
+
 * LatticeCoordinate[mu]                              -----DONE
 
-
-* Stencil operator support                           -----Initial thoughts
+* Stencil operator support                           -----Initial thoughts, trial implementation DONE.
+                                                     -----some simple tests that Stencil matches Cshift.
 
 * Subset support, slice sums etc...                  -----Only need slice sum?
                                                      -----Generic cartesian subslicing?
                                                      -----Array ranges / boost extents?
-                                                     -----Multigrid grid transferral.
+                                                     -----Multigrid grid transferral?
                                                      -----Suggests generalised cartesian subblocking
-                                                          sums, returning modified grid.
+                                                          sums, returning modified grid?
+					             -----What should interface be?
 
 i)  Two classes of subset;   red black parity subsetting (pick checkerboard).
                              cartesian sub-block subsetting
@@ -74,12 +83,11 @@ iii) No general permutation map.
 
 * NERSC Lattice loading, plaquette test
 
-
 * I/O support
   - MPI IO?
   - BinaryWriter, TextWriter etc...
   - protocol buffers?
-  - 
+
 // Cartesian grid inheritance
 //            Grid::GridBase
 //                     |

From ab9a764bb1aa7c234729a8022ad6a835d79506f5 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 14 Apr 2015 22:40:40 +0100
Subject: [PATCH 050/429] Reduce now going through MPI.

---
 Grid_Communicator.h      | 28 +++++++++++++---
 Grid_Lattice.h           | 69 ++++++++++++++++++++++++++++++++++++++--
 Grid_QCD.h               | 35 --------------------
 Grid_communicator_mpi.cc |  9 +++---
 Grid_cshift_common.h     |  2 --
 Grid_math_types.h        | 24 +++++++-------
 Grid_simd.h              |  8 ++---
 Grid_vComplexD.h         |  2 +-
 Grid_vComplexF.h         |  2 +-
 Grid_vRealD.h            |  2 +-
 Grid_vRealF.h            |  2 +-
 TODO                     | 14 +++++---
 12 files changed, 123 insertions(+), 74 deletions(-)

diff --git a/Grid_Communicator.h b/Grid_Communicator.h
index c7e71f12..0e15765c 100644
--- a/Grid_Communicator.h
+++ b/Grid_Communicator.h
@@ -42,13 +42,31 @@ class CartesianCommunicator {
     ////////////////////////////////////////////////////////////
     // Reduction
     ////////////////////////////////////////////////////////////
-    void GlobalSum(float &);
-    void GlobalSumVector(float *,int N);
+    void GlobalSum(RealF &);
+    void GlobalSumVector(RealF *,int N);
 
-    void GlobalSum(double &);
-    void GlobalSumVector(double *,int N);
+    void GlobalSum(RealD &);
+    void GlobalSumVector(RealD *,int N);
+
+    void GlobalSum(ComplexF &c)
+    {
+      GlobalSumVector((float *)&c,2);
+    }
+    void GlobalSumVector(ComplexF *c,int N)
+    {
+      GlobalSumVector((float *)c,2*N);
+    }
+
+    void GlobalSum(ComplexD &c)
+    {
+      GlobalSumVector((double *)&c,2);
+    }
+    void GlobalSumVector(ComplexD *c,int N)
+    {
+      GlobalSumVector((double *)c,2*N);
+    }
     
-    template<class obj> void GlobalSumObj(obj &o){
+    template<class obj> void GlobalSum(obj &o){
 
       typedef typename obj::scalar_type scalar_type;
       int words = sizeof(obj)/sizeof(scalar_type);
diff --git a/Grid_Lattice.h b/Grid_Lattice.h
index 1e2c820e..a94a13a4 100644
--- a/Grid_Lattice.h
+++ b/Grid_Lattice.h
@@ -407,18 +407,83 @@ public:
         return ret;
     }
 
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    // Reduction operations
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
     template<class vobj>
     inline RealD norm2(const Lattice<vobj> &arg){
 
       typedef typename vobj::scalar_type scalar;
-      decltype(localInnerProduct(arg._odata[0],arg._odata[0])) vnrm=zero;
+      decltype(innerProduct(arg._odata[0],arg._odata[0])) vnrm=zero;
 
       scalar nrm;
+      //FIXME make this loop parallelisable
       for(int ss=0;ss<arg._grid->oSites(); ss++){
-	vnrm = vnrm + localInnerProduct(arg._odata[ss],arg._odata[ss]);
+	vnrm = vnrm + innerProduct(arg._odata[ss],arg._odata[ss]);
       }
       nrm = Reduce(TensorRemove(vnrm));
+      arg._grid->GlobalSum(nrm);
       return real(nrm);
     }
+
+    template<class vobj>
+    inline auto innerProduct(const Lattice<vobj> &left,const Lattice<vobj> &right) ->decltype(innerProduct(left._odata[0],right._odata[0]))
+    {
+      typedef typename vobj::scalar_type scalar;
+      decltype(innerProduct(left._odata[0],right._odata[0])) vnrm=zero;
+
+      scalar nrm;
+      //FIXME make this loop parallelisable
+      for(int ss=0;ss<left._grid->oSites(); ss++){
+	vnrm = vnrm + innerProduct(left._odata[ss],right._odata[ss]);
+      }
+      nrm = Reduce(TensorRemove(vnrm));
+      right._grid->GlobalSum(nrm);
+      return nrm;
+    }
+
+    /////////////////////////////////////////////////////
+    // Non site reduced routines
+    /////////////////////////////////////////////////////
+
+    // localNorm2,
+    template<class vobj>
+    inline auto localNorm2 (const Lattice<vobj> &rhs)-> Lattice<typename vobj::tensor_reduced>
+    {
+      Lattice<typename vobj::tensor_reduced> ret(rhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<rhs._grid->oSites(); ss++){
+	  ret._odata[ss]=innerProduct(rhs,rhs);
+        }
+        return ret;
+    }
+
+    // localInnerProduct
+    template<class vobj>
+    inline auto localInnerProduct (const Lattice<vobj> &lhs,const Lattice<vobj> &rhs)
+      -> Lattice<typename vobj::tensor_reduced>
+    {
+      Lattice<typename vobj::tensor_reduced> ret(rhs._grid);
+#pragma omp parallel for
+      for(int ss=0;ss<rhs._grid->oSites(); ss++){
+	ret._odata[ss]=innerProduct(lhs._odata[ss],rhs._odata[ss]);
+      }
+      return ret;
+    }
+    
+    // outerProduct Scalar x Scalar -> Scalar
+    //              Vector x Vector -> Matrix
+    template<class ll,class rr>
+    inline auto outerProduct (const Lattice<ll> &lhs,const Lattice<rr> &rhs) -> Lattice<decltype(outerProduct(lhs._odata[0],rhs._odata[0]))>
+    {
+        Lattice<decltype(outerProduct(lhs._odata[0],rhs._odata[0]))> ret(rhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<rhs._grid->oSites(); ss++){
+            ret._odata[ss]=outerProduct(lhs._odata[ss],rhs._odata[ss]);
+        }
+        return ret;
+     }
+
+
 }
 #endif
diff --git a/Grid_QCD.h b/Grid_QCD.h
index 1d52f911..72a8ee67 100644
--- a/Grid_QCD.h
+++ b/Grid_QCD.h
@@ -57,41 +57,6 @@ namespace QCD {
     typedef Lattice<vSpinVector>       LatticeSpinVector;
     typedef Lattice<vColourVector>     LatticeColourVector;
 
-    // localNorm2,
-    template<class tt>
-    inline LatticeComplex localNorm2 (const Lattice<tt> &rhs)
-    {
-        LatticeComplex ret(rhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<rhs._grid->oSites(); ss++){
-            ret._odata[ss]=trace(adj(rhs)*rhs);
-        }
-        return ret;
-    }
-    // localInnerProduct
-    template<class tt>
-    inline LatticeComplex localInnerProduct (const Lattice<tt> &lhs,const Lattice<tt> &rhs)
-    {
-        LatticeComplex ret(rhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<rhs._grid->oSites(); ss++){
-            ret._odata[ss]=localInnerProduct(lhs._odata[ss],rhs._odata[ss]);
-        }
-        return ret;
-    }
-    
-    // outerProduct Scalar x Scalar -> Scalar
-    //              Vector x Vector -> Matrix
-    template<class ll,class rr>
-    inline auto outerProduct (const Lattice<ll> &lhs,const Lattice<rr> &rhs) -> Lattice<decltype(outerProduct(lhs._odata[0],rhs._odata[0]))>
-    {
-        Lattice<decltype(outerProduct(lhs._odata[0],rhs._odata[0]))> ret(rhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<rhs._grid->oSites(); ss++){
-            ret._odata[ss]=outerProduct(lhs._odata[ss],rhs._odata[ss]);
-        }
-        return ret;
-     }
 
     // FIXME for debug; deprecate this
    inline void LatticeCoordinate(LatticeInteger &l,int mu){
diff --git a/Grid_communicator_mpi.cc b/Grid_communicator_mpi.cc
index a4943478..dba5f039 100644
--- a/Grid_communicator_mpi.cc
+++ b/Grid_communicator_mpi.cc
@@ -29,21 +29,20 @@ CartesianCommunicator::CartesianCommunicator(std::vector<int> &processors)
 }
 
 void CartesianCommunicator::GlobalSum(float &f){
-  MPI_Allreduce(&f,&f,1,MPI_FLOAT,MPI_SUM,communicator);
+  MPI_Allreduce(MPI_IN_PLACE,&f,1,MPI_FLOAT,MPI_SUM,communicator);
 }
 void CartesianCommunicator::GlobalSumVector(float *f,int N)
 {
-  MPI_Allreduce(f,f,N,MPI_FLOAT,MPI_SUM,communicator);
+  MPI_Allreduce(MPI_IN_PLACE,f,N,MPI_FLOAT,MPI_SUM,communicator);
 }
 void CartesianCommunicator::GlobalSum(double &d)
 {
-  MPI_Allreduce(&d,&d,1,MPI_DOUBLE,MPI_SUM,communicator);
+  MPI_Allreduce(MPI_IN_PLACE,&d,1,MPI_DOUBLE,MPI_SUM,communicator);
 }
 void CartesianCommunicator::GlobalSumVector(double *d,int N)
 {
-  MPI_Allreduce(d,d,N,MPI_DOUBLE,MPI_SUM,communicator);
+  MPI_Allreduce(MPI_IN_PLACE,d,N,MPI_DOUBLE,MPI_SUM,communicator);
 }
-
 void CartesianCommunicator::ShiftedRanks(int dim,int shift,int &source,int &dest)
 {
   MPI_Cart_shift(communicator,dim,shift,&source,&dest);
diff --git a/Grid_cshift_common.h b/Grid_cshift_common.h
index e962f6ea..2910151c 100644
--- a/Grid_cshift_common.h
+++ b/Grid_cshift_common.h
@@ -8,7 +8,6 @@ friend void Gather_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllo
 {
   int rd = rhs._grid->_rdimensions[dimension];
 
-  printf("Gather_plane_simple buf size %d rhs size %d \n",buffer.size(),rhs._odata.size());fflush(stdout);
   if ( !rhs._grid->CheckerBoarded(dimension) ) {
 
     int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
@@ -19,7 +18,6 @@ friend void Gather_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllo
 #pragma omp parallel for collapse(2)
     for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
       for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-	printf("Gather_plane_simple %d ; %d %d %d\n",bo,so,o,b);fflush(stdout);
 	buffer[bo++]=rhs._odata[so+o+b];
       }
       o +=rhs._grid->_slice_stride[dimension];
diff --git a/Grid_math_types.h b/Grid_math_types.h
index e3432645..01fc7bf2 100644
--- a/Grid_math_types.h
+++ b/Grid_math_types.h
@@ -903,38 +903,38 @@ template<class l,int N> inline iMatrix<l,N> operator - (Integer lhs,const iMatri
 
 
     ///////////////////////////////////////////////////////////////////////////////////////
-    // localInnerProduct Scalar x Scalar -> Scalar
-    // localInnerProduct Vector x Vector -> Scalar
-    // localInnerProduct Matrix x Matrix -> Scalar
+    // innerProduct Scalar x Scalar -> Scalar
+    // innerProduct Vector x Vector -> Scalar
+    // innerProduct Matrix x Matrix -> Scalar
     ///////////////////////////////////////////////////////////////////////////////////////
     template<class l,class r,int N> inline
-    auto localInnerProduct (const iVector<l,N>& lhs,const iVector<r,N>& rhs) -> iScalar<decltype(localInnerProduct(lhs._internal[0],rhs._internal[0]))>
+    auto innerProduct (const iVector<l,N>& lhs,const iVector<r,N>& rhs) -> iScalar<decltype(innerProduct(lhs._internal[0],rhs._internal[0]))>
     {
-        typedef decltype(localInnerProduct(lhs._internal[0],rhs._internal[0])) ret_t;
+        typedef decltype(innerProduct(lhs._internal[0],rhs._internal[0])) ret_t;
         iScalar<ret_t> ret=zero;
         for(int c1=0;c1<N;c1++){
-            ret._internal += localInnerProduct(lhs._internal[c1],rhs._internal[c1]);
+            ret._internal += innerProduct(lhs._internal[c1],rhs._internal[c1]);
         }
         return ret;
     }
     template<class l,class r,int N> inline
-    auto localInnerProduct (const iMatrix<l,N>& lhs,const iMatrix<r,N>& rhs) -> iScalar<decltype(localInnerProduct(lhs._internal[0][0],rhs._internal[0][0]))>
+    auto innerProduct (const iMatrix<l,N>& lhs,const iMatrix<r,N>& rhs) -> iScalar<decltype(innerProduct(lhs._internal[0][0],rhs._internal[0][0]))>
     {
-        typedef decltype(localInnerProduct(lhs._internal[0][0],rhs._internal[0][0])) ret_t;
+        typedef decltype(innerProduct(lhs._internal[0][0],rhs._internal[0][0])) ret_t;
         iScalar<ret_t> ret=zero;
         iScalar<ret_t> tmp;
         for(int c1=0;c1<N;c1++){
         for(int c2=0;c2<N;c2++){
-	  ret._internal+=localInnerProduct(lhs._internal[c1][c2],rhs._internal[c1][c2]);
+	  ret._internal+=innerProduct(lhs._internal[c1][c2],rhs._internal[c1][c2]);
         }}
         return ret;
     }
     template<class l,class r> inline
-    auto localInnerProduct (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(localInnerProduct(lhs._internal,rhs._internal))>
+    auto innerProduct (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(innerProduct(lhs._internal,rhs._internal))>
     {
-        typedef decltype(localInnerProduct(lhs._internal,rhs._internal)) ret_t;
+        typedef decltype(innerProduct(lhs._internal,rhs._internal)) ret_t;
         iScalar<ret_t> ret;
-        ret._internal = localInnerProduct(lhs._internal,rhs._internal);
+        ret._internal = innerProduct(lhs._internal,rhs._internal);
         return ret;
     }
 
diff --git a/Grid_simd.h b/Grid_simd.h
index 23061632..4f9f4b0d 100644
--- a/Grid_simd.h
+++ b/Grid_simd.h
@@ -33,10 +33,10 @@ namespace Grid {
 
   inline RealF adj(const RealF  & r){ return r; }
   inline RealF conj(const RealF  & r){ return r; }
-  inline ComplexD localInnerProduct(const ComplexD & l, const ComplexD & r) { return conj(l)*r; }
-  inline ComplexF localInnerProduct(const ComplexF & l, const ComplexF & r) { return conj(l)*r; }
-  inline RealD localInnerProduct(const RealD & l, const RealD & r) { return l*r; }
-  inline RealF localInnerProduct(const RealF & l, const RealF & r) { return l*r; }
+  inline ComplexD innerProduct(const ComplexD & l, const ComplexD & r) { return conj(l)*r; }
+  inline ComplexF innerProduct(const ComplexF & l, const ComplexF & r) { return conj(l)*r; }
+  inline RealD innerProduct(const RealD & l, const RealD & r) { return l*r; }
+  inline RealF innerProduct(const RealF & l, const RealF & r) { return l*r; }
 
     ////////////////////////////////////////////////////////////////////////////////
     //Provide support functions for basic real and complex data types required by Grid
diff --git a/Grid_vComplexD.h b/Grid_vComplexD.h
index f1c335e8..133e354b 100644
--- a/Grid_vComplexD.h
+++ b/Grid_vComplexD.h
@@ -313,7 +313,7 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
     };
 
 
-    inline vComplexD localInnerProduct(const vComplexD & l, const vComplexD & r) { return conj(l)*r; }
+    inline vComplexD innerProduct(const vComplexD & l, const vComplexD & r) { return conj(l)*r; }
 
 
     typedef  vComplexD vDComplex;
diff --git a/Grid_vComplexF.h b/Grid_vComplexF.h
index 18cf1749..9dc5f051 100644
--- a/Grid_vComplexF.h
+++ b/Grid_vComplexF.h
@@ -362,7 +362,7 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
 
     };
 
-    inline vComplexF localInnerProduct(const vComplexF & l, const vComplexF & r) 
+    inline vComplexF innerProduct(const vComplexF & l, const vComplexF & r) 
     {
       return conj(l)*r; 
     }
diff --git a/Grid_vRealD.h b/Grid_vRealD.h
index c51e41cf..84c80698 100644
--- a/Grid_vRealD.h
+++ b/Grid_vRealD.h
@@ -240,7 +240,7 @@ namespace Grid {
         static int Nsimd(void) { return sizeof(dvec)/sizeof(double);}
     };
 
-   inline vRealD localInnerProduct(const vRealD & l, const vRealD & r) { return conj(l)*r; }
+   inline vRealD innerProduct(const vRealD & l, const vRealD & r) { return conj(l)*r; }
     inline void zeroit(vRealD &z){ vzero(z);}
 
     inline vRealD outerProduct(const vRealD &l, const vRealD& r)
diff --git a/Grid_vRealF.h b/Grid_vRealF.h
index 0fe68f43..aeb454f0 100644
--- a/Grid_vRealF.h
+++ b/Grid_vRealF.h
@@ -260,7 +260,7 @@ friend inline void vstore(const vRealF &ret, float *a){
     public:
         static inline int Nsimd(void) { return sizeof(fvec)/sizeof(float);}
     };
-    inline vRealF localInnerProduct(const vRealF & l, const vRealF & r) { return conj(l)*r; }
+    inline vRealF innerProduct(const vRealF & l, const vRealF & r) { return conj(l)*r; }
     inline void  zeroit(vRealF &z){ vzero(z);}
 
     inline vRealF outerProduct(const vRealF &l, const vRealF& r)
diff --git a/TODO b/TODO
index 2c8fdcf8..010a06b7 100644
--- a/TODO
+++ b/TODO
@@ -1,15 +1,14 @@
 
+AUDITS:
 * FIXME audit
-
+* const audit
 * Replace vset with a call to merge.; 
-
 * care in Gmerge,Gextract over vset .
-
-* Const audit
-
 * extract / merge extra implementation removal      
 
 
+FUNCTIONALITY:
+
 * Conditional execution, where etc...                -----DONE, simple test
 
 * Integer relational support                         -----DONE
@@ -22,6 +21,11 @@
 
 * Stencil operator support                           -----Initial thoughts, trial implementation DONE.
                                                      -----some simple tests that Stencil matches Cshift.
+                                                     -----do all permute in comms phase, so that copy permute
+						     -----cases move into a buffer.
+
+* CovariantShift support                             -----Use a class to store gauge field? (parallel transport?)
+
 
 * Subset support, slice sums etc...                  -----Only need slice sum?
                                                      -----Generic cartesian subslicing?

From cab7ef9bc22ea99982e7ff2cd1f7b342d71cf375 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 14 Apr 2015 23:20:16 +0100
Subject: [PATCH 051/429] Some bug fixes

---
 Grid_main.cc            |  8 ++++----
 Grid_math_type_mapper.h | 10 ++++++++++
 Grid_math_types.h       |  3 +++
 Grid_simd.h             | 12 +++++++++---
 Grid_vComplexF.h        |  4 +++-
 Grid_vRealD.h           |  3 +++
 Grid_vRealF.h           |  3 +++
 7 files changed, 35 insertions(+), 8 deletions(-)

diff --git a/Grid_main.cc b/Grid_main.cc
index 13c3d9ee..8cac55f5 100644
--- a/Grid_main.cc
+++ b/Grid_main.cc
@@ -14,10 +14,10 @@ int main (int argc, char ** argv)
   std::vector<int> simd_layout(4);
   
   std::vector<int> mpi_layout(4);
-  mpi_layout[0]=2;
-  mpi_layout[1]=2;
-  mpi_layout[2]=2;
-  mpi_layout[3]=2;
+  mpi_layout[0]=1;
+  mpi_layout[1]=1;
+  mpi_layout[2]=1;
+  mpi_layout[3]=1;
 
 #ifdef AVX512
  for(int omp=128;omp<236;omp+=16){
diff --git a/Grid_math_type_mapper.h b/Grid_math_type_mapper.h
index 856f2c6b..4e7120b1 100644
--- a/Grid_math_type_mapper.h
+++ b/Grid_math_type_mapper.h
@@ -10,6 +10,7 @@ namespace Grid {
     typedef typename T::scalar_type scalar_type;
     typedef typename T::vector_type vector_type;
     typedef typename T::tensor_reduced tensor_reduced;
+    //    enum { TensorLevel = T::TensorLevel };
   };
 
 //////////////////////////////////////////////////////////////////////////////////
@@ -20,24 +21,28 @@ namespace Grid {
     typedef RealF scalar_type;
     typedef RealF vector_type;
     typedef RealF tensor_reduced ;
+    //    enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<RealD> {
   public:
     typedef RealD scalar_type;
     typedef RealD vector_type;
     typedef RealD tensor_reduced;
+    //    enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<ComplexF> {
   public:
     typedef ComplexF scalar_type;
     typedef ComplexF vector_type;
     typedef ComplexF tensor_reduced;
+    //    enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<ComplexD> {
   public:
     typedef ComplexD scalar_type;
     typedef ComplexD vector_type;
     typedef ComplexD tensor_reduced;
+    //    enum { TensorLevel = 0 };
   };
 
   template<> class GridTypeMapper<vRealF> {
@@ -45,30 +50,35 @@ namespace Grid {
     typedef RealF  scalar_type;
     typedef vRealF vector_type;
     typedef vRealF tensor_reduced;
+    //    enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<vRealD> {
   public:
     typedef RealD  scalar_type;
     typedef vRealD vector_type;
     typedef vRealD tensor_reduced;
+    //enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<vComplexF> {
   public:
     typedef ComplexF  scalar_type;
     typedef vComplexF vector_type;
     typedef vComplexF tensor_reduced;
+    //enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<vComplexD> {
   public:
     typedef ComplexD  scalar_type;
     typedef vComplexD vector_type;
     typedef vComplexD tensor_reduced;
+    //enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<vInteger> {
   public:
     typedef Integer  scalar_type;
     typedef vInteger vector_type;
     typedef vInteger tensor_reduced;
+    //enum { TensorLevel = 0 };
   };
 
   // Again terminate the recursion.
diff --git a/Grid_math_types.h b/Grid_math_types.h
index 01fc7bf2..db90bda7 100644
--- a/Grid_math_types.h
+++ b/Grid_math_types.h
@@ -19,6 +19,7 @@ public:
   typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
   typedef typename GridTypeMapper<vtype>::vector_type vector_type;
   typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
+  //enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
   typedef iScalar<tensor_reduced_v> tensor_reduced;
 
 
@@ -78,6 +79,7 @@ public:
   typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
   typedef typename GridTypeMapper<vtype>::vector_type vector_type;
   typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
+  //  enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
   typedef iScalar<tensor_reduced_v> tensor_reduced;
 
 
@@ -136,6 +138,7 @@ public:
   typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
   typedef typename GridTypeMapper<vtype>::vector_type vector_type;
   typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
+  //  enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
   typedef iScalar<tensor_reduced_v> tensor_reduced;
 
   iMatrix(Zero &z){ *this = zero; };
diff --git a/Grid_simd.h b/Grid_simd.h
index 4f9f4b0d..f0cf6c08 100644
--- a/Grid_simd.h
+++ b/Grid_simd.h
@@ -273,10 +273,16 @@ namespace Grid {
    inline void add (vRealD *__restrict__ y,const vRealD *__restrict__ l,const RealD *__restrict__ r){ *y = (*l) + (*r); }
 
   // Default precision
-  typedef RealD  Real;
-  typedef std::complex<Real>  Complex;
-
+#ifdef GRID_DEFAULT_PRECISION_DOUBLE
+  typedef RealD   Real;
   typedef vRealD vReal;
   typedef vComplexD vComplex;
+  typedef std::complex<Real>  Complex;
+#else
+  typedef RealF  Real;
+  typedef vRealF vReal;
+  typedef vComplexF vComplex;
+  typedef std::complex<Real>  Complex;
+#endif
 }
 #endif
diff --git a/Grid_vComplexF.h b/Grid_vComplexF.h
index 9dc5f051..75c47248 100644
--- a/Grid_vComplexF.h
+++ b/Grid_vComplexF.h
@@ -373,6 +373,8 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
     {
         return l*r;
     }
-
+    inline vComplexF trace(const vComplexF &arg){
+        return arg;
+    }
 }
 #endif
diff --git a/Grid_vRealD.h b/Grid_vRealD.h
index 84c80698..86f738f3 100644
--- a/Grid_vRealD.h
+++ b/Grid_vRealD.h
@@ -247,6 +247,9 @@ namespace Grid {
     {
         return l*r;
     }
+    inline vRealD trace(const vRealD &arg){
+        return arg;
+    }
 
 
 }
diff --git a/Grid_vRealF.h b/Grid_vRealF.h
index aeb454f0..b7c127ce 100644
--- a/Grid_vRealF.h
+++ b/Grid_vRealF.h
@@ -268,6 +268,9 @@ friend inline void vstore(const vRealF &ret, float *a){
         return l*r;
     }
 
+    inline vRealF trace(const vRealF &arg){
+        return arg;
+    }
 
     
 }

From fddb904b4cb9d3a21c0d537f9859308db6177326 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 15 Apr 2015 12:03:38 +0100
Subject: [PATCH 052/429] Typo in capital

---
 Makefile.am | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/Makefile.am b/Makefile.am
index e306eb56..1e70ac1b 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -49,6 +49,6 @@ bin_PROGRAMS = Grid_main test_Grid_stencil
 Grid_main_SOURCES = Grid_main.cc
 Grid_main_LDADD = libGrid.a
 
-test_Grid_stencil_SOURCES = test_Grid_Stencil.cc
+test_Grid_stencil_SOURCES = test_Grid_stencil.cc
 test_Grid_stencil_LDADD = libGrid.a
 

From 933c54d9c419183c934b0b23b819c3a083d8af90 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 16 Apr 2015 14:47:28 +0100
Subject: [PATCH 053/429] Improving the trace support to support any index
 tracing and simplifying implmentation in some ways

---
 Grid_QCD.h              |  10 +-
 Grid_config.h           |   4 +-
 Grid_main.cc            |  21 +++-
 Grid_math_type_mapper.h |  41 +++++---
 Grid_math_types.h       | 224 ++++++++++++++++++++++++++++++++++++++--
 Grid_simd.h             |  24 +++--
 Grid_vComplexD.h        |  18 ++--
 Grid_vComplexF.h        |  20 ++--
 Grid_vInteger.h         |  14 +--
 Grid_vRealD.h           |  14 +--
 Grid_vRealF.h           |  14 +--
 test_Grid_stencil.cc    |   4 +-
 12 files changed, 321 insertions(+), 87 deletions(-)

diff --git a/Grid_QCD.h b/Grid_QCD.h
index 72a8ee67..2f411f16 100644
--- a/Grid_QCD.h
+++ b/Grid_QCD.h
@@ -7,12 +7,11 @@ namespace QCD {
     static const int Ns=4;
     static const int CbRed  =0;
     static const int CbBlack=1;
-
+    
     // QCD iMatrix types
     template<typename vtype> using iSinglet          = iScalar<iScalar<vtype> > ;
     template<typename vtype> using iSpinMatrix       = iMatrix<iScalar<vtype>, Ns>;
     template<typename vtype> using iSpinColourMatrix = iMatrix<iMatrix<vtype, Nc>, Ns>;
-
     template<typename vtype> using iColourMatrix     = iScalar<iMatrix<vtype, Nc>> ;
 
     template<typename vtype> using iSpinVector       = iVector<iScalar<vtype>, Ns>;
@@ -20,10 +19,8 @@ namespace QCD {
     template<typename vtype> using iSpinColourVector = iVector<iVector<vtype, Nc>, Ns>;
 
     typedef iSinglet<Complex >          TComplex;    // This is painful. Tensor singlet complex type.
-    typedef iSinglet<vComplex >         vTComplex;
-    typedef iSinglet<Real >             TReal;    // This is painful. Tensor singlet complex type.
-
-
+    typedef iSinglet<vComplex >         vTComplex;   // what if we don't know the tensor structure
+    typedef iSinglet<Real >             TReal;       // Shouldn't need these.
     typedef iSinglet<vInteger >         vTInteger;
 
     typedef iSpinMatrix<Complex >       SpinMatrix;
@@ -44,7 +41,6 @@ namespace QCD {
     typedef iSpinColourVector<vComplex > vSpinColourVector;
     
     typedef Lattice<vTComplex>            LatticeComplex;
-
     typedef Lattice<vInteger>            LatticeInteger; // Predicates for "where"
     
     typedef Lattice<vColourMatrix>     LatticeColourMatrix;
diff --git a/Grid_config.h b/Grid_config.h
index bb885708..5faa2157 100644
--- a/Grid_config.h
+++ b/Grid_config.h
@@ -73,8 +73,8 @@
 /* Define to the version of this package. */
 #define PACKAGE_VERSION "1.0"
 
-/* SSE2 */
-/* #undef SSE2 */
+/* SSE4 */
+/* #undef SSE4 */
 
 /* Define to 1 if you have the ANSI C header files. */
 #define STDC_HEADERS 1
diff --git a/Grid_main.cc b/Grid_main.cc
index 8cac55f5..e122fcbe 100644
--- a/Grid_main.cc
+++ b/Grid_main.cc
@@ -152,6 +152,8 @@ int main (int argc, char ** argv)
     cMat= adj(cMat);
     cm=adj(cm);
     scm=adj(scm);
+    scm=transpose(scm);
+    scm=transposeIndex<1>(scm);
     
 //    Foo = Foo+scalar; // LatticeColourMatrix+Scalar
 //    Foo = Foo*scalar; // LatticeColourMatrix*Scalar
@@ -162,6 +164,23 @@ int main (int argc, char ** argv)
     
     LatticeComplex trscMat(&Fine);
     trscMat = trace(scMat); // Trace
+
+    {
+      TComplex      c;
+      ColourMatrix c_m;   
+      SpinMatrix   s_m;   
+      SpinColourMatrix sc_m; 
+
+      s_m = traceIndex<1>(sc_m);
+      c_m = traceIndex<2>(sc_m);
+
+      c   = traceIndex<2>(s_m);
+      c   = traceIndex<1>(c_m);
+
+      printf("c. Level %d\n",c_m.TensorLevel);
+      printf("c. Level %d\n",c_m._internal.TensorLevel);
+
+    }
     
     FooBar = Bar;
  
@@ -203,7 +222,7 @@ int main (int argc, char ** argv)
         peekSite(bar,Bar,coor);
         for(int r=0;r<3;r++){
         for(int c=0;c<3;c++){
-	  cout<<"bar "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "<<bar._internal._internal[r][c]<<std::endl;
+	  //	  cout<<"bar "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "<<bar._internal._internal[r][c]<<std::endl;
 	}}
       }}}}
     } 
diff --git a/Grid_math_type_mapper.h b/Grid_math_type_mapper.h
index 4e7120b1..7f6d85f5 100644
--- a/Grid_math_type_mapper.h
+++ b/Grid_math_type_mapper.h
@@ -1,8 +1,22 @@
 #ifndef GRID_MATH_TYPE_MAPPER_H
 #define GRID_MATH_TYPE_MAPPER_H
+
 namespace Grid {
+
+
 //////////////////////////////////////////////////////////////////////////////////
 // Want to recurse: GridTypeMapper<Matrix<vComplexD> >::scalar_type == ComplexD.
+// Use of a helper class like this allows us to template specialise and "dress"
+// other classes such as RealD == double, ComplexD == std::complex<double> with these
+// traits.
+//
+// It is possible that we could do this more elegantly if I introduced a 
+// queryable trait in iScalar, iMatrix and iVector and used the query on vtype in 
+// place of the type mapper?
+//
+// Not sure how to do this, but probably could be done with a research effort
+// to study C++11's type_traits.h file. (std::enable_if<isGridTensorType<vtype> >)
+//
 //////////////////////////////////////////////////////////////////////////////////
   
   template <class T> class GridTypeMapper {
@@ -10,7 +24,7 @@ namespace Grid {
     typedef typename T::scalar_type scalar_type;
     typedef typename T::vector_type vector_type;
     typedef typename T::tensor_reduced tensor_reduced;
-    //    enum { TensorLevel = T::TensorLevel };
+    enum { TensorLevel = T::TensorLevel };
   };
 
 //////////////////////////////////////////////////////////////////////////////////
@@ -21,28 +35,28 @@ namespace Grid {
     typedef RealF scalar_type;
     typedef RealF vector_type;
     typedef RealF tensor_reduced ;
-    //    enum { TensorLevel = 0 };
+    enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<RealD> {
   public:
     typedef RealD scalar_type;
     typedef RealD vector_type;
     typedef RealD tensor_reduced;
-    //    enum { TensorLevel = 0 };
+    enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<ComplexF> {
   public:
     typedef ComplexF scalar_type;
     typedef ComplexF vector_type;
     typedef ComplexF tensor_reduced;
-    //    enum { TensorLevel = 0 };
+    enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<ComplexD> {
   public:
     typedef ComplexD scalar_type;
     typedef ComplexD vector_type;
     typedef ComplexD tensor_reduced;
-    //    enum { TensorLevel = 0 };
+    enum { TensorLevel = 0 };
   };
 
   template<> class GridTypeMapper<vRealF> {
@@ -50,44 +64,37 @@ namespace Grid {
     typedef RealF  scalar_type;
     typedef vRealF vector_type;
     typedef vRealF tensor_reduced;
-    //    enum { TensorLevel = 0 };
+    enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<vRealD> {
   public:
     typedef RealD  scalar_type;
     typedef vRealD vector_type;
     typedef vRealD tensor_reduced;
-    //enum { TensorLevel = 0 };
+    enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<vComplexF> {
   public:
     typedef ComplexF  scalar_type;
     typedef vComplexF vector_type;
     typedef vComplexF tensor_reduced;
-    //enum { TensorLevel = 0 };
+    enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<vComplexD> {
   public:
     typedef ComplexD  scalar_type;
     typedef vComplexD vector_type;
     typedef vComplexD tensor_reduced;
-    //enum { TensorLevel = 0 };
+    enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<vInteger> {
   public:
     typedef Integer  scalar_type;
     typedef vInteger vector_type;
     typedef vInteger tensor_reduced;
-    //enum { TensorLevel = 0 };
+    enum { TensorLevel = 0 };
   };
 
-  // Again terminate the recursion.
-  inline vRealD    TensorRemove(vRealD    arg){ return arg;}
-  inline vRealF    TensorRemove(vRealF    arg){ return arg;}
-  inline vComplexF TensorRemove(vComplexF arg){ return arg;}
-  inline vComplexD TensorRemove(vComplexD arg){ return arg;}
-  inline vInteger  TensorRemove(vInteger  arg){ return arg;}
-
 }
 
 #endif
diff --git a/Grid_math_types.h b/Grid_math_types.h
index db90bda7..58de9c15 100644
--- a/Grid_math_types.h
+++ b/Grid_math_types.h
@@ -2,15 +2,76 @@
 #define GRID_MATH_TYPES_H
 
 #include <Grid_math_type_mapper.h>
+#include <type_traits>
 
 namespace Grid {
 
+  // First some of my own traits
+  template<typename T> struct isGridTensor {
+    static const bool value = true;
+    static const bool notvalue = false;
+  };
+
+  template<> struct isGridTensor<RealD > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<RealF > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<ComplexD > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<ComplexF > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<Integer > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<vRealD > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<vRealF > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<vComplexD > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<vComplexF > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<vInteger > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+
+  // Match the index
+  template<typename T,int Level> struct matchGridTensorIndex {
+    static const bool value = (Level==T::TensorLevel);
+    static const bool notvalue = (Level!=T::TensorLevel);
+  };
+
 
 ///////////////////////////////////////////////////
 // Scalar, Vector, Matrix objects.
 // These can be composed to form tensor products of internal indices.
 ///////////////////////////////////////////////////
-    
+
+  // Terminates the recursion for temoval of all Grids tensors
+  inline vRealD    TensorRemove(vRealD    arg){ return arg;}
+  inline vRealF    TensorRemove(vRealF    arg){ return arg;}
+  inline vComplexF TensorRemove(vComplexF arg){ return arg;}
+  inline vComplexD TensorRemove(vComplexD arg){ return arg;}
+  inline vInteger  TensorRemove(vInteger  arg){ return arg;}
+
 template<class vtype> class iScalar
 {
 public:
@@ -19,9 +80,12 @@ public:
   typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
   typedef typename GridTypeMapper<vtype>::vector_type vector_type;
   typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
-  //enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
   typedef iScalar<tensor_reduced_v> tensor_reduced;
 
+  enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
+
+  // Scalar no action
+  //  template<int Level> using tensor_reduce_level = typename iScalar<GridTypeMapper<vtype>::tensor_reduce_level<Level> >;
 
     iScalar(){};
     
@@ -79,9 +143,9 @@ public:
   typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
   typedef typename GridTypeMapper<vtype>::vector_type vector_type;
   typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
-  //  enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
-  typedef iScalar<tensor_reduced_v> tensor_reduced;
 
+  enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
+  typedef iScalar<tensor_reduced_v> tensor_reduced;
 
     iVector(Zero &z){ *this = zero; };
     iVector() {};
@@ -137,8 +201,10 @@ public:
 
   typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
   typedef typename GridTypeMapper<vtype>::vector_type vector_type;
+
   typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
-  //  enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
+
+  enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
   typedef iScalar<tensor_reduced_v> tensor_reduced;
 
   iMatrix(Zero &z){ *this = zero; };
@@ -1018,6 +1084,137 @@ template<class vtype,int N> inline iMatrix<vtype,N> adj(const iMatrix<vtype,N> &
     }}
     return ret;
 }
+/////////////////////////////////////////////////////////////////
+// Transpose
+/////////////////////////////////////////////////////////////////
+template<class vtype,int N>
+  inline typename std::enable_if<isGridTensor<vtype>::value, iMatrix<vtype,N> >::type 
+  transpose(iMatrix<vtype,N> arg)
+  {
+    iMatrix<vtype,N> ret;
+    for(int i=0;i<N;i++){
+      for(int j=0;j<N;j++){
+	ret._internal[i][j] = transpose(arg._internal[j][i]); // NB recurses
+      }}
+    return ret;
+  }
+template<class vtype,int N>
+  inline typename std::enable_if<isGridTensor<vtype>::notvalue, iMatrix<vtype,N> >::type 
+  transpose(iMatrix<vtype,N> arg)
+  {
+    iMatrix<vtype,N> ret;
+    for(int i=0;i<N;i++){
+      for(int j=0;j<N;j++){
+	ret._internal[i][j] = arg._internal[j][i]; // Stop recursion if not a tensor type
+      }}
+    return ret;
+  }
+
+template<class vtype,int N>
+  inline typename std::enable_if<isGridTensor<vtype>::value, iScalar<vtype> >::type 
+  transpose(iScalar<vtype> arg)
+  {
+    iScalar<vtype> ret;
+    ret._internal = transpose(arg._internal); // NB recurses
+    return ret;
+  }
+
+template<class vtype,int N>
+  inline typename std::enable_if<isGridTensor<vtype>::notvalue, iScalar<vtype> >::type 
+  transpose(iScalar<vtype> arg)
+  {
+    iScalar<vtype> ret;
+    ret._internal = arg._internal; // NB recursion stops
+    return ret;
+  }
+
+
+
+/*
+ *  No need to implement transpose on the primitive types
+ *  Not sure that this idiom is any more elegant that the trace idiom below however!
+inline ComplexD transpose(ComplexD &rhs){  return rhs;}
+inline ComplexF transpose(ComplexF &rhs){  return rhs;}
+inline RealD transpose(RealD &rhs){  return rhs;}
+inline RealF transpose(RealF &rhs){  return rhs;}
+ */
+
+////////////////////////////////////////////////////////////////////////////////////////////
+// Transpose a specific index
+////////////////////////////////////////////////////////////////////////////////////////////
+template<int Level,class vtype,int N> inline 
+  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value, iMatrix<vtype,N> >::type 
+transposeIndex (const iMatrix<vtype,N> &arg)
+{
+  iMatrix<vtype,N> ret;
+  for(int i=0;i<N;i++){
+    for(int j=0;j<N;j++){
+      ret._internal[i][j] = arg._internal[j][i]; 
+  }}
+  return ret;
+}
+// or not
+template<int Level,class vtype,int N> inline 
+typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue, iMatrix<vtype,N> >::type 
+transposeIndex (const iMatrix<vtype,N> &arg)
+{
+  iMatrix<vtype,N> ret;
+  for(int i=0;i<N;i++){
+    for(int j=0;j<N;j++){
+      ret._internal[i][j] = transposeIndex<Level>(arg._internal[i][j]); 
+  }}
+  return ret;
+}
+template<int Level,class vtype> inline 
+typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue, iScalar<vtype> >::type 
+transposeIndex (const iScalar<vtype> &arg)
+{
+  return transposeIndex<Level>(arg._internal);
+}
+////////////////////////////////
+// Trace a specific index 
+////////////////////////////////
+
+template<int Level> inline ComplexF traceIndex(const ComplexF arg) { return arg;}
+template<int Level> inline ComplexD traceIndex(const ComplexD arg) { return arg;}
+template<int Level> inline RealF traceIndex(const RealF arg) { return arg;}
+template<int Level> inline RealD traceIndex(const RealD arg) { return arg;}
+
+template<int Level,class vtype> inline 
+auto traceIndex(const iScalar<vtype> &arg) -> iScalar<decltype(traceIndex<Level>(arg._internal)) >
+{
+  iScalar<decltype(traceIndex<Level>(arg._internal))> ret;
+  ret._internal = traceIndex<Level>(arg._internal);
+  return ret;
+}
+
+// If we hit the right index, return scalar and trace it with no further recursion
+template<int Level,class vtype,int N> inline 
+auto traceIndex(const iMatrix<vtype,N> &arg) ->
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value,  // Index matches
+                                                    iScalar<vtype> >::type                              // return scalar
+{
+  iScalar<vtype> ret;
+  zeroit(ret._internal);
+  for(int i=0;i<N;i++){
+    ret._internal = ret._internal + arg._internal[i][i];
+  }
+  return ret;
+}
+
+// not this level, so recurse
+template<int Level,class vtype,int N> inline 
+auto traceIndex(const iMatrix<vtype,N> &arg) ->
+  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,// No index match
+         iMatrix<decltype(traceIndex<Level>(arg._internal[0][0])),N> >::type     // return matrix
+{
+  iMatrix<decltype(traceIndex<Level>(arg._internal[0][0])),N> ret;
+  for(int i=0;i<N;i++){
+  for(int j=0;j<N;j++){
+    ret._internal[i][j] = traceIndex<Level>(arg._internal[i][j]);
+  }}
+  return ret;
+}
 
 /////////////////////////////////////////////////////////////////
 // Can only take the real/imag part of scalar objects, since
@@ -1075,17 +1272,24 @@ template<class itype,int N> inline auto imag(const iVector<itype,N> &z) -> iVect
     // Trace of scalar and matrix
     /////////////////////////////////
 
-inline Complex trace( const Complex &arg){
+inline ComplexF trace( const ComplexF &arg){
     return arg;
 }
-//inline vComplex trace(const vComplex &arg){
-//    return arg;
-//}
+inline ComplexD trace( const ComplexD &arg){
+    return arg;
+}
+inline RealF trace( const RealF &arg){
+    return arg;
+}
+inline RealD trace( const RealD &arg){
+    return arg;
+}
+
 template<class vtype,int N>
 inline auto trace(const iMatrix<vtype,N> &arg) -> iScalar<decltype(trace(arg._internal[0][0]))>
 {
     iScalar<decltype( trace(arg._internal[0][0] )) > ret;
-    ZeroIt(ret._internal);
+    zeroit(ret._internal);
     for(int i=0;i<N;i++){
         ret._internal=ret._internal+trace(arg._internal[i][i]);
     }
diff --git a/Grid_simd.h b/Grid_simd.h
index f0cf6c08..ae05c3b2 100644
--- a/Grid_simd.h
+++ b/Grid_simd.h
@@ -12,7 +12,7 @@
 ////////////////////////////////////////////////////////////////////////
     
 
-#ifdef SSE2
+#ifdef SSE4
 #include <pmmintrin.h>
 #endif
 #if defined(AVX1) || defined (AVX2)
@@ -33,6 +33,12 @@ namespace Grid {
 
   inline RealF adj(const RealF  & r){ return r; }
   inline RealF conj(const RealF  & r){ return r; }
+  inline RealF real(const RealF  & r){ return r; }
+
+  inline RealD adj(const RealD  & r){ return r; }
+  inline RealD conj(const RealD  & r){ return r; }
+  inline RealD real(const RealD  & r){ return r; }
+
   inline ComplexD innerProduct(const ComplexD & l, const ComplexD & r) { return conj(l)*r; }
   inline ComplexF innerProduct(const ComplexF & l, const ComplexF & r) { return conj(l)*r; }
   inline RealD innerProduct(const RealD & l, const RealD & r) { return l*r; }
@@ -60,8 +66,6 @@ namespace Grid {
     inline void mult(RealD * __restrict__ y,const RealD * __restrict__ l,const RealD *__restrict__ r){ *y = (*l) * (*r);}
     inline void sub (RealD * __restrict__ y,const RealD * __restrict__ l,const RealD *__restrict__ r){ *y = (*l) - (*r);}
     inline void add (RealD * __restrict__ y,const RealD * __restrict__ l,const RealD *__restrict__ r){ *y = (*l) + (*r);}
-    inline RealD adj(const RealD & r){ return r; }  // No-op for real
-    inline RealD conj(const RealD & r){ return r; }
     
     inline void mac (RealF * __restrict__ y,const RealF * __restrict__ a,const RealF *__restrict__ x){  *y = (*a) * (*x)+(*y); }
     inline void mult(RealF * __restrict__ y,const RealF * __restrict__ l,const RealF *__restrict__ r){ *y = (*l) * (*r); }
@@ -72,14 +76,14 @@ namespace Grid {
 
   class Zero{};
   static Zero zero;
-  template<class itype> inline void ZeroIt(itype &arg){ arg=zero;};
-  template<>            inline void ZeroIt(ComplexF &arg){ arg=0; };
-  template<>            inline void ZeroIt(ComplexD &arg){ arg=0; };
-  template<>            inline void ZeroIt(RealF &arg){ arg=0; };
-  template<>            inline void ZeroIt(RealD &arg){ arg=0; };
+  template<class itype> inline void zeroit(itype &arg){ arg=zero;};
+  template<>            inline void zeroit(ComplexF &arg){ arg=0; };
+  template<>            inline void zeroit(ComplexD &arg){ arg=0; };
+  template<>            inline void zeroit(RealF &arg){ arg=0; };
+  template<>            inline void zeroit(RealD &arg){ arg=0; };
 
 
-#if defined (SSE2)
+#if defined (SSE4)
     typedef __m128 fvec;
     typedef __m128d dvec;
     typedef __m128 cvec;
@@ -202,7 +206,7 @@ inline void Gpermute(vsimd &y,const vsimd &b,int perm){
       case 1: y.v = _mm256_shuffle_ps(b.v,b.v,_MM_SHUFFLE(1,0,3,2)); break;
       case 0: y.v = _mm256_permute2f128_ps(b.v,b.v,0x01); break;
 #endif
-#ifdef SSE2
+#ifdef SSE4
       case 1: y.v = _mm_shuffle_ps(b.v,b.v,_MM_SHUFFLE(2,3,0,1)); break;
       case 0: y.v = _mm_shuffle_ps(b.v,b.v,_MM_SHUFFLE(1,0,3,2));break;
 #endif
diff --git a/Grid_vComplexD.h b/Grid_vComplexD.h
index 133e354b..ffcff3a9 100644
--- a/Grid_vComplexD.h
+++ b/Grid_vComplexD.h
@@ -49,7 +49,7 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_add_pd(a.v,b.v);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_add_pd(a.v,b.v);
 #endif
 #ifdef AVX512
@@ -67,7 +67,7 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_sub_pd(a.v,b.v);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_sub_pd(a.v,b.v);
 #endif
 #ifdef AVX512
@@ -114,7 +114,7 @@ namespace Grid {
             ymm1 = _mm256_mul_pd(ymm1,ymm2);       // ymm1 <- br ai, ai bi
             ret.v= _mm256_addsub_pd(ymm0,ymm1);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             zvec ymm0,ymm1,ymm2;
             ymm0 = _mm_shuffle_pd(a.v,a.v,0x0); // ymm0 <- ar ar,
             ymm0 = _mm_mul_pd(ymm0,b.v);        // ymm0 <- ar bi, ar br
@@ -200,14 +200,14 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_set_pd(ig,rl,ig,rl);
 #endif
-#ifdef SSE2
-            ret.v = _mm_set_pd(a,b);
+#ifdef SSE4
+            ret.v = _mm_set_pd(ig,rl);
 #endif
 #ifdef AVX512
             ret.v = _mm512_set_pd(ig,rl,ig,rl,ig,rl,ig,rl);
 #endif
 #ifdef QPX
-            ret.v = {a,b,a,b};
+            ret.v = {ig,rl,ig,rl};
 #endif
         }
 
@@ -215,7 +215,7 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_set_pd(a[1].imag(),a[1].real(),a[0].imag(),a[0].real());
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_set_pd(a[0].imag(),a[0].real());
 #endif
 #ifdef AVX512
@@ -231,7 +231,7 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
 #if defined (AVX1)|| defined (AVX2)
        _mm256_store_pd((double *)a,ret.v);
 #endif
-#ifdef SSE2
+#ifdef SSE4
        _mm_store_pd((double *)a,ret.v);
 #endif
 #ifdef AVX512
@@ -257,7 +257,7 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
             __m256d tmp = _mm256_addsub_pd(ret.v,_mm256_shuffle_pd(in.v,in.v,0x5));
              ret.v=_mm256_shuffle_pd(tmp,tmp,0x5);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_addsub_pd(ret.v,in.v);
 #endif
 #ifdef AVX512
diff --git a/Grid_vComplexF.h b/Grid_vComplexF.h
index 75c47248..42fe5163 100644
--- a/Grid_vComplexF.h
+++ b/Grid_vComplexF.h
@@ -63,7 +63,7 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_add_ps(a.v,b.v);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_add_ps(a.v,b.v);
 #endif
 #ifdef AVX512
@@ -81,7 +81,7 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_sub_ps(a.v,b.v);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_sub_ps(a.v,b.v);
 #endif
 #ifdef AVX512
@@ -120,7 +120,7 @@ namespace Grid {
             ymm1 = _mm256_mul_ps(ymm1,ymm2);       // ymm1 <- br ai, ai bi
             ret.v= _mm256_addsub_ps(ymm0,ymm1);    
 #endif
-#ifdef SSE2
+#ifdef SSE4
             cvec ymm0,ymm1,ymm2;
             ymm0 = _mm_shuffle_ps(a.v,a.v,_MM_SHUFFLE(2,2,0,0)); // ymm0 <- ar ar,
             ymm0 = _mm_mul_ps(ymm0,b.v);        // ymm0 <- ar bi, ar br
@@ -156,8 +156,8 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_set_ps(a[3].imag(),a[3].real(),a[2].imag(),a[2].real(),a[1].imag(),a[1].real(),a[0].imag(),a[0].real());
 #endif
-#ifdef SSE2
-            ret.v = _mm_set_ps(a[1].imag, a[1].real(),a[0].imag(),a[0].real());
+#ifdef SSE4
+            ret.v = _mm_set_ps(a[1].imag(), a[1].real(),a[0].imag(),a[0].real());
 #endif
 #ifdef AVX512
             ret.v = _mm512_set_ps(a[7].imag(),a[7].real(),a[6].imag(),a[6].real(),a[5].imag(),a[5].real(),a[4].imag(),a[4].real(),a[3].imag(),a[3].real(),a[2].imag(),a[2].real(),a[1].imag(),a[1].real(),a[0].imag(),a[0].real());
@@ -181,7 +181,7 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
 #if defined (AVX1)|| defined (AVX2)
         _mm256_store_ps((float *)a,ret.v);
 #endif
-#ifdef SSE2
+#ifdef SSE4
         _mm_store_ps((float *)a,ret.v);
 #endif
 #ifdef AVX512
@@ -201,7 +201,7 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_set_ps(b,a,b,a,b,a,b,a);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_set_ps(a,b,a,b);
 #endif
 #ifdef AVX512
@@ -213,7 +213,11 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
         }
        friend inline ComplexF Reduce(const vComplexF & in)
        {
+#ifdef SSE4
+#error
+#endif
 #if defined (AVX1) || defined(AVX2)
+	 // FIXME this is inefficient and use 
          __attribute__ ((aligned(32))) float c_[8];
          _mm256_store_ps(c_,in.v);
          return ComplexF(c_[0]+c_[2]+c_[4]+c_[6],c_[1]+c_[3]+c_[5]+c_[7]);
@@ -305,7 +309,7 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
              tmp = _mm256_addsub_ps(ret.v,_mm256_shuffle_ps(in.v,in.v,_MM_SHUFFLE(2,3,0,1))); // ymm1 <- br,bi
              ret.v=_mm256_shuffle_ps(tmp,tmp,_MM_SHUFFLE(2,3,0,1));
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_addsub_ps(ret.v,in.v);
 #endif
 #ifdef AVX512
diff --git a/Grid_vInteger.h b/Grid_vInteger.h
index 82adbd8e..672d4938 100644
--- a/Grid_vInteger.h
+++ b/Grid_vInteger.h
@@ -48,7 +48,7 @@ namespace Grid {
 #if defined (AVX2)
             ret.v = _mm256_add_epi32(a.v,b.v);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_add_epi32(a.v,b.v);
 #endif
 #ifdef AVX512
@@ -78,7 +78,7 @@ namespace Grid {
 #if defined (AVX2)
             ret.v = _mm256_sub_epi32(a.v,b.v);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_sub_epi32(a.v,b.v);
 #endif
 #ifdef AVX512
@@ -108,7 +108,7 @@ namespace Grid {
 #if defined (AVX2)
             ret.v = _mm256_mul_epi32(a.v,b.v);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_mul_epi32(a.v,b.v);
 #endif
 #ifdef AVX512
@@ -147,7 +147,7 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_set1_epi32(a);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_set1_epi32(a);
 #endif
 #ifdef AVX512
@@ -161,7 +161,7 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
 	  ret.v = _mm256_set_epi32(a[7],a[6],a[5],a[4],a[3],a[2],a[1],a[0]);
 #endif
-#ifdef SSE2
+#ifdef SSE4
 	  ret.v = _mm_set_epi32(a[0],a[1],a[2],a[3]);
 #endif
 #ifdef AVX512
@@ -177,8 +177,8 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
 	  _mm256_store_si256((__m256i*)a,ret.v);
 #endif
-#ifdef SSE2
-	  _mm_store_si128(a,ret.v);
+#ifdef SSE4
+	  _mm_store_si128((__m128i *)a,ret.v);
 #endif
 #ifdef AVX512
 	  _mm512_store_si512(a,ret.v);
diff --git a/Grid_vRealD.h b/Grid_vRealD.h
index 86f738f3..0f88f7da 100644
--- a/Grid_vRealD.h
+++ b/Grid_vRealD.h
@@ -24,7 +24,7 @@ namespace Grid {
         friend inline vRealD conj(const vRealD &in){ return in; }
         
         friend inline void mac (vRealD &y,const vRealD a,const vRealD x){
-#if defined (AVX1) || defined (SSE2)
+#if defined (AVX1) || defined (SSE4)
             y = a*x+y;
 #endif
 #ifdef AVX2     // AVX 2 introduced FMA support. FMA4 eliminates a copy, but AVX only has FMA3
@@ -55,7 +55,7 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_add_pd(a.v,b.v);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_add_pd(a.v,b.v);
 #endif
 #ifdef AVX512
@@ -72,7 +72,7 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_sub_pd(a.v,b.v);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_sub_pd(a.v,b.v);
 #endif
 #ifdef AVX512
@@ -90,7 +90,7 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_mul_pd(a.v,b.v);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_mul_pd(a.v,b.v);
 #endif
 #ifdef AVX512
@@ -133,7 +133,7 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_set_pd(a,a,a,a);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_set_pd(a,a);
 #endif
 #ifdef AVX512
@@ -147,7 +147,7 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_set_pd(a[3],a[2],a[1],a[0]);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_set_pd(a[0],a[1]);
 #endif
 #ifdef AVX512
@@ -163,7 +163,7 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             _mm256_store_pd(a,ret.v);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             _mm_store_pd(a,ret.v);
 #endif
 #ifdef AVX512
diff --git a/Grid_vRealF.h b/Grid_vRealF.h
index b7c127ce..78c37177 100644
--- a/Grid_vRealF.h
+++ b/Grid_vRealF.h
@@ -26,7 +26,7 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_add_ps(a.v,b.v);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_add_ps(a.v,b.v);
 #endif
 #ifdef AVX512
@@ -48,7 +48,7 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_sub_ps(a.v,b.v);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_sub_ps(a.v,b.v);
 #endif
 #ifdef AVX512
@@ -70,7 +70,7 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_mul_ps(a.v,b.v);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_mul_ps(a.v,b.v);
 #endif
 #ifdef AVX512
@@ -96,7 +96,7 @@ namespace Grid {
         friend inline vRealF conj(const vRealF &in){ return in; }
 
         friend inline void mac (vRealF &y,const vRealF a,const vRealF x){
-#if defined (AVX1) || defined (SSE2)
+#if defined (AVX1) || defined (SSE4)
             y = a*x+y;
 #endif
 #ifdef AVX2     // AVX 2 introduced FMA support. FMA4 eliminates a copy, but AVX only has FMA3
@@ -158,7 +158,7 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_set_ps(a,a,a,a,a,a,a,a);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_set_ps(a,a,a,a);
 #endif
 #ifdef AVX512
@@ -175,7 +175,7 @@ namespace Grid {
 #if defined (AVX1)|| defined (AVX2)
             ret.v = _mm256_set_ps(a[7],a[6],a[5],a[4],a[3],a[2],a[1],a[0]);
 #endif
-#ifdef SSE2
+#ifdef SSE4
             ret.v = _mm_set_ps(a[0],a[1],a[2],a[3]);
 #endif
 #ifdef AVX512
@@ -195,7 +195,7 @@ friend inline void vstore(const vRealF &ret, float *a){
 #if defined (AVX1)|| defined (AVX2)
 	_mm256_store_ps(a,ret.v);
 #endif
-#ifdef SSE2
+#ifdef SSE4
 	_mm_store_ps(a,ret.v);
 #endif
 #ifdef AVX512
diff --git a/test_Grid_stencil.cc b/test_Grid_stencil.cc
index 8077562f..fd2a9503 100644
--- a/test_Grid_stencil.cc
+++ b/test_Grid_stencil.cc
@@ -29,14 +29,14 @@ int main (int argc, char ** argv)
     
 #ifdef AVX512
     simd_layout[0] = 1;
-    simd_layout[1] = 1;
+    simd_layout[1] = 2;
     simd_layout[2] = 2;
     simd_layout[3] = 2;
 #endif
 #if defined (AVX1)|| defined (AVX2)
     simd_layout[0] = 1;
     simd_layout[1] = 1;
-    simd_layout[2] = 1;
+    simd_layout[2] = 2;
     simd_layout[3] = 2;
 #endif
 #if defined (SSE2)

From 1972eea128295b5091cf28339c5322d0baa1f82b Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 16 Apr 2015 14:48:21 +0100
Subject: [PATCH 054/429] spin trace type work

---
 Grid_config.h.in |  4 ++--
 Makefile.in      |  6 +++---
 TODO             | 17 ++++++++++-------
 configure        |  6 +++---
 configure.ac     |  6 +++---
 5 files changed, 21 insertions(+), 18 deletions(-)

diff --git a/Grid_config.h.in b/Grid_config.h.in
index 91948fa2..660edd5d 100644
--- a/Grid_config.h.in
+++ b/Grid_config.h.in
@@ -72,8 +72,8 @@
 /* Define to the version of this package. */
 #undef PACKAGE_VERSION
 
-/* SSE2 */
-#undef SSE2
+/* SSE4 */
+#undef SSE4
 
 /* Define to 1 if you have the ANSI C header files. */
 #undef STDC_HEADERS
diff --git a/Makefile.in b/Makefile.in
index 11b86688..ce8b74be 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -158,7 +158,7 @@ PROGRAMS = $(bin_PROGRAMS)
 am_Grid_main_OBJECTS = Grid_main.$(OBJEXT)
 Grid_main_OBJECTS = $(am_Grid_main_OBJECTS)
 Grid_main_DEPENDENCIES = libGrid.a
-am_test_Grid_stencil_OBJECTS = test_Grid_Stencil.$(OBJEXT)
+am_test_Grid_stencil_OBJECTS = test_Grid_stencil.$(OBJEXT)
 test_Grid_stencil_OBJECTS = $(am_test_Grid_stencil_OBJECTS)
 test_Grid_stencil_DEPENDENCIES = libGrid.a
 AM_V_P = $(am__v_P_@AM_V@)
@@ -370,7 +370,7 @@ include_HEADERS = Grid_config.h\
 
 Grid_main_SOURCES = Grid_main.cc
 Grid_main_LDADD = libGrid.a
-test_Grid_stencil_SOURCES = test_Grid_Stencil.cc
+test_Grid_stencil_SOURCES = test_Grid_stencil.cc
 test_Grid_stencil_LDADD = libGrid.a
 all: Grid_config.h
 	$(MAKE) $(AM_MAKEFLAGS) all-am
@@ -523,7 +523,7 @@ distclean-compile:
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_init.Po@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_main.Po@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_stencil_common.Po@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/test_Grid_Stencil.Po@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/test_Grid_stencil.Po@am__quote@
 
 .cc.o:
 @am__fastdepCXX_TRUE@	$(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
diff --git a/TODO b/TODO
index 010a06b7..5f21f520 100644
--- a/TODO
+++ b/TODO
@@ -7,18 +7,16 @@ AUDITS:
 * extract / merge extra implementation removal      
 
 
+BUILD:
+* Test infrastructure
+* subdirs lib, tests ??
+
 FUNCTIONALITY:
-
 * Conditional execution, where etc...                -----DONE, simple test
-
 * Integer relational support                         -----DONE
-
 * Coordinate information, integers etc...            -----DONE
-
 * Integer type padding/union to vector.              -----DONE 
-
 * LatticeCoordinate[mu]                              -----DONE
-
 * Stencil operator support                           -----Initial thoughts, trial implementation DONE.
                                                      -----some simple tests that Stencil matches Cshift.
                                                      -----do all permute in comms phase, so that copy permute
@@ -27,6 +25,12 @@ FUNCTIONALITY:
 * CovariantShift support                             -----Use a class to store gauge field? (parallel transport?)
 
 
+POST-SFW call:
+
+* TraceColor, TraceSpin.
+* How to define simple matrix operations, such as flavour matrices?
+
+
 * Subset support, slice sums etc...                  -----Only need slice sum?
                                                      -----Generic cartesian subslicing?
                                                      -----Array ranges / boost extents?
@@ -34,7 +38,6 @@ FUNCTIONALITY:
                                                      -----Suggests generalised cartesian subblocking
                                                           sums, returning modified grid?
 					             -----What should interface be?
-
 i)  Two classes of subset;   red black parity subsetting (pick checkerboard).
                              cartesian sub-block subsetting
 
diff --git a/configure b/configure
index c6f93abc..1adf726f 100755
--- a/configure
+++ b/configure
@@ -5009,10 +5009,10 @@ fi
 
 
 case ${ac_SIMD} in
-     SSE2)
-       echo Configuring for SSE2
+     SSE4)
+       echo Configuring for SSE4
 
-$as_echo "#define SSE2 1" >>confdefs.h
+$as_echo "#define SSE4 1" >>confdefs.h
 
      ;;
      AVX)
diff --git a/configure.ac b/configure.ac
index bfe745e5..83405ca8 100644
--- a/configure.ac
+++ b/configure.ac
@@ -29,9 +29,9 @@ AC_CHECK_FUNCS([gettimeofday])
 AC_ARG_ENABLE([simd],[AC_HELP_STRING([--enable-simd=SSE|AVX|AVX2|AVX512],[Select instructions])],[ac_SIMD=${enable_simd}],[ac_SIMD=AVX2])
 
 case ${ac_SIMD} in
-     SSE2)
-       echo Configuring for SSE2
-       AC_DEFINE([SSE2],[1],[SSE2] )
+     SSE4)
+       echo Configuring for SSE4
+       AC_DEFINE([SSE4],[1],[SSE4] )
      ;;
      AVX)
        echo Configuring for AVX

From 6b04dd4a5d7aac3e88be18409c7a7fdcd773dd18 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 16 Apr 2015 15:20:19 +0100
Subject: [PATCH 055/429] Better code

---
 Grid_Lattice.h    |   4 +-
 Grid_math_types.h | 105 ++++++++++++++++++++++++----------------------
 2 files changed, 57 insertions(+), 52 deletions(-)

diff --git a/Grid_Lattice.h b/Grid_Lattice.h
index a94a13a4..8075df94 100644
--- a/Grid_Lattice.h
+++ b/Grid_Lattice.h
@@ -156,7 +156,8 @@ public:
         }
     };
 
-    // FIXME for debug; deprecate this
+    // FIXME for debug; deprecate this; made obscelete by 
+    // LatticeCoordinate();
     friend void lex_sites(Lattice<vobj> &l){
       Real *v_ptr = (Real *)&l._odata[0];
       size_t o_len = l._grid->oSites();
@@ -211,6 +212,7 @@ public:
         return *this;
     }
     
+    // FIXME trace type structure is weird
     inline friend Lattice<iScalar<vComplex> > _trace(const Lattice<vobj> &lhs){
         Lattice<iScalar<vComplex> > ret(lhs._grid);
 #pragma omp parallel for
diff --git a/Grid_math_types.h b/Grid_math_types.h
index 58de9c15..9a9f1274 100644
--- a/Grid_math_types.h
+++ b/Grid_math_types.h
@@ -1059,6 +1059,23 @@ template<class vtype> inline iScalar<vtype> conj(const iScalar<vtype>&r)
     ret._internal = conj(r._internal);
     return ret;
 }
+template<class vtype,int N> inline iVector<vtype,N> conj(const iVector<vtype,N>&r)
+{
+  iVector<vtype,N> ret;
+  for(int i=0;i<N;i++){
+    ret._internal[i] = conj(r._internal[i]);
+  }
+  return ret;
+}
+template<class vtype,int N> inline iMatrix<vtype,N> conj(const iMatrix<vtype,N>&r)
+{
+  iMatrix<vtype,N> ret;
+  for(int i=0;i<N;i++){
+  for(int j=0;j<N;j++){
+    ret._internal[i][j] = conj(r._internal[i][j]);
+  }}
+  return ret;
+}
 
 // Adj function for scalar, vector, matrix
 template<class vtype> inline iScalar<vtype> adj(const iScalar<vtype>&r)
@@ -1085,8 +1102,14 @@ template<class vtype,int N> inline iMatrix<vtype,N> adj(const iMatrix<vtype,N> &
     return ret;
 }
 /////////////////////////////////////////////////////////////////
-// Transpose
+// Transpose all indices
 /////////////////////////////////////////////////////////////////
+
+inline ComplexD transpose(ComplexD &rhs){  return rhs;}
+inline ComplexF transpose(ComplexF &rhs){  return rhs;}
+inline RealD transpose(RealD &rhs){  return rhs;}
+inline RealF transpose(RealF &rhs){  return rhs;}
+
 template<class vtype,int N>
   inline typename std::enable_if<isGridTensor<vtype>::value, iMatrix<vtype,N> >::type 
   transpose(iMatrix<vtype,N> arg)
@@ -1128,19 +1151,9 @@ template<class vtype,int N>
     return ret;
   }
 
-
-
-/*
- *  No need to implement transpose on the primitive types
- *  Not sure that this idiom is any more elegant that the trace idiom below however!
-inline ComplexD transpose(ComplexD &rhs){  return rhs;}
-inline ComplexF transpose(ComplexF &rhs){  return rhs;}
-inline RealD transpose(RealD &rhs){  return rhs;}
-inline RealF transpose(RealF &rhs){  return rhs;}
- */
-
 ////////////////////////////////////////////////////////////////////////////////////////////
-// Transpose a specific index
+// Transpose a specific index; instructive to compare this style of recursion termination
+// to that of adj; which is easiers?
 ////////////////////////////////////////////////////////////////////////////////////////////
 template<int Level,class vtype,int N> inline 
   typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value, iMatrix<vtype,N> >::type 
@@ -1171,15 +1184,40 @@ transposeIndex (const iScalar<vtype> &arg)
 {
   return transposeIndex<Level>(arg._internal);
 }
-////////////////////////////////
-// Trace a specific index 
-////////////////////////////////
+
+//////////////////////////////////////////////////////////////////
+// Traces: both all indices and a specific index 
+/////////////////////////////////////////////////////////////////
+
+inline ComplexF trace( const ComplexF &arg){    return arg;}
+inline ComplexD trace( const ComplexD &arg){    return arg;}
+inline RealF trace( const RealF &arg){    return arg;}
+inline RealD trace( const RealD &arg){    return arg;}
 
 template<int Level> inline ComplexF traceIndex(const ComplexF arg) { return arg;}
 template<int Level> inline ComplexD traceIndex(const ComplexD arg) { return arg;}
 template<int Level> inline RealF traceIndex(const RealF arg) { return arg;}
 template<int Level> inline RealD traceIndex(const RealD arg) { return arg;}
 
+template<class vtype,int N>
+inline auto trace(const iMatrix<vtype,N> &arg) -> iScalar<decltype(trace(arg._internal[0][0]))>
+{
+    iScalar<decltype( trace(arg._internal[0][0] )) > ret;
+    zeroit(ret._internal);
+    for(int i=0;i<N;i++){
+        ret._internal=ret._internal+trace(arg._internal[i][i]);
+    }
+    return ret;
+}
+template<class vtype>
+inline auto trace(const iScalar<vtype> &arg) -> iScalar<decltype(trace(arg._internal))>
+{
+    iScalar<decltype(trace(arg._internal))> ret;
+    ret._internal=trace(arg._internal);
+    return ret;
+}
+
+// Specific indices.
 template<int Level,class vtype> inline 
 auto traceIndex(const iScalar<vtype> &arg) -> iScalar<decltype(traceIndex<Level>(arg._internal)) >
 {
@@ -1268,41 +1306,6 @@ template<class itype,int N> inline auto imag(const iVector<itype,N> &z) -> iVect
     return ret;
 }
 
-    /////////////////////////////////
-    // Trace of scalar and matrix
-    /////////////////////////////////
-
-inline ComplexF trace( const ComplexF &arg){
-    return arg;
-}
-inline ComplexD trace( const ComplexD &arg){
-    return arg;
-}
-inline RealF trace( const RealF &arg){
-    return arg;
-}
-inline RealD trace( const RealD &arg){
-    return arg;
-}
-
-template<class vtype,int N>
-inline auto trace(const iMatrix<vtype,N> &arg) -> iScalar<decltype(trace(arg._internal[0][0]))>
-{
-    iScalar<decltype( trace(arg._internal[0][0] )) > ret;
-    zeroit(ret._internal);
-    for(int i=0;i<N;i++){
-        ret._internal=ret._internal+trace(arg._internal[i][i]);
-    }
-    return ret;
-}
-template<class vtype>
-inline auto trace(const iScalar<vtype> &arg) -> iScalar<decltype(trace(arg._internal))>
-{
-    iScalar<decltype(trace(arg._internal))> ret;
-    ret._internal=trace(arg._internal);
-    return ret;
-}
-
 };
     
 #endif

From 3b9110b5db1f9d6d695f713a281ed27f41619c67 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 16 Apr 2015 17:22:52 +0100
Subject: [PATCH 056/429] SSE flag changed

---
 Grid_main.cc | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/Grid_main.cc b/Grid_main.cc
index e122fcbe..291cde25 100644
--- a/Grid_main.cc
+++ b/Grid_main.cc
@@ -48,7 +48,7 @@ int main (int argc, char ** argv)
     simd_layout[2] = 2;
     simd_layout[3] = 2;
 #endif
-#if defined (SSE2)
+#if defined (SSE4)
     simd_layout[0] = 1;
     simd_layout[1] = 1;
     simd_layout[2] = 1;

From 1408a3c0f9fe3f117dd42008cc257991fc2e8e3a Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 12:17:13 +0100
Subject: [PATCH 057/429] Got traceIndex, transposeIndex fully working. Need to
 think about peekIndex interface and () based indexing.

---
 Grid_summation.h | 12 ++++++++++++
 1 file changed, 12 insertions(+)
 create mode 100644 Grid_summation.h

diff --git a/Grid_summation.h b/Grid_summation.h
new file mode 100644
index 00000000..dff2df61
--- /dev/null
+++ b/Grid_summation.h
@@ -0,0 +1,12 @@
+#ifndef GRID_SUMMATION_H
+#define GRID_SUMMATION_H
+
+template<class vobj>
+inline void sumBlocks(Lattice<vobj> &coarseData,const Lattice<vobj &fineData)
+{
+  GridBase * fine  = findData._grid;
+  GridBase * coarse= findData._grid;
+  
+  return;
+}
+#endif

From 56b36311871c28d93237fe8f2e6823e18e216f5c Mon Sep 17 00:00:00 2001
From: paboyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 12:21:37 +0100
Subject: [PATCH 058/429] Update README.md

---
 README.md | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index 42537417..74c03a33 100644
--- a/README.md
+++ b/README.md
@@ -12,7 +12,8 @@ array indices to both MPI tasks and SIMD processing elements.
 The transformation is based on the observation that Cartesian array processing involves
 identical processing to be performed on different regions of the Cartesian array.
 
-The library will (eventually) both geometrically decompose into MPI tasks and across SIMD lanes.
+The library will both geometrically decompose into MPI tasks and across SIMD lanes.
+Local vector loops are parallelised with OpenMP pragmas.
 
 Data parallel array operations can then be specified with a SINGLE data parallel paradigm, but
 optimally use MPI, OpenMP and SIMD parallelism under the hood. This is a significant simplification
@@ -30,7 +31,7 @@ The corresponding scalar types are named
 
   RealF, RealD, ComplexF, ComplexD
 
-MPI parallelism is UNIMPLEMENTED and for now only OpenMP and SIMD parallelism is present in the library.
+MPI, OpenMP, and SIMD parallelism are present in the library.
 
    You can give `configure' initial values for configuration parameters
 by setting variables in the command line or in the environment.  Here

From 5d1b866e7a473d894883425bb9d097e6bb3c3690 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 12:40:55 +0100
Subject: [PATCH 059/429] typo

---
 README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 74c03a33..80714d76 100644
--- a/README.md
+++ b/README.md
@@ -35,7 +35,7 @@ MPI, OpenMP, and SIMD parallelism are present in the library.
 
    You can give `configure' initial values for configuration parameters
 by setting variables in the command line or in the environment.  Here
-is are examples:
+are examples:
 
      ./configure CXX=clang++ CXXFLAGS="-std=c++11 -O3 -mavx" --enable-simd=AVX1
 

From e25f10566c83f4d1bf5ccc9ecd3b850b5fc57ab4 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 14:36:01 +0100
Subject: [PATCH 060/429] peekIndex update

---
 Grid_Lattice.h       | 170 ++++++++++++++++++++++++----
 Grid_QCD.h           |  73 ++++++++----
 Grid_main.cc         |  85 +++++++++++---
 Grid_math_types.h    | 263 ++++++++++++++++++++++++++++++++++++++++---
 Grid_simd.h          |   1 +
 Grid_vRealD.h        |   5 +-
 Grid_vRealF.h        |   4 +-
 TODO                 |  58 +++++++---
 test_Grid_stencil.cc |  12 +-
 9 files changed, 562 insertions(+), 109 deletions(-)

diff --git a/Grid_Lattice.h b/Grid_Lattice.h
index 8075df94..b595ae0a 100644
--- a/Grid_Lattice.h
+++ b/Grid_Lattice.h
@@ -6,7 +6,23 @@
 
 namespace Grid {
 
-  extern int GridCshiftPermuteMap[4][16];
+// TODO: Indexing ()
+//       mac,real,imag
+//
+// Functionality required:
+//     -=,+=,*=,()
+//     add,+,sub,-,mult,mac,*
+//     adj,conj
+//     real,imag
+//     transpose,transposeIndex  
+//     trace,traceIndex
+//     peekIndex
+//     innerProduct,outerProduct,
+//     localNorm2
+//     localInnerProduct
+//     
+
+extern int GridCshiftPermuteMap[4][16];
 
 template<class vobj>
 class Lattice
@@ -41,6 +57,9 @@ public:
     template<class obj1,class obj2,class obj3>
     friend void mult(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs);
 
+    template<class obj1,class obj2,class obj3>
+    friend void mac(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs);
+
     template<class obj1,class obj2,class obj3>
     friend void sub(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs);
 
@@ -212,26 +231,6 @@ public:
         return *this;
     }
     
-    // FIXME trace type structure is weird
-    inline friend Lattice<iScalar<vComplex> > _trace(const Lattice<vobj> &lhs){
-        Lattice<iScalar<vComplex> > ret(lhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-            ret._odata[ss] = trace(lhs._odata[ss]);
-        }
-        return ret;
-    };
-    
-    inline friend Lattice<iScalar<iScalar< vComplex > > > trace(const Lattice<vobj> &lhs){
-        Lattice<iScalar< iScalar<vComplex> >  > ret(lhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-            ret._odata[ss] = trace(lhs._odata[ss]);
-        }
-        return ret;
-    };
-
-    
     inline friend Lattice<vobj> adj(const Lattice<vobj> &lhs){
         Lattice<vobj> ret(lhs._grid);
 #pragma omp parallel for
@@ -241,6 +240,16 @@ public:
         return ret;
     };
 
+    inline friend Lattice<vobj> transpose(const Lattice<vobj> &lhs){
+        Lattice<vobj> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+            ret._odata[ss] = transpose(lhs._odata[ss]);
+        }
+        return ret;
+    };
+
+
     inline friend Lattice<vobj> conj(const Lattice<vobj> &lhs){
         Lattice<vobj> ret(lhs._grid);
 #pragma omp parallel for
@@ -304,6 +313,17 @@ public:
 	  mult(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
         }
     }
+
+    template<class obj1,class obj2,class obj3>
+    void mac(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
+        conformable(lhs,rhs);
+	uint32_t vec_len = lhs._grid->oSites();
+#pragma omp parallel for
+        for(int ss=0;ss<vec_len;ss++){
+	  mac(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
+        }
+    }
+
     template<class obj1,class obj2,class obj3>
     void sub(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
         conformable(lhs,rhs);
@@ -409,6 +429,82 @@ public:
         return ret;
     }
 
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    // Trace
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    template<class vobj>
+    inline auto trace(const Lattice<vobj> &lhs)
+      -> Lattice<decltype(trace(lhs._odata[0]))>
+    {
+      Lattice<decltype(trace(lhs._odata[0]))> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+            ret._odata[ss] = trace(lhs._odata[ss]);
+        }
+        return ret;
+    };
+    
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    // Index level dependent operations
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    template<int Index,class vobj>
+    inline auto traceIndex(const Lattice<vobj> &lhs)
+      -> Lattice<decltype(traceIndex<Index>(lhs._odata[0]))>
+    {
+      Lattice<decltype(traceIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
+#pragma omp parallel for
+      for(int ss=0;ss<lhs._grid->oSites();ss++){
+	ret._odata[ss] = traceIndex<Index>(lhs._odata[ss]);
+      }
+      return ret;
+    };
+    template<int Index,class vobj>
+    inline auto transposeIndex(const Lattice<vobj> &lhs)
+      -> Lattice<decltype(transposeIndex<Index>(lhs._odata[0]))>
+    {
+      Lattice<decltype(transposeIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+            ret._odata[ss] = transposeIndex<Index>(lhs._odata[ss]);
+        }
+        return ret;
+    };
+
+    // Fixme; this is problematic since the number of args is variable and 
+    // may mismatch...
+    template<int Index,class vobj>
+    inline auto peekIndex(const Lattice<vobj> &lhs)
+      -> Lattice<decltype(peekIndex<Index>(lhs._odata[0]))>
+    {
+      Lattice<decltype(peekIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+            ret._odata[ss] = peekIndex<Index>(lhs._odata[ss]);
+        }
+        return ret;
+    };
+    template<int Index,class vobj>
+      inline auto peekIndex(const Lattice<vobj> &lhs,int i)
+      -> Lattice<decltype(peekIndex<Index>(lhs._odata[0],i))>
+    {
+      Lattice<decltype(peekIndex<Index>(lhs._odata[0],i))> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+	  ret._odata[ss] = peekIndex<Index>(lhs._odata[ss],i);
+        }
+        return ret;
+    };
+    template<int Index,class vobj>
+      inline auto peekIndex(const Lattice<vobj> &lhs,int i,int j)
+      -> Lattice<decltype(peekIndex<Index>(lhs._odata[0],i,j))>
+    {
+      Lattice<decltype(peekIndex<Index>(lhs._odata[0],i,j))> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+	  ret._odata[ss] = peekIndex<Index>(lhs._odata[ss],i,j);
+        }
+        return ret;
+    };
     ////////////////////////////////////////////////////////////////////////////////////////////////////
     // Reduction operations
     ////////////////////////////////////////////////////////////////////////////////////////////////////
@@ -416,14 +512,16 @@ public:
     inline RealD norm2(const Lattice<vobj> &arg){
 
       typedef typename vobj::scalar_type scalar;
+      typedef typename vobj::vector_type vector;
       decltype(innerProduct(arg._odata[0],arg._odata[0])) vnrm=zero;
-
       scalar nrm;
       //FIXME make this loop parallelisable
+      vnrm=zero;
       for(int ss=0;ss<arg._grid->oSites(); ss++){
 	vnrm = vnrm + innerProduct(arg._odata[ss],arg._odata[ss]);
       }
-      nrm = Reduce(TensorRemove(vnrm));
+      vector vvnrm =TensorRemove(vnrm) ;
+      nrm = Reduce(vvnrm);
       arg._grid->GlobalSum(nrm);
       return real(nrm);
     }
@@ -439,7 +537,7 @@ public:
       for(int ss=0;ss<left._grid->oSites(); ss++){
 	vnrm = vnrm + innerProduct(left._odata[ss],right._odata[ss]);
       }
-      nrm = Reduce(TensorRemove(vnrm));
+      nrm = Reduce(vnrm);
       right._grid->GlobalSum(nrm);
       return nrm;
     }
@@ -455,10 +553,32 @@ public:
       Lattice<typename vobj::tensor_reduced> ret(rhs._grid);
 #pragma omp parallel for
         for(int ss=0;ss<rhs._grid->oSites(); ss++){
-	  ret._odata[ss]=innerProduct(rhs,rhs);
+	  ret._odata[ss]=innerProduct(rhs._odata[ss],rhs._odata[ss]);
         }
         return ret;
     }
+    
+    template<class vobj>
+    inline auto real(const Lattice<vobj> &z) -> Lattice<decltype(real(z._odata[0]))>
+    {
+      Lattice<decltype(real(z._odata[0]))> ret(z._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<z._grid->oSites();ss++){
+            ret._odata[ss] = real(z._odata[ss]);
+        }
+      return ret;
+    }
+
+    template<class vobj>
+    inline auto imag(const Lattice<vobj> &z) -> Lattice<decltype(imag(z._odata[0]))>
+    {
+      Lattice<decltype(imag(z._odata[0]))> ret(z._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<z._grid->oSites();ss++){
+            ret._odata[ss] = imag(z._odata[ss]);
+        }
+      return ret;
+    }
 
     // localInnerProduct
     template<class vobj>
diff --git a/Grid_QCD.h b/Grid_QCD.h
index 2f411f16..6f85d692 100644
--- a/Grid_QCD.h
+++ b/Grid_QCD.h
@@ -5,57 +5,80 @@ namespace QCD {
 
     static const int Nc=3;
     static const int Ns=4;
+    static const int Nd=4;
+
     static const int CbRed  =0;
     static const int CbBlack=1;
     
+    //////////////////////////////////////////////////////////////////////////////
     // QCD iMatrix types
-    template<typename vtype> using iSinglet          = iScalar<iScalar<vtype> > ;
-    template<typename vtype> using iSpinMatrix       = iMatrix<iScalar<vtype>, Ns>;
-    template<typename vtype> using iSpinColourMatrix = iMatrix<iMatrix<vtype, Nc>, Ns>;
-    template<typename vtype> using iColourMatrix     = iScalar<iMatrix<vtype, Nc>> ;
+    // Index conventions:                            Lorentz x Spin x Colour
+    //
+    // ChrisK very keen to add extra space for Gparity doubling.
+    //
+    // Also add domain wall index, in a way where Wilson operator 
+    // naturally distributes across the 5th dimensions.
+    //////////////////////////////////////////////////////////////////////////////
+    template<typename vtype> using iSinglet          = iScalar<iScalar<iScalar<vtype> > >;
+    template<typename vtype> using iSpinMatrix       = iScalar<iMatrix<iScalar<vtype>, Ns> >;
+    template<typename vtype> using iSpinColourMatrix = iScalar<iMatrix<iMatrix<vtype, Nc>, Ns> >;
+    template<typename vtype> using iColourMatrix     = iScalar<iScalar<iMatrix<vtype, Nc> > > ;
+    template<typename vtype> using iLorentzColourMatrix = iVector<iScalar<iMatrix<vtype, Nc> >, Nd > ;
 
-    template<typename vtype> using iSpinVector       = iVector<iScalar<vtype>, Ns>;
-    template<typename vtype> using iColourVector     = iScalar<iVector<vtype, Nc> >;
-    template<typename vtype> using iSpinColourVector = iVector<iVector<vtype, Nc>, Ns>;
 
-    typedef iSinglet<Complex >          TComplex;    // This is painful. Tensor singlet complex type.
-    typedef iSinglet<vComplex >         vTComplex;   // what if we don't know the tensor structure
-    typedef iSinglet<Real >             TReal;       // Shouldn't need these.
-    typedef iSinglet<vInteger >         vTInteger;
+    template<typename vtype> using iSpinVector       = iScalar<iVector<iScalar<vtype>, Ns> >;
+    template<typename vtype> using iColourVector     = iScalar<iScalar<iVector<vtype, Nc> > >;
+    template<typename vtype> using iSpinColourVector = iScalar<iVector<iVector<vtype, Nc>, Ns> >;
 
-    typedef iSpinMatrix<Complex >       SpinMatrix;
-    typedef iColourMatrix<Complex >     ColourMatrix;
-    typedef iSpinColourMatrix<Complex > SpinColourMatrix;
+    typedef iSpinMatrix<Complex >          SpinMatrix;
+    typedef iColourMatrix<Complex >        ColourMatrix;
+    typedef iSpinColourMatrix<Complex >    SpinColourMatrix;
+    typedef iLorentzColourMatrix<Complex > LorentzColourMatrix;
 
     typedef iSpinVector<Complex >       SpinVector;
     typedef iColourVector<Complex >     ColourVector;
     typedef iSpinColourVector<Complex > SpinColourVector;
 
     
-    typedef iSpinMatrix<vComplex >       vSpinMatrix;
-    typedef iColourMatrix<vComplex >     vColourMatrix;
-    typedef iSpinColourMatrix<vComplex > vSpinColourMatrix;
+    typedef iSpinMatrix<vComplex >          vSpinMatrix;
+    typedef iColourMatrix<vComplex >        vColourMatrix;
+    typedef iSpinColourMatrix<vComplex >    vSpinColourMatrix;
+    typedef iLorentzColourMatrix<vComplex > vLorentzColourMatrix;
     
     typedef iSpinVector<vComplex >       vSpinVector;
     typedef iColourVector<vComplex >     vColourVector;
     typedef iSpinColourVector<vComplex > vSpinColourVector;
     
-    typedef Lattice<vTComplex>            LatticeComplex;
+    typedef iSinglet<Complex >          TComplex;    // This is painful. Tensor singlet complex type.
+    typedef iSinglet<vComplex >         vTComplex;   // what if we don't know the tensor structure
+    typedef iSinglet<Real >             TReal;       // Shouldn't need these; can I make it work without?
+    typedef iSinglet<vReal >            vTReal;      
+    typedef iSinglet<vInteger >         vTInteger;
+    typedef iSinglet<Integer >          TInteger;
+
+    typedef Lattice<vTReal>              LatticeReal;
+    typedef Lattice<vTComplex>           LatticeComplex;
     typedef Lattice<vInteger>            LatticeInteger; // Predicates for "where"
     
     typedef Lattice<vColourMatrix>     LatticeColourMatrix;
     typedef Lattice<vSpinMatrix>       LatticeSpinMatrix;
-    typedef Lattice<vSpinColourMatrix> LatticePropagator;
-    typedef LatticePropagator LatticeSpinColourMatrix;
+    typedef Lattice<vSpinColourMatrix> LatticeSpinColourMatrix;
 
-    typedef Lattice<vSpinColourVector> LatticeFermion;
     typedef Lattice<vSpinColourVector> LatticeSpinColourVector;
     typedef Lattice<vSpinVector>       LatticeSpinVector;
     typedef Lattice<vColourVector>     LatticeColourVector;
 
+    ///////////////////////////////////////////
+    // Physical names for things
+    ///////////////////////////////////////////
+    typedef Lattice<vSpinColourVector> LatticeFermion;
+    typedef Lattice<vSpinColourMatrix> LatticePropagator;
+    typedef Lattice<vLorentzColourMatrix> LatticeGaugeField;
 
-    // FIXME for debug; deprecate this
-   inline void LatticeCoordinate(LatticeInteger &l,int mu){
+
+    
+
+     inline void LatticeCoordinate(LatticeInteger &l,int mu){
       GridBase *grid = l._grid;
       int Nsimd = grid->iSites();
       std::vector<int> gcoor;
@@ -63,8 +86,8 @@ namespace QCD {
       std::vector<Integer *> mergeptr(Nsimd);
       for(int o=0;o<grid->oSites();o++){
 	for(int i=0;i<grid->iSites();i++){
-	  //	  RankIndexToGlobalCoor(grid->ThisRank(),o,i,gcoor);
-	  grid->RankIndexToGlobalCoor(0,o,i,gcoor);
+	  grid->RankIndexToGlobalCoor(grid->ThisRank(),o,i,gcoor);
+	  //	  grid->RankIndexToGlobalCoor(0,o,i,gcoor);
 	  mergebuf[i]=gcoor[mu];
 	  mergeptr[i]=&mergebuf[i];
 	}
diff --git a/Grid_main.cc b/Grid_main.cc
index 291cde25..0d6c5039 100644
--- a/Grid_main.cc
+++ b/Grid_main.cc
@@ -84,6 +84,8 @@ int main (int argc, char ** argv)
     LatticeSpinColourMatrix scMat(&Fine);
     
     LatticeComplex scalar(&Fine);
+    LatticeReal    rscalar(&Fine);
+    LatticeReal    iscalar(&Fine);
 
     SpinMatrix GammaFive;
     iSpinMatrix<vComplex> iGammaFive;
@@ -110,6 +112,44 @@ int main (int argc, char ** argv)
     
     cMat = outerProduct(cVec,cVec);
     scalar = localInnerProduct(cVec,cVec);
+
+    scalar += scalar;
+    scalar -= scalar;
+    scalar *= scalar;
+    add(scalar,scalar,scalar);
+    sub(scalar,scalar,scalar);
+    mult(scalar,scalar,scalar);
+    mac(scalar,scalar,scalar);
+    scalar = scalar+scalar;
+    scalar = scalar-scalar;
+    scalar = scalar*scalar;
+
+    scalar=outerProduct(scalar,scalar);
+
+    scalar=adj(scalar);
+
+    //    rscalar=real(scalar);
+    //    iscalar=imag(scalar);
+    //    scalar =cmplx(rscalar,iscalar);
+
+    scalar=transpose(scalar);
+    scalar=transposeIndex<1>(scalar);
+    scalar=traceIndex<1>(scalar);
+    scalar=peekIndex<1>(cVec,0);
+    scalar=trace(scalar);
+    scalar=localInnerProduct(cVec,cVec);
+    scalar=localNorm2(cVec);
+
+//     -=,+=,*=,()
+//     add,+,sub,-,mult,mac,*
+//     adj,conj
+//     real,imag
+//     transpose,transposeIndex  
+//     trace,traceIndex
+//     peekIndex
+//     innerProduct,outerProduct,
+//     localNorm2
+//     localInnerProduct
     
     scMat = sMat*scMat;  // LatticeSpinColourMatrix = LatticeSpinMatrix       * LatticeSpinColourMatrix
 
@@ -171,15 +211,21 @@ int main (int argc, char ** argv)
       SpinMatrix   s_m;   
       SpinColourMatrix sc_m; 
 
-      s_m = traceIndex<1>(sc_m);
-      c_m = traceIndex<2>(sc_m);
+      s_m = traceIndex<1>(sc_m); // Map to traceColour
+      c_m = traceIndex<2>(sc_m); // map to traceSpin
 
-      c   = traceIndex<2>(s_m);
+      c   = traceIndex<2>(s_m); 
       c   = traceIndex<1>(c_m);
+      
+      s_m = peekIndex<1>(scm,0,0);
+      c_m = peekIndex<2>(scm,1,2);
 
       printf("c. Level %d\n",c_m.TensorLevel);
-      printf("c. Level %d\n",c_m._internal.TensorLevel);
-
+      printf("c. Level %d\n",c_m().TensorLevel);
+      printf("c. Level %d\n",c_m()().TensorLevel);
+      
+      c_m()() = scm()(0,0); //ColourComponents of CM <= ColourComponents of SpinColourMatrix
+      scm()(1,1) = cm()();  //ColourComponents of CM <= ColourComponents of SpinColourMatrix
     }
     
     FooBar = Bar;
@@ -244,6 +290,9 @@ int main (int argc, char ** argv)
     int ncall=100;
     int Nc = Grid::QCD::Nc;
 
+    LatticeGaugeField U(&Fine);
+    //    LatticeColourMatrix Uy = U(yDir);
+
     flops = ncall*1.0*volume*(8*Nc*Nc*Nc);
     bytes = ncall*1.0*volume*Nc*Nc    *2*3*sizeof(Grid::Real);
     if ( Fine.IsBoss() ) {
@@ -373,38 +422,38 @@ int main (int argc, char ** argv)
         mdiff = shifted1-shifted2;
         amdiff=adj(mdiff);
         ColourMatrix prod = amdiff*mdiff;
-        TReal Ttr=real(trace(prod));
-        double nn=Ttr._internal._internal;
+        Real Ttr=real(trace(prod));
+        double nn=Ttr;
         if ( nn > 0 )
             cout<<"Shift real trace fail "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<endl;
      
 
         for(int r=0;r<3;r++){
         for(int c=0;c<3;c++){
-            diff =shifted1._internal._internal[r][c]-shifted2._internal._internal[r][c];
+            diff =shifted1()()(r,c)-shifted2()()(r,c);
             nn=real(conj(diff)*diff);
             if ( nn > 0 )
                 cout<<"Shift fail (shifted1/shifted2-ref) "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "
-                    <<shifted1._internal._internal[r][c]<<" "<<shifted2._internal._internal[r][c]
-                    << " "<< foo._internal._internal[r][c]<< " lex expect " << lex_coor << " lex "<<lex<<endl;
+                    <<shifted1()()(r,c)<<" "<<shifted2()()(r,c)
+                    << " "<< foo()()(r,c)<< " lex expect " << lex_coor << " lex "<<lex<<endl;
             else if(0)
                 cout<<"Shift pass 1vs2 "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "
-                    <<shifted1._internal._internal[r][c]<<" "<<shifted2._internal._internal[r][c]
-                    << " "<< foo._internal._internal[r][c]<< " lex expect " << lex_coor << " lex "<<lex<<endl;
+                    <<shifted1()()(r,c)<<" "<<shifted2()()(r,c)
+                    << " "<< foo()()(r,c)<< " lex expect " << lex_coor << " lex "<<lex<<endl;
         }}
         
         for(int r=0;r<3;r++){
         for(int c=0;c<3;c++){
-            diff =shifted3._internal._internal[r][c]-shifted2._internal._internal[r][c];
+            diff =shifted3()()(r,c)-shifted2()()(r,c);
             nn=real(conj(diff)*diff);
             if ( nn > 0 )
                 cout<<"Shift rb fail (shifted3/shifted2-ref) "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "
-                <<shifted3._internal._internal[r][c]<<" "<<shifted2._internal._internal[r][c]
-                << " "<< foo._internal._internal[r][c]<< " lex expect " << lex_coor << " lex "<<lex<<endl;
+                <<shifted3()()(r,c)<<" "<<shifted2()()(r,c)
+                << " "<< foo()()(r,c)<< " lex expect " << lex_coor << " lex "<<lex<<endl;
             else if(0)
                 cout<<"Shift rb pass 3vs2 "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "
-                <<shifted3._internal._internal[r][c]<<" "<<shifted2._internal._internal[r][c]
-                << " "<< foo._internal._internal[r][c]<< " lex expect " << lex_coor << " lex "<<lex<<endl;
+                <<shifted3()()(r,c)<<" "<<shifted2()()(r,c)
+                << " "<< foo()()(r,c)<< " lex expect " << lex_coor << " lex "<<lex<<endl;
         }}
         peekSite(bar,Bar,coor);
                     
@@ -412,7 +461,7 @@ int main (int argc, char ** argv)
         foobar2 = foo*bar;
         for(int r=0;r<Nc;r++){
         for(int c=0;c<Nc;c++){
-            diff =foobar2._internal._internal[r][c]-foobar1._internal._internal[r][c];
+            diff =foobar2()()(r,c)-foobar1()()(r,c);
             nrm = nrm + real(conj(diff)*diff);
         }}
     }}}}
diff --git a/Grid_math_types.h b/Grid_math_types.h
index 9a9f1274..0153f46b 100644
--- a/Grid_math_types.h
+++ b/Grid_math_types.h
@@ -4,6 +4,10 @@
 #include <Grid_math_type_mapper.h>
 #include <type_traits>
 
+//
+// Indexing; want to be able to dereference and 
+// obtain either an lvalue or an rvalue.
+//
 namespace Grid {
 
   // First some of my own traits
@@ -58,6 +62,16 @@ namespace Grid {
     static const bool value = (Level==T::TensorLevel);
     static const bool notvalue = (Level!=T::TensorLevel);
   };
+  // What is the vtype
+  template<typename T> struct isComplex {
+    static const bool value = false;
+  };
+  template<> struct isComplex<ComplexF> {
+    static const bool value = true;
+  };
+  template<> struct isComplex<ComplexD> {
+    static const bool value = true;
+  };
 
 
 ///////////////////////////////////////////////////
@@ -65,12 +79,7 @@ namespace Grid {
 // These can be composed to form tensor products of internal indices.
 ///////////////////////////////////////////////////
 
-  // Terminates the recursion for temoval of all Grids tensors
-  inline vRealD    TensorRemove(vRealD    arg){ return arg;}
-  inline vRealF    TensorRemove(vRealF    arg){ return arg;}
-  inline vComplexF TensorRemove(vComplexF arg){ return arg;}
-  inline vComplexD TensorRemove(vComplexD arg){ return arg;}
-  inline vInteger  TensorRemove(vInteger  arg){ return arg;}
+
 
 template<class vtype> class iScalar
 {
@@ -109,10 +118,6 @@ public:
     friend void merge(iScalar<vtype> &in,std::vector<scalar_type *> &out){
       merge(in._internal,out); // extract advances the pointers in out
     }
-    friend inline iScalar<vtype>::vector_type TensorRemove(iScalar<vtype> arg)
-    {
-      return TensorRemove(arg._internal);
-    }
 
     // Unary negation
     friend inline iScalar<vtype> operator -(const iScalar<vtype> &r) {
@@ -133,7 +138,23 @@ public:
         *this = (*this)+r;
         return *this;
     }
+    
+    inline vtype & operator ()(void) {
+      return _internal;
+    }
+
+    operator ComplexD () const { return(TensorRemove(_internal)); };
+    operator RealD () const { return(real(TensorRemove(_internal))); }
+
 };
+///////////////////////////////////////////////////////////
+// Allows to turn scalar<scalar<scalar<double>>>> back to double.
+///////////////////////////////////////////////////////////
+template<class T> inline typename std::enable_if<isGridTensor<T>::notvalue, T>::type TensorRemove(T arg) { return arg;}
+template<class vtype> inline auto TensorRemove(iScalar<vtype> arg) -> decltype(TensorRemove(arg._internal))
+{
+  return TensorRemove(arg._internal);
+}
     
 template<class vtype,int N> class iVector
 {
@@ -148,7 +169,8 @@ public:
   typedef iScalar<tensor_reduced_v> tensor_reduced;
 
     iVector(Zero &z){ *this = zero; };
-    iVector() {};
+    iVector() {};// Empty constructure
+
     iVector<vtype,N> & operator= (Zero &hero){
         zeroit(*this);
         return *this;
@@ -192,6 +214,9 @@ public:
         *this = (*this)+r;
         return *this;
     }
+    inline vtype & operator ()(int i) {
+      return _internal[i];
+    }
 };
     
 template<class vtype,int N> class iMatrix
@@ -262,6 +287,9 @@ public:
     *this = (*this)+r;
     return *this;
   }
+  inline vtype & operator ()(int i,int j) {
+    return _internal[i][j];
+  }
 
 };
 
@@ -801,6 +829,33 @@ template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,
 }
 template<class l,int N> inline iMatrix<l,N> operator * (double lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
 
+////////////////////////////////////////////////////////////////////
+// Complex support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,ComplexD rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l> inline iScalar<l> operator * (ComplexD lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,ComplexD rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l,int N> inline iVector<l,N> operator * (ComplexD lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,ComplexD rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator * (ComplexD lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
+
 ////////////////////////////////////////////////////////////////////
 // Integer support; cast to "scalar_type" through constructor
 ////////////////////////////////////////////////////////////////////
@@ -1101,6 +1156,9 @@ template<class vtype,int N> inline iMatrix<vtype,N> adj(const iMatrix<vtype,N> &
     }}
     return ret;
 }
+
+
+
 /////////////////////////////////////////////////////////////////
 // Transpose all indices
 /////////////////////////////////////////////////////////////////
@@ -1133,7 +1191,7 @@ template<class vtype,int N>
     return ret;
   }
 
-template<class vtype,int N>
+template<class vtype>
   inline typename std::enable_if<isGridTensor<vtype>::value, iScalar<vtype> >::type 
   transpose(iScalar<vtype> arg)
   {
@@ -1142,7 +1200,7 @@ template<class vtype,int N>
     return ret;
   }
 
-template<class vtype,int N>
+template<class vtype>
   inline typename std::enable_if<isGridTensor<vtype>::notvalue, iScalar<vtype> >::type 
   transpose(iScalar<vtype> arg)
   {
@@ -1151,6 +1209,7 @@ template<class vtype,int N>
     return ret;
   }
 
+
 ////////////////////////////////////////////////////////////////////////////////////////////
 // Transpose a specific index; instructive to compare this style of recursion termination
 // to that of adj; which is easiers?
@@ -1182,7 +1241,15 @@ template<int Level,class vtype> inline
 typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue, iScalar<vtype> >::type 
 transposeIndex (const iScalar<vtype> &arg)
 {
-  return transposeIndex<Level>(arg._internal);
+  iScalar<vtype> ret;
+  ret._internal=transposeIndex<Level>(arg._internal);
+  return ret;
+}
+template<int Level,class vtype> inline 
+typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value, iScalar<vtype> >::type 
+transposeIndex (const iScalar<vtype> &arg)
+{
+  return arg;
 }
 
 //////////////////////////////////////////////////////////////////
@@ -1216,8 +1283,10 @@ inline auto trace(const iScalar<vtype> &arg) -> iScalar<decltype(trace(arg._inte
     ret._internal=trace(arg._internal);
     return ret;
 }
-
-// Specific indices.
+////////////////////////////////////////////////////////////////////////////////////////////////////////
+// Trace Specific indices.
+////////////////////////////////////////////////////////////////////////////////////////////////////////
+/*
 template<int Level,class vtype> inline 
 auto traceIndex(const iScalar<vtype> &arg) -> iScalar<decltype(traceIndex<Level>(arg._internal)) >
 {
@@ -1225,6 +1294,25 @@ auto traceIndex(const iScalar<vtype> &arg) -> iScalar<decltype(traceIndex<Level>
   ret._internal = traceIndex<Level>(arg._internal);
   return ret;
 }
+*/
+template<int Level,class vtype> inline auto 
+traceIndex (const iScalar<vtype> &arg) ->
+typename 
+std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue, 
+  iScalar<decltype(traceIndex<Level>(arg._internal))> >::type 
+
+{
+  iScalar<decltype(traceIndex<Level>(arg._internal))> ret;
+  ret._internal=traceIndex<Level>(arg._internal);
+  return ret;
+}
+template<int Level,class vtype> inline auto
+traceIndex (const iScalar<vtype> &arg) ->
+typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value, 
+                        iScalar<vtype> >::type 
+{
+  return arg;
+}
 
 // If we hit the right index, return scalar and trace it with no further recursion
 template<int Level,class vtype,int N> inline 
@@ -1254,6 +1342,149 @@ auto traceIndex(const iMatrix<vtype,N> &arg) ->
   return ret;
 }
 
+//////////////////////////////////////////////////////////////////////////////
+// Peek on a specific index; returns a scalar in that index, tensor inherits rest
+//////////////////////////////////////////////////////////////////////////////
+// If we hit the right index, return scalar with no further recursion
+
+//template<int Level> inline ComplexF peekIndex(const ComplexF arg) { return arg;}
+//template<int Level> inline ComplexD peekIndex(const ComplexD arg) { return arg;}
+//template<int Level> inline RealF peekIndex(const RealF arg) { return arg;}
+//template<int Level> inline RealD peekIndex(const RealD arg) { return arg;}
+
+// Scalar peek, no indices
+template<int Level,class vtype> inline 
+  auto peekIndex(const iScalar<vtype> &arg) -> 
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value,  // Index matches
+  iScalar<vtype> >::type                              // return scalar
+{
+  return arg;
+}
+// Vector peek, one index
+template<int Level,class vtype,int N> inline 
+  auto peekIndex(const iVector<vtype,N> &arg,int i) -> 
+  typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::value,  // Index matches
+  iScalar<vtype> >::type                              // return scalar
+{
+  iScalar<vtype> ret;                              // return scalar
+  ret._internal = arg._internal[i];
+  return ret;
+}
+// Matrix peek, two indices
+template<int Level,class vtype,int N> inline 
+  auto peekIndex(const iMatrix<vtype,N> &arg,int i,int j) -> 
+  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value,  // Index matches
+  iScalar<vtype> >::type                              // return scalar
+{
+  iScalar<vtype> ret;                              // return scalar
+  ret._internal = arg._internal[i][j];
+  return ret;
+}
+
+/////////////
+// No match peek for scalar,vector,matrix must forward on either 0,1,2 args. Must have 9 routines with notvalue
+/////////////
+// scalar
+template<int Level,class vtype> inline 
+  auto peekIndex(const iScalar<vtype> &arg) ->                                     // Scalar 0 index  
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does NOT match
+  iScalar<decltype(peekIndex<Level>(arg._internal))> >::type                       
+{
+  iScalar<decltype(peekIndex<Level>(arg._internal))> ret;
+  ret._internal= peekIndex<Level>(arg._internal);
+  return ret;
+}
+template<int Level,class vtype> inline 
+  auto peekIndex(const iScalar<vtype> &arg,int i) ->                             // Scalar 1 index
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does NOT match
+  iScalar<decltype(peekIndex<Level>(arg._internal,i))> >::type                       
+{
+  iScalar<decltype(peekIndex<Level>(arg._internal,i))> ret;
+  ret._internal=peekIndex<Level>(arg._internal,i);
+  return ret;
+}
+template<int Level,class vtype> inline 
+  auto peekIndex(const iScalar<vtype> &arg,int i,int j) ->                         // Scalar, 2 index
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does NOT match
+  iScalar<decltype(peekIndex<Level>(arg._internal,i,j))> >::type                       
+{
+  iScalar<decltype(peekIndex<Level>(arg._internal,i,j))> ret;
+  ret._internal=peekIndex<Level>(arg._internal,i,j);
+  return ret;
+}
+// vector
+template<int Level,class vtype,int N> inline 
+auto peekIndex(const iVector<vtype,N> &arg) -> 
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does not match
+  iVector<decltype(peekIndex<Level>(arg._internal[0])),N> >::type                       
+{
+  iVector<decltype(peekIndex<Level>(arg._internal[0])),N> ret;
+  for(int ii=0;ii<N;ii++){
+    ret._internal[ii]=peekIndex<Level>(arg._internal[ii]);
+  }
+  return ret;
+}
+template<int Level,class vtype,int N> inline 
+  auto peekIndex(const iVector<vtype,N> &arg,int i) -> 
+  typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,  // Index does not match
+  iVector<decltype(peekIndex<Level>(arg._internal[0],i)),N> >::type                       
+{
+  iVector<decltype(peekIndex<Level>(arg._internal[0],i)),N> ret;
+  for(int ii=0;ii<N;ii++){
+    ret._internal[ii]=peekIndex<Level>(arg._internal[ii],i);
+  }
+  return ret;
+}
+template<int Level,class vtype,int N> inline 
+  auto peekIndex(const iVector<vtype,N> &arg,int i,int j) -> 
+  typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,  // Index does not match
+  iVector<decltype(peekIndex<Level>(arg._internal[0],i,j)),N> >::type                       
+{
+  iVector<decltype(peekIndex<Level>(arg._internal[0],i,j)),N> ret;
+  for(int ii=0;ii<N;ii++){
+    ret._internal[ii]=peekIndex<Level>(arg._internal[ii],i,j);
+  }
+  return ret;
+}
+// matrix
+template<int Level,class vtype,int N> inline 
+auto peekIndex(const iMatrix<vtype,N> &arg) -> 
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does not match
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0])),N> >::type                       
+{
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0])),N> ret;
+  for(int ii=0;ii<N;ii++){
+  for(int jj=0;jj<N;jj++){
+    ret._internal[ii][jj]=peekIndex<Level>(arg._internal[ii][jj]);// Could avoid this because peeking a scalar is dumb
+  }}
+  return ret;
+}
+template<int Level,class vtype,int N> inline 
+  auto peekIndex(const iMatrix<vtype,N> &arg,int i) -> 
+  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,  // Index does not match
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0],i)),N> >::type                       
+{
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0],i)),N> ret;
+  for(int ii=0;ii<N;ii++){
+  for(int jj=0;jj<N;jj++){
+    ret._internal[ii][jj]=peekIndex<Level>(arg._internal[ii][jj],i);
+  }}
+  return ret;
+}
+template<int Level,class vtype,int N> inline 
+  auto peekIndex(const iMatrix<vtype,N> &arg,int i,int j) -> 
+  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,  // Index does not match
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0],i,j)),N> >::type                       
+{
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0],i,j)),N> ret;
+  for(int ii=0;ii<N;ii++){
+  for(int jj=0;jj<N;jj++){
+    ret._internal[ii][jj]=peekIndex<Level>(arg._internal[ii][jj],i,j);
+  }}
+  return ret;
+}
+
+
 /////////////////////////////////////////////////////////////////
 // Can only take the real/imag part of scalar objects, since
 // lattice objects of different complex nature are non-conformable.
diff --git a/Grid_simd.h b/Grid_simd.h
index ae05c3b2..dbde0fd7 100644
--- a/Grid_simd.h
+++ b/Grid_simd.h
@@ -235,6 +235,7 @@ inline void Gpermute(vsimd &y,const vsimd &b,int perm){
 #include <Grid_vComplexF.h>
 #include <Grid_vComplexD.h>
 
+
 namespace Grid {
 
   // NB: Template the following on "type Complex" and then implement *,+,- for 
diff --git a/Grid_vRealD.h b/Grid_vRealD.h
index 0f88f7da..0eb6d9c1 100644
--- a/Grid_vRealD.h
+++ b/Grid_vRealD.h
@@ -240,7 +240,7 @@ namespace Grid {
         static int Nsimd(void) { return sizeof(dvec)/sizeof(double);}
     };
 
-   inline vRealD innerProduct(const vRealD & l, const vRealD & r) { return conj(l)*r; }
+    inline vRealD innerProduct(const vRealD & l, const vRealD & r) { return conj(l)*r; }
     inline void zeroit(vRealD &z){ vzero(z);}
 
     inline vRealD outerProduct(const vRealD &l, const vRealD& r)
@@ -250,6 +250,9 @@ namespace Grid {
     inline vRealD trace(const vRealD &arg){
         return arg;
     }
+    inline vRealD real(const vRealD &arg){
+        return arg;
+    }
 
 
 }
diff --git a/Grid_vRealF.h b/Grid_vRealF.h
index 78c37177..7831a6c7 100644
--- a/Grid_vRealF.h
+++ b/Grid_vRealF.h
@@ -267,10 +267,12 @@ friend inline void vstore(const vRealF &ret, float *a){
     {
         return l*r;
     }
-
     inline vRealF trace(const vRealF &arg){
         return arg;
     }
+    inline vRealF real(const vRealF &arg){
+        return arg;
+    }
 
     
 }
diff --git a/TODO b/TODO
index 5f21f520..33032277 100644
--- a/TODO
+++ b/TODO
@@ -1,36 +1,60 @@
-
-AUDITS:
-* FIXME audit
-* const audit
-* Replace vset with a call to merge.; 
-* care in Gmerge,Gextract over vset .
-* extract / merge extra implementation removal      
-
-
-BUILD:
-* Test infrastructure
-* subdirs lib, tests ??
-
 FUNCTIONALITY:
 * Conditional execution, where etc...                -----DONE, simple test
 * Integer relational support                         -----DONE
 * Coordinate information, integers etc...            -----DONE
 * Integer type padding/union to vector.              -----DONE 
 * LatticeCoordinate[mu]                              -----DONE
+
+AUDITS:
+// Lattice support audit                 Tested in Grid_main.cc
+//
+//     -=,+=,*=                           Y
+//     add,+,sub,-,mult,mac,*             Y
+//     innerProduct,norm2                 Y
+//     localInnerProduct,outerProduct,    Y
+//     adj,conj                           Y
+//     transpose,                         Y
+//     trace                              Y
+//
+//     transposeIndex                     Y
+//     traceIndex                         Y
+//     peekIndex                          N; #args
+//
+//     real,imag                          missing, semantic thought needed on real/im support.
+//                                        perhaps I just keep everything complex?
+// 
+
+* FIXME audit
+* const audit
+* Replace vset with a call to merge.; 
+* care in Gmerge,Gextract over vset .
+* extract / merge extra implementation removal      
+
+BUILD:
+* Test infrastructure
+* subdirs lib, tests ??
+
+* How to do U[mu] ... lorentz part of type structure or not. more like chroma if not.
+* Passing a Grid into construct iVector<LatticeColourMatrix,4>??
+*
+
 * Stencil operator support                           -----Initial thoughts, trial implementation DONE.
                                                      -----some simple tests that Stencil matches Cshift.
                                                      -----do all permute in comms phase, so that copy permute
 						     -----cases move into a buffer.
+						     -----allow transform in/out buffers spproj
 
 * CovariantShift support                             -----Use a class to store gauge field? (parallel transport?)
 
+POST-SFW call April 16:
 
-POST-SFW call:
+* TraceColor, TraceSpin.                             ----- DONE (traceIndex<1>,traceIndex<2>, transposeIndex<1>,transposeIndex<2>)
+                                                     ----- Implement mapping between traceColour and traceSpin and traceIndex<1/2>.
+
+* expose traceIndex, peekIndex, transposeIndex etc at the Lattice Level -- DONE
 
-* TraceColor, TraceSpin.
 * How to define simple matrix operations, such as flavour matrices?
 
-
 * Subset support, slice sums etc...                  -----Only need slice sum?
                                                      -----Generic cartesian subslicing?
                                                      -----Array ranges / boost extents?
@@ -38,10 +62,10 @@ POST-SFW call:
                                                      -----Suggests generalised cartesian subblocking
                                                           sums, returning modified grid?
 					             -----What should interface be?
+
 i)  Two classes of subset;   red black parity subsetting (pick checkerboard).
                              cartesian sub-block subsetting
 
-
 ii) Need to be able to project one Grid to another Grid.
 
 Interface: (?)
diff --git a/test_Grid_stencil.cc b/test_Grid_stencil.cc
index fd2a9503..3ed32f3a 100644
--- a/test_Grid_stencil.cc
+++ b/test_Grid_stencil.cc
@@ -107,7 +107,7 @@ int main (int argc, char ** argv)
 
 	Real nrmC = norm2(Check);
 	Real nrmB = norm2(Bar);
-	Real nrm = norm2(Check-Bar);
+	Real nrm  = norm2(Check-Bar);
 	printf("N2diff = %le (%le, %le) \n",nrm,nrmC,nrmB);fflush(stdout);
 
 	Real snrmC =0;
@@ -127,18 +127,18 @@ int main (int argc, char ** argv)
 
 	  for(int r=0;r<3;r++){
 	  for(int c=0;c<3;c++){
-            diff =check._internal._internal[r][c]-bar._internal._internal[r][c];
+            diff =check()()(r,c)-bar()()(r,c);
             double nn=real(conj(diff)*diff);
             if ( nn > 0){
 	      printf("Coor (%d %d %d %d) \t rc %d%d \t %le %le %le\n",
 		     coor[0],coor[1],coor[2],coor[3],r,c,
 		     nn,
-		     real(check._internal._internal[r][c]),
-		     real(bar._internal._internal[r][c])
+		     real(check()()(r,c)),
+		     real(bar()()(r,c))
 		     );
 	    }
-	    snrmC=snrmC+real(conj(check._internal._internal[r][c])*check._internal._internal[r][c]);
-	    snrmB=snrmB+real(conj(bar._internal._internal[r][c])*bar._internal._internal[r][c]);
+	    snrmC=snrmC+real(conj(check()()(r,c))*check()()(r,c));
+	    snrmB=snrmB+real(conj(bar()()(r,c))*bar()()(r,c));
 	    snrm=snrm+nn;
 	  }}
 	 

From 3e3df092bbdcbea363735c8175b978504857c6e1 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 14:55:00 +0100
Subject: [PATCH 061/429] Reorg of build structure

---
 Grid.h => lib/Grid.h                          |  0
 Grid_Cartesian.h => lib/Grid_Cartesian.h      |  0
 .../Grid_Communicator.h                       |  0
 Grid_Lattice.h => lib/Grid_Lattice.h          |  0
 Grid_QCD.h => lib/Grid_QCD.h                  |  0
 .../Grid_aligned_allocator.h                  |  0
 .../Grid_communicator_fake.cc                 |  0
 .../Grid_communicator_mpi.cc                  |  0
 Grid_comparison.h => lib/Grid_comparison.h    |  0
 Grid_config.h => lib/Grid_config.h            |  0
 Grid_config.h.in => lib/Grid_config.h.in      |  0
 Grid_cshift.h => lib/Grid_cshift.h            |  0
 .../Grid_cshift_common.h                      |  0
 Grid_cshift_mpi.h => lib/Grid_cshift_mpi.h    |  0
 Grid_cshift_none.h => lib/Grid_cshift_none.h  |  0
 Grid_init.cc => lib/Grid_init.cc              |  0
 .../Grid_math_type_mapper.h                   |  0
 Grid_math_types.h => lib/Grid_math_types.h    |  0
 Grid_predicated.h => lib/Grid_predicated.h    |  0
 Grid_simd.h => lib/Grid_simd.h                |  0
 Grid_stencil.h => lib/Grid_stencil.h          |  0
 .../Grid_stencil_common.cc                    |  0
 Grid_summation.h => lib/Grid_summation.h      |  0
 Grid_vComplexD.h => lib/Grid_vComplexD.h      |  0
 Grid_vComplexF.h => lib/Grid_vComplexF.h      |  0
 Grid_vInteger.h => lib/Grid_vInteger.h        |  0
 Grid_vRealD.h => lib/Grid_vRealD.h            |  0
 Grid_vRealF.h => lib/Grid_vRealF.h            |  0
 lib/Makefile.am                               | 44 +++++++++++++++++++
 Grid_main.cc => tests/Grid_main.cc            |  0
 tests/Makefile.am                             | 15 +++++++
 .../test_Grid_jacobi.cc                       |  0
 .../test_Grid_stencil.cc                      |  0
 33 files changed, 59 insertions(+)
 rename Grid.h => lib/Grid.h (100%)
 rename Grid_Cartesian.h => lib/Grid_Cartesian.h (100%)
 rename Grid_Communicator.h => lib/Grid_Communicator.h (100%)
 rename Grid_Lattice.h => lib/Grid_Lattice.h (100%)
 rename Grid_QCD.h => lib/Grid_QCD.h (100%)
 rename Grid_aligned_allocator.h => lib/Grid_aligned_allocator.h (100%)
 rename Grid_communicator_fake.cc => lib/Grid_communicator_fake.cc (100%)
 rename Grid_communicator_mpi.cc => lib/Grid_communicator_mpi.cc (100%)
 rename Grid_comparison.h => lib/Grid_comparison.h (100%)
 rename Grid_config.h => lib/Grid_config.h (100%)
 rename Grid_config.h.in => lib/Grid_config.h.in (100%)
 rename Grid_cshift.h => lib/Grid_cshift.h (100%)
 rename Grid_cshift_common.h => lib/Grid_cshift_common.h (100%)
 rename Grid_cshift_mpi.h => lib/Grid_cshift_mpi.h (100%)
 rename Grid_cshift_none.h => lib/Grid_cshift_none.h (100%)
 rename Grid_init.cc => lib/Grid_init.cc (100%)
 rename Grid_math_type_mapper.h => lib/Grid_math_type_mapper.h (100%)
 rename Grid_math_types.h => lib/Grid_math_types.h (100%)
 rename Grid_predicated.h => lib/Grid_predicated.h (100%)
 rename Grid_simd.h => lib/Grid_simd.h (100%)
 rename Grid_stencil.h => lib/Grid_stencil.h (100%)
 rename Grid_stencil_common.cc => lib/Grid_stencil_common.cc (100%)
 rename Grid_summation.h => lib/Grid_summation.h (100%)
 rename Grid_vComplexD.h => lib/Grid_vComplexD.h (100%)
 rename Grid_vComplexF.h => lib/Grid_vComplexF.h (100%)
 rename Grid_vInteger.h => lib/Grid_vInteger.h (100%)
 rename Grid_vRealD.h => lib/Grid_vRealD.h (100%)
 rename Grid_vRealF.h => lib/Grid_vRealF.h (100%)
 create mode 100644 lib/Makefile.am
 rename Grid_main.cc => tests/Grid_main.cc (100%)
 create mode 100644 tests/Makefile.am
 rename test_Grid_jacobi.cc => tests/test_Grid_jacobi.cc (100%)
 rename test_Grid_stencil.cc => tests/test_Grid_stencil.cc (100%)

diff --git a/Grid.h b/lib/Grid.h
similarity index 100%
rename from Grid.h
rename to lib/Grid.h
diff --git a/Grid_Cartesian.h b/lib/Grid_Cartesian.h
similarity index 100%
rename from Grid_Cartesian.h
rename to lib/Grid_Cartesian.h
diff --git a/Grid_Communicator.h b/lib/Grid_Communicator.h
similarity index 100%
rename from Grid_Communicator.h
rename to lib/Grid_Communicator.h
diff --git a/Grid_Lattice.h b/lib/Grid_Lattice.h
similarity index 100%
rename from Grid_Lattice.h
rename to lib/Grid_Lattice.h
diff --git a/Grid_QCD.h b/lib/Grid_QCD.h
similarity index 100%
rename from Grid_QCD.h
rename to lib/Grid_QCD.h
diff --git a/Grid_aligned_allocator.h b/lib/Grid_aligned_allocator.h
similarity index 100%
rename from Grid_aligned_allocator.h
rename to lib/Grid_aligned_allocator.h
diff --git a/Grid_communicator_fake.cc b/lib/Grid_communicator_fake.cc
similarity index 100%
rename from Grid_communicator_fake.cc
rename to lib/Grid_communicator_fake.cc
diff --git a/Grid_communicator_mpi.cc b/lib/Grid_communicator_mpi.cc
similarity index 100%
rename from Grid_communicator_mpi.cc
rename to lib/Grid_communicator_mpi.cc
diff --git a/Grid_comparison.h b/lib/Grid_comparison.h
similarity index 100%
rename from Grid_comparison.h
rename to lib/Grid_comparison.h
diff --git a/Grid_config.h b/lib/Grid_config.h
similarity index 100%
rename from Grid_config.h
rename to lib/Grid_config.h
diff --git a/Grid_config.h.in b/lib/Grid_config.h.in
similarity index 100%
rename from Grid_config.h.in
rename to lib/Grid_config.h.in
diff --git a/Grid_cshift.h b/lib/Grid_cshift.h
similarity index 100%
rename from Grid_cshift.h
rename to lib/Grid_cshift.h
diff --git a/Grid_cshift_common.h b/lib/Grid_cshift_common.h
similarity index 100%
rename from Grid_cshift_common.h
rename to lib/Grid_cshift_common.h
diff --git a/Grid_cshift_mpi.h b/lib/Grid_cshift_mpi.h
similarity index 100%
rename from Grid_cshift_mpi.h
rename to lib/Grid_cshift_mpi.h
diff --git a/Grid_cshift_none.h b/lib/Grid_cshift_none.h
similarity index 100%
rename from Grid_cshift_none.h
rename to lib/Grid_cshift_none.h
diff --git a/Grid_init.cc b/lib/Grid_init.cc
similarity index 100%
rename from Grid_init.cc
rename to lib/Grid_init.cc
diff --git a/Grid_math_type_mapper.h b/lib/Grid_math_type_mapper.h
similarity index 100%
rename from Grid_math_type_mapper.h
rename to lib/Grid_math_type_mapper.h
diff --git a/Grid_math_types.h b/lib/Grid_math_types.h
similarity index 100%
rename from Grid_math_types.h
rename to lib/Grid_math_types.h
diff --git a/Grid_predicated.h b/lib/Grid_predicated.h
similarity index 100%
rename from Grid_predicated.h
rename to lib/Grid_predicated.h
diff --git a/Grid_simd.h b/lib/Grid_simd.h
similarity index 100%
rename from Grid_simd.h
rename to lib/Grid_simd.h
diff --git a/Grid_stencil.h b/lib/Grid_stencil.h
similarity index 100%
rename from Grid_stencil.h
rename to lib/Grid_stencil.h
diff --git a/Grid_stencil_common.cc b/lib/Grid_stencil_common.cc
similarity index 100%
rename from Grid_stencil_common.cc
rename to lib/Grid_stencil_common.cc
diff --git a/Grid_summation.h b/lib/Grid_summation.h
similarity index 100%
rename from Grid_summation.h
rename to lib/Grid_summation.h
diff --git a/Grid_vComplexD.h b/lib/Grid_vComplexD.h
similarity index 100%
rename from Grid_vComplexD.h
rename to lib/Grid_vComplexD.h
diff --git a/Grid_vComplexF.h b/lib/Grid_vComplexF.h
similarity index 100%
rename from Grid_vComplexF.h
rename to lib/Grid_vComplexF.h
diff --git a/Grid_vInteger.h b/lib/Grid_vInteger.h
similarity index 100%
rename from Grid_vInteger.h
rename to lib/Grid_vInteger.h
diff --git a/Grid_vRealD.h b/lib/Grid_vRealD.h
similarity index 100%
rename from Grid_vRealD.h
rename to lib/Grid_vRealD.h
diff --git a/Grid_vRealF.h b/lib/Grid_vRealF.h
similarity index 100%
rename from Grid_vRealF.h
rename to lib/Grid_vRealF.h
diff --git a/lib/Makefile.am b/lib/Makefile.am
new file mode 100644
index 00000000..8cb9a0be
--- /dev/null
+++ b/lib/Makefile.am
@@ -0,0 +1,44 @@
+# additional include paths necessary to compile the C++ library
+AM_CXXFLAGS = -I$(top_srcdir)/
+
+
+extra_sources=
+if BUILD_COMMS_MPI
+  extra_sources+=Grid_communicator_mpi.cc
+  extra_sources+=Grid_stencil_common.cc
+endif
+if BUILD_COMMS_NONE
+  extra_sources+=Grid_communicator_fake.cc
+  extra_sources+=Grid_stencil_common.cc
+endif
+
+#
+# Libraries
+#
+lib_LIBRARIES = libGrid.a
+libGrid_a_SOURCES = Grid_init.cc $(extra_sources)
+
+#
+# Include files
+#
+include_HEADERS = Grid_config.h\
+	Grid.h\
+	Grid_simd.h\
+	Grid_vComplexD.h\
+	Grid_vComplexF.h\
+	Grid_vRealD.h\
+	Grid_vRealF.h\
+	Grid_Cartesian.h\
+	Grid_Lattice.h\
+	Grid_Communicator.h\
+	Grid_QCD.h\
+	Grid_aligned_allocator.h\
+	Grid_cshift.h\
+	Grid_cshift_common.h\
+	Grid_cshift_mpi.h\
+	Grid_cshift_none.h\
+	Grid_stencil.h\
+	Grid_math_types.h
+
+
+
diff --git a/Grid_main.cc b/tests/Grid_main.cc
similarity index 100%
rename from Grid_main.cc
rename to tests/Grid_main.cc
diff --git a/tests/Makefile.am b/tests/Makefile.am
new file mode 100644
index 00000000..2bf8cd1a
--- /dev/null
+++ b/tests/Makefile.am
@@ -0,0 +1,15 @@
+# additional include paths necessary to compile the C++ library
+AM_CXXFLAGS = -I$(top_srcdir)/lib
+AM_LDFLAGS = -L$(top_srcdir)/lib
+
+#
+# Test code
+#
+bin_PROGRAMS = Grid_main test_Grid_stencil
+
+Grid_main_SOURCES = Grid_main.cc
+Grid_main_LDADD = -lGrid
+
+test_Grid_stencil_SOURCES = test_Grid_stencil.cc
+test_Grid_stencil_LDADD = -lGrid
+
diff --git a/test_Grid_jacobi.cc b/tests/test_Grid_jacobi.cc
similarity index 100%
rename from test_Grid_jacobi.cc
rename to tests/test_Grid_jacobi.cc
diff --git a/test_Grid_stencil.cc b/tests/test_Grid_stencil.cc
similarity index 100%
rename from test_Grid_stencil.cc
rename to tests/test_Grid_stencil.cc

From 388b735fd036a1990fd66e1bc5831811c727bc07 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 14:56:05 +0100
Subject: [PATCH 062/429] Build reorg

---
 Makefile.am          |  52 +-------
 Makefile.in          | 311 +++++--------------------------------------
 configure            |  12 +-
 configure.ac         |   6 +-
 lib/Grid_config.h    |   4 +-
 lib/Grid_config.h.in |   2 +-
 6 files changed, 51 insertions(+), 336 deletions(-)

diff --git a/Makefile.am b/Makefile.am
index 1e70ac1b..a55b2b92 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -1,54 +1,4 @@
 # additional include paths necessary to compile the C++ library
 AM_CXXFLAGS = -I$(top_srcdir)/
-
-
-extra_sources=
-if BUILD_COMMS_MPI
-  extra_sources+=Grid_communicator_mpi.cc
-  extra_sources+=Grid_stencil_common.cc
-endif
-if BUILD_COMMS_NONE
-  extra_sources+=Grid_communicator_fake.cc
-  extra_sources+=Grid_stencil_common.cc
-endif
-
-#
-# Libraries
-#
-lib_LIBRARIES = libGrid.a
-libGrid_a_SOURCES = Grid_init.cc $(extra_sources)
-
-#
-# Include files
-#
-include_HEADERS = Grid_config.h\
-	Grid.h\
-	Grid_simd.h\
-	Grid_vComplexD.h\
-	Grid_vComplexF.h\
-	Grid_vRealD.h\
-	Grid_vRealF.h\
-	Grid_Cartesian.h\
-	Grid_Lattice.h\
-	Grid_Communicator.h\
-	Grid_QCD.h\
-	Grid_aligned_allocator.h\
-	Grid_cshift.h\
-	Grid_cshift_common.h\
-	Grid_cshift_mpi.h\
-	Grid_cshift_none.h\
-	Grid_stencil.h\
-	Grid_math_types.h
-
-
-#
-# Test code
-#
-bin_PROGRAMS = Grid_main test_Grid_stencil
-
-Grid_main_SOURCES = Grid_main.cc
-Grid_main_LDADD = libGrid.a
-
-test_Grid_stencil_SOURCES = test_Grid_stencil.cc
-test_Grid_stencil_LDADD = libGrid.a
+SUBDIRS = lib tests
 
diff --git a/Makefile.in b/Makefile.in
index ce8b74be..3ca9a01f 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -15,7 +15,6 @@
 @SET_MAKE@
 
 
-
 VPATH = @srcdir@
 am__is_gnu_make = { \
   if test -z '$(MAKELEVEL)'; then \
@@ -92,16 +91,13 @@ POST_UNINSTALL = :
 @BUILD_COMMS_MPI_TRUE@	Grid_stencil_common.cc
 @BUILD_COMMS_NONE_TRUE@am__append_2 = Grid_communicator_fake.cc \
 @BUILD_COMMS_NONE_TRUE@	Grid_stencil_common.cc
-bin_PROGRAMS = Grid_main$(EXEEXT) test_Grid_stencil$(EXEEXT)
-subdir = .
+subdir = lib
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/configure.ac
 am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
 	$(ACLOCAL_M4)
-DIST_COMMON = $(srcdir)/Makefile.am $(top_srcdir)/configure \
-	$(am__configure_deps) $(include_HEADERS) $(am__DIST_COMMON)
-am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \
- configure.lineno config.status.lineno
+DIST_COMMON = $(srcdir)/Makefile.am $(include_HEADERS) \
+	$(am__DIST_COMMON)
 mkinstalldirs = $(install_sh) -d
 CONFIG_HEADER = Grid_config.h
 CONFIG_CLEAN_FILES =
@@ -133,8 +129,7 @@ am__uninstall_files_from_dir = { \
     || { echo " ( cd '$$dir' && rm -f" $$files ")"; \
          $(am__cd) "$$dir" && rm -f $$files; }; \
   }
-am__installdirs = "$(DESTDIR)$(libdir)" "$(DESTDIR)$(bindir)" \
-	"$(DESTDIR)$(includedir)"
+am__installdirs = "$(DESTDIR)$(libdir)" "$(DESTDIR)$(includedir)"
 LIBRARIES = $(lib_LIBRARIES)
 AR = ar
 ARFLAGS = cru
@@ -154,13 +149,6 @@ am__libGrid_a_SOURCES_DIST = Grid_init.cc Grid_communicator_mpi.cc \
 am__objects_3 = $(am__objects_1) $(am__objects_2)
 am_libGrid_a_OBJECTS = Grid_init.$(OBJEXT) $(am__objects_3)
 libGrid_a_OBJECTS = $(am_libGrid_a_OBJECTS)
-PROGRAMS = $(bin_PROGRAMS)
-am_Grid_main_OBJECTS = Grid_main.$(OBJEXT)
-Grid_main_OBJECTS = $(am_Grid_main_OBJECTS)
-Grid_main_DEPENDENCIES = libGrid.a
-am_test_Grid_stencil_OBJECTS = test_Grid_stencil.$(OBJEXT)
-test_Grid_stencil_OBJECTS = $(am_test_Grid_stencil_OBJECTS)
-test_Grid_stencil_DEPENDENCIES = libGrid.a
 AM_V_P = $(am__v_P_@AM_V@)
 am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
 am__v_P_0 = false
@@ -190,10 +178,8 @@ AM_V_CXXLD = $(am__v_CXXLD_@AM_V@)
 am__v_CXXLD_ = $(am__v_CXXLD_@AM_DEFAULT_V@)
 am__v_CXXLD_0 = @echo "  CXXLD   " $@;
 am__v_CXXLD_1 = 
-SOURCES = $(libGrid_a_SOURCES) $(Grid_main_SOURCES) \
-	$(test_Grid_stencil_SOURCES)
-DIST_SOURCES = $(am__libGrid_a_SOURCES_DIST) $(Grid_main_SOURCES) \
-	$(test_Grid_stencil_SOURCES)
+SOURCES = $(libGrid_a_SOURCES)
+DIST_SOURCES = $(am__libGrid_a_SOURCES_DIST)
 am__can_run_installinfo = \
   case $$AM_UPDATE_INFO_DIR in \
     n|no|NO) false;; \
@@ -220,28 +206,9 @@ am__define_uniq_tagged_files = \
   done | $(am__uniquify_input)`
 ETAGS = etags
 CTAGS = ctags
-CSCOPE = cscope
-AM_RECURSIVE_TARGETS = cscope
 am__DIST_COMMON = $(srcdir)/Grid_config.h.in $(srcdir)/Makefile.in \
-	AUTHORS COPYING ChangeLog INSTALL NEWS README TODO compile \
-	depcomp install-sh missing
+	$(top_srcdir)/depcomp
 DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
-distdir = $(PACKAGE)-$(VERSION)
-top_distdir = $(distdir)
-am__remove_distdir = \
-  if test -d "$(distdir)"; then \
-    find "$(distdir)" -type d ! -perm -200 -exec chmod u+w {} ';' \
-      && rm -rf "$(distdir)" \
-      || { sleep 5 && rm -rf "$(distdir)"; }; \
-  else :; fi
-am__post_remove_distdir = $(am__remove_distdir)
-DIST_ARCHIVES = $(distdir).tar.gz
-GZIP_ENV = --best
-DIST_TARGETS = dist-gzip
-distuninstallcheck_listfiles = find . -type f -print
-am__distuninstallcheck_listfiles = $(distuninstallcheck_listfiles) \
-  | sed 's|^\./|$(prefix)/|' | grep -v '$(infodir)/dir$$'
-distcleancheck_listfiles = find . -type f -print
 ACLOCAL = @ACLOCAL@
 AMTAR = @AMTAR@
 AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
@@ -368,47 +335,39 @@ include_HEADERS = Grid_config.h\
 	Grid_stencil.h\
 	Grid_math_types.h
 
-Grid_main_SOURCES = Grid_main.cc
-Grid_main_LDADD = libGrid.a
-test_Grid_stencil_SOURCES = test_Grid_stencil.cc
-test_Grid_stencil_LDADD = libGrid.a
 all: Grid_config.h
 	$(MAKE) $(AM_MAKEFLAGS) all-am
 
 .SUFFIXES:
 .SUFFIXES: .cc .o .obj
-am--refresh: Makefile
-	@:
 $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	@for dep in $?; do \
 	  case '$(am__configure_deps)' in \
 	    *$$dep*) \
-	      echo ' cd $(srcdir) && $(AUTOMAKE) --gnu'; \
-	      $(am__cd) $(srcdir) && $(AUTOMAKE) --gnu \
-		&& exit 0; \
+	      ( cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh ) \
+	        && { if test -f $@; then exit 0; else break; fi; }; \
 	      exit 1;; \
 	  esac; \
 	done; \
-	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu Makefile'; \
+	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu lib/Makefile'; \
 	$(am__cd) $(top_srcdir) && \
-	  $(AUTOMAKE) --gnu Makefile
+	  $(AUTOMAKE) --gnu lib/Makefile
 Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
 	@case '$?' in \
 	  *config.status*) \
-	    echo ' $(SHELL) ./config.status'; \
-	    $(SHELL) ./config.status;; \
+	    cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh;; \
 	  *) \
-	    echo ' cd $(top_builddir) && $(SHELL) ./config.status $@ $(am__depfiles_maybe)'; \
-	    cd $(top_builddir) && $(SHELL) ./config.status $@ $(am__depfiles_maybe);; \
+	    echo ' cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__depfiles_maybe)'; \
+	    cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__depfiles_maybe);; \
 	esac;
 
 $(top_builddir)/config.status: $(top_srcdir)/configure $(CONFIG_STATUS_DEPENDENCIES)
-	$(SHELL) ./config.status --recheck
+	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
 
 $(top_srcdir)/configure:  $(am__configure_deps)
-	$(am__cd) $(srcdir) && $(AUTOCONF)
+	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
 $(ACLOCAL_M4):  $(am__aclocal_m4_deps)
-	$(am__cd) $(srcdir) && $(ACLOCAL) $(ACLOCAL_AMFLAGS)
+	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
 $(am__aclocal_m4_deps):
 
 Grid_config.h: stamp-h1
@@ -417,7 +376,7 @@ Grid_config.h: stamp-h1
 
 stamp-h1: $(srcdir)/Grid_config.h.in $(top_builddir)/config.status
 	@rm -f stamp-h1
-	cd $(top_builddir) && $(SHELL) ./config.status Grid_config.h
+	cd $(top_builddir) && $(SHELL) ./config.status lib/Grid_config.h
 $(srcdir)/Grid_config.h.in:  $(am__configure_deps) 
 	($(am__cd) $(top_srcdir) && $(AUTOHEADER))
 	rm -f stamp-h1
@@ -461,56 +420,6 @@ libGrid.a: $(libGrid_a_OBJECTS) $(libGrid_a_DEPENDENCIES) $(EXTRA_libGrid_a_DEPE
 	$(AM_V_at)-rm -f libGrid.a
 	$(AM_V_AR)$(libGrid_a_AR) libGrid.a $(libGrid_a_OBJECTS) $(libGrid_a_LIBADD)
 	$(AM_V_at)$(RANLIB) libGrid.a
-install-binPROGRAMS: $(bin_PROGRAMS)
-	@$(NORMAL_INSTALL)
-	@list='$(bin_PROGRAMS)'; test -n "$(bindir)" || list=; \
-	if test -n "$$list"; then \
-	  echo " $(MKDIR_P) '$(DESTDIR)$(bindir)'"; \
-	  $(MKDIR_P) "$(DESTDIR)$(bindir)" || exit 1; \
-	fi; \
-	for p in $$list; do echo "$$p $$p"; done | \
-	sed 's/$(EXEEXT)$$//' | \
-	while read p p1; do if test -f $$p \
-	  ; then echo "$$p"; echo "$$p"; else :; fi; \
-	done | \
-	sed -e 'p;s,.*/,,;n;h' \
-	    -e 's|.*|.|' \
-	    -e 'p;x;s,.*/,,;s/$(EXEEXT)$$//;$(transform);s/$$/$(EXEEXT)/' | \
-	sed 'N;N;N;s,\n, ,g' | \
-	$(AWK) 'BEGIN { files["."] = ""; dirs["."] = 1 } \
-	  { d=$$3; if (dirs[d] != 1) { print "d", d; dirs[d] = 1 } \
-	    if ($$2 == $$4) files[d] = files[d] " " $$1; \
-	    else { print "f", $$3 "/" $$4, $$1; } } \
-	  END { for (d in files) print "f", d, files[d] }' | \
-	while read type dir files; do \
-	    if test "$$dir" = .; then dir=; else dir=/$$dir; fi; \
-	    test -z "$$files" || { \
-	      echo " $(INSTALL_PROGRAM_ENV) $(INSTALL_PROGRAM) $$files '$(DESTDIR)$(bindir)$$dir'"; \
-	      $(INSTALL_PROGRAM_ENV) $(INSTALL_PROGRAM) $$files "$(DESTDIR)$(bindir)$$dir" || exit $$?; \
-	    } \
-	; done
-
-uninstall-binPROGRAMS:
-	@$(NORMAL_UNINSTALL)
-	@list='$(bin_PROGRAMS)'; test -n "$(bindir)" || list=; \
-	files=`for p in $$list; do echo "$$p"; done | \
-	  sed -e 'h;s,^.*/,,;s/$(EXEEXT)$$//;$(transform)' \
-	      -e 's/$$/$(EXEEXT)/' \
-	`; \
-	test -n "$$list" || exit 0; \
-	echo " ( cd '$(DESTDIR)$(bindir)' && rm -f" $$files ")"; \
-	cd "$(DESTDIR)$(bindir)" && rm -f $$files
-
-clean-binPROGRAMS:
-	-test -z "$(bin_PROGRAMS)" || rm -f $(bin_PROGRAMS)
-
-Grid_main$(EXEEXT): $(Grid_main_OBJECTS) $(Grid_main_DEPENDENCIES) $(EXTRA_Grid_main_DEPENDENCIES) 
-	@rm -f Grid_main$(EXEEXT)
-	$(AM_V_CXXLD)$(CXXLINK) $(Grid_main_OBJECTS) $(Grid_main_LDADD) $(LIBS)
-
-test_Grid_stencil$(EXEEXT): $(test_Grid_stencil_OBJECTS) $(test_Grid_stencil_DEPENDENCIES) $(EXTRA_test_Grid_stencil_DEPENDENCIES) 
-	@rm -f test_Grid_stencil$(EXEEXT)
-	$(AM_V_CXXLD)$(CXXLINK) $(test_Grid_stencil_OBJECTS) $(test_Grid_stencil_LDADD) $(LIBS)
 
 mostlyclean-compile:
 	-rm -f *.$(OBJEXT)
@@ -521,9 +430,7 @@ distclean-compile:
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_communicator_fake.Po@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_communicator_mpi.Po@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_init.Po@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_main.Po@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_stencil_common.Po@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/test_Grid_stencil.Po@am__quote@
 
 .cc.o:
 @am__fastdepCXX_TRUE@	$(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
@@ -593,12 +500,6 @@ GTAGS:
 	here=`$(am__cd) $(top_builddir) && pwd` \
 	  && $(am__cd) $(top_srcdir) \
 	  && gtags -i $(GTAGS_ARGS) "$$here"
-cscope: cscope.files
-	test ! -s cscope.files \
-	  || $(CSCOPE) -b -q $(AM_CSCOPEFLAGS) $(CSCOPEFLAGS) -i cscope.files $(CSCOPE_ARGS)
-clean-cscope:
-	-rm -f cscope.files
-cscope.files: clean-cscope cscopelist
 cscopelist: cscopelist-am
 
 cscopelist-am: $(am__tagged_files)
@@ -617,11 +518,8 @@ cscopelist-am: $(am__tagged_files)
 
 distclean-tags:
 	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
-	-rm -f cscope.out cscope.in.out cscope.po.out cscope.files
 
 distdir: $(DISTFILES)
-	$(am__remove_distdir)
-	test -d "$(distdir)" || mkdir "$(distdir)"
 	@srcdirstrip=`echo "$(srcdir)" | sed 's/[].[^$$\\*]/\\\\&/g'`; \
 	topsrcdirstrip=`echo "$(top_srcdir)" | sed 's/[].[^$$\\*]/\\\\&/g'`; \
 	list='$(DISTFILES)'; \
@@ -651,142 +549,11 @@ distdir: $(DISTFILES)
 	    || exit 1; \
 	  fi; \
 	done
-	-test -n "$(am__skip_mode_fix)" \
-	|| find "$(distdir)" -type d ! -perm -755 \
-		-exec chmod u+rwx,go+rx {} \; -o \
-	  ! -type d ! -perm -444 -links 1 -exec chmod a+r {} \; -o \
-	  ! -type d ! -perm -400 -exec chmod a+r {} \; -o \
-	  ! -type d ! -perm -444 -exec $(install_sh) -c -m a+r {} {} \; \
-	|| chmod -R a+r "$(distdir)"
-dist-gzip: distdir
-	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
-	$(am__post_remove_distdir)
-
-dist-bzip2: distdir
-	tardir=$(distdir) && $(am__tar) | BZIP2=$${BZIP2--9} bzip2 -c >$(distdir).tar.bz2
-	$(am__post_remove_distdir)
-
-dist-lzip: distdir
-	tardir=$(distdir) && $(am__tar) | lzip -c $${LZIP_OPT--9} >$(distdir).tar.lz
-	$(am__post_remove_distdir)
-
-dist-xz: distdir
-	tardir=$(distdir) && $(am__tar) | XZ_OPT=$${XZ_OPT--e} xz -c >$(distdir).tar.xz
-	$(am__post_remove_distdir)
-
-dist-tarZ: distdir
-	@echo WARNING: "Support for distribution archives compressed with" \
-		       "legacy program 'compress' is deprecated." >&2
-	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
-	tardir=$(distdir) && $(am__tar) | compress -c >$(distdir).tar.Z
-	$(am__post_remove_distdir)
-
-dist-shar: distdir
-	@echo WARNING: "Support for shar distribution archives is" \
-	               "deprecated." >&2
-	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
-	shar $(distdir) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).shar.gz
-	$(am__post_remove_distdir)
-
-dist-zip: distdir
-	-rm -f $(distdir).zip
-	zip -rq $(distdir).zip $(distdir)
-	$(am__post_remove_distdir)
-
-dist dist-all:
-	$(MAKE) $(AM_MAKEFLAGS) $(DIST_TARGETS) am__post_remove_distdir='@:'
-	$(am__post_remove_distdir)
-
-# This target untars the dist file and tries a VPATH configuration.  Then
-# it guarantees that the distribution is self-contained by making another
-# tarfile.
-distcheck: dist
-	case '$(DIST_ARCHIVES)' in \
-	*.tar.gz*) \
-	  GZIP=$(GZIP_ENV) gzip -dc $(distdir).tar.gz | $(am__untar) ;;\
-	*.tar.bz2*) \
-	  bzip2 -dc $(distdir).tar.bz2 | $(am__untar) ;;\
-	*.tar.lz*) \
-	  lzip -dc $(distdir).tar.lz | $(am__untar) ;;\
-	*.tar.xz*) \
-	  xz -dc $(distdir).tar.xz | $(am__untar) ;;\
-	*.tar.Z*) \
-	  uncompress -c $(distdir).tar.Z | $(am__untar) ;;\
-	*.shar.gz*) \
-	  GZIP=$(GZIP_ENV) gzip -dc $(distdir).shar.gz | unshar ;;\
-	*.zip*) \
-	  unzip $(distdir).zip ;;\
-	esac
-	chmod -R a-w $(distdir)
-	chmod u+w $(distdir)
-	mkdir $(distdir)/_build $(distdir)/_build/sub $(distdir)/_inst
-	chmod a-w $(distdir)
-	test -d $(distdir)/_build || exit 0; \
-	dc_install_base=`$(am__cd) $(distdir)/_inst && pwd | sed -e 's,^[^:\\/]:[\\/],/,'` \
-	  && dc_destdir="$${TMPDIR-/tmp}/am-dc-$$$$/" \
-	  && am__cwd=`pwd` \
-	  && $(am__cd) $(distdir)/_build/sub \
-	  && ../../configure \
-	    $(AM_DISTCHECK_CONFIGURE_FLAGS) \
-	    $(DISTCHECK_CONFIGURE_FLAGS) \
-	    --srcdir=../.. --prefix="$$dc_install_base" \
-	  && $(MAKE) $(AM_MAKEFLAGS) \
-	  && $(MAKE) $(AM_MAKEFLAGS) dvi \
-	  && $(MAKE) $(AM_MAKEFLAGS) check \
-	  && $(MAKE) $(AM_MAKEFLAGS) install \
-	  && $(MAKE) $(AM_MAKEFLAGS) installcheck \
-	  && $(MAKE) $(AM_MAKEFLAGS) uninstall \
-	  && $(MAKE) $(AM_MAKEFLAGS) distuninstallcheck_dir="$$dc_install_base" \
-	        distuninstallcheck \
-	  && chmod -R a-w "$$dc_install_base" \
-	  && ({ \
-	       (cd ../.. && umask 077 && mkdir "$$dc_destdir") \
-	       && $(MAKE) $(AM_MAKEFLAGS) DESTDIR="$$dc_destdir" install \
-	       && $(MAKE) $(AM_MAKEFLAGS) DESTDIR="$$dc_destdir" uninstall \
-	       && $(MAKE) $(AM_MAKEFLAGS) DESTDIR="$$dc_destdir" \
-	            distuninstallcheck_dir="$$dc_destdir" distuninstallcheck; \
-	      } || { rm -rf "$$dc_destdir"; exit 1; }) \
-	  && rm -rf "$$dc_destdir" \
-	  && $(MAKE) $(AM_MAKEFLAGS) dist \
-	  && rm -rf $(DIST_ARCHIVES) \
-	  && $(MAKE) $(AM_MAKEFLAGS) distcleancheck \
-	  && cd "$$am__cwd" \
-	  || exit 1
-	$(am__post_remove_distdir)
-	@(echo "$(distdir) archives ready for distribution: "; \
-	  list='$(DIST_ARCHIVES)'; for i in $$list; do echo $$i; done) | \
-	  sed -e 1h -e 1s/./=/g -e 1p -e 1x -e '$$p' -e '$$x'
-distuninstallcheck:
-	@test -n '$(distuninstallcheck_dir)' || { \
-	  echo 'ERROR: trying to run $@ with an empty' \
-	       '$$(distuninstallcheck_dir)' >&2; \
-	  exit 1; \
-	}; \
-	$(am__cd) '$(distuninstallcheck_dir)' || { \
-	  echo 'ERROR: cannot chdir into $(distuninstallcheck_dir)' >&2; \
-	  exit 1; \
-	}; \
-	test `$(am__distuninstallcheck_listfiles) | wc -l` -eq 0 \
-	   || { echo "ERROR: files left after uninstall:" ; \
-	        if test -n "$(DESTDIR)"; then \
-	          echo "  (check DESTDIR support)"; \
-	        fi ; \
-	        $(distuninstallcheck_listfiles) ; \
-	        exit 1; } >&2
-distcleancheck: distclean
-	@if test '$(srcdir)' = . ; then \
-	  echo "ERROR: distcleancheck can only run from a VPATH build" ; \
-	  exit 1 ; \
-	fi
-	@test `$(distcleancheck_listfiles) | wc -l` -eq 0 \
-	  || { echo "ERROR: files left in build directory after distclean:" ; \
-	       $(distcleancheck_listfiles) ; \
-	       exit 1; } >&2
 check-am: all-am
 check: check-am
-all-am: Makefile $(LIBRARIES) $(PROGRAMS) $(HEADERS) Grid_config.h
+all-am: Makefile $(LIBRARIES) $(HEADERS) Grid_config.h
 installdirs:
-	for dir in "$(DESTDIR)$(libdir)" "$(DESTDIR)$(bindir)" "$(DESTDIR)$(includedir)"; do \
+	for dir in "$(DESTDIR)$(libdir)" "$(DESTDIR)$(includedir)"; do \
 	  test -z "$$dir" || $(MKDIR_P) "$$dir"; \
 	done
 install: install-am
@@ -821,11 +588,9 @@ maintainer-clean-generic:
 	@echo "it deletes files that may require special tools to rebuild."
 clean: clean-am
 
-clean-am: clean-binPROGRAMS clean-generic clean-libLIBRARIES \
-	mostlyclean-am
+clean-am: clean-generic clean-libLIBRARIES mostlyclean-am
 
 distclean: distclean-am
-	-rm -f $(am__CONFIG_DISTCLEAN_FILES)
 	-rm -rf ./$(DEPDIR)
 	-rm -f Makefile
 distclean-am: clean-am distclean-compile distclean-generic \
@@ -849,7 +614,7 @@ install-dvi: install-dvi-am
 
 install-dvi-am:
 
-install-exec-am: install-binPROGRAMS install-libLIBRARIES
+install-exec-am: install-libLIBRARIES
 
 install-html: install-html-am
 
@@ -872,8 +637,6 @@ install-ps-am:
 installcheck-am:
 
 maintainer-clean: maintainer-clean-am
-	-rm -f $(am__CONFIG_DISTCLEAN_FILES)
-	-rm -rf $(top_srcdir)/autom4te.cache
 	-rm -rf ./$(DEPDIR)
 	-rm -f Makefile
 maintainer-clean-am: distclean-am maintainer-clean-generic
@@ -890,29 +653,23 @@ ps: ps-am
 
 ps-am:
 
-uninstall-am: uninstall-binPROGRAMS uninstall-includeHEADERS \
-	uninstall-libLIBRARIES
+uninstall-am: uninstall-includeHEADERS uninstall-libLIBRARIES
 
 .MAKE: all install-am install-strip
 
-.PHONY: CTAGS GTAGS TAGS all all-am am--refresh check check-am clean \
-	clean-binPROGRAMS clean-cscope clean-generic \
-	clean-libLIBRARIES cscope cscopelist-am ctags ctags-am dist \
-	dist-all dist-bzip2 dist-gzip dist-lzip dist-shar dist-tarZ \
-	dist-xz dist-zip distcheck distclean distclean-compile \
-	distclean-generic distclean-hdr distclean-tags distcleancheck \
-	distdir distuninstallcheck dvi dvi-am html html-am info \
-	info-am install install-am install-binPROGRAMS install-data \
-	install-data-am install-dvi install-dvi-am install-exec \
-	install-exec-am install-html install-html-am \
-	install-includeHEADERS install-info install-info-am \
-	install-libLIBRARIES install-man install-pdf install-pdf-am \
-	install-ps install-ps-am install-strip installcheck \
-	installcheck-am installdirs maintainer-clean \
+.PHONY: CTAGS GTAGS TAGS all all-am check check-am clean clean-generic \
+	clean-libLIBRARIES cscopelist-am ctags ctags-am distclean \
+	distclean-compile distclean-generic distclean-hdr \
+	distclean-tags distdir dvi dvi-am html html-am info info-am \
+	install install-am install-data install-data-am install-dvi \
+	install-dvi-am install-exec install-exec-am install-html \
+	install-html-am install-includeHEADERS install-info \
+	install-info-am install-libLIBRARIES install-man install-pdf \
+	install-pdf-am install-ps install-ps-am install-strip \
+	installcheck installcheck-am installdirs maintainer-clean \
 	maintainer-clean-generic mostlyclean mostlyclean-compile \
 	mostlyclean-generic pdf pdf-am ps ps-am tags tags-am uninstall \
-	uninstall-am uninstall-binPROGRAMS uninstall-includeHEADERS \
-	uninstall-libLIBRARIES
+	uninstall-am uninstall-includeHEADERS uninstall-libLIBRARIES
 
 .PRECIOUS: Makefile
 
diff --git a/configure b/configure
index 1adf726f..fc0b6453 100755
--- a/configure
+++ b/configure
@@ -585,7 +585,7 @@ PACKAGE_STRING='Grid 1.0'
 PACKAGE_BUGREPORT='paboyle@ph.ed.ac.uk'
 PACKAGE_URL=''
 
-ac_unique_file="Grid.h"
+ac_unique_file="lib/Grid.h"
 # Factoring default headers for most tests.
 ac_includes_default="\
 #include <stdio.h>
@@ -2920,7 +2920,7 @@ fi
 
 
 
-ac_config_headers="$ac_config_headers Grid_config.h"
+ac_config_headers="$ac_config_headers lib/Grid_config.h"
 
 
 # Checks for programs.
@@ -5085,6 +5085,10 @@ fi
 
 ac_config_files="$ac_config_files Makefile"
 
+ac_config_files="$ac_config_files lib/Makefile"
+
+ac_config_files="$ac_config_files tests/Makefile"
+
 cat >confcache <<\_ACEOF
 # This file is a shell script that caches the results of configure
 # tests run on this system so they can be shared between configure
@@ -5822,9 +5826,11 @@ cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 for ac_config_target in $ac_config_targets
 do
   case $ac_config_target in
-    "Grid_config.h") CONFIG_HEADERS="$CONFIG_HEADERS Grid_config.h" ;;
+    "lib/Grid_config.h") CONFIG_HEADERS="$CONFIG_HEADERS lib/Grid_config.h" ;;
     "depfiles") CONFIG_COMMANDS="$CONFIG_COMMANDS depfiles" ;;
     "Makefile") CONFIG_FILES="$CONFIG_FILES Makefile" ;;
+    "lib/Makefile") CONFIG_FILES="$CONFIG_FILES lib/Makefile" ;;
+    "tests/Makefile") CONFIG_FILES="$CONFIG_FILES tests/Makefile" ;;
 
   *) as_fn_error $? "invalid argument: \`$ac_config_target'" "$LINENO" 5;;
   esac
diff --git a/configure.ac b/configure.ac
index 83405ca8..07c1974e 100644
--- a/configure.ac
+++ b/configure.ac
@@ -2,8 +2,8 @@
 AC_INIT([Grid], [1.0], [paboyle@ph.ed.ac.uk])
 AM_INIT_AUTOMAKE
 AC_CONFIG_MACRO_DIR([m4])
-AC_CONFIG_SRCDIR([Grid.h])
-AC_CONFIG_HEADERS([Grid_config.h])
+AC_CONFIG_SRCDIR([lib/Grid.h])
+AC_CONFIG_HEADERS([lib/Grid_config.h])
 
 # Checks for programs.
 AC_PROG_CXX
@@ -72,4 +72,6 @@ AM_CONDITIONAL(BUILD_COMMS_NONE,[ test "X${ac_COMMS}X" == "XnoneX" ])
 
 
 AC_CONFIG_FILES(Makefile)
+AC_CONFIG_FILES(lib/Makefile)
+AC_CONFIG_FILES(tests/Makefile)
 AC_OUTPUT
diff --git a/lib/Grid_config.h b/lib/Grid_config.h
index 5faa2157..cb676b54 100644
--- a/lib/Grid_config.h
+++ b/lib/Grid_config.h
@@ -1,5 +1,5 @@
-/* Grid_config.h.  Generated from Grid_config.h.in by configure.  */
-/* Grid_config.h.in.  Generated from configure.ac by autoheader.  */
+/* lib/Grid_config.h.  Generated from Grid_config.h.in by configure.  */
+/* lib/Grid_config.h.in.  Generated from configure.ac by autoheader.  */
 
 /* AVX */
 #define AVX1 1
diff --git a/lib/Grid_config.h.in b/lib/Grid_config.h.in
index 660edd5d..467ee6e4 100644
--- a/lib/Grid_config.h.in
+++ b/lib/Grid_config.h.in
@@ -1,4 +1,4 @@
-/* Grid_config.h.in.  Generated from configure.ac by autoheader.  */
+/* lib/Grid_config.h.in.  Generated from configure.ac by autoheader.  */
 
 /* AVX */
 #undef AVX1

From f678be5f94e5a4f698ee42eec91dd7bea9bc1800 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 16:17:41 +0100
Subject: [PATCH 063/429] Shaken out the peekIndex support. Hardwire constants
 "SpinIndex, ColourIndex" and LorentzIndex in Grid_QCD.h

---
 lib/Grid_Lattice.h    |  28 ++++++++++
 lib/Grid_QCD.h        |  11 +++-
 lib/Grid_math_types.h | 119 ++++++++++++++++++++++++++++++++++++++++++
 tests/Grid_main.cc    |  12 ++++-
 4 files changed, 166 insertions(+), 4 deletions(-)

diff --git a/lib/Grid_Lattice.h b/lib/Grid_Lattice.h
index b595ae0a..b9f544b5 100644
--- a/lib/Grid_Lattice.h
+++ b/lib/Grid_Lattice.h
@@ -505,6 +505,34 @@ public:
         }
         return ret;
     };
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    // Poke internal indices of a Lattice object
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    template<int Index,class vobj> inline
+    void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0]))> & rhs)
+    {
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+	  pokeIndex<Index>(lhs._odata[ss],rhs._odata[ss]);
+	}      
+    }
+    template<int Index,class vobj> inline
+    void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0],0))> & rhs,int i)
+    {
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+	  pokeIndex<Index>(lhs._odata[ss],rhs._odata[ss],i);
+	}      
+    }
+    template<int Index,class vobj> inline
+    void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0],0,0))> & rhs,int i,int j)
+    {
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+	  pokeIndex<Index>(lhs._odata[ss],rhs._odata[ss],i,j);
+	}      
+    }
+
     ////////////////////////////////////////////////////////////////////////////////////////////////////
     // Reduction operations
     ////////////////////////////////////////////////////////////////////////////////////////////////////
diff --git a/lib/Grid_QCD.h b/lib/Grid_QCD.h
index 6f85d692..c8473c28 100644
--- a/lib/Grid_QCD.h
+++ b/lib/Grid_QCD.h
@@ -1,6 +1,7 @@
 #ifndef GRID_QCD_H
 #define GRID_QCD_H
 namespace Grid{
+
 namespace QCD {
 
     static const int Nc=3;
@@ -13,12 +14,18 @@ namespace QCD {
     //////////////////////////////////////////////////////////////////////////////
     // QCD iMatrix types
     // Index conventions:                            Lorentz x Spin x Colour
-    //
+    //////////////////////////////////////////////////////////////////////////////
+    static const int ColourIndex = 1;
+    static const int SpinIndex   = 2;
+    static const int LorentzIndex= 3;
+
     // ChrisK very keen to add extra space for Gparity doubling.
     //
     // Also add domain wall index, in a way where Wilson operator 
     // naturally distributes across the 5th dimensions.
-    //////////////////////////////////////////////////////////////////////////////
+    //
+    // That probably makes for GridRedBlack4dCartesian grid.
+
     template<typename vtype> using iSinglet          = iScalar<iScalar<iScalar<vtype> > >;
     template<typename vtype> using iSpinMatrix       = iScalar<iMatrix<iScalar<vtype>, Ns> >;
     template<typename vtype> using iSpinColourMatrix = iScalar<iMatrix<iMatrix<vtype, Nc>, Ns> >;
diff --git a/lib/Grid_math_types.h b/lib/Grid_math_types.h
index 0153f46b..3429d7f6 100644
--- a/lib/Grid_math_types.h
+++ b/lib/Grid_math_types.h
@@ -1485,6 +1485,125 @@ template<int Level,class vtype,int N> inline
 }
 
 
+//////////////////////////////////////////////////////////////////////////////
+// Poke a specific index; 
+//////////////////////////////////////////////////////////////////////////////
+
+// Scalar poke
+template<int Level,class vtype> inline 
+  void pokeIndex(iScalar<vtype> &ret, 
+		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value,iScalar<vtype> >::type &arg)
+{
+  ret._internal = arg._internal;
+}
+// Vector poke, one index
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iVector<vtype,N> &ret, 
+		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::value,iScalar<vtype> >::type &arg,int i)
+{
+  ret._internal[i] = arg._internal;
+}
+// Vector poke, two indices
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iMatrix<vtype,N> &ret, 
+		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value,iScalar<vtype> >::type &arg,int i,int j)
+{
+  ret._internal[i][j] = arg._internal;
+}
+
+/////////////
+// No match poke for scalar,vector,matrix must forward on either 0,1,2 args. Must have 9 routines with notvalue
+/////////////
+// scalar
+template<int Level,class vtype> inline 
+  void pokeIndex(iScalar<vtype> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,iScalar<decltype(peekIndex<Level>(ret._internal))> >::type &arg)
+		 
+{
+  pokeIndex<Level>(ret._internal,arg._internal);
+}
+template<int Level,class vtype> inline 
+  void pokeIndex(iScalar<vtype> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,iScalar<decltype(peekIndex<Level>(ret._internal,0))> >::type &arg,
+		 int i)
+		 
+{
+  pokeIndex<Level>(ret._internal,arg._internal,i);
+}
+template<int Level,class vtype> inline 
+  void pokeIndex(iScalar<vtype> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,iScalar<decltype(peekIndex<Level>(ret._internal,0,0))> >::type &arg,
+		 int i,int j)
+		 
+{
+  pokeIndex<Level>(ret._internal,arg._internal,i,j);
+}
+
+// Vector
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iVector<vtype,N> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,iVector<decltype(peekIndex<Level>(ret._internal)),N> >::type &arg)
+		 
+{
+  for(int ii=0;ii<N;ii++){
+    pokeIndex<Level>(ret._internal[ii],arg._internal[ii]);
+  }
+}
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iVector<vtype,N> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,iVector<decltype(peekIndex<Level>(ret._internal,0)),N> >::type &arg,
+		 int i)
+		 
+{
+  for(int ii=0;ii<N;ii++){
+    pokeIndex<Level>(ret._internal[ii],arg._internal[ii],i);
+  }
+}
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iVector<vtype,N> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,iVector<decltype(peekIndex<Level>(ret._internal,0,0)),N> >::type &arg,
+		 int i,int j)
+		 
+{
+  for(int ii=0;ii<N;ii++){
+    pokeIndex<Level>(ret._internal[ii],arg._internal[ii],i,j);
+  }
+}
+
+// Matrix
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iMatrix<vtype,N> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,iMatrix<decltype(peekIndex<Level>(ret._internal)),N> >::type &arg)
+		 
+{
+  for(int ii=0;ii<N;ii++){
+  for(int jj=0;jj<N;jj++){
+    pokeIndex<Level>(ret._internal[ii][jj],arg._internal[ii][jj]);
+  }}
+}
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iMatrix<vtype,N> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,iMatrix<decltype(peekIndex<Level>(ret._internal,0)),N> >::type &arg,
+		 int i)
+		 
+{
+  for(int ii=0;ii<N;ii++){
+  for(int jj=0;jj<N;jj++){
+    pokeIndex<Level>(ret._internal[ii][jj],arg._internal[ii][jj],i);
+  }}
+}
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iMatrix<vtype,N> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,iMatrix<decltype(peekIndex<Level>(ret._internal,0,0)),N> >::type &arg,
+		 int i,int j)
+		 
+{
+  for(int ii=0;ii<N;ii++){
+  for(int jj=0;jj<N;jj++){
+    pokeIndex<Level>(ret._internal[ii][jj],arg._internal[ii][jj],i,j);
+  }}
+}
+
 /////////////////////////////////////////////////////////////////
 // Can only take the real/imag part of scalar objects, since
 // lattice objects of different complex nature are non-conformable.
diff --git a/tests/Grid_main.cc b/tests/Grid_main.cc
index 0d6c5039..d6032dd9 100644
--- a/tests/Grid_main.cc
+++ b/tests/Grid_main.cc
@@ -131,11 +131,14 @@ int main (int argc, char ** argv)
     //    rscalar=real(scalar);
     //    iscalar=imag(scalar);
     //    scalar =cmplx(rscalar,iscalar);
+    pokeIndex<1>(cVec,scalar,1);
+
 
     scalar=transpose(scalar);
     scalar=transposeIndex<1>(scalar);
     scalar=traceIndex<1>(scalar);
     scalar=peekIndex<1>(cVec,0);
+
     scalar=trace(scalar);
     scalar=localInnerProduct(cVec,cVec);
     scalar=localNorm2(cVec);
@@ -205,7 +208,7 @@ int main (int argc, char ** argv)
     LatticeComplex trscMat(&Fine);
     trscMat = trace(scMat); // Trace
 
-    {
+    { // Peek-ology and Poke-ology, with a little app-ology
       TComplex      c;
       ColourMatrix c_m;   
       SpinMatrix   s_m;   
@@ -224,9 +227,14 @@ int main (int argc, char ** argv)
       printf("c. Level %d\n",c_m().TensorLevel);
       printf("c. Level %d\n",c_m()().TensorLevel);
       
-      c_m()() = scm()(0,0); //ColourComponents of CM <= ColourComponents of SpinColourMatrix
+      c_m()()    = scm()(0,0); //ColourComponents of CM <= ColourComponents of SpinColourMatrix
       scm()(1,1) = cm()();  //ColourComponents of CM <= ColourComponents of SpinColourMatrix
+      c          = scm()(1,1)(1,2);
+      scm()(1,1)(2,1) = c;
+
+      pokeIndex<1> (c_m,c,0,0);
     }
+
     
     FooBar = Bar;
  

From 1674f899e0a87487f1b00c78b911ee45978acb2d Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 16:42:47 +0100
Subject: [PATCH 064/429] Cleaing up

---
 lib/Grid_Communicator.h |  1 +
 lib/Grid_QCD.h          | 57 ++++++++++++++++++++++++++++++++++++++++-
 tests/Grid_main.cc      |  3 ++-
 3 files changed, 59 insertions(+), 2 deletions(-)

diff --git a/lib/Grid_Communicator.h b/lib/Grid_Communicator.h
index 0e15765c..845c0123 100644
--- a/lib/Grid_Communicator.h
+++ b/lib/Grid_Communicator.h
@@ -1,5 +1,6 @@
 #ifndef GRID_COMMUNICATOR_H
 #define GRID_COMMUNICATOR_H
+
 ///////////////////////////////////
 // Processor layout information
 ///////////////////////////////////
diff --git a/lib/Grid_QCD.h b/lib/Grid_QCD.h
index c8473c28..51ef1774 100644
--- a/lib/Grid_QCD.h
+++ b/lib/Grid_QCD.h
@@ -18,6 +18,7 @@ namespace QCD {
     static const int ColourIndex = 1;
     static const int SpinIndex   = 2;
     static const int LorentzIndex= 3;
+    
 
     // ChrisK very keen to add extra space for Gparity doubling.
     //
@@ -83,8 +84,62 @@ namespace QCD {
     typedef Lattice<vLorentzColourMatrix> LatticeGaugeField;
 
 
-    
+    //////////////////////////////////////////////////////////////////////////////
+    // Peek and Poke named after physics attributes
+    //////////////////////////////////////////////////////////////////////////////
+    //spin
+    template<class vobj> auto peekSpin(const vobj &rhs,int i) -> decltype(peekIndex<SpinIndex>(rhs,0))
+    {
+      return peekIndex<SpinIndex>(rhs,i);
+    }
+    template<class vobj> auto peekSpin(const vobj &rhs,int i,int j) -> decltype(peekIndex<SpinIndex>(rhs,0,0))
+    {
+      return peekIndex<SpinIndex>(rhs,i,j);
+    }
+    template<class vobj> auto peekSpin(const Lattice<vobj> &rhs,int i) -> decltype(peekIndex<SpinIndex>(rhs,0))
+    {
+      return peekIndex<SpinIndex>(rhs,i);
+    }
+    template<class vobj> auto peekSpin(const Lattice<vobj> &rhs,int i,int j) -> decltype(peekIndex<SpinIndex>(rhs,0,0))
+    {
+      return peekIndex<SpinIndex>(rhs,i,j);
+    }
+    //colour
+    template<class vobj> auto peekColour(const vobj &rhs,int i) -> decltype(peekIndex<ColourIndex>(rhs,0))
+    {
+      return peekIndex<ColourIndex>(rhs,i);
+    }
+    template<class vobj> auto peekColour(const vobj &rhs,int i,int j) -> decltype(peekIndex<ColourIndex>(rhs,0,0))
+    {
+      return peekIndex<ColourIndex>(rhs,i,j);
+    }
+    template<class vobj> auto peekColour(const Lattice<vobj> &rhs,int i) -> decltype(peekIndex<ColourIndex>(rhs,0))
+    {
+      return peekIndex<ColourIndex>(rhs,i);
+    }
+    template<class vobj> auto peekColour(const Lattice<vobj> &rhs,int i,int j) -> decltype(peekIndex<ColourIndex>(rhs,0,0))
+    {
+      return peekIndex<ColourIndex>(rhs,i,j);
+    }
+    //lorentz
+    template<class vobj> auto peekLorentz(const vobj &rhs,int i) -> decltype(peekIndex<LorentzIndex>(rhs,0))
+    {
+      return peekIndex<LorentzIndex>(rhs,i);
+    }
+    template<class vobj> auto peekLorentz(const vobj &rhs,int i,int j) -> decltype(peekIndex<LorentzIndex>(rhs,0,0))
+    {
+      return peekIndex<LorentzIndex>(rhs,i,j);
+    }
+    template<class vobj> auto peekLorentz(const Lattice<vobj> &rhs,int i) -> decltype(peekIndex<LorentzIndex>(rhs,0))
+    {
+      return peekIndex<LorentzIndex>(rhs,i);
+    }
+    template<class vobj> auto peekLorentz(const Lattice<vobj> &rhs,int i,int j) -> decltype(peekIndex<LorentzIndex>(rhs,0,0))
+    {
+      return peekIndex<LorentzIndex>(rhs,i,j);
+    }
 
+    // FIXME this is rather generic and should find a way to place it earlier.
      inline void LatticeCoordinate(LatticeInteger &l,int mu){
       GridBase *grid = l._grid;
       int Nsimd = grid->iSites();
diff --git a/tests/Grid_main.cc b/tests/Grid_main.cc
index d6032dd9..e58d20e2 100644
--- a/tests/Grid_main.cc
+++ b/tests/Grid_main.cc
@@ -222,6 +222,7 @@ int main (int argc, char ** argv)
       
       s_m = peekIndex<1>(scm,0,0);
       c_m = peekIndex<2>(scm,1,2);
+      c_m = peekSpin(scm,1,2);
 
       printf("c. Level %d\n",c_m.TensorLevel);
       printf("c. Level %d\n",c_m().TensorLevel);
@@ -299,7 +300,7 @@ int main (int argc, char ** argv)
     int Nc = Grid::QCD::Nc;
 
     LatticeGaugeField U(&Fine);
-    //    LatticeColourMatrix Uy = U(yDir);
+    LatticeColourMatrix Uy = peekLorentz(U,1);
 
     flops = ncall*1.0*volume*(8*Nc*Nc*Nc);
     bytes = ncall*1.0*volume*Nc*Nc    *2*3*sizeof(Grid::Real);

From 18a885d195f330a7b5b5d48530cd8743a4081374 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 17:07:09 +0100
Subject: [PATCH 065/429] Renaming

---
 lib/{Grid_Cartesian.h => Grid_cartesian.h} | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 rename lib/{Grid_Cartesian.h => Grid_cartesian.h} (100%)

diff --git a/lib/Grid_Cartesian.h b/lib/Grid_cartesian.h
similarity index 100%
rename from lib/Grid_Cartesian.h
rename to lib/Grid_cartesian.h

From ac181abc95ebf08d50be04c5709457fc5e45c6af Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 17:07:37 +0100
Subject: [PATCH 066/429] Rename

---
 lib/{Grid_Communicator.h => Grid_communicator.h} | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 rename lib/{Grid_Communicator.h => Grid_communicator.h} (100%)

diff --git a/lib/Grid_Communicator.h b/lib/Grid_communicator.h
similarity index 100%
rename from lib/Grid_Communicator.h
rename to lib/Grid_communicator.h

From 2c9e5aa0544581bbb07a2389dbc3f9a39fb31506 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 17:09:24 +0100
Subject: [PATCH 067/429] Clean up capitalisation

---
 lib/{Grid_Lattice.h => Grid_lattice.h}                    | 0
 lib/{Grid_math_type_mapper.h => Grid_math_types_mapper.h} | 0
 2 files changed, 0 insertions(+), 0 deletions(-)
 rename lib/{Grid_Lattice.h => Grid_lattice.h} (100%)
 rename lib/{Grid_math_type_mapper.h => Grid_math_types_mapper.h} (100%)

diff --git a/lib/Grid_Lattice.h b/lib/Grid_lattice.h
similarity index 100%
rename from lib/Grid_Lattice.h
rename to lib/Grid_lattice.h
diff --git a/lib/Grid_math_type_mapper.h b/lib/Grid_math_types_mapper.h
similarity index 100%
rename from lib/Grid_math_type_mapper.h
rename to lib/Grid_math_types_mapper.h

From 08f20da1030cde852108c5b98c36b58204285ec4 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 17:09:48 +0100
Subject: [PATCH 068/429] Clean up caps.

---
 lib/Grid.h           | 4 ++--
 lib/Grid_cartesian.h | 3 ++-
 2 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/lib/Grid.h b/lib/Grid.h
index 2a1072ea..50fe3102 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -44,8 +44,8 @@
 #include <Grid_aligned_allocator.h>
 #include <Grid_simd.h>
 #include <Grid_math_types.h>
-#include <Grid_Cartesian.h>
-#include <Grid_Lattice.h>
+#include <Grid_cartesian.h>
+#include <Grid_lattice.h>
 #include <Grid_comparison.h>
 #include <Grid_stencil.h>
 #include <Grid_QCD.h>
diff --git a/lib/Grid_cartesian.h b/lib/Grid_cartesian.h
index c891edcc..bfb78018 100644
--- a/lib/Grid_cartesian.h
+++ b/lib/Grid_cartesian.h
@@ -2,7 +2,8 @@
 #define GRID_CARTESIAN_H
 
 #include <Grid.h>
-#include <Grid_Communicator.h>
+#include <Grid_communicator.h>
+
 namespace Grid{
     
 /////////////////////////////////////////////////////////////////////////////////////////

From f7d80aac7fc6e8d6ddb8caee60e56b34c5a51194 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 17:10:45 +0100
Subject: [PATCH 069/429] Rename

---
 lib/Grid_math_types.h | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/lib/Grid_math_types.h b/lib/Grid_math_types.h
index 3429d7f6..1042a95b 100644
--- a/lib/Grid_math_types.h
+++ b/lib/Grid_math_types.h
@@ -1,9 +1,10 @@
 #ifndef GRID_MATH_TYPES_H
 #define GRID_MATH_TYPES_H
 
-#include <Grid_math_type_mapper.h>
 #include <type_traits>
 
+#include <Grid_math_types_mapper.h>
+
 //
 // Indexing; want to be able to dereference and 
 // obtain either an lvalue or an rvalue.

From 8195d302dc24c1f132488d1922a7b7f3bcc07c7f Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 18:36:48 +0100
Subject: [PATCH 070/429] Reorganise to keep files smaller

---
 TODO                           | 129 +++---
 lib/Grid.h                     |   6 +-
 lib/Grid_cartesian.h           | 398 +-----------------
 lib/Grid_cartesian_base.h      | 200 +++++++++
 lib/Grid_cartesian_full.h      |  95 +++++
 lib/Grid_cartesian_red_black.h | 121 ++++++
 lib/Grid_cshift_mpi.h          |   4 +-
 lib/Grid_cshift_none.h         |   4 +-
 lib/Grid_lattice.h             |   3 -
 lib/Grid_math.h                | 649 ++++++++++++++++++++++++++++
 lib/Grid_math_arith.h          | 745 +++++++++++++++++++++++++++++++++
 lib/Grid_math_tensors.h        | 224 ++++++++++
 lib/Grid_math_traits.h         | 165 ++++++++
 lib/Grid_vComplexD.h           |   6 +-
 lib/Grid_vComplexF.h           |   5 +-
 lib/Grid_vInteger.h            |   6 +-
 lib/Grid_vRealD.h              |   6 +-
 lib/Grid_vRealF.h              |   5 +-
 lib/Makefile.am                |  29 +-
 19 files changed, 2308 insertions(+), 492 deletions(-)
 create mode 100644 lib/Grid_cartesian_base.h
 create mode 100644 lib/Grid_cartesian_full.h
 create mode 100644 lib/Grid_cartesian_red_black.h
 create mode 100644 lib/Grid_math.h
 create mode 100644 lib/Grid_math_arith.h
 create mode 100644 lib/Grid_math_tensors.h
 create mode 100644 lib/Grid_math_traits.h

diff --git a/TODO b/TODO
index 33032277..bfe506f7 100644
--- a/TODO
+++ b/TODO
@@ -4,6 +4,70 @@ FUNCTIONALITY:
 * Coordinate information, integers etc...            -----DONE
 * Integer type padding/union to vector.              -----DONE 
 * LatticeCoordinate[mu]                              -----DONE
+* expose traceIndex, peekIndex, transposeIndex etc at the Lattice Level -- DONE
+* TraceColor, TraceSpin.                             ----- DONE (traceIndex<1>,traceIndex<2>, transposeIndex<1>,transposeIndex<2>)
+                                                     ----- Implement mapping between traceColour and traceSpin and traceIndex<1/2>.
+* How to do U[mu] ... lorentz part of type structure or not. more like chroma if not. -- DONE
+
+* subdirs lib, tests ??                              ----- DONE
+
+Not done, or just incomplete
+
+* Consider switch std::vector to boost arrays.
+  boost::multi_array<type, 3> A()...    to replace multi1d, multi2d etc..
+
+
+* How to define simple matrix operations, such as flavour matrices?
+
+* Dirac, Pauli, SU subgroup, etc.. * Gamma/Dirac structures
+* Fourspin, two spin project
+
+* su3 exponentiation, log etc.. [Jamie's code?]
+
+* Stencil operator support                           -----Initial thoughts, trial implementation DONE.
+                                                     -----some simple tests that Stencil matches Cshift.
+                                                     -----do all permute in comms phase, so that copy permute
+						     -----cases move into a buffer.
+						     -----allow transform in/out buffers spproj
+
+* CovariantShift support                             -----Use a class to store gauge field? (parallel transport?)
+
+* Subset support, slice sums etc...                  -----Only need slice sum?
+                                                     -----Generic cartesian subslicing?
+                                                     -----Array ranges / boost extents?
+                                                     -----Multigrid grid transferral?
+                                                     -----Suggests generalised cartesian subblocking
+                                                          sums, returning modified grid?
+					             -----What should interface be?
+
+* Grid transferral
+  * pickCheckerboard, pickSubPlane, pickSubBlock,
+  *                    sumSubPlane, sumSubBlocks
+
+* rb4d support.
+
+* Check for missing functionality                    - partially audited against QDP++ layout
+
+* Optimise the extract/merge SIMD routines; Azusa??
+
+ - I have collated into single location at least.
+ - Need to use _mm_*insert/extract routines.
+
+* Conformable test in Cshift routines.
+
+
+
+* Broadcast, reduction tests. innerProduct, localInnerProduct
+
+* QDP++ regression suite and comparative benchmark
+
+* I/O support
+
+* NERSC Lattice loading, plaquette test
+
+  - MPI IO?
+  - BinaryWriter, TextWriter etc...
+  - protocol buffers?
 
 AUDITS:
 // Lattice support audit                 Tested in Grid_main.cc
@@ -18,7 +82,7 @@ AUDITS:
 //
 //     transposeIndex                     Y
 //     traceIndex                         Y
-//     peekIndex                          N; #args
+//     peekIndex                          Y
 //
 //     real,imag                          missing, semantic thought needed on real/im support.
 //                                        perhaps I just keep everything complex?
@@ -29,47 +93,15 @@ AUDITS:
 * Replace vset with a call to merge.; 
 * care in Gmerge,Gextract over vset .
 * extract / merge extra implementation removal      
-
-BUILD:
 * Test infrastructure
-* subdirs lib, tests ??
 
-* How to do U[mu] ... lorentz part of type structure or not. more like chroma if not.
-* Passing a Grid into construct iVector<LatticeColourMatrix,4>??
-*
-
-* Stencil operator support                           -----Initial thoughts, trial implementation DONE.
-                                                     -----some simple tests that Stencil matches Cshift.
-                                                     -----do all permute in comms phase, so that copy permute
-						     -----cases move into a buffer.
-						     -----allow transform in/out buffers spproj
-
-* CovariantShift support                             -----Use a class to store gauge field? (parallel transport?)
-
-POST-SFW call April 16:
-
-* TraceColor, TraceSpin.                             ----- DONE (traceIndex<1>,traceIndex<2>, transposeIndex<1>,transposeIndex<2>)
-                                                     ----- Implement mapping between traceColour and traceSpin and traceIndex<1/2>.
-
-* expose traceIndex, peekIndex, transposeIndex etc at the Lattice Level -- DONE
-
-* How to define simple matrix operations, such as flavour matrices?
-
-* Subset support, slice sums etc...                  -----Only need slice sum?
-                                                     -----Generic cartesian subslicing?
-                                                     -----Array ranges / boost extents?
-                                                     -----Multigrid grid transferral?
-                                                     -----Suggests generalised cartesian subblocking
-                                                          sums, returning modified grid?
-					             -----What should interface be?
-
-i)  Two classes of subset;   red black parity subsetting (pick checkerboard).
+[ More on subsets and grid transfers ]
+i)  Three classes of subset;   red black parity subsetting (pick checkerboard).
                              cartesian sub-block subsetting
+                             rbNd 
 
 ii) Need to be able to project one Grid to another Grid.
 
-Interface: (?)
-
 Lattice<vobj> coarse_data SubBlockSum (GridBase *CoarseGrid, Lattice<vobj> &fine_data)
 
 Operation ensure either:
@@ -89,35 +121,12 @@ Instead of subsetting
 
 iii) No general permutation map.
 
-* Consider switch std::vector to boost arrays.
-  boost::multi_array<type, 3> A()...    to replace multi1d, multi2d etc..
 
-*? Cell definition <-> sliceSum.
+ ? Cell definition <-> sliceSum.
  ? Replicated arrays.
 
-* Check for missing functionality                    - partially audited against QDP++ layout
 
-* Optimise the extract/merge SIMD routines; Azusa??
 
- - I have collated into single location at least.
- - Need to use _mm_*insert/extract routines.
-
-* Conformable test in Cshift routines.
-
-* Gamma/Dirac structures
-
-* Fourspin, two spin project
-
-* Broadcast, reduction tests. innerProduct, localInnerProduct
-
-* QDP++ regression suite and comparative benchmark
-
-* NERSC Lattice loading, plaquette test
-
-* I/O support
-  - MPI IO?
-  - BinaryWriter, TextWriter etc...
-  - protocol buffers?
 
 // Cartesian grid inheritance
 //            Grid::GridBase
diff --git a/lib/Grid.h b/lib/Grid.h
index 50fe3102..c340d678 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -7,8 +7,8 @@
 //
 
 
-#ifndef GRID_V3_H
-#define GRID_V3_H
+#ifndef GRID_H
+#define GRID_H
 
 #include <stdio.h>
 #include <complex>
@@ -43,7 +43,7 @@
 
 #include <Grid_aligned_allocator.h>
 #include <Grid_simd.h>
-#include <Grid_math_types.h>
+#include <Grid_math.h>
 #include <Grid_cartesian.h>
 #include <Grid_lattice.h>
 #include <Grid_comparison.h>
diff --git a/lib/Grid_cartesian.h b/lib/Grid_cartesian.h
index bfb78018..dbac98a8 100644
--- a/lib/Grid_cartesian.h
+++ b/lib/Grid_cartesian.h
@@ -1,400 +1,8 @@
 #ifndef GRID_CARTESIAN_H
 #define GRID_CARTESIAN_H
 
-#include <Grid.h>
-#include <Grid_communicator.h>
+#include <Grid_cartesian_base.h>
+#include <Grid_cartesian_full.h>
+#include <Grid_cartesian_red_black.h> 
 
-namespace Grid{
-    
-/////////////////////////////////////////////////////////////////////////////////////////
-// Grid Support.
-/////////////////////////////////////////////////////////////////////////////////////////
-
-class GridBase : public CartesianCommunicator {
-public:
-
- // Give Lattice access
- template<class object> friend class Lattice;
-  
- GridBase(std::vector<int> & processor_grid) : CartesianCommunicator(processor_grid) {};
-
-        
- //FIXME 
- // protected:
- // Lattice wide random support. not yet fully implemented. Need seed strategy
- // and one generator per site.
- // std::default_random_engine generator;
- //    static std::mt19937  generator( 9 );
- 
-    //////////////////////////////////////////////////////////////////////
-    // Commicator provides information on the processor grid
-    //////////////////////////////////////////////////////////////////////
-    //    unsigned long _ndimension;
-    //    std::vector<int> _processors; // processor grid
-    //    int              _processor;  // linear processor rank
-    //    std::vector<int> _processor_coor;  // linear processor rank
-    //////////////////////////////////////////////////////////////////////
-
-    // Physics Grid information.
-    std::vector<int> _simd_layout;     // Which dimensions get relayed out over simd lanes.
-    std::vector<int> _fdimensions;// Global dimensions of array prior to cb removal
-    std::vector<int> _gdimensions;// Global dimensions of array after cb removal
-    std::vector<int> _ldimensions;// local dimensions of array with processor images removed
-    std::vector<int> _rdimensions;// Reduced local dimensions with simd lane images and processor images removed 
-    std::vector<int> _ostride;    // Outer stride for each dimension
-    std::vector<int> _istride;    // Inner stride i.e. within simd lane
-    int _osites;                  // _isites*_osites = product(dimensions).
-    int _isites;
-    std::vector<int> _slice_block;   // subslice information
-    std::vector<int> _slice_stride;
-    std::vector<int> _slice_nblock;
-
-    // Might need these at some point
-    //    std::vector<int> _lstart;     // local start of array in gcoors. _processor_coor[d]*_ldimensions[d]
-    //    std::vector<int> _lend;       // local end of array in gcoors    _processor_coor[d]*_ldimensions[d]+_ldimensions_[d]-1
-
-public:
-
-    ////////////////////////////////////////////////////////////////
-    // Checkerboarding interface is virtual and overridden by 
-    // GridCartesian / GridRedBlackCartesian
-    ////////////////////////////////////////////////////////////////
-    virtual int CheckerBoarded(int dim)=0;
-    virtual int CheckerBoard(std::vector<int> site)=0;
-    virtual int CheckerBoardDestination(int source_cb,int shift)=0;
-    virtual int CheckerBoardShift(int source_cb,int dim,int shift,int osite)=0;
-    inline int  CheckerBoardFromOindex (int Oindex){
-      std::vector<int> ocoor;
-      oCoorFromOindex(ocoor,Oindex); 
-      int ss=0;
-      for(int d=0;d<_ndimension;d++){
-	ss=ss+ocoor[d];
-      }      
-      return ss&0x1;
-    }
-
-    //////////////////////////////////////////////////////////////////////////////////////////////
-    // Local layout calculations
-    //////////////////////////////////////////////////////////////////////////////////////////////
-    // These routines are key. Subdivide the linearised cartesian index into
-    //      "inner" index identifying which simd lane of object<vFcomplex> is associated with coord
-    //      "outer" index identifying which element of _odata in class "Lattice" is associated with coord.
-    //
-    // Compared to, say, Blitz++ we simply need to store BOTH an inner stride and an outer
-    // stride per dimension. The cost of evaluating the indexing information is doubled for an n-dimensional
-    // coordinate. Note, however, for data parallel operations the "inner" indexing cost is not paid and all
-    // lanes are operated upon simultaneously.
-  
-    virtual int oIndex(std::vector<int> &coor)
-    {
-        int idx=0;
-	// Works with either global or local coordinates
-        for(int d=0;d<_ndimension;d++) idx+=_ostride[d]*(coor[d]%_rdimensions[d]);
-        return idx;
-    }
-    inline int oIndexReduced(std::vector<int> &ocoor)
-    {
-      int idx=0; 
-      // ocoor is already reduced so can eliminate the modulo operation
-      // for fast indexing and inline the routine
-      for(int d=0;d<_ndimension;d++) idx+=_ostride[d]*ocoor[d];
-      return idx;
-    }
-    inline void oCoorFromOindex (std::vector<int>& coor,int Oindex){
-      coor.resize(_ndimension);
-      for(int d=0;d<_ndimension;d++){
-	coor[d] = Oindex % _rdimensions[d];
-	Oindex  = Oindex / _rdimensions[d];
-      }
-    }
-
-    //////////////////////////////////////////////////////////
-    // SIMD lane addressing
-    //////////////////////////////////////////////////////////
-    inline int iIndex(std::vector<int> &lcoor)
-    {
-        int idx=0;
-        for(int d=0;d<_ndimension;d++) idx+=_istride[d]*(lcoor[d]/_rdimensions[d]);
-        return idx;
-    }
-    inline void iCoorFromIindex(std::vector<int> &coor,int lane)
-    {
-      coor.resize(_ndimension);
-      for(int d=0;d<_ndimension;d++){
-	coor[d] = lane % _simd_layout[d];
-	lane    = lane / _simd_layout[d];
-      }
-    }
-    inline int PermuteDim(int dimension){
-      return _simd_layout[dimension]>1;
-    }
-    inline int PermuteType(int dimension){
-      int permute_type=0;
-      for(int d=_ndimension-1;d>dimension;d--){
-	if (_simd_layout[d]>1 ) permute_type++;
-      }
-      return permute_type;
-    }
-    ////////////////////////////////////////////////////////////////
-    // Array sizing queries
-    ////////////////////////////////////////////////////////////////
-
-    inline int iSites(void) { return _isites; };
-    inline int Nsimd(void)  { return _isites; };// Synonymous with iSites
-    inline int oSites(void) { return _osites; };
-    inline int lSites(void) { return _isites*_osites; }; 
-    inline int gSites(void) { return _isites*_osites*_Nprocessors; }; 
-    inline int Nd    (void) { return _ndimension;};
-    inline const std::vector<int> &FullDimensions(void)         { return _fdimensions;};
-    inline const std::vector<int> &GlobalDimensions(void)       { return _gdimensions;};
-    inline const std::vector<int> &LocalDimensions(void)        { return _ldimensions;};
-    inline const std::vector<int> &VirtualLocalDimensions(void) { return _ldimensions;};
-
-    ////////////////////////////////////////////////////////////////
-    // Global addressing
-    ////////////////////////////////////////////////////////////////
-    void RankIndexToGlobalCoor(int rank, int o_idx, int i_idx , std::vector<int> &gcoor)
-    {
-      gcoor.resize(_ndimension);
-      std::vector<int> coor(_ndimension);
-
-      ProcessorCoorFromRank(rank,coor);
-      for(int mu=0;mu<_ndimension;mu++) gcoor[mu] = _ldimensions[mu]&coor[mu];
-
-      iCoorFromIindex(coor,i_idx);
-      for(int mu=0;mu<_ndimension;mu++) gcoor[mu] += _rdimensions[mu]&coor[mu];
-
-      oCoorFromOindex (coor,o_idx);
-      for(int mu=0;mu<_ndimension;mu++) gcoor[mu] += coor[mu];
-      
-    }
-    void RankIndexCbToFullGlobalCoor(int rank, int o_idx, int i_idx, int cb,std::vector<int> &fcoor)
-    {
-      RankIndexToGlobalCoor(rank,o_idx,i_idx ,fcoor);
-      if(CheckerBoarded(0)){
-	fcoor[0] = fcoor[0]*2+cb;
-      }
-    }
-    void ProcessorCoorLocalCoorToGlobalCoor(std::vector<int> &Pcoor,std::vector<int> &Lcoor,std::vector<int> &gcoor)
-    {
-      gcoor.resize(_ndimension);
-      for(int mu=0;mu<_ndimension;mu++) gcoor[mu] = Pcoor[mu]*_ldimensions[mu]+Lcoor[mu];
-    }
-    void GlobalCoorToProcessorCoorLocalCoor(std::vector<int> &pcoor,std::vector<int> &lcoor,const std::vector<int> &gcoor)
-    {
-      pcoor.resize(_ndimension);
-      lcoor.resize(_ndimension);
-      for(int mu=0;mu<_ndimension;mu++){
-	pcoor[mu] = gcoor[mu]/_ldimensions[mu];
-	lcoor[mu] = gcoor[mu]%_ldimensions[mu];
-      }
-    }
-    void GlobalCoorToRankIndex(int &rank, int &o_idx, int &i_idx ,const std::vector<int> &gcoor)
-    {
-      std::vector<int> pcoor;
-      std::vector<int> lcoor;
-      GlobalCoorToProcessorCoorLocalCoor(pcoor,lcoor,gcoor);
-      rank = RankFromProcessorCoor(pcoor);
-      i_idx= iIndex(lcoor);
-      o_idx= oIndex(lcoor);
-    }
-
-};
-
-class GridCartesian: public GridBase {
-
-public:
-
-    virtual int CheckerBoarded(int dim){
-      return 0;
-    }
-    virtual int CheckerBoard(std::vector<int> site){
-        return 0;
-    }
-    virtual int CheckerBoardDestination(int cb,int shift){
-        return 0;
-    }
-    virtual int CheckerBoardShift(int source_cb,int dim,int shift, int osite){
-        return shift;
-    }
-    GridCartesian(std::vector<int> &dimensions,
-		  std::vector<int> &simd_layout,
-		  std::vector<int> &processor_grid
-		  ) : GridBase(processor_grid)
-    {
-        ///////////////////////
-        // Grid information
-        ///////////////////////
-        _ndimension = dimensions.size();
-            
-        _fdimensions.resize(_ndimension);
-        _gdimensions.resize(_ndimension);
-        _ldimensions.resize(_ndimension);
-        _rdimensions.resize(_ndimension);
-        _simd_layout.resize(_ndimension);
-            
-        _ostride.resize(_ndimension);
-        _istride.resize(_ndimension);
-            
-        _osites = 1;
-        _isites = 1;
-        for(int d=0;d<_ndimension;d++){
-	  _fdimensions[d] = dimensions[d]; // Global dimensions
-	  _gdimensions[d] = _fdimensions[d]; // Global dimensions
-	  _simd_layout[d] = simd_layout[d];
-
-	  //FIXME check for exact division
-
-	  // Use a reduced simd grid
-	  _ldimensions[d]= _gdimensions[d]/_processors[d];  //local dimensions
-	  _rdimensions[d]= _ldimensions[d]/_simd_layout[d]; //overdecomposition
-	  _osites *= _rdimensions[d];
-	  _isites *= _simd_layout[d];
-                
-	  // Addressing support
-	  if ( d==0 ) {
-	    _ostride[d] = 1;
-	    _istride[d] = 1;
-	  } else {
-	    _ostride[d] = _ostride[d-1]*_rdimensions[d-1];
-	    _istride[d] = _istride[d-1]*_simd_layout[d-1];
-	  }
-        }
-        
-        ///////////////////////
-        // subplane information
-        ///////////////////////
-        _slice_block.resize(_ndimension);
-        _slice_stride.resize(_ndimension);
-        _slice_nblock.resize(_ndimension);
-            
-        int block =1;
-        int nblock=1;
-        for(int d=0;d<_ndimension;d++) nblock*=_rdimensions[d];
-            
-        for(int d=0;d<_ndimension;d++){
-            nblock/=_rdimensions[d];
-            _slice_block[d] =block;
-            _slice_stride[d]=_ostride[d]*_rdimensions[d];
-            _slice_nblock[d]=nblock;
-            block = block*_rdimensions[d];
-        }
-
-    };
-};
- 
-// Specialise this for red black grids storing half the data like a chess board.
-class GridRedBlackCartesian : public GridBase
-{
-public:
-    virtual int CheckerBoarded(int dim){
-      if( dim==0) return 1;
-      else return 0;
-    }
-    virtual int CheckerBoard(std::vector<int> site){
-      return (site[0]+site[1]+site[2]+site[3])&0x1;
-    }
-
-    // Depending on the cb of site, we toggle source cb.
-    // for block #b, element #e = (b, e)
-    // we need 
-    virtual int CheckerBoardShift(int source_cb,int dim,int shift,int osite){
-
-      if(dim != 0) return shift;
-
-      int fulldim =_fdimensions[0];
-      shift = (shift+fulldim)%fulldim;
-
-      // Probably faster with table lookup;
-      // or by looping over x,y,z and multiply rather than computing checkerboard.
-      int ocb=CheckerBoardFromOindex(osite);
-	  
-      if ( (source_cb+ocb)&1 ) {
-	return (shift)/2;
-      } else {
-	return (shift+1)/2;
-      }
-    }
-
-    virtual int CheckerBoardDestination(int source_cb,int shift){
-        if ((shift+_fdimensions[0])&0x1) {
-            return 1-source_cb;
-        } else {
-            return source_cb;
-        }
-    };
-    GridRedBlackCartesian(std::vector<int> &dimensions,
-			  std::vector<int> &simd_layout,
-			  std::vector<int> &processor_grid) : GridBase(processor_grid)
-    {
-    ///////////////////////
-    // Grid information
-    ///////////////////////
-        _ndimension = dimensions.size();
-        
-        _fdimensions.resize(_ndimension);
-        _gdimensions.resize(_ndimension);
-        _ldimensions.resize(_ndimension);
-        _rdimensions.resize(_ndimension);
-        _simd_layout.resize(_ndimension);
-        
-        _ostride.resize(_ndimension);
-        _istride.resize(_ndimension);
-        
-        _osites = 1;
-        _isites = 1;
-        for(int d=0;d<_ndimension;d++){
-            _fdimensions[d] = dimensions[d];
-            _gdimensions[d] = _fdimensions[d];
-            if (d==0) _gdimensions[0] = _gdimensions[0]/2; // Remove a checkerboard
-            _ldimensions[d] = _gdimensions[d]/_processors[d];
-                
-            // Use a reduced simd grid
-            _simd_layout[d] = simd_layout[d];
-            _rdimensions[d]= _ldimensions[d]/_simd_layout[d];
-
-            _osites *= _rdimensions[d];
-            _isites *= _simd_layout[d];
-                
-            // Addressing support
-            if ( d==0 ) {
-                _ostride[d] = 1;
-                _istride[d] = 1;
-            } else {
-                _ostride[d] = _ostride[d-1]*_rdimensions[d-1];
-                _istride[d] = _istride[d-1]*_simd_layout[d-1];
-            }
-        }
-            
-        ////////////////////////////////////////////////////////////////////////////////////////////
-        // subplane information
-        ////////////////////////////////////////////////////////////////////////////////////////////
-        _slice_block.resize(_ndimension);
-        _slice_stride.resize(_ndimension);
-        _slice_nblock.resize(_ndimension);
-        
-        int block =1;
-        int nblock=1;
-        for(int d=0;d<_ndimension;d++) nblock*=_rdimensions[d];
-            
-        for(int d=0;d<_ndimension;d++){
-            nblock/=_rdimensions[d];
-            _slice_block[d] =block;
-            _slice_stride[d]=_ostride[d]*_rdimensions[d];
-            _slice_nblock[d]=nblock;
-            block = block*_rdimensions[d];
-        }
-            
-    };
-protected:
-    virtual int oIndex(std::vector<int> &coor)
-    {
-        int idx=_ostride[0]*((coor[0]/2)%_rdimensions[0]);
-        for(int d=1;d<_ndimension;d++) idx+=_ostride[d]*(coor[d]%_rdimensions[d]);
-        return idx;
-    };
-        
-};
-
-}
 #endif
diff --git a/lib/Grid_cartesian_base.h b/lib/Grid_cartesian_base.h
new file mode 100644
index 00000000..2731d9a5
--- /dev/null
+++ b/lib/Grid_cartesian_base.h
@@ -0,0 +1,200 @@
+#ifndef GRID_CARTESIAN_BASE_H
+#define GRID_CARTESIAN_BASE_H
+
+#include <Grid.h>
+#include <Grid_communicator.h>
+
+namespace Grid{
+
+class GridBase : public CartesianCommunicator {
+public:
+
+ // Give Lattice access
+ template<class object> friend class Lattice;
+  
+ GridBase(std::vector<int> & processor_grid) : CartesianCommunicator(processor_grid) {};
+
+        
+ //FIXME 
+ // protected:
+ // Lattice wide random support. not yet fully implemented. Need seed strategy
+ // and one generator per site.
+ // std::default_random_engine generator;
+ //    static std::mt19937  generator( 9 );
+ 
+    //////////////////////////////////////////////////////////////////////
+    // Commicator provides information on the processor grid
+    //////////////////////////////////////////////////////////////////////
+    //    unsigned long _ndimension;
+    //    std::vector<int> _processors; // processor grid
+    //    int              _processor;  // linear processor rank
+    //    std::vector<int> _processor_coor;  // linear processor rank
+    //////////////////////////////////////////////////////////////////////
+
+    // Physics Grid information.
+    std::vector<int> _simd_layout;     // Which dimensions get relayed out over simd lanes.
+    std::vector<int> _fdimensions;// Global dimensions of array prior to cb removal
+    std::vector<int> _gdimensions;// Global dimensions of array after cb removal
+    std::vector<int> _ldimensions;// local dimensions of array with processor images removed
+    std::vector<int> _rdimensions;// Reduced local dimensions with simd lane images and processor images removed 
+    std::vector<int> _ostride;    // Outer stride for each dimension
+    std::vector<int> _istride;    // Inner stride i.e. within simd lane
+    int _osites;                  // _isites*_osites = product(dimensions).
+    int _isites;
+    std::vector<int> _slice_block;   // subslice information
+    std::vector<int> _slice_stride;
+    std::vector<int> _slice_nblock;
+
+    // Might need these at some point
+    //    std::vector<int> _lstart;     // local start of array in gcoors. _processor_coor[d]*_ldimensions[d]
+    //    std::vector<int> _lend;       // local end of array in gcoors    _processor_coor[d]*_ldimensions[d]+_ldimensions_[d]-1
+
+public:
+
+    ////////////////////////////////////////////////////////////////
+    // Checkerboarding interface is virtual and overridden by 
+    // GridCartesian / GridRedBlackCartesian
+    ////////////////////////////////////////////////////////////////
+    virtual int CheckerBoarded(int dim)=0;
+    virtual int CheckerBoard(std::vector<int> site)=0;
+    virtual int CheckerBoardDestination(int source_cb,int shift)=0;
+    virtual int CheckerBoardShift(int source_cb,int dim,int shift,int osite)=0;
+    inline int  CheckerBoardFromOindex (int Oindex){
+      std::vector<int> ocoor;
+      oCoorFromOindex(ocoor,Oindex); 
+      int ss=0;
+      for(int d=0;d<_ndimension;d++){
+	ss=ss+ocoor[d];
+      }      
+      return ss&0x1;
+    }
+
+    //////////////////////////////////////////////////////////////////////////////////////////////
+    // Local layout calculations
+    //////////////////////////////////////////////////////////////////////////////////////////////
+    // These routines are key. Subdivide the linearised cartesian index into
+    //      "inner" index identifying which simd lane of object<vFcomplex> is associated with coord
+    //      "outer" index identifying which element of _odata in class "Lattice" is associated with coord.
+    //
+    // Compared to, say, Blitz++ we simply need to store BOTH an inner stride and an outer
+    // stride per dimension. The cost of evaluating the indexing information is doubled for an n-dimensional
+    // coordinate. Note, however, for data parallel operations the "inner" indexing cost is not paid and all
+    // lanes are operated upon simultaneously.
+  
+    virtual int oIndex(std::vector<int> &coor)
+    {
+        int idx=0;
+	// Works with either global or local coordinates
+        for(int d=0;d<_ndimension;d++) idx+=_ostride[d]*(coor[d]%_rdimensions[d]);
+        return idx;
+    }
+    inline int oIndexReduced(std::vector<int> &ocoor)
+    {
+      int idx=0; 
+      // ocoor is already reduced so can eliminate the modulo operation
+      // for fast indexing and inline the routine
+      for(int d=0;d<_ndimension;d++) idx+=_ostride[d]*ocoor[d];
+      return idx;
+    }
+    inline void oCoorFromOindex (std::vector<int>& coor,int Oindex){
+      coor.resize(_ndimension);
+      for(int d=0;d<_ndimension;d++){
+	coor[d] = Oindex % _rdimensions[d];
+	Oindex  = Oindex / _rdimensions[d];
+      }
+    }
+
+    //////////////////////////////////////////////////////////
+    // SIMD lane addressing
+    //////////////////////////////////////////////////////////
+    inline int iIndex(std::vector<int> &lcoor)
+    {
+        int idx=0;
+        for(int d=0;d<_ndimension;d++) idx+=_istride[d]*(lcoor[d]/_rdimensions[d]);
+        return idx;
+    }
+    inline void iCoorFromIindex(std::vector<int> &coor,int lane)
+    {
+      coor.resize(_ndimension);
+      for(int d=0;d<_ndimension;d++){
+	coor[d] = lane % _simd_layout[d];
+	lane    = lane / _simd_layout[d];
+      }
+    }
+    inline int PermuteDim(int dimension){
+      return _simd_layout[dimension]>1;
+    }
+    inline int PermuteType(int dimension){
+      int permute_type=0;
+      for(int d=_ndimension-1;d>dimension;d--){
+	if (_simd_layout[d]>1 ) permute_type++;
+      }
+      return permute_type;
+    }
+    ////////////////////////////////////////////////////////////////
+    // Array sizing queries
+    ////////////////////////////////////////////////////////////////
+
+    inline int iSites(void) { return _isites; };
+    inline int Nsimd(void)  { return _isites; };// Synonymous with iSites
+    inline int oSites(void) { return _osites; };
+    inline int lSites(void) { return _isites*_osites; }; 
+    inline int gSites(void) { return _isites*_osites*_Nprocessors; }; 
+    inline int Nd    (void) { return _ndimension;};
+    inline const std::vector<int> &FullDimensions(void)         { return _fdimensions;};
+    inline const std::vector<int> &GlobalDimensions(void)       { return _gdimensions;};
+    inline const std::vector<int> &LocalDimensions(void)        { return _ldimensions;};
+    inline const std::vector<int> &VirtualLocalDimensions(void) { return _ldimensions;};
+
+    ////////////////////////////////////////////////////////////////
+    // Global addressing
+    ////////////////////////////////////////////////////////////////
+    void RankIndexToGlobalCoor(int rank, int o_idx, int i_idx , std::vector<int> &gcoor)
+    {
+      gcoor.resize(_ndimension);
+      std::vector<int> coor(_ndimension);
+
+      ProcessorCoorFromRank(rank,coor);
+      for(int mu=0;mu<_ndimension;mu++) gcoor[mu] = _ldimensions[mu]&coor[mu];
+
+      iCoorFromIindex(coor,i_idx);
+      for(int mu=0;mu<_ndimension;mu++) gcoor[mu] += _rdimensions[mu]&coor[mu];
+
+      oCoorFromOindex (coor,o_idx);
+      for(int mu=0;mu<_ndimension;mu++) gcoor[mu] += coor[mu];
+      
+    }
+    void RankIndexCbToFullGlobalCoor(int rank, int o_idx, int i_idx, int cb,std::vector<int> &fcoor)
+    {
+      RankIndexToGlobalCoor(rank,o_idx,i_idx ,fcoor);
+      if(CheckerBoarded(0)){
+	fcoor[0] = fcoor[0]*2+cb;
+      }
+    }
+    void ProcessorCoorLocalCoorToGlobalCoor(std::vector<int> &Pcoor,std::vector<int> &Lcoor,std::vector<int> &gcoor)
+    {
+      gcoor.resize(_ndimension);
+      for(int mu=0;mu<_ndimension;mu++) gcoor[mu] = Pcoor[mu]*_ldimensions[mu]+Lcoor[mu];
+    }
+    void GlobalCoorToProcessorCoorLocalCoor(std::vector<int> &pcoor,std::vector<int> &lcoor,const std::vector<int> &gcoor)
+    {
+      pcoor.resize(_ndimension);
+      lcoor.resize(_ndimension);
+      for(int mu=0;mu<_ndimension;mu++){
+	pcoor[mu] = gcoor[mu]/_ldimensions[mu];
+	lcoor[mu] = gcoor[mu]%_ldimensions[mu];
+      }
+    }
+    void GlobalCoorToRankIndex(int &rank, int &o_idx, int &i_idx ,const std::vector<int> &gcoor)
+    {
+      std::vector<int> pcoor;
+      std::vector<int> lcoor;
+      GlobalCoorToProcessorCoorLocalCoor(pcoor,lcoor,gcoor);
+      rank = RankFromProcessorCoor(pcoor);
+      i_idx= iIndex(lcoor);
+      o_idx= oIndex(lcoor);
+    }
+
+};
+}
+#endif
diff --git a/lib/Grid_cartesian_full.h b/lib/Grid_cartesian_full.h
new file mode 100644
index 00000000..383bf14f
--- /dev/null
+++ b/lib/Grid_cartesian_full.h
@@ -0,0 +1,95 @@
+#ifndef GRID_CARTESIAN_FULL_H
+#define GRID_CARTESIAN_FULL_H
+
+namespace Grid{
+    
+/////////////////////////////////////////////////////////////////////////////////////////
+// Grid Support.
+/////////////////////////////////////////////////////////////////////////////////////////
+
+
+class GridCartesian: public GridBase {
+
+public:
+
+    virtual int CheckerBoarded(int dim){
+      return 0;
+    }
+    virtual int CheckerBoard(std::vector<int> site){
+        return 0;
+    }
+    virtual int CheckerBoardDestination(int cb,int shift){
+        return 0;
+    }
+    virtual int CheckerBoardShift(int source_cb,int dim,int shift, int osite){
+        return shift;
+    }
+    GridCartesian(std::vector<int> &dimensions,
+		  std::vector<int> &simd_layout,
+		  std::vector<int> &processor_grid
+		  ) : GridBase(processor_grid)
+    {
+        ///////////////////////
+        // Grid information
+        ///////////////////////
+        _ndimension = dimensions.size();
+            
+        _fdimensions.resize(_ndimension);
+        _gdimensions.resize(_ndimension);
+        _ldimensions.resize(_ndimension);
+        _rdimensions.resize(_ndimension);
+        _simd_layout.resize(_ndimension);
+            
+        _ostride.resize(_ndimension);
+        _istride.resize(_ndimension);
+            
+        _osites = 1;
+        _isites = 1;
+        for(int d=0;d<_ndimension;d++){
+	  _fdimensions[d] = dimensions[d]; // Global dimensions
+	  _gdimensions[d] = _fdimensions[d]; // Global dimensions
+	  _simd_layout[d] = simd_layout[d];
+
+	  //FIXME check for exact division
+
+	  // Use a reduced simd grid
+	  _ldimensions[d]= _gdimensions[d]/_processors[d];  //local dimensions
+	  _rdimensions[d]= _ldimensions[d]/_simd_layout[d]; //overdecomposition
+	  _osites *= _rdimensions[d];
+	  _isites *= _simd_layout[d];
+                
+	  // Addressing support
+	  if ( d==0 ) {
+	    _ostride[d] = 1;
+	    _istride[d] = 1;
+	  } else {
+	    _ostride[d] = _ostride[d-1]*_rdimensions[d-1];
+	    _istride[d] = _istride[d-1]*_simd_layout[d-1];
+	  }
+        }
+        
+        ///////////////////////
+        // subplane information
+        ///////////////////////
+        _slice_block.resize(_ndimension);
+        _slice_stride.resize(_ndimension);
+        _slice_nblock.resize(_ndimension);
+            
+        int block =1;
+        int nblock=1;
+        for(int d=0;d<_ndimension;d++) nblock*=_rdimensions[d];
+            
+        for(int d=0;d<_ndimension;d++){
+            nblock/=_rdimensions[d];
+            _slice_block[d] =block;
+            _slice_stride[d]=_ostride[d]*_rdimensions[d];
+            _slice_nblock[d]=nblock;
+            block = block*_rdimensions[d];
+        }
+
+    };
+};
+
+
+}
+#endif
diff --git a/lib/Grid_cartesian_red_black.h b/lib/Grid_cartesian_red_black.h
new file mode 100644
index 00000000..e85fd3f7
--- /dev/null
+++ b/lib/Grid_cartesian_red_black.h
@@ -0,0 +1,121 @@
+#ifndef GRID_CARTESIAN_RED_BLACK_H
+#define GRID_CARTESIAN_RED_BLACK_H
+
+
+namespace Grid {
+
+// Specialise this for red black grids storing half the data like a chess board.
+class GridRedBlackCartesian : public GridBase
+{
+public:
+    virtual int CheckerBoarded(int dim){
+      if( dim==0) return 1;
+      else return 0;
+    }
+    virtual int CheckerBoard(std::vector<int> site){
+      return (site[0]+site[1]+site[2]+site[3])&0x1;
+    }
+
+    // Depending on the cb of site, we toggle source cb.
+    // for block #b, element #e = (b, e)
+    // we need 
+    virtual int CheckerBoardShift(int source_cb,int dim,int shift,int osite){
+
+      if(dim != 0) return shift;
+
+      int fulldim =_fdimensions[0];
+      shift = (shift+fulldim)%fulldim;
+
+      // Probably faster with table lookup;
+      // or by looping over x,y,z and multiply rather than computing checkerboard.
+      int ocb=CheckerBoardFromOindex(osite);
+	  
+      if ( (source_cb+ocb)&1 ) {
+	return (shift)/2;
+      } else {
+	return (shift+1)/2;
+      }
+    }
+
+    virtual int CheckerBoardDestination(int source_cb,int shift){
+        if ((shift+_fdimensions[0])&0x1) {
+            return 1-source_cb;
+        } else {
+            return source_cb;
+        }
+    };
+    GridRedBlackCartesian(std::vector<int> &dimensions,
+			  std::vector<int> &simd_layout,
+			  std::vector<int> &processor_grid) : GridBase(processor_grid)
+    {
+    ///////////////////////
+    // Grid information
+    ///////////////////////
+        _ndimension = dimensions.size();
+        
+        _fdimensions.resize(_ndimension);
+        _gdimensions.resize(_ndimension);
+        _ldimensions.resize(_ndimension);
+        _rdimensions.resize(_ndimension);
+        _simd_layout.resize(_ndimension);
+        
+        _ostride.resize(_ndimension);
+        _istride.resize(_ndimension);
+        
+        _osites = 1;
+        _isites = 1;
+        for(int d=0;d<_ndimension;d++){
+            _fdimensions[d] = dimensions[d];
+            _gdimensions[d] = _fdimensions[d];
+            if (d==0) _gdimensions[0] = _gdimensions[0]/2; // Remove a checkerboard
+            _ldimensions[d] = _gdimensions[d]/_processors[d];
+                
+            // Use a reduced simd grid
+            _simd_layout[d] = simd_layout[d];
+            _rdimensions[d]= _ldimensions[d]/_simd_layout[d];
+
+            _osites *= _rdimensions[d];
+            _isites *= _simd_layout[d];
+                
+            // Addressing support
+            if ( d==0 ) {
+                _ostride[d] = 1;
+                _istride[d] = 1;
+            } else {
+                _ostride[d] = _ostride[d-1]*_rdimensions[d-1];
+                _istride[d] = _istride[d-1]*_simd_layout[d-1];
+            }
+        }
+            
+        ////////////////////////////////////////////////////////////////////////////////////////////
+        // subplane information
+        ////////////////////////////////////////////////////////////////////////////////////////////
+        _slice_block.resize(_ndimension);
+        _slice_stride.resize(_ndimension);
+        _slice_nblock.resize(_ndimension);
+        
+        int block =1;
+        int nblock=1;
+        for(int d=0;d<_ndimension;d++) nblock*=_rdimensions[d];
+            
+        for(int d=0;d<_ndimension;d++){
+            nblock/=_rdimensions[d];
+            _slice_block[d] =block;
+            _slice_stride[d]=_ostride[d]*_rdimensions[d];
+            _slice_nblock[d]=nblock;
+            block = block*_rdimensions[d];
+        }
+            
+    };
+protected:
+    virtual int oIndex(std::vector<int> &coor)
+    {
+        int idx=_ostride[0]*((coor[0]/2)%_rdimensions[0]);
+        for(int d=1;d<_ndimension;d++) idx+=_ostride[d]*(coor[d]%_rdimensions[d]);
+        return idx;
+    };
+        
+};
+
+}
+#endif
diff --git a/lib/Grid_cshift_mpi.h b/lib/Grid_cshift_mpi.h
index 4ca8f6b5..f51becf3 100644
--- a/lib/Grid_cshift_mpi.h
+++ b/lib/Grid_cshift_mpi.h
@@ -1,5 +1,5 @@
-#ifndef _GRID_MPI_CSHIFT_H_
-#define _GRID_MPI_CSHIFT_H_
+#ifndef _GRID_CSHIFT_MPI_H_
+#define _GRID_CSHIFT_MPI_H_
 
 #ifndef MAX
 #define MAX(x,y) ((x)>(y)?(x):(y))
diff --git a/lib/Grid_cshift_none.h b/lib/Grid_cshift_none.h
index 402e52b4..7f4f7bc1 100644
--- a/lib/Grid_cshift_none.h
+++ b/lib/Grid_cshift_none.h
@@ -1,5 +1,5 @@
-#ifndef _GRID_NONE_CSHIFT_H_
-#define _GRID_NONE_CSHIFT_H_
+#ifndef _GRID_CSHIFT_NONE_H_
+#define _GRID_CSHIFT_NONE_H_
 
 friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
 {
diff --git a/lib/Grid_lattice.h b/lib/Grid_lattice.h
index b9f544b5..be95caa0 100644
--- a/lib/Grid_lattice.h
+++ b/lib/Grid_lattice.h
@@ -1,9 +1,6 @@
 #ifndef GRID_LATTICE_H
 #define GRID_LATTICE_H
 
-#include "Grid.h"
-
-
 namespace Grid {
 
 // TODO: Indexing ()
diff --git a/lib/Grid_math.h b/lib/Grid_math.h
new file mode 100644
index 00000000..bad20054
--- /dev/null
+++ b/lib/Grid_math.h
@@ -0,0 +1,649 @@
+#ifndef GRID_MATH_H
+#define GRID_MATH_H
+
+#include <Grid_math_traits.h>
+#include <Grid_math_tensors.h>
+#include <Grid_math_arith.h>
+
+//
+// Indexing; want to be able to dereference and 
+// obtain either an lvalue or an rvalue.
+//
+namespace Grid {
+
+
+
+
+    ///////////////////////////////////////////////////////////////////////////////////////
+    // innerProduct Scalar x Scalar -> Scalar
+    // innerProduct Vector x Vector -> Scalar
+    // innerProduct Matrix x Matrix -> Scalar
+    ///////////////////////////////////////////////////////////////////////////////////////
+    template<class l,class r,int N> inline
+    auto innerProduct (const iVector<l,N>& lhs,const iVector<r,N>& rhs) -> iScalar<decltype(innerProduct(lhs._internal[0],rhs._internal[0]))>
+    {
+        typedef decltype(innerProduct(lhs._internal[0],rhs._internal[0])) ret_t;
+        iScalar<ret_t> ret=zero;
+        for(int c1=0;c1<N;c1++){
+            ret._internal += innerProduct(lhs._internal[c1],rhs._internal[c1]);
+        }
+        return ret;
+    }
+    template<class l,class r,int N> inline
+    auto innerProduct (const iMatrix<l,N>& lhs,const iMatrix<r,N>& rhs) -> iScalar<decltype(innerProduct(lhs._internal[0][0],rhs._internal[0][0]))>
+    {
+        typedef decltype(innerProduct(lhs._internal[0][0],rhs._internal[0][0])) ret_t;
+        iScalar<ret_t> ret=zero;
+        iScalar<ret_t> tmp;
+        for(int c1=0;c1<N;c1++){
+        for(int c2=0;c2<N;c2++){
+	  ret._internal+=innerProduct(lhs._internal[c1][c2],rhs._internal[c1][c2]);
+        }}
+        return ret;
+    }
+    template<class l,class r> inline
+    auto innerProduct (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(innerProduct(lhs._internal,rhs._internal))>
+    {
+        typedef decltype(innerProduct(lhs._internal,rhs._internal)) ret_t;
+        iScalar<ret_t> ret;
+        ret._internal = innerProduct(lhs._internal,rhs._internal);
+        return ret;
+    }
+
+    ///////////////////////////////////////////////////////////////////////////////////////
+    // outerProduct Scalar x Scalar -> Scalar
+    //              Vector x Vector -> Matrix
+    ///////////////////////////////////////////////////////////////////////////////////////
+
+template<class l,class r,int N> inline
+auto outerProduct (const iVector<l,N>& lhs,const iVector<r,N>& rhs) -> iMatrix<decltype(outerProduct(lhs._internal[0],rhs._internal[0])),N>
+{
+    typedef decltype(outerProduct(lhs._internal[0],rhs._internal[0])) ret_t;
+    iMatrix<ret_t,N> ret;
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        ret._internal[c1][c2] = outerProduct(lhs._internal[c1],rhs._internal[c2]);
+    }}
+    return ret;
+}
+template<class l,class r> inline
+auto outerProduct (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(outerProduct(lhs._internal,rhs._internal))>
+{
+    typedef decltype(outerProduct(lhs._internal,rhs._internal)) ret_t;
+    iScalar<ret_t> ret;
+    ret._internal = outerProduct(lhs._internal,rhs._internal);
+    return ret;
+}
+
+inline ComplexF outerProduct(const ComplexF &l, const ComplexF& r)
+{
+  return l*r;
+}
+inline ComplexD outerProduct(const ComplexD &l, const ComplexD& r)
+{
+  return l*r;
+}
+inline RealF outerProduct(const RealF &l, const RealF& r)
+{
+  return l*r;
+}
+inline RealD outerProduct(const RealD &l, const RealD& r)
+{
+  return l*r;
+}
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    /////////////////////////////////////////// CONJ         ///////////////////////////////////////////
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+ 
+// Conj function for scalar, vector, matrix
+template<class vtype> inline iScalar<vtype> conj(const iScalar<vtype>&r)
+{
+    iScalar<vtype> ret;
+    ret._internal = conj(r._internal);
+    return ret;
+}
+template<class vtype,int N> inline iVector<vtype,N> conj(const iVector<vtype,N>&r)
+{
+  iVector<vtype,N> ret;
+  for(int i=0;i<N;i++){
+    ret._internal[i] = conj(r._internal[i]);
+  }
+  return ret;
+}
+template<class vtype,int N> inline iMatrix<vtype,N> conj(const iMatrix<vtype,N>&r)
+{
+  iMatrix<vtype,N> ret;
+  for(int i=0;i<N;i++){
+  for(int j=0;j<N;j++){
+    ret._internal[i][j] = conj(r._internal[i][j]);
+  }}
+  return ret;
+}
+
+// Adj function for scalar, vector, matrix
+template<class vtype> inline iScalar<vtype> adj(const iScalar<vtype>&r)
+{
+    iScalar<vtype> ret;
+    ret._internal = adj(r._internal);
+    return ret;
+}
+template<class vtype,int N> inline iVector<vtype,N> adj(const iVector<vtype,N>&r)
+{
+    iVector<vtype,N> ret;
+    for(int i=0;i<N;i++){
+        ret._internal[i] = adj(r._internal[i]);
+    }
+    return ret;
+}
+template<class vtype,int N> inline iMatrix<vtype,N> adj(const iMatrix<vtype,N> &arg)
+{
+    iMatrix<vtype,N> ret;
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        ret._internal[c1][c2]=adj(arg._internal[c2][c1]);
+    }}
+    return ret;
+}
+
+
+
+/////////////////////////////////////////////////////////////////
+// Transpose all indices
+/////////////////////////////////////////////////////////////////
+
+inline ComplexD transpose(ComplexD &rhs){  return rhs;}
+inline ComplexF transpose(ComplexF &rhs){  return rhs;}
+inline RealD transpose(RealD &rhs){  return rhs;}
+inline RealF transpose(RealF &rhs){  return rhs;}
+
+template<class vtype,int N>
+  inline typename std::enable_if<isGridTensor<vtype>::value, iMatrix<vtype,N> >::type 
+  transpose(iMatrix<vtype,N> arg)
+  {
+    iMatrix<vtype,N> ret;
+    for(int i=0;i<N;i++){
+      for(int j=0;j<N;j++){
+	ret._internal[i][j] = transpose(arg._internal[j][i]); // NB recurses
+      }}
+    return ret;
+  }
+template<class vtype,int N>
+  inline typename std::enable_if<isGridTensor<vtype>::notvalue, iMatrix<vtype,N> >::type 
+  transpose(iMatrix<vtype,N> arg)
+  {
+    iMatrix<vtype,N> ret;
+    for(int i=0;i<N;i++){
+      for(int j=0;j<N;j++){
+	ret._internal[i][j] = arg._internal[j][i]; // Stop recursion if not a tensor type
+      }}
+    return ret;
+  }
+
+template<class vtype>
+  inline typename std::enable_if<isGridTensor<vtype>::value, iScalar<vtype> >::type 
+  transpose(iScalar<vtype> arg)
+  {
+    iScalar<vtype> ret;
+    ret._internal = transpose(arg._internal); // NB recurses
+    return ret;
+  }
+
+template<class vtype>
+  inline typename std::enable_if<isGridTensor<vtype>::notvalue, iScalar<vtype> >::type 
+  transpose(iScalar<vtype> arg)
+  {
+    iScalar<vtype> ret;
+    ret._internal = arg._internal; // NB recursion stops
+    return ret;
+  }
+
+
+////////////////////////////////////////////////////////////////////////////////////////////
+// Transpose a specific index; instructive to compare this style of recursion termination
+// to that of adj; which is easiers?
+////////////////////////////////////////////////////////////////////////////////////////////
+template<int Level,class vtype,int N> inline 
+  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value, iMatrix<vtype,N> >::type 
+transposeIndex (const iMatrix<vtype,N> &arg)
+{
+  iMatrix<vtype,N> ret;
+  for(int i=0;i<N;i++){
+    for(int j=0;j<N;j++){
+      ret._internal[i][j] = arg._internal[j][i]; 
+  }}
+  return ret;
+}
+// or not
+template<int Level,class vtype,int N> inline 
+typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue, iMatrix<vtype,N> >::type 
+transposeIndex (const iMatrix<vtype,N> &arg)
+{
+  iMatrix<vtype,N> ret;
+  for(int i=0;i<N;i++){
+    for(int j=0;j<N;j++){
+      ret._internal[i][j] = transposeIndex<Level>(arg._internal[i][j]); 
+  }}
+  return ret;
+}
+template<int Level,class vtype> inline 
+typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue, iScalar<vtype> >::type 
+transposeIndex (const iScalar<vtype> &arg)
+{
+  iScalar<vtype> ret;
+  ret._internal=transposeIndex<Level>(arg._internal);
+  return ret;
+}
+template<int Level,class vtype> inline 
+typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value, iScalar<vtype> >::type 
+transposeIndex (const iScalar<vtype> &arg)
+{
+  return arg;
+}
+
+//////////////////////////////////////////////////////////////////
+// Traces: both all indices and a specific index 
+/////////////////////////////////////////////////////////////////
+
+inline ComplexF trace( const ComplexF &arg){    return arg;}
+inline ComplexD trace( const ComplexD &arg){    return arg;}
+inline RealF trace( const RealF &arg){    return arg;}
+inline RealD trace( const RealD &arg){    return arg;}
+
+template<int Level> inline ComplexF traceIndex(const ComplexF arg) { return arg;}
+template<int Level> inline ComplexD traceIndex(const ComplexD arg) { return arg;}
+template<int Level> inline RealF traceIndex(const RealF arg) { return arg;}
+template<int Level> inline RealD traceIndex(const RealD arg) { return arg;}
+
+template<class vtype,int N>
+inline auto trace(const iMatrix<vtype,N> &arg) -> iScalar<decltype(trace(arg._internal[0][0]))>
+{
+    iScalar<decltype( trace(arg._internal[0][0] )) > ret;
+    zeroit(ret._internal);
+    for(int i=0;i<N;i++){
+        ret._internal=ret._internal+trace(arg._internal[i][i]);
+    }
+    return ret;
+}
+template<class vtype>
+inline auto trace(const iScalar<vtype> &arg) -> iScalar<decltype(trace(arg._internal))>
+{
+    iScalar<decltype(trace(arg._internal))> ret;
+    ret._internal=trace(arg._internal);
+    return ret;
+}
+////////////////////////////////////////////////////////////////////////////////////////////////////////
+// Trace Specific indices.
+////////////////////////////////////////////////////////////////////////////////////////////////////////
+/*
+template<int Level,class vtype> inline 
+auto traceIndex(const iScalar<vtype> &arg) -> iScalar<decltype(traceIndex<Level>(arg._internal)) >
+{
+  iScalar<decltype(traceIndex<Level>(arg._internal))> ret;
+  ret._internal = traceIndex<Level>(arg._internal);
+  return ret;
+}
+*/
+template<int Level,class vtype> inline auto 
+traceIndex (const iScalar<vtype> &arg) ->
+typename 
+std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue, 
+  iScalar<decltype(traceIndex<Level>(arg._internal))> >::type 
+
+{
+  iScalar<decltype(traceIndex<Level>(arg._internal))> ret;
+  ret._internal=traceIndex<Level>(arg._internal);
+  return ret;
+}
+template<int Level,class vtype> inline auto
+traceIndex (const iScalar<vtype> &arg) ->
+typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value, 
+                        iScalar<vtype> >::type 
+{
+  return arg;
+}
+
+// If we hit the right index, return scalar and trace it with no further recursion
+template<int Level,class vtype,int N> inline 
+auto traceIndex(const iMatrix<vtype,N> &arg) ->
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value,  // Index matches
+                                                    iScalar<vtype> >::type                              // return scalar
+{
+  iScalar<vtype> ret;
+  zeroit(ret._internal);
+  for(int i=0;i<N;i++){
+    ret._internal = ret._internal + arg._internal[i][i];
+  }
+  return ret;
+}
+
+// not this level, so recurse
+template<int Level,class vtype,int N> inline 
+auto traceIndex(const iMatrix<vtype,N> &arg) ->
+  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,// No index match
+         iMatrix<decltype(traceIndex<Level>(arg._internal[0][0])),N> >::type     // return matrix
+{
+  iMatrix<decltype(traceIndex<Level>(arg._internal[0][0])),N> ret;
+  for(int i=0;i<N;i++){
+  for(int j=0;j<N;j++){
+    ret._internal[i][j] = traceIndex<Level>(arg._internal[i][j]);
+  }}
+  return ret;
+}
+
+//////////////////////////////////////////////////////////////////////////////
+// Peek on a specific index; returns a scalar in that index, tensor inherits rest
+//////////////////////////////////////////////////////////////////////////////
+// If we hit the right index, return scalar with no further recursion
+
+//template<int Level> inline ComplexF peekIndex(const ComplexF arg) { return arg;}
+//template<int Level> inline ComplexD peekIndex(const ComplexD arg) { return arg;}
+//template<int Level> inline RealF peekIndex(const RealF arg) { return arg;}
+//template<int Level> inline RealD peekIndex(const RealD arg) { return arg;}
+
+// Scalar peek, no indices
+template<int Level,class vtype> inline 
+  auto peekIndex(const iScalar<vtype> &arg) -> 
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value,  // Index matches
+  iScalar<vtype> >::type                              // return scalar
+{
+  return arg;
+}
+// Vector peek, one index
+template<int Level,class vtype,int N> inline 
+  auto peekIndex(const iVector<vtype,N> &arg,int i) -> 
+  typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::value,  // Index matches
+  iScalar<vtype> >::type                              // return scalar
+{
+  iScalar<vtype> ret;                              // return scalar
+  ret._internal = arg._internal[i];
+  return ret;
+}
+// Matrix peek, two indices
+template<int Level,class vtype,int N> inline 
+  auto peekIndex(const iMatrix<vtype,N> &arg,int i,int j) -> 
+  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value,  // Index matches
+  iScalar<vtype> >::type                              // return scalar
+{
+  iScalar<vtype> ret;                              // return scalar
+  ret._internal = arg._internal[i][j];
+  return ret;
+}
+
+/////////////
+// No match peek for scalar,vector,matrix must forward on either 0,1,2 args. Must have 9 routines with notvalue
+/////////////
+// scalar
+template<int Level,class vtype> inline 
+  auto peekIndex(const iScalar<vtype> &arg) ->                                     // Scalar 0 index  
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does NOT match
+  iScalar<decltype(peekIndex<Level>(arg._internal))> >::type                       
+{
+  iScalar<decltype(peekIndex<Level>(arg._internal))> ret;
+  ret._internal= peekIndex<Level>(arg._internal);
+  return ret;
+}
+template<int Level,class vtype> inline 
+  auto peekIndex(const iScalar<vtype> &arg,int i) ->                             // Scalar 1 index
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does NOT match
+  iScalar<decltype(peekIndex<Level>(arg._internal,i))> >::type                       
+{
+  iScalar<decltype(peekIndex<Level>(arg._internal,i))> ret;
+  ret._internal=peekIndex<Level>(arg._internal,i);
+  return ret;
+}
+template<int Level,class vtype> inline 
+  auto peekIndex(const iScalar<vtype> &arg,int i,int j) ->                         // Scalar, 2 index
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does NOT match
+  iScalar<decltype(peekIndex<Level>(arg._internal,i,j))> >::type                       
+{
+  iScalar<decltype(peekIndex<Level>(arg._internal,i,j))> ret;
+  ret._internal=peekIndex<Level>(arg._internal,i,j);
+  return ret;
+}
+// vector
+template<int Level,class vtype,int N> inline 
+auto peekIndex(const iVector<vtype,N> &arg) -> 
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does not match
+  iVector<decltype(peekIndex<Level>(arg._internal[0])),N> >::type                       
+{
+  iVector<decltype(peekIndex<Level>(arg._internal[0])),N> ret;
+  for(int ii=0;ii<N;ii++){
+    ret._internal[ii]=peekIndex<Level>(arg._internal[ii]);
+  }
+  return ret;
+}
+template<int Level,class vtype,int N> inline 
+  auto peekIndex(const iVector<vtype,N> &arg,int i) -> 
+  typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,  // Index does not match
+  iVector<decltype(peekIndex<Level>(arg._internal[0],i)),N> >::type                       
+{
+  iVector<decltype(peekIndex<Level>(arg._internal[0],i)),N> ret;
+  for(int ii=0;ii<N;ii++){
+    ret._internal[ii]=peekIndex<Level>(arg._internal[ii],i);
+  }
+  return ret;
+}
+template<int Level,class vtype,int N> inline 
+  auto peekIndex(const iVector<vtype,N> &arg,int i,int j) -> 
+  typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,  // Index does not match
+  iVector<decltype(peekIndex<Level>(arg._internal[0],i,j)),N> >::type                       
+{
+  iVector<decltype(peekIndex<Level>(arg._internal[0],i,j)),N> ret;
+  for(int ii=0;ii<N;ii++){
+    ret._internal[ii]=peekIndex<Level>(arg._internal[ii],i,j);
+  }
+  return ret;
+}
+// matrix
+template<int Level,class vtype,int N> inline 
+auto peekIndex(const iMatrix<vtype,N> &arg) -> 
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does not match
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0])),N> >::type                       
+{
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0])),N> ret;
+  for(int ii=0;ii<N;ii++){
+  for(int jj=0;jj<N;jj++){
+    ret._internal[ii][jj]=peekIndex<Level>(arg._internal[ii][jj]);// Could avoid this because peeking a scalar is dumb
+  }}
+  return ret;
+}
+template<int Level,class vtype,int N> inline 
+  auto peekIndex(const iMatrix<vtype,N> &arg,int i) -> 
+  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,  // Index does not match
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0],i)),N> >::type                       
+{
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0],i)),N> ret;
+  for(int ii=0;ii<N;ii++){
+  for(int jj=0;jj<N;jj++){
+    ret._internal[ii][jj]=peekIndex<Level>(arg._internal[ii][jj],i);
+  }}
+  return ret;
+}
+template<int Level,class vtype,int N> inline 
+  auto peekIndex(const iMatrix<vtype,N> &arg,int i,int j) -> 
+  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,  // Index does not match
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0],i,j)),N> >::type                       
+{
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0],i,j)),N> ret;
+  for(int ii=0;ii<N;ii++){
+  for(int jj=0;jj<N;jj++){
+    ret._internal[ii][jj]=peekIndex<Level>(arg._internal[ii][jj],i,j);
+  }}
+  return ret;
+}
+
+
+//////////////////////////////////////////////////////////////////////////////
+// Poke a specific index; 
+//////////////////////////////////////////////////////////////////////////////
+
+// Scalar poke
+template<int Level,class vtype> inline 
+  void pokeIndex(iScalar<vtype> &ret, 
+		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value,iScalar<vtype> >::type &arg)
+{
+  ret._internal = arg._internal;
+}
+// Vector poke, one index
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iVector<vtype,N> &ret, 
+		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::value,iScalar<vtype> >::type &arg,int i)
+{
+  ret._internal[i] = arg._internal;
+}
+// Vector poke, two indices
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iMatrix<vtype,N> &ret, 
+		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value,iScalar<vtype> >::type &arg,int i,int j)
+{
+  ret._internal[i][j] = arg._internal;
+}
+
+/////////////
+// No match poke for scalar,vector,matrix must forward on either 0,1,2 args. Must have 9 routines with notvalue
+/////////////
+// scalar
+template<int Level,class vtype> inline 
+  void pokeIndex(iScalar<vtype> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,iScalar<decltype(peekIndex<Level>(ret._internal))> >::type &arg)
+		 
+{
+  pokeIndex<Level>(ret._internal,arg._internal);
+}
+template<int Level,class vtype> inline 
+  void pokeIndex(iScalar<vtype> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,iScalar<decltype(peekIndex<Level>(ret._internal,0))> >::type &arg,
+		 int i)
+		 
+{
+  pokeIndex<Level>(ret._internal,arg._internal,i);
+}
+template<int Level,class vtype> inline 
+  void pokeIndex(iScalar<vtype> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,iScalar<decltype(peekIndex<Level>(ret._internal,0,0))> >::type &arg,
+		 int i,int j)
+		 
+{
+  pokeIndex<Level>(ret._internal,arg._internal,i,j);
+}
+
+// Vector
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iVector<vtype,N> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,iVector<decltype(peekIndex<Level>(ret._internal)),N> >::type &arg)
+		 
+{
+  for(int ii=0;ii<N;ii++){
+    pokeIndex<Level>(ret._internal[ii],arg._internal[ii]);
+  }
+}
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iVector<vtype,N> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,iVector<decltype(peekIndex<Level>(ret._internal,0)),N> >::type &arg,
+		 int i)
+		 
+{
+  for(int ii=0;ii<N;ii++){
+    pokeIndex<Level>(ret._internal[ii],arg._internal[ii],i);
+  }
+}
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iVector<vtype,N> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,iVector<decltype(peekIndex<Level>(ret._internal,0,0)),N> >::type &arg,
+		 int i,int j)
+		 
+{
+  for(int ii=0;ii<N;ii++){
+    pokeIndex<Level>(ret._internal[ii],arg._internal[ii],i,j);
+  }
+}
+
+// Matrix
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iMatrix<vtype,N> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,iMatrix<decltype(peekIndex<Level>(ret._internal)),N> >::type &arg)
+		 
+{
+  for(int ii=0;ii<N;ii++){
+  for(int jj=0;jj<N;jj++){
+    pokeIndex<Level>(ret._internal[ii][jj],arg._internal[ii][jj]);
+  }}
+}
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iMatrix<vtype,N> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,iMatrix<decltype(peekIndex<Level>(ret._internal,0)),N> >::type &arg,
+		 int i)
+		 
+{
+  for(int ii=0;ii<N;ii++){
+  for(int jj=0;jj<N;jj++){
+    pokeIndex<Level>(ret._internal[ii][jj],arg._internal[ii][jj],i);
+  }}
+}
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iMatrix<vtype,N> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,iMatrix<decltype(peekIndex<Level>(ret._internal,0,0)),N> >::type &arg,
+		 int i,int j)
+		 
+{
+  for(int ii=0;ii<N;ii++){
+  for(int jj=0;jj<N;jj++){
+    pokeIndex<Level>(ret._internal[ii][jj],arg._internal[ii][jj],i,j);
+  }}
+}
+
+/////////////////////////////////////////////////////////////////
+// Can only take the real/imag part of scalar objects, since
+// lattice objects of different complex nature are non-conformable.
+/////////////////////////////////////////////////////////////////
+template<class itype> inline auto real(const iScalar<itype> &z) -> iScalar<decltype(real(z._internal))>
+{
+    iScalar<decltype(real(z._internal))> ret;
+    ret._internal = real(z._internal);
+    return ret;
+}
+template<class itype,int N> inline auto real(const iMatrix<itype,N> &z) -> iMatrix<decltype(real(z._internal[0][0])),N>
+{
+    iMatrix<decltype(real(z._internal[0][0])),N> ret;
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        ret._internal[c1][c2] = real(z._internal[c1][c2]);
+    }}
+    return ret;
+}
+template<class itype,int N> inline auto real(const iVector<itype,N> &z) -> iVector<decltype(real(z._internal[0])),N>
+{
+    iVector<decltype(real(z._internal[0])),N> ret;
+    for(int c1=0;c1<N;c1++){
+        ret._internal[c1] = real(z._internal[c1]);
+    }
+    return ret;
+}
+    
+template<class itype> inline auto imag(const iScalar<itype> &z) -> iScalar<decltype(imag(z._internal))>
+{
+    iScalar<decltype(imag(z._internal))> ret;
+    ret._internal = imag(z._internal);
+    return ret;
+}
+template<class itype,int N> inline auto imag(const iMatrix<itype,N> &z) -> iMatrix<decltype(imag(z._internal[0][0])),N>
+{
+    iMatrix<decltype(imag(z._internal[0][0])),N> ret;
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        ret._internal[c1][c2] = imag(z._internal[c1][c2]);
+    }}
+    return ret;
+}
+template<class itype,int N> inline auto imag(const iVector<itype,N> &z) -> iVector<decltype(imag(z._internal[0])),N>
+{
+    iVector<decltype(imag(z._internal[0])),N> ret;
+    for(int c1=0;c1<N;c1++){
+        ret._internal[c1] = imag(z._internal[c1]);
+    }
+    return ret;
+}
+
+};
+    
+#endif
diff --git a/lib/Grid_math_arith.h b/lib/Grid_math_arith.h
new file mode 100644
index 00000000..aebaaee0
--- /dev/null
+++ b/lib/Grid_math_arith.h
@@ -0,0 +1,745 @@
+#ifndef GRID_MATH_ARITH_H
+#define GRID_MATH_ARITH_H
+
+namespace Grid {
+
+    
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    /////////////////////////////////////////// ADD         ///////////////////////////////////////////
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    
+
+// ADD is simple for now; cannot mix types and straightforward template
+// Scalar +/- Scalar
+// Vector +/- Vector
+// Matrix +/- Matrix
+  template<class vtype,class ltype,class rtype> inline void add(iScalar<vtype> * __restrict__ ret,
+								const iScalar<ltype> * __restrict__ lhs,
+								const iScalar<rtype> * __restrict__ rhs)
+  {
+    add(&ret->_internal,&lhs->_internal,&rhs->_internal);
+  }
+  template<class vtype,class ltype,class rtype,int N> inline void add(iVector<vtype,N> * __restrict__ ret,
+								      const iVector<ltype,N> * __restrict__ lhs,
+								      const iVector<rtype,N> * __restrict__ rhs)
+  {
+    for(int c=0;c<N;c++){
+      ret->_internal[c]=lhs->_internal[c]+rhs->_internal[c];
+    }
+    return;
+  }
+  
+  template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
+									const iMatrix<ltype,N> * __restrict__ lhs,
+									const iMatrix<rtype,N> * __restrict__ rhs)
+  {
+    for(int c2=0;c2<N;c2++){
+      for(int c1=0;c1<N;c1++){
+        add(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal[c1][c2]);
+      }}
+    return;
+  }
+  template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
+									const iScalar<ltype>   * __restrict__ lhs,
+									const iMatrix<rtype,N> * __restrict__ rhs)
+  {
+    for(int c2=0;c2<N;c2++){
+      for(int c1=0;c1<N;c1++){
+        add(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
+      }}
+    return;
+  }
+  template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
+									const iMatrix<ltype,N> * __restrict__ lhs,
+									const iScalar<rtype>   * __restrict__ rhs)
+  {
+    for(int c2=0;c2<N;c2++){
+      for(int c1=0;c1<N;c1++){
+        if ( c1==c2)
+	  add(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
+        else
+	  ret->_internal[c1][c2]=lhs->_internal[c1][c2];
+      }}
+    return;
+  }
+
+  // Need to figure multi-precision.
+  template<class Mytype>  Mytype timesI(Mytype &r)
+    {
+      iScalar<Complex> i;
+      i._internal = Complex(0,1);
+      return r*i;
+    }
+
+  // + operator for scalar, vector, matrix
+  template<class ltype,class rtype>
+    //inline auto operator + (iScalar<ltype>& lhs,iScalar<rtype>&& rhs) -> iScalar<decltype(lhs._internal + rhs._internal)>
+    inline auto operator + (const iScalar<ltype>& lhs,const iScalar<rtype>& rhs) -> iScalar<decltype(lhs._internal + rhs._internal)>
+  {
+    typedef iScalar<decltype(lhs._internal+rhs._internal)> ret_t;
+    ret_t ret;
+    add(&ret,&lhs,&rhs);
+    return ret;
+  }
+  template<class ltype,class rtype,int N>
+    inline auto operator + (const iVector<ltype,N>& lhs,const iVector<rtype,N>& rhs) ->iVector<decltype(lhs._internal[0]+rhs._internal[0]),N>
+    {
+      typedef iVector<decltype(lhs._internal[0]+rhs._internal[0]),N> ret_t;
+    ret_t ret;
+    add(&ret,&lhs,&rhs);
+    return ret;
+    }
+  template<class ltype,class rtype,int N>
+    inline auto operator + (const iMatrix<ltype,N>& lhs,const iMatrix<rtype,N>& rhs) ->iMatrix<decltype(lhs._internal[0][0]+rhs._internal[0][0]),N>
+    {
+      typedef iMatrix<decltype(lhs._internal[0][0]+rhs._internal[0][0]),N> ret_t;
+      ret_t ret;
+      add(&ret,&lhs,&rhs);
+      return ret;
+    }
+  template<class ltype,class rtype,int N>
+inline auto operator + (const iScalar<ltype>& lhs,const iMatrix<rtype,N>& rhs)->iMatrix<decltype(lhs._internal+rhs._internal[0][0]),N>
+    {
+      typedef iMatrix<decltype(lhs._internal+rhs._internal[0][0]),N> ret_t;
+      ret_t ret;
+      add(&ret,&lhs,&rhs);
+      return ret;
+    }
+
+  template<class ltype,class rtype,int N>
+    inline auto operator + (const iMatrix<ltype,N>& lhs,const iScalar<rtype>& rhs)->iMatrix<decltype(lhs._internal[0][0]+rhs._internal),N>
+    {
+      typedef iMatrix<decltype(lhs._internal[0][0]+rhs._internal),N> ret_t;
+      ret_t ret;
+      add(&ret,&lhs,&rhs);
+      return ret;
+    }
+
+
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    /////////////////////////////////////////// SUB         ///////////////////////////////////////////
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    
+
+// SUB is simple for now; cannot mix types and straightforward template
+// Scalar +/- Scalar
+// Vector +/- Vector
+// Matrix +/- Matrix
+// Matrix /- scalar
+template<class vtype,class ltype,class rtype> inline void sub(iScalar<vtype> * __restrict__ ret,
+                                                              const iScalar<ltype> * __restrict__ lhs,
+                                                              const iScalar<rtype> * __restrict__ rhs)
+{
+    sub(&ret->_internal,&lhs->_internal,&rhs->_internal);
+}
+
+template<class vtype,class ltype,class rtype,int N> inline void sub(iVector<vtype,N> * __restrict__ ret,
+                                                                    const iVector<ltype,N> * __restrict__ lhs,
+                                                                    const iVector<rtype,N> * __restrict__ rhs)
+{
+    for(int c=0;c<N;c++){
+        ret->_internal[c]=lhs->_internal[c]-rhs->_internal[c];
+    }
+    return;
+}
+template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
+                                                                     const iMatrix<ltype,N> * __restrict__ lhs,
+                                                                     const iMatrix<rtype,N> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        sub(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal[c1][c2]);
+    }}
+    return;
+}
+template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
+                                                                     const iScalar<ltype> * __restrict__ lhs,
+                                                                     const iMatrix<rtype,N> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        if ( c1!=c2) {
+            sub(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
+        } else {
+            // Fails -- need unary minus. Catalogue other unops?
+            ret->_internal[c1][c2]=zero;
+            ret->_internal[c1][c2]=ret->_internal[c1][c2]-rhs->_internal[c1][c2];
+
+        }
+    }}
+    return;
+}
+template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
+                                                                     const iMatrix<ltype,N> * __restrict__ lhs,
+                                                                     const iScalar<rtype> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        if ( c1!=c2)
+            sub(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
+        else
+            ret->_internal[c1][c2]=lhs->_internal[c1][c2];
+    }}
+    return;
+}
+
+template<class v> void vprefetch(const iScalar<v> &vv)
+{
+  vprefetch(vv._internal);
+}
+template<class v,int N> void vprefetch(const iVector<v,N> &vv)
+{
+  for(int i=0;i<N;i++){
+    vprefetch(vv._internal[i]);
+  }
+}
+template<class v,int N> void vprefetch(const iMatrix<v,N> &vv)
+{
+  for(int i=0;i<N;i++){
+  for(int j=0;j<N;j++){
+    vprefetch(vv._internal[i][j]);
+  }}
+}
+
+    // - operator for scalar, vector, matrix
+template<class ltype,class rtype> inline auto
+operator - (const iScalar<ltype>& lhs, const iScalar<rtype>& rhs) -> iScalar<decltype(lhs._internal - rhs._internal)>
+{
+    typedef iScalar<decltype(lhs._internal-rhs._internal)> ret_t;
+    ret_t ret;
+    sub(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class ltype,class rtype,int N>
+inline auto operator - (const iVector<ltype,N>& lhs,const iVector<rtype,N>& rhs) ->iVector<decltype(lhs._internal[0]-rhs._internal[0]),N>
+{
+    typedef iVector<decltype(lhs._internal[0]-rhs._internal[0]),N> ret_t;
+    ret_t ret;
+    sub(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class ltype,class rtype,int N>
+inline auto operator - (const iMatrix<ltype,N>& lhs,const iMatrix<rtype,N>& rhs) ->iMatrix<decltype(lhs._internal[0][0]-rhs._internal[0][0]),N>
+{
+    typedef iMatrix<decltype(lhs._internal[0][0]-rhs._internal[0][0]),N> ret_t;
+    ret_t ret;
+    sub(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class ltype,class rtype,int N>
+inline auto operator - (const iScalar<ltype>& lhs,const iMatrix<rtype,N>& rhs)->iMatrix<decltype(lhs._internal-rhs._internal[0][0]),N>
+{
+    typedef iMatrix<decltype(lhs._internal-rhs._internal[0][0]),N> ret_t;
+    ret_t ret;
+    sub(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class ltype,class rtype,int N>
+inline auto operator - (const iMatrix<ltype,N>& lhs,const iScalar<rtype>& rhs)->iMatrix<decltype(lhs._internal[0][0]-rhs._internal),N>
+{
+    typedef iMatrix<decltype(lhs._internal[0][0]-rhs._internal),N> ret_t;
+    ret_t ret;
+    sub(&ret,&lhs,&rhs);
+    return ret;
+}
+
+///////////////////////////////////////////////////////////////////////////////////////////////////
+/////////////////////////////////////////// MAC         ///////////////////////////////////////////
+///////////////////////////////////////////////////////////////////////////////////////////////////
+
+    ///////////////////////////
+    // Legal multiplication table
+    ///////////////////////////
+    // scal x scal = scal
+    // mat x  mat  = mat
+    // mat  x scal = mat
+    // scal x mat  = mat
+    // mat  x vec  = vec
+    // vec  x scal = vec
+    // scal x vec  = vec
+    ///////////////////////////
+template<class rtype,class vtype,class mtype>
+inline  void mac(iScalar<rtype> * __restrict__ ret,const iScalar<vtype> * __restrict__ lhs,const iScalar<mtype> * __restrict__ rhs)
+{
+    mac(&ret->_internal,&lhs->_internal,&rhs->_internal);
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+    for(int c3=0;c3<N;c3++){
+        mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c3],&rhs->_internal[c3][c2]);
+    }}}
+    return;
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs){
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
+    }}
+    return;
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iScalar<ltype> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        mac(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
+    }}
+    return;
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mac(iVector<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iVector<rtype,N> * __restrict__ rhs)
+{
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        mac(&ret->_internal[c1],&lhs->_internal[c1][c2],&rhs->_internal[c2]);
+    }}
+    return;
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mac(iVector<rrtype,N> * __restrict__ ret,const iScalar<ltype> * __restrict__ lhs,const iVector<rtype,N> * __restrict__ rhs)
+{
+    for(int c1=0;c1<N;c1++){
+        mac(&ret->_internal[c1],&lhs->_internal,&rhs->_internal[c1]);
+    }
+    return;
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mac(iVector<rrtype,N> * __restrict__ ret,const iVector<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs)
+{
+    for(int c1=0;c1<N;c1++){
+        mac(&ret->_internal[c1],&lhs->_internal[c1],&rhs->_internal);
+    }
+    return;
+}
+
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    /////////////////////////////////////////// MUL         ///////////////////////////////////////////
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+
+    
+template<class rtype,class vtype,class mtype>
+inline void mult(iScalar<rtype> * __restrict__ ret,const iScalar<mtype> * __restrict__ lhs,const iScalar<vtype> * __restrict__ rhs){
+    mult(&ret->_internal,&lhs->_internal,&rhs->_internal);
+}
+
+template<class rrtype,class ltype,class rtype,int N>
+inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        mult(&ret->_internal[c1][c2],&lhs->_internal[c1][0],&rhs->_internal[0][c2]);
+        for(int c3=1;c3<N;c3++){
+            mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c3],&rhs->_internal[c3][c2]);
+        }
+    }}
+    return;
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        mult(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
+    }}
+    return;
+}
+
+template<class rrtype,class ltype,class rtype, int N>
+inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iScalar<ltype>   * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        mult(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
+    }}
+    return;
+}
+// Matrix left multiplies vector
+template<class rtype,class vtype,class mtype,int N>
+inline void mult(iVector<rtype,N> * __restrict__ ret,const iMatrix<mtype,N> * __restrict__ lhs,const iVector<vtype,N> * __restrict__ rhs)
+{
+    for(int c1=0;c1<N;c1++){
+        mult(&ret->_internal[c1],&lhs->_internal[c1][0],&rhs->_internal[0]);
+        for(int c2=1;c2<N;c2++){
+            mac(&ret->_internal[c1],&lhs->_internal[c1][c2],&rhs->_internal[c2]);
+        }
+    }
+    return;
+}
+template<class rtype,class vtype,class mtype,int N>
+inline void mult(iVector<rtype,N> * __restrict__ ret,
+                 const iScalar<mtype>   * __restrict__ lhs,
+                 const iVector<vtype,N> * __restrict__ rhs){
+    for(int c1=0;c1<N;c1++){
+        mult(&ret->_internal[c1],&lhs->_internal,&rhs->_internal[c1]);
+    }
+}
+template<class rtype,class vtype,class mtype,int N>
+inline void mult(iVector<rtype,N> * __restrict__ ret,
+                 const iVector<vtype,N> * __restrict__ rhs,
+                 const iScalar<mtype> * __restrict__ lhs){
+    mult(ret,lhs,rhs);
+}
+    
+
+
+template<class rtype,class vtype,class mtype,int N> inline
+iVector<rtype,N> operator * (const iMatrix<mtype,N>& lhs,const iVector<vtype,N>& rhs)
+{
+    iVector<rtype,N> ret;
+    mult(&ret,&lhs,&rhs);
+    return ret;
+}
+
+template<class rtype,class vtype,class mtype,int N> inline
+iVector<rtype,N> operator * (const iScalar<mtype>& lhs,const iVector<vtype,N>& rhs)
+{
+    iVector<rtype,N> ret;
+    mult(&ret,&lhs,&rhs);
+    return ret;
+}
+
+template<class rtype,class vtype,class mtype,int N> inline
+iVector<rtype,N> operator * (const iVector<mtype,N>& lhs,const iScalar<vtype>& rhs)
+{
+    iVector<rtype,N> ret;
+    mult(&ret,&lhs,&rhs);
+    return ret;
+}
+    
+    //////////////////////////////////////////////////////////////////
+    // Glue operators to mult routines. Must resolve return type cleverly from typeof(internal)
+    // since nesting matrix<scalar> x matrix<matrix>-> matrix<matrix>
+    // while         matrix<scalar> x matrix<scalar>-> matrix<scalar>
+    // so return type depends on argument types in nasty way.
+    //////////////////////////////////////////////////////////////////
+    // scal x scal = scal
+    // mat x  mat  = mat
+    // mat  x scal = mat
+    // scal x mat  = mat
+    // mat  x vec  = vec
+    // vec  x scal = vec
+    // scal x vec  = vec
+    //
+    // We can special case scalar_type ??
+template<class l,class r>
+inline auto operator * (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(lhs._internal * rhs._internal)>
+{
+    typedef iScalar<decltype(lhs._internal*rhs._internal)> ret_t;
+    ret_t ret;
+    mult(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class l,class r,int N> inline
+auto operator * (const iMatrix<l,N>& lhs,const iMatrix<r,N>& rhs) -> iMatrix<decltype(lhs._internal[0][0]*rhs._internal[0][0]),N>
+{
+    typedef decltype(lhs._internal[0][0]*rhs._internal[0][0]) ret_t;
+    iMatrix<ret_t,N> ret;
+    mult(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class l,class r, int N> inline
+auto operator * (const iMatrix<r,N>& lhs,const iScalar<l>& rhs) -> iMatrix<decltype(lhs._internal[0][0]*rhs._internal),N>
+{
+    typedef decltype(lhs._internal[0][0]*rhs._internal) ret_t;
+        
+    iMatrix<ret_t,N> ret;
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        mult(&ret._internal[c1][c2],&lhs._internal[c1][c2],&rhs._internal);
+    }}
+    return ret;
+}
+template<class l,class r,int N> inline
+auto operator * (const iScalar<l>& lhs,const iMatrix<r,N>& rhs) -> iMatrix<decltype(lhs._internal*rhs._internal[0][0]),N>
+{
+    typedef decltype(lhs._internal*rhs._internal[0][0]) ret_t;
+    iMatrix<ret_t,N> ret;
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        mult(&ret._internal[c1][c2],&lhs._internal,&rhs._internal[c1][c2]);
+    }}
+    return ret;
+}
+template<class l,class r,int N> inline
+auto operator * (const iMatrix<l,N>& lhs,const iVector<r,N>& rhs) -> iVector<decltype(lhs._internal[0][0]*rhs._internal[0]),N>
+{
+    typedef decltype(lhs._internal[0][0]*rhs._internal[0]) ret_t;
+    iVector<ret_t,N> ret;
+    for(int c1=0;c1<N;c1++){
+        mult(&ret._internal[c1],&lhs._internal[c1][0],&rhs._internal[0]);
+        for(int c2=1;c2<N;c2++){
+            mac(&ret._internal[c1],&lhs._internal[c1][c2],&rhs._internal[c2]);
+        }
+    }
+    return ret;
+}
+template<class l,class r,int N> inline
+auto operator * (const iScalar<l>& lhs,const iVector<r,N>& rhs) -> iVector<decltype(lhs._internal*rhs._internal[0]),N>
+{
+    typedef decltype(lhs._internal*rhs._internal[0]) ret_t;
+    iVector<ret_t,N> ret;
+    for(int c1=0;c1<N;c1++){
+        mult(&ret._internal[c1],&lhs._internal,&rhs._internal[c1]);
+    }
+    return ret;
+}
+template<class l,class r,int N> inline
+auto operator * (const iVector<l,N>& lhs,const iScalar<r>& rhs) -> iVector<decltype(lhs._internal[0]*rhs._internal),N>
+{
+    typedef decltype(lhs._internal[0]*rhs._internal) ret_t;
+    iVector<ret_t,N> ret;
+    for(int c1=0;c1<N;c1++){
+        mult(&ret._internal[c1],&lhs._internal[c1],&rhs._internal);
+    }
+    return ret;
+}
+
+
+//////////////////////////////////////////////////////////////////////////////////////////
+// Must support native C++ types Integer, Complex, Real
+//////////////////////////////////////////////////////////////////////////////////////////
+
+// multiplication by fundamental scalar type
+template<class l,int N> inline iScalar<l> operator * (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
+{
+  typename iScalar<l>::tensor_reduced srhs(rhs);
+  return lhs*srhs;
+}
+template<class l,int N> inline iScalar<l> operator * (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
+{
+  typename iVector<l,N>::tensor_reduced srhs(rhs);
+  return lhs*srhs;
+}
+template<class l,int N> inline iVector<l,N> operator * (const typename iScalar<l>::scalar_type lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type &rhs) 
+{
+  typename iMatrix<l,N>::tensor_reduced srhs(rhs);
+  return lhs*srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator * (const typename iScalar<l>::scalar_type & lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
+
+////////////////////////////////////////////////////////////////////
+// Double support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l> inline iScalar<l> operator * (double lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l,int N> inline iVector<l,N> operator * (double lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator * (double lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
+
+////////////////////////////////////////////////////////////////////
+// Complex support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,ComplexD rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l> inline iScalar<l> operator * (ComplexD lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,ComplexD rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l,int N> inline iVector<l,N> operator * (ComplexD lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,ComplexD rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator * (ComplexD lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
+
+////////////////////////////////////////////////////////////////////
+// Integer support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l> inline iScalar<l> operator * (Integer lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l,int N> inline iVector<l,N> operator * (Integer lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator * (Integer lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
+
+
+
+///////////////////////////////////////////////////////////////////////////////////////////////
+// addition by fundamental scalar type applies to matrix(down diag) and scalar
+///////////////////////////////////////////////////////////////////////////////////////////////
+template<class l,int N> inline iScalar<l> operator + (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
+{
+  typename iScalar<l>::tensor_reduced srhs(rhs);
+  return lhs+srhs;
+}
+template<class l,int N> inline iScalar<l> operator + (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
+{
+  typename iMatrix<l,N>::tensor_reduced srhs(rhs);
+  return lhs+srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator + (const typename iScalar<l>::scalar_type lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
+
+////////////////////////////////////////////////////////////////////
+// Double support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator + (const iScalar<l>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs+srhs;
+}
+template<class l> inline iScalar<l> operator + (double lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs+srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator + (double lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
+
+
+// Integer support cast to scalar type through constructor
+
+
+template<class l> inline iScalar<l> operator + (const iScalar<l>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs+srhs;
+}
+
+template<class l> inline iScalar<l> operator + (Integer lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs+srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator + (Integer lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
+
+
+///////////////////////////////////////////////////////////////////////////////////////////////
+// subtraction of fundamental scalar type applies to matrix(down diag) and scalar
+///////////////////////////////////////////////////////////////////////////////////////////////
+template<class l,int N> inline iScalar<l> operator - (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
+{
+  typename iScalar<l>::tensor_reduced srhs(rhs);
+  return lhs-srhs;
+}
+template<class l,int N> inline iScalar<l> operator - (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) 
+{
+  typename iScalar<l>::tensor_reduced slhs(lhs);
+  return slhs-rhs;
+}
+
+template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
+{
+  typename iScalar<l>::tensor_reduced srhs(rhs);
+  return lhs-srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator - (const typename iScalar<l>::scalar_type lhs,const iMatrix<l,N>& rhs) 
+{
+  typename iScalar<l>::tensor_reduced slhs(lhs);
+  return slhs-rhs;
+}
+
+////////////////////////////////////////////////////////////////////
+// Double support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator - (const iScalar<l>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs-srhs;
+}
+template<class l> inline iScalar<l> operator - (double lhs,const iScalar<l>& rhs) 
+{
+  typename iScalar<l>::scalar_type t(lhs);
+  typename iScalar<l>::tensor_reduced slhs(t);
+  return slhs-rhs;
+}
+
+template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs-srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator - (double lhs,const iMatrix<l,N>& rhs) 
+{
+  typename iScalar<l>::scalar_type t(lhs);
+  typename iScalar<l>::tensor_reduced slhs(t);
+  return slhs-rhs;
+}
+
+////////////////////////////////////////////////////////////////////
+// Integer support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator - (const iScalar<l>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs-srhs;
+}
+template<class l> inline iScalar<l> operator - (Integer lhs,const iScalar<l>& rhs) 
+{
+  typename iScalar<l>::scalar_type t(lhs);
+  typename iScalar<l>::tensor_reduced slhs(t);
+  return slhs-rhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs-srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator - (Integer lhs,const iMatrix<l,N>& rhs) 
+{
+  typename iScalar<l>::scalar_type t(lhs);
+  typename iScalar<l>::tensor_reduced slhs(t);
+  return slhs-rhs;
+}
+
+}
+#endif
+
diff --git a/lib/Grid_math_tensors.h b/lib/Grid_math_tensors.h
new file mode 100644
index 00000000..98b9d22e
--- /dev/null
+++ b/lib/Grid_math_tensors.h
@@ -0,0 +1,224 @@
+#ifndef GRID_MATH_TENSORS_H
+#define GRID_MATH_TENSORS_H
+
+namespace Grid {
+
+///////////////////////////////////////////////////
+// Scalar, Vector, Matrix objects.
+// These can be composed to form tensor products of internal indices.
+///////////////////////////////////////////////////
+
+template<class vtype> class iScalar
+{
+public:
+  vtype _internal;
+
+  typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
+  typedef typename GridTypeMapper<vtype>::vector_type vector_type;
+  typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
+  typedef iScalar<tensor_reduced_v> tensor_reduced;
+
+  enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
+
+  // Scalar no action
+  //  template<int Level> using tensor_reduce_level = typename iScalar<GridTypeMapper<vtype>::tensor_reduce_level<Level> >;
+
+    iScalar(){};
+    
+    iScalar(scalar_type s) : _internal(s) {};// recurse down and hit the constructor for vector_type
+
+    iScalar(Zero &z){ *this = zero; };
+
+    iScalar<vtype> & operator= (const Zero &hero){
+        zeroit(*this);
+        return *this;
+    }
+    friend void zeroit(iScalar<vtype> &that){
+        zeroit(that._internal);
+    }
+    friend void permute(iScalar<vtype> &out,const iScalar<vtype> &in,int permutetype){
+      permute(out._internal,in._internal,permutetype);
+    }
+    friend void extract(const iScalar<vtype> &in,std::vector<scalar_type *> &out){
+      extract(in._internal,out); // extract advances the pointers in out
+    }
+    friend void merge(iScalar<vtype> &in,std::vector<scalar_type *> &out){
+      merge(in._internal,out); // extract advances the pointers in out
+    }
+
+    // Unary negation
+    friend inline iScalar<vtype> operator -(const iScalar<vtype> &r) {
+        iScalar<vtype> ret;
+        ret._internal= -r._internal;
+        return ret;
+    }
+    // *=,+=,-= operators inherit from corresponding "*,-,+" behaviour
+    inline iScalar<vtype> &operator *=(const iScalar<vtype> &r) {
+        *this = (*this)*r;
+        return *this;
+    }
+    inline iScalar<vtype> &operator -=(const iScalar<vtype> &r) {
+        *this = (*this)-r;
+        return *this;
+    }
+    inline iScalar<vtype> &operator +=(const iScalar<vtype> &r) {
+        *this = (*this)+r;
+        return *this;
+    }
+    
+    inline vtype & operator ()(void) {
+      return _internal;
+    }
+
+    operator ComplexD () const { return(TensorRemove(_internal)); };
+    operator RealD () const { return(real(TensorRemove(_internal))); }
+
+};
+///////////////////////////////////////////////////////////
+// Allows to turn scalar<scalar<scalar<double>>>> back to double.
+///////////////////////////////////////////////////////////
+template<class T> inline typename std::enable_if<isGridTensor<T>::notvalue, T>::type TensorRemove(T arg) { return arg;}
+template<class vtype> inline auto TensorRemove(iScalar<vtype> arg) -> decltype(TensorRemove(arg._internal))
+{
+  return TensorRemove(arg._internal);
+}
+    
+template<class vtype,int N> class iVector
+{
+public:
+  vtype _internal[N];
+
+  typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
+  typedef typename GridTypeMapper<vtype>::vector_type vector_type;
+  typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
+
+  enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
+  typedef iScalar<tensor_reduced_v> tensor_reduced;
+
+    iVector(Zero &z){ *this = zero; };
+    iVector() {};// Empty constructure
+
+    iVector<vtype,N> & operator= (Zero &hero){
+        zeroit(*this);
+        return *this;
+    }
+    friend void zeroit(iVector<vtype,N> &that){
+        for(int i=0;i<N;i++){
+            zeroit(that._internal[i]);
+        }
+    }
+    friend void permute(iVector<vtype,N> &out,const iVector<vtype,N> &in,int permutetype){
+      for(int i=0;i<N;i++){
+	permute(out._internal[i],in._internal[i],permutetype);
+      }
+    }
+    friend void extract(const iVector<vtype,N> &in,std::vector<scalar_type *> &out){
+      for(int i=0;i<N;i++){
+	extract(in._internal[i],out);// extract advances pointers in out
+      }
+    }
+    friend void merge(iVector<vtype,N> &in,std::vector<scalar_type *> &out){
+      for(int i=0;i<N;i++){
+	merge(in._internal[i],out);// extract advances pointers in out
+      }
+    }
+    // Unary negation
+    friend inline iVector<vtype,N> operator -(const iVector<vtype,N> &r) {
+        iVector<vtype,N> ret;
+        for(int i=0;i<N;i++) ret._internal[i]= -r._internal[i];
+        return ret;
+    }
+    // *=,+=,-= operators inherit from corresponding "*,-,+" behaviour
+    inline iVector<vtype,N> &operator *=(const iScalar<vtype> &r) {
+        *this = (*this)*r;
+        return *this;
+    }
+    inline iVector<vtype,N> &operator -=(const iVector<vtype,N> &r) {
+        *this = (*this)-r;
+        return *this;
+    }
+    inline iVector<vtype,N> &operator +=(const iVector<vtype,N> &r) {
+        *this = (*this)+r;
+        return *this;
+    }
+    inline vtype & operator ()(int i) {
+      return _internal[i];
+    }
+};
+    
+template<class vtype,int N> class iMatrix
+{
+public:
+  vtype _internal[N][N];
+
+  typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
+  typedef typename GridTypeMapper<vtype>::vector_type vector_type;
+
+  typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
+
+  enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
+  typedef iScalar<tensor_reduced_v> tensor_reduced;
+
+  iMatrix(Zero &z){ *this = zero; };
+  iMatrix() {};
+  iMatrix<vtype,N> & operator= (Zero &hero){
+    zeroit(*this);
+    return *this;
+  }
+  friend void zeroit(iMatrix<vtype,N> &that){
+    for(int i=0;i<N;i++){
+      for(int j=0;j<N;j++){
+	zeroit(that._internal[i][j]);
+    }}
+  }
+  friend void permute(iMatrix<vtype,N> &out,const iMatrix<vtype,N> &in,int permutetype){
+    for(int i=0;i<N;i++){
+      for(int j=0;j<N;j++){
+	permute(out._internal[i][j],in._internal[i][j],permutetype);
+    }}
+  }
+  friend void extract(const iMatrix<vtype,N> &in,std::vector<scalar_type *> &out){
+    for(int i=0;i<N;i++){
+      for(int j=0;j<N;j++){
+	extract(in._internal[i][j],out);// extract advances pointers in out
+    }}
+  }
+  friend void merge(iMatrix<vtype,N> &in,std::vector<scalar_type *> &out){
+    for(int i=0;i<N;i++){
+      for(int j=0;j<N;j++){
+	merge(in._internal[i][j],out);// extract advances pointers in out
+    }}
+  }
+  // Unary negation
+  friend inline iMatrix<vtype,N> operator -(const iMatrix<vtype,N> &r) {
+    iMatrix<vtype,N> ret;
+    for(int i=0;i<N;i++){
+      for(int j=0;j<N;j++){
+	ret._internal[i][j]= -r._internal[i][j];
+    }}
+    return ret;
+  }
+  // *=,+=,-= operators inherit from corresponding "*,-,+" behaviour
+  template<class T>
+  inline iMatrix<vtype,N> &operator *=(const T &r) {
+    *this = (*this)*r;
+    return *this;
+  }
+  template<class T>
+  inline iMatrix<vtype,N> &operator -=(const T &r) {
+    *this = (*this)-r;
+    return *this;
+  }
+  template<class T>
+  inline iMatrix<vtype,N> &operator +=(const T &r) {
+    *this = (*this)+r;
+    return *this;
+  }
+  inline vtype & operator ()(int i,int j) {
+    return _internal[i][j];
+  }
+
+};
+
+}
+#endif
diff --git a/lib/Grid_math_traits.h b/lib/Grid_math_traits.h
new file mode 100644
index 00000000..8ecc2712
--- /dev/null
+++ b/lib/Grid_math_traits.h
@@ -0,0 +1,165 @@
+#ifndef GRID_MATH_TRAITS_H
+#define GRID_MATH_TRAITS_H
+
+#include <type_traits>
+
+namespace Grid {
+
+//////////////////////////////////////////////////////////////////////////////////
+// Want to recurse: GridTypeMapper<Matrix<vComplexD> >::scalar_type == ComplexD.
+// Use of a helper class like this allows us to template specialise and "dress"
+// other classes such as RealD == double, ComplexD == std::complex<double> with these
+// traits.
+//
+// It is possible that we could do this more elegantly if I introduced a 
+// queryable trait in iScalar, iMatrix and iVector and used the query on vtype in 
+// place of the type mapper?
+//
+// Not sure how to do this, but probably could be done with a research effort
+// to study C++11's type_traits.h file. (std::enable_if<isGridTensorType<vtype> >)
+//
+//////////////////////////////////////////////////////////////////////////////////
+  
+  template <class T> class GridTypeMapper {
+  public:
+    typedef typename T::scalar_type scalar_type;
+    typedef typename T::vector_type vector_type;
+    typedef typename T::tensor_reduced tensor_reduced;
+    enum { TensorLevel = T::TensorLevel };
+  };
+
+//////////////////////////////////////////////////////////////////////////////////
+// Recursion stops with these template specialisations
+//////////////////////////////////////////////////////////////////////////////////
+  template<> class GridTypeMapper<RealF> {
+  public:
+    typedef RealF scalar_type;
+    typedef RealF vector_type;
+    typedef RealF tensor_reduced ;
+    enum { TensorLevel = 0 };
+  };
+  template<> class GridTypeMapper<RealD> {
+  public:
+    typedef RealD scalar_type;
+    typedef RealD vector_type;
+    typedef RealD tensor_reduced;
+    enum { TensorLevel = 0 };
+  };
+  template<> class GridTypeMapper<ComplexF> {
+  public:
+    typedef ComplexF scalar_type;
+    typedef ComplexF vector_type;
+    typedef ComplexF tensor_reduced;
+    enum { TensorLevel = 0 };
+  };
+  template<> class GridTypeMapper<ComplexD> {
+  public:
+    typedef ComplexD scalar_type;
+    typedef ComplexD vector_type;
+    typedef ComplexD tensor_reduced;
+    enum { TensorLevel = 0 };
+  };
+
+  template<> class GridTypeMapper<vRealF> {
+  public:
+    typedef RealF  scalar_type;
+    typedef vRealF vector_type;
+    typedef vRealF tensor_reduced;
+    enum { TensorLevel = 0 };
+  };
+  template<> class GridTypeMapper<vRealD> {
+  public:
+    typedef RealD  scalar_type;
+    typedef vRealD vector_type;
+    typedef vRealD tensor_reduced;
+    enum { TensorLevel = 0 };
+  };
+  template<> class GridTypeMapper<vComplexF> {
+  public:
+    typedef ComplexF  scalar_type;
+    typedef vComplexF vector_type;
+    typedef vComplexF tensor_reduced;
+    enum { TensorLevel = 0 };
+  };
+  template<> class GridTypeMapper<vComplexD> {
+  public:
+    typedef ComplexD  scalar_type;
+    typedef vComplexD vector_type;
+    typedef vComplexD tensor_reduced;
+    enum { TensorLevel = 0 };
+  };
+  template<> class GridTypeMapper<vInteger> {
+  public:
+    typedef Integer  scalar_type;
+    typedef vInteger vector_type;
+    typedef vInteger tensor_reduced;
+    enum { TensorLevel = 0 };
+  };
+
+  // First some of my own traits
+  template<typename T> struct isGridTensor {
+    static const bool value = true;
+    static const bool notvalue = false;
+  };
+
+  template<> struct isGridTensor<RealD > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<RealF > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<ComplexD > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<ComplexF > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<Integer > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<vRealD > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<vRealF > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<vComplexD > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<vComplexF > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<vInteger > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+
+  // Match the index
+  template<typename T,int Level> struct matchGridTensorIndex {
+    static const bool value = (Level==T::TensorLevel);
+    static const bool notvalue = (Level!=T::TensorLevel);
+  };
+  // What is the vtype
+  template<typename T> struct isComplex {
+    static const bool value = false;
+  };
+  template<> struct isComplex<ComplexF> {
+    static const bool value = true;
+  };
+  template<> struct isComplex<ComplexD> {
+    static const bool value = true;
+  };
+
+
+}
+
+#endif
diff --git a/lib/Grid_vComplexD.h b/lib/Grid_vComplexD.h
index ffcff3a9..313eef4a 100644
--- a/lib/Grid_vComplexD.h
+++ b/lib/Grid_vComplexD.h
@@ -1,7 +1,5 @@
-#ifndef VCOMPLEXD_H
-#define  VCOMPLEXD_H
-#include "Grid.h"
-#include "Grid_vComplexF.h"
+#ifndef GRID_VCOMPLEXD_H
+#define  GRID_VCOMPLEXD_H
 
 namespace Grid {
     class vComplexD {
diff --git a/lib/Grid_vComplexF.h b/lib/Grid_vComplexF.h
index 42fe5163..9fac656c 100644
--- a/lib/Grid_vComplexF.h
+++ b/lib/Grid_vComplexF.h
@@ -1,6 +1,5 @@
-#ifndef VCOMPLEXF
-#define VCOMPLEXF
-#include "Grid.h"
+#ifndef GRID_VCOMPLEXF
+#define GRID_VCOMPLEXF
 
 namespace Grid {
 
diff --git a/lib/Grid_vInteger.h b/lib/Grid_vInteger.h
index 672d4938..5dc77444 100644
--- a/lib/Grid_vInteger.h
+++ b/lib/Grid_vInteger.h
@@ -1,7 +1,5 @@
-#ifndef VINTEGER_H
-#define VINTEGER_H
-
-#include "Grid.h"
+#ifndef GRID_VINTEGER_H
+#define GRID_VINTEGER_H
 
 namespace Grid {
 
diff --git a/lib/Grid_vRealD.h b/lib/Grid_vRealD.h
index 0eb6d9c1..ab848916 100644
--- a/lib/Grid_vRealD.h
+++ b/lib/Grid_vRealD.h
@@ -1,7 +1,5 @@
-#ifndef VREALD_H
-#define VREALD_H
-
-#include "Grid.h"
+#ifndef GRID_VREALD_H
+#define GRID_VREALD_H
 
 namespace Grid {
     class vRealD  {
diff --git a/lib/Grid_vRealF.h b/lib/Grid_vRealF.h
index 7831a6c7..7d71c528 100644
--- a/lib/Grid_vRealF.h
+++ b/lib/Grid_vRealF.h
@@ -1,7 +1,6 @@
-#ifndef VREALF_H
-#define VREALF_H
+#ifndef GRID_VREALF_H
+#define GRID_VREALF_H
 
-#include "Grid.h"
 
 namespace Grid {
     class vRealF  {
diff --git a/lib/Makefile.am b/lib/Makefile.am
index 8cb9a0be..93625c68 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -23,22 +23,33 @@ libGrid_a_SOURCES = Grid_init.cc $(extra_sources)
 #
 include_HEADERS = Grid_config.h\
 	Grid.h\
-	Grid_simd.h\
-	Grid_vComplexD.h\
-	Grid_vComplexF.h\
-	Grid_vRealD.h\
-	Grid_vRealF.h\
-	Grid_Cartesian.h\
-	Grid_Lattice.h\
-	Grid_Communicator.h\
 	Grid_QCD.h\
 	Grid_aligned_allocator.h\
+	Grid_cartesian.h\
+	Grid_cartesian_base.h\
+	Grid_cartesian_full.h\
+	Grid_cartesian_red_black.h\
+	Grid_communicator.h\
+	Grid_comparison.h\
+	Grid_config.h\
 	Grid_cshift.h\
 	Grid_cshift_common.h\
 	Grid_cshift_mpi.h\
 	Grid_cshift_none.h\
+	Grid_lattice.h\
+	Grid_math.h\
+	Grid_math_arith.h\
+	Grid_math_tensors.h\
+	Grid_math_traits.h\
+	Grid_predicated.h\
+	Grid_simd.h\
 	Grid_stencil.h\
-	Grid_math_types.h
+	Grid_summation.h\
+	Grid_vComplexD.h\
+	Grid_vComplexF.h\
+	Grid_vInteger.h\
+	Grid_vRealD.h\
+	Grid_vRealF.h
 
 
 

From 3931ad65c8a0ad6f137d5b50fb274cf62e0ba4ca Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 18:37:22 +0100
Subject: [PATCH 071/429] Reorg

---
 lib/Grid_math_types.h | 1662 -----------------------------------------
 1 file changed, 1662 deletions(-)
 delete mode 100644 lib/Grid_math_types.h

diff --git a/lib/Grid_math_types.h b/lib/Grid_math_types.h
deleted file mode 100644
index 1042a95b..00000000
--- a/lib/Grid_math_types.h
+++ /dev/null
@@ -1,1662 +0,0 @@
-#ifndef GRID_MATH_TYPES_H
-#define GRID_MATH_TYPES_H
-
-#include <type_traits>
-
-#include <Grid_math_types_mapper.h>
-
-//
-// Indexing; want to be able to dereference and 
-// obtain either an lvalue or an rvalue.
-//
-namespace Grid {
-
-  // First some of my own traits
-  template<typename T> struct isGridTensor {
-    static const bool value = true;
-    static const bool notvalue = false;
-  };
-
-  template<> struct isGridTensor<RealD > {
-    static const bool value = false;
-    static const bool notvalue = true;
-  };
-  template<> struct isGridTensor<RealF > {
-    static const bool value = false;
-    static const bool notvalue = true;
-  };
-  template<> struct isGridTensor<ComplexD > {
-    static const bool value = false;
-    static const bool notvalue = true;
-  };
-  template<> struct isGridTensor<ComplexF > {
-    static const bool value = false;
-    static const bool notvalue = true;
-  };
-  template<> struct isGridTensor<Integer > {
-    static const bool value = false;
-    static const bool notvalue = true;
-  };
-  template<> struct isGridTensor<vRealD > {
-    static const bool value = false;
-    static const bool notvalue = true;
-  };
-  template<> struct isGridTensor<vRealF > {
-    static const bool value = false;
-    static const bool notvalue = true;
-  };
-  template<> struct isGridTensor<vComplexD > {
-    static const bool value = false;
-    static const bool notvalue = true;
-  };
-  template<> struct isGridTensor<vComplexF > {
-    static const bool value = false;
-    static const bool notvalue = true;
-  };
-  template<> struct isGridTensor<vInteger > {
-    static const bool value = false;
-    static const bool notvalue = true;
-  };
-
-  // Match the index
-  template<typename T,int Level> struct matchGridTensorIndex {
-    static const bool value = (Level==T::TensorLevel);
-    static const bool notvalue = (Level!=T::TensorLevel);
-  };
-  // What is the vtype
-  template<typename T> struct isComplex {
-    static const bool value = false;
-  };
-  template<> struct isComplex<ComplexF> {
-    static const bool value = true;
-  };
-  template<> struct isComplex<ComplexD> {
-    static const bool value = true;
-  };
-
-
-///////////////////////////////////////////////////
-// Scalar, Vector, Matrix objects.
-// These can be composed to form tensor products of internal indices.
-///////////////////////////////////////////////////
-
-
-
-template<class vtype> class iScalar
-{
-public:
-  vtype _internal;
-
-  typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
-  typedef typename GridTypeMapper<vtype>::vector_type vector_type;
-  typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
-  typedef iScalar<tensor_reduced_v> tensor_reduced;
-
-  enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
-
-  // Scalar no action
-  //  template<int Level> using tensor_reduce_level = typename iScalar<GridTypeMapper<vtype>::tensor_reduce_level<Level> >;
-
-    iScalar(){};
-    
-    iScalar(scalar_type s) : _internal(s) {};// recurse down and hit the constructor for vector_type
-
-    iScalar(Zero &z){ *this = zero; };
-
-    iScalar<vtype> & operator= (const Zero &hero){
-        zeroit(*this);
-        return *this;
-    }
-    friend void zeroit(iScalar<vtype> &that){
-        zeroit(that._internal);
-    }
-    friend void permute(iScalar<vtype> &out,const iScalar<vtype> &in,int permutetype){
-      permute(out._internal,in._internal,permutetype);
-    }
-    friend void extract(const iScalar<vtype> &in,std::vector<scalar_type *> &out){
-      extract(in._internal,out); // extract advances the pointers in out
-    }
-    friend void merge(iScalar<vtype> &in,std::vector<scalar_type *> &out){
-      merge(in._internal,out); // extract advances the pointers in out
-    }
-
-    // Unary negation
-    friend inline iScalar<vtype> operator -(const iScalar<vtype> &r) {
-        iScalar<vtype> ret;
-        ret._internal= -r._internal;
-        return ret;
-    }
-    // *=,+=,-= operators inherit from corresponding "*,-,+" behaviour
-    inline iScalar<vtype> &operator *=(const iScalar<vtype> &r) {
-        *this = (*this)*r;
-        return *this;
-    }
-    inline iScalar<vtype> &operator -=(const iScalar<vtype> &r) {
-        *this = (*this)-r;
-        return *this;
-    }
-    inline iScalar<vtype> &operator +=(const iScalar<vtype> &r) {
-        *this = (*this)+r;
-        return *this;
-    }
-    
-    inline vtype & operator ()(void) {
-      return _internal;
-    }
-
-    operator ComplexD () const { return(TensorRemove(_internal)); };
-    operator RealD () const { return(real(TensorRemove(_internal))); }
-
-};
-///////////////////////////////////////////////////////////
-// Allows to turn scalar<scalar<scalar<double>>>> back to double.
-///////////////////////////////////////////////////////////
-template<class T> inline typename std::enable_if<isGridTensor<T>::notvalue, T>::type TensorRemove(T arg) { return arg;}
-template<class vtype> inline auto TensorRemove(iScalar<vtype> arg) -> decltype(TensorRemove(arg._internal))
-{
-  return TensorRemove(arg._internal);
-}
-    
-template<class vtype,int N> class iVector
-{
-public:
-  vtype _internal[N];
-
-  typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
-  typedef typename GridTypeMapper<vtype>::vector_type vector_type;
-  typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
-
-  enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
-  typedef iScalar<tensor_reduced_v> tensor_reduced;
-
-    iVector(Zero &z){ *this = zero; };
-    iVector() {};// Empty constructure
-
-    iVector<vtype,N> & operator= (Zero &hero){
-        zeroit(*this);
-        return *this;
-    }
-    friend void zeroit(iVector<vtype,N> &that){
-        for(int i=0;i<N;i++){
-            zeroit(that._internal[i]);
-        }
-    }
-    friend void permute(iVector<vtype,N> &out,const iVector<vtype,N> &in,int permutetype){
-      for(int i=0;i<N;i++){
-	permute(out._internal[i],in._internal[i],permutetype);
-      }
-    }
-    friend void extract(const iVector<vtype,N> &in,std::vector<scalar_type *> &out){
-      for(int i=0;i<N;i++){
-	extract(in._internal[i],out);// extract advances pointers in out
-      }
-    }
-    friend void merge(iVector<vtype,N> &in,std::vector<scalar_type *> &out){
-      for(int i=0;i<N;i++){
-	merge(in._internal[i],out);// extract advances pointers in out
-      }
-    }
-    // Unary negation
-    friend inline iVector<vtype,N> operator -(const iVector<vtype,N> &r) {
-        iVector<vtype,N> ret;
-        for(int i=0;i<N;i++) ret._internal[i]= -r._internal[i];
-        return ret;
-    }
-    // *=,+=,-= operators inherit from corresponding "*,-,+" behaviour
-    inline iVector<vtype,N> &operator *=(const iScalar<vtype> &r) {
-        *this = (*this)*r;
-        return *this;
-    }
-    inline iVector<vtype,N> &operator -=(const iVector<vtype,N> &r) {
-        *this = (*this)-r;
-        return *this;
-    }
-    inline iVector<vtype,N> &operator +=(const iVector<vtype,N> &r) {
-        *this = (*this)+r;
-        return *this;
-    }
-    inline vtype & operator ()(int i) {
-      return _internal[i];
-    }
-};
-    
-template<class vtype,int N> class iMatrix
-{
-public:
-  vtype _internal[N][N];
-
-  typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
-  typedef typename GridTypeMapper<vtype>::vector_type vector_type;
-
-  typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
-
-  enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
-  typedef iScalar<tensor_reduced_v> tensor_reduced;
-
-  iMatrix(Zero &z){ *this = zero; };
-  iMatrix() {};
-  iMatrix<vtype,N> & operator= (Zero &hero){
-    zeroit(*this);
-    return *this;
-  }
-  friend void zeroit(iMatrix<vtype,N> &that){
-    for(int i=0;i<N;i++){
-      for(int j=0;j<N;j++){
-	zeroit(that._internal[i][j]);
-    }}
-  }
-  friend void permute(iMatrix<vtype,N> &out,const iMatrix<vtype,N> &in,int permutetype){
-    for(int i=0;i<N;i++){
-      for(int j=0;j<N;j++){
-	permute(out._internal[i][j],in._internal[i][j],permutetype);
-    }}
-  }
-  friend void extract(const iMatrix<vtype,N> &in,std::vector<scalar_type *> &out){
-    for(int i=0;i<N;i++){
-      for(int j=0;j<N;j++){
-	extract(in._internal[i][j],out);// extract advances pointers in out
-    }}
-  }
-  friend void merge(iMatrix<vtype,N> &in,std::vector<scalar_type *> &out){
-    for(int i=0;i<N;i++){
-      for(int j=0;j<N;j++){
-	merge(in._internal[i][j],out);// extract advances pointers in out
-    }}
-  }
-  // Unary negation
-  friend inline iMatrix<vtype,N> operator -(const iMatrix<vtype,N> &r) {
-    iMatrix<vtype,N> ret;
-    for(int i=0;i<N;i++){
-      for(int j=0;j<N;j++){
-	ret._internal[i][j]= -r._internal[i][j];
-    }}
-    return ret;
-  }
-  // *=,+=,-= operators inherit from corresponding "*,-,+" behaviour
-  template<class T>
-  inline iMatrix<vtype,N> &operator *=(const T &r) {
-    *this = (*this)*r;
-    return *this;
-  }
-  template<class T>
-  inline iMatrix<vtype,N> &operator -=(const T &r) {
-    *this = (*this)-r;
-    return *this;
-  }
-  template<class T>
-  inline iMatrix<vtype,N> &operator +=(const T &r) {
-    *this = (*this)+r;
-    return *this;
-  }
-  inline vtype & operator ()(int i,int j) {
-    return _internal[i][j];
-  }
-
-};
-
-    
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-    /////////////////////////////////////////// ADD         ///////////////////////////////////////////
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-    
-
-// ADD is simple for now; cannot mix types and straightforward template
-// Scalar +/- Scalar
-// Vector +/- Vector
-// Matrix +/- Matrix
-template<class vtype,class ltype,class rtype> inline void add(iScalar<vtype> * __restrict__ ret,
-                                                              const iScalar<ltype> * __restrict__ lhs,
-                                                              const iScalar<rtype> * __restrict__ rhs)
-{
-    add(&ret->_internal,&lhs->_internal,&rhs->_internal);
-}
-template<class vtype,class ltype,class rtype,int N> inline void add(iVector<vtype,N> * __restrict__ ret,
-                                                                    const iVector<ltype,N> * __restrict__ lhs,
-                                                                    const iVector<rtype,N> * __restrict__ rhs)
-{
-    for(int c=0;c<N;c++){
-        ret->_internal[c]=lhs->_internal[c]+rhs->_internal[c];
-    }
-    return;
-}
-template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
-                                                                      const iMatrix<ltype,N> * __restrict__ lhs,
-                                                                      const iMatrix<rtype,N> * __restrict__ rhs)
-{
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        add(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal[c1][c2]);
-    }}
-    return;
-}
-template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
-                                                                      const iScalar<ltype>   * __restrict__ lhs,
-                                                                      const iMatrix<rtype,N> * __restrict__ rhs)
-{
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        add(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
-    }}
-    return;
-}
-template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
-                                                                      const iMatrix<ltype,N> * __restrict__ lhs,
-                                                                      const iScalar<rtype>   * __restrict__ rhs)
-{
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        if ( c1==c2)
-            add(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
-        else
-            ret->_internal[c1][c2]=lhs->_internal[c1][c2];
-    }}
-    return;
-}
-// Need to figure multi-precision.
-template<class Mytype>  Mytype timesI(Mytype &r)
-{
-    iScalar<Complex> i;
-    i._internal = Complex(0,1);
-    return r*i;
-}
-
-                // + operator for scalar, vector, matrix
-template<class ltype,class rtype>
-//inline auto operator + (iScalar<ltype>& lhs,iScalar<rtype>&& rhs) -> iScalar<decltype(lhs._internal + rhs._internal)>
-inline auto operator + (const iScalar<ltype>& lhs,const iScalar<rtype>& rhs) -> iScalar<decltype(lhs._internal + rhs._internal)>
-{
-    typedef iScalar<decltype(lhs._internal+rhs._internal)> ret_t;
-    ret_t ret;
-    add(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator + (const iVector<ltype,N>& lhs,const iVector<rtype,N>& rhs) ->iVector<decltype(lhs._internal[0]+rhs._internal[0]),N>
-{
-    typedef iVector<decltype(lhs._internal[0]+rhs._internal[0]),N> ret_t;
-    ret_t ret;
-    add(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator + (const iMatrix<ltype,N>& lhs,const iMatrix<rtype,N>& rhs) ->iMatrix<decltype(lhs._internal[0][0]+rhs._internal[0][0]),N>
-{
-    typedef iMatrix<decltype(lhs._internal[0][0]+rhs._internal[0][0]),N> ret_t;
-    ret_t ret;
-    add(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator + (const iScalar<ltype>& lhs,const iMatrix<rtype,N>& rhs)->iMatrix<decltype(lhs._internal+rhs._internal[0][0]),N>
-{
-    typedef iMatrix<decltype(lhs._internal+rhs._internal[0][0]),N> ret_t;
-    ret_t ret;
-    add(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator + (const iMatrix<ltype,N>& lhs,const iScalar<rtype>& rhs)->iMatrix<decltype(lhs._internal[0][0]+rhs._internal),N>
-{
-    typedef iMatrix<decltype(lhs._internal[0][0]+rhs._internal),N> ret_t;
-    ret_t ret;
-    add(&ret,&lhs,&rhs);
-    return ret;
-}
-
-
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-    /////////////////////////////////////////// SUB         ///////////////////////////////////////////
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-    
-
-// SUB is simple for now; cannot mix types and straightforward template
-// Scalar +/- Scalar
-// Vector +/- Vector
-// Matrix +/- Matrix
-// Matrix /- scalar
-template<class vtype,class ltype,class rtype> inline void sub(iScalar<vtype> * __restrict__ ret,
-                                                              const iScalar<ltype> * __restrict__ lhs,
-                                                              const iScalar<rtype> * __restrict__ rhs)
-{
-    sub(&ret->_internal,&lhs->_internal,&rhs->_internal);
-}
-
-template<class vtype,class ltype,class rtype,int N> inline void sub(iVector<vtype,N> * __restrict__ ret,
-                                                                    const iVector<ltype,N> * __restrict__ lhs,
-                                                                    const iVector<rtype,N> * __restrict__ rhs)
-{
-    for(int c=0;c<N;c++){
-        ret->_internal[c]=lhs->_internal[c]-rhs->_internal[c];
-    }
-    return;
-}
-template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
-                                                                     const iMatrix<ltype,N> * __restrict__ lhs,
-                                                                     const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        sub(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal[c1][c2]);
-    }}
-    return;
-}
-template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
-                                                                     const iScalar<ltype> * __restrict__ lhs,
-                                                                     const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        if ( c1!=c2) {
-            sub(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
-        } else {
-            // Fails -- need unary minus. Catalogue other unops?
-            ret->_internal[c1][c2]=zero;
-            ret->_internal[c1][c2]=ret->_internal[c1][c2]-rhs->_internal[c1][c2];
-
-        }
-    }}
-    return;
-}
-template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
-                                                                     const iMatrix<ltype,N> * __restrict__ lhs,
-                                                                     const iScalar<rtype> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        if ( c1!=c2)
-            sub(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
-        else
-            ret->_internal[c1][c2]=lhs->_internal[c1][c2];
-    }}
-    return;
-}
-
-template<class v> void vprefetch(const iScalar<v> &vv)
-{
-  vprefetch(vv._internal);
-}
-template<class v,int N> void vprefetch(const iVector<v,N> &vv)
-{
-  for(int i=0;i<N;i++){
-    vprefetch(vv._internal[i]);
-  }
-}
-template<class v,int N> void vprefetch(const iMatrix<v,N> &vv)
-{
-  for(int i=0;i<N;i++){
-  for(int j=0;j<N;j++){
-    vprefetch(vv._internal[i][j]);
-  }}
-}
-
-    // - operator for scalar, vector, matrix
-template<class ltype,class rtype> inline auto
-operator - (const iScalar<ltype>& lhs, const iScalar<rtype>& rhs) -> iScalar<decltype(lhs._internal - rhs._internal)>
-{
-    typedef iScalar<decltype(lhs._internal-rhs._internal)> ret_t;
-    ret_t ret;
-    sub(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator - (const iVector<ltype,N>& lhs,const iVector<rtype,N>& rhs) ->iVector<decltype(lhs._internal[0]-rhs._internal[0]),N>
-{
-    typedef iVector<decltype(lhs._internal[0]-rhs._internal[0]),N> ret_t;
-    ret_t ret;
-    sub(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator - (const iMatrix<ltype,N>& lhs,const iMatrix<rtype,N>& rhs) ->iMatrix<decltype(lhs._internal[0][0]-rhs._internal[0][0]),N>
-{
-    typedef iMatrix<decltype(lhs._internal[0][0]-rhs._internal[0][0]),N> ret_t;
-    ret_t ret;
-    sub(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator - (const iScalar<ltype>& lhs,const iMatrix<rtype,N>& rhs)->iMatrix<decltype(lhs._internal-rhs._internal[0][0]),N>
-{
-    typedef iMatrix<decltype(lhs._internal-rhs._internal[0][0]),N> ret_t;
-    ret_t ret;
-    sub(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator - (const iMatrix<ltype,N>& lhs,const iScalar<rtype>& rhs)->iMatrix<decltype(lhs._internal[0][0]-rhs._internal),N>
-{
-    typedef iMatrix<decltype(lhs._internal[0][0]-rhs._internal),N> ret_t;
-    ret_t ret;
-    sub(&ret,&lhs,&rhs);
-    return ret;
-}
-
-///////////////////////////////////////////////////////////////////////////////////////////////////
-/////////////////////////////////////////// MAC         ///////////////////////////////////////////
-///////////////////////////////////////////////////////////////////////////////////////////////////
-
-    ///////////////////////////
-    // Legal multiplication table
-    ///////////////////////////
-    // scal x scal = scal
-    // mat x  mat  = mat
-    // mat  x scal = mat
-    // scal x mat  = mat
-    // mat  x vec  = vec
-    // vec  x scal = vec
-    // scal x vec  = vec
-    ///////////////////////////
-template<class rtype,class vtype,class mtype>
-inline  void mac(iScalar<rtype> * __restrict__ ret,const iScalar<vtype> * __restrict__ lhs,const iScalar<mtype> * __restrict__ rhs)
-{
-    mac(&ret->_internal,&lhs->_internal,&rhs->_internal);
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-    for(int c3=0;c3<N;c3++){
-        mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c3],&rhs->_internal[c3][c2]);
-    }}}
-    return;
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs){
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
-    }}
-    return;
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iScalar<ltype> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        mac(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
-    }}
-    return;
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iVector<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iVector<rtype,N> * __restrict__ rhs)
-{
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        mac(&ret->_internal[c1],&lhs->_internal[c1][c2],&rhs->_internal[c2]);
-    }}
-    return;
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iVector<rrtype,N> * __restrict__ ret,const iScalar<ltype> * __restrict__ lhs,const iVector<rtype,N> * __restrict__ rhs)
-{
-    for(int c1=0;c1<N;c1++){
-        mac(&ret->_internal[c1],&lhs->_internal,&rhs->_internal[c1]);
-    }
-    return;
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iVector<rrtype,N> * __restrict__ ret,const iVector<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs)
-{
-    for(int c1=0;c1<N;c1++){
-        mac(&ret->_internal[c1],&lhs->_internal[c1],&rhs->_internal);
-    }
-    return;
-}
-
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-    /////////////////////////////////////////// MUL         ///////////////////////////////////////////
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-
-    
-template<class rtype,class vtype,class mtype>
-inline void mult(iScalar<rtype> * __restrict__ ret,const iScalar<mtype> * __restrict__ lhs,const iScalar<vtype> * __restrict__ rhs){
-    mult(&ret->_internal,&lhs->_internal,&rhs->_internal);
-}
-
-template<class rrtype,class ltype,class rtype,int N>
-inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        mult(&ret->_internal[c1][c2],&lhs->_internal[c1][0],&rhs->_internal[0][c2]);
-        for(int c3=1;c3<N;c3++){
-            mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c3],&rhs->_internal[c3][c2]);
-        }
-    }}
-    return;
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        mult(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
-    }}
-    return;
-}
-
-template<class rrtype,class ltype,class rtype, int N>
-inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iScalar<ltype>   * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        mult(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
-    }}
-    return;
-}
-// Matrix left multiplies vector
-template<class rtype,class vtype,class mtype,int N>
-inline void mult(iVector<rtype,N> * __restrict__ ret,const iMatrix<mtype,N> * __restrict__ lhs,const iVector<vtype,N> * __restrict__ rhs)
-{
-    for(int c1=0;c1<N;c1++){
-        mult(&ret->_internal[c1],&lhs->_internal[c1][0],&rhs->_internal[0]);
-        for(int c2=1;c2<N;c2++){
-            mac(&ret->_internal[c1],&lhs->_internal[c1][c2],&rhs->_internal[c2]);
-        }
-    }
-    return;
-}
-template<class rtype,class vtype,class mtype,int N>
-inline void mult(iVector<rtype,N> * __restrict__ ret,
-                 const iScalar<mtype>   * __restrict__ lhs,
-                 const iVector<vtype,N> * __restrict__ rhs){
-    for(int c1=0;c1<N;c1++){
-        mult(&ret->_internal[c1],&lhs->_internal,&rhs->_internal[c1]);
-    }
-}
-template<class rtype,class vtype,class mtype,int N>
-inline void mult(iVector<rtype,N> * __restrict__ ret,
-                 const iVector<vtype,N> * __restrict__ rhs,
-                 const iScalar<mtype> * __restrict__ lhs){
-    mult(ret,lhs,rhs);
-}
-    
-
-
-template<class rtype,class vtype,class mtype,int N> inline
-iVector<rtype,N> operator * (const iMatrix<mtype,N>& lhs,const iVector<vtype,N>& rhs)
-{
-    iVector<rtype,N> ret;
-    mult(&ret,&lhs,&rhs);
-    return ret;
-}
-
-template<class rtype,class vtype,class mtype,int N> inline
-iVector<rtype,N> operator * (const iScalar<mtype>& lhs,const iVector<vtype,N>& rhs)
-{
-    iVector<rtype,N> ret;
-    mult(&ret,&lhs,&rhs);
-    return ret;
-}
-
-template<class rtype,class vtype,class mtype,int N> inline
-iVector<rtype,N> operator * (const iVector<mtype,N>& lhs,const iScalar<vtype>& rhs)
-{
-    iVector<rtype,N> ret;
-    mult(&ret,&lhs,&rhs);
-    return ret;
-}
-    
-    //////////////////////////////////////////////////////////////////
-    // Glue operators to mult routines. Must resolve return type cleverly from typeof(internal)
-    // since nesting matrix<scalar> x matrix<matrix>-> matrix<matrix>
-    // while         matrix<scalar> x matrix<scalar>-> matrix<scalar>
-    // so return type depends on argument types in nasty way.
-    //////////////////////////////////////////////////////////////////
-    // scal x scal = scal
-    // mat x  mat  = mat
-    // mat  x scal = mat
-    // scal x mat  = mat
-    // mat  x vec  = vec
-    // vec  x scal = vec
-    // scal x vec  = vec
-    //
-    // We can special case scalar_type ??
-template<class l,class r>
-inline auto operator * (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(lhs._internal * rhs._internal)>
-{
-    typedef iScalar<decltype(lhs._internal*rhs._internal)> ret_t;
-    ret_t ret;
-    mult(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class l,class r,int N> inline
-auto operator * (const iMatrix<l,N>& lhs,const iMatrix<r,N>& rhs) -> iMatrix<decltype(lhs._internal[0][0]*rhs._internal[0][0]),N>
-{
-    typedef decltype(lhs._internal[0][0]*rhs._internal[0][0]) ret_t;
-    iMatrix<ret_t,N> ret;
-    mult(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class l,class r, int N> inline
-auto operator * (const iMatrix<r,N>& lhs,const iScalar<l>& rhs) -> iMatrix<decltype(lhs._internal[0][0]*rhs._internal),N>
-{
-    typedef decltype(lhs._internal[0][0]*rhs._internal) ret_t;
-        
-    iMatrix<ret_t,N> ret;
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        mult(&ret._internal[c1][c2],&lhs._internal[c1][c2],&rhs._internal);
-    }}
-    return ret;
-}
-template<class l,class r,int N> inline
-auto operator * (const iScalar<l>& lhs,const iMatrix<r,N>& rhs) -> iMatrix<decltype(lhs._internal*rhs._internal[0][0]),N>
-{
-    typedef decltype(lhs._internal*rhs._internal[0][0]) ret_t;
-    iMatrix<ret_t,N> ret;
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        mult(&ret._internal[c1][c2],&lhs._internal,&rhs._internal[c1][c2]);
-    }}
-    return ret;
-}
-template<class l,class r,int N> inline
-auto operator * (const iMatrix<l,N>& lhs,const iVector<r,N>& rhs) -> iVector<decltype(lhs._internal[0][0]*rhs._internal[0]),N>
-{
-    typedef decltype(lhs._internal[0][0]*rhs._internal[0]) ret_t;
-    iVector<ret_t,N> ret;
-    for(int c1=0;c1<N;c1++){
-        mult(&ret._internal[c1],&lhs._internal[c1][0],&rhs._internal[0]);
-        for(int c2=1;c2<N;c2++){
-            mac(&ret._internal[c1],&lhs._internal[c1][c2],&rhs._internal[c2]);
-        }
-    }
-    return ret;
-}
-template<class l,class r,int N> inline
-auto operator * (const iScalar<l>& lhs,const iVector<r,N>& rhs) -> iVector<decltype(lhs._internal*rhs._internal[0]),N>
-{
-    typedef decltype(lhs._internal*rhs._internal[0]) ret_t;
-    iVector<ret_t,N> ret;
-    for(int c1=0;c1<N;c1++){
-        mult(&ret._internal[c1],&lhs._internal,&rhs._internal[c1]);
-    }
-    return ret;
-}
-template<class l,class r,int N> inline
-auto operator * (const iVector<l,N>& lhs,const iScalar<r>& rhs) -> iVector<decltype(lhs._internal[0]*rhs._internal),N>
-{
-    typedef decltype(lhs._internal[0]*rhs._internal) ret_t;
-    iVector<ret_t,N> ret;
-    for(int c1=0;c1<N;c1++){
-        mult(&ret._internal[c1],&lhs._internal[c1],&rhs._internal);
-    }
-    return ret;
-}
-//////////////////////////////////////////////////////////////////////////////////////////
-// Must support native C++ types Integer, Complex, Real
-//////////////////////////////////////////////////////////////////////////////////////////
-
-// multiplication by fundamental scalar type
-template<class l,int N> inline iScalar<l> operator * (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
-{
-  typename iScalar<l>::tensor_reduced srhs(rhs);
-  return lhs*srhs;
-}
-template<class l,int N> inline iScalar<l> operator * (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
-
-template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
-{
-  typename iVector<l,N>::tensor_reduced srhs(rhs);
-  return lhs*srhs;
-}
-template<class l,int N> inline iVector<l,N> operator * (const typename iScalar<l>::scalar_type lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
-
-template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type &rhs) 
-{
-  typename iMatrix<l,N>::tensor_reduced srhs(rhs);
-  return lhs*srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator * (const typename iScalar<l>::scalar_type & lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
-
-////////////////////////////////////////////////////////////////////
-// Double support; cast to "scalar_type" through constructor
-////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,double rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs*srhs;
-}
-template<class l> inline iScalar<l> operator * (double lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
-
-template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,double rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs*srhs;
-}
-template<class l,int N> inline iVector<l,N> operator * (double lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
-
-template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,double rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs*srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator * (double lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
-
-////////////////////////////////////////////////////////////////////
-// Complex support; cast to "scalar_type" through constructor
-////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,ComplexD rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs*srhs;
-}
-template<class l> inline iScalar<l> operator * (ComplexD lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
-
-template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,ComplexD rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs*srhs;
-}
-template<class l,int N> inline iVector<l,N> operator * (ComplexD lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
-
-template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,ComplexD rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs*srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator * (ComplexD lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
-
-////////////////////////////////////////////////////////////////////
-// Integer support; cast to "scalar_type" through constructor
-////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,Integer rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs*srhs;
-}
-template<class l> inline iScalar<l> operator * (Integer lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
-
-template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,Integer rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs*srhs;
-}
-template<class l,int N> inline iVector<l,N> operator * (Integer lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
-
-template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,Integer rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs*srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator * (Integer lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
-
-
-
-///////////////////////////////////////////////////////////////////////////////////////////////
-// addition by fundamental scalar type applies to matrix(down diag) and scalar
-///////////////////////////////////////////////////////////////////////////////////////////////
-template<class l,int N> inline iScalar<l> operator + (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
-{
-  typename iScalar<l>::tensor_reduced srhs(rhs);
-  return lhs+srhs;
-}
-template<class l,int N> inline iScalar<l> operator + (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
-
-template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
-{
-  typename iMatrix<l,N>::tensor_reduced srhs(rhs);
-  return lhs+srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator + (const typename iScalar<l>::scalar_type lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
-
-////////////////////////////////////////////////////////////////////
-// Double support; cast to "scalar_type" through constructor
-////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator + (const iScalar<l>& lhs,double rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs+srhs;
-}
-template<class l> inline iScalar<l> operator + (double lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
-
-template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,double rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs+srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator + (double lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
-
-////////////////////////////////////////////////////////////////////
-// Integer support; cast to "scalar_type" through constructor
-////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator + (const iScalar<l>& lhs,Integer rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs+srhs;
-}
-template<class l> inline iScalar<l> operator + (Integer lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
-
-template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,Integer rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs+srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator + (Integer lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
-
-
-///////////////////////////////////////////////////////////////////////////////////////////////
-// subtraction of fundamental scalar type applies to matrix(down diag) and scalar
-///////////////////////////////////////////////////////////////////////////////////////////////
-template<class l,int N> inline iScalar<l> operator - (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
-{
-  typename iScalar<l>::tensor_reduced srhs(rhs);
-  return lhs-srhs;
-}
-template<class l,int N> inline iScalar<l> operator - (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) 
-{
-  typename iScalar<l>::tensor_reduced slhs(lhs);
-  return slhs-rhs;
-}
-
-template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
-{
-  typename iScalar<l>::tensor_reduced srhs(rhs);
-  return lhs-srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator - (const typename iScalar<l>::scalar_type lhs,const iMatrix<l,N>& rhs) 
-{
-  typename iScalar<l>::tensor_reduced slhs(lhs);
-  return slhs-rhs;
-}
-
-////////////////////////////////////////////////////////////////////
-// Double support; cast to "scalar_type" through constructor
-////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator - (const iScalar<l>& lhs,double rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs-srhs;
-}
-template<class l> inline iScalar<l> operator - (double lhs,const iScalar<l>& rhs) 
-{
-  typename iScalar<l>::scalar_type t(lhs);
-  typename iScalar<l>::tensor_reduced slhs(t);
-  return slhs-rhs;
-}
-
-template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,double rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs-srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator - (double lhs,const iMatrix<l,N>& rhs) 
-{
-  typename iScalar<l>::scalar_type t(lhs);
-  typename iScalar<l>::tensor_reduced slhs(t);
-  return slhs-rhs;
-}
-
-////////////////////////////////////////////////////////////////////
-// Integer support; cast to "scalar_type" through constructor
-////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator - (const iScalar<l>& lhs,Integer rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs-srhs;
-}
-template<class l> inline iScalar<l> operator - (Integer lhs,const iScalar<l>& rhs) 
-{
-  typename iScalar<l>::scalar_type t(lhs);
-  typename iScalar<l>::tensor_reduced slhs(t);
-  return slhs-rhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,Integer rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs-srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator - (Integer lhs,const iMatrix<l,N>& rhs) 
-{
-  typename iScalar<l>::scalar_type t(lhs);
-  typename iScalar<l>::tensor_reduced slhs(t);
-  return slhs-rhs;
-}
-
-
-
-
-    ///////////////////////////////////////////////////////////////////////////////////////
-    // innerProduct Scalar x Scalar -> Scalar
-    // innerProduct Vector x Vector -> Scalar
-    // innerProduct Matrix x Matrix -> Scalar
-    ///////////////////////////////////////////////////////////////////////////////////////
-    template<class l,class r,int N> inline
-    auto innerProduct (const iVector<l,N>& lhs,const iVector<r,N>& rhs) -> iScalar<decltype(innerProduct(lhs._internal[0],rhs._internal[0]))>
-    {
-        typedef decltype(innerProduct(lhs._internal[0],rhs._internal[0])) ret_t;
-        iScalar<ret_t> ret=zero;
-        for(int c1=0;c1<N;c1++){
-            ret._internal += innerProduct(lhs._internal[c1],rhs._internal[c1]);
-        }
-        return ret;
-    }
-    template<class l,class r,int N> inline
-    auto innerProduct (const iMatrix<l,N>& lhs,const iMatrix<r,N>& rhs) -> iScalar<decltype(innerProduct(lhs._internal[0][0],rhs._internal[0][0]))>
-    {
-        typedef decltype(innerProduct(lhs._internal[0][0],rhs._internal[0][0])) ret_t;
-        iScalar<ret_t> ret=zero;
-        iScalar<ret_t> tmp;
-        for(int c1=0;c1<N;c1++){
-        for(int c2=0;c2<N;c2++){
-	  ret._internal+=innerProduct(lhs._internal[c1][c2],rhs._internal[c1][c2]);
-        }}
-        return ret;
-    }
-    template<class l,class r> inline
-    auto innerProduct (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(innerProduct(lhs._internal,rhs._internal))>
-    {
-        typedef decltype(innerProduct(lhs._internal,rhs._internal)) ret_t;
-        iScalar<ret_t> ret;
-        ret._internal = innerProduct(lhs._internal,rhs._internal);
-        return ret;
-    }
-
-    ///////////////////////////////////////////////////////////////////////////////////////
-    // outerProduct Scalar x Scalar -> Scalar
-    //              Vector x Vector -> Matrix
-    ///////////////////////////////////////////////////////////////////////////////////////
-
-template<class l,class r,int N> inline
-auto outerProduct (const iVector<l,N>& lhs,const iVector<r,N>& rhs) -> iMatrix<decltype(outerProduct(lhs._internal[0],rhs._internal[0])),N>
-{
-    typedef decltype(outerProduct(lhs._internal[0],rhs._internal[0])) ret_t;
-    iMatrix<ret_t,N> ret;
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        ret._internal[c1][c2] = outerProduct(lhs._internal[c1],rhs._internal[c2]);
-    }}
-    return ret;
-}
-template<class l,class r> inline
-auto outerProduct (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(outerProduct(lhs._internal,rhs._internal))>
-{
-    typedef decltype(outerProduct(lhs._internal,rhs._internal)) ret_t;
-    iScalar<ret_t> ret;
-    ret._internal = outerProduct(lhs._internal,rhs._internal);
-    return ret;
-}
-
-inline ComplexF outerProduct(const ComplexF &l, const ComplexF& r)
-{
-  return l*r;
-}
-inline ComplexD outerProduct(const ComplexD &l, const ComplexD& r)
-{
-  return l*r;
-}
-inline RealF outerProduct(const RealF &l, const RealF& r)
-{
-  return l*r;
-}
-inline RealD outerProduct(const RealD &l, const RealD& r)
-{
-  return l*r;
-}
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-    /////////////////////////////////////////// CONJ         ///////////////////////////////////////////
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
- 
-// Conj function for scalar, vector, matrix
-template<class vtype> inline iScalar<vtype> conj(const iScalar<vtype>&r)
-{
-    iScalar<vtype> ret;
-    ret._internal = conj(r._internal);
-    return ret;
-}
-template<class vtype,int N> inline iVector<vtype,N> conj(const iVector<vtype,N>&r)
-{
-  iVector<vtype,N> ret;
-  for(int i=0;i<N;i++){
-    ret._internal[i] = conj(r._internal[i]);
-  }
-  return ret;
-}
-template<class vtype,int N> inline iMatrix<vtype,N> conj(const iMatrix<vtype,N>&r)
-{
-  iMatrix<vtype,N> ret;
-  for(int i=0;i<N;i++){
-  for(int j=0;j<N;j++){
-    ret._internal[i][j] = conj(r._internal[i][j]);
-  }}
-  return ret;
-}
-
-// Adj function for scalar, vector, matrix
-template<class vtype> inline iScalar<vtype> adj(const iScalar<vtype>&r)
-{
-    iScalar<vtype> ret;
-    ret._internal = adj(r._internal);
-    return ret;
-}
-template<class vtype,int N> inline iVector<vtype,N> adj(const iVector<vtype,N>&r)
-{
-    iVector<vtype,N> ret;
-    for(int i=0;i<N;i++){
-        ret._internal[i] = adj(r._internal[i]);
-    }
-    return ret;
-}
-template<class vtype,int N> inline iMatrix<vtype,N> adj(const iMatrix<vtype,N> &arg)
-{
-    iMatrix<vtype,N> ret;
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        ret._internal[c1][c2]=adj(arg._internal[c2][c1]);
-    }}
-    return ret;
-}
-
-
-
-/////////////////////////////////////////////////////////////////
-// Transpose all indices
-/////////////////////////////////////////////////////////////////
-
-inline ComplexD transpose(ComplexD &rhs){  return rhs;}
-inline ComplexF transpose(ComplexF &rhs){  return rhs;}
-inline RealD transpose(RealD &rhs){  return rhs;}
-inline RealF transpose(RealF &rhs){  return rhs;}
-
-template<class vtype,int N>
-  inline typename std::enable_if<isGridTensor<vtype>::value, iMatrix<vtype,N> >::type 
-  transpose(iMatrix<vtype,N> arg)
-  {
-    iMatrix<vtype,N> ret;
-    for(int i=0;i<N;i++){
-      for(int j=0;j<N;j++){
-	ret._internal[i][j] = transpose(arg._internal[j][i]); // NB recurses
-      }}
-    return ret;
-  }
-template<class vtype,int N>
-  inline typename std::enable_if<isGridTensor<vtype>::notvalue, iMatrix<vtype,N> >::type 
-  transpose(iMatrix<vtype,N> arg)
-  {
-    iMatrix<vtype,N> ret;
-    for(int i=0;i<N;i++){
-      for(int j=0;j<N;j++){
-	ret._internal[i][j] = arg._internal[j][i]; // Stop recursion if not a tensor type
-      }}
-    return ret;
-  }
-
-template<class vtype>
-  inline typename std::enable_if<isGridTensor<vtype>::value, iScalar<vtype> >::type 
-  transpose(iScalar<vtype> arg)
-  {
-    iScalar<vtype> ret;
-    ret._internal = transpose(arg._internal); // NB recurses
-    return ret;
-  }
-
-template<class vtype>
-  inline typename std::enable_if<isGridTensor<vtype>::notvalue, iScalar<vtype> >::type 
-  transpose(iScalar<vtype> arg)
-  {
-    iScalar<vtype> ret;
-    ret._internal = arg._internal; // NB recursion stops
-    return ret;
-  }
-
-
-////////////////////////////////////////////////////////////////////////////////////////////
-// Transpose a specific index; instructive to compare this style of recursion termination
-// to that of adj; which is easiers?
-////////////////////////////////////////////////////////////////////////////////////////////
-template<int Level,class vtype,int N> inline 
-  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value, iMatrix<vtype,N> >::type 
-transposeIndex (const iMatrix<vtype,N> &arg)
-{
-  iMatrix<vtype,N> ret;
-  for(int i=0;i<N;i++){
-    for(int j=0;j<N;j++){
-      ret._internal[i][j] = arg._internal[j][i]; 
-  }}
-  return ret;
-}
-// or not
-template<int Level,class vtype,int N> inline 
-typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue, iMatrix<vtype,N> >::type 
-transposeIndex (const iMatrix<vtype,N> &arg)
-{
-  iMatrix<vtype,N> ret;
-  for(int i=0;i<N;i++){
-    for(int j=0;j<N;j++){
-      ret._internal[i][j] = transposeIndex<Level>(arg._internal[i][j]); 
-  }}
-  return ret;
-}
-template<int Level,class vtype> inline 
-typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue, iScalar<vtype> >::type 
-transposeIndex (const iScalar<vtype> &arg)
-{
-  iScalar<vtype> ret;
-  ret._internal=transposeIndex<Level>(arg._internal);
-  return ret;
-}
-template<int Level,class vtype> inline 
-typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value, iScalar<vtype> >::type 
-transposeIndex (const iScalar<vtype> &arg)
-{
-  return arg;
-}
-
-//////////////////////////////////////////////////////////////////
-// Traces: both all indices and a specific index 
-/////////////////////////////////////////////////////////////////
-
-inline ComplexF trace( const ComplexF &arg){    return arg;}
-inline ComplexD trace( const ComplexD &arg){    return arg;}
-inline RealF trace( const RealF &arg){    return arg;}
-inline RealD trace( const RealD &arg){    return arg;}
-
-template<int Level> inline ComplexF traceIndex(const ComplexF arg) { return arg;}
-template<int Level> inline ComplexD traceIndex(const ComplexD arg) { return arg;}
-template<int Level> inline RealF traceIndex(const RealF arg) { return arg;}
-template<int Level> inline RealD traceIndex(const RealD arg) { return arg;}
-
-template<class vtype,int N>
-inline auto trace(const iMatrix<vtype,N> &arg) -> iScalar<decltype(trace(arg._internal[0][0]))>
-{
-    iScalar<decltype( trace(arg._internal[0][0] )) > ret;
-    zeroit(ret._internal);
-    for(int i=0;i<N;i++){
-        ret._internal=ret._internal+trace(arg._internal[i][i]);
-    }
-    return ret;
-}
-template<class vtype>
-inline auto trace(const iScalar<vtype> &arg) -> iScalar<decltype(trace(arg._internal))>
-{
-    iScalar<decltype(trace(arg._internal))> ret;
-    ret._internal=trace(arg._internal);
-    return ret;
-}
-////////////////////////////////////////////////////////////////////////////////////////////////////////
-// Trace Specific indices.
-////////////////////////////////////////////////////////////////////////////////////////////////////////
-/*
-template<int Level,class vtype> inline 
-auto traceIndex(const iScalar<vtype> &arg) -> iScalar<decltype(traceIndex<Level>(arg._internal)) >
-{
-  iScalar<decltype(traceIndex<Level>(arg._internal))> ret;
-  ret._internal = traceIndex<Level>(arg._internal);
-  return ret;
-}
-*/
-template<int Level,class vtype> inline auto 
-traceIndex (const iScalar<vtype> &arg) ->
-typename 
-std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue, 
-  iScalar<decltype(traceIndex<Level>(arg._internal))> >::type 
-
-{
-  iScalar<decltype(traceIndex<Level>(arg._internal))> ret;
-  ret._internal=traceIndex<Level>(arg._internal);
-  return ret;
-}
-template<int Level,class vtype> inline auto
-traceIndex (const iScalar<vtype> &arg) ->
-typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value, 
-                        iScalar<vtype> >::type 
-{
-  return arg;
-}
-
-// If we hit the right index, return scalar and trace it with no further recursion
-template<int Level,class vtype,int N> inline 
-auto traceIndex(const iMatrix<vtype,N> &arg) ->
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value,  // Index matches
-                                                    iScalar<vtype> >::type                              // return scalar
-{
-  iScalar<vtype> ret;
-  zeroit(ret._internal);
-  for(int i=0;i<N;i++){
-    ret._internal = ret._internal + arg._internal[i][i];
-  }
-  return ret;
-}
-
-// not this level, so recurse
-template<int Level,class vtype,int N> inline 
-auto traceIndex(const iMatrix<vtype,N> &arg) ->
-  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,// No index match
-         iMatrix<decltype(traceIndex<Level>(arg._internal[0][0])),N> >::type     // return matrix
-{
-  iMatrix<decltype(traceIndex<Level>(arg._internal[0][0])),N> ret;
-  for(int i=0;i<N;i++){
-  for(int j=0;j<N;j++){
-    ret._internal[i][j] = traceIndex<Level>(arg._internal[i][j]);
-  }}
-  return ret;
-}
-
-//////////////////////////////////////////////////////////////////////////////
-// Peek on a specific index; returns a scalar in that index, tensor inherits rest
-//////////////////////////////////////////////////////////////////////////////
-// If we hit the right index, return scalar with no further recursion
-
-//template<int Level> inline ComplexF peekIndex(const ComplexF arg) { return arg;}
-//template<int Level> inline ComplexD peekIndex(const ComplexD arg) { return arg;}
-//template<int Level> inline RealF peekIndex(const RealF arg) { return arg;}
-//template<int Level> inline RealD peekIndex(const RealD arg) { return arg;}
-
-// Scalar peek, no indices
-template<int Level,class vtype> inline 
-  auto peekIndex(const iScalar<vtype> &arg) -> 
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value,  // Index matches
-  iScalar<vtype> >::type                              // return scalar
-{
-  return arg;
-}
-// Vector peek, one index
-template<int Level,class vtype,int N> inline 
-  auto peekIndex(const iVector<vtype,N> &arg,int i) -> 
-  typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::value,  // Index matches
-  iScalar<vtype> >::type                              // return scalar
-{
-  iScalar<vtype> ret;                              // return scalar
-  ret._internal = arg._internal[i];
-  return ret;
-}
-// Matrix peek, two indices
-template<int Level,class vtype,int N> inline 
-  auto peekIndex(const iMatrix<vtype,N> &arg,int i,int j) -> 
-  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value,  // Index matches
-  iScalar<vtype> >::type                              // return scalar
-{
-  iScalar<vtype> ret;                              // return scalar
-  ret._internal = arg._internal[i][j];
-  return ret;
-}
-
-/////////////
-// No match peek for scalar,vector,matrix must forward on either 0,1,2 args. Must have 9 routines with notvalue
-/////////////
-// scalar
-template<int Level,class vtype> inline 
-  auto peekIndex(const iScalar<vtype> &arg) ->                                     // Scalar 0 index  
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does NOT match
-  iScalar<decltype(peekIndex<Level>(arg._internal))> >::type                       
-{
-  iScalar<decltype(peekIndex<Level>(arg._internal))> ret;
-  ret._internal= peekIndex<Level>(arg._internal);
-  return ret;
-}
-template<int Level,class vtype> inline 
-  auto peekIndex(const iScalar<vtype> &arg,int i) ->                             // Scalar 1 index
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does NOT match
-  iScalar<decltype(peekIndex<Level>(arg._internal,i))> >::type                       
-{
-  iScalar<decltype(peekIndex<Level>(arg._internal,i))> ret;
-  ret._internal=peekIndex<Level>(arg._internal,i);
-  return ret;
-}
-template<int Level,class vtype> inline 
-  auto peekIndex(const iScalar<vtype> &arg,int i,int j) ->                         // Scalar, 2 index
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does NOT match
-  iScalar<decltype(peekIndex<Level>(arg._internal,i,j))> >::type                       
-{
-  iScalar<decltype(peekIndex<Level>(arg._internal,i,j))> ret;
-  ret._internal=peekIndex<Level>(arg._internal,i,j);
-  return ret;
-}
-// vector
-template<int Level,class vtype,int N> inline 
-auto peekIndex(const iVector<vtype,N> &arg) -> 
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does not match
-  iVector<decltype(peekIndex<Level>(arg._internal[0])),N> >::type                       
-{
-  iVector<decltype(peekIndex<Level>(arg._internal[0])),N> ret;
-  for(int ii=0;ii<N;ii++){
-    ret._internal[ii]=peekIndex<Level>(arg._internal[ii]);
-  }
-  return ret;
-}
-template<int Level,class vtype,int N> inline 
-  auto peekIndex(const iVector<vtype,N> &arg,int i) -> 
-  typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,  // Index does not match
-  iVector<decltype(peekIndex<Level>(arg._internal[0],i)),N> >::type                       
-{
-  iVector<decltype(peekIndex<Level>(arg._internal[0],i)),N> ret;
-  for(int ii=0;ii<N;ii++){
-    ret._internal[ii]=peekIndex<Level>(arg._internal[ii],i);
-  }
-  return ret;
-}
-template<int Level,class vtype,int N> inline 
-  auto peekIndex(const iVector<vtype,N> &arg,int i,int j) -> 
-  typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,  // Index does not match
-  iVector<decltype(peekIndex<Level>(arg._internal[0],i,j)),N> >::type                       
-{
-  iVector<decltype(peekIndex<Level>(arg._internal[0],i,j)),N> ret;
-  for(int ii=0;ii<N;ii++){
-    ret._internal[ii]=peekIndex<Level>(arg._internal[ii],i,j);
-  }
-  return ret;
-}
-// matrix
-template<int Level,class vtype,int N> inline 
-auto peekIndex(const iMatrix<vtype,N> &arg) -> 
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does not match
-  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0])),N> >::type                       
-{
-  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0])),N> ret;
-  for(int ii=0;ii<N;ii++){
-  for(int jj=0;jj<N;jj++){
-    ret._internal[ii][jj]=peekIndex<Level>(arg._internal[ii][jj]);// Could avoid this because peeking a scalar is dumb
-  }}
-  return ret;
-}
-template<int Level,class vtype,int N> inline 
-  auto peekIndex(const iMatrix<vtype,N> &arg,int i) -> 
-  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,  // Index does not match
-  iMatrix<decltype(peekIndex<Level>(arg._internal[0],i)),N> >::type                       
-{
-  iMatrix<decltype(peekIndex<Level>(arg._internal[0],i)),N> ret;
-  for(int ii=0;ii<N;ii++){
-  for(int jj=0;jj<N;jj++){
-    ret._internal[ii][jj]=peekIndex<Level>(arg._internal[ii][jj],i);
-  }}
-  return ret;
-}
-template<int Level,class vtype,int N> inline 
-  auto peekIndex(const iMatrix<vtype,N> &arg,int i,int j) -> 
-  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,  // Index does not match
-  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0],i,j)),N> >::type                       
-{
-  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0],i,j)),N> ret;
-  for(int ii=0;ii<N;ii++){
-  for(int jj=0;jj<N;jj++){
-    ret._internal[ii][jj]=peekIndex<Level>(arg._internal[ii][jj],i,j);
-  }}
-  return ret;
-}
-
-
-//////////////////////////////////////////////////////////////////////////////
-// Poke a specific index; 
-//////////////////////////////////////////////////////////////////////////////
-
-// Scalar poke
-template<int Level,class vtype> inline 
-  void pokeIndex(iScalar<vtype> &ret, 
-		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value,iScalar<vtype> >::type &arg)
-{
-  ret._internal = arg._internal;
-}
-// Vector poke, one index
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iVector<vtype,N> &ret, 
-		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::value,iScalar<vtype> >::type &arg,int i)
-{
-  ret._internal[i] = arg._internal;
-}
-// Vector poke, two indices
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iMatrix<vtype,N> &ret, 
-		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value,iScalar<vtype> >::type &arg,int i,int j)
-{
-  ret._internal[i][j] = arg._internal;
-}
-
-/////////////
-// No match poke for scalar,vector,matrix must forward on either 0,1,2 args. Must have 9 routines with notvalue
-/////////////
-// scalar
-template<int Level,class vtype> inline 
-  void pokeIndex(iScalar<vtype> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,iScalar<decltype(peekIndex<Level>(ret._internal))> >::type &arg)
-		 
-{
-  pokeIndex<Level>(ret._internal,arg._internal);
-}
-template<int Level,class vtype> inline 
-  void pokeIndex(iScalar<vtype> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,iScalar<decltype(peekIndex<Level>(ret._internal,0))> >::type &arg,
-		 int i)
-		 
-{
-  pokeIndex<Level>(ret._internal,arg._internal,i);
-}
-template<int Level,class vtype> inline 
-  void pokeIndex(iScalar<vtype> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,iScalar<decltype(peekIndex<Level>(ret._internal,0,0))> >::type &arg,
-		 int i,int j)
-		 
-{
-  pokeIndex<Level>(ret._internal,arg._internal,i,j);
-}
-
-// Vector
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iVector<vtype,N> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,iVector<decltype(peekIndex<Level>(ret._internal)),N> >::type &arg)
-		 
-{
-  for(int ii=0;ii<N;ii++){
-    pokeIndex<Level>(ret._internal[ii],arg._internal[ii]);
-  }
-}
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iVector<vtype,N> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,iVector<decltype(peekIndex<Level>(ret._internal,0)),N> >::type &arg,
-		 int i)
-		 
-{
-  for(int ii=0;ii<N;ii++){
-    pokeIndex<Level>(ret._internal[ii],arg._internal[ii],i);
-  }
-}
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iVector<vtype,N> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,iVector<decltype(peekIndex<Level>(ret._internal,0,0)),N> >::type &arg,
-		 int i,int j)
-		 
-{
-  for(int ii=0;ii<N;ii++){
-    pokeIndex<Level>(ret._internal[ii],arg._internal[ii],i,j);
-  }
-}
-
-// Matrix
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iMatrix<vtype,N> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,iMatrix<decltype(peekIndex<Level>(ret._internal)),N> >::type &arg)
-		 
-{
-  for(int ii=0;ii<N;ii++){
-  for(int jj=0;jj<N;jj++){
-    pokeIndex<Level>(ret._internal[ii][jj],arg._internal[ii][jj]);
-  }}
-}
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iMatrix<vtype,N> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,iMatrix<decltype(peekIndex<Level>(ret._internal,0)),N> >::type &arg,
-		 int i)
-		 
-{
-  for(int ii=0;ii<N;ii++){
-  for(int jj=0;jj<N;jj++){
-    pokeIndex<Level>(ret._internal[ii][jj],arg._internal[ii][jj],i);
-  }}
-}
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iMatrix<vtype,N> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,iMatrix<decltype(peekIndex<Level>(ret._internal,0,0)),N> >::type &arg,
-		 int i,int j)
-		 
-{
-  for(int ii=0;ii<N;ii++){
-  for(int jj=0;jj<N;jj++){
-    pokeIndex<Level>(ret._internal[ii][jj],arg._internal[ii][jj],i,j);
-  }}
-}
-
-/////////////////////////////////////////////////////////////////
-// Can only take the real/imag part of scalar objects, since
-// lattice objects of different complex nature are non-conformable.
-/////////////////////////////////////////////////////////////////
-template<class itype> inline auto real(const iScalar<itype> &z) -> iScalar<decltype(real(z._internal))>
-{
-    iScalar<decltype(real(z._internal))> ret;
-    ret._internal = real(z._internal);
-    return ret;
-}
-template<class itype,int N> inline auto real(const iMatrix<itype,N> &z) -> iMatrix<decltype(real(z._internal[0][0])),N>
-{
-    iMatrix<decltype(real(z._internal[0][0])),N> ret;
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        ret._internal[c1][c2] = real(z._internal[c1][c2]);
-    }}
-    return ret;
-}
-template<class itype,int N> inline auto real(const iVector<itype,N> &z) -> iVector<decltype(real(z._internal[0])),N>
-{
-    iVector<decltype(real(z._internal[0])),N> ret;
-    for(int c1=0;c1<N;c1++){
-        ret._internal[c1] = real(z._internal[c1]);
-    }
-    return ret;
-}
-    
-template<class itype> inline auto imag(const iScalar<itype> &z) -> iScalar<decltype(imag(z._internal))>
-{
-    iScalar<decltype(imag(z._internal))> ret;
-    ret._internal = imag(z._internal);
-    return ret;
-}
-template<class itype,int N> inline auto imag(const iMatrix<itype,N> &z) -> iMatrix<decltype(imag(z._internal[0][0])),N>
-{
-    iMatrix<decltype(imag(z._internal[0][0])),N> ret;
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        ret._internal[c1][c2] = imag(z._internal[c1][c2]);
-    }}
-    return ret;
-}
-template<class itype,int N> inline auto imag(const iVector<itype,N> &z) -> iVector<decltype(imag(z._internal[0])),N>
-{
-    iVector<decltype(imag(z._internal[0])),N> ret;
-    for(int c1=0;c1<N;c1++){
-        ret._internal[c1] = imag(z._internal[c1]);
-    }
-    return ret;
-}
-
-};
-    
-#endif

From 62ee8e1cb3ab37cb6a273c1d03da327962edb892 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 18:37:56 +0100
Subject: [PATCH 072/429] Cleanup

---
 lib/Grid_math_types_mapper.h | 100 -----------------------------------
 1 file changed, 100 deletions(-)
 delete mode 100644 lib/Grid_math_types_mapper.h

diff --git a/lib/Grid_math_types_mapper.h b/lib/Grid_math_types_mapper.h
deleted file mode 100644
index 7f6d85f5..00000000
--- a/lib/Grid_math_types_mapper.h
+++ /dev/null
@@ -1,100 +0,0 @@
-#ifndef GRID_MATH_TYPE_MAPPER_H
-#define GRID_MATH_TYPE_MAPPER_H
-
-namespace Grid {
-
-
-//////////////////////////////////////////////////////////////////////////////////
-// Want to recurse: GridTypeMapper<Matrix<vComplexD> >::scalar_type == ComplexD.
-// Use of a helper class like this allows us to template specialise and "dress"
-// other classes such as RealD == double, ComplexD == std::complex<double> with these
-// traits.
-//
-// It is possible that we could do this more elegantly if I introduced a 
-// queryable trait in iScalar, iMatrix and iVector and used the query on vtype in 
-// place of the type mapper?
-//
-// Not sure how to do this, but probably could be done with a research effort
-// to study C++11's type_traits.h file. (std::enable_if<isGridTensorType<vtype> >)
-//
-//////////////////////////////////////////////////////////////////////////////////
-  
-  template <class T> class GridTypeMapper {
-  public:
-    typedef typename T::scalar_type scalar_type;
-    typedef typename T::vector_type vector_type;
-    typedef typename T::tensor_reduced tensor_reduced;
-    enum { TensorLevel = T::TensorLevel };
-  };
-
-//////////////////////////////////////////////////////////////////////////////////
-// Recursion stops with these template specialisations
-//////////////////////////////////////////////////////////////////////////////////
-  template<> class GridTypeMapper<RealF> {
-  public:
-    typedef RealF scalar_type;
-    typedef RealF vector_type;
-    typedef RealF tensor_reduced ;
-    enum { TensorLevel = 0 };
-  };
-  template<> class GridTypeMapper<RealD> {
-  public:
-    typedef RealD scalar_type;
-    typedef RealD vector_type;
-    typedef RealD tensor_reduced;
-    enum { TensorLevel = 0 };
-  };
-  template<> class GridTypeMapper<ComplexF> {
-  public:
-    typedef ComplexF scalar_type;
-    typedef ComplexF vector_type;
-    typedef ComplexF tensor_reduced;
-    enum { TensorLevel = 0 };
-  };
-  template<> class GridTypeMapper<ComplexD> {
-  public:
-    typedef ComplexD scalar_type;
-    typedef ComplexD vector_type;
-    typedef ComplexD tensor_reduced;
-    enum { TensorLevel = 0 };
-  };
-
-  template<> class GridTypeMapper<vRealF> {
-  public:
-    typedef RealF  scalar_type;
-    typedef vRealF vector_type;
-    typedef vRealF tensor_reduced;
-    enum { TensorLevel = 0 };
-  };
-  template<> class GridTypeMapper<vRealD> {
-  public:
-    typedef RealD  scalar_type;
-    typedef vRealD vector_type;
-    typedef vRealD tensor_reduced;
-    enum { TensorLevel = 0 };
-  };
-  template<> class GridTypeMapper<vComplexF> {
-  public:
-    typedef ComplexF  scalar_type;
-    typedef vComplexF vector_type;
-    typedef vComplexF tensor_reduced;
-    enum { TensorLevel = 0 };
-  };
-  template<> class GridTypeMapper<vComplexD> {
-  public:
-    typedef ComplexD  scalar_type;
-    typedef vComplexD vector_type;
-    typedef vComplexD tensor_reduced;
-    enum { TensorLevel = 0 };
-  };
-  template<> class GridTypeMapper<vInteger> {
-  public:
-    typedef Integer  scalar_type;
-    typedef vInteger vector_type;
-    typedef vInteger tensor_reduced;
-    enum { TensorLevel = 0 };
-  };
-
-}
-
-#endif

From 520af214af0f3ec7a11716c17f22af5aa205a968 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 18:47:43 +0100
Subject: [PATCH 073/429] splitting into smaller, multiple files for
 readability and easy find.

---
 lib/Grid_math.h           | 647 +-------------------------------------
 lib/Grid_math_inner.h     |  41 +++
 lib/Grid_math_outer.h     |  47 +++
 lib/Grid_math_peek.h      | 149 +++++++++
 lib/Grid_math_poke.h      | 126 ++++++++
 lib/Grid_math_reality.h   | 115 +++++++
 lib/Grid_math_trace.h     |  95 ++++++
 lib/Grid_math_transpose.h | 101 ++++++
 8 files changed, 681 insertions(+), 640 deletions(-)
 create mode 100644 lib/Grid_math_inner.h
 create mode 100644 lib/Grid_math_outer.h
 create mode 100644 lib/Grid_math_peek.h
 create mode 100644 lib/Grid_math_poke.h
 create mode 100644 lib/Grid_math_reality.h
 create mode 100644 lib/Grid_math_trace.h
 create mode 100644 lib/Grid_math_transpose.h

diff --git a/lib/Grid_math.h b/lib/Grid_math.h
index bad20054..b1e04c19 100644
--- a/lib/Grid_math.h
+++ b/lib/Grid_math.h
@@ -4,646 +4,13 @@
 #include <Grid_math_traits.h>
 #include <Grid_math_tensors.h>
 #include <Grid_math_arith.h>
+#include <Grid_math_inner.h>
+#include <Grid_math_outer.h>
+#include <Grid_math_transpose.h>
+#include <Grid_math_trace.h>
+#include <Grid_math_peek.h>
+#include <Grid_math_poke.h>
+#include <Grid_math_reality.h>
 
-//
-// Indexing; want to be able to dereference and 
-// obtain either an lvalue or an rvalue.
-//
-namespace Grid {
-
-
-
-
-    ///////////////////////////////////////////////////////////////////////////////////////
-    // innerProduct Scalar x Scalar -> Scalar
-    // innerProduct Vector x Vector -> Scalar
-    // innerProduct Matrix x Matrix -> Scalar
-    ///////////////////////////////////////////////////////////////////////////////////////
-    template<class l,class r,int N> inline
-    auto innerProduct (const iVector<l,N>& lhs,const iVector<r,N>& rhs) -> iScalar<decltype(innerProduct(lhs._internal[0],rhs._internal[0]))>
-    {
-        typedef decltype(innerProduct(lhs._internal[0],rhs._internal[0])) ret_t;
-        iScalar<ret_t> ret=zero;
-        for(int c1=0;c1<N;c1++){
-            ret._internal += innerProduct(lhs._internal[c1],rhs._internal[c1]);
-        }
-        return ret;
-    }
-    template<class l,class r,int N> inline
-    auto innerProduct (const iMatrix<l,N>& lhs,const iMatrix<r,N>& rhs) -> iScalar<decltype(innerProduct(lhs._internal[0][0],rhs._internal[0][0]))>
-    {
-        typedef decltype(innerProduct(lhs._internal[0][0],rhs._internal[0][0])) ret_t;
-        iScalar<ret_t> ret=zero;
-        iScalar<ret_t> tmp;
-        for(int c1=0;c1<N;c1++){
-        for(int c2=0;c2<N;c2++){
-	  ret._internal+=innerProduct(lhs._internal[c1][c2],rhs._internal[c1][c2]);
-        }}
-        return ret;
-    }
-    template<class l,class r> inline
-    auto innerProduct (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(innerProduct(lhs._internal,rhs._internal))>
-    {
-        typedef decltype(innerProduct(lhs._internal,rhs._internal)) ret_t;
-        iScalar<ret_t> ret;
-        ret._internal = innerProduct(lhs._internal,rhs._internal);
-        return ret;
-    }
-
-    ///////////////////////////////////////////////////////////////////////////////////////
-    // outerProduct Scalar x Scalar -> Scalar
-    //              Vector x Vector -> Matrix
-    ///////////////////////////////////////////////////////////////////////////////////////
-
-template<class l,class r,int N> inline
-auto outerProduct (const iVector<l,N>& lhs,const iVector<r,N>& rhs) -> iMatrix<decltype(outerProduct(lhs._internal[0],rhs._internal[0])),N>
-{
-    typedef decltype(outerProduct(lhs._internal[0],rhs._internal[0])) ret_t;
-    iMatrix<ret_t,N> ret;
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        ret._internal[c1][c2] = outerProduct(lhs._internal[c1],rhs._internal[c2]);
-    }}
-    return ret;
-}
-template<class l,class r> inline
-auto outerProduct (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(outerProduct(lhs._internal,rhs._internal))>
-{
-    typedef decltype(outerProduct(lhs._internal,rhs._internal)) ret_t;
-    iScalar<ret_t> ret;
-    ret._internal = outerProduct(lhs._internal,rhs._internal);
-    return ret;
-}
-
-inline ComplexF outerProduct(const ComplexF &l, const ComplexF& r)
-{
-  return l*r;
-}
-inline ComplexD outerProduct(const ComplexD &l, const ComplexD& r)
-{
-  return l*r;
-}
-inline RealF outerProduct(const RealF &l, const RealF& r)
-{
-  return l*r;
-}
-inline RealD outerProduct(const RealD &l, const RealD& r)
-{
-  return l*r;
-}
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-    /////////////////////////////////////////// CONJ         ///////////////////////////////////////////
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
- 
-// Conj function for scalar, vector, matrix
-template<class vtype> inline iScalar<vtype> conj(const iScalar<vtype>&r)
-{
-    iScalar<vtype> ret;
-    ret._internal = conj(r._internal);
-    return ret;
-}
-template<class vtype,int N> inline iVector<vtype,N> conj(const iVector<vtype,N>&r)
-{
-  iVector<vtype,N> ret;
-  for(int i=0;i<N;i++){
-    ret._internal[i] = conj(r._internal[i]);
-  }
-  return ret;
-}
-template<class vtype,int N> inline iMatrix<vtype,N> conj(const iMatrix<vtype,N>&r)
-{
-  iMatrix<vtype,N> ret;
-  for(int i=0;i<N;i++){
-  for(int j=0;j<N;j++){
-    ret._internal[i][j] = conj(r._internal[i][j]);
-  }}
-  return ret;
-}
-
-// Adj function for scalar, vector, matrix
-template<class vtype> inline iScalar<vtype> adj(const iScalar<vtype>&r)
-{
-    iScalar<vtype> ret;
-    ret._internal = adj(r._internal);
-    return ret;
-}
-template<class vtype,int N> inline iVector<vtype,N> adj(const iVector<vtype,N>&r)
-{
-    iVector<vtype,N> ret;
-    for(int i=0;i<N;i++){
-        ret._internal[i] = adj(r._internal[i]);
-    }
-    return ret;
-}
-template<class vtype,int N> inline iMatrix<vtype,N> adj(const iMatrix<vtype,N> &arg)
-{
-    iMatrix<vtype,N> ret;
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        ret._internal[c1][c2]=adj(arg._internal[c2][c1]);
-    }}
-    return ret;
-}
-
-
-
-/////////////////////////////////////////////////////////////////
-// Transpose all indices
-/////////////////////////////////////////////////////////////////
-
-inline ComplexD transpose(ComplexD &rhs){  return rhs;}
-inline ComplexF transpose(ComplexF &rhs){  return rhs;}
-inline RealD transpose(RealD &rhs){  return rhs;}
-inline RealF transpose(RealF &rhs){  return rhs;}
-
-template<class vtype,int N>
-  inline typename std::enable_if<isGridTensor<vtype>::value, iMatrix<vtype,N> >::type 
-  transpose(iMatrix<vtype,N> arg)
-  {
-    iMatrix<vtype,N> ret;
-    for(int i=0;i<N;i++){
-      for(int j=0;j<N;j++){
-	ret._internal[i][j] = transpose(arg._internal[j][i]); // NB recurses
-      }}
-    return ret;
-  }
-template<class vtype,int N>
-  inline typename std::enable_if<isGridTensor<vtype>::notvalue, iMatrix<vtype,N> >::type 
-  transpose(iMatrix<vtype,N> arg)
-  {
-    iMatrix<vtype,N> ret;
-    for(int i=0;i<N;i++){
-      for(int j=0;j<N;j++){
-	ret._internal[i][j] = arg._internal[j][i]; // Stop recursion if not a tensor type
-      }}
-    return ret;
-  }
-
-template<class vtype>
-  inline typename std::enable_if<isGridTensor<vtype>::value, iScalar<vtype> >::type 
-  transpose(iScalar<vtype> arg)
-  {
-    iScalar<vtype> ret;
-    ret._internal = transpose(arg._internal); // NB recurses
-    return ret;
-  }
-
-template<class vtype>
-  inline typename std::enable_if<isGridTensor<vtype>::notvalue, iScalar<vtype> >::type 
-  transpose(iScalar<vtype> arg)
-  {
-    iScalar<vtype> ret;
-    ret._internal = arg._internal; // NB recursion stops
-    return ret;
-  }
-
-
-////////////////////////////////////////////////////////////////////////////////////////////
-// Transpose a specific index; instructive to compare this style of recursion termination
-// to that of adj; which is easiers?
-////////////////////////////////////////////////////////////////////////////////////////////
-template<int Level,class vtype,int N> inline 
-  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value, iMatrix<vtype,N> >::type 
-transposeIndex (const iMatrix<vtype,N> &arg)
-{
-  iMatrix<vtype,N> ret;
-  for(int i=0;i<N;i++){
-    for(int j=0;j<N;j++){
-      ret._internal[i][j] = arg._internal[j][i]; 
-  }}
-  return ret;
-}
-// or not
-template<int Level,class vtype,int N> inline 
-typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue, iMatrix<vtype,N> >::type 
-transposeIndex (const iMatrix<vtype,N> &arg)
-{
-  iMatrix<vtype,N> ret;
-  for(int i=0;i<N;i++){
-    for(int j=0;j<N;j++){
-      ret._internal[i][j] = transposeIndex<Level>(arg._internal[i][j]); 
-  }}
-  return ret;
-}
-template<int Level,class vtype> inline 
-typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue, iScalar<vtype> >::type 
-transposeIndex (const iScalar<vtype> &arg)
-{
-  iScalar<vtype> ret;
-  ret._internal=transposeIndex<Level>(arg._internal);
-  return ret;
-}
-template<int Level,class vtype> inline 
-typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value, iScalar<vtype> >::type 
-transposeIndex (const iScalar<vtype> &arg)
-{
-  return arg;
-}
-
-//////////////////////////////////////////////////////////////////
-// Traces: both all indices and a specific index 
-/////////////////////////////////////////////////////////////////
-
-inline ComplexF trace( const ComplexF &arg){    return arg;}
-inline ComplexD trace( const ComplexD &arg){    return arg;}
-inline RealF trace( const RealF &arg){    return arg;}
-inline RealD trace( const RealD &arg){    return arg;}
-
-template<int Level> inline ComplexF traceIndex(const ComplexF arg) { return arg;}
-template<int Level> inline ComplexD traceIndex(const ComplexD arg) { return arg;}
-template<int Level> inline RealF traceIndex(const RealF arg) { return arg;}
-template<int Level> inline RealD traceIndex(const RealD arg) { return arg;}
-
-template<class vtype,int N>
-inline auto trace(const iMatrix<vtype,N> &arg) -> iScalar<decltype(trace(arg._internal[0][0]))>
-{
-    iScalar<decltype( trace(arg._internal[0][0] )) > ret;
-    zeroit(ret._internal);
-    for(int i=0;i<N;i++){
-        ret._internal=ret._internal+trace(arg._internal[i][i]);
-    }
-    return ret;
-}
-template<class vtype>
-inline auto trace(const iScalar<vtype> &arg) -> iScalar<decltype(trace(arg._internal))>
-{
-    iScalar<decltype(trace(arg._internal))> ret;
-    ret._internal=trace(arg._internal);
-    return ret;
-}
-////////////////////////////////////////////////////////////////////////////////////////////////////////
-// Trace Specific indices.
-////////////////////////////////////////////////////////////////////////////////////////////////////////
-/*
-template<int Level,class vtype> inline 
-auto traceIndex(const iScalar<vtype> &arg) -> iScalar<decltype(traceIndex<Level>(arg._internal)) >
-{
-  iScalar<decltype(traceIndex<Level>(arg._internal))> ret;
-  ret._internal = traceIndex<Level>(arg._internal);
-  return ret;
-}
-*/
-template<int Level,class vtype> inline auto 
-traceIndex (const iScalar<vtype> &arg) ->
-typename 
-std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue, 
-  iScalar<decltype(traceIndex<Level>(arg._internal))> >::type 
-
-{
-  iScalar<decltype(traceIndex<Level>(arg._internal))> ret;
-  ret._internal=traceIndex<Level>(arg._internal);
-  return ret;
-}
-template<int Level,class vtype> inline auto
-traceIndex (const iScalar<vtype> &arg) ->
-typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value, 
-                        iScalar<vtype> >::type 
-{
-  return arg;
-}
-
-// If we hit the right index, return scalar and trace it with no further recursion
-template<int Level,class vtype,int N> inline 
-auto traceIndex(const iMatrix<vtype,N> &arg) ->
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value,  // Index matches
-                                                    iScalar<vtype> >::type                              // return scalar
-{
-  iScalar<vtype> ret;
-  zeroit(ret._internal);
-  for(int i=0;i<N;i++){
-    ret._internal = ret._internal + arg._internal[i][i];
-  }
-  return ret;
-}
-
-// not this level, so recurse
-template<int Level,class vtype,int N> inline 
-auto traceIndex(const iMatrix<vtype,N> &arg) ->
-  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,// No index match
-         iMatrix<decltype(traceIndex<Level>(arg._internal[0][0])),N> >::type     // return matrix
-{
-  iMatrix<decltype(traceIndex<Level>(arg._internal[0][0])),N> ret;
-  for(int i=0;i<N;i++){
-  for(int j=0;j<N;j++){
-    ret._internal[i][j] = traceIndex<Level>(arg._internal[i][j]);
-  }}
-  return ret;
-}
-
-//////////////////////////////////////////////////////////////////////////////
-// Peek on a specific index; returns a scalar in that index, tensor inherits rest
-//////////////////////////////////////////////////////////////////////////////
-// If we hit the right index, return scalar with no further recursion
-
-//template<int Level> inline ComplexF peekIndex(const ComplexF arg) { return arg;}
-//template<int Level> inline ComplexD peekIndex(const ComplexD arg) { return arg;}
-//template<int Level> inline RealF peekIndex(const RealF arg) { return arg;}
-//template<int Level> inline RealD peekIndex(const RealD arg) { return arg;}
-
-// Scalar peek, no indices
-template<int Level,class vtype> inline 
-  auto peekIndex(const iScalar<vtype> &arg) -> 
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value,  // Index matches
-  iScalar<vtype> >::type                              // return scalar
-{
-  return arg;
-}
-// Vector peek, one index
-template<int Level,class vtype,int N> inline 
-  auto peekIndex(const iVector<vtype,N> &arg,int i) -> 
-  typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::value,  // Index matches
-  iScalar<vtype> >::type                              // return scalar
-{
-  iScalar<vtype> ret;                              // return scalar
-  ret._internal = arg._internal[i];
-  return ret;
-}
-// Matrix peek, two indices
-template<int Level,class vtype,int N> inline 
-  auto peekIndex(const iMatrix<vtype,N> &arg,int i,int j) -> 
-  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value,  // Index matches
-  iScalar<vtype> >::type                              // return scalar
-{
-  iScalar<vtype> ret;                              // return scalar
-  ret._internal = arg._internal[i][j];
-  return ret;
-}
-
-/////////////
-// No match peek for scalar,vector,matrix must forward on either 0,1,2 args. Must have 9 routines with notvalue
-/////////////
-// scalar
-template<int Level,class vtype> inline 
-  auto peekIndex(const iScalar<vtype> &arg) ->                                     // Scalar 0 index  
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does NOT match
-  iScalar<decltype(peekIndex<Level>(arg._internal))> >::type                       
-{
-  iScalar<decltype(peekIndex<Level>(arg._internal))> ret;
-  ret._internal= peekIndex<Level>(arg._internal);
-  return ret;
-}
-template<int Level,class vtype> inline 
-  auto peekIndex(const iScalar<vtype> &arg,int i) ->                             // Scalar 1 index
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does NOT match
-  iScalar<decltype(peekIndex<Level>(arg._internal,i))> >::type                       
-{
-  iScalar<decltype(peekIndex<Level>(arg._internal,i))> ret;
-  ret._internal=peekIndex<Level>(arg._internal,i);
-  return ret;
-}
-template<int Level,class vtype> inline 
-  auto peekIndex(const iScalar<vtype> &arg,int i,int j) ->                         // Scalar, 2 index
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does NOT match
-  iScalar<decltype(peekIndex<Level>(arg._internal,i,j))> >::type                       
-{
-  iScalar<decltype(peekIndex<Level>(arg._internal,i,j))> ret;
-  ret._internal=peekIndex<Level>(arg._internal,i,j);
-  return ret;
-}
-// vector
-template<int Level,class vtype,int N> inline 
-auto peekIndex(const iVector<vtype,N> &arg) -> 
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does not match
-  iVector<decltype(peekIndex<Level>(arg._internal[0])),N> >::type                       
-{
-  iVector<decltype(peekIndex<Level>(arg._internal[0])),N> ret;
-  for(int ii=0;ii<N;ii++){
-    ret._internal[ii]=peekIndex<Level>(arg._internal[ii]);
-  }
-  return ret;
-}
-template<int Level,class vtype,int N> inline 
-  auto peekIndex(const iVector<vtype,N> &arg,int i) -> 
-  typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,  // Index does not match
-  iVector<decltype(peekIndex<Level>(arg._internal[0],i)),N> >::type                       
-{
-  iVector<decltype(peekIndex<Level>(arg._internal[0],i)),N> ret;
-  for(int ii=0;ii<N;ii++){
-    ret._internal[ii]=peekIndex<Level>(arg._internal[ii],i);
-  }
-  return ret;
-}
-template<int Level,class vtype,int N> inline 
-  auto peekIndex(const iVector<vtype,N> &arg,int i,int j) -> 
-  typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,  // Index does not match
-  iVector<decltype(peekIndex<Level>(arg._internal[0],i,j)),N> >::type                       
-{
-  iVector<decltype(peekIndex<Level>(arg._internal[0],i,j)),N> ret;
-  for(int ii=0;ii<N;ii++){
-    ret._internal[ii]=peekIndex<Level>(arg._internal[ii],i,j);
-  }
-  return ret;
-}
-// matrix
-template<int Level,class vtype,int N> inline 
-auto peekIndex(const iMatrix<vtype,N> &arg) -> 
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does not match
-  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0])),N> >::type                       
-{
-  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0])),N> ret;
-  for(int ii=0;ii<N;ii++){
-  for(int jj=0;jj<N;jj++){
-    ret._internal[ii][jj]=peekIndex<Level>(arg._internal[ii][jj]);// Could avoid this because peeking a scalar is dumb
-  }}
-  return ret;
-}
-template<int Level,class vtype,int N> inline 
-  auto peekIndex(const iMatrix<vtype,N> &arg,int i) -> 
-  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,  // Index does not match
-  iMatrix<decltype(peekIndex<Level>(arg._internal[0],i)),N> >::type                       
-{
-  iMatrix<decltype(peekIndex<Level>(arg._internal[0],i)),N> ret;
-  for(int ii=0;ii<N;ii++){
-  for(int jj=0;jj<N;jj++){
-    ret._internal[ii][jj]=peekIndex<Level>(arg._internal[ii][jj],i);
-  }}
-  return ret;
-}
-template<int Level,class vtype,int N> inline 
-  auto peekIndex(const iMatrix<vtype,N> &arg,int i,int j) -> 
-  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,  // Index does not match
-  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0],i,j)),N> >::type                       
-{
-  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0],i,j)),N> ret;
-  for(int ii=0;ii<N;ii++){
-  for(int jj=0;jj<N;jj++){
-    ret._internal[ii][jj]=peekIndex<Level>(arg._internal[ii][jj],i,j);
-  }}
-  return ret;
-}
-
-
-//////////////////////////////////////////////////////////////////////////////
-// Poke a specific index; 
-//////////////////////////////////////////////////////////////////////////////
-
-// Scalar poke
-template<int Level,class vtype> inline 
-  void pokeIndex(iScalar<vtype> &ret, 
-		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value,iScalar<vtype> >::type &arg)
-{
-  ret._internal = arg._internal;
-}
-// Vector poke, one index
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iVector<vtype,N> &ret, 
-		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::value,iScalar<vtype> >::type &arg,int i)
-{
-  ret._internal[i] = arg._internal;
-}
-// Vector poke, two indices
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iMatrix<vtype,N> &ret, 
-		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value,iScalar<vtype> >::type &arg,int i,int j)
-{
-  ret._internal[i][j] = arg._internal;
-}
-
-/////////////
-// No match poke for scalar,vector,matrix must forward on either 0,1,2 args. Must have 9 routines with notvalue
-/////////////
-// scalar
-template<int Level,class vtype> inline 
-  void pokeIndex(iScalar<vtype> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,iScalar<decltype(peekIndex<Level>(ret._internal))> >::type &arg)
-		 
-{
-  pokeIndex<Level>(ret._internal,arg._internal);
-}
-template<int Level,class vtype> inline 
-  void pokeIndex(iScalar<vtype> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,iScalar<decltype(peekIndex<Level>(ret._internal,0))> >::type &arg,
-		 int i)
-		 
-{
-  pokeIndex<Level>(ret._internal,arg._internal,i);
-}
-template<int Level,class vtype> inline 
-  void pokeIndex(iScalar<vtype> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,iScalar<decltype(peekIndex<Level>(ret._internal,0,0))> >::type &arg,
-		 int i,int j)
-		 
-{
-  pokeIndex<Level>(ret._internal,arg._internal,i,j);
-}
-
-// Vector
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iVector<vtype,N> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,iVector<decltype(peekIndex<Level>(ret._internal)),N> >::type &arg)
-		 
-{
-  for(int ii=0;ii<N;ii++){
-    pokeIndex<Level>(ret._internal[ii],arg._internal[ii]);
-  }
-}
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iVector<vtype,N> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,iVector<decltype(peekIndex<Level>(ret._internal,0)),N> >::type &arg,
-		 int i)
-		 
-{
-  for(int ii=0;ii<N;ii++){
-    pokeIndex<Level>(ret._internal[ii],arg._internal[ii],i);
-  }
-}
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iVector<vtype,N> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,iVector<decltype(peekIndex<Level>(ret._internal,0,0)),N> >::type &arg,
-		 int i,int j)
-		 
-{
-  for(int ii=0;ii<N;ii++){
-    pokeIndex<Level>(ret._internal[ii],arg._internal[ii],i,j);
-  }
-}
-
-// Matrix
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iMatrix<vtype,N> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,iMatrix<decltype(peekIndex<Level>(ret._internal)),N> >::type &arg)
-		 
-{
-  for(int ii=0;ii<N;ii++){
-  for(int jj=0;jj<N;jj++){
-    pokeIndex<Level>(ret._internal[ii][jj],arg._internal[ii][jj]);
-  }}
-}
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iMatrix<vtype,N> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,iMatrix<decltype(peekIndex<Level>(ret._internal,0)),N> >::type &arg,
-		 int i)
-		 
-{
-  for(int ii=0;ii<N;ii++){
-  for(int jj=0;jj<N;jj++){
-    pokeIndex<Level>(ret._internal[ii][jj],arg._internal[ii][jj],i);
-  }}
-}
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iMatrix<vtype,N> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,iMatrix<decltype(peekIndex<Level>(ret._internal,0,0)),N> >::type &arg,
-		 int i,int j)
-		 
-{
-  for(int ii=0;ii<N;ii++){
-  for(int jj=0;jj<N;jj++){
-    pokeIndex<Level>(ret._internal[ii][jj],arg._internal[ii][jj],i,j);
-  }}
-}
-
-/////////////////////////////////////////////////////////////////
-// Can only take the real/imag part of scalar objects, since
-// lattice objects of different complex nature are non-conformable.
-/////////////////////////////////////////////////////////////////
-template<class itype> inline auto real(const iScalar<itype> &z) -> iScalar<decltype(real(z._internal))>
-{
-    iScalar<decltype(real(z._internal))> ret;
-    ret._internal = real(z._internal);
-    return ret;
-}
-template<class itype,int N> inline auto real(const iMatrix<itype,N> &z) -> iMatrix<decltype(real(z._internal[0][0])),N>
-{
-    iMatrix<decltype(real(z._internal[0][0])),N> ret;
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        ret._internal[c1][c2] = real(z._internal[c1][c2]);
-    }}
-    return ret;
-}
-template<class itype,int N> inline auto real(const iVector<itype,N> &z) -> iVector<decltype(real(z._internal[0])),N>
-{
-    iVector<decltype(real(z._internal[0])),N> ret;
-    for(int c1=0;c1<N;c1++){
-        ret._internal[c1] = real(z._internal[c1]);
-    }
-    return ret;
-}
-    
-template<class itype> inline auto imag(const iScalar<itype> &z) -> iScalar<decltype(imag(z._internal))>
-{
-    iScalar<decltype(imag(z._internal))> ret;
-    ret._internal = imag(z._internal);
-    return ret;
-}
-template<class itype,int N> inline auto imag(const iMatrix<itype,N> &z) -> iMatrix<decltype(imag(z._internal[0][0])),N>
-{
-    iMatrix<decltype(imag(z._internal[0][0])),N> ret;
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        ret._internal[c1][c2] = imag(z._internal[c1][c2]);
-    }}
-    return ret;
-}
-template<class itype,int N> inline auto imag(const iVector<itype,N> &z) -> iVector<decltype(imag(z._internal[0])),N>
-{
-    iVector<decltype(imag(z._internal[0])),N> ret;
-    for(int c1=0;c1<N;c1++){
-        ret._internal[c1] = imag(z._internal[c1]);
-    }
-    return ret;
-}
-
-};
     
 #endif
diff --git a/lib/Grid_math_inner.h b/lib/Grid_math_inner.h
new file mode 100644
index 00000000..88387db7
--- /dev/null
+++ b/lib/Grid_math_inner.h
@@ -0,0 +1,41 @@
+#ifndef GRID_MATH_INNER_H
+#define GRID_MATH_INNER_H
+namespace Grid {
+    ///////////////////////////////////////////////////////////////////////////////////////
+    // innerProduct Scalar x Scalar -> Scalar
+    // innerProduct Vector x Vector -> Scalar
+    // innerProduct Matrix x Matrix -> Scalar
+    ///////////////////////////////////////////////////////////////////////////////////////
+    template<class l,class r,int N> inline
+    auto innerProduct (const iVector<l,N>& lhs,const iVector<r,N>& rhs) -> iScalar<decltype(innerProduct(lhs._internal[0],rhs._internal[0]))>
+    {
+        typedef decltype(innerProduct(lhs._internal[0],rhs._internal[0])) ret_t;
+        iScalar<ret_t> ret=zero;
+        for(int c1=0;c1<N;c1++){
+            ret._internal += innerProduct(lhs._internal[c1],rhs._internal[c1]);
+        }
+        return ret;
+    }
+    template<class l,class r,int N> inline
+    auto innerProduct (const iMatrix<l,N>& lhs,const iMatrix<r,N>& rhs) -> iScalar<decltype(innerProduct(lhs._internal[0][0],rhs._internal[0][0]))>
+    {
+        typedef decltype(innerProduct(lhs._internal[0][0],rhs._internal[0][0])) ret_t;
+        iScalar<ret_t> ret=zero;
+        iScalar<ret_t> tmp;
+        for(int c1=0;c1<N;c1++){
+        for(int c2=0;c2<N;c2++){
+	  ret._internal+=innerProduct(lhs._internal[c1][c2],rhs._internal[c1][c2]);
+        }}
+        return ret;
+    }
+    template<class l,class r> inline
+    auto innerProduct (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(innerProduct(lhs._internal,rhs._internal))>
+    {
+        typedef decltype(innerProduct(lhs._internal,rhs._internal)) ret_t;
+        iScalar<ret_t> ret;
+        ret._internal = innerProduct(lhs._internal,rhs._internal);
+        return ret;
+    }
+
+}
+#endif
diff --git a/lib/Grid_math_outer.h b/lib/Grid_math_outer.h
new file mode 100644
index 00000000..816f6ad8
--- /dev/null
+++ b/lib/Grid_math_outer.h
@@ -0,0 +1,47 @@
+#ifndef GRID_MATH_OUTER_H
+#define GRID_MATH_OUTER_H
+namespace Grid {
+    ///////////////////////////////////////////////////////////////////////////////////////
+    // outerProduct Scalar x Scalar -> Scalar
+    //              Vector x Vector -> Matrix
+    ///////////////////////////////////////////////////////////////////////////////////////
+
+template<class l,class r,int N> inline
+auto outerProduct (const iVector<l,N>& lhs,const iVector<r,N>& rhs) -> iMatrix<decltype(outerProduct(lhs._internal[0],rhs._internal[0])),N>
+{
+    typedef decltype(outerProduct(lhs._internal[0],rhs._internal[0])) ret_t;
+    iMatrix<ret_t,N> ret;
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        ret._internal[c1][c2] = outerProduct(lhs._internal[c1],rhs._internal[c2]);
+    }}
+    return ret;
+}
+template<class l,class r> inline
+auto outerProduct (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(outerProduct(lhs._internal,rhs._internal))>
+{
+    typedef decltype(outerProduct(lhs._internal,rhs._internal)) ret_t;
+    iScalar<ret_t> ret;
+    ret._internal = outerProduct(lhs._internal,rhs._internal);
+    return ret;
+}
+
+inline ComplexF outerProduct(const ComplexF &l, const ComplexF& r)
+{
+  return l*r;
+}
+inline ComplexD outerProduct(const ComplexD &l, const ComplexD& r)
+{
+  return l*r;
+}
+inline RealF outerProduct(const RealF &l, const RealF& r)
+{
+  return l*r;
+}
+inline RealD outerProduct(const RealD &l, const RealD& r)
+{
+  return l*r;
+}
+
+}
+#endif
diff --git a/lib/Grid_math_peek.h b/lib/Grid_math_peek.h
new file mode 100644
index 00000000..b76fa84d
--- /dev/null
+++ b/lib/Grid_math_peek.h
@@ -0,0 +1,149 @@
+#ifndef GRID_MATH_PEEK_H
+#define GRID_MATH_PEEK_H
+namespace Grid {
+
+//////////////////////////////////////////////////////////////////////////////
+// Peek on a specific index; returns a scalar in that index, tensor inherits rest
+//////////////////////////////////////////////////////////////////////////////
+// If we hit the right index, return scalar with no further recursion
+
+//template<int Level> inline ComplexF peekIndex(const ComplexF arg) { return arg;}
+//template<int Level> inline ComplexD peekIndex(const ComplexD arg) { return arg;}
+//template<int Level> inline RealF peekIndex(const RealF arg) { return arg;}
+//template<int Level> inline RealD peekIndex(const RealD arg) { return arg;}
+
+// Scalar peek, no indices
+template<int Level,class vtype> inline 
+  auto peekIndex(const iScalar<vtype> &arg) -> 
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value,  // Index matches
+  iScalar<vtype> >::type                              // return scalar
+{
+  return arg;
+}
+// Vector peek, one index
+template<int Level,class vtype,int N> inline 
+  auto peekIndex(const iVector<vtype,N> &arg,int i) -> 
+  typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::value,  // Index matches
+  iScalar<vtype> >::type                              // return scalar
+{
+  iScalar<vtype> ret;                              // return scalar
+  ret._internal = arg._internal[i];
+  return ret;
+}
+// Matrix peek, two indices
+template<int Level,class vtype,int N> inline 
+  auto peekIndex(const iMatrix<vtype,N> &arg,int i,int j) -> 
+  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value,  // Index matches
+  iScalar<vtype> >::type                              // return scalar
+{
+  iScalar<vtype> ret;                              // return scalar
+  ret._internal = arg._internal[i][j];
+  return ret;
+}
+
+/////////////
+// No match peek for scalar,vector,matrix must forward on either 0,1,2 args. Must have 9 routines with notvalue
+/////////////
+// scalar
+template<int Level,class vtype> inline 
+  auto peekIndex(const iScalar<vtype> &arg) ->                                     // Scalar 0 index  
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does NOT match
+  iScalar<decltype(peekIndex<Level>(arg._internal))> >::type                       
+{
+  iScalar<decltype(peekIndex<Level>(arg._internal))> ret;
+  ret._internal= peekIndex<Level>(arg._internal);
+  return ret;
+}
+template<int Level,class vtype> inline 
+  auto peekIndex(const iScalar<vtype> &arg,int i) ->                             // Scalar 1 index
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does NOT match
+  iScalar<decltype(peekIndex<Level>(arg._internal,i))> >::type                       
+{
+  iScalar<decltype(peekIndex<Level>(arg._internal,i))> ret;
+  ret._internal=peekIndex<Level>(arg._internal,i);
+  return ret;
+}
+template<int Level,class vtype> inline 
+  auto peekIndex(const iScalar<vtype> &arg,int i,int j) ->                         // Scalar, 2 index
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does NOT match
+  iScalar<decltype(peekIndex<Level>(arg._internal,i,j))> >::type                       
+{
+  iScalar<decltype(peekIndex<Level>(arg._internal,i,j))> ret;
+  ret._internal=peekIndex<Level>(arg._internal,i,j);
+  return ret;
+}
+// vector
+template<int Level,class vtype,int N> inline 
+auto peekIndex(const iVector<vtype,N> &arg) -> 
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does not match
+  iVector<decltype(peekIndex<Level>(arg._internal[0])),N> >::type                       
+{
+  iVector<decltype(peekIndex<Level>(arg._internal[0])),N> ret;
+  for(int ii=0;ii<N;ii++){
+    ret._internal[ii]=peekIndex<Level>(arg._internal[ii]);
+  }
+  return ret;
+}
+template<int Level,class vtype,int N> inline 
+  auto peekIndex(const iVector<vtype,N> &arg,int i) -> 
+  typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,  // Index does not match
+  iVector<decltype(peekIndex<Level>(arg._internal[0],i)),N> >::type                       
+{
+  iVector<decltype(peekIndex<Level>(arg._internal[0],i)),N> ret;
+  for(int ii=0;ii<N;ii++){
+    ret._internal[ii]=peekIndex<Level>(arg._internal[ii],i);
+  }
+  return ret;
+}
+template<int Level,class vtype,int N> inline 
+  auto peekIndex(const iVector<vtype,N> &arg,int i,int j) -> 
+  typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,  // Index does not match
+  iVector<decltype(peekIndex<Level>(arg._internal[0],i,j)),N> >::type                       
+{
+  iVector<decltype(peekIndex<Level>(arg._internal[0],i,j)),N> ret;
+  for(int ii=0;ii<N;ii++){
+    ret._internal[ii]=peekIndex<Level>(arg._internal[ii],i,j);
+  }
+  return ret;
+}
+// matrix
+template<int Level,class vtype,int N> inline 
+auto peekIndex(const iMatrix<vtype,N> &arg) -> 
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does not match
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0])),N> >::type                       
+{
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0])),N> ret;
+  for(int ii=0;ii<N;ii++){
+  for(int jj=0;jj<N;jj++){
+    ret._internal[ii][jj]=peekIndex<Level>(arg._internal[ii][jj]);// Could avoid this because peeking a scalar is dumb
+  }}
+  return ret;
+}
+template<int Level,class vtype,int N> inline 
+  auto peekIndex(const iMatrix<vtype,N> &arg,int i) -> 
+  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,  // Index does not match
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0],i)),N> >::type                       
+{
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0],i)),N> ret;
+  for(int ii=0;ii<N;ii++){
+  for(int jj=0;jj<N;jj++){
+    ret._internal[ii][jj]=peekIndex<Level>(arg._internal[ii][jj],i);
+  }}
+  return ret;
+}
+template<int Level,class vtype,int N> inline 
+  auto peekIndex(const iMatrix<vtype,N> &arg,int i,int j) -> 
+  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,  // Index does not match
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0],i,j)),N> >::type                       
+{
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0],i,j)),N> ret;
+  for(int ii=0;ii<N;ii++){
+  for(int jj=0;jj<N;jj++){
+    ret._internal[ii][jj]=peekIndex<Level>(arg._internal[ii][jj],i,j);
+  }}
+  return ret;
+}
+
+
+}
+#endif
diff --git a/lib/Grid_math_poke.h b/lib/Grid_math_poke.h
new file mode 100644
index 00000000..3f1cb89e
--- /dev/null
+++ b/lib/Grid_math_poke.h
@@ -0,0 +1,126 @@
+#ifndef GRID_MATH_POKE_H
+#define GRID_MATH_POKE_H
+namespace Grid {
+
+//////////////////////////////////////////////////////////////////////////////
+// Poke a specific index; 
+//////////////////////////////////////////////////////////////////////////////
+
+// Scalar poke
+template<int Level,class vtype> inline 
+  void pokeIndex(iScalar<vtype> &ret, 
+		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value,iScalar<vtype> >::type &arg)
+{
+  ret._internal = arg._internal;
+}
+// Vector poke, one index
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iVector<vtype,N> &ret, 
+		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::value,iScalar<vtype> >::type &arg,int i)
+{
+  ret._internal[i] = arg._internal;
+}
+// Vector poke, two indices
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iMatrix<vtype,N> &ret, 
+		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value,iScalar<vtype> >::type &arg,int i,int j)
+{
+  ret._internal[i][j] = arg._internal;
+}
+
+/////////////
+// No match poke for scalar,vector,matrix must forward on either 0,1,2 args. Must have 9 routines with notvalue
+/////////////
+// scalar
+template<int Level,class vtype> inline 
+  void pokeIndex(iScalar<vtype> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,iScalar<decltype(peekIndex<Level>(ret._internal))> >::type &arg)
+		 
+{
+  pokeIndex<Level>(ret._internal,arg._internal);
+}
+template<int Level,class vtype> inline 
+  void pokeIndex(iScalar<vtype> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,iScalar<decltype(peekIndex<Level>(ret._internal,0))> >::type &arg,
+		 int i)
+		 
+{
+  pokeIndex<Level>(ret._internal,arg._internal,i);
+}
+template<int Level,class vtype> inline 
+  void pokeIndex(iScalar<vtype> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,iScalar<decltype(peekIndex<Level>(ret._internal,0,0))> >::type &arg,
+		 int i,int j)
+		 
+{
+  pokeIndex<Level>(ret._internal,arg._internal,i,j);
+}
+
+// Vector
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iVector<vtype,N> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,iVector<decltype(peekIndex<Level>(ret._internal)),N> >::type &arg)
+		 
+{
+  for(int ii=0;ii<N;ii++){
+    pokeIndex<Level>(ret._internal[ii],arg._internal[ii]);
+  }
+}
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iVector<vtype,N> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,iVector<decltype(peekIndex<Level>(ret._internal,0)),N> >::type &arg,
+		 int i)
+		 
+{
+  for(int ii=0;ii<N;ii++){
+    pokeIndex<Level>(ret._internal[ii],arg._internal[ii],i);
+  }
+}
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iVector<vtype,N> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,iVector<decltype(peekIndex<Level>(ret._internal,0,0)),N> >::type &arg,
+		 int i,int j)
+		 
+{
+  for(int ii=0;ii<N;ii++){
+    pokeIndex<Level>(ret._internal[ii],arg._internal[ii],i,j);
+  }
+}
+
+// Matrix
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iMatrix<vtype,N> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,iMatrix<decltype(peekIndex<Level>(ret._internal)),N> >::type &arg)
+		 
+{
+  for(int ii=0;ii<N;ii++){
+  for(int jj=0;jj<N;jj++){
+    pokeIndex<Level>(ret._internal[ii][jj],arg._internal[ii][jj]);
+  }}
+}
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iMatrix<vtype,N> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,iMatrix<decltype(peekIndex<Level>(ret._internal,0)),N> >::type &arg,
+		 int i)
+		 
+{
+  for(int ii=0;ii<N;ii++){
+  for(int jj=0;jj<N;jj++){
+    pokeIndex<Level>(ret._internal[ii][jj],arg._internal[ii][jj],i);
+  }}
+}
+template<int Level,class vtype,int N> inline 
+  void pokeIndex(iMatrix<vtype,N> &ret,
+		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,iMatrix<decltype(peekIndex<Level>(ret._internal,0,0)),N> >::type &arg,
+		 int i,int j)
+		 
+{
+  for(int ii=0;ii<N;ii++){
+  for(int jj=0;jj<N;jj++){
+    pokeIndex<Level>(ret._internal[ii][jj],arg._internal[ii][jj],i,j);
+  }}
+}
+
+
+}
+#endif
diff --git a/lib/Grid_math_reality.h b/lib/Grid_math_reality.h
new file mode 100644
index 00000000..1f676c11
--- /dev/null
+++ b/lib/Grid_math_reality.h
@@ -0,0 +1,115 @@
+#ifndef GRID_MATH_REALITY_H
+#define GRID_MATH_REALITY_H
+namespace Grid {
+
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    /////////////////////////////////////////// CONJ         ///////////////////////////////////////////
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+ 
+// Conj function for scalar, vector, matrix
+template<class vtype> inline iScalar<vtype> conj(const iScalar<vtype>&r)
+{
+    iScalar<vtype> ret;
+    ret._internal = conj(r._internal);
+    return ret;
+}
+template<class vtype,int N> inline iVector<vtype,N> conj(const iVector<vtype,N>&r)
+{
+  iVector<vtype,N> ret;
+  for(int i=0;i<N;i++){
+    ret._internal[i] = conj(r._internal[i]);
+  }
+  return ret;
+}
+template<class vtype,int N> inline iMatrix<vtype,N> conj(const iMatrix<vtype,N>&r)
+{
+  iMatrix<vtype,N> ret;
+  for(int i=0;i<N;i++){
+  for(int j=0;j<N;j++){
+    ret._internal[i][j] = conj(r._internal[i][j]);
+  }}
+  return ret;
+}
+
+// Adj function for scalar, vector, matrix
+template<class vtype> inline iScalar<vtype> adj(const iScalar<vtype>&r)
+{
+    iScalar<vtype> ret;
+    ret._internal = adj(r._internal);
+    return ret;
+}
+template<class vtype,int N> inline iVector<vtype,N> adj(const iVector<vtype,N>&r)
+{
+    iVector<vtype,N> ret;
+    for(int i=0;i<N;i++){
+        ret._internal[i] = adj(r._internal[i]);
+    }
+    return ret;
+}
+template<class vtype,int N> inline iMatrix<vtype,N> adj(const iMatrix<vtype,N> &arg)
+{
+    iMatrix<vtype,N> ret;
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        ret._internal[c1][c2]=adj(arg._internal[c2][c1]);
+    }}
+    return ret;
+}
+
+
+
+/////////////////////////////////////////////////////////////////
+// Can only take the real/imag part of scalar objects, since
+// lattice objects of different complex nature are non-conformable.
+/////////////////////////////////////////////////////////////////
+template<class itype> inline auto real(const iScalar<itype> &z) -> iScalar<decltype(real(z._internal))>
+{
+    iScalar<decltype(real(z._internal))> ret;
+    ret._internal = real(z._internal);
+    return ret;
+}
+template<class itype,int N> inline auto real(const iMatrix<itype,N> &z) -> iMatrix<decltype(real(z._internal[0][0])),N>
+{
+    iMatrix<decltype(real(z._internal[0][0])),N> ret;
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        ret._internal[c1][c2] = real(z._internal[c1][c2]);
+    }}
+    return ret;
+}
+template<class itype,int N> inline auto real(const iVector<itype,N> &z) -> iVector<decltype(real(z._internal[0])),N>
+{
+    iVector<decltype(real(z._internal[0])),N> ret;
+    for(int c1=0;c1<N;c1++){
+        ret._internal[c1] = real(z._internal[c1]);
+    }
+    return ret;
+}
+    
+template<class itype> inline auto imag(const iScalar<itype> &z) -> iScalar<decltype(imag(z._internal))>
+{
+    iScalar<decltype(imag(z._internal))> ret;
+    ret._internal = imag(z._internal);
+    return ret;
+}
+template<class itype,int N> inline auto imag(const iMatrix<itype,N> &z) -> iMatrix<decltype(imag(z._internal[0][0])),N>
+{
+    iMatrix<decltype(imag(z._internal[0][0])),N> ret;
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        ret._internal[c1][c2] = imag(z._internal[c1][c2]);
+    }}
+    return ret;
+}
+template<class itype,int N> inline auto imag(const iVector<itype,N> &z) -> iVector<decltype(imag(z._internal[0])),N>
+{
+    iVector<decltype(imag(z._internal[0])),N> ret;
+    for(int c1=0;c1<N;c1++){
+        ret._internal[c1] = imag(z._internal[c1]);
+    }
+    return ret;
+}
+
+
+}
+#endif
diff --git a/lib/Grid_math_trace.h b/lib/Grid_math_trace.h
new file mode 100644
index 00000000..db8202e6
--- /dev/null
+++ b/lib/Grid_math_trace.h
@@ -0,0 +1,95 @@
+#ifndef GRID_MATH_TRACE_H
+#define GRID_MATH_TRACE_H
+namespace Grid {
+//////////////////////////////////////////////////////////////////
+// Traces: both all indices and a specific index 
+/////////////////////////////////////////////////////////////////
+
+inline ComplexF trace( const ComplexF &arg){    return arg;}
+inline ComplexD trace( const ComplexD &arg){    return arg;}
+inline RealF trace( const RealF &arg){    return arg;}
+inline RealD trace( const RealD &arg){    return arg;}
+
+template<int Level> inline ComplexF traceIndex(const ComplexF arg) { return arg;}
+template<int Level> inline ComplexD traceIndex(const ComplexD arg) { return arg;}
+template<int Level> inline RealF traceIndex(const RealF arg) { return arg;}
+template<int Level> inline RealD traceIndex(const RealD arg) { return arg;}
+
+template<class vtype,int N>
+inline auto trace(const iMatrix<vtype,N> &arg) -> iScalar<decltype(trace(arg._internal[0][0]))>
+{
+    iScalar<decltype( trace(arg._internal[0][0] )) > ret;
+    zeroit(ret._internal);
+    for(int i=0;i<N;i++){
+        ret._internal=ret._internal+trace(arg._internal[i][i]);
+    }
+    return ret;
+}
+template<class vtype>
+inline auto trace(const iScalar<vtype> &arg) -> iScalar<decltype(trace(arg._internal))>
+{
+    iScalar<decltype(trace(arg._internal))> ret;
+    ret._internal=trace(arg._internal);
+    return ret;
+}
+////////////////////////////////////////////////////////////////////////////////////////////////////////
+// Trace Specific indices.
+////////////////////////////////////////////////////////////////////////////////////////////////////////
+/*
+template<int Level,class vtype> inline 
+auto traceIndex(const iScalar<vtype> &arg) -> iScalar<decltype(traceIndex<Level>(arg._internal)) >
+{
+  iScalar<decltype(traceIndex<Level>(arg._internal))> ret;
+  ret._internal = traceIndex<Level>(arg._internal);
+  return ret;
+}
+*/
+template<int Level,class vtype> inline auto 
+traceIndex (const iScalar<vtype> &arg) ->
+typename 
+std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue, 
+  iScalar<decltype(traceIndex<Level>(arg._internal))> >::type 
+
+{
+  iScalar<decltype(traceIndex<Level>(arg._internal))> ret;
+  ret._internal=traceIndex<Level>(arg._internal);
+  return ret;
+}
+template<int Level,class vtype> inline auto
+traceIndex (const iScalar<vtype> &arg) ->
+typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value, 
+                        iScalar<vtype> >::type 
+{
+  return arg;
+}
+
+// If we hit the right index, return scalar and trace it with no further recursion
+template<int Level,class vtype,int N> inline 
+auto traceIndex(const iMatrix<vtype,N> &arg) ->
+  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value,  // Index matches
+                                                    iScalar<vtype> >::type                              // return scalar
+{
+  iScalar<vtype> ret;
+  zeroit(ret._internal);
+  for(int i=0;i<N;i++){
+    ret._internal = ret._internal + arg._internal[i][i];
+  }
+  return ret;
+}
+
+// not this level, so recurse
+template<int Level,class vtype,int N> inline 
+auto traceIndex(const iMatrix<vtype,N> &arg) ->
+  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,// No index match
+         iMatrix<decltype(traceIndex<Level>(arg._internal[0][0])),N> >::type     // return matrix
+{
+  iMatrix<decltype(traceIndex<Level>(arg._internal[0][0])),N> ret;
+  for(int i=0;i<N;i++){
+  for(int j=0;j<N;j++){
+    ret._internal[i][j] = traceIndex<Level>(arg._internal[i][j]);
+  }}
+  return ret;
+}
+
+}
+#endif
diff --git a/lib/Grid_math_transpose.h b/lib/Grid_math_transpose.h
new file mode 100644
index 00000000..76f80b1b
--- /dev/null
+++ b/lib/Grid_math_transpose.h
@@ -0,0 +1,101 @@
+#ifndef GRID_MATH_TRANSPOSE_H
+#define GRID_MATH_TRANSPOSE_H
+namespace Grid {
+
+
+
+/////////////////////////////////////////////////////////////////
+// Transpose all indices
+/////////////////////////////////////////////////////////////////
+
+inline ComplexD transpose(ComplexD &rhs){  return rhs;}
+inline ComplexF transpose(ComplexF &rhs){  return rhs;}
+inline RealD transpose(RealD &rhs){  return rhs;}
+inline RealF transpose(RealF &rhs){  return rhs;}
+
+template<class vtype,int N>
+  inline typename std::enable_if<isGridTensor<vtype>::value, iMatrix<vtype,N> >::type 
+  transpose(iMatrix<vtype,N> arg)
+  {
+    iMatrix<vtype,N> ret;
+    for(int i=0;i<N;i++){
+      for(int j=0;j<N;j++){
+	ret._internal[i][j] = transpose(arg._internal[j][i]); // NB recurses
+      }}
+    return ret;
+  }
+template<class vtype,int N>
+  inline typename std::enable_if<isGridTensor<vtype>::notvalue, iMatrix<vtype,N> >::type 
+  transpose(iMatrix<vtype,N> arg)
+  {
+    iMatrix<vtype,N> ret;
+    for(int i=0;i<N;i++){
+      for(int j=0;j<N;j++){
+	ret._internal[i][j] = arg._internal[j][i]; // Stop recursion if not a tensor type
+      }}
+    return ret;
+  }
+
+template<class vtype>
+  inline typename std::enable_if<isGridTensor<vtype>::value, iScalar<vtype> >::type 
+  transpose(iScalar<vtype> arg)
+  {
+    iScalar<vtype> ret;
+    ret._internal = transpose(arg._internal); // NB recurses
+    return ret;
+  }
+
+template<class vtype>
+  inline typename std::enable_if<isGridTensor<vtype>::notvalue, iScalar<vtype> >::type 
+  transpose(iScalar<vtype> arg)
+  {
+    iScalar<vtype> ret;
+    ret._internal = arg._internal; // NB recursion stops
+    return ret;
+  }
+
+
+////////////////////////////////////////////////////////////////////////////////////////////
+// Transpose a specific index; instructive to compare this style of recursion termination
+// to that of adj; which is easiers?
+////////////////////////////////////////////////////////////////////////////////////////////
+template<int Level,class vtype,int N> inline 
+  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value, iMatrix<vtype,N> >::type 
+transposeIndex (const iMatrix<vtype,N> &arg)
+{
+  iMatrix<vtype,N> ret;
+  for(int i=0;i<N;i++){
+    for(int j=0;j<N;j++){
+      ret._internal[i][j] = arg._internal[j][i]; 
+  }}
+  return ret;
+}
+// or not
+template<int Level,class vtype,int N> inline 
+typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue, iMatrix<vtype,N> >::type 
+transposeIndex (const iMatrix<vtype,N> &arg)
+{
+  iMatrix<vtype,N> ret;
+  for(int i=0;i<N;i++){
+    for(int j=0;j<N;j++){
+      ret._internal[i][j] = transposeIndex<Level>(arg._internal[i][j]); 
+  }}
+  return ret;
+}
+template<int Level,class vtype> inline 
+typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue, iScalar<vtype> >::type 
+transposeIndex (const iScalar<vtype> &arg)
+{
+  iScalar<vtype> ret;
+  ret._internal=transposeIndex<Level>(arg._internal);
+  return ret;
+}
+template<int Level,class vtype> inline 
+typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value, iScalar<vtype> >::type 
+transposeIndex (const iScalar<vtype> &arg)
+{
+  return arg;
+}
+
+}
+#endif

From 0fce5237923bbb41b7b60ad63b669cdacd2b8e1e Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 18:54:30 +0100
Subject: [PATCH 074/429] Split up into multiple files

---
 lib/Grid_math_arith.h        | 744 +----------------------------------
 lib/Grid_math_arith_add.h    | 122 ++++++
 lib/Grid_math_arith_mac.h    |  81 ++++
 lib/Grid_math_arith_mul.h    | 189 +++++++++
 lib/Grid_math_arith_scalar.h | 258 ++++++++++++
 lib/Grid_math_arith_sub.h    | 134 +++++++
 6 files changed, 789 insertions(+), 739 deletions(-)
 create mode 100644 lib/Grid_math_arith_add.h
 create mode 100644 lib/Grid_math_arith_mac.h
 create mode 100644 lib/Grid_math_arith_mul.h
 create mode 100644 lib/Grid_math_arith_scalar.h
 create mode 100644 lib/Grid_math_arith_sub.h

diff --git a/lib/Grid_math_arith.h b/lib/Grid_math_arith.h
index aebaaee0..e7b6b4ac 100644
--- a/lib/Grid_math_arith.h
+++ b/lib/Grid_math_arith.h
@@ -1,745 +1,11 @@
 #ifndef GRID_MATH_ARITH_H
 #define GRID_MATH_ARITH_H
 
-namespace Grid {
+#include <Grid_math_arith_add.h>
+#include <Grid_math_arith_sub.h>
+#include <Grid_math_arith_mac.h>
+#include <Grid_math_arith_mul.h>
+#include <Grid_math_arith_scalar.h>
 
-    
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-    /////////////////////////////////////////// ADD         ///////////////////////////////////////////
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-    
-
-// ADD is simple for now; cannot mix types and straightforward template
-// Scalar +/- Scalar
-// Vector +/- Vector
-// Matrix +/- Matrix
-  template<class vtype,class ltype,class rtype> inline void add(iScalar<vtype> * __restrict__ ret,
-								const iScalar<ltype> * __restrict__ lhs,
-								const iScalar<rtype> * __restrict__ rhs)
-  {
-    add(&ret->_internal,&lhs->_internal,&rhs->_internal);
-  }
-  template<class vtype,class ltype,class rtype,int N> inline void add(iVector<vtype,N> * __restrict__ ret,
-								      const iVector<ltype,N> * __restrict__ lhs,
-								      const iVector<rtype,N> * __restrict__ rhs)
-  {
-    for(int c=0;c<N;c++){
-      ret->_internal[c]=lhs->_internal[c]+rhs->_internal[c];
-    }
-    return;
-  }
-  
-  template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
-									const iMatrix<ltype,N> * __restrict__ lhs,
-									const iMatrix<rtype,N> * __restrict__ rhs)
-  {
-    for(int c2=0;c2<N;c2++){
-      for(int c1=0;c1<N;c1++){
-        add(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal[c1][c2]);
-      }}
-    return;
-  }
-  template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
-									const iScalar<ltype>   * __restrict__ lhs,
-									const iMatrix<rtype,N> * __restrict__ rhs)
-  {
-    for(int c2=0;c2<N;c2++){
-      for(int c1=0;c1<N;c1++){
-        add(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
-      }}
-    return;
-  }
-  template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
-									const iMatrix<ltype,N> * __restrict__ lhs,
-									const iScalar<rtype>   * __restrict__ rhs)
-  {
-    for(int c2=0;c2<N;c2++){
-      for(int c1=0;c1<N;c1++){
-        if ( c1==c2)
-	  add(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
-        else
-	  ret->_internal[c1][c2]=lhs->_internal[c1][c2];
-      }}
-    return;
-  }
-
-  // Need to figure multi-precision.
-  template<class Mytype>  Mytype timesI(Mytype &r)
-    {
-      iScalar<Complex> i;
-      i._internal = Complex(0,1);
-      return r*i;
-    }
-
-  // + operator for scalar, vector, matrix
-  template<class ltype,class rtype>
-    //inline auto operator + (iScalar<ltype>& lhs,iScalar<rtype>&& rhs) -> iScalar<decltype(lhs._internal + rhs._internal)>
-    inline auto operator + (const iScalar<ltype>& lhs,const iScalar<rtype>& rhs) -> iScalar<decltype(lhs._internal + rhs._internal)>
-  {
-    typedef iScalar<decltype(lhs._internal+rhs._internal)> ret_t;
-    ret_t ret;
-    add(&ret,&lhs,&rhs);
-    return ret;
-  }
-  template<class ltype,class rtype,int N>
-    inline auto operator + (const iVector<ltype,N>& lhs,const iVector<rtype,N>& rhs) ->iVector<decltype(lhs._internal[0]+rhs._internal[0]),N>
-    {
-      typedef iVector<decltype(lhs._internal[0]+rhs._internal[0]),N> ret_t;
-    ret_t ret;
-    add(&ret,&lhs,&rhs);
-    return ret;
-    }
-  template<class ltype,class rtype,int N>
-    inline auto operator + (const iMatrix<ltype,N>& lhs,const iMatrix<rtype,N>& rhs) ->iMatrix<decltype(lhs._internal[0][0]+rhs._internal[0][0]),N>
-    {
-      typedef iMatrix<decltype(lhs._internal[0][0]+rhs._internal[0][0]),N> ret_t;
-      ret_t ret;
-      add(&ret,&lhs,&rhs);
-      return ret;
-    }
-  template<class ltype,class rtype,int N>
-inline auto operator + (const iScalar<ltype>& lhs,const iMatrix<rtype,N>& rhs)->iMatrix<decltype(lhs._internal+rhs._internal[0][0]),N>
-    {
-      typedef iMatrix<decltype(lhs._internal+rhs._internal[0][0]),N> ret_t;
-      ret_t ret;
-      add(&ret,&lhs,&rhs);
-      return ret;
-    }
-
-  template<class ltype,class rtype,int N>
-    inline auto operator + (const iMatrix<ltype,N>& lhs,const iScalar<rtype>& rhs)->iMatrix<decltype(lhs._internal[0][0]+rhs._internal),N>
-    {
-      typedef iMatrix<decltype(lhs._internal[0][0]+rhs._internal),N> ret_t;
-      ret_t ret;
-      add(&ret,&lhs,&rhs);
-      return ret;
-    }
-
-
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-    /////////////////////////////////////////// SUB         ///////////////////////////////////////////
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-    
-
-// SUB is simple for now; cannot mix types and straightforward template
-// Scalar +/- Scalar
-// Vector +/- Vector
-// Matrix +/- Matrix
-// Matrix /- scalar
-template<class vtype,class ltype,class rtype> inline void sub(iScalar<vtype> * __restrict__ ret,
-                                                              const iScalar<ltype> * __restrict__ lhs,
-                                                              const iScalar<rtype> * __restrict__ rhs)
-{
-    sub(&ret->_internal,&lhs->_internal,&rhs->_internal);
-}
-
-template<class vtype,class ltype,class rtype,int N> inline void sub(iVector<vtype,N> * __restrict__ ret,
-                                                                    const iVector<ltype,N> * __restrict__ lhs,
-                                                                    const iVector<rtype,N> * __restrict__ rhs)
-{
-    for(int c=0;c<N;c++){
-        ret->_internal[c]=lhs->_internal[c]-rhs->_internal[c];
-    }
-    return;
-}
-template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
-                                                                     const iMatrix<ltype,N> * __restrict__ lhs,
-                                                                     const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        sub(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal[c1][c2]);
-    }}
-    return;
-}
-template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
-                                                                     const iScalar<ltype> * __restrict__ lhs,
-                                                                     const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        if ( c1!=c2) {
-            sub(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
-        } else {
-            // Fails -- need unary minus. Catalogue other unops?
-            ret->_internal[c1][c2]=zero;
-            ret->_internal[c1][c2]=ret->_internal[c1][c2]-rhs->_internal[c1][c2];
-
-        }
-    }}
-    return;
-}
-template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
-                                                                     const iMatrix<ltype,N> * __restrict__ lhs,
-                                                                     const iScalar<rtype> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        if ( c1!=c2)
-            sub(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
-        else
-            ret->_internal[c1][c2]=lhs->_internal[c1][c2];
-    }}
-    return;
-}
-
-template<class v> void vprefetch(const iScalar<v> &vv)
-{
-  vprefetch(vv._internal);
-}
-template<class v,int N> void vprefetch(const iVector<v,N> &vv)
-{
-  for(int i=0;i<N;i++){
-    vprefetch(vv._internal[i]);
-  }
-}
-template<class v,int N> void vprefetch(const iMatrix<v,N> &vv)
-{
-  for(int i=0;i<N;i++){
-  for(int j=0;j<N;j++){
-    vprefetch(vv._internal[i][j]);
-  }}
-}
-
-    // - operator for scalar, vector, matrix
-template<class ltype,class rtype> inline auto
-operator - (const iScalar<ltype>& lhs, const iScalar<rtype>& rhs) -> iScalar<decltype(lhs._internal - rhs._internal)>
-{
-    typedef iScalar<decltype(lhs._internal-rhs._internal)> ret_t;
-    ret_t ret;
-    sub(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator - (const iVector<ltype,N>& lhs,const iVector<rtype,N>& rhs) ->iVector<decltype(lhs._internal[0]-rhs._internal[0]),N>
-{
-    typedef iVector<decltype(lhs._internal[0]-rhs._internal[0]),N> ret_t;
-    ret_t ret;
-    sub(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator - (const iMatrix<ltype,N>& lhs,const iMatrix<rtype,N>& rhs) ->iMatrix<decltype(lhs._internal[0][0]-rhs._internal[0][0]),N>
-{
-    typedef iMatrix<decltype(lhs._internal[0][0]-rhs._internal[0][0]),N> ret_t;
-    ret_t ret;
-    sub(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator - (const iScalar<ltype>& lhs,const iMatrix<rtype,N>& rhs)->iMatrix<decltype(lhs._internal-rhs._internal[0][0]),N>
-{
-    typedef iMatrix<decltype(lhs._internal-rhs._internal[0][0]),N> ret_t;
-    ret_t ret;
-    sub(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class ltype,class rtype,int N>
-inline auto operator - (const iMatrix<ltype,N>& lhs,const iScalar<rtype>& rhs)->iMatrix<decltype(lhs._internal[0][0]-rhs._internal),N>
-{
-    typedef iMatrix<decltype(lhs._internal[0][0]-rhs._internal),N> ret_t;
-    ret_t ret;
-    sub(&ret,&lhs,&rhs);
-    return ret;
-}
-
-///////////////////////////////////////////////////////////////////////////////////////////////////
-/////////////////////////////////////////// MAC         ///////////////////////////////////////////
-///////////////////////////////////////////////////////////////////////////////////////////////////
-
-    ///////////////////////////
-    // Legal multiplication table
-    ///////////////////////////
-    // scal x scal = scal
-    // mat x  mat  = mat
-    // mat  x scal = mat
-    // scal x mat  = mat
-    // mat  x vec  = vec
-    // vec  x scal = vec
-    // scal x vec  = vec
-    ///////////////////////////
-template<class rtype,class vtype,class mtype>
-inline  void mac(iScalar<rtype> * __restrict__ ret,const iScalar<vtype> * __restrict__ lhs,const iScalar<mtype> * __restrict__ rhs)
-{
-    mac(&ret->_internal,&lhs->_internal,&rhs->_internal);
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-    for(int c3=0;c3<N;c3++){
-        mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c3],&rhs->_internal[c3][c2]);
-    }}}
-    return;
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs){
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
-    }}
-    return;
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iScalar<ltype> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        mac(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
-    }}
-    return;
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iVector<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iVector<rtype,N> * __restrict__ rhs)
-{
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        mac(&ret->_internal[c1],&lhs->_internal[c1][c2],&rhs->_internal[c2]);
-    }}
-    return;
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iVector<rrtype,N> * __restrict__ ret,const iScalar<ltype> * __restrict__ lhs,const iVector<rtype,N> * __restrict__ rhs)
-{
-    for(int c1=0;c1<N;c1++){
-        mac(&ret->_internal[c1],&lhs->_internal,&rhs->_internal[c1]);
-    }
-    return;
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iVector<rrtype,N> * __restrict__ ret,const iVector<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs)
-{
-    for(int c1=0;c1<N;c1++){
-        mac(&ret->_internal[c1],&lhs->_internal[c1],&rhs->_internal);
-    }
-    return;
-}
-
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-    /////////////////////////////////////////// MUL         ///////////////////////////////////////////
-    ///////////////////////////////////////////////////////////////////////////////////////////////////
-
-    
-template<class rtype,class vtype,class mtype>
-inline void mult(iScalar<rtype> * __restrict__ ret,const iScalar<mtype> * __restrict__ lhs,const iScalar<vtype> * __restrict__ rhs){
-    mult(&ret->_internal,&lhs->_internal,&rhs->_internal);
-}
-
-template<class rrtype,class ltype,class rtype,int N>
-inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        mult(&ret->_internal[c1][c2],&lhs->_internal[c1][0],&rhs->_internal[0][c2]);
-        for(int c3=1;c3<N;c3++){
-            mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c3],&rhs->_internal[c3][c2]);
-        }
-    }}
-    return;
-}
-template<class rrtype,class ltype,class rtype,int N>
-inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        mult(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
-    }}
-    return;
-}
-
-template<class rrtype,class ltype,class rtype, int N>
-inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iScalar<ltype>   * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
-        mult(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
-    }}
-    return;
-}
-// Matrix left multiplies vector
-template<class rtype,class vtype,class mtype,int N>
-inline void mult(iVector<rtype,N> * __restrict__ ret,const iMatrix<mtype,N> * __restrict__ lhs,const iVector<vtype,N> * __restrict__ rhs)
-{
-    for(int c1=0;c1<N;c1++){
-        mult(&ret->_internal[c1],&lhs->_internal[c1][0],&rhs->_internal[0]);
-        for(int c2=1;c2<N;c2++){
-            mac(&ret->_internal[c1],&lhs->_internal[c1][c2],&rhs->_internal[c2]);
-        }
-    }
-    return;
-}
-template<class rtype,class vtype,class mtype,int N>
-inline void mult(iVector<rtype,N> * __restrict__ ret,
-                 const iScalar<mtype>   * __restrict__ lhs,
-                 const iVector<vtype,N> * __restrict__ rhs){
-    for(int c1=0;c1<N;c1++){
-        mult(&ret->_internal[c1],&lhs->_internal,&rhs->_internal[c1]);
-    }
-}
-template<class rtype,class vtype,class mtype,int N>
-inline void mult(iVector<rtype,N> * __restrict__ ret,
-                 const iVector<vtype,N> * __restrict__ rhs,
-                 const iScalar<mtype> * __restrict__ lhs){
-    mult(ret,lhs,rhs);
-}
-    
-
-
-template<class rtype,class vtype,class mtype,int N> inline
-iVector<rtype,N> operator * (const iMatrix<mtype,N>& lhs,const iVector<vtype,N>& rhs)
-{
-    iVector<rtype,N> ret;
-    mult(&ret,&lhs,&rhs);
-    return ret;
-}
-
-template<class rtype,class vtype,class mtype,int N> inline
-iVector<rtype,N> operator * (const iScalar<mtype>& lhs,const iVector<vtype,N>& rhs)
-{
-    iVector<rtype,N> ret;
-    mult(&ret,&lhs,&rhs);
-    return ret;
-}
-
-template<class rtype,class vtype,class mtype,int N> inline
-iVector<rtype,N> operator * (const iVector<mtype,N>& lhs,const iScalar<vtype>& rhs)
-{
-    iVector<rtype,N> ret;
-    mult(&ret,&lhs,&rhs);
-    return ret;
-}
-    
-    //////////////////////////////////////////////////////////////////
-    // Glue operators to mult routines. Must resolve return type cleverly from typeof(internal)
-    // since nesting matrix<scalar> x matrix<matrix>-> matrix<matrix>
-    // while         matrix<scalar> x matrix<scalar>-> matrix<scalar>
-    // so return type depends on argument types in nasty way.
-    //////////////////////////////////////////////////////////////////
-    // scal x scal = scal
-    // mat x  mat  = mat
-    // mat  x scal = mat
-    // scal x mat  = mat
-    // mat  x vec  = vec
-    // vec  x scal = vec
-    // scal x vec  = vec
-    //
-    // We can special case scalar_type ??
-template<class l,class r>
-inline auto operator * (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(lhs._internal * rhs._internal)>
-{
-    typedef iScalar<decltype(lhs._internal*rhs._internal)> ret_t;
-    ret_t ret;
-    mult(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class l,class r,int N> inline
-auto operator * (const iMatrix<l,N>& lhs,const iMatrix<r,N>& rhs) -> iMatrix<decltype(lhs._internal[0][0]*rhs._internal[0][0]),N>
-{
-    typedef decltype(lhs._internal[0][0]*rhs._internal[0][0]) ret_t;
-    iMatrix<ret_t,N> ret;
-    mult(&ret,&lhs,&rhs);
-    return ret;
-}
-template<class l,class r, int N> inline
-auto operator * (const iMatrix<r,N>& lhs,const iScalar<l>& rhs) -> iMatrix<decltype(lhs._internal[0][0]*rhs._internal),N>
-{
-    typedef decltype(lhs._internal[0][0]*rhs._internal) ret_t;
-        
-    iMatrix<ret_t,N> ret;
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        mult(&ret._internal[c1][c2],&lhs._internal[c1][c2],&rhs._internal);
-    }}
-    return ret;
-}
-template<class l,class r,int N> inline
-auto operator * (const iScalar<l>& lhs,const iMatrix<r,N>& rhs) -> iMatrix<decltype(lhs._internal*rhs._internal[0][0]),N>
-{
-    typedef decltype(lhs._internal*rhs._internal[0][0]) ret_t;
-    iMatrix<ret_t,N> ret;
-    for(int c1=0;c1<N;c1++){
-    for(int c2=0;c2<N;c2++){
-        mult(&ret._internal[c1][c2],&lhs._internal,&rhs._internal[c1][c2]);
-    }}
-    return ret;
-}
-template<class l,class r,int N> inline
-auto operator * (const iMatrix<l,N>& lhs,const iVector<r,N>& rhs) -> iVector<decltype(lhs._internal[0][0]*rhs._internal[0]),N>
-{
-    typedef decltype(lhs._internal[0][0]*rhs._internal[0]) ret_t;
-    iVector<ret_t,N> ret;
-    for(int c1=0;c1<N;c1++){
-        mult(&ret._internal[c1],&lhs._internal[c1][0],&rhs._internal[0]);
-        for(int c2=1;c2<N;c2++){
-            mac(&ret._internal[c1],&lhs._internal[c1][c2],&rhs._internal[c2]);
-        }
-    }
-    return ret;
-}
-template<class l,class r,int N> inline
-auto operator * (const iScalar<l>& lhs,const iVector<r,N>& rhs) -> iVector<decltype(lhs._internal*rhs._internal[0]),N>
-{
-    typedef decltype(lhs._internal*rhs._internal[0]) ret_t;
-    iVector<ret_t,N> ret;
-    for(int c1=0;c1<N;c1++){
-        mult(&ret._internal[c1],&lhs._internal,&rhs._internal[c1]);
-    }
-    return ret;
-}
-template<class l,class r,int N> inline
-auto operator * (const iVector<l,N>& lhs,const iScalar<r>& rhs) -> iVector<decltype(lhs._internal[0]*rhs._internal),N>
-{
-    typedef decltype(lhs._internal[0]*rhs._internal) ret_t;
-    iVector<ret_t,N> ret;
-    for(int c1=0;c1<N;c1++){
-        mult(&ret._internal[c1],&lhs._internal[c1],&rhs._internal);
-    }
-    return ret;
-}
-
-
-//////////////////////////////////////////////////////////////////////////////////////////
-// Must support native C++ types Integer, Complex, Real
-//////////////////////////////////////////////////////////////////////////////////////////
-
-// multiplication by fundamental scalar type
-template<class l,int N> inline iScalar<l> operator * (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
-{
-  typename iScalar<l>::tensor_reduced srhs(rhs);
-  return lhs*srhs;
-}
-template<class l,int N> inline iScalar<l> operator * (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
-
-template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
-{
-  typename iVector<l,N>::tensor_reduced srhs(rhs);
-  return lhs*srhs;
-}
-template<class l,int N> inline iVector<l,N> operator * (const typename iScalar<l>::scalar_type lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
-
-template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type &rhs) 
-{
-  typename iMatrix<l,N>::tensor_reduced srhs(rhs);
-  return lhs*srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator * (const typename iScalar<l>::scalar_type & lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
-
-////////////////////////////////////////////////////////////////////
-// Double support; cast to "scalar_type" through constructor
-////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,double rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs*srhs;
-}
-template<class l> inline iScalar<l> operator * (double lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
-
-template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,double rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs*srhs;
-}
-template<class l,int N> inline iVector<l,N> operator * (double lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
-
-template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,double rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs*srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator * (double lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
-
-////////////////////////////////////////////////////////////////////
-// Complex support; cast to "scalar_type" through constructor
-////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,ComplexD rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs*srhs;
-}
-template<class l> inline iScalar<l> operator * (ComplexD lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
-
-template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,ComplexD rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs*srhs;
-}
-template<class l,int N> inline iVector<l,N> operator * (ComplexD lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
-
-template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,ComplexD rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs*srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator * (ComplexD lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
-
-////////////////////////////////////////////////////////////////////
-// Integer support; cast to "scalar_type" through constructor
-////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,Integer rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs*srhs;
-}
-template<class l> inline iScalar<l> operator * (Integer lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
-
-template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,Integer rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs*srhs;
-}
-template<class l,int N> inline iVector<l,N> operator * (Integer lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
-
-template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,Integer rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs*srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator * (Integer lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
-
-
-
-///////////////////////////////////////////////////////////////////////////////////////////////
-// addition by fundamental scalar type applies to matrix(down diag) and scalar
-///////////////////////////////////////////////////////////////////////////////////////////////
-template<class l,int N> inline iScalar<l> operator + (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
-{
-  typename iScalar<l>::tensor_reduced srhs(rhs);
-  return lhs+srhs;
-}
-template<class l,int N> inline iScalar<l> operator + (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
-
-template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
-{
-  typename iMatrix<l,N>::tensor_reduced srhs(rhs);
-  return lhs+srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator + (const typename iScalar<l>::scalar_type lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
-
-////////////////////////////////////////////////////////////////////
-// Double support; cast to "scalar_type" through constructor
-////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator + (const iScalar<l>& lhs,double rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs+srhs;
-}
-template<class l> inline iScalar<l> operator + (double lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
-
-template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,double rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs+srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator + (double lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
-
-
-// Integer support cast to scalar type through constructor
-
-
-template<class l> inline iScalar<l> operator + (const iScalar<l>& lhs,Integer rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs+srhs;
-}
-
-template<class l> inline iScalar<l> operator + (Integer lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
-
-template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,Integer rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs+srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator + (Integer lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
-
-
-///////////////////////////////////////////////////////////////////////////////////////////////
-// subtraction of fundamental scalar type applies to matrix(down diag) and scalar
-///////////////////////////////////////////////////////////////////////////////////////////////
-template<class l,int N> inline iScalar<l> operator - (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
-{
-  typename iScalar<l>::tensor_reduced srhs(rhs);
-  return lhs-srhs;
-}
-template<class l,int N> inline iScalar<l> operator - (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) 
-{
-  typename iScalar<l>::tensor_reduced slhs(lhs);
-  return slhs-rhs;
-}
-
-template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
-{
-  typename iScalar<l>::tensor_reduced srhs(rhs);
-  return lhs-srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator - (const typename iScalar<l>::scalar_type lhs,const iMatrix<l,N>& rhs) 
-{
-  typename iScalar<l>::tensor_reduced slhs(lhs);
-  return slhs-rhs;
-}
-
-////////////////////////////////////////////////////////////////////
-// Double support; cast to "scalar_type" through constructor
-////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator - (const iScalar<l>& lhs,double rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs-srhs;
-}
-template<class l> inline iScalar<l> operator - (double lhs,const iScalar<l>& rhs) 
-{
-  typename iScalar<l>::scalar_type t(lhs);
-  typename iScalar<l>::tensor_reduced slhs(t);
-  return slhs-rhs;
-}
-
-template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,double rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs-srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator - (double lhs,const iMatrix<l,N>& rhs) 
-{
-  typename iScalar<l>::scalar_type t(lhs);
-  typename iScalar<l>::tensor_reduced slhs(t);
-  return slhs-rhs;
-}
-
-////////////////////////////////////////////////////////////////////
-// Integer support; cast to "scalar_type" through constructor
-////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator - (const iScalar<l>& lhs,Integer rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs-srhs;
-}
-template<class l> inline iScalar<l> operator - (Integer lhs,const iScalar<l>& rhs) 
-{
-  typename iScalar<l>::scalar_type t(lhs);
-  typename iScalar<l>::tensor_reduced slhs(t);
-  return slhs-rhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,Integer rhs) 
-{
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
-  return lhs-srhs;
-}
-template<class l,int N> inline iMatrix<l,N> operator - (Integer lhs,const iMatrix<l,N>& rhs) 
-{
-  typename iScalar<l>::scalar_type t(lhs);
-  typename iScalar<l>::tensor_reduced slhs(t);
-  return slhs-rhs;
-}
-
-}
 #endif
 
diff --git a/lib/Grid_math_arith_add.h b/lib/Grid_math_arith_add.h
new file mode 100644
index 00000000..7e62e07b
--- /dev/null
+++ b/lib/Grid_math_arith_add.h
@@ -0,0 +1,122 @@
+#ifndef GRID_MATH_ARITH_ADD_H
+#define GRID_MATH_ARITH_ADD_H
+
+namespace Grid {
+
+    
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    /////////////////////////////////////////// ADD         ///////////////////////////////////////////
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    
+
+// ADD is simple for now; cannot mix types and straightforward template
+// Scalar +/- Scalar
+// Vector +/- Vector
+// Matrix +/- Matrix
+  template<class vtype,class ltype,class rtype> inline void add(iScalar<vtype> * __restrict__ ret,
+								const iScalar<ltype> * __restrict__ lhs,
+								const iScalar<rtype> * __restrict__ rhs)
+  {
+    add(&ret->_internal,&lhs->_internal,&rhs->_internal);
+  }
+  template<class vtype,class ltype,class rtype,int N> inline void add(iVector<vtype,N> * __restrict__ ret,
+								      const iVector<ltype,N> * __restrict__ lhs,
+								      const iVector<rtype,N> * __restrict__ rhs)
+  {
+    for(int c=0;c<N;c++){
+      ret->_internal[c]=lhs->_internal[c]+rhs->_internal[c];
+    }
+    return;
+  }
+  
+  template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
+									const iMatrix<ltype,N> * __restrict__ lhs,
+									const iMatrix<rtype,N> * __restrict__ rhs)
+  {
+    for(int c2=0;c2<N;c2++){
+      for(int c1=0;c1<N;c1++){
+        add(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal[c1][c2]);
+      }}
+    return;
+  }
+  template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
+									const iScalar<ltype>   * __restrict__ lhs,
+									const iMatrix<rtype,N> * __restrict__ rhs)
+  {
+    for(int c2=0;c2<N;c2++){
+      for(int c1=0;c1<N;c1++){
+        add(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
+      }}
+    return;
+  }
+  template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
+									const iMatrix<ltype,N> * __restrict__ lhs,
+									const iScalar<rtype>   * __restrict__ rhs)
+  {
+    for(int c2=0;c2<N;c2++){
+      for(int c1=0;c1<N;c1++){
+        if ( c1==c2)
+	  add(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
+        else
+	  ret->_internal[c1][c2]=lhs->_internal[c1][c2];
+      }}
+    return;
+  }
+
+  // Need to figure multi-precision.
+  template<class Mytype>  Mytype timesI(Mytype &r)
+    {
+      iScalar<Complex> i;
+      i._internal = Complex(0,1);
+      return r*i;
+    }
+
+  // + operator for scalar, vector, matrix
+  template<class ltype,class rtype>
+    //inline auto operator + (iScalar<ltype>& lhs,iScalar<rtype>&& rhs) -> iScalar<decltype(lhs._internal + rhs._internal)>
+    inline auto operator + (const iScalar<ltype>& lhs,const iScalar<rtype>& rhs) -> iScalar<decltype(lhs._internal + rhs._internal)>
+  {
+    typedef iScalar<decltype(lhs._internal+rhs._internal)> ret_t;
+    ret_t ret;
+    add(&ret,&lhs,&rhs);
+    return ret;
+  }
+  template<class ltype,class rtype,int N>
+    inline auto operator + (const iVector<ltype,N>& lhs,const iVector<rtype,N>& rhs) ->iVector<decltype(lhs._internal[0]+rhs._internal[0]),N>
+    {
+      typedef iVector<decltype(lhs._internal[0]+rhs._internal[0]),N> ret_t;
+    ret_t ret;
+    add(&ret,&lhs,&rhs);
+    return ret;
+    }
+  template<class ltype,class rtype,int N>
+    inline auto operator + (const iMatrix<ltype,N>& lhs,const iMatrix<rtype,N>& rhs) ->iMatrix<decltype(lhs._internal[0][0]+rhs._internal[0][0]),N>
+    {
+      typedef iMatrix<decltype(lhs._internal[0][0]+rhs._internal[0][0]),N> ret_t;
+      ret_t ret;
+      add(&ret,&lhs,&rhs);
+      return ret;
+    }
+  template<class ltype,class rtype,int N>
+inline auto operator + (const iScalar<ltype>& lhs,const iMatrix<rtype,N>& rhs)->iMatrix<decltype(lhs._internal+rhs._internal[0][0]),N>
+    {
+      typedef iMatrix<decltype(lhs._internal+rhs._internal[0][0]),N> ret_t;
+      ret_t ret;
+      add(&ret,&lhs,&rhs);
+      return ret;
+    }
+
+  template<class ltype,class rtype,int N>
+    inline auto operator + (const iMatrix<ltype,N>& lhs,const iScalar<rtype>& rhs)->iMatrix<decltype(lhs._internal[0][0]+rhs._internal),N>
+    {
+      typedef iMatrix<decltype(lhs._internal[0][0]+rhs._internal),N> ret_t;
+      ret_t ret;
+      add(&ret,&lhs,&rhs);
+      return ret;
+    }
+
+
+
+}
+
+#endif
diff --git a/lib/Grid_math_arith_mac.h b/lib/Grid_math_arith_mac.h
new file mode 100644
index 00000000..59cb9a5e
--- /dev/null
+++ b/lib/Grid_math_arith_mac.h
@@ -0,0 +1,81 @@
+#ifndef GRID_MATH_ARITH_MAC_H
+#define GRID_MATH_ARITH_MAC_H
+
+namespace Grid {
+
+
+///////////////////////////////////////////////////////////////////////////////////////////////////
+/////////////////////////////////////////// MAC         ///////////////////////////////////////////
+///////////////////////////////////////////////////////////////////////////////////////////////////
+
+    ///////////////////////////
+    // Legal multiplication table
+    ///////////////////////////
+    // scal x scal = scal
+    // mat x  mat  = mat
+    // mat  x scal = mat
+    // scal x mat  = mat
+    // mat  x vec  = vec
+    // vec  x scal = vec
+    // scal x vec  = vec
+    ///////////////////////////
+template<class rtype,class vtype,class mtype>
+inline  void mac(iScalar<rtype> * __restrict__ ret,const iScalar<vtype> * __restrict__ lhs,const iScalar<mtype> * __restrict__ rhs)
+{
+    mac(&ret->_internal,&lhs->_internal,&rhs->_internal);
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+    for(int c3=0;c3<N;c3++){
+        mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c3],&rhs->_internal[c3][c2]);
+    }}}
+    return;
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs){
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
+    }}
+    return;
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iScalar<ltype> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        mac(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
+    }}
+    return;
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mac(iVector<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iVector<rtype,N> * __restrict__ rhs)
+{
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        mac(&ret->_internal[c1],&lhs->_internal[c1][c2],&rhs->_internal[c2]);
+    }}
+    return;
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mac(iVector<rrtype,N> * __restrict__ ret,const iScalar<ltype> * __restrict__ lhs,const iVector<rtype,N> * __restrict__ rhs)
+{
+    for(int c1=0;c1<N;c1++){
+        mac(&ret->_internal[c1],&lhs->_internal,&rhs->_internal[c1]);
+    }
+    return;
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mac(iVector<rrtype,N> * __restrict__ ret,const iVector<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs)
+{
+    for(int c1=0;c1<N;c1++){
+        mac(&ret->_internal[c1],&lhs->_internal[c1],&rhs->_internal);
+    }
+    return;
+}
+
+
+}
+
+#endif
diff --git a/lib/Grid_math_arith_mul.h b/lib/Grid_math_arith_mul.h
new file mode 100644
index 00000000..c8bb0b2c
--- /dev/null
+++ b/lib/Grid_math_arith_mul.h
@@ -0,0 +1,189 @@
+#ifndef GRID_MATH_ARITH_MUL_H
+#define GRID_MATH_ARITH_MUL_H
+
+namespace Grid {
+
+
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    /////////////////////////////////////////// MUL         ///////////////////////////////////////////
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+
+    
+template<class rtype,class vtype,class mtype>
+inline void mult(iScalar<rtype> * __restrict__ ret,const iScalar<mtype> * __restrict__ lhs,const iScalar<vtype> * __restrict__ rhs){
+    mult(&ret->_internal,&lhs->_internal,&rhs->_internal);
+}
+
+template<class rrtype,class ltype,class rtype,int N>
+inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        mult(&ret->_internal[c1][c2],&lhs->_internal[c1][0],&rhs->_internal[0][c2]);
+        for(int c3=1;c3<N;c3++){
+            mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c3],&rhs->_internal[c3][c2]);
+        }
+    }}
+    return;
+}
+template<class rrtype,class ltype,class rtype,int N>
+inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        mult(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
+    }}
+    return;
+}
+
+template<class rrtype,class ltype,class rtype, int N>
+inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iScalar<ltype>   * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        mult(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
+    }}
+    return;
+}
+// Matrix left multiplies vector
+template<class rtype,class vtype,class mtype,int N>
+inline void mult(iVector<rtype,N> * __restrict__ ret,const iMatrix<mtype,N> * __restrict__ lhs,const iVector<vtype,N> * __restrict__ rhs)
+{
+    for(int c1=0;c1<N;c1++){
+        mult(&ret->_internal[c1],&lhs->_internal[c1][0],&rhs->_internal[0]);
+        for(int c2=1;c2<N;c2++){
+            mac(&ret->_internal[c1],&lhs->_internal[c1][c2],&rhs->_internal[c2]);
+        }
+    }
+    return;
+}
+template<class rtype,class vtype,class mtype,int N>
+inline void mult(iVector<rtype,N> * __restrict__ ret,
+                 const iScalar<mtype>   * __restrict__ lhs,
+                 const iVector<vtype,N> * __restrict__ rhs){
+    for(int c1=0;c1<N;c1++){
+        mult(&ret->_internal[c1],&lhs->_internal,&rhs->_internal[c1]);
+    }
+}
+template<class rtype,class vtype,class mtype,int N>
+inline void mult(iVector<rtype,N> * __restrict__ ret,
+                 const iVector<vtype,N> * __restrict__ rhs,
+                 const iScalar<mtype> * __restrict__ lhs){
+    mult(ret,lhs,rhs);
+}
+    
+
+
+template<class rtype,class vtype,class mtype,int N> inline
+iVector<rtype,N> operator * (const iMatrix<mtype,N>& lhs,const iVector<vtype,N>& rhs)
+{
+    iVector<rtype,N> ret;
+    mult(&ret,&lhs,&rhs);
+    return ret;
+}
+
+template<class rtype,class vtype,class mtype,int N> inline
+iVector<rtype,N> operator * (const iScalar<mtype>& lhs,const iVector<vtype,N>& rhs)
+{
+    iVector<rtype,N> ret;
+    mult(&ret,&lhs,&rhs);
+    return ret;
+}
+
+template<class rtype,class vtype,class mtype,int N> inline
+iVector<rtype,N> operator * (const iVector<mtype,N>& lhs,const iScalar<vtype>& rhs)
+{
+    iVector<rtype,N> ret;
+    mult(&ret,&lhs,&rhs);
+    return ret;
+}
+    
+    //////////////////////////////////////////////////////////////////
+    // Glue operators to mult routines. Must resolve return type cleverly from typeof(internal)
+    // since nesting matrix<scalar> x matrix<matrix>-> matrix<matrix>
+    // while         matrix<scalar> x matrix<scalar>-> matrix<scalar>
+    // so return type depends on argument types in nasty way.
+    //////////////////////////////////////////////////////////////////
+    // scal x scal = scal
+    // mat x  mat  = mat
+    // mat  x scal = mat
+    // scal x mat  = mat
+    // mat  x vec  = vec
+    // vec  x scal = vec
+    // scal x vec  = vec
+    //
+    // We can special case scalar_type ??
+template<class l,class r>
+inline auto operator * (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(lhs._internal * rhs._internal)>
+{
+    typedef iScalar<decltype(lhs._internal*rhs._internal)> ret_t;
+    ret_t ret;
+    mult(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class l,class r,int N> inline
+auto operator * (const iMatrix<l,N>& lhs,const iMatrix<r,N>& rhs) -> iMatrix<decltype(lhs._internal[0][0]*rhs._internal[0][0]),N>
+{
+    typedef decltype(lhs._internal[0][0]*rhs._internal[0][0]) ret_t;
+    iMatrix<ret_t,N> ret;
+    mult(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class l,class r, int N> inline
+auto operator * (const iMatrix<r,N>& lhs,const iScalar<l>& rhs) -> iMatrix<decltype(lhs._internal[0][0]*rhs._internal),N>
+{
+    typedef decltype(lhs._internal[0][0]*rhs._internal) ret_t;
+        
+    iMatrix<ret_t,N> ret;
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        mult(&ret._internal[c1][c2],&lhs._internal[c1][c2],&rhs._internal);
+    }}
+    return ret;
+}
+template<class l,class r,int N> inline
+auto operator * (const iScalar<l>& lhs,const iMatrix<r,N>& rhs) -> iMatrix<decltype(lhs._internal*rhs._internal[0][0]),N>
+{
+    typedef decltype(lhs._internal*rhs._internal[0][0]) ret_t;
+    iMatrix<ret_t,N> ret;
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
+        mult(&ret._internal[c1][c2],&lhs._internal,&rhs._internal[c1][c2]);
+    }}
+    return ret;
+}
+template<class l,class r,int N> inline
+auto operator * (const iMatrix<l,N>& lhs,const iVector<r,N>& rhs) -> iVector<decltype(lhs._internal[0][0]*rhs._internal[0]),N>
+{
+    typedef decltype(lhs._internal[0][0]*rhs._internal[0]) ret_t;
+    iVector<ret_t,N> ret;
+    for(int c1=0;c1<N;c1++){
+        mult(&ret._internal[c1],&lhs._internal[c1][0],&rhs._internal[0]);
+        for(int c2=1;c2<N;c2++){
+            mac(&ret._internal[c1],&lhs._internal[c1][c2],&rhs._internal[c2]);
+        }
+    }
+    return ret;
+}
+template<class l,class r,int N> inline
+auto operator * (const iScalar<l>& lhs,const iVector<r,N>& rhs) -> iVector<decltype(lhs._internal*rhs._internal[0]),N>
+{
+    typedef decltype(lhs._internal*rhs._internal[0]) ret_t;
+    iVector<ret_t,N> ret;
+    for(int c1=0;c1<N;c1++){
+        mult(&ret._internal[c1],&lhs._internal,&rhs._internal[c1]);
+    }
+    return ret;
+}
+template<class l,class r,int N> inline
+auto operator * (const iVector<l,N>& lhs,const iScalar<r>& rhs) -> iVector<decltype(lhs._internal[0]*rhs._internal),N>
+{
+    typedef decltype(lhs._internal[0]*rhs._internal) ret_t;
+    iVector<ret_t,N> ret;
+    for(int c1=0;c1<N;c1++){
+        mult(&ret._internal[c1],&lhs._internal[c1],&rhs._internal);
+    }
+    return ret;
+}
+
+
+}
+
+#endif
diff --git a/lib/Grid_math_arith_scalar.h b/lib/Grid_math_arith_scalar.h
new file mode 100644
index 00000000..2adb2836
--- /dev/null
+++ b/lib/Grid_math_arith_scalar.h
@@ -0,0 +1,258 @@
+#ifndef GRID_MATH_ARITH_SCALAR_H
+#define GRID_MATH_ARITH_SCALAR_H
+
+namespace Grid {
+
+
+//////////////////////////////////////////////////////////////////////////////////////////
+// Must support native C++ types Integer, Complex, Real
+//////////////////////////////////////////////////////////////////////////////////////////
+
+// multiplication by fundamental scalar type
+template<class l,int N> inline iScalar<l> operator * (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
+{
+  typename iScalar<l>::tensor_reduced srhs(rhs);
+  return lhs*srhs;
+}
+template<class l,int N> inline iScalar<l> operator * (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
+{
+  typename iVector<l,N>::tensor_reduced srhs(rhs);
+  return lhs*srhs;
+}
+template<class l,int N> inline iVector<l,N> operator * (const typename iScalar<l>::scalar_type lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type &rhs) 
+{
+  typename iMatrix<l,N>::tensor_reduced srhs(rhs);
+  return lhs*srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator * (const typename iScalar<l>::scalar_type & lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
+
+////////////////////////////////////////////////////////////////////
+// Double support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l> inline iScalar<l> operator * (double lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l,int N> inline iVector<l,N> operator * (double lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator * (double lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
+
+////////////////////////////////////////////////////////////////////
+// Complex support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,ComplexD rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l> inline iScalar<l> operator * (ComplexD lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,ComplexD rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l,int N> inline iVector<l,N> operator * (ComplexD lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,ComplexD rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator * (ComplexD lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
+
+////////////////////////////////////////////////////////////////////
+// Integer support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l> inline iScalar<l> operator * (Integer lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l,int N> inline iVector<l,N> operator * (Integer lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs*srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator * (Integer lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
+
+
+
+///////////////////////////////////////////////////////////////////////////////////////////////
+// addition by fundamental scalar type applies to matrix(down diag) and scalar
+///////////////////////////////////////////////////////////////////////////////////////////////
+template<class l,int N> inline iScalar<l> operator + (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
+{
+  typename iScalar<l>::tensor_reduced srhs(rhs);
+  return lhs+srhs;
+}
+template<class l,int N> inline iScalar<l> operator + (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
+{
+  typename iMatrix<l,N>::tensor_reduced srhs(rhs);
+  return lhs+srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator + (const typename iScalar<l>::scalar_type lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
+
+////////////////////////////////////////////////////////////////////
+// Double support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator + (const iScalar<l>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs+srhs;
+}
+template<class l> inline iScalar<l> operator + (double lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs+srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator + (double lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
+
+
+// Integer support cast to scalar type through constructor
+
+
+template<class l> inline iScalar<l> operator + (const iScalar<l>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs+srhs;
+}
+
+template<class l> inline iScalar<l> operator + (Integer lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
+
+template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs+srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator + (Integer lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
+
+
+///////////////////////////////////////////////////////////////////////////////////////////////
+// subtraction of fundamental scalar type applies to matrix(down diag) and scalar
+///////////////////////////////////////////////////////////////////////////////////////////////
+template<class l,int N> inline iScalar<l> operator - (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
+{
+  typename iScalar<l>::tensor_reduced srhs(rhs);
+  return lhs-srhs;
+}
+template<class l,int N> inline iScalar<l> operator - (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) 
+{
+  typename iScalar<l>::tensor_reduced slhs(lhs);
+  return slhs-rhs;
+}
+
+template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
+{
+  typename iScalar<l>::tensor_reduced srhs(rhs);
+  return lhs-srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator - (const typename iScalar<l>::scalar_type lhs,const iMatrix<l,N>& rhs) 
+{
+  typename iScalar<l>::tensor_reduced slhs(lhs);
+  return slhs-rhs;
+}
+
+////////////////////////////////////////////////////////////////////
+// Double support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator - (const iScalar<l>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs-srhs;
+}
+template<class l> inline iScalar<l> operator - (double lhs,const iScalar<l>& rhs) 
+{
+  typename iScalar<l>::scalar_type t(lhs);
+  typename iScalar<l>::tensor_reduced slhs(t);
+  return slhs-rhs;
+}
+
+template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,double rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs-srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator - (double lhs,const iMatrix<l,N>& rhs) 
+{
+  typename iScalar<l>::scalar_type t(lhs);
+  typename iScalar<l>::tensor_reduced slhs(t);
+  return slhs-rhs;
+}
+
+////////////////////////////////////////////////////////////////////
+// Integer support; cast to "scalar_type" through constructor
+////////////////////////////////////////////////////////////////////
+template<class l> inline iScalar<l> operator - (const iScalar<l>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs-srhs;
+}
+template<class l> inline iScalar<l> operator - (Integer lhs,const iScalar<l>& rhs) 
+{
+  typename iScalar<l>::scalar_type t(lhs);
+  typename iScalar<l>::tensor_reduced slhs(t);
+  return slhs-rhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,Integer rhs) 
+{
+  typename iScalar<l>::scalar_type t(rhs);
+  typename iScalar<l>::tensor_reduced srhs(t);
+  return lhs-srhs;
+}
+template<class l,int N> inline iMatrix<l,N> operator - (Integer lhs,const iMatrix<l,N>& rhs) 
+{
+  typename iScalar<l>::scalar_type t(lhs);
+  typename iScalar<l>::tensor_reduced slhs(t);
+  return slhs-rhs;
+}
+
+
+}
+#endif
diff --git a/lib/Grid_math_arith_sub.h b/lib/Grid_math_arith_sub.h
new file mode 100644
index 00000000..3a58bf4c
--- /dev/null
+++ b/lib/Grid_math_arith_sub.h
@@ -0,0 +1,134 @@
+#ifndef GRID_MATH_ARITH_SUB_H
+#define GRID_MATH_ARITH_SUB_H
+
+namespace Grid {
+
+
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    /////////////////////////////////////////// SUB         ///////////////////////////////////////////
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    
+
+// SUB is simple for now; cannot mix types and straightforward template
+// Scalar +/- Scalar
+// Vector +/- Vector
+// Matrix +/- Matrix
+// Matrix /- scalar
+template<class vtype,class ltype,class rtype> inline void sub(iScalar<vtype> * __restrict__ ret,
+                                                              const iScalar<ltype> * __restrict__ lhs,
+                                                              const iScalar<rtype> * __restrict__ rhs)
+{
+    sub(&ret->_internal,&lhs->_internal,&rhs->_internal);
+}
+
+template<class vtype,class ltype,class rtype,int N> inline void sub(iVector<vtype,N> * __restrict__ ret,
+                                                                    const iVector<ltype,N> * __restrict__ lhs,
+                                                                    const iVector<rtype,N> * __restrict__ rhs)
+{
+    for(int c=0;c<N;c++){
+        ret->_internal[c]=lhs->_internal[c]-rhs->_internal[c];
+    }
+    return;
+}
+template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
+                                                                     const iMatrix<ltype,N> * __restrict__ lhs,
+                                                                     const iMatrix<rtype,N> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        sub(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal[c1][c2]);
+    }}
+    return;
+}
+template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
+                                                                     const iScalar<ltype> * __restrict__ lhs,
+                                                                     const iMatrix<rtype,N> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        if ( c1!=c2) {
+            sub(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
+        } else {
+            // Fails -- need unary minus. Catalogue other unops?
+            ret->_internal[c1][c2]=zero;
+            ret->_internal[c1][c2]=ret->_internal[c1][c2]-rhs->_internal[c1][c2];
+
+        }
+    }}
+    return;
+}
+template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
+                                                                     const iMatrix<ltype,N> * __restrict__ lhs,
+                                                                     const iScalar<rtype> * __restrict__ rhs){
+    for(int c2=0;c2<N;c2++){
+    for(int c1=0;c1<N;c1++){
+        if ( c1!=c2)
+            sub(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
+        else
+            ret->_internal[c1][c2]=lhs->_internal[c1][c2];
+    }}
+    return;
+}
+
+template<class v> void vprefetch(const iScalar<v> &vv)
+{
+  vprefetch(vv._internal);
+}
+template<class v,int N> void vprefetch(const iVector<v,N> &vv)
+{
+  for(int i=0;i<N;i++){
+    vprefetch(vv._internal[i]);
+  }
+}
+template<class v,int N> void vprefetch(const iMatrix<v,N> &vv)
+{
+  for(int i=0;i<N;i++){
+  for(int j=0;j<N;j++){
+    vprefetch(vv._internal[i][j]);
+  }}
+}
+
+    // - operator for scalar, vector, matrix
+template<class ltype,class rtype> inline auto
+operator - (const iScalar<ltype>& lhs, const iScalar<rtype>& rhs) -> iScalar<decltype(lhs._internal - rhs._internal)>
+{
+    typedef iScalar<decltype(lhs._internal-rhs._internal)> ret_t;
+    ret_t ret;
+    sub(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class ltype,class rtype,int N>
+inline auto operator - (const iVector<ltype,N>& lhs,const iVector<rtype,N>& rhs) ->iVector<decltype(lhs._internal[0]-rhs._internal[0]),N>
+{
+    typedef iVector<decltype(lhs._internal[0]-rhs._internal[0]),N> ret_t;
+    ret_t ret;
+    sub(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class ltype,class rtype,int N>
+inline auto operator - (const iMatrix<ltype,N>& lhs,const iMatrix<rtype,N>& rhs) ->iMatrix<decltype(lhs._internal[0][0]-rhs._internal[0][0]),N>
+{
+    typedef iMatrix<decltype(lhs._internal[0][0]-rhs._internal[0][0]),N> ret_t;
+    ret_t ret;
+    sub(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class ltype,class rtype,int N>
+inline auto operator - (const iScalar<ltype>& lhs,const iMatrix<rtype,N>& rhs)->iMatrix<decltype(lhs._internal-rhs._internal[0][0]),N>
+{
+    typedef iMatrix<decltype(lhs._internal-rhs._internal[0][0]),N> ret_t;
+    ret_t ret;
+    sub(&ret,&lhs,&rhs);
+    return ret;
+}
+template<class ltype,class rtype,int N>
+inline auto operator - (const iMatrix<ltype,N>& lhs,const iScalar<rtype>& rhs)->iMatrix<decltype(lhs._internal[0][0]-rhs._internal),N>
+{
+    typedef iMatrix<decltype(lhs._internal[0][0]-rhs._internal),N> ret_t;
+    ret_t ret;
+    sub(&ret,&lhs,&rhs);
+    return ret;
+}
+
+
+}
+
+#endif

From d964d01d6aeb0b223a596f3712b1fc22ef52b834 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 20:44:19 +0100
Subject: [PATCH 075/429] Shrinking and organising the files

---
 lib/Grid.h                              |   2 +-
 lib/Grid_QCD.h                          |  24 -
 lib/Grid_cshift_common.h                |  19 +-
 lib/Grid_cshift_mpi.h                   |  15 +-
 lib/Grid_cshift_none.h                  |   6 +-
 lib/Grid_lattice.h                      | 630 ++----------------------
 lib/Grid_lattice_arith.h                | 160 ++++++
 lib/Grid_lattice_conformable.h          |  14 +
 lib/Grid_lattice_coordinate.h           |  45 ++
 lib/Grid_lattice_local.h                |  54 ++
 lib/Grid_lattice_peekpoke.h             | 143 ++++++
 lib/Grid_lattice_reality.h              |  52 ++
 lib/Grid_lattice_reduction.h            |  45 ++
 lib/Grid_lattice_rng.h                  |  32 ++
 lib/Grid_lattice_trace.h                |  43 ++
 lib/Grid_lattice_transfer.h             |  46 ++
 lib/Grid_lattice_transpose.h            |  39 ++
 lib/{Grid_predicated.h => Grid_where.h} |  19 +-
 18 files changed, 750 insertions(+), 638 deletions(-)
 create mode 100644 lib/Grid_lattice_arith.h
 create mode 100644 lib/Grid_lattice_conformable.h
 create mode 100644 lib/Grid_lattice_coordinate.h
 create mode 100644 lib/Grid_lattice_local.h
 create mode 100644 lib/Grid_lattice_peekpoke.h
 create mode 100644 lib/Grid_lattice_reality.h
 create mode 100644 lib/Grid_lattice_reduction.h
 create mode 100644 lib/Grid_lattice_rng.h
 create mode 100644 lib/Grid_lattice_trace.h
 create mode 100644 lib/Grid_lattice_transfer.h
 create mode 100644 lib/Grid_lattice_transpose.h
 rename lib/{Grid_predicated.h => Grid_where.h} (77%)

diff --git a/lib/Grid.h b/lib/Grid.h
index c340d678..093302cb 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -1,5 +1,5 @@
 //
-//  Grid.cpp
+//  Grid.h
 //  simd
 //
 //  Created by Peter Boyle on 09/05/2014.
diff --git a/lib/Grid_QCD.h b/lib/Grid_QCD.h
index 51ef1774..c405662d 100644
--- a/lib/Grid_QCD.h
+++ b/lib/Grid_QCD.h
@@ -139,30 +139,6 @@ namespace QCD {
       return peekIndex<LorentzIndex>(rhs,i,j);
     }
 
-    // FIXME this is rather generic and should find a way to place it earlier.
-     inline void LatticeCoordinate(LatticeInteger &l,int mu){
-      GridBase *grid = l._grid;
-      int Nsimd = grid->iSites();
-      std::vector<int> gcoor;
-      std::vector<Integer> mergebuf(Nsimd);
-      std::vector<Integer *> mergeptr(Nsimd);
-      for(int o=0;o<grid->oSites();o++){
-	for(int i=0;i<grid->iSites();i++){
-	  grid->RankIndexToGlobalCoor(grid->ThisRank(),o,i,gcoor);
-	  //	  grid->RankIndexToGlobalCoor(0,o,i,gcoor);
-	  mergebuf[i]=gcoor[mu];
-	  mergeptr[i]=&mergebuf[i];
-	}
-	merge(l._odata[o],mergeptr);
-      }
-    };
-
-#include <Grid_predicated.h>
-
-#if 0
-
-#endif
-
 }   //namespace QCD
 } // Grid
 #endif
diff --git a/lib/Grid_cshift_common.h b/lib/Grid_cshift_common.h
index 2910151c..8c4f1e1d 100644
--- a/lib/Grid_cshift_common.h
+++ b/lib/Grid_cshift_common.h
@@ -1,10 +1,11 @@
 #ifndef _GRID_CSHIFT_COMMON_H_
 #define _GRID_CSHIFT_COMMON_H_
 
+namespace Grid {
 //////////////////////////////////////////////////////
 // Gather for when there is no need to SIMD split
 //////////////////////////////////////////////////////
-friend void Gather_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllocator<vobj> > &buffer,             int dimension,int plane,int cbmask)
+template<class vobj> void Gather_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllocator<vobj> > &buffer,             int dimension,int plane,int cbmask)
 {
   int rd = rhs._grid->_rdimensions[dimension];
 
@@ -48,7 +49,7 @@ friend void Gather_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllo
 //////////////////////////////////////////////////////
 // Gather for when there *is* need to SIMD split
 //////////////////////////////////////////////////////
-friend void Gather_plane_extract(Lattice<vobj> &rhs,std::vector<scalar_type *> pointers,int dimension,int plane,int cbmask)
+ template<class vobj,class scalar_type> void Gather_plane_extract(Lattice<vobj> &rhs,std::vector<scalar_type *> pointers,int dimension,int plane,int cbmask)
 {
   int rd = rhs._grid->_rdimensions[dimension];
 
@@ -91,7 +92,7 @@ friend void Gather_plane_extract(Lattice<vobj> &rhs,std::vector<scalar_type *> p
 //////////////////////////////////////////////////////
 // Scatter for when there is no need to SIMD split
 //////////////////////////////////////////////////////
-friend void Scatter_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllocator<vobj> > &buffer,             int dimension,int plane,int cbmask)
+template<class vobj> void Scatter_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllocator<vobj> > &buffer,             int dimension,int plane,int cbmask)
 {
   int rd = rhs._grid->_rdimensions[dimension];
 
@@ -134,7 +135,7 @@ friend void Scatter_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAll
 //////////////////////////////////////////////////////
 // Scatter for when there *is* need to SIMD split
 //////////////////////////////////////////////////////
-friend void Scatter_plane_merge(Lattice<vobj> &rhs,std::vector<scalar_type *> pointers,int dimension,int plane,int cbmask)
+ template<class vobj,class scalar_type> void Scatter_plane_merge(Lattice<vobj> &rhs,std::vector<scalar_type *> pointers,int dimension,int plane,int cbmask)
 {
   int rd = rhs._grid->_rdimensions[dimension];
 
@@ -177,7 +178,7 @@ friend void Scatter_plane_merge(Lattice<vobj> &rhs,std::vector<scalar_type *> po
 //////////////////////////////////////////////////////
 // local to node block strided copies
 //////////////////////////////////////////////////////
-friend void Copy_plane(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int dimension,int lplane,int rplane,int cbmask)
+template<class vobj> void Copy_plane(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int dimension,int lplane,int rplane,int cbmask)
 {
   int rd = rhs._grid->_rdimensions[dimension];
 
@@ -219,7 +220,7 @@ friend void Copy_plane(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int dimension,int
   }
 }
 
-friend void Copy_plane_permute(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int dimension,int lplane,int rplane,int cbmask,int permute_type)
+template<class vobj> void Copy_plane_permute(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int dimension,int lplane,int rplane,int cbmask,int permute_type)
 {
   int rd = rhs._grid->_rdimensions[dimension];
 
@@ -265,7 +266,7 @@ friend void Copy_plane_permute(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int dimens
 //////////////////////////////////////////////////////
 // Local to node Cshift
 //////////////////////////////////////////////////////
-friend void Cshift_local(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimension,int shift)
+template<class vobj> void Cshift_local(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimension,int shift)
 {
   int sshift[2];
 
@@ -280,7 +281,7 @@ friend void Cshift_local(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimension,int
   }
 }
 
-friend Lattice<vobj> Cshift_local(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int shift,int cbmask)
+template<class vobj> Lattice<vobj> Cshift_local(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int shift,int cbmask)
 {
   GridBase *grid = rhs._grid;
   int fd = grid->_fdimensions[dimension];
@@ -322,5 +323,5 @@ friend Lattice<vobj> Cshift_local(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dime
   }
   return ret;
 }
-
+}
 #endif
diff --git a/lib/Grid_cshift_mpi.h b/lib/Grid_cshift_mpi.h
index f51becf3..c078c7d8 100644
--- a/lib/Grid_cshift_mpi.h
+++ b/lib/Grid_cshift_mpi.h
@@ -6,7 +6,9 @@
 #define MIN(x,y) ((x)>(y)?(y):(x))
 #endif
 
-friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
+namespace Grid { 
+
+template<class vobj> Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
 {
   typedef typename vobj::vector_type vector_type;
   typedef typename vobj::scalar_type scalar_type;
@@ -37,7 +39,7 @@ friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
   return ret;
 }
 
-friend void Cshift_comms(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimension,int shift)
+template<class vobj> void Cshift_comms(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimension,int shift)
 {
   int sshift[2];
 
@@ -52,7 +54,7 @@ friend void Cshift_comms(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimension,int
   }
 }
 
-friend void Cshift_comms_simd(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimension,int shift)
+template<class vobj> void Cshift_comms_simd(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimension,int shift)
 {
   int sshift[2];
 
@@ -67,8 +69,7 @@ friend void Cshift_comms_simd(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimensio
   }
 }
 
-
-friend void Cshift_comms(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int shift,int cbmask)
+template<class vobj> void Cshift_comms(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int shift,int cbmask)
 {
   typedef typename vobj::vector_type vector_type;
   typedef typename vobj::scalar_type scalar_type;
@@ -127,8 +128,7 @@ friend void Cshift_comms(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int
   }
 }
 
-
-friend void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int shift,int cbmask)
+template<class vobj> void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int shift,int cbmask)
 {
   GridBase *grid=rhs._grid;
   const int Nsimd = grid->Nsimd();
@@ -260,4 +260,5 @@ friend void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimensi
     }
   }
 }
+}
 #endif
diff --git a/lib/Grid_cshift_none.h b/lib/Grid_cshift_none.h
index 7f4f7bc1..3485a63e 100644
--- a/lib/Grid_cshift_none.h
+++ b/lib/Grid_cshift_none.h
@@ -1,12 +1,12 @@
 #ifndef _GRID_CSHIFT_NONE_H_
 #define _GRID_CSHIFT_NONE_H_
-
-friend Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
+namespace Grid {
+template<class vobj> Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
 {
   Lattice<vobj> ret(rhs._grid);
   ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
   Cshift_local(ret,rhs,dimension,shift);
   return ret;
 }
-        
+}
 #endif
diff --git a/lib/Grid_lattice.h b/lib/Grid_lattice.h
index be95caa0..cc115c3c 100644
--- a/lib/Grid_lattice.h
+++ b/lib/Grid_lattice.h
@@ -3,10 +3,10 @@
 
 namespace Grid {
 
-// TODO: Indexing ()
+// TODO: 
 //       mac,real,imag
-//
-// Functionality required:
+
+// Functionality:
 //     -=,+=,*=,()
 //     add,+,sub,-,mult,mac,*
 //     adj,conj
@@ -17,7 +17,6 @@ namespace Grid {
 //     innerProduct,outerProduct,
 //     localNorm2
 //     localInnerProduct
-//     
 
 extern int GridCshiftPermuteMap[4][16];
 
@@ -33,51 +32,39 @@ public:
     typedef typename vobj::scalar_type scalar_type;
     typedef typename vobj::vector_type vector_type;
 
+    //////////////////////////////////////////////////////////////////
+    // Constructor requires "grid" passed.
+    // what about a default grid?
+    //////////////////////////////////////////////////////////////////
     Lattice(GridBase *grid) : _grid(grid) {
         _odata.reserve(_grid->oSites());
         assert((((uint64_t)&_odata[0])&0xF) ==0);
         checkerboard=0;
     }
 
-#include <Grid_cshift.h>
-   
-    template<class obj1,class obj2>
-    friend void conformable(const Lattice<obj1> &lhs,const Lattice<obj2> &rhs);
+    template<class sobj> inline Lattice<vobj> & operator = (const sobj & r){
+#pragma omp parallel for
+        for(int ss=0;ss<_grid->oSites();ss++){
+            this->_odata[ss]=r;
+        }
+        return *this;
+    }
 
-    // FIXME Performance difference between operator * and mult is troubling.
-    // Auto move constructor seems to lose surprisingly much.
-
-    // Site wise binary operations
-    // We eliminate a temporary object assignment if use the mult,add,sub routines.
-    // For the operator versions we rely on move constructor to eliminate the
-    // vector copy back.
-    template<class obj1,class obj2,class obj3>
-    friend void mult(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs);
-
-    template<class obj1,class obj2,class obj3>
-    friend void mac(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs);
-
-    template<class obj1,class obj2,class obj3>
-    friend void sub(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs);
-
-    template<class obj1,class obj2,class obj3>
-    friend void add(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs);
+    // *=,+=,-= operators inherit behvour from correspond */+/- operation
+    template<class T> inline Lattice<vobj> &operator *=(const T &r) {
+        *this = (*this)*r;
+        return *this;
+    }
 
+    template<class T> inline Lattice<vobj> &operator -=(const T &r) {
+        *this = (*this)-r;
+        return *this;
+    }
+    template<class T> inline Lattice<vobj> &operator +=(const T &r) {
+        *this = (*this)+r;
+        return *this;
+    }
     
-    friend void axpy(Lattice<vobj> &ret,double a,const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
-        conformable(lhs,rhs);
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-            axpy(&ret._odata[ss],a,&lhs._odata[ss],&rhs._odata[ss]);
-        }
-    }
-    friend void axpy(Lattice<vobj> &ret,std::complex<double> a,const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
-        conformable(lhs,rhs);
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-            axpy(&ret._odata[ss],a,&lhs._odata[ss],&rhs._odata[ss]);
-        }
-    }
     inline friend Lattice<vobj> operator / (const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
         conformable(lhs,rhs);
         Lattice<vobj> ret(lhs._grid);
@@ -87,550 +74,23 @@ public:
         }
         return ret;
     };
-
-    template<class sobj>
-    inline Lattice<vobj> & operator = (const sobj & r){
-#pragma omp parallel for
-        for(int ss=0;ss<_grid->oSites();ss++){
-            this->_odata[ss]=r;
-        }
-        return *this;
-    }
-    
-    // Poke a scalar object into the SIMD array
-    template<class sobj>
-    friend void pokeSite(const sobj &s,Lattice<vobj> &l,std::vector<int> &site){
-
-      GridBase *grid=l._grid;
-
-      typedef typename vobj::scalar_type scalar_type;
-      typedef typename vobj::vector_type vector_type;
-
-      int Nsimd = grid->Nsimd();
-
-      assert( l.checkerboard== l._grid->CheckerBoard(site));
-      assert( sizeof(sobj)*Nsimd == sizeof(vobj));
-
-      int rank,odx,idx;
-      grid->GlobalCoorToRankIndex(rank,odx,idx,site);
-
-      // Optional to broadcast from node 0.
-      grid->Broadcast(0,s);
-
-      std::vector<sobj> buf(Nsimd);
-      std::vector<scalar_type *> pointers(Nsimd);  
-      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
-
-      // extract-modify-merge cycle is easiest way and this is not perf critical
-      extract(l._odata[odx],pointers);
-      
-      buf[idx] = s;
-
-      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
-      merge(l._odata[odx],pointers);
-
-      return;
-    };
-    
-    // Peek a scalar object from the SIMD array
-    template<class sobj>
-    friend void peekSite(sobj &s,Lattice<vobj> &l,std::vector<int> &site){
-        
-      GridBase *grid=l._grid;
-
-      typedef typename vobj::scalar_type scalar_type;
-      typedef typename vobj::vector_type vector_type;
-
-      int Nsimd = grid->Nsimd();
-
-      assert( l.checkerboard== l._grid->CheckerBoard(site));
-      assert( sizeof(sobj)*Nsimd == sizeof(vobj));
-
-      int rank,odx,idx;
-      grid->GlobalCoorToRankIndex(rank,odx,idx,site);
-      std::vector<sobj> buf(Nsimd);
-      std::vector<scalar_type *> pointers(Nsimd);  
-      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
-
-      extract(l._odata[odx],pointers);
-      
-      s = buf[idx];
-      grid->Broadcast(rank,s);
-
-      return;
-    };
-    
-    // FIXME Randomise; deprecate this
-    friend void random(Lattice<vobj> &l){
-        Real *v_ptr = (Real *)&l._odata[0];
-        size_t v_len = l._grid->oSites()*sizeof(vobj);
-        size_t d_len = v_len/sizeof(Real);
-	
-        for(int i=0;i<d_len;i++){
-
-            v_ptr[i]=drand48();
-        }
-    };
-
-    // FIXME for debug; deprecate this; made obscelete by 
-    // LatticeCoordinate();
-    friend void lex_sites(Lattice<vobj> &l){
-      Real *v_ptr = (Real *)&l._odata[0];
-      size_t o_len = l._grid->oSites();
-      size_t v_len = sizeof(vobj)/sizeof(vRealF);
-      size_t vec_len = vRealF::Nsimd();
-
-      for(int i=0;i<o_len;i++){
-	for(int j=0;j<v_len;j++){
-          for(int vv=0;vv<vec_len;vv+=2){
-	    v_ptr[i*v_len*vec_len+j*vec_len+vv  ]= i+vv*500;
-	    v_ptr[i*v_len*vec_len+j*vec_len+vv+1]= i+vv*500;
-	  }
-	}}
-    }
-    
-    // FIXME Implement a consistent seed management strategy
-    friend void gaussian(Lattice<vobj> &l){
-        // Zero mean, unit variance.
-        std::normal_distribution<double> distribution(0.0,1.0);
-        Real *v_ptr = (Real *)&l._odata[0];
-        size_t v_len = l._grid->oSites()*sizeof(vobj);
-        size_t d_len = v_len/sizeof(Real);
-
-        for(int i=0;i<d_len;i++){
-	  v_ptr[i]= drand48();
-        }
-    };
-
-    // Unary functions and Unops
-    friend inline Lattice<vobj> operator -(const Lattice<vobj> &r) {
-        Lattice<vobj> ret(r._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<r._grid->oSites();ss++){
-            ret._odata[ss]= -r._odata[ss];
-        }
-        return ret;
-    }
-    // *=,+=,-= operators inherit behvour from correspond */+/- operation
-    template<class T>
-    inline Lattice<vobj> &operator *=(const T &r) {
-        *this = (*this)*r;
-        return *this;
-    }
-    template<class T>
-    inline Lattice<vobj> &operator -=(const T &r) {
-        *this = (*this)-r;
-        return *this;
-    }
-    template<class T>
-    inline Lattice<vobj> &operator +=(const T &r) {
-        *this = (*this)+r;
-        return *this;
-    }
-    
-    inline friend Lattice<vobj> adj(const Lattice<vobj> &lhs){
-        Lattice<vobj> ret(lhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-            ret._odata[ss] = adj(lhs._odata[ss]);
-        }
-        return ret;
-    };
-
-    inline friend Lattice<vobj> transpose(const Lattice<vobj> &lhs){
-        Lattice<vobj> ret(lhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-            ret._odata[ss] = transpose(lhs._odata[ss]);
-        }
-        return ret;
-    };
-
-
-    inline friend Lattice<vobj> conj(const Lattice<vobj> &lhs){
-        Lattice<vobj> ret(lhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-            ret._odata[ss] = conj(lhs._odata[ss]);
-        }
-        return ret;
-    };
-
-    // remove and insert a half checkerboard
-    friend void pickCheckerboard(int cb,Lattice<vobj> &half,const Lattice<vobj> &full){
-      half.checkerboard = cb;
-      int ssh=0;
-#pragma omp parallel for
-      for(int ss=0;ss<full._grid->oSites();ss++){
-	std::vector<int> coor;
-	int cbos;
-	
-	full._grid->oCoorFromOindex(coor,ss);
-	cbos=half._grid->CheckerBoard(coor);
-	
-	if (cbos==cb) {
-
-	  half._odata[ssh] = full._odata[ss];
-	  ssh++;
-	}
-      }
-    }
-    friend void setCheckerboard(Lattice<vobj> &full,const Lattice<vobj> &half){
-      int cb = half.checkerboard;
-      int ssh=0;
-#pragma omp parallel for
-      for(int ss=0;ss<full._grid->oSites();ss++){
-	std::vector<int> coor;
-	int cbos;
-	
-	full._grid->oCoorFromOindex(coor,ss);
-	cbos=half._grid->CheckerBoard(coor);
-
-	if (cbos==cb) {
-	  full._odata[ss]=half._odata[ssh];
-	  ssh++;
-	}
-      }
-    }
-}; // class Lattice
-
-    template<class obj1,class obj2>
-    void conformable(const Lattice<obj1> &lhs,const Lattice<obj2> &rhs)
-    {
-        assert(lhs._grid == rhs._grid);
-        assert(lhs.checkerboard == rhs.checkerboard);
-    }
-
-    template<class obj1,class obj2,class obj3>
-    void mult(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
-        conformable(lhs,rhs);
-	uint32_t vec_len = lhs._grid->oSites();
-#pragma omp parallel for
-        for(int ss=0;ss<vec_len;ss++){
-	  mult(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
-        }
-    }
-
-    template<class obj1,class obj2,class obj3>
-    void mac(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
-        conformable(lhs,rhs);
-	uint32_t vec_len = lhs._grid->oSites();
-#pragma omp parallel for
-        for(int ss=0;ss<vec_len;ss++){
-	  mac(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
-        }
-    }
-
-    template<class obj1,class obj2,class obj3>
-    void sub(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
-        conformable(lhs,rhs);
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-            sub(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
-        }
-    }
-    template<class obj1,class obj2,class obj3>
-    void add(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
-        conformable(lhs,rhs);
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-            add(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
-        }
-    }
-
-    // Lattice BinOp Lattice,
-    template<class left,class right>
-    inline auto operator * (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]*rhs._odata[0])>
-    {
-        //NB mult performs conformable check. Do not reapply here for performance.
-      Lattice<decltype(lhs._odata[0]*rhs._odata[0])> ret(rhs._grid);
-      mult(ret,lhs,rhs);
-      return ret;
-    }
-    template<class left,class right>
-    inline auto operator + (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]*rhs._odata[0])>
-    {
-        //NB mult performs conformable check. Do not reapply here for performance.
-        Lattice<decltype(lhs._odata[0]*rhs._odata[0])> ret(rhs._grid);
-        add(ret,lhs,rhs);
-        return ret;
-    }
-    template<class left,class right>
-    inline auto operator - (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]*rhs._odata[0])>
-    {
-        //NB mult performs conformable check. Do not reapply here for performance.
-        Lattice<decltype(lhs._odata[0]*rhs._odata[0])> ret(rhs._grid);
-        sub(ret,lhs,rhs);
-        return ret;
-    }
-
-    // Scalar BinOp Lattice ;generate return type
-    template<class left,class right>
-    inline auto operator * (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs*rhs._odata[0])>
-    {
-        Lattice<decltype(lhs*rhs._odata[0])> ret(rhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<rhs._grid->oSites(); ss++){
-            ret._odata[ss]=lhs*rhs._odata[ss];
-        }
-        return ret;
-    }
-    template<class left,class right>
-    inline auto operator + (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs*rhs._odata[0])>
-    {
-        Lattice<decltype(lhs*rhs._odata[0])> ret(rhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<rhs._grid->oSites(); ss++){
-            ret._odata[ss]=lhs+rhs._odata[ss];
-        }
-        return ret;
-    }
-    template<class left,class right>
-    inline auto operator - (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs*rhs._odata[0])>
-    {
-        Lattice<decltype(lhs*rhs._odata[0])> ret(rhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<rhs._grid->oSites(); ss++){
-            ret._odata[ss]=lhs-rhs._odata[ss];
-        }
-        return ret;
-    }
-    template<class left,class right>
-    inline auto operator * (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]*rhs)>
-    {
-        Lattice<decltype(lhs._odata[0]*rhs)> ret(lhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites(); ss++){
-            ret._odata[ss]=lhs._odata[ss]*rhs;
-        }
-        return ret;
-    }
-    template<class left,class right>
-    inline auto operator + (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]*rhs)>
-    {
-        Lattice<decltype(lhs._odata[0]*rhs)> ret(lhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<rhs._grid->oSites(); ss++){
-            ret._odata[ss]=lhs._odata[ss]+rhs;
-        }
-        return ret;
-    }
-    template<class left,class right>
-    inline auto operator - (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]*rhs)>
-    {
-        Lattice<decltype(lhs._odata[0]*rhs)> ret(lhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<rhs._grid->oSites(); ss++){
-            ret._odata[ss]=lhs._odata[ss]-rhs;
-        }
-        return ret;
-    }
-
-    ////////////////////////////////////////////////////////////////////////////////////////////////////
-    // Trace
-    ////////////////////////////////////////////////////////////////////////////////////////////////////
-    template<class vobj>
-    inline auto trace(const Lattice<vobj> &lhs)
-      -> Lattice<decltype(trace(lhs._odata[0]))>
-    {
-      Lattice<decltype(trace(lhs._odata[0]))> ret(lhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-            ret._odata[ss] = trace(lhs._odata[ss]);
-        }
-        return ret;
-    };
-    
-    ////////////////////////////////////////////////////////////////////////////////////////////////////
-    // Index level dependent operations
-    ////////////////////////////////////////////////////////////////////////////////////////////////////
-    template<int Index,class vobj>
-    inline auto traceIndex(const Lattice<vobj> &lhs)
-      -> Lattice<decltype(traceIndex<Index>(lhs._odata[0]))>
-    {
-      Lattice<decltype(traceIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
-#pragma omp parallel for
-      for(int ss=0;ss<lhs._grid->oSites();ss++){
-	ret._odata[ss] = traceIndex<Index>(lhs._odata[ss]);
-      }
-      return ret;
-    };
-    template<int Index,class vobj>
-    inline auto transposeIndex(const Lattice<vobj> &lhs)
-      -> Lattice<decltype(transposeIndex<Index>(lhs._odata[0]))>
-    {
-      Lattice<decltype(transposeIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-            ret._odata[ss] = transposeIndex<Index>(lhs._odata[ss]);
-        }
-        return ret;
-    };
-
-    // Fixme; this is problematic since the number of args is variable and 
-    // may mismatch...
-    template<int Index,class vobj>
-    inline auto peekIndex(const Lattice<vobj> &lhs)
-      -> Lattice<decltype(peekIndex<Index>(lhs._odata[0]))>
-    {
-      Lattice<decltype(peekIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-            ret._odata[ss] = peekIndex<Index>(lhs._odata[ss]);
-        }
-        return ret;
-    };
-    template<int Index,class vobj>
-      inline auto peekIndex(const Lattice<vobj> &lhs,int i)
-      -> Lattice<decltype(peekIndex<Index>(lhs._odata[0],i))>
-    {
-      Lattice<decltype(peekIndex<Index>(lhs._odata[0],i))> ret(lhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-	  ret._odata[ss] = peekIndex<Index>(lhs._odata[ss],i);
-        }
-        return ret;
-    };
-    template<int Index,class vobj>
-      inline auto peekIndex(const Lattice<vobj> &lhs,int i,int j)
-      -> Lattice<decltype(peekIndex<Index>(lhs._odata[0],i,j))>
-    {
-      Lattice<decltype(peekIndex<Index>(lhs._odata[0],i,j))> ret(lhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-	  ret._odata[ss] = peekIndex<Index>(lhs._odata[ss],i,j);
-        }
-        return ret;
-    };
-    ////////////////////////////////////////////////////////////////////////////////////////////////////
-    // Poke internal indices of a Lattice object
-    ////////////////////////////////////////////////////////////////////////////////////////////////////
-    template<int Index,class vobj> inline
-    void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0]))> & rhs)
-    {
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-	  pokeIndex<Index>(lhs._odata[ss],rhs._odata[ss]);
-	}      
-    }
-    template<int Index,class vobj> inline
-    void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0],0))> & rhs,int i)
-    {
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-	  pokeIndex<Index>(lhs._odata[ss],rhs._odata[ss],i);
-	}      
-    }
-    template<int Index,class vobj> inline
-    void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0],0,0))> & rhs,int i,int j)
-    {
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-	  pokeIndex<Index>(lhs._odata[ss],rhs._odata[ss],i,j);
-	}      
-    }
-
-    ////////////////////////////////////////////////////////////////////////////////////////////////////
-    // Reduction operations
-    ////////////////////////////////////////////////////////////////////////////////////////////////////
-    template<class vobj>
-    inline RealD norm2(const Lattice<vobj> &arg){
-
-      typedef typename vobj::scalar_type scalar;
-      typedef typename vobj::vector_type vector;
-      decltype(innerProduct(arg._odata[0],arg._odata[0])) vnrm=zero;
-      scalar nrm;
-      //FIXME make this loop parallelisable
-      vnrm=zero;
-      for(int ss=0;ss<arg._grid->oSites(); ss++){
-	vnrm = vnrm + innerProduct(arg._odata[ss],arg._odata[ss]);
-      }
-      vector vvnrm =TensorRemove(vnrm) ;
-      nrm = Reduce(vvnrm);
-      arg._grid->GlobalSum(nrm);
-      return real(nrm);
-    }
-
-    template<class vobj>
-    inline auto innerProduct(const Lattice<vobj> &left,const Lattice<vobj> &right) ->decltype(innerProduct(left._odata[0],right._odata[0]))
-    {
-      typedef typename vobj::scalar_type scalar;
-      decltype(innerProduct(left._odata[0],right._odata[0])) vnrm=zero;
-
-      scalar nrm;
-      //FIXME make this loop parallelisable
-      for(int ss=0;ss<left._grid->oSites(); ss++){
-	vnrm = vnrm + innerProduct(left._odata[ss],right._odata[ss]);
-      }
-      nrm = Reduce(vnrm);
-      right._grid->GlobalSum(nrm);
-      return nrm;
-    }
-
-    /////////////////////////////////////////////////////
-    // Non site reduced routines
-    /////////////////////////////////////////////////////
-
-    // localNorm2,
-    template<class vobj>
-    inline auto localNorm2 (const Lattice<vobj> &rhs)-> Lattice<typename vobj::tensor_reduced>
-    {
-      Lattice<typename vobj::tensor_reduced> ret(rhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<rhs._grid->oSites(); ss++){
-	  ret._odata[ss]=innerProduct(rhs._odata[ss],rhs._odata[ss]);
-        }
-        return ret;
-    }
-    
-    template<class vobj>
-    inline auto real(const Lattice<vobj> &z) -> Lattice<decltype(real(z._odata[0]))>
-    {
-      Lattice<decltype(real(z._odata[0]))> ret(z._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<z._grid->oSites();ss++){
-            ret._odata[ss] = real(z._odata[ss]);
-        }
-      return ret;
-    }
-
-    template<class vobj>
-    inline auto imag(const Lattice<vobj> &z) -> Lattice<decltype(imag(z._odata[0]))>
-    {
-      Lattice<decltype(imag(z._odata[0]))> ret(z._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<z._grid->oSites();ss++){
-            ret._odata[ss] = imag(z._odata[ss]);
-        }
-      return ret;
-    }
-
-    // localInnerProduct
-    template<class vobj>
-    inline auto localInnerProduct (const Lattice<vobj> &lhs,const Lattice<vobj> &rhs)
-      -> Lattice<typename vobj::tensor_reduced>
-    {
-      Lattice<typename vobj::tensor_reduced> ret(rhs._grid);
-#pragma omp parallel for
-      for(int ss=0;ss<rhs._grid->oSites(); ss++){
-	ret._odata[ss]=innerProduct(lhs._odata[ss],rhs._odata[ss]);
-      }
-      return ret;
-    }
-    
-    // outerProduct Scalar x Scalar -> Scalar
-    //              Vector x Vector -> Matrix
-    template<class ll,class rr>
-    inline auto outerProduct (const Lattice<ll> &lhs,const Lattice<rr> &rhs) -> Lattice<decltype(outerProduct(lhs._odata[0],rhs._odata[0]))>
-    {
-        Lattice<decltype(outerProduct(lhs._odata[0],rhs._odata[0]))> ret(rhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<rhs._grid->oSites(); ss++){
-            ret._odata[ss]=outerProduct(lhs._odata[ss],rhs._odata[ss]);
-        }
-        return ret;
-     }
-
-
+ }; // class Lattice
 }
+
+#include <Grid_lattice_conformable.h>
+#include <Grid_lattice_arith.h>
+#include <Grid_lattice_trace.h>
+#include <Grid_lattice_transpose.h>
+#include <Grid_lattice_local.h>
+#include <Grid_lattice_reduction.h>
+#include <Grid_lattice_peekpoke.h>
+#include <Grid_lattice_reality.h>
+#include <Grid_cshift.h>
+#include <Grid_where.h>
+#include <Grid_lattice_coordinate.h>
+#include <Grid_lattice_rng.h>
+#include <Grid_lattice_transfer.h>
+
+
+
 #endif
diff --git a/lib/Grid_lattice_arith.h b/lib/Grid_lattice_arith.h
new file mode 100644
index 00000000..468dca79
--- /dev/null
+++ b/lib/Grid_lattice_arith.h
@@ -0,0 +1,160 @@
+#ifndef GRID_LATTICE_ARITH_H
+#define GRID_LATTICE_ARITH_H
+
+namespace Grid {
+
+  template<class vobj>
+  inline Lattice<vobj> operator -(const Lattice<vobj> &r)
+  {
+    Lattice<vobj> ret(r._grid);
+#pragma omp parallel for
+    for(int ss=0;ss<r._grid->oSites();ss++){
+      ret._odata[ss]= -r._odata[ss];
+    }
+    return ret;
+  }
+  
+  template<class vobj>
+  inline void axpy(Lattice<vobj> &ret,double a,const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
+    conformable(lhs,rhs);
+#pragma omp parallel for
+    for(int ss=0;ss<lhs._grid->oSites();ss++){
+      axpy(&ret._odata[ss],a,&lhs._odata[ss],&rhs._odata[ss]);
+    }
+  }
+  template<class vobj>
+  inline void axpy(Lattice<vobj> &ret,std::complex<double> a,const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
+    conformable(lhs,rhs);
+#pragma omp parallel for
+    for(int ss=0;ss<lhs._grid->oSites();ss++){
+      axpy(&ret._odata[ss],a,&lhs._odata[ss],&rhs._odata[ss]);
+    }
+  }
+  
+  
+  template<class obj1,class obj2,class obj3>
+    void mult(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
+    conformable(lhs,rhs);
+    uint32_t vec_len = lhs._grid->oSites();
+#pragma omp parallel for
+    for(int ss=0;ss<vec_len;ss++){
+      mult(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
+    }
+  }
+  
+  template<class obj1,class obj2,class obj3>
+    void mac(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
+    conformable(lhs,rhs);
+    uint32_t vec_len = lhs._grid->oSites();
+#pragma omp parallel for
+    for(int ss=0;ss<vec_len;ss++){
+      mac(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
+    }
+  }
+  
+  template<class obj1,class obj2,class obj3>
+    void sub(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
+    conformable(lhs,rhs);
+#pragma omp parallel for
+    for(int ss=0;ss<lhs._grid->oSites();ss++){
+      sub(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
+    }
+  }
+  template<class obj1,class obj2,class obj3>
+    void add(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
+    conformable(lhs,rhs);
+#pragma omp parallel for
+    for(int ss=0;ss<lhs._grid->oSites();ss++){
+      add(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
+    }
+  }
+  
+  // Lattice BinOp Lattice,
+  template<class left,class right>
+    inline auto operator * (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]*rhs._odata[0])>
+  {
+    //NB mult performs conformable check. Do not reapply here for performance.
+    Lattice<decltype(lhs._odata[0]*rhs._odata[0])> ret(rhs._grid);
+    mult(ret,lhs,rhs);
+    return ret;
+  }
+  template<class left,class right>
+    inline auto operator + (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]*rhs._odata[0])>
+  {
+    //NB mult performs conformable check. Do not reapply here for performance.
+    Lattice<decltype(lhs._odata[0]*rhs._odata[0])> ret(rhs._grid);
+    add(ret,lhs,rhs);
+    return ret;
+  }
+  template<class left,class right>
+    inline auto operator - (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]*rhs._odata[0])>
+  {
+    //NB mult performs conformable check. Do not reapply here for performance.
+    Lattice<decltype(lhs._odata[0]*rhs._odata[0])> ret(rhs._grid);
+    sub(ret,lhs,rhs);
+    return ret;
+  }
+  
+  // Scalar BinOp Lattice ;generate return type
+  template<class left,class right>
+    inline auto operator * (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs*rhs._odata[0])>
+  {
+    Lattice<decltype(lhs*rhs._odata[0])> ret(rhs._grid);
+#pragma omp parallel for
+    for(int ss=0;ss<rhs._grid->oSites(); ss++){
+      ret._odata[ss]=lhs*rhs._odata[ss];
+    }
+        return ret;
+  }
+  template<class left,class right>
+    inline auto operator + (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs*rhs._odata[0])>
+    {
+      Lattice<decltype(lhs*rhs._odata[0])> ret(rhs._grid);
+#pragma omp parallel for
+      for(int ss=0;ss<rhs._grid->oSites(); ss++){
+	ret._odata[ss]=lhs+rhs._odata[ss];
+      }
+        return ret;
+    }
+  template<class left,class right>
+    inline auto operator - (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs*rhs._odata[0])>
+  {
+    Lattice<decltype(lhs*rhs._odata[0])> ret(rhs._grid);
+#pragma omp parallel for
+    for(int ss=0;ss<rhs._grid->oSites(); ss++){
+      ret._odata[ss]=lhs-rhs._odata[ss];
+    }
+    return ret;
+  }
+    template<class left,class right>
+      inline auto operator * (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]*rhs)>
+    {
+      Lattice<decltype(lhs._odata[0]*rhs)> ret(lhs._grid);
+#pragma omp parallel for
+      for(int ss=0;ss<lhs._grid->oSites(); ss++){
+            ret._odata[ss]=lhs._odata[ss]*rhs;
+      }
+      return ret;
+    }
+    template<class left,class right>
+      inline auto operator + (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]*rhs)>
+    {
+        Lattice<decltype(lhs._odata[0]*rhs)> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<rhs._grid->oSites(); ss++){
+	  ret._odata[ss]=lhs._odata[ss]+rhs;
+        }
+        return ret;
+    }
+    template<class left,class right>
+      inline auto operator - (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]*rhs)>
+    {
+      Lattice<decltype(lhs._odata[0]*rhs)> ret(lhs._grid);
+#pragma omp parallel for
+      for(int ss=0;ss<rhs._grid->oSites(); ss++){
+	ret._odata[ss]=lhs._odata[ss]-rhs;
+      }
+      return ret;
+    }
+}
+#endif
diff --git a/lib/Grid_lattice_conformable.h b/lib/Grid_lattice_conformable.h
new file mode 100644
index 00000000..b9af33b4
--- /dev/null
+++ b/lib/Grid_lattice_conformable.h
@@ -0,0 +1,14 @@
+#ifndef GRID_LATTICE_CONFORMABLE_H
+#define GRID_LATTICE_CONFORMABLE_H
+
+namespace Grid {
+
+    template<class obj1,class obj2>
+    void conformable(const Lattice<obj1> &lhs,const Lattice<obj2> &rhs)
+    {
+        assert(lhs._grid == rhs._grid);
+        assert(lhs.checkerboard == rhs.checkerboard);
+    }
+
+}
+#endif
diff --git a/lib/Grid_lattice_coordinate.h b/lib/Grid_lattice_coordinate.h
new file mode 100644
index 00000000..0a71ec8b
--- /dev/null
+++ b/lib/Grid_lattice_coordinate.h
@@ -0,0 +1,45 @@
+#ifndef GRID_LATTICE_COORDINATE_H
+#define GRID_LATTICE_COORDINATE_H
+
+namespace Grid {
+
+    template<class iobj> inline void LatticeCoordinate(Lattice<iobj> &l,int mu)
+    {
+      GridBase *grid = l._grid;
+      int Nsimd = grid->iSites();
+      std::vector<int> gcoor;
+      std::vector<Integer> mergebuf(Nsimd);
+      std::vector<Integer *> mergeptr(Nsimd);
+      vInteger vI;
+      for(int o=0;o<grid->oSites();o++){
+	for(int i=0;i<grid->iSites();i++){
+	  grid->RankIndexToGlobalCoor(grid->ThisRank(),o,i,gcoor);
+	  //	  grid->RankIndexToGlobalCoor(0,o,i,gcoor);
+	  mergebuf[i]=gcoor[mu];
+	  mergeptr[i]=&mergebuf[i];
+	}
+	merge(vI,mergeptr);
+	l._odata[o]=vI;
+      }
+    };
+
+    // LatticeCoordinate();
+    // FIXME for debug; deprecate this; made obscelete by 
+    template<class vobj> void lex_sites(Lattice<vobj> &l){
+      Real *v_ptr = (Real *)&l._odata[0];
+      size_t o_len = l._grid->oSites();
+      size_t v_len = sizeof(vobj)/sizeof(vRealF);
+      size_t vec_len = vRealF::Nsimd();
+
+      for(int i=0;i<o_len;i++){
+	for(int j=0;j<v_len;j++){
+          for(int vv=0;vv<vec_len;vv+=2){
+	    v_ptr[i*v_len*vec_len+j*vec_len+vv  ]= i+vv*500;
+	    v_ptr[i*v_len*vec_len+j*vec_len+vv+1]= i+vv*500;
+	  }
+	}}
+    }
+
+
+}
+#endif
diff --git a/lib/Grid_lattice_local.h b/lib/Grid_lattice_local.h
new file mode 100644
index 00000000..181f5be2
--- /dev/null
+++ b/lib/Grid_lattice_local.h
@@ -0,0 +1,54 @@
+#ifndef GRID_LATTICE_LOCALREDUCTION_H
+#define GRID_LATTICE_LOCALREDUCTION_H
+
+///////////////////////////////////////////////
+// localInner, localNorm, outerProduct
+///////////////////////////////////////////////
+
+namespace Grid {
+
+    /////////////////////////////////////////////////////
+    // Non site, reduced locally reduced routines
+    /////////////////////////////////////////////////////
+
+    // localNorm2,
+    template<class vobj>
+    inline auto localNorm2 (const Lattice<vobj> &rhs)-> Lattice<typename vobj::tensor_reduced>
+    {
+      Lattice<typename vobj::tensor_reduced> ret(rhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<rhs._grid->oSites(); ss++){
+	  ret._odata[ss]=innerProduct(rhs._odata[ss],rhs._odata[ss]);
+        }
+        return ret;
+    }
+    
+    // localInnerProduct
+    template<class vobj>
+    inline auto localInnerProduct (const Lattice<vobj> &lhs,const Lattice<vobj> &rhs)
+      -> Lattice<typename vobj::tensor_reduced>
+    {
+      Lattice<typename vobj::tensor_reduced> ret(rhs._grid);
+#pragma omp parallel for
+      for(int ss=0;ss<rhs._grid->oSites(); ss++){
+	ret._odata[ss]=innerProduct(lhs._odata[ss],rhs._odata[ss]);
+      }
+      return ret;
+    }
+    
+    // outerProduct Scalar x Scalar -> Scalar
+    //              Vector x Vector -> Matrix
+    template<class ll,class rr>
+    inline auto outerProduct (const Lattice<ll> &lhs,const Lattice<rr> &rhs) -> Lattice<decltype(outerProduct(lhs._odata[0],rhs._odata[0]))>
+    {
+        Lattice<decltype(outerProduct(lhs._odata[0],rhs._odata[0]))> ret(rhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<rhs._grid->oSites(); ss++){
+            ret._odata[ss]=outerProduct(lhs._odata[ss],rhs._odata[ss]);
+        }
+        return ret;
+     }
+
+}
+
+#endif
diff --git a/lib/Grid_lattice_peekpoke.h b/lib/Grid_lattice_peekpoke.h
new file mode 100644
index 00000000..41212037
--- /dev/null
+++ b/lib/Grid_lattice_peekpoke.h
@@ -0,0 +1,143 @@
+#ifndef GRID_LATTICE_PEEK_H
+#define GRID_LATTICE_PEEK_H
+
+///////////////////////////////////////////////
+// Peeking and poking around
+///////////////////////////////////////////////
+
+namespace Grid {
+
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    // Peek internal indices of a Lattice object
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    template<int Index,class vobj>
+    inline auto peekIndex(const Lattice<vobj> &lhs)
+      -> Lattice<decltype(peekIndex<Index>(lhs._odata[0]))>
+    {
+      Lattice<decltype(peekIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+            ret._odata[ss] = peekIndex<Index>(lhs._odata[ss]);
+        }
+        return ret;
+    };
+    template<int Index,class vobj>
+      inline auto peekIndex(const Lattice<vobj> &lhs,int i)
+      -> Lattice<decltype(peekIndex<Index>(lhs._odata[0],i))>
+    {
+      Lattice<decltype(peekIndex<Index>(lhs._odata[0],i))> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+	  ret._odata[ss] = peekIndex<Index>(lhs._odata[ss],i);
+        }
+        return ret;
+    };
+    template<int Index,class vobj>
+      inline auto peekIndex(const Lattice<vobj> &lhs,int i,int j)
+      -> Lattice<decltype(peekIndex<Index>(lhs._odata[0],i,j))>
+    {
+      Lattice<decltype(peekIndex<Index>(lhs._odata[0],i,j))> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+	  ret._odata[ss] = peekIndex<Index>(lhs._odata[ss],i,j);
+        }
+        return ret;
+    };
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    // Poke internal indices of a Lattice object
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    template<int Index,class vobj> inline
+    void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0]))> & rhs)
+    {
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+	  pokeIndex<Index>(lhs._odata[ss],rhs._odata[ss]);
+	}      
+    }
+    template<int Index,class vobj> inline
+    void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0],0))> & rhs,int i)
+    {
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+	  pokeIndex<Index>(lhs._odata[ss],rhs._odata[ss],i);
+	}      
+    }
+    template<int Index,class vobj> inline
+    void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0],0,0))> & rhs,int i,int j)
+    {
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+	  pokeIndex<Index>(lhs._odata[ss],rhs._odata[ss],i,j);
+	}      
+    }
+
+    //////////////////////////////////////////////////////
+    // Poke a scalar object into the SIMD array
+    //////////////////////////////////////////////////////
+    template<class vobj,class sobj>
+    void pokeSite(const sobj &s,Lattice<vobj> &l,std::vector<int> &site){
+
+      GridBase *grid=l._grid;
+
+      typedef typename vobj::scalar_type scalar_type;
+      typedef typename vobj::vector_type vector_type;
+
+      int Nsimd = grid->Nsimd();
+
+      assert( l.checkerboard== l._grid->CheckerBoard(site));
+      assert( sizeof(sobj)*Nsimd == sizeof(vobj));
+
+      int rank,odx,idx;
+      grid->GlobalCoorToRankIndex(rank,odx,idx,site);
+
+      // Optional to broadcast from node 0.
+      grid->Broadcast(0,s);
+
+      std::vector<sobj> buf(Nsimd);
+      std::vector<scalar_type *> pointers(Nsimd);  
+      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
+
+      // extract-modify-merge cycle is easiest way and this is not perf critical
+      extract(l._odata[odx],pointers);
+      
+      buf[idx] = s;
+
+      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
+      merge(l._odata[odx],pointers);
+
+      return;
+    };
+
+    //////////////////////////////////////////////////////////
+    // Peek a scalar object from the SIMD array
+    //////////////////////////////////////////////////////////
+    template<class vobj,class sobj>
+      void peekSite(sobj &s,Lattice<vobj> &l,std::vector<int> &site){
+        
+      GridBase *grid=l._grid;
+
+      typedef typename vobj::scalar_type scalar_type;
+      typedef typename vobj::vector_type vector_type;
+
+      int Nsimd = grid->Nsimd();
+
+      assert( l.checkerboard== l._grid->CheckerBoard(site));
+      assert( sizeof(sobj)*Nsimd == sizeof(vobj));
+
+      int rank,odx,idx;
+      grid->GlobalCoorToRankIndex(rank,odx,idx,site);
+      std::vector<sobj> buf(Nsimd);
+      std::vector<scalar_type *> pointers(Nsimd);  
+      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
+
+      extract(l._odata[odx],pointers);
+      
+      s = buf[idx];
+      grid->Broadcast(rank,s);
+
+      return;
+    };
+
+}
+#endif
+
diff --git a/lib/Grid_lattice_reality.h b/lib/Grid_lattice_reality.h
new file mode 100644
index 00000000..a3a7a27f
--- /dev/null
+++ b/lib/Grid_lattice_reality.h
@@ -0,0 +1,52 @@
+#ifndef GRID_LATTICE_REALITY_H
+#define GRID_LATTICE_REALITY_H
+
+
+// FIXME .. this is the sector of the code 
+// I am most worried about the directions
+// The choice of burying complex in the SIMD
+// is making the use of "real" and "imag" very cumbersome
+
+namespace Grid {
+
+    template<class vobj> inline Lattice<vobj> adj(const Lattice<vobj> &lhs){
+        Lattice<vobj> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+            ret._odata[ss] = adj(lhs._odata[ss]);
+        }
+        return ret;
+    };
+
+    template<class vobj> inline Lattice<vobj> conj(const Lattice<vobj> &lhs){
+        Lattice<vobj> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+            ret._odata[ss] = conj(lhs._odata[ss]);
+        }
+        return ret;
+    };
+
+    template<class vobj> inline auto real(const Lattice<vobj> &z) -> Lattice<decltype(real(z._odata[0]))>
+    {
+      Lattice<decltype(real(z._odata[0]))> ret(z._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<z._grid->oSites();ss++){
+            ret._odata[ss] = real(z._odata[ss]);
+        }
+      return ret;
+    }
+
+    template<class vobj> inline auto imag(const Lattice<vobj> &z) -> Lattice<decltype(imag(z._odata[0]))>
+    {
+      Lattice<decltype(imag(z._odata[0]))> ret(z._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<z._grid->oSites();ss++){
+            ret._odata[ss] = imag(z._odata[ss]);
+        }
+      return ret;
+    }
+
+
+}
+#endif
diff --git a/lib/Grid_lattice_reduction.h b/lib/Grid_lattice_reduction.h
new file mode 100644
index 00000000..8167fab0
--- /dev/null
+++ b/lib/Grid_lattice_reduction.h
@@ -0,0 +1,45 @@
+#ifndef GRID_LATTICE_REDUCTION_H
+#define GRID_LATTICE_REDUCTION_H
+
+namespace Grid {
+
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    // Reduction operations
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    template<class vobj>
+    inline RealD norm2(const Lattice<vobj> &arg){
+
+      typedef typename vobj::scalar_type scalar;
+      typedef typename vobj::vector_type vector;
+      decltype(innerProduct(arg._odata[0],arg._odata[0])) vnrm=zero;
+      scalar nrm;
+      //FIXME make this loop parallelisable
+      vnrm=zero;
+      for(int ss=0;ss<arg._grid->oSites(); ss++){
+	vnrm = vnrm + innerProduct(arg._odata[ss],arg._odata[ss]);
+      }
+      vector vvnrm =TensorRemove(vnrm) ;
+      nrm = Reduce(vvnrm);
+      arg._grid->GlobalSum(nrm);
+      return real(nrm);
+    }
+
+    template<class vobj>
+    inline auto innerProduct(const Lattice<vobj> &left,const Lattice<vobj> &right) ->decltype(innerProduct(left._odata[0],right._odata[0]))
+    {
+      typedef typename vobj::scalar_type scalar;
+      decltype(innerProduct(left._odata[0],right._odata[0])) vnrm=zero;
+
+      scalar nrm;
+      //FIXME make this loop parallelisable
+      for(int ss=0;ss<left._grid->oSites(); ss++){
+	vnrm = vnrm + innerProduct(left._odata[ss],right._odata[ss]);
+      }
+      nrm = Reduce(vnrm);
+      right._grid->GlobalSum(nrm);
+      return nrm;
+    }
+
+}
+#endif
+
diff --git a/lib/Grid_lattice_rng.h b/lib/Grid_lattice_rng.h
new file mode 100644
index 00000000..636bfb72
--- /dev/null
+++ b/lib/Grid_lattice_rng.h
@@ -0,0 +1,32 @@
+#ifndef GRID_LATTICE_RNG_H
+#define GRID_LATTICE_RNG_H
+
+namespace Grid {
+
+    // FIXME Randomise; deprecate this
+    template <class vobj> inline void random(Lattice<vobj> &l){
+        Real *v_ptr = (Real *)&l._odata[0];
+        size_t v_len = l._grid->oSites()*sizeof(vobj);
+        size_t d_len = v_len/sizeof(Real);
+	
+        for(int i=0;i<d_len;i++){
+
+            v_ptr[i]=drand48();
+        }
+    };
+    
+    // FIXME Implement a consistent seed management strategy
+    template <class vobj> inline void gaussian(Lattice<vobj> &l){
+        // Zero mean, unit variance.
+        std::normal_distribution<double> distribution(0.0,1.0);
+        Real *v_ptr = (Real *)&l._odata[0];
+        size_t v_len = l._grid->oSites()*sizeof(vobj);
+        size_t d_len = v_len/sizeof(Real);
+
+        for(int i=0;i<d_len;i++){
+	  v_ptr[i]= drand48();
+        }
+    };
+
+}
+#endif
diff --git a/lib/Grid_lattice_trace.h b/lib/Grid_lattice_trace.h
new file mode 100644
index 00000000..8c7607d8
--- /dev/null
+++ b/lib/Grid_lattice_trace.h
@@ -0,0 +1,43 @@
+#ifndef GRID_LATTICE_TRACE_H
+#define GRID_LATTICE_TRACE_H
+
+///////////////////////////////////////////////
+// Tracing, transposing, peeking, poking
+///////////////////////////////////////////////
+
+namespace Grid {
+
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    // Trace
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    template<class vobj>
+    inline auto trace(const Lattice<vobj> &lhs)
+      -> Lattice<decltype(trace(lhs._odata[0]))>
+    {
+      Lattice<decltype(trace(lhs._odata[0]))> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+            ret._odata[ss] = trace(lhs._odata[ss]);
+        }
+        return ret;
+    };
+    
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    // Trace Index level dependent operation
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    template<int Index,class vobj>
+    inline auto traceIndex(const Lattice<vobj> &lhs)
+      -> Lattice<decltype(traceIndex<Index>(lhs._odata[0]))>
+    {
+      Lattice<decltype(traceIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
+#pragma omp parallel for
+      for(int ss=0;ss<lhs._grid->oSites();ss++){
+	ret._odata[ss] = traceIndex<Index>(lhs._odata[ss]);
+      }
+      return ret;
+    };
+
+
+}
+#endif
+
diff --git a/lib/Grid_lattice_transfer.h b/lib/Grid_lattice_transfer.h
new file mode 100644
index 00000000..ecb3016a
--- /dev/null
+++ b/lib/Grid_lattice_transfer.h
@@ -0,0 +1,46 @@
+#ifndef GRID_LATTICE_TRANSFER_H
+#define GRID_LATTICE_TRANSFER_H
+
+namespace Grid {
+
+  ////////////////////////////////////////////////////////////////////////////////////////////
+  // remove and insert a half checkerboard
+  ////////////////////////////////////////////////////////////////////////////////////////////
+  template<class vobj> inline void pickCheckerboard(int cb,Lattice<vobj> &half,const Lattice<vobj> &full){
+    half.checkerboard = cb;
+    int ssh=0;
+#pragma omp parallel for
+    for(int ss=0;ss<full._grid->oSites();ss++){
+      std::vector<int> coor;
+      int cbos;
+      
+      full._grid->oCoorFromOindex(coor,ss);
+      cbos=half._grid->CheckerBoard(coor);
+      
+      if (cbos==cb) {
+	
+	half._odata[ssh] = full._odata[ss];
+	ssh++;
+      }
+    }
+  }
+  template<class vobj> inline void setCheckerboard(Lattice<vobj> &full,const Lattice<vobj> &half){
+    int cb = half.checkerboard;
+    int ssh=0;
+#pragma omp parallel for
+    for(int ss=0;ss<full._grid->oSites();ss++){
+      std::vector<int> coor;
+      int cbos;
+      
+      full._grid->oCoorFromOindex(coor,ss);
+      cbos=half._grid->CheckerBoard(coor);
+      
+      if (cbos==cb) {
+	full._odata[ss]=half._odata[ssh];
+	ssh++;
+      }
+    }
+  }
+  
+}
+#endif
diff --git a/lib/Grid_lattice_transpose.h b/lib/Grid_lattice_transpose.h
new file mode 100644
index 00000000..7166d2b5
--- /dev/null
+++ b/lib/Grid_lattice_transpose.h
@@ -0,0 +1,39 @@
+#ifndef GRID_LATTICE_TRANSPOSE_H
+#define GRID_LATTICE_TRANSPOSE_H
+
+///////////////////////////////////////////////
+// Transpose
+///////////////////////////////////////////////
+
+namespace Grid {
+
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    // Transpose
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+  template<class vobj>
+    inline Lattice<vobj> transpose(const Lattice<vobj> &lhs){
+        Lattice<vobj> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+            ret._odata[ss] = transpose(lhs._odata[ss]);
+        }
+        return ret;
+    };
+    
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    // Index level dependent transpose
+    ////////////////////////////////////////////////////////////////////////////////////////////////////
+    template<int Index,class vobj>
+    inline auto transposeIndex(const Lattice<vobj> &lhs)
+      -> Lattice<decltype(transposeIndex<Index>(lhs._odata[0]))>
+    {
+      Lattice<decltype(transposeIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+            ret._odata[ss] = transposeIndex<Index>(lhs._odata[ss]);
+        }
+        return ret;
+    };
+
+}
+#endif
diff --git a/lib/Grid_predicated.h b/lib/Grid_where.h
similarity index 77%
rename from lib/Grid_predicated.h
rename to lib/Grid_where.h
index c6f29204..3a7052ad 100644
--- a/lib/Grid_predicated.h
+++ b/lib/Grid_where.h
@@ -1,14 +1,14 @@
-#ifndef GRID_PREDICATED_H
-#define GRID_PREDICATED_H
-
+#ifndef GRID_WHERE_H
+#define GRID_WHERE_H
+namespace Grid {
 // Must implement the predicate gating the 
 // Must be able to reduce the predicate down to a single vInteger per site.
 // Must be able to require the type be iScalar x iScalar x ....
 //                              give a GetVtype method in iScalar
 //                              and blow away the tensor structures.
 //
-template<class vobj>
-inline void where(Lattice<vobj> &ret,const LatticeInteger &predicate,Lattice<vobj> &iftrue,Lattice<vobj> &iffalse)
+template<class vobj,class iobj>
+inline void where(Lattice<vobj> &ret,const Lattice<iobj> &predicate,Lattice<vobj> &iftrue,Lattice<vobj> &iffalse)
 {
   conformable(iftrue,iffalse);
   conformable(iftrue,predicate);
@@ -17,6 +17,7 @@ inline void where(Lattice<vobj> &ret,const LatticeInteger &predicate,Lattice<vob
   GridBase *grid=iftrue._grid;
   typedef typename vobj::scalar_type scalar_type;
   typedef typename vobj::vector_type vector_type;
+  typedef typename iobj::vector_type mask_type;
 
   const int Nsimd = grid->Nsimd();
   const int words = sizeof(vobj)/sizeof(vector_type);
@@ -35,7 +36,7 @@ inline void where(Lattice<vobj> &ret,const LatticeInteger &predicate,Lattice<vob
     for(int s=0;s<Nsimd;s++) pointers[s] = & falsevals[s][0];
     extract(iffalse._odata[ss]  ,pointers);
 
-    extract(predicate._odata[ss],mask);
+    extract(TensorRemove(predicate._odata[ss]),mask);
 
     for(int s=0;s<Nsimd;s++){
       if (mask[s]) pointers[s]=&truevals[s][0];
@@ -46,8 +47,8 @@ inline void where(Lattice<vobj> &ret,const LatticeInteger &predicate,Lattice<vob
   }
 }
 
-template<class vobj>
-inline Lattice<vobj> where(const LatticeInteger &predicate,Lattice<vobj> &iftrue,Lattice<vobj> &iffalse)
+template<class vobj,class iobj>
+inline Lattice<vobj> where(const Lattice<iobj> &predicate,Lattice<vobj> &iftrue,Lattice<vobj> &iffalse)
 {
   conformable(iftrue,iffalse);
   conformable(iftrue,predicate);
@@ -58,5 +59,5 @@ inline Lattice<vobj> where(const LatticeInteger &predicate,Lattice<vobj> &iftrue
 
   return ret;
 }
-
+}
 #endif

From e6ec92d0e48f9d8c49f77ec38c75afd6ebd12124 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 20:45:00 +0100
Subject: [PATCH 076/429] More files, shorter each.

---
 lib/Makefile.am   | 31 ++++++++++++++++++++++++++-----
 tests/Makefile.am |  1 -
 2 files changed, 26 insertions(+), 6 deletions(-)

diff --git a/lib/Makefile.am b/lib/Makefile.am
index 93625c68..dab5c6d3 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -1,7 +1,6 @@
 # additional include paths necessary to compile the C++ library
 AM_CXXFLAGS = -I$(top_srcdir)/
 
-
 extra_sources=
 if BUILD_COMMS_MPI
   extra_sources+=Grid_communicator_mpi.cc
@@ -21,7 +20,7 @@ libGrid_a_SOURCES = Grid_init.cc $(extra_sources)
 #
 # Include files
 #
-include_HEADERS = Grid_config.h\
+include_HEADERS =\
 	Grid.h\
 	Grid_QCD.h\
 	Grid_aligned_allocator.h\
@@ -37,11 +36,33 @@ include_HEADERS = Grid_config.h\
 	Grid_cshift_mpi.h\
 	Grid_cshift_none.h\
 	Grid_lattice.h\
+	Grid_lattice_arith.h\
+	Grid_lattice_conformable.h\
+	Grid_lattice_coordinate.h\
+	Grid_lattice_local.h\
+	Grid_lattice_peekpoke.h\
+	Grid_lattice_reality.h\
+	Grid_lattice_reduction.h
+	Grid_lattice_rng.h\
+	Grid_lattice_trace.h\
+	Grid_lattice_transfer.h\
+	Grid_lattice_transpose.h\
 	Grid_math.h\
 	Grid_math_arith.h\
+	Grid_math_arith_add.h\
+	Grid_math_arith_mac.h\
+	Grid_math_arith_mul.h\
+	Grid_math_arith_scalar.h\
+	Grid_math_arith_sub.h\
+	Grid_math_inner.h\
+	Grid_math_outer.h\
+	Grid_math_peek.h\
+	Grid_math_poke.h\
+	Grid_math_reality.h\
 	Grid_math_tensors.h\
+	Grid_math_trace.h\
 	Grid_math_traits.h\
-	Grid_predicated.h\
+	Grid_math_transpose.h\
 	Grid_simd.h\
 	Grid_stencil.h\
 	Grid_summation.h\
@@ -49,7 +70,7 @@ include_HEADERS = Grid_config.h\
 	Grid_vComplexF.h\
 	Grid_vInteger.h\
 	Grid_vRealD.h\
-	Grid_vRealF.h
-
+	Grid_vRealF.h\
+	Grid_where.h
 
 
diff --git a/tests/Makefile.am b/tests/Makefile.am
index 2bf8cd1a..ec0703b1 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -12,4 +12,3 @@ Grid_main_LDADD = -lGrid
 
 test_Grid_stencil_SOURCES = test_Grid_stencil.cc
 test_Grid_stencil_LDADD = -lGrid
-

From a17ce0695b78c597266581fbd2e47514b9311d89 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 20:52:40 +0100
Subject: [PATCH 077/429] Clean up

---
 Makefile.in | 616 +++++++++++++++++++++++++++++-----------------------
 TODO        |  12 +
 2 files changed, 358 insertions(+), 270 deletions(-)

diff --git a/Makefile.in b/Makefile.in
index 3ca9a01f..99b67fff 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -13,8 +13,6 @@
 # PARTICULAR PURPOSE.
 
 @SET_MAKE@
-
-
 VPATH = @srcdir@
 am__is_gnu_make = { \
   if test -z '$(MAKELEVEL)'; then \
@@ -87,68 +85,19 @@ POST_INSTALL = :
 NORMAL_UNINSTALL = :
 PRE_UNINSTALL = :
 POST_UNINSTALL = :
-@BUILD_COMMS_MPI_TRUE@am__append_1 = Grid_communicator_mpi.cc \
-@BUILD_COMMS_MPI_TRUE@	Grid_stencil_common.cc
-@BUILD_COMMS_NONE_TRUE@am__append_2 = Grid_communicator_fake.cc \
-@BUILD_COMMS_NONE_TRUE@	Grid_stencil_common.cc
-subdir = lib
+subdir = .
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/configure.ac
 am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
 	$(ACLOCAL_M4)
-DIST_COMMON = $(srcdir)/Makefile.am $(include_HEADERS) \
-	$(am__DIST_COMMON)
+DIST_COMMON = $(srcdir)/Makefile.am $(top_srcdir)/configure \
+	$(am__configure_deps) $(am__DIST_COMMON)
+am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \
+ configure.lineno config.status.lineno
 mkinstalldirs = $(install_sh) -d
-CONFIG_HEADER = Grid_config.h
+CONFIG_HEADER = $(top_builddir)/lib/Grid_config.h
 CONFIG_CLEAN_FILES =
 CONFIG_CLEAN_VPATH_FILES =
-am__vpath_adj_setup = srcdirstrip=`echo "$(srcdir)" | sed 's|.|.|g'`;
-am__vpath_adj = case $$p in \
-    $(srcdir)/*) f=`echo "$$p" | sed "s|^$$srcdirstrip/||"`;; \
-    *) f=$$p;; \
-  esac;
-am__strip_dir = f=`echo $$p | sed -e 's|^.*/||'`;
-am__install_max = 40
-am__nobase_strip_setup = \
-  srcdirstrip=`echo "$(srcdir)" | sed 's/[].[^$$\\*|]/\\\\&/g'`
-am__nobase_strip = \
-  for p in $$list; do echo "$$p"; done | sed -e "s|$$srcdirstrip/||"
-am__nobase_list = $(am__nobase_strip_setup); \
-  for p in $$list; do echo "$$p $$p"; done | \
-  sed "s| $$srcdirstrip/| |;"' / .*\//!s/ .*/ ./; s,\( .*\)/[^/]*$$,\1,' | \
-  $(AWK) 'BEGIN { files["."] = "" } { files[$$2] = files[$$2] " " $$1; \
-    if (++n[$$2] == $(am__install_max)) \
-      { print $$2, files[$$2]; n[$$2] = 0; files[$$2] = "" } } \
-    END { for (dir in files) print dir, files[dir] }'
-am__base_list = \
-  sed '$$!N;$$!N;$$!N;$$!N;$$!N;$$!N;$$!N;s/\n/ /g' | \
-  sed '$$!N;$$!N;$$!N;$$!N;s/\n/ /g'
-am__uninstall_files_from_dir = { \
-  test -z "$$files" \
-    || { test ! -d "$$dir" && test ! -f "$$dir" && test ! -r "$$dir"; } \
-    || { echo " ( cd '$$dir' && rm -f" $$files ")"; \
-         $(am__cd) "$$dir" && rm -f $$files; }; \
-  }
-am__installdirs = "$(DESTDIR)$(libdir)" "$(DESTDIR)$(includedir)"
-LIBRARIES = $(lib_LIBRARIES)
-AR = ar
-ARFLAGS = cru
-AM_V_AR = $(am__v_AR_@AM_V@)
-am__v_AR_ = $(am__v_AR_@AM_DEFAULT_V@)
-am__v_AR_0 = @echo "  AR      " $@;
-am__v_AR_1 = 
-libGrid_a_AR = $(AR) $(ARFLAGS)
-libGrid_a_LIBADD =
-am__libGrid_a_SOURCES_DIST = Grid_init.cc Grid_communicator_mpi.cc \
-	Grid_stencil_common.cc Grid_communicator_fake.cc
-@BUILD_COMMS_MPI_TRUE@am__objects_1 = Grid_communicator_mpi.$(OBJEXT) \
-@BUILD_COMMS_MPI_TRUE@	Grid_stencil_common.$(OBJEXT)
-@BUILD_COMMS_NONE_TRUE@am__objects_2 =  \
-@BUILD_COMMS_NONE_TRUE@	Grid_communicator_fake.$(OBJEXT) \
-@BUILD_COMMS_NONE_TRUE@	Grid_stencil_common.$(OBJEXT)
-am__objects_3 = $(am__objects_1) $(am__objects_2)
-am_libGrid_a_OBJECTS = Grid_init.$(OBJEXT) $(am__objects_3)
-libGrid_a_OBJECTS = $(am_libGrid_a_OBJECTS)
 AM_V_P = $(am__v_P_@AM_V@)
 am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
 am__v_P_0 = false
@@ -161,33 +110,30 @@ AM_V_at = $(am__v_at_@AM_V@)
 am__v_at_ = $(am__v_at_@AM_DEFAULT_V@)
 am__v_at_0 = @
 am__v_at_1 = 
-DEFAULT_INCLUDES = -I.@am__isrc@
-depcomp = $(SHELL) $(top_srcdir)/depcomp
-am__depfiles_maybe = depfiles
-am__mv = mv -f
-CXXCOMPILE = $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) \
-	$(AM_CPPFLAGS) $(CPPFLAGS) $(AM_CXXFLAGS) $(CXXFLAGS)
-AM_V_CXX = $(am__v_CXX_@AM_V@)
-am__v_CXX_ = $(am__v_CXX_@AM_DEFAULT_V@)
-am__v_CXX_0 = @echo "  CXX     " $@;
-am__v_CXX_1 = 
-CXXLD = $(CXX)
-CXXLINK = $(CXXLD) $(AM_CXXFLAGS) $(CXXFLAGS) $(AM_LDFLAGS) $(LDFLAGS) \
-	-o $@
-AM_V_CXXLD = $(am__v_CXXLD_@AM_V@)
-am__v_CXXLD_ = $(am__v_CXXLD_@AM_DEFAULT_V@)
-am__v_CXXLD_0 = @echo "  CXXLD   " $@;
-am__v_CXXLD_1 = 
-SOURCES = $(libGrid_a_SOURCES)
-DIST_SOURCES = $(am__libGrid_a_SOURCES_DIST)
+SOURCES =
+DIST_SOURCES =
+RECURSIVE_TARGETS = all-recursive check-recursive cscopelist-recursive \
+	ctags-recursive dvi-recursive html-recursive info-recursive \
+	install-data-recursive install-dvi-recursive \
+	install-exec-recursive install-html-recursive \
+	install-info-recursive install-pdf-recursive \
+	install-ps-recursive install-recursive installcheck-recursive \
+	installdirs-recursive pdf-recursive ps-recursive \
+	tags-recursive uninstall-recursive
 am__can_run_installinfo = \
   case $$AM_UPDATE_INFO_DIR in \
     n|no|NO) false;; \
     *) (install-info --version) >/dev/null 2>&1;; \
   esac
-HEADERS = $(include_HEADERS)
-am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) \
-	$(LISP)Grid_config.h.in
+RECURSIVE_CLEAN_TARGETS = mostlyclean-recursive clean-recursive	\
+  distclean-recursive maintainer-clean-recursive
+am__recursive_targets = \
+  $(RECURSIVE_TARGETS) \
+  $(RECURSIVE_CLEAN_TARGETS) \
+  $(am__extra_recursive_targets)
+AM_RECURSIVE_TARGETS = $(am__recursive_targets:-recursive=) TAGS CTAGS \
+	cscope distdir dist dist-all distcheck
+am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) $(LISP)
 # Read a list of newline-separated strings from the standard input,
 # and print each of them once, without duplicates.  Input order is
 # *not* preserved.
@@ -206,9 +152,52 @@ am__define_uniq_tagged_files = \
   done | $(am__uniquify_input)`
 ETAGS = etags
 CTAGS = ctags
-am__DIST_COMMON = $(srcdir)/Grid_config.h.in $(srcdir)/Makefile.in \
-	$(top_srcdir)/depcomp
+CSCOPE = cscope
+DIST_SUBDIRS = $(SUBDIRS)
+am__DIST_COMMON = $(srcdir)/Makefile.in AUTHORS COPYING ChangeLog \
+	INSTALL NEWS README TODO compile depcomp install-sh missing
 DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
+distdir = $(PACKAGE)-$(VERSION)
+top_distdir = $(distdir)
+am__remove_distdir = \
+  if test -d "$(distdir)"; then \
+    find "$(distdir)" -type d ! -perm -200 -exec chmod u+w {} ';' \
+      && rm -rf "$(distdir)" \
+      || { sleep 5 && rm -rf "$(distdir)"; }; \
+  else :; fi
+am__post_remove_distdir = $(am__remove_distdir)
+am__relativize = \
+  dir0=`pwd`; \
+  sed_first='s,^\([^/]*\)/.*$$,\1,'; \
+  sed_rest='s,^[^/]*/*,,'; \
+  sed_last='s,^.*/\([^/]*\)$$,\1,'; \
+  sed_butlast='s,/*[^/]*$$,,'; \
+  while test -n "$$dir1"; do \
+    first=`echo "$$dir1" | sed -e "$$sed_first"`; \
+    if test "$$first" != "."; then \
+      if test "$$first" = ".."; then \
+        dir2=`echo "$$dir0" | sed -e "$$sed_last"`/"$$dir2"; \
+        dir0=`echo "$$dir0" | sed -e "$$sed_butlast"`; \
+      else \
+        first2=`echo "$$dir2" | sed -e "$$sed_first"`; \
+        if test "$$first2" = "$$first"; then \
+          dir2=`echo "$$dir2" | sed -e "$$sed_rest"`; \
+        else \
+          dir2="../$$dir2"; \
+        fi; \
+        dir0="$$dir0"/"$$first"; \
+      fi; \
+    fi; \
+    dir1=`echo "$$dir1" | sed -e "$$sed_rest"`; \
+  done; \
+  reldir="$$dir2"
+DIST_ARCHIVES = $(distdir).tar.gz
+GZIP_ENV = --best
+DIST_TARGETS = dist-gzip
+distuninstallcheck_listfiles = find . -type f -print
+am__distuninstallcheck_listfiles = $(distuninstallcheck_listfiles) \
+  | sed 's|^\./|$(prefix)/|' | grep -v '$(infodir)/dir$$'
+distcleancheck_listfiles = find . -type f -print
 ACLOCAL = @ACLOCAL@
 AMTAR = @AMTAR@
 AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
@@ -305,176 +294,99 @@ top_srcdir = @top_srcdir@
 
 # additional include paths necessary to compile the C++ library
 AM_CXXFLAGS = -I$(top_srcdir)/
-extra_sources = $(am__append_1) $(am__append_2)
-
-#
-# Libraries
-#
-lib_LIBRARIES = libGrid.a
-libGrid_a_SOURCES = Grid_init.cc $(extra_sources)
-
-#
-# Include files
-#
-include_HEADERS = Grid_config.h\
-	Grid.h\
-	Grid_simd.h\
-	Grid_vComplexD.h\
-	Grid_vComplexF.h\
-	Grid_vRealD.h\
-	Grid_vRealF.h\
-	Grid_Cartesian.h\
-	Grid_Lattice.h\
-	Grid_Communicator.h\
-	Grid_QCD.h\
-	Grid_aligned_allocator.h\
-	Grid_cshift.h\
-	Grid_cshift_common.h\
-	Grid_cshift_mpi.h\
-	Grid_cshift_none.h\
-	Grid_stencil.h\
-	Grid_math_types.h
-
-all: Grid_config.h
-	$(MAKE) $(AM_MAKEFLAGS) all-am
+SUBDIRS = lib tests
+all: all-recursive
 
 .SUFFIXES:
-.SUFFIXES: .cc .o .obj
+am--refresh: Makefile
+	@:
 $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	@for dep in $?; do \
 	  case '$(am__configure_deps)' in \
 	    *$$dep*) \
-	      ( cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh ) \
-	        && { if test -f $@; then exit 0; else break; fi; }; \
+	      echo ' cd $(srcdir) && $(AUTOMAKE) --gnu'; \
+	      $(am__cd) $(srcdir) && $(AUTOMAKE) --gnu \
+		&& exit 0; \
 	      exit 1;; \
 	  esac; \
 	done; \
-	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu lib/Makefile'; \
+	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu Makefile'; \
 	$(am__cd) $(top_srcdir) && \
-	  $(AUTOMAKE) --gnu lib/Makefile
+	  $(AUTOMAKE) --gnu Makefile
 Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
 	@case '$?' in \
 	  *config.status*) \
-	    cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh;; \
+	    echo ' $(SHELL) ./config.status'; \
+	    $(SHELL) ./config.status;; \
 	  *) \
-	    echo ' cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__depfiles_maybe)'; \
-	    cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__depfiles_maybe);; \
+	    echo ' cd $(top_builddir) && $(SHELL) ./config.status $@ $(am__depfiles_maybe)'; \
+	    cd $(top_builddir) && $(SHELL) ./config.status $@ $(am__depfiles_maybe);; \
 	esac;
 
 $(top_builddir)/config.status: $(top_srcdir)/configure $(CONFIG_STATUS_DEPENDENCIES)
-	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
+	$(SHELL) ./config.status --recheck
 
 $(top_srcdir)/configure:  $(am__configure_deps)
-	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
+	$(am__cd) $(srcdir) && $(AUTOCONF)
 $(ACLOCAL_M4):  $(am__aclocal_m4_deps)
-	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
+	$(am__cd) $(srcdir) && $(ACLOCAL) $(ACLOCAL_AMFLAGS)
 $(am__aclocal_m4_deps):
 
-Grid_config.h: stamp-h1
-	@test -f $@ || rm -f stamp-h1
-	@test -f $@ || $(MAKE) $(AM_MAKEFLAGS) stamp-h1
-
-stamp-h1: $(srcdir)/Grid_config.h.in $(top_builddir)/config.status
-	@rm -f stamp-h1
-	cd $(top_builddir) && $(SHELL) ./config.status lib/Grid_config.h
-$(srcdir)/Grid_config.h.in:  $(am__configure_deps) 
-	($(am__cd) $(top_srcdir) && $(AUTOHEADER))
-	rm -f stamp-h1
-	touch $@
-
-distclean-hdr:
-	-rm -f Grid_config.h stamp-h1
-install-libLIBRARIES: $(lib_LIBRARIES)
-	@$(NORMAL_INSTALL)
-	@list='$(lib_LIBRARIES)'; test -n "$(libdir)" || list=; \
-	list2=; for p in $$list; do \
-	  if test -f $$p; then \
-	    list2="$$list2 $$p"; \
-	  else :; fi; \
-	done; \
-	test -z "$$list2" || { \
-	  echo " $(MKDIR_P) '$(DESTDIR)$(libdir)'"; \
-	  $(MKDIR_P) "$(DESTDIR)$(libdir)" || exit 1; \
-	  echo " $(INSTALL_DATA) $$list2 '$(DESTDIR)$(libdir)'"; \
-	  $(INSTALL_DATA) $$list2 "$(DESTDIR)$(libdir)" || exit $$?; }
-	@$(POST_INSTALL)
-	@list='$(lib_LIBRARIES)'; test -n "$(libdir)" || list=; \
-	for p in $$list; do \
-	  if test -f $$p; then \
-	    $(am__strip_dir) \
-	    echo " ( cd '$(DESTDIR)$(libdir)' && $(RANLIB) $$f )"; \
-	    ( cd "$(DESTDIR)$(libdir)" && $(RANLIB) $$f ) || exit $$?; \
-	  else :; fi; \
-	done
-
-uninstall-libLIBRARIES:
-	@$(NORMAL_UNINSTALL)
-	@list='$(lib_LIBRARIES)'; test -n "$(libdir)" || list=; \
-	files=`for p in $$list; do echo $$p; done | sed -e 's|^.*/||'`; \
-	dir='$(DESTDIR)$(libdir)'; $(am__uninstall_files_from_dir)
-
-clean-libLIBRARIES:
-	-test -z "$(lib_LIBRARIES)" || rm -f $(lib_LIBRARIES)
-
-libGrid.a: $(libGrid_a_OBJECTS) $(libGrid_a_DEPENDENCIES) $(EXTRA_libGrid_a_DEPENDENCIES) 
-	$(AM_V_at)-rm -f libGrid.a
-	$(AM_V_AR)$(libGrid_a_AR) libGrid.a $(libGrid_a_OBJECTS) $(libGrid_a_LIBADD)
-	$(AM_V_at)$(RANLIB) libGrid.a
-
-mostlyclean-compile:
-	-rm -f *.$(OBJEXT)
-
-distclean-compile:
-	-rm -f *.tab.c
-
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_communicator_fake.Po@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_communicator_mpi.Po@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_init.Po@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/Grid_stencil_common.Po@am__quote@
-
-.cc.o:
-@am__fastdepCXX_TRUE@	$(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ $<
-@am__fastdepCXX_TRUE@	$(AM_V_at)$(am__mv) $(DEPDIR)/$*.Tpo $(DEPDIR)/$*.Po
-@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
-@AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
-@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ $<
-
-.cc.obj:
-@am__fastdepCXX_TRUE@	$(AM_V_CXX)$(CXXCOMPILE) -MT $@ -MD -MP -MF $(DEPDIR)/$*.Tpo -c -o $@ `$(CYGPATH_W) '$<'`
-@am__fastdepCXX_TRUE@	$(AM_V_at)$(am__mv) $(DEPDIR)/$*.Tpo $(DEPDIR)/$*.Po
-@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
-@AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
-@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ `$(CYGPATH_W) '$<'`
-install-includeHEADERS: $(include_HEADERS)
-	@$(NORMAL_INSTALL)
-	@list='$(include_HEADERS)'; test -n "$(includedir)" || list=; \
-	if test -n "$$list"; then \
-	  echo " $(MKDIR_P) '$(DESTDIR)$(includedir)'"; \
-	  $(MKDIR_P) "$(DESTDIR)$(includedir)" || exit 1; \
+# This directory's subdirectories are mostly independent; you can cd
+# into them and run 'make' without going through this Makefile.
+# To change the values of 'make' variables: instead of editing Makefiles,
+# (1) if the variable is set in 'config.status', edit 'config.status'
+#     (which will cause the Makefiles to be regenerated when you run 'make');
+# (2) otherwise, pass the desired values on the 'make' command line.
+$(am__recursive_targets):
+	@fail=; \
+	if $(am__make_keepgoing); then \
+	  failcom='fail=yes'; \
+	else \
+	  failcom='exit 1'; \
 	fi; \
-	for p in $$list; do \
-	  if test -f "$$p"; then d=; else d="$(srcdir)/"; fi; \
-	  echo "$$d$$p"; \
-	done | $(am__base_list) | \
-	while read files; do \
-	  echo " $(INSTALL_HEADER) $$files '$(DESTDIR)$(includedir)'"; \
-	  $(INSTALL_HEADER) $$files "$(DESTDIR)$(includedir)" || exit $$?; \
-	done
-
-uninstall-includeHEADERS:
-	@$(NORMAL_UNINSTALL)
-	@list='$(include_HEADERS)'; test -n "$(includedir)" || list=; \
-	files=`for p in $$list; do echo $$p; done | sed -e 's|^.*/||'`; \
-	dir='$(DESTDIR)$(includedir)'; $(am__uninstall_files_from_dir)
+	dot_seen=no; \
+	target=`echo $@ | sed s/-recursive//`; \
+	case "$@" in \
+	  distclean-* | maintainer-clean-*) list='$(DIST_SUBDIRS)' ;; \
+	  *) list='$(SUBDIRS)' ;; \
+	esac; \
+	for subdir in $$list; do \
+	  echo "Making $$target in $$subdir"; \
+	  if test "$$subdir" = "."; then \
+	    dot_seen=yes; \
+	    local_target="$$target-am"; \
+	  else \
+	    local_target="$$target"; \
+	  fi; \
+	  ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) $$local_target) \
+	  || eval $$failcom; \
+	done; \
+	if test "$$dot_seen" = "no"; then \
+	  $(MAKE) $(AM_MAKEFLAGS) "$$target-am" || exit 1; \
+	fi; test -z "$$fail"
 
 ID: $(am__tagged_files)
 	$(am__define_uniq_tagged_files); mkid -fID $$unique
-tags: tags-am
+tags: tags-recursive
 TAGS: tags
 
 tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
 	set x; \
 	here=`pwd`; \
+	if ($(ETAGS) --etags-include --version) >/dev/null 2>&1; then \
+	  include_option=--etags-include; \
+	  empty_fix=.; \
+	else \
+	  include_option=--include; \
+	  empty_fix=; \
+	fi; \
+	list='$(SUBDIRS)'; for subdir in $$list; do \
+	  if test "$$subdir" = .; then :; else \
+	    test ! -f $$subdir/TAGS || \
+	      set "$$@" "$$include_option=$$here/$$subdir/TAGS"; \
+	  fi; \
+	done; \
 	$(am__define_uniq_tagged_files); \
 	shift; \
 	if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
@@ -487,7 +399,7 @@ tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
 	      $$unique; \
 	  fi; \
 	fi
-ctags: ctags-am
+ctags: ctags-recursive
 
 CTAGS: ctags
 ctags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
@@ -500,7 +412,13 @@ GTAGS:
 	here=`$(am__cd) $(top_builddir) && pwd` \
 	  && $(am__cd) $(top_srcdir) \
 	  && gtags -i $(GTAGS_ARGS) "$$here"
-cscopelist: cscopelist-am
+cscope: cscope.files
+	test ! -s cscope.files \
+	  || $(CSCOPE) -b -q $(AM_CSCOPEFLAGS) $(CSCOPEFLAGS) -i cscope.files $(CSCOPE_ARGS)
+clean-cscope:
+	-rm -f cscope.files
+cscope.files: clean-cscope cscopelist
+cscopelist: cscopelist-recursive
 
 cscopelist-am: $(am__tagged_files)
 	list='$(am__tagged_files)'; \
@@ -518,8 +436,11 @@ cscopelist-am: $(am__tagged_files)
 
 distclean-tags:
 	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
+	-rm -f cscope.out cscope.in.out cscope.po.out cscope.files
 
 distdir: $(DISTFILES)
+	$(am__remove_distdir)
+	test -d "$(distdir)" || mkdir "$(distdir)"
 	@srcdirstrip=`echo "$(srcdir)" | sed 's/[].[^$$\\*]/\\\\&/g'`; \
 	topsrcdirstrip=`echo "$(top_srcdir)" | sed 's/[].[^$$\\*]/\\\\&/g'`; \
 	list='$(DISTFILES)'; \
@@ -549,22 +470,176 @@ distdir: $(DISTFILES)
 	    || exit 1; \
 	  fi; \
 	done
-check-am: all-am
-check: check-am
-all-am: Makefile $(LIBRARIES) $(HEADERS) Grid_config.h
-installdirs:
-	for dir in "$(DESTDIR)$(libdir)" "$(DESTDIR)$(includedir)"; do \
-	  test -z "$$dir" || $(MKDIR_P) "$$dir"; \
+	@list='$(DIST_SUBDIRS)'; for subdir in $$list; do \
+	  if test "$$subdir" = .; then :; else \
+	    $(am__make_dryrun) \
+	      || test -d "$(distdir)/$$subdir" \
+	      || $(MKDIR_P) "$(distdir)/$$subdir" \
+	      || exit 1; \
+	    dir1=$$subdir; dir2="$(distdir)/$$subdir"; \
+	    $(am__relativize); \
+	    new_distdir=$$reldir; \
+	    dir1=$$subdir; dir2="$(top_distdir)"; \
+	    $(am__relativize); \
+	    new_top_distdir=$$reldir; \
+	    echo " (cd $$subdir && $(MAKE) $(AM_MAKEFLAGS) top_distdir="$$new_top_distdir" distdir="$$new_distdir" \\"; \
+	    echo "     am__remove_distdir=: am__skip_length_check=: am__skip_mode_fix=: distdir)"; \
+	    ($(am__cd) $$subdir && \
+	      $(MAKE) $(AM_MAKEFLAGS) \
+	        top_distdir="$$new_top_distdir" \
+	        distdir="$$new_distdir" \
+		am__remove_distdir=: \
+		am__skip_length_check=: \
+		am__skip_mode_fix=: \
+	        distdir) \
+	      || exit 1; \
+	  fi; \
 	done
-install: install-am
-install-exec: install-exec-am
-install-data: install-data-am
-uninstall: uninstall-am
+	-test -n "$(am__skip_mode_fix)" \
+	|| find "$(distdir)" -type d ! -perm -755 \
+		-exec chmod u+rwx,go+rx {} \; -o \
+	  ! -type d ! -perm -444 -links 1 -exec chmod a+r {} \; -o \
+	  ! -type d ! -perm -400 -exec chmod a+r {} \; -o \
+	  ! -type d ! -perm -444 -exec $(install_sh) -c -m a+r {} {} \; \
+	|| chmod -R a+r "$(distdir)"
+dist-gzip: distdir
+	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
+	$(am__post_remove_distdir)
+
+dist-bzip2: distdir
+	tardir=$(distdir) && $(am__tar) | BZIP2=$${BZIP2--9} bzip2 -c >$(distdir).tar.bz2
+	$(am__post_remove_distdir)
+
+dist-lzip: distdir
+	tardir=$(distdir) && $(am__tar) | lzip -c $${LZIP_OPT--9} >$(distdir).tar.lz
+	$(am__post_remove_distdir)
+
+dist-xz: distdir
+	tardir=$(distdir) && $(am__tar) | XZ_OPT=$${XZ_OPT--e} xz -c >$(distdir).tar.xz
+	$(am__post_remove_distdir)
+
+dist-tarZ: distdir
+	@echo WARNING: "Support for distribution archives compressed with" \
+		       "legacy program 'compress' is deprecated." >&2
+	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
+	tardir=$(distdir) && $(am__tar) | compress -c >$(distdir).tar.Z
+	$(am__post_remove_distdir)
+
+dist-shar: distdir
+	@echo WARNING: "Support for shar distribution archives is" \
+	               "deprecated." >&2
+	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
+	shar $(distdir) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).shar.gz
+	$(am__post_remove_distdir)
+
+dist-zip: distdir
+	-rm -f $(distdir).zip
+	zip -rq $(distdir).zip $(distdir)
+	$(am__post_remove_distdir)
+
+dist dist-all:
+	$(MAKE) $(AM_MAKEFLAGS) $(DIST_TARGETS) am__post_remove_distdir='@:'
+	$(am__post_remove_distdir)
+
+# This target untars the dist file and tries a VPATH configuration.  Then
+# it guarantees that the distribution is self-contained by making another
+# tarfile.
+distcheck: dist
+	case '$(DIST_ARCHIVES)' in \
+	*.tar.gz*) \
+	  GZIP=$(GZIP_ENV) gzip -dc $(distdir).tar.gz | $(am__untar) ;;\
+	*.tar.bz2*) \
+	  bzip2 -dc $(distdir).tar.bz2 | $(am__untar) ;;\
+	*.tar.lz*) \
+	  lzip -dc $(distdir).tar.lz | $(am__untar) ;;\
+	*.tar.xz*) \
+	  xz -dc $(distdir).tar.xz | $(am__untar) ;;\
+	*.tar.Z*) \
+	  uncompress -c $(distdir).tar.Z | $(am__untar) ;;\
+	*.shar.gz*) \
+	  GZIP=$(GZIP_ENV) gzip -dc $(distdir).shar.gz | unshar ;;\
+	*.zip*) \
+	  unzip $(distdir).zip ;;\
+	esac
+	chmod -R a-w $(distdir)
+	chmod u+w $(distdir)
+	mkdir $(distdir)/_build $(distdir)/_build/sub $(distdir)/_inst
+	chmod a-w $(distdir)
+	test -d $(distdir)/_build || exit 0; \
+	dc_install_base=`$(am__cd) $(distdir)/_inst && pwd | sed -e 's,^[^:\\/]:[\\/],/,'` \
+	  && dc_destdir="$${TMPDIR-/tmp}/am-dc-$$$$/" \
+	  && am__cwd=`pwd` \
+	  && $(am__cd) $(distdir)/_build/sub \
+	  && ../../configure \
+	    $(AM_DISTCHECK_CONFIGURE_FLAGS) \
+	    $(DISTCHECK_CONFIGURE_FLAGS) \
+	    --srcdir=../.. --prefix="$$dc_install_base" \
+	  && $(MAKE) $(AM_MAKEFLAGS) \
+	  && $(MAKE) $(AM_MAKEFLAGS) dvi \
+	  && $(MAKE) $(AM_MAKEFLAGS) check \
+	  && $(MAKE) $(AM_MAKEFLAGS) install \
+	  && $(MAKE) $(AM_MAKEFLAGS) installcheck \
+	  && $(MAKE) $(AM_MAKEFLAGS) uninstall \
+	  && $(MAKE) $(AM_MAKEFLAGS) distuninstallcheck_dir="$$dc_install_base" \
+	        distuninstallcheck \
+	  && chmod -R a-w "$$dc_install_base" \
+	  && ({ \
+	       (cd ../.. && umask 077 && mkdir "$$dc_destdir") \
+	       && $(MAKE) $(AM_MAKEFLAGS) DESTDIR="$$dc_destdir" install \
+	       && $(MAKE) $(AM_MAKEFLAGS) DESTDIR="$$dc_destdir" uninstall \
+	       && $(MAKE) $(AM_MAKEFLAGS) DESTDIR="$$dc_destdir" \
+	            distuninstallcheck_dir="$$dc_destdir" distuninstallcheck; \
+	      } || { rm -rf "$$dc_destdir"; exit 1; }) \
+	  && rm -rf "$$dc_destdir" \
+	  && $(MAKE) $(AM_MAKEFLAGS) dist \
+	  && rm -rf $(DIST_ARCHIVES) \
+	  && $(MAKE) $(AM_MAKEFLAGS) distcleancheck \
+	  && cd "$$am__cwd" \
+	  || exit 1
+	$(am__post_remove_distdir)
+	@(echo "$(distdir) archives ready for distribution: "; \
+	  list='$(DIST_ARCHIVES)'; for i in $$list; do echo $$i; done) | \
+	  sed -e 1h -e 1s/./=/g -e 1p -e 1x -e '$$p' -e '$$x'
+distuninstallcheck:
+	@test -n '$(distuninstallcheck_dir)' || { \
+	  echo 'ERROR: trying to run $@ with an empty' \
+	       '$$(distuninstallcheck_dir)' >&2; \
+	  exit 1; \
+	}; \
+	$(am__cd) '$(distuninstallcheck_dir)' || { \
+	  echo 'ERROR: cannot chdir into $(distuninstallcheck_dir)' >&2; \
+	  exit 1; \
+	}; \
+	test `$(am__distuninstallcheck_listfiles) | wc -l` -eq 0 \
+	   || { echo "ERROR: files left after uninstall:" ; \
+	        if test -n "$(DESTDIR)"; then \
+	          echo "  (check DESTDIR support)"; \
+	        fi ; \
+	        $(distuninstallcheck_listfiles) ; \
+	        exit 1; } >&2
+distcleancheck: distclean
+	@if test '$(srcdir)' = . ; then \
+	  echo "ERROR: distcleancheck can only run from a VPATH build" ; \
+	  exit 1 ; \
+	fi
+	@test `$(distcleancheck_listfiles) | wc -l` -eq 0 \
+	  || { echo "ERROR: files left in build directory after distclean:" ; \
+	       $(distcleancheck_listfiles) ; \
+	       exit 1; } >&2
+check-am: all-am
+check: check-recursive
+all-am: Makefile
+installdirs: installdirs-recursive
+installdirs-am:
+install: install-recursive
+install-exec: install-exec-recursive
+install-data: install-data-recursive
+uninstall: uninstall-recursive
 
 install-am: all-am
 	@$(MAKE) $(AM_MAKEFLAGS) install-exec-am install-data-am
 
-installcheck: installcheck-am
+installcheck: installcheck-recursive
 install-strip:
 	if test -z '$(STRIP)'; then \
 	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
@@ -586,90 +661,91 @@ distclean-generic:
 maintainer-clean-generic:
 	@echo "This command is intended for maintainers to use"
 	@echo "it deletes files that may require special tools to rebuild."
-clean: clean-am
+clean: clean-recursive
 
-clean-am: clean-generic clean-libLIBRARIES mostlyclean-am
+clean-am: clean-generic mostlyclean-am
 
-distclean: distclean-am
-	-rm -rf ./$(DEPDIR)
+distclean: distclean-recursive
+	-rm -f $(am__CONFIG_DISTCLEAN_FILES)
 	-rm -f Makefile
-distclean-am: clean-am distclean-compile distclean-generic \
-	distclean-hdr distclean-tags
+distclean-am: clean-am distclean-generic distclean-tags
 
-dvi: dvi-am
+dvi: dvi-recursive
 
 dvi-am:
 
-html: html-am
+html: html-recursive
 
 html-am:
 
-info: info-am
+info: info-recursive
 
 info-am:
 
-install-data-am: install-includeHEADERS
+install-data-am:
 
-install-dvi: install-dvi-am
+install-dvi: install-dvi-recursive
 
 install-dvi-am:
 
-install-exec-am: install-libLIBRARIES
+install-exec-am:
 
-install-html: install-html-am
+install-html: install-html-recursive
 
 install-html-am:
 
-install-info: install-info-am
+install-info: install-info-recursive
 
 install-info-am:
 
 install-man:
 
-install-pdf: install-pdf-am
+install-pdf: install-pdf-recursive
 
 install-pdf-am:
 
-install-ps: install-ps-am
+install-ps: install-ps-recursive
 
 install-ps-am:
 
 installcheck-am:
 
-maintainer-clean: maintainer-clean-am
-	-rm -rf ./$(DEPDIR)
+maintainer-clean: maintainer-clean-recursive
+	-rm -f $(am__CONFIG_DISTCLEAN_FILES)
+	-rm -rf $(top_srcdir)/autom4te.cache
 	-rm -f Makefile
 maintainer-clean-am: distclean-am maintainer-clean-generic
 
-mostlyclean: mostlyclean-am
+mostlyclean: mostlyclean-recursive
 
-mostlyclean-am: mostlyclean-compile mostlyclean-generic
+mostlyclean-am: mostlyclean-generic
 
-pdf: pdf-am
+pdf: pdf-recursive
 
 pdf-am:
 
-ps: ps-am
+ps: ps-recursive
 
 ps-am:
 
-uninstall-am: uninstall-includeHEADERS uninstall-libLIBRARIES
+uninstall-am:
 
-.MAKE: all install-am install-strip
+.MAKE: $(am__recursive_targets) install-am install-strip
 
-.PHONY: CTAGS GTAGS TAGS all all-am check check-am clean clean-generic \
-	clean-libLIBRARIES cscopelist-am ctags ctags-am distclean \
-	distclean-compile distclean-generic distclean-hdr \
-	distclean-tags distdir dvi dvi-am html html-am info info-am \
-	install install-am install-data install-data-am install-dvi \
-	install-dvi-am install-exec install-exec-am install-html \
-	install-html-am install-includeHEADERS install-info \
-	install-info-am install-libLIBRARIES install-man install-pdf \
-	install-pdf-am install-ps install-ps-am install-strip \
-	installcheck installcheck-am installdirs maintainer-clean \
-	maintainer-clean-generic mostlyclean mostlyclean-compile \
-	mostlyclean-generic pdf pdf-am ps ps-am tags tags-am uninstall \
-	uninstall-am uninstall-includeHEADERS uninstall-libLIBRARIES
+.PHONY: $(am__recursive_targets) CTAGS GTAGS TAGS all all-am \
+	am--refresh check check-am clean clean-cscope clean-generic \
+	cscope cscopelist-am ctags ctags-am dist dist-all dist-bzip2 \
+	dist-gzip dist-lzip dist-shar dist-tarZ dist-xz dist-zip \
+	distcheck distclean distclean-generic distclean-tags \
+	distcleancheck distdir distuninstallcheck dvi dvi-am html \
+	html-am info info-am install install-am install-data \
+	install-data-am install-dvi install-dvi-am install-exec \
+	install-exec-am install-html install-html-am install-info \
+	install-info-am install-man install-pdf install-pdf-am \
+	install-ps install-ps-am install-strip installcheck \
+	installcheck-am installdirs installdirs-am maintainer-clean \
+	maintainer-clean-generic mostlyclean mostlyclean-generic pdf \
+	pdf-am ps ps-am tags tags-am uninstall uninstall-am
 
 .PRECIOUS: Makefile
 
diff --git a/TODO b/TODO
index bfe506f7..ee0e0aeb 100644
--- a/TODO
+++ b/TODO
@@ -10,6 +10,18 @@ FUNCTIONALITY:
 * How to do U[mu] ... lorentz part of type structure or not. more like chroma if not. -- DONE
 
 * subdirs lib, tests ??                              ----- DONE
+  - lib/math
+  - lib/cartesian
+  - lib/cshift
+  - lib/stencil
+  - lib/communicator
+  - lib/algorithms
+  - lib/fileio
+
+  - lib/qcd
+  - lib/qcd/actions
+  - lib/qcd/measurements
+  - lib/qcd/
 
 Not done, or just incomplete
 

From aee6669d0baeb1eb5118d82a915ab9b752cbfad5 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 21:22:50 +0100
Subject: [PATCH 078/429] Build reorg with which I am a bit happier

---
 lib/{ => cartesian}/Grid_cartesian_base.h        | 0
 lib/{ => cartesian}/Grid_cartesian_full.h        | 0
 lib/{ => cartesian}/Grid_cartesian_red_black.h   | 0
 lib/{ => communicator}/Grid_communicator_fake.cc | 0
 lib/{ => communicator}/Grid_communicator_mpi.cc  | 0
 lib/{ => cshift}/Grid_cshift_common.h            | 0
 lib/{ => cshift}/Grid_cshift_mpi.h               | 0
 lib/{ => cshift}/Grid_cshift_none.h              | 0
 lib/{ => lattice}/Grid_lattice_arith.h           | 0
 lib/{ => lattice}/Grid_lattice_conformable.h     | 0
 lib/{ => lattice}/Grid_lattice_coordinate.h      | 0
 lib/{ => lattice}/Grid_lattice_local.h           | 0
 lib/{ => lattice}/Grid_lattice_peekpoke.h        | 0
 lib/{ => lattice}/Grid_lattice_reality.h         | 0
 lib/{ => lattice}/Grid_lattice_reduction.h       | 0
 lib/{ => lattice}/Grid_lattice_rng.h             | 0
 lib/{ => lattice}/Grid_lattice_trace.h           | 0
 lib/{ => lattice}/Grid_lattice_transfer.h        | 0
 lib/{ => lattice}/Grid_lattice_transpose.h       | 0
 lib/{ => math}/Grid_math_arith.h                 | 0
 lib/{ => math}/Grid_math_arith_add.h             | 0
 lib/{ => math}/Grid_math_arith_mac.h             | 0
 lib/{ => math}/Grid_math_arith_mul.h             | 0
 lib/{ => math}/Grid_math_arith_scalar.h          | 0
 lib/{ => math}/Grid_math_arith_sub.h             | 0
 lib/{ => math}/Grid_math_inner.h                 | 0
 lib/{ => math}/Grid_math_outer.h                 | 0
 lib/{ => math}/Grid_math_peek.h                  | 0
 lib/{ => math}/Grid_math_poke.h                  | 0
 lib/{ => math}/Grid_math_reality.h               | 0
 lib/{ => math}/Grid_math_tensors.h               | 0
 lib/{ => math}/Grid_math_trace.h                 | 0
 lib/{ => math}/Grid_math_traits.h                | 0
 lib/{ => math}/Grid_math_transpose.h             | 0
 lib/{ => qcd}/Grid_QCD.h                         | 0
 lib/{ => simd}/Grid_vComplexD.h                  | 0
 lib/{ => simd}/Grid_vComplexF.h                  | 0
 lib/{ => simd}/Grid_vInteger.h                   | 0
 lib/{ => simd}/Grid_vRealD.h                     | 0
 lib/{ => simd}/Grid_vRealF.h                     | 0
 lib/{ => stencil}/Grid_stencil_common.cc         | 0
 41 files changed, 0 insertions(+), 0 deletions(-)
 rename lib/{ => cartesian}/Grid_cartesian_base.h (100%)
 rename lib/{ => cartesian}/Grid_cartesian_full.h (100%)
 rename lib/{ => cartesian}/Grid_cartesian_red_black.h (100%)
 rename lib/{ => communicator}/Grid_communicator_fake.cc (100%)
 rename lib/{ => communicator}/Grid_communicator_mpi.cc (100%)
 rename lib/{ => cshift}/Grid_cshift_common.h (100%)
 rename lib/{ => cshift}/Grid_cshift_mpi.h (100%)
 rename lib/{ => cshift}/Grid_cshift_none.h (100%)
 rename lib/{ => lattice}/Grid_lattice_arith.h (100%)
 rename lib/{ => lattice}/Grid_lattice_conformable.h (100%)
 rename lib/{ => lattice}/Grid_lattice_coordinate.h (100%)
 rename lib/{ => lattice}/Grid_lattice_local.h (100%)
 rename lib/{ => lattice}/Grid_lattice_peekpoke.h (100%)
 rename lib/{ => lattice}/Grid_lattice_reality.h (100%)
 rename lib/{ => lattice}/Grid_lattice_reduction.h (100%)
 rename lib/{ => lattice}/Grid_lattice_rng.h (100%)
 rename lib/{ => lattice}/Grid_lattice_trace.h (100%)
 rename lib/{ => lattice}/Grid_lattice_transfer.h (100%)
 rename lib/{ => lattice}/Grid_lattice_transpose.h (100%)
 rename lib/{ => math}/Grid_math_arith.h (100%)
 rename lib/{ => math}/Grid_math_arith_add.h (100%)
 rename lib/{ => math}/Grid_math_arith_mac.h (100%)
 rename lib/{ => math}/Grid_math_arith_mul.h (100%)
 rename lib/{ => math}/Grid_math_arith_scalar.h (100%)
 rename lib/{ => math}/Grid_math_arith_sub.h (100%)
 rename lib/{ => math}/Grid_math_inner.h (100%)
 rename lib/{ => math}/Grid_math_outer.h (100%)
 rename lib/{ => math}/Grid_math_peek.h (100%)
 rename lib/{ => math}/Grid_math_poke.h (100%)
 rename lib/{ => math}/Grid_math_reality.h (100%)
 rename lib/{ => math}/Grid_math_tensors.h (100%)
 rename lib/{ => math}/Grid_math_trace.h (100%)
 rename lib/{ => math}/Grid_math_traits.h (100%)
 rename lib/{ => math}/Grid_math_transpose.h (100%)
 rename lib/{ => qcd}/Grid_QCD.h (100%)
 rename lib/{ => simd}/Grid_vComplexD.h (100%)
 rename lib/{ => simd}/Grid_vComplexF.h (100%)
 rename lib/{ => simd}/Grid_vInteger.h (100%)
 rename lib/{ => simd}/Grid_vRealD.h (100%)
 rename lib/{ => simd}/Grid_vRealF.h (100%)
 rename lib/{ => stencil}/Grid_stencil_common.cc (100%)

diff --git a/lib/Grid_cartesian_base.h b/lib/cartesian/Grid_cartesian_base.h
similarity index 100%
rename from lib/Grid_cartesian_base.h
rename to lib/cartesian/Grid_cartesian_base.h
diff --git a/lib/Grid_cartesian_full.h b/lib/cartesian/Grid_cartesian_full.h
similarity index 100%
rename from lib/Grid_cartesian_full.h
rename to lib/cartesian/Grid_cartesian_full.h
diff --git a/lib/Grid_cartesian_red_black.h b/lib/cartesian/Grid_cartesian_red_black.h
similarity index 100%
rename from lib/Grid_cartesian_red_black.h
rename to lib/cartesian/Grid_cartesian_red_black.h
diff --git a/lib/Grid_communicator_fake.cc b/lib/communicator/Grid_communicator_fake.cc
similarity index 100%
rename from lib/Grid_communicator_fake.cc
rename to lib/communicator/Grid_communicator_fake.cc
diff --git a/lib/Grid_communicator_mpi.cc b/lib/communicator/Grid_communicator_mpi.cc
similarity index 100%
rename from lib/Grid_communicator_mpi.cc
rename to lib/communicator/Grid_communicator_mpi.cc
diff --git a/lib/Grid_cshift_common.h b/lib/cshift/Grid_cshift_common.h
similarity index 100%
rename from lib/Grid_cshift_common.h
rename to lib/cshift/Grid_cshift_common.h
diff --git a/lib/Grid_cshift_mpi.h b/lib/cshift/Grid_cshift_mpi.h
similarity index 100%
rename from lib/Grid_cshift_mpi.h
rename to lib/cshift/Grid_cshift_mpi.h
diff --git a/lib/Grid_cshift_none.h b/lib/cshift/Grid_cshift_none.h
similarity index 100%
rename from lib/Grid_cshift_none.h
rename to lib/cshift/Grid_cshift_none.h
diff --git a/lib/Grid_lattice_arith.h b/lib/lattice/Grid_lattice_arith.h
similarity index 100%
rename from lib/Grid_lattice_arith.h
rename to lib/lattice/Grid_lattice_arith.h
diff --git a/lib/Grid_lattice_conformable.h b/lib/lattice/Grid_lattice_conformable.h
similarity index 100%
rename from lib/Grid_lattice_conformable.h
rename to lib/lattice/Grid_lattice_conformable.h
diff --git a/lib/Grid_lattice_coordinate.h b/lib/lattice/Grid_lattice_coordinate.h
similarity index 100%
rename from lib/Grid_lattice_coordinate.h
rename to lib/lattice/Grid_lattice_coordinate.h
diff --git a/lib/Grid_lattice_local.h b/lib/lattice/Grid_lattice_local.h
similarity index 100%
rename from lib/Grid_lattice_local.h
rename to lib/lattice/Grid_lattice_local.h
diff --git a/lib/Grid_lattice_peekpoke.h b/lib/lattice/Grid_lattice_peekpoke.h
similarity index 100%
rename from lib/Grid_lattice_peekpoke.h
rename to lib/lattice/Grid_lattice_peekpoke.h
diff --git a/lib/Grid_lattice_reality.h b/lib/lattice/Grid_lattice_reality.h
similarity index 100%
rename from lib/Grid_lattice_reality.h
rename to lib/lattice/Grid_lattice_reality.h
diff --git a/lib/Grid_lattice_reduction.h b/lib/lattice/Grid_lattice_reduction.h
similarity index 100%
rename from lib/Grid_lattice_reduction.h
rename to lib/lattice/Grid_lattice_reduction.h
diff --git a/lib/Grid_lattice_rng.h b/lib/lattice/Grid_lattice_rng.h
similarity index 100%
rename from lib/Grid_lattice_rng.h
rename to lib/lattice/Grid_lattice_rng.h
diff --git a/lib/Grid_lattice_trace.h b/lib/lattice/Grid_lattice_trace.h
similarity index 100%
rename from lib/Grid_lattice_trace.h
rename to lib/lattice/Grid_lattice_trace.h
diff --git a/lib/Grid_lattice_transfer.h b/lib/lattice/Grid_lattice_transfer.h
similarity index 100%
rename from lib/Grid_lattice_transfer.h
rename to lib/lattice/Grid_lattice_transfer.h
diff --git a/lib/Grid_lattice_transpose.h b/lib/lattice/Grid_lattice_transpose.h
similarity index 100%
rename from lib/Grid_lattice_transpose.h
rename to lib/lattice/Grid_lattice_transpose.h
diff --git a/lib/Grid_math_arith.h b/lib/math/Grid_math_arith.h
similarity index 100%
rename from lib/Grid_math_arith.h
rename to lib/math/Grid_math_arith.h
diff --git a/lib/Grid_math_arith_add.h b/lib/math/Grid_math_arith_add.h
similarity index 100%
rename from lib/Grid_math_arith_add.h
rename to lib/math/Grid_math_arith_add.h
diff --git a/lib/Grid_math_arith_mac.h b/lib/math/Grid_math_arith_mac.h
similarity index 100%
rename from lib/Grid_math_arith_mac.h
rename to lib/math/Grid_math_arith_mac.h
diff --git a/lib/Grid_math_arith_mul.h b/lib/math/Grid_math_arith_mul.h
similarity index 100%
rename from lib/Grid_math_arith_mul.h
rename to lib/math/Grid_math_arith_mul.h
diff --git a/lib/Grid_math_arith_scalar.h b/lib/math/Grid_math_arith_scalar.h
similarity index 100%
rename from lib/Grid_math_arith_scalar.h
rename to lib/math/Grid_math_arith_scalar.h
diff --git a/lib/Grid_math_arith_sub.h b/lib/math/Grid_math_arith_sub.h
similarity index 100%
rename from lib/Grid_math_arith_sub.h
rename to lib/math/Grid_math_arith_sub.h
diff --git a/lib/Grid_math_inner.h b/lib/math/Grid_math_inner.h
similarity index 100%
rename from lib/Grid_math_inner.h
rename to lib/math/Grid_math_inner.h
diff --git a/lib/Grid_math_outer.h b/lib/math/Grid_math_outer.h
similarity index 100%
rename from lib/Grid_math_outer.h
rename to lib/math/Grid_math_outer.h
diff --git a/lib/Grid_math_peek.h b/lib/math/Grid_math_peek.h
similarity index 100%
rename from lib/Grid_math_peek.h
rename to lib/math/Grid_math_peek.h
diff --git a/lib/Grid_math_poke.h b/lib/math/Grid_math_poke.h
similarity index 100%
rename from lib/Grid_math_poke.h
rename to lib/math/Grid_math_poke.h
diff --git a/lib/Grid_math_reality.h b/lib/math/Grid_math_reality.h
similarity index 100%
rename from lib/Grid_math_reality.h
rename to lib/math/Grid_math_reality.h
diff --git a/lib/Grid_math_tensors.h b/lib/math/Grid_math_tensors.h
similarity index 100%
rename from lib/Grid_math_tensors.h
rename to lib/math/Grid_math_tensors.h
diff --git a/lib/Grid_math_trace.h b/lib/math/Grid_math_trace.h
similarity index 100%
rename from lib/Grid_math_trace.h
rename to lib/math/Grid_math_trace.h
diff --git a/lib/Grid_math_traits.h b/lib/math/Grid_math_traits.h
similarity index 100%
rename from lib/Grid_math_traits.h
rename to lib/math/Grid_math_traits.h
diff --git a/lib/Grid_math_transpose.h b/lib/math/Grid_math_transpose.h
similarity index 100%
rename from lib/Grid_math_transpose.h
rename to lib/math/Grid_math_transpose.h
diff --git a/lib/Grid_QCD.h b/lib/qcd/Grid_QCD.h
similarity index 100%
rename from lib/Grid_QCD.h
rename to lib/qcd/Grid_QCD.h
diff --git a/lib/Grid_vComplexD.h b/lib/simd/Grid_vComplexD.h
similarity index 100%
rename from lib/Grid_vComplexD.h
rename to lib/simd/Grid_vComplexD.h
diff --git a/lib/Grid_vComplexF.h b/lib/simd/Grid_vComplexF.h
similarity index 100%
rename from lib/Grid_vComplexF.h
rename to lib/simd/Grid_vComplexF.h
diff --git a/lib/Grid_vInteger.h b/lib/simd/Grid_vInteger.h
similarity index 100%
rename from lib/Grid_vInteger.h
rename to lib/simd/Grid_vInteger.h
diff --git a/lib/Grid_vRealD.h b/lib/simd/Grid_vRealD.h
similarity index 100%
rename from lib/Grid_vRealD.h
rename to lib/simd/Grid_vRealD.h
diff --git a/lib/Grid_vRealF.h b/lib/simd/Grid_vRealF.h
similarity index 100%
rename from lib/Grid_vRealF.h
rename to lib/simd/Grid_vRealF.h
diff --git a/lib/Grid_stencil_common.cc b/lib/stencil/Grid_stencil_common.cc
similarity index 100%
rename from lib/Grid_stencil_common.cc
rename to lib/stencil/Grid_stencil_common.cc

From 62fec044193331356c082433d3a3624bd62f322e Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 21:23:32 +0100
Subject: [PATCH 079/429] Reorganisation

---
 configure.ac               |  2 +-
 lib/Grid.h                 |  4 +++-
 lib/Grid_cartesian.h       |  6 +++---
 lib/Grid_cshift.h          |  9 +++++----
 lib/Grid_lattice.h         | 24 +++++++++++-------------
 lib/Grid_math.h            | 20 ++++++++++----------
 lib/Grid_simd.h            | 10 +++++-----
 lib/math/Grid_math_arith.h | 10 +++++-----
 8 files changed, 43 insertions(+), 42 deletions(-)

diff --git a/configure.ac b/configure.ac
index 07c1974e..97fdbf85 100644
--- a/configure.ac
+++ b/configure.ac
@@ -1,6 +1,6 @@
 # Process this file with autoconf to produce a configure script.
 AC_INIT([Grid], [1.0], [paboyle@ph.ed.ac.uk])
-AM_INIT_AUTOMAKE
+AM_INIT_AUTOMAKE(subdir-objects)
 AC_CONFIG_MACRO_DIR([m4])
 AC_CONFIG_SRCDIR([lib/Grid.h])
 AC_CONFIG_HEADERS([lib/Grid_config.h])
diff --git a/lib/Grid.h b/lib/Grid.h
index 093302cb..c456bf8a 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -47,8 +47,10 @@
 #include <Grid_cartesian.h>
 #include <Grid_lattice.h>
 #include <Grid_comparison.h>
+#include <Grid_cshift.h>
+#include <Grid_where.h>
 #include <Grid_stencil.h>
-#include <Grid_QCD.h>
+#include <qcd/Grid_QCD.h>
 
 namespace Grid {
 
diff --git a/lib/Grid_cartesian.h b/lib/Grid_cartesian.h
index dbac98a8..c01be20a 100644
--- a/lib/Grid_cartesian.h
+++ b/lib/Grid_cartesian.h
@@ -1,8 +1,8 @@
 #ifndef GRID_CARTESIAN_H
 #define GRID_CARTESIAN_H
 
-#include <Grid_cartesian_base.h>
-#include <Grid_cartesian_full.h>
-#include <Grid_cartesian_red_black.h> 
+#include <cartesian/Grid_cartesian_base.h>
+#include <cartesian/Grid_cartesian_full.h>
+#include <cartesian/Grid_cartesian_red_black.h> 
 
 #endif
diff --git a/lib/Grid_cshift.h b/lib/Grid_cshift.h
index 2df2370e..1c601b7d 100644
--- a/lib/Grid_cshift.h
+++ b/lib/Grid_cshift.h
@@ -1,16 +1,17 @@
 #ifndef _GRID_CSHIFT_H_
 #define _GRID_CSHIFT_H_
-#include <Grid_cshift_common.h>
+
+#include <cshift/Grid_cshift_common.h>
 
 #ifdef GRID_COMMS_NONE
-#include <Grid_cshift_none.h>
+#include <cshift/Grid_cshift_none.h>
 #endif
 
 #ifdef GRID_COMMS_FAKE
-#include <Grid_cshift_fake.h>
+#include <cshift/Grid_cshift_fake.h>
 #endif
 
 #ifdef GRID_COMMS_MPI
-#include <Grid_cshift_mpi.h>
+#include <cshift/Grid_cshift_mpi.h>
 #endif 
 #endif
diff --git a/lib/Grid_lattice.h b/lib/Grid_lattice.h
index cc115c3c..03bacf2d 100644
--- a/lib/Grid_lattice.h
+++ b/lib/Grid_lattice.h
@@ -77,19 +77,17 @@ public:
  }; // class Lattice
 }
 
-#include <Grid_lattice_conformable.h>
-#include <Grid_lattice_arith.h>
-#include <Grid_lattice_trace.h>
-#include <Grid_lattice_transpose.h>
-#include <Grid_lattice_local.h>
-#include <Grid_lattice_reduction.h>
-#include <Grid_lattice_peekpoke.h>
-#include <Grid_lattice_reality.h>
-#include <Grid_cshift.h>
-#include <Grid_where.h>
-#include <Grid_lattice_coordinate.h>
-#include <Grid_lattice_rng.h>
-#include <Grid_lattice_transfer.h>
+#include <lattice/Grid_lattice_conformable.h>
+#include <lattice/Grid_lattice_arith.h>
+#include <lattice/Grid_lattice_trace.h>
+#include <lattice/Grid_lattice_transpose.h>
+#include <lattice/Grid_lattice_local.h>
+#include <lattice/Grid_lattice_reduction.h>
+#include <lattice/Grid_lattice_peekpoke.h>
+#include <lattice/Grid_lattice_reality.h>
+#include <lattice/Grid_lattice_coordinate.h>
+#include <lattice/Grid_lattice_rng.h>
+#include <lattice/Grid_lattice_transfer.h>
 
 
 
diff --git a/lib/Grid_math.h b/lib/Grid_math.h
index b1e04c19..17bc09a5 100644
--- a/lib/Grid_math.h
+++ b/lib/Grid_math.h
@@ -1,16 +1,16 @@
 #ifndef GRID_MATH_H
 #define GRID_MATH_H
 
-#include <Grid_math_traits.h>
-#include <Grid_math_tensors.h>
-#include <Grid_math_arith.h>
-#include <Grid_math_inner.h>
-#include <Grid_math_outer.h>
-#include <Grid_math_transpose.h>
-#include <Grid_math_trace.h>
-#include <Grid_math_peek.h>
-#include <Grid_math_poke.h>
-#include <Grid_math_reality.h>
+#include <math/Grid_math_traits.h>
+#include <math/Grid_math_tensors.h>
+#include <math/Grid_math_arith.h>
+#include <math/Grid_math_inner.h>
+#include <math/Grid_math_outer.h>
+#include <math/Grid_math_transpose.h>
+#include <math/Grid_math_trace.h>
+#include <math/Grid_math_peek.h>
+#include <math/Grid_math_poke.h>
+#include <math/Grid_math_reality.h>
 
     
 #endif
diff --git a/lib/Grid_simd.h b/lib/Grid_simd.h
index dbde0fd7..9bf10fc9 100644
--- a/lib/Grid_simd.h
+++ b/lib/Grid_simd.h
@@ -229,11 +229,11 @@ inline void Gpermute(vsimd &y,const vsimd &b,int perm){
     };
 };
 
-#include <Grid_vInteger.h>
-#include <Grid_vRealF.h>
-#include <Grid_vRealD.h>
-#include <Grid_vComplexF.h>
-#include <Grid_vComplexD.h>
+#include <simd/Grid_vInteger.h>
+#include <simd/Grid_vRealF.h>
+#include <simd/Grid_vRealD.h>
+#include <simd/Grid_vComplexF.h>
+#include <simd/Grid_vComplexD.h>
 
 
 namespace Grid {
diff --git a/lib/math/Grid_math_arith.h b/lib/math/Grid_math_arith.h
index e7b6b4ac..ca90ba88 100644
--- a/lib/math/Grid_math_arith.h
+++ b/lib/math/Grid_math_arith.h
@@ -1,11 +1,11 @@
 #ifndef GRID_MATH_ARITH_H
 #define GRID_MATH_ARITH_H
 
-#include <Grid_math_arith_add.h>
-#include <Grid_math_arith_sub.h>
-#include <Grid_math_arith_mac.h>
-#include <Grid_math_arith_mul.h>
-#include <Grid_math_arith_scalar.h>
+#include <math/Grid_math_arith_add.h>
+#include <math/Grid_math_arith_sub.h>
+#include <math/Grid_math_arith_mac.h>
+#include <math/Grid_math_arith_mul.h>
+#include <math/Grid_math_arith_scalar.h>
 
 #endif
 

From 1556c2ba3f38d98145a162ad021a6946bb4e913e Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 21:24:10 +0100
Subject: [PATCH 080/429] Finishing the reorg

---
 lib/Grid_init.cc   |  0
 lib/Makefile.am    | 85 +++++++++++++++++++++++-----------------------
 tests/Grid_main.cc |  2 +-
 3 files changed, 44 insertions(+), 43 deletions(-)
 mode change 100755 => 100644 lib/Grid_init.cc

diff --git a/lib/Grid_init.cc b/lib/Grid_init.cc
old mode 100755
new mode 100644
diff --git a/lib/Makefile.am b/lib/Makefile.am
index dab5c6d3..20c1f867 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -3,12 +3,12 @@ AM_CXXFLAGS = -I$(top_srcdir)/
 
 extra_sources=
 if BUILD_COMMS_MPI
-  extra_sources+=Grid_communicator_mpi.cc
-  extra_sources+=Grid_stencil_common.cc
+  extra_sources+=communicator/Grid_communicator_mpi.cc
+  extra_sources+=stencil/Grid_stencil_common.cc
 endif
 if BUILD_COMMS_NONE
-  extra_sources+=Grid_communicator_fake.cc
-  extra_sources+=Grid_stencil_common.cc
+  extra_sources+=communicator/Grid_communicator_fake.cc
+  extra_sources+=stencil/Grid_stencil_common.cc
 endif
 
 #
@@ -22,55 +22,56 @@ libGrid_a_SOURCES = Grid_init.cc $(extra_sources)
 #
 include_HEADERS =\
 	Grid.h\
-	Grid_QCD.h\
 	Grid_aligned_allocator.h\
 	Grid_cartesian.h\
-	Grid_cartesian_base.h\
-	Grid_cartesian_full.h\
-	Grid_cartesian_red_black.h\
 	Grid_communicator.h\
 	Grid_comparison.h\
 	Grid_config.h\
 	Grid_cshift.h\
-	Grid_cshift_common.h\
-	Grid_cshift_mpi.h\
-	Grid_cshift_none.h\
 	Grid_lattice.h\
-	Grid_lattice_arith.h\
-	Grid_lattice_conformable.h\
-	Grid_lattice_coordinate.h\
-	Grid_lattice_local.h\
-	Grid_lattice_peekpoke.h\
-	Grid_lattice_reality.h\
-	Grid_lattice_reduction.h
-	Grid_lattice_rng.h\
-	Grid_lattice_trace.h\
-	Grid_lattice_transfer.h\
-	Grid_lattice_transpose.h\
 	Grid_math.h\
-	Grid_math_arith.h\
-	Grid_math_arith_add.h\
-	Grid_math_arith_mac.h\
-	Grid_math_arith_mul.h\
-	Grid_math_arith_scalar.h\
-	Grid_math_arith_sub.h\
-	Grid_math_inner.h\
-	Grid_math_outer.h\
-	Grid_math_peek.h\
-	Grid_math_poke.h\
-	Grid_math_reality.h\
-	Grid_math_tensors.h\
-	Grid_math_trace.h\
-	Grid_math_traits.h\
-	Grid_math_transpose.h\
 	Grid_simd.h\
 	Grid_stencil.h\
 	Grid_summation.h\
-	Grid_vComplexD.h\
-	Grid_vComplexF.h\
-	Grid_vInteger.h\
-	Grid_vRealD.h\
-	Grid_vRealF.h\
 	Grid_where.h
 
+nobase_include_HEADERS=\
+	cartesian/Grid_cartesian_base.h\
+	cartesian/Grid_cartesian_full.h\
+	cartesian/Grid_cartesian_red_black.h\
+	cshift/Grid_cshift_common.h\
+	cshift/Grid_cshift_mpi.h\
+	cshift/Grid_cshift_none.h\
+	lattice/Grid_lattice_arith.h\
+	lattice/Grid_lattice_conformable.h\
+	lattice/Grid_lattice_coordinate.h\
+	lattice/Grid_lattice_local.h\
+	lattice/Grid_lattice_peekpoke.h\
+	lattice/Grid_lattice_reality.h\
+	lattice/Grid_lattice_reduction.h\
+	lattice/Grid_lattice_rng.h\
+	lattice/Grid_lattice_trace.h\
+	lattice/Grid_lattice_transfer.h\
+	lattice/Grid_lattice_transpose.h\
+	math/Grid_math_arith.h\
+	math/Grid_math_arith_add.h\
+	math/Grid_math_arith_mac.h\
+	math/Grid_math_arith_mul.h\
+	math/Grid_math_arith_scalar.h\
+	math/Grid_math_arith_sub.h\
+	math/Grid_math_inner.h\
+	math/Grid_math_outer.h\
+	math/Grid_math_peek.h\
+	math/Grid_math_poke.h\
+	math/Grid_math_reality.h\
+	math/Grid_math_tensors.h\
+	math/Grid_math_trace.h\
+	math/Grid_math_traits.h\
+	math/Grid_math_transpose.h\
+	qcd/Grid_QCD.h\
+	simd/Grid_vComplexD.h\
+	simd/Grid_vComplexF.h\
+	simd/Grid_vInteger.h\
+	simd/Grid_vRealD.h\
+	simd/Grid_vRealF.h
 
diff --git a/tests/Grid_main.cc b/tests/Grid_main.cc
index e58d20e2..bd206ea1 100644
--- a/tests/Grid_main.cc
+++ b/tests/Grid_main.cc
@@ -277,7 +277,7 @@ int main (int argc, char ** argv)
         peekSite(bar,Bar,coor);
         for(int r=0;r<3;r++){
         for(int c=0;c<3;c++){
-	  //	  cout<<"bar "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "<<bar._internal._internal[r][c]<<std::endl;
+	  cout<<"bar "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "<<bar()()(r,c)<<std::endl;
 	}}
       }}}}
     } 

From 4e1a3aee821c23f3cd0d77636db90b7ed7b58fdf Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 22:16:31 +0100
Subject: [PATCH 081/429] Update

---
 TODO | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/TODO b/TODO
index ee0e0aeb..867ca248 100644
--- a/TODO
+++ b/TODO
@@ -16,12 +16,12 @@ FUNCTIONALITY:
   - lib/stencil
   - lib/communicator
   - lib/algorithms
-  - lib/fileio
-
   - lib/qcd
+
+  - lib/io/   -- GridLog, GridIn, GridErr, GridDebug, GridMessage
   - lib/qcd/actions
   - lib/qcd/measurements
-  - lib/qcd/
+
 
 Not done, or just incomplete
 

From f64d39ab5734817f1b2939f5865722cac83eefd5 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 18 Apr 2015 22:17:01 +0100
Subject: [PATCH 082/429] Split all OMP directives into lattice subdir for easy
 maintainance of parallelism and future OMP 4.0 offload.

---
 lib/Grid_comparison.h                 | 132 +------------------------
 lib/Grid_init.cc                      |   2 +-
 lib/lattice/Grid_lattice_comparison.h | 133 ++++++++++++++++++++++++++
 3 files changed, 138 insertions(+), 129 deletions(-)
 create mode 100644 lib/lattice/Grid_lattice_comparison.h

diff --git a/lib/Grid_comparison.h b/lib/Grid_comparison.h
index c890d342..b91f6f91 100644
--- a/lib/Grid_comparison.h
+++ b/lib/Grid_comparison.h
@@ -1,5 +1,6 @@
 #ifndef GRID_COMPARISON_H
 #define GRID_COMPARISON_H
+
 namespace Grid {
 
     // Generic list of functors
@@ -132,133 +133,8 @@ namespace Grid {
     {
       return Comparison(sne<Integer,Integer>(),lhs,rhs);
     }
-
-    //////////////////////////////////////////////////////////////////////////
-    // relational operators
-    // 
-    // Support <,>,<=,>=,==,!=
-    //
-    //Query supporting bitwise &, |, ^, !
-    //Query supporting logical &&, ||, 
-    //////////////////////////////////////////////////////////////////////////
-    template<class vfunctor,class lobj,class robj> 
-    inline Lattice<vInteger> LLComparison(vfunctor op,const Lattice<lobj> &lhs,const Lattice<robj> &rhs)
-    {
-      Lattice<vInteger> ret(rhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<rhs._grid->oSites(); ss++){
-	  ret._odata[ss]=op(lhs._odata[ss],rhs._odata[ss]);
-        }
-        return ret;
-    }
-    template<class vfunctor,class lobj,class robj> 
-    inline Lattice<vInteger> LSComparison(vfunctor op,const Lattice<lobj> &lhs,const robj &rhs)
-    {
-      Lattice<vInteger> ret(lhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites(); ss++){
-	  ret._odata[ss]=op(lhs._odata[ss],rhs);
-        }
-        return ret;
-    }
-    template<class vfunctor,class lobj,class robj> 
-    inline Lattice<vInteger> SLComparison(vfunctor op,const lobj &lhs,const Lattice<robj> &rhs)
-    {
-      Lattice<vInteger> ret(rhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<rhs._grid->oSites(); ss++){
-	  ret._odata[ss]=op(lhs._odata[ss],rhs);
-        }
-        return ret;
-    }
-
-    // Less than
-   template<class lobj,class robj>
-   inline Lattice<vInteger> operator < (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
-     return LLComparison(vlt<lobj,robj>(),lhs,rhs);
-   }
-   template<class lobj,class robj>
-   inline Lattice<vInteger> operator < (const Lattice<lobj> & lhs, const robj & rhs) {
-     return LSComparison(vlt<lobj,robj>(),lhs,rhs);
-   }
-   template<class lobj,class robj>
-   inline Lattice<vInteger> operator < (const lobj & lhs, const Lattice<robj> & rhs) {
-     return SLComparison(vlt<lobj,robj>(),lhs,rhs);
-   }
-
-   // Less than equal
-   template<class lobj,class robj>
-   inline Lattice<vInteger> operator <= (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
-     return LLComparison(vle<lobj,robj>(),lhs,rhs);
-   }
-   template<class lobj,class robj>
-   inline Lattice<vInteger> operator <= (const Lattice<lobj> & lhs, const robj & rhs) {
-     return LSComparison(vle<lobj,robj>(),lhs,rhs);
-   }
-   template<class lobj,class robj>
-   inline Lattice<vInteger> operator <= (const lobj & lhs, const Lattice<robj> & rhs) {
-     return SLComparison(vle<lobj,robj>(),lhs,rhs);
-   }
-
-   // Greater than 
-   template<class lobj,class robj>
-   inline Lattice<vInteger> operator > (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
-     return LLComparison(vgt<lobj,robj>(),lhs,rhs);
-   }
-   template<class lobj,class robj>
-   inline Lattice<vInteger> operator > (const Lattice<lobj> & lhs, const robj & rhs) {
-     return LSComparison(vgt<lobj,robj>(),lhs,rhs);
-   }
-   template<class lobj,class robj>
-   inline Lattice<vInteger> operator > (const lobj & lhs, const Lattice<robj> & rhs) {
-     return SLComparison(vgt<lobj,robj>(),lhs,rhs);
-   }
-
-
-   // Greater than equal
-   template<class lobj,class robj>
-   inline Lattice<vInteger> operator >= (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
-     return LLComparison(vge<lobj,robj>(),lhs,rhs);
-   }
-   template<class lobj,class robj>
-   inline Lattice<vInteger> operator >= (const Lattice<lobj> & lhs, const robj & rhs) {
-     return LSComparison(vge<lobj,robj>(),lhs,rhs);
-   }
-   template<class lobj,class robj>
-   inline Lattice<vInteger> operator >= (const lobj & lhs, const Lattice<robj> & rhs) {
-     return SLComparison(vge<lobj,robj>(),lhs,rhs);
-   }
-
-
-   // equal
-   template<class lobj,class robj>
-   inline Lattice<vInteger> operator == (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
-     return LLComparison(veq<lobj,robj>(),lhs,rhs);
-   }
-   template<class lobj,class robj>
-   inline Lattice<vInteger> operator == (const Lattice<lobj> & lhs, const robj & rhs) {
-     return LSComparison(veq<lobj,robj>(),lhs,rhs);
-   }
-   template<class lobj,class robj>
-   inline Lattice<vInteger> operator == (const lobj & lhs, const Lattice<robj> & rhs) {
-     return SLComparison(veq<lobj,robj>(),lhs,rhs);
-   }
-
-
-   // not equal
-   template<class lobj,class robj>
-   inline Lattice<vInteger> operator != (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
-     return LLComparison(vne<lobj,robj>(),lhs,rhs);
-   }
-   template<class lobj,class robj>
-   inline Lattice<vInteger> operator != (const Lattice<lobj> & lhs, const robj & rhs) {
-     return LSComparison(vne<lobj,robj>(),lhs,rhs);
-   }
-   template<class lobj,class robj>
-   inline Lattice<vInteger> operator != (const lobj & lhs, const Lattice<robj> & rhs) {
-     return SLComparison(vne<lobj,robj>(),lhs,rhs);
-   }
-
-
 }
+
+#include <lattice/Grid_lattice_comparison.h>
+
 #endif
diff --git a/lib/Grid_init.cc b/lib/Grid_init.cc
index 9e8b8b96..ee11a982 100644
--- a/lib/Grid_init.cc
+++ b/lib/Grid_init.cc
@@ -11,7 +11,7 @@
 #include <sys/time.h>
 #include <signal.h>
 
-#include "Grid.h"
+#include <Grid.h>
 
 #undef __X86_64
 namespace Grid {
diff --git a/lib/lattice/Grid_lattice_comparison.h b/lib/lattice/Grid_lattice_comparison.h
new file mode 100644
index 00000000..82c98572
--- /dev/null
+++ b/lib/lattice/Grid_lattice_comparison.h
@@ -0,0 +1,133 @@
+#ifndef GRID_LATTICE_COMPARISON_H
+#define GRID_LATTICE_COMPARISON_H
+
+namespace Grid {
+
+    //////////////////////////////////////////////////////////////////////////
+    // relational operators
+    // 
+    // Support <,>,<=,>=,==,!=
+    //
+    //Query supporting bitwise &, |, ^, !
+    //Query supporting logical &&, ||, 
+    //////////////////////////////////////////////////////////////////////////
+    template<class vfunctor,class lobj,class robj> 
+    inline Lattice<vInteger> LLComparison(vfunctor op,const Lattice<lobj> &lhs,const Lattice<robj> &rhs)
+    {
+      Lattice<vInteger> ret(rhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<rhs._grid->oSites(); ss++){
+	  ret._odata[ss]=op(lhs._odata[ss],rhs._odata[ss]);
+        }
+        return ret;
+    }
+    template<class vfunctor,class lobj,class robj> 
+    inline Lattice<vInteger> LSComparison(vfunctor op,const Lattice<lobj> &lhs,const robj &rhs)
+    {
+      Lattice<vInteger> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites(); ss++){
+	  ret._odata[ss]=op(lhs._odata[ss],rhs);
+        }
+        return ret;
+    }
+    template<class vfunctor,class lobj,class robj> 
+    inline Lattice<vInteger> SLComparison(vfunctor op,const lobj &lhs,const Lattice<robj> &rhs)
+    {
+      Lattice<vInteger> ret(rhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<rhs._grid->oSites(); ss++){
+	  ret._odata[ss]=op(lhs._odata[ss],rhs);
+        }
+        return ret;
+    }
+
+    // Less than
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator < (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
+     return LLComparison(vlt<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator < (const Lattice<lobj> & lhs, const robj & rhs) {
+     return LSComparison(vlt<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator < (const lobj & lhs, const Lattice<robj> & rhs) {
+     return SLComparison(vlt<lobj,robj>(),lhs,rhs);
+   }
+
+   // Less than equal
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator <= (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
+     return LLComparison(vle<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator <= (const Lattice<lobj> & lhs, const robj & rhs) {
+     return LSComparison(vle<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator <= (const lobj & lhs, const Lattice<robj> & rhs) {
+     return SLComparison(vle<lobj,robj>(),lhs,rhs);
+   }
+
+   // Greater than 
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator > (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
+     return LLComparison(vgt<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator > (const Lattice<lobj> & lhs, const robj & rhs) {
+     return LSComparison(vgt<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator > (const lobj & lhs, const Lattice<robj> & rhs) {
+     return SLComparison(vgt<lobj,robj>(),lhs,rhs);
+   }
+
+
+   // Greater than equal
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator >= (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
+     return LLComparison(vge<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator >= (const Lattice<lobj> & lhs, const robj & rhs) {
+     return LSComparison(vge<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator >= (const lobj & lhs, const Lattice<robj> & rhs) {
+     return SLComparison(vge<lobj,robj>(),lhs,rhs);
+   }
+
+
+   // equal
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator == (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
+     return LLComparison(veq<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator == (const Lattice<lobj> & lhs, const robj & rhs) {
+     return LSComparison(veq<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator == (const lobj & lhs, const Lattice<robj> & rhs) {
+     return SLComparison(veq<lobj,robj>(),lhs,rhs);
+   }
+
+
+   // not equal
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator != (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
+     return LLComparison(vne<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator != (const Lattice<lobj> & lhs, const robj & rhs) {
+     return LSComparison(vne<lobj,robj>(),lhs,rhs);
+   }
+   template<class lobj,class robj>
+   inline Lattice<vInteger> operator != (const lobj & lhs, const Lattice<robj> & rhs) {
+     return SLComparison(vne<lobj,robj>(),lhs,rhs);
+   }
+
+}
+#endif

From 650410cb2f82f2b1356c85ad7ab9f4defcd5c763 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sun, 19 Apr 2015 14:55:16 +0100
Subject: [PATCH 083/429] Update to task list

---
 TODO | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/TODO b/TODO
index 867ca248..a8fa5ce9 100644
--- a/TODO
+++ b/TODO
@@ -17,21 +17,22 @@ FUNCTIONALITY:
   - lib/communicator
   - lib/algorithms
   - lib/qcd
-
+ future
   - lib/io/   -- GridLog, GridIn, GridErr, GridDebug, GridMessage
   - lib/qcd/actions
   - lib/qcd/measurements
 
 
 Not done, or just incomplete
+* random number generation
 
 * Consider switch std::vector to boost arrays.
   boost::multi_array<type, 3> A()...    to replace multi1d, multi2d etc..
 
-
 * How to define simple matrix operations, such as flavour matrices?
 
 * Dirac, Pauli, SU subgroup, etc.. * Gamma/Dirac structures
+
 * Fourspin, two spin project
 
 * su3 exponentiation, log etc.. [Jamie's code?]

From a5b0c492d75f176d98f24563f320af2459775599 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sun, 19 Apr 2015 14:55:58 +0100
Subject: [PATCH 084/429] Rework of RNG to use C++11 random. Should work
 correctly maintaining parallel RNG across a machine. If a "fixedSeed" is
 used, randoms should be reproducible across different machine decomposition
 since the generators are physically indexed and assigned in lexico ordering.

---
 lib/cartesian/Grid_cartesian_base.h      |  51 +++++---
 lib/cartesian/Grid_cartesian_full.h      |   6 +-
 lib/cartesian/Grid_cartesian_red_black.h |   9 +-
 lib/lattice/Grid_lattice_conformable.h   |   4 +
 lib/lattice/Grid_lattice_rng.h           | 144 +++++++++++++++++++----
 tests/Grid_main.cc                       |  25 ++--
 tests/test_Grid_stencil.cc               |   7 +-
 7 files changed, 191 insertions(+), 55 deletions(-)

diff --git a/lib/cartesian/Grid_cartesian_base.h b/lib/cartesian/Grid_cartesian_base.h
index 2731d9a5..9ee10293 100644
--- a/lib/cartesian/Grid_cartesian_base.h
+++ b/lib/cartesian/Grid_cartesian_base.h
@@ -6,22 +6,25 @@
 
 namespace Grid{
 
-class GridBase : public CartesianCommunicator {
+  class GridRNG ; // Forward declaration;
+
+  class GridBase : public CartesianCommunicator {
 public:
 
- // Give Lattice access
- template<class object> friend class Lattice;
-  
- GridBase(std::vector<int> & processor_grid) : CartesianCommunicator(processor_grid) {};
+    // Give Lattice access
+    template<class object> friend class Lattice;
 
-        
- //FIXME 
- // protected:
- // Lattice wide random support. not yet fully implemented. Need seed strategy
- // and one generator per site.
- // std::default_random_engine generator;
- //    static std::mt19937  generator( 9 );
- 
+    GridRNG *_rng;
+
+    GridBase(std::vector<int> & processor_grid) : CartesianCommunicator(processor_grid) {};
+            
+    //FIXME 
+    // protected:
+    // Lattice wide random support. not yet fully implemented. Need seed strategy
+    // and one generator per site.
+    // std::default_random_engine generator;
+    //    static std::mt19937  generator( 9 );
+    
     //////////////////////////////////////////////////////////////////////
     // Commicator provides information on the processor grid
     //////////////////////////////////////////////////////////////////////
@@ -32,7 +35,7 @@ public:
     //////////////////////////////////////////////////////////////////////
 
     // Physics Grid information.
-    std::vector<int> _simd_layout;     // Which dimensions get relayed out over simd lanes.
+    std::vector<int> _simd_layout;// Which dimensions get relayed out over simd lanes.
     std::vector<int> _fdimensions;// Global dimensions of array prior to cb removal
     std::vector<int> _gdimensions;// Global dimensions of array after cb removal
     std::vector<int> _ldimensions;// local dimensions of array with processor images removed
@@ -41,6 +44,8 @@ public:
     std::vector<int> _istride;    // Inner stride i.e. within simd lane
     int _osites;                  // _isites*_osites = product(dimensions).
     int _isites;
+    int _fsites;                  // _isites*_osites = product(dimensions).
+    int _gsites;
     std::vector<int> _slice_block;   // subslice information
     std::vector<int> _slice_stride;
     std::vector<int> _slice_nblock;
@@ -149,6 +154,21 @@ public:
     ////////////////////////////////////////////////////////////////
     // Global addressing
     ////////////////////////////////////////////////////////////////
+    void GlobalIndexToGlobalCoor(int gidx,std::vector<int> &gcoor){
+      gcoor.resize(_ndimension);
+      for(int d=0;d<_ndimension;d++){
+	gcoor[d] = gidx % _gdimensions[d];
+	gidx     = gidx / _gdimensions[d];
+      }
+    }
+    void GlobalCoorToGlobalIndex(const std::vector<int> & gcoor,int & gidx){
+      gidx=0;
+      int mult=1;
+      for(int mu=0;mu<_ndimension;mu++) {
+	gidx+=mult*gcoor[mu];
+	mult*=_gdimensions[mu];
+      }
+    }
     void RankIndexToGlobalCoor(int rank, int o_idx, int i_idx , std::vector<int> &gcoor)
     {
       gcoor.resize(_ndimension);
@@ -194,7 +214,8 @@ public:
       i_idx= iIndex(lcoor);
       o_idx= oIndex(lcoor);
     }
-
 };
+
+
 }
 #endif
diff --git a/lib/cartesian/Grid_cartesian_full.h b/lib/cartesian/Grid_cartesian_full.h
index 383bf14f..73bd08b3 100644
--- a/lib/cartesian/Grid_cartesian_full.h
+++ b/lib/cartesian/Grid_cartesian_full.h
@@ -43,12 +43,14 @@ public:
         _ostride.resize(_ndimension);
         _istride.resize(_ndimension);
             
-        _osites = 1;
-        _isites = 1;
+        _fsites = _gsites = _osites = _isites = 1;
+
         for(int d=0;d<_ndimension;d++){
 	  _fdimensions[d] = dimensions[d]; // Global dimensions
 	  _gdimensions[d] = _fdimensions[d]; // Global dimensions
 	  _simd_layout[d] = simd_layout[d];
+	  _fsites = _fsites * _fdimensions[d];
+	  _gsites = _gsites * _gdimensions[d];
 
 	  //FIXME check for exact division
 
diff --git a/lib/cartesian/Grid_cartesian_red_black.h b/lib/cartesian/Grid_cartesian_red_black.h
index e85fd3f7..55bd1f20 100644
--- a/lib/cartesian/Grid_cartesian_red_black.h
+++ b/lib/cartesian/Grid_cartesian_red_black.h
@@ -62,14 +62,17 @@ public:
         _ostride.resize(_ndimension);
         _istride.resize(_ndimension);
         
-        _osites = 1;
-        _isites = 1;
+        _fsites = _gsites = _osites = _isites = 1;
+
         for(int d=0;d<_ndimension;d++){
             _fdimensions[d] = dimensions[d];
             _gdimensions[d] = _fdimensions[d];
+	    _fsites = _fsites * _fdimensions[d];
+	    _gsites = _gsites * _gdimensions[d];
+                
             if (d==0) _gdimensions[0] = _gdimensions[0]/2; // Remove a checkerboard
             _ldimensions[d] = _gdimensions[d]/_processors[d];
-                
+
             // Use a reduced simd grid
             _simd_layout[d] = simd_layout[d];
             _rdimensions[d]= _ldimensions[d]/_simd_layout[d];
diff --git a/lib/lattice/Grid_lattice_conformable.h b/lib/lattice/Grid_lattice_conformable.h
index b9af33b4..faa8c7a7 100644
--- a/lib/lattice/Grid_lattice_conformable.h
+++ b/lib/lattice/Grid_lattice_conformable.h
@@ -9,6 +9,10 @@ namespace Grid {
         assert(lhs._grid == rhs._grid);
         assert(lhs.checkerboard == rhs.checkerboard);
     }
+    void inline conformable(const GridBase *lhs,GridBase *rhs)
+    {
+        assert(lhs == rhs);
+    }
 
 }
 #endif
diff --git a/lib/lattice/Grid_lattice_rng.h b/lib/lattice/Grid_lattice_rng.h
index 636bfb72..24627c92 100644
--- a/lib/lattice/Grid_lattice_rng.h
+++ b/lib/lattice/Grid_lattice_rng.h
@@ -1,32 +1,136 @@
 #ifndef GRID_LATTICE_RNG_H
 #define GRID_LATTICE_RNG_H
 
+#include <random>
+
 namespace Grid {
 
-    // FIXME Randomise; deprecate this
-    template <class vobj> inline void random(Lattice<vobj> &l){
-        Real *v_ptr = (Real *)&l._odata[0];
-        size_t v_len = l._grid->oSites()*sizeof(vobj);
-        size_t d_len = v_len/sizeof(Real);
-	
-        for(int i=0;i<d_len;i++){
+  // Wrap seed_seq to give common interface with random_device
+  class fixedSeed {
+  public:
 
-            v_ptr[i]=drand48();
-        }
-    };
+    typedef std::seed_seq::result_type result_type;
+
+    std::seed_seq src;
     
-    // FIXME Implement a consistent seed management strategy
-    template <class vobj> inline void gaussian(Lattice<vobj> &l){
-        // Zero mean, unit variance.
-        std::normal_distribution<double> distribution(0.0,1.0);
-        Real *v_ptr = (Real *)&l._odata[0];
-        size_t v_len = l._grid->oSites()*sizeof(vobj);
-        size_t d_len = v_len/sizeof(Real);
+    fixedSeed(std::vector<int> &seeds) : src(seeds.begin(),seeds.end()) {};
 
-        for(int i=0;i<d_len;i++){
-	  v_ptr[i]= drand48();
-        }
+    result_type operator () (void){
+
+      std::vector<result_type> list(1);
+
+      src.generate(list.begin(),list.end());
+
+      return list[0];
+
+    }
+
+  };
+  class GridRNG {
+  public:
+    // One generator per site.
+    std::vector<std::ranlux48>             _generators;
+    
+    // Uniform and Gaussian distributions from these generators.
+    std::uniform_real_distribution<double> _uniform;
+    std::normal_distribution<double>       _gaussian;
+
+    GridBase *_grid;
+    int _vol;
+
+    int generator_idx(int os,int is){
+      return is*_grid->oSites()+os;
+    }
+
+    GridRNG(GridBase *grid) : _uniform{0,1}, _gaussian(0.0,1.0) {
+      _grid=grid;
+      _vol =_grid->iSites()*_grid->oSites();
+      _generators.resize(_vol);
+      //      SeedFixedIntegers(seeds);
+      // worst case we seed properly but non-deterministically
+      SeedRandomDevice();
+    }
+
+    // FIXME: drive seeding from node zero and transmit to all
+    // to get unique randoms on each node
+    void SeedRandomDevice(void){
+      std::random_device rd;
+      Seed(rd);
+    }
+    void SeedFixedIntegers(std::vector<int> &seeds){
+      fixedSeed src(seeds);
+      Seed(src);
+    }
+
+    // This loop could be made faster to avoid the Ahmdahl by
+    // i)  seed generators on each timeslice, for x=y=z=0;
+    // ii) seed generators on each z for x=y=0
+    // iii)seed generators on each y,z for x=0
+    // iv) seed generators on each y,z,x 
+    // made possible by physical indexing.
+    template<class source> void Seed(source &src)
+    {
+      std::vector<int> gcoor;
+
+      for(int gidx=0;gidx<_grid->_gsites;gidx++){
+	int rank,o_idx,i_idx;
+	_grid->GlobalIndexToGlobalCoor(gidx,gcoor);
+	_grid->GlobalCoorToRankIndex(rank,o_idx,i_idx,gcoor);
+
+	int l_idx=generator_idx(o_idx,i_idx);
+
+	typename source::result_type init = src();
+
+	_grid->Broadcast(0,(void *)&init,sizeof(init));
+	if( rank == _grid->ThisRank() ){
+	  _generators[l_idx] = std::ranlux48(init);
+	}
+      }
+    }    
+
+    //FIXME implement generic IO and create state save/restore
+    //void SaveState(const std::string<char> &file);
+    //void LoadState(const std::string<char> &file);
+
+    template <class vobj,class distribution> inline void fill(Lattice<vobj> &l,distribution &dist){
+
+      typedef typename vobj::scalar_type scalar_type;
+      typedef typename vobj::vector_type vector_type;
+      
+      conformable(_grid,l._grid);
+
+      int     Nsimd =_grid->Nsimd();
+      int     osites=_grid->oSites();
+
+      int words = sizeof(vobj)/sizeof(vector_type);
+      std::vector<std::vector<scalar_type> > buf(Nsimd,std::vector<scalar_type>(words));
+      std::vector<scalar_type *> pointers(Nsimd);  
+      
+      for(int ss=0;ss<osites;ss++){
+
+	for(int si=0;si<Nsimd;si++){
+
+	  int gdx = generator_idx(ss,si); // index of generator state
+
+	  pointers[si] = (scalar_type *)&buf[si][0];
+	  for(int idx=0;idx<words;idx++){
+	    pointers[si][idx] = dist(_generators[gdx]);
+	  }
+
+	}
+	// merge into SIMD lanes
+	merge(l._odata[ss],pointers);
+      }
     };
+  };
 
+  // FIXME Implement a consistent seed management strategy
+  template <class vobj> inline void random(GridRNG &rng,Lattice<vobj> &l){
+    rng.fill(l,rng._uniform);
+  }
+
+  template <class vobj> inline void gaussian(GridRNG &rng,Lattice<vobj> &l){
+    rng.fill(l,rng._gaussian);
+  }
 }
 #endif
diff --git a/tests/Grid_main.cc b/tests/Grid_main.cc
index bd206ea1..83cacf92 100644
--- a/tests/Grid_main.cc
+++ b/tests/Grid_main.cc
@@ -57,7 +57,8 @@ int main (int argc, char ** argv)
     
     GridCartesian Fine(latt_size,simd_layout,mpi_layout);
     GridRedBlackCartesian rbFine(latt_size,simd_layout,mpi_layout);
-    
+    GridRNG       FineRNG(&Fine);
+
     LatticeColourMatrix Foo(&Fine);
     LatticeColourMatrix Bar(&Fine);
 
@@ -91,17 +92,17 @@ int main (int argc, char ** argv)
     iSpinMatrix<vComplex> iGammaFive;
     ColourMatrix cmat;
     
-    random(Foo);
-    gaussian(Bar);
-    random(scFoo);
-    random(scBar);
+    random(FineRNG,Foo);
+    gaussian(FineRNG,Bar);
+    random(FineRNG,scFoo);
+    random(FineRNG,scBar);
 
-    random(cMat);
-    random(sMat);
-    random(scMat);
-    random(cVec);
-    random(sVec);
-    random(scVec);
+    random(FineRNG,cMat);
+    random(FineRNG,sMat);
+    random(FineRNG,scMat);
+    random(FineRNG,cVec);
+    random(FineRNG,sVec);
+    random(FineRNG,scVec);
 
     fflush(stdout);
     cVec = cMat * cVec;  // LatticeColourVector     = LatticeColourMatrix     * LatticeColourVector
@@ -277,7 +278,7 @@ int main (int argc, char ** argv)
         peekSite(bar,Bar,coor);
         for(int r=0;r<3;r++){
         for(int c=0;c<3;c++){
-	  cout<<"bar "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "<<bar()()(r,c)<<std::endl;
+	  //	  cout<<"bar "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "<<bar()()(r,c)<<std::endl;
 	}}
       }}}}
     } 
diff --git a/tests/test_Grid_stencil.cc b/tests/test_Grid_stencil.cc
index 3ed32f3a..917cb558 100644
--- a/tests/test_Grid_stencil.cc
+++ b/tests/test_Grid_stencil.cc
@@ -48,14 +48,15 @@ int main (int argc, char ** argv)
     
     GridCartesian Fine(latt_size,simd_layout,mpi_layout);
     GridRedBlackCartesian rbFine(latt_size,simd_layout,mpi_layout);
-    
+    GridRNG       fRNG(&Fine);
+
     LatticeColourMatrix Foo(&Fine);
     LatticeColourMatrix Bar(&Fine);
     LatticeColourMatrix Check(&Fine);
     LatticeColourMatrix Diff(&Fine);
     
-    random(Foo);
-    gaussian(Bar);
+    random(fRNG,Foo);
+    gaussian(fRNG,Bar);
 
 
     for(int dir=0;dir<4;dir++){

From 1851327d19eb5433ecc27f6c6f5562bfb1df3d84 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 22 Apr 2015 22:46:48 +0100
Subject: [PATCH 085/429] Got the NERSC IO working and fixed a bug in cshift.

---
 configure                                  |  22 +
 configure.ac                               |   2 +
 lib/Grid.h                                 |   1 +
 lib/Grid_communicator.h                    |   7 +-
 lib/Grid_comparison.h                      |  24 +-
 lib/Grid_config.h                          |   6 +
 lib/Grid_config.h.in                       |   6 +
 lib/Grid_lattice.h                         |  12 +-
 lib/cartesian/Grid_cartesian_base.h        |  41 +-
 lib/communicator/Grid_communicator_fake.cc |   1 +
 lib/communicator/Grid_communicator_mpi.cc  |   3 +
 lib/cshift/Grid_cshift_mpi.h               |   8 +-
 lib/lattice/Grid_lattice_coordinate.h      |   9 +-
 lib/lattice/Grid_lattice_peekpoke.h        |  84 +++-
 lib/lattice/Grid_lattice_reduction.h       |  30 ++
 lib/math/Grid_math_tensors.h               |  41 +-
 lib/math/Grid_math_traits.h                |  14 +
 lib/parallelIO/GridNerscIO.h               | 524 +++++++++++++++++++++
 lib/qcd/Grid_QCD.h                         |   4 +-
 tests/Grid_cshift.cc                       |  84 ++++
 tests/Grid_nersc_io.cc                     |  62 +++
 tests/Makefile.am                          |   8 +-
 22 files changed, 925 insertions(+), 68 deletions(-)
 create mode 100644 lib/parallelIO/GridNerscIO.h
 create mode 100644 tests/Grid_cshift.cc
 create mode 100644 tests/Grid_nersc_io.cc

diff --git a/configure b/configure
index fc0b6453..48fafcf1 100755
--- a/configure
+++ b/configure
@@ -4999,6 +4999,28 @@ _ACEOF
 fi
 done
 
+for ac_func in ntohll
+do :
+  ac_fn_c_check_func "$LINENO" "ntohll" "ac_cv_func_ntohll"
+if test "x$ac_cv_func_ntohll" = xyes; then :
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_NTOHLL 1
+_ACEOF
+
+fi
+done
+
+for ac_func in be64toh
+do :
+  ac_fn_c_check_func "$LINENO" "be64toh" "ac_cv_func_be64toh"
+if test "x$ac_cv_func_be64toh" = xyes; then :
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_BE64TOH 1
+_ACEOF
+
+fi
+done
+
 
 # Check whether --enable-simd was given.
 if test "${enable_simd+set}" = set; then :
diff --git a/configure.ac b/configure.ac
index 97fdbf85..21f40f6c 100644
--- a/configure.ac
+++ b/configure.ac
@@ -25,6 +25,8 @@ AC_TYPE_UINT64_T
 
 # Checks for library functions.
 AC_CHECK_FUNCS([gettimeofday])
+AC_CHECK_FUNCS([ntohll])
+AC_CHECK_FUNCS([be64toh])
 
 AC_ARG_ENABLE([simd],[AC_HELP_STRING([--enable-simd=SSE|AVX|AVX2|AVX512],[Select instructions])],[ac_SIMD=${enable_simd}],[ac_SIMD=AVX2])
 
diff --git a/lib/Grid.h b/lib/Grid.h
index c456bf8a..7e12490a 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -51,6 +51,7 @@
 #include <Grid_where.h>
 #include <Grid_stencil.h>
 #include <qcd/Grid_QCD.h>
+#include <parallelIO/GridNerscIO.h>
 
 namespace Grid {
 
diff --git a/lib/Grid_communicator.h b/lib/Grid_communicator.h
index 845c0123..f5baa413 100644
--- a/lib/Grid_communicator.h
+++ b/lib/Grid_communicator.h
@@ -35,6 +35,7 @@ class CartesianCommunicator {
     // Grid information queries
     /////////////////////////////////
     int                      IsBoss(void)            { return _processor==0; };
+    int                      BossRank(void)          { return 0; };
     int                      ThisRank(void)          { return _processor; };
     const std::vector<int> & ThisProcessorCoor(void) { return _processor_coor; };
     const std::vector<int> & ProcessorGrid(void)     { return _processors; };
@@ -49,6 +50,8 @@ class CartesianCommunicator {
     void GlobalSum(RealD &);
     void GlobalSumVector(RealD *,int N);
 
+    void GlobalSum(uint32_t &);
+
     void GlobalSum(ComplexF &c)
     {
       GlobalSumVector((float *)&c,2);
@@ -68,12 +71,10 @@ class CartesianCommunicator {
     }
     
     template<class obj> void GlobalSum(obj &o){
-
       typedef typename obj::scalar_type scalar_type;
       int words = sizeof(obj)/sizeof(scalar_type);
-
       scalar_type * ptr = (scalar_type *)& o;
-      GlobalSum(ptr,words);
+      GlobalSumVector(ptr,words);
     }
     ////////////////////////////////////////////////////////////
     // Face exchange
diff --git a/lib/Grid_comparison.h b/lib/Grid_comparison.h
index b91f6f91..530817c6 100644
--- a/lib/Grid_comparison.h
+++ b/lib/Grid_comparison.h
@@ -8,42 +8,42 @@ namespace Grid {
     public:
       vInteger operator()(const lobj &lhs, const robj &rhs)
 	{ 
-	  return lhs == rhs;
+	  return TensorRemove(lhs) == TensorRemove(rhs);
 	}
     };
     template<class lobj,class robj> class vne {
     public:
       vInteger operator()(const lobj &lhs, const robj &rhs)
 	{ 
-	  return lhs != rhs;
+	  return TensorRemove(lhs) != TensorRemove(rhs);
 	}
     };
     template<class lobj,class robj> class vlt {
     public:
       vInteger operator()(const lobj &lhs, const robj &rhs)
 	{ 
-	  return lhs < rhs;
+	  return TensorRemove(lhs) < TensorRemove(rhs);
 	}
     };
     template<class lobj,class robj> class vle {
     public:
       vInteger operator()(const lobj &lhs, const robj &rhs)
 	{ 
-	  return lhs <= rhs;
+	  return TensorRemove(lhs) <= TensorRemove(rhs);
 	}
     };
     template<class lobj,class robj> class vgt {
     public:
       vInteger operator()(const lobj &lhs, const robj &rhs)
 	{ 
-	  return lhs > rhs;
+	  return TensorRemove(lhs) > TensorRemove(rhs);
 	}
     };
     template<class lobj,class robj> class vge {
     public:
       vInteger operator()(const lobj &lhs, const robj &rhs)
 	{ 
-	  return lhs >= rhs;
+	  return TensorRemove(lhs) >= TensorRemove(rhs);
 	}
     };
 
@@ -52,42 +52,42 @@ namespace Grid {
     public:
       Integer operator()(const lobj &lhs, const robj &rhs)
 	{ 
-	  return lhs == rhs;
+	  return TensorRemove(lhs) == TensorRemove(rhs);
 	}
     };
     template<class lobj,class robj> class sne {
     public:
       Integer operator()(const lobj &lhs, const robj &rhs)
 	{ 
-	  return lhs != rhs;
+	  return TensorRemove(lhs) != TensorRemove(rhs);
 	}
     };
     template<class lobj,class robj> class slt {
     public:
       Integer operator()(const lobj &lhs, const robj &rhs)
 	{ 
-	  return lhs < rhs;
+	  return TensorRemove(lhs) < TensorRemove(rhs);
 	}
     };
     template<class lobj,class robj> class sle {
     public:
       Integer operator()(const lobj &lhs, const robj &rhs)
 	{ 
-	  return lhs <= rhs;
+	  return TensorRemove(lhs) <= TensorRemove(rhs);
 	}
     };
     template<class lobj,class robj> class sgt {
     public:
       Integer operator()(const lobj &lhs, const robj &rhs)
 	{ 
-	  return lhs > rhs;
+	  return TensorRemove(lhs) > TensorRemove(rhs);
 	}
     };
     template<class lobj,class robj> class sge {
     public:
       Integer operator()(const lobj &lhs, const robj &rhs)
 	{ 
-	  return lhs >= rhs;
+	  return TensorRemove(lhs) >= TensorRemove(rhs);
 	}
     };
 
diff --git a/lib/Grid_config.h b/lib/Grid_config.h
index cb676b54..2171ea6b 100644
--- a/lib/Grid_config.h
+++ b/lib/Grid_config.h
@@ -16,6 +16,9 @@
 /* GRID_COMMS_NONE */
 /* #undef GRID_COMMS_NONE */
 
+/* Define to 1 if you have the `be64toh' function. */
+/* #undef HAVE_BE64TOH */
+
 /* Define to 1 if you have the `gettimeofday' function. */
 #define HAVE_GETTIMEOFDAY 1
 
@@ -31,6 +34,9 @@
 /* Define to 1 if you have the <memory.h> header file. */
 #define HAVE_MEMORY_H 1
 
+/* Define to 1 if you have the `ntohll' function. */
+/* #undef HAVE_NTOHLL */
+
 /* Define to 1 if you have the <stdint.h> header file. */
 #define HAVE_STDINT_H 1
 
diff --git a/lib/Grid_config.h.in b/lib/Grid_config.h.in
index 467ee6e4..2381525f 100644
--- a/lib/Grid_config.h.in
+++ b/lib/Grid_config.h.in
@@ -15,6 +15,9 @@
 /* GRID_COMMS_NONE */
 #undef GRID_COMMS_NONE
 
+/* Define to 1 if you have the `be64toh' function. */
+#undef HAVE_BE64TOH
+
 /* Define to 1 if you have the `gettimeofday' function. */
 #undef HAVE_GETTIMEOFDAY
 
@@ -30,6 +33,9 @@
 /* Define to 1 if you have the <memory.h> header file. */
 #undef HAVE_MEMORY_H
 
+/* Define to 1 if you have the `ntohll' function. */
+#undef HAVE_NTOHLL
+
 /* Define to 1 if you have the <stdint.h> header file. */
 #undef HAVE_STDINT_H
 
diff --git a/lib/Grid_lattice.h b/lib/Grid_lattice.h
index 03bacf2d..7256459c 100644
--- a/lib/Grid_lattice.h
+++ b/lib/Grid_lattice.h
@@ -37,7 +37,8 @@ public:
     // what about a default grid?
     //////////////////////////////////////////////////////////////////
     Lattice(GridBase *grid) : _grid(grid) {
-        _odata.reserve(_grid->oSites());
+      //        _odata.reserve(_grid->oSites());
+        _odata.resize(_grid->oSites());
         assert((((uint64_t)&_odata[0])&0xF) ==0);
         checkerboard=0;
     }
@@ -49,6 +50,15 @@ public:
         }
         return *this;
     }
+    template<class robj> inline Lattice<vobj> & operator = (const Lattice<robj> & r){
+      conformable(*this,r);
+      std::cout<<"Lattice operator ="<<std::endl;
+
+        for(int ss=0;ss<_grid->oSites();ss++){
+            this->_odata[ss]=r._odata[ss];
+        }
+        return *this;
+    }
 
     // *=,+=,-= operators inherit behvour from correspond */+/- operation
     template<class T> inline Lattice<vobj> &operator *=(const T &r) {
diff --git a/lib/cartesian/Grid_cartesian_base.h b/lib/cartesian/Grid_cartesian_base.h
index 9ee10293..aff9375a 100644
--- a/lib/cartesian/Grid_cartesian_base.h
+++ b/lib/cartesian/Grid_cartesian_base.h
@@ -6,34 +6,23 @@
 
 namespace Grid{
 
-  class GridRNG ; // Forward declaration;
-
+  //////////////////////////////////////////////////////////////////////
+  // Commicator provides information on the processor grid
+  //////////////////////////////////////////////////////////////////////
+  //    unsigned long _ndimension;
+  //    std::vector<int> _processors; // processor grid
+  //    int              _processor;  // linear processor rank
+  //    std::vector<int> _processor_coor;  // linear processor rank
+  //////////////////////////////////////////////////////////////////////
   class GridBase : public CartesianCommunicator {
+
 public:
 
     // Give Lattice access
     template<class object> friend class Lattice;
 
-    GridRNG *_rng;
-
     GridBase(std::vector<int> & processor_grid) : CartesianCommunicator(processor_grid) {};
             
-    //FIXME 
-    // protected:
-    // Lattice wide random support. not yet fully implemented. Need seed strategy
-    // and one generator per site.
-    // std::default_random_engine generator;
-    //    static std::mt19937  generator( 9 );
-    
-    //////////////////////////////////////////////////////////////////////
-    // Commicator provides information on the processor grid
-    //////////////////////////////////////////////////////////////////////
-    //    unsigned long _ndimension;
-    //    std::vector<int> _processors; // processor grid
-    //    int              _processor;  // linear processor rank
-    //    std::vector<int> _processor_coor;  // linear processor rank
-    //////////////////////////////////////////////////////////////////////
-
     // Physics Grid information.
     std::vector<int> _simd_layout;// Which dimensions get relayed out over simd lanes.
     std::vector<int> _fdimensions;// Global dimensions of array prior to cb removal
@@ -101,6 +90,13 @@ public:
       for(int d=0;d<_ndimension;d++) idx+=_ostride[d]*ocoor[d];
       return idx;
     }
+    inline void CoorFromIndex (std::vector<int>& coor,int index,std::vector<int> &dims){
+      coor.resize(_ndimension);
+      for(int d=0;d<_ndimension;d++){
+	coor[d] = index % dims[d];
+	index   = index / dims[d];
+      }
+    }
     inline void oCoorFromOindex (std::vector<int>& coor,int Oindex){
       coor.resize(_ndimension);
       for(int d=0;d<_ndimension;d++){
@@ -146,6 +142,7 @@ public:
     inline int lSites(void) { return _isites*_osites; }; 
     inline int gSites(void) { return _isites*_osites*_Nprocessors; }; 
     inline int Nd    (void) { return _ndimension;};
+
     inline const std::vector<int> &FullDimensions(void)         { return _fdimensions;};
     inline const std::vector<int> &GlobalDimensions(void)       { return _gdimensions;};
     inline const std::vector<int> &LocalDimensions(void)        { return _ldimensions;};
@@ -175,10 +172,10 @@ public:
       std::vector<int> coor(_ndimension);
 
       ProcessorCoorFromRank(rank,coor);
-      for(int mu=0;mu<_ndimension;mu++) gcoor[mu] = _ldimensions[mu]&coor[mu];
+      for(int mu=0;mu<_ndimension;mu++) gcoor[mu] = _ldimensions[mu]*coor[mu];
 
       iCoorFromIindex(coor,i_idx);
-      for(int mu=0;mu<_ndimension;mu++) gcoor[mu] += _rdimensions[mu]&coor[mu];
+      for(int mu=0;mu<_ndimension;mu++) gcoor[mu] += _rdimensions[mu]*coor[mu];
 
       oCoorFromOindex (coor,o_idx);
       for(int mu=0;mu<_ndimension;mu++) gcoor[mu] += coor[mu];
diff --git a/lib/communicator/Grid_communicator_fake.cc b/lib/communicator/Grid_communicator_fake.cc
index 2b94ea4b..5f4a4024 100644
--- a/lib/communicator/Grid_communicator_fake.cc
+++ b/lib/communicator/Grid_communicator_fake.cc
@@ -17,6 +17,7 @@ CartesianCommunicator::CartesianCommunicator(std::vector<int> &processors)
 void CartesianCommunicator::GlobalSum(float &){}
 void CartesianCommunicator::GlobalSumVector(float *,int N){}
 void CartesianCommunicator::GlobalSum(double &){}
+void CartesianCommunicator::GlobalSum(uint32_t &){}
 void CartesianCommunicator::GlobalSumVector(double *,int N){}
 
 // Basic Halo comms primitive
diff --git a/lib/communicator/Grid_communicator_mpi.cc b/lib/communicator/Grid_communicator_mpi.cc
index dba5f039..bd53bd9c 100644
--- a/lib/communicator/Grid_communicator_mpi.cc
+++ b/lib/communicator/Grid_communicator_mpi.cc
@@ -28,6 +28,9 @@ CartesianCommunicator::CartesianCommunicator(std::vector<int> &processors)
   assert(Size==_Nprocessors);
 }
 
+void CartesianCommunicator::GlobalSum(uint32_t &u){
+  MPI_Allreduce(MPI_IN_PLACE,&u,1,MPI_UINT32_T,MPI_SUM,communicator);
+}
 void CartesianCommunicator::GlobalSum(float &f){
   MPI_Allreduce(MPI_IN_PLACE,&f,1,MPI_FLOAT,MPI_SUM,communicator);
 }
diff --git a/lib/cshift/Grid_cshift_mpi.h b/lib/cshift/Grid_cshift_mpi.h
index c078c7d8..60894ef3 100644
--- a/lib/cshift/Grid_cshift_mpi.h
+++ b/lib/cshift/Grid_cshift_mpi.h
@@ -188,8 +188,8 @@ template<class vobj> void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &r
 
     if ( comm_any ) {
 
-      for(int i=0;i<Nsimd;i++){
-	pointers[i] = (scalar_type *)&send_buf_extract[i][0];
+      for(int i=0;i<Nsimd;i++){         // there is a reversal in extract merge
+	pointers[i] = (scalar_type *)&send_buf_extract[Nsimd-1-i][0];
       }
       Gather_plane_extract(rhs,pointers,dimension,sx,cbmask);
 
@@ -238,9 +238,9 @@ template<class vobj> void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &r
       for(int i=0;i<Nsimd;i++){
 	if ( permute_slice ) {
 	  PermuteMap=i^toggle_bit;
-	  pointers[i] = rpointers[PermuteMap];
+	  pointers[Nsimd-1-i] = rpointers[PermuteMap];
 	} else {
-	  pointers[i] = rpointers[i];
+	  pointers[Nsimd-1-i] = rpointers[i];
 	}
       }
 
diff --git a/lib/lattice/Grid_lattice_coordinate.h b/lib/lattice/Grid_lattice_coordinate.h
index 0a71ec8b..8fba8b1b 100644
--- a/lib/lattice/Grid_lattice_coordinate.h
+++ b/lib/lattice/Grid_lattice_coordinate.h
@@ -5,16 +5,17 @@ namespace Grid {
 
     template<class iobj> inline void LatticeCoordinate(Lattice<iobj> &l,int mu)
     {
+      typedef typename iobj::scalar_type scalar_type;
+      typedef typename iobj::vector_type vector_type;
       GridBase *grid = l._grid;
       int Nsimd = grid->iSites();
       std::vector<int> gcoor;
-      std::vector<Integer> mergebuf(Nsimd);
-      std::vector<Integer *> mergeptr(Nsimd);
-      vInteger vI;
+      std::vector<scalar_type> mergebuf(Nsimd);
+      std::vector<scalar_type *> mergeptr(Nsimd);
+      vector_type vI;
       for(int o=0;o<grid->oSites();o++){
 	for(int i=0;i<grid->iSites();i++){
 	  grid->RankIndexToGlobalCoor(grid->ThisRank(),o,i,gcoor);
-	  //	  grid->RankIndexToGlobalCoor(0,o,i,gcoor);
 	  mergebuf[i]=gcoor[mu];
 	  mergeptr[i]=&mergebuf[i];
 	}
diff --git a/lib/lattice/Grid_lattice_peekpoke.h b/lib/lattice/Grid_lattice_peekpoke.h
index 41212037..93245016 100644
--- a/lib/lattice/Grid_lattice_peekpoke.h
+++ b/lib/lattice/Grid_lattice_peekpoke.h
@@ -43,6 +43,7 @@ namespace Grid {
         }
         return ret;
     };
+
     ////////////////////////////////////////////////////////////////////////////////////////////////////
     // Poke internal indices of a Lattice object
     ////////////////////////////////////////////////////////////////////////////////////////////////////
@@ -88,26 +89,26 @@ namespace Grid {
       assert( sizeof(sobj)*Nsimd == sizeof(vobj));
 
       int rank,odx,idx;
-      grid->GlobalCoorToRankIndex(rank,odx,idx,site);
-
       // Optional to broadcast from node 0.
-      grid->Broadcast(0,s);
+      grid->GlobalCoorToRankIndex(rank,odx,idx,site);
+      grid->Broadcast(grid->BossRank(),s);
 
       std::vector<sobj> buf(Nsimd);
       std::vector<scalar_type *> pointers(Nsimd);  
-      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
 
       // extract-modify-merge cycle is easiest way and this is not perf critical
-      extract(l._odata[odx],pointers);
-      
-      buf[idx] = s;
-
-      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
-      merge(l._odata[odx],pointers);
+      if ( rank == grid->ThisRank() ) {
+	for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
+	extract(l._odata[odx],pointers);
+	buf[idx] = s;
+	for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
+	merge(l._odata[odx],pointers);
+      }
 
       return;
     };
 
+
     //////////////////////////////////////////////////////////
     // Peek a scalar object from the SIMD array
     //////////////////////////////////////////////////////////
@@ -138,6 +139,69 @@ namespace Grid {
       return;
     };
 
+    //////////////////////////////////////////////////////////
+    // Peek a scalar object from the SIMD array
+    //////////////////////////////////////////////////////////
+    template<class vobj,class sobj>
+    void peekLocalSite(sobj &s,Lattice<vobj> &l,std::vector<int> &site){
+        
+      GridBase *grid=l._grid;
+
+      typedef typename vobj::scalar_type scalar_type;
+      typedef typename vobj::vector_type vector_type;
+
+      int Nsimd = grid->Nsimd();
+
+      assert( l.checkerboard== l._grid->CheckerBoard(site));
+      assert( sizeof(sobj)*Nsimd == sizeof(vobj));
+
+      int odx,idx;
+      idx= grid->iIndex(site);
+      odx= grid->oIndex(site);
+
+      std::vector<sobj> buf(Nsimd);
+      std::vector<scalar_type *> pointers(Nsimd);  
+      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
+
+      extract(l._odata[odx],pointers);
+      
+      s = buf[idx];
+
+      return;
+    };
+
+    template<class vobj,class sobj>
+    void pokeLocalSite(const sobj &s,Lattice<vobj> &l,std::vector<int> &site){
+
+      GridBase *grid=l._grid;
+
+      typedef typename vobj::scalar_type scalar_type;
+      typedef typename vobj::vector_type vector_type;
+
+      int Nsimd = grid->Nsimd();
+
+      assert( l.checkerboard== l._grid->CheckerBoard(site));
+      assert( sizeof(sobj)*Nsimd == sizeof(vobj));
+
+      int odx,idx;
+      idx= grid->iIndex(site);
+      odx= grid->oIndex(site);
+
+      std::vector<sobj> buf(Nsimd);
+      std::vector<scalar_type *> pointers(Nsimd);  
+      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
+
+      // extract-modify-merge cycle is easiest way and this is not perf critical
+      extract(l._odata[odx],pointers);
+      
+      buf[idx] = s;
+
+      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
+      merge(l._odata[odx],pointers);
+
+      return;
+    };
+
 }
 #endif
 
diff --git a/lib/lattice/Grid_lattice_reduction.h b/lib/lattice/Grid_lattice_reduction.h
index 8167fab0..7d34d2ea 100644
--- a/lib/lattice/Grid_lattice_reduction.h
+++ b/lib/lattice/Grid_lattice_reduction.h
@@ -40,6 +40,36 @@ namespace Grid {
       return nrm;
     }
 
+    template<class vobj>
+      inline typename vobj::scalar_object sum(const Lattice<vobj> &arg){
+
+      GridBase *grid=arg._grid;
+      int Nsimd = grid->Nsimd();
+
+      typedef typename vobj::scalar_object sobj;
+      typedef typename vobj::scalar_type   scalar_type;
+
+      vobj vsum;
+      sobj ssum;
+
+      vsum=zero;
+      ssum=zero;
+      for(int ss=0;ss<arg._grid->oSites(); ss++){
+	vsum = vsum + arg._odata[ss];
+      }
+      
+      std::vector<sobj>               buf(Nsimd);
+      std::vector<scalar_type *> pointers(Nsimd);  
+      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
+      extract(vsum,pointers);
+
+      for(int i=0;i<Nsimd;i++) ssum = ssum + buf[i];
+
+      arg._grid->GlobalSum(ssum);
+
+      return ssum;
+    }
+
 }
 #endif
 
diff --git a/lib/math/Grid_math_tensors.h b/lib/math/Grid_math_tensors.h
index 98b9d22e..0ee30970 100644
--- a/lib/math/Grid_math_tensors.h
+++ b/lib/math/Grid_math_tensors.h
@@ -13,10 +13,12 @@ template<class vtype> class iScalar
 public:
   vtype _internal;
 
-  typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
+  typedef typename GridTypeMapper<vtype>::scalar_type   scalar_type;
   typedef typename GridTypeMapper<vtype>::vector_type vector_type;
   typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
   typedef iScalar<tensor_reduced_v> tensor_reduced;
+  typedef typename GridTypeMapper<vtype>::scalar_object recurse_scalar_object;
+  typedef iScalar<recurse_scalar_object> scalar_object;
 
   enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
 
@@ -27,7 +29,7 @@ public:
     
     iScalar(scalar_type s) : _internal(s) {};// recurse down and hit the constructor for vector_type
 
-    iScalar(Zero &z){ *this = zero; };
+    iScalar(const Zero &z){ *this = zero; };
 
     iScalar<vtype> & operator= (const Zero &hero){
         zeroit(*this);
@@ -73,11 +75,18 @@ public:
     operator ComplexD () const { return(TensorRemove(_internal)); };
     operator RealD () const { return(real(TensorRemove(_internal))); }
 
+
+    template<class T,typename std::enable_if<isGridTensor<T>::notvalue, T>::type* = nullptr > inline auto operator = (T arg) -> iScalar<vtype>
+    { 
+      _internal = arg;
+      return *this;
+    }
+
 };
 ///////////////////////////////////////////////////////////
 // Allows to turn scalar<scalar<scalar<double>>>> back to double.
 ///////////////////////////////////////////////////////////
-template<class T> inline typename std::enable_if<isGridTensor<T>::notvalue, T>::type TensorRemove(T arg) { return arg;}
+template<class T>     inline typename std::enable_if<isGridTensor<T>::notvalue, T>::type TensorRemove(T arg) { return arg;}
 template<class vtype> inline auto TensorRemove(iScalar<vtype> arg) -> decltype(TensorRemove(arg._internal))
 {
   return TensorRemove(arg._internal);
@@ -91,14 +100,17 @@ public:
   typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
   typedef typename GridTypeMapper<vtype>::vector_type vector_type;
   typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
+  typedef typename GridTypeMapper<vtype>::scalar_object recurse_scalar_object;
+  typedef iScalar<tensor_reduced_v> tensor_reduced;
+  typedef iVector<recurse_scalar_object,N> scalar_object;
+
 
   enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
-  typedef iScalar<tensor_reduced_v> tensor_reduced;
 
-    iVector(Zero &z){ *this = zero; };
+    iVector(const Zero &z){ *this = zero; };
     iVector() {};// Empty constructure
 
-    iVector<vtype,N> & operator= (Zero &hero){
+    iVector<vtype,N> & operator= (const Zero &hero){
         zeroit(*this);
         return *this;
     }
@@ -153,18 +165,27 @@ public:
 
   typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
   typedef typename GridTypeMapper<vtype>::vector_type vector_type;
-
   typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
+  typedef typename GridTypeMapper<vtype>::scalar_object recurse_scalar_object;
+  typedef iScalar<tensor_reduced_v> tensor_reduced;
+  typedef iMatrix<recurse_scalar_object,N> scalar_object;
 
   enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
-  typedef iScalar<tensor_reduced_v> tensor_reduced;
 
-  iMatrix(Zero &z){ *this = zero; };
+  iMatrix(const Zero &z){ *this = zero; };
   iMatrix() {};
-  iMatrix<vtype,N> & operator= (Zero &hero){
+  iMatrix<vtype,N> & operator= (const Zero &hero){
     zeroit(*this);
     return *this;
   }
+  template<class T,typename std::enable_if<isGridTensor<T>::notvalue, T>::type* = nullptr > inline auto operator = (T arg) -> iMatrix<vtype,N>
+    { 
+      zeroit(*this);
+      for(int i=0;i<N;i++)
+	_internal[i][i] = arg;
+      return *this;
+    }
+
   friend void zeroit(iMatrix<vtype,N> &that){
     for(int i=0;i<N;i++){
       for(int j=0;j<N;j++){
diff --git a/lib/math/Grid_math_traits.h b/lib/math/Grid_math_traits.h
index 8ecc2712..353cdd88 100644
--- a/lib/math/Grid_math_traits.h
+++ b/lib/math/Grid_math_traits.h
@@ -25,6 +25,7 @@ namespace Grid {
     typedef typename T::scalar_type scalar_type;
     typedef typename T::vector_type vector_type;
     typedef typename T::tensor_reduced tensor_reduced;
+    typedef typename T::scalar_object scalar_object;
     enum { TensorLevel = T::TensorLevel };
   };
 
@@ -36,6 +37,7 @@ namespace Grid {
     typedef RealF scalar_type;
     typedef RealF vector_type;
     typedef RealF tensor_reduced ;
+    typedef RealF scalar_object;
     enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<RealD> {
@@ -43,6 +45,7 @@ namespace Grid {
     typedef RealD scalar_type;
     typedef RealD vector_type;
     typedef RealD tensor_reduced;
+    typedef RealD scalar_object;
     enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<ComplexF> {
@@ -50,6 +53,7 @@ namespace Grid {
     typedef ComplexF scalar_type;
     typedef ComplexF vector_type;
     typedef ComplexF tensor_reduced;
+    typedef ComplexF scalar_object;
     enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<ComplexD> {
@@ -57,6 +61,7 @@ namespace Grid {
     typedef ComplexD scalar_type;
     typedef ComplexD vector_type;
     typedef ComplexD tensor_reduced;
+    typedef ComplexD scalar_object;
     enum { TensorLevel = 0 };
   };
 
@@ -65,6 +70,7 @@ namespace Grid {
     typedef RealF  scalar_type;
     typedef vRealF vector_type;
     typedef vRealF tensor_reduced;
+    typedef RealF  scalar_object;
     enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<vRealD> {
@@ -72,6 +78,7 @@ namespace Grid {
     typedef RealD  scalar_type;
     typedef vRealD vector_type;
     typedef vRealD tensor_reduced;
+    typedef RealD  scalar_object;
     enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<vComplexF> {
@@ -79,6 +86,7 @@ namespace Grid {
     typedef ComplexF  scalar_type;
     typedef vComplexF vector_type;
     typedef vComplexF tensor_reduced;
+    typedef ComplexF  scalar_object;
     enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<vComplexD> {
@@ -86,6 +94,7 @@ namespace Grid {
     typedef ComplexD  scalar_type;
     typedef vComplexD vector_type;
     typedef vComplexD tensor_reduced;
+    typedef ComplexD  scalar_object;
     enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<vInteger> {
@@ -93,6 +102,7 @@ namespace Grid {
     typedef Integer  scalar_type;
     typedef vInteger vector_type;
     typedef vInteger tensor_reduced;
+    typedef Integer  scalar_object;
     enum { TensorLevel = 0 };
   };
 
@@ -102,6 +112,10 @@ namespace Grid {
     static const bool notvalue = false;
   };
 
+  template<> struct isGridTensor<int > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
   template<> struct isGridTensor<RealD > {
     static const bool value = false;
     static const bool notvalue = true;
diff --git a/lib/parallelIO/GridNerscIO.h b/lib/parallelIO/GridNerscIO.h
new file mode 100644
index 00000000..51820357
--- /dev/null
+++ b/lib/parallelIO/GridNerscIO.h
@@ -0,0 +1,524 @@
+#ifndef GRID_NERSC_IO_H
+#define GRID_NERSC_IO_H
+
+#include <algorithm>
+#include <iostream>
+#include <iomanip>
+#include <fstream>
+
+#include <arpa/inet.h>
+
+namespace Grid {
+
+  using namespace QCD;
+
+////////////////////////////////////////////////////////////////////////////////
+// Some data types for intermediate storage
+////////////////////////////////////////////////////////////////////////////////
+  template<typename vtype> using iLorentzColour2x3 = iVector<iVector<iVector<vtype, Nc>, 2>, 4 >;
+typedef iLorentzColour2x3<Complex>  LorentzColour2x3;
+typedef iLorentzColour2x3<ComplexF> LorentzColour2x3F;
+typedef iLorentzColour2x3<ComplexD> LorentzColour2x3D;
+
+////////////////////////////////////////////////////////////////////////////////
+// header specification/interpretation
+////////////////////////////////////////////////////////////////////////////////
+class NerscField {
+ public:
+    // header strings (not in order)
+    int dimension[4];
+    std::string boundary[4]; 
+    int data_start;
+    std::string hdr_version;
+    std::string storage_format;
+    // Checks on data
+    double link_trace;
+    double plaquette;
+    uint32_t checksum;
+    unsigned int sequence_number;
+    std::string data_type;
+    std::string ensemble_id ;
+    std::string ensemble_label ;
+    std::string creator ;
+    std::string creator_hardware ;
+    std::string creation_date ;
+    std::string archive_date ;
+    std::string floating_point;
+};
+
+
+////////////////////////////////////////////////////////////////////////////////
+// Write and read from fstream; comput header offset for payload
+////////////////////////////////////////////////////////////////////////////////
+inline unsigned int writeNerscHeader(NerscField &field,std::string file)
+{
+  std::ofstream fout(file,std::ios::out);
+  
+  fout.seekp(0,std::ios::beg);
+  fout << "BEGIN_HEADER"      << std::endl;
+  fout << "HDR_VERSION = "    << field.hdr_version    << std::endl;
+  fout << "DATATYPE = "       << field.data_type      << std::endl;
+  fout << "STORAGE_FORMAT = " << field.storage_format << std::endl;
+
+  for(int i=0;i<4;i++){
+    fout << "DIMENSION_" << i+1 << " = " << field.dimension[i] << std::endl ;
+  }
+  // just to keep the space and write it later
+  fout << "LINK_TRACE = " << std::setprecision(10) << field.link_trace << std::endl;
+  fout << "PLAQUETTE  = " << std::setprecision(10) << field.plaquette  << std::endl;
+  for(int i=0;i<4;i++){
+    fout << "BOUNDARY_"<<i+1<<" = " << field.boundary[i] << std::endl;
+  }
+  fout << "CHECKSUM = "<< std::hex << std::setw(16) << 0 << field.checksum << std::endl;
+
+  fout << "ENSEMBLE_ID = "     << field.ensemble_id      << std::endl;
+  fout << "ENSEMBLE_LABEL = "  << field.ensemble_label   << std::endl;
+  fout << "SEQUENCE_NUMBER = " << field.sequence_number  << std::endl;
+  fout << "CREATOR = "         << field.creator          << std::endl;
+  fout << "CREATOR_HARDWARE = "<< field.creator_hardware << std::endl;
+  fout << "CREATION_DATE = "   << field.creation_date    << std::endl;
+  fout << "ARCHIVE_DATE = "    << field.archive_date     << std::endl;
+  fout << "FLOATING_POINT = "  << field.floating_point   << std::endl;
+  fout << "END_HEADER"         << std::endl;
+  field.data_start = fout.tellp();
+  return field.data_start;
+}
+
+// A little helper
+inline void removeWhitespace(std::string &key)
+{
+  key.erase(std::remove_if(key.begin(), key.end(), ::isspace),key.end());
+}
+// for the header-reader
+inline int readNerscHeader(std::string file,GridBase *grid,  NerscField &field)
+{
+  int offset=0;
+  std::map<std::string,std::string> header;
+  std::string line;
+
+  //////////////////////////////////////////////////
+  // read the header
+  //////////////////////////////////////////////////
+  std::ifstream fin(file);
+
+  getline(fin,line); // read one line and insist is 
+
+  removeWhitespace(line);
+  assert(line==std::string("BEGIN_HEADER"));
+
+  do {
+    getline(fin,line); // read one line
+    int eq = line.find("=");
+    if(eq >0) {
+      std::string key=line.substr(0,eq);
+      std::string val=line.substr(eq+1);
+      removeWhitespace(key);
+      removeWhitespace(val);
+      
+      header[key] = val;
+    }
+  } while( line.find("END_HEADER") == std::string::npos );
+
+  field.data_start = fin.tellg();
+
+  //////////////////////////////////////////////////
+  // chomp the values
+  //////////////////////////////////////////////////
+
+  field.hdr_version    = header[std::string("HDR_VERSION")];
+  field.data_type      = header[std::string("DATATYPE")];
+  field.storage_format = header[std::string("STORAGE_FORMAT")];
+  
+  field.dimension[0] = std::stol(header[std::string("DIMENSION_1")]);
+  field.dimension[1] = std::stol(header[std::string("DIMENSION_2")]);
+  field.dimension[2] = std::stol(header[std::string("DIMENSION_3")]);
+  field.dimension[3] = std::stol(header[std::string("DIMENSION_4")]);
+
+  assert(grid->_ndimension == 4);
+  for(int d=0;d<4;d++){
+    assert(grid->_fdimensions[d]==field.dimension[d]);
+  }
+
+  field.link_trace = std::stod(header[std::string("LINK_TRACE")]);
+  field.plaquette  = std::stod(header[std::string("PLAQUETTE")]);
+
+  field.boundary[0] = header[std::string("BOUNDARY_1")];
+  field.boundary[1] = header[std::string("BOUNDARY_2")];
+  field.boundary[2] = header[std::string("BOUNDARY_3")];
+  field.boundary[3] = header[std::string("BOUNDARY_4")];
+
+  field.checksum = std::stoul(header[std::string("CHECKSUM")],0,16);
+  field.ensemble_id      = header[std::string("ENSEMBLE_ID")];
+  field.ensemble_label   = header[std::string("ENSEMBLE_LABEL")];
+  field.sequence_number  = std::stol(header[std::string("SEQUENCE_NUMBER")]);
+  field.creator          = header[std::string("CREATOR")];
+  field.creator_hardware = header[std::string("CREATOR_HARDWARE")];
+  field.creation_date    = header[std::string("CREATION_DATE")];
+  field.archive_date     = header[std::string("ARCHIVE_DATE")];
+  field.floating_point   = header[std::string("FLOATING_POINT")];
+
+  return field.data_start;
+}
+
+
+//////////////////////////////////////////////////////////////////////
+// Utilities
+//////////////////////////////////////////////////////////////////////
+inline void reconstruct3(LorentzColourMatrix & cm)
+{
+  const int x=0;
+  const int y=1;
+  const int z=2;
+  for(int mu=0;mu<4;mu++){
+    cm(mu)()(2,x) = adj(cm(mu)()(0,y)*cm(mu)()(1,z)-cm(mu)()(0,z)*cm(mu)()(1,y)); //x= yz-zy
+    cm(mu)()(2,y) = adj(cm(mu)()(0,z)*cm(mu)()(1,x)-cm(mu)()(0,x)*cm(mu)()(1,z)); //y= zx-xz
+    cm(mu)()(2,z) = adj(cm(mu)()(0,x)*cm(mu)()(1,y)-cm(mu)()(0,y)*cm(mu)()(1,x)); //z= xy-yx
+  }
+}
+
+
+ void inline be32toh(void *file_object,uint32_t bytes)
+ {
+   uint32_t * f = (uint32_t *)file_object;
+   for(int i=0;i*sizeof(uint32_t)<bytes;i++){  
+     f[i] = ntohl(f[i]);
+   }
+ }
+ void inline le32toh(void *file_object,uint32_t bytes)
+ {
+   uint32_t *fp = (uint32_t *)file_object;
+
+   uint32_t f;
+
+   for(int i=0;i*sizeof(uint32_t)<bytes;i++){  
+     f = fp[i];
+     // got network order and the network to host
+     f = ((f&0xFF)<<24) | ((f&0xFF00)<<8) | ((f&0xFF0000)>>8) | ((f&0xFF000000UL)>>24) ; 
+     fp[i] = ntohl(f);
+   }
+ }
+ void inline be64toh(void *file_object,uint32_t bytes)
+ {
+   uint64_t * f = (uint64_t *)file_object;
+   for(int i=0;i*sizeof(uint64_t)<bytes;i++){  
+     f[i] = ntohll(f[i]);
+   }
+ }
+ void inline le64toh(void *file_object,uint32_t bytes)
+ {
+   uint64_t *fp = (uint64_t *)file_object;
+   uint64_t f,g;
+
+   for(int i=0;i*sizeof(uint64_t)<bytes;i++){  
+     f = fp[i];
+     // got network order and the network to host
+     g = ((f&0xFF)<<24) | ((f&0xFF00)<<8) | ((f&0xFF0000)>>8) | ((f&0xFF000000UL)>>24) ; 
+     g = g << 32;
+     f = f >> 32;
+     g|= ((f&0xFF)<<24) | ((f&0xFF00)<<8) | ((f&0xFF0000)>>8) | ((f&0xFF000000UL)>>24) ; 
+     fp[i] = ntohl(g);
+   }
+ }
+
+inline void NerscChecksum(uint32_t *buf,uint32_t buf_size,uint32_t &csum)
+{
+  for(int i=0;i*sizeof(uint32_t)<buf_size;i++){
+    csum=csum+buf[i];
+  }
+}
+
+  template<class fobj,class sobj>
+  struct NerscSimpleMunger{
+    void operator() (fobj &in,sobj &out,uint32_t &csum){
+
+      for(int mu=0;mu<4;mu++){
+      for(int i=0;i<3;i++){
+      for(int j=0;j<3;j++){
+	out(mu)()(i,j) = in(mu)()(i,j);
+      }}}
+
+      NerscChecksum((uint32_t *)&in,sizeof(in),csum); 
+    };
+  };
+  template<class fobj,class sobj>
+  struct NerscSimpleUnmunger{
+    void operator() (sobj &in,fobj &out,uint32_t &csum){
+      for(int mu=0;mu<4;mu++){
+      for(int i=0;i<3;i++){
+      for(int j=0;j<3;j++){
+	out(mu)()(i,j) = in(mu)()(i,j);
+      }}}
+      NerscChecksum((uint32_t *)&out,sizeof(out),csum); 
+    };
+  };
+ 
+ template<class fobj,class sobj>
+ struct Nersc3x2munger{
+   void operator() (fobj &in,sobj &out,uint32_t &csum){
+     
+     NerscChecksum((uint32_t *)&in,sizeof(in),csum); 
+
+     for(int mu=0;mu<4;mu++){
+       for(int i=0;i<2;i++){
+       for(int j=0;j<3;j++){
+	   out(mu)()(i,j) = in(mu)(i)(j);
+       }}
+     }
+     reconstruct3(out);
+   }
+ };
+
+ template<class fobj,class sobj>
+ struct Nersc3x2unmunger{
+
+  void operator() (sobj &in,fobj &out,uint32_t &csum){
+
+    NerscChecksum((uint32_t *)&out,sizeof(out),csum); 
+
+    for(int mu=0;mu<4;mu++){
+      for(int i=0;i<2;i++){
+      for(int j=0;j<3;j++){
+	out(mu)(i)(j) = in(mu)()(i,j);
+      }}
+    }
+  }
+};
+
+////////////////////////////////////////////////////////////////////////////
+// Template wizardry to map types to strings for NERSC in an extensible way
+////////////////////////////////////////////////////////////////////////////
+ template<class vobj> struct NerscDataType {
+   static void DataType     (std::string &str) { str = std::string("4D_BINARY_UNKNOWN"); };
+   static void FloatingPoint(std::string &str) { str = std::string("IEEE64BIG"); };
+ };
+
+ template<> struct NerscDataType<iColourMatrix<ComplexD> > {
+   static void DataType     (std::string &str) { str = std::string("4D_SU3_GAUGE_3X3"); };
+   static void FloatingPoint(std::string &str) { str = std::string("IEEE64BIG");};
+ };
+
+ template<> struct NerscDataType<iColourMatrix<ComplexF> > {
+   static void DataType     (std::string &str) { str = std::string("4D_SU3_GAUGE_3X3"); };
+   static void FloatingPoint(std::string &str) { str = std::string("IEEE32BIG");};
+ };
+
+//////////////////////////////////////////////////////////////////////
+// Bit and Physical Checksumming and QA of data
+//////////////////////////////////////////////////////////////////////
+/*
+template<class vobj> inline uint32_t NerscChecksum(Lattice<vobj> & data)
+{
+  uint32_t sum;
+  for(int ss=0;ss<data._grid->Osites();ss++){
+    uint32_t *iptr = (uint32_t *)& data._odata[0] ;
+    for(int i=0;i<sizeof(vobj);i+=sizeof(uint32_t)){
+      sum=sum+iptr[i];
+    }
+  }
+  data._grid->globalSum(sum);
+  return sum;
+}
+*/
+template<class vobj> inline void NerscPhysicalCharacteristics(Lattice<vobj> & data,NerscField &header)
+{
+  header.data_type      = NerscDataType<vobj>::DataType;
+  header.floating_point = NerscDataType<vobj>::FloatingPoint;
+  return;
+}
+
+ template<> inline void NerscPhysicalCharacteristics(LatticeGaugeField & data,NerscField &header)
+{
+  NerscDataType<decltype(data._odata[0])>::DataType(header.data_type);
+  NerscDataType<decltype(data._odata[0])>::FloatingPoint(header.floating_point);
+  header.link_trace=1.0;
+  header.plaquette =1.0;
+}
+
+template<class vobj> inline void NerscStatisics(Lattice<vobj> & data,NerscField &header)
+{
+  assert(data._grid->_ndimension==4);
+
+  for(int d=0;d<4;d++)
+    header.dimension[d] = data._grid->_fdimensions[d];
+
+  // compute checksum and any physical properties contained for this type
+  //  header.checksum = NerscChecksum(data);
+
+  NerscPhysicalCharacteristics(data,header);
+}
+
+/////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+// Now the meat: the object readers
+/////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+template<class vobj,class sobj,class fobj,class munger>
+inline void readNerscObject(Lattice<vobj> &Umu,std::string file,munger munge,int offset,std::string &format)
+{
+  GridBase *grid = Umu._grid;
+
+  int ieee32big = (format == std::string("IEEE32BIG"));
+  int ieee32    = (format == std::string("IEEE32"));
+  int ieee64big = (format == std::string("IEEE64BIG"));
+  int ieee64    = (format == std::string("IEEE64"));
+
+  // Find the location of each site and send to primary node
+  //  for(int site=0; site < Layout::vol(); ++site){
+  //     multi1d<int> coord = crtesn(site, Layout::lattSize());
+  //     for(int dd=0; dd<Nd; dd++){        /* dir */
+  //        cfg_in.readArray(su3_buffer, float_size, mat_size);
+  //
+  // Above from Chroma; defines loop order now that NERSC doc no longer
+  // available (how short sighted is that?)
+  {
+    std::ifstream fin(file,std::ios::binary|std::ios::in);
+    fin.seekg(offset);
+
+    Umu = zero;
+    uint32_t csum=0;
+    fobj file_object;
+    sobj munged;
+    
+    for(int t=0;t<grid->_fdimensions[3];t++){
+    for(int z=0;z<grid->_fdimensions[2];z++){
+    for(int y=0;y<grid->_fdimensions[1];y++){
+    for(int x=0;x<grid->_fdimensions[0];x++){
+
+      std::vector<int> site({x,y,z,t});
+
+      if ( grid->IsBoss() ) {
+	fin.read((char *)&file_object,sizeof(file_object));
+
+	if(ieee32big) be32toh((void *)&file_object,sizeof(file_object));
+	if(ieee32)    le32toh((void *)&file_object,sizeof(file_object));
+	if(ieee64big) be64toh((void *)&file_object,sizeof(file_object));
+	if(ieee64)    le64toh((void *)&file_object,sizeof(file_object));
+
+	munge(file_object,munged,csum);
+      }
+      // The boss who read the file has their value poked
+      pokeSite(munged,Umu,site);
+    }}}}
+  }
+}
+
+template<class vobj,class sobj,class fobj,class munger>
+inline void writeNerscObject(Lattice<vobj> &Umu,std::string file,munger munge,int offset,
+                       int sequence,double lt,double pl)
+{
+  GridBase *grid = Umu._grid;
+  NerscField header;
+  
+  //////////////////////////////////////////////////
+  // First write the header; this is in wrong place
+  //////////////////////////////////////////////////
+  assert(grid->_ndimension == 4);
+  for(int d=0;d<4;d++){
+    header.dimension[d]=grid->_fdimensions[d];
+    header.boundary [d]=std::string("PERIODIC");; 
+  }
+  header.hdr_version=std::string("WHATDAHECK");
+  //  header.storage_format=storage_format<vobj>::string; // use template specialisation
+  //  header.data_type=data_type<vobj>::string;
+  header.storage_format=std::string("debug");
+  header.data_type     =std::string("debug");
+
+  //FIXME; use template specialisation to fill these out
+  header.link_trace   =lt;
+  header.plaquette    =pl;
+  header.checksum     =0;
+
+  //
+  header.sequence_number =sequence;
+  header.ensemble_id     =std::string("UKQCD");
+  header.ensemble_label  =std::string("UKQCD");
+  header.creator         =std::string("Tadahito");
+  header.creator_hardware=std::string("BlueGene/Q");
+  header.creation_date   =std::string("AnnoDomini");
+  header.archive_date    =std::string("AnnoDomini");
+  header.floating_point  =std::string("IEEE64BIG");
+  //  header.data_start=;
+  //  unsigned int checksum;
+
+  //////////////////////////////////////////////////
+  // Now write the body
+  //////////////////////////////////////////////////
+  {
+    std::ofstream fout(file,std::ios::binary|std::ios::in);
+    fout.seekp(offset);
+
+    Umu = zero;
+    uint32_t csum=0;
+    fobj file_object;
+    sobj unmunged;
+    for(int t=0;t<grid->_fdimensions[3];t++){
+    for(int z=0;z<grid->_fdimensions[2];z++){
+    for(int y=0;y<grid->_fdimensions[1];y++){
+    for(int x=0;x<grid->_fdimensions[0];x++){
+      std::vector<int> site({x,y,z,t});
+      peekSite(unmunged,Umu,site);
+      munge(unmunged,file_object,csum);
+      // broadcast & insert
+      fout.write((char *)&file_object,sizeof(file_object));
+    }}}}
+  }
+}
+
+
+
+inline void readNerscConfiguration(LatticeGaugeField &Umu,NerscField& header,std::string file)
+{
+  GridBase *grid = Umu._grid;
+
+  int offset = readNerscHeader(file,Umu._grid,header);
+
+  std::string format(header.floating_point);
+
+  int ieee32big = (format == std::string("IEEE32BIG"));
+  int ieee32    = (format == std::string("IEEE32"));
+  int ieee64big = (format == std::string("IEEE64BIG"));
+  int ieee64    = (format == std::string("IEEE64"));
+
+  // depending on datatype, set up munger;
+  // munger is a function of <floating point, Real, data_type>
+  if ( header.data_type == std::string("4D_SU3_GAUGE") ) {
+    if ( ieee32 || ieee32big ) {
+      readNerscObject<vLorentzColourMatrix, LorentzColourMatrix, LorentzColour2x3F> 
+	(Umu,file,
+	 Nersc3x2munger<LorentzColour2x3F,LorentzColourMatrix>(),
+	 offset,format);
+    }
+    if ( ieee64 || ieee64big ) {
+      readNerscObject<vLorentzColourMatrix, LorentzColourMatrix, LorentzColour2x3D> 
+	(Umu,file,
+	 Nersc3x2munger<LorentzColour2x3D,LorentzColourMatrix>(),
+	 offset,format);
+    }
+  } else if ( header.data_type == std::string("4D_SU3_GAUGE_3X3") ) {
+    if ( ieee32 || ieee32big ) {
+      readNerscObject<vLorentzColourMatrix,LorentzColourMatrix,LorentzColourMatrixF>
+	(Umu,file,NerscSimpleMunger<LorentzColourMatrixF,LorentzColourMatrix>(),offset,format);
+    }
+    if ( ieee64 || ieee64big ) {
+      readNerscObject<vLorentzColourMatrix,LorentzColourMatrix,LorentzColourMatrixD>
+	(Umu,file,NerscSimpleMunger<LorentzColourMatrixD,LorentzColourMatrix>(),offset,format);
+    }
+  } else {
+    assert(0);
+  }
+
+}
+
+template<class vobj>
+inline void writeNerscConfiguration(Lattice<vobj> &Umu,NerscField &header,std::string file)
+{
+  GridBase &grid = Umu._grid;
+  
+  NerscStatisics(Umu,header);
+
+  int offset = writeNerscHeader(header,file);
+
+  writeNerscObject(Umu,NerscSimpleMunger<vobj,vobj>(),offset);
+}
+
+
+}
+#endif
diff --git a/lib/qcd/Grid_QCD.h b/lib/qcd/Grid_QCD.h
index c405662d..b30b7a67 100644
--- a/lib/qcd/Grid_QCD.h
+++ b/lib/qcd/Grid_QCD.h
@@ -42,6 +42,8 @@ namespace QCD {
     typedef iColourMatrix<Complex >        ColourMatrix;
     typedef iSpinColourMatrix<Complex >    SpinColourMatrix;
     typedef iLorentzColourMatrix<Complex > LorentzColourMatrix;
+    typedef iLorentzColourMatrix<ComplexF > LorentzColourMatrixF;
+    typedef iLorentzColourMatrix<ComplexD > LorentzColourMatrixD;
 
     typedef iSpinVector<Complex >       SpinVector;
     typedef iColourVector<Complex >     ColourVector;
@@ -66,7 +68,7 @@ namespace QCD {
 
     typedef Lattice<vTReal>              LatticeReal;
     typedef Lattice<vTComplex>           LatticeComplex;
-    typedef Lattice<vInteger>            LatticeInteger; // Predicates for "where"
+    typedef Lattice<vTInteger>            LatticeInteger; // Predicates for "where"
     
     typedef Lattice<vColourMatrix>     LatticeColourMatrix;
     typedef Lattice<vSpinMatrix>       LatticeSpinMatrix;
diff --git a/tests/Grid_cshift.cc b/tests/Grid_cshift.cc
new file mode 100644
index 00000000..3fe2b9cc
--- /dev/null
+++ b/tests/Grid_cshift.cc
@@ -0,0 +1,84 @@
+#include <Grid.h>
+#include <parallelIO/GridNerscIO.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> simd_layout({1,1,2,2});
+  std::vector<int> mpi_layout ({2,2,2,2});
+  std::vector<int> latt_size  ({8,8,8,16});
+    
+  GridCartesian     Fine(latt_size,simd_layout,mpi_layout);
+  GridRNG           FineRNG(&Fine);
+
+  LatticeComplex U(&Fine);
+  LatticeComplex ShiftU(&Fine);
+
+  LatticeComplex lex(&Fine);
+  lex=zero;
+  Integer stride =1;
+  {
+    double nrm;
+    LatticeComplex coor(&Fine);
+
+    for(int d=0;d<4;d++){
+      LatticeCoordinate(coor,d);
+      lex = lex + coor*stride;
+      stride=stride*latt_size[d];
+    }
+    U=lex;
+  }
+
+  TComplex cm;
+  
+  for(int dir=0;dir<4;dir++){
+    for(int shift=0;shift<latt_size[dir];shift++){
+      if ( Fine.IsBoss() ) 
+	std::cout<<"Shifting by "<<shift<<" in direction"<<dir<<std::endl;
+
+	ShiftU  = Cshift(U,dir,shift);    // Shift everything
+
+	std::vector<int> coor(4);
+
+	for(coor[3]=0;coor[3]<latt_size[3];coor[3]++){
+	for(coor[2]=0;coor[2]<latt_size[2];coor[2]++){
+	for(coor[1]=0;coor[1]<latt_size[1];coor[1]++){
+	for(coor[0]=0;coor[0]<latt_size[0];coor[0]++){
+	  
+	  peekSite(cm,ShiftU,coor);
+
+	  double nrm=norm2(U);
+
+	  std::vector<int> scoor(coor);
+	  scoor[dir] = (scoor[dir]+shift)%latt_size[dir];
+	  
+	  Integer slex = scoor[0]
+	    + latt_size[0]*scoor[1]
+	    + latt_size[0]*latt_size[1]*scoor[2]
+	    + latt_size[0]*latt_size[1]*latt_size[2]*scoor[3];
+
+	  Complex scm(slex);
+	  
+	  nrm = abs(scm-cm()()());
+	  std::vector<int> peer(4);
+	  int index=real(cm);
+	  Fine.CoorFromIndex(peer,index,latt_size);
+
+	  if (nrm > 0){
+	    cout<<"FAIL shift "<< shift<<" in dir "<< dir<<" ["<<coor[0]<<","<<coor[1]<<","<<coor[2]<<","<<coor[3]<<"] = "<< cm()()()<<" expect "<<scm<<"  "<<nrm<<endl;
+	    cout<<"Got    "<<index<<" " << peer[0]<<","<<peer[1]<<","<<peer[2]<<","<<peer[3]<<std::endl;
+	    index=real(scm);
+	    Fine.CoorFromIndex(peer,index,latt_size);
+	    cout<<"Expect "<<index<<" " << peer[0]<<","<<peer[1]<<","<<peer[2]<<","<<peer[3]<<std::endl;
+	  }
+	}}}}
+    }
+  }
+
+  Grid_finalize();
+}
diff --git a/tests/Grid_nersc_io.cc b/tests/Grid_nersc_io.cc
new file mode 100644
index 00000000..1d965404
--- /dev/null
+++ b/tests/Grid_nersc_io.cc
@@ -0,0 +1,62 @@
+#include <Grid.h>
+#include <parallelIO/GridNerscIO.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> simd_layout({1,1,2,2});
+  std::vector<int> mpi_layout ({2,1,1,2});
+  std::vector<int> latt_size  ({16,16,16,32});
+    
+  GridCartesian     Fine(latt_size,simd_layout,mpi_layout);
+  GridRNG           FineRNG(&Fine);
+  LatticeGaugeField Umu(&Fine);
+
+  std::vector<LatticeColourMatrix> U(4,&Fine);
+  
+  NerscField header;
+  
+  std::string file("./ckpoint_lat.4000");
+  readNerscConfiguration(Umu,header,file);
+
+  for(int mu=0;mu<Nd;mu++){
+    U[mu] = peekIndex<3>(Umu,mu);
+  }
+
+  // Painful ; fix syntactical niceness
+  LatticeComplex LinkTrace(&Fine);
+  LinkTrace=zero;
+  for(int mu=0;mu<Nd;mu++){
+    LinkTrace = LinkTrace + trace(U[mu]);
+  }
+
+  // (1+2+3)=6 = N(N-1)/2 terms
+  LatticeComplex Plaq(&Fine);
+  Plaq = zero;
+  for(int mu=1;mu<Nd;mu++){
+    for(int nu=0;nu<mu;nu++){
+      Plaq = Plaq + trace(U[mu]*Cshift(U[nu],mu,1)*adj(Cshift(U[mu],nu,1))*adj(U[nu]));
+    }
+  }
+
+
+  double vol = Fine.gSites();
+
+  Complex PlaqScale(1.0/vol/6.0/3.0);
+  TComplex Tp = sum(Plaq);
+  Complex p  = TensorRemove(Tp);
+  std::cout << "calculated plaquettes " <<p*PlaqScale<<std::endl;
+
+  Complex LinkTraceScale(1.0/vol/4.0/3.0);
+  TComplex Tl = sum(LinkTrace);
+  Complex l  = TensorRemove(Tl);
+  std::cout << "calculated link trace " <<l*LinkTraceScale<<std::endl;
+
+  Grid_finalize();
+}
diff --git a/tests/Makefile.am b/tests/Makefile.am
index ec0703b1..0be04e52 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -5,10 +5,16 @@ AM_LDFLAGS = -L$(top_srcdir)/lib
 #
 # Test code
 #
-bin_PROGRAMS = Grid_main test_Grid_stencil
+bin_PROGRAMS = Grid_main test_Grid_stencil Grid_nersc_io Grid_cshift
 
 Grid_main_SOURCES = Grid_main.cc
 Grid_main_LDADD = -lGrid
 
+Grid_nersc_io_SOURCES = Grid_nersc_io.cc
+Grid_nersc_io_LDADD = -lGrid
+
+Grid_cshift_SOURCES = Grid_cshift.cc
+Grid_cshift_LDADD = -lGrid
+
 test_Grid_stencil_SOURCES = test_Grid_stencil.cc
 test_Grid_stencil_LDADD = -lGrid

From eb58297a43404ef297e98bd4684e93beaa5da0a3 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 23 Apr 2015 07:51:15 +0100
Subject: [PATCH 086/429] Fixing endian on linux I hope

---
 configure.ac                 |  3 +--
 lib/parallelIO/GridNerscIO.h | 22 ++++++++++++++--------
 2 files changed, 15 insertions(+), 10 deletions(-)

diff --git a/configure.ac b/configure.ac
index 21f40f6c..647b5e30 100644
--- a/configure.ac
+++ b/configure.ac
@@ -17,6 +17,7 @@ AC_PROG_RANLIB
 AC_CHECK_HEADERS(stdint.h)
 AC_CHECK_HEADERS(malloc/malloc.h)
 AC_CHECK_HEADERS(malloc.h)
+AC_CHECK_HEADERS(endian.h)
 
 # Checks for typedefs, structures, and compiler characteristics.
 AC_TYPE_SIZE_T
@@ -25,8 +26,6 @@ AC_TYPE_UINT64_T
 
 # Checks for library functions.
 AC_CHECK_FUNCS([gettimeofday])
-AC_CHECK_FUNCS([ntohll])
-AC_CHECK_FUNCS([be64toh])
 
 AC_ARG_ENABLE([simd],[AC_HELP_STRING([--enable-simd=SSE|AVX|AVX2|AVX512],[Select instructions])],[ac_SIMD=${enable_simd}],[ac_SIMD=AVX2])
 
diff --git a/lib/parallelIO/GridNerscIO.h b/lib/parallelIO/GridNerscIO.h
index 51820357..5d1e2e41 100644
--- a/lib/parallelIO/GridNerscIO.h
+++ b/lib/parallelIO/GridNerscIO.h
@@ -6,7 +6,13 @@
 #include <iomanip>
 #include <fstream>
 
+#ifdef HAVE_ENDIAN_H
+#include <endian.h>
 #include <arpa/inet.h>
+#define ntohll be64toh
+#else 
+#include <arpa/inet.h>
+#endif
 
 namespace Grid {
 
@@ -177,14 +183,14 @@ inline void reconstruct3(LorentzColourMatrix & cm)
 }
 
 
- void inline be32toh(void *file_object,uint32_t bytes)
+ void inline be32toh_v(void *file_object,uint32_t bytes)
  {
    uint32_t * f = (uint32_t *)file_object;
    for(int i=0;i*sizeof(uint32_t)<bytes;i++){  
      f[i] = ntohl(f[i]);
    }
  }
- void inline le32toh(void *file_object,uint32_t bytes)
+ void inline le32toh_v(void *file_object,uint32_t bytes)
  {
    uint32_t *fp = (uint32_t *)file_object;
 
@@ -197,14 +203,14 @@ inline void reconstruct3(LorentzColourMatrix & cm)
      fp[i] = ntohl(f);
    }
  }
- void inline be64toh(void *file_object,uint32_t bytes)
+ void inline be64toh_v(void *file_object,uint32_t bytes)
  {
    uint64_t * f = (uint64_t *)file_object;
    for(int i=0;i*sizeof(uint64_t)<bytes;i++){  
      f[i] = ntohll(f[i]);
    }
  }
- void inline le64toh(void *file_object,uint32_t bytes)
+ void inline le64toh_v(void *file_object,uint32_t bytes)
  {
    uint64_t *fp = (uint64_t *)file_object;
    uint64_t f,g;
@@ -387,10 +393,10 @@ inline void readNerscObject(Lattice<vobj> &Umu,std::string file,munger munge,int
       if ( grid->IsBoss() ) {
 	fin.read((char *)&file_object,sizeof(file_object));
 
-	if(ieee32big) be32toh((void *)&file_object,sizeof(file_object));
-	if(ieee32)    le32toh((void *)&file_object,sizeof(file_object));
-	if(ieee64big) be64toh((void *)&file_object,sizeof(file_object));
-	if(ieee64)    le64toh((void *)&file_object,sizeof(file_object));
+	if(ieee32big) be32toh_v((void *)&file_object,sizeof(file_object));
+	if(ieee32)    le32toh_v((void *)&file_object,sizeof(file_object));
+	if(ieee64big) be64toh_v((void *)&file_object,sizeof(file_object));
+	if(ieee64)    le64toh_v((void *)&file_object,sizeof(file_object));
 
 	munge(file_object,munged,csum);
       }

From 73c0db82d5fd50300d9ebcb3280bcb69f7cf7e21 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 23 Apr 2015 08:02:51 +0100
Subject: [PATCH 087/429] Better description of Intel's many ISA targets

---
 configure.ac | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/configure.ac b/configure.ac
index 647b5e30..b6e73bcb 100644
--- a/configure.ac
+++ b/configure.ac
@@ -27,7 +27,9 @@ AC_TYPE_UINT64_T
 # Checks for library functions.
 AC_CHECK_FUNCS([gettimeofday])
 
-AC_ARG_ENABLE([simd],[AC_HELP_STRING([--enable-simd=SSE|AVX|AVX2|AVX512],[Select instructions])],[ac_SIMD=${enable_simd}],[ac_SIMD=AVX2])
+AC_ARG_ENABLE([simd],[AC_HELP_STRING([--enable-simd=SSE|AVX|AVX2|AVX512|MIC],\
+	[Select instructions to be SSE4.0, AVX 1.0, AVX 2.0+FMA, AVX 512, MIC])],\
+	[ac_SIMD=${enable_simd}],[ac_SIMD=AVX2])
 
 case ${ac_SIMD} in
      SSE4)
@@ -42,8 +44,8 @@ case ${ac_SIMD} in
        echo Configuring for AVX2
        AC_DEFINE([AVX2],[1],[AVX2] )
      ;;
-     AVX512)
-       echo Configuring for AVX512
+     AVX512|MIC)
+       echo Configuring for AVX512 and MIC
        AC_DEFINE([AVX512],[1],[AVX512] )
      ;;
      *)

From a9e574dd27174e4ed1455d3531009bb97f1b75d6 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 23 Apr 2015 08:31:40 +0100
Subject: [PATCH 088/429] Snippets from Guido to optimise Reduce

---
 lib/simd/Grid_vComplexF.h | 39 ++++++++++++++++++++++++++++++++-------
 1 file changed, 32 insertions(+), 7 deletions(-)

diff --git a/lib/simd/Grid_vComplexF.h b/lib/simd/Grid_vComplexF.h
index 9fac656c..d931964c 100644
--- a/lib/simd/Grid_vComplexF.h
+++ b/lib/simd/Grid_vComplexF.h
@@ -213,19 +213,44 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
        friend inline ComplexF Reduce(const vComplexF & in)
        {
 #ifdef SSE4
-#error
+	   union {
+	     __m128 v1;    // SSE 4 x float vector
+	     float f[4];  // scalar array of 4 floats
+	   } u128;
+	   u128.v1= _mm_add_ps(v, _mm_shuffle_ps(v, v, 0b01001110)); // FIXME Prefer to use _MM_SHUFFLE macros
+	   return ComplexF(u128.f[0], u128.f[1]);
 #endif
-#if defined (AVX1) || defined(AVX2)
-	 // FIXME this is inefficient and use 
-         __attribute__ ((aligned(32))) float c_[8];
-         _mm256_store_ps(c_,in.v);
-         return ComplexF(c_[0]+c_[2]+c_[4]+c_[6],c_[1]+c_[3]+c_[5]+c_[7]);
-
+#ifdef AVX1
+	   //it would be better passing 2 arguments to saturate the vector lanes
+	   union {
+	     __m256 v1;    
+	     float f[8];  
+	   } u256;
+	   //SWAP lanes
+	   __m256 t0 = _mm256_permute2f128_ps(in.v, in.v, 1);
+	   __m256 t1 = _mm256_permute_ps(in.v  , 0b11011000);//real (0,2,1,3)
+	   __m256 t2 = _mm256_permute_ps(t0 , 0b10001101);//imag (1,3,0,2)
+	   t0 = _mm256_blend_ps(t1, t2, 0b0101000001010000);// (0,0,1,1,0,0,1,1)
+	   t1 = _mm256_hadd_ps(t0,t0);
+	   u256.v1 = _mm256_hadd_ps(t1, t1);
+	   return ComplexF(u256.f[0], u256.f[4]);
+#endif
+#ifdef AVX2
+	   union {
+	     __m256 v1;    
+	     float f[8];  
+	   } u256;
+	   const __m256i mask= _mm256_set_epi32( 7, 5, 3, 1, 6, 4, 2, 0);
+	   __m256 tmp1 = _mm256_permutevar8x32_ps(in.v, mask);
+	   __m256 tmp2 = _mm256_hadd_ps(tmp1, tmp1);
+	   u256.v1 = _mm256_hadd_ps(tmp2, tmp2);
+	   return ComplexF(u256.f[0], u256.f[4]);
 #endif
 #ifdef AVX512
             return ComplexF(_mm512_mask_reduce_add_ps(0x5555, in.v),_mm512_mask_reduce_add_ps(0xAAAA, in.v));
 #endif
 #ifdef QPX
+#error
 #endif
         }
 

From 2f8431ab03a4b26e4af6c18b5088c6c65d8e4aa8 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 23 Apr 2015 11:04:19 +0100
Subject: [PATCH 089/429] Consolidate index to coor in a single routine

---
 lib/cartesian/Grid_cartesian_base.h | 30 ++++++++++++++---------------
 1 file changed, 14 insertions(+), 16 deletions(-)

diff --git a/lib/cartesian/Grid_cartesian_base.h b/lib/cartesian/Grid_cartesian_base.h
index aff9375a..6ccfd175 100644
--- a/lib/cartesian/Grid_cartesian_base.h
+++ b/lib/cartesian/Grid_cartesian_base.h
@@ -90,20 +90,26 @@ public:
       for(int d=0;d<_ndimension;d++) idx+=_ostride[d]*ocoor[d];
       return idx;
     }
-    inline void CoorFromIndex (std::vector<int>& coor,int index,std::vector<int> &dims){
-      coor.resize(_ndimension);
+    static inline void CoorFromIndex (std::vector<int>& coor,int index,std::vector<int> &dims){
+      int nd= dims.size();
+      coor.resize(nd);
       for(int d=0;d<_ndimension;d++){
 	coor[d] = index % dims[d];
 	index   = index / dims[d];
       }
     }
-    inline void oCoorFromOindex (std::vector<int>& coor,int Oindex){
-      coor.resize(_ndimension);
+    static inline void IndexFromCoor (std::vector<int>& coor,int &index,std::vector<int> &dims){
+      int nd=dims.size();
+      int stride=1;
+      index=0;
       for(int d=0;d<_ndimension;d++){
-	coor[d] = Oindex % _rdimensions[d];
-	Oindex  = Oindex / _rdimensions[d];
+	index = index+stride*coor[d];
+	stride=stride*dims[d];
       }
     }
+    inline void oCoorFromOindex (std::vector<int>& coor,int Oindex){
+      CoorFromIndex(coor,Oindex,_rdimensions);
+    }
 
     //////////////////////////////////////////////////////////
     // SIMD lane addressing
@@ -116,11 +122,7 @@ public:
     }
     inline void iCoorFromIindex(std::vector<int> &coor,int lane)
     {
-      coor.resize(_ndimension);
-      for(int d=0;d<_ndimension;d++){
-	coor[d] = lane % _simd_layout[d];
-	lane    = lane / _simd_layout[d];
-      }
+      CoorFromIndex(coor,lane,_simd_layout);
     }
     inline int PermuteDim(int dimension){
       return _simd_layout[dimension]>1;
@@ -152,11 +154,7 @@ public:
     // Global addressing
     ////////////////////////////////////////////////////////////////
     void GlobalIndexToGlobalCoor(int gidx,std::vector<int> &gcoor){
-      gcoor.resize(_ndimension);
-      for(int d=0;d<_ndimension;d++){
-	gcoor[d] = gidx % _gdimensions[d];
-	gidx     = gidx / _gdimensions[d];
-      }
+      CoorFromIndex(gcoor,gidx,_gdimensions);
     }
     void GlobalCoorToGlobalIndex(const std::vector<int> & gcoor,int & gidx){
       gidx=0;

From b7416d79e3ca82299b74f20931d3c663f9b870c0 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 23 Apr 2015 11:04:59 +0100
Subject: [PATCH 090/429] Begginings of slice summation and subblocking

---
 TODO                 |  4 +--
 configure            | 46 ++++++++++++++-------------------
 lib/Grid_config.h    |  7 ++---
 lib/Grid_config.h.in |  7 ++---
 lib/Grid_stencil.h   |  6 ++---
 lib/Grid_summation.h | 61 +++++++++++++++++++++++++++++++++++++++++++-
 6 files changed, 88 insertions(+), 43 deletions(-)

diff --git a/TODO b/TODO
index a8fa5ce9..256f2609 100644
--- a/TODO
+++ b/TODO
@@ -10,7 +10,7 @@ FUNCTIONALITY:
 * How to do U[mu] ... lorentz part of type structure or not. more like chroma if not. -- DONE
 
 * subdirs lib, tests ??                              ----- DONE
-  - lib/math
+  - lib/math        
   - lib/cartesian
   - lib/cshift
   - lib/stencil
@@ -26,7 +26,7 @@ FUNCTIONALITY:
 Not done, or just incomplete
 * random number generation
 
-* Consider switch std::vector to boost arrays.
+* Consider switch std::vector to boost arrays or something lighter weight
   boost::multi_array<type, 3> A()...    to replace multi1d, multi2d etc..
 
 * How to define simple matrix operations, such as flavour matrices?
diff --git a/configure b/configure
index 48fafcf1..9e554925 100755
--- a/configure
+++ b/configure
@@ -1365,8 +1365,9 @@ Optional Features:
   --disable-dependency-tracking
                           speeds up one-time build
   --disable-openmp        do not use OpenMP
-  --enable-simd=SSE|AVX|AVX2|AVX512
-                          Select instructions
+  --enable-simd=SSE|AVX|AVX2|AVX512|MIC
+                          Select instructions to be SSE4.0, AVX 1.0, AVX
+                          2.0+FMA, AVX 512, MIC
   --enable-comms=none|mpi Select communications
 
 Some influential environment variables:
@@ -4945,6 +4946,18 @@ fi
 
 done
 
+for ac_header in endian.h
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "endian.h" "ac_cv_header_endian_h" "$ac_includes_default"
+if test "x$ac_cv_header_endian_h" = xyes; then :
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_ENDIAN_H 1
+_ACEOF
+
+fi
+
+done
+
 
 # Checks for typedefs, structures, and compiler characteristics.
 ac_fn_c_check_type "$LINENO" "size_t" "ac_cv_type_size_t" "$ac_includes_default"
@@ -4999,32 +5012,11 @@ _ACEOF
 fi
 done
 
-for ac_func in ntohll
-do :
-  ac_fn_c_check_func "$LINENO" "ntohll" "ac_cv_func_ntohll"
-if test "x$ac_cv_func_ntohll" = xyes; then :
-  cat >>confdefs.h <<_ACEOF
-#define HAVE_NTOHLL 1
-_ACEOF
-
-fi
-done
-
-for ac_func in be64toh
-do :
-  ac_fn_c_check_func "$LINENO" "be64toh" "ac_cv_func_be64toh"
-if test "x$ac_cv_func_be64toh" = xyes; then :
-  cat >>confdefs.h <<_ACEOF
-#define HAVE_BE64TOH 1
-_ACEOF
-
-fi
-done
-
 
 # Check whether --enable-simd was given.
 if test "${enable_simd+set}" = set; then :
-  enableval=$enable_simd; ac_SIMD=${enable_simd}
+  enableval=$enable_simd; \
+	ac_SIMD=${enable_simd}
 else
   ac_SIMD=AVX2
 fi
@@ -5049,8 +5041,8 @@ $as_echo "#define AVX1 1" >>confdefs.h
 $as_echo "#define AVX2 1" >>confdefs.h
 
      ;;
-     AVX512)
-       echo Configuring for AVX512
+     AVX512|MIC)
+       echo Configuring for AVX512 and MIC
 
 $as_echo "#define AVX512 1" >>confdefs.h
 
diff --git a/lib/Grid_config.h b/lib/Grid_config.h
index 2171ea6b..4152540e 100644
--- a/lib/Grid_config.h
+++ b/lib/Grid_config.h
@@ -16,8 +16,8 @@
 /* GRID_COMMS_NONE */
 /* #undef GRID_COMMS_NONE */
 
-/* Define to 1 if you have the `be64toh' function. */
-/* #undef HAVE_BE64TOH */
+/* Define to 1 if you have the <endian.h> header file. */
+/* #undef HAVE_ENDIAN_H */
 
 /* Define to 1 if you have the `gettimeofday' function. */
 #define HAVE_GETTIMEOFDAY 1
@@ -34,9 +34,6 @@
 /* Define to 1 if you have the <memory.h> header file. */
 #define HAVE_MEMORY_H 1
 
-/* Define to 1 if you have the `ntohll' function. */
-/* #undef HAVE_NTOHLL */
-
 /* Define to 1 if you have the <stdint.h> header file. */
 #define HAVE_STDINT_H 1
 
diff --git a/lib/Grid_config.h.in b/lib/Grid_config.h.in
index 2381525f..2dc0bda4 100644
--- a/lib/Grid_config.h.in
+++ b/lib/Grid_config.h.in
@@ -15,8 +15,8 @@
 /* GRID_COMMS_NONE */
 #undef GRID_COMMS_NONE
 
-/* Define to 1 if you have the `be64toh' function. */
-#undef HAVE_BE64TOH
+/* Define to 1 if you have the <endian.h> header file. */
+#undef HAVE_ENDIAN_H
 
 /* Define to 1 if you have the `gettimeofday' function. */
 #undef HAVE_GETTIMEOFDAY
@@ -33,9 +33,6 @@
 /* Define to 1 if you have the <memory.h> header file. */
 #undef HAVE_MEMORY_H
 
-/* Define to 1 if you have the `ntohll' function. */
-#undef HAVE_NTOHLL
-
 /* Define to 1 if you have the <stdint.h> header file. */
 #undef HAVE_STDINT_H
 
diff --git a/lib/Grid_stencil.h b/lib/Grid_stencil.h
index d8debac0..7bafb879 100644
--- a/lib/Grid_stencil.h
+++ b/lib/Grid_stencil.h
@@ -283,7 +283,7 @@ namespace Grid {
 	    if ( comm_any ) {
 
 	      for(int i=0;i<Nsimd;i++){
-		pointers[i] = (scalar_type *)&send_buf_extract[i][0];
+		pointers[Nsimd-1-i] = (scalar_type *)&send_buf_extract[i][0];
 	      }
 	      Gather_plane_extract(rhs,pointers,dimension,sx,cbmask);
 	      
@@ -332,9 +332,9 @@ namespace Grid {
 	      for(int i=0;i<Nsimd;i++){
 		if ( permute_slice ) {
 		  PermuteMap=i^toggle_bit;
-		  pointers[i] = rpointers[PermuteMap];
+		  pointers[Nsimd-1-i] = rpointers[PermuteMap];
 		} else {
-		  pointers[i] = rpointers[i];
+		  pointers[Nsimd-1-i] = rpointers[i];
 		}
 	      }
 
diff --git a/lib/Grid_summation.h b/lib/Grid_summation.h
index dff2df61..366cef06 100644
--- a/lib/Grid_summation.h
+++ b/lib/Grid_summation.h
@@ -1,12 +1,71 @@
 #ifndef GRID_SUMMATION_H
 #define GRID_SUMMATION_H
 
+void subdivides(GridBase *coarse,GridBase *fine)
+{
+  assert(coarse->_ndimension == fine->_ndimension);
+
+  int _ndimension = coarse->_ndimension;
+
+  // local and global volumes subdivide cleanly after SIMDization
+  for(int d=0;d<_ndimension;d++){
+    assert((fine->_rdimensions[d] / coarse->_rdimensions[d])* coarse->_rdimensions[d]==fine->_rdimensions[d]); 
+    assert((fine->_ldimensions[d] / coarse->_ldimensions[d])* coarse->_ldimensions[d]==fine->_ldimensions[d]); 
+    assert((fine->_gdimensions[d] / coarse->_gdimensions[d])* coarse->_gdimensions[d]==fine->_gdimensions[d]); 
+    assert((fine->_fdimensions[d] / coarse->_fdimensions[d])* coarse->_fdimensions[d]==fine->_fdimensions[d]); 
+  }
+}
+// Generic name : Coarsen?
+//              : SubMeshSum?
+//
 template<class vobj>
-inline void sumBlocks(Lattice<vobj> &coarseData,const Lattice<vobj &fineData)
+inline void sumBlocks(Lattice<vobj> &coarseData,const Lattice<vobj> &fineData)
 {
   GridBase * fine  = findData._grid;
   GridBase * coarse= findData._grid;
+
+  subdivides(coars,fine); // require they map
+
+  int _ndimension = coarse->_ndimension;
   
+  std::vector<bool> replicated(_ndimension,false);
+  std::vector<int>  block_r   (_dimension);
+  std::vector<int>  block_f   (_dimension);
+
+  ///////////////////////////////////////////////////////////
+  // Detect whether the result is replicated in dimension d
+  ///////////////////////////////////////////////////////////
+  for(int d=0 ; d<_ndimension;d++){
+    if ( (_fdimensions[d] == 1) && (coarse->_processors[d]>1) ) {
+      replicated[d]=true;
+    }
+    block_r[d] = fine->_rdimensions[d] / coarse->_rdimensions[d];
+    block_l[d] = fine->_ldimensions[d] / coarse->_ldimensions[d];
+    block_f[d] = fine->_fdimensions[d] / coarse->_fdimensions[d];
+  }
+
+  coaseData=zero;
+
+  //FIXME Bagel's strategy: loop over fine sites
+  // identify corresponding coarse site, but coarse sites are 
+  // divided across threads. Not so easy to do in openmp but
+  // there must be a way
+  for(int sf=0;sf<fine->oSites();sf++){
+    
+    int sc;
+    vobj sum=zero;
+    std::vector<int> coor_c(_ndimension);
+    std::vector<int> coor_f(_ndimension);
+
+    GridBase::CoorFromIndex(coor_f,sf,fine->_rdimensions);
+
+    for(int d=0;d<_ndimension;d++) coor_c[d]=coor_f[d]/fine->_rdimensions;
+
+    GridBase::IndexFromCoor(coor_c,sc,coarse->_rdimensions);
+
+    coarseData._odata[sc]=coarseData._odata[sc]+fineData._odata[sf];
+
+  }
   return;
 }
 #endif

From 62e8d2d12730c3b642c85dfd80e4a08eabafa82e Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 23 Apr 2015 15:13:00 +0100
Subject: [PATCH 091/429] Slice summation working. May move this into
 lattice/Grid_lattice_reduction however

---
 lib/Grid_lattice.h                   |   1 +
 lib/Grid_simd.h                      |  36 ++++-----
 lib/Grid_summation.h                 | 113 ++++++++++++++++++++-------
 lib/cartesian/Grid_cartesian_base.h  |  41 +++++-----
 lib/lattice/Grid_lattice_reduction.h |   5 +-
 lib/math/Grid_math_tensors.h         |  32 ++++++++
 tests/Grid_nersc_io.cc               |  25 +++++-
 7 files changed, 181 insertions(+), 72 deletions(-)

diff --git a/lib/Grid_lattice.h b/lib/Grid_lattice.h
index 7256459c..ed875bc0 100644
--- a/lib/Grid_lattice.h
+++ b/lib/Grid_lattice.h
@@ -98,6 +98,7 @@ public:
 #include <lattice/Grid_lattice_coordinate.h>
 #include <lattice/Grid_lattice_rng.h>
 #include <lattice/Grid_lattice_transfer.h>
+#include <Grid_summation.h>
 
 
 
diff --git a/lib/Grid_simd.h b/lib/Grid_simd.h
index 9bf10fc9..17755fe0 100644
--- a/lib/Grid_simd.h
+++ b/lib/Grid_simd.h
@@ -126,16 +126,16 @@ namespace Grid {
 /////////////////////////////////////////////////////////////////
 // Generic extract/merge/permute
 /////////////////////////////////////////////////////////////////
+
 template<class vsimd,class scalar>
 inline void Gextract(const vsimd &y,std::vector<scalar *> &extracted){
-  // FIXME: bounce off stack is painful
-  // temporary hack while I figure out better way.
-  // There are intrinsics to do this work without the storage.
+  // FIXME: bounce off memory is painful
   int Nextr=extracted.size();
   int Nsimd=vsimd::Nsimd();
   int s=Nsimd/Nextr;
 
   std::vector<scalar,alignedAllocator<scalar> > buf(Nsimd); 
+
   vstore(y,&buf[0]);
   for(int i=0;i<Nextr;i++){
     *extracted[i] = buf[i*s];
@@ -143,6 +143,19 @@ inline void Gextract(const vsimd &y,std::vector<scalar *> &extracted){
   }
 };
 template<class vsimd,class scalar>
+inline void Gextract(const vsimd &y,std::vector<scalar> &extracted){
+  int Nextr=extracted.size();
+  int Nsimd=vsimd::Nsimd();
+  int s=Nsimd/Nextr;
+
+  std::vector<scalar,alignedAllocator<scalar> > buf(Nsimd); 
+
+  vstore(y,&buf[0]);
+  for(int i=0;i<Nextr;i++){
+    extracted[i] = buf[i*s];
+  }
+};
+template<class vsimd,class scalar>
 inline void Gmerge(vsimd &y,std::vector<scalar *> &extracted){
   int Nextr=extracted.size();
   int Nsimd=vsimd::Nsimd();
@@ -158,23 +171,6 @@ inline void Gmerge(vsimd &y,std::vector<scalar *> &extracted){
   vset(y,&buf[0]); 
 };
 template<class vsimd,class scalar>
-inline void Gextract(const vsimd &y,std::vector<scalar> &extracted){
-  // FIXME: bounce off stack is painful
-  // temporary hack while I figure out better way.
-  // There are intrinsics to do this work without the storage.
-  int Nextr=extracted.size();
-  int Nsimd=vsimd::Nsimd();
-  int s=Nsimd/Nextr;
-
-  std::vector<scalar,alignedAllocator<scalar> > buf(Nsimd); 
-
-  vstore(y,&buf[0]);
-
-  for(int i=0;i<Nextr;i++){
-    extracted[i] = buf[i*s];
-  }
-};
-template<class vsimd,class scalar>
 inline void Gmerge(vsimd &y,std::vector<scalar> &extracted){
   int Nextr=extracted.size();
   int Nsimd=vsimd::Nsimd();
diff --git a/lib/Grid_summation.h b/lib/Grid_summation.h
index 366cef06..12eeb2bc 100644
--- a/lib/Grid_summation.h
+++ b/lib/Grid_summation.h
@@ -1,7 +1,8 @@
 #ifndef GRID_SUMMATION_H
 #define GRID_SUMMATION_H
+namespace Grid {
 
-void subdivides(GridBase *coarse,GridBase *fine)
+inline void subdivides(GridBase *coarse,GridBase *fine)
 {
   assert(coarse->_ndimension == fine->_ndimension);
 
@@ -9,58 +10,43 @@ void subdivides(GridBase *coarse,GridBase *fine)
 
   // local and global volumes subdivide cleanly after SIMDization
   for(int d=0;d<_ndimension;d++){
+    assert(coarse->_processors[d]  == fine->_processors[d]);
+    assert(coarse->_simd_layout[d] == fine->_simd_layout[d]);
     assert((fine->_rdimensions[d] / coarse->_rdimensions[d])* coarse->_rdimensions[d]==fine->_rdimensions[d]); 
-    assert((fine->_ldimensions[d] / coarse->_ldimensions[d])* coarse->_ldimensions[d]==fine->_ldimensions[d]); 
-    assert((fine->_gdimensions[d] / coarse->_gdimensions[d])* coarse->_gdimensions[d]==fine->_gdimensions[d]); 
-    assert((fine->_fdimensions[d] / coarse->_fdimensions[d])* coarse->_fdimensions[d]==fine->_fdimensions[d]); 
   }
 }
+
+// useful in multigrid project;
 // Generic name : Coarsen?
-//              : SubMeshSum?
-//
 template<class vobj>
 inline void sumBlocks(Lattice<vobj> &coarseData,const Lattice<vobj> &fineData)
 {
-  GridBase * fine  = findData._grid;
-  GridBase * coarse= findData._grid;
+  GridBase * fine  = fineData._grid;
+  GridBase * coarse= coarseData._grid;
 
-  subdivides(coars,fine); // require they map
+  subdivides(coarse,fine); // require they map
 
   int _ndimension = coarse->_ndimension;
   
-  std::vector<bool> replicated(_ndimension,false);
-  std::vector<int>  block_r   (_dimension);
-  std::vector<int>  block_f   (_dimension);
-
+  std::vector<int>  block_r      (_ndimension);
+  
   ///////////////////////////////////////////////////////////
   // Detect whether the result is replicated in dimension d
   ///////////////////////////////////////////////////////////
+  
   for(int d=0 ; d<_ndimension;d++){
-    if ( (_fdimensions[d] == 1) && (coarse->_processors[d]>1) ) {
-      replicated[d]=true;
-    }
     block_r[d] = fine->_rdimensions[d] / coarse->_rdimensions[d];
-    block_l[d] = fine->_ldimensions[d] / coarse->_ldimensions[d];
-    block_f[d] = fine->_fdimensions[d] / coarse->_fdimensions[d];
   }
 
-  coaseData=zero;
-
-  //FIXME Bagel's strategy: loop over fine sites
-  // identify corresponding coarse site, but coarse sites are 
-  // divided across threads. Not so easy to do in openmp but
-  // there must be a way
+  coarseData=zero;
   for(int sf=0;sf<fine->oSites();sf++){
     
     int sc;
-    vobj sum=zero;
     std::vector<int> coor_c(_ndimension);
     std::vector<int> coor_f(_ndimension);
 
     GridBase::CoorFromIndex(coor_f,sf,fine->_rdimensions);
-
-    for(int d=0;d<_ndimension;d++) coor_c[d]=coor_f[d]/fine->_rdimensions;
-
+    for(int d=0;d<_ndimension;d++) coor_c[d]=coor_f[d]/block_r[d];
     GridBase::IndexFromCoor(coor_c,sc,coarse->_rdimensions);
 
     coarseData._odata[sc]=coarseData._odata[sc]+fineData._odata[sf];
@@ -68,4 +54,75 @@ inline void sumBlocks(Lattice<vobj> &coarseData,const Lattice<vobj> &fineData)
   }
   return;
 }
+
+template<class vobj> inline void sliceSum(const Lattice<vobj> &Data,std::vector<typename vobj::scalar_object> &result,int orthogdim)
+{
+  typedef typename vobj::scalar_object sobj;
+
+  GridBase  *grid = Data._grid;
+  const int    Nd = grid->_ndimension;
+  const int Nsimd = grid->Nsimd();
+
+  assert(orthogdim >= 0);
+  assert(orthogdim < Nd);
+
+  int fd=grid->_fdimensions[orthogdim];
+  int ld=grid->_ldimensions[orthogdim];
+  int rd=grid->_rdimensions[orthogdim];
+
+  std::vector<vobj,alignedAllocator<vobj> > lvSum(rd); // will locally sum vectors first
+  std::vector<sobj> lsSum(ld,sobj(zero)); // sum across these down to scalars
+  std::vector<sobj> extracted(Nsimd);     // splitting the SIMD
+
+  result.resize(fd); // And then global sum to return the same vector to every node for IO to file
+  for(int r=0;r<rd;r++){
+    lvSum[r]=zero;
+  }
+
+  std::vector<int>  coor(Nd);  
+
+  // sum over reduced dimension planes, breaking out orthog dir
+  for(int ss=0;ss<grid->oSites();ss++){
+    GridBase::CoorFromIndex(coor,ss,grid->_rdimensions);
+    int r = coor[orthogdim];
+    lvSum[r]=lvSum[r]+Data._odata[ss];
+  }  
+
+  // Sum across simd lanes in the plane, breaking out orthog dir.
+  std::vector<int> icoor(Nd);
+
+  for(int rt=0;rt<rd;rt++){
+
+    extract(lvSum[rt],extracted);
+
+    for(int idx=0;idx<Nsimd;idx++){
+
+      grid->iCoorFromIindex(icoor,idx);
+
+      int ldx =rt+icoor[orthogdim]*rd;
+
+      lsSum[ldx]=lsSum[ldx]+extracted[idx];
+
+    }
+  }
+  
+  // sum over nodes.
+  sobj gsum;
+  for(int t=0;t<fd;t++){
+    int pt = t/ld; // processor plane
+    int lt = t%ld;
+    if ( pt == grid->_processor_coor[orthogdim] ) {
+      gsum=lsSum[lt];
+    } else {
+      gsum=zero;
+    }
+
+    grid->GlobalSum(gsum);
+
+    result[t]=gsum;
+  }
+
+}
+
+}
 #endif
diff --git a/lib/cartesian/Grid_cartesian_base.h b/lib/cartesian/Grid_cartesian_base.h
index 6ccfd175..6fd55399 100644
--- a/lib/cartesian/Grid_cartesian_base.h
+++ b/lib/cartesian/Grid_cartesian_base.h
@@ -93,7 +93,7 @@ public:
     static inline void CoorFromIndex (std::vector<int>& coor,int index,std::vector<int> &dims){
       int nd= dims.size();
       coor.resize(nd);
-      for(int d=0;d<_ndimension;d++){
+      for(int d=0;d<nd;d++){
 	coor[d] = index % dims[d];
 	index   = index / dims[d];
       }
@@ -102,7 +102,7 @@ public:
       int nd=dims.size();
       int stride=1;
       index=0;
-      for(int d=0;d<_ndimension;d++){
+      for(int d=0;d<nd;d++){
 	index = index+stride*coor[d];
 	stride=stride*dims[d];
       }
@@ -164,6 +164,25 @@ public:
 	mult*=_gdimensions[mu];
       }
     }
+    void GlobalCoorToProcessorCoorLocalCoor(std::vector<int> &pcoor,std::vector<int> &lcoor,const std::vector<int> &gcoor)
+    {
+      pcoor.resize(_ndimension);
+      lcoor.resize(_ndimension);
+      for(int mu=0;mu<_ndimension;mu++){
+	pcoor[mu] = gcoor[mu]/_ldimensions[mu];
+	lcoor[mu] = gcoor[mu]%_ldimensions[mu];
+      }
+    }
+    void GlobalCoorToRankIndex(int &rank, int &o_idx, int &i_idx ,const std::vector<int> &gcoor)
+    {
+      std::vector<int> pcoor;
+      std::vector<int> lcoor;
+      GlobalCoorToProcessorCoorLocalCoor(pcoor,lcoor,gcoor);
+      rank = RankFromProcessorCoor(pcoor);
+      i_idx= iIndex(lcoor);
+      o_idx= oIndex(lcoor);
+    }
+
     void RankIndexToGlobalCoor(int rank, int o_idx, int i_idx , std::vector<int> &gcoor)
     {
       gcoor.resize(_ndimension);
@@ -191,24 +210,6 @@ public:
       gcoor.resize(_ndimension);
       for(int mu=0;mu<_ndimension;mu++) gcoor[mu] = Pcoor[mu]*_ldimensions[mu]+Lcoor[mu];
     }
-    void GlobalCoorToProcessorCoorLocalCoor(std::vector<int> &pcoor,std::vector<int> &lcoor,const std::vector<int> &gcoor)
-    {
-      pcoor.resize(_ndimension);
-      lcoor.resize(_ndimension);
-      for(int mu=0;mu<_ndimension;mu++){
-	pcoor[mu] = gcoor[mu]/_ldimensions[mu];
-	lcoor[mu] = gcoor[mu]%_ldimensions[mu];
-      }
-    }
-    void GlobalCoorToRankIndex(int &rank, int &o_idx, int &i_idx ,const std::vector<int> &gcoor)
-    {
-      std::vector<int> pcoor;
-      std::vector<int> lcoor;
-      GlobalCoorToProcessorCoorLocalCoor(pcoor,lcoor,gcoor);
-      rank = RankFromProcessorCoor(pcoor);
-      i_idx= iIndex(lcoor);
-      o_idx= oIndex(lcoor);
-    }
 };
 
 
diff --git a/lib/lattice/Grid_lattice_reduction.h b/lib/lattice/Grid_lattice_reduction.h
index 7d34d2ea..b96d0f05 100644
--- a/lib/lattice/Grid_lattice_reduction.h
+++ b/lib/lattice/Grid_lattice_reduction.h
@@ -25,7 +25,9 @@ namespace Grid {
     }
 
     template<class vobj>
-    inline auto innerProduct(const Lattice<vobj> &left,const Lattice<vobj> &right) ->decltype(innerProduct(left._odata[0],right._odata[0]))
+    inline ComplexD innerProduct(const Lattice<vobj> &left,const Lattice<vobj> &right) 
+      //    inline auto innerProduct(const Lattice<vobj> &left,const Lattice<vobj> &right) 
+      //->decltype(innerProduct(left._odata[0],right._odata[0]))
     {
       typedef typename vobj::scalar_type scalar;
       decltype(innerProduct(left._odata[0],right._odata[0])) vnrm=zero;
@@ -54,6 +56,7 @@ namespace Grid {
 
       vsum=zero;
       ssum=zero;
+      //FIXME make this loop parallelisable
       for(int ss=0;ss<arg._grid->oSites(); ss++){
 	vsum = vsum + arg._odata[ss];
       }
diff --git a/lib/math/Grid_math_tensors.h b/lib/math/Grid_math_tensors.h
index 0ee30970..379e61d2 100644
--- a/lib/math/Grid_math_tensors.h
+++ b/lib/math/Grid_math_tensors.h
@@ -241,5 +241,37 @@ public:
 
 };
 
+template<class vobj> inline 
+void extract(const vobj &vec,std::vector<typename vobj::scalar_object> &extracted)
+{
+  typedef typename vobj::scalar_type scalar_type ;
+  typedef typename vobj::vector_type vector_type ;
+
+  int Nsimd=vobj::vector_type::Nsimd();
+  
+  extracted.resize(Nsimd);
+
+  std::vector<scalar_type *> pointers(Nsimd);
+  for(int i=0;i<Nsimd;i++) 
+    pointers[i] =(scalar_type *)& extracted[i];
+  
+  extract(vec,pointers);
+}
+template<class vobj> inline 
+void merge(vobj &vec,std::vector<typename vobj::scalar_object> &extracted)
+{
+  typedef typename vobj::scalar_type scalar_type ;
+  typedef typename vobj::vector_type vector_type ;
+  
+  int Nsimd=vobj::vector_type::Nsimd();
+  assert(extracted.size()==Nsimd);
+
+  std::vector<scalar_type *> pointers(Nsimd);
+  for(int i=0;i<Nsimd;i++) 
+    pointers[i] =(scalar_type *)& extracted[i];
+  
+  merge(vec,pointers);
+}
+
 }
 #endif
diff --git a/tests/Grid_nersc_io.cc b/tests/Grid_nersc_io.cc
index 1d965404..35cad831 100644
--- a/tests/Grid_nersc_io.cc
+++ b/tests/Grid_nersc_io.cc
@@ -11,8 +11,10 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> simd_layout({1,1,2,2});
-  std::vector<int> mpi_layout ({2,1,1,2});
+  std::vector<int> mpi_layout ({1,1,1,1});
   std::vector<int> latt_size  ({16,16,16,32});
+  int orthodir=3;
+  int orthosz =latt_size[orthodir];
     
   GridCartesian     Fine(latt_size,simd_layout,mpi_layout);
   GridRNG           FineRNG(&Fine);
@@ -45,14 +47,31 @@ int main (int argc, char ** argv)
     }
   }
 
-
   double vol = Fine.gSites();
-
   Complex PlaqScale(1.0/vol/6.0/3.0);
+
+  std::vector<TComplex> Plaq_T(orthosz);
+  sliceSum(Plaq,Plaq_T,Nd-1);
+  int Nt = Plaq_T.size();
+
+  TComplex Plaq_T_sum=zero;
+  for(int t=0;t<Nt;t++){
+    Plaq_T_sum = Plaq_T_sum+Plaq_T[t];
+    Complex Pt=TensorRemove(Plaq_T[t]);
+    std::cout << "sliced ["<<t<<"]" <<Pt*PlaqScale*Real(Nt)<<std::endl;
+  }
+
+  {
+    Complex Pt = TensorRemove(Plaq_T_sum);
+    std::cout << "total " <<Pt*PlaqScale<<std::endl;
+  }  
+
+
   TComplex Tp = sum(Plaq);
   Complex p  = TensorRemove(Tp);
   std::cout << "calculated plaquettes " <<p*PlaqScale<<std::endl;
 
+
   Complex LinkTraceScale(1.0/vol/4.0/3.0);
   TComplex Tl = sum(LinkTrace);
   Complex l  = TensorRemove(Tl);

From afe6c4f64f3003526c955c47c1a35037e23c88f6 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 23 Apr 2015 20:41:22 +0100
Subject: [PATCH 092/429] move

---
 lib/qcd/{Grid_QCD.h => Grid_qcd.h} | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 rename lib/qcd/{Grid_QCD.h => Grid_qcd.h} (100%)

diff --git a/lib/qcd/Grid_QCD.h b/lib/qcd/Grid_qcd.h
similarity index 100%
rename from lib/qcd/Grid_QCD.h
rename to lib/qcd/Grid_qcd.h

From 4b4dcc4c1369ea4d5a25db4fc7947fda97a70114 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 23 Apr 2015 20:42:09 +0100
Subject: [PATCH 093/429] Rename Grid_QCD

---
 lib/Grid.h | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/lib/Grid.h b/lib/Grid.h
index 7e12490a..2a666c3b 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -50,16 +50,21 @@
 #include <Grid_cshift.h>
 #include <Grid_where.h>
 #include <Grid_stencil.h>
-#include <qcd/Grid_QCD.h>
+#include <qcd/Grid_qcd.h>
 #include <parallelIO/GridNerscIO.h>
 
 namespace Grid {
 
   void Grid_init(int *argc,char ***argv);
   void Grid_finalize(void);
-  double usecond(void);
   void Grid_sa_signal_handler(int sig,siginfo_t *si,void * ptr);
   void Grid_debug_handler_init(void);
+  void Grid_quiesce_nodes(void);
+  void Grid_unquiesce_nodes(void);
+
+  // C++11 time facilities better?
+  double usecond(void);
+
 
 };
 

From e2e3ea57428cd2cf9190e392d914085bcf597c48 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 23 Apr 2015 20:42:30 +0100
Subject: [PATCH 094/429] Reorganised the TODO. Really getting somewhere

---
 TODO                   | 298 ++++++++++++-----------------------------
 lib/Grid_init.cc       |  61 +++++++--
 lib/Grid_summation.h   |  94 ++++++++++++-
 tests/Grid_nersc_io.cc |  13 +-
 4 files changed, 239 insertions(+), 227 deletions(-)

diff --git a/TODO b/TODO
index 256f2609..12e33302 100644
--- a/TODO
+++ b/TODO
@@ -1,3 +1,77 @@
+* - BinaryWriter, TextWriter etc...
+  - protocol buffers? replace xml
+
+* Stencil operator support                           -----Initial thoughts, trial implementation DONE.
+                                                     -----some simple tests that Stencil matches Cshift.
+                                                     -----do all permute in comms phase, so that copy permute
+						     -----cases move into a buffer.
+						     -----allow transform in/out buffers spproj
+
+
+* CovariantShift support                             -----Use a class to store gauge field? (parallel transport?)
+
+* Consider switch std::vector to boost arrays or something lighter weight
+  boost::multi_array<type, 3> A()...    to replace multi1d, multi2d etc..
+
+* How to define simple matrix operations, such as flavour matrices?
+
+* Make the Tensor types and Complex etc... play more nicely.
+
+* Dirac, Pauli, SU subgroup, etc.. * Gamma/Dirac structures
+
+* Fourspin, two spin project
+
+* su3 exponentiation & log etc.. [Jamie's code?]
+  TaProj
+
+* Parallel MPI2 IO
+
+* rb4d support.
+
+* Check for missing functionality                    - partially audited against QDP++ layout
+
+* Optimise the extract/merge SIMD routines; Azusa??
+ - I have collated into single location at least.
+ - Need to use _mm_*insert/extract routines.
+
+* Conformable test in Cshift routines.
+
+* QDP++ regression suite and comparative benchmark
+
+AUDITS:
+
+* FIXME audit
+* const audit
+* Replace vset with a call to merge.; 
+* care in Gmerge,Gextract over vset .
+* extract / merge extra implementation removal      
+* Test infrastructure
+
+   // TODO
+   //
+   // Base class to share common code between vRealF, VComplexF etc...
+  //
+   // Unary functions
+   // cos,sin, tan, acos, asin, cosh, acosh, tanh, sinh, // Scalar<vReal> only arg
+   // exp, log, sqrt, fabs
+   //
+  // transposeColor, transposeSpin,
+   // adjColor, adjSpin,
+   //
+  // copyMask.
+   //
+   // localMaxAbs
+   //
+   // Fourier transform equivalent.
+
+* LinearOperator
+
+  LinearSolver
+
+  Polynomial etc...
+
+======================================================================================================
+
 FUNCTIONALITY:
 * Conditional execution, where etc...                -----DONE, simple test
 * Integer relational support                         -----DONE
@@ -22,222 +96,24 @@ FUNCTIONALITY:
   - lib/qcd/actions
   - lib/qcd/measurements
 
+* Subset support, slice sums etc...                  -----DONE
+  sliceSum(orthog)
+  sum
+  innerProduct
+  norm2
 
-Not done, or just incomplete
-* random number generation
+* Subgrid Transferral                                -----DONE
+  subBlock (coarseLattice,fineLattice)
+  projectBlockBasis  
+  promoteBlockBasis
 
-* Consider switch std::vector to boost arrays or something lighter weight
-  boost::multi_array<type, 3> A()...    to replace multi1d, multi2d etc..
+* random number generation                           ----- DONE
 
-* How to define simple matrix operations, such as flavour matrices?
-
-* Dirac, Pauli, SU subgroup, etc.. * Gamma/Dirac structures
-
-* Fourspin, two spin project
-
-* su3 exponentiation, log etc.. [Jamie's code?]
-
-* Stencil operator support                           -----Initial thoughts, trial implementation DONE.
-                                                     -----some simple tests that Stencil matches Cshift.
-                                                     -----do all permute in comms phase, so that copy permute
-						     -----cases move into a buffer.
-						     -----allow transform in/out buffers spproj
-
-* CovariantShift support                             -----Use a class to store gauge field? (parallel transport?)
-
-* Subset support, slice sums etc...                  -----Only need slice sum?
-                                                     -----Generic cartesian subslicing?
-                                                     -----Array ranges / boost extents?
-                                                     -----Multigrid grid transferral?
-                                                     -----Suggests generalised cartesian subblocking
-                                                          sums, returning modified grid?
-					             -----What should interface be?
-
-* Grid transferral
-  * pickCheckerboard, pickSubPlane, pickSubBlock,
-  *                    sumSubPlane, sumSubBlocks
-
-* rb4d support.
-
-* Check for missing functionality                    - partially audited against QDP++ layout
-
-* Optimise the extract/merge SIMD routines; Azusa??
-
- - I have collated into single location at least.
- - Need to use _mm_*insert/extract routines.
-
-* Conformable test in Cshift routines.
-
-
-
-* Broadcast, reduction tests. innerProduct, localInnerProduct
-
-* QDP++ regression suite and comparative benchmark
+* Broadcast, reduction tests. innerProduct, localInnerProduct --- DONE
 
 * I/O support
-
-* NERSC Lattice loading, plaquette test
-
-  - MPI IO?
-  - BinaryWriter, TextWriter etc...
-  - protocol buffers?
-
-AUDITS:
-// Lattice support audit                 Tested in Grid_main.cc
-//
-//     -=,+=,*=                           Y
-//     add,+,sub,-,mult,mac,*             Y
-//     innerProduct,norm2                 Y
-//     localInnerProduct,outerProduct,    Y
-//     adj,conj                           Y
-//     transpose,                         Y
-//     trace                              Y
-//
-//     transposeIndex                     Y
-//     traceIndex                         Y
-//     peekIndex                          Y
-//
-//     real,imag                          missing, semantic thought needed on real/im support.
-//                                        perhaps I just keep everything complex?
-// 
-
-* FIXME audit
-* const audit
-* Replace vset with a call to merge.; 
-* care in Gmerge,Gextract over vset .
-* extract / merge extra implementation removal      
-* Test infrastructure
-
-[ More on subsets and grid transfers ]
-i)  Three classes of subset;   red black parity subsetting (pick checkerboard).
-                             cartesian sub-block subsetting
-                             rbNd 
-
-ii) Need to be able to project one Grid to another Grid.
-
-Lattice<vobj> coarse_data SubBlockSum (GridBase *CoarseGrid, Lattice<vobj> &fine_data)
-
-Operation ensure either:
- rd[dim] divide rd[dim] fine_data
-
-This will give a distributed array over mpi ranks in a given dim IF coarse gd != 1 and _processors[d]>1
-Dimension can be *replicated* on all ranks in dimension. Need a "replicated" option on GridCartesian etc..
-
-This will give "slice" summation and fourier projection assistance.
-
-    Generic concept is to subdivide (based on RD so applies to red/black or full).
-    Return a type on SUB-grid from CellSum TOP-grid
-    SUB-grid need not distribute but be replicated in some dims if that is how the
-    cartesian communicator works.
-
-Instead of subsetting 
-
-iii) No general permutation map.
+* NERSC Lattice loading, plaquette test             ------- DONE single node 
 
 
- ? Cell definition <-> sliceSum.
- ? Replicated arrays.
+* Controling std::cout                              ------- DONE
 
-
-
-
-// Cartesian grid inheritance
-//            Grid::GridBase
-//                     |
-//           __________|___________
-//          |                      |
-// Grid::GridCartesian   Grid::GridCartesianRedBlack
-//
-// TODO: document the following as an API guaranteed public interface
-
-    /* 
-     *       Rough map of functionality against QDP++ Layout
-     *
-     *       Param     |     Grid                     |     QDP++             
-     *       -----------------------------------------
-     *                 |                              |
-     *        void     |     oSites, iSites, lSites   |  sitesOnNode 
-     *        void     |     gSites                   |  vol
-     *                 |                              |
-     *        gcoor    |     oIndex, iIndex           |  linearSiteIndex // no virtual node in QDP
-     *        lcoor    |                              |
-     * 
-     *        void     |     CheckerBoarded           |  -        // No checkerboarded in QDP
-     *        void     |     FullDimensions           |  lattSize
-     *        void     |     GlobalDimensions         |  lattSize // No checkerboarded in QDP
-     *        void     |     LocalDimensions          |  subgridLattSize
-     *        void     |     VirtualLocalDimensions   |  subgridLattSize // no virtual node in QDP
-     *                 |                              |
-     *       int x 3   |     oiSiteRankToGlobal       |  siteCoords
-     *                 |     ProcessorCoorLocalCoorToGlobalCoor | 
-     *                 |                              |
-     *     vector<int> |     GlobalCoorToRankIndex   |  nodeNumber(coord)
-     *     vector<int> |     GlobalCoorToProcessorCoorLocalCoor|  nodeCoord(coord)
-     *                 |                              |
-     *     void        |     Processors               |  logicalSize    // returns cart array shape
-     *     void        |     ThisRank        |  nodeNumber();  // returns this node rank
-     *     void        |     ThisProcessorCoor        |    // returns this node coor
-     *     void        |     isBoss(void)             |  primaryNode();
-     *                 |                              |
-     *                 |     RankFromProcessorCoor    |  getLogicalCoorFrom(node)
-     *                 |     ProcessorCoorFromRank    |  getNodeNumberFrom(logical_coord)
-     */
-  // Work out whether to permute 
-  // ABCDEFGH ->   AE BF CG DH       permute              wrap num
-  //
-  // Shift 0       AE BF CG DH       0 0 0 0    ABCDEFGH   0   0
-  // Shift 1       BF CG DH AE       0 0 0 1    BCDEFGHA   0   1
-  // Shift 2       CG DH AE BF       0 0 1 1    CDEFGHAB   0   2
-  // Shift 3       DH AE BF CG       0 1 1 1    DEFGHABC   0   3
-  // Shift 4       AE BF CG DH       1 1 1 1    EFGHABCD   1   0 
-  // Shift 5       BF CG DH AE       1 1 1 0    FGHACBDE   1   1 
-  // Shift 6       CG DH AE BF       1 1 0 0    GHABCDEF   1   2
-  // Shift 7       DH AE BF CG       1 0 0 0    HABCDEFG   1   3
-
-  // Suppose 4way simd in one dim.
-  // ABCDEFGH ->   AECG BFDH      permute              wrap num
-
-  // Shift 0       AECG BFDH      0,00 0,00 ABCDEFGH         0     0
-  // Shift 1       BFDH CGEA      0,00 1,01 BCDEFGHA         0     1
-  // Shift 2       CGEA DHFB      1,01 1,01 CDEFGHAB         1     0
-  // Shift 3       DHFB EAGC      1,01 1,11 DEFGHABC         1     1
-  // Shift 4       EAGC FBHD      1,11 1,11 EFGHABCD         2     0 
-  // Shift 5       FBHD GCAE      1,11 1,10 FGHABCDE         2     1
-  // Shift 6       GCAE HDBF      1,10 1,10 GHABCDEF         3     0
-  // Shift 7       HDBF AECG      1,10 0,00 HABCDEFG         3     1
-
-  // Generalisation to 8 way simd, 16 way simd required.
-  //
-  // Need log2 Nway masks. consisting of 
-  //	    1 bit  256 bit granule
-  //	    2 bit  128 bit granule
-  //        4 bits 64  bit granule
-  //        8 bits 32  bit granules
-  //
-  //        15 bits....
-    // TODO
-    //
-    // Base class to share common code between vRealF, VComplexF etc...
-    //
-    // lattice Broad cast assignment
-    //
-    // where() support
-    // implement with masks, and/or? Type of the mask & boolean support?
-    //
-    // Unary functions
-    // cos,sin, tan, acos, asin, cosh, acosh, tanh, sinh, // Scalar<vReal> only arg
-    // exp, log, sqrt, fabs
-    //
-    // transposeColor, transposeSpin,
-    // adjColor, adjSpin,
-    // traceColor, traceSpin.
-    // peekColor, peekSpin + pokeColor PokeSpin
-    //
-    // copyMask.
-    //
-    // localMaxAbs
-    //
-    // norm2,
-    // sumMulti equivalent.
-    // Fourier transform equivalent.
-    //
diff --git a/lib/Grid_init.cc b/lib/Grid_init.cc
index ee11a982..762f7568 100644
--- a/lib/Grid_init.cc
+++ b/lib/Grid_init.cc
@@ -10,18 +10,57 @@
 #include <sys/stat.h> 
 #include <sys/time.h>
 #include <signal.h>
-
+#include <iostream>
 #include <Grid.h>
 
 #undef __X86_64
+#define MAC
+
+#ifdef MAC
+#include <execinfo.h>
+#endif
+
 namespace Grid {
 
+  std::streambuf *Grid_saved_stream_buf;
+#if 0
+  void Grid_quiesce_nodes(void)
+  {
+#ifdef GRID_COMMS_MPI
+    int me;
+    MPI_Comm_rank(MPI_COMM_WORLD,&me);
+    std::streambuf* Grid_saved_stream_buf = std::cout.rdbuf();
+    if ( me ) { 
+      std::ofstream file("log.node");
+      std::cout.rdbuf(file.rdbuf());
+    }
+#endif
+  }
+#endif
+  void Grid_quiesce_nodes(void)
+  {
+#ifdef GRID_COMMS_MPI
+    int me;
+    MPI_Comm_rank(MPI_COMM_WORLD,&me);
+    if ( me ) { 
+      std::cout.setstate(std::ios::badbit);
+    }
+#endif
+  }
+  void Grid_unquiesce_nodes(void)
+  {
+#ifdef GRID_COMMS_MPI
+    std::cout.clear();
+#endif
+  }
+
 void Grid_init(int *argc,char ***argv)
 {
 #ifdef GRID_COMMS_MPI
   MPI_Init(argc,argv);
 #endif
   Grid_debug_handler_init();
+  Grid_quiesce_nodes();
 }
 void Grid_finalize(void)
 {
@@ -35,6 +74,10 @@ double usecond(void) {
   return 1.0*tv.tv_usec + 1.0e6*tv.tv_sec;
 }
 
+
+#define _NBACKTRACE (256)
+void * Grid_backtrace_buffer[_NBACKTRACE];
+
 void Grid_sa_signal_handler(int sig,siginfo_t *si,void * ptr)
 {
   printf("Caught signal %d\n",si->si_signo);
@@ -43,10 +86,8 @@ void Grid_sa_signal_handler(int sig,siginfo_t *si,void * ptr)
 
 #ifdef __X86_64
     ucontext_t * uc= (ucontext_t *)ptr;
-
   struct sigcontext *sc = (struct sigcontext *)&uc->uc_mcontext;
   printf("  instruction %llx\n",(uint64_t)sc->rip);
-
 #define REG(A)  printf("  %s %lx\n",#A, sc-> A);
   REG(rdi);
   REG(rsi);
@@ -68,14 +109,14 @@ void Grid_sa_signal_handler(int sig,siginfo_t *si,void * ptr)
   REG(r14);
   REG(r15);
 #endif
-
-  fflush(stdout);
-
-  if ( si->si_signo == SIGSEGV ) {
-    printf("Grid_sa_signal_handler: Oops... this was a sigsegv you naughty naughty programmer. Goodbye\n");
-    fflush(stdout);
-    exit(-1);
+#ifdef MAC
+  int symbols    = backtrace        (Grid_backtrace_buffer,_NBACKTRACE);
+  char **strings = backtrace_symbols(Grid_backtrace_buffer,symbols);
+  for (int i = 0; i < symbols; i++){
+    printf ("%s\n", strings[i]);
   }
+#endif
+  exit(0);
   return;
 };
 
diff --git a/lib/Grid_summation.h b/lib/Grid_summation.h
index 12eeb2bc..c500c5f1 100644
--- a/lib/Grid_summation.h
+++ b/lib/Grid_summation.h
@@ -16,6 +16,96 @@ inline void subdivides(GridBase *coarse,GridBase *fine)
   }
 }
 
+template<class vobj,int nbasis>
+inline void projectBlockBasis(Lattice<iVector<vComplex,nbasis > > &coarseData,
+			      const             Lattice<vobj>   &fineData,
+			      const std::vector<Lattice<vobj> > &Basis)
+{
+  GridBase * fine  = fineData._grid;
+  GridBase * coarse= coarseData._grid;
+  int  _ndimension = coarse->_ndimension;
+
+  // checks
+  assert( nbasis == Basis.size() );
+  subdivides(coarse,fine); 
+  for(int i=0;i<nbasis;i++){
+    conformable(Basis,fineData);
+  }
+
+  std::vector<int>  block_r      (_ndimension);
+  
+  for(int d=0 ; d<_ndimension;d++){
+    block_r[d] = fine->_rdimensions[d] / coarse->_rdimensions[d];
+  }
+
+  coarseData=zero;
+
+  // Loop with a cache friendly loop ordering
+  for(int sf=0;sf<fine->oSites();sf++){
+
+    int sc;
+    std::vector<int> coor_c(_ndimension);
+    std::vector<int> coor_f(_ndimension);
+    GridBase::CoorFromIndex(coor_f,sf,fine->_rdimensions);
+    for(int d=0;d<_ndimension;d++) coor_c[d]=coor_f[d]/block_r[d];
+    GridBase::IndexFromCoor(coor_c,sc,coarse->_rdimensions);
+
+    for(int i=0;i<nbasis;i++) {
+      
+      coarseData._odata[sc][i]=coarseData._odata[sc][i]
+	+ innerProduct(Basis[i]._odata[sf],fineData._odata[sf]);
+
+    }
+  }
+  return;
+  
+}
+
+
+template<class vobj,int nbasis>
+inline void promoteBlockBasis(const Lattice<iVector<vComplex,nbasis > > &coarseData,
+			                        Lattice<vobj>   &fineData,
+			      const std::vector<Lattice<vobj> > &Basis)
+{
+  GridBase * fine  = fineData._grid;
+  GridBase * coarse= coarseData._grid;
+  int  _ndimension = coarse->_ndimension;
+
+  // checks
+  assert( nbasis == Basis.size() );
+  subdivides(coarse,fine); 
+  for(int i=0;i<nbasis;i++){
+    conformable(Basis,fineData);
+  }
+
+  std::vector<int>  block_r      (_ndimension);
+  
+  for(int d=0 ; d<_ndimension;d++){
+    block_r[d] = fine->_rdimensions[d] / coarse->_rdimensions[d];
+  }
+
+  // Loop with a cache friendly loop ordering
+  for(int sf=0;sf<fine->oSites();sf++){
+
+    int sc;
+    std::vector<int> coor_c(_ndimension);
+    std::vector<int> coor_f(_ndimension);
+
+    GridBase::CoorFromIndex(coor_f,sf,fine->_rdimensions);
+    for(int d=0;d<_ndimension;d++) coor_c[d]=coor_f[d]/block_r[d];
+    GridBase::IndexFromCoor(coor_c,sc,coarse->_rdimensions);
+
+    for(int i=0;i<nbasis;i++) {
+
+      if(i==0) fineData._odata[sf]=                    coarseData._odata[sc][i]*Basis[i]._odata[sf];
+      else     fineData._odata[sf]=fineData._odata[sf]+coarseData._odata[sc][i]*Basis[i]._odata[sf];
+
+    }
+  }
+  return;
+  
+}
+
 // useful in multigrid project;
 // Generic name : Coarsen?
 template<class vobj>
@@ -30,10 +120,6 @@ inline void sumBlocks(Lattice<vobj> &coarseData,const Lattice<vobj> &fineData)
   
   std::vector<int>  block_r      (_ndimension);
   
-  ///////////////////////////////////////////////////////////
-  // Detect whether the result is replicated in dimension d
-  ///////////////////////////////////////////////////////////
-  
   for(int d=0 ; d<_ndimension;d++){
     block_r[d] = fine->_rdimensions[d] / coarse->_rdimensions[d];
   }
diff --git a/tests/Grid_nersc_io.cc b/tests/Grid_nersc_io.cc
index 35cad831..f9faf7ef 100644
--- a/tests/Grid_nersc_io.cc
+++ b/tests/Grid_nersc_io.cc
@@ -11,12 +11,15 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> simd_layout({1,1,2,2});
-  std::vector<int> mpi_layout ({1,1,1,1});
+  std::vector<int> mpi_layout ({2,2,2,2});
   std::vector<int> latt_size  ({16,16,16,32});
+  std::vector<int> clatt_size  ({4,4,4,8});
   int orthodir=3;
   int orthosz =latt_size[orthodir];
     
   GridCartesian     Fine(latt_size,simd_layout,mpi_layout);
+  GridCartesian     Coarse(clatt_size,simd_layout,mpi_layout);
+
   GridRNG           FineRNG(&Fine);
   LatticeGaugeField Umu(&Fine);
 
@@ -40,6 +43,7 @@ int main (int argc, char ** argv)
 
   // (1+2+3)=6 = N(N-1)/2 terms
   LatticeComplex Plaq(&Fine);
+  LatticeComplex cPlaq(&Coarse);
   Plaq = zero;
   for(int mu=1;mu<Nd;mu++){
     for(int nu=0;nu<mu;nu++){
@@ -47,6 +51,7 @@ int main (int argc, char ** argv)
     }
   }
 
+  
   double vol = Fine.gSites();
   Complex PlaqScale(1.0/vol/6.0/3.0);
 
@@ -66,7 +71,6 @@ int main (int argc, char ** argv)
     std::cout << "total " <<Pt*PlaqScale<<std::endl;
   }  
 
-
   TComplex Tp = sum(Plaq);
   Complex p  = TensorRemove(Tp);
   std::cout << "calculated plaquettes " <<p*PlaqScale<<std::endl;
@@ -77,5 +81,10 @@ int main (int argc, char ** argv)
   Complex l  = TensorRemove(Tl);
   std::cout << "calculated link trace " <<l*LinkTraceScale<<std::endl;
 
+  sumBlocks(cPlaq,Plaq);
+  TComplex TcP = sum(cPlaq);
+  Complex ll= TensorRemove(TcP);
+  std::cout << "coarsened plaquettes sum to " <<ll*PlaqScale<<std::endl;
+  
   Grid_finalize();
 }

From b8eef54fa743476693fd1a3056caebe4063f0d4a Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 24 Apr 2015 18:20:03 +0100
Subject: [PATCH 095/429] First implementation of Dirac matrices as a Gamma
 class.

---
 lib/Grid_simd.h                               |   7 +-
 lib/math/Grid_math_arith_add.h                |   7 -
 lib/math/Grid_math_reality.h                  |  59 ++-
 lib/math/Grid_math_tensors.h                  |  23 +
 lib/qcd/Grid_qcd.h                            |   3 +
 lib/qcd/Grid_qcd_dirac.h                      | 393 ++++++++++++++++++
 lib/simd/Grid_vComplexD.h                     |  66 ++-
 lib/simd/Grid_vComplexF.h                     |  45 +-
 tests/Grid_gamma.cc                           |  95 +++++
 .../{test_Grid_stencil.cc => Grid_stencil.cc} |   0
 tests/Makefile.am                             |   9 +-
 11 files changed, 683 insertions(+), 24 deletions(-)
 create mode 100644 lib/qcd/Grid_qcd_dirac.h
 create mode 100644 tests/Grid_gamma.cc
 rename tests/{test_Grid_stencil.cc => Grid_stencil.cc} (100%)

diff --git a/lib/Grid_simd.h b/lib/Grid_simd.h
index 17755fe0..d8788e33 100644
--- a/lib/Grid_simd.h
+++ b/lib/Grid_simd.h
@@ -61,7 +61,12 @@ namespace Grid {
     inline void add (ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) + (*r); }
     inline ComplexF  adj(const ComplexF& r ){ return(conj(r)); }
     //conj already supported for complex
-    
+
+    inline ComplexF timesI(const ComplexF r)     { return(r*ComplexF(0.0,1.0));}
+    inline ComplexF timesMinusI(const ComplexF r){ return(r*ComplexF(0.0,-1.0));}
+    inline ComplexD timesI(const ComplexD r)     { return(r*ComplexD(0.0,1.0));}
+    inline ComplexD timesMinusI(const ComplexD r){ return(r*ComplexD(0.0,-1.0));}
+
     inline void mac (RealD * __restrict__ y,const RealD * __restrict__ a,const RealD *__restrict__ x){  *y = (*a) * (*x)+(*y);}
     inline void mult(RealD * __restrict__ y,const RealD * __restrict__ l,const RealD *__restrict__ r){ *y = (*l) * (*r);}
     inline void sub (RealD * __restrict__ y,const RealD * __restrict__ l,const RealD *__restrict__ r){ *y = (*l) - (*r);}
diff --git a/lib/math/Grid_math_arith_add.h b/lib/math/Grid_math_arith_add.h
index 7e62e07b..93449402 100644
--- a/lib/math/Grid_math_arith_add.h
+++ b/lib/math/Grid_math_arith_add.h
@@ -63,13 +63,6 @@ namespace Grid {
     return;
   }
 
-  // Need to figure multi-precision.
-  template<class Mytype>  Mytype timesI(Mytype &r)
-    {
-      iScalar<Complex> i;
-      i._internal = Complex(0,1);
-      return r*i;
-    }
 
   // + operator for scalar, vector, matrix
   template<class ltype,class rtype>
diff --git a/lib/math/Grid_math_reality.h b/lib/math/Grid_math_reality.h
index 1f676c11..1bf34f2a 100644
--- a/lib/math/Grid_math_reality.h
+++ b/lib/math/Grid_math_reality.h
@@ -5,8 +5,63 @@ namespace Grid {
     ///////////////////////////////////////////////////////////////////////////////////////////////////
     /////////////////////////////////////////// CONJ         ///////////////////////////////////////////
     ///////////////////////////////////////////////////////////////////////////////////////////////////
- 
+
+#ifdef GRID_WARN_SUBOPTIMAL
+#warning "Optimisation alert switch over to two argument form to avoid copy back in perf critical timesI "
+#endif     
+/////////////////////////////////////////////// 
+// multiply by I; make recursive.
+/////////////////////////////////////////////// 
+template<class vtype> inline iScalar<vtype> timesI(const iScalar<vtype>&r) 
+{
+    iScalar<vtype> ret;
+    ret._internal = timesI(r._internal);
+    return ret;
+}
+template<class vtype,int N> inline iVector<vtype,N> timesI(const iVector<vtype,N>&r) 
+{
+  iVector<vtype,N> ret;
+  for(int i=0;i<N;i++){
+    ret._internal[i] = timesI(r._internal[i]);
+  }
+  return ret;
+}
+template<class vtype,int N> inline iMatrix<vtype,N> timesI(const iMatrix<vtype,N>&r)
+{
+  iMatrix<vtype,N> ret;
+  for(int i=0;i<N;i++){
+  for(int j=0;j<N;j++){
+    ret._internal[i][j] = timesI(r._internal[i][j]);
+  }}
+  return ret;
+}
+
+template<class vtype> inline iScalar<vtype> timesMinusI(const iScalar<vtype>&r) 
+{
+    iScalar<vtype> ret;
+    ret._internal = timesMinusI(r._internal);
+    return ret;
+}
+template<class vtype,int N> inline iVector<vtype,N> timesMinusI(const iVector<vtype,N>&r) 
+{
+  iVector<vtype,N> ret;
+  for(int i=0;i<N;i++){
+    ret._internal[i] = timesMinusI(r._internal[i]);
+  }
+  return ret;
+}
+template<class vtype,int N> inline iMatrix<vtype,N> timesMinusI(const iMatrix<vtype,N>&r)
+{
+  iMatrix<vtype,N> ret;
+  for(int i=0;i<N;i++){
+  for(int j=0;j<N;j++){
+    ret._internal[i][j] = timesMinusI(r._internal[i][j]);
+  }}
+  return ret;
+}
+/////////////////////////////////////////////// 
 // Conj function for scalar, vector, matrix
+/////////////////////////////////////////////// 
 template<class vtype> inline iScalar<vtype> conj(const iScalar<vtype>&r)
 {
     iScalar<vtype> ret;
@@ -31,7 +86,9 @@ template<class vtype,int N> inline iMatrix<vtype,N> conj(const iMatrix<vtype,N>&
   return ret;
 }
 
+/////////////////////////////////////////////// 
 // Adj function for scalar, vector, matrix
+/////////////////////////////////////////////// 
 template<class vtype> inline iScalar<vtype> adj(const iScalar<vtype>&r)
 {
     iScalar<vtype> ret;
diff --git a/lib/math/Grid_math_tensors.h b/lib/math/Grid_math_tensors.h
index 379e61d2..8bca6805 100644
--- a/lib/math/Grid_math_tensors.h
+++ b/lib/math/Grid_math_tensors.h
@@ -72,6 +72,13 @@ public:
       return _internal;
     }
 
+    inline const vtype & operator ()(void) const {
+      return _internal;
+    }
+    //    inline vtype && operator ()(void) {
+    //      return _internal;
+    //    }
+
     operator ComplexD () const { return(TensorRemove(_internal)); };
     operator RealD () const { return(real(TensorRemove(_internal))); }
 
@@ -156,6 +163,12 @@ public:
     inline vtype & operator ()(int i) {
       return _internal[i];
     }
+    inline const vtype & operator ()(int i) const {
+      return _internal[i];
+    }
+    //    inline vtype && operator ()(int i) {
+    //      return _internal[i];
+    //    }
 };
     
 template<class vtype,int N> class iMatrix
@@ -235,12 +248,22 @@ public:
     *this = (*this)+r;
     return *this;
   }
+
+  // returns an lvalue reference
   inline vtype & operator ()(int i,int j) {
     return _internal[i][j];
   }
+  inline const vtype & operator ()(int i,int j) const {
+    return _internal[i][j];
+  }
+
+  //  inline vtype && operator ()(int i,int j) {
+  //    return _internal[i][j];
+  //  }
 
 };
 
+
 template<class vobj> inline 
 void extract(const vobj &vec,std::vector<typename vobj::scalar_object> &extracted)
 {
diff --git a/lib/qcd/Grid_qcd.h b/lib/qcd/Grid_qcd.h
index b30b7a67..a5c4320e 100644
--- a/lib/qcd/Grid_qcd.h
+++ b/lib/qcd/Grid_qcd.h
@@ -143,4 +143,7 @@ namespace QCD {
 
 }   //namespace QCD
 } // Grid
+
+#include <qcd/Grid_qcd_dirac.h>
+
 #endif
diff --git a/lib/qcd/Grid_qcd_dirac.h b/lib/qcd/Grid_qcd_dirac.h
new file mode 100644
index 00000000..86e3870d
--- /dev/null
+++ b/lib/qcd/Grid_qcd_dirac.h
@@ -0,0 +1,393 @@
+#ifndef GRID_QCD_DIRAC_H
+#define GRID_QCD_DIRAC_H
+namespace Grid{
+
+namespace QCD {
+
+  class Gamma {
+
+  public:
+
+    const int Ns=4;
+    
+    enum GammaMatrix {
+      Identity,                        
+      GammaX, 
+      GammaY, 
+      GammaZ, 
+      GammaT,  
+      Gamma5,
+      //      GammaXGamma5, 
+      //      GammaYGamma5, 
+      //      GammaZGamma5, 
+      //      GammaTGamma5,  
+      //      SigmaXY,
+      //      SigmaXZ,
+      //      SigmaYZ,
+      //      SigmaXT,
+      //      SigmaYT,
+      //      SigmaZT,
+      MinusIdentity,                        
+      MinusGammaX, 
+      MinusGammaY, 
+      MinusGammaZ, 
+      MinusGammaT,  
+      MinusGamma5
+      //      MinusGammaXGamma5, easiest to form by composition
+      //      MinusGammaYGamma5, as performance is not critical for these
+      //      MinusGammaZGamma5, 
+      //      MinusGammaTGamma5,  
+      //      MinusSigmaXY,
+      //      MinusSigmaXZ,
+      //      MinusSigmaYZ,
+      //      MinusSigmaXT,
+      //      MinusSigmaYT,
+      //      MinusSigmaZT
+    };
+
+    Gamma (GammaMatrix g) { _g=g; }
+
+    GammaMatrix _g;
+
+
+  };
+
+
+    /* Gx
+     *  0 0  0  i    
+     *  0 0  i  0    
+     *  0 -i 0  0
+     * -i 0  0  0
+     */
+    template<class vtype> inline void rmultMinusGammaX(iMatrix<vtype,Ns> &ret,const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(i,0) = timesI(rhs(i,3));
+	ret(i,1) = timesI(rhs(i,2));
+	ret(i,2) = timesMinusI(rhs(i,1));
+	ret(i,3) = timesMinusI(rhs(i,0));
+      }
+    };
+    template<class vtype> inline void rmultGammaX(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(i,0) = timesMinusI(rhs(i,3));
+	ret(i,1) = timesMinusI(rhs(i,2));
+	ret(i,2) = timesI(rhs(i,1));
+	ret(i,3) = timesI(rhs(i,1));
+      }
+    };
+
+    template<class vtype> inline void multGammaX(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+      ret._internal[0] = timesI(rhs._internal[3]);
+      ret._internal[1] = timesI(rhs._internal[2]);
+      ret._internal[2] = timesMinusI(rhs._internal[1]);
+      ret._internal[3] = timesMinusI(rhs._internal[0]);
+    };
+    template<class vtype> inline void multMinusGammaX(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+	ret(0) = timesMinusI(rhs(3));
+	ret(1) = timesMinusI(rhs(2));
+	ret(2) = timesI(rhs(1));
+	ret(3) = timesI(rhs(0));
+    };
+    template<class vtype> inline void multGammaX(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(0,i) = timesI(rhs(3,i));
+	ret(1,i) = timesI(rhs._internal[2][i]);
+	ret(2,i) = timesMinusI(rhs._internal[1][i]);
+	ret(3,i) = timesMinusI(rhs._internal[0][i]);
+      }
+    };
+    template<class vtype> inline void multMinusGammaX(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(0,i) = timesMinusI(rhs(3,i));
+	ret(1,i) = timesMinusI(rhs(2,i));
+	ret(2,i) = timesI(rhs(1,i));
+	ret(3,i) = timesI(rhs(0,i));
+      }
+    };
+
+
+
+
+    /*Gy
+     *  0 0  0  -1  [0] -+ [3]
+     *  0 0  1  0   [1] +- [2]
+     *  0 1  0  0
+     * -1 0  0  0
+     */
+    template<class vtype> inline void multGammaY(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+      ret(0) = -rhs(3);
+      ret(1) =  rhs(2);
+      ret(2) =  rhs(1);
+      ret(3) = -rhs(0);
+    };
+    template<class vtype> inline void multMinusGammaY(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+      ret(0) =  rhs(3);
+      ret(1) = -rhs(2);
+      ret(2) = -rhs(1);
+      ret(3) =  rhs(0);
+    };
+    template<class vtype> inline void multGammaY(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(0,i) = -rhs(3,i);
+	ret(1,i) =  rhs(2,i);
+	ret(2,i) =  rhs(1,i);
+	ret(3,i) = -rhs(0,i);
+      }
+    };
+    template<class vtype> inline void multMinusGammaY(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(0,i) =  rhs(3,i);
+	ret(1,i) = -rhs(2,i);
+	ret(2,i) = -rhs(1,i);
+	ret(3,i) =  rhs(0,i);
+      }
+    };
+    /*Gz
+     *  0 0  i  0   [0]+-i[2]
+     *  0 0  0 -i   [1]-+i[3]
+     * -i 0  0  0
+     *  0 i  0  0
+     */
+    template<class vtype> inline void multGammaZ(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+      ret(0) = timesI(rhs(2));
+      ret(1) =timesMinusI(rhs(3));
+      ret(2) =timesMinusI(rhs(0));
+      ret(3) = timesI(rhs(1));
+    };
+    template<class vtype> inline void multMinusGammaZ(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+      ret(0) = timesMinusI(rhs(2));
+      ret(1) = timesI(rhs(3));
+      ret(2) = timesI(rhs(0));
+      ret(3) = timesMinusI(rhs(1));
+    };
+    template<class vtype> inline void multGammaZ(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(0,i) = timesI(rhs(2,i));
+	ret(1,i) =timesMinusI(rhs(3,i));
+	ret(2,i) =timesMinusI(rhs(0,i));
+	ret(3,i) = timesI(rhs(1,i));
+      }
+    };
+    template<class vtype> inline void multMinusGammaZ(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(0,i) = timesMinusI(rhs(2,i));
+	ret(1,i) = timesI(rhs(3,i));
+	ret(2,i) = timesI(rhs(0,i));
+	ret(3,i) = timesMinusI(rhs(1,i));
+      }
+    };
+    /*Gt
+     *  0 0  1  0 [0]+-[2]
+     *  0 0  0  1 [1]+-[3]
+     *  1 0  0  0
+     *  0 1  0  0
+     */
+    template<class vtype> inline void multGammaT(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+      ret(0) = rhs(2);
+      ret(1) = rhs(3);
+      ret(2) = rhs(0);
+      ret(3) = rhs(1);
+    };
+    template<class vtype> inline void multMinusGammaT(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+      ret(0) =-rhs(2);
+      ret(1) =-rhs(3);
+      ret(2) =-rhs(0);
+      ret(3) =-rhs(1);
+    };
+    template<class vtype> inline void multGammaT(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(0,i) = rhs(2,i);
+	ret(1,i) = rhs(3,i);
+	ret(2,i) = rhs(0,i);
+	ret(3,i) = rhs(1,i);
+      }
+    };
+    template<class vtype> inline void multMinusGammaT(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(0,i) =-rhs(2,i);
+	ret(1,i) =-rhs(3,i);
+	ret(2,i) =-rhs(0,i);
+	ret(3,i) =-rhs(1,i);
+      }
+    };
+    /*G5
+     *  1 0  0  0 [0]+-[2]
+     *  0 1  0  0 [1]+-[3]
+     *  0 0 -1  0
+     *  0 0  0 -1
+     */
+    template<class vtype> inline void multGamma5(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+      ret(0) = rhs(0);
+      ret(1) = rhs(1);
+      ret(2) =-rhs(2);
+      ret(3) =-rhs(3);
+    };
+    template<class vtype> inline void multMinusGamma5(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+      ret(0) =-rhs(0);
+      ret(1) =-rhs(1);
+      ret(2) = rhs(2);
+      ret(3) = rhs(3);
+    };
+    template<class vtype> inline void multGamma5(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(0,i) = rhs(0,i);
+	ret(1,i) = rhs(1,i);
+	ret(2,i) =-rhs(2,i);
+	ret(3,i) =-rhs(3,i);
+      }
+    };
+    template<class vtype> inline void multMinusGamma5(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(0,i) =-rhs(0,i);
+	ret(1,i) =-rhs(1,i);
+	ret(2,i) = rhs(2,i);
+	ret(3,i) = rhs(3,i);
+      }
+    };
+
+
+
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    // Operator * : first case this is not a spin index, so recurse
+    ///////////////////////////////////////////////////////////////////////////////////////////////////
+    
+    // FIXME
+    //
+    // Optimisation; switch over to a "multGammaX(ret._internal,arg._internal)" style early and
+    // note that doing so from the lattice operator will avoid copy back and case switch overhead, as
+    // was done for the tensor math operator to remove operator * notation early
+    //
+#ifdef GRID_WARN_SUBOPTIMAL
+#warning "Optimisation alert switch over to multGammaX early "
+#endif     
+
+    template<class vtype> inline auto operator * ( const Gamma &G,const iScalar<vtype> &arg) ->
+      typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type 
+
+	{
+	  iScalar<vtype> ret;
+	  ret._internal=G*arg._internal;
+	  return ret;
+	}
+    template<class vtype,int N> inline auto operator * ( const Gamma &G,const iVector<vtype,N> &arg) ->
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type 
+	{
+	  iVector<vtype,N> ret;
+	  ret._internal=G*arg._internal;
+	  return ret;
+	}
+    template<class vtype,int N> inline auto operator * ( const Gamma &G,const iMatrix<vtype,N> &arg) ->
+      typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type 
+	{
+	  iMatrix<vtype,N> ret;
+	  ret._internal=G*arg._internal;
+	  return ret;
+	}
+    ////////////////////////////////////////////////////////
+    // When we hit the spin index this matches and we stop
+    ////////////////////////////////////////////////////////
+    template<class vtype> inline auto operator * ( const Gamma &G,const iMatrix<vtype,Ns> &arg) ->
+      typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,Ns>,SpinIndex>::value,iMatrix<vtype,Ns> >::type 
+      {
+	iMatrix<vtype,Ns> ret;
+	switch (G._g) {
+	case Gamma::Identity:
+	  ret = arg;
+	  break;
+	case Gamma::MinusIdentity:
+	  ret = -arg;
+	  break;
+	case Gamma::GammaX:
+	  multGammaX(ret,arg);
+	  break;
+	case Gamma::MinusGammaX:
+	  multMinusGammaX(ret,arg);
+	  break;
+	case Gamma::GammaY:
+	  multGammaY(ret,arg);
+	  break;
+	case Gamma::MinusGammaY:
+	  multMinusGammaY(ret,arg);
+	  break;
+	case Gamma::GammaZ:
+	  multGammaZ(ret,arg);
+	  break;
+	case Gamma::MinusGammaZ:
+	  multMinusGammaZ(ret,arg);
+	  break;
+	case Gamma::GammaT:
+	  multGammaT(ret,arg);
+	  break;
+	case Gamma::MinusGammaT:
+	  multMinusGammaT(ret,arg);
+	  break;
+	case Gamma::Gamma5:
+	  multGamma5(ret,arg);
+	  break;
+	case Gamma::MinusGamma5:
+	  multMinusGamma5(ret,arg);
+	  break;
+	default:
+	  assert(0);
+	  break;
+	}
+	return ret;
+      }
+    /* Output from test
+./Grid_gamma 
+Identity((1,0),(0,0),(0,0),(0,0))
+        ((0,0),(1,0),(0,0),(0,0))
+        ((0,0),(0,0),(1,0),(0,0))
+        ((0,0),(0,0),(0,0),(1,0))   OK
+
+GammaX  ((0,0),(0,0),(0,0),(0,1))
+        ((0,0),(0,0),(0,1),(0,0))
+        ((0,0),(0,-1),(0,0),(0,0))
+        ((0,-1),(0,0),(0,0),(0,0))  OK
+    * Gx
+    *  0 0  0  i    
+    *  0 0  i  0    
+    *  0 -i 0  0
+    * -i 0  0  0
+
+GammaY  ((-0,-0),(-0,-0),(-0,-0),(-1,-0))
+        ((0,0),(0,0),(1,0),(0,0))
+        ((0,0),(1,0),(0,0),(0,0))          OK
+        ((-1,-0),(-0,-0),(-0,-0),(-0,-0))
+     *Gy
+     *  0 0  0  -1  [0] -+ [3]
+     *  0 0  1  0   [1] +- [2]
+     *  0 1  0  0
+     * -1 0  0  0
+
+GammaZ  ((0,0),(0,0),(0,1),(0,0))
+        ((0,0),(0,0),(0,0),(0,-1))
+        ((0,-1),(0,0),(0,0),(0,0))
+        ((0,0),(0,1),(0,0),(0,0))   OK
+     *  0 0  i  0   [0]+-i[2]
+     *  0 0  0 -i   [1]-+i[3]
+     * -i 0  0  0
+     *  0 i  0  0
+
+GammaT  ((0,0),(0,0),(1,0),(0,0))
+        ((0,0),(0,0),(0,0),(1,0))  OK
+        ((1,0),(0,0),(0,0),(0,0))
+        ((0,0),(1,0),(0,0),(0,0))
+     *  0 0  1  0 [0]+-[2]
+     *  0 0  0  1 [1]+-[3]
+     *  1 0  0  0
+     *  0 1  0  0
+
+Gamma5  ((1,0),(0,0),(0,0),(0,0))
+        ((0,0),(1,0),(0,0),(0,0))
+        ((-0,-0),(-0,-0),(-1,-0),(-0,-0))
+        ((-0,-0),(-0,-0),(-0,-0),(-1,-0))
+     *  1 0  0  0 [0]+-[2]
+     *  0 1  0  0 [1]+-[3]  OK
+     *  0 0 -1  0
+     *  0 0  0 -1
+     */
+
+}   //namespace QCD
+} // Grid
+#endif
diff --git a/lib/simd/Grid_vComplexD.h b/lib/simd/Grid_vComplexD.h
index 313eef4a..402d303c 100644
--- a/lib/simd/Grid_vComplexD.h
+++ b/lib/simd/Grid_vComplexD.h
@@ -101,13 +101,13 @@ namespace Grid {
              IF IMM0[2] = 0
              THEN DEST[191:128]=SRC1[191:128] ELSE DEST[191:128]=SRC1[255:192] FI;
              IF IMM0[3] = 0
-             THEN DEST[255:192]=SRC2[191:128] ELSE DEST[255:192]=SRC2[255:192] FI;
+             THEN DEST[255:192]=SRC2[191:128] ELSE DEST[255:192]=SRC2[255:192] FI; // Ox5 r<->i   ; 0xC unchanged
              */
 #if defined (AVX1)|| defined (AVX2)
             zvec ymm0,ymm1,ymm2;
             ymm0 = _mm256_shuffle_pd(a.v,a.v,0x0); // ymm0 <- ar ar, ar,ar b'00,00
             ymm0 = _mm256_mul_pd(ymm0,b.v);        // ymm0 <- ar bi, ar br
-            ymm1 = _mm256_shuffle_pd(b.v,b.v,0x5); // ymm1 <- br,bi  b'01,01
+            ymm1 = _mm256_shuffle_pd(b.v,b.v,0x5); // ymm1 <- br,bi  b'01,01 
             ymm2 = _mm256_shuffle_pd(a.v,a.v,0xF); // ymm2 <- ai,ai  b'11,11
             ymm1 = _mm256_mul_pd(ymm1,ymm2);       // ymm1 <- br ai, ai bi
             ret.v= _mm256_addsub_pd(ymm0,ymm1);
@@ -140,7 +140,7 @@ namespace Grid {
              * publisher = {ACM},
              * address = {New York, NY, USA},
              * keywords = {autovectorization, fourier transform, program generation, simd, super-optimization},
- *                } 
+	     *                } 
              */
             zvec vzero,ymm0,ymm1,real,imag;
             vzero =  _mm512_setzero();
@@ -252,23 +252,71 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
             vComplexD ret ; vzero(ret);
 #if defined (AVX1)|| defined (AVX2)
             // addsubps 0, inv=>0+in.v[3] 0-in.v[2], 0+in.v[1], 0-in.v[0], ...
-            __m256d tmp = _mm256_addsub_pd(ret.v,_mm256_shuffle_pd(in.v,in.v,0x5));
-             ret.v=_mm256_shuffle_pd(tmp,tmp,0x5);
+	    //            __m256d tmp = _mm256_addsub_pd(ret.v,_mm256_shuffle_pd(in.v,in.v,0x5));
+	    //             ret.v=_mm256_shuffle_pd(tmp,tmp,0x5);
+            ret.v = _mm256_addsub_pd(ret.v,in.v);
 #endif
 #ifdef SSE4
             ret.v = _mm_addsub_pd(ret.v,in.v);
 #endif
 #ifdef AVX512
-             // Xeon does not have fmaddsub or addsub 
-             // with mask 0xa (1010), v[0] -v[1] v[2] -v[3] ....
-             ret.v = _mm512_mask_sub_pd(in.v, 0xaaaa,ret.v, in.v);
-             
+	    ret.v = _mm512_mask_sub_pd(in.v, 0xaaaa,ret.v, in.v);             
 #endif
 #ifdef QPX
 	     assert(0);
 #endif
             return ret;
         }
+
+        friend inline vComplexD timesI(const vComplexD &in){
+	  vComplexD ret; vzero(ret);
+#if defined (AVX1)|| defined (AVX2)
+	  cvec tmp =_mm256_addsub_ps(ret.v,in.v); // r,-i
+	  /*
+             IF IMM0[0] = 0
+             THEN DEST[63:0]=SRC1[63:0] ELSE DEST[63:0]=SRC1[127:64] FI;
+             IF IMM0[1] = 0
+             THEN DEST[127:64]=SRC2[63:0] ELSE DEST[127:64]=SRC2[127:64] FI;
+             IF IMM0[2] = 0
+             THEN DEST[191:128]=SRC1[191:128] ELSE DEST[191:128]=SRC1[255:192] FI;
+             IF IMM0[3] = 0
+             THEN DEST[255:192]=SRC2[191:128] ELSE DEST[255:192]=SRC2[255:192] FI;
+	  */
+          ret.v    =_mm256_shuffle_ps(tmp,tmp,0x5);
+#endif
+#ifdef SSE4
+	  cvec tmp =_mm_addsub_ps(ret.v,in.v); // r,-i
+          ret.v    =_mm_shuffle_ps(tmp,tmp,0x5);
+#endif
+#ifdef AVX512
+          ret.v = _mm512_mask_sub_ps(in.v,0xaaaa,ret.v,in.v); // real -imag 
+	  ret.v = _mm512_swizzle_ps(ret.v, _MM_SWIZ_REG_CDAB);// OK
+#endif
+#ifdef QPX
+            assert(0);
+#endif
+	  return ret;
+	}
+        friend inline vComplexD timesMinusI(const vComplexD &in){
+	  vComplexD ret; vzero(ret);
+#if defined (AVX1)|| defined (AVX2)
+	  cvec tmp =_mm256_shuffle_ps(in.v,in.v,0x5);
+          ret.v    =_mm256_addsub_ps(ret.v,tmp); // i,-r
+#endif
+#ifdef SSE4
+	  cvec tmp =_mm_shuffle_ps(in.v,in.v,0x5);
+          ret.v    =_mm_addsub_ps(ret.v,tmp); // r,-i
+#endif
+#ifdef AVX512
+          cvec tmp = _mm512_swizzle_ps(in.v, _MM_SWIZ_REG_CDAB);// OK
+	  ret.v    = _mm512_mask_sub_ps(tmp,0xaaaa,ret.v,tmp); // real -imag 
+#endif
+#ifdef QPX
+            assert(0);
+#endif
+	  return ret;
+	}
+
 // REDUCE FIXME must be a cleaner implementation
        friend inline ComplexD Reduce(const vComplexD & in)
        { 
diff --git a/lib/simd/Grid_vComplexF.h b/lib/simd/Grid_vComplexF.h
index d931964c..4c31ad04 100644
--- a/lib/simd/Grid_vComplexF.h
+++ b/lib/simd/Grid_vComplexF.h
@@ -329,9 +329,10 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
         friend inline vComplexF conj(const vComplexF &in){
             vComplexF ret ; vzero(ret);
 #if defined (AVX1)|| defined (AVX2)
-             cvec tmp;
-             tmp = _mm256_addsub_ps(ret.v,_mm256_shuffle_ps(in.v,in.v,_MM_SHUFFLE(2,3,0,1))); // ymm1 <- br,bi
-             ret.v=_mm256_shuffle_ps(tmp,tmp,_MM_SHUFFLE(2,3,0,1));
+	    //             cvec tmp;
+	    //             tmp = _mm256_addsub_ps(ret.v,_mm256_shuffle_ps(in.v,in.v,_MM_SHUFFLE(2,3,0,1))); // ymm1 <- br,bi
+	    //             ret.v=_mm256_shuffle_ps(tmp,tmp,_MM_SHUFFLE(2,3,0,1));
+	    ret.v = _mm256_addsub_ps(ret.v,in.v);
 #endif
 #ifdef SSE4
             ret.v = _mm_addsub_ps(ret.v,in.v);
@@ -344,6 +345,44 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
 #endif
             return ret;
         }
+        friend inline vComplexF timesI(const vComplexF &in){
+	  vComplexF ret; vzero(ret);
+#if defined (AVX1)|| defined (AVX2)
+	  cvec tmp =_mm256_addsub_ps(ret.v,in.v); // r,-i
+          ret.v = _mm256_shuffle_ps(tmp,tmp,0x5);
+#endif
+#ifdef SSE4
+	  cvec tmp =_mm_addsub_ps(ret.v,in.v); // r,-i
+          ret.v = _mm_shuffle_ps(tmp,tmp,0x5);
+#endif
+#ifdef AVX512
+          ret.v = _mm512_mask_sub_ps(in.v,0xaaaa,ret.v,in.v); // real -imag 
+	  ret.v = _mm512_swizzle_ps(ret.v, _MM_SWIZ_REG_CDAB);// OK
+#endif
+#ifdef QPX
+            assert(0);
+#endif
+	  return ret;
+	}
+        friend inline vComplexF timesMinusI(const vComplexF &in){
+	  vComplexF ret; vzero(ret);
+#if defined (AVX1)|| defined (AVX2)
+	  cvec tmp =_mm256_shuffle_ps(in.v,in.v,0x5);
+          ret.v = _mm256_addsub_ps(ret.v,tmp); // i,-r
+#endif
+#ifdef SSE4
+	  cvec tmp =_mm_shuffle_ps(in.v,in.v,0x5);
+          ret.v = _mm_addsub_ps(ret.v,tmp); // r,-i
+#endif
+#ifdef AVX512
+          cvec tmp = _mm512_swizzle_ps(in.v, _MM_SWIZ_REG_CDAB);// OK
+	  ret.v    = _mm512_mask_sub_ps(tmp,0xaaaa,ret.v,tmp); // real -imag 
+#endif
+#ifdef QPX
+            assert(0);
+#endif
+	  return ret;
+	}
         
         // Unary negation
         friend inline vComplexF operator -(const vComplexF &r) {
diff --git a/tests/Grid_gamma.cc b/tests/Grid_gamma.cc
new file mode 100644
index 00000000..9c213352
--- /dev/null
+++ b/tests/Grid_gamma.cc
@@ -0,0 +1,95 @@
+#include <Grid.h>
+#include <parallelIO/GridNerscIO.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> simd_layout({1,1,2,2});
+  std::vector<int> mpi_layout ({1,1,1,1});
+  std::vector<int> latt_size  ({8,8,8,8});
+    
+  GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+
+  GridRNG           RNG(&Grid);
+
+  SpinMatrix ident=zero;
+  SpinMatrix ll=zero;
+  SpinMatrix rr=zero;
+  SpinMatrix result;
+
+  for(int a=0;a<Ns;a++){
+    ident()(a,a) = 1.0;
+  }
+
+  Gamma::GammaMatrix g [] = {
+    Gamma::Identity,
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT,
+    Gamma::Gamma5,
+    Gamma::MinusIdentity,
+    Gamma::MinusGammaX,
+    Gamma::MinusGammaY,
+    Gamma::MinusGammaZ,
+    Gamma::MinusGammaT,
+    Gamma::MinusGamma5
+  };
+  const char *list[] = { 
+    "Identity ",
+    "GammaX   ",
+    "GammaY   ",
+    "GammaZ   ",
+    "GammaT   ",
+    "Gamma5   ",
+    "-Identity",
+    "-GammaX  ",
+    "-GammaY  ",
+    "-GammaZ  ",
+    "-GammaT  ",
+    "-Gamma5  ",
+    "         "
+  };
+  //  result == ll*Gamma(g[0])*rr;
+  //  result == ll*Gamma(g[0]);
+
+  for(int mu=0;mu<12;mu++){
+
+    result = Gamma(g[mu])* ident;
+
+    for(int i=0;i<Ns;i++){
+
+      if(i==0) std::cout << list[mu];
+      else     std::cout << list[12];
+
+      std::cout<<"(";
+      for(int j=0;j<Ns;j++){
+	if ( abs(result()(i,j)())==0 ) {
+	  std::cout<< " 0";
+	} else if ( abs(result()(i,j)() - ComplexF(0,1))==0){
+	  std::cout<< " i";
+	} else if ( abs(result()(i,j)() + ComplexF(0,1))==0){
+	  std::cout<< "-i";
+	} else if ( abs(result()(i,j)() - ComplexF(1,0))==0){
+	  std::cout<< " 1";
+	} else if ( abs(result()(i,j)() + ComplexF(1,0))==0){
+	  std::cout<< "-1";
+	}
+	std::cout<< ((j==Ns-1) ? ")" : "," );
+      }
+      std::cout << std::endl;
+    }
+
+    std::cout << std::endl;
+
+  }
+
+  
+  Grid_finalize();
+}
diff --git a/tests/test_Grid_stencil.cc b/tests/Grid_stencil.cc
similarity index 100%
rename from tests/test_Grid_stencil.cc
rename to tests/Grid_stencil.cc
diff --git a/tests/Makefile.am b/tests/Makefile.am
index 0be04e52..8afac541 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -5,7 +5,7 @@ AM_LDFLAGS = -L$(top_srcdir)/lib
 #
 # Test code
 #
-bin_PROGRAMS = Grid_main test_Grid_stencil Grid_nersc_io Grid_cshift
+bin_PROGRAMS = Grid_main Grid_stencil Grid_nersc_io Grid_cshift Grid_gamma
 
 Grid_main_SOURCES = Grid_main.cc
 Grid_main_LDADD = -lGrid
@@ -16,5 +16,8 @@ Grid_nersc_io_LDADD = -lGrid
 Grid_cshift_SOURCES = Grid_cshift.cc
 Grid_cshift_LDADD = -lGrid
 
-test_Grid_stencil_SOURCES = test_Grid_stencil.cc
-test_Grid_stencil_LDADD = -lGrid
+Grid_gamma_SOURCES = Grid_gamma.cc
+Grid_gamma_LDADD = -lGrid
+
+Grid_stencil_SOURCES = Grid_stencil.cc
+Grid_stencil_LDADD = -lGrid

From 74432432b65ab434e1e086a6af69e848b80fe9ec Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 24 Apr 2015 18:40:44 +0100
Subject: [PATCH 096/429] Moved code from summation into transfer and reduction

---
 lib/Grid_summation.h                 | 208 ---------------------------
 lib/lattice/Grid_lattice_reduction.h |  76 ++++++++++
 lib/lattice/Grid_lattice_transfer.h  | 140 ++++++++++++++++++
 lib/qcd/Grid_qcd_dirac.h             | 174 +++++++++++++++-------
 4 files changed, 336 insertions(+), 262 deletions(-)

diff --git a/lib/Grid_summation.h b/lib/Grid_summation.h
index c500c5f1..936ca8ef 100644
--- a/lib/Grid_summation.h
+++ b/lib/Grid_summation.h
@@ -2,213 +2,5 @@
 #define GRID_SUMMATION_H
 namespace Grid {
 
-inline void subdivides(GridBase *coarse,GridBase *fine)
-{
-  assert(coarse->_ndimension == fine->_ndimension);
-
-  int _ndimension = coarse->_ndimension;
-
-  // local and global volumes subdivide cleanly after SIMDization
-  for(int d=0;d<_ndimension;d++){
-    assert(coarse->_processors[d]  == fine->_processors[d]);
-    assert(coarse->_simd_layout[d] == fine->_simd_layout[d]);
-    assert((fine->_rdimensions[d] / coarse->_rdimensions[d])* coarse->_rdimensions[d]==fine->_rdimensions[d]); 
-  }
-}
-
-template<class vobj,int nbasis>
-inline void projectBlockBasis(Lattice<iVector<vComplex,nbasis > > &coarseData,
-			      const             Lattice<vobj>   &fineData,
-			      const std::vector<Lattice<vobj> > &Basis)
-{
-  GridBase * fine  = fineData._grid;
-  GridBase * coarse= coarseData._grid;
-  int  _ndimension = coarse->_ndimension;
-
-  // checks
-  assert( nbasis == Basis.size() );
-  subdivides(coarse,fine); 
-  for(int i=0;i<nbasis;i++){
-    conformable(Basis,fineData);
-  }
-
-  std::vector<int>  block_r      (_ndimension);
-  
-  for(int d=0 ; d<_ndimension;d++){
-    block_r[d] = fine->_rdimensions[d] / coarse->_rdimensions[d];
-  }
-
-  coarseData=zero;
-
-  // Loop with a cache friendly loop ordering
-  for(int sf=0;sf<fine->oSites();sf++){
-
-    int sc;
-    std::vector<int> coor_c(_ndimension);
-    std::vector<int> coor_f(_ndimension);
-    GridBase::CoorFromIndex(coor_f,sf,fine->_rdimensions);
-    for(int d=0;d<_ndimension;d++) coor_c[d]=coor_f[d]/block_r[d];
-    GridBase::IndexFromCoor(coor_c,sc,coarse->_rdimensions);
-
-    for(int i=0;i<nbasis;i++) {
-      
-      coarseData._odata[sc][i]=coarseData._odata[sc][i]
-	+ innerProduct(Basis[i]._odata[sf],fineData._odata[sf]);
-
-    }
-  }
-  return;
-  
-}
-
-
-template<class vobj,int nbasis>
-inline void promoteBlockBasis(const Lattice<iVector<vComplex,nbasis > > &coarseData,
-			                        Lattice<vobj>   &fineData,
-			      const std::vector<Lattice<vobj> > &Basis)
-{
-  GridBase * fine  = fineData._grid;
-  GridBase * coarse= coarseData._grid;
-  int  _ndimension = coarse->_ndimension;
-
-  // checks
-  assert( nbasis == Basis.size() );
-  subdivides(coarse,fine); 
-  for(int i=0;i<nbasis;i++){
-    conformable(Basis,fineData);
-  }
-
-  std::vector<int>  block_r      (_ndimension);
-  
-  for(int d=0 ; d<_ndimension;d++){
-    block_r[d] = fine->_rdimensions[d] / coarse->_rdimensions[d];
-  }
-
-  // Loop with a cache friendly loop ordering
-  for(int sf=0;sf<fine->oSites();sf++){
-
-    int sc;
-    std::vector<int> coor_c(_ndimension);
-    std::vector<int> coor_f(_ndimension);
-
-    GridBase::CoorFromIndex(coor_f,sf,fine->_rdimensions);
-    for(int d=0;d<_ndimension;d++) coor_c[d]=coor_f[d]/block_r[d];
-    GridBase::IndexFromCoor(coor_c,sc,coarse->_rdimensions);
-
-    for(int i=0;i<nbasis;i++) {
-
-      if(i==0) fineData._odata[sf]=                    coarseData._odata[sc][i]*Basis[i]._odata[sf];
-      else     fineData._odata[sf]=fineData._odata[sf]+coarseData._odata[sc][i]*Basis[i]._odata[sf];
-
-    }
-  }
-  return;
-  
-}
-
-// useful in multigrid project;
-// Generic name : Coarsen?
-template<class vobj>
-inline void sumBlocks(Lattice<vobj> &coarseData,const Lattice<vobj> &fineData)
-{
-  GridBase * fine  = fineData._grid;
-  GridBase * coarse= coarseData._grid;
-
-  subdivides(coarse,fine); // require they map
-
-  int _ndimension = coarse->_ndimension;
-  
-  std::vector<int>  block_r      (_ndimension);
-  
-  for(int d=0 ; d<_ndimension;d++){
-    block_r[d] = fine->_rdimensions[d] / coarse->_rdimensions[d];
-  }
-
-  coarseData=zero;
-  for(int sf=0;sf<fine->oSites();sf++){
-    
-    int sc;
-    std::vector<int> coor_c(_ndimension);
-    std::vector<int> coor_f(_ndimension);
-
-    GridBase::CoorFromIndex(coor_f,sf,fine->_rdimensions);
-    for(int d=0;d<_ndimension;d++) coor_c[d]=coor_f[d]/block_r[d];
-    GridBase::IndexFromCoor(coor_c,sc,coarse->_rdimensions);
-
-    coarseData._odata[sc]=coarseData._odata[sc]+fineData._odata[sf];
-
-  }
-  return;
-}
-
-template<class vobj> inline void sliceSum(const Lattice<vobj> &Data,std::vector<typename vobj::scalar_object> &result,int orthogdim)
-{
-  typedef typename vobj::scalar_object sobj;
-
-  GridBase  *grid = Data._grid;
-  const int    Nd = grid->_ndimension;
-  const int Nsimd = grid->Nsimd();
-
-  assert(orthogdim >= 0);
-  assert(orthogdim < Nd);
-
-  int fd=grid->_fdimensions[orthogdim];
-  int ld=grid->_ldimensions[orthogdim];
-  int rd=grid->_rdimensions[orthogdim];
-
-  std::vector<vobj,alignedAllocator<vobj> > lvSum(rd); // will locally sum vectors first
-  std::vector<sobj> lsSum(ld,sobj(zero)); // sum across these down to scalars
-  std::vector<sobj> extracted(Nsimd);     // splitting the SIMD
-
-  result.resize(fd); // And then global sum to return the same vector to every node for IO to file
-  for(int r=0;r<rd;r++){
-    lvSum[r]=zero;
-  }
-
-  std::vector<int>  coor(Nd);  
-
-  // sum over reduced dimension planes, breaking out orthog dir
-  for(int ss=0;ss<grid->oSites();ss++){
-    GridBase::CoorFromIndex(coor,ss,grid->_rdimensions);
-    int r = coor[orthogdim];
-    lvSum[r]=lvSum[r]+Data._odata[ss];
-  }  
-
-  // Sum across simd lanes in the plane, breaking out orthog dir.
-  std::vector<int> icoor(Nd);
-
-  for(int rt=0;rt<rd;rt++){
-
-    extract(lvSum[rt],extracted);
-
-    for(int idx=0;idx<Nsimd;idx++){
-
-      grid->iCoorFromIindex(icoor,idx);
-
-      int ldx =rt+icoor[orthogdim]*rd;
-
-      lsSum[ldx]=lsSum[ldx]+extracted[idx];
-
-    }
-  }
-  
-  // sum over nodes.
-  sobj gsum;
-  for(int t=0;t<fd;t++){
-    int pt = t/ld; // processor plane
-    int lt = t%ld;
-    if ( pt == grid->_processor_coor[orthogdim] ) {
-      gsum=lsSum[lt];
-    } else {
-      gsum=zero;
-    }
-
-    grid->GlobalSum(gsum);
-
-    result[t]=gsum;
-  }
-
-}
-
 }
 #endif
diff --git a/lib/lattice/Grid_lattice_reduction.h b/lib/lattice/Grid_lattice_reduction.h
index b96d0f05..bec436f6 100644
--- a/lib/lattice/Grid_lattice_reduction.h
+++ b/lib/lattice/Grid_lattice_reduction.h
@@ -2,6 +2,9 @@
 #define GRID_LATTICE_REDUCTION_H
 
 namespace Grid {
+#ifdef GRID_WARN_SUBOPTIMAL
+#warning "Optimisation alert all these reduction loops are NOT threaded "
+#endif     
 
     ////////////////////////////////////////////////////////////////////////////////////////////////////
     // Reduction operations
@@ -73,6 +76,79 @@ namespace Grid {
       return ssum;
     }
 
+
+
+
+template<class vobj> inline void sliceSum(const Lattice<vobj> &Data,std::vector<typename vobj::scalar_object> &result,int orthogdim)
+{
+  typedef typename vobj::scalar_object sobj;
+
+  GridBase  *grid = Data._grid;
+  const int    Nd = grid->_ndimension;
+  const int Nsimd = grid->Nsimd();
+
+  assert(orthogdim >= 0);
+  assert(orthogdim < Nd);
+
+  int fd=grid->_fdimensions[orthogdim];
+  int ld=grid->_ldimensions[orthogdim];
+  int rd=grid->_rdimensions[orthogdim];
+
+  std::vector<vobj,alignedAllocator<vobj> > lvSum(rd); // will locally sum vectors first
+  std::vector<sobj> lsSum(ld,sobj(zero)); // sum across these down to scalars
+  std::vector<sobj> extracted(Nsimd);     // splitting the SIMD
+
+  result.resize(fd); // And then global sum to return the same vector to every node for IO to file
+  for(int r=0;r<rd;r++){
+    lvSum[r]=zero;
+  }
+
+  std::vector<int>  coor(Nd);  
+
+  // sum over reduced dimension planes, breaking out orthog dir
+  for(int ss=0;ss<grid->oSites();ss++){
+    GridBase::CoorFromIndex(coor,ss,grid->_rdimensions);
+    int r = coor[orthogdim];
+    lvSum[r]=lvSum[r]+Data._odata[ss];
+  }  
+
+  // Sum across simd lanes in the plane, breaking out orthog dir.
+  std::vector<int> icoor(Nd);
+
+  for(int rt=0;rt<rd;rt++){
+
+    extract(lvSum[rt],extracted);
+
+    for(int idx=0;idx<Nsimd;idx++){
+
+      grid->iCoorFromIindex(icoor,idx);
+
+      int ldx =rt+icoor[orthogdim]*rd;
+
+      lsSum[ldx]=lsSum[ldx]+extracted[idx];
+
+    }
+  }
+  
+  // sum over nodes.
+  sobj gsum;
+  for(int t=0;t<fd;t++){
+    int pt = t/ld; // processor plane
+    int lt = t%ld;
+    if ( pt == grid->_processor_coor[orthogdim] ) {
+      gsum=lsSum[lt];
+    } else {
+      gsum=zero;
+    }
+
+    grid->GlobalSum(gsum);
+
+    result[t]=gsum;
+  }
+
+}
+
+
 }
 #endif
 
diff --git a/lib/lattice/Grid_lattice_transfer.h b/lib/lattice/Grid_lattice_transfer.h
index ecb3016a..53973741 100644
--- a/lib/lattice/Grid_lattice_transfer.h
+++ b/lib/lattice/Grid_lattice_transfer.h
@@ -3,6 +3,20 @@
 
 namespace Grid {
 
+inline void subdivides(GridBase *coarse,GridBase *fine)
+{
+  assert(coarse->_ndimension == fine->_ndimension);
+
+  int _ndimension = coarse->_ndimension;
+
+  // local and global volumes subdivide cleanly after SIMDization
+  for(int d=0;d<_ndimension;d++){
+    assert(coarse->_processors[d]  == fine->_processors[d]);
+    assert(coarse->_simd_layout[d] == fine->_simd_layout[d]);
+    assert((fine->_rdimensions[d] / coarse->_rdimensions[d])* coarse->_rdimensions[d]==fine->_rdimensions[d]); 
+  }
+}
+
   ////////////////////////////////////////////////////////////////////////////////////////////
   // remove and insert a half checkerboard
   ////////////////////////////////////////////////////////////////////////////////////////////
@@ -42,5 +56,131 @@ namespace Grid {
     }
   }
   
+
+template<class vobj,int nbasis>
+inline void projectBlockBasis(Lattice<iVector<vComplex,nbasis > > &coarseData,
+			      const             Lattice<vobj>   &fineData,
+			      const std::vector<Lattice<vobj> > &Basis)
+{
+  GridBase * fine  = fineData._grid;
+  GridBase * coarse= coarseData._grid;
+  int  _ndimension = coarse->_ndimension;
+
+  // checks
+  assert( nbasis == Basis.size() );
+  subdivides(coarse,fine); 
+  for(int i=0;i<nbasis;i++){
+    conformable(Basis,fineData);
+  }
+
+  std::vector<int>  block_r      (_ndimension);
+  
+  for(int d=0 ; d<_ndimension;d++){
+    block_r[d] = fine->_rdimensions[d] / coarse->_rdimensions[d];
+  }
+
+  coarseData=zero;
+
+  // Loop with a cache friendly loop ordering
+  for(int sf=0;sf<fine->oSites();sf++){
+
+    int sc;
+    std::vector<int> coor_c(_ndimension);
+    std::vector<int> coor_f(_ndimension);
+    GridBase::CoorFromIndex(coor_f,sf,fine->_rdimensions);
+    for(int d=0;d<_ndimension;d++) coor_c[d]=coor_f[d]/block_r[d];
+    GridBase::IndexFromCoor(coor_c,sc,coarse->_rdimensions);
+
+    for(int i=0;i<nbasis;i++) {
+      
+      coarseData._odata[sc][i]=coarseData._odata[sc][i]
+	+ innerProduct(Basis[i]._odata[sf],fineData._odata[sf]);
+
+    }
+  }
+  return;
+  
+}
+
+
+template<class vobj,int nbasis>
+inline void promoteBlockBasis(const Lattice<iVector<vComplex,nbasis > > &coarseData,
+			                        Lattice<vobj>   &fineData,
+			      const std::vector<Lattice<vobj> > &Basis)
+{
+  GridBase * fine  = fineData._grid;
+  GridBase * coarse= coarseData._grid;
+  int  _ndimension = coarse->_ndimension;
+
+  // checks
+  assert( nbasis == Basis.size() );
+  subdivides(coarse,fine); 
+  for(int i=0;i<nbasis;i++){
+    conformable(Basis,fineData);
+  }
+
+  std::vector<int>  block_r      (_ndimension);
+  
+  for(int d=0 ; d<_ndimension;d++){
+    block_r[d] = fine->_rdimensions[d] / coarse->_rdimensions[d];
+  }
+
+  // Loop with a cache friendly loop ordering
+  for(int sf=0;sf<fine->oSites();sf++){
+
+    int sc;
+    std::vector<int> coor_c(_ndimension);
+    std::vector<int> coor_f(_ndimension);
+
+    GridBase::CoorFromIndex(coor_f,sf,fine->_rdimensions);
+    for(int d=0;d<_ndimension;d++) coor_c[d]=coor_f[d]/block_r[d];
+    GridBase::IndexFromCoor(coor_c,sc,coarse->_rdimensions);
+
+    for(int i=0;i<nbasis;i++) {
+
+      if(i==0) fineData._odata[sf]=                    coarseData._odata[sc][i]*Basis[i]._odata[sf];
+      else     fineData._odata[sf]=fineData._odata[sf]+coarseData._odata[sc][i]*Basis[i]._odata[sf];
+
+    }
+  }
+  return;
+  
+}
+
+// useful in multigrid project;
+// Generic name : Coarsen?
+template<class vobj>
+inline void sumBlocks(Lattice<vobj> &coarseData,const Lattice<vobj> &fineData)
+{
+  GridBase * fine  = fineData._grid;
+  GridBase * coarse= coarseData._grid;
+
+  subdivides(coarse,fine); // require they map
+
+  int _ndimension = coarse->_ndimension;
+  
+  std::vector<int>  block_r      (_ndimension);
+  
+  for(int d=0 ; d<_ndimension;d++){
+    block_r[d] = fine->_rdimensions[d] / coarse->_rdimensions[d];
+  }
+
+  coarseData=zero;
+  for(int sf=0;sf<fine->oSites();sf++){
+    
+    int sc;
+    std::vector<int> coor_c(_ndimension);
+    std::vector<int> coor_f(_ndimension);
+
+    GridBase::CoorFromIndex(coor_f,sf,fine->_rdimensions);
+    for(int d=0;d<_ndimension;d++) coor_c[d]=coor_f[d]/block_r[d];
+    GridBase::IndexFromCoor(coor_c,sc,coarse->_rdimensions);
+
+    coarseData._odata[sc]=coarseData._odata[sc]+fineData._odata[sf];
+
+  }
+  return;
+}
+
 }
 #endif
diff --git a/lib/qcd/Grid_qcd_dirac.h b/lib/qcd/Grid_qcd_dirac.h
index 86e3870d..ef2a4cd8 100644
--- a/lib/qcd/Grid_qcd_dirac.h
+++ b/lib/qcd/Grid_qcd_dirac.h
@@ -59,6 +59,8 @@ namespace QCD {
      *  0 -i 0  0
      * -i 0  0  0
      */
+  // right multiplication makes sense for matrix args, not for vector since there is 
+  // no concept of row versus columnar indices
     template<class vtype> inline void rmultMinusGammaX(iMatrix<vtype,Ns> &ret,const iMatrix<vtype,Ns> &rhs){
       for(int i=0;i<Ns;i++){
 	ret(i,0) = timesI(rhs(i,3));
@@ -75,19 +77,6 @@ namespace QCD {
 	ret(i,3) = timesI(rhs(i,1));
       }
     };
-
-    template<class vtype> inline void multGammaX(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
-      ret._internal[0] = timesI(rhs._internal[3]);
-      ret._internal[1] = timesI(rhs._internal[2]);
-      ret._internal[2] = timesMinusI(rhs._internal[1]);
-      ret._internal[3] = timesMinusI(rhs._internal[0]);
-    };
-    template<class vtype> inline void multMinusGammaX(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
-	ret(0) = timesMinusI(rhs(3));
-	ret(1) = timesMinusI(rhs(2));
-	ret(2) = timesI(rhs(1));
-	ret(3) = timesI(rhs(0));
-    };
     template<class vtype> inline void multGammaX(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
       for(int i=0;i<Ns;i++){
 	ret(0,i) = timesI(rhs(3,i));
@@ -105,7 +94,18 @@ namespace QCD {
       }
     };
 
-
+    template<class vtype> inline void multGammaX(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+      ret._internal[0] = timesI(rhs._internal[3]);
+      ret._internal[1] = timesI(rhs._internal[2]);
+      ret._internal[2] = timesMinusI(rhs._internal[1]);
+      ret._internal[3] = timesMinusI(rhs._internal[0]);
+    };
+    template<class vtype> inline void multMinusGammaX(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+	ret(0) = timesMinusI(rhs(3));
+	ret(1) = timesMinusI(rhs(2));
+	ret(2) = timesI(rhs(1));
+	ret(3) = timesI(rhs(0));
+    };
 
 
     /*Gy
@@ -114,17 +114,21 @@ namespace QCD {
      *  0 1  0  0
      * -1 0  0  0
      */
-    template<class vtype> inline void multGammaY(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
-      ret(0) = -rhs(3);
-      ret(1) =  rhs(2);
-      ret(2) =  rhs(1);
-      ret(3) = -rhs(0);
+    template<class vtype> inline void rmultGammaY(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(i,0) = -rhs(i,3);
+	ret(i,1) =  rhs(i,2);
+	ret(i,2) =  rhs(i,1);
+	ret(i,3) = -rhs(i,0);
+      }
     };
-    template<class vtype> inline void multMinusGammaY(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
-      ret(0) =  rhs(3);
-      ret(1) = -rhs(2);
-      ret(2) = -rhs(1);
-      ret(3) =  rhs(0);
+    template<class vtype> inline void rmultMinusGammaY(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(i,0) =  rhs(i,3);
+	ret(i,1) = -rhs(i,2);
+	ret(i,2) = -rhs(i,1);
+	ret(i,3) =  rhs(i,0);
+      }
     };
     template<class vtype> inline void multGammaY(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
       for(int i=0;i<Ns;i++){
@@ -142,23 +146,39 @@ namespace QCD {
 	ret(3,i) =  rhs(0,i);
       }
     };
+    template<class vtype> inline void multGammaY(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+      ret(0) = -rhs(3);
+      ret(1) =  rhs(2);
+      ret(2) =  rhs(1);
+      ret(3) = -rhs(0);
+    };
+    template<class vtype> inline void multMinusGammaY(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+      ret(0) =  rhs(3);
+      ret(1) = -rhs(2);
+      ret(2) = -rhs(1);
+      ret(3) =  rhs(0);
+    };
     /*Gz
      *  0 0  i  0   [0]+-i[2]
      *  0 0  0 -i   [1]-+i[3]
      * -i 0  0  0
      *  0 i  0  0
      */
-    template<class vtype> inline void multGammaZ(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
-      ret(0) = timesI(rhs(2));
-      ret(1) =timesMinusI(rhs(3));
-      ret(2) =timesMinusI(rhs(0));
-      ret(3) = timesI(rhs(1));
+    template<class vtype> inline void rmultGammaZ(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(i,0) = timesMinusI(rhs(i,2));
+	ret(i,1) =      timesI(rhs(i,3));
+	ret(i,2) =      timesI(rhs(i,0));
+	ret(i,3) = timesMinusI(rhs(i,1));
+      }
     };
-    template<class vtype> inline void multMinusGammaZ(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
-      ret(0) = timesMinusI(rhs(2));
-      ret(1) = timesI(rhs(3));
-      ret(2) = timesI(rhs(0));
-      ret(3) = timesMinusI(rhs(1));
+    template<class vtype> inline void rmultMinusGammaZ(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(i,0) =      timesI(rhs(i,2));
+	ret(i,1) = timesMinusI(rhs(i,3));
+	ret(i,2) = timesMinusI(rhs(i,0));
+	ret(i,3) =      timesI(rhs(i,1));
+      }
     };
     template<class vtype> inline void multGammaZ(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
       for(int i=0;i<Ns;i++){
@@ -176,23 +196,39 @@ namespace QCD {
 	ret(3,i) = timesMinusI(rhs(1,i));
       }
     };
+    template<class vtype> inline void multGammaZ(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+      ret(0) = timesI(rhs(2));
+      ret(1) =timesMinusI(rhs(3));
+      ret(2) =timesMinusI(rhs(0));
+      ret(3) = timesI(rhs(1));
+    };
+    template<class vtype> inline void multMinusGammaZ(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+      ret(0) = timesMinusI(rhs(2));
+      ret(1) = timesI(rhs(3));
+      ret(2) = timesI(rhs(0));
+      ret(3) = timesMinusI(rhs(1));
+    };
     /*Gt
      *  0 0  1  0 [0]+-[2]
      *  0 0  0  1 [1]+-[3]
      *  1 0  0  0
      *  0 1  0  0
      */
-    template<class vtype> inline void multGammaT(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
-      ret(0) = rhs(2);
-      ret(1) = rhs(3);
-      ret(2) = rhs(0);
-      ret(3) = rhs(1);
+    template<class vtype> inline void rmultGammaT(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(i,0) = rhs(i,2);
+	ret(i,1) = rhs(i,3);
+	ret(i,2) = rhs(i,0);
+	ret(i,3) = rhs(i,1);
+      }
     };
-    template<class vtype> inline void multMinusGammaT(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
-      ret(0) =-rhs(2);
-      ret(1) =-rhs(3);
-      ret(2) =-rhs(0);
-      ret(3) =-rhs(1);
+    template<class vtype> inline void rmultMinusGammaT(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(i,0) =- rhs(i,2);
+	ret(i,1) =- rhs(i,3);
+	ret(i,2) =- rhs(i,0);
+	ret(i,3) =- rhs(i,1);
+      }
     };
     template<class vtype> inline void multGammaT(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
       for(int i=0;i<Ns;i++){
@@ -210,24 +246,41 @@ namespace QCD {
 	ret(3,i) =-rhs(1,i);
       }
     };
+    template<class vtype> inline void multGammaT(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+      ret(0) = rhs(2);
+      ret(1) = rhs(3);
+      ret(2) = rhs(0);
+      ret(3) = rhs(1);
+    };
+    template<class vtype> inline void multMinusGammaT(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+      ret(0) =-rhs(2);
+      ret(1) =-rhs(3);
+      ret(2) =-rhs(0);
+      ret(3) =-rhs(1);
+    };
     /*G5
      *  1 0  0  0 [0]+-[2]
      *  0 1  0  0 [1]+-[3]
      *  0 0 -1  0
      *  0 0  0 -1
      */
-    template<class vtype> inline void multGamma5(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
-      ret(0) = rhs(0);
-      ret(1) = rhs(1);
-      ret(2) =-rhs(2);
-      ret(3) =-rhs(3);
+    template<class vtype> inline void rmultGamma5(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(i,0) = rhs(i,0);
+	ret(i,1) = rhs(i,1);
+	ret(i,2) =-rhs(i,2);
+	ret(i,3) =-rhs(i,3);
+      }
     };
-    template<class vtype> inline void multMinusGamma5(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
-      ret(0) =-rhs(0);
-      ret(1) =-rhs(1);
-      ret(2) = rhs(2);
-      ret(3) = rhs(3);
+    template<class vtype> inline void rmultMinusGamma5(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
+      for(int i=0;i<Ns;i++){
+	ret(i,0) =-rhs(i,0);
+	ret(i,1) =-rhs(i,1);
+	ret(i,2) = rhs(i,2);
+	ret(i,3) = rhs(i,3);
+      }
     };
+
     template<class vtype> inline void multGamma5(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
       for(int i=0;i<Ns;i++){
 	ret(0,i) = rhs(0,i);
@@ -245,6 +298,19 @@ namespace QCD {
       }
     };
 
+    template<class vtype> inline void multGamma5(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+      ret(0) = rhs(0);
+      ret(1) = rhs(1);
+      ret(2) =-rhs(2);
+      ret(3) =-rhs(3);
+    };
+    template<class vtype> inline void multMinusGamma5(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+      ret(0) =-rhs(0);
+      ret(1) =-rhs(1);
+      ret(2) = rhs(2);
+      ret(3) = rhs(3);
+    };
+
 
 
     ///////////////////////////////////////////////////////////////////////////////////////////////////

From 750dd5f5fd8a4f182fe38577538c68e0f9f9e932 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 24 Apr 2015 18:41:34 +0100
Subject: [PATCH 097/429] Cleared the code out from Grid_summation to
 lattice/Grid_lattice_transfer.h

---
 lib/Grid_summation.h | 6 ------
 1 file changed, 6 deletions(-)
 delete mode 100644 lib/Grid_summation.h

diff --git a/lib/Grid_summation.h b/lib/Grid_summation.h
deleted file mode 100644
index 936ca8ef..00000000
--- a/lib/Grid_summation.h
+++ /dev/null
@@ -1,6 +0,0 @@
-#ifndef GRID_SUMMATION_H
-#define GRID_SUMMATION_H
-namespace Grid {
-
-}
-#endif

From f2ac20e7abf42449aa29b9544dab659f93e572e4 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 24 Apr 2015 18:42:44 +0100
Subject: [PATCH 098/429] Removed summation

---
 lib/Grid_lattice.h | 1 -
 1 file changed, 1 deletion(-)

diff --git a/lib/Grid_lattice.h b/lib/Grid_lattice.h
index ed875bc0..7256459c 100644
--- a/lib/Grid_lattice.h
+++ b/lib/Grid_lattice.h
@@ -98,7 +98,6 @@ public:
 #include <lattice/Grid_lattice_coordinate.h>
 #include <lattice/Grid_lattice_rng.h>
 #include <lattice/Grid_lattice_transfer.h>
-#include <Grid_summation.h>
 
 
 

From 71d5927a6635ca4768cce3b12905b9f1b419ee4a Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 24 Apr 2015 19:08:29 +0100
Subject: [PATCH 099/429] Vectors now too and right multiple of matrix with
 gamma

---
 lib/Makefile.am          |  13 ++--
 lib/qcd/Grid_qcd_dirac.h | 156 +++++++++++++++++++++++++++++++++++----
 tests/Grid_gamma.cc      |   8 +-
 3 files changed, 156 insertions(+), 21 deletions(-)

diff --git a/lib/Makefile.am b/lib/Makefile.am
index 20c1f867..9034ce9c 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -4,18 +4,21 @@ AM_CXXFLAGS = -I$(top_srcdir)/
 extra_sources=
 if BUILD_COMMS_MPI
   extra_sources+=communicator/Grid_communicator_mpi.cc
-  extra_sources+=stencil/Grid_stencil_common.cc
 endif
+
 if BUILD_COMMS_NONE
   extra_sources+=communicator/Grid_communicator_fake.cc
-  extra_sources+=stencil/Grid_stencil_common.cc
 endif
 
 #
 # Libraries
 #
 lib_LIBRARIES = libGrid.a
-libGrid_a_SOURCES = Grid_init.cc $(extra_sources)
+libGrid_a_SOURCES =\
+	 Grid_init.cc\
+	stencil/Grid_stencil_common.cc\
+	qcd/Grid_qcd_dirac.cc\
+	$(extra_sources)
 
 #
 # Include files
@@ -32,7 +35,6 @@ include_HEADERS =\
 	Grid_math.h\
 	Grid_simd.h\
 	Grid_stencil.h\
-	Grid_summation.h\
 	Grid_where.h
 
 nobase_include_HEADERS=\
@@ -68,7 +70,8 @@ nobase_include_HEADERS=\
 	math/Grid_math_trace.h\
 	math/Grid_math_traits.h\
 	math/Grid_math_transpose.h\
-	qcd/Grid_QCD.h\
+	qcd/Grid_qcd.h\
+	qcd/Grid_qcd_dirac.h\
 	simd/Grid_vComplexD.h\
 	simd/Grid_vComplexF.h\
 	simd/Grid_vInteger.h\
diff --git a/lib/qcd/Grid_qcd_dirac.h b/lib/qcd/Grid_qcd_dirac.h
index ef2a4cd8..945178e9 100644
--- a/lib/qcd/Grid_qcd_dirac.h
+++ b/lib/qcd/Grid_qcd_dirac.h
@@ -44,12 +44,14 @@ namespace QCD {
       //      MinusSigmaYT,
       //      MinusSigmaZT
     };
+    
+    static GammaMatrix GammaMatrices[];
+    static const char *GammaMatrixNames[];
 
     Gamma (GammaMatrix g) { _g=g; }
 
     GammaMatrix _g;
 
-
   };
 
 
@@ -94,13 +96,13 @@ namespace QCD {
       }
     };
 
-    template<class vtype> inline void multGammaX(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+    template<class vtype> inline void multGammaX(iVector<vtype,Ns> &ret, const iVector<vtype,Ns> &rhs){
       ret._internal[0] = timesI(rhs._internal[3]);
       ret._internal[1] = timesI(rhs._internal[2]);
       ret._internal[2] = timesMinusI(rhs._internal[1]);
       ret._internal[3] = timesMinusI(rhs._internal[0]);
     };
-    template<class vtype> inline void multMinusGammaX(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+    template<class vtype> inline void multMinusGammaX(iVector<vtype,Ns> &ret, const iVector<vtype,Ns> &rhs){
 	ret(0) = timesMinusI(rhs(3));
 	ret(1) = timesMinusI(rhs(2));
 	ret(2) = timesI(rhs(1));
@@ -146,13 +148,13 @@ namespace QCD {
 	ret(3,i) =  rhs(0,i);
       }
     };
-    template<class vtype> inline void multGammaY(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+    template<class vtype> inline void multGammaY(iVector<vtype,Ns> &ret, const iVector<vtype,Ns> &rhs){
       ret(0) = -rhs(3);
       ret(1) =  rhs(2);
       ret(2) =  rhs(1);
       ret(3) = -rhs(0);
     };
-    template<class vtype> inline void multMinusGammaY(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+    template<class vtype> inline void multMinusGammaY(iVector<vtype,Ns> &ret, const iVector<vtype,Ns> &rhs){
       ret(0) =  rhs(3);
       ret(1) = -rhs(2);
       ret(2) = -rhs(1);
@@ -196,13 +198,13 @@ namespace QCD {
 	ret(3,i) = timesMinusI(rhs(1,i));
       }
     };
-    template<class vtype> inline void multGammaZ(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+    template<class vtype> inline void multGammaZ(iVector<vtype,Ns> &ret, const iVector<vtype,Ns> &rhs){
       ret(0) = timesI(rhs(2));
       ret(1) =timesMinusI(rhs(3));
       ret(2) =timesMinusI(rhs(0));
       ret(3) = timesI(rhs(1));
     };
-    template<class vtype> inline void multMinusGammaZ(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+    template<class vtype> inline void multMinusGammaZ(iVector<vtype,Ns> &ret, const iVector<vtype,Ns> &rhs){
       ret(0) = timesMinusI(rhs(2));
       ret(1) = timesI(rhs(3));
       ret(2) = timesI(rhs(0));
@@ -246,13 +248,13 @@ namespace QCD {
 	ret(3,i) =-rhs(1,i);
       }
     };
-    template<class vtype> inline void multGammaT(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+    template<class vtype> inline void multGammaT(iVector<vtype,Ns> &ret, const iVector<vtype,Ns> &rhs){
       ret(0) = rhs(2);
       ret(1) = rhs(3);
       ret(2) = rhs(0);
       ret(3) = rhs(1);
     };
-    template<class vtype> inline void multMinusGammaT(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+    template<class vtype> inline void multMinusGammaT(iVector<vtype,Ns> &ret, const iVector<vtype,Ns> &rhs){
       ret(0) =-rhs(2);
       ret(1) =-rhs(3);
       ret(2) =-rhs(0);
@@ -298,13 +300,13 @@ namespace QCD {
       }
     };
 
-    template<class vtype> inline void multGamma5(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+    template<class vtype> inline void multGamma5(iVector<vtype,Ns> &ret, const iVector<vtype,Ns> &rhs){
       ret(0) = rhs(0);
       ret(1) = rhs(1);
       ret(2) =-rhs(2);
       ret(3) =-rhs(3);
     };
-    template<class vtype> inline void multMinusGamma5(iVector<vtype,Ns> &ret, iVector<vtype,Ns> &rhs){
+    template<class vtype> inline void multMinusGamma5(iVector<vtype,Ns> &ret, const iVector<vtype,Ns> &rhs){
       ret(0) =-rhs(0);
       ret(1) =-rhs(1);
       ret(2) = rhs(2);
@@ -313,6 +315,10 @@ namespace QCD {
 
 
 
+#ifdef GRID_WARN_SUBOPTIMAL
+#warning "Optimisation alert switch over to multGammaX early "
+#endif     
+
     ///////////////////////////////////////////////////////////////////////////////////////////////////
     // Operator * : first case this is not a spin index, so recurse
     ///////////////////////////////////////////////////////////////////////////////////////////////////
@@ -323,10 +329,8 @@ namespace QCD {
     // note that doing so from the lattice operator will avoid copy back and case switch overhead, as
     // was done for the tensor math operator to remove operator * notation early
     //
-#ifdef GRID_WARN_SUBOPTIMAL
-#warning "Optimisation alert switch over to multGammaX early "
-#endif     
 
+    //left multiply
     template<class vtype> inline auto operator * ( const Gamma &G,const iScalar<vtype> &arg) ->
       typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type 
 
@@ -349,6 +353,32 @@ namespace QCD {
 	  ret._internal=G*arg._internal;
 	  return ret;
 	}
+
+
+    //right multiply
+    template<class vtype> inline auto operator * (const iScalar<vtype> &arg, const Gamma &G) ->
+      typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type 
+
+	{
+	  iScalar<vtype> ret;
+	  ret._internal=arg._internal*G;
+	  return ret;
+	}
+    template<class vtype,int N> inline auto operator * (const iVector<vtype,N> &arg, const Gamma &G) ->
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type 
+	{
+	  iVector<vtype,N> ret;
+	  ret._internal=arg._internal*G;
+	  return ret;
+	}
+    template<class vtype,int N> inline auto operator * (const iMatrix<vtype,N> &arg, const Gamma &G) ->
+      typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type 
+	{
+	  iMatrix<vtype,N> ret;
+	  ret._internal=arg._internal*G;
+	  return ret;
+	}
+
     ////////////////////////////////////////////////////////
     // When we hit the spin index this matches and we stop
     ////////////////////////////////////////////////////////
@@ -399,6 +429,104 @@ namespace QCD {
 	}
 	return ret;
       }
+    // Could have used type trait for Matrix/vector and then an enable if to share code
+    template<class vtype> inline auto operator * ( const Gamma &G,const iVector<vtype,Ns> &arg) ->
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type 
+      {
+	iVector<vtype,Ns> ret;
+	switch (G._g) {
+	case Gamma::Identity:
+	  ret = arg;
+	  break;
+	case Gamma::MinusIdentity:
+	  ret = -arg;
+	  break;
+	case Gamma::GammaX:
+	  multGammaX(ret,arg);
+	  break;
+	case Gamma::MinusGammaX:
+	  multMinusGammaX(ret,arg);
+	  break;
+	case Gamma::GammaY:
+	  multGammaY(ret,arg);
+	  break;
+	case Gamma::MinusGammaY:
+	  multMinusGammaY(ret,arg);
+	  break;
+	case Gamma::GammaZ:
+	  multGammaZ(ret,arg);
+	  break;
+	case Gamma::MinusGammaZ:
+	  multMinusGammaZ(ret,arg);
+	  break;
+	case Gamma::GammaT:
+	  multGammaT(ret,arg);
+	  break;
+	case Gamma::MinusGammaT:
+	  multMinusGammaT(ret,arg);
+	  break;
+	case Gamma::Gamma5:
+	  multGamma5(ret,arg);
+	  break;
+	case Gamma::MinusGamma5:
+	  multMinusGamma5(ret,arg);
+	  break;
+	default:
+	  assert(0);
+	  break;
+	}
+	return ret;
+      }
+
+    template<class vtype> inline auto operator * (const iMatrix<vtype,Ns> &arg, const Gamma &G) ->
+      typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,Ns>,SpinIndex>::value,iMatrix<vtype,Ns> >::type 
+      {
+	iMatrix<vtype,Ns> ret;
+	switch (G._g) {
+	case Gamma::Identity:
+	  ret = arg;
+	  break;
+	case Gamma::MinusIdentity:
+	  ret = -arg;
+	  break;
+	case Gamma::GammaX:
+	  rmultGammaX(ret,arg);
+	  break;
+	case Gamma::MinusGammaX:
+	  rmultMinusGammaX(ret,arg);
+	  break;
+	case Gamma::GammaY:
+	  rmultGammaY(ret,arg);
+	  break;
+	case Gamma::MinusGammaY:
+	  rmultMinusGammaY(ret,arg);
+	  break;
+	case Gamma::GammaZ:
+	  rmultGammaZ(ret,arg);
+	  break;
+	case Gamma::MinusGammaZ:
+	  rmultMinusGammaZ(ret,arg);
+	  break;
+	case Gamma::GammaT:
+	  rmultGammaT(ret,arg);
+	  break;
+	case Gamma::MinusGammaT:
+	  rmultMinusGammaT(ret,arg);
+	  break;
+	case Gamma::Gamma5:
+	  rmultGamma5(ret,arg);
+	  break;
+	case Gamma::MinusGamma5:
+	  rmultMinusGamma5(ret,arg);
+	  break;
+	default:
+	  assert(0);
+	  break;
+	}
+	return ret;
+      }
+
+
     /* Output from test
 ./Grid_gamma 
 Identity((1,0),(0,0),(0,0),(0,0))
diff --git a/tests/Grid_gamma.cc b/tests/Grid_gamma.cc
index 9c213352..d3a33039 100644
--- a/tests/Grid_gamma.cc
+++ b/tests/Grid_gamma.cc
@@ -23,6 +23,9 @@ int main (int argc, char ** argv)
   SpinMatrix rr=zero;
   SpinMatrix result;
 
+  SpinVector lv=zero;
+  SpinVector rv=zero;
+
   for(int a=0;a<Ns;a++){
     ident()(a,a) = 1.0;
   }
@@ -56,8 +59,9 @@ int main (int argc, char ** argv)
     "-Gamma5  ",
     "         "
   };
-  //  result == ll*Gamma(g[0])*rr;
-  //  result == ll*Gamma(g[0]);
+  result =ll*Gamma(g[0])*rr;
+  result =ll*Gamma(g[0]);
+  rv = Gamma(g[0])*lv;
 
   for(int mu=0;mu<12;mu++){
 

From 2a67214f9de3b81ef3e038f46bc0d1960385d619 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 24 Apr 2015 19:12:14 +0100
Subject: [PATCH 100/429] static names and enum list

---
 lib/qcd/Grid_qcd_dirac.cc | 39 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 39 insertions(+)
 create mode 100644 lib/qcd/Grid_qcd_dirac.cc

diff --git a/lib/qcd/Grid_qcd_dirac.cc b/lib/qcd/Grid_qcd_dirac.cc
new file mode 100644
index 00000000..6526dfd3
--- /dev/null
+++ b/lib/qcd/Grid_qcd_dirac.cc
@@ -0,0 +1,39 @@
+#include <Grid.h>
+
+namespace Grid {
+
+  namespace QCD {
+
+    Gamma::GammaMatrix  Gamma::GammaMatrices [] = {
+      Gamma::Identity,
+      Gamma::GammaX,
+      Gamma::GammaY,
+      Gamma::GammaZ,
+      Gamma::GammaT,
+      Gamma::Gamma5,
+      Gamma::MinusIdentity,
+      Gamma::MinusGammaX,
+      Gamma::MinusGammaY,
+      Gamma::MinusGammaZ,
+      Gamma::MinusGammaT,
+      Gamma::MinusGamma5
+    };
+    const char *Gamma::GammaMatrixNames[] = { 
+      "Identity ",
+      "GammaX   ",
+      "GammaY   ",
+      "GammaZ   ",
+      "GammaT   ",
+      "Gamma5   ",
+      "-Identity",
+      "-GammaX  ",
+      "-GammaY  ",
+      "-GammaZ  ",
+      "-GammaT  ",
+      "-Gamma5  ",
+      "         "
+    };
+
+
+  }
+}

From fc32450360c24a7be9d5daeda257e04be9b61e71 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 24 Apr 2015 20:21:40 +0100
Subject: [PATCH 101/429] Improved the gamma quite a bit. Serial rng's which
 are set on node zero and broadcaste

---
 lib/Grid_communicator.h                    |   2 +
 lib/communicator/Grid_communicator_fake.cc |   3 +
 lib/communicator/Grid_communicator_mpi.cc  |   9 ++
 lib/lattice/Grid_lattice_rng.h             | 107 +++++++++++++++++----
 lib/math/Grid_math_inner.h                 |   7 ++
 lib/qcd/Grid_qcd_dirac.h                   |   8 +-
 tests/Grid_cshift.cc                       |   3 +-
 tests/Grid_gamma.cc                        |  33 ++++++-
 tests/Grid_main.cc                         |   3 +-
 tests/Grid_nersc_io.cc                     |   1 -
 tests/Grid_stencil.cc                      |   3 +-
 11 files changed, 151 insertions(+), 28 deletions(-)

diff --git a/lib/Grid_communicator.h b/lib/Grid_communicator.h
index f5baa413..84d080ae 100644
--- a/lib/Grid_communicator.h
+++ b/lib/Grid_communicator.h
@@ -99,6 +99,8 @@ class CartesianCommunicator {
       Broadcast(root,(void *)&data,sizeof(data));
     };
 
+    static void BroadcastWorld(int root,void* data, int bytes);
+
 }; 
 }
 
diff --git a/lib/communicator/Grid_communicator_fake.cc b/lib/communicator/Grid_communicator_fake.cc
index 5f4a4024..c5b77620 100644
--- a/lib/communicator/Grid_communicator_fake.cc
+++ b/lib/communicator/Grid_communicator_fake.cc
@@ -37,6 +37,9 @@ void CartesianCommunicator::Barrier(void)
 void CartesianCommunicator::Broadcast(int root,void* data, int bytes)
 {
 }
+void CartesianCommunicator::BroadcastWorld(int root,void* data, int bytes)
+{
+}
 
 
 void CartesianCommunicator::ShiftedRanks(int dim,int shift,int &source,int &dest)
diff --git a/lib/communicator/Grid_communicator_mpi.cc b/lib/communicator/Grid_communicator_mpi.cc
index bd53bd9c..6a3f69f2 100644
--- a/lib/communicator/Grid_communicator_mpi.cc
+++ b/lib/communicator/Grid_communicator_mpi.cc
@@ -92,5 +92,14 @@ void CartesianCommunicator::Broadcast(int root,void* data, int bytes)
 	    communicator);
 }
 
+void CartesianCommunicator::BroadcastWorld(int root,void* data, int bytes)
+{
+  MPI_Bcast(data,
+	    bytes,
+	    MPI_BYTE,
+	    root,
+	    MPI_COMM_WORLD);
+}
+
 }
 
diff --git a/lib/lattice/Grid_lattice_rng.h b/lib/lattice/Grid_lattice_rng.h
index 24627c92..85d8c471 100644
--- a/lib/lattice/Grid_lattice_rng.h
+++ b/lib/lattice/Grid_lattice_rng.h
@@ -26,7 +26,70 @@ namespace Grid {
     }
 
   };
-  class GridRNG {
+  class GridRNGbase {
+
+  public:
+
+   GridRNGbase() : _uniform{0,1}, _gaussian(0.0,1.0) {};
+
+    int _seeded;
+    // One generator per site.
+    // Uniform and Gaussian distributions from these generators.
+    std::vector<std::ranlux48>             _generators;
+    std::uniform_real_distribution<double> _uniform;
+    std::normal_distribution<double>       _gaussian;
+
+
+  };
+
+  class GridSerialRNG : public GridRNGbase {
+  public:
+
+    // FIXME ... do we require lockstep draws of randoms 
+    // from all nodes keeping seeds consistent.
+    // place a barrier/broadcast in the fill routine
+    template<class source> void Seed(source &src)
+    {
+      typename source::result_type init = src();
+      CartesianCommunicator::BroadcastWorld(0,(void *)&init,sizeof(init));
+      _generators[0] = std::ranlux48(init);
+      _seeded=1;
+    }    
+
+    GridSerialRNG() : GridRNGbase() {
+      _generators.resize(1);
+      _seeded=0;
+    }
+
+
+    template <class sobj,class distribution> inline void fill(sobj &l,distribution &dist){
+
+      typedef typename sobj::scalar_type scalar_type;
+ 
+      int words = sizeof(sobj)/sizeof(scalar_type);
+
+      scalar_type *buf = (scalar_type *) & l;
+
+      for(int idx=0;idx<words;idx++){
+	buf[idx] = dist(_generators[0]);
+      }
+      
+      CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
+
+    };
+
+    void SeedRandomDevice(void){
+      std::random_device rd;
+      Seed(rd);
+    }
+    void SeedFixedIntegers(std::vector<int> &seeds){
+      fixedSeed src(seeds);
+      Seed(src);
+    }
+
+  };
+
+  class GridParallelRNG : public GridRNGbase {
   public:
     // One generator per site.
     std::vector<std::ranlux48>             _generators;
@@ -42,25 +105,13 @@ namespace Grid {
       return is*_grid->oSites()+os;
     }
 
-    GridRNG(GridBase *grid) : _uniform{0,1}, _gaussian(0.0,1.0) {
+    GridParallelRNG(GridBase *grid) : GridRNGbase() {
       _grid=grid;
       _vol =_grid->iSites()*_grid->oSites();
       _generators.resize(_vol);
-      //      SeedFixedIntegers(seeds);
-      // worst case we seed properly but non-deterministically
-      SeedRandomDevice();
+      _seeded=0;
     }
 
-    // FIXME: drive seeding from node zero and transmit to all
-    // to get unique randoms on each node
-    void SeedRandomDevice(void){
-      std::random_device rd;
-      Seed(rd);
-    }
-    void SeedFixedIntegers(std::vector<int> &seeds){
-      fixedSeed src(seeds);
-      Seed(src);
-    }
 
     // This loop could be made faster to avoid the Ahmdahl by
     // i)  seed generators on each timeslice, for x=y=z=0;
@@ -86,6 +137,7 @@ namespace Grid {
 	  _generators[l_idx] = std::ranlux48(init);
 	}
       }
+      _seeded=1;
     }    
 
     //FIXME implement generic IO and create state save/restore
@@ -122,15 +174,34 @@ namespace Grid {
 	merge(l._odata[ss],pointers);
       }
     };
+
+    void SeedRandomDevice(void){
+      std::random_device rd;
+      Seed(rd);
+    }
+    void SeedFixedIntegers(std::vector<int> &seeds){
+      fixedSeed src(seeds);
+      Seed(src);
+    }
+
   };
 
-  // FIXME Implement a consistent seed management strategy
-  template <class vobj> inline void random(GridRNG &rng,Lattice<vobj> &l){
+
+  template <class vobj> inline void random(GridParallelRNG &rng,Lattice<vobj> &l){
     rng.fill(l,rng._uniform);
   }
 
-  template <class vobj> inline void gaussian(GridRNG &rng,Lattice<vobj> &l){
+  template <class vobj> inline void gaussian(GridParallelRNG &rng,Lattice<vobj> &l){
     rng.fill(l,rng._gaussian);
   }
+
+
+  template <class sobj> inline void random(GridSerialRNG &rng,sobj &l){
+    rng.fill(l,rng._uniform);
+  }
+  template <class sobj> inline void gaussian(GridSerialRNG &rng,sobj &l){
+    rng.fill(l,rng._gaussian);
+  }
+
 }
 #endif
diff --git a/lib/math/Grid_math_inner.h b/lib/math/Grid_math_inner.h
index 88387db7..465501e1 100644
--- a/lib/math/Grid_math_inner.h
+++ b/lib/math/Grid_math_inner.h
@@ -6,6 +6,13 @@ namespace Grid {
     // innerProduct Vector x Vector -> Scalar
     // innerProduct Matrix x Matrix -> Scalar
     ///////////////////////////////////////////////////////////////////////////////////////
+    template<class sobj> inline RealD norm2l(sobj &arg){
+      typedef typename sobj::scalar_type scalar;
+      decltype(innerProduct(arg,arg)) nrm;
+      nrm = innerProduct(arg,arg);
+      return real(nrm);
+    }
+
     template<class l,class r,int N> inline
     auto innerProduct (const iVector<l,N>& lhs,const iVector<r,N>& rhs) -> iScalar<decltype(innerProduct(lhs._internal[0],rhs._internal[0]))>
     {
diff --git a/lib/qcd/Grid_qcd_dirac.h b/lib/qcd/Grid_qcd_dirac.h
index 945178e9..03dc9cc5 100644
--- a/lib/qcd/Grid_qcd_dirac.h
+++ b/lib/qcd/Grid_qcd_dirac.h
@@ -76,15 +76,15 @@ namespace QCD {
 	ret(i,0) = timesMinusI(rhs(i,3));
 	ret(i,1) = timesMinusI(rhs(i,2));
 	ret(i,2) = timesI(rhs(i,1));
-	ret(i,3) = timesI(rhs(i,1));
+	ret(i,3) = timesI(rhs(i,0));
       }
     };
     template<class vtype> inline void multGammaX(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
       for(int i=0;i<Ns;i++){
 	ret(0,i) = timesI(rhs(3,i));
-	ret(1,i) = timesI(rhs._internal[2][i]);
-	ret(2,i) = timesMinusI(rhs._internal[1][i]);
-	ret(3,i) = timesMinusI(rhs._internal[0][i]);
+	ret(1,i) = timesI(rhs(2,i));
+	ret(2,i) = timesMinusI(rhs(1,i));
+	ret(3,i) = timesMinusI(rhs(0,i));
       }
     };
     template<class vtype> inline void multMinusGammaX(iMatrix<vtype,Ns> &ret, const iMatrix<vtype,Ns> &rhs){
diff --git a/tests/Grid_cshift.cc b/tests/Grid_cshift.cc
index 3fe2b9cc..995756d6 100644
--- a/tests/Grid_cshift.cc
+++ b/tests/Grid_cshift.cc
@@ -14,7 +14,8 @@ int main (int argc, char ** argv)
   std::vector<int> latt_size  ({8,8,8,16});
     
   GridCartesian     Fine(latt_size,simd_layout,mpi_layout);
-  GridRNG           FineRNG(&Fine);
+  GridParallelRNG           FineRNG(&Fine);
+  FineRNG.SeedRandomDevice();
 
   LatticeComplex U(&Fine);
   LatticeComplex ShiftU(&Fine);
diff --git a/tests/Grid_gamma.cc b/tests/Grid_gamma.cc
index d3a33039..758f9faa 100644
--- a/tests/Grid_gamma.cc
+++ b/tests/Grid_gamma.cc
@@ -16,9 +16,15 @@ int main (int argc, char ** argv)
     
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
 
-  GridRNG           RNG(&Grid);
+  GridParallelRNG          pRNG(&Grid);
+  pRNG.SeedRandomDevice();
+
+  GridSerialRNG            sRNG;
+  sRNG.SeedRandomDevice();
 
   SpinMatrix ident=zero;
+  SpinMatrix rnd  ; random(sRNG,rnd);
+
   SpinMatrix ll=zero;
   SpinMatrix rr=zero;
   SpinMatrix result;
@@ -94,6 +100,29 @@ int main (int argc, char ** argv)
 
   }
 
-  
+  std::cout << "Testing Gamma^2 - 1 = 0"<<std::endl;
+  for(int mu=0;mu<6;mu++){
+    result =  Gamma(g[mu])* ident * Gamma(g[mu]);
+    result = result - ident;
+    double mag = TensorRemove(norm2l(result));
+    std::cout << list[mu]<<" " << mag<<std::endl;
+  }
+
+  std::cout << "Testing (MinusGamma + G )M = 0"<<std::endl;
+  for(int mu=0;mu<6;mu++){
+    result =          rnd * Gamma(g[mu]);
+    result = result + rnd * Gamma(g[mu+6]);
+    double mag = TensorRemove(norm2l(result));
+    std::cout << list[mu]<<" " << mag<<std::endl;
+  }
+
+  std::cout << "Testing M(MinusGamma + G )  = 0"<<std::endl;
+  for(int mu=0;mu<6;mu++){
+    result =           Gamma(g[mu])  *rnd;
+    result = result +  Gamma(g[mu+6])*rnd;
+    double mag = TensorRemove(norm2l(result));
+    std::cout << list[mu]<<" " << mag<<std::endl;
+  }
+
   Grid_finalize();
 }
diff --git a/tests/Grid_main.cc b/tests/Grid_main.cc
index 83cacf92..af13836e 100644
--- a/tests/Grid_main.cc
+++ b/tests/Grid_main.cc
@@ -57,7 +57,8 @@ int main (int argc, char ** argv)
     
     GridCartesian Fine(latt_size,simd_layout,mpi_layout);
     GridRedBlackCartesian rbFine(latt_size,simd_layout,mpi_layout);
-    GridRNG       FineRNG(&Fine);
+    GridParallelRNG       FineRNG(&Fine);
+    FineRNG.SeedRandomDevice();
 
     LatticeColourMatrix Foo(&Fine);
     LatticeColourMatrix Bar(&Fine);
diff --git a/tests/Grid_nersc_io.cc b/tests/Grid_nersc_io.cc
index f9faf7ef..69c7f84f 100644
--- a/tests/Grid_nersc_io.cc
+++ b/tests/Grid_nersc_io.cc
@@ -20,7 +20,6 @@ int main (int argc, char ** argv)
   GridCartesian     Fine(latt_size,simd_layout,mpi_layout);
   GridCartesian     Coarse(clatt_size,simd_layout,mpi_layout);
 
-  GridRNG           FineRNG(&Fine);
   LatticeGaugeField Umu(&Fine);
 
   std::vector<LatticeColourMatrix> U(4,&Fine);
diff --git a/tests/Grid_stencil.cc b/tests/Grid_stencil.cc
index 917cb558..f35581f6 100644
--- a/tests/Grid_stencil.cc
+++ b/tests/Grid_stencil.cc
@@ -48,7 +48,8 @@ int main (int argc, char ** argv)
     
     GridCartesian Fine(latt_size,simd_layout,mpi_layout);
     GridRedBlackCartesian rbFine(latt_size,simd_layout,mpi_layout);
-    GridRNG       fRNG(&Fine);
+    GridParallelRNG       fRNG(&Fine);
+    fRNG.SeedRandomDevice();
 
     LatticeColourMatrix Foo(&Fine);
     LatticeColourMatrix Bar(&Fine);

From dc970c644210a30e0ad17d3d0834005398f61ee2 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 24 Apr 2015 22:56:37 +0100
Subject: [PATCH 102/429] Dirac done ; remove from TODO

---
 TODO | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/TODO b/TODO
index 12e33302..d2f8423c 100644
--- a/TODO
+++ b/TODO
@@ -17,7 +17,7 @@
 
 * Make the Tensor types and Complex etc... play more nicely.
 
-* Dirac, Pauli, SU subgroup, etc.. * Gamma/Dirac structures
+* Pauli, SU subgroup, etc.. 
 
 * Fourspin, two spin project
 
@@ -73,6 +73,7 @@ AUDITS:
 ======================================================================================================
 
 FUNCTIONALITY:
+* Dirac Gamma/Dirac structures ---- DONE
 * Conditional execution, where etc...                -----DONE, simple test
 * Integer relational support                         -----DONE
 * Coordinate information, integers etc...            -----DONE

From 2d8cf9e456964d5c197a50acfdb1c4ec23422e81 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 25 Apr 2015 12:54:06 +0100
Subject: [PATCH 103/429] Added two spinor functionality required to support
 the Wilson hopping term.

---
 TODO                       |   22 +-
 lib/math/Grid_math_inner.h |    2 +-
 lib/qcd/Grid_qcd.h         |   68 ++-
 lib/qcd/Grid_qcd_2spinor.h | 1082 ++++++++++++++++++++++++++++++++++++
 tests/Grid_gamma.cc        |  112 ++--
 5 files changed, 1227 insertions(+), 59 deletions(-)
 create mode 100644 lib/qcd/Grid_qcd_2spinor.h

diff --git a/TODO b/TODO
index d2f8423c..b12502eb 100644
--- a/TODO
+++ b/TODO
@@ -13,7 +13,26 @@
 * Consider switch std::vector to boost arrays or something lighter weight
   boost::multi_array<type, 3> A()...    to replace multi1d, multi2d etc..
 
-* How to define simple matrix operations, such as flavour matrices?
+* TensorRemove is a hack, come up with a long term rationalised approach to Complex vs. Scalar<Scalar<Scalar<Complex > > >
+  QDP forces use of "toDouble" to get back to non tensor scalar. This role is presently taken TensorRemove, but I
+  want to introduce a syntax that does not require this.
+
+* norm2l is a hack. figure out syntax error and make this norm2. c.f. tests/Grid_gamma.cc
+
+*  std::vector replacement;
+
+  Had to change "reserve" back to "resize" on std::vector in Lattice class.
+
+  This forces the constructor call on EVERY element of the array with negative
+  performance effects on temporaries.
+  
+  The reversion was required because copy constructur has to work.
+
+  CONCLUSION: I must implement a similar to vector without construction/fill on
+  resize. Find out if valarray or alternative works differently prior to 
+  doing this since there may still be something I can use..
+
+* Flavour matrices?
 
 * Make the Tensor types and Complex etc... play more nicely.
 
@@ -36,7 +55,6 @@
 
 * Conformable test in Cshift routines.
 
-* QDP++ regression suite and comparative benchmark
 
 AUDITS:
 
diff --git a/lib/math/Grid_math_inner.h b/lib/math/Grid_math_inner.h
index 465501e1..71b5e3b8 100644
--- a/lib/math/Grid_math_inner.h
+++ b/lib/math/Grid_math_inner.h
@@ -6,7 +6,7 @@ namespace Grid {
     // innerProduct Vector x Vector -> Scalar
     // innerProduct Matrix x Matrix -> Scalar
     ///////////////////////////////////////////////////////////////////////////////////////
-    template<class sobj> inline RealD norm2l(sobj &arg){
+    template<class sobj> inline RealD norm2l(const sobj &arg){
       typedef typename sobj::scalar_type scalar;
       decltype(innerProduct(arg,arg)) nrm;
       nrm = innerProduct(arg,arg);
diff --git a/lib/qcd/Grid_qcd.h b/lib/qcd/Grid_qcd.h
index a5c4320e..f331bd88 100644
--- a/lib/qcd/Grid_qcd.h
+++ b/lib/qcd/Grid_qcd.h
@@ -7,6 +7,7 @@ namespace QCD {
     static const int Nc=3;
     static const int Ns=4;
     static const int Nd=4;
+    static const int Nhs=2; // half spinor
 
     static const int CbRed  =0;
     static const int CbBlack=1;
@@ -38,16 +39,21 @@ namespace QCD {
     template<typename vtype> using iColourVector     = iScalar<iScalar<iVector<vtype, Nc> > >;
     template<typename vtype> using iSpinColourVector = iScalar<iVector<iVector<vtype, Nc>, Ns> >;
 
-    typedef iSpinMatrix<Complex >          SpinMatrix;
-    typedef iColourMatrix<Complex >        ColourMatrix;
-    typedef iSpinColourMatrix<Complex >    SpinColourMatrix;
-    typedef iLorentzColourMatrix<Complex > LorentzColourMatrix;
+    template<typename vtype> using iHalfSpinVector       = iScalar<iVector<iScalar<vtype>, Nhs> >;
+    template<typename vtype> using iHalfSpinColourVector = iScalar<iVector<iVector<vtype, Nc>, Nhs> >;
+
+    typedef iSpinMatrix<Complex >           SpinMatrix;
+    typedef iColourMatrix<Complex >         ColourMatrix;
+    typedef iSpinColourMatrix<Complex >     SpinColourMatrix;
+    typedef iLorentzColourMatrix<Complex >  LorentzColourMatrix;
     typedef iLorentzColourMatrix<ComplexF > LorentzColourMatrixF;
     typedef iLorentzColourMatrix<ComplexD > LorentzColourMatrixD;
 
-    typedef iSpinVector<Complex >       SpinVector;
-    typedef iColourVector<Complex >     ColourVector;
-    typedef iSpinColourVector<Complex > SpinColourVector;
+    typedef iSpinVector<Complex >           SpinVector;
+    typedef iColourVector<Complex >         ColourVector;
+    typedef iSpinColourVector<Complex >     SpinColourVector;
+    typedef iHalfSpinVector<Complex >       HalfSpinVector;
+    typedef iHalfSpinColourVector<Complex > HalfSpinColourVector;
 
     
     typedef iSpinMatrix<vComplex >          vSpinMatrix;
@@ -55,36 +61,46 @@ namespace QCD {
     typedef iSpinColourMatrix<vComplex >    vSpinColourMatrix;
     typedef iLorentzColourMatrix<vComplex > vLorentzColourMatrix;
     
-    typedef iSpinVector<vComplex >       vSpinVector;
-    typedef iColourVector<vComplex >     vColourVector;
-    typedef iSpinColourVector<vComplex > vSpinColourVector;
+    typedef iSpinVector<vComplex >           vSpinVector;
+    typedef iColourVector<vComplex >         vColourVector;
+    typedef iSpinColourVector<vComplex >     vSpinColourVector;
+    typedef iHalfSpinVector<vComplex >       vHalfSpinVector;
+    typedef iHalfSpinColourVector<vComplex > vHalfSpinColourVector;
     
-    typedef iSinglet<Complex >          TComplex;    // This is painful. Tensor singlet complex type.
-    typedef iSinglet<vComplex >         vTComplex;   // what if we don't know the tensor structure
-    typedef iSinglet<Real >             TReal;       // Shouldn't need these; can I make it work without?
-    typedef iSinglet<vReal >            vTReal;      
-    typedef iSinglet<vInteger >         vTInteger;
-    typedef iSinglet<Integer >          TInteger;
+    typedef iSinglet<Complex >         TComplex;    // FIXME This is painful. Tensor singlet complex type.
+    typedef iSinglet<vComplex >        vTComplex;   // what if we don't know the tensor structure
+    typedef iSinglet<Real >            TReal;       // Shouldn't need these; can I make it work without?
+    typedef iSinglet<vReal >           vTReal;      
+    typedef iSinglet<vInteger >        vTInteger;
+    typedef iSinglet<Integer >         TInteger;
 
-    typedef Lattice<vTReal>              LatticeReal;
-    typedef Lattice<vTComplex>           LatticeComplex;
-    typedef Lattice<vTInteger>            LatticeInteger; // Predicates for "where"
+    typedef Lattice<vTReal>            LatticeReal;
+    typedef Lattice<vTComplex>         LatticeComplex;
+    typedef Lattice<vTInteger>         LatticeInteger; // Predicates for "where"
     
     typedef Lattice<vColourMatrix>     LatticeColourMatrix;
     typedef Lattice<vSpinMatrix>       LatticeSpinMatrix;
     typedef Lattice<vSpinColourMatrix> LatticeSpinColourMatrix;
 
-    typedef Lattice<vSpinColourVector> LatticeSpinColourVector;
-    typedef Lattice<vSpinVector>       LatticeSpinVector;
-    typedef Lattice<vColourVector>     LatticeColourVector;
+    typedef Lattice<vSpinVector>           LatticeSpinVector;
+    typedef Lattice<vColourVector>         LatticeColourVector;
+    typedef Lattice<vSpinColourVector>     LatticeSpinColourVector;
+    typedef Lattice<vHalfSpinVector>       LatticeHalfSpinVector;
+    typedef Lattice<vHalfSpinColourVector> LatticeHalfSpinColourVector;
 
     ///////////////////////////////////////////
     // Physical names for things
     ///////////////////////////////////////////
-    typedef Lattice<vSpinColourVector> LatticeFermion;
-    typedef Lattice<vSpinColourMatrix> LatticePropagator;
-    typedef Lattice<vLorentzColourMatrix> LatticeGaugeField;
+    typedef Lattice<vHalfSpinColourVector> LatticeHalfFermion;
+    typedef Lattice<vSpinColourVector>     LatticeFermion;
 
+    typedef Lattice<vSpinColourMatrix>     LatticePropagator;
+    typedef Lattice<vLorentzColourMatrix>  LatticeGaugeField;
+
+    // Uhgg... typing this hurt  ;)
+    // (my keyboard got burning hot when I typed this, must be the anti-Fermion)
+    typedef Lattice<vColourVector>         LatticeStaggeredFermion;    
+    typedef Lattice<vColourMatrix>         LatticeStaggeredPropagator; 
 
     //////////////////////////////////////////////////////////////////////////////
     // Peek and Poke named after physics attributes
@@ -145,5 +161,7 @@ namespace QCD {
 } // Grid
 
 #include <qcd/Grid_qcd_dirac.h>
+#include <qcd/Grid_qcd_2spinor.h>
+//#include <qcd/Grid_qcd_pauli.h>
 
 #endif
diff --git a/lib/qcd/Grid_qcd_2spinor.h b/lib/qcd/Grid_qcd_2spinor.h
new file mode 100644
index 00000000..c19264f5
--- /dev/null
+++ b/lib/qcd/Grid_qcd_2spinor.h
@@ -0,0 +1,1082 @@
+#ifndef GRID_QCD_TWOSPIN_H
+#define GRID_QCD_TWOSPIN_H
+namespace Grid{
+
+namespace QCD {
+  
+
+  //////////////////////////////////////////////////////////////////////////////////////////////////////
+  // Normalisation alert; the g5   project is 1/2(1+-G5) 
+  //                      the xyzt projects are (1+-Gxyzt)
+  //
+  // * xyzt project
+  //
+  // This is because this is how the Wilson operator is normally written as
+  // (m+4r) - \frac{1}{2} D_{hop}
+  // and / or
+  // 1 - \frac{1}{2 m+8r} D_{hop} = 1 - kappa D_{hop}
+  //
+  // Note that the free, critical hopping parameter kappa is then 1/8 th for r=1.
+  //
+  // However, the xyzt 2spin "projectors" are not really projectors because they do not
+  // square to 1, however the ChiralProjector is a true projector.
+  //
+  // For this reason there is NO provision in Grid of a four spinor result from the
+  // xyzt projectors. They are intended to be used only in combination with "reconstruct" in the
+  // wilson dirac operator and closely related actions.
+  //
+  // I also do NOT provide lattice wide operators of these, since the dirac operator is best implemented
+  // via Stencils and single site variants will be needed only for the cache friendly high perf dirac operator.
+  //
+  // * chiral project
+  //
+  // Both four spinor and two spinor result variants are provided.
+  //
+  // The four spinor project will be recursively provided to Lattice wide routines, and likely used in 
+  // the domain wall and mobius implementations.
+  //
+  //////////////////////////////////////////////////////////////////////////////////////////////////////
+
+  /* Gx
+   *  0 0  0  i    [0]+-i[3]
+   *  0 0  i  0    [1]+-i[2]
+   *  0 -i 0  0
+   * -i 0  0  0
+   */
+  template<class vtype> inline void 
+    spProjXp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+    {
+      // To fail is not to err (Cryptic clue: suggest to Google SFINAE ;) )
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      hspin(0)=fspin(0)+timesI(fspin(3));
+      hspin(1)=fspin(1)+timesI(fspin(2));
+    }
+  template<class vtype> inline void spProjXm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      hspin(0)=fspin(0)-timesI(fspin(3));
+      hspin(1)=fspin(1)-timesI(fspin(2));
+    }
+
+      //  0 0  0  -1  [0] -+ [3]
+      //  0 0  1  0   [1] +- [2]
+      //  0 1  0  0
+      // -1 0  0  0
+  template<class vtype> inline void spProjYp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      hspin(0)=fspin(0)-fspin(3);
+      hspin(1)=fspin(1)+fspin(2);
+    }
+  template<class vtype> inline void spProjYm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      hspin(0)=fspin(0)+fspin(3);
+      hspin(1)=fspin(1)-fspin(2);
+    }
+	    /*Gz
+	     *  0 0  i  0   [0]+-i[2]
+	     *  0 0  0 -i   [1]-+i[3]
+	     * -i 0  0  0
+	     *  0 i  0  0
+	     */
+  template<class vtype> inline void spProjZp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      hspin(0)=fspin(0)+timesI(fspin(2));
+      hspin(1)=fspin(1)-timesI(fspin(3));
+    }
+  template<class vtype> inline void spProjZm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      hspin(0)=fspin(0)-timesI(fspin(2));
+      hspin(1)=fspin(1)+timesI(fspin(3));
+    }
+	    /*Gt
+	     *  0 0  1  0 [0]+-[2]
+	     *  0 0  0  1 [1]+-[3]
+	     *  1 0  0  0
+	     *  0 1  0  0
+	     */
+  template<class vtype> inline void spProjTp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      hspin(0)=fspin(0)+fspin(2);
+      hspin(1)=fspin(1)+fspin(3);
+    }
+  template<class vtype> inline void spProjTm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      hspin(0)=fspin(0)-fspin(2);
+      hspin(1)=fspin(1)-fspin(3);
+    }
+	    /*G5
+	     *  1 0  0  0 
+	     *  0 1  0  0 
+	     *  0 0 -1  0
+	     *  0 0  0 -1
+	     */
+
+  template<class vtype> inline void spProj5p (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      hspin(0)=fspin(0);
+      hspin(1)=fspin(1);
+    }
+  template<class vtype> inline void spProj5m (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      hspin(0)=fspin(2);
+      hspin(1)=fspin(3);
+    }
+  
+  //  template<class vtype> inline void fspProj5p (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype> inline void spProj5p (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      rfspin(0)=fspin(0);
+      rfspin(1)=fspin(1);
+      rfspin(2)=zero;
+      rfspin(3)=zero;
+    }
+  //  template<class vtype> inline void fspProj5m (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype> inline void spProj5m (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      rfspin(0)=zero;
+      rfspin(1)=zero;
+      rfspin(2)=fspin(2);
+      rfspin(3)=fspin(3);
+    }
+
+  ////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+  // Reconstruction routines to move back again to four spin
+  ////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+  /* Gx
+   *  0 0  0  i    [0]+-i[3]
+   *  0 0  i  0    [1]+-i[2]
+   *  0 -i 0  0  -i[1]+-[2]   == -i ([0]+-i[3]) = -i (1)
+   * -i 0  0  0  -i[0]+-[3]   == -i ([1]+-i[2]) = -i (0)
+   */
+  template<class vtype> inline void spReconXp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)=hspin(0);
+      fspin(1)=hspin(1);
+      fspin(2)=timesMinusI(hspin(1));
+      fspin(3)=timesMinusI(hspin(0));
+    }
+  template<class vtype> inline void spReconXm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)=hspin(0);
+      fspin(1)=hspin(1);
+      fspin(2)=timesI(hspin(1));
+      fspin(3)=timesI(hspin(0));
+    }
+  template<class vtype> inline void accumReconXp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)+=hspin(0);
+      fspin(1)+=hspin(1);
+      fspin(2)-=timesI(hspin(1));
+      fspin(3)-=timesI(hspin(0));
+    }
+  template<class vtype> inline void accumReconXm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)+=hspin(0);
+      fspin(1)+=hspin(1);
+      fspin(2)+=timesI(hspin(1));
+      fspin(3)+=timesI(hspin(0));
+    }
+
+      //  0 0  0  -1  [0] -+ [3]
+      //  0 0  1  0   [1] +- [2]
+      //  0 1  0  0              == 1(1)
+      // -1 0  0  0              ==-1(0)
+
+  template<class vtype> inline void spReconYp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)=hspin(0);
+      fspin(1)=hspin(1);
+      fspin(2)= hspin(1);
+      fspin(3)=-hspin(0);//Unary minus?
+    }
+  template<class vtype> inline void spReconYm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)=hspin(0);
+      fspin(1)=hspin(1);
+      fspin(2)=-hspin(1);
+      fspin(3)= hspin(0);
+    }
+  template<class vtype> inline void accumReconYp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)+=hspin(0);
+      fspin(1)+=hspin(1);
+      fspin(2)+=hspin(1);
+      fspin(3)-=hspin(0);
+    }
+  template<class vtype> inline void accumReconYm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)+=hspin(0);
+      fspin(1)+=hspin(1);
+      fspin(2)-=hspin(1);
+      fspin(3)+=hspin(0);
+    }
+
+	    /*Gz
+	     *  0 0  i  0   [0]+-i[2]
+	     *  0 0  0 -i   [1]-+i[3]
+	     * -i 0  0  0     => -i (0)
+	     *  0 i  0  0     =>  i (1)
+	     */
+  template<class vtype> inline void spReconZp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)=hspin(0);
+      fspin(1)=hspin(1);
+      fspin(2)=timesMinusI(hspin(0));
+      fspin(3)=timesI(hspin(1));
+    }
+  template<class vtype> inline void spReconZm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)=hspin(0);
+      fspin(1)=hspin(1);
+      fspin(2)=     timesI(hspin(0));
+      fspin(3)=timesMinusI(hspin(1));
+    }
+  template<class vtype> inline void accumReconZp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)+=hspin(0);
+      fspin(1)+=hspin(1);
+      fspin(2)-=timesI(hspin(0));
+      fspin(3)+=timesI(hspin(1));
+    }
+  template<class vtype> inline void accumReconZm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)+=hspin(0);
+      fspin(1)+=hspin(1);
+      fspin(2)+=timesI(hspin(0));
+      fspin(3)-=timesI(hspin(1));
+    }
+	    /*Gt
+	     *  0 0  1  0 [0]+-[2]
+	     *  0 0  0  1 [1]+-[3]
+	     *  1 0  0  0    => (0)
+	     *  0 1  0  0    => (1)
+	     */
+  template<class vtype> inline void spReconTp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)=hspin(0);
+      fspin(1)=hspin(1);
+      fspin(2)=hspin(0);
+      fspin(3)=hspin(1);
+    }
+  template<class vtype> inline void spReconTm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)=hspin(0);
+      fspin(1)=hspin(1);
+      fspin(2)=-hspin(0);
+      fspin(3)=-hspin(1);
+    }
+  template<class vtype> inline void accumReconTp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)+=hspin(0);
+      fspin(1)+=hspin(1);
+      fspin(2)+=hspin(0);
+      fspin(3)+=hspin(1);
+    }
+  template<class vtype> inline void accumReconTm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)+=hspin(0);
+      fspin(1)+=hspin(1);
+      fspin(2)-=hspin(0);
+      fspin(3)-=hspin(1);
+    }
+	    /*G5
+	     *  1 0  0  0 
+	     *  0 1  0  0 
+	     *  0 0 -1  0
+	     *  0 0  0 -1
+	     */
+  template<class vtype> inline void spRecon5p (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)=hspin(0)+hspin(0); // add is lower latency than mul
+      fspin(1)=hspin(1)+hspin(1); // probably no measurable diffence though
+      fspin(2)=zero;
+      fspin(3)=zero;
+    }
+  template<class vtype> inline void spRecon5m (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)=zero;
+      fspin(1)=zero;
+      fspin(2)=hspin(0)+hspin(0);
+      fspin(3)=hspin(1)+hspin(1);
+    }
+  template<class vtype> inline void accumRecon5p (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(0)+=hspin(0)+hspin(0);
+      fspin(1)+=hspin(1)+hspin(1);
+    }
+  template<class vtype> inline void accumRecon5m (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+    {
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      fspin(2)+=hspin(0)+hspin(0);
+      fspin(3)+=hspin(1)+hspin(1);
+    }
+
+  //////////////////////////////////////////////////////////////////////////////////////////////
+  // Recursively apply these until we hit the spin index
+  //////////////////////////////////////////////////////////////////////////////////////////////
+
+  //////////
+  // Xp
+  //////////
+  template<class rtype,class vtype> inline void spProjXp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spProjXp(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spProjXp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spProjXp(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spProjXp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spProjXp(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+  template<class rtype,class vtype> inline void spReconXp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spReconXp(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spReconXp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spReconXp(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spReconXp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spReconXp(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+  template<class rtype,class vtype> inline void accumReconXp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    accumReconXp(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void accumReconXp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      accumReconXp(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void accumReconXp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      accumReconXp(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+
+  ////////
+  // Xm
+  ////////
+  template<class rtype,class vtype> inline void spProjXm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spProjXm(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spProjXm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spProjXm(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spProjXm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spProjXm(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+  template<class rtype,class vtype> inline void spReconXm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spReconXm(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spReconXm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spReconXm(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spReconXm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spReconXm(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+  template<class rtype,class vtype> inline void accumReconXm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    accumReconXm(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void accumReconXm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      accumReconXm(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void accumReconXm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      accumReconXm(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+
+  ////////
+  // Yp
+  ////////
+  template<class rtype,class vtype> inline void spProjYp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spProjYp(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spProjYp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spProjYp(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spProjYp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spProjYp(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+  template<class rtype,class vtype> inline void spReconYp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spReconYp(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spReconYp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spReconYp(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spReconYp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spReconYp(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+  template<class rtype,class vtype> inline void accumReconYp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    accumReconYp(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void accumReconYp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      accumReconYp(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void accumReconYp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      accumReconYp(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+  ////////
+  // Ym
+  ////////
+  template<class rtype,class vtype> inline void spProjYm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spProjYm(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spProjYm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spProjYm(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spProjYm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spProjYm(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+  template<class rtype,class vtype> inline void spReconYm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spReconYm(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spReconYm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spReconYm(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spReconYm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spReconYm(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+  template<class rtype,class vtype> inline void accumReconYm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    accumReconYm(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void accumReconYm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      accumReconYm(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void accumReconYm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      accumReconYm(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+  ////////
+  // Zp
+  ////////
+  template<class rtype,class vtype> inline void spProjZp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spProjZp(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spProjZp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spProjZp(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spProjZp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spProjZp(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+  template<class rtype,class vtype> inline void spReconZp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spReconZp(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spReconZp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spReconZp(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spReconZp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spReconZp(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+  template<class rtype,class vtype> inline void accumReconZp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    accumReconZp(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void accumReconZp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      accumReconZp(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void accumReconZp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      accumReconZp(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+  ////////
+  // Zm
+  ////////
+  template<class rtype,class vtype> inline void spProjZm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spProjZm(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spProjZm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spProjZm(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spProjZm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spProjZm(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+  template<class rtype,class vtype> inline void spReconZm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spReconZm(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spReconZm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spReconZm(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spReconZm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spReconZm(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+  template<class rtype,class vtype> inline void accumReconZm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    accumReconZm(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void accumReconZm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      accumReconZm(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void accumReconZm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      accumReconZm(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+  ////////
+  // Tp
+  ////////
+  template<class rtype,class vtype> inline void spProjTp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spProjTp(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spProjTp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spProjTp(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spProjTp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spProjTp(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+  template<class rtype,class vtype> inline void spReconTp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spReconTp(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spReconTp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spReconTp(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spReconTp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spReconTp(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+  template<class rtype,class vtype> inline void accumReconTp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    accumReconTp(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void accumReconTp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      accumReconTp(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void accumReconTp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      accumReconTp(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+  ////////
+  // Tm
+  ////////
+  template<class rtype,class vtype> inline void spProjTm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spProjTm(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spProjTm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spProjTm(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spProjTm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spProjTm(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+  template<class rtype,class vtype> inline void spReconTm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spReconTm(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spReconTm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spReconTm(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spReconTm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spReconTm(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+  template<class rtype,class vtype> inline void accumReconTm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    accumReconTm(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void accumReconTm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      accumReconTm(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void accumReconTm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      accumReconTm(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+  ////////
+  // 5p
+  ////////
+  template<class rtype,class vtype> inline void spProj5p (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spProj5p(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spProj5p (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spProj5p(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spProj5p (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spProj5p(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+  template<class rtype,class vtype> inline void spRecon5p (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spRecon5p(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spRecon5p (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spRecon5p(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spRecon5p (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spRecon5p(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+  template<class rtype,class vtype> inline void accumRecon5p (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    accumRecon5p(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void accumRecon5p (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      accumRecon5p(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void accumRecon5p (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      accumRecon5p(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+  // four spinor projectors for chiral proj
+  //  template<class vtype> inline void fspProj5p (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
+  template<class vtype> inline void spProj5p (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spProj5p(hspin._internal,fspin._internal);
+  }
+  //  template<class vtype,int N> inline void fspProj5p (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class vtype,int N> inline void spProj5p (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spProj5p(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  //  template<class vtype,int N> inline void fspProj5p (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class vtype,int N> inline void spProj5p (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spProj5p(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+  ////////
+  // 5m
+  ////////
+  template<class rtype,class vtype> inline void spProj5m (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spProj5m(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spProj5m (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spProj5m(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spProj5m (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spProj5m(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+  template<class rtype,class vtype> inline void spRecon5m (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spRecon5m(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void spRecon5m (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spRecon5m(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void spRecon5m (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spRecon5m(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+  template<class rtype,class vtype> inline void accumRecon5m (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    accumRecon5m(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> inline void accumRecon5m (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      accumRecon5m(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  template<class rtype,class vtype,int N> inline void accumRecon5m (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      accumRecon5m(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+  // four spinor projectors for chiral proj
+  //  template<class vtype> inline void fspProj5m (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
+  template<class vtype> inline void spProj5m (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    spProj5m(hspin._internal,fspin._internal);
+  }
+  //  template<class vtype,int N> inline void fspProj5m (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class vtype,int N> inline void spProj5m (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++) {
+      spProj5m(hspin._internal[i],fspin._internal[i]);
+    }
+  }
+  //  template<class vtype,int N> inline void fspProj5m (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class vtype,int N> inline void spProj5m (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  {
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    for(int i=0;i<N;i++){ 
+    for(int j=0;j<N;j++){
+      spProj5m(hspin._internal[i][j],fspin._internal[i][j]);
+    }}
+  }
+
+
+
+
+}   //namespace QCD
+} // Grid
+#endif
diff --git a/tests/Grid_gamma.cc b/tests/Grid_gamma.cc
index 758f9faa..3b5a8430 100644
--- a/tests/Grid_gamma.cc
+++ b/tests/Grid_gamma.cc
@@ -29,42 +29,16 @@ int main (int argc, char ** argv)
   SpinMatrix rr=zero;
   SpinMatrix result;
 
-  SpinVector lv=zero;
-  SpinVector rv=zero;
+  SpinVector lv; random(sRNG,lv);
+  SpinVector rv; random(sRNG,rv);
 
   for(int a=0;a<Ns;a++){
     ident()(a,a) = 1.0;
   }
 
-  Gamma::GammaMatrix g [] = {
-    Gamma::Identity,
-    Gamma::GammaX,
-    Gamma::GammaY,
-    Gamma::GammaZ,
-    Gamma::GammaT,
-    Gamma::Gamma5,
-    Gamma::MinusIdentity,
-    Gamma::MinusGammaX,
-    Gamma::MinusGammaY,
-    Gamma::MinusGammaZ,
-    Gamma::MinusGammaT,
-    Gamma::MinusGamma5
-  };
-  const char *list[] = { 
-    "Identity ",
-    "GammaX   ",
-    "GammaY   ",
-    "GammaZ   ",
-    "GammaT   ",
-    "Gamma5   ",
-    "-Identity",
-    "-GammaX  ",
-    "-GammaY  ",
-    "-GammaZ  ",
-    "-GammaT  ",
-    "-Gamma5  ",
-    "         "
-  };
+  const Gamma::GammaMatrix *g = Gamma::GammaMatrices;
+  const char **list          = Gamma::GammaMatrixNames;
+
   result =ll*Gamma(g[0])*rr;
   result =ll*Gamma(g[0]);
   rv = Gamma(g[0])*lv;
@@ -124,5 +98,81 @@ int main (int argc, char ** argv)
     std::cout << list[mu]<<" " << mag<<std::endl;
   }
 
+  // Testing spins and reconstructs
+  SpinVector     recon; random(sRNG,rv);
+  SpinVector      full;
+  HalfSpinVector hsp,hsm;
+
+  // Xp
+  double mag;
+  spProjXp(hsm,rv);
+  spReconXp(recon,hsm);
+  full = rv + Gamma(Gamma::GammaX) *rv;
+  mag = TensorRemove(norm2l(full-recon));
+  std::cout << "Xp "<< mag<<std::endl;
+
+  // Xm
+  spProjXm(hsm,rv);
+  spReconXm(recon,hsm);
+  full = rv - Gamma(Gamma::GammaX) *rv;
+  mag = TensorRemove(norm2l(full-recon));
+  std::cout << "Xm "<< mag<<std::endl;
+
+  // Yp
+  spProjYp(hsm,rv);
+  spReconYp(recon,hsm);
+  full = rv + Gamma(Gamma::GammaY) *rv;
+  mag = TensorRemove(norm2l(full-recon));
+  std::cout << "Yp "<< mag<<std::endl;
+
+  // Ym
+  spProjYm(hsm,rv);
+  spReconYm(recon,hsm);
+  full = rv - Gamma(Gamma::GammaY) *rv;
+  mag = TensorRemove(norm2l(full-recon));
+  std::cout << "Ym "<< mag<<std::endl;
+
+  // Zp
+  spProjZp(hsm,rv);
+  spReconZp(recon,hsm);
+  full = rv + Gamma(Gamma::GammaZ) *rv;
+  mag = TensorRemove(norm2l(full-recon));
+  std::cout << "Zp "<< mag<<std::endl;
+
+  // Zm
+  spProjZm(hsm,rv);
+  spReconZm(recon,hsm);
+  full = rv - Gamma(Gamma::GammaZ) *rv;
+  mag = TensorRemove(norm2l(full-recon));
+  std::cout << "Zm "<< mag<<std::endl;
+
+  // Tp
+  spProjTp(hsm,rv);
+  spReconTp(recon,hsm);
+  full = rv + Gamma(Gamma::GammaT) *rv;
+  mag = TensorRemove(norm2l(full-recon));
+  std::cout << "Tp "<< mag<<std::endl;
+
+  // Tm
+  spProjTm(hsm,rv);
+  spReconTm(recon,hsm);
+  full = rv - Gamma(Gamma::GammaT) *rv;
+  mag = TensorRemove(norm2l(full-recon));
+  std::cout << "Tm "<< mag<<std::endl;
+
+  // 5p
+  spProj5p(hsm,rv);
+  spRecon5p(recon,hsm);
+  full = rv + Gamma(Gamma::Gamma5) *rv;
+  mag = TensorRemove(norm2l(full-recon));
+  std::cout << "5p "<< mag<<std::endl;
+
+  // 5m
+  spProj5m(hsm,rv);
+  spRecon5m(recon,hsm);
+  full = rv - Gamma(Gamma::Gamma5) *rv;
+  mag = TensorRemove(norm2l(full-recon));
+  std::cout << "5m "<< mag<<std::endl;
+
   Grid_finalize();
 }

From d5fd34b6e8472956ac2018ce20fe00627025dbfa Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 25 Apr 2015 13:04:26 +0100
Subject: [PATCH 104/429] Update to TODO list

---
 TODO | 64 ++++++++++++++++++++++++++++++++++++------------------------
 1 file changed, 38 insertions(+), 26 deletions(-)

diff --git a/TODO b/TODO
index b12502eb..064e99b4 100644
--- a/TODO
+++ b/TODO
@@ -1,5 +1,5 @@
 * - BinaryWriter, TextWriter etc...
-  - protocol buffers? replace xml
+  - protocol buffers? replace xmlReader/Writer ec..
 
 * Stencil operator support                           -----Initial thoughts, trial implementation DONE.
                                                      -----some simple tests that Stencil matches Cshift.
@@ -10,15 +10,14 @@
 
 * CovariantShift support                             -----Use a class to store gauge field? (parallel transport?)
 
+
 * Consider switch std::vector to boost arrays or something lighter weight
   boost::multi_array<type, 3> A()...    to replace multi1d, multi2d etc..
 
-* TensorRemove is a hack, come up with a long term rationalised approach to Complex vs. Scalar<Scalar<Scalar<Complex > > >
-  QDP forces use of "toDouble" to get back to non tensor scalar. This role is presently taken TensorRemove, but I
-  want to introduce a syntax that does not require this.
 
 * norm2l is a hack. figure out syntax error and make this norm2. c.f. tests/Grid_gamma.cc
 
+
 *  std::vector replacement;
 
   Had to change "reserve" back to "resize" on std::vector in Lattice class.
@@ -32,30 +31,47 @@
   resize. Find out if valarray or alternative works differently prior to 
   doing this since there may still be something I can use..
 
-* Flavour matrices?
 
 * Make the Tensor types and Complex etc... play more nicely.
+* TensorRemove is a hack, come up with a long term rationalised approach to Complex vs. Scalar<Scalar<Scalar<Complex > > >
+  QDP forces use of "toDouble" to get back to non tensor scalar. This role is presently taken TensorRemove, but I
+  want to introduce a syntax that does not require this.
 
-* Pauli, SU subgroup, etc.. 
-
-* Fourspin, two spin project
-
-* su3 exponentiation & log etc.. [Jamie's code?]
-  TaProj
-
-* Parallel MPI2 IO
-
-* rb4d support.
-
-* Check for missing functionality                    - partially audited against QDP++ layout
 
 * Optimise the extract/merge SIMD routines; Azusa??
  - I have collated into single location at least.
  - Need to use _mm_*insert/extract routines.
 
+
 * Conformable test in Cshift routines.
 
 
+* Flavour matrices?
+* Pauli, SU subgroup, etc.. 
+* su3 exponentiation & log etc.. [Jamie's code?]
+* TaProj
+
+
+* Fourspin, two spin project --- DONE
+
+
+* Parallel MPI2 IO
+
+* rb4d support for 5th dimension in Mobius.
+
+
+* Check for missing functionality                    - partially audited against QDP++ layout
+
+Algorithms
+* LinearOperator
+* LinearSolver
+* Polynomial 
+* Eigen
+* Pcg
+* fPcg
+* MCR
+* etc..
+
 AUDITS:
 
 * FIXME audit
@@ -68,25 +84,20 @@ AUDITS:
    // TODO
    //
    // Base class to share common code between vRealF, VComplexF etc...
-  //
+   //
    // Unary functions
    // cos,sin, tan, acos, asin, cosh, acosh, tanh, sinh, // Scalar<vReal> only arg
    // exp, log, sqrt, fabs
    //
-  // transposeColor, transposeSpin,
+   // transposeColor, transposeSpin,
    // adjColor, adjSpin,
    //
-  // copyMask.
+   // copyMask.
    //
    // localMaxAbs
    //
    // Fourier transform equivalent.
 
-* LinearOperator
-
-  LinearSolver
-
-  Polynomial etc...
 
 ======================================================================================================
 
@@ -102,6 +113,8 @@ FUNCTIONALITY:
                                                      ----- Implement mapping between traceColour and traceSpin and traceIndex<1/2>.
 * How to do U[mu] ... lorentz part of type structure or not. more like chroma if not. -- DONE
 
+* Twospin/Fourspin/Gamma/Proj/Recon                  ----- DONE
+
 * subdirs lib, tests ??                              ----- DONE
   - lib/math        
   - lib/cartesian
@@ -133,6 +146,5 @@ FUNCTIONALITY:
 * I/O support
 * NERSC Lattice loading, plaquette test             ------- DONE single node 
 
-
 * Controling std::cout                              ------- DONE
 

From c678f2d255c63f9ddd3cd9f2d74cf6473497a0bd Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 25 Apr 2015 14:33:02 +0100
Subject: [PATCH 105/429] Starting the implementation of wilson; incomplete and
 committing non-functional code which is not yet included from elsewhere or
 linked to the build system.

---
 TODO                           |  10 +--
 lib/qcd/Grid_qcd_wilson_dop.cc | 157 +++++++++++++++++++++++++++++++++
 lib/qcd/Grid_qcd_wilson_dop.h  |  62 +++++++++++++
 3 files changed, 224 insertions(+), 5 deletions(-)
 create mode 100644 lib/qcd/Grid_qcd_wilson_dop.cc
 create mode 100644 lib/qcd/Grid_qcd_wilson_dop.h

diff --git a/TODO b/TODO
index 064e99b4..46c12838 100644
--- a/TODO
+++ b/TODO
@@ -1,5 +1,6 @@
 * - BinaryWriter, TextWriter etc...
-  - protocol buffers? replace xmlReader/Writer ec..
+  - use protocol buffers? replace xmlReader/Writer ec..
+  - Binary use htonll, htonl
 
 * Stencil operator support                           -----Initial thoughts, trial implementation DONE.
                                                      -----some simple tests that Stencil matches Cshift.
@@ -15,10 +16,9 @@
   boost::multi_array<type, 3> A()...    to replace multi1d, multi2d etc..
 
 
-* norm2l is a hack. figure out syntax error and make this norm2. c.f. tests/Grid_gamma.cc
+* norm2l is a hack. figure out syntax error and make this norm2 c.f. tests/Grid_gamma.cc
 
-
-*  std::vector replacement;
+****  std::vector replacement;
 
   Had to change "reserve" back to "resize" on std::vector in Lattice class.
 
@@ -59,7 +59,6 @@
 
 * rb4d support for 5th dimension in Mobius.
 
-
 * Check for missing functionality                    - partially audited against QDP++ layout
 
 Algorithms
@@ -70,6 +69,7 @@ Algorithms
 * Pcg
 * fPcg
 * MCR
+* HDCG
 * etc..
 
 AUDITS:
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
new file mode 100644
index 00000000..73b19c12
--- /dev/null
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -0,0 +1,157 @@
+#ifnfdef GRID_QCD_WILSON_DOP_H
+#define  GRID_QCD_WILSON_DOP_H
+
+#include <Grid.h>
+
+namespace Grid {
+namespace QCD {
+
+
+const std::vector<int> WilsonMatrix::directions   ({0,1,2,3, 0, 1, 2, 3,0});
+const std::vector<int> WilsonMatrix::displacements({1,1,1,1,-1,-1,-1,-1,0});
+
+  // Should be in header?
+static const int WilsonMatrix::Xp = 0;
+static const int WilsonMatrix::Yp = 1;
+static const int WilsonMatrix::Zp = 2;
+static const int WilsonMatrix::Tp = 3;
+static const int WilsonMatrix::Xm = 4;
+static const int WilsonMatrix::Ym = 5;
+static const int WilsonMatrix::Zm = 6;
+static const int WilsonMatrix::Tm = 7;
+static const int WilsonMatrix::X0 = 8;
+static const int WilsonMatrix::npoint=9;
+
+
+WilsonMatrix::WilsonMatrix(LatticeGaugeField &_Umu,int _mass)
+  : Stencil((&Umu._grid,npoint,0,directions,displacements),
+    mass(_mass),
+    Umu(_Umu)
+{
+  // Allocate the required comms buffer
+  grid = _Umu._grid;
+  comm_buf.resize(Stencil._unified_buffer_size);
+}
+void WilsonMatrix::multiply(const LatticeFermion &in, LatticeFermion &out)
+{
+  
+}
+void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
+{
+  Stencil.HaloExchange(in,comm_buf);
+
+  for(int ss=0;ss<_grid->oSites();ss++){
+
+    int offset,local;
+
+    vSpinColourVector result;
+    vHalfSpinColourVector UChi;
+
+    // Xp
+    offset = Stencil._offsets [Xp][ss];
+    local  = Stencil._is_local[Xp][ss];
+    if ( local ) {
+      Uchi = U[]*spProjXp(in._odata[offset]);
+    } else {
+      Uchi = U[]*comm_buf._odata[offset]
+    }
+    result = ReconXp(Uchi);
+
+    // Yp
+    offset = Stencil._offsets [Yp][ss];
+    local  = Stencil._is_local[Yp][ss];
+    if ( local ) {
+      Uchi = U[]*spProjYp(in._odata[offset]);
+    } else {
+      Uchi = U[]*comm_buf._odata[offset]
+    }
+    result+= ReconYp(Uchi);
+
+    // Zp
+    offset = Stencil._offsets [Zp][ss];
+    local  = Stencil._is_local[Zp][ss];
+    if ( local ) {
+      Uchi = U[]*spProjZp(in._odata[offset]);
+    } else {
+      Uchi = U[]*comm_buf._odata[offset]
+    }
+    result+= ReconZp(Uchi);
+
+    // Tp
+    offset = Stencil._offsets [Tp][ss];
+    local  = Stencil._is_local[Tp][ss];
+    if ( local ) {
+      Uchi = U[]*spProjTp(in._odata[offset]);
+    } else {
+      Uchi = U[]*comm_buf._odata[offset]
+    }
+    result+= ReconTp(Uchi);
+
+    // Xm
+    offset = Stencil._offsets [Xm][ss];
+    local  = Stencil._is_local[Xm][ss];
+    if ( local ) {
+      Uchi = U[]*spProjXm(in._odata[offset]);
+    } else {
+      Uchi = U[]*comm_buf._odata[offset]
+    }
+    result+= ReconXm(Uchi);
+
+    // Ym
+    offset = Stencil._offsets [Ym][ss];
+    local  = Stencil._is_local[Ym][ss];
+    if ( local ) {
+      Uchi = U[]*spProjYm(in._odata[offset]);
+    } else {
+      Uchi = U[]*comm_buf._odata[offset]
+    }
+    result+= ReconYm(Uchi);
+
+    // Zm
+    offset = Stencil._offsets [Zm][ss];
+    local  = Stencil._is_local[Zm][ss];
+    if ( local ) {
+      Uchi = U[]*spProjZm(in._odata[offset]);
+    } else {
+      Uchi = U[]*comm_buf._odata[offset]
+    }
+    result+= ReconZm(Uchi);
+
+    // Tm
+    offset = Stencil._offsets [Tm][ss];
+    local  = Stencil._is_local[Tm][ss];
+    if ( local ) {
+      Uchi = U[]*spProjTm(in._odata[offset]);
+    } else {
+      Uchi = U[]*comm_buf._odata[offset]
+    }
+    result+= ReconTm(Uchi);
+
+    out._odata[ss] = result;
+  }
+
+}
+void WilsonMatrix::Dw(const LatticeFermion &in, LatticeFermion &out)
+{
+
+}
+void WilsonMatrix::MpcDag   (const LatticeFermion &in, LatticeFermion &out)
+{
+
+}
+void WilsonMatrix::Mpc      (const LatticeFermion &in, LatticeFermion &out)
+{
+
+}
+void WilsonMatrix::MpcDagMpc(const LatticeFermion &in, LatticeFermion &out)
+{
+
+}
+void WilsonMatrix::MDagM    (const LatticeFermion &in, LatticeFermion &out)
+{
+
+}
+
+
+}}
+#endif
diff --git a/lib/qcd/Grid_qcd_wilson_dop.h b/lib/qcd/Grid_qcd_wilson_dop.h
new file mode 100644
index 00000000..f6775842
--- /dev/null
+++ b/lib/qcd/Grid_qcd_wilson_dop.h
@@ -0,0 +1,62 @@
+#ifnfdef GRID_QCD_WILSON_DOP_H
+#define  GRID_QCD_WILSON_DOP_H
+
+#include <Grid.h>
+
+namespace Grid {
+
+  namespace QCD {
+
+
+    template<class vtype> class LinearOperatorBase {
+    public:
+      void multiply(const Lattice<vtype> &in, Lattice<vtype> &out){ assert(0);}
+    };
+
+    class WilsonMatrix : public LinearOperatorBase<LatticeFermion>
+    {
+      //NB r=1;
+    public:
+      double                        mass;
+      GridBase                     *grid;
+
+      // Copy of the gauge field 
+      LatticeGaugeField             Umu;
+
+      //Defines the stencil
+      CartesianStencil              Stencil; 
+      static const int npoint=9;
+      static const std::vector<int> directions   ;
+      static const std::vector<int> displacements;
+
+      static const int Xp,Xm,Yp,Ym,Zp,Zm,Tp,Tm;
+
+      // Comms buffer
+      std::vector<vSpinColourVector,alignedAllocator<vSpinColourVector> >  comm_buf;
+
+      // Constructor
+      WilsonMatrix(LatticeGaugeField &Umu,int mass);
+
+      // override multiply
+      void multiply(const LatticeFermion &in, LatticeFermion &out);
+
+      // non-hermitian hopping term; half cb or both
+      void Dhop(const LatticeFermion &in, LatticeFermion &out);
+
+      // m+4r -1/2 Dhop; both cb's
+      void Dw(const LatticeFermion &in, LatticeFermion &out);
+
+      // half checkerboard operaions
+      void MpcDag   (const LatticeFermion &in, LatticeFermion &out);
+      void Mpc      (const LatticeFermion &in, LatticeFermion &out);
+      void MpcDagMpc(const LatticeFermion &in, LatticeFermion &out);
+
+      // full checkerboard hermitian
+      void MDagM    (const LatticeFermion &in, LatticeFermion &out);
+
+      
+    };
+
+  }
+}
+#endif

From 35cfef2129f4a60ffec6916b9ef0f820150467f1 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sun, 26 Apr 2015 15:51:09 +0100
Subject: [PATCH 106/429] Big updates with progress towards wilson matrix

---
 TODO                                 |  18 ++-
 lib/Grid.h                           |   1 +
 lib/Grid_config.h                    |   4 +-
 lib/Grid_simd.h                      |  15 +-
 lib/Grid_stencil.h                   | 159 +++++++++++++-----
 lib/Makefile.am                      |   1 +
 lib/lattice/Grid_lattice_arith.h     |  24 +--
 lib/lattice/Grid_lattice_reduction.h |   9 +-
 lib/lattice/Grid_lattice_rng.h       |  67 +++++++-
 lib/math/Grid_math_arith_scalar.h    | 100 ++++++------
 lib/math/Grid_math_inner.h           |   6 +-
 lib/math/Grid_math_tensors.h         |  39 +++--
 lib/qcd/Grid_qcd.h                   | 233 +++++++++++++++++++++------
 lib/qcd/Grid_qcd_dirac.h             |  17 +-
 lib/qcd/Grid_qcd_wilson_dop.cc       | 201 +++++++++++++++--------
 lib/qcd/Grid_qcd_wilson_dop.h        |  12 +-
 lib/simd/Grid_vComplexD.h            |  51 +++---
 lib/simd/Grid_vComplexF.h            |  36 +++--
 lib/simd/Grid_vRealD.h               |  11 +-
 lib/simd/Grid_vRealF.h               |  11 +-
 lib/stencil/Grid_stencil_common.cc   |   3 +-
 tests/Grid_gamma.cc                  |  24 ++-
 tests/Grid_nersc_io.cc               |   4 +-
 tests/Grid_simd.cc                   | 166 +++++++++++++++++++
 tests/Grid_stencil.cc                |  74 +++------
 tests/Grid_wilson.cc                 |  69 ++++++++
 tests/Makefile.am                    |   8 +-
 27 files changed, 1008 insertions(+), 355 deletions(-)
 create mode 100644 tests/Grid_simd.cc
 create mode 100644 tests/Grid_wilson.cc

diff --git a/TODO b/TODO
index 46c12838..98dfc3b8 100644
--- a/TODO
+++ b/TODO
@@ -2,6 +2,10 @@
   - use protocol buffers? replace xmlReader/Writer ec..
   - Binary use htonll, htonl
 
+* Reduce implemention is poor
+* Bug in SeedFixedIntegers gives same output on each site.
+* Bug in RNG with complex numbers ; only filling real values; need helper function -- DONE
+
 * Stencil operator support                           -----Initial thoughts, trial implementation DONE.
                                                      -----some simple tests that Stencil matches Cshift.
                                                      -----do all permute in comms phase, so that copy permute
@@ -11,6 +15,7 @@
 
 * CovariantShift support                             -----Use a class to store gauge field? (parallel transport?)
 
+* Strong test for norm2, conj and all primitive types.
 
 * Consider switch std::vector to boost arrays or something lighter weight
   boost::multi_array<type, 3> A()...    to replace multi1d, multi2d etc..
@@ -33,10 +38,21 @@
 
 
 * Make the Tensor types and Complex etc... play more nicely.
-* TensorRemove is a hack, come up with a long term rationalised approach to Complex vs. Scalar<Scalar<Scalar<Complex > > >
+
+  - TensorRemove is a hack, come up with a long term rationalised approach to Complex vs. Scalar<Scalar<Scalar<Complex > > >
   QDP forces use of "toDouble" to get back to non tensor scalar. This role is presently taken TensorRemove, but I
   want to introduce a syntax that does not require this.
 
+  - Reductions that contract indices on a site should always demote the tensor structure.
+    norm2(), innerProduct.
+
+  - Result of Sum(), SliceSum // spatial sums
+              trace, traceIndex etc.. do not.
+
+  - problem arises because "trace" returns Lattice<TComplex> moving everything down to Scalar,
+    and then Sum and SliceSum to not remove the Scalars. This would be fixed if we 
+    template specialize the scalar scalar scalar sum and SliceSum,  on the basis of being
+    pure scalar.
 
 * Optimise the extract/merge SIMD routines; Azusa??
  - I have collated into single location at least.
diff --git a/lib/Grid.h b/lib/Grid.h
index 2a666c3b..d402b545 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -11,6 +11,7 @@
 #define GRID_H
 
 #include <stdio.h>
+
 #include <complex>
 #include <vector>
 #include <iostream>
diff --git a/lib/Grid_config.h b/lib/Grid_config.h
index 4152540e..f3d7016d 100644
--- a/lib/Grid_config.h
+++ b/lib/Grid_config.h
@@ -2,7 +2,7 @@
 /* lib/Grid_config.h.in.  Generated from configure.ac by autoheader.  */
 
 /* AVX */
-#define AVX1 1
+/* #undef AVX1 */
 
 /* AVX2 */
 /* #undef AVX2 */
@@ -77,7 +77,7 @@
 #define PACKAGE_VERSION "1.0"
 
 /* SSE4 */
-/* #undef SSE4 */
+#define SSE4 1
 
 /* Define to 1 if you have the ANSI C header files. */
 #define STDC_HEADERS 1
diff --git a/lib/Grid_simd.h b/lib/Grid_simd.h
index d8788e33..a059cc23 100644
--- a/lib/Grid_simd.h
+++ b/lib/Grid_simd.h
@@ -26,10 +26,15 @@ namespace Grid {
 
   typedef  float  RealF;
   typedef  double RealD;
-  
+#ifdef GRID_DEFAULT_PRECISION_DOUBLE
+  typedef RealD   Real;
+#else
+  typedef RealF  Real;
+#endif
+
   typedef std::complex<RealF> ComplexF;
   typedef std::complex<RealD> ComplexD;
-
+  typedef std::complex<Real>  Complex;
 
   inline RealF adj(const RealF  & r){ return r; }
   inline RealF conj(const RealF  & r){ return r; }
@@ -63,8 +68,8 @@ namespace Grid {
     //conj already supported for complex
 
     inline ComplexF timesI(const ComplexF r)     { return(r*ComplexF(0.0,1.0));}
-    inline ComplexF timesMinusI(const ComplexF r){ return(r*ComplexF(0.0,-1.0));}
     inline ComplexD timesI(const ComplexD r)     { return(r*ComplexD(0.0,1.0));}
+    inline ComplexF timesMinusI(const ComplexF r){ return(r*ComplexF(0.0,-1.0));}
     inline ComplexD timesMinusI(const ComplexD r){ return(r*ComplexD(0.0,-1.0));}
 
     inline void mac (RealD * __restrict__ y,const RealD * __restrict__ a,const RealD *__restrict__ x){  *y = (*a) * (*x)+(*y);}
@@ -280,15 +285,11 @@ namespace Grid {
 
   // Default precision
 #ifdef GRID_DEFAULT_PRECISION_DOUBLE
-  typedef RealD   Real;
   typedef vRealD vReal;
   typedef vComplexD vComplex;
-  typedef std::complex<Real>  Complex;
 #else
-  typedef RealF  Real;
   typedef vRealF vReal;
   typedef vComplexF vComplex;
-  typedef std::complex<Real>  Complex;
 #endif
 }
 #endif
diff --git a/lib/Grid_stencil.h b/lib/Grid_stencil.h
index 7bafb879..83db0a80 100644
--- a/lib/Grid_stencil.h
+++ b/lib/Grid_stencil.h
@@ -47,6 +47,101 @@ namespace Grid {
     int from_rank;
   } ;
 
+
+///////////////////////////////////////////////////////////////////
+// Gather for when there is no need to SIMD split with compression
+///////////////////////////////////////////////////////////////////
+template<class vobj,class cobj,class compressor> void 
+Gather_plane_simple (Lattice<vobj> &rhs,std::vector<cobj,alignedAllocator<cobj> > &buffer,int dimension,int plane,int cbmask,compressor &compress)
+{
+  int rd = rhs._grid->_rdimensions[dimension];
+
+  if ( !rhs._grid->CheckerBoarded(dimension) ) {
+
+    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+    int o   = 0;                                    // relative offset to base within plane
+    int bo  = 0;                                    // offset in buffer
+
+    // Simple block stride gather of SIMD objects
+#pragma omp parallel for collapse(2)
+    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+	buffer[bo++]=compress(rhs._odata[so+o+b]);
+      }
+      o +=rhs._grid->_slice_stride[dimension];
+    }
+
+  } else { 
+
+    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+    int o   = 0;                                      // relative offset to base within plane
+    int bo  = 0;                                      // offset in buffer
+
+#pragma omp parallel for collapse(2)
+    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+
+	int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);// Could easily be a table lookup
+	if ( ocb &cbmask ) {
+	  buffer[bo]=compress(rhs._odata[so+o+b]);
+	  bo++;
+	}
+
+      }
+      o +=rhs._grid->_slice_stride[dimension];
+    }
+  }
+}
+
+///////////////////////////////////////////////////////////////////
+// Gather for when there *is* need to SIMD split with compression
+///////////////////////////////////////////////////////////////////
+template<class cobj,class vobj,class compressor> void 
+  Gather_plane_extract(Lattice<vobj> &rhs,std::vector<typename cobj::scalar_type *> pointers,int dimension,int plane,int cbmask,compressor &compress)
+{
+  int rd = rhs._grid->_rdimensions[dimension];
+
+  if ( !rhs._grid->CheckerBoarded(dimension) ) {
+
+    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+    int o   = 0;                                    // relative offset to base within plane
+    int bo  = 0;                                    // offset in buffer
+
+    // Simple block stride gather of SIMD objects
+#pragma omp parallel for collapse(2)
+    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+	cobj temp;
+	temp=compress(rhs._odata[so+o+b]);
+	extract(temp,pointers);
+      }
+      o +=rhs._grid->_slice_stride[dimension];
+    }
+
+  } else { 
+
+    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+    int o   = 0;                                      // relative offset to base within plane
+    int bo  = 0;                                      // offset in buffer
+    
+#pragma omp parallel for collapse(2)
+    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+
+	int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);
+	if ( ocb & cbmask ) {
+	  cobj temp; 
+	  temp =compress(rhs._odata[so+o+b]);
+	  extract(temp,pointers);
+	}
+
+      }
+      o +=rhs._grid->_slice_stride[dimension];
+    }
+  }
+}
+
+
   class CartesianStencil { // Stencil runs along coordinate axes only; NO diagonal fill in.
   public:
 
@@ -86,8 +181,8 @@ namespace Grid {
 
       // Could allow a functional munging of the halo to another type during the comms.
       // this could implement the 16bit/32bit/64bit compression.
-      template<class vobj> void HaloExchange(Lattice<vobj> &source,
-					     std::vector<vobj,alignedAllocator<vobj> > &u_comm_buf)
+      template<class vobj,class cobj, class compressor> void 
+	HaloExchange(Lattice<vobj> &source,std::vector<cobj,alignedAllocator<cobj> > &u_comm_buf,compressor &compress)
       {
 	// conformable(source._grid,_grid);
 	assert(source._grid==_grid);
@@ -95,12 +190,10 @@ namespace Grid {
 	int u_comm_offset=0;
 
 	// Gather all comms buffers
-	typedef typename vobj::vector_type vector_type;
-	typedef typename vobj::scalar_type scalar_type;
-
 	for(int point = 0 ; point < _npoints; point++) {
 
-	  printf("Point %d \n",point);fflush(stdout);
+	  compress.Point(point);
+
 	  int dimension    = _directions[point];
 	  int displacement = _distances[point];
 	  
@@ -126,33 +219,30 @@ namespace Grid {
 	    sshift[1] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,1);
 	    if ( sshift[0] == sshift[1] ) {
 	      if (splice_dim) {
-		printf("splice 0x3 \n");fflush(stdout);
-		GatherStartCommsSimd(source,dimension,shift,0x3,u_comm_buf,u_comm_offset);
+		GatherStartCommsSimd(source,dimension,shift,0x3,u_comm_buf,u_comm_offset,compress);
 	      } else { 
-		printf("NO splice 0x3 \n");fflush(stdout);
-		GatherStartComms(source,dimension,shift,0x3,u_comm_buf,u_comm_offset);
+		GatherStartComms(source,dimension,shift,0x3,u_comm_buf,u_comm_offset,compress);
 	      }
 	    } else {
 	      if(splice_dim){
-		printf("splice 0x1,2 \n");fflush(stdout);
-		GatherStartCommsSimd(source,dimension,shift,0x1,u_comm_buf,u_comm_offset);// if checkerboard is unfavourable take two passes
-		GatherStartCommsSimd(source,dimension,shift,0x2,u_comm_buf,u_comm_offset);// both with block stride loop iteration
+		GatherStartCommsSimd(source,dimension,shift,0x1,u_comm_buf,u_comm_offset,compress);// if checkerboard is unfavourable take two passes
+		GatherStartCommsSimd(source,dimension,shift,0x2,u_comm_buf,u_comm_offset,compress);// both with block stride loop iteration
 	      } else {
-		printf("NO splice 0x1,2 \n");fflush(stdout);
-		GatherStartComms(source,dimension,shift,0x1,u_comm_buf,u_comm_offset);
-		GatherStartComms(source,dimension,shift,0x2,u_comm_buf,u_comm_offset);
+		GatherStartComms(source,dimension,shift,0x1,u_comm_buf,u_comm_offset,compress);
+		GatherStartComms(source,dimension,shift,0x2,u_comm_buf,u_comm_offset,compress);
 	      }
 	    }
 	  }
 	}
       }
 
-      template<class vobj> void GatherStartComms(Lattice<vobj> &rhs,int dimension,int shift,int cbmask,
-						 std::vector<vobj,alignedAllocator<vobj> > &u_comm_buf,
-						 int &u_comm_offset)
+      template<class vobj,class cobj, class compressor> 
+        void GatherStartComms(Lattice<vobj> &rhs,int dimension,int shift,int cbmask,
+			      std::vector<cobj,alignedAllocator<cobj> > &u_comm_buf,
+			      int &u_comm_offset,compressor & compress)
 	{
-	  typedef typename vobj::vector_type vector_type;
-	  typedef typename vobj::scalar_type scalar_type;
+	  typedef typename cobj::vector_type vector_type;
+	  typedef typename cobj::scalar_type scalar_type;
 	  
 	  GridBase *grid=_grid;
 	  assert(rhs._grid==_grid);
@@ -169,31 +259,26 @@ namespace Grid {
 	  
 	  int buffer_size = _grid->_slice_nblock[dimension]*_grid->_slice_block[dimension];
 	  
-	  std::vector<vobj,alignedAllocator<vobj> > send_buf(buffer_size); // hmm...
-	  std::vector<vobj,alignedAllocator<vobj> > recv_buf(buffer_size);
+	  std::vector<cobj,alignedAllocator<cobj> > send_buf(buffer_size); // hmm...
+	  std::vector<cobj,alignedAllocator<cobj> > recv_buf(buffer_size);
 	  
 	  int cb= (cbmask==0x2)? 1 : 0;
 	  int sshift= _grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,cb);
 	  
 	  for(int x=0;x<rd;x++){       
 	    
-	    printf("GatherStartComms x %d/%d\n",x,rd);fflush(stdout);
 	    int offnode = ( x+sshift >= rd );
 	    int sx        = (x+sshift)%rd;
 	    int comm_proc = (x+sshift)/rd;
 	    
 	    if (offnode) {
 	      
-	      printf("GatherStartComms offnode x %d\n",x);fflush(stdout);
 	      int words = send_buf.size();
 	      if (cbmask != 0x3) words=words>>1;
 	    
-	      int bytes = words * sizeof(vobj);
+	      int bytes = words * sizeof(cobj);
 
-	      printf("Gather_plane_simple dimension %d sx %d cbmask %d\n",dimension,sx,cbmask);fflush(stdout);
-	      Gather_plane_simple (rhs,send_buf,dimension,sx,cbmask);
-
-	      printf("GatherStartComms gathered offnode x %d\n",x);fflush(stdout);
+	      Gather_plane_simple (rhs,send_buf,dimension,sx,cbmask,compress);
 
 	      int rank           = _grid->_processor;
 	      int recv_from_rank;
@@ -219,14 +304,14 @@ namespace Grid {
 	}
 
 
-      template<class vobj>
+      template<class vobj,class cobj, class compressor> 
 	void  GatherStartCommsSimd(Lattice<vobj> &rhs,int dimension,int shift,int cbmask,
-				   std::vector<vobj,alignedAllocator<vobj> > &u_comm_buf,
-				   int &u_comm_offset)
+				   std::vector<cobj,alignedAllocator<cobj> > &u_comm_buf,
+				   int &u_comm_offset,compressor &compress)
 	{
 	  const int Nsimd = _grid->Nsimd();
-	  typedef typename vobj::vector_type vector_type;
-	  typedef typename vobj::scalar_type scalar_type;
+	  typedef typename cobj::vector_type vector_type;
+	  typedef typename cobj::scalar_type scalar_type;
 	  
 	  int fd = _grid->_fdimensions[dimension];
 	  int rd = _grid->_rdimensions[dimension];
@@ -245,7 +330,7 @@ namespace Grid {
 	  // Simd direction uses an extract/merge pair
 	  ///////////////////////////////////////////////
 	  int buffer_size = _grid->_slice_nblock[dimension]*_grid->_slice_block[dimension];
-	  int words = sizeof(vobj)/sizeof(vector_type);
+	  int words = sizeof(cobj)/sizeof(vector_type);
 
 	  /*   FIXME ALTERNATE BUFFER DETERMINATION */
 	  std::vector<std::vector<scalar_type> > send_buf_extract(Nsimd,std::vector<scalar_type>(buffer_size*words) ); 
@@ -285,7 +370,7 @@ namespace Grid {
 	      for(int i=0;i<Nsimd;i++){
 		pointers[Nsimd-1-i] = (scalar_type *)&send_buf_extract[i][0];
 	      }
-	      Gather_plane_extract(rhs,pointers,dimension,sx,cbmask);
+	      Gather_plane_extract<cobj>(rhs,pointers,dimension,sx,cbmask,compress);
 	      
 	      for(int i=0;i<Nsimd;i++){
 		
diff --git a/lib/Makefile.am b/lib/Makefile.am
index 9034ce9c..590a13f8 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -18,6 +18,7 @@ libGrid_a_SOURCES =\
 	 Grid_init.cc\
 	stencil/Grid_stencil_common.cc\
 	qcd/Grid_qcd_dirac.cc\
+	qcd/Grid_qcd_wilson_dop.cc\
 	$(extra_sources)
 
 #
diff --git a/lib/lattice/Grid_lattice_arith.h b/lib/lattice/Grid_lattice_arith.h
index 468dca79..26ed6a29 100644
--- a/lib/lattice/Grid_lattice_arith.h
+++ b/lib/lattice/Grid_lattice_arith.h
@@ -79,18 +79,18 @@ namespace Grid {
     return ret;
   }
   template<class left,class right>
-    inline auto operator + (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]*rhs._odata[0])>
+    inline auto operator + (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]+rhs._odata[0])>
   {
     //NB mult performs conformable check. Do not reapply here for performance.
-    Lattice<decltype(lhs._odata[0]*rhs._odata[0])> ret(rhs._grid);
+    Lattice<decltype(lhs._odata[0]+rhs._odata[0])> ret(rhs._grid);
     add(ret,lhs,rhs);
     return ret;
   }
   template<class left,class right>
-    inline auto operator - (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]*rhs._odata[0])>
+    inline auto operator - (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]-rhs._odata[0])>
   {
     //NB mult performs conformable check. Do not reapply here for performance.
-    Lattice<decltype(lhs._odata[0]*rhs._odata[0])> ret(rhs._grid);
+    Lattice<decltype(lhs._odata[0]-rhs._odata[0])> ret(rhs._grid);
     sub(ret,lhs,rhs);
     return ret;
   }
@@ -107,9 +107,9 @@ namespace Grid {
         return ret;
   }
   template<class left,class right>
-    inline auto operator + (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs*rhs._odata[0])>
+    inline auto operator + (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs+rhs._odata[0])>
     {
-      Lattice<decltype(lhs*rhs._odata[0])> ret(rhs._grid);
+      Lattice<decltype(lhs+rhs._odata[0])> ret(rhs._grid);
 #pragma omp parallel for
       for(int ss=0;ss<rhs._grid->oSites(); ss++){
 	ret._odata[ss]=lhs+rhs._odata[ss];
@@ -117,9 +117,9 @@ namespace Grid {
         return ret;
     }
   template<class left,class right>
-    inline auto operator - (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs*rhs._odata[0])>
+    inline auto operator - (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs-rhs._odata[0])>
   {
-    Lattice<decltype(lhs*rhs._odata[0])> ret(rhs._grid);
+    Lattice<decltype(lhs-rhs._odata[0])> ret(rhs._grid);
 #pragma omp parallel for
     for(int ss=0;ss<rhs._grid->oSites(); ss++){
       ret._odata[ss]=lhs-rhs._odata[ss];
@@ -137,9 +137,9 @@ namespace Grid {
       return ret;
     }
     template<class left,class right>
-      inline auto operator + (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]*rhs)>
+      inline auto operator + (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]+rhs)>
     {
-        Lattice<decltype(lhs._odata[0]*rhs)> ret(lhs._grid);
+        Lattice<decltype(lhs._odata[0]+rhs)> ret(lhs._grid);
 #pragma omp parallel for
         for(int ss=0;ss<rhs._grid->oSites(); ss++){
 	  ret._odata[ss]=lhs._odata[ss]+rhs;
@@ -147,9 +147,9 @@ namespace Grid {
         return ret;
     }
     template<class left,class right>
-      inline auto operator - (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]*rhs)>
+      inline auto operator - (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]-rhs)>
     {
-      Lattice<decltype(lhs._odata[0]*rhs)> ret(lhs._grid);
+      Lattice<decltype(lhs._odata[0]-rhs)> ret(lhs._grid);
 #pragma omp parallel for
       for(int ss=0;ss<rhs._grid->oSites(); ss++){
 	ret._odata[ss]=lhs._odata[ss]-rhs;
diff --git a/lib/lattice/Grid_lattice_reduction.h b/lib/lattice/Grid_lattice_reduction.h
index bec436f6..4491e8ec 100644
--- a/lib/lattice/Grid_lattice_reduction.h
+++ b/lib/lattice/Grid_lattice_reduction.h
@@ -14,7 +14,7 @@ namespace Grid {
 
       typedef typename vobj::scalar_type scalar;
       typedef typename vobj::vector_type vector;
-      decltype(innerProduct(arg._odata[0],arg._odata[0])) vnrm=zero;
+      decltype(innerProduct(arg._odata[0],arg._odata[0])) vnrm; 
       scalar nrm;
       //FIXME make this loop parallelisable
       vnrm=zero;
@@ -33,10 +33,11 @@ namespace Grid {
       //->decltype(innerProduct(left._odata[0],right._odata[0]))
     {
       typedef typename vobj::scalar_type scalar;
-      decltype(innerProduct(left._odata[0],right._odata[0])) vnrm=zero;
+      decltype(innerProduct(left._odata[0],right._odata[0])) vnrm; 
 
       scalar nrm;
       //FIXME make this loop parallelisable
+      vnrm=zero;
       for(int ss=0;ss<left._grid->oSites(); ss++){
 	vnrm = vnrm + innerProduct(left._odata[ss],right._odata[ss]);
       }
@@ -94,8 +95,10 @@ template<class vobj> inline void sliceSum(const Lattice<vobj> &Data,std::vector<
   int ld=grid->_ldimensions[orthogdim];
   int rd=grid->_rdimensions[orthogdim];
 
+  sobj szero; szero=zero;
+
   std::vector<vobj,alignedAllocator<vobj> > lvSum(rd); // will locally sum vectors first
-  std::vector<sobj> lsSum(ld,sobj(zero)); // sum across these down to scalars
+  std::vector<sobj> lsSum(ld,szero); // sum across these down to scalars
   std::vector<sobj> extracted(Nsimd);     // splitting the SIMD
 
   result.resize(fd); // And then global sum to return the same vector to every node for IO to file
diff --git a/lib/lattice/Grid_lattice_rng.h b/lib/lattice/Grid_lattice_rng.h
index 85d8c471..0842c80c 100644
--- a/lib/lattice/Grid_lattice_rng.h
+++ b/lib/lattice/Grid_lattice_rng.h
@@ -26,6 +26,8 @@ namespace Grid {
     }
 
   };
+  
+
   class GridRNGbase {
 
   public:
@@ -62,6 +64,21 @@ namespace Grid {
     }
 
 
+    // real scalars are one component
+    template<class scalar,class distribution> void fillScalar(scalar &s,distribution &dist)
+    {
+      s=dist(_generators[0]);
+    }
+    template<class distribution> void fillScalar(ComplexF &s,distribution &dist)
+    {
+      s=ComplexF(dist(_generators[0]),dist(_generators[0]));
+    }
+    template<class distribution> void fillScalar(ComplexD &s,distribution &dist)
+    {
+      s=ComplexD(dist(_generators[0]),dist(_generators[0]));
+    }
+
+
     template <class sobj,class distribution> inline void fill(sobj &l,distribution &dist){
 
       typedef typename sobj::scalar_type scalar_type;
@@ -71,13 +88,60 @@ namespace Grid {
       scalar_type *buf = (scalar_type *) & l;
 
       for(int idx=0;idx<words;idx++){
-	buf[idx] = dist(_generators[0]);
+	fillScalar(buf[idx],dist);
       }
       
       CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
 
     };
 
+    template <class distribution>  inline void fill(ComplexF &l,distribution &dist){
+      fillScalar(l,dist);
+      CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
+    }
+    template <class distribution>  inline void fill(ComplexD &l,distribution &dist){
+      fillScalar(l,dist);
+      CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
+    }
+    template <class distribution>  inline void fill(RealF &l,distribution &dist){
+      fillScalar(l,dist);
+      CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
+    }
+    template <class distribution>  inline void fill(RealD &l,distribution &dist){
+      fillScalar(l,dist);
+      CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
+    }
+    // vector fill
+    template <class distribution>  inline void fill(vComplexF &l,distribution &dist){
+      RealF *pointer=(RealF *)&l;
+      for(int i=0;i<2*vComplexF::Nsimd();i++){
+	fillScalar(pointer[i],dist);
+      }
+      CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
+    }
+    template <class distribution>  inline void fill(vComplexD &l,distribution &dist){
+      RealD *pointer=(RealD *)&l;
+      for(int i=0;i<2*vComplexD::Nsimd();i++){
+	fillScalar(pointer[i],dist);
+      }
+      CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
+    }
+    template <class distribution>  inline void fill(vRealF &l,distribution &dist){
+      RealF *pointer=(RealF *)&l;
+      for(int i=0;i<vRealF::Nsimd();i++){
+	fillScalar(pointer[i],dist);
+      }
+      CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
+    }
+    template <class distribution>  inline void fill(vRealD &l,distribution &dist){
+      RealD *pointer=(RealD *)&l;
+      for(int i=0;i<vRealD::Nsimd();i++){
+	fillScalar(pointer[i],dist);
+      }
+      CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
+    }
+
+
     void SeedRandomDevice(void){
       std::random_device rd;
       Seed(rd);
@@ -186,7 +250,6 @@ namespace Grid {
 
   };
 
-
   template <class vobj> inline void random(GridParallelRNG &rng,Lattice<vobj> &l){
     rng.fill(l,rng._uniform);
   }
diff --git a/lib/math/Grid_math_arith_scalar.h b/lib/math/Grid_math_arith_scalar.h
index 2adb2836..69c9a169 100644
--- a/lib/math/Grid_math_arith_scalar.h
+++ b/lib/math/Grid_math_arith_scalar.h
@@ -11,21 +11,21 @@ namespace Grid {
 // multiplication by fundamental scalar type
 template<class l,int N> inline iScalar<l> operator * (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
 {
-  typename iScalar<l>::tensor_reduced srhs(rhs);
+  typename iScalar<l>::tensor_reduced srhs; srhs=rhs;
   return lhs*srhs;
 }
 template<class l,int N> inline iScalar<l> operator * (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
 
 template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
 {
-  typename iVector<l,N>::tensor_reduced srhs(rhs);
+  typename iVector<l,N>::tensor_reduced srhs; srhs=rhs;
   return lhs*srhs;
 }
 template<class l,int N> inline iVector<l,N> operator * (const typename iScalar<l>::scalar_type lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
 
 template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type &rhs) 
 {
-  typename iMatrix<l,N>::tensor_reduced srhs(rhs);
+  typename iMatrix<l,N>::tensor_reduced srhs; srhs=rhs;
   return lhs*srhs;
 }
 template<class l,int N> inline iMatrix<l,N> operator * (const typename iScalar<l>::scalar_type & lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
@@ -35,24 +35,24 @@ template<class l,int N> inline iMatrix<l,N> operator * (const typename iScalar<l
 ////////////////////////////////////////////////////////////////////
 template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,double rhs) 
 {
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
+  typename iScalar<l>::scalar_type t; t=rhs;
+  typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs*srhs;
 }
 template<class l> inline iScalar<l> operator * (double lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
 
 template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,double rhs) 
 {
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
+  typename iScalar<l>::scalar_type t;t=rhs;
+  typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs*srhs;
 }
 template<class l,int N> inline iVector<l,N> operator * (double lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
 
 template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,double rhs) 
 {
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
+  typename iScalar<l>::scalar_type t;t=rhs;
+  typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs*srhs;
 }
 template<class l,int N> inline iMatrix<l,N> operator * (double lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
@@ -62,24 +62,26 @@ template<class l,int N> inline iMatrix<l,N> operator * (double lhs,const iMatrix
 ////////////////////////////////////////////////////////////////////
 template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,ComplexD rhs) 
 {
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
+  typename iScalar<l>::scalar_type t;t=rhs;
+  typename iScalar<l>::tensor_reduced srhs;srhs=t;
+  
+  
   return lhs*srhs;
 }
 template<class l> inline iScalar<l> operator * (ComplexD lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
 
 template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,ComplexD rhs) 
 {
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
+  typename iScalar<l>::scalar_type t;t=rhs;
+  typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs*srhs;
 }
 template<class l,int N> inline iVector<l,N> operator * (ComplexD lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
 
 template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,ComplexD rhs) 
 {
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
+  typename iScalar<l>::scalar_type t;t=rhs;
+  typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs*srhs;
 }
 template<class l,int N> inline iMatrix<l,N> operator * (ComplexD lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
@@ -89,24 +91,24 @@ template<class l,int N> inline iMatrix<l,N> operator * (ComplexD lhs,const iMatr
 ////////////////////////////////////////////////////////////////////
 template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,Integer rhs) 
 {
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
+  typename iScalar<l>::scalar_type t;  t=rhs;
+  typename iScalar<l>::tensor_reduced srhs; srhs=t;
   return lhs*srhs;
 }
 template<class l> inline iScalar<l> operator * (Integer lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
 
 template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,Integer rhs) 
 {
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
+  typename iScalar<l>::scalar_type t;t=rhs;
+  typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs*srhs;
 }
 template<class l,int N> inline iVector<l,N> operator * (Integer lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
 
 template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,Integer rhs) 
 {
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
+  typename iScalar<l>::scalar_type t;t=rhs;
+  typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs*srhs;
 }
 template<class l,int N> inline iMatrix<l,N> operator * (Integer lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
@@ -118,14 +120,14 @@ template<class l,int N> inline iMatrix<l,N> operator * (Integer lhs,const iMatri
 ///////////////////////////////////////////////////////////////////////////////////////////////
 template<class l,int N> inline iScalar<l> operator + (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
 {
-  typename iScalar<l>::tensor_reduced srhs(rhs);
+  typename iScalar<l>::tensor_reduced srhs; srhs=rhs;
   return lhs+srhs;
 }
 template<class l,int N> inline iScalar<l> operator + (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
 
 template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
 {
-  typename iMatrix<l,N>::tensor_reduced srhs(rhs);
+  typename iMatrix<l,N>::tensor_reduced srhs; srhs=rhs;
   return lhs+srhs;
 }
 template<class l,int N> inline iMatrix<l,N> operator + (const typename iScalar<l>::scalar_type lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
@@ -135,16 +137,16 @@ template<class l,int N> inline iMatrix<l,N> operator + (const typename iScalar<l
 ////////////////////////////////////////////////////////////////////
 template<class l> inline iScalar<l> operator + (const iScalar<l>& lhs,double rhs) 
 {
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
+  typename iScalar<l>::scalar_type t; t=rhs;
+  typename iScalar<l>::tensor_reduced srhs; srhs=t;
   return lhs+srhs;
 }
 template<class l> inline iScalar<l> operator + (double lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
 
 template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,double rhs) 
 {
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
+  typename iScalar<l>::scalar_type t;t=rhs;
+  typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs+srhs;
 }
 template<class l,int N> inline iMatrix<l,N> operator + (double lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
@@ -155,8 +157,8 @@ template<class l,int N> inline iMatrix<l,N> operator + (double lhs,const iMatrix
 
 template<class l> inline iScalar<l> operator + (const iScalar<l>& lhs,Integer rhs) 
 {
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
+  typename iScalar<l>::scalar_type t; t=rhs;
+  typename iScalar<l>::tensor_reduced srhs; srhs=t;
   return lhs+srhs;
 }
 
@@ -164,8 +166,8 @@ template<class l> inline iScalar<l> operator + (Integer lhs,const iScalar<l>& rh
 
 template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,Integer rhs) 
 {
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
+  typename iScalar<l>::scalar_type t;t=rhs;
+  typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs+srhs;
 }
 template<class l,int N> inline iMatrix<l,N> operator + (Integer lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
@@ -176,23 +178,23 @@ template<class l,int N> inline iMatrix<l,N> operator + (Integer lhs,const iMatri
 ///////////////////////////////////////////////////////////////////////////////////////////////
 template<class l,int N> inline iScalar<l> operator - (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
 {
-  typename iScalar<l>::tensor_reduced srhs(rhs);
+  typename iScalar<l>::tensor_reduced srhs; srhs=rhs;
   return lhs-srhs;
 }
 template<class l,int N> inline iScalar<l> operator - (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) 
 {
-  typename iScalar<l>::tensor_reduced slhs(lhs);
+  typename iScalar<l>::tensor_reduced slhs;slhs=lhs;
   return slhs-rhs;
 }
 
 template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
 {
-  typename iScalar<l>::tensor_reduced srhs(rhs);
+  typename iScalar<l>::tensor_reduced srhs; srhs=rhs;
   return lhs-srhs;
 }
 template<class l,int N> inline iMatrix<l,N> operator - (const typename iScalar<l>::scalar_type lhs,const iMatrix<l,N>& rhs) 
 {
-  typename iScalar<l>::tensor_reduced slhs(lhs);
+  typename iScalar<l>::tensor_reduced slhs;slhs=lhs;
   return slhs-rhs;
 }
 
@@ -201,27 +203,27 @@ template<class l,int N> inline iMatrix<l,N> operator - (const typename iScalar<l
 ////////////////////////////////////////////////////////////////////
 template<class l> inline iScalar<l> operator - (const iScalar<l>& lhs,double rhs) 
 {
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
+  typename iScalar<l>::scalar_type t; t=rhs;
+  typename iScalar<l>::tensor_reduced srhs; srhs=t;
   return lhs-srhs;
 }
 template<class l> inline iScalar<l> operator - (double lhs,const iScalar<l>& rhs) 
 {
   typename iScalar<l>::scalar_type t(lhs);
-  typename iScalar<l>::tensor_reduced slhs(t);
+  typename iScalar<l>::tensor_reduced slhs;slhs=t;
   return slhs-rhs;
 }
 
 template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,double rhs) 
 {
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
+  typename iScalar<l>::scalar_type t;t=rhs;
+  typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs-srhs;
 }
 template<class l,int N> inline iMatrix<l,N> operator - (double lhs,const iMatrix<l,N>& rhs) 
 {
   typename iScalar<l>::scalar_type t(lhs);
-  typename iScalar<l>::tensor_reduced slhs(t);
+  typename iScalar<l>::tensor_reduced slhs;slhs=t;
   return slhs-rhs;
 }
 
@@ -230,26 +232,26 @@ template<class l,int N> inline iMatrix<l,N> operator - (double lhs,const iMatrix
 ////////////////////////////////////////////////////////////////////
 template<class l> inline iScalar<l> operator - (const iScalar<l>& lhs,Integer rhs) 
 {
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
+  typename iScalar<l>::scalar_type t; t=rhs;
+  typename iScalar<l>::tensor_reduced srhs; srhs=t;
   return lhs-srhs;
 }
 template<class l> inline iScalar<l> operator - (Integer lhs,const iScalar<l>& rhs) 
 {
-  typename iScalar<l>::scalar_type t(lhs);
-  typename iScalar<l>::tensor_reduced slhs(t);
+  typename iScalar<l>::scalar_type t;t=lhs;
+  typename iScalar<l>::tensor_reduced slhs;slhs=t;
   return slhs-rhs;
 }
 template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,Integer rhs) 
 {
-  typename iScalar<l>::scalar_type t(rhs);
-  typename iScalar<l>::tensor_reduced srhs(t);
+  typename iScalar<l>::scalar_type t;t=rhs;
+  typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs-srhs;
 }
 template<class l,int N> inline iMatrix<l,N> operator - (Integer lhs,const iMatrix<l,N>& rhs) 
 {
-  typename iScalar<l>::scalar_type t(lhs);
-  typename iScalar<l>::tensor_reduced slhs(t);
+  typename iScalar<l>::scalar_type t;t=lhs;
+  typename iScalar<l>::tensor_reduced slhs;slhs=t;
   return slhs-rhs;
 }
 
diff --git a/lib/math/Grid_math_inner.h b/lib/math/Grid_math_inner.h
index 71b5e3b8..1e44032c 100644
--- a/lib/math/Grid_math_inner.h
+++ b/lib/math/Grid_math_inner.h
@@ -17,7 +17,8 @@ namespace Grid {
     auto innerProduct (const iVector<l,N>& lhs,const iVector<r,N>& rhs) -> iScalar<decltype(innerProduct(lhs._internal[0],rhs._internal[0]))>
     {
         typedef decltype(innerProduct(lhs._internal[0],rhs._internal[0])) ret_t;
-        iScalar<ret_t> ret=zero;
+        iScalar<ret_t> ret;
+	ret=zero;
         for(int c1=0;c1<N;c1++){
             ret._internal += innerProduct(lhs._internal[c1],rhs._internal[c1]);
         }
@@ -27,8 +28,9 @@ namespace Grid {
     auto innerProduct (const iMatrix<l,N>& lhs,const iMatrix<r,N>& rhs) -> iScalar<decltype(innerProduct(lhs._internal[0][0],rhs._internal[0][0]))>
     {
         typedef decltype(innerProduct(lhs._internal[0][0],rhs._internal[0][0])) ret_t;
-        iScalar<ret_t> ret=zero;
+        iScalar<ret_t> ret;
         iScalar<ret_t> tmp;
+	ret=zero;
         for(int c1=0;c1<N;c1++){
         for(int c2=0;c2<N;c2++){
 	  ret._internal+=innerProduct(lhs._internal[c1][c2],rhs._internal[c1][c2]);
diff --git a/lib/math/Grid_math_tensors.h b/lib/math/Grid_math_tensors.h
index 8bca6805..cf5a5293 100644
--- a/lib/math/Grid_math_tensors.h
+++ b/lib/math/Grid_math_tensors.h
@@ -8,6 +8,16 @@ namespace Grid {
 // These can be composed to form tensor products of internal indices.
 ///////////////////////////////////////////////////
 
+// It is useful to NOT have any constructors
+// so that these classes assert "is_pod<class> == true"
+// because then the standard C++ valarray container eliminates fill overhead on new allocation and 
+// non-move copying.
+//
+// However note that doing this eliminates some syntactical sugar such as 
+// calling the constructor explicitly or implicitly
+//
+#define TENSOR_IS_POD
+
 template<class vtype> class iScalar
 {
 public:
@@ -25,16 +35,22 @@ public:
   // Scalar no action
   //  template<int Level> using tensor_reduce_level = typename iScalar<GridTypeMapper<vtype>::tensor_reduce_level<Level> >;
 
-    iScalar(){};
-    
-    iScalar(scalar_type s) : _internal(s) {};// recurse down and hit the constructor for vector_type
-
-    iScalar(const Zero &z){ *this = zero; };
+#ifndef TENSOR_IS_POD
+  iScalar(){;};
+  iScalar(scalar_type s) : _internal(s) {};// recurse down and hit the constructor for vector_type
+  iScalar(const Zero &z){ *this = zero; };
+#endif
 
     iScalar<vtype> & operator= (const Zero &hero){
-        zeroit(*this);
-        return *this;
+      zeroit(*this);
+      return *this;
     }
+    iScalar<vtype> & operator= (const scalar_type s){
+      _internal=s;
+      return *this;
+    }
+
+
     friend void zeroit(iScalar<vtype> &that){
         zeroit(that._internal);
     }
@@ -114,8 +130,10 @@ public:
 
   enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
 
-    iVector(const Zero &z){ *this = zero; };
-    iVector() {};// Empty constructure
+#ifndef TENSOR_IS_POD
+  iVector(const Zero &z){ *this = zero; };
+  iVector() {};// Empty constructure
+#endif
 
     iVector<vtype,N> & operator= (const Zero &hero){
         zeroit(*this);
@@ -185,8 +203,11 @@ public:
 
   enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
 
+#ifndef TENSOR_IS_POD
   iMatrix(const Zero &z){ *this = zero; };
   iMatrix() {};
+#endif
+
   iMatrix<vtype,N> & operator= (const Zero &hero){
     zeroit(*this);
     return *this;
diff --git a/lib/qcd/Grid_qcd.h b/lib/qcd/Grid_qcd.h
index f331bd88..0bf68632 100644
--- a/lib/qcd/Grid_qcd.h
+++ b/lib/qcd/Grid_qcd.h
@@ -8,6 +8,7 @@ namespace QCD {
     static const int Ns=4;
     static const int Nd=4;
     static const int Nhs=2; // half spinor
+    static const int Nds=8; // double stored gauge field
 
     static const int CbRed  =0;
     static const int CbBlack=1;
@@ -28,79 +29,216 @@ namespace QCD {
     //
     // That probably makes for GridRedBlack4dCartesian grid.
 
-    template<typename vtype> using iSinglet          = iScalar<iScalar<iScalar<vtype> > >;
-    template<typename vtype> using iSpinMatrix       = iScalar<iMatrix<iScalar<vtype>, Ns> >;
-    template<typename vtype> using iSpinColourMatrix = iScalar<iMatrix<iMatrix<vtype, Nc>, Ns> >;
-    template<typename vtype> using iColourMatrix     = iScalar<iScalar<iMatrix<vtype, Nc> > > ;
-    template<typename vtype> using iLorentzColourMatrix = iVector<iScalar<iMatrix<vtype, Nc> >, Nd > ;
+    // s,sp,c,spc,lc
+    template<typename vtype> using iSinglet                   = iScalar<iScalar<iScalar<vtype> > >;
+    template<typename vtype> using iSpinMatrix                = iScalar<iMatrix<iScalar<vtype>, Ns> >;
+    template<typename vtype> using iColourMatrix              = iScalar<iScalar<iMatrix<vtype, Nc> > > ;
+    template<typename vtype> using iSpinColourMatrix          = iScalar<iMatrix<iMatrix<vtype, Nc>, Ns> >;
+    template<typename vtype> using iLorentzColourMatrix       = iVector<iScalar<iMatrix<vtype, Nc> >, Nd > ;
+    template<typename vtype> using iDoubleStoredColourMatrix  = iVector<iScalar<iMatrix<vtype, Nc> >, Nds > ;
+    template<typename vtype> using iSpinVector                = iScalar<iVector<iScalar<vtype>, Ns> >;
+    template<typename vtype> using iColourVector              = iScalar<iScalar<iVector<vtype, Nc> > >;
+    template<typename vtype> using iSpinColourVector          = iScalar<iVector<iVector<vtype, Nc>, Ns> >;
+    template<typename vtype> using iHalfSpinVector            = iScalar<iVector<iScalar<vtype>, Nhs> >;
+    template<typename vtype> using iHalfSpinColourVector      = iScalar<iVector<iVector<vtype, Nc>, Nhs> >;
 
+    // Spin matrix
+    typedef iSpinMatrix<Complex  >          SpinMatrix;
+    typedef iSpinMatrix<ComplexF >          SpinMatrixF;
+    typedef iSpinMatrix<ComplexD >          SpinMatrixD;
 
-    template<typename vtype> using iSpinVector       = iScalar<iVector<iScalar<vtype>, Ns> >;
-    template<typename vtype> using iColourVector     = iScalar<iScalar<iVector<vtype, Nc> > >;
-    template<typename vtype> using iSpinColourVector = iScalar<iVector<iVector<vtype, Nc>, Ns> >;
+    typedef iSpinMatrix<vComplex >          vSpinMatrix;
+    typedef iSpinMatrix<vComplexF>          vSpinMatrixF;
+    typedef iSpinMatrix<vComplexD>          vSpinMatrixD;
 
-    template<typename vtype> using iHalfSpinVector       = iScalar<iVector<iScalar<vtype>, Nhs> >;
-    template<typename vtype> using iHalfSpinColourVector = iScalar<iVector<iVector<vtype, Nc>, Nhs> >;
+    // Colour Matrix
+    typedef iColourMatrix<Complex  >        ColourMatrix;
+    typedef iColourMatrix<ComplexF >        ColourMatrixF;
+    typedef iColourMatrix<ComplexD >        ColourMatrixD;
 
-    typedef iSpinMatrix<Complex >           SpinMatrix;
-    typedef iColourMatrix<Complex >         ColourMatrix;
-    typedef iSpinColourMatrix<Complex >     SpinColourMatrix;
-    typedef iLorentzColourMatrix<Complex >  LorentzColourMatrix;
+    typedef iColourMatrix<vComplex >        vColourMatrix;
+    typedef iColourMatrix<vComplexF>        vColourMatrixF;
+    typedef iColourMatrix<vComplexD>        vColourMatrixD;
+
+    // SpinColour matrix
+    typedef iSpinColourMatrix<Complex  >    SpinColourMatrix;
+    typedef iSpinColourMatrix<ComplexF >    SpinColourMatrixF;
+    typedef iSpinColourMatrix<ComplexD >    SpinColourMatrixD;
+
+    typedef iSpinColourMatrix<vComplex >    vSpinColourMatrix;
+    typedef iSpinColourMatrix<vComplexF>    vSpinColourMatrixF;
+    typedef iSpinColourMatrix<vComplexD>    vSpinColourMatrixD;
+
+    // LorentzColour
+    typedef iLorentzColourMatrix<Complex  > LorentzColourMatrix;
     typedef iLorentzColourMatrix<ComplexF > LorentzColourMatrixF;
     typedef iLorentzColourMatrix<ComplexD > LorentzColourMatrixD;
 
-    typedef iSpinVector<Complex >           SpinVector;
-    typedef iColourVector<Complex >         ColourVector;
-    typedef iSpinColourVector<Complex >     SpinColourVector;
-    typedef iHalfSpinVector<Complex >       HalfSpinVector;
-    typedef iHalfSpinColourVector<Complex > HalfSpinColourVector;
-
-    
-    typedef iSpinMatrix<vComplex >          vSpinMatrix;
-    typedef iColourMatrix<vComplex >        vColourMatrix;
-    typedef iSpinColourMatrix<vComplex >    vSpinColourMatrix;
     typedef iLorentzColourMatrix<vComplex > vLorentzColourMatrix;
-    
+    typedef iLorentzColourMatrix<vComplexF> vLorentzColourMatrixF;
+    typedef iLorentzColourMatrix<vComplexD> vLorentzColourMatrixD;
+
+    // DoubleStored gauge field
+    typedef iDoubleStoredColourMatrix<Complex  > DoubleStoredColourMatrix;
+    typedef iDoubleStoredColourMatrix<ComplexF > DoubleStoredColourMatrixF;
+    typedef iDoubleStoredColourMatrix<ComplexD > DoubleStoredColourMatrixD;
+
+    typedef iDoubleStoredColourMatrix<vComplex > vDoubleStoredColourMatrix;
+    typedef iDoubleStoredColourMatrix<vComplexF> vDoubleStoredColourMatrixF;
+    typedef iDoubleStoredColourMatrix<vComplexD> vDoubleStoredColourMatrixD;
+
+    // Spin vector
+    typedef iSpinVector<Complex >           SpinVector;
+    typedef iSpinVector<ComplexF>           SpinVectorF;
+    typedef iSpinVector<ComplexD>           SpinVectorD;
+
     typedef iSpinVector<vComplex >           vSpinVector;
+    typedef iSpinVector<vComplexF>           vSpinVectorF;
+    typedef iSpinVector<vComplexD>           vSpinVectorD;
+
+    // Colour vector
+    typedef iColourVector<Complex >         ColourVector;
+    typedef iColourVector<ComplexF>         ColourVectorF;
+    typedef iColourVector<ComplexD>         ColourVectorD;
+
     typedef iColourVector<vComplex >         vColourVector;
+    typedef iColourVector<vComplexF>         vColourVectorF;
+    typedef iColourVector<vComplexD>         vColourVectorD;
+
+    // SpinColourVector
+    typedef iSpinColourVector<Complex >     SpinColourVector;
+    typedef iSpinColourVector<ComplexF>     SpinColourVectorF;
+    typedef iSpinColourVector<ComplexD>     SpinColourVectorD;
+
     typedef iSpinColourVector<vComplex >     vSpinColourVector;
+    typedef iSpinColourVector<vComplexF>     vSpinColourVectorF;
+    typedef iSpinColourVector<vComplexD>     vSpinColourVectorD;
+
+    // HalfSpin vector
+    typedef iHalfSpinVector<Complex >       HalfSpinVector;
+    typedef iHalfSpinVector<ComplexF>       HalfSpinVectorF;
+    typedef iHalfSpinVector<ComplexD>       HalfSpinVectorD;
+
     typedef iHalfSpinVector<vComplex >       vHalfSpinVector;
-    typedef iHalfSpinColourVector<vComplex > vHalfSpinColourVector;
+    typedef iHalfSpinVector<vComplexF>       vHalfSpinVectorF;
+    typedef iHalfSpinVector<vComplexD>       vHalfSpinVectorD;
+
+    // HalfSpinColour vector
+    typedef iHalfSpinColourVector<Complex > HalfSpinColourVector;
+    typedef iHalfSpinColourVector<ComplexF> HalfSpinColourVectorF;
+    typedef iHalfSpinColourVector<ComplexD> HalfSpinColourVectorD;
     
+    typedef iHalfSpinColourVector<vComplex > vHalfSpinColourVector;
+    typedef iHalfSpinColourVector<vComplexF> vHalfSpinColourVectorF;
+    typedef iHalfSpinColourVector<vComplexD> vHalfSpinColourVectorD;
+    
+    // singlets
     typedef iSinglet<Complex >         TComplex;    // FIXME This is painful. Tensor singlet complex type.
-    typedef iSinglet<vComplex >        vTComplex;   // what if we don't know the tensor structure
+    typedef iSinglet<ComplexF>         TComplexF;    // FIXME This is painful. Tensor singlet complex type.
+    typedef iSinglet<ComplexD>         TComplexD;    // FIXME This is painful. Tensor singlet complex type.
+
+    typedef iSinglet<vComplex >        vTComplex ;   // what if we don't know the tensor structure
+    typedef iSinglet<vComplexF>        vTComplexF;   // what if we don't know the tensor structure
+    typedef iSinglet<vComplexD>        vTComplexD;   // what if we don't know the tensor structure
+
     typedef iSinglet<Real >            TReal;       // Shouldn't need these; can I make it work without?
+    typedef iSinglet<RealF>            TRealF;       // Shouldn't need these; can I make it work without?
+    typedef iSinglet<RealD>            TRealD;       // Shouldn't need these; can I make it work without?
+
     typedef iSinglet<vReal >           vTReal;      
-    typedef iSinglet<vInteger >        vTInteger;
+    typedef iSinglet<vRealF>           vTRealF;      
+    typedef iSinglet<vRealD>           vTRealD;      
+
+    typedef iSinglet<vInteger>         vTInteger;
     typedef iSinglet<Integer >         TInteger;
 
-    typedef Lattice<vTReal>            LatticeReal;
-    typedef Lattice<vTComplex>         LatticeComplex;
-    typedef Lattice<vTInteger>         LatticeInteger; // Predicates for "where"
-    
-    typedef Lattice<vColourMatrix>     LatticeColourMatrix;
-    typedef Lattice<vSpinMatrix>       LatticeSpinMatrix;
-    typedef Lattice<vSpinColourMatrix> LatticeSpinColourMatrix;
 
-    typedef Lattice<vSpinVector>           LatticeSpinVector;
-    typedef Lattice<vColourVector>         LatticeColourVector;
-    typedef Lattice<vSpinColourVector>     LatticeSpinColourVector;
-    typedef Lattice<vHalfSpinVector>       LatticeHalfSpinVector;
-    typedef Lattice<vHalfSpinColourVector> LatticeHalfSpinColourVector;
+    // Lattices of these
+    typedef Lattice<vColourMatrix>          LatticeColourMatrix;
+    typedef Lattice<vColourMatrixF>         LatticeColourMatrixF;
+    typedef Lattice<vColourMatrixD>         LatticeColourMatrixD;
+
+    typedef Lattice<vSpinMatrix>            LatticeSpinMatrix;
+    typedef Lattice<vSpinMatrixF>           LatticeSpinMatrixF;
+    typedef Lattice<vSpinMatrixD>           LatticeSpinMatrixD;
+
+    typedef Lattice<vSpinColourMatrix>      LatticeSpinColourMatrix;
+    typedef Lattice<vSpinColourMatrixF>     LatticeSpinColourMatrixF;
+    typedef Lattice<vSpinColourMatrixD>     LatticeSpinColourMatrixD;
+
+
+    typedef Lattice<vLorentzColourMatrix>  LatticeLorentzColourMatrix;
+    typedef Lattice<vLorentzColourMatrixF> LatticeLorentzColourMatrixF;
+    typedef Lattice<vLorentzColourMatrixD> LatticeLorentzColourMatrixD;
+
+    // DoubleStored gauge field
+    typedef Lattice<vDoubleStoredColourMatrix>  LatticeDoubleStoredColourMatrix;
+    typedef Lattice<vDoubleStoredColourMatrixF> LatticeDoubleStoredColourMatrixF;
+    typedef Lattice<vDoubleStoredColourMatrixD> LatticeDoubleStoredColourMatrixD;
+
+    typedef Lattice<vSpinVector>            LatticeSpinVector;
+    typedef Lattice<vSpinVectorF>           LatticeSpinVectorF;
+    typedef Lattice<vSpinVectorD>           LatticeSpinVectorD;
+
+    typedef Lattice<vColourVector>          LatticeColourVector;
+    typedef Lattice<vColourVectorF>         LatticeColourVectorF;
+    typedef Lattice<vColourVectorD>         LatticeColourVectorD;
+
+    typedef Lattice<vSpinColourVector>      LatticeSpinColourVector;
+    typedef Lattice<vSpinColourVectorF>     LatticeSpinColourVectorF;
+    typedef Lattice<vSpinColourVectorD>     LatticeSpinColourVectorD;
+
+    typedef Lattice<vHalfSpinVector>        LatticeHalfSpinVector;
+    typedef Lattice<vHalfSpinVectorF>       LatticeHalfSpinVectorF;
+    typedef Lattice<vHalfSpinVectorD>       LatticeHalfSpinVectorD;
+
+    typedef Lattice<vHalfSpinColourVector>  LatticeHalfSpinColourVector;
+    typedef Lattice<vHalfSpinColourVectorF> LatticeHalfSpinColourVectorF;
+    typedef Lattice<vHalfSpinColourVectorD> LatticeHalfSpinColourVectorD;
+
+    typedef Lattice<vTReal>            LatticeReal;
+    typedef Lattice<vTRealF>           LatticeRealF;
+    typedef Lattice<vTRealD>           LatticeRealD;
+
+    typedef Lattice<vTComplex>         LatticeComplex;
+    typedef Lattice<vTComplexF>        LatticeComplexF;
+    typedef Lattice<vTComplexD>        LatticeComplexD;
+
+    typedef Lattice<vTInteger>         LatticeInteger; // Predicates for "where"
+
 
     ///////////////////////////////////////////
     // Physical names for things
     ///////////////////////////////////////////
-    typedef Lattice<vHalfSpinColourVector> LatticeHalfFermion;
-    typedef Lattice<vSpinColourVector>     LatticeFermion;
+    typedef LatticeHalfSpinColourVector  LatticeHalfFermion;
+    typedef LatticeHalfSpinColourVectorF LatticeHalfFermionF;
+    typedef LatticeHalfSpinColourVectorF LatticeHalfFermionD;
 
-    typedef Lattice<vSpinColourMatrix>     LatticePropagator;
-    typedef Lattice<vLorentzColourMatrix>  LatticeGaugeField;
+    typedef LatticeSpinColourVector      LatticeFermion;
+    typedef LatticeSpinColourVectorF     LatticeFermionF;
+    typedef LatticeSpinColourVectorD     LatticeFermionD;
+
+    typedef LatticeSpinColourMatrix                LatticePropagator;
+    typedef LatticeSpinColourMatrixF               LatticePropagatorF;
+    typedef LatticeSpinColourMatrixD               LatticePropagatorD;
+
+    typedef LatticeLorentzColourMatrix             LatticeGaugeField;
+    typedef LatticeLorentzColourMatrixF            LatticeGaugeFieldF;
+    typedef LatticeLorentzColourMatrixD            LatticeGaugeFieldD;
+
+    typedef LatticeDoubleStoredColourMatrix        LatticeDoubledGaugeField;
+    typedef LatticeDoubleStoredColourMatrixF       LatticeDoubledGaugeFieldF;
+    typedef LatticeDoubleStoredColourMatrixD       LatticeDoubledGaugeFieldD;
 
     // Uhgg... typing this hurt  ;)
     // (my keyboard got burning hot when I typed this, must be the anti-Fermion)
-    typedef Lattice<vColourVector>         LatticeStaggeredFermion;    
-    typedef Lattice<vColourMatrix>         LatticeStaggeredPropagator; 
+    typedef Lattice<vColourVector>          LatticeStaggeredFermion;    
+    typedef Lattice<vColourVectorF>         LatticeStaggeredFermionF;    
+    typedef Lattice<vColourVectorD>         LatticeStaggeredFermionD;    
+
+    typedef Lattice<vColourMatrix>          LatticeStaggeredPropagator; 
+    typedef Lattice<vColourMatrixF>         LatticeStaggeredPropagatorF; 
+    typedef Lattice<vColourMatrixD>         LatticeStaggeredPropagatorD; 
 
     //////////////////////////////////////////////////////////////////////////////
     // Peek and Poke named after physics attributes
@@ -157,11 +295,14 @@ namespace QCD {
       return peekIndex<LorentzIndex>(rhs,i,j);
     }
 
+    // FIXME transpose Colour, transpose Spin, traceColour traceSpin
+
 }   //namespace QCD
 } // Grid
 
 #include <qcd/Grid_qcd_dirac.h>
 #include <qcd/Grid_qcd_2spinor.h>
 //#include <qcd/Grid_qcd_pauli.h>
+#include <qcd/Grid_qcd_wilson_dop.h>
 
 #endif
diff --git a/lib/qcd/Grid_qcd_dirac.h b/lib/qcd/Grid_qcd_dirac.h
index 03dc9cc5..2623e683 100644
--- a/lib/qcd/Grid_qcd_dirac.h
+++ b/lib/qcd/Grid_qcd_dirac.h
@@ -17,8 +17,14 @@ namespace QCD {
       GammaZ, 
       GammaT,  
       Gamma5,
-      //      GammaXGamma5, 
-      //      GammaYGamma5, 
+      MinusIdentity,                        
+      MinusGammaX, 
+      MinusGammaY, 
+      MinusGammaZ, 
+      MinusGammaT,  
+      MinusGamma5
+      //      GammaXGamma5, // Rest are composite (willing to take hit for two calls sequentially)
+      //      GammaYGamma5, // as they are less commonly used.
       //      GammaZGamma5, 
       //      GammaTGamma5,  
       //      SigmaXY,
@@ -27,12 +33,6 @@ namespace QCD {
       //      SigmaXT,
       //      SigmaYT,
       //      SigmaZT,
-      MinusIdentity,                        
-      MinusGammaX, 
-      MinusGammaY, 
-      MinusGammaZ, 
-      MinusGammaT,  
-      MinusGamma5
       //      MinusGammaXGamma5, easiest to form by composition
       //      MinusGammaYGamma5, as performance is not critical for these
       //      MinusGammaZGamma5, 
@@ -54,7 +54,6 @@ namespace QCD {
 
   };
 
-
     /* Gx
      *  0 0  0  i    
      *  0 0  i  0    
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index 73b19c12..aba21417 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -1,157 +1,220 @@
-#ifnfdef GRID_QCD_WILSON_DOP_H
-#define  GRID_QCD_WILSON_DOP_H
 
 #include <Grid.h>
 
 namespace Grid {
 namespace QCD {
 
-
 const std::vector<int> WilsonMatrix::directions   ({0,1,2,3, 0, 1, 2, 3,0});
 const std::vector<int> WilsonMatrix::displacements({1,1,1,1,-1,-1,-1,-1,0});
 
   // Should be in header?
-static const int WilsonMatrix::Xp = 0;
-static const int WilsonMatrix::Yp = 1;
-static const int WilsonMatrix::Zp = 2;
-static const int WilsonMatrix::Tp = 3;
-static const int WilsonMatrix::Xm = 4;
-static const int WilsonMatrix::Ym = 5;
-static const int WilsonMatrix::Zm = 6;
-static const int WilsonMatrix::Tm = 7;
-static const int WilsonMatrix::X0 = 8;
-static const int WilsonMatrix::npoint=9;
+const int WilsonMatrix::Xp = 0;
+const int WilsonMatrix::Yp = 1;
+const int WilsonMatrix::Zp = 2;
+const int WilsonMatrix::Tp = 3;
+const int WilsonMatrix::Xm = 4;
+const int WilsonMatrix::Ym = 5;
+const int WilsonMatrix::Zm = 6;
+const int WilsonMatrix::Tm = 7;
+  //const int WilsonMatrix::X0 = 8;
 
+  class WilsonCompressor {
+  public:
+    int mu;
+    
+    void Point(int p) { mu=p;};
+    vHalfSpinColourVector operator () (vSpinColourVector &in)
+    {
+      vHalfSpinColourVector ret;
+      switch(mu) {
+      case WilsonMatrix::Xp:
+	spProjXp(ret,in);
+	break;
+      case WilsonMatrix::Yp:
+	spProjYp(ret,in);
+	break;
+      case WilsonMatrix::Zp:
+	spProjZp(ret,in);
+	break;
+      case WilsonMatrix::Tp:
+	spProjTp(ret,in);
+	break;
+      case WilsonMatrix::Xm:
+	spProjXm(ret,in);
+	break;
+      case WilsonMatrix::Ym:
+	spProjYm(ret,in);
+	break;
+      case WilsonMatrix::Zm:
+	spProjZm(ret,in);
+	break;
+      case WilsonMatrix::Tm:
+	spProjTm(ret,in);
+	break;
+      default: 
+	assert(0);
+	break;
+      }
+      return ret;
+    }
+  };
 
-WilsonMatrix::WilsonMatrix(LatticeGaugeField &_Umu,int _mass)
-  : Stencil((&Umu._grid,npoint,0,directions,displacements),
+  WilsonMatrix::WilsonMatrix(LatticeGaugeField &_Umu,double _mass)
+  : Stencil(Umu._grid,npoint,0,directions,displacements),
     mass(_mass),
-    Umu(_Umu)
+    Umu(_Umu._grid)
 {
   // Allocate the required comms buffer
   grid = _Umu._grid;
   comm_buf.resize(Stencil._unified_buffer_size);
+  DoubleStore(Umu,_Umu);
 }
+      
+void WilsonMatrix::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu)
+{
+  LatticeColourMatrix U(grid);
+
+  for(int mu=0;mu<Nd;mu++){
+    U = peekIndex<LorentzIndex>(Umu,mu);
+    pokeIndex<LorentzIndex>(Uds,U,mu);
+    U = adj(Cshift(U,mu,-1));
+    pokeIndex<LorentzIndex>(Uds,U,mu+4);
+  }
+}
+
 void WilsonMatrix::multiply(const LatticeFermion &in, LatticeFermion &out)
 {
-  
+  Dhop(in,out);
+  return;
 }
+
 void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
 {
-  Stencil.HaloExchange(in,comm_buf);
+  //  Stencil.HaloExchange(in,comm_buf);
 
-  for(int ss=0;ss<_grid->oSites();ss++){
+  for(int ss=0;ss<grid->oSites();ss++){
 
     int offset,local;
 
     vSpinColourVector result;
-    vHalfSpinColourVector UChi;
-
+    vHalfSpinColourVector  chi;    
+    vHalfSpinColourVector Uchi;
+    vHalfSpinColourVector *chi_p;
     // Xp
     offset = Stencil._offsets [Xp][ss];
     local  = Stencil._is_local[Xp][ss];
-    if ( local ) {
-      Uchi = U[]*spProjXp(in._odata[offset]);
-    } else {
-      Uchi = U[]*comm_buf._odata[offset]
-    }
-    result = ReconXp(Uchi);
 
+    chi_p  = &comm_buf[offset];
+    if ( local ) {
+      spProjXp(chi,in._odata[offset]);
+      chi_p = &chi;
+    } 
+    mult(&(Uchi()),&(Umu._odata[ss](Xp)),&(*chi_p)());
+    spReconXp(result,Uchi);
+
+#if 0
     // Yp
     offset = Stencil._offsets [Yp][ss];
     local  = Stencil._is_local[Yp][ss];
+    chi_p  = &comm_buf[offset];
     if ( local ) {
-      Uchi = U[]*spProjYp(in._odata[offset]);
-    } else {
-      Uchi = U[]*comm_buf._odata[offset]
-    }
-    result+= ReconYp(Uchi);
+      spProjYp(chi,in._odata[offset]);
+      chi_p = &chi;
+    } 
+    mult(&(Uchi()),&(Umu._odata[ss](Yp)),&(*chi_p)());
+    accumReconYp(result,Uchi);
 
     // Zp
     offset = Stencil._offsets [Zp][ss];
     local  = Stencil._is_local[Zp][ss];
+    chi_p  = &comm_buf[offset];
     if ( local ) {
-      Uchi = U[]*spProjZp(in._odata[offset]);
-    } else {
-      Uchi = U[]*comm_buf._odata[offset]
-    }
-    result+= ReconZp(Uchi);
+      spProjZp(chi,in._odata[offset]);
+      chi_p = &chi;
+    } 
+    mult(&(Uchi()),&(Umu._odata[ss](Zp)),&(*chi_p)() );
+    accumReconZp(result,Uchi);
 
     // Tp
     offset = Stencil._offsets [Tp][ss];
     local  = Stencil._is_local[Tp][ss];
+    chi_p  = &comm_buf[offset];
     if ( local ) {
-      Uchi = U[]*spProjTp(in._odata[offset]);
-    } else {
-      Uchi = U[]*comm_buf._odata[offset]
-    }
-    result+= ReconTp(Uchi);
+      spProjTp(chi,in._odata[offset]);
+      chi_p = &chi;
+    } 
+    mult(&(Uchi()),&(Umu._odata[ss](Tp)),&(*chi_p)());
+    accumReconTp(result,Uchi);
 
     // Xm
     offset = Stencil._offsets [Xm][ss];
     local  = Stencil._is_local[Xm][ss];
+    chi_p  = &comm_buf[offset];
     if ( local ) {
-      Uchi = U[]*spProjXm(in._odata[offset]);
-    } else {
-      Uchi = U[]*comm_buf._odata[offset]
-    }
-    result+= ReconXm(Uchi);
+      spProjXm(chi,in._odata[offset]);
+      chi_p = &chi;
+    } 
+    mult(&(Uchi()),&(Umu._odata[ss](Xm)),&(*chi_p)());
+    accumReconXm(result,Uchi);
 
     // Ym
     offset = Stencil._offsets [Ym][ss];
     local  = Stencil._is_local[Ym][ss];
+    chi_p  = &comm_buf[offset];
     if ( local ) {
-      Uchi = U[]*spProjYm(in._odata[offset]);
-    } else {
-      Uchi = U[]*comm_buf._odata[offset]
-    }
-    result+= ReconYm(Uchi);
+      spProjYm(chi,in._odata[offset]);
+      chi_p = &chi;
+    } 
+    mult(&(Uchi()),&(Umu._odata[ss](Ym)),&(*chi_p)());
+    accumReconYm(result,Uchi);
 
     // Zm
     offset = Stencil._offsets [Zm][ss];
     local  = Stencil._is_local[Zm][ss];
+    chi_p  = &comm_buf[offset];
     if ( local ) {
-      Uchi = U[]*spProjZm(in._odata[offset]);
-    } else {
-      Uchi = U[]*comm_buf._odata[offset]
-    }
-    result+= ReconZm(Uchi);
+      spProjZm(chi,in._odata[offset]);
+      chi_p = &chi;
+    } 
+    mult(&(Uchi()),&(Umu._odata[ss](Zm)),&(*chi_p)());
+    accumReconZm(result,Uchi);
 
     // Tm
     offset = Stencil._offsets [Tm][ss];
     local  = Stencil._is_local[Tm][ss];
+    chi_p  = &comm_buf[offset];
     if ( local ) {
-      Uchi = U[]*spProjTm(in._odata[offset]);
-    } else {
-      Uchi = U[]*comm_buf._odata[offset]
-    }
-    result+= ReconTm(Uchi);
-
+      spProjTm(chi,in._odata[offset]);
+      chi_p = &chi;
+    } 
+    mult(&(Uchi()),&(Umu._odata[ss](Tm)),&(*chi_p)());
+    accumReconTm(result,Uchi);
+#endif
     out._odata[ss] = result;
   }
 
 }
 void WilsonMatrix::Dw(const LatticeFermion &in, LatticeFermion &out)
 {
-
+  return;
 }
 void WilsonMatrix::MpcDag   (const LatticeFermion &in, LatticeFermion &out)
 {
-
+  return;
 }
 void WilsonMatrix::Mpc      (const LatticeFermion &in, LatticeFermion &out)
 {
-
+  return;
 }
 void WilsonMatrix::MpcDagMpc(const LatticeFermion &in, LatticeFermion &out)
 {
-
+  return;
 }
 void WilsonMatrix::MDagM    (const LatticeFermion &in, LatticeFermion &out)
 {
-
+  return;
 }
 
 
 }}
-#endif
+
diff --git a/lib/qcd/Grid_qcd_wilson_dop.h b/lib/qcd/Grid_qcd_wilson_dop.h
index f6775842..dc3c80b9 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.h
+++ b/lib/qcd/Grid_qcd_wilson_dop.h
@@ -1,4 +1,4 @@
-#ifnfdef GRID_QCD_WILSON_DOP_H
+#ifndef  GRID_QCD_WILSON_DOP_H
 #define  GRID_QCD_WILSON_DOP_H
 
 #include <Grid.h>
@@ -21,21 +21,23 @@ namespace Grid {
       GridBase                     *grid;
 
       // Copy of the gauge field 
-      LatticeGaugeField             Umu;
+      LatticeDoubledGaugeField             Umu;
 
       //Defines the stencil
       CartesianStencil              Stencil; 
       static const int npoint=9;
       static const std::vector<int> directions   ;
       static const std::vector<int> displacements;
-
       static const int Xp,Xm,Yp,Ym,Zp,Zm,Tp,Tm;
 
       // Comms buffer
-      std::vector<vSpinColourVector,alignedAllocator<vSpinColourVector> >  comm_buf;
+      std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  comm_buf;
 
       // Constructor
-      WilsonMatrix(LatticeGaugeField &Umu,int mass);
+      WilsonMatrix(LatticeGaugeField &Umu,double mass);
+
+      // DoubleStore
+      void DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu);
 
       // override multiply
       void multiply(const LatticeFermion &in, LatticeFermion &out);
diff --git a/lib/simd/Grid_vComplexD.h b/lib/simd/Grid_vComplexD.h
index 402d303c..76295fdf 100644
--- a/lib/simd/Grid_vComplexD.h
+++ b/lib/simd/Grid_vComplexD.h
@@ -251,13 +251,13 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
         friend inline vComplexD conj(const vComplexD &in){
             vComplexD ret ; vzero(ret);
 #if defined (AVX1)|| defined (AVX2)
-            // addsubps 0, inv=>0+in.v[3] 0-in.v[2], 0+in.v[1], 0-in.v[0], ...
-	    //            __m256d tmp = _mm256_addsub_pd(ret.v,_mm256_shuffle_pd(in.v,in.v,0x5));
-	    //             ret.v=_mm256_shuffle_pd(tmp,tmp,0x5);
-            ret.v = _mm256_addsub_pd(ret.v,in.v);
+	    //	    addsubps 0, inv=>0+in.v[3] 0-in.v[2], 0+in.v[1], 0-in.v[0], ...
+	    zvec tmp = _mm256_addsub_pd(ret.v,_mm256_shuffle_pd(in.v,in.v,0x5));
+	    ret.v    =_mm256_shuffle_pd(tmp,tmp,0x5);
 #endif
 #ifdef SSE4
-            ret.v = _mm_addsub_pd(ret.v,in.v);
+	    zvec tmp = _mm_addsub_pd(ret.v,_mm_shuffle_pd(in.v,in.v,0x1));
+	    ret.v    = _mm_shuffle_pd(tmp,tmp,0x1);
 #endif
 #ifdef AVX512
 	    ret.v = _mm512_mask_sub_pd(in.v, 0xaaaa,ret.v, in.v);             
@@ -268,48 +268,41 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
             return ret;
         }
 
-        friend inline vComplexD timesI(const vComplexD &in){
+        friend inline vComplexD timesMinusI(const vComplexD &in){
 	  vComplexD ret; vzero(ret);
+	  vComplexD tmp;
 #if defined (AVX1)|| defined (AVX2)
-	  cvec tmp =_mm256_addsub_ps(ret.v,in.v); // r,-i
-	  /*
-             IF IMM0[0] = 0
-             THEN DEST[63:0]=SRC1[63:0] ELSE DEST[63:0]=SRC1[127:64] FI;
-             IF IMM0[1] = 0
-             THEN DEST[127:64]=SRC2[63:0] ELSE DEST[127:64]=SRC2[127:64] FI;
-             IF IMM0[2] = 0
-             THEN DEST[191:128]=SRC1[191:128] ELSE DEST[191:128]=SRC1[255:192] FI;
-             IF IMM0[3] = 0
-             THEN DEST[255:192]=SRC2[191:128] ELSE DEST[255:192]=SRC2[255:192] FI;
-	  */
-          ret.v    =_mm256_shuffle_ps(tmp,tmp,0x5);
+	  tmp.v    =_mm256_addsub_pd(ret.v,in.v); // r,-i
+          ret.v    =_mm256_shuffle_pd(tmp.v,tmp.v,0x5);
 #endif
 #ifdef SSE4
-	  cvec tmp =_mm_addsub_ps(ret.v,in.v); // r,-i
-          ret.v    =_mm_shuffle_ps(tmp,tmp,0x5);
+	  tmp.v    =_mm_addsub_pd(ret.v,in.v); // r,-i
+          ret.v    =_mm_shuffle_pd(tmp.v,tmp.v,0x1);
 #endif
 #ifdef AVX512
-          ret.v = _mm512_mask_sub_ps(in.v,0xaaaa,ret.v,in.v); // real -imag 
-	  ret.v = _mm512_swizzle_ps(ret.v, _MM_SWIZ_REG_CDAB);// OK
+          ret.v = _mm512_mask_sub_pd(in.v,0xaaaa,ret.v,in.v); // real -imag 
+	  ret.v = _mm512_swizzle_pd(ret.v, _MM_SWIZ_REG_CDAB);// OK
 #endif
 #ifdef QPX
             assert(0);
 #endif
 	  return ret;
 	}
-        friend inline vComplexD timesMinusI(const vComplexD &in){
+
+        friend inline vComplexD timesI(const vComplexD &in){
 	  vComplexD ret; vzero(ret);
+	  vComplexD tmp;
 #if defined (AVX1)|| defined (AVX2)
-	  cvec tmp =_mm256_shuffle_ps(in.v,in.v,0x5);
-          ret.v    =_mm256_addsub_ps(ret.v,tmp); // i,-r
+	  tmp.v    =_mm256_shuffle_pd(in.v,in.v,0x5);
+          ret.v    =_mm256_addsub_pd(ret.v,tmp.v); // i,-r
 #endif
 #ifdef SSE4
-	  cvec tmp =_mm_shuffle_ps(in.v,in.v,0x5);
-          ret.v    =_mm_addsub_ps(ret.v,tmp); // r,-i
+	  tmp.v    =_mm_shuffle_pd(in.v,in.v,0x1);
+          ret.v    =_mm_addsub_pd(ret.v,tmp.v); // r,-i
 #endif
 #ifdef AVX512
-          cvec tmp = _mm512_swizzle_ps(in.v, _MM_SWIZ_REG_CDAB);// OK
-	  ret.v    = _mm512_mask_sub_ps(tmp,0xaaaa,ret.v,tmp); // real -imag 
+          tmp.v    = _mm512_swizzle_pd(in.v, _MM_SWIZ_REG_CDAB);// OK
+	  ret.v    = _mm512_mask_sub_pd(tmp.v,0xaaaa,ret.v,tmp.v); // real -imag 
 #endif
 #ifdef QPX
             assert(0);
diff --git a/lib/simd/Grid_vComplexF.h b/lib/simd/Grid_vComplexF.h
index 4c31ad04..fb9c5a67 100644
--- a/lib/simd/Grid_vComplexF.h
+++ b/lib/simd/Grid_vComplexF.h
@@ -214,10 +214,10 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
        {
 #ifdef SSE4
 	   union {
-	     __m128 v1;    // SSE 4 x float vector
+	     cvec v1;    // SSE 4 x float vector
 	     float f[4];  // scalar array of 4 floats
 	   } u128;
-	   u128.v1= _mm_add_ps(v, _mm_shuffle_ps(v, v, 0b01001110)); // FIXME Prefer to use _MM_SHUFFLE macros
+	   u128.v1= _mm_add_ps(in.v, _mm_shuffle_ps(in.v,in.v, 0b01001110)); // FIXME Prefer to use _MM_SHUFFLE macros
 	   return ComplexF(u128.f[0], u128.f[1]);
 #endif
 #ifdef AVX1
@@ -329,13 +329,15 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
         friend inline vComplexF conj(const vComplexF &in){
             vComplexF ret ; vzero(ret);
 #if defined (AVX1)|| defined (AVX2)
-	    //             cvec tmp;
-	    //             tmp = _mm256_addsub_ps(ret.v,_mm256_shuffle_ps(in.v,in.v,_MM_SHUFFLE(2,3,0,1))); // ymm1 <- br,bi
-	    //             ret.v=_mm256_shuffle_ps(tmp,tmp,_MM_SHUFFLE(2,3,0,1));
-	    ret.v = _mm256_addsub_ps(ret.v,in.v);
+	    cvec tmp;
+	    tmp = _mm256_addsub_ps(ret.v,_mm256_shuffle_ps(in.v,in.v,_MM_SHUFFLE(2,3,0,1))); // ymm1 <- br,bi
+	    ret.v=_mm256_shuffle_ps(tmp,tmp,_MM_SHUFFLE(2,3,0,1));
+
 #endif
 #ifdef SSE4
-            ret.v = _mm_addsub_ps(ret.v,in.v);
+	    cvec tmp;
+	    tmp = _mm_addsub_ps(ret.v,_mm_shuffle_ps(in.v,in.v,_MM_SHUFFLE(2,3,0,1))); // ymm1 <- br,bi
+	    ret.v=_mm_shuffle_ps(tmp,tmp,_MM_SHUFFLE(2,3,0,1));
 #endif
 #ifdef AVX512
             ret.v = _mm512_mask_sub_ps(in.v,0xaaaa,ret.v,in.v); // Zero out 0+real 0-imag 
@@ -345,15 +347,16 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
 #endif
             return ret;
         }
-        friend inline vComplexF timesI(const vComplexF &in){
-	  vComplexF ret; vzero(ret);
+        friend inline vComplexF timesMinusI(const vComplexF &in){
+	  vComplexF ret; 
+	  vzero(ret);
 #if defined (AVX1)|| defined (AVX2)
 	  cvec tmp =_mm256_addsub_ps(ret.v,in.v); // r,-i
-          ret.v = _mm256_shuffle_ps(tmp,tmp,0x5);
+          ret.v = _mm256_shuffle_ps(tmp,tmp,_MM_SHUFFLE(2,3,0,1)); //-i,r
 #endif
 #ifdef SSE4
 	  cvec tmp =_mm_addsub_ps(ret.v,in.v); // r,-i
-          ret.v = _mm_shuffle_ps(tmp,tmp,0x5);
+          ret.v = _mm_shuffle_ps(tmp,tmp,_MM_SHUFFLE(2,3,0,1));
 #endif
 #ifdef AVX512
           ret.v = _mm512_mask_sub_ps(in.v,0xaaaa,ret.v,in.v); // real -imag 
@@ -364,14 +367,14 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
 #endif
 	  return ret;
 	}
-        friend inline vComplexF timesMinusI(const vComplexF &in){
+        friend inline vComplexF timesI(const vComplexF &in){
 	  vComplexF ret; vzero(ret);
 #if defined (AVX1)|| defined (AVX2)
-	  cvec tmp =_mm256_shuffle_ps(in.v,in.v,0x5);
-          ret.v = _mm256_addsub_ps(ret.v,tmp); // i,-r
+	  cvec tmp =_mm256_shuffle_ps(in.v,in.v,_MM_SHUFFLE(2,3,0,1));//i,r
+          ret.v    =_mm256_addsub_ps(ret.v,tmp);     //i,-r
 #endif
 #ifdef SSE4
-	  cvec tmp =_mm_shuffle_ps(in.v,in.v,0x5);
+	  cvec tmp =_mm_shuffle_ps(in.v,in.v,_MM_SHUFFLE(2,3,0,1));
           ret.v = _mm_addsub_ps(ret.v,tmp); // r,-i
 #endif
 #ifdef AVX512
@@ -443,5 +446,8 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
     inline vComplexF trace(const vComplexF &arg){
         return arg;
     }
+
+
 }
+
 #endif
diff --git a/lib/simd/Grid_vRealD.h b/lib/simd/Grid_vRealD.h
index ab848916..f6d68963 100644
--- a/lib/simd/Grid_vRealD.h
+++ b/lib/simd/Grid_vRealD.h
@@ -146,7 +146,7 @@ namespace Grid {
             ret.v = _mm256_set_pd(a[3],a[2],a[1],a[0]);
 #endif
 #ifdef SSE4
-            ret.v = _mm_set_pd(a[0],a[1]);
+            ret.v = _mm_set_pd(a[1],a[0]);
 #endif
 #ifdef AVX512
             ret.v = _mm512_set_pd(a[7],a[6],a[5],a[4],a[3],a[2],a[1],a[0]);
@@ -186,6 +186,15 @@ namespace Grid {
 
        friend inline RealD Reduce(const vRealD & in)
        {
+#if defined (SSE4)
+	 // FIXME Hack
+	 const RealD * ptr =(const RealD *)  &in;
+	 RealD ret = 0; 
+	 for(int i=0;i<vRealD::Nsimd();i++){
+	   ret = ret+ptr[i];
+	 }
+	 return ret;
+#endif
 #if defined (AVX1) || defined(AVX2)
 	 typedef union  {
 	   uint64_t l;
diff --git a/lib/simd/Grid_vRealF.h b/lib/simd/Grid_vRealF.h
index 7d71c528..86895492 100644
--- a/lib/simd/Grid_vRealF.h
+++ b/lib/simd/Grid_vRealF.h
@@ -175,7 +175,7 @@ namespace Grid {
             ret.v = _mm256_set_ps(a[7],a[6],a[5],a[4],a[3],a[2],a[1],a[0]);
 #endif
 #ifdef SSE4
-            ret.v = _mm_set_ps(a[0],a[1],a[2],a[3]);
+            ret.v = _mm_set_ps(a[3],a[2],a[1],a[0]);
 #endif
 #ifdef AVX512
             ret.v = _mm512_set_ps( a[15],a[14],a[13],a[12],a[11],a[10],a[9],a[8],
@@ -220,6 +220,15 @@ friend inline void vstore(const vRealF &ret, float *a){
         }
        friend inline RealF Reduce(const vRealF & in)
        {
+#if defined (SSE4)
+	 // FIXME Hack
+	 const RealF * ptr = (const RealF *) &in;
+	 RealF ret = 0; 
+	 for(int i=0;i<vRealF::Nsimd();i++){
+	   ret = ret+ptr[i];
+	 }
+	 return ret;
+#endif
 #if defined (AVX1) || defined(AVX2)
             __attribute__ ((aligned(32))) float c_[16];
             __m256 tmp = _mm256_permute2f128_ps(in.v,in.v,0x01);
diff --git a/lib/stencil/Grid_stencil_common.cc b/lib/stencil/Grid_stencil_common.cc
index 01bdce5d..c245aafb 100644
--- a/lib/stencil/Grid_stencil_common.cc
+++ b/lib/stencil/Grid_stencil_common.cc
@@ -121,8 +121,9 @@ namespace Grid {
       int simd_layout     = _grid->_simd_layout[dimension];
       int comm_dim        = _grid->_processors[dimension] >1 ;
       
-      assert(simd_layout==1);
+      //      assert(simd_layout==1); // Why?
       assert(comm_dim==1);
+      shift = (shift + fd) %fd;
       assert(shift>=0);
       assert(shift<fd);
       
diff --git a/tests/Grid_gamma.cc b/tests/Grid_gamma.cc
index 3b5a8430..ab6307f3 100644
--- a/tests/Grid_gamma.cc
+++ b/tests/Grid_gamma.cc
@@ -5,6 +5,11 @@ using namespace std;
 using namespace Grid;
 using namespace Grid::QCD;
 
+template<class d>
+struct scal {
+  d internal;
+};
+
 
 int main (int argc, char ** argv)
 {
@@ -22,22 +27,33 @@ int main (int argc, char ** argv)
   GridSerialRNG            sRNG;
   sRNG.SeedRandomDevice();
 
-  SpinMatrix ident=zero;
+  SpinMatrix ident; ident=zero;
   SpinMatrix rnd  ; random(sRNG,rnd);
 
-  SpinMatrix ll=zero;
-  SpinMatrix rr=zero;
+  SpinMatrix ll; ll=zero;
+  SpinMatrix rr; rr=zero;
   SpinMatrix result;
 
   SpinVector lv; random(sRNG,lv);
   SpinVector rv; random(sRNG,rv);
 
+  std::cout << " Is pod " << std::is_pod<SpinVector>::value  << std::endl;
+  std::cout << " Is pod double   " << std::is_pod<double>::value  << std::endl;
+  std::cout << " Is pod ComplexF " << std::is_pod<ComplexF>::value  << std::endl;
+  std::cout << " Is pod scal<double> " << std::is_pod<scal<double> >::value  << std::endl;
+  std::cout << " Is pod Scalar<double> " << std::is_pod<iScalar<double> >::value  << std::endl;
+  std::cout << " Is pod Scalar<ComplexF> " << std::is_pod<iScalar<ComplexF> >::value  << std::endl;
+  std::cout << " Is pod Scalar<vComplexF> " << std::is_pod<iScalar<vComplexF> >::value  << std::endl;
+  std::cout << " Is pod Scalar<vComplexD> " << std::is_pod<iScalar<vComplexD> >::value  << std::endl;
+  std::cout << " Is pod Scalar<vRealF> " << std::is_pod<iScalar<vRealF> >::value  << std::endl;
+  std::cout << " Is pod Scalar<vRealD> " << std::is_pod<iScalar<vRealD> >::value  << std::endl;
+
   for(int a=0;a<Ns;a++){
     ident()(a,a) = 1.0;
   }
 
   const Gamma::GammaMatrix *g = Gamma::GammaMatrices;
-  const char **list          = Gamma::GammaMatrixNames;
+  const char **list           = Gamma::GammaMatrixNames;
 
   result =ll*Gamma(g[0])*rr;
   result =ll*Gamma(g[0]);
diff --git a/tests/Grid_nersc_io.cc b/tests/Grid_nersc_io.cc
index 69c7f84f..06e51d41 100644
--- a/tests/Grid_nersc_io.cc
+++ b/tests/Grid_nersc_io.cc
@@ -49,7 +49,6 @@ int main (int argc, char ** argv)
       Plaq = Plaq + trace(U[mu]*Cshift(U[nu],mu,1)*adj(Cshift(U[mu],nu,1))*adj(U[nu]));
     }
   }
-
   
   double vol = Fine.gSites();
   Complex PlaqScale(1.0/vol/6.0/3.0);
@@ -58,7 +57,8 @@ int main (int argc, char ** argv)
   sliceSum(Plaq,Plaq_T,Nd-1);
   int Nt = Plaq_T.size();
 
-  TComplex Plaq_T_sum=zero;
+  TComplex Plaq_T_sum; 
+  Plaq_T_sum=zero;
   for(int t=0;t<Nt;t++){
     Plaq_T_sum = Plaq_T_sum+Plaq_T[t];
     Complex Pt=TensorRemove(Plaq_T[t]);
diff --git a/tests/Grid_simd.cc b/tests/Grid_simd.cc
new file mode 100644
index 00000000..3badb3a5
--- /dev/null
+++ b/tests/Grid_simd.cc
@@ -0,0 +1,166 @@
+#include <Grid.h>
+#include <parallelIO/GridNerscIO.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+class funcPlus {
+public:
+  funcPlus() {};
+  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = i1+i2;}
+  std::string name(void) const { return std::string("Plus"); }
+};
+class funcMinus {
+public:
+  funcMinus() {};
+  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = i1-i2;}
+  std::string name(void) const { return std::string("Minus"); }
+};
+class funcTimes {
+public:
+  funcTimes() {};
+  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = i1*i2;}
+  std::string name(void) const { return std::string("Times"); }
+};
+class funcConj {
+public:
+  funcConj() {};
+  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = conj(i1);}
+  std::string name(void) const { return std::string("Conj"); }
+};
+class funcAdj {
+public:
+  funcAdj() {};
+  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = adj(i1);}
+  std::string name(void) const { return std::string("Adj"); }
+};
+
+class funcTimesI {
+public:
+  funcTimesI() {};
+  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = timesI(i1);}
+  std::string name(void) const { return std::string("timesI"); }
+};
+
+class funcTimesMinusI {
+public:
+  funcTimesMinusI() {};
+  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = timesMinusI(i1);}
+  std::string name(void) const { return std::string("timesMinusI"); }
+};
+
+template<class scal, class vec,class functor > 
+void Tester(const functor &func)
+{
+  GridSerialRNG          sRNG;
+  sRNG.SeedRandomDevice();
+  
+  int Nsimd = vec::Nsimd();
+
+  std::vector<scal> input1(Nsimd);
+  std::vector<scal> input2(Nsimd);
+  std::vector<scal> result(Nsimd);
+  std::vector<scal> reference(Nsimd);
+
+  std::vector<vec,alignedAllocator<vec> > buf(3);
+  vec & v_input1 = buf[0];
+  vec & v_input2 = buf[1];
+  vec & v_result = buf[2];
+
+
+  for(int i=0;i<Nsimd;i++){
+    random(sRNG,input1[i]);
+    random(sRNG,input2[i]);
+    random(sRNG,result[i]);
+  }
+
+  Gmerge(v_input1,input1);
+  Gmerge(v_input2,input2);
+  Gmerge(v_result,result);
+
+  func(v_result,v_input1,v_input2);
+
+  for(int i=0;i<Nsimd;i++) {
+    func(reference[i],input1[i],input2[i]);
+  }
+
+  Gextract(v_result,result);
+  std::cout << " " << func.name()<<std::endl;
+
+  int ok=0;
+  for(int i=0;i<Nsimd;i++){
+    if ( abs(reference[i]-result[i])>0){
+      std::cout<< "*****" << std::endl;
+      std::cout<< "["<<i<<"] "<< abs(reference[i]-result[i]) << " " <<reference[i]<< " " << result[i]<<std::endl;
+      ok++;
+    }
+  }
+  if ( ok==0 ) std::cout << " OK!" <<std::endl;
+
+}
+
+
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> simd_layout({1,1,2,2});
+  std::vector<int> mpi_layout ({1,1,1,1});
+  std::vector<int> latt_size  ({8,8,8,8});
+    
+  GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+  std::vector<int> seeds({1,2,3,4});
+
+  // Insist that operations on random scalars gives
+  // identical results to on vectors.
+
+  Tester<RealD,vRealD>(funcPlus());
+
+  std::cout << "==================================="<<  std::endl;
+  std::cout << "Testing vComplexF "<<std::endl;
+  std::cout << "==================================="<<  std::endl;
+
+  Tester<ComplexF,vComplexF>(funcTimesI());
+  Tester<ComplexF,vComplexF>(funcTimesMinusI());
+  Tester<ComplexF,vComplexF>(funcPlus());
+  Tester<ComplexF,vComplexF>(funcMinus());
+  Tester<ComplexF,vComplexF>(funcTimes());
+  Tester<ComplexF,vComplexF>(funcConj());
+  Tester<ComplexF,vComplexF>(funcAdj());
+
+  std::cout << "==================================="<<  std::endl;
+  std::cout << "Testing vComplexD "<<std::endl;
+  std::cout << "==================================="<<  std::endl;
+
+
+  Tester<ComplexD,vComplexD>(funcTimesI());
+  Tester<ComplexD,vComplexD>(funcTimesMinusI());
+  Tester<ComplexD,vComplexD>(funcPlus());
+  Tester<ComplexD,vComplexD>(funcMinus());
+  Tester<ComplexD,vComplexD>(funcTimes());
+  Tester<ComplexD,vComplexD>(funcConj());
+  Tester<ComplexD,vComplexD>(funcAdj());
+
+  std::cout << "==================================="<<  std::endl;
+  std::cout << "Testing vRealF "<<std::endl;
+  std::cout << "==================================="<<  std::endl;
+
+
+  Tester<RealF,vRealF>(funcPlus());
+  Tester<RealF,vRealF>(funcMinus());
+  Tester<RealF,vRealF>(funcTimes());
+  Tester<RealF,vRealF>(funcAdj());
+
+  std::cout << "==================================="<<  std::endl;
+  std::cout << "Testing vRealD "<<std::endl;
+  std::cout << "==================================="<<  std::endl;
+
+  Tester<RealD,vRealD>(funcPlus());
+  Tester<RealD,vRealD>(funcMinus());
+  Tester<RealD,vRealD>(funcTimes());
+  Tester<RealD,vRealD>(funcAdj());
+
+  Grid_finalize();
+}
diff --git a/tests/Grid_stencil.cc b/tests/Grid_stencil.cc
index f35581f6..a7529045 100644
--- a/tests/Grid_stencil.cc
+++ b/tests/Grid_stencil.cc
@@ -4,60 +4,37 @@ using namespace std;
 using namespace Grid;
 using namespace Grid::QCD;
 
+template<class vobj>
+class SimpleCompressor {
+public:
+  void Point(int) {};
 
+  vobj operator() (vobj &arg) {
+    return arg;
+  }
+};
 int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> latt_size  (4);
-  std::vector<int> simd_layout(4);
-  std::vector<int> mpi_layout (4);
+  std::vector<int> simd_layout({1,1,2,2});
+  std::vector<int> mpi_layout ({2,2,2,2});
+  std::vector<int> latt_size  ({8,8,8,8});
 
-  int omp=1;
-  int lat=8;
-
-  mpi_layout[0]=1;
-  mpi_layout[1]=1;
-  mpi_layout[2]=1;
-  mpi_layout[3]=1;
-
-    latt_size[0] = lat;
-    latt_size[1] = lat;
-    latt_size[2] = lat;
-    latt_size[3] = lat;
-    double volume = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
+  double volume = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
     
-#ifdef AVX512
-    simd_layout[0] = 1;
-    simd_layout[1] = 2;
-    simd_layout[2] = 2;
-    simd_layout[3] = 2;
-#endif
-#if defined (AVX1)|| defined (AVX2)
-    simd_layout[0] = 1;
-    simd_layout[1] = 1;
-    simd_layout[2] = 2;
-    simd_layout[3] = 2;
-#endif
-#if defined (SSE2)
-    simd_layout[0] = 1;
-    simd_layout[1] = 1;
-    simd_layout[2] = 1;
-    simd_layout[3] = 2;
-#endif
-    
-    GridCartesian Fine(latt_size,simd_layout,mpi_layout);
-    GridRedBlackCartesian rbFine(latt_size,simd_layout,mpi_layout);
-    GridParallelRNG       fRNG(&Fine);
-    fRNG.SeedRandomDevice();
-
-    LatticeColourMatrix Foo(&Fine);
-    LatticeColourMatrix Bar(&Fine);
-    LatticeColourMatrix Check(&Fine);
-    LatticeColourMatrix Diff(&Fine);
-    
-    random(fRNG,Foo);
-    gaussian(fRNG,Bar);
+  GridCartesian Fine(latt_size,simd_layout,mpi_layout);
+  GridRedBlackCartesian rbFine(latt_size,simd_layout,mpi_layout);
+  GridParallelRNG       fRNG(&Fine);
+  fRNG.SeedRandomDevice();
+  
+  LatticeColourMatrix Foo(&Fine);
+  LatticeColourMatrix Bar(&Fine);
+  LatticeColourMatrix Check(&Fine);
+  LatticeColourMatrix Diff(&Fine);
+  
+  random(fRNG,Foo);
+  gaussian(fRNG,Bar);
 
 
     for(int dir=0;dir<4;dir++){
@@ -86,7 +63,8 @@ int main (int argc, char ** argv)
 	fflush(stdout);
 	std::vector<vColourMatrix,alignedAllocator<vColourMatrix> >  comm_buf(myStencil._unified_buffer_size);
 	printf("calling halo exchange\n");fflush(stdout);
-	myStencil.HaloExchange(Foo,comm_buf);
+	SimpleCompressor<vColourMatrix> compress;
+	myStencil.HaloExchange(Foo,comm_buf,compress);
 
 	Bar = Cshift(Foo,dir,disp);
 
diff --git a/tests/Grid_wilson.cc b/tests/Grid_wilson.cc
new file mode 100644
index 00000000..ade5263d
--- /dev/null
+++ b/tests/Grid_wilson.cc
@@ -0,0 +1,69 @@
+#include <Grid.h>
+#include <parallelIO/GridNerscIO.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> simd_layout({1,1,2,2});
+  std::vector<int> mpi_layout ({1,1,1,1});
+  std::vector<int> latt_size  ({8,8,8,8});
+    
+  GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+  std::vector<int> seeds({1,2,3,4});
+
+  GridParallelRNG          pRNG(&Grid);
+  //  pRNG.SeedFixedIntegers(seeds);
+  pRNG.SeedRandomDevice();
+
+  LatticeFermion src(&Grid); random(pRNG,src);
+  LatticeFermion result(&Grid); result=zero;
+  LatticeFermion    ref(&Grid);    ref=zero;
+  LatticeGaugeField Umu(&Grid); random(pRNG,Umu);
+  std::vector<LatticeColourMatrix> U(4,&Grid);
+
+  for(int mu=0;mu<Nd;mu++){
+    U[mu] = 1.0;
+    pokeIndex<3>(Umu,U[mu],mu);
+  }
+  
+  {
+    int mu=0;
+    //    ref =  src + Gamma(Gamma::GammaX)* src ; // 1-gamma_x
+    ref =  src;
+    ref = U[0]*Cshift(ref,0,1);
+  }
+
+  RealD mass=0.1;
+  WilsonMatrix Dw(Umu,mass);
+  std::cout << "Calling Dw"<<std::endl;
+  Dw.multiply(src,result);
+  std::cout << "Called Dw"<<std::endl;
+  std::cout << "norm result "<< norm2(result)<<std::endl;
+  std::cout << "norm ref    "<< norm2(ref)<<std::endl;
+
+  for(int ss=0;ss<10;ss++ ){
+    for(int i=0;i<Ns;i++){
+      for(int j=0;j<Nc;j++){
+	ComplexF * ref_p = (ComplexF *)&ref._odata[ss]()(i)(j);
+	ComplexF * res_p = (ComplexF *)&result._odata[ss]()(i)(j);
+	std::cout << ss<< " "<<i<<" "<<j<<" "<< (*ref_p)<<" " <<(*res_p)<<std::endl;
+      }
+    }
+  }
+
+  ref = ref -result;
+  std::cout << "norm diff   "<< norm2(ref)<<std::endl;
+
+  Grid_finalize();
+}
diff --git a/tests/Makefile.am b/tests/Makefile.am
index 8afac541..aaaff613 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -5,7 +5,7 @@ AM_LDFLAGS = -L$(top_srcdir)/lib
 #
 # Test code
 #
-bin_PROGRAMS = Grid_main Grid_stencil Grid_nersc_io Grid_cshift Grid_gamma
+bin_PROGRAMS = Grid_main Grid_stencil Grid_nersc_io Grid_cshift Grid_gamma Grid_wilson Grid_simd
 
 Grid_main_SOURCES = Grid_main.cc
 Grid_main_LDADD = -lGrid
@@ -21,3 +21,9 @@ Grid_gamma_LDADD = -lGrid
 
 Grid_stencil_SOURCES = Grid_stencil.cc
 Grid_stencil_LDADD = -lGrid
+
+Grid_wilson_SOURCES = Grid_wilson.cc
+Grid_wilson_LDADD = -lGrid
+
+Grid_simd_SOURCES = Grid_simd.cc
+Grid_simd_LDADD = -lGrid

From 0b7d389258ef83847a1bad9f9a5d2fabab063336 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 27 Apr 2015 13:45:07 +0100
Subject: [PATCH 107/429] Reworking CSHIFT and Stencil. Implementing Wilson and
 discovered rework is required

---
 TODO                                |   4 +
 lib/Grid_config.h                   |   4 +-
 lib/Grid_stencil.h                  | 156 ++++++++++++----------------
 lib/cartesian/Grid_cartesian_base.h |   6 +-
 lib/cshift/Grid_cshift_common.h     |   4 +-
 lib/cshift/Grid_cshift_mpi.h        | 132 ++++++++---------------
 lib/qcd/Grid_qcd_wilson_dop.cc      |  20 +++-
 tests/Grid_cshift.cc                |   9 +-
 tests/Grid_simd.cc                  |   2 -
 tests/Grid_stencil.cc               |   2 +-
 tests/Grid_wilson.cc                |  41 ++++++--
 11 files changed, 176 insertions(+), 204 deletions(-)

diff --git a/TODO b/TODO
index 98dfc3b8..ef3940ea 100644
--- a/TODO
+++ b/TODO
@@ -2,6 +2,10 @@
   - use protocol buffers? replace xmlReader/Writer ec..
   - Binary use htonll, htonl
 
+* Stencil -- do the permute for locally permuted in halo exchange.
+  - BUG cshift mpi; the "s" indexing is weird in the Cshift_comms_simd
+    as simd_layout or not is confusing
+
 * Reduce implemention is poor
 * Bug in SeedFixedIntegers gives same output on each site.
 * Bug in RNG with complex numbers ; only filling real values; need helper function -- DONE
diff --git a/lib/Grid_config.h b/lib/Grid_config.h
index f3d7016d..4152540e 100644
--- a/lib/Grid_config.h
+++ b/lib/Grid_config.h
@@ -2,7 +2,7 @@
 /* lib/Grid_config.h.in.  Generated from configure.ac by autoheader.  */
 
 /* AVX */
-/* #undef AVX1 */
+#define AVX1 1
 
 /* AVX2 */
 /* #undef AVX2 */
@@ -77,7 +77,7 @@
 #define PACKAGE_VERSION "1.0"
 
 /* SSE4 */
-#define SSE4 1
+/* #undef SSE4 */
 
 /* Define to 1 if you have the ANSI C header files. */
 #define STDC_HEADERS 1
diff --git a/lib/Grid_stencil.h b/lib/Grid_stencil.h
index 83db0a80..afb11db6 100644
--- a/lib/Grid_stencil.h
+++ b/lib/Grid_stencil.h
@@ -52,7 +52,7 @@ namespace Grid {
 // Gather for when there is no need to SIMD split with compression
 ///////////////////////////////////////////////////////////////////
 template<class vobj,class cobj,class compressor> void 
-Gather_plane_simple (Lattice<vobj> &rhs,std::vector<cobj,alignedAllocator<cobj> > &buffer,int dimension,int plane,int cbmask,compressor &compress)
+Gather_plane_simple (const Lattice<vobj> &rhs,std::vector<cobj,alignedAllocator<cobj> > &buffer,int dimension,int plane,int cbmask,compressor &compress)
 {
   int rd = rhs._grid->_rdimensions[dimension];
 
@@ -97,7 +97,7 @@ Gather_plane_simple (Lattice<vobj> &rhs,std::vector<cobj,alignedAllocator<cobj>
 // Gather for when there *is* need to SIMD split with compression
 ///////////////////////////////////////////////////////////////////
 template<class cobj,class vobj,class compressor> void 
-  Gather_plane_extract(Lattice<vobj> &rhs,std::vector<typename cobj::scalar_type *> pointers,int dimension,int plane,int cbmask,compressor &compress)
+Gather_plane_extract(const Lattice<vobj> &rhs,std::vector<typename cobj::scalar_type *> pointers,int dimension,int plane,int cbmask,compressor &compress)
 {
   int rd = rhs._grid->_rdimensions[dimension];
 
@@ -182,7 +182,7 @@ template<class cobj,class vobj,class compressor> void
       // Could allow a functional munging of the halo to another type during the comms.
       // this could implement the 16bit/32bit/64bit compression.
       template<class vobj,class cobj, class compressor> void 
-	HaloExchange(Lattice<vobj> &source,std::vector<cobj,alignedAllocator<cobj> > &u_comm_buf,compressor &compress)
+	HaloExchange(const Lattice<vobj> &source,std::vector<cobj,alignedAllocator<cobj> > &u_comm_buf,compressor &compress)
       {
 	// conformable(source._grid,_grid);
 	assert(source._grid==_grid);
@@ -237,7 +237,7 @@ template<class cobj,class vobj,class compressor> void
       }
 
       template<class vobj,class cobj, class compressor> 
-        void GatherStartComms(Lattice<vobj> &rhs,int dimension,int shift,int cbmask,
+        void GatherStartComms(const Lattice<vobj> &rhs,int dimension,int shift,int cbmask,
 			      std::vector<cobj,alignedAllocator<cobj> > &u_comm_buf,
 			      int &u_comm_offset,compressor & compress)
 	{
@@ -250,6 +250,7 @@ template<class cobj,class vobj,class compressor> void
 
 	  int fd              = _grid->_fdimensions[dimension];
 	  int rd              = _grid->_rdimensions[dimension];
+	  int pd              = _grid->_processors[dimension];
 	  int simd_layout     = _grid->_simd_layout[dimension];
 	  int comm_dim        = _grid->_processors[dimension] >1 ;
 	  assert(simd_layout==1);
@@ -267,11 +268,10 @@ template<class cobj,class vobj,class compressor> void
 	  
 	  for(int x=0;x<rd;x++){       
 	    
-	    int offnode = ( x+sshift >= rd );
 	    int sx        = (x+sshift)%rd;
-	    int comm_proc = (x+sshift)/rd;
-	    
-	    if (offnode) {
+	    int comm_proc = ((x+sshift)/rd)%pd;
+
+	    if (comm_proc) {
 	      
 	      int words = send_buf.size();
 	      if (cbmask != 0x3) words=words>>1;
@@ -284,7 +284,8 @@ template<class cobj,class vobj,class compressor> void
 	      int recv_from_rank;
 	      int xmit_to_rank;
 	      _grid->ShiftedRanks(dimension,comm_proc,xmit_to_rank,recv_from_rank);
-	      
+	      assert (xmit_to_rank != _grid->ThisRank());
+	      assert (recv_from_rank != _grid->ThisRank());
 	      //      FIXME Implement asynchronous send & also avoid buffer copy
 	      _grid->SendToRecvFrom((void *)&send_buf[0],
 				   xmit_to_rank,
@@ -293,7 +294,11 @@ template<class cobj,class vobj,class compressor> void
 				   bytes);
 	      printf("GatherStartComms communicated offnode x %d\n",x);fflush(stdout);
 
-	      printf("GatherStartComms inserting %d buf size %d\n",u_comm_offset,buffer_size);fflush(stdout);
+	      printf("GatherStartComms inserting %le to u_comm_offset %d buf size %d for dim %d shift %d\n",
+		     *( (RealF *) &recv_buf[0]),
+		     u_comm_offset,buffer_size,
+		     dimension,shift
+		     ); fflush(stdout);
 	      for(int i=0;i<buffer_size;i++){
 		u_comm_buf[u_comm_offset+i]=recv_buf[i];
 	      }
@@ -305,17 +310,19 @@ template<class cobj,class vobj,class compressor> void
 
 
       template<class vobj,class cobj, class compressor> 
-	void  GatherStartCommsSimd(Lattice<vobj> &rhs,int dimension,int shift,int cbmask,
+	void  GatherStartCommsSimd(const Lattice<vobj> &rhs,int dimension,int shift,int cbmask,
 				   std::vector<cobj,alignedAllocator<cobj> > &u_comm_buf,
 				   int &u_comm_offset,compressor &compress)
 	{
 	  const int Nsimd = _grid->Nsimd();
+
 	  typedef typename cobj::vector_type vector_type;
 	  typedef typename cobj::scalar_type scalar_type;
 	  
 	  int fd = _grid->_fdimensions[dimension];
 	  int rd = _grid->_rdimensions[dimension];
 	  int ld = _grid->_ldimensions[dimension];
+	  int pd              = _grid->_processors[dimension];
 	  int simd_layout     = _grid->_simd_layout[dimension];
 	  int comm_dim        = _grid->_processors[dimension] >1 ;
 
@@ -346,89 +353,62 @@ template<class cobj,class vobj,class compressor> void
 
 	  int cb    = (cbmask==0x2)? 1 : 0;
 	  int sshift= _grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,cb);
-	  
-	  std::vector<int> comm_offnode(simd_layout);
-	  std::vector<int> comm_proc   (simd_layout);  //relative processor coord in dim=dimension
-	  std::vector<int> icoor(_grid->Nd());
-	  
-	  for(int x=0;x<rd;x++){
+
+	  // loop over outer coord planes orthog to dim
+	  for(int x=0;x<rd;x++){       
 	    
-	    int comm_any = 0;
-	    for(int s=0;s<simd_layout;s++) {
-	      int shifted_x   = x+s*rd+sshift;
-	      comm_offnode[s] = shifted_x >= ld;
-	      comm_any        = comm_any | comm_offnode[s];
-	      comm_proc[s]    = shifted_x/ld;
+	    // FIXME call local permute copy if none are offnode.
+	    for(int i=0;i<Nsimd;i++){       
+	      pointers[i] = (scalar_type *)&send_buf_extract[i][0];
 	    }
-    
-	    int o    = 0;
-	    int bo   = x*_grid->_ostride[dimension];
 	    int sx   = (x+sshift)%rd;
-	    
-	    if ( comm_any ) {
 
-	      for(int i=0;i<Nsimd;i++){
-		pointers[Nsimd-1-i] = (scalar_type *)&send_buf_extract[i][0];
+	    std::cout<< "Gathering "<< x <<std::endl;
+	    Gather_plane_extract<cobj>(rhs,pointers,dimension,sx,cbmask,compress);
+	    std::cout<< "Gathered "<<std::endl;
+	    for(int i=0;i<Nsimd;i++){
+      
+	      int inner_bit = (Nsimd>>(permute_type+1));
+	      int ic= (i&inner_bit)? 1:0;
+
+	      int my_coor          = rd*ic + x;
+	      int nbr_coor         = my_coor+sshift;
+	      int nbr_proc = ((nbr_coor)/ld) % pd;// relative shift in processors
+
+	      int nbr_ic   = (nbr_coor%ld)/rd;    // inner coord of peer
+	      int nbr_ox   = (nbr_coor%rd);       // outer coord of peer
+	      int nbr_lane = (i&(~inner_bit));
+
+	      int recv_from_rank;
+	      int xmit_to_rank;
+
+	      if (nbr_ic) nbr_lane|=inner_bit;
+	      assert (sx == nbr_ox);
+
+	      if(nbr_proc){
+
+		std::cout<< "MPI sending "<<std::endl;
+		_grid->ShiftedRanks(dimension,nbr_proc,xmit_to_rank,recv_from_rank); 
+
+		_grid->SendToRecvFrom((void *)&send_buf_extract[nbr_lane][0],
+				     xmit_to_rank,
+				     (void *)&recv_buf_extract[i][0],
+				     recv_from_rank,
+				     bytes);
+		std::cout<< "MPI complete "<<std::endl;
+
+		rpointers[i] = (scalar_type *)&recv_buf_extract[i][0];
+	      } else { 
+		rpointers[i] = (scalar_type *)&send_buf_extract[nbr_lane][0];
 	      }
-	      Gather_plane_extract<cobj>(rhs,pointers,dimension,sx,cbmask,compress);
-	      
-	      for(int i=0;i<Nsimd;i++){
-		
-		int s;
-		_grid->iCoorFromIindex(icoor,i);
-		s = icoor[dimension];
-		
-		if(comm_offnode[s]){
-		  
-		  int rank           = _grid->_processor;
-		  int recv_from_rank;
-		  int xmit_to_rank;
-
-		  _grid->ShiftedRanks(dimension,comm_proc[s],xmit_to_rank,recv_from_rank);
-	  
-
-		  _grid->SendToRecvFrom((void *)&send_buf_extract[i][0],
-					xmit_to_rank,
-					(void *)&recv_buf_extract[i][0],
-					recv_from_rank,
-					bytes);
-
-		  rpointers[i] = (scalar_type *)&recv_buf_extract[i][0];
-		  
-		} else { 
-		  
-		  rpointers[i] = (scalar_type *)&send_buf_extract[i][0];
-
-		}
-		
-	      }
-
-	      // Permute by swizzling pointers in merge
-	      int permute_slice=0;
-	      int lshift=sshift%ld;
-	      int wrap  =lshift/rd;
-	      int  num  =lshift%rd;
-
-	      if ( x< rd-num ) permute_slice=wrap;
-	      else permute_slice = 1-wrap;
-
-	      int toggle_bit = (Nsimd>>(permute_type+1));
-	      int PermuteMap;
-	      for(int i=0;i<Nsimd;i++){
-		if ( permute_slice ) {
-		  PermuteMap=i^toggle_bit;
-		  pointers[Nsimd-1-i] = rpointers[PermuteMap];
-		} else {
-		  pointers[Nsimd-1-i] = rpointers[i];
-		}
-	      }
-
-	      // Here we don't want to scatter, just place into a buffer.
-	      for(int i=0;i<buffer_size;i++){
-		merge(u_comm_buf[u_comm_offset+i],pointers);
-	      }
-
 	    }
+
+	    // Here we don't want to scatter, just place into a buffer.
+	    std::cout<< "merging "<<std::endl;
+	    for(int i=0;i<buffer_size;i++){
+	      merge(u_comm_buf[u_comm_offset+i],rpointers);
+	    }
+
 	  }
 	}
   };
diff --git a/lib/cartesian/Grid_cartesian_base.h b/lib/cartesian/Grid_cartesian_base.h
index 6fd55399..c297cbb9 100644
--- a/lib/cartesian/Grid_cartesian_base.h
+++ b/lib/cartesian/Grid_cartesian_base.h
@@ -98,6 +98,9 @@ public:
 	index   = index / dims[d];
       }
     }
+    inline void oCoorFromOindex (std::vector<int>& coor,int Oindex){
+      CoorFromIndex(coor,Oindex,_rdimensions);
+    }
     static inline void IndexFromCoor (std::vector<int>& coor,int &index,std::vector<int> &dims){
       int nd=dims.size();
       int stride=1;
@@ -107,9 +110,6 @@ public:
 	stride=stride*dims[d];
       }
     }
-    inline void oCoorFromOindex (std::vector<int>& coor,int Oindex){
-      CoorFromIndex(coor,Oindex,_rdimensions);
-    }
 
     //////////////////////////////////////////////////////////
     // SIMD lane addressing
diff --git a/lib/cshift/Grid_cshift_common.h b/lib/cshift/Grid_cshift_common.h
index 8c4f1e1d..eebc895e 100644
--- a/lib/cshift/Grid_cshift_common.h
+++ b/lib/cshift/Grid_cshift_common.h
@@ -5,7 +5,7 @@ namespace Grid {
 //////////////////////////////////////////////////////
 // Gather for when there is no need to SIMD split
 //////////////////////////////////////////////////////
-template<class vobj> void Gather_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllocator<vobj> > &buffer,             int dimension,int plane,int cbmask)
+template<class vobj> void Gather_plane_simple (const Lattice<vobj> &rhs,std::vector<vobj,alignedAllocator<vobj> > &buffer,             int dimension,int plane,int cbmask)
 {
   int rd = rhs._grid->_rdimensions[dimension];
 
@@ -49,7 +49,7 @@ template<class vobj> void Gather_plane_simple (Lattice<vobj> &rhs,std::vector<vo
 //////////////////////////////////////////////////////
 // Gather for when there *is* need to SIMD split
 //////////////////////////////////////////////////////
- template<class vobj,class scalar_type> void Gather_plane_extract(Lattice<vobj> &rhs,std::vector<scalar_type *> pointers,int dimension,int plane,int cbmask)
+ template<class vobj,class scalar_type> void Gather_plane_extract(const Lattice<vobj> &rhs,std::vector<scalar_type *> pointers,int dimension,int plane,int cbmask)
 {
   int rd = rhs._grid->_rdimensions[dimension];
 
diff --git a/lib/cshift/Grid_cshift_mpi.h b/lib/cshift/Grid_cshift_mpi.h
index 60894ef3..ab9237ad 100644
--- a/lib/cshift/Grid_cshift_mpi.h
+++ b/lib/cshift/Grid_cshift_mpi.h
@@ -79,6 +79,7 @@ template<class vobj> void Cshift_comms(Lattice<vobj> &ret,Lattice<vobj> &rhs,int
 
   int fd              = rhs._grid->_fdimensions[dimension];
   int rd              = rhs._grid->_rdimensions[dimension];
+  int pd              = rhs._grid->_processors[dimension];
   int simd_layout     = rhs._grid->_simd_layout[dimension];
   int comm_dim        = rhs._grid->_processors[dimension] >1 ;
   assert(simd_layout==1);
@@ -95,11 +96,10 @@ template<class vobj> void Cshift_comms(Lattice<vobj> &ret,Lattice<vobj> &rhs,int
 
   for(int x=0;x<rd;x++){       
 
-    int offnode = ( x+sshift >= rd );
-    int sx        = (x+sshift)%rd;
-    int comm_proc = (x+sshift)/rd;
+    int sx        =  (x+sshift)%rd;
+    int comm_proc = ((x+sshift)/rd)%pd;
     
-    if (!offnode) {
+    if (comm_proc==0) {
 
       Copy_plane(ret,rhs,dimension,x,sx,cbmask); 
 
@@ -138,6 +138,7 @@ template<class vobj> void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &r
   int fd = grid->_fdimensions[dimension];
   int rd = grid->_rdimensions[dimension];
   int ld = grid->_ldimensions[dimension];
+  int pd = grid->_processors[dimension];
   int simd_layout     = grid->_simd_layout[dimension];
   int comm_dim        = grid->_processors[dimension] >1 ;
 
@@ -164,101 +165,58 @@ template<class vobj> void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &r
   ///////////////////////////////////////////
   // Work out what to send where
   ///////////////////////////////////////////
-
   int cb    = (cbmask==0x2)? 1 : 0;
   int sshift= grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,cb);
-  
-  std::vector<int> comm_offnode(simd_layout);
-  std::vector<int> comm_proc   (simd_layout);  //relative processor coord in dim=dimension
-  std::vector<int> icoor(grid->Nd());
 
+  // loop over outer coord planes orthog to dim
   for(int x=0;x<rd;x++){       
 
-    int comm_any = 0;
-    for(int s=0;s<simd_layout;s++) {
-      int shifted_x   = x+s*rd+sshift;
-      comm_offnode[s] = shifted_x >= ld; 
-      comm_any        = comm_any | comm_offnode[s];
-      comm_proc[s]    = shifted_x/ld;     
+  // FIXME call local permute copy if none are offnode.
+
+    for(int i=0;i<Nsimd;i++){       
+      pointers[i] = (scalar_type *)&send_buf_extract[i][0];
     }
-    
-    int o    = 0;
-    int bo   = x*grid->_ostride[dimension];
     int sx   = (x+sshift)%rd;
+    Gather_plane_extract(rhs,pointers,dimension,sx,cbmask);
 
-    if ( comm_any ) {
+    for(int i=0;i<Nsimd;i++){
+      
+      int inner_bit = (Nsimd>>(permute_type+1));
+      int ic= (i&inner_bit)? 1:0;
 
-      for(int i=0;i<Nsimd;i++){         // there is a reversal in extract merge
-	pointers[i] = (scalar_type *)&send_buf_extract[Nsimd-1-i][0];
+      int my_coor          = rd*ic + x;
+      int nbr_coor         = my_coor+sshift;
+      int nbr_proc = ((nbr_coor)/ld) % pd;// relative shift in processors
+
+      int nbr_ic   = (nbr_coor%ld)/rd;    // inner coord of peer
+      int nbr_ox   = (nbr_coor%rd);       // outer coord of peer
+      int nbr_lane = (i&(~inner_bit));
+
+      int recv_from_rank;
+      int xmit_to_rank;
+
+      if (nbr_ic) nbr_lane|=inner_bit;
+
+      assert (sx == nbr_ox);
+
+      if(nbr_proc){
+	grid->ShiftedRanks(dimension,nbr_proc,xmit_to_rank,recv_from_rank); 
+
+	grid->SendToRecvFrom((void *)&send_buf_extract[nbr_lane][0],
+			     xmit_to_rank,
+			     (void *)&recv_buf_extract[i][0],
+			     recv_from_rank,
+			     bytes);
+
+	rpointers[i] = (scalar_type *)&recv_buf_extract[i][0];
+      } else { 
+	rpointers[i] = (scalar_type *)&send_buf_extract[nbr_lane][0];
       }
-      Gather_plane_extract(rhs,pointers,dimension,sx,cbmask);
-
-      for(int i=0;i<Nsimd;i++){
-
-
-	int s;
-	grid->iCoorFromIindex(icoor,i);
-	s = icoor[dimension];
-
-	if(comm_offnode[s]){
-
-	  int rank           = grid->_processor;
-	  int recv_from_rank;
-	  int xmit_to_rank;
-	  grid->ShiftedRanks(dimension,comm_proc[s],xmit_to_rank,recv_from_rank);
-	  
-
-	  grid->SendToRecvFrom((void *)&send_buf_extract[i][0],
-			    xmit_to_rank,
-			    (void *)&recv_buf_extract[i][0],
-			    recv_from_rank,
-			    bytes);
-
-	  rpointers[i] = (scalar_type *)&recv_buf_extract[i][0];
-
-	} else { 
-
-	  rpointers[i] = (scalar_type *)&send_buf_extract[i][0];
-
-	}
-
-      }
-
-      // Permute by swizzling pointers in merge
-      int permute_slice=0;
-      int lshift=sshift%ld;
-      int wrap  =lshift/rd;
-      int  num  =lshift%rd;
-
-      if ( x< rd-num ) permute_slice=wrap;
-      else permute_slice = 1-wrap;
-
-      int toggle_bit = (Nsimd>>(permute_type+1));
-      int PermuteMap;
-      for(int i=0;i<Nsimd;i++){
-	if ( permute_slice ) {
-	  PermuteMap=i^toggle_bit;
-	  pointers[Nsimd-1-i] = rpointers[PermuteMap];
-	} else {
-	  pointers[Nsimd-1-i] = rpointers[i];
-	}
-      }
-
-      Scatter_plane_merge(ret,pointers,dimension,x,cbmask);
-
-    } else { 
-
-      int permute_slice=0;
-      int wrap = sshift/rd;
-      int  num = sshift%rd;
-      if ( x< rd-num ) permute_slice=wrap;
-      else permute_slice = 1-wrap;
-
-      if ( permute_slice ) Copy_plane_permute(ret,rhs,dimension,x,sx,cbmask,permute_type);
-      else                 Copy_plane(ret,rhs,dimension,x,sx,cbmask); 
 
     }
+    Scatter_plane_merge(ret,rpointers,dimension,x,cbmask);
   }
-}
+
+ }
 }
 #endif
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index aba21417..b1d6d525 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -22,8 +22,12 @@ const int WilsonMatrix::Tm = 7;
   public:
     int mu;
     
-    void Point(int p) { mu=p;};
-    vHalfSpinColourVector operator () (vSpinColourVector &in)
+    void Point(int p) { 
+      mu=p;
+      std::cout << "WilsonCompressor.Point " << mu<<std::endl;
+    };
+
+    vHalfSpinColourVector operator () (const vSpinColourVector &in)
     {
       vHalfSpinColourVector ret;
       switch(mu) {
@@ -90,7 +94,8 @@ void WilsonMatrix::multiply(const LatticeFermion &in, LatticeFermion &out)
 
 void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
 {
-  //  Stencil.HaloExchange(in,comm_buf);
+  WilsonCompressor compressor;
+  Stencil.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
 
   for(int ss=0;ss<grid->oSites();ss++){
 
@@ -100,10 +105,13 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     vHalfSpinColourVector  chi;    
     vHalfSpinColourVector Uchi;
     vHalfSpinColourVector *chi_p;
+
+    result=zero;
+
+#if 0
     // Xp
     offset = Stencil._offsets [Xp][ss];
     local  = Stencil._is_local[Xp][ss];
-
     chi_p  = &comm_buf[offset];
     if ( local ) {
       spProjXp(chi,in._odata[offset]);
@@ -112,7 +120,6 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     mult(&(Uchi()),&(Umu._odata[ss](Xp)),&(*chi_p)());
     spReconXp(result,Uchi);
 
-#if 0
     // Yp
     offset = Stencil._offsets [Yp][ss];
     local  = Stencil._is_local[Yp][ss];
@@ -145,6 +152,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     } 
     mult(&(Uchi()),&(Umu._odata[ss](Tp)),&(*chi_p)());
     accumReconTp(result,Uchi);
+#endif
 
     // Xm
     offset = Stencil._offsets [Xm][ss];
@@ -156,6 +164,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     } 
     mult(&(Uchi()),&(Umu._odata[ss](Xm)),&(*chi_p)());
     accumReconXm(result,Uchi);
+#if 0
 
     // Ym
     offset = Stencil._offsets [Ym][ss];
@@ -190,6 +199,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     mult(&(Uchi()),&(Umu._odata[ss](Tm)),&(*chi_p)());
     accumReconTm(result,Uchi);
 #endif
+
     out._odata[ss] = result;
   }
 
diff --git a/tests/Grid_cshift.cc b/tests/Grid_cshift.cc
index 995756d6..c0863965 100644
--- a/tests/Grid_cshift.cc
+++ b/tests/Grid_cshift.cc
@@ -1,7 +1,6 @@
 #include <Grid.h>
 #include <parallelIO/GridNerscIO.h>
 
-using namespace std;
 using namespace Grid;
 using namespace Grid::QCD;
 
@@ -10,7 +9,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> simd_layout({1,1,2,2});
-  std::vector<int> mpi_layout ({2,2,2,2});
+  std::vector<int> mpi_layout ({2,2,1,4});
   std::vector<int> latt_size  ({8,8,8,16});
     
   GridCartesian     Fine(latt_size,simd_layout,mpi_layout);
@@ -71,11 +70,11 @@ int main (int argc, char ** argv)
 	  Fine.CoorFromIndex(peer,index,latt_size);
 
 	  if (nrm > 0){
-	    cout<<"FAIL shift "<< shift<<" in dir "<< dir<<" ["<<coor[0]<<","<<coor[1]<<","<<coor[2]<<","<<coor[3]<<"] = "<< cm()()()<<" expect "<<scm<<"  "<<nrm<<endl;
-	    cout<<"Got    "<<index<<" " << peer[0]<<","<<peer[1]<<","<<peer[2]<<","<<peer[3]<<std::endl;
+	    std::cerr<<"FAIL shift "<< shift<<" in dir "<< dir<<" ["<<coor[0]<<","<<coor[1]<<","<<coor[2]<<","<<coor[3]<<"] = "<< cm()()()<<" expect "<<scm<<"  "<<nrm<<std::endl;
+	    std::cerr<<"Got    "<<index<<" " << peer[0]<<","<<peer[1]<<","<<peer[2]<<","<<peer[3]<<std::endl;
 	    index=real(scm);
 	    Fine.CoorFromIndex(peer,index,latt_size);
-	    cout<<"Expect "<<index<<" " << peer[0]<<","<<peer[1]<<","<<peer[2]<<","<<peer[3]<<std::endl;
+	    std::cerr<<"Expect "<<index<<" " << peer[0]<<","<<peer[1]<<","<<peer[2]<<","<<peer[3]<<std::endl;
 	  }
 	}}}}
     }
diff --git a/tests/Grid_simd.cc b/tests/Grid_simd.cc
index 3badb3a5..364b13e8 100644
--- a/tests/Grid_simd.cc
+++ b/tests/Grid_simd.cc
@@ -116,8 +116,6 @@ int main (int argc, char ** argv)
   // Insist that operations on random scalars gives
   // identical results to on vectors.
 
-  Tester<RealD,vRealD>(funcPlus());
-
   std::cout << "==================================="<<  std::endl;
   std::cout << "Testing vComplexF "<<std::endl;
   std::cout << "==================================="<<  std::endl;
diff --git a/tests/Grid_stencil.cc b/tests/Grid_stencil.cc
index a7529045..c06c0d16 100644
--- a/tests/Grid_stencil.cc
+++ b/tests/Grid_stencil.cc
@@ -9,7 +9,7 @@ class SimpleCompressor {
 public:
   void Point(int) {};
 
-  vobj operator() (vobj &arg) {
+  vobj operator() (const vobj &arg) {
     return arg;
   }
 };
diff --git a/tests/Grid_wilson.cc b/tests/Grid_wilson.cc
index ade5263d..f6e86f5d 100644
--- a/tests/Grid_wilson.cc
+++ b/tests/Grid_wilson.cc
@@ -1,5 +1,4 @@
 #include <Grid.h>
-#include <parallelIO/GridNerscIO.h>
 
 using namespace std;
 using namespace Grid;
@@ -10,13 +9,19 @@ struct scal {
   d internal;
 };
 
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
 
 int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
   std::vector<int> simd_layout({1,1,2,2});
-  std::vector<int> mpi_layout ({1,1,1,1});
+  std::vector<int> mpi_layout ({2,1,1,2});
   std::vector<int> latt_size  ({8,8,8,8});
     
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
@@ -29,19 +34,37 @@ int main (int argc, char ** argv)
   LatticeFermion src(&Grid); random(pRNG,src);
   LatticeFermion result(&Grid); result=zero;
   LatticeFermion    ref(&Grid);    ref=zero;
+  LatticeFermion    tmp(&Grid);    tmp=zero;
   LatticeGaugeField Umu(&Grid); random(pRNG,Umu);
   std::vector<LatticeColourMatrix> U(4,&Grid);
 
   for(int mu=0;mu<Nd;mu++){
-    U[mu] = 1.0;
-    pokeIndex<3>(Umu,U[mu],mu);
+    //    U[mu] = 1.0;
+    //    pokeIndex<3>(Umu,U[mu],mu);
+    U[mu] = peekIndex<3>(Umu,mu);
   }
   
-  {
-    int mu=0;
-    //    ref =  src + Gamma(Gamma::GammaX)* src ; // 1-gamma_x
-    ref =  src;
-    ref = U[0]*Cshift(ref,0,1);
+  std::vector<int> mask({0,0,0,0,1,0,0,0});
+  { // Naive wilson implementation
+    ref = zero;
+    for(int mu=0;mu<Nd;mu++){
+
+      //    ref =  src + Gamma(Gamma::GammaX)* src ; // 1-gamma_x
+      if( mask[mu] ) {
+	tmp = U[mu]*Cshift(src,mu,1);
+	for(int i=0;i<ref._odata.size();i++){
+	  ref._odata[i]+= tmp._odata[i] + Gamma(Gmu[mu])*tmp._odata[i]; ;
+	}
+      }
+
+      if( mask[mu+4] ){
+	tmp =adj(U[mu])*src;
+	tmp =Cshift(tmp,mu,-1);
+	for(int i=0;i<ref._odata.size();i++){
+	  ref._odata[i]+= tmp._odata[i] - Gamma(Gmu[mu])*tmp._odata[i]; ;
+	}
+      }
+    }
   }
 
   RealD mass=0.1;

From b0485894b34a4cb970e1132fbe706615852e3639 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 28 Apr 2015 08:11:59 +0100
Subject: [PATCH 108/429] Shaken out stencil to the point where I think wilson
 dslash is correct. Need to audit code carefully, consolidate between stencil
 and cshift, and then benchmark and optimise.

---
 lib/Grid_comparison.h                 |   6 +-
 lib/Grid_extract.h                    | 193 ++++++++++++++++
 lib/Grid_lattice.h                    |   1 +
 lib/Grid_simd.h                       |  62 -----
 lib/Grid_stencil.h                    | 222 ++++++------------
 lib/Grid_where.h                      |  21 +-
 lib/cshift/Grid_cshift_common.h       | 320 +++++++++++---------------
 lib/cshift/Grid_cshift_mpi.h          |  20 +-
 lib/lattice/Grid_lattice_coordinate.h |  10 +-
 lib/lattice/Grid_lattice_peekpoke.h   |  27 +--
 lib/lattice/Grid_lattice_reduction.h  |   4 +-
 lib/lattice/Grid_lattice_rng.h        |  82 ++++---
 lib/math/Grid_math_tensors.h          |  59 -----
 lib/math/Grid_math_traits.h           |  12 +-
 lib/qcd/Grid_qcd_wilson_dop.cc        |  97 ++++++--
 lib/simd/Grid_vComplexD.h             |   2 +
 lib/simd/Grid_vComplexF.h             |   3 +-
 lib/simd/Grid_vInteger.h              |   3 +-
 lib/simd/Grid_vRealD.h                |   3 +-
 lib/simd/Grid_vRealF.h                |   3 +-
 lib/stencil/Grid_stencil_common.cc    |   7 +-
 tests/Grid_simd.cc                    |   8 +-
 tests/Grid_stencil.cc                 |  36 +--
 tests/Grid_wilson.cc                  |   3 +-
 24 files changed, 599 insertions(+), 605 deletions(-)
 create mode 100644 lib/Grid_extract.h

diff --git a/lib/Grid_comparison.h b/lib/Grid_comparison.h
index 530817c6..53dec1a0 100644
--- a/lib/Grid_comparison.h
+++ b/lib/Grid_comparison.h
@@ -101,12 +101,12 @@ namespace Grid {
       std::vector<Integer> vlhs(vInteger::Nsimd());   // Use functors to reduce this to single implementation
       std::vector<Integer> vrhs(vInteger::Nsimd());
       vInteger ret;
-      extract(lhs,vlhs);
-      extract(rhs,vrhs);
+      extract<vInteger,Integer>(lhs,vlhs);
+      extract<vInteger,Integer>(rhs,vrhs);
       for(int s=0;s<vInteger::Nsimd();s++){
 	vlhs[s] = sop(vlhs[s],vrhs[s]);
       }
-      merge(ret,vlhs);
+      merge<vInteger,Integer>(ret,vlhs);
       return ret;
     }
     inline vInteger operator < (const vInteger & lhs, const vInteger & rhs)
diff --git a/lib/Grid_extract.h b/lib/Grid_extract.h
new file mode 100644
index 00000000..e29fbee7
--- /dev/null
+++ b/lib/Grid_extract.h
@@ -0,0 +1,193 @@
+#ifndef GRID_EXTRACT_H
+#define GRID_EXTRACT_H
+/////////////////////////////////////////////////////////////////
+// Generic extract/merge/permute
+/////////////////////////////////////////////////////////////////
+
+namespace Grid{
+
+////////////////////////////////////////////////////////////////////////////////////////////////
+// Extract/merge a fundamental vector type, to pointer array with offset
+////////////////////////////////////////////////////////////////////////////////////////////////
+
+template<class vsimd,class scalar>
+inline void extract(typename std::enable_if<isGridTensor<vsimd>::notvalue, const vsimd >::type * y, 
+		    std::vector<scalar *> &extracted,int offset){
+  // FIXME: bounce off memory is painful
+  int Nextr=extracted.size();
+  int Nsimd=vsimd::Nsimd();
+  int s=Nsimd/Nextr;
+
+  scalar*buf = (scalar *)y;
+  for(int i=0;i<Nextr;i++){
+    extracted[i][offset] = buf[i*s];
+  }
+};
+////////////////////////////////////////////////////////////////////////
+// Merge simd vector from array of scalars to pointer array with offset
+////////////////////////////////////////////////////////////////////////
+template<class vsimd,class scalar>
+inline void merge(typename std::enable_if<isGridTensor<vsimd>::notvalue, vsimd >::type * y, 
+		  std::vector<scalar *> &extracted,int offset){
+  int Nextr=extracted.size();
+  int Nsimd=vsimd::Nsimd();
+  int s=Nsimd/Nextr;
+
+  scalar *buf =(scalar *) y;
+  for(int i=0;i<Nextr;i++){
+    for(int ii=0;ii<s;ii++){
+      buf[i*s+ii]=extracted[i][offset];
+    }
+  }
+
+};
+
+
+////////////////////////////////////////////////////////////////////////////////////////////////
+// Extract a fundamental vector type to scalar array 
+////////////////////////////////////////////////////////////////////////////////////////////////
+template<class vsimd,class scalar>
+inline void extract(typename std::enable_if<isGridTensor<vsimd>::notvalue, const vsimd >::type  &y,std::vector<scalar> &extracted){
+
+  int Nextr=extracted.size();
+  int Nsimd=vsimd::Nsimd();
+  int s=Nsimd/Nextr;
+
+  scalar *buf = (scalar *)&y;
+  for(int i=0;i<Nextr;i++){
+    for(int ii=0;ii<s;ii++){
+      extracted[i]=buf[i*s+ii];
+    }
+  }
+
+};
+
+////////////////////////////////////////////////////////////////////////
+// Merge simd vector from array of scalars
+////////////////////////////////////////////////////////////////////////
+template<class vsimd,class scalar>
+inline void merge(typename std::enable_if<isGridTensor<vsimd>::notvalue, vsimd >::type  &y,std::vector<scalar> &extracted){
+  int Nextr=extracted.size();
+  int Nsimd=vsimd::Nsimd();
+  int s=Nsimd/Nextr;
+  scalar *buf = (scalar *)&y;
+
+  for(int i=0;i<Nextr;i++){
+    for(int ii=0;ii<s;ii++){
+      buf[i*s+ii]=extracted[i];
+    }
+  }
+
+};
+template<class vsimd,class scalar>
+inline void AmergeA(typename std::enable_if<isGridTensor<vsimd>::notvalue, vsimd >::type  &y,std::vector<scalar> &extracted){
+  int Nextr=extracted.size();
+  int Nsimd=vsimd::Nsimd();
+  int s=Nsimd/Nextr;
+
+  scalar *buf = (scalar *)&y;
+  for(int i=0;i<Nextr;i++){
+    for(int ii=0;ii<s;ii++){
+      buf[i*s+ii]=extracted[i];
+    }
+  }
+};
+
+////////////////////////////////////////////////////////////////////////
+// Extract to contiguous array scalar object
+////////////////////////////////////////////////////////////////////////
+template<class vobj> inline void extract(const vobj &vec,std::vector<typename vobj::scalar_object> &extracted)
+{
+  typedef typename vobj::scalar_type scalar_type ;
+  typedef typename vobj::vector_type vector_type ;
+
+  const int Nsimd=vobj::vector_type::Nsimd();
+  const int words=sizeof(vobj)/sizeof(vector_type);
+
+  extracted.resize(Nsimd);
+
+  std::vector<scalar_type *> pointers(Nsimd);
+  for(int i=0;i<Nsimd;i++) 
+    pointers[i] =(scalar_type *)& extracted[i];
+
+  vector_type *vp = (vector_type *)&vec;
+  for(int w=0;w<words;w++){
+    extract<vector_type,scalar_type>(&vp[w],pointers,w);
+  }
+}
+////////////////////////////////////////////////////////////////////////
+// Extract to a bunch of scalar object pointers, with offset
+////////////////////////////////////////////////////////////////////////
+template<class vobj> inline 
+void extract(const vobj &vec,std::vector<typename vobj::scalar_object *> &extracted, int offset)
+{
+
+  typedef typename vobj::scalar_type scalar_type ;
+  typedef typename vobj::vector_type vector_type ;
+
+  const int words=sizeof(vobj)/sizeof(vector_type);
+  const int Nsimd=vobj::vector_type::Nsimd();
+
+  assert(extracted.size()==Nsimd);
+
+  std::vector<scalar_type *> pointers(Nsimd);
+  for(int i=0;i<Nsimd;i++) {
+    pointers[i] =(scalar_type *)& extracted[i][offset];
+  }
+
+  vector_type *vp = (vector_type *)&vec;
+  for(int w=0;w<words;w++){
+    extract<vector_type,scalar_type>(&vp[w],pointers,w);
+  }
+}
+
+////////////////////////////////////////////////////////////////////////
+// Merge a contiguous array of scalar objects
+////////////////////////////////////////////////////////////////////////
+template<class vobj> inline 
+void merge(vobj &vec,std::vector<typename vobj::scalar_object> &extracted)
+{
+  typedef typename vobj::scalar_type scalar_type ;
+  typedef typename vobj::vector_type vector_type ;
+  
+  const int Nsimd=vobj::vector_type::Nsimd();
+  const int words=sizeof(vobj)/sizeof(vector_type);
+
+  assert(extracted.size()==Nsimd);
+
+  std::vector<scalar_type *> pointers(Nsimd);
+  for(int i=0;i<Nsimd;i++) 
+    pointers[i] =(scalar_type *)& extracted[i];
+  
+  vector_type *vp = (vector_type *)&vec;
+  for(int w=0;w<words;w++){
+    merge<vector_type,scalar_type>(&vp[w],pointers,w);
+  }
+}
+
+////////////////////////////////////////////////////////////////////////
+// Merge a bunch of different scalar object pointers, with offset
+////////////////////////////////////////////////////////////////////////
+template<class vobj> inline 
+void merge(vobj &vec,std::vector<typename vobj::scalar_object *> &extracted,int offset)
+{
+  typedef typename vobj::scalar_type scalar_type ;
+  typedef typename vobj::vector_type vector_type ;
+  
+  const int Nsimd=vobj::vector_type::Nsimd();
+  const int words=sizeof(vobj)/sizeof(vector_type);
+
+  assert(extracted.size()==Nsimd);
+
+  std::vector<scalar_type *> pointers(Nsimd);
+  for(int i=0;i<Nsimd;i++) 
+    pointers[i] =(scalar_type *)& extracted[i][offset];
+  
+  vector_type *vp = (vector_type *)&vec;
+  assert((void *)vp!=NULL);
+  for(int w=0;w<words;w++){
+    merge<vector_type,scalar_type>(&vp[w],pointers,w);
+  }
+}
+}
+#endif
diff --git a/lib/Grid_lattice.h b/lib/Grid_lattice.h
index 7256459c..c76fa0a9 100644
--- a/lib/Grid_lattice.h
+++ b/lib/Grid_lattice.h
@@ -95,6 +95,7 @@ public:
 #include <lattice/Grid_lattice_reduction.h>
 #include <lattice/Grid_lattice_peekpoke.h>
 #include <lattice/Grid_lattice_reality.h>
+#include <Grid_extract.h>
 #include <lattice/Grid_lattice_coordinate.h>
 #include <lattice/Grid_lattice_rng.h>
 #include <lattice/Grid_lattice_transfer.h>
diff --git a/lib/Grid_simd.h b/lib/Grid_simd.h
index a059cc23..4e33604d 100644
--- a/lib/Grid_simd.h
+++ b/lib/Grid_simd.h
@@ -133,68 +133,6 @@ namespace Grid {
 #endif
 
 
-/////////////////////////////////////////////////////////////////
-// Generic extract/merge/permute
-/////////////////////////////////////////////////////////////////
-
-template<class vsimd,class scalar>
-inline void Gextract(const vsimd &y,std::vector<scalar *> &extracted){
-  // FIXME: bounce off memory is painful
-  int Nextr=extracted.size();
-  int Nsimd=vsimd::Nsimd();
-  int s=Nsimd/Nextr;
-
-  std::vector<scalar,alignedAllocator<scalar> > buf(Nsimd); 
-
-  vstore(y,&buf[0]);
-  for(int i=0;i<Nextr;i++){
-    *extracted[i] = buf[i*s];
-    extracted[i]++;
-  }
-};
-template<class vsimd,class scalar>
-inline void Gextract(const vsimd &y,std::vector<scalar> &extracted){
-  int Nextr=extracted.size();
-  int Nsimd=vsimd::Nsimd();
-  int s=Nsimd/Nextr;
-
-  std::vector<scalar,alignedAllocator<scalar> > buf(Nsimd); 
-
-  vstore(y,&buf[0]);
-  for(int i=0;i<Nextr;i++){
-    extracted[i] = buf[i*s];
-  }
-};
-template<class vsimd,class scalar>
-inline void Gmerge(vsimd &y,std::vector<scalar *> &extracted){
-  int Nextr=extracted.size();
-  int Nsimd=vsimd::Nsimd();
-  int s=Nsimd/Nextr;
-
-  std::vector<scalar> buf(Nsimd); 
-  for(int i=0;i<Nextr;i++){
-    for(int ii=0;ii<s;ii++){
-      buf[i*s+ii]=*extracted[i];
-    }
-    extracted[i]++;
-  }
-  vset(y,&buf[0]); 
-};
-template<class vsimd,class scalar>
-inline void Gmerge(vsimd &y,std::vector<scalar> &extracted){
-  int Nextr=extracted.size();
-  int Nsimd=vsimd::Nsimd();
-  int s=Nsimd/Nextr;
-
-  std::vector<scalar> buf(Nsimd); 
-  for(int i=0;i<Nextr;i++){
-    for(int ii=0;ii<s;ii++){
-      buf[i*s+ii]=extracted[i];
-    }
-  }
-  vset(y,&buf[0]); 
-};
-
 //////////////////////////////////////////////////////////
 // Permute
 // Permute 0 every ABCDEFGH -> BA DC FE HG
diff --git a/lib/Grid_stencil.h b/lib/Grid_stencil.h
index afb11db6..fdc8ece9 100644
--- a/lib/Grid_stencil.h
+++ b/lib/Grid_stencil.h
@@ -48,100 +48,6 @@ namespace Grid {
   } ;
 
 
-///////////////////////////////////////////////////////////////////
-// Gather for when there is no need to SIMD split with compression
-///////////////////////////////////////////////////////////////////
-template<class vobj,class cobj,class compressor> void 
-Gather_plane_simple (const Lattice<vobj> &rhs,std::vector<cobj,alignedAllocator<cobj> > &buffer,int dimension,int plane,int cbmask,compressor &compress)
-{
-  int rd = rhs._grid->_rdimensions[dimension];
-
-  if ( !rhs._grid->CheckerBoarded(dimension) ) {
-
-    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-    int o   = 0;                                    // relative offset to base within plane
-    int bo  = 0;                                    // offset in buffer
-
-    // Simple block stride gather of SIMD objects
-#pragma omp parallel for collapse(2)
-    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-	buffer[bo++]=compress(rhs._odata[so+o+b]);
-      }
-      o +=rhs._grid->_slice_stride[dimension];
-    }
-
-  } else { 
-
-    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-    int o   = 0;                                      // relative offset to base within plane
-    int bo  = 0;                                      // offset in buffer
-
-#pragma omp parallel for collapse(2)
-    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-
-	int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);// Could easily be a table lookup
-	if ( ocb &cbmask ) {
-	  buffer[bo]=compress(rhs._odata[so+o+b]);
-	  bo++;
-	}
-
-      }
-      o +=rhs._grid->_slice_stride[dimension];
-    }
-  }
-}
-
-///////////////////////////////////////////////////////////////////
-// Gather for when there *is* need to SIMD split with compression
-///////////////////////////////////////////////////////////////////
-template<class cobj,class vobj,class compressor> void 
-Gather_plane_extract(const Lattice<vobj> &rhs,std::vector<typename cobj::scalar_type *> pointers,int dimension,int plane,int cbmask,compressor &compress)
-{
-  int rd = rhs._grid->_rdimensions[dimension];
-
-  if ( !rhs._grid->CheckerBoarded(dimension) ) {
-
-    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-    int o   = 0;                                    // relative offset to base within plane
-    int bo  = 0;                                    // offset in buffer
-
-    // Simple block stride gather of SIMD objects
-#pragma omp parallel for collapse(2)
-    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-	cobj temp;
-	temp=compress(rhs._odata[so+o+b]);
-	extract(temp,pointers);
-      }
-      o +=rhs._grid->_slice_stride[dimension];
-    }
-
-  } else { 
-
-    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-    int o   = 0;                                      // relative offset to base within plane
-    int bo  = 0;                                      // offset in buffer
-    
-#pragma omp parallel for collapse(2)
-    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-
-	int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);
-	if ( ocb & cbmask ) {
-	  cobj temp; 
-	  temp =compress(rhs._odata[so+o+b]);
-	  extract(temp,pointers);
-	}
-
-      }
-      o +=rhs._grid->_slice_stride[dimension];
-    }
-  }
-}
-
-
   class CartesianStencil { // Stencil runs along coordinate axes only; NO diagonal fill in.
   public:
 
@@ -184,6 +90,7 @@ Gather_plane_extract(const Lattice<vobj> &rhs,std::vector<typename cobj::scalar_
       template<class vobj,class cobj, class compressor> void 
 	HaloExchange(const Lattice<vobj> &source,std::vector<cobj,alignedAllocator<cobj> > &u_comm_buf,compressor &compress)
       {
+	std::cout<< "HaloExchange comm_buf.size()="<< u_comm_buf.size()<<" unified_buffer_size"<< _unified_buffer_size<< std::endl;
 	// conformable(source._grid,_grid);
 	assert(source._grid==_grid);
 	if (u_comm_buf.size() != _unified_buffer_size ) u_comm_buf.resize(_unified_buffer_size);
@@ -234,6 +141,7 @@ Gather_plane_extract(const Lattice<vobj> &rhs,std::vector<typename cobj::scalar_
 	    }
 	  }
 	}
+	std::cout<< "HaloExchange complete"<< std::endl;
       }
 
       template<class vobj,class cobj, class compressor> 
@@ -318,6 +226,7 @@ Gather_plane_extract(const Lattice<vobj> &rhs,std::vector<typename cobj::scalar_
 
 	  typedef typename cobj::vector_type vector_type;
 	  typedef typename cobj::scalar_type scalar_type;
+	  typedef typename cobj::scalar_object scalar_object;
 	  
 	  int fd = _grid->_fdimensions[dimension];
 	  int rd = _grid->_rdimensions[dimension];
@@ -340,12 +249,12 @@ Gather_plane_extract(const Lattice<vobj> &rhs,std::vector<typename cobj::scalar_
 	  int words = sizeof(cobj)/sizeof(vector_type);
 
 	  /*   FIXME ALTERNATE BUFFER DETERMINATION */
-	  std::vector<std::vector<scalar_type> > send_buf_extract(Nsimd,std::vector<scalar_type>(buffer_size*words) ); 
-	  std::vector<std::vector<scalar_type> > recv_buf_extract(Nsimd,std::vector<scalar_type>(buffer_size*words) );
-	  int bytes = buffer_size*words*sizeof(scalar_type);
+	  std::vector<std::vector<scalar_object> > send_buf_extract(Nsimd,std::vector<scalar_object>(buffer_size) ); 
+	  std::vector<std::vector<scalar_object> > recv_buf_extract(Nsimd,std::vector<scalar_object>(buffer_size) );
+	  int bytes = buffer_size*sizeof(scalar_object);
 
-	  std::vector<scalar_type *> pointers(Nsimd);  //
-	  std::vector<scalar_type *> rpointers(Nsimd); // received pointers
+	  std::vector<scalar_object *> pointers(Nsimd);  //
+	  std::vector<scalar_object *> rpointers(Nsimd); // received pointers
 	  
 	  ///////////////////////////////////////////
 	  // Work out what to send where
@@ -353,62 +262,77 @@ Gather_plane_extract(const Lattice<vobj> &rhs,std::vector<typename cobj::scalar_
 
 	  int cb    = (cbmask==0x2)? 1 : 0;
 	  int sshift= _grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,cb);
-
+	  
 	  // loop over outer coord planes orthog to dim
 	  for(int x=0;x<rd;x++){       
-	    
-	    // FIXME call local permute copy if none are offnode.
-	    for(int i=0;i<Nsimd;i++){       
-	      pointers[i] = (scalar_type *)&send_buf_extract[i][0];
-	    }
-	    int sx   = (x+sshift)%rd;
 
-	    std::cout<< "Gathering "<< x <<std::endl;
-	    Gather_plane_extract<cobj>(rhs,pointers,dimension,sx,cbmask,compress);
-	    std::cout<< "Gathered "<<std::endl;
-	    for(int i=0;i<Nsimd;i++){
-      
-	      int inner_bit = (Nsimd>>(permute_type+1));
-	      int ic= (i&inner_bit)? 1:0;
-
-	      int my_coor          = rd*ic + x;
-	      int nbr_coor         = my_coor+sshift;
-	      int nbr_proc = ((nbr_coor)/ld) % pd;// relative shift in processors
-
-	      int nbr_ic   = (nbr_coor%ld)/rd;    // inner coord of peer
-	      int nbr_ox   = (nbr_coor%rd);       // outer coord of peer
-	      int nbr_lane = (i&(~inner_bit));
-
-	      int recv_from_rank;
-	      int xmit_to_rank;
-
-	      if (nbr_ic) nbr_lane|=inner_bit;
-	      assert (sx == nbr_ox);
-
-	      if(nbr_proc){
-
-		std::cout<< "MPI sending "<<std::endl;
-		_grid->ShiftedRanks(dimension,nbr_proc,xmit_to_rank,recv_from_rank); 
-
-		_grid->SendToRecvFrom((void *)&send_buf_extract[nbr_lane][0],
-				     xmit_to_rank,
-				     (void *)&recv_buf_extract[i][0],
-				     recv_from_rank,
-				     bytes);
-		std::cout<< "MPI complete "<<std::endl;
-
-		rpointers[i] = (scalar_type *)&recv_buf_extract[i][0];
-	      } else { 
-		rpointers[i] = (scalar_type *)&send_buf_extract[nbr_lane][0];
+	    int any_offnode = ( ((x+sshift)%fd) >= rd );
+	    std::cout<<"any_offnode ="<<any_offnode<<std::endl;
+	    if ( any_offnode ) {
+	      // FIXME call local permute copy if none are offnode.
+	      for(int i=0;i<Nsimd;i++){       
+		pointers[i] = &send_buf_extract[i][0];
 	      }
-	    }
+	      int sx   = (x+sshift)%rd;
+	      
+	      std::cout<< "Gathering "<< x <<std::endl;
+	      Gather_plane_extract<cobj>(rhs,pointers,dimension,sx,cbmask,compress);
+	      std::cout<< "Gathered "<<std::endl;
+	      for(int i=0;i<Nsimd;i++){
+		
+		std::vector<int> icoor;
+		_grid->iCoorFromIindex(icoor,i);
 
-	    // Here we don't want to scatter, just place into a buffer.
-	    std::cout<< "merging "<<std::endl;
-	    for(int i=0;i<buffer_size;i++){
-	      merge(u_comm_buf[u_comm_offset+i],rpointers);
-	    }
+		int inner_bit = (Nsimd>>(permute_type+1));
+		int ic= (i&inner_bit)? 1:0;
+		assert(ic==icoor[dimension]);
 
+		int my_coor          = rd*ic + x;
+		int nbr_coor         = my_coor+sshift;
+		int nbr_proc = ((nbr_coor)/ld) % pd;// relative shift in processors
+		int nbr_lcoor= (nbr_coor%ld);
+		int nbr_ic   = (nbr_lcoor)/rd;    // inner coord of peer
+		int nbr_ox   = (nbr_lcoor%rd);    // outer coord of peer
+		int nbr_lane = (i&(~inner_bit));
+		
+		int recv_from_rank;
+		int xmit_to_rank;
+		
+		if (nbr_ic) nbr_lane|=inner_bit;
+		assert (sx == nbr_ox);
+
+   std::cout<<"nbr_proc "<<nbr_proc<< " x "<<x<<" nbr_x "<<nbr_ox << " lane "<<i << " nbr_lane "<<nbr_lane
+	    << " nbr_ic "<<nbr_ic  << " mycoor "<< my_coor<< " nbr_coor "<<nbr_coor<<std::endl;
+		
+		if(nbr_proc){
+		  
+		  std::cout<< "MPI sending "<<std::endl;
+		  _grid->ShiftedRanks(dimension,nbr_proc,xmit_to_rank,recv_from_rank); 
+		  
+		  _grid->SendToRecvFrom((void *)&send_buf_extract[nbr_lane][0],
+					xmit_to_rank,
+					(void *)&recv_buf_extract[i][0],
+					recv_from_rank,
+					bytes);
+		  std::cout<< "MPI complete "<<std::endl;
+		  
+		  rpointers[i] = &recv_buf_extract[i][0];
+		  std::cout<<"lane "<<i<<" data "<<*( (Real *) rpointers[i])<<std::endl;
+		} else { 
+		  rpointers[i] = &send_buf_extract[nbr_lane][0];
+		  std::cout<<"lane "<<i<<" data "<<*( (Real *) rpointers[i])<<std::endl;
+		}
+	      }
+
+	      // Here we don't want to scatter, just place into a buffer.
+	      std::cout<< "merging u_comm_offset "<< u_comm_offset<<" comm_buf_size" << u_comm_buf.size() <<std::endl;
+
+	      for(int i=0;i<buffer_size;i++){
+		assert(u_comm_offset+i<_unified_buffer_size);
+		merge(u_comm_buf[u_comm_offset+i],rpointers,i);
+	      }
+	      u_comm_offset+=buffer_size;
+	    }
 	  }
 	}
   };
diff --git a/lib/Grid_where.h b/lib/Grid_where.h
index 3a7052ad..588c0596 100644
--- a/lib/Grid_where.h
+++ b/lib/Grid_where.h
@@ -15,6 +15,7 @@ inline void where(Lattice<vobj> &ret,const Lattice<iobj> &predicate,Lattice<vobj
   conformable(iftrue,ret);
 
   GridBase *grid=iftrue._grid;
+  typedef typename vobj::scalar_object scalar_object;
   typedef typename vobj::scalar_type scalar_type;
   typedef typename vobj::vector_type vector_type;
   typedef typename iobj::vector_type mask_type;
@@ -23,27 +24,21 @@ inline void where(Lattice<vobj> &ret,const Lattice<iobj> &predicate,Lattice<vobj
   const int words = sizeof(vobj)/sizeof(vector_type);
 
   std::vector<Integer> mask(Nsimd);
-  std::vector<std::vector<scalar_type> > truevals (Nsimd,std::vector<scalar_type>(words) );
-  std::vector<std::vector<scalar_type> > falsevals(Nsimd,std::vector<scalar_type>(words) );
-  std::vector<scalar_type *> pointers(Nsimd);
+  std::vector<scalar_object> truevals (Nsimd);
+  std::vector<scalar_object> falsevals(Nsimd);
 
 #pragma omp parallel for
   for(int ss=0;ss<iftrue._grid->oSites(); ss++){
 
-    for(int s=0;s<Nsimd;s++) pointers[s] = & truevals[s][0];
-    extract(iftrue._odata[ss]   ,pointers);
-
-    for(int s=0;s<Nsimd;s++) pointers[s] = & falsevals[s][0];
-    extract(iffalse._odata[ss]  ,pointers);
-
-    extract(TensorRemove(predicate._odata[ss]),mask);
+    extract(iftrue._odata[ss]   ,truevals);
+    extract(iffalse._odata[ss]  ,falsevals);
+    extract<vInteger,Integer>(TensorRemove(predicate._odata[ss]),mask);
 
     for(int s=0;s<Nsimd;s++){
-      if (mask[s]) pointers[s]=&truevals[s][0];
-      else         pointers[s]=&falsevals[s][0];
+      if (mask[s]) falsevals[s]=truevals[s];
     }
 
-    merge(ret._odata[ss],pointers);
+    merge(ret._odata[ss],falsevals);
   }
 }
 
diff --git a/lib/cshift/Grid_cshift_common.h b/lib/cshift/Grid_cshift_common.h
index eebc895e..6ae6df71 100644
--- a/lib/cshift/Grid_cshift_common.h
+++ b/lib/cshift/Grid_cshift_common.h
@@ -2,176 +2,155 @@
 #define _GRID_CSHIFT_COMMON_H_
 
 namespace Grid {
+
+template<class vobj>
+class SimpleCompressor {
+public:
+  void Point(int) {};
+
+  vobj operator() (const vobj &arg) {
+    return arg;
+  }
+};
+
+///////////////////////////////////////////////////////////////////
+// Gather for when there is no need to SIMD split with compression
+///////////////////////////////////////////////////////////////////
+template<class vobj,class cobj,class compressor> void 
+Gather_plane_simple (const Lattice<vobj> &rhs,std::vector<cobj,alignedAllocator<cobj> > &buffer,int dimension,int plane,int cbmask,compressor &compress)
+{
+  int rd = rhs._grid->_rdimensions[dimension];
+
+  if ( !rhs._grid->CheckerBoarded(dimension) ) {
+    cbmask = 0x3;
+  }
+
+  int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+  int o   = 0;                                      // relative offset to base within plane
+  int bo  = 0;                                      // offset in buffer
+  
+#pragma omp parallel for collapse(2)
+  for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+    for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+      int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);// Could easily be a table lookup
+      if ( ocb &cbmask ) {
+	buffer[bo]=compress(rhs._odata[so+o+b]);
+	bo++;
+      }
+    }
+    o +=rhs._grid->_slice_stride[dimension];
+  }
+}
+
+
+///////////////////////////////////////////////////////////////////
+// Gather for when there *is* need to SIMD split with compression
+///////////////////////////////////////////////////////////////////
+template<class cobj,class vobj,class compressor> void 
+Gather_plane_extract(const Lattice<vobj> &rhs,std::vector<typename cobj::scalar_object *> pointers,int dimension,int plane,int cbmask,compressor &compress)
+{
+  int rd = rhs._grid->_rdimensions[dimension];
+
+  if ( !rhs._grid->CheckerBoarded(dimension) ) {
+    cbmask = 0x3;
+  }
+
+  int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+  int o   = 0;                                      // relative offset to base within plane
+  int bo  = 0;                                      // offset in buffer
+    
+#pragma omp parallel for collapse(2)
+  for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+    for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+
+      int offset = b+n*rhs._grid->_slice_block[dimension];
+
+      int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);
+      if ( ocb & cbmask ) {
+	cobj temp; 
+	temp =compress(rhs._odata[so+o+b]);
+	extract<cobj>(temp,pointers,offset);
+      }
+      
+    }
+    o +=rhs._grid->_slice_stride[dimension];
+  }
+}
+
 //////////////////////////////////////////////////////
 // Gather for when there is no need to SIMD split
 //////////////////////////////////////////////////////
 template<class vobj> void Gather_plane_simple (const Lattice<vobj> &rhs,std::vector<vobj,alignedAllocator<vobj> > &buffer,             int dimension,int plane,int cbmask)
 {
-  int rd = rhs._grid->_rdimensions[dimension];
-
-  if ( !rhs._grid->CheckerBoarded(dimension) ) {
-
-    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-    int o   = 0;                                    // relative offset to base within plane
-    int bo  = 0;                                    // offset in buffer
-
-    // Simple block stride gather of SIMD objects
-#pragma omp parallel for collapse(2)
-    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-	buffer[bo++]=rhs._odata[so+o+b];
-      }
-      o +=rhs._grid->_slice_stride[dimension];
-    }
-
-  } else { 
-
-    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-    int o   = 0;                                      // relative offset to base within plane
-    int bo  = 0;                                      // offset in buffer
-
-#pragma omp parallel for collapse(2)
-    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-
-	int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);// Could easily be a table lookup
-	if ( ocb &cbmask ) {
-	  buffer[bo]=rhs._odata[so+o+b];
-	  bo++;
-	}
-
-      }
-      o +=rhs._grid->_slice_stride[dimension];
-    }
-  }
+  SimpleCompressor<vobj> dontcompress;
+  Gather_plane_simple (rhs,buffer,dimension,plane,cbmask,dontcompress);
 }
 
 //////////////////////////////////////////////////////
 // Gather for when there *is* need to SIMD split
 //////////////////////////////////////////////////////
- template<class vobj,class scalar_type> void Gather_plane_extract(const Lattice<vobj> &rhs,std::vector<scalar_type *> pointers,int dimension,int plane,int cbmask)
+template<class vobj> void Gather_plane_extract(const Lattice<vobj> &rhs,std::vector<typename vobj::scalar_object *> pointers,int dimension,int plane,int cbmask)
 {
-  int rd = rhs._grid->_rdimensions[dimension];
-
-  if ( !rhs._grid->CheckerBoarded(dimension) ) {
-
-    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-    int o   = 0;                                    // relative offset to base within plane
-    int bo  = 0;                                    // offset in buffer
-
-    // Simple block stride gather of SIMD objects
-#pragma omp parallel for collapse(2)
-    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-	extract(rhs._odata[so+o+b],pointers);
-      }
-      o +=rhs._grid->_slice_stride[dimension];
-    }
-
-  } else { 
-
-    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-    int o   = 0;                                      // relative offset to base within plane
-    int bo  = 0;                                      // offset in buffer
-    
-#pragma omp parallel for collapse(2)
-    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-
-	int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);
-	if ( ocb & cbmask ) {
-	  extract(rhs._odata[so+o+b],pointers);
-	}
-
-      }
-      o +=rhs._grid->_slice_stride[dimension];
-    }
-  }
+  SimpleCompressor<vobj> dontcompress;
+  Gather_plane_extract<vobj,vobj,decltype(dontcompress)>(rhs,pointers,dimension,plane,cbmask,dontcompress);
 }
 
 //////////////////////////////////////////////////////
 // Scatter for when there is no need to SIMD split
 //////////////////////////////////////////////////////
-template<class vobj> void Scatter_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllocator<vobj> > &buffer,             int dimension,int plane,int cbmask)
+template<class vobj> void Scatter_plane_simple (Lattice<vobj> &rhs,std::vector<vobj,alignedAllocator<vobj> > &buffer, int dimension,int plane,int cbmask)
 {
   int rd = rhs._grid->_rdimensions[dimension];
 
   if ( !rhs._grid->CheckerBoarded(dimension) ) {
+    cbmask=0x3;
+  }
 
-    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-    int o   = 0;                                    // relative offset to base within plane
-    int bo  = 0;                                    // offset in buffer
-
-    // Simple block stride gather of SIMD objects
-#pragma omp parallel for collapse(2)
-    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-	rhs._odata[so+o+b]=buffer[bo++];
-      }
-      o +=rhs._grid->_slice_stride[dimension];
-    }
-
-  } else { 
-
-    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-    int o   = 0;                                      // relative offset to base within plane
-    int bo  = 0;                                      // offset in buffer
+  int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+  int o   = 0;                                      // relative offset to base within plane
+  int bo  = 0;                                      // offset in buffer
     
 #pragma omp parallel for collapse(2)
-    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-
-	int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);// Could easily be a table lookup
-	if ( ocb & cbmask ) {
-	  rhs._odata[so+o+b]=buffer[bo++];
-	}
+  for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+    for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
 
+      int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);// Could easily be a table lookup
+      if ( ocb & cbmask ) {
+	rhs._odata[so+o+b]=buffer[bo++];
       }
-      o +=rhs._grid->_slice_stride[dimension];
+
     }
+    o +=rhs._grid->_slice_stride[dimension];
   }
 }
 
 //////////////////////////////////////////////////////
 // Scatter for when there *is* need to SIMD split
 //////////////////////////////////////////////////////
- template<class vobj,class scalar_type> void Scatter_plane_merge(Lattice<vobj> &rhs,std::vector<scalar_type *> pointers,int dimension,int plane,int cbmask)
+ template<class vobj,class cobj> void Scatter_plane_merge(Lattice<vobj> &rhs,std::vector<cobj *> pointers,int dimension,int plane,int cbmask)
 {
   int rd = rhs._grid->_rdimensions[dimension];
 
   if ( !rhs._grid->CheckerBoarded(dimension) ) {
+    cbmask=0x3;
+  }
 
-    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-    int o   = 0;                                    // relative offset to base within plane
-    int bo  = 0;                                    // offset in buffer
-
-    // Simple block stride gather of SIMD objects
-#pragma omp parallel for collapse(2)
-    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-	merge(rhs._odata[so+o+b],pointers);
-      }
-      o +=rhs._grid->_slice_stride[dimension];
-    }
-
-  } else { 
-
-    int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-    int o   = 0;                                      // relative offset to base within plane
-    int bo  = 0;                                      // offset in buffer
+  int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+  int o   = 0;                                      // relative offset to base within plane
+  int bo  = 0;                                      // offset in buffer
     
 #pragma omp parallel for collapse(2)
-    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-
-	int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);
-	if ( ocb&cbmask ) {
-	  merge(rhs._odata[so+o+b],pointers);
-	}
+  for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+    for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
 
+      int offset = b+n*rhs._grid->_slice_block[dimension];
+      int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);
+      if ( ocb&cbmask ) {
+	merge(rhs._odata[so+o+b],pointers,offset);
       }
-      o +=rhs._grid->_slice_stride[dimension];
+      
     }
+    o +=rhs._grid->_slice_stride[dimension];
   }
 }
 
@@ -183,40 +162,26 @@ template<class vobj> void Copy_plane(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int
   int rd = rhs._grid->_rdimensions[dimension];
 
   if ( !rhs._grid->CheckerBoarded(dimension) ) {
+    cbmask=0x3;
+  }
 
-    int o   = 0;                                     // relative offset to base within plane
-    int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-    int lo  = lplane*lhs._grid->_ostride[dimension]; // offset in buffer
 
-  // Simple block stride gather of SIMD objects
+  int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+  int lo  = lplane*lhs._grid->_ostride[dimension]; // base offset for start of plane 
+  int o   = 0;                                     // relative offset to base within plane
+  
 #pragma omp parallel for collapse(2)
-    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+  for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+    for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+      
+      int ocb=1<<lhs._grid->CheckerBoardFromOindex(o+b);
+      
+      if ( ocb&cbmask ) {
 	lhs._odata[lo+o+b]=rhs._odata[ro+o+b];
       }
-      o +=rhs._grid->_slice_stride[dimension];
+      
     }
-
-  } else {
-
-    int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-    int lo  = lplane*lhs._grid->_ostride[dimension]; // base offset for start of plane 
-    int o   = 0;                                     // relative offset to base within plane
-
-#pragma omp parallel for collapse(2)
-    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-
-	int ocb=1<<lhs._grid->CheckerBoardFromOindex(o+b);
-
-	if ( ocb&cbmask ) {
-	  lhs._odata[lo+o+b]=rhs._odata[ro+o+b];
-	}
-
-      }
-      o +=rhs._grid->_slice_stride[dimension];
-    }
-
+    o +=rhs._grid->_slice_stride[dimension];
   }
 }
 
@@ -224,42 +189,25 @@ template<class vobj> void Copy_plane_permute(Lattice<vobj>& lhs,Lattice<vobj> &r
 {
   int rd = rhs._grid->_rdimensions[dimension];
 
-
   if ( !rhs._grid->CheckerBoarded(dimension) ) {
+    cbmask=0x3;
+  }
 
-    int o   = 0;                                     // relative offset to base within plane
-    int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-    int lo  = lplane*rhs._grid->_ostride[dimension]; // offset in buffer
-
-  // Simple block stride gather of SIMD objects
+  int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
+  int lo  = lplane*lhs._grid->_ostride[dimension]; // base offset for start of plane 
+  int o   = 0;                                     // relative offset to base within plane
+  
 #pragma omp parallel for collapse(2)
-    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+  for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
+    for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+      
+      int ocb=1<<lhs._grid->CheckerBoardFromOindex(o+b);
+      if ( ocb&cbmask ) {
 	permute(lhs._odata[lo+o+b],rhs._odata[ro+o+b],permute_type);
       }
-      o +=rhs._grid->_slice_stride[dimension];
+      
     }
-
-  } else {
-
-    int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-    int lo  = lplane*lhs._grid->_ostride[dimension]; // base offset for start of plane 
-    int o   = 0;                                     // relative offset to base within plane
-    
-#pragma omp parallel for collapse(2)
-    for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-      for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-
-	int ocb=1<<lhs._grid->CheckerBoardFromOindex(o+b);
-
-	if ( ocb&cbmask ) {
-	  permute(lhs._odata[lo+o+b],rhs._odata[ro+o+b],permute_type);
-	}
-
-      }
-      o +=rhs._grid->_slice_stride[dimension];
-    }
-
+    o +=rhs._grid->_slice_stride[dimension];
   }
 }
 
diff --git a/lib/cshift/Grid_cshift_mpi.h b/lib/cshift/Grid_cshift_mpi.h
index ab9237ad..7c77ebc9 100644
--- a/lib/cshift/Grid_cshift_mpi.h
+++ b/lib/cshift/Grid_cshift_mpi.h
@@ -133,6 +133,7 @@ template<class vobj> void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &r
   GridBase *grid=rhs._grid;
   const int Nsimd = grid->Nsimd();
   typedef typename vobj::vector_type vector_type;
+  typedef typename vobj::scalar_object scalar_object;
   typedef typename vobj::scalar_type scalar_type;
    
   int fd = grid->_fdimensions[dimension];
@@ -155,12 +156,12 @@ template<class vobj> void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &r
   int buffer_size = grid->_slice_nblock[dimension]*grid->_slice_block[dimension];
   int words = sizeof(vobj)/sizeof(vector_type);
 
-  std::vector<std::vector<scalar_type> > send_buf_extract(Nsimd,std::vector<scalar_type>(buffer_size*words) );
-  std::vector<std::vector<scalar_type> > recv_buf_extract(Nsimd,std::vector<scalar_type>(buffer_size*words) );
-  int bytes = buffer_size*words*sizeof(scalar_type);
+  std::vector<std::vector<scalar_object> > send_buf_extract(Nsimd,std::vector<scalar_object>(buffer_size) );
+  std::vector<std::vector<scalar_object> > recv_buf_extract(Nsimd,std::vector<scalar_object>(buffer_size) );
+  int bytes = buffer_size*sizeof(scalar_object);
 
-  std::vector<scalar_type *> pointers(Nsimd);  // 
-  std::vector<scalar_type *> rpointers(Nsimd); // received pointers
+  std::vector<scalar_object *>  pointers(Nsimd);  // 
+  std::vector<scalar_object *> rpointers(Nsimd); // received pointers
 
   ///////////////////////////////////////////
   // Work out what to send where
@@ -171,10 +172,9 @@ template<class vobj> void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &r
   // loop over outer coord planes orthog to dim
   for(int x=0;x<rd;x++){       
 
-  // FIXME call local permute copy if none are offnode.
-
+    // FIXME call local permute copy if none are offnode.
     for(int i=0;i<Nsimd;i++){       
-      pointers[i] = (scalar_type *)&send_buf_extract[i][0];
+      pointers[i] = &send_buf_extract[i][0];
     }
     int sx   = (x+sshift)%rd;
     Gather_plane_extract(rhs,pointers,dimension,sx,cbmask);
@@ -208,9 +208,9 @@ template<class vobj> void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &r
 			     recv_from_rank,
 			     bytes);
 
-	rpointers[i] = (scalar_type *)&recv_buf_extract[i][0];
+	rpointers[i] = &recv_buf_extract[i][0];
       } else { 
-	rpointers[i] = (scalar_type *)&send_buf_extract[nbr_lane][0];
+	rpointers[i] = &send_buf_extract[nbr_lane][0];
       }
 
     }
diff --git a/lib/lattice/Grid_lattice_coordinate.h b/lib/lattice/Grid_lattice_coordinate.h
index 8fba8b1b..a9b381fa 100644
--- a/lib/lattice/Grid_lattice_coordinate.h
+++ b/lib/lattice/Grid_lattice_coordinate.h
@@ -5,21 +5,23 @@ namespace Grid {
 
     template<class iobj> inline void LatticeCoordinate(Lattice<iobj> &l,int mu)
     {
+      typedef typename iobj::scalar_object scalar_object;
       typedef typename iobj::scalar_type scalar_type;
       typedef typename iobj::vector_type vector_type;
+
       GridBase *grid = l._grid;
       int Nsimd = grid->iSites();
+
       std::vector<int> gcoor;
       std::vector<scalar_type> mergebuf(Nsimd);
-      std::vector<scalar_type *> mergeptr(Nsimd);
+
       vector_type vI;
       for(int o=0;o<grid->oSites();o++){
 	for(int i=0;i<grid->iSites();i++){
 	  grid->RankIndexToGlobalCoor(grid->ThisRank(),o,i,gcoor);
-	  mergebuf[i]=gcoor[mu];
-	  mergeptr[i]=&mergebuf[i];
+	  mergebuf[i]=(Integer)gcoor[mu];
 	}
-	merge(vI,mergeptr);
+	AmergeA<vector_type,scalar_type>(vI,mergebuf);
 	l._odata[o]=vI;
       }
     };
diff --git a/lib/lattice/Grid_lattice_peekpoke.h b/lib/lattice/Grid_lattice_peekpoke.h
index 93245016..ae87c5d8 100644
--- a/lib/lattice/Grid_lattice_peekpoke.h
+++ b/lib/lattice/Grid_lattice_peekpoke.h
@@ -94,15 +94,12 @@ namespace Grid {
       grid->Broadcast(grid->BossRank(),s);
 
       std::vector<sobj> buf(Nsimd);
-      std::vector<scalar_type *> pointers(Nsimd);  
 
       // extract-modify-merge cycle is easiest way and this is not perf critical
       if ( rank == grid->ThisRank() ) {
-	for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
-	extract(l._odata[odx],pointers);
+	extract(l._odata[odx],buf);
 	buf[idx] = s;
-	for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
-	merge(l._odata[odx],pointers);
+	merge(l._odata[odx],buf);
       }
 
       return;
@@ -127,13 +124,12 @@ namespace Grid {
 
       int rank,odx,idx;
       grid->GlobalCoorToRankIndex(rank,odx,idx,site);
-      std::vector<sobj> buf(Nsimd);
-      std::vector<scalar_type *> pointers(Nsimd);  
-      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
 
-      extract(l._odata[odx],pointers);
-      
+      std::vector<sobj> buf(Nsimd);
+      extract(l._odata[odx],buf);
+
       s = buf[idx];
+
       grid->Broadcast(rank,s);
 
       return;
@@ -160,10 +156,8 @@ namespace Grid {
       odx= grid->oIndex(site);
 
       std::vector<sobj> buf(Nsimd);
-      std::vector<scalar_type *> pointers(Nsimd);  
-      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
 
-      extract(l._odata[odx],pointers);
+      extract(l._odata[odx],buf);
       
       s = buf[idx];
 
@@ -188,16 +182,13 @@ namespace Grid {
       odx= grid->oIndex(site);
 
       std::vector<sobj> buf(Nsimd);
-      std::vector<scalar_type *> pointers(Nsimd);  
-      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
 
       // extract-modify-merge cycle is easiest way and this is not perf critical
-      extract(l._odata[odx],pointers);
+      extract(l._odata[odx],buf);
       
       buf[idx] = s;
 
-      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
-      merge(l._odata[odx],pointers);
+      merge(l._odata[odx],buf);
 
       return;
     };
diff --git a/lib/lattice/Grid_lattice_reduction.h b/lib/lattice/Grid_lattice_reduction.h
index 4491e8ec..6679f29b 100644
--- a/lib/lattice/Grid_lattice_reduction.h
+++ b/lib/lattice/Grid_lattice_reduction.h
@@ -66,9 +66,7 @@ namespace Grid {
       }
       
       std::vector<sobj>               buf(Nsimd);
-      std::vector<scalar_type *> pointers(Nsimd);  
-      for(int i=0;i<Nsimd;i++) pointers[i] = (scalar_type *)&buf[i];
-      extract(vsum,pointers);
+      extract(vsum,buf);
 
       for(int i=0;i<Nsimd;i++) ssum = ssum + buf[i];
 
diff --git a/lib/lattice/Grid_lattice_rng.h b/lib/lattice/Grid_lattice_rng.h
index 0842c80c..f3cc0ddc 100644
--- a/lib/lattice/Grid_lattice_rng.h
+++ b/lib/lattice/Grid_lattice_rng.h
@@ -26,8 +26,21 @@ namespace Grid {
     }
 
   };
-  
 
+  // real scalars are one component
+  template<class scalar,class distribution,class generator> void fillScalar(scalar &s,distribution &dist,generator & gen)
+  {
+    s=dist(gen);
+  }
+  template<class distribution,class generator> void fillScalar(ComplexF &s,distribution &dist, generator &gen)
+  {
+    s=ComplexF(dist(gen),dist(gen));
+  }
+  template<class distribution,class generator> void fillScalar(ComplexD &s,distribution &dist,generator &gen)
+  {
+    s=ComplexD(dist(gen),dist(gen));
+  }
+  
   class GridRNGbase {
 
   public:
@@ -64,20 +77,6 @@ namespace Grid {
     }
 
 
-    // real scalars are one component
-    template<class scalar,class distribution> void fillScalar(scalar &s,distribution &dist)
-    {
-      s=dist(_generators[0]);
-    }
-    template<class distribution> void fillScalar(ComplexF &s,distribution &dist)
-    {
-      s=ComplexF(dist(_generators[0]),dist(_generators[0]));
-    }
-    template<class distribution> void fillScalar(ComplexD &s,distribution &dist)
-    {
-      s=ComplexD(dist(_generators[0]),dist(_generators[0]));
-    }
-
 
     template <class sobj,class distribution> inline void fill(sobj &l,distribution &dist){
 
@@ -88,7 +87,7 @@ namespace Grid {
       scalar_type *buf = (scalar_type *) & l;
 
       for(int idx=0;idx<words;idx++){
-	fillScalar(buf[idx],dist);
+	fillScalar(buf[idx],dist,_generators[0]);
       }
       
       CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
@@ -96,47 +95,47 @@ namespace Grid {
     };
 
     template <class distribution>  inline void fill(ComplexF &l,distribution &dist){
-      fillScalar(l,dist);
+      fillScalar(l,dist,_generators[0]);
       CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
     }
     template <class distribution>  inline void fill(ComplexD &l,distribution &dist){
-      fillScalar(l,dist);
+      fillScalar(l,dist,_generators[0]);
       CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
     }
     template <class distribution>  inline void fill(RealF &l,distribution &dist){
-      fillScalar(l,dist);
+      fillScalar(l,dist,_generators[0]);
       CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
     }
     template <class distribution>  inline void fill(RealD &l,distribution &dist){
-      fillScalar(l,dist);
+      fillScalar(l,dist,_generators[0]);
       CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
     }
     // vector fill
     template <class distribution>  inline void fill(vComplexF &l,distribution &dist){
       RealF *pointer=(RealF *)&l;
       for(int i=0;i<2*vComplexF::Nsimd();i++){
-	fillScalar(pointer[i],dist);
+	fillScalar(pointer[i],dist,_generators[0]);
       }
       CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
     }
     template <class distribution>  inline void fill(vComplexD &l,distribution &dist){
       RealD *pointer=(RealD *)&l;
       for(int i=0;i<2*vComplexD::Nsimd();i++){
-	fillScalar(pointer[i],dist);
+	fillScalar(pointer[i],dist,_generators[0]);
       }
       CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
     }
     template <class distribution>  inline void fill(vRealF &l,distribution &dist){
       RealF *pointer=(RealF *)&l;
       for(int i=0;i<vRealF::Nsimd();i++){
-	fillScalar(pointer[i],dist);
+	fillScalar(pointer[i],dist,_generators[0]);
       }
       CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
     }
     template <class distribution>  inline void fill(vRealD &l,distribution &dist){
       RealD *pointer=(RealD *)&l;
       for(int i=0;i<vRealD::Nsimd();i++){
-	fillScalar(pointer[i],dist);
+	fillScalar(pointer[i],dist,_generators[0]);
       }
       CartesianCommunicator::BroadcastWorld(0,(void *)&l,sizeof(l));
     }
@@ -187,18 +186,31 @@ namespace Grid {
     {
       std::vector<int> gcoor;
 
-      for(int gidx=0;gidx<_grid->_gsites;gidx++){
+      int gsites = _grid->_gsites;
+
+      typename source::result_type init = src();
+      std::ranlux48 pseeder(init);
+      std::uniform_int_distribution<uint64_t> ui;
+
+      for(int gidx=0;gidx<gsites;gidx++){
+
 	int rank,o_idx,i_idx;
 	_grid->GlobalIndexToGlobalCoor(gidx,gcoor);
 	_grid->GlobalCoorToRankIndex(rank,o_idx,i_idx,gcoor);
 
 	int l_idx=generator_idx(o_idx,i_idx);
+	
+	std::vector<int> site_seeds(4);
+	for(int i=0;i<4;i++){
+	  site_seeds[i]= ui(pseeder);
+	}
 
-	typename source::result_type init = src();
+	_grid->Broadcast(0,(void *)&site_seeds[0],sizeof(int)*site_seeds.size());
 
-	_grid->Broadcast(0,(void *)&init,sizeof(init));
 	if( rank == _grid->ThisRank() ){
-	  _generators[l_idx] = std::ranlux48(init);
+	  fixedSeed ssrc(site_seeds);
+	  typename source::result_type sinit = ssrc();
+	  _generators[l_idx] = std::ranlux48(sinit);
 	}
       }
       _seeded=1;
@@ -210,6 +222,7 @@ namespace Grid {
 
     template <class vobj,class distribution> inline void fill(Lattice<vobj> &l,distribution &dist){
 
+      typedef typename vobj::scalar_object scalar_object;
       typedef typename vobj::scalar_type scalar_type;
       typedef typename vobj::vector_type vector_type;
       
@@ -217,25 +230,22 @@ namespace Grid {
 
       int     Nsimd =_grid->Nsimd();
       int     osites=_grid->oSites();
+      int words=sizeof(scalar_object)/sizeof(scalar_type);
 
-      int words = sizeof(vobj)/sizeof(vector_type);
-      std::vector<std::vector<scalar_type> > buf(Nsimd,std::vector<scalar_type>(words));
-      std::vector<scalar_type *> pointers(Nsimd);  
+      std::vector<scalar_object> buf(Nsimd);
       
       for(int ss=0;ss<osites;ss++){
-
 	for(int si=0;si<Nsimd;si++){
 
 	  int gdx = generator_idx(ss,si); // index of generator state
-
-	  pointers[si] = (scalar_type *)&buf[si][0];
+	  scalar_type *pointer = (scalar_type *)&buf[si];
 	  for(int idx=0;idx<words;idx++){
-	    pointers[si][idx] = dist(_generators[gdx]);
+	    fillScalar(pointer[idx],dist,_generators[gdx]);
 	  }
 
 	}
 	// merge into SIMD lanes
-	merge(l._odata[ss],pointers);
+	merge(l._odata[ss],buf);
       }
     };
 
diff --git a/lib/math/Grid_math_tensors.h b/lib/math/Grid_math_tensors.h
index cf5a5293..b08d0c83 100644
--- a/lib/math/Grid_math_tensors.h
+++ b/lib/math/Grid_math_tensors.h
@@ -57,12 +57,6 @@ public:
     friend void permute(iScalar<vtype> &out,const iScalar<vtype> &in,int permutetype){
       permute(out._internal,in._internal,permutetype);
     }
-    friend void extract(const iScalar<vtype> &in,std::vector<scalar_type *> &out){
-      extract(in._internal,out); // extract advances the pointers in out
-    }
-    friend void merge(iScalar<vtype> &in,std::vector<scalar_type *> &out){
-      merge(in._internal,out); // extract advances the pointers in out
-    }
 
     // Unary negation
     friend inline iScalar<vtype> operator -(const iScalar<vtype> &r) {
@@ -149,16 +143,6 @@ public:
 	permute(out._internal[i],in._internal[i],permutetype);
       }
     }
-    friend void extract(const iVector<vtype,N> &in,std::vector<scalar_type *> &out){
-      for(int i=0;i<N;i++){
-	extract(in._internal[i],out);// extract advances pointers in out
-      }
-    }
-    friend void merge(iVector<vtype,N> &in,std::vector<scalar_type *> &out){
-      for(int i=0;i<N;i++){
-	merge(in._internal[i],out);// extract advances pointers in out
-      }
-    }
     // Unary negation
     friend inline iVector<vtype,N> operator -(const iVector<vtype,N> &r) {
         iVector<vtype,N> ret;
@@ -232,18 +216,6 @@ public:
 	permute(out._internal[i][j],in._internal[i][j],permutetype);
     }}
   }
-  friend void extract(const iMatrix<vtype,N> &in,std::vector<scalar_type *> &out){
-    for(int i=0;i<N;i++){
-      for(int j=0;j<N;j++){
-	extract(in._internal[i][j],out);// extract advances pointers in out
-    }}
-  }
-  friend void merge(iMatrix<vtype,N> &in,std::vector<scalar_type *> &out){
-    for(int i=0;i<N;i++){
-      for(int j=0;j<N;j++){
-	merge(in._internal[i][j],out);// extract advances pointers in out
-    }}
-  }
   // Unary negation
   friend inline iMatrix<vtype,N> operator -(const iMatrix<vtype,N> &r) {
     iMatrix<vtype,N> ret;
@@ -285,37 +257,6 @@ public:
 };
 
 
-template<class vobj> inline 
-void extract(const vobj &vec,std::vector<typename vobj::scalar_object> &extracted)
-{
-  typedef typename vobj::scalar_type scalar_type ;
-  typedef typename vobj::vector_type vector_type ;
-
-  int Nsimd=vobj::vector_type::Nsimd();
-  
-  extracted.resize(Nsimd);
-
-  std::vector<scalar_type *> pointers(Nsimd);
-  for(int i=0;i<Nsimd;i++) 
-    pointers[i] =(scalar_type *)& extracted[i];
-  
-  extract(vec,pointers);
-}
-template<class vobj> inline 
-void merge(vobj &vec,std::vector<typename vobj::scalar_object> &extracted)
-{
-  typedef typename vobj::scalar_type scalar_type ;
-  typedef typename vobj::vector_type vector_type ;
-  
-  int Nsimd=vobj::vector_type::Nsimd();
-  assert(extracted.size()==Nsimd);
-
-  std::vector<scalar_type *> pointers(Nsimd);
-  for(int i=0;i<Nsimd;i++) 
-    pointers[i] =(scalar_type *)& extracted[i];
-  
-  merge(vec,pointers);
-}
 
 }
 #endif
diff --git a/lib/math/Grid_math_traits.h b/lib/math/Grid_math_traits.h
index 353cdd88..c47dca10 100644
--- a/lib/math/Grid_math_traits.h
+++ b/lib/math/Grid_math_traits.h
@@ -64,6 +64,14 @@ namespace Grid {
     typedef ComplexD scalar_object;
     enum { TensorLevel = 0 };
   };
+  template<> class GridTypeMapper<Integer> {
+  public:
+    typedef Integer scalar_type;
+    typedef Integer vector_type;
+    typedef Integer tensor_reduced;
+    typedef Integer scalar_object;
+    enum { TensorLevel = 0 };
+  };
 
   template<> class GridTypeMapper<vRealF> {
   public:
@@ -99,10 +107,10 @@ namespace Grid {
   };
   template<> class GridTypeMapper<vInteger> {
   public:
-    typedef Integer  scalar_type;
+    typedef  Integer scalar_type;
     typedef vInteger vector_type;
     typedef vInteger tensor_reduced;
-    typedef Integer  scalar_object;
+    typedef  Integer scalar_object;
     enum { TensorLevel = 0 };
   };
 
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index b1d6d525..1e2741b4 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -99,106 +99,159 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
 
   for(int ss=0;ss<grid->oSites();ss++){
 
-    int offset,local;
+    int offset,local,perm, ptype;
 
     vSpinColourVector result;
     vHalfSpinColourVector  chi;    
+    vHalfSpinColourVector  tmp;    
     vHalfSpinColourVector Uchi;
     vHalfSpinColourVector *chi_p;
 
     result=zero;
 
-#if 0
     // Xp
     offset = Stencil._offsets [Xp][ss];
     local  = Stencil._is_local[Xp][ss];
+    perm   = Stencil._permute[Xp][ss];
+    ptype  = Stencil._permute_type[Xp];
     chi_p  = &comm_buf[offset];
     if ( local ) {
-      spProjXp(chi,in._odata[offset]);
       chi_p = &chi;
-    } 
+      spProjXp(chi,in._odata[offset]);
+      if ( perm ) {
+	permute(tmp,chi,ptype);
+	chi_p = &tmp;
+      }
+    }
     mult(&(Uchi()),&(Umu._odata[ss](Xp)),&(*chi_p)());
     spReconXp(result,Uchi);
 
     // Yp
     offset = Stencil._offsets [Yp][ss];
     local  = Stencil._is_local[Yp][ss];
+    perm   = Stencil._permute[Yp][ss];
+    ptype  = Stencil._permute_type[Yp];
     chi_p  = &comm_buf[offset];
     if ( local ) {
-      spProjYp(chi,in._odata[offset]);
       chi_p = &chi;
-    } 
+      spProjYp(chi,in._odata[offset]);
+      if ( perm ) {
+	permute(tmp,chi,ptype);
+	chi_p = &tmp;
+      }
+    }
     mult(&(Uchi()),&(Umu._odata[ss](Yp)),&(*chi_p)());
     accumReconYp(result,Uchi);
 
     // Zp
     offset = Stencil._offsets [Zp][ss];
     local  = Stencil._is_local[Zp][ss];
+    perm   = Stencil._permute[Zp][ss];
+    ptype  = Stencil._permute_type[Zp];
     chi_p  = &comm_buf[offset];
+
     if ( local ) {
-      spProjZp(chi,in._odata[offset]);
       chi_p = &chi;
-    } 
-    mult(&(Uchi()),&(Umu._odata[ss](Zp)),&(*chi_p)() );
+      spProjZp(chi,in._odata[offset]);
+      if ( perm ) {
+	permute(tmp,chi,ptype);
+	chi_p = &tmp;
+      }
+    }
+    mult(&(Uchi()),&(Umu._odata[ss](Zp)),&(*chi_p)());
     accumReconZp(result,Uchi);
 
     // Tp
     offset = Stencil._offsets [Tp][ss];
     local  = Stencil._is_local[Tp][ss];
+    perm   = Stencil._permute[Tp][ss];
+    ptype  = Stencil._permute_type[Tp];
     chi_p  = &comm_buf[offset];
+
     if ( local ) {
-      spProjTp(chi,in._odata[offset]);
       chi_p = &chi;
-    } 
+      spProjTp(chi,in._odata[offset]);
+      if ( perm ) {
+	permute(tmp,chi,ptype);
+	chi_p = &tmp;
+      }
+    }
     mult(&(Uchi()),&(Umu._odata[ss](Tp)),&(*chi_p)());
     accumReconTp(result,Uchi);
-#endif
 
     // Xm
     offset = Stencil._offsets [Xm][ss];
     local  = Stencil._is_local[Xm][ss];
+    perm   = Stencil._permute[Xm][ss];
+    ptype  = Stencil._permute_type[Xm];
     chi_p  = &comm_buf[offset];
     if ( local ) {
-      spProjXm(chi,in._odata[offset]);
       chi_p = &chi;
-    } 
+      spProjXm(chi,in._odata[offset]);
+      if ( perm ) {
+	permute(tmp,chi,ptype);
+	chi_p = &tmp;
+      }
+    }
+    std::cout<<"Xm for site  "<<ss<<" l "<<local<<" p "<<perm<<" chi "<<Reduce(TensorRemove(innerProduct(*chi_p,*chi_p)))<<std::endl;
     mult(&(Uchi()),&(Umu._odata[ss](Xm)),&(*chi_p)());
     accumReconXm(result,Uchi);
-#if 0
+
 
     // Ym
     offset = Stencil._offsets [Ym][ss];
     local  = Stencil._is_local[Ym][ss];
+    perm   = Stencil._permute[Ym][ss];
+    ptype  = Stencil._permute_type[Ym];
     chi_p  = &comm_buf[offset];
+
     if ( local ) {
-      spProjYm(chi,in._odata[offset]);
       chi_p = &chi;
-    } 
+      spProjYm(chi,in._odata[offset]);
+      if ( perm ) {
+	permute(tmp,chi,ptype);
+	chi_p = &tmp;
+      }
+    }
     mult(&(Uchi()),&(Umu._odata[ss](Ym)),&(*chi_p)());
     accumReconYm(result,Uchi);
 
     // Zm
     offset = Stencil._offsets [Zm][ss];
     local  = Stencil._is_local[Zm][ss];
+    perm   = Stencil._permute[Zm][ss];
+    ptype  = Stencil._permute_type[Zm];
     chi_p  = &comm_buf[offset];
+
     if ( local ) {
-      spProjZm(chi,in._odata[offset]);
       chi_p = &chi;
-    } 
+      spProjZm(chi,in._odata[offset]);
+      if ( perm ) {
+	permute(tmp,chi,ptype);
+	chi_p = &tmp;
+      }
+    }
     mult(&(Uchi()),&(Umu._odata[ss](Zm)),&(*chi_p)());
     accumReconZm(result,Uchi);
 
     // Tm
     offset = Stencil._offsets [Tm][ss];
     local  = Stencil._is_local[Tm][ss];
+    perm   = Stencil._permute[Tm][ss];
+    ptype  = Stencil._permute_type[Tm];
     chi_p  = &comm_buf[offset];
+
     if ( local ) {
-      spProjTm(chi,in._odata[offset]);
       chi_p = &chi;
-    } 
+      spProjTm(chi,in._odata[offset]);
+      if ( perm ) {
+	permute(tmp,chi,ptype);
+	chi_p = &tmp;
+      }
+    }
     mult(&(Uchi()),&(Umu._odata[ss](Tm)),&(*chi_p)());
     accumReconTm(result,Uchi);
-#endif
+
 
     out._odata[ss] = result;
   }
diff --git a/lib/simd/Grid_vComplexD.h b/lib/simd/Grid_vComplexD.h
index 76295fdf..337b82c7 100644
--- a/lib/simd/Grid_vComplexD.h
+++ b/lib/simd/Grid_vComplexD.h
@@ -163,6 +163,7 @@ namespace Grid {
 	// all subtypes; may not be a good assumption, but could
 	// add the vector width as a template param for BG/Q for example
 	////////////////////////////////////////////////////////////////////
+	/*
 	friend inline void permute(vComplexD &y,vComplexD b,int perm)
 	{
 	  Gpermute<vComplexD>(y,b,perm);
@@ -183,6 +184,7 @@ namespace Grid {
 	{
 	  Gextract<vComplexD,ComplexD>(y,extracted);
 	}
+	*/
 
         ///////////////////////
         // Splat
diff --git a/lib/simd/Grid_vComplexF.h b/lib/simd/Grid_vComplexF.h
index fb9c5a67..c42e4029 100644
--- a/lib/simd/Grid_vComplexF.h
+++ b/lib/simd/Grid_vComplexF.h
@@ -412,6 +412,7 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
       {
 	Gpermute<vComplexF>(y,b,perm);
       }
+      /*
       friend inline void merge(vComplexF &y,std::vector<ComplexF *> &extracted)
       {
 	Gmerge<vComplexF,ComplexF >(y,extracted);
@@ -428,7 +429,7 @@ friend inline void vstore(const vComplexF &ret, ComplexF *a){
       {
 	Gextract<vComplexF,ComplexF>(y,extracted);
       }
-
+      */
 
     };
 
diff --git a/lib/simd/Grid_vInteger.h b/lib/simd/Grid_vInteger.h
index 5dc77444..6035aea1 100644
--- a/lib/simd/Grid_vInteger.h
+++ b/lib/simd/Grid_vInteger.h
@@ -221,6 +221,7 @@ namespace Grid {
       {
 	Gpermute<vInteger>(y,b,perm);
       }
+      /*
       friend inline void merge(vInteger &y,std::vector<Integer *> &extracted)
       {
 	Gmerge<vInteger,Integer>(y,extracted);
@@ -237,7 +238,7 @@ namespace Grid {
       {
 	Gextract<vInteger,Integer>(y,extracted);
       }
-
+      */
 
     public:
         static inline int Nsimd(void) { return sizeof(ivec)/sizeof(Integer);}
diff --git a/lib/simd/Grid_vRealD.h b/lib/simd/Grid_vRealD.h
index f6d68963..36e189e1 100644
--- a/lib/simd/Grid_vRealD.h
+++ b/lib/simd/Grid_vRealD.h
@@ -105,6 +105,7 @@ namespace Grid {
 	// all subtypes; may not be a good assumption, but could
 	// add the vector width as a template param for BG/Q for example
 	////////////////////////////////////////////////////////////////////
+	/*
 	friend inline void permute(vRealD &y,vRealD b,int perm)
 	{
 	  Gpermute<vRealD>(y,b,perm);
@@ -125,7 +126,7 @@ namespace Grid {
 	{
 	  Gextract<vRealD,RealD>(y,extracted);
 	}
-
+	*/
         
         friend inline void vsplat(vRealD &ret,double a){
 #if defined (AVX1)|| defined (AVX2)
diff --git a/lib/simd/Grid_vRealF.h b/lib/simd/Grid_vRealF.h
index 86895492..70f76bc0 100644
--- a/lib/simd/Grid_vRealF.h
+++ b/lib/simd/Grid_vRealF.h
@@ -127,6 +127,7 @@ namespace Grid {
 	// all subtypes; may not be a good assumption, but could
 	// add the vector width as a template param for BG/Q for example
 	////////////////////////////////////////////////////////////////////
+	/*
 	friend inline void permute(vRealF &y,vRealF b,int perm)
 	{
 	  Gpermute<vRealF>(y,b,perm);
@@ -147,7 +148,7 @@ namespace Grid {
 	{
 	  Gextract<vRealF,RealF>(y,extracted);
 	}
-
+	*/
 
         
         /////////////////////////////////////////////////////
diff --git a/lib/stencil/Grid_stencil_common.cc b/lib/stencil/Grid_stencil_common.cc
index c245aafb..fda4ef2e 100644
--- a/lib/stencil/Grid_stencil_common.cc
+++ b/lib/stencil/Grid_stencil_common.cc
@@ -118,6 +118,7 @@ namespace Grid {
       
       int fd              = _grid->_fdimensions[dimension];
       int rd              = _grid->_rdimensions[dimension];
+      int pd              = _grid->_processors[dimension];
       int simd_layout     = _grid->_simd_layout[dimension];
       int comm_dim        = _grid->_processors[dimension] >1 ;
       
@@ -136,10 +137,10 @@ namespace Grid {
       
       for(int x=0;x<rd;x++){       
 	
-	int offnode = ( x+sshift >= rd );
+	int comm_proc = ((x+sshift)/rd)%pd;
+	int offnode =   (comm_proc!=0);
 	int sx        = (x+sshift)%rd;
-	int comm_proc = (x+sshift)/rd;
-	  
+
 	if (!offnode) {
 	  
 	  int permute_slice=0;
diff --git a/tests/Grid_simd.cc b/tests/Grid_simd.cc
index 364b13e8..1de2ad1f 100644
--- a/tests/Grid_simd.cc
+++ b/tests/Grid_simd.cc
@@ -75,9 +75,9 @@ void Tester(const functor &func)
     random(sRNG,result[i]);
   }
 
-  Gmerge(v_input1,input1);
-  Gmerge(v_input2,input2);
-  Gmerge(v_result,result);
+  merge<vec,scal>(v_input1,input1);
+  merge<vec,scal>(v_input2,input2);
+  merge<vec,scal>(v_result,result);
 
   func(v_result,v_input1,v_input2);
 
@@ -85,7 +85,7 @@ void Tester(const functor &func)
     func(reference[i],input1[i],input2[i]);
   }
 
-  Gextract(v_result,result);
+  extract<vec,scal>(v_result,result);
   std::cout << " " << func.name()<<std::endl;
 
   int ok=0;
diff --git a/tests/Grid_stencil.cc b/tests/Grid_stencil.cc
index c06c0d16..eb661487 100644
--- a/tests/Grid_stencil.cc
+++ b/tests/Grid_stencil.cc
@@ -4,21 +4,12 @@ using namespace std;
 using namespace Grid;
 using namespace Grid::QCD;
 
-template<class vobj>
-class SimpleCompressor {
-public:
-  void Point(int) {};
-
-  vobj operator() (const vobj &arg) {
-    return arg;
-  }
-};
 int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
   std::vector<int> simd_layout({1,1,2,2});
-  std::vector<int> mpi_layout ({2,2,2,2});
+  std::vector<int> mpi_layout ({2,2,1,2});
   std::vector<int> latt_size  ({8,8,8,8});
 
   double volume = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
@@ -26,7 +17,9 @@ int main (int argc, char ** argv)
   GridCartesian Fine(latt_size,simd_layout,mpi_layout);
   GridRedBlackCartesian rbFine(latt_size,simd_layout,mpi_layout);
   GridParallelRNG       fRNG(&Fine);
-  fRNG.SeedRandomDevice();
+  //  fRNG.SeedRandomDevice();
+  std::vector<int> seeds({1,2,3,4});
+  fRNG.SeedFixedIntegers(seeds);
   
   LatticeColourMatrix Foo(&Fine);
   LatticeColourMatrix Bar(&Fine);
@@ -38,8 +31,9 @@ int main (int argc, char ** argv)
 
 
     for(int dir=0;dir<4;dir++){
-      for(int disp=0;disp<Fine._rdimensions[dir];disp++){
+      for(int disp=0;disp<Fine._fdimensions[dir];disp++){
 
+	std::cout << "Using stencil to shift dim "<<dir<< " by "<<disp<<std::endl;
 	// start to test the Cartesian npoint stencil infrastructure
 	int npoint=1;
 	std::vector<int> directions(npoint,dir);
@@ -47,22 +41,13 @@ int main (int argc, char ** argv)
 
 	CartesianStencil myStencil(&Fine,npoint,0,directions,displacements);
 
-	printf("STENCIL: osites %d %d dir %d disp %d\n",Fine.oSites(),(int)myStencil._offsets[0].size(),dir,disp);
 	std::vector<int> ocoor(4);
 	for(int o=0;o<Fine.oSites();o++){
 	  Fine.oCoorFromOindex(ocoor,o);
 	  ocoor[dir]=(ocoor[dir]+disp)%Fine._rdimensions[dir];
-	  int nbr = Fine.oIndexReduced(ocoor);
-	  int stcl= myStencil._offsets[0][o];
-	  if(nbr!=stcl){
-	    printf("STENCIL: nbr %d stencil._offset %d\n",nbr,stcl);
-	  }
 	}
 	
-	printf("allocating %d buffers\n",myStencil._unified_buffer_size);
-	fflush(stdout);
 	std::vector<vColourMatrix,alignedAllocator<vColourMatrix> >  comm_buf(myStencil._unified_buffer_size);
-	printf("calling halo exchange\n");fflush(stdout);
 	SimpleCompressor<vColourMatrix> compress;
 	myStencil.HaloExchange(Foo,comm_buf,compress);
 
@@ -81,14 +66,12 @@ int main (int argc, char ** argv)
 	    Check._odata[i] = Foo._odata[offset];
 	  else 
 	    Check._odata[i] = comm_buf[offset];
-	  
-
 	}
 
 	Real nrmC = norm2(Check);
 	Real nrmB = norm2(Bar);
 	Real nrm  = norm2(Check-Bar);
-	printf("N2diff = %le (%le, %le) \n",nrm,nrmC,nrmB);fflush(stdout);
+	std::cout<<"N2diff ="<<nrm<<" "<<nrmC<<" " <<nrmB<<std::endl;
 
 	Real snrmC =0;
 	Real snrmB =0;
@@ -110,10 +93,11 @@ int main (int argc, char ** argv)
             diff =check()()(r,c)-bar()()(r,c);
             double nn=real(conj(diff)*diff);
             if ( nn > 0){
-	      printf("Coor (%d %d %d %d) \t rc %d%d \t %le %le %le\n",
+	      printf("Coor (%d %d %d %d) \t rc %d%d \t %le (%le,%le) %le\n",
 		     coor[0],coor[1],coor[2],coor[3],r,c,
 		     nn,
 		     real(check()()(r,c)),
+		     imag(check()()(r,c)),
 		     real(bar()()(r,c))
 		     );
 	    }
@@ -124,7 +108,7 @@ int main (int argc, char ** argv)
 	 
 	}}}}
 
-	printf("scalar N2diff = %le (%le, %le) \n",snrm,snrmC,snrmB);fflush(stdout);
+	std::cout<<"scalar N2diff = "<<snrm<<" " <<snrmC<<" "<<snrmB<<std::endl;
 
 
       }
diff --git a/tests/Grid_wilson.cc b/tests/Grid_wilson.cc
index f6e86f5d..b510e9d2 100644
--- a/tests/Grid_wilson.cc
+++ b/tests/Grid_wilson.cc
@@ -28,6 +28,7 @@ int main (int argc, char ** argv)
   std::vector<int> seeds({1,2,3,4});
 
   GridParallelRNG          pRNG(&Grid);
+  //  std::vector<int> seeds({1,2,3,4});
   //  pRNG.SeedFixedIntegers(seeds);
   pRNG.SeedRandomDevice();
 
@@ -44,7 +45,7 @@ int main (int argc, char ** argv)
     U[mu] = peekIndex<3>(Umu,mu);
   }
   
-  std::vector<int> mask({0,0,0,0,1,0,0,0});
+  std::vector<int> mask({1,1,1,1,1,1,1,1});
   { // Naive wilson implementation
     ref = zero;
     for(int mu=0;mu<Nd;mu++){

From dcc23faa4a25c5d5b353461dd9028e8127fbf663 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 29 Apr 2015 06:23:56 +0100
Subject: [PATCH 109/429] Fixed the stencil sector and Wilson now agrees
 between stencil based implementation and the cshift based implementation.
 Managed to reduce the volume of code in this sector a little, but
 consolidation would be good, perhaps taking common logic out into simple
 helper functions

---
 lib/Grid_stencil.h                 | 40 ++++++------------------------
 lib/qcd/Grid_qcd_wilson_dop.cc     |  2 --
 lib/stencil/Grid_stencil_common.cc | 21 ++++++----------
 tests/Grid_wilson.cc               |  5 ++--
 4 files changed, 17 insertions(+), 51 deletions(-)

diff --git a/lib/Grid_stencil.h b/lib/Grid_stencil.h
index fdc8ece9..4d32575f 100644
--- a/lib/Grid_stencil.h
+++ b/lib/Grid_stencil.h
@@ -39,13 +39,6 @@
 
 namespace Grid {
 
-  struct CommsRequest { 
-    int words;
-    int unified_buffer_offset;
-    int tag;
-    int to_rank;
-    int from_rank;
-  } ;
 
 
   class CartesianStencil { // Stencil runs along coordinate axes only; NO diagonal fill in.
@@ -69,7 +62,6 @@ namespace Grid {
       int _unified_buffer_size;
       int _request_count;
 
-      std::vector<CommsRequest>  CommsRequests;
 
       CartesianStencil(GridBase *grid,
 		       int npoints,
@@ -90,7 +82,6 @@ namespace Grid {
       template<class vobj,class cobj, class compressor> void 
 	HaloExchange(const Lattice<vobj> &source,std::vector<cobj,alignedAllocator<cobj> > &u_comm_buf,compressor &compress)
       {
-	std::cout<< "HaloExchange comm_buf.size()="<< u_comm_buf.size()<<" unified_buffer_size"<< _unified_buffer_size<< std::endl;
 	// conformable(source._grid,_grid);
 	assert(source._grid==_grid);
 	if (u_comm_buf.size() != _unified_buffer_size ) u_comm_buf.resize(_unified_buffer_size);
@@ -141,7 +132,6 @@ namespace Grid {
 	    }
 	  }
 	}
-	std::cout<< "HaloExchange complete"<< std::endl;
       }
 
       template<class vobj,class cobj, class compressor> 
@@ -194,24 +184,18 @@ namespace Grid {
 	      _grid->ShiftedRanks(dimension,comm_proc,xmit_to_rank,recv_from_rank);
 	      assert (xmit_to_rank != _grid->ThisRank());
 	      assert (recv_from_rank != _grid->ThisRank());
+
 	      //      FIXME Implement asynchronous send & also avoid buffer copy
 	      _grid->SendToRecvFrom((void *)&send_buf[0],
 				   xmit_to_rank,
 				   (void *)&recv_buf[0],
 				   recv_from_rank,
 				   bytes);
-	      printf("GatherStartComms communicated offnode x %d\n",x);fflush(stdout);
 
-	      printf("GatherStartComms inserting %le to u_comm_offset %d buf size %d for dim %d shift %d\n",
-		     *( (RealF *) &recv_buf[0]),
-		     u_comm_offset,buffer_size,
-		     dimension,shift
-		     ); fflush(stdout);
 	      for(int i=0;i<buffer_size;i++){
 		u_comm_buf[u_comm_offset+i]=recv_buf[i];
 	      }
 	      u_comm_offset+=buffer_size;
-	      printf("GatherStartComms inserted x %d\n",x);fflush(stdout);
 	    }
 	  }
 	}
@@ -248,7 +232,7 @@ namespace Grid {
 	  int buffer_size = _grid->_slice_nblock[dimension]*_grid->_slice_block[dimension];
 	  int words = sizeof(cobj)/sizeof(vector_type);
 
-	  /*   FIXME ALTERNATE BUFFER DETERMINATION */
+	  /* FIXME ALTERNATE BUFFER DETERMINATION ; possibly slow to allocate*/
 	  std::vector<std::vector<scalar_object> > send_buf_extract(Nsimd,std::vector<scalar_object>(buffer_size) ); 
 	  std::vector<std::vector<scalar_object> > recv_buf_extract(Nsimd,std::vector<scalar_object>(buffer_size) );
 	  int bytes = buffer_size*sizeof(scalar_object);
@@ -267,25 +251,21 @@ namespace Grid {
 	  for(int x=0;x<rd;x++){       
 
 	    int any_offnode = ( ((x+sshift)%fd) >= rd );
-	    std::cout<<"any_offnode ="<<any_offnode<<std::endl;
+
 	    if ( any_offnode ) {
-	      // FIXME call local permute copy if none are offnode.
+
 	      for(int i=0;i<Nsimd;i++){       
 		pointers[i] = &send_buf_extract[i][0];
 	      }
 	      int sx   = (x+sshift)%rd;
 	      
-	      std::cout<< "Gathering "<< x <<std::endl;
 	      Gather_plane_extract<cobj>(rhs,pointers,dimension,sx,cbmask,compress);
-	      std::cout<< "Gathered "<<std::endl;
+
 	      for(int i=0;i<Nsimd;i++){
 		
-		std::vector<int> icoor;
-		_grid->iCoorFromIindex(icoor,i);
 
 		int inner_bit = (Nsimd>>(permute_type+1));
 		int ic= (i&inner_bit)? 1:0;
-		assert(ic==icoor[dimension]);
 
 		int my_coor          = rd*ic + x;
 		int nbr_coor         = my_coor+sshift;
@@ -301,12 +281,9 @@ namespace Grid {
 		if (nbr_ic) nbr_lane|=inner_bit;
 		assert (sx == nbr_ox);
 
-   std::cout<<"nbr_proc "<<nbr_proc<< " x "<<x<<" nbr_x "<<nbr_ox << " lane "<<i << " nbr_lane "<<nbr_lane
-	    << " nbr_ic "<<nbr_ic  << " mycoor "<< my_coor<< " nbr_coor "<<nbr_coor<<std::endl;
 		
 		if(nbr_proc){
 		  
-		  std::cout<< "MPI sending "<<std::endl;
 		  _grid->ShiftedRanks(dimension,nbr_proc,xmit_to_rank,recv_from_rank); 
 		  
 		  _grid->SendToRecvFrom((void *)&send_buf_extract[nbr_lane][0],
@@ -314,23 +291,20 @@ namespace Grid {
 					(void *)&recv_buf_extract[i][0],
 					recv_from_rank,
 					bytes);
-		  std::cout<< "MPI complete "<<std::endl;
 		  
 		  rpointers[i] = &recv_buf_extract[i][0];
-		  std::cout<<"lane "<<i<<" data "<<*( (Real *) rpointers[i])<<std::endl;
+
 		} else { 
 		  rpointers[i] = &send_buf_extract[nbr_lane][0];
-		  std::cout<<"lane "<<i<<" data "<<*( (Real *) rpointers[i])<<std::endl;
 		}
 	      }
 
 	      // Here we don't want to scatter, just place into a buffer.
-	      std::cout<< "merging u_comm_offset "<< u_comm_offset<<" comm_buf_size" << u_comm_buf.size() <<std::endl;
-
 	      for(int i=0;i<buffer_size;i++){
 		assert(u_comm_offset+i<_unified_buffer_size);
 		merge(u_comm_buf[u_comm_offset+i],rpointers,i);
 	      }
+
 	      u_comm_offset+=buffer_size;
 	    }
 	  }
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index 1e2741b4..efe72ca5 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -24,7 +24,6 @@ const int WilsonMatrix::Tm = 7;
     
     void Point(int p) { 
       mu=p;
-      std::cout << "WilsonCompressor.Point " << mu<<std::endl;
     };
 
     vHalfSpinColourVector operator () (const vSpinColourVector &in)
@@ -193,7 +192,6 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
 	chi_p = &tmp;
       }
     }
-    std::cout<<"Xm for site  "<<ss<<" l "<<local<<" p "<<perm<<" chi "<<Reduce(TensorRemove(innerProduct(*chi_p,*chi_p)))<<std::endl;
     mult(&(Uchi()),&(Umu._odata[ss](Xm)),&(*chi_p)());
     accumReconXm(result,Uchi);
 
diff --git a/lib/stencil/Grid_stencil_common.cc b/lib/stencil/Grid_stencil_common.cc
index fda4ef2e..52daf466 100644
--- a/lib/stencil/Grid_stencil_common.cc
+++ b/lib/stencil/Grid_stencil_common.cc
@@ -19,7 +19,6 @@ namespace Grid {
       _distances  = distances;
       _unified_buffer_size=0;
       _request_count =0;
-      CommsRequests.resize(0);
 
       int osites  = _grid->oSites();
 
@@ -117,6 +116,7 @@ namespace Grid {
       GridBase *grid=_grid;
       
       int fd              = _grid->_fdimensions[dimension];
+      int ld              = _grid->_ldimensions[dimension];
       int rd              = _grid->_rdimensions[dimension];
       int pd              = _grid->_processors[dimension];
       int simd_layout     = _grid->_simd_layout[dimension];
@@ -137,9 +137,10 @@ namespace Grid {
       
       for(int x=0;x<rd;x++){       
 	
-	int comm_proc = ((x+sshift)/rd)%pd;
-	int offnode =   (comm_proc!=0);
-	int sx        = (x+sshift)%rd;
+	int offnode = (((x+sshift)%fd) >= rd ); 
+	//	int comm_proc   = ((x+sshift)/ld)%pd;        
+	//	int offnode     = (comm_proc!=0);
+	int sx          = (x+sshift)%rd;
 
 	if (!offnode) {
 	  
@@ -157,17 +158,9 @@ namespace Grid {
 	  int recv_from_rank;
 	  int xmit_to_rank;
 
-	  CommsRequest cr;
-
-	  cr.tag                   = _request_count++;
-	  cr.words                 = words;
-	  cr.unified_buffer_offset = _unified_buffer_size;
+	  int unified_buffer_offset = _unified_buffer_size;
 	  _unified_buffer_size    += words;
-	  grid->ShiftedRanks(dimension,comm_proc,cr.to_rank,cr.from_rank);
-
-	  CommsRequests.push_back(cr);
-
-	  ScatterPlane(point,dimension,x,cbmask,cr.unified_buffer_offset); // permute/extract/merge is done in comms phase
+	  ScatterPlane(point,dimension,x,cbmask,unified_buffer_offset); // permute/extract/merge is done in comms phase
 	  
 	}
       }
diff --git a/tests/Grid_wilson.cc b/tests/Grid_wilson.cc
index b510e9d2..4f926689 100644
--- a/tests/Grid_wilson.cc
+++ b/tests/Grid_wilson.cc
@@ -21,7 +21,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> simd_layout({1,1,2,2});
-  std::vector<int> mpi_layout ({2,1,1,2});
+  std::vector<int> mpi_layout ({2,2,2,2});
   std::vector<int> latt_size  ({8,8,8,8});
     
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
@@ -76,7 +76,8 @@ int main (int argc, char ** argv)
   std::cout << "norm result "<< norm2(result)<<std::endl;
   std::cout << "norm ref    "<< norm2(ref)<<std::endl;
 
-  for(int ss=0;ss<10;ss++ ){
+  //  for(int ss=0;ss<10;ss++ ){
+  for(int ss=0;ss<0;ss++ ){
     for(int i=0;i<Ns;i++){
       for(int j=0;j<Nc;j++){
 	ComplexF * ref_p = (ComplexF *)&ref._odata[ss]()(i)(j);

From d8ffa09e3b22efa4be9dc08b3f5ec7c15ac24d40 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 29 Apr 2015 06:50:18 +0100
Subject: [PATCH 110/429] Benchmark wilson dhop now; 14.6GF on one core, not as
 fast as SU(3)xSU(3) [23GF] but still not too shabby. Disassembling output
 shows ugly sequences in the permute sector. Could comparatively benchmark
 with and without the if-else structure to see how much I'm losing.

Drops to 9GF as it falls out of cache. Moving to Lebesgue ordering should help there. Substantive progress.
---
 tests/Grid_wilson.cc | 21 ++++++++++++++++++---
 1 file changed, 18 insertions(+), 3 deletions(-)

diff --git a/tests/Grid_wilson.cc b/tests/Grid_wilson.cc
index 4f926689..ff1a9010 100644
--- a/tests/Grid_wilson.cc
+++ b/tests/Grid_wilson.cc
@@ -21,8 +21,8 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> simd_layout({1,1,2,2});
-  std::vector<int> mpi_layout ({2,2,2,2});
-  std::vector<int> latt_size  ({8,8,8,8});
+  std::vector<int> mpi_layout ({1,1,1,1});
+  std::vector<int> latt_size  ({4,4,8,8});
     
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
   std::vector<int> seeds({1,2,3,4});
@@ -39,6 +39,11 @@ int main (int argc, char ** argv)
   LatticeGaugeField Umu(&Grid); random(pRNG,Umu);
   std::vector<LatticeColourMatrix> U(4,&Grid);
 
+  double volume=1;
+  for(int mu=0;mu<Nd;mu++){
+    volume=volume*latt_size[mu];
+  }  
+
   for(int mu=0;mu<Nd;mu++){
     //    U[mu] = 1.0;
     //    pokeIndex<3>(Umu,U[mu],mu);
@@ -70,11 +75,21 @@ int main (int argc, char ** argv)
 
   RealD mass=0.1;
   WilsonMatrix Dw(Umu,mass);
+  
   std::cout << "Calling Dw"<<std::endl;
-  Dw.multiply(src,result);
+  int ncall=100;
+  double t0=usecond();
+  for(int i=0;i<ncall;i++){
+    Dw.multiply(src,result);
+  }
+  double t1=usecond();
+  double flops=1320*volume*ncall;
+
+  
   std::cout << "Called Dw"<<std::endl;
   std::cout << "norm result "<< norm2(result)<<std::endl;
   std::cout << "norm ref    "<< norm2(ref)<<std::endl;
+  std::cout << "mflop/s = "<< flops/(t1-t0)<<std::endl;
 
   //  for(int ss=0;ss<10;ss++ ){
   for(int ss=0;ss<0;ss++ ){

From 24fc71b2e943b9e3af4dbe74162923775b02eb83 Mon Sep 17 00:00:00 2001
From: mspraggs <matthew.spraggs@gmail.com>
Date: Wed, 29 Apr 2015 23:44:03 +0100
Subject: [PATCH 111/429] Added <map> include to GridNerscIO.h

Adding this allows clang to compile Grid to completion.
---
 lib/parallelIO/GridNerscIO.h | 1 +
 1 file changed, 1 insertion(+)

diff --git a/lib/parallelIO/GridNerscIO.h b/lib/parallelIO/GridNerscIO.h
index 5d1e2e41..4dfa7979 100644
--- a/lib/parallelIO/GridNerscIO.h
+++ b/lib/parallelIO/GridNerscIO.h
@@ -5,6 +5,7 @@
 #include <iostream>
 #include <iomanip>
 #include <fstream>
+#include <map>
 
 #ifdef HAVE_ENDIAN_H
 #include <endian.h>

From c0ead9479181f04cf9825d79ce143b2dfc3eb151 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 30 Apr 2015 16:39:06 +0100
Subject: [PATCH 112/429] Integrated Lebesgue code and been playing with
 alternate implementations of the wilson dop without any particular success in
 increasing the performance.

---
 lib/Grid_stencil.h                 |  18 ++++
 lib/qcd/Grid_qcd_dirac.cc          |   6 +-
 lib/qcd/Grid_qcd_wilson_dop.cc     | 150 +++++++++++++++--------------
 lib/qcd/Grid_qcd_wilson_dop.h      |   2 +
 lib/stencil/Grid_stencil_common.cc |  92 ++++++++++++++++++
 tests/Grid_wilson.cc               |   6 +-
 6 files changed, 196 insertions(+), 78 deletions(-)

diff --git a/lib/Grid_stencil.h b/lib/Grid_stencil.h
index 4d32575f..fa537f65 100644
--- a/lib/Grid_stencil.h
+++ b/lib/Grid_stencil.h
@@ -44,6 +44,24 @@ namespace Grid {
   class CartesianStencil { // Stencil runs along coordinate axes only; NO diagonal fill in.
   public:
 
+      typedef uint32_t StencilInteger;
+      
+
+
+      StencilInteger alignup(StencilInteger n){
+	n--;           // 1000 0011 --> 1000 0010
+	n |= n >> 1;   // 1000 0010 | 0100 0001 = 1100 0011
+	n |= n >> 2;   // 1100 0011 | 0011 0000 = 1111 0011
+	n |= n >> 4;   // 1111 0011 | 0000 1111 = 1111 1111
+	n |= n >> 8;   // ... (At this point all bits are 1, so further bitwise-or
+	n |= n >> 16;  //      operations produce no effect.)
+	n++;           // 1111 1111 --> 1 0000 0000
+	return n;
+      };
+      void LebesgueOrder(void);
+
+      std::vector<StencilInteger> _LebesgueReorder;
+
       int                               _checkerboard;
       int                               _npoints; // Move to template param?
       GridBase *                        _grid;
diff --git a/lib/qcd/Grid_qcd_dirac.cc b/lib/qcd/Grid_qcd_dirac.cc
index 6526dfd3..9514a2a6 100644
--- a/lib/qcd/Grid_qcd_dirac.cc
+++ b/lib/qcd/Grid_qcd_dirac.cc
@@ -34,6 +34,10 @@ namespace Grid {
       "         "
     };
 
-
+    //    void sprojMul( vHalfSpinColourVector &out,vColourMatrix &u, vSpinColourVector &in){
+    //      vHalfSpinColourVector hspin;
+    //      spProjXp(hspin,in);
+    //      mult(&out,&u,&hspin);
+    //    }
   }
 }
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index efe72ca5..7ce41e57 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -96,17 +96,17 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
   WilsonCompressor compressor;
   Stencil.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
 
-  for(int ss=0;ss<grid->oSites();ss++){
+  vHalfSpinColourVector  tmp;    
+  vHalfSpinColourVector  chi;    
+  vSpinColourVector result;
+  vHalfSpinColourVector Uchi;
+  vHalfSpinColourVector *chi_p;
+  int offset,local,perm, ptype;
 
-    int offset,local,perm, ptype;
+  for(int sss=0;sss<grid->oSites();sss++){
 
-    vSpinColourVector result;
-    vHalfSpinColourVector  chi;    
-    vHalfSpinColourVector  tmp;    
-    vHalfSpinColourVector Uchi;
-    vHalfSpinColourVector *chi_p;
-
-    result=zero;
+    int ss = sss;
+	//int ss = Stencil._LebesgueReorder[sss];
 
     // Xp
     offset = Stencil._offsets [Xp][ss];
@@ -114,15 +114,16 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     perm   = Stencil._permute[Xp][ss];
     ptype  = Stencil._permute_type[Xp];
     chi_p  = &comm_buf[offset];
-    if ( local ) {
-      chi_p = &chi;
+    if ( local && perm ) 
+    {
+      spProjXp(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
       spProjXp(chi,in._odata[offset]);
-      if ( perm ) {
-	permute(tmp,chi,ptype);
-	chi_p = &tmp;
-      }
+    } else { 
+      chi=comm_buf[offset];
     }
-    mult(&(Uchi()),&(Umu._odata[ss](Xp)),&(*chi_p)());
+    mult(&Uchi(),&Umu._odata[ss](Xp),&chi());
     spReconXp(result,Uchi);
 
     // Yp
@@ -130,16 +131,17 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     local  = Stencil._is_local[Yp][ss];
     perm   = Stencil._permute[Yp][ss];
     ptype  = Stencil._permute_type[Yp];
-    chi_p  = &comm_buf[offset];
-    if ( local ) {
-      chi_p = &chi;
+
+    if ( local && perm ) 
+    {
+      spProjYp(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
       spProjYp(chi,in._odata[offset]);
-      if ( perm ) {
-	permute(tmp,chi,ptype);
-	chi_p = &tmp;
-      }
+    } else { 
+      chi=comm_buf[offset];
     }
-    mult(&(Uchi()),&(Umu._odata[ss](Yp)),&(*chi_p)());
+    mult(&Uchi(),&Umu._odata[ss](Yp),&chi());
     accumReconYp(result,Uchi);
 
     // Zp
@@ -147,17 +149,17 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     local  = Stencil._is_local[Zp][ss];
     perm   = Stencil._permute[Zp][ss];
     ptype  = Stencil._permute_type[Zp];
-    chi_p  = &comm_buf[offset];
 
-    if ( local ) {
-      chi_p = &chi;
+    if ( local && perm ) 
+    {
+      spProjZp(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
       spProjZp(chi,in._odata[offset]);
-      if ( perm ) {
-	permute(tmp,chi,ptype);
-	chi_p = &tmp;
-      }
+    } else { 
+      chi=comm_buf[offset];
     }
-    mult(&(Uchi()),&(Umu._odata[ss](Zp)),&(*chi_p)());
+    mult(&Uchi(),&Umu._odata[ss](Zp),&chi());
     accumReconZp(result,Uchi);
 
     // Tp
@@ -165,17 +167,17 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     local  = Stencil._is_local[Tp][ss];
     perm   = Stencil._permute[Tp][ss];
     ptype  = Stencil._permute_type[Tp];
-    chi_p  = &comm_buf[offset];
 
-    if ( local ) {
-      chi_p = &chi;
+    if ( local && perm ) 
+    {
+      spProjTp(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
       spProjTp(chi,in._odata[offset]);
-      if ( perm ) {
-	permute(tmp,chi,ptype);
-	chi_p = &tmp;
-      }
+    } else { 
+      chi=comm_buf[offset];
     }
-    mult(&(Uchi()),&(Umu._odata[ss](Tp)),&(*chi_p)());
+    mult(&Uchi(),&Umu._odata[ss](Tp),&chi());
     accumReconTp(result,Uchi);
 
     // Xm
@@ -183,16 +185,17 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     local  = Stencil._is_local[Xm][ss];
     perm   = Stencil._permute[Xm][ss];
     ptype  = Stencil._permute_type[Xm];
-    chi_p  = &comm_buf[offset];
-    if ( local ) {
-      chi_p = &chi;
+
+    if ( local && perm ) 
+    {
+      spProjXm(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
       spProjXm(chi,in._odata[offset]);
-      if ( perm ) {
-	permute(tmp,chi,ptype);
-	chi_p = &tmp;
-      }
+    } else { 
+      chi=comm_buf[offset];
     }
-    mult(&(Uchi()),&(Umu._odata[ss](Xm)),&(*chi_p)());
+    mult(&Uchi(),&Umu._odata[ss](Xm),&chi());
     accumReconXm(result,Uchi);
 
 
@@ -201,17 +204,17 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     local  = Stencil._is_local[Ym][ss];
     perm   = Stencil._permute[Ym][ss];
     ptype  = Stencil._permute_type[Ym];
-    chi_p  = &comm_buf[offset];
 
-    if ( local ) {
-      chi_p = &chi;
+    if ( local && perm ) 
+    {
+      spProjYm(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
       spProjYm(chi,in._odata[offset]);
-      if ( perm ) {
-	permute(tmp,chi,ptype);
-	chi_p = &tmp;
-      }
+    } else { 
+      chi=comm_buf[offset];
     }
-    mult(&(Uchi()),&(Umu._odata[ss](Ym)),&(*chi_p)());
+    mult(&Uchi(),&Umu._odata[ss](Ym),&chi());
     accumReconYm(result,Uchi);
 
     // Zm
@@ -219,17 +222,17 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     local  = Stencil._is_local[Zm][ss];
     perm   = Stencil._permute[Zm][ss];
     ptype  = Stencil._permute_type[Zm];
-    chi_p  = &comm_buf[offset];
 
-    if ( local ) {
-      chi_p = &chi;
+    if ( local && perm ) 
+    {
+      spProjZm(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
       spProjZm(chi,in._odata[offset]);
-      if ( perm ) {
-	permute(tmp,chi,ptype);
-	chi_p = &tmp;
-      }
+    } else { 
+      chi=comm_buf[offset];
     }
-    mult(&(Uchi()),&(Umu._odata[ss](Zm)),&(*chi_p)());
+    mult(&Uchi(),&Umu._odata[ss](Zm),&chi());
     accumReconZm(result,Uchi);
 
     // Tm
@@ -237,24 +240,25 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     local  = Stencil._is_local[Tm][ss];
     perm   = Stencil._permute[Tm][ss];
     ptype  = Stencil._permute_type[Tm];
-    chi_p  = &comm_buf[offset];
 
-    if ( local ) {
-      chi_p = &chi;
+    if ( local && perm ) 
+    {
+      spProjTm(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
       spProjTm(chi,in._odata[offset]);
-      if ( perm ) {
-	permute(tmp,chi,ptype);
-	chi_p = &tmp;
-      }
+    } else { 
+      chi=comm_buf[offset];
     }
-    mult(&(Uchi()),&(Umu._odata[ss](Tm)),&(*chi_p)());
+    mult(&Uchi(),&Umu._odata[ss](Tm),&chi());
     accumReconTm(result,Uchi);
 
-
     out._odata[ss] = result;
   }
 
 }
+
+
 void WilsonMatrix::Dw(const LatticeFermion &in, LatticeFermion &out)
 {
   return;
diff --git a/lib/qcd/Grid_qcd_wilson_dop.h b/lib/qcd/Grid_qcd_wilson_dop.h
index dc3c80b9..900f1801 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.h
+++ b/lib/qcd/Grid_qcd_wilson_dop.h
@@ -48,6 +48,8 @@ namespace Grid {
       // m+4r -1/2 Dhop; both cb's
       void Dw(const LatticeFermion &in, LatticeFermion &out);
 
+      typedef iScalar<iMatrix<vComplex, Nc> > matrix;
+
       // half checkerboard operaions
       void MpcDag   (const LatticeFermion &in, LatticeFermion &out);
       void Mpc      (const LatticeFermion &in, LatticeFermion &out);
diff --git a/lib/stencil/Grid_stencil_common.cc b/lib/stencil/Grid_stencil_common.cc
index 52daf466..6cbcc890 100644
--- a/lib/stencil/Grid_stencil_common.cc
+++ b/lib/stencil/Grid_stencil_common.cc
@@ -2,6 +2,96 @@
 
 namespace Grid {
 
+
+
+void CartesianStencil::LebesgueOrder(void) 
+{
+  _LebesgueReorder.resize(0);
+  
+  // Align up dimensions to power of two.
+  const StencilInteger one=1;
+  StencilInteger ND = _grid->_ndimension;
+  std::vector<StencilInteger> dims(ND);
+  std::vector<StencilInteger> adims(ND);
+  std::vector<std::vector<StencilInteger> > bitlist(ND);
+
+
+  for(StencilInteger mu=0;mu<ND;mu++){
+    dims[mu] = _grid->_rdimensions[mu];
+    assert ( dims[mu] != 0 );
+    adims[mu] = alignup(dims[mu]);
+  }
+
+  // List which bits of padded volume coordinate contribute; this strategy 
+  // i) avoids recursion 
+  // ii) has loop lengths at most the width of a 32 bit word.
+  int sitebit=0;
+  int split=24;
+  for(int mu=0;mu<ND;mu++){   // mu 0 takes bit 0; mu 1 takes bit 1 etc...
+    for(int bit=0;bit<split;bit++){
+    StencilInteger mask = one<<bit;
+      if ( mask&(adims[mu]-1) ){
+	bitlist[mu].push_back(sitebit);
+	sitebit++;
+      }
+    }
+  }
+  for(int bit=split;bit<32;bit++){
+    StencilInteger mask = one<<bit;
+    for(int mu=0;mu<ND;mu++){   // mu 0 takes bit 0; mu 1 takes bit 1 etc...
+      if ( mask&(adims[mu]-1) ){
+	bitlist[mu].push_back(sitebit);
+	sitebit++;
+      }
+    }
+  }
+
+  // Work out padded and unpadded volumes
+  StencilInteger avol = 1;
+  for(int mu=0;mu<ND;mu++) avol = avol * adims[mu];
+
+  StencilInteger vol = 1;
+  for(int mu=0;mu<ND;mu++) vol = vol * dims[mu];
+  
+  // Loop over padded volume, following Lebesgue curve
+  // We interleave the bits from sequential "mu".
+  std::vector<StencilInteger> ax(ND);
+  
+  for(StencilInteger asite=0;asite<avol;asite++){
+
+    // Start with zero and collect bits
+    for(int mu=0;mu<ND;mu++) ax[mu] = 0;
+
+    int contained = 1;
+    for(int mu=0;mu<ND;mu++){
+
+      // Build the coordinate on the aligned volume
+      for(int bit=0;bit<bitlist[mu].size();bit++){
+	int sbit=bitlist[mu][bit];
+
+	if(asite&(one<<sbit)){
+	  ax[mu]|=one<<bit;
+	}
+      }
+
+      // Is it contained in original box
+      if ( ax[mu]>dims[mu]-1 ) contained = 0;
+
+    }
+
+    if ( contained ) {
+      int site = ax[0]
+	+        dims[0]*ax[1]
+        +dims[0]*dims[1]*ax[2]
+        +dims[0]*dims[1]*dims[2]*ax[3];
+
+      _LebesgueReorder.push_back(site);
+    }
+  }
+
+  assert( _LebesgueReorder.size() == vol );
+}
+
   CartesianStencil::CartesianStencil(GridBase *grid,
 				     int npoints,
 				     int checkerboard,
@@ -20,6 +110,8 @@ namespace Grid {
       _unified_buffer_size=0;
       _request_count =0;
 
+      LebesgueOrder();
+
       int osites  = _grid->oSites();
 
       for(int i=0;i<npoints;i++){
diff --git a/tests/Grid_wilson.cc b/tests/Grid_wilson.cc
index ff1a9010..c64ea88c 100644
--- a/tests/Grid_wilson.cc
+++ b/tests/Grid_wilson.cc
@@ -22,7 +22,7 @@ int main (int argc, char ** argv)
 
   std::vector<int> simd_layout({1,1,2,2});
   std::vector<int> mpi_layout ({1,1,1,1});
-  std::vector<int> latt_size  ({4,4,8,8});
+  std::vector<int> latt_size  ({8,8,8,8});
     
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
   std::vector<int> seeds({1,2,3,4});
@@ -45,9 +45,7 @@ int main (int argc, char ** argv)
   }  
 
   for(int mu=0;mu<Nd;mu++){
-    //    U[mu] = 1.0;
-    //    pokeIndex<3>(Umu,U[mu],mu);
-    U[mu] = peekIndex<3>(Umu,mu);
+    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
   }
   
   std::vector<int> mask({1,1,1,1,1,1,1,1});

From bdf18941a227fac51770e63363f6c0300646df77 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 1 May 2015 10:57:33 +0100
Subject: [PATCH 113/429] Improving the byte swap support for portability

---
 configure                    | 71 ++++++++++++++++++++++++++++++++++++
 configure.ac                 |  3 ++
 lib/Grid_config.h            |  8 ++++
 lib/Grid_config.h.in         |  8 ++++
 lib/parallelIO/GridNerscIO.h | 38 +++++++++++++++++--
 5 files changed, 125 insertions(+), 3 deletions(-)

diff --git a/configure b/configure
index 9e554925..6a6c4331 100755
--- a/configure
+++ b/configure
@@ -1785,6 +1785,52 @@ $as_echo "$ac_res" >&6; }
 
 } # ac_fn_c_check_header_compile
 
+# ac_fn_c_check_decl LINENO SYMBOL VAR INCLUDES
+# ---------------------------------------------
+# Tests whether SYMBOL is declared in INCLUDES, setting cache variable VAR
+# accordingly.
+ac_fn_c_check_decl ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  as_decl_name=`echo $2|sed 's/ *(.*//'`
+  as_decl_use=`echo $2|sed -e 's/(/((/' -e 's/)/) 0&/' -e 's/,/) 0& (/g'`
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $as_decl_name is declared" >&5
+$as_echo_n "checking whether $as_decl_name is declared... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+#ifndef $as_decl_name
+#ifdef __cplusplus
+  (void) $as_decl_use;
+#else
+  (void) $as_decl_name;
+#endif
+#endif
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  eval "$3=yes"
+else
+  eval "$3=no"
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_decl
+
 # ac_fn_c_check_type LINENO TYPE VAR INCLUDES
 # -------------------------------------------
 # Tests whether TYPE exists after having included INCLUDES, setting cache
@@ -4958,6 +5004,31 @@ fi
 
 done
 
+#AC_CHECK_HEADERS(machine/endian.h)
+ac_fn_c_check_decl "$LINENO" "ntohll" "ac_cv_have_decl_ntohll" "#include <arpa/inet.h>
+"
+if test "x$ac_cv_have_decl_ntohll" = xyes; then :
+  ac_have_decl=1
+else
+  ac_have_decl=0
+fi
+
+cat >>confdefs.h <<_ACEOF
+#define HAVE_DECL_NTOHLL $ac_have_decl
+_ACEOF
+
+ac_fn_c_check_decl "$LINENO" "be64toh" "ac_cv_have_decl_be64toh" "#include <arpa/inet.h>
+"
+if test "x$ac_cv_have_decl_be64toh" = xyes; then :
+  ac_have_decl=1
+else
+  ac_have_decl=0
+fi
+
+cat >>confdefs.h <<_ACEOF
+#define HAVE_DECL_BE64TOH $ac_have_decl
+_ACEOF
+
 
 # Checks for typedefs, structures, and compiler characteristics.
 ac_fn_c_check_type "$LINENO" "size_t" "ac_cv_type_size_t" "$ac_includes_default"
diff --git a/configure.ac b/configure.ac
index b6e73bcb..ffd37d87 100644
--- a/configure.ac
+++ b/configure.ac
@@ -18,6 +18,9 @@ AC_CHECK_HEADERS(stdint.h)
 AC_CHECK_HEADERS(malloc/malloc.h)
 AC_CHECK_HEADERS(malloc.h)
 AC_CHECK_HEADERS(endian.h)
+#AC_CHECK_HEADERS(machine/endian.h)
+AC_CHECK_DECLS([ntohll],[], [], [[#include <arpa/inet.h>]])
+AC_CHECK_DECLS([be64toh],[], [], [[#include <arpa/inet.h>]])
 
 # Checks for typedefs, structures, and compiler characteristics.
 AC_TYPE_SIZE_T
diff --git a/lib/Grid_config.h b/lib/Grid_config.h
index 4152540e..c743bba0 100644
--- a/lib/Grid_config.h
+++ b/lib/Grid_config.h
@@ -16,6 +16,14 @@
 /* GRID_COMMS_NONE */
 /* #undef GRID_COMMS_NONE */
 
+/* Define to 1 if you have the declaration of `be64toh', and to 0 if you
+   don't. */
+#define HAVE_DECL_BE64TOH 0
+
+/* Define to 1 if you have the declaration of `ntohll', and to 0 if you don't.
+   */
+#define HAVE_DECL_NTOHLL 1
+
 /* Define to 1 if you have the <endian.h> header file. */
 /* #undef HAVE_ENDIAN_H */
 
diff --git a/lib/Grid_config.h.in b/lib/Grid_config.h.in
index 2dc0bda4..437a6095 100644
--- a/lib/Grid_config.h.in
+++ b/lib/Grid_config.h.in
@@ -15,6 +15,14 @@
 /* GRID_COMMS_NONE */
 #undef GRID_COMMS_NONE
 
+/* Define to 1 if you have the declaration of `be64toh', and to 0 if you
+   don't. */
+#undef HAVE_DECL_BE64TOH
+
+/* Define to 1 if you have the declaration of `ntohll', and to 0 if you don't.
+   */
+#undef HAVE_DECL_NTOHLL
+
 /* Define to 1 if you have the <endian.h> header file. */
 #undef HAVE_ENDIAN_H
 
diff --git a/lib/parallelIO/GridNerscIO.h b/lib/parallelIO/GridNerscIO.h
index 4dfa7979..4094ab70 100644
--- a/lib/parallelIO/GridNerscIO.h
+++ b/lib/parallelIO/GridNerscIO.h
@@ -9,10 +9,42 @@
 
 #ifdef HAVE_ENDIAN_H
 #include <endian.h>
+#endif
+
+
 #include <arpa/inet.h>
-#define ntohll be64toh
+
+// 64bit endian swap is a portability pain
+#ifndef __has_builtin         // Optional of course.
+#define __has_builtin(x) 0  // Compatibility with non-clang compilers.
+#endif
+
+#if HAVE_DECL_BE64TOH 
+#undef Grid_ntohll
+#define Grid_ntohll be64toh
+#endif
+
+#if HAVE_DECL_NTOHLL
+#undef  Grid_ntohll
+#define Grid_ntohll ntohll
+#endif
+
+#ifndef Grid_ntohll
+
+#if BYTE_ORDER == BIG_ENDIAN 
+
+#define Grid_ntohll(A) (A)
+
 #else 
-#include <arpa/inet.h>
+
+#if __has_builtin(__builtin_bswap64)
+#define Grid_ntohll(A) __builtin_bswap64(A)
+#else
+#error
+#endif
+
+#endif
+
 #endif
 
 namespace Grid {
@@ -208,7 +240,7 @@ inline void reconstruct3(LorentzColourMatrix & cm)
  {
    uint64_t * f = (uint64_t *)file_object;
    for(int i=0;i*sizeof(uint64_t)<bytes;i++){  
-     f[i] = ntohll(f[i]);
+     f[i] = Grid_ntohll(f[i]);
    }
  }
  void inline le64toh_v(void *file_object,uint32_t bytes)

From 6a39089a432302e0956d2f9d02d235be8e60fc5a Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 2 May 2015 17:52:36 +0100
Subject: [PATCH 114/429] Starting a benchmarking sub dir

---
 Makefile.am                          | 2 +-
 Makefile.in                          | 2 +-
 {tests => benchmarks}/Grid_wilson.cc | 0
 configure                            | 3 +++
 configure.ac                         | 1 +
 tests/Makefile.am                    | 5 +----
 6 files changed, 7 insertions(+), 6 deletions(-)
 rename {tests => benchmarks}/Grid_wilson.cc (100%)

diff --git a/Makefile.am b/Makefile.am
index a55b2b92..fae10e5d 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -1,4 +1,4 @@
 # additional include paths necessary to compile the C++ library
 AM_CXXFLAGS = -I$(top_srcdir)/
-SUBDIRS = lib tests
+SUBDIRS = lib tests benchmarks
 
diff --git a/Makefile.in b/Makefile.in
index 99b67fff..c9257ab2 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -294,7 +294,7 @@ top_srcdir = @top_srcdir@
 
 # additional include paths necessary to compile the C++ library
 AM_CXXFLAGS = -I$(top_srcdir)/
-SUBDIRS = lib tests
+SUBDIRS = lib tests benchmarks
 all: all-recursive
 
 .SUFFIXES:
diff --git a/tests/Grid_wilson.cc b/benchmarks/Grid_wilson.cc
similarity index 100%
rename from tests/Grid_wilson.cc
rename to benchmarks/Grid_wilson.cc
diff --git a/configure b/configure
index 6a6c4331..3d25f6a9 100755
--- a/configure
+++ b/configure
@@ -5174,6 +5174,8 @@ ac_config_files="$ac_config_files lib/Makefile"
 
 ac_config_files="$ac_config_files tests/Makefile"
 
+ac_config_files="$ac_config_files benchmarks/Makefile"
+
 cat >confcache <<\_ACEOF
 # This file is a shell script that caches the results of configure
 # tests run on this system so they can be shared between configure
@@ -5916,6 +5918,7 @@ do
     "Makefile") CONFIG_FILES="$CONFIG_FILES Makefile" ;;
     "lib/Makefile") CONFIG_FILES="$CONFIG_FILES lib/Makefile" ;;
     "tests/Makefile") CONFIG_FILES="$CONFIG_FILES tests/Makefile" ;;
+    "benchmarks/Makefile") CONFIG_FILES="$CONFIG_FILES benchmarks/Makefile" ;;
 
   *) as_fn_error $? "invalid argument: \`$ac_config_target'" "$LINENO" 5;;
   esac
diff --git a/configure.ac b/configure.ac
index ffd37d87..ce73ffd9 100644
--- a/configure.ac
+++ b/configure.ac
@@ -80,4 +80,5 @@ AM_CONDITIONAL(BUILD_COMMS_NONE,[ test "X${ac_COMMS}X" == "XnoneX" ])
 AC_CONFIG_FILES(Makefile)
 AC_CONFIG_FILES(lib/Makefile)
 AC_CONFIG_FILES(tests/Makefile)
+AC_CONFIG_FILES(benchmarks/Makefile)
 AC_OUTPUT
diff --git a/tests/Makefile.am b/tests/Makefile.am
index aaaff613..b129d3d9 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -5,7 +5,7 @@ AM_LDFLAGS = -L$(top_srcdir)/lib
 #
 # Test code
 #
-bin_PROGRAMS = Grid_main Grid_stencil Grid_nersc_io Grid_cshift Grid_gamma Grid_wilson Grid_simd
+bin_PROGRAMS = Grid_main Grid_stencil Grid_nersc_io Grid_cshift Grid_gamma  Grid_simd
 
 Grid_main_SOURCES = Grid_main.cc
 Grid_main_LDADD = -lGrid
@@ -22,8 +22,5 @@ Grid_gamma_LDADD = -lGrid
 Grid_stencil_SOURCES = Grid_stencil.cc
 Grid_stencil_LDADD = -lGrid
 
-Grid_wilson_SOURCES = Grid_wilson.cc
-Grid_wilson_LDADD = -lGrid
-
 Grid_simd_SOURCES = Grid_simd.cc
 Grid_simd_LDADD = -lGrid

From ea52562527020556237340b0437d966264eef7ef Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 2 May 2015 23:42:30 +0100
Subject: [PATCH 115/429] Added a comms benchmark

---
 benchmarks/Grid_comms.cc                   | 83 +++++++++++++++++++++
 benchmarks/Grid_wilson.cc                  | 10 +--
 lib/Grid_communicator.h                    | 12 +++-
 lib/communicator/Grid_communicator_fake.cc | 15 +++-
 lib/communicator/Grid_communicator_mpi.cc  | 84 +++++++++++++++-------
 5 files changed, 173 insertions(+), 31 deletions(-)
 create mode 100644 benchmarks/Grid_comms.cc

diff --git a/benchmarks/Grid_comms.cc b/benchmarks/Grid_comms.cc
new file mode 100644
index 00000000..ecc1a92e
--- /dev/null
+++ b/benchmarks/Grid_comms.cc
@@ -0,0 +1,83 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> simd_layout({1,1,2,2});
+  std::vector<int> mpi_layout ({1,2,2,1});
+
+
+  std::cout << "  L  "<<"\t\t"<<" Ls  "<<"\t\t"<<"bytes"<<"\t\t"<<"MB/s uni"<<"\t\t"<<"MB/s bidi"<<std::endl;
+  int Nloop=10;
+  for(int lat=4;lat<=16;lat+=4){
+    for(int Ls=1;Ls<=16;Ls*=2){
+
+      int bytes=lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD);
+      double start=usecond();
+      int ncomm=0;
+      for(int i=0;i<Nloop;i++){
+	std::vector<int> latt_size  ({lat,lat,lat,lat});
+    
+	GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+
+
+	std::vector<std::vector<HalfSpinColourVectorD> > xbuf(8,std::vector<HalfSpinColourVectorD>(lat*lat*lat*Ls));
+	std::vector<std::vector<HalfSpinColourVectorD> > rbuf(8,std::vector<HalfSpinColourVectorD>(lat*lat*lat*Ls));
+	std::vector<CartesianCommunicator::CommsRequest_t> requests;
+
+	ncomm=0;
+	for(int mu=0;mu<4;mu++){
+	
+	  if (mpi_layout[mu]>1 ) {
+	  
+	    ncomm++;
+	    int comm_proc=1;
+	    int xmit_to_rank;
+	    int recv_from_rank;
+	    
+	    Grid.ShiftedRanks(mu,comm_proc,xmit_to_rank,recv_from_rank);
+	    Grid.SendToRecvFromBegin(requests,
+				   (void *)&xbuf[mu][0],
+				   xmit_to_rank,
+				   (void *)&rbuf[mu][0],
+				   recv_from_rank,
+				   bytes);
+	
+	    comm_proc = mpi_layout[mu]-1;
+	  
+	    Grid.ShiftedRanks(mu,comm_proc,xmit_to_rank,recv_from_rank);
+	    Grid.SendToRecvFromBegin(requests,
+				     (void *)&xbuf[mu+4][0],
+				     xmit_to_rank,
+				     (void *)&rbuf[mu+4][0],
+				     recv_from_rank,
+				     bytes);
+	  
+	  }
+	}
+
+	Grid.SendToRecvFromComplete(requests);
+	Grid.Barrier();
+      }
+
+      double stop=usecond();
+
+      double xbytes    = Nloop*bytes*2*ncomm;
+      double rbytes    = xbytes;
+      double bidibytes = xbytes+rbytes;
+
+      double time = stop-start;
+
+      std::cout << lat<<"\t\t"<<Ls<<"\t\t"<<bytes<<"\t\t"<<xbytes/time<<"\t\t"<<bidibytes/time<<std::endl;
+    }
+    
+  }
+
+
+  Grid_finalize();
+}
diff --git a/benchmarks/Grid_wilson.cc b/benchmarks/Grid_wilson.cc
index c64ea88c..51255fcb 100644
--- a/benchmarks/Grid_wilson.cc
+++ b/benchmarks/Grid_wilson.cc
@@ -35,6 +35,7 @@ int main (int argc, char ** argv)
   LatticeFermion src(&Grid); random(pRNG,src);
   LatticeFermion result(&Grid); result=zero;
   LatticeFermion    ref(&Grid);    ref=zero;
+  LatticeFermion    err(&Grid);    
   LatticeFermion    tmp(&Grid);    tmp=zero;
   LatticeGaugeField Umu(&Grid); random(pRNG,Umu);
   std::vector<LatticeColourMatrix> U(4,&Grid);
@@ -82,12 +83,14 @@ int main (int argc, char ** argv)
   }
   double t1=usecond();
   double flops=1320*volume*ncall;
-
   
   std::cout << "Called Dw"<<std::endl;
   std::cout << "norm result "<< norm2(result)<<std::endl;
   std::cout << "norm ref    "<< norm2(ref)<<std::endl;
-  std::cout << "mflop/s = "<< flops/(t1-t0)<<std::endl;
+  std::cout << "mflop/s =   "<< flops/(t1-t0)<<std::endl;
+  err = ref -result;
+  std::cout << "norm diff   "<< norm2(err)<<std::endl;
+
 
   //  for(int ss=0;ss<10;ss++ ){
   for(int ss=0;ss<0;ss++ ){
@@ -100,8 +103,5 @@ int main (int argc, char ** argv)
     }
   }
 
-  ref = ref -result;
-  std::cout << "norm diff   "<< norm2(ref)<<std::endl;
-
   Grid_finalize();
 }
diff --git a/lib/Grid_communicator.h b/lib/Grid_communicator.h
index 84d080ae..661f39b8 100644
--- a/lib/Grid_communicator.h
+++ b/lib/Grid_communicator.h
@@ -21,6 +21,9 @@ class CartesianCommunicator {
 
 #ifdef GRID_COMMS_MPI
     MPI_Comm communicator;
+    typedef MPI_Request CommsRequest_t;
+#else 
+    typedef int CommsRequest_t;
 #endif
 
     // Constructor
@@ -77,13 +80,20 @@ class CartesianCommunicator {
       GlobalSumVector(ptr,words);
     }
     ////////////////////////////////////////////////////////////
-    // Face exchange
+    // Face exchange, buffer swap in translational invariant way
     ////////////////////////////////////////////////////////////
     void SendToRecvFrom(void *xmit,
 			int xmit_to_rank,
 			void *recv,
 			int recv_from_rank,
 			int bytes);
+    void SendToRecvFromBegin(std::vector<CommsRequest_t> &list,
+			 void *xmit,
+			 int xmit_to_rank,
+			 void *recv,
+			 int recv_from_rank,
+			 int bytes);
+    void SendToRecvFromComplete(std::vector<CommsRequest_t> &waitall);
 
     ////////////////////////////////////////////////////////////
     // Barrier
diff --git a/lib/communicator/Grid_communicator_fake.cc b/lib/communicator/Grid_communicator_fake.cc
index c5b77620..962283ee 100644
--- a/lib/communicator/Grid_communicator_fake.cc
+++ b/lib/communicator/Grid_communicator_fake.cc
@@ -20,7 +20,7 @@ void CartesianCommunicator::GlobalSum(double &){}
 void CartesianCommunicator::GlobalSum(uint32_t &){}
 void CartesianCommunicator::GlobalSumVector(double *,int N){}
 
-// Basic Halo comms primitive
+// Basic Halo comms primitive -- should never call in single node
 void CartesianCommunicator::SendToRecvFrom(void *xmit,
 					   int dest,
 					   void *recv,
@@ -29,6 +29,19 @@ void CartesianCommunicator::SendToRecvFrom(void *xmit,
 {
   exit(-1);
 }
+void CartesianCommunicator::SendToRecvFromBegin(std::vector<CommsRequest_t> &list,
+						void *xmit,
+						int dest,
+						void *recv,
+						int from,
+						int bytes)
+{
+  exit(-1);
+}
+void CartesianCommunicator::SendToRecvFromComplete(std::vector<CommsRequest_t> &list)
+{
+  exit(-1);
+}
 
 void CartesianCommunicator::Barrier(void)
 {
diff --git a/lib/communicator/Grid_communicator_mpi.cc b/lib/communicator/Grid_communicator_mpi.cc
index 6a3f69f2..c76fb76f 100644
--- a/lib/communicator/Grid_communicator_mpi.cc
+++ b/lib/communicator/Grid_communicator_mpi.cc
@@ -29,37 +29,45 @@ CartesianCommunicator::CartesianCommunicator(std::vector<int> &processors)
 }
 
 void CartesianCommunicator::GlobalSum(uint32_t &u){
-  MPI_Allreduce(MPI_IN_PLACE,&u,1,MPI_UINT32_T,MPI_SUM,communicator);
+  int ierr=MPI_Allreduce(MPI_IN_PLACE,&u,1,MPI_UINT32_T,MPI_SUM,communicator);
+  assert(ierr==0);
 }
 void CartesianCommunicator::GlobalSum(float &f){
-  MPI_Allreduce(MPI_IN_PLACE,&f,1,MPI_FLOAT,MPI_SUM,communicator);
+  int ierr=MPI_Allreduce(MPI_IN_PLACE,&f,1,MPI_FLOAT,MPI_SUM,communicator);
+  assert(ierr==0);
 }
 void CartesianCommunicator::GlobalSumVector(float *f,int N)
 {
-  MPI_Allreduce(MPI_IN_PLACE,f,N,MPI_FLOAT,MPI_SUM,communicator);
+  int ierr=MPI_Allreduce(MPI_IN_PLACE,f,N,MPI_FLOAT,MPI_SUM,communicator);
+  assert(ierr==0);
 }
 void CartesianCommunicator::GlobalSum(double &d)
 {
-  MPI_Allreduce(MPI_IN_PLACE,&d,1,MPI_DOUBLE,MPI_SUM,communicator);
+  int ierr = MPI_Allreduce(MPI_IN_PLACE,&d,1,MPI_DOUBLE,MPI_SUM,communicator);
+  assert(ierr==0);
 }
 void CartesianCommunicator::GlobalSumVector(double *d,int N)
 {
-  MPI_Allreduce(MPI_IN_PLACE,d,N,MPI_DOUBLE,MPI_SUM,communicator);
+  int ierr = MPI_Allreduce(MPI_IN_PLACE,d,N,MPI_DOUBLE,MPI_SUM,communicator);
+  assert(ierr==0);
 }
 void CartesianCommunicator::ShiftedRanks(int dim,int shift,int &source,int &dest)
 {
-  MPI_Cart_shift(communicator,dim,shift,&source,&dest);
+  int ierr=MPI_Cart_shift(communicator,dim,shift,&source,&dest);
+  assert(ierr==0);
 }
 int CartesianCommunicator::RankFromProcessorCoor(std::vector<int> &coor)
 {
   int rank;
-  MPI_Cart_rank  (communicator, &coor[0], &rank);
+  int ierr=MPI_Cart_rank  (communicator, &coor[0], &rank);
+  assert(ierr==0);
   return rank;
 }
 void  CartesianCommunicator::ProcessorCoorFromRank(int rank, std::vector<int> &coor)
 {
   coor.resize(_ndimension);
-  MPI_Cart_coords  (communicator, rank, _ndimension,&coor[0]);
+  int ierr=MPI_Cart_coords  (communicator, rank, _ndimension,&coor[0]);
+  assert(ierr==0);
 }
 
 // Basic Halo comms primitive
@@ -69,36 +77,64 @@ void CartesianCommunicator::SendToRecvFrom(void *xmit,
 					   int from,
 					   int bytes)
 {
-  MPI_Request reqs[2];
-  MPI_Status OkeyDokey[2];
+  std::vector<CommsRequest_t> reqs(0);
+  SendToRecvFromBegin(reqs,xmit,dest,recv,from,bytes);
+  SendToRecvFromComplete(reqs);
+}
+// Basic Halo comms primitive
+  void CartesianCommunicator::SendToRecvFromBegin(std::vector<CommsRequest_t> &list,
+						  void *xmit,
+						  int dest,
+						  void *recv,
+						  int from,
+						  int bytes)
+{
+  MPI_Request xrq;
+  MPI_Request rrq;
   int rank = _processor;
-  MPI_Isend(xmit, bytes, MPI_CHAR,dest,_processor,communicator,&reqs[0]);
-  MPI_Irecv(recv, bytes, MPI_CHAR,from,from,communicator,&reqs[1]);
-  MPI_Waitall(2,reqs,OkeyDokey);
+  int ierr;
+  ierr=MPI_Isend(xmit, bytes, MPI_CHAR,dest,_processor,communicator,&xrq);
+  ierr|=MPI_Irecv(recv, bytes, MPI_CHAR,from,from,communicator,&rrq);
+  
+  assert(ierr==0);
 
+  list.push_back(xrq);
+  list.push_back(rrq);
+
+}
+void CartesianCommunicator::SendToRecvFromComplete(std::vector<CommsRequest_t> &list)
+{
+  int nreq=list.size();
+  std::vector<MPI_Status> status(nreq);
+  int ierr = MPI_Waitall(nreq,&list[0],&status[0]);
+
+  assert(ierr==0);
 }
 
 void CartesianCommunicator::Barrier(void)
 {
-  MPI_Barrier(communicator);
+  int ierr = MPI_Barrier(communicator);
+  assert(ierr==0);
 }
 
 void CartesianCommunicator::Broadcast(int root,void* data, int bytes)
 {
-  MPI_Bcast(data,
-	    bytes,
-	    MPI_BYTE,
-	    root,
-	    communicator);
+  int ierr=MPI_Bcast(data,
+		     bytes,
+		     MPI_BYTE,
+		     root,
+		     communicator);
+  assert(ierr==0);
 }
 
 void CartesianCommunicator::BroadcastWorld(int root,void* data, int bytes)
 {
-  MPI_Bcast(data,
-	    bytes,
-	    MPI_BYTE,
-	    root,
-	    MPI_COMM_WORLD);
+  int ierr= MPI_Bcast(data,
+		      bytes,
+		      MPI_BYTE,
+		      root,
+		      MPI_COMM_WORLD);
+  assert(ierr==0);
 }
 
 }

From 253362f9783831d01191eca518efdd764160c4ec Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 2 May 2015 23:51:43 +0100
Subject: [PATCH 116/429] Added a comms benchmark

---
 benchmarks/Grid_comms.cc | 87 +++++++++++++++++++++++++++++++++++++++-
 1 file changed, 85 insertions(+), 2 deletions(-)

diff --git a/benchmarks/Grid_comms.cc b/benchmarks/Grid_comms.cc
index ecc1a92e..9963a6bf 100644
--- a/benchmarks/Grid_comms.cc
+++ b/benchmarks/Grid_comms.cc
@@ -12,8 +12,16 @@ int main (int argc, char ** argv)
   std::vector<int> mpi_layout ({1,2,2,1});
 
 
-  std::cout << "  L  "<<"\t\t"<<" Ls  "<<"\t\t"<<"bytes"<<"\t\t"<<"MB/s uni"<<"\t\t"<<"MB/s bidi"<<std::endl;
   int Nloop=10;
+  int nmu=0;
+  for(int mu=0;mu<4;mu++) if (mpi_layout[mu]>1) nmu++;
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "= Benchmarking concurrent halo exchange in "<<nmu<<" dimensions"<<std::endl;
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<" Ls  "<<"\t\t"<<"bytes"<<"\t\t"<<"MB/s uni"<<"\t\t"<<"MB/s bidi"<<std::endl;
+
+
+
   for(int lat=4;lat<=16;lat+=4){
     for(int Ls=1;Ls<=16;Ls*=2){
 
@@ -75,8 +83,83 @@ int main (int argc, char ** argv)
 
       std::cout << lat<<"\t\t"<<Ls<<"\t\t"<<bytes<<"\t\t"<<xbytes/time<<"\t\t"<<bidibytes/time<<std::endl;
     }
+  }    
+
+
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "= Benchmarking sequential halo exchange in "<<nmu<<" dimensions"<<std::endl;
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<" Ls  "<<"\t\t"<<"bytes"<<"\t\t"<<"MB/s uni"<<"\t\t"<<"MB/s bidi"<<std::endl;
+
+
+
+  for(int lat=4;lat<=16;lat+=4){
+    for(int Ls=1;Ls<=16;Ls*=2){
+
+      int bytes=lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD);
+      double start=usecond();
+      int ncomm=0;
+      for(int i=0;i<Nloop;i++){
+	std::vector<int> latt_size  ({lat,lat,lat,lat});
     
-  }
+	GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+
+
+	std::vector<std::vector<HalfSpinColourVectorD> > xbuf(8,std::vector<HalfSpinColourVectorD>(lat*lat*lat*Ls));
+	std::vector<std::vector<HalfSpinColourVectorD> > rbuf(8,std::vector<HalfSpinColourVectorD>(lat*lat*lat*Ls));
+
+	ncomm=0;
+	for(int mu=0;mu<4;mu++){
+	
+	  if (mpi_layout[mu]>1 ) {
+	  
+	    ncomm++;
+	    int comm_proc=1;
+	    int xmit_to_rank;
+	    int recv_from_rank;
+	    
+	    {
+	      std::vector<CartesianCommunicator::CommsRequest_t> requests;
+	      Grid.ShiftedRanks(mu,comm_proc,xmit_to_rank,recv_from_rank);
+	      Grid.SendToRecvFromBegin(requests,
+				       (void *)&xbuf[mu][0],
+				       xmit_to_rank,
+				       (void *)&rbuf[mu][0],
+				       recv_from_rank,
+				       bytes);
+	      Grid.SendToRecvFromComplete(requests);
+	    }
+
+	    comm_proc = mpi_layout[mu]-1;
+	  
+	    {
+	      std::vector<CartesianCommunicator::CommsRequest_t> requests;
+	      Grid.ShiftedRanks(mu,comm_proc,xmit_to_rank,recv_from_rank);
+	      Grid.SendToRecvFromBegin(requests,
+				       (void *)&xbuf[mu+4][0],
+				       xmit_to_rank,
+				       (void *)&rbuf[mu+4][0],
+				       recv_from_rank,
+				       bytes);
+	      Grid.SendToRecvFromComplete(requests);
+	    }
+	  }
+	}
+	Grid.Barrier();
+      }
+
+      double stop=usecond();
+
+      double xbytes    = Nloop*bytes*2*ncomm;
+      double rbytes    = xbytes;
+      double bidibytes = xbytes+rbytes;
+
+      double time = stop-start;
+
+      std::cout << lat<<"\t\t"<<Ls<<"\t\t"<<bytes<<"\t\t"<<xbytes/time<<"\t\t"<<bidibytes/time<<std::endl;
+    }
+  }  
+
 
 
   Grid_finalize();

From 9d93d1e6d49fd26c843a5a5559c831124a9c5df4 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sun, 3 May 2015 09:44:47 +0100
Subject: [PATCH 117/429] Comms and memory benchmarks added

---
 benchmarks/Grid_comms.cc                  |  38 +++---
 benchmarks/Grid_memory_bandwidth.cc       | 150 ++++++++++++++++++++++
 lib/Grid.h                                |   1 +
 lib/Grid_lattice.h                        |   7 +-
 lib/communicator/Grid_communicator_mpi.cc |   2 +-
 lib/lattice/Grid_lattice_arith.h          | 120 ++++++++++++++---
 lib/math/Grid_math_arith_mac.h            |   3 +-
 lib/math/Grid_math_arith_mul.h            |   1 -
 lib/math/Grid_math_tensors.h              |   8 +-
 lib/simd/Grid_vComplexD.h                 |   2 +-
 lib/simd/Grid_vComplexF.h                 |   2 +-
 lib/simd/Grid_vRealD.h                    |   5 +-
 lib/simd/Grid_vRealF.h                    |   6 +-
 tests/Grid_gamma.cc                       |  14 +-
 14 files changed, 300 insertions(+), 59 deletions(-)
 create mode 100644 benchmarks/Grid_memory_bandwidth.cc

diff --git a/benchmarks/Grid_comms.cc b/benchmarks/Grid_comms.cc
index 9963a6bf..ef7c5491 100644
--- a/benchmarks/Grid_comms.cc
+++ b/benchmarks/Grid_comms.cc
@@ -25,17 +25,19 @@ int main (int argc, char ** argv)
   for(int lat=4;lat<=16;lat+=4){
     for(int Ls=1;Ls<=16;Ls*=2){
 
+      std::vector<int> latt_size  ({lat,lat,lat,lat});
+
+      GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+
+      std::vector<std::vector<HalfSpinColourVectorD> > xbuf(8,std::vector<HalfSpinColourVectorD>(lat*lat*lat*Ls));
+      std::vector<std::vector<HalfSpinColourVectorD> > rbuf(8,std::vector<HalfSpinColourVectorD>(lat*lat*lat*Ls));
+
+      int ncomm;
       int bytes=lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD);
+
       double start=usecond();
-      int ncomm=0;
       for(int i=0;i<Nloop;i++){
-	std::vector<int> latt_size  ({lat,lat,lat,lat});
-    
-	GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
 
-
-	std::vector<std::vector<HalfSpinColourVectorD> > xbuf(8,std::vector<HalfSpinColourVectorD>(lat*lat*lat*Ls));
-	std::vector<std::vector<HalfSpinColourVectorD> > rbuf(8,std::vector<HalfSpinColourVectorD>(lat*lat*lat*Ls));
 	std::vector<CartesianCommunicator::CommsRequest_t> requests;
 
 	ncomm=0;
@@ -68,11 +70,10 @@ int main (int argc, char ** argv)
 	  
 	  }
 	}
-
 	Grid.SendToRecvFromComplete(requests);
 	Grid.Barrier();
-      }
 
+      }
       double stop=usecond();
 
       double xbytes    = Nloop*bytes*2*ncomm;
@@ -96,18 +97,20 @@ int main (int argc, char ** argv)
   for(int lat=4;lat<=16;lat+=4){
     for(int Ls=1;Ls<=16;Ls*=2){
 
+      std::vector<int> latt_size  ({lat,lat,lat,lat});
+
+      GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+
+      std::vector<std::vector<HalfSpinColourVectorD> > xbuf(8,std::vector<HalfSpinColourVectorD>(lat*lat*lat*Ls));
+      std::vector<std::vector<HalfSpinColourVectorD> > rbuf(8,std::vector<HalfSpinColourVectorD>(lat*lat*lat*Ls));
+
+
+      int ncomm;
       int bytes=lat*lat*lat*Ls*sizeof(HalfSpinColourVectorD);
+
       double start=usecond();
-      int ncomm=0;
       for(int i=0;i<Nloop;i++){
-	std::vector<int> latt_size  ({lat,lat,lat,lat});
     
-	GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
-
-
-	std::vector<std::vector<HalfSpinColourVectorD> > xbuf(8,std::vector<HalfSpinColourVectorD>(lat*lat*lat*Ls));
-	std::vector<std::vector<HalfSpinColourVectorD> > rbuf(8,std::vector<HalfSpinColourVectorD>(lat*lat*lat*Ls));
-
 	ncomm=0;
 	for(int mu=0;mu<4;mu++){
 	
@@ -131,7 +134,6 @@ int main (int argc, char ** argv)
 	    }
 
 	    comm_proc = mpi_layout[mu]-1;
-	  
 	    {
 	      std::vector<CartesianCommunicator::CommsRequest_t> requests;
 	      Grid.ShiftedRanks(mu,comm_proc,xmit_to_rank,recv_from_rank);
diff --git a/benchmarks/Grid_memory_bandwidth.cc b/benchmarks/Grid_memory_bandwidth.cc
new file mode 100644
index 00000000..e4e81445
--- /dev/null
+++ b/benchmarks/Grid_memory_bandwidth.cc
@@ -0,0 +1,150 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> simd_layout({1,2,2,2});
+  std::vector<int> mpi_layout ({1,1,1,1});
+
+  const int Nvec=8;
+  typedef Lattice< iVector< vReal,Nvec> > LatticeVec;
+
+  int Nloop=100;
+
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "= Benchmarking AXPY bandwidth"<<std::endl;
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"MB/s"<<std::endl;
+
+  
+  for(int lat=4;lat<=32;lat+=4){
+
+      std::vector<int> latt_size  ({lat,lat,lat,lat});
+
+      GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+
+      GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
+
+      LatticeVec z(&Grid); random(pRNG,z);
+      LatticeVec x(&Grid); random(pRNG,x);
+      LatticeVec y(&Grid); random(pRNG,y);
+      double a=1.0;
+
+
+      double start=usecond();
+      for(int i=0;i<Nloop;i++){
+	//	z=a*x+y;
+	//   inline void axpy(Lattice<vobj> &ret,double a,const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
+	axpy(z,a,x,y);
+      }
+      double stop=usecond();
+      double time = stop-start;
+      
+      double bytes=3*lat*lat*lat*lat*Nvec*sizeof(Real)*Nloop;
+      std::cout << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
+
+    }
+
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "= Benchmarking a*x + y bandwidth"<<std::endl;
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"MB/s"<<std::endl;
+
+  
+  for(int lat=4;lat<=32;lat+=4){
+
+      std::vector<int> latt_size  ({lat,lat,lat,lat});
+
+      GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+
+      GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
+
+      LatticeVec z(&Grid); random(pRNG,z);
+      LatticeVec x(&Grid); random(pRNG,x);
+      LatticeVec y(&Grid); random(pRNG,y);
+      double a=1.0;
+
+
+      double start=usecond();
+      for(int i=0;i<Nloop;i++){
+	z=a*x+y;
+      }
+      double stop=usecond();
+      double time = stop-start;
+      
+      double bytes=3*lat*lat*lat*lat*Nvec*sizeof(Real)*Nloop;
+      std::cout << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
+
+    }
+
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "= Benchmarking COPY bandwidth"<<std::endl;
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"MB/s"<<std::endl;
+
+
+  for(int lat=4;lat<=32;lat+=4){
+
+      std::vector<int> latt_size  ({lat,lat,lat,lat});
+
+      GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+
+      GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
+
+      LatticeVec z(&Grid); random(pRNG,z);
+      LatticeVec x(&Grid); random(pRNG,x);
+      LatticeVec y(&Grid); random(pRNG,y);
+      RealD a=1.0;
+
+
+      double start=usecond();
+      for(int i=0;i<Nloop;i++){
+	x=z;
+      }
+      double stop=usecond();
+      double time = stop-start;
+      
+      double bytes=2*lat*lat*lat*lat*Nvec*sizeof(Real)*Nloop;
+      std::cout << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
+
+  }
+
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "= Benchmarking READ bandwidth"<<std::endl;
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"\t\t"<<"bytes"<<"\t\t"<<"MB/s"<<std::endl;
+
+
+  for(int lat=4;lat<=32;lat+=4){
+
+      std::vector<int> latt_size  ({lat,lat,lat,lat});
+
+      GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+
+      GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
+
+      LatticeVec z(&Grid); random(pRNG,z);
+      LatticeVec x(&Grid); random(pRNG,x);
+      LatticeVec y(&Grid); random(pRNG,y);
+      RealD a=1.0;
+      ComplexD nn;
+
+      double start=usecond();
+      for(int i=0;i<Nloop;i++){
+	nn=norm2(x);
+      }
+      double stop=usecond();
+      double time = stop-start;
+      
+      double bytes=lat*lat*lat*lat*Nvec*sizeof(Real)*Nloop;
+      std::cout << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
+
+  }    
+
+  Grid_finalize();
+}
diff --git a/lib/Grid.h b/lib/Grid.h
index d402b545..05095f48 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -14,6 +14,7 @@
 
 #include <complex>
 #include <vector>
+#include <valarray>
 #include <iostream>
 #include <cassert>
 #include <random>
diff --git a/lib/Grid_lattice.h b/lib/Grid_lattice.h
index c76fa0a9..cf1c30c5 100644
--- a/lib/Grid_lattice.h
+++ b/lib/Grid_lattice.h
@@ -26,7 +26,8 @@ class Lattice
 public:
     GridBase *_grid;
     int checkerboard;
-    std::vector<vobj,alignedAllocator<vobj> > _odata;
+    //std::vector<vobj,alignedAllocator<vobj> > _odata;
+    std::valarray<vobj> _odata;
 public:
 
     typedef typename vobj::scalar_type scalar_type;
@@ -36,9 +37,9 @@ public:
     // Constructor requires "grid" passed.
     // what about a default grid?
     //////////////////////////////////////////////////////////////////
-    Lattice(GridBase *grid) : _grid(grid) {
+ Lattice(GridBase *grid) : _grid(grid), _odata(_grid->oSites()) {
       //        _odata.reserve(_grid->oSites());
-        _odata.resize(_grid->oSites());
+      //        _odata.resize(_grid->oSites());
         assert((((uint64_t)&_odata[0])&0xF) ==0);
         checkerboard=0;
     }
diff --git a/lib/communicator/Grid_communicator_mpi.cc b/lib/communicator/Grid_communicator_mpi.cc
index c76fb76f..6ef05c3d 100644
--- a/lib/communicator/Grid_communicator_mpi.cc
+++ b/lib/communicator/Grid_communicator_mpi.cc
@@ -93,7 +93,7 @@ void CartesianCommunicator::SendToRecvFrom(void *xmit,
   MPI_Request rrq;
   int rank = _processor;
   int ierr;
-  ierr=MPI_Isend(xmit, bytes, MPI_CHAR,dest,_processor,communicator,&xrq);
+  ierr =MPI_Isend(xmit, bytes, MPI_CHAR,dest,_processor,communicator,&xrq);
   ierr|=MPI_Irecv(recv, bytes, MPI_CHAR,from,from,communicator,&rrq);
   
   assert(ierr==0);
diff --git a/lib/lattice/Grid_lattice_arith.h b/lib/lattice/Grid_lattice_arith.h
index 26ed6a29..4d859173 100644
--- a/lib/lattice/Grid_lattice_arith.h
+++ b/lib/lattice/Grid_lattice_arith.h
@@ -3,6 +3,9 @@
 
 namespace Grid {
 
+  //////////////////////////////////////////////////////////////////////////////////////////////////////
+  // unary negation
+  //////////////////////////////////////////////////////////////////////////////////////////////////////
   template<class vobj>
   inline Lattice<vobj> operator -(const Lattice<vobj> &r)
   {
@@ -13,25 +16,10 @@ namespace Grid {
     }
     return ret;
   }
-  
-  template<class vobj>
-  inline void axpy(Lattice<vobj> &ret,double a,const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
-    conformable(lhs,rhs);
-#pragma omp parallel for
-    for(int ss=0;ss<lhs._grid->oSites();ss++){
-      axpy(&ret._odata[ss],a,&lhs._odata[ss],&rhs._odata[ss]);
-    }
-  }
-  template<class vobj>
-  inline void axpy(Lattice<vobj> &ret,std::complex<double> a,const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
-    conformable(lhs,rhs);
-#pragma omp parallel for
-    for(int ss=0;ss<lhs._grid->oSites();ss++){
-      axpy(&ret._odata[ss],a,&lhs._odata[ss],&rhs._odata[ss]);
-    }
-  }
-  
-  
+
+  //////////////////////////////////////////////////////////////////////////////////////////////////////
+  //  avoid copy back routines for mult, mac, sub, add
+  //////////////////////////////////////////////////////////////////////////////////////////////////////
   template<class obj1,class obj2,class obj3>
     void mult(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
     conformable(lhs,rhs);
@@ -69,7 +57,89 @@ namespace Grid {
     }
   }
   
+  //////////////////////////////////////////////////////////////////////////////////////////////////////
+  //  avoid copy back routines for mult, mac, sub, add
+  //////////////////////////////////////////////////////////////////////////////////////////////////////
+  template<class obj1,class obj2,class obj3>
+    void mult(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
+    conformable(lhs,rhs);
+    uint32_t vec_len = lhs._grid->oSites();
+#pragma omp parallel for
+    for(int ss=0;ss<vec_len;ss++){
+      mult(&ret._odata[ss],&lhs._odata[ss],&rhs);
+    }
+  }
+  
+  template<class obj1,class obj2,class obj3>
+    void mac(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
+    conformable(lhs,rhs);
+    uint32_t vec_len = lhs._grid->oSites();
+#pragma omp parallel for
+    for(int ss=0;ss<vec_len;ss++){
+      mac(&ret._odata[ss],&lhs._odata[ss],&rhs);
+    }
+  }
+  
+  template<class obj1,class obj2,class obj3>
+    void sub(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
+    conformable(lhs,rhs);
+#pragma omp parallel for
+    for(int ss=0;ss<lhs._grid->oSites();ss++){
+      sub(&ret._odata[ss],&lhs._odata[ss],&rhs);
+    }
+  }
+  template<class obj1,class obj2,class obj3>
+    void add(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
+    conformable(lhs,rhs);
+#pragma omp parallel for
+    for(int ss=0;ss<lhs._grid->oSites();ss++){
+      add(&ret._odata[ss],&lhs._odata[ss],&rhs);
+    }
+  }
+
+  //////////////////////////////////////////////////////////////////////////////////////////////////////
+  //  avoid copy back routines for mult, mac, sub, add
+  //////////////////////////////////////////////////////////////////////////////////////////////////////
+  template<class obj1,class obj2,class obj3>
+    void mult(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
+    conformable(lhs,rhs);
+    uint32_t vec_len = lhs._grid->oSites();
+#pragma omp parallel for
+    for(int ss=0;ss<vec_len;ss++){
+      mult(&ret._odata[ss],&lhs,&rhs._odata[ss]);
+    }
+  }
+  
+  template<class obj1,class obj2,class obj3>
+    void mac(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
+    conformable(lhs,rhs);
+    uint32_t vec_len = lhs._grid->oSites();
+#pragma omp parallel for
+    for(int ss=0;ss<vec_len;ss++){
+      mac(&ret._odata[ss],&lhs,&rhs._odata[ss]);
+    }
+  }
+  
+  template<class obj1,class obj2,class obj3>
+    void sub(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
+    conformable(lhs,rhs);
+#pragma omp parallel for
+    for(int ss=0;ss<lhs._grid->oSites();ss++){
+      sub(&ret._odata[ss],&lhs,&rhs._odata[ss]);
+    }
+  }
+  template<class obj1,class obj2,class obj3>
+    void add(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
+    conformable(lhs,rhs);
+#pragma omp parallel for
+    for(int ss=0;ss<lhs._grid->oSites();ss++){
+      add(&ret._odata[ss],&lhs,&rhs._odata[ss]);
+    }
+  }
+  
+  /////////////////////////////////////////////////////////////////////////////////////
   // Lattice BinOp Lattice,
+  /////////////////////////////////////////////////////////////////////////////////////
   template<class left,class right>
     inline auto operator * (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]*rhs._odata[0])>
   {
@@ -156,5 +226,17 @@ namespace Grid {
       }
       return ret;
     }
+
+  template<class sobj,class vobj>
+  inline void axpy(Lattice<vobj> &ret,sobj a,const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
+    conformable(lhs,rhs);
+    vobj tmp;
+#pragma omp parallel for
+    for(int ss=0;ss<lhs._grid->oSites();ss++){
+      tmp = a*lhs._odata[ss];
+      ret._odata[ss]= tmp+rhs._odata[ss];
+    }
+  }
+
 }
 #endif
diff --git a/lib/math/Grid_math_arith_mac.h b/lib/math/Grid_math_arith_mac.h
index 59cb9a5e..6260b98f 100644
--- a/lib/math/Grid_math_arith_mac.h
+++ b/lib/math/Grid_math_arith_mac.h
@@ -7,6 +7,7 @@ namespace Grid {
 ///////////////////////////////////////////////////////////////////////////////////////////////////
 /////////////////////////////////////////// MAC         ///////////////////////////////////////////
 ///////////////////////////////////////////////////////////////////////////////////////////////////
+    ///////////////////////////
 
     ///////////////////////////
     // Legal multiplication table
@@ -74,8 +75,6 @@ inline void mac(iVector<rrtype,N> * __restrict__ ret,const iVector<ltype,N> * __
     }
     return;
 }
-
-
 }
 
 #endif
diff --git a/lib/math/Grid_math_arith_mul.h b/lib/math/Grid_math_arith_mul.h
index c8bb0b2c..66c8b121 100644
--- a/lib/math/Grid_math_arith_mul.h
+++ b/lib/math/Grid_math_arith_mul.h
@@ -7,7 +7,6 @@ namespace Grid {
     ///////////////////////////////////////////////////////////////////////////////////////////////////
     /////////////////////////////////////////// MUL         ///////////////////////////////////////////
     ///////////////////////////////////////////////////////////////////////////////////////////////////
-
     
 template<class rtype,class vtype,class mtype>
 inline void mult(iScalar<rtype> * __restrict__ ret,const iScalar<mtype> * __restrict__ lhs,const iScalar<vtype> * __restrict__ rhs){
diff --git a/lib/math/Grid_math_tensors.h b/lib/math/Grid_math_tensors.h
index b08d0c83..25da7954 100644
--- a/lib/math/Grid_math_tensors.h
+++ b/lib/math/Grid_math_tensors.h
@@ -16,7 +16,7 @@ namespace Grid {
 // However note that doing this eliminates some syntactical sugar such as 
 // calling the constructor explicitly or implicitly
 //
-#define TENSOR_IS_POD
+#undef TENSOR_IS_POD
 
 template<class vtype> class iScalar
 {
@@ -36,7 +36,7 @@ public:
   //  template<int Level> using tensor_reduce_level = typename iScalar<GridTypeMapper<vtype>::tensor_reduce_level<Level> >;
 
 #ifndef TENSOR_IS_POD
-  iScalar(){;};
+  iScalar()=default;
   iScalar(scalar_type s) : _internal(s) {};// recurse down and hit the constructor for vector_type
   iScalar(const Zero &z){ *this = zero; };
 #endif
@@ -126,7 +126,7 @@ public:
 
 #ifndef TENSOR_IS_POD
   iVector(const Zero &z){ *this = zero; };
-  iVector() {};// Empty constructure
+  iVector() =default;
 #endif
 
     iVector<vtype,N> & operator= (const Zero &hero){
@@ -189,7 +189,7 @@ public:
 
 #ifndef TENSOR_IS_POD
   iMatrix(const Zero &z){ *this = zero; };
-  iMatrix() {};
+  iMatrix() =default;
 #endif
 
   iMatrix<vtype,N> & operator= (const Zero &hero){
diff --git a/lib/simd/Grid_vComplexD.h b/lib/simd/Grid_vComplexD.h
index 337b82c7..208e2640 100644
--- a/lib/simd/Grid_vComplexD.h
+++ b/lib/simd/Grid_vComplexD.h
@@ -13,7 +13,7 @@ namespace Grid {
             vzero(*this);
             return (*this);
         }
-        vComplexD(){};
+        vComplexD()=default;
         vComplexD(ComplexD a){
 	  vsplat(*this,a);
 	};
diff --git a/lib/simd/Grid_vComplexF.h b/lib/simd/Grid_vComplexF.h
index c42e4029..9c7922e3 100644
--- a/lib/simd/Grid_vComplexF.h
+++ b/lib/simd/Grid_vComplexF.h
@@ -28,7 +28,7 @@ namespace Grid {
             vzero(*this);
             return (*this);
         }
-        vComplexF(){};
+        vComplexF()=default;
         vComplexF(ComplexF a){
 	  vsplat(*this,a);
 	};
diff --git a/lib/simd/Grid_vRealD.h b/lib/simd/Grid_vRealD.h
index 36e189e1..0bdaa5e4 100644
--- a/lib/simd/Grid_vRealD.h
+++ b/lib/simd/Grid_vRealD.h
@@ -10,10 +10,13 @@ namespace Grid {
 	typedef dvec  vector_type;
 	typedef RealD scalar_type;
 
-        vRealD(){};
+        vRealD()=default;
         vRealD(RealD a){
 	  vsplat(*this,a);
 	};
+        vRealD(Zero &zero){
+	  zeroit(*this);
+	}
 
         friend inline void mult(vRealD * __restrict__ y,const vRealD * __restrict__ l,const vRealD *__restrict__ r) {*y = (*l) * (*r);}
         friend inline void sub (vRealD * __restrict__ y,const vRealD * __restrict__ l,const vRealD *__restrict__ r) {*y = (*l) - (*r);}
diff --git a/lib/simd/Grid_vRealF.h b/lib/simd/Grid_vRealF.h
index 70f76bc0..e8d89cec 100644
--- a/lib/simd/Grid_vRealF.h
+++ b/lib/simd/Grid_vRealF.h
@@ -8,14 +8,16 @@ namespace Grid {
         fvec v;
 
     public:
-
 	typedef fvec  vector_type;
 	typedef RealF scalar_type;
 
-        vRealF(){};
+        vRealF()=default;
         vRealF(RealF a){
 	  vsplat(*this,a);
 	};
+        vRealF(Zero &zero){
+	  zeroit(*this);
+	}
         ////////////////////////////////////
         // Arithmetic operator overloads +,-,*
         ////////////////////////////////////
diff --git a/tests/Grid_gamma.cc b/tests/Grid_gamma.cc
index ab6307f3..af735856 100644
--- a/tests/Grid_gamma.cc
+++ b/tests/Grid_gamma.cc
@@ -5,11 +5,10 @@ using namespace std;
 using namespace Grid;
 using namespace Grid::QCD;
 
-template<class d>
-struct scal {
-  d internal;
-};
-
+//template<class vobj> class is_pod< iScalar<vobj> >
+//{
+//
+//};
 
 int main (int argc, char ** argv)
 {
@@ -40,13 +39,16 @@ int main (int argc, char ** argv)
   std::cout << " Is pod " << std::is_pod<SpinVector>::value  << std::endl;
   std::cout << " Is pod double   " << std::is_pod<double>::value  << std::endl;
   std::cout << " Is pod ComplexF " << std::is_pod<ComplexF>::value  << std::endl;
-  std::cout << " Is pod scal<double> " << std::is_pod<scal<double> >::value  << std::endl;
+  std::cout << " Is triv double " << std::is_trivially_default_constructible<double>::value  << std::endl;
+  std::cout << " Is triv ComplexF " << std::is_trivially_default_constructible<ComplexF>::value  << std::endl;
   std::cout << " Is pod Scalar<double> " << std::is_pod<iScalar<double> >::value  << std::endl;
   std::cout << " Is pod Scalar<ComplexF> " << std::is_pod<iScalar<ComplexF> >::value  << std::endl;
   std::cout << " Is pod Scalar<vComplexF> " << std::is_pod<iScalar<vComplexF> >::value  << std::endl;
   std::cout << " Is pod Scalar<vComplexD> " << std::is_pod<iScalar<vComplexD> >::value  << std::endl;
   std::cout << " Is pod Scalar<vRealF> " << std::is_pod<iScalar<vRealF> >::value  << std::endl;
   std::cout << " Is pod Scalar<vRealD> " << std::is_pod<iScalar<vRealD> >::value  << std::endl;
+  std::cout << " Is triv Scalar<double> " <<std::is_trivially_default_constructible<iScalar<double> >::value << std::endl;
+  std::cout << " Is triv Scalar<vComplexD> "<<std::is_trivially_default_constructible<iScalar<vComplexD> >::value  << std::endl;
 
   for(int a=0;a<Ns;a++){
     ident()(a,a) = 1.0;

From 2b46ad38e25f61583ac7adc906b7520cccbdd268 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sun, 3 May 2015 09:48:13 +0100
Subject: [PATCH 118/429] Back to vector for now; cost of init loop is clear in
 the a*x + y loop in memory benchmark and must move to better container class.

---
 lib/Grid_lattice.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/lib/Grid_lattice.h b/lib/Grid_lattice.h
index cf1c30c5..75a80ef8 100644
--- a/lib/Grid_lattice.h
+++ b/lib/Grid_lattice.h
@@ -26,8 +26,8 @@ class Lattice
 public:
     GridBase *_grid;
     int checkerboard;
-    //std::vector<vobj,alignedAllocator<vobj> > _odata;
-    std::valarray<vobj> _odata;
+    std::vector<vobj,alignedAllocator<vobj> > _odata;
+    //std::valarray<vobj> _odata;
 public:
 
     typedef typename vobj::scalar_type scalar_type;

From 0e8415de1b2d2a61a742fd667dabe2f19a01ed5f Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 5 May 2015 17:56:42 +0100
Subject: [PATCH 119/429] Added a makefile

---
 benchmarks/Makefile.am | 19 +++++++++++++++++++
 1 file changed, 19 insertions(+)
 create mode 100644 benchmarks/Makefile.am

diff --git a/benchmarks/Makefile.am b/benchmarks/Makefile.am
new file mode 100644
index 00000000..d6fad22d
--- /dev/null
+++ b/benchmarks/Makefile.am
@@ -0,0 +1,19 @@
+# additional include paths necessary to compile the C++ library
+AM_CXXFLAGS = -I$(top_srcdir)/lib
+AM_LDFLAGS = -L$(top_srcdir)/lib
+
+#
+# Test code
+#
+bin_PROGRAMS = Grid_wilson Grid_comms Grid_memory_bandwidth
+
+
+Grid_wilson_SOURCES = Grid_wilson.cc
+Grid_wilson_LDADD = -lGrid
+
+Grid_comms_SOURCES = Grid_comms.cc
+Grid_comms_LDADD = -lGrid
+
+Grid_memory_bandwidth_SOURCES = Grid_memory_bandwidth.cc
+Grid_memory_bandwidth_LDADD = -lGrid
+

From b720222d981a77d5e6d9459f05a0b1bf90fee967 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 5 May 2015 18:08:53 +0100
Subject: [PATCH 120/429] Updated bandwidth test

---
 benchmarks/Grid_memory_bandwidth.cc | 95 +++++++++++++++--------------
 1 file changed, 48 insertions(+), 47 deletions(-)

diff --git a/benchmarks/Grid_memory_bandwidth.cc b/benchmarks/Grid_memory_bandwidth.cc
index e4e81445..dc25fe51 100644
--- a/benchmarks/Grid_memory_bandwidth.cc
+++ b/benchmarks/Grid_memory_bandwidth.cc
@@ -14,26 +14,26 @@ int main (int argc, char ** argv)
   const int Nvec=8;
   typedef Lattice< iVector< vReal,Nvec> > LatticeVec;
 
-  int Nloop=100;
+  int Nloop=1000;
 
   std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "= Benchmarking AXPY bandwidth"<<std::endl;
+  std::cout << "= Benchmarking fused AXPY bandwidth"<<std::endl;
   std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"MB/s"<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s"<<std::endl;
+  std::cout << "----------------------------------------------------------"<<std::endl;
 
-  
-  for(int lat=4;lat<=32;lat+=4){
+  for(int lat=4;lat<=24;lat+=4){
 
       std::vector<int> latt_size  ({lat,lat,lat,lat});
 
       GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
 
-      GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
+      //GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
 
-      LatticeVec z(&Grid); random(pRNG,z);
-      LatticeVec x(&Grid); random(pRNG,x);
-      LatticeVec y(&Grid); random(pRNG,y);
-      double a=1.0;
+      LatticeVec z(&Grid); //random(pRNG,z);
+      LatticeVec x(&Grid); //random(pRNG,x);
+      LatticeVec y(&Grid); //random(pRNG,y);
+      double a=2.0;
 
 
       double start=usecond();
@@ -43,31 +43,32 @@ int main (int argc, char ** argv)
 	axpy(z,a,x,y);
       }
       double stop=usecond();
-      double time = stop-start;
+      double time = (stop-start)/Nloop/1000;
       
-      double bytes=3*lat*lat*lat*lat*Nvec*sizeof(Real)*Nloop;
-      std::cout << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
+      double bytes=3*lat*lat*lat*lat*Nvec*sizeof(Real);
+      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
 
     }
 
   std::cout << "===================================================================================================="<<std::endl;
   std::cout << "= Benchmarking a*x + y bandwidth"<<std::endl;
   std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"MB/s"<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s"<<"\t\tGB/s (eff)"<<std::endl;
+  std::cout << "----------------------------------------------------------"<<std::endl;
 
   
-  for(int lat=4;lat<=32;lat+=4){
+  for(int lat=4;lat<=24;lat+=4){
 
       std::vector<int> latt_size  ({lat,lat,lat,lat});
 
       GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
 
-      GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
+      //GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
 
-      LatticeVec z(&Grid); random(pRNG,z);
-      LatticeVec x(&Grid); random(pRNG,x);
-      LatticeVec y(&Grid); random(pRNG,y);
-      double a=1.0;
+      LatticeVec z(&Grid); //random(pRNG,z);
+      LatticeVec x(&Grid); //random(pRNG,x);
+      LatticeVec y(&Grid); //random(pRNG,y);
+      double a=2.0;
 
 
       double start=usecond();
@@ -75,63 +76,63 @@ int main (int argc, char ** argv)
 	z=a*x+y;
       }
       double stop=usecond();
-      double time = stop-start;
+      double time = (stop-start)/Nloop/1000;
       
-      double bytes=3*lat*lat*lat*lat*Nvec*sizeof(Real)*Nloop;
-      std::cout << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
+      double effbytes=3*lat*lat*lat*lat*Nvec*sizeof(Real);
+      double bytes=5*lat*lat*lat*lat*Nvec*sizeof(Real);
+      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<"\t\t"<<effbytes/time<<std::endl;
 
     }
 
   std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "= Benchmarking COPY bandwidth"<<std::endl;
+  std::cout << "= Benchmarking SCALE bandwidth"<<std::endl;
   std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"MB/s"<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s"<<std::endl;
 
-
-  for(int lat=4;lat<=32;lat+=4){
+  for(int lat=4;lat<=24;lat+=4){
 
       std::vector<int> latt_size  ({lat,lat,lat,lat});
 
       GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
 
-      GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
+      //GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
 
-      LatticeVec z(&Grid); random(pRNG,z);
-      LatticeVec x(&Grid); random(pRNG,x);
-      LatticeVec y(&Grid); random(pRNG,y);
-      RealD a=1.0;
+      LatticeVec z(&Grid); //random(pRNG,z);
+      LatticeVec x(&Grid); //random(pRNG,x);
+      LatticeVec y(&Grid); //random(pRNG,y);
+      RealD a=2.0;
 
 
       double start=usecond();
       for(int i=0;i<Nloop;i++){
-	x=z;
+	x=a*z;
       }
       double stop=usecond();
-      double time = stop-start;
+      double time = (stop-start)/Nloop/1000;
       
-      double bytes=2*lat*lat*lat*lat*Nvec*sizeof(Real)*Nloop;
-      std::cout << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
+      double bytes=2*lat*lat*lat*lat*Nvec*sizeof(Real);
+      std::cout <<std::setprecision(3) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
 
   }
 
   std::cout << "===================================================================================================="<<std::endl;
   std::cout << "= Benchmarking READ bandwidth"<<std::endl;
   std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"\t\t"<<"bytes"<<"\t\t"<<"MB/s"<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s"<<std::endl;
+  std::cout << "----------------------------------------------------------"<<std::endl;
 
-
-  for(int lat=4;lat<=32;lat+=4){
+  for(int lat=4;lat<=24;lat+=4){
 
       std::vector<int> latt_size  ({lat,lat,lat,lat});
 
       GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
 
-      GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
+      //GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
 
-      LatticeVec z(&Grid); random(pRNG,z);
-      LatticeVec x(&Grid); random(pRNG,x);
-      LatticeVec y(&Grid); random(pRNG,y);
-      RealD a=1.0;
+      LatticeVec z(&Grid); //random(pRNG,z);
+      LatticeVec x(&Grid); //random(pRNG,x);
+      LatticeVec y(&Grid); //random(pRNG,y);
+      RealD a=2.0;
       ComplexD nn;
 
       double start=usecond();
@@ -139,10 +140,10 @@ int main (int argc, char ** argv)
 	nn=norm2(x);
       }
       double stop=usecond();
-      double time = stop-start;
+      double time = (stop-start)/Nloop/1000;
       
-      double bytes=lat*lat*lat*lat*Nvec*sizeof(Real)*Nloop;
-      std::cout << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
+      double bytes=lat*lat*lat*lat*Nvec*sizeof(Real);
+      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
 
   }    
 

From 249165d1b29c3fc38f6f85655a3051dba7ded578 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 5 May 2015 18:09:28 +0100
Subject: [PATCH 121/429] Added streaming stores

---
 lib/simd/Grid_vComplexD.h | 15 +++++++++++++++
 lib/simd/Grid_vComplexF.h | 29 ++++++++++++++++++++++-------
 lib/simd/Grid_vInteger.h  |  4 ++++
 lib/simd/Grid_vRealD.h    | 15 +++++++++++++++
 lib/simd/Grid_vRealF.h    | 15 +++++++++++++++
 5 files changed, 71 insertions(+), 7 deletions(-)

diff --git a/lib/simd/Grid_vComplexD.h b/lib/simd/Grid_vComplexD.h
index 208e2640..f0108d59 100644
--- a/lib/simd/Grid_vComplexD.h
+++ b/lib/simd/Grid_vComplexD.h
@@ -242,6 +242,21 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
 	assert(0);
 #endif
         }
+        friend inline void vstream(vComplexD &out,const vComplexD &in){
+#if defined (AVX1)|| defined (AVX2)
+	  _mm256_stream_pd((double *)&out.v,in.v);
+#endif
+#ifdef SSE4
+	  _mm_stream_pd((double *)&out.v,in.v);
+#endif
+#ifdef AVX512
+	  _mm512_stream_pd((double *)&out.v,in.v);
+	  //Note v has a3 a2 a1 a0
+#endif
+#ifdef QPX
+	  assert(0);
+#endif
+	}
       friend inline void vprefetch(const vComplexD &v)
         {
             _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
diff --git a/lib/simd/Grid_vComplexF.h b/lib/simd/Grid_vComplexF.h
index 9c7922e3..5f52fc53 100644
--- a/lib/simd/Grid_vComplexF.h
+++ b/lib/simd/Grid_vComplexF.h
@@ -176,21 +176,36 @@ namespace Grid {
             vsplat(ret,a,b);
         }
 
-friend inline void vstore(const vComplexF &ret, ComplexF *a){
+	friend inline void vstore(const vComplexF &ret, ComplexF *a){
 #if defined (AVX1)|| defined (AVX2)
-        _mm256_store_ps((float *)a,ret.v);
+	  _mm256_store_ps((float *)a,ret.v);
 #endif
 #ifdef SSE4
-        _mm_store_ps((float *)a,ret.v);
+	  _mm_store_ps((float *)a,ret.v);
 #endif
 #ifdef AVX512
-	_mm512_store_ps((float *)a,ret.v);
-//Note v has a3 a2 a1 a0
+	  _mm512_store_ps((float *)a,ret.v);
+	  //Note v has a3 a2 a1 a0
 #endif
 #ifdef QPX
-	assert(0);
+	  assert(0);
 #endif
-}
+	}
+        friend inline void vstream(vComplexF &out,const vComplexF &in){
+#if defined (AVX1)|| defined (AVX2)
+	  _mm256_stream_ps((float *)&out.v,in.v);
+#endif
+#ifdef SSE4
+	  _mm_stream_ps((float *)&out.v,in.v);
+#endif
+#ifdef AVX512
+	  _mm512_stream_ps((float *)&out.v,in.v);
+	  //Note v has a3 a2 a1 a0
+#endif
+#ifdef QPX
+	  assert(0);
+#endif
+	}
       friend inline void vprefetch(const vComplexF &v)
         {
             _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
diff --git a/lib/simd/Grid_vInteger.h b/lib/simd/Grid_vInteger.h
index 6035aea1..5a429a9c 100644
--- a/lib/simd/Grid_vInteger.h
+++ b/lib/simd/Grid_vInteger.h
@@ -186,6 +186,10 @@ namespace Grid {
 #endif
         }
 
+	friend inline void vstream(vInteger & out,const vInteger &in){
+	  out=in;
+        }
+
         friend inline void vprefetch(const vInteger &v)
         {
             _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
diff --git a/lib/simd/Grid_vRealD.h b/lib/simd/Grid_vRealD.h
index 0bdaa5e4..ee51fa03 100644
--- a/lib/simd/Grid_vRealD.h
+++ b/lib/simd/Grid_vRealD.h
@@ -174,6 +174,21 @@ namespace Grid {
 #endif
 #ifdef QPX
 	    assert(0);
+#endif
+	}
+        friend inline void vstream(vRealD &out,const vRealD &in){
+#if defined (AVX1)|| defined (AVX2)
+	  _mm256_stream_pd((double *)&out.v,in.v);
+#endif
+#ifdef SSE4
+	  _mm_stream_pd((double *)&out.v,in.v);
+#endif
+#ifdef AVX512
+	  _mm512_stream_pd((double *)&out.v,in.v);
+	  //Note v has a3 a2 a1 a0
+#endif
+#ifdef QPX
+	  assert(0);
 #endif
 	}
         friend inline void vprefetch(const vRealD &v)
diff --git a/lib/simd/Grid_vRealF.h b/lib/simd/Grid_vRealF.h
index e8d89cec..48e5c134 100644
--- a/lib/simd/Grid_vRealF.h
+++ b/lib/simd/Grid_vRealF.h
@@ -208,6 +208,21 @@ friend inline void vstore(const vRealF &ret, float *a){
 	assert(0);
 #endif
         }
+        friend inline void vstream(vRealF &out,const vRealF &in){
+#if defined (AVX1)|| defined (AVX2)
+	  _mm256_stream_ps((float *)&out.v,in.v);
+#endif
+#ifdef SSE4
+	  _mm_stream_ps((float *)&out.v,in.v);
+#endif
+#ifdef AVX512
+	  _mm512_stream_ps((float *)&out.v,in.v);
+	  //Note v has a3 a2 a1 a0
+#endif
+#ifdef QPX
+	  assert(0);
+#endif
+	}
 
 
         friend inline void vprefetch(const vRealF &v)

From cd990ba13d000f19f0321fdc5582417caf6f9024 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 5 May 2015 18:13:06 +0100
Subject: [PATCH 122/429] Streaming store option

---
 lib/math/Grid_math_tensors.h | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/lib/math/Grid_math_tensors.h b/lib/math/Grid_math_tensors.h
index 25da7954..564d5cb6 100644
--- a/lib/math/Grid_math_tensors.h
+++ b/lib/math/Grid_math_tensors.h
@@ -49,6 +49,9 @@ public:
       _internal=s;
       return *this;
     }
+    friend void vstream(iScalar<vtype> &out,const iScalar<vtype> &in){
+      vstream(out._internal,in._internal);
+    }
 
 
     friend void zeroit(iScalar<vtype> &that){
@@ -138,6 +141,12 @@ public:
             zeroit(that._internal[i]);
         }
     }
+    friend void vstream(iVector<vtype,N> &out,const iVector<vtype,N> &in){
+      for(int i=0;i<N;i++){
+	vstream(out._internal[i],in._internal[i]);
+      }
+    }
+
     friend void permute(iVector<vtype,N> &out,const iVector<vtype,N> &in,int permutetype){
       for(int i=0;i<N;i++){
 	permute(out._internal[i],in._internal[i],permutetype);
@@ -210,6 +219,12 @@ public:
 	zeroit(that._internal[i][j]);
     }}
   }
+  friend void vstream(iMatrix<vtype,N> &out,const iMatrix<vtype,N> &in){
+      for(int i=0;i<N;i++){
+      for(int j=0;j<N;j++){
+	vstream(out._internal[i][j],in._internal[i][j]);
+      }}
+    }
   friend void permute(iMatrix<vtype,N> &out,const iMatrix<vtype,N> &in,int permutetype){
     for(int i=0;i<N;i++){
       for(int j=0;j<N;j++){

From cb4b82b09f29c0134ad06315a34a39f8b339dfa9 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 5 May 2015 18:14:09 +0100
Subject: [PATCH 123/429] streaming store cases

---
 lib/lattice/Grid_lattice_arith.h | 125 +++++++++++++++++++------------
 1 file changed, 76 insertions(+), 49 deletions(-)

diff --git a/lib/lattice/Grid_lattice_arith.h b/lib/lattice/Grid_lattice_arith.h
index 4d859173..63ad3335 100644
--- a/lib/lattice/Grid_lattice_arith.h
+++ b/lib/lattice/Grid_lattice_arith.h
@@ -12,7 +12,7 @@ namespace Grid {
     Lattice<vobj> ret(r._grid);
 #pragma omp parallel for
     for(int ss=0;ss<r._grid->oSites();ss++){
-      ret._odata[ss]= -r._odata[ss];
+      vstream(ret._odata[ss], -r._odata[ss]);
     }
     return ret;
   }
@@ -23,20 +23,22 @@ namespace Grid {
   template<class obj1,class obj2,class obj3>
     void mult(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
     conformable(lhs,rhs);
-    uint32_t vec_len = lhs._grid->oSites();
 #pragma omp parallel for
-    for(int ss=0;ss<vec_len;ss++){
-      mult(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
+    for(int ss=0;ss<lhs._grid->oSites();ss++){
+      obj1 tmp;
+      mult(&tmp,&lhs._odata[ss],&rhs._odata[ss]);
+      vstream(ret._odata[ss],tmp);
     }
   }
   
   template<class obj1,class obj2,class obj3>
     void mac(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
     conformable(lhs,rhs);
-    uint32_t vec_len = lhs._grid->oSites();
 #pragma omp parallel for
-    for(int ss=0;ss<vec_len;ss++){
-      mac(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
+    for(int ss=0;ss<lhs._grid->oSites();ss++){
+      obj1 tmp;
+      mac(&tmp,&lhs._odata[ss],&rhs._odata[ss]);
+      vstream(ret._odata[ss],tmp);
     }
   }
   
@@ -45,7 +47,9 @@ namespace Grid {
     conformable(lhs,rhs);
 #pragma omp parallel for
     for(int ss=0;ss<lhs._grid->oSites();ss++){
-      sub(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
+      obj1 tmp;
+      sub(&tmp,&lhs._odata[ss],&rhs._odata[ss]);
+      vstream(ret._odata[ss],tmp);
     }
   }
   template<class obj1,class obj2,class obj3>
@@ -53,7 +57,9 @@ namespace Grid {
     conformable(lhs,rhs);
 #pragma omp parallel for
     for(int ss=0;ss<lhs._grid->oSites();ss++){
-      add(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
+      obj1 tmp;
+      add(&tmp,&lhs._odata[ss],&rhs._odata[ss]);
+      vstream(ret._odata[ss],tmp);
     }
   }
   
@@ -62,88 +68,100 @@ namespace Grid {
   //////////////////////////////////////////////////////////////////////////////////////////////////////
   template<class obj1,class obj2,class obj3>
     void mult(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
-    conformable(lhs,rhs);
-    uint32_t vec_len = lhs._grid->oSites();
+    conformable(lhs,ret);
 #pragma omp parallel for
-    for(int ss=0;ss<vec_len;ss++){
-      mult(&ret._odata[ss],&lhs._odata[ss],&rhs);
+    for(int ss=0;ss<lhs._grid->oSites();ss++){
+      obj1 tmp;
+      mult(&tmp,&lhs._odata[ss],&rhs);
+      vstream(ret._odata[ss],tmp);
     }
   }
   
   template<class obj1,class obj2,class obj3>
     void mac(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
-    conformable(lhs,rhs);
-    uint32_t vec_len = lhs._grid->oSites();
+    conformable(lhs,ret);
 #pragma omp parallel for
-    for(int ss=0;ss<vec_len;ss++){
-      mac(&ret._odata[ss],&lhs._odata[ss],&rhs);
+    for(int ss=0;ss<lhs._grid->oSites();ss++){
+      obj1 tmp;
+      mac(&tmp,&lhs._odata[ss],&rhs);
+      vstream(ret._odata[ss],tmp);
     }
   }
   
   template<class obj1,class obj2,class obj3>
     void sub(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
-    conformable(lhs,rhs);
+    conformable(lhs,ret);
 #pragma omp parallel for
     for(int ss=0;ss<lhs._grid->oSites();ss++){
-      sub(&ret._odata[ss],&lhs._odata[ss],&rhs);
+      obj1 tmp;
+      sub(&tmp,&lhs._odata[ss],&rhs);
+      vstream(ret._odata[ss],tmp);
     }
   }
   template<class obj1,class obj2,class obj3>
     void add(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
-    conformable(lhs,rhs);
+    conformable(lhs,ret);
 #pragma omp parallel for
     for(int ss=0;ss<lhs._grid->oSites();ss++){
-      add(&ret._odata[ss],&lhs._odata[ss],&rhs);
+      obj1 tmp;
+      add(&tmp,&lhs._odata[ss],&rhs);
+      vstream(ret._odata[ss],tmp);
     }
   }
 
   //////////////////////////////////////////////////////////////////////////////////////////////////////
   //  avoid copy back routines for mult, mac, sub, add
   //////////////////////////////////////////////////////////////////////////////////////////////////////
-  template<class obj1,class obj2,class obj3>
+    template<class obj1,class obj2,class obj3>
     void mult(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
-    conformable(lhs,rhs);
-    uint32_t vec_len = lhs._grid->oSites();
+    conformable(ret,rhs);
 #pragma omp parallel for
-    for(int ss=0;ss<vec_len;ss++){
-      mult(&ret._odata[ss],&lhs,&rhs._odata[ss]);
+    for(int ss=0;ss<rhs._grid->oSites();ss++){
+      obj1 tmp;
+      mult(&tmp,&lhs,&rhs._odata[ss]);
+      vstream(ret._odata[ss],tmp);
     }
   }
   
   template<class obj1,class obj2,class obj3>
     void mac(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
-    conformable(lhs,rhs);
-    uint32_t vec_len = lhs._grid->oSites();
+    conformable(ret,rhs);
 #pragma omp parallel for
-    for(int ss=0;ss<vec_len;ss++){
-      mac(&ret._odata[ss],&lhs,&rhs._odata[ss]);
+    for(int ss=0;ss<rhs._grid->oSites();ss++){
+      obj1 tmp;
+      mac(&tmp,&lhs,&rhs._odata[ss]);
+      vstream(ret._odata[ss],tmp);
     }
   }
   
   template<class obj1,class obj2,class obj3>
     void sub(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
-    conformable(lhs,rhs);
+    conformable(ret,rhs);
 #pragma omp parallel for
-    for(int ss=0;ss<lhs._grid->oSites();ss++){
-      sub(&ret._odata[ss],&lhs,&rhs._odata[ss]);
+    for(int ss=0;ss<rhs._grid->oSites();ss++){
+      obj1 tmp;
+      sub(&tmp,&lhs,&rhs._odata[ss]);
+      vstream(ret._odata[ss],tmp);
     }
   }
   template<class obj1,class obj2,class obj3>
     void add(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
-    conformable(lhs,rhs);
+    conformable(ret,rhs);
 #pragma omp parallel for
-    for(int ss=0;ss<lhs._grid->oSites();ss++){
-      add(&ret._odata[ss],&lhs,&rhs._odata[ss]);
+    for(int ss=0;ss<rhs._grid->oSites();ss++){
+      obj1 tmp;
+      add(&tmp,&lhs,&rhs._odata[ss]);
+      vstream(ret._odata[ss],tmp);
     }
   }
   
   /////////////////////////////////////////////////////////////////////////////////////
   // Lattice BinOp Lattice,
+  //NB mult performs conformable check. Do not reapply here for performance.
   /////////////////////////////////////////////////////////////////////////////////////
   template<class left,class right>
     inline auto operator * (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]*rhs._odata[0])>
   {
-    //NB mult performs conformable check. Do not reapply here for performance.
     Lattice<decltype(lhs._odata[0]*rhs._odata[0])> ret(rhs._grid);
     mult(ret,lhs,rhs);
     return ret;
@@ -151,7 +169,6 @@ namespace Grid {
   template<class left,class right>
     inline auto operator + (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]+rhs._odata[0])>
   {
-    //NB mult performs conformable check. Do not reapply here for performance.
     Lattice<decltype(lhs._odata[0]+rhs._odata[0])> ret(rhs._grid);
     add(ret,lhs,rhs);
     return ret;
@@ -159,7 +176,6 @@ namespace Grid {
   template<class left,class right>
     inline auto operator - (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]-rhs._odata[0])>
   {
-    //NB mult performs conformable check. Do not reapply here for performance.
     Lattice<decltype(lhs._odata[0]-rhs._odata[0])> ret(rhs._grid);
     sub(ret,lhs,rhs);
     return ret;
@@ -172,9 +188,11 @@ namespace Grid {
     Lattice<decltype(lhs*rhs._odata[0])> ret(rhs._grid);
 #pragma omp parallel for
     for(int ss=0;ss<rhs._grid->oSites(); ss++){
-      ret._odata[ss]=lhs*rhs._odata[ss];
+      decltype(lhs*rhs._odata[0]) tmp=lhs*rhs._odata[ss]; 
+      vstream(ret._odata[ss],tmp);
+	   //      ret._odata[ss]=lhs*rhs._odata[ss];
     }
-        return ret;
+    return ret;
   }
   template<class left,class right>
     inline auto operator + (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs+rhs._odata[0])>
@@ -182,7 +200,9 @@ namespace Grid {
       Lattice<decltype(lhs+rhs._odata[0])> ret(rhs._grid);
 #pragma omp parallel for
       for(int ss=0;ss<rhs._grid->oSites(); ss++){
-	ret._odata[ss]=lhs+rhs._odata[ss];
+	decltype(lhs+rhs._odata[0]) tmp =lhs-rhs._odata[ss];  
+	vstream(ret._odata[ss],tmp);
+	//	ret._odata[ss]=lhs+rhs._odata[ss];
       }
         return ret;
     }
@@ -192,7 +212,9 @@ namespace Grid {
     Lattice<decltype(lhs-rhs._odata[0])> ret(rhs._grid);
 #pragma omp parallel for
     for(int ss=0;ss<rhs._grid->oSites(); ss++){
-      ret._odata[ss]=lhs-rhs._odata[ss];
+      decltype(lhs-rhs._odata[0]) tmp=lhs-rhs._odata[ss];  
+      vstream(ret._odata[ss],tmp);
+      //      ret._odata[ss]=lhs-rhs._odata[ss];
     }
     return ret;
   }
@@ -202,7 +224,9 @@ namespace Grid {
       Lattice<decltype(lhs._odata[0]*rhs)> ret(lhs._grid);
 #pragma omp parallel for
       for(int ss=0;ss<lhs._grid->oSites(); ss++){
-            ret._odata[ss]=lhs._odata[ss]*rhs;
+	decltype(lhs._odata[0]*rhs) tmp =lhs._odata[ss]*rhs;
+	vstream(ret._odata[ss],tmp);
+	//            ret._odata[ss]=lhs._odata[ss]*rhs;
       }
       return ret;
     }
@@ -212,7 +236,9 @@ namespace Grid {
         Lattice<decltype(lhs._odata[0]+rhs)> ret(lhs._grid);
 #pragma omp parallel for
         for(int ss=0;ss<rhs._grid->oSites(); ss++){
-	  ret._odata[ss]=lhs._odata[ss]+rhs;
+	  decltype(lhs._odata[0]+rhs) tmp=lhs._odata[ss]+rhs; 
+	  vstream(ret._odata[ss],tmp);
+	  //	  ret._odata[ss]=lhs._odata[ss]+rhs;
         }
         return ret;
     }
@@ -222,7 +248,9 @@ namespace Grid {
       Lattice<decltype(lhs._odata[0]-rhs)> ret(lhs._grid);
 #pragma omp parallel for
       for(int ss=0;ss<rhs._grid->oSites(); ss++){
-	ret._odata[ss]=lhs._odata[ss]-rhs;
+	  decltype(lhs._odata[0]-rhs) tmp=lhs._odata[ss]-rhs;
+	  vstream(ret._odata[ss],tmp);
+	  //	ret._odata[ss]=lhs._odata[ss]-rhs;
       }
       return ret;
     }
@@ -230,11 +258,10 @@ namespace Grid {
   template<class sobj,class vobj>
   inline void axpy(Lattice<vobj> &ret,sobj a,const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
     conformable(lhs,rhs);
-    vobj tmp;
 #pragma omp parallel for
     for(int ss=0;ss<lhs._grid->oSites();ss++){
-      tmp = a*lhs._odata[ss];
-      ret._odata[ss]= tmp+rhs._odata[ss];
+      vobj tmp = a*lhs._odata[ss];
+      vstream(ret._odata[ss],tmp+rhs._odata[ss]);
     }
   }
 

From cdd5cdeda2113f6bf1af466835513de5f7839f38 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@t002.ph.ed.ac.uk>
Date: Tue, 5 May 2015 22:09:22 +0100
Subject: [PATCH 124/429] Cleaned up for Linux

---
 Makefile.in                         |  393 +--
 TODO                                |    4 +-
 aclocal.m4                          |  707 ++--
 benchmarks/Grid_memory_bandwidth.cc |   24 +-
 configure                           | 5011 ++++++++++++++++-----------
 lib/Grid_config.h                   |   17 +-
 lib/Grid_config.h.in                |    3 -
 lib/Grid_init.cc                    |    2 +-
 lib/qcd/Grid_qcd_wilson_dop.cc      |    2 +-
 tests/Grid_gamma.cc                 |    8 +-
 10 files changed, 3378 insertions(+), 2793 deletions(-)

diff --git a/Makefile.in b/Makefile.in
index c9257ab2..f9e18fe4 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -1,8 +1,9 @@
-# Makefile.in generated by automake 1.15 from Makefile.am.
+# Makefile.in generated by automake 1.11.1 from Makefile.am.
 # @configure_input@
 
-# Copyright (C) 1994-2014 Free Software Foundation, Inc.
-
+# Copyright (C) 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002,
+# 2003, 2004, 2005, 2006, 2007, 2008, 2009  Free Software Foundation,
+# Inc.
 # This Makefile.in is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
@@ -14,61 +15,6 @@
 
 @SET_MAKE@
 VPATH = @srcdir@
-am__is_gnu_make = { \
-  if test -z '$(MAKELEVEL)'; then \
-    false; \
-  elif test -n '$(MAKE_HOST)'; then \
-    true; \
-  elif test -n '$(MAKE_VERSION)' && test -n '$(CURDIR)'; then \
-    true; \
-  else \
-    false; \
-  fi; \
-}
-am__make_running_with_option = \
-  case $${target_option-} in \
-      ?) ;; \
-      *) echo "am__make_running_with_option: internal error: invalid" \
-              "target option '$${target_option-}' specified" >&2; \
-         exit 1;; \
-  esac; \
-  has_opt=no; \
-  sane_makeflags=$$MAKEFLAGS; \
-  if $(am__is_gnu_make); then \
-    sane_makeflags=$$MFLAGS; \
-  else \
-    case $$MAKEFLAGS in \
-      *\\[\ \	]*) \
-        bs=\\; \
-        sane_makeflags=`printf '%s\n' "$$MAKEFLAGS" \
-          | sed "s/$$bs$$bs[$$bs $$bs	]*//g"`;; \
-    esac; \
-  fi; \
-  skip_next=no; \
-  strip_trailopt () \
-  { \
-    flg=`printf '%s\n' "$$flg" | sed "s/$$1.*$$//"`; \
-  }; \
-  for flg in $$sane_makeflags; do \
-    test $$skip_next = yes && { skip_next=no; continue; }; \
-    case $$flg in \
-      *=*|--*) continue;; \
-        -*I) strip_trailopt 'I'; skip_next=yes;; \
-      -*I?*) strip_trailopt 'I';; \
-        -*O) strip_trailopt 'O'; skip_next=yes;; \
-      -*O?*) strip_trailopt 'O';; \
-        -*l) strip_trailopt 'l'; skip_next=yes;; \
-      -*l?*) strip_trailopt 'l';; \
-      -[dEDm]) skip_next=yes;; \
-      -[JT]) skip_next=yes;; \
-    esac; \
-    case $$flg in \
-      *$$target_option*) has_opt=yes; break;; \
-    esac; \
-  done; \
-  test $$has_opt = yes
-am__make_dryrun = (target_option=n; $(am__make_running_with_option))
-am__make_keepgoing = (target_option=k; $(am__make_running_with_option))
 pkgdatadir = $(datadir)/@PACKAGE@
 pkgincludedir = $(includedir)/@PACKAGE@
 pkglibdir = $(libdir)/@PACKAGE@
@@ -86,86 +32,43 @@ NORMAL_UNINSTALL = :
 PRE_UNINSTALL = :
 POST_UNINSTALL = :
 subdir = .
+DIST_COMMON = README $(am__configure_deps) $(srcdir)/Makefile.am \
+	$(srcdir)/Makefile.in $(top_srcdir)/configure AUTHORS COPYING \
+	ChangeLog INSTALL NEWS TODO compile depcomp install-sh missing
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/configure.ac
 am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
 	$(ACLOCAL_M4)
-DIST_COMMON = $(srcdir)/Makefile.am $(top_srcdir)/configure \
-	$(am__configure_deps) $(am__DIST_COMMON)
 am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \
  configure.lineno config.status.lineno
 mkinstalldirs = $(install_sh) -d
 CONFIG_HEADER = $(top_builddir)/lib/Grid_config.h
 CONFIG_CLEAN_FILES =
 CONFIG_CLEAN_VPATH_FILES =
-AM_V_P = $(am__v_P_@AM_V@)
-am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
-am__v_P_0 = false
-am__v_P_1 = :
-AM_V_GEN = $(am__v_GEN_@AM_V@)
-am__v_GEN_ = $(am__v_GEN_@AM_DEFAULT_V@)
-am__v_GEN_0 = @echo "  GEN     " $@;
-am__v_GEN_1 = 
-AM_V_at = $(am__v_at_@AM_V@)
-am__v_at_ = $(am__v_at_@AM_DEFAULT_V@)
-am__v_at_0 = @
-am__v_at_1 = 
 SOURCES =
 DIST_SOURCES =
-RECURSIVE_TARGETS = all-recursive check-recursive cscopelist-recursive \
-	ctags-recursive dvi-recursive html-recursive info-recursive \
-	install-data-recursive install-dvi-recursive \
-	install-exec-recursive install-html-recursive \
-	install-info-recursive install-pdf-recursive \
-	install-ps-recursive install-recursive installcheck-recursive \
-	installdirs-recursive pdf-recursive ps-recursive \
-	tags-recursive uninstall-recursive
-am__can_run_installinfo = \
-  case $$AM_UPDATE_INFO_DIR in \
-    n|no|NO) false;; \
-    *) (install-info --version) >/dev/null 2>&1;; \
-  esac
+RECURSIVE_TARGETS = all-recursive check-recursive dvi-recursive \
+	html-recursive info-recursive install-data-recursive \
+	install-dvi-recursive install-exec-recursive \
+	install-html-recursive install-info-recursive \
+	install-pdf-recursive install-ps-recursive install-recursive \
+	installcheck-recursive installdirs-recursive pdf-recursive \
+	ps-recursive uninstall-recursive
 RECURSIVE_CLEAN_TARGETS = mostlyclean-recursive clean-recursive	\
   distclean-recursive maintainer-clean-recursive
-am__recursive_targets = \
-  $(RECURSIVE_TARGETS) \
-  $(RECURSIVE_CLEAN_TARGETS) \
-  $(am__extra_recursive_targets)
-AM_RECURSIVE_TARGETS = $(am__recursive_targets:-recursive=) TAGS CTAGS \
-	cscope distdir dist dist-all distcheck
-am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) $(LISP)
-# Read a list of newline-separated strings from the standard input,
-# and print each of them once, without duplicates.  Input order is
-# *not* preserved.
-am__uniquify_input = $(AWK) '\
-  BEGIN { nonempty = 0; } \
-  { items[$$0] = 1; nonempty = 1; } \
-  END { if (nonempty) { for (i in items) print i; }; } \
-'
-# Make sure the list of sources is unique.  This is necessary because,
-# e.g., the same source file might be shared among _SOURCES variables
-# for different programs/libraries.
-am__define_uniq_tagged_files = \
-  list='$(am__tagged_files)'; \
-  unique=`for i in $$list; do \
-    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
-  done | $(am__uniquify_input)`
+AM_RECURSIVE_TARGETS = $(RECURSIVE_TARGETS:-recursive=) \
+	$(RECURSIVE_CLEAN_TARGETS:-recursive=) tags TAGS ctags CTAGS \
+	distdir dist dist-all distcheck
 ETAGS = etags
 CTAGS = ctags
-CSCOPE = cscope
 DIST_SUBDIRS = $(SUBDIRS)
-am__DIST_COMMON = $(srcdir)/Makefile.in AUTHORS COPYING ChangeLog \
-	INSTALL NEWS README TODO compile depcomp install-sh missing
 DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
 distdir = $(PACKAGE)-$(VERSION)
 top_distdir = $(distdir)
 am__remove_distdir = \
-  if test -d "$(distdir)"; then \
-    find "$(distdir)" -type d ! -perm -200 -exec chmod u+w {} ';' \
-      && rm -rf "$(distdir)" \
-      || { sleep 5 && rm -rf "$(distdir)"; }; \
-  else :; fi
-am__post_remove_distdir = $(am__remove_distdir)
+  { test ! -d "$(distdir)" \
+    || { find "$(distdir)" -type d ! -perm -200 -exec chmod u+w {} ';' \
+         && rm -fr "$(distdir)"; }; }
 am__relativize = \
   dir0=`pwd`; \
   sed_first='s,^\([^/]*\)/.*$$,\1,'; \
@@ -193,14 +96,10 @@ am__relativize = \
   reldir="$$dir2"
 DIST_ARCHIVES = $(distdir).tar.gz
 GZIP_ENV = --best
-DIST_TARGETS = dist-gzip
 distuninstallcheck_listfiles = find . -type f -print
-am__distuninstallcheck_listfiles = $(distuninstallcheck_listfiles) \
-  | sed 's|^\./|$(prefix)/|' | grep -v '$(infodir)/dir$$'
 distcleancheck_listfiles = find . -type f -print
 ACLOCAL = @ACLOCAL@
 AMTAR = @AMTAR@
-AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
 AUTOCONF = @AUTOCONF@
 AUTOHEADER = @AUTOHEADER@
 AUTOMAKE = @AUTOMAKE@
@@ -240,7 +139,6 @@ PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
 PACKAGE_NAME = @PACKAGE_NAME@
 PACKAGE_STRING = @PACKAGE_STRING@
 PACKAGE_TARNAME = @PACKAGE_TARNAME@
-PACKAGE_URL = @PACKAGE_URL@
 PACKAGE_VERSION = @PACKAGE_VERSION@
 PATH_SEPARATOR = @PATH_SEPARATOR@
 RANLIB = @RANLIB@
@@ -298,7 +196,7 @@ SUBDIRS = lib tests benchmarks
 all: all-recursive
 
 .SUFFIXES:
-am--refresh: Makefile
+am--refresh:
 	@:
 $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	@for dep in $?; do \
@@ -313,6 +211,7 @@ $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu Makefile'; \
 	$(am__cd) $(top_srcdir) && \
 	  $(AUTOMAKE) --gnu Makefile
+.PRECIOUS: Makefile
 Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
 	@case '$?' in \
 	  *config.status*) \
@@ -333,25 +232,22 @@ $(ACLOCAL_M4):  $(am__aclocal_m4_deps)
 $(am__aclocal_m4_deps):
 
 # This directory's subdirectories are mostly independent; you can cd
-# into them and run 'make' without going through this Makefile.
-# To change the values of 'make' variables: instead of editing Makefiles,
-# (1) if the variable is set in 'config.status', edit 'config.status'
-#     (which will cause the Makefiles to be regenerated when you run 'make');
-# (2) otherwise, pass the desired values on the 'make' command line.
-$(am__recursive_targets):
-	@fail=; \
-	if $(am__make_keepgoing); then \
-	  failcom='fail=yes'; \
-	else \
-	  failcom='exit 1'; \
-	fi; \
+# into them and run `make' without going through this Makefile.
+# To change the values of `make' variables: instead of editing Makefiles,
+# (1) if the variable is set in `config.status', edit `config.status'
+#     (which will cause the Makefiles to be regenerated when you run `make');
+# (2) otherwise, pass the desired values on the `make' command line.
+$(RECURSIVE_TARGETS):
+	@fail= failcom='exit 1'; \
+	for f in x $$MAKEFLAGS; do \
+	  case $$f in \
+	    *=* | --[!k]*);; \
+	    *k*) failcom='fail=yes';; \
+	  esac; \
+	done; \
 	dot_seen=no; \
 	target=`echo $@ | sed s/-recursive//`; \
-	case "$@" in \
-	  distclean-* | maintainer-clean-*) list='$(DIST_SUBDIRS)' ;; \
-	  *) list='$(SUBDIRS)' ;; \
-	esac; \
-	for subdir in $$list; do \
+	list='$(SUBDIRS)'; for subdir in $$list; do \
 	  echo "Making $$target in $$subdir"; \
 	  if test "$$subdir" = "."; then \
 	    dot_seen=yes; \
@@ -366,12 +262,57 @@ $(am__recursive_targets):
 	  $(MAKE) $(AM_MAKEFLAGS) "$$target-am" || exit 1; \
 	fi; test -z "$$fail"
 
-ID: $(am__tagged_files)
-	$(am__define_uniq_tagged_files); mkid -fID $$unique
-tags: tags-recursive
-TAGS: tags
+$(RECURSIVE_CLEAN_TARGETS):
+	@fail= failcom='exit 1'; \
+	for f in x $$MAKEFLAGS; do \
+	  case $$f in \
+	    *=* | --[!k]*);; \
+	    *k*) failcom='fail=yes';; \
+	  esac; \
+	done; \
+	dot_seen=no; \
+	case "$@" in \
+	  distclean-* | maintainer-clean-*) list='$(DIST_SUBDIRS)' ;; \
+	  *) list='$(SUBDIRS)' ;; \
+	esac; \
+	rev=''; for subdir in $$list; do \
+	  if test "$$subdir" = "."; then :; else \
+	    rev="$$subdir $$rev"; \
+	  fi; \
+	done; \
+	rev="$$rev ."; \
+	target=`echo $@ | sed s/-recursive//`; \
+	for subdir in $$rev; do \
+	  echo "Making $$target in $$subdir"; \
+	  if test "$$subdir" = "."; then \
+	    local_target="$$target-am"; \
+	  else \
+	    local_target="$$target"; \
+	  fi; \
+	  ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) $$local_target) \
+	  || eval $$failcom; \
+	done && test -z "$$fail"
+tags-recursive:
+	list='$(SUBDIRS)'; for subdir in $$list; do \
+	  test "$$subdir" = . || ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) tags); \
+	done
+ctags-recursive:
+	list='$(SUBDIRS)'; for subdir in $$list; do \
+	  test "$$subdir" = . || ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) ctags); \
+	done
 
-tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
+ID: $(HEADERS) $(SOURCES) $(LISP) $(TAGS_FILES)
+	list='$(SOURCES) $(HEADERS) $(LISP) $(TAGS_FILES)'; \
+	unique=`for i in $$list; do \
+	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+	  done | \
+	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
+	      END { if (nonempty) { for (i in files) print i; }; }'`; \
+	mkid -fID $$unique
+tags: TAGS
+
+TAGS: tags-recursive $(HEADERS) $(SOURCES)  $(TAGS_DEPENDENCIES) \
+		$(TAGS_FILES) $(LISP)
 	set x; \
 	here=`pwd`; \
 	if ($(ETAGS) --etags-include --version) >/dev/null 2>&1; then \
@@ -387,7 +328,12 @@ tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
 	      set "$$@" "$$include_option=$$here/$$subdir/TAGS"; \
 	  fi; \
 	done; \
-	$(am__define_uniq_tagged_files); \
+	list='$(SOURCES) $(HEADERS)  $(LISP) $(TAGS_FILES)'; \
+	unique=`for i in $$list; do \
+	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+	  done | \
+	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
+	      END { if (nonempty) { for (i in files) print i; }; }'`; \
 	shift; \
 	if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
 	  test -n "$$unique" || unique=$$empty_fix; \
@@ -399,11 +345,15 @@ tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
 	      $$unique; \
 	  fi; \
 	fi
-ctags: ctags-recursive
-
-CTAGS: ctags
-ctags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
-	$(am__define_uniq_tagged_files); \
+ctags: CTAGS
+CTAGS: ctags-recursive $(HEADERS) $(SOURCES)  $(TAGS_DEPENDENCIES) \
+		$(TAGS_FILES) $(LISP)
+	list='$(SOURCES) $(HEADERS)  $(LISP) $(TAGS_FILES)'; \
+	unique=`for i in $$list; do \
+	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+	  done | \
+	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
+	      END { if (nonempty) { for (i in files) print i; }; }'`; \
 	test -z "$(CTAGS_ARGS)$$unique" \
 	  || $(CTAGS) $(CTAGSFLAGS) $(AM_CTAGSFLAGS) $(CTAGS_ARGS) \
 	     $$unique
@@ -412,31 +362,9 @@ GTAGS:
 	here=`$(am__cd) $(top_builddir) && pwd` \
 	  && $(am__cd) $(top_srcdir) \
 	  && gtags -i $(GTAGS_ARGS) "$$here"
-cscope: cscope.files
-	test ! -s cscope.files \
-	  || $(CSCOPE) -b -q $(AM_CSCOPEFLAGS) $(CSCOPEFLAGS) -i cscope.files $(CSCOPE_ARGS)
-clean-cscope:
-	-rm -f cscope.files
-cscope.files: clean-cscope cscopelist
-cscopelist: cscopelist-recursive
-
-cscopelist-am: $(am__tagged_files)
-	list='$(am__tagged_files)'; \
-	case "$(srcdir)" in \
-	  [\\/]* | ?:[\\/]*) sdir="$(srcdir)" ;; \
-	  *) sdir=$(subdir)/$(srcdir) ;; \
-	esac; \
-	for i in $$list; do \
-	  if test -f "$$i"; then \
-	    echo "$(subdir)/$$i"; \
-	  else \
-	    echo "$$sdir/$$i"; \
-	  fi; \
-	done >> $(top_builddir)/cscope.files
 
 distclean-tags:
 	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
-	-rm -f cscope.out cscope.in.out cscope.po.out cscope.files
 
 distdir: $(DISTFILES)
 	$(am__remove_distdir)
@@ -472,10 +400,13 @@ distdir: $(DISTFILES)
 	done
 	@list='$(DIST_SUBDIRS)'; for subdir in $$list; do \
 	  if test "$$subdir" = .; then :; else \
-	    $(am__make_dryrun) \
-	      || test -d "$(distdir)/$$subdir" \
-	      || $(MKDIR_P) "$(distdir)/$$subdir" \
-	      || exit 1; \
+	    test -d "$(distdir)/$$subdir" \
+	    || $(MKDIR_P) "$(distdir)/$$subdir" \
+	    || exit 1; \
+	  fi; \
+	done
+	@list='$(DIST_SUBDIRS)'; for subdir in $$list; do \
+	  if test "$$subdir" = .; then :; else \
 	    dir1=$$subdir; dir2="$(distdir)/$$subdir"; \
 	    $(am__relativize); \
 	    new_distdir=$$reldir; \
@@ -504,42 +435,36 @@ distdir: $(DISTFILES)
 	|| chmod -R a+r "$(distdir)"
 dist-gzip: distdir
 	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 
 dist-bzip2: distdir
-	tardir=$(distdir) && $(am__tar) | BZIP2=$${BZIP2--9} bzip2 -c >$(distdir).tar.bz2
-	$(am__post_remove_distdir)
+	tardir=$(distdir) && $(am__tar) | bzip2 -9 -c >$(distdir).tar.bz2
+	$(am__remove_distdir)
 
-dist-lzip: distdir
-	tardir=$(distdir) && $(am__tar) | lzip -c $${LZIP_OPT--9} >$(distdir).tar.lz
-	$(am__post_remove_distdir)
+dist-lzma: distdir
+	tardir=$(distdir) && $(am__tar) | lzma -9 -c >$(distdir).tar.lzma
+	$(am__remove_distdir)
 
 dist-xz: distdir
-	tardir=$(distdir) && $(am__tar) | XZ_OPT=$${XZ_OPT--e} xz -c >$(distdir).tar.xz
-	$(am__post_remove_distdir)
+	tardir=$(distdir) && $(am__tar) | xz -c >$(distdir).tar.xz
+	$(am__remove_distdir)
 
 dist-tarZ: distdir
-	@echo WARNING: "Support for distribution archives compressed with" \
-		       "legacy program 'compress' is deprecated." >&2
-	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	tardir=$(distdir) && $(am__tar) | compress -c >$(distdir).tar.Z
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 
 dist-shar: distdir
-	@echo WARNING: "Support for shar distribution archives is" \
-	               "deprecated." >&2
-	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	shar $(distdir) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).shar.gz
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 
 dist-zip: distdir
 	-rm -f $(distdir).zip
 	zip -rq $(distdir).zip $(distdir)
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 
-dist dist-all:
-	$(MAKE) $(AM_MAKEFLAGS) $(DIST_TARGETS) am__post_remove_distdir='@:'
-	$(am__post_remove_distdir)
+dist dist-all: distdir
+	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
+	$(am__remove_distdir)
 
 # This target untars the dist file and tries a VPATH configuration.  Then
 # it guarantees that the distribution is self-contained by making another
@@ -550,8 +475,8 @@ distcheck: dist
 	  GZIP=$(GZIP_ENV) gzip -dc $(distdir).tar.gz | $(am__untar) ;;\
 	*.tar.bz2*) \
 	  bzip2 -dc $(distdir).tar.bz2 | $(am__untar) ;;\
-	*.tar.lz*) \
-	  lzip -dc $(distdir).tar.lz | $(am__untar) ;;\
+	*.tar.lzma*) \
+	  lzma -dc $(distdir).tar.lzma | $(am__untar) ;;\
 	*.tar.xz*) \
 	  xz -dc $(distdir).tar.xz | $(am__untar) ;;\
 	*.tar.Z*) \
@@ -561,19 +486,17 @@ distcheck: dist
 	*.zip*) \
 	  unzip $(distdir).zip ;;\
 	esac
-	chmod -R a-w $(distdir)
-	chmod u+w $(distdir)
-	mkdir $(distdir)/_build $(distdir)/_build/sub $(distdir)/_inst
+	chmod -R a-w $(distdir); chmod u+w $(distdir)
+	mkdir $(distdir)/_build
+	mkdir $(distdir)/_inst
 	chmod a-w $(distdir)
 	test -d $(distdir)/_build || exit 0; \
 	dc_install_base=`$(am__cd) $(distdir)/_inst && pwd | sed -e 's,^[^:\\/]:[\\/],/,'` \
 	  && dc_destdir="$${TMPDIR-/tmp}/am-dc-$$$$/" \
 	  && am__cwd=`pwd` \
-	  && $(am__cd) $(distdir)/_build/sub \
-	  && ../../configure \
-	    $(AM_DISTCHECK_CONFIGURE_FLAGS) \
+	  && $(am__cd) $(distdir)/_build \
+	  && ../configure --srcdir=.. --prefix="$$dc_install_base" \
 	    $(DISTCHECK_CONFIGURE_FLAGS) \
-	    --srcdir=../.. --prefix="$$dc_install_base" \
 	  && $(MAKE) $(AM_MAKEFLAGS) \
 	  && $(MAKE) $(AM_MAKEFLAGS) dvi \
 	  && $(MAKE) $(AM_MAKEFLAGS) check \
@@ -596,21 +519,13 @@ distcheck: dist
 	  && $(MAKE) $(AM_MAKEFLAGS) distcleancheck \
 	  && cd "$$am__cwd" \
 	  || exit 1
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 	@(echo "$(distdir) archives ready for distribution: "; \
 	  list='$(DIST_ARCHIVES)'; for i in $$list; do echo $$i; done) | \
 	  sed -e 1h -e 1s/./=/g -e 1p -e 1x -e '$$p' -e '$$x'
 distuninstallcheck:
-	@test -n '$(distuninstallcheck_dir)' || { \
-	  echo 'ERROR: trying to run $@ with an empty' \
-	       '$$(distuninstallcheck_dir)' >&2; \
-	  exit 1; \
-	}; \
-	$(am__cd) '$(distuninstallcheck_dir)' || { \
-	  echo 'ERROR: cannot chdir into $(distuninstallcheck_dir)' >&2; \
-	  exit 1; \
-	}; \
-	test `$(am__distuninstallcheck_listfiles) | wc -l` -eq 0 \
+	@$(am__cd) '$(distuninstallcheck_dir)' \
+	&& test `$(distuninstallcheck_listfiles) | wc -l` -le 1 \
 	   || { echo "ERROR: files left after uninstall:" ; \
 	        if test -n "$(DESTDIR)"; then \
 	          echo "  (check DESTDIR support)"; \
@@ -641,15 +556,10 @@ install-am: all-am
 
 installcheck: installcheck-recursive
 install-strip:
-	if test -z '$(STRIP)'; then \
-	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
-	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
-	      install; \
-	else \
-	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
-	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
-	    "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'" install; \
-	fi
+	$(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
+	  install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
+	  `test -z '$(STRIP)' || \
+	    echo "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'"` install
 mostlyclean-generic:
 
 clean-generic:
@@ -730,24 +640,23 @@ ps-am:
 
 uninstall-am:
 
-.MAKE: $(am__recursive_targets) install-am install-strip
+.MAKE: $(RECURSIVE_CLEAN_TARGETS) $(RECURSIVE_TARGETS) ctags-recursive \
+	install-am install-strip tags-recursive
 
-.PHONY: $(am__recursive_targets) CTAGS GTAGS TAGS all all-am \
-	am--refresh check check-am clean clean-cscope clean-generic \
-	cscope cscopelist-am ctags ctags-am dist dist-all dist-bzip2 \
-	dist-gzip dist-lzip dist-shar dist-tarZ dist-xz dist-zip \
-	distcheck distclean distclean-generic distclean-tags \
-	distcleancheck distdir distuninstallcheck dvi dvi-am html \
-	html-am info info-am install install-am install-data \
-	install-data-am install-dvi install-dvi-am install-exec \
-	install-exec-am install-html install-html-am install-info \
-	install-info-am install-man install-pdf install-pdf-am \
-	install-ps install-ps-am install-strip installcheck \
-	installcheck-am installdirs installdirs-am maintainer-clean \
+.PHONY: $(RECURSIVE_CLEAN_TARGETS) $(RECURSIVE_TARGETS) CTAGS GTAGS \
+	all all-am am--refresh check check-am clean clean-generic \
+	ctags ctags-recursive dist dist-all dist-bzip2 dist-gzip \
+	dist-lzma dist-shar dist-tarZ dist-xz dist-zip distcheck \
+	distclean distclean-generic distclean-tags distcleancheck \
+	distdir distuninstallcheck dvi dvi-am html html-am info \
+	info-am install install-am install-data install-data-am \
+	install-dvi install-dvi-am install-exec install-exec-am \
+	install-html install-html-am install-info install-info-am \
+	install-man install-pdf install-pdf-am install-ps \
+	install-ps-am install-strip installcheck installcheck-am \
+	installdirs installdirs-am maintainer-clean \
 	maintainer-clean-generic mostlyclean mostlyclean-generic pdf \
-	pdf-am ps ps-am tags tags-am uninstall uninstall-am
-
-.PRECIOUS: Makefile
+	pdf-am ps ps-am tags tags-recursive uninstall uninstall-am
 
 
 # Tell versions [3.59,3.63) of GNU make to not export all variables.
diff --git a/TODO b/TODO
index ef3940ea..ee1dd897 100644
--- a/TODO
+++ b/TODO
@@ -2,9 +2,7 @@
   - use protocol buffers? replace xmlReader/Writer ec..
   - Binary use htonll, htonl
 
-* Stencil -- do the permute for locally permuted in halo exchange.
-  - BUG cshift mpi; the "s" indexing is weird in the Cshift_comms_simd
-    as simd_layout or not is confusing
+* Stencil -- DONE
 
 * Reduce implemention is poor
 * Bug in SeedFixedIntegers gives same output on each site.
diff --git a/aclocal.m4 b/aclocal.m4
index f3018f6c..bcc213b4 100644
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -1,7 +1,7 @@
-# generated automatically by aclocal 1.15 -*- Autoconf -*-
-
-# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+# generated automatically by aclocal 1.11.1 -*- Autoconf -*-
 
+# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
+# 2005, 2006, 2007, 2008, 2009  Free Software Foundation, Inc.
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
@@ -11,16 +11,15 @@
 # even the implied warranty of MERCHANTABILITY or FITNESS FOR A
 # PARTICULAR PURPOSE.
 
-m4_ifndef([AC_CONFIG_MACRO_DIRS], [m4_defun([_AM_CONFIG_MACRO_DIRS], [])m4_defun([AC_CONFIG_MACRO_DIRS], [_AM_CONFIG_MACRO_DIRS($@)])])
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
-m4_if(m4_defn([AC_AUTOCONF_VERSION]), [2.69],,
-[m4_warning([this file was generated for autoconf 2.69.
+m4_if(m4_defn([AC_AUTOCONF_VERSION]), [2.63],,
+[m4_warning([this file was generated for autoconf 2.63.
 You have another version of autoconf.  It may work, but is not guaranteed to.
 If you have problems, you may need to regenerate the build system entirely.
-To do so, use the procedure documented by the package, typically 'autoreconf'.])])
+To do so, use the procedure documented by the package, typically `autoreconf'.])])
 
-# Copyright (C) 2002-2014 Free Software Foundation, Inc.
+# Copyright (C) 2002, 2003, 2005, 2006, 2007, 2008  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -32,10 +31,10 @@ To do so, use the procedure documented by the package, typically 'autoreconf'.])
 # generated from the m4 files accompanying Automake X.Y.
 # (This private macro should not be called outside this file.)
 AC_DEFUN([AM_AUTOMAKE_VERSION],
-[am__api_version='1.15'
+[am__api_version='1.11'
 dnl Some users find AM_AUTOMAKE_VERSION and mistake it for a way to
 dnl require some minimum version.  Point them to the right macro.
-m4_if([$1], [1.15], [],
+m4_if([$1], [1.11.1], [],
       [AC_FATAL([Do not call $0, use AM_INIT_AUTOMAKE([$1]).])])dnl
 ])
 
@@ -51,22 +50,22 @@ m4_define([_AM_AUTOCONF_VERSION], [])
 # Call AM_AUTOMAKE_VERSION and AM_AUTOMAKE_VERSION so they can be traced.
 # This function is AC_REQUIREd by AM_INIT_AUTOMAKE.
 AC_DEFUN([AM_SET_CURRENT_AUTOMAKE_VERSION],
-[AM_AUTOMAKE_VERSION([1.15])dnl
+[AM_AUTOMAKE_VERSION([1.11.1])dnl
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
 _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 
 # AM_AUX_DIR_EXPAND                                         -*- Autoconf -*-
 
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2001, 2003, 2005  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
 # For projects using AC_CONFIG_AUX_DIR([foo]), Autoconf sets
-# $ac_aux_dir to '$srcdir/foo'.  In other projects, it is set to
-# '$srcdir', '$srcdir/..', or '$srcdir/../..'.
+# $ac_aux_dir to `$srcdir/foo'.  In other projects, it is set to
+# `$srcdir', `$srcdir/..', or `$srcdir/../..'.
 #
 # Of course, Automake must honor this variable whenever it calls a
 # tool from the auxiliary directory.  The problem is that $srcdir (and
@@ -85,7 +84,7 @@ _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 #
 # The reason of the latter failure is that $top_srcdir and $ac_aux_dir
 # are both prefixed by $srcdir.  In an in-source build this is usually
-# harmless because $srcdir is '.', but things will broke when you
+# harmless because $srcdir is `.', but things will broke when you
 # start a VPATH build or use an absolute $srcdir.
 #
 # So we could use something similar to $top_srcdir/$ac_aux_dir/missing,
@@ -103,26 +102,30 @@ _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 # configured tree to be moved without reconfiguration.
 
 AC_DEFUN([AM_AUX_DIR_EXPAND],
-[AC_REQUIRE([AC_CONFIG_AUX_DIR_DEFAULT])dnl
-# Expand $ac_aux_dir to an absolute path.
-am_aux_dir=`cd "$ac_aux_dir" && pwd`
+[dnl Rely on autoconf to set up CDPATH properly.
+AC_PREREQ([2.50])dnl
+# expand $ac_aux_dir to an absolute path
+am_aux_dir=`cd $ac_aux_dir && pwd`
 ])
 
 # AM_CONDITIONAL                                            -*- Autoconf -*-
 
-# Copyright (C) 1997-2014 Free Software Foundation, Inc.
+# Copyright (C) 1997, 2000, 2001, 2003, 2004, 2005, 2006, 2008
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 9
+
 # AM_CONDITIONAL(NAME, SHELL-CONDITION)
 # -------------------------------------
 # Define a conditional.
 AC_DEFUN([AM_CONDITIONAL],
-[AC_PREREQ([2.52])dnl
- m4_if([$1], [TRUE],  [AC_FATAL([$0: invalid condition: $1])],
-       [$1], [FALSE], [AC_FATAL([$0: invalid condition: $1])])dnl
+[AC_PREREQ(2.52)dnl
+ ifelse([$1], [TRUE],  [AC_FATAL([$0: invalid condition: $1])],
+	[$1], [FALSE], [AC_FATAL([$0: invalid condition: $1])])dnl
 AC_SUBST([$1_TRUE])dnl
 AC_SUBST([$1_FALSE])dnl
 _AM_SUBST_NOTMAKE([$1_TRUE])dnl
@@ -141,14 +144,16 @@ AC_CONFIG_COMMANDS_PRE(
 Usually this means the macro was only invoked conditionally.]])
 fi])])
 
-# Copyright (C) 1999-2014 Free Software Foundation, Inc.
+# Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2009
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 10
 
-# There are a few dirty hacks below to avoid letting 'AC_PROG_CC' be
+# There are a few dirty hacks below to avoid letting `AC_PROG_CC' be
 # written in clear, in which case automake, when reading aclocal.m4,
 # will think it sees a *use*, and therefore will trigger all it's
 # C support machinery.  Also note that it means that autoscan, seeing
@@ -158,7 +163,7 @@ fi])])
 # _AM_DEPENDENCIES(NAME)
 # ----------------------
 # See how the compiler implements dependency checking.
-# NAME is "CC", "CXX", "OBJC", "OBJCXX", "UPC", or "GJC".
+# NAME is "CC", "CXX", "GCJ", or "OBJC".
 # We try a few techniques and use that to set a single cache variable.
 #
 # We don't AC_REQUIRE the corresponding AC_PROG_CC since the latter was
@@ -171,13 +176,12 @@ AC_REQUIRE([AM_OUTPUT_DEPENDENCY_COMMANDS])dnl
 AC_REQUIRE([AM_MAKE_INCLUDE])dnl
 AC_REQUIRE([AM_DEP_TRACK])dnl
 
-m4_if([$1], [CC],   [depcc="$CC"   am_compiler_list=],
-      [$1], [CXX],  [depcc="$CXX"  am_compiler_list=],
-      [$1], [OBJC], [depcc="$OBJC" am_compiler_list='gcc3 gcc'],
-      [$1], [OBJCXX], [depcc="$OBJCXX" am_compiler_list='gcc3 gcc'],
-      [$1], [UPC],  [depcc="$UPC"  am_compiler_list=],
-      [$1], [GCJ],  [depcc="$GCJ"  am_compiler_list='gcc3 gcc'],
-                    [depcc="$$1"   am_compiler_list=])
+ifelse([$1], CC,   [depcc="$CC"   am_compiler_list=],
+       [$1], CXX,  [depcc="$CXX"  am_compiler_list=],
+       [$1], OBJC, [depcc="$OBJC" am_compiler_list='gcc3 gcc'],
+       [$1], UPC,  [depcc="$UPC"  am_compiler_list=],
+       [$1], GCJ,  [depcc="$GCJ"  am_compiler_list='gcc3 gcc'],
+                   [depcc="$$1"   am_compiler_list=])
 
 AC_CACHE_CHECK([dependency style of $depcc],
                [am_cv_$1_dependencies_compiler_type],
@@ -185,9 +189,8 @@ AC_CACHE_CHECK([dependency style of $depcc],
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named 'D' -- because '-MD' means "put the output
-  # in D".
-  rm -rf conftest.dir
+  # making a dummy file named `D' -- because `-MD' means `put the output
+  # in D'.
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -226,16 +229,16 @@ AC_CACHE_CHECK([dependency style of $depcc],
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
-      # Solaris 10 /bin/sh.
-      echo '/* dummy */' > sub/conftst$i.h
+      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
+      # Solaris 8's {/usr,}/bin/sh.
+      touch sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with '-c' and '-o' for the sake of the "dashmstdout"
+    # We check with `-c' and `-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle '-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs.
+    # handle `-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -244,16 +247,16 @@ AC_CACHE_CHECK([dependency style of $depcc],
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # After this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested.
+      # after this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
-      # This compiler won't grok '-c -o', but also, the minuso test has
+    msvisualcpp | msvcmsys)
+      # This compiler won't grok `-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -301,7 +304,7 @@ AM_CONDITIONAL([am__fastdep$1], [
 # AM_SET_DEPDIR
 # -------------
 # Choose a directory name for dependency files.
-# This macro is AC_REQUIREd in _AM_DEPENDENCIES.
+# This macro is AC_REQUIREd in _AM_DEPENDENCIES
 AC_DEFUN([AM_SET_DEPDIR],
 [AC_REQUIRE([AM_SET_LEADING_DOT])dnl
 AC_SUBST([DEPDIR], ["${am__leading_dot}deps"])dnl
@@ -311,39 +314,34 @@ AC_SUBST([DEPDIR], ["${am__leading_dot}deps"])dnl
 # AM_DEP_TRACK
 # ------------
 AC_DEFUN([AM_DEP_TRACK],
-[AC_ARG_ENABLE([dependency-tracking], [dnl
-AS_HELP_STRING(
-  [--enable-dependency-tracking],
-  [do not reject slow dependency extractors])
-AS_HELP_STRING(
-  [--disable-dependency-tracking],
-  [speeds up one-time build])])
+[AC_ARG_ENABLE(dependency-tracking,
+[  --disable-dependency-tracking  speeds up one-time build
+  --enable-dependency-tracking   do not reject slow dependency extractors])
 if test "x$enable_dependency_tracking" != xno; then
   am_depcomp="$ac_aux_dir/depcomp"
   AMDEPBACKSLASH='\'
-  am__nodep='_no'
 fi
 AM_CONDITIONAL([AMDEP], [test "x$enable_dependency_tracking" != xno])
 AC_SUBST([AMDEPBACKSLASH])dnl
 _AM_SUBST_NOTMAKE([AMDEPBACKSLASH])dnl
-AC_SUBST([am__nodep])dnl
-_AM_SUBST_NOTMAKE([am__nodep])dnl
 ])
 
 # Generate code to set up dependency tracking.              -*- Autoconf -*-
 
-# Copyright (C) 1999-2014 Free Software Foundation, Inc.
+# Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2008
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+#serial 5
 
 # _AM_OUTPUT_DEPENDENCY_COMMANDS
 # ------------------------------
 AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
 [{
-  # Older Autoconf quotes --file arguments for eval, but not when files
+  # Autoconf 2.62 quotes --file arguments for eval, but not when files
   # are listed without --file.  Let's play safe and only enable the eval
   # if we detect the quoting.
   case $CONFIG_FILES in
@@ -356,7 +354,7 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
     # Strip MF so we end up with the name of the file.
     mf=`echo "$mf" | sed -e 's/:.*$//'`
     # Check whether this is an Automake generated Makefile or not.
-    # We used to match only the files named 'Makefile.in', but
+    # We used to match only the files named `Makefile.in', but
     # some people rename them; so instead we look at the file content.
     # Grep'ing the first line is not enough: some people post-process
     # each Makefile.in and add a new line on top of each file to say so.
@@ -368,19 +366,21 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
       continue
     fi
     # Extract the definition of DEPDIR, am__include, and am__quote
-    # from the Makefile without running 'make'.
+    # from the Makefile without running `make'.
     DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
     test -z "$DEPDIR" && continue
     am__include=`sed -n 's/^am__include = //p' < "$mf"`
-    test -z "$am__include" && continue
+    test -z "am__include" && continue
     am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
+    # When using ansi2knr, U may be empty or an underscore; expand it
+    U=`sed -n 's/^U = //p' < "$mf"`
     # Find all dependency output files, they are included files with
     # $(DEPDIR) in their names.  We invoke sed twice because it is the
     # simplest approach to changing $(DEPDIR) to its actual value in the
     # expansion.
     for file in `sed -n "
       s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
-	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
+	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g' -e 's/\$U/'"$U"'/g'`; do
       # Make sure the directory exists.
       test -f "$dirpart/$file" && continue
       fdir=`AS_DIRNAME(["$file"])`
@@ -398,7 +398,7 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
 # This macro should only be invoked once -- use via AC_REQUIRE.
 #
 # This code is only required when automatic dependency tracking
-# is enabled.  FIXME.  This creates each '.P' file that we will
+# is enabled.  FIXME.  This creates each `.P' file that we will
 # need in order to bootstrap the dependency handling code.
 AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 [AC_CONFIG_COMMANDS([depfiles],
@@ -408,21 +408,18 @@ AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 
 # Do all the work for Automake.                             -*- Autoconf -*-
 
-# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
+# 2005, 2006, 2008, 2009 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 16
+
 # This macro actually does too much.  Some checks are only needed if
 # your package does certain things.  But this isn't really a big deal.
 
-dnl Redefine AC_PROG_CC to automatically invoke _AM_PROG_CC_C_O.
-m4_define([AC_PROG_CC],
-m4_defn([AC_PROG_CC])
-[_AM_PROG_CC_C_O
-])
-
 # AM_INIT_AUTOMAKE(PACKAGE, VERSION, [NO-DEFINE])
 # AM_INIT_AUTOMAKE([OPTIONS])
 # -----------------------------------------------
@@ -435,7 +432,7 @@ m4_defn([AC_PROG_CC])
 # arguments mandatory, and then we can depend on a new Autoconf
 # release and drop the old call support.
 AC_DEFUN([AM_INIT_AUTOMAKE],
-[AC_PREREQ([2.65])dnl
+[AC_PREREQ([2.62])dnl
 dnl Autoconf wants to disallow AM_ names.  We explicitly allow
 dnl the ones we care about.
 m4_pattern_allow([^AM_[A-Z]+FLAGS$])dnl
@@ -464,42 +461,33 @@ AC_SUBST([CYGPATH_W])
 # Define the identity of the package.
 dnl Distinguish between old-style and new-style calls.
 m4_ifval([$2],
-[AC_DIAGNOSE([obsolete],
-             [$0: two- and three-arguments forms are deprecated.])
-m4_ifval([$3], [_AM_SET_OPTION([no-define])])dnl
+[m4_ifval([$3], [_AM_SET_OPTION([no-define])])dnl
  AC_SUBST([PACKAGE], [$1])dnl
  AC_SUBST([VERSION], [$2])],
 [_AM_SET_OPTIONS([$1])dnl
 dnl Diagnose old-style AC_INIT with new-style AM_AUTOMAKE_INIT.
-m4_if(
-  m4_ifdef([AC_PACKAGE_NAME], [ok]):m4_ifdef([AC_PACKAGE_VERSION], [ok]),
-  [ok:ok],,
+m4_if(m4_ifdef([AC_PACKAGE_NAME], 1)m4_ifdef([AC_PACKAGE_VERSION], 1), 11,,
   [m4_fatal([AC_INIT should be called with package and version arguments])])dnl
  AC_SUBST([PACKAGE], ['AC_PACKAGE_TARNAME'])dnl
  AC_SUBST([VERSION], ['AC_PACKAGE_VERSION'])])dnl
 
 _AM_IF_OPTION([no-define],,
-[AC_DEFINE_UNQUOTED([PACKAGE], ["$PACKAGE"], [Name of package])
- AC_DEFINE_UNQUOTED([VERSION], ["$VERSION"], [Version number of package])])dnl
+[AC_DEFINE_UNQUOTED(PACKAGE, "$PACKAGE", [Name of package])
+ AC_DEFINE_UNQUOTED(VERSION, "$VERSION", [Version number of package])])dnl
 
 # Some tools Automake needs.
 AC_REQUIRE([AM_SANITY_CHECK])dnl
 AC_REQUIRE([AC_ARG_PROGRAM])dnl
-AM_MISSING_PROG([ACLOCAL], [aclocal-${am__api_version}])
-AM_MISSING_PROG([AUTOCONF], [autoconf])
-AM_MISSING_PROG([AUTOMAKE], [automake-${am__api_version}])
-AM_MISSING_PROG([AUTOHEADER], [autoheader])
-AM_MISSING_PROG([MAKEINFO], [makeinfo])
+AM_MISSING_PROG(ACLOCAL, aclocal-${am__api_version})
+AM_MISSING_PROG(AUTOCONF, autoconf)
+AM_MISSING_PROG(AUTOMAKE, automake-${am__api_version})
+AM_MISSING_PROG(AUTOHEADER, autoheader)
+AM_MISSING_PROG(MAKEINFO, makeinfo)
 AC_REQUIRE([AM_PROG_INSTALL_SH])dnl
 AC_REQUIRE([AM_PROG_INSTALL_STRIP])dnl
-AC_REQUIRE([AC_PROG_MKDIR_P])dnl
-# For better backward compatibility.  To be removed once Automake 1.9.x
-# dies out for good.  For more background, see:
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
-AC_SUBST([mkdir_p], ['$(MKDIR_P)'])
-# We need awk for the "check" target (and possibly the TAP driver).  The
-# system "awk" is bad on some platforms.
+AC_REQUIRE([AM_PROG_MKDIR_P])dnl
+# We need awk for the "check" target.  The system "awk" is bad on
+# some platforms.
 AC_REQUIRE([AC_PROG_AWK])dnl
 AC_REQUIRE([AC_PROG_MAKE_SET])dnl
 AC_REQUIRE([AM_SET_LEADING_DOT])dnl
@@ -508,82 +496,34 @@ _AM_IF_OPTION([tar-ustar], [_AM_PROG_TAR([ustar])],
 			     [_AM_PROG_TAR([v7])])])
 _AM_IF_OPTION([no-dependencies],,
 [AC_PROVIDE_IFELSE([AC_PROG_CC],
-		  [_AM_DEPENDENCIES([CC])],
-		  [m4_define([AC_PROG_CC],
-			     m4_defn([AC_PROG_CC])[_AM_DEPENDENCIES([CC])])])dnl
+		  [_AM_DEPENDENCIES(CC)],
+		  [define([AC_PROG_CC],
+			  defn([AC_PROG_CC])[_AM_DEPENDENCIES(CC)])])dnl
 AC_PROVIDE_IFELSE([AC_PROG_CXX],
-		  [_AM_DEPENDENCIES([CXX])],
-		  [m4_define([AC_PROG_CXX],
-			     m4_defn([AC_PROG_CXX])[_AM_DEPENDENCIES([CXX])])])dnl
+		  [_AM_DEPENDENCIES(CXX)],
+		  [define([AC_PROG_CXX],
+			  defn([AC_PROG_CXX])[_AM_DEPENDENCIES(CXX)])])dnl
 AC_PROVIDE_IFELSE([AC_PROG_OBJC],
-		  [_AM_DEPENDENCIES([OBJC])],
-		  [m4_define([AC_PROG_OBJC],
-			     m4_defn([AC_PROG_OBJC])[_AM_DEPENDENCIES([OBJC])])])dnl
-AC_PROVIDE_IFELSE([AC_PROG_OBJCXX],
-		  [_AM_DEPENDENCIES([OBJCXX])],
-		  [m4_define([AC_PROG_OBJCXX],
-			     m4_defn([AC_PROG_OBJCXX])[_AM_DEPENDENCIES([OBJCXX])])])dnl
+		  [_AM_DEPENDENCIES(OBJC)],
+		  [define([AC_PROG_OBJC],
+			  defn([AC_PROG_OBJC])[_AM_DEPENDENCIES(OBJC)])])dnl
 ])
-AC_REQUIRE([AM_SILENT_RULES])dnl
-dnl The testsuite driver may need to know about EXEEXT, so add the
-dnl 'am__EXEEXT' conditional if _AM_COMPILER_EXEEXT was seen.  This
-dnl macro is hooked onto _AC_COMPILER_EXEEXT early, see below.
+_AM_IF_OPTION([silent-rules], [AC_REQUIRE([AM_SILENT_RULES])])dnl
+dnl The `parallel-tests' driver may need to know about EXEEXT, so add the
+dnl `am__EXEEXT' conditional if _AM_COMPILER_EXEEXT was seen.  This macro
+dnl is hooked onto _AC_COMPILER_EXEEXT early, see below.
 AC_CONFIG_COMMANDS_PRE(dnl
 [m4_provide_if([_AM_COMPILER_EXEEXT],
   [AM_CONDITIONAL([am__EXEEXT], [test -n "$EXEEXT"])])])dnl
-
-# POSIX will say in a future version that running "rm -f" with no argument
-# is OK; and we want to be able to make that assumption in our Makefile
-# recipes.  So use an aggressive probe to check that the usage we want is
-# actually supported "in the wild" to an acceptable degree.
-# See automake bug#10828.
-# To make any issue more visible, cause the running configure to be aborted
-# by default if the 'rm' program in use doesn't match our expectations; the
-# user can still override this though.
-if rm -f && rm -fr && rm -rf; then : OK; else
-  cat >&2 <<'END'
-Oops!
-
-Your 'rm' program seems unable to run without file operands specified
-on the command line, even when the '-f' option is present.  This is contrary
-to the behaviour of most rm programs out there, and not conforming with
-the upcoming POSIX standard: <http://austingroupbugs.net/view.php?id=542>
-
-Please tell bug-automake@gnu.org about your system, including the value
-of your $PATH and any error possibly output before this message.  This
-can help us improve future automake versions.
-
-END
-  if test x"$ACCEPT_INFERIOR_RM_PROGRAM" = x"yes"; then
-    echo 'Configuration will proceed anyway, since you have set the' >&2
-    echo 'ACCEPT_INFERIOR_RM_PROGRAM variable to "yes"' >&2
-    echo >&2
-  else
-    cat >&2 <<'END'
-Aborting the configuration process, to ensure you take notice of the issue.
-
-You can download and install GNU coreutils to get an 'rm' implementation
-that behaves properly: <http://www.gnu.org/software/coreutils/>.
-
-If you want to complete the configuration process using your problematic
-'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
-to "yes", and re-run configure.
-
-END
-    AC_MSG_ERROR([Your 'rm' program is bad, sorry.])
-  fi
-fi
-dnl The trailing newline in this macro's definition is deliberate, for
-dnl backward compatibility and to allow trailing 'dnl'-style comments
-dnl after the AM_INIT_AUTOMAKE invocation. See automake bug#16841.
 ])
 
-dnl Hook into '_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
+dnl Hook into `_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
 dnl add the conditional right here, as _AC_COMPILER_EXEEXT may be further
 dnl mangled by Autoconf and run in a shell conditional statement.
 m4_define([_AC_COMPILER_EXEEXT],
 m4_defn([_AC_COMPILER_EXEEXT])[m4_provide([_AM_COMPILER_EXEEXT])])
 
+
 # When config.status generates a header, we must update the stamp-h file.
 # This file resides in the same directory as the config header
 # that is generated.  The stamp files are numbered to have different names.
@@ -605,7 +545,7 @@ for _am_header in $config_headers :; do
 done
 echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_count])
 
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2001, 2003, 2005, 2008  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -616,7 +556,7 @@ echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_co
 # Define $install_sh.
 AC_DEFUN([AM_PROG_INSTALL_SH],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
-if test x"${install_sh+set}" != xset; then
+if test x"${install_sh}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -624,14 +564,16 @@ if test x"${install_sh+set}" != xset; then
     install_sh="\${SHELL} $am_aux_dir/install-sh"
   esac
 fi
-AC_SUBST([install_sh])])
+AC_SUBST(install_sh)])
 
-# Copyright (C) 2003-2014 Free Software Foundation, Inc.
+# Copyright (C) 2003, 2005  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 2
+
 # Check whether the underlying file-system supports filenames
 # with a leading dot.  For instance MS-DOS doesn't.
 AC_DEFUN([AM_SET_LEADING_DOT],
@@ -647,12 +589,14 @@ AC_SUBST([am__leading_dot])])
 
 # Check to see how 'make' treats includes.	            -*- Autoconf -*-
 
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2001, 2002, 2003, 2005, 2009  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 4
+
 # AM_MAKE_INCLUDE()
 # -----------------
 # Check to see how make treats includes.
@@ -670,7 +614,7 @@ am__quote=
 _am_result=none
 # First try GNU make style include.
 echo "include confinc" > confmf
-# Ignore all kinds of additional output from 'make'.
+# Ignore all kinds of additional output from `make'.
 case `$am_make -s -f confmf 2> /dev/null` in #(
 *the\ am__doit\ target*)
   am__include=include
@@ -697,12 +641,15 @@ rm -f confinc confmf
 
 # Fake the existence of programs that GNU maintainers use.  -*- Autoconf -*-
 
-# Copyright (C) 1997-2014 Free Software Foundation, Inc.
+# Copyright (C) 1997, 1999, 2000, 2001, 2003, 2004, 2005, 2008
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 6
+
 # AM_MISSING_PROG(NAME, PROGRAM)
 # ------------------------------
 AC_DEFUN([AM_MISSING_PROG],
@@ -710,10 +657,11 @@ AC_DEFUN([AM_MISSING_PROG],
 $1=${$1-"${am_missing_run}$2"}
 AC_SUBST($1)])
 
+
 # AM_MISSING_HAS_RUN
 # ------------------
-# Define MISSING if not defined so far and test if it is modern enough.
-# If it is, set am_missing_run to use it, otherwise, to nothing.
+# Define MISSING if not defined so far and test if it supports --run.
+# If it does, set am_missing_run to use it, otherwise, to nothing.
 AC_DEFUN([AM_MISSING_HAS_RUN],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
 AC_REQUIRE_AUX_FILE([missing])dnl
@@ -726,35 +674,63 @@ if test x"${MISSING+set}" != xset; then
   esac
 fi
 # Use eval to expand $SHELL
-if eval "$MISSING --is-lightweight"; then
-  am_missing_run="$MISSING "
+if eval "$MISSING --run true"; then
+  am_missing_run="$MISSING --run "
 else
   am_missing_run=
-  AC_MSG_WARN(['missing' script is too old or missing])
+  AC_MSG_WARN([`missing' script is too old or missing])
 fi
 ])
 
-# Helper functions for option handling.                     -*- Autoconf -*-
-
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2003, 2004, 2005, 2006  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# AM_PROG_MKDIR_P
+# ---------------
+# Check for `mkdir -p'.
+AC_DEFUN([AM_PROG_MKDIR_P],
+[AC_PREREQ([2.60])dnl
+AC_REQUIRE([AC_PROG_MKDIR_P])dnl
+dnl Automake 1.8 to 1.9.6 used to define mkdir_p.  We now use MKDIR_P,
+dnl while keeping a definition of mkdir_p for backward compatibility.
+dnl @MKDIR_P@ is magic: AC_OUTPUT adjusts its value for each Makefile.
+dnl However we cannot define mkdir_p as $(MKDIR_P) for the sake of
+dnl Makefile.ins that do not define MKDIR_P, so we do our own
+dnl adjustment using top_builddir (which is defined more often than
+dnl MKDIR_P).
+AC_SUBST([mkdir_p], ["$MKDIR_P"])dnl
+case $mkdir_p in
+  [[\\/$]]* | ?:[[\\/]]*) ;;
+  */*) mkdir_p="\$(top_builddir)/$mkdir_p" ;;
+esac
+])
+
+# Helper functions for option handling.                     -*- Autoconf -*-
+
+# Copyright (C) 2001, 2002, 2003, 2005, 2008  Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# serial 4
+
 # _AM_MANGLE_OPTION(NAME)
 # -----------------------
 AC_DEFUN([_AM_MANGLE_OPTION],
 [[_AM_OPTION_]m4_bpatsubst($1, [[^a-zA-Z0-9_]], [_])])
 
 # _AM_SET_OPTION(NAME)
-# --------------------
+# ------------------------------
 # Set option NAME.  Presently that only means defining a flag for this option.
 AC_DEFUN([_AM_SET_OPTION],
-[m4_define(_AM_MANGLE_OPTION([$1]), [1])])
+[m4_define(_AM_MANGLE_OPTION([$1]), 1)])
 
 # _AM_SET_OPTIONS(OPTIONS)
-# ------------------------
+# ----------------------------------
 # OPTIONS is a space-separated list of Automake options.
 AC_DEFUN([_AM_SET_OPTIONS],
 [m4_foreach_w([_AM_Option], [$1], [_AM_SET_OPTION(_AM_Option)])])
@@ -765,82 +741,24 @@ AC_DEFUN([_AM_SET_OPTIONS],
 AC_DEFUN([_AM_IF_OPTION],
 [m4_ifset(_AM_MANGLE_OPTION([$1]), [$2], [$3])])
 
-# Copyright (C) 1999-2014 Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# _AM_PROG_CC_C_O
-# ---------------
-# Like AC_PROG_CC_C_O, but changed for automake.  We rewrite AC_PROG_CC
-# to automatically call this.
-AC_DEFUN([_AM_PROG_CC_C_O],
-[AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
-AC_REQUIRE_AUX_FILE([compile])dnl
-AC_LANG_PUSH([C])dnl
-AC_CACHE_CHECK(
-  [whether $CC understands -c and -o together],
-  [am_cv_prog_cc_c_o],
-  [AC_LANG_CONFTEST([AC_LANG_PROGRAM([])])
-  # Make sure it works both with $CC and with simple cc.
-  # Following AC_PROG_CC_C_O, we do the test twice because some
-  # compilers refuse to overwrite an existing .o file with -o,
-  # though they will create one.
-  am_cv_prog_cc_c_o=yes
-  for am_i in 1 2; do
-    if AM_RUN_LOG([$CC -c conftest.$ac_ext -o conftest2.$ac_objext]) \
-         && test -f conftest2.$ac_objext; then
-      : OK
-    else
-      am_cv_prog_cc_c_o=no
-      break
-    fi
-  done
-  rm -f core conftest*
-  unset am_i])
-if test "$am_cv_prog_cc_c_o" != yes; then
-   # Losing compiler, so override with the script.
-   # FIXME: It is wrong to rewrite CC.
-   # But if we don't then we get into trouble of one sort or another.
-   # A longer-term fix would be to have automake use am__CC in this case,
-   # and then we could set am__CC="\$(top_srcdir)/compile \$(CC)"
-   CC="$am_aux_dir/compile $CC"
-fi
-AC_LANG_POP([C])])
-
-# For backward compatibility.
-AC_DEFUN_ONCE([AM_PROG_CC_C_O], [AC_REQUIRE([AC_PROG_CC])])
-
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# AM_RUN_LOG(COMMAND)
-# -------------------
-# Run COMMAND, save the exit status in ac_status, and log it.
-# (This has been adapted from Autoconf's _AC_RUN_LOG macro.)
-AC_DEFUN([AM_RUN_LOG],
-[{ echo "$as_me:$LINENO: $1" >&AS_MESSAGE_LOG_FD
-   ($1) >&AS_MESSAGE_LOG_FD 2>&AS_MESSAGE_LOG_FD
-   ac_status=$?
-   echo "$as_me:$LINENO: \$? = $ac_status" >&AS_MESSAGE_LOG_FD
-   (exit $ac_status); }])
-
 # Check to make sure that the build environment is sane.    -*- Autoconf -*-
 
-# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+# Copyright (C) 1996, 1997, 2000, 2001, 2003, 2005, 2008
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 5
+
 # AM_SANITY_CHECK
 # ---------------
 AC_DEFUN([AM_SANITY_CHECK],
 [AC_MSG_CHECKING([whether build environment is sane])
+# Just in case
+sleep 1
+echo timestamp > conftest.file
 # Reject unsafe characters in $srcdir or the absolute working directory
 # name.  Accept space and tab only in the latter.
 am_lf='
@@ -851,40 +769,32 @@ case `pwd` in
 esac
 case $srcdir in
   *[[\\\"\#\$\&\'\`$am_lf\ \	]]*)
-    AC_MSG_ERROR([unsafe srcdir value: '$srcdir']);;
+    AC_MSG_ERROR([unsafe srcdir value: `$srcdir']);;
 esac
 
-# Do 'set' in a subshell so we don't clobber the current shell's
+# Do `set' in a subshell so we don't clobber the current shell's
 # arguments.  Must try -L first in case configure is actually a
 # symlink; some systems play weird games with the mod time of symlinks
 # (eg FreeBSD returns the mod time of the symlink's containing
 # directory).
 if (
-   am_has_slept=no
-   for am_try in 1 2; do
-     echo "timestamp, slept: $am_has_slept" > conftest.file
-     set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
-     if test "$[*]" = "X"; then
-	# -L didn't work.
-	set X `ls -t "$srcdir/configure" conftest.file`
-     fi
-     if test "$[*]" != "X $srcdir/configure conftest.file" \
-	&& test "$[*]" != "X conftest.file $srcdir/configure"; then
+   set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
+   if test "$[*]" = "X"; then
+      # -L didn't work.
+      set X `ls -t "$srcdir/configure" conftest.file`
+   fi
+   rm -f conftest.file
+   if test "$[*]" != "X $srcdir/configure conftest.file" \
+      && test "$[*]" != "X conftest.file $srcdir/configure"; then
+
+      # If neither matched, then we have a broken ls.  This can happen
+      # if, for instance, CONFIG_SHELL is bash and it inherits a
+      # broken ls alias from the environment.  This has actually
+      # happened.  Such a system could not be considered "sane".
+      AC_MSG_ERROR([ls -t appears to fail.  Make sure there is not a broken
+alias in your environment])
+   fi
 
-	# If neither matched, then we have a broken ls.  This can happen
-	# if, for instance, CONFIG_SHELL is bash and it inherits a
-	# broken ls alias from the environment.  This has actually
-	# happened.  Such a system could not be considered "sane".
-	AC_MSG_ERROR([ls -t appears to fail.  Make sure there is not a broken
-  alias in your environment])
-     fi
-     if test "$[2]" = conftest.file || test $am_try -eq 2; then
-       break
-     fi
-     # Just in case.
-     sleep 1
-     am_has_slept=yes
-   done
    test "$[2]" = conftest.file
    )
 then
@@ -894,85 +804,9 @@ else
    AC_MSG_ERROR([newly created file is older than distributed files!
 Check your system clock])
 fi
-AC_MSG_RESULT([yes])
-# If we didn't sleep, we still need to ensure time stamps of config.status and
-# generated files are strictly newer.
-am_sleep_pid=
-if grep 'slept: no' conftest.file >/dev/null 2>&1; then
-  ( sleep 1 ) &
-  am_sleep_pid=$!
-fi
-AC_CONFIG_COMMANDS_PRE(
-  [AC_MSG_CHECKING([that generated files are newer than configure])
-   if test -n "$am_sleep_pid"; then
-     # Hide warnings about reused PIDs.
-     wait $am_sleep_pid 2>/dev/null
-   fi
-   AC_MSG_RESULT([done])])
-rm -f conftest.file
-])
+AC_MSG_RESULT(yes)])
 
-# Copyright (C) 2009-2014 Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# AM_SILENT_RULES([DEFAULT])
-# --------------------------
-# Enable less verbose build rules; with the default set to DEFAULT
-# ("yes" being less verbose, "no" or empty being verbose).
-AC_DEFUN([AM_SILENT_RULES],
-[AC_ARG_ENABLE([silent-rules], [dnl
-AS_HELP_STRING(
-  [--enable-silent-rules],
-  [less verbose build output (undo: "make V=1")])
-AS_HELP_STRING(
-  [--disable-silent-rules],
-  [verbose build output (undo: "make V=0")])dnl
-])
-case $enable_silent_rules in @%:@ (((
-  yes) AM_DEFAULT_VERBOSITY=0;;
-   no) AM_DEFAULT_VERBOSITY=1;;
-    *) AM_DEFAULT_VERBOSITY=m4_if([$1], [yes], [0], [1]);;
-esac
-dnl
-dnl A few 'make' implementations (e.g., NonStop OS and NextStep)
-dnl do not support nested variable expansions.
-dnl See automake bug#9928 and bug#10237.
-am_make=${MAKE-make}
-AC_CACHE_CHECK([whether $am_make supports nested variables],
-   [am_cv_make_support_nested_variables],
-   [if AS_ECHO([['TRUE=$(BAR$(V))
-BAR0=false
-BAR1=true
-V=1
-am__doit:
-	@$(TRUE)
-.PHONY: am__doit']]) | $am_make -f - >/dev/null 2>&1; then
-  am_cv_make_support_nested_variables=yes
-else
-  am_cv_make_support_nested_variables=no
-fi])
-if test $am_cv_make_support_nested_variables = yes; then
-  dnl Using '$V' instead of '$(V)' breaks IRIX make.
-  AM_V='$(V)'
-  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
-else
-  AM_V=$AM_DEFAULT_VERBOSITY
-  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
-fi
-AC_SUBST([AM_V])dnl
-AM_SUBST_NOTMAKE([AM_V])dnl
-AC_SUBST([AM_DEFAULT_V])dnl
-AM_SUBST_NOTMAKE([AM_DEFAULT_V])dnl
-AC_SUBST([AM_DEFAULT_VERBOSITY])dnl
-AM_BACKSLASH='\'
-AC_SUBST([AM_BACKSLASH])dnl
-_AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
-])
-
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2001, 2003, 2005  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -980,32 +814,34 @@ _AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
 
 # AM_PROG_INSTALL_STRIP
 # ---------------------
-# One issue with vendor 'install' (even GNU) is that you can't
+# One issue with vendor `install' (even GNU) is that you can't
 # specify the program used to strip binaries.  This is especially
 # annoying in cross-compiling environments, where the build's strip
 # is unlikely to handle the host's binaries.
 # Fortunately install-sh will honor a STRIPPROG variable, so we
-# always use install-sh in "make install-strip", and initialize
+# always use install-sh in `make install-strip', and initialize
 # STRIPPROG with the value of the STRIP variable (set by the user).
 AC_DEFUN([AM_PROG_INSTALL_STRIP],
 [AC_REQUIRE([AM_PROG_INSTALL_SH])dnl
-# Installed binaries are usually stripped using 'strip' when the user
-# run "make install-strip".  However 'strip' might not be the right
+# Installed binaries are usually stripped using `strip' when the user
+# run `make install-strip'.  However `strip' might not be the right
 # tool to use in cross-compilation environments, therefore Automake
-# will honor the 'STRIP' environment variable to overrule this program.
-dnl Don't test for $cross_compiling = yes, because it might be 'maybe'.
+# will honor the `STRIP' environment variable to overrule this program.
+dnl Don't test for $cross_compiling = yes, because it might be `maybe'.
 if test "$cross_compiling" != no; then
   AC_CHECK_TOOL([STRIP], [strip], :)
 fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 AC_SUBST([INSTALL_STRIP_PROGRAM])])
 
-# Copyright (C) 2006-2014 Free Software Foundation, Inc.
+# Copyright (C) 2006, 2008  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 2
+
 # _AM_SUBST_NOTMAKE(VARIABLE)
 # ---------------------------
 # Prevent Automake from outputting VARIABLE = @VARIABLE@ in Makefile.in.
@@ -1013,22 +849,24 @@ AC_SUBST([INSTALL_STRIP_PROGRAM])])
 AC_DEFUN([_AM_SUBST_NOTMAKE])
 
 # AM_SUBST_NOTMAKE(VARIABLE)
-# --------------------------
+# ---------------------------
 # Public sister of _AM_SUBST_NOTMAKE.
 AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 
 # Check how to create a tarball.                            -*- Autoconf -*-
 
-# Copyright (C) 2004-2014 Free Software Foundation, Inc.
+# Copyright (C) 2004, 2005  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 2
+
 # _AM_PROG_TAR(FORMAT)
 # --------------------
 # Check how to create a tarball in format FORMAT.
-# FORMAT should be one of 'v7', 'ustar', or 'pax'.
+# FORMAT should be one of `v7', `ustar', or `pax'.
 #
 # Substitute a variable $(am__tar) that is a command
 # writing to stdout a FORMAT-tarball containing the directory
@@ -1038,114 +876,75 @@ AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 # Substitute a variable $(am__untar) that extract such
 # a tarball read from stdin.
 #     $(am__untar) < result.tar
-#
 AC_DEFUN([_AM_PROG_TAR],
-[# Always define AMTAR for backward compatibility.  Yes, it's still used
-# in the wild :-(  We should find a proper way to deprecate it ...
-AC_SUBST([AMTAR], ['$${TAR-tar}'])
-
-# We'll loop over all known methods to create a tar archive until one works.
-_am_tools='gnutar m4_if([$1], [ustar], [plaintar]) pax cpio none'
-
+[# Always define AMTAR for backward compatibility.
+AM_MISSING_PROG([AMTAR], [tar])
 m4_if([$1], [v7],
-  [am__tar='$${TAR-tar} chof - "$$tardir"' am__untar='$${TAR-tar} xf -'],
+     [am__tar='${AMTAR} chof - "$$tardir"'; am__untar='${AMTAR} xf -'],
+     [m4_case([$1], [ustar],, [pax],,
+              [m4_fatal([Unknown tar format])])
+AC_MSG_CHECKING([how to create a $1 tar archive])
+# Loop over all known methods to create a tar archive until one works.
+_am_tools='gnutar m4_if([$1], [ustar], [plaintar]) pax cpio none'
+_am_tools=${am_cv_prog_tar_$1-$_am_tools}
+# Do not fold the above two line into one, because Tru64 sh and
+# Solaris sh will not grok spaces in the rhs of `-'.
+for _am_tool in $_am_tools
+do
+  case $_am_tool in
+  gnutar)
+    for _am_tar in tar gnutar gtar;
+    do
+      AM_RUN_LOG([$_am_tar --version]) && break
+    done
+    am__tar="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$$tardir"'
+    am__tar_="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$tardir"'
+    am__untar="$_am_tar -xf -"
+    ;;
+  plaintar)
+    # Must skip GNU tar: if it does not support --format= it doesn't create
+    # ustar tarball either.
+    (tar --version) >/dev/null 2>&1 && continue
+    am__tar='tar chf - "$$tardir"'
+    am__tar_='tar chf - "$tardir"'
+    am__untar='tar xf -'
+    ;;
+  pax)
+    am__tar='pax -L -x $1 -w "$$tardir"'
+    am__tar_='pax -L -x $1 -w "$tardir"'
+    am__untar='pax -r'
+    ;;
+  cpio)
+    am__tar='find "$$tardir" -print | cpio -o -H $1 -L'
+    am__tar_='find "$tardir" -print | cpio -o -H $1 -L'
+    am__untar='cpio -i -H $1 -d'
+    ;;
+  none)
+    am__tar=false
+    am__tar_=false
+    am__untar=false
+    ;;
+  esac
 
-  [m4_case([$1],
-    [ustar],
-     [# The POSIX 1988 'ustar' format is defined with fixed-size fields.
-      # There is notably a 21 bits limit for the UID and the GID.  In fact,
-      # the 'pax' utility can hang on bigger UID/GID (see automake bug#8343
-      # and bug#13588).
-      am_max_uid=2097151 # 2^21 - 1
-      am_max_gid=$am_max_uid
-      # The $UID and $GID variables are not portable, so we need to resort
-      # to the POSIX-mandated id(1) utility.  Errors in the 'id' calls
-      # below are definitely unexpected, so allow the users to see them
-      # (that is, avoid stderr redirection).
-      am_uid=`id -u || echo unknown`
-      am_gid=`id -g || echo unknown`
-      AC_MSG_CHECKING([whether UID '$am_uid' is supported by ustar format])
-      if test $am_uid -le $am_max_uid; then
-         AC_MSG_RESULT([yes])
-      else
-         AC_MSG_RESULT([no])
-         _am_tools=none
-      fi
-      AC_MSG_CHECKING([whether GID '$am_gid' is supported by ustar format])
-      if test $am_gid -le $am_max_gid; then
-         AC_MSG_RESULT([yes])
-      else
-        AC_MSG_RESULT([no])
-        _am_tools=none
-      fi],
+  # If the value was cached, stop now.  We just wanted to have am__tar
+  # and am__untar set.
+  test -n "${am_cv_prog_tar_$1}" && break
 
-  [pax],
-    [],
-
-  [m4_fatal([Unknown tar format])])
-
-  AC_MSG_CHECKING([how to create a $1 tar archive])
-
-  # Go ahead even if we have the value already cached.  We do so because we
-  # need to set the values for the 'am__tar' and 'am__untar' variables.
-  _am_tools=${am_cv_prog_tar_$1-$_am_tools}
-
-  for _am_tool in $_am_tools; do
-    case $_am_tool in
-    gnutar)
-      for _am_tar in tar gnutar gtar; do
-        AM_RUN_LOG([$_am_tar --version]) && break
-      done
-      am__tar="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$$tardir"'
-      am__tar_="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$tardir"'
-      am__untar="$_am_tar -xf -"
-      ;;
-    plaintar)
-      # Must skip GNU tar: if it does not support --format= it doesn't create
-      # ustar tarball either.
-      (tar --version) >/dev/null 2>&1 && continue
-      am__tar='tar chf - "$$tardir"'
-      am__tar_='tar chf - "$tardir"'
-      am__untar='tar xf -'
-      ;;
-    pax)
-      am__tar='pax -L -x $1 -w "$$tardir"'
-      am__tar_='pax -L -x $1 -w "$tardir"'
-      am__untar='pax -r'
-      ;;
-    cpio)
-      am__tar='find "$$tardir" -print | cpio -o -H $1 -L'
-      am__tar_='find "$tardir" -print | cpio -o -H $1 -L'
-      am__untar='cpio -i -H $1 -d'
-      ;;
-    none)
-      am__tar=false
-      am__tar_=false
-      am__untar=false
-      ;;
-    esac
-
-    # If the value was cached, stop now.  We just wanted to have am__tar
-    # and am__untar set.
-    test -n "${am_cv_prog_tar_$1}" && break
-
-    # tar/untar a dummy directory, and stop if the command works.
-    rm -rf conftest.dir
-    mkdir conftest.dir
-    echo GrepMe > conftest.dir/file
-    AM_RUN_LOG([tardir=conftest.dir && eval $am__tar_ >conftest.tar])
-    rm -rf conftest.dir
-    if test -s conftest.tar; then
-      AM_RUN_LOG([$am__untar <conftest.tar])
-      AM_RUN_LOG([cat conftest.dir/file])
-      grep GrepMe conftest.dir/file >/dev/null 2>&1 && break
-    fi
-  done
+  # tar/untar a dummy directory, and stop if the command works
   rm -rf conftest.dir
+  mkdir conftest.dir
+  echo GrepMe > conftest.dir/file
+  AM_RUN_LOG([tardir=conftest.dir && eval $am__tar_ >conftest.tar])
+  rm -rf conftest.dir
+  if test -s conftest.tar; then
+    AM_RUN_LOG([$am__untar <conftest.tar])
+    grep GrepMe conftest.dir/file >/dev/null 2>&1 && break
+  fi
+done
+rm -rf conftest.dir
 
-  AC_CACHE_VAL([am_cv_prog_tar_$1], [am_cv_prog_tar_$1=$_am_tool])
-  AC_MSG_RESULT([$am_cv_prog_tar_$1])])
-
+AC_CACHE_VAL([am_cv_prog_tar_$1], [am_cv_prog_tar_$1=$_am_tool])
+AC_MSG_RESULT([$am_cv_prog_tar_$1])])
 AC_SUBST([am__tar])
 AC_SUBST([am__untar])
 ]) # _AM_PROG_TAR
diff --git a/benchmarks/Grid_memory_bandwidth.cc b/benchmarks/Grid_memory_bandwidth.cc
index dc25fe51..3254e95b 100644
--- a/benchmarks/Grid_memory_bandwidth.cc
+++ b/benchmarks/Grid_memory_bandwidth.cc
@@ -22,7 +22,7 @@ int main (int argc, char ** argv)
   std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s"<<std::endl;
   std::cout << "----------------------------------------------------------"<<std::endl;
 
-  for(int lat=4;lat<=24;lat+=4){
+  for(int lat=4;lat<=32;lat+=4){
 
       std::vector<int> latt_size  ({lat,lat,lat,lat});
 
@@ -43,10 +43,10 @@ int main (int argc, char ** argv)
 	axpy(z,a,x,y);
       }
       double stop=usecond();
-      double time = (stop-start)/Nloop/1000;
+      double time = (stop-start)/Nloop*1000;
       
       double bytes=3*lat*lat*lat*lat*Nvec*sizeof(Real);
-      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
+      std::cout<<std::setprecision(2) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
 
     }
 
@@ -57,7 +57,7 @@ int main (int argc, char ** argv)
   std::cout << "----------------------------------------------------------"<<std::endl;
 
   
-  for(int lat=4;lat<=24;lat+=4){
+  for(int lat=4;lat<=32;lat+=4){
 
       std::vector<int> latt_size  ({lat,lat,lat,lat});
 
@@ -76,11 +76,11 @@ int main (int argc, char ** argv)
 	z=a*x+y;
       }
       double stop=usecond();
-      double time = (stop-start)/Nloop/1000;
+      double time = (stop-start)/Nloop*1000;
       
       double effbytes=3*lat*lat*lat*lat*Nvec*sizeof(Real);
       double bytes=5*lat*lat*lat*lat*Nvec*sizeof(Real);
-      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<"\t\t"<<effbytes/time<<std::endl;
+      std::cout<<std::setprecision(2) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<"\t\t"<<effbytes/time<<std::endl;
 
     }
 
@@ -89,7 +89,7 @@ int main (int argc, char ** argv)
   std::cout << "===================================================================================================="<<std::endl;
   std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s"<<std::endl;
 
-  for(int lat=4;lat<=24;lat+=4){
+  for(int lat=4;lat<=32;lat+=4){
 
       std::vector<int> latt_size  ({lat,lat,lat,lat});
 
@@ -108,10 +108,10 @@ int main (int argc, char ** argv)
 	x=a*z;
       }
       double stop=usecond();
-      double time = (stop-start)/Nloop/1000;
+      double time = (stop-start)/Nloop*1000;
       
       double bytes=2*lat*lat*lat*lat*Nvec*sizeof(Real);
-      std::cout <<std::setprecision(3) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
+      std::cout <<std::setprecision(2) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
 
   }
 
@@ -121,7 +121,7 @@ int main (int argc, char ** argv)
   std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s"<<std::endl;
   std::cout << "----------------------------------------------------------"<<std::endl;
 
-  for(int lat=4;lat<=24;lat+=4){
+  for(int lat=4;lat<=32;lat+=4){
 
       std::vector<int> latt_size  ({lat,lat,lat,lat});
 
@@ -140,10 +140,10 @@ int main (int argc, char ** argv)
 	nn=norm2(x);
       }
       double stop=usecond();
-      double time = (stop-start)/Nloop/1000;
+      double time = (stop-start)/Nloop*1000;
       
       double bytes=lat*lat*lat*lat*Nvec*sizeof(Real);
-      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
+      std::cout<<std::setprecision(2) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
 
   }    
 
diff --git a/configure b/configure
index 3d25f6a9..3e71855d 100755
--- a/configure
+++ b/configure
@@ -1,22 +1,20 @@
 #! /bin/sh
 # Guess values for system-dependent variables and create Makefiles.
-# Generated by GNU Autoconf 2.69 for Grid 1.0.
+# Generated by GNU Autoconf 2.63 for Grid 1.0.
 #
 # Report bugs to <paboyle@ph.ed.ac.uk>.
 #
-#
-# Copyright (C) 1992-1996, 1998-2012 Free Software Foundation, Inc.
-#
-#
+# Copyright (C) 1992, 1993, 1994, 1995, 1996, 1998, 1999, 2000, 2001,
+# 2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
 # This configure script is free software; the Free Software Foundation
 # gives unlimited permission to copy, distribute and modify it.
-## -------------------- ##
-## M4sh Initialization. ##
-## -------------------- ##
+## --------------------- ##
+## M4sh Initialization.  ##
+## --------------------- ##
 
 # Be more Bourne compatible
 DUALCASE=1; export DUALCASE # for MKS sh
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
   emulate sh
   NULLCMD=:
   # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
@@ -24,15 +22,23 @@ if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
   alias -g '${1+"$@"}'='"$@"'
   setopt NO_GLOB_SUBST
 else
-  case `(set -o) 2>/dev/null` in #(
-  *posix*) :
-    set -o posix ;; #(
-  *) :
-     ;;
+  case `(set -o) 2>/dev/null` in
+  *posix*) set -o posix ;;
 esac
+
 fi
 
 
+
+
+# PATH needs CR
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
+
 as_nl='
 '
 export as_nl
@@ -40,13 +46,7 @@ export as_nl
 as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
-# Prefer a ksh shell builtin over an external printf program on Solaris,
-# but without wasting forks for bash or zsh.
-if test -z "$BASH_VERSION$ZSH_VERSION" \
-    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
-  as_echo='print -r --'
-  as_echo_n='print -rn --'
-elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+if (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
   as_echo='printf %s\n'
   as_echo_n='printf %s'
 else
@@ -57,7 +57,7 @@ else
     as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
     as_echo_n_body='eval
       arg=$1;
-      case $arg in #(
+      case $arg in
       *"$as_nl"*)
 	expr "X$arg" : "X\\(.*\\)$as_nl";
 	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
@@ -80,6 +80,13 @@ if test "${PATH_SEPARATOR+set}" != set; then
   }
 fi
 
+# Support unset when possible.
+if ( (MAIL=60; unset MAIL) || exit) >/dev/null 2>&1; then
+  as_unset=unset
+else
+  as_unset=false
+fi
+
 
 # IFS
 # We need space, tab and new line, in precisely that order.  Quoting is
@@ -89,16 +96,15 @@ fi
 IFS=" ""	$as_nl"
 
 # Find who we are.  Look in the path if we contain no directory separator.
-as_myself=
-case $0 in #((
+case $0 in
   *[\\/]* ) as_myself=$0 ;;
   *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
 for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
-  done
+  test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+done
 IFS=$as_save_IFS
 
      ;;
@@ -110,16 +116,12 @@ if test "x$as_myself" = x; then
 fi
 if test ! -f "$as_myself"; then
   $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
-  exit 1
+  { (exit 1); exit 1; }
 fi
 
-# Unset variables that we do not need and which cause bugs (e.g. in
-# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
-# suppresses any "Segmentation fault" message there.  '((' could
-# trigger a bug in pdksh 5.2.14.
-for as_var in BASH_ENV ENV MAIL MAILPATH
-do eval test x\${$as_var+set} = xset \
-  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
+# Work around bugs in pre-3.0 UWIN ksh.
+for as_var in ENV MAIL MAILPATH
+do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
 done
 PS1='$ '
 PS2='> '
@@ -131,294 +133,7 @@ export LC_ALL
 LANGUAGE=C
 export LANGUAGE
 
-# CDPATH.
-(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
-
-# Use a proper internal environment variable to ensure we don't fall
-  # into an infinite loop, continuously re-executing ourselves.
-  if test x"${_as_can_reexec}" != xno && test "x$CONFIG_SHELL" != x; then
-    _as_can_reexec=no; export _as_can_reexec;
-    # We cannot yet assume a decent shell, so we have to provide a
-# neutralization value for shells without unset; and this also
-# works around shells that cannot unset nonexistent variables.
-# Preserve -v and -x to the replacement shell.
-BASH_ENV=/dev/null
-ENV=/dev/null
-(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
-case $- in # ((((
-  *v*x* | *x*v* ) as_opts=-vx ;;
-  *v* ) as_opts=-v ;;
-  *x* ) as_opts=-x ;;
-  * ) as_opts= ;;
-esac
-exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
-# Admittedly, this is quite paranoid, since all the known shells bail
-# out after a failed `exec'.
-$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
-as_fn_exit 255
-  fi
-  # We don't want this to propagate to other subprocesses.
-          { _as_can_reexec=; unset _as_can_reexec;}
-if test "x$CONFIG_SHELL" = x; then
-  as_bourne_compatible="if test -n \"\${ZSH_VERSION+set}\" && (emulate sh) >/dev/null 2>&1; then :
-  emulate sh
-  NULLCMD=:
-  # Pre-4.2 versions of Zsh do word splitting on \${1+\"\$@\"}, which
-  # is contrary to our usage.  Disable this feature.
-  alias -g '\${1+\"\$@\"}'='\"\$@\"'
-  setopt NO_GLOB_SUBST
-else
-  case \`(set -o) 2>/dev/null\` in #(
-  *posix*) :
-    set -o posix ;; #(
-  *) :
-     ;;
-esac
-fi
-"
-  as_required="as_fn_return () { (exit \$1); }
-as_fn_success () { as_fn_return 0; }
-as_fn_failure () { as_fn_return 1; }
-as_fn_ret_success () { return 0; }
-as_fn_ret_failure () { return 1; }
-
-exitcode=0
-as_fn_success || { exitcode=1; echo as_fn_success failed.; }
-as_fn_failure && { exitcode=1; echo as_fn_failure succeeded.; }
-as_fn_ret_success || { exitcode=1; echo as_fn_ret_success failed.; }
-as_fn_ret_failure && { exitcode=1; echo as_fn_ret_failure succeeded.; }
-if ( set x; as_fn_ret_success y && test x = \"\$1\" ); then :
-
-else
-  exitcode=1; echo positional parameters were not saved.
-fi
-test x\$exitcode = x0 || exit 1
-test -x / || exit 1"
-  as_suggested="  as_lineno_1=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_1a=\$LINENO
-  as_lineno_2=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_2a=\$LINENO
-  eval 'test \"x\$as_lineno_1'\$as_run'\" != \"x\$as_lineno_2'\$as_run'\" &&
-  test \"x\`expr \$as_lineno_1'\$as_run' + 1\`\" = \"x\$as_lineno_2'\$as_run'\"' || exit 1
-test \$(( 1 + 1 )) = 2 || exit 1"
-  if (eval "$as_required") 2>/dev/null; then :
-  as_have_required=yes
-else
-  as_have_required=no
-fi
-  if test x$as_have_required = xyes && (eval "$as_suggested") 2>/dev/null; then :
-
-else
-  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-as_found=false
-for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-  as_found=:
-  case $as_dir in #(
-	 /*)
-	   for as_base in sh bash ksh sh5; do
-	     # Try only shells that exist, to save several forks.
-	     as_shell=$as_dir/$as_base
-	     if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
-		    { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$as_shell"; } 2>/dev/null; then :
-  CONFIG_SHELL=$as_shell as_have_required=yes
-		   if { $as_echo "$as_bourne_compatible""$as_suggested" | as_run=a "$as_shell"; } 2>/dev/null; then :
-  break 2
-fi
-fi
-	   done;;
-       esac
-  as_found=false
-done
-$as_found || { if { test -f "$SHELL" || test -f "$SHELL.exe"; } &&
-	      { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$SHELL"; } 2>/dev/null; then :
-  CONFIG_SHELL=$SHELL as_have_required=yes
-fi; }
-IFS=$as_save_IFS
-
-
-      if test "x$CONFIG_SHELL" != x; then :
-  export CONFIG_SHELL
-             # We cannot yet assume a decent shell, so we have to provide a
-# neutralization value for shells without unset; and this also
-# works around shells that cannot unset nonexistent variables.
-# Preserve -v and -x to the replacement shell.
-BASH_ENV=/dev/null
-ENV=/dev/null
-(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
-case $- in # ((((
-  *v*x* | *x*v* ) as_opts=-vx ;;
-  *v* ) as_opts=-v ;;
-  *x* ) as_opts=-x ;;
-  * ) as_opts= ;;
-esac
-exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
-# Admittedly, this is quite paranoid, since all the known shells bail
-# out after a failed `exec'.
-$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
-exit 255
-fi
-
-    if test x$as_have_required = xno; then :
-  $as_echo "$0: This script requires a shell more modern than all"
-  $as_echo "$0: the shells that I found on your system."
-  if test x${ZSH_VERSION+set} = xset ; then
-    $as_echo "$0: In particular, zsh $ZSH_VERSION has bugs and should"
-    $as_echo "$0: be upgraded to zsh 4.3.4 or later."
-  else
-    $as_echo "$0: Please tell bug-autoconf@gnu.org and
-$0: paboyle@ph.ed.ac.uk about your system, including any
-$0: error possibly output before this message. Then install
-$0: a modern shell, or manually run the script under such a
-$0: shell if you do have one."
-  fi
-  exit 1
-fi
-fi
-fi
-SHELL=${CONFIG_SHELL-/bin/sh}
-export SHELL
-# Unset more variables known to interfere with behavior of common tools.
-CLICOLOR_FORCE= GREP_OPTIONS=
-unset CLICOLOR_FORCE GREP_OPTIONS
-
-## --------------------- ##
-## M4sh Shell Functions. ##
-## --------------------- ##
-# as_fn_unset VAR
-# ---------------
-# Portably unset VAR.
-as_fn_unset ()
-{
-  { eval $1=; unset $1;}
-}
-as_unset=as_fn_unset
-
-# as_fn_set_status STATUS
-# -----------------------
-# Set $? to STATUS, without forking.
-as_fn_set_status ()
-{
-  return $1
-} # as_fn_set_status
-
-# as_fn_exit STATUS
-# -----------------
-# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
-as_fn_exit ()
-{
-  set +e
-  as_fn_set_status $1
-  exit $1
-} # as_fn_exit
-
-# as_fn_mkdir_p
-# -------------
-# Create "$as_dir" as a directory, including parents if necessary.
-as_fn_mkdir_p ()
-{
-
-  case $as_dir in #(
-  -*) as_dir=./$as_dir;;
-  esac
-  test -d "$as_dir" || eval $as_mkdir_p || {
-    as_dirs=
-    while :; do
-      case $as_dir in #(
-      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
-      *) as_qdir=$as_dir;;
-      esac
-      as_dirs="'$as_qdir' $as_dirs"
-      as_dir=`$as_dirname -- "$as_dir" ||
-$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_dir" : 'X\(//\)[^/]' \| \
-	 X"$as_dir" : 'X\(//\)$' \| \
-	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_dir" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      test -d "$as_dir" && break
-    done
-    test -z "$as_dirs" || eval "mkdir $as_dirs"
-  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
-
-
-} # as_fn_mkdir_p
-
-# as_fn_executable_p FILE
-# -----------------------
-# Test if FILE is an executable regular file.
-as_fn_executable_p ()
-{
-  test -f "$1" && test -x "$1"
-} # as_fn_executable_p
-# as_fn_append VAR VALUE
-# ----------------------
-# Append the text in VALUE to the end of the definition contained in VAR. Take
-# advantage of any shell optimizations that allow amortized linear growth over
-# repeated appends, instead of the typical quadratic growth present in naive
-# implementations.
-if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
-  eval 'as_fn_append ()
-  {
-    eval $1+=\$2
-  }'
-else
-  as_fn_append ()
-  {
-    eval $1=\$$1\$2
-  }
-fi # as_fn_append
-
-# as_fn_arith ARG...
-# ------------------
-# Perform arithmetic evaluation on the ARGs, and store the result in the
-# global $as_val. Take advantage of shells that can avoid forks. The arguments
-# must be portable across $(()) and expr.
-if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
-  eval 'as_fn_arith ()
-  {
-    as_val=$(( $* ))
-  }'
-else
-  as_fn_arith ()
-  {
-    as_val=`expr "$@" || test $? -eq 1`
-  }
-fi # as_fn_arith
-
-
-# as_fn_error STATUS ERROR [LINENO LOG_FD]
-# ----------------------------------------
-# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
-# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
-# script with STATUS, using 1 if that was 0.
-as_fn_error ()
-{
-  as_status=$1; test $as_status -eq 0 && as_status=1
-  if test "$4"; then
-    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
-  fi
-  $as_echo "$as_me: error: $2" >&2
-  as_fn_exit $as_status
-} # as_fn_error
-
+# Required to use basename.
 if expr a : '\(a\)' >/dev/null 2>&1 &&
    test "X`expr 00001 : '.*\(...\)'`" = X001; then
   as_expr=expr
@@ -432,12 +147,8 @@ else
   as_basename=false
 fi
 
-if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
-  as_dirname=dirname
-else
-  as_dirname=false
-fi
 
+# Name of the executable.
 as_me=`$as_basename -- "$0" ||
 $as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
 	 X"$0" : 'X\(//\)$' \| \
@@ -457,19 +168,295 @@ $as_echo X/"$0" |
 	  }
 	  s/.*/./; q'`
 
-# Avoid depending upon Character Ranges.
-as_cr_letters='abcdefghijklmnopqrstuvwxyz'
-as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
-as_cr_Letters=$as_cr_letters$as_cr_LETTERS
-as_cr_digits='0123456789'
-as_cr_alnum=$as_cr_Letters$as_cr_digits
+# CDPATH.
+$as_unset CDPATH
 
 
-  as_lineno_1=$LINENO as_lineno_1a=$LINENO
-  as_lineno_2=$LINENO as_lineno_2a=$LINENO
-  eval 'test "x$as_lineno_1'$as_run'" != "x$as_lineno_2'$as_run'" &&
-  test "x`expr $as_lineno_1'$as_run' + 1`" = "x$as_lineno_2'$as_run'"' || {
-  # Blame Lee E. McMahon (1931-1989) for sed's syntax.  :-)
+if test "x$CONFIG_SHELL" = x; then
+  if (eval ":") 2>/dev/null; then
+  as_have_required=yes
+else
+  as_have_required=no
+fi
+
+  if test $as_have_required = yes &&	 (eval ":
+(as_func_return () {
+  (exit \$1)
+}
+as_func_success () {
+  as_func_return 0
+}
+as_func_failure () {
+  as_func_return 1
+}
+as_func_ret_success () {
+  return 0
+}
+as_func_ret_failure () {
+  return 1
+}
+
+exitcode=0
+if as_func_success; then
+  :
+else
+  exitcode=1
+  echo as_func_success failed.
+fi
+
+if as_func_failure; then
+  exitcode=1
+  echo as_func_failure succeeded.
+fi
+
+if as_func_ret_success; then
+  :
+else
+  exitcode=1
+  echo as_func_ret_success failed.
+fi
+
+if as_func_ret_failure; then
+  exitcode=1
+  echo as_func_ret_failure succeeded.
+fi
+
+if ( set x; as_func_ret_success y && test x = \"\$1\" ); then
+  :
+else
+  exitcode=1
+  echo positional parameters were not saved.
+fi
+
+test \$exitcode = 0) || { (exit 1); exit 1; }
+
+(
+  as_lineno_1=\$LINENO
+  as_lineno_2=\$LINENO
+  test \"x\$as_lineno_1\" != \"x\$as_lineno_2\" &&
+  test \"x\`expr \$as_lineno_1 + 1\`\" = \"x\$as_lineno_2\") || { (exit 1); exit 1; }
+") 2> /dev/null; then
+  :
+else
+  as_candidate_shells=
+    as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+  case $as_dir in
+	 /*)
+	   for as_base in sh bash ksh sh5; do
+	     as_candidate_shells="$as_candidate_shells $as_dir/$as_base"
+	   done;;
+       esac
+done
+IFS=$as_save_IFS
+
+
+      for as_shell in $as_candidate_shells $SHELL; do
+	 # Try only shells that exist, to save several forks.
+	 if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
+		{ ("$as_shell") 2> /dev/null <<\_ASEOF
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
+  emulate sh
+  NULLCMD=:
+  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '${1+"$@"}'='"$@"'
+  setopt NO_GLOB_SUBST
+else
+  case `(set -o) 2>/dev/null` in
+  *posix*) set -o posix ;;
+esac
+
+fi
+
+
+:
+_ASEOF
+}; then
+  CONFIG_SHELL=$as_shell
+	       as_have_required=yes
+	       if { "$as_shell" 2> /dev/null <<\_ASEOF
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
+  emulate sh
+  NULLCMD=:
+  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '${1+"$@"}'='"$@"'
+  setopt NO_GLOB_SUBST
+else
+  case `(set -o) 2>/dev/null` in
+  *posix*) set -o posix ;;
+esac
+
+fi
+
+
+:
+(as_func_return () {
+  (exit $1)
+}
+as_func_success () {
+  as_func_return 0
+}
+as_func_failure () {
+  as_func_return 1
+}
+as_func_ret_success () {
+  return 0
+}
+as_func_ret_failure () {
+  return 1
+}
+
+exitcode=0
+if as_func_success; then
+  :
+else
+  exitcode=1
+  echo as_func_success failed.
+fi
+
+if as_func_failure; then
+  exitcode=1
+  echo as_func_failure succeeded.
+fi
+
+if as_func_ret_success; then
+  :
+else
+  exitcode=1
+  echo as_func_ret_success failed.
+fi
+
+if as_func_ret_failure; then
+  exitcode=1
+  echo as_func_ret_failure succeeded.
+fi
+
+if ( set x; as_func_ret_success y && test x = "$1" ); then
+  :
+else
+  exitcode=1
+  echo positional parameters were not saved.
+fi
+
+test $exitcode = 0) || { (exit 1); exit 1; }
+
+(
+  as_lineno_1=$LINENO
+  as_lineno_2=$LINENO
+  test "x$as_lineno_1" != "x$as_lineno_2" &&
+  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2") || { (exit 1); exit 1; }
+
+_ASEOF
+}; then
+  break
+fi
+
+fi
+
+      done
+
+      if test "x$CONFIG_SHELL" != x; then
+  for as_var in BASH_ENV ENV
+	do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
+	done
+	export CONFIG_SHELL
+	exec "$CONFIG_SHELL" "$as_myself" ${1+"$@"}
+fi
+
+
+    if test $as_have_required = no; then
+  echo This script requires a shell more modern than all the
+      echo shells that I found on your system.  Please install a
+      echo modern shell, or manually run the script under such a
+      echo shell if you do have one.
+      { (exit 1); exit 1; }
+fi
+
+
+fi
+
+fi
+
+
+
+(eval "as_func_return () {
+  (exit \$1)
+}
+as_func_success () {
+  as_func_return 0
+}
+as_func_failure () {
+  as_func_return 1
+}
+as_func_ret_success () {
+  return 0
+}
+as_func_ret_failure () {
+  return 1
+}
+
+exitcode=0
+if as_func_success; then
+  :
+else
+  exitcode=1
+  echo as_func_success failed.
+fi
+
+if as_func_failure; then
+  exitcode=1
+  echo as_func_failure succeeded.
+fi
+
+if as_func_ret_success; then
+  :
+else
+  exitcode=1
+  echo as_func_ret_success failed.
+fi
+
+if as_func_ret_failure; then
+  exitcode=1
+  echo as_func_ret_failure succeeded.
+fi
+
+if ( set x; as_func_ret_success y && test x = \"\$1\" ); then
+  :
+else
+  exitcode=1
+  echo positional parameters were not saved.
+fi
+
+test \$exitcode = 0") || {
+  echo No shell found that supports shell functions.
+  echo Please tell bug-autoconf@gnu.org about your system,
+  echo including any error possibly output before this message.
+  echo This can help us improve future autoconf versions.
+  echo Configuration will now proceed without shell functions.
+}
+
+
+
+  as_lineno_1=$LINENO
+  as_lineno_2=$LINENO
+  test "x$as_lineno_1" != "x$as_lineno_2" &&
+  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2" || {
+
+  # Create $as_me.lineno as a copy of $as_myself, but with $LINENO
+  # uniformly replaced by the line number.  The first 'sed' inserts a
+  # line-number line after each line using $LINENO; the second 'sed'
+  # does the real work.  The second script uses 'N' to pair each
+  # line-number line with the line containing $LINENO, and appends
+  # trailing '-' during substitution so that $LINENO is not a special
+  # case at line end.
+  # (Raja R Harinath suggested sed '=', and Paul Eggert wrote the
+  # scripts with optimization help from Paolo Bonzini.  Blame Lee
+  # E. McMahon (1931-1989) for sed's syntax.  :-)
   sed -n '
     p
     /[$]LINENO/=
@@ -486,12 +473,9 @@ as_cr_alnum=$as_cr_Letters$as_cr_digits
       s/-\n.*//
     ' >$as_me.lineno &&
   chmod +x "$as_me.lineno" ||
-    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2; as_fn_exit 1; }
+    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2
+   { (exit 1); exit 1; }; }
 
-  # If we had to re-execute with $CONFIG_SHELL, we're ensured to have
-  # already done that, so ensure we don't try to do so again and fall
-  # in an infinite loop.  This has already happened in practice.
-  _as_can_reexec=no; export _as_can_reexec
   # Don't try to exec as it changes $[0], causing all sort of problems
   # (the dirname of $[0] is not the place where we might find the
   # original and so on.  Autoconf is especially sensitive to this).
@@ -500,18 +484,29 @@ as_cr_alnum=$as_cr_Letters$as_cr_digits
   exit
 }
 
+
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
+
 ECHO_C= ECHO_N= ECHO_T=
-case `echo -n x` in #(((((
+case `echo -n x` in
 -n*)
-  case `echo 'xy\c'` in
+  case `echo 'x\c'` in
   *c*) ECHO_T='	';;	# ECHO_T is single tab character.
-  xy)  ECHO_C='\c';;
-  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
-       ECHO_T='	';;
+  *)   ECHO_C='\c';;
   esac;;
 *)
   ECHO_N='-n';;
 esac
+if expr a : '\(a\)' >/dev/null 2>&1 &&
+   test "X`expr 00001 : '.*\(...\)'`" = X001; then
+  as_expr=expr
+else
+  as_expr=false
+fi
 
 rm -f conf$$ conf$$.exe conf$$.file
 if test -d conf$$.dir; then
@@ -526,29 +521,49 @@ if (echo >conf$$.file) 2>/dev/null; then
     # ... but there are two gotchas:
     # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
     # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
-    # In both cases, we have to default to `cp -pR'.
+    # In both cases, we have to default to `cp -p'.
     ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
-      as_ln_s='cp -pR'
+      as_ln_s='cp -p'
   elif ln conf$$.file conf$$ 2>/dev/null; then
     as_ln_s=ln
   else
-    as_ln_s='cp -pR'
+    as_ln_s='cp -p'
   fi
 else
-  as_ln_s='cp -pR'
+  as_ln_s='cp -p'
 fi
 rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
 rmdir conf$$.dir 2>/dev/null
 
 if mkdir -p . 2>/dev/null; then
-  as_mkdir_p='mkdir -p "$as_dir"'
+  as_mkdir_p=:
 else
   test -d ./-p && rmdir ./-p
   as_mkdir_p=false
 fi
 
-as_test_x='test -x'
-as_executable_p=as_fn_executable_p
+if test -x / >/dev/null 2>&1; then
+  as_test_x='test -x'
+else
+  if ls -dL / >/dev/null 2>&1; then
+    as_ls_L_option=L
+  else
+    as_ls_L_option=
+  fi
+  as_test_x='
+    eval sh -c '\''
+      if test -d "$1"; then
+	test -d "$1/.";
+      else
+	case $1 in
+	-*)set "./$1";;
+	esac;
+	case `ls -ld'$as_ls_L_option' "$1" 2>/dev/null` in
+	???[sx]*):;;*)false;;esac;fi
+    '\'' sh
+  '
+fi
+as_executable_p=$as_test_x
 
 # Sed expression to map a string onto a valid CPP name.
 as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
@@ -557,11 +572,11 @@ as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
 as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
 
 
-test -n "$DJDIR" || exec 7<&0 </dev/null
-exec 6>&1
+
+exec 7<&0 </dev/null 6>&1
 
 # Name of the host.
-# hostname on some systems (SVR3.2, old GNU/Linux) returns a bogus exit status,
+# hostname on some systems (SVR3.2, Linux) returns a bogus exit status,
 # so uname gets run too.
 ac_hostname=`(hostname || uname -n) 2>/dev/null | sed 1q`
 
@@ -576,6 +591,7 @@ cross_compiling=no
 subdirs=
 MFLAGS=
 MAKEFLAGS=
+SHELL=${CONFIG_SHELL-/bin/sh}
 
 # Identity of this package.
 PACKAGE_NAME='Grid'
@@ -583,7 +599,6 @@ PACKAGE_TARNAME='grid'
 PACKAGE_VERSION='1.0'
 PACKAGE_STRING='Grid 1.0'
 PACKAGE_BUGREPORT='paboyle@ph.ed.ac.uk'
-PACKAGE_URL=''
 
 ac_unique_file="lib/Grid.h"
 # Factoring default headers for most tests.
@@ -644,7 +659,6 @@ CC
 am__fastdepCXX_FALSE
 am__fastdepCXX_TRUE
 CXXDEPMODE
-am__nodep
 AMDEPBACKSLASH
 AMDEP_FALSE
 AMDEP_TRUE
@@ -658,10 +672,6 @@ CPPFLAGS
 LDFLAGS
 CXXFLAGS
 CXX
-AM_BACKSLASH
-AM_DEFAULT_VERBOSITY
-AM_DEFAULT_V
-AM_V
 am__untar
 am__tar
 AMTAR
@@ -715,7 +725,6 @@ bindir
 program_transform_name
 prefix
 exec_prefix
-PACKAGE_URL
 PACKAGE_BUGREPORT
 PACKAGE_STRING
 PACKAGE_VERSION
@@ -726,7 +735,6 @@ SHELL'
 ac_subst_files=''
 ac_user_opts='
 enable_option_checking
-enable_silent_rules
 enable_dependency_tracking
 enable_openmp
 enable_simd
@@ -806,9 +814,8 @@ do
   fi
 
   case $ac_option in
-  *=?*) ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
-  *=)   ac_optarg= ;;
-  *)    ac_optarg=yes ;;
+  *=*)	ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
+  *)	ac_optarg=yes ;;
   esac
 
   # Accept the important Cygnus configure options, so we can diagnose typos.
@@ -853,7 +860,8 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*disable-\(.*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid feature name: $ac_useropt"
+      { $as_echo "$as_me: error: invalid feature name: $ac_useropt" >&2
+   { (exit 1); exit 1; }; }
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -879,7 +887,8 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*enable-\([^=]*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid feature name: $ac_useropt"
+      { $as_echo "$as_me: error: invalid feature name: $ac_useropt" >&2
+   { (exit 1); exit 1; }; }
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1083,7 +1092,8 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*with-\([^=]*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid package name: $ac_useropt"
+      { $as_echo "$as_me: error: invalid package name: $ac_useropt" >&2
+   { (exit 1); exit 1; }; }
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1099,7 +1109,8 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*without-\(.*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid package name: $ac_useropt"
+      { $as_echo "$as_me: error: invalid package name: $ac_useropt" >&2
+   { (exit 1); exit 1; }; }
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1129,17 +1140,17 @@ do
   | --x-librar=* | --x-libra=* | --x-libr=* | --x-lib=* | --x-li=* | --x-l=*)
     x_libraries=$ac_optarg ;;
 
-  -*) as_fn_error $? "unrecognized option: \`$ac_option'
-Try \`$0 --help' for more information"
+  -*) { $as_echo "$as_me: error: unrecognized option: $ac_option
+Try \`$0 --help' for more information." >&2
+   { (exit 1); exit 1; }; }
     ;;
 
   *=*)
     ac_envvar=`expr "x$ac_option" : 'x\([^=]*\)='`
     # Reject names that are not valid shell variable names.
-    case $ac_envvar in #(
-      '' | [0-9]* | *[!_$as_cr_alnum]* )
-      as_fn_error $? "invalid variable name: \`$ac_envvar'" ;;
-    esac
+    expr "x$ac_envvar" : ".*[^_$as_cr_alnum]" >/dev/null &&
+      { $as_echo "$as_me: error: invalid variable name: $ac_envvar" >&2
+   { (exit 1); exit 1; }; }
     eval $ac_envvar=\$ac_optarg
     export $ac_envvar ;;
 
@@ -1148,7 +1159,7 @@ Try \`$0 --help' for more information"
     $as_echo "$as_me: WARNING: you should use --build, --host, --target" >&2
     expr "x$ac_option" : ".*[^-._$as_cr_alnum]" >/dev/null &&
       $as_echo "$as_me: WARNING: invalid host type: $ac_option" >&2
-    : "${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}"
+    : ${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}
     ;;
 
   esac
@@ -1156,13 +1167,15 @@ done
 
 if test -n "$ac_prev"; then
   ac_option=--`echo $ac_prev | sed 's/_/-/g'`
-  as_fn_error $? "missing argument to $ac_option"
+  { $as_echo "$as_me: error: missing argument to $ac_option" >&2
+   { (exit 1); exit 1; }; }
 fi
 
 if test -n "$ac_unrecognized_opts"; then
   case $enable_option_checking in
     no) ;;
-    fatal) as_fn_error $? "unrecognized options: $ac_unrecognized_opts" ;;
+    fatal) { $as_echo "$as_me: error: unrecognized options: $ac_unrecognized_opts" >&2
+   { (exit 1); exit 1; }; } ;;
     *)     $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2 ;;
   esac
 fi
@@ -1185,7 +1198,8 @@ do
     [\\/$]* | ?:[\\/]* )  continue;;
     NONE | '' ) case $ac_var in *prefix ) continue;; esac;;
   esac
-  as_fn_error $? "expected an absolute directory name for --$ac_var: $ac_val"
+  { $as_echo "$as_me: error: expected an absolute directory name for --$ac_var: $ac_val" >&2
+   { (exit 1); exit 1; }; }
 done
 
 # There might be people who depend on the old broken behavior: `$host'
@@ -1199,6 +1213,8 @@ target=$target_alias
 if test "x$host_alias" != x; then
   if test "x$build_alias" = x; then
     cross_compiling=maybe
+    $as_echo "$as_me: WARNING: If you wanted to set the --build type, don't use --host.
+    If a cross compiler is detected then cross compile mode will be used." >&2
   elif test "x$build_alias" != "x$host_alias"; then
     cross_compiling=yes
   fi
@@ -1213,9 +1229,11 @@ test "$silent" = yes && exec 6>/dev/null
 ac_pwd=`pwd` && test -n "$ac_pwd" &&
 ac_ls_di=`ls -di .` &&
 ac_pwd_ls_di=`cd "$ac_pwd" && ls -di .` ||
-  as_fn_error $? "working directory cannot be determined"
+  { $as_echo "$as_me: error: working directory cannot be determined" >&2
+   { (exit 1); exit 1; }; }
 test "X$ac_ls_di" = "X$ac_pwd_ls_di" ||
-  as_fn_error $? "pwd does not report name of working directory"
+  { $as_echo "$as_me: error: pwd does not report name of working directory" >&2
+   { (exit 1); exit 1; }; }
 
 
 # Find the source files, if location was not specified.
@@ -1254,11 +1272,13 @@ else
 fi
 if test ! -r "$srcdir/$ac_unique_file"; then
   test "$ac_srcdir_defaulted" = yes && srcdir="$ac_confdir or .."
-  as_fn_error $? "cannot find sources ($ac_unique_file) in $srcdir"
+  { $as_echo "$as_me: error: cannot find sources ($ac_unique_file) in $srcdir" >&2
+   { (exit 1); exit 1; }; }
 fi
 ac_msg="sources are in $srcdir, but \`cd $srcdir' does not work"
 ac_abs_confdir=`(
-	cd "$srcdir" && test -r "./$ac_unique_file" || as_fn_error $? "$ac_msg"
+	cd "$srcdir" && test -r "./$ac_unique_file" || { $as_echo "$as_me: error: $ac_msg" >&2
+   { (exit 1); exit 1; }; }
 	pwd)`
 # When building in place, set srcdir=.
 if test "$ac_abs_confdir" = "$ac_pwd"; then
@@ -1298,7 +1318,7 @@ Configuration:
       --help=short        display options specific to this package
       --help=recursive    display the short help of all the included packages
   -V, --version           display version information and exit
-  -q, --quiet, --silent   do not print \`checking ...' messages
+  -q, --quiet, --silent   do not print \`checking...' messages
       --cache-file=FILE   cache test results in FILE [disabled]
   -C, --config-cache      alias for \`--cache-file=config.cache'
   -n, --no-create         do not create output files
@@ -1358,12 +1378,8 @@ Optional Features:
   --disable-option-checking  ignore unrecognized --enable/--with options
   --disable-FEATURE       do not include FEATURE (same as --enable-FEATURE=no)
   --enable-FEATURE[=ARG]  include FEATURE [ARG=yes]
-  --enable-silent-rules   less verbose build output (undo: "make V=1")
-  --disable-silent-rules  verbose build output (undo: "make V=0")
-  --enable-dependency-tracking
-                          do not reject slow dependency extractors
-  --disable-dependency-tracking
-                          speeds up one-time build
+  --disable-dependency-tracking  speeds up one-time build
+  --enable-dependency-tracking   do not reject slow dependency extractors
   --disable-openmp        do not use OpenMP
   --enable-simd=SSE|AVX|AVX2|AVX512|MIC
                           Select instructions to be SSE4.0, AVX 1.0, AVX
@@ -1376,7 +1392,7 @@ Some influential environment variables:
   LDFLAGS     linker flags, e.g. -L<lib dir> if you have libraries in a
               nonstandard directory <lib dir>
   LIBS        libraries to pass to the linker, e.g. -l<library>
-  CPPFLAGS    (Objective) C/C++ preprocessor flags, e.g. -I<include dir> if
+  CPPFLAGS    C/C++/Objective C preprocessor flags, e.g. -I<include dir> if
               you have headers in a nonstandard directory <include dir>
   CC          C compiler command
   CFLAGS      C compiler flags
@@ -1449,568 +1465,21 @@ test -n "$ac_init_help" && exit $ac_status
 if $ac_init_version; then
   cat <<\_ACEOF
 Grid configure 1.0
-generated by GNU Autoconf 2.69
+generated by GNU Autoconf 2.63
 
-Copyright (C) 2012 Free Software Foundation, Inc.
+Copyright (C) 1992, 1993, 1994, 1995, 1996, 1998, 1999, 2000, 2001,
+2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
 This configure script is free software; the Free Software Foundation
 gives unlimited permission to copy, distribute and modify it.
 _ACEOF
   exit
 fi
-
-## ------------------------ ##
-## Autoconf initialization. ##
-## ------------------------ ##
-
-# ac_fn_cxx_try_compile LINENO
-# ----------------------------
-# Try to compile conftest.$ac_ext, and return whether this succeeded.
-ac_fn_cxx_try_compile ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  rm -f conftest.$ac_objext
-  if { { ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compile") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_retval=1
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_cxx_try_compile
-
-# ac_fn_c_try_compile LINENO
-# --------------------------
-# Try to compile conftest.$ac_ext, and return whether this succeeded.
-ac_fn_c_try_compile ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  rm -f conftest.$ac_objext
-  if { { ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compile") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_retval=1
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_c_try_compile
-
-# ac_fn_c_try_link LINENO
-# -----------------------
-# Try to link conftest.$ac_ext, and return whether this succeeded.
-ac_fn_c_try_link ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  rm -f conftest.$ac_objext conftest$ac_exeext
-  if { { ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_link") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 test -x conftest$ac_exeext
-       }; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_retval=1
-fi
-  # Delete the IPA/IPO (Inter Procedural Analysis/Optimization) information
-  # created by the PGI compiler (conftest_ipa8_conftest.oo), as it would
-  # interfere with the next link command; also delete a directory that is
-  # left behind by Apple's compiler.  We do this before executing the actions.
-  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_c_try_link
-
-# ac_fn_c_try_cpp LINENO
-# ----------------------
-# Try to preprocess conftest.$ac_ext, and return whether this succeeded.
-ac_fn_c_try_cpp ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  if { { ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } > conftest.i && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-    ac_retval=1
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_c_try_cpp
-
-# ac_fn_c_check_header_mongrel LINENO HEADER VAR INCLUDES
-# -------------------------------------------------------
-# Tests whether HEADER exists, giving a warning if it cannot be compiled using
-# the include files in INCLUDES and setting the cache variable VAR
-# accordingly.
-ac_fn_c_check_header_mongrel ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  if eval \${$3+:} false; then :
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 usability" >&5
-$as_echo_n "checking $2 usability... " >&6; }
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-#include <$2>
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  ac_header_compiler=yes
-else
-  ac_header_compiler=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 presence" >&5
-$as_echo_n "checking $2 presence... " >&6; }
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <$2>
-_ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
-  ac_header_preproc=yes
-else
-  ac_header_preproc=no
-fi
-rm -f conftest.err conftest.i conftest.$ac_ext
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in #((
-  yes:no: )
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $2: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $2:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $2: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
-( $as_echo "## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##"
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  eval "$3=\$ac_header_compiler"
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_check_header_mongrel
-
-# ac_fn_c_try_run LINENO
-# ----------------------
-# Try to link conftest.$ac_ext, and return whether this succeeded. Assumes
-# that executables *can* be run.
-ac_fn_c_try_run ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  if { { ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_link") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && { ac_try='./conftest$ac_exeext'
-  { { case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; }; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: program exited with status $ac_status" >&5
-       $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-       ac_retval=$ac_status
-fi
-  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_c_try_run
-
-# ac_fn_c_check_header_compile LINENO HEADER VAR INCLUDES
-# -------------------------------------------------------
-# Tests whether HEADER exists and can be compiled using the include files in
-# INCLUDES, setting the cache variable VAR accordingly.
-ac_fn_c_check_header_compile ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-#include <$2>
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  eval "$3=yes"
-else
-  eval "$3=no"
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_check_header_compile
-
-# ac_fn_c_check_decl LINENO SYMBOL VAR INCLUDES
-# ---------------------------------------------
-# Tests whether SYMBOL is declared in INCLUDES, setting cache variable VAR
-# accordingly.
-ac_fn_c_check_decl ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  as_decl_name=`echo $2|sed 's/ *(.*//'`
-  as_decl_use=`echo $2|sed -e 's/(/((/' -e 's/)/) 0&/' -e 's/,/) 0& (/g'`
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $as_decl_name is declared" >&5
-$as_echo_n "checking whether $as_decl_name is declared... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-int
-main ()
-{
-#ifndef $as_decl_name
-#ifdef __cplusplus
-  (void) $as_decl_use;
-#else
-  (void) $as_decl_name;
-#endif
-#endif
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  eval "$3=yes"
-else
-  eval "$3=no"
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_check_decl
-
-# ac_fn_c_check_type LINENO TYPE VAR INCLUDES
-# -------------------------------------------
-# Tests whether TYPE exists after having included INCLUDES, setting cache
-# variable VAR accordingly.
-ac_fn_c_check_type ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  eval "$3=no"
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-int
-main ()
-{
-if (sizeof ($2))
-	 return 0;
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-int
-main ()
-{
-if (sizeof (($2)))
-	    return 0;
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-
-else
-  eval "$3=yes"
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_check_type
-
-# ac_fn_c_find_uintX_t LINENO BITS VAR
-# ------------------------------------
-# Finds an unsigned integer type with width BITS, setting cache variable VAR
-# accordingly.
-ac_fn_c_find_uintX_t ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for uint$2_t" >&5
-$as_echo_n "checking for uint$2_t... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  eval "$3=no"
-     # Order is important - never check a type that is potentially smaller
-     # than half of the expected target width.
-     for ac_type in uint$2_t 'unsigned int' 'unsigned long int' \
-	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
-       cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-static int test_array [1 - 2 * !((($ac_type) -1 >> ($2 / 2 - 1)) >> ($2 / 2 - 1) == 3)];
-test_array [0] = 0;
-return test_array [0];
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  case $ac_type in #(
-  uint$2_t) :
-    eval "$3=yes" ;; #(
-  *) :
-    eval "$3=\$ac_type" ;;
-esac
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-       if eval test \"x\$"$3"\" = x"no"; then :
-
-else
-  break
-fi
-     done
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_find_uintX_t
-
-# ac_fn_c_check_func LINENO FUNC VAR
-# ----------------------------------
-# Tests whether FUNC exists, setting the cache variable VAR accordingly
-ac_fn_c_check_func ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-/* Define $2 to an innocuous variant, in case <limits.h> declares $2.
-   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
-#define $2 innocuous_$2
-
-/* System header to define __stub macros and hopefully few prototypes,
-    which can conflict with char $2 (); below.
-    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
-    <limits.h> exists even on freestanding compilers.  */
-
-#ifdef __STDC__
-# include <limits.h>
-#else
-# include <assert.h>
-#endif
-
-#undef $2
-
-/* Override any GCC internal prototype to avoid an error.
-   Use char because int might match the return type of a GCC
-   builtin and then its argument prototype would still apply.  */
-#ifdef __cplusplus
-extern "C"
-#endif
-char $2 ();
-/* The GNU C library defines this for functions which it implements
-    to always fail with ENOSYS.  Some functions are actually named
-    something starting with __ and the normal name is an alias.  */
-#if defined __stub_$2 || defined __stub___$2
-choke me
-#endif
-
-int
-main ()
-{
-return $2 ();
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_link "$LINENO"; then :
-  eval "$3=yes"
-else
-  eval "$3=no"
-fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_check_func
 cat >config.log <<_ACEOF
 This file contains any messages produced by compilers while
 running configure, to aid debugging if configure makes a mistake.
 
 It was created by Grid $as_me 1.0, which was
-generated by GNU Autoconf 2.69.  Invocation command line was
+generated by GNU Autoconf 2.63.  Invocation command line was
 
   $ $0 $@
 
@@ -2046,8 +1515,8 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    $as_echo "PATH: $as_dir"
-  done
+  $as_echo "PATH: $as_dir"
+done
 IFS=$as_save_IFS
 
 } >&5
@@ -2084,9 +1553,9 @@ do
       ac_arg=`$as_echo "$ac_arg" | sed "s/'/'\\\\\\\\''/g"` ;;
     esac
     case $ac_pass in
-    1) as_fn_append ac_configure_args0 " '$ac_arg'" ;;
+    1) ac_configure_args0="$ac_configure_args0 '$ac_arg'" ;;
     2)
-      as_fn_append ac_configure_args1 " '$ac_arg'"
+      ac_configure_args1="$ac_configure_args1 '$ac_arg'"
       if test $ac_must_keep_next = true; then
 	ac_must_keep_next=false # Got value, back to normal.
       else
@@ -2102,13 +1571,13 @@ do
 	  -* ) ac_must_keep_next=true ;;
 	esac
       fi
-      as_fn_append ac_configure_args " '$ac_arg'"
+      ac_configure_args="$ac_configure_args '$ac_arg'"
       ;;
     esac
   done
 done
-{ ac_configure_args0=; unset ac_configure_args0;}
-{ ac_configure_args1=; unset ac_configure_args1;}
+$as_unset ac_configure_args0 || test "${ac_configure_args0+set}" != set || { ac_configure_args0=; export ac_configure_args0; }
+$as_unset ac_configure_args1 || test "${ac_configure_args1+set}" != set || { ac_configure_args1=; export ac_configure_args1; }
 
 # When interrupted or exit'd, cleanup temporary files, and complete
 # config.log.  We remove comments because anyway the quotes in there
@@ -2120,9 +1589,11 @@ trap 'exit_status=$?
   {
     echo
 
-    $as_echo "## ---------------- ##
+    cat <<\_ASBOX
+## ---------------- ##
 ## Cache variables. ##
-## ---------------- ##"
+## ---------------- ##
+_ASBOX
     echo
     # The following way of writing the cache mishandles newlines in values,
 (
@@ -2131,13 +1602,13 @@ trap 'exit_status=$?
     case $ac_val in #(
     *${as_nl}*)
       case $ac_var in #(
-      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
+      *_cv_*) { $as_echo "$as_me:$LINENO: WARNING: cache variable $ac_var contains a newline" >&5
 $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
       esac
       case $ac_var in #(
       _ | IFS | as_nl) ;; #(
       BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
-      *) { eval $ac_var=; unset $ac_var;} ;;
+      *) $as_unset $ac_var ;;
       esac ;;
     esac
   done
@@ -2156,9 +1627,11 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
 )
     echo
 
-    $as_echo "## ----------------- ##
+    cat <<\_ASBOX
+## ----------------- ##
 ## Output variables. ##
-## ----------------- ##"
+## ----------------- ##
+_ASBOX
     echo
     for ac_var in $ac_subst_vars
     do
@@ -2171,9 +1644,11 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
     echo
 
     if test -n "$ac_subst_files"; then
-      $as_echo "## ------------------- ##
+      cat <<\_ASBOX
+## ------------------- ##
 ## File substitutions. ##
-## ------------------- ##"
+## ------------------- ##
+_ASBOX
       echo
       for ac_var in $ac_subst_files
       do
@@ -2187,9 +1662,11 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
     fi
 
     if test -s confdefs.h; then
-      $as_echo "## ----------- ##
+      cat <<\_ASBOX
+## ----------- ##
 ## confdefs.h. ##
-## ----------- ##"
+## ----------- ##
+_ASBOX
       echo
       cat confdefs.h
       echo
@@ -2203,39 +1680,37 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
     exit $exit_status
 ' 0
 for ac_signal in 1 2 13 15; do
-  trap 'ac_signal='$ac_signal'; as_fn_exit 1' $ac_signal
+  trap 'ac_signal='$ac_signal'; { (exit 1); exit 1; }' $ac_signal
 done
 ac_signal=0
 
 # confdefs.h avoids OS command line length limits that DEFS can exceed.
 rm -f -r conftest* confdefs.h
 
-$as_echo "/* confdefs.h */" > confdefs.h
-
 # Predefined preprocessor variables.
 
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_NAME "$PACKAGE_NAME"
 _ACEOF
 
+
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_TARNAME "$PACKAGE_TARNAME"
 _ACEOF
 
+
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_VERSION "$PACKAGE_VERSION"
 _ACEOF
 
+
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_STRING "$PACKAGE_STRING"
 _ACEOF
 
-cat >>confdefs.h <<_ACEOF
-#define PACKAGE_BUGREPORT "$PACKAGE_BUGREPORT"
-_ACEOF
 
 cat >>confdefs.h <<_ACEOF
-#define PACKAGE_URL "$PACKAGE_URL"
+#define PACKAGE_BUGREPORT "$PACKAGE_BUGREPORT"
 _ACEOF
 
 
@@ -2244,12 +1719,7 @@ _ACEOF
 ac_site_file1=NONE
 ac_site_file2=NONE
 if test -n "$CONFIG_SITE"; then
-  # We do not want a PATH search for config.site.
-  case $CONFIG_SITE in #((
-    -*)  ac_site_file1=./$CONFIG_SITE;;
-    */*) ac_site_file1=$CONFIG_SITE;;
-    *)   ac_site_file1=./$CONFIG_SITE;;
-  esac
+  ac_site_file1=$CONFIG_SITE
 elif test "x$prefix" != xNONE; then
   ac_site_file1=$prefix/share/config.site
   ac_site_file2=$prefix/etc/config.site
@@ -2260,23 +1730,19 @@ fi
 for ac_site_file in "$ac_site_file1" "$ac_site_file2"
 do
   test "x$ac_site_file" = xNONE && continue
-  if test /dev/null != "$ac_site_file" && test -r "$ac_site_file"; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: loading site script $ac_site_file" >&5
+  if test -r "$ac_site_file"; then
+    { $as_echo "$as_me:$LINENO: loading site script $ac_site_file" >&5
 $as_echo "$as_me: loading site script $ac_site_file" >&6;}
     sed 's/^/| /' "$ac_site_file" >&5
-    . "$ac_site_file" \
-      || { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "failed to load site script $ac_site_file
-See \`config.log' for more details" "$LINENO" 5; }
+    . "$ac_site_file"
   fi
 done
 
 if test -r "$cache_file"; then
-  # Some versions of bash will fail to source /dev/null (special files
-  # actually), so we avoid doing that.  DJGPP emulates it as a regular file.
-  if test /dev/null != "$cache_file" && test -f "$cache_file"; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: loading cache $cache_file" >&5
+  # Some versions of bash will fail to source /dev/null (special
+  # files actually), so we avoid doing that.
+  if test -f "$cache_file"; then
+    { $as_echo "$as_me:$LINENO: loading cache $cache_file" >&5
 $as_echo "$as_me: loading cache $cache_file" >&6;}
     case $cache_file in
       [\\/]* | ?:[\\/]* ) . "$cache_file";;
@@ -2284,7 +1750,7 @@ $as_echo "$as_me: loading cache $cache_file" >&6;}
     esac
   fi
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: creating cache $cache_file" >&5
+  { $as_echo "$as_me:$LINENO: creating cache $cache_file" >&5
 $as_echo "$as_me: creating cache $cache_file" >&6;}
   >$cache_file
 fi
@@ -2299,11 +1765,11 @@ for ac_var in $ac_precious_vars; do
   eval ac_new_val=\$ac_env_${ac_var}_value
   case $ac_old_set,$ac_new_set in
     set,)
-      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
+      { $as_echo "$as_me:$LINENO: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
 $as_echo "$as_me: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&2;}
       ac_cache_corrupted=: ;;
     ,set)
-      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was not set in the previous run" >&5
+      { $as_echo "$as_me:$LINENO: error: \`$ac_var' was not set in the previous run" >&5
 $as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
       ac_cache_corrupted=: ;;
     ,);;
@@ -2313,17 +1779,17 @@ $as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
 	ac_old_val_w=`echo x $ac_old_val`
 	ac_new_val_w=`echo x $ac_new_val`
 	if test "$ac_old_val_w" != "$ac_new_val_w"; then
-	  { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' has changed since the previous run:" >&5
+	  { $as_echo "$as_me:$LINENO: error: \`$ac_var' has changed since the previous run:" >&5
 $as_echo "$as_me: error: \`$ac_var' has changed since the previous run:" >&2;}
 	  ac_cache_corrupted=:
 	else
-	  { $as_echo "$as_me:${as_lineno-$LINENO}: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
+	  { $as_echo "$as_me:$LINENO: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
 $as_echo "$as_me: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&2;}
 	  eval $ac_var=\$ac_old_val
 	fi
-	{ $as_echo "$as_me:${as_lineno-$LINENO}:   former value:  \`$ac_old_val'" >&5
+	{ $as_echo "$as_me:$LINENO:   former value:  \`$ac_old_val'" >&5
 $as_echo "$as_me:   former value:  \`$ac_old_val'" >&2;}
-	{ $as_echo "$as_me:${as_lineno-$LINENO}:   current value: \`$ac_new_val'" >&5
+	{ $as_echo "$as_me:$LINENO:   current value: \`$ac_new_val'" >&5
 $as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
       fi;;
   esac
@@ -2335,20 +1801,43 @@ $as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
     esac
     case " $ac_configure_args " in
       *" '$ac_arg' "*) ;; # Avoid dups.  Use of quotes ensures accuracy.
-      *) as_fn_append ac_configure_args " '$ac_arg'" ;;
+      *) ac_configure_args="$ac_configure_args '$ac_arg'" ;;
     esac
   fi
 done
 if $ac_cache_corrupted; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+  { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-  { $as_echo "$as_me:${as_lineno-$LINENO}: error: changes in the environment can compromise the build" >&5
+  { $as_echo "$as_me:$LINENO: error: changes in the environment can compromise the build" >&5
 $as_echo "$as_me: error: changes in the environment can compromise the build" >&2;}
-  as_fn_error $? "run \`make distclean' and/or \`rm $cache_file' and start over" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: run \`make distclean' and/or \`rm $cache_file' and start over" >&5
+$as_echo "$as_me: error: run \`make distclean' and/or \`rm $cache_file' and start over" >&2;}
+   { (exit 1); exit 1; }; }
 fi
-## -------------------- ##
-## Main body of script. ##
-## -------------------- ##
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
 
 ac_ext=c
 ac_cpp='$CPP $CPPFLAGS'
@@ -2357,7 +1846,7 @@ ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
 
-am__api_version='1.15'
+am__api_version='1.11'
 
 ac_aux_dir=
 for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
@@ -2376,7 +1865,9 @@ for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
   fi
 done
 if test -z "$ac_aux_dir"; then
-  as_fn_error $? "cannot find install-sh, install.sh, or shtool in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: cannot find install-sh or install.sh in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" >&5
+$as_echo "$as_me: error: cannot find install-sh or install.sh in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" >&2;}
+   { (exit 1); exit 1; }; }
 fi
 
 # These three variables are undocumented and unsupported,
@@ -2402,10 +1893,10 @@ ac_configure="$SHELL $ac_aux_dir/configure"  # Please don't use this var.
 # OS/2's system install, which has a completely different semantic
 # ./install, which can be erroneously created by make from ./install.sh.
 # Reject install programs that cannot install multiple files.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a BSD-compatible install" >&5
+{ $as_echo "$as_me:$LINENO: checking for a BSD-compatible install" >&5
 $as_echo_n "checking for a BSD-compatible install... " >&6; }
 if test -z "$INSTALL"; then
-if ${ac_cv_path_install+:} false; then :
+if test "${ac_cv_path_install+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
@@ -2413,11 +1904,11 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    # Account for people who put trailing slashes in PATH elements.
-case $as_dir/ in #((
-  ./ | .// | /[cC]/* | \
+  # Account for people who put trailing slashes in PATH elements.
+case $as_dir/ in
+  ./ | .// | /cC/* | \
   /etc/* | /usr/sbin/* | /usr/etc/* | /sbin/* | /usr/afsws/bin/* | \
-  ?:[\\/]os2[\\/]install[\\/]* | ?:[\\/]OS2[\\/]INSTALL[\\/]* | \
+  ?:\\/os2\\/install\\/* | ?:\\/OS2\\/INSTALL\\/* | \
   /usr/ucb/* ) ;;
   *)
     # OSF1 and SCO ODT 3.0 have their own names for install.
@@ -2425,7 +1916,7 @@ case $as_dir/ in #((
     # by default.
     for ac_prog in ginstall scoinst install; do
       for ac_exec_ext in '' $ac_executable_extensions; do
-	if as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext"; then
+	if { test -f "$as_dir/$ac_prog$ac_exec_ext" && $as_test_x "$as_dir/$ac_prog$ac_exec_ext"; }; then
 	  if test $ac_prog = install &&
 	    grep dspmsg "$as_dir/$ac_prog$ac_exec_ext" >/dev/null 2>&1; then
 	    # AIX install.  It has an incompatible calling convention.
@@ -2454,7 +1945,7 @@ case $as_dir/ in #((
     ;;
 esac
 
-  done
+done
 IFS=$as_save_IFS
 
 rm -rf conftest.one conftest.two conftest.dir
@@ -2470,7 +1961,7 @@ fi
     INSTALL=$ac_install_sh
   fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $INSTALL" >&5
+{ $as_echo "$as_me:$LINENO: result: $INSTALL" >&5
 $as_echo "$INSTALL" >&6; }
 
 # Use test -z because SunOS4 sh mishandles braces in ${var-val}.
@@ -2481,73 +1972,68 @@ test -z "$INSTALL_SCRIPT" && INSTALL_SCRIPT='${INSTALL}'
 
 test -z "$INSTALL_DATA" && INSTALL_DATA='${INSTALL} -m 644'
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether build environment is sane" >&5
+{ $as_echo "$as_me:$LINENO: checking whether build environment is sane" >&5
 $as_echo_n "checking whether build environment is sane... " >&6; }
+# Just in case
+sleep 1
+echo timestamp > conftest.file
 # Reject unsafe characters in $srcdir or the absolute working directory
 # name.  Accept space and tab only in the latter.
 am_lf='
 '
 case `pwd` in
   *[\\\"\#\$\&\'\`$am_lf]*)
-    as_fn_error $? "unsafe absolute working directory name" "$LINENO" 5;;
+    { { $as_echo "$as_me:$LINENO: error: unsafe absolute working directory name" >&5
+$as_echo "$as_me: error: unsafe absolute working directory name" >&2;}
+   { (exit 1); exit 1; }; };;
 esac
 case $srcdir in
   *[\\\"\#\$\&\'\`$am_lf\ \	]*)
-    as_fn_error $? "unsafe srcdir value: '$srcdir'" "$LINENO" 5;;
+    { { $as_echo "$as_me:$LINENO: error: unsafe srcdir value: \`$srcdir'" >&5
+$as_echo "$as_me: error: unsafe srcdir value: \`$srcdir'" >&2;}
+   { (exit 1); exit 1; }; };;
 esac
 
-# Do 'set' in a subshell so we don't clobber the current shell's
+# Do `set' in a subshell so we don't clobber the current shell's
 # arguments.  Must try -L first in case configure is actually a
 # symlink; some systems play weird games with the mod time of symlinks
 # (eg FreeBSD returns the mod time of the symlink's containing
 # directory).
 if (
-   am_has_slept=no
-   for am_try in 1 2; do
-     echo "timestamp, slept: $am_has_slept" > conftest.file
-     set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
-     if test "$*" = "X"; then
-	# -L didn't work.
-	set X `ls -t "$srcdir/configure" conftest.file`
-     fi
-     if test "$*" != "X $srcdir/configure conftest.file" \
-	&& test "$*" != "X conftest.file $srcdir/configure"; then
+   set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
+   if test "$*" = "X"; then
+      # -L didn't work.
+      set X `ls -t "$srcdir/configure" conftest.file`
+   fi
+   rm -f conftest.file
+   if test "$*" != "X $srcdir/configure conftest.file" \
+      && test "$*" != "X conftest.file $srcdir/configure"; then
+
+      # If neither matched, then we have a broken ls.  This can happen
+      # if, for instance, CONFIG_SHELL is bash and it inherits a
+      # broken ls alias from the environment.  This has actually
+      # happened.  Such a system could not be considered "sane".
+      { { $as_echo "$as_me:$LINENO: error: ls -t appears to fail.  Make sure there is not a broken
+alias in your environment" >&5
+$as_echo "$as_me: error: ls -t appears to fail.  Make sure there is not a broken
+alias in your environment" >&2;}
+   { (exit 1); exit 1; }; }
+   fi
 
-	# If neither matched, then we have a broken ls.  This can happen
-	# if, for instance, CONFIG_SHELL is bash and it inherits a
-	# broken ls alias from the environment.  This has actually
-	# happened.  Such a system could not be considered "sane".
-	as_fn_error $? "ls -t appears to fail.  Make sure there is not a broken
-  alias in your environment" "$LINENO" 5
-     fi
-     if test "$2" = conftest.file || test $am_try -eq 2; then
-       break
-     fi
-     # Just in case.
-     sleep 1
-     am_has_slept=yes
-   done
    test "$2" = conftest.file
    )
 then
    # Ok.
    :
 else
-   as_fn_error $? "newly created file is older than distributed files!
-Check your system clock" "$LINENO" 5
+   { { $as_echo "$as_me:$LINENO: error: newly created file is older than distributed files!
+Check your system clock" >&5
+$as_echo "$as_me: error: newly created file is older than distributed files!
+Check your system clock" >&2;}
+   { (exit 1); exit 1; }; }
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+{ $as_echo "$as_me:$LINENO: result: yes" >&5
 $as_echo "yes" >&6; }
-# If we didn't sleep, we still need to ensure time stamps of config.status and
-# generated files are strictly newer.
-am_sleep_pid=
-if grep 'slept: no' conftest.file >/dev/null 2>&1; then
-  ( sleep 1 ) &
-  am_sleep_pid=$!
-fi
-
-rm -f conftest.file
-
 test "$program_prefix" != NONE &&
   program_transform_name="s&^&$program_prefix&;$program_transform_name"
 # Use a double $ so make ignores it.
@@ -2558,8 +2044,8 @@ test "$program_suffix" != NONE &&
 ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
 program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
 
-# Expand $ac_aux_dir to an absolute path.
-am_aux_dir=`cd "$ac_aux_dir" && pwd`
+# expand $ac_aux_dir to an absolute path
+am_aux_dir=`cd $ac_aux_dir && pwd`
 
 if test x"${MISSING+set}" != xset; then
   case $am_aux_dir in
@@ -2570,15 +2056,15 @@ if test x"${MISSING+set}" != xset; then
   esac
 fi
 # Use eval to expand $SHELL
-if eval "$MISSING --is-lightweight"; then
-  am_missing_run="$MISSING "
+if eval "$MISSING --run true"; then
+  am_missing_run="$MISSING --run "
 else
   am_missing_run=
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: 'missing' script is too old or missing" >&5
-$as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
+  { $as_echo "$as_me:$LINENO: WARNING: \`missing' script is too old or missing" >&5
+$as_echo "$as_me: WARNING: \`missing' script is too old or missing" >&2;}
 fi
 
-if test x"${install_sh+set}" != xset; then
+if test x"${install_sh}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -2587,17 +2073,17 @@ if test x"${install_sh+set}" != xset; then
   esac
 fi
 
-# Installed binaries are usually stripped using 'strip' when the user
-# run "make install-strip".  However 'strip' might not be the right
+# Installed binaries are usually stripped using `strip' when the user
+# run `make install-strip'.  However `strip' might not be the right
 # tool to use in cross-compilation environments, therefore Automake
-# will honor the 'STRIP' environment variable to overrule this program.
+# will honor the `STRIP' environment variable to overrule this program.
 if test "$cross_compiling" != no; then
   if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}strip", so it can be a program name with args.
 set dummy ${ac_tool_prefix}strip; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_STRIP+:} false; then :
+if test "${ac_cv_prog_STRIP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$STRIP"; then
@@ -2608,24 +2094,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_STRIP="${ac_tool_prefix}strip"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 STRIP=$ac_cv_prog_STRIP
 if test -n "$STRIP"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $STRIP" >&5
+  { $as_echo "$as_me:$LINENO: result: $STRIP" >&5
 $as_echo "$STRIP" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2635,9 +2121,9 @@ if test -z "$ac_cv_prog_STRIP"; then
   ac_ct_STRIP=$STRIP
   # Extract the first word of "strip", so it can be a program name with args.
 set dummy strip; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_STRIP+:} false; then :
+if test "${ac_cv_prog_ac_ct_STRIP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_STRIP"; then
@@ -2648,24 +2134,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_ac_ct_STRIP="strip"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_STRIP=$ac_cv_prog_ac_ct_STRIP
 if test -n "$ac_ct_STRIP"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_STRIP" >&5
+  { $as_echo "$as_me:$LINENO: result: $ac_ct_STRIP" >&5
 $as_echo "$ac_ct_STRIP" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2674,7 +2160,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -2687,10 +2173,10 @@ fi
 fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a thread-safe mkdir -p" >&5
+{ $as_echo "$as_me:$LINENO: checking for a thread-safe mkdir -p" >&5
 $as_echo_n "checking for a thread-safe mkdir -p... " >&6; }
 if test -z "$MKDIR_P"; then
-  if ${ac_cv_path_mkdir+:} false; then :
+  if test "${ac_cv_path_mkdir+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
@@ -2698,9 +2184,9 @@ for as_dir in $PATH$PATH_SEPARATOR/opt/sfw/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_prog in mkdir gmkdir; do
+  for ac_prog in mkdir gmkdir; do
 	 for ac_exec_ext in '' $ac_executable_extensions; do
-	   as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext" || continue
+	   { test -f "$as_dir/$ac_prog$ac_exec_ext" && $as_test_x "$as_dir/$ac_prog$ac_exec_ext"; } || continue
 	   case `"$as_dir/$ac_prog$ac_exec_ext" --version 2>&1` in #(
 	     'mkdir (GNU coreutils) '* | \
 	     'mkdir (coreutils) '* | \
@@ -2710,12 +2196,11 @@ do
 	   esac
 	 done
        done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 
-  test -d ./--version && rmdir ./--version
   if test "${ac_cv_path_mkdir+set}" = set; then
     MKDIR_P="$ac_cv_path_mkdir -p"
   else
@@ -2723,19 +2208,26 @@ fi
     # value for MKDIR_P within a source directory, because that will
     # break other packages using the cache if that directory is
     # removed, or if the value is a relative name.
+    test -d ./--version && rmdir ./--version
     MKDIR_P="$ac_install_sh -d"
   fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $MKDIR_P" >&5
+{ $as_echo "$as_me:$LINENO: result: $MKDIR_P" >&5
 $as_echo "$MKDIR_P" >&6; }
 
+mkdir_p="$MKDIR_P"
+case $mkdir_p in
+  [\\/$]* | ?:[\\/]*) ;;
+  */*) mkdir_p="\$(top_builddir)/$mkdir_p" ;;
+esac
+
 for ac_prog in gawk mawk nawk awk
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_AWK+:} false; then :
+if test "${ac_cv_prog_AWK+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$AWK"; then
@@ -2746,24 +2238,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_AWK="$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 AWK=$ac_cv_prog_AWK
 if test -n "$AWK"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $AWK" >&5
+  { $as_echo "$as_me:$LINENO: result: $AWK" >&5
 $as_echo "$AWK" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2771,11 +2263,11 @@ fi
   test -n "$AWK" && break
 done
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether ${MAKE-make} sets \$(MAKE)" >&5
+{ $as_echo "$as_me:$LINENO: checking whether ${MAKE-make} sets \$(MAKE)" >&5
 $as_echo_n "checking whether ${MAKE-make} sets \$(MAKE)... " >&6; }
 set x ${MAKE-make}
 ac_make=`$as_echo "$2" | sed 's/+/p/g; s/[^a-zA-Z0-9_]/_/g'`
-if eval \${ac_cv_prog_make_${ac_make}_set+:} false; then :
+if { as_var=ac_cv_prog_make_${ac_make}_set; eval "test \"\${$as_var+set}\" = set"; }; then
   $as_echo_n "(cached) " >&6
 else
   cat >conftest.make <<\_ACEOF
@@ -2783,7 +2275,7 @@ SHELL = /bin/sh
 all:
 	@echo '@@@%%%=$(MAKE)=@@@%%%'
 _ACEOF
-# GNU make sometimes prints "make[1]: Entering ...", which would confuse us.
+# GNU make sometimes prints "make[1]: Entering...", which would confuse us.
 case `${MAKE-make} -f conftest.make 2>/dev/null` in
   *@@@%%%=?*=@@@%%%*)
     eval ac_cv_prog_make_${ac_make}_set=yes;;
@@ -2793,11 +2285,11 @@ esac
 rm -f conftest.make
 fi
 if eval test \$ac_cv_prog_make_${ac_make}_set = yes; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+  { $as_echo "$as_me:$LINENO: result: yes" >&5
 $as_echo "yes" >&6; }
   SET_MAKE=
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
   SET_MAKE="MAKE=${MAKE-make}"
 fi
@@ -2811,52 +2303,15 @@ else
 fi
 rmdir .tst 2>/dev/null
 
-# Check whether --enable-silent-rules was given.
-if test "${enable_silent_rules+set}" = set; then :
-  enableval=$enable_silent_rules;
-fi
-
-case $enable_silent_rules in # (((
-  yes) AM_DEFAULT_VERBOSITY=0;;
-   no) AM_DEFAULT_VERBOSITY=1;;
-    *) AM_DEFAULT_VERBOSITY=1;;
-esac
-am_make=${MAKE-make}
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $am_make supports nested variables" >&5
-$as_echo_n "checking whether $am_make supports nested variables... " >&6; }
-if ${am_cv_make_support_nested_variables+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if $as_echo 'TRUE=$(BAR$(V))
-BAR0=false
-BAR1=true
-V=1
-am__doit:
-	@$(TRUE)
-.PHONY: am__doit' | $am_make -f - >/dev/null 2>&1; then
-  am_cv_make_support_nested_variables=yes
-else
-  am_cv_make_support_nested_variables=no
-fi
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_make_support_nested_variables" >&5
-$as_echo "$am_cv_make_support_nested_variables" >&6; }
-if test $am_cv_make_support_nested_variables = yes; then
-    AM_V='$(V)'
-  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
-else
-  AM_V=$AM_DEFAULT_VERBOSITY
-  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
-fi
-AM_BACKSLASH='\'
-
 if test "`cd $srcdir && pwd`" != "`pwd`"; then
   # Use -I$(srcdir) only when $(srcdir) != ., so that make's output
   # is not polluted with repeated "-I."
   am__isrc=' -I$(srcdir)'
   # test to see if srcdir already configured
   if test -f $srcdir/config.status; then
-    as_fn_error $? "source directory already configured; run \"make distclean\" there first" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: source directory already configured; run \"make distclean\" there first" >&5
+$as_echo "$as_me: error: source directory already configured; run \"make distclean\" there first" >&2;}
+   { (exit 1); exit 1; }; }
   fi
 fi
 
@@ -2900,72 +2355,19 @@ AUTOHEADER=${AUTOHEADER-"${am_missing_run}autoheader"}
 
 MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 
-# For better backward compatibility.  To be removed once Automake 1.9.x
-# dies out for good.  For more background, see:
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
-mkdir_p='$(MKDIR_P)'
+# We need awk for the "check" target.  The system "awk" is bad on
+# some platforms.
+# Always define AMTAR for backward compatibility.
 
-# We need awk for the "check" target (and possibly the TAP driver).  The
-# system "awk" is bad on some platforms.
-# Always define AMTAR for backward compatibility.  Yes, it's still used
-# in the wild :-(  We should find a proper way to deprecate it ...
-AMTAR='$${TAR-tar}'
+AMTAR=${AMTAR-"${am_missing_run}tar"}
 
-
-# We'll loop over all known methods to create a tar archive until one works.
-_am_tools='gnutar  pax cpio none'
-
-am__tar='$${TAR-tar} chof - "$$tardir"' am__untar='$${TAR-tar} xf -'
+am__tar='${AMTAR} chof - "$$tardir"'; am__untar='${AMTAR} xf -'
 
 
 
 
 
 
-# POSIX will say in a future version that running "rm -f" with no argument
-# is OK; and we want to be able to make that assumption in our Makefile
-# recipes.  So use an aggressive probe to check that the usage we want is
-# actually supported "in the wild" to an acceptable degree.
-# See automake bug#10828.
-# To make any issue more visible, cause the running configure to be aborted
-# by default if the 'rm' program in use doesn't match our expectations; the
-# user can still override this though.
-if rm -f && rm -fr && rm -rf; then : OK; else
-  cat >&2 <<'END'
-Oops!
-
-Your 'rm' program seems unable to run without file operands specified
-on the command line, even when the '-f' option is present.  This is contrary
-to the behaviour of most rm programs out there, and not conforming with
-the upcoming POSIX standard: <http://austingroupbugs.net/view.php?id=542>
-
-Please tell bug-automake@gnu.org about your system, including the value
-of your $PATH and any error possibly output before this message.  This
-can help us improve future automake versions.
-
-END
-  if test x"$ACCEPT_INFERIOR_RM_PROGRAM" = x"yes"; then
-    echo 'Configuration will proceed anyway, since you have set the' >&2
-    echo 'ACCEPT_INFERIOR_RM_PROGRAM variable to "yes"' >&2
-    echo >&2
-  else
-    cat >&2 <<'END'
-Aborting the configuration process, to ensure you take notice of the issue.
-
-You can download and install GNU coreutils to get an 'rm' implementation
-that behaves properly: <http://www.gnu.org/software/coreutils/>.
-
-If you want to complete the configuration process using your problematic
-'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
-to "yes", and re-run configure.
-
-END
-    as_fn_error $? "Your 'rm' program is bad, sorry." "$LINENO" 5
-  fi
-fi
-
-
 
 ac_config_headers="$ac_config_headers lib/Grid_config.h"
 
@@ -2985,9 +2387,9 @@ if test -z "$CXX"; then
   do
     # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
 set dummy $ac_tool_prefix$ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CXX+:} false; then :
+if test "${ac_cv_prog_CXX+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CXX"; then
@@ -2998,24 +2400,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_CXX="$ac_tool_prefix$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 CXX=$ac_cv_prog_CXX
 if test -n "$CXX"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CXX" >&5
+  { $as_echo "$as_me:$LINENO: result: $CXX" >&5
 $as_echo "$CXX" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3029,9 +2431,9 @@ if test -z "$CXX"; then
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_CXX+:} false; then :
+if test "${ac_cv_prog_ac_ct_CXX+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CXX"; then
@@ -3042,24 +2444,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_ac_ct_CXX="$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CXX=$ac_cv_prog_ac_ct_CXX
 if test -n "$ac_ct_CXX"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CXX" >&5
+  { $as_echo "$as_me:$LINENO: result: $ac_ct_CXX" >&5
 $as_echo "$ac_ct_CXX" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3072,7 +2474,7 @@ done
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -3083,31 +2485,48 @@ fi
   fi
 fi
 # Provide some information about the compiler.
-$as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler version" >&5
+$as_echo "$as_me:$LINENO: checking for C++ compiler version" >&5
 set X $ac_compile
 ac_compiler=$2
-for ac_option in --version -v -V -qversion; do
-  { { ac_try="$ac_compiler $ac_option >&5"
+{ (ac_try="$ac_compiler --version >&5"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler --version >&5") 2>&5
   ac_status=$?
-  if test -s conftest.err; then
-    sed '10a\
-... rest of stderr output deleted ...
-         10q' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-  fi
-  rm -f conftest.er1 conftest.err
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }
-done
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
+{ (ac_try="$ac_compiler -v >&5"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler -v >&5") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
+{ (ac_try="$ac_compiler -V >&5"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler -V >&5") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
 
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3123,8 +2542,8 @@ ac_clean_files="$ac_clean_files a.out a.out.dSYM a.exe b.out"
 # Try to create an executable without -o first, disregard a.out.
 # It will help us diagnose broken compilers, and finding out an intuition
 # of exeext.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether the C++ compiler works" >&5
-$as_echo_n "checking whether the C++ compiler works... " >&6; }
+{ $as_echo "$as_me:$LINENO: checking for C++ compiler default output file name" >&5
+$as_echo_n "checking for C++ compiler default output file name... " >&6; }
 ac_link_default=`$as_echo "$ac_link" | sed 's/ -o *conftest[^ ]*//'`
 
 # The possible output files:
@@ -3140,17 +2559,17 @@ do
 done
 rm -f $ac_rmfiles
 
-if { { ac_try="$ac_link_default"
+if { (ac_try="$ac_link_default"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
   (eval "$ac_link_default") 2>&5
   ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; then :
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; then
   # Autoconf-2.13 could set the ac_cv_exeext variable to `no'.
 # So ignore a value of `no', otherwise this would lead to `EXEEXT = no'
 # in a Makefile.  We should not override ac_cv_exeext if it was cached,
@@ -3167,7 +2586,7 @@ do
 	# certainly right.
 	break;;
     *.* )
-	if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
+        if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
 	then :; else
 	   ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
 	fi
@@ -3186,41 +2605,84 @@ test "$ac_cv_exeext" = no && ac_cv_exeext=
 else
   ac_file=''
 fi
-if test -z "$ac_file"; then :
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-$as_echo "$as_me: failed program was:" >&5
+
+{ $as_echo "$as_me:$LINENO: result: $ac_file" >&5
+$as_echo "$ac_file" >&6; }
+if test -z "$ac_file"; then
+  $as_echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
 
-{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error 77 "C++ compiler cannot create executables
-See \`config.log' for more details" "$LINENO" 5; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
-$as_echo "yes" >&6; }
+{ { $as_echo "$as_me:$LINENO: error: C++ compiler cannot create executables
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: C++ compiler cannot create executables
+See \`config.log' for more details." >&2;}
+   { (exit 77); exit 77; }; }; }
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler default output file name" >&5
-$as_echo_n "checking for C++ compiler default output file name... " >&6; }
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_file" >&5
-$as_echo "$ac_file" >&6; }
+
 ac_exeext=$ac_cv_exeext
 
+# Check that the compiler produces executables we can run.  If not, either
+# the compiler is broken, or we cross compile.
+{ $as_echo "$as_me:$LINENO: checking whether the C++ compiler works" >&5
+$as_echo_n "checking whether the C++ compiler works... " >&6; }
+# FIXME: These cross compiler hacks should be removed for Autoconf 3.0
+# If not cross compiling, check that we can run a simple program.
+if test "$cross_compiling" != yes; then
+  if { ac_try='./$ac_file'
+  { (case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; }; then
+    cross_compiling=no
+  else
+    if test "$cross_compiling" = maybe; then
+	cross_compiling=yes
+    else
+	{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+{ { $as_echo "$as_me:$LINENO: error: cannot run C++ compiled programs.
+If you meant to cross compile, use \`--host'.
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: cannot run C++ compiled programs.
+If you meant to cross compile, use \`--host'.
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
+    fi
+  fi
+fi
+{ $as_echo "$as_me:$LINENO: result: yes" >&5
+$as_echo "yes" >&6; }
+
 rm -f -r a.out a.out.dSYM a.exe conftest$ac_cv_exeext b.out
 ac_clean_files=$ac_clean_files_save
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of executables" >&5
+# Check that the compiler produces executables we can run.  If not, either
+# the compiler is broken, or we cross compile.
+{ $as_echo "$as_me:$LINENO: checking whether we are cross compiling" >&5
+$as_echo_n "checking whether we are cross compiling... " >&6; }
+{ $as_echo "$as_me:$LINENO: result: $cross_compiling" >&5
+$as_echo "$cross_compiling" >&6; }
+
+{ $as_echo "$as_me:$LINENO: checking for suffix of executables" >&5
 $as_echo_n "checking for suffix of executables... " >&6; }
-if { { ac_try="$ac_link"
+if { (ac_try="$ac_link"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
   (eval "$ac_link") 2>&5
   ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; then :
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; then
   # If both `conftest.exe' and `conftest' are `present' (well, observable)
 # catch `conftest.exe'.  For instance with Cygwin, `ls conftest' will
 # work properly (i.e., refer to `conftest.exe'), while it won't with
@@ -3235,83 +2697,32 @@ for ac_file in conftest.exe conftest conftest.*; do
   esac
 done
 else
-  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+  { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "cannot compute suffix of executables: cannot compile and link
-See \`config.log' for more details" "$LINENO" 5; }
+{ { $as_echo "$as_me:$LINENO: error: cannot compute suffix of executables: cannot compile and link
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: cannot compute suffix of executables: cannot compile and link
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
 fi
-rm -f conftest conftest$ac_cv_exeext
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_exeext" >&5
+
+rm -f conftest$ac_cv_exeext
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_exeext" >&5
 $as_echo "$ac_cv_exeext" >&6; }
 
 rm -f conftest.$ac_ext
 EXEEXT=$ac_cv_exeext
 ac_exeext=$EXEEXT
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <stdio.h>
-int
-main ()
-{
-FILE *f = fopen ("conftest.out", "w");
- return ferror (f) || fclose (f) != 0;
-
-  ;
-  return 0;
-}
-_ACEOF
-ac_clean_files="$ac_clean_files conftest.out"
-# Check that the compiler produces executables we can run.  If not, either
-# the compiler is broken, or we cross compile.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are cross compiling" >&5
-$as_echo_n "checking whether we are cross compiling... " >&6; }
-if test "$cross_compiling" != yes; then
-  { { ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_link") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }
-  if { ac_try='./conftest$ac_cv_exeext'
-  { { case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; }; then
-    cross_compiling=no
-  else
-    if test "$cross_compiling" = maybe; then
-	cross_compiling=yes
-    else
-	{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "cannot run C++ compiled programs.
-If you meant to cross compile, use \`--host'.
-See \`config.log' for more details" "$LINENO" 5; }
-    fi
-  fi
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $cross_compiling" >&5
-$as_echo "$cross_compiling" >&6; }
-
-rm -f conftest.$ac_ext conftest$ac_cv_exeext conftest.out
-ac_clean_files=$ac_clean_files_save
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of object files" >&5
+{ $as_echo "$as_me:$LINENO: checking for suffix of object files" >&5
 $as_echo_n "checking for suffix of object files... " >&6; }
-if ${ac_cv_objext+:} false; then :
+if test "${ac_cv_objext+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3323,17 +2734,17 @@ main ()
 }
 _ACEOF
 rm -f conftest.o conftest.obj
-if { { ac_try="$ac_compile"
+if { (ac_try="$ac_compile"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
   (eval "$ac_compile") 2>&5
   ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; then :
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; then
   for ac_file in conftest.o conftest.obj conftest.*; do
   test -f "$ac_file" || continue;
   case $ac_file in
@@ -3346,23 +2757,31 @@ else
   $as_echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
 
-{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "cannot compute suffix of object files: cannot compile
-See \`config.log' for more details" "$LINENO" 5; }
+{ { $as_echo "$as_me:$LINENO: error: cannot compute suffix of object files: cannot compile
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: cannot compute suffix of object files: cannot compile
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
 fi
+
 rm -f conftest.$ac_cv_objext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_objext" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_objext" >&5
 $as_echo "$ac_cv_objext" >&6; }
 OBJEXT=$ac_cv_objext
 ac_objext=$OBJEXT
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C++ compiler" >&5
+{ $as_echo "$as_me:$LINENO: checking whether we are using the GNU C++ compiler" >&5
 $as_echo_n "checking whether we are using the GNU C++ compiler... " >&6; }
-if ${ac_cv_cxx_compiler_gnu+:} false; then :
+if test "${ac_cv_cxx_compiler_gnu+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3376,16 +2795,37 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_compiler_gnu=yes
 else
-  ac_compiler_gnu=no
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_compiler_gnu=no
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 ac_cv_cxx_compiler_gnu=$ac_compiler_gnu
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_cxx_compiler_gnu" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_cxx_compiler_gnu" >&5
 $as_echo "$ac_cv_cxx_compiler_gnu" >&6; }
 if test $ac_compiler_gnu = yes; then
   GXX=yes
@@ -3394,16 +2834,20 @@ else
 fi
 ac_test_CXXFLAGS=${CXXFLAGS+set}
 ac_save_CXXFLAGS=$CXXFLAGS
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX accepts -g" >&5
+{ $as_echo "$as_me:$LINENO: checking whether $CXX accepts -g" >&5
 $as_echo_n "checking whether $CXX accepts -g... " >&6; }
-if ${ac_cv_prog_cxx_g+:} false; then :
+if test "${ac_cv_prog_cxx_g+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   ac_save_cxx_werror_flag=$ac_cxx_werror_flag
    ac_cxx_werror_flag=yes
    ac_cv_prog_cxx_g=no
    CXXFLAGS="-g"
-   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+   cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3414,11 +2858,35 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_prog_cxx_g=yes
 else
-  CXXFLAGS=""
-      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	CXXFLAGS=""
+      cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3429,12 +2897,36 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  :
 else
-  ac_cxx_werror_flag=$ac_save_cxx_werror_flag
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cxx_werror_flag=$ac_save_cxx_werror_flag
 	 CXXFLAGS="-g"
-	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+	 cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3445,17 +2937,42 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_prog_cxx_g=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
    ac_cxx_werror_flag=$ac_save_cxx_werror_flag
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cxx_g" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cxx_g" >&5
 $as_echo "$ac_cv_prog_cxx_g" >&6; }
 if test "$ac_test_CXXFLAGS" = set; then
   CXXFLAGS=$ac_save_CXXFLAGS
@@ -3489,14 +3006,14 @@ am__doit:
 .PHONY: am__doit
 END
 # If we don't find an include directive, just comment out the code.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for style of include used by $am_make" >&5
+{ $as_echo "$as_me:$LINENO: checking for style of include used by $am_make" >&5
 $as_echo_n "checking for style of include used by $am_make... " >&6; }
 am__include="#"
 am__quote=
 _am_result=none
 # First try GNU make style include.
 echo "include confinc" > confmf
-# Ignore all kinds of additional output from 'make'.
+# Ignore all kinds of additional output from `make'.
 case `$am_make -s -f confmf 2> /dev/null` in #(
 *the\ am__doit\ target*)
   am__include=include
@@ -3517,19 +3034,18 @@ if test "$am__include" = "#"; then
 fi
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $_am_result" >&5
+{ $as_echo "$as_me:$LINENO: result: $_am_result" >&5
 $as_echo "$_am_result" >&6; }
 rm -f confinc confmf
 
 # Check whether --enable-dependency-tracking was given.
-if test "${enable_dependency_tracking+set}" = set; then :
+if test "${enable_dependency_tracking+set}" = set; then
   enableval=$enable_dependency_tracking;
 fi
 
 if test "x$enable_dependency_tracking" != xno; then
   am_depcomp="$ac_aux_dir/depcomp"
   AMDEPBACKSLASH='\'
-  am__nodep='_no'
 fi
  if test "x$enable_dependency_tracking" != xno; then
   AMDEP_TRUE=
@@ -3543,18 +3059,17 @@ fi
 
 depcc="$CXX"  am_compiler_list=
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
+{ $as_echo "$as_me:$LINENO: checking dependency style of $depcc" >&5
 $as_echo_n "checking dependency style of $depcc... " >&6; }
-if ${am_cv_CXX_dependencies_compiler_type+:} false; then :
+if test "${am_cv_CXX_dependencies_compiler_type+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named 'D' -- because '-MD' means "put the output
-  # in D".
-  rm -rf conftest.dir
+  # making a dummy file named `D' -- because `-MD' means `put the output
+  # in D'.
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -3588,16 +3103,16 @@ else
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
-      # Solaris 10 /bin/sh.
-      echo '/* dummy */' > sub/conftst$i.h
+      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
+      # Solaris 8's {/usr,}/bin/sh.
+      touch sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with '-c' and '-o' for the sake of the "dashmstdout"
+    # We check with `-c' and `-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle '-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs.
+    # handle `-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -3606,16 +3121,16 @@ else
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # After this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested.
+      # after this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
-      # This compiler won't grok '-c -o', but also, the minuso test has
+    msvisualcpp | msvcmsys)
+      # This compiler won't grok `-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -3654,7 +3169,7 @@ else
 fi
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CXX_dependencies_compiler_type" >&5
+{ $as_echo "$as_me:$LINENO: result: $am_cv_CXX_dependencies_compiler_type" >&5
 $as_echo "$am_cv_CXX_dependencies_compiler_type" >&6; }
 CXXDEPMODE=depmode=$am_cv_CXX_dependencies_compiler_type
 
@@ -3677,9 +3192,9 @@ ac_compiler_gnu=$ac_cv_c_compiler_gnu
 if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}gcc", so it can be a program name with args.
 set dummy ${ac_tool_prefix}gcc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
+if test "${ac_cv_prog_CC+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3690,24 +3205,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_CC="${ac_tool_prefix}gcc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+  { $as_echo "$as_me:$LINENO: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3717,9 +3232,9 @@ if test -z "$ac_cv_prog_CC"; then
   ac_ct_CC=$CC
   # Extract the first word of "gcc", so it can be a program name with args.
 set dummy gcc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_CC+:} false; then :
+if test "${ac_cv_prog_ac_ct_CC+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CC"; then
@@ -3730,24 +3245,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_ac_ct_CC="gcc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CC=$ac_cv_prog_ac_ct_CC
 if test -n "$ac_ct_CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
+  { $as_echo "$as_me:$LINENO: result: $ac_ct_CC" >&5
 $as_echo "$ac_ct_CC" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3756,7 +3271,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -3770,9 +3285,9 @@ if test -z "$CC"; then
           if test -n "$ac_tool_prefix"; then
     # Extract the first word of "${ac_tool_prefix}cc", so it can be a program name with args.
 set dummy ${ac_tool_prefix}cc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
+if test "${ac_cv_prog_CC+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3783,24 +3298,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_CC="${ac_tool_prefix}cc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+  { $as_echo "$as_me:$LINENO: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3810,9 +3325,9 @@ fi
 if test -z "$CC"; then
   # Extract the first word of "cc", so it can be a program name with args.
 set dummy cc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
+if test "${ac_cv_prog_CC+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3824,18 +3339,18 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     if test "$as_dir/$ac_word$ac_exec_ext" = "/usr/ucb/cc"; then
        ac_prog_rejected=yes
        continue
      fi
     ac_cv_prog_CC="cc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 if test $ac_prog_rejected = yes; then
@@ -3854,10 +3369,10 @@ fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+  { $as_echo "$as_me:$LINENO: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3869,9 +3384,9 @@ if test -z "$CC"; then
   do
     # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
 set dummy $ac_tool_prefix$ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
+if test "${ac_cv_prog_CC+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3882,24 +3397,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_CC="$ac_tool_prefix$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+  { $as_echo "$as_me:$LINENO: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3913,9 +3428,9 @@ if test -z "$CC"; then
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_CC+:} false; then :
+if test "${ac_cv_prog_ac_ct_CC+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CC"; then
@@ -3926,24 +3441,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_ac_ct_CC="$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CC=$ac_cv_prog_ac_ct_CC
 if test -n "$ac_ct_CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
+  { $as_echo "$as_me:$LINENO: result: $ac_ct_CC" >&5
 $as_echo "$ac_ct_CC" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3956,7 +3471,7 @@ done
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -3967,42 +3482,62 @@ fi
 fi
 
 
-test -z "$CC" && { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+test -z "$CC" && { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "no acceptable C compiler found in \$PATH
-See \`config.log' for more details" "$LINENO" 5; }
+{ { $as_echo "$as_me:$LINENO: error: no acceptable C compiler found in \$PATH
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: no acceptable C compiler found in \$PATH
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
 
 # Provide some information about the compiler.
-$as_echo "$as_me:${as_lineno-$LINENO}: checking for C compiler version" >&5
+$as_echo "$as_me:$LINENO: checking for C compiler version" >&5
 set X $ac_compile
 ac_compiler=$2
-for ac_option in --version -v -V -qversion; do
-  { { ac_try="$ac_compiler $ac_option >&5"
+{ (ac_try="$ac_compiler --version >&5"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler --version >&5") 2>&5
   ac_status=$?
-  if test -s conftest.err; then
-    sed '10a\
-... rest of stderr output deleted ...
-         10q' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-  fi
-  rm -f conftest.er1 conftest.err
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }
-done
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
+{ (ac_try="$ac_compiler -v >&5"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler -v >&5") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
+{ (ac_try="$ac_compiler -V >&5"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler -V >&5") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C compiler" >&5
+{ $as_echo "$as_me:$LINENO: checking whether we are using the GNU C compiler" >&5
 $as_echo_n "checking whether we are using the GNU C compiler... " >&6; }
-if ${ac_cv_c_compiler_gnu+:} false; then :
+if test "${ac_cv_c_compiler_gnu+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -4016,16 +3551,37 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_compiler_gnu=yes
 else
-  ac_compiler_gnu=no
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_compiler_gnu=no
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 ac_cv_c_compiler_gnu=$ac_compiler_gnu
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_c_compiler_gnu" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_compiler_gnu" >&5
 $as_echo "$ac_cv_c_compiler_gnu" >&6; }
 if test $ac_compiler_gnu = yes; then
   GCC=yes
@@ -4034,16 +3590,20 @@ else
 fi
 ac_test_CFLAGS=${CFLAGS+set}
 ac_save_CFLAGS=$CFLAGS
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC accepts -g" >&5
+{ $as_echo "$as_me:$LINENO: checking whether $CC accepts -g" >&5
 $as_echo_n "checking whether $CC accepts -g... " >&6; }
-if ${ac_cv_prog_cc_g+:} false; then :
+if test "${ac_cv_prog_cc_g+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   ac_save_c_werror_flag=$ac_c_werror_flag
    ac_c_werror_flag=yes
    ac_cv_prog_cc_g=no
    CFLAGS="-g"
-   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+   cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -4054,11 +3614,35 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_prog_cc_g=yes
 else
-  CFLAGS=""
-      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	CFLAGS=""
+      cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -4069,12 +3653,36 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  :
 else
-  ac_c_werror_flag=$ac_save_c_werror_flag
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_c_werror_flag=$ac_save_c_werror_flag
 	 CFLAGS="-g"
-	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+	 cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -4085,17 +3693,42 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_prog_cc_g=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
    ac_c_werror_flag=$ac_save_c_werror_flag
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_g" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cc_g" >&5
 $as_echo "$ac_cv_prog_cc_g" >&6; }
 if test "$ac_test_CFLAGS" = set; then
   CFLAGS=$ac_save_CFLAGS
@@ -4112,18 +3745,23 @@ else
     CFLAGS=
   fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to accept ISO C89" >&5
+{ $as_echo "$as_me:$LINENO: checking for $CC option to accept ISO C89" >&5
 $as_echo_n "checking for $CC option to accept ISO C89... " >&6; }
-if ${ac_cv_prog_cc_c89+:} false; then :
+if test "${ac_cv_prog_cc_c89+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   ac_cv_prog_cc_c89=no
 ac_save_CC=$CC
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <stdarg.h>
 #include <stdio.h>
-struct stat;
+#include <sys/types.h>
+#include <sys/stat.h>
 /* Most of the following tests are stolen from RCS 5.7's src/conf.sh.  */
 struct buf { int x; };
 FILE * (*rcsopen) (struct buf *, struct stat *, int);
@@ -4175,9 +3813,32 @@ for ac_arg in '' -qlanglvl=extc89 -qlanglvl=ansi -std \
 	-Ae "-Aa -D_HPUX_SOURCE" "-Xc -D__EXTENSIONS__"
 do
   CC="$ac_save_CC $ac_arg"
-  if ac_fn_c_try_compile "$LINENO"; then :
+  rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_prog_cc_c89=$ac_arg
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
 fi
+
 rm -f core conftest.err conftest.$ac_objext
   test "x$ac_cv_prog_cc_c89" != "xno" && break
 done
@@ -4188,19 +3849,17 @@ fi
 # AC_CACHE_VAL
 case "x$ac_cv_prog_cc_c89" in
   x)
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: none needed" >&5
+    { $as_echo "$as_me:$LINENO: result: none needed" >&5
 $as_echo "none needed" >&6; } ;;
   xno)
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: unsupported" >&5
+    { $as_echo "$as_me:$LINENO: result: unsupported" >&5
 $as_echo "unsupported" >&6; } ;;
   *)
     CC="$CC $ac_cv_prog_cc_c89"
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_c89" >&5
+    { $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cc_c89" >&5
 $as_echo "$ac_cv_prog_cc_c89" >&6; } ;;
 esac
-if test "x$ac_cv_prog_cc_c89" != xno; then :
 
-fi
 
 ac_ext=c
 ac_cpp='$CPP $CPPFLAGS'
@@ -4208,79 +3867,19 @@ ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
 ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC understands -c and -o together" >&5
-$as_echo_n "checking whether $CC understands -c and -o together... " >&6; }
-if ${am_cv_prog_cc_c_o+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-  # Make sure it works both with $CC and with simple cc.
-  # Following AC_PROG_CC_C_O, we do the test twice because some
-  # compilers refuse to overwrite an existing .o file with -o,
-  # though they will create one.
-  am_cv_prog_cc_c_o=yes
-  for am_i in 1 2; do
-    if { echo "$as_me:$LINENO: $CC -c conftest.$ac_ext -o conftest2.$ac_objext" >&5
-   ($CC -c conftest.$ac_ext -o conftest2.$ac_objext) >&5 2>&5
-   ac_status=$?
-   echo "$as_me:$LINENO: \$? = $ac_status" >&5
-   (exit $ac_status); } \
-         && test -f conftest2.$ac_objext; then
-      : OK
-    else
-      am_cv_prog_cc_c_o=no
-      break
-    fi
-  done
-  rm -f core conftest*
-  unset am_i
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_prog_cc_c_o" >&5
-$as_echo "$am_cv_prog_cc_c_o" >&6; }
-if test "$am_cv_prog_cc_c_o" != yes; then
-   # Losing compiler, so override with the script.
-   # FIXME: It is wrong to rewrite CC.
-   # But if we don't then we get into trouble of one sort or another.
-   # A longer-term fix would be to have automake use am__CC in this case,
-   # and then we could set am__CC="\$(top_srcdir)/compile \$(CC)"
-   CC="$am_aux_dir/compile $CC"
-fi
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-
-
 depcc="$CC"   am_compiler_list=
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
+{ $as_echo "$as_me:$LINENO: checking dependency style of $depcc" >&5
 $as_echo_n "checking dependency style of $depcc... " >&6; }
-if ${am_cv_CC_dependencies_compiler_type+:} false; then :
+if test "${am_cv_CC_dependencies_compiler_type+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named 'D' -- because '-MD' means "put the output
-  # in D".
-  rm -rf conftest.dir
+  # making a dummy file named `D' -- because `-MD' means `put the output
+  # in D'.
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -4314,16 +3913,16 @@ else
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
-      # Solaris 10 /bin/sh.
-      echo '/* dummy */' > sub/conftst$i.h
+      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
+      # Solaris 8's {/usr,}/bin/sh.
+      touch sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with '-c' and '-o' for the sake of the "dashmstdout"
+    # We check with `-c' and `-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle '-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs.
+    # handle `-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -4332,16 +3931,16 @@ else
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # After this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested.
+      # after this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
-      # This compiler won't grok '-c -o', but also, the minuso test has
+    msvisualcpp | msvcmsys)
+      # This compiler won't grok `-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -4380,7 +3979,7 @@ else
 fi
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CC_dependencies_compiler_type" >&5
+{ $as_echo "$as_me:$LINENO: result: $am_cv_CC_dependencies_compiler_type" >&5
 $as_echo "$am_cv_CC_dependencies_compiler_type" >&6; }
 CCDEPMODE=depmode=$am_cv_CC_dependencies_compiler_type
 
@@ -4399,18 +3998,17 @@ fi
 
   OPENMP_CFLAGS=
   # Check whether --enable-openmp was given.
-if test "${enable_openmp+set}" = set; then :
+if test "${enable_openmp+set}" = set; then
   enableval=$enable_openmp;
 fi
 
   if test "$enable_openmp" != no; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to support OpenMP" >&5
+    { $as_echo "$as_me:$LINENO: checking for $CC option to support OpenMP" >&5
 $as_echo_n "checking for $CC option to support OpenMP... " >&6; }
-if ${ac_cv_prog_c_openmp+:} false; then :
+if test "${ac_cv_prog_c_openmp+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
+  cat >conftest.$ac_ext <<_ACEOF
 
 #ifndef _OPENMP
  choke me
@@ -4419,16 +4017,37 @@ else
 int main () { return omp_get_num_threads (); }
 
 _ACEOF
-if ac_fn_c_try_link "$LINENO"; then :
+rm -f conftest.$ac_objext conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_link") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 $as_test_x conftest$ac_exeext
+       }; then
   ac_cv_prog_c_openmp='none needed'
 else
-  ac_cv_prog_c_openmp='unsupported'
-	  	  	  	  	  	  	                                	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp -homp \
-                           -Popenmp --openmp; do
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cv_prog_c_openmp='unsupported'
+	  	  	  	  	  	  	  	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp; do
 	    ac_save_CFLAGS=$CFLAGS
 	    CFLAGS="$CFLAGS $ac_option"
-	    cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
+	    cat >conftest.$ac_ext <<_ACEOF
 
 #ifndef _OPENMP
  choke me
@@ -4437,27 +4056,56 @@ else
 int main () { return omp_get_num_threads (); }
 
 _ACEOF
-if ac_fn_c_try_link "$LINENO"; then :
+rm -f conftest.$ac_objext conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_link") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 $as_test_x conftest$ac_exeext
+       }; then
   ac_cv_prog_c_openmp=$ac_option
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
 fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
+
+rm -rf conftest.dSYM
+rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
+      conftest$ac_exeext conftest.$ac_ext
 	    CFLAGS=$ac_save_CFLAGS
 	    if test "$ac_cv_prog_c_openmp" != unsupported; then
 	      break
 	    fi
 	  done
 fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
+
+rm -rf conftest.dSYM
+rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
+      conftest$ac_exeext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_c_openmp" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_c_openmp" >&5
 $as_echo "$ac_cv_prog_c_openmp" >&6; }
     case $ac_cv_prog_c_openmp in #(
       "none needed" | unsupported)
-	;; #(
+        ;; #(
       *)
-	OPENMP_CFLAGS=$ac_cv_prog_c_openmp ;;
+        OPENMP_CFLAGS=$ac_cv_prog_c_openmp ;;
     esac
   fi
 
@@ -4465,9 +4113,9 @@ $as_echo "$ac_cv_prog_c_openmp" >&6; }
 if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}ranlib", so it can be a program name with args.
 set dummy ${ac_tool_prefix}ranlib; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_RANLIB+:} false; then :
+if test "${ac_cv_prog_RANLIB+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$RANLIB"; then
@@ -4478,24 +4126,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_RANLIB="${ac_tool_prefix}ranlib"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 RANLIB=$ac_cv_prog_RANLIB
 if test -n "$RANLIB"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $RANLIB" >&5
+  { $as_echo "$as_me:$LINENO: result: $RANLIB" >&5
 $as_echo "$RANLIB" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -4505,9 +4153,9 @@ if test -z "$ac_cv_prog_RANLIB"; then
   ac_ct_RANLIB=$RANLIB
   # Extract the first word of "ranlib", so it can be a program name with args.
 set dummy ranlib; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_RANLIB+:} false; then :
+if test "${ac_cv_prog_ac_ct_RANLIB+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_RANLIB"; then
@@ -4518,24 +4166,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_ac_ct_RANLIB="ranlib"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_RANLIB=$ac_cv_prog_ac_ct_RANLIB
 if test -n "$ac_ct_RANLIB"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_RANLIB" >&5
+  { $as_echo "$as_me:$LINENO: result: $ac_ct_RANLIB" >&5
 $as_echo "$ac_ct_RANLIB" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -4544,7 +4192,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -4564,14 +4212,14 @@ ac_cpp='$CPP $CPPFLAGS'
 ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
 ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking how to run the C preprocessor" >&5
+{ $as_echo "$as_me:$LINENO: checking how to run the C preprocessor" >&5
 $as_echo_n "checking how to run the C preprocessor... " >&6; }
 # On Suns, sometimes $CPP names a directory.
 if test -n "$CPP" && test -d "$CPP"; then
   CPP=
 fi
 if test -z "$CPP"; then
-  if ${ac_cv_prog_CPP+:} false; then :
+  if test "${ac_cv_prog_CPP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
       # Double quotes because CPP needs to be expanded
@@ -4586,7 +4234,11 @@ do
   # <limits.h> exists even on freestanding compilers.
   # On the NeXT, cc -E runs the code through the compiler's parser,
   # not just through cpp. "Syntax error" is here to catch this case.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #ifdef __STDC__
 # include <limits.h>
@@ -4595,34 +4247,78 @@ do
 #endif
 		     Syntax error
 _ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
-
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  :
 else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
   # Broken: fails on valid input.
 continue
 fi
-rm -f conftest.err conftest.i conftest.$ac_ext
+
+rm -f conftest.err conftest.$ac_ext
 
   # OK, works on sane cases.  Now check whether nonexistent headers
   # can be detected and how.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <ac_nonexistent.h>
 _ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
   # Broken: success on invalid input.
 continue
 else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
   # Passes both tests.
 ac_preproc_ok=:
 break
 fi
-rm -f conftest.err conftest.i conftest.$ac_ext
+
+rm -f conftest.err conftest.$ac_ext
 
 done
 # Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
-rm -f conftest.i conftest.err conftest.$ac_ext
-if $ac_preproc_ok; then :
+rm -f conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then
   break
 fi
 
@@ -4634,7 +4330,7 @@ fi
 else
   ac_cv_prog_CPP=$CPP
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $CPP" >&5
+{ $as_echo "$as_me:$LINENO: result: $CPP" >&5
 $as_echo "$CPP" >&6; }
 ac_preproc_ok=false
 for ac_c_preproc_warn_flag in '' yes
@@ -4645,7 +4341,11 @@ do
   # <limits.h> exists even on freestanding compilers.
   # On the NeXT, cc -E runs the code through the compiler's parser,
   # not just through cpp. "Syntax error" is here to catch this case.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #ifdef __STDC__
 # include <limits.h>
@@ -4654,40 +4354,87 @@ do
 #endif
 		     Syntax error
 _ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
-
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  :
 else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
   # Broken: fails on valid input.
 continue
 fi
-rm -f conftest.err conftest.i conftest.$ac_ext
+
+rm -f conftest.err conftest.$ac_ext
 
   # OK, works on sane cases.  Now check whether nonexistent headers
   # can be detected and how.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <ac_nonexistent.h>
 _ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
   # Broken: success on invalid input.
 continue
 else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
   # Passes both tests.
 ac_preproc_ok=:
 break
 fi
-rm -f conftest.err conftest.i conftest.$ac_ext
+
+rm -f conftest.err conftest.$ac_ext
 
 done
 # Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
-rm -f conftest.i conftest.err conftest.$ac_ext
-if $ac_preproc_ok; then :
-
+rm -f conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then
+  :
 else
-  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+  { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "C preprocessor \"$CPP\" fails sanity check
-See \`config.log' for more details" "$LINENO" 5; }
+{ { $as_echo "$as_me:$LINENO: error: C preprocessor \"$CPP\" fails sanity check
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: C preprocessor \"$CPP\" fails sanity check
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
 fi
 
 ac_ext=c
@@ -4697,9 +4444,9 @@ ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for grep that handles long lines and -e" >&5
+{ $as_echo "$as_me:$LINENO: checking for grep that handles long lines and -e" >&5
 $as_echo_n "checking for grep that handles long lines and -e... " >&6; }
-if ${ac_cv_path_GREP+:} false; then :
+if test "${ac_cv_path_GREP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -z "$GREP"; then
@@ -4710,10 +4457,10 @@ for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_prog in grep ggrep; do
+  for ac_prog in grep ggrep; do
     for ac_exec_ext in '' $ac_executable_extensions; do
       ac_path_GREP="$as_dir/$ac_prog$ac_exec_ext"
-      as_fn_executable_p "$ac_path_GREP" || continue
+      { test -f "$ac_path_GREP" && $as_test_x "$ac_path_GREP"; } || continue
 # Check for GNU ac_path_GREP and select it if it is found.
   # Check for GNU $ac_path_GREP
 case `"$ac_path_GREP" --version 2>&1` in
@@ -4730,7 +4477,7 @@ case `"$ac_path_GREP" --version 2>&1` in
     $as_echo 'GREP' >> "conftest.nl"
     "$ac_path_GREP" -e 'GREP$' -e '-(cannot match)-' < "conftest.nl" >"conftest.out" 2>/dev/null || break
     diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
-    as_fn_arith $ac_count + 1 && ac_count=$as_val
+    ac_count=`expr $ac_count + 1`
     if test $ac_count -gt ${ac_path_GREP_max-0}; then
       # Best one so far, save it but keep looking for a better one
       ac_cv_path_GREP="$ac_path_GREP"
@@ -4745,24 +4492,26 @@ esac
       $ac_path_GREP_found && break 3
     done
   done
-  done
+done
 IFS=$as_save_IFS
   if test -z "$ac_cv_path_GREP"; then
-    as_fn_error $? "no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&5
+$as_echo "$as_me: error: no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&2;}
+   { (exit 1); exit 1; }; }
   fi
 else
   ac_cv_path_GREP=$GREP
 fi
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_GREP" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_path_GREP" >&5
 $as_echo "$ac_cv_path_GREP" >&6; }
  GREP="$ac_cv_path_GREP"
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for egrep" >&5
+{ $as_echo "$as_me:$LINENO: checking for egrep" >&5
 $as_echo_n "checking for egrep... " >&6; }
-if ${ac_cv_path_EGREP+:} false; then :
+if test "${ac_cv_path_EGREP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if echo a | $GREP -E '(a|b)' >/dev/null 2>&1
@@ -4776,10 +4525,10 @@ for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_prog in egrep; do
+  for ac_prog in egrep; do
     for ac_exec_ext in '' $ac_executable_extensions; do
       ac_path_EGREP="$as_dir/$ac_prog$ac_exec_ext"
-      as_fn_executable_p "$ac_path_EGREP" || continue
+      { test -f "$ac_path_EGREP" && $as_test_x "$ac_path_EGREP"; } || continue
 # Check for GNU ac_path_EGREP and select it if it is found.
   # Check for GNU $ac_path_EGREP
 case `"$ac_path_EGREP" --version 2>&1` in
@@ -4796,7 +4545,7 @@ case `"$ac_path_EGREP" --version 2>&1` in
     $as_echo 'EGREP' >> "conftest.nl"
     "$ac_path_EGREP" 'EGREP$' < "conftest.nl" >"conftest.out" 2>/dev/null || break
     diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
-    as_fn_arith $ac_count + 1 && ac_count=$as_val
+    ac_count=`expr $ac_count + 1`
     if test $ac_count -gt ${ac_path_EGREP_max-0}; then
       # Best one so far, save it but keep looking for a better one
       ac_cv_path_EGREP="$ac_path_EGREP"
@@ -4811,10 +4560,12 @@ esac
       $ac_path_EGREP_found && break 3
     done
   done
-  done
+done
 IFS=$as_save_IFS
   if test -z "$ac_cv_path_EGREP"; then
-    as_fn_error $? "no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&5
+$as_echo "$as_me: error: no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&2;}
+   { (exit 1); exit 1; }; }
   fi
 else
   ac_cv_path_EGREP=$EGREP
@@ -4822,17 +4573,21 @@ fi
 
    fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_EGREP" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_path_EGREP" >&5
 $as_echo "$ac_cv_path_EGREP" >&6; }
  EGREP="$ac_cv_path_EGREP"
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for ANSI C header files" >&5
+{ $as_echo "$as_me:$LINENO: checking for ANSI C header files" >&5
 $as_echo_n "checking for ANSI C header files... " >&6; }
-if ${ac_cv_header_stdc+:} false; then :
+if test "${ac_cv_header_stdc+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <stdlib.h>
 #include <stdarg.h>
@@ -4847,23 +4602,48 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_header_stdc=yes
 else
-  ac_cv_header_stdc=no
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cv_header_stdc=no
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 
 if test $ac_cv_header_stdc = yes; then
   # SunOS 4.x string.h does not declare mem*, contrary to ANSI.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <string.h>
 
 _ACEOF
 if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
-  $EGREP "memchr" >/dev/null 2>&1; then :
-
+  $EGREP "memchr" >/dev/null 2>&1; then
+  :
 else
   ac_cv_header_stdc=no
 fi
@@ -4873,14 +4653,18 @@ fi
 
 if test $ac_cv_header_stdc = yes; then
   # ISC 2.0.2 stdlib.h does not declare free, contrary to ANSI.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <stdlib.h>
 
 _ACEOF
 if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
-  $EGREP "free" >/dev/null 2>&1; then :
-
+  $EGREP "free" >/dev/null 2>&1; then
+  :
 else
   ac_cv_header_stdc=no
 fi
@@ -4890,10 +4674,14 @@ fi
 
 if test $ac_cv_header_stdc = yes; then
   # /bin/cc in Irix-4.0.5 gets non-ANSI ctype macros unless using -ansi.
-  if test "$cross_compiling" = yes; then :
+  if test "$cross_compiling" = yes; then
   :
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <ctype.h>
 #include <stdlib.h>
@@ -4920,33 +4708,118 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_run "$LINENO"; then :
-
+rm -f conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && { ac_try='./conftest$ac_exeext'
+  { (case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; }; then
+  :
 else
-  ac_cv_header_stdc=no
+  $as_echo "$as_me: program exited with status $ac_status" >&5
+$as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+( exit $ac_status )
+ac_cv_header_stdc=no
 fi
-rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
-  conftest.$ac_objext conftest.beam conftest.$ac_ext
+rm -rf conftest.dSYM
+rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext conftest.$ac_objext conftest.$ac_ext
 fi
 
+
 fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_header_stdc" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_header_stdc" >&5
 $as_echo "$ac_cv_header_stdc" >&6; }
 if test $ac_cv_header_stdc = yes; then
 
-$as_echo "#define STDC_HEADERS 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define STDC_HEADERS 1
+_ACEOF
 
 fi
 
 # On IRIX 5.3, sys/types and inttypes.h are conflicting.
+
+
+
+
+
+
+
+
+
 for ac_header in sys/types.h sys/stat.h stdlib.h string.h memory.h strings.h \
 		  inttypes.h stdint.h unistd.h
-do :
-  as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-ac_fn_c_check_header_compile "$LINENO" "$ac_header" "$as_ac_Header" "$ac_includes_default
-"
-if eval test \"x\$"$as_ac_Header"\" = x"yes"; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  eval "$as_ac_Header=yes"
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	eval "$as_ac_Header=no"
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
 #define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
@@ -4956,48 +4829,604 @@ fi
 done
 
 
+
 for ac_header in stdint.h
-do :
-  ac_fn_c_check_header_mongrel "$LINENO" "stdint.h" "ac_cv_header_stdint_h" "$ac_includes_default"
-if test "x$ac_cv_header_stdint_h" = xyes; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
+$as_echo_n "checking $ac_header usability... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_header_compiler=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_header_compiler=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
+$as_echo_n "checking $ac_header presence... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <$ac_header>
+_ACEOF
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  ac_header_preproc=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_header_preproc=no
+fi
+
+rm -f conftest.err conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
+  yes:no: )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
+    ac_header_preproc=yes
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
+$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
+    ( cat <<\_ASBOX
+## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##
+_ASBOX
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  eval "$as_ac_Header=\$ac_header_preproc"
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+
+fi
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_STDINT_H 1
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
 
 fi
 
 done
 
+
 for ac_header in malloc/malloc.h
-do :
-  ac_fn_c_check_header_mongrel "$LINENO" "malloc/malloc.h" "ac_cv_header_malloc_malloc_h" "$ac_includes_default"
-if test "x$ac_cv_header_malloc_malloc_h" = xyes; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
+$as_echo_n "checking $ac_header usability... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_header_compiler=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_header_compiler=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
+$as_echo_n "checking $ac_header presence... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <$ac_header>
+_ACEOF
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  ac_header_preproc=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_header_preproc=no
+fi
+
+rm -f conftest.err conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
+  yes:no: )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
+    ac_header_preproc=yes
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
+$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
+    ( cat <<\_ASBOX
+## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##
+_ASBOX
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  eval "$as_ac_Header=\$ac_header_preproc"
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+
+fi
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_MALLOC_MALLOC_H 1
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
 
 fi
 
 done
 
+
 for ac_header in malloc.h
-do :
-  ac_fn_c_check_header_mongrel "$LINENO" "malloc.h" "ac_cv_header_malloc_h" "$ac_includes_default"
-if test "x$ac_cv_header_malloc_h" = xyes; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
+$as_echo_n "checking $ac_header usability... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_header_compiler=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_header_compiler=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
+$as_echo_n "checking $ac_header presence... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <$ac_header>
+_ACEOF
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  ac_header_preproc=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_header_preproc=no
+fi
+
+rm -f conftest.err conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
+  yes:no: )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
+    ac_header_preproc=yes
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
+$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
+    ( cat <<\_ASBOX
+## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##
+_ASBOX
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  eval "$as_ac_Header=\$ac_header_preproc"
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+
+fi
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_MALLOC_H 1
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
 
 fi
 
 done
 
+
 for ac_header in endian.h
-do :
-  ac_fn_c_check_header_mongrel "$LINENO" "endian.h" "ac_cv_header_endian_h" "$ac_includes_default"
-if test "x$ac_cv_header_endian_h" = xyes; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
+$as_echo_n "checking $ac_header usability... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_header_compiler=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_header_compiler=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
+$as_echo_n "checking $ac_header presence... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <$ac_header>
+_ACEOF
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  ac_header_preproc=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_header_preproc=no
+fi
+
+rm -f conftest.err conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
+  yes:no: )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
+    ac_header_preproc=yes
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
+$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
+    ( cat <<\_ASBOX
+## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##
+_ASBOX
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  eval "$as_ac_Header=\$ac_header_preproc"
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+
+fi
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_ENDIAN_H 1
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
 
 fi
@@ -5005,35 +5434,244 @@ fi
 done
 
 #AC_CHECK_HEADERS(machine/endian.h)
-ac_fn_c_check_decl "$LINENO" "ntohll" "ac_cv_have_decl_ntohll" "#include <arpa/inet.h>
-"
-if test "x$ac_cv_have_decl_ntohll" = xyes; then :
-  ac_have_decl=1
+{ $as_echo "$as_me:$LINENO: checking whether ntohll is declared" >&5
+$as_echo_n "checking whether ntohll is declared... " >&6; }
+if test "${ac_cv_have_decl_ntohll+set}" = set; then
+  $as_echo_n "(cached) " >&6
 else
-  ac_have_decl=0
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <arpa/inet.h>
+
+int
+main ()
+{
+#ifndef ntohll
+  (void) ntohll;
+#endif
+
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_cv_have_decl_ntohll=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cv_have_decl_ntohll=no
 fi
 
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_have_decl_ntohll" >&5
+$as_echo "$ac_cv_have_decl_ntohll" >&6; }
+if test "x$ac_cv_have_decl_ntohll" = x""yes; then
+
 cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_NTOHLL $ac_have_decl
+#define HAVE_DECL_NTOHLL 1
 _ACEOF
 
-ac_fn_c_check_decl "$LINENO" "be64toh" "ac_cv_have_decl_be64toh" "#include <arpa/inet.h>
-"
-if test "x$ac_cv_have_decl_be64toh" = xyes; then :
-  ac_have_decl=1
+
 else
-  ac_have_decl=0
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_DECL_NTOHLL 0
+_ACEOF
+
+
 fi
 
-cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_BE64TOH $ac_have_decl
+
+{ $as_echo "$as_me:$LINENO: checking whether be64toh is declared" >&5
+$as_echo_n "checking whether be64toh is declared... " >&6; }
+if test "${ac_cv_have_decl_be64toh+set}" = set; then
+  $as_echo_n "(cached) " >&6
+else
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
 _ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <arpa/inet.h>
+
+int
+main ()
+{
+#ifndef be64toh
+  (void) be64toh;
+#endif
+
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_cv_have_decl_be64toh=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cv_have_decl_be64toh=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_have_decl_be64toh" >&5
+$as_echo "$ac_cv_have_decl_be64toh" >&6; }
+if test "x$ac_cv_have_decl_be64toh" = x""yes; then
+
+cat >>confdefs.h <<_ACEOF
+#define HAVE_DECL_BE64TOH 1
+_ACEOF
+
+
+else
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_DECL_BE64TOH 0
+_ACEOF
+
+
+fi
+
 
 
 # Checks for typedefs, structures, and compiler characteristics.
-ac_fn_c_check_type "$LINENO" "size_t" "ac_cv_type_size_t" "$ac_includes_default"
-if test "x$ac_cv_type_size_t" = xyes; then :
+{ $as_echo "$as_me:$LINENO: checking for size_t" >&5
+$as_echo_n "checking for size_t... " >&6; }
+if test "${ac_cv_type_size_t+set}" = set; then
+  $as_echo_n "(cached) " >&6
+else
+  ac_cv_type_size_t=no
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+if (sizeof (size_t))
+       return 0;
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+if (sizeof ((size_t)))
+	  return 0;
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  :
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
 
+	ac_cv_type_size_t=yes
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_type_size_t" >&5
+$as_echo "$ac_cv_type_size_t" >&6; }
+if test "x$ac_cv_type_size_t" = x""yes; then
+  :
 else
 
 cat >>confdefs.h <<_ACEOF
@@ -5042,12 +5680,75 @@ _ACEOF
 
 fi
 
-ac_fn_c_find_uintX_t "$LINENO" "32" "ac_cv_c_uint32_t"
-case $ac_cv_c_uint32_t in #(
+
+  { $as_echo "$as_me:$LINENO: checking for uint32_t" >&5
+$as_echo_n "checking for uint32_t... " >&6; }
+if test "${ac_cv_c_uint32_t+set}" = set; then
+  $as_echo_n "(cached) " >&6
+else
+  ac_cv_c_uint32_t=no
+     for ac_type in 'uint32_t' 'unsigned int' 'unsigned long int' \
+	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
+       cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+static int test_array [1 - 2 * !(($ac_type) -1 >> (32 - 1) == 1)];
+test_array [0] = 0
+
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  case $ac_type in
+  uint32_t) ac_cv_c_uint32_t=yes ;;
+  *) ac_cv_c_uint32_t=$ac_type ;;
+esac
+
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+       test "$ac_cv_c_uint32_t" != no && break
+     done
+fi
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_uint32_t" >&5
+$as_echo "$ac_cv_c_uint32_t" >&6; }
+  case $ac_cv_c_uint32_t in #(
   no|yes) ;; #(
   *)
 
-$as_echo "#define _UINT32_T 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define _UINT32_T 1
+_ACEOF
 
 
 cat >>confdefs.h <<_ACEOF
@@ -5056,12 +5757,75 @@ _ACEOF
 ;;
   esac
 
-ac_fn_c_find_uintX_t "$LINENO" "64" "ac_cv_c_uint64_t"
-case $ac_cv_c_uint64_t in #(
+
+  { $as_echo "$as_me:$LINENO: checking for uint64_t" >&5
+$as_echo_n "checking for uint64_t... " >&6; }
+if test "${ac_cv_c_uint64_t+set}" = set; then
+  $as_echo_n "(cached) " >&6
+else
+  ac_cv_c_uint64_t=no
+     for ac_type in 'uint64_t' 'unsigned int' 'unsigned long int' \
+	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
+       cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+static int test_array [1 - 2 * !(($ac_type) -1 >> (64 - 1) == 1)];
+test_array [0] = 0
+
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  case $ac_type in
+  uint64_t) ac_cv_c_uint64_t=yes ;;
+  *) ac_cv_c_uint64_t=$ac_type ;;
+esac
+
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+       test "$ac_cv_c_uint64_t" != no && break
+     done
+fi
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_uint64_t" >&5
+$as_echo "$ac_cv_c_uint64_t" >&6; }
+  case $ac_cv_c_uint64_t in #(
   no|yes) ;; #(
   *)
 
-$as_echo "#define _UINT64_T 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define _UINT64_T 1
+_ACEOF
 
 
 cat >>confdefs.h <<_ACEOF
@@ -5072,12 +5836,102 @@ _ACEOF
 
 
 # Checks for library functions.
+
 for ac_func in gettimeofday
-do :
-  ac_fn_c_check_func "$LINENO" "gettimeofday" "ac_cv_func_gettimeofday"
-if test "x$ac_cv_func_gettimeofday" = xyes; then :
+do
+as_ac_var=`$as_echo "ac_cv_func_$ac_func" | $as_tr_sh`
+{ $as_echo "$as_me:$LINENO: checking for $ac_func" >&5
+$as_echo_n "checking for $ac_func... " >&6; }
+if { as_var=$as_ac_var; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+/* Define $ac_func to an innocuous variant, in case <limits.h> declares $ac_func.
+   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
+#define $ac_func innocuous_$ac_func
+
+/* System header to define __stub macros and hopefully few prototypes,
+    which can conflict with char $ac_func (); below.
+    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
+    <limits.h> exists even on freestanding compilers.  */
+
+#ifdef __STDC__
+# include <limits.h>
+#else
+# include <assert.h>
+#endif
+
+#undef $ac_func
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char $ac_func ();
+/* The GNU C library defines this for functions which it implements
+    to always fail with ENOSYS.  Some functions are actually named
+    something starting with __ and the normal name is an alias.  */
+#if defined __stub_$ac_func || defined __stub___$ac_func
+choke me
+#endif
+
+int
+main ()
+{
+return $ac_func ();
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_link") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 $as_test_x conftest$ac_exeext
+       }; then
+  eval "$as_ac_var=yes"
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	eval "$as_ac_var=no"
+fi
+
+rm -rf conftest.dSYM
+rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
+      conftest$ac_exeext conftest.$ac_ext
+fi
+ac_res=`eval 'as_val=${'$as_ac_var'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+as_val=`eval 'as_val=${'$as_ac_var'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_GETTIMEOFDAY 1
+#define `$as_echo "HAVE_$ac_func" | $as_tr_cpp` 1
 _ACEOF
 
 fi
@@ -5085,7 +5939,7 @@ done
 
 
 # Check whether --enable-simd was given.
-if test "${enable_simd+set}" = set; then :
+if test "${enable_simd+set}" = set; then
   enableval=$enable_simd; \
 	ac_SIMD=${enable_simd}
 else
@@ -5097,35 +5951,45 @@ case ${ac_SIMD} in
      SSE4)
        echo Configuring for SSE4
 
-$as_echo "#define SSE4 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define SSE4 1
+_ACEOF
 
      ;;
      AVX)
        echo Configuring for AVX
 
-$as_echo "#define AVX1 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define AVX1 1
+_ACEOF
 
      ;;
      AVX2)
        echo Configuring for AVX2
 
-$as_echo "#define AVX2 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define AVX2 1
+_ACEOF
 
      ;;
      AVX512|MIC)
        echo Configuring for AVX512 and MIC
 
-$as_echo "#define AVX512 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define AVX512 1
+_ACEOF
 
      ;;
      *)
-     as_fn_error $? "${ac_SIMD} unsupported --enable-simd option" "$LINENO" 5;
+     { { $as_echo "$as_me:$LINENO: error: ${ac_SIMD} unsupported --enable-simd option" >&5
+$as_echo "$as_me: error: ${ac_SIMD} unsupported --enable-simd option" >&2;}
+   { (exit 1); exit 1; }; };
      ;;
 esac
 
 
 # Check whether --enable-comms was given.
-if test "${enable_comms+set}" = set; then :
+if test "${enable_comms+set}" = set; then
   enableval=$enable_comms; ac_COMMS=${enable_comms}
 else
   ac_COMMS=none
@@ -5136,17 +6000,23 @@ case ${ac_COMMS} in
      none)
        echo Configuring for NO communications
 
-$as_echo "#define GRID_COMMS_NONE 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define GRID_COMMS_NONE 1
+_ACEOF
 
      ;;
      mpi)
        echo Configuring for MPI communications
 
-$as_echo "#define GRID_COMMS_MPI 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define GRID_COMMS_MPI 1
+_ACEOF
 
      ;;
      *)
-     as_fn_error $? "${ac_COMMS} unsupported --enable-comms option" "$LINENO" 5;
+     { { $as_echo "$as_me:$LINENO: error: ${ac_COMMS} unsupported --enable-comms option" >&5
+$as_echo "$as_me: error: ${ac_COMMS} unsupported --enable-comms option" >&2;}
+   { (exit 1); exit 1; }; };
      ;;
 esac
 
@@ -5203,13 +6073,13 @@ _ACEOF
     case $ac_val in #(
     *${as_nl}*)
       case $ac_var in #(
-      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
+      *_cv_*) { $as_echo "$as_me:$LINENO: WARNING: cache variable $ac_var contains a newline" >&5
 $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
       esac
       case $ac_var in #(
       _ | IFS | as_nl) ;; #(
       BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
-      *) { eval $ac_var=; unset $ac_var;} ;;
+      *) $as_unset $ac_var ;;
       esac ;;
     esac
   done
@@ -5217,8 +6087,8 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
   (set) 2>&1 |
     case $as_nl`(ac_space=' '; set) 2>&1` in #(
     *${as_nl}ac_space=\ *)
-      # `set' does not quote correctly, so add quotes: double-quote
-      # substitution turns \\\\ into \\, and sed turns \\ into \.
+      # `set' does not quote correctly, so add quotes (double-quote
+      # substitution turns \\\\ into \\, and sed turns \\ into \).
       sed -n \
 	"s/'/'\\\\''/g;
 	  s/^\\([_$as_cr_alnum]*_cv_[_$as_cr_alnum]*\\)=\\(.*\\)/\\1='\\2'/p"
@@ -5240,23 +6110,12 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
      :end' >>confcache
 if diff "$cache_file" confcache >/dev/null 2>&1; then :; else
   if test -w "$cache_file"; then
-    if test "x$cache_file" != "x/dev/null"; then
-      { $as_echo "$as_me:${as_lineno-$LINENO}: updating cache $cache_file" >&5
+    test "x$cache_file" != "x/dev/null" &&
+      { $as_echo "$as_me:$LINENO: updating cache $cache_file" >&5
 $as_echo "$as_me: updating cache $cache_file" >&6;}
-      if test ! -f "$cache_file" || test -h "$cache_file"; then
-	cat confcache >"$cache_file"
-      else
-        case $cache_file in #(
-        */* | ?:*)
-	  mv -f confcache "$cache_file"$$ &&
-	  mv -f "$cache_file"$$ "$cache_file" ;; #(
-        *)
-	  mv -f confcache "$cache_file" ;;
-	esac
-      fi
-    fi
+    cat confcache >$cache_file
   else
-    { $as_echo "$as_me:${as_lineno-$LINENO}: not updating unwritable cache $cache_file" >&5
+    { $as_echo "$as_me:$LINENO: not updating unwritable cache $cache_file" >&5
 $as_echo "$as_me: not updating unwritable cache $cache_file" >&6;}
   fi
 fi
@@ -5270,29 +6129,20 @@ DEFS=-DHAVE_CONFIG_H
 
 ac_libobjs=
 ac_ltlibobjs=
-U=
 for ac_i in : $LIBOBJS; do test "x$ac_i" = x: && continue
   # 1. Remove the extension, and $U if already installed.
   ac_script='s/\$U\././;s/\.o$//;s/\.obj$//'
   ac_i=`$as_echo "$ac_i" | sed "$ac_script"`
   # 2. Prepend LIBOBJDIR.  When used with automake>=1.10 LIBOBJDIR
   #    will be set to the directory where LIBOBJS objects are built.
-  as_fn_append ac_libobjs " \${LIBOBJDIR}$ac_i\$U.$ac_objext"
-  as_fn_append ac_ltlibobjs " \${LIBOBJDIR}$ac_i"'$U.lo'
+  ac_libobjs="$ac_libobjs \${LIBOBJDIR}$ac_i\$U.$ac_objext"
+  ac_ltlibobjs="$ac_ltlibobjs \${LIBOBJDIR}$ac_i"'$U.lo'
 done
 LIBOBJS=$ac_libobjs
 
 LTLIBOBJS=$ac_ltlibobjs
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking that generated files are newer than configure" >&5
-$as_echo_n "checking that generated files are newer than configure... " >&6; }
-   if test -n "$am_sleep_pid"; then
-     # Hide warnings about reused PIDs.
-     wait $am_sleep_pid 2>/dev/null
-   fi
-   { $as_echo "$as_me:${as_lineno-$LINENO}: result: done" >&5
-$as_echo "done" >&6; }
  if test -n "$EXEEXT"; then
   am__EXEEXT_TRUE=
   am__EXEEXT_FALSE='#'
@@ -5302,34 +6152,48 @@ else
 fi
 
 if test -z "${AMDEP_TRUE}" && test -z "${AMDEP_FALSE}"; then
-  as_fn_error $? "conditional \"AMDEP\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: conditional \"AMDEP\" was never defined.
+Usually this means the macro was only invoked conditionally." >&5
+$as_echo "$as_me: error: conditional \"AMDEP\" was never defined.
+Usually this means the macro was only invoked conditionally." >&2;}
+   { (exit 1); exit 1; }; }
 fi
 if test -z "${am__fastdepCXX_TRUE}" && test -z "${am__fastdepCXX_FALSE}"; then
-  as_fn_error $? "conditional \"am__fastdepCXX\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: conditional \"am__fastdepCXX\" was never defined.
+Usually this means the macro was only invoked conditionally." >&5
+$as_echo "$as_me: error: conditional \"am__fastdepCXX\" was never defined.
+Usually this means the macro was only invoked conditionally." >&2;}
+   { (exit 1); exit 1; }; }
 fi
 if test -z "${am__fastdepCC_TRUE}" && test -z "${am__fastdepCC_FALSE}"; then
-  as_fn_error $? "conditional \"am__fastdepCC\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: conditional \"am__fastdepCC\" was never defined.
+Usually this means the macro was only invoked conditionally." >&5
+$as_echo "$as_me: error: conditional \"am__fastdepCC\" was never defined.
+Usually this means the macro was only invoked conditionally." >&2;}
+   { (exit 1); exit 1; }; }
 fi
 if test -z "${BUILD_COMMS_MPI_TRUE}" && test -z "${BUILD_COMMS_MPI_FALSE}"; then
-  as_fn_error $? "conditional \"BUILD_COMMS_MPI\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: conditional \"BUILD_COMMS_MPI\" was never defined.
+Usually this means the macro was only invoked conditionally." >&5
+$as_echo "$as_me: error: conditional \"BUILD_COMMS_MPI\" was never defined.
+Usually this means the macro was only invoked conditionally." >&2;}
+   { (exit 1); exit 1; }; }
 fi
 if test -z "${BUILD_COMMS_NONE_TRUE}" && test -z "${BUILD_COMMS_NONE_FALSE}"; then
-  as_fn_error $? "conditional \"BUILD_COMMS_NONE\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: conditional \"BUILD_COMMS_NONE\" was never defined.
+Usually this means the macro was only invoked conditionally." >&5
+$as_echo "$as_me: error: conditional \"BUILD_COMMS_NONE\" was never defined.
+Usually this means the macro was only invoked conditionally." >&2;}
+   { (exit 1); exit 1; }; }
 fi
 
-: "${CONFIG_STATUS=./config.status}"
+: ${CONFIG_STATUS=./config.status}
 ac_write_fail=0
 ac_clean_files_save=$ac_clean_files
 ac_clean_files="$ac_clean_files $CONFIG_STATUS"
-{ $as_echo "$as_me:${as_lineno-$LINENO}: creating $CONFIG_STATUS" >&5
+{ $as_echo "$as_me:$LINENO: creating $CONFIG_STATUS" >&5
 $as_echo "$as_me: creating $CONFIG_STATUS" >&6;}
-as_write_fail=0
-cat >$CONFIG_STATUS <<_ASEOF || as_write_fail=1
+cat >$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 #! $SHELL
 # Generated by $as_me.
 # Run this file to recreate the current configuration.
@@ -5339,18 +6203,17 @@ cat >$CONFIG_STATUS <<_ASEOF || as_write_fail=1
 debug=false
 ac_cs_recheck=false
 ac_cs_silent=false
-
 SHELL=\${CONFIG_SHELL-$SHELL}
-export SHELL
-_ASEOF
-cat >>$CONFIG_STATUS <<\_ASEOF || as_write_fail=1
-## -------------------- ##
-## M4sh Initialization. ##
-## -------------------- ##
+_ACEOF
+
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+## --------------------- ##
+## M4sh Initialization.  ##
+## --------------------- ##
 
 # Be more Bourne compatible
 DUALCASE=1; export DUALCASE # for MKS sh
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
   emulate sh
   NULLCMD=:
   # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
@@ -5358,15 +6221,23 @@ if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
   alias -g '${1+"$@"}'='"$@"'
   setopt NO_GLOB_SUBST
 else
-  case `(set -o) 2>/dev/null` in #(
-  *posix*) :
-    set -o posix ;; #(
-  *) :
-     ;;
+  case `(set -o) 2>/dev/null` in
+  *posix*) set -o posix ;;
 esac
+
 fi
 
 
+
+
+# PATH needs CR
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
+
 as_nl='
 '
 export as_nl
@@ -5374,13 +6245,7 @@ export as_nl
 as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
-# Prefer a ksh shell builtin over an external printf program on Solaris,
-# but without wasting forks for bash or zsh.
-if test -z "$BASH_VERSION$ZSH_VERSION" \
-    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
-  as_echo='print -r --'
-  as_echo_n='print -rn --'
-elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+if (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
   as_echo='printf %s\n'
   as_echo_n='printf %s'
 else
@@ -5391,7 +6256,7 @@ else
     as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
     as_echo_n_body='eval
       arg=$1;
-      case $arg in #(
+      case $arg in
       *"$as_nl"*)
 	expr "X$arg" : "X\\(.*\\)$as_nl";
 	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
@@ -5414,6 +6279,13 @@ if test "${PATH_SEPARATOR+set}" != set; then
   }
 fi
 
+# Support unset when possible.
+if ( (MAIL=60; unset MAIL) || exit) >/dev/null 2>&1; then
+  as_unset=unset
+else
+  as_unset=false
+fi
+
 
 # IFS
 # We need space, tab and new line, in precisely that order.  Quoting is
@@ -5423,16 +6295,15 @@ fi
 IFS=" ""	$as_nl"
 
 # Find who we are.  Look in the path if we contain no directory separator.
-as_myself=
-case $0 in #((
+case $0 in
   *[\\/]* ) as_myself=$0 ;;
   *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
 for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
-  done
+  test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+done
 IFS=$as_save_IFS
 
      ;;
@@ -5444,16 +6315,12 @@ if test "x$as_myself" = x; then
 fi
 if test ! -f "$as_myself"; then
   $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
-  exit 1
+  { (exit 1); exit 1; }
 fi
 
-# Unset variables that we do not need and which cause bugs (e.g. in
-# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
-# suppresses any "Segmentation fault" message there.  '((' could
-# trigger a bug in pdksh 5.2.14.
-for as_var in BASH_ENV ENV MAIL MAILPATH
-do eval test x\${$as_var+set} = xset \
-  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
+# Work around bugs in pre-3.0 UWIN ksh.
+for as_var in ENV MAIL MAILPATH
+do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
 done
 PS1='$ '
 PS2='> '
@@ -5465,89 +6332,7 @@ export LC_ALL
 LANGUAGE=C
 export LANGUAGE
 
-# CDPATH.
-(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
-
-
-# as_fn_error STATUS ERROR [LINENO LOG_FD]
-# ----------------------------------------
-# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
-# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
-# script with STATUS, using 1 if that was 0.
-as_fn_error ()
-{
-  as_status=$1; test $as_status -eq 0 && as_status=1
-  if test "$4"; then
-    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
-  fi
-  $as_echo "$as_me: error: $2" >&2
-  as_fn_exit $as_status
-} # as_fn_error
-
-
-# as_fn_set_status STATUS
-# -----------------------
-# Set $? to STATUS, without forking.
-as_fn_set_status ()
-{
-  return $1
-} # as_fn_set_status
-
-# as_fn_exit STATUS
-# -----------------
-# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
-as_fn_exit ()
-{
-  set +e
-  as_fn_set_status $1
-  exit $1
-} # as_fn_exit
-
-# as_fn_unset VAR
-# ---------------
-# Portably unset VAR.
-as_fn_unset ()
-{
-  { eval $1=; unset $1;}
-}
-as_unset=as_fn_unset
-# as_fn_append VAR VALUE
-# ----------------------
-# Append the text in VALUE to the end of the definition contained in VAR. Take
-# advantage of any shell optimizations that allow amortized linear growth over
-# repeated appends, instead of the typical quadratic growth present in naive
-# implementations.
-if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
-  eval 'as_fn_append ()
-  {
-    eval $1+=\$2
-  }'
-else
-  as_fn_append ()
-  {
-    eval $1=\$$1\$2
-  }
-fi # as_fn_append
-
-# as_fn_arith ARG...
-# ------------------
-# Perform arithmetic evaluation on the ARGs, and store the result in the
-# global $as_val. Take advantage of shells that can avoid forks. The arguments
-# must be portable across $(()) and expr.
-if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
-  eval 'as_fn_arith ()
-  {
-    as_val=$(( $* ))
-  }'
-else
-  as_fn_arith ()
-  {
-    as_val=`expr "$@" || test $? -eq 1`
-  }
-fi # as_fn_arith
-
-
+# Required to use basename.
 if expr a : '\(a\)' >/dev/null 2>&1 &&
    test "X`expr 00001 : '.*\(...\)'`" = X001; then
   as_expr=expr
@@ -5561,12 +6346,8 @@ else
   as_basename=false
 fi
 
-if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
-  as_dirname=dirname
-else
-  as_dirname=false
-fi
 
+# Name of the executable.
 as_me=`$as_basename -- "$0" ||
 $as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
 	 X"$0" : 'X\(//\)$' \| \
@@ -5586,25 +6367,76 @@ $as_echo X/"$0" |
 	  }
 	  s/.*/./; q'`
 
-# Avoid depending upon Character Ranges.
-as_cr_letters='abcdefghijklmnopqrstuvwxyz'
-as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
-as_cr_Letters=$as_cr_letters$as_cr_LETTERS
-as_cr_digits='0123456789'
-as_cr_alnum=$as_cr_Letters$as_cr_digits
+# CDPATH.
+$as_unset CDPATH
+
+
+
+  as_lineno_1=$LINENO
+  as_lineno_2=$LINENO
+  test "x$as_lineno_1" != "x$as_lineno_2" &&
+  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2" || {
+
+  # Create $as_me.lineno as a copy of $as_myself, but with $LINENO
+  # uniformly replaced by the line number.  The first 'sed' inserts a
+  # line-number line after each line using $LINENO; the second 'sed'
+  # does the real work.  The second script uses 'N' to pair each
+  # line-number line with the line containing $LINENO, and appends
+  # trailing '-' during substitution so that $LINENO is not a special
+  # case at line end.
+  # (Raja R Harinath suggested sed '=', and Paul Eggert wrote the
+  # scripts with optimization help from Paolo Bonzini.  Blame Lee
+  # E. McMahon (1931-1989) for sed's syntax.  :-)
+  sed -n '
+    p
+    /[$]LINENO/=
+  ' <$as_myself |
+    sed '
+      s/[$]LINENO.*/&-/
+      t lineno
+      b
+      :lineno
+      N
+      :loop
+      s/[$]LINENO\([^'$as_cr_alnum'_].*\n\)\(.*\)/\2\1\2/
+      t loop
+      s/-\n.*//
+    ' >$as_me.lineno &&
+  chmod +x "$as_me.lineno" ||
+    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2
+   { (exit 1); exit 1; }; }
+
+  # Don't try to exec as it changes $[0], causing all sort of problems
+  # (the dirname of $[0] is not the place where we might find the
+  # original and so on.  Autoconf is especially sensitive to this).
+  . "./$as_me.lineno"
+  # Exit status is that of the last command.
+  exit
+}
+
+
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
 
 ECHO_C= ECHO_N= ECHO_T=
-case `echo -n x` in #(((((
+case `echo -n x` in
 -n*)
-  case `echo 'xy\c'` in
+  case `echo 'x\c'` in
   *c*) ECHO_T='	';;	# ECHO_T is single tab character.
-  xy)  ECHO_C='\c';;
-  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
-       ECHO_T='	';;
+  *)   ECHO_C='\c';;
   esac;;
 *)
   ECHO_N='-n';;
 esac
+if expr a : '\(a\)' >/dev/null 2>&1 &&
+   test "X`expr 00001 : '.*\(...\)'`" = X001; then
+  as_expr=expr
+else
+  as_expr=false
+fi
 
 rm -f conf$$ conf$$.exe conf$$.file
 if test -d conf$$.dir; then
@@ -5619,85 +6451,49 @@ if (echo >conf$$.file) 2>/dev/null; then
     # ... but there are two gotchas:
     # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
     # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
-    # In both cases, we have to default to `cp -pR'.
+    # In both cases, we have to default to `cp -p'.
     ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
-      as_ln_s='cp -pR'
+      as_ln_s='cp -p'
   elif ln conf$$.file conf$$ 2>/dev/null; then
     as_ln_s=ln
   else
-    as_ln_s='cp -pR'
+    as_ln_s='cp -p'
   fi
 else
-  as_ln_s='cp -pR'
+  as_ln_s='cp -p'
 fi
 rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
 rmdir conf$$.dir 2>/dev/null
 
-
-# as_fn_mkdir_p
-# -------------
-# Create "$as_dir" as a directory, including parents if necessary.
-as_fn_mkdir_p ()
-{
-
-  case $as_dir in #(
-  -*) as_dir=./$as_dir;;
-  esac
-  test -d "$as_dir" || eval $as_mkdir_p || {
-    as_dirs=
-    while :; do
-      case $as_dir in #(
-      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
-      *) as_qdir=$as_dir;;
-      esac
-      as_dirs="'$as_qdir' $as_dirs"
-      as_dir=`$as_dirname -- "$as_dir" ||
-$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_dir" : 'X\(//\)[^/]' \| \
-	 X"$as_dir" : 'X\(//\)$' \| \
-	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_dir" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      test -d "$as_dir" && break
-    done
-    test -z "$as_dirs" || eval "mkdir $as_dirs"
-  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
-
-
-} # as_fn_mkdir_p
 if mkdir -p . 2>/dev/null; then
-  as_mkdir_p='mkdir -p "$as_dir"'
+  as_mkdir_p=:
 else
   test -d ./-p && rmdir ./-p
   as_mkdir_p=false
 fi
 
-
-# as_fn_executable_p FILE
-# -----------------------
-# Test if FILE is an executable regular file.
-as_fn_executable_p ()
-{
-  test -f "$1" && test -x "$1"
-} # as_fn_executable_p
-as_test_x='test -x'
-as_executable_p=as_fn_executable_p
+if test -x / >/dev/null 2>&1; then
+  as_test_x='test -x'
+else
+  if ls -dL / >/dev/null 2>&1; then
+    as_ls_L_option=L
+  else
+    as_ls_L_option=
+  fi
+  as_test_x='
+    eval sh -c '\''
+      if test -d "$1"; then
+	test -d "$1/.";
+      else
+	case $1 in
+	-*)set "./$1";;
+	esac;
+	case `ls -ld'$as_ls_L_option' "$1" 2>/dev/null` in
+	???[sx]*):;;*)false;;esac;fi
+    '\'' sh
+  '
+fi
+as_executable_p=$as_test_x
 
 # Sed expression to map a string onto a valid CPP name.
 as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
@@ -5707,19 +6503,13 @@ as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
 
 
 exec 6>&1
-## ----------------------------------- ##
-## Main body of $CONFIG_STATUS script. ##
-## ----------------------------------- ##
-_ASEOF
-test $as_write_fail = 0 && chmod +x $CONFIG_STATUS || ac_write_fail=1
 
-cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-# Save the log message, to keep $0 and so on meaningful, and to
+# Save the log message, to keep $[0] and so on meaningful, and to
 # report actual input values of CONFIG_FILES etc. instead of their
 # values after options handling.
 ac_log="
 This file was extended by Grid $as_me 1.0, which was
-generated by GNU Autoconf 2.69.  Invocation command line was
+generated by GNU Autoconf 2.63.  Invocation command line was
 
   CONFIG_FILES    = $CONFIG_FILES
   CONFIG_HEADERS  = $CONFIG_HEADERS
@@ -5751,15 +6541,13 @@ _ACEOF
 
 cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 ac_cs_usage="\
-\`$as_me' instantiates files and other configuration actions
-from templates according to the current configuration.  Unless the files
-and actions are specified as TAGs, all are instantiated by default.
+\`$as_me' instantiates files from templates according to the
+current configuration.
 
-Usage: $0 [OPTION]... [TAG]...
+Usage: $0 [OPTION]... [FILE]...
 
   -h, --help       print this help, then exit
   -V, --version    print version number and configuration settings, then exit
-      --config     print configuration, then exit
   -q, --quiet, --silent
                    do not print progress messages
   -d, --debug      don't remove temporary files
@@ -5778,17 +6566,16 @@ $config_headers
 Configuration commands:
 $config_commands
 
-Report bugs to <paboyle@ph.ed.ac.uk>."
+Report bugs to <bug-autoconf@gnu.org>."
 
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-ac_cs_config="`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`"
 ac_cs_version="\\
 Grid config.status 1.0
-configured by $0, generated by GNU Autoconf 2.69,
-  with options \\"\$ac_cs_config\\"
+configured by $0, generated by GNU Autoconf 2.63,
+  with options \\"`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`\\"
 
-Copyright (C) 2012 Free Software Foundation, Inc.
+Copyright (C) 2008 Free Software Foundation, Inc.
 This config.status script is free software; the Free Software Foundation
 gives unlimited permission to copy, distribute and modify it."
 
@@ -5806,16 +6593,11 @@ ac_need_defaults=:
 while test $# != 0
 do
   case $1 in
-  --*=?*)
+  --*=*)
     ac_option=`expr "X$1" : 'X\([^=]*\)='`
     ac_optarg=`expr "X$1" : 'X[^=]*=\(.*\)'`
     ac_shift=:
     ;;
-  --*=)
-    ac_option=`expr "X$1" : 'X\([^=]*\)='`
-    ac_optarg=
-    ac_shift=:
-    ;;
   *)
     ac_option=$1
     ac_optarg=$2
@@ -5829,29 +6611,27 @@ do
     ac_cs_recheck=: ;;
   --version | --versio | --versi | --vers | --ver | --ve | --v | -V )
     $as_echo "$ac_cs_version"; exit ;;
-  --config | --confi | --conf | --con | --co | --c )
-    $as_echo "$ac_cs_config"; exit ;;
   --debug | --debu | --deb | --de | --d | -d )
     debug=: ;;
   --file | --fil | --fi | --f )
     $ac_shift
     case $ac_optarg in
     *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
-    '') as_fn_error $? "missing file argument" ;;
     esac
-    as_fn_append CONFIG_FILES " '$ac_optarg'"
+    CONFIG_FILES="$CONFIG_FILES '$ac_optarg'"
     ac_need_defaults=false;;
   --header | --heade | --head | --hea )
     $ac_shift
     case $ac_optarg in
     *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
     esac
-    as_fn_append CONFIG_HEADERS " '$ac_optarg'"
+    CONFIG_HEADERS="$CONFIG_HEADERS '$ac_optarg'"
     ac_need_defaults=false;;
   --he | --h)
     # Conflict between --help and --header
-    as_fn_error $? "ambiguous option: \`$1'
-Try \`$0 --help' for more information.";;
+    { $as_echo "$as_me: error: ambiguous option: $1
+Try \`$0 --help' for more information." >&2
+   { (exit 1); exit 1; }; };;
   --help | --hel | -h )
     $as_echo "$ac_cs_usage"; exit ;;
   -q | -quiet | --quiet | --quie | --qui | --qu | --q \
@@ -5859,10 +6639,11 @@ Try \`$0 --help' for more information.";;
     ac_cs_silent=: ;;
 
   # This is an error.
-  -*) as_fn_error $? "unrecognized option: \`$1'
-Try \`$0 --help' for more information." ;;
+  -*) { $as_echo "$as_me: error: unrecognized option: $1
+Try \`$0 --help' for more information." >&2
+   { (exit 1); exit 1; }; } ;;
 
-  *) as_fn_append ac_config_targets " $1"
+  *) ac_config_targets="$ac_config_targets $1"
      ac_need_defaults=false ;;
 
   esac
@@ -5879,7 +6660,7 @@ fi
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 if \$ac_cs_recheck; then
-  set X $SHELL '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
+  set X '$SHELL' '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
   shift
   \$as_echo "running CONFIG_SHELL=$SHELL \$*" >&6
   CONFIG_SHELL='$SHELL'
@@ -5920,7 +6701,9 @@ do
     "tests/Makefile") CONFIG_FILES="$CONFIG_FILES tests/Makefile" ;;
     "benchmarks/Makefile") CONFIG_FILES="$CONFIG_FILES benchmarks/Makefile" ;;
 
-  *) as_fn_error $? "invalid argument: \`$ac_config_target'" "$LINENO" 5;;
+  *) { { $as_echo "$as_me:$LINENO: error: invalid argument: $ac_config_target" >&5
+$as_echo "$as_me: error: invalid argument: $ac_config_target" >&2;}
+   { (exit 1); exit 1; }; };;
   esac
 done
 
@@ -5943,24 +6726,26 @@ fi
 # after its creation but before its name has been assigned to `$tmp'.
 $debug ||
 {
-  tmp= ac_tmp=
+  tmp=
   trap 'exit_status=$?
-  : "${ac_tmp:=$tmp}"
-  { test ! -d "$ac_tmp" || rm -fr "$ac_tmp"; } && exit $exit_status
+  { test -z "$tmp" || test ! -d "$tmp" || rm -fr "$tmp"; } && exit $exit_status
 ' 0
-  trap 'as_fn_exit 1' 1 2 13 15
+  trap '{ (exit 1); exit 1; }' 1 2 13 15
 }
 # Create a (secure) tmp directory for tmp files.
 
 {
   tmp=`(umask 077 && mktemp -d "./confXXXXXX") 2>/dev/null` &&
-  test -d "$tmp"
+  test -n "$tmp" && test -d "$tmp"
 }  ||
 {
   tmp=./conf$$-$RANDOM
   (umask 077 && mkdir "$tmp")
-} || as_fn_error $? "cannot create a temporary directory in ." "$LINENO" 5
-ac_tmp=$tmp
+} ||
+{
+   $as_echo "$as_me: cannot create a temporary directory in ." >&2
+   { (exit 1); exit 1; }
+}
 
 # Set up the scripts for CONFIG_FILES section.
 # No need to generate them if there are no CONFIG_FILES.
@@ -5968,13 +6753,7 @@ ac_tmp=$tmp
 if test -n "$CONFIG_FILES"; then
 
 
-ac_cr=`echo X | tr X '\015'`
-# On cygwin, bash can eat \r inside `` if the user requested igncr.
-# But we know of no other shell where ac_cr would be empty at this
-# point, so we can use a bashism as a fallback.
-if test "x$ac_cr" = x; then
-  eval ac_cr=\$\'\\r\'
-fi
+ac_cr='
'
 ac_cs_awk_cr=`$AWK 'BEGIN { print "a\rb" }' </dev/null 2>/dev/null`
 if test "$ac_cs_awk_cr" = "a${ac_cr}b"; then
   ac_cs_awk_cr='\\r'
@@ -5982,7 +6761,7 @@ else
   ac_cs_awk_cr=$ac_cr
 fi
 
-echo 'BEGIN {' >"$ac_tmp/subs1.awk" &&
+echo 'BEGIN {' >"$tmp/subs1.awk" &&
 _ACEOF
 
 
@@ -5991,18 +6770,24 @@ _ACEOF
   echo "$ac_subst_vars" | sed 's/.*/&!$&$ac_delim/' &&
   echo "_ACEOF"
 } >conf$$subs.sh ||
-  as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
-ac_delim_num=`echo "$ac_subst_vars" | grep -c '^'`
+  { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
+$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
+   { (exit 1); exit 1; }; }
+ac_delim_num=`echo "$ac_subst_vars" | grep -c '$'`
 ac_delim='%!_!# '
 for ac_last_try in false false false false false :; do
   . ./conf$$subs.sh ||
-    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
+$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
+   { (exit 1); exit 1; }; }
 
   ac_delim_n=`sed -n "s/.*$ac_delim\$/X/p" conf$$subs.awk | grep -c X`
   if test $ac_delim_n = $ac_delim_num; then
     break
   elif $ac_last_try; then
-    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
+$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
+   { (exit 1); exit 1; }; }
   else
     ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
   fi
@@ -6010,7 +6795,7 @@ done
 rm -f conf$$subs.sh
 
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-cat >>"\$ac_tmp/subs1.awk" <<\\_ACAWK &&
+cat >>"\$tmp/subs1.awk" <<\\_ACAWK &&
 _ACEOF
 sed -n '
 h
@@ -6024,7 +6809,7 @@ s/'"$ac_delim"'$//
 t delim
 :nl
 h
-s/\(.\{148\}\)..*/\1/
+s/\(.\{148\}\).*/\1/
 t more1
 s/["\\]/\\&/g; s/^/"/; s/$/\\n"\\/
 p
@@ -6038,7 +6823,7 @@ s/.\{148\}//
 t nl
 :delim
 h
-s/\(.\{148\}\)..*/\1/
+s/\(.\{148\}\).*/\1/
 t more2
 s/["\\]/\\&/g; s/^/"/; s/$/"/
 p
@@ -6058,7 +6843,7 @@ t delim
 rm -f conf$$subs.awk
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 _ACAWK
-cat >>"\$ac_tmp/subs1.awk" <<_ACAWK &&
+cat >>"\$tmp/subs1.awk" <<_ACAWK &&
   for (key in S) S_is_set[key] = 1
   FS = ""
 
@@ -6090,29 +6875,23 @@ if sed "s/$ac_cr//" < /dev/null > /dev/null 2>&1; then
   sed "s/$ac_cr\$//; s/$ac_cr/$ac_cs_awk_cr/g"
 else
   cat
-fi < "$ac_tmp/subs1.awk" > "$ac_tmp/subs.awk" \
-  || as_fn_error $? "could not setup config files machinery" "$LINENO" 5
+fi < "$tmp/subs1.awk" > "$tmp/subs.awk" \
+  || { { $as_echo "$as_me:$LINENO: error: could not setup config files machinery" >&5
+$as_echo "$as_me: error: could not setup config files machinery" >&2;}
+   { (exit 1); exit 1; }; }
 _ACEOF
 
-# VPATH may cause trouble with some makes, so we remove sole $(srcdir),
-# ${srcdir} and @srcdir@ entries from VPATH if srcdir is ".", strip leading and
+# VPATH may cause trouble with some makes, so we remove $(srcdir),
+# ${srcdir} and @srcdir@ from VPATH if srcdir is ".", strip leading and
 # trailing colons and then remove the whole line if VPATH becomes empty
 # (actually we leave an empty line to preserve line numbers).
 if test "x$srcdir" = x.; then
-  ac_vpsub='/^[	 ]*VPATH[	 ]*=[	 ]*/{
-h
-s///
-s/^/:/
-s/[	 ]*$/:/
-s/:\$(srcdir):/:/g
-s/:\${srcdir}:/:/g
-s/:@srcdir@:/:/g
-s/^:*//
+  ac_vpsub='/^[	 ]*VPATH[	 ]*=/{
+s/:*\$(srcdir):*/:/
+s/:*\${srcdir}:*/:/
+s/:*@srcdir@:*/:/
+s/^\([^=]*=[	 ]*\):*/\1/
 s/:*$//
-x
-s/\(=[	 ]*\).*/\1/
-G
-s/\n//
 s/^[^=]*=[	 ]*$//
 }'
 fi
@@ -6124,7 +6903,7 @@ fi # test -n "$CONFIG_FILES"
 # No need to generate them if there are no CONFIG_HEADERS.
 # This happens for instance with `./config.status Makefile'.
 if test -n "$CONFIG_HEADERS"; then
-cat >"$ac_tmp/defines.awk" <<\_ACAWK ||
+cat >"$tmp/defines.awk" <<\_ACAWK ||
 BEGIN {
 _ACEOF
 
@@ -6136,11 +6915,13 @@ _ACEOF
 # handling of long lines.
 ac_delim='%!_!# '
 for ac_last_try in false false :; do
-  ac_tt=`sed -n "/$ac_delim/p" confdefs.h`
-  if test -z "$ac_tt"; then
+  ac_t=`sed -n "/$ac_delim/p" confdefs.h`
+  if test -z "$ac_t"; then
     break
   elif $ac_last_try; then
-    as_fn_error $? "could not make $CONFIG_HEADERS" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_HEADERS" >&5
+$as_echo "$as_me: error: could not make $CONFIG_HEADERS" >&2;}
+   { (exit 1); exit 1; }; }
   else
     ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
   fi
@@ -6225,7 +7006,9 @@ cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 _ACAWK
 _ACEOF
 cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-  as_fn_error $? "could not setup config headers machinery" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: could not setup config headers machinery" >&5
+$as_echo "$as_me: error: could not setup config headers machinery" >&2;}
+   { (exit 1); exit 1; }; }
 fi # test -n "$CONFIG_HEADERS"
 
 
@@ -6238,7 +7021,9 @@ do
   esac
   case $ac_mode$ac_tag in
   :[FHL]*:*);;
-  :L* | :C*:*) as_fn_error $? "invalid tag \`$ac_tag'" "$LINENO" 5;;
+  :L* | :C*:*) { { $as_echo "$as_me:$LINENO: error: invalid tag $ac_tag" >&5
+$as_echo "$as_me: error: invalid tag $ac_tag" >&2;}
+   { (exit 1); exit 1; }; };;
   :[FH]-) ac_tag=-:-;;
   :[FH]*) ac_tag=$ac_tag:$ac_tag.in;;
   esac
@@ -6257,7 +7042,7 @@ do
     for ac_f
     do
       case $ac_f in
-      -) ac_f="$ac_tmp/stdin";;
+      -) ac_f="$tmp/stdin";;
       *) # Look for the file first in the build tree, then in the source tree
 	 # (if the path is not absolute).  The absolute path cannot be DOS-style,
 	 # because $ac_f cannot contain `:'.
@@ -6266,10 +7051,12 @@ do
 	   [\\/$]*) false;;
 	   *) test -f "$srcdir/$ac_f" && ac_f="$srcdir/$ac_f";;
 	   esac ||
-	   as_fn_error 1 "cannot find input file: \`$ac_f'" "$LINENO" 5;;
+	   { { $as_echo "$as_me:$LINENO: error: cannot find input file: $ac_f" >&5
+$as_echo "$as_me: error: cannot find input file: $ac_f" >&2;}
+   { (exit 1); exit 1; }; };;
       esac
       case $ac_f in *\'*) ac_f=`$as_echo "$ac_f" | sed "s/'/'\\\\\\\\''/g"`;; esac
-      as_fn_append ac_file_inputs " '$ac_f'"
+      ac_file_inputs="$ac_file_inputs '$ac_f'"
     done
 
     # Let's still pretend it is `configure' which instantiates (i.e., don't
@@ -6280,7 +7067,7 @@ do
 	`' by configure.'
     if test x"$ac_file" != x-; then
       configure_input="$ac_file.  $configure_input"
-      { $as_echo "$as_me:${as_lineno-$LINENO}: creating $ac_file" >&5
+      { $as_echo "$as_me:$LINENO: creating $ac_file" >&5
 $as_echo "$as_me: creating $ac_file" >&6;}
     fi
     # Neutralize special characters interpreted by sed in replacement strings.
@@ -6292,8 +7079,10 @@ $as_echo "$as_me: creating $ac_file" >&6;}
     esac
 
     case $ac_tag in
-    *:-:* | *:-) cat >"$ac_tmp/stdin" \
-      || as_fn_error $? "could not create $ac_file" "$LINENO" 5 ;;
+    *:-:* | *:-) cat >"$tmp/stdin" \
+      || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; } ;;
     esac
     ;;
   esac
@@ -6321,7 +7110,47 @@ $as_echo X"$ac_file" |
 	    q
 	  }
 	  s/.*/./; q'`
-  as_dir="$ac_dir"; as_fn_mkdir_p
+  { as_dir="$ac_dir"
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || { $as_mkdir_p && mkdir -p "$as_dir"; } || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || { { $as_echo "$as_me:$LINENO: error: cannot create directory $as_dir" >&5
+$as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
+   { (exit 1); exit 1; }; }; }
   ac_builddir=.
 
 case "$ac_dir" in
@@ -6378,6 +7207,7 @@ cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 # If the template does not know about datarootdir, expand it.
 # FIXME: This hack should be removed a few years after 2.60.
 ac_datarootdir_hack=; ac_datarootdir_seen=
+
 ac_sed_dataroot='
 /datarootdir/ {
   p
@@ -6387,11 +7217,12 @@ ac_sed_dataroot='
 /@docdir@/p
 /@infodir@/p
 /@localedir@/p
-/@mandir@/p'
+/@mandir@/p
+'
 case `eval "sed -n \"\$ac_sed_dataroot\" $ac_file_inputs"` in
 *datarootdir*) ac_datarootdir_seen=yes;;
 *@datadir@*|*@docdir@*|*@infodir@*|*@localedir@*|*@mandir@*)
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
+  { $as_echo "$as_me:$LINENO: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
 $as_echo "$as_me: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&2;}
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
@@ -6401,7 +7232,7 @@ cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
   s&@infodir@&$infodir&g
   s&@localedir@&$localedir&g
   s&@mandir@&$mandir&g
-  s&\\\${datarootdir}&$datarootdir&g' ;;
+    s&\\\${datarootdir}&$datarootdir&g' ;;
 esac
 _ACEOF
 
@@ -6429,24 +7260,27 @@ s&@INSTALL@&$ac_INSTALL&;t t
 s&@MKDIR_P@&$ac_MKDIR_P&;t t
 $ac_datarootdir_hack
 "
-eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$ac_tmp/subs.awk" \
-  >$ac_tmp/out || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$tmp/subs.awk" >$tmp/out \
+  || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; }
 
 test -z "$ac_datarootdir_hack$ac_datarootdir_seen" &&
-  { ac_out=`sed -n '/\${datarootdir}/p' "$ac_tmp/out"`; test -n "$ac_out"; } &&
-  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' \
-      "$ac_tmp/out"`; test -z "$ac_out"; } &&
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file contains a reference to the variable \`datarootdir'
-which seems to be undefined.  Please make sure it is defined" >&5
+  { ac_out=`sed -n '/\${datarootdir}/p' "$tmp/out"`; test -n "$ac_out"; } &&
+  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' "$tmp/out"`; test -z "$ac_out"; } &&
+  { $as_echo "$as_me:$LINENO: WARNING: $ac_file contains a reference to the variable \`datarootdir'
+which seems to be undefined.  Please make sure it is defined." >&5
 $as_echo "$as_me: WARNING: $ac_file contains a reference to the variable \`datarootdir'
-which seems to be undefined.  Please make sure it is defined" >&2;}
+which seems to be undefined.  Please make sure it is defined." >&2;}
 
-  rm -f "$ac_tmp/stdin"
+  rm -f "$tmp/stdin"
   case $ac_file in
-  -) cat "$ac_tmp/out" && rm -f "$ac_tmp/out";;
-  *) rm -f "$ac_file" && mv "$ac_tmp/out" "$ac_file";;
+  -) cat "$tmp/out" && rm -f "$tmp/out";;
+  *) rm -f "$ac_file" && mv "$tmp/out" "$ac_file";;
   esac \
-  || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+  || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; }
  ;;
   :H)
   #
@@ -6455,21 +7289,27 @@ which seems to be undefined.  Please make sure it is defined" >&2;}
   if test x"$ac_file" != x-; then
     {
       $as_echo "/* $configure_input  */" \
-      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs"
-    } >"$ac_tmp/config.h" \
-      || as_fn_error $? "could not create $ac_file" "$LINENO" 5
-    if diff "$ac_file" "$ac_tmp/config.h" >/dev/null 2>&1; then
-      { $as_echo "$as_me:${as_lineno-$LINENO}: $ac_file is unchanged" >&5
+      && eval '$AWK -f "$tmp/defines.awk"' "$ac_file_inputs"
+    } >"$tmp/config.h" \
+      || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; }
+    if diff "$ac_file" "$tmp/config.h" >/dev/null 2>&1; then
+      { $as_echo "$as_me:$LINENO: $ac_file is unchanged" >&5
 $as_echo "$as_me: $ac_file is unchanged" >&6;}
     else
       rm -f "$ac_file"
-      mv "$ac_tmp/config.h" "$ac_file" \
-	|| as_fn_error $? "could not create $ac_file" "$LINENO" 5
+      mv "$tmp/config.h" "$ac_file" \
+	|| { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; }
     fi
   else
     $as_echo "/* $configure_input  */" \
-      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs" \
-      || as_fn_error $? "could not create -" "$LINENO" 5
+      && eval '$AWK -f "$tmp/defines.awk"' "$ac_file_inputs" \
+      || { { $as_echo "$as_me:$LINENO: error: could not create -" >&5
+$as_echo "$as_me: error: could not create -" >&2;}
+   { (exit 1); exit 1; }; }
   fi
 # Compute "$ac_file"'s index in $config_headers.
 _am_arg="$ac_file"
@@ -6507,7 +7347,7 @@ $as_echo X"$_am_arg" |
 	  s/.*/./; q'`/stamp-h$_am_stamp_count
  ;;
 
-  :C)  { $as_echo "$as_me:${as_lineno-$LINENO}: executing $ac_file commands" >&5
+  :C)  { $as_echo "$as_me:$LINENO: executing $ac_file commands" >&5
 $as_echo "$as_me: executing $ac_file commands" >&6;}
  ;;
   esac
@@ -6515,7 +7355,7 @@ $as_echo "$as_me: executing $ac_file commands" >&6;}
 
   case $ac_file$ac_mode in
     "depfiles":C) test x"$AMDEP_TRUE" != x"" || {
-  # Older Autoconf quotes --file arguments for eval, but not when files
+  # Autoconf 2.62 quotes --file arguments for eval, but not when files
   # are listed without --file.  Let's play safe and only enable the eval
   # if we detect the quoting.
   case $CONFIG_FILES in
@@ -6528,7 +7368,7 @@ $as_echo "$as_me: executing $ac_file commands" >&6;}
     # Strip MF so we end up with the name of the file.
     mf=`echo "$mf" | sed -e 's/:.*$//'`
     # Check whether this is an Automake generated Makefile or not.
-    # We used to match only the files named 'Makefile.in', but
+    # We used to match only the files named `Makefile.in', but
     # some people rename them; so instead we look at the file content.
     # Grep'ing the first line is not enough: some people post-process
     # each Makefile.in and add a new line on top of each file to say so.
@@ -6562,19 +7402,21 @@ $as_echo X"$mf" |
       continue
     fi
     # Extract the definition of DEPDIR, am__include, and am__quote
-    # from the Makefile without running 'make'.
+    # from the Makefile without running `make'.
     DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
     test -z "$DEPDIR" && continue
     am__include=`sed -n 's/^am__include = //p' < "$mf"`
-    test -z "$am__include" && continue
+    test -z "am__include" && continue
     am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
+    # When using ansi2knr, U may be empty or an underscore; expand it
+    U=`sed -n 's/^U = //p' < "$mf"`
     # Find all dependency output files, they are included files with
     # $(DEPDIR) in their names.  We invoke sed twice because it is the
     # simplest approach to changing $(DEPDIR) to its actual value in the
     # expansion.
     for file in `sed -n "
       s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
-	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
+	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g' -e 's/\$U/'"$U"'/g'`; do
       # Make sure the directory exists.
       test -f "$dirpart/$file" && continue
       fdir=`$as_dirname -- "$file" ||
@@ -6600,7 +7442,47 @@ $as_echo X"$file" |
 	    q
 	  }
 	  s/.*/./; q'`
-      as_dir=$dirpart/$fdir; as_fn_mkdir_p
+      { as_dir=$dirpart/$fdir
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || { $as_mkdir_p && mkdir -p "$as_dir"; } || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || { { $as_echo "$as_me:$LINENO: error: cannot create directory $as_dir" >&5
+$as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
+   { (exit 1); exit 1; }; }; }
       # echo "creating $dirpart/$file"
       echo '# dummy' > "$dirpart/$file"
     done
@@ -6612,12 +7494,15 @@ $as_echo X"$file" |
 done # for ac_tag
 
 
-as_fn_exit 0
+{ (exit 0); exit 0; }
 _ACEOF
+chmod +x $CONFIG_STATUS
 ac_clean_files=$ac_clean_files_save
 
 test $ac_write_fail = 0 ||
-  as_fn_error $? "write failure creating $CONFIG_STATUS" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: write failure creating $CONFIG_STATUS" >&5
+$as_echo "$as_me: error: write failure creating $CONFIG_STATUS" >&2;}
+   { (exit 1); exit 1; }; }
 
 
 # configure is writing to config.log, and then calls config.status.
@@ -6638,10 +7523,10 @@ if test "$no_create" != yes; then
   exec 5>>config.log
   # Use ||, not &&, to avoid exiting from the if with $? = 1, which
   # would make configure fail if this is the last instruction.
-  $ac_cs_success || as_fn_exit 1
+  $ac_cs_success || { (exit 1); exit 1; }
 fi
 if test -n "$ac_unrecognized_opts" && test "$enable_option_checking" != no; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
+  { $as_echo "$as_me:$LINENO: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
 $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2;}
 fi
 
diff --git a/lib/Grid_config.h b/lib/Grid_config.h
index c743bba0..fe3a5e60 100644
--- a/lib/Grid_config.h
+++ b/lib/Grid_config.h
@@ -11,21 +11,21 @@
 /* #undef AVX512 */
 
 /* GRID_COMMS_MPI */
-#define GRID_COMMS_MPI 1
+/* #undef GRID_COMMS_MPI */
 
 /* GRID_COMMS_NONE */
-/* #undef GRID_COMMS_NONE */
+#define GRID_COMMS_NONE 1
 
 /* Define to 1 if you have the declaration of `be64toh', and to 0 if you
    don't. */
-#define HAVE_DECL_BE64TOH 0
+#define HAVE_DECL_BE64TOH 1
 
 /* Define to 1 if you have the declaration of `ntohll', and to 0 if you don't.
    */
-#define HAVE_DECL_NTOHLL 1
+#define HAVE_DECL_NTOHLL 0
 
 /* Define to 1 if you have the <endian.h> header file. */
-/* #undef HAVE_ENDIAN_H */
+#define HAVE_ENDIAN_H 1
 
 /* Define to 1 if you have the `gettimeofday' function. */
 #define HAVE_GETTIMEOFDAY 1
@@ -34,10 +34,10 @@
 #define HAVE_INTTYPES_H 1
 
 /* Define to 1 if you have the <malloc.h> header file. */
-/* #undef HAVE_MALLOC_H */
+#define HAVE_MALLOC_H 1
 
 /* Define to 1 if you have the <malloc/malloc.h> header file. */
-#define HAVE_MALLOC_MALLOC_H 1
+/* #undef HAVE_MALLOC_MALLOC_H */
 
 /* Define to 1 if you have the <memory.h> header file. */
 #define HAVE_MEMORY_H 1
@@ -78,9 +78,6 @@
 /* Define to the one symbol short name of this package. */
 #define PACKAGE_TARNAME "grid"
 
-/* Define to the home page for this package. */
-#define PACKAGE_URL ""
-
 /* Define to the version of this package. */
 #define PACKAGE_VERSION "1.0"
 
diff --git a/lib/Grid_config.h.in b/lib/Grid_config.h.in
index 437a6095..4b094250 100644
--- a/lib/Grid_config.h.in
+++ b/lib/Grid_config.h.in
@@ -77,9 +77,6 @@
 /* Define to the one symbol short name of this package. */
 #undef PACKAGE_TARNAME
 
-/* Define to the home page for this package. */
-#undef PACKAGE_URL
-
 /* Define to the version of this package. */
 #undef PACKAGE_VERSION
 
diff --git a/lib/Grid_init.cc b/lib/Grid_init.cc
index 762f7568..47159d05 100644
--- a/lib/Grid_init.cc
+++ b/lib/Grid_init.cc
@@ -81,7 +81,7 @@ void * Grid_backtrace_buffer[_NBACKTRACE];
 void Grid_sa_signal_handler(int sig,siginfo_t *si,void * ptr)
 {
   printf("Caught signal %d\n",si->si_signo);
-  printf("  mem address %llx\n",(uint64_t)si->si_addr);
+  printf("  mem address %lx\n",(uint64_t)si->si_addr);
   printf("         code %d\n",si->si_code);
 
 #ifdef __X86_64
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index 7ce41e57..ddc73c4a 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -253,7 +253,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     mult(&Uchi(),&Umu._odata[ss](Tm),&chi());
     accumReconTm(result,Uchi);
 
-    out._odata[ss] = result;
+    vstream(out._odata[ss],result);
   }
 
 }
diff --git a/tests/Grid_gamma.cc b/tests/Grid_gamma.cc
index af735856..9cda8855 100644
--- a/tests/Grid_gamma.cc
+++ b/tests/Grid_gamma.cc
@@ -39,16 +39,16 @@ int main (int argc, char ** argv)
   std::cout << " Is pod " << std::is_pod<SpinVector>::value  << std::endl;
   std::cout << " Is pod double   " << std::is_pod<double>::value  << std::endl;
   std::cout << " Is pod ComplexF " << std::is_pod<ComplexF>::value  << std::endl;
-  std::cout << " Is triv double " << std::is_trivially_default_constructible<double>::value  << std::endl;
-  std::cout << " Is triv ComplexF " << std::is_trivially_default_constructible<ComplexF>::value  << std::endl;
+  std::cout << " Is triv double " << std::has_trivial_default_constructor<double>::value  << std::endl;
+  std::cout << " Is triv ComplexF " << std::has_trivial_default_constructor<ComplexF>::value  << std::endl;
   std::cout << " Is pod Scalar<double> " << std::is_pod<iScalar<double> >::value  << std::endl;
   std::cout << " Is pod Scalar<ComplexF> " << std::is_pod<iScalar<ComplexF> >::value  << std::endl;
   std::cout << " Is pod Scalar<vComplexF> " << std::is_pod<iScalar<vComplexF> >::value  << std::endl;
   std::cout << " Is pod Scalar<vComplexD> " << std::is_pod<iScalar<vComplexD> >::value  << std::endl;
   std::cout << " Is pod Scalar<vRealF> " << std::is_pod<iScalar<vRealF> >::value  << std::endl;
   std::cout << " Is pod Scalar<vRealD> " << std::is_pod<iScalar<vRealD> >::value  << std::endl;
-  std::cout << " Is triv Scalar<double> " <<std::is_trivially_default_constructible<iScalar<double> >::value << std::endl;
-  std::cout << " Is triv Scalar<vComplexD> "<<std::is_trivially_default_constructible<iScalar<vComplexD> >::value  << std::endl;
+  std::cout << " Is triv Scalar<double> " <<std::has_trivial_default_constructor<iScalar<double> >::value << std::endl;
+  std::cout << " Is triv Scalar<vComplexD> "<<std::has_trivial_default_constructor<iScalar<vComplexD> >::value  << std::endl;
 
   for(int a=0;a<Ns;a++){
     ident()(a,a) = 1.0;

From 52403d587c7f9e8508c2dc2a0583944f8750ebd3 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 6 May 2015 06:37:21 +0100
Subject: [PATCH 125/429] Wilson perf improvements with Gauge prefetching

---
 Makefile.in                    |  393 ++-
 aclocal.m4                     |  705 +++--
 benchmarks/Grid_wilson.cc      |    2 +-
 configure                      | 5011 +++++++++++++-------------------
 lib/Grid_config.h              |   17 +-
 lib/Grid_config.h.in           |    3 +
 lib/math/Grid_math_tensors.h   |   11 +
 lib/qcd/Grid_qcd_wilson_dop.cc |   27 +-
 lib/simd/Grid_vComplexD.h      |    2 +-
 lib/simd/Grid_vComplexF.h      |    2 +-
 lib/simd/Grid_vInteger.h       |    2 +-
 lib/simd/Grid_vRealD.h         |    2 +-
 lib/simd/Grid_vRealF.h         |    2 +-
 tests/Grid_gamma.cc            |   26 +-
 14 files changed, 2819 insertions(+), 3386 deletions(-)

diff --git a/Makefile.in b/Makefile.in
index f9e18fe4..c9257ab2 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -1,9 +1,8 @@
-# Makefile.in generated by automake 1.11.1 from Makefile.am.
+# Makefile.in generated by automake 1.15 from Makefile.am.
 # @configure_input@
 
-# Copyright (C) 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002,
-# 2003, 2004, 2005, 2006, 2007, 2008, 2009  Free Software Foundation,
-# Inc.
+# Copyright (C) 1994-2014 Free Software Foundation, Inc.
+
 # This Makefile.in is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
@@ -15,6 +14,61 @@
 
 @SET_MAKE@
 VPATH = @srcdir@
+am__is_gnu_make = { \
+  if test -z '$(MAKELEVEL)'; then \
+    false; \
+  elif test -n '$(MAKE_HOST)'; then \
+    true; \
+  elif test -n '$(MAKE_VERSION)' && test -n '$(CURDIR)'; then \
+    true; \
+  else \
+    false; \
+  fi; \
+}
+am__make_running_with_option = \
+  case $${target_option-} in \
+      ?) ;; \
+      *) echo "am__make_running_with_option: internal error: invalid" \
+              "target option '$${target_option-}' specified" >&2; \
+         exit 1;; \
+  esac; \
+  has_opt=no; \
+  sane_makeflags=$$MAKEFLAGS; \
+  if $(am__is_gnu_make); then \
+    sane_makeflags=$$MFLAGS; \
+  else \
+    case $$MAKEFLAGS in \
+      *\\[\ \	]*) \
+        bs=\\; \
+        sane_makeflags=`printf '%s\n' "$$MAKEFLAGS" \
+          | sed "s/$$bs$$bs[$$bs $$bs	]*//g"`;; \
+    esac; \
+  fi; \
+  skip_next=no; \
+  strip_trailopt () \
+  { \
+    flg=`printf '%s\n' "$$flg" | sed "s/$$1.*$$//"`; \
+  }; \
+  for flg in $$sane_makeflags; do \
+    test $$skip_next = yes && { skip_next=no; continue; }; \
+    case $$flg in \
+      *=*|--*) continue;; \
+        -*I) strip_trailopt 'I'; skip_next=yes;; \
+      -*I?*) strip_trailopt 'I';; \
+        -*O) strip_trailopt 'O'; skip_next=yes;; \
+      -*O?*) strip_trailopt 'O';; \
+        -*l) strip_trailopt 'l'; skip_next=yes;; \
+      -*l?*) strip_trailopt 'l';; \
+      -[dEDm]) skip_next=yes;; \
+      -[JT]) skip_next=yes;; \
+    esac; \
+    case $$flg in \
+      *$$target_option*) has_opt=yes; break;; \
+    esac; \
+  done; \
+  test $$has_opt = yes
+am__make_dryrun = (target_option=n; $(am__make_running_with_option))
+am__make_keepgoing = (target_option=k; $(am__make_running_with_option))
 pkgdatadir = $(datadir)/@PACKAGE@
 pkgincludedir = $(includedir)/@PACKAGE@
 pkglibdir = $(libdir)/@PACKAGE@
@@ -32,43 +86,86 @@ NORMAL_UNINSTALL = :
 PRE_UNINSTALL = :
 POST_UNINSTALL = :
 subdir = .
-DIST_COMMON = README $(am__configure_deps) $(srcdir)/Makefile.am \
-	$(srcdir)/Makefile.in $(top_srcdir)/configure AUTHORS COPYING \
-	ChangeLog INSTALL NEWS TODO compile depcomp install-sh missing
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/configure.ac
 am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
 	$(ACLOCAL_M4)
+DIST_COMMON = $(srcdir)/Makefile.am $(top_srcdir)/configure \
+	$(am__configure_deps) $(am__DIST_COMMON)
 am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \
  configure.lineno config.status.lineno
 mkinstalldirs = $(install_sh) -d
 CONFIG_HEADER = $(top_builddir)/lib/Grid_config.h
 CONFIG_CLEAN_FILES =
 CONFIG_CLEAN_VPATH_FILES =
+AM_V_P = $(am__v_P_@AM_V@)
+am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
+am__v_P_0 = false
+am__v_P_1 = :
+AM_V_GEN = $(am__v_GEN_@AM_V@)
+am__v_GEN_ = $(am__v_GEN_@AM_DEFAULT_V@)
+am__v_GEN_0 = @echo "  GEN     " $@;
+am__v_GEN_1 = 
+AM_V_at = $(am__v_at_@AM_V@)
+am__v_at_ = $(am__v_at_@AM_DEFAULT_V@)
+am__v_at_0 = @
+am__v_at_1 = 
 SOURCES =
 DIST_SOURCES =
-RECURSIVE_TARGETS = all-recursive check-recursive dvi-recursive \
-	html-recursive info-recursive install-data-recursive \
-	install-dvi-recursive install-exec-recursive \
-	install-html-recursive install-info-recursive \
-	install-pdf-recursive install-ps-recursive install-recursive \
-	installcheck-recursive installdirs-recursive pdf-recursive \
-	ps-recursive uninstall-recursive
+RECURSIVE_TARGETS = all-recursive check-recursive cscopelist-recursive \
+	ctags-recursive dvi-recursive html-recursive info-recursive \
+	install-data-recursive install-dvi-recursive \
+	install-exec-recursive install-html-recursive \
+	install-info-recursive install-pdf-recursive \
+	install-ps-recursive install-recursive installcheck-recursive \
+	installdirs-recursive pdf-recursive ps-recursive \
+	tags-recursive uninstall-recursive
+am__can_run_installinfo = \
+  case $$AM_UPDATE_INFO_DIR in \
+    n|no|NO) false;; \
+    *) (install-info --version) >/dev/null 2>&1;; \
+  esac
 RECURSIVE_CLEAN_TARGETS = mostlyclean-recursive clean-recursive	\
   distclean-recursive maintainer-clean-recursive
-AM_RECURSIVE_TARGETS = $(RECURSIVE_TARGETS:-recursive=) \
-	$(RECURSIVE_CLEAN_TARGETS:-recursive=) tags TAGS ctags CTAGS \
-	distdir dist dist-all distcheck
+am__recursive_targets = \
+  $(RECURSIVE_TARGETS) \
+  $(RECURSIVE_CLEAN_TARGETS) \
+  $(am__extra_recursive_targets)
+AM_RECURSIVE_TARGETS = $(am__recursive_targets:-recursive=) TAGS CTAGS \
+	cscope distdir dist dist-all distcheck
+am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) $(LISP)
+# Read a list of newline-separated strings from the standard input,
+# and print each of them once, without duplicates.  Input order is
+# *not* preserved.
+am__uniquify_input = $(AWK) '\
+  BEGIN { nonempty = 0; } \
+  { items[$$0] = 1; nonempty = 1; } \
+  END { if (nonempty) { for (i in items) print i; }; } \
+'
+# Make sure the list of sources is unique.  This is necessary because,
+# e.g., the same source file might be shared among _SOURCES variables
+# for different programs/libraries.
+am__define_uniq_tagged_files = \
+  list='$(am__tagged_files)'; \
+  unique=`for i in $$list; do \
+    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+  done | $(am__uniquify_input)`
 ETAGS = etags
 CTAGS = ctags
+CSCOPE = cscope
 DIST_SUBDIRS = $(SUBDIRS)
+am__DIST_COMMON = $(srcdir)/Makefile.in AUTHORS COPYING ChangeLog \
+	INSTALL NEWS README TODO compile depcomp install-sh missing
 DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
 distdir = $(PACKAGE)-$(VERSION)
 top_distdir = $(distdir)
 am__remove_distdir = \
-  { test ! -d "$(distdir)" \
-    || { find "$(distdir)" -type d ! -perm -200 -exec chmod u+w {} ';' \
-         && rm -fr "$(distdir)"; }; }
+  if test -d "$(distdir)"; then \
+    find "$(distdir)" -type d ! -perm -200 -exec chmod u+w {} ';' \
+      && rm -rf "$(distdir)" \
+      || { sleep 5 && rm -rf "$(distdir)"; }; \
+  else :; fi
+am__post_remove_distdir = $(am__remove_distdir)
 am__relativize = \
   dir0=`pwd`; \
   sed_first='s,^\([^/]*\)/.*$$,\1,'; \
@@ -96,10 +193,14 @@ am__relativize = \
   reldir="$$dir2"
 DIST_ARCHIVES = $(distdir).tar.gz
 GZIP_ENV = --best
+DIST_TARGETS = dist-gzip
 distuninstallcheck_listfiles = find . -type f -print
+am__distuninstallcheck_listfiles = $(distuninstallcheck_listfiles) \
+  | sed 's|^\./|$(prefix)/|' | grep -v '$(infodir)/dir$$'
 distcleancheck_listfiles = find . -type f -print
 ACLOCAL = @ACLOCAL@
 AMTAR = @AMTAR@
+AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
 AUTOCONF = @AUTOCONF@
 AUTOHEADER = @AUTOHEADER@
 AUTOMAKE = @AUTOMAKE@
@@ -139,6 +240,7 @@ PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
 PACKAGE_NAME = @PACKAGE_NAME@
 PACKAGE_STRING = @PACKAGE_STRING@
 PACKAGE_TARNAME = @PACKAGE_TARNAME@
+PACKAGE_URL = @PACKAGE_URL@
 PACKAGE_VERSION = @PACKAGE_VERSION@
 PATH_SEPARATOR = @PATH_SEPARATOR@
 RANLIB = @RANLIB@
@@ -196,7 +298,7 @@ SUBDIRS = lib tests benchmarks
 all: all-recursive
 
 .SUFFIXES:
-am--refresh:
+am--refresh: Makefile
 	@:
 $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	@for dep in $?; do \
@@ -211,7 +313,6 @@ $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu Makefile'; \
 	$(am__cd) $(top_srcdir) && \
 	  $(AUTOMAKE) --gnu Makefile
-.PRECIOUS: Makefile
 Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
 	@case '$?' in \
 	  *config.status*) \
@@ -232,22 +333,25 @@ $(ACLOCAL_M4):  $(am__aclocal_m4_deps)
 $(am__aclocal_m4_deps):
 
 # This directory's subdirectories are mostly independent; you can cd
-# into them and run `make' without going through this Makefile.
-# To change the values of `make' variables: instead of editing Makefiles,
-# (1) if the variable is set in `config.status', edit `config.status'
-#     (which will cause the Makefiles to be regenerated when you run `make');
-# (2) otherwise, pass the desired values on the `make' command line.
-$(RECURSIVE_TARGETS):
-	@fail= failcom='exit 1'; \
-	for f in x $$MAKEFLAGS; do \
-	  case $$f in \
-	    *=* | --[!k]*);; \
-	    *k*) failcom='fail=yes';; \
-	  esac; \
-	done; \
+# into them and run 'make' without going through this Makefile.
+# To change the values of 'make' variables: instead of editing Makefiles,
+# (1) if the variable is set in 'config.status', edit 'config.status'
+#     (which will cause the Makefiles to be regenerated when you run 'make');
+# (2) otherwise, pass the desired values on the 'make' command line.
+$(am__recursive_targets):
+	@fail=; \
+	if $(am__make_keepgoing); then \
+	  failcom='fail=yes'; \
+	else \
+	  failcom='exit 1'; \
+	fi; \
 	dot_seen=no; \
 	target=`echo $@ | sed s/-recursive//`; \
-	list='$(SUBDIRS)'; for subdir in $$list; do \
+	case "$@" in \
+	  distclean-* | maintainer-clean-*) list='$(DIST_SUBDIRS)' ;; \
+	  *) list='$(SUBDIRS)' ;; \
+	esac; \
+	for subdir in $$list; do \
 	  echo "Making $$target in $$subdir"; \
 	  if test "$$subdir" = "."; then \
 	    dot_seen=yes; \
@@ -262,57 +366,12 @@ $(RECURSIVE_TARGETS):
 	  $(MAKE) $(AM_MAKEFLAGS) "$$target-am" || exit 1; \
 	fi; test -z "$$fail"
 
-$(RECURSIVE_CLEAN_TARGETS):
-	@fail= failcom='exit 1'; \
-	for f in x $$MAKEFLAGS; do \
-	  case $$f in \
-	    *=* | --[!k]*);; \
-	    *k*) failcom='fail=yes';; \
-	  esac; \
-	done; \
-	dot_seen=no; \
-	case "$@" in \
-	  distclean-* | maintainer-clean-*) list='$(DIST_SUBDIRS)' ;; \
-	  *) list='$(SUBDIRS)' ;; \
-	esac; \
-	rev=''; for subdir in $$list; do \
-	  if test "$$subdir" = "."; then :; else \
-	    rev="$$subdir $$rev"; \
-	  fi; \
-	done; \
-	rev="$$rev ."; \
-	target=`echo $@ | sed s/-recursive//`; \
-	for subdir in $$rev; do \
-	  echo "Making $$target in $$subdir"; \
-	  if test "$$subdir" = "."; then \
-	    local_target="$$target-am"; \
-	  else \
-	    local_target="$$target"; \
-	  fi; \
-	  ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) $$local_target) \
-	  || eval $$failcom; \
-	done && test -z "$$fail"
-tags-recursive:
-	list='$(SUBDIRS)'; for subdir in $$list; do \
-	  test "$$subdir" = . || ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) tags); \
-	done
-ctags-recursive:
-	list='$(SUBDIRS)'; for subdir in $$list; do \
-	  test "$$subdir" = . || ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) ctags); \
-	done
+ID: $(am__tagged_files)
+	$(am__define_uniq_tagged_files); mkid -fID $$unique
+tags: tags-recursive
+TAGS: tags
 
-ID: $(HEADERS) $(SOURCES) $(LISP) $(TAGS_FILES)
-	list='$(SOURCES) $(HEADERS) $(LISP) $(TAGS_FILES)'; \
-	unique=`for i in $$list; do \
-	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
-	  done | \
-	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
-	      END { if (nonempty) { for (i in files) print i; }; }'`; \
-	mkid -fID $$unique
-tags: TAGS
-
-TAGS: tags-recursive $(HEADERS) $(SOURCES)  $(TAGS_DEPENDENCIES) \
-		$(TAGS_FILES) $(LISP)
+tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
 	set x; \
 	here=`pwd`; \
 	if ($(ETAGS) --etags-include --version) >/dev/null 2>&1; then \
@@ -328,12 +387,7 @@ TAGS: tags-recursive $(HEADERS) $(SOURCES)  $(TAGS_DEPENDENCIES) \
 	      set "$$@" "$$include_option=$$here/$$subdir/TAGS"; \
 	  fi; \
 	done; \
-	list='$(SOURCES) $(HEADERS)  $(LISP) $(TAGS_FILES)'; \
-	unique=`for i in $$list; do \
-	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
-	  done | \
-	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
-	      END { if (nonempty) { for (i in files) print i; }; }'`; \
+	$(am__define_uniq_tagged_files); \
 	shift; \
 	if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
 	  test -n "$$unique" || unique=$$empty_fix; \
@@ -345,15 +399,11 @@ TAGS: tags-recursive $(HEADERS) $(SOURCES)  $(TAGS_DEPENDENCIES) \
 	      $$unique; \
 	  fi; \
 	fi
-ctags: CTAGS
-CTAGS: ctags-recursive $(HEADERS) $(SOURCES)  $(TAGS_DEPENDENCIES) \
-		$(TAGS_FILES) $(LISP)
-	list='$(SOURCES) $(HEADERS)  $(LISP) $(TAGS_FILES)'; \
-	unique=`for i in $$list; do \
-	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
-	  done | \
-	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
-	      END { if (nonempty) { for (i in files) print i; }; }'`; \
+ctags: ctags-recursive
+
+CTAGS: ctags
+ctags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
+	$(am__define_uniq_tagged_files); \
 	test -z "$(CTAGS_ARGS)$$unique" \
 	  || $(CTAGS) $(CTAGSFLAGS) $(AM_CTAGSFLAGS) $(CTAGS_ARGS) \
 	     $$unique
@@ -362,9 +412,31 @@ GTAGS:
 	here=`$(am__cd) $(top_builddir) && pwd` \
 	  && $(am__cd) $(top_srcdir) \
 	  && gtags -i $(GTAGS_ARGS) "$$here"
+cscope: cscope.files
+	test ! -s cscope.files \
+	  || $(CSCOPE) -b -q $(AM_CSCOPEFLAGS) $(CSCOPEFLAGS) -i cscope.files $(CSCOPE_ARGS)
+clean-cscope:
+	-rm -f cscope.files
+cscope.files: clean-cscope cscopelist
+cscopelist: cscopelist-recursive
+
+cscopelist-am: $(am__tagged_files)
+	list='$(am__tagged_files)'; \
+	case "$(srcdir)" in \
+	  [\\/]* | ?:[\\/]*) sdir="$(srcdir)" ;; \
+	  *) sdir=$(subdir)/$(srcdir) ;; \
+	esac; \
+	for i in $$list; do \
+	  if test -f "$$i"; then \
+	    echo "$(subdir)/$$i"; \
+	  else \
+	    echo "$$sdir/$$i"; \
+	  fi; \
+	done >> $(top_builddir)/cscope.files
 
 distclean-tags:
 	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
+	-rm -f cscope.out cscope.in.out cscope.po.out cscope.files
 
 distdir: $(DISTFILES)
 	$(am__remove_distdir)
@@ -400,13 +472,10 @@ distdir: $(DISTFILES)
 	done
 	@list='$(DIST_SUBDIRS)'; for subdir in $$list; do \
 	  if test "$$subdir" = .; then :; else \
-	    test -d "$(distdir)/$$subdir" \
-	    || $(MKDIR_P) "$(distdir)/$$subdir" \
-	    || exit 1; \
-	  fi; \
-	done
-	@list='$(DIST_SUBDIRS)'; for subdir in $$list; do \
-	  if test "$$subdir" = .; then :; else \
+	    $(am__make_dryrun) \
+	      || test -d "$(distdir)/$$subdir" \
+	      || $(MKDIR_P) "$(distdir)/$$subdir" \
+	      || exit 1; \
 	    dir1=$$subdir; dir2="$(distdir)/$$subdir"; \
 	    $(am__relativize); \
 	    new_distdir=$$reldir; \
@@ -435,36 +504,42 @@ distdir: $(DISTFILES)
 	|| chmod -R a+r "$(distdir)"
 dist-gzip: distdir
 	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
-	$(am__remove_distdir)
+	$(am__post_remove_distdir)
 
 dist-bzip2: distdir
-	tardir=$(distdir) && $(am__tar) | bzip2 -9 -c >$(distdir).tar.bz2
-	$(am__remove_distdir)
+	tardir=$(distdir) && $(am__tar) | BZIP2=$${BZIP2--9} bzip2 -c >$(distdir).tar.bz2
+	$(am__post_remove_distdir)
 
-dist-lzma: distdir
-	tardir=$(distdir) && $(am__tar) | lzma -9 -c >$(distdir).tar.lzma
-	$(am__remove_distdir)
+dist-lzip: distdir
+	tardir=$(distdir) && $(am__tar) | lzip -c $${LZIP_OPT--9} >$(distdir).tar.lz
+	$(am__post_remove_distdir)
 
 dist-xz: distdir
-	tardir=$(distdir) && $(am__tar) | xz -c >$(distdir).tar.xz
-	$(am__remove_distdir)
+	tardir=$(distdir) && $(am__tar) | XZ_OPT=$${XZ_OPT--e} xz -c >$(distdir).tar.xz
+	$(am__post_remove_distdir)
 
 dist-tarZ: distdir
+	@echo WARNING: "Support for distribution archives compressed with" \
+		       "legacy program 'compress' is deprecated." >&2
+	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	tardir=$(distdir) && $(am__tar) | compress -c >$(distdir).tar.Z
-	$(am__remove_distdir)
+	$(am__post_remove_distdir)
 
 dist-shar: distdir
+	@echo WARNING: "Support for shar distribution archives is" \
+	               "deprecated." >&2
+	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	shar $(distdir) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).shar.gz
-	$(am__remove_distdir)
+	$(am__post_remove_distdir)
 
 dist-zip: distdir
 	-rm -f $(distdir).zip
 	zip -rq $(distdir).zip $(distdir)
-	$(am__remove_distdir)
+	$(am__post_remove_distdir)
 
-dist dist-all: distdir
-	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
-	$(am__remove_distdir)
+dist dist-all:
+	$(MAKE) $(AM_MAKEFLAGS) $(DIST_TARGETS) am__post_remove_distdir='@:'
+	$(am__post_remove_distdir)
 
 # This target untars the dist file and tries a VPATH configuration.  Then
 # it guarantees that the distribution is self-contained by making another
@@ -475,8 +550,8 @@ distcheck: dist
 	  GZIP=$(GZIP_ENV) gzip -dc $(distdir).tar.gz | $(am__untar) ;;\
 	*.tar.bz2*) \
 	  bzip2 -dc $(distdir).tar.bz2 | $(am__untar) ;;\
-	*.tar.lzma*) \
-	  lzma -dc $(distdir).tar.lzma | $(am__untar) ;;\
+	*.tar.lz*) \
+	  lzip -dc $(distdir).tar.lz | $(am__untar) ;;\
 	*.tar.xz*) \
 	  xz -dc $(distdir).tar.xz | $(am__untar) ;;\
 	*.tar.Z*) \
@@ -486,17 +561,19 @@ distcheck: dist
 	*.zip*) \
 	  unzip $(distdir).zip ;;\
 	esac
-	chmod -R a-w $(distdir); chmod u+w $(distdir)
-	mkdir $(distdir)/_build
-	mkdir $(distdir)/_inst
+	chmod -R a-w $(distdir)
+	chmod u+w $(distdir)
+	mkdir $(distdir)/_build $(distdir)/_build/sub $(distdir)/_inst
 	chmod a-w $(distdir)
 	test -d $(distdir)/_build || exit 0; \
 	dc_install_base=`$(am__cd) $(distdir)/_inst && pwd | sed -e 's,^[^:\\/]:[\\/],/,'` \
 	  && dc_destdir="$${TMPDIR-/tmp}/am-dc-$$$$/" \
 	  && am__cwd=`pwd` \
-	  && $(am__cd) $(distdir)/_build \
-	  && ../configure --srcdir=.. --prefix="$$dc_install_base" \
+	  && $(am__cd) $(distdir)/_build/sub \
+	  && ../../configure \
+	    $(AM_DISTCHECK_CONFIGURE_FLAGS) \
 	    $(DISTCHECK_CONFIGURE_FLAGS) \
+	    --srcdir=../.. --prefix="$$dc_install_base" \
 	  && $(MAKE) $(AM_MAKEFLAGS) \
 	  && $(MAKE) $(AM_MAKEFLAGS) dvi \
 	  && $(MAKE) $(AM_MAKEFLAGS) check \
@@ -519,13 +596,21 @@ distcheck: dist
 	  && $(MAKE) $(AM_MAKEFLAGS) distcleancheck \
 	  && cd "$$am__cwd" \
 	  || exit 1
-	$(am__remove_distdir)
+	$(am__post_remove_distdir)
 	@(echo "$(distdir) archives ready for distribution: "; \
 	  list='$(DIST_ARCHIVES)'; for i in $$list; do echo $$i; done) | \
 	  sed -e 1h -e 1s/./=/g -e 1p -e 1x -e '$$p' -e '$$x'
 distuninstallcheck:
-	@$(am__cd) '$(distuninstallcheck_dir)' \
-	&& test `$(distuninstallcheck_listfiles) | wc -l` -le 1 \
+	@test -n '$(distuninstallcheck_dir)' || { \
+	  echo 'ERROR: trying to run $@ with an empty' \
+	       '$$(distuninstallcheck_dir)' >&2; \
+	  exit 1; \
+	}; \
+	$(am__cd) '$(distuninstallcheck_dir)' || { \
+	  echo 'ERROR: cannot chdir into $(distuninstallcheck_dir)' >&2; \
+	  exit 1; \
+	}; \
+	test `$(am__distuninstallcheck_listfiles) | wc -l` -eq 0 \
 	   || { echo "ERROR: files left after uninstall:" ; \
 	        if test -n "$(DESTDIR)"; then \
 	          echo "  (check DESTDIR support)"; \
@@ -556,10 +641,15 @@ install-am: all-am
 
 installcheck: installcheck-recursive
 install-strip:
-	$(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
-	  install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
-	  `test -z '$(STRIP)' || \
-	    echo "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'"` install
+	if test -z '$(STRIP)'; then \
+	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
+	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
+	      install; \
+	else \
+	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
+	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
+	    "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'" install; \
+	fi
 mostlyclean-generic:
 
 clean-generic:
@@ -640,23 +730,24 @@ ps-am:
 
 uninstall-am:
 
-.MAKE: $(RECURSIVE_CLEAN_TARGETS) $(RECURSIVE_TARGETS) ctags-recursive \
-	install-am install-strip tags-recursive
+.MAKE: $(am__recursive_targets) install-am install-strip
 
-.PHONY: $(RECURSIVE_CLEAN_TARGETS) $(RECURSIVE_TARGETS) CTAGS GTAGS \
-	all all-am am--refresh check check-am clean clean-generic \
-	ctags ctags-recursive dist dist-all dist-bzip2 dist-gzip \
-	dist-lzma dist-shar dist-tarZ dist-xz dist-zip distcheck \
-	distclean distclean-generic distclean-tags distcleancheck \
-	distdir distuninstallcheck dvi dvi-am html html-am info \
-	info-am install install-am install-data install-data-am \
-	install-dvi install-dvi-am install-exec install-exec-am \
-	install-html install-html-am install-info install-info-am \
-	install-man install-pdf install-pdf-am install-ps \
-	install-ps-am install-strip installcheck installcheck-am \
-	installdirs installdirs-am maintainer-clean \
+.PHONY: $(am__recursive_targets) CTAGS GTAGS TAGS all all-am \
+	am--refresh check check-am clean clean-cscope clean-generic \
+	cscope cscopelist-am ctags ctags-am dist dist-all dist-bzip2 \
+	dist-gzip dist-lzip dist-shar dist-tarZ dist-xz dist-zip \
+	distcheck distclean distclean-generic distclean-tags \
+	distcleancheck distdir distuninstallcheck dvi dvi-am html \
+	html-am info info-am install install-am install-data \
+	install-data-am install-dvi install-dvi-am install-exec \
+	install-exec-am install-html install-html-am install-info \
+	install-info-am install-man install-pdf install-pdf-am \
+	install-ps install-ps-am install-strip installcheck \
+	installcheck-am installdirs installdirs-am maintainer-clean \
 	maintainer-clean-generic mostlyclean mostlyclean-generic pdf \
-	pdf-am ps ps-am tags tags-recursive uninstall uninstall-am
+	pdf-am ps ps-am tags tags-am uninstall uninstall-am
+
+.PRECIOUS: Makefile
 
 
 # Tell versions [3.59,3.63) of GNU make to not export all variables.
diff --git a/aclocal.m4 b/aclocal.m4
index bcc213b4..f3018f6c 100644
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -1,7 +1,7 @@
-# generated automatically by aclocal 1.11.1 -*- Autoconf -*-
+# generated automatically by aclocal 1.15 -*- Autoconf -*-
+
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 
-# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
-# 2005, 2006, 2007, 2008, 2009  Free Software Foundation, Inc.
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
@@ -11,15 +11,16 @@
 # even the implied warranty of MERCHANTABILITY or FITNESS FOR A
 # PARTICULAR PURPOSE.
 
+m4_ifndef([AC_CONFIG_MACRO_DIRS], [m4_defun([_AM_CONFIG_MACRO_DIRS], [])m4_defun([AC_CONFIG_MACRO_DIRS], [_AM_CONFIG_MACRO_DIRS($@)])])
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
-m4_if(m4_defn([AC_AUTOCONF_VERSION]), [2.63],,
-[m4_warning([this file was generated for autoconf 2.63.
+m4_if(m4_defn([AC_AUTOCONF_VERSION]), [2.69],,
+[m4_warning([this file was generated for autoconf 2.69.
 You have another version of autoconf.  It may work, but is not guaranteed to.
 If you have problems, you may need to regenerate the build system entirely.
-To do so, use the procedure documented by the package, typically `autoreconf'.])])
+To do so, use the procedure documented by the package, typically 'autoreconf'.])])
 
-# Copyright (C) 2002, 2003, 2005, 2006, 2007, 2008  Free Software Foundation, Inc.
+# Copyright (C) 2002-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -31,10 +32,10 @@ To do so, use the procedure documented by the package, typically `autoreconf'.])
 # generated from the m4 files accompanying Automake X.Y.
 # (This private macro should not be called outside this file.)
 AC_DEFUN([AM_AUTOMAKE_VERSION],
-[am__api_version='1.11'
+[am__api_version='1.15'
 dnl Some users find AM_AUTOMAKE_VERSION and mistake it for a way to
 dnl require some minimum version.  Point them to the right macro.
-m4_if([$1], [1.11.1], [],
+m4_if([$1], [1.15], [],
       [AC_FATAL([Do not call $0, use AM_INIT_AUTOMAKE([$1]).])])dnl
 ])
 
@@ -50,22 +51,22 @@ m4_define([_AM_AUTOCONF_VERSION], [])
 # Call AM_AUTOMAKE_VERSION and AM_AUTOMAKE_VERSION so they can be traced.
 # This function is AC_REQUIREd by AM_INIT_AUTOMAKE.
 AC_DEFUN([AM_SET_CURRENT_AUTOMAKE_VERSION],
-[AM_AUTOMAKE_VERSION([1.11.1])dnl
+[AM_AUTOMAKE_VERSION([1.15])dnl
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
 _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 
 # AM_AUX_DIR_EXPAND                                         -*- Autoconf -*-
 
-# Copyright (C) 2001, 2003, 2005  Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
 # For projects using AC_CONFIG_AUX_DIR([foo]), Autoconf sets
-# $ac_aux_dir to `$srcdir/foo'.  In other projects, it is set to
-# `$srcdir', `$srcdir/..', or `$srcdir/../..'.
+# $ac_aux_dir to '$srcdir/foo'.  In other projects, it is set to
+# '$srcdir', '$srcdir/..', or '$srcdir/../..'.
 #
 # Of course, Automake must honor this variable whenever it calls a
 # tool from the auxiliary directory.  The problem is that $srcdir (and
@@ -84,7 +85,7 @@ _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 #
 # The reason of the latter failure is that $top_srcdir and $ac_aux_dir
 # are both prefixed by $srcdir.  In an in-source build this is usually
-# harmless because $srcdir is `.', but things will broke when you
+# harmless because $srcdir is '.', but things will broke when you
 # start a VPATH build or use an absolute $srcdir.
 #
 # So we could use something similar to $top_srcdir/$ac_aux_dir/missing,
@@ -102,30 +103,26 @@ _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 # configured tree to be moved without reconfiguration.
 
 AC_DEFUN([AM_AUX_DIR_EXPAND],
-[dnl Rely on autoconf to set up CDPATH properly.
-AC_PREREQ([2.50])dnl
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+[AC_REQUIRE([AC_CONFIG_AUX_DIR_DEFAULT])dnl
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 ])
 
 # AM_CONDITIONAL                                            -*- Autoconf -*-
 
-# Copyright (C) 1997, 2000, 2001, 2003, 2004, 2005, 2006, 2008
-# Free Software Foundation, Inc.
+# Copyright (C) 1997-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 9
-
 # AM_CONDITIONAL(NAME, SHELL-CONDITION)
 # -------------------------------------
 # Define a conditional.
 AC_DEFUN([AM_CONDITIONAL],
-[AC_PREREQ(2.52)dnl
- ifelse([$1], [TRUE],  [AC_FATAL([$0: invalid condition: $1])],
-	[$1], [FALSE], [AC_FATAL([$0: invalid condition: $1])])dnl
+[AC_PREREQ([2.52])dnl
+ m4_if([$1], [TRUE],  [AC_FATAL([$0: invalid condition: $1])],
+       [$1], [FALSE], [AC_FATAL([$0: invalid condition: $1])])dnl
 AC_SUBST([$1_TRUE])dnl
 AC_SUBST([$1_FALSE])dnl
 _AM_SUBST_NOTMAKE([$1_TRUE])dnl
@@ -144,16 +141,14 @@ AC_CONFIG_COMMANDS_PRE(
 Usually this means the macro was only invoked conditionally.]])
 fi])])
 
-# Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2009
-# Free Software Foundation, Inc.
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 10
 
-# There are a few dirty hacks below to avoid letting `AC_PROG_CC' be
+# There are a few dirty hacks below to avoid letting 'AC_PROG_CC' be
 # written in clear, in which case automake, when reading aclocal.m4,
 # will think it sees a *use*, and therefore will trigger all it's
 # C support machinery.  Also note that it means that autoscan, seeing
@@ -163,7 +158,7 @@ fi])])
 # _AM_DEPENDENCIES(NAME)
 # ----------------------
 # See how the compiler implements dependency checking.
-# NAME is "CC", "CXX", "GCJ", or "OBJC".
+# NAME is "CC", "CXX", "OBJC", "OBJCXX", "UPC", or "GJC".
 # We try a few techniques and use that to set a single cache variable.
 #
 # We don't AC_REQUIRE the corresponding AC_PROG_CC since the latter was
@@ -176,12 +171,13 @@ AC_REQUIRE([AM_OUTPUT_DEPENDENCY_COMMANDS])dnl
 AC_REQUIRE([AM_MAKE_INCLUDE])dnl
 AC_REQUIRE([AM_DEP_TRACK])dnl
 
-ifelse([$1], CC,   [depcc="$CC"   am_compiler_list=],
-       [$1], CXX,  [depcc="$CXX"  am_compiler_list=],
-       [$1], OBJC, [depcc="$OBJC" am_compiler_list='gcc3 gcc'],
-       [$1], UPC,  [depcc="$UPC"  am_compiler_list=],
-       [$1], GCJ,  [depcc="$GCJ"  am_compiler_list='gcc3 gcc'],
-                   [depcc="$$1"   am_compiler_list=])
+m4_if([$1], [CC],   [depcc="$CC"   am_compiler_list=],
+      [$1], [CXX],  [depcc="$CXX"  am_compiler_list=],
+      [$1], [OBJC], [depcc="$OBJC" am_compiler_list='gcc3 gcc'],
+      [$1], [OBJCXX], [depcc="$OBJCXX" am_compiler_list='gcc3 gcc'],
+      [$1], [UPC],  [depcc="$UPC"  am_compiler_list=],
+      [$1], [GCJ],  [depcc="$GCJ"  am_compiler_list='gcc3 gcc'],
+                    [depcc="$$1"   am_compiler_list=])
 
 AC_CACHE_CHECK([dependency style of $depcc],
                [am_cv_$1_dependencies_compiler_type],
@@ -189,8 +185,9 @@ AC_CACHE_CHECK([dependency style of $depcc],
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named `D' -- because `-MD' means `put the output
-  # in D'.
+  # making a dummy file named 'D' -- because '-MD' means "put the output
+  # in D".
+  rm -rf conftest.dir
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -229,16 +226,16 @@ AC_CACHE_CHECK([dependency style of $depcc],
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
-      # Solaris 8's {/usr,}/bin/sh.
-      touch sub/conftst$i.h
+      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
+      # Solaris 10 /bin/sh.
+      echo '/* dummy */' > sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with `-c' and `-o' for the sake of the "dashmstdout"
+    # We check with '-c' and '-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle `-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs
+    # handle '-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs.
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -247,16 +244,16 @@ AC_CACHE_CHECK([dependency style of $depcc],
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # after this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested
+      # After this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested.
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvisualcpp | msvcmsys)
-      # This compiler won't grok `-c -o', but also, the minuso test has
+    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
+      # This compiler won't grok '-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -304,7 +301,7 @@ AM_CONDITIONAL([am__fastdep$1], [
 # AM_SET_DEPDIR
 # -------------
 # Choose a directory name for dependency files.
-# This macro is AC_REQUIREd in _AM_DEPENDENCIES
+# This macro is AC_REQUIREd in _AM_DEPENDENCIES.
 AC_DEFUN([AM_SET_DEPDIR],
 [AC_REQUIRE([AM_SET_LEADING_DOT])dnl
 AC_SUBST([DEPDIR], ["${am__leading_dot}deps"])dnl
@@ -314,34 +311,39 @@ AC_SUBST([DEPDIR], ["${am__leading_dot}deps"])dnl
 # AM_DEP_TRACK
 # ------------
 AC_DEFUN([AM_DEP_TRACK],
-[AC_ARG_ENABLE(dependency-tracking,
-[  --disable-dependency-tracking  speeds up one-time build
-  --enable-dependency-tracking   do not reject slow dependency extractors])
+[AC_ARG_ENABLE([dependency-tracking], [dnl
+AS_HELP_STRING(
+  [--enable-dependency-tracking],
+  [do not reject slow dependency extractors])
+AS_HELP_STRING(
+  [--disable-dependency-tracking],
+  [speeds up one-time build])])
 if test "x$enable_dependency_tracking" != xno; then
   am_depcomp="$ac_aux_dir/depcomp"
   AMDEPBACKSLASH='\'
+  am__nodep='_no'
 fi
 AM_CONDITIONAL([AMDEP], [test "x$enable_dependency_tracking" != xno])
 AC_SUBST([AMDEPBACKSLASH])dnl
 _AM_SUBST_NOTMAKE([AMDEPBACKSLASH])dnl
+AC_SUBST([am__nodep])dnl
+_AM_SUBST_NOTMAKE([am__nodep])dnl
 ])
 
 # Generate code to set up dependency tracking.              -*- Autoconf -*-
 
-# Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2008
-# Free Software Foundation, Inc.
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-#serial 5
 
 # _AM_OUTPUT_DEPENDENCY_COMMANDS
 # ------------------------------
 AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
 [{
-  # Autoconf 2.62 quotes --file arguments for eval, but not when files
+  # Older Autoconf quotes --file arguments for eval, but not when files
   # are listed without --file.  Let's play safe and only enable the eval
   # if we detect the quoting.
   case $CONFIG_FILES in
@@ -354,7 +356,7 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
     # Strip MF so we end up with the name of the file.
     mf=`echo "$mf" | sed -e 's/:.*$//'`
     # Check whether this is an Automake generated Makefile or not.
-    # We used to match only the files named `Makefile.in', but
+    # We used to match only the files named 'Makefile.in', but
     # some people rename them; so instead we look at the file content.
     # Grep'ing the first line is not enough: some people post-process
     # each Makefile.in and add a new line on top of each file to say so.
@@ -366,21 +368,19 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
       continue
     fi
     # Extract the definition of DEPDIR, am__include, and am__quote
-    # from the Makefile without running `make'.
+    # from the Makefile without running 'make'.
     DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
     test -z "$DEPDIR" && continue
     am__include=`sed -n 's/^am__include = //p' < "$mf"`
-    test -z "am__include" && continue
+    test -z "$am__include" && continue
     am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
-    # When using ansi2knr, U may be empty or an underscore; expand it
-    U=`sed -n 's/^U = //p' < "$mf"`
     # Find all dependency output files, they are included files with
     # $(DEPDIR) in their names.  We invoke sed twice because it is the
     # simplest approach to changing $(DEPDIR) to its actual value in the
     # expansion.
     for file in `sed -n "
       s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
-	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g' -e 's/\$U/'"$U"'/g'`; do
+	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
       # Make sure the directory exists.
       test -f "$dirpart/$file" && continue
       fdir=`AS_DIRNAME(["$file"])`
@@ -398,7 +398,7 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
 # This macro should only be invoked once -- use via AC_REQUIRE.
 #
 # This code is only required when automatic dependency tracking
-# is enabled.  FIXME.  This creates each `.P' file that we will
+# is enabled.  FIXME.  This creates each '.P' file that we will
 # need in order to bootstrap the dependency handling code.
 AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 [AC_CONFIG_COMMANDS([depfiles],
@@ -408,18 +408,21 @@ AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 
 # Do all the work for Automake.                             -*- Autoconf -*-
 
-# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
-# 2005, 2006, 2008, 2009 Free Software Foundation, Inc.
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 16
-
 # This macro actually does too much.  Some checks are only needed if
 # your package does certain things.  But this isn't really a big deal.
 
+dnl Redefine AC_PROG_CC to automatically invoke _AM_PROG_CC_C_O.
+m4_define([AC_PROG_CC],
+m4_defn([AC_PROG_CC])
+[_AM_PROG_CC_C_O
+])
+
 # AM_INIT_AUTOMAKE(PACKAGE, VERSION, [NO-DEFINE])
 # AM_INIT_AUTOMAKE([OPTIONS])
 # -----------------------------------------------
@@ -432,7 +435,7 @@ AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 # arguments mandatory, and then we can depend on a new Autoconf
 # release and drop the old call support.
 AC_DEFUN([AM_INIT_AUTOMAKE],
-[AC_PREREQ([2.62])dnl
+[AC_PREREQ([2.65])dnl
 dnl Autoconf wants to disallow AM_ names.  We explicitly allow
 dnl the ones we care about.
 m4_pattern_allow([^AM_[A-Z]+FLAGS$])dnl
@@ -461,33 +464,42 @@ AC_SUBST([CYGPATH_W])
 # Define the identity of the package.
 dnl Distinguish between old-style and new-style calls.
 m4_ifval([$2],
-[m4_ifval([$3], [_AM_SET_OPTION([no-define])])dnl
+[AC_DIAGNOSE([obsolete],
+             [$0: two- and three-arguments forms are deprecated.])
+m4_ifval([$3], [_AM_SET_OPTION([no-define])])dnl
  AC_SUBST([PACKAGE], [$1])dnl
  AC_SUBST([VERSION], [$2])],
 [_AM_SET_OPTIONS([$1])dnl
 dnl Diagnose old-style AC_INIT with new-style AM_AUTOMAKE_INIT.
-m4_if(m4_ifdef([AC_PACKAGE_NAME], 1)m4_ifdef([AC_PACKAGE_VERSION], 1), 11,,
+m4_if(
+  m4_ifdef([AC_PACKAGE_NAME], [ok]):m4_ifdef([AC_PACKAGE_VERSION], [ok]),
+  [ok:ok],,
   [m4_fatal([AC_INIT should be called with package and version arguments])])dnl
  AC_SUBST([PACKAGE], ['AC_PACKAGE_TARNAME'])dnl
  AC_SUBST([VERSION], ['AC_PACKAGE_VERSION'])])dnl
 
 _AM_IF_OPTION([no-define],,
-[AC_DEFINE_UNQUOTED(PACKAGE, "$PACKAGE", [Name of package])
- AC_DEFINE_UNQUOTED(VERSION, "$VERSION", [Version number of package])])dnl
+[AC_DEFINE_UNQUOTED([PACKAGE], ["$PACKAGE"], [Name of package])
+ AC_DEFINE_UNQUOTED([VERSION], ["$VERSION"], [Version number of package])])dnl
 
 # Some tools Automake needs.
 AC_REQUIRE([AM_SANITY_CHECK])dnl
 AC_REQUIRE([AC_ARG_PROGRAM])dnl
-AM_MISSING_PROG(ACLOCAL, aclocal-${am__api_version})
-AM_MISSING_PROG(AUTOCONF, autoconf)
-AM_MISSING_PROG(AUTOMAKE, automake-${am__api_version})
-AM_MISSING_PROG(AUTOHEADER, autoheader)
-AM_MISSING_PROG(MAKEINFO, makeinfo)
+AM_MISSING_PROG([ACLOCAL], [aclocal-${am__api_version}])
+AM_MISSING_PROG([AUTOCONF], [autoconf])
+AM_MISSING_PROG([AUTOMAKE], [automake-${am__api_version}])
+AM_MISSING_PROG([AUTOHEADER], [autoheader])
+AM_MISSING_PROG([MAKEINFO], [makeinfo])
 AC_REQUIRE([AM_PROG_INSTALL_SH])dnl
 AC_REQUIRE([AM_PROG_INSTALL_STRIP])dnl
-AC_REQUIRE([AM_PROG_MKDIR_P])dnl
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
+AC_REQUIRE([AC_PROG_MKDIR_P])dnl
+# For better backward compatibility.  To be removed once Automake 1.9.x
+# dies out for good.  For more background, see:
+# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
+# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
+AC_SUBST([mkdir_p], ['$(MKDIR_P)'])
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
 AC_REQUIRE([AC_PROG_AWK])dnl
 AC_REQUIRE([AC_PROG_MAKE_SET])dnl
 AC_REQUIRE([AM_SET_LEADING_DOT])dnl
@@ -496,34 +508,82 @@ _AM_IF_OPTION([tar-ustar], [_AM_PROG_TAR([ustar])],
 			     [_AM_PROG_TAR([v7])])])
 _AM_IF_OPTION([no-dependencies],,
 [AC_PROVIDE_IFELSE([AC_PROG_CC],
-		  [_AM_DEPENDENCIES(CC)],
-		  [define([AC_PROG_CC],
-			  defn([AC_PROG_CC])[_AM_DEPENDENCIES(CC)])])dnl
+		  [_AM_DEPENDENCIES([CC])],
+		  [m4_define([AC_PROG_CC],
+			     m4_defn([AC_PROG_CC])[_AM_DEPENDENCIES([CC])])])dnl
 AC_PROVIDE_IFELSE([AC_PROG_CXX],
-		  [_AM_DEPENDENCIES(CXX)],
-		  [define([AC_PROG_CXX],
-			  defn([AC_PROG_CXX])[_AM_DEPENDENCIES(CXX)])])dnl
+		  [_AM_DEPENDENCIES([CXX])],
+		  [m4_define([AC_PROG_CXX],
+			     m4_defn([AC_PROG_CXX])[_AM_DEPENDENCIES([CXX])])])dnl
 AC_PROVIDE_IFELSE([AC_PROG_OBJC],
-		  [_AM_DEPENDENCIES(OBJC)],
-		  [define([AC_PROG_OBJC],
-			  defn([AC_PROG_OBJC])[_AM_DEPENDENCIES(OBJC)])])dnl
+		  [_AM_DEPENDENCIES([OBJC])],
+		  [m4_define([AC_PROG_OBJC],
+			     m4_defn([AC_PROG_OBJC])[_AM_DEPENDENCIES([OBJC])])])dnl
+AC_PROVIDE_IFELSE([AC_PROG_OBJCXX],
+		  [_AM_DEPENDENCIES([OBJCXX])],
+		  [m4_define([AC_PROG_OBJCXX],
+			     m4_defn([AC_PROG_OBJCXX])[_AM_DEPENDENCIES([OBJCXX])])])dnl
 ])
-_AM_IF_OPTION([silent-rules], [AC_REQUIRE([AM_SILENT_RULES])])dnl
-dnl The `parallel-tests' driver may need to know about EXEEXT, so add the
-dnl `am__EXEEXT' conditional if _AM_COMPILER_EXEEXT was seen.  This macro
-dnl is hooked onto _AC_COMPILER_EXEEXT early, see below.
+AC_REQUIRE([AM_SILENT_RULES])dnl
+dnl The testsuite driver may need to know about EXEEXT, so add the
+dnl 'am__EXEEXT' conditional if _AM_COMPILER_EXEEXT was seen.  This
+dnl macro is hooked onto _AC_COMPILER_EXEEXT early, see below.
 AC_CONFIG_COMMANDS_PRE(dnl
 [m4_provide_if([_AM_COMPILER_EXEEXT],
   [AM_CONDITIONAL([am__EXEEXT], [test -n "$EXEEXT"])])])dnl
+
+# POSIX will say in a future version that running "rm -f" with no argument
+# is OK; and we want to be able to make that assumption in our Makefile
+# recipes.  So use an aggressive probe to check that the usage we want is
+# actually supported "in the wild" to an acceptable degree.
+# See automake bug#10828.
+# To make any issue more visible, cause the running configure to be aborted
+# by default if the 'rm' program in use doesn't match our expectations; the
+# user can still override this though.
+if rm -f && rm -fr && rm -rf; then : OK; else
+  cat >&2 <<'END'
+Oops!
+
+Your 'rm' program seems unable to run without file operands specified
+on the command line, even when the '-f' option is present.  This is contrary
+to the behaviour of most rm programs out there, and not conforming with
+the upcoming POSIX standard: <http://austingroupbugs.net/view.php?id=542>
+
+Please tell bug-automake@gnu.org about your system, including the value
+of your $PATH and any error possibly output before this message.  This
+can help us improve future automake versions.
+
+END
+  if test x"$ACCEPT_INFERIOR_RM_PROGRAM" = x"yes"; then
+    echo 'Configuration will proceed anyway, since you have set the' >&2
+    echo 'ACCEPT_INFERIOR_RM_PROGRAM variable to "yes"' >&2
+    echo >&2
+  else
+    cat >&2 <<'END'
+Aborting the configuration process, to ensure you take notice of the issue.
+
+You can download and install GNU coreutils to get an 'rm' implementation
+that behaves properly: <http://www.gnu.org/software/coreutils/>.
+
+If you want to complete the configuration process using your problematic
+'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
+to "yes", and re-run configure.
+
+END
+    AC_MSG_ERROR([Your 'rm' program is bad, sorry.])
+  fi
+fi
+dnl The trailing newline in this macro's definition is deliberate, for
+dnl backward compatibility and to allow trailing 'dnl'-style comments
+dnl after the AM_INIT_AUTOMAKE invocation. See automake bug#16841.
 ])
 
-dnl Hook into `_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
+dnl Hook into '_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
 dnl add the conditional right here, as _AC_COMPILER_EXEEXT may be further
 dnl mangled by Autoconf and run in a shell conditional statement.
 m4_define([_AC_COMPILER_EXEEXT],
 m4_defn([_AC_COMPILER_EXEEXT])[m4_provide([_AM_COMPILER_EXEEXT])])
 
-
 # When config.status generates a header, we must update the stamp-h file.
 # This file resides in the same directory as the config header
 # that is generated.  The stamp files are numbered to have different names.
@@ -545,7 +605,7 @@ for _am_header in $config_headers :; do
 done
 echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_count])
 
-# Copyright (C) 2001, 2003, 2005, 2008  Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -556,7 +616,7 @@ echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_co
 # Define $install_sh.
 AC_DEFUN([AM_PROG_INSTALL_SH],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -564,16 +624,14 @@ if test x"${install_sh}" != xset; then
     install_sh="\${SHELL} $am_aux_dir/install-sh"
   esac
 fi
-AC_SUBST(install_sh)])
+AC_SUBST([install_sh])])
 
-# Copyright (C) 2003, 2005  Free Software Foundation, Inc.
+# Copyright (C) 2003-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 2
-
 # Check whether the underlying file-system supports filenames
 # with a leading dot.  For instance MS-DOS doesn't.
 AC_DEFUN([AM_SET_LEADING_DOT],
@@ -589,14 +647,12 @@ AC_SUBST([am__leading_dot])])
 
 # Check to see how 'make' treats includes.	            -*- Autoconf -*-
 
-# Copyright (C) 2001, 2002, 2003, 2005, 2009  Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 4
-
 # AM_MAKE_INCLUDE()
 # -----------------
 # Check to see how make treats includes.
@@ -614,7 +670,7 @@ am__quote=
 _am_result=none
 # First try GNU make style include.
 echo "include confinc" > confmf
-# Ignore all kinds of additional output from `make'.
+# Ignore all kinds of additional output from 'make'.
 case `$am_make -s -f confmf 2> /dev/null` in #(
 *the\ am__doit\ target*)
   am__include=include
@@ -641,15 +697,12 @@ rm -f confinc confmf
 
 # Fake the existence of programs that GNU maintainers use.  -*- Autoconf -*-
 
-# Copyright (C) 1997, 1999, 2000, 2001, 2003, 2004, 2005, 2008
-# Free Software Foundation, Inc.
+# Copyright (C) 1997-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 6
-
 # AM_MISSING_PROG(NAME, PROGRAM)
 # ------------------------------
 AC_DEFUN([AM_MISSING_PROG],
@@ -657,11 +710,10 @@ AC_DEFUN([AM_MISSING_PROG],
 $1=${$1-"${am_missing_run}$2"}
 AC_SUBST($1)])
 
-
 # AM_MISSING_HAS_RUN
 # ------------------
-# Define MISSING if not defined so far and test if it supports --run.
-# If it does, set am_missing_run to use it, otherwise, to nothing.
+# Define MISSING if not defined so far and test if it is modern enough.
+# If it is, set am_missing_run to use it, otherwise, to nothing.
 AC_DEFUN([AM_MISSING_HAS_RUN],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
 AC_REQUIRE_AUX_FILE([missing])dnl
@@ -674,63 +726,35 @@ if test x"${MISSING+set}" != xset; then
   esac
 fi
 # Use eval to expand $SHELL
-if eval "$MISSING --run true"; then
-  am_missing_run="$MISSING --run "
+if eval "$MISSING --is-lightweight"; then
+  am_missing_run="$MISSING "
 else
   am_missing_run=
-  AC_MSG_WARN([`missing' script is too old or missing])
+  AC_MSG_WARN(['missing' script is too old or missing])
 fi
 ])
 
-# Copyright (C) 2003, 2004, 2005, 2006  Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# AM_PROG_MKDIR_P
-# ---------------
-# Check for `mkdir -p'.
-AC_DEFUN([AM_PROG_MKDIR_P],
-[AC_PREREQ([2.60])dnl
-AC_REQUIRE([AC_PROG_MKDIR_P])dnl
-dnl Automake 1.8 to 1.9.6 used to define mkdir_p.  We now use MKDIR_P,
-dnl while keeping a definition of mkdir_p for backward compatibility.
-dnl @MKDIR_P@ is magic: AC_OUTPUT adjusts its value for each Makefile.
-dnl However we cannot define mkdir_p as $(MKDIR_P) for the sake of
-dnl Makefile.ins that do not define MKDIR_P, so we do our own
-dnl adjustment using top_builddir (which is defined more often than
-dnl MKDIR_P).
-AC_SUBST([mkdir_p], ["$MKDIR_P"])dnl
-case $mkdir_p in
-  [[\\/$]]* | ?:[[\\/]]*) ;;
-  */*) mkdir_p="\$(top_builddir)/$mkdir_p" ;;
-esac
-])
-
 # Helper functions for option handling.                     -*- Autoconf -*-
 
-# Copyright (C) 2001, 2002, 2003, 2005, 2008  Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 4
-
 # _AM_MANGLE_OPTION(NAME)
 # -----------------------
 AC_DEFUN([_AM_MANGLE_OPTION],
 [[_AM_OPTION_]m4_bpatsubst($1, [[^a-zA-Z0-9_]], [_])])
 
 # _AM_SET_OPTION(NAME)
-# ------------------------------
+# --------------------
 # Set option NAME.  Presently that only means defining a flag for this option.
 AC_DEFUN([_AM_SET_OPTION],
-[m4_define(_AM_MANGLE_OPTION([$1]), 1)])
+[m4_define(_AM_MANGLE_OPTION([$1]), [1])])
 
 # _AM_SET_OPTIONS(OPTIONS)
-# ----------------------------------
+# ------------------------
 # OPTIONS is a space-separated list of Automake options.
 AC_DEFUN([_AM_SET_OPTIONS],
 [m4_foreach_w([_AM_Option], [$1], [_AM_SET_OPTION(_AM_Option)])])
@@ -741,24 +765,82 @@ AC_DEFUN([_AM_SET_OPTIONS],
 AC_DEFUN([_AM_IF_OPTION],
 [m4_ifset(_AM_MANGLE_OPTION([$1]), [$2], [$3])])
 
-# Check to make sure that the build environment is sane.    -*- Autoconf -*-
-
-# Copyright (C) 1996, 1997, 2000, 2001, 2003, 2005, 2008
-# Free Software Foundation, Inc.
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 5
+# _AM_PROG_CC_C_O
+# ---------------
+# Like AC_PROG_CC_C_O, but changed for automake.  We rewrite AC_PROG_CC
+# to automatically call this.
+AC_DEFUN([_AM_PROG_CC_C_O],
+[AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
+AC_REQUIRE_AUX_FILE([compile])dnl
+AC_LANG_PUSH([C])dnl
+AC_CACHE_CHECK(
+  [whether $CC understands -c and -o together],
+  [am_cv_prog_cc_c_o],
+  [AC_LANG_CONFTEST([AC_LANG_PROGRAM([])])
+  # Make sure it works both with $CC and with simple cc.
+  # Following AC_PROG_CC_C_O, we do the test twice because some
+  # compilers refuse to overwrite an existing .o file with -o,
+  # though they will create one.
+  am_cv_prog_cc_c_o=yes
+  for am_i in 1 2; do
+    if AM_RUN_LOG([$CC -c conftest.$ac_ext -o conftest2.$ac_objext]) \
+         && test -f conftest2.$ac_objext; then
+      : OK
+    else
+      am_cv_prog_cc_c_o=no
+      break
+    fi
+  done
+  rm -f core conftest*
+  unset am_i])
+if test "$am_cv_prog_cc_c_o" != yes; then
+   # Losing compiler, so override with the script.
+   # FIXME: It is wrong to rewrite CC.
+   # But if we don't then we get into trouble of one sort or another.
+   # A longer-term fix would be to have automake use am__CC in this case,
+   # and then we could set am__CC="\$(top_srcdir)/compile \$(CC)"
+   CC="$am_aux_dir/compile $CC"
+fi
+AC_LANG_POP([C])])
+
+# For backward compatibility.
+AC_DEFUN_ONCE([AM_PROG_CC_C_O], [AC_REQUIRE([AC_PROG_CC])])
+
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# AM_RUN_LOG(COMMAND)
+# -------------------
+# Run COMMAND, save the exit status in ac_status, and log it.
+# (This has been adapted from Autoconf's _AC_RUN_LOG macro.)
+AC_DEFUN([AM_RUN_LOG],
+[{ echo "$as_me:$LINENO: $1" >&AS_MESSAGE_LOG_FD
+   ($1) >&AS_MESSAGE_LOG_FD 2>&AS_MESSAGE_LOG_FD
+   ac_status=$?
+   echo "$as_me:$LINENO: \$? = $ac_status" >&AS_MESSAGE_LOG_FD
+   (exit $ac_status); }])
+
+# Check to make sure that the build environment is sane.    -*- Autoconf -*-
+
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
 
 # AM_SANITY_CHECK
 # ---------------
 AC_DEFUN([AM_SANITY_CHECK],
 [AC_MSG_CHECKING([whether build environment is sane])
-# Just in case
-sleep 1
-echo timestamp > conftest.file
 # Reject unsafe characters in $srcdir or the absolute working directory
 # name.  Accept space and tab only in the latter.
 am_lf='
@@ -769,32 +851,40 @@ case `pwd` in
 esac
 case $srcdir in
   *[[\\\"\#\$\&\'\`$am_lf\ \	]]*)
-    AC_MSG_ERROR([unsafe srcdir value: `$srcdir']);;
+    AC_MSG_ERROR([unsafe srcdir value: '$srcdir']);;
 esac
 
-# Do `set' in a subshell so we don't clobber the current shell's
+# Do 'set' in a subshell so we don't clobber the current shell's
 # arguments.  Must try -L first in case configure is actually a
 # symlink; some systems play weird games with the mod time of symlinks
 # (eg FreeBSD returns the mod time of the symlink's containing
 # directory).
 if (
-   set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
-   if test "$[*]" = "X"; then
-      # -L didn't work.
-      set X `ls -t "$srcdir/configure" conftest.file`
-   fi
-   rm -f conftest.file
-   if test "$[*]" != "X $srcdir/configure conftest.file" \
-      && test "$[*]" != "X conftest.file $srcdir/configure"; then
-
-      # If neither matched, then we have a broken ls.  This can happen
-      # if, for instance, CONFIG_SHELL is bash and it inherits a
-      # broken ls alias from the environment.  This has actually
-      # happened.  Such a system could not be considered "sane".
-      AC_MSG_ERROR([ls -t appears to fail.  Make sure there is not a broken
-alias in your environment])
-   fi
+   am_has_slept=no
+   for am_try in 1 2; do
+     echo "timestamp, slept: $am_has_slept" > conftest.file
+     set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
+     if test "$[*]" = "X"; then
+	# -L didn't work.
+	set X `ls -t "$srcdir/configure" conftest.file`
+     fi
+     if test "$[*]" != "X $srcdir/configure conftest.file" \
+	&& test "$[*]" != "X conftest.file $srcdir/configure"; then
 
+	# If neither matched, then we have a broken ls.  This can happen
+	# if, for instance, CONFIG_SHELL is bash and it inherits a
+	# broken ls alias from the environment.  This has actually
+	# happened.  Such a system could not be considered "sane".
+	AC_MSG_ERROR([ls -t appears to fail.  Make sure there is not a broken
+  alias in your environment])
+     fi
+     if test "$[2]" = conftest.file || test $am_try -eq 2; then
+       break
+     fi
+     # Just in case.
+     sleep 1
+     am_has_slept=yes
+   done
    test "$[2]" = conftest.file
    )
 then
@@ -804,9 +894,85 @@ else
    AC_MSG_ERROR([newly created file is older than distributed files!
 Check your system clock])
 fi
-AC_MSG_RESULT(yes)])
+AC_MSG_RESULT([yes])
+# If we didn't sleep, we still need to ensure time stamps of config.status and
+# generated files are strictly newer.
+am_sleep_pid=
+if grep 'slept: no' conftest.file >/dev/null 2>&1; then
+  ( sleep 1 ) &
+  am_sleep_pid=$!
+fi
+AC_CONFIG_COMMANDS_PRE(
+  [AC_MSG_CHECKING([that generated files are newer than configure])
+   if test -n "$am_sleep_pid"; then
+     # Hide warnings about reused PIDs.
+     wait $am_sleep_pid 2>/dev/null
+   fi
+   AC_MSG_RESULT([done])])
+rm -f conftest.file
+])
 
-# Copyright (C) 2001, 2003, 2005  Free Software Foundation, Inc.
+# Copyright (C) 2009-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# AM_SILENT_RULES([DEFAULT])
+# --------------------------
+# Enable less verbose build rules; with the default set to DEFAULT
+# ("yes" being less verbose, "no" or empty being verbose).
+AC_DEFUN([AM_SILENT_RULES],
+[AC_ARG_ENABLE([silent-rules], [dnl
+AS_HELP_STRING(
+  [--enable-silent-rules],
+  [less verbose build output (undo: "make V=1")])
+AS_HELP_STRING(
+  [--disable-silent-rules],
+  [verbose build output (undo: "make V=0")])dnl
+])
+case $enable_silent_rules in @%:@ (((
+  yes) AM_DEFAULT_VERBOSITY=0;;
+   no) AM_DEFAULT_VERBOSITY=1;;
+    *) AM_DEFAULT_VERBOSITY=m4_if([$1], [yes], [0], [1]);;
+esac
+dnl
+dnl A few 'make' implementations (e.g., NonStop OS and NextStep)
+dnl do not support nested variable expansions.
+dnl See automake bug#9928 and bug#10237.
+am_make=${MAKE-make}
+AC_CACHE_CHECK([whether $am_make supports nested variables],
+   [am_cv_make_support_nested_variables],
+   [if AS_ECHO([['TRUE=$(BAR$(V))
+BAR0=false
+BAR1=true
+V=1
+am__doit:
+	@$(TRUE)
+.PHONY: am__doit']]) | $am_make -f - >/dev/null 2>&1; then
+  am_cv_make_support_nested_variables=yes
+else
+  am_cv_make_support_nested_variables=no
+fi])
+if test $am_cv_make_support_nested_variables = yes; then
+  dnl Using '$V' instead of '$(V)' breaks IRIX make.
+  AM_V='$(V)'
+  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
+else
+  AM_V=$AM_DEFAULT_VERBOSITY
+  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
+fi
+AC_SUBST([AM_V])dnl
+AM_SUBST_NOTMAKE([AM_V])dnl
+AC_SUBST([AM_DEFAULT_V])dnl
+AM_SUBST_NOTMAKE([AM_DEFAULT_V])dnl
+AC_SUBST([AM_DEFAULT_VERBOSITY])dnl
+AM_BACKSLASH='\'
+AC_SUBST([AM_BACKSLASH])dnl
+_AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
+])
+
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -814,34 +980,32 @@ AC_MSG_RESULT(yes)])
 
 # AM_PROG_INSTALL_STRIP
 # ---------------------
-# One issue with vendor `install' (even GNU) is that you can't
+# One issue with vendor 'install' (even GNU) is that you can't
 # specify the program used to strip binaries.  This is especially
 # annoying in cross-compiling environments, where the build's strip
 # is unlikely to handle the host's binaries.
 # Fortunately install-sh will honor a STRIPPROG variable, so we
-# always use install-sh in `make install-strip', and initialize
+# always use install-sh in "make install-strip", and initialize
 # STRIPPROG with the value of the STRIP variable (set by the user).
 AC_DEFUN([AM_PROG_INSTALL_STRIP],
 [AC_REQUIRE([AM_PROG_INSTALL_SH])dnl
-# Installed binaries are usually stripped using `strip' when the user
-# run `make install-strip'.  However `strip' might not be the right
+# Installed binaries are usually stripped using 'strip' when the user
+# run "make install-strip".  However 'strip' might not be the right
 # tool to use in cross-compilation environments, therefore Automake
-# will honor the `STRIP' environment variable to overrule this program.
-dnl Don't test for $cross_compiling = yes, because it might be `maybe'.
+# will honor the 'STRIP' environment variable to overrule this program.
+dnl Don't test for $cross_compiling = yes, because it might be 'maybe'.
 if test "$cross_compiling" != no; then
   AC_CHECK_TOOL([STRIP], [strip], :)
 fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 AC_SUBST([INSTALL_STRIP_PROGRAM])])
 
-# Copyright (C) 2006, 2008  Free Software Foundation, Inc.
+# Copyright (C) 2006-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 2
-
 # _AM_SUBST_NOTMAKE(VARIABLE)
 # ---------------------------
 # Prevent Automake from outputting VARIABLE = @VARIABLE@ in Makefile.in.
@@ -849,24 +1013,22 @@ AC_SUBST([INSTALL_STRIP_PROGRAM])])
 AC_DEFUN([_AM_SUBST_NOTMAKE])
 
 # AM_SUBST_NOTMAKE(VARIABLE)
-# ---------------------------
+# --------------------------
 # Public sister of _AM_SUBST_NOTMAKE.
 AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 
 # Check how to create a tarball.                            -*- Autoconf -*-
 
-# Copyright (C) 2004, 2005  Free Software Foundation, Inc.
+# Copyright (C) 2004-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 2
-
 # _AM_PROG_TAR(FORMAT)
 # --------------------
 # Check how to create a tarball in format FORMAT.
-# FORMAT should be one of `v7', `ustar', or `pax'.
+# FORMAT should be one of 'v7', 'ustar', or 'pax'.
 #
 # Substitute a variable $(am__tar) that is a command
 # writing to stdout a FORMAT-tarball containing the directory
@@ -876,75 +1038,114 @@ AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 # Substitute a variable $(am__untar) that extract such
 # a tarball read from stdin.
 #     $(am__untar) < result.tar
+#
 AC_DEFUN([_AM_PROG_TAR],
-[# Always define AMTAR for backward compatibility.
-AM_MISSING_PROG([AMTAR], [tar])
-m4_if([$1], [v7],
-     [am__tar='${AMTAR} chof - "$$tardir"'; am__untar='${AMTAR} xf -'],
-     [m4_case([$1], [ustar],, [pax],,
-              [m4_fatal([Unknown tar format])])
-AC_MSG_CHECKING([how to create a $1 tar archive])
-# Loop over all known methods to create a tar archive until one works.
+[# Always define AMTAR for backward compatibility.  Yes, it's still used
+# in the wild :-(  We should find a proper way to deprecate it ...
+AC_SUBST([AMTAR], ['$${TAR-tar}'])
+
+# We'll loop over all known methods to create a tar archive until one works.
 _am_tools='gnutar m4_if([$1], [ustar], [plaintar]) pax cpio none'
-_am_tools=${am_cv_prog_tar_$1-$_am_tools}
-# Do not fold the above two line into one, because Tru64 sh and
-# Solaris sh will not grok spaces in the rhs of `-'.
-for _am_tool in $_am_tools
-do
-  case $_am_tool in
-  gnutar)
-    for _am_tar in tar gnutar gtar;
-    do
-      AM_RUN_LOG([$_am_tar --version]) && break
-    done
-    am__tar="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$$tardir"'
-    am__tar_="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$tardir"'
-    am__untar="$_am_tar -xf -"
-    ;;
-  plaintar)
-    # Must skip GNU tar: if it does not support --format= it doesn't create
-    # ustar tarball either.
-    (tar --version) >/dev/null 2>&1 && continue
-    am__tar='tar chf - "$$tardir"'
-    am__tar_='tar chf - "$tardir"'
-    am__untar='tar xf -'
-    ;;
-  pax)
-    am__tar='pax -L -x $1 -w "$$tardir"'
-    am__tar_='pax -L -x $1 -w "$tardir"'
-    am__untar='pax -r'
-    ;;
-  cpio)
-    am__tar='find "$$tardir" -print | cpio -o -H $1 -L'
-    am__tar_='find "$tardir" -print | cpio -o -H $1 -L'
-    am__untar='cpio -i -H $1 -d'
-    ;;
-  none)
-    am__tar=false
-    am__tar_=false
-    am__untar=false
-    ;;
-  esac
 
-  # If the value was cached, stop now.  We just wanted to have am__tar
-  # and am__untar set.
-  test -n "${am_cv_prog_tar_$1}" && break
+m4_if([$1], [v7],
+  [am__tar='$${TAR-tar} chof - "$$tardir"' am__untar='$${TAR-tar} xf -'],
 
-  # tar/untar a dummy directory, and stop if the command works
+  [m4_case([$1],
+    [ustar],
+     [# The POSIX 1988 'ustar' format is defined with fixed-size fields.
+      # There is notably a 21 bits limit for the UID and the GID.  In fact,
+      # the 'pax' utility can hang on bigger UID/GID (see automake bug#8343
+      # and bug#13588).
+      am_max_uid=2097151 # 2^21 - 1
+      am_max_gid=$am_max_uid
+      # The $UID and $GID variables are not portable, so we need to resort
+      # to the POSIX-mandated id(1) utility.  Errors in the 'id' calls
+      # below are definitely unexpected, so allow the users to see them
+      # (that is, avoid stderr redirection).
+      am_uid=`id -u || echo unknown`
+      am_gid=`id -g || echo unknown`
+      AC_MSG_CHECKING([whether UID '$am_uid' is supported by ustar format])
+      if test $am_uid -le $am_max_uid; then
+         AC_MSG_RESULT([yes])
+      else
+         AC_MSG_RESULT([no])
+         _am_tools=none
+      fi
+      AC_MSG_CHECKING([whether GID '$am_gid' is supported by ustar format])
+      if test $am_gid -le $am_max_gid; then
+         AC_MSG_RESULT([yes])
+      else
+        AC_MSG_RESULT([no])
+        _am_tools=none
+      fi],
+
+  [pax],
+    [],
+
+  [m4_fatal([Unknown tar format])])
+
+  AC_MSG_CHECKING([how to create a $1 tar archive])
+
+  # Go ahead even if we have the value already cached.  We do so because we
+  # need to set the values for the 'am__tar' and 'am__untar' variables.
+  _am_tools=${am_cv_prog_tar_$1-$_am_tools}
+
+  for _am_tool in $_am_tools; do
+    case $_am_tool in
+    gnutar)
+      for _am_tar in tar gnutar gtar; do
+        AM_RUN_LOG([$_am_tar --version]) && break
+      done
+      am__tar="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$$tardir"'
+      am__tar_="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$tardir"'
+      am__untar="$_am_tar -xf -"
+      ;;
+    plaintar)
+      # Must skip GNU tar: if it does not support --format= it doesn't create
+      # ustar tarball either.
+      (tar --version) >/dev/null 2>&1 && continue
+      am__tar='tar chf - "$$tardir"'
+      am__tar_='tar chf - "$tardir"'
+      am__untar='tar xf -'
+      ;;
+    pax)
+      am__tar='pax -L -x $1 -w "$$tardir"'
+      am__tar_='pax -L -x $1 -w "$tardir"'
+      am__untar='pax -r'
+      ;;
+    cpio)
+      am__tar='find "$$tardir" -print | cpio -o -H $1 -L'
+      am__tar_='find "$tardir" -print | cpio -o -H $1 -L'
+      am__untar='cpio -i -H $1 -d'
+      ;;
+    none)
+      am__tar=false
+      am__tar_=false
+      am__untar=false
+      ;;
+    esac
+
+    # If the value was cached, stop now.  We just wanted to have am__tar
+    # and am__untar set.
+    test -n "${am_cv_prog_tar_$1}" && break
+
+    # tar/untar a dummy directory, and stop if the command works.
+    rm -rf conftest.dir
+    mkdir conftest.dir
+    echo GrepMe > conftest.dir/file
+    AM_RUN_LOG([tardir=conftest.dir && eval $am__tar_ >conftest.tar])
+    rm -rf conftest.dir
+    if test -s conftest.tar; then
+      AM_RUN_LOG([$am__untar <conftest.tar])
+      AM_RUN_LOG([cat conftest.dir/file])
+      grep GrepMe conftest.dir/file >/dev/null 2>&1 && break
+    fi
+  done
   rm -rf conftest.dir
-  mkdir conftest.dir
-  echo GrepMe > conftest.dir/file
-  AM_RUN_LOG([tardir=conftest.dir && eval $am__tar_ >conftest.tar])
-  rm -rf conftest.dir
-  if test -s conftest.tar; then
-    AM_RUN_LOG([$am__untar <conftest.tar])
-    grep GrepMe conftest.dir/file >/dev/null 2>&1 && break
-  fi
-done
-rm -rf conftest.dir
 
-AC_CACHE_VAL([am_cv_prog_tar_$1], [am_cv_prog_tar_$1=$_am_tool])
-AC_MSG_RESULT([$am_cv_prog_tar_$1])])
+  AC_CACHE_VAL([am_cv_prog_tar_$1], [am_cv_prog_tar_$1=$_am_tool])
+  AC_MSG_RESULT([$am_cv_prog_tar_$1])])
+
 AC_SUBST([am__tar])
 AC_SUBST([am__untar])
 ]) # _AM_PROG_TAR
diff --git a/benchmarks/Grid_wilson.cc b/benchmarks/Grid_wilson.cc
index 51255fcb..ecd7b65b 100644
--- a/benchmarks/Grid_wilson.cc
+++ b/benchmarks/Grid_wilson.cc
@@ -76,7 +76,7 @@ int main (int argc, char ** argv)
   WilsonMatrix Dw(Umu,mass);
   
   std::cout << "Calling Dw"<<std::endl;
-  int ncall=100;
+  int ncall=10000;
   double t0=usecond();
   for(int i=0;i<ncall;i++){
     Dw.multiply(src,result);
diff --git a/configure b/configure
index 3e71855d..3d25f6a9 100755
--- a/configure
+++ b/configure
@@ -1,20 +1,22 @@
 #! /bin/sh
 # Guess values for system-dependent variables and create Makefiles.
-# Generated by GNU Autoconf 2.63 for Grid 1.0.
+# Generated by GNU Autoconf 2.69 for Grid 1.0.
 #
 # Report bugs to <paboyle@ph.ed.ac.uk>.
 #
-# Copyright (C) 1992, 1993, 1994, 1995, 1996, 1998, 1999, 2000, 2001,
-# 2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
+#
+# Copyright (C) 1992-1996, 1998-2012 Free Software Foundation, Inc.
+#
+#
 # This configure script is free software; the Free Software Foundation
 # gives unlimited permission to copy, distribute and modify it.
-## --------------------- ##
-## M4sh Initialization.  ##
-## --------------------- ##
+## -------------------- ##
+## M4sh Initialization. ##
+## -------------------- ##
 
 # Be more Bourne compatible
 DUALCASE=1; export DUALCASE # for MKS sh
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
   emulate sh
   NULLCMD=:
   # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
@@ -22,23 +24,15 @@ if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
   alias -g '${1+"$@"}'='"$@"'
   setopt NO_GLOB_SUBST
 else
-  case `(set -o) 2>/dev/null` in
-  *posix*) set -o posix ;;
+  case `(set -o) 2>/dev/null` in #(
+  *posix*) :
+    set -o posix ;; #(
+  *) :
+     ;;
 esac
-
 fi
 
 
-
-
-# PATH needs CR
-# Avoid depending upon Character Ranges.
-as_cr_letters='abcdefghijklmnopqrstuvwxyz'
-as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
-as_cr_Letters=$as_cr_letters$as_cr_LETTERS
-as_cr_digits='0123456789'
-as_cr_alnum=$as_cr_Letters$as_cr_digits
-
 as_nl='
 '
 export as_nl
@@ -46,7 +40,13 @@ export as_nl
 as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
-if (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+# Prefer a ksh shell builtin over an external printf program on Solaris,
+# but without wasting forks for bash or zsh.
+if test -z "$BASH_VERSION$ZSH_VERSION" \
+    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
+  as_echo='print -r --'
+  as_echo_n='print -rn --'
+elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
   as_echo='printf %s\n'
   as_echo_n='printf %s'
 else
@@ -57,7 +57,7 @@ else
     as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
     as_echo_n_body='eval
       arg=$1;
-      case $arg in
+      case $arg in #(
       *"$as_nl"*)
 	expr "X$arg" : "X\\(.*\\)$as_nl";
 	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
@@ -80,13 +80,6 @@ if test "${PATH_SEPARATOR+set}" != set; then
   }
 fi
 
-# Support unset when possible.
-if ( (MAIL=60; unset MAIL) || exit) >/dev/null 2>&1; then
-  as_unset=unset
-else
-  as_unset=false
-fi
-
 
 # IFS
 # We need space, tab and new line, in precisely that order.  Quoting is
@@ -96,15 +89,16 @@ fi
 IFS=" ""	$as_nl"
 
 # Find who we are.  Look in the path if we contain no directory separator.
-case $0 in
+as_myself=
+case $0 in #((
   *[\\/]* ) as_myself=$0 ;;
   *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
 for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
-done
+    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+  done
 IFS=$as_save_IFS
 
      ;;
@@ -116,12 +110,16 @@ if test "x$as_myself" = x; then
 fi
 if test ! -f "$as_myself"; then
   $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
-  { (exit 1); exit 1; }
+  exit 1
 fi
 
-# Work around bugs in pre-3.0 UWIN ksh.
-for as_var in ENV MAIL MAILPATH
-do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
+# Unset variables that we do not need and which cause bugs (e.g. in
+# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
+# suppresses any "Segmentation fault" message there.  '((' could
+# trigger a bug in pdksh 5.2.14.
+for as_var in BASH_ENV ENV MAIL MAILPATH
+do eval test x\${$as_var+set} = xset \
+  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
 done
 PS1='$ '
 PS2='> '
@@ -133,7 +131,294 @@ export LC_ALL
 LANGUAGE=C
 export LANGUAGE
 
-# Required to use basename.
+# CDPATH.
+(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
+
+# Use a proper internal environment variable to ensure we don't fall
+  # into an infinite loop, continuously re-executing ourselves.
+  if test x"${_as_can_reexec}" != xno && test "x$CONFIG_SHELL" != x; then
+    _as_can_reexec=no; export _as_can_reexec;
+    # We cannot yet assume a decent shell, so we have to provide a
+# neutralization value for shells without unset; and this also
+# works around shells that cannot unset nonexistent variables.
+# Preserve -v and -x to the replacement shell.
+BASH_ENV=/dev/null
+ENV=/dev/null
+(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
+case $- in # ((((
+  *v*x* | *x*v* ) as_opts=-vx ;;
+  *v* ) as_opts=-v ;;
+  *x* ) as_opts=-x ;;
+  * ) as_opts= ;;
+esac
+exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
+# Admittedly, this is quite paranoid, since all the known shells bail
+# out after a failed `exec'.
+$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
+as_fn_exit 255
+  fi
+  # We don't want this to propagate to other subprocesses.
+          { _as_can_reexec=; unset _as_can_reexec;}
+if test "x$CONFIG_SHELL" = x; then
+  as_bourne_compatible="if test -n \"\${ZSH_VERSION+set}\" && (emulate sh) >/dev/null 2>&1; then :
+  emulate sh
+  NULLCMD=:
+  # Pre-4.2 versions of Zsh do word splitting on \${1+\"\$@\"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '\${1+\"\$@\"}'='\"\$@\"'
+  setopt NO_GLOB_SUBST
+else
+  case \`(set -o) 2>/dev/null\` in #(
+  *posix*) :
+    set -o posix ;; #(
+  *) :
+     ;;
+esac
+fi
+"
+  as_required="as_fn_return () { (exit \$1); }
+as_fn_success () { as_fn_return 0; }
+as_fn_failure () { as_fn_return 1; }
+as_fn_ret_success () { return 0; }
+as_fn_ret_failure () { return 1; }
+
+exitcode=0
+as_fn_success || { exitcode=1; echo as_fn_success failed.; }
+as_fn_failure && { exitcode=1; echo as_fn_failure succeeded.; }
+as_fn_ret_success || { exitcode=1; echo as_fn_ret_success failed.; }
+as_fn_ret_failure && { exitcode=1; echo as_fn_ret_failure succeeded.; }
+if ( set x; as_fn_ret_success y && test x = \"\$1\" ); then :
+
+else
+  exitcode=1; echo positional parameters were not saved.
+fi
+test x\$exitcode = x0 || exit 1
+test -x / || exit 1"
+  as_suggested="  as_lineno_1=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_1a=\$LINENO
+  as_lineno_2=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_2a=\$LINENO
+  eval 'test \"x\$as_lineno_1'\$as_run'\" != \"x\$as_lineno_2'\$as_run'\" &&
+  test \"x\`expr \$as_lineno_1'\$as_run' + 1\`\" = \"x\$as_lineno_2'\$as_run'\"' || exit 1
+test \$(( 1 + 1 )) = 2 || exit 1"
+  if (eval "$as_required") 2>/dev/null; then :
+  as_have_required=yes
+else
+  as_have_required=no
+fi
+  if test x$as_have_required = xyes && (eval "$as_suggested") 2>/dev/null; then :
+
+else
+  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+as_found=false
+for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+  as_found=:
+  case $as_dir in #(
+	 /*)
+	   for as_base in sh bash ksh sh5; do
+	     # Try only shells that exist, to save several forks.
+	     as_shell=$as_dir/$as_base
+	     if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
+		    { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$as_shell"; } 2>/dev/null; then :
+  CONFIG_SHELL=$as_shell as_have_required=yes
+		   if { $as_echo "$as_bourne_compatible""$as_suggested" | as_run=a "$as_shell"; } 2>/dev/null; then :
+  break 2
+fi
+fi
+	   done;;
+       esac
+  as_found=false
+done
+$as_found || { if { test -f "$SHELL" || test -f "$SHELL.exe"; } &&
+	      { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$SHELL"; } 2>/dev/null; then :
+  CONFIG_SHELL=$SHELL as_have_required=yes
+fi; }
+IFS=$as_save_IFS
+
+
+      if test "x$CONFIG_SHELL" != x; then :
+  export CONFIG_SHELL
+             # We cannot yet assume a decent shell, so we have to provide a
+# neutralization value for shells without unset; and this also
+# works around shells that cannot unset nonexistent variables.
+# Preserve -v and -x to the replacement shell.
+BASH_ENV=/dev/null
+ENV=/dev/null
+(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
+case $- in # ((((
+  *v*x* | *x*v* ) as_opts=-vx ;;
+  *v* ) as_opts=-v ;;
+  *x* ) as_opts=-x ;;
+  * ) as_opts= ;;
+esac
+exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
+# Admittedly, this is quite paranoid, since all the known shells bail
+# out after a failed `exec'.
+$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
+exit 255
+fi
+
+    if test x$as_have_required = xno; then :
+  $as_echo "$0: This script requires a shell more modern than all"
+  $as_echo "$0: the shells that I found on your system."
+  if test x${ZSH_VERSION+set} = xset ; then
+    $as_echo "$0: In particular, zsh $ZSH_VERSION has bugs and should"
+    $as_echo "$0: be upgraded to zsh 4.3.4 or later."
+  else
+    $as_echo "$0: Please tell bug-autoconf@gnu.org and
+$0: paboyle@ph.ed.ac.uk about your system, including any
+$0: error possibly output before this message. Then install
+$0: a modern shell, or manually run the script under such a
+$0: shell if you do have one."
+  fi
+  exit 1
+fi
+fi
+fi
+SHELL=${CONFIG_SHELL-/bin/sh}
+export SHELL
+# Unset more variables known to interfere with behavior of common tools.
+CLICOLOR_FORCE= GREP_OPTIONS=
+unset CLICOLOR_FORCE GREP_OPTIONS
+
+## --------------------- ##
+## M4sh Shell Functions. ##
+## --------------------- ##
+# as_fn_unset VAR
+# ---------------
+# Portably unset VAR.
+as_fn_unset ()
+{
+  { eval $1=; unset $1;}
+}
+as_unset=as_fn_unset
+
+# as_fn_set_status STATUS
+# -----------------------
+# Set $? to STATUS, without forking.
+as_fn_set_status ()
+{
+  return $1
+} # as_fn_set_status
+
+# as_fn_exit STATUS
+# -----------------
+# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
+as_fn_exit ()
+{
+  set +e
+  as_fn_set_status $1
+  exit $1
+} # as_fn_exit
+
+# as_fn_mkdir_p
+# -------------
+# Create "$as_dir" as a directory, including parents if necessary.
+as_fn_mkdir_p ()
+{
+
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || eval $as_mkdir_p || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
+
+
+} # as_fn_mkdir_p
+
+# as_fn_executable_p FILE
+# -----------------------
+# Test if FILE is an executable regular file.
+as_fn_executable_p ()
+{
+  test -f "$1" && test -x "$1"
+} # as_fn_executable_p
+# as_fn_append VAR VALUE
+# ----------------------
+# Append the text in VALUE to the end of the definition contained in VAR. Take
+# advantage of any shell optimizations that allow amortized linear growth over
+# repeated appends, instead of the typical quadratic growth present in naive
+# implementations.
+if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
+  eval 'as_fn_append ()
+  {
+    eval $1+=\$2
+  }'
+else
+  as_fn_append ()
+  {
+    eval $1=\$$1\$2
+  }
+fi # as_fn_append
+
+# as_fn_arith ARG...
+# ------------------
+# Perform arithmetic evaluation on the ARGs, and store the result in the
+# global $as_val. Take advantage of shells that can avoid forks. The arguments
+# must be portable across $(()) and expr.
+if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
+  eval 'as_fn_arith ()
+  {
+    as_val=$(( $* ))
+  }'
+else
+  as_fn_arith ()
+  {
+    as_val=`expr "$@" || test $? -eq 1`
+  }
+fi # as_fn_arith
+
+
+# as_fn_error STATUS ERROR [LINENO LOG_FD]
+# ----------------------------------------
+# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
+# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
+# script with STATUS, using 1 if that was 0.
+as_fn_error ()
+{
+  as_status=$1; test $as_status -eq 0 && as_status=1
+  if test "$4"; then
+    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
+  fi
+  $as_echo "$as_me: error: $2" >&2
+  as_fn_exit $as_status
+} # as_fn_error
+
 if expr a : '\(a\)' >/dev/null 2>&1 &&
    test "X`expr 00001 : '.*\(...\)'`" = X001; then
   as_expr=expr
@@ -147,8 +432,12 @@ else
   as_basename=false
 fi
 
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
 
-# Name of the executable.
 as_me=`$as_basename -- "$0" ||
 $as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
 	 X"$0" : 'X\(//\)$' \| \
@@ -168,295 +457,19 @@ $as_echo X/"$0" |
 	  }
 	  s/.*/./; q'`
 
-# CDPATH.
-$as_unset CDPATH
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
 
 
-if test "x$CONFIG_SHELL" = x; then
-  if (eval ":") 2>/dev/null; then
-  as_have_required=yes
-else
-  as_have_required=no
-fi
-
-  if test $as_have_required = yes &&	 (eval ":
-(as_func_return () {
-  (exit \$1)
-}
-as_func_success () {
-  as_func_return 0
-}
-as_func_failure () {
-  as_func_return 1
-}
-as_func_ret_success () {
-  return 0
-}
-as_func_ret_failure () {
-  return 1
-}
-
-exitcode=0
-if as_func_success; then
-  :
-else
-  exitcode=1
-  echo as_func_success failed.
-fi
-
-if as_func_failure; then
-  exitcode=1
-  echo as_func_failure succeeded.
-fi
-
-if as_func_ret_success; then
-  :
-else
-  exitcode=1
-  echo as_func_ret_success failed.
-fi
-
-if as_func_ret_failure; then
-  exitcode=1
-  echo as_func_ret_failure succeeded.
-fi
-
-if ( set x; as_func_ret_success y && test x = \"\$1\" ); then
-  :
-else
-  exitcode=1
-  echo positional parameters were not saved.
-fi
-
-test \$exitcode = 0) || { (exit 1); exit 1; }
-
-(
-  as_lineno_1=\$LINENO
-  as_lineno_2=\$LINENO
-  test \"x\$as_lineno_1\" != \"x\$as_lineno_2\" &&
-  test \"x\`expr \$as_lineno_1 + 1\`\" = \"x\$as_lineno_2\") || { (exit 1); exit 1; }
-") 2> /dev/null; then
-  :
-else
-  as_candidate_shells=
-    as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-  case $as_dir in
-	 /*)
-	   for as_base in sh bash ksh sh5; do
-	     as_candidate_shells="$as_candidate_shells $as_dir/$as_base"
-	   done;;
-       esac
-done
-IFS=$as_save_IFS
-
-
-      for as_shell in $as_candidate_shells $SHELL; do
-	 # Try only shells that exist, to save several forks.
-	 if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
-		{ ("$as_shell") 2> /dev/null <<\_ASEOF
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
-  emulate sh
-  NULLCMD=:
-  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
-  # is contrary to our usage.  Disable this feature.
-  alias -g '${1+"$@"}'='"$@"'
-  setopt NO_GLOB_SUBST
-else
-  case `(set -o) 2>/dev/null` in
-  *posix*) set -o posix ;;
-esac
-
-fi
-
-
-:
-_ASEOF
-}; then
-  CONFIG_SHELL=$as_shell
-	       as_have_required=yes
-	       if { "$as_shell" 2> /dev/null <<\_ASEOF
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
-  emulate sh
-  NULLCMD=:
-  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
-  # is contrary to our usage.  Disable this feature.
-  alias -g '${1+"$@"}'='"$@"'
-  setopt NO_GLOB_SUBST
-else
-  case `(set -o) 2>/dev/null` in
-  *posix*) set -o posix ;;
-esac
-
-fi
-
-
-:
-(as_func_return () {
-  (exit $1)
-}
-as_func_success () {
-  as_func_return 0
-}
-as_func_failure () {
-  as_func_return 1
-}
-as_func_ret_success () {
-  return 0
-}
-as_func_ret_failure () {
-  return 1
-}
-
-exitcode=0
-if as_func_success; then
-  :
-else
-  exitcode=1
-  echo as_func_success failed.
-fi
-
-if as_func_failure; then
-  exitcode=1
-  echo as_func_failure succeeded.
-fi
-
-if as_func_ret_success; then
-  :
-else
-  exitcode=1
-  echo as_func_ret_success failed.
-fi
-
-if as_func_ret_failure; then
-  exitcode=1
-  echo as_func_ret_failure succeeded.
-fi
-
-if ( set x; as_func_ret_success y && test x = "$1" ); then
-  :
-else
-  exitcode=1
-  echo positional parameters were not saved.
-fi
-
-test $exitcode = 0) || { (exit 1); exit 1; }
-
-(
-  as_lineno_1=$LINENO
-  as_lineno_2=$LINENO
-  test "x$as_lineno_1" != "x$as_lineno_2" &&
-  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2") || { (exit 1); exit 1; }
-
-_ASEOF
-}; then
-  break
-fi
-
-fi
-
-      done
-
-      if test "x$CONFIG_SHELL" != x; then
-  for as_var in BASH_ENV ENV
-	do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
-	done
-	export CONFIG_SHELL
-	exec "$CONFIG_SHELL" "$as_myself" ${1+"$@"}
-fi
-
-
-    if test $as_have_required = no; then
-  echo This script requires a shell more modern than all the
-      echo shells that I found on your system.  Please install a
-      echo modern shell, or manually run the script under such a
-      echo shell if you do have one.
-      { (exit 1); exit 1; }
-fi
-
-
-fi
-
-fi
-
-
-
-(eval "as_func_return () {
-  (exit \$1)
-}
-as_func_success () {
-  as_func_return 0
-}
-as_func_failure () {
-  as_func_return 1
-}
-as_func_ret_success () {
-  return 0
-}
-as_func_ret_failure () {
-  return 1
-}
-
-exitcode=0
-if as_func_success; then
-  :
-else
-  exitcode=1
-  echo as_func_success failed.
-fi
-
-if as_func_failure; then
-  exitcode=1
-  echo as_func_failure succeeded.
-fi
-
-if as_func_ret_success; then
-  :
-else
-  exitcode=1
-  echo as_func_ret_success failed.
-fi
-
-if as_func_ret_failure; then
-  exitcode=1
-  echo as_func_ret_failure succeeded.
-fi
-
-if ( set x; as_func_ret_success y && test x = \"\$1\" ); then
-  :
-else
-  exitcode=1
-  echo positional parameters were not saved.
-fi
-
-test \$exitcode = 0") || {
-  echo No shell found that supports shell functions.
-  echo Please tell bug-autoconf@gnu.org about your system,
-  echo including any error possibly output before this message.
-  echo This can help us improve future autoconf versions.
-  echo Configuration will now proceed without shell functions.
-}
-
-
-
-  as_lineno_1=$LINENO
-  as_lineno_2=$LINENO
-  test "x$as_lineno_1" != "x$as_lineno_2" &&
-  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2" || {
-
-  # Create $as_me.lineno as a copy of $as_myself, but with $LINENO
-  # uniformly replaced by the line number.  The first 'sed' inserts a
-  # line-number line after each line using $LINENO; the second 'sed'
-  # does the real work.  The second script uses 'N' to pair each
-  # line-number line with the line containing $LINENO, and appends
-  # trailing '-' during substitution so that $LINENO is not a special
-  # case at line end.
-  # (Raja R Harinath suggested sed '=', and Paul Eggert wrote the
-  # scripts with optimization help from Paolo Bonzini.  Blame Lee
-  # E. McMahon (1931-1989) for sed's syntax.  :-)
+  as_lineno_1=$LINENO as_lineno_1a=$LINENO
+  as_lineno_2=$LINENO as_lineno_2a=$LINENO
+  eval 'test "x$as_lineno_1'$as_run'" != "x$as_lineno_2'$as_run'" &&
+  test "x`expr $as_lineno_1'$as_run' + 1`" = "x$as_lineno_2'$as_run'"' || {
+  # Blame Lee E. McMahon (1931-1989) for sed's syntax.  :-)
   sed -n '
     p
     /[$]LINENO/=
@@ -473,9 +486,12 @@ test \$exitcode = 0") || {
       s/-\n.*//
     ' >$as_me.lineno &&
   chmod +x "$as_me.lineno" ||
-    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2
-   { (exit 1); exit 1; }; }
+    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2; as_fn_exit 1; }
 
+  # If we had to re-execute with $CONFIG_SHELL, we're ensured to have
+  # already done that, so ensure we don't try to do so again and fall
+  # in an infinite loop.  This has already happened in practice.
+  _as_can_reexec=no; export _as_can_reexec
   # Don't try to exec as it changes $[0], causing all sort of problems
   # (the dirname of $[0] is not the place where we might find the
   # original and so on.  Autoconf is especially sensitive to this).
@@ -484,29 +500,18 @@ test \$exitcode = 0") || {
   exit
 }
 
-
-if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
-  as_dirname=dirname
-else
-  as_dirname=false
-fi
-
 ECHO_C= ECHO_N= ECHO_T=
-case `echo -n x` in
+case `echo -n x` in #(((((
 -n*)
-  case `echo 'x\c'` in
+  case `echo 'xy\c'` in
   *c*) ECHO_T='	';;	# ECHO_T is single tab character.
-  *)   ECHO_C='\c';;
+  xy)  ECHO_C='\c';;
+  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
+       ECHO_T='	';;
   esac;;
 *)
   ECHO_N='-n';;
 esac
-if expr a : '\(a\)' >/dev/null 2>&1 &&
-   test "X`expr 00001 : '.*\(...\)'`" = X001; then
-  as_expr=expr
-else
-  as_expr=false
-fi
 
 rm -f conf$$ conf$$.exe conf$$.file
 if test -d conf$$.dir; then
@@ -521,49 +526,29 @@ if (echo >conf$$.file) 2>/dev/null; then
     # ... but there are two gotchas:
     # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
     # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
-    # In both cases, we have to default to `cp -p'.
+    # In both cases, we have to default to `cp -pR'.
     ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
-      as_ln_s='cp -p'
+      as_ln_s='cp -pR'
   elif ln conf$$.file conf$$ 2>/dev/null; then
     as_ln_s=ln
   else
-    as_ln_s='cp -p'
+    as_ln_s='cp -pR'
   fi
 else
-  as_ln_s='cp -p'
+  as_ln_s='cp -pR'
 fi
 rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
 rmdir conf$$.dir 2>/dev/null
 
 if mkdir -p . 2>/dev/null; then
-  as_mkdir_p=:
+  as_mkdir_p='mkdir -p "$as_dir"'
 else
   test -d ./-p && rmdir ./-p
   as_mkdir_p=false
 fi
 
-if test -x / >/dev/null 2>&1; then
-  as_test_x='test -x'
-else
-  if ls -dL / >/dev/null 2>&1; then
-    as_ls_L_option=L
-  else
-    as_ls_L_option=
-  fi
-  as_test_x='
-    eval sh -c '\''
-      if test -d "$1"; then
-	test -d "$1/.";
-      else
-	case $1 in
-	-*)set "./$1";;
-	esac;
-	case `ls -ld'$as_ls_L_option' "$1" 2>/dev/null` in
-	???[sx]*):;;*)false;;esac;fi
-    '\'' sh
-  '
-fi
-as_executable_p=$as_test_x
+as_test_x='test -x'
+as_executable_p=as_fn_executable_p
 
 # Sed expression to map a string onto a valid CPP name.
 as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
@@ -572,11 +557,11 @@ as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
 as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
 
 
-
-exec 7<&0 </dev/null 6>&1
+test -n "$DJDIR" || exec 7<&0 </dev/null
+exec 6>&1
 
 # Name of the host.
-# hostname on some systems (SVR3.2, Linux) returns a bogus exit status,
+# hostname on some systems (SVR3.2, old GNU/Linux) returns a bogus exit status,
 # so uname gets run too.
 ac_hostname=`(hostname || uname -n) 2>/dev/null | sed 1q`
 
@@ -591,7 +576,6 @@ cross_compiling=no
 subdirs=
 MFLAGS=
 MAKEFLAGS=
-SHELL=${CONFIG_SHELL-/bin/sh}
 
 # Identity of this package.
 PACKAGE_NAME='Grid'
@@ -599,6 +583,7 @@ PACKAGE_TARNAME='grid'
 PACKAGE_VERSION='1.0'
 PACKAGE_STRING='Grid 1.0'
 PACKAGE_BUGREPORT='paboyle@ph.ed.ac.uk'
+PACKAGE_URL=''
 
 ac_unique_file="lib/Grid.h"
 # Factoring default headers for most tests.
@@ -659,6 +644,7 @@ CC
 am__fastdepCXX_FALSE
 am__fastdepCXX_TRUE
 CXXDEPMODE
+am__nodep
 AMDEPBACKSLASH
 AMDEP_FALSE
 AMDEP_TRUE
@@ -672,6 +658,10 @@ CPPFLAGS
 LDFLAGS
 CXXFLAGS
 CXX
+AM_BACKSLASH
+AM_DEFAULT_VERBOSITY
+AM_DEFAULT_V
+AM_V
 am__untar
 am__tar
 AMTAR
@@ -725,6 +715,7 @@ bindir
 program_transform_name
 prefix
 exec_prefix
+PACKAGE_URL
 PACKAGE_BUGREPORT
 PACKAGE_STRING
 PACKAGE_VERSION
@@ -735,6 +726,7 @@ SHELL'
 ac_subst_files=''
 ac_user_opts='
 enable_option_checking
+enable_silent_rules
 enable_dependency_tracking
 enable_openmp
 enable_simd
@@ -814,8 +806,9 @@ do
   fi
 
   case $ac_option in
-  *=*)	ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
-  *)	ac_optarg=yes ;;
+  *=?*) ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
+  *=)   ac_optarg= ;;
+  *)    ac_optarg=yes ;;
   esac
 
   # Accept the important Cygnus configure options, so we can diagnose typos.
@@ -860,8 +853,7 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*disable-\(.*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid feature name: $ac_useropt" >&2
-   { (exit 1); exit 1; }; }
+      as_fn_error $? "invalid feature name: $ac_useropt"
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -887,8 +879,7 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*enable-\([^=]*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid feature name: $ac_useropt" >&2
-   { (exit 1); exit 1; }; }
+      as_fn_error $? "invalid feature name: $ac_useropt"
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1092,8 +1083,7 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*with-\([^=]*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid package name: $ac_useropt" >&2
-   { (exit 1); exit 1; }; }
+      as_fn_error $? "invalid package name: $ac_useropt"
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1109,8 +1099,7 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*without-\(.*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid package name: $ac_useropt" >&2
-   { (exit 1); exit 1; }; }
+      as_fn_error $? "invalid package name: $ac_useropt"
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1140,17 +1129,17 @@ do
   | --x-librar=* | --x-libra=* | --x-libr=* | --x-lib=* | --x-li=* | --x-l=*)
     x_libraries=$ac_optarg ;;
 
-  -*) { $as_echo "$as_me: error: unrecognized option: $ac_option
-Try \`$0 --help' for more information." >&2
-   { (exit 1); exit 1; }; }
+  -*) as_fn_error $? "unrecognized option: \`$ac_option'
+Try \`$0 --help' for more information"
     ;;
 
   *=*)
     ac_envvar=`expr "x$ac_option" : 'x\([^=]*\)='`
     # Reject names that are not valid shell variable names.
-    expr "x$ac_envvar" : ".*[^_$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid variable name: $ac_envvar" >&2
-   { (exit 1); exit 1; }; }
+    case $ac_envvar in #(
+      '' | [0-9]* | *[!_$as_cr_alnum]* )
+      as_fn_error $? "invalid variable name: \`$ac_envvar'" ;;
+    esac
     eval $ac_envvar=\$ac_optarg
     export $ac_envvar ;;
 
@@ -1159,7 +1148,7 @@ Try \`$0 --help' for more information." >&2
     $as_echo "$as_me: WARNING: you should use --build, --host, --target" >&2
     expr "x$ac_option" : ".*[^-._$as_cr_alnum]" >/dev/null &&
       $as_echo "$as_me: WARNING: invalid host type: $ac_option" >&2
-    : ${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}
+    : "${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}"
     ;;
 
   esac
@@ -1167,15 +1156,13 @@ done
 
 if test -n "$ac_prev"; then
   ac_option=--`echo $ac_prev | sed 's/_/-/g'`
-  { $as_echo "$as_me: error: missing argument to $ac_option" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "missing argument to $ac_option"
 fi
 
 if test -n "$ac_unrecognized_opts"; then
   case $enable_option_checking in
     no) ;;
-    fatal) { $as_echo "$as_me: error: unrecognized options: $ac_unrecognized_opts" >&2
-   { (exit 1); exit 1; }; } ;;
+    fatal) as_fn_error $? "unrecognized options: $ac_unrecognized_opts" ;;
     *)     $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2 ;;
   esac
 fi
@@ -1198,8 +1185,7 @@ do
     [\\/$]* | ?:[\\/]* )  continue;;
     NONE | '' ) case $ac_var in *prefix ) continue;; esac;;
   esac
-  { $as_echo "$as_me: error: expected an absolute directory name for --$ac_var: $ac_val" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "expected an absolute directory name for --$ac_var: $ac_val"
 done
 
 # There might be people who depend on the old broken behavior: `$host'
@@ -1213,8 +1199,6 @@ target=$target_alias
 if test "x$host_alias" != x; then
   if test "x$build_alias" = x; then
     cross_compiling=maybe
-    $as_echo "$as_me: WARNING: If you wanted to set the --build type, don't use --host.
-    If a cross compiler is detected then cross compile mode will be used." >&2
   elif test "x$build_alias" != "x$host_alias"; then
     cross_compiling=yes
   fi
@@ -1229,11 +1213,9 @@ test "$silent" = yes && exec 6>/dev/null
 ac_pwd=`pwd` && test -n "$ac_pwd" &&
 ac_ls_di=`ls -di .` &&
 ac_pwd_ls_di=`cd "$ac_pwd" && ls -di .` ||
-  { $as_echo "$as_me: error: working directory cannot be determined" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "working directory cannot be determined"
 test "X$ac_ls_di" = "X$ac_pwd_ls_di" ||
-  { $as_echo "$as_me: error: pwd does not report name of working directory" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "pwd does not report name of working directory"
 
 
 # Find the source files, if location was not specified.
@@ -1272,13 +1254,11 @@ else
 fi
 if test ! -r "$srcdir/$ac_unique_file"; then
   test "$ac_srcdir_defaulted" = yes && srcdir="$ac_confdir or .."
-  { $as_echo "$as_me: error: cannot find sources ($ac_unique_file) in $srcdir" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "cannot find sources ($ac_unique_file) in $srcdir"
 fi
 ac_msg="sources are in $srcdir, but \`cd $srcdir' does not work"
 ac_abs_confdir=`(
-	cd "$srcdir" && test -r "./$ac_unique_file" || { $as_echo "$as_me: error: $ac_msg" >&2
-   { (exit 1); exit 1; }; }
+	cd "$srcdir" && test -r "./$ac_unique_file" || as_fn_error $? "$ac_msg"
 	pwd)`
 # When building in place, set srcdir=.
 if test "$ac_abs_confdir" = "$ac_pwd"; then
@@ -1318,7 +1298,7 @@ Configuration:
       --help=short        display options specific to this package
       --help=recursive    display the short help of all the included packages
   -V, --version           display version information and exit
-  -q, --quiet, --silent   do not print \`checking...' messages
+  -q, --quiet, --silent   do not print \`checking ...' messages
       --cache-file=FILE   cache test results in FILE [disabled]
   -C, --config-cache      alias for \`--cache-file=config.cache'
   -n, --no-create         do not create output files
@@ -1378,8 +1358,12 @@ Optional Features:
   --disable-option-checking  ignore unrecognized --enable/--with options
   --disable-FEATURE       do not include FEATURE (same as --enable-FEATURE=no)
   --enable-FEATURE[=ARG]  include FEATURE [ARG=yes]
-  --disable-dependency-tracking  speeds up one-time build
-  --enable-dependency-tracking   do not reject slow dependency extractors
+  --enable-silent-rules   less verbose build output (undo: "make V=1")
+  --disable-silent-rules  verbose build output (undo: "make V=0")
+  --enable-dependency-tracking
+                          do not reject slow dependency extractors
+  --disable-dependency-tracking
+                          speeds up one-time build
   --disable-openmp        do not use OpenMP
   --enable-simd=SSE|AVX|AVX2|AVX512|MIC
                           Select instructions to be SSE4.0, AVX 1.0, AVX
@@ -1392,7 +1376,7 @@ Some influential environment variables:
   LDFLAGS     linker flags, e.g. -L<lib dir> if you have libraries in a
               nonstandard directory <lib dir>
   LIBS        libraries to pass to the linker, e.g. -l<library>
-  CPPFLAGS    C/C++/Objective C preprocessor flags, e.g. -I<include dir> if
+  CPPFLAGS    (Objective) C/C++ preprocessor flags, e.g. -I<include dir> if
               you have headers in a nonstandard directory <include dir>
   CC          C compiler command
   CFLAGS      C compiler flags
@@ -1465,21 +1449,568 @@ test -n "$ac_init_help" && exit $ac_status
 if $ac_init_version; then
   cat <<\_ACEOF
 Grid configure 1.0
-generated by GNU Autoconf 2.63
+generated by GNU Autoconf 2.69
 
-Copyright (C) 1992, 1993, 1994, 1995, 1996, 1998, 1999, 2000, 2001,
-2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
+Copyright (C) 2012 Free Software Foundation, Inc.
 This configure script is free software; the Free Software Foundation
 gives unlimited permission to copy, distribute and modify it.
 _ACEOF
   exit
 fi
+
+## ------------------------ ##
+## Autoconf initialization. ##
+## ------------------------ ##
+
+# ac_fn_cxx_try_compile LINENO
+# ----------------------------
+# Try to compile conftest.$ac_ext, and return whether this succeeded.
+ac_fn_cxx_try_compile ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  rm -f conftest.$ac_objext
+  if { { ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compile") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_retval=1
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_cxx_try_compile
+
+# ac_fn_c_try_compile LINENO
+# --------------------------
+# Try to compile conftest.$ac_ext, and return whether this succeeded.
+ac_fn_c_try_compile ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  rm -f conftest.$ac_objext
+  if { { ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compile") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_retval=1
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_compile
+
+# ac_fn_c_try_link LINENO
+# -----------------------
+# Try to link conftest.$ac_ext, and return whether this succeeded.
+ac_fn_c_try_link ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  rm -f conftest.$ac_objext conftest$ac_exeext
+  if { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 test -x conftest$ac_exeext
+       }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_retval=1
+fi
+  # Delete the IPA/IPO (Inter Procedural Analysis/Optimization) information
+  # created by the PGI compiler (conftest_ipa8_conftest.oo), as it would
+  # interfere with the next link command; also delete a directory that is
+  # left behind by Apple's compiler.  We do this before executing the actions.
+  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_link
+
+# ac_fn_c_try_cpp LINENO
+# ----------------------
+# Try to preprocess conftest.$ac_ext, and return whether this succeeded.
+ac_fn_c_try_cpp ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if { { ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } > conftest.i && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+    ac_retval=1
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_cpp
+
+# ac_fn_c_check_header_mongrel LINENO HEADER VAR INCLUDES
+# -------------------------------------------------------
+# Tests whether HEADER exists, giving a warning if it cannot be compiled using
+# the include files in INCLUDES and setting the cache variable VAR
+# accordingly.
+ac_fn_c_check_header_mongrel ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if eval \${$3+:} false; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 usability" >&5
+$as_echo_n "checking $2 usability... " >&6; }
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+#include <$2>
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_header_compiler=yes
+else
+  ac_header_compiler=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 presence" >&5
+$as_echo_n "checking $2 presence... " >&6; }
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <$2>
+_ACEOF
+if ac_fn_c_try_cpp "$LINENO"; then :
+  ac_header_preproc=yes
+else
+  ac_header_preproc=no
+fi
+rm -f conftest.err conftest.i conftest.$ac_ext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in #((
+  yes:no: )
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $2: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $2:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $2: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
+( $as_echo "## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##"
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  eval "$3=\$ac_header_compiler"
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_header_mongrel
+
+# ac_fn_c_try_run LINENO
+# ----------------------
+# Try to link conftest.$ac_ext, and return whether this succeeded. Assumes
+# that executables *can* be run.
+ac_fn_c_try_run ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && { ac_try='./conftest$ac_exeext'
+  { { case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: program exited with status $ac_status" >&5
+       $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+       ac_retval=$ac_status
+fi
+  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_run
+
+# ac_fn_c_check_header_compile LINENO HEADER VAR INCLUDES
+# -------------------------------------------------------
+# Tests whether HEADER exists and can be compiled using the include files in
+# INCLUDES, setting the cache variable VAR accordingly.
+ac_fn_c_check_header_compile ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+#include <$2>
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  eval "$3=yes"
+else
+  eval "$3=no"
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_header_compile
+
+# ac_fn_c_check_decl LINENO SYMBOL VAR INCLUDES
+# ---------------------------------------------
+# Tests whether SYMBOL is declared in INCLUDES, setting cache variable VAR
+# accordingly.
+ac_fn_c_check_decl ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  as_decl_name=`echo $2|sed 's/ *(.*//'`
+  as_decl_use=`echo $2|sed -e 's/(/((/' -e 's/)/) 0&/' -e 's/,/) 0& (/g'`
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $as_decl_name is declared" >&5
+$as_echo_n "checking whether $as_decl_name is declared... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+#ifndef $as_decl_name
+#ifdef __cplusplus
+  (void) $as_decl_use;
+#else
+  (void) $as_decl_name;
+#endif
+#endif
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  eval "$3=yes"
+else
+  eval "$3=no"
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_decl
+
+# ac_fn_c_check_type LINENO TYPE VAR INCLUDES
+# -------------------------------------------
+# Tests whether TYPE exists after having included INCLUDES, setting cache
+# variable VAR accordingly.
+ac_fn_c_check_type ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  eval "$3=no"
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+if (sizeof ($2))
+	 return 0;
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+if (sizeof (($2)))
+	    return 0;
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+
+else
+  eval "$3=yes"
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_type
+
+# ac_fn_c_find_uintX_t LINENO BITS VAR
+# ------------------------------------
+# Finds an unsigned integer type with width BITS, setting cache variable VAR
+# accordingly.
+ac_fn_c_find_uintX_t ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for uint$2_t" >&5
+$as_echo_n "checking for uint$2_t... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  eval "$3=no"
+     # Order is important - never check a type that is potentially smaller
+     # than half of the expected target width.
+     for ac_type in uint$2_t 'unsigned int' 'unsigned long int' \
+	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
+       cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+static int test_array [1 - 2 * !((($ac_type) -1 >> ($2 / 2 - 1)) >> ($2 / 2 - 1) == 3)];
+test_array [0] = 0;
+return test_array [0];
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  case $ac_type in #(
+  uint$2_t) :
+    eval "$3=yes" ;; #(
+  *) :
+    eval "$3=\$ac_type" ;;
+esac
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+       if eval test \"x\$"$3"\" = x"no"; then :
+
+else
+  break
+fi
+     done
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_find_uintX_t
+
+# ac_fn_c_check_func LINENO FUNC VAR
+# ----------------------------------
+# Tests whether FUNC exists, setting the cache variable VAR accordingly
+ac_fn_c_check_func ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+/* Define $2 to an innocuous variant, in case <limits.h> declares $2.
+   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
+#define $2 innocuous_$2
+
+/* System header to define __stub macros and hopefully few prototypes,
+    which can conflict with char $2 (); below.
+    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
+    <limits.h> exists even on freestanding compilers.  */
+
+#ifdef __STDC__
+# include <limits.h>
+#else
+# include <assert.h>
+#endif
+
+#undef $2
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char $2 ();
+/* The GNU C library defines this for functions which it implements
+    to always fail with ENOSYS.  Some functions are actually named
+    something starting with __ and the normal name is an alias.  */
+#if defined __stub_$2 || defined __stub___$2
+choke me
+#endif
+
+int
+main ()
+{
+return $2 ();
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  eval "$3=yes"
+else
+  eval "$3=no"
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_func
 cat >config.log <<_ACEOF
 This file contains any messages produced by compilers while
 running configure, to aid debugging if configure makes a mistake.
 
 It was created by Grid $as_me 1.0, which was
-generated by GNU Autoconf 2.63.  Invocation command line was
+generated by GNU Autoconf 2.69.  Invocation command line was
 
   $ $0 $@
 
@@ -1515,8 +2046,8 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  $as_echo "PATH: $as_dir"
-done
+    $as_echo "PATH: $as_dir"
+  done
 IFS=$as_save_IFS
 
 } >&5
@@ -1553,9 +2084,9 @@ do
       ac_arg=`$as_echo "$ac_arg" | sed "s/'/'\\\\\\\\''/g"` ;;
     esac
     case $ac_pass in
-    1) ac_configure_args0="$ac_configure_args0 '$ac_arg'" ;;
+    1) as_fn_append ac_configure_args0 " '$ac_arg'" ;;
     2)
-      ac_configure_args1="$ac_configure_args1 '$ac_arg'"
+      as_fn_append ac_configure_args1 " '$ac_arg'"
       if test $ac_must_keep_next = true; then
 	ac_must_keep_next=false # Got value, back to normal.
       else
@@ -1571,13 +2102,13 @@ do
 	  -* ) ac_must_keep_next=true ;;
 	esac
       fi
-      ac_configure_args="$ac_configure_args '$ac_arg'"
+      as_fn_append ac_configure_args " '$ac_arg'"
       ;;
     esac
   done
 done
-$as_unset ac_configure_args0 || test "${ac_configure_args0+set}" != set || { ac_configure_args0=; export ac_configure_args0; }
-$as_unset ac_configure_args1 || test "${ac_configure_args1+set}" != set || { ac_configure_args1=; export ac_configure_args1; }
+{ ac_configure_args0=; unset ac_configure_args0;}
+{ ac_configure_args1=; unset ac_configure_args1;}
 
 # When interrupted or exit'd, cleanup temporary files, and complete
 # config.log.  We remove comments because anyway the quotes in there
@@ -1589,11 +2120,9 @@ trap 'exit_status=$?
   {
     echo
 
-    cat <<\_ASBOX
-## ---------------- ##
+    $as_echo "## ---------------- ##
 ## Cache variables. ##
-## ---------------- ##
-_ASBOX
+## ---------------- ##"
     echo
     # The following way of writing the cache mishandles newlines in values,
 (
@@ -1602,13 +2131,13 @@ _ASBOX
     case $ac_val in #(
     *${as_nl}*)
       case $ac_var in #(
-      *_cv_*) { $as_echo "$as_me:$LINENO: WARNING: cache variable $ac_var contains a newline" >&5
+      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
 $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
       esac
       case $ac_var in #(
       _ | IFS | as_nl) ;; #(
       BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
-      *) $as_unset $ac_var ;;
+      *) { eval $ac_var=; unset $ac_var;} ;;
       esac ;;
     esac
   done
@@ -1627,11 +2156,9 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
 )
     echo
 
-    cat <<\_ASBOX
-## ----------------- ##
+    $as_echo "## ----------------- ##
 ## Output variables. ##
-## ----------------- ##
-_ASBOX
+## ----------------- ##"
     echo
     for ac_var in $ac_subst_vars
     do
@@ -1644,11 +2171,9 @@ _ASBOX
     echo
 
     if test -n "$ac_subst_files"; then
-      cat <<\_ASBOX
-## ------------------- ##
+      $as_echo "## ------------------- ##
 ## File substitutions. ##
-## ------------------- ##
-_ASBOX
+## ------------------- ##"
       echo
       for ac_var in $ac_subst_files
       do
@@ -1662,11 +2187,9 @@ _ASBOX
     fi
 
     if test -s confdefs.h; then
-      cat <<\_ASBOX
-## ----------- ##
+      $as_echo "## ----------- ##
 ## confdefs.h. ##
-## ----------- ##
-_ASBOX
+## ----------- ##"
       echo
       cat confdefs.h
       echo
@@ -1680,46 +2203,53 @@ _ASBOX
     exit $exit_status
 ' 0
 for ac_signal in 1 2 13 15; do
-  trap 'ac_signal='$ac_signal'; { (exit 1); exit 1; }' $ac_signal
+  trap 'ac_signal='$ac_signal'; as_fn_exit 1' $ac_signal
 done
 ac_signal=0
 
 # confdefs.h avoids OS command line length limits that DEFS can exceed.
 rm -f -r conftest* confdefs.h
 
+$as_echo "/* confdefs.h */" > confdefs.h
+
 # Predefined preprocessor variables.
 
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_NAME "$PACKAGE_NAME"
 _ACEOF
 
-
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_TARNAME "$PACKAGE_TARNAME"
 _ACEOF
 
-
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_VERSION "$PACKAGE_VERSION"
 _ACEOF
 
-
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_STRING "$PACKAGE_STRING"
 _ACEOF
 
-
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_BUGREPORT "$PACKAGE_BUGREPORT"
 _ACEOF
 
+cat >>confdefs.h <<_ACEOF
+#define PACKAGE_URL "$PACKAGE_URL"
+_ACEOF
+
 
 # Let the site file select an alternate cache file if it wants to.
 # Prefer an explicitly selected file to automatically selected ones.
 ac_site_file1=NONE
 ac_site_file2=NONE
 if test -n "$CONFIG_SITE"; then
-  ac_site_file1=$CONFIG_SITE
+  # We do not want a PATH search for config.site.
+  case $CONFIG_SITE in #((
+    -*)  ac_site_file1=./$CONFIG_SITE;;
+    */*) ac_site_file1=$CONFIG_SITE;;
+    *)   ac_site_file1=./$CONFIG_SITE;;
+  esac
 elif test "x$prefix" != xNONE; then
   ac_site_file1=$prefix/share/config.site
   ac_site_file2=$prefix/etc/config.site
@@ -1730,19 +2260,23 @@ fi
 for ac_site_file in "$ac_site_file1" "$ac_site_file2"
 do
   test "x$ac_site_file" = xNONE && continue
-  if test -r "$ac_site_file"; then
-    { $as_echo "$as_me:$LINENO: loading site script $ac_site_file" >&5
+  if test /dev/null != "$ac_site_file" && test -r "$ac_site_file"; then
+    { $as_echo "$as_me:${as_lineno-$LINENO}: loading site script $ac_site_file" >&5
 $as_echo "$as_me: loading site script $ac_site_file" >&6;}
     sed 's/^/| /' "$ac_site_file" >&5
-    . "$ac_site_file"
+    . "$ac_site_file" \
+      || { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "failed to load site script $ac_site_file
+See \`config.log' for more details" "$LINENO" 5; }
   fi
 done
 
 if test -r "$cache_file"; then
-  # Some versions of bash will fail to source /dev/null (special
-  # files actually), so we avoid doing that.
-  if test -f "$cache_file"; then
-    { $as_echo "$as_me:$LINENO: loading cache $cache_file" >&5
+  # Some versions of bash will fail to source /dev/null (special files
+  # actually), so we avoid doing that.  DJGPP emulates it as a regular file.
+  if test /dev/null != "$cache_file" && test -f "$cache_file"; then
+    { $as_echo "$as_me:${as_lineno-$LINENO}: loading cache $cache_file" >&5
 $as_echo "$as_me: loading cache $cache_file" >&6;}
     case $cache_file in
       [\\/]* | ?:[\\/]* ) . "$cache_file";;
@@ -1750,7 +2284,7 @@ $as_echo "$as_me: loading cache $cache_file" >&6;}
     esac
   fi
 else
-  { $as_echo "$as_me:$LINENO: creating cache $cache_file" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: creating cache $cache_file" >&5
 $as_echo "$as_me: creating cache $cache_file" >&6;}
   >$cache_file
 fi
@@ -1765,11 +2299,11 @@ for ac_var in $ac_precious_vars; do
   eval ac_new_val=\$ac_env_${ac_var}_value
   case $ac_old_set,$ac_new_set in
     set,)
-      { $as_echo "$as_me:$LINENO: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
+      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
 $as_echo "$as_me: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&2;}
       ac_cache_corrupted=: ;;
     ,set)
-      { $as_echo "$as_me:$LINENO: error: \`$ac_var' was not set in the previous run" >&5
+      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was not set in the previous run" >&5
 $as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
       ac_cache_corrupted=: ;;
     ,);;
@@ -1779,17 +2313,17 @@ $as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
 	ac_old_val_w=`echo x $ac_old_val`
 	ac_new_val_w=`echo x $ac_new_val`
 	if test "$ac_old_val_w" != "$ac_new_val_w"; then
-	  { $as_echo "$as_me:$LINENO: error: \`$ac_var' has changed since the previous run:" >&5
+	  { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' has changed since the previous run:" >&5
 $as_echo "$as_me: error: \`$ac_var' has changed since the previous run:" >&2;}
 	  ac_cache_corrupted=:
 	else
-	  { $as_echo "$as_me:$LINENO: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
+	  { $as_echo "$as_me:${as_lineno-$LINENO}: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
 $as_echo "$as_me: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&2;}
 	  eval $ac_var=\$ac_old_val
 	fi
-	{ $as_echo "$as_me:$LINENO:   former value:  \`$ac_old_val'" >&5
+	{ $as_echo "$as_me:${as_lineno-$LINENO}:   former value:  \`$ac_old_val'" >&5
 $as_echo "$as_me:   former value:  \`$ac_old_val'" >&2;}
-	{ $as_echo "$as_me:$LINENO:   current value: \`$ac_new_val'" >&5
+	{ $as_echo "$as_me:${as_lineno-$LINENO}:   current value: \`$ac_new_val'" >&5
 $as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
       fi;;
   esac
@@ -1801,43 +2335,20 @@ $as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
     esac
     case " $ac_configure_args " in
       *" '$ac_arg' "*) ;; # Avoid dups.  Use of quotes ensures accuracy.
-      *) ac_configure_args="$ac_configure_args '$ac_arg'" ;;
+      *) as_fn_append ac_configure_args " '$ac_arg'" ;;
     esac
   fi
 done
 if $ac_cache_corrupted; then
-  { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-  { $as_echo "$as_me:$LINENO: error: changes in the environment can compromise the build" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: error: changes in the environment can compromise the build" >&5
 $as_echo "$as_me: error: changes in the environment can compromise the build" >&2;}
-  { { $as_echo "$as_me:$LINENO: error: run \`make distclean' and/or \`rm $cache_file' and start over" >&5
-$as_echo "$as_me: error: run \`make distclean' and/or \`rm $cache_file' and start over" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "run \`make distclean' and/or \`rm $cache_file' and start over" "$LINENO" 5
 fi
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+## -------------------- ##
+## Main body of script. ##
+## -------------------- ##
 
 ac_ext=c
 ac_cpp='$CPP $CPPFLAGS'
@@ -1846,7 +2357,7 @@ ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
 
-am__api_version='1.11'
+am__api_version='1.15'
 
 ac_aux_dir=
 for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
@@ -1865,9 +2376,7 @@ for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
   fi
 done
 if test -z "$ac_aux_dir"; then
-  { { $as_echo "$as_me:$LINENO: error: cannot find install-sh or install.sh in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" >&5
-$as_echo "$as_me: error: cannot find install-sh or install.sh in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "cannot find install-sh, install.sh, or shtool in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" "$LINENO" 5
 fi
 
 # These three variables are undocumented and unsupported,
@@ -1893,10 +2402,10 @@ ac_configure="$SHELL $ac_aux_dir/configure"  # Please don't use this var.
 # OS/2's system install, which has a completely different semantic
 # ./install, which can be erroneously created by make from ./install.sh.
 # Reject install programs that cannot install multiple files.
-{ $as_echo "$as_me:$LINENO: checking for a BSD-compatible install" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a BSD-compatible install" >&5
 $as_echo_n "checking for a BSD-compatible install... " >&6; }
 if test -z "$INSTALL"; then
-if test "${ac_cv_path_install+set}" = set; then
+if ${ac_cv_path_install+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
@@ -1904,11 +2413,11 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  # Account for people who put trailing slashes in PATH elements.
-case $as_dir/ in
-  ./ | .// | /cC/* | \
+    # Account for people who put trailing slashes in PATH elements.
+case $as_dir/ in #((
+  ./ | .// | /[cC]/* | \
   /etc/* | /usr/sbin/* | /usr/etc/* | /sbin/* | /usr/afsws/bin/* | \
-  ?:\\/os2\\/install\\/* | ?:\\/OS2\\/INSTALL\\/* | \
+  ?:[\\/]os2[\\/]install[\\/]* | ?:[\\/]OS2[\\/]INSTALL[\\/]* | \
   /usr/ucb/* ) ;;
   *)
     # OSF1 and SCO ODT 3.0 have their own names for install.
@@ -1916,7 +2425,7 @@ case $as_dir/ in
     # by default.
     for ac_prog in ginstall scoinst install; do
       for ac_exec_ext in '' $ac_executable_extensions; do
-	if { test -f "$as_dir/$ac_prog$ac_exec_ext" && $as_test_x "$as_dir/$ac_prog$ac_exec_ext"; }; then
+	if as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext"; then
 	  if test $ac_prog = install &&
 	    grep dspmsg "$as_dir/$ac_prog$ac_exec_ext" >/dev/null 2>&1; then
 	    # AIX install.  It has an incompatible calling convention.
@@ -1945,7 +2454,7 @@ case $as_dir/ in
     ;;
 esac
 
-done
+  done
 IFS=$as_save_IFS
 
 rm -rf conftest.one conftest.two conftest.dir
@@ -1961,7 +2470,7 @@ fi
     INSTALL=$ac_install_sh
   fi
 fi
-{ $as_echo "$as_me:$LINENO: result: $INSTALL" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $INSTALL" >&5
 $as_echo "$INSTALL" >&6; }
 
 # Use test -z because SunOS4 sh mishandles braces in ${var-val}.
@@ -1972,68 +2481,73 @@ test -z "$INSTALL_SCRIPT" && INSTALL_SCRIPT='${INSTALL}'
 
 test -z "$INSTALL_DATA" && INSTALL_DATA='${INSTALL} -m 644'
 
-{ $as_echo "$as_me:$LINENO: checking whether build environment is sane" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether build environment is sane" >&5
 $as_echo_n "checking whether build environment is sane... " >&6; }
-# Just in case
-sleep 1
-echo timestamp > conftest.file
 # Reject unsafe characters in $srcdir or the absolute working directory
 # name.  Accept space and tab only in the latter.
 am_lf='
 '
 case `pwd` in
   *[\\\"\#\$\&\'\`$am_lf]*)
-    { { $as_echo "$as_me:$LINENO: error: unsafe absolute working directory name" >&5
-$as_echo "$as_me: error: unsafe absolute working directory name" >&2;}
-   { (exit 1); exit 1; }; };;
+    as_fn_error $? "unsafe absolute working directory name" "$LINENO" 5;;
 esac
 case $srcdir in
   *[\\\"\#\$\&\'\`$am_lf\ \	]*)
-    { { $as_echo "$as_me:$LINENO: error: unsafe srcdir value: \`$srcdir'" >&5
-$as_echo "$as_me: error: unsafe srcdir value: \`$srcdir'" >&2;}
-   { (exit 1); exit 1; }; };;
+    as_fn_error $? "unsafe srcdir value: '$srcdir'" "$LINENO" 5;;
 esac
 
-# Do `set' in a subshell so we don't clobber the current shell's
+# Do 'set' in a subshell so we don't clobber the current shell's
 # arguments.  Must try -L first in case configure is actually a
 # symlink; some systems play weird games with the mod time of symlinks
 # (eg FreeBSD returns the mod time of the symlink's containing
 # directory).
 if (
-   set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
-   if test "$*" = "X"; then
-      # -L didn't work.
-      set X `ls -t "$srcdir/configure" conftest.file`
-   fi
-   rm -f conftest.file
-   if test "$*" != "X $srcdir/configure conftest.file" \
-      && test "$*" != "X conftest.file $srcdir/configure"; then
-
-      # If neither matched, then we have a broken ls.  This can happen
-      # if, for instance, CONFIG_SHELL is bash and it inherits a
-      # broken ls alias from the environment.  This has actually
-      # happened.  Such a system could not be considered "sane".
-      { { $as_echo "$as_me:$LINENO: error: ls -t appears to fail.  Make sure there is not a broken
-alias in your environment" >&5
-$as_echo "$as_me: error: ls -t appears to fail.  Make sure there is not a broken
-alias in your environment" >&2;}
-   { (exit 1); exit 1; }; }
-   fi
+   am_has_slept=no
+   for am_try in 1 2; do
+     echo "timestamp, slept: $am_has_slept" > conftest.file
+     set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
+     if test "$*" = "X"; then
+	# -L didn't work.
+	set X `ls -t "$srcdir/configure" conftest.file`
+     fi
+     if test "$*" != "X $srcdir/configure conftest.file" \
+	&& test "$*" != "X conftest.file $srcdir/configure"; then
 
+	# If neither matched, then we have a broken ls.  This can happen
+	# if, for instance, CONFIG_SHELL is bash and it inherits a
+	# broken ls alias from the environment.  This has actually
+	# happened.  Such a system could not be considered "sane".
+	as_fn_error $? "ls -t appears to fail.  Make sure there is not a broken
+  alias in your environment" "$LINENO" 5
+     fi
+     if test "$2" = conftest.file || test $am_try -eq 2; then
+       break
+     fi
+     # Just in case.
+     sleep 1
+     am_has_slept=yes
+   done
    test "$2" = conftest.file
    )
 then
    # Ok.
    :
 else
-   { { $as_echo "$as_me:$LINENO: error: newly created file is older than distributed files!
-Check your system clock" >&5
-$as_echo "$as_me: error: newly created file is older than distributed files!
-Check your system clock" >&2;}
-   { (exit 1); exit 1; }; }
+   as_fn_error $? "newly created file is older than distributed files!
+Check your system clock" "$LINENO" 5
 fi
-{ $as_echo "$as_me:$LINENO: result: yes" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
 $as_echo "yes" >&6; }
+# If we didn't sleep, we still need to ensure time stamps of config.status and
+# generated files are strictly newer.
+am_sleep_pid=
+if grep 'slept: no' conftest.file >/dev/null 2>&1; then
+  ( sleep 1 ) &
+  am_sleep_pid=$!
+fi
+
+rm -f conftest.file
+
 test "$program_prefix" != NONE &&
   program_transform_name="s&^&$program_prefix&;$program_transform_name"
 # Use a double $ so make ignores it.
@@ -2044,8 +2558,8 @@ test "$program_suffix" != NONE &&
 ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
 program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
 
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 
 if test x"${MISSING+set}" != xset; then
   case $am_aux_dir in
@@ -2056,15 +2570,15 @@ if test x"${MISSING+set}" != xset; then
   esac
 fi
 # Use eval to expand $SHELL
-if eval "$MISSING --run true"; then
-  am_missing_run="$MISSING --run "
+if eval "$MISSING --is-lightweight"; then
+  am_missing_run="$MISSING "
 else
   am_missing_run=
-  { $as_echo "$as_me:$LINENO: WARNING: \`missing' script is too old or missing" >&5
-$as_echo "$as_me: WARNING: \`missing' script is too old or missing" >&2;}
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: 'missing' script is too old or missing" >&5
+$as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
 fi
 
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -2073,17 +2587,17 @@ if test x"${install_sh}" != xset; then
   esac
 fi
 
-# Installed binaries are usually stripped using `strip' when the user
-# run `make install-strip'.  However `strip' might not be the right
+# Installed binaries are usually stripped using 'strip' when the user
+# run "make install-strip".  However 'strip' might not be the right
 # tool to use in cross-compilation environments, therefore Automake
-# will honor the `STRIP' environment variable to overrule this program.
+# will honor the 'STRIP' environment variable to overrule this program.
 if test "$cross_compiling" != no; then
   if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}strip", so it can be a program name with args.
 set dummy ${ac_tool_prefix}strip; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_STRIP+set}" = set; then
+if ${ac_cv_prog_STRIP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$STRIP"; then
@@ -2094,24 +2608,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_STRIP="${ac_tool_prefix}strip"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 STRIP=$ac_cv_prog_STRIP
 if test -n "$STRIP"; then
-  { $as_echo "$as_me:$LINENO: result: $STRIP" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $STRIP" >&5
 $as_echo "$STRIP" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2121,9 +2635,9 @@ if test -z "$ac_cv_prog_STRIP"; then
   ac_ct_STRIP=$STRIP
   # Extract the first word of "strip", so it can be a program name with args.
 set dummy strip; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_ac_ct_STRIP+set}" = set; then
+if ${ac_cv_prog_ac_ct_STRIP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_STRIP"; then
@@ -2134,24 +2648,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_ac_ct_STRIP="strip"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_STRIP=$ac_cv_prog_ac_ct_STRIP
 if test -n "$ac_ct_STRIP"; then
-  { $as_echo "$as_me:$LINENO: result: $ac_ct_STRIP" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_STRIP" >&5
 $as_echo "$ac_ct_STRIP" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2160,7 +2674,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -2173,10 +2687,10 @@ fi
 fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 
-{ $as_echo "$as_me:$LINENO: checking for a thread-safe mkdir -p" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a thread-safe mkdir -p" >&5
 $as_echo_n "checking for a thread-safe mkdir -p... " >&6; }
 if test -z "$MKDIR_P"; then
-  if test "${ac_cv_path_mkdir+set}" = set; then
+  if ${ac_cv_path_mkdir+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
@@ -2184,9 +2698,9 @@ for as_dir in $PATH$PATH_SEPARATOR/opt/sfw/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_prog in mkdir gmkdir; do
+    for ac_prog in mkdir gmkdir; do
 	 for ac_exec_ext in '' $ac_executable_extensions; do
-	   { test -f "$as_dir/$ac_prog$ac_exec_ext" && $as_test_x "$as_dir/$ac_prog$ac_exec_ext"; } || continue
+	   as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext" || continue
 	   case `"$as_dir/$ac_prog$ac_exec_ext" --version 2>&1` in #(
 	     'mkdir (GNU coreutils) '* | \
 	     'mkdir (coreutils) '* | \
@@ -2196,11 +2710,12 @@ do
 	   esac
 	 done
        done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 
+  test -d ./--version && rmdir ./--version
   if test "${ac_cv_path_mkdir+set}" = set; then
     MKDIR_P="$ac_cv_path_mkdir -p"
   else
@@ -2208,26 +2723,19 @@ fi
     # value for MKDIR_P within a source directory, because that will
     # break other packages using the cache if that directory is
     # removed, or if the value is a relative name.
-    test -d ./--version && rmdir ./--version
     MKDIR_P="$ac_install_sh -d"
   fi
 fi
-{ $as_echo "$as_me:$LINENO: result: $MKDIR_P" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $MKDIR_P" >&5
 $as_echo "$MKDIR_P" >&6; }
 
-mkdir_p="$MKDIR_P"
-case $mkdir_p in
-  [\\/$]* | ?:[\\/]*) ;;
-  */*) mkdir_p="\$(top_builddir)/$mkdir_p" ;;
-esac
-
 for ac_prog in gawk mawk nawk awk
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_AWK+set}" = set; then
+if ${ac_cv_prog_AWK+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$AWK"; then
@@ -2238,24 +2746,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_AWK="$ac_prog"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 AWK=$ac_cv_prog_AWK
 if test -n "$AWK"; then
-  { $as_echo "$as_me:$LINENO: result: $AWK" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $AWK" >&5
 $as_echo "$AWK" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2263,11 +2771,11 @@ fi
   test -n "$AWK" && break
 done
 
-{ $as_echo "$as_me:$LINENO: checking whether ${MAKE-make} sets \$(MAKE)" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether ${MAKE-make} sets \$(MAKE)" >&5
 $as_echo_n "checking whether ${MAKE-make} sets \$(MAKE)... " >&6; }
 set x ${MAKE-make}
 ac_make=`$as_echo "$2" | sed 's/+/p/g; s/[^a-zA-Z0-9_]/_/g'`
-if { as_var=ac_cv_prog_make_${ac_make}_set; eval "test \"\${$as_var+set}\" = set"; }; then
+if eval \${ac_cv_prog_make_${ac_make}_set+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   cat >conftest.make <<\_ACEOF
@@ -2275,7 +2783,7 @@ SHELL = /bin/sh
 all:
 	@echo '@@@%%%=$(MAKE)=@@@%%%'
 _ACEOF
-# GNU make sometimes prints "make[1]: Entering...", which would confuse us.
+# GNU make sometimes prints "make[1]: Entering ...", which would confuse us.
 case `${MAKE-make} -f conftest.make 2>/dev/null` in
   *@@@%%%=?*=@@@%%%*)
     eval ac_cv_prog_make_${ac_make}_set=yes;;
@@ -2285,11 +2793,11 @@ esac
 rm -f conftest.make
 fi
 if eval test \$ac_cv_prog_make_${ac_make}_set = yes; then
-  { $as_echo "$as_me:$LINENO: result: yes" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
 $as_echo "yes" >&6; }
   SET_MAKE=
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
   SET_MAKE="MAKE=${MAKE-make}"
 fi
@@ -2303,15 +2811,52 @@ else
 fi
 rmdir .tst 2>/dev/null
 
+# Check whether --enable-silent-rules was given.
+if test "${enable_silent_rules+set}" = set; then :
+  enableval=$enable_silent_rules;
+fi
+
+case $enable_silent_rules in # (((
+  yes) AM_DEFAULT_VERBOSITY=0;;
+   no) AM_DEFAULT_VERBOSITY=1;;
+    *) AM_DEFAULT_VERBOSITY=1;;
+esac
+am_make=${MAKE-make}
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $am_make supports nested variables" >&5
+$as_echo_n "checking whether $am_make supports nested variables... " >&6; }
+if ${am_cv_make_support_nested_variables+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if $as_echo 'TRUE=$(BAR$(V))
+BAR0=false
+BAR1=true
+V=1
+am__doit:
+	@$(TRUE)
+.PHONY: am__doit' | $am_make -f - >/dev/null 2>&1; then
+  am_cv_make_support_nested_variables=yes
+else
+  am_cv_make_support_nested_variables=no
+fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_make_support_nested_variables" >&5
+$as_echo "$am_cv_make_support_nested_variables" >&6; }
+if test $am_cv_make_support_nested_variables = yes; then
+    AM_V='$(V)'
+  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
+else
+  AM_V=$AM_DEFAULT_VERBOSITY
+  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
+fi
+AM_BACKSLASH='\'
+
 if test "`cd $srcdir && pwd`" != "`pwd`"; then
   # Use -I$(srcdir) only when $(srcdir) != ., so that make's output
   # is not polluted with repeated "-I."
   am__isrc=' -I$(srcdir)'
   # test to see if srcdir already configured
   if test -f $srcdir/config.status; then
-    { { $as_echo "$as_me:$LINENO: error: source directory already configured; run \"make distclean\" there first" >&5
-$as_echo "$as_me: error: source directory already configured; run \"make distclean\" there first" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "source directory already configured; run \"make distclean\" there first" "$LINENO" 5
   fi
 fi
 
@@ -2355,19 +2900,72 @@ AUTOHEADER=${AUTOHEADER-"${am_missing_run}autoheader"}
 
 MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
-# Always define AMTAR for backward compatibility.
+# For better backward compatibility.  To be removed once Automake 1.9.x
+# dies out for good.  For more background, see:
+# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
+# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
+mkdir_p='$(MKDIR_P)'
 
-AMTAR=${AMTAR-"${am_missing_run}tar"}
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
+# Always define AMTAR for backward compatibility.  Yes, it's still used
+# in the wild :-(  We should find a proper way to deprecate it ...
+AMTAR='$${TAR-tar}'
 
-am__tar='${AMTAR} chof - "$$tardir"'; am__untar='${AMTAR} xf -'
+
+# We'll loop over all known methods to create a tar archive until one works.
+_am_tools='gnutar  pax cpio none'
+
+am__tar='$${TAR-tar} chof - "$$tardir"' am__untar='$${TAR-tar} xf -'
 
 
 
 
 
 
+# POSIX will say in a future version that running "rm -f" with no argument
+# is OK; and we want to be able to make that assumption in our Makefile
+# recipes.  So use an aggressive probe to check that the usage we want is
+# actually supported "in the wild" to an acceptable degree.
+# See automake bug#10828.
+# To make any issue more visible, cause the running configure to be aborted
+# by default if the 'rm' program in use doesn't match our expectations; the
+# user can still override this though.
+if rm -f && rm -fr && rm -rf; then : OK; else
+  cat >&2 <<'END'
+Oops!
+
+Your 'rm' program seems unable to run without file operands specified
+on the command line, even when the '-f' option is present.  This is contrary
+to the behaviour of most rm programs out there, and not conforming with
+the upcoming POSIX standard: <http://austingroupbugs.net/view.php?id=542>
+
+Please tell bug-automake@gnu.org about your system, including the value
+of your $PATH and any error possibly output before this message.  This
+can help us improve future automake versions.
+
+END
+  if test x"$ACCEPT_INFERIOR_RM_PROGRAM" = x"yes"; then
+    echo 'Configuration will proceed anyway, since you have set the' >&2
+    echo 'ACCEPT_INFERIOR_RM_PROGRAM variable to "yes"' >&2
+    echo >&2
+  else
+    cat >&2 <<'END'
+Aborting the configuration process, to ensure you take notice of the issue.
+
+You can download and install GNU coreutils to get an 'rm' implementation
+that behaves properly: <http://www.gnu.org/software/coreutils/>.
+
+If you want to complete the configuration process using your problematic
+'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
+to "yes", and re-run configure.
+
+END
+    as_fn_error $? "Your 'rm' program is bad, sorry." "$LINENO" 5
+  fi
+fi
+
+
 
 ac_config_headers="$ac_config_headers lib/Grid_config.h"
 
@@ -2387,9 +2985,9 @@ if test -z "$CXX"; then
   do
     # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
 set dummy $ac_tool_prefix$ac_prog; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_CXX+set}" = set; then
+if ${ac_cv_prog_CXX+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CXX"; then
@@ -2400,24 +2998,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_CXX="$ac_tool_prefix$ac_prog"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 CXX=$ac_cv_prog_CXX
 if test -n "$CXX"; then
-  { $as_echo "$as_me:$LINENO: result: $CXX" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CXX" >&5
 $as_echo "$CXX" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2431,9 +3029,9 @@ if test -z "$CXX"; then
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_ac_ct_CXX+set}" = set; then
+if ${ac_cv_prog_ac_ct_CXX+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CXX"; then
@@ -2444,24 +3042,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_ac_ct_CXX="$ac_prog"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CXX=$ac_cv_prog_ac_ct_CXX
 if test -n "$ac_ct_CXX"; then
-  { $as_echo "$as_me:$LINENO: result: $ac_ct_CXX" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CXX" >&5
 $as_echo "$ac_ct_CXX" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2474,7 +3072,7 @@ done
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -2485,48 +3083,31 @@ fi
   fi
 fi
 # Provide some information about the compiler.
-$as_echo "$as_me:$LINENO: checking for C++ compiler version" >&5
+$as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler version" >&5
 set X $ac_compile
 ac_compiler=$2
-{ (ac_try="$ac_compiler --version >&5"
+for ac_option in --version -v -V -qversion; do
+  { { ac_try="$ac_compiler $ac_option >&5"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler --version >&5") 2>&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
-{ (ac_try="$ac_compiler -v >&5"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler -v >&5") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
-{ (ac_try="$ac_compiler -V >&5"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler -V >&5") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
+  if test -s conftest.err; then
+    sed '10a\
+... rest of stderr output deleted ...
+         10q' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+  fi
+  rm -f conftest.er1 conftest.err
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }
+done
 
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2542,8 +3123,8 @@ ac_clean_files="$ac_clean_files a.out a.out.dSYM a.exe b.out"
 # Try to create an executable without -o first, disregard a.out.
 # It will help us diagnose broken compilers, and finding out an intuition
 # of exeext.
-{ $as_echo "$as_me:$LINENO: checking for C++ compiler default output file name" >&5
-$as_echo_n "checking for C++ compiler default output file name... " >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether the C++ compiler works" >&5
+$as_echo_n "checking whether the C++ compiler works... " >&6; }
 ac_link_default=`$as_echo "$ac_link" | sed 's/ -o *conftest[^ ]*//'`
 
 # The possible output files:
@@ -2559,17 +3140,17 @@ do
 done
 rm -f $ac_rmfiles
 
-if { (ac_try="$ac_link_default"
+if { { ac_try="$ac_link_default"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
   (eval "$ac_link_default") 2>&5
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; then
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then :
   # Autoconf-2.13 could set the ac_cv_exeext variable to `no'.
 # So ignore a value of `no', otherwise this would lead to `EXEEXT = no'
 # in a Makefile.  We should not override ac_cv_exeext if it was cached,
@@ -2586,7 +3167,7 @@ do
 	# certainly right.
 	break;;
     *.* )
-        if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
+	if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
 	then :; else
 	   ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
 	fi
@@ -2605,84 +3186,41 @@ test "$ac_cv_exeext" = no && ac_cv_exeext=
 else
   ac_file=''
 fi
-
-{ $as_echo "$as_me:$LINENO: result: $ac_file" >&5
-$as_echo "$ac_file" >&6; }
-if test -z "$ac_file"; then
-  $as_echo "$as_me: failed program was:" >&5
+if test -z "$ac_file"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+$as_echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
 
-{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: C++ compiler cannot create executables
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: C++ compiler cannot create executables
-See \`config.log' for more details." >&2;}
-   { (exit 77); exit 77; }; }; }
-fi
-
-ac_exeext=$ac_cv_exeext
-
-# Check that the compiler produces executables we can run.  If not, either
-# the compiler is broken, or we cross compile.
-{ $as_echo "$as_me:$LINENO: checking whether the C++ compiler works" >&5
-$as_echo_n "checking whether the C++ compiler works... " >&6; }
-# FIXME: These cross compiler hacks should be removed for Autoconf 3.0
-# If not cross compiling, check that we can run a simple program.
-if test "$cross_compiling" != yes; then
-  if { ac_try='./$ac_file'
-  { (case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; }; then
-    cross_compiling=no
-  else
-    if test "$cross_compiling" = maybe; then
-	cross_compiling=yes
-    else
-	{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: cannot run C++ compiled programs.
-If you meant to cross compile, use \`--host'.
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: cannot run C++ compiled programs.
-If you meant to cross compile, use \`--host'.
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
-    fi
-  fi
-fi
-{ $as_echo "$as_me:$LINENO: result: yes" >&5
+as_fn_error 77 "C++ compiler cannot create executables
+See \`config.log' for more details" "$LINENO" 5; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
 $as_echo "yes" >&6; }
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler default output file name" >&5
+$as_echo_n "checking for C++ compiler default output file name... " >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_file" >&5
+$as_echo "$ac_file" >&6; }
+ac_exeext=$ac_cv_exeext
 
 rm -f -r a.out a.out.dSYM a.exe conftest$ac_cv_exeext b.out
 ac_clean_files=$ac_clean_files_save
-# Check that the compiler produces executables we can run.  If not, either
-# the compiler is broken, or we cross compile.
-{ $as_echo "$as_me:$LINENO: checking whether we are cross compiling" >&5
-$as_echo_n "checking whether we are cross compiling... " >&6; }
-{ $as_echo "$as_me:$LINENO: result: $cross_compiling" >&5
-$as_echo "$cross_compiling" >&6; }
-
-{ $as_echo "$as_me:$LINENO: checking for suffix of executables" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of executables" >&5
 $as_echo_n "checking for suffix of executables... " >&6; }
-if { (ac_try="$ac_link"
+if { { ac_try="$ac_link"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
   (eval "$ac_link") 2>&5
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; then
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then :
   # If both `conftest.exe' and `conftest' are `present' (well, observable)
 # catch `conftest.exe'.  For instance with Cygwin, `ls conftest' will
 # work properly (i.e., refer to `conftest.exe'), while it won't with
@@ -2697,32 +3235,83 @@ for ac_file in conftest.exe conftest conftest.*; do
   esac
 done
 else
-  { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: cannot compute suffix of executables: cannot compile and link
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: cannot compute suffix of executables: cannot compile and link
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
+as_fn_error $? "cannot compute suffix of executables: cannot compile and link
+See \`config.log' for more details" "$LINENO" 5; }
 fi
-
-rm -f conftest$ac_cv_exeext
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_exeext" >&5
+rm -f conftest conftest$ac_cv_exeext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_exeext" >&5
 $as_echo "$ac_cv_exeext" >&6; }
 
 rm -f conftest.$ac_ext
 EXEEXT=$ac_cv_exeext
 ac_exeext=$EXEEXT
-{ $as_echo "$as_me:$LINENO: checking for suffix of object files" >&5
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <stdio.h>
+int
+main ()
+{
+FILE *f = fopen ("conftest.out", "w");
+ return ferror (f) || fclose (f) != 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+ac_clean_files="$ac_clean_files conftest.out"
+# Check that the compiler produces executables we can run.  If not, either
+# the compiler is broken, or we cross compile.
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are cross compiling" >&5
+$as_echo_n "checking whether we are cross compiling... " >&6; }
+if test "$cross_compiling" != yes; then
+  { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }
+  if { ac_try='./conftest$ac_cv_exeext'
+  { { case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; }; then
+    cross_compiling=no
+  else
+    if test "$cross_compiling" = maybe; then
+	cross_compiling=yes
+    else
+	{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "cannot run C++ compiled programs.
+If you meant to cross compile, use \`--host'.
+See \`config.log' for more details" "$LINENO" 5; }
+    fi
+  fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $cross_compiling" >&5
+$as_echo "$cross_compiling" >&6; }
+
+rm -f conftest.$ac_ext conftest$ac_cv_exeext conftest.out
+ac_clean_files=$ac_clean_files_save
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of object files" >&5
 $as_echo_n "checking for suffix of object files... " >&6; }
-if test "${ac_cv_objext+set}" = set; then
+if ${ac_cv_objext+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2734,17 +3323,17 @@ main ()
 }
 _ACEOF
 rm -f conftest.o conftest.obj
-if { (ac_try="$ac_compile"
+if { { ac_try="$ac_compile"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
   (eval "$ac_compile") 2>&5
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; then
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then :
   for ac_file in conftest.o conftest.obj conftest.*; do
   test -f "$ac_file" || continue;
   case $ac_file in
@@ -2757,31 +3346,23 @@ else
   $as_echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
 
-{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: cannot compute suffix of object files: cannot compile
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: cannot compute suffix of object files: cannot compile
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
+as_fn_error $? "cannot compute suffix of object files: cannot compile
+See \`config.log' for more details" "$LINENO" 5; }
 fi
-
 rm -f conftest.$ac_cv_objext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_objext" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_objext" >&5
 $as_echo "$ac_cv_objext" >&6; }
 OBJEXT=$ac_cv_objext
 ac_objext=$OBJEXT
-{ $as_echo "$as_me:$LINENO: checking whether we are using the GNU C++ compiler" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C++ compiler" >&5
 $as_echo_n "checking whether we are using the GNU C++ compiler... " >&6; }
-if test "${ac_cv_cxx_compiler_gnu+set}" = set; then
+if ${ac_cv_cxx_compiler_gnu+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2795,37 +3376,16 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_cxx_try_compile "$LINENO"; then :
   ac_compiler_gnu=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_compiler_gnu=no
+  ac_compiler_gnu=no
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 ac_cv_cxx_compiler_gnu=$ac_compiler_gnu
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_cxx_compiler_gnu" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_cxx_compiler_gnu" >&5
 $as_echo "$ac_cv_cxx_compiler_gnu" >&6; }
 if test $ac_compiler_gnu = yes; then
   GXX=yes
@@ -2834,20 +3394,16 @@ else
 fi
 ac_test_CXXFLAGS=${CXXFLAGS+set}
 ac_save_CXXFLAGS=$CXXFLAGS
-{ $as_echo "$as_me:$LINENO: checking whether $CXX accepts -g" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX accepts -g" >&5
 $as_echo_n "checking whether $CXX accepts -g... " >&6; }
-if test "${ac_cv_prog_cxx_g+set}" = set; then
+if ${ac_cv_prog_cxx_g+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   ac_save_cxx_werror_flag=$ac_cxx_werror_flag
    ac_cxx_werror_flag=yes
    ac_cv_prog_cxx_g=no
    CXXFLAGS="-g"
-   cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2858,35 +3414,11 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_cxx_try_compile "$LINENO"; then :
   ac_cv_prog_cxx_g=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	CXXFLAGS=""
-      cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  CXXFLAGS=""
+      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2897,36 +3429,12 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+if ac_fn_cxx_try_compile "$LINENO"; then :
 
-	ac_cxx_werror_flag=$ac_save_cxx_werror_flag
+else
+  ac_cxx_werror_flag=$ac_save_cxx_werror_flag
 	 CXXFLAGS="-g"
-	 cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2937,42 +3445,17 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_cxx_try_compile "$LINENO"; then :
   ac_cv_prog_cxx_g=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
    ac_cxx_werror_flag=$ac_save_cxx_werror_flag
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cxx_g" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cxx_g" >&5
 $as_echo "$ac_cv_prog_cxx_g" >&6; }
 if test "$ac_test_CXXFLAGS" = set; then
   CXXFLAGS=$ac_save_CXXFLAGS
@@ -3006,14 +3489,14 @@ am__doit:
 .PHONY: am__doit
 END
 # If we don't find an include directive, just comment out the code.
-{ $as_echo "$as_me:$LINENO: checking for style of include used by $am_make" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for style of include used by $am_make" >&5
 $as_echo_n "checking for style of include used by $am_make... " >&6; }
 am__include="#"
 am__quote=
 _am_result=none
 # First try GNU make style include.
 echo "include confinc" > confmf
-# Ignore all kinds of additional output from `make'.
+# Ignore all kinds of additional output from 'make'.
 case `$am_make -s -f confmf 2> /dev/null` in #(
 *the\ am__doit\ target*)
   am__include=include
@@ -3034,18 +3517,19 @@ if test "$am__include" = "#"; then
 fi
 
 
-{ $as_echo "$as_me:$LINENO: result: $_am_result" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $_am_result" >&5
 $as_echo "$_am_result" >&6; }
 rm -f confinc confmf
 
 # Check whether --enable-dependency-tracking was given.
-if test "${enable_dependency_tracking+set}" = set; then
+if test "${enable_dependency_tracking+set}" = set; then :
   enableval=$enable_dependency_tracking;
 fi
 
 if test "x$enable_dependency_tracking" != xno; then
   am_depcomp="$ac_aux_dir/depcomp"
   AMDEPBACKSLASH='\'
+  am__nodep='_no'
 fi
  if test "x$enable_dependency_tracking" != xno; then
   AMDEP_TRUE=
@@ -3059,17 +3543,18 @@ fi
 
 depcc="$CXX"  am_compiler_list=
 
-{ $as_echo "$as_me:$LINENO: checking dependency style of $depcc" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
 $as_echo_n "checking dependency style of $depcc... " >&6; }
-if test "${am_cv_CXX_dependencies_compiler_type+set}" = set; then
+if ${am_cv_CXX_dependencies_compiler_type+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named `D' -- because `-MD' means `put the output
-  # in D'.
+  # making a dummy file named 'D' -- because '-MD' means "put the output
+  # in D".
+  rm -rf conftest.dir
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -3103,16 +3588,16 @@ else
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
-      # Solaris 8's {/usr,}/bin/sh.
-      touch sub/conftst$i.h
+      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
+      # Solaris 10 /bin/sh.
+      echo '/* dummy */' > sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with `-c' and `-o' for the sake of the "dashmstdout"
+    # We check with '-c' and '-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle `-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs
+    # handle '-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs.
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -3121,16 +3606,16 @@ else
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # after this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested
+      # After this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested.
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvisualcpp | msvcmsys)
-      # This compiler won't grok `-c -o', but also, the minuso test has
+    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
+      # This compiler won't grok '-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -3169,7 +3654,7 @@ else
 fi
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $am_cv_CXX_dependencies_compiler_type" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CXX_dependencies_compiler_type" >&5
 $as_echo "$am_cv_CXX_dependencies_compiler_type" >&6; }
 CXXDEPMODE=depmode=$am_cv_CXX_dependencies_compiler_type
 
@@ -3192,9 +3677,9 @@ ac_compiler_gnu=$ac_cv_c_compiler_gnu
 if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}gcc", so it can be a program name with args.
 set dummy ${ac_tool_prefix}gcc; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_CC+set}" = set; then
+if ${ac_cv_prog_CC+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3205,24 +3690,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_CC="${ac_tool_prefix}gcc"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:$LINENO: result: $CC" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3232,9 +3717,9 @@ if test -z "$ac_cv_prog_CC"; then
   ac_ct_CC=$CC
   # Extract the first word of "gcc", so it can be a program name with args.
 set dummy gcc; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_ac_ct_CC+set}" = set; then
+if ${ac_cv_prog_ac_ct_CC+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CC"; then
@@ -3245,24 +3730,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_ac_ct_CC="gcc"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CC=$ac_cv_prog_ac_ct_CC
 if test -n "$ac_ct_CC"; then
-  { $as_echo "$as_me:$LINENO: result: $ac_ct_CC" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
 $as_echo "$ac_ct_CC" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3271,7 +3756,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -3285,9 +3770,9 @@ if test -z "$CC"; then
           if test -n "$ac_tool_prefix"; then
     # Extract the first word of "${ac_tool_prefix}cc", so it can be a program name with args.
 set dummy ${ac_tool_prefix}cc; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_CC+set}" = set; then
+if ${ac_cv_prog_CC+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3298,24 +3783,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_CC="${ac_tool_prefix}cc"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:$LINENO: result: $CC" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3325,9 +3810,9 @@ fi
 if test -z "$CC"; then
   # Extract the first word of "cc", so it can be a program name with args.
 set dummy cc; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_CC+set}" = set; then
+if ${ac_cv_prog_CC+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3339,18 +3824,18 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     if test "$as_dir/$ac_word$ac_exec_ext" = "/usr/ucb/cc"; then
        ac_prog_rejected=yes
        continue
      fi
     ac_cv_prog_CC="cc"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 if test $ac_prog_rejected = yes; then
@@ -3369,10 +3854,10 @@ fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:$LINENO: result: $CC" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3384,9 +3869,9 @@ if test -z "$CC"; then
   do
     # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
 set dummy $ac_tool_prefix$ac_prog; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_CC+set}" = set; then
+if ${ac_cv_prog_CC+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3397,24 +3882,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_CC="$ac_tool_prefix$ac_prog"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:$LINENO: result: $CC" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3428,9 +3913,9 @@ if test -z "$CC"; then
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_ac_ct_CC+set}" = set; then
+if ${ac_cv_prog_ac_ct_CC+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CC"; then
@@ -3441,24 +3926,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_ac_ct_CC="$ac_prog"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CC=$ac_cv_prog_ac_ct_CC
 if test -n "$ac_ct_CC"; then
-  { $as_echo "$as_me:$LINENO: result: $ac_ct_CC" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
 $as_echo "$ac_ct_CC" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3471,7 +3956,7 @@ done
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -3482,62 +3967,42 @@ fi
 fi
 
 
-test -z "$CC" && { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+test -z "$CC" && { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: no acceptable C compiler found in \$PATH
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: no acceptable C compiler found in \$PATH
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
+as_fn_error $? "no acceptable C compiler found in \$PATH
+See \`config.log' for more details" "$LINENO" 5; }
 
 # Provide some information about the compiler.
-$as_echo "$as_me:$LINENO: checking for C compiler version" >&5
+$as_echo "$as_me:${as_lineno-$LINENO}: checking for C compiler version" >&5
 set X $ac_compile
 ac_compiler=$2
-{ (ac_try="$ac_compiler --version >&5"
+for ac_option in --version -v -V -qversion; do
+  { { ac_try="$ac_compiler $ac_option >&5"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler --version >&5") 2>&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
-{ (ac_try="$ac_compiler -v >&5"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler -v >&5") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
-{ (ac_try="$ac_compiler -V >&5"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler -V >&5") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
+  if test -s conftest.err; then
+    sed '10a\
+... rest of stderr output deleted ...
+         10q' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+  fi
+  rm -f conftest.er1 conftest.err
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }
+done
 
-{ $as_echo "$as_me:$LINENO: checking whether we are using the GNU C compiler" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C compiler" >&5
 $as_echo_n "checking whether we are using the GNU C compiler... " >&6; }
-if test "${ac_cv_c_compiler_gnu+set}" = set; then
+if ${ac_cv_c_compiler_gnu+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -3551,37 +4016,16 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_c_try_compile "$LINENO"; then :
   ac_compiler_gnu=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_compiler_gnu=no
+  ac_compiler_gnu=no
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 ac_cv_c_compiler_gnu=$ac_compiler_gnu
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_compiler_gnu" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_c_compiler_gnu" >&5
 $as_echo "$ac_cv_c_compiler_gnu" >&6; }
 if test $ac_compiler_gnu = yes; then
   GCC=yes
@@ -3590,20 +4034,16 @@ else
 fi
 ac_test_CFLAGS=${CFLAGS+set}
 ac_save_CFLAGS=$CFLAGS
-{ $as_echo "$as_me:$LINENO: checking whether $CC accepts -g" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC accepts -g" >&5
 $as_echo_n "checking whether $CC accepts -g... " >&6; }
-if test "${ac_cv_prog_cc_g+set}" = set; then
+if ${ac_cv_prog_cc_g+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   ac_save_c_werror_flag=$ac_c_werror_flag
    ac_c_werror_flag=yes
    ac_cv_prog_cc_g=no
    CFLAGS="-g"
-   cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -3614,35 +4054,11 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_c_try_compile "$LINENO"; then :
   ac_cv_prog_cc_g=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	CFLAGS=""
-      cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  CFLAGS=""
+      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -3653,36 +4069,12 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+if ac_fn_c_try_compile "$LINENO"; then :
 
-	ac_c_werror_flag=$ac_save_c_werror_flag
+else
+  ac_c_werror_flag=$ac_save_c_werror_flag
 	 CFLAGS="-g"
-	 cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -3693,42 +4085,17 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_c_try_compile "$LINENO"; then :
   ac_cv_prog_cc_g=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
    ac_c_werror_flag=$ac_save_c_werror_flag
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cc_g" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_g" >&5
 $as_echo "$ac_cv_prog_cc_g" >&6; }
 if test "$ac_test_CFLAGS" = set; then
   CFLAGS=$ac_save_CFLAGS
@@ -3745,23 +4112,18 @@ else
     CFLAGS=
   fi
 fi
-{ $as_echo "$as_me:$LINENO: checking for $CC option to accept ISO C89" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to accept ISO C89" >&5
 $as_echo_n "checking for $CC option to accept ISO C89... " >&6; }
-if test "${ac_cv_prog_cc_c89+set}" = set; then
+if ${ac_cv_prog_cc_c89+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   ac_cv_prog_cc_c89=no
 ac_save_CC=$CC
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <stdarg.h>
 #include <stdio.h>
-#include <sys/types.h>
-#include <sys/stat.h>
+struct stat;
 /* Most of the following tests are stolen from RCS 5.7's src/conf.sh.  */
 struct buf { int x; };
 FILE * (*rcsopen) (struct buf *, struct stat *, int);
@@ -3813,32 +4175,9 @@ for ac_arg in '' -qlanglvl=extc89 -qlanglvl=ansi -std \
 	-Ae "-Aa -D_HPUX_SOURCE" "-Xc -D__EXTENSIONS__"
 do
   CC="$ac_save_CC $ac_arg"
-  rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+  if ac_fn_c_try_compile "$LINENO"; then :
   ac_cv_prog_cc_c89=$ac_arg
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
 fi
-
 rm -f core conftest.err conftest.$ac_objext
   test "x$ac_cv_prog_cc_c89" != "xno" && break
 done
@@ -3849,17 +4188,19 @@ fi
 # AC_CACHE_VAL
 case "x$ac_cv_prog_cc_c89" in
   x)
-    { $as_echo "$as_me:$LINENO: result: none needed" >&5
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: none needed" >&5
 $as_echo "none needed" >&6; } ;;
   xno)
-    { $as_echo "$as_me:$LINENO: result: unsupported" >&5
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: unsupported" >&5
 $as_echo "unsupported" >&6; } ;;
   *)
     CC="$CC $ac_cv_prog_cc_c89"
-    { $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cc_c89" >&5
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_c89" >&5
 $as_echo "$ac_cv_prog_cc_c89" >&6; } ;;
 esac
+if test "x$ac_cv_prog_cc_c89" != xno; then :
 
+fi
 
 ac_ext=c
 ac_cpp='$CPP $CPPFLAGS'
@@ -3867,19 +4208,79 @@ ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
 ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC understands -c and -o together" >&5
+$as_echo_n "checking whether $CC understands -c and -o together... " >&6; }
+if ${am_cv_prog_cc_c_o+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+  # Make sure it works both with $CC and with simple cc.
+  # Following AC_PROG_CC_C_O, we do the test twice because some
+  # compilers refuse to overwrite an existing .o file with -o,
+  # though they will create one.
+  am_cv_prog_cc_c_o=yes
+  for am_i in 1 2; do
+    if { echo "$as_me:$LINENO: $CC -c conftest.$ac_ext -o conftest2.$ac_objext" >&5
+   ($CC -c conftest.$ac_ext -o conftest2.$ac_objext) >&5 2>&5
+   ac_status=$?
+   echo "$as_me:$LINENO: \$? = $ac_status" >&5
+   (exit $ac_status); } \
+         && test -f conftest2.$ac_objext; then
+      : OK
+    else
+      am_cv_prog_cc_c_o=no
+      break
+    fi
+  done
+  rm -f core conftest*
+  unset am_i
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_prog_cc_c_o" >&5
+$as_echo "$am_cv_prog_cc_c_o" >&6; }
+if test "$am_cv_prog_cc_c_o" != yes; then
+   # Losing compiler, so override with the script.
+   # FIXME: It is wrong to rewrite CC.
+   # But if we don't then we get into trouble of one sort or another.
+   # A longer-term fix would be to have automake use am__CC in this case,
+   # and then we could set am__CC="\$(top_srcdir)/compile \$(CC)"
+   CC="$am_aux_dir/compile $CC"
+fi
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+
+
 depcc="$CC"   am_compiler_list=
 
-{ $as_echo "$as_me:$LINENO: checking dependency style of $depcc" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
 $as_echo_n "checking dependency style of $depcc... " >&6; }
-if test "${am_cv_CC_dependencies_compiler_type+set}" = set; then
+if ${am_cv_CC_dependencies_compiler_type+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named `D' -- because `-MD' means `put the output
-  # in D'.
+  # making a dummy file named 'D' -- because '-MD' means "put the output
+  # in D".
+  rm -rf conftest.dir
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -3913,16 +4314,16 @@ else
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
-      # Solaris 8's {/usr,}/bin/sh.
-      touch sub/conftst$i.h
+      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
+      # Solaris 10 /bin/sh.
+      echo '/* dummy */' > sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with `-c' and `-o' for the sake of the "dashmstdout"
+    # We check with '-c' and '-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle `-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs
+    # handle '-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs.
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -3931,16 +4332,16 @@ else
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # after this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested
+      # After this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested.
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvisualcpp | msvcmsys)
-      # This compiler won't grok `-c -o', but also, the minuso test has
+    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
+      # This compiler won't grok '-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -3979,7 +4380,7 @@ else
 fi
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $am_cv_CC_dependencies_compiler_type" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CC_dependencies_compiler_type" >&5
 $as_echo "$am_cv_CC_dependencies_compiler_type" >&6; }
 CCDEPMODE=depmode=$am_cv_CC_dependencies_compiler_type
 
@@ -3998,17 +4399,18 @@ fi
 
   OPENMP_CFLAGS=
   # Check whether --enable-openmp was given.
-if test "${enable_openmp+set}" = set; then
+if test "${enable_openmp+set}" = set; then :
   enableval=$enable_openmp;
 fi
 
   if test "$enable_openmp" != no; then
-    { $as_echo "$as_me:$LINENO: checking for $CC option to support OpenMP" >&5
+    { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to support OpenMP" >&5
 $as_echo_n "checking for $CC option to support OpenMP... " >&6; }
-if test "${ac_cv_prog_c_openmp+set}" = set; then
+if ${ac_cv_prog_c_openmp+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
 
 #ifndef _OPENMP
  choke me
@@ -4017,37 +4419,16 @@ else
 int main () { return omp_get_num_threads (); }
 
 _ACEOF
-rm -f conftest.$ac_objext conftest$ac_exeext
-if { (ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_link") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 $as_test_x conftest$ac_exeext
-       }; then
+if ac_fn_c_try_link "$LINENO"; then :
   ac_cv_prog_c_openmp='none needed'
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_cv_prog_c_openmp='unsupported'
-	  	  	  	  	  	  	  	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp; do
+  ac_cv_prog_c_openmp='unsupported'
+	  	  	  	  	  	  	                                	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp -homp \
+                           -Popenmp --openmp; do
 	    ac_save_CFLAGS=$CFLAGS
 	    CFLAGS="$CFLAGS $ac_option"
-	    cat >conftest.$ac_ext <<_ACEOF
+	    cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
 
 #ifndef _OPENMP
  choke me
@@ -4056,56 +4437,27 @@ sed 's/^/| /' conftest.$ac_ext >&5
 int main () { return omp_get_num_threads (); }
 
 _ACEOF
-rm -f conftest.$ac_objext conftest$ac_exeext
-if { (ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_link") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 $as_test_x conftest$ac_exeext
-       }; then
+if ac_fn_c_try_link "$LINENO"; then :
   ac_cv_prog_c_openmp=$ac_option
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
 fi
-
-rm -rf conftest.dSYM
-rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
-      conftest$ac_exeext conftest.$ac_ext
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
 	    CFLAGS=$ac_save_CFLAGS
 	    if test "$ac_cv_prog_c_openmp" != unsupported; then
 	      break
 	    fi
 	  done
 fi
-
-rm -rf conftest.dSYM
-rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
-      conftest$ac_exeext conftest.$ac_ext
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_c_openmp" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_c_openmp" >&5
 $as_echo "$ac_cv_prog_c_openmp" >&6; }
     case $ac_cv_prog_c_openmp in #(
       "none needed" | unsupported)
-        ;; #(
+	;; #(
       *)
-        OPENMP_CFLAGS=$ac_cv_prog_c_openmp ;;
+	OPENMP_CFLAGS=$ac_cv_prog_c_openmp ;;
     esac
   fi
 
@@ -4113,9 +4465,9 @@ $as_echo "$ac_cv_prog_c_openmp" >&6; }
 if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}ranlib", so it can be a program name with args.
 set dummy ${ac_tool_prefix}ranlib; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_RANLIB+set}" = set; then
+if ${ac_cv_prog_RANLIB+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$RANLIB"; then
@@ -4126,24 +4478,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_RANLIB="${ac_tool_prefix}ranlib"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 RANLIB=$ac_cv_prog_RANLIB
 if test -n "$RANLIB"; then
-  { $as_echo "$as_me:$LINENO: result: $RANLIB" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $RANLIB" >&5
 $as_echo "$RANLIB" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -4153,9 +4505,9 @@ if test -z "$ac_cv_prog_RANLIB"; then
   ac_ct_RANLIB=$RANLIB
   # Extract the first word of "ranlib", so it can be a program name with args.
 set dummy ranlib; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_ac_ct_RANLIB+set}" = set; then
+if ${ac_cv_prog_ac_ct_RANLIB+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_RANLIB"; then
@@ -4166,24 +4518,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_ac_ct_RANLIB="ranlib"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_RANLIB=$ac_cv_prog_ac_ct_RANLIB
 if test -n "$ac_ct_RANLIB"; then
-  { $as_echo "$as_me:$LINENO: result: $ac_ct_RANLIB" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_RANLIB" >&5
 $as_echo "$ac_ct_RANLIB" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -4192,7 +4544,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -4212,14 +4564,14 @@ ac_cpp='$CPP $CPPFLAGS'
 ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
 ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
-{ $as_echo "$as_me:$LINENO: checking how to run the C preprocessor" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking how to run the C preprocessor" >&5
 $as_echo_n "checking how to run the C preprocessor... " >&6; }
 # On Suns, sometimes $CPP names a directory.
 if test -n "$CPP" && test -d "$CPP"; then
   CPP=
 fi
 if test -z "$CPP"; then
-  if test "${ac_cv_prog_CPP+set}" = set; then
+  if ${ac_cv_prog_CPP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
       # Double quotes because CPP needs to be expanded
@@ -4234,11 +4586,7 @@ do
   # <limits.h> exists even on freestanding compilers.
   # On the NeXT, cc -E runs the code through the compiler's parser,
   # not just through cpp. "Syntax error" is here to catch this case.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #ifdef __STDC__
 # include <limits.h>
@@ -4247,78 +4595,34 @@ cat >>conftest.$ac_ext <<_ACEOF
 #endif
 		     Syntax error
 _ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+if ac_fn_c_try_cpp "$LINENO"; then :
 
+else
   # Broken: fails on valid input.
 continue
 fi
-
-rm -f conftest.err conftest.$ac_ext
+rm -f conftest.err conftest.i conftest.$ac_ext
 
   # OK, works on sane cases.  Now check whether nonexistent headers
   # can be detected and how.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <ac_nonexistent.h>
 _ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
+if ac_fn_c_try_cpp "$LINENO"; then :
   # Broken: success on invalid input.
 continue
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
   # Passes both tests.
 ac_preproc_ok=:
 break
 fi
-
-rm -f conftest.err conftest.$ac_ext
+rm -f conftest.err conftest.i conftest.$ac_ext
 
 done
 # Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
-rm -f conftest.err conftest.$ac_ext
-if $ac_preproc_ok; then
+rm -f conftest.i conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then :
   break
 fi
 
@@ -4330,7 +4634,7 @@ fi
 else
   ac_cv_prog_CPP=$CPP
 fi
-{ $as_echo "$as_me:$LINENO: result: $CPP" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $CPP" >&5
 $as_echo "$CPP" >&6; }
 ac_preproc_ok=false
 for ac_c_preproc_warn_flag in '' yes
@@ -4341,11 +4645,7 @@ do
   # <limits.h> exists even on freestanding compilers.
   # On the NeXT, cc -E runs the code through the compiler's parser,
   # not just through cpp. "Syntax error" is here to catch this case.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #ifdef __STDC__
 # include <limits.h>
@@ -4354,87 +4654,40 @@ cat >>conftest.$ac_ext <<_ACEOF
 #endif
 		     Syntax error
 _ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+if ac_fn_c_try_cpp "$LINENO"; then :
 
+else
   # Broken: fails on valid input.
 continue
 fi
-
-rm -f conftest.err conftest.$ac_ext
+rm -f conftest.err conftest.i conftest.$ac_ext
 
   # OK, works on sane cases.  Now check whether nonexistent headers
   # can be detected and how.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <ac_nonexistent.h>
 _ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
+if ac_fn_c_try_cpp "$LINENO"; then :
   # Broken: success on invalid input.
 continue
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
   # Passes both tests.
 ac_preproc_ok=:
 break
 fi
-
-rm -f conftest.err conftest.$ac_ext
+rm -f conftest.err conftest.i conftest.$ac_ext
 
 done
 # Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
-rm -f conftest.err conftest.$ac_ext
-if $ac_preproc_ok; then
-  :
+rm -f conftest.i conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then :
+
 else
-  { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: C preprocessor \"$CPP\" fails sanity check
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: C preprocessor \"$CPP\" fails sanity check
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
+as_fn_error $? "C preprocessor \"$CPP\" fails sanity check
+See \`config.log' for more details" "$LINENO" 5; }
 fi
 
 ac_ext=c
@@ -4444,9 +4697,9 @@ ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
 
-{ $as_echo "$as_me:$LINENO: checking for grep that handles long lines and -e" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for grep that handles long lines and -e" >&5
 $as_echo_n "checking for grep that handles long lines and -e... " >&6; }
-if test "${ac_cv_path_GREP+set}" = set; then
+if ${ac_cv_path_GREP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -z "$GREP"; then
@@ -4457,10 +4710,10 @@ for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_prog in grep ggrep; do
+    for ac_prog in grep ggrep; do
     for ac_exec_ext in '' $ac_executable_extensions; do
       ac_path_GREP="$as_dir/$ac_prog$ac_exec_ext"
-      { test -f "$ac_path_GREP" && $as_test_x "$ac_path_GREP"; } || continue
+      as_fn_executable_p "$ac_path_GREP" || continue
 # Check for GNU ac_path_GREP and select it if it is found.
   # Check for GNU $ac_path_GREP
 case `"$ac_path_GREP" --version 2>&1` in
@@ -4477,7 +4730,7 @@ case `"$ac_path_GREP" --version 2>&1` in
     $as_echo 'GREP' >> "conftest.nl"
     "$ac_path_GREP" -e 'GREP$' -e '-(cannot match)-' < "conftest.nl" >"conftest.out" 2>/dev/null || break
     diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
-    ac_count=`expr $ac_count + 1`
+    as_fn_arith $ac_count + 1 && ac_count=$as_val
     if test $ac_count -gt ${ac_path_GREP_max-0}; then
       # Best one so far, save it but keep looking for a better one
       ac_cv_path_GREP="$ac_path_GREP"
@@ -4492,26 +4745,24 @@ esac
       $ac_path_GREP_found && break 3
     done
   done
-done
+  done
 IFS=$as_save_IFS
   if test -z "$ac_cv_path_GREP"; then
-    { { $as_echo "$as_me:$LINENO: error: no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&5
-$as_echo "$as_me: error: no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
   fi
 else
   ac_cv_path_GREP=$GREP
 fi
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_path_GREP" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_GREP" >&5
 $as_echo "$ac_cv_path_GREP" >&6; }
  GREP="$ac_cv_path_GREP"
 
 
-{ $as_echo "$as_me:$LINENO: checking for egrep" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for egrep" >&5
 $as_echo_n "checking for egrep... " >&6; }
-if test "${ac_cv_path_EGREP+set}" = set; then
+if ${ac_cv_path_EGREP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if echo a | $GREP -E '(a|b)' >/dev/null 2>&1
@@ -4525,10 +4776,10 @@ for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_prog in egrep; do
+    for ac_prog in egrep; do
     for ac_exec_ext in '' $ac_executable_extensions; do
       ac_path_EGREP="$as_dir/$ac_prog$ac_exec_ext"
-      { test -f "$ac_path_EGREP" && $as_test_x "$ac_path_EGREP"; } || continue
+      as_fn_executable_p "$ac_path_EGREP" || continue
 # Check for GNU ac_path_EGREP and select it if it is found.
   # Check for GNU $ac_path_EGREP
 case `"$ac_path_EGREP" --version 2>&1` in
@@ -4545,7 +4796,7 @@ case `"$ac_path_EGREP" --version 2>&1` in
     $as_echo 'EGREP' >> "conftest.nl"
     "$ac_path_EGREP" 'EGREP$' < "conftest.nl" >"conftest.out" 2>/dev/null || break
     diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
-    ac_count=`expr $ac_count + 1`
+    as_fn_arith $ac_count + 1 && ac_count=$as_val
     if test $ac_count -gt ${ac_path_EGREP_max-0}; then
       # Best one so far, save it but keep looking for a better one
       ac_cv_path_EGREP="$ac_path_EGREP"
@@ -4560,12 +4811,10 @@ esac
       $ac_path_EGREP_found && break 3
     done
   done
-done
+  done
 IFS=$as_save_IFS
   if test -z "$ac_cv_path_EGREP"; then
-    { { $as_echo "$as_me:$LINENO: error: no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&5
-$as_echo "$as_me: error: no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
   fi
 else
   ac_cv_path_EGREP=$EGREP
@@ -4573,21 +4822,17 @@ fi
 
    fi
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_path_EGREP" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_EGREP" >&5
 $as_echo "$ac_cv_path_EGREP" >&6; }
  EGREP="$ac_cv_path_EGREP"
 
 
-{ $as_echo "$as_me:$LINENO: checking for ANSI C header files" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for ANSI C header files" >&5
 $as_echo_n "checking for ANSI C header files... " >&6; }
-if test "${ac_cv_header_stdc+set}" = set; then
+if ${ac_cv_header_stdc+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <stdlib.h>
 #include <stdarg.h>
@@ -4602,48 +4847,23 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_c_try_compile "$LINENO"; then :
   ac_cv_header_stdc=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_cv_header_stdc=no
+  ac_cv_header_stdc=no
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 
 if test $ac_cv_header_stdc = yes; then
   # SunOS 4.x string.h does not declare mem*, contrary to ANSI.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <string.h>
 
 _ACEOF
 if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
-  $EGREP "memchr" >/dev/null 2>&1; then
-  :
+  $EGREP "memchr" >/dev/null 2>&1; then :
+
 else
   ac_cv_header_stdc=no
 fi
@@ -4653,18 +4873,14 @@ fi
 
 if test $ac_cv_header_stdc = yes; then
   # ISC 2.0.2 stdlib.h does not declare free, contrary to ANSI.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <stdlib.h>
 
 _ACEOF
 if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
-  $EGREP "free" >/dev/null 2>&1; then
-  :
+  $EGREP "free" >/dev/null 2>&1; then :
+
 else
   ac_cv_header_stdc=no
 fi
@@ -4674,14 +4890,10 @@ fi
 
 if test $ac_cv_header_stdc = yes; then
   # /bin/cc in Irix-4.0.5 gets non-ANSI ctype macros unless using -ansi.
-  if test "$cross_compiling" = yes; then
+  if test "$cross_compiling" = yes; then :
   :
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <ctype.h>
 #include <stdlib.h>
@@ -4708,118 +4920,33 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest$ac_exeext
-if { (ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_link") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && { ac_try='./conftest$ac_exeext'
-  { (case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; }; then
-  :
+if ac_fn_c_try_run "$LINENO"; then :
+
 else
-  $as_echo "$as_me: program exited with status $ac_status" >&5
-$as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-( exit $ac_status )
-ac_cv_header_stdc=no
+  ac_cv_header_stdc=no
 fi
-rm -rf conftest.dSYM
-rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext conftest.$ac_objext conftest.$ac_ext
+rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
+  conftest.$ac_objext conftest.beam conftest.$ac_ext
 fi
 
-
 fi
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_header_stdc" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_header_stdc" >&5
 $as_echo "$ac_cv_header_stdc" >&6; }
 if test $ac_cv_header_stdc = yes; then
 
-cat >>confdefs.h <<\_ACEOF
-#define STDC_HEADERS 1
-_ACEOF
+$as_echo "#define STDC_HEADERS 1" >>confdefs.h
 
 fi
 
 # On IRIX 5.3, sys/types and inttypes.h are conflicting.
-
-
-
-
-
-
-
-
-
 for ac_header in sys/types.h sys/stat.h stdlib.h string.h memory.h strings.h \
 		  inttypes.h stdint.h unistd.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  eval "$as_ac_Header=yes"
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	eval "$as_ac_Header=no"
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+ac_fn_c_check_header_compile "$LINENO" "$ac_header" "$as_ac_Header" "$ac_includes_default
+"
+if eval test \"x\$"$as_ac_Header"\" = x"yes"; then :
   cat >>confdefs.h <<_ACEOF
 #define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
@@ -4829,604 +4956,48 @@ fi
 done
 
 
-
 for ac_header in stdint.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
-$as_echo_n "checking $ac_header usability... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_header_compiler=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_header_compiler=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
-$as_echo_n "checking $ac_header presence... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <$ac_header>
-_ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  ac_header_preproc=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-  ac_header_preproc=no
-fi
-
-rm -f conftest.err conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
-  yes:no: )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
-    ac_header_preproc=yes
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
-$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
-    ( cat <<\_ASBOX
-## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##
-_ASBOX
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  eval "$as_ac_Header=\$ac_header_preproc"
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-
-fi
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "stdint.h" "ac_cv_header_stdint_h" "$ac_includes_default"
+if test "x$ac_cv_header_stdint_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+#define HAVE_STDINT_H 1
 _ACEOF
 
 fi
 
 done
 
-
 for ac_header in malloc/malloc.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
-$as_echo_n "checking $ac_header usability... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_header_compiler=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_header_compiler=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
-$as_echo_n "checking $ac_header presence... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <$ac_header>
-_ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  ac_header_preproc=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-  ac_header_preproc=no
-fi
-
-rm -f conftest.err conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
-  yes:no: )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
-    ac_header_preproc=yes
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
-$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
-    ( cat <<\_ASBOX
-## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##
-_ASBOX
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  eval "$as_ac_Header=\$ac_header_preproc"
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-
-fi
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "malloc/malloc.h" "ac_cv_header_malloc_malloc_h" "$ac_includes_default"
+if test "x$ac_cv_header_malloc_malloc_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+#define HAVE_MALLOC_MALLOC_H 1
 _ACEOF
 
 fi
 
 done
 
-
 for ac_header in malloc.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
-$as_echo_n "checking $ac_header usability... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_header_compiler=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_header_compiler=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
-$as_echo_n "checking $ac_header presence... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <$ac_header>
-_ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  ac_header_preproc=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-  ac_header_preproc=no
-fi
-
-rm -f conftest.err conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
-  yes:no: )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
-    ac_header_preproc=yes
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
-$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
-    ( cat <<\_ASBOX
-## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##
-_ASBOX
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  eval "$as_ac_Header=\$ac_header_preproc"
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-
-fi
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "malloc.h" "ac_cv_header_malloc_h" "$ac_includes_default"
+if test "x$ac_cv_header_malloc_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+#define HAVE_MALLOC_H 1
 _ACEOF
 
 fi
 
 done
 
-
 for ac_header in endian.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
-$as_echo_n "checking $ac_header usability... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_header_compiler=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_header_compiler=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
-$as_echo_n "checking $ac_header presence... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <$ac_header>
-_ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  ac_header_preproc=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-  ac_header_preproc=no
-fi
-
-rm -f conftest.err conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
-  yes:no: )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
-    ac_header_preproc=yes
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
-$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
-    ( cat <<\_ASBOX
-## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##
-_ASBOX
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  eval "$as_ac_Header=\$ac_header_preproc"
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-
-fi
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "endian.h" "ac_cv_header_endian_h" "$ac_includes_default"
+if test "x$ac_cv_header_endian_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+#define HAVE_ENDIAN_H 1
 _ACEOF
 
 fi
@@ -5434,244 +5005,35 @@ fi
 done
 
 #AC_CHECK_HEADERS(machine/endian.h)
-{ $as_echo "$as_me:$LINENO: checking whether ntohll is declared" >&5
-$as_echo_n "checking whether ntohll is declared... " >&6; }
-if test "${ac_cv_have_decl_ntohll+set}" = set; then
-  $as_echo_n "(cached) " >&6
+ac_fn_c_check_decl "$LINENO" "ntohll" "ac_cv_have_decl_ntohll" "#include <arpa/inet.h>
+"
+if test "x$ac_cv_have_decl_ntohll" = xyes; then :
+  ac_have_decl=1
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <arpa/inet.h>
-
-int
-main ()
-{
-#ifndef ntohll
-  (void) ntohll;
-#endif
-
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_cv_have_decl_ntohll=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_cv_have_decl_ntohll=no
+  ac_have_decl=0
 fi
 
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_have_decl_ntohll" >&5
-$as_echo "$ac_cv_have_decl_ntohll" >&6; }
-if test "x$ac_cv_have_decl_ntohll" = x""yes; then
-
 cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_NTOHLL 1
+#define HAVE_DECL_NTOHLL $ac_have_decl
 _ACEOF
 
-
+ac_fn_c_check_decl "$LINENO" "be64toh" "ac_cv_have_decl_be64toh" "#include <arpa/inet.h>
+"
+if test "x$ac_cv_have_decl_be64toh" = xyes; then :
+  ac_have_decl=1
 else
-  cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_NTOHLL 0
-_ACEOF
-
-
+  ac_have_decl=0
 fi
 
-
-{ $as_echo "$as_me:$LINENO: checking whether be64toh is declared" >&5
-$as_echo_n "checking whether be64toh is declared... " >&6; }
-if test "${ac_cv_have_decl_be64toh+set}" = set; then
-  $as_echo_n "(cached) " >&6
-else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <arpa/inet.h>
-
-int
-main ()
-{
-#ifndef be64toh
-  (void) be64toh;
-#endif
-
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_cv_have_decl_be64toh=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_cv_have_decl_be64toh=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_have_decl_be64toh" >&5
-$as_echo "$ac_cv_have_decl_be64toh" >&6; }
-if test "x$ac_cv_have_decl_be64toh" = x""yes; then
-
 cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_BE64TOH 1
+#define HAVE_DECL_BE64TOH $ac_have_decl
 _ACEOF
 
 
-else
-  cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_BE64TOH 0
-_ACEOF
-
-
-fi
-
-
-
 # Checks for typedefs, structures, and compiler characteristics.
-{ $as_echo "$as_me:$LINENO: checking for size_t" >&5
-$as_echo_n "checking for size_t... " >&6; }
-if test "${ac_cv_type_size_t+set}" = set; then
-  $as_echo_n "(cached) " >&6
-else
-  ac_cv_type_size_t=no
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-if (sizeof (size_t))
-       return 0;
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-if (sizeof ((size_t)))
-	  return 0;
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+ac_fn_c_check_type "$LINENO" "size_t" "ac_cv_type_size_t" "$ac_includes_default"
+if test "x$ac_cv_type_size_t" = xyes; then :
 
-	ac_cv_type_size_t=yes
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_type_size_t" >&5
-$as_echo "$ac_cv_type_size_t" >&6; }
-if test "x$ac_cv_type_size_t" = x""yes; then
-  :
 else
 
 cat >>confdefs.h <<_ACEOF
@@ -5680,75 +5042,12 @@ _ACEOF
 
 fi
 
-
-  { $as_echo "$as_me:$LINENO: checking for uint32_t" >&5
-$as_echo_n "checking for uint32_t... " >&6; }
-if test "${ac_cv_c_uint32_t+set}" = set; then
-  $as_echo_n "(cached) " >&6
-else
-  ac_cv_c_uint32_t=no
-     for ac_type in 'uint32_t' 'unsigned int' 'unsigned long int' \
-	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
-       cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-static int test_array [1 - 2 * !(($ac_type) -1 >> (32 - 1) == 1)];
-test_array [0] = 0
-
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  case $ac_type in
-  uint32_t) ac_cv_c_uint32_t=yes ;;
-  *) ac_cv_c_uint32_t=$ac_type ;;
-esac
-
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-       test "$ac_cv_c_uint32_t" != no && break
-     done
-fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_uint32_t" >&5
-$as_echo "$ac_cv_c_uint32_t" >&6; }
-  case $ac_cv_c_uint32_t in #(
+ac_fn_c_find_uintX_t "$LINENO" "32" "ac_cv_c_uint32_t"
+case $ac_cv_c_uint32_t in #(
   no|yes) ;; #(
   *)
 
-cat >>confdefs.h <<\_ACEOF
-#define _UINT32_T 1
-_ACEOF
+$as_echo "#define _UINT32_T 1" >>confdefs.h
 
 
 cat >>confdefs.h <<_ACEOF
@@ -5757,75 +5056,12 @@ _ACEOF
 ;;
   esac
 
-
-  { $as_echo "$as_me:$LINENO: checking for uint64_t" >&5
-$as_echo_n "checking for uint64_t... " >&6; }
-if test "${ac_cv_c_uint64_t+set}" = set; then
-  $as_echo_n "(cached) " >&6
-else
-  ac_cv_c_uint64_t=no
-     for ac_type in 'uint64_t' 'unsigned int' 'unsigned long int' \
-	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
-       cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-static int test_array [1 - 2 * !(($ac_type) -1 >> (64 - 1) == 1)];
-test_array [0] = 0
-
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  case $ac_type in
-  uint64_t) ac_cv_c_uint64_t=yes ;;
-  *) ac_cv_c_uint64_t=$ac_type ;;
-esac
-
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-       test "$ac_cv_c_uint64_t" != no && break
-     done
-fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_uint64_t" >&5
-$as_echo "$ac_cv_c_uint64_t" >&6; }
-  case $ac_cv_c_uint64_t in #(
+ac_fn_c_find_uintX_t "$LINENO" "64" "ac_cv_c_uint64_t"
+case $ac_cv_c_uint64_t in #(
   no|yes) ;; #(
   *)
 
-cat >>confdefs.h <<\_ACEOF
-#define _UINT64_T 1
-_ACEOF
+$as_echo "#define _UINT64_T 1" >>confdefs.h
 
 
 cat >>confdefs.h <<_ACEOF
@@ -5836,102 +5072,12 @@ _ACEOF
 
 
 # Checks for library functions.
-
 for ac_func in gettimeofday
-do
-as_ac_var=`$as_echo "ac_cv_func_$ac_func" | $as_tr_sh`
-{ $as_echo "$as_me:$LINENO: checking for $ac_func" >&5
-$as_echo_n "checking for $ac_func... " >&6; }
-if { as_var=$as_ac_var; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-/* Define $ac_func to an innocuous variant, in case <limits.h> declares $ac_func.
-   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
-#define $ac_func innocuous_$ac_func
-
-/* System header to define __stub macros and hopefully few prototypes,
-    which can conflict with char $ac_func (); below.
-    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
-    <limits.h> exists even on freestanding compilers.  */
-
-#ifdef __STDC__
-# include <limits.h>
-#else
-# include <assert.h>
-#endif
-
-#undef $ac_func
-
-/* Override any GCC internal prototype to avoid an error.
-   Use char because int might match the return type of a GCC
-   builtin and then its argument prototype would still apply.  */
-#ifdef __cplusplus
-extern "C"
-#endif
-char $ac_func ();
-/* The GNU C library defines this for functions which it implements
-    to always fail with ENOSYS.  Some functions are actually named
-    something starting with __ and the normal name is an alias.  */
-#if defined __stub_$ac_func || defined __stub___$ac_func
-choke me
-#endif
-
-int
-main ()
-{
-return $ac_func ();
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext conftest$ac_exeext
-if { (ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_link") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 $as_test_x conftest$ac_exeext
-       }; then
-  eval "$as_ac_var=yes"
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	eval "$as_ac_var=no"
-fi
-
-rm -rf conftest.dSYM
-rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
-      conftest$ac_exeext conftest.$ac_ext
-fi
-ac_res=`eval 'as_val=${'$as_ac_var'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-as_val=`eval 'as_val=${'$as_ac_var'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_c_check_func "$LINENO" "gettimeofday" "ac_cv_func_gettimeofday"
+if test "x$ac_cv_func_gettimeofday" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_func" | $as_tr_cpp` 1
+#define HAVE_GETTIMEOFDAY 1
 _ACEOF
 
 fi
@@ -5939,7 +5085,7 @@ done
 
 
 # Check whether --enable-simd was given.
-if test "${enable_simd+set}" = set; then
+if test "${enable_simd+set}" = set; then :
   enableval=$enable_simd; \
 	ac_SIMD=${enable_simd}
 else
@@ -5951,45 +5097,35 @@ case ${ac_SIMD} in
      SSE4)
        echo Configuring for SSE4
 
-cat >>confdefs.h <<\_ACEOF
-#define SSE4 1
-_ACEOF
+$as_echo "#define SSE4 1" >>confdefs.h
 
      ;;
      AVX)
        echo Configuring for AVX
 
-cat >>confdefs.h <<\_ACEOF
-#define AVX1 1
-_ACEOF
+$as_echo "#define AVX1 1" >>confdefs.h
 
      ;;
      AVX2)
        echo Configuring for AVX2
 
-cat >>confdefs.h <<\_ACEOF
-#define AVX2 1
-_ACEOF
+$as_echo "#define AVX2 1" >>confdefs.h
 
      ;;
      AVX512|MIC)
        echo Configuring for AVX512 and MIC
 
-cat >>confdefs.h <<\_ACEOF
-#define AVX512 1
-_ACEOF
+$as_echo "#define AVX512 1" >>confdefs.h
 
      ;;
      *)
-     { { $as_echo "$as_me:$LINENO: error: ${ac_SIMD} unsupported --enable-simd option" >&5
-$as_echo "$as_me: error: ${ac_SIMD} unsupported --enable-simd option" >&2;}
-   { (exit 1); exit 1; }; };
+     as_fn_error $? "${ac_SIMD} unsupported --enable-simd option" "$LINENO" 5;
      ;;
 esac
 
 
 # Check whether --enable-comms was given.
-if test "${enable_comms+set}" = set; then
+if test "${enable_comms+set}" = set; then :
   enableval=$enable_comms; ac_COMMS=${enable_comms}
 else
   ac_COMMS=none
@@ -6000,23 +5136,17 @@ case ${ac_COMMS} in
      none)
        echo Configuring for NO communications
 
-cat >>confdefs.h <<\_ACEOF
-#define GRID_COMMS_NONE 1
-_ACEOF
+$as_echo "#define GRID_COMMS_NONE 1" >>confdefs.h
 
      ;;
      mpi)
        echo Configuring for MPI communications
 
-cat >>confdefs.h <<\_ACEOF
-#define GRID_COMMS_MPI 1
-_ACEOF
+$as_echo "#define GRID_COMMS_MPI 1" >>confdefs.h
 
      ;;
      *)
-     { { $as_echo "$as_me:$LINENO: error: ${ac_COMMS} unsupported --enable-comms option" >&5
-$as_echo "$as_me: error: ${ac_COMMS} unsupported --enable-comms option" >&2;}
-   { (exit 1); exit 1; }; };
+     as_fn_error $? "${ac_COMMS} unsupported --enable-comms option" "$LINENO" 5;
      ;;
 esac
 
@@ -6073,13 +5203,13 @@ _ACEOF
     case $ac_val in #(
     *${as_nl}*)
       case $ac_var in #(
-      *_cv_*) { $as_echo "$as_me:$LINENO: WARNING: cache variable $ac_var contains a newline" >&5
+      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
 $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
       esac
       case $ac_var in #(
       _ | IFS | as_nl) ;; #(
       BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
-      *) $as_unset $ac_var ;;
+      *) { eval $ac_var=; unset $ac_var;} ;;
       esac ;;
     esac
   done
@@ -6087,8 +5217,8 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
   (set) 2>&1 |
     case $as_nl`(ac_space=' '; set) 2>&1` in #(
     *${as_nl}ac_space=\ *)
-      # `set' does not quote correctly, so add quotes (double-quote
-      # substitution turns \\\\ into \\, and sed turns \\ into \).
+      # `set' does not quote correctly, so add quotes: double-quote
+      # substitution turns \\\\ into \\, and sed turns \\ into \.
       sed -n \
 	"s/'/'\\\\''/g;
 	  s/^\\([_$as_cr_alnum]*_cv_[_$as_cr_alnum]*\\)=\\(.*\\)/\\1='\\2'/p"
@@ -6110,12 +5240,23 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
      :end' >>confcache
 if diff "$cache_file" confcache >/dev/null 2>&1; then :; else
   if test -w "$cache_file"; then
-    test "x$cache_file" != "x/dev/null" &&
-      { $as_echo "$as_me:$LINENO: updating cache $cache_file" >&5
+    if test "x$cache_file" != "x/dev/null"; then
+      { $as_echo "$as_me:${as_lineno-$LINENO}: updating cache $cache_file" >&5
 $as_echo "$as_me: updating cache $cache_file" >&6;}
-    cat confcache >$cache_file
+      if test ! -f "$cache_file" || test -h "$cache_file"; then
+	cat confcache >"$cache_file"
+      else
+        case $cache_file in #(
+        */* | ?:*)
+	  mv -f confcache "$cache_file"$$ &&
+	  mv -f "$cache_file"$$ "$cache_file" ;; #(
+        *)
+	  mv -f confcache "$cache_file" ;;
+	esac
+      fi
+    fi
   else
-    { $as_echo "$as_me:$LINENO: not updating unwritable cache $cache_file" >&5
+    { $as_echo "$as_me:${as_lineno-$LINENO}: not updating unwritable cache $cache_file" >&5
 $as_echo "$as_me: not updating unwritable cache $cache_file" >&6;}
   fi
 fi
@@ -6129,20 +5270,29 @@ DEFS=-DHAVE_CONFIG_H
 
 ac_libobjs=
 ac_ltlibobjs=
+U=
 for ac_i in : $LIBOBJS; do test "x$ac_i" = x: && continue
   # 1. Remove the extension, and $U if already installed.
   ac_script='s/\$U\././;s/\.o$//;s/\.obj$//'
   ac_i=`$as_echo "$ac_i" | sed "$ac_script"`
   # 2. Prepend LIBOBJDIR.  When used with automake>=1.10 LIBOBJDIR
   #    will be set to the directory where LIBOBJS objects are built.
-  ac_libobjs="$ac_libobjs \${LIBOBJDIR}$ac_i\$U.$ac_objext"
-  ac_ltlibobjs="$ac_ltlibobjs \${LIBOBJDIR}$ac_i"'$U.lo'
+  as_fn_append ac_libobjs " \${LIBOBJDIR}$ac_i\$U.$ac_objext"
+  as_fn_append ac_ltlibobjs " \${LIBOBJDIR}$ac_i"'$U.lo'
 done
 LIBOBJS=$ac_libobjs
 
 LTLIBOBJS=$ac_ltlibobjs
 
 
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking that generated files are newer than configure" >&5
+$as_echo_n "checking that generated files are newer than configure... " >&6; }
+   if test -n "$am_sleep_pid"; then
+     # Hide warnings about reused PIDs.
+     wait $am_sleep_pid 2>/dev/null
+   fi
+   { $as_echo "$as_me:${as_lineno-$LINENO}: result: done" >&5
+$as_echo "done" >&6; }
  if test -n "$EXEEXT"; then
   am__EXEEXT_TRUE=
   am__EXEEXT_FALSE='#'
@@ -6152,48 +5302,34 @@ else
 fi
 
 if test -z "${AMDEP_TRUE}" && test -z "${AMDEP_FALSE}"; then
-  { { $as_echo "$as_me:$LINENO: error: conditional \"AMDEP\" was never defined.
-Usually this means the macro was only invoked conditionally." >&5
-$as_echo "$as_me: error: conditional \"AMDEP\" was never defined.
-Usually this means the macro was only invoked conditionally." >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "conditional \"AMDEP\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 if test -z "${am__fastdepCXX_TRUE}" && test -z "${am__fastdepCXX_FALSE}"; then
-  { { $as_echo "$as_me:$LINENO: error: conditional \"am__fastdepCXX\" was never defined.
-Usually this means the macro was only invoked conditionally." >&5
-$as_echo "$as_me: error: conditional \"am__fastdepCXX\" was never defined.
-Usually this means the macro was only invoked conditionally." >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "conditional \"am__fastdepCXX\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 if test -z "${am__fastdepCC_TRUE}" && test -z "${am__fastdepCC_FALSE}"; then
-  { { $as_echo "$as_me:$LINENO: error: conditional \"am__fastdepCC\" was never defined.
-Usually this means the macro was only invoked conditionally." >&5
-$as_echo "$as_me: error: conditional \"am__fastdepCC\" was never defined.
-Usually this means the macro was only invoked conditionally." >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "conditional \"am__fastdepCC\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 if test -z "${BUILD_COMMS_MPI_TRUE}" && test -z "${BUILD_COMMS_MPI_FALSE}"; then
-  { { $as_echo "$as_me:$LINENO: error: conditional \"BUILD_COMMS_MPI\" was never defined.
-Usually this means the macro was only invoked conditionally." >&5
-$as_echo "$as_me: error: conditional \"BUILD_COMMS_MPI\" was never defined.
-Usually this means the macro was only invoked conditionally." >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "conditional \"BUILD_COMMS_MPI\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 if test -z "${BUILD_COMMS_NONE_TRUE}" && test -z "${BUILD_COMMS_NONE_FALSE}"; then
-  { { $as_echo "$as_me:$LINENO: error: conditional \"BUILD_COMMS_NONE\" was never defined.
-Usually this means the macro was only invoked conditionally." >&5
-$as_echo "$as_me: error: conditional \"BUILD_COMMS_NONE\" was never defined.
-Usually this means the macro was only invoked conditionally." >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "conditional \"BUILD_COMMS_NONE\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 
-: ${CONFIG_STATUS=./config.status}
+: "${CONFIG_STATUS=./config.status}"
 ac_write_fail=0
 ac_clean_files_save=$ac_clean_files
 ac_clean_files="$ac_clean_files $CONFIG_STATUS"
-{ $as_echo "$as_me:$LINENO: creating $CONFIG_STATUS" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: creating $CONFIG_STATUS" >&5
 $as_echo "$as_me: creating $CONFIG_STATUS" >&6;}
-cat >$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+as_write_fail=0
+cat >$CONFIG_STATUS <<_ASEOF || as_write_fail=1
 #! $SHELL
 # Generated by $as_me.
 # Run this file to recreate the current configuration.
@@ -6203,17 +5339,18 @@ cat >$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 debug=false
 ac_cs_recheck=false
 ac_cs_silent=false
-SHELL=\${CONFIG_SHELL-$SHELL}
-_ACEOF
 
-cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-## --------------------- ##
-## M4sh Initialization.  ##
-## --------------------- ##
+SHELL=\${CONFIG_SHELL-$SHELL}
+export SHELL
+_ASEOF
+cat >>$CONFIG_STATUS <<\_ASEOF || as_write_fail=1
+## -------------------- ##
+## M4sh Initialization. ##
+## -------------------- ##
 
 # Be more Bourne compatible
 DUALCASE=1; export DUALCASE # for MKS sh
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
   emulate sh
   NULLCMD=:
   # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
@@ -6221,23 +5358,15 @@ if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
   alias -g '${1+"$@"}'='"$@"'
   setopt NO_GLOB_SUBST
 else
-  case `(set -o) 2>/dev/null` in
-  *posix*) set -o posix ;;
+  case `(set -o) 2>/dev/null` in #(
+  *posix*) :
+    set -o posix ;; #(
+  *) :
+     ;;
 esac
-
 fi
 
 
-
-
-# PATH needs CR
-# Avoid depending upon Character Ranges.
-as_cr_letters='abcdefghijklmnopqrstuvwxyz'
-as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
-as_cr_Letters=$as_cr_letters$as_cr_LETTERS
-as_cr_digits='0123456789'
-as_cr_alnum=$as_cr_Letters$as_cr_digits
-
 as_nl='
 '
 export as_nl
@@ -6245,7 +5374,13 @@ export as_nl
 as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
-if (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+# Prefer a ksh shell builtin over an external printf program on Solaris,
+# but without wasting forks for bash or zsh.
+if test -z "$BASH_VERSION$ZSH_VERSION" \
+    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
+  as_echo='print -r --'
+  as_echo_n='print -rn --'
+elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
   as_echo='printf %s\n'
   as_echo_n='printf %s'
 else
@@ -6256,7 +5391,7 @@ else
     as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
     as_echo_n_body='eval
       arg=$1;
-      case $arg in
+      case $arg in #(
       *"$as_nl"*)
 	expr "X$arg" : "X\\(.*\\)$as_nl";
 	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
@@ -6279,13 +5414,6 @@ if test "${PATH_SEPARATOR+set}" != set; then
   }
 fi
 
-# Support unset when possible.
-if ( (MAIL=60; unset MAIL) || exit) >/dev/null 2>&1; then
-  as_unset=unset
-else
-  as_unset=false
-fi
-
 
 # IFS
 # We need space, tab and new line, in precisely that order.  Quoting is
@@ -6295,15 +5423,16 @@ fi
 IFS=" ""	$as_nl"
 
 # Find who we are.  Look in the path if we contain no directory separator.
-case $0 in
+as_myself=
+case $0 in #((
   *[\\/]* ) as_myself=$0 ;;
   *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
 for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
-done
+    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+  done
 IFS=$as_save_IFS
 
      ;;
@@ -6315,12 +5444,16 @@ if test "x$as_myself" = x; then
 fi
 if test ! -f "$as_myself"; then
   $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
-  { (exit 1); exit 1; }
+  exit 1
 fi
 
-# Work around bugs in pre-3.0 UWIN ksh.
-for as_var in ENV MAIL MAILPATH
-do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
+# Unset variables that we do not need and which cause bugs (e.g. in
+# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
+# suppresses any "Segmentation fault" message there.  '((' could
+# trigger a bug in pdksh 5.2.14.
+for as_var in BASH_ENV ENV MAIL MAILPATH
+do eval test x\${$as_var+set} = xset \
+  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
 done
 PS1='$ '
 PS2='> '
@@ -6332,7 +5465,89 @@ export LC_ALL
 LANGUAGE=C
 export LANGUAGE
 
-# Required to use basename.
+# CDPATH.
+(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
+
+
+# as_fn_error STATUS ERROR [LINENO LOG_FD]
+# ----------------------------------------
+# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
+# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
+# script with STATUS, using 1 if that was 0.
+as_fn_error ()
+{
+  as_status=$1; test $as_status -eq 0 && as_status=1
+  if test "$4"; then
+    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
+  fi
+  $as_echo "$as_me: error: $2" >&2
+  as_fn_exit $as_status
+} # as_fn_error
+
+
+# as_fn_set_status STATUS
+# -----------------------
+# Set $? to STATUS, without forking.
+as_fn_set_status ()
+{
+  return $1
+} # as_fn_set_status
+
+# as_fn_exit STATUS
+# -----------------
+# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
+as_fn_exit ()
+{
+  set +e
+  as_fn_set_status $1
+  exit $1
+} # as_fn_exit
+
+# as_fn_unset VAR
+# ---------------
+# Portably unset VAR.
+as_fn_unset ()
+{
+  { eval $1=; unset $1;}
+}
+as_unset=as_fn_unset
+# as_fn_append VAR VALUE
+# ----------------------
+# Append the text in VALUE to the end of the definition contained in VAR. Take
+# advantage of any shell optimizations that allow amortized linear growth over
+# repeated appends, instead of the typical quadratic growth present in naive
+# implementations.
+if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
+  eval 'as_fn_append ()
+  {
+    eval $1+=\$2
+  }'
+else
+  as_fn_append ()
+  {
+    eval $1=\$$1\$2
+  }
+fi # as_fn_append
+
+# as_fn_arith ARG...
+# ------------------
+# Perform arithmetic evaluation on the ARGs, and store the result in the
+# global $as_val. Take advantage of shells that can avoid forks. The arguments
+# must be portable across $(()) and expr.
+if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
+  eval 'as_fn_arith ()
+  {
+    as_val=$(( $* ))
+  }'
+else
+  as_fn_arith ()
+  {
+    as_val=`expr "$@" || test $? -eq 1`
+  }
+fi # as_fn_arith
+
+
 if expr a : '\(a\)' >/dev/null 2>&1 &&
    test "X`expr 00001 : '.*\(...\)'`" = X001; then
   as_expr=expr
@@ -6346,8 +5561,12 @@ else
   as_basename=false
 fi
 
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
 
-# Name of the executable.
 as_me=`$as_basename -- "$0" ||
 $as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
 	 X"$0" : 'X\(//\)$' \| \
@@ -6367,76 +5586,25 @@ $as_echo X/"$0" |
 	  }
 	  s/.*/./; q'`
 
-# CDPATH.
-$as_unset CDPATH
-
-
-
-  as_lineno_1=$LINENO
-  as_lineno_2=$LINENO
-  test "x$as_lineno_1" != "x$as_lineno_2" &&
-  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2" || {
-
-  # Create $as_me.lineno as a copy of $as_myself, but with $LINENO
-  # uniformly replaced by the line number.  The first 'sed' inserts a
-  # line-number line after each line using $LINENO; the second 'sed'
-  # does the real work.  The second script uses 'N' to pair each
-  # line-number line with the line containing $LINENO, and appends
-  # trailing '-' during substitution so that $LINENO is not a special
-  # case at line end.
-  # (Raja R Harinath suggested sed '=', and Paul Eggert wrote the
-  # scripts with optimization help from Paolo Bonzini.  Blame Lee
-  # E. McMahon (1931-1989) for sed's syntax.  :-)
-  sed -n '
-    p
-    /[$]LINENO/=
-  ' <$as_myself |
-    sed '
-      s/[$]LINENO.*/&-/
-      t lineno
-      b
-      :lineno
-      N
-      :loop
-      s/[$]LINENO\([^'$as_cr_alnum'_].*\n\)\(.*\)/\2\1\2/
-      t loop
-      s/-\n.*//
-    ' >$as_me.lineno &&
-  chmod +x "$as_me.lineno" ||
-    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2
-   { (exit 1); exit 1; }; }
-
-  # Don't try to exec as it changes $[0], causing all sort of problems
-  # (the dirname of $[0] is not the place where we might find the
-  # original and so on.  Autoconf is especially sensitive to this).
-  . "./$as_me.lineno"
-  # Exit status is that of the last command.
-  exit
-}
-
-
-if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
-  as_dirname=dirname
-else
-  as_dirname=false
-fi
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
 
 ECHO_C= ECHO_N= ECHO_T=
-case `echo -n x` in
+case `echo -n x` in #(((((
 -n*)
-  case `echo 'x\c'` in
+  case `echo 'xy\c'` in
   *c*) ECHO_T='	';;	# ECHO_T is single tab character.
-  *)   ECHO_C='\c';;
+  xy)  ECHO_C='\c';;
+  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
+       ECHO_T='	';;
   esac;;
 *)
   ECHO_N='-n';;
 esac
-if expr a : '\(a\)' >/dev/null 2>&1 &&
-   test "X`expr 00001 : '.*\(...\)'`" = X001; then
-  as_expr=expr
-else
-  as_expr=false
-fi
 
 rm -f conf$$ conf$$.exe conf$$.file
 if test -d conf$$.dir; then
@@ -6451,49 +5619,85 @@ if (echo >conf$$.file) 2>/dev/null; then
     # ... but there are two gotchas:
     # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
     # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
-    # In both cases, we have to default to `cp -p'.
+    # In both cases, we have to default to `cp -pR'.
     ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
-      as_ln_s='cp -p'
+      as_ln_s='cp -pR'
   elif ln conf$$.file conf$$ 2>/dev/null; then
     as_ln_s=ln
   else
-    as_ln_s='cp -p'
+    as_ln_s='cp -pR'
   fi
 else
-  as_ln_s='cp -p'
+  as_ln_s='cp -pR'
 fi
 rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
 rmdir conf$$.dir 2>/dev/null
 
+
+# as_fn_mkdir_p
+# -------------
+# Create "$as_dir" as a directory, including parents if necessary.
+as_fn_mkdir_p ()
+{
+
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || eval $as_mkdir_p || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
+
+
+} # as_fn_mkdir_p
 if mkdir -p . 2>/dev/null; then
-  as_mkdir_p=:
+  as_mkdir_p='mkdir -p "$as_dir"'
 else
   test -d ./-p && rmdir ./-p
   as_mkdir_p=false
 fi
 
-if test -x / >/dev/null 2>&1; then
-  as_test_x='test -x'
-else
-  if ls -dL / >/dev/null 2>&1; then
-    as_ls_L_option=L
-  else
-    as_ls_L_option=
-  fi
-  as_test_x='
-    eval sh -c '\''
-      if test -d "$1"; then
-	test -d "$1/.";
-      else
-	case $1 in
-	-*)set "./$1";;
-	esac;
-	case `ls -ld'$as_ls_L_option' "$1" 2>/dev/null` in
-	???[sx]*):;;*)false;;esac;fi
-    '\'' sh
-  '
-fi
-as_executable_p=$as_test_x
+
+# as_fn_executable_p FILE
+# -----------------------
+# Test if FILE is an executable regular file.
+as_fn_executable_p ()
+{
+  test -f "$1" && test -x "$1"
+} # as_fn_executable_p
+as_test_x='test -x'
+as_executable_p=as_fn_executable_p
 
 # Sed expression to map a string onto a valid CPP name.
 as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
@@ -6503,13 +5707,19 @@ as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
 
 
 exec 6>&1
+## ----------------------------------- ##
+## Main body of $CONFIG_STATUS script. ##
+## ----------------------------------- ##
+_ASEOF
+test $as_write_fail = 0 && chmod +x $CONFIG_STATUS || ac_write_fail=1
 
-# Save the log message, to keep $[0] and so on meaningful, and to
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+# Save the log message, to keep $0 and so on meaningful, and to
 # report actual input values of CONFIG_FILES etc. instead of their
 # values after options handling.
 ac_log="
 This file was extended by Grid $as_me 1.0, which was
-generated by GNU Autoconf 2.63.  Invocation command line was
+generated by GNU Autoconf 2.69.  Invocation command line was
 
   CONFIG_FILES    = $CONFIG_FILES
   CONFIG_HEADERS  = $CONFIG_HEADERS
@@ -6541,13 +5751,15 @@ _ACEOF
 
 cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 ac_cs_usage="\
-\`$as_me' instantiates files from templates according to the
-current configuration.
+\`$as_me' instantiates files and other configuration actions
+from templates according to the current configuration.  Unless the files
+and actions are specified as TAGs, all are instantiated by default.
 
-Usage: $0 [OPTION]... [FILE]...
+Usage: $0 [OPTION]... [TAG]...
 
   -h, --help       print this help, then exit
   -V, --version    print version number and configuration settings, then exit
+      --config     print configuration, then exit
   -q, --quiet, --silent
                    do not print progress messages
   -d, --debug      don't remove temporary files
@@ -6566,16 +5778,17 @@ $config_headers
 Configuration commands:
 $config_commands
 
-Report bugs to <bug-autoconf@gnu.org>."
+Report bugs to <paboyle@ph.ed.ac.uk>."
 
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+ac_cs_config="`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`"
 ac_cs_version="\\
 Grid config.status 1.0
-configured by $0, generated by GNU Autoconf 2.63,
-  with options \\"`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`\\"
+configured by $0, generated by GNU Autoconf 2.69,
+  with options \\"\$ac_cs_config\\"
 
-Copyright (C) 2008 Free Software Foundation, Inc.
+Copyright (C) 2012 Free Software Foundation, Inc.
 This config.status script is free software; the Free Software Foundation
 gives unlimited permission to copy, distribute and modify it."
 
@@ -6593,11 +5806,16 @@ ac_need_defaults=:
 while test $# != 0
 do
   case $1 in
-  --*=*)
+  --*=?*)
     ac_option=`expr "X$1" : 'X\([^=]*\)='`
     ac_optarg=`expr "X$1" : 'X[^=]*=\(.*\)'`
     ac_shift=:
     ;;
+  --*=)
+    ac_option=`expr "X$1" : 'X\([^=]*\)='`
+    ac_optarg=
+    ac_shift=:
+    ;;
   *)
     ac_option=$1
     ac_optarg=$2
@@ -6611,27 +5829,29 @@ do
     ac_cs_recheck=: ;;
   --version | --versio | --versi | --vers | --ver | --ve | --v | -V )
     $as_echo "$ac_cs_version"; exit ;;
+  --config | --confi | --conf | --con | --co | --c )
+    $as_echo "$ac_cs_config"; exit ;;
   --debug | --debu | --deb | --de | --d | -d )
     debug=: ;;
   --file | --fil | --fi | --f )
     $ac_shift
     case $ac_optarg in
     *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
+    '') as_fn_error $? "missing file argument" ;;
     esac
-    CONFIG_FILES="$CONFIG_FILES '$ac_optarg'"
+    as_fn_append CONFIG_FILES " '$ac_optarg'"
     ac_need_defaults=false;;
   --header | --heade | --head | --hea )
     $ac_shift
     case $ac_optarg in
     *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
     esac
-    CONFIG_HEADERS="$CONFIG_HEADERS '$ac_optarg'"
+    as_fn_append CONFIG_HEADERS " '$ac_optarg'"
     ac_need_defaults=false;;
   --he | --h)
     # Conflict between --help and --header
-    { $as_echo "$as_me: error: ambiguous option: $1
-Try \`$0 --help' for more information." >&2
-   { (exit 1); exit 1; }; };;
+    as_fn_error $? "ambiguous option: \`$1'
+Try \`$0 --help' for more information.";;
   --help | --hel | -h )
     $as_echo "$ac_cs_usage"; exit ;;
   -q | -quiet | --quiet | --quie | --qui | --qu | --q \
@@ -6639,11 +5859,10 @@ Try \`$0 --help' for more information." >&2
     ac_cs_silent=: ;;
 
   # This is an error.
-  -*) { $as_echo "$as_me: error: unrecognized option: $1
-Try \`$0 --help' for more information." >&2
-   { (exit 1); exit 1; }; } ;;
+  -*) as_fn_error $? "unrecognized option: \`$1'
+Try \`$0 --help' for more information." ;;
 
-  *) ac_config_targets="$ac_config_targets $1"
+  *) as_fn_append ac_config_targets " $1"
      ac_need_defaults=false ;;
 
   esac
@@ -6660,7 +5879,7 @@ fi
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 if \$ac_cs_recheck; then
-  set X '$SHELL' '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
+  set X $SHELL '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
   shift
   \$as_echo "running CONFIG_SHELL=$SHELL \$*" >&6
   CONFIG_SHELL='$SHELL'
@@ -6701,9 +5920,7 @@ do
     "tests/Makefile") CONFIG_FILES="$CONFIG_FILES tests/Makefile" ;;
     "benchmarks/Makefile") CONFIG_FILES="$CONFIG_FILES benchmarks/Makefile" ;;
 
-  *) { { $as_echo "$as_me:$LINENO: error: invalid argument: $ac_config_target" >&5
-$as_echo "$as_me: error: invalid argument: $ac_config_target" >&2;}
-   { (exit 1); exit 1; }; };;
+  *) as_fn_error $? "invalid argument: \`$ac_config_target'" "$LINENO" 5;;
   esac
 done
 
@@ -6726,26 +5943,24 @@ fi
 # after its creation but before its name has been assigned to `$tmp'.
 $debug ||
 {
-  tmp=
+  tmp= ac_tmp=
   trap 'exit_status=$?
-  { test -z "$tmp" || test ! -d "$tmp" || rm -fr "$tmp"; } && exit $exit_status
+  : "${ac_tmp:=$tmp}"
+  { test ! -d "$ac_tmp" || rm -fr "$ac_tmp"; } && exit $exit_status
 ' 0
-  trap '{ (exit 1); exit 1; }' 1 2 13 15
+  trap 'as_fn_exit 1' 1 2 13 15
 }
 # Create a (secure) tmp directory for tmp files.
 
 {
   tmp=`(umask 077 && mktemp -d "./confXXXXXX") 2>/dev/null` &&
-  test -n "$tmp" && test -d "$tmp"
+  test -d "$tmp"
 }  ||
 {
   tmp=./conf$$-$RANDOM
   (umask 077 && mkdir "$tmp")
-} ||
-{
-   $as_echo "$as_me: cannot create a temporary directory in ." >&2
-   { (exit 1); exit 1; }
-}
+} || as_fn_error $? "cannot create a temporary directory in ." "$LINENO" 5
+ac_tmp=$tmp
 
 # Set up the scripts for CONFIG_FILES section.
 # No need to generate them if there are no CONFIG_FILES.
@@ -6753,7 +5968,13 @@ $debug ||
 if test -n "$CONFIG_FILES"; then
 
 
-ac_cr='
'
+ac_cr=`echo X | tr X '\015'`
+# On cygwin, bash can eat \r inside `` if the user requested igncr.
+# But we know of no other shell where ac_cr would be empty at this
+# point, so we can use a bashism as a fallback.
+if test "x$ac_cr" = x; then
+  eval ac_cr=\$\'\\r\'
+fi
 ac_cs_awk_cr=`$AWK 'BEGIN { print "a\rb" }' </dev/null 2>/dev/null`
 if test "$ac_cs_awk_cr" = "a${ac_cr}b"; then
   ac_cs_awk_cr='\\r'
@@ -6761,7 +5982,7 @@ else
   ac_cs_awk_cr=$ac_cr
 fi
 
-echo 'BEGIN {' >"$tmp/subs1.awk" &&
+echo 'BEGIN {' >"$ac_tmp/subs1.awk" &&
 _ACEOF
 
 
@@ -6770,24 +5991,18 @@ _ACEOF
   echo "$ac_subst_vars" | sed 's/.*/&!$&$ac_delim/' &&
   echo "_ACEOF"
 } >conf$$subs.sh ||
-  { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
-$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
-   { (exit 1); exit 1; }; }
-ac_delim_num=`echo "$ac_subst_vars" | grep -c '$'`
+  as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
+ac_delim_num=`echo "$ac_subst_vars" | grep -c '^'`
 ac_delim='%!_!# '
 for ac_last_try in false false false false false :; do
   . ./conf$$subs.sh ||
-    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
-$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
 
   ac_delim_n=`sed -n "s/.*$ac_delim\$/X/p" conf$$subs.awk | grep -c X`
   if test $ac_delim_n = $ac_delim_num; then
     break
   elif $ac_last_try; then
-    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
-$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
   else
     ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
   fi
@@ -6795,7 +6010,7 @@ done
 rm -f conf$$subs.sh
 
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-cat >>"\$tmp/subs1.awk" <<\\_ACAWK &&
+cat >>"\$ac_tmp/subs1.awk" <<\\_ACAWK &&
 _ACEOF
 sed -n '
 h
@@ -6809,7 +6024,7 @@ s/'"$ac_delim"'$//
 t delim
 :nl
 h
-s/\(.\{148\}\).*/\1/
+s/\(.\{148\}\)..*/\1/
 t more1
 s/["\\]/\\&/g; s/^/"/; s/$/\\n"\\/
 p
@@ -6823,7 +6038,7 @@ s/.\{148\}//
 t nl
 :delim
 h
-s/\(.\{148\}\).*/\1/
+s/\(.\{148\}\)..*/\1/
 t more2
 s/["\\]/\\&/g; s/^/"/; s/$/"/
 p
@@ -6843,7 +6058,7 @@ t delim
 rm -f conf$$subs.awk
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 _ACAWK
-cat >>"\$tmp/subs1.awk" <<_ACAWK &&
+cat >>"\$ac_tmp/subs1.awk" <<_ACAWK &&
   for (key in S) S_is_set[key] = 1
   FS = ""
 
@@ -6875,23 +6090,29 @@ if sed "s/$ac_cr//" < /dev/null > /dev/null 2>&1; then
   sed "s/$ac_cr\$//; s/$ac_cr/$ac_cs_awk_cr/g"
 else
   cat
-fi < "$tmp/subs1.awk" > "$tmp/subs.awk" \
-  || { { $as_echo "$as_me:$LINENO: error: could not setup config files machinery" >&5
-$as_echo "$as_me: error: could not setup config files machinery" >&2;}
-   { (exit 1); exit 1; }; }
+fi < "$ac_tmp/subs1.awk" > "$ac_tmp/subs.awk" \
+  || as_fn_error $? "could not setup config files machinery" "$LINENO" 5
 _ACEOF
 
-# VPATH may cause trouble with some makes, so we remove $(srcdir),
-# ${srcdir} and @srcdir@ from VPATH if srcdir is ".", strip leading and
+# VPATH may cause trouble with some makes, so we remove sole $(srcdir),
+# ${srcdir} and @srcdir@ entries from VPATH if srcdir is ".", strip leading and
 # trailing colons and then remove the whole line if VPATH becomes empty
 # (actually we leave an empty line to preserve line numbers).
 if test "x$srcdir" = x.; then
-  ac_vpsub='/^[	 ]*VPATH[	 ]*=/{
-s/:*\$(srcdir):*/:/
-s/:*\${srcdir}:*/:/
-s/:*@srcdir@:*/:/
-s/^\([^=]*=[	 ]*\):*/\1/
+  ac_vpsub='/^[	 ]*VPATH[	 ]*=[	 ]*/{
+h
+s///
+s/^/:/
+s/[	 ]*$/:/
+s/:\$(srcdir):/:/g
+s/:\${srcdir}:/:/g
+s/:@srcdir@:/:/g
+s/^:*//
 s/:*$//
+x
+s/\(=[	 ]*\).*/\1/
+G
+s/\n//
 s/^[^=]*=[	 ]*$//
 }'
 fi
@@ -6903,7 +6124,7 @@ fi # test -n "$CONFIG_FILES"
 # No need to generate them if there are no CONFIG_HEADERS.
 # This happens for instance with `./config.status Makefile'.
 if test -n "$CONFIG_HEADERS"; then
-cat >"$tmp/defines.awk" <<\_ACAWK ||
+cat >"$ac_tmp/defines.awk" <<\_ACAWK ||
 BEGIN {
 _ACEOF
 
@@ -6915,13 +6136,11 @@ _ACEOF
 # handling of long lines.
 ac_delim='%!_!# '
 for ac_last_try in false false :; do
-  ac_t=`sed -n "/$ac_delim/p" confdefs.h`
-  if test -z "$ac_t"; then
+  ac_tt=`sed -n "/$ac_delim/p" confdefs.h`
+  if test -z "$ac_tt"; then
     break
   elif $ac_last_try; then
-    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_HEADERS" >&5
-$as_echo "$as_me: error: could not make $CONFIG_HEADERS" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "could not make $CONFIG_HEADERS" "$LINENO" 5
   else
     ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
   fi
@@ -7006,9 +6225,7 @@ cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 _ACAWK
 _ACEOF
 cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-  { { $as_echo "$as_me:$LINENO: error: could not setup config headers machinery" >&5
-$as_echo "$as_me: error: could not setup config headers machinery" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "could not setup config headers machinery" "$LINENO" 5
 fi # test -n "$CONFIG_HEADERS"
 
 
@@ -7021,9 +6238,7 @@ do
   esac
   case $ac_mode$ac_tag in
   :[FHL]*:*);;
-  :L* | :C*:*) { { $as_echo "$as_me:$LINENO: error: invalid tag $ac_tag" >&5
-$as_echo "$as_me: error: invalid tag $ac_tag" >&2;}
-   { (exit 1); exit 1; }; };;
+  :L* | :C*:*) as_fn_error $? "invalid tag \`$ac_tag'" "$LINENO" 5;;
   :[FH]-) ac_tag=-:-;;
   :[FH]*) ac_tag=$ac_tag:$ac_tag.in;;
   esac
@@ -7042,7 +6257,7 @@ $as_echo "$as_me: error: invalid tag $ac_tag" >&2;}
     for ac_f
     do
       case $ac_f in
-      -) ac_f="$tmp/stdin";;
+      -) ac_f="$ac_tmp/stdin";;
       *) # Look for the file first in the build tree, then in the source tree
 	 # (if the path is not absolute).  The absolute path cannot be DOS-style,
 	 # because $ac_f cannot contain `:'.
@@ -7051,12 +6266,10 @@ $as_echo "$as_me: error: invalid tag $ac_tag" >&2;}
 	   [\\/$]*) false;;
 	   *) test -f "$srcdir/$ac_f" && ac_f="$srcdir/$ac_f";;
 	   esac ||
-	   { { $as_echo "$as_me:$LINENO: error: cannot find input file: $ac_f" >&5
-$as_echo "$as_me: error: cannot find input file: $ac_f" >&2;}
-   { (exit 1); exit 1; }; };;
+	   as_fn_error 1 "cannot find input file: \`$ac_f'" "$LINENO" 5;;
       esac
       case $ac_f in *\'*) ac_f=`$as_echo "$ac_f" | sed "s/'/'\\\\\\\\''/g"`;; esac
-      ac_file_inputs="$ac_file_inputs '$ac_f'"
+      as_fn_append ac_file_inputs " '$ac_f'"
     done
 
     # Let's still pretend it is `configure' which instantiates (i.e., don't
@@ -7067,7 +6280,7 @@ $as_echo "$as_me: error: cannot find input file: $ac_f" >&2;}
 	`' by configure.'
     if test x"$ac_file" != x-; then
       configure_input="$ac_file.  $configure_input"
-      { $as_echo "$as_me:$LINENO: creating $ac_file" >&5
+      { $as_echo "$as_me:${as_lineno-$LINENO}: creating $ac_file" >&5
 $as_echo "$as_me: creating $ac_file" >&6;}
     fi
     # Neutralize special characters interpreted by sed in replacement strings.
@@ -7079,10 +6292,8 @@ $as_echo "$as_me: creating $ac_file" >&6;}
     esac
 
     case $ac_tag in
-    *:-:* | *:-) cat >"$tmp/stdin" \
-      || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; } ;;
+    *:-:* | *:-) cat >"$ac_tmp/stdin" \
+      || as_fn_error $? "could not create $ac_file" "$LINENO" 5 ;;
     esac
     ;;
   esac
@@ -7110,47 +6321,7 @@ $as_echo X"$ac_file" |
 	    q
 	  }
 	  s/.*/./; q'`
-  { as_dir="$ac_dir"
-  case $as_dir in #(
-  -*) as_dir=./$as_dir;;
-  esac
-  test -d "$as_dir" || { $as_mkdir_p && mkdir -p "$as_dir"; } || {
-    as_dirs=
-    while :; do
-      case $as_dir in #(
-      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
-      *) as_qdir=$as_dir;;
-      esac
-      as_dirs="'$as_qdir' $as_dirs"
-      as_dir=`$as_dirname -- "$as_dir" ||
-$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_dir" : 'X\(//\)[^/]' \| \
-	 X"$as_dir" : 'X\(//\)$' \| \
-	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_dir" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      test -d "$as_dir" && break
-    done
-    test -z "$as_dirs" || eval "mkdir $as_dirs"
-  } || test -d "$as_dir" || { { $as_echo "$as_me:$LINENO: error: cannot create directory $as_dir" >&5
-$as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
-   { (exit 1); exit 1; }; }; }
+  as_dir="$ac_dir"; as_fn_mkdir_p
   ac_builddir=.
 
 case "$ac_dir" in
@@ -7207,7 +6378,6 @@ cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 # If the template does not know about datarootdir, expand it.
 # FIXME: This hack should be removed a few years after 2.60.
 ac_datarootdir_hack=; ac_datarootdir_seen=
-
 ac_sed_dataroot='
 /datarootdir/ {
   p
@@ -7217,12 +6387,11 @@ ac_sed_dataroot='
 /@docdir@/p
 /@infodir@/p
 /@localedir@/p
-/@mandir@/p
-'
+/@mandir@/p'
 case `eval "sed -n \"\$ac_sed_dataroot\" $ac_file_inputs"` in
 *datarootdir*) ac_datarootdir_seen=yes;;
 *@datadir@*|*@docdir@*|*@infodir@*|*@localedir@*|*@mandir@*)
-  { $as_echo "$as_me:$LINENO: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
 $as_echo "$as_me: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&2;}
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
@@ -7232,7 +6401,7 @@ cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
   s&@infodir@&$infodir&g
   s&@localedir@&$localedir&g
   s&@mandir@&$mandir&g
-    s&\\\${datarootdir}&$datarootdir&g' ;;
+  s&\\\${datarootdir}&$datarootdir&g' ;;
 esac
 _ACEOF
 
@@ -7260,27 +6429,24 @@ s&@INSTALL@&$ac_INSTALL&;t t
 s&@MKDIR_P@&$ac_MKDIR_P&;t t
 $ac_datarootdir_hack
 "
-eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$tmp/subs.awk" >$tmp/out \
-  || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; }
+eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$ac_tmp/subs.awk" \
+  >$ac_tmp/out || as_fn_error $? "could not create $ac_file" "$LINENO" 5
 
 test -z "$ac_datarootdir_hack$ac_datarootdir_seen" &&
-  { ac_out=`sed -n '/\${datarootdir}/p' "$tmp/out"`; test -n "$ac_out"; } &&
-  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' "$tmp/out"`; test -z "$ac_out"; } &&
-  { $as_echo "$as_me:$LINENO: WARNING: $ac_file contains a reference to the variable \`datarootdir'
-which seems to be undefined.  Please make sure it is defined." >&5
+  { ac_out=`sed -n '/\${datarootdir}/p' "$ac_tmp/out"`; test -n "$ac_out"; } &&
+  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' \
+      "$ac_tmp/out"`; test -z "$ac_out"; } &&
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file contains a reference to the variable \`datarootdir'
+which seems to be undefined.  Please make sure it is defined" >&5
 $as_echo "$as_me: WARNING: $ac_file contains a reference to the variable \`datarootdir'
-which seems to be undefined.  Please make sure it is defined." >&2;}
+which seems to be undefined.  Please make sure it is defined" >&2;}
 
-  rm -f "$tmp/stdin"
+  rm -f "$ac_tmp/stdin"
   case $ac_file in
-  -) cat "$tmp/out" && rm -f "$tmp/out";;
-  *) rm -f "$ac_file" && mv "$tmp/out" "$ac_file";;
+  -) cat "$ac_tmp/out" && rm -f "$ac_tmp/out";;
+  *) rm -f "$ac_file" && mv "$ac_tmp/out" "$ac_file";;
   esac \
-  || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; }
+  || as_fn_error $? "could not create $ac_file" "$LINENO" 5
  ;;
   :H)
   #
@@ -7289,27 +6455,21 @@ $as_echo "$as_me: error: could not create $ac_file" >&2;}
   if test x"$ac_file" != x-; then
     {
       $as_echo "/* $configure_input  */" \
-      && eval '$AWK -f "$tmp/defines.awk"' "$ac_file_inputs"
-    } >"$tmp/config.h" \
-      || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; }
-    if diff "$ac_file" "$tmp/config.h" >/dev/null 2>&1; then
-      { $as_echo "$as_me:$LINENO: $ac_file is unchanged" >&5
+      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs"
+    } >"$ac_tmp/config.h" \
+      || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+    if diff "$ac_file" "$ac_tmp/config.h" >/dev/null 2>&1; then
+      { $as_echo "$as_me:${as_lineno-$LINENO}: $ac_file is unchanged" >&5
 $as_echo "$as_me: $ac_file is unchanged" >&6;}
     else
       rm -f "$ac_file"
-      mv "$tmp/config.h" "$ac_file" \
-	|| { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; }
+      mv "$ac_tmp/config.h" "$ac_file" \
+	|| as_fn_error $? "could not create $ac_file" "$LINENO" 5
     fi
   else
     $as_echo "/* $configure_input  */" \
-      && eval '$AWK -f "$tmp/defines.awk"' "$ac_file_inputs" \
-      || { { $as_echo "$as_me:$LINENO: error: could not create -" >&5
-$as_echo "$as_me: error: could not create -" >&2;}
-   { (exit 1); exit 1; }; }
+      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs" \
+      || as_fn_error $? "could not create -" "$LINENO" 5
   fi
 # Compute "$ac_file"'s index in $config_headers.
 _am_arg="$ac_file"
@@ -7347,7 +6507,7 @@ $as_echo X"$_am_arg" |
 	  s/.*/./; q'`/stamp-h$_am_stamp_count
  ;;
 
-  :C)  { $as_echo "$as_me:$LINENO: executing $ac_file commands" >&5
+  :C)  { $as_echo "$as_me:${as_lineno-$LINENO}: executing $ac_file commands" >&5
 $as_echo "$as_me: executing $ac_file commands" >&6;}
  ;;
   esac
@@ -7355,7 +6515,7 @@ $as_echo "$as_me: executing $ac_file commands" >&6;}
 
   case $ac_file$ac_mode in
     "depfiles":C) test x"$AMDEP_TRUE" != x"" || {
-  # Autoconf 2.62 quotes --file arguments for eval, but not when files
+  # Older Autoconf quotes --file arguments for eval, but not when files
   # are listed without --file.  Let's play safe and only enable the eval
   # if we detect the quoting.
   case $CONFIG_FILES in
@@ -7368,7 +6528,7 @@ $as_echo "$as_me: executing $ac_file commands" >&6;}
     # Strip MF so we end up with the name of the file.
     mf=`echo "$mf" | sed -e 's/:.*$//'`
     # Check whether this is an Automake generated Makefile or not.
-    # We used to match only the files named `Makefile.in', but
+    # We used to match only the files named 'Makefile.in', but
     # some people rename them; so instead we look at the file content.
     # Grep'ing the first line is not enough: some people post-process
     # each Makefile.in and add a new line on top of each file to say so.
@@ -7402,21 +6562,19 @@ $as_echo X"$mf" |
       continue
     fi
     # Extract the definition of DEPDIR, am__include, and am__quote
-    # from the Makefile without running `make'.
+    # from the Makefile without running 'make'.
     DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
     test -z "$DEPDIR" && continue
     am__include=`sed -n 's/^am__include = //p' < "$mf"`
-    test -z "am__include" && continue
+    test -z "$am__include" && continue
     am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
-    # When using ansi2knr, U may be empty or an underscore; expand it
-    U=`sed -n 's/^U = //p' < "$mf"`
     # Find all dependency output files, they are included files with
     # $(DEPDIR) in their names.  We invoke sed twice because it is the
     # simplest approach to changing $(DEPDIR) to its actual value in the
     # expansion.
     for file in `sed -n "
       s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
-	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g' -e 's/\$U/'"$U"'/g'`; do
+	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
       # Make sure the directory exists.
       test -f "$dirpart/$file" && continue
       fdir=`$as_dirname -- "$file" ||
@@ -7442,47 +6600,7 @@ $as_echo X"$file" |
 	    q
 	  }
 	  s/.*/./; q'`
-      { as_dir=$dirpart/$fdir
-  case $as_dir in #(
-  -*) as_dir=./$as_dir;;
-  esac
-  test -d "$as_dir" || { $as_mkdir_p && mkdir -p "$as_dir"; } || {
-    as_dirs=
-    while :; do
-      case $as_dir in #(
-      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
-      *) as_qdir=$as_dir;;
-      esac
-      as_dirs="'$as_qdir' $as_dirs"
-      as_dir=`$as_dirname -- "$as_dir" ||
-$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_dir" : 'X\(//\)[^/]' \| \
-	 X"$as_dir" : 'X\(//\)$' \| \
-	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_dir" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      test -d "$as_dir" && break
-    done
-    test -z "$as_dirs" || eval "mkdir $as_dirs"
-  } || test -d "$as_dir" || { { $as_echo "$as_me:$LINENO: error: cannot create directory $as_dir" >&5
-$as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
-   { (exit 1); exit 1; }; }; }
+      as_dir=$dirpart/$fdir; as_fn_mkdir_p
       # echo "creating $dirpart/$file"
       echo '# dummy' > "$dirpart/$file"
     done
@@ -7494,15 +6612,12 @@ $as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
 done # for ac_tag
 
 
-{ (exit 0); exit 0; }
+as_fn_exit 0
 _ACEOF
-chmod +x $CONFIG_STATUS
 ac_clean_files=$ac_clean_files_save
 
 test $ac_write_fail = 0 ||
-  { { $as_echo "$as_me:$LINENO: error: write failure creating $CONFIG_STATUS" >&5
-$as_echo "$as_me: error: write failure creating $CONFIG_STATUS" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "write failure creating $CONFIG_STATUS" "$LINENO" 5
 
 
 # configure is writing to config.log, and then calls config.status.
@@ -7523,10 +6638,10 @@ if test "$no_create" != yes; then
   exec 5>>config.log
   # Use ||, not &&, to avoid exiting from the if with $? = 1, which
   # would make configure fail if this is the last instruction.
-  $ac_cs_success || { (exit 1); exit 1; }
+  $ac_cs_success || as_fn_exit 1
 fi
 if test -n "$ac_unrecognized_opts" && test "$enable_option_checking" != no; then
-  { $as_echo "$as_me:$LINENO: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
 $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2;}
 fi
 
diff --git a/lib/Grid_config.h b/lib/Grid_config.h
index fe3a5e60..c743bba0 100644
--- a/lib/Grid_config.h
+++ b/lib/Grid_config.h
@@ -11,21 +11,21 @@
 /* #undef AVX512 */
 
 /* GRID_COMMS_MPI */
-/* #undef GRID_COMMS_MPI */
+#define GRID_COMMS_MPI 1
 
 /* GRID_COMMS_NONE */
-#define GRID_COMMS_NONE 1
+/* #undef GRID_COMMS_NONE */
 
 /* Define to 1 if you have the declaration of `be64toh', and to 0 if you
    don't. */
-#define HAVE_DECL_BE64TOH 1
+#define HAVE_DECL_BE64TOH 0
 
 /* Define to 1 if you have the declaration of `ntohll', and to 0 if you don't.
    */
-#define HAVE_DECL_NTOHLL 0
+#define HAVE_DECL_NTOHLL 1
 
 /* Define to 1 if you have the <endian.h> header file. */
-#define HAVE_ENDIAN_H 1
+/* #undef HAVE_ENDIAN_H */
 
 /* Define to 1 if you have the `gettimeofday' function. */
 #define HAVE_GETTIMEOFDAY 1
@@ -34,10 +34,10 @@
 #define HAVE_INTTYPES_H 1
 
 /* Define to 1 if you have the <malloc.h> header file. */
-#define HAVE_MALLOC_H 1
+/* #undef HAVE_MALLOC_H */
 
 /* Define to 1 if you have the <malloc/malloc.h> header file. */
-/* #undef HAVE_MALLOC_MALLOC_H */
+#define HAVE_MALLOC_MALLOC_H 1
 
 /* Define to 1 if you have the <memory.h> header file. */
 #define HAVE_MEMORY_H 1
@@ -78,6 +78,9 @@
 /* Define to the one symbol short name of this package. */
 #define PACKAGE_TARNAME "grid"
 
+/* Define to the home page for this package. */
+#define PACKAGE_URL ""
+
 /* Define to the version of this package. */
 #define PACKAGE_VERSION "1.0"
 
diff --git a/lib/Grid_config.h.in b/lib/Grid_config.h.in
index 4b094250..437a6095 100644
--- a/lib/Grid_config.h.in
+++ b/lib/Grid_config.h.in
@@ -77,6 +77,9 @@
 /* Define to the one symbol short name of this package. */
 #undef PACKAGE_TARNAME
 
+/* Define to the home page for this package. */
+#undef PACKAGE_URL
+
 /* Define to the version of this package. */
 #undef PACKAGE_VERSION
 
diff --git a/lib/math/Grid_math_tensors.h b/lib/math/Grid_math_tensors.h
index 564d5cb6..09739555 100644
--- a/lib/math/Grid_math_tensors.h
+++ b/lib/math/Grid_math_tensors.h
@@ -57,6 +57,9 @@ public:
     friend void zeroit(iScalar<vtype> &that){
         zeroit(that._internal);
     }
+    friend void prefetch(iScalar<vtype> &that){
+      prefetch(that._internal);
+    }
     friend void permute(iScalar<vtype> &out,const iScalar<vtype> &in,int permutetype){
       permute(out._internal,in._internal,permutetype);
     }
@@ -141,6 +144,9 @@ public:
             zeroit(that._internal[i]);
         }
     }
+    friend void prefetch(iVector<vtype,N> &that){
+      for(int i=0;i<N;i++) prefetch(that._internal[i]);
+    }
     friend void vstream(iVector<vtype,N> &out,const iVector<vtype,N> &in){
       for(int i=0;i<N;i++){
 	vstream(out._internal[i],in._internal[i]);
@@ -219,6 +225,11 @@ public:
 	zeroit(that._internal[i][j]);
     }}
   }
+  friend void prefetch(iMatrix<vtype,N> &that){
+    for(int i=0;i<N;i++) 
+    for(int j=0;j<N;j++) 
+      prefetch(that._internal[i][j]);
+  }
   friend void vstream(iMatrix<vtype,N> &out,const iMatrix<vtype,N> &in){
       for(int i=0;i<N;i++){
       for(int j=0;j<N;j++){
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index ddc73c4a..37bc4bda 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -106,7 +106,9 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
   for(int sss=0;sss<grid->oSites();sss++){
 
     int ss = sss;
-	//int ss = Stencil._LebesgueReorder[sss];
+    int ssu= sss;
+    //int ss = 0;
+    //int ss = Stencil._LebesgueReorder[sss];
 
     // Xp
     offset = Stencil._offsets [Xp][ss];
@@ -123,7 +125,8 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ss](Xp),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Xp),&chi());
+    prefetch(Umu._odata[ssu](Yp));
     spReconXp(result,Uchi);
 
     // Yp
@@ -141,7 +144,8 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ss](Yp),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Yp),&chi());
+    prefetch(Umu._odata[ssu](Zp));
     accumReconYp(result,Uchi);
 
     // Zp
@@ -159,7 +163,8 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ss](Zp),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Zp),&chi());
+    prefetch(Umu._odata[ssu](Tp));
     accumReconZp(result,Uchi);
 
     // Tp
@@ -177,7 +182,8 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ss](Tp),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Tp),&chi());
+    prefetch(Umu._odata[ssu](Xm));
     accumReconTp(result,Uchi);
 
     // Xm
@@ -195,7 +201,8 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ss](Xm),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Xm),&chi());
+    prefetch(Umu._odata[ssu](Ym));
     accumReconXm(result,Uchi);
 
 
@@ -214,7 +221,8 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ss](Ym),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Ym),&chi());
+    prefetch(Umu._odata[ssu](Zm));
     accumReconYm(result,Uchi);
 
     // Zm
@@ -232,7 +240,8 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ss](Zm),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Zm),&chi());
+    prefetch(Umu._odata[ssu](Tm));
     accumReconZm(result,Uchi);
 
     // Tm
@@ -250,7 +259,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ss](Tm),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Tm),&chi());
     accumReconTm(result,Uchi);
 
     vstream(out._odata[ss],result);
diff --git a/lib/simd/Grid_vComplexD.h b/lib/simd/Grid_vComplexD.h
index f0108d59..5dbb28e5 100644
--- a/lib/simd/Grid_vComplexD.h
+++ b/lib/simd/Grid_vComplexD.h
@@ -257,7 +257,7 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
 	  assert(0);
 #endif
 	}
-      friend inline void vprefetch(const vComplexD &v)
+      friend inline void prefetch(const vComplexD &v)
         {
             _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
         }
diff --git a/lib/simd/Grid_vComplexF.h b/lib/simd/Grid_vComplexF.h
index 5f52fc53..66a1e1bf 100644
--- a/lib/simd/Grid_vComplexF.h
+++ b/lib/simd/Grid_vComplexF.h
@@ -206,7 +206,7 @@ namespace Grid {
 	  assert(0);
 #endif
 	}
-      friend inline void vprefetch(const vComplexF &v)
+      friend inline void prefetch(const vComplexF &v)
         {
             _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
         }
diff --git a/lib/simd/Grid_vInteger.h b/lib/simd/Grid_vInteger.h
index 5a429a9c..b7950873 100644
--- a/lib/simd/Grid_vInteger.h
+++ b/lib/simd/Grid_vInteger.h
@@ -190,7 +190,7 @@ namespace Grid {
 	  out=in;
         }
 
-        friend inline void vprefetch(const vInteger &v)
+        friend inline void prefetch(const vInteger &v)
         {
             _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
         }
diff --git a/lib/simd/Grid_vRealD.h b/lib/simd/Grid_vRealD.h
index ee51fa03..8041c1f9 100644
--- a/lib/simd/Grid_vRealD.h
+++ b/lib/simd/Grid_vRealD.h
@@ -191,7 +191,7 @@ namespace Grid {
 	  assert(0);
 #endif
 	}
-        friend inline void vprefetch(const vRealD &v)
+        friend inline void prefetch(const vRealD &v)
         {
             _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
         }
diff --git a/lib/simd/Grid_vRealF.h b/lib/simd/Grid_vRealF.h
index 48e5c134..e1d5679a 100644
--- a/lib/simd/Grid_vRealF.h
+++ b/lib/simd/Grid_vRealF.h
@@ -225,7 +225,7 @@ friend inline void vstore(const vRealF &ret, float *a){
 	}
 
 
-        friend inline void vprefetch(const vRealF &v)
+        friend inline void prefetch(const vRealF &v)
         {
             _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
         }
diff --git a/tests/Grid_gamma.cc b/tests/Grid_gamma.cc
index 9cda8855..3f350bc5 100644
--- a/tests/Grid_gamma.cc
+++ b/tests/Grid_gamma.cc
@@ -36,19 +36,19 @@ int main (int argc, char ** argv)
   SpinVector lv; random(sRNG,lv);
   SpinVector rv; random(sRNG,rv);
 
-  std::cout << " Is pod " << std::is_pod<SpinVector>::value  << std::endl;
-  std::cout << " Is pod double   " << std::is_pod<double>::value  << std::endl;
-  std::cout << " Is pod ComplexF " << std::is_pod<ComplexF>::value  << std::endl;
-  std::cout << " Is triv double " << std::has_trivial_default_constructor<double>::value  << std::endl;
-  std::cout << " Is triv ComplexF " << std::has_trivial_default_constructor<ComplexF>::value  << std::endl;
-  std::cout << " Is pod Scalar<double> " << std::is_pod<iScalar<double> >::value  << std::endl;
-  std::cout << " Is pod Scalar<ComplexF> " << std::is_pod<iScalar<ComplexF> >::value  << std::endl;
-  std::cout << " Is pod Scalar<vComplexF> " << std::is_pod<iScalar<vComplexF> >::value  << std::endl;
-  std::cout << " Is pod Scalar<vComplexD> " << std::is_pod<iScalar<vComplexD> >::value  << std::endl;
-  std::cout << " Is pod Scalar<vRealF> " << std::is_pod<iScalar<vRealF> >::value  << std::endl;
-  std::cout << " Is pod Scalar<vRealD> " << std::is_pod<iScalar<vRealD> >::value  << std::endl;
-  std::cout << " Is triv Scalar<double> " <<std::has_trivial_default_constructor<iScalar<double> >::value << std::endl;
-  std::cout << " Is triv Scalar<vComplexD> "<<std::has_trivial_default_constructor<iScalar<vComplexD> >::value  << std::endl;
+  //  std::cout << " Is pod " << std::is_pod<SpinVector>::value  << std::endl;
+  //  std::cout << " Is pod double   " << std::is_pod<double>::value  << std::endl;
+  //  std::cout << " Is pod ComplexF " << std::is_pod<ComplexF>::value  << std::endl;
+  //  std::cout << " Is triv double " << std::has_trivial_default_constructor<double>::value  << std::endl;
+  //  std::cout << " Is triv ComplexF " << std::has_trivial_default_constructor<ComplexF>::value  << std::endl;
+  //  std::cout << " Is pod Scalar<double> " << std::is_pod<iScalar<double> >::value  << std::endl;
+  //  std::cout << " Is pod Scalar<ComplexF> " << std::is_pod<iScalar<ComplexF> >::value  << std::endl;
+  //  std::cout << " Is pod Scalar<vComplexF> " << std::is_pod<iScalar<vComplexF> >::value  << std::endl;
+  //  std::cout << " Is pod Scalar<vComplexD> " << std::is_pod<iScalar<vComplexD> >::value  << std::endl;
+  //  std::cout << " Is pod Scalar<vRealF> " << std::is_pod<iScalar<vRealF> >::value  << std::endl;
+  //  std::cout << " Is pod Scalar<vRealD> " << std::is_pod<iScalar<vRealD> >::value  << std::endl;
+  //  std::cout << " Is triv Scalar<double> " <<std::has_trivial_default_constructor<iScalar<double> >::value << std::endl;
+  //  std::cout << " Is triv Scalar<vComplexD> "<<std::has_trivial_default_constructor<iScalar<vComplexD> >::value  << std::endl;
 
   for(int a=0;a<Ns;a++){
     ident()(a,a) = 1.0;

From 9ed1fb45e157d3a5de2f9feee8e39b55ae280e47 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@t002.ph.ed.ac.uk>
Date: Sun, 10 May 2015 15:13:50 +0100
Subject: [PATCH 126/429] Updated todo list

---
 TODO | 77 ++++++++++++++++++------------------------------------------
 1 file changed, 23 insertions(+), 54 deletions(-)

diff --git a/TODO b/TODO
index ee1dd897..9dc8fada 100644
--- a/TODO
+++ b/TODO
@@ -2,44 +2,17 @@
   - use protocol buffers? replace xmlReader/Writer ec..
   - Binary use htonll, htonl
 
-* Stencil -- DONE
+* CovariantShift support                             -----Use a class to store gauge field? (parallel transport?)
+
+* Strong test for norm2, conj and all primitive types. -- Grid_simd test is almost there
 
 * Reduce implemention is poor
 * Bug in SeedFixedIntegers gives same output on each site.
 * Bug in RNG with complex numbers ; only filling real values; need helper function -- DONE
 
-* Stencil operator support                           -----Initial thoughts, trial implementation DONE.
-                                                     -----some simple tests that Stencil matches Cshift.
-                                                     -----do all permute in comms phase, so that copy permute
-						     -----cases move into a buffer.
-						     -----allow transform in/out buffers spproj
-
-
-* CovariantShift support                             -----Use a class to store gauge field? (parallel transport?)
-
-* Strong test for norm2, conj and all primitive types.
-
-* Consider switch std::vector to boost arrays or something lighter weight
-  boost::multi_array<type, 3> A()...    to replace multi1d, multi2d etc..
-
-
 * norm2l is a hack. figure out syntax error and make this norm2 c.f. tests/Grid_gamma.cc
 
-****  std::vector replacement;
-
-  Had to change "reserve" back to "resize" on std::vector in Lattice class.
-
-  This forces the constructor call on EVERY element of the array with negative
-  performance effects on temporaries.
-  
-  The reversion was required because copy constructur has to work.
-
-  CONCLUSION: I must implement a similar to vector without construction/fill on
-  resize. Find out if valarray or alternative works differently prior to 
-  doing this since there may still be something I can use..
-
-
-* Make the Tensor types and Complex etc... play more nicely.
+** Make the Tensor types and Complex etc... play more nicely.
 
   - TensorRemove is a hack, come up with a long term rationalised approach to Complex vs. Scalar<Scalar<Scalar<Complex > > >
   QDP forces use of "toDouble" to get back to non tensor scalar. This role is presently taken TensorRemove, but I
@@ -60,24 +33,35 @@
  - I have collated into single location at least.
  - Need to use _mm_*insert/extract routines.
 
-
 * Conformable test in Cshift routines.
 
-
 * Flavour matrices?
 * Pauli, SU subgroup, etc.. 
 * su3 exponentiation & log etc.. [Jamie's code?]
 * TaProj
 
-
-* Fourspin, two spin project --- DONE
-
+* FFTnD ?
 
 * Parallel MPI2 IO
+  Plaquette checks into nersc reader.
 
 * rb4d support for 5th dimension in Mobius.
 
 * Check for missing functionality                    - partially audited against QDP++ layout
+   // Base class to share common code between vRealF, VComplexF etc...
+   //
+   // Unary functions
+   // cos,sin, tan, acos, asin, cosh, acosh, tanh, sinh, // Scalar<vReal> only arg
+   // exp, log, sqrt, fabs
+   //
+   // transposeColor, transposeSpin,
+   // adjColor, adjSpin,
+   //
+   // copyMask.
+   //
+   // localMaxAbs
+   //
+   // Fourier transform equivalent.
 
 Algorithms
 * LinearOperator
@@ -97,29 +81,14 @@ AUDITS:
 * Replace vset with a call to merge.; 
 * care in Gmerge,Gextract over vset .
 * extract / merge extra implementation removal      
-* Test infrastructure
-
-   // TODO
-   //
-   // Base class to share common code between vRealF, VComplexF etc...
-   //
-   // Unary functions
-   // cos,sin, tan, acos, asin, cosh, acosh, tanh, sinh, // Scalar<vReal> only arg
-   // exp, log, sqrt, fabs
-   //
-   // transposeColor, transposeSpin,
-   // adjColor, adjSpin,
-   //
-   // copyMask.
-   //
-   // localMaxAbs
-   //
-   // Fourier transform equivalent.
 
 
 ======================================================================================================
 
 FUNCTIONALITY:
+* Stencil -- DONE
+* Test infrastructure -- DONE
+* Fourspin, two spin project --- DONE
 * Dirac Gamma/Dirac structures ---- DONE
 * Conditional execution, where etc...                -----DONE, simple test
 * Integer relational support                         -----DONE

From 3657f2303dc96818f62a7ed01a34fb7cc3143f56 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@t002.ph.ed.ac.uk>
Date: Sun, 10 May 2015 15:18:04 +0100
Subject: [PATCH 127/429] ET ready benchmark with bytes counted assuming loop
 interchange

---
 benchmarks/Grid_memory_bandwidth.cc | 14 ++++++--------
 1 file changed, 6 insertions(+), 8 deletions(-)

diff --git a/benchmarks/Grid_memory_bandwidth.cc b/benchmarks/Grid_memory_bandwidth.cc
index 3254e95b..fb6db6d8 100644
--- a/benchmarks/Grid_memory_bandwidth.cc
+++ b/benchmarks/Grid_memory_bandwidth.cc
@@ -38,7 +38,6 @@ int main (int argc, char ** argv)
 
       double start=usecond();
       for(int i=0;i<Nloop;i++){
-	//	z=a*x+y;
 	//   inline void axpy(Lattice<vobj> &ret,double a,const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
 	axpy(z,a,x,y);
       }
@@ -53,7 +52,7 @@ int main (int argc, char ** argv)
   std::cout << "===================================================================================================="<<std::endl;
   std::cout << "= Benchmarking a*x + y bandwidth"<<std::endl;
   std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s"<<"\t\tGB/s (eff)"<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s"<<std::endl;
   std::cout << "----------------------------------------------------------"<<std::endl;
 
   
@@ -73,14 +72,13 @@ int main (int argc, char ** argv)
 
       double start=usecond();
       for(int i=0;i<Nloop;i++){
-	z=a*x+y;
+	z=a*x-y;
       }
       double stop=usecond();
       double time = (stop-start)/Nloop*1000;
-      
-      double effbytes=3*lat*lat*lat*lat*Nvec*sizeof(Real);
-      double bytes=5*lat*lat*lat*lat*Nvec*sizeof(Real);
-      std::cout<<std::setprecision(2) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<"\t\t"<<effbytes/time<<std::endl;
+     
+      double bytes=3*lat*lat*lat*lat*Nvec*sizeof(Real);
+      std::cout<<std::setprecision(2) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
 
     }
 
@@ -105,7 +103,7 @@ int main (int argc, char ** argv)
 
       double start=usecond();
       for(int i=0;i<Nloop;i++){
-	x=a*z;
+	z=a*x;
       }
       double stop=usecond();
       double time = (stop-start)/Nloop*1000;

From 463c31ae090e5bfdda809a7dbfa0b10ee834d648 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@t002.ph.ed.ac.uk>
Date: Sun, 10 May 2015 15:22:31 +0100
Subject: [PATCH 128/429] Bringing expression templates for faster vector loops

---
 lib/Grid_lattice.h | 105 +++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 101 insertions(+), 4 deletions(-)

diff --git a/lib/Grid_lattice.h b/lib/Grid_lattice.h
index 75a80ef8..861a9e24 100644
--- a/lib/Grid_lattice.h
+++ b/lib/Grid_lattice.h
@@ -20,18 +20,106 @@ namespace Grid {
 
 extern int GridCshiftPermuteMap[4][16];
 
+////////////////////////////////////////////////
+// Basic expressions used in Expression Template
+////////////////////////////////////////////////
+
+class LatticeBase {};
+class LatticeExpressionBase {};
+
+template <typename Op, typename T1>                           
+class LatticeUnaryExpression  : public std::pair<Op,std::tuple<T1> > , public LatticeExpressionBase {
+ public:
+ LatticeUnaryExpression(const std::pair<Op,std::tuple<T1> > &arg): std::pair<Op,std::tuple<T1> >(arg) {};
+};
+
+template <typename Op, typename T1, typename T2>              
+class LatticeBinaryExpression : public std::pair<Op,std::tuple<T1,T2> > , public LatticeExpressionBase {
+ public:
+ LatticeBinaryExpression(const std::pair<Op,std::tuple<T1,T2> > &arg): std::pair<Op,std::tuple<T1,T2> >(arg) {};
+};
+
+template <typename Op, typename T1, typename T2, typename T3> 
+class LatticeTrinaryExpression :public std::pair<Op,std::tuple<T1,T2,T3> >, public LatticeExpressionBase {
+ public:
+ LatticeTrinaryExpression(const std::pair<Op,std::tuple<T1,T2,T3> > &arg): std::pair<Op,std::tuple<T1,T2,T3> >(arg) {};
+};
+
 template<class vobj>
-class Lattice
+class Lattice : public LatticeBase
 {
 public:
+
     GridBase *_grid;
     int checkerboard;
     std::vector<vobj,alignedAllocator<vobj> > _odata;
-    //std::valarray<vobj> _odata;
-public:
 
+public:
     typedef typename vobj::scalar_type scalar_type;
     typedef typename vobj::vector_type vector_type;
+    typedef vobj vector_object;
+    
+  ////////////////////////////////////////////////////////////////////////////////
+  // Expression Template closure support
+  ////////////////////////////////////////////////////////////////////////////////
+  template <typename Op, typename T1>                         inline Lattice<vobj> & operator=(const LatticeUnaryExpression<Op,T1> &expr)
+  {
+#pragma omp parallel for
+    for(int ss=0;ss<_grid->oSites();ss++){
+      vobj tmp= eval(ss,expr);
+      vstream(_odata[ss] ,tmp);
+    }
+    return *this;
+  }
+  template <typename Op, typename T1,typename T2>             inline Lattice<vobj> & operator=(const LatticeBinaryExpression<Op,T1,T2> &expr)
+  {
+#pragma omp parallel for
+    for(int ss=0;ss<_grid->oSites();ss++){
+      vobj tmp= eval(ss,expr);
+      vstream(_odata[ss] ,tmp);
+    }
+    return *this;
+  }
+  template <typename Op, typename T1,typename T2,typename T3> inline Lattice<vobj> & operator=(const LatticeTrinaryExpression<Op,T1,T2,T3> &expr)
+  {
+#pragma omp parallel for
+    for(int ss=0;ss<_grid->oSites();ss++){
+      vobj tmp= eval(ss,expr);
+      vstream(_odata[ss] ,tmp);
+    }
+    return *this;
+  }
+  //GridFromExpression is tricky to do
+  template<class Op,class T1>
+    Lattice(const LatticeUnaryExpression<Op,T1> & expr):    _grid(nullptr){
+    GridFromExpression(_grid,expr);
+    assert(_grid!=nullptr);
+    _odata.resize(_grid->oSites());
+#pragma omp parallel for
+    for(int ss=0;ss<_grid->oSites();ss++){
+      _odata[ss] = eval(ss,expr);
+    }
+  };
+  template<class Op,class T1, class T2>
+  Lattice(const LatticeBinaryExpression<Op,T1,T2> & expr):    _grid(nullptr){
+    GridFromExpression(_grid,expr);
+    assert(_grid!=nullptr);
+    _odata.resize(_grid->oSites());
+#pragma omp parallel for
+    for(int ss=0;ss<_grid->oSites();ss++){
+      _odata[ss] = eval(ss,expr);
+    }
+  };
+  template<class Op,class T1, class T2, class T3>
+  Lattice(const LatticeTrinaryExpression<Op,T1,T2,T3> & expr):    _grid(nullptr){
+    GridFromExpression(_grid,expr);
+    assert(_grid!=nullptr);
+    _odata.resize(_grid->oSites());
+#pragma omp parallel for
+    for(int ss=0;ss<_grid->oSites();ss++){
+      _odata[ss] = eval(ss,expr);
+    }
+  };
 
     //////////////////////////////////////////////////////////////////
     // Constructor requires "grid" passed.
@@ -54,7 +142,7 @@ public:
     template<class robj> inline Lattice<vobj> & operator = (const Lattice<robj> & r){
       conformable(*this,r);
       std::cout<<"Lattice operator ="<<std::endl;
-
+#pragma omp parallel for
         for(int ss=0;ss<_grid->oSites();ss++){
             this->_odata[ss]=r._odata[ss];
         }
@@ -88,8 +176,17 @@ public:
  }; // class Lattice
 }
 
+#define GRID_LATTICE_EXPRESSION_TEMPLATES
+
 #include <lattice/Grid_lattice_conformable.h>
+
+#ifdef GRID_LATTICE_EXPRESSION_TEMPLATES
+#include <lattice/Grid_lattice_ET.h>
+#else 
+#include <lattice/Grid_lattice_overload.h>
+#endif
 #include <lattice/Grid_lattice_arith.h>
+
 #include <lattice/Grid_lattice_trace.h>
 #include <lattice/Grid_lattice_transpose.h>
 #include <lattice/Grid_lattice_local.h>

From e02cbaa0162d9081937ea1987901b151e380eefa Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@t002.ph.ed.ac.uk>
Date: Sun, 10 May 2015 15:23:09 +0100
Subject: [PATCH 129/429] Fixing breakage in the Comms non compile

---
 lib/communicator/Grid_communicator_fake.cc | 20 +++++++++++---------
 1 file changed, 11 insertions(+), 9 deletions(-)

diff --git a/lib/communicator/Grid_communicator_fake.cc b/lib/communicator/Grid_communicator_fake.cc
index 962283ee..1cad9c1f 100644
--- a/lib/communicator/Grid_communicator_fake.cc
+++ b/lib/communicator/Grid_communicator_fake.cc
@@ -3,15 +3,17 @@ namespace Grid {
 
 CartesianCommunicator::CartesianCommunicator(std::vector<int> &processors)
 {
-  _ndimension = _processors.size();
-  _processor_coor.resize(_ndimension);
   _processors = processors;
+  _ndimension = processors.size();
+  _processor_coor.resize(_ndimension);
   
   // Require 1^N processor grid for fake
-  for(int d=0;d<_ndimension;d++) if(_processors[d]!=1) exit(-1);
-
-  _processor = 0;// I am the one. The only one..
-  for(int d=0;d<_ndimension;d++) _processor_coor[d] = 0;
+  _Nprocessors=1;
+  _processor = 0;
+  for(int d=0;d<_ndimension;d++) {
+    assert(_processors[d]==1);
+    _processor_coor[d] = 0;
+  }
 }
 
 void CartesianCommunicator::GlobalSum(float &){}
@@ -57,12 +59,12 @@ void CartesianCommunicator::BroadcastWorld(int root,void* data, int bytes)
 
 void CartesianCommunicator::ShiftedRanks(int dim,int shift,int &source,int &dest)
 {
-  source =1;
-  dest=1;
+  source =0;
+  dest=0;
 }
 int CartesianCommunicator::RankFromProcessorCoor(std::vector<int> &coor)
 {
-  return 1;
+  return 0;
 }
 void  CartesianCommunicator::ProcessorCoorFromRank(int rank, std::vector<int> &coor)
 {

From 4a8fd55f52fa9335c4de3b6d01bd09be338ffd9f Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@t002.ph.ed.ac.uk>
Date: Sun, 10 May 2015 15:23:49 +0100
Subject: [PATCH 130/429] Moving operator stuff into separate file so that we
 can switch on/off replacement with expression templates

---
 lib/lattice/Grid_lattice_arith.h | 113 -------------------------------
 1 file changed, 113 deletions(-)

diff --git a/lib/lattice/Grid_lattice_arith.h b/lib/lattice/Grid_lattice_arith.h
index 63ad3335..93801be5 100644
--- a/lib/lattice/Grid_lattice_arith.h
+++ b/lib/lattice/Grid_lattice_arith.h
@@ -3,19 +3,6 @@
 
 namespace Grid {
 
-  //////////////////////////////////////////////////////////////////////////////////////////////////////
-  // unary negation
-  //////////////////////////////////////////////////////////////////////////////////////////////////////
-  template<class vobj>
-  inline Lattice<vobj> operator -(const Lattice<vobj> &r)
-  {
-    Lattice<vobj> ret(r._grid);
-#pragma omp parallel for
-    for(int ss=0;ss<r._grid->oSites();ss++){
-      vstream(ret._odata[ss], -r._odata[ss]);
-    }
-    return ret;
-  }
 
   //////////////////////////////////////////////////////////////////////////////////////////////////////
   //  avoid copy back routines for mult, mac, sub, add
@@ -155,106 +142,6 @@ namespace Grid {
     }
   }
   
-  /////////////////////////////////////////////////////////////////////////////////////
-  // Lattice BinOp Lattice,
-  //NB mult performs conformable check. Do not reapply here for performance.
-  /////////////////////////////////////////////////////////////////////////////////////
-  template<class left,class right>
-    inline auto operator * (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]*rhs._odata[0])>
-  {
-    Lattice<decltype(lhs._odata[0]*rhs._odata[0])> ret(rhs._grid);
-    mult(ret,lhs,rhs);
-    return ret;
-  }
-  template<class left,class right>
-    inline auto operator + (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]+rhs._odata[0])>
-  {
-    Lattice<decltype(lhs._odata[0]+rhs._odata[0])> ret(rhs._grid);
-    add(ret,lhs,rhs);
-    return ret;
-  }
-  template<class left,class right>
-    inline auto operator - (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]-rhs._odata[0])>
-  {
-    Lattice<decltype(lhs._odata[0]-rhs._odata[0])> ret(rhs._grid);
-    sub(ret,lhs,rhs);
-    return ret;
-  }
-  
-  // Scalar BinOp Lattice ;generate return type
-  template<class left,class right>
-    inline auto operator * (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs*rhs._odata[0])>
-  {
-    Lattice<decltype(lhs*rhs._odata[0])> ret(rhs._grid);
-#pragma omp parallel for
-    for(int ss=0;ss<rhs._grid->oSites(); ss++){
-      decltype(lhs*rhs._odata[0]) tmp=lhs*rhs._odata[ss]; 
-      vstream(ret._odata[ss],tmp);
-	   //      ret._odata[ss]=lhs*rhs._odata[ss];
-    }
-    return ret;
-  }
-  template<class left,class right>
-    inline auto operator + (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs+rhs._odata[0])>
-    {
-      Lattice<decltype(lhs+rhs._odata[0])> ret(rhs._grid);
-#pragma omp parallel for
-      for(int ss=0;ss<rhs._grid->oSites(); ss++){
-	decltype(lhs+rhs._odata[0]) tmp =lhs-rhs._odata[ss];  
-	vstream(ret._odata[ss],tmp);
-	//	ret._odata[ss]=lhs+rhs._odata[ss];
-      }
-        return ret;
-    }
-  template<class left,class right>
-    inline auto operator - (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs-rhs._odata[0])>
-  {
-    Lattice<decltype(lhs-rhs._odata[0])> ret(rhs._grid);
-#pragma omp parallel for
-    for(int ss=0;ss<rhs._grid->oSites(); ss++){
-      decltype(lhs-rhs._odata[0]) tmp=lhs-rhs._odata[ss];  
-      vstream(ret._odata[ss],tmp);
-      //      ret._odata[ss]=lhs-rhs._odata[ss];
-    }
-    return ret;
-  }
-    template<class left,class right>
-      inline auto operator * (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]*rhs)>
-    {
-      Lattice<decltype(lhs._odata[0]*rhs)> ret(lhs._grid);
-#pragma omp parallel for
-      for(int ss=0;ss<lhs._grid->oSites(); ss++){
-	decltype(lhs._odata[0]*rhs) tmp =lhs._odata[ss]*rhs;
-	vstream(ret._odata[ss],tmp);
-	//            ret._odata[ss]=lhs._odata[ss]*rhs;
-      }
-      return ret;
-    }
-    template<class left,class right>
-      inline auto operator + (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]+rhs)>
-    {
-        Lattice<decltype(lhs._odata[0]+rhs)> ret(lhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<rhs._grid->oSites(); ss++){
-	  decltype(lhs._odata[0]+rhs) tmp=lhs._odata[ss]+rhs; 
-	  vstream(ret._odata[ss],tmp);
-	  //	  ret._odata[ss]=lhs._odata[ss]+rhs;
-        }
-        return ret;
-    }
-    template<class left,class right>
-      inline auto operator - (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]-rhs)>
-    {
-      Lattice<decltype(lhs._odata[0]-rhs)> ret(lhs._grid);
-#pragma omp parallel for
-      for(int ss=0;ss<rhs._grid->oSites(); ss++){
-	  decltype(lhs._odata[0]-rhs) tmp=lhs._odata[ss]-rhs;
-	  vstream(ret._odata[ss],tmp);
-	  //	ret._odata[ss]=lhs._odata[ss]-rhs;
-      }
-      return ret;
-    }
-
   template<class sobj,class vobj>
   inline void axpy(Lattice<vobj> &ret,sobj a,const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
     conformable(lhs,rhs);

From 961fbb27186df91460fee501535cbf2d32d4a900 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@t002.ph.ed.ac.uk>
Date: Sun, 10 May 2015 15:24:37 +0100
Subject: [PATCH 131/429] Assertion should never hit, but did due to a bug

---
 lib/lattice/Grid_lattice_reduction.h | 1 +
 1 file changed, 1 insertion(+)

diff --git a/lib/lattice/Grid_lattice_reduction.h b/lib/lattice/Grid_lattice_reduction.h
index 6679f29b..679effb0 100644
--- a/lib/lattice/Grid_lattice_reduction.h
+++ b/lib/lattice/Grid_lattice_reduction.h
@@ -83,6 +83,7 @@ template<class vobj> inline void sliceSum(const Lattice<vobj> &Data,std::vector<
   typedef typename vobj::scalar_object sobj;
 
   GridBase  *grid = Data._grid;
+  assert(grid!=NULL);
   const int    Nd = grid->_ndimension;
   const int Nsimd = grid->Nsimd();
 

From dc7132af717f894f7d0d52d3926aec820b9fb40d Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@t002.ph.ed.ac.uk>
Date: Sun, 10 May 2015 15:25:23 +0100
Subject: [PATCH 132/429] Small tweak to enable benchmarking to suppress gauge
 field bandwidth as a test. This is a short term hack while I benchmark.

---
 lib/qcd/Grid_qcd_wilson_dop.cc | 17 +++++++++--------
 1 file changed, 9 insertions(+), 8 deletions(-)

diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index ddc73c4a..a9ac6346 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -106,6 +106,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
   for(int sss=0;sss<grid->oSites();sss++){
 
     int ss = sss;
+    int ssu= sss;
 	//int ss = Stencil._LebesgueReorder[sss];
 
     // Xp
@@ -123,7 +124,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ss](Xp),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Xp),&chi());
     spReconXp(result,Uchi);
 
     // Yp
@@ -141,7 +142,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ss](Yp),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Yp),&chi());
     accumReconYp(result,Uchi);
 
     // Zp
@@ -159,7 +160,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ss](Zp),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Zp),&chi());
     accumReconZp(result,Uchi);
 
     // Tp
@@ -177,7 +178,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ss](Tp),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Tp),&chi());
     accumReconTp(result,Uchi);
 
     // Xm
@@ -195,7 +196,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ss](Xm),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Xm),&chi());
     accumReconXm(result,Uchi);
 
 
@@ -214,7 +215,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ss](Ym),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Ym),&chi());
     accumReconYm(result,Uchi);
 
     // Zm
@@ -232,7 +233,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ss](Zm),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Zm),&chi());
     accumReconZm(result,Uchi);
 
     // Tm
@@ -250,7 +251,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ss](Tm),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Tm),&chi());
     accumReconTm(result,Uchi);
 
     vstream(out._odata[ss],result);

From cd90f55536a958dcf5550d424868e330f5fe12fe Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@t002.ph.ed.ac.uk>
Date: Sun, 10 May 2015 15:26:06 +0100
Subject: [PATCH 133/429] Single node default. Should expose this as command
 line args, but haven't sorted out Grid_initialize to handle this. Should put
 this on the TODO list.

---
 tests/Grid_cshift.cc | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tests/Grid_cshift.cc b/tests/Grid_cshift.cc
index c0863965..fcee7b70 100644
--- a/tests/Grid_cshift.cc
+++ b/tests/Grid_cshift.cc
@@ -9,7 +9,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> simd_layout({1,1,2,2});
-  std::vector<int> mpi_layout ({2,2,1,4});
+  std::vector<int> mpi_layout ({1,1,1,1});
   std::vector<int> latt_size  ({8,8,8,16});
     
   GridCartesian     Fine(latt_size,simd_layout,mpi_layout);

From 7119bce9f37147b55ccfb671c775d87d0d81a973 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@t002.ph.ed.ac.uk>
Date: Sun, 10 May 2015 15:27:38 +0100
Subject: [PATCH 134/429] Default to single node. Move to command line args.

---
 tests/Grid_nersc_io.cc | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/tests/Grid_nersc_io.cc b/tests/Grid_nersc_io.cc
index 06e51d41..95f9db88 100644
--- a/tests/Grid_nersc_io.cc
+++ b/tests/Grid_nersc_io.cc
@@ -11,7 +11,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> simd_layout({1,1,2,2});
-  std::vector<int> mpi_layout ({2,2,2,2});
+  std::vector<int> mpi_layout ({1,1,1,1});
   std::vector<int> latt_size  ({16,16,16,32});
   std::vector<int> clatt_size  ({4,4,4,8});
   int orthodir=3;
@@ -52,6 +52,7 @@ int main (int argc, char ** argv)
   
   double vol = Fine.gSites();
   Complex PlaqScale(1.0/vol/6.0/3.0);
+  std::cout <<"PlaqScale" << PlaqScale<<std::endl;
 
   std::vector<TComplex> Plaq_T(orthosz);
   sliceSum(Plaq,Plaq_T,Nd-1);

From 79c51ac51f5f420db55e35ed53efb221fc9b6e5f Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@t002.ph.ed.ac.uk>
Date: Sun, 10 May 2015 15:30:29 +0100
Subject: [PATCH 135/429] Hack; must bring norm2 into the unary operator list.
 ET's are still incomplete.

---
 tests/Grid_stencil.cc | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/tests/Grid_stencil.cc b/tests/Grid_stencil.cc
index eb661487..64a4ee9d 100644
--- a/tests/Grid_stencil.cc
+++ b/tests/Grid_stencil.cc
@@ -70,7 +70,8 @@ int main (int argc, char ** argv)
 
 	Real nrmC = norm2(Check);
 	Real nrmB = norm2(Bar);
-	Real nrm  = norm2(Check-Bar);
+	Diff = Check-Bar;
+	Real nrm  = norm2(Diff);
 	std::cout<<"N2diff ="<<nrm<<" "<<nrmC<<" " <<nrmB<<std::endl;
 
 	Real snrmC =0;

From 5a7751d9df47366060a6f61c303ddf2f40824c6b Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@t002.ph.ed.ac.uk>
Date: Sun, 10 May 2015 15:32:56 +0100
Subject: [PATCH 136/429] Updated TODO list

---
 TODO | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/TODO b/TODO
index 9dc8fada..0ef20444 100644
--- a/TODO
+++ b/TODO
@@ -2,6 +2,17 @@
   - use protocol buffers? replace xmlReader/Writer ec..
   - Binary use htonll, htonl
 
+* Expression template engine:
+
+   - Audit
+   - Introduce base clase for Grid Tensors.
+   - Introduce norm2 unary op.
+   - Introduce conversion automatic from expression to Lattice<vobj>
+   -
+
+* Command line args for geometry, simd, etc. layout. Is it necessary to have
+  user pass these? Is this a QCD specific?
+
 * CovariantShift support                             -----Use a class to store gauge field? (parallel transport?)
 
 * Strong test for norm2, conj and all primitive types. -- Grid_simd test is almost there

From 1ab92563b912c1a0103bcec0f4b44f5bea428485 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@t002.ph.ed.ac.uk>
Date: Sun, 10 May 2015 15:34:20 +0100
Subject: [PATCH 137/429] Expression template engin

---
 lib/lattice/Grid_lattice_ET.h | 227 ++++++++++++++++++++++++++++++++++
 1 file changed, 227 insertions(+)
 create mode 100644 lib/lattice/Grid_lattice_ET.h

diff --git a/lib/lattice/Grid_lattice_ET.h b/lib/lattice/Grid_lattice_ET.h
new file mode 100644
index 00000000..d4fd7d05
--- /dev/null
+++ b/lib/lattice/Grid_lattice_ET.h
@@ -0,0 +1,227 @@
+
+#ifndef GRID_LATTICE_ET_H
+#define GRID_LATTICE_ET_H
+
+#include <iostream>
+#include <vector>
+#include <tuple>
+#include <typeinfo>
+
+namespace Grid {
+
+////////////////////////////////////////////
+// recursive evaluation of expressions; Could
+// switch to generic approach with variadics, a la
+// Antonin's Lat Sim but the repack to variadic with popped
+// from tuple is hideous; C++14 introduces std::make_index_sequence for this
+////////////////////////////////////////////
+
+
+//leaf eval of lattice ; should enable if protect using traits
+
+template <typename T> using is_lattice      = std::is_base_of<LatticeBase,T >;
+
+template <typename T> using is_lattice_expr = std::is_base_of<LatticeExpressionBase,T >;
+
+template<class sobj>
+inline sobj eval(const unsigned int ss, const sobj &arg)
+{
+  return arg;
+}
+template<class lobj>
+inline const lobj &eval(const unsigned int ss, const Lattice<lobj> &arg)
+{
+    return arg._odata[ss];
+}
+
+// handle nodes in syntax tree
+template <typename Op, typename T1>
+auto inline eval(const unsigned int ss, const LatticeUnaryExpression<Op,T1 > &expr) // eval one operand
+  -> decltype(expr.first.func(eval(ss,std::get<0>(expr.second))))
+{
+  return expr.first.func(eval(ss,std::get<0>(expr.second)));
+}
+
+template <typename Op, typename T1, typename T2>
+auto inline eval(const unsigned int ss, const LatticeBinaryExpression<Op,T1,T2> &expr) // eval two operands
+  -> decltype(expr.first.func(eval(ss,std::get<0>(expr.second)),eval(ss,std::get<1>(expr.second))))
+{
+  return expr.first.func(eval(ss,std::get<0>(expr.second)),eval(ss,std::get<1>(expr.second)));
+}
+
+template <typename Op, typename T1, typename T2, typename T3>
+auto inline eval(const unsigned int ss, const LatticeTrinaryExpression<Op,T1,T2,T3 > &expr) // eval three operands
+  -> decltype(expr.first.func(eval(ss,std::get<0>(expr.second)),eval(ss,std::get<1>(expr.second)),eval(ss,std::get<2>(expr.second))))
+{
+  return expr.first.func(eval(ss,std::get<0>(expr.second)),eval(ss,std::get<1>(expr.second)),eval(ss,std::get<2>(expr.second)) );
+}
+
+//////////////////////////////////////////////////////////////////////////
+// Obtain the grid from an expression, ensuring conformable. This must follow a tree recursion
+//////////////////////////////////////////////////////////////////////////
+template<class T1, typename std::enable_if<is_lattice<T1>::value, T1>::type * =nullptr >
+inline void GridFromExpression(GridBase * &grid,const T1& lat)   // Lattice leaf
+{
+  if ( grid ) {
+    conformable(grid,lat._grid);
+  } 
+  grid=lat._grid;
+}
+template <typename Op, typename T1>
+inline void GridFromExpression(GridBase * &grid,const LatticeUnaryExpression<Op,T1 > &expr)
+{
+  GridFromExpression(grid,std::get<0>(expr.second));// recurse 
+}
+
+template <typename Op, typename T1, typename T2>
+inline void GridFromExpression(GridBase * &grid,const LatticeBinaryExpression<Op,T1,T2> &expr) 
+{
+  GridFromExpression(grid,std::get<0>(expr.second));// recurse
+  GridFromExpression(grid,std::get<1>(expr.second));
+}
+template <typename Op, typename T1, typename T2, typename T3>
+inline void GridFromExpression( GridBase * &grid,const LatticeTrinaryExpression<Op,T1,T2,T3 > &expr) 
+{
+  GridFromExpression(grid,std::get<0>(expr.second));// recurse
+  GridFromExpression(grid,std::get<1>(expr.second));
+  GridFromExpression(grid,std::get<2>(expr.second));
+}
+template<class T1,typename std::enable_if<!is_lattice<T1>::value, T1>::type * = nullptr >
+inline void GridFromExpression(GridBase * &grid,const T1& notlat)   // non-lattice leaf
+{
+}
+
+////////////////////////////////////////////
+// Unary operators and funcs
+////////////////////////////////////////////
+#define GridUnopClass(name,ret)\
+template <class arg> struct name\
+{\
+  static auto inline func(const arg a)-> decltype(ret) { return ret; } \
+};
+
+GridUnopClass(UnarySub,-a);
+GridUnopClass(UnaryAdj,adj(a));
+GridUnopClass(UnaryConj,conj(a));
+GridUnopClass(UnaryTrace,trace(a));
+GridUnopClass(UnaryTranspose,transpose(a));
+
+////////////////////////////////////////////
+// Binary operators
+////////////////////////////////////////////
+#define GridBinOpClass(name,combination)\
+template <class left,class right>\
+struct name\
+{\
+  static auto inline func(const left &lhs,const right &rhs)-> decltype(combination) const \
+    {\
+      return combination;\
+    }\
+}
+GridBinOpClass(BinaryAdd,lhs+rhs);
+GridBinOpClass(BinarySub,lhs-rhs);
+GridBinOpClass(BinaryMul,lhs*rhs);
+
+////////////////////////////////////////////
+// Operator syntactical glue
+////////////////////////////////////////////
+ 
+#define GRID_UNOP(name)   name<decltype(eval(0, arg))>
+#define GRID_BINOP(name)  name<decltype(eval(0, lhs)), decltype(eval(0, rhs))>
+#define GRID_TRINOP(name) name<decltype(eval(0, pred)), decltype(eval(0, lhs)), decltype(eval(0, rhs))>
+
+#define GRID_DEF_UNOP(op, name)\
+template <typename T1,\
+  typename std::enable_if<is_lattice<T1>::value||is_lattice_expr<T1>::value, T1>::type* = nullptr> inline auto op(const T1 &arg) \
+  -> decltype(LatticeUnaryExpression<GRID_UNOP(name),const T1&>(std::make_pair(GRID_UNOP(name)(),std::forward_as_tuple(arg)))) \
+{ return LatticeUnaryExpression<GRID_UNOP(name), const T1 &>(std::make_pair(GRID_UNOP(name)(),std::forward_as_tuple(arg))); }
+
+#define GRID_BINOP_LEFT(op, name)\
+template <typename T1,typename T2,\
+          typename std::enable_if<is_lattice<T1>::value||is_lattice_expr<T1>::value, T1>::type* = nullptr>\
+inline auto op(const T1 &lhs,const T2&rhs) \
+  -> decltype(LatticeBinaryExpression<GRID_BINOP(name),const T1&,const T2 &>(std::make_pair(GRID_BINOP(name)(),\
+											    std::forward_as_tuple(lhs, rhs)))) \
+{\
+ return LatticeBinaryExpression<GRID_BINOP(name), const T1 &, const T2 &>(std::make_pair(GRID_BINOP(name)(),\
+									  std::forward_as_tuple(lhs, rhs))); \
+}
+
+#define GRID_BINOP_RIGHT(op, name)\
+ template <typename T1,typename T2,\
+           typename std::enable_if<!is_lattice<T1>::value && !is_lattice_expr<T1>::value, T1>::type* = nullptr,\
+           typename std::enable_if< is_lattice<T2>::value ||  is_lattice_expr<T2>::value, T2>::type* = nullptr> \
+inline auto op(const T1 &lhs,const T2&rhs)			\
+  -> decltype(LatticeBinaryExpression<GRID_BINOP(name),const T1&,const T2 &>(std::make_pair(GRID_BINOP(name)(),\
+											    std::forward_as_tuple(lhs, rhs)))) \
+{\
+ return LatticeBinaryExpression<GRID_BINOP(name), const T1 &, const T2 &>(std::make_pair(GRID_BINOP(name)(),\
+								          std::forward_as_tuple(lhs, rhs))); \
+}
+
+#define GRID_DEF_BINOP(op, name)\
+ GRID_BINOP_LEFT(op,name);\
+ GRID_BINOP_RIGHT(op,name);
+
+
+#define GRID_DEF_TRINOP(op, name)\
+template <typename T1,typename T2,typename T3> inline auto op(const T1 &pred,const T2&lhs,const T3 &rhs) \
+  -> decltype(LatticeTrinaryExpression<GRID_TRINOP(name),const T1&,const T2 &,const T3&>(std::make_pair(GRID_TRINOP(name)(),\
+										   std::forward_as_tuple(pred,lhs,rhs)))) \
+{\
+  return LatticeTrinaryExpression<GRID_TRINOP(name), const T1 &, const T2 &,const T3&>(std::make_pair(GRID_TRINOP(name)(), \
+										 std::forward_as_tuple(pred,lhs, rhs))); \
+}
+////////////////////////
+//Operator definitions
+////////////////////////
+
+GRID_DEF_UNOP(operator -,UnarySub);
+GRID_DEF_UNOP(adj,UnaryAdj);
+GRID_DEF_UNOP(conj,UnaryConj);
+GRID_DEF_UNOP(trace,UnaryTrace);
+GRID_DEF_UNOP(transpose,UnaryTranspose);
+
+GRID_DEF_BINOP(operator+,BinaryAdd);
+GRID_DEF_BINOP(operator-,BinarySub);
+GRID_DEF_BINOP(operator*,BinaryMul);
+
+#undef GRID_UNOP
+#undef GRID_BINOP
+#undef GRID_TRINOP
+
+#undef GRID_DEF_UNOP
+#undef GRID_DEF_BINOP
+#undef GRID_DEF_TRINOP
+
+}
+
+#if 0
+using namespace Grid;
+ 	      
+ int main(int argc,char **argv){
+   
+   Lattice<double> v1(16);
+   Lattice<double> v2(16);
+   Lattice<double> v3(16);
+
+   BinaryAdd<double,double> tmp;
+   LatticeBinaryExpression<BinaryAdd<double,double>,Lattice<double> &,Lattice<double> &> 
+     expr(std::make_pair(tmp,
+	  std::forward_as_tuple(v1,v2)));
+   tmp.func(eval(0,v1),eval(0,v2));
+
+   auto var = v1+v2;
+   std::cout<<typeid(var).name()<<std::endl;
+
+   v3=v1+v2;
+   v3=v1+v2+v1*v2;
+ };
+
+void testit(Lattice<double> &v1,Lattice<double> &v2,Lattice<double> &v3)
+{
+   v3=v1+v2+v1*v2;
+}
+#endif
+
+#endif

From 1ec1b4ee440922ecaa5299a2b46bda01a8008d41 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@t002.ph.ed.ac.uk>
Date: Sun, 10 May 2015 15:35:30 +0100
Subject: [PATCH 138/429] Expression template hack

---
 lib/lattice/Grid_lattice_overload.h | 120 ++++++++++++++++++++++++++++
 1 file changed, 120 insertions(+)
 create mode 100644 lib/lattice/Grid_lattice_overload.h

diff --git a/lib/lattice/Grid_lattice_overload.h b/lib/lattice/Grid_lattice_overload.h
new file mode 100644
index 00000000..512e45d1
--- /dev/null
+++ b/lib/lattice/Grid_lattice_overload.h
@@ -0,0 +1,120 @@
+#ifndef GRID_LATTICE_OVERLOAD_H
+#define GRID_LATTICE_OVERLOAD_H
+
+namespace Grid {
+
+  //////////////////////////////////////////////////////////////////////////////////////////////////////
+  // unary negation
+  //////////////////////////////////////////////////////////////////////////////////////////////////////
+  template<class vobj>
+  inline Lattice<vobj> operator -(const Lattice<vobj> &r)
+  {
+    Lattice<vobj> ret(r._grid);
+#pragma omp parallel for
+    for(int ss=0;ss<r._grid->oSites();ss++){
+      vstream(ret._odata[ss], -r._odata[ss]);
+    }
+    return ret;
+  } 
+  /////////////////////////////////////////////////////////////////////////////////////
+  // Lattice BinOp Lattice,
+  //NB mult performs conformable check. Do not reapply here for performance.
+  /////////////////////////////////////////////////////////////////////////////////////
+  template<class left,class right>
+    inline auto operator * (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]*rhs._odata[0])>
+  {
+    Lattice<decltype(lhs._odata[0]*rhs._odata[0])> ret(rhs._grid);
+    mult(ret,lhs,rhs);
+    return ret;
+  }
+  template<class left,class right>
+    inline auto operator + (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]+rhs._odata[0])>
+  {
+    Lattice<decltype(lhs._odata[0]+rhs._odata[0])> ret(rhs._grid);
+    add(ret,lhs,rhs);
+    return ret;
+  }
+  template<class left,class right>
+    inline auto operator - (const Lattice<left> &lhs,const Lattice<right> &rhs)-> Lattice<decltype(lhs._odata[0]-rhs._odata[0])>
+  {
+    Lattice<decltype(lhs._odata[0]-rhs._odata[0])> ret(rhs._grid);
+    sub(ret,lhs,rhs);
+    return ret;
+  }
+  
+  // Scalar BinOp Lattice ;generate return type
+  template<class left,class right>
+  inline auto operator * (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs*rhs._odata[0])>
+  {
+    Lattice<decltype(lhs*rhs._odata[0])> ret(rhs._grid);
+#pragma omp parallel for
+    for(int ss=0;ss<rhs._grid->oSites(); ss++){
+      decltype(lhs*rhs._odata[0]) tmp=lhs*rhs._odata[ss]; 
+      vstream(ret._odata[ss],tmp);
+	   //      ret._odata[ss]=lhs*rhs._odata[ss];
+    }
+    return ret;
+  }
+  template<class left,class right>
+    inline auto operator + (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs+rhs._odata[0])>
+    {
+      Lattice<decltype(lhs+rhs._odata[0])> ret(rhs._grid);
+#pragma omp parallel for
+      for(int ss=0;ss<rhs._grid->oSites(); ss++){
+	decltype(lhs+rhs._odata[0]) tmp =lhs-rhs._odata[ss];  
+	vstream(ret._odata[ss],tmp);
+	//	ret._odata[ss]=lhs+rhs._odata[ss];
+      }
+        return ret;
+    }
+  template<class left,class right>
+    inline auto operator - (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs-rhs._odata[0])>
+  {
+    Lattice<decltype(lhs-rhs._odata[0])> ret(rhs._grid);
+#pragma omp parallel for
+    for(int ss=0;ss<rhs._grid->oSites(); ss++){
+      decltype(lhs-rhs._odata[0]) tmp=lhs-rhs._odata[ss];  
+      vstream(ret._odata[ss],tmp);
+      //      ret._odata[ss]=lhs-rhs._odata[ss];
+    }
+    return ret;
+  }
+    template<class left,class right>
+      inline auto operator * (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]*rhs)>
+    {
+      Lattice<decltype(lhs._odata[0]*rhs)> ret(lhs._grid);
+#pragma omp parallel for
+      for(int ss=0;ss<lhs._grid->oSites(); ss++){
+	decltype(lhs._odata[0]*rhs) tmp =lhs._odata[ss]*rhs;
+	vstream(ret._odata[ss],tmp);
+	//            ret._odata[ss]=lhs._odata[ss]*rhs;
+      }
+      return ret;
+    }
+    template<class left,class right>
+      inline auto operator + (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]+rhs)>
+    {
+        Lattice<decltype(lhs._odata[0]+rhs)> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<rhs._grid->oSites(); ss++){
+	  decltype(lhs._odata[0]+rhs) tmp=lhs._odata[ss]+rhs; 
+	  vstream(ret._odata[ss],tmp);
+	  //	  ret._odata[ss]=lhs._odata[ss]+rhs;
+        }
+        return ret;
+    }
+    template<class left,class right>
+      inline auto operator - (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]-rhs)>
+    {
+      Lattice<decltype(lhs._odata[0]-rhs)> ret(lhs._grid);
+#pragma omp parallel for
+      for(int ss=0;ss<rhs._grid->oSites(); ss++){
+	  decltype(lhs._odata[0]-rhs) tmp=lhs._odata[ss]-rhs;
+	  vstream(ret._odata[ss],tmp);
+	  //	ret._odata[ss]=lhs._odata[ss]-rhs;
+      }
+      return ret;
+    }
+
+
+}

From 2203c6e5976e31eb83e9337f6bdbbf80152f1a6e Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@frontend03.ecdf.ed.ac.uk>
Date: Sun, 10 May 2015 23:29:21 +0100
Subject: [PATCH 139/429] Lots of changes required to compile for MIC under
 ICPC

---
 Makefile.in                           |  393 +-
 TODO                                  |    2 +
 aclocal.m4                            |  707 ++--
 configure                             | 5011 +++++++++++++++----------
 lib/Grid_config.h                     |   23 +-
 lib/Grid_config.h.in                  |    3 -
 lib/Grid_lattice.h                    |    2 +-
 lib/Grid_simd.h                       |    9 +-
 lib/cshift/Grid_cshift_common.h       |   26 +-
 lib/lattice/Grid_lattice_coordinate.h |   15 +-
 lib/lattice/Grid_lattice_overload.h   |    1 +
 lib/math/Grid_math_tensors.h          |    8 +-
 lib/qcd/Grid_qcd_wilson_dop.cc        |   37 +-
 lib/simd/Grid_vComplexD.h             |   17 +-
 lib/simd/Grid_vComplexF.h             |    5 +-
 lib/simd/Grid_vInteger.h              |    7 +-
 lib/simd/Grid_vRealD.h                |    2 +-
 lib/simd/Grid_vRealF.h                |    3 +-
 tests/Grid_gamma.cc                   |    3 +-
 tests/Grid_main.cc                    |   22 +-
 20 files changed, 3438 insertions(+), 2858 deletions(-)

diff --git a/Makefile.in b/Makefile.in
index c9257ab2..f9e18fe4 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -1,8 +1,9 @@
-# Makefile.in generated by automake 1.15 from Makefile.am.
+# Makefile.in generated by automake 1.11.1 from Makefile.am.
 # @configure_input@
 
-# Copyright (C) 1994-2014 Free Software Foundation, Inc.
-
+# Copyright (C) 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002,
+# 2003, 2004, 2005, 2006, 2007, 2008, 2009  Free Software Foundation,
+# Inc.
 # This Makefile.in is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
@@ -14,61 +15,6 @@
 
 @SET_MAKE@
 VPATH = @srcdir@
-am__is_gnu_make = { \
-  if test -z '$(MAKELEVEL)'; then \
-    false; \
-  elif test -n '$(MAKE_HOST)'; then \
-    true; \
-  elif test -n '$(MAKE_VERSION)' && test -n '$(CURDIR)'; then \
-    true; \
-  else \
-    false; \
-  fi; \
-}
-am__make_running_with_option = \
-  case $${target_option-} in \
-      ?) ;; \
-      *) echo "am__make_running_with_option: internal error: invalid" \
-              "target option '$${target_option-}' specified" >&2; \
-         exit 1;; \
-  esac; \
-  has_opt=no; \
-  sane_makeflags=$$MAKEFLAGS; \
-  if $(am__is_gnu_make); then \
-    sane_makeflags=$$MFLAGS; \
-  else \
-    case $$MAKEFLAGS in \
-      *\\[\ \	]*) \
-        bs=\\; \
-        sane_makeflags=`printf '%s\n' "$$MAKEFLAGS" \
-          | sed "s/$$bs$$bs[$$bs $$bs	]*//g"`;; \
-    esac; \
-  fi; \
-  skip_next=no; \
-  strip_trailopt () \
-  { \
-    flg=`printf '%s\n' "$$flg" | sed "s/$$1.*$$//"`; \
-  }; \
-  for flg in $$sane_makeflags; do \
-    test $$skip_next = yes && { skip_next=no; continue; }; \
-    case $$flg in \
-      *=*|--*) continue;; \
-        -*I) strip_trailopt 'I'; skip_next=yes;; \
-      -*I?*) strip_trailopt 'I';; \
-        -*O) strip_trailopt 'O'; skip_next=yes;; \
-      -*O?*) strip_trailopt 'O';; \
-        -*l) strip_trailopt 'l'; skip_next=yes;; \
-      -*l?*) strip_trailopt 'l';; \
-      -[dEDm]) skip_next=yes;; \
-      -[JT]) skip_next=yes;; \
-    esac; \
-    case $$flg in \
-      *$$target_option*) has_opt=yes; break;; \
-    esac; \
-  done; \
-  test $$has_opt = yes
-am__make_dryrun = (target_option=n; $(am__make_running_with_option))
-am__make_keepgoing = (target_option=k; $(am__make_running_with_option))
 pkgdatadir = $(datadir)/@PACKAGE@
 pkgincludedir = $(includedir)/@PACKAGE@
 pkglibdir = $(libdir)/@PACKAGE@
@@ -86,86 +32,43 @@ NORMAL_UNINSTALL = :
 PRE_UNINSTALL = :
 POST_UNINSTALL = :
 subdir = .
+DIST_COMMON = README $(am__configure_deps) $(srcdir)/Makefile.am \
+	$(srcdir)/Makefile.in $(top_srcdir)/configure AUTHORS COPYING \
+	ChangeLog INSTALL NEWS TODO compile depcomp install-sh missing
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/configure.ac
 am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
 	$(ACLOCAL_M4)
-DIST_COMMON = $(srcdir)/Makefile.am $(top_srcdir)/configure \
-	$(am__configure_deps) $(am__DIST_COMMON)
 am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \
  configure.lineno config.status.lineno
 mkinstalldirs = $(install_sh) -d
 CONFIG_HEADER = $(top_builddir)/lib/Grid_config.h
 CONFIG_CLEAN_FILES =
 CONFIG_CLEAN_VPATH_FILES =
-AM_V_P = $(am__v_P_@AM_V@)
-am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
-am__v_P_0 = false
-am__v_P_1 = :
-AM_V_GEN = $(am__v_GEN_@AM_V@)
-am__v_GEN_ = $(am__v_GEN_@AM_DEFAULT_V@)
-am__v_GEN_0 = @echo "  GEN     " $@;
-am__v_GEN_1 = 
-AM_V_at = $(am__v_at_@AM_V@)
-am__v_at_ = $(am__v_at_@AM_DEFAULT_V@)
-am__v_at_0 = @
-am__v_at_1 = 
 SOURCES =
 DIST_SOURCES =
-RECURSIVE_TARGETS = all-recursive check-recursive cscopelist-recursive \
-	ctags-recursive dvi-recursive html-recursive info-recursive \
-	install-data-recursive install-dvi-recursive \
-	install-exec-recursive install-html-recursive \
-	install-info-recursive install-pdf-recursive \
-	install-ps-recursive install-recursive installcheck-recursive \
-	installdirs-recursive pdf-recursive ps-recursive \
-	tags-recursive uninstall-recursive
-am__can_run_installinfo = \
-  case $$AM_UPDATE_INFO_DIR in \
-    n|no|NO) false;; \
-    *) (install-info --version) >/dev/null 2>&1;; \
-  esac
+RECURSIVE_TARGETS = all-recursive check-recursive dvi-recursive \
+	html-recursive info-recursive install-data-recursive \
+	install-dvi-recursive install-exec-recursive \
+	install-html-recursive install-info-recursive \
+	install-pdf-recursive install-ps-recursive install-recursive \
+	installcheck-recursive installdirs-recursive pdf-recursive \
+	ps-recursive uninstall-recursive
 RECURSIVE_CLEAN_TARGETS = mostlyclean-recursive clean-recursive	\
   distclean-recursive maintainer-clean-recursive
-am__recursive_targets = \
-  $(RECURSIVE_TARGETS) \
-  $(RECURSIVE_CLEAN_TARGETS) \
-  $(am__extra_recursive_targets)
-AM_RECURSIVE_TARGETS = $(am__recursive_targets:-recursive=) TAGS CTAGS \
-	cscope distdir dist dist-all distcheck
-am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) $(LISP)
-# Read a list of newline-separated strings from the standard input,
-# and print each of them once, without duplicates.  Input order is
-# *not* preserved.
-am__uniquify_input = $(AWK) '\
-  BEGIN { nonempty = 0; } \
-  { items[$$0] = 1; nonempty = 1; } \
-  END { if (nonempty) { for (i in items) print i; }; } \
-'
-# Make sure the list of sources is unique.  This is necessary because,
-# e.g., the same source file might be shared among _SOURCES variables
-# for different programs/libraries.
-am__define_uniq_tagged_files = \
-  list='$(am__tagged_files)'; \
-  unique=`for i in $$list; do \
-    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
-  done | $(am__uniquify_input)`
+AM_RECURSIVE_TARGETS = $(RECURSIVE_TARGETS:-recursive=) \
+	$(RECURSIVE_CLEAN_TARGETS:-recursive=) tags TAGS ctags CTAGS \
+	distdir dist dist-all distcheck
 ETAGS = etags
 CTAGS = ctags
-CSCOPE = cscope
 DIST_SUBDIRS = $(SUBDIRS)
-am__DIST_COMMON = $(srcdir)/Makefile.in AUTHORS COPYING ChangeLog \
-	INSTALL NEWS README TODO compile depcomp install-sh missing
 DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
 distdir = $(PACKAGE)-$(VERSION)
 top_distdir = $(distdir)
 am__remove_distdir = \
-  if test -d "$(distdir)"; then \
-    find "$(distdir)" -type d ! -perm -200 -exec chmod u+w {} ';' \
-      && rm -rf "$(distdir)" \
-      || { sleep 5 && rm -rf "$(distdir)"; }; \
-  else :; fi
-am__post_remove_distdir = $(am__remove_distdir)
+  { test ! -d "$(distdir)" \
+    || { find "$(distdir)" -type d ! -perm -200 -exec chmod u+w {} ';' \
+         && rm -fr "$(distdir)"; }; }
 am__relativize = \
   dir0=`pwd`; \
   sed_first='s,^\([^/]*\)/.*$$,\1,'; \
@@ -193,14 +96,10 @@ am__relativize = \
   reldir="$$dir2"
 DIST_ARCHIVES = $(distdir).tar.gz
 GZIP_ENV = --best
-DIST_TARGETS = dist-gzip
 distuninstallcheck_listfiles = find . -type f -print
-am__distuninstallcheck_listfiles = $(distuninstallcheck_listfiles) \
-  | sed 's|^\./|$(prefix)/|' | grep -v '$(infodir)/dir$$'
 distcleancheck_listfiles = find . -type f -print
 ACLOCAL = @ACLOCAL@
 AMTAR = @AMTAR@
-AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
 AUTOCONF = @AUTOCONF@
 AUTOHEADER = @AUTOHEADER@
 AUTOMAKE = @AUTOMAKE@
@@ -240,7 +139,6 @@ PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
 PACKAGE_NAME = @PACKAGE_NAME@
 PACKAGE_STRING = @PACKAGE_STRING@
 PACKAGE_TARNAME = @PACKAGE_TARNAME@
-PACKAGE_URL = @PACKAGE_URL@
 PACKAGE_VERSION = @PACKAGE_VERSION@
 PATH_SEPARATOR = @PATH_SEPARATOR@
 RANLIB = @RANLIB@
@@ -298,7 +196,7 @@ SUBDIRS = lib tests benchmarks
 all: all-recursive
 
 .SUFFIXES:
-am--refresh: Makefile
+am--refresh:
 	@:
 $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	@for dep in $?; do \
@@ -313,6 +211,7 @@ $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu Makefile'; \
 	$(am__cd) $(top_srcdir) && \
 	  $(AUTOMAKE) --gnu Makefile
+.PRECIOUS: Makefile
 Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
 	@case '$?' in \
 	  *config.status*) \
@@ -333,25 +232,22 @@ $(ACLOCAL_M4):  $(am__aclocal_m4_deps)
 $(am__aclocal_m4_deps):
 
 # This directory's subdirectories are mostly independent; you can cd
-# into them and run 'make' without going through this Makefile.
-# To change the values of 'make' variables: instead of editing Makefiles,
-# (1) if the variable is set in 'config.status', edit 'config.status'
-#     (which will cause the Makefiles to be regenerated when you run 'make');
-# (2) otherwise, pass the desired values on the 'make' command line.
-$(am__recursive_targets):
-	@fail=; \
-	if $(am__make_keepgoing); then \
-	  failcom='fail=yes'; \
-	else \
-	  failcom='exit 1'; \
-	fi; \
+# into them and run `make' without going through this Makefile.
+# To change the values of `make' variables: instead of editing Makefiles,
+# (1) if the variable is set in `config.status', edit `config.status'
+#     (which will cause the Makefiles to be regenerated when you run `make');
+# (2) otherwise, pass the desired values on the `make' command line.
+$(RECURSIVE_TARGETS):
+	@fail= failcom='exit 1'; \
+	for f in x $$MAKEFLAGS; do \
+	  case $$f in \
+	    *=* | --[!k]*);; \
+	    *k*) failcom='fail=yes';; \
+	  esac; \
+	done; \
 	dot_seen=no; \
 	target=`echo $@ | sed s/-recursive//`; \
-	case "$@" in \
-	  distclean-* | maintainer-clean-*) list='$(DIST_SUBDIRS)' ;; \
-	  *) list='$(SUBDIRS)' ;; \
-	esac; \
-	for subdir in $$list; do \
+	list='$(SUBDIRS)'; for subdir in $$list; do \
 	  echo "Making $$target in $$subdir"; \
 	  if test "$$subdir" = "."; then \
 	    dot_seen=yes; \
@@ -366,12 +262,57 @@ $(am__recursive_targets):
 	  $(MAKE) $(AM_MAKEFLAGS) "$$target-am" || exit 1; \
 	fi; test -z "$$fail"
 
-ID: $(am__tagged_files)
-	$(am__define_uniq_tagged_files); mkid -fID $$unique
-tags: tags-recursive
-TAGS: tags
+$(RECURSIVE_CLEAN_TARGETS):
+	@fail= failcom='exit 1'; \
+	for f in x $$MAKEFLAGS; do \
+	  case $$f in \
+	    *=* | --[!k]*);; \
+	    *k*) failcom='fail=yes';; \
+	  esac; \
+	done; \
+	dot_seen=no; \
+	case "$@" in \
+	  distclean-* | maintainer-clean-*) list='$(DIST_SUBDIRS)' ;; \
+	  *) list='$(SUBDIRS)' ;; \
+	esac; \
+	rev=''; for subdir in $$list; do \
+	  if test "$$subdir" = "."; then :; else \
+	    rev="$$subdir $$rev"; \
+	  fi; \
+	done; \
+	rev="$$rev ."; \
+	target=`echo $@ | sed s/-recursive//`; \
+	for subdir in $$rev; do \
+	  echo "Making $$target in $$subdir"; \
+	  if test "$$subdir" = "."; then \
+	    local_target="$$target-am"; \
+	  else \
+	    local_target="$$target"; \
+	  fi; \
+	  ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) $$local_target) \
+	  || eval $$failcom; \
+	done && test -z "$$fail"
+tags-recursive:
+	list='$(SUBDIRS)'; for subdir in $$list; do \
+	  test "$$subdir" = . || ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) tags); \
+	done
+ctags-recursive:
+	list='$(SUBDIRS)'; for subdir in $$list; do \
+	  test "$$subdir" = . || ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) ctags); \
+	done
 
-tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
+ID: $(HEADERS) $(SOURCES) $(LISP) $(TAGS_FILES)
+	list='$(SOURCES) $(HEADERS) $(LISP) $(TAGS_FILES)'; \
+	unique=`for i in $$list; do \
+	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+	  done | \
+	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
+	      END { if (nonempty) { for (i in files) print i; }; }'`; \
+	mkid -fID $$unique
+tags: TAGS
+
+TAGS: tags-recursive $(HEADERS) $(SOURCES)  $(TAGS_DEPENDENCIES) \
+		$(TAGS_FILES) $(LISP)
 	set x; \
 	here=`pwd`; \
 	if ($(ETAGS) --etags-include --version) >/dev/null 2>&1; then \
@@ -387,7 +328,12 @@ tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
 	      set "$$@" "$$include_option=$$here/$$subdir/TAGS"; \
 	  fi; \
 	done; \
-	$(am__define_uniq_tagged_files); \
+	list='$(SOURCES) $(HEADERS)  $(LISP) $(TAGS_FILES)'; \
+	unique=`for i in $$list; do \
+	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+	  done | \
+	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
+	      END { if (nonempty) { for (i in files) print i; }; }'`; \
 	shift; \
 	if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
 	  test -n "$$unique" || unique=$$empty_fix; \
@@ -399,11 +345,15 @@ tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
 	      $$unique; \
 	  fi; \
 	fi
-ctags: ctags-recursive
-
-CTAGS: ctags
-ctags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
-	$(am__define_uniq_tagged_files); \
+ctags: CTAGS
+CTAGS: ctags-recursive $(HEADERS) $(SOURCES)  $(TAGS_DEPENDENCIES) \
+		$(TAGS_FILES) $(LISP)
+	list='$(SOURCES) $(HEADERS)  $(LISP) $(TAGS_FILES)'; \
+	unique=`for i in $$list; do \
+	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+	  done | \
+	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
+	      END { if (nonempty) { for (i in files) print i; }; }'`; \
 	test -z "$(CTAGS_ARGS)$$unique" \
 	  || $(CTAGS) $(CTAGSFLAGS) $(AM_CTAGSFLAGS) $(CTAGS_ARGS) \
 	     $$unique
@@ -412,31 +362,9 @@ GTAGS:
 	here=`$(am__cd) $(top_builddir) && pwd` \
 	  && $(am__cd) $(top_srcdir) \
 	  && gtags -i $(GTAGS_ARGS) "$$here"
-cscope: cscope.files
-	test ! -s cscope.files \
-	  || $(CSCOPE) -b -q $(AM_CSCOPEFLAGS) $(CSCOPEFLAGS) -i cscope.files $(CSCOPE_ARGS)
-clean-cscope:
-	-rm -f cscope.files
-cscope.files: clean-cscope cscopelist
-cscopelist: cscopelist-recursive
-
-cscopelist-am: $(am__tagged_files)
-	list='$(am__tagged_files)'; \
-	case "$(srcdir)" in \
-	  [\\/]* | ?:[\\/]*) sdir="$(srcdir)" ;; \
-	  *) sdir=$(subdir)/$(srcdir) ;; \
-	esac; \
-	for i in $$list; do \
-	  if test -f "$$i"; then \
-	    echo "$(subdir)/$$i"; \
-	  else \
-	    echo "$$sdir/$$i"; \
-	  fi; \
-	done >> $(top_builddir)/cscope.files
 
 distclean-tags:
 	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
-	-rm -f cscope.out cscope.in.out cscope.po.out cscope.files
 
 distdir: $(DISTFILES)
 	$(am__remove_distdir)
@@ -472,10 +400,13 @@ distdir: $(DISTFILES)
 	done
 	@list='$(DIST_SUBDIRS)'; for subdir in $$list; do \
 	  if test "$$subdir" = .; then :; else \
-	    $(am__make_dryrun) \
-	      || test -d "$(distdir)/$$subdir" \
-	      || $(MKDIR_P) "$(distdir)/$$subdir" \
-	      || exit 1; \
+	    test -d "$(distdir)/$$subdir" \
+	    || $(MKDIR_P) "$(distdir)/$$subdir" \
+	    || exit 1; \
+	  fi; \
+	done
+	@list='$(DIST_SUBDIRS)'; for subdir in $$list; do \
+	  if test "$$subdir" = .; then :; else \
 	    dir1=$$subdir; dir2="$(distdir)/$$subdir"; \
 	    $(am__relativize); \
 	    new_distdir=$$reldir; \
@@ -504,42 +435,36 @@ distdir: $(DISTFILES)
 	|| chmod -R a+r "$(distdir)"
 dist-gzip: distdir
 	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 
 dist-bzip2: distdir
-	tardir=$(distdir) && $(am__tar) | BZIP2=$${BZIP2--9} bzip2 -c >$(distdir).tar.bz2
-	$(am__post_remove_distdir)
+	tardir=$(distdir) && $(am__tar) | bzip2 -9 -c >$(distdir).tar.bz2
+	$(am__remove_distdir)
 
-dist-lzip: distdir
-	tardir=$(distdir) && $(am__tar) | lzip -c $${LZIP_OPT--9} >$(distdir).tar.lz
-	$(am__post_remove_distdir)
+dist-lzma: distdir
+	tardir=$(distdir) && $(am__tar) | lzma -9 -c >$(distdir).tar.lzma
+	$(am__remove_distdir)
 
 dist-xz: distdir
-	tardir=$(distdir) && $(am__tar) | XZ_OPT=$${XZ_OPT--e} xz -c >$(distdir).tar.xz
-	$(am__post_remove_distdir)
+	tardir=$(distdir) && $(am__tar) | xz -c >$(distdir).tar.xz
+	$(am__remove_distdir)
 
 dist-tarZ: distdir
-	@echo WARNING: "Support for distribution archives compressed with" \
-		       "legacy program 'compress' is deprecated." >&2
-	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	tardir=$(distdir) && $(am__tar) | compress -c >$(distdir).tar.Z
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 
 dist-shar: distdir
-	@echo WARNING: "Support for shar distribution archives is" \
-	               "deprecated." >&2
-	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	shar $(distdir) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).shar.gz
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 
 dist-zip: distdir
 	-rm -f $(distdir).zip
 	zip -rq $(distdir).zip $(distdir)
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 
-dist dist-all:
-	$(MAKE) $(AM_MAKEFLAGS) $(DIST_TARGETS) am__post_remove_distdir='@:'
-	$(am__post_remove_distdir)
+dist dist-all: distdir
+	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
+	$(am__remove_distdir)
 
 # This target untars the dist file and tries a VPATH configuration.  Then
 # it guarantees that the distribution is self-contained by making another
@@ -550,8 +475,8 @@ distcheck: dist
 	  GZIP=$(GZIP_ENV) gzip -dc $(distdir).tar.gz | $(am__untar) ;;\
 	*.tar.bz2*) \
 	  bzip2 -dc $(distdir).tar.bz2 | $(am__untar) ;;\
-	*.tar.lz*) \
-	  lzip -dc $(distdir).tar.lz | $(am__untar) ;;\
+	*.tar.lzma*) \
+	  lzma -dc $(distdir).tar.lzma | $(am__untar) ;;\
 	*.tar.xz*) \
 	  xz -dc $(distdir).tar.xz | $(am__untar) ;;\
 	*.tar.Z*) \
@@ -561,19 +486,17 @@ distcheck: dist
 	*.zip*) \
 	  unzip $(distdir).zip ;;\
 	esac
-	chmod -R a-w $(distdir)
-	chmod u+w $(distdir)
-	mkdir $(distdir)/_build $(distdir)/_build/sub $(distdir)/_inst
+	chmod -R a-w $(distdir); chmod u+w $(distdir)
+	mkdir $(distdir)/_build
+	mkdir $(distdir)/_inst
 	chmod a-w $(distdir)
 	test -d $(distdir)/_build || exit 0; \
 	dc_install_base=`$(am__cd) $(distdir)/_inst && pwd | sed -e 's,^[^:\\/]:[\\/],/,'` \
 	  && dc_destdir="$${TMPDIR-/tmp}/am-dc-$$$$/" \
 	  && am__cwd=`pwd` \
-	  && $(am__cd) $(distdir)/_build/sub \
-	  && ../../configure \
-	    $(AM_DISTCHECK_CONFIGURE_FLAGS) \
+	  && $(am__cd) $(distdir)/_build \
+	  && ../configure --srcdir=.. --prefix="$$dc_install_base" \
 	    $(DISTCHECK_CONFIGURE_FLAGS) \
-	    --srcdir=../.. --prefix="$$dc_install_base" \
 	  && $(MAKE) $(AM_MAKEFLAGS) \
 	  && $(MAKE) $(AM_MAKEFLAGS) dvi \
 	  && $(MAKE) $(AM_MAKEFLAGS) check \
@@ -596,21 +519,13 @@ distcheck: dist
 	  && $(MAKE) $(AM_MAKEFLAGS) distcleancheck \
 	  && cd "$$am__cwd" \
 	  || exit 1
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 	@(echo "$(distdir) archives ready for distribution: "; \
 	  list='$(DIST_ARCHIVES)'; for i in $$list; do echo $$i; done) | \
 	  sed -e 1h -e 1s/./=/g -e 1p -e 1x -e '$$p' -e '$$x'
 distuninstallcheck:
-	@test -n '$(distuninstallcheck_dir)' || { \
-	  echo 'ERROR: trying to run $@ with an empty' \
-	       '$$(distuninstallcheck_dir)' >&2; \
-	  exit 1; \
-	}; \
-	$(am__cd) '$(distuninstallcheck_dir)' || { \
-	  echo 'ERROR: cannot chdir into $(distuninstallcheck_dir)' >&2; \
-	  exit 1; \
-	}; \
-	test `$(am__distuninstallcheck_listfiles) | wc -l` -eq 0 \
+	@$(am__cd) '$(distuninstallcheck_dir)' \
+	&& test `$(distuninstallcheck_listfiles) | wc -l` -le 1 \
 	   || { echo "ERROR: files left after uninstall:" ; \
 	        if test -n "$(DESTDIR)"; then \
 	          echo "  (check DESTDIR support)"; \
@@ -641,15 +556,10 @@ install-am: all-am
 
 installcheck: installcheck-recursive
 install-strip:
-	if test -z '$(STRIP)'; then \
-	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
-	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
-	      install; \
-	else \
-	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
-	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
-	    "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'" install; \
-	fi
+	$(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
+	  install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
+	  `test -z '$(STRIP)' || \
+	    echo "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'"` install
 mostlyclean-generic:
 
 clean-generic:
@@ -730,24 +640,23 @@ ps-am:
 
 uninstall-am:
 
-.MAKE: $(am__recursive_targets) install-am install-strip
+.MAKE: $(RECURSIVE_CLEAN_TARGETS) $(RECURSIVE_TARGETS) ctags-recursive \
+	install-am install-strip tags-recursive
 
-.PHONY: $(am__recursive_targets) CTAGS GTAGS TAGS all all-am \
-	am--refresh check check-am clean clean-cscope clean-generic \
-	cscope cscopelist-am ctags ctags-am dist dist-all dist-bzip2 \
-	dist-gzip dist-lzip dist-shar dist-tarZ dist-xz dist-zip \
-	distcheck distclean distclean-generic distclean-tags \
-	distcleancheck distdir distuninstallcheck dvi dvi-am html \
-	html-am info info-am install install-am install-data \
-	install-data-am install-dvi install-dvi-am install-exec \
-	install-exec-am install-html install-html-am install-info \
-	install-info-am install-man install-pdf install-pdf-am \
-	install-ps install-ps-am install-strip installcheck \
-	installcheck-am installdirs installdirs-am maintainer-clean \
+.PHONY: $(RECURSIVE_CLEAN_TARGETS) $(RECURSIVE_TARGETS) CTAGS GTAGS \
+	all all-am am--refresh check check-am clean clean-generic \
+	ctags ctags-recursive dist dist-all dist-bzip2 dist-gzip \
+	dist-lzma dist-shar dist-tarZ dist-xz dist-zip distcheck \
+	distclean distclean-generic distclean-tags distcleancheck \
+	distdir distuninstallcheck dvi dvi-am html html-am info \
+	info-am install install-am install-data install-data-am \
+	install-dvi install-dvi-am install-exec install-exec-am \
+	install-html install-html-am install-info install-info-am \
+	install-man install-pdf install-pdf-am install-ps \
+	install-ps-am install-strip installcheck installcheck-am \
+	installdirs installdirs-am maintainer-clean \
 	maintainer-clean-generic mostlyclean mostlyclean-generic pdf \
-	pdf-am ps ps-am tags tags-am uninstall uninstall-am
-
-.PRECIOUS: Makefile
+	pdf-am ps ps-am tags tags-recursive uninstall uninstall-am
 
 
 # Tell versions [3.59,3.63) of GNU make to not export all variables.
diff --git a/TODO b/TODO
index 0ef20444..57769ae6 100644
--- a/TODO
+++ b/TODO
@@ -2,6 +2,8 @@
   - use protocol buffers? replace xmlReader/Writer ec..
   - Binary use htonll, htonl
 
+* Had to hack assignment to 1.0 in the tests/Grid_gamma test
+y
 * Expression template engine:
 
    - Audit
diff --git a/aclocal.m4 b/aclocal.m4
index f3018f6c..bcc213b4 100644
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -1,7 +1,7 @@
-# generated automatically by aclocal 1.15 -*- Autoconf -*-
-
-# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+# generated automatically by aclocal 1.11.1 -*- Autoconf -*-
 
+# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
+# 2005, 2006, 2007, 2008, 2009  Free Software Foundation, Inc.
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
@@ -11,16 +11,15 @@
 # even the implied warranty of MERCHANTABILITY or FITNESS FOR A
 # PARTICULAR PURPOSE.
 
-m4_ifndef([AC_CONFIG_MACRO_DIRS], [m4_defun([_AM_CONFIG_MACRO_DIRS], [])m4_defun([AC_CONFIG_MACRO_DIRS], [_AM_CONFIG_MACRO_DIRS($@)])])
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
-m4_if(m4_defn([AC_AUTOCONF_VERSION]), [2.69],,
-[m4_warning([this file was generated for autoconf 2.69.
+m4_if(m4_defn([AC_AUTOCONF_VERSION]), [2.63],,
+[m4_warning([this file was generated for autoconf 2.63.
 You have another version of autoconf.  It may work, but is not guaranteed to.
 If you have problems, you may need to regenerate the build system entirely.
-To do so, use the procedure documented by the package, typically 'autoreconf'.])])
+To do so, use the procedure documented by the package, typically `autoreconf'.])])
 
-# Copyright (C) 2002-2014 Free Software Foundation, Inc.
+# Copyright (C) 2002, 2003, 2005, 2006, 2007, 2008  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -32,10 +31,10 @@ To do so, use the procedure documented by the package, typically 'autoreconf'.])
 # generated from the m4 files accompanying Automake X.Y.
 # (This private macro should not be called outside this file.)
 AC_DEFUN([AM_AUTOMAKE_VERSION],
-[am__api_version='1.15'
+[am__api_version='1.11'
 dnl Some users find AM_AUTOMAKE_VERSION and mistake it for a way to
 dnl require some minimum version.  Point them to the right macro.
-m4_if([$1], [1.15], [],
+m4_if([$1], [1.11.1], [],
       [AC_FATAL([Do not call $0, use AM_INIT_AUTOMAKE([$1]).])])dnl
 ])
 
@@ -51,22 +50,22 @@ m4_define([_AM_AUTOCONF_VERSION], [])
 # Call AM_AUTOMAKE_VERSION and AM_AUTOMAKE_VERSION so they can be traced.
 # This function is AC_REQUIREd by AM_INIT_AUTOMAKE.
 AC_DEFUN([AM_SET_CURRENT_AUTOMAKE_VERSION],
-[AM_AUTOMAKE_VERSION([1.15])dnl
+[AM_AUTOMAKE_VERSION([1.11.1])dnl
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
 _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 
 # AM_AUX_DIR_EXPAND                                         -*- Autoconf -*-
 
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2001, 2003, 2005  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
 # For projects using AC_CONFIG_AUX_DIR([foo]), Autoconf sets
-# $ac_aux_dir to '$srcdir/foo'.  In other projects, it is set to
-# '$srcdir', '$srcdir/..', or '$srcdir/../..'.
+# $ac_aux_dir to `$srcdir/foo'.  In other projects, it is set to
+# `$srcdir', `$srcdir/..', or `$srcdir/../..'.
 #
 # Of course, Automake must honor this variable whenever it calls a
 # tool from the auxiliary directory.  The problem is that $srcdir (and
@@ -85,7 +84,7 @@ _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 #
 # The reason of the latter failure is that $top_srcdir and $ac_aux_dir
 # are both prefixed by $srcdir.  In an in-source build this is usually
-# harmless because $srcdir is '.', but things will broke when you
+# harmless because $srcdir is `.', but things will broke when you
 # start a VPATH build or use an absolute $srcdir.
 #
 # So we could use something similar to $top_srcdir/$ac_aux_dir/missing,
@@ -103,26 +102,30 @@ _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 # configured tree to be moved without reconfiguration.
 
 AC_DEFUN([AM_AUX_DIR_EXPAND],
-[AC_REQUIRE([AC_CONFIG_AUX_DIR_DEFAULT])dnl
-# Expand $ac_aux_dir to an absolute path.
-am_aux_dir=`cd "$ac_aux_dir" && pwd`
+[dnl Rely on autoconf to set up CDPATH properly.
+AC_PREREQ([2.50])dnl
+# expand $ac_aux_dir to an absolute path
+am_aux_dir=`cd $ac_aux_dir && pwd`
 ])
 
 # AM_CONDITIONAL                                            -*- Autoconf -*-
 
-# Copyright (C) 1997-2014 Free Software Foundation, Inc.
+# Copyright (C) 1997, 2000, 2001, 2003, 2004, 2005, 2006, 2008
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 9
+
 # AM_CONDITIONAL(NAME, SHELL-CONDITION)
 # -------------------------------------
 # Define a conditional.
 AC_DEFUN([AM_CONDITIONAL],
-[AC_PREREQ([2.52])dnl
- m4_if([$1], [TRUE],  [AC_FATAL([$0: invalid condition: $1])],
-       [$1], [FALSE], [AC_FATAL([$0: invalid condition: $1])])dnl
+[AC_PREREQ(2.52)dnl
+ ifelse([$1], [TRUE],  [AC_FATAL([$0: invalid condition: $1])],
+	[$1], [FALSE], [AC_FATAL([$0: invalid condition: $1])])dnl
 AC_SUBST([$1_TRUE])dnl
 AC_SUBST([$1_FALSE])dnl
 _AM_SUBST_NOTMAKE([$1_TRUE])dnl
@@ -141,14 +144,16 @@ AC_CONFIG_COMMANDS_PRE(
 Usually this means the macro was only invoked conditionally.]])
 fi])])
 
-# Copyright (C) 1999-2014 Free Software Foundation, Inc.
+# Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2009
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 10
 
-# There are a few dirty hacks below to avoid letting 'AC_PROG_CC' be
+# There are a few dirty hacks below to avoid letting `AC_PROG_CC' be
 # written in clear, in which case automake, when reading aclocal.m4,
 # will think it sees a *use*, and therefore will trigger all it's
 # C support machinery.  Also note that it means that autoscan, seeing
@@ -158,7 +163,7 @@ fi])])
 # _AM_DEPENDENCIES(NAME)
 # ----------------------
 # See how the compiler implements dependency checking.
-# NAME is "CC", "CXX", "OBJC", "OBJCXX", "UPC", or "GJC".
+# NAME is "CC", "CXX", "GCJ", or "OBJC".
 # We try a few techniques and use that to set a single cache variable.
 #
 # We don't AC_REQUIRE the corresponding AC_PROG_CC since the latter was
@@ -171,13 +176,12 @@ AC_REQUIRE([AM_OUTPUT_DEPENDENCY_COMMANDS])dnl
 AC_REQUIRE([AM_MAKE_INCLUDE])dnl
 AC_REQUIRE([AM_DEP_TRACK])dnl
 
-m4_if([$1], [CC],   [depcc="$CC"   am_compiler_list=],
-      [$1], [CXX],  [depcc="$CXX"  am_compiler_list=],
-      [$1], [OBJC], [depcc="$OBJC" am_compiler_list='gcc3 gcc'],
-      [$1], [OBJCXX], [depcc="$OBJCXX" am_compiler_list='gcc3 gcc'],
-      [$1], [UPC],  [depcc="$UPC"  am_compiler_list=],
-      [$1], [GCJ],  [depcc="$GCJ"  am_compiler_list='gcc3 gcc'],
-                    [depcc="$$1"   am_compiler_list=])
+ifelse([$1], CC,   [depcc="$CC"   am_compiler_list=],
+       [$1], CXX,  [depcc="$CXX"  am_compiler_list=],
+       [$1], OBJC, [depcc="$OBJC" am_compiler_list='gcc3 gcc'],
+       [$1], UPC,  [depcc="$UPC"  am_compiler_list=],
+       [$1], GCJ,  [depcc="$GCJ"  am_compiler_list='gcc3 gcc'],
+                   [depcc="$$1"   am_compiler_list=])
 
 AC_CACHE_CHECK([dependency style of $depcc],
                [am_cv_$1_dependencies_compiler_type],
@@ -185,9 +189,8 @@ AC_CACHE_CHECK([dependency style of $depcc],
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named 'D' -- because '-MD' means "put the output
-  # in D".
-  rm -rf conftest.dir
+  # making a dummy file named `D' -- because `-MD' means `put the output
+  # in D'.
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -226,16 +229,16 @@ AC_CACHE_CHECK([dependency style of $depcc],
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
-      # Solaris 10 /bin/sh.
-      echo '/* dummy */' > sub/conftst$i.h
+      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
+      # Solaris 8's {/usr,}/bin/sh.
+      touch sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with '-c' and '-o' for the sake of the "dashmstdout"
+    # We check with `-c' and `-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle '-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs.
+    # handle `-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -244,16 +247,16 @@ AC_CACHE_CHECK([dependency style of $depcc],
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # After this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested.
+      # after this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
-      # This compiler won't grok '-c -o', but also, the minuso test has
+    msvisualcpp | msvcmsys)
+      # This compiler won't grok `-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -301,7 +304,7 @@ AM_CONDITIONAL([am__fastdep$1], [
 # AM_SET_DEPDIR
 # -------------
 # Choose a directory name for dependency files.
-# This macro is AC_REQUIREd in _AM_DEPENDENCIES.
+# This macro is AC_REQUIREd in _AM_DEPENDENCIES
 AC_DEFUN([AM_SET_DEPDIR],
 [AC_REQUIRE([AM_SET_LEADING_DOT])dnl
 AC_SUBST([DEPDIR], ["${am__leading_dot}deps"])dnl
@@ -311,39 +314,34 @@ AC_SUBST([DEPDIR], ["${am__leading_dot}deps"])dnl
 # AM_DEP_TRACK
 # ------------
 AC_DEFUN([AM_DEP_TRACK],
-[AC_ARG_ENABLE([dependency-tracking], [dnl
-AS_HELP_STRING(
-  [--enable-dependency-tracking],
-  [do not reject slow dependency extractors])
-AS_HELP_STRING(
-  [--disable-dependency-tracking],
-  [speeds up one-time build])])
+[AC_ARG_ENABLE(dependency-tracking,
+[  --disable-dependency-tracking  speeds up one-time build
+  --enable-dependency-tracking   do not reject slow dependency extractors])
 if test "x$enable_dependency_tracking" != xno; then
   am_depcomp="$ac_aux_dir/depcomp"
   AMDEPBACKSLASH='\'
-  am__nodep='_no'
 fi
 AM_CONDITIONAL([AMDEP], [test "x$enable_dependency_tracking" != xno])
 AC_SUBST([AMDEPBACKSLASH])dnl
 _AM_SUBST_NOTMAKE([AMDEPBACKSLASH])dnl
-AC_SUBST([am__nodep])dnl
-_AM_SUBST_NOTMAKE([am__nodep])dnl
 ])
 
 # Generate code to set up dependency tracking.              -*- Autoconf -*-
 
-# Copyright (C) 1999-2014 Free Software Foundation, Inc.
+# Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2008
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+#serial 5
 
 # _AM_OUTPUT_DEPENDENCY_COMMANDS
 # ------------------------------
 AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
 [{
-  # Older Autoconf quotes --file arguments for eval, but not when files
+  # Autoconf 2.62 quotes --file arguments for eval, but not when files
   # are listed without --file.  Let's play safe and only enable the eval
   # if we detect the quoting.
   case $CONFIG_FILES in
@@ -356,7 +354,7 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
     # Strip MF so we end up with the name of the file.
     mf=`echo "$mf" | sed -e 's/:.*$//'`
     # Check whether this is an Automake generated Makefile or not.
-    # We used to match only the files named 'Makefile.in', but
+    # We used to match only the files named `Makefile.in', but
     # some people rename them; so instead we look at the file content.
     # Grep'ing the first line is not enough: some people post-process
     # each Makefile.in and add a new line on top of each file to say so.
@@ -368,19 +366,21 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
       continue
     fi
     # Extract the definition of DEPDIR, am__include, and am__quote
-    # from the Makefile without running 'make'.
+    # from the Makefile without running `make'.
     DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
     test -z "$DEPDIR" && continue
     am__include=`sed -n 's/^am__include = //p' < "$mf"`
-    test -z "$am__include" && continue
+    test -z "am__include" && continue
     am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
+    # When using ansi2knr, U may be empty or an underscore; expand it
+    U=`sed -n 's/^U = //p' < "$mf"`
     # Find all dependency output files, they are included files with
     # $(DEPDIR) in their names.  We invoke sed twice because it is the
     # simplest approach to changing $(DEPDIR) to its actual value in the
     # expansion.
     for file in `sed -n "
       s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
-	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
+	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g' -e 's/\$U/'"$U"'/g'`; do
       # Make sure the directory exists.
       test -f "$dirpart/$file" && continue
       fdir=`AS_DIRNAME(["$file"])`
@@ -398,7 +398,7 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
 # This macro should only be invoked once -- use via AC_REQUIRE.
 #
 # This code is only required when automatic dependency tracking
-# is enabled.  FIXME.  This creates each '.P' file that we will
+# is enabled.  FIXME.  This creates each `.P' file that we will
 # need in order to bootstrap the dependency handling code.
 AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 [AC_CONFIG_COMMANDS([depfiles],
@@ -408,21 +408,18 @@ AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 
 # Do all the work for Automake.                             -*- Autoconf -*-
 
-# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
+# 2005, 2006, 2008, 2009 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 16
+
 # This macro actually does too much.  Some checks are only needed if
 # your package does certain things.  But this isn't really a big deal.
 
-dnl Redefine AC_PROG_CC to automatically invoke _AM_PROG_CC_C_O.
-m4_define([AC_PROG_CC],
-m4_defn([AC_PROG_CC])
-[_AM_PROG_CC_C_O
-])
-
 # AM_INIT_AUTOMAKE(PACKAGE, VERSION, [NO-DEFINE])
 # AM_INIT_AUTOMAKE([OPTIONS])
 # -----------------------------------------------
@@ -435,7 +432,7 @@ m4_defn([AC_PROG_CC])
 # arguments mandatory, and then we can depend on a new Autoconf
 # release and drop the old call support.
 AC_DEFUN([AM_INIT_AUTOMAKE],
-[AC_PREREQ([2.65])dnl
+[AC_PREREQ([2.62])dnl
 dnl Autoconf wants to disallow AM_ names.  We explicitly allow
 dnl the ones we care about.
 m4_pattern_allow([^AM_[A-Z]+FLAGS$])dnl
@@ -464,42 +461,33 @@ AC_SUBST([CYGPATH_W])
 # Define the identity of the package.
 dnl Distinguish between old-style and new-style calls.
 m4_ifval([$2],
-[AC_DIAGNOSE([obsolete],
-             [$0: two- and three-arguments forms are deprecated.])
-m4_ifval([$3], [_AM_SET_OPTION([no-define])])dnl
+[m4_ifval([$3], [_AM_SET_OPTION([no-define])])dnl
  AC_SUBST([PACKAGE], [$1])dnl
  AC_SUBST([VERSION], [$2])],
 [_AM_SET_OPTIONS([$1])dnl
 dnl Diagnose old-style AC_INIT with new-style AM_AUTOMAKE_INIT.
-m4_if(
-  m4_ifdef([AC_PACKAGE_NAME], [ok]):m4_ifdef([AC_PACKAGE_VERSION], [ok]),
-  [ok:ok],,
+m4_if(m4_ifdef([AC_PACKAGE_NAME], 1)m4_ifdef([AC_PACKAGE_VERSION], 1), 11,,
   [m4_fatal([AC_INIT should be called with package and version arguments])])dnl
  AC_SUBST([PACKAGE], ['AC_PACKAGE_TARNAME'])dnl
  AC_SUBST([VERSION], ['AC_PACKAGE_VERSION'])])dnl
 
 _AM_IF_OPTION([no-define],,
-[AC_DEFINE_UNQUOTED([PACKAGE], ["$PACKAGE"], [Name of package])
- AC_DEFINE_UNQUOTED([VERSION], ["$VERSION"], [Version number of package])])dnl
+[AC_DEFINE_UNQUOTED(PACKAGE, "$PACKAGE", [Name of package])
+ AC_DEFINE_UNQUOTED(VERSION, "$VERSION", [Version number of package])])dnl
 
 # Some tools Automake needs.
 AC_REQUIRE([AM_SANITY_CHECK])dnl
 AC_REQUIRE([AC_ARG_PROGRAM])dnl
-AM_MISSING_PROG([ACLOCAL], [aclocal-${am__api_version}])
-AM_MISSING_PROG([AUTOCONF], [autoconf])
-AM_MISSING_PROG([AUTOMAKE], [automake-${am__api_version}])
-AM_MISSING_PROG([AUTOHEADER], [autoheader])
-AM_MISSING_PROG([MAKEINFO], [makeinfo])
+AM_MISSING_PROG(ACLOCAL, aclocal-${am__api_version})
+AM_MISSING_PROG(AUTOCONF, autoconf)
+AM_MISSING_PROG(AUTOMAKE, automake-${am__api_version})
+AM_MISSING_PROG(AUTOHEADER, autoheader)
+AM_MISSING_PROG(MAKEINFO, makeinfo)
 AC_REQUIRE([AM_PROG_INSTALL_SH])dnl
 AC_REQUIRE([AM_PROG_INSTALL_STRIP])dnl
-AC_REQUIRE([AC_PROG_MKDIR_P])dnl
-# For better backward compatibility.  To be removed once Automake 1.9.x
-# dies out for good.  For more background, see:
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
-AC_SUBST([mkdir_p], ['$(MKDIR_P)'])
-# We need awk for the "check" target (and possibly the TAP driver).  The
-# system "awk" is bad on some platforms.
+AC_REQUIRE([AM_PROG_MKDIR_P])dnl
+# We need awk for the "check" target.  The system "awk" is bad on
+# some platforms.
 AC_REQUIRE([AC_PROG_AWK])dnl
 AC_REQUIRE([AC_PROG_MAKE_SET])dnl
 AC_REQUIRE([AM_SET_LEADING_DOT])dnl
@@ -508,82 +496,34 @@ _AM_IF_OPTION([tar-ustar], [_AM_PROG_TAR([ustar])],
 			     [_AM_PROG_TAR([v7])])])
 _AM_IF_OPTION([no-dependencies],,
 [AC_PROVIDE_IFELSE([AC_PROG_CC],
-		  [_AM_DEPENDENCIES([CC])],
-		  [m4_define([AC_PROG_CC],
-			     m4_defn([AC_PROG_CC])[_AM_DEPENDENCIES([CC])])])dnl
+		  [_AM_DEPENDENCIES(CC)],
+		  [define([AC_PROG_CC],
+			  defn([AC_PROG_CC])[_AM_DEPENDENCIES(CC)])])dnl
 AC_PROVIDE_IFELSE([AC_PROG_CXX],
-		  [_AM_DEPENDENCIES([CXX])],
-		  [m4_define([AC_PROG_CXX],
-			     m4_defn([AC_PROG_CXX])[_AM_DEPENDENCIES([CXX])])])dnl
+		  [_AM_DEPENDENCIES(CXX)],
+		  [define([AC_PROG_CXX],
+			  defn([AC_PROG_CXX])[_AM_DEPENDENCIES(CXX)])])dnl
 AC_PROVIDE_IFELSE([AC_PROG_OBJC],
-		  [_AM_DEPENDENCIES([OBJC])],
-		  [m4_define([AC_PROG_OBJC],
-			     m4_defn([AC_PROG_OBJC])[_AM_DEPENDENCIES([OBJC])])])dnl
-AC_PROVIDE_IFELSE([AC_PROG_OBJCXX],
-		  [_AM_DEPENDENCIES([OBJCXX])],
-		  [m4_define([AC_PROG_OBJCXX],
-			     m4_defn([AC_PROG_OBJCXX])[_AM_DEPENDENCIES([OBJCXX])])])dnl
+		  [_AM_DEPENDENCIES(OBJC)],
+		  [define([AC_PROG_OBJC],
+			  defn([AC_PROG_OBJC])[_AM_DEPENDENCIES(OBJC)])])dnl
 ])
-AC_REQUIRE([AM_SILENT_RULES])dnl
-dnl The testsuite driver may need to know about EXEEXT, so add the
-dnl 'am__EXEEXT' conditional if _AM_COMPILER_EXEEXT was seen.  This
-dnl macro is hooked onto _AC_COMPILER_EXEEXT early, see below.
+_AM_IF_OPTION([silent-rules], [AC_REQUIRE([AM_SILENT_RULES])])dnl
+dnl The `parallel-tests' driver may need to know about EXEEXT, so add the
+dnl `am__EXEEXT' conditional if _AM_COMPILER_EXEEXT was seen.  This macro
+dnl is hooked onto _AC_COMPILER_EXEEXT early, see below.
 AC_CONFIG_COMMANDS_PRE(dnl
 [m4_provide_if([_AM_COMPILER_EXEEXT],
   [AM_CONDITIONAL([am__EXEEXT], [test -n "$EXEEXT"])])])dnl
-
-# POSIX will say in a future version that running "rm -f" with no argument
-# is OK; and we want to be able to make that assumption in our Makefile
-# recipes.  So use an aggressive probe to check that the usage we want is
-# actually supported "in the wild" to an acceptable degree.
-# See automake bug#10828.
-# To make any issue more visible, cause the running configure to be aborted
-# by default if the 'rm' program in use doesn't match our expectations; the
-# user can still override this though.
-if rm -f && rm -fr && rm -rf; then : OK; else
-  cat >&2 <<'END'
-Oops!
-
-Your 'rm' program seems unable to run without file operands specified
-on the command line, even when the '-f' option is present.  This is contrary
-to the behaviour of most rm programs out there, and not conforming with
-the upcoming POSIX standard: <http://austingroupbugs.net/view.php?id=542>
-
-Please tell bug-automake@gnu.org about your system, including the value
-of your $PATH and any error possibly output before this message.  This
-can help us improve future automake versions.
-
-END
-  if test x"$ACCEPT_INFERIOR_RM_PROGRAM" = x"yes"; then
-    echo 'Configuration will proceed anyway, since you have set the' >&2
-    echo 'ACCEPT_INFERIOR_RM_PROGRAM variable to "yes"' >&2
-    echo >&2
-  else
-    cat >&2 <<'END'
-Aborting the configuration process, to ensure you take notice of the issue.
-
-You can download and install GNU coreutils to get an 'rm' implementation
-that behaves properly: <http://www.gnu.org/software/coreutils/>.
-
-If you want to complete the configuration process using your problematic
-'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
-to "yes", and re-run configure.
-
-END
-    AC_MSG_ERROR([Your 'rm' program is bad, sorry.])
-  fi
-fi
-dnl The trailing newline in this macro's definition is deliberate, for
-dnl backward compatibility and to allow trailing 'dnl'-style comments
-dnl after the AM_INIT_AUTOMAKE invocation. See automake bug#16841.
 ])
 
-dnl Hook into '_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
+dnl Hook into `_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
 dnl add the conditional right here, as _AC_COMPILER_EXEEXT may be further
 dnl mangled by Autoconf and run in a shell conditional statement.
 m4_define([_AC_COMPILER_EXEEXT],
 m4_defn([_AC_COMPILER_EXEEXT])[m4_provide([_AM_COMPILER_EXEEXT])])
 
+
 # When config.status generates a header, we must update the stamp-h file.
 # This file resides in the same directory as the config header
 # that is generated.  The stamp files are numbered to have different names.
@@ -605,7 +545,7 @@ for _am_header in $config_headers :; do
 done
 echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_count])
 
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2001, 2003, 2005, 2008  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -616,7 +556,7 @@ echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_co
 # Define $install_sh.
 AC_DEFUN([AM_PROG_INSTALL_SH],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
-if test x"${install_sh+set}" != xset; then
+if test x"${install_sh}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -624,14 +564,16 @@ if test x"${install_sh+set}" != xset; then
     install_sh="\${SHELL} $am_aux_dir/install-sh"
   esac
 fi
-AC_SUBST([install_sh])])
+AC_SUBST(install_sh)])
 
-# Copyright (C) 2003-2014 Free Software Foundation, Inc.
+# Copyright (C) 2003, 2005  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 2
+
 # Check whether the underlying file-system supports filenames
 # with a leading dot.  For instance MS-DOS doesn't.
 AC_DEFUN([AM_SET_LEADING_DOT],
@@ -647,12 +589,14 @@ AC_SUBST([am__leading_dot])])
 
 # Check to see how 'make' treats includes.	            -*- Autoconf -*-
 
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2001, 2002, 2003, 2005, 2009  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 4
+
 # AM_MAKE_INCLUDE()
 # -----------------
 # Check to see how make treats includes.
@@ -670,7 +614,7 @@ am__quote=
 _am_result=none
 # First try GNU make style include.
 echo "include confinc" > confmf
-# Ignore all kinds of additional output from 'make'.
+# Ignore all kinds of additional output from `make'.
 case `$am_make -s -f confmf 2> /dev/null` in #(
 *the\ am__doit\ target*)
   am__include=include
@@ -697,12 +641,15 @@ rm -f confinc confmf
 
 # Fake the existence of programs that GNU maintainers use.  -*- Autoconf -*-
 
-# Copyright (C) 1997-2014 Free Software Foundation, Inc.
+# Copyright (C) 1997, 1999, 2000, 2001, 2003, 2004, 2005, 2008
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 6
+
 # AM_MISSING_PROG(NAME, PROGRAM)
 # ------------------------------
 AC_DEFUN([AM_MISSING_PROG],
@@ -710,10 +657,11 @@ AC_DEFUN([AM_MISSING_PROG],
 $1=${$1-"${am_missing_run}$2"}
 AC_SUBST($1)])
 
+
 # AM_MISSING_HAS_RUN
 # ------------------
-# Define MISSING if not defined so far and test if it is modern enough.
-# If it is, set am_missing_run to use it, otherwise, to nothing.
+# Define MISSING if not defined so far and test if it supports --run.
+# If it does, set am_missing_run to use it, otherwise, to nothing.
 AC_DEFUN([AM_MISSING_HAS_RUN],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
 AC_REQUIRE_AUX_FILE([missing])dnl
@@ -726,35 +674,63 @@ if test x"${MISSING+set}" != xset; then
   esac
 fi
 # Use eval to expand $SHELL
-if eval "$MISSING --is-lightweight"; then
-  am_missing_run="$MISSING "
+if eval "$MISSING --run true"; then
+  am_missing_run="$MISSING --run "
 else
   am_missing_run=
-  AC_MSG_WARN(['missing' script is too old or missing])
+  AC_MSG_WARN([`missing' script is too old or missing])
 fi
 ])
 
-# Helper functions for option handling.                     -*- Autoconf -*-
-
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2003, 2004, 2005, 2006  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# AM_PROG_MKDIR_P
+# ---------------
+# Check for `mkdir -p'.
+AC_DEFUN([AM_PROG_MKDIR_P],
+[AC_PREREQ([2.60])dnl
+AC_REQUIRE([AC_PROG_MKDIR_P])dnl
+dnl Automake 1.8 to 1.9.6 used to define mkdir_p.  We now use MKDIR_P,
+dnl while keeping a definition of mkdir_p for backward compatibility.
+dnl @MKDIR_P@ is magic: AC_OUTPUT adjusts its value for each Makefile.
+dnl However we cannot define mkdir_p as $(MKDIR_P) for the sake of
+dnl Makefile.ins that do not define MKDIR_P, so we do our own
+dnl adjustment using top_builddir (which is defined more often than
+dnl MKDIR_P).
+AC_SUBST([mkdir_p], ["$MKDIR_P"])dnl
+case $mkdir_p in
+  [[\\/$]]* | ?:[[\\/]]*) ;;
+  */*) mkdir_p="\$(top_builddir)/$mkdir_p" ;;
+esac
+])
+
+# Helper functions for option handling.                     -*- Autoconf -*-
+
+# Copyright (C) 2001, 2002, 2003, 2005, 2008  Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# serial 4
+
 # _AM_MANGLE_OPTION(NAME)
 # -----------------------
 AC_DEFUN([_AM_MANGLE_OPTION],
 [[_AM_OPTION_]m4_bpatsubst($1, [[^a-zA-Z0-9_]], [_])])
 
 # _AM_SET_OPTION(NAME)
-# --------------------
+# ------------------------------
 # Set option NAME.  Presently that only means defining a flag for this option.
 AC_DEFUN([_AM_SET_OPTION],
-[m4_define(_AM_MANGLE_OPTION([$1]), [1])])
+[m4_define(_AM_MANGLE_OPTION([$1]), 1)])
 
 # _AM_SET_OPTIONS(OPTIONS)
-# ------------------------
+# ----------------------------------
 # OPTIONS is a space-separated list of Automake options.
 AC_DEFUN([_AM_SET_OPTIONS],
 [m4_foreach_w([_AM_Option], [$1], [_AM_SET_OPTION(_AM_Option)])])
@@ -765,82 +741,24 @@ AC_DEFUN([_AM_SET_OPTIONS],
 AC_DEFUN([_AM_IF_OPTION],
 [m4_ifset(_AM_MANGLE_OPTION([$1]), [$2], [$3])])
 
-# Copyright (C) 1999-2014 Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# _AM_PROG_CC_C_O
-# ---------------
-# Like AC_PROG_CC_C_O, but changed for automake.  We rewrite AC_PROG_CC
-# to automatically call this.
-AC_DEFUN([_AM_PROG_CC_C_O],
-[AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
-AC_REQUIRE_AUX_FILE([compile])dnl
-AC_LANG_PUSH([C])dnl
-AC_CACHE_CHECK(
-  [whether $CC understands -c and -o together],
-  [am_cv_prog_cc_c_o],
-  [AC_LANG_CONFTEST([AC_LANG_PROGRAM([])])
-  # Make sure it works both with $CC and with simple cc.
-  # Following AC_PROG_CC_C_O, we do the test twice because some
-  # compilers refuse to overwrite an existing .o file with -o,
-  # though they will create one.
-  am_cv_prog_cc_c_o=yes
-  for am_i in 1 2; do
-    if AM_RUN_LOG([$CC -c conftest.$ac_ext -o conftest2.$ac_objext]) \
-         && test -f conftest2.$ac_objext; then
-      : OK
-    else
-      am_cv_prog_cc_c_o=no
-      break
-    fi
-  done
-  rm -f core conftest*
-  unset am_i])
-if test "$am_cv_prog_cc_c_o" != yes; then
-   # Losing compiler, so override with the script.
-   # FIXME: It is wrong to rewrite CC.
-   # But if we don't then we get into trouble of one sort or another.
-   # A longer-term fix would be to have automake use am__CC in this case,
-   # and then we could set am__CC="\$(top_srcdir)/compile \$(CC)"
-   CC="$am_aux_dir/compile $CC"
-fi
-AC_LANG_POP([C])])
-
-# For backward compatibility.
-AC_DEFUN_ONCE([AM_PROG_CC_C_O], [AC_REQUIRE([AC_PROG_CC])])
-
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# AM_RUN_LOG(COMMAND)
-# -------------------
-# Run COMMAND, save the exit status in ac_status, and log it.
-# (This has been adapted from Autoconf's _AC_RUN_LOG macro.)
-AC_DEFUN([AM_RUN_LOG],
-[{ echo "$as_me:$LINENO: $1" >&AS_MESSAGE_LOG_FD
-   ($1) >&AS_MESSAGE_LOG_FD 2>&AS_MESSAGE_LOG_FD
-   ac_status=$?
-   echo "$as_me:$LINENO: \$? = $ac_status" >&AS_MESSAGE_LOG_FD
-   (exit $ac_status); }])
-
 # Check to make sure that the build environment is sane.    -*- Autoconf -*-
 
-# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+# Copyright (C) 1996, 1997, 2000, 2001, 2003, 2005, 2008
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 5
+
 # AM_SANITY_CHECK
 # ---------------
 AC_DEFUN([AM_SANITY_CHECK],
 [AC_MSG_CHECKING([whether build environment is sane])
+# Just in case
+sleep 1
+echo timestamp > conftest.file
 # Reject unsafe characters in $srcdir or the absolute working directory
 # name.  Accept space and tab only in the latter.
 am_lf='
@@ -851,40 +769,32 @@ case `pwd` in
 esac
 case $srcdir in
   *[[\\\"\#\$\&\'\`$am_lf\ \	]]*)
-    AC_MSG_ERROR([unsafe srcdir value: '$srcdir']);;
+    AC_MSG_ERROR([unsafe srcdir value: `$srcdir']);;
 esac
 
-# Do 'set' in a subshell so we don't clobber the current shell's
+# Do `set' in a subshell so we don't clobber the current shell's
 # arguments.  Must try -L first in case configure is actually a
 # symlink; some systems play weird games with the mod time of symlinks
 # (eg FreeBSD returns the mod time of the symlink's containing
 # directory).
 if (
-   am_has_slept=no
-   for am_try in 1 2; do
-     echo "timestamp, slept: $am_has_slept" > conftest.file
-     set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
-     if test "$[*]" = "X"; then
-	# -L didn't work.
-	set X `ls -t "$srcdir/configure" conftest.file`
-     fi
-     if test "$[*]" != "X $srcdir/configure conftest.file" \
-	&& test "$[*]" != "X conftest.file $srcdir/configure"; then
+   set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
+   if test "$[*]" = "X"; then
+      # -L didn't work.
+      set X `ls -t "$srcdir/configure" conftest.file`
+   fi
+   rm -f conftest.file
+   if test "$[*]" != "X $srcdir/configure conftest.file" \
+      && test "$[*]" != "X conftest.file $srcdir/configure"; then
+
+      # If neither matched, then we have a broken ls.  This can happen
+      # if, for instance, CONFIG_SHELL is bash and it inherits a
+      # broken ls alias from the environment.  This has actually
+      # happened.  Such a system could not be considered "sane".
+      AC_MSG_ERROR([ls -t appears to fail.  Make sure there is not a broken
+alias in your environment])
+   fi
 
-	# If neither matched, then we have a broken ls.  This can happen
-	# if, for instance, CONFIG_SHELL is bash and it inherits a
-	# broken ls alias from the environment.  This has actually
-	# happened.  Such a system could not be considered "sane".
-	AC_MSG_ERROR([ls -t appears to fail.  Make sure there is not a broken
-  alias in your environment])
-     fi
-     if test "$[2]" = conftest.file || test $am_try -eq 2; then
-       break
-     fi
-     # Just in case.
-     sleep 1
-     am_has_slept=yes
-   done
    test "$[2]" = conftest.file
    )
 then
@@ -894,85 +804,9 @@ else
    AC_MSG_ERROR([newly created file is older than distributed files!
 Check your system clock])
 fi
-AC_MSG_RESULT([yes])
-# If we didn't sleep, we still need to ensure time stamps of config.status and
-# generated files are strictly newer.
-am_sleep_pid=
-if grep 'slept: no' conftest.file >/dev/null 2>&1; then
-  ( sleep 1 ) &
-  am_sleep_pid=$!
-fi
-AC_CONFIG_COMMANDS_PRE(
-  [AC_MSG_CHECKING([that generated files are newer than configure])
-   if test -n "$am_sleep_pid"; then
-     # Hide warnings about reused PIDs.
-     wait $am_sleep_pid 2>/dev/null
-   fi
-   AC_MSG_RESULT([done])])
-rm -f conftest.file
-])
+AC_MSG_RESULT(yes)])
 
-# Copyright (C) 2009-2014 Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# AM_SILENT_RULES([DEFAULT])
-# --------------------------
-# Enable less verbose build rules; with the default set to DEFAULT
-# ("yes" being less verbose, "no" or empty being verbose).
-AC_DEFUN([AM_SILENT_RULES],
-[AC_ARG_ENABLE([silent-rules], [dnl
-AS_HELP_STRING(
-  [--enable-silent-rules],
-  [less verbose build output (undo: "make V=1")])
-AS_HELP_STRING(
-  [--disable-silent-rules],
-  [verbose build output (undo: "make V=0")])dnl
-])
-case $enable_silent_rules in @%:@ (((
-  yes) AM_DEFAULT_VERBOSITY=0;;
-   no) AM_DEFAULT_VERBOSITY=1;;
-    *) AM_DEFAULT_VERBOSITY=m4_if([$1], [yes], [0], [1]);;
-esac
-dnl
-dnl A few 'make' implementations (e.g., NonStop OS and NextStep)
-dnl do not support nested variable expansions.
-dnl See automake bug#9928 and bug#10237.
-am_make=${MAKE-make}
-AC_CACHE_CHECK([whether $am_make supports nested variables],
-   [am_cv_make_support_nested_variables],
-   [if AS_ECHO([['TRUE=$(BAR$(V))
-BAR0=false
-BAR1=true
-V=1
-am__doit:
-	@$(TRUE)
-.PHONY: am__doit']]) | $am_make -f - >/dev/null 2>&1; then
-  am_cv_make_support_nested_variables=yes
-else
-  am_cv_make_support_nested_variables=no
-fi])
-if test $am_cv_make_support_nested_variables = yes; then
-  dnl Using '$V' instead of '$(V)' breaks IRIX make.
-  AM_V='$(V)'
-  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
-else
-  AM_V=$AM_DEFAULT_VERBOSITY
-  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
-fi
-AC_SUBST([AM_V])dnl
-AM_SUBST_NOTMAKE([AM_V])dnl
-AC_SUBST([AM_DEFAULT_V])dnl
-AM_SUBST_NOTMAKE([AM_DEFAULT_V])dnl
-AC_SUBST([AM_DEFAULT_VERBOSITY])dnl
-AM_BACKSLASH='\'
-AC_SUBST([AM_BACKSLASH])dnl
-_AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
-])
-
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2001, 2003, 2005  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -980,32 +814,34 @@ _AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
 
 # AM_PROG_INSTALL_STRIP
 # ---------------------
-# One issue with vendor 'install' (even GNU) is that you can't
+# One issue with vendor `install' (even GNU) is that you can't
 # specify the program used to strip binaries.  This is especially
 # annoying in cross-compiling environments, where the build's strip
 # is unlikely to handle the host's binaries.
 # Fortunately install-sh will honor a STRIPPROG variable, so we
-# always use install-sh in "make install-strip", and initialize
+# always use install-sh in `make install-strip', and initialize
 # STRIPPROG with the value of the STRIP variable (set by the user).
 AC_DEFUN([AM_PROG_INSTALL_STRIP],
 [AC_REQUIRE([AM_PROG_INSTALL_SH])dnl
-# Installed binaries are usually stripped using 'strip' when the user
-# run "make install-strip".  However 'strip' might not be the right
+# Installed binaries are usually stripped using `strip' when the user
+# run `make install-strip'.  However `strip' might not be the right
 # tool to use in cross-compilation environments, therefore Automake
-# will honor the 'STRIP' environment variable to overrule this program.
-dnl Don't test for $cross_compiling = yes, because it might be 'maybe'.
+# will honor the `STRIP' environment variable to overrule this program.
+dnl Don't test for $cross_compiling = yes, because it might be `maybe'.
 if test "$cross_compiling" != no; then
   AC_CHECK_TOOL([STRIP], [strip], :)
 fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 AC_SUBST([INSTALL_STRIP_PROGRAM])])
 
-# Copyright (C) 2006-2014 Free Software Foundation, Inc.
+# Copyright (C) 2006, 2008  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 2
+
 # _AM_SUBST_NOTMAKE(VARIABLE)
 # ---------------------------
 # Prevent Automake from outputting VARIABLE = @VARIABLE@ in Makefile.in.
@@ -1013,22 +849,24 @@ AC_SUBST([INSTALL_STRIP_PROGRAM])])
 AC_DEFUN([_AM_SUBST_NOTMAKE])
 
 # AM_SUBST_NOTMAKE(VARIABLE)
-# --------------------------
+# ---------------------------
 # Public sister of _AM_SUBST_NOTMAKE.
 AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 
 # Check how to create a tarball.                            -*- Autoconf -*-
 
-# Copyright (C) 2004-2014 Free Software Foundation, Inc.
+# Copyright (C) 2004, 2005  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 2
+
 # _AM_PROG_TAR(FORMAT)
 # --------------------
 # Check how to create a tarball in format FORMAT.
-# FORMAT should be one of 'v7', 'ustar', or 'pax'.
+# FORMAT should be one of `v7', `ustar', or `pax'.
 #
 # Substitute a variable $(am__tar) that is a command
 # writing to stdout a FORMAT-tarball containing the directory
@@ -1038,114 +876,75 @@ AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 # Substitute a variable $(am__untar) that extract such
 # a tarball read from stdin.
 #     $(am__untar) < result.tar
-#
 AC_DEFUN([_AM_PROG_TAR],
-[# Always define AMTAR for backward compatibility.  Yes, it's still used
-# in the wild :-(  We should find a proper way to deprecate it ...
-AC_SUBST([AMTAR], ['$${TAR-tar}'])
-
-# We'll loop over all known methods to create a tar archive until one works.
-_am_tools='gnutar m4_if([$1], [ustar], [plaintar]) pax cpio none'
-
+[# Always define AMTAR for backward compatibility.
+AM_MISSING_PROG([AMTAR], [tar])
 m4_if([$1], [v7],
-  [am__tar='$${TAR-tar} chof - "$$tardir"' am__untar='$${TAR-tar} xf -'],
+     [am__tar='${AMTAR} chof - "$$tardir"'; am__untar='${AMTAR} xf -'],
+     [m4_case([$1], [ustar],, [pax],,
+              [m4_fatal([Unknown tar format])])
+AC_MSG_CHECKING([how to create a $1 tar archive])
+# Loop over all known methods to create a tar archive until one works.
+_am_tools='gnutar m4_if([$1], [ustar], [plaintar]) pax cpio none'
+_am_tools=${am_cv_prog_tar_$1-$_am_tools}
+# Do not fold the above two line into one, because Tru64 sh and
+# Solaris sh will not grok spaces in the rhs of `-'.
+for _am_tool in $_am_tools
+do
+  case $_am_tool in
+  gnutar)
+    for _am_tar in tar gnutar gtar;
+    do
+      AM_RUN_LOG([$_am_tar --version]) && break
+    done
+    am__tar="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$$tardir"'
+    am__tar_="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$tardir"'
+    am__untar="$_am_tar -xf -"
+    ;;
+  plaintar)
+    # Must skip GNU tar: if it does not support --format= it doesn't create
+    # ustar tarball either.
+    (tar --version) >/dev/null 2>&1 && continue
+    am__tar='tar chf - "$$tardir"'
+    am__tar_='tar chf - "$tardir"'
+    am__untar='tar xf -'
+    ;;
+  pax)
+    am__tar='pax -L -x $1 -w "$$tardir"'
+    am__tar_='pax -L -x $1 -w "$tardir"'
+    am__untar='pax -r'
+    ;;
+  cpio)
+    am__tar='find "$$tardir" -print | cpio -o -H $1 -L'
+    am__tar_='find "$tardir" -print | cpio -o -H $1 -L'
+    am__untar='cpio -i -H $1 -d'
+    ;;
+  none)
+    am__tar=false
+    am__tar_=false
+    am__untar=false
+    ;;
+  esac
 
-  [m4_case([$1],
-    [ustar],
-     [# The POSIX 1988 'ustar' format is defined with fixed-size fields.
-      # There is notably a 21 bits limit for the UID and the GID.  In fact,
-      # the 'pax' utility can hang on bigger UID/GID (see automake bug#8343
-      # and bug#13588).
-      am_max_uid=2097151 # 2^21 - 1
-      am_max_gid=$am_max_uid
-      # The $UID and $GID variables are not portable, so we need to resort
-      # to the POSIX-mandated id(1) utility.  Errors in the 'id' calls
-      # below are definitely unexpected, so allow the users to see them
-      # (that is, avoid stderr redirection).
-      am_uid=`id -u || echo unknown`
-      am_gid=`id -g || echo unknown`
-      AC_MSG_CHECKING([whether UID '$am_uid' is supported by ustar format])
-      if test $am_uid -le $am_max_uid; then
-         AC_MSG_RESULT([yes])
-      else
-         AC_MSG_RESULT([no])
-         _am_tools=none
-      fi
-      AC_MSG_CHECKING([whether GID '$am_gid' is supported by ustar format])
-      if test $am_gid -le $am_max_gid; then
-         AC_MSG_RESULT([yes])
-      else
-        AC_MSG_RESULT([no])
-        _am_tools=none
-      fi],
+  # If the value was cached, stop now.  We just wanted to have am__tar
+  # and am__untar set.
+  test -n "${am_cv_prog_tar_$1}" && break
 
-  [pax],
-    [],
-
-  [m4_fatal([Unknown tar format])])
-
-  AC_MSG_CHECKING([how to create a $1 tar archive])
-
-  # Go ahead even if we have the value already cached.  We do so because we
-  # need to set the values for the 'am__tar' and 'am__untar' variables.
-  _am_tools=${am_cv_prog_tar_$1-$_am_tools}
-
-  for _am_tool in $_am_tools; do
-    case $_am_tool in
-    gnutar)
-      for _am_tar in tar gnutar gtar; do
-        AM_RUN_LOG([$_am_tar --version]) && break
-      done
-      am__tar="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$$tardir"'
-      am__tar_="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$tardir"'
-      am__untar="$_am_tar -xf -"
-      ;;
-    plaintar)
-      # Must skip GNU tar: if it does not support --format= it doesn't create
-      # ustar tarball either.
-      (tar --version) >/dev/null 2>&1 && continue
-      am__tar='tar chf - "$$tardir"'
-      am__tar_='tar chf - "$tardir"'
-      am__untar='tar xf -'
-      ;;
-    pax)
-      am__tar='pax -L -x $1 -w "$$tardir"'
-      am__tar_='pax -L -x $1 -w "$tardir"'
-      am__untar='pax -r'
-      ;;
-    cpio)
-      am__tar='find "$$tardir" -print | cpio -o -H $1 -L'
-      am__tar_='find "$tardir" -print | cpio -o -H $1 -L'
-      am__untar='cpio -i -H $1 -d'
-      ;;
-    none)
-      am__tar=false
-      am__tar_=false
-      am__untar=false
-      ;;
-    esac
-
-    # If the value was cached, stop now.  We just wanted to have am__tar
-    # and am__untar set.
-    test -n "${am_cv_prog_tar_$1}" && break
-
-    # tar/untar a dummy directory, and stop if the command works.
-    rm -rf conftest.dir
-    mkdir conftest.dir
-    echo GrepMe > conftest.dir/file
-    AM_RUN_LOG([tardir=conftest.dir && eval $am__tar_ >conftest.tar])
-    rm -rf conftest.dir
-    if test -s conftest.tar; then
-      AM_RUN_LOG([$am__untar <conftest.tar])
-      AM_RUN_LOG([cat conftest.dir/file])
-      grep GrepMe conftest.dir/file >/dev/null 2>&1 && break
-    fi
-  done
+  # tar/untar a dummy directory, and stop if the command works
   rm -rf conftest.dir
+  mkdir conftest.dir
+  echo GrepMe > conftest.dir/file
+  AM_RUN_LOG([tardir=conftest.dir && eval $am__tar_ >conftest.tar])
+  rm -rf conftest.dir
+  if test -s conftest.tar; then
+    AM_RUN_LOG([$am__untar <conftest.tar])
+    grep GrepMe conftest.dir/file >/dev/null 2>&1 && break
+  fi
+done
+rm -rf conftest.dir
 
-  AC_CACHE_VAL([am_cv_prog_tar_$1], [am_cv_prog_tar_$1=$_am_tool])
-  AC_MSG_RESULT([$am_cv_prog_tar_$1])])
-
+AC_CACHE_VAL([am_cv_prog_tar_$1], [am_cv_prog_tar_$1=$_am_tool])
+AC_MSG_RESULT([$am_cv_prog_tar_$1])])
 AC_SUBST([am__tar])
 AC_SUBST([am__untar])
 ]) # _AM_PROG_TAR
diff --git a/configure b/configure
index 3d25f6a9..3e71855d 100755
--- a/configure
+++ b/configure
@@ -1,22 +1,20 @@
 #! /bin/sh
 # Guess values for system-dependent variables and create Makefiles.
-# Generated by GNU Autoconf 2.69 for Grid 1.0.
+# Generated by GNU Autoconf 2.63 for Grid 1.0.
 #
 # Report bugs to <paboyle@ph.ed.ac.uk>.
 #
-#
-# Copyright (C) 1992-1996, 1998-2012 Free Software Foundation, Inc.
-#
-#
+# Copyright (C) 1992, 1993, 1994, 1995, 1996, 1998, 1999, 2000, 2001,
+# 2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
 # This configure script is free software; the Free Software Foundation
 # gives unlimited permission to copy, distribute and modify it.
-## -------------------- ##
-## M4sh Initialization. ##
-## -------------------- ##
+## --------------------- ##
+## M4sh Initialization.  ##
+## --------------------- ##
 
 # Be more Bourne compatible
 DUALCASE=1; export DUALCASE # for MKS sh
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
   emulate sh
   NULLCMD=:
   # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
@@ -24,15 +22,23 @@ if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
   alias -g '${1+"$@"}'='"$@"'
   setopt NO_GLOB_SUBST
 else
-  case `(set -o) 2>/dev/null` in #(
-  *posix*) :
-    set -o posix ;; #(
-  *) :
-     ;;
+  case `(set -o) 2>/dev/null` in
+  *posix*) set -o posix ;;
 esac
+
 fi
 
 
+
+
+# PATH needs CR
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
+
 as_nl='
 '
 export as_nl
@@ -40,13 +46,7 @@ export as_nl
 as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
-# Prefer a ksh shell builtin over an external printf program on Solaris,
-# but without wasting forks for bash or zsh.
-if test -z "$BASH_VERSION$ZSH_VERSION" \
-    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
-  as_echo='print -r --'
-  as_echo_n='print -rn --'
-elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+if (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
   as_echo='printf %s\n'
   as_echo_n='printf %s'
 else
@@ -57,7 +57,7 @@ else
     as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
     as_echo_n_body='eval
       arg=$1;
-      case $arg in #(
+      case $arg in
       *"$as_nl"*)
 	expr "X$arg" : "X\\(.*\\)$as_nl";
 	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
@@ -80,6 +80,13 @@ if test "${PATH_SEPARATOR+set}" != set; then
   }
 fi
 
+# Support unset when possible.
+if ( (MAIL=60; unset MAIL) || exit) >/dev/null 2>&1; then
+  as_unset=unset
+else
+  as_unset=false
+fi
+
 
 # IFS
 # We need space, tab and new line, in precisely that order.  Quoting is
@@ -89,16 +96,15 @@ fi
 IFS=" ""	$as_nl"
 
 # Find who we are.  Look in the path if we contain no directory separator.
-as_myself=
-case $0 in #((
+case $0 in
   *[\\/]* ) as_myself=$0 ;;
   *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
 for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
-  done
+  test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+done
 IFS=$as_save_IFS
 
      ;;
@@ -110,16 +116,12 @@ if test "x$as_myself" = x; then
 fi
 if test ! -f "$as_myself"; then
   $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
-  exit 1
+  { (exit 1); exit 1; }
 fi
 
-# Unset variables that we do not need and which cause bugs (e.g. in
-# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
-# suppresses any "Segmentation fault" message there.  '((' could
-# trigger a bug in pdksh 5.2.14.
-for as_var in BASH_ENV ENV MAIL MAILPATH
-do eval test x\${$as_var+set} = xset \
-  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
+# Work around bugs in pre-3.0 UWIN ksh.
+for as_var in ENV MAIL MAILPATH
+do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
 done
 PS1='$ '
 PS2='> '
@@ -131,294 +133,7 @@ export LC_ALL
 LANGUAGE=C
 export LANGUAGE
 
-# CDPATH.
-(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
-
-# Use a proper internal environment variable to ensure we don't fall
-  # into an infinite loop, continuously re-executing ourselves.
-  if test x"${_as_can_reexec}" != xno && test "x$CONFIG_SHELL" != x; then
-    _as_can_reexec=no; export _as_can_reexec;
-    # We cannot yet assume a decent shell, so we have to provide a
-# neutralization value for shells without unset; and this also
-# works around shells that cannot unset nonexistent variables.
-# Preserve -v and -x to the replacement shell.
-BASH_ENV=/dev/null
-ENV=/dev/null
-(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
-case $- in # ((((
-  *v*x* | *x*v* ) as_opts=-vx ;;
-  *v* ) as_opts=-v ;;
-  *x* ) as_opts=-x ;;
-  * ) as_opts= ;;
-esac
-exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
-# Admittedly, this is quite paranoid, since all the known shells bail
-# out after a failed `exec'.
-$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
-as_fn_exit 255
-  fi
-  # We don't want this to propagate to other subprocesses.
-          { _as_can_reexec=; unset _as_can_reexec;}
-if test "x$CONFIG_SHELL" = x; then
-  as_bourne_compatible="if test -n \"\${ZSH_VERSION+set}\" && (emulate sh) >/dev/null 2>&1; then :
-  emulate sh
-  NULLCMD=:
-  # Pre-4.2 versions of Zsh do word splitting on \${1+\"\$@\"}, which
-  # is contrary to our usage.  Disable this feature.
-  alias -g '\${1+\"\$@\"}'='\"\$@\"'
-  setopt NO_GLOB_SUBST
-else
-  case \`(set -o) 2>/dev/null\` in #(
-  *posix*) :
-    set -o posix ;; #(
-  *) :
-     ;;
-esac
-fi
-"
-  as_required="as_fn_return () { (exit \$1); }
-as_fn_success () { as_fn_return 0; }
-as_fn_failure () { as_fn_return 1; }
-as_fn_ret_success () { return 0; }
-as_fn_ret_failure () { return 1; }
-
-exitcode=0
-as_fn_success || { exitcode=1; echo as_fn_success failed.; }
-as_fn_failure && { exitcode=1; echo as_fn_failure succeeded.; }
-as_fn_ret_success || { exitcode=1; echo as_fn_ret_success failed.; }
-as_fn_ret_failure && { exitcode=1; echo as_fn_ret_failure succeeded.; }
-if ( set x; as_fn_ret_success y && test x = \"\$1\" ); then :
-
-else
-  exitcode=1; echo positional parameters were not saved.
-fi
-test x\$exitcode = x0 || exit 1
-test -x / || exit 1"
-  as_suggested="  as_lineno_1=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_1a=\$LINENO
-  as_lineno_2=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_2a=\$LINENO
-  eval 'test \"x\$as_lineno_1'\$as_run'\" != \"x\$as_lineno_2'\$as_run'\" &&
-  test \"x\`expr \$as_lineno_1'\$as_run' + 1\`\" = \"x\$as_lineno_2'\$as_run'\"' || exit 1
-test \$(( 1 + 1 )) = 2 || exit 1"
-  if (eval "$as_required") 2>/dev/null; then :
-  as_have_required=yes
-else
-  as_have_required=no
-fi
-  if test x$as_have_required = xyes && (eval "$as_suggested") 2>/dev/null; then :
-
-else
-  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-as_found=false
-for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-  as_found=:
-  case $as_dir in #(
-	 /*)
-	   for as_base in sh bash ksh sh5; do
-	     # Try only shells that exist, to save several forks.
-	     as_shell=$as_dir/$as_base
-	     if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
-		    { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$as_shell"; } 2>/dev/null; then :
-  CONFIG_SHELL=$as_shell as_have_required=yes
-		   if { $as_echo "$as_bourne_compatible""$as_suggested" | as_run=a "$as_shell"; } 2>/dev/null; then :
-  break 2
-fi
-fi
-	   done;;
-       esac
-  as_found=false
-done
-$as_found || { if { test -f "$SHELL" || test -f "$SHELL.exe"; } &&
-	      { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$SHELL"; } 2>/dev/null; then :
-  CONFIG_SHELL=$SHELL as_have_required=yes
-fi; }
-IFS=$as_save_IFS
-
-
-      if test "x$CONFIG_SHELL" != x; then :
-  export CONFIG_SHELL
-             # We cannot yet assume a decent shell, so we have to provide a
-# neutralization value for shells without unset; and this also
-# works around shells that cannot unset nonexistent variables.
-# Preserve -v and -x to the replacement shell.
-BASH_ENV=/dev/null
-ENV=/dev/null
-(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
-case $- in # ((((
-  *v*x* | *x*v* ) as_opts=-vx ;;
-  *v* ) as_opts=-v ;;
-  *x* ) as_opts=-x ;;
-  * ) as_opts= ;;
-esac
-exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
-# Admittedly, this is quite paranoid, since all the known shells bail
-# out after a failed `exec'.
-$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
-exit 255
-fi
-
-    if test x$as_have_required = xno; then :
-  $as_echo "$0: This script requires a shell more modern than all"
-  $as_echo "$0: the shells that I found on your system."
-  if test x${ZSH_VERSION+set} = xset ; then
-    $as_echo "$0: In particular, zsh $ZSH_VERSION has bugs and should"
-    $as_echo "$0: be upgraded to zsh 4.3.4 or later."
-  else
-    $as_echo "$0: Please tell bug-autoconf@gnu.org and
-$0: paboyle@ph.ed.ac.uk about your system, including any
-$0: error possibly output before this message. Then install
-$0: a modern shell, or manually run the script under such a
-$0: shell if you do have one."
-  fi
-  exit 1
-fi
-fi
-fi
-SHELL=${CONFIG_SHELL-/bin/sh}
-export SHELL
-# Unset more variables known to interfere with behavior of common tools.
-CLICOLOR_FORCE= GREP_OPTIONS=
-unset CLICOLOR_FORCE GREP_OPTIONS
-
-## --------------------- ##
-## M4sh Shell Functions. ##
-## --------------------- ##
-# as_fn_unset VAR
-# ---------------
-# Portably unset VAR.
-as_fn_unset ()
-{
-  { eval $1=; unset $1;}
-}
-as_unset=as_fn_unset
-
-# as_fn_set_status STATUS
-# -----------------------
-# Set $? to STATUS, without forking.
-as_fn_set_status ()
-{
-  return $1
-} # as_fn_set_status
-
-# as_fn_exit STATUS
-# -----------------
-# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
-as_fn_exit ()
-{
-  set +e
-  as_fn_set_status $1
-  exit $1
-} # as_fn_exit
-
-# as_fn_mkdir_p
-# -------------
-# Create "$as_dir" as a directory, including parents if necessary.
-as_fn_mkdir_p ()
-{
-
-  case $as_dir in #(
-  -*) as_dir=./$as_dir;;
-  esac
-  test -d "$as_dir" || eval $as_mkdir_p || {
-    as_dirs=
-    while :; do
-      case $as_dir in #(
-      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
-      *) as_qdir=$as_dir;;
-      esac
-      as_dirs="'$as_qdir' $as_dirs"
-      as_dir=`$as_dirname -- "$as_dir" ||
-$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_dir" : 'X\(//\)[^/]' \| \
-	 X"$as_dir" : 'X\(//\)$' \| \
-	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_dir" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      test -d "$as_dir" && break
-    done
-    test -z "$as_dirs" || eval "mkdir $as_dirs"
-  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
-
-
-} # as_fn_mkdir_p
-
-# as_fn_executable_p FILE
-# -----------------------
-# Test if FILE is an executable regular file.
-as_fn_executable_p ()
-{
-  test -f "$1" && test -x "$1"
-} # as_fn_executable_p
-# as_fn_append VAR VALUE
-# ----------------------
-# Append the text in VALUE to the end of the definition contained in VAR. Take
-# advantage of any shell optimizations that allow amortized linear growth over
-# repeated appends, instead of the typical quadratic growth present in naive
-# implementations.
-if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
-  eval 'as_fn_append ()
-  {
-    eval $1+=\$2
-  }'
-else
-  as_fn_append ()
-  {
-    eval $1=\$$1\$2
-  }
-fi # as_fn_append
-
-# as_fn_arith ARG...
-# ------------------
-# Perform arithmetic evaluation on the ARGs, and store the result in the
-# global $as_val. Take advantage of shells that can avoid forks. The arguments
-# must be portable across $(()) and expr.
-if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
-  eval 'as_fn_arith ()
-  {
-    as_val=$(( $* ))
-  }'
-else
-  as_fn_arith ()
-  {
-    as_val=`expr "$@" || test $? -eq 1`
-  }
-fi # as_fn_arith
-
-
-# as_fn_error STATUS ERROR [LINENO LOG_FD]
-# ----------------------------------------
-# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
-# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
-# script with STATUS, using 1 if that was 0.
-as_fn_error ()
-{
-  as_status=$1; test $as_status -eq 0 && as_status=1
-  if test "$4"; then
-    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
-  fi
-  $as_echo "$as_me: error: $2" >&2
-  as_fn_exit $as_status
-} # as_fn_error
-
+# Required to use basename.
 if expr a : '\(a\)' >/dev/null 2>&1 &&
    test "X`expr 00001 : '.*\(...\)'`" = X001; then
   as_expr=expr
@@ -432,12 +147,8 @@ else
   as_basename=false
 fi
 
-if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
-  as_dirname=dirname
-else
-  as_dirname=false
-fi
 
+# Name of the executable.
 as_me=`$as_basename -- "$0" ||
 $as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
 	 X"$0" : 'X\(//\)$' \| \
@@ -457,19 +168,295 @@ $as_echo X/"$0" |
 	  }
 	  s/.*/./; q'`
 
-# Avoid depending upon Character Ranges.
-as_cr_letters='abcdefghijklmnopqrstuvwxyz'
-as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
-as_cr_Letters=$as_cr_letters$as_cr_LETTERS
-as_cr_digits='0123456789'
-as_cr_alnum=$as_cr_Letters$as_cr_digits
+# CDPATH.
+$as_unset CDPATH
 
 
-  as_lineno_1=$LINENO as_lineno_1a=$LINENO
-  as_lineno_2=$LINENO as_lineno_2a=$LINENO
-  eval 'test "x$as_lineno_1'$as_run'" != "x$as_lineno_2'$as_run'" &&
-  test "x`expr $as_lineno_1'$as_run' + 1`" = "x$as_lineno_2'$as_run'"' || {
-  # Blame Lee E. McMahon (1931-1989) for sed's syntax.  :-)
+if test "x$CONFIG_SHELL" = x; then
+  if (eval ":") 2>/dev/null; then
+  as_have_required=yes
+else
+  as_have_required=no
+fi
+
+  if test $as_have_required = yes &&	 (eval ":
+(as_func_return () {
+  (exit \$1)
+}
+as_func_success () {
+  as_func_return 0
+}
+as_func_failure () {
+  as_func_return 1
+}
+as_func_ret_success () {
+  return 0
+}
+as_func_ret_failure () {
+  return 1
+}
+
+exitcode=0
+if as_func_success; then
+  :
+else
+  exitcode=1
+  echo as_func_success failed.
+fi
+
+if as_func_failure; then
+  exitcode=1
+  echo as_func_failure succeeded.
+fi
+
+if as_func_ret_success; then
+  :
+else
+  exitcode=1
+  echo as_func_ret_success failed.
+fi
+
+if as_func_ret_failure; then
+  exitcode=1
+  echo as_func_ret_failure succeeded.
+fi
+
+if ( set x; as_func_ret_success y && test x = \"\$1\" ); then
+  :
+else
+  exitcode=1
+  echo positional parameters were not saved.
+fi
+
+test \$exitcode = 0) || { (exit 1); exit 1; }
+
+(
+  as_lineno_1=\$LINENO
+  as_lineno_2=\$LINENO
+  test \"x\$as_lineno_1\" != \"x\$as_lineno_2\" &&
+  test \"x\`expr \$as_lineno_1 + 1\`\" = \"x\$as_lineno_2\") || { (exit 1); exit 1; }
+") 2> /dev/null; then
+  :
+else
+  as_candidate_shells=
+    as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+  case $as_dir in
+	 /*)
+	   for as_base in sh bash ksh sh5; do
+	     as_candidate_shells="$as_candidate_shells $as_dir/$as_base"
+	   done;;
+       esac
+done
+IFS=$as_save_IFS
+
+
+      for as_shell in $as_candidate_shells $SHELL; do
+	 # Try only shells that exist, to save several forks.
+	 if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
+		{ ("$as_shell") 2> /dev/null <<\_ASEOF
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
+  emulate sh
+  NULLCMD=:
+  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '${1+"$@"}'='"$@"'
+  setopt NO_GLOB_SUBST
+else
+  case `(set -o) 2>/dev/null` in
+  *posix*) set -o posix ;;
+esac
+
+fi
+
+
+:
+_ASEOF
+}; then
+  CONFIG_SHELL=$as_shell
+	       as_have_required=yes
+	       if { "$as_shell" 2> /dev/null <<\_ASEOF
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
+  emulate sh
+  NULLCMD=:
+  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '${1+"$@"}'='"$@"'
+  setopt NO_GLOB_SUBST
+else
+  case `(set -o) 2>/dev/null` in
+  *posix*) set -o posix ;;
+esac
+
+fi
+
+
+:
+(as_func_return () {
+  (exit $1)
+}
+as_func_success () {
+  as_func_return 0
+}
+as_func_failure () {
+  as_func_return 1
+}
+as_func_ret_success () {
+  return 0
+}
+as_func_ret_failure () {
+  return 1
+}
+
+exitcode=0
+if as_func_success; then
+  :
+else
+  exitcode=1
+  echo as_func_success failed.
+fi
+
+if as_func_failure; then
+  exitcode=1
+  echo as_func_failure succeeded.
+fi
+
+if as_func_ret_success; then
+  :
+else
+  exitcode=1
+  echo as_func_ret_success failed.
+fi
+
+if as_func_ret_failure; then
+  exitcode=1
+  echo as_func_ret_failure succeeded.
+fi
+
+if ( set x; as_func_ret_success y && test x = "$1" ); then
+  :
+else
+  exitcode=1
+  echo positional parameters were not saved.
+fi
+
+test $exitcode = 0) || { (exit 1); exit 1; }
+
+(
+  as_lineno_1=$LINENO
+  as_lineno_2=$LINENO
+  test "x$as_lineno_1" != "x$as_lineno_2" &&
+  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2") || { (exit 1); exit 1; }
+
+_ASEOF
+}; then
+  break
+fi
+
+fi
+
+      done
+
+      if test "x$CONFIG_SHELL" != x; then
+  for as_var in BASH_ENV ENV
+	do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
+	done
+	export CONFIG_SHELL
+	exec "$CONFIG_SHELL" "$as_myself" ${1+"$@"}
+fi
+
+
+    if test $as_have_required = no; then
+  echo This script requires a shell more modern than all the
+      echo shells that I found on your system.  Please install a
+      echo modern shell, or manually run the script under such a
+      echo shell if you do have one.
+      { (exit 1); exit 1; }
+fi
+
+
+fi
+
+fi
+
+
+
+(eval "as_func_return () {
+  (exit \$1)
+}
+as_func_success () {
+  as_func_return 0
+}
+as_func_failure () {
+  as_func_return 1
+}
+as_func_ret_success () {
+  return 0
+}
+as_func_ret_failure () {
+  return 1
+}
+
+exitcode=0
+if as_func_success; then
+  :
+else
+  exitcode=1
+  echo as_func_success failed.
+fi
+
+if as_func_failure; then
+  exitcode=1
+  echo as_func_failure succeeded.
+fi
+
+if as_func_ret_success; then
+  :
+else
+  exitcode=1
+  echo as_func_ret_success failed.
+fi
+
+if as_func_ret_failure; then
+  exitcode=1
+  echo as_func_ret_failure succeeded.
+fi
+
+if ( set x; as_func_ret_success y && test x = \"\$1\" ); then
+  :
+else
+  exitcode=1
+  echo positional parameters were not saved.
+fi
+
+test \$exitcode = 0") || {
+  echo No shell found that supports shell functions.
+  echo Please tell bug-autoconf@gnu.org about your system,
+  echo including any error possibly output before this message.
+  echo This can help us improve future autoconf versions.
+  echo Configuration will now proceed without shell functions.
+}
+
+
+
+  as_lineno_1=$LINENO
+  as_lineno_2=$LINENO
+  test "x$as_lineno_1" != "x$as_lineno_2" &&
+  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2" || {
+
+  # Create $as_me.lineno as a copy of $as_myself, but with $LINENO
+  # uniformly replaced by the line number.  The first 'sed' inserts a
+  # line-number line after each line using $LINENO; the second 'sed'
+  # does the real work.  The second script uses 'N' to pair each
+  # line-number line with the line containing $LINENO, and appends
+  # trailing '-' during substitution so that $LINENO is not a special
+  # case at line end.
+  # (Raja R Harinath suggested sed '=', and Paul Eggert wrote the
+  # scripts with optimization help from Paolo Bonzini.  Blame Lee
+  # E. McMahon (1931-1989) for sed's syntax.  :-)
   sed -n '
     p
     /[$]LINENO/=
@@ -486,12 +473,9 @@ as_cr_alnum=$as_cr_Letters$as_cr_digits
       s/-\n.*//
     ' >$as_me.lineno &&
   chmod +x "$as_me.lineno" ||
-    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2; as_fn_exit 1; }
+    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2
+   { (exit 1); exit 1; }; }
 
-  # If we had to re-execute with $CONFIG_SHELL, we're ensured to have
-  # already done that, so ensure we don't try to do so again and fall
-  # in an infinite loop.  This has already happened in practice.
-  _as_can_reexec=no; export _as_can_reexec
   # Don't try to exec as it changes $[0], causing all sort of problems
   # (the dirname of $[0] is not the place where we might find the
   # original and so on.  Autoconf is especially sensitive to this).
@@ -500,18 +484,29 @@ as_cr_alnum=$as_cr_Letters$as_cr_digits
   exit
 }
 
+
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
+
 ECHO_C= ECHO_N= ECHO_T=
-case `echo -n x` in #(((((
+case `echo -n x` in
 -n*)
-  case `echo 'xy\c'` in
+  case `echo 'x\c'` in
   *c*) ECHO_T='	';;	# ECHO_T is single tab character.
-  xy)  ECHO_C='\c';;
-  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
-       ECHO_T='	';;
+  *)   ECHO_C='\c';;
   esac;;
 *)
   ECHO_N='-n';;
 esac
+if expr a : '\(a\)' >/dev/null 2>&1 &&
+   test "X`expr 00001 : '.*\(...\)'`" = X001; then
+  as_expr=expr
+else
+  as_expr=false
+fi
 
 rm -f conf$$ conf$$.exe conf$$.file
 if test -d conf$$.dir; then
@@ -526,29 +521,49 @@ if (echo >conf$$.file) 2>/dev/null; then
     # ... but there are two gotchas:
     # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
     # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
-    # In both cases, we have to default to `cp -pR'.
+    # In both cases, we have to default to `cp -p'.
     ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
-      as_ln_s='cp -pR'
+      as_ln_s='cp -p'
   elif ln conf$$.file conf$$ 2>/dev/null; then
     as_ln_s=ln
   else
-    as_ln_s='cp -pR'
+    as_ln_s='cp -p'
   fi
 else
-  as_ln_s='cp -pR'
+  as_ln_s='cp -p'
 fi
 rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
 rmdir conf$$.dir 2>/dev/null
 
 if mkdir -p . 2>/dev/null; then
-  as_mkdir_p='mkdir -p "$as_dir"'
+  as_mkdir_p=:
 else
   test -d ./-p && rmdir ./-p
   as_mkdir_p=false
 fi
 
-as_test_x='test -x'
-as_executable_p=as_fn_executable_p
+if test -x / >/dev/null 2>&1; then
+  as_test_x='test -x'
+else
+  if ls -dL / >/dev/null 2>&1; then
+    as_ls_L_option=L
+  else
+    as_ls_L_option=
+  fi
+  as_test_x='
+    eval sh -c '\''
+      if test -d "$1"; then
+	test -d "$1/.";
+      else
+	case $1 in
+	-*)set "./$1";;
+	esac;
+	case `ls -ld'$as_ls_L_option' "$1" 2>/dev/null` in
+	???[sx]*):;;*)false;;esac;fi
+    '\'' sh
+  '
+fi
+as_executable_p=$as_test_x
 
 # Sed expression to map a string onto a valid CPP name.
 as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
@@ -557,11 +572,11 @@ as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
 as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
 
 
-test -n "$DJDIR" || exec 7<&0 </dev/null
-exec 6>&1
+
+exec 7<&0 </dev/null 6>&1
 
 # Name of the host.
-# hostname on some systems (SVR3.2, old GNU/Linux) returns a bogus exit status,
+# hostname on some systems (SVR3.2, Linux) returns a bogus exit status,
 # so uname gets run too.
 ac_hostname=`(hostname || uname -n) 2>/dev/null | sed 1q`
 
@@ -576,6 +591,7 @@ cross_compiling=no
 subdirs=
 MFLAGS=
 MAKEFLAGS=
+SHELL=${CONFIG_SHELL-/bin/sh}
 
 # Identity of this package.
 PACKAGE_NAME='Grid'
@@ -583,7 +599,6 @@ PACKAGE_TARNAME='grid'
 PACKAGE_VERSION='1.0'
 PACKAGE_STRING='Grid 1.0'
 PACKAGE_BUGREPORT='paboyle@ph.ed.ac.uk'
-PACKAGE_URL=''
 
 ac_unique_file="lib/Grid.h"
 # Factoring default headers for most tests.
@@ -644,7 +659,6 @@ CC
 am__fastdepCXX_FALSE
 am__fastdepCXX_TRUE
 CXXDEPMODE
-am__nodep
 AMDEPBACKSLASH
 AMDEP_FALSE
 AMDEP_TRUE
@@ -658,10 +672,6 @@ CPPFLAGS
 LDFLAGS
 CXXFLAGS
 CXX
-AM_BACKSLASH
-AM_DEFAULT_VERBOSITY
-AM_DEFAULT_V
-AM_V
 am__untar
 am__tar
 AMTAR
@@ -715,7 +725,6 @@ bindir
 program_transform_name
 prefix
 exec_prefix
-PACKAGE_URL
 PACKAGE_BUGREPORT
 PACKAGE_STRING
 PACKAGE_VERSION
@@ -726,7 +735,6 @@ SHELL'
 ac_subst_files=''
 ac_user_opts='
 enable_option_checking
-enable_silent_rules
 enable_dependency_tracking
 enable_openmp
 enable_simd
@@ -806,9 +814,8 @@ do
   fi
 
   case $ac_option in
-  *=?*) ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
-  *=)   ac_optarg= ;;
-  *)    ac_optarg=yes ;;
+  *=*)	ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
+  *)	ac_optarg=yes ;;
   esac
 
   # Accept the important Cygnus configure options, so we can diagnose typos.
@@ -853,7 +860,8 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*disable-\(.*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid feature name: $ac_useropt"
+      { $as_echo "$as_me: error: invalid feature name: $ac_useropt" >&2
+   { (exit 1); exit 1; }; }
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -879,7 +887,8 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*enable-\([^=]*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid feature name: $ac_useropt"
+      { $as_echo "$as_me: error: invalid feature name: $ac_useropt" >&2
+   { (exit 1); exit 1; }; }
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1083,7 +1092,8 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*with-\([^=]*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid package name: $ac_useropt"
+      { $as_echo "$as_me: error: invalid package name: $ac_useropt" >&2
+   { (exit 1); exit 1; }; }
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1099,7 +1109,8 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*without-\(.*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid package name: $ac_useropt"
+      { $as_echo "$as_me: error: invalid package name: $ac_useropt" >&2
+   { (exit 1); exit 1; }; }
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1129,17 +1140,17 @@ do
   | --x-librar=* | --x-libra=* | --x-libr=* | --x-lib=* | --x-li=* | --x-l=*)
     x_libraries=$ac_optarg ;;
 
-  -*) as_fn_error $? "unrecognized option: \`$ac_option'
-Try \`$0 --help' for more information"
+  -*) { $as_echo "$as_me: error: unrecognized option: $ac_option
+Try \`$0 --help' for more information." >&2
+   { (exit 1); exit 1; }; }
     ;;
 
   *=*)
     ac_envvar=`expr "x$ac_option" : 'x\([^=]*\)='`
     # Reject names that are not valid shell variable names.
-    case $ac_envvar in #(
-      '' | [0-9]* | *[!_$as_cr_alnum]* )
-      as_fn_error $? "invalid variable name: \`$ac_envvar'" ;;
-    esac
+    expr "x$ac_envvar" : ".*[^_$as_cr_alnum]" >/dev/null &&
+      { $as_echo "$as_me: error: invalid variable name: $ac_envvar" >&2
+   { (exit 1); exit 1; }; }
     eval $ac_envvar=\$ac_optarg
     export $ac_envvar ;;
 
@@ -1148,7 +1159,7 @@ Try \`$0 --help' for more information"
     $as_echo "$as_me: WARNING: you should use --build, --host, --target" >&2
     expr "x$ac_option" : ".*[^-._$as_cr_alnum]" >/dev/null &&
       $as_echo "$as_me: WARNING: invalid host type: $ac_option" >&2
-    : "${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}"
+    : ${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}
     ;;
 
   esac
@@ -1156,13 +1167,15 @@ done
 
 if test -n "$ac_prev"; then
   ac_option=--`echo $ac_prev | sed 's/_/-/g'`
-  as_fn_error $? "missing argument to $ac_option"
+  { $as_echo "$as_me: error: missing argument to $ac_option" >&2
+   { (exit 1); exit 1; }; }
 fi
 
 if test -n "$ac_unrecognized_opts"; then
   case $enable_option_checking in
     no) ;;
-    fatal) as_fn_error $? "unrecognized options: $ac_unrecognized_opts" ;;
+    fatal) { $as_echo "$as_me: error: unrecognized options: $ac_unrecognized_opts" >&2
+   { (exit 1); exit 1; }; } ;;
     *)     $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2 ;;
   esac
 fi
@@ -1185,7 +1198,8 @@ do
     [\\/$]* | ?:[\\/]* )  continue;;
     NONE | '' ) case $ac_var in *prefix ) continue;; esac;;
   esac
-  as_fn_error $? "expected an absolute directory name for --$ac_var: $ac_val"
+  { $as_echo "$as_me: error: expected an absolute directory name for --$ac_var: $ac_val" >&2
+   { (exit 1); exit 1; }; }
 done
 
 # There might be people who depend on the old broken behavior: `$host'
@@ -1199,6 +1213,8 @@ target=$target_alias
 if test "x$host_alias" != x; then
   if test "x$build_alias" = x; then
     cross_compiling=maybe
+    $as_echo "$as_me: WARNING: If you wanted to set the --build type, don't use --host.
+    If a cross compiler is detected then cross compile mode will be used." >&2
   elif test "x$build_alias" != "x$host_alias"; then
     cross_compiling=yes
   fi
@@ -1213,9 +1229,11 @@ test "$silent" = yes && exec 6>/dev/null
 ac_pwd=`pwd` && test -n "$ac_pwd" &&
 ac_ls_di=`ls -di .` &&
 ac_pwd_ls_di=`cd "$ac_pwd" && ls -di .` ||
-  as_fn_error $? "working directory cannot be determined"
+  { $as_echo "$as_me: error: working directory cannot be determined" >&2
+   { (exit 1); exit 1; }; }
 test "X$ac_ls_di" = "X$ac_pwd_ls_di" ||
-  as_fn_error $? "pwd does not report name of working directory"
+  { $as_echo "$as_me: error: pwd does not report name of working directory" >&2
+   { (exit 1); exit 1; }; }
 
 
 # Find the source files, if location was not specified.
@@ -1254,11 +1272,13 @@ else
 fi
 if test ! -r "$srcdir/$ac_unique_file"; then
   test "$ac_srcdir_defaulted" = yes && srcdir="$ac_confdir or .."
-  as_fn_error $? "cannot find sources ($ac_unique_file) in $srcdir"
+  { $as_echo "$as_me: error: cannot find sources ($ac_unique_file) in $srcdir" >&2
+   { (exit 1); exit 1; }; }
 fi
 ac_msg="sources are in $srcdir, but \`cd $srcdir' does not work"
 ac_abs_confdir=`(
-	cd "$srcdir" && test -r "./$ac_unique_file" || as_fn_error $? "$ac_msg"
+	cd "$srcdir" && test -r "./$ac_unique_file" || { $as_echo "$as_me: error: $ac_msg" >&2
+   { (exit 1); exit 1; }; }
 	pwd)`
 # When building in place, set srcdir=.
 if test "$ac_abs_confdir" = "$ac_pwd"; then
@@ -1298,7 +1318,7 @@ Configuration:
       --help=short        display options specific to this package
       --help=recursive    display the short help of all the included packages
   -V, --version           display version information and exit
-  -q, --quiet, --silent   do not print \`checking ...' messages
+  -q, --quiet, --silent   do not print \`checking...' messages
       --cache-file=FILE   cache test results in FILE [disabled]
   -C, --config-cache      alias for \`--cache-file=config.cache'
   -n, --no-create         do not create output files
@@ -1358,12 +1378,8 @@ Optional Features:
   --disable-option-checking  ignore unrecognized --enable/--with options
   --disable-FEATURE       do not include FEATURE (same as --enable-FEATURE=no)
   --enable-FEATURE[=ARG]  include FEATURE [ARG=yes]
-  --enable-silent-rules   less verbose build output (undo: "make V=1")
-  --disable-silent-rules  verbose build output (undo: "make V=0")
-  --enable-dependency-tracking
-                          do not reject slow dependency extractors
-  --disable-dependency-tracking
-                          speeds up one-time build
+  --disable-dependency-tracking  speeds up one-time build
+  --enable-dependency-tracking   do not reject slow dependency extractors
   --disable-openmp        do not use OpenMP
   --enable-simd=SSE|AVX|AVX2|AVX512|MIC
                           Select instructions to be SSE4.0, AVX 1.0, AVX
@@ -1376,7 +1392,7 @@ Some influential environment variables:
   LDFLAGS     linker flags, e.g. -L<lib dir> if you have libraries in a
               nonstandard directory <lib dir>
   LIBS        libraries to pass to the linker, e.g. -l<library>
-  CPPFLAGS    (Objective) C/C++ preprocessor flags, e.g. -I<include dir> if
+  CPPFLAGS    C/C++/Objective C preprocessor flags, e.g. -I<include dir> if
               you have headers in a nonstandard directory <include dir>
   CC          C compiler command
   CFLAGS      C compiler flags
@@ -1449,568 +1465,21 @@ test -n "$ac_init_help" && exit $ac_status
 if $ac_init_version; then
   cat <<\_ACEOF
 Grid configure 1.0
-generated by GNU Autoconf 2.69
+generated by GNU Autoconf 2.63
 
-Copyright (C) 2012 Free Software Foundation, Inc.
+Copyright (C) 1992, 1993, 1994, 1995, 1996, 1998, 1999, 2000, 2001,
+2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
 This configure script is free software; the Free Software Foundation
 gives unlimited permission to copy, distribute and modify it.
 _ACEOF
   exit
 fi
-
-## ------------------------ ##
-## Autoconf initialization. ##
-## ------------------------ ##
-
-# ac_fn_cxx_try_compile LINENO
-# ----------------------------
-# Try to compile conftest.$ac_ext, and return whether this succeeded.
-ac_fn_cxx_try_compile ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  rm -f conftest.$ac_objext
-  if { { ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compile") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_retval=1
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_cxx_try_compile
-
-# ac_fn_c_try_compile LINENO
-# --------------------------
-# Try to compile conftest.$ac_ext, and return whether this succeeded.
-ac_fn_c_try_compile ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  rm -f conftest.$ac_objext
-  if { { ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compile") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_retval=1
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_c_try_compile
-
-# ac_fn_c_try_link LINENO
-# -----------------------
-# Try to link conftest.$ac_ext, and return whether this succeeded.
-ac_fn_c_try_link ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  rm -f conftest.$ac_objext conftest$ac_exeext
-  if { { ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_link") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 test -x conftest$ac_exeext
-       }; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_retval=1
-fi
-  # Delete the IPA/IPO (Inter Procedural Analysis/Optimization) information
-  # created by the PGI compiler (conftest_ipa8_conftest.oo), as it would
-  # interfere with the next link command; also delete a directory that is
-  # left behind by Apple's compiler.  We do this before executing the actions.
-  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_c_try_link
-
-# ac_fn_c_try_cpp LINENO
-# ----------------------
-# Try to preprocess conftest.$ac_ext, and return whether this succeeded.
-ac_fn_c_try_cpp ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  if { { ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } > conftest.i && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-    ac_retval=1
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_c_try_cpp
-
-# ac_fn_c_check_header_mongrel LINENO HEADER VAR INCLUDES
-# -------------------------------------------------------
-# Tests whether HEADER exists, giving a warning if it cannot be compiled using
-# the include files in INCLUDES and setting the cache variable VAR
-# accordingly.
-ac_fn_c_check_header_mongrel ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  if eval \${$3+:} false; then :
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 usability" >&5
-$as_echo_n "checking $2 usability... " >&6; }
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-#include <$2>
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  ac_header_compiler=yes
-else
-  ac_header_compiler=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 presence" >&5
-$as_echo_n "checking $2 presence... " >&6; }
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <$2>
-_ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
-  ac_header_preproc=yes
-else
-  ac_header_preproc=no
-fi
-rm -f conftest.err conftest.i conftest.$ac_ext
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in #((
-  yes:no: )
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $2: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $2:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $2: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
-( $as_echo "## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##"
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  eval "$3=\$ac_header_compiler"
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_check_header_mongrel
-
-# ac_fn_c_try_run LINENO
-# ----------------------
-# Try to link conftest.$ac_ext, and return whether this succeeded. Assumes
-# that executables *can* be run.
-ac_fn_c_try_run ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  if { { ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_link") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && { ac_try='./conftest$ac_exeext'
-  { { case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; }; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: program exited with status $ac_status" >&5
-       $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-       ac_retval=$ac_status
-fi
-  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_c_try_run
-
-# ac_fn_c_check_header_compile LINENO HEADER VAR INCLUDES
-# -------------------------------------------------------
-# Tests whether HEADER exists and can be compiled using the include files in
-# INCLUDES, setting the cache variable VAR accordingly.
-ac_fn_c_check_header_compile ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-#include <$2>
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  eval "$3=yes"
-else
-  eval "$3=no"
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_check_header_compile
-
-# ac_fn_c_check_decl LINENO SYMBOL VAR INCLUDES
-# ---------------------------------------------
-# Tests whether SYMBOL is declared in INCLUDES, setting cache variable VAR
-# accordingly.
-ac_fn_c_check_decl ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  as_decl_name=`echo $2|sed 's/ *(.*//'`
-  as_decl_use=`echo $2|sed -e 's/(/((/' -e 's/)/) 0&/' -e 's/,/) 0& (/g'`
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $as_decl_name is declared" >&5
-$as_echo_n "checking whether $as_decl_name is declared... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-int
-main ()
-{
-#ifndef $as_decl_name
-#ifdef __cplusplus
-  (void) $as_decl_use;
-#else
-  (void) $as_decl_name;
-#endif
-#endif
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  eval "$3=yes"
-else
-  eval "$3=no"
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_check_decl
-
-# ac_fn_c_check_type LINENO TYPE VAR INCLUDES
-# -------------------------------------------
-# Tests whether TYPE exists after having included INCLUDES, setting cache
-# variable VAR accordingly.
-ac_fn_c_check_type ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  eval "$3=no"
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-int
-main ()
-{
-if (sizeof ($2))
-	 return 0;
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-int
-main ()
-{
-if (sizeof (($2)))
-	    return 0;
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-
-else
-  eval "$3=yes"
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_check_type
-
-# ac_fn_c_find_uintX_t LINENO BITS VAR
-# ------------------------------------
-# Finds an unsigned integer type with width BITS, setting cache variable VAR
-# accordingly.
-ac_fn_c_find_uintX_t ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for uint$2_t" >&5
-$as_echo_n "checking for uint$2_t... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  eval "$3=no"
-     # Order is important - never check a type that is potentially smaller
-     # than half of the expected target width.
-     for ac_type in uint$2_t 'unsigned int' 'unsigned long int' \
-	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
-       cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-static int test_array [1 - 2 * !((($ac_type) -1 >> ($2 / 2 - 1)) >> ($2 / 2 - 1) == 3)];
-test_array [0] = 0;
-return test_array [0];
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  case $ac_type in #(
-  uint$2_t) :
-    eval "$3=yes" ;; #(
-  *) :
-    eval "$3=\$ac_type" ;;
-esac
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-       if eval test \"x\$"$3"\" = x"no"; then :
-
-else
-  break
-fi
-     done
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_find_uintX_t
-
-# ac_fn_c_check_func LINENO FUNC VAR
-# ----------------------------------
-# Tests whether FUNC exists, setting the cache variable VAR accordingly
-ac_fn_c_check_func ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-/* Define $2 to an innocuous variant, in case <limits.h> declares $2.
-   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
-#define $2 innocuous_$2
-
-/* System header to define __stub macros and hopefully few prototypes,
-    which can conflict with char $2 (); below.
-    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
-    <limits.h> exists even on freestanding compilers.  */
-
-#ifdef __STDC__
-# include <limits.h>
-#else
-# include <assert.h>
-#endif
-
-#undef $2
-
-/* Override any GCC internal prototype to avoid an error.
-   Use char because int might match the return type of a GCC
-   builtin and then its argument prototype would still apply.  */
-#ifdef __cplusplus
-extern "C"
-#endif
-char $2 ();
-/* The GNU C library defines this for functions which it implements
-    to always fail with ENOSYS.  Some functions are actually named
-    something starting with __ and the normal name is an alias.  */
-#if defined __stub_$2 || defined __stub___$2
-choke me
-#endif
-
-int
-main ()
-{
-return $2 ();
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_link "$LINENO"; then :
-  eval "$3=yes"
-else
-  eval "$3=no"
-fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_check_func
 cat >config.log <<_ACEOF
 This file contains any messages produced by compilers while
 running configure, to aid debugging if configure makes a mistake.
 
 It was created by Grid $as_me 1.0, which was
-generated by GNU Autoconf 2.69.  Invocation command line was
+generated by GNU Autoconf 2.63.  Invocation command line was
 
   $ $0 $@
 
@@ -2046,8 +1515,8 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    $as_echo "PATH: $as_dir"
-  done
+  $as_echo "PATH: $as_dir"
+done
 IFS=$as_save_IFS
 
 } >&5
@@ -2084,9 +1553,9 @@ do
       ac_arg=`$as_echo "$ac_arg" | sed "s/'/'\\\\\\\\''/g"` ;;
     esac
     case $ac_pass in
-    1) as_fn_append ac_configure_args0 " '$ac_arg'" ;;
+    1) ac_configure_args0="$ac_configure_args0 '$ac_arg'" ;;
     2)
-      as_fn_append ac_configure_args1 " '$ac_arg'"
+      ac_configure_args1="$ac_configure_args1 '$ac_arg'"
       if test $ac_must_keep_next = true; then
 	ac_must_keep_next=false # Got value, back to normal.
       else
@@ -2102,13 +1571,13 @@ do
 	  -* ) ac_must_keep_next=true ;;
 	esac
       fi
-      as_fn_append ac_configure_args " '$ac_arg'"
+      ac_configure_args="$ac_configure_args '$ac_arg'"
       ;;
     esac
   done
 done
-{ ac_configure_args0=; unset ac_configure_args0;}
-{ ac_configure_args1=; unset ac_configure_args1;}
+$as_unset ac_configure_args0 || test "${ac_configure_args0+set}" != set || { ac_configure_args0=; export ac_configure_args0; }
+$as_unset ac_configure_args1 || test "${ac_configure_args1+set}" != set || { ac_configure_args1=; export ac_configure_args1; }
 
 # When interrupted or exit'd, cleanup temporary files, and complete
 # config.log.  We remove comments because anyway the quotes in there
@@ -2120,9 +1589,11 @@ trap 'exit_status=$?
   {
     echo
 
-    $as_echo "## ---------------- ##
+    cat <<\_ASBOX
+## ---------------- ##
 ## Cache variables. ##
-## ---------------- ##"
+## ---------------- ##
+_ASBOX
     echo
     # The following way of writing the cache mishandles newlines in values,
 (
@@ -2131,13 +1602,13 @@ trap 'exit_status=$?
     case $ac_val in #(
     *${as_nl}*)
       case $ac_var in #(
-      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
+      *_cv_*) { $as_echo "$as_me:$LINENO: WARNING: cache variable $ac_var contains a newline" >&5
 $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
       esac
       case $ac_var in #(
       _ | IFS | as_nl) ;; #(
       BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
-      *) { eval $ac_var=; unset $ac_var;} ;;
+      *) $as_unset $ac_var ;;
       esac ;;
     esac
   done
@@ -2156,9 +1627,11 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
 )
     echo
 
-    $as_echo "## ----------------- ##
+    cat <<\_ASBOX
+## ----------------- ##
 ## Output variables. ##
-## ----------------- ##"
+## ----------------- ##
+_ASBOX
     echo
     for ac_var in $ac_subst_vars
     do
@@ -2171,9 +1644,11 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
     echo
 
     if test -n "$ac_subst_files"; then
-      $as_echo "## ------------------- ##
+      cat <<\_ASBOX
+## ------------------- ##
 ## File substitutions. ##
-## ------------------- ##"
+## ------------------- ##
+_ASBOX
       echo
       for ac_var in $ac_subst_files
       do
@@ -2187,9 +1662,11 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
     fi
 
     if test -s confdefs.h; then
-      $as_echo "## ----------- ##
+      cat <<\_ASBOX
+## ----------- ##
 ## confdefs.h. ##
-## ----------- ##"
+## ----------- ##
+_ASBOX
       echo
       cat confdefs.h
       echo
@@ -2203,39 +1680,37 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
     exit $exit_status
 ' 0
 for ac_signal in 1 2 13 15; do
-  trap 'ac_signal='$ac_signal'; as_fn_exit 1' $ac_signal
+  trap 'ac_signal='$ac_signal'; { (exit 1); exit 1; }' $ac_signal
 done
 ac_signal=0
 
 # confdefs.h avoids OS command line length limits that DEFS can exceed.
 rm -f -r conftest* confdefs.h
 
-$as_echo "/* confdefs.h */" > confdefs.h
-
 # Predefined preprocessor variables.
 
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_NAME "$PACKAGE_NAME"
 _ACEOF
 
+
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_TARNAME "$PACKAGE_TARNAME"
 _ACEOF
 
+
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_VERSION "$PACKAGE_VERSION"
 _ACEOF
 
+
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_STRING "$PACKAGE_STRING"
 _ACEOF
 
-cat >>confdefs.h <<_ACEOF
-#define PACKAGE_BUGREPORT "$PACKAGE_BUGREPORT"
-_ACEOF
 
 cat >>confdefs.h <<_ACEOF
-#define PACKAGE_URL "$PACKAGE_URL"
+#define PACKAGE_BUGREPORT "$PACKAGE_BUGREPORT"
 _ACEOF
 
 
@@ -2244,12 +1719,7 @@ _ACEOF
 ac_site_file1=NONE
 ac_site_file2=NONE
 if test -n "$CONFIG_SITE"; then
-  # We do not want a PATH search for config.site.
-  case $CONFIG_SITE in #((
-    -*)  ac_site_file1=./$CONFIG_SITE;;
-    */*) ac_site_file1=$CONFIG_SITE;;
-    *)   ac_site_file1=./$CONFIG_SITE;;
-  esac
+  ac_site_file1=$CONFIG_SITE
 elif test "x$prefix" != xNONE; then
   ac_site_file1=$prefix/share/config.site
   ac_site_file2=$prefix/etc/config.site
@@ -2260,23 +1730,19 @@ fi
 for ac_site_file in "$ac_site_file1" "$ac_site_file2"
 do
   test "x$ac_site_file" = xNONE && continue
-  if test /dev/null != "$ac_site_file" && test -r "$ac_site_file"; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: loading site script $ac_site_file" >&5
+  if test -r "$ac_site_file"; then
+    { $as_echo "$as_me:$LINENO: loading site script $ac_site_file" >&5
 $as_echo "$as_me: loading site script $ac_site_file" >&6;}
     sed 's/^/| /' "$ac_site_file" >&5
-    . "$ac_site_file" \
-      || { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "failed to load site script $ac_site_file
-See \`config.log' for more details" "$LINENO" 5; }
+    . "$ac_site_file"
   fi
 done
 
 if test -r "$cache_file"; then
-  # Some versions of bash will fail to source /dev/null (special files
-  # actually), so we avoid doing that.  DJGPP emulates it as a regular file.
-  if test /dev/null != "$cache_file" && test -f "$cache_file"; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: loading cache $cache_file" >&5
+  # Some versions of bash will fail to source /dev/null (special
+  # files actually), so we avoid doing that.
+  if test -f "$cache_file"; then
+    { $as_echo "$as_me:$LINENO: loading cache $cache_file" >&5
 $as_echo "$as_me: loading cache $cache_file" >&6;}
     case $cache_file in
       [\\/]* | ?:[\\/]* ) . "$cache_file";;
@@ -2284,7 +1750,7 @@ $as_echo "$as_me: loading cache $cache_file" >&6;}
     esac
   fi
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: creating cache $cache_file" >&5
+  { $as_echo "$as_me:$LINENO: creating cache $cache_file" >&5
 $as_echo "$as_me: creating cache $cache_file" >&6;}
   >$cache_file
 fi
@@ -2299,11 +1765,11 @@ for ac_var in $ac_precious_vars; do
   eval ac_new_val=\$ac_env_${ac_var}_value
   case $ac_old_set,$ac_new_set in
     set,)
-      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
+      { $as_echo "$as_me:$LINENO: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
 $as_echo "$as_me: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&2;}
       ac_cache_corrupted=: ;;
     ,set)
-      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was not set in the previous run" >&5
+      { $as_echo "$as_me:$LINENO: error: \`$ac_var' was not set in the previous run" >&5
 $as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
       ac_cache_corrupted=: ;;
     ,);;
@@ -2313,17 +1779,17 @@ $as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
 	ac_old_val_w=`echo x $ac_old_val`
 	ac_new_val_w=`echo x $ac_new_val`
 	if test "$ac_old_val_w" != "$ac_new_val_w"; then
-	  { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' has changed since the previous run:" >&5
+	  { $as_echo "$as_me:$LINENO: error: \`$ac_var' has changed since the previous run:" >&5
 $as_echo "$as_me: error: \`$ac_var' has changed since the previous run:" >&2;}
 	  ac_cache_corrupted=:
 	else
-	  { $as_echo "$as_me:${as_lineno-$LINENO}: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
+	  { $as_echo "$as_me:$LINENO: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
 $as_echo "$as_me: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&2;}
 	  eval $ac_var=\$ac_old_val
 	fi
-	{ $as_echo "$as_me:${as_lineno-$LINENO}:   former value:  \`$ac_old_val'" >&5
+	{ $as_echo "$as_me:$LINENO:   former value:  \`$ac_old_val'" >&5
 $as_echo "$as_me:   former value:  \`$ac_old_val'" >&2;}
-	{ $as_echo "$as_me:${as_lineno-$LINENO}:   current value: \`$ac_new_val'" >&5
+	{ $as_echo "$as_me:$LINENO:   current value: \`$ac_new_val'" >&5
 $as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
       fi;;
   esac
@@ -2335,20 +1801,43 @@ $as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
     esac
     case " $ac_configure_args " in
       *" '$ac_arg' "*) ;; # Avoid dups.  Use of quotes ensures accuracy.
-      *) as_fn_append ac_configure_args " '$ac_arg'" ;;
+      *) ac_configure_args="$ac_configure_args '$ac_arg'" ;;
     esac
   fi
 done
 if $ac_cache_corrupted; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+  { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-  { $as_echo "$as_me:${as_lineno-$LINENO}: error: changes in the environment can compromise the build" >&5
+  { $as_echo "$as_me:$LINENO: error: changes in the environment can compromise the build" >&5
 $as_echo "$as_me: error: changes in the environment can compromise the build" >&2;}
-  as_fn_error $? "run \`make distclean' and/or \`rm $cache_file' and start over" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: run \`make distclean' and/or \`rm $cache_file' and start over" >&5
+$as_echo "$as_me: error: run \`make distclean' and/or \`rm $cache_file' and start over" >&2;}
+   { (exit 1); exit 1; }; }
 fi
-## -------------------- ##
-## Main body of script. ##
-## -------------------- ##
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
 
 ac_ext=c
 ac_cpp='$CPP $CPPFLAGS'
@@ -2357,7 +1846,7 @@ ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
 
-am__api_version='1.15'
+am__api_version='1.11'
 
 ac_aux_dir=
 for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
@@ -2376,7 +1865,9 @@ for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
   fi
 done
 if test -z "$ac_aux_dir"; then
-  as_fn_error $? "cannot find install-sh, install.sh, or shtool in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: cannot find install-sh or install.sh in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" >&5
+$as_echo "$as_me: error: cannot find install-sh or install.sh in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" >&2;}
+   { (exit 1); exit 1; }; }
 fi
 
 # These three variables are undocumented and unsupported,
@@ -2402,10 +1893,10 @@ ac_configure="$SHELL $ac_aux_dir/configure"  # Please don't use this var.
 # OS/2's system install, which has a completely different semantic
 # ./install, which can be erroneously created by make from ./install.sh.
 # Reject install programs that cannot install multiple files.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a BSD-compatible install" >&5
+{ $as_echo "$as_me:$LINENO: checking for a BSD-compatible install" >&5
 $as_echo_n "checking for a BSD-compatible install... " >&6; }
 if test -z "$INSTALL"; then
-if ${ac_cv_path_install+:} false; then :
+if test "${ac_cv_path_install+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
@@ -2413,11 +1904,11 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    # Account for people who put trailing slashes in PATH elements.
-case $as_dir/ in #((
-  ./ | .// | /[cC]/* | \
+  # Account for people who put trailing slashes in PATH elements.
+case $as_dir/ in
+  ./ | .// | /cC/* | \
   /etc/* | /usr/sbin/* | /usr/etc/* | /sbin/* | /usr/afsws/bin/* | \
-  ?:[\\/]os2[\\/]install[\\/]* | ?:[\\/]OS2[\\/]INSTALL[\\/]* | \
+  ?:\\/os2\\/install\\/* | ?:\\/OS2\\/INSTALL\\/* | \
   /usr/ucb/* ) ;;
   *)
     # OSF1 and SCO ODT 3.0 have their own names for install.
@@ -2425,7 +1916,7 @@ case $as_dir/ in #((
     # by default.
     for ac_prog in ginstall scoinst install; do
       for ac_exec_ext in '' $ac_executable_extensions; do
-	if as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext"; then
+	if { test -f "$as_dir/$ac_prog$ac_exec_ext" && $as_test_x "$as_dir/$ac_prog$ac_exec_ext"; }; then
 	  if test $ac_prog = install &&
 	    grep dspmsg "$as_dir/$ac_prog$ac_exec_ext" >/dev/null 2>&1; then
 	    # AIX install.  It has an incompatible calling convention.
@@ -2454,7 +1945,7 @@ case $as_dir/ in #((
     ;;
 esac
 
-  done
+done
 IFS=$as_save_IFS
 
 rm -rf conftest.one conftest.two conftest.dir
@@ -2470,7 +1961,7 @@ fi
     INSTALL=$ac_install_sh
   fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $INSTALL" >&5
+{ $as_echo "$as_me:$LINENO: result: $INSTALL" >&5
 $as_echo "$INSTALL" >&6; }
 
 # Use test -z because SunOS4 sh mishandles braces in ${var-val}.
@@ -2481,73 +1972,68 @@ test -z "$INSTALL_SCRIPT" && INSTALL_SCRIPT='${INSTALL}'
 
 test -z "$INSTALL_DATA" && INSTALL_DATA='${INSTALL} -m 644'
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether build environment is sane" >&5
+{ $as_echo "$as_me:$LINENO: checking whether build environment is sane" >&5
 $as_echo_n "checking whether build environment is sane... " >&6; }
+# Just in case
+sleep 1
+echo timestamp > conftest.file
 # Reject unsafe characters in $srcdir or the absolute working directory
 # name.  Accept space and tab only in the latter.
 am_lf='
 '
 case `pwd` in
   *[\\\"\#\$\&\'\`$am_lf]*)
-    as_fn_error $? "unsafe absolute working directory name" "$LINENO" 5;;
+    { { $as_echo "$as_me:$LINENO: error: unsafe absolute working directory name" >&5
+$as_echo "$as_me: error: unsafe absolute working directory name" >&2;}
+   { (exit 1); exit 1; }; };;
 esac
 case $srcdir in
   *[\\\"\#\$\&\'\`$am_lf\ \	]*)
-    as_fn_error $? "unsafe srcdir value: '$srcdir'" "$LINENO" 5;;
+    { { $as_echo "$as_me:$LINENO: error: unsafe srcdir value: \`$srcdir'" >&5
+$as_echo "$as_me: error: unsafe srcdir value: \`$srcdir'" >&2;}
+   { (exit 1); exit 1; }; };;
 esac
 
-# Do 'set' in a subshell so we don't clobber the current shell's
+# Do `set' in a subshell so we don't clobber the current shell's
 # arguments.  Must try -L first in case configure is actually a
 # symlink; some systems play weird games with the mod time of symlinks
 # (eg FreeBSD returns the mod time of the symlink's containing
 # directory).
 if (
-   am_has_slept=no
-   for am_try in 1 2; do
-     echo "timestamp, slept: $am_has_slept" > conftest.file
-     set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
-     if test "$*" = "X"; then
-	# -L didn't work.
-	set X `ls -t "$srcdir/configure" conftest.file`
-     fi
-     if test "$*" != "X $srcdir/configure conftest.file" \
-	&& test "$*" != "X conftest.file $srcdir/configure"; then
+   set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
+   if test "$*" = "X"; then
+      # -L didn't work.
+      set X `ls -t "$srcdir/configure" conftest.file`
+   fi
+   rm -f conftest.file
+   if test "$*" != "X $srcdir/configure conftest.file" \
+      && test "$*" != "X conftest.file $srcdir/configure"; then
+
+      # If neither matched, then we have a broken ls.  This can happen
+      # if, for instance, CONFIG_SHELL is bash and it inherits a
+      # broken ls alias from the environment.  This has actually
+      # happened.  Such a system could not be considered "sane".
+      { { $as_echo "$as_me:$LINENO: error: ls -t appears to fail.  Make sure there is not a broken
+alias in your environment" >&5
+$as_echo "$as_me: error: ls -t appears to fail.  Make sure there is not a broken
+alias in your environment" >&2;}
+   { (exit 1); exit 1; }; }
+   fi
 
-	# If neither matched, then we have a broken ls.  This can happen
-	# if, for instance, CONFIG_SHELL is bash and it inherits a
-	# broken ls alias from the environment.  This has actually
-	# happened.  Such a system could not be considered "sane".
-	as_fn_error $? "ls -t appears to fail.  Make sure there is not a broken
-  alias in your environment" "$LINENO" 5
-     fi
-     if test "$2" = conftest.file || test $am_try -eq 2; then
-       break
-     fi
-     # Just in case.
-     sleep 1
-     am_has_slept=yes
-   done
    test "$2" = conftest.file
    )
 then
    # Ok.
    :
 else
-   as_fn_error $? "newly created file is older than distributed files!
-Check your system clock" "$LINENO" 5
+   { { $as_echo "$as_me:$LINENO: error: newly created file is older than distributed files!
+Check your system clock" >&5
+$as_echo "$as_me: error: newly created file is older than distributed files!
+Check your system clock" >&2;}
+   { (exit 1); exit 1; }; }
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+{ $as_echo "$as_me:$LINENO: result: yes" >&5
 $as_echo "yes" >&6; }
-# If we didn't sleep, we still need to ensure time stamps of config.status and
-# generated files are strictly newer.
-am_sleep_pid=
-if grep 'slept: no' conftest.file >/dev/null 2>&1; then
-  ( sleep 1 ) &
-  am_sleep_pid=$!
-fi
-
-rm -f conftest.file
-
 test "$program_prefix" != NONE &&
   program_transform_name="s&^&$program_prefix&;$program_transform_name"
 # Use a double $ so make ignores it.
@@ -2558,8 +2044,8 @@ test "$program_suffix" != NONE &&
 ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
 program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
 
-# Expand $ac_aux_dir to an absolute path.
-am_aux_dir=`cd "$ac_aux_dir" && pwd`
+# expand $ac_aux_dir to an absolute path
+am_aux_dir=`cd $ac_aux_dir && pwd`
 
 if test x"${MISSING+set}" != xset; then
   case $am_aux_dir in
@@ -2570,15 +2056,15 @@ if test x"${MISSING+set}" != xset; then
   esac
 fi
 # Use eval to expand $SHELL
-if eval "$MISSING --is-lightweight"; then
-  am_missing_run="$MISSING "
+if eval "$MISSING --run true"; then
+  am_missing_run="$MISSING --run "
 else
   am_missing_run=
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: 'missing' script is too old or missing" >&5
-$as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
+  { $as_echo "$as_me:$LINENO: WARNING: \`missing' script is too old or missing" >&5
+$as_echo "$as_me: WARNING: \`missing' script is too old or missing" >&2;}
 fi
 
-if test x"${install_sh+set}" != xset; then
+if test x"${install_sh}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -2587,17 +2073,17 @@ if test x"${install_sh+set}" != xset; then
   esac
 fi
 
-# Installed binaries are usually stripped using 'strip' when the user
-# run "make install-strip".  However 'strip' might not be the right
+# Installed binaries are usually stripped using `strip' when the user
+# run `make install-strip'.  However `strip' might not be the right
 # tool to use in cross-compilation environments, therefore Automake
-# will honor the 'STRIP' environment variable to overrule this program.
+# will honor the `STRIP' environment variable to overrule this program.
 if test "$cross_compiling" != no; then
   if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}strip", so it can be a program name with args.
 set dummy ${ac_tool_prefix}strip; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_STRIP+:} false; then :
+if test "${ac_cv_prog_STRIP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$STRIP"; then
@@ -2608,24 +2094,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_STRIP="${ac_tool_prefix}strip"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 STRIP=$ac_cv_prog_STRIP
 if test -n "$STRIP"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $STRIP" >&5
+  { $as_echo "$as_me:$LINENO: result: $STRIP" >&5
 $as_echo "$STRIP" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2635,9 +2121,9 @@ if test -z "$ac_cv_prog_STRIP"; then
   ac_ct_STRIP=$STRIP
   # Extract the first word of "strip", so it can be a program name with args.
 set dummy strip; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_STRIP+:} false; then :
+if test "${ac_cv_prog_ac_ct_STRIP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_STRIP"; then
@@ -2648,24 +2134,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_ac_ct_STRIP="strip"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_STRIP=$ac_cv_prog_ac_ct_STRIP
 if test -n "$ac_ct_STRIP"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_STRIP" >&5
+  { $as_echo "$as_me:$LINENO: result: $ac_ct_STRIP" >&5
 $as_echo "$ac_ct_STRIP" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2674,7 +2160,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -2687,10 +2173,10 @@ fi
 fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a thread-safe mkdir -p" >&5
+{ $as_echo "$as_me:$LINENO: checking for a thread-safe mkdir -p" >&5
 $as_echo_n "checking for a thread-safe mkdir -p... " >&6; }
 if test -z "$MKDIR_P"; then
-  if ${ac_cv_path_mkdir+:} false; then :
+  if test "${ac_cv_path_mkdir+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
@@ -2698,9 +2184,9 @@ for as_dir in $PATH$PATH_SEPARATOR/opt/sfw/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_prog in mkdir gmkdir; do
+  for ac_prog in mkdir gmkdir; do
 	 for ac_exec_ext in '' $ac_executable_extensions; do
-	   as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext" || continue
+	   { test -f "$as_dir/$ac_prog$ac_exec_ext" && $as_test_x "$as_dir/$ac_prog$ac_exec_ext"; } || continue
 	   case `"$as_dir/$ac_prog$ac_exec_ext" --version 2>&1` in #(
 	     'mkdir (GNU coreutils) '* | \
 	     'mkdir (coreutils) '* | \
@@ -2710,12 +2196,11 @@ do
 	   esac
 	 done
        done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 
-  test -d ./--version && rmdir ./--version
   if test "${ac_cv_path_mkdir+set}" = set; then
     MKDIR_P="$ac_cv_path_mkdir -p"
   else
@@ -2723,19 +2208,26 @@ fi
     # value for MKDIR_P within a source directory, because that will
     # break other packages using the cache if that directory is
     # removed, or if the value is a relative name.
+    test -d ./--version && rmdir ./--version
     MKDIR_P="$ac_install_sh -d"
   fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $MKDIR_P" >&5
+{ $as_echo "$as_me:$LINENO: result: $MKDIR_P" >&5
 $as_echo "$MKDIR_P" >&6; }
 
+mkdir_p="$MKDIR_P"
+case $mkdir_p in
+  [\\/$]* | ?:[\\/]*) ;;
+  */*) mkdir_p="\$(top_builddir)/$mkdir_p" ;;
+esac
+
 for ac_prog in gawk mawk nawk awk
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_AWK+:} false; then :
+if test "${ac_cv_prog_AWK+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$AWK"; then
@@ -2746,24 +2238,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_AWK="$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 AWK=$ac_cv_prog_AWK
 if test -n "$AWK"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $AWK" >&5
+  { $as_echo "$as_me:$LINENO: result: $AWK" >&5
 $as_echo "$AWK" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2771,11 +2263,11 @@ fi
   test -n "$AWK" && break
 done
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether ${MAKE-make} sets \$(MAKE)" >&5
+{ $as_echo "$as_me:$LINENO: checking whether ${MAKE-make} sets \$(MAKE)" >&5
 $as_echo_n "checking whether ${MAKE-make} sets \$(MAKE)... " >&6; }
 set x ${MAKE-make}
 ac_make=`$as_echo "$2" | sed 's/+/p/g; s/[^a-zA-Z0-9_]/_/g'`
-if eval \${ac_cv_prog_make_${ac_make}_set+:} false; then :
+if { as_var=ac_cv_prog_make_${ac_make}_set; eval "test \"\${$as_var+set}\" = set"; }; then
   $as_echo_n "(cached) " >&6
 else
   cat >conftest.make <<\_ACEOF
@@ -2783,7 +2275,7 @@ SHELL = /bin/sh
 all:
 	@echo '@@@%%%=$(MAKE)=@@@%%%'
 _ACEOF
-# GNU make sometimes prints "make[1]: Entering ...", which would confuse us.
+# GNU make sometimes prints "make[1]: Entering...", which would confuse us.
 case `${MAKE-make} -f conftest.make 2>/dev/null` in
   *@@@%%%=?*=@@@%%%*)
     eval ac_cv_prog_make_${ac_make}_set=yes;;
@@ -2793,11 +2285,11 @@ esac
 rm -f conftest.make
 fi
 if eval test \$ac_cv_prog_make_${ac_make}_set = yes; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+  { $as_echo "$as_me:$LINENO: result: yes" >&5
 $as_echo "yes" >&6; }
   SET_MAKE=
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
   SET_MAKE="MAKE=${MAKE-make}"
 fi
@@ -2811,52 +2303,15 @@ else
 fi
 rmdir .tst 2>/dev/null
 
-# Check whether --enable-silent-rules was given.
-if test "${enable_silent_rules+set}" = set; then :
-  enableval=$enable_silent_rules;
-fi
-
-case $enable_silent_rules in # (((
-  yes) AM_DEFAULT_VERBOSITY=0;;
-   no) AM_DEFAULT_VERBOSITY=1;;
-    *) AM_DEFAULT_VERBOSITY=1;;
-esac
-am_make=${MAKE-make}
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $am_make supports nested variables" >&5
-$as_echo_n "checking whether $am_make supports nested variables... " >&6; }
-if ${am_cv_make_support_nested_variables+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if $as_echo 'TRUE=$(BAR$(V))
-BAR0=false
-BAR1=true
-V=1
-am__doit:
-	@$(TRUE)
-.PHONY: am__doit' | $am_make -f - >/dev/null 2>&1; then
-  am_cv_make_support_nested_variables=yes
-else
-  am_cv_make_support_nested_variables=no
-fi
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_make_support_nested_variables" >&5
-$as_echo "$am_cv_make_support_nested_variables" >&6; }
-if test $am_cv_make_support_nested_variables = yes; then
-    AM_V='$(V)'
-  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
-else
-  AM_V=$AM_DEFAULT_VERBOSITY
-  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
-fi
-AM_BACKSLASH='\'
-
 if test "`cd $srcdir && pwd`" != "`pwd`"; then
   # Use -I$(srcdir) only when $(srcdir) != ., so that make's output
   # is not polluted with repeated "-I."
   am__isrc=' -I$(srcdir)'
   # test to see if srcdir already configured
   if test -f $srcdir/config.status; then
-    as_fn_error $? "source directory already configured; run \"make distclean\" there first" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: source directory already configured; run \"make distclean\" there first" >&5
+$as_echo "$as_me: error: source directory already configured; run \"make distclean\" there first" >&2;}
+   { (exit 1); exit 1; }; }
   fi
 fi
 
@@ -2900,72 +2355,19 @@ AUTOHEADER=${AUTOHEADER-"${am_missing_run}autoheader"}
 
 MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 
-# For better backward compatibility.  To be removed once Automake 1.9.x
-# dies out for good.  For more background, see:
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
-mkdir_p='$(MKDIR_P)'
+# We need awk for the "check" target.  The system "awk" is bad on
+# some platforms.
+# Always define AMTAR for backward compatibility.
 
-# We need awk for the "check" target (and possibly the TAP driver).  The
-# system "awk" is bad on some platforms.
-# Always define AMTAR for backward compatibility.  Yes, it's still used
-# in the wild :-(  We should find a proper way to deprecate it ...
-AMTAR='$${TAR-tar}'
+AMTAR=${AMTAR-"${am_missing_run}tar"}
 
-
-# We'll loop over all known methods to create a tar archive until one works.
-_am_tools='gnutar  pax cpio none'
-
-am__tar='$${TAR-tar} chof - "$$tardir"' am__untar='$${TAR-tar} xf -'
+am__tar='${AMTAR} chof - "$$tardir"'; am__untar='${AMTAR} xf -'
 
 
 
 
 
 
-# POSIX will say in a future version that running "rm -f" with no argument
-# is OK; and we want to be able to make that assumption in our Makefile
-# recipes.  So use an aggressive probe to check that the usage we want is
-# actually supported "in the wild" to an acceptable degree.
-# See automake bug#10828.
-# To make any issue more visible, cause the running configure to be aborted
-# by default if the 'rm' program in use doesn't match our expectations; the
-# user can still override this though.
-if rm -f && rm -fr && rm -rf; then : OK; else
-  cat >&2 <<'END'
-Oops!
-
-Your 'rm' program seems unable to run without file operands specified
-on the command line, even when the '-f' option is present.  This is contrary
-to the behaviour of most rm programs out there, and not conforming with
-the upcoming POSIX standard: <http://austingroupbugs.net/view.php?id=542>
-
-Please tell bug-automake@gnu.org about your system, including the value
-of your $PATH and any error possibly output before this message.  This
-can help us improve future automake versions.
-
-END
-  if test x"$ACCEPT_INFERIOR_RM_PROGRAM" = x"yes"; then
-    echo 'Configuration will proceed anyway, since you have set the' >&2
-    echo 'ACCEPT_INFERIOR_RM_PROGRAM variable to "yes"' >&2
-    echo >&2
-  else
-    cat >&2 <<'END'
-Aborting the configuration process, to ensure you take notice of the issue.
-
-You can download and install GNU coreutils to get an 'rm' implementation
-that behaves properly: <http://www.gnu.org/software/coreutils/>.
-
-If you want to complete the configuration process using your problematic
-'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
-to "yes", and re-run configure.
-
-END
-    as_fn_error $? "Your 'rm' program is bad, sorry." "$LINENO" 5
-  fi
-fi
-
-
 
 ac_config_headers="$ac_config_headers lib/Grid_config.h"
 
@@ -2985,9 +2387,9 @@ if test -z "$CXX"; then
   do
     # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
 set dummy $ac_tool_prefix$ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CXX+:} false; then :
+if test "${ac_cv_prog_CXX+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CXX"; then
@@ -2998,24 +2400,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_CXX="$ac_tool_prefix$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 CXX=$ac_cv_prog_CXX
 if test -n "$CXX"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CXX" >&5
+  { $as_echo "$as_me:$LINENO: result: $CXX" >&5
 $as_echo "$CXX" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3029,9 +2431,9 @@ if test -z "$CXX"; then
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_CXX+:} false; then :
+if test "${ac_cv_prog_ac_ct_CXX+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CXX"; then
@@ -3042,24 +2444,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_ac_ct_CXX="$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CXX=$ac_cv_prog_ac_ct_CXX
 if test -n "$ac_ct_CXX"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CXX" >&5
+  { $as_echo "$as_me:$LINENO: result: $ac_ct_CXX" >&5
 $as_echo "$ac_ct_CXX" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3072,7 +2474,7 @@ done
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -3083,31 +2485,48 @@ fi
   fi
 fi
 # Provide some information about the compiler.
-$as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler version" >&5
+$as_echo "$as_me:$LINENO: checking for C++ compiler version" >&5
 set X $ac_compile
 ac_compiler=$2
-for ac_option in --version -v -V -qversion; do
-  { { ac_try="$ac_compiler $ac_option >&5"
+{ (ac_try="$ac_compiler --version >&5"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler --version >&5") 2>&5
   ac_status=$?
-  if test -s conftest.err; then
-    sed '10a\
-... rest of stderr output deleted ...
-         10q' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-  fi
-  rm -f conftest.er1 conftest.err
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }
-done
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
+{ (ac_try="$ac_compiler -v >&5"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler -v >&5") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
+{ (ac_try="$ac_compiler -V >&5"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler -V >&5") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
 
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3123,8 +2542,8 @@ ac_clean_files="$ac_clean_files a.out a.out.dSYM a.exe b.out"
 # Try to create an executable without -o first, disregard a.out.
 # It will help us diagnose broken compilers, and finding out an intuition
 # of exeext.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether the C++ compiler works" >&5
-$as_echo_n "checking whether the C++ compiler works... " >&6; }
+{ $as_echo "$as_me:$LINENO: checking for C++ compiler default output file name" >&5
+$as_echo_n "checking for C++ compiler default output file name... " >&6; }
 ac_link_default=`$as_echo "$ac_link" | sed 's/ -o *conftest[^ ]*//'`
 
 # The possible output files:
@@ -3140,17 +2559,17 @@ do
 done
 rm -f $ac_rmfiles
 
-if { { ac_try="$ac_link_default"
+if { (ac_try="$ac_link_default"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
   (eval "$ac_link_default") 2>&5
   ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; then :
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; then
   # Autoconf-2.13 could set the ac_cv_exeext variable to `no'.
 # So ignore a value of `no', otherwise this would lead to `EXEEXT = no'
 # in a Makefile.  We should not override ac_cv_exeext if it was cached,
@@ -3167,7 +2586,7 @@ do
 	# certainly right.
 	break;;
     *.* )
-	if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
+        if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
 	then :; else
 	   ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
 	fi
@@ -3186,41 +2605,84 @@ test "$ac_cv_exeext" = no && ac_cv_exeext=
 else
   ac_file=''
 fi
-if test -z "$ac_file"; then :
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-$as_echo "$as_me: failed program was:" >&5
+
+{ $as_echo "$as_me:$LINENO: result: $ac_file" >&5
+$as_echo "$ac_file" >&6; }
+if test -z "$ac_file"; then
+  $as_echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
 
-{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error 77 "C++ compiler cannot create executables
-See \`config.log' for more details" "$LINENO" 5; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
-$as_echo "yes" >&6; }
+{ { $as_echo "$as_me:$LINENO: error: C++ compiler cannot create executables
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: C++ compiler cannot create executables
+See \`config.log' for more details." >&2;}
+   { (exit 77); exit 77; }; }; }
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler default output file name" >&5
-$as_echo_n "checking for C++ compiler default output file name... " >&6; }
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_file" >&5
-$as_echo "$ac_file" >&6; }
+
 ac_exeext=$ac_cv_exeext
 
+# Check that the compiler produces executables we can run.  If not, either
+# the compiler is broken, or we cross compile.
+{ $as_echo "$as_me:$LINENO: checking whether the C++ compiler works" >&5
+$as_echo_n "checking whether the C++ compiler works... " >&6; }
+# FIXME: These cross compiler hacks should be removed for Autoconf 3.0
+# If not cross compiling, check that we can run a simple program.
+if test "$cross_compiling" != yes; then
+  if { ac_try='./$ac_file'
+  { (case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; }; then
+    cross_compiling=no
+  else
+    if test "$cross_compiling" = maybe; then
+	cross_compiling=yes
+    else
+	{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+{ { $as_echo "$as_me:$LINENO: error: cannot run C++ compiled programs.
+If you meant to cross compile, use \`--host'.
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: cannot run C++ compiled programs.
+If you meant to cross compile, use \`--host'.
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
+    fi
+  fi
+fi
+{ $as_echo "$as_me:$LINENO: result: yes" >&5
+$as_echo "yes" >&6; }
+
 rm -f -r a.out a.out.dSYM a.exe conftest$ac_cv_exeext b.out
 ac_clean_files=$ac_clean_files_save
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of executables" >&5
+# Check that the compiler produces executables we can run.  If not, either
+# the compiler is broken, or we cross compile.
+{ $as_echo "$as_me:$LINENO: checking whether we are cross compiling" >&5
+$as_echo_n "checking whether we are cross compiling... " >&6; }
+{ $as_echo "$as_me:$LINENO: result: $cross_compiling" >&5
+$as_echo "$cross_compiling" >&6; }
+
+{ $as_echo "$as_me:$LINENO: checking for suffix of executables" >&5
 $as_echo_n "checking for suffix of executables... " >&6; }
-if { { ac_try="$ac_link"
+if { (ac_try="$ac_link"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
   (eval "$ac_link") 2>&5
   ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; then :
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; then
   # If both `conftest.exe' and `conftest' are `present' (well, observable)
 # catch `conftest.exe'.  For instance with Cygwin, `ls conftest' will
 # work properly (i.e., refer to `conftest.exe'), while it won't with
@@ -3235,83 +2697,32 @@ for ac_file in conftest.exe conftest conftest.*; do
   esac
 done
 else
-  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+  { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "cannot compute suffix of executables: cannot compile and link
-See \`config.log' for more details" "$LINENO" 5; }
+{ { $as_echo "$as_me:$LINENO: error: cannot compute suffix of executables: cannot compile and link
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: cannot compute suffix of executables: cannot compile and link
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
 fi
-rm -f conftest conftest$ac_cv_exeext
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_exeext" >&5
+
+rm -f conftest$ac_cv_exeext
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_exeext" >&5
 $as_echo "$ac_cv_exeext" >&6; }
 
 rm -f conftest.$ac_ext
 EXEEXT=$ac_cv_exeext
 ac_exeext=$EXEEXT
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <stdio.h>
-int
-main ()
-{
-FILE *f = fopen ("conftest.out", "w");
- return ferror (f) || fclose (f) != 0;
-
-  ;
-  return 0;
-}
-_ACEOF
-ac_clean_files="$ac_clean_files conftest.out"
-# Check that the compiler produces executables we can run.  If not, either
-# the compiler is broken, or we cross compile.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are cross compiling" >&5
-$as_echo_n "checking whether we are cross compiling... " >&6; }
-if test "$cross_compiling" != yes; then
-  { { ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_link") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }
-  if { ac_try='./conftest$ac_cv_exeext'
-  { { case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; }; then
-    cross_compiling=no
-  else
-    if test "$cross_compiling" = maybe; then
-	cross_compiling=yes
-    else
-	{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "cannot run C++ compiled programs.
-If you meant to cross compile, use \`--host'.
-See \`config.log' for more details" "$LINENO" 5; }
-    fi
-  fi
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $cross_compiling" >&5
-$as_echo "$cross_compiling" >&6; }
-
-rm -f conftest.$ac_ext conftest$ac_cv_exeext conftest.out
-ac_clean_files=$ac_clean_files_save
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of object files" >&5
+{ $as_echo "$as_me:$LINENO: checking for suffix of object files" >&5
 $as_echo_n "checking for suffix of object files... " >&6; }
-if ${ac_cv_objext+:} false; then :
+if test "${ac_cv_objext+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3323,17 +2734,17 @@ main ()
 }
 _ACEOF
 rm -f conftest.o conftest.obj
-if { { ac_try="$ac_compile"
+if { (ac_try="$ac_compile"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
   (eval "$ac_compile") 2>&5
   ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; then :
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; then
   for ac_file in conftest.o conftest.obj conftest.*; do
   test -f "$ac_file" || continue;
   case $ac_file in
@@ -3346,23 +2757,31 @@ else
   $as_echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
 
-{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "cannot compute suffix of object files: cannot compile
-See \`config.log' for more details" "$LINENO" 5; }
+{ { $as_echo "$as_me:$LINENO: error: cannot compute suffix of object files: cannot compile
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: cannot compute suffix of object files: cannot compile
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
 fi
+
 rm -f conftest.$ac_cv_objext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_objext" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_objext" >&5
 $as_echo "$ac_cv_objext" >&6; }
 OBJEXT=$ac_cv_objext
 ac_objext=$OBJEXT
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C++ compiler" >&5
+{ $as_echo "$as_me:$LINENO: checking whether we are using the GNU C++ compiler" >&5
 $as_echo_n "checking whether we are using the GNU C++ compiler... " >&6; }
-if ${ac_cv_cxx_compiler_gnu+:} false; then :
+if test "${ac_cv_cxx_compiler_gnu+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3376,16 +2795,37 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_compiler_gnu=yes
 else
-  ac_compiler_gnu=no
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_compiler_gnu=no
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 ac_cv_cxx_compiler_gnu=$ac_compiler_gnu
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_cxx_compiler_gnu" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_cxx_compiler_gnu" >&5
 $as_echo "$ac_cv_cxx_compiler_gnu" >&6; }
 if test $ac_compiler_gnu = yes; then
   GXX=yes
@@ -3394,16 +2834,20 @@ else
 fi
 ac_test_CXXFLAGS=${CXXFLAGS+set}
 ac_save_CXXFLAGS=$CXXFLAGS
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX accepts -g" >&5
+{ $as_echo "$as_me:$LINENO: checking whether $CXX accepts -g" >&5
 $as_echo_n "checking whether $CXX accepts -g... " >&6; }
-if ${ac_cv_prog_cxx_g+:} false; then :
+if test "${ac_cv_prog_cxx_g+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   ac_save_cxx_werror_flag=$ac_cxx_werror_flag
    ac_cxx_werror_flag=yes
    ac_cv_prog_cxx_g=no
    CXXFLAGS="-g"
-   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+   cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3414,11 +2858,35 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_prog_cxx_g=yes
 else
-  CXXFLAGS=""
-      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	CXXFLAGS=""
+      cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3429,12 +2897,36 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  :
 else
-  ac_cxx_werror_flag=$ac_save_cxx_werror_flag
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cxx_werror_flag=$ac_save_cxx_werror_flag
 	 CXXFLAGS="-g"
-	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+	 cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3445,17 +2937,42 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_prog_cxx_g=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
    ac_cxx_werror_flag=$ac_save_cxx_werror_flag
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cxx_g" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cxx_g" >&5
 $as_echo "$ac_cv_prog_cxx_g" >&6; }
 if test "$ac_test_CXXFLAGS" = set; then
   CXXFLAGS=$ac_save_CXXFLAGS
@@ -3489,14 +3006,14 @@ am__doit:
 .PHONY: am__doit
 END
 # If we don't find an include directive, just comment out the code.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for style of include used by $am_make" >&5
+{ $as_echo "$as_me:$LINENO: checking for style of include used by $am_make" >&5
 $as_echo_n "checking for style of include used by $am_make... " >&6; }
 am__include="#"
 am__quote=
 _am_result=none
 # First try GNU make style include.
 echo "include confinc" > confmf
-# Ignore all kinds of additional output from 'make'.
+# Ignore all kinds of additional output from `make'.
 case `$am_make -s -f confmf 2> /dev/null` in #(
 *the\ am__doit\ target*)
   am__include=include
@@ -3517,19 +3034,18 @@ if test "$am__include" = "#"; then
 fi
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $_am_result" >&5
+{ $as_echo "$as_me:$LINENO: result: $_am_result" >&5
 $as_echo "$_am_result" >&6; }
 rm -f confinc confmf
 
 # Check whether --enable-dependency-tracking was given.
-if test "${enable_dependency_tracking+set}" = set; then :
+if test "${enable_dependency_tracking+set}" = set; then
   enableval=$enable_dependency_tracking;
 fi
 
 if test "x$enable_dependency_tracking" != xno; then
   am_depcomp="$ac_aux_dir/depcomp"
   AMDEPBACKSLASH='\'
-  am__nodep='_no'
 fi
  if test "x$enable_dependency_tracking" != xno; then
   AMDEP_TRUE=
@@ -3543,18 +3059,17 @@ fi
 
 depcc="$CXX"  am_compiler_list=
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
+{ $as_echo "$as_me:$LINENO: checking dependency style of $depcc" >&5
 $as_echo_n "checking dependency style of $depcc... " >&6; }
-if ${am_cv_CXX_dependencies_compiler_type+:} false; then :
+if test "${am_cv_CXX_dependencies_compiler_type+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named 'D' -- because '-MD' means "put the output
-  # in D".
-  rm -rf conftest.dir
+  # making a dummy file named `D' -- because `-MD' means `put the output
+  # in D'.
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -3588,16 +3103,16 @@ else
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
-      # Solaris 10 /bin/sh.
-      echo '/* dummy */' > sub/conftst$i.h
+      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
+      # Solaris 8's {/usr,}/bin/sh.
+      touch sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with '-c' and '-o' for the sake of the "dashmstdout"
+    # We check with `-c' and `-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle '-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs.
+    # handle `-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -3606,16 +3121,16 @@ else
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # After this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested.
+      # after this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
-      # This compiler won't grok '-c -o', but also, the minuso test has
+    msvisualcpp | msvcmsys)
+      # This compiler won't grok `-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -3654,7 +3169,7 @@ else
 fi
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CXX_dependencies_compiler_type" >&5
+{ $as_echo "$as_me:$LINENO: result: $am_cv_CXX_dependencies_compiler_type" >&5
 $as_echo "$am_cv_CXX_dependencies_compiler_type" >&6; }
 CXXDEPMODE=depmode=$am_cv_CXX_dependencies_compiler_type
 
@@ -3677,9 +3192,9 @@ ac_compiler_gnu=$ac_cv_c_compiler_gnu
 if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}gcc", so it can be a program name with args.
 set dummy ${ac_tool_prefix}gcc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
+if test "${ac_cv_prog_CC+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3690,24 +3205,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_CC="${ac_tool_prefix}gcc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+  { $as_echo "$as_me:$LINENO: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3717,9 +3232,9 @@ if test -z "$ac_cv_prog_CC"; then
   ac_ct_CC=$CC
   # Extract the first word of "gcc", so it can be a program name with args.
 set dummy gcc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_CC+:} false; then :
+if test "${ac_cv_prog_ac_ct_CC+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CC"; then
@@ -3730,24 +3245,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_ac_ct_CC="gcc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CC=$ac_cv_prog_ac_ct_CC
 if test -n "$ac_ct_CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
+  { $as_echo "$as_me:$LINENO: result: $ac_ct_CC" >&5
 $as_echo "$ac_ct_CC" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3756,7 +3271,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -3770,9 +3285,9 @@ if test -z "$CC"; then
           if test -n "$ac_tool_prefix"; then
     # Extract the first word of "${ac_tool_prefix}cc", so it can be a program name with args.
 set dummy ${ac_tool_prefix}cc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
+if test "${ac_cv_prog_CC+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3783,24 +3298,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_CC="${ac_tool_prefix}cc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+  { $as_echo "$as_me:$LINENO: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3810,9 +3325,9 @@ fi
 if test -z "$CC"; then
   # Extract the first word of "cc", so it can be a program name with args.
 set dummy cc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
+if test "${ac_cv_prog_CC+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3824,18 +3339,18 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     if test "$as_dir/$ac_word$ac_exec_ext" = "/usr/ucb/cc"; then
        ac_prog_rejected=yes
        continue
      fi
     ac_cv_prog_CC="cc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 if test $ac_prog_rejected = yes; then
@@ -3854,10 +3369,10 @@ fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+  { $as_echo "$as_me:$LINENO: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3869,9 +3384,9 @@ if test -z "$CC"; then
   do
     # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
 set dummy $ac_tool_prefix$ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
+if test "${ac_cv_prog_CC+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3882,24 +3397,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_CC="$ac_tool_prefix$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+  { $as_echo "$as_me:$LINENO: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3913,9 +3428,9 @@ if test -z "$CC"; then
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_CC+:} false; then :
+if test "${ac_cv_prog_ac_ct_CC+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CC"; then
@@ -3926,24 +3441,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_ac_ct_CC="$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CC=$ac_cv_prog_ac_ct_CC
 if test -n "$ac_ct_CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
+  { $as_echo "$as_me:$LINENO: result: $ac_ct_CC" >&5
 $as_echo "$ac_ct_CC" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3956,7 +3471,7 @@ done
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -3967,42 +3482,62 @@ fi
 fi
 
 
-test -z "$CC" && { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+test -z "$CC" && { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "no acceptable C compiler found in \$PATH
-See \`config.log' for more details" "$LINENO" 5; }
+{ { $as_echo "$as_me:$LINENO: error: no acceptable C compiler found in \$PATH
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: no acceptable C compiler found in \$PATH
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
 
 # Provide some information about the compiler.
-$as_echo "$as_me:${as_lineno-$LINENO}: checking for C compiler version" >&5
+$as_echo "$as_me:$LINENO: checking for C compiler version" >&5
 set X $ac_compile
 ac_compiler=$2
-for ac_option in --version -v -V -qversion; do
-  { { ac_try="$ac_compiler $ac_option >&5"
+{ (ac_try="$ac_compiler --version >&5"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler --version >&5") 2>&5
   ac_status=$?
-  if test -s conftest.err; then
-    sed '10a\
-... rest of stderr output deleted ...
-         10q' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-  fi
-  rm -f conftest.er1 conftest.err
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }
-done
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
+{ (ac_try="$ac_compiler -v >&5"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler -v >&5") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
+{ (ac_try="$ac_compiler -V >&5"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler -V >&5") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C compiler" >&5
+{ $as_echo "$as_me:$LINENO: checking whether we are using the GNU C compiler" >&5
 $as_echo_n "checking whether we are using the GNU C compiler... " >&6; }
-if ${ac_cv_c_compiler_gnu+:} false; then :
+if test "${ac_cv_c_compiler_gnu+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -4016,16 +3551,37 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_compiler_gnu=yes
 else
-  ac_compiler_gnu=no
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_compiler_gnu=no
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 ac_cv_c_compiler_gnu=$ac_compiler_gnu
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_c_compiler_gnu" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_compiler_gnu" >&5
 $as_echo "$ac_cv_c_compiler_gnu" >&6; }
 if test $ac_compiler_gnu = yes; then
   GCC=yes
@@ -4034,16 +3590,20 @@ else
 fi
 ac_test_CFLAGS=${CFLAGS+set}
 ac_save_CFLAGS=$CFLAGS
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC accepts -g" >&5
+{ $as_echo "$as_me:$LINENO: checking whether $CC accepts -g" >&5
 $as_echo_n "checking whether $CC accepts -g... " >&6; }
-if ${ac_cv_prog_cc_g+:} false; then :
+if test "${ac_cv_prog_cc_g+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   ac_save_c_werror_flag=$ac_c_werror_flag
    ac_c_werror_flag=yes
    ac_cv_prog_cc_g=no
    CFLAGS="-g"
-   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+   cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -4054,11 +3614,35 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_prog_cc_g=yes
 else
-  CFLAGS=""
-      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	CFLAGS=""
+      cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -4069,12 +3653,36 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  :
 else
-  ac_c_werror_flag=$ac_save_c_werror_flag
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_c_werror_flag=$ac_save_c_werror_flag
 	 CFLAGS="-g"
-	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+	 cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -4085,17 +3693,42 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_prog_cc_g=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
    ac_c_werror_flag=$ac_save_c_werror_flag
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_g" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cc_g" >&5
 $as_echo "$ac_cv_prog_cc_g" >&6; }
 if test "$ac_test_CFLAGS" = set; then
   CFLAGS=$ac_save_CFLAGS
@@ -4112,18 +3745,23 @@ else
     CFLAGS=
   fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to accept ISO C89" >&5
+{ $as_echo "$as_me:$LINENO: checking for $CC option to accept ISO C89" >&5
 $as_echo_n "checking for $CC option to accept ISO C89... " >&6; }
-if ${ac_cv_prog_cc_c89+:} false; then :
+if test "${ac_cv_prog_cc_c89+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   ac_cv_prog_cc_c89=no
 ac_save_CC=$CC
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <stdarg.h>
 #include <stdio.h>
-struct stat;
+#include <sys/types.h>
+#include <sys/stat.h>
 /* Most of the following tests are stolen from RCS 5.7's src/conf.sh.  */
 struct buf { int x; };
 FILE * (*rcsopen) (struct buf *, struct stat *, int);
@@ -4175,9 +3813,32 @@ for ac_arg in '' -qlanglvl=extc89 -qlanglvl=ansi -std \
 	-Ae "-Aa -D_HPUX_SOURCE" "-Xc -D__EXTENSIONS__"
 do
   CC="$ac_save_CC $ac_arg"
-  if ac_fn_c_try_compile "$LINENO"; then :
+  rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_prog_cc_c89=$ac_arg
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
 fi
+
 rm -f core conftest.err conftest.$ac_objext
   test "x$ac_cv_prog_cc_c89" != "xno" && break
 done
@@ -4188,19 +3849,17 @@ fi
 # AC_CACHE_VAL
 case "x$ac_cv_prog_cc_c89" in
   x)
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: none needed" >&5
+    { $as_echo "$as_me:$LINENO: result: none needed" >&5
 $as_echo "none needed" >&6; } ;;
   xno)
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: unsupported" >&5
+    { $as_echo "$as_me:$LINENO: result: unsupported" >&5
 $as_echo "unsupported" >&6; } ;;
   *)
     CC="$CC $ac_cv_prog_cc_c89"
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_c89" >&5
+    { $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cc_c89" >&5
 $as_echo "$ac_cv_prog_cc_c89" >&6; } ;;
 esac
-if test "x$ac_cv_prog_cc_c89" != xno; then :
 
-fi
 
 ac_ext=c
 ac_cpp='$CPP $CPPFLAGS'
@@ -4208,79 +3867,19 @@ ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
 ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC understands -c and -o together" >&5
-$as_echo_n "checking whether $CC understands -c and -o together... " >&6; }
-if ${am_cv_prog_cc_c_o+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-  # Make sure it works both with $CC and with simple cc.
-  # Following AC_PROG_CC_C_O, we do the test twice because some
-  # compilers refuse to overwrite an existing .o file with -o,
-  # though they will create one.
-  am_cv_prog_cc_c_o=yes
-  for am_i in 1 2; do
-    if { echo "$as_me:$LINENO: $CC -c conftest.$ac_ext -o conftest2.$ac_objext" >&5
-   ($CC -c conftest.$ac_ext -o conftest2.$ac_objext) >&5 2>&5
-   ac_status=$?
-   echo "$as_me:$LINENO: \$? = $ac_status" >&5
-   (exit $ac_status); } \
-         && test -f conftest2.$ac_objext; then
-      : OK
-    else
-      am_cv_prog_cc_c_o=no
-      break
-    fi
-  done
-  rm -f core conftest*
-  unset am_i
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_prog_cc_c_o" >&5
-$as_echo "$am_cv_prog_cc_c_o" >&6; }
-if test "$am_cv_prog_cc_c_o" != yes; then
-   # Losing compiler, so override with the script.
-   # FIXME: It is wrong to rewrite CC.
-   # But if we don't then we get into trouble of one sort or another.
-   # A longer-term fix would be to have automake use am__CC in this case,
-   # and then we could set am__CC="\$(top_srcdir)/compile \$(CC)"
-   CC="$am_aux_dir/compile $CC"
-fi
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-
-
 depcc="$CC"   am_compiler_list=
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
+{ $as_echo "$as_me:$LINENO: checking dependency style of $depcc" >&5
 $as_echo_n "checking dependency style of $depcc... " >&6; }
-if ${am_cv_CC_dependencies_compiler_type+:} false; then :
+if test "${am_cv_CC_dependencies_compiler_type+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named 'D' -- because '-MD' means "put the output
-  # in D".
-  rm -rf conftest.dir
+  # making a dummy file named `D' -- because `-MD' means `put the output
+  # in D'.
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -4314,16 +3913,16 @@ else
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
-      # Solaris 10 /bin/sh.
-      echo '/* dummy */' > sub/conftst$i.h
+      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
+      # Solaris 8's {/usr,}/bin/sh.
+      touch sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with '-c' and '-o' for the sake of the "dashmstdout"
+    # We check with `-c' and `-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle '-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs.
+    # handle `-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -4332,16 +3931,16 @@ else
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # After this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested.
+      # after this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
-      # This compiler won't grok '-c -o', but also, the minuso test has
+    msvisualcpp | msvcmsys)
+      # This compiler won't grok `-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -4380,7 +3979,7 @@ else
 fi
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CC_dependencies_compiler_type" >&5
+{ $as_echo "$as_me:$LINENO: result: $am_cv_CC_dependencies_compiler_type" >&5
 $as_echo "$am_cv_CC_dependencies_compiler_type" >&6; }
 CCDEPMODE=depmode=$am_cv_CC_dependencies_compiler_type
 
@@ -4399,18 +3998,17 @@ fi
 
   OPENMP_CFLAGS=
   # Check whether --enable-openmp was given.
-if test "${enable_openmp+set}" = set; then :
+if test "${enable_openmp+set}" = set; then
   enableval=$enable_openmp;
 fi
 
   if test "$enable_openmp" != no; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to support OpenMP" >&5
+    { $as_echo "$as_me:$LINENO: checking for $CC option to support OpenMP" >&5
 $as_echo_n "checking for $CC option to support OpenMP... " >&6; }
-if ${ac_cv_prog_c_openmp+:} false; then :
+if test "${ac_cv_prog_c_openmp+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
+  cat >conftest.$ac_ext <<_ACEOF
 
 #ifndef _OPENMP
  choke me
@@ -4419,16 +4017,37 @@ else
 int main () { return omp_get_num_threads (); }
 
 _ACEOF
-if ac_fn_c_try_link "$LINENO"; then :
+rm -f conftest.$ac_objext conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_link") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 $as_test_x conftest$ac_exeext
+       }; then
   ac_cv_prog_c_openmp='none needed'
 else
-  ac_cv_prog_c_openmp='unsupported'
-	  	  	  	  	  	  	                                	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp -homp \
-                           -Popenmp --openmp; do
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cv_prog_c_openmp='unsupported'
+	  	  	  	  	  	  	  	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp; do
 	    ac_save_CFLAGS=$CFLAGS
 	    CFLAGS="$CFLAGS $ac_option"
-	    cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
+	    cat >conftest.$ac_ext <<_ACEOF
 
 #ifndef _OPENMP
  choke me
@@ -4437,27 +4056,56 @@ else
 int main () { return omp_get_num_threads (); }
 
 _ACEOF
-if ac_fn_c_try_link "$LINENO"; then :
+rm -f conftest.$ac_objext conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_link") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 $as_test_x conftest$ac_exeext
+       }; then
   ac_cv_prog_c_openmp=$ac_option
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
 fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
+
+rm -rf conftest.dSYM
+rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
+      conftest$ac_exeext conftest.$ac_ext
 	    CFLAGS=$ac_save_CFLAGS
 	    if test "$ac_cv_prog_c_openmp" != unsupported; then
 	      break
 	    fi
 	  done
 fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
+
+rm -rf conftest.dSYM
+rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
+      conftest$ac_exeext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_c_openmp" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_c_openmp" >&5
 $as_echo "$ac_cv_prog_c_openmp" >&6; }
     case $ac_cv_prog_c_openmp in #(
       "none needed" | unsupported)
-	;; #(
+        ;; #(
       *)
-	OPENMP_CFLAGS=$ac_cv_prog_c_openmp ;;
+        OPENMP_CFLAGS=$ac_cv_prog_c_openmp ;;
     esac
   fi
 
@@ -4465,9 +4113,9 @@ $as_echo "$ac_cv_prog_c_openmp" >&6; }
 if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}ranlib", so it can be a program name with args.
 set dummy ${ac_tool_prefix}ranlib; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_RANLIB+:} false; then :
+if test "${ac_cv_prog_RANLIB+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$RANLIB"; then
@@ -4478,24 +4126,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_RANLIB="${ac_tool_prefix}ranlib"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 RANLIB=$ac_cv_prog_RANLIB
 if test -n "$RANLIB"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $RANLIB" >&5
+  { $as_echo "$as_me:$LINENO: result: $RANLIB" >&5
 $as_echo "$RANLIB" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -4505,9 +4153,9 @@ if test -z "$ac_cv_prog_RANLIB"; then
   ac_ct_RANLIB=$RANLIB
   # Extract the first word of "ranlib", so it can be a program name with args.
 set dummy ranlib; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_RANLIB+:} false; then :
+if test "${ac_cv_prog_ac_ct_RANLIB+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_RANLIB"; then
@@ -4518,24 +4166,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_ac_ct_RANLIB="ranlib"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_RANLIB=$ac_cv_prog_ac_ct_RANLIB
 if test -n "$ac_ct_RANLIB"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_RANLIB" >&5
+  { $as_echo "$as_me:$LINENO: result: $ac_ct_RANLIB" >&5
 $as_echo "$ac_ct_RANLIB" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -4544,7 +4192,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -4564,14 +4212,14 @@ ac_cpp='$CPP $CPPFLAGS'
 ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
 ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking how to run the C preprocessor" >&5
+{ $as_echo "$as_me:$LINENO: checking how to run the C preprocessor" >&5
 $as_echo_n "checking how to run the C preprocessor... " >&6; }
 # On Suns, sometimes $CPP names a directory.
 if test -n "$CPP" && test -d "$CPP"; then
   CPP=
 fi
 if test -z "$CPP"; then
-  if ${ac_cv_prog_CPP+:} false; then :
+  if test "${ac_cv_prog_CPP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
       # Double quotes because CPP needs to be expanded
@@ -4586,7 +4234,11 @@ do
   # <limits.h> exists even on freestanding compilers.
   # On the NeXT, cc -E runs the code through the compiler's parser,
   # not just through cpp. "Syntax error" is here to catch this case.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #ifdef __STDC__
 # include <limits.h>
@@ -4595,34 +4247,78 @@ do
 #endif
 		     Syntax error
 _ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
-
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  :
 else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
   # Broken: fails on valid input.
 continue
 fi
-rm -f conftest.err conftest.i conftest.$ac_ext
+
+rm -f conftest.err conftest.$ac_ext
 
   # OK, works on sane cases.  Now check whether nonexistent headers
   # can be detected and how.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <ac_nonexistent.h>
 _ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
   # Broken: success on invalid input.
 continue
 else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
   # Passes both tests.
 ac_preproc_ok=:
 break
 fi
-rm -f conftest.err conftest.i conftest.$ac_ext
+
+rm -f conftest.err conftest.$ac_ext
 
 done
 # Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
-rm -f conftest.i conftest.err conftest.$ac_ext
-if $ac_preproc_ok; then :
+rm -f conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then
   break
 fi
 
@@ -4634,7 +4330,7 @@ fi
 else
   ac_cv_prog_CPP=$CPP
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $CPP" >&5
+{ $as_echo "$as_me:$LINENO: result: $CPP" >&5
 $as_echo "$CPP" >&6; }
 ac_preproc_ok=false
 for ac_c_preproc_warn_flag in '' yes
@@ -4645,7 +4341,11 @@ do
   # <limits.h> exists even on freestanding compilers.
   # On the NeXT, cc -E runs the code through the compiler's parser,
   # not just through cpp. "Syntax error" is here to catch this case.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #ifdef __STDC__
 # include <limits.h>
@@ -4654,40 +4354,87 @@ do
 #endif
 		     Syntax error
 _ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
-
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  :
 else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
   # Broken: fails on valid input.
 continue
 fi
-rm -f conftest.err conftest.i conftest.$ac_ext
+
+rm -f conftest.err conftest.$ac_ext
 
   # OK, works on sane cases.  Now check whether nonexistent headers
   # can be detected and how.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <ac_nonexistent.h>
 _ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
   # Broken: success on invalid input.
 continue
 else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
   # Passes both tests.
 ac_preproc_ok=:
 break
 fi
-rm -f conftest.err conftest.i conftest.$ac_ext
+
+rm -f conftest.err conftest.$ac_ext
 
 done
 # Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
-rm -f conftest.i conftest.err conftest.$ac_ext
-if $ac_preproc_ok; then :
-
+rm -f conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then
+  :
 else
-  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+  { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "C preprocessor \"$CPP\" fails sanity check
-See \`config.log' for more details" "$LINENO" 5; }
+{ { $as_echo "$as_me:$LINENO: error: C preprocessor \"$CPP\" fails sanity check
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: C preprocessor \"$CPP\" fails sanity check
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
 fi
 
 ac_ext=c
@@ -4697,9 +4444,9 @@ ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for grep that handles long lines and -e" >&5
+{ $as_echo "$as_me:$LINENO: checking for grep that handles long lines and -e" >&5
 $as_echo_n "checking for grep that handles long lines and -e... " >&6; }
-if ${ac_cv_path_GREP+:} false; then :
+if test "${ac_cv_path_GREP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -z "$GREP"; then
@@ -4710,10 +4457,10 @@ for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_prog in grep ggrep; do
+  for ac_prog in grep ggrep; do
     for ac_exec_ext in '' $ac_executable_extensions; do
       ac_path_GREP="$as_dir/$ac_prog$ac_exec_ext"
-      as_fn_executable_p "$ac_path_GREP" || continue
+      { test -f "$ac_path_GREP" && $as_test_x "$ac_path_GREP"; } || continue
 # Check for GNU ac_path_GREP and select it if it is found.
   # Check for GNU $ac_path_GREP
 case `"$ac_path_GREP" --version 2>&1` in
@@ -4730,7 +4477,7 @@ case `"$ac_path_GREP" --version 2>&1` in
     $as_echo 'GREP' >> "conftest.nl"
     "$ac_path_GREP" -e 'GREP$' -e '-(cannot match)-' < "conftest.nl" >"conftest.out" 2>/dev/null || break
     diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
-    as_fn_arith $ac_count + 1 && ac_count=$as_val
+    ac_count=`expr $ac_count + 1`
     if test $ac_count -gt ${ac_path_GREP_max-0}; then
       # Best one so far, save it but keep looking for a better one
       ac_cv_path_GREP="$ac_path_GREP"
@@ -4745,24 +4492,26 @@ esac
       $ac_path_GREP_found && break 3
     done
   done
-  done
+done
 IFS=$as_save_IFS
   if test -z "$ac_cv_path_GREP"; then
-    as_fn_error $? "no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&5
+$as_echo "$as_me: error: no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&2;}
+   { (exit 1); exit 1; }; }
   fi
 else
   ac_cv_path_GREP=$GREP
 fi
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_GREP" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_path_GREP" >&5
 $as_echo "$ac_cv_path_GREP" >&6; }
  GREP="$ac_cv_path_GREP"
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for egrep" >&5
+{ $as_echo "$as_me:$LINENO: checking for egrep" >&5
 $as_echo_n "checking for egrep... " >&6; }
-if ${ac_cv_path_EGREP+:} false; then :
+if test "${ac_cv_path_EGREP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if echo a | $GREP -E '(a|b)' >/dev/null 2>&1
@@ -4776,10 +4525,10 @@ for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_prog in egrep; do
+  for ac_prog in egrep; do
     for ac_exec_ext in '' $ac_executable_extensions; do
       ac_path_EGREP="$as_dir/$ac_prog$ac_exec_ext"
-      as_fn_executable_p "$ac_path_EGREP" || continue
+      { test -f "$ac_path_EGREP" && $as_test_x "$ac_path_EGREP"; } || continue
 # Check for GNU ac_path_EGREP and select it if it is found.
   # Check for GNU $ac_path_EGREP
 case `"$ac_path_EGREP" --version 2>&1` in
@@ -4796,7 +4545,7 @@ case `"$ac_path_EGREP" --version 2>&1` in
     $as_echo 'EGREP' >> "conftest.nl"
     "$ac_path_EGREP" 'EGREP$' < "conftest.nl" >"conftest.out" 2>/dev/null || break
     diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
-    as_fn_arith $ac_count + 1 && ac_count=$as_val
+    ac_count=`expr $ac_count + 1`
     if test $ac_count -gt ${ac_path_EGREP_max-0}; then
       # Best one so far, save it but keep looking for a better one
       ac_cv_path_EGREP="$ac_path_EGREP"
@@ -4811,10 +4560,12 @@ esac
       $ac_path_EGREP_found && break 3
     done
   done
-  done
+done
 IFS=$as_save_IFS
   if test -z "$ac_cv_path_EGREP"; then
-    as_fn_error $? "no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&5
+$as_echo "$as_me: error: no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&2;}
+   { (exit 1); exit 1; }; }
   fi
 else
   ac_cv_path_EGREP=$EGREP
@@ -4822,17 +4573,21 @@ fi
 
    fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_EGREP" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_path_EGREP" >&5
 $as_echo "$ac_cv_path_EGREP" >&6; }
  EGREP="$ac_cv_path_EGREP"
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for ANSI C header files" >&5
+{ $as_echo "$as_me:$LINENO: checking for ANSI C header files" >&5
 $as_echo_n "checking for ANSI C header files... " >&6; }
-if ${ac_cv_header_stdc+:} false; then :
+if test "${ac_cv_header_stdc+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <stdlib.h>
 #include <stdarg.h>
@@ -4847,23 +4602,48 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_header_stdc=yes
 else
-  ac_cv_header_stdc=no
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cv_header_stdc=no
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 
 if test $ac_cv_header_stdc = yes; then
   # SunOS 4.x string.h does not declare mem*, contrary to ANSI.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <string.h>
 
 _ACEOF
 if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
-  $EGREP "memchr" >/dev/null 2>&1; then :
-
+  $EGREP "memchr" >/dev/null 2>&1; then
+  :
 else
   ac_cv_header_stdc=no
 fi
@@ -4873,14 +4653,18 @@ fi
 
 if test $ac_cv_header_stdc = yes; then
   # ISC 2.0.2 stdlib.h does not declare free, contrary to ANSI.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <stdlib.h>
 
 _ACEOF
 if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
-  $EGREP "free" >/dev/null 2>&1; then :
-
+  $EGREP "free" >/dev/null 2>&1; then
+  :
 else
   ac_cv_header_stdc=no
 fi
@@ -4890,10 +4674,14 @@ fi
 
 if test $ac_cv_header_stdc = yes; then
   # /bin/cc in Irix-4.0.5 gets non-ANSI ctype macros unless using -ansi.
-  if test "$cross_compiling" = yes; then :
+  if test "$cross_compiling" = yes; then
   :
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <ctype.h>
 #include <stdlib.h>
@@ -4920,33 +4708,118 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_run "$LINENO"; then :
-
+rm -f conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && { ac_try='./conftest$ac_exeext'
+  { (case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; }; then
+  :
 else
-  ac_cv_header_stdc=no
+  $as_echo "$as_me: program exited with status $ac_status" >&5
+$as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+( exit $ac_status )
+ac_cv_header_stdc=no
 fi
-rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
-  conftest.$ac_objext conftest.beam conftest.$ac_ext
+rm -rf conftest.dSYM
+rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext conftest.$ac_objext conftest.$ac_ext
 fi
 
+
 fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_header_stdc" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_header_stdc" >&5
 $as_echo "$ac_cv_header_stdc" >&6; }
 if test $ac_cv_header_stdc = yes; then
 
-$as_echo "#define STDC_HEADERS 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define STDC_HEADERS 1
+_ACEOF
 
 fi
 
 # On IRIX 5.3, sys/types and inttypes.h are conflicting.
+
+
+
+
+
+
+
+
+
 for ac_header in sys/types.h sys/stat.h stdlib.h string.h memory.h strings.h \
 		  inttypes.h stdint.h unistd.h
-do :
-  as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-ac_fn_c_check_header_compile "$LINENO" "$ac_header" "$as_ac_Header" "$ac_includes_default
-"
-if eval test \"x\$"$as_ac_Header"\" = x"yes"; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  eval "$as_ac_Header=yes"
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	eval "$as_ac_Header=no"
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
 #define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
@@ -4956,48 +4829,604 @@ fi
 done
 
 
+
 for ac_header in stdint.h
-do :
-  ac_fn_c_check_header_mongrel "$LINENO" "stdint.h" "ac_cv_header_stdint_h" "$ac_includes_default"
-if test "x$ac_cv_header_stdint_h" = xyes; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
+$as_echo_n "checking $ac_header usability... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_header_compiler=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_header_compiler=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
+$as_echo_n "checking $ac_header presence... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <$ac_header>
+_ACEOF
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  ac_header_preproc=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_header_preproc=no
+fi
+
+rm -f conftest.err conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
+  yes:no: )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
+    ac_header_preproc=yes
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
+$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
+    ( cat <<\_ASBOX
+## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##
+_ASBOX
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  eval "$as_ac_Header=\$ac_header_preproc"
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+
+fi
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_STDINT_H 1
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
 
 fi
 
 done
 
+
 for ac_header in malloc/malloc.h
-do :
-  ac_fn_c_check_header_mongrel "$LINENO" "malloc/malloc.h" "ac_cv_header_malloc_malloc_h" "$ac_includes_default"
-if test "x$ac_cv_header_malloc_malloc_h" = xyes; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
+$as_echo_n "checking $ac_header usability... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_header_compiler=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_header_compiler=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
+$as_echo_n "checking $ac_header presence... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <$ac_header>
+_ACEOF
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  ac_header_preproc=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_header_preproc=no
+fi
+
+rm -f conftest.err conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
+  yes:no: )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
+    ac_header_preproc=yes
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
+$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
+    ( cat <<\_ASBOX
+## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##
+_ASBOX
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  eval "$as_ac_Header=\$ac_header_preproc"
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+
+fi
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_MALLOC_MALLOC_H 1
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
 
 fi
 
 done
 
+
 for ac_header in malloc.h
-do :
-  ac_fn_c_check_header_mongrel "$LINENO" "malloc.h" "ac_cv_header_malloc_h" "$ac_includes_default"
-if test "x$ac_cv_header_malloc_h" = xyes; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
+$as_echo_n "checking $ac_header usability... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_header_compiler=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_header_compiler=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
+$as_echo_n "checking $ac_header presence... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <$ac_header>
+_ACEOF
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  ac_header_preproc=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_header_preproc=no
+fi
+
+rm -f conftest.err conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
+  yes:no: )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
+    ac_header_preproc=yes
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
+$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
+    ( cat <<\_ASBOX
+## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##
+_ASBOX
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  eval "$as_ac_Header=\$ac_header_preproc"
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+
+fi
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_MALLOC_H 1
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
 
 fi
 
 done
 
+
 for ac_header in endian.h
-do :
-  ac_fn_c_check_header_mongrel "$LINENO" "endian.h" "ac_cv_header_endian_h" "$ac_includes_default"
-if test "x$ac_cv_header_endian_h" = xyes; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
+$as_echo_n "checking $ac_header usability... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_header_compiler=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_header_compiler=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
+$as_echo_n "checking $ac_header presence... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <$ac_header>
+_ACEOF
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  ac_header_preproc=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_header_preproc=no
+fi
+
+rm -f conftest.err conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
+  yes:no: )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
+    ac_header_preproc=yes
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
+$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
+    ( cat <<\_ASBOX
+## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##
+_ASBOX
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  eval "$as_ac_Header=\$ac_header_preproc"
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+
+fi
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_ENDIAN_H 1
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
 
 fi
@@ -5005,35 +5434,244 @@ fi
 done
 
 #AC_CHECK_HEADERS(machine/endian.h)
-ac_fn_c_check_decl "$LINENO" "ntohll" "ac_cv_have_decl_ntohll" "#include <arpa/inet.h>
-"
-if test "x$ac_cv_have_decl_ntohll" = xyes; then :
-  ac_have_decl=1
+{ $as_echo "$as_me:$LINENO: checking whether ntohll is declared" >&5
+$as_echo_n "checking whether ntohll is declared... " >&6; }
+if test "${ac_cv_have_decl_ntohll+set}" = set; then
+  $as_echo_n "(cached) " >&6
 else
-  ac_have_decl=0
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <arpa/inet.h>
+
+int
+main ()
+{
+#ifndef ntohll
+  (void) ntohll;
+#endif
+
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_cv_have_decl_ntohll=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cv_have_decl_ntohll=no
 fi
 
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_have_decl_ntohll" >&5
+$as_echo "$ac_cv_have_decl_ntohll" >&6; }
+if test "x$ac_cv_have_decl_ntohll" = x""yes; then
+
 cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_NTOHLL $ac_have_decl
+#define HAVE_DECL_NTOHLL 1
 _ACEOF
 
-ac_fn_c_check_decl "$LINENO" "be64toh" "ac_cv_have_decl_be64toh" "#include <arpa/inet.h>
-"
-if test "x$ac_cv_have_decl_be64toh" = xyes; then :
-  ac_have_decl=1
+
 else
-  ac_have_decl=0
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_DECL_NTOHLL 0
+_ACEOF
+
+
 fi
 
-cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_BE64TOH $ac_have_decl
+
+{ $as_echo "$as_me:$LINENO: checking whether be64toh is declared" >&5
+$as_echo_n "checking whether be64toh is declared... " >&6; }
+if test "${ac_cv_have_decl_be64toh+set}" = set; then
+  $as_echo_n "(cached) " >&6
+else
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
 _ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <arpa/inet.h>
+
+int
+main ()
+{
+#ifndef be64toh
+  (void) be64toh;
+#endif
+
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_cv_have_decl_be64toh=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cv_have_decl_be64toh=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_have_decl_be64toh" >&5
+$as_echo "$ac_cv_have_decl_be64toh" >&6; }
+if test "x$ac_cv_have_decl_be64toh" = x""yes; then
+
+cat >>confdefs.h <<_ACEOF
+#define HAVE_DECL_BE64TOH 1
+_ACEOF
+
+
+else
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_DECL_BE64TOH 0
+_ACEOF
+
+
+fi
+
 
 
 # Checks for typedefs, structures, and compiler characteristics.
-ac_fn_c_check_type "$LINENO" "size_t" "ac_cv_type_size_t" "$ac_includes_default"
-if test "x$ac_cv_type_size_t" = xyes; then :
+{ $as_echo "$as_me:$LINENO: checking for size_t" >&5
+$as_echo_n "checking for size_t... " >&6; }
+if test "${ac_cv_type_size_t+set}" = set; then
+  $as_echo_n "(cached) " >&6
+else
+  ac_cv_type_size_t=no
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+if (sizeof (size_t))
+       return 0;
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+if (sizeof ((size_t)))
+	  return 0;
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  :
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
 
+	ac_cv_type_size_t=yes
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_type_size_t" >&5
+$as_echo "$ac_cv_type_size_t" >&6; }
+if test "x$ac_cv_type_size_t" = x""yes; then
+  :
 else
 
 cat >>confdefs.h <<_ACEOF
@@ -5042,12 +5680,75 @@ _ACEOF
 
 fi
 
-ac_fn_c_find_uintX_t "$LINENO" "32" "ac_cv_c_uint32_t"
-case $ac_cv_c_uint32_t in #(
+
+  { $as_echo "$as_me:$LINENO: checking for uint32_t" >&5
+$as_echo_n "checking for uint32_t... " >&6; }
+if test "${ac_cv_c_uint32_t+set}" = set; then
+  $as_echo_n "(cached) " >&6
+else
+  ac_cv_c_uint32_t=no
+     for ac_type in 'uint32_t' 'unsigned int' 'unsigned long int' \
+	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
+       cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+static int test_array [1 - 2 * !(($ac_type) -1 >> (32 - 1) == 1)];
+test_array [0] = 0
+
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  case $ac_type in
+  uint32_t) ac_cv_c_uint32_t=yes ;;
+  *) ac_cv_c_uint32_t=$ac_type ;;
+esac
+
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+       test "$ac_cv_c_uint32_t" != no && break
+     done
+fi
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_uint32_t" >&5
+$as_echo "$ac_cv_c_uint32_t" >&6; }
+  case $ac_cv_c_uint32_t in #(
   no|yes) ;; #(
   *)
 
-$as_echo "#define _UINT32_T 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define _UINT32_T 1
+_ACEOF
 
 
 cat >>confdefs.h <<_ACEOF
@@ -5056,12 +5757,75 @@ _ACEOF
 ;;
   esac
 
-ac_fn_c_find_uintX_t "$LINENO" "64" "ac_cv_c_uint64_t"
-case $ac_cv_c_uint64_t in #(
+
+  { $as_echo "$as_me:$LINENO: checking for uint64_t" >&5
+$as_echo_n "checking for uint64_t... " >&6; }
+if test "${ac_cv_c_uint64_t+set}" = set; then
+  $as_echo_n "(cached) " >&6
+else
+  ac_cv_c_uint64_t=no
+     for ac_type in 'uint64_t' 'unsigned int' 'unsigned long int' \
+	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
+       cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+static int test_array [1 - 2 * !(($ac_type) -1 >> (64 - 1) == 1)];
+test_array [0] = 0
+
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  case $ac_type in
+  uint64_t) ac_cv_c_uint64_t=yes ;;
+  *) ac_cv_c_uint64_t=$ac_type ;;
+esac
+
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+       test "$ac_cv_c_uint64_t" != no && break
+     done
+fi
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_uint64_t" >&5
+$as_echo "$ac_cv_c_uint64_t" >&6; }
+  case $ac_cv_c_uint64_t in #(
   no|yes) ;; #(
   *)
 
-$as_echo "#define _UINT64_T 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define _UINT64_T 1
+_ACEOF
 
 
 cat >>confdefs.h <<_ACEOF
@@ -5072,12 +5836,102 @@ _ACEOF
 
 
 # Checks for library functions.
+
 for ac_func in gettimeofday
-do :
-  ac_fn_c_check_func "$LINENO" "gettimeofday" "ac_cv_func_gettimeofday"
-if test "x$ac_cv_func_gettimeofday" = xyes; then :
+do
+as_ac_var=`$as_echo "ac_cv_func_$ac_func" | $as_tr_sh`
+{ $as_echo "$as_me:$LINENO: checking for $ac_func" >&5
+$as_echo_n "checking for $ac_func... " >&6; }
+if { as_var=$as_ac_var; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+/* Define $ac_func to an innocuous variant, in case <limits.h> declares $ac_func.
+   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
+#define $ac_func innocuous_$ac_func
+
+/* System header to define __stub macros and hopefully few prototypes,
+    which can conflict with char $ac_func (); below.
+    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
+    <limits.h> exists even on freestanding compilers.  */
+
+#ifdef __STDC__
+# include <limits.h>
+#else
+# include <assert.h>
+#endif
+
+#undef $ac_func
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char $ac_func ();
+/* The GNU C library defines this for functions which it implements
+    to always fail with ENOSYS.  Some functions are actually named
+    something starting with __ and the normal name is an alias.  */
+#if defined __stub_$ac_func || defined __stub___$ac_func
+choke me
+#endif
+
+int
+main ()
+{
+return $ac_func ();
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_link") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 $as_test_x conftest$ac_exeext
+       }; then
+  eval "$as_ac_var=yes"
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	eval "$as_ac_var=no"
+fi
+
+rm -rf conftest.dSYM
+rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
+      conftest$ac_exeext conftest.$ac_ext
+fi
+ac_res=`eval 'as_val=${'$as_ac_var'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+as_val=`eval 'as_val=${'$as_ac_var'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_GETTIMEOFDAY 1
+#define `$as_echo "HAVE_$ac_func" | $as_tr_cpp` 1
 _ACEOF
 
 fi
@@ -5085,7 +5939,7 @@ done
 
 
 # Check whether --enable-simd was given.
-if test "${enable_simd+set}" = set; then :
+if test "${enable_simd+set}" = set; then
   enableval=$enable_simd; \
 	ac_SIMD=${enable_simd}
 else
@@ -5097,35 +5951,45 @@ case ${ac_SIMD} in
      SSE4)
        echo Configuring for SSE4
 
-$as_echo "#define SSE4 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define SSE4 1
+_ACEOF
 
      ;;
      AVX)
        echo Configuring for AVX
 
-$as_echo "#define AVX1 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define AVX1 1
+_ACEOF
 
      ;;
      AVX2)
        echo Configuring for AVX2
 
-$as_echo "#define AVX2 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define AVX2 1
+_ACEOF
 
      ;;
      AVX512|MIC)
        echo Configuring for AVX512 and MIC
 
-$as_echo "#define AVX512 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define AVX512 1
+_ACEOF
 
      ;;
      *)
-     as_fn_error $? "${ac_SIMD} unsupported --enable-simd option" "$LINENO" 5;
+     { { $as_echo "$as_me:$LINENO: error: ${ac_SIMD} unsupported --enable-simd option" >&5
+$as_echo "$as_me: error: ${ac_SIMD} unsupported --enable-simd option" >&2;}
+   { (exit 1); exit 1; }; };
      ;;
 esac
 
 
 # Check whether --enable-comms was given.
-if test "${enable_comms+set}" = set; then :
+if test "${enable_comms+set}" = set; then
   enableval=$enable_comms; ac_COMMS=${enable_comms}
 else
   ac_COMMS=none
@@ -5136,17 +6000,23 @@ case ${ac_COMMS} in
      none)
        echo Configuring for NO communications
 
-$as_echo "#define GRID_COMMS_NONE 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define GRID_COMMS_NONE 1
+_ACEOF
 
      ;;
      mpi)
        echo Configuring for MPI communications
 
-$as_echo "#define GRID_COMMS_MPI 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define GRID_COMMS_MPI 1
+_ACEOF
 
      ;;
      *)
-     as_fn_error $? "${ac_COMMS} unsupported --enable-comms option" "$LINENO" 5;
+     { { $as_echo "$as_me:$LINENO: error: ${ac_COMMS} unsupported --enable-comms option" >&5
+$as_echo "$as_me: error: ${ac_COMMS} unsupported --enable-comms option" >&2;}
+   { (exit 1); exit 1; }; };
      ;;
 esac
 
@@ -5203,13 +6073,13 @@ _ACEOF
     case $ac_val in #(
     *${as_nl}*)
       case $ac_var in #(
-      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
+      *_cv_*) { $as_echo "$as_me:$LINENO: WARNING: cache variable $ac_var contains a newline" >&5
 $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
       esac
       case $ac_var in #(
       _ | IFS | as_nl) ;; #(
       BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
-      *) { eval $ac_var=; unset $ac_var;} ;;
+      *) $as_unset $ac_var ;;
       esac ;;
     esac
   done
@@ -5217,8 +6087,8 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
   (set) 2>&1 |
     case $as_nl`(ac_space=' '; set) 2>&1` in #(
     *${as_nl}ac_space=\ *)
-      # `set' does not quote correctly, so add quotes: double-quote
-      # substitution turns \\\\ into \\, and sed turns \\ into \.
+      # `set' does not quote correctly, so add quotes (double-quote
+      # substitution turns \\\\ into \\, and sed turns \\ into \).
       sed -n \
 	"s/'/'\\\\''/g;
 	  s/^\\([_$as_cr_alnum]*_cv_[_$as_cr_alnum]*\\)=\\(.*\\)/\\1='\\2'/p"
@@ -5240,23 +6110,12 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
      :end' >>confcache
 if diff "$cache_file" confcache >/dev/null 2>&1; then :; else
   if test -w "$cache_file"; then
-    if test "x$cache_file" != "x/dev/null"; then
-      { $as_echo "$as_me:${as_lineno-$LINENO}: updating cache $cache_file" >&5
+    test "x$cache_file" != "x/dev/null" &&
+      { $as_echo "$as_me:$LINENO: updating cache $cache_file" >&5
 $as_echo "$as_me: updating cache $cache_file" >&6;}
-      if test ! -f "$cache_file" || test -h "$cache_file"; then
-	cat confcache >"$cache_file"
-      else
-        case $cache_file in #(
-        */* | ?:*)
-	  mv -f confcache "$cache_file"$$ &&
-	  mv -f "$cache_file"$$ "$cache_file" ;; #(
-        *)
-	  mv -f confcache "$cache_file" ;;
-	esac
-      fi
-    fi
+    cat confcache >$cache_file
   else
-    { $as_echo "$as_me:${as_lineno-$LINENO}: not updating unwritable cache $cache_file" >&5
+    { $as_echo "$as_me:$LINENO: not updating unwritable cache $cache_file" >&5
 $as_echo "$as_me: not updating unwritable cache $cache_file" >&6;}
   fi
 fi
@@ -5270,29 +6129,20 @@ DEFS=-DHAVE_CONFIG_H
 
 ac_libobjs=
 ac_ltlibobjs=
-U=
 for ac_i in : $LIBOBJS; do test "x$ac_i" = x: && continue
   # 1. Remove the extension, and $U if already installed.
   ac_script='s/\$U\././;s/\.o$//;s/\.obj$//'
   ac_i=`$as_echo "$ac_i" | sed "$ac_script"`
   # 2. Prepend LIBOBJDIR.  When used with automake>=1.10 LIBOBJDIR
   #    will be set to the directory where LIBOBJS objects are built.
-  as_fn_append ac_libobjs " \${LIBOBJDIR}$ac_i\$U.$ac_objext"
-  as_fn_append ac_ltlibobjs " \${LIBOBJDIR}$ac_i"'$U.lo'
+  ac_libobjs="$ac_libobjs \${LIBOBJDIR}$ac_i\$U.$ac_objext"
+  ac_ltlibobjs="$ac_ltlibobjs \${LIBOBJDIR}$ac_i"'$U.lo'
 done
 LIBOBJS=$ac_libobjs
 
 LTLIBOBJS=$ac_ltlibobjs
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking that generated files are newer than configure" >&5
-$as_echo_n "checking that generated files are newer than configure... " >&6; }
-   if test -n "$am_sleep_pid"; then
-     # Hide warnings about reused PIDs.
-     wait $am_sleep_pid 2>/dev/null
-   fi
-   { $as_echo "$as_me:${as_lineno-$LINENO}: result: done" >&5
-$as_echo "done" >&6; }
  if test -n "$EXEEXT"; then
   am__EXEEXT_TRUE=
   am__EXEEXT_FALSE='#'
@@ -5302,34 +6152,48 @@ else
 fi
 
 if test -z "${AMDEP_TRUE}" && test -z "${AMDEP_FALSE}"; then
-  as_fn_error $? "conditional \"AMDEP\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: conditional \"AMDEP\" was never defined.
+Usually this means the macro was only invoked conditionally." >&5
+$as_echo "$as_me: error: conditional \"AMDEP\" was never defined.
+Usually this means the macro was only invoked conditionally." >&2;}
+   { (exit 1); exit 1; }; }
 fi
 if test -z "${am__fastdepCXX_TRUE}" && test -z "${am__fastdepCXX_FALSE}"; then
-  as_fn_error $? "conditional \"am__fastdepCXX\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: conditional \"am__fastdepCXX\" was never defined.
+Usually this means the macro was only invoked conditionally." >&5
+$as_echo "$as_me: error: conditional \"am__fastdepCXX\" was never defined.
+Usually this means the macro was only invoked conditionally." >&2;}
+   { (exit 1); exit 1; }; }
 fi
 if test -z "${am__fastdepCC_TRUE}" && test -z "${am__fastdepCC_FALSE}"; then
-  as_fn_error $? "conditional \"am__fastdepCC\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: conditional \"am__fastdepCC\" was never defined.
+Usually this means the macro was only invoked conditionally." >&5
+$as_echo "$as_me: error: conditional \"am__fastdepCC\" was never defined.
+Usually this means the macro was only invoked conditionally." >&2;}
+   { (exit 1); exit 1; }; }
 fi
 if test -z "${BUILD_COMMS_MPI_TRUE}" && test -z "${BUILD_COMMS_MPI_FALSE}"; then
-  as_fn_error $? "conditional \"BUILD_COMMS_MPI\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: conditional \"BUILD_COMMS_MPI\" was never defined.
+Usually this means the macro was only invoked conditionally." >&5
+$as_echo "$as_me: error: conditional \"BUILD_COMMS_MPI\" was never defined.
+Usually this means the macro was only invoked conditionally." >&2;}
+   { (exit 1); exit 1; }; }
 fi
 if test -z "${BUILD_COMMS_NONE_TRUE}" && test -z "${BUILD_COMMS_NONE_FALSE}"; then
-  as_fn_error $? "conditional \"BUILD_COMMS_NONE\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: conditional \"BUILD_COMMS_NONE\" was never defined.
+Usually this means the macro was only invoked conditionally." >&5
+$as_echo "$as_me: error: conditional \"BUILD_COMMS_NONE\" was never defined.
+Usually this means the macro was only invoked conditionally." >&2;}
+   { (exit 1); exit 1; }; }
 fi
 
-: "${CONFIG_STATUS=./config.status}"
+: ${CONFIG_STATUS=./config.status}
 ac_write_fail=0
 ac_clean_files_save=$ac_clean_files
 ac_clean_files="$ac_clean_files $CONFIG_STATUS"
-{ $as_echo "$as_me:${as_lineno-$LINENO}: creating $CONFIG_STATUS" >&5
+{ $as_echo "$as_me:$LINENO: creating $CONFIG_STATUS" >&5
 $as_echo "$as_me: creating $CONFIG_STATUS" >&6;}
-as_write_fail=0
-cat >$CONFIG_STATUS <<_ASEOF || as_write_fail=1
+cat >$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 #! $SHELL
 # Generated by $as_me.
 # Run this file to recreate the current configuration.
@@ -5339,18 +6203,17 @@ cat >$CONFIG_STATUS <<_ASEOF || as_write_fail=1
 debug=false
 ac_cs_recheck=false
 ac_cs_silent=false
-
 SHELL=\${CONFIG_SHELL-$SHELL}
-export SHELL
-_ASEOF
-cat >>$CONFIG_STATUS <<\_ASEOF || as_write_fail=1
-## -------------------- ##
-## M4sh Initialization. ##
-## -------------------- ##
+_ACEOF
+
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+## --------------------- ##
+## M4sh Initialization.  ##
+## --------------------- ##
 
 # Be more Bourne compatible
 DUALCASE=1; export DUALCASE # for MKS sh
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
   emulate sh
   NULLCMD=:
   # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
@@ -5358,15 +6221,23 @@ if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
   alias -g '${1+"$@"}'='"$@"'
   setopt NO_GLOB_SUBST
 else
-  case `(set -o) 2>/dev/null` in #(
-  *posix*) :
-    set -o posix ;; #(
-  *) :
-     ;;
+  case `(set -o) 2>/dev/null` in
+  *posix*) set -o posix ;;
 esac
+
 fi
 
 
+
+
+# PATH needs CR
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
+
 as_nl='
 '
 export as_nl
@@ -5374,13 +6245,7 @@ export as_nl
 as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
-# Prefer a ksh shell builtin over an external printf program on Solaris,
-# but without wasting forks for bash or zsh.
-if test -z "$BASH_VERSION$ZSH_VERSION" \
-    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
-  as_echo='print -r --'
-  as_echo_n='print -rn --'
-elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+if (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
   as_echo='printf %s\n'
   as_echo_n='printf %s'
 else
@@ -5391,7 +6256,7 @@ else
     as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
     as_echo_n_body='eval
       arg=$1;
-      case $arg in #(
+      case $arg in
       *"$as_nl"*)
 	expr "X$arg" : "X\\(.*\\)$as_nl";
 	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
@@ -5414,6 +6279,13 @@ if test "${PATH_SEPARATOR+set}" != set; then
   }
 fi
 
+# Support unset when possible.
+if ( (MAIL=60; unset MAIL) || exit) >/dev/null 2>&1; then
+  as_unset=unset
+else
+  as_unset=false
+fi
+
 
 # IFS
 # We need space, tab and new line, in precisely that order.  Quoting is
@@ -5423,16 +6295,15 @@ fi
 IFS=" ""	$as_nl"
 
 # Find who we are.  Look in the path if we contain no directory separator.
-as_myself=
-case $0 in #((
+case $0 in
   *[\\/]* ) as_myself=$0 ;;
   *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
 for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
-  done
+  test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+done
 IFS=$as_save_IFS
 
      ;;
@@ -5444,16 +6315,12 @@ if test "x$as_myself" = x; then
 fi
 if test ! -f "$as_myself"; then
   $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
-  exit 1
+  { (exit 1); exit 1; }
 fi
 
-# Unset variables that we do not need and which cause bugs (e.g. in
-# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
-# suppresses any "Segmentation fault" message there.  '((' could
-# trigger a bug in pdksh 5.2.14.
-for as_var in BASH_ENV ENV MAIL MAILPATH
-do eval test x\${$as_var+set} = xset \
-  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
+# Work around bugs in pre-3.0 UWIN ksh.
+for as_var in ENV MAIL MAILPATH
+do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
 done
 PS1='$ '
 PS2='> '
@@ -5465,89 +6332,7 @@ export LC_ALL
 LANGUAGE=C
 export LANGUAGE
 
-# CDPATH.
-(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
-
-
-# as_fn_error STATUS ERROR [LINENO LOG_FD]
-# ----------------------------------------
-# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
-# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
-# script with STATUS, using 1 if that was 0.
-as_fn_error ()
-{
-  as_status=$1; test $as_status -eq 0 && as_status=1
-  if test "$4"; then
-    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
-  fi
-  $as_echo "$as_me: error: $2" >&2
-  as_fn_exit $as_status
-} # as_fn_error
-
-
-# as_fn_set_status STATUS
-# -----------------------
-# Set $? to STATUS, without forking.
-as_fn_set_status ()
-{
-  return $1
-} # as_fn_set_status
-
-# as_fn_exit STATUS
-# -----------------
-# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
-as_fn_exit ()
-{
-  set +e
-  as_fn_set_status $1
-  exit $1
-} # as_fn_exit
-
-# as_fn_unset VAR
-# ---------------
-# Portably unset VAR.
-as_fn_unset ()
-{
-  { eval $1=; unset $1;}
-}
-as_unset=as_fn_unset
-# as_fn_append VAR VALUE
-# ----------------------
-# Append the text in VALUE to the end of the definition contained in VAR. Take
-# advantage of any shell optimizations that allow amortized linear growth over
-# repeated appends, instead of the typical quadratic growth present in naive
-# implementations.
-if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
-  eval 'as_fn_append ()
-  {
-    eval $1+=\$2
-  }'
-else
-  as_fn_append ()
-  {
-    eval $1=\$$1\$2
-  }
-fi # as_fn_append
-
-# as_fn_arith ARG...
-# ------------------
-# Perform arithmetic evaluation on the ARGs, and store the result in the
-# global $as_val. Take advantage of shells that can avoid forks. The arguments
-# must be portable across $(()) and expr.
-if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
-  eval 'as_fn_arith ()
-  {
-    as_val=$(( $* ))
-  }'
-else
-  as_fn_arith ()
-  {
-    as_val=`expr "$@" || test $? -eq 1`
-  }
-fi # as_fn_arith
-
-
+# Required to use basename.
 if expr a : '\(a\)' >/dev/null 2>&1 &&
    test "X`expr 00001 : '.*\(...\)'`" = X001; then
   as_expr=expr
@@ -5561,12 +6346,8 @@ else
   as_basename=false
 fi
 
-if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
-  as_dirname=dirname
-else
-  as_dirname=false
-fi
 
+# Name of the executable.
 as_me=`$as_basename -- "$0" ||
 $as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
 	 X"$0" : 'X\(//\)$' \| \
@@ -5586,25 +6367,76 @@ $as_echo X/"$0" |
 	  }
 	  s/.*/./; q'`
 
-# Avoid depending upon Character Ranges.
-as_cr_letters='abcdefghijklmnopqrstuvwxyz'
-as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
-as_cr_Letters=$as_cr_letters$as_cr_LETTERS
-as_cr_digits='0123456789'
-as_cr_alnum=$as_cr_Letters$as_cr_digits
+# CDPATH.
+$as_unset CDPATH
+
+
+
+  as_lineno_1=$LINENO
+  as_lineno_2=$LINENO
+  test "x$as_lineno_1" != "x$as_lineno_2" &&
+  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2" || {
+
+  # Create $as_me.lineno as a copy of $as_myself, but with $LINENO
+  # uniformly replaced by the line number.  The first 'sed' inserts a
+  # line-number line after each line using $LINENO; the second 'sed'
+  # does the real work.  The second script uses 'N' to pair each
+  # line-number line with the line containing $LINENO, and appends
+  # trailing '-' during substitution so that $LINENO is not a special
+  # case at line end.
+  # (Raja R Harinath suggested sed '=', and Paul Eggert wrote the
+  # scripts with optimization help from Paolo Bonzini.  Blame Lee
+  # E. McMahon (1931-1989) for sed's syntax.  :-)
+  sed -n '
+    p
+    /[$]LINENO/=
+  ' <$as_myself |
+    sed '
+      s/[$]LINENO.*/&-/
+      t lineno
+      b
+      :lineno
+      N
+      :loop
+      s/[$]LINENO\([^'$as_cr_alnum'_].*\n\)\(.*\)/\2\1\2/
+      t loop
+      s/-\n.*//
+    ' >$as_me.lineno &&
+  chmod +x "$as_me.lineno" ||
+    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2
+   { (exit 1); exit 1; }; }
+
+  # Don't try to exec as it changes $[0], causing all sort of problems
+  # (the dirname of $[0] is not the place where we might find the
+  # original and so on.  Autoconf is especially sensitive to this).
+  . "./$as_me.lineno"
+  # Exit status is that of the last command.
+  exit
+}
+
+
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
 
 ECHO_C= ECHO_N= ECHO_T=
-case `echo -n x` in #(((((
+case `echo -n x` in
 -n*)
-  case `echo 'xy\c'` in
+  case `echo 'x\c'` in
   *c*) ECHO_T='	';;	# ECHO_T is single tab character.
-  xy)  ECHO_C='\c';;
-  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
-       ECHO_T='	';;
+  *)   ECHO_C='\c';;
   esac;;
 *)
   ECHO_N='-n';;
 esac
+if expr a : '\(a\)' >/dev/null 2>&1 &&
+   test "X`expr 00001 : '.*\(...\)'`" = X001; then
+  as_expr=expr
+else
+  as_expr=false
+fi
 
 rm -f conf$$ conf$$.exe conf$$.file
 if test -d conf$$.dir; then
@@ -5619,85 +6451,49 @@ if (echo >conf$$.file) 2>/dev/null; then
     # ... but there are two gotchas:
     # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
     # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
-    # In both cases, we have to default to `cp -pR'.
+    # In both cases, we have to default to `cp -p'.
     ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
-      as_ln_s='cp -pR'
+      as_ln_s='cp -p'
   elif ln conf$$.file conf$$ 2>/dev/null; then
     as_ln_s=ln
   else
-    as_ln_s='cp -pR'
+    as_ln_s='cp -p'
   fi
 else
-  as_ln_s='cp -pR'
+  as_ln_s='cp -p'
 fi
 rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
 rmdir conf$$.dir 2>/dev/null
 
-
-# as_fn_mkdir_p
-# -------------
-# Create "$as_dir" as a directory, including parents if necessary.
-as_fn_mkdir_p ()
-{
-
-  case $as_dir in #(
-  -*) as_dir=./$as_dir;;
-  esac
-  test -d "$as_dir" || eval $as_mkdir_p || {
-    as_dirs=
-    while :; do
-      case $as_dir in #(
-      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
-      *) as_qdir=$as_dir;;
-      esac
-      as_dirs="'$as_qdir' $as_dirs"
-      as_dir=`$as_dirname -- "$as_dir" ||
-$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_dir" : 'X\(//\)[^/]' \| \
-	 X"$as_dir" : 'X\(//\)$' \| \
-	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_dir" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      test -d "$as_dir" && break
-    done
-    test -z "$as_dirs" || eval "mkdir $as_dirs"
-  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
-
-
-} # as_fn_mkdir_p
 if mkdir -p . 2>/dev/null; then
-  as_mkdir_p='mkdir -p "$as_dir"'
+  as_mkdir_p=:
 else
   test -d ./-p && rmdir ./-p
   as_mkdir_p=false
 fi
 
-
-# as_fn_executable_p FILE
-# -----------------------
-# Test if FILE is an executable regular file.
-as_fn_executable_p ()
-{
-  test -f "$1" && test -x "$1"
-} # as_fn_executable_p
-as_test_x='test -x'
-as_executable_p=as_fn_executable_p
+if test -x / >/dev/null 2>&1; then
+  as_test_x='test -x'
+else
+  if ls -dL / >/dev/null 2>&1; then
+    as_ls_L_option=L
+  else
+    as_ls_L_option=
+  fi
+  as_test_x='
+    eval sh -c '\''
+      if test -d "$1"; then
+	test -d "$1/.";
+      else
+	case $1 in
+	-*)set "./$1";;
+	esac;
+	case `ls -ld'$as_ls_L_option' "$1" 2>/dev/null` in
+	???[sx]*):;;*)false;;esac;fi
+    '\'' sh
+  '
+fi
+as_executable_p=$as_test_x
 
 # Sed expression to map a string onto a valid CPP name.
 as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
@@ -5707,19 +6503,13 @@ as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
 
 
 exec 6>&1
-## ----------------------------------- ##
-## Main body of $CONFIG_STATUS script. ##
-## ----------------------------------- ##
-_ASEOF
-test $as_write_fail = 0 && chmod +x $CONFIG_STATUS || ac_write_fail=1
 
-cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-# Save the log message, to keep $0 and so on meaningful, and to
+# Save the log message, to keep $[0] and so on meaningful, and to
 # report actual input values of CONFIG_FILES etc. instead of their
 # values after options handling.
 ac_log="
 This file was extended by Grid $as_me 1.0, which was
-generated by GNU Autoconf 2.69.  Invocation command line was
+generated by GNU Autoconf 2.63.  Invocation command line was
 
   CONFIG_FILES    = $CONFIG_FILES
   CONFIG_HEADERS  = $CONFIG_HEADERS
@@ -5751,15 +6541,13 @@ _ACEOF
 
 cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 ac_cs_usage="\
-\`$as_me' instantiates files and other configuration actions
-from templates according to the current configuration.  Unless the files
-and actions are specified as TAGs, all are instantiated by default.
+\`$as_me' instantiates files from templates according to the
+current configuration.
 
-Usage: $0 [OPTION]... [TAG]...
+Usage: $0 [OPTION]... [FILE]...
 
   -h, --help       print this help, then exit
   -V, --version    print version number and configuration settings, then exit
-      --config     print configuration, then exit
   -q, --quiet, --silent
                    do not print progress messages
   -d, --debug      don't remove temporary files
@@ -5778,17 +6566,16 @@ $config_headers
 Configuration commands:
 $config_commands
 
-Report bugs to <paboyle@ph.ed.ac.uk>."
+Report bugs to <bug-autoconf@gnu.org>."
 
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-ac_cs_config="`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`"
 ac_cs_version="\\
 Grid config.status 1.0
-configured by $0, generated by GNU Autoconf 2.69,
-  with options \\"\$ac_cs_config\\"
+configured by $0, generated by GNU Autoconf 2.63,
+  with options \\"`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`\\"
 
-Copyright (C) 2012 Free Software Foundation, Inc.
+Copyright (C) 2008 Free Software Foundation, Inc.
 This config.status script is free software; the Free Software Foundation
 gives unlimited permission to copy, distribute and modify it."
 
@@ -5806,16 +6593,11 @@ ac_need_defaults=:
 while test $# != 0
 do
   case $1 in
-  --*=?*)
+  --*=*)
     ac_option=`expr "X$1" : 'X\([^=]*\)='`
     ac_optarg=`expr "X$1" : 'X[^=]*=\(.*\)'`
     ac_shift=:
     ;;
-  --*=)
-    ac_option=`expr "X$1" : 'X\([^=]*\)='`
-    ac_optarg=
-    ac_shift=:
-    ;;
   *)
     ac_option=$1
     ac_optarg=$2
@@ -5829,29 +6611,27 @@ do
     ac_cs_recheck=: ;;
   --version | --versio | --versi | --vers | --ver | --ve | --v | -V )
     $as_echo "$ac_cs_version"; exit ;;
-  --config | --confi | --conf | --con | --co | --c )
-    $as_echo "$ac_cs_config"; exit ;;
   --debug | --debu | --deb | --de | --d | -d )
     debug=: ;;
   --file | --fil | --fi | --f )
     $ac_shift
     case $ac_optarg in
     *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
-    '') as_fn_error $? "missing file argument" ;;
     esac
-    as_fn_append CONFIG_FILES " '$ac_optarg'"
+    CONFIG_FILES="$CONFIG_FILES '$ac_optarg'"
     ac_need_defaults=false;;
   --header | --heade | --head | --hea )
     $ac_shift
     case $ac_optarg in
     *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
     esac
-    as_fn_append CONFIG_HEADERS " '$ac_optarg'"
+    CONFIG_HEADERS="$CONFIG_HEADERS '$ac_optarg'"
     ac_need_defaults=false;;
   --he | --h)
     # Conflict between --help and --header
-    as_fn_error $? "ambiguous option: \`$1'
-Try \`$0 --help' for more information.";;
+    { $as_echo "$as_me: error: ambiguous option: $1
+Try \`$0 --help' for more information." >&2
+   { (exit 1); exit 1; }; };;
   --help | --hel | -h )
     $as_echo "$ac_cs_usage"; exit ;;
   -q | -quiet | --quiet | --quie | --qui | --qu | --q \
@@ -5859,10 +6639,11 @@ Try \`$0 --help' for more information.";;
     ac_cs_silent=: ;;
 
   # This is an error.
-  -*) as_fn_error $? "unrecognized option: \`$1'
-Try \`$0 --help' for more information." ;;
+  -*) { $as_echo "$as_me: error: unrecognized option: $1
+Try \`$0 --help' for more information." >&2
+   { (exit 1); exit 1; }; } ;;
 
-  *) as_fn_append ac_config_targets " $1"
+  *) ac_config_targets="$ac_config_targets $1"
      ac_need_defaults=false ;;
 
   esac
@@ -5879,7 +6660,7 @@ fi
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 if \$ac_cs_recheck; then
-  set X $SHELL '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
+  set X '$SHELL' '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
   shift
   \$as_echo "running CONFIG_SHELL=$SHELL \$*" >&6
   CONFIG_SHELL='$SHELL'
@@ -5920,7 +6701,9 @@ do
     "tests/Makefile") CONFIG_FILES="$CONFIG_FILES tests/Makefile" ;;
     "benchmarks/Makefile") CONFIG_FILES="$CONFIG_FILES benchmarks/Makefile" ;;
 
-  *) as_fn_error $? "invalid argument: \`$ac_config_target'" "$LINENO" 5;;
+  *) { { $as_echo "$as_me:$LINENO: error: invalid argument: $ac_config_target" >&5
+$as_echo "$as_me: error: invalid argument: $ac_config_target" >&2;}
+   { (exit 1); exit 1; }; };;
   esac
 done
 
@@ -5943,24 +6726,26 @@ fi
 # after its creation but before its name has been assigned to `$tmp'.
 $debug ||
 {
-  tmp= ac_tmp=
+  tmp=
   trap 'exit_status=$?
-  : "${ac_tmp:=$tmp}"
-  { test ! -d "$ac_tmp" || rm -fr "$ac_tmp"; } && exit $exit_status
+  { test -z "$tmp" || test ! -d "$tmp" || rm -fr "$tmp"; } && exit $exit_status
 ' 0
-  trap 'as_fn_exit 1' 1 2 13 15
+  trap '{ (exit 1); exit 1; }' 1 2 13 15
 }
 # Create a (secure) tmp directory for tmp files.
 
 {
   tmp=`(umask 077 && mktemp -d "./confXXXXXX") 2>/dev/null` &&
-  test -d "$tmp"
+  test -n "$tmp" && test -d "$tmp"
 }  ||
 {
   tmp=./conf$$-$RANDOM
   (umask 077 && mkdir "$tmp")
-} || as_fn_error $? "cannot create a temporary directory in ." "$LINENO" 5
-ac_tmp=$tmp
+} ||
+{
+   $as_echo "$as_me: cannot create a temporary directory in ." >&2
+   { (exit 1); exit 1; }
+}
 
 # Set up the scripts for CONFIG_FILES section.
 # No need to generate them if there are no CONFIG_FILES.
@@ -5968,13 +6753,7 @@ ac_tmp=$tmp
 if test -n "$CONFIG_FILES"; then
 
 
-ac_cr=`echo X | tr X '\015'`
-# On cygwin, bash can eat \r inside `` if the user requested igncr.
-# But we know of no other shell where ac_cr would be empty at this
-# point, so we can use a bashism as a fallback.
-if test "x$ac_cr" = x; then
-  eval ac_cr=\$\'\\r\'
-fi
+ac_cr='
'
 ac_cs_awk_cr=`$AWK 'BEGIN { print "a\rb" }' </dev/null 2>/dev/null`
 if test "$ac_cs_awk_cr" = "a${ac_cr}b"; then
   ac_cs_awk_cr='\\r'
@@ -5982,7 +6761,7 @@ else
   ac_cs_awk_cr=$ac_cr
 fi
 
-echo 'BEGIN {' >"$ac_tmp/subs1.awk" &&
+echo 'BEGIN {' >"$tmp/subs1.awk" &&
 _ACEOF
 
 
@@ -5991,18 +6770,24 @@ _ACEOF
   echo "$ac_subst_vars" | sed 's/.*/&!$&$ac_delim/' &&
   echo "_ACEOF"
 } >conf$$subs.sh ||
-  as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
-ac_delim_num=`echo "$ac_subst_vars" | grep -c '^'`
+  { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
+$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
+   { (exit 1); exit 1; }; }
+ac_delim_num=`echo "$ac_subst_vars" | grep -c '$'`
 ac_delim='%!_!# '
 for ac_last_try in false false false false false :; do
   . ./conf$$subs.sh ||
-    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
+$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
+   { (exit 1); exit 1; }; }
 
   ac_delim_n=`sed -n "s/.*$ac_delim\$/X/p" conf$$subs.awk | grep -c X`
   if test $ac_delim_n = $ac_delim_num; then
     break
   elif $ac_last_try; then
-    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
+$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
+   { (exit 1); exit 1; }; }
   else
     ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
   fi
@@ -6010,7 +6795,7 @@ done
 rm -f conf$$subs.sh
 
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-cat >>"\$ac_tmp/subs1.awk" <<\\_ACAWK &&
+cat >>"\$tmp/subs1.awk" <<\\_ACAWK &&
 _ACEOF
 sed -n '
 h
@@ -6024,7 +6809,7 @@ s/'"$ac_delim"'$//
 t delim
 :nl
 h
-s/\(.\{148\}\)..*/\1/
+s/\(.\{148\}\).*/\1/
 t more1
 s/["\\]/\\&/g; s/^/"/; s/$/\\n"\\/
 p
@@ -6038,7 +6823,7 @@ s/.\{148\}//
 t nl
 :delim
 h
-s/\(.\{148\}\)..*/\1/
+s/\(.\{148\}\).*/\1/
 t more2
 s/["\\]/\\&/g; s/^/"/; s/$/"/
 p
@@ -6058,7 +6843,7 @@ t delim
 rm -f conf$$subs.awk
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 _ACAWK
-cat >>"\$ac_tmp/subs1.awk" <<_ACAWK &&
+cat >>"\$tmp/subs1.awk" <<_ACAWK &&
   for (key in S) S_is_set[key] = 1
   FS = ""
 
@@ -6090,29 +6875,23 @@ if sed "s/$ac_cr//" < /dev/null > /dev/null 2>&1; then
   sed "s/$ac_cr\$//; s/$ac_cr/$ac_cs_awk_cr/g"
 else
   cat
-fi < "$ac_tmp/subs1.awk" > "$ac_tmp/subs.awk" \
-  || as_fn_error $? "could not setup config files machinery" "$LINENO" 5
+fi < "$tmp/subs1.awk" > "$tmp/subs.awk" \
+  || { { $as_echo "$as_me:$LINENO: error: could not setup config files machinery" >&5
+$as_echo "$as_me: error: could not setup config files machinery" >&2;}
+   { (exit 1); exit 1; }; }
 _ACEOF
 
-# VPATH may cause trouble with some makes, so we remove sole $(srcdir),
-# ${srcdir} and @srcdir@ entries from VPATH if srcdir is ".", strip leading and
+# VPATH may cause trouble with some makes, so we remove $(srcdir),
+# ${srcdir} and @srcdir@ from VPATH if srcdir is ".", strip leading and
 # trailing colons and then remove the whole line if VPATH becomes empty
 # (actually we leave an empty line to preserve line numbers).
 if test "x$srcdir" = x.; then
-  ac_vpsub='/^[	 ]*VPATH[	 ]*=[	 ]*/{
-h
-s///
-s/^/:/
-s/[	 ]*$/:/
-s/:\$(srcdir):/:/g
-s/:\${srcdir}:/:/g
-s/:@srcdir@:/:/g
-s/^:*//
+  ac_vpsub='/^[	 ]*VPATH[	 ]*=/{
+s/:*\$(srcdir):*/:/
+s/:*\${srcdir}:*/:/
+s/:*@srcdir@:*/:/
+s/^\([^=]*=[	 ]*\):*/\1/
 s/:*$//
-x
-s/\(=[	 ]*\).*/\1/
-G
-s/\n//
 s/^[^=]*=[	 ]*$//
 }'
 fi
@@ -6124,7 +6903,7 @@ fi # test -n "$CONFIG_FILES"
 # No need to generate them if there are no CONFIG_HEADERS.
 # This happens for instance with `./config.status Makefile'.
 if test -n "$CONFIG_HEADERS"; then
-cat >"$ac_tmp/defines.awk" <<\_ACAWK ||
+cat >"$tmp/defines.awk" <<\_ACAWK ||
 BEGIN {
 _ACEOF
 
@@ -6136,11 +6915,13 @@ _ACEOF
 # handling of long lines.
 ac_delim='%!_!# '
 for ac_last_try in false false :; do
-  ac_tt=`sed -n "/$ac_delim/p" confdefs.h`
-  if test -z "$ac_tt"; then
+  ac_t=`sed -n "/$ac_delim/p" confdefs.h`
+  if test -z "$ac_t"; then
     break
   elif $ac_last_try; then
-    as_fn_error $? "could not make $CONFIG_HEADERS" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_HEADERS" >&5
+$as_echo "$as_me: error: could not make $CONFIG_HEADERS" >&2;}
+   { (exit 1); exit 1; }; }
   else
     ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
   fi
@@ -6225,7 +7006,9 @@ cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 _ACAWK
 _ACEOF
 cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-  as_fn_error $? "could not setup config headers machinery" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: could not setup config headers machinery" >&5
+$as_echo "$as_me: error: could not setup config headers machinery" >&2;}
+   { (exit 1); exit 1; }; }
 fi # test -n "$CONFIG_HEADERS"
 
 
@@ -6238,7 +7021,9 @@ do
   esac
   case $ac_mode$ac_tag in
   :[FHL]*:*);;
-  :L* | :C*:*) as_fn_error $? "invalid tag \`$ac_tag'" "$LINENO" 5;;
+  :L* | :C*:*) { { $as_echo "$as_me:$LINENO: error: invalid tag $ac_tag" >&5
+$as_echo "$as_me: error: invalid tag $ac_tag" >&2;}
+   { (exit 1); exit 1; }; };;
   :[FH]-) ac_tag=-:-;;
   :[FH]*) ac_tag=$ac_tag:$ac_tag.in;;
   esac
@@ -6257,7 +7042,7 @@ do
     for ac_f
     do
       case $ac_f in
-      -) ac_f="$ac_tmp/stdin";;
+      -) ac_f="$tmp/stdin";;
       *) # Look for the file first in the build tree, then in the source tree
 	 # (if the path is not absolute).  The absolute path cannot be DOS-style,
 	 # because $ac_f cannot contain `:'.
@@ -6266,10 +7051,12 @@ do
 	   [\\/$]*) false;;
 	   *) test -f "$srcdir/$ac_f" && ac_f="$srcdir/$ac_f";;
 	   esac ||
-	   as_fn_error 1 "cannot find input file: \`$ac_f'" "$LINENO" 5;;
+	   { { $as_echo "$as_me:$LINENO: error: cannot find input file: $ac_f" >&5
+$as_echo "$as_me: error: cannot find input file: $ac_f" >&2;}
+   { (exit 1); exit 1; }; };;
       esac
       case $ac_f in *\'*) ac_f=`$as_echo "$ac_f" | sed "s/'/'\\\\\\\\''/g"`;; esac
-      as_fn_append ac_file_inputs " '$ac_f'"
+      ac_file_inputs="$ac_file_inputs '$ac_f'"
     done
 
     # Let's still pretend it is `configure' which instantiates (i.e., don't
@@ -6280,7 +7067,7 @@ do
 	`' by configure.'
     if test x"$ac_file" != x-; then
       configure_input="$ac_file.  $configure_input"
-      { $as_echo "$as_me:${as_lineno-$LINENO}: creating $ac_file" >&5
+      { $as_echo "$as_me:$LINENO: creating $ac_file" >&5
 $as_echo "$as_me: creating $ac_file" >&6;}
     fi
     # Neutralize special characters interpreted by sed in replacement strings.
@@ -6292,8 +7079,10 @@ $as_echo "$as_me: creating $ac_file" >&6;}
     esac
 
     case $ac_tag in
-    *:-:* | *:-) cat >"$ac_tmp/stdin" \
-      || as_fn_error $? "could not create $ac_file" "$LINENO" 5 ;;
+    *:-:* | *:-) cat >"$tmp/stdin" \
+      || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; } ;;
     esac
     ;;
   esac
@@ -6321,7 +7110,47 @@ $as_echo X"$ac_file" |
 	    q
 	  }
 	  s/.*/./; q'`
-  as_dir="$ac_dir"; as_fn_mkdir_p
+  { as_dir="$ac_dir"
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || { $as_mkdir_p && mkdir -p "$as_dir"; } || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || { { $as_echo "$as_me:$LINENO: error: cannot create directory $as_dir" >&5
+$as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
+   { (exit 1); exit 1; }; }; }
   ac_builddir=.
 
 case "$ac_dir" in
@@ -6378,6 +7207,7 @@ cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 # If the template does not know about datarootdir, expand it.
 # FIXME: This hack should be removed a few years after 2.60.
 ac_datarootdir_hack=; ac_datarootdir_seen=
+
 ac_sed_dataroot='
 /datarootdir/ {
   p
@@ -6387,11 +7217,12 @@ ac_sed_dataroot='
 /@docdir@/p
 /@infodir@/p
 /@localedir@/p
-/@mandir@/p'
+/@mandir@/p
+'
 case `eval "sed -n \"\$ac_sed_dataroot\" $ac_file_inputs"` in
 *datarootdir*) ac_datarootdir_seen=yes;;
 *@datadir@*|*@docdir@*|*@infodir@*|*@localedir@*|*@mandir@*)
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
+  { $as_echo "$as_me:$LINENO: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
 $as_echo "$as_me: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&2;}
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
@@ -6401,7 +7232,7 @@ cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
   s&@infodir@&$infodir&g
   s&@localedir@&$localedir&g
   s&@mandir@&$mandir&g
-  s&\\\${datarootdir}&$datarootdir&g' ;;
+    s&\\\${datarootdir}&$datarootdir&g' ;;
 esac
 _ACEOF
 
@@ -6429,24 +7260,27 @@ s&@INSTALL@&$ac_INSTALL&;t t
 s&@MKDIR_P@&$ac_MKDIR_P&;t t
 $ac_datarootdir_hack
 "
-eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$ac_tmp/subs.awk" \
-  >$ac_tmp/out || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$tmp/subs.awk" >$tmp/out \
+  || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; }
 
 test -z "$ac_datarootdir_hack$ac_datarootdir_seen" &&
-  { ac_out=`sed -n '/\${datarootdir}/p' "$ac_tmp/out"`; test -n "$ac_out"; } &&
-  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' \
-      "$ac_tmp/out"`; test -z "$ac_out"; } &&
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file contains a reference to the variable \`datarootdir'
-which seems to be undefined.  Please make sure it is defined" >&5
+  { ac_out=`sed -n '/\${datarootdir}/p' "$tmp/out"`; test -n "$ac_out"; } &&
+  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' "$tmp/out"`; test -z "$ac_out"; } &&
+  { $as_echo "$as_me:$LINENO: WARNING: $ac_file contains a reference to the variable \`datarootdir'
+which seems to be undefined.  Please make sure it is defined." >&5
 $as_echo "$as_me: WARNING: $ac_file contains a reference to the variable \`datarootdir'
-which seems to be undefined.  Please make sure it is defined" >&2;}
+which seems to be undefined.  Please make sure it is defined." >&2;}
 
-  rm -f "$ac_tmp/stdin"
+  rm -f "$tmp/stdin"
   case $ac_file in
-  -) cat "$ac_tmp/out" && rm -f "$ac_tmp/out";;
-  *) rm -f "$ac_file" && mv "$ac_tmp/out" "$ac_file";;
+  -) cat "$tmp/out" && rm -f "$tmp/out";;
+  *) rm -f "$ac_file" && mv "$tmp/out" "$ac_file";;
   esac \
-  || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+  || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; }
  ;;
   :H)
   #
@@ -6455,21 +7289,27 @@ which seems to be undefined.  Please make sure it is defined" >&2;}
   if test x"$ac_file" != x-; then
     {
       $as_echo "/* $configure_input  */" \
-      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs"
-    } >"$ac_tmp/config.h" \
-      || as_fn_error $? "could not create $ac_file" "$LINENO" 5
-    if diff "$ac_file" "$ac_tmp/config.h" >/dev/null 2>&1; then
-      { $as_echo "$as_me:${as_lineno-$LINENO}: $ac_file is unchanged" >&5
+      && eval '$AWK -f "$tmp/defines.awk"' "$ac_file_inputs"
+    } >"$tmp/config.h" \
+      || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; }
+    if diff "$ac_file" "$tmp/config.h" >/dev/null 2>&1; then
+      { $as_echo "$as_me:$LINENO: $ac_file is unchanged" >&5
 $as_echo "$as_me: $ac_file is unchanged" >&6;}
     else
       rm -f "$ac_file"
-      mv "$ac_tmp/config.h" "$ac_file" \
-	|| as_fn_error $? "could not create $ac_file" "$LINENO" 5
+      mv "$tmp/config.h" "$ac_file" \
+	|| { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; }
     fi
   else
     $as_echo "/* $configure_input  */" \
-      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs" \
-      || as_fn_error $? "could not create -" "$LINENO" 5
+      && eval '$AWK -f "$tmp/defines.awk"' "$ac_file_inputs" \
+      || { { $as_echo "$as_me:$LINENO: error: could not create -" >&5
+$as_echo "$as_me: error: could not create -" >&2;}
+   { (exit 1); exit 1; }; }
   fi
 # Compute "$ac_file"'s index in $config_headers.
 _am_arg="$ac_file"
@@ -6507,7 +7347,7 @@ $as_echo X"$_am_arg" |
 	  s/.*/./; q'`/stamp-h$_am_stamp_count
  ;;
 
-  :C)  { $as_echo "$as_me:${as_lineno-$LINENO}: executing $ac_file commands" >&5
+  :C)  { $as_echo "$as_me:$LINENO: executing $ac_file commands" >&5
 $as_echo "$as_me: executing $ac_file commands" >&6;}
  ;;
   esac
@@ -6515,7 +7355,7 @@ $as_echo "$as_me: executing $ac_file commands" >&6;}
 
   case $ac_file$ac_mode in
     "depfiles":C) test x"$AMDEP_TRUE" != x"" || {
-  # Older Autoconf quotes --file arguments for eval, but not when files
+  # Autoconf 2.62 quotes --file arguments for eval, but not when files
   # are listed without --file.  Let's play safe and only enable the eval
   # if we detect the quoting.
   case $CONFIG_FILES in
@@ -6528,7 +7368,7 @@ $as_echo "$as_me: executing $ac_file commands" >&6;}
     # Strip MF so we end up with the name of the file.
     mf=`echo "$mf" | sed -e 's/:.*$//'`
     # Check whether this is an Automake generated Makefile or not.
-    # We used to match only the files named 'Makefile.in', but
+    # We used to match only the files named `Makefile.in', but
     # some people rename them; so instead we look at the file content.
     # Grep'ing the first line is not enough: some people post-process
     # each Makefile.in and add a new line on top of each file to say so.
@@ -6562,19 +7402,21 @@ $as_echo X"$mf" |
       continue
     fi
     # Extract the definition of DEPDIR, am__include, and am__quote
-    # from the Makefile without running 'make'.
+    # from the Makefile without running `make'.
     DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
     test -z "$DEPDIR" && continue
     am__include=`sed -n 's/^am__include = //p' < "$mf"`
-    test -z "$am__include" && continue
+    test -z "am__include" && continue
     am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
+    # When using ansi2knr, U may be empty or an underscore; expand it
+    U=`sed -n 's/^U = //p' < "$mf"`
     # Find all dependency output files, they are included files with
     # $(DEPDIR) in their names.  We invoke sed twice because it is the
     # simplest approach to changing $(DEPDIR) to its actual value in the
     # expansion.
     for file in `sed -n "
       s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
-	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
+	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g' -e 's/\$U/'"$U"'/g'`; do
       # Make sure the directory exists.
       test -f "$dirpart/$file" && continue
       fdir=`$as_dirname -- "$file" ||
@@ -6600,7 +7442,47 @@ $as_echo X"$file" |
 	    q
 	  }
 	  s/.*/./; q'`
-      as_dir=$dirpart/$fdir; as_fn_mkdir_p
+      { as_dir=$dirpart/$fdir
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || { $as_mkdir_p && mkdir -p "$as_dir"; } || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || { { $as_echo "$as_me:$LINENO: error: cannot create directory $as_dir" >&5
+$as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
+   { (exit 1); exit 1; }; }; }
       # echo "creating $dirpart/$file"
       echo '# dummy' > "$dirpart/$file"
     done
@@ -6612,12 +7494,15 @@ $as_echo X"$file" |
 done # for ac_tag
 
 
-as_fn_exit 0
+{ (exit 0); exit 0; }
 _ACEOF
+chmod +x $CONFIG_STATUS
 ac_clean_files=$ac_clean_files_save
 
 test $ac_write_fail = 0 ||
-  as_fn_error $? "write failure creating $CONFIG_STATUS" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: write failure creating $CONFIG_STATUS" >&5
+$as_echo "$as_me: error: write failure creating $CONFIG_STATUS" >&2;}
+   { (exit 1); exit 1; }; }
 
 
 # configure is writing to config.log, and then calls config.status.
@@ -6638,10 +7523,10 @@ if test "$no_create" != yes; then
   exec 5>>config.log
   # Use ||, not &&, to avoid exiting from the if with $? = 1, which
   # would make configure fail if this is the last instruction.
-  $ac_cs_success || as_fn_exit 1
+  $ac_cs_success || { (exit 1); exit 1; }
 fi
 if test -n "$ac_unrecognized_opts" && test "$enable_option_checking" != no; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
+  { $as_echo "$as_me:$LINENO: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
 $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2;}
 fi
 
diff --git a/lib/Grid_config.h b/lib/Grid_config.h
index c743bba0..8b51adce 100644
--- a/lib/Grid_config.h
+++ b/lib/Grid_config.h
@@ -2,42 +2,42 @@
 /* lib/Grid_config.h.in.  Generated from configure.ac by autoheader.  */
 
 /* AVX */
-#define AVX1 1
+/* #undef AVX1 */
 
 /* AVX2 */
 /* #undef AVX2 */
 
 /* AVX512 */
-/* #undef AVX512 */
+#define AVX512 1
 
 /* GRID_COMMS_MPI */
-#define GRID_COMMS_MPI 1
+/* #undef GRID_COMMS_MPI */
 
 /* GRID_COMMS_NONE */
-/* #undef GRID_COMMS_NONE */
+#define GRID_COMMS_NONE 1
 
 /* Define to 1 if you have the declaration of `be64toh', and to 0 if you
    don't. */
-#define HAVE_DECL_BE64TOH 0
+#define HAVE_DECL_BE64TOH 1
 
 /* Define to 1 if you have the declaration of `ntohll', and to 0 if you don't.
    */
-#define HAVE_DECL_NTOHLL 1
+#define HAVE_DECL_NTOHLL 0
 
 /* Define to 1 if you have the <endian.h> header file. */
-/* #undef HAVE_ENDIAN_H */
+#define HAVE_ENDIAN_H 1
 
 /* Define to 1 if you have the `gettimeofday' function. */
-#define HAVE_GETTIMEOFDAY 1
+/* #undef HAVE_GETTIMEOFDAY */
 
 /* Define to 1 if you have the <inttypes.h> header file. */
 #define HAVE_INTTYPES_H 1
 
 /* Define to 1 if you have the <malloc.h> header file. */
-/* #undef HAVE_MALLOC_H */
+#define HAVE_MALLOC_H 1
 
 /* Define to 1 if you have the <malloc/malloc.h> header file. */
-#define HAVE_MALLOC_MALLOC_H 1
+/* #undef HAVE_MALLOC_MALLOC_H */
 
 /* Define to 1 if you have the <memory.h> header file. */
 #define HAVE_MEMORY_H 1
@@ -78,9 +78,6 @@
 /* Define to the one symbol short name of this package. */
 #define PACKAGE_TARNAME "grid"
 
-/* Define to the home page for this package. */
-#define PACKAGE_URL ""
-
 /* Define to the version of this package. */
 #define PACKAGE_VERSION "1.0"
 
diff --git a/lib/Grid_config.h.in b/lib/Grid_config.h.in
index 437a6095..4b094250 100644
--- a/lib/Grid_config.h.in
+++ b/lib/Grid_config.h.in
@@ -77,9 +77,6 @@
 /* Define to the one symbol short name of this package. */
 #undef PACKAGE_TARNAME
 
-/* Define to the home page for this package. */
-#undef PACKAGE_URL
-
 /* Define to the version of this package. */
 #undef PACKAGE_VERSION
 
diff --git a/lib/Grid_lattice.h b/lib/Grid_lattice.h
index 861a9e24..356a25b8 100644
--- a/lib/Grid_lattice.h
+++ b/lib/Grid_lattice.h
@@ -176,7 +176,7 @@ public:
  }; // class Lattice
 }
 
-#define GRID_LATTICE_EXPRESSION_TEMPLATES
+#undef GRID_LATTICE_EXPRESSION_TEMPLATES
 
 #include <lattice/Grid_lattice_conformable.h>
 
diff --git a/lib/Grid_simd.h b/lib/Grid_simd.h
index 4e33604d..6695eb0d 100644
--- a/lib/Grid_simd.h
+++ b/lib/Grid_simd.h
@@ -20,6 +20,7 @@
 #endif
 #ifdef AVX512
 #include <immintrin.h>
+#include <zmmintrin.h>
 #endif
 
 namespace Grid {
@@ -160,10 +161,10 @@ inline void Gpermute(vsimd &y,const vsimd &b,int perm){
         // Permute 1 every abcd efgh ijkl mnop -> cdab ghef jkij opmn 
         // Permute 2 every abcd efgh ijkl mnop -> efgh abcd mnop ijkl
         // Permute 3 every abcd efgh ijkl mnop -> ijkl mnop abcd efgh
-      case 3: y.v = _mm512_swizzle_ps(b.v,_MM_SWIZ_REG_CDAB); break;
-      case 2: y.v = _mm512_swizzle_ps(b.v,_MM_SWIZ_REG_BADC); break;
-      case 1: y.v = _mm512_permute4f128_ps(b.v,(_MM_PERM_ENUM)_MM_SHUFFLE(2,3,0,1)); break;
-      case 0: y.v = _mm512_permute4f128_ps(b.v,(_MM_PERM_ENUM)_MM_SHUFFLE(1,0,3,2)); break;
+      case 3: y.v =(decltype(y.v)) _mm512_swizzle_ps((__m512)b.v,_MM_SWIZ_REG_CDAB); break;
+      case 2: y.v =(decltype(y.v)) _mm512_swizzle_ps((__m512)b.v,_MM_SWIZ_REG_BADC); break;
+      case 1: y.v =(decltype(y.v)) _mm512_permute4f128_ps((__m512)b.v,(_MM_PERM_ENUM)_MM_SHUFFLE(2,3,0,1)); break;
+      case 0: y.v =(decltype(y.v)) _mm512_permute4f128_ps((__m512)b.v,(_MM_PERM_ENUM)_MM_SHUFFLE(1,0,3,2)); break;
 #endif
 #ifdef QPX
 #error not implemented
diff --git a/lib/cshift/Grid_cshift_common.h b/lib/cshift/Grid_cshift_common.h
index 6ae6df71..8320d2d0 100644
--- a/lib/cshift/Grid_cshift_common.h
+++ b/lib/cshift/Grid_cshift_common.h
@@ -26,19 +26,18 @@ Gather_plane_simple (const Lattice<vobj> &rhs,std::vector<cobj,alignedAllocator<
   }
 
   int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-  int o   = 0;                                      // relative offset to base within plane
   int bo  = 0;                                      // offset in buffer
   
 #pragma omp parallel for collapse(2)
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+      int o = n*rhs._grid->_slice_stride[dimension];
       int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);// Could easily be a table lookup
       if ( ocb &cbmask ) {
 	buffer[bo]=compress(rhs._odata[so+o+b]);
 	bo++;
       }
     }
-    o +=rhs._grid->_slice_stride[dimension];
   }
 }
 
@@ -56,13 +55,13 @@ Gather_plane_extract(const Lattice<vobj> &rhs,std::vector<typename cobj::scalar_
   }
 
   int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-  int o   = 0;                                      // relative offset to base within plane
   int bo  = 0;                                      // offset in buffer
     
 #pragma omp parallel for collapse(2)
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
 
+      int o=n*rhs._grid->_slice_stride[dimension];
       int offset = b+n*rhs._grid->_slice_block[dimension];
 
       int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);
@@ -71,9 +70,7 @@ Gather_plane_extract(const Lattice<vobj> &rhs,std::vector<typename cobj::scalar_
 	temp =compress(rhs._odata[so+o+b]);
 	extract<cobj>(temp,pointers,offset);
       }
-      
     }
-    o +=rhs._grid->_slice_stride[dimension];
   }
 }
 
@@ -107,20 +104,17 @@ template<class vobj> void Scatter_plane_simple (Lattice<vobj> &rhs,std::vector<v
   }
 
   int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-  int o   = 0;                                      // relative offset to base within plane
   int bo  = 0;                                      // offset in buffer
     
 #pragma omp parallel for collapse(2)
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-
+      int o=n*rhs._grid->_slice_stride[dimension];
       int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);// Could easily be a table lookup
       if ( ocb & cbmask ) {
 	rhs._odata[so+o+b]=buffer[bo++];
       }
-
     }
-    o +=rhs._grid->_slice_stride[dimension];
   }
 }
 
@@ -136,21 +130,18 @@ template<class vobj> void Scatter_plane_simple (Lattice<vobj> &rhs,std::vector<v
   }
 
   int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-  int o   = 0;                                      // relative offset to base within plane
-  int bo  = 0;                                      // offset in buffer
     
 #pragma omp parallel for collapse(2)
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
 
+      int o      = n*rhs._grid->_slice_stride[dimension];
       int offset = b+n*rhs._grid->_slice_block[dimension];
       int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);
       if ( ocb&cbmask ) {
 	merge(rhs._odata[so+o+b],pointers,offset);
       }
-      
     }
-    o +=rhs._grid->_slice_stride[dimension];
   }
 }
 
@@ -168,20 +159,18 @@ template<class vobj> void Copy_plane(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int
 
   int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
   int lo  = lplane*lhs._grid->_ostride[dimension]; // base offset for start of plane 
-  int o   = 0;                                     // relative offset to base within plane
   
 #pragma omp parallel for collapse(2)
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
       
+      int o =n*rhs._grid->_slice_stride[dimension];
       int ocb=1<<lhs._grid->CheckerBoardFromOindex(o+b);
-      
       if ( ocb&cbmask ) {
 	lhs._odata[lo+o+b]=rhs._odata[ro+o+b];
       }
       
     }
-    o +=rhs._grid->_slice_stride[dimension];
   }
 }
 
@@ -195,19 +184,16 @@ template<class vobj> void Copy_plane_permute(Lattice<vobj>& lhs,Lattice<vobj> &r
 
   int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
   int lo  = lplane*lhs._grid->_ostride[dimension]; // base offset for start of plane 
-  int o   = 0;                                     // relative offset to base within plane
   
 #pragma omp parallel for collapse(2)
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-      
+      int o =n*rhs._grid->_slice_stride[dimension];
       int ocb=1<<lhs._grid->CheckerBoardFromOindex(o+b);
       if ( ocb&cbmask ) {
 	permute(lhs._odata[lo+o+b],rhs._odata[ro+o+b],permute_type);
       }
-      
     }
-    o +=rhs._grid->_slice_stride[dimension];
   }
 }
 
diff --git a/lib/lattice/Grid_lattice_coordinate.h b/lib/lattice/Grid_lattice_coordinate.h
index a9b381fa..c49aed89 100644
--- a/lib/lattice/Grid_lattice_coordinate.h
+++ b/lib/lattice/Grid_lattice_coordinate.h
@@ -3,6 +3,17 @@
 
 namespace Grid {
 
+  /*
+  depbase=`echo Grid_main.o | sed 's|[^/]*$|.deps/&|;s|\.o$||'`;\
+        icpc -DHAVE_CONFIG_H -I. -I../lib    -I../lib -mmic -O3 -std=c++11 -fopenmp -MT Grid_main.o -MD -MP -MF $depbase.Tpo -c -o Grid_main.o Grid_main.cc &&\
+        mv -f $depbase.Tpo $depbase.Po
+	  ../lib/lattice/Grid_lattice_coordinate.h(25): error: no suitable user-defined conversion from "vector_type" to "const Grid::iScalar<Grid::iScalar<Grid::iScalar<Grid::vInteger>>>" exists
+    l._odata[o]=vI;
+                    ^
+          detected during instantiation of "void Grid::LatticeCoordinate(Grid::Lattice<iobj> &, int) [with iobj=Grid::QCD::vTInteger]" at line 283 of "Grid_main.cc"
+
+	    compilation aborted for Grid_main.cc (code 2)
+*/
     template<class iobj> inline void LatticeCoordinate(Lattice<iobj> &l,int mu)
     {
       typedef typename iobj::scalar_object scalar_object;
@@ -21,8 +32,8 @@ namespace Grid {
 	  grid->RankIndexToGlobalCoor(grid->ThisRank(),o,i,gcoor);
 	  mergebuf[i]=(Integer)gcoor[mu];
 	}
-	AmergeA<vector_type,scalar_type>(vI,mergebuf);
-	l._odata[o]=vI;
+	merge<vector_type,scalar_type>(vI,mergebuf);
+	l._odata[o]._internal._internal._internal=vI;
       }
     };
 
diff --git a/lib/lattice/Grid_lattice_overload.h b/lib/lattice/Grid_lattice_overload.h
index 512e45d1..c4dd3373 100644
--- a/lib/lattice/Grid_lattice_overload.h
+++ b/lib/lattice/Grid_lattice_overload.h
@@ -118,3 +118,4 @@ namespace Grid {
 
 
 }
+#endif
diff --git a/lib/math/Grid_math_tensors.h b/lib/math/Grid_math_tensors.h
index 09739555..97048d62 100644
--- a/lib/math/Grid_math_tensors.h
+++ b/lib/math/Grid_math_tensors.h
@@ -45,10 +45,6 @@ public:
       zeroit(*this);
       return *this;
     }
-    iScalar<vtype> & operator= (const scalar_type s){
-      _internal=s;
-      return *this;
-    }
     friend void vstream(iScalar<vtype> &out,const iScalar<vtype> &in){
       vstream(out._internal,in._internal);
     }
@@ -98,10 +94,10 @@ public:
     operator ComplexD () const { return(TensorRemove(_internal)); };
     operator RealD () const { return(real(TensorRemove(_internal))); }
 
-
+    // convert from a something to a scalar
     template<class T,typename std::enable_if<isGridTensor<T>::notvalue, T>::type* = nullptr > inline auto operator = (T arg) -> iScalar<vtype>
     { 
-      _internal = arg;
+      _internal = vtype(arg);
       return *this;
     }
 
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index 71b049ec..41718798 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -107,12 +107,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
 
     int ss = sss;
     int ssu= sss;
-<<<<<<< HEAD
-	//int ss = Stencil._LebesgueReorder[sss];
-=======
-    //int ss = 0;
     //int ss = Stencil._LebesgueReorder[sss];
->>>>>>> 55ccb8ccf4aa6b9d6e36e71aa19f89db0463254f
 
     // Xp
     offset = Stencil._offsets [Xp][ss];
@@ -130,10 +125,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
       chi=comm_buf[offset];
     }
     mult(&Uchi(),&Umu._odata[ssu](Xp),&chi());
-<<<<<<< HEAD
-=======
-    prefetch(Umu._odata[ssu](Yp));
->>>>>>> 55ccb8ccf4aa6b9d6e36e71aa19f89db0463254f
+    //prefetch(Umu._odata[ssu](Yp));
     spReconXp(result,Uchi);
 
     // Yp
@@ -152,10 +144,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
       chi=comm_buf[offset];
     }
     mult(&Uchi(),&Umu._odata[ssu](Yp),&chi());
-<<<<<<< HEAD
-=======
-    prefetch(Umu._odata[ssu](Zp));
->>>>>>> 55ccb8ccf4aa6b9d6e36e71aa19f89db0463254f
+    //    prefetch(Umu._odata[ssu](Zp));
     accumReconYp(result,Uchi);
 
     // Zp
@@ -174,10 +163,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
       chi=comm_buf[offset];
     }
     mult(&Uchi(),&Umu._odata[ssu](Zp),&chi());
-<<<<<<< HEAD
-=======
-    prefetch(Umu._odata[ssu](Tp));
->>>>>>> 55ccb8ccf4aa6b9d6e36e71aa19f89db0463254f
+    //    prefetch(Umu._odata[ssu](Tp));
     accumReconZp(result,Uchi);
 
     // Tp
@@ -196,10 +182,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
       chi=comm_buf[offset];
     }
     mult(&Uchi(),&Umu._odata[ssu](Tp),&chi());
-<<<<<<< HEAD
-=======
-    prefetch(Umu._odata[ssu](Xm));
->>>>>>> 55ccb8ccf4aa6b9d6e36e71aa19f89db0463254f
+    //    prefetch(Umu._odata[ssu](Xm));
     accumReconTp(result,Uchi);
 
     // Xm
@@ -218,10 +201,6 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
       chi=comm_buf[offset];
     }
     mult(&Uchi(),&Umu._odata[ssu](Xm),&chi());
-<<<<<<< HEAD
-=======
-    prefetch(Umu._odata[ssu](Ym));
->>>>>>> 55ccb8ccf4aa6b9d6e36e71aa19f89db0463254f
     accumReconXm(result,Uchi);
 
 
@@ -241,10 +220,6 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
       chi=comm_buf[offset];
     }
     mult(&Uchi(),&Umu._odata[ssu](Ym),&chi());
-<<<<<<< HEAD
-=======
-    prefetch(Umu._odata[ssu](Zm));
->>>>>>> 55ccb8ccf4aa6b9d6e36e71aa19f89db0463254f
     accumReconYm(result,Uchi);
 
     // Zm
@@ -263,10 +238,6 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
       chi=comm_buf[offset];
     }
     mult(&Uchi(),&Umu._odata[ssu](Zm),&chi());
-<<<<<<< HEAD
-=======
-    prefetch(Umu._odata[ssu](Tm));
->>>>>>> 55ccb8ccf4aa6b9d6e36e71aa19f89db0463254f
     accumReconZm(result,Uchi);
 
     // Tm
diff --git a/lib/simd/Grid_vComplexD.h b/lib/simd/Grid_vComplexD.h
index 5dbb28e5..ef5da859 100644
--- a/lib/simd/Grid_vComplexD.h
+++ b/lib/simd/Grid_vComplexD.h
@@ -143,10 +143,10 @@ namespace Grid {
 	     *                } 
              */
             zvec vzero,ymm0,ymm1,real,imag;
-            vzero =  _mm512_setzero();
+            vzero =(zvec)_mm512_setzero();
             ymm0 =  _mm512_swizzle_pd(a.v, _MM_SWIZ_REG_CDAB); // 
-            real =  _mm512_mask_or_epi64(a.v, 0xAAAA,vzero, ymm0);
-            imag =  _mm512_mask_sub_pd(a.v, 0x5555,vzero, ymm0);
+            real =(zvec)_mm512_mask_or_epi64((__m512i)a.v, 0xAA,(__m512i)vzero,(__m512i) ymm0);
+            imag =  _mm512_mask_sub_pd(a.v, 0x55,vzero, ymm0);
             ymm1 =  _mm512_mul_pd(real, b.v);
             ymm0 =  _mm512_swizzle_pd(b.v, _MM_SWIZ_REG_CDAB); // OK
             ret.v=  _mm512_fmadd_pd(ymm0,imag,ymm1);
@@ -250,7 +250,8 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
 	  _mm_stream_pd((double *)&out.v,in.v);
 #endif
 #ifdef AVX512
-	  _mm512_stream_pd((double *)&out.v,in.v);
+	  _mm512_storenrngo_pd((double *)&out.v,in.v);
+	  //	  _mm512_stream_pd((double *)&out.v,in.v);
 	  //Note v has a3 a2 a1 a0
 #endif
 #ifdef QPX
@@ -277,7 +278,7 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
 	    ret.v    = _mm_shuffle_pd(tmp,tmp,0x1);
 #endif
 #ifdef AVX512
-	    ret.v = _mm512_mask_sub_pd(in.v, 0xaaaa,ret.v, in.v);             
+	    ret.v = _mm512_mask_sub_pd(in.v, 0xaa,ret.v, in.v);             
 #endif
 #ifdef QPX
 	     assert(0);
@@ -297,7 +298,7 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
           ret.v    =_mm_shuffle_pd(tmp.v,tmp.v,0x1);
 #endif
 #ifdef AVX512
-          ret.v = _mm512_mask_sub_pd(in.v,0xaaaa,ret.v,in.v); // real -imag 
+          ret.v = _mm512_mask_sub_pd(in.v,0xaa,ret.v,in.v); // real -imag 
 	  ret.v = _mm512_swizzle_pd(ret.v, _MM_SWIZ_REG_CDAB);// OK
 #endif
 #ifdef QPX
@@ -319,7 +320,7 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
 #endif
 #ifdef AVX512
           tmp.v    = _mm512_swizzle_pd(in.v, _MM_SWIZ_REG_CDAB);// OK
-	  ret.v    = _mm512_mask_sub_pd(tmp.v,0xaaaa,ret.v,tmp.v); // real -imag 
+	  ret.v    = _mm512_mask_sub_pd(tmp.v,0xaa,ret.v,tmp.v); // real -imag 
 #endif
 #ifdef QPX
             assert(0);
@@ -337,7 +338,7 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
 	 return ComplexD(c_[0]+c_[2],c_[1]+c_[3]);
 #endif 
 #ifdef AVX512
-            return ComplexD(_mm512_mask_reduce_add_pd(0x5555, in.v),_mm512_mask_reduce_add_pd(0xAAAA, in.v));
+            return ComplexD(_mm512_mask_reduce_add_pd(0x55, in.v),_mm512_mask_reduce_add_pd(0xAA, in.v));
 #endif 
 #ifdef QPX
 #endif
diff --git a/lib/simd/Grid_vComplexF.h b/lib/simd/Grid_vComplexF.h
index 66a1e1bf..34510333 100644
--- a/lib/simd/Grid_vComplexF.h
+++ b/lib/simd/Grid_vComplexF.h
@@ -133,7 +133,7 @@ namespace Grid {
             cvec vzero,ymm0,ymm1,real, imag;
             vzero = _mm512_setzero();
             ymm0  = _mm512_swizzle_ps(a.v, _MM_SWIZ_REG_CDAB); // 
-            real  = _mm512_mask_or_epi32(a.v, 0xAAAA,vzero, ymm0);
+            real  = (__m512)_mm512_mask_or_epi32((__m512i)a.v, 0xAAAA,(__m512i)vzero,(__m512i)ymm0);
             imag  = _mm512_mask_sub_ps(a.v, 0x5555,vzero, ymm0);
             ymm1  = _mm512_mul_ps(real, b.v);
             ymm0  = _mm512_swizzle_ps(b.v, _MM_SWIZ_REG_CDAB); // OK
@@ -199,7 +199,8 @@ namespace Grid {
 	  _mm_stream_ps((float *)&out.v,in.v);
 #endif
 #ifdef AVX512
-	  _mm512_stream_ps((float *)&out.v,in.v);
+	  _mm512_storenrngo_ps((float *)&out.v,in.v);
+  //	  _mm512_stream_ps((float *)&out.v,in.v);
 	  //Note v has a3 a2 a1 a0
 #endif
 #ifdef QPX
diff --git a/lib/simd/Grid_vInteger.h b/lib/simd/Grid_vInteger.h
index b7950873..7d558fe7 100644
--- a/lib/simd/Grid_vInteger.h
+++ b/lib/simd/Grid_vInteger.h
@@ -3,8 +3,10 @@
 
 namespace Grid {
 
-#define _mm256_set_m128i(hi,lo) _mm256_insertf128_si256(_mm256_castsi128_si256(lo),(hi),1)
 // _mm256_set_m128i(hi,lo); // not defined in all versions of immintrin.h
+#ifndef _mm256_set_m128i
+#define _mm256_set_m128i(hi,lo) _mm256_insertf128_si256(_mm256_castsi128_si256(lo),(hi),1)
+#endif
 
   typedef uint32_t Integer;
 
@@ -110,7 +112,8 @@ namespace Grid {
             ret.v = _mm_mul_epi32(a.v,b.v);
 #endif
 #ifdef AVX512
-            ret.v = _mm512_mul_epi32(a.v,b.v);
+	    //            ret.v = _mm512_mul_epi32(a.v,b.v);
+	    ret.v = _mm512_mullo_epi32(a.v,b.v);
 #endif
 #ifdef QPX
             // Implement as array of ints is only option
diff --git a/lib/simd/Grid_vRealD.h b/lib/simd/Grid_vRealD.h
index 8041c1f9..4d5e2f57 100644
--- a/lib/simd/Grid_vRealD.h
+++ b/lib/simd/Grid_vRealD.h
@@ -184,7 +184,7 @@ namespace Grid {
 	  _mm_stream_pd((double *)&out.v,in.v);
 #endif
 #ifdef AVX512
-	  _mm512_stream_pd((double *)&out.v,in.v);
+	  _mm512_storenrngo_pd((double *)&out.v,in.v);
 	  //Note v has a3 a2 a1 a0
 #endif
 #ifdef QPX
diff --git a/lib/simd/Grid_vRealF.h b/lib/simd/Grid_vRealF.h
index e1d5679a..79e1b91f 100644
--- a/lib/simd/Grid_vRealF.h
+++ b/lib/simd/Grid_vRealF.h
@@ -216,7 +216,8 @@ friend inline void vstore(const vRealF &ret, float *a){
 	  _mm_stream_ps((float *)&out.v,in.v);
 #endif
 #ifdef AVX512
-	  _mm512_stream_ps((float *)&out.v,in.v);
+	  _mm512_storenrngo_ps((float *)&out.v,in.v);
+	  //	  _mm512_stream_ps((float *)&out.v,in.v);
 	  //Note v has a3 a2 a1 a0
 #endif
 #ifdef QPX
diff --git a/tests/Grid_gamma.cc b/tests/Grid_gamma.cc
index 3f350bc5..731b2d10 100644
--- a/tests/Grid_gamma.cc
+++ b/tests/Grid_gamma.cc
@@ -50,8 +50,9 @@ int main (int argc, char ** argv)
   //  std::cout << " Is triv Scalar<double> " <<std::has_trivial_default_constructor<iScalar<double> >::value << std::endl;
   //  std::cout << " Is triv Scalar<vComplexD> "<<std::has_trivial_default_constructor<iScalar<vComplexD> >::value  << std::endl;
 
+  std::complex<float> c(1.0);
   for(int a=0;a<Ns;a++){
-    ident()(a,a) = 1.0;
+    ident()(a,a) = c;
   }
 
   const Gamma::GammaMatrix *g = Gamma::GammaMatrices;
diff --git a/tests/Grid_main.cc b/tests/Grid_main.cc
index af13836e..00b63e70 100644
--- a/tests/Grid_main.cc
+++ b/tests/Grid_main.cc
@@ -4,6 +4,22 @@ using namespace std;
 using namespace Grid;
 using namespace Grid::QCD;
 
+/*
+ Grid_main.cc(232): error: no suitable user-defined conversion from "Grid::iScalar<Grid::iMatrix<Grid::iScalar<Grid::Complex>, 4>>" to "const Grid::iScalar<Grid::iScalar<Grid::iMatrix<Grid::Complex, 3>>>" exists
+c_m = peekIdiot<SpinColourMatrix>(scm,1,2);
+*/
+template<class vobj> auto peekIdiot(const vobj &rhs,int i,int j) -> decltype(peekIndex<2>(rhs,0,0))
+{
+  return peekIndex<2>(rhs,i,j);
+}
+template<class vobj> auto peekDumKopf(const vobj &rhs,int i,int j) -> decltype(peekIndex<3>(rhs,0,0))
+{
+  return peekIndex<3>(rhs,i,j);
+}
+template<class vobj> auto peekDumKopf(const vobj &rhs,int i) -> decltype(peekIndex<3>(rhs,0))
+{
+  return peekIndex<3>(rhs,i);
+}
 
 int main (int argc, char ** argv)
 {
@@ -224,7 +240,8 @@ int main (int argc, char ** argv)
       
       s_m = peekIndex<1>(scm,0,0);
       c_m = peekIndex<2>(scm,1,2);
-      c_m = peekSpin(scm,1,2);
+      //      c_m = peekSpin<SpinColourMatrix>(scm,1,2);
+      c_m = peekIdiot<SpinColourMatrix>(scm,1,2);
 
       printf("c. Level %d\n",c_m.TensorLevel);
       printf("c. Level %d\n",c_m().TensorLevel);
@@ -302,7 +319,8 @@ int main (int argc, char ** argv)
     int Nc = Grid::QCD::Nc;
 
     LatticeGaugeField U(&Fine);
-    LatticeColourMatrix Uy = peekLorentz(U,1);
+    //    LatticeColourMatrix Uy = peekLorentz(U,1);
+    LatticeColourMatrix Uy = peekDumKopf(U,1);
 
     flops = ncall*1.0*volume*(8*Nc*Nc*Nc);
     bytes = ncall*1.0*volume*Nc*Nc    *2*3*sizeof(Grid::Real);

From 06dcbed6b15860d9c3510d3286dc7fba83339509 Mon Sep 17 00:00:00 2001
From: paboyle <paboyle@frontend03.ecdf.ed.ac.uk>
Date: Mon, 11 May 2015 09:44:50 +0100
Subject: [PATCH 140/429] Updated to do list

---
 TODO | 31 ++++++++++++++-----------------
 1 file changed, 14 insertions(+), 17 deletions(-)

diff --git a/TODO b/TODO
index 57769ae6..980e331c 100644
--- a/TODO
+++ b/TODO
@@ -2,41 +2,39 @@
   - use protocol buffers? replace xmlReader/Writer ec..
   - Binary use htonll, htonl
 
+*** Hacks and bug fixes to clean up
 * Had to hack assignment to 1.0 in the tests/Grid_gamma test
-y
+* norm2l is a hack. figure out syntax error and make this norm2 c.f. tests/Grid_gamma.cc
+* Reduce implemention is poor
+* Bug in SeedFixedIntegers gives same output on each site.
+* Bug in RNG with complex numbers ; only filling real values; need helper function -- DONE
+* Conformable test in Cshift routines.
+
+*** Functionality
+
+* Command line args for geometry, simd, etc. layout. Is it necessary to have
+  user pass these? Is this a QCD specific?
+
+* Strong test for norm2, conj and all primitive types. -- Grid_simd test is almost there
+
 * Expression template engine:
 
    - Audit
    - Introduce base clase for Grid Tensors.
    - Introduce norm2 unary op.
    - Introduce conversion automatic from expression to Lattice<vobj>
-   -
 
-* Command line args for geometry, simd, etc. layout. Is it necessary to have
-  user pass these? Is this a QCD specific?
 
 * CovariantShift support                             -----Use a class to store gauge field? (parallel transport?)
 
-* Strong test for norm2, conj and all primitive types. -- Grid_simd test is almost there
-
-* Reduce implemention is poor
-* Bug in SeedFixedIntegers gives same output on each site.
-* Bug in RNG with complex numbers ; only filling real values; need helper function -- DONE
-
-* norm2l is a hack. figure out syntax error and make this norm2 c.f. tests/Grid_gamma.cc
-
 ** Make the Tensor types and Complex etc... play more nicely.
-
   - TensorRemove is a hack, come up with a long term rationalised approach to Complex vs. Scalar<Scalar<Scalar<Complex > > >
   QDP forces use of "toDouble" to get back to non tensor scalar. This role is presently taken TensorRemove, but I
   want to introduce a syntax that does not require this.
-
   - Reductions that contract indices on a site should always demote the tensor structure.
     norm2(), innerProduct.
-
   - Result of Sum(), SliceSum // spatial sums
               trace, traceIndex etc.. do not.
-
   - problem arises because "trace" returns Lattice<TComplex> moving everything down to Scalar,
     and then Sum and SliceSum to not remove the Scalars. This would be fixed if we 
     template specialize the scalar scalar scalar sum and SliceSum,  on the basis of being
@@ -46,7 +44,6 @@ y
  - I have collated into single location at least.
  - Need to use _mm_*insert/extract routines.
 
-* Conformable test in Cshift routines.
 
 * Flavour matrices?
 * Pauli, SU subgroup, etc.. 

From b42453d1fd76e41a7ed4ac4f538c67b5edd34aba Mon Sep 17 00:00:00 2001
From: paboyle <paboyle@ph.ed.ac.uk>
Date: Mon, 11 May 2015 12:43:10 +0100
Subject: [PATCH 141/429] Command line args and a general clean up

---
 TODO                                          |  6 +-
 benchmarks/Grid_comms.cc                      |  6 +-
 benchmarks/Grid_memory_bandwidth.cc           |  7 +-
 benchmarks/Grid_wilson.cc                     |  8 +-
 lib/Grid.h                                    |  8 ++
 lib/Grid_aligned_allocator.h                  | 15 ++-
 lib/Grid_comparison.h                         |  6 ++
 lib/Grid_cshift.h                             |  4 -
 lib/Grid_init.cc                              | 96 +++++++++++++++----
 lib/Makefile.am                               |  2 +-
 ...ator_fake.cc => Grid_communicator_none.cc} |  6 +-
 lib/math/Grid_math_arith_sub.h                | 18 ----
 lib/math/Grid_math_tensors.h                  | 44 +++++----
 lib/qcd/Grid_qcd_wilson_dop.cc                |  1 +
 tests/Grid_gamma.cc                           |  8 +-
 tests/Grid_main.cc                            | 32 +------
 tests/Grid_nersc_io.cc                        |  8 +-
 tests/Grid_simd.cc                            |  8 +-
 tests/Grid_stencil.cc                         |  8 +-
 19 files changed, 179 insertions(+), 112 deletions(-)
 rename lib/communicator/{Grid_communicator_fake.cc => Grid_communicator_none.cc} (97%)

diff --git a/TODO b/TODO
index 980e331c..1f5be83f 100644
--- a/TODO
+++ b/TODO
@@ -5,13 +5,17 @@
 *** Hacks and bug fixes to clean up
 * Had to hack assignment to 1.0 in the tests/Grid_gamma test
 * norm2l is a hack. figure out syntax error and make this norm2 c.f. tests/Grid_gamma.cc
-* Reduce implemention is poor
+
+* Reduce implemention is poor ; need threaded reductions; OMP isn't able to do it for generic objects.
+  
 * Bug in SeedFixedIntegers gives same output on each site.
 * Bug in RNG with complex numbers ; only filling real values; need helper function -- DONE
 * Conformable test in Cshift routines.
 
 *** Functionality
 
+* Implement where to take template scheme.
+
 * Command line args for geometry, simd, etc. layout. Is it necessary to have
   user pass these? Is this a QCD specific?
 
diff --git a/benchmarks/Grid_comms.cc b/benchmarks/Grid_comms.cc
index ef7c5491..b1e3848a 100644
--- a/benchmarks/Grid_comms.cc
+++ b/benchmarks/Grid_comms.cc
@@ -8,9 +8,11 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> simd_layout({1,1,2,2});
-  std::vector<int> mpi_layout ({1,2,2,1});
+  std::vector<int> latt_size;
+  std::vector<int> simd_layout;
+  std::vector<int> mpi_layout;
 
+  GridParseLayout(argv,argc,mpi_layout,simd_layout,latt_size);
 
   int Nloop=10;
   int nmu=0;
diff --git a/benchmarks/Grid_memory_bandwidth.cc b/benchmarks/Grid_memory_bandwidth.cc
index fb6db6d8..eaffe83e 100644
--- a/benchmarks/Grid_memory_bandwidth.cc
+++ b/benchmarks/Grid_memory_bandwidth.cc
@@ -8,8 +8,11 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> simd_layout({1,2,2,2});
-  std::vector<int> mpi_layout ({1,1,1,1});
+  std::vector<int> tmp_latt_size;
+  std::vector<int> simd_layout;
+  std::vector<int> mpi_layout;
+
+  GridParseLayout(argv,argc,mpi_layout,simd_layout,tmp_latt_size);
 
   const int Nvec=8;
   typedef Lattice< iVector< vReal,Nvec> > LatticeVec;
diff --git a/benchmarks/Grid_wilson.cc b/benchmarks/Grid_wilson.cc
index ecd7b65b..d0b96d1c 100644
--- a/benchmarks/Grid_wilson.cc
+++ b/benchmarks/Grid_wilson.cc
@@ -20,9 +20,11 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> simd_layout({1,1,2,2});
-  std::vector<int> mpi_layout ({1,1,1,1});
-  std::vector<int> latt_size  ({8,8,8,8});
+  std::vector<int> latt_size;
+  std::vector<int> simd_layout;
+  std::vector<int> mpi_layout;
+
+  GridParseLayout(argv,argc,mpi_layout,simd_layout,latt_size);
     
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
   std::vector<int> seeds({1,2,3,4});
diff --git a/lib/Grid.h b/lib/Grid.h
index 05095f48..4b452051 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -57,6 +57,7 @@
 
 namespace Grid {
 
+
   void Grid_init(int *argc,char ***argv);
   void Grid_finalize(void);
   void Grid_sa_signal_handler(int sig,siginfo_t *si,void * ptr);
@@ -68,6 +69,13 @@ namespace Grid {
   double usecond(void);
 
 
+  // Common parsing chores
+  std::string GridCmdOptionPayload(char ** begin, char ** end, const std::string & option);
+  bool        GridCmdOptionExists(char** begin, char** end, const std::string& option);
+  void        GridParseIntVector(std::string &str,std::vector<int> & vec);
+  void        GridParseLayout(char **argv,int argc,std::vector<int> &mpi,std::vector<int> &simd,std::vector<int> &latt);
+
+
 };
 
 #endif
diff --git a/lib/Grid_aligned_allocator.h b/lib/Grid_aligned_allocator.h
index 9008105a..f0b93172 100644
--- a/lib/Grid_aligned_allocator.h
+++ b/lib/Grid_aligned_allocator.h
@@ -20,31 +20,36 @@ public:
   typedef _Tp        value_type;
 
   template<typename _Tp1>  struct rebind { typedef alignedAllocator<_Tp1> other; };
+
   alignedAllocator() throw() { }
+
   alignedAllocator(const alignedAllocator&) throw() { }
+
   template<typename _Tp1> alignedAllocator(const alignedAllocator<_Tp1>&) throw() { }
+
   ~alignedAllocator() throw() { }
-  pointer address(reference __x) const { return &__x; }
+
+  pointer       address(reference __x)       const { return &__x; }
   const_pointer address(const_reference __x) const { return &__x; }
+
   size_type  max_size() const throw() { return size_t(-1) / sizeof(_Tp); }
-  // Should override allocate and deallocate
+
   pointer allocate(size_type __n, const void* = 0)
   { 
-    //_Tp * ptr = (_Tp *) memalign(sizeof(_Tp),__n*sizeof(_Tp));
-    // _Tp * ptr = (_Tp *) memalign(128,__n*sizeof(_Tp));
 #ifdef AVX512
     _Tp * ptr = (_Tp *) memalign(128,__n*sizeof(_Tp));
 #else
     _Tp * ptr = (_Tp *) _mm_malloc(__n*sizeof(_Tp),128);
 #endif
-
     return ptr;
   }
+
   void deallocate(pointer __p, size_type) { 
     free(__p); 
   }
   void construct(pointer __p, const _Tp& __val) { };
   void construct(pointer __p) { };
+
   void destroy(pointer __p) { };
 };
 
diff --git a/lib/Grid_comparison.h b/lib/Grid_comparison.h
index 53dec1a0..45311831 100644
--- a/lib/Grid_comparison.h
+++ b/lib/Grid_comparison.h
@@ -3,6 +3,12 @@
 
 namespace Grid {
 
+  /////////////////////////////////////////
+  // This implementation is a bit poor.
+  // Only support logical operations (== etc)
+  // on scalar objects. Strip any tensor structures.
+  // Should guard this with isGridTensor<> enable if?
+  /////////////////////////////////////////
     // Generic list of functors
     template<class lobj,class robj> class veq {
     public:
diff --git a/lib/Grid_cshift.h b/lib/Grid_cshift.h
index 1c601b7d..10c7a3c4 100644
--- a/lib/Grid_cshift.h
+++ b/lib/Grid_cshift.h
@@ -7,10 +7,6 @@
 #include <cshift/Grid_cshift_none.h>
 #endif
 
-#ifdef GRID_COMMS_FAKE
-#include <cshift/Grid_cshift_fake.h>
-#endif
-
 #ifdef GRID_COMMS_MPI
 #include <cshift/Grid_cshift_mpi.h>
 #endif 
diff --git a/lib/Grid_init.cc b/lib/Grid_init.cc
index 47159d05..2382dc15 100644
--- a/lib/Grid_init.cc
+++ b/lib/Grid_init.cc
@@ -12,6 +12,7 @@
 #include <signal.h>
 #include <iostream>
 #include <Grid.h>
+#include <algorithm>
 
 #undef __X86_64
 #define MAC
@@ -22,21 +23,6 @@
 
 namespace Grid {
 
-  std::streambuf *Grid_saved_stream_buf;
-#if 0
-  void Grid_quiesce_nodes(void)
-  {
-#ifdef GRID_COMMS_MPI
-    int me;
-    MPI_Comm_rank(MPI_COMM_WORLD,&me);
-    std::streambuf* Grid_saved_stream_buf = std::cout.rdbuf();
-    if ( me ) { 
-      std::ofstream file("log.node");
-      std::cout.rdbuf(file.rdbuf());
-    }
-#endif
-  }
-#endif
   void Grid_quiesce_nodes(void)
   {
 #ifdef GRID_COMMS_MPI
@@ -54,18 +40,95 @@ namespace Grid {
 #endif
   }
 
+std::string GridCmdOptionPayload(char ** begin, char ** end, const std::string & option)
+{
+  char ** itr = std::find(begin, end, option);
+  if (itr != end && ++itr != end) {
+    std::string payload(*itr);
+    return payload;
+  }
+  return std::string("");
+}
+bool GridCmdOptionExists(char** begin, char** end, const std::string& option)
+{
+  return std::find(begin, end, option) != end;
+}
 void Grid_init(int *argc,char ***argv)
 {
 #ifdef GRID_COMMS_MPI
   MPI_Init(argc,argv);
 #endif
-  Grid_debug_handler_init();
+  // Parse command line args.
   Grid_quiesce_nodes();
+
 }
+
+void GridCmdOptionIntVector(std::string &str,std::vector<int> & vec)
+{
+  vec.resize(0);
+  std::stringstream ss(str);
+  int i;
+  while (ss >> i){
+    vec.push_back(i);
+    if (ss.peek() == ',')
+      ss.ignore();
+  }    
+  return;
+}
+
+void GridParseLayout(char **argv,int argc,std::vector<int> &mpi,std::vector<int> &simd,std::vector<int> &latt)
+{
+  mpi =std::vector<int>({1,1,1,1});
+#if defined(AVX) || defined (AVX2)
+  simd=std::vector<int>({1,1,2,2});
+#endif
+#if defined(SSE4)
+  simd=std::vector<int>({1,1,1,2});
+#endif
+#if defined(AVX512)
+  simd=std::vector<int>({1,2,2,2});
+#endif
+  latt=std::vector<int>({8,8,8,8});
+  
+  std::string arg;
+  if( GridCmdOptionExists(argv,argv+argc,"--mpi-layout") ){
+    arg = GridCmdOptionPayload(argv,argv+argc,"--mpi-layout");
+    GridCmdOptionIntVector(arg,mpi);
+  }
+  if( GridCmdOptionExists(argv,argv+argc,"--simd-layout") ){
+    arg= GridCmdOptionPayload(argv,argv+argc,"--simd-layout");
+    GridCmdOptionIntVector(arg,simd);
+  }
+  if( GridCmdOptionExists(argv,argv+argc,"--lattice") ){
+    arg= GridCmdOptionPayload(argv,argv+argc,"--lattice");
+    GridCmdOptionIntVector(arg,latt);
+  }
+  std::cout<<"MPI layout";
+  for(int i=0;i<mpi.size();i++){
+    std::cout<<mpi[i]<<" ";
+  }
+  std::cout<<std::endl;
+  
+  std::cout<<"SIMD layout";
+  for(int i=0;i<simd.size();i++){
+    std::cout<<simd[i]<<" ";
+  }
+  std::cout<<std::endl;
+  
+  std::cout<<"Grid ";
+  for(int i=0;i<latt.size();i++){
+    std::cout<<latt[i]<<" ";
+  }
+  std::cout<<std::endl;
+  
+}
+
+
 void Grid_finalize(void)
 {
 #ifdef GRID_COMMS_MPI
   MPI_Finalize();
+  Grid_unquiesce_nodes();
 #endif
 }
 double usecond(void) {
@@ -74,7 +137,6 @@ double usecond(void) {
   return 1.0*tv.tv_usec + 1.0e6*tv.tv_sec;
 }
 
-
 #define _NBACKTRACE (256)
 void * Grid_backtrace_buffer[_NBACKTRACE];
 
diff --git a/lib/Makefile.am b/lib/Makefile.am
index 590a13f8..46795a33 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -7,7 +7,7 @@ if BUILD_COMMS_MPI
 endif
 
 if BUILD_COMMS_NONE
-  extra_sources+=communicator/Grid_communicator_fake.cc
+  extra_sources+=communicator/Grid_communicator_none.cc
 endif
 
 #
diff --git a/lib/communicator/Grid_communicator_fake.cc b/lib/communicator/Grid_communicator_none.cc
similarity index 97%
rename from lib/communicator/Grid_communicator_fake.cc
rename to lib/communicator/Grid_communicator_none.cc
index 1cad9c1f..90eb26cc 100644
--- a/lib/communicator/Grid_communicator_fake.cc
+++ b/lib/communicator/Grid_communicator_none.cc
@@ -29,7 +29,7 @@ void CartesianCommunicator::SendToRecvFrom(void *xmit,
 					   int from,
 					   int bytes)
 {
-  exit(-1);
+  assert(0);
 }
 void CartesianCommunicator::SendToRecvFromBegin(std::vector<CommsRequest_t> &list,
 						void *xmit,
@@ -38,11 +38,11 @@ void CartesianCommunicator::SendToRecvFromBegin(std::vector<CommsRequest_t> &lis
 						int from,
 						int bytes)
 {
-  exit(-1);
+  assert(0);
 }
 void CartesianCommunicator::SendToRecvFromComplete(std::vector<CommsRequest_t> &list)
 {
-  exit(-1);
+  assert(0);
 }
 
 void CartesianCommunicator::Barrier(void)
diff --git a/lib/math/Grid_math_arith_sub.h b/lib/math/Grid_math_arith_sub.h
index 3a58bf4c..de4ef9e5 100644
--- a/lib/math/Grid_math_arith_sub.h
+++ b/lib/math/Grid_math_arith_sub.h
@@ -68,24 +68,6 @@ template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vty
     return;
 }
 
-template<class v> void vprefetch(const iScalar<v> &vv)
-{
-  vprefetch(vv._internal);
-}
-template<class v,int N> void vprefetch(const iVector<v,N> &vv)
-{
-  for(int i=0;i<N;i++){
-    vprefetch(vv._internal[i]);
-  }
-}
-template<class v,int N> void vprefetch(const iMatrix<v,N> &vv)
-{
-  for(int i=0;i<N;i++){
-  for(int j=0;j<N;j++){
-    vprefetch(vv._internal[i][j]);
-  }}
-}
-
     // - operator for scalar, vector, matrix
 template<class ltype,class rtype> inline auto
 operator - (const iScalar<ltype>& lhs, const iScalar<rtype>& rhs) -> iScalar<decltype(lhs._internal - rhs._internal)>
diff --git a/lib/math/Grid_math_tensors.h b/lib/math/Grid_math_tensors.h
index 97048d62..b71b0047 100644
--- a/lib/math/Grid_math_tensors.h
+++ b/lib/math/Grid_math_tensors.h
@@ -16,9 +16,9 @@ namespace Grid {
 // However note that doing this eliminates some syntactical sugar such as 
 // calling the constructor explicitly or implicitly
 //
-#undef TENSOR_IS_POD
+class GridTensorBase {};
 
-template<class vtype> class iScalar
+template<class vtype> class iScalar :public GridTensorBase
 {
 public:
   vtype _internal;
@@ -34,12 +34,9 @@ public:
 
   // Scalar no action
   //  template<int Level> using tensor_reduce_level = typename iScalar<GridTypeMapper<vtype>::tensor_reduce_level<Level> >;
-
-#ifndef TENSOR_IS_POD
   iScalar()=default;
   iScalar(scalar_type s) : _internal(s) {};// recurse down and hit the constructor for vector_type
   iScalar(const Zero &z){ *this = zero; };
-#endif
 
     iScalar<vtype> & operator= (const Zero &hero){
       zeroit(*this);
@@ -87,15 +84,12 @@ public:
     inline const vtype & operator ()(void) const {
       return _internal;
     }
-    //    inline vtype && operator ()(void) {
-    //      return _internal;
-    //    }
 
     operator ComplexD () const { return(TensorRemove(_internal)); };
     operator RealD () const { return(real(TensorRemove(_internal))); }
 
     // convert from a something to a scalar
-    template<class T,typename std::enable_if<isGridTensor<T>::notvalue, T>::type* = nullptr > inline auto operator = (T arg) -> iScalar<vtype>
+    template<class T,typename std::enable_if<!isGridTensor<T>::value, T>::type* = nullptr > inline auto operator = (T arg) -> iScalar<vtype>
     { 
       _internal = vtype(arg);
       return *this;
@@ -105,13 +99,13 @@ public:
 ///////////////////////////////////////////////////////////
 // Allows to turn scalar<scalar<scalar<double>>>> back to double.
 ///////////////////////////////////////////////////////////
-template<class T>     inline typename std::enable_if<isGridTensor<T>::notvalue, T>::type TensorRemove(T arg) { return arg;}
+template<class T>     inline typename std::enable_if<!isGridTensor<T>::value, T>::type TensorRemove(T arg) { return arg;}
 template<class vtype> inline auto TensorRemove(iScalar<vtype> arg) -> decltype(TensorRemove(arg._internal))
 {
   return TensorRemove(arg._internal);
 }
     
-template<class vtype,int N> class iVector
+template<class vtype,int N> class iVector :public GridTensorBase
 {
 public:
   vtype _internal[N];
@@ -125,11 +119,8 @@ public:
 
 
   enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
-
-#ifndef TENSOR_IS_POD
   iVector(const Zero &z){ *this = zero; };
   iVector() =default;
-#endif
 
     iVector<vtype,N> & operator= (const Zero &hero){
         zeroit(*this);
@@ -184,7 +175,7 @@ public:
     //    }
 };
     
-template<class vtype,int N> class iMatrix
+template<class vtype,int N> class iMatrix :public GridTensorBase
 {
 public:
   vtype _internal[N][N];
@@ -198,16 +189,16 @@ public:
 
   enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
 
-#ifndef TENSOR_IS_POD
+
   iMatrix(const Zero &z){ *this = zero; };
   iMatrix() =default;
-#endif
+
 
   iMatrix<vtype,N> & operator= (const Zero &hero){
     zeroit(*this);
     return *this;
   }
-  template<class T,typename std::enable_if<isGridTensor<T>::notvalue, T>::type* = nullptr > inline auto operator = (T arg) -> iMatrix<vtype,N>
+  template<class T,typename std::enable_if<!isGridTensor<T>::value, T>::type* = nullptr > inline auto operator = (T arg) -> iMatrix<vtype,N>
     { 
       zeroit(*this);
       for(int i=0;i<N;i++)
@@ -278,6 +269,23 @@ public:
 
 };
 
+template<class v> void vprefetch(const iScalar<v> &vv)
+{
+  vprefetch(vv._internal);
+}
+template<class v,int N> void vprefetch(const iVector<v,N> &vv)
+{
+  for(int i=0;i<N;i++){
+    vprefetch(vv._internal[i]);
+  }
+}
+template<class v,int N> void vprefetch(const iMatrix<v,N> &vv)
+{
+  for(int i=0;i<N;i++){
+  for(int j=0;j<N;j++){
+    vprefetch(vv._internal[i][j]);
+  }}
+}
 
 
 }
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index 41718798..5054e9ce 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -103,6 +103,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
   vHalfSpinColourVector *chi_p;
   int offset,local,perm, ptype;
 
+#pragma omp parallel for
   for(int sss=0;sss<grid->oSites();sss++){
 
     int ss = sss;
diff --git a/tests/Grid_gamma.cc b/tests/Grid_gamma.cc
index 731b2d10..2fd1d864 100644
--- a/tests/Grid_gamma.cc
+++ b/tests/Grid_gamma.cc
@@ -14,9 +14,11 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> simd_layout({1,1,2,2});
-  std::vector<int> mpi_layout ({1,1,1,1});
-  std::vector<int> latt_size  ({8,8,8,8});
+  std::vector<int> latt_size;
+  std::vector<int> simd_layout;
+  std::vector<int> mpi_layout;
+
+  GridParseLayout(argv,argc,mpi_layout,simd_layout,latt_size);
     
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
 
diff --git a/tests/Grid_main.cc b/tests/Grid_main.cc
index 00b63e70..179694c6 100644
--- a/tests/Grid_main.cc
+++ b/tests/Grid_main.cc
@@ -25,15 +25,12 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> latt_size(4);
+  std::vector<int> latt_size;
+  std::vector<int> simd_layout;
+  std::vector<int> mpi_layout;
 
-  std::vector<int> simd_layout(4);
-  
-  std::vector<int> mpi_layout(4);
-  mpi_layout[0]=1;
-  mpi_layout[1]=1;
-  mpi_layout[2]=1;
-  mpi_layout[3]=1;
+  GridParseLayout(argv,argc,mpi_layout,simd_layout,latt_size);
+  latt_size.resize(4);
 
 #ifdef AVX512
  for(int omp=128;omp<236;omp+=16){
@@ -52,25 +49,6 @@ int main (int argc, char ** argv)
     latt_size[3] = lat;
     double volume = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
     
-#ifdef AVX512
-    simd_layout[0] = 1;
-    simd_layout[1] = 2;
-    simd_layout[2] = 2;
-    simd_layout[3] = 2;
-#endif
-#if defined (AVX1)|| defined (AVX2)
-    simd_layout[0] = 1;
-    simd_layout[1] = 1;
-    simd_layout[2] = 2;
-    simd_layout[3] = 2;
-#endif
-#if defined (SSE4)
-    simd_layout[0] = 1;
-    simd_layout[1] = 1;
-    simd_layout[2] = 1;
-    simd_layout[3] = 2;
-#endif
-    
     GridCartesian Fine(latt_size,simd_layout,mpi_layout);
     GridRedBlackCartesian rbFine(latt_size,simd_layout,mpi_layout);
     GridParallelRNG       FineRNG(&Fine);
diff --git a/tests/Grid_nersc_io.cc b/tests/Grid_nersc_io.cc
index 95f9db88..71cedd42 100644
--- a/tests/Grid_nersc_io.cc
+++ b/tests/Grid_nersc_io.cc
@@ -10,8 +10,12 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> simd_layout({1,1,2,2});
-  std::vector<int> mpi_layout ({1,1,1,1});
+  std::vector<int> tmp_latt_size;
+  std::vector<int> simd_layout;
+  std::vector<int> mpi_layout;
+
+  GridParseLayout(argv,argc,mpi_layout,simd_layout,tmp_latt_size);
+
   std::vector<int> latt_size  ({16,16,16,32});
   std::vector<int> clatt_size  ({4,4,4,8});
   int orthodir=3;
diff --git a/tests/Grid_simd.cc b/tests/Grid_simd.cc
index 1de2ad1f..fe90af2e 100644
--- a/tests/Grid_simd.cc
+++ b/tests/Grid_simd.cc
@@ -106,9 +106,11 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> simd_layout({1,1,2,2});
-  std::vector<int> mpi_layout ({1,1,1,1});
-  std::vector<int> latt_size  ({8,8,8,8});
+  std::vector<int> latt_size;
+  std::vector<int> simd_layout;
+  std::vector<int> mpi_layout;
+
+  GridParseLayout(argv,argc,mpi_layout,simd_layout,latt_size);
     
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
   std::vector<int> seeds({1,2,3,4});
diff --git a/tests/Grid_stencil.cc b/tests/Grid_stencil.cc
index 64a4ee9d..16acf294 100644
--- a/tests/Grid_stencil.cc
+++ b/tests/Grid_stencil.cc
@@ -8,9 +8,11 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> simd_layout({1,1,2,2});
-  std::vector<int> mpi_layout ({2,2,1,2});
-  std::vector<int> latt_size  ({8,8,8,8});
+  std::vector<int> latt_size;
+  std::vector<int> simd_layout;
+  std::vector<int> mpi_layout;
+
+  GridParseLayout(argv,argc,mpi_layout,simd_layout,latt_size);
 
   double volume = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
     

From 4eb08ac9de57cb72faea4c0bb42584d43514f07e Mon Sep 17 00:00:00 2001
From: paboyle <paboyle@ph.ed.ac.uk>
Date: Mon, 11 May 2015 12:56:27 +0100
Subject: [PATCH 142/429] CML parse

---
 tests/Grid_cshift.cc | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/tests/Grid_cshift.cc b/tests/Grid_cshift.cc
index fcee7b70..619808d6 100644
--- a/tests/Grid_cshift.cc
+++ b/tests/Grid_cshift.cc
@@ -8,9 +8,10 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> simd_layout({1,1,2,2});
-  std::vector<int> mpi_layout ({1,1,1,1});
-  std::vector<int> latt_size  ({8,8,8,16});
+  std::vector<int> simd_layout;
+  std::vector<int> mpi_layout;
+  std::vector<int> latt_size;
+  GridParseLayout(argv,argc,mpi_layout,simd_layout,latt_size);
     
   GridCartesian     Fine(latt_size,simd_layout,mpi_layout);
   GridParallelRNG           FineRNG(&Fine);

From b613ed0bb817a9600063b87ad8607862abec3134 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 11 May 2015 14:36:48 +0100
Subject: [PATCH 143/429] Got command line args working

---
 Makefile.in                         |  393 ++-
 aclocal.m4                          |  705 ++--
 benchmarks/Grid_comms.cc            |    2 +-
 benchmarks/Grid_memory_bandwidth.cc |    2 +-
 benchmarks/Grid_wilson.cc           |    2 +-
 configure                           | 5011 +++++++++++----------------
 lib/Grid.h                          |    6 +-
 lib/Grid_config.h                   |   23 +-
 lib/Grid_config.h.in                |    3 +
 lib/Grid_init.cc                    |   61 +-
 lib/math/Grid_math_tensors.h        |    6 +-
 tests/Grid_cshift.cc                |   10 +-
 tests/Grid_gamma.cc                 |    2 +-
 tests/Grid_main.cc                  |    2 +-
 tests/Grid_nersc_io.cc              |    2 +-
 tests/Grid_simd.cc                  |    2 +-
 tests/Grid_stencil.cc               |    2 +-
 17 files changed, 2827 insertions(+), 3407 deletions(-)

diff --git a/Makefile.in b/Makefile.in
index f9e18fe4..c9257ab2 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -1,9 +1,8 @@
-# Makefile.in generated by automake 1.11.1 from Makefile.am.
+# Makefile.in generated by automake 1.15 from Makefile.am.
 # @configure_input@
 
-# Copyright (C) 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002,
-# 2003, 2004, 2005, 2006, 2007, 2008, 2009  Free Software Foundation,
-# Inc.
+# Copyright (C) 1994-2014 Free Software Foundation, Inc.
+
 # This Makefile.in is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
@@ -15,6 +14,61 @@
 
 @SET_MAKE@
 VPATH = @srcdir@
+am__is_gnu_make = { \
+  if test -z '$(MAKELEVEL)'; then \
+    false; \
+  elif test -n '$(MAKE_HOST)'; then \
+    true; \
+  elif test -n '$(MAKE_VERSION)' && test -n '$(CURDIR)'; then \
+    true; \
+  else \
+    false; \
+  fi; \
+}
+am__make_running_with_option = \
+  case $${target_option-} in \
+      ?) ;; \
+      *) echo "am__make_running_with_option: internal error: invalid" \
+              "target option '$${target_option-}' specified" >&2; \
+         exit 1;; \
+  esac; \
+  has_opt=no; \
+  sane_makeflags=$$MAKEFLAGS; \
+  if $(am__is_gnu_make); then \
+    sane_makeflags=$$MFLAGS; \
+  else \
+    case $$MAKEFLAGS in \
+      *\\[\ \	]*) \
+        bs=\\; \
+        sane_makeflags=`printf '%s\n' "$$MAKEFLAGS" \
+          | sed "s/$$bs$$bs[$$bs $$bs	]*//g"`;; \
+    esac; \
+  fi; \
+  skip_next=no; \
+  strip_trailopt () \
+  { \
+    flg=`printf '%s\n' "$$flg" | sed "s/$$1.*$$//"`; \
+  }; \
+  for flg in $$sane_makeflags; do \
+    test $$skip_next = yes && { skip_next=no; continue; }; \
+    case $$flg in \
+      *=*|--*) continue;; \
+        -*I) strip_trailopt 'I'; skip_next=yes;; \
+      -*I?*) strip_trailopt 'I';; \
+        -*O) strip_trailopt 'O'; skip_next=yes;; \
+      -*O?*) strip_trailopt 'O';; \
+        -*l) strip_trailopt 'l'; skip_next=yes;; \
+      -*l?*) strip_trailopt 'l';; \
+      -[dEDm]) skip_next=yes;; \
+      -[JT]) skip_next=yes;; \
+    esac; \
+    case $$flg in \
+      *$$target_option*) has_opt=yes; break;; \
+    esac; \
+  done; \
+  test $$has_opt = yes
+am__make_dryrun = (target_option=n; $(am__make_running_with_option))
+am__make_keepgoing = (target_option=k; $(am__make_running_with_option))
 pkgdatadir = $(datadir)/@PACKAGE@
 pkgincludedir = $(includedir)/@PACKAGE@
 pkglibdir = $(libdir)/@PACKAGE@
@@ -32,43 +86,86 @@ NORMAL_UNINSTALL = :
 PRE_UNINSTALL = :
 POST_UNINSTALL = :
 subdir = .
-DIST_COMMON = README $(am__configure_deps) $(srcdir)/Makefile.am \
-	$(srcdir)/Makefile.in $(top_srcdir)/configure AUTHORS COPYING \
-	ChangeLog INSTALL NEWS TODO compile depcomp install-sh missing
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/configure.ac
 am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
 	$(ACLOCAL_M4)
+DIST_COMMON = $(srcdir)/Makefile.am $(top_srcdir)/configure \
+	$(am__configure_deps) $(am__DIST_COMMON)
 am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \
  configure.lineno config.status.lineno
 mkinstalldirs = $(install_sh) -d
 CONFIG_HEADER = $(top_builddir)/lib/Grid_config.h
 CONFIG_CLEAN_FILES =
 CONFIG_CLEAN_VPATH_FILES =
+AM_V_P = $(am__v_P_@AM_V@)
+am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
+am__v_P_0 = false
+am__v_P_1 = :
+AM_V_GEN = $(am__v_GEN_@AM_V@)
+am__v_GEN_ = $(am__v_GEN_@AM_DEFAULT_V@)
+am__v_GEN_0 = @echo "  GEN     " $@;
+am__v_GEN_1 = 
+AM_V_at = $(am__v_at_@AM_V@)
+am__v_at_ = $(am__v_at_@AM_DEFAULT_V@)
+am__v_at_0 = @
+am__v_at_1 = 
 SOURCES =
 DIST_SOURCES =
-RECURSIVE_TARGETS = all-recursive check-recursive dvi-recursive \
-	html-recursive info-recursive install-data-recursive \
-	install-dvi-recursive install-exec-recursive \
-	install-html-recursive install-info-recursive \
-	install-pdf-recursive install-ps-recursive install-recursive \
-	installcheck-recursive installdirs-recursive pdf-recursive \
-	ps-recursive uninstall-recursive
+RECURSIVE_TARGETS = all-recursive check-recursive cscopelist-recursive \
+	ctags-recursive dvi-recursive html-recursive info-recursive \
+	install-data-recursive install-dvi-recursive \
+	install-exec-recursive install-html-recursive \
+	install-info-recursive install-pdf-recursive \
+	install-ps-recursive install-recursive installcheck-recursive \
+	installdirs-recursive pdf-recursive ps-recursive \
+	tags-recursive uninstall-recursive
+am__can_run_installinfo = \
+  case $$AM_UPDATE_INFO_DIR in \
+    n|no|NO) false;; \
+    *) (install-info --version) >/dev/null 2>&1;; \
+  esac
 RECURSIVE_CLEAN_TARGETS = mostlyclean-recursive clean-recursive	\
   distclean-recursive maintainer-clean-recursive
-AM_RECURSIVE_TARGETS = $(RECURSIVE_TARGETS:-recursive=) \
-	$(RECURSIVE_CLEAN_TARGETS:-recursive=) tags TAGS ctags CTAGS \
-	distdir dist dist-all distcheck
+am__recursive_targets = \
+  $(RECURSIVE_TARGETS) \
+  $(RECURSIVE_CLEAN_TARGETS) \
+  $(am__extra_recursive_targets)
+AM_RECURSIVE_TARGETS = $(am__recursive_targets:-recursive=) TAGS CTAGS \
+	cscope distdir dist dist-all distcheck
+am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) $(LISP)
+# Read a list of newline-separated strings from the standard input,
+# and print each of them once, without duplicates.  Input order is
+# *not* preserved.
+am__uniquify_input = $(AWK) '\
+  BEGIN { nonempty = 0; } \
+  { items[$$0] = 1; nonempty = 1; } \
+  END { if (nonempty) { for (i in items) print i; }; } \
+'
+# Make sure the list of sources is unique.  This is necessary because,
+# e.g., the same source file might be shared among _SOURCES variables
+# for different programs/libraries.
+am__define_uniq_tagged_files = \
+  list='$(am__tagged_files)'; \
+  unique=`for i in $$list; do \
+    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+  done | $(am__uniquify_input)`
 ETAGS = etags
 CTAGS = ctags
+CSCOPE = cscope
 DIST_SUBDIRS = $(SUBDIRS)
+am__DIST_COMMON = $(srcdir)/Makefile.in AUTHORS COPYING ChangeLog \
+	INSTALL NEWS README TODO compile depcomp install-sh missing
 DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
 distdir = $(PACKAGE)-$(VERSION)
 top_distdir = $(distdir)
 am__remove_distdir = \
-  { test ! -d "$(distdir)" \
-    || { find "$(distdir)" -type d ! -perm -200 -exec chmod u+w {} ';' \
-         && rm -fr "$(distdir)"; }; }
+  if test -d "$(distdir)"; then \
+    find "$(distdir)" -type d ! -perm -200 -exec chmod u+w {} ';' \
+      && rm -rf "$(distdir)" \
+      || { sleep 5 && rm -rf "$(distdir)"; }; \
+  else :; fi
+am__post_remove_distdir = $(am__remove_distdir)
 am__relativize = \
   dir0=`pwd`; \
   sed_first='s,^\([^/]*\)/.*$$,\1,'; \
@@ -96,10 +193,14 @@ am__relativize = \
   reldir="$$dir2"
 DIST_ARCHIVES = $(distdir).tar.gz
 GZIP_ENV = --best
+DIST_TARGETS = dist-gzip
 distuninstallcheck_listfiles = find . -type f -print
+am__distuninstallcheck_listfiles = $(distuninstallcheck_listfiles) \
+  | sed 's|^\./|$(prefix)/|' | grep -v '$(infodir)/dir$$'
 distcleancheck_listfiles = find . -type f -print
 ACLOCAL = @ACLOCAL@
 AMTAR = @AMTAR@
+AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
 AUTOCONF = @AUTOCONF@
 AUTOHEADER = @AUTOHEADER@
 AUTOMAKE = @AUTOMAKE@
@@ -139,6 +240,7 @@ PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
 PACKAGE_NAME = @PACKAGE_NAME@
 PACKAGE_STRING = @PACKAGE_STRING@
 PACKAGE_TARNAME = @PACKAGE_TARNAME@
+PACKAGE_URL = @PACKAGE_URL@
 PACKAGE_VERSION = @PACKAGE_VERSION@
 PATH_SEPARATOR = @PATH_SEPARATOR@
 RANLIB = @RANLIB@
@@ -196,7 +298,7 @@ SUBDIRS = lib tests benchmarks
 all: all-recursive
 
 .SUFFIXES:
-am--refresh:
+am--refresh: Makefile
 	@:
 $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	@for dep in $?; do \
@@ -211,7 +313,6 @@ $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu Makefile'; \
 	$(am__cd) $(top_srcdir) && \
 	  $(AUTOMAKE) --gnu Makefile
-.PRECIOUS: Makefile
 Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
 	@case '$?' in \
 	  *config.status*) \
@@ -232,22 +333,25 @@ $(ACLOCAL_M4):  $(am__aclocal_m4_deps)
 $(am__aclocal_m4_deps):
 
 # This directory's subdirectories are mostly independent; you can cd
-# into them and run `make' without going through this Makefile.
-# To change the values of `make' variables: instead of editing Makefiles,
-# (1) if the variable is set in `config.status', edit `config.status'
-#     (which will cause the Makefiles to be regenerated when you run `make');
-# (2) otherwise, pass the desired values on the `make' command line.
-$(RECURSIVE_TARGETS):
-	@fail= failcom='exit 1'; \
-	for f in x $$MAKEFLAGS; do \
-	  case $$f in \
-	    *=* | --[!k]*);; \
-	    *k*) failcom='fail=yes';; \
-	  esac; \
-	done; \
+# into them and run 'make' without going through this Makefile.
+# To change the values of 'make' variables: instead of editing Makefiles,
+# (1) if the variable is set in 'config.status', edit 'config.status'
+#     (which will cause the Makefiles to be regenerated when you run 'make');
+# (2) otherwise, pass the desired values on the 'make' command line.
+$(am__recursive_targets):
+	@fail=; \
+	if $(am__make_keepgoing); then \
+	  failcom='fail=yes'; \
+	else \
+	  failcom='exit 1'; \
+	fi; \
 	dot_seen=no; \
 	target=`echo $@ | sed s/-recursive//`; \
-	list='$(SUBDIRS)'; for subdir in $$list; do \
+	case "$@" in \
+	  distclean-* | maintainer-clean-*) list='$(DIST_SUBDIRS)' ;; \
+	  *) list='$(SUBDIRS)' ;; \
+	esac; \
+	for subdir in $$list; do \
 	  echo "Making $$target in $$subdir"; \
 	  if test "$$subdir" = "."; then \
 	    dot_seen=yes; \
@@ -262,57 +366,12 @@ $(RECURSIVE_TARGETS):
 	  $(MAKE) $(AM_MAKEFLAGS) "$$target-am" || exit 1; \
 	fi; test -z "$$fail"
 
-$(RECURSIVE_CLEAN_TARGETS):
-	@fail= failcom='exit 1'; \
-	for f in x $$MAKEFLAGS; do \
-	  case $$f in \
-	    *=* | --[!k]*);; \
-	    *k*) failcom='fail=yes';; \
-	  esac; \
-	done; \
-	dot_seen=no; \
-	case "$@" in \
-	  distclean-* | maintainer-clean-*) list='$(DIST_SUBDIRS)' ;; \
-	  *) list='$(SUBDIRS)' ;; \
-	esac; \
-	rev=''; for subdir in $$list; do \
-	  if test "$$subdir" = "."; then :; else \
-	    rev="$$subdir $$rev"; \
-	  fi; \
-	done; \
-	rev="$$rev ."; \
-	target=`echo $@ | sed s/-recursive//`; \
-	for subdir in $$rev; do \
-	  echo "Making $$target in $$subdir"; \
-	  if test "$$subdir" = "."; then \
-	    local_target="$$target-am"; \
-	  else \
-	    local_target="$$target"; \
-	  fi; \
-	  ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) $$local_target) \
-	  || eval $$failcom; \
-	done && test -z "$$fail"
-tags-recursive:
-	list='$(SUBDIRS)'; for subdir in $$list; do \
-	  test "$$subdir" = . || ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) tags); \
-	done
-ctags-recursive:
-	list='$(SUBDIRS)'; for subdir in $$list; do \
-	  test "$$subdir" = . || ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) ctags); \
-	done
+ID: $(am__tagged_files)
+	$(am__define_uniq_tagged_files); mkid -fID $$unique
+tags: tags-recursive
+TAGS: tags
 
-ID: $(HEADERS) $(SOURCES) $(LISP) $(TAGS_FILES)
-	list='$(SOURCES) $(HEADERS) $(LISP) $(TAGS_FILES)'; \
-	unique=`for i in $$list; do \
-	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
-	  done | \
-	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
-	      END { if (nonempty) { for (i in files) print i; }; }'`; \
-	mkid -fID $$unique
-tags: TAGS
-
-TAGS: tags-recursive $(HEADERS) $(SOURCES)  $(TAGS_DEPENDENCIES) \
-		$(TAGS_FILES) $(LISP)
+tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
 	set x; \
 	here=`pwd`; \
 	if ($(ETAGS) --etags-include --version) >/dev/null 2>&1; then \
@@ -328,12 +387,7 @@ TAGS: tags-recursive $(HEADERS) $(SOURCES)  $(TAGS_DEPENDENCIES) \
 	      set "$$@" "$$include_option=$$here/$$subdir/TAGS"; \
 	  fi; \
 	done; \
-	list='$(SOURCES) $(HEADERS)  $(LISP) $(TAGS_FILES)'; \
-	unique=`for i in $$list; do \
-	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
-	  done | \
-	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
-	      END { if (nonempty) { for (i in files) print i; }; }'`; \
+	$(am__define_uniq_tagged_files); \
 	shift; \
 	if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
 	  test -n "$$unique" || unique=$$empty_fix; \
@@ -345,15 +399,11 @@ TAGS: tags-recursive $(HEADERS) $(SOURCES)  $(TAGS_DEPENDENCIES) \
 	      $$unique; \
 	  fi; \
 	fi
-ctags: CTAGS
-CTAGS: ctags-recursive $(HEADERS) $(SOURCES)  $(TAGS_DEPENDENCIES) \
-		$(TAGS_FILES) $(LISP)
-	list='$(SOURCES) $(HEADERS)  $(LISP) $(TAGS_FILES)'; \
-	unique=`for i in $$list; do \
-	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
-	  done | \
-	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
-	      END { if (nonempty) { for (i in files) print i; }; }'`; \
+ctags: ctags-recursive
+
+CTAGS: ctags
+ctags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
+	$(am__define_uniq_tagged_files); \
 	test -z "$(CTAGS_ARGS)$$unique" \
 	  || $(CTAGS) $(CTAGSFLAGS) $(AM_CTAGSFLAGS) $(CTAGS_ARGS) \
 	     $$unique
@@ -362,9 +412,31 @@ GTAGS:
 	here=`$(am__cd) $(top_builddir) && pwd` \
 	  && $(am__cd) $(top_srcdir) \
 	  && gtags -i $(GTAGS_ARGS) "$$here"
+cscope: cscope.files
+	test ! -s cscope.files \
+	  || $(CSCOPE) -b -q $(AM_CSCOPEFLAGS) $(CSCOPEFLAGS) -i cscope.files $(CSCOPE_ARGS)
+clean-cscope:
+	-rm -f cscope.files
+cscope.files: clean-cscope cscopelist
+cscopelist: cscopelist-recursive
+
+cscopelist-am: $(am__tagged_files)
+	list='$(am__tagged_files)'; \
+	case "$(srcdir)" in \
+	  [\\/]* | ?:[\\/]*) sdir="$(srcdir)" ;; \
+	  *) sdir=$(subdir)/$(srcdir) ;; \
+	esac; \
+	for i in $$list; do \
+	  if test -f "$$i"; then \
+	    echo "$(subdir)/$$i"; \
+	  else \
+	    echo "$$sdir/$$i"; \
+	  fi; \
+	done >> $(top_builddir)/cscope.files
 
 distclean-tags:
 	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
+	-rm -f cscope.out cscope.in.out cscope.po.out cscope.files
 
 distdir: $(DISTFILES)
 	$(am__remove_distdir)
@@ -400,13 +472,10 @@ distdir: $(DISTFILES)
 	done
 	@list='$(DIST_SUBDIRS)'; for subdir in $$list; do \
 	  if test "$$subdir" = .; then :; else \
-	    test -d "$(distdir)/$$subdir" \
-	    || $(MKDIR_P) "$(distdir)/$$subdir" \
-	    || exit 1; \
-	  fi; \
-	done
-	@list='$(DIST_SUBDIRS)'; for subdir in $$list; do \
-	  if test "$$subdir" = .; then :; else \
+	    $(am__make_dryrun) \
+	      || test -d "$(distdir)/$$subdir" \
+	      || $(MKDIR_P) "$(distdir)/$$subdir" \
+	      || exit 1; \
 	    dir1=$$subdir; dir2="$(distdir)/$$subdir"; \
 	    $(am__relativize); \
 	    new_distdir=$$reldir; \
@@ -435,36 +504,42 @@ distdir: $(DISTFILES)
 	|| chmod -R a+r "$(distdir)"
 dist-gzip: distdir
 	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
-	$(am__remove_distdir)
+	$(am__post_remove_distdir)
 
 dist-bzip2: distdir
-	tardir=$(distdir) && $(am__tar) | bzip2 -9 -c >$(distdir).tar.bz2
-	$(am__remove_distdir)
+	tardir=$(distdir) && $(am__tar) | BZIP2=$${BZIP2--9} bzip2 -c >$(distdir).tar.bz2
+	$(am__post_remove_distdir)
 
-dist-lzma: distdir
-	tardir=$(distdir) && $(am__tar) | lzma -9 -c >$(distdir).tar.lzma
-	$(am__remove_distdir)
+dist-lzip: distdir
+	tardir=$(distdir) && $(am__tar) | lzip -c $${LZIP_OPT--9} >$(distdir).tar.lz
+	$(am__post_remove_distdir)
 
 dist-xz: distdir
-	tardir=$(distdir) && $(am__tar) | xz -c >$(distdir).tar.xz
-	$(am__remove_distdir)
+	tardir=$(distdir) && $(am__tar) | XZ_OPT=$${XZ_OPT--e} xz -c >$(distdir).tar.xz
+	$(am__post_remove_distdir)
 
 dist-tarZ: distdir
+	@echo WARNING: "Support for distribution archives compressed with" \
+		       "legacy program 'compress' is deprecated." >&2
+	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	tardir=$(distdir) && $(am__tar) | compress -c >$(distdir).tar.Z
-	$(am__remove_distdir)
+	$(am__post_remove_distdir)
 
 dist-shar: distdir
+	@echo WARNING: "Support for shar distribution archives is" \
+	               "deprecated." >&2
+	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	shar $(distdir) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).shar.gz
-	$(am__remove_distdir)
+	$(am__post_remove_distdir)
 
 dist-zip: distdir
 	-rm -f $(distdir).zip
 	zip -rq $(distdir).zip $(distdir)
-	$(am__remove_distdir)
+	$(am__post_remove_distdir)
 
-dist dist-all: distdir
-	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
-	$(am__remove_distdir)
+dist dist-all:
+	$(MAKE) $(AM_MAKEFLAGS) $(DIST_TARGETS) am__post_remove_distdir='@:'
+	$(am__post_remove_distdir)
 
 # This target untars the dist file and tries a VPATH configuration.  Then
 # it guarantees that the distribution is self-contained by making another
@@ -475,8 +550,8 @@ distcheck: dist
 	  GZIP=$(GZIP_ENV) gzip -dc $(distdir).tar.gz | $(am__untar) ;;\
 	*.tar.bz2*) \
 	  bzip2 -dc $(distdir).tar.bz2 | $(am__untar) ;;\
-	*.tar.lzma*) \
-	  lzma -dc $(distdir).tar.lzma | $(am__untar) ;;\
+	*.tar.lz*) \
+	  lzip -dc $(distdir).tar.lz | $(am__untar) ;;\
 	*.tar.xz*) \
 	  xz -dc $(distdir).tar.xz | $(am__untar) ;;\
 	*.tar.Z*) \
@@ -486,17 +561,19 @@ distcheck: dist
 	*.zip*) \
 	  unzip $(distdir).zip ;;\
 	esac
-	chmod -R a-w $(distdir); chmod u+w $(distdir)
-	mkdir $(distdir)/_build
-	mkdir $(distdir)/_inst
+	chmod -R a-w $(distdir)
+	chmod u+w $(distdir)
+	mkdir $(distdir)/_build $(distdir)/_build/sub $(distdir)/_inst
 	chmod a-w $(distdir)
 	test -d $(distdir)/_build || exit 0; \
 	dc_install_base=`$(am__cd) $(distdir)/_inst && pwd | sed -e 's,^[^:\\/]:[\\/],/,'` \
 	  && dc_destdir="$${TMPDIR-/tmp}/am-dc-$$$$/" \
 	  && am__cwd=`pwd` \
-	  && $(am__cd) $(distdir)/_build \
-	  && ../configure --srcdir=.. --prefix="$$dc_install_base" \
+	  && $(am__cd) $(distdir)/_build/sub \
+	  && ../../configure \
+	    $(AM_DISTCHECK_CONFIGURE_FLAGS) \
 	    $(DISTCHECK_CONFIGURE_FLAGS) \
+	    --srcdir=../.. --prefix="$$dc_install_base" \
 	  && $(MAKE) $(AM_MAKEFLAGS) \
 	  && $(MAKE) $(AM_MAKEFLAGS) dvi \
 	  && $(MAKE) $(AM_MAKEFLAGS) check \
@@ -519,13 +596,21 @@ distcheck: dist
 	  && $(MAKE) $(AM_MAKEFLAGS) distcleancheck \
 	  && cd "$$am__cwd" \
 	  || exit 1
-	$(am__remove_distdir)
+	$(am__post_remove_distdir)
 	@(echo "$(distdir) archives ready for distribution: "; \
 	  list='$(DIST_ARCHIVES)'; for i in $$list; do echo $$i; done) | \
 	  sed -e 1h -e 1s/./=/g -e 1p -e 1x -e '$$p' -e '$$x'
 distuninstallcheck:
-	@$(am__cd) '$(distuninstallcheck_dir)' \
-	&& test `$(distuninstallcheck_listfiles) | wc -l` -le 1 \
+	@test -n '$(distuninstallcheck_dir)' || { \
+	  echo 'ERROR: trying to run $@ with an empty' \
+	       '$$(distuninstallcheck_dir)' >&2; \
+	  exit 1; \
+	}; \
+	$(am__cd) '$(distuninstallcheck_dir)' || { \
+	  echo 'ERROR: cannot chdir into $(distuninstallcheck_dir)' >&2; \
+	  exit 1; \
+	}; \
+	test `$(am__distuninstallcheck_listfiles) | wc -l` -eq 0 \
 	   || { echo "ERROR: files left after uninstall:" ; \
 	        if test -n "$(DESTDIR)"; then \
 	          echo "  (check DESTDIR support)"; \
@@ -556,10 +641,15 @@ install-am: all-am
 
 installcheck: installcheck-recursive
 install-strip:
-	$(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
-	  install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
-	  `test -z '$(STRIP)' || \
-	    echo "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'"` install
+	if test -z '$(STRIP)'; then \
+	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
+	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
+	      install; \
+	else \
+	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
+	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
+	    "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'" install; \
+	fi
 mostlyclean-generic:
 
 clean-generic:
@@ -640,23 +730,24 @@ ps-am:
 
 uninstall-am:
 
-.MAKE: $(RECURSIVE_CLEAN_TARGETS) $(RECURSIVE_TARGETS) ctags-recursive \
-	install-am install-strip tags-recursive
+.MAKE: $(am__recursive_targets) install-am install-strip
 
-.PHONY: $(RECURSIVE_CLEAN_TARGETS) $(RECURSIVE_TARGETS) CTAGS GTAGS \
-	all all-am am--refresh check check-am clean clean-generic \
-	ctags ctags-recursive dist dist-all dist-bzip2 dist-gzip \
-	dist-lzma dist-shar dist-tarZ dist-xz dist-zip distcheck \
-	distclean distclean-generic distclean-tags distcleancheck \
-	distdir distuninstallcheck dvi dvi-am html html-am info \
-	info-am install install-am install-data install-data-am \
-	install-dvi install-dvi-am install-exec install-exec-am \
-	install-html install-html-am install-info install-info-am \
-	install-man install-pdf install-pdf-am install-ps \
-	install-ps-am install-strip installcheck installcheck-am \
-	installdirs installdirs-am maintainer-clean \
+.PHONY: $(am__recursive_targets) CTAGS GTAGS TAGS all all-am \
+	am--refresh check check-am clean clean-cscope clean-generic \
+	cscope cscopelist-am ctags ctags-am dist dist-all dist-bzip2 \
+	dist-gzip dist-lzip dist-shar dist-tarZ dist-xz dist-zip \
+	distcheck distclean distclean-generic distclean-tags \
+	distcleancheck distdir distuninstallcheck dvi dvi-am html \
+	html-am info info-am install install-am install-data \
+	install-data-am install-dvi install-dvi-am install-exec \
+	install-exec-am install-html install-html-am install-info \
+	install-info-am install-man install-pdf install-pdf-am \
+	install-ps install-ps-am install-strip installcheck \
+	installcheck-am installdirs installdirs-am maintainer-clean \
 	maintainer-clean-generic mostlyclean mostlyclean-generic pdf \
-	pdf-am ps ps-am tags tags-recursive uninstall uninstall-am
+	pdf-am ps ps-am tags tags-am uninstall uninstall-am
+
+.PRECIOUS: Makefile
 
 
 # Tell versions [3.59,3.63) of GNU make to not export all variables.
diff --git a/aclocal.m4 b/aclocal.m4
index bcc213b4..f3018f6c 100644
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -1,7 +1,7 @@
-# generated automatically by aclocal 1.11.1 -*- Autoconf -*-
+# generated automatically by aclocal 1.15 -*- Autoconf -*-
+
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 
-# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
-# 2005, 2006, 2007, 2008, 2009  Free Software Foundation, Inc.
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
@@ -11,15 +11,16 @@
 # even the implied warranty of MERCHANTABILITY or FITNESS FOR A
 # PARTICULAR PURPOSE.
 
+m4_ifndef([AC_CONFIG_MACRO_DIRS], [m4_defun([_AM_CONFIG_MACRO_DIRS], [])m4_defun([AC_CONFIG_MACRO_DIRS], [_AM_CONFIG_MACRO_DIRS($@)])])
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
-m4_if(m4_defn([AC_AUTOCONF_VERSION]), [2.63],,
-[m4_warning([this file was generated for autoconf 2.63.
+m4_if(m4_defn([AC_AUTOCONF_VERSION]), [2.69],,
+[m4_warning([this file was generated for autoconf 2.69.
 You have another version of autoconf.  It may work, but is not guaranteed to.
 If you have problems, you may need to regenerate the build system entirely.
-To do so, use the procedure documented by the package, typically `autoreconf'.])])
+To do so, use the procedure documented by the package, typically 'autoreconf'.])])
 
-# Copyright (C) 2002, 2003, 2005, 2006, 2007, 2008  Free Software Foundation, Inc.
+# Copyright (C) 2002-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -31,10 +32,10 @@ To do so, use the procedure documented by the package, typically `autoreconf'.])
 # generated from the m4 files accompanying Automake X.Y.
 # (This private macro should not be called outside this file.)
 AC_DEFUN([AM_AUTOMAKE_VERSION],
-[am__api_version='1.11'
+[am__api_version='1.15'
 dnl Some users find AM_AUTOMAKE_VERSION and mistake it for a way to
 dnl require some minimum version.  Point them to the right macro.
-m4_if([$1], [1.11.1], [],
+m4_if([$1], [1.15], [],
       [AC_FATAL([Do not call $0, use AM_INIT_AUTOMAKE([$1]).])])dnl
 ])
 
@@ -50,22 +51,22 @@ m4_define([_AM_AUTOCONF_VERSION], [])
 # Call AM_AUTOMAKE_VERSION and AM_AUTOMAKE_VERSION so they can be traced.
 # This function is AC_REQUIREd by AM_INIT_AUTOMAKE.
 AC_DEFUN([AM_SET_CURRENT_AUTOMAKE_VERSION],
-[AM_AUTOMAKE_VERSION([1.11.1])dnl
+[AM_AUTOMAKE_VERSION([1.15])dnl
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
 _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 
 # AM_AUX_DIR_EXPAND                                         -*- Autoconf -*-
 
-# Copyright (C) 2001, 2003, 2005  Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
 # For projects using AC_CONFIG_AUX_DIR([foo]), Autoconf sets
-# $ac_aux_dir to `$srcdir/foo'.  In other projects, it is set to
-# `$srcdir', `$srcdir/..', or `$srcdir/../..'.
+# $ac_aux_dir to '$srcdir/foo'.  In other projects, it is set to
+# '$srcdir', '$srcdir/..', or '$srcdir/../..'.
 #
 # Of course, Automake must honor this variable whenever it calls a
 # tool from the auxiliary directory.  The problem is that $srcdir (and
@@ -84,7 +85,7 @@ _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 #
 # The reason of the latter failure is that $top_srcdir and $ac_aux_dir
 # are both prefixed by $srcdir.  In an in-source build this is usually
-# harmless because $srcdir is `.', but things will broke when you
+# harmless because $srcdir is '.', but things will broke when you
 # start a VPATH build or use an absolute $srcdir.
 #
 # So we could use something similar to $top_srcdir/$ac_aux_dir/missing,
@@ -102,30 +103,26 @@ _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 # configured tree to be moved without reconfiguration.
 
 AC_DEFUN([AM_AUX_DIR_EXPAND],
-[dnl Rely on autoconf to set up CDPATH properly.
-AC_PREREQ([2.50])dnl
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+[AC_REQUIRE([AC_CONFIG_AUX_DIR_DEFAULT])dnl
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 ])
 
 # AM_CONDITIONAL                                            -*- Autoconf -*-
 
-# Copyright (C) 1997, 2000, 2001, 2003, 2004, 2005, 2006, 2008
-# Free Software Foundation, Inc.
+# Copyright (C) 1997-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 9
-
 # AM_CONDITIONAL(NAME, SHELL-CONDITION)
 # -------------------------------------
 # Define a conditional.
 AC_DEFUN([AM_CONDITIONAL],
-[AC_PREREQ(2.52)dnl
- ifelse([$1], [TRUE],  [AC_FATAL([$0: invalid condition: $1])],
-	[$1], [FALSE], [AC_FATAL([$0: invalid condition: $1])])dnl
+[AC_PREREQ([2.52])dnl
+ m4_if([$1], [TRUE],  [AC_FATAL([$0: invalid condition: $1])],
+       [$1], [FALSE], [AC_FATAL([$0: invalid condition: $1])])dnl
 AC_SUBST([$1_TRUE])dnl
 AC_SUBST([$1_FALSE])dnl
 _AM_SUBST_NOTMAKE([$1_TRUE])dnl
@@ -144,16 +141,14 @@ AC_CONFIG_COMMANDS_PRE(
 Usually this means the macro was only invoked conditionally.]])
 fi])])
 
-# Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2009
-# Free Software Foundation, Inc.
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 10
 
-# There are a few dirty hacks below to avoid letting `AC_PROG_CC' be
+# There are a few dirty hacks below to avoid letting 'AC_PROG_CC' be
 # written in clear, in which case automake, when reading aclocal.m4,
 # will think it sees a *use*, and therefore will trigger all it's
 # C support machinery.  Also note that it means that autoscan, seeing
@@ -163,7 +158,7 @@ fi])])
 # _AM_DEPENDENCIES(NAME)
 # ----------------------
 # See how the compiler implements dependency checking.
-# NAME is "CC", "CXX", "GCJ", or "OBJC".
+# NAME is "CC", "CXX", "OBJC", "OBJCXX", "UPC", or "GJC".
 # We try a few techniques and use that to set a single cache variable.
 #
 # We don't AC_REQUIRE the corresponding AC_PROG_CC since the latter was
@@ -176,12 +171,13 @@ AC_REQUIRE([AM_OUTPUT_DEPENDENCY_COMMANDS])dnl
 AC_REQUIRE([AM_MAKE_INCLUDE])dnl
 AC_REQUIRE([AM_DEP_TRACK])dnl
 
-ifelse([$1], CC,   [depcc="$CC"   am_compiler_list=],
-       [$1], CXX,  [depcc="$CXX"  am_compiler_list=],
-       [$1], OBJC, [depcc="$OBJC" am_compiler_list='gcc3 gcc'],
-       [$1], UPC,  [depcc="$UPC"  am_compiler_list=],
-       [$1], GCJ,  [depcc="$GCJ"  am_compiler_list='gcc3 gcc'],
-                   [depcc="$$1"   am_compiler_list=])
+m4_if([$1], [CC],   [depcc="$CC"   am_compiler_list=],
+      [$1], [CXX],  [depcc="$CXX"  am_compiler_list=],
+      [$1], [OBJC], [depcc="$OBJC" am_compiler_list='gcc3 gcc'],
+      [$1], [OBJCXX], [depcc="$OBJCXX" am_compiler_list='gcc3 gcc'],
+      [$1], [UPC],  [depcc="$UPC"  am_compiler_list=],
+      [$1], [GCJ],  [depcc="$GCJ"  am_compiler_list='gcc3 gcc'],
+                    [depcc="$$1"   am_compiler_list=])
 
 AC_CACHE_CHECK([dependency style of $depcc],
                [am_cv_$1_dependencies_compiler_type],
@@ -189,8 +185,9 @@ AC_CACHE_CHECK([dependency style of $depcc],
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named `D' -- because `-MD' means `put the output
-  # in D'.
+  # making a dummy file named 'D' -- because '-MD' means "put the output
+  # in D".
+  rm -rf conftest.dir
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -229,16 +226,16 @@ AC_CACHE_CHECK([dependency style of $depcc],
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
-      # Solaris 8's {/usr,}/bin/sh.
-      touch sub/conftst$i.h
+      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
+      # Solaris 10 /bin/sh.
+      echo '/* dummy */' > sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with `-c' and `-o' for the sake of the "dashmstdout"
+    # We check with '-c' and '-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle `-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs
+    # handle '-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs.
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -247,16 +244,16 @@ AC_CACHE_CHECK([dependency style of $depcc],
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # after this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested
+      # After this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested.
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvisualcpp | msvcmsys)
-      # This compiler won't grok `-c -o', but also, the minuso test has
+    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
+      # This compiler won't grok '-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -304,7 +301,7 @@ AM_CONDITIONAL([am__fastdep$1], [
 # AM_SET_DEPDIR
 # -------------
 # Choose a directory name for dependency files.
-# This macro is AC_REQUIREd in _AM_DEPENDENCIES
+# This macro is AC_REQUIREd in _AM_DEPENDENCIES.
 AC_DEFUN([AM_SET_DEPDIR],
 [AC_REQUIRE([AM_SET_LEADING_DOT])dnl
 AC_SUBST([DEPDIR], ["${am__leading_dot}deps"])dnl
@@ -314,34 +311,39 @@ AC_SUBST([DEPDIR], ["${am__leading_dot}deps"])dnl
 # AM_DEP_TRACK
 # ------------
 AC_DEFUN([AM_DEP_TRACK],
-[AC_ARG_ENABLE(dependency-tracking,
-[  --disable-dependency-tracking  speeds up one-time build
-  --enable-dependency-tracking   do not reject slow dependency extractors])
+[AC_ARG_ENABLE([dependency-tracking], [dnl
+AS_HELP_STRING(
+  [--enable-dependency-tracking],
+  [do not reject slow dependency extractors])
+AS_HELP_STRING(
+  [--disable-dependency-tracking],
+  [speeds up one-time build])])
 if test "x$enable_dependency_tracking" != xno; then
   am_depcomp="$ac_aux_dir/depcomp"
   AMDEPBACKSLASH='\'
+  am__nodep='_no'
 fi
 AM_CONDITIONAL([AMDEP], [test "x$enable_dependency_tracking" != xno])
 AC_SUBST([AMDEPBACKSLASH])dnl
 _AM_SUBST_NOTMAKE([AMDEPBACKSLASH])dnl
+AC_SUBST([am__nodep])dnl
+_AM_SUBST_NOTMAKE([am__nodep])dnl
 ])
 
 # Generate code to set up dependency tracking.              -*- Autoconf -*-
 
-# Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2008
-# Free Software Foundation, Inc.
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-#serial 5
 
 # _AM_OUTPUT_DEPENDENCY_COMMANDS
 # ------------------------------
 AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
 [{
-  # Autoconf 2.62 quotes --file arguments for eval, but not when files
+  # Older Autoconf quotes --file arguments for eval, but not when files
   # are listed without --file.  Let's play safe and only enable the eval
   # if we detect the quoting.
   case $CONFIG_FILES in
@@ -354,7 +356,7 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
     # Strip MF so we end up with the name of the file.
     mf=`echo "$mf" | sed -e 's/:.*$//'`
     # Check whether this is an Automake generated Makefile or not.
-    # We used to match only the files named `Makefile.in', but
+    # We used to match only the files named 'Makefile.in', but
     # some people rename them; so instead we look at the file content.
     # Grep'ing the first line is not enough: some people post-process
     # each Makefile.in and add a new line on top of each file to say so.
@@ -366,21 +368,19 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
       continue
     fi
     # Extract the definition of DEPDIR, am__include, and am__quote
-    # from the Makefile without running `make'.
+    # from the Makefile without running 'make'.
     DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
     test -z "$DEPDIR" && continue
     am__include=`sed -n 's/^am__include = //p' < "$mf"`
-    test -z "am__include" && continue
+    test -z "$am__include" && continue
     am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
-    # When using ansi2knr, U may be empty or an underscore; expand it
-    U=`sed -n 's/^U = //p' < "$mf"`
     # Find all dependency output files, they are included files with
     # $(DEPDIR) in their names.  We invoke sed twice because it is the
     # simplest approach to changing $(DEPDIR) to its actual value in the
     # expansion.
     for file in `sed -n "
       s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
-	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g' -e 's/\$U/'"$U"'/g'`; do
+	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
       # Make sure the directory exists.
       test -f "$dirpart/$file" && continue
       fdir=`AS_DIRNAME(["$file"])`
@@ -398,7 +398,7 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
 # This macro should only be invoked once -- use via AC_REQUIRE.
 #
 # This code is only required when automatic dependency tracking
-# is enabled.  FIXME.  This creates each `.P' file that we will
+# is enabled.  FIXME.  This creates each '.P' file that we will
 # need in order to bootstrap the dependency handling code.
 AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 [AC_CONFIG_COMMANDS([depfiles],
@@ -408,18 +408,21 @@ AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 
 # Do all the work for Automake.                             -*- Autoconf -*-
 
-# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
-# 2005, 2006, 2008, 2009 Free Software Foundation, Inc.
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 16
-
 # This macro actually does too much.  Some checks are only needed if
 # your package does certain things.  But this isn't really a big deal.
 
+dnl Redefine AC_PROG_CC to automatically invoke _AM_PROG_CC_C_O.
+m4_define([AC_PROG_CC],
+m4_defn([AC_PROG_CC])
+[_AM_PROG_CC_C_O
+])
+
 # AM_INIT_AUTOMAKE(PACKAGE, VERSION, [NO-DEFINE])
 # AM_INIT_AUTOMAKE([OPTIONS])
 # -----------------------------------------------
@@ -432,7 +435,7 @@ AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 # arguments mandatory, and then we can depend on a new Autoconf
 # release and drop the old call support.
 AC_DEFUN([AM_INIT_AUTOMAKE],
-[AC_PREREQ([2.62])dnl
+[AC_PREREQ([2.65])dnl
 dnl Autoconf wants to disallow AM_ names.  We explicitly allow
 dnl the ones we care about.
 m4_pattern_allow([^AM_[A-Z]+FLAGS$])dnl
@@ -461,33 +464,42 @@ AC_SUBST([CYGPATH_W])
 # Define the identity of the package.
 dnl Distinguish between old-style and new-style calls.
 m4_ifval([$2],
-[m4_ifval([$3], [_AM_SET_OPTION([no-define])])dnl
+[AC_DIAGNOSE([obsolete],
+             [$0: two- and three-arguments forms are deprecated.])
+m4_ifval([$3], [_AM_SET_OPTION([no-define])])dnl
  AC_SUBST([PACKAGE], [$1])dnl
  AC_SUBST([VERSION], [$2])],
 [_AM_SET_OPTIONS([$1])dnl
 dnl Diagnose old-style AC_INIT with new-style AM_AUTOMAKE_INIT.
-m4_if(m4_ifdef([AC_PACKAGE_NAME], 1)m4_ifdef([AC_PACKAGE_VERSION], 1), 11,,
+m4_if(
+  m4_ifdef([AC_PACKAGE_NAME], [ok]):m4_ifdef([AC_PACKAGE_VERSION], [ok]),
+  [ok:ok],,
   [m4_fatal([AC_INIT should be called with package and version arguments])])dnl
  AC_SUBST([PACKAGE], ['AC_PACKAGE_TARNAME'])dnl
  AC_SUBST([VERSION], ['AC_PACKAGE_VERSION'])])dnl
 
 _AM_IF_OPTION([no-define],,
-[AC_DEFINE_UNQUOTED(PACKAGE, "$PACKAGE", [Name of package])
- AC_DEFINE_UNQUOTED(VERSION, "$VERSION", [Version number of package])])dnl
+[AC_DEFINE_UNQUOTED([PACKAGE], ["$PACKAGE"], [Name of package])
+ AC_DEFINE_UNQUOTED([VERSION], ["$VERSION"], [Version number of package])])dnl
 
 # Some tools Automake needs.
 AC_REQUIRE([AM_SANITY_CHECK])dnl
 AC_REQUIRE([AC_ARG_PROGRAM])dnl
-AM_MISSING_PROG(ACLOCAL, aclocal-${am__api_version})
-AM_MISSING_PROG(AUTOCONF, autoconf)
-AM_MISSING_PROG(AUTOMAKE, automake-${am__api_version})
-AM_MISSING_PROG(AUTOHEADER, autoheader)
-AM_MISSING_PROG(MAKEINFO, makeinfo)
+AM_MISSING_PROG([ACLOCAL], [aclocal-${am__api_version}])
+AM_MISSING_PROG([AUTOCONF], [autoconf])
+AM_MISSING_PROG([AUTOMAKE], [automake-${am__api_version}])
+AM_MISSING_PROG([AUTOHEADER], [autoheader])
+AM_MISSING_PROG([MAKEINFO], [makeinfo])
 AC_REQUIRE([AM_PROG_INSTALL_SH])dnl
 AC_REQUIRE([AM_PROG_INSTALL_STRIP])dnl
-AC_REQUIRE([AM_PROG_MKDIR_P])dnl
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
+AC_REQUIRE([AC_PROG_MKDIR_P])dnl
+# For better backward compatibility.  To be removed once Automake 1.9.x
+# dies out for good.  For more background, see:
+# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
+# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
+AC_SUBST([mkdir_p], ['$(MKDIR_P)'])
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
 AC_REQUIRE([AC_PROG_AWK])dnl
 AC_REQUIRE([AC_PROG_MAKE_SET])dnl
 AC_REQUIRE([AM_SET_LEADING_DOT])dnl
@@ -496,34 +508,82 @@ _AM_IF_OPTION([tar-ustar], [_AM_PROG_TAR([ustar])],
 			     [_AM_PROG_TAR([v7])])])
 _AM_IF_OPTION([no-dependencies],,
 [AC_PROVIDE_IFELSE([AC_PROG_CC],
-		  [_AM_DEPENDENCIES(CC)],
-		  [define([AC_PROG_CC],
-			  defn([AC_PROG_CC])[_AM_DEPENDENCIES(CC)])])dnl
+		  [_AM_DEPENDENCIES([CC])],
+		  [m4_define([AC_PROG_CC],
+			     m4_defn([AC_PROG_CC])[_AM_DEPENDENCIES([CC])])])dnl
 AC_PROVIDE_IFELSE([AC_PROG_CXX],
-		  [_AM_DEPENDENCIES(CXX)],
-		  [define([AC_PROG_CXX],
-			  defn([AC_PROG_CXX])[_AM_DEPENDENCIES(CXX)])])dnl
+		  [_AM_DEPENDENCIES([CXX])],
+		  [m4_define([AC_PROG_CXX],
+			     m4_defn([AC_PROG_CXX])[_AM_DEPENDENCIES([CXX])])])dnl
 AC_PROVIDE_IFELSE([AC_PROG_OBJC],
-		  [_AM_DEPENDENCIES(OBJC)],
-		  [define([AC_PROG_OBJC],
-			  defn([AC_PROG_OBJC])[_AM_DEPENDENCIES(OBJC)])])dnl
+		  [_AM_DEPENDENCIES([OBJC])],
+		  [m4_define([AC_PROG_OBJC],
+			     m4_defn([AC_PROG_OBJC])[_AM_DEPENDENCIES([OBJC])])])dnl
+AC_PROVIDE_IFELSE([AC_PROG_OBJCXX],
+		  [_AM_DEPENDENCIES([OBJCXX])],
+		  [m4_define([AC_PROG_OBJCXX],
+			     m4_defn([AC_PROG_OBJCXX])[_AM_DEPENDENCIES([OBJCXX])])])dnl
 ])
-_AM_IF_OPTION([silent-rules], [AC_REQUIRE([AM_SILENT_RULES])])dnl
-dnl The `parallel-tests' driver may need to know about EXEEXT, so add the
-dnl `am__EXEEXT' conditional if _AM_COMPILER_EXEEXT was seen.  This macro
-dnl is hooked onto _AC_COMPILER_EXEEXT early, see below.
+AC_REQUIRE([AM_SILENT_RULES])dnl
+dnl The testsuite driver may need to know about EXEEXT, so add the
+dnl 'am__EXEEXT' conditional if _AM_COMPILER_EXEEXT was seen.  This
+dnl macro is hooked onto _AC_COMPILER_EXEEXT early, see below.
 AC_CONFIG_COMMANDS_PRE(dnl
 [m4_provide_if([_AM_COMPILER_EXEEXT],
   [AM_CONDITIONAL([am__EXEEXT], [test -n "$EXEEXT"])])])dnl
+
+# POSIX will say in a future version that running "rm -f" with no argument
+# is OK; and we want to be able to make that assumption in our Makefile
+# recipes.  So use an aggressive probe to check that the usage we want is
+# actually supported "in the wild" to an acceptable degree.
+# See automake bug#10828.
+# To make any issue more visible, cause the running configure to be aborted
+# by default if the 'rm' program in use doesn't match our expectations; the
+# user can still override this though.
+if rm -f && rm -fr && rm -rf; then : OK; else
+  cat >&2 <<'END'
+Oops!
+
+Your 'rm' program seems unable to run without file operands specified
+on the command line, even when the '-f' option is present.  This is contrary
+to the behaviour of most rm programs out there, and not conforming with
+the upcoming POSIX standard: <http://austingroupbugs.net/view.php?id=542>
+
+Please tell bug-automake@gnu.org about your system, including the value
+of your $PATH and any error possibly output before this message.  This
+can help us improve future automake versions.
+
+END
+  if test x"$ACCEPT_INFERIOR_RM_PROGRAM" = x"yes"; then
+    echo 'Configuration will proceed anyway, since you have set the' >&2
+    echo 'ACCEPT_INFERIOR_RM_PROGRAM variable to "yes"' >&2
+    echo >&2
+  else
+    cat >&2 <<'END'
+Aborting the configuration process, to ensure you take notice of the issue.
+
+You can download and install GNU coreutils to get an 'rm' implementation
+that behaves properly: <http://www.gnu.org/software/coreutils/>.
+
+If you want to complete the configuration process using your problematic
+'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
+to "yes", and re-run configure.
+
+END
+    AC_MSG_ERROR([Your 'rm' program is bad, sorry.])
+  fi
+fi
+dnl The trailing newline in this macro's definition is deliberate, for
+dnl backward compatibility and to allow trailing 'dnl'-style comments
+dnl after the AM_INIT_AUTOMAKE invocation. See automake bug#16841.
 ])
 
-dnl Hook into `_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
+dnl Hook into '_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
 dnl add the conditional right here, as _AC_COMPILER_EXEEXT may be further
 dnl mangled by Autoconf and run in a shell conditional statement.
 m4_define([_AC_COMPILER_EXEEXT],
 m4_defn([_AC_COMPILER_EXEEXT])[m4_provide([_AM_COMPILER_EXEEXT])])
 
-
 # When config.status generates a header, we must update the stamp-h file.
 # This file resides in the same directory as the config header
 # that is generated.  The stamp files are numbered to have different names.
@@ -545,7 +605,7 @@ for _am_header in $config_headers :; do
 done
 echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_count])
 
-# Copyright (C) 2001, 2003, 2005, 2008  Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -556,7 +616,7 @@ echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_co
 # Define $install_sh.
 AC_DEFUN([AM_PROG_INSTALL_SH],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -564,16 +624,14 @@ if test x"${install_sh}" != xset; then
     install_sh="\${SHELL} $am_aux_dir/install-sh"
   esac
 fi
-AC_SUBST(install_sh)])
+AC_SUBST([install_sh])])
 
-# Copyright (C) 2003, 2005  Free Software Foundation, Inc.
+# Copyright (C) 2003-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 2
-
 # Check whether the underlying file-system supports filenames
 # with a leading dot.  For instance MS-DOS doesn't.
 AC_DEFUN([AM_SET_LEADING_DOT],
@@ -589,14 +647,12 @@ AC_SUBST([am__leading_dot])])
 
 # Check to see how 'make' treats includes.	            -*- Autoconf -*-
 
-# Copyright (C) 2001, 2002, 2003, 2005, 2009  Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 4
-
 # AM_MAKE_INCLUDE()
 # -----------------
 # Check to see how make treats includes.
@@ -614,7 +670,7 @@ am__quote=
 _am_result=none
 # First try GNU make style include.
 echo "include confinc" > confmf
-# Ignore all kinds of additional output from `make'.
+# Ignore all kinds of additional output from 'make'.
 case `$am_make -s -f confmf 2> /dev/null` in #(
 *the\ am__doit\ target*)
   am__include=include
@@ -641,15 +697,12 @@ rm -f confinc confmf
 
 # Fake the existence of programs that GNU maintainers use.  -*- Autoconf -*-
 
-# Copyright (C) 1997, 1999, 2000, 2001, 2003, 2004, 2005, 2008
-# Free Software Foundation, Inc.
+# Copyright (C) 1997-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 6
-
 # AM_MISSING_PROG(NAME, PROGRAM)
 # ------------------------------
 AC_DEFUN([AM_MISSING_PROG],
@@ -657,11 +710,10 @@ AC_DEFUN([AM_MISSING_PROG],
 $1=${$1-"${am_missing_run}$2"}
 AC_SUBST($1)])
 
-
 # AM_MISSING_HAS_RUN
 # ------------------
-# Define MISSING if not defined so far and test if it supports --run.
-# If it does, set am_missing_run to use it, otherwise, to nothing.
+# Define MISSING if not defined so far and test if it is modern enough.
+# If it is, set am_missing_run to use it, otherwise, to nothing.
 AC_DEFUN([AM_MISSING_HAS_RUN],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
 AC_REQUIRE_AUX_FILE([missing])dnl
@@ -674,63 +726,35 @@ if test x"${MISSING+set}" != xset; then
   esac
 fi
 # Use eval to expand $SHELL
-if eval "$MISSING --run true"; then
-  am_missing_run="$MISSING --run "
+if eval "$MISSING --is-lightweight"; then
+  am_missing_run="$MISSING "
 else
   am_missing_run=
-  AC_MSG_WARN([`missing' script is too old or missing])
+  AC_MSG_WARN(['missing' script is too old or missing])
 fi
 ])
 
-# Copyright (C) 2003, 2004, 2005, 2006  Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# AM_PROG_MKDIR_P
-# ---------------
-# Check for `mkdir -p'.
-AC_DEFUN([AM_PROG_MKDIR_P],
-[AC_PREREQ([2.60])dnl
-AC_REQUIRE([AC_PROG_MKDIR_P])dnl
-dnl Automake 1.8 to 1.9.6 used to define mkdir_p.  We now use MKDIR_P,
-dnl while keeping a definition of mkdir_p for backward compatibility.
-dnl @MKDIR_P@ is magic: AC_OUTPUT adjusts its value for each Makefile.
-dnl However we cannot define mkdir_p as $(MKDIR_P) for the sake of
-dnl Makefile.ins that do not define MKDIR_P, so we do our own
-dnl adjustment using top_builddir (which is defined more often than
-dnl MKDIR_P).
-AC_SUBST([mkdir_p], ["$MKDIR_P"])dnl
-case $mkdir_p in
-  [[\\/$]]* | ?:[[\\/]]*) ;;
-  */*) mkdir_p="\$(top_builddir)/$mkdir_p" ;;
-esac
-])
-
 # Helper functions for option handling.                     -*- Autoconf -*-
 
-# Copyright (C) 2001, 2002, 2003, 2005, 2008  Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 4
-
 # _AM_MANGLE_OPTION(NAME)
 # -----------------------
 AC_DEFUN([_AM_MANGLE_OPTION],
 [[_AM_OPTION_]m4_bpatsubst($1, [[^a-zA-Z0-9_]], [_])])
 
 # _AM_SET_OPTION(NAME)
-# ------------------------------
+# --------------------
 # Set option NAME.  Presently that only means defining a flag for this option.
 AC_DEFUN([_AM_SET_OPTION],
-[m4_define(_AM_MANGLE_OPTION([$1]), 1)])
+[m4_define(_AM_MANGLE_OPTION([$1]), [1])])
 
 # _AM_SET_OPTIONS(OPTIONS)
-# ----------------------------------
+# ------------------------
 # OPTIONS is a space-separated list of Automake options.
 AC_DEFUN([_AM_SET_OPTIONS],
 [m4_foreach_w([_AM_Option], [$1], [_AM_SET_OPTION(_AM_Option)])])
@@ -741,24 +765,82 @@ AC_DEFUN([_AM_SET_OPTIONS],
 AC_DEFUN([_AM_IF_OPTION],
 [m4_ifset(_AM_MANGLE_OPTION([$1]), [$2], [$3])])
 
-# Check to make sure that the build environment is sane.    -*- Autoconf -*-
-
-# Copyright (C) 1996, 1997, 2000, 2001, 2003, 2005, 2008
-# Free Software Foundation, Inc.
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 5
+# _AM_PROG_CC_C_O
+# ---------------
+# Like AC_PROG_CC_C_O, but changed for automake.  We rewrite AC_PROG_CC
+# to automatically call this.
+AC_DEFUN([_AM_PROG_CC_C_O],
+[AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
+AC_REQUIRE_AUX_FILE([compile])dnl
+AC_LANG_PUSH([C])dnl
+AC_CACHE_CHECK(
+  [whether $CC understands -c and -o together],
+  [am_cv_prog_cc_c_o],
+  [AC_LANG_CONFTEST([AC_LANG_PROGRAM([])])
+  # Make sure it works both with $CC and with simple cc.
+  # Following AC_PROG_CC_C_O, we do the test twice because some
+  # compilers refuse to overwrite an existing .o file with -o,
+  # though they will create one.
+  am_cv_prog_cc_c_o=yes
+  for am_i in 1 2; do
+    if AM_RUN_LOG([$CC -c conftest.$ac_ext -o conftest2.$ac_objext]) \
+         && test -f conftest2.$ac_objext; then
+      : OK
+    else
+      am_cv_prog_cc_c_o=no
+      break
+    fi
+  done
+  rm -f core conftest*
+  unset am_i])
+if test "$am_cv_prog_cc_c_o" != yes; then
+   # Losing compiler, so override with the script.
+   # FIXME: It is wrong to rewrite CC.
+   # But if we don't then we get into trouble of one sort or another.
+   # A longer-term fix would be to have automake use am__CC in this case,
+   # and then we could set am__CC="\$(top_srcdir)/compile \$(CC)"
+   CC="$am_aux_dir/compile $CC"
+fi
+AC_LANG_POP([C])])
+
+# For backward compatibility.
+AC_DEFUN_ONCE([AM_PROG_CC_C_O], [AC_REQUIRE([AC_PROG_CC])])
+
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# AM_RUN_LOG(COMMAND)
+# -------------------
+# Run COMMAND, save the exit status in ac_status, and log it.
+# (This has been adapted from Autoconf's _AC_RUN_LOG macro.)
+AC_DEFUN([AM_RUN_LOG],
+[{ echo "$as_me:$LINENO: $1" >&AS_MESSAGE_LOG_FD
+   ($1) >&AS_MESSAGE_LOG_FD 2>&AS_MESSAGE_LOG_FD
+   ac_status=$?
+   echo "$as_me:$LINENO: \$? = $ac_status" >&AS_MESSAGE_LOG_FD
+   (exit $ac_status); }])
+
+# Check to make sure that the build environment is sane.    -*- Autoconf -*-
+
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
 
 # AM_SANITY_CHECK
 # ---------------
 AC_DEFUN([AM_SANITY_CHECK],
 [AC_MSG_CHECKING([whether build environment is sane])
-# Just in case
-sleep 1
-echo timestamp > conftest.file
 # Reject unsafe characters in $srcdir or the absolute working directory
 # name.  Accept space and tab only in the latter.
 am_lf='
@@ -769,32 +851,40 @@ case `pwd` in
 esac
 case $srcdir in
   *[[\\\"\#\$\&\'\`$am_lf\ \	]]*)
-    AC_MSG_ERROR([unsafe srcdir value: `$srcdir']);;
+    AC_MSG_ERROR([unsafe srcdir value: '$srcdir']);;
 esac
 
-# Do `set' in a subshell so we don't clobber the current shell's
+# Do 'set' in a subshell so we don't clobber the current shell's
 # arguments.  Must try -L first in case configure is actually a
 # symlink; some systems play weird games with the mod time of symlinks
 # (eg FreeBSD returns the mod time of the symlink's containing
 # directory).
 if (
-   set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
-   if test "$[*]" = "X"; then
-      # -L didn't work.
-      set X `ls -t "$srcdir/configure" conftest.file`
-   fi
-   rm -f conftest.file
-   if test "$[*]" != "X $srcdir/configure conftest.file" \
-      && test "$[*]" != "X conftest.file $srcdir/configure"; then
-
-      # If neither matched, then we have a broken ls.  This can happen
-      # if, for instance, CONFIG_SHELL is bash and it inherits a
-      # broken ls alias from the environment.  This has actually
-      # happened.  Such a system could not be considered "sane".
-      AC_MSG_ERROR([ls -t appears to fail.  Make sure there is not a broken
-alias in your environment])
-   fi
+   am_has_slept=no
+   for am_try in 1 2; do
+     echo "timestamp, slept: $am_has_slept" > conftest.file
+     set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
+     if test "$[*]" = "X"; then
+	# -L didn't work.
+	set X `ls -t "$srcdir/configure" conftest.file`
+     fi
+     if test "$[*]" != "X $srcdir/configure conftest.file" \
+	&& test "$[*]" != "X conftest.file $srcdir/configure"; then
 
+	# If neither matched, then we have a broken ls.  This can happen
+	# if, for instance, CONFIG_SHELL is bash and it inherits a
+	# broken ls alias from the environment.  This has actually
+	# happened.  Such a system could not be considered "sane".
+	AC_MSG_ERROR([ls -t appears to fail.  Make sure there is not a broken
+  alias in your environment])
+     fi
+     if test "$[2]" = conftest.file || test $am_try -eq 2; then
+       break
+     fi
+     # Just in case.
+     sleep 1
+     am_has_slept=yes
+   done
    test "$[2]" = conftest.file
    )
 then
@@ -804,9 +894,85 @@ else
    AC_MSG_ERROR([newly created file is older than distributed files!
 Check your system clock])
 fi
-AC_MSG_RESULT(yes)])
+AC_MSG_RESULT([yes])
+# If we didn't sleep, we still need to ensure time stamps of config.status and
+# generated files are strictly newer.
+am_sleep_pid=
+if grep 'slept: no' conftest.file >/dev/null 2>&1; then
+  ( sleep 1 ) &
+  am_sleep_pid=$!
+fi
+AC_CONFIG_COMMANDS_PRE(
+  [AC_MSG_CHECKING([that generated files are newer than configure])
+   if test -n "$am_sleep_pid"; then
+     # Hide warnings about reused PIDs.
+     wait $am_sleep_pid 2>/dev/null
+   fi
+   AC_MSG_RESULT([done])])
+rm -f conftest.file
+])
 
-# Copyright (C) 2001, 2003, 2005  Free Software Foundation, Inc.
+# Copyright (C) 2009-2014 Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# AM_SILENT_RULES([DEFAULT])
+# --------------------------
+# Enable less verbose build rules; with the default set to DEFAULT
+# ("yes" being less verbose, "no" or empty being verbose).
+AC_DEFUN([AM_SILENT_RULES],
+[AC_ARG_ENABLE([silent-rules], [dnl
+AS_HELP_STRING(
+  [--enable-silent-rules],
+  [less verbose build output (undo: "make V=1")])
+AS_HELP_STRING(
+  [--disable-silent-rules],
+  [verbose build output (undo: "make V=0")])dnl
+])
+case $enable_silent_rules in @%:@ (((
+  yes) AM_DEFAULT_VERBOSITY=0;;
+   no) AM_DEFAULT_VERBOSITY=1;;
+    *) AM_DEFAULT_VERBOSITY=m4_if([$1], [yes], [0], [1]);;
+esac
+dnl
+dnl A few 'make' implementations (e.g., NonStop OS and NextStep)
+dnl do not support nested variable expansions.
+dnl See automake bug#9928 and bug#10237.
+am_make=${MAKE-make}
+AC_CACHE_CHECK([whether $am_make supports nested variables],
+   [am_cv_make_support_nested_variables],
+   [if AS_ECHO([['TRUE=$(BAR$(V))
+BAR0=false
+BAR1=true
+V=1
+am__doit:
+	@$(TRUE)
+.PHONY: am__doit']]) | $am_make -f - >/dev/null 2>&1; then
+  am_cv_make_support_nested_variables=yes
+else
+  am_cv_make_support_nested_variables=no
+fi])
+if test $am_cv_make_support_nested_variables = yes; then
+  dnl Using '$V' instead of '$(V)' breaks IRIX make.
+  AM_V='$(V)'
+  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
+else
+  AM_V=$AM_DEFAULT_VERBOSITY
+  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
+fi
+AC_SUBST([AM_V])dnl
+AM_SUBST_NOTMAKE([AM_V])dnl
+AC_SUBST([AM_DEFAULT_V])dnl
+AM_SUBST_NOTMAKE([AM_DEFAULT_V])dnl
+AC_SUBST([AM_DEFAULT_VERBOSITY])dnl
+AM_BACKSLASH='\'
+AC_SUBST([AM_BACKSLASH])dnl
+_AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
+])
+
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -814,34 +980,32 @@ AC_MSG_RESULT(yes)])
 
 # AM_PROG_INSTALL_STRIP
 # ---------------------
-# One issue with vendor `install' (even GNU) is that you can't
+# One issue with vendor 'install' (even GNU) is that you can't
 # specify the program used to strip binaries.  This is especially
 # annoying in cross-compiling environments, where the build's strip
 # is unlikely to handle the host's binaries.
 # Fortunately install-sh will honor a STRIPPROG variable, so we
-# always use install-sh in `make install-strip', and initialize
+# always use install-sh in "make install-strip", and initialize
 # STRIPPROG with the value of the STRIP variable (set by the user).
 AC_DEFUN([AM_PROG_INSTALL_STRIP],
 [AC_REQUIRE([AM_PROG_INSTALL_SH])dnl
-# Installed binaries are usually stripped using `strip' when the user
-# run `make install-strip'.  However `strip' might not be the right
+# Installed binaries are usually stripped using 'strip' when the user
+# run "make install-strip".  However 'strip' might not be the right
 # tool to use in cross-compilation environments, therefore Automake
-# will honor the `STRIP' environment variable to overrule this program.
-dnl Don't test for $cross_compiling = yes, because it might be `maybe'.
+# will honor the 'STRIP' environment variable to overrule this program.
+dnl Don't test for $cross_compiling = yes, because it might be 'maybe'.
 if test "$cross_compiling" != no; then
   AC_CHECK_TOOL([STRIP], [strip], :)
 fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 AC_SUBST([INSTALL_STRIP_PROGRAM])])
 
-# Copyright (C) 2006, 2008  Free Software Foundation, Inc.
+# Copyright (C) 2006-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 2
-
 # _AM_SUBST_NOTMAKE(VARIABLE)
 # ---------------------------
 # Prevent Automake from outputting VARIABLE = @VARIABLE@ in Makefile.in.
@@ -849,24 +1013,22 @@ AC_SUBST([INSTALL_STRIP_PROGRAM])])
 AC_DEFUN([_AM_SUBST_NOTMAKE])
 
 # AM_SUBST_NOTMAKE(VARIABLE)
-# ---------------------------
+# --------------------------
 # Public sister of _AM_SUBST_NOTMAKE.
 AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 
 # Check how to create a tarball.                            -*- Autoconf -*-
 
-# Copyright (C) 2004, 2005  Free Software Foundation, Inc.
+# Copyright (C) 2004-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
-# serial 2
-
 # _AM_PROG_TAR(FORMAT)
 # --------------------
 # Check how to create a tarball in format FORMAT.
-# FORMAT should be one of `v7', `ustar', or `pax'.
+# FORMAT should be one of 'v7', 'ustar', or 'pax'.
 #
 # Substitute a variable $(am__tar) that is a command
 # writing to stdout a FORMAT-tarball containing the directory
@@ -876,75 +1038,114 @@ AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 # Substitute a variable $(am__untar) that extract such
 # a tarball read from stdin.
 #     $(am__untar) < result.tar
+#
 AC_DEFUN([_AM_PROG_TAR],
-[# Always define AMTAR for backward compatibility.
-AM_MISSING_PROG([AMTAR], [tar])
-m4_if([$1], [v7],
-     [am__tar='${AMTAR} chof - "$$tardir"'; am__untar='${AMTAR} xf -'],
-     [m4_case([$1], [ustar],, [pax],,
-              [m4_fatal([Unknown tar format])])
-AC_MSG_CHECKING([how to create a $1 tar archive])
-# Loop over all known methods to create a tar archive until one works.
+[# Always define AMTAR for backward compatibility.  Yes, it's still used
+# in the wild :-(  We should find a proper way to deprecate it ...
+AC_SUBST([AMTAR], ['$${TAR-tar}'])
+
+# We'll loop over all known methods to create a tar archive until one works.
 _am_tools='gnutar m4_if([$1], [ustar], [plaintar]) pax cpio none'
-_am_tools=${am_cv_prog_tar_$1-$_am_tools}
-# Do not fold the above two line into one, because Tru64 sh and
-# Solaris sh will not grok spaces in the rhs of `-'.
-for _am_tool in $_am_tools
-do
-  case $_am_tool in
-  gnutar)
-    for _am_tar in tar gnutar gtar;
-    do
-      AM_RUN_LOG([$_am_tar --version]) && break
-    done
-    am__tar="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$$tardir"'
-    am__tar_="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$tardir"'
-    am__untar="$_am_tar -xf -"
-    ;;
-  plaintar)
-    # Must skip GNU tar: if it does not support --format= it doesn't create
-    # ustar tarball either.
-    (tar --version) >/dev/null 2>&1 && continue
-    am__tar='tar chf - "$$tardir"'
-    am__tar_='tar chf - "$tardir"'
-    am__untar='tar xf -'
-    ;;
-  pax)
-    am__tar='pax -L -x $1 -w "$$tardir"'
-    am__tar_='pax -L -x $1 -w "$tardir"'
-    am__untar='pax -r'
-    ;;
-  cpio)
-    am__tar='find "$$tardir" -print | cpio -o -H $1 -L'
-    am__tar_='find "$tardir" -print | cpio -o -H $1 -L'
-    am__untar='cpio -i -H $1 -d'
-    ;;
-  none)
-    am__tar=false
-    am__tar_=false
-    am__untar=false
-    ;;
-  esac
 
-  # If the value was cached, stop now.  We just wanted to have am__tar
-  # and am__untar set.
-  test -n "${am_cv_prog_tar_$1}" && break
+m4_if([$1], [v7],
+  [am__tar='$${TAR-tar} chof - "$$tardir"' am__untar='$${TAR-tar} xf -'],
 
-  # tar/untar a dummy directory, and stop if the command works
+  [m4_case([$1],
+    [ustar],
+     [# The POSIX 1988 'ustar' format is defined with fixed-size fields.
+      # There is notably a 21 bits limit for the UID and the GID.  In fact,
+      # the 'pax' utility can hang on bigger UID/GID (see automake bug#8343
+      # and bug#13588).
+      am_max_uid=2097151 # 2^21 - 1
+      am_max_gid=$am_max_uid
+      # The $UID and $GID variables are not portable, so we need to resort
+      # to the POSIX-mandated id(1) utility.  Errors in the 'id' calls
+      # below are definitely unexpected, so allow the users to see them
+      # (that is, avoid stderr redirection).
+      am_uid=`id -u || echo unknown`
+      am_gid=`id -g || echo unknown`
+      AC_MSG_CHECKING([whether UID '$am_uid' is supported by ustar format])
+      if test $am_uid -le $am_max_uid; then
+         AC_MSG_RESULT([yes])
+      else
+         AC_MSG_RESULT([no])
+         _am_tools=none
+      fi
+      AC_MSG_CHECKING([whether GID '$am_gid' is supported by ustar format])
+      if test $am_gid -le $am_max_gid; then
+         AC_MSG_RESULT([yes])
+      else
+        AC_MSG_RESULT([no])
+        _am_tools=none
+      fi],
+
+  [pax],
+    [],
+
+  [m4_fatal([Unknown tar format])])
+
+  AC_MSG_CHECKING([how to create a $1 tar archive])
+
+  # Go ahead even if we have the value already cached.  We do so because we
+  # need to set the values for the 'am__tar' and 'am__untar' variables.
+  _am_tools=${am_cv_prog_tar_$1-$_am_tools}
+
+  for _am_tool in $_am_tools; do
+    case $_am_tool in
+    gnutar)
+      for _am_tar in tar gnutar gtar; do
+        AM_RUN_LOG([$_am_tar --version]) && break
+      done
+      am__tar="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$$tardir"'
+      am__tar_="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$tardir"'
+      am__untar="$_am_tar -xf -"
+      ;;
+    plaintar)
+      # Must skip GNU tar: if it does not support --format= it doesn't create
+      # ustar tarball either.
+      (tar --version) >/dev/null 2>&1 && continue
+      am__tar='tar chf - "$$tardir"'
+      am__tar_='tar chf - "$tardir"'
+      am__untar='tar xf -'
+      ;;
+    pax)
+      am__tar='pax -L -x $1 -w "$$tardir"'
+      am__tar_='pax -L -x $1 -w "$tardir"'
+      am__untar='pax -r'
+      ;;
+    cpio)
+      am__tar='find "$$tardir" -print | cpio -o -H $1 -L'
+      am__tar_='find "$tardir" -print | cpio -o -H $1 -L'
+      am__untar='cpio -i -H $1 -d'
+      ;;
+    none)
+      am__tar=false
+      am__tar_=false
+      am__untar=false
+      ;;
+    esac
+
+    # If the value was cached, stop now.  We just wanted to have am__tar
+    # and am__untar set.
+    test -n "${am_cv_prog_tar_$1}" && break
+
+    # tar/untar a dummy directory, and stop if the command works.
+    rm -rf conftest.dir
+    mkdir conftest.dir
+    echo GrepMe > conftest.dir/file
+    AM_RUN_LOG([tardir=conftest.dir && eval $am__tar_ >conftest.tar])
+    rm -rf conftest.dir
+    if test -s conftest.tar; then
+      AM_RUN_LOG([$am__untar <conftest.tar])
+      AM_RUN_LOG([cat conftest.dir/file])
+      grep GrepMe conftest.dir/file >/dev/null 2>&1 && break
+    fi
+  done
   rm -rf conftest.dir
-  mkdir conftest.dir
-  echo GrepMe > conftest.dir/file
-  AM_RUN_LOG([tardir=conftest.dir && eval $am__tar_ >conftest.tar])
-  rm -rf conftest.dir
-  if test -s conftest.tar; then
-    AM_RUN_LOG([$am__untar <conftest.tar])
-    grep GrepMe conftest.dir/file >/dev/null 2>&1 && break
-  fi
-done
-rm -rf conftest.dir
 
-AC_CACHE_VAL([am_cv_prog_tar_$1], [am_cv_prog_tar_$1=$_am_tool])
-AC_MSG_RESULT([$am_cv_prog_tar_$1])])
+  AC_CACHE_VAL([am_cv_prog_tar_$1], [am_cv_prog_tar_$1=$_am_tool])
+  AC_MSG_RESULT([$am_cv_prog_tar_$1])])
+
 AC_SUBST([am__tar])
 AC_SUBST([am__untar])
 ]) # _AM_PROG_TAR
diff --git a/benchmarks/Grid_comms.cc b/benchmarks/Grid_comms.cc
index b1e3848a..5317e127 100644
--- a/benchmarks/Grid_comms.cc
+++ b/benchmarks/Grid_comms.cc
@@ -12,7 +12,7 @@ int main (int argc, char ** argv)
   std::vector<int> simd_layout;
   std::vector<int> mpi_layout;
 
-  GridParseLayout(argv,argc,mpi_layout,simd_layout,latt_size);
+  GridParseLayout(argv,argc,latt_size,simd_layout,mpi_layout);
 
   int Nloop=10;
   int nmu=0;
diff --git a/benchmarks/Grid_memory_bandwidth.cc b/benchmarks/Grid_memory_bandwidth.cc
index eaffe83e..dd6d1816 100644
--- a/benchmarks/Grid_memory_bandwidth.cc
+++ b/benchmarks/Grid_memory_bandwidth.cc
@@ -12,7 +12,7 @@ int main (int argc, char ** argv)
   std::vector<int> simd_layout;
   std::vector<int> mpi_layout;
 
-  GridParseLayout(argv,argc,mpi_layout,simd_layout,tmp_latt_size);
+  GridParseLayout(argv,argc,tmp_latt_size,simd_layout,mpi_layout);
 
   const int Nvec=8;
   typedef Lattice< iVector< vReal,Nvec> > LatticeVec;
diff --git a/benchmarks/Grid_wilson.cc b/benchmarks/Grid_wilson.cc
index d0b96d1c..ea3f597a 100644
--- a/benchmarks/Grid_wilson.cc
+++ b/benchmarks/Grid_wilson.cc
@@ -24,7 +24,7 @@ int main (int argc, char ** argv)
   std::vector<int> simd_layout;
   std::vector<int> mpi_layout;
 
-  GridParseLayout(argv,argc,mpi_layout,simd_layout,latt_size);
+  GridParseLayout(argv,argc,latt_size,simd_layout,mpi_layout);
     
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
   std::vector<int> seeds({1,2,3,4});
diff --git a/configure b/configure
index 3e71855d..3d25f6a9 100755
--- a/configure
+++ b/configure
@@ -1,20 +1,22 @@
 #! /bin/sh
 # Guess values for system-dependent variables and create Makefiles.
-# Generated by GNU Autoconf 2.63 for Grid 1.0.
+# Generated by GNU Autoconf 2.69 for Grid 1.0.
 #
 # Report bugs to <paboyle@ph.ed.ac.uk>.
 #
-# Copyright (C) 1992, 1993, 1994, 1995, 1996, 1998, 1999, 2000, 2001,
-# 2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
+#
+# Copyright (C) 1992-1996, 1998-2012 Free Software Foundation, Inc.
+#
+#
 # This configure script is free software; the Free Software Foundation
 # gives unlimited permission to copy, distribute and modify it.
-## --------------------- ##
-## M4sh Initialization.  ##
-## --------------------- ##
+## -------------------- ##
+## M4sh Initialization. ##
+## -------------------- ##
 
 # Be more Bourne compatible
 DUALCASE=1; export DUALCASE # for MKS sh
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
   emulate sh
   NULLCMD=:
   # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
@@ -22,23 +24,15 @@ if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
   alias -g '${1+"$@"}'='"$@"'
   setopt NO_GLOB_SUBST
 else
-  case `(set -o) 2>/dev/null` in
-  *posix*) set -o posix ;;
+  case `(set -o) 2>/dev/null` in #(
+  *posix*) :
+    set -o posix ;; #(
+  *) :
+     ;;
 esac
-
 fi
 
 
-
-
-# PATH needs CR
-# Avoid depending upon Character Ranges.
-as_cr_letters='abcdefghijklmnopqrstuvwxyz'
-as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
-as_cr_Letters=$as_cr_letters$as_cr_LETTERS
-as_cr_digits='0123456789'
-as_cr_alnum=$as_cr_Letters$as_cr_digits
-
 as_nl='
 '
 export as_nl
@@ -46,7 +40,13 @@ export as_nl
 as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
-if (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+# Prefer a ksh shell builtin over an external printf program on Solaris,
+# but without wasting forks for bash or zsh.
+if test -z "$BASH_VERSION$ZSH_VERSION" \
+    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
+  as_echo='print -r --'
+  as_echo_n='print -rn --'
+elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
   as_echo='printf %s\n'
   as_echo_n='printf %s'
 else
@@ -57,7 +57,7 @@ else
     as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
     as_echo_n_body='eval
       arg=$1;
-      case $arg in
+      case $arg in #(
       *"$as_nl"*)
 	expr "X$arg" : "X\\(.*\\)$as_nl";
 	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
@@ -80,13 +80,6 @@ if test "${PATH_SEPARATOR+set}" != set; then
   }
 fi
 
-# Support unset when possible.
-if ( (MAIL=60; unset MAIL) || exit) >/dev/null 2>&1; then
-  as_unset=unset
-else
-  as_unset=false
-fi
-
 
 # IFS
 # We need space, tab and new line, in precisely that order.  Quoting is
@@ -96,15 +89,16 @@ fi
 IFS=" ""	$as_nl"
 
 # Find who we are.  Look in the path if we contain no directory separator.
-case $0 in
+as_myself=
+case $0 in #((
   *[\\/]* ) as_myself=$0 ;;
   *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
 for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
-done
+    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+  done
 IFS=$as_save_IFS
 
      ;;
@@ -116,12 +110,16 @@ if test "x$as_myself" = x; then
 fi
 if test ! -f "$as_myself"; then
   $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
-  { (exit 1); exit 1; }
+  exit 1
 fi
 
-# Work around bugs in pre-3.0 UWIN ksh.
-for as_var in ENV MAIL MAILPATH
-do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
+# Unset variables that we do not need and which cause bugs (e.g. in
+# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
+# suppresses any "Segmentation fault" message there.  '((' could
+# trigger a bug in pdksh 5.2.14.
+for as_var in BASH_ENV ENV MAIL MAILPATH
+do eval test x\${$as_var+set} = xset \
+  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
 done
 PS1='$ '
 PS2='> '
@@ -133,7 +131,294 @@ export LC_ALL
 LANGUAGE=C
 export LANGUAGE
 
-# Required to use basename.
+# CDPATH.
+(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
+
+# Use a proper internal environment variable to ensure we don't fall
+  # into an infinite loop, continuously re-executing ourselves.
+  if test x"${_as_can_reexec}" != xno && test "x$CONFIG_SHELL" != x; then
+    _as_can_reexec=no; export _as_can_reexec;
+    # We cannot yet assume a decent shell, so we have to provide a
+# neutralization value for shells without unset; and this also
+# works around shells that cannot unset nonexistent variables.
+# Preserve -v and -x to the replacement shell.
+BASH_ENV=/dev/null
+ENV=/dev/null
+(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
+case $- in # ((((
+  *v*x* | *x*v* ) as_opts=-vx ;;
+  *v* ) as_opts=-v ;;
+  *x* ) as_opts=-x ;;
+  * ) as_opts= ;;
+esac
+exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
+# Admittedly, this is quite paranoid, since all the known shells bail
+# out after a failed `exec'.
+$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
+as_fn_exit 255
+  fi
+  # We don't want this to propagate to other subprocesses.
+          { _as_can_reexec=; unset _as_can_reexec;}
+if test "x$CONFIG_SHELL" = x; then
+  as_bourne_compatible="if test -n \"\${ZSH_VERSION+set}\" && (emulate sh) >/dev/null 2>&1; then :
+  emulate sh
+  NULLCMD=:
+  # Pre-4.2 versions of Zsh do word splitting on \${1+\"\$@\"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '\${1+\"\$@\"}'='\"\$@\"'
+  setopt NO_GLOB_SUBST
+else
+  case \`(set -o) 2>/dev/null\` in #(
+  *posix*) :
+    set -o posix ;; #(
+  *) :
+     ;;
+esac
+fi
+"
+  as_required="as_fn_return () { (exit \$1); }
+as_fn_success () { as_fn_return 0; }
+as_fn_failure () { as_fn_return 1; }
+as_fn_ret_success () { return 0; }
+as_fn_ret_failure () { return 1; }
+
+exitcode=0
+as_fn_success || { exitcode=1; echo as_fn_success failed.; }
+as_fn_failure && { exitcode=1; echo as_fn_failure succeeded.; }
+as_fn_ret_success || { exitcode=1; echo as_fn_ret_success failed.; }
+as_fn_ret_failure && { exitcode=1; echo as_fn_ret_failure succeeded.; }
+if ( set x; as_fn_ret_success y && test x = \"\$1\" ); then :
+
+else
+  exitcode=1; echo positional parameters were not saved.
+fi
+test x\$exitcode = x0 || exit 1
+test -x / || exit 1"
+  as_suggested="  as_lineno_1=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_1a=\$LINENO
+  as_lineno_2=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_2a=\$LINENO
+  eval 'test \"x\$as_lineno_1'\$as_run'\" != \"x\$as_lineno_2'\$as_run'\" &&
+  test \"x\`expr \$as_lineno_1'\$as_run' + 1\`\" = \"x\$as_lineno_2'\$as_run'\"' || exit 1
+test \$(( 1 + 1 )) = 2 || exit 1"
+  if (eval "$as_required") 2>/dev/null; then :
+  as_have_required=yes
+else
+  as_have_required=no
+fi
+  if test x$as_have_required = xyes && (eval "$as_suggested") 2>/dev/null; then :
+
+else
+  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+as_found=false
+for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+  as_found=:
+  case $as_dir in #(
+	 /*)
+	   for as_base in sh bash ksh sh5; do
+	     # Try only shells that exist, to save several forks.
+	     as_shell=$as_dir/$as_base
+	     if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
+		    { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$as_shell"; } 2>/dev/null; then :
+  CONFIG_SHELL=$as_shell as_have_required=yes
+		   if { $as_echo "$as_bourne_compatible""$as_suggested" | as_run=a "$as_shell"; } 2>/dev/null; then :
+  break 2
+fi
+fi
+	   done;;
+       esac
+  as_found=false
+done
+$as_found || { if { test -f "$SHELL" || test -f "$SHELL.exe"; } &&
+	      { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$SHELL"; } 2>/dev/null; then :
+  CONFIG_SHELL=$SHELL as_have_required=yes
+fi; }
+IFS=$as_save_IFS
+
+
+      if test "x$CONFIG_SHELL" != x; then :
+  export CONFIG_SHELL
+             # We cannot yet assume a decent shell, so we have to provide a
+# neutralization value for shells without unset; and this also
+# works around shells that cannot unset nonexistent variables.
+# Preserve -v and -x to the replacement shell.
+BASH_ENV=/dev/null
+ENV=/dev/null
+(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
+case $- in # ((((
+  *v*x* | *x*v* ) as_opts=-vx ;;
+  *v* ) as_opts=-v ;;
+  *x* ) as_opts=-x ;;
+  * ) as_opts= ;;
+esac
+exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
+# Admittedly, this is quite paranoid, since all the known shells bail
+# out after a failed `exec'.
+$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
+exit 255
+fi
+
+    if test x$as_have_required = xno; then :
+  $as_echo "$0: This script requires a shell more modern than all"
+  $as_echo "$0: the shells that I found on your system."
+  if test x${ZSH_VERSION+set} = xset ; then
+    $as_echo "$0: In particular, zsh $ZSH_VERSION has bugs and should"
+    $as_echo "$0: be upgraded to zsh 4.3.4 or later."
+  else
+    $as_echo "$0: Please tell bug-autoconf@gnu.org and
+$0: paboyle@ph.ed.ac.uk about your system, including any
+$0: error possibly output before this message. Then install
+$0: a modern shell, or manually run the script under such a
+$0: shell if you do have one."
+  fi
+  exit 1
+fi
+fi
+fi
+SHELL=${CONFIG_SHELL-/bin/sh}
+export SHELL
+# Unset more variables known to interfere with behavior of common tools.
+CLICOLOR_FORCE= GREP_OPTIONS=
+unset CLICOLOR_FORCE GREP_OPTIONS
+
+## --------------------- ##
+## M4sh Shell Functions. ##
+## --------------------- ##
+# as_fn_unset VAR
+# ---------------
+# Portably unset VAR.
+as_fn_unset ()
+{
+  { eval $1=; unset $1;}
+}
+as_unset=as_fn_unset
+
+# as_fn_set_status STATUS
+# -----------------------
+# Set $? to STATUS, without forking.
+as_fn_set_status ()
+{
+  return $1
+} # as_fn_set_status
+
+# as_fn_exit STATUS
+# -----------------
+# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
+as_fn_exit ()
+{
+  set +e
+  as_fn_set_status $1
+  exit $1
+} # as_fn_exit
+
+# as_fn_mkdir_p
+# -------------
+# Create "$as_dir" as a directory, including parents if necessary.
+as_fn_mkdir_p ()
+{
+
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || eval $as_mkdir_p || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
+
+
+} # as_fn_mkdir_p
+
+# as_fn_executable_p FILE
+# -----------------------
+# Test if FILE is an executable regular file.
+as_fn_executable_p ()
+{
+  test -f "$1" && test -x "$1"
+} # as_fn_executable_p
+# as_fn_append VAR VALUE
+# ----------------------
+# Append the text in VALUE to the end of the definition contained in VAR. Take
+# advantage of any shell optimizations that allow amortized linear growth over
+# repeated appends, instead of the typical quadratic growth present in naive
+# implementations.
+if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
+  eval 'as_fn_append ()
+  {
+    eval $1+=\$2
+  }'
+else
+  as_fn_append ()
+  {
+    eval $1=\$$1\$2
+  }
+fi # as_fn_append
+
+# as_fn_arith ARG...
+# ------------------
+# Perform arithmetic evaluation on the ARGs, and store the result in the
+# global $as_val. Take advantage of shells that can avoid forks. The arguments
+# must be portable across $(()) and expr.
+if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
+  eval 'as_fn_arith ()
+  {
+    as_val=$(( $* ))
+  }'
+else
+  as_fn_arith ()
+  {
+    as_val=`expr "$@" || test $? -eq 1`
+  }
+fi # as_fn_arith
+
+
+# as_fn_error STATUS ERROR [LINENO LOG_FD]
+# ----------------------------------------
+# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
+# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
+# script with STATUS, using 1 if that was 0.
+as_fn_error ()
+{
+  as_status=$1; test $as_status -eq 0 && as_status=1
+  if test "$4"; then
+    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
+  fi
+  $as_echo "$as_me: error: $2" >&2
+  as_fn_exit $as_status
+} # as_fn_error
+
 if expr a : '\(a\)' >/dev/null 2>&1 &&
    test "X`expr 00001 : '.*\(...\)'`" = X001; then
   as_expr=expr
@@ -147,8 +432,12 @@ else
   as_basename=false
 fi
 
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
 
-# Name of the executable.
 as_me=`$as_basename -- "$0" ||
 $as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
 	 X"$0" : 'X\(//\)$' \| \
@@ -168,295 +457,19 @@ $as_echo X/"$0" |
 	  }
 	  s/.*/./; q'`
 
-# CDPATH.
-$as_unset CDPATH
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
 
 
-if test "x$CONFIG_SHELL" = x; then
-  if (eval ":") 2>/dev/null; then
-  as_have_required=yes
-else
-  as_have_required=no
-fi
-
-  if test $as_have_required = yes &&	 (eval ":
-(as_func_return () {
-  (exit \$1)
-}
-as_func_success () {
-  as_func_return 0
-}
-as_func_failure () {
-  as_func_return 1
-}
-as_func_ret_success () {
-  return 0
-}
-as_func_ret_failure () {
-  return 1
-}
-
-exitcode=0
-if as_func_success; then
-  :
-else
-  exitcode=1
-  echo as_func_success failed.
-fi
-
-if as_func_failure; then
-  exitcode=1
-  echo as_func_failure succeeded.
-fi
-
-if as_func_ret_success; then
-  :
-else
-  exitcode=1
-  echo as_func_ret_success failed.
-fi
-
-if as_func_ret_failure; then
-  exitcode=1
-  echo as_func_ret_failure succeeded.
-fi
-
-if ( set x; as_func_ret_success y && test x = \"\$1\" ); then
-  :
-else
-  exitcode=1
-  echo positional parameters were not saved.
-fi
-
-test \$exitcode = 0) || { (exit 1); exit 1; }
-
-(
-  as_lineno_1=\$LINENO
-  as_lineno_2=\$LINENO
-  test \"x\$as_lineno_1\" != \"x\$as_lineno_2\" &&
-  test \"x\`expr \$as_lineno_1 + 1\`\" = \"x\$as_lineno_2\") || { (exit 1); exit 1; }
-") 2> /dev/null; then
-  :
-else
-  as_candidate_shells=
-    as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-  case $as_dir in
-	 /*)
-	   for as_base in sh bash ksh sh5; do
-	     as_candidate_shells="$as_candidate_shells $as_dir/$as_base"
-	   done;;
-       esac
-done
-IFS=$as_save_IFS
-
-
-      for as_shell in $as_candidate_shells $SHELL; do
-	 # Try only shells that exist, to save several forks.
-	 if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
-		{ ("$as_shell") 2> /dev/null <<\_ASEOF
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
-  emulate sh
-  NULLCMD=:
-  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
-  # is contrary to our usage.  Disable this feature.
-  alias -g '${1+"$@"}'='"$@"'
-  setopt NO_GLOB_SUBST
-else
-  case `(set -o) 2>/dev/null` in
-  *posix*) set -o posix ;;
-esac
-
-fi
-
-
-:
-_ASEOF
-}; then
-  CONFIG_SHELL=$as_shell
-	       as_have_required=yes
-	       if { "$as_shell" 2> /dev/null <<\_ASEOF
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
-  emulate sh
-  NULLCMD=:
-  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
-  # is contrary to our usage.  Disable this feature.
-  alias -g '${1+"$@"}'='"$@"'
-  setopt NO_GLOB_SUBST
-else
-  case `(set -o) 2>/dev/null` in
-  *posix*) set -o posix ;;
-esac
-
-fi
-
-
-:
-(as_func_return () {
-  (exit $1)
-}
-as_func_success () {
-  as_func_return 0
-}
-as_func_failure () {
-  as_func_return 1
-}
-as_func_ret_success () {
-  return 0
-}
-as_func_ret_failure () {
-  return 1
-}
-
-exitcode=0
-if as_func_success; then
-  :
-else
-  exitcode=1
-  echo as_func_success failed.
-fi
-
-if as_func_failure; then
-  exitcode=1
-  echo as_func_failure succeeded.
-fi
-
-if as_func_ret_success; then
-  :
-else
-  exitcode=1
-  echo as_func_ret_success failed.
-fi
-
-if as_func_ret_failure; then
-  exitcode=1
-  echo as_func_ret_failure succeeded.
-fi
-
-if ( set x; as_func_ret_success y && test x = "$1" ); then
-  :
-else
-  exitcode=1
-  echo positional parameters were not saved.
-fi
-
-test $exitcode = 0) || { (exit 1); exit 1; }
-
-(
-  as_lineno_1=$LINENO
-  as_lineno_2=$LINENO
-  test "x$as_lineno_1" != "x$as_lineno_2" &&
-  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2") || { (exit 1); exit 1; }
-
-_ASEOF
-}; then
-  break
-fi
-
-fi
-
-      done
-
-      if test "x$CONFIG_SHELL" != x; then
-  for as_var in BASH_ENV ENV
-	do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
-	done
-	export CONFIG_SHELL
-	exec "$CONFIG_SHELL" "$as_myself" ${1+"$@"}
-fi
-
-
-    if test $as_have_required = no; then
-  echo This script requires a shell more modern than all the
-      echo shells that I found on your system.  Please install a
-      echo modern shell, or manually run the script under such a
-      echo shell if you do have one.
-      { (exit 1); exit 1; }
-fi
-
-
-fi
-
-fi
-
-
-
-(eval "as_func_return () {
-  (exit \$1)
-}
-as_func_success () {
-  as_func_return 0
-}
-as_func_failure () {
-  as_func_return 1
-}
-as_func_ret_success () {
-  return 0
-}
-as_func_ret_failure () {
-  return 1
-}
-
-exitcode=0
-if as_func_success; then
-  :
-else
-  exitcode=1
-  echo as_func_success failed.
-fi
-
-if as_func_failure; then
-  exitcode=1
-  echo as_func_failure succeeded.
-fi
-
-if as_func_ret_success; then
-  :
-else
-  exitcode=1
-  echo as_func_ret_success failed.
-fi
-
-if as_func_ret_failure; then
-  exitcode=1
-  echo as_func_ret_failure succeeded.
-fi
-
-if ( set x; as_func_ret_success y && test x = \"\$1\" ); then
-  :
-else
-  exitcode=1
-  echo positional parameters were not saved.
-fi
-
-test \$exitcode = 0") || {
-  echo No shell found that supports shell functions.
-  echo Please tell bug-autoconf@gnu.org about your system,
-  echo including any error possibly output before this message.
-  echo This can help us improve future autoconf versions.
-  echo Configuration will now proceed without shell functions.
-}
-
-
-
-  as_lineno_1=$LINENO
-  as_lineno_2=$LINENO
-  test "x$as_lineno_1" != "x$as_lineno_2" &&
-  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2" || {
-
-  # Create $as_me.lineno as a copy of $as_myself, but with $LINENO
-  # uniformly replaced by the line number.  The first 'sed' inserts a
-  # line-number line after each line using $LINENO; the second 'sed'
-  # does the real work.  The second script uses 'N' to pair each
-  # line-number line with the line containing $LINENO, and appends
-  # trailing '-' during substitution so that $LINENO is not a special
-  # case at line end.
-  # (Raja R Harinath suggested sed '=', and Paul Eggert wrote the
-  # scripts with optimization help from Paolo Bonzini.  Blame Lee
-  # E. McMahon (1931-1989) for sed's syntax.  :-)
+  as_lineno_1=$LINENO as_lineno_1a=$LINENO
+  as_lineno_2=$LINENO as_lineno_2a=$LINENO
+  eval 'test "x$as_lineno_1'$as_run'" != "x$as_lineno_2'$as_run'" &&
+  test "x`expr $as_lineno_1'$as_run' + 1`" = "x$as_lineno_2'$as_run'"' || {
+  # Blame Lee E. McMahon (1931-1989) for sed's syntax.  :-)
   sed -n '
     p
     /[$]LINENO/=
@@ -473,9 +486,12 @@ test \$exitcode = 0") || {
       s/-\n.*//
     ' >$as_me.lineno &&
   chmod +x "$as_me.lineno" ||
-    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2
-   { (exit 1); exit 1; }; }
+    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2; as_fn_exit 1; }
 
+  # If we had to re-execute with $CONFIG_SHELL, we're ensured to have
+  # already done that, so ensure we don't try to do so again and fall
+  # in an infinite loop.  This has already happened in practice.
+  _as_can_reexec=no; export _as_can_reexec
   # Don't try to exec as it changes $[0], causing all sort of problems
   # (the dirname of $[0] is not the place where we might find the
   # original and so on.  Autoconf is especially sensitive to this).
@@ -484,29 +500,18 @@ test \$exitcode = 0") || {
   exit
 }
 
-
-if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
-  as_dirname=dirname
-else
-  as_dirname=false
-fi
-
 ECHO_C= ECHO_N= ECHO_T=
-case `echo -n x` in
+case `echo -n x` in #(((((
 -n*)
-  case `echo 'x\c'` in
+  case `echo 'xy\c'` in
   *c*) ECHO_T='	';;	# ECHO_T is single tab character.
-  *)   ECHO_C='\c';;
+  xy)  ECHO_C='\c';;
+  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
+       ECHO_T='	';;
   esac;;
 *)
   ECHO_N='-n';;
 esac
-if expr a : '\(a\)' >/dev/null 2>&1 &&
-   test "X`expr 00001 : '.*\(...\)'`" = X001; then
-  as_expr=expr
-else
-  as_expr=false
-fi
 
 rm -f conf$$ conf$$.exe conf$$.file
 if test -d conf$$.dir; then
@@ -521,49 +526,29 @@ if (echo >conf$$.file) 2>/dev/null; then
     # ... but there are two gotchas:
     # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
     # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
-    # In both cases, we have to default to `cp -p'.
+    # In both cases, we have to default to `cp -pR'.
     ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
-      as_ln_s='cp -p'
+      as_ln_s='cp -pR'
   elif ln conf$$.file conf$$ 2>/dev/null; then
     as_ln_s=ln
   else
-    as_ln_s='cp -p'
+    as_ln_s='cp -pR'
   fi
 else
-  as_ln_s='cp -p'
+  as_ln_s='cp -pR'
 fi
 rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
 rmdir conf$$.dir 2>/dev/null
 
 if mkdir -p . 2>/dev/null; then
-  as_mkdir_p=:
+  as_mkdir_p='mkdir -p "$as_dir"'
 else
   test -d ./-p && rmdir ./-p
   as_mkdir_p=false
 fi
 
-if test -x / >/dev/null 2>&1; then
-  as_test_x='test -x'
-else
-  if ls -dL / >/dev/null 2>&1; then
-    as_ls_L_option=L
-  else
-    as_ls_L_option=
-  fi
-  as_test_x='
-    eval sh -c '\''
-      if test -d "$1"; then
-	test -d "$1/.";
-      else
-	case $1 in
-	-*)set "./$1";;
-	esac;
-	case `ls -ld'$as_ls_L_option' "$1" 2>/dev/null` in
-	???[sx]*):;;*)false;;esac;fi
-    '\'' sh
-  '
-fi
-as_executable_p=$as_test_x
+as_test_x='test -x'
+as_executable_p=as_fn_executable_p
 
 # Sed expression to map a string onto a valid CPP name.
 as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
@@ -572,11 +557,11 @@ as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
 as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
 
 
-
-exec 7<&0 </dev/null 6>&1
+test -n "$DJDIR" || exec 7<&0 </dev/null
+exec 6>&1
 
 # Name of the host.
-# hostname on some systems (SVR3.2, Linux) returns a bogus exit status,
+# hostname on some systems (SVR3.2, old GNU/Linux) returns a bogus exit status,
 # so uname gets run too.
 ac_hostname=`(hostname || uname -n) 2>/dev/null | sed 1q`
 
@@ -591,7 +576,6 @@ cross_compiling=no
 subdirs=
 MFLAGS=
 MAKEFLAGS=
-SHELL=${CONFIG_SHELL-/bin/sh}
 
 # Identity of this package.
 PACKAGE_NAME='Grid'
@@ -599,6 +583,7 @@ PACKAGE_TARNAME='grid'
 PACKAGE_VERSION='1.0'
 PACKAGE_STRING='Grid 1.0'
 PACKAGE_BUGREPORT='paboyle@ph.ed.ac.uk'
+PACKAGE_URL=''
 
 ac_unique_file="lib/Grid.h"
 # Factoring default headers for most tests.
@@ -659,6 +644,7 @@ CC
 am__fastdepCXX_FALSE
 am__fastdepCXX_TRUE
 CXXDEPMODE
+am__nodep
 AMDEPBACKSLASH
 AMDEP_FALSE
 AMDEP_TRUE
@@ -672,6 +658,10 @@ CPPFLAGS
 LDFLAGS
 CXXFLAGS
 CXX
+AM_BACKSLASH
+AM_DEFAULT_VERBOSITY
+AM_DEFAULT_V
+AM_V
 am__untar
 am__tar
 AMTAR
@@ -725,6 +715,7 @@ bindir
 program_transform_name
 prefix
 exec_prefix
+PACKAGE_URL
 PACKAGE_BUGREPORT
 PACKAGE_STRING
 PACKAGE_VERSION
@@ -735,6 +726,7 @@ SHELL'
 ac_subst_files=''
 ac_user_opts='
 enable_option_checking
+enable_silent_rules
 enable_dependency_tracking
 enable_openmp
 enable_simd
@@ -814,8 +806,9 @@ do
   fi
 
   case $ac_option in
-  *=*)	ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
-  *)	ac_optarg=yes ;;
+  *=?*) ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
+  *=)   ac_optarg= ;;
+  *)    ac_optarg=yes ;;
   esac
 
   # Accept the important Cygnus configure options, so we can diagnose typos.
@@ -860,8 +853,7 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*disable-\(.*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid feature name: $ac_useropt" >&2
-   { (exit 1); exit 1; }; }
+      as_fn_error $? "invalid feature name: $ac_useropt"
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -887,8 +879,7 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*enable-\([^=]*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid feature name: $ac_useropt" >&2
-   { (exit 1); exit 1; }; }
+      as_fn_error $? "invalid feature name: $ac_useropt"
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1092,8 +1083,7 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*with-\([^=]*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid package name: $ac_useropt" >&2
-   { (exit 1); exit 1; }; }
+      as_fn_error $? "invalid package name: $ac_useropt"
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1109,8 +1099,7 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*without-\(.*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid package name: $ac_useropt" >&2
-   { (exit 1); exit 1; }; }
+      as_fn_error $? "invalid package name: $ac_useropt"
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1140,17 +1129,17 @@ do
   | --x-librar=* | --x-libra=* | --x-libr=* | --x-lib=* | --x-li=* | --x-l=*)
     x_libraries=$ac_optarg ;;
 
-  -*) { $as_echo "$as_me: error: unrecognized option: $ac_option
-Try \`$0 --help' for more information." >&2
-   { (exit 1); exit 1; }; }
+  -*) as_fn_error $? "unrecognized option: \`$ac_option'
+Try \`$0 --help' for more information"
     ;;
 
   *=*)
     ac_envvar=`expr "x$ac_option" : 'x\([^=]*\)='`
     # Reject names that are not valid shell variable names.
-    expr "x$ac_envvar" : ".*[^_$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid variable name: $ac_envvar" >&2
-   { (exit 1); exit 1; }; }
+    case $ac_envvar in #(
+      '' | [0-9]* | *[!_$as_cr_alnum]* )
+      as_fn_error $? "invalid variable name: \`$ac_envvar'" ;;
+    esac
     eval $ac_envvar=\$ac_optarg
     export $ac_envvar ;;
 
@@ -1159,7 +1148,7 @@ Try \`$0 --help' for more information." >&2
     $as_echo "$as_me: WARNING: you should use --build, --host, --target" >&2
     expr "x$ac_option" : ".*[^-._$as_cr_alnum]" >/dev/null &&
       $as_echo "$as_me: WARNING: invalid host type: $ac_option" >&2
-    : ${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}
+    : "${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}"
     ;;
 
   esac
@@ -1167,15 +1156,13 @@ done
 
 if test -n "$ac_prev"; then
   ac_option=--`echo $ac_prev | sed 's/_/-/g'`
-  { $as_echo "$as_me: error: missing argument to $ac_option" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "missing argument to $ac_option"
 fi
 
 if test -n "$ac_unrecognized_opts"; then
   case $enable_option_checking in
     no) ;;
-    fatal) { $as_echo "$as_me: error: unrecognized options: $ac_unrecognized_opts" >&2
-   { (exit 1); exit 1; }; } ;;
+    fatal) as_fn_error $? "unrecognized options: $ac_unrecognized_opts" ;;
     *)     $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2 ;;
   esac
 fi
@@ -1198,8 +1185,7 @@ do
     [\\/$]* | ?:[\\/]* )  continue;;
     NONE | '' ) case $ac_var in *prefix ) continue;; esac;;
   esac
-  { $as_echo "$as_me: error: expected an absolute directory name for --$ac_var: $ac_val" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "expected an absolute directory name for --$ac_var: $ac_val"
 done
 
 # There might be people who depend on the old broken behavior: `$host'
@@ -1213,8 +1199,6 @@ target=$target_alias
 if test "x$host_alias" != x; then
   if test "x$build_alias" = x; then
     cross_compiling=maybe
-    $as_echo "$as_me: WARNING: If you wanted to set the --build type, don't use --host.
-    If a cross compiler is detected then cross compile mode will be used." >&2
   elif test "x$build_alias" != "x$host_alias"; then
     cross_compiling=yes
   fi
@@ -1229,11 +1213,9 @@ test "$silent" = yes && exec 6>/dev/null
 ac_pwd=`pwd` && test -n "$ac_pwd" &&
 ac_ls_di=`ls -di .` &&
 ac_pwd_ls_di=`cd "$ac_pwd" && ls -di .` ||
-  { $as_echo "$as_me: error: working directory cannot be determined" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "working directory cannot be determined"
 test "X$ac_ls_di" = "X$ac_pwd_ls_di" ||
-  { $as_echo "$as_me: error: pwd does not report name of working directory" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "pwd does not report name of working directory"
 
 
 # Find the source files, if location was not specified.
@@ -1272,13 +1254,11 @@ else
 fi
 if test ! -r "$srcdir/$ac_unique_file"; then
   test "$ac_srcdir_defaulted" = yes && srcdir="$ac_confdir or .."
-  { $as_echo "$as_me: error: cannot find sources ($ac_unique_file) in $srcdir" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "cannot find sources ($ac_unique_file) in $srcdir"
 fi
 ac_msg="sources are in $srcdir, but \`cd $srcdir' does not work"
 ac_abs_confdir=`(
-	cd "$srcdir" && test -r "./$ac_unique_file" || { $as_echo "$as_me: error: $ac_msg" >&2
-   { (exit 1); exit 1; }; }
+	cd "$srcdir" && test -r "./$ac_unique_file" || as_fn_error $? "$ac_msg"
 	pwd)`
 # When building in place, set srcdir=.
 if test "$ac_abs_confdir" = "$ac_pwd"; then
@@ -1318,7 +1298,7 @@ Configuration:
       --help=short        display options specific to this package
       --help=recursive    display the short help of all the included packages
   -V, --version           display version information and exit
-  -q, --quiet, --silent   do not print \`checking...' messages
+  -q, --quiet, --silent   do not print \`checking ...' messages
       --cache-file=FILE   cache test results in FILE [disabled]
   -C, --config-cache      alias for \`--cache-file=config.cache'
   -n, --no-create         do not create output files
@@ -1378,8 +1358,12 @@ Optional Features:
   --disable-option-checking  ignore unrecognized --enable/--with options
   --disable-FEATURE       do not include FEATURE (same as --enable-FEATURE=no)
   --enable-FEATURE[=ARG]  include FEATURE [ARG=yes]
-  --disable-dependency-tracking  speeds up one-time build
-  --enable-dependency-tracking   do not reject slow dependency extractors
+  --enable-silent-rules   less verbose build output (undo: "make V=1")
+  --disable-silent-rules  verbose build output (undo: "make V=0")
+  --enable-dependency-tracking
+                          do not reject slow dependency extractors
+  --disable-dependency-tracking
+                          speeds up one-time build
   --disable-openmp        do not use OpenMP
   --enable-simd=SSE|AVX|AVX2|AVX512|MIC
                           Select instructions to be SSE4.0, AVX 1.0, AVX
@@ -1392,7 +1376,7 @@ Some influential environment variables:
   LDFLAGS     linker flags, e.g. -L<lib dir> if you have libraries in a
               nonstandard directory <lib dir>
   LIBS        libraries to pass to the linker, e.g. -l<library>
-  CPPFLAGS    C/C++/Objective C preprocessor flags, e.g. -I<include dir> if
+  CPPFLAGS    (Objective) C/C++ preprocessor flags, e.g. -I<include dir> if
               you have headers in a nonstandard directory <include dir>
   CC          C compiler command
   CFLAGS      C compiler flags
@@ -1465,21 +1449,568 @@ test -n "$ac_init_help" && exit $ac_status
 if $ac_init_version; then
   cat <<\_ACEOF
 Grid configure 1.0
-generated by GNU Autoconf 2.63
+generated by GNU Autoconf 2.69
 
-Copyright (C) 1992, 1993, 1994, 1995, 1996, 1998, 1999, 2000, 2001,
-2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
+Copyright (C) 2012 Free Software Foundation, Inc.
 This configure script is free software; the Free Software Foundation
 gives unlimited permission to copy, distribute and modify it.
 _ACEOF
   exit
 fi
+
+## ------------------------ ##
+## Autoconf initialization. ##
+## ------------------------ ##
+
+# ac_fn_cxx_try_compile LINENO
+# ----------------------------
+# Try to compile conftest.$ac_ext, and return whether this succeeded.
+ac_fn_cxx_try_compile ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  rm -f conftest.$ac_objext
+  if { { ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compile") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_retval=1
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_cxx_try_compile
+
+# ac_fn_c_try_compile LINENO
+# --------------------------
+# Try to compile conftest.$ac_ext, and return whether this succeeded.
+ac_fn_c_try_compile ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  rm -f conftest.$ac_objext
+  if { { ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compile") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_retval=1
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_compile
+
+# ac_fn_c_try_link LINENO
+# -----------------------
+# Try to link conftest.$ac_ext, and return whether this succeeded.
+ac_fn_c_try_link ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  rm -f conftest.$ac_objext conftest$ac_exeext
+  if { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 test -x conftest$ac_exeext
+       }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_retval=1
+fi
+  # Delete the IPA/IPO (Inter Procedural Analysis/Optimization) information
+  # created by the PGI compiler (conftest_ipa8_conftest.oo), as it would
+  # interfere with the next link command; also delete a directory that is
+  # left behind by Apple's compiler.  We do this before executing the actions.
+  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_link
+
+# ac_fn_c_try_cpp LINENO
+# ----------------------
+# Try to preprocess conftest.$ac_ext, and return whether this succeeded.
+ac_fn_c_try_cpp ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if { { ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } > conftest.i && {
+	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+    ac_retval=1
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_cpp
+
+# ac_fn_c_check_header_mongrel LINENO HEADER VAR INCLUDES
+# -------------------------------------------------------
+# Tests whether HEADER exists, giving a warning if it cannot be compiled using
+# the include files in INCLUDES and setting the cache variable VAR
+# accordingly.
+ac_fn_c_check_header_mongrel ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if eval \${$3+:} false; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 usability" >&5
+$as_echo_n "checking $2 usability... " >&6; }
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+#include <$2>
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_header_compiler=yes
+else
+  ac_header_compiler=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 presence" >&5
+$as_echo_n "checking $2 presence... " >&6; }
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <$2>
+_ACEOF
+if ac_fn_c_try_cpp "$LINENO"; then :
+  ac_header_preproc=yes
+else
+  ac_header_preproc=no
+fi
+rm -f conftest.err conftest.i conftest.$ac_ext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in #((
+  yes:no: )
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $2: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $2:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $2: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
+( $as_echo "## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##"
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  eval "$3=\$ac_header_compiler"
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_header_mongrel
+
+# ac_fn_c_try_run LINENO
+# ----------------------
+# Try to link conftest.$ac_ext, and return whether this succeeded. Assumes
+# that executables *can* be run.
+ac_fn_c_try_run ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && { ac_try='./conftest$ac_exeext'
+  { { case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: program exited with status $ac_status" >&5
+       $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+       ac_retval=$ac_status
+fi
+  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_run
+
+# ac_fn_c_check_header_compile LINENO HEADER VAR INCLUDES
+# -------------------------------------------------------
+# Tests whether HEADER exists and can be compiled using the include files in
+# INCLUDES, setting the cache variable VAR accordingly.
+ac_fn_c_check_header_compile ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+#include <$2>
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  eval "$3=yes"
+else
+  eval "$3=no"
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_header_compile
+
+# ac_fn_c_check_decl LINENO SYMBOL VAR INCLUDES
+# ---------------------------------------------
+# Tests whether SYMBOL is declared in INCLUDES, setting cache variable VAR
+# accordingly.
+ac_fn_c_check_decl ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  as_decl_name=`echo $2|sed 's/ *(.*//'`
+  as_decl_use=`echo $2|sed -e 's/(/((/' -e 's/)/) 0&/' -e 's/,/) 0& (/g'`
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $as_decl_name is declared" >&5
+$as_echo_n "checking whether $as_decl_name is declared... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+#ifndef $as_decl_name
+#ifdef __cplusplus
+  (void) $as_decl_use;
+#else
+  (void) $as_decl_name;
+#endif
+#endif
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  eval "$3=yes"
+else
+  eval "$3=no"
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_decl
+
+# ac_fn_c_check_type LINENO TYPE VAR INCLUDES
+# -------------------------------------------
+# Tests whether TYPE exists after having included INCLUDES, setting cache
+# variable VAR accordingly.
+ac_fn_c_check_type ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  eval "$3=no"
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+if (sizeof ($2))
+	 return 0;
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+if (sizeof (($2)))
+	    return 0;
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+
+else
+  eval "$3=yes"
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_type
+
+# ac_fn_c_find_uintX_t LINENO BITS VAR
+# ------------------------------------
+# Finds an unsigned integer type with width BITS, setting cache variable VAR
+# accordingly.
+ac_fn_c_find_uintX_t ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for uint$2_t" >&5
+$as_echo_n "checking for uint$2_t... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  eval "$3=no"
+     # Order is important - never check a type that is potentially smaller
+     # than half of the expected target width.
+     for ac_type in uint$2_t 'unsigned int' 'unsigned long int' \
+	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
+       cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+static int test_array [1 - 2 * !((($ac_type) -1 >> ($2 / 2 - 1)) >> ($2 / 2 - 1) == 3)];
+test_array [0] = 0;
+return test_array [0];
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  case $ac_type in #(
+  uint$2_t) :
+    eval "$3=yes" ;; #(
+  *) :
+    eval "$3=\$ac_type" ;;
+esac
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+       if eval test \"x\$"$3"\" = x"no"; then :
+
+else
+  break
+fi
+     done
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_find_uintX_t
+
+# ac_fn_c_check_func LINENO FUNC VAR
+# ----------------------------------
+# Tests whether FUNC exists, setting the cache variable VAR accordingly
+ac_fn_c_check_func ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+/* Define $2 to an innocuous variant, in case <limits.h> declares $2.
+   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
+#define $2 innocuous_$2
+
+/* System header to define __stub macros and hopefully few prototypes,
+    which can conflict with char $2 (); below.
+    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
+    <limits.h> exists even on freestanding compilers.  */
+
+#ifdef __STDC__
+# include <limits.h>
+#else
+# include <assert.h>
+#endif
+
+#undef $2
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char $2 ();
+/* The GNU C library defines this for functions which it implements
+    to always fail with ENOSYS.  Some functions are actually named
+    something starting with __ and the normal name is an alias.  */
+#if defined __stub_$2 || defined __stub___$2
+choke me
+#endif
+
+int
+main ()
+{
+return $2 ();
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  eval "$3=yes"
+else
+  eval "$3=no"
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_check_func
 cat >config.log <<_ACEOF
 This file contains any messages produced by compilers while
 running configure, to aid debugging if configure makes a mistake.
 
 It was created by Grid $as_me 1.0, which was
-generated by GNU Autoconf 2.63.  Invocation command line was
+generated by GNU Autoconf 2.69.  Invocation command line was
 
   $ $0 $@
 
@@ -1515,8 +2046,8 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  $as_echo "PATH: $as_dir"
-done
+    $as_echo "PATH: $as_dir"
+  done
 IFS=$as_save_IFS
 
 } >&5
@@ -1553,9 +2084,9 @@ do
       ac_arg=`$as_echo "$ac_arg" | sed "s/'/'\\\\\\\\''/g"` ;;
     esac
     case $ac_pass in
-    1) ac_configure_args0="$ac_configure_args0 '$ac_arg'" ;;
+    1) as_fn_append ac_configure_args0 " '$ac_arg'" ;;
     2)
-      ac_configure_args1="$ac_configure_args1 '$ac_arg'"
+      as_fn_append ac_configure_args1 " '$ac_arg'"
       if test $ac_must_keep_next = true; then
 	ac_must_keep_next=false # Got value, back to normal.
       else
@@ -1571,13 +2102,13 @@ do
 	  -* ) ac_must_keep_next=true ;;
 	esac
       fi
-      ac_configure_args="$ac_configure_args '$ac_arg'"
+      as_fn_append ac_configure_args " '$ac_arg'"
       ;;
     esac
   done
 done
-$as_unset ac_configure_args0 || test "${ac_configure_args0+set}" != set || { ac_configure_args0=; export ac_configure_args0; }
-$as_unset ac_configure_args1 || test "${ac_configure_args1+set}" != set || { ac_configure_args1=; export ac_configure_args1; }
+{ ac_configure_args0=; unset ac_configure_args0;}
+{ ac_configure_args1=; unset ac_configure_args1;}
 
 # When interrupted or exit'd, cleanup temporary files, and complete
 # config.log.  We remove comments because anyway the quotes in there
@@ -1589,11 +2120,9 @@ trap 'exit_status=$?
   {
     echo
 
-    cat <<\_ASBOX
-## ---------------- ##
+    $as_echo "## ---------------- ##
 ## Cache variables. ##
-## ---------------- ##
-_ASBOX
+## ---------------- ##"
     echo
     # The following way of writing the cache mishandles newlines in values,
 (
@@ -1602,13 +2131,13 @@ _ASBOX
     case $ac_val in #(
     *${as_nl}*)
       case $ac_var in #(
-      *_cv_*) { $as_echo "$as_me:$LINENO: WARNING: cache variable $ac_var contains a newline" >&5
+      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
 $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
       esac
       case $ac_var in #(
       _ | IFS | as_nl) ;; #(
       BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
-      *) $as_unset $ac_var ;;
+      *) { eval $ac_var=; unset $ac_var;} ;;
       esac ;;
     esac
   done
@@ -1627,11 +2156,9 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
 )
     echo
 
-    cat <<\_ASBOX
-## ----------------- ##
+    $as_echo "## ----------------- ##
 ## Output variables. ##
-## ----------------- ##
-_ASBOX
+## ----------------- ##"
     echo
     for ac_var in $ac_subst_vars
     do
@@ -1644,11 +2171,9 @@ _ASBOX
     echo
 
     if test -n "$ac_subst_files"; then
-      cat <<\_ASBOX
-## ------------------- ##
+      $as_echo "## ------------------- ##
 ## File substitutions. ##
-## ------------------- ##
-_ASBOX
+## ------------------- ##"
       echo
       for ac_var in $ac_subst_files
       do
@@ -1662,11 +2187,9 @@ _ASBOX
     fi
 
     if test -s confdefs.h; then
-      cat <<\_ASBOX
-## ----------- ##
+      $as_echo "## ----------- ##
 ## confdefs.h. ##
-## ----------- ##
-_ASBOX
+## ----------- ##"
       echo
       cat confdefs.h
       echo
@@ -1680,46 +2203,53 @@ _ASBOX
     exit $exit_status
 ' 0
 for ac_signal in 1 2 13 15; do
-  trap 'ac_signal='$ac_signal'; { (exit 1); exit 1; }' $ac_signal
+  trap 'ac_signal='$ac_signal'; as_fn_exit 1' $ac_signal
 done
 ac_signal=0
 
 # confdefs.h avoids OS command line length limits that DEFS can exceed.
 rm -f -r conftest* confdefs.h
 
+$as_echo "/* confdefs.h */" > confdefs.h
+
 # Predefined preprocessor variables.
 
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_NAME "$PACKAGE_NAME"
 _ACEOF
 
-
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_TARNAME "$PACKAGE_TARNAME"
 _ACEOF
 
-
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_VERSION "$PACKAGE_VERSION"
 _ACEOF
 
-
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_STRING "$PACKAGE_STRING"
 _ACEOF
 
-
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_BUGREPORT "$PACKAGE_BUGREPORT"
 _ACEOF
 
+cat >>confdefs.h <<_ACEOF
+#define PACKAGE_URL "$PACKAGE_URL"
+_ACEOF
+
 
 # Let the site file select an alternate cache file if it wants to.
 # Prefer an explicitly selected file to automatically selected ones.
 ac_site_file1=NONE
 ac_site_file2=NONE
 if test -n "$CONFIG_SITE"; then
-  ac_site_file1=$CONFIG_SITE
+  # We do not want a PATH search for config.site.
+  case $CONFIG_SITE in #((
+    -*)  ac_site_file1=./$CONFIG_SITE;;
+    */*) ac_site_file1=$CONFIG_SITE;;
+    *)   ac_site_file1=./$CONFIG_SITE;;
+  esac
 elif test "x$prefix" != xNONE; then
   ac_site_file1=$prefix/share/config.site
   ac_site_file2=$prefix/etc/config.site
@@ -1730,19 +2260,23 @@ fi
 for ac_site_file in "$ac_site_file1" "$ac_site_file2"
 do
   test "x$ac_site_file" = xNONE && continue
-  if test -r "$ac_site_file"; then
-    { $as_echo "$as_me:$LINENO: loading site script $ac_site_file" >&5
+  if test /dev/null != "$ac_site_file" && test -r "$ac_site_file"; then
+    { $as_echo "$as_me:${as_lineno-$LINENO}: loading site script $ac_site_file" >&5
 $as_echo "$as_me: loading site script $ac_site_file" >&6;}
     sed 's/^/| /' "$ac_site_file" >&5
-    . "$ac_site_file"
+    . "$ac_site_file" \
+      || { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "failed to load site script $ac_site_file
+See \`config.log' for more details" "$LINENO" 5; }
   fi
 done
 
 if test -r "$cache_file"; then
-  # Some versions of bash will fail to source /dev/null (special
-  # files actually), so we avoid doing that.
-  if test -f "$cache_file"; then
-    { $as_echo "$as_me:$LINENO: loading cache $cache_file" >&5
+  # Some versions of bash will fail to source /dev/null (special files
+  # actually), so we avoid doing that.  DJGPP emulates it as a regular file.
+  if test /dev/null != "$cache_file" && test -f "$cache_file"; then
+    { $as_echo "$as_me:${as_lineno-$LINENO}: loading cache $cache_file" >&5
 $as_echo "$as_me: loading cache $cache_file" >&6;}
     case $cache_file in
       [\\/]* | ?:[\\/]* ) . "$cache_file";;
@@ -1750,7 +2284,7 @@ $as_echo "$as_me: loading cache $cache_file" >&6;}
     esac
   fi
 else
-  { $as_echo "$as_me:$LINENO: creating cache $cache_file" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: creating cache $cache_file" >&5
 $as_echo "$as_me: creating cache $cache_file" >&6;}
   >$cache_file
 fi
@@ -1765,11 +2299,11 @@ for ac_var in $ac_precious_vars; do
   eval ac_new_val=\$ac_env_${ac_var}_value
   case $ac_old_set,$ac_new_set in
     set,)
-      { $as_echo "$as_me:$LINENO: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
+      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
 $as_echo "$as_me: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&2;}
       ac_cache_corrupted=: ;;
     ,set)
-      { $as_echo "$as_me:$LINENO: error: \`$ac_var' was not set in the previous run" >&5
+      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was not set in the previous run" >&5
 $as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
       ac_cache_corrupted=: ;;
     ,);;
@@ -1779,17 +2313,17 @@ $as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
 	ac_old_val_w=`echo x $ac_old_val`
 	ac_new_val_w=`echo x $ac_new_val`
 	if test "$ac_old_val_w" != "$ac_new_val_w"; then
-	  { $as_echo "$as_me:$LINENO: error: \`$ac_var' has changed since the previous run:" >&5
+	  { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' has changed since the previous run:" >&5
 $as_echo "$as_me: error: \`$ac_var' has changed since the previous run:" >&2;}
 	  ac_cache_corrupted=:
 	else
-	  { $as_echo "$as_me:$LINENO: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
+	  { $as_echo "$as_me:${as_lineno-$LINENO}: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
 $as_echo "$as_me: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&2;}
 	  eval $ac_var=\$ac_old_val
 	fi
-	{ $as_echo "$as_me:$LINENO:   former value:  \`$ac_old_val'" >&5
+	{ $as_echo "$as_me:${as_lineno-$LINENO}:   former value:  \`$ac_old_val'" >&5
 $as_echo "$as_me:   former value:  \`$ac_old_val'" >&2;}
-	{ $as_echo "$as_me:$LINENO:   current value: \`$ac_new_val'" >&5
+	{ $as_echo "$as_me:${as_lineno-$LINENO}:   current value: \`$ac_new_val'" >&5
 $as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
       fi;;
   esac
@@ -1801,43 +2335,20 @@ $as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
     esac
     case " $ac_configure_args " in
       *" '$ac_arg' "*) ;; # Avoid dups.  Use of quotes ensures accuracy.
-      *) ac_configure_args="$ac_configure_args '$ac_arg'" ;;
+      *) as_fn_append ac_configure_args " '$ac_arg'" ;;
     esac
   fi
 done
 if $ac_cache_corrupted; then
-  { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-  { $as_echo "$as_me:$LINENO: error: changes in the environment can compromise the build" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: error: changes in the environment can compromise the build" >&5
 $as_echo "$as_me: error: changes in the environment can compromise the build" >&2;}
-  { { $as_echo "$as_me:$LINENO: error: run \`make distclean' and/or \`rm $cache_file' and start over" >&5
-$as_echo "$as_me: error: run \`make distclean' and/or \`rm $cache_file' and start over" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "run \`make distclean' and/or \`rm $cache_file' and start over" "$LINENO" 5
 fi
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+## -------------------- ##
+## Main body of script. ##
+## -------------------- ##
 
 ac_ext=c
 ac_cpp='$CPP $CPPFLAGS'
@@ -1846,7 +2357,7 @@ ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
 
-am__api_version='1.11'
+am__api_version='1.15'
 
 ac_aux_dir=
 for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
@@ -1865,9 +2376,7 @@ for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
   fi
 done
 if test -z "$ac_aux_dir"; then
-  { { $as_echo "$as_me:$LINENO: error: cannot find install-sh or install.sh in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" >&5
-$as_echo "$as_me: error: cannot find install-sh or install.sh in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "cannot find install-sh, install.sh, or shtool in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" "$LINENO" 5
 fi
 
 # These three variables are undocumented and unsupported,
@@ -1893,10 +2402,10 @@ ac_configure="$SHELL $ac_aux_dir/configure"  # Please don't use this var.
 # OS/2's system install, which has a completely different semantic
 # ./install, which can be erroneously created by make from ./install.sh.
 # Reject install programs that cannot install multiple files.
-{ $as_echo "$as_me:$LINENO: checking for a BSD-compatible install" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a BSD-compatible install" >&5
 $as_echo_n "checking for a BSD-compatible install... " >&6; }
 if test -z "$INSTALL"; then
-if test "${ac_cv_path_install+set}" = set; then
+if ${ac_cv_path_install+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
@@ -1904,11 +2413,11 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  # Account for people who put trailing slashes in PATH elements.
-case $as_dir/ in
-  ./ | .// | /cC/* | \
+    # Account for people who put trailing slashes in PATH elements.
+case $as_dir/ in #((
+  ./ | .// | /[cC]/* | \
   /etc/* | /usr/sbin/* | /usr/etc/* | /sbin/* | /usr/afsws/bin/* | \
-  ?:\\/os2\\/install\\/* | ?:\\/OS2\\/INSTALL\\/* | \
+  ?:[\\/]os2[\\/]install[\\/]* | ?:[\\/]OS2[\\/]INSTALL[\\/]* | \
   /usr/ucb/* ) ;;
   *)
     # OSF1 and SCO ODT 3.0 have their own names for install.
@@ -1916,7 +2425,7 @@ case $as_dir/ in
     # by default.
     for ac_prog in ginstall scoinst install; do
       for ac_exec_ext in '' $ac_executable_extensions; do
-	if { test -f "$as_dir/$ac_prog$ac_exec_ext" && $as_test_x "$as_dir/$ac_prog$ac_exec_ext"; }; then
+	if as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext"; then
 	  if test $ac_prog = install &&
 	    grep dspmsg "$as_dir/$ac_prog$ac_exec_ext" >/dev/null 2>&1; then
 	    # AIX install.  It has an incompatible calling convention.
@@ -1945,7 +2454,7 @@ case $as_dir/ in
     ;;
 esac
 
-done
+  done
 IFS=$as_save_IFS
 
 rm -rf conftest.one conftest.two conftest.dir
@@ -1961,7 +2470,7 @@ fi
     INSTALL=$ac_install_sh
   fi
 fi
-{ $as_echo "$as_me:$LINENO: result: $INSTALL" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $INSTALL" >&5
 $as_echo "$INSTALL" >&6; }
 
 # Use test -z because SunOS4 sh mishandles braces in ${var-val}.
@@ -1972,68 +2481,73 @@ test -z "$INSTALL_SCRIPT" && INSTALL_SCRIPT='${INSTALL}'
 
 test -z "$INSTALL_DATA" && INSTALL_DATA='${INSTALL} -m 644'
 
-{ $as_echo "$as_me:$LINENO: checking whether build environment is sane" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether build environment is sane" >&5
 $as_echo_n "checking whether build environment is sane... " >&6; }
-# Just in case
-sleep 1
-echo timestamp > conftest.file
 # Reject unsafe characters in $srcdir or the absolute working directory
 # name.  Accept space and tab only in the latter.
 am_lf='
 '
 case `pwd` in
   *[\\\"\#\$\&\'\`$am_lf]*)
-    { { $as_echo "$as_me:$LINENO: error: unsafe absolute working directory name" >&5
-$as_echo "$as_me: error: unsafe absolute working directory name" >&2;}
-   { (exit 1); exit 1; }; };;
+    as_fn_error $? "unsafe absolute working directory name" "$LINENO" 5;;
 esac
 case $srcdir in
   *[\\\"\#\$\&\'\`$am_lf\ \	]*)
-    { { $as_echo "$as_me:$LINENO: error: unsafe srcdir value: \`$srcdir'" >&5
-$as_echo "$as_me: error: unsafe srcdir value: \`$srcdir'" >&2;}
-   { (exit 1); exit 1; }; };;
+    as_fn_error $? "unsafe srcdir value: '$srcdir'" "$LINENO" 5;;
 esac
 
-# Do `set' in a subshell so we don't clobber the current shell's
+# Do 'set' in a subshell so we don't clobber the current shell's
 # arguments.  Must try -L first in case configure is actually a
 # symlink; some systems play weird games with the mod time of symlinks
 # (eg FreeBSD returns the mod time of the symlink's containing
 # directory).
 if (
-   set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
-   if test "$*" = "X"; then
-      # -L didn't work.
-      set X `ls -t "$srcdir/configure" conftest.file`
-   fi
-   rm -f conftest.file
-   if test "$*" != "X $srcdir/configure conftest.file" \
-      && test "$*" != "X conftest.file $srcdir/configure"; then
-
-      # If neither matched, then we have a broken ls.  This can happen
-      # if, for instance, CONFIG_SHELL is bash and it inherits a
-      # broken ls alias from the environment.  This has actually
-      # happened.  Such a system could not be considered "sane".
-      { { $as_echo "$as_me:$LINENO: error: ls -t appears to fail.  Make sure there is not a broken
-alias in your environment" >&5
-$as_echo "$as_me: error: ls -t appears to fail.  Make sure there is not a broken
-alias in your environment" >&2;}
-   { (exit 1); exit 1; }; }
-   fi
+   am_has_slept=no
+   for am_try in 1 2; do
+     echo "timestamp, slept: $am_has_slept" > conftest.file
+     set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
+     if test "$*" = "X"; then
+	# -L didn't work.
+	set X `ls -t "$srcdir/configure" conftest.file`
+     fi
+     if test "$*" != "X $srcdir/configure conftest.file" \
+	&& test "$*" != "X conftest.file $srcdir/configure"; then
 
+	# If neither matched, then we have a broken ls.  This can happen
+	# if, for instance, CONFIG_SHELL is bash and it inherits a
+	# broken ls alias from the environment.  This has actually
+	# happened.  Such a system could not be considered "sane".
+	as_fn_error $? "ls -t appears to fail.  Make sure there is not a broken
+  alias in your environment" "$LINENO" 5
+     fi
+     if test "$2" = conftest.file || test $am_try -eq 2; then
+       break
+     fi
+     # Just in case.
+     sleep 1
+     am_has_slept=yes
+   done
    test "$2" = conftest.file
    )
 then
    # Ok.
    :
 else
-   { { $as_echo "$as_me:$LINENO: error: newly created file is older than distributed files!
-Check your system clock" >&5
-$as_echo "$as_me: error: newly created file is older than distributed files!
-Check your system clock" >&2;}
-   { (exit 1); exit 1; }; }
+   as_fn_error $? "newly created file is older than distributed files!
+Check your system clock" "$LINENO" 5
 fi
-{ $as_echo "$as_me:$LINENO: result: yes" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
 $as_echo "yes" >&6; }
+# If we didn't sleep, we still need to ensure time stamps of config.status and
+# generated files are strictly newer.
+am_sleep_pid=
+if grep 'slept: no' conftest.file >/dev/null 2>&1; then
+  ( sleep 1 ) &
+  am_sleep_pid=$!
+fi
+
+rm -f conftest.file
+
 test "$program_prefix" != NONE &&
   program_transform_name="s&^&$program_prefix&;$program_transform_name"
 # Use a double $ so make ignores it.
@@ -2044,8 +2558,8 @@ test "$program_suffix" != NONE &&
 ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
 program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
 
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 
 if test x"${MISSING+set}" != xset; then
   case $am_aux_dir in
@@ -2056,15 +2570,15 @@ if test x"${MISSING+set}" != xset; then
   esac
 fi
 # Use eval to expand $SHELL
-if eval "$MISSING --run true"; then
-  am_missing_run="$MISSING --run "
+if eval "$MISSING --is-lightweight"; then
+  am_missing_run="$MISSING "
 else
   am_missing_run=
-  { $as_echo "$as_me:$LINENO: WARNING: \`missing' script is too old or missing" >&5
-$as_echo "$as_me: WARNING: \`missing' script is too old or missing" >&2;}
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: 'missing' script is too old or missing" >&5
+$as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
 fi
 
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -2073,17 +2587,17 @@ if test x"${install_sh}" != xset; then
   esac
 fi
 
-# Installed binaries are usually stripped using `strip' when the user
-# run `make install-strip'.  However `strip' might not be the right
+# Installed binaries are usually stripped using 'strip' when the user
+# run "make install-strip".  However 'strip' might not be the right
 # tool to use in cross-compilation environments, therefore Automake
-# will honor the `STRIP' environment variable to overrule this program.
+# will honor the 'STRIP' environment variable to overrule this program.
 if test "$cross_compiling" != no; then
   if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}strip", so it can be a program name with args.
 set dummy ${ac_tool_prefix}strip; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_STRIP+set}" = set; then
+if ${ac_cv_prog_STRIP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$STRIP"; then
@@ -2094,24 +2608,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_STRIP="${ac_tool_prefix}strip"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 STRIP=$ac_cv_prog_STRIP
 if test -n "$STRIP"; then
-  { $as_echo "$as_me:$LINENO: result: $STRIP" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $STRIP" >&5
 $as_echo "$STRIP" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2121,9 +2635,9 @@ if test -z "$ac_cv_prog_STRIP"; then
   ac_ct_STRIP=$STRIP
   # Extract the first word of "strip", so it can be a program name with args.
 set dummy strip; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_ac_ct_STRIP+set}" = set; then
+if ${ac_cv_prog_ac_ct_STRIP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_STRIP"; then
@@ -2134,24 +2648,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_ac_ct_STRIP="strip"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_STRIP=$ac_cv_prog_ac_ct_STRIP
 if test -n "$ac_ct_STRIP"; then
-  { $as_echo "$as_me:$LINENO: result: $ac_ct_STRIP" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_STRIP" >&5
 $as_echo "$ac_ct_STRIP" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2160,7 +2674,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -2173,10 +2687,10 @@ fi
 fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 
-{ $as_echo "$as_me:$LINENO: checking for a thread-safe mkdir -p" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a thread-safe mkdir -p" >&5
 $as_echo_n "checking for a thread-safe mkdir -p... " >&6; }
 if test -z "$MKDIR_P"; then
-  if test "${ac_cv_path_mkdir+set}" = set; then
+  if ${ac_cv_path_mkdir+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
@@ -2184,9 +2698,9 @@ for as_dir in $PATH$PATH_SEPARATOR/opt/sfw/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_prog in mkdir gmkdir; do
+    for ac_prog in mkdir gmkdir; do
 	 for ac_exec_ext in '' $ac_executable_extensions; do
-	   { test -f "$as_dir/$ac_prog$ac_exec_ext" && $as_test_x "$as_dir/$ac_prog$ac_exec_ext"; } || continue
+	   as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext" || continue
 	   case `"$as_dir/$ac_prog$ac_exec_ext" --version 2>&1` in #(
 	     'mkdir (GNU coreutils) '* | \
 	     'mkdir (coreutils) '* | \
@@ -2196,11 +2710,12 @@ do
 	   esac
 	 done
        done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 
+  test -d ./--version && rmdir ./--version
   if test "${ac_cv_path_mkdir+set}" = set; then
     MKDIR_P="$ac_cv_path_mkdir -p"
   else
@@ -2208,26 +2723,19 @@ fi
     # value for MKDIR_P within a source directory, because that will
     # break other packages using the cache if that directory is
     # removed, or if the value is a relative name.
-    test -d ./--version && rmdir ./--version
     MKDIR_P="$ac_install_sh -d"
   fi
 fi
-{ $as_echo "$as_me:$LINENO: result: $MKDIR_P" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $MKDIR_P" >&5
 $as_echo "$MKDIR_P" >&6; }
 
-mkdir_p="$MKDIR_P"
-case $mkdir_p in
-  [\\/$]* | ?:[\\/]*) ;;
-  */*) mkdir_p="\$(top_builddir)/$mkdir_p" ;;
-esac
-
 for ac_prog in gawk mawk nawk awk
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_AWK+set}" = set; then
+if ${ac_cv_prog_AWK+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$AWK"; then
@@ -2238,24 +2746,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_AWK="$ac_prog"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 AWK=$ac_cv_prog_AWK
 if test -n "$AWK"; then
-  { $as_echo "$as_me:$LINENO: result: $AWK" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $AWK" >&5
 $as_echo "$AWK" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2263,11 +2771,11 @@ fi
   test -n "$AWK" && break
 done
 
-{ $as_echo "$as_me:$LINENO: checking whether ${MAKE-make} sets \$(MAKE)" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether ${MAKE-make} sets \$(MAKE)" >&5
 $as_echo_n "checking whether ${MAKE-make} sets \$(MAKE)... " >&6; }
 set x ${MAKE-make}
 ac_make=`$as_echo "$2" | sed 's/+/p/g; s/[^a-zA-Z0-9_]/_/g'`
-if { as_var=ac_cv_prog_make_${ac_make}_set; eval "test \"\${$as_var+set}\" = set"; }; then
+if eval \${ac_cv_prog_make_${ac_make}_set+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   cat >conftest.make <<\_ACEOF
@@ -2275,7 +2783,7 @@ SHELL = /bin/sh
 all:
 	@echo '@@@%%%=$(MAKE)=@@@%%%'
 _ACEOF
-# GNU make sometimes prints "make[1]: Entering...", which would confuse us.
+# GNU make sometimes prints "make[1]: Entering ...", which would confuse us.
 case `${MAKE-make} -f conftest.make 2>/dev/null` in
   *@@@%%%=?*=@@@%%%*)
     eval ac_cv_prog_make_${ac_make}_set=yes;;
@@ -2285,11 +2793,11 @@ esac
 rm -f conftest.make
 fi
 if eval test \$ac_cv_prog_make_${ac_make}_set = yes; then
-  { $as_echo "$as_me:$LINENO: result: yes" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
 $as_echo "yes" >&6; }
   SET_MAKE=
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
   SET_MAKE="MAKE=${MAKE-make}"
 fi
@@ -2303,15 +2811,52 @@ else
 fi
 rmdir .tst 2>/dev/null
 
+# Check whether --enable-silent-rules was given.
+if test "${enable_silent_rules+set}" = set; then :
+  enableval=$enable_silent_rules;
+fi
+
+case $enable_silent_rules in # (((
+  yes) AM_DEFAULT_VERBOSITY=0;;
+   no) AM_DEFAULT_VERBOSITY=1;;
+    *) AM_DEFAULT_VERBOSITY=1;;
+esac
+am_make=${MAKE-make}
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $am_make supports nested variables" >&5
+$as_echo_n "checking whether $am_make supports nested variables... " >&6; }
+if ${am_cv_make_support_nested_variables+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if $as_echo 'TRUE=$(BAR$(V))
+BAR0=false
+BAR1=true
+V=1
+am__doit:
+	@$(TRUE)
+.PHONY: am__doit' | $am_make -f - >/dev/null 2>&1; then
+  am_cv_make_support_nested_variables=yes
+else
+  am_cv_make_support_nested_variables=no
+fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_make_support_nested_variables" >&5
+$as_echo "$am_cv_make_support_nested_variables" >&6; }
+if test $am_cv_make_support_nested_variables = yes; then
+    AM_V='$(V)'
+  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
+else
+  AM_V=$AM_DEFAULT_VERBOSITY
+  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
+fi
+AM_BACKSLASH='\'
+
 if test "`cd $srcdir && pwd`" != "`pwd`"; then
   # Use -I$(srcdir) only when $(srcdir) != ., so that make's output
   # is not polluted with repeated "-I."
   am__isrc=' -I$(srcdir)'
   # test to see if srcdir already configured
   if test -f $srcdir/config.status; then
-    { { $as_echo "$as_me:$LINENO: error: source directory already configured; run \"make distclean\" there first" >&5
-$as_echo "$as_me: error: source directory already configured; run \"make distclean\" there first" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "source directory already configured; run \"make distclean\" there first" "$LINENO" 5
   fi
 fi
 
@@ -2355,19 +2900,72 @@ AUTOHEADER=${AUTOHEADER-"${am_missing_run}autoheader"}
 
 MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
-# Always define AMTAR for backward compatibility.
+# For better backward compatibility.  To be removed once Automake 1.9.x
+# dies out for good.  For more background, see:
+# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
+# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
+mkdir_p='$(MKDIR_P)'
 
-AMTAR=${AMTAR-"${am_missing_run}tar"}
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
+# Always define AMTAR for backward compatibility.  Yes, it's still used
+# in the wild :-(  We should find a proper way to deprecate it ...
+AMTAR='$${TAR-tar}'
 
-am__tar='${AMTAR} chof - "$$tardir"'; am__untar='${AMTAR} xf -'
+
+# We'll loop over all known methods to create a tar archive until one works.
+_am_tools='gnutar  pax cpio none'
+
+am__tar='$${TAR-tar} chof - "$$tardir"' am__untar='$${TAR-tar} xf -'
 
 
 
 
 
 
+# POSIX will say in a future version that running "rm -f" with no argument
+# is OK; and we want to be able to make that assumption in our Makefile
+# recipes.  So use an aggressive probe to check that the usage we want is
+# actually supported "in the wild" to an acceptable degree.
+# See automake bug#10828.
+# To make any issue more visible, cause the running configure to be aborted
+# by default if the 'rm' program in use doesn't match our expectations; the
+# user can still override this though.
+if rm -f && rm -fr && rm -rf; then : OK; else
+  cat >&2 <<'END'
+Oops!
+
+Your 'rm' program seems unable to run without file operands specified
+on the command line, even when the '-f' option is present.  This is contrary
+to the behaviour of most rm programs out there, and not conforming with
+the upcoming POSIX standard: <http://austingroupbugs.net/view.php?id=542>
+
+Please tell bug-automake@gnu.org about your system, including the value
+of your $PATH and any error possibly output before this message.  This
+can help us improve future automake versions.
+
+END
+  if test x"$ACCEPT_INFERIOR_RM_PROGRAM" = x"yes"; then
+    echo 'Configuration will proceed anyway, since you have set the' >&2
+    echo 'ACCEPT_INFERIOR_RM_PROGRAM variable to "yes"' >&2
+    echo >&2
+  else
+    cat >&2 <<'END'
+Aborting the configuration process, to ensure you take notice of the issue.
+
+You can download and install GNU coreutils to get an 'rm' implementation
+that behaves properly: <http://www.gnu.org/software/coreutils/>.
+
+If you want to complete the configuration process using your problematic
+'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
+to "yes", and re-run configure.
+
+END
+    as_fn_error $? "Your 'rm' program is bad, sorry." "$LINENO" 5
+  fi
+fi
+
+
 
 ac_config_headers="$ac_config_headers lib/Grid_config.h"
 
@@ -2387,9 +2985,9 @@ if test -z "$CXX"; then
   do
     # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
 set dummy $ac_tool_prefix$ac_prog; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_CXX+set}" = set; then
+if ${ac_cv_prog_CXX+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CXX"; then
@@ -2400,24 +2998,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_CXX="$ac_tool_prefix$ac_prog"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 CXX=$ac_cv_prog_CXX
 if test -n "$CXX"; then
-  { $as_echo "$as_me:$LINENO: result: $CXX" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CXX" >&5
 $as_echo "$CXX" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2431,9 +3029,9 @@ if test -z "$CXX"; then
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_ac_ct_CXX+set}" = set; then
+if ${ac_cv_prog_ac_ct_CXX+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CXX"; then
@@ -2444,24 +3042,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_ac_ct_CXX="$ac_prog"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CXX=$ac_cv_prog_ac_ct_CXX
 if test -n "$ac_ct_CXX"; then
-  { $as_echo "$as_me:$LINENO: result: $ac_ct_CXX" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CXX" >&5
 $as_echo "$ac_ct_CXX" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2474,7 +3072,7 @@ done
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -2485,48 +3083,31 @@ fi
   fi
 fi
 # Provide some information about the compiler.
-$as_echo "$as_me:$LINENO: checking for C++ compiler version" >&5
+$as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler version" >&5
 set X $ac_compile
 ac_compiler=$2
-{ (ac_try="$ac_compiler --version >&5"
+for ac_option in --version -v -V -qversion; do
+  { { ac_try="$ac_compiler $ac_option >&5"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler --version >&5") 2>&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
-{ (ac_try="$ac_compiler -v >&5"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler -v >&5") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
-{ (ac_try="$ac_compiler -V >&5"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler -V >&5") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
+  if test -s conftest.err; then
+    sed '10a\
+... rest of stderr output deleted ...
+         10q' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+  fi
+  rm -f conftest.er1 conftest.err
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }
+done
 
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2542,8 +3123,8 @@ ac_clean_files="$ac_clean_files a.out a.out.dSYM a.exe b.out"
 # Try to create an executable without -o first, disregard a.out.
 # It will help us diagnose broken compilers, and finding out an intuition
 # of exeext.
-{ $as_echo "$as_me:$LINENO: checking for C++ compiler default output file name" >&5
-$as_echo_n "checking for C++ compiler default output file name... " >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether the C++ compiler works" >&5
+$as_echo_n "checking whether the C++ compiler works... " >&6; }
 ac_link_default=`$as_echo "$ac_link" | sed 's/ -o *conftest[^ ]*//'`
 
 # The possible output files:
@@ -2559,17 +3140,17 @@ do
 done
 rm -f $ac_rmfiles
 
-if { (ac_try="$ac_link_default"
+if { { ac_try="$ac_link_default"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
   (eval "$ac_link_default") 2>&5
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; then
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then :
   # Autoconf-2.13 could set the ac_cv_exeext variable to `no'.
 # So ignore a value of `no', otherwise this would lead to `EXEEXT = no'
 # in a Makefile.  We should not override ac_cv_exeext if it was cached,
@@ -2586,7 +3167,7 @@ do
 	# certainly right.
 	break;;
     *.* )
-        if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
+	if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
 	then :; else
 	   ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
 	fi
@@ -2605,84 +3186,41 @@ test "$ac_cv_exeext" = no && ac_cv_exeext=
 else
   ac_file=''
 fi
-
-{ $as_echo "$as_me:$LINENO: result: $ac_file" >&5
-$as_echo "$ac_file" >&6; }
-if test -z "$ac_file"; then
-  $as_echo "$as_me: failed program was:" >&5
+if test -z "$ac_file"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+$as_echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
 
-{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: C++ compiler cannot create executables
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: C++ compiler cannot create executables
-See \`config.log' for more details." >&2;}
-   { (exit 77); exit 77; }; }; }
-fi
-
-ac_exeext=$ac_cv_exeext
-
-# Check that the compiler produces executables we can run.  If not, either
-# the compiler is broken, or we cross compile.
-{ $as_echo "$as_me:$LINENO: checking whether the C++ compiler works" >&5
-$as_echo_n "checking whether the C++ compiler works... " >&6; }
-# FIXME: These cross compiler hacks should be removed for Autoconf 3.0
-# If not cross compiling, check that we can run a simple program.
-if test "$cross_compiling" != yes; then
-  if { ac_try='./$ac_file'
-  { (case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; }; then
-    cross_compiling=no
-  else
-    if test "$cross_compiling" = maybe; then
-	cross_compiling=yes
-    else
-	{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: cannot run C++ compiled programs.
-If you meant to cross compile, use \`--host'.
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: cannot run C++ compiled programs.
-If you meant to cross compile, use \`--host'.
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
-    fi
-  fi
-fi
-{ $as_echo "$as_me:$LINENO: result: yes" >&5
+as_fn_error 77 "C++ compiler cannot create executables
+See \`config.log' for more details" "$LINENO" 5; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
 $as_echo "yes" >&6; }
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler default output file name" >&5
+$as_echo_n "checking for C++ compiler default output file name... " >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_file" >&5
+$as_echo "$ac_file" >&6; }
+ac_exeext=$ac_cv_exeext
 
 rm -f -r a.out a.out.dSYM a.exe conftest$ac_cv_exeext b.out
 ac_clean_files=$ac_clean_files_save
-# Check that the compiler produces executables we can run.  If not, either
-# the compiler is broken, or we cross compile.
-{ $as_echo "$as_me:$LINENO: checking whether we are cross compiling" >&5
-$as_echo_n "checking whether we are cross compiling... " >&6; }
-{ $as_echo "$as_me:$LINENO: result: $cross_compiling" >&5
-$as_echo "$cross_compiling" >&6; }
-
-{ $as_echo "$as_me:$LINENO: checking for suffix of executables" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of executables" >&5
 $as_echo_n "checking for suffix of executables... " >&6; }
-if { (ac_try="$ac_link"
+if { { ac_try="$ac_link"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
   (eval "$ac_link") 2>&5
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; then
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then :
   # If both `conftest.exe' and `conftest' are `present' (well, observable)
 # catch `conftest.exe'.  For instance with Cygwin, `ls conftest' will
 # work properly (i.e., refer to `conftest.exe'), while it won't with
@@ -2697,32 +3235,83 @@ for ac_file in conftest.exe conftest conftest.*; do
   esac
 done
 else
-  { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: cannot compute suffix of executables: cannot compile and link
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: cannot compute suffix of executables: cannot compile and link
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
+as_fn_error $? "cannot compute suffix of executables: cannot compile and link
+See \`config.log' for more details" "$LINENO" 5; }
 fi
-
-rm -f conftest$ac_cv_exeext
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_exeext" >&5
+rm -f conftest conftest$ac_cv_exeext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_exeext" >&5
 $as_echo "$ac_cv_exeext" >&6; }
 
 rm -f conftest.$ac_ext
 EXEEXT=$ac_cv_exeext
 ac_exeext=$EXEEXT
-{ $as_echo "$as_me:$LINENO: checking for suffix of object files" >&5
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <stdio.h>
+int
+main ()
+{
+FILE *f = fopen ("conftest.out", "w");
+ return ferror (f) || fclose (f) != 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+ac_clean_files="$ac_clean_files conftest.out"
+# Check that the compiler produces executables we can run.  If not, either
+# the compiler is broken, or we cross compile.
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are cross compiling" >&5
+$as_echo_n "checking whether we are cross compiling... " >&6; }
+if test "$cross_compiling" != yes; then
+  { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }
+  if { ac_try='./conftest$ac_cv_exeext'
+  { { case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; }; then
+    cross_compiling=no
+  else
+    if test "$cross_compiling" = maybe; then
+	cross_compiling=yes
+    else
+	{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "cannot run C++ compiled programs.
+If you meant to cross compile, use \`--host'.
+See \`config.log' for more details" "$LINENO" 5; }
+    fi
+  fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $cross_compiling" >&5
+$as_echo "$cross_compiling" >&6; }
+
+rm -f conftest.$ac_ext conftest$ac_cv_exeext conftest.out
+ac_clean_files=$ac_clean_files_save
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of object files" >&5
 $as_echo_n "checking for suffix of object files... " >&6; }
-if test "${ac_cv_objext+set}" = set; then
+if ${ac_cv_objext+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2734,17 +3323,17 @@ main ()
 }
 _ACEOF
 rm -f conftest.o conftest.obj
-if { (ac_try="$ac_compile"
+if { { ac_try="$ac_compile"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
   (eval "$ac_compile") 2>&5
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; then
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then :
   for ac_file in conftest.o conftest.obj conftest.*; do
   test -f "$ac_file" || continue;
   case $ac_file in
@@ -2757,31 +3346,23 @@ else
   $as_echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
 
-{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: cannot compute suffix of object files: cannot compile
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: cannot compute suffix of object files: cannot compile
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
+as_fn_error $? "cannot compute suffix of object files: cannot compile
+See \`config.log' for more details" "$LINENO" 5; }
 fi
-
 rm -f conftest.$ac_cv_objext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_objext" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_objext" >&5
 $as_echo "$ac_cv_objext" >&6; }
 OBJEXT=$ac_cv_objext
 ac_objext=$OBJEXT
-{ $as_echo "$as_me:$LINENO: checking whether we are using the GNU C++ compiler" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C++ compiler" >&5
 $as_echo_n "checking whether we are using the GNU C++ compiler... " >&6; }
-if test "${ac_cv_cxx_compiler_gnu+set}" = set; then
+if ${ac_cv_cxx_compiler_gnu+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2795,37 +3376,16 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_cxx_try_compile "$LINENO"; then :
   ac_compiler_gnu=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_compiler_gnu=no
+  ac_compiler_gnu=no
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 ac_cv_cxx_compiler_gnu=$ac_compiler_gnu
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_cxx_compiler_gnu" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_cxx_compiler_gnu" >&5
 $as_echo "$ac_cv_cxx_compiler_gnu" >&6; }
 if test $ac_compiler_gnu = yes; then
   GXX=yes
@@ -2834,20 +3394,16 @@ else
 fi
 ac_test_CXXFLAGS=${CXXFLAGS+set}
 ac_save_CXXFLAGS=$CXXFLAGS
-{ $as_echo "$as_me:$LINENO: checking whether $CXX accepts -g" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX accepts -g" >&5
 $as_echo_n "checking whether $CXX accepts -g... " >&6; }
-if test "${ac_cv_prog_cxx_g+set}" = set; then
+if ${ac_cv_prog_cxx_g+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   ac_save_cxx_werror_flag=$ac_cxx_werror_flag
    ac_cxx_werror_flag=yes
    ac_cv_prog_cxx_g=no
    CXXFLAGS="-g"
-   cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2858,35 +3414,11 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_cxx_try_compile "$LINENO"; then :
   ac_cv_prog_cxx_g=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	CXXFLAGS=""
-      cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  CXXFLAGS=""
+      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2897,36 +3429,12 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+if ac_fn_cxx_try_compile "$LINENO"; then :
 
-	ac_cxx_werror_flag=$ac_save_cxx_werror_flag
+else
+  ac_cxx_werror_flag=$ac_save_cxx_werror_flag
 	 CXXFLAGS="-g"
-	 cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2937,42 +3445,17 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_cxx_try_compile "$LINENO"; then :
   ac_cv_prog_cxx_g=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
    ac_cxx_werror_flag=$ac_save_cxx_werror_flag
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cxx_g" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cxx_g" >&5
 $as_echo "$ac_cv_prog_cxx_g" >&6; }
 if test "$ac_test_CXXFLAGS" = set; then
   CXXFLAGS=$ac_save_CXXFLAGS
@@ -3006,14 +3489,14 @@ am__doit:
 .PHONY: am__doit
 END
 # If we don't find an include directive, just comment out the code.
-{ $as_echo "$as_me:$LINENO: checking for style of include used by $am_make" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for style of include used by $am_make" >&5
 $as_echo_n "checking for style of include used by $am_make... " >&6; }
 am__include="#"
 am__quote=
 _am_result=none
 # First try GNU make style include.
 echo "include confinc" > confmf
-# Ignore all kinds of additional output from `make'.
+# Ignore all kinds of additional output from 'make'.
 case `$am_make -s -f confmf 2> /dev/null` in #(
 *the\ am__doit\ target*)
   am__include=include
@@ -3034,18 +3517,19 @@ if test "$am__include" = "#"; then
 fi
 
 
-{ $as_echo "$as_me:$LINENO: result: $_am_result" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $_am_result" >&5
 $as_echo "$_am_result" >&6; }
 rm -f confinc confmf
 
 # Check whether --enable-dependency-tracking was given.
-if test "${enable_dependency_tracking+set}" = set; then
+if test "${enable_dependency_tracking+set}" = set; then :
   enableval=$enable_dependency_tracking;
 fi
 
 if test "x$enable_dependency_tracking" != xno; then
   am_depcomp="$ac_aux_dir/depcomp"
   AMDEPBACKSLASH='\'
+  am__nodep='_no'
 fi
  if test "x$enable_dependency_tracking" != xno; then
   AMDEP_TRUE=
@@ -3059,17 +3543,18 @@ fi
 
 depcc="$CXX"  am_compiler_list=
 
-{ $as_echo "$as_me:$LINENO: checking dependency style of $depcc" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
 $as_echo_n "checking dependency style of $depcc... " >&6; }
-if test "${am_cv_CXX_dependencies_compiler_type+set}" = set; then
+if ${am_cv_CXX_dependencies_compiler_type+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named `D' -- because `-MD' means `put the output
-  # in D'.
+  # making a dummy file named 'D' -- because '-MD' means "put the output
+  # in D".
+  rm -rf conftest.dir
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -3103,16 +3588,16 @@ else
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
-      # Solaris 8's {/usr,}/bin/sh.
-      touch sub/conftst$i.h
+      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
+      # Solaris 10 /bin/sh.
+      echo '/* dummy */' > sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with `-c' and `-o' for the sake of the "dashmstdout"
+    # We check with '-c' and '-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle `-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs
+    # handle '-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs.
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -3121,16 +3606,16 @@ else
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # after this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested
+      # After this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested.
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvisualcpp | msvcmsys)
-      # This compiler won't grok `-c -o', but also, the minuso test has
+    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
+      # This compiler won't grok '-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -3169,7 +3654,7 @@ else
 fi
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $am_cv_CXX_dependencies_compiler_type" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CXX_dependencies_compiler_type" >&5
 $as_echo "$am_cv_CXX_dependencies_compiler_type" >&6; }
 CXXDEPMODE=depmode=$am_cv_CXX_dependencies_compiler_type
 
@@ -3192,9 +3677,9 @@ ac_compiler_gnu=$ac_cv_c_compiler_gnu
 if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}gcc", so it can be a program name with args.
 set dummy ${ac_tool_prefix}gcc; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_CC+set}" = set; then
+if ${ac_cv_prog_CC+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3205,24 +3690,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_CC="${ac_tool_prefix}gcc"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:$LINENO: result: $CC" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3232,9 +3717,9 @@ if test -z "$ac_cv_prog_CC"; then
   ac_ct_CC=$CC
   # Extract the first word of "gcc", so it can be a program name with args.
 set dummy gcc; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_ac_ct_CC+set}" = set; then
+if ${ac_cv_prog_ac_ct_CC+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CC"; then
@@ -3245,24 +3730,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_ac_ct_CC="gcc"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CC=$ac_cv_prog_ac_ct_CC
 if test -n "$ac_ct_CC"; then
-  { $as_echo "$as_me:$LINENO: result: $ac_ct_CC" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
 $as_echo "$ac_ct_CC" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3271,7 +3756,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -3285,9 +3770,9 @@ if test -z "$CC"; then
           if test -n "$ac_tool_prefix"; then
     # Extract the first word of "${ac_tool_prefix}cc", so it can be a program name with args.
 set dummy ${ac_tool_prefix}cc; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_CC+set}" = set; then
+if ${ac_cv_prog_CC+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3298,24 +3783,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_CC="${ac_tool_prefix}cc"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:$LINENO: result: $CC" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3325,9 +3810,9 @@ fi
 if test -z "$CC"; then
   # Extract the first word of "cc", so it can be a program name with args.
 set dummy cc; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_CC+set}" = set; then
+if ${ac_cv_prog_CC+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3339,18 +3824,18 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     if test "$as_dir/$ac_word$ac_exec_ext" = "/usr/ucb/cc"; then
        ac_prog_rejected=yes
        continue
      fi
     ac_cv_prog_CC="cc"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 if test $ac_prog_rejected = yes; then
@@ -3369,10 +3854,10 @@ fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:$LINENO: result: $CC" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3384,9 +3869,9 @@ if test -z "$CC"; then
   do
     # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
 set dummy $ac_tool_prefix$ac_prog; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_CC+set}" = set; then
+if ${ac_cv_prog_CC+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CC"; then
@@ -3397,24 +3882,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_CC="$ac_tool_prefix$ac_prog"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 CC=$ac_cv_prog_CC
 if test -n "$CC"; then
-  { $as_echo "$as_me:$LINENO: result: $CC" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
 $as_echo "$CC" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3428,9 +3913,9 @@ if test -z "$CC"; then
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_ac_ct_CC+set}" = set; then
+if ${ac_cv_prog_ac_ct_CC+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CC"; then
@@ -3441,24 +3926,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_ac_ct_CC="$ac_prog"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CC=$ac_cv_prog_ac_ct_CC
 if test -n "$ac_ct_CC"; then
-  { $as_echo "$as_me:$LINENO: result: $ac_ct_CC" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
 $as_echo "$ac_ct_CC" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3471,7 +3956,7 @@ done
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -3482,62 +3967,42 @@ fi
 fi
 
 
-test -z "$CC" && { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+test -z "$CC" && { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: no acceptable C compiler found in \$PATH
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: no acceptable C compiler found in \$PATH
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
+as_fn_error $? "no acceptable C compiler found in \$PATH
+See \`config.log' for more details" "$LINENO" 5; }
 
 # Provide some information about the compiler.
-$as_echo "$as_me:$LINENO: checking for C compiler version" >&5
+$as_echo "$as_me:${as_lineno-$LINENO}: checking for C compiler version" >&5
 set X $ac_compile
 ac_compiler=$2
-{ (ac_try="$ac_compiler --version >&5"
+for ac_option in --version -v -V -qversion; do
+  { { ac_try="$ac_compiler $ac_option >&5"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler --version >&5") 2>&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
-{ (ac_try="$ac_compiler -v >&5"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler -v >&5") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
-{ (ac_try="$ac_compiler -V >&5"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler -V >&5") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
+  if test -s conftest.err; then
+    sed '10a\
+... rest of stderr output deleted ...
+         10q' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+  fi
+  rm -f conftest.er1 conftest.err
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }
+done
 
-{ $as_echo "$as_me:$LINENO: checking whether we are using the GNU C compiler" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C compiler" >&5
 $as_echo_n "checking whether we are using the GNU C compiler... " >&6; }
-if test "${ac_cv_c_compiler_gnu+set}" = set; then
+if ${ac_cv_c_compiler_gnu+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -3551,37 +4016,16 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_c_try_compile "$LINENO"; then :
   ac_compiler_gnu=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_compiler_gnu=no
+  ac_compiler_gnu=no
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 ac_cv_c_compiler_gnu=$ac_compiler_gnu
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_compiler_gnu" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_c_compiler_gnu" >&5
 $as_echo "$ac_cv_c_compiler_gnu" >&6; }
 if test $ac_compiler_gnu = yes; then
   GCC=yes
@@ -3590,20 +4034,16 @@ else
 fi
 ac_test_CFLAGS=${CFLAGS+set}
 ac_save_CFLAGS=$CFLAGS
-{ $as_echo "$as_me:$LINENO: checking whether $CC accepts -g" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC accepts -g" >&5
 $as_echo_n "checking whether $CC accepts -g... " >&6; }
-if test "${ac_cv_prog_cc_g+set}" = set; then
+if ${ac_cv_prog_cc_g+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   ac_save_c_werror_flag=$ac_c_werror_flag
    ac_c_werror_flag=yes
    ac_cv_prog_cc_g=no
    CFLAGS="-g"
-   cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -3614,35 +4054,11 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_c_try_compile "$LINENO"; then :
   ac_cv_prog_cc_g=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	CFLAGS=""
-      cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  CFLAGS=""
+      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -3653,36 +4069,12 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+if ac_fn_c_try_compile "$LINENO"; then :
 
-	ac_c_werror_flag=$ac_save_c_werror_flag
+else
+  ac_c_werror_flag=$ac_save_c_werror_flag
 	 CFLAGS="-g"
-	 cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -3693,42 +4085,17 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_c_try_compile "$LINENO"; then :
   ac_cv_prog_cc_g=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
    ac_c_werror_flag=$ac_save_c_werror_flag
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cc_g" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_g" >&5
 $as_echo "$ac_cv_prog_cc_g" >&6; }
 if test "$ac_test_CFLAGS" = set; then
   CFLAGS=$ac_save_CFLAGS
@@ -3745,23 +4112,18 @@ else
     CFLAGS=
   fi
 fi
-{ $as_echo "$as_me:$LINENO: checking for $CC option to accept ISO C89" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to accept ISO C89" >&5
 $as_echo_n "checking for $CC option to accept ISO C89... " >&6; }
-if test "${ac_cv_prog_cc_c89+set}" = set; then
+if ${ac_cv_prog_cc_c89+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   ac_cv_prog_cc_c89=no
 ac_save_CC=$CC
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <stdarg.h>
 #include <stdio.h>
-#include <sys/types.h>
-#include <sys/stat.h>
+struct stat;
 /* Most of the following tests are stolen from RCS 5.7's src/conf.sh.  */
 struct buf { int x; };
 FILE * (*rcsopen) (struct buf *, struct stat *, int);
@@ -3813,32 +4175,9 @@ for ac_arg in '' -qlanglvl=extc89 -qlanglvl=ansi -std \
 	-Ae "-Aa -D_HPUX_SOURCE" "-Xc -D__EXTENSIONS__"
 do
   CC="$ac_save_CC $ac_arg"
-  rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+  if ac_fn_c_try_compile "$LINENO"; then :
   ac_cv_prog_cc_c89=$ac_arg
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
 fi
-
 rm -f core conftest.err conftest.$ac_objext
   test "x$ac_cv_prog_cc_c89" != "xno" && break
 done
@@ -3849,17 +4188,19 @@ fi
 # AC_CACHE_VAL
 case "x$ac_cv_prog_cc_c89" in
   x)
-    { $as_echo "$as_me:$LINENO: result: none needed" >&5
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: none needed" >&5
 $as_echo "none needed" >&6; } ;;
   xno)
-    { $as_echo "$as_me:$LINENO: result: unsupported" >&5
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: unsupported" >&5
 $as_echo "unsupported" >&6; } ;;
   *)
     CC="$CC $ac_cv_prog_cc_c89"
-    { $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cc_c89" >&5
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_c89" >&5
 $as_echo "$ac_cv_prog_cc_c89" >&6; } ;;
 esac
+if test "x$ac_cv_prog_cc_c89" != xno; then :
 
+fi
 
 ac_ext=c
 ac_cpp='$CPP $CPPFLAGS'
@@ -3867,19 +4208,79 @@ ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
 ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC understands -c and -o together" >&5
+$as_echo_n "checking whether $CC understands -c and -o together... " >&6; }
+if ${am_cv_prog_cc_c_o+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+  # Make sure it works both with $CC and with simple cc.
+  # Following AC_PROG_CC_C_O, we do the test twice because some
+  # compilers refuse to overwrite an existing .o file with -o,
+  # though they will create one.
+  am_cv_prog_cc_c_o=yes
+  for am_i in 1 2; do
+    if { echo "$as_me:$LINENO: $CC -c conftest.$ac_ext -o conftest2.$ac_objext" >&5
+   ($CC -c conftest.$ac_ext -o conftest2.$ac_objext) >&5 2>&5
+   ac_status=$?
+   echo "$as_me:$LINENO: \$? = $ac_status" >&5
+   (exit $ac_status); } \
+         && test -f conftest2.$ac_objext; then
+      : OK
+    else
+      am_cv_prog_cc_c_o=no
+      break
+    fi
+  done
+  rm -f core conftest*
+  unset am_i
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_prog_cc_c_o" >&5
+$as_echo "$am_cv_prog_cc_c_o" >&6; }
+if test "$am_cv_prog_cc_c_o" != yes; then
+   # Losing compiler, so override with the script.
+   # FIXME: It is wrong to rewrite CC.
+   # But if we don't then we get into trouble of one sort or another.
+   # A longer-term fix would be to have automake use am__CC in this case,
+   # and then we could set am__CC="\$(top_srcdir)/compile \$(CC)"
+   CC="$am_aux_dir/compile $CC"
+fi
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+
+
 depcc="$CC"   am_compiler_list=
 
-{ $as_echo "$as_me:$LINENO: checking dependency style of $depcc" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
 $as_echo_n "checking dependency style of $depcc... " >&6; }
-if test "${am_cv_CC_dependencies_compiler_type+set}" = set; then
+if ${am_cv_CC_dependencies_compiler_type+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named `D' -- because `-MD' means `put the output
-  # in D'.
+  # making a dummy file named 'D' -- because '-MD' means "put the output
+  # in D".
+  rm -rf conftest.dir
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -3913,16 +4314,16 @@ else
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
-      # Solaris 8's {/usr,}/bin/sh.
-      touch sub/conftst$i.h
+      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
+      # Solaris 10 /bin/sh.
+      echo '/* dummy */' > sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with `-c' and `-o' for the sake of the "dashmstdout"
+    # We check with '-c' and '-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle `-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs
+    # handle '-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs.
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -3931,16 +4332,16 @@ else
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # after this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested
+      # After this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested.
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvisualcpp | msvcmsys)
-      # This compiler won't grok `-c -o', but also, the minuso test has
+    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
+      # This compiler won't grok '-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -3979,7 +4380,7 @@ else
 fi
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $am_cv_CC_dependencies_compiler_type" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CC_dependencies_compiler_type" >&5
 $as_echo "$am_cv_CC_dependencies_compiler_type" >&6; }
 CCDEPMODE=depmode=$am_cv_CC_dependencies_compiler_type
 
@@ -3998,17 +4399,18 @@ fi
 
   OPENMP_CFLAGS=
   # Check whether --enable-openmp was given.
-if test "${enable_openmp+set}" = set; then
+if test "${enable_openmp+set}" = set; then :
   enableval=$enable_openmp;
 fi
 
   if test "$enable_openmp" != no; then
-    { $as_echo "$as_me:$LINENO: checking for $CC option to support OpenMP" >&5
+    { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to support OpenMP" >&5
 $as_echo_n "checking for $CC option to support OpenMP... " >&6; }
-if test "${ac_cv_prog_c_openmp+set}" = set; then
+if ${ac_cv_prog_c_openmp+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
 
 #ifndef _OPENMP
  choke me
@@ -4017,37 +4419,16 @@ else
 int main () { return omp_get_num_threads (); }
 
 _ACEOF
-rm -f conftest.$ac_objext conftest$ac_exeext
-if { (ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_link") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 $as_test_x conftest$ac_exeext
-       }; then
+if ac_fn_c_try_link "$LINENO"; then :
   ac_cv_prog_c_openmp='none needed'
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_cv_prog_c_openmp='unsupported'
-	  	  	  	  	  	  	  	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp; do
+  ac_cv_prog_c_openmp='unsupported'
+	  	  	  	  	  	  	                                	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp -homp \
+                           -Popenmp --openmp; do
 	    ac_save_CFLAGS=$CFLAGS
 	    CFLAGS="$CFLAGS $ac_option"
-	    cat >conftest.$ac_ext <<_ACEOF
+	    cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
 
 #ifndef _OPENMP
  choke me
@@ -4056,56 +4437,27 @@ sed 's/^/| /' conftest.$ac_ext >&5
 int main () { return omp_get_num_threads (); }
 
 _ACEOF
-rm -f conftest.$ac_objext conftest$ac_exeext
-if { (ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_link") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 $as_test_x conftest$ac_exeext
-       }; then
+if ac_fn_c_try_link "$LINENO"; then :
   ac_cv_prog_c_openmp=$ac_option
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
 fi
-
-rm -rf conftest.dSYM
-rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
-      conftest$ac_exeext conftest.$ac_ext
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
 	    CFLAGS=$ac_save_CFLAGS
 	    if test "$ac_cv_prog_c_openmp" != unsupported; then
 	      break
 	    fi
 	  done
 fi
-
-rm -rf conftest.dSYM
-rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
-      conftest$ac_exeext conftest.$ac_ext
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_c_openmp" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_c_openmp" >&5
 $as_echo "$ac_cv_prog_c_openmp" >&6; }
     case $ac_cv_prog_c_openmp in #(
       "none needed" | unsupported)
-        ;; #(
+	;; #(
       *)
-        OPENMP_CFLAGS=$ac_cv_prog_c_openmp ;;
+	OPENMP_CFLAGS=$ac_cv_prog_c_openmp ;;
     esac
   fi
 
@@ -4113,9 +4465,9 @@ $as_echo "$ac_cv_prog_c_openmp" >&6; }
 if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}ranlib", so it can be a program name with args.
 set dummy ${ac_tool_prefix}ranlib; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_RANLIB+set}" = set; then
+if ${ac_cv_prog_RANLIB+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$RANLIB"; then
@@ -4126,24 +4478,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_RANLIB="${ac_tool_prefix}ranlib"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 RANLIB=$ac_cv_prog_RANLIB
 if test -n "$RANLIB"; then
-  { $as_echo "$as_me:$LINENO: result: $RANLIB" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $RANLIB" >&5
 $as_echo "$RANLIB" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -4153,9 +4505,9 @@ if test -z "$ac_cv_prog_RANLIB"; then
   ac_ct_RANLIB=$RANLIB
   # Extract the first word of "ranlib", so it can be a program name with args.
 set dummy ranlib; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_ac_ct_RANLIB+set}" = set; then
+if ${ac_cv_prog_ac_ct_RANLIB+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_RANLIB"; then
@@ -4166,24 +4518,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_ac_ct_RANLIB="ranlib"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_RANLIB=$ac_cv_prog_ac_ct_RANLIB
 if test -n "$ac_ct_RANLIB"; then
-  { $as_echo "$as_me:$LINENO: result: $ac_ct_RANLIB" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_RANLIB" >&5
 $as_echo "$ac_ct_RANLIB" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -4192,7 +4544,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -4212,14 +4564,14 @@ ac_cpp='$CPP $CPPFLAGS'
 ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
 ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
-{ $as_echo "$as_me:$LINENO: checking how to run the C preprocessor" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking how to run the C preprocessor" >&5
 $as_echo_n "checking how to run the C preprocessor... " >&6; }
 # On Suns, sometimes $CPP names a directory.
 if test -n "$CPP" && test -d "$CPP"; then
   CPP=
 fi
 if test -z "$CPP"; then
-  if test "${ac_cv_prog_CPP+set}" = set; then
+  if ${ac_cv_prog_CPP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
       # Double quotes because CPP needs to be expanded
@@ -4234,11 +4586,7 @@ do
   # <limits.h> exists even on freestanding compilers.
   # On the NeXT, cc -E runs the code through the compiler's parser,
   # not just through cpp. "Syntax error" is here to catch this case.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #ifdef __STDC__
 # include <limits.h>
@@ -4247,78 +4595,34 @@ cat >>conftest.$ac_ext <<_ACEOF
 #endif
 		     Syntax error
 _ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+if ac_fn_c_try_cpp "$LINENO"; then :
 
+else
   # Broken: fails on valid input.
 continue
 fi
-
-rm -f conftest.err conftest.$ac_ext
+rm -f conftest.err conftest.i conftest.$ac_ext
 
   # OK, works on sane cases.  Now check whether nonexistent headers
   # can be detected and how.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <ac_nonexistent.h>
 _ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
+if ac_fn_c_try_cpp "$LINENO"; then :
   # Broken: success on invalid input.
 continue
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
   # Passes both tests.
 ac_preproc_ok=:
 break
 fi
-
-rm -f conftest.err conftest.$ac_ext
+rm -f conftest.err conftest.i conftest.$ac_ext
 
 done
 # Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
-rm -f conftest.err conftest.$ac_ext
-if $ac_preproc_ok; then
+rm -f conftest.i conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then :
   break
 fi
 
@@ -4330,7 +4634,7 @@ fi
 else
   ac_cv_prog_CPP=$CPP
 fi
-{ $as_echo "$as_me:$LINENO: result: $CPP" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $CPP" >&5
 $as_echo "$CPP" >&6; }
 ac_preproc_ok=false
 for ac_c_preproc_warn_flag in '' yes
@@ -4341,11 +4645,7 @@ do
   # <limits.h> exists even on freestanding compilers.
   # On the NeXT, cc -E runs the code through the compiler's parser,
   # not just through cpp. "Syntax error" is here to catch this case.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #ifdef __STDC__
 # include <limits.h>
@@ -4354,87 +4654,40 @@ cat >>conftest.$ac_ext <<_ACEOF
 #endif
 		     Syntax error
 _ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+if ac_fn_c_try_cpp "$LINENO"; then :
 
+else
   # Broken: fails on valid input.
 continue
 fi
-
-rm -f conftest.err conftest.$ac_ext
+rm -f conftest.err conftest.i conftest.$ac_ext
 
   # OK, works on sane cases.  Now check whether nonexistent headers
   # can be detected and how.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <ac_nonexistent.h>
 _ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
+if ac_fn_c_try_cpp "$LINENO"; then :
   # Broken: success on invalid input.
 continue
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
   # Passes both tests.
 ac_preproc_ok=:
 break
 fi
-
-rm -f conftest.err conftest.$ac_ext
+rm -f conftest.err conftest.i conftest.$ac_ext
 
 done
 # Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
-rm -f conftest.err conftest.$ac_ext
-if $ac_preproc_ok; then
-  :
+rm -f conftest.i conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then :
+
 else
-  { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: C preprocessor \"$CPP\" fails sanity check
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: C preprocessor \"$CPP\" fails sanity check
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
+as_fn_error $? "C preprocessor \"$CPP\" fails sanity check
+See \`config.log' for more details" "$LINENO" 5; }
 fi
 
 ac_ext=c
@@ -4444,9 +4697,9 @@ ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
 
-{ $as_echo "$as_me:$LINENO: checking for grep that handles long lines and -e" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for grep that handles long lines and -e" >&5
 $as_echo_n "checking for grep that handles long lines and -e... " >&6; }
-if test "${ac_cv_path_GREP+set}" = set; then
+if ${ac_cv_path_GREP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -z "$GREP"; then
@@ -4457,10 +4710,10 @@ for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_prog in grep ggrep; do
+    for ac_prog in grep ggrep; do
     for ac_exec_ext in '' $ac_executable_extensions; do
       ac_path_GREP="$as_dir/$ac_prog$ac_exec_ext"
-      { test -f "$ac_path_GREP" && $as_test_x "$ac_path_GREP"; } || continue
+      as_fn_executable_p "$ac_path_GREP" || continue
 # Check for GNU ac_path_GREP and select it if it is found.
   # Check for GNU $ac_path_GREP
 case `"$ac_path_GREP" --version 2>&1` in
@@ -4477,7 +4730,7 @@ case `"$ac_path_GREP" --version 2>&1` in
     $as_echo 'GREP' >> "conftest.nl"
     "$ac_path_GREP" -e 'GREP$' -e '-(cannot match)-' < "conftest.nl" >"conftest.out" 2>/dev/null || break
     diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
-    ac_count=`expr $ac_count + 1`
+    as_fn_arith $ac_count + 1 && ac_count=$as_val
     if test $ac_count -gt ${ac_path_GREP_max-0}; then
       # Best one so far, save it but keep looking for a better one
       ac_cv_path_GREP="$ac_path_GREP"
@@ -4492,26 +4745,24 @@ esac
       $ac_path_GREP_found && break 3
     done
   done
-done
+  done
 IFS=$as_save_IFS
   if test -z "$ac_cv_path_GREP"; then
-    { { $as_echo "$as_me:$LINENO: error: no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&5
-$as_echo "$as_me: error: no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
   fi
 else
   ac_cv_path_GREP=$GREP
 fi
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_path_GREP" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_GREP" >&5
 $as_echo "$ac_cv_path_GREP" >&6; }
  GREP="$ac_cv_path_GREP"
 
 
-{ $as_echo "$as_me:$LINENO: checking for egrep" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for egrep" >&5
 $as_echo_n "checking for egrep... " >&6; }
-if test "${ac_cv_path_EGREP+set}" = set; then
+if ${ac_cv_path_EGREP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if echo a | $GREP -E '(a|b)' >/dev/null 2>&1
@@ -4525,10 +4776,10 @@ for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_prog in egrep; do
+    for ac_prog in egrep; do
     for ac_exec_ext in '' $ac_executable_extensions; do
       ac_path_EGREP="$as_dir/$ac_prog$ac_exec_ext"
-      { test -f "$ac_path_EGREP" && $as_test_x "$ac_path_EGREP"; } || continue
+      as_fn_executable_p "$ac_path_EGREP" || continue
 # Check for GNU ac_path_EGREP and select it if it is found.
   # Check for GNU $ac_path_EGREP
 case `"$ac_path_EGREP" --version 2>&1` in
@@ -4545,7 +4796,7 @@ case `"$ac_path_EGREP" --version 2>&1` in
     $as_echo 'EGREP' >> "conftest.nl"
     "$ac_path_EGREP" 'EGREP$' < "conftest.nl" >"conftest.out" 2>/dev/null || break
     diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
-    ac_count=`expr $ac_count + 1`
+    as_fn_arith $ac_count + 1 && ac_count=$as_val
     if test $ac_count -gt ${ac_path_EGREP_max-0}; then
       # Best one so far, save it but keep looking for a better one
       ac_cv_path_EGREP="$ac_path_EGREP"
@@ -4560,12 +4811,10 @@ esac
       $ac_path_EGREP_found && break 3
     done
   done
-done
+  done
 IFS=$as_save_IFS
   if test -z "$ac_cv_path_EGREP"; then
-    { { $as_echo "$as_me:$LINENO: error: no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&5
-$as_echo "$as_me: error: no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
   fi
 else
   ac_cv_path_EGREP=$EGREP
@@ -4573,21 +4822,17 @@ fi
 
    fi
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_path_EGREP" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_EGREP" >&5
 $as_echo "$ac_cv_path_EGREP" >&6; }
  EGREP="$ac_cv_path_EGREP"
 
 
-{ $as_echo "$as_me:$LINENO: checking for ANSI C header files" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for ANSI C header files" >&5
 $as_echo_n "checking for ANSI C header files... " >&6; }
-if test "${ac_cv_header_stdc+set}" = set; then
+if ${ac_cv_header_stdc+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <stdlib.h>
 #include <stdarg.h>
@@ -4602,48 +4847,23 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_c_try_compile "$LINENO"; then :
   ac_cv_header_stdc=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_cv_header_stdc=no
+  ac_cv_header_stdc=no
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 
 if test $ac_cv_header_stdc = yes; then
   # SunOS 4.x string.h does not declare mem*, contrary to ANSI.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <string.h>
 
 _ACEOF
 if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
-  $EGREP "memchr" >/dev/null 2>&1; then
-  :
+  $EGREP "memchr" >/dev/null 2>&1; then :
+
 else
   ac_cv_header_stdc=no
 fi
@@ -4653,18 +4873,14 @@ fi
 
 if test $ac_cv_header_stdc = yes; then
   # ISC 2.0.2 stdlib.h does not declare free, contrary to ANSI.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <stdlib.h>
 
 _ACEOF
 if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
-  $EGREP "free" >/dev/null 2>&1; then
-  :
+  $EGREP "free" >/dev/null 2>&1; then :
+
 else
   ac_cv_header_stdc=no
 fi
@@ -4674,14 +4890,10 @@ fi
 
 if test $ac_cv_header_stdc = yes; then
   # /bin/cc in Irix-4.0.5 gets non-ANSI ctype macros unless using -ansi.
-  if test "$cross_compiling" = yes; then
+  if test "$cross_compiling" = yes; then :
   :
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <ctype.h>
 #include <stdlib.h>
@@ -4708,118 +4920,33 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest$ac_exeext
-if { (ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_link") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && { ac_try='./conftest$ac_exeext'
-  { (case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; }; then
-  :
+if ac_fn_c_try_run "$LINENO"; then :
+
 else
-  $as_echo "$as_me: program exited with status $ac_status" >&5
-$as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-( exit $ac_status )
-ac_cv_header_stdc=no
+  ac_cv_header_stdc=no
 fi
-rm -rf conftest.dSYM
-rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext conftest.$ac_objext conftest.$ac_ext
+rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
+  conftest.$ac_objext conftest.beam conftest.$ac_ext
 fi
 
-
 fi
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_header_stdc" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_header_stdc" >&5
 $as_echo "$ac_cv_header_stdc" >&6; }
 if test $ac_cv_header_stdc = yes; then
 
-cat >>confdefs.h <<\_ACEOF
-#define STDC_HEADERS 1
-_ACEOF
+$as_echo "#define STDC_HEADERS 1" >>confdefs.h
 
 fi
 
 # On IRIX 5.3, sys/types and inttypes.h are conflicting.
-
-
-
-
-
-
-
-
-
 for ac_header in sys/types.h sys/stat.h stdlib.h string.h memory.h strings.h \
 		  inttypes.h stdint.h unistd.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  eval "$as_ac_Header=yes"
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	eval "$as_ac_Header=no"
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+ac_fn_c_check_header_compile "$LINENO" "$ac_header" "$as_ac_Header" "$ac_includes_default
+"
+if eval test \"x\$"$as_ac_Header"\" = x"yes"; then :
   cat >>confdefs.h <<_ACEOF
 #define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
@@ -4829,604 +4956,48 @@ fi
 done
 
 
-
 for ac_header in stdint.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
-$as_echo_n "checking $ac_header usability... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_header_compiler=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_header_compiler=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
-$as_echo_n "checking $ac_header presence... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <$ac_header>
-_ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  ac_header_preproc=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-  ac_header_preproc=no
-fi
-
-rm -f conftest.err conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
-  yes:no: )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
-    ac_header_preproc=yes
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
-$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
-    ( cat <<\_ASBOX
-## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##
-_ASBOX
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  eval "$as_ac_Header=\$ac_header_preproc"
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-
-fi
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "stdint.h" "ac_cv_header_stdint_h" "$ac_includes_default"
+if test "x$ac_cv_header_stdint_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+#define HAVE_STDINT_H 1
 _ACEOF
 
 fi
 
 done
 
-
 for ac_header in malloc/malloc.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
-$as_echo_n "checking $ac_header usability... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_header_compiler=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_header_compiler=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
-$as_echo_n "checking $ac_header presence... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <$ac_header>
-_ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  ac_header_preproc=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-  ac_header_preproc=no
-fi
-
-rm -f conftest.err conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
-  yes:no: )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
-    ac_header_preproc=yes
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
-$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
-    ( cat <<\_ASBOX
-## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##
-_ASBOX
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  eval "$as_ac_Header=\$ac_header_preproc"
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-
-fi
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "malloc/malloc.h" "ac_cv_header_malloc_malloc_h" "$ac_includes_default"
+if test "x$ac_cv_header_malloc_malloc_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+#define HAVE_MALLOC_MALLOC_H 1
 _ACEOF
 
 fi
 
 done
 
-
 for ac_header in malloc.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
-$as_echo_n "checking $ac_header usability... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_header_compiler=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_header_compiler=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
-$as_echo_n "checking $ac_header presence... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <$ac_header>
-_ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  ac_header_preproc=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-  ac_header_preproc=no
-fi
-
-rm -f conftest.err conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
-  yes:no: )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
-    ac_header_preproc=yes
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
-$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
-    ( cat <<\_ASBOX
-## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##
-_ASBOX
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  eval "$as_ac_Header=\$ac_header_preproc"
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-
-fi
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "malloc.h" "ac_cv_header_malloc_h" "$ac_includes_default"
+if test "x$ac_cv_header_malloc_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+#define HAVE_MALLOC_H 1
 _ACEOF
 
 fi
 
 done
 
-
 for ac_header in endian.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
-$as_echo_n "checking $ac_header usability... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_header_compiler=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_header_compiler=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
-$as_echo_n "checking $ac_header presence... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <$ac_header>
-_ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  ac_header_preproc=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-  ac_header_preproc=no
-fi
-
-rm -f conftest.err conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in
-  yes:no: )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
-    ac_header_preproc=yes
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
-$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
-    ( cat <<\_ASBOX
-## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##
-_ASBOX
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  eval "$as_ac_Header=\$ac_header_preproc"
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-
-fi
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_c_check_header_mongrel "$LINENO" "endian.h" "ac_cv_header_endian_h" "$ac_includes_default"
+if test "x$ac_cv_header_endian_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+#define HAVE_ENDIAN_H 1
 _ACEOF
 
 fi
@@ -5434,244 +5005,35 @@ fi
 done
 
 #AC_CHECK_HEADERS(machine/endian.h)
-{ $as_echo "$as_me:$LINENO: checking whether ntohll is declared" >&5
-$as_echo_n "checking whether ntohll is declared... " >&6; }
-if test "${ac_cv_have_decl_ntohll+set}" = set; then
-  $as_echo_n "(cached) " >&6
+ac_fn_c_check_decl "$LINENO" "ntohll" "ac_cv_have_decl_ntohll" "#include <arpa/inet.h>
+"
+if test "x$ac_cv_have_decl_ntohll" = xyes; then :
+  ac_have_decl=1
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <arpa/inet.h>
-
-int
-main ()
-{
-#ifndef ntohll
-  (void) ntohll;
-#endif
-
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_cv_have_decl_ntohll=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_cv_have_decl_ntohll=no
+  ac_have_decl=0
 fi
 
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_have_decl_ntohll" >&5
-$as_echo "$ac_cv_have_decl_ntohll" >&6; }
-if test "x$ac_cv_have_decl_ntohll" = x""yes; then
-
 cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_NTOHLL 1
+#define HAVE_DECL_NTOHLL $ac_have_decl
 _ACEOF
 
-
+ac_fn_c_check_decl "$LINENO" "be64toh" "ac_cv_have_decl_be64toh" "#include <arpa/inet.h>
+"
+if test "x$ac_cv_have_decl_be64toh" = xyes; then :
+  ac_have_decl=1
 else
-  cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_NTOHLL 0
-_ACEOF
-
-
+  ac_have_decl=0
 fi
 
-
-{ $as_echo "$as_me:$LINENO: checking whether be64toh is declared" >&5
-$as_echo_n "checking whether be64toh is declared... " >&6; }
-if test "${ac_cv_have_decl_be64toh+set}" = set; then
-  $as_echo_n "(cached) " >&6
-else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <arpa/inet.h>
-
-int
-main ()
-{
-#ifndef be64toh
-  (void) be64toh;
-#endif
-
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_cv_have_decl_be64toh=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_cv_have_decl_be64toh=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_have_decl_be64toh" >&5
-$as_echo "$ac_cv_have_decl_be64toh" >&6; }
-if test "x$ac_cv_have_decl_be64toh" = x""yes; then
-
 cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_BE64TOH 1
+#define HAVE_DECL_BE64TOH $ac_have_decl
 _ACEOF
 
 
-else
-  cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_BE64TOH 0
-_ACEOF
-
-
-fi
-
-
-
 # Checks for typedefs, structures, and compiler characteristics.
-{ $as_echo "$as_me:$LINENO: checking for size_t" >&5
-$as_echo_n "checking for size_t... " >&6; }
-if test "${ac_cv_type_size_t+set}" = set; then
-  $as_echo_n "(cached) " >&6
-else
-  ac_cv_type_size_t=no
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-if (sizeof (size_t))
-       return 0;
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-if (sizeof ((size_t)))
-	  return 0;
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+ac_fn_c_check_type "$LINENO" "size_t" "ac_cv_type_size_t" "$ac_includes_default"
+if test "x$ac_cv_type_size_t" = xyes; then :
 
-	ac_cv_type_size_t=yes
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_type_size_t" >&5
-$as_echo "$ac_cv_type_size_t" >&6; }
-if test "x$ac_cv_type_size_t" = x""yes; then
-  :
 else
 
 cat >>confdefs.h <<_ACEOF
@@ -5680,75 +5042,12 @@ _ACEOF
 
 fi
 
-
-  { $as_echo "$as_me:$LINENO: checking for uint32_t" >&5
-$as_echo_n "checking for uint32_t... " >&6; }
-if test "${ac_cv_c_uint32_t+set}" = set; then
-  $as_echo_n "(cached) " >&6
-else
-  ac_cv_c_uint32_t=no
-     for ac_type in 'uint32_t' 'unsigned int' 'unsigned long int' \
-	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
-       cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-static int test_array [1 - 2 * !(($ac_type) -1 >> (32 - 1) == 1)];
-test_array [0] = 0
-
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  case $ac_type in
-  uint32_t) ac_cv_c_uint32_t=yes ;;
-  *) ac_cv_c_uint32_t=$ac_type ;;
-esac
-
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-       test "$ac_cv_c_uint32_t" != no && break
-     done
-fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_uint32_t" >&5
-$as_echo "$ac_cv_c_uint32_t" >&6; }
-  case $ac_cv_c_uint32_t in #(
+ac_fn_c_find_uintX_t "$LINENO" "32" "ac_cv_c_uint32_t"
+case $ac_cv_c_uint32_t in #(
   no|yes) ;; #(
   *)
 
-cat >>confdefs.h <<\_ACEOF
-#define _UINT32_T 1
-_ACEOF
+$as_echo "#define _UINT32_T 1" >>confdefs.h
 
 
 cat >>confdefs.h <<_ACEOF
@@ -5757,75 +5056,12 @@ _ACEOF
 ;;
   esac
 
-
-  { $as_echo "$as_me:$LINENO: checking for uint64_t" >&5
-$as_echo_n "checking for uint64_t... " >&6; }
-if test "${ac_cv_c_uint64_t+set}" = set; then
-  $as_echo_n "(cached) " >&6
-else
-  ac_cv_c_uint64_t=no
-     for ac_type in 'uint64_t' 'unsigned int' 'unsigned long int' \
-	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
-       cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-static int test_array [1 - 2 * !(($ac_type) -1 >> (64 - 1) == 1)];
-test_array [0] = 0
-
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  case $ac_type in
-  uint64_t) ac_cv_c_uint64_t=yes ;;
-  *) ac_cv_c_uint64_t=$ac_type ;;
-esac
-
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-       test "$ac_cv_c_uint64_t" != no && break
-     done
-fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_uint64_t" >&5
-$as_echo "$ac_cv_c_uint64_t" >&6; }
-  case $ac_cv_c_uint64_t in #(
+ac_fn_c_find_uintX_t "$LINENO" "64" "ac_cv_c_uint64_t"
+case $ac_cv_c_uint64_t in #(
   no|yes) ;; #(
   *)
 
-cat >>confdefs.h <<\_ACEOF
-#define _UINT64_T 1
-_ACEOF
+$as_echo "#define _UINT64_T 1" >>confdefs.h
 
 
 cat >>confdefs.h <<_ACEOF
@@ -5836,102 +5072,12 @@ _ACEOF
 
 
 # Checks for library functions.
-
 for ac_func in gettimeofday
-do
-as_ac_var=`$as_echo "ac_cv_func_$ac_func" | $as_tr_sh`
-{ $as_echo "$as_me:$LINENO: checking for $ac_func" >&5
-$as_echo_n "checking for $ac_func... " >&6; }
-if { as_var=$as_ac_var; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-/* Define $ac_func to an innocuous variant, in case <limits.h> declares $ac_func.
-   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
-#define $ac_func innocuous_$ac_func
-
-/* System header to define __stub macros and hopefully few prototypes,
-    which can conflict with char $ac_func (); below.
-    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
-    <limits.h> exists even on freestanding compilers.  */
-
-#ifdef __STDC__
-# include <limits.h>
-#else
-# include <assert.h>
-#endif
-
-#undef $ac_func
-
-/* Override any GCC internal prototype to avoid an error.
-   Use char because int might match the return type of a GCC
-   builtin and then its argument prototype would still apply.  */
-#ifdef __cplusplus
-extern "C"
-#endif
-char $ac_func ();
-/* The GNU C library defines this for functions which it implements
-    to always fail with ENOSYS.  Some functions are actually named
-    something starting with __ and the normal name is an alias.  */
-#if defined __stub_$ac_func || defined __stub___$ac_func
-choke me
-#endif
-
-int
-main ()
-{
-return $ac_func ();
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext conftest$ac_exeext
-if { (ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_link") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 $as_test_x conftest$ac_exeext
-       }; then
-  eval "$as_ac_var=yes"
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	eval "$as_ac_var=no"
-fi
-
-rm -rf conftest.dSYM
-rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
-      conftest$ac_exeext conftest.$ac_ext
-fi
-ac_res=`eval 'as_val=${'$as_ac_var'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-as_val=`eval 'as_val=${'$as_ac_var'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_c_check_func "$LINENO" "gettimeofday" "ac_cv_func_gettimeofday"
+if test "x$ac_cv_func_gettimeofday" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_func" | $as_tr_cpp` 1
+#define HAVE_GETTIMEOFDAY 1
 _ACEOF
 
 fi
@@ -5939,7 +5085,7 @@ done
 
 
 # Check whether --enable-simd was given.
-if test "${enable_simd+set}" = set; then
+if test "${enable_simd+set}" = set; then :
   enableval=$enable_simd; \
 	ac_SIMD=${enable_simd}
 else
@@ -5951,45 +5097,35 @@ case ${ac_SIMD} in
      SSE4)
        echo Configuring for SSE4
 
-cat >>confdefs.h <<\_ACEOF
-#define SSE4 1
-_ACEOF
+$as_echo "#define SSE4 1" >>confdefs.h
 
      ;;
      AVX)
        echo Configuring for AVX
 
-cat >>confdefs.h <<\_ACEOF
-#define AVX1 1
-_ACEOF
+$as_echo "#define AVX1 1" >>confdefs.h
 
      ;;
      AVX2)
        echo Configuring for AVX2
 
-cat >>confdefs.h <<\_ACEOF
-#define AVX2 1
-_ACEOF
+$as_echo "#define AVX2 1" >>confdefs.h
 
      ;;
      AVX512|MIC)
        echo Configuring for AVX512 and MIC
 
-cat >>confdefs.h <<\_ACEOF
-#define AVX512 1
-_ACEOF
+$as_echo "#define AVX512 1" >>confdefs.h
 
      ;;
      *)
-     { { $as_echo "$as_me:$LINENO: error: ${ac_SIMD} unsupported --enable-simd option" >&5
-$as_echo "$as_me: error: ${ac_SIMD} unsupported --enable-simd option" >&2;}
-   { (exit 1); exit 1; }; };
+     as_fn_error $? "${ac_SIMD} unsupported --enable-simd option" "$LINENO" 5;
      ;;
 esac
 
 
 # Check whether --enable-comms was given.
-if test "${enable_comms+set}" = set; then
+if test "${enable_comms+set}" = set; then :
   enableval=$enable_comms; ac_COMMS=${enable_comms}
 else
   ac_COMMS=none
@@ -6000,23 +5136,17 @@ case ${ac_COMMS} in
      none)
        echo Configuring for NO communications
 
-cat >>confdefs.h <<\_ACEOF
-#define GRID_COMMS_NONE 1
-_ACEOF
+$as_echo "#define GRID_COMMS_NONE 1" >>confdefs.h
 
      ;;
      mpi)
        echo Configuring for MPI communications
 
-cat >>confdefs.h <<\_ACEOF
-#define GRID_COMMS_MPI 1
-_ACEOF
+$as_echo "#define GRID_COMMS_MPI 1" >>confdefs.h
 
      ;;
      *)
-     { { $as_echo "$as_me:$LINENO: error: ${ac_COMMS} unsupported --enable-comms option" >&5
-$as_echo "$as_me: error: ${ac_COMMS} unsupported --enable-comms option" >&2;}
-   { (exit 1); exit 1; }; };
+     as_fn_error $? "${ac_COMMS} unsupported --enable-comms option" "$LINENO" 5;
      ;;
 esac
 
@@ -6073,13 +5203,13 @@ _ACEOF
     case $ac_val in #(
     *${as_nl}*)
       case $ac_var in #(
-      *_cv_*) { $as_echo "$as_me:$LINENO: WARNING: cache variable $ac_var contains a newline" >&5
+      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
 $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
       esac
       case $ac_var in #(
       _ | IFS | as_nl) ;; #(
       BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
-      *) $as_unset $ac_var ;;
+      *) { eval $ac_var=; unset $ac_var;} ;;
       esac ;;
     esac
   done
@@ -6087,8 +5217,8 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
   (set) 2>&1 |
     case $as_nl`(ac_space=' '; set) 2>&1` in #(
     *${as_nl}ac_space=\ *)
-      # `set' does not quote correctly, so add quotes (double-quote
-      # substitution turns \\\\ into \\, and sed turns \\ into \).
+      # `set' does not quote correctly, so add quotes: double-quote
+      # substitution turns \\\\ into \\, and sed turns \\ into \.
       sed -n \
 	"s/'/'\\\\''/g;
 	  s/^\\([_$as_cr_alnum]*_cv_[_$as_cr_alnum]*\\)=\\(.*\\)/\\1='\\2'/p"
@@ -6110,12 +5240,23 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
      :end' >>confcache
 if diff "$cache_file" confcache >/dev/null 2>&1; then :; else
   if test -w "$cache_file"; then
-    test "x$cache_file" != "x/dev/null" &&
-      { $as_echo "$as_me:$LINENO: updating cache $cache_file" >&5
+    if test "x$cache_file" != "x/dev/null"; then
+      { $as_echo "$as_me:${as_lineno-$LINENO}: updating cache $cache_file" >&5
 $as_echo "$as_me: updating cache $cache_file" >&6;}
-    cat confcache >$cache_file
+      if test ! -f "$cache_file" || test -h "$cache_file"; then
+	cat confcache >"$cache_file"
+      else
+        case $cache_file in #(
+        */* | ?:*)
+	  mv -f confcache "$cache_file"$$ &&
+	  mv -f "$cache_file"$$ "$cache_file" ;; #(
+        *)
+	  mv -f confcache "$cache_file" ;;
+	esac
+      fi
+    fi
   else
-    { $as_echo "$as_me:$LINENO: not updating unwritable cache $cache_file" >&5
+    { $as_echo "$as_me:${as_lineno-$LINENO}: not updating unwritable cache $cache_file" >&5
 $as_echo "$as_me: not updating unwritable cache $cache_file" >&6;}
   fi
 fi
@@ -6129,20 +5270,29 @@ DEFS=-DHAVE_CONFIG_H
 
 ac_libobjs=
 ac_ltlibobjs=
+U=
 for ac_i in : $LIBOBJS; do test "x$ac_i" = x: && continue
   # 1. Remove the extension, and $U if already installed.
   ac_script='s/\$U\././;s/\.o$//;s/\.obj$//'
   ac_i=`$as_echo "$ac_i" | sed "$ac_script"`
   # 2. Prepend LIBOBJDIR.  When used with automake>=1.10 LIBOBJDIR
   #    will be set to the directory where LIBOBJS objects are built.
-  ac_libobjs="$ac_libobjs \${LIBOBJDIR}$ac_i\$U.$ac_objext"
-  ac_ltlibobjs="$ac_ltlibobjs \${LIBOBJDIR}$ac_i"'$U.lo'
+  as_fn_append ac_libobjs " \${LIBOBJDIR}$ac_i\$U.$ac_objext"
+  as_fn_append ac_ltlibobjs " \${LIBOBJDIR}$ac_i"'$U.lo'
 done
 LIBOBJS=$ac_libobjs
 
 LTLIBOBJS=$ac_ltlibobjs
 
 
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking that generated files are newer than configure" >&5
+$as_echo_n "checking that generated files are newer than configure... " >&6; }
+   if test -n "$am_sleep_pid"; then
+     # Hide warnings about reused PIDs.
+     wait $am_sleep_pid 2>/dev/null
+   fi
+   { $as_echo "$as_me:${as_lineno-$LINENO}: result: done" >&5
+$as_echo "done" >&6; }
  if test -n "$EXEEXT"; then
   am__EXEEXT_TRUE=
   am__EXEEXT_FALSE='#'
@@ -6152,48 +5302,34 @@ else
 fi
 
 if test -z "${AMDEP_TRUE}" && test -z "${AMDEP_FALSE}"; then
-  { { $as_echo "$as_me:$LINENO: error: conditional \"AMDEP\" was never defined.
-Usually this means the macro was only invoked conditionally." >&5
-$as_echo "$as_me: error: conditional \"AMDEP\" was never defined.
-Usually this means the macro was only invoked conditionally." >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "conditional \"AMDEP\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 if test -z "${am__fastdepCXX_TRUE}" && test -z "${am__fastdepCXX_FALSE}"; then
-  { { $as_echo "$as_me:$LINENO: error: conditional \"am__fastdepCXX\" was never defined.
-Usually this means the macro was only invoked conditionally." >&5
-$as_echo "$as_me: error: conditional \"am__fastdepCXX\" was never defined.
-Usually this means the macro was only invoked conditionally." >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "conditional \"am__fastdepCXX\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 if test -z "${am__fastdepCC_TRUE}" && test -z "${am__fastdepCC_FALSE}"; then
-  { { $as_echo "$as_me:$LINENO: error: conditional \"am__fastdepCC\" was never defined.
-Usually this means the macro was only invoked conditionally." >&5
-$as_echo "$as_me: error: conditional \"am__fastdepCC\" was never defined.
-Usually this means the macro was only invoked conditionally." >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "conditional \"am__fastdepCC\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 if test -z "${BUILD_COMMS_MPI_TRUE}" && test -z "${BUILD_COMMS_MPI_FALSE}"; then
-  { { $as_echo "$as_me:$LINENO: error: conditional \"BUILD_COMMS_MPI\" was never defined.
-Usually this means the macro was only invoked conditionally." >&5
-$as_echo "$as_me: error: conditional \"BUILD_COMMS_MPI\" was never defined.
-Usually this means the macro was only invoked conditionally." >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "conditional \"BUILD_COMMS_MPI\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 if test -z "${BUILD_COMMS_NONE_TRUE}" && test -z "${BUILD_COMMS_NONE_FALSE}"; then
-  { { $as_echo "$as_me:$LINENO: error: conditional \"BUILD_COMMS_NONE\" was never defined.
-Usually this means the macro was only invoked conditionally." >&5
-$as_echo "$as_me: error: conditional \"BUILD_COMMS_NONE\" was never defined.
-Usually this means the macro was only invoked conditionally." >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "conditional \"BUILD_COMMS_NONE\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 
-: ${CONFIG_STATUS=./config.status}
+: "${CONFIG_STATUS=./config.status}"
 ac_write_fail=0
 ac_clean_files_save=$ac_clean_files
 ac_clean_files="$ac_clean_files $CONFIG_STATUS"
-{ $as_echo "$as_me:$LINENO: creating $CONFIG_STATUS" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: creating $CONFIG_STATUS" >&5
 $as_echo "$as_me: creating $CONFIG_STATUS" >&6;}
-cat >$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+as_write_fail=0
+cat >$CONFIG_STATUS <<_ASEOF || as_write_fail=1
 #! $SHELL
 # Generated by $as_me.
 # Run this file to recreate the current configuration.
@@ -6203,17 +5339,18 @@ cat >$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 debug=false
 ac_cs_recheck=false
 ac_cs_silent=false
-SHELL=\${CONFIG_SHELL-$SHELL}
-_ACEOF
 
-cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-## --------------------- ##
-## M4sh Initialization.  ##
-## --------------------- ##
+SHELL=\${CONFIG_SHELL-$SHELL}
+export SHELL
+_ASEOF
+cat >>$CONFIG_STATUS <<\_ASEOF || as_write_fail=1
+## -------------------- ##
+## M4sh Initialization. ##
+## -------------------- ##
 
 # Be more Bourne compatible
 DUALCASE=1; export DUALCASE # for MKS sh
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
   emulate sh
   NULLCMD=:
   # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
@@ -6221,23 +5358,15 @@ if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
   alias -g '${1+"$@"}'='"$@"'
   setopt NO_GLOB_SUBST
 else
-  case `(set -o) 2>/dev/null` in
-  *posix*) set -o posix ;;
+  case `(set -o) 2>/dev/null` in #(
+  *posix*) :
+    set -o posix ;; #(
+  *) :
+     ;;
 esac
-
 fi
 
 
-
-
-# PATH needs CR
-# Avoid depending upon Character Ranges.
-as_cr_letters='abcdefghijklmnopqrstuvwxyz'
-as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
-as_cr_Letters=$as_cr_letters$as_cr_LETTERS
-as_cr_digits='0123456789'
-as_cr_alnum=$as_cr_Letters$as_cr_digits
-
 as_nl='
 '
 export as_nl
@@ -6245,7 +5374,13 @@ export as_nl
 as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
-if (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+# Prefer a ksh shell builtin over an external printf program on Solaris,
+# but without wasting forks for bash or zsh.
+if test -z "$BASH_VERSION$ZSH_VERSION" \
+    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
+  as_echo='print -r --'
+  as_echo_n='print -rn --'
+elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
   as_echo='printf %s\n'
   as_echo_n='printf %s'
 else
@@ -6256,7 +5391,7 @@ else
     as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
     as_echo_n_body='eval
       arg=$1;
-      case $arg in
+      case $arg in #(
       *"$as_nl"*)
 	expr "X$arg" : "X\\(.*\\)$as_nl";
 	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
@@ -6279,13 +5414,6 @@ if test "${PATH_SEPARATOR+set}" != set; then
   }
 fi
 
-# Support unset when possible.
-if ( (MAIL=60; unset MAIL) || exit) >/dev/null 2>&1; then
-  as_unset=unset
-else
-  as_unset=false
-fi
-
 
 # IFS
 # We need space, tab and new line, in precisely that order.  Quoting is
@@ -6295,15 +5423,16 @@ fi
 IFS=" ""	$as_nl"
 
 # Find who we are.  Look in the path if we contain no directory separator.
-case $0 in
+as_myself=
+case $0 in #((
   *[\\/]* ) as_myself=$0 ;;
   *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
 for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
-done
+    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+  done
 IFS=$as_save_IFS
 
      ;;
@@ -6315,12 +5444,16 @@ if test "x$as_myself" = x; then
 fi
 if test ! -f "$as_myself"; then
   $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
-  { (exit 1); exit 1; }
+  exit 1
 fi
 
-# Work around bugs in pre-3.0 UWIN ksh.
-for as_var in ENV MAIL MAILPATH
-do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
+# Unset variables that we do not need and which cause bugs (e.g. in
+# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
+# suppresses any "Segmentation fault" message there.  '((' could
+# trigger a bug in pdksh 5.2.14.
+for as_var in BASH_ENV ENV MAIL MAILPATH
+do eval test x\${$as_var+set} = xset \
+  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
 done
 PS1='$ '
 PS2='> '
@@ -6332,7 +5465,89 @@ export LC_ALL
 LANGUAGE=C
 export LANGUAGE
 
-# Required to use basename.
+# CDPATH.
+(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
+
+
+# as_fn_error STATUS ERROR [LINENO LOG_FD]
+# ----------------------------------------
+# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
+# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
+# script with STATUS, using 1 if that was 0.
+as_fn_error ()
+{
+  as_status=$1; test $as_status -eq 0 && as_status=1
+  if test "$4"; then
+    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
+  fi
+  $as_echo "$as_me: error: $2" >&2
+  as_fn_exit $as_status
+} # as_fn_error
+
+
+# as_fn_set_status STATUS
+# -----------------------
+# Set $? to STATUS, without forking.
+as_fn_set_status ()
+{
+  return $1
+} # as_fn_set_status
+
+# as_fn_exit STATUS
+# -----------------
+# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
+as_fn_exit ()
+{
+  set +e
+  as_fn_set_status $1
+  exit $1
+} # as_fn_exit
+
+# as_fn_unset VAR
+# ---------------
+# Portably unset VAR.
+as_fn_unset ()
+{
+  { eval $1=; unset $1;}
+}
+as_unset=as_fn_unset
+# as_fn_append VAR VALUE
+# ----------------------
+# Append the text in VALUE to the end of the definition contained in VAR. Take
+# advantage of any shell optimizations that allow amortized linear growth over
+# repeated appends, instead of the typical quadratic growth present in naive
+# implementations.
+if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
+  eval 'as_fn_append ()
+  {
+    eval $1+=\$2
+  }'
+else
+  as_fn_append ()
+  {
+    eval $1=\$$1\$2
+  }
+fi # as_fn_append
+
+# as_fn_arith ARG...
+# ------------------
+# Perform arithmetic evaluation on the ARGs, and store the result in the
+# global $as_val. Take advantage of shells that can avoid forks. The arguments
+# must be portable across $(()) and expr.
+if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
+  eval 'as_fn_arith ()
+  {
+    as_val=$(( $* ))
+  }'
+else
+  as_fn_arith ()
+  {
+    as_val=`expr "$@" || test $? -eq 1`
+  }
+fi # as_fn_arith
+
+
 if expr a : '\(a\)' >/dev/null 2>&1 &&
    test "X`expr 00001 : '.*\(...\)'`" = X001; then
   as_expr=expr
@@ -6346,8 +5561,12 @@ else
   as_basename=false
 fi
 
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
 
-# Name of the executable.
 as_me=`$as_basename -- "$0" ||
 $as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
 	 X"$0" : 'X\(//\)$' \| \
@@ -6367,76 +5586,25 @@ $as_echo X/"$0" |
 	  }
 	  s/.*/./; q'`
 
-# CDPATH.
-$as_unset CDPATH
-
-
-
-  as_lineno_1=$LINENO
-  as_lineno_2=$LINENO
-  test "x$as_lineno_1" != "x$as_lineno_2" &&
-  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2" || {
-
-  # Create $as_me.lineno as a copy of $as_myself, but with $LINENO
-  # uniformly replaced by the line number.  The first 'sed' inserts a
-  # line-number line after each line using $LINENO; the second 'sed'
-  # does the real work.  The second script uses 'N' to pair each
-  # line-number line with the line containing $LINENO, and appends
-  # trailing '-' during substitution so that $LINENO is not a special
-  # case at line end.
-  # (Raja R Harinath suggested sed '=', and Paul Eggert wrote the
-  # scripts with optimization help from Paolo Bonzini.  Blame Lee
-  # E. McMahon (1931-1989) for sed's syntax.  :-)
-  sed -n '
-    p
-    /[$]LINENO/=
-  ' <$as_myself |
-    sed '
-      s/[$]LINENO.*/&-/
-      t lineno
-      b
-      :lineno
-      N
-      :loop
-      s/[$]LINENO\([^'$as_cr_alnum'_].*\n\)\(.*\)/\2\1\2/
-      t loop
-      s/-\n.*//
-    ' >$as_me.lineno &&
-  chmod +x "$as_me.lineno" ||
-    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2
-   { (exit 1); exit 1; }; }
-
-  # Don't try to exec as it changes $[0], causing all sort of problems
-  # (the dirname of $[0] is not the place where we might find the
-  # original and so on.  Autoconf is especially sensitive to this).
-  . "./$as_me.lineno"
-  # Exit status is that of the last command.
-  exit
-}
-
-
-if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
-  as_dirname=dirname
-else
-  as_dirname=false
-fi
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
 
 ECHO_C= ECHO_N= ECHO_T=
-case `echo -n x` in
+case `echo -n x` in #(((((
 -n*)
-  case `echo 'x\c'` in
+  case `echo 'xy\c'` in
   *c*) ECHO_T='	';;	# ECHO_T is single tab character.
-  *)   ECHO_C='\c';;
+  xy)  ECHO_C='\c';;
+  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
+       ECHO_T='	';;
   esac;;
 *)
   ECHO_N='-n';;
 esac
-if expr a : '\(a\)' >/dev/null 2>&1 &&
-   test "X`expr 00001 : '.*\(...\)'`" = X001; then
-  as_expr=expr
-else
-  as_expr=false
-fi
 
 rm -f conf$$ conf$$.exe conf$$.file
 if test -d conf$$.dir; then
@@ -6451,49 +5619,85 @@ if (echo >conf$$.file) 2>/dev/null; then
     # ... but there are two gotchas:
     # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
     # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
-    # In both cases, we have to default to `cp -p'.
+    # In both cases, we have to default to `cp -pR'.
     ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
-      as_ln_s='cp -p'
+      as_ln_s='cp -pR'
   elif ln conf$$.file conf$$ 2>/dev/null; then
     as_ln_s=ln
   else
-    as_ln_s='cp -p'
+    as_ln_s='cp -pR'
   fi
 else
-  as_ln_s='cp -p'
+  as_ln_s='cp -pR'
 fi
 rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
 rmdir conf$$.dir 2>/dev/null
 
+
+# as_fn_mkdir_p
+# -------------
+# Create "$as_dir" as a directory, including parents if necessary.
+as_fn_mkdir_p ()
+{
+
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || eval $as_mkdir_p || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
+
+
+} # as_fn_mkdir_p
 if mkdir -p . 2>/dev/null; then
-  as_mkdir_p=:
+  as_mkdir_p='mkdir -p "$as_dir"'
 else
   test -d ./-p && rmdir ./-p
   as_mkdir_p=false
 fi
 
-if test -x / >/dev/null 2>&1; then
-  as_test_x='test -x'
-else
-  if ls -dL / >/dev/null 2>&1; then
-    as_ls_L_option=L
-  else
-    as_ls_L_option=
-  fi
-  as_test_x='
-    eval sh -c '\''
-      if test -d "$1"; then
-	test -d "$1/.";
-      else
-	case $1 in
-	-*)set "./$1";;
-	esac;
-	case `ls -ld'$as_ls_L_option' "$1" 2>/dev/null` in
-	???[sx]*):;;*)false;;esac;fi
-    '\'' sh
-  '
-fi
-as_executable_p=$as_test_x
+
+# as_fn_executable_p FILE
+# -----------------------
+# Test if FILE is an executable regular file.
+as_fn_executable_p ()
+{
+  test -f "$1" && test -x "$1"
+} # as_fn_executable_p
+as_test_x='test -x'
+as_executable_p=as_fn_executable_p
 
 # Sed expression to map a string onto a valid CPP name.
 as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
@@ -6503,13 +5707,19 @@ as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
 
 
 exec 6>&1
+## ----------------------------------- ##
+## Main body of $CONFIG_STATUS script. ##
+## ----------------------------------- ##
+_ASEOF
+test $as_write_fail = 0 && chmod +x $CONFIG_STATUS || ac_write_fail=1
 
-# Save the log message, to keep $[0] and so on meaningful, and to
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+# Save the log message, to keep $0 and so on meaningful, and to
 # report actual input values of CONFIG_FILES etc. instead of their
 # values after options handling.
 ac_log="
 This file was extended by Grid $as_me 1.0, which was
-generated by GNU Autoconf 2.63.  Invocation command line was
+generated by GNU Autoconf 2.69.  Invocation command line was
 
   CONFIG_FILES    = $CONFIG_FILES
   CONFIG_HEADERS  = $CONFIG_HEADERS
@@ -6541,13 +5751,15 @@ _ACEOF
 
 cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 ac_cs_usage="\
-\`$as_me' instantiates files from templates according to the
-current configuration.
+\`$as_me' instantiates files and other configuration actions
+from templates according to the current configuration.  Unless the files
+and actions are specified as TAGs, all are instantiated by default.
 
-Usage: $0 [OPTION]... [FILE]...
+Usage: $0 [OPTION]... [TAG]...
 
   -h, --help       print this help, then exit
   -V, --version    print version number and configuration settings, then exit
+      --config     print configuration, then exit
   -q, --quiet, --silent
                    do not print progress messages
   -d, --debug      don't remove temporary files
@@ -6566,16 +5778,17 @@ $config_headers
 Configuration commands:
 $config_commands
 
-Report bugs to <bug-autoconf@gnu.org>."
+Report bugs to <paboyle@ph.ed.ac.uk>."
 
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+ac_cs_config="`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`"
 ac_cs_version="\\
 Grid config.status 1.0
-configured by $0, generated by GNU Autoconf 2.63,
-  with options \\"`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`\\"
+configured by $0, generated by GNU Autoconf 2.69,
+  with options \\"\$ac_cs_config\\"
 
-Copyright (C) 2008 Free Software Foundation, Inc.
+Copyright (C) 2012 Free Software Foundation, Inc.
 This config.status script is free software; the Free Software Foundation
 gives unlimited permission to copy, distribute and modify it."
 
@@ -6593,11 +5806,16 @@ ac_need_defaults=:
 while test $# != 0
 do
   case $1 in
-  --*=*)
+  --*=?*)
     ac_option=`expr "X$1" : 'X\([^=]*\)='`
     ac_optarg=`expr "X$1" : 'X[^=]*=\(.*\)'`
     ac_shift=:
     ;;
+  --*=)
+    ac_option=`expr "X$1" : 'X\([^=]*\)='`
+    ac_optarg=
+    ac_shift=:
+    ;;
   *)
     ac_option=$1
     ac_optarg=$2
@@ -6611,27 +5829,29 @@ do
     ac_cs_recheck=: ;;
   --version | --versio | --versi | --vers | --ver | --ve | --v | -V )
     $as_echo "$ac_cs_version"; exit ;;
+  --config | --confi | --conf | --con | --co | --c )
+    $as_echo "$ac_cs_config"; exit ;;
   --debug | --debu | --deb | --de | --d | -d )
     debug=: ;;
   --file | --fil | --fi | --f )
     $ac_shift
     case $ac_optarg in
     *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
+    '') as_fn_error $? "missing file argument" ;;
     esac
-    CONFIG_FILES="$CONFIG_FILES '$ac_optarg'"
+    as_fn_append CONFIG_FILES " '$ac_optarg'"
     ac_need_defaults=false;;
   --header | --heade | --head | --hea )
     $ac_shift
     case $ac_optarg in
     *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
     esac
-    CONFIG_HEADERS="$CONFIG_HEADERS '$ac_optarg'"
+    as_fn_append CONFIG_HEADERS " '$ac_optarg'"
     ac_need_defaults=false;;
   --he | --h)
     # Conflict between --help and --header
-    { $as_echo "$as_me: error: ambiguous option: $1
-Try \`$0 --help' for more information." >&2
-   { (exit 1); exit 1; }; };;
+    as_fn_error $? "ambiguous option: \`$1'
+Try \`$0 --help' for more information.";;
   --help | --hel | -h )
     $as_echo "$ac_cs_usage"; exit ;;
   -q | -quiet | --quiet | --quie | --qui | --qu | --q \
@@ -6639,11 +5859,10 @@ Try \`$0 --help' for more information." >&2
     ac_cs_silent=: ;;
 
   # This is an error.
-  -*) { $as_echo "$as_me: error: unrecognized option: $1
-Try \`$0 --help' for more information." >&2
-   { (exit 1); exit 1; }; } ;;
+  -*) as_fn_error $? "unrecognized option: \`$1'
+Try \`$0 --help' for more information." ;;
 
-  *) ac_config_targets="$ac_config_targets $1"
+  *) as_fn_append ac_config_targets " $1"
      ac_need_defaults=false ;;
 
   esac
@@ -6660,7 +5879,7 @@ fi
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 if \$ac_cs_recheck; then
-  set X '$SHELL' '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
+  set X $SHELL '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
   shift
   \$as_echo "running CONFIG_SHELL=$SHELL \$*" >&6
   CONFIG_SHELL='$SHELL'
@@ -6701,9 +5920,7 @@ do
     "tests/Makefile") CONFIG_FILES="$CONFIG_FILES tests/Makefile" ;;
     "benchmarks/Makefile") CONFIG_FILES="$CONFIG_FILES benchmarks/Makefile" ;;
 
-  *) { { $as_echo "$as_me:$LINENO: error: invalid argument: $ac_config_target" >&5
-$as_echo "$as_me: error: invalid argument: $ac_config_target" >&2;}
-   { (exit 1); exit 1; }; };;
+  *) as_fn_error $? "invalid argument: \`$ac_config_target'" "$LINENO" 5;;
   esac
 done
 
@@ -6726,26 +5943,24 @@ fi
 # after its creation but before its name has been assigned to `$tmp'.
 $debug ||
 {
-  tmp=
+  tmp= ac_tmp=
   trap 'exit_status=$?
-  { test -z "$tmp" || test ! -d "$tmp" || rm -fr "$tmp"; } && exit $exit_status
+  : "${ac_tmp:=$tmp}"
+  { test ! -d "$ac_tmp" || rm -fr "$ac_tmp"; } && exit $exit_status
 ' 0
-  trap '{ (exit 1); exit 1; }' 1 2 13 15
+  trap 'as_fn_exit 1' 1 2 13 15
 }
 # Create a (secure) tmp directory for tmp files.
 
 {
   tmp=`(umask 077 && mktemp -d "./confXXXXXX") 2>/dev/null` &&
-  test -n "$tmp" && test -d "$tmp"
+  test -d "$tmp"
 }  ||
 {
   tmp=./conf$$-$RANDOM
   (umask 077 && mkdir "$tmp")
-} ||
-{
-   $as_echo "$as_me: cannot create a temporary directory in ." >&2
-   { (exit 1); exit 1; }
-}
+} || as_fn_error $? "cannot create a temporary directory in ." "$LINENO" 5
+ac_tmp=$tmp
 
 # Set up the scripts for CONFIG_FILES section.
 # No need to generate them if there are no CONFIG_FILES.
@@ -6753,7 +5968,13 @@ $debug ||
 if test -n "$CONFIG_FILES"; then
 
 
-ac_cr='
'
+ac_cr=`echo X | tr X '\015'`
+# On cygwin, bash can eat \r inside `` if the user requested igncr.
+# But we know of no other shell where ac_cr would be empty at this
+# point, so we can use a bashism as a fallback.
+if test "x$ac_cr" = x; then
+  eval ac_cr=\$\'\\r\'
+fi
 ac_cs_awk_cr=`$AWK 'BEGIN { print "a\rb" }' </dev/null 2>/dev/null`
 if test "$ac_cs_awk_cr" = "a${ac_cr}b"; then
   ac_cs_awk_cr='\\r'
@@ -6761,7 +5982,7 @@ else
   ac_cs_awk_cr=$ac_cr
 fi
 
-echo 'BEGIN {' >"$tmp/subs1.awk" &&
+echo 'BEGIN {' >"$ac_tmp/subs1.awk" &&
 _ACEOF
 
 
@@ -6770,24 +5991,18 @@ _ACEOF
   echo "$ac_subst_vars" | sed 's/.*/&!$&$ac_delim/' &&
   echo "_ACEOF"
 } >conf$$subs.sh ||
-  { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
-$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
-   { (exit 1); exit 1; }; }
-ac_delim_num=`echo "$ac_subst_vars" | grep -c '$'`
+  as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
+ac_delim_num=`echo "$ac_subst_vars" | grep -c '^'`
 ac_delim='%!_!# '
 for ac_last_try in false false false false false :; do
   . ./conf$$subs.sh ||
-    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
-$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
 
   ac_delim_n=`sed -n "s/.*$ac_delim\$/X/p" conf$$subs.awk | grep -c X`
   if test $ac_delim_n = $ac_delim_num; then
     break
   elif $ac_last_try; then
-    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
-$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
   else
     ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
   fi
@@ -6795,7 +6010,7 @@ done
 rm -f conf$$subs.sh
 
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-cat >>"\$tmp/subs1.awk" <<\\_ACAWK &&
+cat >>"\$ac_tmp/subs1.awk" <<\\_ACAWK &&
 _ACEOF
 sed -n '
 h
@@ -6809,7 +6024,7 @@ s/'"$ac_delim"'$//
 t delim
 :nl
 h
-s/\(.\{148\}\).*/\1/
+s/\(.\{148\}\)..*/\1/
 t more1
 s/["\\]/\\&/g; s/^/"/; s/$/\\n"\\/
 p
@@ -6823,7 +6038,7 @@ s/.\{148\}//
 t nl
 :delim
 h
-s/\(.\{148\}\).*/\1/
+s/\(.\{148\}\)..*/\1/
 t more2
 s/["\\]/\\&/g; s/^/"/; s/$/"/
 p
@@ -6843,7 +6058,7 @@ t delim
 rm -f conf$$subs.awk
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 _ACAWK
-cat >>"\$tmp/subs1.awk" <<_ACAWK &&
+cat >>"\$ac_tmp/subs1.awk" <<_ACAWK &&
   for (key in S) S_is_set[key] = 1
   FS = ""
 
@@ -6875,23 +6090,29 @@ if sed "s/$ac_cr//" < /dev/null > /dev/null 2>&1; then
   sed "s/$ac_cr\$//; s/$ac_cr/$ac_cs_awk_cr/g"
 else
   cat
-fi < "$tmp/subs1.awk" > "$tmp/subs.awk" \
-  || { { $as_echo "$as_me:$LINENO: error: could not setup config files machinery" >&5
-$as_echo "$as_me: error: could not setup config files machinery" >&2;}
-   { (exit 1); exit 1; }; }
+fi < "$ac_tmp/subs1.awk" > "$ac_tmp/subs.awk" \
+  || as_fn_error $? "could not setup config files machinery" "$LINENO" 5
 _ACEOF
 
-# VPATH may cause trouble with some makes, so we remove $(srcdir),
-# ${srcdir} and @srcdir@ from VPATH if srcdir is ".", strip leading and
+# VPATH may cause trouble with some makes, so we remove sole $(srcdir),
+# ${srcdir} and @srcdir@ entries from VPATH if srcdir is ".", strip leading and
 # trailing colons and then remove the whole line if VPATH becomes empty
 # (actually we leave an empty line to preserve line numbers).
 if test "x$srcdir" = x.; then
-  ac_vpsub='/^[	 ]*VPATH[	 ]*=/{
-s/:*\$(srcdir):*/:/
-s/:*\${srcdir}:*/:/
-s/:*@srcdir@:*/:/
-s/^\([^=]*=[	 ]*\):*/\1/
+  ac_vpsub='/^[	 ]*VPATH[	 ]*=[	 ]*/{
+h
+s///
+s/^/:/
+s/[	 ]*$/:/
+s/:\$(srcdir):/:/g
+s/:\${srcdir}:/:/g
+s/:@srcdir@:/:/g
+s/^:*//
 s/:*$//
+x
+s/\(=[	 ]*\).*/\1/
+G
+s/\n//
 s/^[^=]*=[	 ]*$//
 }'
 fi
@@ -6903,7 +6124,7 @@ fi # test -n "$CONFIG_FILES"
 # No need to generate them if there are no CONFIG_HEADERS.
 # This happens for instance with `./config.status Makefile'.
 if test -n "$CONFIG_HEADERS"; then
-cat >"$tmp/defines.awk" <<\_ACAWK ||
+cat >"$ac_tmp/defines.awk" <<\_ACAWK ||
 BEGIN {
 _ACEOF
 
@@ -6915,13 +6136,11 @@ _ACEOF
 # handling of long lines.
 ac_delim='%!_!# '
 for ac_last_try in false false :; do
-  ac_t=`sed -n "/$ac_delim/p" confdefs.h`
-  if test -z "$ac_t"; then
+  ac_tt=`sed -n "/$ac_delim/p" confdefs.h`
+  if test -z "$ac_tt"; then
     break
   elif $ac_last_try; then
-    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_HEADERS" >&5
-$as_echo "$as_me: error: could not make $CONFIG_HEADERS" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "could not make $CONFIG_HEADERS" "$LINENO" 5
   else
     ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
   fi
@@ -7006,9 +6225,7 @@ cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 _ACAWK
 _ACEOF
 cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-  { { $as_echo "$as_me:$LINENO: error: could not setup config headers machinery" >&5
-$as_echo "$as_me: error: could not setup config headers machinery" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "could not setup config headers machinery" "$LINENO" 5
 fi # test -n "$CONFIG_HEADERS"
 
 
@@ -7021,9 +6238,7 @@ do
   esac
   case $ac_mode$ac_tag in
   :[FHL]*:*);;
-  :L* | :C*:*) { { $as_echo "$as_me:$LINENO: error: invalid tag $ac_tag" >&5
-$as_echo "$as_me: error: invalid tag $ac_tag" >&2;}
-   { (exit 1); exit 1; }; };;
+  :L* | :C*:*) as_fn_error $? "invalid tag \`$ac_tag'" "$LINENO" 5;;
   :[FH]-) ac_tag=-:-;;
   :[FH]*) ac_tag=$ac_tag:$ac_tag.in;;
   esac
@@ -7042,7 +6257,7 @@ $as_echo "$as_me: error: invalid tag $ac_tag" >&2;}
     for ac_f
     do
       case $ac_f in
-      -) ac_f="$tmp/stdin";;
+      -) ac_f="$ac_tmp/stdin";;
       *) # Look for the file first in the build tree, then in the source tree
 	 # (if the path is not absolute).  The absolute path cannot be DOS-style,
 	 # because $ac_f cannot contain `:'.
@@ -7051,12 +6266,10 @@ $as_echo "$as_me: error: invalid tag $ac_tag" >&2;}
 	   [\\/$]*) false;;
 	   *) test -f "$srcdir/$ac_f" && ac_f="$srcdir/$ac_f";;
 	   esac ||
-	   { { $as_echo "$as_me:$LINENO: error: cannot find input file: $ac_f" >&5
-$as_echo "$as_me: error: cannot find input file: $ac_f" >&2;}
-   { (exit 1); exit 1; }; };;
+	   as_fn_error 1 "cannot find input file: \`$ac_f'" "$LINENO" 5;;
       esac
       case $ac_f in *\'*) ac_f=`$as_echo "$ac_f" | sed "s/'/'\\\\\\\\''/g"`;; esac
-      ac_file_inputs="$ac_file_inputs '$ac_f'"
+      as_fn_append ac_file_inputs " '$ac_f'"
     done
 
     # Let's still pretend it is `configure' which instantiates (i.e., don't
@@ -7067,7 +6280,7 @@ $as_echo "$as_me: error: cannot find input file: $ac_f" >&2;}
 	`' by configure.'
     if test x"$ac_file" != x-; then
       configure_input="$ac_file.  $configure_input"
-      { $as_echo "$as_me:$LINENO: creating $ac_file" >&5
+      { $as_echo "$as_me:${as_lineno-$LINENO}: creating $ac_file" >&5
 $as_echo "$as_me: creating $ac_file" >&6;}
     fi
     # Neutralize special characters interpreted by sed in replacement strings.
@@ -7079,10 +6292,8 @@ $as_echo "$as_me: creating $ac_file" >&6;}
     esac
 
     case $ac_tag in
-    *:-:* | *:-) cat >"$tmp/stdin" \
-      || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; } ;;
+    *:-:* | *:-) cat >"$ac_tmp/stdin" \
+      || as_fn_error $? "could not create $ac_file" "$LINENO" 5 ;;
     esac
     ;;
   esac
@@ -7110,47 +6321,7 @@ $as_echo X"$ac_file" |
 	    q
 	  }
 	  s/.*/./; q'`
-  { as_dir="$ac_dir"
-  case $as_dir in #(
-  -*) as_dir=./$as_dir;;
-  esac
-  test -d "$as_dir" || { $as_mkdir_p && mkdir -p "$as_dir"; } || {
-    as_dirs=
-    while :; do
-      case $as_dir in #(
-      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
-      *) as_qdir=$as_dir;;
-      esac
-      as_dirs="'$as_qdir' $as_dirs"
-      as_dir=`$as_dirname -- "$as_dir" ||
-$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_dir" : 'X\(//\)[^/]' \| \
-	 X"$as_dir" : 'X\(//\)$' \| \
-	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_dir" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      test -d "$as_dir" && break
-    done
-    test -z "$as_dirs" || eval "mkdir $as_dirs"
-  } || test -d "$as_dir" || { { $as_echo "$as_me:$LINENO: error: cannot create directory $as_dir" >&5
-$as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
-   { (exit 1); exit 1; }; }; }
+  as_dir="$ac_dir"; as_fn_mkdir_p
   ac_builddir=.
 
 case "$ac_dir" in
@@ -7207,7 +6378,6 @@ cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 # If the template does not know about datarootdir, expand it.
 # FIXME: This hack should be removed a few years after 2.60.
 ac_datarootdir_hack=; ac_datarootdir_seen=
-
 ac_sed_dataroot='
 /datarootdir/ {
   p
@@ -7217,12 +6387,11 @@ ac_sed_dataroot='
 /@docdir@/p
 /@infodir@/p
 /@localedir@/p
-/@mandir@/p
-'
+/@mandir@/p'
 case `eval "sed -n \"\$ac_sed_dataroot\" $ac_file_inputs"` in
 *datarootdir*) ac_datarootdir_seen=yes;;
 *@datadir@*|*@docdir@*|*@infodir@*|*@localedir@*|*@mandir@*)
-  { $as_echo "$as_me:$LINENO: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
 $as_echo "$as_me: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&2;}
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
@@ -7232,7 +6401,7 @@ cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
   s&@infodir@&$infodir&g
   s&@localedir@&$localedir&g
   s&@mandir@&$mandir&g
-    s&\\\${datarootdir}&$datarootdir&g' ;;
+  s&\\\${datarootdir}&$datarootdir&g' ;;
 esac
 _ACEOF
 
@@ -7260,27 +6429,24 @@ s&@INSTALL@&$ac_INSTALL&;t t
 s&@MKDIR_P@&$ac_MKDIR_P&;t t
 $ac_datarootdir_hack
 "
-eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$tmp/subs.awk" >$tmp/out \
-  || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; }
+eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$ac_tmp/subs.awk" \
+  >$ac_tmp/out || as_fn_error $? "could not create $ac_file" "$LINENO" 5
 
 test -z "$ac_datarootdir_hack$ac_datarootdir_seen" &&
-  { ac_out=`sed -n '/\${datarootdir}/p' "$tmp/out"`; test -n "$ac_out"; } &&
-  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' "$tmp/out"`; test -z "$ac_out"; } &&
-  { $as_echo "$as_me:$LINENO: WARNING: $ac_file contains a reference to the variable \`datarootdir'
-which seems to be undefined.  Please make sure it is defined." >&5
+  { ac_out=`sed -n '/\${datarootdir}/p' "$ac_tmp/out"`; test -n "$ac_out"; } &&
+  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' \
+      "$ac_tmp/out"`; test -z "$ac_out"; } &&
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file contains a reference to the variable \`datarootdir'
+which seems to be undefined.  Please make sure it is defined" >&5
 $as_echo "$as_me: WARNING: $ac_file contains a reference to the variable \`datarootdir'
-which seems to be undefined.  Please make sure it is defined." >&2;}
+which seems to be undefined.  Please make sure it is defined" >&2;}
 
-  rm -f "$tmp/stdin"
+  rm -f "$ac_tmp/stdin"
   case $ac_file in
-  -) cat "$tmp/out" && rm -f "$tmp/out";;
-  *) rm -f "$ac_file" && mv "$tmp/out" "$ac_file";;
+  -) cat "$ac_tmp/out" && rm -f "$ac_tmp/out";;
+  *) rm -f "$ac_file" && mv "$ac_tmp/out" "$ac_file";;
   esac \
-  || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; }
+  || as_fn_error $? "could not create $ac_file" "$LINENO" 5
  ;;
   :H)
   #
@@ -7289,27 +6455,21 @@ $as_echo "$as_me: error: could not create $ac_file" >&2;}
   if test x"$ac_file" != x-; then
     {
       $as_echo "/* $configure_input  */" \
-      && eval '$AWK -f "$tmp/defines.awk"' "$ac_file_inputs"
-    } >"$tmp/config.h" \
-      || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; }
-    if diff "$ac_file" "$tmp/config.h" >/dev/null 2>&1; then
-      { $as_echo "$as_me:$LINENO: $ac_file is unchanged" >&5
+      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs"
+    } >"$ac_tmp/config.h" \
+      || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+    if diff "$ac_file" "$ac_tmp/config.h" >/dev/null 2>&1; then
+      { $as_echo "$as_me:${as_lineno-$LINENO}: $ac_file is unchanged" >&5
 $as_echo "$as_me: $ac_file is unchanged" >&6;}
     else
       rm -f "$ac_file"
-      mv "$tmp/config.h" "$ac_file" \
-	|| { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; }
+      mv "$ac_tmp/config.h" "$ac_file" \
+	|| as_fn_error $? "could not create $ac_file" "$LINENO" 5
     fi
   else
     $as_echo "/* $configure_input  */" \
-      && eval '$AWK -f "$tmp/defines.awk"' "$ac_file_inputs" \
-      || { { $as_echo "$as_me:$LINENO: error: could not create -" >&5
-$as_echo "$as_me: error: could not create -" >&2;}
-   { (exit 1); exit 1; }; }
+      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs" \
+      || as_fn_error $? "could not create -" "$LINENO" 5
   fi
 # Compute "$ac_file"'s index in $config_headers.
 _am_arg="$ac_file"
@@ -7347,7 +6507,7 @@ $as_echo X"$_am_arg" |
 	  s/.*/./; q'`/stamp-h$_am_stamp_count
  ;;
 
-  :C)  { $as_echo "$as_me:$LINENO: executing $ac_file commands" >&5
+  :C)  { $as_echo "$as_me:${as_lineno-$LINENO}: executing $ac_file commands" >&5
 $as_echo "$as_me: executing $ac_file commands" >&6;}
  ;;
   esac
@@ -7355,7 +6515,7 @@ $as_echo "$as_me: executing $ac_file commands" >&6;}
 
   case $ac_file$ac_mode in
     "depfiles":C) test x"$AMDEP_TRUE" != x"" || {
-  # Autoconf 2.62 quotes --file arguments for eval, but not when files
+  # Older Autoconf quotes --file arguments for eval, but not when files
   # are listed without --file.  Let's play safe and only enable the eval
   # if we detect the quoting.
   case $CONFIG_FILES in
@@ -7368,7 +6528,7 @@ $as_echo "$as_me: executing $ac_file commands" >&6;}
     # Strip MF so we end up with the name of the file.
     mf=`echo "$mf" | sed -e 's/:.*$//'`
     # Check whether this is an Automake generated Makefile or not.
-    # We used to match only the files named `Makefile.in', but
+    # We used to match only the files named 'Makefile.in', but
     # some people rename them; so instead we look at the file content.
     # Grep'ing the first line is not enough: some people post-process
     # each Makefile.in and add a new line on top of each file to say so.
@@ -7402,21 +6562,19 @@ $as_echo X"$mf" |
       continue
     fi
     # Extract the definition of DEPDIR, am__include, and am__quote
-    # from the Makefile without running `make'.
+    # from the Makefile without running 'make'.
     DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
     test -z "$DEPDIR" && continue
     am__include=`sed -n 's/^am__include = //p' < "$mf"`
-    test -z "am__include" && continue
+    test -z "$am__include" && continue
     am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
-    # When using ansi2knr, U may be empty or an underscore; expand it
-    U=`sed -n 's/^U = //p' < "$mf"`
     # Find all dependency output files, they are included files with
     # $(DEPDIR) in their names.  We invoke sed twice because it is the
     # simplest approach to changing $(DEPDIR) to its actual value in the
     # expansion.
     for file in `sed -n "
       s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
-	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g' -e 's/\$U/'"$U"'/g'`; do
+	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
       # Make sure the directory exists.
       test -f "$dirpart/$file" && continue
       fdir=`$as_dirname -- "$file" ||
@@ -7442,47 +6600,7 @@ $as_echo X"$file" |
 	    q
 	  }
 	  s/.*/./; q'`
-      { as_dir=$dirpart/$fdir
-  case $as_dir in #(
-  -*) as_dir=./$as_dir;;
-  esac
-  test -d "$as_dir" || { $as_mkdir_p && mkdir -p "$as_dir"; } || {
-    as_dirs=
-    while :; do
-      case $as_dir in #(
-      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
-      *) as_qdir=$as_dir;;
-      esac
-      as_dirs="'$as_qdir' $as_dirs"
-      as_dir=`$as_dirname -- "$as_dir" ||
-$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_dir" : 'X\(//\)[^/]' \| \
-	 X"$as_dir" : 'X\(//\)$' \| \
-	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_dir" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      test -d "$as_dir" && break
-    done
-    test -z "$as_dirs" || eval "mkdir $as_dirs"
-  } || test -d "$as_dir" || { { $as_echo "$as_me:$LINENO: error: cannot create directory $as_dir" >&5
-$as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
-   { (exit 1); exit 1; }; }; }
+      as_dir=$dirpart/$fdir; as_fn_mkdir_p
       # echo "creating $dirpart/$file"
       echo '# dummy' > "$dirpart/$file"
     done
@@ -7494,15 +6612,12 @@ $as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
 done # for ac_tag
 
 
-{ (exit 0); exit 0; }
+as_fn_exit 0
 _ACEOF
-chmod +x $CONFIG_STATUS
 ac_clean_files=$ac_clean_files_save
 
 test $ac_write_fail = 0 ||
-  { { $as_echo "$as_me:$LINENO: error: write failure creating $CONFIG_STATUS" >&5
-$as_echo "$as_me: error: write failure creating $CONFIG_STATUS" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "write failure creating $CONFIG_STATUS" "$LINENO" 5
 
 
 # configure is writing to config.log, and then calls config.status.
@@ -7523,10 +6638,10 @@ if test "$no_create" != yes; then
   exec 5>>config.log
   # Use ||, not &&, to avoid exiting from the if with $? = 1, which
   # would make configure fail if this is the last instruction.
-  $ac_cs_success || { (exit 1); exit 1; }
+  $ac_cs_success || as_fn_exit 1
 fi
 if test -n "$ac_unrecognized_opts" && test "$enable_option_checking" != no; then
-  { $as_echo "$as_me:$LINENO: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
 $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2;}
 fi
 
diff --git a/lib/Grid.h b/lib/Grid.h
index 4b452051..cce56c0c 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -73,7 +73,11 @@ namespace Grid {
   std::string GridCmdOptionPayload(char ** begin, char ** end, const std::string & option);
   bool        GridCmdOptionExists(char** begin, char** end, const std::string& option);
   void        GridParseIntVector(std::string &str,std::vector<int> & vec);
-  void        GridParseLayout(char **argv,int argc,std::vector<int> &mpi,std::vector<int> &simd,std::vector<int> &latt);
+
+  void GridParseLayout(char **argv,int argc,
+		       std::vector<int> &simd,
+		       std::vector<int> &latt,
+		       std::vector<int> &mpi);
 
 
 };
diff --git a/lib/Grid_config.h b/lib/Grid_config.h
index 8b51adce..c743bba0 100644
--- a/lib/Grid_config.h
+++ b/lib/Grid_config.h
@@ -2,42 +2,42 @@
 /* lib/Grid_config.h.in.  Generated from configure.ac by autoheader.  */
 
 /* AVX */
-/* #undef AVX1 */
+#define AVX1 1
 
 /* AVX2 */
 /* #undef AVX2 */
 
 /* AVX512 */
-#define AVX512 1
+/* #undef AVX512 */
 
 /* GRID_COMMS_MPI */
-/* #undef GRID_COMMS_MPI */
+#define GRID_COMMS_MPI 1
 
 /* GRID_COMMS_NONE */
-#define GRID_COMMS_NONE 1
+/* #undef GRID_COMMS_NONE */
 
 /* Define to 1 if you have the declaration of `be64toh', and to 0 if you
    don't. */
-#define HAVE_DECL_BE64TOH 1
+#define HAVE_DECL_BE64TOH 0
 
 /* Define to 1 if you have the declaration of `ntohll', and to 0 if you don't.
    */
-#define HAVE_DECL_NTOHLL 0
+#define HAVE_DECL_NTOHLL 1
 
 /* Define to 1 if you have the <endian.h> header file. */
-#define HAVE_ENDIAN_H 1
+/* #undef HAVE_ENDIAN_H */
 
 /* Define to 1 if you have the `gettimeofday' function. */
-/* #undef HAVE_GETTIMEOFDAY */
+#define HAVE_GETTIMEOFDAY 1
 
 /* Define to 1 if you have the <inttypes.h> header file. */
 #define HAVE_INTTYPES_H 1
 
 /* Define to 1 if you have the <malloc.h> header file. */
-#define HAVE_MALLOC_H 1
+/* #undef HAVE_MALLOC_H */
 
 /* Define to 1 if you have the <malloc/malloc.h> header file. */
-/* #undef HAVE_MALLOC_MALLOC_H */
+#define HAVE_MALLOC_MALLOC_H 1
 
 /* Define to 1 if you have the <memory.h> header file. */
 #define HAVE_MEMORY_H 1
@@ -78,6 +78,9 @@
 /* Define to the one symbol short name of this package. */
 #define PACKAGE_TARNAME "grid"
 
+/* Define to the home page for this package. */
+#define PACKAGE_URL ""
+
 /* Define to the version of this package. */
 #define PACKAGE_VERSION "1.0"
 
diff --git a/lib/Grid_config.h.in b/lib/Grid_config.h.in
index 4b094250..437a6095 100644
--- a/lib/Grid_config.h.in
+++ b/lib/Grid_config.h.in
@@ -77,6 +77,9 @@
 /* Define to the one symbol short name of this package. */
 #undef PACKAGE_TARNAME
 
+/* Define to the home page for this package. */
+#undef PACKAGE_URL
+
 /* Define to the version of this package. */
 #undef PACKAGE_VERSION
 
diff --git a/lib/Grid_init.cc b/lib/Grid_init.cc
index 2382dc15..e80bbe64 100644
--- a/lib/Grid_init.cc
+++ b/lib/Grid_init.cc
@@ -76,54 +76,57 @@ void GridCmdOptionIntVector(std::string &str,std::vector<int> & vec)
   return;
 }
 
-void GridParseLayout(char **argv,int argc,std::vector<int> &mpi,std::vector<int> &simd,std::vector<int> &latt)
+void GridParseLayout(char **argv,int argc,
+		     std::vector<int> &latt,
+		     std::vector<int> &simd,
+		     std::vector<int> &mpi)
 {
   mpi =std::vector<int>({1,1,1,1});
-#if defined(AVX) || defined (AVX2)
-  simd=std::vector<int>({1,1,2,2});
-#endif
+  latt=std::vector<int>({8,8,8,8});
+
 #if defined(SSE4)
   simd=std::vector<int>({1,1,1,2});
 #endif
+#if defined(AVX1) || defined (AVX2)
+  simd=std::vector<int>({1,1,2,2});
+#endif
 #if defined(AVX512)
   simd=std::vector<int>({1,2,2,2});
 #endif
-  latt=std::vector<int>({8,8,8,8});
+
   
   std::string arg;
-  if( GridCmdOptionExists(argv,argv+argc,"--mpi-layout") ){
-    arg = GridCmdOptionPayload(argv,argv+argc,"--mpi-layout");
+  if( GridCmdOptionExists(argv,argv+argc,"--mpi") ){
+    arg = GridCmdOptionPayload(argv,argv+argc,"--mpi");
     GridCmdOptionIntVector(arg,mpi);
+    std::cout<<"MPI  ";
+    for(int i=0;i<mpi.size();i++){
+      std::cout<<mpi[i]<<" ";
+    }
+    std::cout<<std::endl;
   }
-  if( GridCmdOptionExists(argv,argv+argc,"--simd-layout") ){
-    arg= GridCmdOptionPayload(argv,argv+argc,"--simd-layout");
+  if( GridCmdOptionExists(argv,argv+argc,"--simd") ){
+    arg= GridCmdOptionPayload(argv,argv+argc,"--simd");
     GridCmdOptionIntVector(arg,simd);
+    std::cout<<"SIMD ";
+    for(int i=0;i<simd.size();i++){
+      std::cout<<simd[i]<<" ";
+    }
+    std::cout<<std::endl;
   }
-  if( GridCmdOptionExists(argv,argv+argc,"--lattice") ){
-    arg= GridCmdOptionPayload(argv,argv+argc,"--lattice");
+  if( GridCmdOptionExists(argv,argv+argc,"--grid") ){
+    arg= GridCmdOptionPayload(argv,argv+argc,"--grid");
     GridCmdOptionIntVector(arg,latt);
+    std::cout<<"Grid ";
+    for(int i=0;i<latt.size();i++){
+      std::cout<<latt[i]<<" ";
+    }
+    std::cout<<std::endl;
   }
-  std::cout<<"MPI layout";
-  for(int i=0;i<mpi.size();i++){
-    std::cout<<mpi[i]<<" ";
-  }
-  std::cout<<std::endl;
   
-  std::cout<<"SIMD layout";
-  for(int i=0;i<simd.size();i++){
-    std::cout<<simd[i]<<" ";
-  }
-  std::cout<<std::endl;
-  
-  std::cout<<"Grid ";
-  for(int i=0;i<latt.size();i++){
-    std::cout<<latt[i]<<" ";
-  }
-  std::cout<<std::endl;
   
 }
-
-
+  
 void Grid_finalize(void)
 {
 #ifdef GRID_COMMS_MPI
diff --git a/lib/math/Grid_math_tensors.h b/lib/math/Grid_math_tensors.h
index b71b0047..0c1d67d1 100644
--- a/lib/math/Grid_math_tensors.h
+++ b/lib/math/Grid_math_tensors.h
@@ -18,7 +18,7 @@ namespace Grid {
 //
 class GridTensorBase {};
 
-template<class vtype> class iScalar :public GridTensorBase
+template<class vtype> class iScalar 
 {
 public:
   vtype _internal;
@@ -105,7 +105,7 @@ template<class vtype> inline auto TensorRemove(iScalar<vtype> arg) -> decltype(T
   return TensorRemove(arg._internal);
 }
     
-template<class vtype,int N> class iVector :public GridTensorBase
+template<class vtype,int N> class iVector 
 {
 public:
   vtype _internal[N];
@@ -175,7 +175,7 @@ public:
     //    }
 };
     
-template<class vtype,int N> class iMatrix :public GridTensorBase
+template<class vtype,int N> class iMatrix 
 {
 public:
   vtype _internal[N][N];
diff --git a/tests/Grid_cshift.cc b/tests/Grid_cshift.cc
index 619808d6..0eb6285a 100644
--- a/tests/Grid_cshift.cc
+++ b/tests/Grid_cshift.cc
@@ -11,11 +11,11 @@ int main (int argc, char ** argv)
   std::vector<int> simd_layout;
   std::vector<int> mpi_layout;
   std::vector<int> latt_size;
-  GridParseLayout(argv,argc,mpi_layout,simd_layout,latt_size);
-    
-  GridCartesian     Fine(latt_size,simd_layout,mpi_layout);
-  GridParallelRNG           FineRNG(&Fine);
-  FineRNG.SeedRandomDevice();
+
+  GridParseLayout(argv,argc,latt_size,simd_layout,mpi_layout);
+  GridCartesian        Fine(latt_size,simd_layout,mpi_layout);
+
+  GridParallelRNG      FineRNG(&Fine);  FineRNG.SeedRandomDevice();
 
   LatticeComplex U(&Fine);
   LatticeComplex ShiftU(&Fine);
diff --git a/tests/Grid_gamma.cc b/tests/Grid_gamma.cc
index 2fd1d864..f8f86242 100644
--- a/tests/Grid_gamma.cc
+++ b/tests/Grid_gamma.cc
@@ -18,7 +18,7 @@ int main (int argc, char ** argv)
   std::vector<int> simd_layout;
   std::vector<int> mpi_layout;
 
-  GridParseLayout(argv,argc,mpi_layout,simd_layout,latt_size);
+  GridParseLayout(argv,argc,latt_size,simd_layout,mpi_layout);
     
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
 
diff --git a/tests/Grid_main.cc b/tests/Grid_main.cc
index 179694c6..9a7aa70b 100644
--- a/tests/Grid_main.cc
+++ b/tests/Grid_main.cc
@@ -29,7 +29,7 @@ int main (int argc, char ** argv)
   std::vector<int> simd_layout;
   std::vector<int> mpi_layout;
 
-  GridParseLayout(argv,argc,mpi_layout,simd_layout,latt_size);
+  GridParseLayout(argv,argc,latt_size,simd_layout,mpi_layout);
   latt_size.resize(4);
 
 #ifdef AVX512
diff --git a/tests/Grid_nersc_io.cc b/tests/Grid_nersc_io.cc
index 71cedd42..e8ff7209 100644
--- a/tests/Grid_nersc_io.cc
+++ b/tests/Grid_nersc_io.cc
@@ -14,7 +14,7 @@ int main (int argc, char ** argv)
   std::vector<int> simd_layout;
   std::vector<int> mpi_layout;
 
-  GridParseLayout(argv,argc,mpi_layout,simd_layout,tmp_latt_size);
+  GridParseLayout(argv,argc,tmp_latt_size,simd_layout,mpi_layout);
 
   std::vector<int> latt_size  ({16,16,16,32});
   std::vector<int> clatt_size  ({4,4,4,8});
diff --git a/tests/Grid_simd.cc b/tests/Grid_simd.cc
index fe90af2e..357e0fe8 100644
--- a/tests/Grid_simd.cc
+++ b/tests/Grid_simd.cc
@@ -110,7 +110,7 @@ int main (int argc, char ** argv)
   std::vector<int> simd_layout;
   std::vector<int> mpi_layout;
 
-  GridParseLayout(argv,argc,mpi_layout,simd_layout,latt_size);
+  GridParseLayout(argv,argc,latt_size,simd_layout,mpi_layout);
     
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
   std::vector<int> seeds({1,2,3,4});
diff --git a/tests/Grid_stencil.cc b/tests/Grid_stencil.cc
index 16acf294..c4b516dd 100644
--- a/tests/Grid_stencil.cc
+++ b/tests/Grid_stencil.cc
@@ -12,7 +12,7 @@ int main (int argc, char ** argv)
   std::vector<int> simd_layout;
   std::vector<int> mpi_layout;
 
-  GridParseLayout(argv,argc,mpi_layout,simd_layout,latt_size);
+  GridParseLayout(argv,argc,latt_size,simd_layout,mpi_layout);
 
   double volume = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
     

From c8dc8ff89171ce86fc5cd5fe8be151ad02622722 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 11 May 2015 18:59:03 +0100
Subject: [PATCH 144/429] Adding a better controlled threading class, preparing
 to force in deterministic reduction.

---
 benchmarks/Grid_comms.cc                 |   7 +-
 benchmarks/Grid_memory_bandwidth.cc      |   9 +-
 benchmarks/Grid_wilson.cc                |   9 +-
 lib/Grid.h                               |  13 ++-
 lib/Grid_init.cc                         | 125 ++++++++++++++---------
 lib/Grid_threads.h                       |  80 +++++++++++++++
 lib/cartesian/Grid_cartesian_base.h      |   5 +-
 lib/cartesian/Grid_cartesian_red_black.h |   2 +-
 tests/Grid_cshift.cc                     |   7 +-
 tests/Grid_gamma.cc                      |  10 +-
 tests/Grid_main.cc                       |   7 +-
 tests/Grid_nersc_io.cc                   |   7 +-
 tests/Grid_simd.cc                       |   8 +-
 tests/Grid_stencil.cc                    |   7 +-
 14 files changed, 199 insertions(+), 97 deletions(-)
 create mode 100644 lib/Grid_threads.h

diff --git a/benchmarks/Grid_comms.cc b/benchmarks/Grid_comms.cc
index 5317e127..3dedfb5b 100644
--- a/benchmarks/Grid_comms.cc
+++ b/benchmarks/Grid_comms.cc
@@ -8,11 +8,8 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> latt_size;
-  std::vector<int> simd_layout;
-  std::vector<int> mpi_layout;
-
-  GridParseLayout(argv,argc,latt_size,simd_layout,mpi_layout);
+  std::vector<int> simd_layout = GridDefaultSimd();
+  std::vector<int> mpi_layout  = GridDefaultMpi();
 
   int Nloop=10;
   int nmu=0;
diff --git a/benchmarks/Grid_memory_bandwidth.cc b/benchmarks/Grid_memory_bandwidth.cc
index dd6d1816..cb444128 100644
--- a/benchmarks/Grid_memory_bandwidth.cc
+++ b/benchmarks/Grid_memory_bandwidth.cc
@@ -8,17 +8,14 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> tmp_latt_size;
-  std::vector<int> simd_layout;
-  std::vector<int> mpi_layout;
-
-  GridParseLayout(argv,argc,tmp_latt_size,simd_layout,mpi_layout);
-
   const int Nvec=8;
   typedef Lattice< iVector< vReal,Nvec> > LatticeVec;
 
   int Nloop=1000;
 
+  std::vector<int> simd_layout = GridDefaultSimd();
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+
   std::cout << "===================================================================================================="<<std::endl;
   std::cout << "= Benchmarking fused AXPY bandwidth"<<std::endl;
   std::cout << "===================================================================================================="<<std::endl;
diff --git a/benchmarks/Grid_wilson.cc b/benchmarks/Grid_wilson.cc
index ea3f597a..6c6e9626 100644
--- a/benchmarks/Grid_wilson.cc
+++ b/benchmarks/Grid_wilson.cc
@@ -20,13 +20,12 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> latt_size;
-  std::vector<int> simd_layout;
-  std::vector<int> mpi_layout;
 
-  GridParseLayout(argv,argc,latt_size,simd_layout,mpi_layout);
-    
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd();
+  std::vector<int> mpi_layout  = GridDefaultMpi();
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+
   std::vector<int> seeds({1,2,3,4});
 
   GridParallelRNG          pRNG(&Grid);
diff --git a/lib/Grid.h b/lib/Grid.h
index cce56c0c..da04438a 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -45,8 +45,11 @@
 
 #include <Grid_aligned_allocator.h>
 #include <Grid_simd.h>
-#include <Grid_math.h>
+#include <Grid_threads.h>
+
 #include <Grid_cartesian.h>
+
+#include <Grid_math.h>
 #include <Grid_lattice.h>
 #include <Grid_comparison.h>
 #include <Grid_cshift.h>
@@ -60,6 +63,7 @@ namespace Grid {
 
   void Grid_init(int *argc,char ***argv);
   void Grid_finalize(void);
+  // internal, controled with --handle
   void Grid_sa_signal_handler(int sig,siginfo_t *si,void * ptr);
   void Grid_debug_handler_init(void);
   void Grid_quiesce_nodes(void);
@@ -68,6 +72,11 @@ namespace Grid {
   // C++11 time facilities better?
   double usecond(void);
 
+  const std::vector<int> &GridDefaultSimd(void);
+  const std::vector<int> &GridDefaultLatt(void);
+  const std::vector<int> &GridDefaultMpi(void);
+  const int              &GridThreads(void)  ;
+  void                 GridSetThreads(int t) ;
 
   // Common parsing chores
   std::string GridCmdOptionPayload(char ** begin, char ** end, const std::string & option);
@@ -75,8 +84,8 @@ namespace Grid {
   void        GridParseIntVector(std::string &str,std::vector<int> & vec);
 
   void GridParseLayout(char **argv,int argc,
-		       std::vector<int> &simd,
 		       std::vector<int> &latt,
+		       std::vector<int> &simd,
 		       std::vector<int> &mpi);
 
 
diff --git a/lib/Grid_init.cc b/lib/Grid_init.cc
index e80bbe64..9e977064 100644
--- a/lib/Grid_init.cc
+++ b/lib/Grid_init.cc
@@ -1,5 +1,5 @@
 /****************************************************************************/
-/* PAB: Signal magic. Processor state dump is x86-64 specific               */
+/* pab: Signal magic. Processor state dump is x86-64 specific               */
 /****************************************************************************/
 
 #include <stdlib.h>
@@ -23,23 +23,25 @@
 
 namespace Grid {
 
-  void Grid_quiesce_nodes(void)
-  {
-#ifdef GRID_COMMS_MPI
-    int me;
-    MPI_Comm_rank(MPI_COMM_WORLD,&me);
-    if ( me ) { 
-      std::cout.setstate(std::ios::badbit);
-    }
-#endif
-  }
-  void Grid_unquiesce_nodes(void)
-  {
-#ifdef GRID_COMMS_MPI
-    std::cout.clear();
-#endif
-  }
+  //////////////////////////////////////////////////////
+  // Convenience functions to access stadard command line arg
+  // driven parallelism controls
+  //////////////////////////////////////////////////////
+  static std::vector<int> Grid_default_simd;
+  static std::vector<int> Grid_default_latt;
+  static std::vector<int> Grid_default_mpi;
 
+  int GridThread::_threads;
+
+
+  const std::vector<int> &GridDefaultSimd(void)     {return Grid_default_simd;};
+  const std::vector<int> &GridDefaultLatt(void)     {return Grid_default_latt;};
+  const std::vector<int> &GridDefaultMpi(void)      {return Grid_default_mpi;};
+
+
+  ////////////////////////////////////////////////////////////
+  // Command line parsing assist for stock controls
+  ////////////////////////////////////////////////////////////
 std::string GridCmdOptionPayload(char ** begin, char ** end, const std::string & option)
 {
   char ** itr = std::find(begin, end, option);
@@ -53,15 +55,6 @@ bool GridCmdOptionExists(char** begin, char** end, const std::string& option)
 {
   return std::find(begin, end, option) != end;
 }
-void Grid_init(int *argc,char ***argv)
-{
-#ifdef GRID_COMMS_MPI
-  MPI_Init(argc,argv);
-#endif
-  // Parse command line args.
-  Grid_quiesce_nodes();
-
-}
 
 void GridCmdOptionIntVector(std::string &str,std::vector<int> & vec)
 {
@@ -70,7 +63,7 @@ void GridCmdOptionIntVector(std::string &str,std::vector<int> & vec)
   int i;
   while (ss >> i){
     vec.push_back(i);
-    if (ss.peek() == ',')
+    if(std::ispunct(ss.peek()))
       ss.ignore();
   }    
   return;
@@ -94,38 +87,74 @@ void GridParseLayout(char **argv,int argc,
   simd=std::vector<int>({1,2,2,2});
 #endif
 
-  
+  GridThread::SetMaxThreads();
+
   std::string arg;
   if( GridCmdOptionExists(argv,argv+argc,"--mpi") ){
     arg = GridCmdOptionPayload(argv,argv+argc,"--mpi");
     GridCmdOptionIntVector(arg,mpi);
-    std::cout<<"MPI  ";
-    for(int i=0;i<mpi.size();i++){
-      std::cout<<mpi[i]<<" ";
-    }
-    std::cout<<std::endl;
   }
   if( GridCmdOptionExists(argv,argv+argc,"--simd") ){
     arg= GridCmdOptionPayload(argv,argv+argc,"--simd");
     GridCmdOptionIntVector(arg,simd);
-    std::cout<<"SIMD ";
-    for(int i=0;i<simd.size();i++){
-      std::cout<<simd[i]<<" ";
-    }
-    std::cout<<std::endl;
   }
   if( GridCmdOptionExists(argv,argv+argc,"--grid") ){
     arg= GridCmdOptionPayload(argv,argv+argc,"--grid");
     GridCmdOptionIntVector(arg,latt);
-    std::cout<<"Grid ";
-    for(int i=0;i<latt.size();i++){
-      std::cout<<latt[i]<<" ";
-    }
-    std::cout<<std::endl;
   }
-  
-  
+  if( GridCmdOptionExists(argv,argv+argc,"--omp") ){
+    std::vector<int> ompthreads(0);
+    arg= GridCmdOptionPayload(argv,argv+argc,"--omp");
+    GridCmdOptionIntVector(arg,ompthreads);
+    assert(ompthreads.size()==1);
+    GridThread::SetThreads(ompthreads[0]);
+  }
+
 }
+
+  /////////////////////////////////////////////////////////
+  /////////////////////////////////////////////////////////
+void Grid_init(int *argc,char ***argv)
+{
+#ifdef GRID_COMMS_MPI
+  MPI_Init(argc,argv);
+#endif
+  // Parse command line args.
+
+  if( GridCmdOptionExists(*argv,*argv+*argc,"--debug-signals") ){
+    Grid_debug_handler_init();
+  }
+  if( !GridCmdOptionExists(*argv,*argv+*argc,"--debug-stdout") ){
+    Grid_quiesce_nodes();
+  }
+  GridParseLayout(*argv,*argc,
+		  Grid_default_latt,
+		  Grid_default_simd,
+		  Grid_default_mpi);
+
+}
+
+
+  ////////////////////////////////////////////////////////////
+  // Verbose limiter on MPI tasks
+  ////////////////////////////////////////////////////////////
+  void Grid_quiesce_nodes(void)
+  {
+#ifdef GRID_COMMS_MPI
+    int me;
+    MPI_Comm_rank(MPI_COMM_WORLD,&me);
+    if ( me ) { 
+      std::cout.setstate(std::ios::badbit);
+    }
+#endif
+  }
+  void Grid_unquiesce_nodes(void)
+  {
+#ifdef GRID_COMMS_MPI
+    std::cout.clear();
+#endif
+  }
+
   
 void Grid_finalize(void)
 {
@@ -146,14 +175,14 @@ void * Grid_backtrace_buffer[_NBACKTRACE];
 void Grid_sa_signal_handler(int sig,siginfo_t *si,void * ptr)
 {
   printf("Caught signal %d\n",si->si_signo);
-  printf("  mem address %lx\n",(uint64_t)si->si_addr);
+  printf("  mem address %llx\n",(unsigned long long)si->si_addr);
   printf("         code %d\n",si->si_code);
 
 #ifdef __X86_64
     ucontext_t * uc= (ucontext_t *)ptr;
   struct sigcontext *sc = (struct sigcontext *)&uc->uc_mcontext;
-  printf("  instruction %llx\n",(uint64_t)sc->rip);
-#define REG(A)  printf("  %s %lx\n",#A, sc-> A);
+  printf("  instruction %llx\n",(unsigned long long)sc->rip);
+#define REG(A)  printf("  %s %lx\n",#A,sc-> A);
   REG(rdi);
   REG(rsi);
   REG(rbp);
diff --git a/lib/Grid_threads.h b/lib/Grid_threads.h
new file mode 100644
index 00000000..6c5f17e0
--- /dev/null
+++ b/lib/Grid_threads.h
@@ -0,0 +1,80 @@
+#ifndef GRID_THREADS_H
+#define GRID_THREADS_H
+
+#ifdef HAVE_OPENMP
+#include <omp.h>
+#define PARALLEL_FOR_LOOP _Pragma("omp parallel for")
+#define PARALLEL_NESTED_LOOP(n) _Pragma("omp parallel for collapse(" #n ")")
+#else
+#define PARALLEL_FOR_LOOP 
+#define PARALLEL_NESTED_LOOP(n) 
+#endif
+
+namespace Grid {
+
+  // Introduce a class to gain deterministic bit reproducible reduction.
+  // make static; perhaps just a namespace is required.
+
+class GridThread {
+ public:
+  static int _threads;
+
+  static void SetThreads(int thr) { 
+#ifdef HAVE_OPENMP
+    _threads = MIN(thr,omp_get_max_threads()) ;
+    omp_set_num_threads(_threads);
+#else 
+    _threads = 1;
+#endif
+  };
+  static void SetMaxThreads(void) { 
+#ifdef HAVE_OPENMP
+    _threads = omp_get_max_threads();
+    omp_set_num_threads(_threads);
+#else 
+    _threads = 1;
+#endif
+  };
+  static int GetThreads(void) { return _threads; };
+  static int SumArraySize(void) {return _threads;};
+
+  static void GetWork(int nwork, int me, int & mywork, int & myoff){
+    int basework = nwork/_threads;
+    int backfill = _threads-(nwork%_threads);
+    if ( me >= _threads ) { 
+      mywork = myoff = 0;
+    } else { 
+      mywork = (nwork+me)/_threads;
+      myoff  = basework * me;
+      if ( me > backfill ) 
+	myoff+= (me-backfill);
+    }
+    return;
+  };
+
+  static void GetWorkBarrier(int nwork, int &me, int & mywork, int & myoff){
+    me     = ThreadBarrier();
+    GetWork(nwork,me,mywork,myoff);
+  };
+
+  static int  ThreadBarrier(void) {
+#ifdef HAVE_OPENMP
+#pragma omp barrier
+    return omp_get_thread_num();
+#else
+    return 0;
+#endif
+  };
+  
+  template<class obj> static void ThreadSum( std::vector<obj> &sum_array,obj &val,int me){
+    sum_array[me] = val;
+    val=zero;
+    ThreadBarrier();
+    for(int i=0;i<_threads;i++) val+= sum_array[i];
+    ThreadBarrier();
+  };
+
+};
+
+}
+#endif
diff --git a/lib/cartesian/Grid_cartesian_base.h b/lib/cartesian/Grid_cartesian_base.h
index c297cbb9..a74773f8 100644
--- a/lib/cartesian/Grid_cartesian_base.h
+++ b/lib/cartesian/Grid_cartesian_base.h
@@ -14,7 +14,7 @@ namespace Grid{
   //    int              _processor;  // linear processor rank
   //    std::vector<int> _processor_coor;  // linear processor rank
   //////////////////////////////////////////////////////////////////////
-  class GridBase : public CartesianCommunicator {
+  class GridBase : public CartesianCommunicator , public GridThread {
 
 public:
 
@@ -22,7 +22,8 @@ public:
     template<class object> friend class Lattice;
 
     GridBase(std::vector<int> & processor_grid) : CartesianCommunicator(processor_grid) {};
-            
+
+
     // Physics Grid information.
     std::vector<int> _simd_layout;// Which dimensions get relayed out over simd lanes.
     std::vector<int> _fdimensions;// Global dimensions of array prior to cb removal
diff --git a/lib/cartesian/Grid_cartesian_red_black.h b/lib/cartesian/Grid_cartesian_red_black.h
index 55bd1f20..475b47c2 100644
--- a/lib/cartesian/Grid_cartesian_red_black.h
+++ b/lib/cartesian/Grid_cartesian_red_black.h
@@ -46,7 +46,7 @@ public:
     };
     GridRedBlackCartesian(std::vector<int> &dimensions,
 			  std::vector<int> &simd_layout,
-			  std::vector<int> &processor_grid) : GridBase(processor_grid)
+			  std::vector<int> &processor_grid ) : GridBase(processor_grid)
     {
     ///////////////////////
     // Grid information
diff --git a/tests/Grid_cshift.cc b/tests/Grid_cshift.cc
index 0eb6285a..e92ff793 100644
--- a/tests/Grid_cshift.cc
+++ b/tests/Grid_cshift.cc
@@ -8,11 +8,10 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> simd_layout;
-  std::vector<int> mpi_layout;
-  std::vector<int> latt_size;
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd();
+  std::vector<int> mpi_layout  = GridDefaultMpi();
 
-  GridParseLayout(argv,argc,latt_size,simd_layout,mpi_layout);
   GridCartesian        Fine(latt_size,simd_layout,mpi_layout);
 
   GridParallelRNG      FineRNG(&Fine);  FineRNG.SeedRandomDevice();
diff --git a/tests/Grid_gamma.cc b/tests/Grid_gamma.cc
index f8f86242..9779ca07 100644
--- a/tests/Grid_gamma.cc
+++ b/tests/Grid_gamma.cc
@@ -14,12 +14,10 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> latt_size;
-  std::vector<int> simd_layout;
-  std::vector<int> mpi_layout;
-
-  GridParseLayout(argv,argc,latt_size,simd_layout,mpi_layout);
-    
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd();
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+     
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
 
   GridParallelRNG          pRNG(&Grid);
diff --git a/tests/Grid_main.cc b/tests/Grid_main.cc
index 9a7aa70b..c3454dfc 100644
--- a/tests/Grid_main.cc
+++ b/tests/Grid_main.cc
@@ -25,11 +25,10 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> latt_size;
-  std::vector<int> simd_layout;
-  std::vector<int> mpi_layout;
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd();
+  std::vector<int> mpi_layout  = GridDefaultMpi();
 
-  GridParseLayout(argv,argc,latt_size,simd_layout,mpi_layout);
   latt_size.resize(4);
 
 #ifdef AVX512
diff --git a/tests/Grid_nersc_io.cc b/tests/Grid_nersc_io.cc
index e8ff7209..73d7edfe 100644
--- a/tests/Grid_nersc_io.cc
+++ b/tests/Grid_nersc_io.cc
@@ -10,12 +10,9 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> tmp_latt_size;
-  std::vector<int> simd_layout;
-  std::vector<int> mpi_layout;
-
-  GridParseLayout(argv,argc,tmp_latt_size,simd_layout,mpi_layout);
 
+  std::vector<int> simd_layout = GridDefaultSimd();
+  std::vector<int> mpi_layout  = GridDefaultMpi();
   std::vector<int> latt_size  ({16,16,16,32});
   std::vector<int> clatt_size  ({4,4,4,8});
   int orthodir=3;
diff --git a/tests/Grid_simd.cc b/tests/Grid_simd.cc
index 357e0fe8..6a957d7e 100644
--- a/tests/Grid_simd.cc
+++ b/tests/Grid_simd.cc
@@ -106,11 +106,9 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> latt_size;
-  std::vector<int> simd_layout;
-  std::vector<int> mpi_layout;
-
-  GridParseLayout(argv,argc,latt_size,simd_layout,mpi_layout);
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd();
+  std::vector<int> mpi_layout  = GridDefaultMpi();
     
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
   std::vector<int> seeds({1,2,3,4});
diff --git a/tests/Grid_stencil.cc b/tests/Grid_stencil.cc
index c4b516dd..d9e779bf 100644
--- a/tests/Grid_stencil.cc
+++ b/tests/Grid_stencil.cc
@@ -8,11 +8,10 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> latt_size;
-  std::vector<int> simd_layout;
-  std::vector<int> mpi_layout;
 
-  GridParseLayout(argv,argc,latt_size,simd_layout,mpi_layout);
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd();
+  std::vector<int> mpi_layout  = GridDefaultMpi();
 
   double volume = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
     

From 6e6843ac6910bbbfc9f3e3dc565304d99c116af5 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 11 May 2015 19:09:49 +0100
Subject: [PATCH 145/429] Moving some things around for pretty

---
 lib/Grid_communicator.h                   | 113 +-----------
 lib/Grid_lattice.h                        | 199 +--------------------
 lib/communicator/Grid_communicator_base.h | 117 +++++++++++++
 lib/lattice/Grid_lattice_base.h           | 203 ++++++++++++++++++++++
 lib/math/Grid_math_inner.h                |   2 +-
 tests/Grid_gamma.cc                       |  26 +--
 6 files changed, 336 insertions(+), 324 deletions(-)
 create mode 100644 lib/communicator/Grid_communicator_base.h
 create mode 100644 lib/lattice/Grid_lattice_base.h

diff --git a/lib/Grid_communicator.h b/lib/Grid_communicator.h
index 661f39b8..cfa6e0a7 100644
--- a/lib/Grid_communicator.h
+++ b/lib/Grid_communicator.h
@@ -1,117 +1,6 @@
 #ifndef GRID_COMMUNICATOR_H
 #define GRID_COMMUNICATOR_H
 
-///////////////////////////////////
-// Processor layout information
-///////////////////////////////////
-#ifdef GRID_COMMS_MPI
-#include <mpi.h>
-#endif
-namespace Grid {
-class CartesianCommunicator {
-  public:    
-
-  // Communicator should know nothing of the physics grid, only processor grid.
-
-    int              _Nprocessors;     // How many in all
-    std::vector<int> _processors;      // Which dimensions get relayed out over processors lanes.
-    int              _processor;       // linear processor rank
-    std::vector<int> _processor_coor;  // linear processor coordinate
-    unsigned long _ndimension;
-
-#ifdef GRID_COMMS_MPI
-    MPI_Comm communicator;
-    typedef MPI_Request CommsRequest_t;
-#else 
-    typedef int CommsRequest_t;
-#endif
-
-    // Constructor
-    CartesianCommunicator(std::vector<int> &pdimensions_in);
-
-    // Wraps MPI_Cart routines
-    void ShiftedRanks(int dim,int shift,int & source, int & dest);
-    int  RankFromProcessorCoor(std::vector<int> &coor);
-    void ProcessorCoorFromRank(int rank,std::vector<int> &coor);
-
-    /////////////////////////////////
-    // Grid information queries
-    /////////////////////////////////
-    int                      IsBoss(void)            { return _processor==0; };
-    int                      BossRank(void)          { return 0; };
-    int                      ThisRank(void)          { return _processor; };
-    const std::vector<int> & ThisProcessorCoor(void) { return _processor_coor; };
-    const std::vector<int> & ProcessorGrid(void)     { return _processors; };
-    int                      ProcessorCount(void)    { return _Nprocessors; };
-
-    ////////////////////////////////////////////////////////////
-    // Reduction
-    ////////////////////////////////////////////////////////////
-    void GlobalSum(RealF &);
-    void GlobalSumVector(RealF *,int N);
-
-    void GlobalSum(RealD &);
-    void GlobalSumVector(RealD *,int N);
-
-    void GlobalSum(uint32_t &);
-
-    void GlobalSum(ComplexF &c)
-    {
-      GlobalSumVector((float *)&c,2);
-    }
-    void GlobalSumVector(ComplexF *c,int N)
-    {
-      GlobalSumVector((float *)c,2*N);
-    }
-
-    void GlobalSum(ComplexD &c)
-    {
-      GlobalSumVector((double *)&c,2);
-    }
-    void GlobalSumVector(ComplexD *c,int N)
-    {
-      GlobalSumVector((double *)c,2*N);
-    }
-    
-    template<class obj> void GlobalSum(obj &o){
-      typedef typename obj::scalar_type scalar_type;
-      int words = sizeof(obj)/sizeof(scalar_type);
-      scalar_type * ptr = (scalar_type *)& o;
-      GlobalSumVector(ptr,words);
-    }
-    ////////////////////////////////////////////////////////////
-    // Face exchange, buffer swap in translational invariant way
-    ////////////////////////////////////////////////////////////
-    void SendToRecvFrom(void *xmit,
-			int xmit_to_rank,
-			void *recv,
-			int recv_from_rank,
-			int bytes);
-    void SendToRecvFromBegin(std::vector<CommsRequest_t> &list,
-			 void *xmit,
-			 int xmit_to_rank,
-			 void *recv,
-			 int recv_from_rank,
-			 int bytes);
-    void SendToRecvFromComplete(std::vector<CommsRequest_t> &waitall);
-
-    ////////////////////////////////////////////////////////////
-    // Barrier
-    ////////////////////////////////////////////////////////////
-    void Barrier(void);
-
-    ////////////////////////////////////////////////////////////
-    // Broadcast a buffer and composite larger
-    ////////////////////////////////////////////////////////////
-    void Broadcast(int root,void* data, int bytes);
-    template<class obj> void Broadcast(int root,obj &data)
-    {
-      Broadcast(root,(void *)&data,sizeof(data));
-    };
-
-    static void BroadcastWorld(int root,void* data, int bytes);
-
-}; 
-}
+#include <communicator/Grid_communicator_base.h>
 
 #endif
diff --git a/lib/Grid_lattice.h b/lib/Grid_lattice.h
index 356a25b8..35664aee 100644
--- a/lib/Grid_lattice.h
+++ b/lib/Grid_lattice.h
@@ -1,203 +1,6 @@
 #ifndef GRID_LATTICE_H
 #define GRID_LATTICE_H
 
-namespace Grid {
-
-// TODO: 
-//       mac,real,imag
-
-// Functionality:
-//     -=,+=,*=,()
-//     add,+,sub,-,mult,mac,*
-//     adj,conj
-//     real,imag
-//     transpose,transposeIndex  
-//     trace,traceIndex
-//     peekIndex
-//     innerProduct,outerProduct,
-//     localNorm2
-//     localInnerProduct
-
-extern int GridCshiftPermuteMap[4][16];
-
-////////////////////////////////////////////////
-// Basic expressions used in Expression Template
-////////////////////////////////////////////////
-
-class LatticeBase {};
-class LatticeExpressionBase {};
-
-template <typename Op, typename T1>                           
-class LatticeUnaryExpression  : public std::pair<Op,std::tuple<T1> > , public LatticeExpressionBase {
- public:
- LatticeUnaryExpression(const std::pair<Op,std::tuple<T1> > &arg): std::pair<Op,std::tuple<T1> >(arg) {};
-};
-
-template <typename Op, typename T1, typename T2>              
-class LatticeBinaryExpression : public std::pair<Op,std::tuple<T1,T2> > , public LatticeExpressionBase {
- public:
- LatticeBinaryExpression(const std::pair<Op,std::tuple<T1,T2> > &arg): std::pair<Op,std::tuple<T1,T2> >(arg) {};
-};
-
-template <typename Op, typename T1, typename T2, typename T3> 
-class LatticeTrinaryExpression :public std::pair<Op,std::tuple<T1,T2,T3> >, public LatticeExpressionBase {
- public:
- LatticeTrinaryExpression(const std::pair<Op,std::tuple<T1,T2,T3> > &arg): std::pair<Op,std::tuple<T1,T2,T3> >(arg) {};
-};
-
-template<class vobj>
-class Lattice : public LatticeBase
-{
-public:
-
-    GridBase *_grid;
-    int checkerboard;
-    std::vector<vobj,alignedAllocator<vobj> > _odata;
-
-public:
-    typedef typename vobj::scalar_type scalar_type;
-    typedef typename vobj::vector_type vector_type;
-    typedef vobj vector_object;
-    
-  ////////////////////////////////////////////////////////////////////////////////
-  // Expression Template closure support
-  ////////////////////////////////////////////////////////////////////////////////
-  template <typename Op, typename T1>                         inline Lattice<vobj> & operator=(const LatticeUnaryExpression<Op,T1> &expr)
-  {
-#pragma omp parallel for
-    for(int ss=0;ss<_grid->oSites();ss++){
-      vobj tmp= eval(ss,expr);
-      vstream(_odata[ss] ,tmp);
-    }
-    return *this;
-  }
-  template <typename Op, typename T1,typename T2>             inline Lattice<vobj> & operator=(const LatticeBinaryExpression<Op,T1,T2> &expr)
-  {
-#pragma omp parallel for
-    for(int ss=0;ss<_grid->oSites();ss++){
-      vobj tmp= eval(ss,expr);
-      vstream(_odata[ss] ,tmp);
-    }
-    return *this;
-  }
-  template <typename Op, typename T1,typename T2,typename T3> inline Lattice<vobj> & operator=(const LatticeTrinaryExpression<Op,T1,T2,T3> &expr)
-  {
-#pragma omp parallel for
-    for(int ss=0;ss<_grid->oSites();ss++){
-      vobj tmp= eval(ss,expr);
-      vstream(_odata[ss] ,tmp);
-    }
-    return *this;
-  }
-  //GridFromExpression is tricky to do
-  template<class Op,class T1>
-    Lattice(const LatticeUnaryExpression<Op,T1> & expr):    _grid(nullptr){
-    GridFromExpression(_grid,expr);
-    assert(_grid!=nullptr);
-    _odata.resize(_grid->oSites());
-#pragma omp parallel for
-    for(int ss=0;ss<_grid->oSites();ss++){
-      _odata[ss] = eval(ss,expr);
-    }
-  };
-  template<class Op,class T1, class T2>
-  Lattice(const LatticeBinaryExpression<Op,T1,T2> & expr):    _grid(nullptr){
-    GridFromExpression(_grid,expr);
-    assert(_grid!=nullptr);
-    _odata.resize(_grid->oSites());
-#pragma omp parallel for
-    for(int ss=0;ss<_grid->oSites();ss++){
-      _odata[ss] = eval(ss,expr);
-    }
-  };
-  template<class Op,class T1, class T2, class T3>
-  Lattice(const LatticeTrinaryExpression<Op,T1,T2,T3> & expr):    _grid(nullptr){
-    GridFromExpression(_grid,expr);
-    assert(_grid!=nullptr);
-    _odata.resize(_grid->oSites());
-#pragma omp parallel for
-    for(int ss=0;ss<_grid->oSites();ss++){
-      _odata[ss] = eval(ss,expr);
-    }
-  };
-
-    //////////////////////////////////////////////////////////////////
-    // Constructor requires "grid" passed.
-    // what about a default grid?
-    //////////////////////////////////////////////////////////////////
- Lattice(GridBase *grid) : _grid(grid), _odata(_grid->oSites()) {
-      //        _odata.reserve(_grid->oSites());
-      //        _odata.resize(_grid->oSites());
-        assert((((uint64_t)&_odata[0])&0xF) ==0);
-        checkerboard=0;
-    }
-
-    template<class sobj> inline Lattice<vobj> & operator = (const sobj & r){
-#pragma omp parallel for
-        for(int ss=0;ss<_grid->oSites();ss++){
-            this->_odata[ss]=r;
-        }
-        return *this;
-    }
-    template<class robj> inline Lattice<vobj> & operator = (const Lattice<robj> & r){
-      conformable(*this,r);
-      std::cout<<"Lattice operator ="<<std::endl;
-#pragma omp parallel for
-        for(int ss=0;ss<_grid->oSites();ss++){
-            this->_odata[ss]=r._odata[ss];
-        }
-        return *this;
-    }
-
-    // *=,+=,-= operators inherit behvour from correspond */+/- operation
-    template<class T> inline Lattice<vobj> &operator *=(const T &r) {
-        *this = (*this)*r;
-        return *this;
-    }
-
-    template<class T> inline Lattice<vobj> &operator -=(const T &r) {
-        *this = (*this)-r;
-        return *this;
-    }
-    template<class T> inline Lattice<vobj> &operator +=(const T &r) {
-        *this = (*this)+r;
-        return *this;
-    }
-    
-    inline friend Lattice<vobj> operator / (const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
-        conformable(lhs,rhs);
-        Lattice<vobj> ret(lhs._grid);
-#pragma omp parallel for
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-            ret._odata[ss] = lhs._odata[ss]/rhs._odata[ss];
-        }
-        return ret;
-    };
- }; // class Lattice
-}
-
-#undef GRID_LATTICE_EXPRESSION_TEMPLATES
-
-#include <lattice/Grid_lattice_conformable.h>
-
-#ifdef GRID_LATTICE_EXPRESSION_TEMPLATES
-#include <lattice/Grid_lattice_ET.h>
-#else 
-#include <lattice/Grid_lattice_overload.h>
-#endif
-#include <lattice/Grid_lattice_arith.h>
-
-#include <lattice/Grid_lattice_trace.h>
-#include <lattice/Grid_lattice_transpose.h>
-#include <lattice/Grid_lattice_local.h>
-#include <lattice/Grid_lattice_reduction.h>
-#include <lattice/Grid_lattice_peekpoke.h>
-#include <lattice/Grid_lattice_reality.h>
-#include <Grid_extract.h>
-#include <lattice/Grid_lattice_coordinate.h>
-#include <lattice/Grid_lattice_rng.h>
-#include <lattice/Grid_lattice_transfer.h>
-
-
+#include <lattice/Grid_lattice_base.h>
 
 #endif
diff --git a/lib/communicator/Grid_communicator_base.h b/lib/communicator/Grid_communicator_base.h
new file mode 100644
index 00000000..47c1f525
--- /dev/null
+++ b/lib/communicator/Grid_communicator_base.h
@@ -0,0 +1,117 @@
+#ifndef GRID_COMMUNICATOR_BASE_H
+#define GRID_COMMUNICATOR_BASE_H
+
+///////////////////////////////////
+// Processor layout information
+///////////////////////////////////
+#ifdef GRID_COMMS_MPI
+#include <mpi.h>
+#endif
+namespace Grid {
+class CartesianCommunicator {
+  public:    
+
+  // Communicator should know nothing of the physics grid, only processor grid.
+
+    int              _Nprocessors;     // How many in all
+    std::vector<int> _processors;      // Which dimensions get relayed out over processors lanes.
+    int              _processor;       // linear processor rank
+    std::vector<int> _processor_coor;  // linear processor coordinate
+    unsigned long _ndimension;
+
+#ifdef GRID_COMMS_MPI
+    MPI_Comm communicator;
+    typedef MPI_Request CommsRequest_t;
+#else 
+    typedef int CommsRequest_t;
+#endif
+
+    // Constructor
+    CartesianCommunicator(std::vector<int> &pdimensions_in);
+
+    // Wraps MPI_Cart routines
+    void ShiftedRanks(int dim,int shift,int & source, int & dest);
+    int  RankFromProcessorCoor(std::vector<int> &coor);
+    void ProcessorCoorFromRank(int rank,std::vector<int> &coor);
+
+    /////////////////////////////////
+    // Grid information queries
+    /////////////////////////////////
+    int                      IsBoss(void)            { return _processor==0; };
+    int                      BossRank(void)          { return 0; };
+    int                      ThisRank(void)          { return _processor; };
+    const std::vector<int> & ThisProcessorCoor(void) { return _processor_coor; };
+    const std::vector<int> & ProcessorGrid(void)     { return _processors; };
+    int                      ProcessorCount(void)    { return _Nprocessors; };
+
+    ////////////////////////////////////////////////////////////
+    // Reduction
+    ////////////////////////////////////////////////////////////
+    void GlobalSum(RealF &);
+    void GlobalSumVector(RealF *,int N);
+
+    void GlobalSum(RealD &);
+    void GlobalSumVector(RealD *,int N);
+
+    void GlobalSum(uint32_t &);
+
+    void GlobalSum(ComplexF &c)
+    {
+      GlobalSumVector((float *)&c,2);
+    }
+    void GlobalSumVector(ComplexF *c,int N)
+    {
+      GlobalSumVector((float *)c,2*N);
+    }
+
+    void GlobalSum(ComplexD &c)
+    {
+      GlobalSumVector((double *)&c,2);
+    }
+    void GlobalSumVector(ComplexD *c,int N)
+    {
+      GlobalSumVector((double *)c,2*N);
+    }
+    
+    template<class obj> void GlobalSum(obj &o){
+      typedef typename obj::scalar_type scalar_type;
+      int words = sizeof(obj)/sizeof(scalar_type);
+      scalar_type * ptr = (scalar_type *)& o;
+      GlobalSumVector(ptr,words);
+    }
+    ////////////////////////////////////////////////////////////
+    // Face exchange, buffer swap in translational invariant way
+    ////////////////////////////////////////////////////////////
+    void SendToRecvFrom(void *xmit,
+			int xmit_to_rank,
+			void *recv,
+			int recv_from_rank,
+			int bytes);
+    void SendToRecvFromBegin(std::vector<CommsRequest_t> &list,
+			 void *xmit,
+			 int xmit_to_rank,
+			 void *recv,
+			 int recv_from_rank,
+			 int bytes);
+    void SendToRecvFromComplete(std::vector<CommsRequest_t> &waitall);
+
+    ////////////////////////////////////////////////////////////
+    // Barrier
+    ////////////////////////////////////////////////////////////
+    void Barrier(void);
+
+    ////////////////////////////////////////////////////////////
+    // Broadcast a buffer and composite larger
+    ////////////////////////////////////////////////////////////
+    void Broadcast(int root,void* data, int bytes);
+    template<class obj> void Broadcast(int root,obj &data)
+    {
+      Broadcast(root,(void *)&data,sizeof(data));
+    };
+
+    static void BroadcastWorld(int root,void* data, int bytes);
+
+}; 
+}
+
+#endif
diff --git a/lib/lattice/Grid_lattice_base.h b/lib/lattice/Grid_lattice_base.h
new file mode 100644
index 00000000..248bf1f7
--- /dev/null
+++ b/lib/lattice/Grid_lattice_base.h
@@ -0,0 +1,203 @@
+#ifndef GRID_LATTICE_BASE_H
+#define GRID_LATTICE_BASE_H
+
+namespace Grid {
+
+// TODO: 
+//       mac,real,imag
+
+// Functionality:
+//     -=,+=,*=,()
+//     add,+,sub,-,mult,mac,*
+//     adj,conj
+//     real,imag
+//     transpose,transposeIndex  
+//     trace,traceIndex
+//     peekIndex
+//     innerProduct,outerProduct,
+//     localNorm2
+//     localInnerProduct
+
+extern int GridCshiftPermuteMap[4][16];
+
+////////////////////////////////////////////////
+// Basic expressions used in Expression Template
+////////////////////////////////////////////////
+
+class LatticeBase {};
+class LatticeExpressionBase {};
+
+template <typename Op, typename T1>                           
+class LatticeUnaryExpression  : public std::pair<Op,std::tuple<T1> > , public LatticeExpressionBase {
+ public:
+ LatticeUnaryExpression(const std::pair<Op,std::tuple<T1> > &arg): std::pair<Op,std::tuple<T1> >(arg) {};
+};
+
+template <typename Op, typename T1, typename T2>              
+class LatticeBinaryExpression : public std::pair<Op,std::tuple<T1,T2> > , public LatticeExpressionBase {
+ public:
+ LatticeBinaryExpression(const std::pair<Op,std::tuple<T1,T2> > &arg): std::pair<Op,std::tuple<T1,T2> >(arg) {};
+};
+
+template <typename Op, typename T1, typename T2, typename T3> 
+class LatticeTrinaryExpression :public std::pair<Op,std::tuple<T1,T2,T3> >, public LatticeExpressionBase {
+ public:
+ LatticeTrinaryExpression(const std::pair<Op,std::tuple<T1,T2,T3> > &arg): std::pair<Op,std::tuple<T1,T2,T3> >(arg) {};
+};
+
+template<class vobj>
+class Lattice : public LatticeBase
+{
+public:
+
+    GridBase *_grid;
+    int checkerboard;
+    std::vector<vobj,alignedAllocator<vobj> > _odata;
+
+public:
+    typedef typename vobj::scalar_type scalar_type;
+    typedef typename vobj::vector_type vector_type;
+    typedef vobj vector_object;
+    
+  ////////////////////////////////////////////////////////////////////////////////
+  // Expression Template closure support
+  ////////////////////////////////////////////////////////////////////////////////
+  template <typename Op, typename T1>                         inline Lattice<vobj> & operator=(const LatticeUnaryExpression<Op,T1> &expr)
+  {
+#pragma omp parallel for
+    for(int ss=0;ss<_grid->oSites();ss++){
+      vobj tmp= eval(ss,expr);
+      vstream(_odata[ss] ,tmp);
+    }
+    return *this;
+  }
+  template <typename Op, typename T1,typename T2>             inline Lattice<vobj> & operator=(const LatticeBinaryExpression<Op,T1,T2> &expr)
+  {
+#pragma omp parallel for
+    for(int ss=0;ss<_grid->oSites();ss++){
+      vobj tmp= eval(ss,expr);
+      vstream(_odata[ss] ,tmp);
+    }
+    return *this;
+  }
+  template <typename Op, typename T1,typename T2,typename T3> inline Lattice<vobj> & operator=(const LatticeTrinaryExpression<Op,T1,T2,T3> &expr)
+  {
+#pragma omp parallel for
+    for(int ss=0;ss<_grid->oSites();ss++){
+      vobj tmp= eval(ss,expr);
+      vstream(_odata[ss] ,tmp);
+    }
+    return *this;
+  }
+  //GridFromExpression is tricky to do
+  template<class Op,class T1>
+    Lattice(const LatticeUnaryExpression<Op,T1> & expr):    _grid(nullptr){
+    GridFromExpression(_grid,expr);
+    assert(_grid!=nullptr);
+    _odata.resize(_grid->oSites());
+#pragma omp parallel for
+    for(int ss=0;ss<_grid->oSites();ss++){
+      _odata[ss] = eval(ss,expr);
+    }
+  };
+  template<class Op,class T1, class T2>
+  Lattice(const LatticeBinaryExpression<Op,T1,T2> & expr):    _grid(nullptr){
+    GridFromExpression(_grid,expr);
+    assert(_grid!=nullptr);
+    _odata.resize(_grid->oSites());
+#pragma omp parallel for
+    for(int ss=0;ss<_grid->oSites();ss++){
+      _odata[ss] = eval(ss,expr);
+    }
+  };
+  template<class Op,class T1, class T2, class T3>
+  Lattice(const LatticeTrinaryExpression<Op,T1,T2,T3> & expr):    _grid(nullptr){
+    GridFromExpression(_grid,expr);
+    assert(_grid!=nullptr);
+    _odata.resize(_grid->oSites());
+#pragma omp parallel for
+    for(int ss=0;ss<_grid->oSites();ss++){
+      _odata[ss] = eval(ss,expr);
+    }
+  };
+
+    //////////////////////////////////////////////////////////////////
+    // Constructor requires "grid" passed.
+    // what about a default grid?
+    //////////////////////////////////////////////////////////////////
+ Lattice(GridBase *grid) : _grid(grid), _odata(_grid->oSites()) {
+      //        _odata.reserve(_grid->oSites());
+      //        _odata.resize(_grid->oSites());
+        assert((((uint64_t)&_odata[0])&0xF) ==0);
+        checkerboard=0;
+    }
+
+    template<class sobj> inline Lattice<vobj> & operator = (const sobj & r){
+#pragma omp parallel for
+        for(int ss=0;ss<_grid->oSites();ss++){
+            this->_odata[ss]=r;
+        }
+        return *this;
+    }
+    template<class robj> inline Lattice<vobj> & operator = (const Lattice<robj> & r){
+      conformable(*this,r);
+      std::cout<<"Lattice operator ="<<std::endl;
+#pragma omp parallel for
+        for(int ss=0;ss<_grid->oSites();ss++){
+            this->_odata[ss]=r._odata[ss];
+        }
+        return *this;
+    }
+
+    // *=,+=,-= operators inherit behvour from correspond */+/- operation
+    template<class T> inline Lattice<vobj> &operator *=(const T &r) {
+        *this = (*this)*r;
+        return *this;
+    }
+
+    template<class T> inline Lattice<vobj> &operator -=(const T &r) {
+        *this = (*this)-r;
+        return *this;
+    }
+    template<class T> inline Lattice<vobj> &operator +=(const T &r) {
+        *this = (*this)+r;
+        return *this;
+    }
+    
+    inline friend Lattice<vobj> operator / (const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
+        conformable(lhs,rhs);
+        Lattice<vobj> ret(lhs._grid);
+#pragma omp parallel for
+        for(int ss=0;ss<lhs._grid->oSites();ss++){
+            ret._odata[ss] = lhs._odata[ss]/rhs._odata[ss];
+        }
+        return ret;
+    };
+ }; // class Lattice
+}
+
+#undef GRID_LATTICE_EXPRESSION_TEMPLATES
+
+#include <lattice/Grid_lattice_conformable.h>
+
+#ifdef GRID_LATTICE_EXPRESSION_TEMPLATES
+#include <lattice/Grid_lattice_ET.h>
+#else 
+#include <lattice/Grid_lattice_overload.h>
+#endif
+#include <lattice/Grid_lattice_arith.h>
+
+#include <lattice/Grid_lattice_trace.h>
+#include <lattice/Grid_lattice_transpose.h>
+#include <lattice/Grid_lattice_local.h>
+#include <lattice/Grid_lattice_reduction.h>
+#include <lattice/Grid_lattice_peekpoke.h>
+#include <lattice/Grid_lattice_reality.h>
+#include <Grid_extract.h>
+#include <lattice/Grid_lattice_coordinate.h>
+#include <lattice/Grid_lattice_rng.h>
+#include <lattice/Grid_lattice_transfer.h>
+
+
+
+#endif
diff --git a/lib/math/Grid_math_inner.h b/lib/math/Grid_math_inner.h
index 1e44032c..bbb35605 100644
--- a/lib/math/Grid_math_inner.h
+++ b/lib/math/Grid_math_inner.h
@@ -6,7 +6,7 @@ namespace Grid {
     // innerProduct Vector x Vector -> Scalar
     // innerProduct Matrix x Matrix -> Scalar
     ///////////////////////////////////////////////////////////////////////////////////////
-    template<class sobj> inline RealD norm2l(const sobj &arg){
+    template<class sobj> inline RealD norm2(const sobj &arg){
       typedef typename sobj::scalar_type scalar;
       decltype(innerProduct(arg,arg)) nrm;
       nrm = innerProduct(arg,arg);
diff --git a/tests/Grid_gamma.cc b/tests/Grid_gamma.cc
index 9779ca07..ba298b19 100644
--- a/tests/Grid_gamma.cc
+++ b/tests/Grid_gamma.cc
@@ -97,7 +97,7 @@ int main (int argc, char ** argv)
   for(int mu=0;mu<6;mu++){
     result =  Gamma(g[mu])* ident * Gamma(g[mu]);
     result = result - ident;
-    double mag = TensorRemove(norm2l(result));
+    double mag = TensorRemove(norm2(result));
     std::cout << list[mu]<<" " << mag<<std::endl;
   }
 
@@ -105,7 +105,7 @@ int main (int argc, char ** argv)
   for(int mu=0;mu<6;mu++){
     result =          rnd * Gamma(g[mu]);
     result = result + rnd * Gamma(g[mu+6]);
-    double mag = TensorRemove(norm2l(result));
+    double mag = TensorRemove(norm2(result));
     std::cout << list[mu]<<" " << mag<<std::endl;
   }
 
@@ -113,7 +113,7 @@ int main (int argc, char ** argv)
   for(int mu=0;mu<6;mu++){
     result =           Gamma(g[mu])  *rnd;
     result = result +  Gamma(g[mu+6])*rnd;
-    double mag = TensorRemove(norm2l(result));
+    double mag = TensorRemove(norm2(result));
     std::cout << list[mu]<<" " << mag<<std::endl;
   }
 
@@ -127,70 +127,70 @@ int main (int argc, char ** argv)
   spProjXp(hsm,rv);
   spReconXp(recon,hsm);
   full = rv + Gamma(Gamma::GammaX) *rv;
-  mag = TensorRemove(norm2l(full-recon));
+  mag = TensorRemove(norm2(full-recon));
   std::cout << "Xp "<< mag<<std::endl;
 
   // Xm
   spProjXm(hsm,rv);
   spReconXm(recon,hsm);
   full = rv - Gamma(Gamma::GammaX) *rv;
-  mag = TensorRemove(norm2l(full-recon));
+  mag = TensorRemove(norm2(full-recon));
   std::cout << "Xm "<< mag<<std::endl;
 
   // Yp
   spProjYp(hsm,rv);
   spReconYp(recon,hsm);
   full = rv + Gamma(Gamma::GammaY) *rv;
-  mag = TensorRemove(norm2l(full-recon));
+  mag = TensorRemove(norm2(full-recon));
   std::cout << "Yp "<< mag<<std::endl;
 
   // Ym
   spProjYm(hsm,rv);
   spReconYm(recon,hsm);
   full = rv - Gamma(Gamma::GammaY) *rv;
-  mag = TensorRemove(norm2l(full-recon));
+  mag = TensorRemove(norm2(full-recon));
   std::cout << "Ym "<< mag<<std::endl;
 
   // Zp
   spProjZp(hsm,rv);
   spReconZp(recon,hsm);
   full = rv + Gamma(Gamma::GammaZ) *rv;
-  mag = TensorRemove(norm2l(full-recon));
+  mag = TensorRemove(norm2(full-recon));
   std::cout << "Zp "<< mag<<std::endl;
 
   // Zm
   spProjZm(hsm,rv);
   spReconZm(recon,hsm);
   full = rv - Gamma(Gamma::GammaZ) *rv;
-  mag = TensorRemove(norm2l(full-recon));
+  mag = TensorRemove(norm2(full-recon));
   std::cout << "Zm "<< mag<<std::endl;
 
   // Tp
   spProjTp(hsm,rv);
   spReconTp(recon,hsm);
   full = rv + Gamma(Gamma::GammaT) *rv;
-  mag = TensorRemove(norm2l(full-recon));
+  mag = TensorRemove(norm2(full-recon));
   std::cout << "Tp "<< mag<<std::endl;
 
   // Tm
   spProjTm(hsm,rv);
   spReconTm(recon,hsm);
   full = rv - Gamma(Gamma::GammaT) *rv;
-  mag = TensorRemove(norm2l(full-recon));
+  mag = TensorRemove(norm2(full-recon));
   std::cout << "Tm "<< mag<<std::endl;
 
   // 5p
   spProj5p(hsm,rv);
   spRecon5p(recon,hsm);
   full = rv + Gamma(Gamma::Gamma5) *rv;
-  mag = TensorRemove(norm2l(full-recon));
+  mag = TensorRemove(norm2(full-recon));
   std::cout << "5p "<< mag<<std::endl;
 
   // 5m
   spProj5m(hsm,rv);
   spRecon5m(recon,hsm);
   full = rv - Gamma(Gamma::Gamma5) *rv;
-  mag = TensorRemove(norm2l(full-recon));
+  mag = TensorRemove(norm2(full-recon));
   std::cout << "5m "<< mag<<std::endl;
 
   Grid_finalize();

From c6baa3e6578887658f3829ad8d63f42b6f19604a Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 12 May 2015 07:51:41 +0100
Subject: [PATCH 146/429] Threading support rework. Placed parallel pragmas as
 macros; implemented deterministic thread reduction in style of BFM.

---
 TODO                                          |  93 ++++++++-------
 lib/Grid.h                                    |   1 -
 lib/Grid_comparison.h                         |   1 +
 lib/Grid_simd.h                               |  12 +-
 lib/Makefile.am                               |   4 +-
 lib/cshift/Grid_cshift_common.h               |  12 +-
 lib/lattice/Grid_lattice_arith.h              |  26 ++--
 lib/lattice/Grid_lattice_base.h               |  18 +--
 lib/lattice/Grid_lattice_comparison.h         |   6 +-
 lib/lattice/Grid_lattice_local.h              |   6 +-
 lib/lattice/Grid_lattice_overload.h           |  14 +--
 lib/lattice/Grid_lattice_peekpoke.h           |  12 +-
 lib/lattice/Grid_lattice_reality.h            |   8 +-
 lib/lattice/Grid_lattice_reduction.h          | 111 ++++++++++--------
 lib/lattice/Grid_lattice_trace.h              |   4 +-
 lib/lattice/Grid_lattice_transfer.h           |   4 +-
 lib/lattice/Grid_lattice_transpose.h          |   4 +-
 .../Grid_lattice_where.h}                     |   6 +-
 lib/math/Grid_math_reality.h                  |  54 +++++++--
 lib/qcd/Grid_qcd_2spinor.h                    |   6 -
 lib/qcd/Grid_qcd_wilson_dop.cc                |   2 +-
 lib/simd/Grid_vComplexD.h                     |  24 +++-
 lib/simd/Grid_vComplexF.h                     |  21 +++-
 lib/simd/Grid_vRealD.h                        |   4 +
 lib/simd/Grid_vRealF.h                        |   4 +
 tests/Grid_gamma.cc                           |   3 +-
 26 files changed, 276 insertions(+), 184 deletions(-)
 rename lib/{Grid_where.h => lattice/Grid_lattice_where.h} (95%)

diff --git a/TODO b/TODO
index 1f5be83f..6a1624f7 100644
--- a/TODO
+++ b/TODO
@@ -1,33 +1,35 @@
+================================================================
+*** Hacks and bug fixes to clean up and Audits
+================================================================
+* Base class to share common code between vRealF, VComplexF etc... 
+
+* Performance check on Guido's reimplementation strategy
+
+* Bug in SeedFixedIntegers gives same output on each site. -- Think I fixed but NOT checked for sure
+
+* FIXME audit
+* const audit
+
+* Replace vset with a call to merge.; 
+* care in Gmerge,Gextract over vset .
+* extract / merge extra implementation removal      
+
+* Strong test for norm2, conj and all primitive types. -- tests/Grid_simd.cc is almost there
+
+================================================================
+*** New Functionality
+================================================================
+
+* Implement where to take template scheme.
+
 * - BinaryWriter, TextWriter etc...
   - use protocol buffers? replace xmlReader/Writer ec..
   - Binary use htonll, htonl
 
-*** Hacks and bug fixes to clean up
-* Had to hack assignment to 1.0 in the tests/Grid_gamma test
-* norm2l is a hack. figure out syntax error and make this norm2 c.f. tests/Grid_gamma.cc
-
-* Reduce implemention is poor ; need threaded reductions; OMP isn't able to do it for generic objects.
-  
-* Bug in SeedFixedIntegers gives same output on each site.
-* Bug in RNG with complex numbers ; only filling real values; need helper function -- DONE
-* Conformable test in Cshift routines.
-
-*** Functionality
-
-* Implement where to take template scheme.
-
-* Command line args for geometry, simd, etc. layout. Is it necessary to have
-  user pass these? Is this a QCD specific?
-
-* Strong test for norm2, conj and all primitive types. -- Grid_simd test is almost there
-
 * Expression template engine:
 
-   - Audit
-   - Introduce base clase for Grid Tensors.
-   - Introduce norm2 unary op.
-   - Introduce conversion automatic from expression to Lattice<vobj>
-
+   -- Audit
+   -- Norm2(expression) problem: introduce norm2 unary op, or Introduce conversion automatic from expression to Lattice<vobj>
 
 * CovariantShift support                             -----Use a class to store gauge field? (parallel transport?)
 
@@ -48,12 +50,10 @@
  - I have collated into single location at least.
  - Need to use _mm_*insert/extract routines.
 
-
 * Flavour matrices?
 * Pauli, SU subgroup, etc.. 
 * su3 exponentiation & log etc.. [Jamie's code?]
 * TaProj
-
 * FFTnD ?
 
 * Parallel MPI2 IO
@@ -62,20 +62,23 @@
 * rb4d support for 5th dimension in Mobius.
 
 * Check for missing functionality                    - partially audited against QDP++ layout
-   // Base class to share common code between vRealF, VComplexF etc...
-   //
    // Unary functions
    // cos,sin, tan, acos, asin, cosh, acosh, tanh, sinh, // Scalar<vReal> only arg
    // exp, log, sqrt, fabs
-   //
    // transposeColor, transposeSpin,
    // adjColor, adjSpin,
-   //
    // copyMask.
-   //
    // localMaxAbs
-   //
    // Fourier transform equivalent.
+Actions
+* Fermion
+  - Wilson
+  - Clover
+  - DomainWall
+  - Mobius
+  - z-Mobius
+* Gauge
+  - Wilson, symanzik, iwasaki
 
 Algorithms
 * LinearOperator
@@ -83,23 +86,21 @@ Algorithms
 * Polynomial 
 * Eigen
 * Pcg
+* Adef2
+* DeflCG
 * fPcg
 * MCR
 * HDCG
+* HMC, Heatbath
 * etc..
 
-AUDITS:
-
-* FIXME audit
-* const audit
-* Replace vset with a call to merge.; 
-* care in Gmerge,Gextract over vset .
-* extract / merge extra implementation removal      
-
-
+======================================================================================================
+FUNCTIONALITY: it pleases me to keep track of things I have done (keeps sane)
 ======================================================================================================
 
-FUNCTIONALITY:
+* Command line args for geometry, simd, etc. layout. Is it necessary to have -- DONE
+  user pass these? Is this a QCD specific?
+
 * Stencil -- DONE
 * Test infrastructure -- DONE
 * Fourspin, two spin project --- DONE
@@ -115,6 +116,7 @@ FUNCTIONALITY:
 * How to do U[mu] ... lorentz part of type structure or not. more like chroma if not. -- DONE
 
 * Twospin/Fourspin/Gamma/Proj/Recon                  ----- DONE
+* norm2l is a hack. figure out syntax error and make this norm2 c.f. tests/Grid_gamma.cc -- DONE
 
 * subdirs lib, tests ??                              ----- DONE
   - lib/math        
@@ -149,3 +151,10 @@ FUNCTIONALITY:
 
 * Controling std::cout                              ------- DONE
 
+* Had to hack assignment to 1.0 in the tests/Grid_gamma test -- DONE
+* Reduce implemention is poor ; need threaded reductions; OMP isn't able to do it for generic objects. -- DONE
+* Bug in RNG with complex numbers ; only filling real values; need helper function -- DONE
+* Conformable test in Cshift routines. -- none needed ; there is only one 
+* Conformable testing in expression templates -- DONE (recursive)
+
+
diff --git a/lib/Grid.h b/lib/Grid.h
index da04438a..4512f443 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -53,7 +53,6 @@
 #include <Grid_lattice.h>
 #include <Grid_comparison.h>
 #include <Grid_cshift.h>
-#include <Grid_where.h>
 #include <Grid_stencil.h>
 #include <qcd/Grid_qcd.h>
 #include <parallelIO/GridNerscIO.h>
diff --git a/lib/Grid_comparison.h b/lib/Grid_comparison.h
index 45311831..3f9c206d 100644
--- a/lib/Grid_comparison.h
+++ b/lib/Grid_comparison.h
@@ -142,5 +142,6 @@ namespace Grid {
 }
 
 #include <lattice/Grid_lattice_comparison.h>
+#include <lattice/Grid_lattice_where.h>
 
 #endif
diff --git a/lib/Grid_simd.h b/lib/Grid_simd.h
index 6695eb0d..39265896 100644
--- a/lib/Grid_simd.h
+++ b/lib/Grid_simd.h
@@ -68,10 +68,14 @@ namespace Grid {
     inline ComplexF  adj(const ComplexF& r ){ return(conj(r)); }
     //conj already supported for complex
 
-    inline ComplexF timesI(const ComplexF r)     { return(r*ComplexF(0.0,1.0));}
-    inline ComplexD timesI(const ComplexD r)     { return(r*ComplexD(0.0,1.0));}
-    inline ComplexF timesMinusI(const ComplexF r){ return(r*ComplexF(0.0,-1.0));}
-    inline ComplexD timesMinusI(const ComplexD r){ return(r*ComplexD(0.0,-1.0));}
+    inline ComplexF timesI(const ComplexF &r)     { return(r*ComplexF(0.0,1.0));}
+    inline ComplexD timesI(const ComplexD &r)     { return(r*ComplexD(0.0,1.0));}
+    inline ComplexF timesMinusI(const ComplexF &r){ return(r*ComplexF(0.0,-1.0));}
+    inline ComplexD timesMinusI(const ComplexD &r){ return(r*ComplexD(0.0,-1.0));}
+    inline void timesI(ComplexF &ret,const ComplexF &r)     { ret = timesI(r);}
+    inline void timesI(ComplexD &ret,const ComplexD &r)     { ret = timesI(r);}
+    inline void timesMinusI(ComplexF &ret,const ComplexF &r){ ret = timesMinusI(r);}
+    inline void timesMinusI(ComplexD &ret,const ComplexD &r){ ret = timesMinusI(r);}
 
     inline void mac (RealD * __restrict__ y,const RealD * __restrict__ a,const RealD *__restrict__ x){  *y = (*a) * (*x)+(*y);}
     inline void mult(RealD * __restrict__ y,const RealD * __restrict__ l,const RealD *__restrict__ r){ *y = (*l) * (*r);}
diff --git a/lib/Makefile.am b/lib/Makefile.am
index 46795a33..f76b0b64 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -35,8 +35,7 @@ include_HEADERS =\
 	Grid_lattice.h\
 	Grid_math.h\
 	Grid_simd.h\
-	Grid_stencil.h\
-	Grid_where.h
+	Grid_stencil.h
 
 nobase_include_HEADERS=\
 	cartesian/Grid_cartesian_base.h\
@@ -56,6 +55,7 @@ nobase_include_HEADERS=\
 	lattice/Grid_lattice_trace.h\
 	lattice/Grid_lattice_transfer.h\
 	lattice/Grid_lattice_transpose.h\
+	lattice/Grid_lattice_where.h\
 	math/Grid_math_arith.h\
 	math/Grid_math_arith_add.h\
 	math/Grid_math_arith_mac.h\
diff --git a/lib/cshift/Grid_cshift_common.h b/lib/cshift/Grid_cshift_common.h
index 8320d2d0..9864df73 100644
--- a/lib/cshift/Grid_cshift_common.h
+++ b/lib/cshift/Grid_cshift_common.h
@@ -28,7 +28,7 @@ Gather_plane_simple (const Lattice<vobj> &rhs,std::vector<cobj,alignedAllocator<
   int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
   int bo  = 0;                                      // offset in buffer
   
-#pragma omp parallel for collapse(2)
+PARALLEL_NESTED_LOOP(2)
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
       int o = n*rhs._grid->_slice_stride[dimension];
@@ -57,7 +57,7 @@ Gather_plane_extract(const Lattice<vobj> &rhs,std::vector<typename cobj::scalar_
   int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
   int bo  = 0;                                      // offset in buffer
     
-#pragma omp parallel for collapse(2)
+PARALLEL_NESTED_LOOP(2)
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
 
@@ -106,7 +106,7 @@ template<class vobj> void Scatter_plane_simple (Lattice<vobj> &rhs,std::vector<v
   int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
   int bo  = 0;                                      // offset in buffer
     
-#pragma omp parallel for collapse(2)
+PARALLEL_NESTED_LOOP(2)
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
       int o=n*rhs._grid->_slice_stride[dimension];
@@ -131,7 +131,7 @@ template<class vobj> void Scatter_plane_simple (Lattice<vobj> &rhs,std::vector<v
 
   int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
     
-#pragma omp parallel for collapse(2)
+PARALLEL_NESTED_LOOP(2)
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
 
@@ -160,7 +160,7 @@ template<class vobj> void Copy_plane(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int
   int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
   int lo  = lplane*lhs._grid->_ostride[dimension]; // base offset for start of plane 
   
-#pragma omp parallel for collapse(2)
+PARALLEL_NESTED_LOOP(2)
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
       
@@ -185,7 +185,7 @@ template<class vobj> void Copy_plane_permute(Lattice<vobj>& lhs,Lattice<vobj> &r
   int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
   int lo  = lplane*lhs._grid->_ostride[dimension]; // base offset for start of plane 
   
-#pragma omp parallel for collapse(2)
+PARALLEL_NESTED_LOOP(2)
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
       int o =n*rhs._grid->_slice_stride[dimension];
diff --git a/lib/lattice/Grid_lattice_arith.h b/lib/lattice/Grid_lattice_arith.h
index 93801be5..c1d58b86 100644
--- a/lib/lattice/Grid_lattice_arith.h
+++ b/lib/lattice/Grid_lattice_arith.h
@@ -10,7 +10,7 @@ namespace Grid {
   template<class obj1,class obj2,class obj3>
     void mult(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
     conformable(lhs,rhs);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
       obj1 tmp;
       mult(&tmp,&lhs._odata[ss],&rhs._odata[ss]);
@@ -21,7 +21,7 @@ namespace Grid {
   template<class obj1,class obj2,class obj3>
     void mac(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
     conformable(lhs,rhs);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
       obj1 tmp;
       mac(&tmp,&lhs._odata[ss],&rhs._odata[ss]);
@@ -32,7 +32,7 @@ namespace Grid {
   template<class obj1,class obj2,class obj3>
     void sub(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
     conformable(lhs,rhs);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
       obj1 tmp;
       sub(&tmp,&lhs._odata[ss],&rhs._odata[ss]);
@@ -42,7 +42,7 @@ namespace Grid {
   template<class obj1,class obj2,class obj3>
     void add(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
     conformable(lhs,rhs);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
       obj1 tmp;
       add(&tmp,&lhs._odata[ss],&rhs._odata[ss]);
@@ -56,7 +56,7 @@ namespace Grid {
   template<class obj1,class obj2,class obj3>
     void mult(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
     conformable(lhs,ret);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
       obj1 tmp;
       mult(&tmp,&lhs._odata[ss],&rhs);
@@ -67,7 +67,7 @@ namespace Grid {
   template<class obj1,class obj2,class obj3>
     void mac(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
     conformable(lhs,ret);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
       obj1 tmp;
       mac(&tmp,&lhs._odata[ss],&rhs);
@@ -78,7 +78,7 @@ namespace Grid {
   template<class obj1,class obj2,class obj3>
     void sub(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
     conformable(lhs,ret);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
       obj1 tmp;
       sub(&tmp,&lhs._odata[ss],&rhs);
@@ -88,7 +88,7 @@ namespace Grid {
   template<class obj1,class obj2,class obj3>
     void add(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
     conformable(lhs,ret);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
       obj1 tmp;
       add(&tmp,&lhs._odata[ss],&rhs);
@@ -102,7 +102,7 @@ namespace Grid {
     template<class obj1,class obj2,class obj3>
     void mult(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
     conformable(ret,rhs);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<rhs._grid->oSites();ss++){
       obj1 tmp;
       mult(&tmp,&lhs,&rhs._odata[ss]);
@@ -113,7 +113,7 @@ namespace Grid {
   template<class obj1,class obj2,class obj3>
     void mac(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
     conformable(ret,rhs);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<rhs._grid->oSites();ss++){
       obj1 tmp;
       mac(&tmp,&lhs,&rhs._odata[ss]);
@@ -124,7 +124,7 @@ namespace Grid {
   template<class obj1,class obj2,class obj3>
     void sub(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
     conformable(ret,rhs);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<rhs._grid->oSites();ss++){
       obj1 tmp;
       sub(&tmp,&lhs,&rhs._odata[ss]);
@@ -134,7 +134,7 @@ namespace Grid {
   template<class obj1,class obj2,class obj3>
     void add(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
     conformable(ret,rhs);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<rhs._grid->oSites();ss++){
       obj1 tmp;
       add(&tmp,&lhs,&rhs._odata[ss]);
@@ -145,7 +145,7 @@ namespace Grid {
   template<class sobj,class vobj>
   inline void axpy(Lattice<vobj> &ret,sobj a,const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
     conformable(lhs,rhs);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
       vobj tmp = a*lhs._odata[ss];
       vstream(ret._odata[ss],tmp+rhs._odata[ss]);
diff --git a/lib/lattice/Grid_lattice_base.h b/lib/lattice/Grid_lattice_base.h
index 248bf1f7..cd376b32 100644
--- a/lib/lattice/Grid_lattice_base.h
+++ b/lib/lattice/Grid_lattice_base.h
@@ -64,7 +64,7 @@ public:
   ////////////////////////////////////////////////////////////////////////////////
   template <typename Op, typename T1>                         inline Lattice<vobj> & operator=(const LatticeUnaryExpression<Op,T1> &expr)
   {
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
       vobj tmp= eval(ss,expr);
       vstream(_odata[ss] ,tmp);
@@ -73,7 +73,7 @@ public:
   }
   template <typename Op, typename T1,typename T2>             inline Lattice<vobj> & operator=(const LatticeBinaryExpression<Op,T1,T2> &expr)
   {
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
       vobj tmp= eval(ss,expr);
       vstream(_odata[ss] ,tmp);
@@ -82,7 +82,7 @@ public:
   }
   template <typename Op, typename T1,typename T2,typename T3> inline Lattice<vobj> & operator=(const LatticeTrinaryExpression<Op,T1,T2,T3> &expr)
   {
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
       vobj tmp= eval(ss,expr);
       vstream(_odata[ss] ,tmp);
@@ -95,7 +95,7 @@ public:
     GridFromExpression(_grid,expr);
     assert(_grid!=nullptr);
     _odata.resize(_grid->oSites());
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
       _odata[ss] = eval(ss,expr);
     }
@@ -105,7 +105,7 @@ public:
     GridFromExpression(_grid,expr);
     assert(_grid!=nullptr);
     _odata.resize(_grid->oSites());
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
       _odata[ss] = eval(ss,expr);
     }
@@ -115,7 +115,7 @@ public:
     GridFromExpression(_grid,expr);
     assert(_grid!=nullptr);
     _odata.resize(_grid->oSites());
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
       _odata[ss] = eval(ss,expr);
     }
@@ -133,7 +133,7 @@ public:
     }
 
     template<class sobj> inline Lattice<vobj> & operator = (const sobj & r){
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<_grid->oSites();ss++){
             this->_odata[ss]=r;
         }
@@ -142,7 +142,7 @@ public:
     template<class robj> inline Lattice<vobj> & operator = (const Lattice<robj> & r){
       conformable(*this,r);
       std::cout<<"Lattice operator ="<<std::endl;
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<_grid->oSites();ss++){
             this->_odata[ss]=r._odata[ss];
         }
@@ -167,7 +167,7 @@ public:
     inline friend Lattice<vobj> operator / (const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
         conformable(lhs,rhs);
         Lattice<vobj> ret(lhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){
             ret._odata[ss] = lhs._odata[ss]/rhs._odata[ss];
         }
diff --git a/lib/lattice/Grid_lattice_comparison.h b/lib/lattice/Grid_lattice_comparison.h
index 82c98572..431cf911 100644
--- a/lib/lattice/Grid_lattice_comparison.h
+++ b/lib/lattice/Grid_lattice_comparison.h
@@ -15,7 +15,7 @@ namespace Grid {
     inline Lattice<vInteger> LLComparison(vfunctor op,const Lattice<lobj> &lhs,const Lattice<robj> &rhs)
     {
       Lattice<vInteger> ret(rhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<rhs._grid->oSites(); ss++){
 	  ret._odata[ss]=op(lhs._odata[ss],rhs._odata[ss]);
         }
@@ -25,7 +25,7 @@ namespace Grid {
     inline Lattice<vInteger> LSComparison(vfunctor op,const Lattice<lobj> &lhs,const robj &rhs)
     {
       Lattice<vInteger> ret(lhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites(); ss++){
 	  ret._odata[ss]=op(lhs._odata[ss],rhs);
         }
@@ -35,7 +35,7 @@ namespace Grid {
     inline Lattice<vInteger> SLComparison(vfunctor op,const lobj &lhs,const Lattice<robj> &rhs)
     {
       Lattice<vInteger> ret(rhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<rhs._grid->oSites(); ss++){
 	  ret._odata[ss]=op(lhs._odata[ss],rhs);
         }
diff --git a/lib/lattice/Grid_lattice_local.h b/lib/lattice/Grid_lattice_local.h
index 181f5be2..7b8bdf9b 100644
--- a/lib/lattice/Grid_lattice_local.h
+++ b/lib/lattice/Grid_lattice_local.h
@@ -16,7 +16,7 @@ namespace Grid {
     inline auto localNorm2 (const Lattice<vobj> &rhs)-> Lattice<typename vobj::tensor_reduced>
     {
       Lattice<typename vobj::tensor_reduced> ret(rhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<rhs._grid->oSites(); ss++){
 	  ret._odata[ss]=innerProduct(rhs._odata[ss],rhs._odata[ss]);
         }
@@ -29,7 +29,7 @@ namespace Grid {
       -> Lattice<typename vobj::tensor_reduced>
     {
       Lattice<typename vobj::tensor_reduced> ret(rhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
       for(int ss=0;ss<rhs._grid->oSites(); ss++){
 	ret._odata[ss]=innerProduct(lhs._odata[ss],rhs._odata[ss]);
       }
@@ -42,7 +42,7 @@ namespace Grid {
     inline auto outerProduct (const Lattice<ll> &lhs,const Lattice<rr> &rhs) -> Lattice<decltype(outerProduct(lhs._odata[0],rhs._odata[0]))>
     {
         Lattice<decltype(outerProduct(lhs._odata[0],rhs._odata[0]))> ret(rhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<rhs._grid->oSites(); ss++){
             ret._odata[ss]=outerProduct(lhs._odata[ss],rhs._odata[ss]);
         }
diff --git a/lib/lattice/Grid_lattice_overload.h b/lib/lattice/Grid_lattice_overload.h
index c4dd3373..24bdd70a 100644
--- a/lib/lattice/Grid_lattice_overload.h
+++ b/lib/lattice/Grid_lattice_overload.h
@@ -10,7 +10,7 @@ namespace Grid {
   inline Lattice<vobj> operator -(const Lattice<vobj> &r)
   {
     Lattice<vobj> ret(r._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<r._grid->oSites();ss++){
       vstream(ret._odata[ss], -r._odata[ss]);
     }
@@ -47,7 +47,7 @@ namespace Grid {
   inline auto operator * (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs*rhs._odata[0])>
   {
     Lattice<decltype(lhs*rhs._odata[0])> ret(rhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<rhs._grid->oSites(); ss++){
       decltype(lhs*rhs._odata[0]) tmp=lhs*rhs._odata[ss]; 
       vstream(ret._odata[ss],tmp);
@@ -59,7 +59,7 @@ namespace Grid {
     inline auto operator + (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs+rhs._odata[0])>
     {
       Lattice<decltype(lhs+rhs._odata[0])> ret(rhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
       for(int ss=0;ss<rhs._grid->oSites(); ss++){
 	decltype(lhs+rhs._odata[0]) tmp =lhs-rhs._odata[ss];  
 	vstream(ret._odata[ss],tmp);
@@ -71,7 +71,7 @@ namespace Grid {
     inline auto operator - (const left &lhs,const Lattice<right> &rhs) -> Lattice<decltype(lhs-rhs._odata[0])>
   {
     Lattice<decltype(lhs-rhs._odata[0])> ret(rhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<rhs._grid->oSites(); ss++){
       decltype(lhs-rhs._odata[0]) tmp=lhs-rhs._odata[ss];  
       vstream(ret._odata[ss],tmp);
@@ -83,7 +83,7 @@ namespace Grid {
       inline auto operator * (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]*rhs)>
     {
       Lattice<decltype(lhs._odata[0]*rhs)> ret(lhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
       for(int ss=0;ss<lhs._grid->oSites(); ss++){
 	decltype(lhs._odata[0]*rhs) tmp =lhs._odata[ss]*rhs;
 	vstream(ret._odata[ss],tmp);
@@ -95,7 +95,7 @@ namespace Grid {
       inline auto operator + (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]+rhs)>
     {
         Lattice<decltype(lhs._odata[0]+rhs)> ret(lhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<rhs._grid->oSites(); ss++){
 	  decltype(lhs._odata[0]+rhs) tmp=lhs._odata[ss]+rhs; 
 	  vstream(ret._odata[ss],tmp);
@@ -107,7 +107,7 @@ namespace Grid {
       inline auto operator - (const Lattice<left> &lhs,const right &rhs) -> Lattice<decltype(lhs._odata[0]-rhs)>
     {
       Lattice<decltype(lhs._odata[0]-rhs)> ret(lhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
       for(int ss=0;ss<rhs._grid->oSites(); ss++){
 	  decltype(lhs._odata[0]-rhs) tmp=lhs._odata[ss]-rhs;
 	  vstream(ret._odata[ss],tmp);
diff --git a/lib/lattice/Grid_lattice_peekpoke.h b/lib/lattice/Grid_lattice_peekpoke.h
index ae87c5d8..ad7dfc50 100644
--- a/lib/lattice/Grid_lattice_peekpoke.h
+++ b/lib/lattice/Grid_lattice_peekpoke.h
@@ -15,7 +15,7 @@ namespace Grid {
       -> Lattice<decltype(peekIndex<Index>(lhs._odata[0]))>
     {
       Lattice<decltype(peekIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){
             ret._odata[ss] = peekIndex<Index>(lhs._odata[ss]);
         }
@@ -26,7 +26,7 @@ namespace Grid {
       -> Lattice<decltype(peekIndex<Index>(lhs._odata[0],i))>
     {
       Lattice<decltype(peekIndex<Index>(lhs._odata[0],i))> ret(lhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){
 	  ret._odata[ss] = peekIndex<Index>(lhs._odata[ss],i);
         }
@@ -37,7 +37,7 @@ namespace Grid {
       -> Lattice<decltype(peekIndex<Index>(lhs._odata[0],i,j))>
     {
       Lattice<decltype(peekIndex<Index>(lhs._odata[0],i,j))> ret(lhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){
 	  ret._odata[ss] = peekIndex<Index>(lhs._odata[ss],i,j);
         }
@@ -50,7 +50,7 @@ namespace Grid {
     template<int Index,class vobj> inline
     void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0]))> & rhs)
     {
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){
 	  pokeIndex<Index>(lhs._odata[ss],rhs._odata[ss]);
 	}      
@@ -58,7 +58,7 @@ namespace Grid {
     template<int Index,class vobj> inline
     void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0],0))> & rhs,int i)
     {
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){
 	  pokeIndex<Index>(lhs._odata[ss],rhs._odata[ss],i);
 	}      
@@ -66,7 +66,7 @@ namespace Grid {
     template<int Index,class vobj> inline
     void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0],0,0))> & rhs,int i,int j)
     {
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){
 	  pokeIndex<Index>(lhs._odata[ss],rhs._odata[ss],i,j);
 	}      
diff --git a/lib/lattice/Grid_lattice_reality.h b/lib/lattice/Grid_lattice_reality.h
index a3a7a27f..cf86d700 100644
--- a/lib/lattice/Grid_lattice_reality.h
+++ b/lib/lattice/Grid_lattice_reality.h
@@ -11,7 +11,7 @@ namespace Grid {
 
     template<class vobj> inline Lattice<vobj> adj(const Lattice<vobj> &lhs){
         Lattice<vobj> ret(lhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){
             ret._odata[ss] = adj(lhs._odata[ss]);
         }
@@ -20,7 +20,7 @@ namespace Grid {
 
     template<class vobj> inline Lattice<vobj> conj(const Lattice<vobj> &lhs){
         Lattice<vobj> ret(lhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){
             ret._odata[ss] = conj(lhs._odata[ss]);
         }
@@ -30,7 +30,7 @@ namespace Grid {
     template<class vobj> inline auto real(const Lattice<vobj> &z) -> Lattice<decltype(real(z._odata[0]))>
     {
       Lattice<decltype(real(z._odata[0]))> ret(z._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<z._grid->oSites();ss++){
             ret._odata[ss] = real(z._odata[ss]);
         }
@@ -40,7 +40,7 @@ namespace Grid {
     template<class vobj> inline auto imag(const Lattice<vobj> &z) -> Lattice<decltype(imag(z._odata[0]))>
     {
       Lattice<decltype(imag(z._odata[0]))> ret(z._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<z._grid->oSites();ss++){
             ret._odata[ss] = imag(z._odata[ss]);
         }
diff --git a/lib/lattice/Grid_lattice_reduction.h b/lib/lattice/Grid_lattice_reduction.h
index 679effb0..a6d3e0dc 100644
--- a/lib/lattice/Grid_lattice_reduction.h
+++ b/lib/lattice/Grid_lattice_reduction.h
@@ -7,69 +7,85 @@ namespace Grid {
 #endif     
 
     ////////////////////////////////////////////////////////////////////////////////////////////////////
-    // Reduction operations
+    // Deterministic Reduction operations
     ////////////////////////////////////////////////////////////////////////////////////////////////////
-    template<class vobj>
-    inline RealD norm2(const Lattice<vobj> &arg){
-
-      typedef typename vobj::scalar_type scalar;
-      typedef typename vobj::vector_type vector;
-      decltype(innerProduct(arg._odata[0],arg._odata[0])) vnrm; 
-      scalar nrm;
-      //FIXME make this loop parallelisable
-      vnrm=zero;
-      for(int ss=0;ss<arg._grid->oSites(); ss++){
-	vnrm = vnrm + innerProduct(arg._odata[ss],arg._odata[ss]);
-      }
-      vector vvnrm =TensorRemove(vnrm) ;
-      nrm = Reduce(vvnrm);
-      arg._grid->GlobalSum(nrm);
-      return real(nrm);
-    }
+  template<class vobj> inline RealD norm2(const Lattice<vobj> &arg){
+    ComplexD nrm = innerProduct(arg,arg);
+    return real(nrm); 
+  }
 
     template<class vobj>
     inline ComplexD innerProduct(const Lattice<vobj> &left,const Lattice<vobj> &right) 
-      //    inline auto innerProduct(const Lattice<vobj> &left,const Lattice<vobj> &right) 
-      //->decltype(innerProduct(left._odata[0],right._odata[0]))
     {
-      typedef typename vobj::scalar_type scalar;
-      decltype(innerProduct(left._odata[0],right._odata[0])) vnrm; 
+      typedef typename vobj::scalar_type scalar_type;
+      typedef typename vobj::vector_type vector_type;
+      vector_type vnrm;
+      scalar_type  nrm;
 
-      scalar nrm;
-      //FIXME make this loop parallelisable
-      vnrm=zero;
-      for(int ss=0;ss<left._grid->oSites(); ss++){
-	vnrm = vnrm + innerProduct(left._odata[ss],right._odata[ss]);
+      GridBase *grid = left._grid;
+
+      std::vector<vector_type,alignedAllocator<vector_type> > sumarray(grid->SumArraySize());
+      for(int i=0;i<grid->SumArraySize();i++){
+	sumarray[i]=zero;
       }
-      nrm = Reduce(vnrm);
+
+PARALLEL_FOR_LOOP
+      for(int thr=0;thr<grid->SumArraySize();thr++){
+
+	int nwork, mywork, myoff;
+	GridThread::GetWork(left._grid->oSites(),thr,mywork,myoff);
+	
+	decltype(innerProduct(left._odata[0],right._odata[0])) vnrm=zero; // private to thread; sub summation
+        for(int ss=myoff;ss<mywork+myoff; ss++){
+	  vnrm = vnrm + innerProduct(left._odata[ss],right._odata[ss]);
+	}
+	sumarray[thr]=TensorRemove(vnrm) ;
+      }
+    
+      vector_type vvnrm; vvnrm=zero;  // sum across threads
+      for(int i=0;i<grid->SumArraySize();i++){
+	vvnrm = vvnrm+sumarray[i];
+      } 
+      nrm = Reduce(vvnrm);// sum across simd
       right._grid->GlobalSum(nrm);
       return nrm;
     }
 
     template<class vobj>
-      inline typename vobj::scalar_object sum(const Lattice<vobj> &arg){
+    inline typename vobj::scalar_object sum(const Lattice<vobj> &arg){
 
       GridBase *grid=arg._grid;
       int Nsimd = grid->Nsimd();
 
-      typedef typename vobj::scalar_object sobj;
-      typedef typename vobj::scalar_type   scalar_type;
-
-      vobj vsum;
-      sobj ssum;
-
-      vsum=zero;
-      ssum=zero;
-      //FIXME make this loop parallelisable
-      for(int ss=0;ss<arg._grid->oSites(); ss++){
-	vsum = vsum + arg._odata[ss];
+      std::vector<vobj,alignedAllocator<vobj> > sumarray(grid->SumArraySize());
+      for(int i=0;i<grid->SumArraySize();i++){
+	sumarray[i]=zero;
       }
-      
+
+PARALLEL_FOR_LOOP
+      for(int thr=0;thr<grid->SumArraySize();thr++){
+	int nwork, mywork, myoff;
+	GridThread::GetWork(grid->oSites(),thr,mywork,myoff);
+
+	vobj vvsum=zero;
+        for(int ss=myoff;ss<mywork+myoff; ss++){
+	  vvsum = vvsum + arg._odata[ss];
+	}
+	sumarray[thr]=vvsum;
+      }
+
+      vobj vsum=zero;  // sum across threads
+      for(int i=0;i<grid->SumArraySize();i++){
+	vsum = vsum+sumarray[i];
+      } 
+
+      typedef typename vobj::scalar_object sobj;
+      sobj ssum=zero;
+
       std::vector<sobj>               buf(Nsimd);
       extract(vsum,buf);
 
       for(int i=0;i<Nsimd;i++) ssum = ssum + buf[i];
-
       arg._grid->GlobalSum(ssum);
 
       return ssum;
@@ -77,13 +93,15 @@ namespace Grid {
 
 
 
-
 template<class vobj> inline void sliceSum(const Lattice<vobj> &Data,std::vector<typename vobj::scalar_object> &result,int orthogdim)
 {
   typedef typename vobj::scalar_object sobj;
-
   GridBase  *grid = Data._grid;
   assert(grid!=NULL);
+
+  // FIXME
+  std::cout<<"WARNING ! SliceSum is unthreaded "<<grid->SumArraySize()<<" threads "<<std::endl;
+
   const int    Nd = grid->_ndimension;
   const int Nsimd = grid->Nsimd();
 
@@ -94,10 +112,8 @@ template<class vobj> inline void sliceSum(const Lattice<vobj> &Data,std::vector<
   int ld=grid->_ldimensions[orthogdim];
   int rd=grid->_rdimensions[orthogdim];
 
-  sobj szero; szero=zero;
-
   std::vector<vobj,alignedAllocator<vobj> > lvSum(rd); // will locally sum vectors first
-  std::vector<sobj> lsSum(ld,szero); // sum across these down to scalars
+  std::vector<sobj> lsSum(ld,zero); // sum across these down to scalars
   std::vector<sobj> extracted(Nsimd);     // splitting the SIMD
 
   result.resize(fd); // And then global sum to return the same vector to every node for IO to file
@@ -108,6 +124,7 @@ template<class vobj> inline void sliceSum(const Lattice<vobj> &Data,std::vector<
   std::vector<int>  coor(Nd);  
 
   // sum over reduced dimension planes, breaking out orthog dir
+
   for(int ss=0;ss<grid->oSites();ss++){
     GridBase::CoorFromIndex(coor,ss,grid->_rdimensions);
     int r = coor[orthogdim];
diff --git a/lib/lattice/Grid_lattice_trace.h b/lib/lattice/Grid_lattice_trace.h
index 8c7607d8..72f984f9 100644
--- a/lib/lattice/Grid_lattice_trace.h
+++ b/lib/lattice/Grid_lattice_trace.h
@@ -15,7 +15,7 @@ namespace Grid {
       -> Lattice<decltype(trace(lhs._odata[0]))>
     {
       Lattice<decltype(trace(lhs._odata[0]))> ret(lhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){
             ret._odata[ss] = trace(lhs._odata[ss]);
         }
@@ -30,7 +30,7 @@ namespace Grid {
       -> Lattice<decltype(traceIndex<Index>(lhs._odata[0]))>
     {
       Lattice<decltype(traceIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
       for(int ss=0;ss<lhs._grid->oSites();ss++){
 	ret._odata[ss] = traceIndex<Index>(lhs._odata[ss]);
       }
diff --git a/lib/lattice/Grid_lattice_transfer.h b/lib/lattice/Grid_lattice_transfer.h
index 53973741..27ae27e1 100644
--- a/lib/lattice/Grid_lattice_transfer.h
+++ b/lib/lattice/Grid_lattice_transfer.h
@@ -23,7 +23,7 @@ inline void subdivides(GridBase *coarse,GridBase *fine)
   template<class vobj> inline void pickCheckerboard(int cb,Lattice<vobj> &half,const Lattice<vobj> &full){
     half.checkerboard = cb;
     int ssh=0;
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<full._grid->oSites();ss++){
       std::vector<int> coor;
       int cbos;
@@ -41,7 +41,7 @@ inline void subdivides(GridBase *coarse,GridBase *fine)
   template<class vobj> inline void setCheckerboard(Lattice<vobj> &full,const Lattice<vobj> &half){
     int cb = half.checkerboard;
     int ssh=0;
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<full._grid->oSites();ss++){
       std::vector<int> coor;
       int cbos;
diff --git a/lib/lattice/Grid_lattice_transpose.h b/lib/lattice/Grid_lattice_transpose.h
index 7166d2b5..c8e7433f 100644
--- a/lib/lattice/Grid_lattice_transpose.h
+++ b/lib/lattice/Grid_lattice_transpose.h
@@ -13,7 +13,7 @@ namespace Grid {
   template<class vobj>
     inline Lattice<vobj> transpose(const Lattice<vobj> &lhs){
         Lattice<vobj> ret(lhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){
             ret._odata[ss] = transpose(lhs._odata[ss]);
         }
@@ -28,7 +28,7 @@ namespace Grid {
       -> Lattice<decltype(transposeIndex<Index>(lhs._odata[0]))>
     {
       Lattice<decltype(transposeIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){
             ret._odata[ss] = transposeIndex<Index>(lhs._odata[ss]);
         }
diff --git a/lib/Grid_where.h b/lib/lattice/Grid_lattice_where.h
similarity index 95%
rename from lib/Grid_where.h
rename to lib/lattice/Grid_lattice_where.h
index 588c0596..84d7471c 100644
--- a/lib/Grid_where.h
+++ b/lib/lattice/Grid_lattice_where.h
@@ -1,5 +1,5 @@
-#ifndef GRID_WHERE_H
-#define GRID_WHERE_H
+#ifndef GRID_LATTICE_WHERE_H
+#define GRID_LATTICE_WHERE_H
 namespace Grid {
 // Must implement the predicate gating the 
 // Must be able to reduce the predicate down to a single vInteger per site.
@@ -27,7 +27,7 @@ inline void where(Lattice<vobj> &ret,const Lattice<iobj> &predicate,Lattice<vobj
   std::vector<scalar_object> truevals (Nsimd);
   std::vector<scalar_object> falsevals(Nsimd);
 
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
   for(int ss=0;ss<iftrue._grid->oSites(); ss++){
 
     extract(iftrue._odata[ss]   ,truevals);
diff --git a/lib/math/Grid_math_reality.h b/lib/math/Grid_math_reality.h
index 1bf34f2a..18a2f89b 100644
--- a/lib/math/Grid_math_reality.h
+++ b/lib/math/Grid_math_reality.h
@@ -6,23 +6,20 @@ namespace Grid {
     /////////////////////////////////////////// CONJ         ///////////////////////////////////////////
     ///////////////////////////////////////////////////////////////////////////////////////////////////
 
-#ifdef GRID_WARN_SUBOPTIMAL
-#warning "Optimisation alert switch over to two argument form to avoid copy back in perf critical timesI "
-#endif     
 /////////////////////////////////////////////// 
 // multiply by I; make recursive.
 /////////////////////////////////////////////// 
 template<class vtype> inline iScalar<vtype> timesI(const iScalar<vtype>&r) 
 {
     iScalar<vtype> ret;
-    ret._internal = timesI(r._internal);
+    timesI(ret._internal,r._internal);
     return ret;
 }
 template<class vtype,int N> inline iVector<vtype,N> timesI(const iVector<vtype,N>&r) 
 {
   iVector<vtype,N> ret;
   for(int i=0;i<N;i++){
-    ret._internal[i] = timesI(r._internal[i]);
+    timesI(ret._internal[i],r._internal[i]);
   }
   return ret;
 }
@@ -31,22 +28,41 @@ template<class vtype,int N> inline iMatrix<vtype,N> timesI(const iMatrix<vtype,N
   iMatrix<vtype,N> ret;
   for(int i=0;i<N;i++){
   for(int j=0;j<N;j++){
-    ret._internal[i][j] = timesI(r._internal[i][j]);
+    timesI(ret._internal[i][j],r._internal[i][j]);
   }}
   return ret;
 }
 
+template<class vtype> inline void timesI(iScalar<vtype> &ret,const iScalar<vtype>&r) 
+{
+  timesI(ret._internal,r._internal);
+}
+template<class vtype,int N> inline void timesI(iVector<vtype,N> &ret,const iVector<vtype,N>&r) 
+{
+  for(int i=0;i<N;i++){
+    timesI(ret._internal[i],r._internal[i]);
+  }
+}
+template<class vtype,int N> inline void  timesI(iMatrix<vtype,N> &ret,const iMatrix<vtype,N>&r)
+{
+  for(int i=0;i<N;i++){
+  for(int j=0;j<N;j++){
+    timesI(ret._internal[i][j],r._internal[i][j]);
+  }}
+}
+
+
 template<class vtype> inline iScalar<vtype> timesMinusI(const iScalar<vtype>&r) 
 {
     iScalar<vtype> ret;
-    ret._internal = timesMinusI(r._internal);
+    timesMinusI(ret._internal,r._internal);
     return ret;
 }
 template<class vtype,int N> inline iVector<vtype,N> timesMinusI(const iVector<vtype,N>&r) 
 {
   iVector<vtype,N> ret;
   for(int i=0;i<N;i++){
-    ret._internal[i] = timesMinusI(r._internal[i]);
+    timesMinusI(ret._internal[i],r._internal[i]);
   }
   return ret;
 }
@@ -55,10 +71,30 @@ template<class vtype,int N> inline iMatrix<vtype,N> timesMinusI(const iMatrix<vt
   iMatrix<vtype,N> ret;
   for(int i=0;i<N;i++){
   for(int j=0;j<N;j++){
-    ret._internal[i][j] = timesMinusI(r._internal[i][j]);
+    timesMinusI(ret._internal[i][j],r._internal[i][j]);
   }}
   return ret;
 }
+
+template<class vtype>  inline void timesMinusI(iScalar<vtype> &ret,const iScalar<vtype>&r) 
+{
+  timesMinusI(ret._internal,r._internal);
+}
+template<class vtype,int N> inline void timesMinusI(iVector<vtype,N> &ret,const iVector<vtype,N>&r) 
+{
+  for(int i=0;i<N;i++){
+    timesMinusI(ret._internal[i],r._internal[i]);
+  }
+}
+template<class vtype,int N> inline void  timesMinusI(iMatrix<vtype,N> &ret,const iMatrix<vtype,N>&r)
+{
+  for(int i=0;i<N;i++){
+  for(int j=0;j<N;j++){
+    timesMinusI(ret._internal[i][j],r._internal[i][j]);
+  }}
+}
+
+
 /////////////////////////////////////////////// 
 // Conj function for scalar, vector, matrix
 /////////////////////////////////////////////// 
diff --git a/lib/qcd/Grid_qcd_2spinor.h b/lib/qcd/Grid_qcd_2spinor.h
index c19264f5..fe6a1da8 100644
--- a/lib/qcd/Grid_qcd_2spinor.h
+++ b/lib/qcd/Grid_qcd_2spinor.h
@@ -1,9 +1,7 @@
 #ifndef GRID_QCD_TWOSPIN_H
 #define GRID_QCD_TWOSPIN_H
 namespace Grid{
-
 namespace QCD {
-  
 
   //////////////////////////////////////////////////////////////////////////////////////////////////////
   // Normalisation alert; the g5   project is 1/2(1+-G5) 
@@ -1073,10 +1071,6 @@ namespace QCD {
       spProj5m(hspin._internal[i][j],fspin._internal[i][j]);
     }}
   }
-
-
-
-
 }   //namespace QCD
 } // Grid
 #endif
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index 5054e9ce..f657208a 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -103,7 +103,7 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
   vHalfSpinColourVector *chi_p;
   int offset,local,perm, ptype;
 
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
   for(int sss=0;sss<grid->oSites();sss++){
 
     int ss = sss;
diff --git a/lib/simd/Grid_vComplexD.h b/lib/simd/Grid_vComplexD.h
index ef5da859..f0bc2da4 100644
--- a/lib/simd/Grid_vComplexD.h
+++ b/lib/simd/Grid_vComplexD.h
@@ -13,6 +13,9 @@ namespace Grid {
             vzero(*this);
             return (*this);
         }
+        vComplexD( Zero & z){
+            vzero(*this);
+        }
         vComplexD()=default;
         vComplexD(ComplexD a){
 	  vsplat(*this,a);
@@ -286,8 +289,8 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
             return ret;
         }
 
-        friend inline vComplexD timesMinusI(const vComplexD &in){
-	  vComplexD ret; vzero(ret);
+        friend inline void timesMinusI(vComplexD &ret,const vComplexD &in){
+	  vzero(ret);
 	  vComplexD tmp;
 #if defined (AVX1)|| defined (AVX2)
 	  tmp.v    =_mm256_addsub_pd(ret.v,in.v); // r,-i
@@ -304,11 +307,10 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
 #ifdef QPX
             assert(0);
 #endif
-	  return ret;
 	}
 
-        friend inline vComplexD timesI(const vComplexD &in){
-	  vComplexD ret; vzero(ret);
+        friend inline void timesI(vComplexD &ret, const vComplexD &in){
+	  vzero(ret);
 	  vComplexD tmp;
 #if defined (AVX1)|| defined (AVX2)
 	  tmp.v    =_mm256_shuffle_pd(in.v,in.v,0x5);
@@ -325,9 +327,21 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
 #ifdef QPX
             assert(0);
 #endif
+	}
+
+        friend inline vComplexD timesMinusI(const vComplexD &in){
+	  vComplexD ret; 
+	  timesMinusI(ret,in);
 	  return ret;
 	}
 
+        friend inline vComplexD timesI(const vComplexD &in){
+	  vComplexD ret; 
+	  timesI(ret,in);
+	  return ret;
+	}
+
+
 // REDUCE FIXME must be a cleaner implementation
        friend inline ComplexD Reduce(const vComplexD & in)
        { 
diff --git a/lib/simd/Grid_vComplexF.h b/lib/simd/Grid_vComplexF.h
index 34510333..202dce43 100644
--- a/lib/simd/Grid_vComplexF.h
+++ b/lib/simd/Grid_vComplexF.h
@@ -28,6 +28,9 @@ namespace Grid {
             vzero(*this);
             return (*this);
         }
+        vComplexF( Zero & z){
+            vzero(*this);
+        }
         vComplexF()=default;
         vComplexF(ComplexF a){
 	  vsplat(*this,a);
@@ -363,8 +366,7 @@ namespace Grid {
 #endif
             return ret;
         }
-        friend inline vComplexF timesMinusI(const vComplexF &in){
-	  vComplexF ret; 
+        friend inline void timesMinusI( vComplexF &ret,const vComplexF &in){
 	  vzero(ret);
 #if defined (AVX1)|| defined (AVX2)
 	  cvec tmp =_mm256_addsub_ps(ret.v,in.v); // r,-i
@@ -381,10 +383,9 @@ namespace Grid {
 #ifdef QPX
             assert(0);
 #endif
-	  return ret;
 	}
-        friend inline vComplexF timesI(const vComplexF &in){
-	  vComplexF ret; vzero(ret);
+        friend inline void timesI(vComplexF &ret,const vComplexF &in){
+	  vzero(ret);
 #if defined (AVX1)|| defined (AVX2)
 	  cvec tmp =_mm256_shuffle_ps(in.v,in.v,_MM_SHUFFLE(2,3,0,1));//i,r
           ret.v    =_mm256_addsub_ps(ret.v,tmp);     //i,-r
@@ -400,8 +401,18 @@ namespace Grid {
 #ifdef QPX
             assert(0);
 #endif
+	}
+        friend inline vComplexF timesMinusI(const vComplexF &in){
+	  vComplexF ret; 
+	  timesMinusI(ret,in);
 	  return ret;
 	}
+        friend inline vComplexF timesI(const vComplexF &in){
+	  vComplexF ret; 
+	  timesI(ret,in);
+	  return ret;
+	}
+
         
         // Unary negation
         friend inline vComplexF operator -(const vComplexF &r) {
diff --git a/lib/simd/Grid_vRealD.h b/lib/simd/Grid_vRealD.h
index 4d5e2f57..aef143f8 100644
--- a/lib/simd/Grid_vRealD.h
+++ b/lib/simd/Grid_vRealD.h
@@ -17,6 +17,10 @@ namespace Grid {
         vRealD(Zero &zero){
 	  zeroit(*this);
 	}
+        vRealD & operator = ( Zero & z){
+	  vzero(*this);
+	  return (*this);
+        }
 
         friend inline void mult(vRealD * __restrict__ y,const vRealD * __restrict__ l,const vRealD *__restrict__ r) {*y = (*l) * (*r);}
         friend inline void sub (vRealD * __restrict__ y,const vRealD * __restrict__ l,const vRealD *__restrict__ r) {*y = (*l) - (*r);}
diff --git a/lib/simd/Grid_vRealF.h b/lib/simd/Grid_vRealF.h
index 79e1b91f..559a9eda 100644
--- a/lib/simd/Grid_vRealF.h
+++ b/lib/simd/Grid_vRealF.h
@@ -18,6 +18,10 @@ namespace Grid {
         vRealF(Zero &zero){
 	  zeroit(*this);
 	}
+        vRealF & operator = ( Zero & z){
+	  vzero(*this);
+	  return (*this);
+        }
         ////////////////////////////////////
         // Arithmetic operator overloads +,-,*
         ////////////////////////////////////
diff --git a/tests/Grid_gamma.cc b/tests/Grid_gamma.cc
index ba298b19..f5582955 100644
--- a/tests/Grid_gamma.cc
+++ b/tests/Grid_gamma.cc
@@ -50,9 +50,8 @@ int main (int argc, char ** argv)
   //  std::cout << " Is triv Scalar<double> " <<std::has_trivial_default_constructor<iScalar<double> >::value << std::endl;
   //  std::cout << " Is triv Scalar<vComplexD> "<<std::has_trivial_default_constructor<iScalar<vComplexD> >::value  << std::endl;
 
-  std::complex<float> c(1.0);
   for(int a=0;a<Ns;a++){
-    ident()(a,a) = c;
+    ident()(a,a) = ComplexF(1.0);
   }
 
   const Gamma::GammaMatrix *g = Gamma::GammaMatrices;

From 556befaaaa98bae74fc3053eb3e54c8269f5fd28 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 12 May 2015 20:41:44 +0100
Subject: [PATCH 147/429] Enhanced SIMD interfacing

---
 benchmarks/Grid_comms.cc            |  2 +-
 benchmarks/Grid_memory_bandwidth.cc |  4 +---
 benchmarks/Grid_wilson.cc           |  2 +-
 lib/Grid.h                          |  2 +-
 lib/Grid_init.cc                    | 34 ++++++++++++++---------------
 lib/lattice/Grid_lattice_ET.h       |  8 +++----
 lib/lattice/Grid_lattice_arith.h    |  2 +-
 lib/lattice/Grid_lattice_base.h     | 14 +++++++-----
 lib/simd/Grid_vComplexF.h           |  6 ++---
 tests/Grid_cshift.cc                |  2 +-
 tests/Grid_gamma.cc                 |  2 +-
 tests/Grid_main.cc                  |  2 +-
 tests/Grid_nersc_io.cc              |  2 +-
 tests/Grid_simd.cc                  |  2 +-
 tests/Grid_stencil.cc               |  2 +-
 15 files changed, 43 insertions(+), 43 deletions(-)

diff --git a/benchmarks/Grid_comms.cc b/benchmarks/Grid_comms.cc
index 3dedfb5b..810cfd06 100644
--- a/benchmarks/Grid_comms.cc
+++ b/benchmarks/Grid_comms.cc
@@ -8,7 +8,7 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::vector<int> simd_layout = GridDefaultSimd();
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexD::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
 
   int Nloop=10;
diff --git a/benchmarks/Grid_memory_bandwidth.cc b/benchmarks/Grid_memory_bandwidth.cc
index cb444128..5a4d1f18 100644
--- a/benchmarks/Grid_memory_bandwidth.cc
+++ b/benchmarks/Grid_memory_bandwidth.cc
@@ -13,7 +13,7 @@ int main (int argc, char ** argv)
 
   int Nloop=1000;
 
-  std::vector<int> simd_layout = GridDefaultSimd();
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vReal::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
 
   std::cout << "===================================================================================================="<<std::endl;
@@ -54,7 +54,6 @@ int main (int argc, char ** argv)
   std::cout << "===================================================================================================="<<std::endl;
   std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s"<<std::endl;
   std::cout << "----------------------------------------------------------"<<std::endl;
-
   
   for(int lat=4;lat<=32;lat+=4){
 
@@ -69,7 +68,6 @@ int main (int argc, char ** argv)
       LatticeVec y(&Grid); //random(pRNG,y);
       double a=2.0;
 
-
       double start=usecond();
       for(int i=0;i<Nloop;i++){
 	z=a*x-y;
diff --git a/benchmarks/Grid_wilson.cc b/benchmarks/Grid_wilson.cc
index 6c6e9626..de31413f 100644
--- a/benchmarks/Grid_wilson.cc
+++ b/benchmarks/Grid_wilson.cc
@@ -22,7 +22,7 @@ int main (int argc, char ** argv)
 
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd();
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
 
diff --git a/lib/Grid.h b/lib/Grid.h
index 4512f443..bd7b371a 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -71,7 +71,7 @@ namespace Grid {
   // C++11 time facilities better?
   double usecond(void);
 
-  const std::vector<int> &GridDefaultSimd(void);
+  const std::vector<int> GridDefaultSimd(int dims,int nsimd);
   const std::vector<int> &GridDefaultLatt(void);
   const std::vector<int> &GridDefaultMpi(void);
   const int              &GridThreads(void)  ;
diff --git a/lib/Grid_init.cc b/lib/Grid_init.cc
index 9e977064..c804e810 100644
--- a/lib/Grid_init.cc
+++ b/lib/Grid_init.cc
@@ -27,14 +27,28 @@ namespace Grid {
   // Convenience functions to access stadard command line arg
   // driven parallelism controls
   //////////////////////////////////////////////////////
-  static std::vector<int> Grid_default_simd;
   static std::vector<int> Grid_default_latt;
   static std::vector<int> Grid_default_mpi;
 
   int GridThread::_threads;
 
+  const std::vector<int> GridDefaultSimd(int dims,int nsimd)
+  {
+    std::vector<int> layout(dims);
+    int nn=nsimd;
+    for(int d=dims-1;d>=0;d--){
+      if ( nn>=2) {
+	layout[d]=2;
+	nn/=2;
+      } else { 
+	layout[d]=1;
+      }
+    }
+    assert(nn==1);
+    return layout;
+  }
 
-  const std::vector<int> &GridDefaultSimd(void)     {return Grid_default_simd;};
+  
   const std::vector<int> &GridDefaultLatt(void)     {return Grid_default_latt;};
   const std::vector<int> &GridDefaultMpi(void)      {return Grid_default_mpi;};
 
@@ -71,22 +85,11 @@ void GridCmdOptionIntVector(std::string &str,std::vector<int> & vec)
 
 void GridParseLayout(char **argv,int argc,
 		     std::vector<int> &latt,
-		     std::vector<int> &simd,
 		     std::vector<int> &mpi)
 {
   mpi =std::vector<int>({1,1,1,1});
   latt=std::vector<int>({8,8,8,8});
 
-#if defined(SSE4)
-  simd=std::vector<int>({1,1,1,2});
-#endif
-#if defined(AVX1) || defined (AVX2)
-  simd=std::vector<int>({1,1,2,2});
-#endif
-#if defined(AVX512)
-  simd=std::vector<int>({1,2,2,2});
-#endif
-
   GridThread::SetMaxThreads();
 
   std::string arg;
@@ -94,10 +97,6 @@ void GridParseLayout(char **argv,int argc,
     arg = GridCmdOptionPayload(argv,argv+argc,"--mpi");
     GridCmdOptionIntVector(arg,mpi);
   }
-  if( GridCmdOptionExists(argv,argv+argc,"--simd") ){
-    arg= GridCmdOptionPayload(argv,argv+argc,"--simd");
-    GridCmdOptionIntVector(arg,simd);
-  }
   if( GridCmdOptionExists(argv,argv+argc,"--grid") ){
     arg= GridCmdOptionPayload(argv,argv+argc,"--grid");
     GridCmdOptionIntVector(arg,latt);
@@ -129,7 +128,6 @@ void Grid_init(int *argc,char ***argv)
   }
   GridParseLayout(*argv,*argc,
 		  Grid_default_latt,
-		  Grid_default_simd,
 		  Grid_default_mpi);
 
 }
diff --git a/lib/lattice/Grid_lattice_ET.h b/lib/lattice/Grid_lattice_ET.h
index d4fd7d05..c0b5f97d 100644
--- a/lib/lattice/Grid_lattice_ET.h
+++ b/lib/lattice/Grid_lattice_ET.h
@@ -67,6 +67,10 @@ inline void GridFromExpression(GridBase * &grid,const T1& lat)   // Lattice leaf
   } 
   grid=lat._grid;
 }
+template<class T1,typename std::enable_if<!is_lattice<T1>::value, T1>::type * = nullptr >
+inline void GridFromExpression(GridBase * &grid,const T1& notlat)   // non-lattice leaf
+{
+}
 template <typename Op, typename T1>
 inline void GridFromExpression(GridBase * &grid,const LatticeUnaryExpression<Op,T1 > &expr)
 {
@@ -86,10 +90,6 @@ inline void GridFromExpression( GridBase * &grid,const LatticeTrinaryExpression<
   GridFromExpression(grid,std::get<1>(expr.second));
   GridFromExpression(grid,std::get<2>(expr.second));
 }
-template<class T1,typename std::enable_if<!is_lattice<T1>::value, T1>::type * = nullptr >
-inline void GridFromExpression(GridBase * &grid,const T1& notlat)   // non-lattice leaf
-{
-}
 
 ////////////////////////////////////////////
 // Unary operators and funcs
diff --git a/lib/lattice/Grid_lattice_arith.h b/lib/lattice/Grid_lattice_arith.h
index c1d58b86..10f31025 100644
--- a/lib/lattice/Grid_lattice_arith.h
+++ b/lib/lattice/Grid_lattice_arith.h
@@ -145,7 +145,7 @@ PARALLEL_FOR_LOOP
   template<class sobj,class vobj>
   inline void axpy(Lattice<vobj> &ret,sobj a,const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
     conformable(lhs,rhs);
-PARALLEL_FOR_LOOP
+#pragma omp parallel for
     for(int ss=0;ss<lhs._grid->oSites();ss++){
       vobj tmp = a*lhs._odata[ss];
       vstream(ret._odata[ss],tmp+rhs._odata[ss]);
diff --git a/lib/lattice/Grid_lattice_base.h b/lib/lattice/Grid_lattice_base.h
index cd376b32..ace7565d 100644
--- a/lib/lattice/Grid_lattice_base.h
+++ b/lib/lattice/Grid_lattice_base.h
@@ -64,7 +64,8 @@ public:
   ////////////////////////////////////////////////////////////////////////////////
   template <typename Op, typename T1>                         inline Lattice<vobj> & operator=(const LatticeUnaryExpression<Op,T1> &expr)
   {
-PARALLEL_FOR_LOOP
+    //PARALLEL_FOR_LOOP
+#pragma omp parallel for
     for(int ss=0;ss<_grid->oSites();ss++){
       vobj tmp= eval(ss,expr);
       vstream(_odata[ss] ,tmp);
@@ -73,7 +74,8 @@ PARALLEL_FOR_LOOP
   }
   template <typename Op, typename T1,typename T2>             inline Lattice<vobj> & operator=(const LatticeBinaryExpression<Op,T1,T2> &expr)
   {
-PARALLEL_FOR_LOOP
+    // PARALLEL_FOR_LOOP
+#pragma omp parallel for
     for(int ss=0;ss<_grid->oSites();ss++){
       vobj tmp= eval(ss,expr);
       vstream(_odata[ss] ,tmp);
@@ -82,7 +84,8 @@ PARALLEL_FOR_LOOP
   }
   template <typename Op, typename T1,typename T2,typename T3> inline Lattice<vobj> & operator=(const LatticeTrinaryExpression<Op,T1,T2,T3> &expr)
   {
-PARALLEL_FOR_LOOP
+    //PARALLEL_FOR_LOOP
+#pragma omp parallel for
     for(int ss=0;ss<_grid->oSites();ss++){
       vobj tmp= eval(ss,expr);
       vstream(_odata[ss] ,tmp);
@@ -176,15 +179,16 @@ PARALLEL_FOR_LOOP
  }; // class Lattice
 }
 
-#undef GRID_LATTICE_EXPRESSION_TEMPLATES
 
 #include <lattice/Grid_lattice_conformable.h>
 
-#ifdef GRID_LATTICE_EXPRESSION_TEMPLATES
+#define GRID_LATTICE_EXPRESSION_TEMPLATES
+#ifdef  GRID_LATTICE_EXPRESSION_TEMPLATES
 #include <lattice/Grid_lattice_ET.h>
 #else 
 #include <lattice/Grid_lattice_overload.h>
 #endif
+
 #include <lattice/Grid_lattice_arith.h>
 
 #include <lattice/Grid_lattice_trace.h>
diff --git a/lib/simd/Grid_vComplexF.h b/lib/simd/Grid_vComplexF.h
index 202dce43..cd757916 100644
--- a/lib/simd/Grid_vComplexF.h
+++ b/lib/simd/Grid_vComplexF.h
@@ -28,9 +28,9 @@ namespace Grid {
             vzero(*this);
             return (*this);
         }
-        vComplexF( Zero & z){
-            vzero(*this);
-        }
+	//        vComplexF( Zero & z){
+	//            vzero(*this);
+	//        }
         vComplexF()=default;
         vComplexF(ComplexF a){
 	  vsplat(*this,a);
diff --git a/tests/Grid_cshift.cc b/tests/Grid_cshift.cc
index e92ff793..d0c83e30 100644
--- a/tests/Grid_cshift.cc
+++ b/tests/Grid_cshift.cc
@@ -9,7 +9,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd();
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
 
   GridCartesian        Fine(latt_size,simd_layout,mpi_layout);
diff --git a/tests/Grid_gamma.cc b/tests/Grid_gamma.cc
index f5582955..e803029b 100644
--- a/tests/Grid_gamma.cc
+++ b/tests/Grid_gamma.cc
@@ -15,7 +15,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd();
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
      
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
diff --git a/tests/Grid_main.cc b/tests/Grid_main.cc
index c3454dfc..7515ebd1 100644
--- a/tests/Grid_main.cc
+++ b/tests/Grid_main.cc
@@ -26,7 +26,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd();
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
 
   latt_size.resize(4);
diff --git a/tests/Grid_nersc_io.cc b/tests/Grid_nersc_io.cc
index 73d7edfe..fbef3cb1 100644
--- a/tests/Grid_nersc_io.cc
+++ b/tests/Grid_nersc_io.cc
@@ -11,7 +11,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
 
-  std::vector<int> simd_layout = GridDefaultSimd();
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
   std::vector<int> latt_size  ({16,16,16,32});
   std::vector<int> clatt_size  ({4,4,4,8});
diff --git a/tests/Grid_simd.cc b/tests/Grid_simd.cc
index 6a957d7e..32f4cdfb 100644
--- a/tests/Grid_simd.cc
+++ b/tests/Grid_simd.cc
@@ -107,7 +107,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd();
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
     
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
diff --git a/tests/Grid_stencil.cc b/tests/Grid_stencil.cc
index d9e779bf..1fdb7265 100644
--- a/tests/Grid_stencil.cc
+++ b/tests/Grid_stencil.cc
@@ -10,7 +10,7 @@ int main (int argc, char ** argv)
 
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd();
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
 
   double volume = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];

From 541d52ab9754db9bfd1eaeef537d45549455aa4f Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 13 May 2015 00:31:00 +0100
Subject: [PATCH 148/429] I have made the Cshift work successfully with open mp
 threading in every routine. Collapse(2) is now working under clang-omp++.

---
 Makefile.in                     |    8 +-
 TODO                            |   23 +-
 aclocal.m4                      |   64 --
 benchmarks/Grid_wilson.cc       |    2 +-
 configure                       | 1004 ++++---------------------------
 configure-commands              |    1 +
 configure.ac                    |    1 +
 lib/Grid.h                      |   10 +-
 lib/Grid_stencil.h              |    6 +-
 lib/Grid_threads.h              |   16 +-
 lib/cshift/Grid_cshift_common.h |   26 +-
 lib/cshift/Grid_cshift_mpi.h    |    4 -
 lib/qcd/Grid_qcd_wilson_dop.cc  |    7 +-
 13 files changed, 166 insertions(+), 1006 deletions(-)

diff --git a/Makefile.in b/Makefile.in
index c9257ab2..c9b848f3 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -205,12 +205,9 @@ AUTOCONF = @AUTOCONF@
 AUTOHEADER = @AUTOHEADER@
 AUTOMAKE = @AUTOMAKE@
 AWK = @AWK@
-CC = @CC@
-CCDEPMODE = @CCDEPMODE@
-CFLAGS = @CFLAGS@
-CPP = @CPP@
 CPPFLAGS = @CPPFLAGS@
 CXX = @CXX@
+CXXCPP = @CXXCPP@
 CXXDEPMODE = @CXXDEPMODE@
 CXXFLAGS = @CXXFLAGS@
 CYGPATH_W = @CYGPATH_W@
@@ -234,7 +231,7 @@ LTLIBOBJS = @LTLIBOBJS@
 MAKEINFO = @MAKEINFO@
 MKDIR_P = @MKDIR_P@
 OBJEXT = @OBJEXT@
-OPENMP_CFLAGS = @OPENMP_CFLAGS@
+OPENMP_CXXFLAGS = @OPENMP_CXXFLAGS@
 PACKAGE = @PACKAGE@
 PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
 PACKAGE_NAME = @PACKAGE_NAME@
@@ -252,7 +249,6 @@ abs_builddir = @abs_builddir@
 abs_srcdir = @abs_srcdir@
 abs_top_builddir = @abs_top_builddir@
 abs_top_srcdir = @abs_top_srcdir@
-ac_ct_CC = @ac_ct_CC@
 ac_ct_CXX = @ac_ct_CXX@
 am__include = @am__include@
 am__leading_dot = @am__leading_dot@
diff --git a/TODO b/TODO
index 6a1624f7..b596c605 100644
--- a/TODO
+++ b/TODO
@@ -2,10 +2,7 @@
 *** Hacks and bug fixes to clean up and Audits
 ================================================================
 * Base class to share common code between vRealF, VComplexF etc... 
-
-* Performance check on Guido's reimplementation strategy
-
-* Bug in SeedFixedIntegers gives same output on each site. -- Think I fixed but NOT checked for sure
+  - Performance check on Guido's reimplementation strategy
 
 * FIXME audit
 * const audit
@@ -20,15 +17,16 @@
 *** New Functionality
 ================================================================
 
-* Implement where to take template scheme.
+* Implement where within expression template scheme.
 
 * - BinaryWriter, TextWriter etc...
   - use protocol buffers? replace xmlReader/Writer ec..
   - Binary use htonll, htonl
 
-* Expression template engine:
+* Bug in SeedFixedIntegers gave same output on each site. -- Think I fixed but NOT checked for sure
+  Implement and use lattice IO to verify this.
 
-   -- Audit
+* Expression template engine: -- DONE
    -- Norm2(expression) problem: introduce norm2 unary op, or Introduce conversion automatic from expression to Lattice<vobj>
 
 * CovariantShift support                             -----Use a class to store gauge field? (parallel transport?)
@@ -56,9 +54,10 @@
 * TaProj
 * FFTnD ?
 
-* Parallel MPI2 IO
-  Plaquette checks into nersc reader.
-
+* Parallel io improvements
+  - optional parallel MPI2 IO
+  - move Plaquette and link trace checks into nersc reader from the Grid_nersc_io.cc test.
+  
 * rb4d support for 5th dimension in Mobius.
 
 * Check for missing functionality                    - partially audited against QDP++ layout
@@ -70,13 +69,15 @@
    // copyMask.
    // localMaxAbs
    // Fourier transform equivalent.
-Actions
+
+Actions -- coherent framework for implementing actions and their forces.
 * Fermion
   - Wilson
   - Clover
   - DomainWall
   - Mobius
   - z-Mobius
+
 * Gauge
   - Wilson, symanzik, iwasaki
 
diff --git a/aclocal.m4 b/aclocal.m4
index f3018f6c..bf79d078 100644
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -765,70 +765,6 @@ AC_DEFUN([_AM_SET_OPTIONS],
 AC_DEFUN([_AM_IF_OPTION],
 [m4_ifset(_AM_MANGLE_OPTION([$1]), [$2], [$3])])
 
-# Copyright (C) 1999-2014 Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# _AM_PROG_CC_C_O
-# ---------------
-# Like AC_PROG_CC_C_O, but changed for automake.  We rewrite AC_PROG_CC
-# to automatically call this.
-AC_DEFUN([_AM_PROG_CC_C_O],
-[AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
-AC_REQUIRE_AUX_FILE([compile])dnl
-AC_LANG_PUSH([C])dnl
-AC_CACHE_CHECK(
-  [whether $CC understands -c and -o together],
-  [am_cv_prog_cc_c_o],
-  [AC_LANG_CONFTEST([AC_LANG_PROGRAM([])])
-  # Make sure it works both with $CC and with simple cc.
-  # Following AC_PROG_CC_C_O, we do the test twice because some
-  # compilers refuse to overwrite an existing .o file with -o,
-  # though they will create one.
-  am_cv_prog_cc_c_o=yes
-  for am_i in 1 2; do
-    if AM_RUN_LOG([$CC -c conftest.$ac_ext -o conftest2.$ac_objext]) \
-         && test -f conftest2.$ac_objext; then
-      : OK
-    else
-      am_cv_prog_cc_c_o=no
-      break
-    fi
-  done
-  rm -f core conftest*
-  unset am_i])
-if test "$am_cv_prog_cc_c_o" != yes; then
-   # Losing compiler, so override with the script.
-   # FIXME: It is wrong to rewrite CC.
-   # But if we don't then we get into trouble of one sort or another.
-   # A longer-term fix would be to have automake use am__CC in this case,
-   # and then we could set am__CC="\$(top_srcdir)/compile \$(CC)"
-   CC="$am_aux_dir/compile $CC"
-fi
-AC_LANG_POP([C])])
-
-# For backward compatibility.
-AC_DEFUN_ONCE([AM_PROG_CC_C_O], [AC_REQUIRE([AC_PROG_CC])])
-
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# AM_RUN_LOG(COMMAND)
-# -------------------
-# Run COMMAND, save the exit status in ac_status, and log it.
-# (This has been adapted from Autoconf's _AC_RUN_LOG macro.)
-AC_DEFUN([AM_RUN_LOG],
-[{ echo "$as_me:$LINENO: $1" >&AS_MESSAGE_LOG_FD
-   ($1) >&AS_MESSAGE_LOG_FD 2>&AS_MESSAGE_LOG_FD
-   ac_status=$?
-   echo "$as_me:$LINENO: \$? = $ac_status" >&AS_MESSAGE_LOG_FD
-   (exit $ac_status); }])
-
 # Check to make sure that the build environment is sane.    -*- Autoconf -*-
 
 # Copyright (C) 1996-2014 Free Software Foundation, Inc.
diff --git a/benchmarks/Grid_wilson.cc b/benchmarks/Grid_wilson.cc
index de31413f..9ce13090 100644
--- a/benchmarks/Grid_wilson.cc
+++ b/benchmarks/Grid_wilson.cc
@@ -77,7 +77,7 @@ int main (int argc, char ** argv)
   WilsonMatrix Dw(Umu,mass);
   
   std::cout << "Calling Dw"<<std::endl;
-  int ncall=10000;
+  int ncall=1000;
   double t0=usecond();
   for(int i=0;i<ncall;i++){
     Dw.multiply(src,result);
diff --git a/configure b/configure
index 3d25f6a9..d5071cd1 100755
--- a/configure
+++ b/configure
@@ -632,15 +632,9 @@ BUILD_COMMS_MPI_FALSE
 BUILD_COMMS_MPI_TRUE
 EGREP
 GREP
-CPP
+CXXCPP
 RANLIB
-OPENMP_CFLAGS
-am__fastdepCC_FALSE
-am__fastdepCC_TRUE
-CCDEPMODE
-ac_ct_CC
-CFLAGS
-CC
+OPENMP_CXXFLAGS
 am__fastdepCXX_FALSE
 am__fastdepCXX_TRUE
 CXXDEPMODE
@@ -741,9 +735,7 @@ LDFLAGS
 LIBS
 CPPFLAGS
 CCC
-CC
-CFLAGS
-CPP'
+CXXCPP'
 
 
 # Initialize some variables set by options.
@@ -1378,9 +1370,7 @@ Some influential environment variables:
   LIBS        libraries to pass to the linker, e.g. -l<library>
   CPPFLAGS    (Objective) C/C++ preprocessor flags, e.g. -I<include dir> if
               you have headers in a nonstandard directory <include dir>
-  CC          C compiler command
-  CFLAGS      C compiler flags
-  CPP         C preprocessor
+  CXXCPP      C++ preprocessor
 
 Use these variables to override the choices made by `configure' or to help
 it to find libraries and programs with nonstandard names/locations.
@@ -1500,48 +1490,10 @@ fi
 
 } # ac_fn_cxx_try_compile
 
-# ac_fn_c_try_compile LINENO
-# --------------------------
-# Try to compile conftest.$ac_ext, and return whether this succeeded.
-ac_fn_c_try_compile ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  rm -f conftest.$ac_objext
-  if { { ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compile") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_retval=1
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_c_try_compile
-
-# ac_fn_c_try_link LINENO
-# -----------------------
+# ac_fn_cxx_try_link LINENO
+# -------------------------
 # Try to link conftest.$ac_ext, and return whether this succeeded.
-ac_fn_c_try_link ()
+ac_fn_cxx_try_link ()
 {
   as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
   rm -f conftest.$ac_objext conftest$ac_exeext
@@ -1561,7 +1513,7 @@ $as_echo "$ac_try_echo"; } >&5
   fi
   $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
   test $ac_status = 0; } && {
-	 test -z "$ac_c_werror_flag" ||
+	 test -z "$ac_cxx_werror_flag" ||
 	 test ! -s conftest.err
        } && test -s conftest$ac_exeext && {
 	 test "$cross_compiling" = yes ||
@@ -1582,12 +1534,12 @@ fi
   eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
   as_fn_set_status $ac_retval
 
-} # ac_fn_c_try_link
+} # ac_fn_cxx_try_link
 
-# ac_fn_c_try_cpp LINENO
-# ----------------------
+# ac_fn_cxx_try_cpp LINENO
+# ------------------------
 # Try to preprocess conftest.$ac_ext, and return whether this succeeded.
-ac_fn_c_try_cpp ()
+ac_fn_cxx_try_cpp ()
 {
   as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
   if { { ac_try="$ac_cpp conftest.$ac_ext"
@@ -1606,7 +1558,7 @@ $as_echo "$ac_try_echo"; } >&5
   fi
   $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
   test $ac_status = 0; } > conftest.i && {
-	 test -z "$ac_c_preproc_warn_flag$ac_c_werror_flag" ||
+	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
 	 test ! -s conftest.err
        }; then :
   ac_retval=0
@@ -1619,14 +1571,14 @@ fi
   eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
   as_fn_set_status $ac_retval
 
-} # ac_fn_c_try_cpp
+} # ac_fn_cxx_try_cpp
 
-# ac_fn_c_check_header_mongrel LINENO HEADER VAR INCLUDES
-# -------------------------------------------------------
+# ac_fn_cxx_check_header_mongrel LINENO HEADER VAR INCLUDES
+# ---------------------------------------------------------
 # Tests whether HEADER exists, giving a warning if it cannot be compiled using
 # the include files in INCLUDES and setting the cache variable VAR
 # accordingly.
-ac_fn_c_check_header_mongrel ()
+ac_fn_cxx_check_header_mongrel ()
 {
   as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
   if eval \${$3+:} false; then :
@@ -1647,7 +1599,7 @@ cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 $4
 #include <$2>
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+if ac_fn_cxx_try_compile "$LINENO"; then :
   ac_header_compiler=yes
 else
   ac_header_compiler=no
@@ -1663,7 +1615,7 @@ cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <$2>
 _ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
+if ac_fn_cxx_try_cpp "$LINENO"; then :
   ac_header_preproc=yes
 else
   ac_header_preproc=no
@@ -1673,7 +1625,7 @@ rm -f conftest.err conftest.i conftest.$ac_ext
 $as_echo "$ac_header_preproc" >&6; }
 
 # So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_c_preproc_warn_flag in #((
+case $ac_header_compiler:$ac_header_preproc:$ac_cxx_preproc_warn_flag in #((
   yes:no: )
     { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&5
 $as_echo "$as_me: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&2;}
@@ -1710,13 +1662,13 @@ $as_echo "$ac_res" >&6; }
 fi
   eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
 
-} # ac_fn_c_check_header_mongrel
+} # ac_fn_cxx_check_header_mongrel
 
-# ac_fn_c_try_run LINENO
-# ----------------------
+# ac_fn_cxx_try_run LINENO
+# ------------------------
 # Try to link conftest.$ac_ext, and return whether this succeeded. Assumes
 # that executables *can* be run.
-ac_fn_c_try_run ()
+ac_fn_cxx_try_run ()
 {
   as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
   if { { ac_try="$ac_link"
@@ -1752,13 +1704,13 @@ fi
   eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
   as_fn_set_status $ac_retval
 
-} # ac_fn_c_try_run
+} # ac_fn_cxx_try_run
 
-# ac_fn_c_check_header_compile LINENO HEADER VAR INCLUDES
-# -------------------------------------------------------
+# ac_fn_cxx_check_header_compile LINENO HEADER VAR INCLUDES
+# ---------------------------------------------------------
 # Tests whether HEADER exists and can be compiled using the include files in
 # INCLUDES, setting the cache variable VAR accordingly.
-ac_fn_c_check_header_compile ()
+ac_fn_cxx_check_header_compile ()
 {
   as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
   { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
@@ -1771,7 +1723,7 @@ else
 $4
 #include <$2>
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+if ac_fn_cxx_try_compile "$LINENO"; then :
   eval "$3=yes"
 else
   eval "$3=no"
@@ -1783,13 +1735,13 @@ eval ac_res=\$$3
 $as_echo "$ac_res" >&6; }
   eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
 
-} # ac_fn_c_check_header_compile
+} # ac_fn_cxx_check_header_compile
 
-# ac_fn_c_check_decl LINENO SYMBOL VAR INCLUDES
-# ---------------------------------------------
+# ac_fn_cxx_check_decl LINENO SYMBOL VAR INCLUDES
+# -----------------------------------------------
 # Tests whether SYMBOL is declared in INCLUDES, setting cache variable VAR
 # accordingly.
-ac_fn_c_check_decl ()
+ac_fn_cxx_check_decl ()
 {
   as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
   as_decl_name=`echo $2|sed 's/ *(.*//'`
@@ -1817,7 +1769,7 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+if ac_fn_cxx_try_compile "$LINENO"; then :
   eval "$3=yes"
 else
   eval "$3=no"
@@ -1829,13 +1781,13 @@ eval ac_res=\$$3
 $as_echo "$ac_res" >&6; }
   eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
 
-} # ac_fn_c_check_decl
+} # ac_fn_cxx_check_decl
 
-# ac_fn_c_check_type LINENO TYPE VAR INCLUDES
-# -------------------------------------------
+# ac_fn_cxx_check_type LINENO TYPE VAR INCLUDES
+# ---------------------------------------------
 # Tests whether TYPE exists after having included INCLUDES, setting cache
 # variable VAR accordingly.
-ac_fn_c_check_type ()
+ac_fn_cxx_check_type ()
 {
   as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
   { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
@@ -1856,7 +1808,7 @@ if (sizeof ($2))
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+if ac_fn_cxx_try_compile "$LINENO"; then :
   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 $4
@@ -1869,7 +1821,7 @@ if (sizeof (($2)))
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+if ac_fn_cxx_try_compile "$LINENO"; then :
 
 else
   eval "$3=yes"
@@ -1883,7 +1835,7 @@ eval ac_res=\$$3
 $as_echo "$ac_res" >&6; }
   eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
 
-} # ac_fn_c_check_type
+} # ac_fn_cxx_check_type
 
 # ac_fn_c_find_uintX_t LINENO BITS VAR
 # ------------------------------------
@@ -1916,7 +1868,7 @@ return test_array [0];
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+if ac_fn_cxx_try_compile "$LINENO"; then :
   case $ac_type in #(
   uint$2_t) :
     eval "$3=yes" ;; #(
@@ -1939,10 +1891,10 @@ $as_echo "$ac_res" >&6; }
 
 } # ac_fn_c_find_uintX_t
 
-# ac_fn_c_check_func LINENO FUNC VAR
-# ----------------------------------
+# ac_fn_cxx_check_func LINENO FUNC VAR
+# ------------------------------------
 # Tests whether FUNC exists, setting the cache variable VAR accordingly
-ac_fn_c_check_func ()
+ac_fn_cxx_check_func ()
 {
   as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
   { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
@@ -1991,7 +1943,7 @@ return $2 ();
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_link "$LINENO"; then :
+if ac_fn_cxx_try_link "$LINENO"; then :
   eval "$3=yes"
 else
   eval "$3=no"
@@ -2004,7 +1956,7 @@ eval ac_res=\$$3
 $as_echo "$ac_res" >&6; }
   eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
 
-} # ac_fn_c_check_func
+} # ac_fn_cxx_check_func
 cat >config.log <<_ACEOF
 This file contains any messages produced by compilers while
 running configure, to aid debugging if configure makes a mistake.
@@ -2971,6 +2923,12 @@ ac_config_headers="$ac_config_headers lib/Grid_config.h"
 
 
 # Checks for programs.
+ac_ext=cpp
+ac_cpp='$CXXCPP $CPPFLAGS'
+ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
+
 ac_ext=cpp
 ac_cpp='$CXXCPP $CPPFLAGS'
 ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
@@ -3472,11 +3430,11 @@ else
     CXXFLAGS=
   fi
 fi
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
+ac_ext=cpp
+ac_cpp='$CXXCPP $CPPFLAGS'
+ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
 DEPDIR="${am__leading_dot}deps"
 
 ac_config_commands="$ac_config_commands depfiles"
@@ -3669,744 +3627,18 @@ else
 fi
 
 
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-if test -n "$ac_tool_prefix"; then
-  # Extract the first word of "${ac_tool_prefix}gcc", so it can be a program name with args.
-set dummy ${ac_tool_prefix}gcc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$CC"; then
-  ac_cv_prog_CC="$CC" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_CC="${ac_tool_prefix}gcc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-CC=$ac_cv_prog_CC
-if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
-$as_echo "$CC" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
 
 
-fi
-if test -z "$ac_cv_prog_CC"; then
-  ac_ct_CC=$CC
-  # Extract the first word of "gcc", so it can be a program name with args.
-set dummy gcc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_CC+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$ac_ct_CC"; then
-  ac_cv_prog_ac_ct_CC="$ac_ct_CC" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_ac_ct_CC="gcc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-ac_ct_CC=$ac_cv_prog_ac_ct_CC
-if test -n "$ac_ct_CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
-$as_echo "$ac_ct_CC" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-  if test "x$ac_ct_CC" = x; then
-    CC=""
-  else
-    case $cross_compiling:$ac_tool_warned in
-yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
-$as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
-ac_tool_warned=yes ;;
-esac
-    CC=$ac_ct_CC
-  fi
-else
-  CC="$ac_cv_prog_CC"
-fi
-
-if test -z "$CC"; then
-          if test -n "$ac_tool_prefix"; then
-    # Extract the first word of "${ac_tool_prefix}cc", so it can be a program name with args.
-set dummy ${ac_tool_prefix}cc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$CC"; then
-  ac_cv_prog_CC="$CC" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_CC="${ac_tool_prefix}cc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-CC=$ac_cv_prog_CC
-if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
-$as_echo "$CC" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-
-  fi
-fi
-if test -z "$CC"; then
-  # Extract the first word of "cc", so it can be a program name with args.
-set dummy cc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$CC"; then
-  ac_cv_prog_CC="$CC" # Let the user override the test.
-else
-  ac_prog_rejected=no
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    if test "$as_dir/$ac_word$ac_exec_ext" = "/usr/ucb/cc"; then
-       ac_prog_rejected=yes
-       continue
-     fi
-    ac_cv_prog_CC="cc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-if test $ac_prog_rejected = yes; then
-  # We found a bogon in the path, so make sure we never use it.
-  set dummy $ac_cv_prog_CC
-  shift
-  if test $# != 0; then
-    # We chose a different compiler from the bogus one.
-    # However, it has the same basename, so the bogon will be chosen
-    # first if we set CC to just the basename; use the full file name.
-    shift
-    ac_cv_prog_CC="$as_dir/$ac_word${1+' '}$@"
-  fi
-fi
-fi
-fi
-CC=$ac_cv_prog_CC
-if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
-$as_echo "$CC" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-
-fi
-if test -z "$CC"; then
-  if test -n "$ac_tool_prefix"; then
-  for ac_prog in cl.exe
-  do
-    # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
-set dummy $ac_tool_prefix$ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$CC"; then
-  ac_cv_prog_CC="$CC" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_CC="$ac_tool_prefix$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-CC=$ac_cv_prog_CC
-if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
-$as_echo "$CC" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-
-    test -n "$CC" && break
-  done
-fi
-if test -z "$CC"; then
-  ac_ct_CC=$CC
-  for ac_prog in cl.exe
-do
-  # Extract the first word of "$ac_prog", so it can be a program name with args.
-set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_CC+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$ac_ct_CC"; then
-  ac_cv_prog_ac_ct_CC="$ac_ct_CC" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_ac_ct_CC="$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-ac_ct_CC=$ac_cv_prog_ac_ct_CC
-if test -n "$ac_ct_CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
-$as_echo "$ac_ct_CC" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-
-  test -n "$ac_ct_CC" && break
-done
-
-  if test "x$ac_ct_CC" = x; then
-    CC=""
-  else
-    case $cross_compiling:$ac_tool_warned in
-yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
-$as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
-ac_tool_warned=yes ;;
-esac
-    CC=$ac_ct_CC
-  fi
-fi
-
-fi
-
-
-test -z "$CC" && { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "no acceptable C compiler found in \$PATH
-See \`config.log' for more details" "$LINENO" 5; }
-
-# Provide some information about the compiler.
-$as_echo "$as_me:${as_lineno-$LINENO}: checking for C compiler version" >&5
-set X $ac_compile
-ac_compiler=$2
-for ac_option in --version -v -V -qversion; do
-  { { ac_try="$ac_compiler $ac_option >&5"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    sed '10a\
-... rest of stderr output deleted ...
-         10q' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-  fi
-  rm -f conftest.er1 conftest.err
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }
-done
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C compiler" >&5
-$as_echo_n "checking whether we are using the GNU C compiler... " >&6; }
-if ${ac_cv_c_compiler_gnu+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-#ifndef __GNUC__
-       choke me
-#endif
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  ac_compiler_gnu=yes
-else
-  ac_compiler_gnu=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-ac_cv_c_compiler_gnu=$ac_compiler_gnu
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_c_compiler_gnu" >&5
-$as_echo "$ac_cv_c_compiler_gnu" >&6; }
-if test $ac_compiler_gnu = yes; then
-  GCC=yes
-else
-  GCC=
-fi
-ac_test_CFLAGS=${CFLAGS+set}
-ac_save_CFLAGS=$CFLAGS
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC accepts -g" >&5
-$as_echo_n "checking whether $CC accepts -g... " >&6; }
-if ${ac_cv_prog_cc_g+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  ac_save_c_werror_flag=$ac_c_werror_flag
-   ac_c_werror_flag=yes
-   ac_cv_prog_cc_g=no
-   CFLAGS="-g"
-   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  ac_cv_prog_cc_g=yes
-else
-  CFLAGS=""
-      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-
-else
-  ac_c_werror_flag=$ac_save_c_werror_flag
-	 CFLAGS="-g"
-	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  ac_cv_prog_cc_g=yes
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-   ac_c_werror_flag=$ac_save_c_werror_flag
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_g" >&5
-$as_echo "$ac_cv_prog_cc_g" >&6; }
-if test "$ac_test_CFLAGS" = set; then
-  CFLAGS=$ac_save_CFLAGS
-elif test $ac_cv_prog_cc_g = yes; then
-  if test "$GCC" = yes; then
-    CFLAGS="-g -O2"
-  else
-    CFLAGS="-g"
-  fi
-else
-  if test "$GCC" = yes; then
-    CFLAGS="-O2"
-  else
-    CFLAGS=
-  fi
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to accept ISO C89" >&5
-$as_echo_n "checking for $CC option to accept ISO C89... " >&6; }
-if ${ac_cv_prog_cc_c89+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  ac_cv_prog_cc_c89=no
-ac_save_CC=$CC
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <stdarg.h>
-#include <stdio.h>
-struct stat;
-/* Most of the following tests are stolen from RCS 5.7's src/conf.sh.  */
-struct buf { int x; };
-FILE * (*rcsopen) (struct buf *, struct stat *, int);
-static char *e (p, i)
-     char **p;
-     int i;
-{
-  return p[i];
-}
-static char *f (char * (*g) (char **, int), char **p, ...)
-{
-  char *s;
-  va_list v;
-  va_start (v,p);
-  s = g (p, va_arg (v,int));
-  va_end (v);
-  return s;
-}
-
-/* OSF 4.0 Compaq cc is some sort of almost-ANSI by default.  It has
-   function prototypes and stuff, but not '\xHH' hex character constants.
-   These don't provoke an error unfortunately, instead are silently treated
-   as 'x'.  The following induces an error, until -std is added to get
-   proper ANSI mode.  Curiously '\x00'!='x' always comes out true, for an
-   array size at least.  It's necessary to write '\x00'==0 to get something
-   that's true only with -std.  */
-int osf4_cc_array ['\x00' == 0 ? 1 : -1];
-
-/* IBM C 6 for AIX is almost-ANSI by default, but it replaces macro parameters
-   inside strings and character constants.  */
-#define FOO(x) 'x'
-int xlc6_cc_array[FOO(a) == 'x' ? 1 : -1];
-
-int test (int i, double x);
-struct s1 {int (*f) (int a);};
-struct s2 {int (*f) (double a);};
-int pairnames (int, char **, FILE *(*)(struct buf *, struct stat *, int), int, int);
-int argc;
-char **argv;
-int
-main ()
-{
-return f (e, argv, 0) != argv[0]  ||  f (e, argv, 1) != argv[1];
-  ;
-  return 0;
-}
-_ACEOF
-for ac_arg in '' -qlanglvl=extc89 -qlanglvl=ansi -std \
-	-Ae "-Aa -D_HPUX_SOURCE" "-Xc -D__EXTENSIONS__"
-do
-  CC="$ac_save_CC $ac_arg"
-  if ac_fn_c_try_compile "$LINENO"; then :
-  ac_cv_prog_cc_c89=$ac_arg
-fi
-rm -f core conftest.err conftest.$ac_objext
-  test "x$ac_cv_prog_cc_c89" != "xno" && break
-done
-rm -f conftest.$ac_ext
-CC=$ac_save_CC
-
-fi
-# AC_CACHE_VAL
-case "x$ac_cv_prog_cc_c89" in
-  x)
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: none needed" >&5
-$as_echo "none needed" >&6; } ;;
-  xno)
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: unsupported" >&5
-$as_echo "unsupported" >&6; } ;;
-  *)
-    CC="$CC $ac_cv_prog_cc_c89"
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_c89" >&5
-$as_echo "$ac_cv_prog_cc_c89" >&6; } ;;
-esac
-if test "x$ac_cv_prog_cc_c89" != xno; then :
-
-fi
-
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC understands -c and -o together" >&5
-$as_echo_n "checking whether $CC understands -c and -o together... " >&6; }
-if ${am_cv_prog_cc_c_o+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-  # Make sure it works both with $CC and with simple cc.
-  # Following AC_PROG_CC_C_O, we do the test twice because some
-  # compilers refuse to overwrite an existing .o file with -o,
-  # though they will create one.
-  am_cv_prog_cc_c_o=yes
-  for am_i in 1 2; do
-    if { echo "$as_me:$LINENO: $CC -c conftest.$ac_ext -o conftest2.$ac_objext" >&5
-   ($CC -c conftest.$ac_ext -o conftest2.$ac_objext) >&5 2>&5
-   ac_status=$?
-   echo "$as_me:$LINENO: \$? = $ac_status" >&5
-   (exit $ac_status); } \
-         && test -f conftest2.$ac_objext; then
-      : OK
-    else
-      am_cv_prog_cc_c_o=no
-      break
-    fi
-  done
-  rm -f core conftest*
-  unset am_i
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_prog_cc_c_o" >&5
-$as_echo "$am_cv_prog_cc_c_o" >&6; }
-if test "$am_cv_prog_cc_c_o" != yes; then
-   # Losing compiler, so override with the script.
-   # FIXME: It is wrong to rewrite CC.
-   # But if we don't then we get into trouble of one sort or another.
-   # A longer-term fix would be to have automake use am__CC in this case,
-   # and then we could set am__CC="\$(top_srcdir)/compile \$(CC)"
-   CC="$am_aux_dir/compile $CC"
-fi
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-
-
-depcc="$CC"   am_compiler_list=
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
-$as_echo_n "checking dependency style of $depcc... " >&6; }
-if ${am_cv_CC_dependencies_compiler_type+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
-  # We make a subdir and do the tests there.  Otherwise we can end up
-  # making bogus files that we don't know about and never remove.  For
-  # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named 'D' -- because '-MD' means "put the output
-  # in D".
-  rm -rf conftest.dir
-  mkdir conftest.dir
-  # Copy depcomp to subdir because otherwise we won't find it if we're
-  # using a relative directory.
-  cp "$am_depcomp" conftest.dir
-  cd conftest.dir
-  # We will build objects and dependencies in a subdirectory because
-  # it helps to detect inapplicable dependency modes.  For instance
-  # both Tru64's cc and ICC support -MD to output dependencies as a
-  # side effect of compilation, but ICC will put the dependencies in
-  # the current directory while Tru64 will put them in the object
-  # directory.
-  mkdir sub
-
-  am_cv_CC_dependencies_compiler_type=none
-  if test "$am_compiler_list" = ""; then
-     am_compiler_list=`sed -n 's/^#*\([a-zA-Z0-9]*\))$/\1/p' < ./depcomp`
-  fi
-  am__universal=false
-  case " $depcc " in #(
-     *\ -arch\ *\ -arch\ *) am__universal=true ;;
-     esac
-
-  for depmode in $am_compiler_list; do
-    # Setup a source with many dependencies, because some compilers
-    # like to wrap large dependency lists on column 80 (with \), and
-    # we should not choose a depcomp mode which is confused by this.
-    #
-    # We need to recreate these files for each test, as the compiler may
-    # overwrite some of them when testing with obscure command lines.
-    # This happens at least with the AIX C compiler.
-    : > sub/conftest.c
-    for i in 1 2 3 4 5 6; do
-      echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
-      # Solaris 10 /bin/sh.
-      echo '/* dummy */' > sub/conftst$i.h
-    done
-    echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
-
-    # We check with '-c' and '-o' for the sake of the "dashmstdout"
-    # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle '-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs.
-    am__obj=sub/conftest.${OBJEXT-o}
-    am__minus_obj="-o $am__obj"
-    case $depmode in
-    gcc)
-      # This depmode causes a compiler race in universal mode.
-      test "$am__universal" = false || continue
-      ;;
-    nosideeffect)
-      # After this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested.
-      if test "x$enable_dependency_tracking" = xyes; then
-	continue
-      else
-	break
-      fi
-      ;;
-    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
-      # This compiler won't grok '-c -o', but also, the minuso test has
-      # not run yet.  These depmodes are late enough in the game, and
-      # so weak that their functioning should not be impacted.
-      am__obj=conftest.${OBJEXT-o}
-      am__minus_obj=
-      ;;
-    none) break ;;
-    esac
-    if depmode=$depmode \
-       source=sub/conftest.c object=$am__obj \
-       depfile=sub/conftest.Po tmpdepfile=sub/conftest.TPo \
-       $SHELL ./depcomp $depcc -c $am__minus_obj sub/conftest.c \
-         >/dev/null 2>conftest.err &&
-       grep sub/conftst1.h sub/conftest.Po > /dev/null 2>&1 &&
-       grep sub/conftst6.h sub/conftest.Po > /dev/null 2>&1 &&
-       grep $am__obj sub/conftest.Po > /dev/null 2>&1 &&
-       ${MAKE-make} -s -f confmf > /dev/null 2>&1; then
-      # icc doesn't choke on unknown options, it will just issue warnings
-      # or remarks (even with -Werror).  So we grep stderr for any message
-      # that says an option was ignored or not supported.
-      # When given -MP, icc 7.0 and 7.1 complain thusly:
-      #   icc: Command line warning: ignoring option '-M'; no argument required
-      # The diagnosis changed in icc 8.0:
-      #   icc: Command line remark: option '-MP' not supported
-      if (grep 'ignoring option' conftest.err ||
-          grep 'not supported' conftest.err) >/dev/null 2>&1; then :; else
-        am_cv_CC_dependencies_compiler_type=$depmode
-        break
-      fi
-    fi
-  done
-
-  cd ..
-  rm -rf conftest.dir
-else
-  am_cv_CC_dependencies_compiler_type=none
-fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CC_dependencies_compiler_type" >&5
-$as_echo "$am_cv_CC_dependencies_compiler_type" >&6; }
-CCDEPMODE=depmode=$am_cv_CC_dependencies_compiler_type
-
- if
-  test "x$enable_dependency_tracking" != xno \
-  && test "$am_cv_CC_dependencies_compiler_type" = gcc3; then
-  am__fastdepCC_TRUE=
-  am__fastdepCC_FALSE='#'
-else
-  am__fastdepCC_TRUE='#'
-  am__fastdepCC_FALSE=
-fi
-
-
-
-
-  OPENMP_CFLAGS=
+  OPENMP_CXXFLAGS=
   # Check whether --enable-openmp was given.
 if test "${enable_openmp+set}" = set; then :
   enableval=$enable_openmp;
 fi
 
   if test "$enable_openmp" != no; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to support OpenMP" >&5
-$as_echo_n "checking for $CC option to support OpenMP... " >&6; }
-if ${ac_cv_prog_c_openmp+:} false; then :
+    { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CXX option to support OpenMP" >&5
+$as_echo_n "checking for $CXX option to support OpenMP... " >&6; }
+if ${ac_cv_prog_cxx_openmp+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
@@ -4419,14 +3651,14 @@ else
 int main () { return omp_get_num_threads (); }
 
 _ACEOF
-if ac_fn_c_try_link "$LINENO"; then :
-  ac_cv_prog_c_openmp='none needed'
+if ac_fn_cxx_try_link "$LINENO"; then :
+  ac_cv_prog_cxx_openmp='none needed'
 else
-  ac_cv_prog_c_openmp='unsupported'
+  ac_cv_prog_cxx_openmp='unsupported'
 	  	  	  	  	  	  	                                	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp -homp \
                            -Popenmp --openmp; do
-	    ac_save_CFLAGS=$CFLAGS
-	    CFLAGS="$CFLAGS $ac_option"
+	    ac_save_CXXFLAGS=$CXXFLAGS
+	    CXXFLAGS="$CXXFLAGS $ac_option"
 	    cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
@@ -4437,13 +3669,13 @@ else
 int main () { return omp_get_num_threads (); }
 
 _ACEOF
-if ac_fn_c_try_link "$LINENO"; then :
-  ac_cv_prog_c_openmp=$ac_option
+if ac_fn_cxx_try_link "$LINENO"; then :
+  ac_cv_prog_cxx_openmp=$ac_option
 fi
 rm -f core conftest.err conftest.$ac_objext \
     conftest$ac_exeext conftest.$ac_ext
-	    CFLAGS=$ac_save_CFLAGS
-	    if test "$ac_cv_prog_c_openmp" != unsupported; then
+	    CXXFLAGS=$ac_save_CXXFLAGS
+	    if test "$ac_cv_prog_cxx_openmp" != unsupported; then
 	      break
 	    fi
 	  done
@@ -4451,13 +3683,13 @@ fi
 rm -f core conftest.err conftest.$ac_objext \
     conftest$ac_exeext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_c_openmp" >&5
-$as_echo "$ac_cv_prog_c_openmp" >&6; }
-    case $ac_cv_prog_c_openmp in #(
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cxx_openmp" >&5
+$as_echo "$ac_cv_prog_cxx_openmp" >&6; }
+    case $ac_cv_prog_cxx_openmp in #(
       "none needed" | unsupported)
 	;; #(
       *)
-	OPENMP_CFLAGS=$ac_cv_prog_c_openmp ;;
+	OPENMP_CXXFLAGS=$ac_cv_prog_cxx_openmp ;;
     esac
   fi
 
@@ -4559,26 +3791,22 @@ fi
 #AX_GCC_VAR_ATTRIBUTE(aligned)
 
 # Checks for header files.
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking how to run the C preprocessor" >&5
-$as_echo_n "checking how to run the C preprocessor... " >&6; }
-# On Suns, sometimes $CPP names a directory.
-if test -n "$CPP" && test -d "$CPP"; then
-  CPP=
-fi
-if test -z "$CPP"; then
-  if ${ac_cv_prog_CPP+:} false; then :
+ac_ext=cpp
+ac_cpp='$CXXCPP $CPPFLAGS'
+ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking how to run the C++ preprocessor" >&5
+$as_echo_n "checking how to run the C++ preprocessor... " >&6; }
+if test -z "$CXXCPP"; then
+  if ${ac_cv_prog_CXXCPP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-      # Double quotes because CPP needs to be expanded
-    for CPP in "$CC -E" "$CC -E -traditional-cpp" "/lib/cpp"
+      # Double quotes because CXXCPP needs to be expanded
+    for CXXCPP in "$CXX -E" "/lib/cpp"
     do
       ac_preproc_ok=false
-for ac_c_preproc_warn_flag in '' yes
+for ac_cxx_preproc_warn_flag in '' yes
 do
   # Use a header file that comes with gcc, so configuring glibc
   # with a fresh cross-compiler works.
@@ -4595,7 +3823,7 @@ do
 #endif
 		     Syntax error
 _ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
+if ac_fn_cxx_try_cpp "$LINENO"; then :
 
 else
   # Broken: fails on valid input.
@@ -4609,7 +3837,7 @@ rm -f conftest.err conftest.i conftest.$ac_ext
 /* end confdefs.h.  */
 #include <ac_nonexistent.h>
 _ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
+if ac_fn_cxx_try_cpp "$LINENO"; then :
   # Broken: success on invalid input.
 continue
 else
@@ -4627,17 +3855,17 @@ if $ac_preproc_ok; then :
 fi
 
     done
-    ac_cv_prog_CPP=$CPP
+    ac_cv_prog_CXXCPP=$CXXCPP
 
 fi
-  CPP=$ac_cv_prog_CPP
+  CXXCPP=$ac_cv_prog_CXXCPP
 else
-  ac_cv_prog_CPP=$CPP
+  ac_cv_prog_CXXCPP=$CXXCPP
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $CPP" >&5
-$as_echo "$CPP" >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $CXXCPP" >&5
+$as_echo "$CXXCPP" >&6; }
 ac_preproc_ok=false
-for ac_c_preproc_warn_flag in '' yes
+for ac_cxx_preproc_warn_flag in '' yes
 do
   # Use a header file that comes with gcc, so configuring glibc
   # with a fresh cross-compiler works.
@@ -4654,7 +3882,7 @@ do
 #endif
 		     Syntax error
 _ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
+if ac_fn_cxx_try_cpp "$LINENO"; then :
 
 else
   # Broken: fails on valid input.
@@ -4668,7 +3896,7 @@ rm -f conftest.err conftest.i conftest.$ac_ext
 /* end confdefs.h.  */
 #include <ac_nonexistent.h>
 _ACEOF
-if ac_fn_c_try_cpp "$LINENO"; then :
+if ac_fn_cxx_try_cpp "$LINENO"; then :
   # Broken: success on invalid input.
 continue
 else
@@ -4686,15 +3914,15 @@ if $ac_preproc_ok; then :
 else
   { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "C preprocessor \"$CPP\" fails sanity check
+as_fn_error $? "C++ preprocessor \"$CXXCPP\" fails sanity check
 See \`config.log' for more details" "$LINENO" 5; }
 fi
 
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
+ac_ext=cpp
+ac_cpp='$CXXCPP $CPPFLAGS'
+ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
 
 
 { $as_echo "$as_me:${as_lineno-$LINENO}: checking for grep that handles long lines and -e" >&5
@@ -4847,7 +4075,7 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
+if ac_fn_cxx_try_compile "$LINENO"; then :
   ac_cv_header_stdc=yes
 else
   ac_cv_header_stdc=no
@@ -4920,7 +4148,7 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_c_try_run "$LINENO"; then :
+if ac_fn_cxx_try_run "$LINENO"; then :
 
 else
   ac_cv_header_stdc=no
@@ -4944,7 +4172,7 @@ for ac_header in sys/types.h sys/stat.h stdlib.h string.h memory.h strings.h \
 		  inttypes.h stdint.h unistd.h
 do :
   as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-ac_fn_c_check_header_compile "$LINENO" "$ac_header" "$as_ac_Header" "$ac_includes_default
+ac_fn_cxx_check_header_compile "$LINENO" "$ac_header" "$as_ac_Header" "$ac_includes_default
 "
 if eval test \"x\$"$as_ac_Header"\" = x"yes"; then :
   cat >>confdefs.h <<_ACEOF
@@ -4958,7 +4186,7 @@ done
 
 for ac_header in stdint.h
 do :
-  ac_fn_c_check_header_mongrel "$LINENO" "stdint.h" "ac_cv_header_stdint_h" "$ac_includes_default"
+  ac_fn_cxx_check_header_mongrel "$LINENO" "stdint.h" "ac_cv_header_stdint_h" "$ac_includes_default"
 if test "x$ac_cv_header_stdint_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
 #define HAVE_STDINT_H 1
@@ -4970,7 +4198,7 @@ done
 
 for ac_header in malloc/malloc.h
 do :
-  ac_fn_c_check_header_mongrel "$LINENO" "malloc/malloc.h" "ac_cv_header_malloc_malloc_h" "$ac_includes_default"
+  ac_fn_cxx_check_header_mongrel "$LINENO" "malloc/malloc.h" "ac_cv_header_malloc_malloc_h" "$ac_includes_default"
 if test "x$ac_cv_header_malloc_malloc_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
 #define HAVE_MALLOC_MALLOC_H 1
@@ -4982,7 +4210,7 @@ done
 
 for ac_header in malloc.h
 do :
-  ac_fn_c_check_header_mongrel "$LINENO" "malloc.h" "ac_cv_header_malloc_h" "$ac_includes_default"
+  ac_fn_cxx_check_header_mongrel "$LINENO" "malloc.h" "ac_cv_header_malloc_h" "$ac_includes_default"
 if test "x$ac_cv_header_malloc_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
 #define HAVE_MALLOC_H 1
@@ -4994,7 +4222,7 @@ done
 
 for ac_header in endian.h
 do :
-  ac_fn_c_check_header_mongrel "$LINENO" "endian.h" "ac_cv_header_endian_h" "$ac_includes_default"
+  ac_fn_cxx_check_header_mongrel "$LINENO" "endian.h" "ac_cv_header_endian_h" "$ac_includes_default"
 if test "x$ac_cv_header_endian_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
 #define HAVE_ENDIAN_H 1
@@ -5005,7 +4233,7 @@ fi
 done
 
 #AC_CHECK_HEADERS(machine/endian.h)
-ac_fn_c_check_decl "$LINENO" "ntohll" "ac_cv_have_decl_ntohll" "#include <arpa/inet.h>
+ac_fn_cxx_check_decl "$LINENO" "ntohll" "ac_cv_have_decl_ntohll" "#include <arpa/inet.h>
 "
 if test "x$ac_cv_have_decl_ntohll" = xyes; then :
   ac_have_decl=1
@@ -5017,7 +4245,7 @@ cat >>confdefs.h <<_ACEOF
 #define HAVE_DECL_NTOHLL $ac_have_decl
 _ACEOF
 
-ac_fn_c_check_decl "$LINENO" "be64toh" "ac_cv_have_decl_be64toh" "#include <arpa/inet.h>
+ac_fn_cxx_check_decl "$LINENO" "be64toh" "ac_cv_have_decl_be64toh" "#include <arpa/inet.h>
 "
 if test "x$ac_cv_have_decl_be64toh" = xyes; then :
   ac_have_decl=1
@@ -5031,7 +4259,7 @@ _ACEOF
 
 
 # Checks for typedefs, structures, and compiler characteristics.
-ac_fn_c_check_type "$LINENO" "size_t" "ac_cv_type_size_t" "$ac_includes_default"
+ac_fn_cxx_check_type "$LINENO" "size_t" "ac_cv_type_size_t" "$ac_includes_default"
 if test "x$ac_cv_type_size_t" = xyes; then :
 
 else
@@ -5074,7 +4302,7 @@ _ACEOF
 # Checks for library functions.
 for ac_func in gettimeofday
 do :
-  ac_fn_c_check_func "$LINENO" "gettimeofday" "ac_cv_func_gettimeofday"
+  ac_fn_cxx_check_func "$LINENO" "gettimeofday" "ac_cv_func_gettimeofday"
 if test "x$ac_cv_func_gettimeofday" = xyes; then :
   cat >>confdefs.h <<_ACEOF
 #define HAVE_GETTIMEOFDAY 1
@@ -5309,10 +4537,6 @@ if test -z "${am__fastdepCXX_TRUE}" && test -z "${am__fastdepCXX_FALSE}"; then
   as_fn_error $? "conditional \"am__fastdepCXX\" was never defined.
 Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
-if test -z "${am__fastdepCC_TRUE}" && test -z "${am__fastdepCC_FALSE}"; then
-  as_fn_error $? "conditional \"am__fastdepCC\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
-fi
 if test -z "${BUILD_COMMS_MPI_TRUE}" && test -z "${BUILD_COMMS_MPI_FALSE}"; then
   as_fn_error $? "conditional \"BUILD_COMMS_MPI\" was never defined.
 Usually this means the macro was only invoked conditionally." "$LINENO" 5
diff --git a/configure-commands b/configure-commands
index d22e9fdd..8ba2dac1 100644
--- a/configure-commands
+++ b/configure-commands
@@ -1,3 +1,4 @@
+CXX=clang-omp++ ./configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp"  --enable-comms=mpi
 CXX=clang++ ./configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx"  --enable-comms=mpi
 CXX=clang++ ./configure --enable-simd=AVX CXXFLAGS="-mavx -g -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx"  --enable-comms=mpi
 CXX=clang++ ./configure --enable-simd=AVX CXXFLAGS="-mavx -g  -std=c++11" LDFLAGS= LIBS=-lmpi  --enable-comms=fake
diff --git a/configure.ac b/configure.ac
index ce73ffd9..86f4b70c 100644
--- a/configure.ac
+++ b/configure.ac
@@ -6,6 +6,7 @@ AC_CONFIG_SRCDIR([lib/Grid.h])
 AC_CONFIG_HEADERS([lib/Grid_config.h])
 
 # Checks for programs.
+AC_LANG(C++)
 AC_PROG_CXX
 AC_OPENMP
 AC_PROG_RANLIB
diff --git a/lib/Grid.h b/lib/Grid.h
index bd7b371a..f46f7bf5 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -24,21 +24,21 @@
 #include <stdio.h>
 #include <signal.h>
 
+#ifndef MAX
+#define MAX(x,y) ((x)>(y)?(x):(y))
+#define MIN(x,y) ((x)>(y)?(y):(x))
+#endif
+
 #include <Grid_config.h>
 
 ////////////////////////////////////////////////////////////
 // Tunable header includes
 ////////////////////////////////////////////////////////////
 
-#ifdef HAVE_OPENMP
-#define OMP
-#include <omp.h>
-#endif
 
 #ifdef HAVE_MALLOC_MALLOC_H
 #include <malloc/malloc.h>
 #endif
-
 #ifdef HAVE_MALLOC_H
 #include <malloc.h>
 #endif
diff --git a/lib/Grid_stencil.h b/lib/Grid_stencil.h
index fa537f65..ce987746 100644
--- a/lib/Grid_stencil.h
+++ b/lib/Grid_stencil.h
@@ -250,7 +250,11 @@ namespace Grid {
 	  int buffer_size = _grid->_slice_nblock[dimension]*_grid->_slice_block[dimension];
 	  int words = sizeof(cobj)/sizeof(vector_type);
 
-	  /* FIXME ALTERNATE BUFFER DETERMINATION ; possibly slow to allocate*/
+	  /*
+	   * possibly slow to allocate
+	   * Doesn't matter in this test, but may want to preallocate in the 
+	   * dirac operators
+	   */
 	  std::vector<std::vector<scalar_object> > send_buf_extract(Nsimd,std::vector<scalar_object>(buffer_size) ); 
 	  std::vector<std::vector<scalar_object> > recv_buf_extract(Nsimd,std::vector<scalar_object>(buffer_size) );
 	  int bytes = buffer_size*sizeof(scalar_object);
diff --git a/lib/Grid_threads.h b/lib/Grid_threads.h
index 6c5f17e0..89a56d80 100644
--- a/lib/Grid_threads.h
+++ b/lib/Grid_threads.h
@@ -1,13 +1,17 @@
 #ifndef GRID_THREADS_H
 #define GRID_THREADS_H
 
-#ifdef HAVE_OPENMP
+#ifdef _OPENMP
+#define GRID_OMP
+#endif
+
+#ifdef GRID_OMP
 #include <omp.h>
 #define PARALLEL_FOR_LOOP _Pragma("omp parallel for")
-#define PARALLEL_NESTED_LOOP(n) _Pragma("omp parallel for collapse(" #n ")")
+#define PARALLEL_NESTED_LOOP2 _Pragma("omp parallel for collapse(2)")
 #else
 #define PARALLEL_FOR_LOOP 
-#define PARALLEL_NESTED_LOOP(n) 
+#define PARALLEL_NESTED_LOOP2
 #endif
 
 namespace Grid {
@@ -20,7 +24,7 @@ class GridThread {
   static int _threads;
 
   static void SetThreads(int thr) { 
-#ifdef HAVE_OPENMP
+#ifdef GRID_OMP
     _threads = MIN(thr,omp_get_max_threads()) ;
     omp_set_num_threads(_threads);
 #else 
@@ -28,7 +32,7 @@ class GridThread {
 #endif
   };
   static void SetMaxThreads(void) { 
-#ifdef HAVE_OPENMP
+#ifdef GRID_OMP
     _threads = omp_get_max_threads();
     omp_set_num_threads(_threads);
 #else 
@@ -58,7 +62,7 @@ class GridThread {
   };
 
   static int  ThreadBarrier(void) {
-#ifdef HAVE_OPENMP
+#ifdef GRID_OMP
 #pragma omp barrier
     return omp_get_thread_num();
 #else
diff --git a/lib/cshift/Grid_cshift_common.h b/lib/cshift/Grid_cshift_common.h
index 9864df73..c212171e 100644
--- a/lib/cshift/Grid_cshift_common.h
+++ b/lib/cshift/Grid_cshift_common.h
@@ -26,16 +26,15 @@ Gather_plane_simple (const Lattice<vobj> &rhs,std::vector<cobj,alignedAllocator<
   }
 
   int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-  int bo  = 0;                                      // offset in buffer
   
-PARALLEL_NESTED_LOOP(2)
+PARALLEL_NESTED_LOOP2
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-      int o = n*rhs._grid->_slice_stride[dimension];
+      int o  = n*rhs._grid->_slice_stride[dimension];
+      int bo = n*rhs._grid->_slice_block[dimension];
       int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);// Could easily be a table lookup
       if ( ocb &cbmask ) {
-	buffer[bo]=compress(rhs._odata[so+o+b]);
-	bo++;
+	buffer[bo+b]=compress(rhs._odata[so+o+b]);
       }
     }
   }
@@ -55,9 +54,8 @@ Gather_plane_extract(const Lattice<vobj> &rhs,std::vector<typename cobj::scalar_
   }
 
   int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-  int bo  = 0;                                      // offset in buffer
     
-PARALLEL_NESTED_LOOP(2)
+PARALLEL_NESTED_LOOP2
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
 
@@ -104,15 +102,15 @@ template<class vobj> void Scatter_plane_simple (Lattice<vobj> &rhs,std::vector<v
   }
 
   int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-  int bo  = 0;                                      // offset in buffer
     
-PARALLEL_NESTED_LOOP(2)
+PARALLEL_NESTED_LOOP2
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-      int o=n*rhs._grid->_slice_stride[dimension];
+      int o   =n*rhs._grid->_slice_stride[dimension];
+      int bo  =n*rhs._grid->_slice_block[dimension];
       int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);// Could easily be a table lookup
       if ( ocb & cbmask ) {
-	rhs._odata[so+o+b]=buffer[bo++];
+	rhs._odata[so+o+b]=buffer[bo+b];
       }
     }
   }
@@ -131,7 +129,7 @@ PARALLEL_NESTED_LOOP(2)
 
   int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
     
-PARALLEL_NESTED_LOOP(2)
+PARALLEL_NESTED_LOOP2
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
 
@@ -160,7 +158,7 @@ template<class vobj> void Copy_plane(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int
   int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
   int lo  = lplane*lhs._grid->_ostride[dimension]; // base offset for start of plane 
   
-PARALLEL_NESTED_LOOP(2)
+PARALLEL_NESTED_LOOP2
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
       
@@ -185,7 +183,7 @@ template<class vobj> void Copy_plane_permute(Lattice<vobj>& lhs,Lattice<vobj> &r
   int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
   int lo  = lplane*lhs._grid->_ostride[dimension]; // base offset for start of plane 
   
-PARALLEL_NESTED_LOOP(2)
+PARALLEL_NESTED_LOOP2
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
       int o =n*rhs._grid->_slice_stride[dimension];
diff --git a/lib/cshift/Grid_cshift_mpi.h b/lib/cshift/Grid_cshift_mpi.h
index 7c77ebc9..569017d8 100644
--- a/lib/cshift/Grid_cshift_mpi.h
+++ b/lib/cshift/Grid_cshift_mpi.h
@@ -1,10 +1,6 @@
 #ifndef _GRID_CSHIFT_MPI_H_
 #define _GRID_CSHIFT_MPI_H_
 
-#ifndef MAX
-#define MAX(x,y) ((x)>(y)?(x):(y))
-#define MIN(x,y) ((x)>(y)?(y):(x))
-#endif
 
 namespace Grid { 
 
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index f657208a..25f1d914 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -100,22 +100,21 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
   vHalfSpinColourVector  chi;    
   vSpinColourVector result;
   vHalfSpinColourVector Uchi;
-  vHalfSpinColourVector *chi_p;
   int offset,local,perm, ptype;
 
 PARALLEL_FOR_LOOP
   for(int sss=0;sss<grid->oSites();sss++){
 
     int ss = sss;
-    int ssu= sss;
-    //int ss = Stencil._LebesgueReorder[sss];
+    int ssu= ss;
+    //    int ss = Stencil._LebesgueReorder[sss];
 
     // Xp
     offset = Stencil._offsets [Xp][ss];
     local  = Stencil._is_local[Xp][ss];
     perm   = Stencil._permute[Xp][ss];
     ptype  = Stencil._permute_type[Xp];
-    chi_p  = &comm_buf[offset];
+
     if ( local && perm ) 
     {
       spProjXp(tmp,in._odata[offset]);

From add4495a4a0241805a22dcdd2c8e3e53542a5113 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 13 May 2015 09:24:10 +0100
Subject: [PATCH 149/429] cout IO for all types

---
 TODO                                | 18 ++++++----
 lib/Grid_simd.h                     | 55 +++++++++++++++++++++++++++++
 lib/lattice/Grid_lattice_base.h     | 20 +++++++++++
 lib/lattice/Grid_lattice_peekpoke.h |  2 +-
 lib/math/Grid_math_tensors.h        | 26 ++++++++++++++
 tests/Makefile.am                   |  5 ++-
 6 files changed, 118 insertions(+), 8 deletions(-)

diff --git a/TODO b/TODO
index b596c605..a75c867d 100644
--- a/TODO
+++ b/TODO
@@ -5,6 +5,7 @@
   - Performance check on Guido's reimplementation strategy
 
 * FIXME audit
+
 * const audit
 
 * Replace vset with a call to merge.; 
@@ -13,6 +14,8 @@
 
 * Strong test for norm2, conj and all primitive types. -- tests/Grid_simd.cc is almost there
 
+* Thread scaling tests Xeon, XeonPhi
+
 ================================================================
 *** New Functionality
 ================================================================
@@ -23,9 +26,6 @@
   - use protocol buffers? replace xmlReader/Writer ec..
   - Binary use htonll, htonl
 
-* Bug in SeedFixedIntegers gave same output on each site. -- Think I fixed but NOT checked for sure
-  Implement and use lattice IO to verify this.
-
 * Expression template engine: -- DONE
    -- Norm2(expression) problem: introduce norm2 unary op, or Introduce conversion automatic from expression to Lattice<vobj>
 
@@ -33,6 +33,7 @@
 
 ** Make the Tensor types and Complex etc... play more nicely.
   - TensorRemove is a hack, come up with a long term rationalised approach to Complex vs. Scalar<Scalar<Scalar<Complex > > >
+
   QDP forces use of "toDouble" to get back to non tensor scalar. This role is presently taken TensorRemove, but I
   want to introduce a syntax that does not require this.
   - Reductions that contract indices on a site should always demote the tensor structure.
@@ -71,6 +72,7 @@
    // Fourier transform equivalent.
 
 Actions -- coherent framework for implementing actions and their forces.
+
 * Fermion
   - Wilson
   - Clover
@@ -81,22 +83,24 @@ Actions -- coherent framework for implementing actions and their forces.
 * Gauge
   - Wilson, symanzik, iwasaki
 
-Algorithms
+Algorithms (lots of reuse/port from BFM)
 * LinearOperator
 * LinearSolver
 * Polynomial 
 * Eigen
+* CG
 * Pcg
 * Adef2
 * DeflCG
 * fPcg
 * MCR
 * HDCG
-* HMC, Heatbath
+* HMC, 
+* Heatbath
 * etc..
 
 ======================================================================================================
-FUNCTIONALITY: it pleases me to keep track of things I have done (keeps sane)
+FUNCTIONALITY: it pleases me to keep track of things I have done (keeps me arguably sane)
 ======================================================================================================
 
 * Command line args for geometry, simd, etc. layout. Is it necessary to have -- DONE
@@ -158,4 +162,6 @@ FUNCTIONALITY: it pleases me to keep track of things I have done (keeps sane)
 * Conformable test in Cshift routines. -- none needed ; there is only one 
 * Conformable testing in expression templates -- DONE (recursive)
 
+* Bug in SeedFixedIntegers gave same output on each site. -- DONE
+  Implement and use lattice IO to verify this.            -- cout for lattice types DONE
 
diff --git a/lib/Grid_simd.h b/lib/Grid_simd.h
index 39265896..7ad7367b 100644
--- a/lib/Grid_simd.h
+++ b/lib/Grid_simd.h
@@ -234,5 +234,60 @@ namespace Grid {
   typedef vRealF vReal;
   typedef vComplexF vComplex;
 #endif
+
+ 
+  inline std::ostream& operator<< (std::ostream& stream, const vComplexF &o){
+    int nn=vComplexF::Nsimd();
+    std::vector<ComplexF,alignedAllocator<ComplexF> > buf(nn);
+    vstore(o,&buf[0]);
+    stream<<"<";
+    for(int i=0;i<nn;i++){
+      stream<<buf[i];
+      if(i<nn-1) stream<<",";
+    }
+    stream<<">";
+    return stream;
+  }
+ 
+  inline std::ostream& operator<< (std::ostream& stream, const vComplexD &o){
+    int nn=vComplexD::Nsimd();
+    std::vector<ComplexD,alignedAllocator<ComplexD> > buf(nn);
+    vstore(o,&buf[0]);
+    stream<<"<";
+    for(int i=0;i<nn;i++){
+      stream<<buf[i];
+      if(i<nn-1) stream<<",";
+    }
+    stream<<">";
+    return stream;
+  }
+
+  inline std::ostream& operator<< (std::ostream& stream, const vRealF &o){
+    int nn=vRealF::Nsimd();
+    std::vector<RealF,alignedAllocator<RealF> > buf(nn);
+    vstore(o,&buf[0]);
+    stream<<"<";
+    for(int i=0;i<nn;i++){
+      stream<<buf[i];
+      if(i<nn-1) stream<<",";
+    }
+    stream<<">";
+    return stream;
+  }
+
+  inline std::ostream& operator<< (std::ostream& stream, const vRealD &o){
+    int nn=vRealD::Nsimd();
+    std::vector<RealD,alignedAllocator<RealD> > buf(nn);
+    vstore(o,&buf[0]);
+    stream<<"<";
+    for(int i=0;i<nn;i++){
+      stream<<buf[i];
+      if(i<nn-1) stream<<",";
+    }
+    stream<<">";
+    return stream;
+  }
+
+
 }
 #endif
diff --git a/lib/lattice/Grid_lattice_base.h b/lib/lattice/Grid_lattice_base.h
index ace7565d..915879cb 100644
--- a/lib/lattice/Grid_lattice_base.h
+++ b/lib/lattice/Grid_lattice_base.h
@@ -176,7 +176,27 @@ PARALLEL_FOR_LOOP
         }
         return ret;
     };
+
  }; // class Lattice
+
+  template<class vobj> inline std::ostream& operator<< (std::ostream& stream, const Lattice<vobj> &o){
+    std::vector<int> gcoor;
+    typedef typename vobj::scalar_object sobj;
+    sobj ss;
+    for(int g=0;g<o._grid->_gsites;g++){
+      o._grid->GlobalIndexToGlobalCoor(g,gcoor);
+      peekSite(ss,o,gcoor);
+      stream<<"[";
+      for(int d=0;d<gcoor.size();d++){
+	stream<<gcoor[d];
+	if(d!=gcoor.size()-1) stream<<",";
+      }
+      stream<<"]\t";
+      stream<<ss<<std::endl;
+    }
+    return stream;
+  }
+
 }
 
 
diff --git a/lib/lattice/Grid_lattice_peekpoke.h b/lib/lattice/Grid_lattice_peekpoke.h
index ad7dfc50..44d2a71f 100644
--- a/lib/lattice/Grid_lattice_peekpoke.h
+++ b/lib/lattice/Grid_lattice_peekpoke.h
@@ -110,7 +110,7 @@ PARALLEL_FOR_LOOP
     // Peek a scalar object from the SIMD array
     //////////////////////////////////////////////////////////
     template<class vobj,class sobj>
-      void peekSite(sobj &s,Lattice<vobj> &l,std::vector<int> &site){
+      void peekSite(sobj &s,const Lattice<vobj> &l,std::vector<int> &site){
         
       GridBase *grid=l._grid;
 
diff --git a/lib/math/Grid_math_tensors.h b/lib/math/Grid_math_tensors.h
index 0c1d67d1..8b3e4b1b 100644
--- a/lib/math/Grid_math_tensors.h
+++ b/lib/math/Grid_math_tensors.h
@@ -95,6 +95,10 @@ public:
       return *this;
     }
 
+    friend std::ostream& operator<< (std::ostream& stream, const iScalar<vtype> &o){
+      stream<< "S {"<<o._internal<<"}";
+      return stream;
+    };
 };
 ///////////////////////////////////////////////////////////
 // Allows to turn scalar<scalar<scalar<double>>>> back to double.
@@ -170,6 +174,15 @@ public:
     inline const vtype & operator ()(int i) const {
       return _internal[i];
     }
+    friend std::ostream& operator<< (std::ostream& stream, const iVector<vtype,N> &o){
+      stream<< "V<"<<N<<">{";
+      for(int i=0;i<N;i++) {
+	stream<<o._internal[i];
+	if (i<N-1)	stream<<",";
+      }
+      stream<<"}";
+      return stream;
+    };
     //    inline vtype && operator ()(int i) {
     //      return _internal[i];
     //    }
@@ -262,6 +275,19 @@ public:
   inline const vtype & operator ()(int i,int j) const {
     return _internal[i][j];
   }
+  friend std::ostream& operator<< (std::ostream& stream, const iMatrix<vtype,N> &o){
+    stream<< "M<"<<N<<">{";
+    for(int i=0;i<N;i++) {
+      stream<< "{";
+      for(int j=0;j<N;j++) {
+	stream<<o._internal[i][j];
+	if (i<N-1)	stream<<",";
+      }
+      stream<<"}\n\t\t";
+    }
+    stream<<"}";
+    return stream;
+  };
 
   //  inline vtype && operator ()(int i,int j) {
   //    return _internal[i][j];
diff --git a/tests/Makefile.am b/tests/Makefile.am
index b129d3d9..6af1ce3b 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -5,11 +5,14 @@ AM_LDFLAGS = -L$(top_srcdir)/lib
 #
 # Test code
 #
-bin_PROGRAMS = Grid_main Grid_stencil Grid_nersc_io Grid_cshift Grid_gamma  Grid_simd
+bin_PROGRAMS = Grid_main Grid_stencil Grid_nersc_io Grid_cshift Grid_gamma  Grid_simd Grid_rng
 
 Grid_main_SOURCES = Grid_main.cc
 Grid_main_LDADD = -lGrid
 
+Grid_rng_SOURCES = Grid_rng.cc
+Grid_rng_LDADD = -lGrid
+
 Grid_nersc_io_SOURCES = Grid_nersc_io.cc
 Grid_nersc_io_LDADD = -lGrid
 

From e6e72d23df1d786ea93f7e4873c5e175f2ec2560 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 13 May 2015 09:24:30 +0100
Subject: [PATCH 150/429] RNG test

---
 tests/Grid_rng.cc | 51 +++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 51 insertions(+)
 create mode 100644 tests/Grid_rng.cc

diff --git a/tests/Grid_rng.cc b/tests/Grid_rng.cc
new file mode 100644
index 00000000..97f6c6b7
--- /dev/null
+++ b/tests/Grid_rng.cc
@@ -0,0 +1,51 @@
+#include <Grid.h>
+#include <parallelIO/GridNerscIO.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+     
+  GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+
+  std::vector<int> seeds({1,2,3,4});
+
+  GridSerialRNG             sRNG;   sRNG.SeedRandomDevice();
+  GridSerialRNG            fsRNG;  fsRNG.SeedFixedIntegers(seeds);
+
+  GridParallelRNG           pRNG(&Grid);   pRNG.SeedRandomDevice();
+  GridParallelRNG          fpRNG(&Grid);  fpRNG.SeedFixedIntegers(seeds);
+
+  SpinMatrix rnd  ; 
+  random(sRNG,rnd);
+  std::cout<<"Random Spin Matrix (random_device)\n"<< rnd<<std::endl;
+
+  random(fsRNG,rnd);
+  std::cout<<"Random Spin Matrix (fixed seed)\n"<< rnd<<std::endl;
+
+  SpinVector rv; 
+  random(sRNG,rv);
+  std::cout<<"Random Spin Vector (random device)\n"<< rv<<std::endl;
+
+  random(fsRNG,rv);
+  std::cout<<"Random Spin Vector (fixed seed)\n"<< rv<<std::endl;
+
+  gaussian(fsRNG,rv);
+  std::cout<<"Gaussian Spin Vector (fixed seed)\n"<< rv<<std::endl;
+
+  LatticeColourVector lcv(&Grid);
+  random(pRNG,lcv);
+  std::cout<<"Random Lattice Colour Vector (random device)\n"<< lcv<<std::endl;
+
+  random(fpRNG,lcv);
+  std::cout<<"Random Lattice Colour Vector (fixed seed)\n"<< lcv<<std::endl;
+
+  Grid_finalize();
+}

From 0097b81778905afa7b391565b6040a9c96103618 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 13 May 2015 10:59:22 +0100
Subject: [PATCH 151/429] OMP dslash working

---
 TODO                           | 77 ++++++++++++++++++----------------
 benchmarks/Grid_wilson.cc      |  2 +-
 lib/qcd/Grid_qcd_wilson_dop.cc | 26 ++++++++----
 lib/qcd/Grid_qcd_wilson_dop.h  | 12 +++---
 4 files changed, 64 insertions(+), 53 deletions(-)

diff --git a/TODO b/TODO
index a75c867d..ed4dedd4 100644
--- a/TODO
+++ b/TODO
@@ -8,60 +8,39 @@
 
 * const audit
 
+Insert/Extract
 * Replace vset with a call to merge.; 
 * care in Gmerge,Gextract over vset .
 * extract / merge extra implementation removal      
-
-* Strong test for norm2, conj and all primitive types. -- tests/Grid_simd.cc is almost there
+* Optimise the extract/merge SIMD routines; Azusa??
+ - I have collated into single location at least.
+ - Need to use _mm_*insert/extract routines.
 
 * Thread scaling tests Xeon, XeonPhi
 
-================================================================
-*** New Functionality
-================================================================
-
-* Implement where within expression template scheme.
-
-* - BinaryWriter, TextWriter etc...
-  - use protocol buffers? replace xmlReader/Writer ec..
-  - Binary use htonll, htonl
-
-* Expression template engine: -- DONE
-   -- Norm2(expression) problem: introduce norm2 unary op, or Introduce conversion automatic from expression to Lattice<vobj>
-
-* CovariantShift support                             -----Use a class to store gauge field? (parallel transport?)
-
 ** Make the Tensor types and Complex etc... play more nicely.
   - TensorRemove is a hack, come up with a long term rationalised approach to Complex vs. Scalar<Scalar<Scalar<Complex > > >
-
   QDP forces use of "toDouble" to get back to non tensor scalar. This role is presently taken TensorRemove, but I
   want to introduce a syntax that does not require this.
+
   - Reductions that contract indices on a site should always demote the tensor structure.
     norm2(), innerProduct.
+
   - Result of Sum(), SliceSum // spatial sums
               trace, traceIndex etc.. do not.
+
   - problem arises because "trace" returns Lattice<TComplex> moving everything down to Scalar,
     and then Sum and SliceSum to not remove the Scalars. This would be fixed if we 
     template specialize the scalar scalar scalar sum and SliceSum,  on the basis of being
     pure scalar.
 
-* Optimise the extract/merge SIMD routines; Azusa??
- - I have collated into single location at least.
- - Need to use _mm_*insert/extract routines.
-
-* Flavour matrices?
-* Pauli, SU subgroup, etc.. 
-* su3 exponentiation & log etc.. [Jamie's code?]
-* TaProj
-* FFTnD ?
-
-* Parallel io improvements
-  - optional parallel MPI2 IO
-  - move Plaquette and link trace checks into nersc reader from the Grid_nersc_io.cc test.
-  
-* rb4d support for 5th dimension in Mobius.
+*** Expression template engine: -- DONE
 
+[   -- Norm2(expression) problem: introduce norm2 unary op, or Introduce conversion automatic from expression to Lattice<vobj>
+* Strong test for norm2, conj and all primitive types. -- tests/Grid_simd.cc is almost there
+* Implement where within expression template scheme.
 * Check for missing functionality                    - partially audited against QDP++ layout
+
    // Unary functions
    // cos,sin, tan, acos, asin, cosh, acosh, tanh, sinh, // Scalar<vReal> only arg
    // exp, log, sqrt, fabs
@@ -69,7 +48,21 @@
    // adjColor, adjSpin,
    // copyMask.
    // localMaxAbs
-   // Fourier transform equivalent.
+   // Fourier transform equivalent.]
+
+================================================================
+*** New Functionality
+================================================================
+
+* - BinaryWriter, TextWriter etc...
+  - use protocol buffers? replace xmlReader/Writer ec..
+  - Binary use htonll, htonl
+
+* CovariantShift support                             -----Use a class to store gauge field? (parallel transport?)
+
+* Parallel io improvements
+  - optional parallel MPI2 IO
+  - move Plaquette and link trace checks into nersc reader from the Grid_nersc_io.cc test.
 
 Actions -- coherent framework for implementing actions and their forces.
 
@@ -80,9 +73,6 @@ Actions -- coherent framework for implementing actions and their forces.
   - Mobius
   - z-Mobius
 
-* Gauge
-  - Wilson, symanzik, iwasaki
-
 Algorithms (lots of reuse/port from BFM)
 * LinearOperator
 * LinearSolver
@@ -97,8 +87,21 @@ Algorithms (lots of reuse/port from BFM)
 * HDCG
 * HMC, 
 * Heatbath
+* Integrators, leapfrog, omelyan, force gradient etc...
 * etc..
 
+* Gauge
+  - Wilson, symanzik, iwasaki
+
+* rb4d support for 5th dimension in Mobius.
+
+* Flavour matrices?
+* Pauli, SU subgroup, etc.. 
+* su3 exponentiation & log etc.. [Jamie's code?]
+* TaProj
+* FFTnD ?
+
+
 ======================================================================================================
 FUNCTIONALITY: it pleases me to keep track of things I have done (keeps me arguably sane)
 ======================================================================================================
diff --git a/benchmarks/Grid_wilson.cc b/benchmarks/Grid_wilson.cc
index 9ce13090..7718eb19 100644
--- a/benchmarks/Grid_wilson.cc
+++ b/benchmarks/Grid_wilson.cc
@@ -80,7 +80,7 @@ int main (int argc, char ** argv)
   int ncall=1000;
   double t0=usecond();
   for(int i=0;i<ncall;i++){
-    Dw.multiply(src,result);
+    Dw.M(src,result);
   }
   double t1=usecond();
   double flops=1320*volume*ncall;
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index 25f1d914..14c48cf2 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -85,7 +85,17 @@ void WilsonMatrix::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeF
   }
 }
 
-void WilsonMatrix::multiply(const LatticeFermion &in, LatticeFermion &out)
+void WilsonMatrix::M(const LatticeFermion &in, LatticeFermion &out)
+{
+  Dhop(in,out);
+  return;
+}
+void WilsonMatrix::Mdag(const LatticeFermion &in, LatticeFermion &out)
+{
+  Dhop(in,out);
+  return;
+}
+void WilsonMatrix::MdagM(const LatticeFermion &in, LatticeFermion &out)
 {
   Dhop(in,out);
   return;
@@ -96,18 +106,18 @@ void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
   WilsonCompressor compressor;
   Stencil.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
 
-  vHalfSpinColourVector  tmp;    
-  vHalfSpinColourVector  chi;    
-  vSpinColourVector result;
-  vHalfSpinColourVector Uchi;
-  int offset,local,perm, ptype;
-
 PARALLEL_FOR_LOOP
   for(int sss=0;sss<grid->oSites();sss++){
 
+    vHalfSpinColourVector  tmp;    
+    vHalfSpinColourVector  chi;    
+    vSpinColourVector result;
+    vHalfSpinColourVector Uchi;
+    int offset,local,perm, ptype;
+
+    //    int ss = Stencil._LebesgueReorder[sss];
     int ss = sss;
     int ssu= ss;
-    //    int ss = Stencil._LebesgueReorder[sss];
 
     // Xp
     offset = Stencil._offsets [Xp][ss];
diff --git a/lib/qcd/Grid_qcd_wilson_dop.h b/lib/qcd/Grid_qcd_wilson_dop.h
index 900f1801..e19cbabe 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.h
+++ b/lib/qcd/Grid_qcd_wilson_dop.h
@@ -3,16 +3,12 @@
 
 #include <Grid.h>
 
+#include <algorithms/LinearOperator.h>
+
 namespace Grid {
 
   namespace QCD {
 
-
-    template<class vtype> class LinearOperatorBase {
-    public:
-      void multiply(const Lattice<vtype> &in, Lattice<vtype> &out){ assert(0);}
-    };
-
     class WilsonMatrix : public LinearOperatorBase<LatticeFermion>
     {
       //NB r=1;
@@ -40,7 +36,9 @@ namespace Grid {
       void DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu);
 
       // override multiply
-      void multiply(const LatticeFermion &in, LatticeFermion &out);
+      virtual void M    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void Mdag (const LatticeFermion &in, LatticeFermion &out);
+      virtual void MdagM(const LatticeFermion &in, LatticeFermion &out);
 
       // non-hermitian hopping term; half cb or both
       void Dhop(const LatticeFermion &in, LatticeFermion &out);

From 5166888c0a69c86476f1021dd108193480a0013b Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 13 May 2015 11:25:34 +0100
Subject: [PATCH 152/429] Linear op added

---
 lib/algorithms/LinearOperator.h | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)
 create mode 100644 lib/algorithms/LinearOperator.h

diff --git a/lib/algorithms/LinearOperator.h b/lib/algorithms/LinearOperator.h
new file mode 100644
index 00000000..5929570f
--- /dev/null
+++ b/lib/algorithms/LinearOperator.h
@@ -0,0 +1,18 @@
+#ifndef  GRID_ALGORITHM_LINEAR_OP_H
+#define  GRID_ALGORITHM_LINEAR_OP_H
+
+#include <Grid.h>
+
+namespace Grid {
+
+  // Red black cases?
+    template<class vtype> class LinearOperatorBase {
+    public:
+      void M(const Lattice<vtype> &in, Lattice<vtype> &out){ assert(0);}
+      void Mdag(const Lattice<vtype> &in, Lattice<vtype> &out){ assert(0);}
+      void MdagM(const Lattice<vtype> &in, Lattice<vtype> &out){ assert(0);}
+    };
+
+}
+
+#endif

From cc6218a692e7ebb03ddaf9cb15d41958f726c52d Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:31:41 +0100
Subject: [PATCH 153/429] strong inline required to force icpc

---
 lib/Grid.h | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/lib/Grid.h b/lib/Grid.h
index f46f7bf5..2936dd27 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -29,13 +29,14 @@
 #define MIN(x,y) ((x)>(y)?(y):(x))
 #endif
 
+#define strong_inline __attribute__((always_inline)) inline
+
 #include <Grid_config.h>
 
 ////////////////////////////////////////////////////////////
 // Tunable header includes
 ////////////////////////////////////////////////////////////
 
-
 #ifdef HAVE_MALLOC_MALLOC_H
 #include <malloc/malloc.h>
 #endif

From 8d1b26dd4b2d97f9b6aba144082f5d50f71971ed Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:32:11 +0100
Subject: [PATCH 154/429] Key of mm_malloc.h

---
 lib/Grid_aligned_allocator.h | 15 +++++++++++----
 1 file changed, 11 insertions(+), 4 deletions(-)

diff --git a/lib/Grid_aligned_allocator.h b/lib/Grid_aligned_allocator.h
index f0b93172..0e74e842 100644
--- a/lib/Grid_aligned_allocator.h
+++ b/lib/Grid_aligned_allocator.h
@@ -2,6 +2,9 @@
 #define GRID_ALIGNED_ALLOCATOR_H
 
 #include <immintrin.h>
+#ifdef HAVE_MM_MALLOC_H
+#include <mm_malloc.h>
+#endif
 
 namespace Grid {
 
@@ -36,16 +39,20 @@ public:
 
   pointer allocate(size_type __n, const void* = 0)
   { 
-#ifdef AVX512
-    _Tp * ptr = (_Tp *) memalign(128,__n*sizeof(_Tp));
-#else
+#ifdef HAVE_MM_MALLOC_H
     _Tp * ptr = (_Tp *) _mm_malloc(__n*sizeof(_Tp),128);
+#else
+    _Tp * ptr = (_Tp *) memalign(128,__n*sizeof(_Tp));
 #endif
     return ptr;
   }
 
   void deallocate(pointer __p, size_type) { 
-    free(__p); 
+#ifdef HAVE_MM_MALLOC_H
+    _mm_free(__p); 
+#else
+    free(__p);
+#endif
   }
   void construct(pointer __p, const _Tp& __val) { };
   void construct(pointer __p) { };

From 4e462209c7ea187a6913b24e830d52575529b9f9 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:32:45 +0100
Subject: [PATCH 155/429] Using boolean logic inside enable_if is more elegant

---
 lib/Grid_extract.h | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/lib/Grid_extract.h b/lib/Grid_extract.h
index e29fbee7..75868463 100644
--- a/lib/Grid_extract.h
+++ b/lib/Grid_extract.h
@@ -11,7 +11,7 @@ namespace Grid{
 ////////////////////////////////////////////////////////////////////////////////////////////////
 
 template<class vsimd,class scalar>
-inline void extract(typename std::enable_if<isGridTensor<vsimd>::notvalue, const vsimd >::type * y, 
+inline void extract(typename std::enable_if<!isGridTensor<vsimd>::value, const vsimd >::type * y, 
 		    std::vector<scalar *> &extracted,int offset){
   // FIXME: bounce off memory is painful
   int Nextr=extracted.size();
@@ -27,7 +27,7 @@ inline void extract(typename std::enable_if<isGridTensor<vsimd>::notvalue, const
 // Merge simd vector from array of scalars to pointer array with offset
 ////////////////////////////////////////////////////////////////////////
 template<class vsimd,class scalar>
-inline void merge(typename std::enable_if<isGridTensor<vsimd>::notvalue, vsimd >::type * y, 
+inline void merge(typename std::enable_if<!isGridTensor<vsimd>::value, vsimd >::type * y, 
 		  std::vector<scalar *> &extracted,int offset){
   int Nextr=extracted.size();
   int Nsimd=vsimd::Nsimd();
@@ -47,7 +47,7 @@ inline void merge(typename std::enable_if<isGridTensor<vsimd>::notvalue, vsimd >
 // Extract a fundamental vector type to scalar array 
 ////////////////////////////////////////////////////////////////////////////////////////////////
 template<class vsimd,class scalar>
-inline void extract(typename std::enable_if<isGridTensor<vsimd>::notvalue, const vsimd >::type  &y,std::vector<scalar> &extracted){
+inline void extract(typename std::enable_if<!isGridTensor<vsimd>::value, const vsimd >::type  &y,std::vector<scalar> &extracted){
 
   int Nextr=extracted.size();
   int Nsimd=vsimd::Nsimd();
@@ -66,7 +66,7 @@ inline void extract(typename std::enable_if<isGridTensor<vsimd>::notvalue, const
 // Merge simd vector from array of scalars
 ////////////////////////////////////////////////////////////////////////
 template<class vsimd,class scalar>
-inline void merge(typename std::enable_if<isGridTensor<vsimd>::notvalue, vsimd >::type  &y,std::vector<scalar> &extracted){
+inline void merge(typename std::enable_if<!isGridTensor<vsimd>::value, vsimd >::type  &y,std::vector<scalar> &extracted){
   int Nextr=extracted.size();
   int Nsimd=vsimd::Nsimd();
   int s=Nsimd/Nextr;
@@ -80,7 +80,7 @@ inline void merge(typename std::enable_if<isGridTensor<vsimd>::notvalue, vsimd >
 
 };
 template<class vsimd,class scalar>
-inline void AmergeA(typename std::enable_if<isGridTensor<vsimd>::notvalue, vsimd >::type  &y,std::vector<scalar> &extracted){
+inline void AmergeA(typename std::enable_if<!isGridTensor<vsimd>::value, vsimd >::type  &y,std::vector<scalar> &extracted){
   int Nextr=extracted.size();
   int Nsimd=vsimd::Nsimd();
   int s=Nsimd/Nextr;

From 051b23fe10e91ecec64f19e78dd1750986a35d92 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:35:02 +0100
Subject: [PATCH 156/429] ICPC and GCC5 fixes

---
 lib/Grid_simd.h | 43 ++++++++++++++++++++++++++++++++-----------
 1 file changed, 32 insertions(+), 11 deletions(-)

diff --git a/lib/Grid_simd.h b/lib/Grid_simd.h
index 7ad7367b..af77591c 100644
--- a/lib/Grid_simd.h
+++ b/lib/Grid_simd.h
@@ -10,17 +10,29 @@
 //
 // Vector types are arch dependent
 ////////////////////////////////////////////////////////////////////////
-    
+
+typedef uint32_t Integer;
 
 #ifdef SSE4
 #include <pmmintrin.h>
 #endif
 #if defined(AVX1) || defined (AVX2)
 #include <immintrin.h>
+    
+// _mm256_set_m128i(hi,lo); // not defined in all versions of immintrin.h
+#ifndef _mm256_set_m128i
+#define _mm256_set_m128i(hi,lo) _mm256_insertf128_si256(_mm256_castsi128_si256(lo),(hi),1)
 #endif
+
+#endif
+
 #ifdef AVX512
 #include <immintrin.h>
-#include <zmmintrin.h>
+#ifndef KNC_ONLY_STORES
+#define  _mm512_storenrngo_ps _mm512_store_ps  // not present in AVX512
+#define  _mm512_storenrngo_pd _mm512_store_pd  // not present in AVX512
+#endif
+
 #endif
 
 namespace Grid {
@@ -148,16 +160,23 @@ namespace Grid {
 //////////////////////////////////////////////////////////
 template<class vsimd>
 inline void Gpermute(vsimd &y,const vsimd &b,int perm){
+	union { 
+	  fvec f;
+	  decltype(vsimd::v) v;
+	} conv;
+	conv.v = b.v;
       switch (perm){
 #if defined(AVX1)||defined(AVX2)
       // 8x32 bits=>3 permutes
-      case 2: y.v = _mm256_shuffle_ps(b.v,b.v,_MM_SHUFFLE(2,3,0,1)); break;
-      case 1: y.v = _mm256_shuffle_ps(b.v,b.v,_MM_SHUFFLE(1,0,3,2)); break;
-      case 0: y.v = _mm256_permute2f128_ps(b.v,b.v,0x01); break;
+      case 2: 
+	conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); 
+	break;
+      case 1: conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2)); break;
+      case 0: conv.f = _mm256_permute2f128_ps(conv.f,conv.f,0x01); break;
 #endif
 #ifdef SSE4
-      case 1: y.v = _mm_shuffle_ps(b.v,b.v,_MM_SHUFFLE(2,3,0,1)); break;
-      case 0: y.v = _mm_shuffle_ps(b.v,b.v,_MM_SHUFFLE(1,0,3,2));break;
+      case 1: conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); break;
+      case 0: conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2));break;
 #endif
 #ifdef AVX512
 	// 16 floats=> permutes
@@ -165,16 +184,18 @@ inline void Gpermute(vsimd &y,const vsimd &b,int perm){
         // Permute 1 every abcd efgh ijkl mnop -> cdab ghef jkij opmn 
         // Permute 2 every abcd efgh ijkl mnop -> efgh abcd mnop ijkl
         // Permute 3 every abcd efgh ijkl mnop -> ijkl mnop abcd efgh
-      case 3: y.v =(decltype(y.v)) _mm512_swizzle_ps((__m512)b.v,_MM_SWIZ_REG_CDAB); break;
-      case 2: y.v =(decltype(y.v)) _mm512_swizzle_ps((__m512)b.v,_MM_SWIZ_REG_BADC); break;
-      case 1: y.v =(decltype(y.v)) _mm512_permute4f128_ps((__m512)b.v,(_MM_PERM_ENUM)_MM_SHUFFLE(2,3,0,1)); break;
-      case 0: y.v =(decltype(y.v)) _mm512_permute4f128_ps((__m512)b.v,(_MM_PERM_ENUM)_MM_SHUFFLE(1,0,3,2)); break;
+      case 3: conv.f = _mm512_swizzle_ps(conv.f,_MM_SWIZ_REG_CDAB); break;
+      case 2: conv.f = _mm512_swizzle_ps(conv.f,_MM_SWIZ_REG_BADC); break;
+      case 1: conv.f = _mm512_permute4f128_ps(conv.f,(_MM_PERM_ENUM)_MM_SHUFFLE(2,3,0,1)); break;
+      case 0: conv.f = _mm512_permute4f128_ps(conv.f,(_MM_PERM_ENUM)_MM_SHUFFLE(1,0,3,2)); break;
 #endif
 #ifdef QPX
 #error not implemented
 #endif
       default: assert(0); break;
       }
+      y.v=conv.v;
+
     };
 };
 

From 6c7eb60d6fe4b6534f604ea93c98483c3d2af104 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:36:22 +0100
Subject: [PATCH 157/429] gcc doesn't like collapse(2) for some reason I can't
 figure

---
 lib/cshift/Grid_cshift_common.h | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/lib/cshift/Grid_cshift_common.h b/lib/cshift/Grid_cshift_common.h
index c212171e..cf1f88af 100644
--- a/lib/cshift/Grid_cshift_common.h
+++ b/lib/cshift/Grid_cshift_common.h
@@ -185,14 +185,15 @@ template<class vobj> void Copy_plane_permute(Lattice<vobj>& lhs,Lattice<vobj> &r
   
 PARALLEL_NESTED_LOOP2
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-    for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-      int o =n*rhs._grid->_slice_stride[dimension];
+  for(int b=0;b<rhs._grid->_slice_block [dimension];b++){
+
+      int o  =n*rhs._grid->_slice_stride[dimension];
       int ocb=1<<lhs._grid->CheckerBoardFromOindex(o+b);
       if ( ocb&cbmask ) {
 	permute(lhs._odata[lo+o+b],rhs._odata[ro+o+b],permute_type);
       }
-    }
-  }
+
+  }}
 }
 
 //////////////////////////////////////////////////////

From c28551f40f36d2c85d4bc3c9ac0a47244ea21c91 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:37:07 +0100
Subject: [PATCH 158/429] Silly formatting change

---
 lib/lattice/Grid_lattice_arith.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/lattice/Grid_lattice_arith.h b/lib/lattice/Grid_lattice_arith.h
index 10f31025..c9599582 100644
--- a/lib/lattice/Grid_lattice_arith.h
+++ b/lib/lattice/Grid_lattice_arith.h
@@ -7,7 +7,7 @@ namespace Grid {
   //////////////////////////////////////////////////////////////////////////////////////////////////////
   //  avoid copy back routines for mult, mac, sub, add
   //////////////////////////////////////////////////////////////////////////////////////////////////////
-  template<class obj1,class obj2,class obj3>
+  template<class obj1,class obj2,class obj3> 
     void mult(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
     conformable(lhs,rhs);
 PARALLEL_FOR_LOOP

From e7d25647e6a1dd098a14fb7d24802962514e29c5 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:38:04 +0100
Subject: [PATCH 159/429] Filed bug report Bug 66153 on GCC-5.
 https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66153

---
 lib/lattice/Grid_lattice_peekpoke.h | 19 ++++++++-----------
 1 file changed, 8 insertions(+), 11 deletions(-)

diff --git a/lib/lattice/Grid_lattice_peekpoke.h b/lib/lattice/Grid_lattice_peekpoke.h
index 44d2a71f..7bffdbb8 100644
--- a/lib/lattice/Grid_lattice_peekpoke.h
+++ b/lib/lattice/Grid_lattice_peekpoke.h
@@ -10,9 +10,8 @@ namespace Grid {
     ////////////////////////////////////////////////////////////////////////////////////////////////////
     // Peek internal indices of a Lattice object
     ////////////////////////////////////////////////////////////////////////////////////////////////////
-    template<int Index,class vobj>
-    inline auto peekIndex(const Lattice<vobj> &lhs)
-      -> Lattice<decltype(peekIndex<Index>(lhs._odata[0]))>
+  template<int Index,class vobj>
+     auto peekIndex(const Lattice<vobj> &lhs) -> Lattice<decltype(peekIndex<Index>(lhs._odata[0]))>
     {
       Lattice<decltype(peekIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
 PARALLEL_FOR_LOOP
@@ -22,8 +21,7 @@ PARALLEL_FOR_LOOP
         return ret;
     };
     template<int Index,class vobj>
-      inline auto peekIndex(const Lattice<vobj> &lhs,int i)
-      -> Lattice<decltype(peekIndex<Index>(lhs._odata[0],i))>
+       auto peekIndex(const Lattice<vobj> &lhs,int i) -> Lattice<decltype(peekIndex<Index>(lhs._odata[0],i))>
     {
       Lattice<decltype(peekIndex<Index>(lhs._odata[0],i))> ret(lhs._grid);
 PARALLEL_FOR_LOOP
@@ -33,8 +31,7 @@ PARALLEL_FOR_LOOP
         return ret;
     };
     template<int Index,class vobj>
-      inline auto peekIndex(const Lattice<vobj> &lhs,int i,int j)
-      -> Lattice<decltype(peekIndex<Index>(lhs._odata[0],i,j))>
+       auto peekIndex(const Lattice<vobj> &lhs,int i,int j) -> Lattice<decltype(peekIndex<Index>(lhs._odata[0],i,j))>
     {
       Lattice<decltype(peekIndex<Index>(lhs._odata[0],i,j))> ret(lhs._grid);
 PARALLEL_FOR_LOOP
@@ -47,7 +44,7 @@ PARALLEL_FOR_LOOP
     ////////////////////////////////////////////////////////////////////////////////////////////////////
     // Poke internal indices of a Lattice object
     ////////////////////////////////////////////////////////////////////////////////////////////////////
-    template<int Index,class vobj> inline
+    template<int Index,class vobj>
     void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0]))> & rhs)
     {
 PARALLEL_FOR_LOOP
@@ -55,7 +52,7 @@ PARALLEL_FOR_LOOP
 	  pokeIndex<Index>(lhs._odata[ss],rhs._odata[ss]);
 	}      
     }
-    template<int Index,class vobj> inline
+    template<int Index,class vobj> 
     void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0],0))> & rhs,int i)
     {
 PARALLEL_FOR_LOOP
@@ -63,8 +60,8 @@ PARALLEL_FOR_LOOP
 	  pokeIndex<Index>(lhs._odata[ss],rhs._odata[ss],i);
 	}      
     }
-    template<int Index,class vobj> inline
-    void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0],0,0))> & rhs,int i,int j)
+    template<int Index,class vobj>
+      void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0],0,0))> & rhs,int i,int j)
     {
 PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){

From 5b46992a15ffed15198f35dd2f1de1000a9d40fa Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:38:54 +0100
Subject: [PATCH 160/429] Formatting change

---
 lib/lattice/Grid_lattice_trace.h | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/lib/lattice/Grid_lattice_trace.h b/lib/lattice/Grid_lattice_trace.h
index 72f984f9..75cc5b87 100644
--- a/lib/lattice/Grid_lattice_trace.h
+++ b/lib/lattice/Grid_lattice_trace.h
@@ -26,8 +26,7 @@ PARALLEL_FOR_LOOP
     // Trace Index level dependent operation
     ////////////////////////////////////////////////////////////////////////////////////////////////////
     template<int Index,class vobj>
-    inline auto traceIndex(const Lattice<vobj> &lhs)
-      -> Lattice<decltype(traceIndex<Index>(lhs._odata[0]))>
+    inline auto traceIndex(const Lattice<vobj> &lhs) -> Lattice<decltype(traceIndex<Index>(lhs._odata[0]))>
     {
       Lattice<decltype(traceIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
 PARALLEL_FOR_LOOP

From adc4f86020dbb21cff29cad6966ba267b9dc4191 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:39:25 +0100
Subject: [PATCH 161/429] Promote to strong inline to force ICPC's hand.
 Annoying.

---
 lib/math/Grid_math_arith_add.h | 22 +++++++++++-----------
 1 file changed, 11 insertions(+), 11 deletions(-)

diff --git a/lib/math/Grid_math_arith_add.h b/lib/math/Grid_math_arith_add.h
index 93449402..19a2f407 100644
--- a/lib/math/Grid_math_arith_add.h
+++ b/lib/math/Grid_math_arith_add.h
@@ -13,13 +13,13 @@ namespace Grid {
 // Scalar +/- Scalar
 // Vector +/- Vector
 // Matrix +/- Matrix
-  template<class vtype,class ltype,class rtype> inline void add(iScalar<vtype> * __restrict__ ret,
+  template<class vtype,class ltype,class rtype> strong_inline void add(iScalar<vtype> * __restrict__ ret,
 								const iScalar<ltype> * __restrict__ lhs,
 								const iScalar<rtype> * __restrict__ rhs)
   {
     add(&ret->_internal,&lhs->_internal,&rhs->_internal);
   }
-  template<class vtype,class ltype,class rtype,int N> inline void add(iVector<vtype,N> * __restrict__ ret,
+  template<class vtype,class ltype,class rtype,int N> strong_inline void add(iVector<vtype,N> * __restrict__ ret,
 								      const iVector<ltype,N> * __restrict__ lhs,
 								      const iVector<rtype,N> * __restrict__ rhs)
   {
@@ -29,7 +29,7 @@ namespace Grid {
     return;
   }
   
-  template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
+  template<class vtype,class ltype,class rtype, int N> strong_inline  void add(iMatrix<vtype,N> * __restrict__ ret,
 									const iMatrix<ltype,N> * __restrict__ lhs,
 									const iMatrix<rtype,N> * __restrict__ rhs)
   {
@@ -39,7 +39,7 @@ namespace Grid {
       }}
     return;
   }
-  template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
+  template<class vtype,class ltype,class rtype, int N> strong_inline  void add(iMatrix<vtype,N> * __restrict__ ret,
 									const iScalar<ltype>   * __restrict__ lhs,
 									const iMatrix<rtype,N> * __restrict__ rhs)
   {
@@ -49,7 +49,7 @@ namespace Grid {
       }}
     return;
   }
-  template<class vtype,class ltype,class rtype, int N> inline  void add(iMatrix<vtype,N> * __restrict__ ret,
+  template<class vtype,class ltype,class rtype, int N> strong_inline  void add(iMatrix<vtype,N> * __restrict__ ret,
 									const iMatrix<ltype,N> * __restrict__ lhs,
 									const iScalar<rtype>   * __restrict__ rhs)
   {
@@ -66,8 +66,8 @@ namespace Grid {
 
   // + operator for scalar, vector, matrix
   template<class ltype,class rtype>
-    //inline auto operator + (iScalar<ltype>& lhs,iScalar<rtype>&& rhs) -> iScalar<decltype(lhs._internal + rhs._internal)>
-    inline auto operator + (const iScalar<ltype>& lhs,const iScalar<rtype>& rhs) -> iScalar<decltype(lhs._internal + rhs._internal)>
+    //strong_inline auto operator + (iScalar<ltype>& lhs,iScalar<rtype>&& rhs) -> iScalar<decltype(lhs._internal + rhs._internal)>
+    strong_inline auto operator + (const iScalar<ltype>& lhs,const iScalar<rtype>& rhs) -> iScalar<decltype(lhs._internal + rhs._internal)>
   {
     typedef iScalar<decltype(lhs._internal+rhs._internal)> ret_t;
     ret_t ret;
@@ -75,7 +75,7 @@ namespace Grid {
     return ret;
   }
   template<class ltype,class rtype,int N>
-    inline auto operator + (const iVector<ltype,N>& lhs,const iVector<rtype,N>& rhs) ->iVector<decltype(lhs._internal[0]+rhs._internal[0]),N>
+    strong_inline auto operator + (const iVector<ltype,N>& lhs,const iVector<rtype,N>& rhs) ->iVector<decltype(lhs._internal[0]+rhs._internal[0]),N>
     {
       typedef iVector<decltype(lhs._internal[0]+rhs._internal[0]),N> ret_t;
     ret_t ret;
@@ -83,7 +83,7 @@ namespace Grid {
     return ret;
     }
   template<class ltype,class rtype,int N>
-    inline auto operator + (const iMatrix<ltype,N>& lhs,const iMatrix<rtype,N>& rhs) ->iMatrix<decltype(lhs._internal[0][0]+rhs._internal[0][0]),N>
+    strong_inline auto operator + (const iMatrix<ltype,N>& lhs,const iMatrix<rtype,N>& rhs) ->iMatrix<decltype(lhs._internal[0][0]+rhs._internal[0][0]),N>
     {
       typedef iMatrix<decltype(lhs._internal[0][0]+rhs._internal[0][0]),N> ret_t;
       ret_t ret;
@@ -91,7 +91,7 @@ namespace Grid {
       return ret;
     }
   template<class ltype,class rtype,int N>
-inline auto operator + (const iScalar<ltype>& lhs,const iMatrix<rtype,N>& rhs)->iMatrix<decltype(lhs._internal+rhs._internal[0][0]),N>
+strong_inline auto operator + (const iScalar<ltype>& lhs,const iMatrix<rtype,N>& rhs)->iMatrix<decltype(lhs._internal+rhs._internal[0][0]),N>
     {
       typedef iMatrix<decltype(lhs._internal+rhs._internal[0][0]),N> ret_t;
       ret_t ret;
@@ -100,7 +100,7 @@ inline auto operator + (const iScalar<ltype>& lhs,const iMatrix<rtype,N>& rhs)->
     }
 
   template<class ltype,class rtype,int N>
-    inline auto operator + (const iMatrix<ltype,N>& lhs,const iScalar<rtype>& rhs)->iMatrix<decltype(lhs._internal[0][0]+rhs._internal),N>
+    strong_inline auto operator + (const iMatrix<ltype,N>& lhs,const iScalar<rtype>& rhs)->iMatrix<decltype(lhs._internal[0][0]+rhs._internal),N>
     {
       typedef iMatrix<decltype(lhs._internal[0][0]+rhs._internal),N> ret_t;
       ret_t ret;

From b38bf82d483a7c1f59ac0e279829d77a62f4b902 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:40:00 +0100
Subject: [PATCH 162/429] Switch to strong_inline macro to force icpc's hand

---
 lib/math/Grid_math_arith_mac.h | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/lib/math/Grid_math_arith_mac.h b/lib/math/Grid_math_arith_mac.h
index 6260b98f..68b0acf1 100644
--- a/lib/math/Grid_math_arith_mac.h
+++ b/lib/math/Grid_math_arith_mac.h
@@ -21,12 +21,12 @@ namespace Grid {
     // scal x vec  = vec
     ///////////////////////////
 template<class rtype,class vtype,class mtype>
-inline  void mac(iScalar<rtype> * __restrict__ ret,const iScalar<vtype> * __restrict__ lhs,const iScalar<mtype> * __restrict__ rhs)
+strong_inline  void mac(iScalar<rtype> * __restrict__ ret,const iScalar<vtype> * __restrict__ lhs,const iScalar<mtype> * __restrict__ rhs)
 {
     mac(&ret->_internal,&lhs->_internal,&rhs->_internal);
 }
 template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
+strong_inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
     for(int c2=0;c2<N;c2++){
     for(int c1=0;c1<N;c1++){
     for(int c3=0;c3<N;c3++){
@@ -35,7 +35,7 @@ inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __
     return;
 }
 template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs){
+strong_inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs){
     for(int c1=0;c1<N;c1++){
     for(int c2=0;c2<N;c2++){
         mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
@@ -43,7 +43,7 @@ inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __
     return;
 }
 template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iScalar<ltype> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
+strong_inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iScalar<ltype> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
     for(int c1=0;c1<N;c1++){
     for(int c2=0;c2<N;c2++){
         mac(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
@@ -51,7 +51,7 @@ inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iScalar<ltype> * __re
     return;
 }
 template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iVector<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iVector<rtype,N> * __restrict__ rhs)
+strong_inline void mac(iVector<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iVector<rtype,N> * __restrict__ rhs)
 {
     for(int c1=0;c1<N;c1++){
     for(int c2=0;c2<N;c2++){
@@ -60,7 +60,7 @@ inline void mac(iVector<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __
     return;
 }
 template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iVector<rrtype,N> * __restrict__ ret,const iScalar<ltype> * __restrict__ lhs,const iVector<rtype,N> * __restrict__ rhs)
+strong_inline void mac(iVector<rrtype,N> * __restrict__ ret,const iScalar<ltype> * __restrict__ lhs,const iVector<rtype,N> * __restrict__ rhs)
 {
     for(int c1=0;c1<N;c1++){
         mac(&ret->_internal[c1],&lhs->_internal,&rhs->_internal[c1]);
@@ -68,7 +68,7 @@ inline void mac(iVector<rrtype,N> * __restrict__ ret,const iScalar<ltype> * __re
     return;
 }
 template<class rrtype,class ltype,class rtype,int N>
-inline void mac(iVector<rrtype,N> * __restrict__ ret,const iVector<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs)
+strong_inline void mac(iVector<rrtype,N> * __restrict__ ret,const iVector<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs)
 {
     for(int c1=0;c1<N;c1++){
         mac(&ret->_internal[c1],&lhs->_internal[c1],&rhs->_internal);

From 8c40dd9c4f9029630d49c3f5cc1e3a191a00ee2f Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:40:31 +0100
Subject: [PATCH 163/429] Force strong_inline to force ipcc's hand

---
 lib/math/Grid_math_arith_mul.h | 34 +++++++++++++++++-----------------
 1 file changed, 17 insertions(+), 17 deletions(-)

diff --git a/lib/math/Grid_math_arith_mul.h b/lib/math/Grid_math_arith_mul.h
index 66c8b121..7b883cf4 100644
--- a/lib/math/Grid_math_arith_mul.h
+++ b/lib/math/Grid_math_arith_mul.h
@@ -9,12 +9,12 @@ namespace Grid {
     ///////////////////////////////////////////////////////////////////////////////////////////////////
     
 template<class rtype,class vtype,class mtype>
-inline void mult(iScalar<rtype> * __restrict__ ret,const iScalar<mtype> * __restrict__ lhs,const iScalar<vtype> * __restrict__ rhs){
+strong_inline void mult(iScalar<rtype> * __restrict__ ret,const iScalar<mtype> * __restrict__ lhs,const iScalar<vtype> * __restrict__ rhs){
     mult(&ret->_internal,&lhs->_internal,&rhs->_internal);
 }
 
 template<class rrtype,class ltype,class rtype,int N>
-inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
+strong_inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
     for(int c2=0;c2<N;c2++){
     for(int c1=0;c1<N;c1++){
         mult(&ret->_internal[c1][c2],&lhs->_internal[c1][0],&rhs->_internal[0][c2]);
@@ -25,7 +25,7 @@ inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * _
     return;
 }
 template<class rrtype,class ltype,class rtype,int N>
-inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs){
+strong_inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs){
     for(int c2=0;c2<N;c2++){
     for(int c1=0;c1<N;c1++){
         mult(&ret->_internal[c1][c2],&lhs->_internal[c1][c2],&rhs->_internal);
@@ -34,7 +34,7 @@ inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * _
 }
 
 template<class rrtype,class ltype,class rtype, int N>
-inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iScalar<ltype>   * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
+strong_inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iScalar<ltype>   * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
     for(int c2=0;c2<N;c2++){
     for(int c1=0;c1<N;c1++){
         mult(&ret->_internal[c1][c2],&lhs->_internal,&rhs->_internal[c1][c2]);
@@ -43,7 +43,7 @@ inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iScalar<ltype>   * _
 }
 // Matrix left multiplies vector
 template<class rtype,class vtype,class mtype,int N>
-inline void mult(iVector<rtype,N> * __restrict__ ret,const iMatrix<mtype,N> * __restrict__ lhs,const iVector<vtype,N> * __restrict__ rhs)
+strong_inline void mult(iVector<rtype,N> * __restrict__ ret,const iMatrix<mtype,N> * __restrict__ lhs,const iVector<vtype,N> * __restrict__ rhs)
 {
     for(int c1=0;c1<N;c1++){
         mult(&ret->_internal[c1],&lhs->_internal[c1][0],&rhs->_internal[0]);
@@ -54,7 +54,7 @@ inline void mult(iVector<rtype,N> * __restrict__ ret,const iMatrix<mtype,N> * __
     return;
 }
 template<class rtype,class vtype,class mtype,int N>
-inline void mult(iVector<rtype,N> * __restrict__ ret,
+strong_inline void mult(iVector<rtype,N> * __restrict__ ret,
                  const iScalar<mtype>   * __restrict__ lhs,
                  const iVector<vtype,N> * __restrict__ rhs){
     for(int c1=0;c1<N;c1++){
@@ -62,7 +62,7 @@ inline void mult(iVector<rtype,N> * __restrict__ ret,
     }
 }
 template<class rtype,class vtype,class mtype,int N>
-inline void mult(iVector<rtype,N> * __restrict__ ret,
+strong_inline void mult(iVector<rtype,N> * __restrict__ ret,
                  const iVector<vtype,N> * __restrict__ rhs,
                  const iScalar<mtype> * __restrict__ lhs){
     mult(ret,lhs,rhs);
@@ -70,7 +70,7 @@ inline void mult(iVector<rtype,N> * __restrict__ ret,
     
 
 
-template<class rtype,class vtype,class mtype,int N> inline
+template<class rtype,class vtype,class mtype,int N> strong_inline
 iVector<rtype,N> operator * (const iMatrix<mtype,N>& lhs,const iVector<vtype,N>& rhs)
 {
     iVector<rtype,N> ret;
@@ -78,7 +78,7 @@ iVector<rtype,N> operator * (const iMatrix<mtype,N>& lhs,const iVector<vtype,N>&
     return ret;
 }
 
-template<class rtype,class vtype,class mtype,int N> inline
+template<class rtype,class vtype,class mtype,int N> strong_inline
 iVector<rtype,N> operator * (const iScalar<mtype>& lhs,const iVector<vtype,N>& rhs)
 {
     iVector<rtype,N> ret;
@@ -86,7 +86,7 @@ iVector<rtype,N> operator * (const iScalar<mtype>& lhs,const iVector<vtype,N>& r
     return ret;
 }
 
-template<class rtype,class vtype,class mtype,int N> inline
+template<class rtype,class vtype,class mtype,int N> strong_inline
 iVector<rtype,N> operator * (const iVector<mtype,N>& lhs,const iScalar<vtype>& rhs)
 {
     iVector<rtype,N> ret;
@@ -110,14 +110,14 @@ iVector<rtype,N> operator * (const iVector<mtype,N>& lhs,const iScalar<vtype>& r
     //
     // We can special case scalar_type ??
 template<class l,class r>
-inline auto operator * (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(lhs._internal * rhs._internal)>
+strong_inline auto operator * (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<decltype(lhs._internal * rhs._internal)>
 {
     typedef iScalar<decltype(lhs._internal*rhs._internal)> ret_t;
     ret_t ret;
     mult(&ret,&lhs,&rhs);
     return ret;
 }
-template<class l,class r,int N> inline
+template<class l,class r,int N> strong_inline
 auto operator * (const iMatrix<l,N>& lhs,const iMatrix<r,N>& rhs) -> iMatrix<decltype(lhs._internal[0][0]*rhs._internal[0][0]),N>
 {
     typedef decltype(lhs._internal[0][0]*rhs._internal[0][0]) ret_t;
@@ -125,7 +125,7 @@ auto operator * (const iMatrix<l,N>& lhs,const iMatrix<r,N>& rhs) -> iMatrix<dec
     mult(&ret,&lhs,&rhs);
     return ret;
 }
-template<class l,class r, int N> inline
+template<class l,class r, int N> strong_inline
 auto operator * (const iMatrix<r,N>& lhs,const iScalar<l>& rhs) -> iMatrix<decltype(lhs._internal[0][0]*rhs._internal),N>
 {
     typedef decltype(lhs._internal[0][0]*rhs._internal) ret_t;
@@ -137,7 +137,7 @@ auto operator * (const iMatrix<r,N>& lhs,const iScalar<l>& rhs) -> iMatrix<declt
     }}
     return ret;
 }
-template<class l,class r,int N> inline
+template<class l,class r,int N> strong_inline
 auto operator * (const iScalar<l>& lhs,const iMatrix<r,N>& rhs) -> iMatrix<decltype(lhs._internal*rhs._internal[0][0]),N>
 {
     typedef decltype(lhs._internal*rhs._internal[0][0]) ret_t;
@@ -148,7 +148,7 @@ auto operator * (const iScalar<l>& lhs,const iMatrix<r,N>& rhs) -> iMatrix<declt
     }}
     return ret;
 }
-template<class l,class r,int N> inline
+template<class l,class r,int N> strong_inline
 auto operator * (const iMatrix<l,N>& lhs,const iVector<r,N>& rhs) -> iVector<decltype(lhs._internal[0][0]*rhs._internal[0]),N>
 {
     typedef decltype(lhs._internal[0][0]*rhs._internal[0]) ret_t;
@@ -161,7 +161,7 @@ auto operator * (const iMatrix<l,N>& lhs,const iVector<r,N>& rhs) -> iVector<dec
     }
     return ret;
 }
-template<class l,class r,int N> inline
+template<class l,class r,int N> strong_inline
 auto operator * (const iScalar<l>& lhs,const iVector<r,N>& rhs) -> iVector<decltype(lhs._internal*rhs._internal[0]),N>
 {
     typedef decltype(lhs._internal*rhs._internal[0]) ret_t;
@@ -171,7 +171,7 @@ auto operator * (const iScalar<l>& lhs,const iVector<r,N>& rhs) -> iVector<declt
     }
     return ret;
 }
-template<class l,class r,int N> inline
+template<class l,class r,int N> strong_inline
 auto operator * (const iVector<l,N>& lhs,const iScalar<r>& rhs) -> iVector<decltype(lhs._internal[0]*rhs._internal),N>
 {
     typedef decltype(lhs._internal[0]*rhs._internal) ret_t;

From cbfa4097b40d94ec1bf7ff798819dd8fc6a0bb39 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:40:59 +0100
Subject: [PATCH 164/429] strong_inline forces ICPC to do it.

---
 lib/math/Grid_math_arith_scalar.h | 96 +++++++++++++++----------------
 1 file changed, 48 insertions(+), 48 deletions(-)

diff --git a/lib/math/Grid_math_arith_scalar.h b/lib/math/Grid_math_arith_scalar.h
index 69c9a169..c6418638 100644
--- a/lib/math/Grid_math_arith_scalar.h
+++ b/lib/math/Grid_math_arith_scalar.h
@@ -9,58 +9,58 @@ namespace Grid {
 //////////////////////////////////////////////////////////////////////////////////////////
 
 // multiplication by fundamental scalar type
-template<class l,int N> inline iScalar<l> operator * (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
+template<class l,int N> strong_inline iScalar<l> operator * (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
 {
   typename iScalar<l>::tensor_reduced srhs; srhs=rhs;
   return lhs*srhs;
 }
-template<class l,int N> inline iScalar<l> operator * (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
+template<class l,int N> strong_inline iScalar<l> operator * (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
 
-template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
+template<class l,int N> strong_inline iVector<l,N> operator * (const iVector<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
 {
   typename iVector<l,N>::tensor_reduced srhs; srhs=rhs;
   return lhs*srhs;
 }
-template<class l,int N> inline iVector<l,N> operator * (const typename iScalar<l>::scalar_type lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
+template<class l,int N> strong_inline iVector<l,N> operator * (const typename iScalar<l>::scalar_type lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
 
-template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type &rhs) 
+template<class l,int N> strong_inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type &rhs) 
 {
   typename iMatrix<l,N>::tensor_reduced srhs; srhs=rhs;
   return lhs*srhs;
 }
-template<class l,int N> inline iMatrix<l,N> operator * (const typename iScalar<l>::scalar_type & lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
+template<class l,int N> strong_inline iMatrix<l,N> operator * (const typename iScalar<l>::scalar_type & lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
 
 ////////////////////////////////////////////////////////////////////
 // Double support; cast to "scalar_type" through constructor
 ////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,double rhs) 
+template<class l> strong_inline iScalar<l> operator * (const iScalar<l>& lhs,double rhs) 
 {
   typename iScalar<l>::scalar_type t; t=rhs;
   typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs*srhs;
 }
-template<class l> inline iScalar<l> operator * (double lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
+template<class l> strong_inline iScalar<l> operator * (double lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
 
-template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,double rhs) 
+template<class l,int N> strong_inline iVector<l,N> operator * (const iVector<l,N>& lhs,double rhs) 
 {
   typename iScalar<l>::scalar_type t;t=rhs;
   typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs*srhs;
 }
-template<class l,int N> inline iVector<l,N> operator * (double lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
+template<class l,int N> strong_inline iVector<l,N> operator * (double lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
 
-template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,double rhs) 
+template<class l,int N> strong_inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,double rhs) 
 {
   typename iScalar<l>::scalar_type t;t=rhs;
   typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs*srhs;
 }
-template<class l,int N> inline iMatrix<l,N> operator * (double lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
+template<class l,int N> strong_inline iMatrix<l,N> operator * (double lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
 
 ////////////////////////////////////////////////////////////////////
 // Complex support; cast to "scalar_type" through constructor
 ////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,ComplexD rhs) 
+template<class l> strong_inline iScalar<l> operator * (const iScalar<l>& lhs,ComplexD rhs) 
 {
   typename iScalar<l>::scalar_type t;t=rhs;
   typename iScalar<l>::tensor_reduced srhs;srhs=t;
@@ -68,131 +68,131 @@ template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,ComplexD r
   
   return lhs*srhs;
 }
-template<class l> inline iScalar<l> operator * (ComplexD lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
+template<class l> strong_inline iScalar<l> operator * (ComplexD lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
 
-template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,ComplexD rhs) 
+template<class l,int N> strong_inline iVector<l,N> operator * (const iVector<l,N>& lhs,ComplexD rhs) 
 {
   typename iScalar<l>::scalar_type t;t=rhs;
   typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs*srhs;
 }
-template<class l,int N> inline iVector<l,N> operator * (ComplexD lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
+template<class l,int N> strong_inline iVector<l,N> operator * (ComplexD lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
 
-template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,ComplexD rhs) 
+template<class l,int N> strong_inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,ComplexD rhs) 
 {
   typename iScalar<l>::scalar_type t;t=rhs;
   typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs*srhs;
 }
-template<class l,int N> inline iMatrix<l,N> operator * (ComplexD lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
+template<class l,int N> strong_inline iMatrix<l,N> operator * (ComplexD lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
 
 ////////////////////////////////////////////////////////////////////
 // Integer support; cast to "scalar_type" through constructor
 ////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator * (const iScalar<l>& lhs,Integer rhs) 
+template<class l> strong_inline iScalar<l> operator * (const iScalar<l>& lhs,Integer rhs) 
 {
   typename iScalar<l>::scalar_type t;  t=rhs;
   typename iScalar<l>::tensor_reduced srhs; srhs=t;
   return lhs*srhs;
 }
-template<class l> inline iScalar<l> operator * (Integer lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
+template<class l> strong_inline iScalar<l> operator * (Integer lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
 
-template<class l,int N> inline iVector<l,N> operator * (const iVector<l,N>& lhs,Integer rhs) 
+template<class l,int N> strong_inline iVector<l,N> operator * (const iVector<l,N>& lhs,Integer rhs) 
 {
   typename iScalar<l>::scalar_type t;t=rhs;
   typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs*srhs;
 }
-template<class l,int N> inline iVector<l,N> operator * (Integer lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
+template<class l,int N> strong_inline iVector<l,N> operator * (Integer lhs,const iVector<l,N>& rhs) {  return rhs*lhs; }
 
-template<class l,int N> inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,Integer rhs) 
+template<class l,int N> strong_inline iMatrix<l,N> operator * (const iMatrix<l,N>& lhs,Integer rhs) 
 {
   typename iScalar<l>::scalar_type t;t=rhs;
   typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs*srhs;
 }
-template<class l,int N> inline iMatrix<l,N> operator * (Integer lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
+template<class l,int N> strong_inline iMatrix<l,N> operator * (Integer lhs,const iMatrix<l,N>& rhs) {  return rhs*lhs; }
 
 
 
 ///////////////////////////////////////////////////////////////////////////////////////////////
 // addition by fundamental scalar type applies to matrix(down diag) and scalar
 ///////////////////////////////////////////////////////////////////////////////////////////////
-template<class l,int N> inline iScalar<l> operator + (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
+template<class l,int N> strong_inline iScalar<l> operator + (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
 {
   typename iScalar<l>::tensor_reduced srhs; srhs=rhs;
   return lhs+srhs;
 }
-template<class l,int N> inline iScalar<l> operator + (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
+template<class l,int N> strong_inline iScalar<l> operator + (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
 
-template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
+template<class l,int N> strong_inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
 {
   typename iMatrix<l,N>::tensor_reduced srhs; srhs=rhs;
   return lhs+srhs;
 }
-template<class l,int N> inline iMatrix<l,N> operator + (const typename iScalar<l>::scalar_type lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
+template<class l,int N> strong_inline iMatrix<l,N> operator + (const typename iScalar<l>::scalar_type lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
 
 ////////////////////////////////////////////////////////////////////
 // Double support; cast to "scalar_type" through constructor
 ////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator + (const iScalar<l>& lhs,double rhs) 
+template<class l> strong_inline iScalar<l> operator + (const iScalar<l>& lhs,double rhs) 
 {
   typename iScalar<l>::scalar_type t; t=rhs;
   typename iScalar<l>::tensor_reduced srhs; srhs=t;
   return lhs+srhs;
 }
-template<class l> inline iScalar<l> operator + (double lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
+template<class l> strong_inline iScalar<l> operator + (double lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
 
-template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,double rhs) 
+template<class l,int N> strong_inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,double rhs) 
 {
   typename iScalar<l>::scalar_type t;t=rhs;
   typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs+srhs;
 }
-template<class l,int N> inline iMatrix<l,N> operator + (double lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
+template<class l,int N> strong_inline iMatrix<l,N> operator + (double lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
 
 
 // Integer support cast to scalar type through constructor
 
 
-template<class l> inline iScalar<l> operator + (const iScalar<l>& lhs,Integer rhs) 
+template<class l> strong_inline iScalar<l> operator + (const iScalar<l>& lhs,Integer rhs) 
 {
   typename iScalar<l>::scalar_type t; t=rhs;
   typename iScalar<l>::tensor_reduced srhs; srhs=t;
   return lhs+srhs;
 }
 
-template<class l> inline iScalar<l> operator + (Integer lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
+template<class l> strong_inline iScalar<l> operator + (Integer lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
 
-template<class l,int N> inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,Integer rhs) 
+template<class l,int N> strong_inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,Integer rhs) 
 {
   typename iScalar<l>::scalar_type t;t=rhs;
   typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs+srhs;
 }
-template<class l,int N> inline iMatrix<l,N> operator + (Integer lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
+template<class l,int N> strong_inline iMatrix<l,N> operator + (Integer lhs,const iMatrix<l,N>& rhs) {  return rhs+lhs; }
 
 
 ///////////////////////////////////////////////////////////////////////////////////////////////
 // subtraction of fundamental scalar type applies to matrix(down diag) and scalar
 ///////////////////////////////////////////////////////////////////////////////////////////////
-template<class l,int N> inline iScalar<l> operator - (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
+template<class l,int N> strong_inline iScalar<l> operator - (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
 {
   typename iScalar<l>::tensor_reduced srhs; srhs=rhs;
   return lhs-srhs;
 }
-template<class l,int N> inline iScalar<l> operator - (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) 
+template<class l,int N> strong_inline iScalar<l> operator - (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) 
 {
   typename iScalar<l>::tensor_reduced slhs;slhs=lhs;
   return slhs-rhs;
 }
 
-template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
+template<class l,int N> strong_inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
 {
   typename iScalar<l>::tensor_reduced srhs; srhs=rhs;
   return lhs-srhs;
 }
-template<class l,int N> inline iMatrix<l,N> operator - (const typename iScalar<l>::scalar_type lhs,const iMatrix<l,N>& rhs) 
+template<class l,int N> strong_inline iMatrix<l,N> operator - (const typename iScalar<l>::scalar_type lhs,const iMatrix<l,N>& rhs) 
 {
   typename iScalar<l>::tensor_reduced slhs;slhs=lhs;
   return slhs-rhs;
@@ -201,26 +201,26 @@ template<class l,int N> inline iMatrix<l,N> operator - (const typename iScalar<l
 ////////////////////////////////////////////////////////////////////
 // Double support; cast to "scalar_type" through constructor
 ////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator - (const iScalar<l>& lhs,double rhs) 
+template<class l> strong_inline iScalar<l> operator - (const iScalar<l>& lhs,double rhs) 
 {
   typename iScalar<l>::scalar_type t; t=rhs;
   typename iScalar<l>::tensor_reduced srhs; srhs=t;
   return lhs-srhs;
 }
-template<class l> inline iScalar<l> operator - (double lhs,const iScalar<l>& rhs) 
+template<class l> strong_inline iScalar<l> operator - (double lhs,const iScalar<l>& rhs) 
 {
   typename iScalar<l>::scalar_type t(lhs);
   typename iScalar<l>::tensor_reduced slhs;slhs=t;
   return slhs-rhs;
 }
 
-template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,double rhs) 
+template<class l,int N> strong_inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,double rhs) 
 {
   typename iScalar<l>::scalar_type t;t=rhs;
   typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs-srhs;
 }
-template<class l,int N> inline iMatrix<l,N> operator - (double lhs,const iMatrix<l,N>& rhs) 
+template<class l,int N> strong_inline iMatrix<l,N> operator - (double lhs,const iMatrix<l,N>& rhs) 
 {
   typename iScalar<l>::scalar_type t(lhs);
   typename iScalar<l>::tensor_reduced slhs;slhs=t;
@@ -230,25 +230,25 @@ template<class l,int N> inline iMatrix<l,N> operator - (double lhs,const iMatrix
 ////////////////////////////////////////////////////////////////////
 // Integer support; cast to "scalar_type" through constructor
 ////////////////////////////////////////////////////////////////////
-template<class l> inline iScalar<l> operator - (const iScalar<l>& lhs,Integer rhs) 
+template<class l> strong_inline iScalar<l> operator - (const iScalar<l>& lhs,Integer rhs) 
 {
   typename iScalar<l>::scalar_type t; t=rhs;
   typename iScalar<l>::tensor_reduced srhs; srhs=t;
   return lhs-srhs;
 }
-template<class l> inline iScalar<l> operator - (Integer lhs,const iScalar<l>& rhs) 
+template<class l> strong_inline iScalar<l> operator - (Integer lhs,const iScalar<l>& rhs) 
 {
   typename iScalar<l>::scalar_type t;t=lhs;
   typename iScalar<l>::tensor_reduced slhs;slhs=t;
   return slhs-rhs;
 }
-template<class l,int N> inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,Integer rhs) 
+template<class l,int N> strong_inline iMatrix<l,N> operator - (const iMatrix<l,N>& lhs,Integer rhs) 
 {
   typename iScalar<l>::scalar_type t;t=rhs;
   typename iScalar<l>::tensor_reduced srhs;srhs=t;
   return lhs-srhs;
 }
-template<class l,int N> inline iMatrix<l,N> operator - (Integer lhs,const iMatrix<l,N>& rhs) 
+template<class l,int N> strong_inline iMatrix<l,N> operator - (Integer lhs,const iMatrix<l,N>& rhs) 
 {
   typename iScalar<l>::scalar_type t;t=lhs;
   typename iScalar<l>::tensor_reduced slhs;slhs=t;

From af6e8f78292459bb1246ec993d0b8f581990d350 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:41:31 +0100
Subject: [PATCH 165/429] Force inlining on ICPC because inline apparently is
 not enoguh

---
 lib/math/Grid_math_arith_sub.h | 20 ++++++++++----------
 1 file changed, 10 insertions(+), 10 deletions(-)

diff --git a/lib/math/Grid_math_arith_sub.h b/lib/math/Grid_math_arith_sub.h
index de4ef9e5..c359349f 100644
--- a/lib/math/Grid_math_arith_sub.h
+++ b/lib/math/Grid_math_arith_sub.h
@@ -14,14 +14,14 @@ namespace Grid {
 // Vector +/- Vector
 // Matrix +/- Matrix
 // Matrix /- scalar
-template<class vtype,class ltype,class rtype> inline void sub(iScalar<vtype> * __restrict__ ret,
+template<class vtype,class ltype,class rtype> strong_inline void sub(iScalar<vtype> * __restrict__ ret,
                                                               const iScalar<ltype> * __restrict__ lhs,
                                                               const iScalar<rtype> * __restrict__ rhs)
 {
     sub(&ret->_internal,&lhs->_internal,&rhs->_internal);
 }
 
-template<class vtype,class ltype,class rtype,int N> inline void sub(iVector<vtype,N> * __restrict__ ret,
+template<class vtype,class ltype,class rtype,int N> strong_inline void sub(iVector<vtype,N> * __restrict__ ret,
                                                                     const iVector<ltype,N> * __restrict__ lhs,
                                                                     const iVector<rtype,N> * __restrict__ rhs)
 {
@@ -30,7 +30,7 @@ template<class vtype,class ltype,class rtype,int N> inline void sub(iVector<vtyp
     }
     return;
 }
-template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
+template<class vtype,class ltype,class rtype, int N> strong_inline void sub(iMatrix<vtype,N> * __restrict__ ret,
                                                                      const iMatrix<ltype,N> * __restrict__ lhs,
                                                                      const iMatrix<rtype,N> * __restrict__ rhs){
     for(int c2=0;c2<N;c2++){
@@ -39,7 +39,7 @@ template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vty
     }}
     return;
 }
-template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
+template<class vtype,class ltype,class rtype, int N> strong_inline void sub(iMatrix<vtype,N> * __restrict__ ret,
                                                                      const iScalar<ltype> * __restrict__ lhs,
                                                                      const iMatrix<rtype,N> * __restrict__ rhs){
     for(int c2=0;c2<N;c2++){
@@ -55,7 +55,7 @@ template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vty
     }}
     return;
 }
-template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vtype,N> * __restrict__ ret,
+template<class vtype,class ltype,class rtype, int N> strong_inline void sub(iMatrix<vtype,N> * __restrict__ ret,
                                                                      const iMatrix<ltype,N> * __restrict__ lhs,
                                                                      const iScalar<rtype> * __restrict__ rhs){
     for(int c2=0;c2<N;c2++){
@@ -69,7 +69,7 @@ template<class vtype,class ltype,class rtype, int N> inline void sub(iMatrix<vty
 }
 
     // - operator for scalar, vector, matrix
-template<class ltype,class rtype> inline auto
+template<class ltype,class rtype> strong_inline auto
 operator - (const iScalar<ltype>& lhs, const iScalar<rtype>& rhs) -> iScalar<decltype(lhs._internal - rhs._internal)>
 {
     typedef iScalar<decltype(lhs._internal-rhs._internal)> ret_t;
@@ -78,7 +78,7 @@ operator - (const iScalar<ltype>& lhs, const iScalar<rtype>& rhs) -> iScalar<dec
     return ret;
 }
 template<class ltype,class rtype,int N>
-inline auto operator - (const iVector<ltype,N>& lhs,const iVector<rtype,N>& rhs) ->iVector<decltype(lhs._internal[0]-rhs._internal[0]),N>
+strong_inline auto operator - (const iVector<ltype,N>& lhs,const iVector<rtype,N>& rhs) ->iVector<decltype(lhs._internal[0]-rhs._internal[0]),N>
 {
     typedef iVector<decltype(lhs._internal[0]-rhs._internal[0]),N> ret_t;
     ret_t ret;
@@ -86,7 +86,7 @@ inline auto operator - (const iVector<ltype,N>& lhs,const iVector<rtype,N>& rhs)
     return ret;
 }
 template<class ltype,class rtype,int N>
-inline auto operator - (const iMatrix<ltype,N>& lhs,const iMatrix<rtype,N>& rhs) ->iMatrix<decltype(lhs._internal[0][0]-rhs._internal[0][0]),N>
+strong_inline auto operator - (const iMatrix<ltype,N>& lhs,const iMatrix<rtype,N>& rhs) ->iMatrix<decltype(lhs._internal[0][0]-rhs._internal[0][0]),N>
 {
     typedef iMatrix<decltype(lhs._internal[0][0]-rhs._internal[0][0]),N> ret_t;
     ret_t ret;
@@ -94,7 +94,7 @@ inline auto operator - (const iMatrix<ltype,N>& lhs,const iMatrix<rtype,N>& rhs)
     return ret;
 }
 template<class ltype,class rtype,int N>
-inline auto operator - (const iScalar<ltype>& lhs,const iMatrix<rtype,N>& rhs)->iMatrix<decltype(lhs._internal-rhs._internal[0][0]),N>
+strong_inline auto operator - (const iScalar<ltype>& lhs,const iMatrix<rtype,N>& rhs)->iMatrix<decltype(lhs._internal-rhs._internal[0][0]),N>
 {
     typedef iMatrix<decltype(lhs._internal-rhs._internal[0][0]),N> ret_t;
     ret_t ret;
@@ -102,7 +102,7 @@ inline auto operator - (const iScalar<ltype>& lhs,const iMatrix<rtype,N>& rhs)->
     return ret;
 }
 template<class ltype,class rtype,int N>
-inline auto operator - (const iMatrix<ltype,N>& lhs,const iScalar<rtype>& rhs)->iMatrix<decltype(lhs._internal[0][0]-rhs._internal),N>
+strong_inline auto operator - (const iMatrix<ltype,N>& lhs,const iScalar<rtype>& rhs)->iMatrix<decltype(lhs._internal[0][0]-rhs._internal),N>
 {
     typedef iMatrix<decltype(lhs._internal[0][0]-rhs._internal),N> ret_t;
     ret_t ret;

From a26fdab7196be3c1a0356bd22c2cbc9b7210bac9 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:42:03 +0100
Subject: [PATCH 166/429] More elegant to do boolean logic inside the enable_if
 construct Should have done that from the beginning and should move this into
 a global edit

---
 lib/math/Grid_math_peek.h | 74 +++++++++++++--------------------------
 1 file changed, 25 insertions(+), 49 deletions(-)

diff --git a/lib/math/Grid_math_peek.h b/lib/math/Grid_math_peek.h
index b76fa84d..5c8d4a93 100644
--- a/lib/math/Grid_math_peek.h
+++ b/lib/math/Grid_math_peek.h
@@ -13,28 +13,22 @@ namespace Grid {
 //template<int Level> inline RealD peekIndex(const RealD arg) { return arg;}
 
 // Scalar peek, no indices
-template<int Level,class vtype> inline 
-  auto peekIndex(const iScalar<vtype> &arg) -> 
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value,  // Index matches
-  iScalar<vtype> >::type                              // return scalar
+template<int Level,class vtype,typename std::enable_if< iScalar<vtype>::TensorLevel == Level >::type * =nullptr> inline 
+  auto peekIndex(const iScalar<vtype> &arg) ->  iScalar<vtype> 
 {
   return arg;
 }
 // Vector peek, one index
-template<int Level,class vtype,int N> inline 
-  auto peekIndex(const iVector<vtype,N> &arg,int i) -> 
-  typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::value,  // Index matches
-  iScalar<vtype> >::type                              // return scalar
+template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::TensorLevel == Level >::type * =nullptr> inline 
+  auto peekIndex(const iVector<vtype,N> &arg,int i) -> iScalar<vtype> // Index matches
 {
   iScalar<vtype> ret;                              // return scalar
   ret._internal = arg._internal[i];
   return ret;
 }
 // Matrix peek, two indices
-template<int Level,class vtype,int N> inline 
-  auto peekIndex(const iMatrix<vtype,N> &arg,int i,int j) -> 
-  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value,  // Index matches
-  iScalar<vtype> >::type                              // return scalar
+template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::TensorLevel == Level >::type * =nullptr> inline 
+  auto peekIndex(const iMatrix<vtype,N> &arg,int i,int j) ->  iScalar<vtype>
 {
   iScalar<vtype> ret;                              // return scalar
   ret._internal = arg._internal[i][j];
@@ -45,38 +39,30 @@ template<int Level,class vtype,int N> inline
 // No match peek for scalar,vector,matrix must forward on either 0,1,2 args. Must have 9 routines with notvalue
 /////////////
 // scalar
-template<int Level,class vtype> inline 
-  auto peekIndex(const iScalar<vtype> &arg) ->                                     // Scalar 0 index  
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does NOT match
-  iScalar<decltype(peekIndex<Level>(arg._internal))> >::type                       
+template<int Level,class vtype,typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+  auto peekIndex(const iScalar<vtype> &arg) -> iScalar<decltype(peekIndex<Level>(arg._internal))>
 {
   iScalar<decltype(peekIndex<Level>(arg._internal))> ret;
   ret._internal= peekIndex<Level>(arg._internal);
   return ret;
 }
-template<int Level,class vtype> inline 
-  auto peekIndex(const iScalar<vtype> &arg,int i) ->                             // Scalar 1 index
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does NOT match
-  iScalar<decltype(peekIndex<Level>(arg._internal,i))> >::type                       
+template<int Level,class vtype, typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+  auto peekIndex(const iScalar<vtype> &arg,int i) ->  iScalar<decltype(peekIndex<Level>(arg._internal,i))> 
 {
   iScalar<decltype(peekIndex<Level>(arg._internal,i))> ret;
   ret._internal=peekIndex<Level>(arg._internal,i);
   return ret;
 }
-template<int Level,class vtype> inline 
-  auto peekIndex(const iScalar<vtype> &arg,int i,int j) ->                         // Scalar, 2 index
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does NOT match
-  iScalar<decltype(peekIndex<Level>(arg._internal,i,j))> >::type                       
+template<int Level,class vtype, typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+  auto peekIndex(const iScalar<vtype> &arg,int i,int j) ->  iScalar<decltype(peekIndex<Level>(arg._internal,i,j))>
 {
   iScalar<decltype(peekIndex<Level>(arg._internal,i,j))> ret;
   ret._internal=peekIndex<Level>(arg._internal,i,j);
   return ret;
 }
 // vector
-template<int Level,class vtype,int N> inline 
-auto peekIndex(const iVector<vtype,N> &arg) -> 
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does not match
-  iVector<decltype(peekIndex<Level>(arg._internal[0])),N> >::type                       
+template<int Level,class vtype,int N, typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+auto peekIndex(const iVector<vtype,N> &arg) ->   iVector<decltype(peekIndex<Level>(arg._internal[0])),N>
 {
   iVector<decltype(peekIndex<Level>(arg._internal[0])),N> ret;
   for(int ii=0;ii<N;ii++){
@@ -84,10 +70,8 @@ auto peekIndex(const iVector<vtype,N> &arg) ->
   }
   return ret;
 }
-template<int Level,class vtype,int N> inline 
-  auto peekIndex(const iVector<vtype,N> &arg,int i) -> 
-  typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,  // Index does not match
-  iVector<decltype(peekIndex<Level>(arg._internal[0],i)),N> >::type                       
+template<int Level,class vtype,int N, typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+  auto peekIndex(const iVector<vtype,N> &arg,int i) ->  iVector<decltype(peekIndex<Level>(arg._internal[0],i)),N>
 {
   iVector<decltype(peekIndex<Level>(arg._internal[0],i)),N> ret;
   for(int ii=0;ii<N;ii++){
@@ -95,10 +79,8 @@ template<int Level,class vtype,int N> inline
   }
   return ret;
 }
-template<int Level,class vtype,int N> inline 
-  auto peekIndex(const iVector<vtype,N> &arg,int i,int j) -> 
-  typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,  // Index does not match
-  iVector<decltype(peekIndex<Level>(arg._internal[0],i,j)),N> >::type                       
+template<int Level,class vtype,int N, typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+  auto peekIndex(const iVector<vtype,N> &arg,int i,int j) ->  iVector<decltype(peekIndex<Level>(arg._internal[0],i,j)),N> 
 {
   iVector<decltype(peekIndex<Level>(arg._internal[0],i,j)),N> ret;
   for(int ii=0;ii<N;ii++){
@@ -107,10 +89,8 @@ template<int Level,class vtype,int N> inline
   return ret;
 }
 // matrix
-template<int Level,class vtype,int N> inline 
-auto peekIndex(const iMatrix<vtype,N> &arg) -> 
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,  // Index does not match
-  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0])),N> >::type                       
+template<int Level,class vtype,int N, typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+auto peekIndex(const iMatrix<vtype,N> &arg) ->   iMatrix<decltype(peekIndex<Level>(arg._internal[0][0])),N> 
 {
   iMatrix<decltype(peekIndex<Level>(arg._internal[0][0])),N> ret;
   for(int ii=0;ii<N;ii++){
@@ -119,22 +99,18 @@ auto peekIndex(const iMatrix<vtype,N> &arg) ->
   }}
   return ret;
 }
-template<int Level,class vtype,int N> inline 
-  auto peekIndex(const iMatrix<vtype,N> &arg,int i) -> 
-  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,  // Index does not match
-  iMatrix<decltype(peekIndex<Level>(arg._internal[0],i)),N> >::type                       
+template<int Level,class vtype,int N, typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+  auto peekIndex(const iMatrix<vtype,N> &arg,int i) ->   iMatrix<decltype(peekIndex<Level>(arg._internal[0][0],i)),N>
 {
-  iMatrix<decltype(peekIndex<Level>(arg._internal[0],i)),N> ret;
+  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0],i)),N> ret;
   for(int ii=0;ii<N;ii++){
   for(int jj=0;jj<N;jj++){
     ret._internal[ii][jj]=peekIndex<Level>(arg._internal[ii][jj],i);
   }}
   return ret;
 }
-template<int Level,class vtype,int N> inline 
-  auto peekIndex(const iMatrix<vtype,N> &arg,int i,int j) -> 
-  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,  // Index does not match
-  iMatrix<decltype(peekIndex<Level>(arg._internal[0][0],i,j)),N> >::type                       
+template<int Level,class vtype,int N, typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+  auto peekIndex(const iMatrix<vtype,N> &arg,int i,int j) ->   iMatrix<decltype(peekIndex<Level>(arg._internal[0][0],i,j)),N>
 {
   iMatrix<decltype(peekIndex<Level>(arg._internal[0][0],i,j)),N> ret;
   for(int ii=0;ii<N;ii++){

From a852d13f03ff82f3bb86b08e4b5bae6e4a1be558 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:42:51 +0100
Subject: [PATCH 167/429] More elegant enable_if

---
 lib/math/Grid_math_poke.h | 73 +++++++++++++--------------------------
 1 file changed, 24 insertions(+), 49 deletions(-)

diff --git a/lib/math/Grid_math_poke.h b/lib/math/Grid_math_poke.h
index 3f1cb89e..a015478e 100644
--- a/lib/math/Grid_math_poke.h
+++ b/lib/math/Grid_math_poke.h
@@ -7,23 +7,20 @@ namespace Grid {
 //////////////////////////////////////////////////////////////////////////////
 
 // Scalar poke
-template<int Level,class vtype> inline 
-  void pokeIndex(iScalar<vtype> &ret, 
-		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value,iScalar<vtype> >::type &arg)
+template<int Level,class vtype,typename std::enable_if< iScalar<vtype>::TensorLevel == Level >::type * =nullptr> inline 
+  void pokeIndex(iScalar<vtype> &ret, const iScalar<vtype> &arg)
 {
   ret._internal = arg._internal;
 }
 // Vector poke, one index
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iVector<vtype,N> &ret, 
-		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::value,iScalar<vtype> >::type &arg,int i)
+template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::TensorLevel == Level >::type * =nullptr> inline 
+  void pokeIndex(iVector<vtype,N> &ret, const iScalar<vtype> &arg,int i)
 {
   ret._internal[i] = arg._internal;
 }
 // Vector poke, two indices
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iMatrix<vtype,N> &ret, 
-		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value,iScalar<vtype> >::type &arg,int i,int j)
+template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::TensorLevel == Level >::type * =nullptr> inline 
+  void pokeIndex(iMatrix<vtype,N> &ret, const iScalar<vtype> &arg,int i,int j)
 {
   ret._internal[i][j] = arg._internal;
 }
@@ -32,55 +29,41 @@ template<int Level,class vtype,int N> inline
 // No match poke for scalar,vector,matrix must forward on either 0,1,2 args. Must have 9 routines with notvalue
 /////////////
 // scalar
-template<int Level,class vtype> inline 
-  void pokeIndex(iScalar<vtype> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,iScalar<decltype(peekIndex<Level>(ret._internal))> >::type &arg)
+template<int Level,class vtype,typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+void pokeIndex(iScalar<vtype> &ret, const iScalar<decltype(peekIndex<Level>(ret._internal))>  &arg)
 		 
 {
   pokeIndex<Level>(ret._internal,arg._internal);
 }
-template<int Level,class vtype> inline 
-  void pokeIndex(iScalar<vtype> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,iScalar<decltype(peekIndex<Level>(ret._internal,0))> >::type &arg,
-		 int i)
+template<int Level,class vtype,typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+  void pokeIndex(iScalar<vtype> &ret, const iScalar<decltype(peekIndex<Level>(ret._internal,0))> &arg, int i)
 		 
 {
   pokeIndex<Level>(ret._internal,arg._internal,i);
 }
-template<int Level,class vtype> inline 
-  void pokeIndex(iScalar<vtype> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue,iScalar<decltype(peekIndex<Level>(ret._internal,0,0))> >::type &arg,
-		 int i,int j)
-		 
+template<int Level,class vtype,typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+  void pokeIndex(iScalar<vtype> &ret, const iScalar<decltype(peekIndex<Level>(ret._internal,0,0))> &arg,int i,int j)
 {
   pokeIndex<Level>(ret._internal,arg._internal,i,j);
 }
 
 // Vector
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iVector<vtype,N> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,iVector<decltype(peekIndex<Level>(ret._internal)),N> >::type &arg)
-		 
+template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+  void pokeIndex(iVector<vtype,N> &ret, iVector<decltype(peekIndex<Level>(ret._internal)),N>  &arg)
 {
   for(int ii=0;ii<N;ii++){
     pokeIndex<Level>(ret._internal[ii],arg._internal[ii]);
   }
 }
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iVector<vtype,N> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,iVector<decltype(peekIndex<Level>(ret._internal,0)),N> >::type &arg,
-		 int i)
-		 
+template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+  void pokeIndex(iVector<vtype,N> &ret, const iVector<decltype(peekIndex<Level>(ret._internal,0)),N> &arg,int i)
 {
   for(int ii=0;ii<N;ii++){
     pokeIndex<Level>(ret._internal[ii],arg._internal[ii],i);
   }
 }
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iVector<vtype,N> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,Level>::notvalue,iVector<decltype(peekIndex<Level>(ret._internal,0,0)),N> >::type &arg,
-		 int i,int j)
-		 
+template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+  void pokeIndex(iVector<vtype,N> &ret, const iVector<decltype(peekIndex<Level>(ret._internal,0,0)),N> &arg,int i,int j)
 {
   for(int ii=0;ii<N;ii++){
     pokeIndex<Level>(ret._internal[ii],arg._internal[ii],i,j);
@@ -88,32 +71,24 @@ template<int Level,class vtype,int N> inline
 }
 
 // Matrix
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iMatrix<vtype,N> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,iMatrix<decltype(peekIndex<Level>(ret._internal)),N> >::type &arg)
-		 
+template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+  void pokeIndex(iMatrix<vtype,N> &ret, const iMatrix<decltype(peekIndex<Level>(ret._internal)),N> &arg)		 
 {
   for(int ii=0;ii<N;ii++){
   for(int jj=0;jj<N;jj++){
     pokeIndex<Level>(ret._internal[ii][jj],arg._internal[ii][jj]);
   }}
 }
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iMatrix<vtype,N> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,iMatrix<decltype(peekIndex<Level>(ret._internal,0)),N> >::type &arg,
-		 int i)
-		 
+template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+  void pokeIndex(iMatrix<vtype,N> &ret, const iMatrix<decltype(peekIndex<Level>(ret._internal,0)),N> &arg,int i)
 {
   for(int ii=0;ii<N;ii++){
   for(int jj=0;jj<N;jj++){
     pokeIndex<Level>(ret._internal[ii][jj],arg._internal[ii][jj],i);
   }}
 }
-template<int Level,class vtype,int N> inline 
-  void pokeIndex(iMatrix<vtype,N> &ret,
-		 const typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,iMatrix<decltype(peekIndex<Level>(ret._internal,0,0)),N> >::type &arg,
-		 int i,int j)
-		 
+template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+  void pokeIndex(iMatrix<vtype,N> &ret, const iMatrix<decltype(peekIndex<Level>(ret._internal,0,0)),N> &arg, int i,int j)
 {
   for(int ii=0;ii<N;ii++){
   for(int jj=0;jj<N;jj++){

From bd721ce1c8ed1111c25d6fc92f82becc91897c3a Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:43:20 +0100
Subject: [PATCH 168/429] Force inlining upon icpc

---
 lib/math/Grid_math_tensors.h | 72 ++++++++++++++++++------------------
 1 file changed, 36 insertions(+), 36 deletions(-)

diff --git a/lib/math/Grid_math_tensors.h b/lib/math/Grid_math_tensors.h
index 8b3e4b1b..a0424576 100644
--- a/lib/math/Grid_math_tensors.h
+++ b/lib/math/Grid_math_tensors.h
@@ -42,46 +42,46 @@ public:
       zeroit(*this);
       return *this;
     }
-    friend void vstream(iScalar<vtype> &out,const iScalar<vtype> &in){
+    friend strong_inline void vstream(iScalar<vtype> &out,const iScalar<vtype> &in){
       vstream(out._internal,in._internal);
     }
 
 
-    friend void zeroit(iScalar<vtype> &that){
+    friend strong_inline void zeroit(iScalar<vtype> &that){
         zeroit(that._internal);
     }
-    friend void prefetch(iScalar<vtype> &that){
+    friend strong_inline void prefetch(iScalar<vtype> &that){
       prefetch(that._internal);
     }
-    friend void permute(iScalar<vtype> &out,const iScalar<vtype> &in,int permutetype){
+    friend strong_inline void permute(iScalar<vtype> &out,const iScalar<vtype> &in,int permutetype){
       permute(out._internal,in._internal,permutetype);
     }
 
     // Unary negation
-    friend inline iScalar<vtype> operator -(const iScalar<vtype> &r) {
+    friend strong_inline iScalar<vtype> operator -(const iScalar<vtype> &r) {
         iScalar<vtype> ret;
         ret._internal= -r._internal;
         return ret;
     }
     // *=,+=,-= operators inherit from corresponding "*,-,+" behaviour
-    inline iScalar<vtype> &operator *=(const iScalar<vtype> &r) {
+    strong_inline iScalar<vtype> &operator *=(const iScalar<vtype> &r) {
         *this = (*this)*r;
         return *this;
     }
-    inline iScalar<vtype> &operator -=(const iScalar<vtype> &r) {
+    strong_inline iScalar<vtype> &operator -=(const iScalar<vtype> &r) {
         *this = (*this)-r;
         return *this;
     }
-    inline iScalar<vtype> &operator +=(const iScalar<vtype> &r) {
+    strong_inline iScalar<vtype> &operator +=(const iScalar<vtype> &r) {
         *this = (*this)+r;
         return *this;
     }
     
-    inline vtype & operator ()(void) {
+    strong_inline vtype & operator ()(void) {
       return _internal;
     }
 
-    inline const vtype & operator ()(void) const {
+    strong_inline const vtype & operator ()(void) const {
       return _internal;
     }
 
@@ -89,7 +89,7 @@ public:
     operator RealD () const { return(real(TensorRemove(_internal))); }
 
     // convert from a something to a scalar
-    template<class T,typename std::enable_if<!isGridTensor<T>::value, T>::type* = nullptr > inline auto operator = (T arg) -> iScalar<vtype>
+    template<class T,typename std::enable_if<!isGridTensor<T>::value, T>::type* = nullptr > strong_inline auto operator = (T arg) -> iScalar<vtype>
     { 
       _internal = vtype(arg);
       return *this;
@@ -103,8 +103,8 @@ public:
 ///////////////////////////////////////////////////////////
 // Allows to turn scalar<scalar<scalar<double>>>> back to double.
 ///////////////////////////////////////////////////////////
-template<class T>     inline typename std::enable_if<!isGridTensor<T>::value, T>::type TensorRemove(T arg) { return arg;}
-template<class vtype> inline auto TensorRemove(iScalar<vtype> arg) -> decltype(TensorRemove(arg._internal))
+template<class T>     strong_inline typename std::enable_if<!isGridTensor<T>::value, T>::type TensorRemove(T arg) { return arg;}
+template<class vtype> strong_inline auto TensorRemove(iScalar<vtype> arg) -> decltype(TensorRemove(arg._internal))
 {
   return TensorRemove(arg._internal);
 }
@@ -130,48 +130,48 @@ public:
         zeroit(*this);
         return *this;
     }
-    friend void zeroit(iVector<vtype,N> &that){
+    friend strong_inline void zeroit(iVector<vtype,N> &that){
         for(int i=0;i<N;i++){
             zeroit(that._internal[i]);
         }
     }
-    friend void prefetch(iVector<vtype,N> &that){
+    friend strong_inline void prefetch(iVector<vtype,N> &that){
       for(int i=0;i<N;i++) prefetch(that._internal[i]);
     }
-    friend void vstream(iVector<vtype,N> &out,const iVector<vtype,N> &in){
+    friend strong_inline void vstream(iVector<vtype,N> &out,const iVector<vtype,N> &in){
       for(int i=0;i<N;i++){
 	vstream(out._internal[i],in._internal[i]);
       }
     }
 
-    friend void permute(iVector<vtype,N> &out,const iVector<vtype,N> &in,int permutetype){
+    friend strong_inline void permute(iVector<vtype,N> &out,const iVector<vtype,N> &in,int permutetype){
       for(int i=0;i<N;i++){
 	permute(out._internal[i],in._internal[i],permutetype);
       }
     }
     // Unary negation
-    friend inline iVector<vtype,N> operator -(const iVector<vtype,N> &r) {
+    friend strong_inline iVector<vtype,N> operator -(const iVector<vtype,N> &r) {
         iVector<vtype,N> ret;
         for(int i=0;i<N;i++) ret._internal[i]= -r._internal[i];
         return ret;
     }
     // *=,+=,-= operators inherit from corresponding "*,-,+" behaviour
-    inline iVector<vtype,N> &operator *=(const iScalar<vtype> &r) {
+    strong_inline iVector<vtype,N> &operator *=(const iScalar<vtype> &r) {
         *this = (*this)*r;
         return *this;
     }
-    inline iVector<vtype,N> &operator -=(const iVector<vtype,N> &r) {
+    strong_inline iVector<vtype,N> &operator -=(const iVector<vtype,N> &r) {
         *this = (*this)-r;
         return *this;
     }
-    inline iVector<vtype,N> &operator +=(const iVector<vtype,N> &r) {
+    strong_inline iVector<vtype,N> &operator +=(const iVector<vtype,N> &r) {
         *this = (*this)+r;
         return *this;
     }
-    inline vtype & operator ()(int i) {
+    strong_inline vtype & operator ()(int i) {
       return _internal[i];
     }
-    inline const vtype & operator ()(int i) const {
+    strong_inline const vtype & operator ()(int i) const {
       return _internal[i];
     }
     friend std::ostream& operator<< (std::ostream& stream, const iVector<vtype,N> &o){
@@ -183,7 +183,7 @@ public:
       stream<<"}";
       return stream;
     };
-    //    inline vtype && operator ()(int i) {
+    //    strong_inline vtype && operator ()(int i) {
     //      return _internal[i];
     //    }
 };
@@ -211,7 +211,7 @@ public:
     zeroit(*this);
     return *this;
   }
-  template<class T,typename std::enable_if<!isGridTensor<T>::value, T>::type* = nullptr > inline auto operator = (T arg) -> iMatrix<vtype,N>
+  template<class T,typename std::enable_if<!isGridTensor<T>::value, T>::type* = nullptr > strong_inline auto operator = (T arg) -> iMatrix<vtype,N>
     { 
       zeroit(*this);
       for(int i=0;i<N;i++)
@@ -219,31 +219,31 @@ public:
       return *this;
     }
 
-  friend void zeroit(iMatrix<vtype,N> &that){
+  friend strong_inline void zeroit(iMatrix<vtype,N> &that){
     for(int i=0;i<N;i++){
       for(int j=0;j<N;j++){
 	zeroit(that._internal[i][j]);
     }}
   }
-  friend void prefetch(iMatrix<vtype,N> &that){
+  friend strong_inline void prefetch(iMatrix<vtype,N> &that){
     for(int i=0;i<N;i++) 
     for(int j=0;j<N;j++) 
       prefetch(that._internal[i][j]);
   }
-  friend void vstream(iMatrix<vtype,N> &out,const iMatrix<vtype,N> &in){
+  friend strong_inline void vstream(iMatrix<vtype,N> &out,const iMatrix<vtype,N> &in){
       for(int i=0;i<N;i++){
       for(int j=0;j<N;j++){
 	vstream(out._internal[i][j],in._internal[i][j]);
       }}
     }
-  friend void permute(iMatrix<vtype,N> &out,const iMatrix<vtype,N> &in,int permutetype){
+  friend strong_inline void permute(iMatrix<vtype,N> &out,const iMatrix<vtype,N> &in,int permutetype){
     for(int i=0;i<N;i++){
       for(int j=0;j<N;j++){
 	permute(out._internal[i][j],in._internal[i][j],permutetype);
     }}
   }
   // Unary negation
-  friend inline iMatrix<vtype,N> operator -(const iMatrix<vtype,N> &r) {
+  friend strong_inline iMatrix<vtype,N> operator -(const iMatrix<vtype,N> &r) {
     iMatrix<vtype,N> ret;
     for(int i=0;i<N;i++){
       for(int j=0;j<N;j++){
@@ -253,26 +253,26 @@ public:
   }
   // *=,+=,-= operators inherit from corresponding "*,-,+" behaviour
   template<class T>
-  inline iMatrix<vtype,N> &operator *=(const T &r) {
+  strong_inline iMatrix<vtype,N> &operator *=(const T &r) {
     *this = (*this)*r;
     return *this;
   }
   template<class T>
-  inline iMatrix<vtype,N> &operator -=(const T &r) {
+  strong_inline iMatrix<vtype,N> &operator -=(const T &r) {
     *this = (*this)-r;
     return *this;
   }
   template<class T>
-  inline iMatrix<vtype,N> &operator +=(const T &r) {
+  strong_inline iMatrix<vtype,N> &operator +=(const T &r) {
     *this = (*this)+r;
     return *this;
   }
 
   // returns an lvalue reference
-  inline vtype & operator ()(int i,int j) {
+  strong_inline vtype & operator ()(int i,int j) {
     return _internal[i][j];
   }
-  inline const vtype & operator ()(int i,int j) const {
+  strong_inline const vtype & operator ()(int i,int j) const {
     return _internal[i][j];
   }
   friend std::ostream& operator<< (std::ostream& stream, const iMatrix<vtype,N> &o){
@@ -289,7 +289,7 @@ public:
     return stream;
   };
 
-  //  inline vtype && operator ()(int i,int j) {
+  //  strong_inline vtype && operator ()(int i,int j) {
   //    return _internal[i][j];
   //  }
 

From 0e7945fe546bb96d5a0f71116ce3abe94ec8909b Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:43:49 +0100
Subject: [PATCH 169/429] Forces inlining upon icpc

---
 lib/qcd/Grid_qcd_2spinor.h | 273 +++++++++++++++++++------------------
 1 file changed, 137 insertions(+), 136 deletions(-)

diff --git a/lib/qcd/Grid_qcd_2spinor.h b/lib/qcd/Grid_qcd_2spinor.h
index fe6a1da8..97de13cf 100644
--- a/lib/qcd/Grid_qcd_2spinor.h
+++ b/lib/qcd/Grid_qcd_2spinor.h
@@ -41,7 +41,8 @@ namespace QCD {
    *  0 -i 0  0
    * -i 0  0  0
    */
-  template<class vtype> inline void 
+
+  template<class vtype> strong_inline void 
     spProjXp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
       // To fail is not to err (Cryptic clue: suggest to Google SFINAE ;) )
@@ -49,7 +50,7 @@ namespace QCD {
       hspin(0)=fspin(0)+timesI(fspin(3));
       hspin(1)=fspin(1)+timesI(fspin(2));
     }
-  template<class vtype> inline void spProjXm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype> strong_inline void spProjXm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)-timesI(fspin(3));
@@ -60,13 +61,13 @@ namespace QCD {
       //  0 0  1  0   [1] +- [2]
       //  0 1  0  0
       // -1 0  0  0
-  template<class vtype> inline void spProjYp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype> strong_inline void spProjYp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)-fspin(3);
       hspin(1)=fspin(1)+fspin(2);
     }
-  template<class vtype> inline void spProjYm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype> strong_inline void spProjYm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)+fspin(3);
@@ -78,13 +79,13 @@ namespace QCD {
 	     * -i 0  0  0
 	     *  0 i  0  0
 	     */
-  template<class vtype> inline void spProjZp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype> strong_inline void spProjZp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)+timesI(fspin(2));
       hspin(1)=fspin(1)-timesI(fspin(3));
     }
-  template<class vtype> inline void spProjZm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype> strong_inline void spProjZm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)-timesI(fspin(2));
@@ -96,13 +97,13 @@ namespace QCD {
 	     *  1 0  0  0
 	     *  0 1  0  0
 	     */
-  template<class vtype> inline void spProjTp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype> strong_inline void spProjTp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)+fspin(2);
       hspin(1)=fspin(1)+fspin(3);
     }
-  template<class vtype> inline void spProjTm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype> strong_inline void spProjTm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)-fspin(2);
@@ -115,21 +116,21 @@ namespace QCD {
 	     *  0 0  0 -1
 	     */
 
-  template<class vtype> inline void spProj5p (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype> strong_inline void spProj5p (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0);
       hspin(1)=fspin(1);
     }
-  template<class vtype> inline void spProj5m (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype> strong_inline void spProj5m (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(2);
       hspin(1)=fspin(3);
     }
   
-  //  template<class vtype> inline void fspProj5p (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
-  template<class vtype> inline void spProj5p (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
+  //  template<class vtype> strong_inline void fspProj5p (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype> strong_inline void spProj5p (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       rfspin(0)=fspin(0);
@@ -137,8 +138,8 @@ namespace QCD {
       rfspin(2)=zero;
       rfspin(3)=zero;
     }
-  //  template<class vtype> inline void fspProj5m (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
-  template<class vtype> inline void spProj5m (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
+  //  template<class vtype> strong_inline void fspProj5m (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype> strong_inline void spProj5m (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       rfspin(0)=zero;
@@ -156,7 +157,7 @@ namespace QCD {
    *  0 -i 0  0  -i[1]+-[2]   == -i ([0]+-i[3]) = -i (1)
    * -i 0  0  0  -i[0]+-[3]   == -i ([1]+-i[2]) = -i (0)
    */
-  template<class vtype> inline void spReconXp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void spReconXp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
@@ -164,7 +165,7 @@ namespace QCD {
       fspin(2)=timesMinusI(hspin(1));
       fspin(3)=timesMinusI(hspin(0));
     }
-  template<class vtype> inline void spReconXm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void spReconXm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
@@ -172,7 +173,7 @@ namespace QCD {
       fspin(2)=timesI(hspin(1));
       fspin(3)=timesI(hspin(0));
     }
-  template<class vtype> inline void accumReconXp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void accumReconXp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
@@ -180,7 +181,7 @@ namespace QCD {
       fspin(2)-=timesI(hspin(1));
       fspin(3)-=timesI(hspin(0));
     }
-  template<class vtype> inline void accumReconXm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void accumReconXm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
@@ -194,7 +195,7 @@ namespace QCD {
       //  0 1  0  0              == 1(1)
       // -1 0  0  0              ==-1(0)
 
-  template<class vtype> inline void spReconYp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void spReconYp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
@@ -202,7 +203,7 @@ namespace QCD {
       fspin(2)= hspin(1);
       fspin(3)=-hspin(0);//Unary minus?
     }
-  template<class vtype> inline void spReconYm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void spReconYm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
@@ -210,7 +211,7 @@ namespace QCD {
       fspin(2)=-hspin(1);
       fspin(3)= hspin(0);
     }
-  template<class vtype> inline void accumReconYp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void accumReconYp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
@@ -218,7 +219,7 @@ namespace QCD {
       fspin(2)+=hspin(1);
       fspin(3)-=hspin(0);
     }
-  template<class vtype> inline void accumReconYm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void accumReconYm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
@@ -233,7 +234,7 @@ namespace QCD {
 	     * -i 0  0  0     => -i (0)
 	     *  0 i  0  0     =>  i (1)
 	     */
-  template<class vtype> inline void spReconZp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void spReconZp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
@@ -241,7 +242,7 @@ namespace QCD {
       fspin(2)=timesMinusI(hspin(0));
       fspin(3)=timesI(hspin(1));
     }
-  template<class vtype> inline void spReconZm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void spReconZm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
@@ -249,7 +250,7 @@ namespace QCD {
       fspin(2)=     timesI(hspin(0));
       fspin(3)=timesMinusI(hspin(1));
     }
-  template<class vtype> inline void accumReconZp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void accumReconZp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
@@ -257,7 +258,7 @@ namespace QCD {
       fspin(2)-=timesI(hspin(0));
       fspin(3)+=timesI(hspin(1));
     }
-  template<class vtype> inline void accumReconZm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void accumReconZm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
@@ -271,7 +272,7 @@ namespace QCD {
 	     *  1 0  0  0    => (0)
 	     *  0 1  0  0    => (1)
 	     */
-  template<class vtype> inline void spReconTp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void spReconTp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
@@ -279,7 +280,7 @@ namespace QCD {
       fspin(2)=hspin(0);
       fspin(3)=hspin(1);
     }
-  template<class vtype> inline void spReconTm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void spReconTm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
@@ -287,7 +288,7 @@ namespace QCD {
       fspin(2)=-hspin(0);
       fspin(3)=-hspin(1);
     }
-  template<class vtype> inline void accumReconTp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void accumReconTp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
@@ -295,7 +296,7 @@ namespace QCD {
       fspin(2)+=hspin(0);
       fspin(3)+=hspin(1);
     }
-  template<class vtype> inline void accumReconTm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void accumReconTm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
@@ -309,7 +310,7 @@ namespace QCD {
 	     *  0 0 -1  0
 	     *  0 0  0 -1
 	     */
-  template<class vtype> inline void spRecon5p (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void spRecon5p (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0)+hspin(0); // add is lower latency than mul
@@ -317,7 +318,7 @@ namespace QCD {
       fspin(2)=zero;
       fspin(3)=zero;
     }
-  template<class vtype> inline void spRecon5m (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void spRecon5m (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=zero;
@@ -325,13 +326,13 @@ namespace QCD {
       fspin(2)=hspin(0)+hspin(0);
       fspin(3)=hspin(1)+hspin(1);
     }
-  template<class vtype> inline void accumRecon5p (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void accumRecon5p (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0)+hspin(0);
       fspin(1)+=hspin(1)+hspin(1);
     }
-  template<class vtype> inline void accumRecon5m (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype> strong_inline void accumRecon5m (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(2)+=hspin(0)+hspin(0);
@@ -345,19 +346,19 @@ namespace QCD {
   //////////
   // Xp
   //////////
-  template<class rtype,class vtype> inline void spProjXp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spProjXp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjXp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spProjXp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjXp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjXp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spProjXp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjXp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -367,19 +368,19 @@ namespace QCD {
   }
 
 
-  template<class rtype,class vtype> inline void spReconXp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spReconXp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconXp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spReconXp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconXp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconXp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spReconXp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconXp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -388,19 +389,19 @@ namespace QCD {
     }}
   }
 
-  template<class rtype,class vtype> inline void accumReconXp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void accumReconXp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconXp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void accumReconXp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconXp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconXp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void accumReconXp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconXp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -414,19 +415,19 @@ namespace QCD {
   ////////
   // Xm
   ////////
-  template<class rtype,class vtype> inline void spProjXm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spProjXm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjXm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spProjXm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjXm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjXm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spProjXm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjXm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -436,19 +437,19 @@ namespace QCD {
   }
 
 
-  template<class rtype,class vtype> inline void spReconXm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spReconXm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconXm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spReconXm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconXm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconXm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spReconXm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconXm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -457,19 +458,19 @@ namespace QCD {
     }}
   }
 
-  template<class rtype,class vtype> inline void accumReconXm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void accumReconXm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconXm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void accumReconXm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconXm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconXm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void accumReconXm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconXm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -483,19 +484,19 @@ namespace QCD {
   ////////
   // Yp
   ////////
-  template<class rtype,class vtype> inline void spProjYp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spProjYp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjYp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spProjYp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjYp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjYp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spProjYp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjYp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -505,19 +506,19 @@ namespace QCD {
   }
 
 
-  template<class rtype,class vtype> inline void spReconYp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spReconYp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconYp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spReconYp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconYp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconYp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spReconYp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconYp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -526,19 +527,19 @@ namespace QCD {
     }}
   }
 
-  template<class rtype,class vtype> inline void accumReconYp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void accumReconYp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconYp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void accumReconYp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconYp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconYp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void accumReconYp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconYp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -551,19 +552,19 @@ namespace QCD {
   ////////
   // Ym
   ////////
-  template<class rtype,class vtype> inline void spProjYm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spProjYm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjYm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spProjYm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjYm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjYm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spProjYm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjYm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -573,19 +574,19 @@ namespace QCD {
   }
 
 
-  template<class rtype,class vtype> inline void spReconYm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spReconYm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconYm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spReconYm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconYm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconYm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spReconYm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconYm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -594,19 +595,19 @@ namespace QCD {
     }}
   }
 
-  template<class rtype,class vtype> inline void accumReconYm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void accumReconYm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconYm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void accumReconYm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconYm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconYm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void accumReconYm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconYm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -619,19 +620,19 @@ namespace QCD {
   ////////
   // Zp
   ////////
-  template<class rtype,class vtype> inline void spProjZp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spProjZp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjZp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spProjZp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjZp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjZp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spProjZp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjZp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -641,19 +642,19 @@ namespace QCD {
   }
 
 
-  template<class rtype,class vtype> inline void spReconZp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spReconZp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconZp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spReconZp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconZp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconZp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spReconZp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconZp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -662,19 +663,19 @@ namespace QCD {
     }}
   }
 
-  template<class rtype,class vtype> inline void accumReconZp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void accumReconZp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconZp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void accumReconZp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconZp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconZp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void accumReconZp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconZp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -687,19 +688,19 @@ namespace QCD {
   ////////
   // Zm
   ////////
-  template<class rtype,class vtype> inline void spProjZm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spProjZm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjZm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spProjZm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjZm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjZm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spProjZm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjZm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -709,19 +710,19 @@ namespace QCD {
   }
 
 
-  template<class rtype,class vtype> inline void spReconZm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spReconZm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconZm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spReconZm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconZm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconZm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spReconZm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconZm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -730,19 +731,19 @@ namespace QCD {
     }}
   }
 
-  template<class rtype,class vtype> inline void accumReconZm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void accumReconZm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconZm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void accumReconZm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconZm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconZm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void accumReconZm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconZm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -755,19 +756,19 @@ namespace QCD {
   ////////
   // Tp
   ////////
-  template<class rtype,class vtype> inline void spProjTp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spProjTp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjTp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spProjTp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjTp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjTp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spProjTp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjTp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -777,19 +778,19 @@ namespace QCD {
   }
 
 
-  template<class rtype,class vtype> inline void spReconTp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spReconTp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconTp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spReconTp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconTp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconTp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spReconTp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconTp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -798,19 +799,19 @@ namespace QCD {
     }}
   }
 
-  template<class rtype,class vtype> inline void accumReconTp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void accumReconTp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconTp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void accumReconTp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconTp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconTp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void accumReconTp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconTp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -823,19 +824,19 @@ namespace QCD {
   ////////
   // Tm
   ////////
-  template<class rtype,class vtype> inline void spProjTm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spProjTm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjTm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spProjTm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjTm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjTm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spProjTm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjTm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -845,19 +846,19 @@ namespace QCD {
   }
 
 
-  template<class rtype,class vtype> inline void spReconTm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spReconTm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconTm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spReconTm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconTm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconTm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spReconTm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconTm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -866,19 +867,19 @@ namespace QCD {
     }}
   }
 
-  template<class rtype,class vtype> inline void accumReconTm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void accumReconTm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconTm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void accumReconTm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconTm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconTm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void accumReconTm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconTm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -891,19 +892,19 @@ namespace QCD {
   ////////
   // 5p
   ////////
-  template<class rtype,class vtype> inline void spProj5p (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spProj5p (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProj5p(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spProj5p (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProj5p (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProj5p(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spProj5p (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProj5p (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -912,19 +913,19 @@ namespace QCD {
     }}
   }
 
-  template<class rtype,class vtype> inline void spRecon5p (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spRecon5p (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spRecon5p(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spRecon5p (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spRecon5p (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spRecon5p(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spRecon5p (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spRecon5p (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -933,19 +934,19 @@ namespace QCD {
     }}
   }
 
-  template<class rtype,class vtype> inline void accumRecon5p (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void accumRecon5p (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumRecon5p(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void accumRecon5p (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumRecon5p (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumRecon5p(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void accumRecon5p (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumRecon5p (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -955,22 +956,22 @@ namespace QCD {
   }
 
   // four spinor projectors for chiral proj
-  //  template<class vtype> inline void fspProj5p (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
-  template<class vtype> inline void spProj5p (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
+  //  template<class vtype> strong_inline void fspProj5p (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
+  template<class vtype> strong_inline void spProj5p (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProj5p(hspin._internal,fspin._internal);
   }
-  //  template<class vtype,int N> inline void fspProj5p (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
-  template<class vtype,int N> inline void spProj5p (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
+  //  template<class vtype,int N> strong_inline void fspProj5p (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class vtype,int N> strong_inline void spProj5p (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProj5p(hspin._internal[i],fspin._internal[i]);
     }
   }
-  //  template<class vtype,int N> inline void fspProj5p (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
-  template<class vtype,int N> inline void spProj5p (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  //  template<class vtype,int N> strong_inline void fspProj5p (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class vtype,int N> strong_inline void spProj5p (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -983,19 +984,19 @@ namespace QCD {
   ////////
   // 5m
   ////////
-  template<class rtype,class vtype> inline void spProj5m (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spProj5m (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProj5m(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spProj5m (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProj5m (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProj5m(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spProj5m (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProj5m (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -1004,19 +1005,19 @@ namespace QCD {
     }}
   }
 
-  template<class rtype,class vtype> inline void spRecon5m (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spRecon5m (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spRecon5m(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void spRecon5m (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spRecon5m (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spRecon5m(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void spRecon5m (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spRecon5m (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -1025,19 +1026,19 @@ namespace QCD {
     }}
   }
 
-  template<class rtype,class vtype> inline void accumRecon5m (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void accumRecon5m (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumRecon5m(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> inline void accumRecon5m (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumRecon5m (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumRecon5m(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> inline void accumRecon5m (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumRecon5m (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
@@ -1048,22 +1049,22 @@ namespace QCD {
 
 
   // four spinor projectors for chiral proj
-  //  template<class vtype> inline void fspProj5m (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
-  template<class vtype> inline void spProj5m (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
+  //  template<class vtype> strong_inline void fspProj5m (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
+  template<class vtype> strong_inline void spProj5m (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProj5m(hspin._internal,fspin._internal);
   }
-  //  template<class vtype,int N> inline void fspProj5m (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
-  template<class vtype,int N> inline void spProj5m (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
+  //  template<class vtype,int N> strong_inline void fspProj5m (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class vtype,int N> strong_inline void spProj5m (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProj5m(hspin._internal[i],fspin._internal[i]);
     }
   }
-  //  template<class vtype,int N> inline void fspProj5m (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
-  template<class vtype,int N> inline void spProj5m (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  //  template<class vtype,int N> strong_inline void fspProj5m (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class vtype,int N> strong_inline void spProj5m (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
     typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 

From 8d77d758c3b5456b997f6a438b91e0a5dea178b8 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:48:04 +0100
Subject: [PATCH 170/429] Parallel for replace

---
 lib/lattice/Grid_lattice_base.h | 9 +++------
 1 file changed, 3 insertions(+), 6 deletions(-)

diff --git a/lib/lattice/Grid_lattice_base.h b/lib/lattice/Grid_lattice_base.h
index 915879cb..64103c5d 100644
--- a/lib/lattice/Grid_lattice_base.h
+++ b/lib/lattice/Grid_lattice_base.h
@@ -64,8 +64,7 @@ public:
   ////////////////////////////////////////////////////////////////////////////////
   template <typename Op, typename T1>                         inline Lattice<vobj> & operator=(const LatticeUnaryExpression<Op,T1> &expr)
   {
-    //PARALLEL_FOR_LOOP
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
       vobj tmp= eval(ss,expr);
       vstream(_odata[ss] ,tmp);
@@ -74,8 +73,7 @@ public:
   }
   template <typename Op, typename T1,typename T2>             inline Lattice<vobj> & operator=(const LatticeBinaryExpression<Op,T1,T2> &expr)
   {
-    // PARALLEL_FOR_LOOP
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
       vobj tmp= eval(ss,expr);
       vstream(_odata[ss] ,tmp);
@@ -84,8 +82,7 @@ public:
   }
   template <typename Op, typename T1,typename T2,typename T3> inline Lattice<vobj> & operator=(const LatticeTrinaryExpression<Op,T1,T2,T3> &expr)
   {
-    //PARALLEL_FOR_LOOP
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
       vobj tmp= eval(ss,expr);
       vstream(_odata[ss] ,tmp);

From 9a120cf5eca4b35a9823038608425e734caab636 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:49:39 +0100
Subject: [PATCH 171/429] ngo store

---
 lib/simd/Grid_vComplexF.h | 1 -
 1 file changed, 1 deletion(-)

diff --git a/lib/simd/Grid_vComplexF.h b/lib/simd/Grid_vComplexF.h
index cd757916..5e3ff4dc 100644
--- a/lib/simd/Grid_vComplexF.h
+++ b/lib/simd/Grid_vComplexF.h
@@ -203,7 +203,6 @@ namespace Grid {
 #endif
 #ifdef AVX512
 	  _mm512_storenrngo_ps((float *)&out.v,in.v);
-  //	  _mm512_stream_ps((float *)&out.v,in.v);
 	  //Note v has a3 a2 a1 a0
 #endif
 #ifdef QPX

From 264850bc160401513714ad2f88ac502f1831e29e Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:50:00 +0100
Subject: [PATCH 172/429] Move platform dependent out to Grid_simd.h

---
 lib/simd/Grid_vInteger.h | 6 ------
 1 file changed, 6 deletions(-)

diff --git a/lib/simd/Grid_vInteger.h b/lib/simd/Grid_vInteger.h
index 7d558fe7..63e8c720 100644
--- a/lib/simd/Grid_vInteger.h
+++ b/lib/simd/Grid_vInteger.h
@@ -3,12 +3,6 @@
 
 namespace Grid {
 
-// _mm256_set_m128i(hi,lo); // not defined in all versions of immintrin.h
-#ifndef _mm256_set_m128i
-#define _mm256_set_m128i(hi,lo) _mm256_insertf128_si256(_mm256_castsi128_si256(lo),(hi),1)
-#endif
-
-  typedef uint32_t Integer;
 
   class vInteger {
     protected:

From 254dee6ac7d1b0c9f190262519e73e34be638df6 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:50:44 +0100
Subject: [PATCH 173/429] GCC and ICPC complained on more careful typeing

---
 lib/simd/Grid_vRealD.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/simd/Grid_vRealD.h b/lib/simd/Grid_vRealD.h
index aef143f8..dbfacb0c 100644
--- a/lib/simd/Grid_vRealD.h
+++ b/lib/simd/Grid_vRealD.h
@@ -237,7 +237,7 @@ namespace Grid {
 	    __m256d tmp  = _mm256_permute2f128_pd(in.v,in.v,0x01); // tmp 1032; in= 3210
             __m256d hadd = _mm256_hadd_pd(in.v,tmp);              // hadd = 1+0,3+2,3+2,1+0
                     hadd = _mm256_hadd_pd(hadd,hadd);             // hadd = 1+0+3+2...
-	    converter.l = _mm256_extract_epi64(hadd,0);
+		    converter.l = _mm256_extract_epi64((ivec)hadd,0);
             return converter.d;
 #endif
 #ifdef AVX512

From 6965a136a0892bc7f216e97c84d0b3bd003e6c77 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:51:15 +0100
Subject: [PATCH 174/429] Remove debug masking

---
 benchmarks/Grid_wilson.cc | 19 +++++++------------
 1 file changed, 7 insertions(+), 12 deletions(-)

diff --git a/benchmarks/Grid_wilson.cc b/benchmarks/Grid_wilson.cc
index 7718eb19..e49d4515 100644
--- a/benchmarks/Grid_wilson.cc
+++ b/benchmarks/Grid_wilson.cc
@@ -50,25 +50,20 @@ int main (int argc, char ** argv)
     U[mu] = peekIndex<LorentzIndex>(Umu,mu);
   }
   
-  std::vector<int> mask({1,1,1,1,1,1,1,1});
   { // Naive wilson implementation
     ref = zero;
     for(int mu=0;mu<Nd;mu++){
 
       //    ref =  src + Gamma(Gamma::GammaX)* src ; // 1-gamma_x
-      if( mask[mu] ) {
-	tmp = U[mu]*Cshift(src,mu,1);
-	for(int i=0;i<ref._odata.size();i++){
-	  ref._odata[i]+= tmp._odata[i] + Gamma(Gmu[mu])*tmp._odata[i]; ;
-	}
+      tmp = U[mu]*Cshift(src,mu,1);
+      for(int i=0;i<ref._odata.size();i++){
+	ref._odata[i]+= tmp._odata[i] + Gamma(Gmu[mu])*tmp._odata[i]; ;
       }
 
-      if( mask[mu+4] ){
-	tmp =adj(U[mu])*src;
-	tmp =Cshift(tmp,mu,-1);
-	for(int i=0;i<ref._odata.size();i++){
-	  ref._odata[i]+= tmp._odata[i] - Gamma(Gmu[mu])*tmp._odata[i]; ;
-	}
+      tmp =adj(U[mu])*src;
+      tmp =Cshift(tmp,mu,-1);
+      for(int i=0;i<ref._odata.size();i++){
+	ref._odata[i]+= tmp._odata[i] - Gamma(Gmu[mu])*tmp._odata[i]; ;
       }
     }
   }

From 100323ab4d60eb89fb8fa36758ca006edf914f85 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 11:52:11 +0100
Subject: [PATCH 175/429] clang++ 3.4/5/7 compile happy for AVX and SSE

icpc compiles happy on MacOSX both with -xCOMMON-AV512 and native AVX

gcc-5 does not compile happy; can work around by renaming lattice peek/poke/transpose/trace templates
relative to tensor ones, but gcc goes into a recursive template instantiation due to
matching error. I think this is a gcc bug and have filed a report https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66153
---
 configure            | 12 ++++++++++++
 configure-commands   |  3 +++
 configure.ac         |  1 +
 lib/Grid_config.h    | 11 +++++++----
 lib/Grid_config.h.in |  3 +++
 5 files changed, 26 insertions(+), 4 deletions(-)

diff --git a/configure b/configure
index d5071cd1..deec23c3 100755
--- a/configure
+++ b/configure
@@ -4196,6 +4196,18 @@ fi
 
 done
 
+for ac_header in mm_malloc.h
+do :
+  ac_fn_cxx_check_header_mongrel "$LINENO" "mm_malloc.h" "ac_cv_header_mm_malloc_h" "$ac_includes_default"
+if test "x$ac_cv_header_mm_malloc_h" = xyes; then :
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_MM_MALLOC_H 1
+_ACEOF
+
+fi
+
+done
+
 for ac_header in malloc/malloc.h
 do :
   ac_fn_cxx_check_header_mongrel "$LINENO" "malloc/malloc.h" "ac_cv_header_malloc_malloc_h" "$ac_includes_default"
diff --git a/configure-commands b/configure-commands
index 8ba2dac1..cf8057a9 100644
--- a/configure-commands
+++ b/configure-commands
@@ -1,5 +1,8 @@
+CXX=icpc ./configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp" --enable-comms=mpi
+
 CXX=clang-omp++ ./configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp"  --enable-comms=mpi
 CXX=clang++ ./configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx"  --enable-comms=mpi
 CXX=clang++ ./configure --enable-simd=AVX CXXFLAGS="-mavx -g -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx"  --enable-comms=mpi
 CXX=clang++ ./configure --enable-simd=AVX CXXFLAGS="-mavx -g  -std=c++11" LDFLAGS= LIBS=-lmpi  --enable-comms=fake
 CXX=clang++ ./configure --enable-simd=AVX CXXFLAGS="-mavx -g  -std=c++11" LDFLAGS= LIBS=-lmpi  --enable-comms=none
+CXX=/usr/local/clang-3.7/bin/clang++ ./configure --enable-simd=AVX512 CXXFLAGS="-mavx512f -O3  -std=c++11" LDFLAGS= LIBS= --enable-comms=none
diff --git a/configure.ac b/configure.ac
index 86f4b70c..97206a01 100644
--- a/configure.ac
+++ b/configure.ac
@@ -16,6 +16,7 @@ AC_PROG_RANLIB
 
 # Checks for header files.
 AC_CHECK_HEADERS(stdint.h)
+AC_CHECK_HEADERS(mm_malloc.h)
 AC_CHECK_HEADERS(malloc/malloc.h)
 AC_CHECK_HEADERS(malloc.h)
 AC_CHECK_HEADERS(endian.h)
diff --git a/lib/Grid_config.h b/lib/Grid_config.h
index c743bba0..f71e202f 100644
--- a/lib/Grid_config.h
+++ b/lib/Grid_config.h
@@ -2,19 +2,19 @@
 /* lib/Grid_config.h.in.  Generated from configure.ac by autoheader.  */
 
 /* AVX */
-#define AVX1 1
+/* #undef AVX1 */
 
 /* AVX2 */
 /* #undef AVX2 */
 
 /* AVX512 */
-/* #undef AVX512 */
+#define AVX512 1
 
 /* GRID_COMMS_MPI */
-#define GRID_COMMS_MPI 1
+/* #undef GRID_COMMS_MPI */
 
 /* GRID_COMMS_NONE */
-/* #undef GRID_COMMS_NONE */
+#define GRID_COMMS_NONE 1
 
 /* Define to 1 if you have the declaration of `be64toh', and to 0 if you
    don't. */
@@ -42,6 +42,9 @@
 /* Define to 1 if you have the <memory.h> header file. */
 #define HAVE_MEMORY_H 1
 
+/* Define to 1 if you have the <mm_malloc.h> header file. */
+#define HAVE_MM_MALLOC_H 1
+
 /* Define to 1 if you have the <stdint.h> header file. */
 #define HAVE_STDINT_H 1
 
diff --git a/lib/Grid_config.h.in b/lib/Grid_config.h.in
index 437a6095..47643530 100644
--- a/lib/Grid_config.h.in
+++ b/lib/Grid_config.h.in
@@ -41,6 +41,9 @@
 /* Define to 1 if you have the <memory.h> header file. */
 #undef HAVE_MEMORY_H
 
+/* Define to 1 if you have the <mm_malloc.h> header file. */
+#undef HAVE_MM_MALLOC_H
+
 /* Define to 1 if you have the <stdint.h> header file. */
 #undef HAVE_STDINT_H
 

From f92fda0cfd9a3ca4fc5537d3d2e0cc0e8a7a11ef Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 12:21:10 +0100
Subject: [PATCH 176/429] Convenience multi-compiler build with out of source
 compile

---
 build-all          | 11 +++++++++++
 configure-all      | 11 +++++++++++
 configure-commands | 34 ++++++++++++++++++++++++++--------
 configure-mic      | 10 ++++++++++
 4 files changed, 58 insertions(+), 8 deletions(-)
 create mode 100755 build-all
 create mode 100755 configure-all
 mode change 100644 => 100755 configure-commands
 create mode 100755 configure-mic

diff --git a/build-all b/build-all
new file mode 100755
index 00000000..cff673ab
--- /dev/null
+++ b/build-all
@@ -0,0 +1,11 @@
+#!/bin/bash 
+DIRS="build-clang-avx build-clang-avx-openmp build-clang-avx-openmp-mpi build-clang-avx-mpi build-icpc-avx512 build-icpc-avx "
+
+for D in $DIRS
+do
+  echo $D
+  mkdir -p $D
+  cd $D
+  make clean all -j 8
+  cd ..
+done
diff --git a/configure-all b/configure-all
new file mode 100755
index 00000000..aa67aaf0
--- /dev/null
+++ b/configure-all
@@ -0,0 +1,11 @@
+#!/bin/bash
+DIRS="build-icpc-avx build-icpc-avx512 build-clang-avx build-clang-avx-openmp build-clang-avx-openmp-mpi build-clang-avx-mpi"
+
+for D in $DIRS
+do
+  echo $D
+  mkdir -p $D
+  cd $D
+  ../configure-commands $D
+  cd ..
+done
diff --git a/configure-commands b/configure-commands
old mode 100644
new mode 100755
index cf8057a9..66d7b19b
--- a/configure-commands
+++ b/configure-commands
@@ -1,8 +1,26 @@
-CXX=icpc ./configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp" --enable-comms=mpi
-
-CXX=clang-omp++ ./configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp"  --enable-comms=mpi
-CXX=clang++ ./configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx"  --enable-comms=mpi
-CXX=clang++ ./configure --enable-simd=AVX CXXFLAGS="-mavx -g -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx"  --enable-comms=mpi
-CXX=clang++ ./configure --enable-simd=AVX CXXFLAGS="-mavx -g  -std=c++11" LDFLAGS= LIBS=-lmpi  --enable-comms=fake
-CXX=clang++ ./configure --enable-simd=AVX CXXFLAGS="-mavx -g  -std=c++11" LDFLAGS= LIBS=-lmpi  --enable-comms=none
-CXX=/usr/local/clang-3.7/bin/clang++ ./configure --enable-simd=AVX512 CXXFLAGS="-mavx512f -O3  -std=c++11" LDFLAGS= LIBS= --enable-comms=none
+#!/bin/bash
+WD=$1
+echo $WD
+case $WD in
+build-icpc-avx)
+  CXX=icpc ../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -I -std=c++11" --enable-comms=none
+  ;;
+build-icpc-avx512)
+  CXX=icpc ../configure --enable-simd=AVX512 CXXFLAGS="-xCOMMON-AVX512 -O3 -I -std=c++11" --host=none  --enable-comms=none
+  ;;
+build-icpc-mic)
+  CXX=icpc ../configure --host=none --enable-simd=AVX512 CXXFLAGS="-mmic -O3 -I -std=c++11" LDFLAGS=-mmic --enable-comms=none
+  ;;
+build-clang-avx)
+CXX=clang++ ../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11"  --enable-comms=none
+  ;;
+build-clang-avx-openmp)
+CXX=clang-omp++ ../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -std=c++11" LDFLAGS="-fopenmp"  --enable-comms=none
+  ;;
+build-clang-avx-openmp-mpi)
+CXX=clang-omp++ ../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp"  --enable-comms=mpi
+;;
+build-clang-avx-mpi)
+CXX=clang++ ../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx "  --enable-comms=mpi
+;;
+esac
diff --git a/configure-mic b/configure-mic
new file mode 100755
index 00000000..668845fe
--- /dev/null
+++ b/configure-mic
@@ -0,0 +1,10 @@
+#!/bin/bash
+DIRS="build-icpc-mic"
+
+for D in $DIRS
+do
+  mkdir -p $D
+ cd $D
+  ../configure-commands
+  cd ..
+done

From a98f3e0f5edf7fe671560de5853716435480c5e9 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 12:21:40 +0100
Subject: [PATCH 177/429] Out of source compile now working

---
 benchmarks/Makefile.am |   2 +-
 lib/Grid_config.h      | 118 -----------------------------------------
 tests/Makefile.am      |   2 +-
 3 files changed, 2 insertions(+), 120 deletions(-)
 delete mode 100644 lib/Grid_config.h

diff --git a/benchmarks/Makefile.am b/benchmarks/Makefile.am
index d6fad22d..76c72279 100644
--- a/benchmarks/Makefile.am
+++ b/benchmarks/Makefile.am
@@ -1,6 +1,6 @@
 # additional include paths necessary to compile the C++ library
 AM_CXXFLAGS = -I$(top_srcdir)/lib
-AM_LDFLAGS = -L$(top_srcdir)/lib
+AM_LDFLAGS = -L$(top_builddir)/lib
 
 #
 # Test code
diff --git a/lib/Grid_config.h b/lib/Grid_config.h
deleted file mode 100644
index f71e202f..00000000
--- a/lib/Grid_config.h
+++ /dev/null
@@ -1,118 +0,0 @@
-/* lib/Grid_config.h.  Generated from Grid_config.h.in by configure.  */
-/* lib/Grid_config.h.in.  Generated from configure.ac by autoheader.  */
-
-/* AVX */
-/* #undef AVX1 */
-
-/* AVX2 */
-/* #undef AVX2 */
-
-/* AVX512 */
-#define AVX512 1
-
-/* GRID_COMMS_MPI */
-/* #undef GRID_COMMS_MPI */
-
-/* GRID_COMMS_NONE */
-#define GRID_COMMS_NONE 1
-
-/* Define to 1 if you have the declaration of `be64toh', and to 0 if you
-   don't. */
-#define HAVE_DECL_BE64TOH 0
-
-/* Define to 1 if you have the declaration of `ntohll', and to 0 if you don't.
-   */
-#define HAVE_DECL_NTOHLL 1
-
-/* Define to 1 if you have the <endian.h> header file. */
-/* #undef HAVE_ENDIAN_H */
-
-/* Define to 1 if you have the `gettimeofday' function. */
-#define HAVE_GETTIMEOFDAY 1
-
-/* Define to 1 if you have the <inttypes.h> header file. */
-#define HAVE_INTTYPES_H 1
-
-/* Define to 1 if you have the <malloc.h> header file. */
-/* #undef HAVE_MALLOC_H */
-
-/* Define to 1 if you have the <malloc/malloc.h> header file. */
-#define HAVE_MALLOC_MALLOC_H 1
-
-/* Define to 1 if you have the <memory.h> header file. */
-#define HAVE_MEMORY_H 1
-
-/* Define to 1 if you have the <mm_malloc.h> header file. */
-#define HAVE_MM_MALLOC_H 1
-
-/* Define to 1 if you have the <stdint.h> header file. */
-#define HAVE_STDINT_H 1
-
-/* Define to 1 if you have the <stdlib.h> header file. */
-#define HAVE_STDLIB_H 1
-
-/* Define to 1 if you have the <strings.h> header file. */
-#define HAVE_STRINGS_H 1
-
-/* Define to 1 if you have the <string.h> header file. */
-#define HAVE_STRING_H 1
-
-/* Define to 1 if you have the <sys/stat.h> header file. */
-#define HAVE_SYS_STAT_H 1
-
-/* Define to 1 if you have the <sys/types.h> header file. */
-#define HAVE_SYS_TYPES_H 1
-
-/* Define to 1 if you have the <unistd.h> header file. */
-#define HAVE_UNISTD_H 1
-
-/* Name of package */
-#define PACKAGE "grid"
-
-/* Define to the address where bug reports for this package should be sent. */
-#define PACKAGE_BUGREPORT "paboyle@ph.ed.ac.uk"
-
-/* Define to the full name of this package. */
-#define PACKAGE_NAME "Grid"
-
-/* Define to the full name and version of this package. */
-#define PACKAGE_STRING "Grid 1.0"
-
-/* Define to the one symbol short name of this package. */
-#define PACKAGE_TARNAME "grid"
-
-/* Define to the home page for this package. */
-#define PACKAGE_URL ""
-
-/* Define to the version of this package. */
-#define PACKAGE_VERSION "1.0"
-
-/* SSE4 */
-/* #undef SSE4 */
-
-/* Define to 1 if you have the ANSI C header files. */
-#define STDC_HEADERS 1
-
-/* Version number of package */
-#define VERSION "1.0"
-
-/* Define for Solaris 2.5.1 so the uint32_t typedef from <sys/synch.h>,
-   <pthread.h>, or <semaphore.h> is not used. If the typedef were allowed, the
-   #define below would cause a syntax error. */
-/* #undef _UINT32_T */
-
-/* Define for Solaris 2.5.1 so the uint64_t typedef from <sys/synch.h>,
-   <pthread.h>, or <semaphore.h> is not used. If the typedef were allowed, the
-   #define below would cause a syntax error. */
-/* #undef _UINT64_T */
-
-/* Define to `unsigned int' if <sys/types.h> does not define. */
-/* #undef size_t */
-
-/* Define to the type of an unsigned integer type of width exactly 32 bits if
-   such a type exists and the standard includes do not define it. */
-/* #undef uint32_t */
-
-/* Define to the type of an unsigned integer type of width exactly 64 bits if
-   such a type exists and the standard includes do not define it. */
-/* #undef uint64_t */
diff --git a/tests/Makefile.am b/tests/Makefile.am
index 6af1ce3b..655bb9c5 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -1,6 +1,6 @@
 # additional include paths necessary to compile the C++ library
 AM_CXXFLAGS = -I$(top_srcdir)/lib
-AM_LDFLAGS = -L$(top_srcdir)/lib
+AM_LDFLAGS = -L$(top_builddir)/lib
 
 #
 # Test code

From 675fd1a065dcb7de1ac8dc3912347c1ebcfe627b Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 12:33:18 +0100
Subject: [PATCH 178/429] Compile options tweak

---
 configure-commands        | 6 +++---
 lib/simd/Grid_vComplexF.h | 1 +
 2 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/configure-commands b/configure-commands
index 66d7b19b..f4c21f30 100755
--- a/configure-commands
+++ b/configure-commands
@@ -3,13 +3,13 @@ WD=$1
 echo $WD
 case $WD in
 build-icpc-avx)
-  CXX=icpc ../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -I -std=c++11" --enable-comms=none
+  CXX=icpc ../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11" --enable-comms=none
   ;;
 build-icpc-avx512)
-  CXX=icpc ../configure --enable-simd=AVX512 CXXFLAGS="-xCOMMON-AVX512 -O3 -I -std=c++11" --host=none  --enable-comms=none
+  CXX=icpc ../configure --enable-simd=AVX512 CXXFLAGS="-xCOMMON-AVX512 -O3  -std=c++11" --host=none  --enable-comms=none
   ;;
 build-icpc-mic)
-  CXX=icpc ../configure --host=none --enable-simd=AVX512 CXXFLAGS="-mmic -O3 -I -std=c++11" LDFLAGS=-mmic --enable-comms=none
+  CXX=icpc ../configure --host=none --enable-simd=AVX512 CXXFLAGS="-mmic -O3  -std=c++11" LDFLAGS=-mmic --enable-comms=none
   ;;
 build-clang-avx)
 CXX=clang++ ../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11"  --enable-comms=none
diff --git a/lib/simd/Grid_vComplexF.h b/lib/simd/Grid_vComplexF.h
index 5e3ff4dc..ac15d19f 100644
--- a/lib/simd/Grid_vComplexF.h
+++ b/lib/simd/Grid_vComplexF.h
@@ -245,6 +245,7 @@ namespace Grid {
 	     float f[8];  
 	   } u256;
 	   //SWAP lanes
+	   // FIXME .. icc complains with lib/lattice/Grid_lattice_reduction.h (49): (col. 20) warning #13211: Immediate parameter to intrinsic call too large
 	   __m256 t0 = _mm256_permute2f128_ps(in.v, in.v, 1);
 	   __m256 t1 = _mm256_permute_ps(in.v  , 0b11011000);//real (0,2,1,3)
 	   __m256 t2 = _mm256_permute_ps(t0 , 0b10001101);//imag (1,3,0,2)

From 516aac65183aa6a2860bb8017b40808c59e91079 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 12:39:53 +0100
Subject: [PATCH 179/429] Log the bug report code into the git repo.

---
 gcc-bug-report/README    | 23 ++++++++++++++++
 gcc-bug-report/broken.cc | 59 ++++++++++++++++++++++++++++++++++++++++
 2 files changed, 82 insertions(+)
 create mode 100644 gcc-bug-report/README
 create mode 100644 gcc-bug-report/broken.cc

diff --git a/gcc-bug-report/README b/gcc-bug-report/README
new file mode 100644
index 00000000..294d0a6c
--- /dev/null
+++ b/gcc-bug-report/README
@@ -0,0 +1,23 @@
+https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66153
+
+Grid code breaks on GCC4.8, 4.9, 5.0 due to the 
+peekIndex<n> operating on lattice.
+
+It erroneously recurses back into the Lattice<obj> variant, even though
+the lattice container is dropped.
+
+Work around is possible; if the Lattice routine is given a disambiguating 
+name prefix, such as 
+
+latPeekIndex<n> 
+
+GCC5 works. 
+
+However this is ugly and for now I have submitted a bug report to see the reaction and
+speed of fixing. 
+
+The simple testcase in this directory is the submitted bug report that encapsulates the
+problem. The test case works with icpc and with clang++, but fails consistently on g++
+current variants.
+
+Peter
\ No newline at end of file
diff --git a/gcc-bug-report/broken.cc b/gcc-bug-report/broken.cc
new file mode 100644
index 00000000..bd403e85
--- /dev/null
+++ b/gcc-bug-report/broken.cc
@@ -0,0 +1,59 @@
+#include <vector>
+#include <complex>
+#include <type_traits>
+#include <iostream>
+
+typedef std::complex<double> ComplexD;
+
+template <class T> class TypeMapper {
+public:
+  enum { NestLevel = T::NestLevel };
+};
+
+template<> class TypeMapper<ComplexD> {
+public:
+  enum { NestLevel = 0 };
+};
+
+template<class obj> class Container {
+ public:
+  std::vector<obj> data;
+  Container(int size) : data (size){};
+};
+
+template<class obj> class Recursive {
+public:
+  enum { NestLevel = TypeMapper<obj>::NestLevel + 1};
+  obj internal;
+};
+
+template<int N,class obj,typename std::enable_if<N==obj::NestLevel >::type * = nullptr > auto function(const obj &arg)-> obj
+{
+  std::cout<<"Leaf "<<obj::NestLevel<<std::endl;
+  return arg;
+}
+template<int N,class obj,typename std::enable_if<N!=obj::NestLevel >::type * = nullptr > auto function(const obj &arg)-> obj
+{
+  std::cout<<"Node "<<obj::NestLevel<<std::endl;
+  obj ret;
+  ret.internal=function<N>(arg.internal);
+  return ret;
+}
+
+template<int N,class obj> auto function(const Container<obj> & arg)-> Container<decltype(function<N>(arg.data[0]))>
+{
+  Container<decltype(function<N>(arg.data[0]))> ret(arg.data.size());
+  for(int ss=0;ss<arg.data.size();ss++){
+    ret.data[ss] = function<N>(arg.data[ss]);
+  }
+  return ret;
+}
+
+
+int main(int argc,char **argv)
+{
+  Container<Recursive<Recursive<ComplexD> > > array(10);
+  Container<Recursive<Recursive<ComplexD> > > ret(10);
+  
+  ret = function<1>(array);
+}

From 7a63bdbd72d15f89137e2ed4d191c7bcc33a624f Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 14:41:19 +0100
Subject: [PATCH 180/429] Added su3 matrix benchmark.

---
 benchmarks/Grid_su3.cc | 139 +++++++++++++++++++++++++++++++++++++++++
 benchmarks/Makefile.am |   5 +-
 2 files changed, 143 insertions(+), 1 deletion(-)
 create mode 100644 benchmarks/Grid_su3.cc

diff --git a/benchmarks/Grid_su3.cc b/benchmarks/Grid_su3.cc
new file mode 100644
index 00000000..7f7bff90
--- /dev/null
+++ b/benchmarks/Grid_su3.cc
@@ -0,0 +1,139 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  int Nloop=1000;
+
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "= Benchmarking SU3xSU3  x= x*y"<<std::endl;
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
+  std::cout << "----------------------------------------------------------"<<std::endl;
+
+  for(int lat=2;lat<=24;lat+=2){
+
+      std::vector<int> latt_size  ({lat,lat,lat,lat});
+
+      GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+      //      GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
+
+      LatticeColourMatrix z(&Grid);// random(pRNG,z);
+      LatticeColourMatrix x(&Grid);// random(pRNG,x);
+      LatticeColourMatrix y(&Grid);// random(pRNG,y);
+
+      double start=usecond();
+      for(int i=0;i<Nloop;i++){
+	x=x*y;
+      }
+      double stop=usecond();
+      double time = (stop-start)/Nloop*1000.0;
+      
+      double bytes=3.0*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
+      double footprint=2.0*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
+      double flops=Nc*Nc*(6.0+8.0+8.0)*lat*lat*lat*lat;
+      std::cout<<std::setprecision(2) << lat<<"\t\t"<<footprint<<"\t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
+
+    }
+
+
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "= Benchmarking SU3xSU3  z= x*y"<<std::endl;
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
+  std::cout << "----------------------------------------------------------"<<std::endl;
+
+  for(int lat=2;lat<=24;lat+=2){
+
+      std::vector<int> latt_size  ({lat,lat,lat,lat});
+
+      GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+      //      GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
+
+      LatticeColourMatrix z(&Grid); //random(pRNG,z);
+      LatticeColourMatrix x(&Grid); //random(pRNG,x);
+      LatticeColourMatrix y(&Grid); //random(pRNG,y);
+
+      double start=usecond();
+      for(int i=0;i<Nloop;i++){
+	z=x*y;
+      }
+      double stop=usecond();
+      double time = (stop-start)/Nloop*1000.0;
+      
+      double bytes=3*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
+      double flops=Nc*Nc*(6+8+8)*lat*lat*lat*lat;
+      std::cout<<std::setprecision(2) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
+
+    }
+
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "= Benchmarking SU3xSU3  mult(z,x,y)"<<std::endl;
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
+  std::cout << "----------------------------------------------------------"<<std::endl;
+
+  for(int lat=2;lat<=24;lat+=2){
+
+      std::vector<int> latt_size  ({lat,lat,lat,lat});
+
+      GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+      //      GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
+
+      LatticeColourMatrix z(&Grid); //random(pRNG,z);
+      LatticeColourMatrix x(&Grid); //random(pRNG,x);
+      LatticeColourMatrix y(&Grid); //random(pRNG,y);
+
+      double start=usecond();
+      for(int i=0;i<Nloop;i++){
+	mult(z,x,y);
+      }
+      double stop=usecond();
+      double time = (stop-start)/Nloop*1000.0;
+      
+      double bytes=3*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
+      double flops=Nc*Nc*(6+8+8)*lat*lat*lat*lat;
+      std::cout<<std::setprecision(2) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
+
+    }
+
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "= Benchmarking SU3xSU3  mac(z,x,y)"<<std::endl;
+  std::cout << "===================================================================================================="<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
+  std::cout << "----------------------------------------------------------"<<std::endl;
+
+  for(int lat=2;lat<=24;lat+=2){
+
+      std::vector<int> latt_size  ({lat,lat,lat,lat});
+
+      GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+      //      GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
+
+      LatticeColourMatrix z(&Grid); //random(pRNG,z);
+      LatticeColourMatrix x(&Grid); //random(pRNG,x);
+      LatticeColourMatrix y(&Grid); //random(pRNG,y);
+
+      double start=usecond();
+      for(int i=0;i<Nloop;i++){
+	mac(z,x,y);
+      }
+      double stop=usecond();
+      double time = (stop-start)/Nloop*1000.0;
+      
+      double bytes=3*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
+      double flops=Nc*Nc*(8+8+8)*lat*lat*lat*lat;
+      std::cout<<std::setprecision(2) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
+
+    }
+
+  Grid_finalize();
+}
diff --git a/benchmarks/Makefile.am b/benchmarks/Makefile.am
index 76c72279..2fe40feb 100644
--- a/benchmarks/Makefile.am
+++ b/benchmarks/Makefile.am
@@ -5,7 +5,7 @@ AM_LDFLAGS = -L$(top_builddir)/lib
 #
 # Test code
 #
-bin_PROGRAMS = Grid_wilson Grid_comms Grid_memory_bandwidth
+bin_PROGRAMS = Grid_wilson Grid_comms Grid_memory_bandwidth Grid_su3
 
 
 Grid_wilson_SOURCES = Grid_wilson.cc
@@ -14,6 +14,9 @@ Grid_wilson_LDADD = -lGrid
 Grid_comms_SOURCES = Grid_comms.cc
 Grid_comms_LDADD = -lGrid
 
+Grid_su3_SOURCES = Grid_su3.cc
+Grid_su3_LDADD = -lGrid
+
 Grid_memory_bandwidth_SOURCES = Grid_memory_bandwidth.cc
 Grid_memory_bandwidth_LDADD = -lGrid
 

From c2ca39635385fa021913cb63054baf7bd8988009 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 15 May 2015 14:41:59 +0100
Subject: [PATCH 181/429] Extra compile targs

---
 configure-commands | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/configure-commands b/configure-commands
index f4c21f30..19d50a4d 100755
--- a/configure-commands
+++ b/configure-commands
@@ -2,6 +2,9 @@
 WD=$1
 echo $WD
 case $WD in
+build-g++-sse4)
+  CXX=g++-5 ../configure --enable-simd=SSE4 CXXFLAGS="-msse4 -O3 -std=c++11" --enable-comms=none
+  ;;
 build-icpc-avx)
   CXX=icpc ../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11" --enable-comms=none
   ;;

From 49f56a25d1ddd7b6191ecc21f039f699ea4c31b3 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 16 May 2015 04:33:10 +0100
Subject: [PATCH 182/429] strong inline

---
 lib/lattice/Grid_lattice_arith.h | 29 +++++++++++++++--------------
 1 file changed, 15 insertions(+), 14 deletions(-)

diff --git a/lib/lattice/Grid_lattice_arith.h b/lib/lattice/Grid_lattice_arith.h
index c9599582..c0bbb2b6 100644
--- a/lib/lattice/Grid_lattice_arith.h
+++ b/lib/lattice/Grid_lattice_arith.h
@@ -7,7 +7,7 @@ namespace Grid {
   //////////////////////////////////////////////////////////////////////////////////////////////////////
   //  avoid copy back routines for mult, mac, sub, add
   //////////////////////////////////////////////////////////////////////////////////////////////////////
-  template<class obj1,class obj2,class obj3> 
+  template<class obj1,class obj2,class obj3> strong_inline
     void mult(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
     conformable(lhs,rhs);
 PARALLEL_FOR_LOOP
@@ -15,10 +15,11 @@ PARALLEL_FOR_LOOP
       obj1 tmp;
       mult(&tmp,&lhs._odata[ss],&rhs._odata[ss]);
       vstream(ret._odata[ss],tmp);
+      //      mult(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
     }
   }
   
-  template<class obj1,class obj2,class obj3>
+  template<class obj1,class obj2,class obj3> strong_inline
     void mac(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
     conformable(lhs,rhs);
 PARALLEL_FOR_LOOP
@@ -29,7 +30,7 @@ PARALLEL_FOR_LOOP
     }
   }
   
-  template<class obj1,class obj2,class obj3>
+  template<class obj1,class obj2,class obj3> strong_inline
     void sub(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
     conformable(lhs,rhs);
 PARALLEL_FOR_LOOP
@@ -39,7 +40,7 @@ PARALLEL_FOR_LOOP
       vstream(ret._odata[ss],tmp);
     }
   }
-  template<class obj1,class obj2,class obj3>
+  template<class obj1,class obj2,class obj3> strong_inline
     void add(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
     conformable(lhs,rhs);
 PARALLEL_FOR_LOOP
@@ -53,7 +54,7 @@ PARALLEL_FOR_LOOP
   //////////////////////////////////////////////////////////////////////////////////////////////////////
   //  avoid copy back routines for mult, mac, sub, add
   //////////////////////////////////////////////////////////////////////////////////////////////////////
-  template<class obj1,class obj2,class obj3>
+  template<class obj1,class obj2,class obj3> strong_inline
     void mult(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
     conformable(lhs,ret);
 PARALLEL_FOR_LOOP
@@ -64,7 +65,7 @@ PARALLEL_FOR_LOOP
     }
   }
   
-  template<class obj1,class obj2,class obj3>
+  template<class obj1,class obj2,class obj3> strong_inline
     void mac(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
     conformable(lhs,ret);
 PARALLEL_FOR_LOOP
@@ -75,7 +76,7 @@ PARALLEL_FOR_LOOP
     }
   }
   
-  template<class obj1,class obj2,class obj3>
+  template<class obj1,class obj2,class obj3> strong_inline
     void sub(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
     conformable(lhs,ret);
 PARALLEL_FOR_LOOP
@@ -85,7 +86,7 @@ PARALLEL_FOR_LOOP
       vstream(ret._odata[ss],tmp);
     }
   }
-  template<class obj1,class obj2,class obj3>
+  template<class obj1,class obj2,class obj3> strong_inline
     void add(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
     conformable(lhs,ret);
 PARALLEL_FOR_LOOP
@@ -99,7 +100,7 @@ PARALLEL_FOR_LOOP
   //////////////////////////////////////////////////////////////////////////////////////////////////////
   //  avoid copy back routines for mult, mac, sub, add
   //////////////////////////////////////////////////////////////////////////////////////////////////////
-    template<class obj1,class obj2,class obj3>
+    template<class obj1,class obj2,class obj3> strong_inline
     void mult(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
     conformable(ret,rhs);
 PARALLEL_FOR_LOOP
@@ -110,7 +111,7 @@ PARALLEL_FOR_LOOP
     }
   }
   
-  template<class obj1,class obj2,class obj3>
+  template<class obj1,class obj2,class obj3> strong_inline
     void mac(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
     conformable(ret,rhs);
 PARALLEL_FOR_LOOP
@@ -121,7 +122,7 @@ PARALLEL_FOR_LOOP
     }
   }
   
-  template<class obj1,class obj2,class obj3>
+  template<class obj1,class obj2,class obj3> strong_inline
     void sub(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
     conformable(ret,rhs);
 PARALLEL_FOR_LOOP
@@ -131,7 +132,7 @@ PARALLEL_FOR_LOOP
       vstream(ret._odata[ss],tmp);
     }
   }
-  template<class obj1,class obj2,class obj3>
+  template<class obj1,class obj2,class obj3> strong_inline
     void add(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
     conformable(ret,rhs);
 PARALLEL_FOR_LOOP
@@ -142,8 +143,8 @@ PARALLEL_FOR_LOOP
     }
   }
   
-  template<class sobj,class vobj>
-  inline void axpy(Lattice<vobj> &ret,sobj a,const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
+  template<class sobj,class vobj> strong_inline
+  void axpy(Lattice<vobj> &ret,sobj a,const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
     conformable(lhs,rhs);
 #pragma omp parallel for
     for(int ss=0;ss<lhs._grid->oSites();ss++){

From 56667e9d3204cbe436e7b27c4cec2f8a93bacd69 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 16 May 2015 04:33:40 +0100
Subject: [PATCH 183/429] more digits

---
 benchmarks/Grid_su3.cc | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/benchmarks/Grid_su3.cc b/benchmarks/Grid_su3.cc
index 7f7bff90..ce3ccaa9 100644
--- a/benchmarks/Grid_su3.cc
+++ b/benchmarks/Grid_su3.cc
@@ -40,7 +40,7 @@ int main (int argc, char ** argv)
       double bytes=3.0*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
       double footprint=2.0*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
       double flops=Nc*Nc*(6.0+8.0+8.0)*lat*lat*lat*lat;
-      std::cout<<std::setprecision(2) << lat<<"\t\t"<<footprint<<"\t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
+      std::cout<<std::setprecision(3) << lat<<"\t\t"<<footprint<<"\t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
 
     }
 
@@ -71,7 +71,7 @@ int main (int argc, char ** argv)
       
       double bytes=3*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
       double flops=Nc*Nc*(6+8+8)*lat*lat*lat*lat;
-      std::cout<<std::setprecision(2) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
+      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
 
     }
 
@@ -101,7 +101,7 @@ int main (int argc, char ** argv)
       
       double bytes=3*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
       double flops=Nc*Nc*(6+8+8)*lat*lat*lat*lat;
-      std::cout<<std::setprecision(2) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
+      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
 
     }
 
@@ -131,7 +131,7 @@ int main (int argc, char ** argv)
       
       double bytes=3*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
       double flops=Nc*Nc*(8+8+8)*lat*lat*lat*lat;
-      std::cout<<std::setprecision(2) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
+      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
 
     }
 

From 9f0e990b4054ee825b4eb4ab171748e5ad2f362e Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 16 May 2015 04:36:22 +0100
Subject: [PATCH 184/429] Optimisation and syntax pretty

---
 lib/lattice/Grid_lattice_base.h  | 20 ++++++++++----------
 lib/lattice/Grid_lattice_trace.h |  2 +-
 lib/math/Grid_math_arith_mac.h   |  5 +++--
 lib/math/Grid_math_arith_mul.h   | 17 ++++++++++++-----
 4 files changed, 26 insertions(+), 18 deletions(-)

diff --git a/lib/lattice/Grid_lattice_base.h b/lib/lattice/Grid_lattice_base.h
index 64103c5d..ae606ae7 100644
--- a/lib/lattice/Grid_lattice_base.h
+++ b/lib/lattice/Grid_lattice_base.h
@@ -62,7 +62,7 @@ public:
   ////////////////////////////////////////////////////////////////////////////////
   // Expression Template closure support
   ////////////////////////////////////////////////////////////////////////////////
-  template <typename Op, typename T1>                         inline Lattice<vobj> & operator=(const LatticeUnaryExpression<Op,T1> &expr)
+  template <typename Op, typename T1>                         strong_inline Lattice<vobj> & operator=(const LatticeUnaryExpression<Op,T1> &expr)
   {
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
@@ -71,7 +71,7 @@ PARALLEL_FOR_LOOP
     }
     return *this;
   }
-  template <typename Op, typename T1,typename T2>             inline Lattice<vobj> & operator=(const LatticeBinaryExpression<Op,T1,T2> &expr)
+  template <typename Op, typename T1,typename T2>             strong_inline Lattice<vobj> & operator=(const LatticeBinaryExpression<Op,T1,T2> &expr)
   {
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
@@ -80,7 +80,7 @@ PARALLEL_FOR_LOOP
     }
     return *this;
   }
-  template <typename Op, typename T1,typename T2,typename T3> inline Lattice<vobj> & operator=(const LatticeTrinaryExpression<Op,T1,T2,T3> &expr)
+  template <typename Op, typename T1,typename T2,typename T3> strong_inline Lattice<vobj> & operator=(const LatticeTrinaryExpression<Op,T1,T2,T3> &expr)
   {
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
@@ -132,14 +132,14 @@ PARALLEL_FOR_LOOP
         checkerboard=0;
     }
 
-    template<class sobj> inline Lattice<vobj> & operator = (const sobj & r){
+    template<class sobj> strong_inline Lattice<vobj> & operator = (const sobj & r){
 PARALLEL_FOR_LOOP
         for(int ss=0;ss<_grid->oSites();ss++){
             this->_odata[ss]=r;
         }
         return *this;
     }
-    template<class robj> inline Lattice<vobj> & operator = (const Lattice<robj> & r){
+    template<class robj> strong_inline Lattice<vobj> & operator = (const Lattice<robj> & r){
       conformable(*this,r);
       std::cout<<"Lattice operator ="<<std::endl;
 PARALLEL_FOR_LOOP
@@ -150,21 +150,21 @@ PARALLEL_FOR_LOOP
     }
 
     // *=,+=,-= operators inherit behvour from correspond */+/- operation
-    template<class T> inline Lattice<vobj> &operator *=(const T &r) {
+    template<class T> strong_inline Lattice<vobj> &operator *=(const T &r) {
         *this = (*this)*r;
         return *this;
     }
 
-    template<class T> inline Lattice<vobj> &operator -=(const T &r) {
+    template<class T> strong_inline Lattice<vobj> &operator -=(const T &r) {
         *this = (*this)-r;
         return *this;
     }
-    template<class T> inline Lattice<vobj> &operator +=(const T &r) {
+    template<class T> strong_inline Lattice<vobj> &operator +=(const T &r) {
         *this = (*this)+r;
         return *this;
     }
     
-    inline friend Lattice<vobj> operator / (const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
+    strong_inline friend Lattice<vobj> operator / (const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
         conformable(lhs,rhs);
         Lattice<vobj> ret(lhs._grid);
 PARALLEL_FOR_LOOP
@@ -176,7 +176,7 @@ PARALLEL_FOR_LOOP
 
  }; // class Lattice
 
-  template<class vobj> inline std::ostream& operator<< (std::ostream& stream, const Lattice<vobj> &o){
+  template<class vobj> std::ostream& operator<< (std::ostream& stream, const Lattice<vobj> &o){
     std::vector<int> gcoor;
     typedef typename vobj::scalar_object sobj;
     sobj ss;
diff --git a/lib/lattice/Grid_lattice_trace.h b/lib/lattice/Grid_lattice_trace.h
index 75cc5b87..4ce26170 100644
--- a/lib/lattice/Grid_lattice_trace.h
+++ b/lib/lattice/Grid_lattice_trace.h
@@ -26,7 +26,7 @@ PARALLEL_FOR_LOOP
     // Trace Index level dependent operation
     ////////////////////////////////////////////////////////////////////////////////////////////////////
     template<int Index,class vobj>
-    inline auto traceIndex(const Lattice<vobj> &lhs) -> Lattice<decltype(traceIndex<Index>(lhs._odata[0]))>
+    inline auto latTraceIndex(const Lattice<vobj> &lhs) -> Lattice<decltype(traceIndex<Index>(lhs._odata[0]))>
     {
       Lattice<decltype(traceIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
 PARALLEL_FOR_LOOP
diff --git a/lib/math/Grid_math_arith_mac.h b/lib/math/Grid_math_arith_mac.h
index 68b0acf1..06b1661d 100644
--- a/lib/math/Grid_math_arith_mac.h
+++ b/lib/math/Grid_math_arith_mac.h
@@ -27,13 +27,14 @@ strong_inline  void mac(iScalar<rtype> * __restrict__ ret,const iScalar<vtype> *
 }
 template<class rrtype,class ltype,class rtype,int N>
 strong_inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
-    for(int c2=0;c2<N;c2++){
-    for(int c1=0;c1<N;c1++){
     for(int c3=0;c3<N;c3++){
+    for(int c1=0;c1<N;c1++){
+    for(int c2=0;c2<N;c2++){
         mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c3],&rhs->_internal[c3][c2]);
     }}}
     return;
 }
+
 template<class rrtype,class ltype,class rtype,int N>
 strong_inline void mac(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs){
     for(int c1=0;c1<N;c1++){
diff --git a/lib/math/Grid_math_arith_mul.h b/lib/math/Grid_math_arith_mul.h
index 7b883cf4..04b05bb4 100644
--- a/lib/math/Grid_math_arith_mul.h
+++ b/lib/math/Grid_math_arith_mul.h
@@ -15,15 +15,22 @@ strong_inline void mult(iScalar<rtype> * __restrict__ ret,const iScalar<mtype> *
 
 template<class rrtype,class ltype,class rtype,int N>
 strong_inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
+  for(int c1=0;c1<N;c1++){
+    int c3=0;
     for(int c2=0;c2<N;c2++){
+      mult(&ret->_internal[c1][c2],&lhs->_internal[c1][c3],&rhs->_internal[c3][c2]);
+    }
+  }
+  for(int c3=1;c3<N;c3++){
     for(int c1=0;c1<N;c1++){
-        mult(&ret->_internal[c1][c2],&lhs->_internal[c1][0],&rhs->_internal[0][c2]);
-        for(int c3=1;c3<N;c3++){
-            mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c3],&rhs->_internal[c3][c2]);
-        }
-    }}
+      for(int c2=0;c2<N;c2++){
+	mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c3],&rhs->_internal[c3][c2]);
+      }
+    }
+  }
     return;
 }
+
 template<class rrtype,class ltype,class rtype,int N>
 strong_inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iScalar<rtype> * __restrict__ rhs){
     for(int c2=0;c2<N;c2++){

From 1247d7aea8b9be73e22c100817abcb4edc3de97d Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 16 May 2015 04:37:26 +0100
Subject: [PATCH 185/429] Pretty syntax

---
 lib/math/Grid_math_trace.h | 35 ++++++++---------------------------
 1 file changed, 8 insertions(+), 27 deletions(-)

diff --git a/lib/math/Grid_math_trace.h b/lib/math/Grid_math_trace.h
index db8202e6..f6778496 100644
--- a/lib/math/Grid_math_trace.h
+++ b/lib/math/Grid_math_trace.h
@@ -35,39 +35,22 @@ inline auto trace(const iScalar<vtype> &arg) -> iScalar<decltype(trace(arg._inte
 ////////////////////////////////////////////////////////////////////////////////////////////////////////
 // Trace Specific indices.
 ////////////////////////////////////////////////////////////////////////////////////////////////////////
-/*
-template<int Level,class vtype> inline 
-auto traceIndex(const iScalar<vtype> &arg) -> iScalar<decltype(traceIndex<Level>(arg._internal)) >
-{
-  iScalar<decltype(traceIndex<Level>(arg._internal))> ret;
-  ret._internal = traceIndex<Level>(arg._internal);
-  return ret;
-}
-*/
-template<int Level,class vtype> inline auto 
-traceIndex (const iScalar<vtype> &arg) ->
-typename 
-std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::notvalue, 
-  iScalar<decltype(traceIndex<Level>(arg._internal))> >::type 
-
+template<int Level,class vtype,typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline auto 
+traceIndex (const iScalar<vtype> &arg) -> iScalar<decltype(traceIndex<Level>(arg._internal))>
 {
   iScalar<decltype(traceIndex<Level>(arg._internal))> ret;
   ret._internal=traceIndex<Level>(arg._internal);
   return ret;
 }
-template<int Level,class vtype> inline auto
-traceIndex (const iScalar<vtype> &arg) ->
-typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value, 
-                        iScalar<vtype> >::type 
+template<int Level,class vtype,typename std::enable_if< iScalar<vtype>::TensorLevel == Level >::type * =nullptr> inline auto
+traceIndex (const iScalar<vtype> &arg) -> iScalar<vtype>
 {
   return arg;
 }
 
 // If we hit the right index, return scalar and trace it with no further recursion
-template<int Level,class vtype,int N> inline 
-auto traceIndex(const iMatrix<vtype,N> &arg) ->
-  typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,Level>::value,  // Index matches
-                                                    iScalar<vtype> >::type                              // return scalar
+template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::TensorLevel == Level >::type * =nullptr> inline 
+auto traceIndex(const iMatrix<vtype,N> &arg) ->  iScalar<vtype> 
 {
   iScalar<vtype> ret;
   zeroit(ret._internal);
@@ -78,10 +61,8 @@ auto traceIndex(const iMatrix<vtype,N> &arg) ->
 }
 
 // not this level, so recurse
-template<int Level,class vtype,int N> inline 
-auto traceIndex(const iMatrix<vtype,N> &arg) ->
-  typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::notvalue,// No index match
-         iMatrix<decltype(traceIndex<Level>(arg._internal[0][0])),N> >::type     // return matrix
+template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+auto traceIndex(const iMatrix<vtype,N> &arg) ->  iMatrix<decltype(traceIndex<Level>(arg._internal[0][0])),N> 
 {
   iMatrix<decltype(traceIndex<Level>(arg._internal[0][0])),N> ret;
   for(int i=0;i<N;i++){

From 9c38a52badfc9e111d6bcecb56961b9cc388d1b1 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 16 May 2015 04:40:28 +0100
Subject: [PATCH 186/429] Update Grid_lattice_trace.h

---
 lib/lattice/Grid_lattice_trace.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/lattice/Grid_lattice_trace.h b/lib/lattice/Grid_lattice_trace.h
index 4ce26170..1b35b412 100644
--- a/lib/lattice/Grid_lattice_trace.h
+++ b/lib/lattice/Grid_lattice_trace.h
@@ -26,7 +26,7 @@ PARALLEL_FOR_LOOP
     // Trace Index level dependent operation
     ////////////////////////////////////////////////////////////////////////////////////////////////////
     template<int Index,class vobj>
-    inline auto latTraceIndex(const Lattice<vobj> &lhs) -> Lattice<decltype(traceIndex<Index>(lhs._odata[0]))>
+    inline auto TraceIndex(const Lattice<vobj> &lhs) -> Lattice<decltype(traceIndex<Index>(lhs._odata[0]))>
     {
       Lattice<decltype(traceIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
 PARALLEL_FOR_LOOP

From 39e7ef1243879503910de6e7c5372e87a0b2563d Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 16 May 2015 05:49:32 +0100
Subject: [PATCH 187/429] Typoo xifed

---
 lib/lattice/Grid_lattice_trace.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/lattice/Grid_lattice_trace.h b/lib/lattice/Grid_lattice_trace.h
index 1b35b412..75cc5b87 100644
--- a/lib/lattice/Grid_lattice_trace.h
+++ b/lib/lattice/Grid_lattice_trace.h
@@ -26,7 +26,7 @@ PARALLEL_FOR_LOOP
     // Trace Index level dependent operation
     ////////////////////////////////////////////////////////////////////////////////////////////////////
     template<int Index,class vobj>
-    inline auto TraceIndex(const Lattice<vobj> &lhs) -> Lattice<decltype(traceIndex<Index>(lhs._odata[0]))>
+    inline auto traceIndex(const Lattice<vobj> &lhs) -> Lattice<decltype(traceIndex<Index>(lhs._odata[0]))>
     {
       Lattice<decltype(traceIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
 PARALLEL_FOR_LOOP

From e2f6745a0ee4088b03b467ce1d25dc6dc3b93d11 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 16 May 2015 06:40:10 +0100
Subject: [PATCH 188/429] Moved things around

---
 configure-all                                 | 11 ------
 build-all => scripts/build-all                |  6 ++--
 scripts/configure-all                         | 12 +++++++
 .../configure-commands                        | 34 ++++++++++++++-----
 configure-mic => scripts/configure-mic        |  0
 5 files changed, 41 insertions(+), 22 deletions(-)
 delete mode 100755 configure-all
 rename build-all => scripts/build-all (86%)
 create mode 100755 scripts/configure-all
 rename configure-commands => scripts/configure-commands (51%)
 rename configure-mic => scripts/configure-mic (100%)

diff --git a/configure-all b/configure-all
deleted file mode 100755
index aa67aaf0..00000000
--- a/configure-all
+++ /dev/null
@@ -1,11 +0,0 @@
-#!/bin/bash
-DIRS="build-icpc-avx build-icpc-avx512 build-clang-avx build-clang-avx-openmp build-clang-avx-openmp-mpi build-clang-avx-mpi"
-
-for D in $DIRS
-do
-  echo $D
-  mkdir -p $D
-  cd $D
-  ../configure-commands $D
-  cd ..
-done
diff --git a/build-all b/scripts/build-all
similarity index 86%
rename from build-all
rename to scripts/build-all
index cff673ab..d9aa8635 100755
--- a/build-all
+++ b/scripts/build-all
@@ -1,11 +1,11 @@
 #!/bin/bash 
+
 DIRS="build-clang-avx build-clang-avx-openmp build-clang-avx-openmp-mpi build-clang-avx-mpi build-icpc-avx512 build-icpc-avx "
 
 for D in $DIRS
 do
   echo $D
-  mkdir -p $D
-  cd $D
+  cd builds/$D
   make clean all -j 8
-  cd ..
+  cd ../../
 done
diff --git a/scripts/configure-all b/scripts/configure-all
new file mode 100755
index 00000000..1f9e758a
--- /dev/null
+++ b/scripts/configure-all
@@ -0,0 +1,12 @@
+#!/bin/bash
+
+DIRS="clang-avx clang-avx-openmp clang-avx-openmp-mpi clang-avx-mpi clang-avx2 clang-avx2-openmp clang-avx2-openmp-mpi clang-avx2-mpi icpc-avx icpc-avx2 icpc-avx512 g++-sse4 g++-avx"
+
+for D in $DIRS
+do
+  echo $D
+  mkdir -p builds/$D
+  cd builds/$D
+  ../../configure-commands $D
+  cd ../..
+done
diff --git a/configure-commands b/scripts/configure-commands
similarity index 51%
rename from configure-commands
rename to scripts/configure-commands
index 19d50a4d..183f9915 100755
--- a/configure-commands
+++ b/scripts/configure-commands
@@ -2,28 +2,46 @@
 WD=$1
 echo $WD
 case $WD in
-build-g++-sse4)
+g++-sse4)
   CXX=g++-5 ../configure --enable-simd=SSE4 CXXFLAGS="-msse4 -O3 -std=c++11" --enable-comms=none
   ;;
-build-icpc-avx)
+g++-avx)
+  CXX=g++-5 ../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11" --enable-comms=none
+  ;;
+icpc-avx)
   CXX=icpc ../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11" --enable-comms=none
   ;;
-build-icpc-avx512)
+icpc-avx2)
+  CXX=icpc ../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -std=c++11" --enable-comms=none
+  ;;
+icpc-avx512)
   CXX=icpc ../configure --enable-simd=AVX512 CXXFLAGS="-xCOMMON-AVX512 -O3  -std=c++11" --host=none  --enable-comms=none
   ;;
-build-icpc-mic)
+icpc-mic)
   CXX=icpc ../configure --host=none --enable-simd=AVX512 CXXFLAGS="-mmic -O3  -std=c++11" LDFLAGS=-mmic --enable-comms=none
   ;;
-build-clang-avx)
+clang-avx)
 CXX=clang++ ../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11"  --enable-comms=none
   ;;
-build-clang-avx-openmp)
+clang-avx2)
+CXX=clang++ ../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -std=c++11"  --enable-comms=none
+  ;;
+clang-avx-openmp)
 CXX=clang-omp++ ../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -std=c++11" LDFLAGS="-fopenmp"  --enable-comms=none
   ;;
-build-clang-avx-openmp-mpi)
+clang-avx2-openmp)
+CXX=clang-omp++ ../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -fopenmp -O3 -std=c++11" LDFLAGS="-fopenmp"  --enable-comms=none
+  ;;
+clang-avx-openmp-mpi)
 CXX=clang-omp++ ../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp"  --enable-comms=mpi
 ;;
-build-clang-avx-mpi)
+clang-avx2-openmp-mpi)
+CXX=clang-omp++ ../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp"  --enable-comms=mpi
+;;
+clang-avx-mpi)
 CXX=clang++ ../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx "  --enable-comms=mpi
 ;;
+clang-avx2-mpi)
+CXX=clang++ ../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx "  --enable-comms=mpi
+;;
 esac
diff --git a/configure-mic b/scripts/configure-mic
similarity index 100%
rename from configure-mic
rename to scripts/configure-mic

From 9b0aae665f066ab8b47a36fef44347dba704c152 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 16 May 2015 07:16:45 +0100
Subject: [PATCH 189/429] Better build automation

---
 gcc-bug-report/broken.cc   |  1 +
 scripts/build-all          | 39 +++++++++++++++++++++++++++++++--
 scripts/configure-all      |  3 +--
 scripts/configure-commands | 45 +++++++++++++++++++++++++-------------
 4 files changed, 69 insertions(+), 19 deletions(-)

diff --git a/gcc-bug-report/broken.cc b/gcc-bug-report/broken.cc
index bd403e85..d23572d3 100644
--- a/gcc-bug-report/broken.cc
+++ b/gcc-bug-report/broken.cc
@@ -56,4 +56,5 @@ int main(int argc,char **argv)
   Container<Recursive<Recursive<ComplexD> > > ret(10);
   
   ret = function<1>(array);
+  ret = function<2>(array);
 }
diff --git a/scripts/build-all b/scripts/build-all
index d9aa8635..d3a978b7 100755
--- a/scripts/build-all
+++ b/scripts/build-all
@@ -1,11 +1,46 @@
 #!/bin/bash 
 
-DIRS="build-clang-avx build-clang-avx-openmp build-clang-avx-openmp-mpi build-clang-avx-mpi build-icpc-avx512 build-icpc-avx "
+DIRS="clang-avx clang-avx-openmp clang-avx-openmp-mpi clang-avx-mpi clang-avx2 clang-avx2-openmp clang-avx2-openmp-mpi clang-avx2-mpi"
+EXTRADIRS="g++-avx g++-sse4 icpc-avx icpc-avx2 icpc-avx512"
+BLACK="\033[30m"
+RED="\033[31m"
+GREEN="\033[32m"
+YELLOW="\033[33m"
+BLUE="\033[34m"
+PINK="\033[35m"
+CYAN="\033[36m"
+WHITE="\033[37m"
+NORMAL="\033[0;39m"
 
 for D in $DIRS
 do
-  echo $D
+
+echo
+echo -e $RED ==============================
+echo -e $GREEN $D
+echo -e $RED ==============================
+echo -e $BLUE
+
   cd builds/$D
   make clean all -j 8
   cd ../../
+echo -e $NORMAL
 done
+
+if [ "X$1" == "Xextra" ]
+then
+for D in $EXTRADIRS
+do
+
+echo
+echo -e $RED ==============================
+echo -e $RED $D
+echo -e $RED ==============================
+echo -e $BLUE
+
+  cd builds/$D
+  make clean all -j 8
+  cd ../../
+echo -e $NORMAL
+done
+fi
\ No newline at end of file
diff --git a/scripts/configure-all b/scripts/configure-all
index 1f9e758a..2f80b60a 100755
--- a/scripts/configure-all
+++ b/scripts/configure-all
@@ -4,9 +4,8 @@ DIRS="clang-avx clang-avx-openmp clang-avx-openmp-mpi clang-avx-mpi clang-avx2 c
 
 for D in $DIRS
 do
-  echo $D
   mkdir -p builds/$D
   cd builds/$D
-  ../../configure-commands $D
+  ../../scripts/configure-commands $D
   cd ../..
 done
diff --git a/scripts/configure-commands b/scripts/configure-commands
index 183f9915..5e83fa89 100755
--- a/scripts/configure-commands
+++ b/scripts/configure-commands
@@ -1,47 +1,62 @@
 #!/bin/bash
 WD=$1
-echo $WD
+BLACK="\033[30m"
+RED="\033[31m"
+GREEN="\033[32m"
+YELLOW="\033[33m"
+BLUE="\033[34m"
+PINK="\033[35m"
+CYAN="\033[36m"
+WHITE="\033[37m"
+NORMAL="\033[0;39m"
+echo
+echo -e $RED ==============================
+echo -e $GREEN $WD
+echo -e $RED ==============================
+echo -e $YELLOW
+
 case $WD in
 g++-sse4)
-  CXX=g++-5 ../configure --enable-simd=SSE4 CXXFLAGS="-msse4 -O3 -std=c++11" --enable-comms=none
+  CXX=g++-5 ../../configure --enable-simd=SSE4 CXXFLAGS="-msse4 -O3 -std=c++11" --enable-comms=none
   ;;
 g++-avx)
-  CXX=g++-5 ../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11" --enable-comms=none
+  CXX=g++-5 ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11" --enable-comms=none
   ;;
 icpc-avx)
-  CXX=icpc ../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11" --enable-comms=none
+  CXX=icpc ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11" --enable-comms=none
   ;;
 icpc-avx2)
-  CXX=icpc ../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -std=c++11" --enable-comms=none
+  CXX=icpc ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -std=c++11" --enable-comms=none
   ;;
 icpc-avx512)
-  CXX=icpc ../configure --enable-simd=AVX512 CXXFLAGS="-xCOMMON-AVX512 -O3  -std=c++11" --host=none  --enable-comms=none
+  CXX=icpc ../../configure --enable-simd=AVX512 CXXFLAGS="-xCOMMON-AVX512 -O3  -std=c++11" --host=none  --enable-comms=none
   ;;
 icpc-mic)
-  CXX=icpc ../configure --host=none --enable-simd=AVX512 CXXFLAGS="-mmic -O3  -std=c++11" LDFLAGS=-mmic --enable-comms=none
+  CXX=icpc ../../configure --host=none --enable-simd=AVX512 CXXFLAGS="-mmic -O3  -std=c++11" LDFLAGS=-mmic --enable-comms=none
   ;;
 clang-avx)
-CXX=clang++ ../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11"  --enable-comms=none
+CXX=clang++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11"  --enable-comms=none
   ;;
 clang-avx2)
-CXX=clang++ ../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -std=c++11"  --enable-comms=none
+CXX=clang++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -std=c++11"  --enable-comms=none
   ;;
 clang-avx-openmp)
-CXX=clang-omp++ ../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -std=c++11" LDFLAGS="-fopenmp"  --enable-comms=none
+CXX=clang-omp++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -std=c++11" LDFLAGS="-fopenmp"  --enable-comms=none
   ;;
 clang-avx2-openmp)
-CXX=clang-omp++ ../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -fopenmp -O3 -std=c++11" LDFLAGS="-fopenmp"  --enable-comms=none
+CXX=clang-omp++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -fopenmp -O3 -std=c++11" LDFLAGS="-fopenmp"  --enable-comms=none
   ;;
 clang-avx-openmp-mpi)
-CXX=clang-omp++ ../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp"  --enable-comms=mpi
+CXX=clang-omp++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp"  --enable-comms=mpi
 ;;
 clang-avx2-openmp-mpi)
-CXX=clang-omp++ ../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp"  --enable-comms=mpi
+CXX=clang-omp++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp"  --enable-comms=mpi
 ;;
 clang-avx-mpi)
-CXX=clang++ ../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx "  --enable-comms=mpi
+CXX=clang++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx "  --enable-comms=mpi
 ;;
 clang-avx2-mpi)
-CXX=clang++ ../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx "  --enable-comms=mpi
+CXX=clang++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx "  --enable-comms=mpi
 ;;
 esac
+echo -e $NORMAL

From dc6b6bdc96965e4c1cf3c438e0f7b3d0c5b6559b Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 16 May 2015 23:35:08 +0100
Subject: [PATCH 190/429] Updating preparing for solvers etc..

---
 benchmarks/Grid_memory_bandwidth.cc      |  20 +-
 benchmarks/Grid_su3.cc                   |  16 +-
 lib/algorithms/LinearOperator.h          | 207 ++++++-
 lib/algorithms/approx/PolynomialApprox.h | 145 +++++
 lib/algorithms/approx/zolotarev.c        | 710 +++++++++++++++++++++++
 lib/algorithms/approx/zolotarev.h        |  85 +++
 lib/qcd/Grid_qcd_wilson_dop.cc           |   5 +-
 lib/qcd/Grid_qcd_wilson_dop.h            |  16 +-
 8 files changed, 1170 insertions(+), 34 deletions(-)
 create mode 100644 lib/algorithms/approx/PolynomialApprox.h
 create mode 100755 lib/algorithms/approx/zolotarev.c
 create mode 100755 lib/algorithms/approx/zolotarev.h

diff --git a/benchmarks/Grid_memory_bandwidth.cc b/benchmarks/Grid_memory_bandwidth.cc
index 5a4d1f18..8ece09e8 100644
--- a/benchmarks/Grid_memory_bandwidth.cc
+++ b/benchmarks/Grid_memory_bandwidth.cc
@@ -19,7 +19,7 @@ int main (int argc, char ** argv)
   std::cout << "===================================================================================================="<<std::endl;
   std::cout << "= Benchmarking fused AXPY bandwidth"<<std::endl;
   std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s"<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s"<<"\t\t"<<"Gflop/s"<<std::endl;
   std::cout << "----------------------------------------------------------"<<std::endl;
 
   for(int lat=4;lat<=32;lat+=4){
@@ -44,15 +44,16 @@ int main (int argc, char ** argv)
       double stop=usecond();
       double time = (stop-start)/Nloop*1000;
       
+      double flops=lat*lat*lat*lat*Nvec*2;// mul,add
       double bytes=3*lat*lat*lat*lat*Nvec*sizeof(Real);
-      std::cout<<std::setprecision(2) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
+      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"   \t\t"<<bytes/time<<"\t\t"<<flops/time<<std::endl;
 
     }
 
   std::cout << "===================================================================================================="<<std::endl;
   std::cout << "= Benchmarking a*x + y bandwidth"<<std::endl;
   std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s"<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s"<<"\t\t"<<"Gflop/s"<<std::endl;
   std::cout << "----------------------------------------------------------"<<std::endl;
   
   for(int lat=4;lat<=32;lat+=4){
@@ -75,15 +76,16 @@ int main (int argc, char ** argv)
       double stop=usecond();
       double time = (stop-start)/Nloop*1000;
      
+      double flops=lat*lat*lat*lat*Nvec*2;// mul,add
       double bytes=3*lat*lat*lat*lat*Nvec*sizeof(Real);
-      std::cout<<std::setprecision(2) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
+      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"   \t\t"<<bytes/time<<"\t\t"<<flops/time<<std::endl;
 
     }
 
   std::cout << "===================================================================================================="<<std::endl;
   std::cout << "= Benchmarking SCALE bandwidth"<<std::endl;
   std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s"<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s"<<"\t\t"<<"Gflop/s"<<std::endl;
 
   for(int lat=4;lat<=32;lat+=4){
 
@@ -107,14 +109,15 @@ int main (int argc, char ** argv)
       double time = (stop-start)/Nloop*1000;
       
       double bytes=2*lat*lat*lat*lat*Nvec*sizeof(Real);
-      std::cout <<std::setprecision(2) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
+      double flops=lat*lat*lat*lat*Nvec*1;// mul
+      std::cout <<std::setprecision(3) << lat<<"\t\t"<<bytes<<"   \t\t"<<bytes/time<<"\t\t"<<flops/time<<std::endl;
 
   }
 
   std::cout << "===================================================================================================="<<std::endl;
   std::cout << "= Benchmarking READ bandwidth"<<std::endl;
   std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s"<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s"<<"\t\t"<<"Gflop/s"<<std::endl;
   std::cout << "----------------------------------------------------------"<<std::endl;
 
   for(int lat=4;lat<=32;lat+=4){
@@ -139,7 +142,8 @@ int main (int argc, char ** argv)
       double time = (stop-start)/Nloop*1000;
       
       double bytes=lat*lat*lat*lat*Nvec*sizeof(Real);
-      std::cout<<std::setprecision(2) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<std::endl;
+      double flops=lat*lat*lat*lat*Nvec*2;// mul,add
+      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"  \t\t"<<bytes/time<<"\t\t"<<flops/time<<std::endl;
 
   }    
 
diff --git a/benchmarks/Grid_su3.cc b/benchmarks/Grid_su3.cc
index ce3ccaa9..eb4dc3a1 100644
--- a/benchmarks/Grid_su3.cc
+++ b/benchmarks/Grid_su3.cc
@@ -16,7 +16,7 @@ int main (int argc, char ** argv)
   std::cout << "===================================================================================================="<<std::endl;
   std::cout << "= Benchmarking SU3xSU3  x= x*y"<<std::endl;
   std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
   std::cout << "----------------------------------------------------------"<<std::endl;
 
   for(int lat=2;lat<=24;lat+=2){
@@ -40,7 +40,7 @@ int main (int argc, char ** argv)
       double bytes=3.0*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
       double footprint=2.0*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
       double flops=Nc*Nc*(6.0+8.0+8.0)*lat*lat*lat*lat;
-      std::cout<<std::setprecision(3) << lat<<"\t\t"<<footprint<<"\t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
+      std::cout<<std::setprecision(3) << lat<<"\t\t"<<footprint<<"    \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
 
     }
 
@@ -48,7 +48,7 @@ int main (int argc, char ** argv)
   std::cout << "===================================================================================================="<<std::endl;
   std::cout << "= Benchmarking SU3xSU3  z= x*y"<<std::endl;
   std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
   std::cout << "----------------------------------------------------------"<<std::endl;
 
   for(int lat=2;lat<=24;lat+=2){
@@ -71,14 +71,14 @@ int main (int argc, char ** argv)
       
       double bytes=3*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
       double flops=Nc*Nc*(6+8+8)*lat*lat*lat*lat;
-      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
+      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"    \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
 
     }
 
   std::cout << "===================================================================================================="<<std::endl;
   std::cout << "= Benchmarking SU3xSU3  mult(z,x,y)"<<std::endl;
   std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
   std::cout << "----------------------------------------------------------"<<std::endl;
 
   for(int lat=2;lat<=24;lat+=2){
@@ -101,14 +101,14 @@ int main (int argc, char ** argv)
       
       double bytes=3*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
       double flops=Nc*Nc*(6+8+8)*lat*lat*lat*lat;
-      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
+      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"    \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
 
     }
 
   std::cout << "===================================================================================================="<<std::endl;
   std::cout << "= Benchmarking SU3xSU3  mac(z,x,y)"<<std::endl;
   std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
+  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
   std::cout << "----------------------------------------------------------"<<std::endl;
 
   for(int lat=2;lat<=24;lat+=2){
@@ -131,7 +131,7 @@ int main (int argc, char ** argv)
       
       double bytes=3*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
       double flops=Nc*Nc*(8+8+8)*lat*lat*lat*lat;
-      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"\t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
+      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"   \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
 
     }
 
diff --git a/lib/algorithms/LinearOperator.h b/lib/algorithms/LinearOperator.h
index 5929570f..c449ee8e 100644
--- a/lib/algorithms/LinearOperator.h
+++ b/lib/algorithms/LinearOperator.h
@@ -5,14 +5,211 @@
 
 namespace Grid {
 
-  // Red black cases?
-    template<class vtype> class LinearOperatorBase {
+  /////////////////////////////////////////////////////////////////////////////////////////////
+  // LinearOperators Take a something and return a something.
+  /////////////////////////////////////////////////////////////////////////////////////////////
+  //
+  // Hopefully linearity is satisfied and the AdjOp is indeed the Hermitian conjugate (transpose if real):
+  //
+  //   i)  F(a x + b y) = aF(x) + b F(y).
+  //  ii)  <x|Op|y> = <y|AdjOp|x>^\ast
+  //
+  // Would be fun to have a test linearity & Herm Conj function!
+  /////////////////////////////////////////////////////////////////////////////////////////////
+    template<class Field> class LinearOperatorBase {
     public:
-      void M(const Lattice<vtype> &in, Lattice<vtype> &out){ assert(0);}
-      void Mdag(const Lattice<vtype> &in, Lattice<vtype> &out){ assert(0);}
-      void MdagM(const Lattice<vtype> &in, Lattice<vtype> &out){ assert(0);}
+      virtual void Op     (const Field &in, Field &out) = 0; // Abstract base
+      virtual void AdjOp  (const Field &in, Field &out) = 0; // Abstract base
     };
 
+  /////////////////////////////////////////////////////////////////////////////////////////////
+  // Hermitian operators are self adjoint and only require Op to be defined, so refine the base
+  /////////////////////////////////////////////////////////////////////////////////////////////
+    template<class Field> class HermitianOperatorBase : public LinearOperatorBase<Field> {
+    public:
+      void AdjOp(const Field &in, Field &out) {
+	return Op(in,out);
+      };
+    };
+
+  /////////////////////////////////////////////////////////////////////////////////////////////
+  // Interface defining what I expect of a general sparse matrix
+  /////////////////////////////////////////////////////////////////////////////////////////////
+    template<class Field> class SparseMatrixBase {
+    public:
+      // Full checkerboar operations
+      virtual void M    (const Field &in, Field &out);
+      virtual void Mdag (const Field &in, Field &out);
+      virtual void MdagM(const Field &in, Field &out);
+      
+      // half checkerboard operaions
+      virtual void Mpc      (const Field &in, Field &out);
+      virtual void MpcDag   (const Field &in, Field &out);
+      virtual void MpcDagMpc(const Field &in, Field &out);
+    };
+
+  /////////////////////////////////////////////////////////////////////////////////////////////
+  // Whereas non hermitian takes a generic sparse matrix (e.g. lattice action)
+  // conforming to sparse matrix interface and builds the full checkerboard non-herm operator
+  // Op and AdjOp distinct.
+  // By sharing the class for Sparse Matrix across multiple operator wrappers, we can share code
+  // between RB and non-RB variants. Sparse matrix is like the fermion action def, and then
+  // the wrappers implement the specialisation of "Op" and "AdjOp" to the cases minimising
+  // replication of code.
+  /////////////////////////////////////////////////////////////////////////////////////////////
+    template<class SparseMatrix,class Field>
+    class NonHermitianOperator : public LinearOperatorBase<Field> {
+      SparseMatrix &_Mat;
+    public:
+      NonHermitianOperator(SparseMatrix &Mat): _Mat(Mat){};
+      void Op     (const Field &in, Field &out){
+	_Mat.M(in,out);
+      }
+      void AdjOp     (const Field &in, Field &out){
+	_Mat.Mdag(in,out);
+      }
+    };
+    
+    ////////////////////////////////////////////////////////////////////////////////////
+    // Redblack Non hermitian wrapper
+    ////////////////////////////////////////////////////////////////////////////////////
+    template<class SparseMatrix,class Field>
+    class NonHermitianRedBlackOperator : public LinearOperatorBase<Field> {
+      SparseMatrix &_Mat;
+    public:
+      NonHermitianRedBlackOperator(SparseMatrix &Mat): _Mat(Mat){};
+      void Op     (const Field &in, Field &out){
+	this->Mpc(in,out);
+      }
+      void AdjOp     (const Field &in, Field &out){ //
+	this->MpcDag(in,out);
+      }
+    };
+
+    ////////////////////////////////////////////////////////////////////////////////////
+    // Hermitian wrapper
+    ////////////////////////////////////////////////////////////////////////////////////
+    template<class SparseMatrix,class Field>
+    class HermitianOperator : public HermitianOperatorBase<Field> {
+      SparseMatrix &_Mat;
+    public:
+      HermitianOperator(SparseMatrix &Mat): _Mat(Mat) {};
+
+      void Op     (const Field &in, Field &out){
+	this->Mpc(in,out);
+      }
+      void AdjOp     (const Field &in, Field &out){ //
+	this->MpcDag(in,out);
+      }
+    };
+
+    ////////////////////////////////////////////////////////////////////////////////////
+    // Hermitian RedBlack wrapper
+    ////////////////////////////////////////////////////////////////////////////////////
+    template<class SparseMatrix,class Field>
+    class HermitianRedBlackOperator : public HermitianOperatorBase<Field> {
+      SparseMatrix &_Mat;
+    public:
+      HermitianRedBlackOperator(SparseMatrix &Mat): _Mat(Mat) {};
+      void Op     (const Field &in, Field &out){
+	this->MpcDagMpc(in,out);
+      }
+    };
+
+
+    /////////////////////////////////////////////////////////////
+    // Base classes for functions of operators
+    /////////////////////////////////////////////////////////////
+    template<class Field> class OperatorFunction {
+    public:
+      virtual void operator() (LinearOperatorBase<Field> *Linop, const Field &in, Field &out) = 0;
+    };
+
+    /////////////////////////////////////////////////////////////
+    // Base classes for polynomial functions of operators ? needed?
+    /////////////////////////////////////////////////////////////
+    template<class Field> class OperatorPolynomial : public OperatorFunction<Field> {
+    public:
+      virtual void operator() (LinearOperatorBase<Field> *Linop,const Field &in, Field &out) = 0;
+    };
+
+    /////////////////////////////////////////////////////////////
+    // Base classes for iterative processes based on operators
+    // single input vec, single output vec.
+    /////////////////////////////////////////////////////////////
+    template<class Field> class IterativeProcess : public OperatorFunction<Field> {
+    public:
+      RealD   Tolerance;
+      Integer MaxIterations;
+      virtual void operator() (LinearOperatorBase<Field> *Linop,const Field &in, Field &out) = 0;
+    };
+
+    /////////////////////////////////////////////////////////////
+    // Grand daddy iterative method
+    /////////////////////////////////////////////////////////////
+    template<class Field> class ConjugateGradient : public IterativeProcess<Field> {
+    public:
+      virtual void operator() (HermitianOperatorBase<Field> *Linop,const Field &in, Field &out) = 0;
+    };
+
+    /////////////////////////////////////////////////////////////
+    // A little more modern
+    /////////////////////////////////////////////////////////////
+    template<class Field> class PreconditionedConjugateGradient : public  IterativeProcess<Field>  {
+    public:
+      void operator() (HermitianOperatorBase<Field> *Linop,
+		       OperatorFunction<Field> *Preconditioner,
+		       const Field &in, 
+		       Field &out) = 0;
+    };
+
+    // FIXME : To think about
+
+    // Chroma functionality list defining LinearOperator
+    /*
+     virtual void operator() (T& chi, const T& psi, enum PlusMinus isign) const = 0;
+     virtual void operator() (T& chi, const T& psi, enum PlusMinus isign, Real epsilon) const
+     virtual const Subset& subset() const = 0;
+     virtual unsigned long nFlops() const { return 0; }
+     virtual void deriv(P& ds_u, const T& chi, const T& psi, enum PlusMinus isign) const
+     class UnprecLinearOperator : public DiffLinearOperator<T,P,Q>
+       const Subset& subset() const {return all;}
+     };
+    */
+
+    // Chroma interface defining GaugeAction
+    /*
+      template<typename P, typename Q>   class GaugeAction
+  virtual const CreateGaugeState<P,Q>& getCreateState() const = 0;
+  virtual GaugeState<P,Q>* createState(const Q& q) const
+  virtual const GaugeBC<P,Q>& getGaugeBC() const
+  virtual const Set& getSet(void) const = 0;
+  virtual void deriv(P& result, const Handle< GaugeState<P,Q> >& state) const 
+  virtual Double S(const Handle< GaugeState<P,Q> >& state) const = 0;
+
+  class LinearGaugeAction : public GaugeAction< multi1d<LatticeColorMatrix>, multi1d<LatticeColorMatrix> >
+  typedef multi1d<LatticeColorMatrix>  P;
+  typedef multi1d<LatticeColorMatrix>  Q;
+  virtual void staple(LatticeColorMatrix& result,
+		      const Handle< GaugeState<P,Q> >& state,
+		      int mu, int cb) const = 0;
+    */
+
+    // Chroma interface defining FermionAction
+    /*
+     template<typename T, typename P, typename Q>  class FermAct4D : public FermionAction<T,P,Q>
+     virtual LinearOperator<T>* linOp(Handle< FermState<T,P,Q> > state) const = 0;
+     virtual LinearOperator<T>* lMdagM(Handle< FermState<T,P,Q> > state) const = 0;
+     virtual LinOpSystemSolver<T>* invLinOp(Handle< FermState<T,P,Q> > state,
+     virtual MdagMSystemSolver<T>* invMdagM(Handle< FermState<T,P,Q> > state,
+     virtual LinOpMultiSystemSolver<T>* mInvLinOp(Handle< FermState<T,P,Q> > state,
+     virtual MdagMMultiSystemSolver<T>* mInvMdagM(Handle< FermState<T,P,Q> > state,
+     virtual MdagMMultiSystemSolverAccumulate<T>* mInvMdagMAcc(Handle< FermState<T,P,Q> > state,
+     virtual SystemSolver<T>* qprop(Handle< FermState<T,P,Q> > state,
+     class DiffFermAct4D : public FermAct4D<T,P,Q>
+     virtual DiffLinearOperator<T,Q,P>* linOp(Handle< FermState<T,P,Q> > state) const = 0;
+     virtual DiffLinearOperator<T,Q,P>* lMdagM(Handle< FermState<T,P,Q> > state) const = 0;
+    */
 }
 
 #endif
diff --git a/lib/algorithms/approx/PolynomialApprox.h b/lib/algorithms/approx/PolynomialApprox.h
new file mode 100644
index 00000000..5f59a078
--- /dev/null
+++ b/lib/algorithms/approx/PolynomialApprox.h
@@ -0,0 +1,145 @@
+#ifndef GRID_POLYNOMIAL_APPROX_H
+#define GRID_POLYNOMIAL_APPROX_H
+
+namespace Grid {
+
+  ////////////////////////////////////////////////////////////////////////////////////////////
+  // Simple general polynomial with user supplied coefficients
+  ////////////////////////////////////////////////////////////////////////////////////////////
+  template<class Field>
+  class Polynomial : public OperatorFunction<Field> {
+  private:
+    std::vector<double> _oeffs;
+  public:
+    Polynomial(std::vector<double> &_Coeffs) : Coeffs(_Coeffs) {};
+
+    // Implement the required interface
+    void operator() (LinearOperatorBase<Field> &Linop, const Field &in, Field &out) {
+
+      Field AtoN = in;
+      out = AtoN*Coeffs[0];
+
+      for(int n=1;n<Coeffs.size();n++){
+	Field Mtmp=AtoN;
+	Linop.Op(Mtmp,AtoN);
+	out=out+AtoN*Coeffs[n];
+      }
+    };
+  };
+
+  ////////////////////////////////////////////////////////////////////////////////////////////
+  // Generic Chebyshev approximations
+  ////////////////////////////////////////////////////////////////////////////////////////////
+  template<class Field>
+  class Chebyshev : public OperatorFunction<Field> {
+  private:
+    std::vector<double> Coeffs;
+    int order;
+    double hi;
+    double lo;
+
+  public:
+    Chebyshev(double _lo,double _hi,int _order, double (* func)(double) ){
+      lo=_lo;
+      hi=_hi;
+      order=_order;
+      
+      if(order < 2) exit(-1);
+      Coeffs.resize(order);
+      for(int j=0;j<order;j++){
+	double s=0;
+	for(int k=0;k<order;k++){
+	  double y=cos(M_PI*(k+0.5)/order);
+	  double x=0.5*(y*(hi-lo)+(hi+lo));
+	  double f=func(x);
+	  s=s+f*cos( j*M_PI*(k+0.5)/order );
+	}
+	Coeffs[j] = s * 2.0/order;
+      }
+    };
+
+    double Evaluate(double x) // Convenience for plotting the approximation
+    {
+      double Tn;
+      double Tnm;
+      double Tnp;
+      
+      double y=( x-0.5*(hi+lo))/(0.5*(hi-lo));
+      
+      double T0=1;
+      double T1=y;
+      
+      double sum;
+      sum = 0.5*Coeffs[0]*T0;
+      sum+= Coeffs[1]*T1;
+      
+      Tn =T1;
+      Tnm=T0;
+      for(int i=2;i<order;i++){
+	Tnp=2*y*Tn-Tnm;
+	Tnm=Tn;
+	Tn =Tnp;
+	sum+= Tn*Coeffs[i];
+      }
+      return sum;
+    };
+
+    // Convenience for plotting the approximation
+    void   PlotApprox(std::ostream &out) {
+      out<<"Polynomial approx ["<<lo<<","<<hi<<"]"<<std::endl;
+      for(double x=lo;x<hi;x+=(hi-lo)/50.0){
+	out <<x<<"\t"<<Evaluate(x)<<std::endl;
+      }
+    };
+
+    // Implement the required interface; could require Lattice base class
+    void operator() (LinearOperatorBase<Field> &Linop, const Field &in, Field &out) {
+
+      Field T0 = in;
+      Field T1 = T0; // Field T1(T0._grid); more efficient but hardwires Lattice class
+      Field T2 = T1;
+      
+      // use a pointer trick to eliminate copies
+      Field *Tnm = &T0;
+      Field *Tn  = &T1;
+      Field *Tnp = &T2;
+      Field y   = in;
+  
+      double xscale = 2.0/(hi-lo);
+      double mscale = -(hi+lo)/(hi-lo);
+
+      Field *T0=Tnm;
+      Field *T1=Tn;
+      
+
+      // Tn=T1 = (xscale M + mscale)in
+      Linop.Op(T0,y);
+
+      T1=y*xscale+in*mscale;
+
+      // sum = .5 c[0] T0 + c[1] T1
+      out = (0.5*coeffs[0])*T0 + coeffs[1]*T1;
+
+      for(int n=2;n<order;n++){
+	
+	Linop.Op(*Tn,y);
+
+	y=xscale*y+mscale*(*Tn);
+
+	*Tnp=2.0*y-(*Tnm);
+	
+	out=out+coeffs[n]* (*Tnp);
+
+	// Cycle pointers to avoid copies
+	Field *swizzle = Tnm;
+	Tnm    =Tn;
+	Tn     =Tnp;
+	Tnp    =swizzle;
+	  
+      }
+    }
+  };
+
+
+}
+#endif
diff --git a/lib/algorithms/approx/zolotarev.c b/lib/algorithms/approx/zolotarev.c
new file mode 100755
index 00000000..dea7a439
--- /dev/null
+++ b/lib/algorithms/approx/zolotarev.c
@@ -0,0 +1,710 @@
+/* -*- Mode: C; comment-column: 22; fill-column: 79; compile-command: "gcc -o zolotarev zolotarev.c -ansi -pedantic -lm -DTEST"; -*- */
+#define VERSION Source Time-stamp: <19-OCT-2004 09:33:22.00 adk@MISSCONTRARY>
+
+/* These C routines evalute the optimal rational approximation to the signum
+ * function for epsilon < |x| < 1 using Zolotarev's theorem.
+ *
+ * To obtain reliable results for high degree approximations (large n) it is
+ * necessary to compute using sufficiently high precision arithmetic. To this
+ * end the code has been parameterised to work with the preprocessor names
+ * INTERNAL_PRECISION and PRECISION set to float, double, or long double as
+ * appropriate. INTERNAL_PRECISION is used in computing the Zolotarev
+ * coefficients, which are converted to PRECISION before being returned to the
+ * caller. Presumably even higher precision could be obtained using GMP or
+ * similar package, but bear in mind that rounding errors might also be
+ * significant in evaluating the resulting polynomial. The convergence criteria
+ * have been written in a precision-independent form. */
+
+#include <math.h>
+#include <stdlib.h>
+#include <stdio.h>
+
+#define MAX(a,b) ((a) > (b) ? (a) : (b))
+#define MIN(a,b) ((a) < (b) ? (a) : (b))
+
+#ifndef INTERNAL_PRECISION
+#define INTERNAL_PRECISION double
+#endif
+
+#include "zolotarev.h"
+#define ZOLOTAREV_INTERNAL
+#undef ZOLOTAREV_DATA
+#define ZOLOTAREV_DATA izd
+#undef ZPRECISION
+#define ZPRECISION INTERNAL_PRECISION
+#include "zolotarev.h"
+#undef ZOLOTAREV_INTERNAL
+
+/* The ANSI standard appears not to know what pi is */
+
+#ifndef M_PI
+#define M_PI ((INTERNAL_PRECISION) 3.141592653589793238462643383279502884197\
+169399375105820974944592307816406286208998628034825342117068)
+#endif
+
+#define ZERO ((INTERNAL_PRECISION) 0)
+#define ONE ((INTERNAL_PRECISION) 1)
+#define TWO ((INTERNAL_PRECISION) 2)
+#define THREE ((INTERNAL_PRECISION) 3)
+#define FOUR ((INTERNAL_PRECISION) 4)
+#define HALF (ONE/TWO)
+
+/* The following obscenity seems to be the simplest (?) way to coerce the C
+ * preprocessor to convert the value of a preprocessor token into a string. */
+
+#define PP2(x) #x
+#define PP1(a,b,c) a ## b(c)
+#define STRINGIFY(name) PP1(PP,2,name)
+
+/* Compute the partial fraction expansion coefficients (alpha) from the
+ * factored form */
+
+static void construct_partfrac(izd *z) {
+  int dn = z -> dn, dd = z -> dd, type = z -> type;
+  int j, k, da = dd + 1 + type;
+  INTERNAL_PRECISION A = z -> A, *a = z -> a, *ap = z -> ap, *alpha;
+  alpha = (INTERNAL_PRECISION*) malloc(da * sizeof(INTERNAL_PRECISION));
+  for (j = 0; j < dd; j++)
+    for (k = 0, alpha[j] = A; k < dd; k++)
+      alpha[j] *=
+	(k < dn ? ap[j] - a[k] : ONE) / (k == j ? ONE : ap[j] - ap[k]);
+  if(type == 1)	      /* implicit pole at zero? */
+    for (k = 0, alpha[dd] = A * (dn > dd ? - a[dd] : ONE); k < dd; k++) {
+      alpha[dd] *= a[k] / ap[k];
+      alpha[k] *= (dn > dd ? ap[k] - a[dd] : ONE) / ap[k];
+    }
+  alpha[da-1] = dn == da - 1 ? A : ZERO;
+  z -> alpha = alpha;
+  z -> da = da;
+  return;
+}
+
+/* Convert factored polynomial into dense polynomial. The input is the overall
+ * factor A and the roots a[i], such that p = A product(x - a[i], i = 1..d) */
+
+static INTERNAL_PRECISION *poly_factored_to_dense(INTERNAL_PRECISION A, 
+						  INTERNAL_PRECISION *a,
+						  int d) {
+  INTERNAL_PRECISION *p;
+  int i, j;
+  p = (INTERNAL_PRECISION *) malloc((d + 2) * sizeof(INTERNAL_PRECISION));
+  p[0] = A;
+  for (i = 0; i < d; i++) {
+    p[i+1] = p[i];
+    for (j = i; j > 0; j--) p[j] = p[j-1] - a[i]*p[j];
+    p[0] *= - a[i];
+  }
+  return p;
+}
+
+/* Convert a rational function of the form R0(x) = x p(x^2)/q(x^2) (type 0) or
+ * R1(x) = p(x^2)/[x q(x^2)] (type 1) into its continued fraction
+ * representation. We assume that 0 <= deg(q) - deg(p) <= 1 for type 0 and 0 <=
+ * deg(p) - deg(q) <= 1 for type 1. On input p and q are in factored form, and
+ * deg(q) = dq, deg(p) = dp.  The output is the continued fraction coefficients
+ * beta, where R(x) = beta[0] x + 1/(beta[1] x + 1/(...)).
+ *
+ * The method used is as follows. There are four cases to consider:
+ *
+ * 0.i.  Type 0, deg p = deg q
+ *
+ * 0.ii. Type 0, deg p = deg q - 1
+ *
+ * 1.i.  Type 1, deg p = deg q
+ *
+ * 1.ii. Type 1, deg p = deg q + 1
+ *
+ * and these are connected by two transformations:
+ *
+ * A. To obtain a continued fraction expansion of type 1 we use a single-step
+ * polynomial division we find beta and r(x) such that p(x) = beta x q(x) +
+ * r(x), with deg(r) = deg(q). This implies that p(x^2) = beta x^2 q(x^2) +
+ * r(x^2), and thus R1(x) = x beta + r(x^2)/(x q(x^2)) = x beta + 1/R0(x)
+ * with R0(x) = x q(x^2)/r(x^2).
+ *
+ * B. A continued fraction expansion of type 0 is obtained in a similar, but
+ * not identical, manner. We use the polynomial division algorithm to compute
+ * the quotient beta and the remainder r that satisfy p(x) = beta q(x) + r(x)
+ * with deg(r) = deg(q) - 1. We thus have x p(x^2) = x beta q(x^2) + x r(x^2),
+ * so R0(x) = x beta + x r(x^2)/q(x^2) = x beta + 1/R1(x) with R1(x) = q(x^2) /
+ * (x r(x^2)).
+ *
+ * Note that the deg(r) must be exactly deg(q) for (A) and deg(q) - 1 for (B)
+ * because p and q have disjoint roots all of multiplicity 1. This means that
+ * the division algorithm requires only a single polynomial subtraction step.
+ *
+ * The transformations between the cases form the following finite state
+ * automaton:
+ *
+ * +------+            +------+            +------+            +------+
+ * |      |            |      | ---(A)---> |      |            |      |
+ * | 0.ii | ---(B)---> | 1.ii |            | 0.i  | <---(A)--- | 1.i  |
+ * |      |            |      | <---(B)--- |      |            |      |
+ * +------+            +------+            +------+            +------+
+ */
+
+static INTERNAL_PRECISION *contfrac_A(INTERNAL_PRECISION *,
+				      INTERNAL_PRECISION *,
+				      INTERNAL_PRECISION *,
+				      INTERNAL_PRECISION *, int, int);
+
+static INTERNAL_PRECISION *contfrac_B(INTERNAL_PRECISION *,
+				      INTERNAL_PRECISION *,
+				      INTERNAL_PRECISION *,
+				      INTERNAL_PRECISION *, int, int);
+
+static void construct_contfrac(izd *z){
+  INTERNAL_PRECISION *r, A = z -> A, *p = z -> a, *q = z -> ap;
+  int dp = z -> dn, dq = z -> dd, type = z -> type;
+
+  z -> db = 2 * dq + 1 + type;
+  z -> beta = (INTERNAL_PRECISION *)
+    malloc(z -> db * sizeof(INTERNAL_PRECISION));
+  p = poly_factored_to_dense(A, p, dp);
+  q = poly_factored_to_dense(ONE, q, dq);
+  r = (INTERNAL_PRECISION *) malloc((MAX(dp,dq) + 1) *
+				    sizeof(INTERNAL_PRECISION));
+  if (type == 0) (void) contfrac_B(z -> beta, p, q, r, dp, dq);
+  else (void) contfrac_A(z -> beta, p, q, r, dp, dq);
+  free(p); free(q); free(r);
+  return;
+}
+
+static INTERNAL_PRECISION *contfrac_A(INTERNAL_PRECISION *beta,
+				      INTERNAL_PRECISION *p,
+				      INTERNAL_PRECISION *q,
+				      INTERNAL_PRECISION *r, int dp, int dq) {
+  INTERNAL_PRECISION quot, *rb;
+  int j;
+
+  /* p(x) = x beta q(x) + r(x); dp = dq or dp = dq + 1 */
+
+  quot = dp == dq ? ZERO : p[dp] / q[dq];
+  r[0] = p[0];
+  for (j = 1; j <= dp; j++) r[j] = p[j] - quot * q[j-1];
+#ifdef DEBUG
+  printf("%s: Continued Fraction form: deg p = %2d, deg q = %2d, beta = %g\n",
+	 __FUNCTION__, dp, dq, (float) quot);
+  for (j = 0; j <= dq + 1; j++)
+    printf("\tp[%2d] = %14.6g\tq[%2d] = %14.6g\tr[%2d] = %14.6g\n",
+	   j, (float) (j > dp ? ZERO : p[j]),
+	   j, (float) (j == 0 ? ZERO : q[j-1]),
+	   j, (float) (j == dp ? ZERO : r[j]));
+#endif /* DEBUG */
+  *(rb = contfrac_B(beta, q, r, p, dq, dq)) = quot;
+  return rb + 1;
+}
+
+static INTERNAL_PRECISION *contfrac_B(INTERNAL_PRECISION *beta,
+				      INTERNAL_PRECISION *p,
+				      INTERNAL_PRECISION *q,
+				      INTERNAL_PRECISION *r, int dp, int dq) {
+  INTERNAL_PRECISION quot, *rb;
+  int j;
+
+  /* p(x) = beta q(x) + r(x); dp = dq or dp = dq - 1 */
+
+  quot = dp == dq ? p[dp] / q[dq] : ZERO;
+  for (j = 0; j < dq; j++) r[j] = p[j] - quot * q[j];
+#ifdef DEBUG
+  printf("%s: Continued Fraction form: deg p = %2d, deg q = %2d, beta = %g\n",
+	 __FUNCTION__, dp, dq, (float) quot);
+  for (j = 0; j <= dq; j++)
+    printf("\tp[%2d] = %14.6g\tq[%2d] = %14.6g\tr[%2d] = %14.6g\n",
+	   j, (float) (j > dp ? ZERO : p[j]),
+	   j, (float) q[j],
+	   j, (float) (j == dq ? ZERO : r[j]));
+#endif /* DEBUG */
+  *(rb = dq > 0 ? contfrac_A(beta, q, r, p, dq, dq-1) : beta) = quot;
+  return rb + 1;
+}
+
+/* The global variable U is used to hold the argument u throughout the AGM
+ * recursion. The global variables F and K are set in the innermost
+ * instantiation of the recursive function AGM to the values of the elliptic
+ * integrals F(u,k) and K(k) respectively. They must be made thread local to
+ * make this code thread-safe in a multithreaded environment. */
+
+static INTERNAL_PRECISION U, F, K;	/* THREAD LOCAL */
+
+/* Recursive implementation of Gauss' arithmetico-geometric mean, which is the
+ * kernel of the method used to compute the Jacobian elliptic functions
+ * sn(u,k), cn(u,k), and dn(u,k) with parameter k (where 0 < k < 1), as well
+ * as the elliptic integral F(s,k) satisfying F(sn(u,k)) = u and the complete
+ * elliptic integral K(k).
+ *
+ * The algorithm used is a recursive implementation of the Gauss (Landen)
+ * transformation.
+ *
+ * The function returns the value of sn(u,k'), where k' is the dual parameter,
+ * and also sets the values of the global variables F and K.  The latter is
+ * used to determine the sign of cn(u,k').
+ *
+ * The algorithm is deemed to have converged when b ceases to increase. This
+ * works whatever INTERNAL_PRECISION is specified. */
+
+static INTERNAL_PRECISION AGM(INTERNAL_PRECISION a,
+			      INTERNAL_PRECISION b,
+			      INTERNAL_PRECISION s) {
+  static INTERNAL_PRECISION pb = -ONE;
+  INTERNAL_PRECISION c, d, xi;
+
+  if (b <= pb) {
+    pb = -ONE;
+    F = asin(s) / a;		/* Here, a is the AGM */
+    K = M_PI / (TWO * a);
+    return sin(U * a);
+  }
+  pb = b;
+  c = a - b;
+  d = a + b;
+  xi = AGM(HALF*d, sqrt(a*b), ONE + c*c == ONE ?
+	   HALF*s*d/a : (a - sqrt(a*a - s*s*c*d))/(c*s));
+  return 2*a*xi / (d + c*xi*xi);
+}
+
+/* Computes sn(u,k), cn(u,k), dn(u,k), F(u,k), and K(k). It is essentially a
+ * wrapper for the routine AGM. The sign of cn(u,k) is defined to be -1 if
+ * K(k) < u < 3*K(k) and +1 otherwise, and thus sign is computed by some quite
+ * unnecessarily obfuscated bit manipulations. */
+
+static void sncndnFK(INTERNAL_PRECISION u, INTERNAL_PRECISION k,
+		     INTERNAL_PRECISION* sn, INTERNAL_PRECISION* cn,
+		     INTERNAL_PRECISION* dn, INTERNAL_PRECISION* elF,
+		     INTERNAL_PRECISION* elK) {
+  int sgn;
+  U = u;
+  *sn = AGM(ONE, sqrt(ONE - k*k), u);
+  sgn = ((int) (fabs(u) / K)) % 4; /* sgn = 0, 1, 2, 3 */
+  sgn ^= sgn >> 1;    /* (sgn & 1) = 0, 1, 1, 0 */
+  sgn = 1 - ((sgn & 1) << 1);	/* sgn = 1, -1, -1, 1 */
+  *cn = ((INTERNAL_PRECISION) sgn) * sqrt(ONE - *sn * *sn);
+  *dn = sqrt(ONE - k*k* *sn * *sn);
+  *elF = F;
+  *elK = K;
+}
+
+/* Compute the coefficients for the optimal rational approximation R(x) to
+ * sgn(x) of degree n over the interval epsilon < |x| < 1 using Zolotarev's
+ * formula. 
+ *
+ * Set type = 0 for the Zolotarev approximation, which is zero at x = 0, and
+ * type = 1 for the approximation which is infinite at x = 0. */
+
+zolotarev_data* bfm_zolotarev(PRECISION epsilon, int n, int type) {
+  INTERNAL_PRECISION A, c, cp, kp, ksq, sn, cn, dn, Kp, Kj, z, z0, t, M, F,
+    l, invlambda, xi, xisq, *tv, s, opl;
+  int m, czero, ts;
+  zolotarev_data *zd;
+  izd *d = (izd*) malloc(sizeof(izd));
+
+  d -> type = type;
+  d -> epsilon = (INTERNAL_PRECISION) epsilon;
+  d -> n = n;
+  d -> dd = n / 2;
+  d -> dn = d -> dd - 1 + n % 2; /* n even: dn = dd - 1, n odd: dn = dd */
+  d -> deg_denom = 2 * d -> dd;
+  d -> deg_num = 2 * d -> dn + 1;
+
+  d -> a = (INTERNAL_PRECISION*) malloc((1 + d -> dn) *
+					sizeof(INTERNAL_PRECISION));
+  d -> ap = (INTERNAL_PRECISION*) malloc(d -> dd *
+					 sizeof(INTERNAL_PRECISION));
+  ksq = d -> epsilon * d -> epsilon;
+  kp = sqrt(ONE - ksq);
+  sncndnFK(ZERO, kp, &sn, &cn, &dn, &F, &Kp); /* compute Kp = K(kp) */
+  z0 = TWO * Kp / (INTERNAL_PRECISION) n;
+  M = ONE;
+  A = ONE / d -> epsilon;
+
+  sncndnFK(HALF * z0, kp, &sn, &cn, &dn, &F, &Kj); /* compute xi */
+  xi = ONE / dn;
+  xisq = xi * xi;
+  invlambda = xi;
+
+  for (m = 0; m < d -> dd; m++) {
+    czero = 2 * (m + 1) == n; /* n even and m = dd -1 */
+
+    z = z0 * ((INTERNAL_PRECISION) m + ONE);
+    sncndnFK(z, kp, &sn, &cn, &dn, &F, &Kj);
+    t = cn / sn;
+    c = - t*t;
+    if (!czero) (d -> a)[d -> dn - 1 - m] = ksq / c;
+
+    z = z0 * ((INTERNAL_PRECISION) m + HALF);
+    sncndnFK(z, kp, &sn, &cn, &dn, &F, &Kj);
+    t = cn / sn;
+    cp = - t*t;
+    (d -> ap)[d -> dd - 1 - m] = ksq / cp;
+
+    M *= (ONE - c) / (ONE - cp);
+    A *= (czero ? -ksq : c) * (ONE - cp) / (cp * (ONE - c));
+    invlambda *= (ONE - c*xisq) / (ONE - cp*xisq);
+  }
+  invlambda /= M;
+  d -> A = TWO / (ONE + invlambda) * A;
+  d -> Delta = (invlambda - ONE) / (invlambda + ONE);
+
+  d -> gamma = (INTERNAL_PRECISION*) malloc((1 + d -> n) *
+					    sizeof(INTERNAL_PRECISION));
+  l = ONE / invlambda;
+  opl = ONE + l;
+  sncndnFK(sqrt( d -> type == 1
+		   ? (THREE + l) / (FOUR * opl)
+		   : (ONE + THREE*l) / (opl*opl*opl)
+	       ), sqrt(ONE - l*l), &sn, &cn, &dn, &F, &Kj);
+  s = M * F;
+  for (m = 0; m < d -> n; m++) {
+    sncndnFK(s + TWO*Kp*m/n, kp, &sn, &cn, &dn, &F, &Kj);
+    d -> gamma[m] = d -> epsilon / dn;
+  }
+
+  /* If R(x) is a Zolotarev rational approximation of degree (n,m) with maximum
+   * error Delta, then (1 - Delta^2) / R(x) is also an optimal Chebyshev
+   * approximation of degree (m,n) */
+
+  if (d -> type == 1) {
+    d -> A = (ONE - d -> Delta * d -> Delta) / d -> A;
+    tv = d -> a; d -> a = d -> ap; d -> ap = tv;
+    ts = d -> dn; d -> dn = d -> dd; d -> dd = ts;
+    ts = d -> deg_num; d -> deg_num = d -> deg_denom; d -> deg_denom = ts;
+  }
+
+  construct_partfrac(d);
+  construct_contfrac(d);
+
+  /* Converting everything to PRECISION for external use only */
+
+  zd = (zolotarev_data*) malloc(sizeof(zolotarev_data));
+  zd -> A = (PRECISION) d -> A;
+  zd -> Delta = (PRECISION) d -> Delta;
+  zd -> epsilon = (PRECISION) d -> epsilon;
+  zd -> n = d -> n;
+  zd -> type = d -> type;
+  zd -> dn = d -> dn;
+  zd -> dd = d -> dd;
+  zd -> da = d -> da;
+  zd -> db = d -> db;
+  zd -> deg_num = d -> deg_num;
+  zd -> deg_denom = d -> deg_denom;
+
+  zd -> a = (PRECISION*) malloc(zd -> dn * sizeof(PRECISION));
+  for (m = 0; m < zd -> dn; m++) zd -> a[m] = (PRECISION) d -> a[m];
+  free(d -> a);
+
+  zd -> ap = (PRECISION*) malloc(zd -> dd * sizeof(PRECISION));
+  for (m = 0; m < zd -> dd; m++) zd -> ap[m] = (PRECISION) d -> ap[m];
+  free(d -> ap);
+
+  zd -> alpha = (PRECISION*) malloc(zd -> da * sizeof(PRECISION));
+  for (m = 0; m < zd -> da; m++) zd -> alpha[m] = (PRECISION) d -> alpha[m];
+  free(d -> alpha);
+
+  zd -> beta = (PRECISION*) malloc(zd -> db * sizeof(PRECISION));
+  for (m = 0; m < zd -> db; m++) zd -> beta[m] = (PRECISION) d -> beta[m];
+  free(d -> beta);
+
+  zd -> gamma = (PRECISION*) malloc(zd -> n * sizeof(PRECISION));
+  for (m = 0; m < zd -> n; m++) zd -> gamma[m] = (PRECISION) d -> gamma[m];
+  free(d -> gamma);
+
+  free(d);
+  return zd;
+}
+
+zolotarev_data* bfm_higham(PRECISION epsilon, int n) {
+  INTERNAL_PRECISION A, M, c, cp, z, z0, t, epssq;
+  int m, czero;
+  zolotarev_data *zd;
+  izd *d = (izd*) malloc(sizeof(izd));
+
+  d -> type = 0;
+  d -> epsilon = (INTERNAL_PRECISION) epsilon;
+  d -> n = n;
+  d -> dd = n / 2;
+  d -> dn = d -> dd - 1 + n % 2; /* n even: dn = dd - 1, n odd: dn = dd */
+  d -> deg_denom = 2 * d -> dd;
+  d -> deg_num = 2 * d -> dn + 1;
+
+  d -> a = (INTERNAL_PRECISION*) malloc((1 + d -> dn) *
+					sizeof(INTERNAL_PRECISION));
+  d -> ap = (INTERNAL_PRECISION*) malloc(d -> dd *
+					 sizeof(INTERNAL_PRECISION));
+  A = (INTERNAL_PRECISION) n;
+  z0 = M_PI / A;
+  A = n % 2 == 0 ? A : ONE / A;
+  M = d -> epsilon * A;
+  epssq = d -> epsilon * d -> epsilon;
+
+  for (m = 0; m < d -> dd; m++) {
+    czero = 2 * (m + 1) == n; /* n even and m = dd - 1*/
+
+    if (!czero) {
+      z = z0 * ((INTERNAL_PRECISION) m + ONE);
+      t = tan(z);
+      c = - t*t;
+      (d -> a)[d -> dn - 1 - m] = c;
+      M *= epssq - c;
+    }
+
+    z = z0 * ((INTERNAL_PRECISION) m + HALF);
+    t = tan(z);
+    cp = - t*t;
+    (d -> ap)[d -> dd - 1 - m] = cp;
+    M /= epssq - cp;
+  }
+
+  d -> gamma = (INTERNAL_PRECISION*) malloc((1 + d -> n) *
+					    sizeof(INTERNAL_PRECISION));
+  for (m = 0; m < d -> n; m++) d -> gamma[m] = ONE;
+
+  d -> A = A;
+  d -> Delta = ONE - M;
+
+  construct_partfrac(d);
+  construct_contfrac(d);
+
+  /* Converting everything to PRECISION for external use only */
+
+  zd = (zolotarev_data*) malloc(sizeof(zolotarev_data));
+  zd -> A = (PRECISION) d -> A;
+  zd -> Delta = (PRECISION) d -> Delta;
+  zd -> epsilon = (PRECISION) d -> epsilon;
+  zd -> n = d -> n;
+  zd -> type = d -> type;
+  zd -> dn = d -> dn;
+  zd -> dd = d -> dd;
+  zd -> da = d -> da;
+  zd -> db = d -> db;
+  zd -> deg_num = d -> deg_num;
+  zd -> deg_denom = d -> deg_denom;
+
+  zd -> a = (PRECISION*) malloc(zd -> dn * sizeof(PRECISION));
+  for (m = 0; m < zd -> dn; m++) zd -> a[m] = (PRECISION) d -> a[m];
+  free(d -> a);
+
+  zd -> ap = (PRECISION*) malloc(zd -> dd * sizeof(PRECISION));
+  for (m = 0; m < zd -> dd; m++) zd -> ap[m] = (PRECISION) d -> ap[m];
+  free(d -> ap);
+
+  zd -> alpha = (PRECISION*) malloc(zd -> da * sizeof(PRECISION));
+  for (m = 0; m < zd -> da; m++) zd -> alpha[m] = (PRECISION) d -> alpha[m];
+  free(d -> alpha);
+
+  zd -> beta = (PRECISION*) malloc(zd -> db * sizeof(PRECISION));
+  for (m = 0; m < zd -> db; m++) zd -> beta[m] = (PRECISION) d -> beta[m];
+  free(d -> beta);
+
+  zd -> gamma = (PRECISION*) malloc(zd -> n * sizeof(PRECISION));
+  for (m = 0; m < zd -> n; m++) zd -> gamma[m] = (PRECISION) d -> gamma[m];
+  free(d -> gamma);
+
+  free(d);
+  return zd;
+}
+
+#ifdef TEST
+
+#undef ZERO
+#define ZERO ((PRECISION) 0)
+#undef ONE
+#define ONE ((PRECISION) 1)
+#undef TWO
+#define TWO ((PRECISION) 2)
+
+/* Evaluate the rational approximation R(x) using the factored form */
+
+static PRECISION zolotarev_eval(PRECISION x, zolotarev_data* rdata) {
+  int m;
+  PRECISION R;
+
+  if (rdata -> type == 0) {
+    R = rdata -> A * x;
+    for (m = 0; m < rdata -> deg_denom/2; m++)
+      R *= (2*(m+1) > rdata -> deg_num ? ONE : x*x - rdata -> a[m]) /
+	(x*x - rdata -> ap[m]);
+  } else {
+    R = rdata -> A / x;
+    for (m = 0; m < rdata -> deg_num/2; m++)
+      R *= (x*x - rdata -> a[m]) /
+	(2*(m+1) > rdata -> deg_denom ? ONE : x*x - rdata -> ap[m]);
+  }
+  return R;
+}
+
+/* Evaluate the rational approximation R(x) using the partial fraction form */
+
+static PRECISION zolotarev_partfrac_eval(PRECISION x, zolotarev_data* rdata) {
+  int m;
+  PRECISION R = rdata -> alpha[rdata -> da - 1];
+  for (m = 0; m < rdata -> dd; m++)
+    R += rdata -> alpha[m] / (x * x - rdata -> ap[m]);
+  if (rdata -> type == 1) R += rdata -> alpha[rdata -> dd] / (x * x);
+  return R * x;
+}    
+
+/* Evaluate the rational approximation R(x) using continued fraction form. 
+ *
+ * If x = 0 and type = 1 then the result should be INF, whereas if x = 0 and
+ * type = 0 then the result should be 0, but division by zero will occur at
+ * intermediate stages of the evaluation. For IEEE implementations with
+ * non-signalling overflow this will work correctly since 1/(1/0) = 1/INF = 0,
+ * but with signalling overflow you will get an error message. */
+
+static PRECISION zolotarev_contfrac_eval(PRECISION x, zolotarev_data* rdata) {
+  int m;
+  PRECISION R = rdata -> beta[0] * x;
+  for (m = 1; m < rdata -> db; m++) R = rdata -> beta[m] * x + ONE / R;
+  return R;
+}    
+
+/* Evaluate the rational approximation R(x) using Cayley form */
+
+static PRECISION zolotarev_cayley_eval(PRECISION x, zolotarev_data* rdata) {
+  int m;
+  PRECISION T;
+
+  T = rdata -> type == 0 ? ONE : -ONE;
+  for (m = 0; m < rdata -> n; m++)
+    T *= (rdata -> gamma[m] - x) / (rdata -> gamma[m] + x);
+  return (ONE - T) / (ONE + T);
+}
+
+/* Test program. Apart from printing out the parameters for R(x) it produces
+ * the following data files for plotting (unless NPLOT is defined):
+ *
+ * zolotarev-fn is a plot of R(x) for |x| < 1.2. This should look like sgn(x).
+ *
+ * zolotarev-err is a plot of the error |R(x) - sgn(x)| scaled by 1/Delta. This
+ * should oscillate deg_num + deg_denom + 2 times between +1 and -1 over the
+ * domain epsilon <= |x| <= 1.
+ *
+ * If ALLPLOTS is defined then zolotarev-partfrac (zolotarev-contfrac) is a
+ * plot of the difference between the values of R(x) computed using the
+ * factored form and the partial fraction (continued fraction) form, scaled by
+ * 1/Delta. It should be zero everywhere. */
+
+int main(int argc, char** argv) {
+  
+  int m, n, plotpts = 5000, type = 0;
+  float eps, x, ypferr, ycferr, ycaylerr, maxypferr, maxycferr, maxycaylerr;
+  zolotarev_data *rdata;
+  PRECISION y;
+  FILE *plot_function, *plot_error, 
+    *plot_partfrac, *plot_contfrac, *plot_cayley;
+
+  if (argc < 3 || argc > 4) {
+    fprintf(stderr, "Usage: %s epsilon n [type]\n", *argv);
+    exit(EXIT_FAILURE);
+  }
+  sscanf(argv[1], "%g", &eps);	/* First argument is epsilon */
+  sscanf(argv[2], "%d", &n);	/* Second argument is n */
+  if (argc == 4) sscanf(argv[3], "%d", &type); /* Third argument is type */
+
+  if (type < 0 || type > 2) {
+    fprintf(stderr, "%s: type must be 0 (Zolotarev R(0) = 0),\n"
+	    "\t\t1 (Zolotarev R(0) = Inf, or 2 (Higham)\n", *argv);
+    exit(EXIT_FAILURE);
+  }
+
+  rdata = type == 2 
+    ? higham((PRECISION) eps, n) 
+    : zolotarev((PRECISION) eps, n, type);
+
+  printf("Zolotarev Test: R(epsilon = %g, n = %d, type = %d)\n\t" 
+	 STRINGIFY(VERSION) "\n\t" STRINGIFY(HVERSION)
+	 "\n\tINTERNAL_PRECISION = " STRINGIFY(INTERNAL_PRECISION)
+	 "\tPRECISION = " STRINGIFY(PRECISION)
+	 "\n\n\tRational approximation of degree (%d,%d), %s at x = 0\n"
+	 "\tDelta = %g (maximum error)\n\n"
+	 "\tA = %g (overall factor)\n",
+	 (float) rdata -> epsilon, rdata -> n, type,
+	 rdata -> deg_num, rdata -> deg_denom,
+	 rdata -> type == 1 ? "infinite" : "zero",
+	 (float) rdata -> Delta, (float) rdata -> A);
+  for (m = 0; m < MIN(rdata -> dd, rdata -> dn); m++)
+    printf("\ta[%2d] = %14.8g\t\ta'[%2d] = %14.8g\n",
+	   m + 1, (float) rdata -> a[m], m + 1, (float) rdata -> ap[m]);
+  if (rdata -> dd > rdata -> dn)
+    printf("\t\t\t\t\ta'[%2d] = %14.8g\n",
+	   rdata -> dn + 1, (float) rdata -> ap[rdata -> dn]);
+  if (rdata -> dd < rdata -> dn)
+    printf("\ta[%2d] = %14.8g\n",
+	   rdata -> dd + 1, (float) rdata -> a[rdata -> dd]);
+
+  printf("\n\tPartial fraction coefficients\n");
+  printf("\talpha[ 0] = %14.8g\n",
+	 (float) rdata -> alpha[rdata -> da - 1]);
+  for (m = 0; m < rdata -> dd; m++)
+    printf("\talpha[%2d] = %14.8g\ta'[%2d] = %14.8g\n",
+	   m + 1, (float) rdata -> alpha[m], m + 1, (float) rdata -> ap[m]);
+  if (rdata -> type == 1)
+    printf("\talpha[%2d] = %14.8g\ta'[%2d] = %14.8g\n",
+	   rdata -> dd + 1, (float) rdata -> alpha[rdata -> dd],
+	   rdata -> dd + 1, (float) ZERO);
+
+  printf("\n\tContinued fraction coefficients\n");
+  for (m = 0; m < rdata -> db; m++)
+    printf("\tbeta[%2d] = %14.8g\n", m, (float) rdata -> beta[m]);
+
+  printf("\n\tCayley transform coefficients\n");
+  for (m = 0; m < rdata -> n; m++)
+    printf("\tgamma[%2d] = %14.8g\n", m, (float) rdata -> gamma[m]);
+
+#ifndef NPLOT
+  plot_function = fopen("zolotarev-fn.dat", "w");
+  plot_error = fopen("zolotarev-err.dat", "w");
+#ifdef ALLPLOTS
+  plot_partfrac = fopen("zolotarev-partfrac.dat", "w");
+  plot_contfrac = fopen("zolotarev-contfrac.dat", "w");
+  plot_cayley = fopen("zolotarev-cayley.dat", "w");
+#endif /* ALLPLOTS */
+  for (m = 0, maxypferr = maxycferr = maxycaylerr = 0.0; m <= plotpts; m++) {
+    x = 2.4 * (float) m / plotpts - 1.2;
+    if (rdata -> type == 0 || fabs(x) * (float) plotpts > 1.0) {
+      /* skip x = 0 for type 1, as R(0) is singular */
+      y = zolotarev_eval((PRECISION) x, rdata);
+      fprintf(plot_function, "%g %g\n", x, (float) y);
+      fprintf(plot_error, "%g %g\n",
+	      x, (float)((y - ((x > 0.0 ? ONE : -ONE))) / rdata -> Delta));
+      ypferr = (float)((zolotarev_partfrac_eval((PRECISION) x, rdata) - y)
+		       / rdata -> Delta);
+      ycferr = (float)((zolotarev_contfrac_eval((PRECISION) x, rdata) - y)
+		       / rdata -> Delta);
+      ycaylerr = (float)((zolotarev_cayley_eval((PRECISION) x, rdata) - y)
+		       / rdata -> Delta);
+      if (fabs(x) < 1.0 && fabs(x) > rdata -> epsilon) {
+	maxypferr = MAX(maxypferr, fabs(ypferr));
+	maxycferr = MAX(maxycferr, fabs(ycferr));
+	maxycaylerr = MAX(maxycaylerr, fabs(ycaylerr));
+      }
+#ifdef ALLPLOTS
+      fprintf(plot_partfrac, "%g %g\n", x, ypferr);
+      fprintf(plot_contfrac, "%g %g\n", x, ycferr);
+      fprintf(plot_cayley, "%g %g\n", x, ycaylerr);
+#endif /* ALLPLOTS */
+    }
+  }
+#ifdef ALLPLOTS
+  fclose(plot_cayley);
+  fclose(plot_contfrac);
+  fclose(plot_partfrac);
+#endif /* ALLPLOTS */
+  fclose(plot_error);
+  fclose(plot_function);
+
+  printf("\n\tMaximum PF error = %g (relative to Delta)\n", maxypferr);
+  printf("\tMaximum CF error = %g (relative to Delta)\n", maxycferr);
+  printf("\tMaximum Cayley error = %g (relative to Delta)\n", maxycaylerr);
+#endif /* NPLOT */
+
+  free(rdata -> a);
+  free(rdata -> ap);
+  free(rdata -> alpha);
+  free(rdata -> beta);
+  free(rdata -> gamma);
+  free(rdata);
+
+  return EXIT_SUCCESS;
+}
+#endif /* TEST */
diff --git a/lib/algorithms/approx/zolotarev.h b/lib/algorithms/approx/zolotarev.h
new file mode 100755
index 00000000..3f0dc58e
--- /dev/null
+++ b/lib/algorithms/approx/zolotarev.h
@@ -0,0 +1,85 @@
+/* -*- Mode: C; comment-column: 22; fill-column: 79; -*- */
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#define HVERSION Header Time-stamp: <14-OCT-2004 09:26:51.00 adk@MISSCONTRARY>
+
+
+#ifndef ZOLOTAREV_INTERNAL
+#ifndef PRECISION
+#define PRECISION double
+#endif
+#define ZPRECISION PRECISION
+#define ZOLOTAREV_DATA zolotarev_data
+#endif
+
+/* This struct contains the coefficients which parameterise an optimal rational
+ * approximation to the signum function.
+ *
+ * The parameterisations used are:
+ *
+ * Factored form for type 0 (R0(0) = 0)
+ *
+ * R0(x) = A * x * prod(x^2 - a[j], j = 0 .. dn-1) / prod(x^2 - ap[j], j = 0
+ * .. dd-1),
+ *
+ * where deg_num = 2*dn + 1 and deg_denom = 2*dd.
+ *
+ * Factored form for type 1 (R1(0) = infinity)
+ *
+ * R1(x) = (A / x) * prod(x^2 - a[j], j = 0 .. dn-1) / prod(x^2 - ap[j], j = 0
+ * .. dd-1),
+ *
+ * where deg_num = 2*dn and deg_denom = 2*dd + 1. 
+ *
+ * Partial fraction form
+ *
+ * R(x) = alpha[da] * x + sum(alpha[j] * x / (x^2 - ap[j]), j = 0 .. da-1)
+ *
+ * where da = dd for type 0 and da = dd + 1 with ap[dd] = 0 for type 1.
+ *
+ * Continued fraction form 
+ *
+ * R(x) = beta[db-1] * x + 1 / (beta[db-2] * x + 1 / (beta[db-3] * x + ...))
+ *
+ * with the final coefficient being beta[0], with d' = 2 * dd + 1 for type 0
+ * and db = 2 * dd + 2 for type 1.
+ *
+ * Cayley form (Chiu's domain wall formulation)
+ *
+ * R(x) = (1 - T(x)) / (1 + T(x))
+ *
+ * where T(x) = prod((x - gamma[j]) / (x + gamma[j]), j = 0 .. n-1)
+ */
+
+typedef struct {
+  ZPRECISION *a,      /* zeros of numerator, a[0 .. dn-1] */
+    *ap,	      /* poles (zeros of denominator), ap[0 .. dd-1] */
+    A,		      /* overall factor */
+    *alpha,	      /* coefficients of partial fraction, alpha[0 .. da-1] */
+    *beta,	      /* coefficients of continued fraction, beta[0 .. db-1] */
+    *gamma,	      /* zeros of numerator of T in Cayley form */
+    Delta,	      /* maximum error, |R(x) - sgn(x)| <= Delta */
+    epsilon;	      /* minimum x value, epsilon < |x| < 1 */
+  int n,	      /* approximation degree */
+    type,	      /* 0: R(0) = 0, 1: R(0) = infinity */
+    dn, dd, da, db,   /* number of elements of a, ap, alpha, and beta */
+    deg_num,	      /* degree of numerator = deg_denom +/- 1 */
+    deg_denom;	      /* degree of denominator */
+} ZOLOTAREV_DATA;
+
+#ifndef ZOLOTAREV_INTERNAL
+
+/* zolotarev(epsilon, n, type) returns a pointer to an initialised
+ * zolotarev_data structure. The arguments must satisfy the constraints that
+ * epsilon > 0, n > 0, and type = 0 or 1. */
+
+ZOLOTAREV_DATA* bfm_higham(PRECISION epsilon, int n) ;
+ZOLOTAREV_DATA* bfm_zolotarev(PRECISION epsilon, int n, int type);
+#endif
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index 14c48cf2..c4d37a87 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -290,10 +290,7 @@ void WilsonMatrix::MpcDagMpc(const LatticeFermion &in, LatticeFermion &out)
 {
   return;
 }
-void WilsonMatrix::MDagM    (const LatticeFermion &in, LatticeFermion &out)
-{
-  return;
-}
+
 
 
 }}
diff --git a/lib/qcd/Grid_qcd_wilson_dop.h b/lib/qcd/Grid_qcd_wilson_dop.h
index e19cbabe..2f5105c8 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.h
+++ b/lib/qcd/Grid_qcd_wilson_dop.h
@@ -9,7 +9,7 @@ namespace Grid {
 
   namespace QCD {
 
-    class WilsonMatrix : public LinearOperatorBase<LatticeFermion>
+    class WilsonMatrix : public SparseMatrixBase<LatticeFermion>
     {
       //NB r=1;
     public:
@@ -40,6 +40,11 @@ namespace Grid {
       virtual void Mdag (const LatticeFermion &in, LatticeFermion &out);
       virtual void MdagM(const LatticeFermion &in, LatticeFermion &out);
 
+      // half checkerboard operaions
+      void Mpc      (const LatticeFermion &in, LatticeFermion &out);
+      void MpcDag   (const LatticeFermion &in, LatticeFermion &out);
+      void MpcDagMpc(const LatticeFermion &in, LatticeFermion &out);
+
       // non-hermitian hopping term; half cb or both
       void Dhop(const LatticeFermion &in, LatticeFermion &out);
 
@@ -48,17 +53,10 @@ namespace Grid {
 
       typedef iScalar<iMatrix<vComplex, Nc> > matrix;
 
-      // half checkerboard operaions
-      void MpcDag   (const LatticeFermion &in, LatticeFermion &out);
-      void Mpc      (const LatticeFermion &in, LatticeFermion &out);
-      void MpcDagMpc(const LatticeFermion &in, LatticeFermion &out);
-
-      // full checkerboard hermitian
-      void MDagM    (const LatticeFermion &in, LatticeFermion &out);
-
       
     };
 
+
   }
 }
 #endif

From 8e99e4671ff538b74a38e1fbc4d07a1dbcf6058a Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sun, 17 May 2015 00:19:03 +0100
Subject: [PATCH 191/429] Working towards solvers

---
 lib/algorithms/LinearOperator.h              | 55 ++++++------
 lib/algorithms/approx/PolynomialApprox.h     |  3 +
 lib/algorithms/iterative/ConjugateGradient.h | 88 ++++++++++++++++++++
 3 files changed, 119 insertions(+), 27 deletions(-)
 create mode 100644 lib/algorithms/iterative/ConjugateGradient.h

diff --git a/lib/algorithms/LinearOperator.h b/lib/algorithms/LinearOperator.h
index c449ee8e..12259236 100644
--- a/lib/algorithms/LinearOperator.h
+++ b/lib/algorithms/LinearOperator.h
@@ -5,6 +5,22 @@
 
 namespace Grid {
 
+  /////////////////////////////////////////////////////////////////////////////////////////////
+  // Interface defining what I expect of a general sparse matrix, such as a Fermion action
+  /////////////////////////////////////////////////////////////////////////////////////////////
+    template<class Field> class SparseMatrixBase {
+    public:
+      // Full checkerboar operations
+      virtual void M    (const Field &in, Field &out);
+      virtual void Mdag (const Field &in, Field &out);
+      virtual RealD MdagM(const Field &in, Field &out);
+      
+      // half checkerboard operaions
+      virtual void Mpc      (const Field &in, Field &out);
+      virtual void MpcDag   (const Field &in, Field &out);
+      virtual RealD MpcDagMpc(const Field &in, Field &out);
+    };
+
   /////////////////////////////////////////////////////////////////////////////////////////////
   // LinearOperators Take a something and return a something.
   /////////////////////////////////////////////////////////////////////////////////////////////
@@ -27,25 +43,13 @@ namespace Grid {
   /////////////////////////////////////////////////////////////////////////////////////////////
     template<class Field> class HermitianOperatorBase : public LinearOperatorBase<Field> {
     public:
+      virtual RealD OpAndNorm(const Field &in, Field &out);
       void AdjOp(const Field &in, Field &out) {
-	return Op(in,out);
+	Op(in,out);
+      };
+      void Op(const Field &in, Field &out) {
+	OpAndNorm(in,out);
       };
-    };
-
-  /////////////////////////////////////////////////////////////////////////////////////////////
-  // Interface defining what I expect of a general sparse matrix
-  /////////////////////////////////////////////////////////////////////////////////////////////
-    template<class Field> class SparseMatrixBase {
-    public:
-      // Full checkerboar operations
-      virtual void M    (const Field &in, Field &out);
-      virtual void Mdag (const Field &in, Field &out);
-      virtual void MdagM(const Field &in, Field &out);
-      
-      // half checkerboard operaions
-      virtual void Mpc      (const Field &in, Field &out);
-      virtual void MpcDag   (const Field &in, Field &out);
-      virtual void MpcDagMpc(const Field &in, Field &out);
     };
 
   /////////////////////////////////////////////////////////////////////////////////////////////
@@ -79,10 +83,10 @@ namespace Grid {
     public:
       NonHermitianRedBlackOperator(SparseMatrix &Mat): _Mat(Mat){};
       void Op     (const Field &in, Field &out){
-	this->Mpc(in,out);
+	_Mat.Mpc(in,out);
       }
       void AdjOp     (const Field &in, Field &out){ //
-	this->MpcDag(in,out);
+	_Mat.MpcDag(in,out);
       }
     };
 
@@ -94,12 +98,8 @@ namespace Grid {
       SparseMatrix &_Mat;
     public:
       HermitianOperator(SparseMatrix &Mat): _Mat(Mat) {};
-
-      void Op     (const Field &in, Field &out){
-	this->Mpc(in,out);
-      }
-      void AdjOp     (const Field &in, Field &out){ //
-	this->MpcDag(in,out);
+      RealD OpAndNorm(const Field &in, Field &out){
+	return _Mat.MdagM(in,out);
       }
     };
 
@@ -111,8 +111,8 @@ namespace Grid {
       SparseMatrix &_Mat;
     public:
       HermitianRedBlackOperator(SparseMatrix &Mat): _Mat(Mat) {};
-      void Op     (const Field &in, Field &out){
-	this->MpcDagMpc(in,out);
+      RealD OpAndNorm(const Field &in, Field &out){
+	return _Mat.MpcDagMpc(in,out);
       }
     };
 
@@ -141,6 +141,7 @@ namespace Grid {
     public:
       RealD   Tolerance;
       Integer MaxIterations;
+      IterativeProcess(RealD tol,Integer maxit) : Tolerance(tol),MaxIterations(maxit) {};
       virtual void operator() (LinearOperatorBase<Field> *Linop,const Field &in, Field &out) = 0;
     };
 
diff --git a/lib/algorithms/approx/PolynomialApprox.h b/lib/algorithms/approx/PolynomialApprox.h
index 5f59a078..5515b325 100644
--- a/lib/algorithms/approx/PolynomialApprox.h
+++ b/lib/algorithms/approx/PolynomialApprox.h
@@ -1,6 +1,9 @@
 #ifndef GRID_POLYNOMIAL_APPROX_H
 #define GRID_POLYNOMIAL_APPROX_H
 
+#include<Grid.h>
+#include<algorithms/LinearOperator.h>
+
 namespace Grid {
 
   ////////////////////////////////////////////////////////////////////////////////////////////
diff --git a/lib/algorithms/iterative/ConjugateGradient.h b/lib/algorithms/iterative/ConjugateGradient.h
new file mode 100644
index 00000000..ad305d2f
--- /dev/null
+++ b/lib/algorithms/iterative/ConjugateGradient.h
@@ -0,0 +1,88 @@
+#ifndef GRID_CONJUGATE_GRADIENT_H
+#define GRID_CONJUGATE_GRADIENT_H
+
+#include<Grid.h>
+#include<algorithms/LinearOperator.h>
+
+namespace Grid {
+
+  template<class Field> class ConjugateGradient : public IterativeProcess<Field> {
+  public:
+
+    ConjugateGradient(RealD tol,Integer maxit): IterativeProces(tol,maxit) {};
+
+    void operator() (HermitianOperatorBase<Field> *Linop,const Field &src, Field &psi){
+
+      RealD residual = Tolerance;
+      RealD cp,c,a,d,b,ssq,qq,b_pred;
+      
+      Field   p(src);
+      Field mmp(src);
+      Field   r(src);
+      
+      //Initial residual computation & set up
+      RealD guess = norm2(psi);
+      
+      Linop.OpAndNorm(psi,mmp,d,b);
+      
+      r= src-mmp;
+      p= r;
+      
+      a  =norm2(p);
+      cp =a;
+      ssq=norm2(src);
+      
+      std::cout <<setprecision(4)<< "ConjugateGradient: guess "<<guess<<std::endl;
+      std::cout <<setprecision(4)<< "ConjugateGradient:   src "<<ssq  <<std::endl;
+      std::cout <<setprecision(4)<< "ConjugateGradient:    mp "<<d    <<std::endl;
+      std::cout <<setprecision(4)<< "ConjugateGradient:   mmp "<<b    <<std::endl;
+      std::cout <<setprecision(4)<< "ConjugateGradient:     r "<<cp   <<std::endl;
+      std::cout <<setprecision(4)<< "ConjugateGradient:     p "<<a    <<std::endl;
+      
+      RealD rsq =  residual* residual*ssq;
+      
+      //Check if guess is really REALLY good :)
+      if ( cp <= rsq ) {
+	return 0;
+      }
+      
+      std::cout << setprecision(4)<< "ConjugateGradient: k=0 residual "<<cp<<" rsq"<<rsq<<std::endl;
+      
+      int k;
+      for (k=1;k<=max_iter;k++){
+	
+	c=cp;
+	
+	Linop.OpAndNorm(p,mmp,d,qq);
+	  
+	a      = c/d;
+	b_pred = a*(a*qq-d)/c;
+	
+	cp = axpy_norm(r,mmp,r,-a);
+	b = cp/c;
+	
+	// Fuse these loops ; should be really easy
+	// Update psi; New (conjugate/M-orthogonal) search direction
+	psi= a*p+psi;
+	p  = p*b+r;
+	  
+	// Stopping condition
+	if ( cp <= rsq ) { 
+	  
+	  Linop.Op(p,mmp);
+	  p=mmp-src;
+	  
+	  RealD mmpnorm = sqrt(norm2(mmp));
+	  RealD psinorm = sqrt(norm2(psi));
+	  RealD srcnorm = sqrt(norm2(src));
+	  RealD resnorm = sqrt(norm2(p));
+	  RealD true_residual = resnorm/srcnorm;
+	  std::cout<<"ConjugateGradient: true   residual  is "<<true_residual<<" sol "<<psinorm<<" src "<<srcnorm<<std::endl;
+	  std::cout<<"ConjugateGradient: target residual was "<<residual<<std::endl;
+	  return k;
+	}
+      }
+    }
+  };
+}
+#endif

From d0e4673a3f0ebdeee032bd5c4298e2814046fb15 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 18 May 2015 07:47:05 +0100
Subject: [PATCH 192/429] Getting closer to having a wilson solver...
 introducing a first and untested cut at Conjugate gradient. Also copied in
 Remez, Zolotarev, Chebyshev from Mike Clark, Tony Kennedy and my BFM package
 respectively since we know we will need these. I wanted the structure of

algorithms/approx
algorithms/iterative

etc.. to start taking shape.
---
 configure                                     |  13 +-
 configure.ac                                  |   2 +-
 lib/Grid.h                                    |  23 +-
 lib/Grid_algorithms.h                         |  34 +
 lib/Grid_config.h.in                          |   3 +
 lib/Makefile.am                               |  40 +-
 lib/algorithms/LinearOperator.h               |  94 +--
 lib/algorithms/SparseMatrix.h                 |  72 ++
 .../{PolynomialApprox.h => Chebyshev.h}       |  14 +-
 lib/algorithms/approx/LICENSE                 |  21 +
 lib/algorithms/approx/README                  |  80 ++
 lib/algorithms/approx/Remez.cc                | 755 ++++++++++++++++++
 lib/algorithms/approx/Remez.h                 | 167 ++++
 .../approx/{zolotarev.c => Zolotarev.cc}      |   0
 lib/algorithms/approx/bigfloat.h              | 368 +++++++++
 lib/algorithms/iterative/ConjugateGradient.h  |  49 +-
 lib/algorithms/iterative/NormalEquations.h    |  34 +
 lib/algorithms/iterative/SchurRedBlack.h      | 106 +++
 lib/algorithms/iterative/TODO                 |  15 +
 lib/qcd/Grid_qcd_wilson_dop.cc                |  44 +-
 lib/qcd/Grid_qcd_wilson_dop.h                 |  19 +-
 tests/Grid_main.cc                            |   2 +-
 22 files changed, 1798 insertions(+), 157 deletions(-)
 create mode 100644 lib/Grid_algorithms.h
 create mode 100644 lib/algorithms/SparseMatrix.h
 rename lib/algorithms/approx/{PolynomialApprox.h => Chebyshev.h} (93%)
 create mode 100644 lib/algorithms/approx/LICENSE
 create mode 100644 lib/algorithms/approx/README
 create mode 100755 lib/algorithms/approx/Remez.cc
 create mode 100755 lib/algorithms/approx/Remez.h
 rename lib/algorithms/approx/{zolotarev.c => Zolotarev.cc} (100%)
 create mode 100755 lib/algorithms/approx/bigfloat.h
 create mode 100644 lib/algorithms/iterative/NormalEquations.h
 create mode 100644 lib/algorithms/iterative/SchurRedBlack.h
 create mode 100644 lib/algorithms/iterative/TODO

diff --git a/configure b/configure
index deec23c3..ee90f6bc 100755
--- a/configure
+++ b/configure
@@ -4244,7 +4244,18 @@ fi
 
 done
 
-#AC_CHECK_HEADERS(machine/endian.h)
+for ac_header in gmp.h
+do :
+  ac_fn_cxx_check_header_mongrel "$LINENO" "gmp.h" "ac_cv_header_gmp_h" "$ac_includes_default"
+if test "x$ac_cv_header_gmp_h" = xyes; then :
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_GMP_H 1
+_ACEOF
+
+fi
+
+done
+
 ac_fn_cxx_check_decl "$LINENO" "ntohll" "ac_cv_have_decl_ntohll" "#include <arpa/inet.h>
 "
 if test "x$ac_cv_have_decl_ntohll" = xyes; then :
diff --git a/configure.ac b/configure.ac
index 97206a01..14fd45bb 100644
--- a/configure.ac
+++ b/configure.ac
@@ -20,7 +20,7 @@ AC_CHECK_HEADERS(mm_malloc.h)
 AC_CHECK_HEADERS(malloc/malloc.h)
 AC_CHECK_HEADERS(malloc.h)
 AC_CHECK_HEADERS(endian.h)
-#AC_CHECK_HEADERS(machine/endian.h)
+AC_CHECK_HEADERS(gmp.h)
 AC_CHECK_DECLS([ntohll],[], [], [[#include <arpa/inet.h>]])
 AC_CHECK_DECLS([be64toh],[], [], [[#include <arpa/inet.h>]])
 
diff --git a/lib/Grid.h b/lib/Grid.h
index 2936dd27..edd94769 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -10,15 +10,17 @@
 #ifndef GRID_H
 #define GRID_H
 
-#include <stdio.h>
+#include <cassert>
 
 #include <complex>
 #include <vector>
-#include <valarray>
+
 #include <iostream>
-#include <cassert>
+#include <iomanip>
 #include <random>
 #include <functional>
+
+#include <stdio.h>
 #include <stdlib.h>
 #include <sys/time.h>
 #include <stdio.h>
@@ -45,16 +47,19 @@
 #endif
 
 #include <Grid_aligned_allocator.h>
+
 #include <Grid_simd.h>
 #include <Grid_threads.h>
 
-#include <Grid_cartesian.h>
-
-#include <Grid_math.h>
-#include <Grid_lattice.h>
+#include <Grid_cartesian.h> // subdir aggregate
+#include <Grid_math.h>      // subdir aggregate
+#include <Grid_lattice.h>   // subdir aggregate
 #include <Grid_comparison.h>
-#include <Grid_cshift.h>
-#include <Grid_stencil.h>
+#include <Grid_cshift.h>    // subdir aggregate
+#include <Grid_stencil.h>   // subdir aggregate
+
+#include <Grid_algorithms.h>// subdir aggregate
+
 #include <qcd/Grid_qcd.h>
 #include <parallelIO/GridNerscIO.h>
 
diff --git a/lib/Grid_algorithms.h b/lib/Grid_algorithms.h
new file mode 100644
index 00000000..b5a2814a
--- /dev/null
+++ b/lib/Grid_algorithms.h
@@ -0,0 +1,34 @@
+#ifndef GRID_ALGORITHMS_H
+#define GRID_ALGORITHMS_H
+
+#include <algorithms/SparseMatrix.h>
+#include <algorithms/LinearOperator.h>
+
+#include <algorithms/iterative/ConjugateGradient.h>
+#include <algorithms/iterative/NormalEquations.h>
+#include <algorithms/iterative/SchurRedBlack.h>
+
+#include <algorithms/approx/Zolotarev.h>
+#include <algorithms/approx/Chebyshev.h>
+#include <algorithms/approx/Remez.h>
+
+// Eigen/lanczos
+// EigCg
+// MCR
+// Pcg
+// Multishift CG
+// Hdcg
+// GCR
+// etc..
+
+// integrator/Leapfrog
+// integrator/Omelyan
+// integrator/ForceGradient
+
+// montecarlo/hmc
+// montecarlo/rhmc
+// montecarlo/metropolis
+// etc...
+
+
+#endif
diff --git a/lib/Grid_config.h.in b/lib/Grid_config.h.in
index 47643530..0ce09cf5 100644
--- a/lib/Grid_config.h.in
+++ b/lib/Grid_config.h.in
@@ -29,6 +29,9 @@
 /* Define to 1 if you have the `gettimeofday' function. */
 #undef HAVE_GETTIMEOFDAY
 
+/* Define to 1 if you have the <gmp.h> header file. */
+#undef HAVE_GMP_H
+
 /* Define to 1 if you have the <inttypes.h> header file. */
 #undef HAVE_INTTYPES_H
 
diff --git a/lib/Makefile.am b/lib/Makefile.am
index f76b0b64..938f7ca1 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -19,35 +19,50 @@ libGrid_a_SOURCES =\
 	stencil/Grid_stencil_common.cc\
 	qcd/Grid_qcd_dirac.cc\
 	qcd/Grid_qcd_wilson_dop.cc\
+	algorithms/approx/Zolotarev.cc\
+	algorithms/approx/Remez.cc\
 	$(extra_sources)
 
 #
 # Include files
 #
-include_HEADERS =\
+nobase_include_HEADERS = algorithms/approx/bigfloat.h\
+	algorithms/approx/Chebyshev.h\
+	algorithms/approx/Remez.h\
+	algorithms/approx/Zolotarev.h\
+	algorithms/iterative/ConjugateGradient.h\
+	algorithms/iterative/NormalEquations.h\
+	algorithms/iterative/SchurRedBlack.h\
+	algorithms/LinearOperator.h\
+	algorithms/SparseMatrix.h\
+	cartesian/Grid_cartesian_base.h\
+	cartesian/Grid_cartesian_full.h\
+	cartesian/Grid_cartesian_red_black.h\
+	communicator/Grid_communicator_base.h\
+	cshift/Grid_cshift_common.h\
+	cshift/Grid_cshift_mpi.h\
+	cshift/Grid_cshift_none.h\
 	Grid.h\
+	Grid_algorithms.h\
 	Grid_aligned_allocator.h\
 	Grid_cartesian.h\
 	Grid_communicator.h\
 	Grid_comparison.h\
-	Grid_config.h\
 	Grid_cshift.h\
+	Grid_extract.h\
 	Grid_lattice.h\
 	Grid_math.h\
 	Grid_simd.h\
-	Grid_stencil.h
-
-nobase_include_HEADERS=\
-	cartesian/Grid_cartesian_base.h\
-	cartesian/Grid_cartesian_full.h\
-	cartesian/Grid_cartesian_red_black.h\
-	cshift/Grid_cshift_common.h\
-	cshift/Grid_cshift_mpi.h\
-	cshift/Grid_cshift_none.h\
+	Grid_stencil.h\
+	Grid_threads.h\
 	lattice/Grid_lattice_arith.h\
+	lattice/Grid_lattice_base.h\
+	lattice/Grid_lattice_comparison.h\
 	lattice/Grid_lattice_conformable.h\
 	lattice/Grid_lattice_coordinate.h\
+	lattice/Grid_lattice_ET.h\
 	lattice/Grid_lattice_local.h\
+	lattice/Grid_lattice_overload.h\
 	lattice/Grid_lattice_peekpoke.h\
 	lattice/Grid_lattice_reality.h\
 	lattice/Grid_lattice_reduction.h\
@@ -71,8 +86,11 @@ nobase_include_HEADERS=\
 	math/Grid_math_trace.h\
 	math/Grid_math_traits.h\
 	math/Grid_math_transpose.h\
+	parallelIO/GridNerscIO.h\
 	qcd/Grid_qcd.h\
+	qcd/Grid_qcd_2spinor.h\
 	qcd/Grid_qcd_dirac.h\
+	qcd/Grid_qcd_wilson_dop.h\
 	simd/Grid_vComplexD.h\
 	simd/Grid_vComplexF.h\
 	simd/Grid_vInteger.h\
diff --git a/lib/algorithms/LinearOperator.h b/lib/algorithms/LinearOperator.h
index 12259236..bb948b95 100644
--- a/lib/algorithms/LinearOperator.h
+++ b/lib/algorithms/LinearOperator.h
@@ -1,26 +1,8 @@
 #ifndef  GRID_ALGORITHM_LINEAR_OP_H
 #define  GRID_ALGORITHM_LINEAR_OP_H
 
-#include <Grid.h>
-
 namespace Grid {
 
-  /////////////////////////////////////////////////////////////////////////////////////////////
-  // Interface defining what I expect of a general sparse matrix, such as a Fermion action
-  /////////////////////////////////////////////////////////////////////////////////////////////
-    template<class Field> class SparseMatrixBase {
-    public:
-      // Full checkerboar operations
-      virtual void M    (const Field &in, Field &out);
-      virtual void Mdag (const Field &in, Field &out);
-      virtual RealD MdagM(const Field &in, Field &out);
-      
-      // half checkerboard operaions
-      virtual void Mpc      (const Field &in, Field &out);
-      virtual void MpcDag   (const Field &in, Field &out);
-      virtual RealD MpcDagMpc(const Field &in, Field &out);
-    };
-
   /////////////////////////////////////////////////////////////////////////////////////////////
   // LinearOperators Take a something and return a something.
   /////////////////////////////////////////////////////////////////////////////////////////////
@@ -61,11 +43,11 @@ namespace Grid {
   // the wrappers implement the specialisation of "Op" and "AdjOp" to the cases minimising
   // replication of code.
   /////////////////////////////////////////////////////////////////////////////////////////////
-    template<class SparseMatrix,class Field>
+    template<class Matrix,class Field>
     class NonHermitianOperator : public LinearOperatorBase<Field> {
-      SparseMatrix &_Mat;
+      Matrix &_Mat;
     public:
-      NonHermitianOperator(SparseMatrix &Mat): _Mat(Mat){};
+      NonHermitianOperator(Matrix &Mat): _Mat(Mat){};
       void Op     (const Field &in, Field &out){
 	_Mat.M(in,out);
       }
@@ -77,11 +59,11 @@ namespace Grid {
     ////////////////////////////////////////////////////////////////////////////////////
     // Redblack Non hermitian wrapper
     ////////////////////////////////////////////////////////////////////////////////////
-    template<class SparseMatrix,class Field>
-    class NonHermitianRedBlackOperator : public LinearOperatorBase<Field> {
-      SparseMatrix &_Mat;
+    template<class Matrix,class Field>
+    class NonHermitianCheckerBoardedOperator : public LinearOperatorBase<Field> {
+      Matrix &_Mat;
     public:
-      NonHermitianRedBlackOperator(SparseMatrix &Mat): _Mat(Mat){};
+      NonHermitianCheckerBoardedOperator(Matrix &Mat): _Mat(Mat){};
       void Op     (const Field &in, Field &out){
 	_Mat.Mpc(in,out);
       }
@@ -93,75 +75,35 @@ namespace Grid {
     ////////////////////////////////////////////////////////////////////////////////////
     // Hermitian wrapper
     ////////////////////////////////////////////////////////////////////////////////////
-    template<class SparseMatrix,class Field>
+    template<class Matrix,class Field>
     class HermitianOperator : public HermitianOperatorBase<Field> {
-      SparseMatrix &_Mat;
+      Matrix &_Mat;
     public:
-      HermitianOperator(SparseMatrix &Mat): _Mat(Mat) {};
+      HermitianOperator(Matrix &Mat): _Mat(Mat) {};
       RealD OpAndNorm(const Field &in, Field &out){
 	return _Mat.MdagM(in,out);
       }
     };
 
     ////////////////////////////////////////////////////////////////////////////////////
-    // Hermitian RedBlack wrapper
+    // Hermitian CheckerBoarded wrapper
     ////////////////////////////////////////////////////////////////////////////////////
-    template<class SparseMatrix,class Field>
-    class HermitianRedBlackOperator : public HermitianOperatorBase<Field> {
-      SparseMatrix &_Mat;
+    template<class Matrix,class Field>
+    class HermitianCheckerBoardedOperator : public HermitianOperatorBase<Field> {
+      Matrix &_Mat;
     public:
-      HermitianRedBlackOperator(SparseMatrix &Mat): _Mat(Mat) {};
-      RealD OpAndNorm(const Field &in, Field &out){
-	return _Mat.MpcDagMpc(in,out);
+      HermitianCheckerBoardedOperator(Matrix &Mat): _Mat(Mat) {};
+      void OpAndNorm(const Field &in, Field &out,RealD &n1,RealD &n2){
+	_Mat.MpcDagMpc(in,out,n1,n2);
       }
     };
 
-
     /////////////////////////////////////////////////////////////
     // Base classes for functions of operators
     /////////////////////////////////////////////////////////////
     template<class Field> class OperatorFunction {
     public:
-      virtual void operator() (LinearOperatorBase<Field> *Linop, const Field &in, Field &out) = 0;
-    };
-
-    /////////////////////////////////////////////////////////////
-    // Base classes for polynomial functions of operators ? needed?
-    /////////////////////////////////////////////////////////////
-    template<class Field> class OperatorPolynomial : public OperatorFunction<Field> {
-    public:
-      virtual void operator() (LinearOperatorBase<Field> *Linop,const Field &in, Field &out) = 0;
-    };
-
-    /////////////////////////////////////////////////////////////
-    // Base classes for iterative processes based on operators
-    // single input vec, single output vec.
-    /////////////////////////////////////////////////////////////
-    template<class Field> class IterativeProcess : public OperatorFunction<Field> {
-    public:
-      RealD   Tolerance;
-      Integer MaxIterations;
-      IterativeProcess(RealD tol,Integer maxit) : Tolerance(tol),MaxIterations(maxit) {};
-      virtual void operator() (LinearOperatorBase<Field> *Linop,const Field &in, Field &out) = 0;
-    };
-
-    /////////////////////////////////////////////////////////////
-    // Grand daddy iterative method
-    /////////////////////////////////////////////////////////////
-    template<class Field> class ConjugateGradient : public IterativeProcess<Field> {
-    public:
-      virtual void operator() (HermitianOperatorBase<Field> *Linop,const Field &in, Field &out) = 0;
-    };
-
-    /////////////////////////////////////////////////////////////
-    // A little more modern
-    /////////////////////////////////////////////////////////////
-    template<class Field> class PreconditionedConjugateGradient : public  IterativeProcess<Field>  {
-    public:
-      void operator() (HermitianOperatorBase<Field> *Linop,
-		       OperatorFunction<Field> *Preconditioner,
-		       const Field &in, 
-		       Field &out) = 0;
+      virtual void operator() (LinearOperatorBase<Field> &Linop, const Field &in, Field &out) = 0;
     };
 
     // FIXME : To think about
diff --git a/lib/algorithms/SparseMatrix.h b/lib/algorithms/SparseMatrix.h
new file mode 100644
index 00000000..6c54fc92
--- /dev/null
+++ b/lib/algorithms/SparseMatrix.h
@@ -0,0 +1,72 @@
+#ifndef  GRID_ALGORITHM_SPARSE_MATRIX_H
+#define  GRID_ALGORITHM_SPARSE_MATRIX_H
+
+#include <Grid.h>
+
+namespace Grid {
+
+  /////////////////////////////////////////////////////////////////////////////////////////////
+  // Interface defining what I expect of a general sparse matrix, such as a Fermion action
+  /////////////////////////////////////////////////////////////////////////////////////////////
+    template<class Field> class SparseMatrixBase {
+    public:
+      // Full checkerboar operations
+      virtual RealD M    (const Field &in, Field &out)=0;
+      virtual RealD Mdag (const Field &in, Field &out)=0;
+      virtual void  MdagM(const Field &in, Field &out,RealD &ni,RealD &no) {
+	Field tmp (in._grid);
+	ni=M(in,tmp);
+	no=Mdag(tmp,out);
+      }
+    };
+
+  /////////////////////////////////////////////////////////////////////////////////////////////
+  // Interface augmented by a red black sparse matrix, such as a Fermion action
+  /////////////////////////////////////////////////////////////////////////////////////////////
+    template<class Field> class CheckerBoardedSparseMatrixBase : public SparseMatrixBase<Field> {
+    public:
+      
+      // half checkerboard operaions
+      virtual  void Meooe    (const Field &in, Field &out)=0;
+      virtual  void Mooee    (const Field &in, Field &out)=0;
+      virtual  void MooeeInv (const Field &in, Field &out)=0;
+
+      virtual  void MeooeDag    (const Field &in, Field &out)=0;
+      virtual  void MooeeDag    (const Field &in, Field &out)=0;
+      virtual  void MooeeInvDag (const Field &in, Field &out)=0;
+
+      // Schur decomp operators
+      virtual  RealD Mpc      (const Field &in, Field &out) {
+	Field tmp(in._grid);
+
+	Meooe(in,tmp);
+	MooeeInv(tmp,out);
+	Meooe(out,tmp);
+
+	Mooee(in,out);
+	out=out-tmp; // axpy_norm
+	RealD n=norm2(out);
+	return n;
+      }
+      virtual  RealD MpcDag   (const Field &in, Field &out){
+	Field tmp(in._grid);
+
+	MeooeDag(in,tmp);
+	MooeeInvDag(tmp,out);
+	MeooeDag(out,tmp);
+
+	MooeeDag(in,out);
+	out=out-tmp; // axpy_norm
+	RealD n=norm2(out);
+	return n;
+      }
+      virtual void MpcDagMpc(const Field &in, Field &out,RealD ni,RealD no) {
+	Field tmp(in._grid);
+	ni=Mpc(in,tmp);
+	no=Mpc(tmp,out);
+      }
+    };
+
+}
+
+#endif
diff --git a/lib/algorithms/approx/PolynomialApprox.h b/lib/algorithms/approx/Chebyshev.h
similarity index 93%
rename from lib/algorithms/approx/PolynomialApprox.h
rename to lib/algorithms/approx/Chebyshev.h
index 5515b325..5c7741e4 100644
--- a/lib/algorithms/approx/PolynomialApprox.h
+++ b/lib/algorithms/approx/Chebyshev.h
@@ -1,5 +1,5 @@
-#ifndef GRID_POLYNOMIAL_APPROX_H
-#define GRID_POLYNOMIAL_APPROX_H
+#ifndef GRID_CHEBYSHEV_H
+#define GRID_CHEBYSHEV_H
 
 #include<Grid.h>
 #include<algorithms/LinearOperator.h>
@@ -12,7 +12,7 @@ namespace Grid {
   template<class Field>
   class Polynomial : public OperatorFunction<Field> {
   private:
-    std::vector<double> _oeffs;
+    std::vector<double> Coeffs;
   public:
     Polynomial(std::vector<double> &_Coeffs) : Coeffs(_Coeffs) {};
 
@@ -111,17 +111,13 @@ namespace Grid {
       double xscale = 2.0/(hi-lo);
       double mscale = -(hi+lo)/(hi-lo);
 
-      Field *T0=Tnm;
-      Field *T1=Tn;
-      
-
       // Tn=T1 = (xscale M + mscale)in
       Linop.Op(T0,y);
 
       T1=y*xscale+in*mscale;
 
       // sum = .5 c[0] T0 + c[1] T1
-      out = (0.5*coeffs[0])*T0 + coeffs[1]*T1;
+      out = (0.5*Coeffs[0])*T0 + Coeffs[1]*T1;
 
       for(int n=2;n<order;n++){
 	
@@ -131,7 +127,7 @@ namespace Grid {
 
 	*Tnp=2.0*y-(*Tnm);
 	
-	out=out+coeffs[n]* (*Tnp);
+	out=out+Coeffs[n]* (*Tnp);
 
 	// Cycle pointers to avoid copies
 	Field *swizzle = Tnm;
diff --git a/lib/algorithms/approx/LICENSE b/lib/algorithms/approx/LICENSE
new file mode 100644
index 00000000..f971c5ed
--- /dev/null
+++ b/lib/algorithms/approx/LICENSE
@@ -0,0 +1,21 @@
+
+Copyright (c) 2011 Michael Clark
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in
+all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+THE SOFTWARE.
+
diff --git a/lib/algorithms/approx/README b/lib/algorithms/approx/README
new file mode 100644
index 00000000..90d8b010
--- /dev/null
+++ b/lib/algorithms/approx/README
@@ -0,0 +1,80 @@
+-----------------------------------------------------------------------------------
+
+PAB. Took Mike Clark's AlgRemez from GitHub and (modified a little) include.
+This is open source and license and readme and comments are preserved consistent
+with the license. Mike, thankyou!
+-----------------------------------------------------------------------------------
+-----------------------------------------------------------------------------------
+AlgRemez
+
+The archive downloadable here contains an implementation of the Remez
+algorithm which calculates optimal rational (and polynomial)
+approximations to the nth root over a given spectral range.  The Remez
+algorithm, although in principle is extremely straightforward to
+program, is quite difficult to get completely correct, e.g., the Maple
+implementation of the algorithm does not always converge to the
+correct answer.
+
+To use this algorithm you need to install GMP, the GNU Multiple
+Precision Library, and when configuring the install, you must include
+the --enable-mpfr option (see the GMP manual for more details).  You
+also have to edit the Makefile for AlgRemez appropriately for your
+system, namely to point corrrectly to the location of the GMP library.
+
+The simple main program included with this archive invokes the
+AlgRemez class to calculate an approximation given by command line
+arguments.  It is invoked by the following
+
+./test y z n d lambda_low lambda_high precision,
+
+where the function to be approximated is f(x) = x^(y/z), with degree
+(n,d) over the spectral range [lambda_low, lambda_high], using
+precision digits of precision in the arithmetic.  So an example would
+be
+
+./test 1 2 5 5 0.0004 64 40
+
+which corresponds to constructing a rational approximation to the
+square root function, with degree (5,5) over the range [0.0004,64]
+with 40 digits of precision used for the arithmetic.  The parameters y
+and z must be positive, the approximation to f(x) = x^(-y/z) is simply
+the inverse of the approximation to f(x) = x^(y/z).  After the
+approximation has been constructed, the roots and poles of the
+rational function are found, and then the partial fraction expansion
+of both the rational function and it's inverse are found, the results
+of which are output to a file called "approx.dat".  In addition, the
+error function of the approximation is output to "error.dat", where it
+can be checked that the resultant approximation satisfies Chebychev's
+criterion, namely all error maxima are equal in magnitude, and
+adjacent maxima are oppostie in sign.  There are some caveats here
+however, the optimal polynomial approximation has complex roots, and
+the root finding implemented here cannot (yet) handle complex roots.
+In addition, the partial fraction expansion of rational approximations
+is only found for the case n = d, i.e., the degree of numerator
+polynomial equals that of the denominator polynomial.  The convention
+for the partial fraction expansion is that polar shifts are always
+written added to x, not subtracted.
+
+To do list
+
+1.  Include an exponential dampening factor in the function to be
+approximated.  This may sound trivial to implement, but for some
+parameters, the algorithm seems to breakdown.  Also, the roots in the
+rational approximation sometimes become complex, which currently
+breaks the stupidly simple root finding code.
+
+2. Make the algorithm faster - it's too slow when running on qcdoc.
+
+3. Add complex root finding.
+
+4. Add more options for error minimisation - currently the code
+minimises the relative error, should add options for absolute error,
+and other norms.
+
+There will be a forthcoming publication concerning the results
+generated by this software, but in the meantime, if you use this
+software, please cite it as
+"M.A. Clark and A.D. Kennedy, https://github.com/mikeaclark/AlgRemez, 2005".
+
+If you have any problems using the software, then please email scientist.mike@gmail.com.
+
diff --git a/lib/algorithms/approx/Remez.cc b/lib/algorithms/approx/Remez.cc
new file mode 100755
index 00000000..a787e766
--- /dev/null
+++ b/lib/algorithms/approx/Remez.cc
@@ -0,0 +1,755 @@
+/*
+
+  Mike Clark - 25th May 2005
+
+  alg_remez.C
+
+  AlgRemez is an implementation of the Remez algorithm, which in this
+  case is used for generating the optimal nth root rational
+  approximation.
+
+  Note this class requires the gnu multiprecision (GNU MP) library.
+
+*/
+
+#include<math.h>
+#include<stdio.h>
+#include<stdlib.h>
+#include<string>
+#include<iostream>
+#include<iomanip>
+#include<cassert>
+
+#include<algorithms/approx/Remez.h>
+
+// Constructor
+AlgRemez::AlgRemez(double lower, double upper, long precision) 
+{
+  prec = precision;
+  bigfloat::setDefaultPrecision(prec);
+
+  apstrt = lower;
+  apend = upper;
+  apwidt = apend - apstrt;
+
+  std::cout<<"Approximation bounds are ["<<apstrt<<","<<apend<<"]\n";
+  std::cout<<"Precision of arithmetic is "<<precision<<std::endl;
+
+  alloc = 0;
+  n = 0;
+  d = 0;
+
+  foundRoots = 0;
+
+  // Only require the approximation spread to be less than 1 ulp
+  tolerance = 1e-15;
+}
+
+// Destructor
+AlgRemez::~AlgRemez()
+{
+  if (alloc) {
+    delete [] param;
+    delete [] roots;
+    delete [] poles;
+    delete [] xx;
+    delete [] mm;
+    delete [] a_power;
+    delete [] a;
+  }
+}
+
+// Free memory and reallocate as necessary
+void AlgRemez::allocate(int num_degree, int den_degree)
+{
+  // Arrays have previously been allocated, deallocate first, then allocate
+  if (alloc) {
+    delete [] param;
+    delete [] roots;
+    delete [] poles;
+    delete [] xx;
+    delete [] mm;
+  }
+
+  // Note use of new and delete in memory allocation - cannot run on qcdsp
+  param = new bigfloat[num_degree+den_degree+1];
+  roots = new bigfloat[num_degree];
+  poles = new bigfloat[den_degree];
+  xx = new bigfloat[num_degree+den_degree+3];
+  mm = new bigfloat[num_degree+den_degree+2];
+
+  if (!alloc) {
+    // The coefficients of the sum in the exponential
+    a = new bigfloat[SUM_MAX];
+    a_power = new int[SUM_MAX];
+  }
+
+  alloc = 1;
+}
+
+// Reset the bounds of the approximation
+void AlgRemez::setBounds(double lower, double upper)
+{
+  apstrt = lower;
+  apend = upper;
+  apwidt = apend - apstrt;
+}
+
+// Generate the rational approximation x^(pnum/pden)
+double AlgRemez::generateApprox(int degree, unsigned long pnum, 
+				unsigned long pden)
+{
+  return generateApprox(degree, degree, pnum, pden);
+}
+
+double AlgRemez::generateApprox(int num_degree, int den_degree, 
+				unsigned long pnum, unsigned long pden)
+{
+  double *a_param = 0;
+  int *a_pow = 0;
+  return generateApprox(num_degree, den_degree, pnum, pden, 0, a_param, a_pow);
+}
+
+// Generate the rational approximation x^(pnum/pden)
+double AlgRemez::generateApprox(int num_degree, int den_degree, 
+				unsigned long pnum, unsigned long pden,
+				int a_len, double *a_param, int *a_pow)
+{
+  std::cout<<"Degree of the approximation is ("<<num_degree<<","<<den_degree<<")\n";
+  std::cout<<"Approximating the function x^("<<pnum<<"/"<<pden<<"\n";
+
+  // Reallocate arrays, since degree has changed
+  if (num_degree != n || den_degree != d) allocate(num_degree,den_degree);
+
+  assert(a_len<=SUM_MAX);
+
+  step = new bigfloat[num_degree+den_degree+2];
+
+  a_length = a_len;
+  for (int j=0; j<a_len; j++) {
+    a[j]= a_param[j];
+    a_power[j] = a_pow[j];
+  }
+
+  power_num = pnum;
+  power_den = pden;
+  spread = 1.0e37;
+  iter = 0;
+
+  n = num_degree;
+  d = den_degree;
+  neq = n + d + 1;
+
+  initialGuess();
+  stpini(step);
+
+  while (spread > tolerance) { //iterate until convergance
+
+    if (iter++%100==0) 
+      std::cout<<"Iteration " <<iter-1<<" spread "<<(double)spread<<" delta "<<(double)delta<<std::endl; 
+
+    equations();
+    assert( delta>= tolerance);
+
+    search(step);
+  }
+
+  int sign;
+  double error = (double)getErr(mm[0],&sign);
+  std::cout<<"Converged at "<<iter<<" iterations; error = "<<error<<std::endl;
+
+  // Once the approximation has been generated, calculate the roots
+  if(!root()) {
+    std::cout<<"Root finding failed\n";
+  } else {
+    foundRoots = 1;
+  }
+  
+  delete [] step;
+
+  // Return the maximum error in the approximation
+  return error;
+}
+
+// Return the partial fraction expansion of the approximation x^(pnum/pden)
+int AlgRemez::getPFE(double *Res, double *Pole, double *Norm) {
+
+  if (n!=d) {
+    std::cout<<"Cannot handle case: Numerator degree neq Denominator degree\n";
+    return 0;
+  }
+
+  if (!alloc) {
+    std::cout<<"Approximation not yet generated\n";
+    return 0;
+  }
+
+  if (!foundRoots) {
+    std::cout<<"Roots not found, so PFE cannot be taken\n";
+    return 0;
+  }
+
+  bigfloat *r = new bigfloat[n];
+  bigfloat *p = new bigfloat[d];
+  
+  for (int i=0; i<n; i++) r[i] = roots[i];
+  for (int i=0; i<d; i++) p[i] = poles[i];
+  
+  // Perform a partial fraction expansion
+  pfe(r, p, norm);
+
+  // Convert to double and return
+  *Norm = (double)norm;
+  for (int i=0; i<n; i++) Res[i] = (double)r[i];
+  for (int i=0; i<d; i++) Pole[i] = (double)p[i];
+
+  delete [] r;
+  delete [] p;
+
+  // Where the smallest shift is located
+  return 0;
+}
+
+// Return the partial fraction expansion of the approximation x^(-pnum/pden)
+int AlgRemez::getIPFE(double *Res, double *Pole, double *Norm) {
+
+  if (n!=d) {
+    std::cout<<"Cannot handle case: Numerator degree neq Denominator degree\n";
+    return 0;
+  }
+
+  if (!alloc) {
+    std::cout<<"Approximation not yet generated\n";
+    return 0;
+  }
+
+  if (!foundRoots) {
+    std::cout<<"Roots not found, so PFE cannot be taken\n";
+    return 0;
+  }
+
+  bigfloat *r = new bigfloat[d];
+  bigfloat *p = new bigfloat[n];
+  
+  // Want the inverse function
+  for (int i=0; i<n; i++) {
+    r[i] = poles[i];
+    p[i] = roots[i];
+  }
+
+  // Perform a partial fraction expansion
+  pfe(r, p, (bigfloat)1l/norm);
+
+  // Convert to double and return
+  *Norm = (double)((bigfloat)1l/(norm));
+  for (int i=0; i<n; i++) {
+    Res[i] = (double)r[i];
+    Pole[i] = (double)p[i];
+  }
+
+  delete [] r;
+  delete [] p;
+
+  // Where the smallest shift is located
+  return 0;
+}
+
+// Initial values of maximal and minimal errors
+void AlgRemez::initialGuess() {
+
+  // Supply initial guesses for solution points
+  long ncheb = neq;			// Degree of Chebyshev error estimate
+  bigfloat a, r;
+
+  // Find ncheb+1 extrema of Chebyshev polynomial
+
+  a = ncheb;
+  mm[0] = apstrt;
+  for (long i = 1; i < ncheb; i++) {
+    r = 0.5 * (1 - cos((M_PI * i)/(double) a));
+    //r *= sqrt_bf(r);
+    r = (exp((double)r)-1.0)/(exp(1.0)-1.0);
+    mm[i] = apstrt + r * apwidt;
+  }
+  mm[ncheb] = apend;
+
+  a = 2.0 * ncheb;
+  for (long i = 0; i <= ncheb; i++) {
+    r = 0.5 * (1 - cos(M_PI * (2*i+1)/(double) a));
+    //r *= sqrt_bf(r); // Squeeze to low end of interval
+    r = (exp((double)r)-1.0)/(exp(1.0)-1.0);
+    xx[i] = apstrt + r * apwidt;
+  }
+}
+
+// Initialise step sizes
+void AlgRemez::stpini(bigfloat *step) {
+  xx[neq+1] = apend;
+  delta = 0.25;
+  step[0] = xx[0] - apstrt;
+  for (int i = 1; i < neq; i++) step[i] = xx[i] - xx[i-1];
+  step[neq] = step[neq-1];
+}
+
+// Search for error maxima and minima
+void AlgRemez::search(bigfloat *step) {
+  bigfloat a, q, xm, ym, xn, yn, xx0, xx1;
+  int i, j, meq, emsign, ensign, steps;
+
+  meq = neq + 1;
+  bigfloat *yy = new bigfloat[meq];
+
+  bigfloat eclose = 1.0e30;
+  bigfloat farther = 0l;
+
+  j = 1;
+  xx0 = apstrt;
+
+  for (i = 0; i < meq; i++) {
+    steps = 0;
+    xx1 = xx[i]; // Next zero
+    if (i == meq-1) xx1 = apend;
+    xm = mm[i];
+    ym = getErr(xm,&emsign);
+    q = step[i];
+    xn = xm + q;
+    if (xn < xx0 || xn >= xx1) {	// Cannot skip over adjacent boundaries
+      q = -q;
+      xn = xm;
+      yn = ym;
+      ensign = emsign;
+    } else {
+      yn = getErr(xn,&ensign);
+      if (yn < ym) {
+	q = -q;
+	xn = xm;
+	yn = ym;
+	ensign = emsign;
+      }
+    }
+  
+    while(yn >= ym) {		// March until error becomes smaller.
+      if (++steps > 10) break;
+      ym = yn;
+      xm = xn;
+      emsign = ensign;
+      a = xm + q;
+      if (a == xm || a <= xx0 || a >= xx1) break;// Must not skip over the zeros either side.
+      xn = a;
+      yn = getErr(xn,&ensign);
+    }
+
+    mm[i] = xm;			// Position of maximum
+    yy[i] = ym;			// Value of maximum
+
+    if (eclose > ym) eclose = ym;
+    if (farther < ym) farther = ym;
+
+    xx0 = xx1; // Walk to next zero.
+  } // end of search loop
+
+  q = (farther - eclose);	// Decrease step size if error spread increased
+  if (eclose != 0.0) q /= eclose; // Relative error spread
+  if (q >= spread) delta *= 0.5; // Spread is increasing; decrease step size
+  spread = q;
+
+  for (i = 0; i < neq; i++) {
+    q = yy[i+1];
+    if (q != 0.0) q = yy[i] / q  - (bigfloat)1l;
+    else q = 0.0625;
+    if (q > (bigfloat)0.25) q = 0.25;
+    q *= mm[i+1] - mm[i];
+    step[i] = q * delta;
+  }
+  step[neq] = step[neq-1];
+  
+  for (i = 0; i < neq; i++) {	// Insert new locations for the zeros.
+    xm = xx[i] - step[i];
+    if (xm <= apstrt) continue;
+    if (xm >= apend) continue;
+    if (xm <= mm[i]) xm = (bigfloat)0.5 * (mm[i] + xx[i]);
+    if (xm >= mm[i+1]) xm = (bigfloat)0.5 * (mm[i+1] + xx[i]);
+    xx[i] = xm;
+  }
+
+  delete [] yy;
+}
+
+// Solve the equations
+void AlgRemez::equations(void) {
+  bigfloat x, y, z;
+  int i, j, ip;
+  bigfloat *aa;
+
+  bigfloat *AA = new bigfloat[(neq)*(neq)];
+  bigfloat *BB = new bigfloat[neq];
+  
+  for (i = 0; i < neq; i++) {	// set up the equations for solution by simq()
+    ip = neq * i;		// offset to 1st element of this row of matrix
+    x = xx[i];			// the guess for this row
+    y = func(x);		// right-hand-side vector
+
+    z = (bigfloat)1l;
+    aa = AA+ip;
+    for (j = 0; j <= n; j++) {
+      *aa++ = z;
+      z *= x;
+    }
+
+    z = (bigfloat)1l;
+    for (j = 0; j < d; j++) {
+      *aa++ = -y * z;
+      z *= x;
+    }
+    BB[i] = y * z;		// Right hand side vector
+  }
+
+  // Solve the simultaneous linear equations.
+  if (simq(AA, BB, param, neq)) {
+    std::cout<<"simq failed\n";
+    exit(0);
+  }
+
+  delete [] AA;
+  delete [] BB;
+
+}
+
+// Evaluate the rational form P(x)/Q(x) using coefficients
+// from the solution vector param
+bigfloat AlgRemez::approx(const bigfloat x) {
+  bigfloat yn, yd;
+  int i;
+
+  // Work backwards toward the constant term.
+  yn = param[n];		// Highest order numerator coefficient
+  for (i = n-1; i >= 0; i--) yn = x * yn  +  param[i]; 
+  yd = x + param[n+d];	// Highest degree coefficient = 1.0
+  for (i = n+d-1; i > n; i--) yd = x * yd  +  param[i];
+
+  return(yn/yd);
+}
+
+// Compute size and sign of the approximation error at x
+bigfloat AlgRemez::getErr(bigfloat x, int *sign) {
+  bigfloat e, f;
+
+  f = func(x);
+  e = approx(x) - f;
+  if (f != 0) e /= f;
+  if (e < (bigfloat)0.0) {
+    *sign = -1;
+    e = -e;
+  }
+  else *sign = 1;
+  
+  return(e);
+}
+
+// Calculate function required for the approximation.
+bigfloat AlgRemez::func(const bigfloat x) {
+
+  bigfloat z = (bigfloat)power_num / (bigfloat)power_den;
+  bigfloat y;
+
+  if (x == (bigfloat)1.0) y = (bigfloat)1.0;
+  else y = pow_bf(x,z);
+
+  if (a_length > 0) {
+    bigfloat sum = 0l;
+    for (int j=0; j<a_length; j++) sum += a[j]*pow_bf(x,a_power[j]);
+    return y * exp_bf(sum);
+  } else {
+    return y;
+  }
+
+}
+
+// Solve the system AX=B
+int AlgRemez::simq(bigfloat A[], bigfloat B[], bigfloat X[], int n) {
+
+  int i, j, ij, ip, ipj, ipk, ipn;
+  int idxpiv, iback;
+  int k, kp, kp1, kpk, kpn;
+  int nip, nkp, nm1;
+  bigfloat em, q, rownrm, big, size, pivot, sum;
+  bigfloat *aa;
+
+  // simq() work vector
+  int *IPS = new int[(neq) * sizeof(int)];
+
+  nm1 = n - 1;
+  // Initialize IPS and X
+  
+  ij = 0;
+  for (i = 0; i < n; i++) {
+    IPS[i] = i;
+    rownrm = 0.0;
+    for(j = 0; j < n; j++) {
+      q = abs_bf(A[ij]);
+      if(rownrm < q) rownrm = q;
+      ++ij;
+    }
+    if (rownrm == (bigfloat)0l) {
+      std::cout<<"simq rownrm=0\n";
+      delete [] IPS;
+      return(1);
+    }
+    X[i] = (bigfloat)1.0 / rownrm;
+  }
+  
+  for (k = 0; k < nm1; k++) {
+    big = 0.0;
+    idxpiv = 0;
+    
+    for (i = k; i < n; i++) {
+      ip = IPS[i];
+      ipk = n*ip + k;
+      size = abs_bf(A[ipk]) * X[ip];
+      if (size > big) {
+	big = size;
+	idxpiv = i;
+      }
+    }
+    
+    if (big == (bigfloat)0l) {
+      std::cout<<"simq big=0\n";
+      delete [] IPS;
+      return(2);
+    }
+    if (idxpiv != k) {
+      j = IPS[k];
+      IPS[k] = IPS[idxpiv];
+      IPS[idxpiv] = j;
+    }
+    kp = IPS[k];
+    kpk = n*kp + k;
+    pivot = A[kpk];
+    kp1 = k+1;
+    for (i = kp1; i < n; i++) {
+      ip = IPS[i];
+      ipk = n*ip + k;
+      em = -A[ipk] / pivot;
+      A[ipk] = -em;
+      nip = n*ip;
+      nkp = n*kp;
+      aa = A+nkp+kp1;
+      for (j = kp1; j < n; j++) {
+	ipj = nip + j;
+	A[ipj] = A[ipj] + em * *aa++;
+      }
+    }
+  }
+  kpn = n * IPS[n-1] + n - 1;	// last element of IPS[n] th row
+  if (A[kpn] == (bigfloat)0l) {
+    std::cout<<"simq A[kpn]=0\n";
+    delete [] IPS;
+    return(3);
+  }
+
+  
+  ip = IPS[0];
+  X[0] = B[ip];
+  for (i = 1; i < n; i++) {
+    ip = IPS[i];
+    ipj = n * ip;
+    sum = 0.0;
+    for (j = 0; j < i; j++) {
+      sum += A[ipj] * X[j];
+      ++ipj;
+    }
+    X[i] = B[ip] - sum;
+  }
+  
+  ipn = n * IPS[n-1] + n - 1;
+  X[n-1] = X[n-1] / A[ipn];
+  
+  for (iback = 1; iback < n; iback++) {
+    //i goes (n-1),...,1
+    i = nm1 - iback;
+    ip = IPS[i];
+    nip = n*ip;
+    sum = 0.0;
+    aa = A+nip+i+1;
+    for (j= i + 1; j < n; j++) 
+      sum += *aa++ * X[j];
+    X[i] = (X[i] - sum) / A[nip+i];
+  }
+  
+  delete [] IPS;
+  return(0);
+}
+
+// Calculate the roots of the approximation
+int AlgRemez::root() {
+
+  long i,j;
+  bigfloat x,dx=0.05;
+  bigfloat upper=1, lower=-100000;
+  bigfloat tol = 1e-20;
+
+  bigfloat *poly = new bigfloat[neq+1];
+
+  // First find the numerator roots
+  for (i=0; i<=n; i++) poly[i] = param[i];
+
+  for (i=n-1; i>=0; i--) {
+    roots[i] = rtnewt(poly,i+1,lower,upper,tol);
+    if (roots[i] == 0.0) {
+      std::cout<<"Failure to converge on root "<<i+1<<"/"<<n<<"\n";
+      return 0;
+    }
+    poly[0] = -poly[0]/roots[i];
+    for (j=1; j<=i; j++) poly[j] = (poly[j-1] - poly[j])/roots[i];
+  }
+  
+ // Now find the denominator roots
+  poly[d] = 1l;
+  for (i=0; i<d; i++) poly[i] = param[n+1+i];
+
+  for (i=d-1; i>=0; i--) {
+    poles[i]=rtnewt(poly,i+1,lower,upper,tol);
+    if (poles[i] == 0.0) {
+      std::cout<<"Failure to converge on pole "<<i+1<<"/"<<d<<"\n";
+      return 0;
+    }
+    poly[0] = -poly[0]/poles[i];
+    for (j=1; j<=i; j++) poly[j] = (poly[j-1] - poly[j])/poles[i];
+  }
+
+  norm = param[n];
+
+  delete [] poly;
+
+  return 1;
+}
+
+// Evaluate the polynomial
+bigfloat AlgRemez::polyEval(bigfloat x, bigfloat *poly, long size) {
+  bigfloat f = poly[size];
+  for (int i=size-1; i>=0; i--) f = f*x + poly[i];
+  return f;
+}
+
+// Evaluate the differential of the polynomial
+bigfloat AlgRemez::polyDiff(bigfloat x, bigfloat *poly, long size) {
+  bigfloat df = (bigfloat)size*poly[size];
+  for (int i=size-1; i>0; i--) df = df*x + (bigfloat)i*poly[i];
+  return df;
+}
+
+
+// Newton's method to calculate roots
+bigfloat AlgRemez::rtnewt(bigfloat *poly, long i, bigfloat x1, 
+			  bigfloat x2, bigfloat xacc) {
+  int j;
+  bigfloat df, dx, f, rtn;
+
+  rtn=(bigfloat)0.5*(x1+x2);
+  for (j=1; j<=JMAX;j++) {
+    f = polyEval(rtn, poly, i);
+    df = polyDiff(rtn, poly, i);
+    dx = f/df;
+    rtn -= dx;
+    if (abs_bf(dx) < xacc) return rtn;
+  }
+  std::cout<<"Maximum number of iterations exceeded in rtnewt\n";
+  return 0.0;
+}
+
+// Evaluate the partial fraction expansion of the rational function
+// with res roots and poles poles.  Result is overwritten on input
+// arrays.
+void AlgRemez::pfe(bigfloat *res, bigfloat *poles, bigfloat norm) {
+  int i,j,small;
+  bigfloat temp;
+  bigfloat *numerator = new bigfloat[n];
+  bigfloat *denominator = new bigfloat[d];
+
+  // Construct the polynomials explicitly 
+  for (i=1; i<n; i++) {
+    numerator[i] = 0l;
+    denominator[i] = 0l;
+  }
+  numerator[0]=1l;
+  denominator[0]=1l;
+
+  for (j=0; j<n; j++) {
+    for (i=n-1; i>=0; i--) {
+      numerator[i] *= -res[j];
+      denominator[i] *= -poles[j];
+      if (i>0) {
+	numerator[i] += numerator[i-1];
+	denominator[i] += denominator[i-1];
+      }
+    }
+  }
+
+  // Convert to proper fraction form.
+  // Fraction is now in the form 1 + n/d, where O(n)+1=O(d)
+  for (i=0; i<n; i++) numerator[i] -= denominator[i];
+
+  // Find the residues of the partial fraction expansion and absorb the
+  // coefficients.
+  for (i=0; i<n; i++) {
+    res[i] = 0l;
+    for (j=n-1; j>=0; j--) {
+      res[i] = poles[i]*res[i]+numerator[j];
+    }
+    for (j=n-1; j>=0; j--) {
+      if (i!=j) res[i] /= poles[i]-poles[j];
+    }
+    res[i] *= norm;
+  }  
+
+  // res now holds the residues
+  j = 0;
+  for (i=0; i<n; i++) poles[i] = -poles[i];
+
+  // Move the ordering of the poles from smallest to largest
+  for (j=0; j<n; j++) {
+    small = j;
+    for (i=j+1; i<n; i++) {
+      if (poles[i] < poles[small]) small = i;
+    }
+    if (small != j) {
+      temp = poles[small];
+      poles[small] = poles[j];
+      poles[j] = temp;
+      temp = res[small];
+      res[small] = res[j];
+      res[j] = temp;
+    }
+  }
+
+  delete [] numerator;
+  delete [] denominator;
+}
+
+double AlgRemez::evaluateApprox(double x) {
+  return (double)approx((bigfloat)x);
+}
+
+double AlgRemez::evaluateInverseApprox(double x) {
+  return 1.0/(double)approx((bigfloat)x);
+}
+
+double AlgRemez::evaluateFunc(double x) {
+  return (double)func((bigfloat)x);
+}
+
+double AlgRemez::evaluateInverseFunc(double x) {
+  return 1.0/(double)func((bigfloat)x);
+}
+
+void AlgRemez::csv(std::ostream & os)
+{
+  double lambda_low = apstrt;
+  double lambda_high= apend;
+  for (double x=lambda_low; x<lambda_high; x*=1.05) {
+    double f = evaluateFunc(x);
+    double r = evaluateApprox(x);
+    os<< x<<","<<r<<","<<f<<","<<r-f<<std::endl;
+  }
+  return;
+}
diff --git a/lib/algorithms/approx/Remez.h b/lib/algorithms/approx/Remez.h
new file mode 100755
index 00000000..906811be
--- /dev/null
+++ b/lib/algorithms/approx/Remez.h
@@ -0,0 +1,167 @@
+/*
+
+  Mike Clark - 25th May 2005
+
+  alg_remez.h
+
+  AlgRemez is an implementation of the Remez algorithm, which in this
+  case is used for generating the optimal nth root rational
+  approximation.
+
+  Note this class requires the gnu multiprecision (GNU MP) library.
+
+*/
+
+#ifndef INCLUDED_ALG_REMEZ_H
+#define INCLUDED_ALG_REMEZ_H
+
+#include <algorithms/approx/bigfloat.h>
+
+#define JMAX 10000 //Maximum number of iterations of Newton's approximation
+#define SUM_MAX 10 // Maximum number of terms in exponential
+
+/*
+ *Usage examples
+  AlgRemez remez(lambda_low,lambda_high,precision);
+  error = remez.generateApprox(n,d,y,z);
+  remez.getPFE(res,pole,&norm);
+  remez.getIPFE(res,pole,&norm);
+  remez.csv(ostream &os);
+ */
+class AlgRemez
+{
+ private:
+  char *cname;
+
+  // The approximation parameters
+  bigfloat *param, *roots, *poles;
+  bigfloat norm;
+
+  // The numerator and denominator degree (n=d)
+  int n, d;
+  
+  // The bounds of the approximation
+  bigfloat apstrt, apwidt, apend;
+
+  // the numerator and denominator of the power we are approximating
+  unsigned long power_num; 
+  unsigned long power_den;
+
+  // Flag to determine whether the arrays have been allocated
+  int alloc;
+
+  // Flag to determine whether the roots have been found
+  int foundRoots;
+
+  // Variables used to calculate the approximation
+  int nd1, iter;
+  bigfloat *xx, *mm, *step;
+  bigfloat delta, spread, tolerance;
+
+  // The exponential summation coefficients
+  bigfloat *a;
+  int *a_power;
+  int a_length;
+
+  // The number of equations we must solve at each iteration (n+d+1)
+  int neq;
+
+  // The precision of the GNU MP library
+  long prec;
+
+  // Initial values of maximal and minmal errors
+  void initialGuess();
+
+  // Solve the equations
+  void equations();
+
+  // Search for error maxima and minima
+  void search(bigfloat *step); 
+
+  // Initialise step sizes
+  void stpini(bigfloat *step);
+
+  // Calculate the roots of the approximation
+  int root();
+
+  // Evaluate the polynomial
+  bigfloat polyEval(bigfloat x, bigfloat *poly, long size);
+  //complex_bf polyEval(complex_bf x, complex_bf *poly, long size);
+
+  // Evaluate the differential of the polynomial
+  bigfloat polyDiff(bigfloat x, bigfloat *poly, long size);
+  //complex_bf polyDiff(complex_bf x, complex_bf *poly, long size);
+
+  // Newton's method to calculate roots
+  bigfloat rtnewt(bigfloat *poly, long i, bigfloat x1, bigfloat x2, bigfloat xacc);
+  //complex_bf rtnewt(complex_bf *poly, long i, bigfloat x1, bigfloat x2, bigfloat xacc);
+
+  // Evaluate the partial fraction expansion of the rational function
+  // with res roots and poles poles.  Result is overwritten on input
+  // arrays.
+  void pfe(bigfloat *res, bigfloat* poles, bigfloat norm);
+
+  // Calculate function required for the approximation
+  bigfloat func(bigfloat x);
+
+  // Compute size and sign of the approximation error at x
+  bigfloat getErr(bigfloat x, int *sign);
+
+  // Solve the system AX=B
+  int simq(bigfloat *A, bigfloat *B, bigfloat *X, int n);
+
+  // Free memory and reallocate as necessary
+  void allocate(int num_degree, int den_degree);
+
+  // Evaluate the rational form P(x)/Q(x) using coefficients from the
+  // solution vector param
+  bigfloat approx(bigfloat x);
+
+ public:
+  
+  // Constructor
+  AlgRemez(double lower, double upper, long prec);
+
+  // Destructor
+  virtual ~AlgRemez();
+
+  // Reset the bounds of the approximation
+  void setBounds(double lower, double upper);
+
+  // Generate the rational approximation x^(pnum/pden)
+  double generateApprox(int num_degree, int den_degree, 
+			unsigned long power_num, unsigned long power_den, 
+			int a_len, double* a_param, int* a_pow);
+  double generateApprox(int num_degree, int den_degree, 
+			unsigned long power_num, unsigned long power_den);
+  double generateApprox(int degree, unsigned long power_num, 
+			unsigned long power_den);
+
+  // Return the partial fraction expansion of the approximation x^(pnum/pden)
+  int getPFE(double *res, double *pole, double *norm);
+
+  // Return the partial fraction expansion of the approximation x^(-pnum/pden)
+  int getIPFE(double *res, double *pole, double *norm);
+
+  // Evaluate the rational form P(x)/Q(x) using coefficients from the
+  // solution vector param
+  double evaluateApprox(double x);
+
+  // Evaluate the rational form Q(x)/P(x) using coefficients from the
+  // solution vector param
+  double evaluateInverseApprox(double x);
+
+  // Calculate function required for the approximation
+  double evaluateFunc(double x);
+
+  // Calculate inverse function required for the approximation
+  double evaluateInverseFunc(double x);
+  
+  // Dump csv of function, approx and error
+  void csv(std::ostream &os);
+};
+
+#endif  // Include guard
+
+
+
diff --git a/lib/algorithms/approx/zolotarev.c b/lib/algorithms/approx/Zolotarev.cc
similarity index 100%
rename from lib/algorithms/approx/zolotarev.c
rename to lib/algorithms/approx/Zolotarev.cc
diff --git a/lib/algorithms/approx/bigfloat.h b/lib/algorithms/approx/bigfloat.h
new file mode 100755
index 00000000..7f59ea33
--- /dev/null
+++ b/lib/algorithms/approx/bigfloat.h
@@ -0,0 +1,368 @@
+/*
+  Mike Clark - 25th May 2005
+
+  bigfloat.h
+
+  Simple C++ wrapper for multiprecision datatype used by AlgRemez
+  algorithm
+*/
+
+#ifndef INCLUDED_BIGFLOAT_H
+#define INCLUDED_BIGFLOAT_H
+
+#ifdef HAVE_GMP
+#include <gmp.h>
+#include <mpf2mpfr.h>
+#include <mpfr.h>
+class bigfloat {
+
+private:
+
+  mpf_t x;
+
+public:
+
+  bigfloat() { mpf_init(x); }
+  bigfloat(const bigfloat& y) { mpf_init_set(x, y.x); }
+  bigfloat(const unsigned long u) { mpf_init_set_ui(x, u); }
+  bigfloat(const long i) { mpf_init_set_si(x, i); }
+  bigfloat(const int i) {mpf_init_set_si(x,(long)i);}
+  bigfloat(const float d) { mpf_init_set_d(x, (double)d); }
+  bigfloat(const double d) { mpf_init_set_d(x, d); }  
+  bigfloat(const char *str) { mpf_init_set_str(x, (char*)str, 10); }
+  ~bigfloat(void) { mpf_clear(x); }
+  operator const double (void) const { return (double)mpf_get_d(x); }
+  static void setDefaultPrecision(unsigned long dprec) {
+    unsigned long bprec =  (unsigned long)(3.321928094 * (double)dprec);
+    mpf_set_default_prec(bprec);
+  }
+
+  void setPrecision(unsigned long dprec) {
+    unsigned long bprec =  (unsigned long)(3.321928094 * (double)dprec);
+    mpf_set_prec(x,bprec);
+  }
+  
+  unsigned long getPrecision(void) const { return mpf_get_prec(x); }
+
+  unsigned long getDefaultPrecision(void) const { return mpf_get_default_prec(); }
+
+  bigfloat& operator=(const bigfloat& y) {
+    mpf_set(x, y.x); 
+    return *this;
+  }
+
+  bigfloat& operator=(const unsigned long y) { 
+    mpf_set_ui(x, y);
+    return *this; 
+  }
+  
+  bigfloat& operator=(const signed long y) {
+    mpf_set_si(x, y); 
+    return *this;
+  }
+  
+  bigfloat& operator=(const float y) {
+    mpf_set_d(x, (double)y); 
+    return *this;
+  }
+
+  bigfloat& operator=(const double y) {
+    mpf_set_d(x, y); 
+    return *this;
+  }
+
+  size_t write(void);
+  size_t read(void);
+
+  /* Arithmetic Functions */
+
+  bigfloat& operator+=(const bigfloat& y) { return *this = *this + y; }
+  bigfloat& operator-=(const bigfloat& y) { return *this = *this - y; }
+  bigfloat& operator*=(const bigfloat& y) { return *this = *this * y; }
+  bigfloat& operator/=(const bigfloat& y) { return *this = *this / y; }
+
+  friend bigfloat operator+(const bigfloat& x, const bigfloat& y) {
+    bigfloat a;
+    mpf_add(a.x,x.x,y.x);
+    return a;
+  }
+
+  friend bigfloat operator+(const bigfloat& x, const unsigned long y) {
+    bigfloat a;
+    mpf_add_ui(a.x,x.x,y);
+    return a;
+  }
+
+  friend bigfloat operator-(const bigfloat& x, const bigfloat& y) {
+    bigfloat a;
+    mpf_sub(a.x,x.x,y.x);
+    return a;
+  }
+  
+  friend bigfloat operator-(const unsigned long x, const bigfloat& y) {
+    bigfloat a;
+    mpf_ui_sub(a.x,x,y.x);
+    return a;
+  }
+  
+  friend bigfloat operator-(const bigfloat& x, const unsigned long y) {
+    bigfloat a;
+    mpf_sub_ui(a.x,x.x,y);
+    return a;
+  }
+
+  friend bigfloat operator-(const bigfloat& x) {
+    bigfloat a;
+    mpf_neg(a.x,x.x);
+    return a;
+  }
+
+  friend bigfloat operator*(const bigfloat& x, const bigfloat& y) {
+    bigfloat a;
+    mpf_mul(a.x,x.x,y.x);
+    return a;
+  }
+
+  friend bigfloat operator*(const bigfloat& x, const unsigned long y) {
+    bigfloat a;
+    mpf_mul_ui(a.x,x.x,y);
+    return a;
+  }
+
+  friend bigfloat operator/(const bigfloat& x, const bigfloat& y){
+    bigfloat a;
+    mpf_div(a.x,x.x,y.x);
+    return a;
+  }
+
+  friend bigfloat operator/(const unsigned long x, const bigfloat& y){
+    bigfloat a;
+    mpf_ui_div(a.x,x,y.x);
+    return a;
+  }
+
+  friend bigfloat operator/(const bigfloat& x, const unsigned long y){
+    bigfloat a;
+    mpf_div_ui(a.x,x.x,y);
+    return a;
+  }
+
+  friend bigfloat sqrt_bf(const bigfloat& x){
+    bigfloat a;
+    mpf_sqrt(a.x,x.x);
+    return a;
+  }
+
+  friend bigfloat sqrt_bf(const unsigned long x){
+    bigfloat a;
+    mpf_sqrt_ui(a.x,x);
+    return a;
+  }
+
+  friend bigfloat abs_bf(const bigfloat& x){
+    bigfloat a;
+    mpf_abs(a.x,x.x);
+    return a;
+  }
+
+  friend bigfloat pow_bf(const bigfloat& a, long power) {
+    bigfloat b;
+    mpf_pow_ui(b.x,a.x,power);
+    return b;
+  }
+
+  friend bigfloat pow_bf(const bigfloat& a, bigfloat &power) {
+    bigfloat b;
+    mpfr_pow(b.x,a.x,power.x,GMP_RNDN);
+    return b;
+  }
+
+  friend bigfloat exp_bf(const bigfloat& a) {
+    bigfloat b;
+    mpfr_exp(b.x,a.x,GMP_RNDN);
+    return b;
+  }
+
+  /* Comparison Functions */
+
+  friend int operator>(const bigfloat& x, const bigfloat& y) {
+    int test;
+    test = mpf_cmp(x.x,y.x);
+    if (test > 0) return 1;
+    else return 0;
+  }
+
+  friend int operator<(const bigfloat& x, const bigfloat& y) {
+    int test;
+    test = mpf_cmp(x.x,y.x);
+    if (test < 0) return 1;
+    else return 0;
+  }
+
+  friend int sgn(const bigfloat&);
+
+};
+#else
+
+typedef double mfloat; 
+class bigfloat {
+private:
+
+  mfloat x;
+
+public:
+
+  bigfloat() { }
+  bigfloat(const bigfloat& y) { x=y.x; }
+  bigfloat(const unsigned long u) { x=u; }
+  bigfloat(const long i) { x=i; }
+  bigfloat(const int i) { x=i;}
+  bigfloat(const float d) { x=d;}
+  bigfloat(const double d) {  x=d;}
+  bigfloat(const char *str) { x=std::stod(std::string(str));}
+  ~bigfloat(void) { }
+  operator double (void) const { return (double)x; }
+  static void setDefaultPrecision(unsigned long dprec) {
+  }
+
+  void setPrecision(unsigned long dprec) {
+  }
+  
+  unsigned long getPrecision(void) const { return 64; }
+  unsigned long getDefaultPrecision(void) const { return 64; }
+
+  bigfloat& operator=(const bigfloat& y)     { x=y.x;    return *this;  }
+  bigfloat& operator=(const unsigned long y) { x=y; return *this; }
+  bigfloat& operator=(const signed long y)   { x=y; return *this; }
+  bigfloat& operator=(const float y)    { x=y; return *this; }
+  bigfloat& operator=(const double y)   { x=y; return *this; }
+
+  size_t write(void);
+  size_t read(void);
+
+  /* Arithmetic Functions */
+
+  bigfloat& operator+=(const bigfloat& y) { return *this = *this + y; }
+  bigfloat& operator-=(const bigfloat& y) { return *this = *this - y; }
+  bigfloat& operator*=(const bigfloat& y) { return *this = *this * y; }
+  bigfloat& operator/=(const bigfloat& y) { return *this = *this / y; }
+
+  friend bigfloat operator+(const bigfloat& x, const bigfloat& y) { 
+    bigfloat a;
+    a.x=x.x+y.x;
+    return a;
+  }
+
+  friend bigfloat operator+(const bigfloat& x, const unsigned long y) {
+    bigfloat a;
+    a.x=x.x+y;
+    return a;
+  }
+
+  friend bigfloat operator-(const bigfloat& x, const bigfloat& y) {
+    bigfloat a;
+    a.x=x.x-y.x;
+    return a;
+  }
+  
+  friend bigfloat operator-(const unsigned long x, const bigfloat& y) {
+    bigfloat bx(x);
+    return bx-y;
+  }
+  
+  friend bigfloat operator-(const bigfloat& x, const unsigned long y) {
+    bigfloat by(y);
+    return x-by;
+  }
+
+  friend bigfloat operator-(const bigfloat& x) {
+    bigfloat a;
+    a.x=-x.x;
+    return a;
+  }
+
+  friend bigfloat operator*(const bigfloat& x, const bigfloat& y) {
+    bigfloat a;
+    a.x=x.x*y.x;
+    return a;
+  }
+
+  friend bigfloat operator*(const bigfloat& x, const unsigned long y) {
+    bigfloat a;
+    a.x=x.x*y;
+    return a;
+  }
+
+  friend bigfloat operator/(const bigfloat& x, const bigfloat& y){
+    bigfloat a;
+    a.x=x.x/y.x;
+    return a;
+  }
+
+  friend bigfloat operator/(const unsigned long x, const bigfloat& y){
+    bigfloat bx(x);
+    return bx/y;
+  }
+
+  friend bigfloat operator/(const bigfloat& x, const unsigned long y){
+    bigfloat by(y);
+    return x/by;
+  }
+
+  friend bigfloat sqrt_bf(const bigfloat& x){
+    bigfloat a;
+    a.x= sqrt(x.x);
+    return a;
+  }
+
+  friend bigfloat sqrt_bf(const unsigned long x){
+    bigfloat a(x);
+    return sqrt_bf(a);
+  }
+
+  friend bigfloat abs_bf(const bigfloat& x){
+    bigfloat a;
+    a.x=abs(x.x);
+    return a;
+  }
+
+  friend bigfloat pow_bf(const bigfloat& a, long power) {
+    bigfloat b;
+    b.x=pow(a.x,power);
+    return b;
+  }
+
+  friend bigfloat pow_bf(const bigfloat& a, bigfloat &power) {
+    bigfloat b;
+    b.x=pow(a.x,power.x);
+    return b;
+  }
+
+  friend bigfloat exp_bf(const bigfloat& a) {
+    bigfloat b;
+    b.x=exp(a.x);
+    return b;
+  }
+
+  /* Comparison Functions */
+  friend int operator>(const bigfloat& x, const bigfloat& y) {
+    return x.x>y.x;
+  }
+
+  friend int operator<(const bigfloat& x, const bigfloat& y) {
+    return x.x<y.x;
+  }
+
+  friend int sgn(const bigfloat& x) {
+    if ( x.x>=0 )  return 1;   
+    else return 0;
+  }
+
+  /* Miscellaneous Functions */
+
+  //  friend bigfloat& random(void);
+};
+
+#endif
+
+#endif
diff --git a/lib/algorithms/iterative/ConjugateGradient.h b/lib/algorithms/iterative/ConjugateGradient.h
index ad305d2f..bcb6ab60 100644
--- a/lib/algorithms/iterative/ConjugateGradient.h
+++ b/lib/algorithms/iterative/ConjugateGradient.h
@@ -1,19 +1,25 @@
 #ifndef GRID_CONJUGATE_GRADIENT_H
 #define GRID_CONJUGATE_GRADIENT_H
 
-#include<Grid.h>
-#include<algorithms/LinearOperator.h>
-
 namespace Grid {
 
-  template<class Field> class ConjugateGradient : public IterativeProcess<Field> {
-  public:
+    /////////////////////////////////////////////////////////////
+    // Base classes for iterative processes based on operators
+    // single input vec, single output vec.
+    /////////////////////////////////////////////////////////////
 
-    ConjugateGradient(RealD tol,Integer maxit): IterativeProces(tol,maxit) {};
+  template<class Field> 
+    class ConjugateGradient :  public OperatorFunction<Field> {
+public:                                                
+    RealD   Tolerance;
+    Integer MaxIterations;
 
-    void operator() (HermitianOperatorBase<Field> *Linop,const Field &src, Field &psi){
+    ConjugateGradient(RealD tol,Integer maxit) : Tolerance(tol), MaxIterations(maxit) { 
+      std::cout << Tolerance<<std::endl;
+    };
+
+    void operator() (HermitianOperatorBase<Field> &Linop,const Field &src, Field &psi){
 
-      RealD residual = Tolerance;
       RealD cp,c,a,d,b,ssq,qq,b_pred;
       
       Field   p(src);
@@ -32,24 +38,24 @@ namespace Grid {
       cp =a;
       ssq=norm2(src);
       
-      std::cout <<setprecision(4)<< "ConjugateGradient: guess "<<guess<<std::endl;
-      std::cout <<setprecision(4)<< "ConjugateGradient:   src "<<ssq  <<std::endl;
-      std::cout <<setprecision(4)<< "ConjugateGradient:    mp "<<d    <<std::endl;
-      std::cout <<setprecision(4)<< "ConjugateGradient:   mmp "<<b    <<std::endl;
-      std::cout <<setprecision(4)<< "ConjugateGradient:     r "<<cp   <<std::endl;
-      std::cout <<setprecision(4)<< "ConjugateGradient:     p "<<a    <<std::endl;
+      std::cout <<std::setprecision(4)<< "ConjugateGradient: guess "<<guess<<std::endl;
+      std::cout <<std::setprecision(4)<< "ConjugateGradient:   src "<<ssq  <<std::endl;
+      std::cout <<std::setprecision(4)<< "ConjugateGradient:    mp "<<d    <<std::endl;
+      std::cout <<std::setprecision(4)<< "ConjugateGradient:   mmp "<<b    <<std::endl;
+      std::cout <<std::setprecision(4)<< "ConjugateGradient:     r "<<cp   <<std::endl;
+      std::cout <<std::setprecision(4)<< "ConjugateGradient:     p "<<a    <<std::endl;
       
-      RealD rsq =  residual* residual*ssq;
+      RealD rsq =  Tolerance* Tolerance*ssq;
       
       //Check if guess is really REALLY good :)
       if ( cp <= rsq ) {
-	return 0;
+	return;
       }
       
-      std::cout << setprecision(4)<< "ConjugateGradient: k=0 residual "<<cp<<" rsq"<<rsq<<std::endl;
+      std::cout << std::setprecision(4)<< "ConjugateGradient: k=0 residual "<<cp<<" rsq"<<rsq<<std::endl;
       
       int k;
-      for (k=1;k<=max_iter;k++){
+      for (k=1;k<=MaxIterations;k++){
 	
 	c=cp;
 	
@@ -62,10 +68,10 @@ namespace Grid {
 	b = cp/c;
 	
 	// Fuse these loops ; should be really easy
-	// Update psi; New (conjugate/M-orthogonal) search direction
 	psi= a*p+psi;
 	p  = p*b+r;
 	  
+	std::cout << "Iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
 	// Stopping condition
 	if ( cp <= rsq ) { 
 	  
@@ -78,10 +84,11 @@ namespace Grid {
 	  RealD resnorm = sqrt(norm2(p));
 	  RealD true_residual = resnorm/srcnorm;
 	  std::cout<<"ConjugateGradient: true   residual  is "<<true_residual<<" sol "<<psinorm<<" src "<<srcnorm<<std::endl;
-	  std::cout<<"ConjugateGradient: target residual was "<<residual<<std::endl;
-	  return k;
+	  std::cout<<"ConjugateGradient: target residual was "<<Tolerance<<std::endl;
 	}
       }
+      std::cout<<"ConjugateGradient did NOT converge"<<std::endl;
+      assert(0);
     }
   };
 }
diff --git a/lib/algorithms/iterative/NormalEquations.h b/lib/algorithms/iterative/NormalEquations.h
new file mode 100644
index 00000000..3f40f8a5
--- /dev/null
+++ b/lib/algorithms/iterative/NormalEquations.h
@@ -0,0 +1,34 @@
+#ifndef GRID_NORMAL_EQUATIONS_H
+#define GRID_NORMAL_EQUATIONS_H
+
+namespace Grid {
+
+  ///////////////////////////////////////////////////////////////////////////////////////////////////////
+  // Take a matrix and form a Red Black solver calling a Herm solver
+  // Use of RB info prevents making SchurRedBlackSolve conform to standard interface
+  ///////////////////////////////////////////////////////////////////////////////////////////////////////
+  template<class Field> class NormalEquations : public OperatorFunction<Field>{
+  private:
+    SparseMatrixBase<Field> & _Matrix;
+    OperatorFunction<Field> & _HermitianSolver;
+
+  public:
+
+    /////////////////////////////////////////////////////
+    // Wrap the usual normal equations Schur trick
+    /////////////////////////////////////////////////////
+  NormalEquations(SparseMatrixBase<Field> &Matrix, OperatorFunction<Field> &HermitianSolver) 
+    :  _Matrix(Matrix), _HermitianSolver(HermitianSolver) {}; 
+
+    void operator() (const Field &in, Field &out){
+ 
+      Field src(in._grid);
+
+      _Matrix.Mdag(in,src);
+      _HermitianSolver(src,out);  // Mdag M out = Mdag in
+ 
+    }     
+  };
+
+}
+#endif
diff --git a/lib/algorithms/iterative/SchurRedBlack.h b/lib/algorithms/iterative/SchurRedBlack.h
new file mode 100644
index 00000000..e6a7b59e
--- /dev/null
+++ b/lib/algorithms/iterative/SchurRedBlack.h
@@ -0,0 +1,106 @@
+#ifndef GRID_SCHUR_RED_BLACK_H
+#define GRID_SCHUR_RED_BLACK_H
+
+  /*
+   * Red black Schur decomposition
+   *
+   *  M = (Mee Meo) =  (1             0 )   (Mee   0               )  (1 Mee^{-1} Meo)
+   *      (Moe Moo)    (Moe Mee^-1    1 )   (0   Moo-Moe Mee^-1 Meo)  (0   1         )
+   *                =         L                     D                     U
+   *
+   * L^-1 = (1              0 )
+   *        (-MoeMee^{-1}   1 )   
+   * L^{dag} = ( 1       Mee^{-dag} Moe^{dag} )
+   *           ( 0       1                    )
+   * L^{-d}  = ( 1      -Mee^{-dag} Moe^{dag} )
+   *           ( 0       1                    )
+   *
+   * U^-1 = (1   -Mee^{-1} Meo)
+   *        (0    1           )
+   * U^{dag} = ( 1                 0)
+   *           (Meo^dag Mee^{-dag} 1)
+   * U^{-dag} = (  1                 0)
+   *            (-Meo^dag Mee^{-dag} 1)
+   ***********************
+   *     M psi = eta
+   ***********************
+   *Odd
+   * i)   (D_oo)^{\dag} D_oo psi_o = (D_oo)^\dag L^{-1}  eta_o
+   *                        eta_o' = D_oo (eta_o - Moe Mee^{-1} eta_e)
+   *Even
+   * ii)  Mee psi_e + Meo psi_o = src_e
+   *
+   *   => sol_e = M_ee^-1 * ( src_e - Meo sol_o )...
+   *
+   */
+
+namespace Grid {
+
+  ///////////////////////////////////////////////////////////////////////////////////////////////////////
+  // Take a matrix and form a Red Black solver calling a Herm solver
+  // Use of RB info prevents making SchurRedBlackSolve conform to standard interface
+  ///////////////////////////////////////////////////////////////////////////////////////////////////////
+  template<class Field> class SchurRedBlackSolve : public OperatorFunction<Field>{
+  private:
+    SparseMatrixBase<Field> & _Matrix;
+    OperatorFunction<Field> & _HermitianRBSolver;
+    int CBfactorise;
+  public:
+
+    /////////////////////////////////////////////////////
+    // Wrap the usual normal equations Schur trick
+    /////////////////////////////////////////////////////
+  SchurRedBlackSolve(SparseMatrixBase<Field> &Matrix, OperatorFunction<Field> &HermitianRBSolver) 
+    :  _Matrix(Matrix), _HermitianRBSolver(HermitianRBSolver) { 
+      CBfactorise=0;
+    };
+
+    void operator() (const Field &in, Field &out){
+
+      // FIXME CGdiagonalMee not implemented virtual function
+      // FIXME need to make eo grid from full grid.
+      // FIXME use CBfactorise to control schur decomp
+      const int Even=0;
+      const int Odd =1;
+
+      // Make a cartesianRedBlack from full Grid
+      GridRedBlackCartesian grid(in._grid);
+ 
+      Field src_e(&grid);
+      Field src_o(&grid);
+      Field sol_e(&grid);
+      Field sol_o(&grid);
+      Field   tmp(&grid);
+      Field  Mtmp(&grid);
+      
+      pickCheckerboard(Even,src_e,in);
+      pickCheckerboard(Odd ,src_o,in);
+
+      /////////////////////////////////////////////////////
+      // src_o = Mdag * (source_o - Moe MeeInv source_e)
+      /////////////////////////////////////////////////////
+      _Matrix.MooeeInv(src_e,tmp);     //    MooeeInv(source[Even],tmp,DaggerNo,Even);
+      _Matrix.Meooe   (tmp,Mtmp);      //    Meo     (tmp,src,Odd,DaggerNo);
+      tmp=src_o-Mtmp;                  //    axpy    (tmp,src,source[Odd],-1.0);
+      _Matrix.MpcDag(tmp,src_o);       //    Mprec(tmp,src,Mtmp,DaggerYes);  
+
+      //////////////////////////////////////////////////////////////
+      // Call the red-black solver
+      //////////////////////////////////////////////////////////////
+      _HermitianRBSolver(src_o,sol_o); //  CGNE_prec_MdagM(solution[Odd],src);
+
+      ///////////////////////////////////////////////////
+      // sol_e = M_ee^-1 * ( src_e - Meo sol_o )...
+      ///////////////////////////////////////////////////
+      _Matrix.Meooe(sol_o,tmp);        // Meo(solution[Odd],tmp,Even,DaggerNo);
+      src_e = src_e-tmp;               // axpy(src,tmp,source[Even],-1.0);
+      _Matrix.MooeeInv(src_e,sol_e);   // MooeeInv(src,solution[Even],DaggerNo,Even);
+     
+      setCheckerboard(out,sol_e);
+      setCheckerboard(out,sol_o);
+ 
+    }     
+  };
+
+}
+#endif
diff --git a/lib/algorithms/iterative/TODO b/lib/algorithms/iterative/TODO
new file mode 100644
index 00000000..ca3bca3b
--- /dev/null
+++ b/lib/algorithms/iterative/TODO
@@ -0,0 +1,15 @@
+- ConjugateGradientMultiShift
+- MCR
+
+- Potentially Useful Boost libraries
+
+- MultiArray
+- Aligned allocator; memory pool
+- Remez -- Mike or Boost?
+- Multiprecision
+- quaternians
+- Tokenize
+- Serialization
+- Regex
+- Proto (ET)
+- uBlas
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index c4d37a87..6a67efe3 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -85,20 +85,40 @@ void WilsonMatrix::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeF
   }
 }
 
-void WilsonMatrix::M(const LatticeFermion &in, LatticeFermion &out)
+RealD WilsonMatrix::M(const LatticeFermion &in, LatticeFermion &out)
 {
   Dhop(in,out);
-  return;
+  return 0.0;
 }
-void WilsonMatrix::Mdag(const LatticeFermion &in, LatticeFermion &out)
+RealD WilsonMatrix::Mdag(const LatticeFermion &in, LatticeFermion &out)
 {
   Dhop(in,out);
-  return;
+  return 0.0;
 }
-void WilsonMatrix::MdagM(const LatticeFermion &in, LatticeFermion &out)
+
+void WilsonMatrix::Meooe(const LatticeFermion &in, LatticeFermion &out)
 {
   Dhop(in,out);
-  return;
+}
+void WilsonMatrix::MeooeDag(const LatticeFermion &in, LatticeFermion &out)
+{
+  Dhop(in,out);
+}
+void WilsonMatrix::Mooee(const LatticeFermion &in, LatticeFermion &out)
+{
+  return ;
+}
+void WilsonMatrix::MooeeInv(const LatticeFermion &in, LatticeFermion &out)
+{
+  return ;
+}
+void WilsonMatrix::MooeeDag(const LatticeFermion &in, LatticeFermion &out)
+{
+  return ;
+}
+void WilsonMatrix::MooeeInvDag(const LatticeFermion &in, LatticeFermion &out)
+{
+  return ;
 }
 
 void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
@@ -278,18 +298,6 @@ void WilsonMatrix::Dw(const LatticeFermion &in, LatticeFermion &out)
 {
   return;
 }
-void WilsonMatrix::MpcDag   (const LatticeFermion &in, LatticeFermion &out)
-{
-  return;
-}
-void WilsonMatrix::Mpc      (const LatticeFermion &in, LatticeFermion &out)
-{
-  return;
-}
-void WilsonMatrix::MpcDagMpc(const LatticeFermion &in, LatticeFermion &out)
-{
-  return;
-}
 
 
 
diff --git a/lib/qcd/Grid_qcd_wilson_dop.h b/lib/qcd/Grid_qcd_wilson_dop.h
index 2f5105c8..1098d330 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.h
+++ b/lib/qcd/Grid_qcd_wilson_dop.h
@@ -1,15 +1,12 @@
 #ifndef  GRID_QCD_WILSON_DOP_H
 #define  GRID_QCD_WILSON_DOP_H
 
-#include <Grid.h>
-
-#include <algorithms/LinearOperator.h>
 
 namespace Grid {
 
   namespace QCD {
 
-    class WilsonMatrix : public SparseMatrixBase<LatticeFermion>
+    class WilsonMatrix : public CheckerBoardedSparseMatrixBase<LatticeFermion>
     {
       //NB r=1;
     public:
@@ -36,14 +33,16 @@ namespace Grid {
       void DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu);
 
       // override multiply
-      virtual void M    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void Mdag (const LatticeFermion &in, LatticeFermion &out);
-      virtual void MdagM(const LatticeFermion &in, LatticeFermion &out);
+      virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out);
+      virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out);
 
       // half checkerboard operaions
-      void Mpc      (const LatticeFermion &in, LatticeFermion &out);
-      void MpcDag   (const LatticeFermion &in, LatticeFermion &out);
-      void MpcDagMpc(const LatticeFermion &in, LatticeFermion &out);
+      virtual void   Meooe       (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
 
       // non-hermitian hopping term; half cb or both
       void Dhop(const LatticeFermion &in, LatticeFermion &out);
diff --git a/tests/Grid_main.cc b/tests/Grid_main.cc
index 7515ebd1..45778922 100644
--- a/tests/Grid_main.cc
+++ b/tests/Grid_main.cc
@@ -48,7 +48,7 @@ int main (int argc, char ** argv)
     latt_size[3] = lat;
     double volume = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
     
-    GridCartesian Fine(latt_size,simd_layout,mpi_layout);
+    GridCartesian           Fine(latt_size,simd_layout,mpi_layout);
     GridRedBlackCartesian rbFine(latt_size,simd_layout,mpi_layout);
     GridParallelRNG       FineRNG(&Fine);
     FineRNG.SeedRandomDevice();

From cee363e28ca55e12621d8e513085b4d2a6273abf Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Mon, 18 May 2015 16:48:14 +0900
Subject: [PATCH 193/429] Corrected some compilation errors (zolotarev.h) and
 SSE4 vsplat and conj to make cshift test pass.

---
 Makefile.in                                   |  43 +--
 aclocal.m4                                    |  61 ++--
 benchmarks/Grid_comms                         | Bin 0 -> 45225 bytes
 benchmarks/Grid_memory_bandwidth              | Bin 0 -> 45026 bytes
 benchmarks/Grid_su3                           | Bin 0 -> 44845 bytes
 benchmarks/Grid_wilson                        | Bin 0 -> 121651 bytes
 configure                                     |  13 +-
 lib/Grid_config.h                             | 121 +++++++
 lib/algorithms/approx/.dirstamp               |   0
 lib/algorithms/approx/Zolotarev.cc            |   6 +-
 .../approx/{zolotarev.h => Zolotarev.h}       |   0
 lib/communicator/.dirstamp                    |   0
 lib/qcd/.dirstamp                             |   0
 lib/simd/Grid_vComplexD.h                     |   3 +
 lib/simd/Grid_vComplexF.h                     |   6 +-
 lib/simd/Grid_vector_types.h                  | 299 ++++++++++++++++++
 lib/stamp-h1                                  |   1 +
 lib/stencil/.dirstamp                         |   0
 tests/Grid_cshift                             | Bin 0 -> 60539 bytes
 tests/Grid_cshift.cc                          |   2 +
 tests/Grid_gamma                              | Bin 0 -> 72670 bytes
 tests/Grid_main                               | Bin 0 -> 143261 bytes
 tests/Grid_nersc_io                           | Bin 0 -> 97496 bytes
 tests/Grid_rng                                | Bin 0 -> 60784 bytes
 tests/Grid_simd                               | Bin 0 -> 98268 bytes
 tests/Grid_stencil                            | Bin 0 -> 97307 bytes
 26 files changed, 482 insertions(+), 73 deletions(-)
 create mode 100755 benchmarks/Grid_comms
 create mode 100755 benchmarks/Grid_memory_bandwidth
 create mode 100755 benchmarks/Grid_su3
 create mode 100755 benchmarks/Grid_wilson
 create mode 100644 lib/Grid_config.h
 create mode 100644 lib/algorithms/approx/.dirstamp
 rename lib/algorithms/approx/{zolotarev.h => Zolotarev.h} (100%)
 create mode 100644 lib/communicator/.dirstamp
 create mode 100644 lib/qcd/.dirstamp
 create mode 100644 lib/simd/Grid_vector_types.h
 create mode 100644 lib/stamp-h1
 create mode 100644 lib/stencil/.dirstamp
 create mode 100755 tests/Grid_cshift
 create mode 100755 tests/Grid_gamma
 create mode 100755 tests/Grid_main
 create mode 100755 tests/Grid_nersc_io
 create mode 100755 tests/Grid_rng
 create mode 100755 tests/Grid_simd
 create mode 100755 tests/Grid_stencil

diff --git a/Makefile.in b/Makefile.in
index c9b848f3..c905f84b 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -1,7 +1,7 @@
-# Makefile.in generated by automake 1.15 from Makefile.am.
+# Makefile.in generated by automake 1.14.1 from Makefile.am.
 # @configure_input@
 
-# Copyright (C) 1994-2014 Free Software Foundation, Inc.
+# Copyright (C) 1994-2013 Free Software Foundation, Inc.
 
 # This Makefile.in is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -14,17 +14,7 @@
 
 @SET_MAKE@
 VPATH = @srcdir@
-am__is_gnu_make = { \
-  if test -z '$(MAKELEVEL)'; then \
-    false; \
-  elif test -n '$(MAKE_HOST)'; then \
-    true; \
-  elif test -n '$(MAKE_VERSION)' && test -n '$(CURDIR)'; then \
-    true; \
-  else \
-    false; \
-  fi; \
-}
+am__is_gnu_make = test -n '$(MAKEFILE_LIST)' && test -n '$(MAKELEVEL)'
 am__make_running_with_option = \
   case $${target_option-} in \
       ?) ;; \
@@ -86,12 +76,14 @@ NORMAL_UNINSTALL = :
 PRE_UNINSTALL = :
 POST_UNINSTALL = :
 subdir = .
+DIST_COMMON = INSTALL NEWS README AUTHORS ChangeLog \
+	$(srcdir)/Makefile.in $(srcdir)/Makefile.am \
+	$(top_srcdir)/configure $(am__configure_deps) COPYING TODO \
+	compile depcomp install-sh missing
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/configure.ac
 am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
 	$(ACLOCAL_M4)
-DIST_COMMON = $(srcdir)/Makefile.am $(top_srcdir)/configure \
-	$(am__configure_deps) $(am__DIST_COMMON)
 am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \
  configure.lineno config.status.lineno
 mkinstalldirs = $(install_sh) -d
@@ -154,8 +146,6 @@ ETAGS = etags
 CTAGS = ctags
 CSCOPE = cscope
 DIST_SUBDIRS = $(SUBDIRS)
-am__DIST_COMMON = $(srcdir)/Makefile.in AUTHORS COPYING ChangeLog \
-	INSTALL NEWS README TODO compile depcomp install-sh missing
 DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
 distdir = $(PACKAGE)-$(VERSION)
 top_distdir = $(distdir)
@@ -309,6 +299,7 @@ $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu Makefile'; \
 	$(am__cd) $(top_srcdir) && \
 	  $(AUTOMAKE) --gnu Makefile
+.PRECIOUS: Makefile
 Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
 	@case '$?' in \
 	  *config.status*) \
@@ -515,15 +506,15 @@ dist-xz: distdir
 	$(am__post_remove_distdir)
 
 dist-tarZ: distdir
-	@echo WARNING: "Support for distribution archives compressed with" \
-		       "legacy program 'compress' is deprecated." >&2
+	@echo WARNING: "Support for shar distribution archives is" \
+	               "deprecated." >&2
 	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	tardir=$(distdir) && $(am__tar) | compress -c >$(distdir).tar.Z
 	$(am__post_remove_distdir)
 
 dist-shar: distdir
-	@echo WARNING: "Support for shar distribution archives is" \
-	               "deprecated." >&2
+	@echo WARNING: "Support for distribution archives compressed with" \
+		       "legacy program 'compress' is deprecated." >&2
 	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	shar $(distdir) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).shar.gz
 	$(am__post_remove_distdir)
@@ -559,17 +550,17 @@ distcheck: dist
 	esac
 	chmod -R a-w $(distdir)
 	chmod u+w $(distdir)
-	mkdir $(distdir)/_build $(distdir)/_build/sub $(distdir)/_inst
+	mkdir $(distdir)/_build $(distdir)/_inst
 	chmod a-w $(distdir)
 	test -d $(distdir)/_build || exit 0; \
 	dc_install_base=`$(am__cd) $(distdir)/_inst && pwd | sed -e 's,^[^:\\/]:[\\/],/,'` \
 	  && dc_destdir="$${TMPDIR-/tmp}/am-dc-$$$$/" \
 	  && am__cwd=`pwd` \
-	  && $(am__cd) $(distdir)/_build/sub \
-	  && ../../configure \
+	  && $(am__cd) $(distdir)/_build \
+	  && ../configure \
 	    $(AM_DISTCHECK_CONFIGURE_FLAGS) \
 	    $(DISTCHECK_CONFIGURE_FLAGS) \
-	    --srcdir=../.. --prefix="$$dc_install_base" \
+	    --srcdir=.. --prefix="$$dc_install_base" \
 	  && $(MAKE) $(AM_MAKEFLAGS) \
 	  && $(MAKE) $(AM_MAKEFLAGS) dvi \
 	  && $(MAKE) $(AM_MAKEFLAGS) check \
@@ -743,8 +734,6 @@ uninstall-am:
 	maintainer-clean-generic mostlyclean mostlyclean-generic pdf \
 	pdf-am ps ps-am tags tags-am uninstall uninstall-am
 
-.PRECIOUS: Makefile
-
 
 # Tell versions [3.59,3.63) of GNU make to not export all variables.
 # Otherwise a system limit (for SysV at least) may be exceeded.
diff --git a/aclocal.m4 b/aclocal.m4
index bf79d078..389763bf 100644
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -1,6 +1,6 @@
-# generated automatically by aclocal 1.15 -*- Autoconf -*-
+# generated automatically by aclocal 1.14.1 -*- Autoconf -*-
 
-# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+# Copyright (C) 1996-2013 Free Software Foundation, Inc.
 
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -20,7 +20,7 @@ You have another version of autoconf.  It may work, but is not guaranteed to.
 If you have problems, you may need to regenerate the build system entirely.
 To do so, use the procedure documented by the package, typically 'autoreconf'.])])
 
-# Copyright (C) 2002-2014 Free Software Foundation, Inc.
+# Copyright (C) 2002-2013 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -32,10 +32,10 @@ To do so, use the procedure documented by the package, typically 'autoreconf'.])
 # generated from the m4 files accompanying Automake X.Y.
 # (This private macro should not be called outside this file.)
 AC_DEFUN([AM_AUTOMAKE_VERSION],
-[am__api_version='1.15'
+[am__api_version='1.14'
 dnl Some users find AM_AUTOMAKE_VERSION and mistake it for a way to
 dnl require some minimum version.  Point them to the right macro.
-m4_if([$1], [1.15], [],
+m4_if([$1], [1.14.1], [],
       [AC_FATAL([Do not call $0, use AM_INIT_AUTOMAKE([$1]).])])dnl
 ])
 
@@ -51,14 +51,14 @@ m4_define([_AM_AUTOCONF_VERSION], [])
 # Call AM_AUTOMAKE_VERSION and AM_AUTOMAKE_VERSION so they can be traced.
 # This function is AC_REQUIREd by AM_INIT_AUTOMAKE.
 AC_DEFUN([AM_SET_CURRENT_AUTOMAKE_VERSION],
-[AM_AUTOMAKE_VERSION([1.15])dnl
+[AM_AUTOMAKE_VERSION([1.14.1])dnl
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
 _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 
 # AM_AUX_DIR_EXPAND                                         -*- Autoconf -*-
 
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2001-2013 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -103,14 +103,15 @@ _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 # configured tree to be moved without reconfiguration.
 
 AC_DEFUN([AM_AUX_DIR_EXPAND],
-[AC_REQUIRE([AC_CONFIG_AUX_DIR_DEFAULT])dnl
-# Expand $ac_aux_dir to an absolute path.
-am_aux_dir=`cd "$ac_aux_dir" && pwd`
+[dnl Rely on autoconf to set up CDPATH properly.
+AC_PREREQ([2.50])dnl
+# expand $ac_aux_dir to an absolute path
+am_aux_dir=`cd $ac_aux_dir && pwd`
 ])
 
 # AM_CONDITIONAL                                            -*- Autoconf -*-
 
-# Copyright (C) 1997-2014 Free Software Foundation, Inc.
+# Copyright (C) 1997-2013 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -141,7 +142,7 @@ AC_CONFIG_COMMANDS_PRE(
 Usually this means the macro was only invoked conditionally.]])
 fi])])
 
-# Copyright (C) 1999-2014 Free Software Foundation, Inc.
+# Copyright (C) 1999-2013 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -332,7 +333,7 @@ _AM_SUBST_NOTMAKE([am__nodep])dnl
 
 # Generate code to set up dependency tracking.              -*- Autoconf -*-
 
-# Copyright (C) 1999-2014 Free Software Foundation, Inc.
+# Copyright (C) 1999-2013 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -408,7 +409,7 @@ AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 
 # Do all the work for Automake.                             -*- Autoconf -*-
 
-# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+# Copyright (C) 1996-2013 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -498,8 +499,8 @@ AC_REQUIRE([AC_PROG_MKDIR_P])dnl
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 AC_SUBST([mkdir_p], ['$(MKDIR_P)'])
-# We need awk for the "check" target (and possibly the TAP driver).  The
-# system "awk" is bad on some platforms.
+# We need awk for the "check" target.  The system "awk" is bad on
+# some platforms.
 AC_REQUIRE([AC_PROG_AWK])dnl
 AC_REQUIRE([AC_PROG_MAKE_SET])dnl
 AC_REQUIRE([AM_SET_LEADING_DOT])dnl
@@ -572,11 +573,7 @@ to "yes", and re-run configure.
 END
     AC_MSG_ERROR([Your 'rm' program is bad, sorry.])
   fi
-fi
-dnl The trailing newline in this macro's definition is deliberate, for
-dnl backward compatibility and to allow trailing 'dnl'-style comments
-dnl after the AM_INIT_AUTOMAKE invocation. See automake bug#16841.
-])
+fi])
 
 dnl Hook into '_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
 dnl add the conditional right here, as _AC_COMPILER_EXEEXT may be further
@@ -605,7 +602,7 @@ for _am_header in $config_headers :; do
 done
 echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_count])
 
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2001-2013 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -616,7 +613,7 @@ echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_co
 # Define $install_sh.
 AC_DEFUN([AM_PROG_INSTALL_SH],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
-if test x"${install_sh+set}" != xset; then
+if test x"${install_sh}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -626,7 +623,7 @@ if test x"${install_sh+set}" != xset; then
 fi
 AC_SUBST([install_sh])])
 
-# Copyright (C) 2003-2014 Free Software Foundation, Inc.
+# Copyright (C) 2003-2013 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -647,7 +644,7 @@ AC_SUBST([am__leading_dot])])
 
 # Check to see how 'make' treats includes.	            -*- Autoconf -*-
 
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2001-2013 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -697,7 +694,7 @@ rm -f confinc confmf
 
 # Fake the existence of programs that GNU maintainers use.  -*- Autoconf -*-
 
-# Copyright (C) 1997-2014 Free Software Foundation, Inc.
+# Copyright (C) 1997-2013 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -736,7 +733,7 @@ fi
 
 # Helper functions for option handling.                     -*- Autoconf -*-
 
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2001-2013 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -767,7 +764,7 @@ AC_DEFUN([_AM_IF_OPTION],
 
 # Check to make sure that the build environment is sane.    -*- Autoconf -*-
 
-# Copyright (C) 1996-2014 Free Software Foundation, Inc.
+# Copyright (C) 1996-2013 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -848,7 +845,7 @@ AC_CONFIG_COMMANDS_PRE(
 rm -f conftest.file
 ])
 
-# Copyright (C) 2009-2014 Free Software Foundation, Inc.
+# Copyright (C) 2009-2013 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -908,7 +905,7 @@ AC_SUBST([AM_BACKSLASH])dnl
 _AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
 ])
 
-# Copyright (C) 2001-2014 Free Software Foundation, Inc.
+# Copyright (C) 2001-2013 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -936,7 +933,7 @@ fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 AC_SUBST([INSTALL_STRIP_PROGRAM])])
 
-# Copyright (C) 2006-2014 Free Software Foundation, Inc.
+# Copyright (C) 2006-2013 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -955,7 +952,7 @@ AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 
 # Check how to create a tarball.                            -*- Autoconf -*-
 
-# Copyright (C) 2004-2014 Free Software Foundation, Inc.
+# Copyright (C) 2004-2013 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
diff --git a/benchmarks/Grid_comms b/benchmarks/Grid_comms
new file mode 100755
index 0000000000000000000000000000000000000000..9013b3da11e6d41d2bd1fdf5976cc5dc80387c21
GIT binary patch
literal 45225
zcmeHwdwf*Y_3z0GqGBdc#Hg)|HrfOwB;jF#sFTOY8J$2%qNP5VOp*yq$;-?H!3PE>
zX_*ew(*CHmz5Lqr<<F`u)LM&ZArT>{zbe>jsalCv>lx!Cz93p*?)SUTK9e((8LGeg
zx%d9l4Rh96Yp=cb+H0@<Jo}v7W_Mnmo|dK<Azist5m=s0ko2pflm!e9MX@Pm%1OA+
zR6egv0yzsmgU=@K3^*o58!%gn$pFm6uPj|Bl%?CmjR8NelaYfS3=IjH-*~XeTOF49
zb<}{0Vig3YqptgJ6)0eXxG~@XDugg*(V>x!>M`AzOgARe8Bme>8_*~p%f?^5jIU3(
z0V8fBP^!gs9Fmuz_BDSq@3*%I6woGa47f!a!hlA)FC!iG{InrgGA@((rLxO-{pbl~
z_R5@AZ{vb_b6T5cw|d*V*UnyBwqW*xc}2dCqEb<A$|vvQC0B`rh8^>OQRnIS(d3M;
zdbi~3vz~szf7ymxep3A1(eoO6UcP{F#rPTZ7Ifz}xCg7#NGru}K7RW*oczpxHvM7m
zx7won-_E`4N8zv9-hJuX?^+A&&pM~yIi*Tjm37yTKHs@xon!TbU{4ydpv<%+F<(zX
ze-8>J<3A4yC!^n$Le7&Z=vgV`)1KBOdd^Hi_oa}3aSHnPQqb3?pl?UM$;$gc3jWn8
z%Jsz*{O_cohf~n!rm(}`Qt&TL!M{F5xele!e{Tx<{VC{!DfC&Fg8vIC=+CB*b1;Se
zM^n((rtn*D3j4gCLZ38r%E|n}l|ugaQ_x>Zq2~`%l(!ijOR{|Dq>z723OPSXQLa!5
zIloGg@2^tGzZCrEC}%1?w>O}Kin3f>73D?9Ctc}n5EO&{h~)3zpwoLqEc!jk-;lt+
z2K-YY$JVFwTP6R0Nq(!$w^q`Fvb;T~*(5~=Z%aADl0O%G4F8Fru=5NX(O4LhdJaf^
zdSuL<pp*YV0>6TB!{={qy5Iti6$=!XYh_zUyUXWyclli|#dXCR*D}vaug~x4s%mxn
zd>)_Ta@5zjnmt{TTwhbw+R^T*cQ>|r^tj|Pu1^1|E|0s}wc4}J)r31g*sobq=bMKd
zI!lUO6zcPLb*!^5yRy#hXtG0@rnPQYXLpy!wZ_}!?{>E`#g%pb`Az<HogT21%tf*r
zz0IC>m$%*Px33W*z4JOfp49@W^Os!Yav|rA8(oda+}+yR(WG;A%x`J!_N_A57MR(Z
z+&(|KkR0?5sO53DAq$_U%kO9cwWMj4yUXS8a(n$g=wV+5+T2FB&)Z~hJL+Ba#V&i@
z0$1&ob#q-%U6iKFBa{)c%VE_vZ#yiZYveEUcKD#C&oj?~(o~no^7t1(X?Lq9M(|aY
z*lVwBirI-q@m<u^(bnni@-Rj>!vgQh_6`uFRl0okWnK~EFDcV?MZvu7EA`?fE0?GT
zJ90x-J{HmCfQCyuJ?(8B&8)xrtCyhaP_nh|M(>&u8oP;p;$j8CU@aX+H8#Ru+Fea<
zf72?VgI@2Y#jZM+r>)b!&ei4Vgx_^^xmS9aX64$ou1-&vucO`F>P7ULVwuc$qq`IK
zh9e?(=rrG4l*aCs)a5{mI{&<;j&A?Nw5Z<{X`u|<w4>E$r}5fvY!iv&id|aba;<6e
zCMh~omcvJ->on6-I_lA|5HVie)upnII_7qE`|YSc;VYi@=9t0i>rvFtRBK7pL{0gU
zYFXz;MaX8->S<r;U*+<2b#-(Zt~9Qy@Ju)aeX6;&b`3j_Hofn_-O8pWmycM9;<~E+
zMlTAm%;W2B^N_7cXF~~5cP@4&=wn*k-d3d(Ug&R;rABjD>-8%>?@HkpK965%^R!9O
z)VWSs>GAu$ZJv&nX7@U>xtsjnj&`Nd-Lx7FvdKgCwvIKPaf-{guB{O@h#aVg?B3ek
z-gd?7>+EiC@+*t0s$6r6ij~DqM`e|(w5ViW?6$OMzJ9-aIauZumB(T8m8#{-D=HnX
zlA^hB5hW#N6n$?xX4vWCO2|N%fq$9!WyyPe9-fX>j?SB|Ocs2^bQzN+FH>X-u5_&b
zvILbh{U7hLFkeqooEQU*`D*R?>rPRS>Lj^nO9Piz#ur2WmhaqwWy?v*RKzn)_nF?4
zsL|(R{FJH30Vu)S3Zd3yr9krc34D^WK;nPewi<U?O0~q@W8VjVrm|Gx{=)BKiI%Rc
z0M0OWTZ4dHI~i*?5{z+|Y^5Z~V2q=ra}8nyV_YYl<uZaXj*@<nK_rypCFvz5x-qVk
z&N;0SjPaH9a)StHjQ6C^Gtu*U8U!{OS2=_W+-johJ|W0Tfr*YOLo5`V=%}_>C^ONS
zw-Ib6x_LdSn&{NY2(>1<+5XE-^pg!{zy=c?jui{5O!O%+s-kq7=%<?KYfbbV6TQbo
zk8K4Yexr#_KQ+Q;6Md>d1iaHkH_uP|P4v&1`0qB+7ntbVP4v@D^c^Pp=_dMsiO#jQ
z5%!ztXBb4l1136Vp|NntL_aG=Rg_^9-PoI9v}U50nfOOd^e>p`V<!69Cc3hrL16R|
zu5FEwZK7KZB4DnGevXNrXQF@6M7Nsg`6haSiGH4mUTmVDZ=#o(=wC9?Z6<nwiLRRH
zg(iBfiN4T8Uv8q$GSM4M^b1V%RVMm{CVHodKHEfJYoZsK=shO-91~p)yqT?rvi7dF
zDr(<=KTG=$M5=+uviF#na2uU}KUkvYJ%FoarVV$*SCJ_)97Q<q9^&khA_oNiQ{w5w
z2L%2@;u*xZ3;esp+4V&F1^!Lq>~bQT1%5kmb~TY6f!{=&?i1-0co%VYEs+L+e~mcC
z7$s6G@J8Y%6SoQc8sh9)BE<q<Mx0$r#47MBh_fq+<O;l!IJ=ODBJhR8*>yxlkArYt
zDRFigkzs+)BF?TNazNnc5@#0?84&oH#Mw1OwhMe3ac-AI`UQS6adrif%>qv+K7)9V
zz(4*PaCQliPJzEioLxbrLEvu?XBQBu75FQ}+4Lhefj>{2O+HdA@Mnm#sYk2=e}Xuh
zawJ#aj}m7Sjwk}(Nt{hLGJ1^Vzn{33_^`n5A<m{6IUw+#5<i#tfWUu9oJ}^eUEtp(
z&ZZja7x*`cvx!DF3;cHCY?_fCf!{=&O)}Cc@GjzPijf9^e~mbsV5C;yjl^dWw+Z|j
z;ujDv7WgvaY*G=cz^@?ArWDBqUc0-wuR+nWQ1sngOYb7Ob2D1ea$%p^-92H%zXn!4
zo~Ve|-UZF_l>y7ttY|k{)lgRE8Y{#q%5Lj*5VSGIxIf1Dw$AuIz&K;S!MHr~BXG79
zS!VW%XtuxF-8-1SCzd}^sUPiA@|GibaD^ItMGd?<T3hcJ8pw6qpv+)@1HsT}o=wp%
z0^<iQmYLOxvg5}n+Ewc8I|0<dSdJQe?Z))$_v~PX>=-_{emA3{!<Lynr0pPP@DGk<
zfL{$hst&$!nVNP$Jvi2l#5>8T_C15dYH**P{uP<N=aPq7t(X>0y$UI{^>C=`_hsD$
zQri1_SUxpyNr49p{<C8$OuuUP=&jH$_*(S9^^u#A#4=Nf_;h$x#aAk>uDGhAzE0in
z=8jBEC?Cn^o~0UmApdT{u*AbUuC2aICvHaMF8q|8?7OsTc!0JIJ)|SJ&k<Um?YQ95
zk{8tAF?H}*dW+?qffmbcz4@IKIMkcp!*mG#)6}4*4r-TL`ffrC(p<2s8r-7>x9Y5b
z+;3IGnfIzeJ((l4z!ADSTP=J~4L(cZ{%kdPSo3^>l#-1}0vD%MSbn@OaC0W%V}YBq
z35NnVPf4>`e!O=#r=h!QQL{Unp@%ku$wOI((Js}HO;ba69tbp`3SO}E{sv9G!t&rg
z{Qf&_5Z9r!{Wi;kZ>8-Gw&oZ2++3#kv%XnXHF5^kc<E!x$$*UFT_-~@Xwd_fU85pD
zZ98)b4rzz}3Fj9Y{Eiy1ZE0H!TJUWxn^J;%cD>BlQ*<$!4-C3!sCQr9D9L?qS$fZc
zem%DeE$mezXR5(>wB<qzJBwnd@s3b~DqbL!s53G&nZY9R$U3xgq8WCf2TUmJyB`@E
z8M&AXyN9w+*bykB7xpa{hXa-NUoqPIq@|PgA2HgI7~gF%+Bo0p810i7-&bO^alWb;
zExi$WaKO{%#AxGuXT@lzkdFhQHW@Uy!-1BGb{N3OkYI<%*oo}$FlE6Ge}}ay{y@jK
zmmG1TRb^)=isdVhJHtQE$NCOVeqVmAaQ3bFY+KHtKVPkekJVQPM=QchqN-)vel@sX
zJ*asHn4rS@+yI(h+M^ZW%a0>&@a0K?mwUq1Q*CG}8<9_SP_t|ssIY9?8?4T6ZK<%d
z?`^Sk4k&1f)%l%hhFieI#%n3uq-PCU12sh0%@JCA3QD6vJJA_XH%IvLTsA~9wOBeJ
z<^2lFZIeh^E>wP94Zf`gpI5`1L<82|hP6V-+_Fv2O>NR*oxw1wot?obs~>^Rd>rcZ
z4M1Ek2$qdEpf*GgF`T7_7umGS=y7`gAX1}$&?qHb-KMC6U`jhgT_`cMN!Y-$ZIdXb
z+Vrj(JgNp>&rti0PzAq5?Sri6)nLTZOGBuEXqt}uUARw|_p{Im_`mSJ<+iF?UTnM+
z^J3*C7nck*T!X%+u7N!^IV;IoNzO`gR+6)loat)xaW(MhXJ8B*&$Qfr5{!buh9-fU
zpDL5!b828;gMSt>y?r}U5%x>CS;8J2R(4Xo2B8{~P~+RV$xTS_AY^AlI4t1-2?qe<
z5_ZZ24HDK$Xp^v5LMtFOWa8j^mcH{b7>(p}{E~GqhKocGDS4q}fc48d-_5+WqNDWB
z$9A!GJA<#Q;Vt=0q^-exixnrvhBH2lMr+Wzm?E@QxK`f_F*h6lKi%*EAcOx_^!gh!
z&`&voKhN(6#q5E9I3hA+6cZBervSqPooU0=kisT-;6XKbPz}5c4}3;=V71!!c=u(_
z;LDL`k&7JPjZB3B5_Pkeyjn64`41Rx+#doTrFCaP$jH~!z(?u&?Sh)VSN(--o7MTX
zI<V=WSO-=e<Pv1&tI<C?!p?m7<>SJED{4{M6*gYz4i#2ja&bumry3ai42*$~;lQtm
z;u-$?Y4Qj@!FD42cZ2lb4Xg_2utECoh8`WS=%jY=Bxr^tRCt5*-wo1#H%R~8Ff2t#
z|J^VEJT9U1-wo1#H%R~8ApLiP^xqA-|Dq)$*Wf|je_8MJ1>G~Ze88@1ukcKa$@EOt
zruHI+Yj*7o&hYO<_QCt|?}i$zv23`ow)(wjRQm_E4BoR$t<Enf8K_?U_XZU``0|*`
z_CpX|b~_*g+QnOX%eWI*3qx$qtA_MxS76cI>6T3_xR|dwgHiO<=%%+qhBMro&*xLx
zWs4vf3HrgNhNkDig&9v0b##U@^Qu83Z*?#~A3{`23#pE_<zk(WQ@qghyQ5De&*5*9
zP0yh|uVF$K)p-r6@+g*xS<0yB33#VQ_o{(Cd4Xe@mQ6Y6N1VZbVxALI>?|h2yl1XR
z{D{yNT|ZXrY45TvA679yEsqWa#!j|u+Ja?^=DkQ7f4S@}#PlWP&#%|oNA+5bO&>Vj
z`YYrBh8_?jjdn#cp5N&_)xm#gvy#PskMS@w3uM`L$h)U!GL@J$058y~?^$x4`ODw4
zT<6R$08!kp7-asB3)y>j#VJ({B9a(smqXABl#~2FKomMnM+hw+lie#q_gUFM#BB0I
z^etjGY568*llfX0ypi|C6y_aGq2{3^^Gf>dpM(}e+CK%;R^c#e=wjW}cf7-tH|GI!
zhMMzH!<g5g3fAlNVnN4z2c*?HN!1fxuVd(OSYu0dY^{z}2<+j_LU%?N>jb*0&M}h&
zV;GnwVh!gm(>ad4t=qkyX+mqMAOyW0Po;XG9*n0-4IhrDN=<(ePn8;fJC3?zT=O;Y
z5#uGeUPQoaV`aLM6igpE1=N0xeT`VpQGyt_keJnR4#k_m*i_6h)$mrq?+Ct_IL)+d
zI*#FSWV*KJE!1=x{r?9rRHP3BT}5DTW<_*gMc~*}%cfSa>GSyUWz%8ih*c2g%0Iu>
z3IoP3ojP;jOjsaUdhKv4ERrHu13H33&d|?AiJVychRjW}9ZINZbu@s!zK*$29i!0B
z9px`|zrj@FMk+&;6YaBM<c4_RBfkYcl8XzRo0EZev?(Ho`$YD%-0Ae)qi<4xuG7fu
z$TjRm4D@V@I3-?gT>pBjbhi5;L=EhPf^T(4VwJL$RHM40*ruxKJHn3ssCGMJP!9H(
zx`TB}3a9@%Wd$hcNVThBXtCYKo@||9OSQ<Pe*{Ok;<y1t)objuMyzbaGIO>;3%;cN
zWCRK%E;DueP<Z?@^Vd1DW5{C1AeNc$GJUe{fJ#0_J41;#p=_et+${=_b;DODy3CiE
z4P3^wrv9eLCRK5-llk<<@)=)d+G4a=*Qi|>qm3^!^J27dzRVbHe3>~4P4u#i^Sv0O
zrPIjSa9nMFj5f~qK#X<@`NT4_A2c)_@A8Rus5^Cn9d3Vv+^OtP4}EFxTOd<x%Wy(a
zo!@||M!g!k1vg{ir4!bW-oFf>MW#I(UOZt1*^80{N3npkY}*Gxtu2=JeaV-Pv3W@k
z=L(pYYzAb|*N~$Af*7gv6&aym3|EM?B$?$}k`qkyTN^0`O@@;#4&yra7O^Uhtt_`-
z2im-{L^?4UFfA~1nI7tJvc|v#W;SF89ucfqU;a5^eR-HB*4LLJG1ixPLW%hGB?V&M
zLJK8bU&gI1`@v;eTdp*%EsHodokR0s()#Jvmj?HzT4R2C=GZ!u%gaxfOfE5tr23z6
znK{o$MmwYQpJ}NXOWY|FH%M43p-sYK30Yz0<>m}7H@^_`Tk2qT6l0yqy4Pp(`!qU+
zJ`2{T^u_>$c|DVBnRtQ`n`->mKiE{|q&eI_02eTvE|#g7a=|+lq-OAU8Z4HL=VJ9~
zUUe=O*^7Nqkqmx0h9s)7>XiE)6g9ywO{-4-9IiNLslmTSb|ZO8RapMWcOVQaN-QqH
zKjP6Bm%N?*DZLg;qwtjtTvHkj^{LjCpPpH)E#IIWK6M^)U1>P&XIxtve)m6GU&a!5
z!mwPVHb_`2p-sYK39W!)w@9ur>9q7g*1YbO#dI@J`ay|#B@wPR=_pz|hEqBsiO%r$
z0f3m3roW20-mQ4hBNm#fK3*)ZUVWeF9UJT=QLZbQnlqHZqC2sEtcOl9w>dy*@Ze!U
z2C=YY$6W@l;7jbP-vpVfOASza8IK5zR&hmWdKp}m@rHg&LYZY;`H{CecwxRFi9-Oh
z(n^?KijNT&389Y&4t>rv>+gEw(^sN*#?;-;yrAv?AVc&CLs$}B=f~+<U&eK6yt370
zVx5{yTjrz%E{xY!5WfW?maFJIgf?PXanmc%H@e3W7|IL$6KmLSVh(O-rrk;@`T|w#
z0f<$qo3r#i1;#2>Oi9G@^y!!B5PQrk!Nb}@+9P<_Sf1XSjAw?<Q;mlYp=9w#pd%cC
zBP_O)y#ps&ovP;5DV{e7(Fr=;1$jwVr-!wFA_}u@M`*IXINe${id^Bfc+`QxTTBiu
zeBkm$Z4|7LKa0ibVXcXJCz&2``h1blVMuc!f#GgyNG<_C9ySw=CE%;^RJjCvDxNBr
zfIH%;=<~Q-{AoN@E&(^iQB6xgUwp*fqR^Cc4I?l%^n;@KPZbL$+-a3Ef#5U>GdqrG
z6z24Jsw_-eJXIEE<gjiIUF=#}m}lduvM{^jsHVc)6(2EPm@p$y7*b?mc&l}jF6VsO
zp=?w*m^O|x@m8w^6TEQ30(a9*`Vv+_eB=c@L)4x_(IV$!35%vv9~j6~L)hMoVh##`
zC2??{Wz*lWL8rB!$L3{TgT46@i~oo|*%bBoJSgZKH$29tqt{kkS8;vCiVD~Dd$_6u
z;{j+U)`vvKTlLR{oUNkP@VqX5JLDUTjICvjy0!Ekf-H*ZXK5S36xoNn2uBFy78w*9
zH~NAW3rwtB;T{)L7F)+1Vgu%IBnYT|_%|fB!|k;Z-o3!Pmw8vlUjHw=Tkx7_Xpt4<
z{{>BVi{A7iYxR60+(U}|1R5aok+WD{LsgR-J`7p00%AUUI#+*pfjDyN2ouKQLdHTl
z$P{9s{APC~nKP^+gW(KMPBJ-zrS}Q3y`=kwb{QBWJ3ygzzI7aV9?|~!B4}U77KZjN
z?^KrVb>97gcYovE&#=YH6zh?4<OuFYe#DDWLOP_`H~)$>!osVfxCtJPTnotb%O$J@
zG@j3xo~H-i%yZxwb?`$KJCfm>vHkFr_Ex?=fMNm=7)CaP^dQ0k3F$<{`z72gVUG^0
zxZz#Zz)P(<6c`1Sfhjof5*A3v2`uq!K(ML+4C+uuXOj({)}5_h(5=9^2an@<mrs3&
zu@yUb^`8ZmS810_;kIv;s)hnRpsJw|TcR`ERO<{sEW{QLs)HY;JA?bw!l%|ugL%}z
zJ3U&KOgD0>@02UTnLTPSz=ROPF$7}z2@=wVA7-TXI39U3Pg7dK9e$XWfe0OhsAs$g
zEGTcu&($^>0?zP3AyA|SOfZ7!EGTLS#tU7!JU)F6H-oM)QlIS0))Q7OcZPqqU(Z_y
zp#u~07HM;gwByAlr!)N9VLiDHLK?BS(j)m%?P;7DNGLfuu{{35&(9ZC`M<`1+e;p=
z=^L>0o`U*xqBPmgPz~(jTcw8X?LmA^=qiZ07KOaI))~5&Muo*`HB9Qn<AL$#1685S
zhbn`ayDEd#e^7wz1hTs_l)1~YEx`3eP+TiRhy^86P<R!TA5<M(^l%kZ{-E;cqTNXP
z5SXZ9P+UWSPjFvp*_OFGC~ocOQD<}GxiVD$&?ktf-|<OhaQY5MXc8#9O%yKAc)mt3
z4`uE^mbjl(Y1~GKYRZQ!w_k#$R}&m`hF}GFhVaNxC>Pu@^FBq#oWW^8Czkup(3L9f
zS&yDyI@m@yCsR1X_lh}pD4>H-K{7G#XoV&*r^bt63RE`RA+*s!$TmUDS*+|=vUP2o
z7^8Bj(jfYVk!b=IjfQ$LZB5_f=-qMS2<77F0FD3zhR2ZLPnO;v!Ya<-Q)=KJV`_N)
zm^%1Q2Bz%MA^%J@_z|8prghK2{1=ZKJF=Seo(B&;JNA#vc;C`-CVtCu{{SuMV0-Yx
zL$_$U4-I7BNVOVv-czg=z73`?AHe0S!??IdaVc!El-VBT<w;&1<AqNJZCGKCc0Y~O
z5OoN@gN%8evMtjFc#k|f_owl0e~V@P(~k0YyEB7hBWu-gdzzy>q6SACmc@U?8dEKN
zS3D>!eBVDAy%Uyq1zd{y&oEY@AC6>Nwk^6JzkB%jC-^ooHM%c2reY4=b$)Qn=#Sxh
zV5`0#{MGko^D+f`W;)BC>UtH1^;bAT#WkVHHKFPGj?ncman5d7S}A!;K9Bvu(M7+m
z2rjxiH2v2&NiqGN;PiWgnZHDKzedK+@;`RHghQWsrfC5l`f&uD4k6rc-g?M<0<C3p
z9R~ySzl}C{$iu=z#4N-A9LAQ=!@~0fp-CqQj~0YwI$@ZC1))MG?2i$Ob%N*~1;<RC
zAUXg+$k7Qsu_VWyWwtl3(j~s5W3@W=vX05w;-7RZmvNJ>jXVbQ+C43sYqf9w1>@@P
zAXv25@$s+L5q!>pGZ(>s*7QX!y?k-d@rQfw(F2}D=<OTx7)x&><3f{oZj4993isk*
zSC+%F?Jc!%2qyw7EK@V{9Ybg)&|49M6o#iAQdYWJzSkngHf<(UQG<hy;FNp^rh3`x
z_y);-2R1QtYeJK1LQAr1%J*A(HLNDkn4E<JYG9z>ar8AD!keT<A60_~(7^bVp8I&%
zMSaQ{9J2KOi8c$2VTV%3H|jXXP2&Cvdg;6hORlZfGPR<XH#XeZaD(GxxoR#hwN_k~
z7h9%&6~8ZAEmIx2mZ@ouFecgI8!*v6;xDq~E=KyxYXOn`a+@W0vL$zEt|j*?R!go6
z$yVTZHIgnt(n)Gys6p&&{U0P1yC}Qh^G<Az6n<Rt*2p65`>5Q=uQ{KB4-aW~106Y8
zK93R8))!$K{n_3-+Ri^AR?{;64B*(T21nI2&A-r+TRjIQsV+iMs^_3A)kP>w<{Ui5
zoP*-j&p~;vn}fn!SA<J_5sKt2LYXp)P#X>YdOW(ELr2Er&msYV_e)$rNNu37PD}2V
zIX#TY-8yHp;MiKE!!1RE4#AzwAEtt2B<h{R?w*nb1g=P2K;Q#H-ljPR1l%%bSir4w
zMhT~GEz;qZB0-<JsYvia-T>m^B>j3Wol>fRklG+|0fAeI=Wd!)EZ~+oHUYQJsTF)%
zi*&f9NYIgs;6pA<r*7EWLs0&PrEeo@$`KgsuPGdaLmQqP&RizWM1`H%4WczT&`~*}
zPlNvxG!q<EkL$JeXx~9avg?WnIHa8ew#}kF*EvIrRXOQ+0l7Lu-`<XrIz!)Q=i>}5
zC^=FS#16*aF)Z<@=R0Wcp97yGd_%4yyaK<Qa%-@r54=9kAkQ<kR9iZRFh+2US;^sL
zCVn^b<^ued@Mb}aW#No5j2jDQJf8N0qaOt^u&jrm^?&@q(PMZlfw5@exq~8p3};Qj
zl@2Zv9~J0Wh9G7bM1g98|J|58oJ(8(p)T)ZCjCgnYeK-ghP;mjstEz_8uC5{Du=;;
zwpfloii38?F%0dqVugp_kQF;SW0t<pVW#2?z61^ycfV!Y<G95*SeE`cQmlW{Q9jz8
zU5%~To9IK~^%xPy(05`)#F@53YT>Ag(eTs~93UGOd$LDIrdNbAAN)bZ(M7-E$aH^2
z+Mkdd<6nB$Lfpvl3p~5X^HA_PaLwD9_EbfvdRHb$w>*>?JP4X|XJ#6P+o5!4`B2w^
z;Qnj%<*iI#GO%ZSvWRoT{|XQ8kLNdy2Ws$rNAS2<NkvdbN9giwM`%f|oT#bgf3*Z2
zVcB=XQVf^Xgf7L|XlLQa>fp!dXOF9e&#C1vS#IA+uJY%xxN=%5M;vL-TE6?Jv-};)
zcL%c%IxY5ha7?NO&wH(o@?(~7d>_nLgctn^K+%pL;L7bW2xjiiLRgq%O}Zb`@7JBk
zA^50vCV9e(?o-3l|C5PIj%eR_3N0h;Z;tTi@55Ne$G6yj1Fa@_z9YCI`^s?UeU9Me
zxg)2kgMUviIf6bDrw6q!gTs-wB0Hg}?XGCFDcT2r#Hcw3S(&Yi`L4-c!DofR^Z{^^
zW9af+s)Hr=4VH~x;^5UZViZgbF3Q)=5rrG730}H1oWHFm*dk_&7~ZGnYkQF$DA|ta
zAA_3td=N&o^$Vj<Rt;rtgGeQ~ckr#5X?qVI17qI{LG8A~kxgjG7=dF7K#^=J67eA-
zc=+hyK=k+@SYJDNu`K;d+VdDi9S`ADY!n<AK@K|4dA#I^<D7$*URFa*P;(USM=HzG
zchdeQj_<sN2D2B_*S9iYz35*5NjR3t>8n4_Qjqn_GZdu?9r(5n#PwBut2=Nkz3aLw
z!WaGu`#9JwnDRU7q!#|snRX1KAJTPsI_*hpt*Ypl(0U+h1w@GjQ0#n^WvU%Lx_uPC
zVf--k?;SwH9qs-{O>i%#*5&V6dcOhD)xpTogMp8Z%TgY!3_E9~2cHPMl7WFg6O)At
z0w1UO&kB5;?mvarWYJ;7<hOy3GP~wq8J@DOB79Yv%F$Qh5AAD6e0yl9;uGC;H+<j-
z?EM6@>(RiYpY%O*+i-Ai;8<GsG$)4H_#S-sSQB27IF8FsoWfTgI;9z2cIkh^%N|GF
zWAN?;74IMLH40M`-NCmdOnAlBb$GQYPYwQwZI#ovm(<`B+G%(JM=d|(|02fFY~Gxy
zhKkj2eSR*NGT%owJQFaa{R(-JoW1Kqw+&N@quR}g)3wDcc?F(f4QU&}WaxwW@cY{D
z@wyL$TD~`cDP?^9jJF58pRf@|aRfh5gJX^mrg-I;;`y`WOz*n<+?{zSAVx4OL9_>+
zK(^sq@~0%oLp0jC)(x%H5SEM%?MQXDHOtah0@-RB=7n6tIYKCc<6;!ge>NwAldwjc
z5=_VAm`q0yQ4ulVbA4J9!jp>N0cRoJV;Dmxwf7+;S5!-NL`~ak`R)K6c`$OR<$HtK
zgX#`?CBkK@#jfE<%EPP?|Ai(xrn27<({-cAa#GRY9%dtystF!<7QW#K{>3rKGs>K*
zeuBq;(9j7Dv)MhJBFgx-_T?v|(Iw$q9?oQEh(7Bb%>tu1f}hW?;iiF#*Cq5xF={ys
z^C6iw8);BOEdbnMcnUKMlJ7vm$RH5y7Q{qWpbU{=mNS$!vcL+PV{Wa6PMvuX?s;+~
z=*$G3bvkjxSyfEl3$|})Kr9Z8tKqfULwr$JEq}VZUc?>f1%~Qs&~8QsJWpg2+t2%$
zScqah3--0zR%nL3S$c+fIujTe6YB)d%Fhs;!Kh_08%Ga%!H8!$1S69}+2JoogSl8=
ze|rszBn}%OLi_&XC|@%vI1CWV$~Nh;C{CA9);t+^f+G4aL!<dj>fCl}OqD56CBZ^p
zWsyT!x6g-Fq~X`|K81yDq6F08G$d=$=X3hoKAcXf!6)HM!4J{c_0t+Jf*T#>e5_wB
zQz7$W@kUmL8pJVo8(gFcE3_)H2+Yz3hHxH+gOYZap6bUlL@Gz<GJVjR><CU~GDmQ6
zHhwZ$5XpwJqVNqUfXzt#@#J#)12%ni@5gBm#O<nNCsw{61!&q^6VBhq2b*L-$Ku3M
z%G+{S0z6NXhd)eyW$9alMWwU+HB0XwnF}QM2w0!LRu}sU)Sx_)rMx+i=PV!gU%+0D
zH|L68PTy@NFPpvgOC%gQ7p@WcFARi~yb)wZb2cnGKCMp(*LU(+m*mg}SP<x{vyZm{
zOP>l|8PfwcbL+75UB?(?(a#th+Io|ZyU<Y3i0boK?NVu=Jds&_ey1!3-^LW4o6y!Q
zeOGhZQ2uvI?+%gCEjSPgF}1pwgZn7DK7Y9sa}LDN2-thktua$r1C5kr<;H~eBP1D_
zNmI<ork*m66|39fRw2`{1CHov4^vt60uW-hz)zu>c+eS7eu!jQeD>RK@eL4Uqf$}Z
zuB9yg8B%Xm>g~nevn)Q?%$f8-FH!s>C_Wv#Q1(s5z=is`l$))~9o(zigF_s>Z7TP)
zli4GW_85Z%?ORBq+v6sj-=(4=ZWM94J=%oKaeL%x_fmV<1B_->--fCrLm@Lc%VZBV
z?9aDqFJPXfhHgP03+@6HQ`*2-rlqeTOHqchemGAyGO_!U`xsgE<`yrO`-td8*(0zQ
z70<WKotCAS5BJUe=!L-1j}k<FflEC1uTH>y=LGKl6L5znaP#$tRGO@x!2R_TaNjV2
zn_nGECA)b7ck2nbzcPW_bpq}yCU7r10rw>nxa}w4W~G`2x`^mN9?0i3NBi}o80h{2
ziizPxzPN`jOE84V@B}C(hMn9#Fc@0Na5pF>hWn417_KBkKPV=Kspc$Zks%td!^vib
zlgKa@&v4Mppbdgycf68AW`-d${4Sp1$L3PpMTT&EDZXcB=p#b_l*m;yKF<gq)&`)M
zy!{<@TqvWG4M)9Cup>9k11!C4;;EeNuZWAh!B)u`NMb+Ud+P*YkDfpnTVsT4*`&Q<
zZ}9|SeAtjmvkizF!K{5x<d?V;5|fN*y<9cM7q{^QLThwUkzcTTgXsN+=)`fN_tz8T
zaw?Zf*$%{w)KcOb;}YY0+TH^bB;IubiQ-rs<h===-B)4w(U)AUu<k=&<7o8cF5wiB
zkZ$kDF$uj$U}^#nGgmaCLx|SCv>V<-bD9`_C>e4AwMk^y7|+1IiV_AW;WgwMS!rg_
z=a9pU+)tH~%`ST9Q-;YY<LAok++`mhH&>2AD09{*d^K}^3wJTy^_=sNSX&};=6fgR
zQLJ|t1$^dg^=Z=s;ykbR0k#}Q)+OZ0S&jBQVkV57Cz^vNFQ#6K?UcAtNaPSJhSA3#
z(aqV<Qkcs1Kh5KA=8?}l(5JM*Vpy`q_9x7*=X3Vy2;#s()Db$nrs+#^et}i$u7~M*
zxsb6WeBs|YhrtRS3*EOh&L6SkMT5lqecvx3t)~1<Jh{L$rC9qd72q<h|5}@ZIXO$7
zi657S1-tQFTPb;l6VQiI`rVjZNc29S<Jw?RIX!>ZlT;>vh%3VLgfdt&9M+y;+lCt8
z8K2Q~FLs13tq5JTliLyxaW6u<=6xu#;V~9AwxYupj%fuoZ@jtVWhg9s3r09c9baN4
zV~Dwm<=d0wT5mtH(C&R0{`<l4xbaCG^nim4hHsMLY9txCSoC&@h7=EB<Fa2mQNJoA
z9r4TbaTf5eh)X+nV%jgoRKKk>*5E(bsr$<!n0Y*Ynm#A|i;SF!NSgk1K)IjG&ph;_
z+^qCmZ5PVMO3#*+z8Nt_rDvNeeIEpxD*gAAzu_^;ksCHKe@)!K`B%8uY=SP?*k;|1
zvcIQVW;ye7382<{mak@Alg)#U%Ln<Kb<gIj>d>9PADX-Sc{o|gfc6OXRB-emclVn*
zejo6VFMGG*5z+q%bazi*gR+Zj674CV!M%cjw$YFLP$hkDtv@5SKZX^rc8D?qd$OZ{
z>>e$NY7wBO_{&5u5ZMDNo`1&LC%0tSHb1EeepplZngdTTvG20J+Y!bje!M*k7x0a8
z2l>m$o|`Oe|0u#3X_rWv7_CEdNtz`|tQ)Y2&F2}Cb~|WztYcNfRaR^^-u4ryjn(#x
z*l=?e4r50ugO4oz^KebamR(>hqkFj{G>>m?YR{km_&~{(NRNGZowiG-U1X-+qtm`v
z6Z~N0Y;KK0{>W+M7_s1*J8}}Pc_SH!cLs;Yn=DL*{V`#m`np<d+o9j83E!BlZw+8W
zU}$k3)W<_2R{{2y7Z-p~Mgo>w+H0sS2RjSL1$avlJ8RfJusK3Y3ml=Z;8?;F`V#@1
zIeKmlKDlyO``+v9h8EjKaA*gc!5bXGrNtFPlOaW^7+P$lSbmMA5YP8Zo-y~Yfkod*
z>poiz%^k|TLvY~%aQUNMnz6fzrO=+;)8I_I@^RUT4Fct%D{-g&{Xy=f;dq;T#qqbq
z@xVe15qANhjei8&#qVeStM>3&<7;<+H>RAG%59-?mm`1O{^s~X#=pw=^OMHUX8iIN
z#7~OFe^z~lXmZ$`SL`cq=JO)6y-i!qiRT5x^PywT@)3Lo0*Re)|Ip3ZHE2<V@GLx@
zKbG#mAwoyUj%P-ASb*);^1W-8;WIPX-^6$)za;Z#ro}m~X}=e(yQci8pDyAI;^@pP
zniEt?(d+5>1FXdB(Hn9v)Sg=t_GFK&b%eXoBJ3Eo9>o)JzO|1NZfbeNcMdn7%g6jB
z5T?ENJJhkRqf{$<<Sdkd+LX%No&pb>%XunJrtpZ|aU9k8#F=))QFzpk-LxTR@MG<l
zU>eB)n>zSfx+9|v<MoATp)nVf=S{UVR0{i>V6@_DbTt)M;Q?4hy)%4X?iZ~1q)>Qv
z0brf8@HJRu@EBVDo7vh`zl9N?(Ama*yEFI(bgPDp==2xSn$y(s5xh@{)_m>t#@SZX
zSgb#Pe-By?zjX!<+@cr7Fn)r->No!``F{|UORSZi_NG;B?ylANny$5}qrIuStIN~w
zx2|%xc33@Y@v-3cl^(0N-Ktod@#)fbA3j*>Q><2})vBCyl45oGaMQTXkIyUDRL=2P
zyW73`O{2Hjn<(GsxuF}vz3$c|^4EBp@D1IC3&rP@Tio5PewVuw-|KCTH=4D{rW0nI
z80J>Vc%55WGH-F0x4Dd$N_?l-F3B;V$Cs8=;rr1ZpV!@9)zQ|5(%`$l9bMqwe0zi7
z9cZ)(sPmT;Q((RLKC~{<;2V{Ef*F_2-Ue~o-)Iw%7kb2SqYYQwW3*<-mgP(3Edw$>
z*KQO1RtbwGv`Lt}{3Ua#Z>)?aE9CYz2zmWde$38-s(;gzDjGkMdwPp1+S@O=1mBB=
zRPiC}qNXMVZshY@@p0~j3-wA}gpcH2Y9-IYg}hdKME%xz+nNgybs_xAU#QHU-PY;F
zbtO0e@J(n1KmFU=_%yHgCQm^j^IFs4Z8iz3b$9tZPWL)|9JzqswzkfiB?MkLQG@f7
z(U@_wr?GqGY<!UypKSJx<NjuTak<Lfy>b=GBVw)RHJ=Rk$A@~Y?&jt$kI!d4ueEh8
zX)-k7`_h8T)!Eh21cr_--?h!xQ*(UhI_6SygvrblS3AB&PW$PGidB-?5f?5@N-mgS
z?DGrp^;naok|tZYunz63zGIoEY0c$b9c=|7fnmNlMAXn{l~>6xXUFNrviO#<8}RvE
z@CnpNTsCpY|0qx7(Anzodrq8fp^|o1#>M>GC;Vn^|F5FaVcc)t9*quU;BTe?=1#)*
zcYg!=WJRe3+>UtL{n2Ov8s_GM(P#tUb{w1Iw?sA{ibnSXHoO>(DoEe(68Hc+0d0Ui
zfXe|l19k%T18xL7a5x$r1uT9A{8&EN0BwM^fc!4h0ALT`FyNhld~fXl@pq%q>@47T
zgl_}f49I=!e!#VW+W|KN4glT_I0`rb*!f{JdI<UT01gB1KZ1Nw&h1CR4>$mLH{b!l
z0l;CvLx7`zqkzg7@`XLDfL1^oU>RToU@c%LAitQ|1K0z&8SqZPe!%U3+X43j4gfBv
z+>fHsG2jg!qa3pT!}Bu{(r#L&q^-?OJ8R11?9FMDb4mX^es^w*MrY|P;+mXaQSv4K
zj(edy>PE@Qy*wxH3d<=sX7?zUoq6%Bx%uZ(0A(^{+*IonIZn>tpBl)}^VPqFOj)_`
z2^C*Q%AC?+Qz~+DD|7NHbF7s)1(i9)6**;#b4E`JOx`ppob|2DZ)bcxU3^3NI;1Oj
zIvSmhvSwD_kStwwPHkK=#VAikT>16Cb3<Ln{v3@y1xOtNDAzpPK8D}Gv(f0&sN2lW
zr1@0j*y8hHnYj73{IAjINAT_OGF9Xh1dLMX`EEgU@e9!?#K!{lybr&EBy{RyO+tSW
z^t=Q*_5T3BT+q*n(<5fv3A%$_*ohlZ`|pD<k^p-UP_JsxH=~Vk(P0GUa~<f1kk41>
z4-@M!Wj^dD6tr130V6P<`#>N2TRfe1t%v+k&{yF;^S@Cal{vW;IeFDN#es}(F>eH>
ze-r7;AXj|v*;Jofz+(e_3F~uumo15|i*p7t;wqHo`){OM{&F<>bL={0u1ZK(kyE@l
zXE;4DY14S+A`R1JBA)|qV#ba1nFoAA-<W@kbUh+nIP+T>-<JB*7w^NphR<d!!##Ce
zoU=WRKFPrHQJ(_5QAG0@f%I2Fw<e)~0(xE&`gG`*3p&r|DAhSu0_J}q=#+mW3CJHd
zuE?<l*!+|WL2@GgPQ;%{1yBd(GSLp^@=&i^$bU~X+J$?{BcPo8O!RG|(P%p4Wajxz
z_CZ^LeF9mVVwRyjAA_6*yel<_`9e-CU%jtL>1UrHT^{^4(O%P$e{K@`g`j6Ap;v>h
zB%xmi`WVWQDBlnIDClP*nNe=qW0T}hr2kyXPozHrdNO_fEcp}jeN)O$q-S!>f<ET<
zOMUn)e+_h)DHced2l{Z5ayda)e;MCDRFj`$5A5Ti6!^ur)9pEV-^#G(Si_n2oPteR
zYEJRjC#gAQfyp&Fw$_|7)G=yW^!MoXQJ>kuzi)w@Jk&GXEiObe?(gkUGPbeUiqnYV
zjf!9H*|==MuMs~H_*DHy2pMBc9#Y`|f%3C2;Nk=_1MqGl_W|K3d7Q`&<W08iMD_wX
z4e)nFiZG&*?MoJ1mYt-t9i}Z^K*;tgk?#TF?)1|{-UG4=u#3nuCo9TXfS;p)Y2eWS
z1>4zE^go(+E@KLSaOXqBWFsN0t=uEhNF<NtnGJ-MBnnpnq}{0QZQwo6DE4(Y%7%RP
zkF|dv$a%O=$AGI$`J7&!G9p&!k4R-Qky0S^!6`(34ur*C!Au`KU1vLw$bKMXn?>X}
z5Ek1?<VtL$tpi+0yL|#=4d7xTci_My&AX3mJAe!WiZl;H6ia#?WA*~M(1>{weG+4y
zV$8D=u~D~UKt2H!HkgEHvI##p0OX5COf;h~<4Zu4r+8uDO#|FL4Hc{uq5l727%!Uf
zTn~I~$t{RrvH3JE?J-=021FjmWe#FQ7?Ux-XWQ9{3)zM!c{eVirz0P@5TY4-Ai9FL
zESw0up+WF9IQ<Vth{_^Tg&(V=0YBv`{7_$dh-MsshP<zp6mj!=AnVy?xXsvX#PAQh
z(N-en+d!z{4*(SozdS}ELNw!*+o9aybk-<!u*2PjSR$N-ia#;M*lUUG!jCPefyhz(
zQ2&aZ7R!N1s=Am+DG)f2;$S{s#)SerXcY?eV$4#qJphEpvQZkXP0zlU$RQw<#(A!C
z3QEVAA2Q8MAWZX9A~qm27N-cxb+}NSHs<4#G53(|w?Ja~>;WQL5Jmq52=h5c<Sf)3
zW8PyL{-rXe`G80*5a!bisI=n3e7c!WK*oGRwmlN5X3Vodo~Bp^=>u?FkpqvDh`9pu
z4imWtKO%hOtPJ8u@AxATI_=H8VTfjIRWi<;i#zU7|Ba;8_<f%@{|^7Qz`rf<Zwvg}
z0{^zazb)`@3;f#x|Nkv8kY^K0*iP4BgM`Bp_RDxfQ)8Z&2rIEzVKC@i$(nKg?LG!9
zB1I^eZWEBRNQUWdn}Aq)iD3MFbuLU9tm3NA(|N1i+8`*LurhFV%kbgxXb0h)={5qf
zsvzvlwh<^?STShnHUfPO3ivfHKM=T9WWeH31TF>40dr<O^fU;F>A48CvV5GLGZ^z~
z&chiFS!@CtbVcHZ-iJgi|DvHDjQo#B_43r_*hB=*SBdf&dSDww;Lo3`$Lp#8?_Xno
z$37ve7Fn-rB>cLBcSv}bgbzx%Pr~OU{D*``B%E}bUalDuo-g472``uMY6)8;TqEJv
zCA>qzyCi&2!hI4xC*eONJR;$w88Uwf&zEq4gqKTrwS+Aau95KT65b)<T@pSh;XVnU
zlkguB9+7a8Y=<)>JYT{E5+<0Re;#z9wcx78?sk8-bzV_fQR(cG@@|2a6zT*;#^Uea
z;t;?Z{OMXe<hRZ(nqO2rd!Ei(y58T_?eniIwHEl@D}8g8*`4-^I=gG`{Ngz+_(Q<0
zh1LT8`QK(MQaYVi*XYR$v1x(7VCr`_0{XjjxXQT0^F2>jr&843;rA5b50TD>Xo;<C
z?=EWW_O>?9#<SoeVyoP~RZ3Cwx^~FZp}$LykqXBlF2r?tTHQ<_?>bxkN|6_TomeTt
z-)Y4adV3Mq(d_oSl_Jk7S4)?>&Es0tjI;&@q4>kFI0xbB#Pdq=m!gFXm%FPAf5TPC
zFz&zBgoMb?-R5mVx(;N4OOeP|&!f@jQ;PV=7l$Ac>*;@rvp-}XYxFy1={8_kZA*gD
z9||HaMt^MJU&IajAR~Q!x(%4zpa8TIF$LMJ#VCA!YEJJ@w~08v6qqlt@k?U*YDD2N
zWQUReo#{5g%)gbuP^*Jjzisr7lB{rhiG8$@-ss2ql@f-HvL9zZ&$1iojsEm%(Aakx
z>5XxrQ!*Ok09ZswwBHI)*#{fxjehreIgqW9>7*L5d<=eWhOkdJaHC(Z&CwHXm+7fL
z^)&RCWaTE@(56OuV;oT|I>#Z&Xyk9CH_G3eklq-V4$1V!IL7oGs}s|I3pCoq7)On9
zykI(r2;4_Ahs631BGR0`R`&C?GQD_qYZAa6E(QZ`!M!=XF>kS*F4&Z396B_vMmhuk
zA#Tj+jd{m(x6WaVzXpGz{{1q&VSlUJCRnX9J)gsyLt_4SBa-<W>9@Do1pD?D0mSB~
zey3j(^M4Rzb9!}^O)#pfbpFSTiqX+n{Yg|Az>PUQA8#{E5wi$v04M05nEx~2rK{36
z8H{<_{#$k2s)N|o@E?Ql5^m^fMtWl&r^)n*{E7G*NK5^V^!$r`45Obi{X5{LiH!8d
zJZ4O$H!8+3qmjQ+k7Gz{&foYu&||kD0>R+dv2om>A47yWz0%X5v-aqKFRz$`k<NgV
z5N}SuaYKV(-?%|fW$0<hFwz<L6eK3iD1X09f2T}uH74jXfz?PT;Z&r@f1Eao4@P=v
zhFBzd84Sp0-$vddLS9b4UB3~}ZcPzJI)g;qXgdataO|5priq8^igGqk8-7OljsB1A
zJ6Zn2TM>cqIQO_sAu)eb)P&m%l(H<{mV};;2LQ?F8HztCJu|j`F$?40Ue}WHPl~N?
z;`t{l#`qLZ&sL0aC!P-L$aP6P{p8sCC7ynYV)Xy<^eKwbugB9*RgC^Ro}Lq1|HacS
ziqXHu({mN0-;Af@O?=rO#?z;vf3ulM;ywcpG>rbs9GQVfAV$9wPd^R)OoN#u?lbVj
z$>>MSkr{XjWAqR4ba<oD?oAOH${C8$o=s%<>zT2?D<4lcR>$L1`fDZ*1B_GjJAHVc
zXoqt|^fZOXmCS)paQH}x$BWFt4kR7VevEpw%E(5*H0asf*B~fHySV{aGJOpBJT7An
z>jg4R;V~F<*evNh=3)*%m2@6sF^30$P|pUrPU@182LP!jU!h_!`h~Y}PdeYPVle1N
zxq2ksDAy^tNmeey&OA0^4t(H9eTJnzRvBr?=bzVQFyvpGLVgSAY05-D<VS`mzc^vP
zhU1B_LjwJONqT<*{Wn7XnMy-KJq`%^#CGwzlrxYZ=Y7FHv3;c~*R*t&YpRl+&`zy_
zKC!&CRWkcuDEKF~uS(GQ05h>&*GYO}KhPrNOl;SjxLcF7)0;r&1K7mCH+83>Uiw$q
zL=?;-!t*Kk-<JHjE}P&p`13GmC6j+c3Odi5Cga}<`Z<_^8RyCkKj%RK(zBcNd~-$a
zct9-VoT~7zJ2OB%5e^HwL`>j_<W~~>;TY)WpnVzhjUJh15hj?)^mGgQsmiFd+nJJ|
zhhl{OjW$8y5dwy%LAQ##_{&Qv=<f;n@w~)%7KwUVZ*LHhwc@HMmxG>6&udc9TT{@#
zk%GQM(9y5Tde4#hzMX<U8xz!Nn28(f7s;;V2>Pi?XM(>~3A({G4zEu^zdZ&0C!o{*
z4UIM-q)du>Q1GL_P4L?^Y!=Y}I~_W|oSrGOBz^O8ozAm}4Ap|J-<z*1KtDGf{o7?W
zK{objyGduiXV>X)GZB6%<@CsXJV{ob0^MpLCin*EEZ1<OO;8R0KNS;5%D0v2be`>H
zuz@}m{KotQ?j!<_jj+76*V_cnGlUGig0A12uRjBw`Sv&31lh394$!TF5`TFH^xP!<
z&tW0|ROLWIdzphNT{6GDR?tsXHYfPWM$koh6YRM|@)sx6?^`M4pK`KphwTaaRD;g;
z)g#;2a#_4JlD{WGpPvbS_*;`ra2o!`KO0UvTNCw>^kJvY&$F=%=K!)?#(Og23^eF9
zpws@NjW%HV;TJ*==*jHX2l~|XGn6|wG>|P8zLSFgH{?fohh=&BnF5CWDfkbkpnpO+
z$aix>z9(V5CdQWrn~>`fS4BBP(mSu!>2oCgLW5t@Z<O@wKo@=}?JT!blp7_#P4Y|i
zl{-Ke_Dt}T0l^QUGxU6=c;&F<H{M?|+Ue=1N`I5>n&S<_uRv$HMpc`@WwTTE3%Vf1
zUCu*?nXJARf=)dfF4g5cB~x*~jdbgEIvuD8UfeL>;cImIGD#1EPWu?|Ng00lQz@rY
z`r#aDpxvOe-aGpmgao61cuB~YLQP;y(2;LKd8cDSnF~3I{$>R|S$Qu8o$X8M)$=vl
zS7!?T&7ji`wXz=1k^1ih-D=X$xP3Z>oYw^%auWO`8yzX_Su5>h)cZ2fsb@oio*O0q
zW`|8=$usc`{C*nkFc5AKxN&(*${CPydSt|LDW~Uho1lny0C0B}I@n}(D^5XQ4m#~q
zEd9jTpSU9h|NWq|-uW)Q2<Vp?{*;3MBgt=*{8qt@)3#7DnVuCX=q}JXE+qO%i{$4^
zF6QtJK}WktsIQ--kpF<>PaM}GDfrVqmt6muf{uI>`gvyxeoqSeh7|NIQobVnXO%4U
zE=eEF)9t`_dl>czdXjyN*!y}u@s3_qRor8tVm@e#qu`~;1j;<cwb;3|vcl<d`MMij
z4p+QL*EMy1XPL|8?GR6?L?rK$Ww&60@UjV%@}$`r5?$BmZ?HN`7Kogh^>;E|cn8z2
zj7MG+7uo9e`|Y3hHO5qhHD0QXz3<r6<@fo!@$Ph!;&NS4wan#o)YZFOO0%a6FA)0h
zkkI9Cb2YW%^+=zRgyeEHceqxzb~L(MUCnq~*XMF~ucemaMZo5w1?3l&LUIx&7oP#@
z&+lA#CfBu2(Vqb}cek~zLlP4v-foR!b-6BIR#9Vj*_Tv9SzU8^FROX!<(Jpl>s|E~
zl}<Z3X#1Fm=6Hh7j~9;>SM^tyRMa@C6h1I?NgDj3wAkgso3I@>x>`N$EB&inp02Kr
zuG%Y`Y_`QtM`hLW<*vC!@Gn`ZDPChPnZH<Asjj>2YVo$Uy%sNNcXr`HY>O1+DlIB4
zn*Xl^T~vwZpI%RweGNJ5{3T_0`0H(g!jQERSC6~R(FA%4-^g|OyWC#C&(UPBbK_#K
zDzPuSvd)L-xh_YC&$Se9dAH%M*$HK->uB}az42tcdF%7|6Sb{($oH`m-*0wZ6N`#_
zRomri^)@!Snta{D2b4y4(`u;MWGG?S!Ki}ztD)|Eb156UTT&M>UT3!Kk_AQ$HoBW#
zO>UomO(N3?UqLS|jel3yc+b_otj@ncf6dzAg?+ttHsAWX6AA;1Yh_zUyY6B1-YeF)
zmdVCe)#~>7(9$}){k{oyu185eQ)MR>vbnW3q2Qji-o*NJQG+hOtHteYO;lmYm3994
zP5yPA9{3VGsK#}px7pJ!8kap@bWuskzw+dg(z;dN7Qd%?nY(>8oXxw;TUX*zR(kw?
zeD0v5rP;kMK>=}W;IlWh7B6hxwAStF@`#3Pv!T)A(1WX_Xs+If%DiX`JVY_ifqnvw
zB)%|6E?ic4q^r&CZBJ0_n)<8tN;iGv;9uxjT2gDgqh8g~(ZzSbX$QMEsZ+@&R?Z=z
zs;zlxryu7;>}%0O`|P!~O|ToAb<E;|Pp_6*Z!H+40o*u#F|o$Q;S;or)>aof{5G5^
z@q1#T5<Xi{I=5<-r)jmPtFpu0g|7+t{A8yxURaqOQoJpzC)(0cU*j4dU+QbBIJ(q}
z10LgDMv889e0)sMtEx0fYpk2+!hgl(Sc=xx>}_{-<CsjWWGS*pDrvlva^Vn6BRa<k
z9w3{&*#lToW6+E*rz>udIbnxUQXb#?h&TI_6t;Gaw{E^m)`D*y`jpO+V(~=}3<@3V
z*za=)ka~Z*QL8*QXxW#2EoQg`)0bRib_vm@;S9PW;>#Z?3|%s>va7@0%>MW*9bK)>
zb}xz`e8{^1AAIQO@+36Fe?2&s%!{{`okl6qje;f$d!Qd!>u&U}DJi)K_v~(6?9b3v
zS~_5x_>SR3+M7*7pE7!Y1mlc5Yzixs$BYmwzsMF72DEk47dF_=lXMQK^H{~id~ZRA
zqrKU)Hi?%|Q|YPkA5cNtMGtZjr*3Xoe&YPb1rIhRTxjP$^papo;Z=41c}@6$N4&G)
zGal|H_7#b3r_Ns<>v$&gyv@E27e0m4+=@PFeCELUT&&y|p<Uz4C(ZhtNUyb0F?;r5
zCMP}~B1Lq;$mO80x?TQS&Y@!Fm^goty{<5;>xOQx$JgY+)VbN?Gv$;};U)9Tt-%AY
zmVIw$cbCVt#@prZc4J^o;%JG2O3i}o$)+0UZA(jCF3jp-4Sd=M-)U*aoC^JTyTjR7
zY`2%pH)0KP?YJDhbQgK+(DS>y982ovx-RnSlT#q`U5+}xaEDgUT1T_p>78r0<64TV
zowLch3YXp9RtKN3ufasS7$Z<UhHqooD-qQoKQ~i@l?`pJ&&4|QSt{BbXBx?7a&+91
zx%!k(ayvvG_PPbG+AE={F*nisv7&MXlhd}2HHmXbIVq1djOtRo=EkicCvHZB?mWU3
zFi7FkG=}5INe&%GPIzNNMH`<W(Ot(b*K85fEVztI3Xi+lr>FJK>-2b5Cv<n?rW9R)
zG^La4K-kiS6OFM|%96T{`7N#8zEu-NBy@XR&G@|kvgb1m<FrhCI~K!y!o(NdDMxkl
z$D{0R&{BRx$hF$D&eenx>obcwZ<o)$EMa+JcuuLX6TCWAi%ZdymX>e=#fenh%qm4E
z7(W`6mbhGN+Wxg&Onpmi8s=#C%O%~iI$yHc8$YDQ0RlcfzKp>#EyXe>#SpCz2+|Aq
z{U*-1yuQxv_9mmBHC9gOo;%7>AF_cp+hK!_dRJU6qRFs5Vrhg~b+dz$NFNr=b*21|
zrd;eo=dz#+9dAdQ%iYAUDY?AprTI-R9LHTLdo`2|Ge>{Zs-#_7X^H5CV=Futirj*=
ztH_!ou)eV7vJA7&j;17zrq5*HwskHKK1sC>6RJ-5eMgsjr6<8s=+_0y#6_x@{WtQH
zNXeA-h$F&SmaSdu>ckQ34jiiXBBwRQNeiX-(=?mtaYcvHeq&plbz^Neg|BiQKB31@
zTvxT<h|0j!&DY(=t_-_abBl^849u{PyVa9qLVe<$2lGUwMTqK{i}j^_;{HXQZ=UZ)
scPBJ6X18*dh`zO{)3g{XF>QB<#u{r2weEGTnA4fpBwQ{jqW4t(H>p34HUIzs

literal 0
HcmV?d00001

diff --git a/benchmarks/Grid_memory_bandwidth b/benchmarks/Grid_memory_bandwidth
new file mode 100755
index 0000000000000000000000000000000000000000..632fb5eb4123b0f247951342fbac8c2b5e1a68c9
GIT binary patch
literal 45026
zcmeHweSB2K)&I>KM1|adVDXhzqb33|8v-VXx*=J(D;tO;U<CAL*=)#al3jN<5PX5)
zCYIYZ%44lnAKTK}_OX@LJl4koDkQuG^-;k-TIw57i@j?^;u}N-_V+z==Wgz9c4Pbe
zKEK~z4(y#dXU?2Cb7tn8nYk}nXSL5wO-T{-kSbgw5LEdC3qxsFLC6g<kmHJl(*%n!
zSvXG^1xyBh7M4S?meS)v&RUN%Ib9m!bo`1_HAHc$g(dX(JPnN+w4h5!G1WH|EIg^j
z0j~z?u^^Zjg36Js`|e^O;#!u_<8BfNVNld!y&Q>0<pz1VL0(Re1unlH>+Peq(XX24
zSEpJ45p1DIC}-E9m}n_f9uCw|y3ff##1@v&;|HlB2zsoSyB_68&VM#!bHYWuzGQkC
z>L0BDuU@#K(Y<13(G`t#(;D5)Ev?g9i)T)oS(NYh<QKAb6Mv$ew_q_VsOxbOAd-0k
zex!2BzvaHm`yW|)Z~bdu&VTa0U+vjF@$CmMqr80lhzGSxab(vzddgD>SAgGi{PwLs
z{q>(YU)XbBQ)J(#+1LKO^G3^O?=5@4lxuy%KH=dpWkN&7Z-0KCcgq^vt<QoyVc3I;
z#wW;G3?UQY{|5pk!heMZCc?KQ;qzz`d`1%f)05DXo=pklpPU3gCkdZ#ljwm4W1@D=
zOM>r0y@}|9N$5XJQt$R8^vjdrX*`(liw~QV(C<%zUkC=KcqV|?;+Lph+miIxe<Z=v
zyiLSsV-k7JPeT9eB=iT8$ay>oUP_{WH|S0H#fKM@$n!=Ld9Fyp|2;5Hqz88rK9MBz
z>zpL?!6b4nOCrx*N%-HHgue;&7YgSHZFko)71eM>JqG@S7usX-_j3A8>ot5Er_bQ@
zwQ=-wK|c<BEFBuXiPNv-^d??!CCA^&+dDA9!ZBL3o%8AA^x2@J@H6~K|K5ofT!+Fw
zE@wBFr;TSd08jL$vn(9LBlrP-pw2mE3fcadf-J9W@-)l-fWsG%WkJ4fmAuHc((Mnp
zd}WOezu)B-WLtHGT<7v}bah2pqo>(b?O4(1(()4L$lgGM&*i9-Z*{Gaok#^hzH~vA
zzX&yW%>^<s^#^>OHP%J*s~k3`6~Z`M9kRE@=aN^seSsE7BbAt66`0`+tns=)VxEp-
ztKD_3X4&2B4p>()mhK|2%XKRQssiRKWf?VlR?90;xudbs<J71;GwK^#{0%zUOe2}o
z;SUfMiUaR~ST08ss_?sf0h<$8v$MhBlLJ17JK%>L)<wWgU*Yh(ojSFxTCOgTtyMGS
z%K25(WhQ!6pxEv4LlD2K$Od`K&0ONZOo-)ZbVU*VGPAXEzB8%`YPSDMr>Dv5@VO{Q
zYv@e(%4QD`yxl&(b&;Fp1kA;nG!Wh0yi#jGBFUMit*8xE`Kb}I4H7Q&x|*9jb<{)E
zOBSGW&=akW74B7LQm2zff=pcimDGFCQPF-ZERd^YSCcoeM)tY97)KtTW2K9_a%F3)
z>~;D4o@Pg*8`-N0crpKKhZp+8z(QS+V1`lf6)pA2rACRWK#|ka5*S_<{gtGwCa$N^
zZzWYVuWn+6V;Wg#mgQAV?!*oClwi#9p-@LX@2SQ_fd+;;x4e+|h-bRDC19n_s;)-U
z{-r)iAQt+;pHQl*02I%Md84a&WuQTJ`FtLqK2U~4W8)6pO(UkRv2v9F1JI<+eWX@8
zowA=Gvmh^SUhQtK!{qe0G`Wb(sgZd-es`<f<Z41cu51?2Ky;l<wgD!i-r;T(yclwU
zdOm$I&s*IA!S7zlhK1i15Kxv!PVX9Fr7I9{H@Q6Zb&fSe=5Pkwo@QZ%!+9&_m(xY$
zO`cV*A&l%_)3gE|h#E*NZmpUe?q;Fh=W+>dzqh5?84%``mC4id3xs)g+nh4FFyCAh
zO%~?Q(9$(EASuc(5z1<6O6S;QbN=+0j94<W*_a6vnu=9$D!Wn)<Bp)E(y&g}Q-q_u
zRP22;>d`EZ(i)G^taPf7$?{XNKO2RAsq9BoT56P#f`4P+)AV&?<)v%R5K!t};3%Yk
zGLPpM0Dkb_9>VhbG~qDEwVdNVjiecf(r2KI#*fM?>+&GdXyF8>|I1UiV!Jj<I2~82
zvxDVl2opK%82l;d(}jyU9LReBTc}i_5HN+ob+s&}Jpd|>2HlSm8Of;!-7gZJ)`WV{
z{WRfe51|L$FA{!+j-;6TKf+&O!0UdR@MZ&E_m_mf(ty|fH{lBn_(^TG3>u9qcgYMk
z8SvD%ddM~4X(Z^Oz<}pn!m^4Dc&b|u76YDist3t{Cz<q6X~64yTgt03;EnxKYrtb*
zMMHxDKPHOB`on-9YrwY}@L2|Yn*kr)sw01c0Z-#h4;u~maXJ$5!v?%@9lFVYKhr?}
zgaKb<z;_w&6Akz+2K-qDe76Bldtp87GvLqGk%;#j@K{ww!$AZ7+$a|31_r!-c0$>z
z0YB3~KVZP0Z@>>4@D~{H!unbU(HNmUrXDg4c$1DqoNd5gXuwZ0;4e1dO$L090iSEY
zUt+))81R=G@WlrFWd^*(fG;-SB?CU!fUh**^9=YJ1O9RYzSe-BV!$^T@KX(VuK_>J
zfR}=wWJ;lo#f>IG>gWz+sE;F43hvI_WgtQ_GGh)%BA1loDo(Z_MQ{TVh5I51m&_uV
zOiXw`g9`~xCAgcxmlK>ua2JCwBA84`coT!qA(%`^cq4<y6HKNf+{WP338vu?_A)q?
zU@{fqS_Xe}D_}Aa;YtR7PVngjTNwN?!DK4J1q}XxU@{S56NBF(cr3x$41S$pG6`XU
z!7mX^rXW0U3W!T~5=<r_+{fT;1k=<H?`QDS1k;occQg3+1k+RxcQN=^1k>?UcoTyk
zC77mmcq4-!Aeg3fxQ)R-BKQXcdl`H;!8EnQwG6(UV4Bk5N(TD~rl}mZF!&aNX$pr6
z7`%dDnz~^VgO?IaQ!|{+;6((}lne_DzK&p;is6Bi)c!dHn+Wb>@GOF9>V@|+xRBtB
z2<~R^<pk3d40kd3B7$k^g*P$y9D-@eg*P&IJi#>8!fgybonV?`VK0MI38pC&u4V8y
zw*aQ86Ru?N=LBC)u!X@N6Fi0B0>G6!>N;u#^%68~2km9I6S%Dnv!sUUr*cPIC-O6*
z`2@0jbsTWw4c+3njHgy%*7RmPgNVY8>>4oI5G7Pmk4E@&jqvdp!c971O?W#f>+{9Q
z?Tio2ugV=;sQwOQXz~%r<s<!=kJyw0H%rO~Qt<BsmDRT1Zs8^igz4G1oT6UKB#WTl
z0KzZp#mVJ@u;qR<ZLu_M10pFnm?bHPSEt^zYYSCK6aUMbc2HKNPn_IFxGe<f^gV-V
zpqG@LQqRZNN-6uL1A{Fnyp0H@j@MCGQub=)Kj7utu68$>1fgZzVw6<ZV!+(AH)9D<
zDPQcO_DR9354u1QxF9ORgvC1s?t*m6;mH1*!grvEI9Uk$wRmyq4W&y;7nfF7N$Wq^
z0JESxpK~W7NqHv6OK~TMJ2hB6W44B@K_(sY3ESwR{v9JkT?or)Q})_IYcp+At}!2x
zly9Y;Z&T~Vquur5y7n9kG3;&6sYFa6Fi}ztN<Hc|V#i~cF{%u`O3EHd*{qQwmY59X
zKZ8J8Ia_F^Ewm(4%KKbW{zBXXnUZo;Z9jpcot%ye?Vy$6_EvP>wmMX$RuMrcW3$^N
zNFj?Vg&y7?tc8}}7Tc#|_O{(oECe!sR94o1wxk?Vt=|%L8Ym04pN`_dMfQu^2N+lN
zAH+-PRj)v-)$=aZLz7VLXC*Vr6V&qttVUct?|w5<J?nTqz2F(&o}gCGJ1`aT?O8p%
zo|X~o`2nv-j@FZMGulH=Lp?W&%OYGV;YLSsqoZ_RpaxCGv4kUsq#lam#u1JUl$WEp
zA%1_3;?5#Eaz^Uoz@fiRtQ_9HU&0#4>G3CDkEF*dNmuBvb<uh*pn7IeJy&V<JeZ)K
zvf=ei8?m0(h!^Vl0pe0NIqaA|bW{yv(T>_E!RLaB1~%qG0b&X}$mOXwF%G*V<*^(_
z&{>vE1D}ZMOU35B^`f_1_(EI<yV0JL3#?tiuz!XVKbS+aQtG^yO&_%u@=MMKbM~Pl
zVNOb4pn3v%QqKpYx3?ilbGr=%)Z@n^k&bSAXfp$u;joKVwe;8?QS8_VUuod7H5rqq
z851dV?+_D7wC<A>dyaIh2aM(sBV~Dx#lWrf`DE;zT&7<Sv4c!~rN?%I)d~1>)q(!|
zTpU-PQw+hWjTN0%Z$+m)$18=(b0n#=(rZ3eF6L$oUMpa2rRl!|M(MA1<wd*F{Tx*-
z-uG)<4L%}w-o{e##%?ip;U>iU#M~RY5bqar<!;0W#oU$q5%-C?o<78?n7eua@qn0n
zEuulOrgWoNW7{OwT;C<ufK82KzgXk$6Kif85Nk@rDVB}m6ls$<rLs$$QqwI?sogJ5
zY3LKDcn8EOdOIS!q@da+1;h9qZ7b;)@2!Wij_h53O2xDiJBk6Q&Jy1GO3acs>r9#z
z=xtMB_Gt34xmd0oQ{VcE2IZdZ(dkC!(hBXCD_@9Dy^a63?wTY$EcNuI)=OeDEcJ?v
zH?N_Qx;*1O7T*b6UU`m*3f_r=#O+3!K+rjQ^r(9LSS0efm~GFg{9Iga&nZSa<1Z-s
zxp?!w&!I&067s}c@BpfZiWG4^F;YDRK0P0%!?%AfUcL*Y<vEo^JQ~C{rPmf3O{VU;
z(B|R+s=4Ec*#0OIZFj+TkNT0dtpD7NM81Jq9I}!|bnAVTS#L}hbqcGvSM{T=ijKbr
zGVRJg)V-|4oftG7C`m)Kb2FQ_Hsz>YIfCW;uZg1Lh~{j53V?V`K98h$n~;ap559zu
z!M*9F!BZEC_fCLG?HAQfAj8YiU*Q#4qQ=|s9jw}nNlxB-2A^~QE7&+2_n+XE+$U9`
zJ!E_S9Xz#qWa=Ul^rPOO)zh_Ncs>5%_1H$Nr=NJCp6StghJ5T%Rx(3MP|w8S^}PSZ
zNbL#odR_)ktvy3N>4#BVboi+LC~n9n$x+;pPnsXa4f&)iqqre{7e{eJK1l=)v#r4~
zT#whHd*k{mWyJP9PQ0MUeo$%c8}dp2!AfQb3F>(tJD<3E9{GGEJzmA@34-Sqcp^+#
znpGIzwJ;O5?ZB`DEWW;s`ll|?l7H%^G6<S~dVwTUn15mf+&>YK6v7;@7n^%xd=yoQ
zIY~ZBD-h?So=1DgAJh=tNIt6Svk`pMC)ht4xg_-}V>y?EUgfvM1rDm%z$&S`+ciM)
zBy@Kc=I(01O?_cHx6&r~Mb<{RCX2n(;<a^_=Wzd215+OLPcLFSU{`vXujU?#(t5s_
zJBRzH>$rbf#QoD!?w?k0|8xuYPd@ISW-<S?eBMTJ`NB=&@*BFu<#M;UeC2*|xu;KD
zzIs4heyya0rJk=%QqTLNOT{VWV+CC8W3zFs9&5sN`Pc$n>&9AeZ5&&PYhY|Gu6K_0
z^7ch`*;BqEw<QHXZYz0Nymx+#>pGuY*As<YDH0~rxi^~<x-Kac$f;D%KT6{*&UtNs
z>dD1dD!Y<7u!+T*13RC^+<{>!byR(h3dcLJnLsiJCNT#FFO{7!9z|mum_!aNTX$f7
z&L?Uyi20`wH_m}cgr5uIBo6E`bS`sX8RWpuA?kgfQu~tluUlEYz3QJxGod@-syZU@
zPw4>R$+-V2*5@~JJ_ff6@?3VXewSDyc3}Y(^<25EKC%kI?sR4zf`f2jpJCMu7xvN-
zc(0}8!Vcq9C*FmfU}ZBFFXz2zaAB6LVW#R+@JjB&HnPs{RS$xvR!<wbunV9e^+Bzk
zBg=-@^8pT*;!NT_Bi574>+weG8FFD$wR$=d)YCq^o(o2-r<-`e#Ain98FFE}QBrGT
zbO5PcQQVLVdn}3@a$y^yxFHu7h~kF$)kbkcF3b)bX5DX=4%g!!(Y<kcyylaU`m2w4
zp?$fCV_nNY3Qa#cABp-C3%m{6`y-~H`I9Ebg836h!2L-j5hwH~^t=lOF!>X$K%75m
zL?g(2tJtWLKS}Bd()u-jlF$=8c3LbKW{nf=oL=~N1h2G&xab=mE>Kc`;%A%^`jcYj
zPcp$xZJHMKC!1LtZP@deU&JQoOn4fc>(KsYC-WOSNXz1V58+xrw7<cAXwJrRw!yh>
z6WiY`>SEiQrQK|IvtmEn+}zT~_BOr&wzZjsosFctC-t0yYxroiRh*Ju1gmc^VxFOh
zd4?k98H$)^C}N(Wh<S!0FLr0t_(-ok<qPH&j*9oTkXPtg|FyQEnMYpX0URE&?HqZ9
zj2hDjUSV{Y2Hq~?mgbP^g90_Lut$9A2>vT-PhWc6#^xmM`O(HE11_ju9P$f?)W@kf
zcp5e~htvyzjQ0!QfJIWja7bOo`9!B7?G4_9z&(f4w2jRkwy}X<IHcx-*ytA?;~N_s
zN<EH#W3zEA9ZFrO`+q9+gi0+PemF&25<9EDSACFF1ou#^xrf6j0{d^=-q0L{dw3Z+
z(S6W%;!OUERzmv5<}_B_F!ykgw!?YyAMgxS<Q~T0+$P>VyvxdF%v{2|)8HO#xW^D@
zZe9bgL=9tH)!SIF_o`dLQ>*6}<Q~q4qSO|xo>vzSucsek#nrQN#ClHS^;{pVXUIK_
z)#|BFP*3OZdLkc<)Sh1wFPP6Tq|w?l<Q{$##YM-F`fpL(kbAf*iW_nd?kH}^JuHsm
zhWM36aYOE58gQ6?_f!wp<DYR-6sO0@Bet)Hc%gl#Me7;b;p|~0GcHI_&yB<D`SanC
z^mrq$=a=AFx&!+N!QKh0&f><LntBS+djL^?v>(Ht{81lb3YtH9m*mn;pcz4@m7Tv4
z5uLyF>U$e3&Ts&C5NKL-!5L}Pan9%v4oJuuwNhPhMr^Z{$c?P}yXHooi8<^VC_T0p
zhg~9Eo?WTSsgazIaaM2Rs5_^D+Jdt{G{eGfnQOjfs-(O%{D^M?t0AF#fhDr%G)O%k
zj+RnhCV$mR7cX-<KZfD9D*>!BmBt)XZ+T8F>FS?q)2}`pp-oqTula5D=a+JuU##0V
z7%98*srr*M;4pTr|Cap2o&D-6osJH>!q=M;`i64llzQX?>c2S8P=mIYD<?It(3yUO
zH6ixE>o9I$kzY8%VxwOOp)he=O16du7g^v@Kr+8zIPm%aMPvK|JMiNBg+po?=M$ZO
z#AiEn)N`00cxiqC2VRHNvp{T9cH2VfIogTWV<zpy3y$G;kb%v^3FH`Z+3+43-&A@7
zDIuYM_=h@`iP5VDP;*7cG1?UzS9@6je*D$>SWX+Nuqp3IA@n&r4uhX~oM?2H@DKnx
zH@kpVIQpsB`|r~MST7F1BG^6rnB2p7yYhvq03W`NJVXKJf%<3;%7mWR?(nd^%M&0V
z3_HALn1_HdcT4Mc(;ca0rOQiiD!sW>zG)ZOh5I4gju+4y+ux%u2?E@jB3lG?J=L%5
zg)`G+W+!`}Fy30zw3EEI2&2pV3IY6#Uaf>j3Xg(};G%D>h*hy&ISdW_(pd080MT95
z4>=b)QBhAPF5OInj0<^SCXiPBG9#A-j0-=6EahAVNK3onP5SRbdr21DB%`ASI+uh?
zmjYwStd1vgfDM|5Ph))fS>&-ktyyglLtO(_5*A_1MA4aPI+-JQ`2tXdkAoz96qo)n
zVU?nc@cW3UTDmuO5NR4y=|uZ7fbcFYxBnTn>kx3eFrd`mQtCojY4tHmt)tWflv)gX
zs&1gvpD4AKQkP)mtp+F+q|_~xdY@9Yl=?Ha9O@EEO~l?nwNvU@oZ6^!C^Z3RU1~9<
ze3Z(kl!P-WHHT7vz}c>P4yDc*L@Jw7Zc2@!)U~*Gqn>yVsh=Um`q(%MF9J>MD5UZ<
zJr1Jn>W@hQOEGIuu>Y^o`#FaDr@>Dq*_1aV<tw`F*?9*Jg<n;Fb}`u%W~PIEP#ny4
za6gZ`dECY0O+4Pn<2EfW^P*5$EnO<5-XPfl3dE@x9Z7L6k0<ds6ER37L=>dnv`w`Z
zjDeO+X-bQLiPdu|RZ{#Ca~P#th$e6@L~c-z;AW>iRE8yZu#HNFLK_g-JDrvG&gU50
zydInorP?t(@?Kpv9z%l;d({8p<@(3^&$zBLy-iYrR1j<)*4S*KD6VYhbCjvB8D5rA
zcRoj&0She(k({wC;G6Hv$ySSX250*r5HMT<D$t+7Fj5H03(c>IEuTd;>>JoHYI6VS
z{!Fc4S&hB(@qJp|S`^wptZr6zyIyvv*+gmY{8OJ+T#G^~K{2gI@#E^c;l+uvUi6&b
zehKT^e;)*FHNRNV(Ji*0f&R3keVO)91@z)?kU~$kA-^KD7%f<qErsr=w1=J~rK00V
zYgTh4&Oele(zni0(znl1%3lxwYy+@kPAGl5_*9V2ffaV06GAR9WGNJOWtd0Gj?aFs
zj7mN-=lJX$D7qCyB%;Euq2LLm=ZH_G?@(CM3Tr-%yx5#j_0|)}sNQm7jxu43Ei?+4
z9R^JQ`Pz8_BkoP#LPM9R^kk&BqNG>6`|1eHuF_+N-OI$tU?VaV$_8~*y|2>1Jd5^1
zrcCPD`4Z`~2wsN|&J7rxRKni*q=^cKf?5>HO~mG?`Zw55oy`%<X9F85fzGCR2sT<2
zvJ7K$E_L<;oQ)kPm)RuJFNxQ92C_**a#7ieju*xDhmf#^vdzcr%04OBH;4j%(e8HH
zl~<+UM}t!5+Ci!3Xd0ffi1Y>~OUl<$@copQvtVsr2XRK3W_jL5dSCxpUx=P_@LQDq
z0(3$HY!`knNx?{}9}{G6f4S6Y|9yd!_bG_3-;c|UeYiLVaLKC|i!D3p@-kg^)8!Q@
z*kh4`dk0!xLus%&h~ELqd5d_9<GU%1I=uT*D7CL%T>F}><g=D^Ww5_h>TFK2m4qdw
z-zLuco22|r%KHq$r%8EV1V+O;b-WnJm4aWV1kT1awFLrx)t@dtHT!A&elJDve8{H+
zNs+zEpcMQz-FK-nsIMO}_CQwtH0aBp&ZNs2$eC_0dDZuKC^k@P3l&s^MpuL;<k&(t
zLB&}+plQK;gx~*o<oN7AmMXKK2u=87srb}{-zyWIRMLNk>i&p|?Io}H-orzDlMG7|
zwp%!);`SEvshi(2K5<mPv5K4l>@OB5Qs_B0Lf8uDnS<nQo?|1AAs*KdY(z7}0~(@_
z7&F9r4Y4nZ@M;KV9vOwKA(#PRhy@y=En4JihPZ>T8LrTvN{#=88dRV`XKGM3L8F$1
z(*Z8qRlhOb-yO3lCne{^!;%Tc=I_|#h&Q}0w$odd*n^}?ov_B+se1Lte~m<7j_5Ii
zx}59^vHV0R<0+JaVY}~VAn%kSxb2XRft4K`Oxc!AB8v+T&;*F6m!fcJS{soc5$|gT
zx%q8+0&Frm&?<zwm<bwq<oHRYM>(4J*Od1Z0>r0YR^H0{W^HQ90r7?xsiF>WP1$D)
zEzY)uR%J`huce-UjFvj9XL+U0z%^b_iMfk5in&WSiMcDf#N1nOD7FMQ^l>P5Gw$iX
zCEi$%+xj1hH#XzW{wJVC#uD7%e*;BuVs<m`^S=cwPR*9!cK=_*+;hYmFCbja%XIQ3
z1wU<z^py16`3QUNsV}`$%o{UWitGwa`}I7NKzI75!oX1*e+o(M%125V^TOPX<?3iC
z7UTCCMCNXq9GCD^YG4dfrDDwwn9-CJIOzAmxn_bWMY`j_;K_8~<s5fir@f>A4=O<Z
z>mm5{ka%2YC3*li0WiHwKECsV3gv`NIo^LJV${)p20E~R6jBWC{z9yQUGTn?f&^x`
zcW0V7{&l1e;0GHrt~zIoEp&|(?9GIBbH=Ef&Ly9gU7j=M3vp?81qK%O2mipCUwV#O
z{wCW|rRSgzhRv{DM9Y@*=Yx!vqgD(EcD~HUW0gHLPvTDUZ43^3=zf|M_Rvphve-j2
z&BrQ~UAEwdFr)PN&C!V1ekSN_owsG%I&a4B_UsBgUl9Dq5P|-@PPyp8uHQp8aV43<
z$@twti7EIkpv262an@ObuyeD{dNJjZ<6kQ<gloZQ?JF3AxFHH1&AO<E<qzUnI8dd6
zijX@QIGDzeX*!aDDx?1_%1;HquQYz&P|>ehzRDPUrt|xTfhuG0na=MUfZS|<TQ8p4
ziF1roFq?Z#Qg9c3y(aA02E~puG4Slld!V4^?i0tqh$MQoIQ2!8So^ZAWS}LpT=`r0
zcBmB22NoP>!r#JzAEd#YH$c{VtQnoxmqI4{c>jdbQ2Mivlpde`pJYj&E=~Chio;%~
z`eq@)?Ja1w6U|oTO;8nWOL?_4RK7hOs5`f&D+hqHZ%a>sd+1HIm-PDfEBlsdTR~pl
z+`VgP^9y_MU9hq*mVU^dVdIU>)+x5@38Rg+(A-R0XhAk#SV$#*6NArF+jl@y6t1nn
z;Kt3`yl<qQZ?K*|CFQ*-mAoh3y^W|!-oj?UF3#z<rMw|Nu+v_0RD7T(^8lV<ItoKc
zTSt?v<fQmRn&{Vc&i(@;LB08ZXn(OAjHz~e9yGj1>N%Mjpw*@wH7GmPNB;tv&e@x#
z&Iv!GLgr)Y=+`kTQr@+7p0}CLl!#gdT!nI}O}RO9erNh-n=&`Me<BWfQq9M(d&C)$
zdf)&UrQDoJPo*0i+73T$rUS@Se}Sru+AS9ZgT9nEn8IL(OT}@cY`t^IW(znXxJ}$}
z8J!zB`}KxN%Iq9<JqfP#Rw&mj?96$pLaAq~8+gG9IjR-a0h4KqyrQVa^E&u~wX-5G
zm-VJU1(t%cr{~DzlsyMdg0SOlMO}9&d@m-lQ2+=+YaBe`iVWq@@k7DLsTZifw$Vjg
z_}7%Tq{vR&)<Hm^fY&}?zwkx#G24X)#CGb23PrW$?L#SX;n9?LX}u9Vj0v*`%dI16
z&|YLu;552Xj^$Qh67IbG?QB6P!*gg)eaWtW-%iy_!IP=J<=1si{R5uAz!Afk{{j+=
z)K}~&C&7BFCd+F$aHFTaJ{-jC0jvAKiY?fq=au3(tAO7Cetr1uhaBCQxC1R8RVaIC
zDOmEk*#1MXE?2_G4+OtH#anq`PN)6yROO}M2Wjvc>3EoUO7NSMz`4P1QUhm@HmT`Q
zV&Y`*>vZ3Y`JH2)D(zgHB5fyH`eA<cvTu#_mY&d5ciWe?;GPq37X!haCpunV*Qe|W
zo=j<hT_1=aR=Exw+2rH0?Fz8kdKFUYtFK{{m7wpV`{^z8n73hyZlR~S4RGm_D!f`T
z3D5r0Tvd+KZ7Jm?bs`?WkxC8*E{6BYq{KN=r~r;VC!3bum+i+{6mG5es$)?Xp)<E<
z<M4>`&On|fEmn0m<4$C+ssg32C$WC}Lj5y6q*8KF{7E;KK(YNZG#}{sX(+{}d?_h|
zw$PXyTZug<Gmv4!=_(xH@|^5#lhAP5F)Mr3qE}HZR!3vv_#qqfT>T1)kwVzw*wka?
znWhY}!wlY1%0983_8hhlnqa#M%?n(BHCJF1_7r23R4KSK-KHQbyr1<J=4(Z0YL2Ar
zx96!+&mfG{o~^9o#PYC|vPXOX>jOOT5xz$JNl#{vw1q|`!nJsCQ^gq>yN?){YQUH5
z5r4>*m%7E;OoA%As2V0zg>uTC_pwd+tF0%3>M&mWPDm*SA)y@-W|Db2gSGKf^#D|`
zp!3e>(xc~Szj%c##CbUtw8@n^)1S2AStWSp<4k=hrv3qC&_f>~q8;(8cx8almjVv=
z08l@n)Xiu^xQ~<_%J_Dc2|5QGDKz$qtB|ILzZH8r;Ec-%MxIH+$#!nn`dZ}TsT`@Z
zRozN2{z)aTwN$gb{q2CzUA5}lr~pNuF|h5Tbd)VvQ9m>CR`on0$2lgnZm_W!rGxD7
zh4wmUGh;9y_GBve=Cp$lkJ3=oKRT4zd3{95#(wbrRcI33XJ|u)dKvT-t|JT{#R_Fi
zGRR_&ktLK-&htiSV&K|;Rh)HWqoO?a-;jc|H>9DhkPcEkPy(%R#otHiBpUjLIG_)G
zD5mc(*xLvNXsV>VjIpGAg~_hnpqULS7)lrRTjDtIyo$B~<I*H0P3pAZJg5xEZg^%I
z!z@Fsd>J>^$SJ9#v{HwDz)IOd*J@5{v`xY4-;!-p=4Il?i=B;Py&2+c6SH}Ak7yHk
z)X;e7d_G4T{SW4B1a7;JjM$5ymz;YlI&=0)oewes7#1wv)pd9RhCFzvofjJ#S7OKP
zG`Pr;!(#g@;6$9;SX`acs<HJETks<^=XVA8*-QEYQ^?9u;v#0{p3m99iawY_(_Sq>
z!TyUdG{XM@Peh#c?0v{)hEJWym_N^2Mtw;Xs)>36vTEc;S=mz;$Y}$aadwCu%h3v|
zViV;k@2Q`BLsH*Klc+kUVS6hn=!usISDoYK&7ilz*vO5WYhuR|S~iq?D7J5572Sz@
zUMN?ov2k!VHJr_bU_(m4S%IdF$xIuVq@*_5NUQlM(m$D0aaJbD$@8dVH9d@FJas(;
zFu%_w&A<wPiBSW;f@Fk-Kt}XdLi6U6-9CboWvT{GpI7Htwy&e+|L#St`4X4gO^>M1
z#AOeu8@Y8rF4p{GXg&?>JZ_r`05kEkiEpOH_j=BkTpW$Iap0R!ACpKQgW#ZEfFhbc
zZpUd93Ci*Ykf-V6<UW#hNFS5bGe|=a6@*4nKZU4-hCoJihCv^+=`*QsW1S_1?xY(@
zuo4oMw86o2v7;8ZMtd`UU(6>No3ERCiBy|zv1}iu=C9vP7J;lNyBB0MT4FnW$H!<#
zrvio@r6~LxTw<yJFaq_%!>BinKph%J{o@g+*AAn;X9ViohEcB>fx2!Ob>j%sHw>ee
zN1(oL81<qNsIMMIZ5@G{I@RFl!pwj?lS9iK^_snKbbkfLK=2N|%mb6f2+k*hmw+)4
zY{Q+$D8Y$6Ab0{81HscL4Fqo!0lg_{AQ)$?VjB@eVr4kpNbm>|48{^1FcP#9!H!rV
z>6NaikT(#)f5j5~!q|$LM9>-Aik}z>CK5pqnDAm!K0Tp*NHsw@JUNl{G>c~?a!2hi
zks&uMOT>1X;>o7e-;fvnm}aGB$YRz^V*6dgnC%>a8O_Ep?PU{M#rA?>%;<1DnPlsc
z*N;{EuJG@WVjL5Aw07Dx#x{4w2uv$9R^i`}c?0WxI_r2p(f-F_d}%3{Ojr-{`f>2E
zEBx^g$5>0-zJD0U-;TiX-!wkqPZ00ufcw$5T#eA~)<*%V%XYFs5e{kk4xi+)8wCtg
z;5p)hS#}WF>Yq`DsnbC4C@089tUg2p8)7+-ts)NH#G!=<RvHPkHDn)U){sbvRu}Cv
zh=*ZN#;%pgxN{pHvsR9PDb=hJ^KW}l^AV(?vTHR@LSgkiWRhR1iE%gW^qkJWUahul
z!n=6HxgDJQ*TmIHs~YtdUT2Kw9BB=nxS3ijx`^W>#*thsbE7wQYwFxYtuT1!|7{&l
zP#w<@Q`nUMMqSvlM$adVuMgAe)285#K*Schpu%|>UteICy1k!<-X`!^&^h%(TEk!m
zkB#o9Dy<)J;+0^9@jGy<DETB%N^A<$i>Z;c(b#x{MZlVzTAq#{Z4GmGoPm`2by|SV
zM<qM3xZvPyfJ4$?Q#m1L`?n;_naH@Vvxo_UJ;NdOb(-543!oWV(X`CN33h4d%58Kk
zv6apu)TLiQkoCK%vC$nJj&KYcx}u@!ZqLcZ#y7X0)~G`}zC;Akp(5TtitqLIp$he?
z9T>k~Vry$4XfP6-Lj+5J>A#9uyZEIVdqpm$`h~;gD?`!#052b-fxyi%Wv34>ds$Sn
zb%oIh-@aWNFONaGq5SdMns6o0T!&0j{c6PgJeOWW--LEkr)R7AM1W4u<ek0|IeMpO
z8ajOm7#cc#9`RqloA~epo9K9rKfn1dm>o7jmP{P8T5yc`xg=h0$6FwXAQoP3%y{%P
zdhV;HhwfVJ+PJui&XGgYcbvY$g2QNa_VW-W<j>x5o(4|^9P(%GDA2(30q$tS)^a=T
zNz?@Zl|2lBxnV+m=#q{jYtN3Jk738FHW1I?uFS|QEd%C=x(1*j|5|1R!d1ZHeqD5a
z(vb|!%@Y;MR~31OZ9OM5aPG3Uh2BHT9-7Z=2>1u+rrZ1AK!+?e|Ivg&!o9+A%v*=1
z6KWerQQv?Jw`uS^5XapOoGrN9Bz2aVaM-x+mk=Af?RRkCX3y)xiBuYP11W(bTvKsm
z7aUA$sj-EMa)`$Y#v?Ev<#7(L;YdIDsKS*-+$;@uafR|_{{?h33jX~QiK1V`HM{>b
zTqpIXA>Xd_4NY0?c>|`O>Z(c{+re&Cbgs_Sjs|cb&^vDu#K%n{ITz>4^KyYGCIq%y
z>S6Sjjf{nD3LeqJ$r_FiEVj_XTwCad0x9n$?T!GRk$H0!J_&J1{Ry5?wT0$c=phpv
z2Cug%3kynnM}vz{+KXF%2=ujxJlu~qzizaz!P)<o(sF?mn%<jEPx1>mx|T|I`c(aN
z6<eWQJLar_pRdMc8x9DBt#=@$`nR#uc8nLzE0ht88y3t(%mhF>KJ^^4<_3HIUGw2z
z&9B_?4h~v1;Y=jlP7>}-SP@PC#{4|WAMZr|vkCL3QT{!Y|4=mlUzMjf1x%oGy5d}U
zBi%1D>f3PCJam7NeIG-+KXlTLyR_}6P}q*)AG#wGw?ZrOFtQHQeHvTnE({AiGmLF6
z9k-V3S+&UgHa%?!e}@xT`qIV!P+2^8<U=1|wpElI56}>?E4b7Bfx4M>(_Q?Q|9v=-
zwJA5j8q!&)HM^qImD%5F>ukY{u)=HYl=8m8{R?^`ODYNbFQmiglEHu(%v2wA#AW1S
zW%i#7+eFe7^4b=FhNjCU5}u@>N4ieo(UKGPlw-EM;{lwe_1cwh)I1RNr_uAEJ%>|m
zX-)9gQ!zuMLy$k4son-UvNuYkbO}sN>0;afE3LM7UShh?gioGyPAf)SWzRbdjr5$v
z%>N`)edAdu0Rmm1pSRnUk0Dz*ctj?=gV~%SmGrlK%trf7`g5S@v8X-&Vi)EFeQOEx
z9Z6VGxbfIv0ciLYl;MkW)>sn4QHaMAzt|rmv)pgF`u{@^t~SkaH9H%c9KKue4Np^j
zi{DjeDy^x!(X_(RT!+t|1{wsD$!;<UE7k<?W!BS9J5883=L)|tufEaaMLg|)9l+NF
zZgCPgE^jqWF|9FFA4gDCR%*8<&t7CLEjQ%m=H|BGOOZ2+<Urn~vZu}wa4ajhDeuyP
z*1376tFJB)^7H9C<ySO1@P$aH>xy|kcb&`u_7T<m27z$1X4y50*&5_t;&S4{qa0eb
z$mM99JJnQtjp>>wWv)g^*E#qatLgGQ)3Q~b6}OnKHjT{VCSet0K5G{HU~Rplr7<8o
zy!c*f9UB|`Qdw(ZDgFDr$ztJJjgA)+78FtS#kyQ^<5b}Jh2}DRpV#GgJDSTpO-(J$
zZl?nc2KC0fYZ<EnS>vnmGZzrUYWDqIjipX!;&STsdvT+K!hT+!(ln#8IbCWdVes95
z9oE2M*8!bxsfFcb^Vr1W0v;!Bzj->z8*QUO3)#ASoWHJTT>ui#_#=E<dwywi^VL`5
z+nwObJ`|kqbP7&Svp-<M$B$>t(z<ju#^N=+`-!sL#d^2O-Bg#0tf?kk@-U~Gylz}q
zf&vk~$SdHdecKqHPIcez%FUy?R(afY24<BGpWkJ7tii`>bLrc`rpqs945tp4;L=1m
zs<_UzqGjbYd?FQ}(51=E;XoaI0k_Q2va$i~VY#MD>Q2WL#+Oq~j=DOZ%kMW`(%9Ha
zI3Aq%x-X-Wy*`f<1RkG%S=~(}IleL+9a7qyGh*aseC?U^M=F{%E80oMftUoHJ51S^
z<{9;rgjW^jR<&o5%eiW<&(oC46ss#Q24MYT(EC5dZw|g=-5etswZy;3bz2KQM=fJS
zpg&@&i2?psb)p7uqbuMVxmsEerJS2~6@7CaFC7YfPemeqNOwIQiFBu7jv~$;B?y&&
z0v_l2Uc_C<uYD#G$xRi6O}N=zi?|yPOLZaMcrX&#huHLXBqE@^>0QtvE<kKSY(ZRu
zxDv4!aV_Evh&R3$i3}hf=mR}%zupHu;%vn9O@lVXZHPA^ei-opUJ2MwFkV&9%mDm3
z=n+?9=Uj`p7I7<LFXD}e+Ymp2xC?PN;sQJ}aS-)c5cdJD#iu~%TSwkwphw(>_zA=t
z5qBfrg!mxhF2n<fyAfwXAAN{Th=t>kNHJm);!4B?i0O-m7Q}6cD-l18xE662VlU!-
zh}#g;J1Jebg)s=&^flVS?H@hwNg?I-MM6qzcFMV9MrUqJ8J$h|^YGj9pOMJrM8HF0
zdO^rRKG6^U0Xjq92wB;4vnE|9p0PT!O}O@)t1h3Oa}lv18VYH*SK3*Pqtob9?rB=R
zf>4A!;#Gh#apn2gCC_fO2&GxsbFwDQ$uiB!%AJ!{P?}XdFROcOaP+;SIy3G|zd!As
zR90>|$`#{2{sgo&y{jovx$>;am}2bjf1nuTF}7d(yC5WM<{Od7t3&`9f@l}Xw;R81
z;6s4Z-}&e2p*GRs<s^*VuQ9fV+EkjA8`N8&)%zf_2mTg`VCY1{1f(|OrzXIYJbel9
z?*M-=4o~ubiQj(UFN~2Rs@u{mORSDS&h$hFc<LW976}1;P(*T-17C?bLYp={P(90m
zfB3CP<OUiK!}~C4J!B`k--$%fjnP2$YzBTy9G>)A4gOugHz1v!*=!N!WM!9TO)AeS
z2&Uafbt6#uPf&gkbo3EUV}CwKG_Vm1pz!oclpiiD##>s9QmB3ZjdGJ<OMZ=0yYyej
zl`G9En3vU;8XR@+Q0Jlyl}krG8~d?iLHYDCUM6pJe6w<GtXyaMeQEb|`DrX}M*4{(
zk;o#XNyd3uT`4psDNy@J9u@g$W;77~@4(YH4&w1AfIpZ3KLPUX2cDj963Vko6jA+C
zfhYc}34!{d<I*fskfxt76(~FMYd(!c&Y_CX2gWv$9*pfFx$Xr0md%lf4{72@5%I}M
z$NWDUiKK#0dbLqMm@CjvFyr2+W=NmA!N>F&<fVFt_g&I)b^_%NqMmrYOhEnn6X2%;
zzb^s49Qf`8c={VLTN2;{z;^+E4#d^lP5QW()5qiKZ>$jic>MFgCz9uXIDLG(KH>c1
z@#*BVz~4B3NuKk7e*$=@DH;f01pFr8<N4TuZ+J4+K9mzZ`5xE@7jeP3xIfjJHR--I
zYnG`q-I|qqZ-$graL*_yt2j8iBFoa4Rg6AHPcwTDtB-M%$;LN*&+{PoV7SGENZQf6
zg|tr?+OFqNLc*^QPDFSRKP!GL{0-MO5PwZ$T0!IR{DqJ~KsOB}#HSNLFUgjH>@`yR
z&2XD&o%Vs@Ci1B|gFL*nMGVGJ87x*AgK-4h46`7EY~r~9E`)N4!4y(*0WOnt25T_x
ziNTpv<{1Dc#FGg4BMenO$d(ZU3p7s*uA?4Uh)Yz0)9^G9F)&k^QUE5zGYPm02Y+-&
z;pfC)BM1IS?fDG=x^>QSWMF7H&`3GA;X+1*Y3ECDDFTv(TxgOcU>c32<)vLl=vP@D
zPt-zzDP*?up}dd4pM@yxOz>ra5>S=kCxiu;`5GQCOQ-E>6VgfvBvZfwC8&W{5mbU7
z6=k6e*YWt(60{gUJU*#~NZNiOZ2_e#Iff<v3}7u)N=KG;dJg^3n#p?3{Q#($k054$
z-i#i}U?GzB!QBw<P%7z(WUyi&Ft!BHij4g^8_Hcq0NsAWw<3jF0*>Q{z7edX=`64$
zQRfj*2w);&8`X0?F2ulv(|Cd8qMU_9_6z_rkrv{%8vqT2CkZ$RfVk0`PdEduqnt;n
z%wzym=2rxeo=I8cE`{Z|kT^|L4|N0O{GQ1E1Rz?^E&#-hmW{$+0Z=_B2{;#BO*x-a
z8T#u=ROU+p=%Y8ptq!r!hzr%zLiGfB&Iuyh#es6lc>};}#Fhdd12lrE26{AwjY6us
zlYrmjM*!Wv7JBfb(eVlaG#c-q1cgZ2V?x?F(~&C1?_EOOir-Hu@qPGS1K(@ldkuWA
zf$uf&y#~J5!1o&XUIX82;Cl^xuYvzN4P;NUFa{gX(&8>27yLkj2YAegaMAZ_@o_2+
z5Gd&QF>x^cZ9ocRv`BdPObbiVPLM+SO%@iP&92xg-c`#|bP__r#ID+2o02Ner%fjX
z+M!bT>Qux-@x!SWirC2j#g<G9MFJgmP*77X6h)6FuAw6Y1Ui19fb$j>XghBV*!OCo
zt(L{uVzXf4?V}Aa1$}Q!dua--qJ_meep8kP>vA7txkG=eP_O?~L~GB$SdBiyVO<U!
zud(*7<@ub=|MMvNckJV8dXx9-M?5~p<53ed`Oo6<r97U=<GDOu!sB`#uj27NJbsAB
zzvc0>Jl@OWH+lRKkB{+q)I?rCk1yr%OdikW@e&@_^LQ1H@8R)7JpL_@pXKph9>2-s
zk9d5H$D{bXI*Z4b@^~hX=kj<7kL!87ipTfxI8NR4r$%R)au=^?X%4iQit>x|3#XY&
zS{P`~(+~pB#b3|G!#u0-r*QFSPEFJEXXF=5E7C{{*9Lqo{=k|-Q*OYq(tpJwtKC{!
zWtFGTD7d2Djl2GNrd;~t!F48-wA+_dXvOnzaEL#N8gQ&Y9PnvzgPy`2YM0L|<TrZ)
zu6+E7(`jJMp_R=o`72u7jdj!TEJHp)4Gw>UkYBf^8T_<3;L~!rz%hV~JfEx4K?Qiq
z+ZYh?>2EO$`S`=HxI%6>^4QZoLcXg(uJ<{bTyjGl%IXl9;t$WdTAeO0?zpqRB+Yop
z4xbNy6qfPO)3-QL5cN5l+)kA9pbA{_S-tcGk&wT_?-%mvkp)+CAikgew=mg1vZ=bA
zDNeNj!U-8i==P5xanbFs4quD}*&Mxmb*cprKU73)BES&18H-}fNEyoqQY|bmz!~N+
zXy_7Oz8qP&9q!TVe>l~`h#%&NN(K$tY26-jv_MC~WRvysy4|KPJ5#9PcAM-vwOuc-
z+s`Gyk*(Cr>wZALpRU_|kgyQ1zng(2o2!@C?d)3axO}`E7b9AaPEQAzWW#k>x8no6
zd>1cI@{^pp{2VRZjs)pcFR%NJO<5YnK~AXGub0=`-yT<9_d~6sR#5j_RGxfueEIu;
zBc14eQ}^2s(|IleoqZcaeE%!RG?wq?{-B$eXU`rO7|^LS1wDQcX=8bPokD*dih^&h
z7VB5NoDM&Vgt5H7&PcyWqoBiVV~Cf36Edm1uK#SlPRZuwU)2@EnZ(!sL|pkkr-f1W
zIkf_G@7x&T>wgxR#`;@b7DnFc(n{@)t6%RwEf=@QkT8~~$1^C*Ze)>A99Ll+D7+3b
z8fr8)Dd_96eRpZFNsFS_`1;>N4h=QEyuNNzdHHzyc=%(KCHeL8^cMvw4E&SjkAjv|
zq?gy%ErYzg-m$tG_4@UG97I`T{rcaP9$beE1f5=khF~3k5*fzwLR+my+NMRVlxYZh
zIXxbQd}I06^|g$;b-h+dms4k<m($@fC`>rL{hN6C4ZOTbUy$<xCcPk!$Dus_$Na5q
z%F@eoHE?Q;mH;Y4VH|#X-3-Amqwm%d?AZfDhF(rb5v<QW9Y?YHBMsU_6(C%Ip9Md?
z{kr|5`JJf#4UZuMVJWvm8Z;!g4v!S4p_Rp{mSH6H*VlDFYs{p-#jg9$SbTbPzY>ej
zi0)Hj@uQ;qnOOX2LHAFw_)J0fJF)oF1l^y+;!lt6S7PyJME7m6_%VWR*JJTx1>Ii9
z;<KXry;!^`==L=hpDpNiGZsHi(CuL?emv}(#fW0*G~9&L?Uyk#4Yw_IyA+F`2s=}2
zM6q-l?h@;E#F&|eJEyvRh{a<x>hs=^ktUoi==0ft#wa;Q(C29^Uf&H5VQGG(3!g=X
zFj`9U%ft0Bm1U0?=y5w^xQ65D@i=3s1dxjRwR%6AcxEf&6v$cEQOhv;y!|1dMDmay
zr`yc*xR){fmI31hdW_2$ws1T>rezE-aXdYSWeo2CAUXB(3w>QS2t3J2uTWAzPp~i%
z0u!FzucV;kTahMw8^`PInvO)GcIkSi$DWK~Dd<Ta{T#)_Gj;y-=YJ{a{DVpO-v@jO
z9x8|*hyTv`x5b^$&~vy<4{`Wi93Ow4Lod`*du!wRQDyky^Fo9%CjR=l5o-i4=K`OS
zV16ZQ*DOXqPAG1)=qy+)G5qlMQf4Ck)58%d)O&IB>lWbYp@aB&eFw+K+W~s+FeSmf
z{weUL1oM=>%QBv-hzqYUJm$+%3(J8TSxAKwN<@Da!(%hB%);n&dP@@eN0Q*XlHhj(
ze<4<21GO4|Xo`hH9IwAmp|2y3Gd_@quXE8H7Di)Xm`MH$7=El!88_Z80scaock63e
z`8IaNLncZ1C`s_Y;r#o!KK1>^0OOB#$ri?WGUtCLHZmrjHbxgE!OsLfQM)AINq*Da
zwTunfK?*-gLjUt5_^u@Qmy_U6Fg)xA9}ihtwL&2lsEOpXGyHIWx`^S&3bk?L@P|qG
zY)*oI8F(5$8@Zl~xjcQGe`Vabor1%H@vslIRtux#%QK;x<J+n=JUvTGp_$>ewDGzD
z_=~_te{ToFl7;6vJ^hV$V|X8UlM%tw^z0MKpUua=KF&>8C=&m+85TxK&!AJN0)8CP
z^YM>i#=;8Vsa=*EEexh-T`ByO;kC5!`YiBN?|!Z~T@NoOp??>6laT>W9|t}=f&EFx
z!j|Oej++<N3_qMv8>erJ8xMZ~p7fx<7p3d@T~2R`>)-LGCmx3d3_n)b6eo`xc;fG!
zZDB&uGx!vC0#Ea+jnA(d4J~}h`FP{xyby;#Y#j1_EavpHfhRp=$H{ps$M2UcjFO)D
zrtk~IB#-|77kjoJ_?Lkv{wB_U0;3j$4;dd07^1W>7$>P`V;@v&z+$fpe4=@J1<|9u
zeZ0N&`4S3qlhEJ9`M1UKzlG7mzbv<~@-B82gxfj3_67}q1;=;l^c=sM<6i)tjZfa*
zA8`CZPG7+3xtPLn;F;b^HU9cIp^u?ad-eAWIUB(OJkjg#-|6(NjDD=pSFZ8l(@yv?
z=cAX`V>%~Gq|Xc(M&iHeDvi&pyui5}pIxKjb^BR_1dWsZOEml<PG18&>7kPAK_6F5
zoKGzuSL|7Jl=%_x)Q{edTE;=Q55HslIa5Qli{VjkTzd}zpAA0o<EK9fpHU*o3Gdgg
z)vM2!tAJ0WpJl+4ek{DdF615A0ldi|pPqa)37@AK9(>})!(Vhf*N@)6=fl8~oV9Us
zRsm1?Z?jrhmGtfgg&zS=dg$(~Ww3s^pY!SFeA;-%i=2;lj)h^^yDCV1#PAFibNMz2
z{$dzp(vOLc6a5^*0X+3DeH()X*k1~F1D{CGPjGq*r#CTbJbRmjegJr~=kfi1oYR}m
zvTzPsGzpcG-YVntU&!z=qztff`glKedlLEwlHj{IpUOBp{B9EZV@dGialV4AVc`;n
zA1e&-dC|Zr7IM60l7(UDoiYm544>fqfxTLb_f0*%HF!0fzn#ur6D=!?xwBJ1_sU{0
zcrP~&GhLA9*%!_!wac=<Wrb{$lX91rRt4<Evh4P-drpjwQi*CV8dlHDVVL4!n39CD
z=yk|TwKuHo=9#SSI_;fd8Se;Ng`vzVV>0O#cI!X;nr$LY(9JFVrSItbxK3ZdA85h1
z5uAc7UstwBw%e+zWm%|m`S6OXA2$KzK$Glj#Ou3$ApuI3>pb$xM$ZaIqg;o#$^EjU
zrIn;)ud3GN&n&sJ5S$Yb$#joTyFVx6o||utpxxuEYiVj)gCYiqy?q@+D$8>hl~!0~
z>w<C!t4Yq&d<-m{JGaVOEmxP$v0I6PbRT6=7mM%*@M63mm*2Rcw8B;<&<#PE<1kVR
z3uGB@YI|18jjraEfd<*-^Lc!g^PLvUJiBd9Sxt>RJ-;a0Z%JOWH_w>Iv{cp7q<v_>
zT3P27yguA$t>=v7!u-Pg8Q;a|$~m~d>2~?7tBA5HU@peZUAGeggV#!2U5+N36L>Sd
zF)jyu4tKzBb6Tq$xLC`~)<yHH{K%dz+dO`GA-+M-gtxkfwWZ3_=(oCK(U_@zS0G;6
zY8(GvK>Yjn^3rHl%%=oo8Q*7c%1(a^8w0`$hx1m5>C}bL^`Lh_^%96X!`R9dE%nKp
z7%MZ)b@NQU2Uj@iWT(R)SQSg;_4wVbGMeXT#GCN(4b+;Kq+bgQW8XB_-_o`&stU~1
zJ`rGZL-%egP3`Kc5t)HRUfJYn)<z$V?CVy^i}*w=YjpVin1$Y!fInWu>T0y|Um6Ip
z3Xqr9HCDzc#ntML?`xSP@CD?0hr2Oegaz}f0yCU}HC`9S8AexyjBky&n%Ts&#<I>g
zo4;$6n+vNN-1Pxh-6BWxtr&>zMeZuIEUa|lqaIBzPko(ZO`HJikpbO!Ch4pMld#^c
zO`~ErW&=GYQDlRqz%+>!86Cq0EgO?^lfyl1lvFR#y4LU+j_;7B&|Im1MWM{&@zHk*
zsI^vi!lA<_SqV9XvZlI)-T<Biv9`iS`>mCgPH2ZFc2uv7PE&rRyAlLwfwR>iJMe(T
zuzuzKDDDdyh11I#T+UlvzBwL;4<EMh2k<lrtJR%QY30L}W~;7{hg?Z@MHxAgYWA$g
zkabAXSwpTcPOh@T1e2?(NXCCbwk^cusdG2W^xd1NHj-44kkgPolJPvt3fRH8krtiB
zhOVynx#G<iwO#j^u`QQlJj{p|!CVq+Uf72x5;VVZmAh(&%=^V(1f$_K7qE}bz$19p
zkkKdmA3y6yth*%Yv8;=3iK;TLGt5^S2ML?Bn5~*1?6WjURBtYt<MTM`$Qa+?@io?2
z-Dm=vitd@P!5*J0ZU%n0XEYbZ>dHz=F>|FbS<Hp*=4O`<)2^<?8L%y=lCR{S{h{<s
z?V~?dySvy*p(=2twvcJH)j@$o%|97rtCo>nis7~t(;kY()M|CCaIZ3(uSA**wM>=}
z^RM0mrDKhNXQ*UzKGvvwt|d{6QRg&#b*mDm^`W_uj89a%QN?N+m_hc#@Ld_Q9SIB$
z22)hj=rEt@u{GDZS`!REl9ao)*v|%`xv)x1j}CO{@KuwHsi809u$=M3@`5PIaIFdy
zIq}(=*inp+*EpPHmg1(VT<7=54cJvQx^(MJOAcBHMw@metOP#kQ>QJawDDTVmfe0V
z0ohk^I13;2SON_7s2r%I#aFb&!xui>SToJZx3#!key0m-?mCy>P*Yr=nv0Ax$%Prg
zO}4ki=aN^seSsDSyl#TA6wj#8$OyxD_&S8k#~RGPSX6<{2enFDE_t+tB(y~iKha`~
zMuB;{wkqS)m=@IqvbAcaTsgmLx~wk|v;}g0iGT%GlV?@Do#YcDI?BomH60C;ThGwC
zi76e<1z$JPyGV0)tZVq1G|H4sm|+|pRb&{kuFz+taqt$?s^8_P^J`_@MP8Td);Oay
zRFKUpu1c~J)CL(36-GA_3#vRb>Kj}94Z~T)*$~=r_}$-WMydkQH4AAbc2b~0!&(d5
zMBdo=^)YJ`q~zbBl5cgbk)3F90P_XK-9Eo{QQQtd*Ha-|(912&$;NgPvs7rN6%MU*
zVwOHhEK00tEi}vWs;2MOi(wW=*GabK0N-^js`4jVUeTA>$l78|5A8ItlS;DFNa9Gb
zzJzt<qkujE23IJ!{oa;lr*5?L%?%XnDM5em307x?4m{OzOfO=}(0s)11&iA{8?EyE
z*wa=O(*H&CG1ss=o#}&#^EAm0Cw-7hc4Ny<pLN3%u`8pav(U_JT6Ci&qgnOXLNfN`
zvb6mz?T#FcjUH!$!KAGzFf^-V7rtD!21^?+#<s`jSm}z>2CO6_8NOF^v1e_u$7yYq
zy?B<{gQt_-;I*n?1iMC!2ifgd5J2AM)lD%vGPE(_m%C`gJ8UKh^5W*zP$`y4{+1>(
zNH|NHo?k%BKnx8y8eIvNv;VyBu4k+~)3I5z4nHWU@)!A6JG_ugUt97+1XxI?*RZoO
b8_pHjl#7n&O2?W;te=caf7+D@nt%I0!DXRr

literal 0
HcmV?d00001

diff --git a/benchmarks/Grid_su3 b/benchmarks/Grid_su3
new file mode 100755
index 0000000000000000000000000000000000000000..72d8a8d0170e828d287c6b87996116ca96fb2d3a
GIT binary patch
literal 44845
zcmeIb3wTu3)i-`}gQ$=R7A;;Xqeh#E#sm<EicUxZXLJH72{Z^e8IlQ%B$=3*#NY*k
zNvzW$O8eTDwzj30x3$(*YHbT@A>kI(s=+H>NxhyiB2f_{73TZ>_St80W-=4szW?X{
zJ<s<%Za8P{wf5R;uf6tdoxRVQ++O0Co1B!S7$I4?N+DMEb(<jRcSXr=7pSDoQchBA
z%4FqiWeg~3xNIVYP>&gKNQyRKy5vg%OvN=TStra&wuy)V&(_H(K@Wz8ge-43*kn|P
znQL^^fQn)j1m>fzdv6mcV3mj%u$u}Y3|Vw&<fD4bHze~7$$SP>r2Ymp>c_hAs*vdw
z$u?lbZ3IfOxDP|}GL*gRuMzQj9arjYdPKtY(hvqT@_h^WsOSGQWJ<<`vb;og8Ezjv
zgDhUTw9d0)#`H_;YA&hsG&Hwf(mHF#B{Qby`n<VQMZGDXyz}NS5*ZCUP69@qC*q>X
znSTB|xBPSR1-EZq`dZqQ-B0wrSv>jeFPN5#%V@WNE3?YgQ=CLv9<FJ)_O3hmxt~@4
zWB1+l;k_SaUj6f~CAN>>T>5=$cFBv5i5tfiDXY?c`}4CKx2&<R{u|iihTSOW)Ho@N
zpkzGy3s5K?{WDZB9=$n%oS!71rzMa-Edf96Y>lJ;<OK9N3FQ1cfgRWx<JD_k0{Ui@
z8;`#|0sq?x%H5WLe^~-L`-2r%Y<MsMe_sOn0tm3i3IV<rSG;;XnxMUYmw<k60{Vso
z`g|h+|8ElT?@OTP(FAlgf&D$;x8jNoPbARi#RU3Xnn3=W5FF2*o&<8j3C7o%3HaL+
z=(#k3KDQ;1e`^ByR`8#zoT;?kQ6)@NDGhZW<dd#+MAO$x{!Qz2dYj}=ll)aN{ByxS
z0dj1eI=@x&uax{&S#Fu6ua@;4oM@929c+_w`XzrR_!xeMi}r6k&4&AM*emtymin~G
zq*b7k-+H=DQY64S;Pcm1Pnp8Ge}>|8uB`VqIDLLsli%r7oY%BC7rIw^e13ORQJu@@
zbNdvhy`t1v<8G4ViqfJwZ-cwSwW7|gr^QckHu_gJxm`8R)$TRUYQ+3tzixiHZ#qh7
z%+GUDsL$WzT~o5~+H#k@x&+Epx4N8-%}s7+i>Jxo?5bmqYs>wYRr}X8y1|k^4cTt;
z)VLd*o(7M<q(zAIOmB3%R|}-vpFh*-M9JQpoGVbctFF#lt#f%VtF3GHtuokVnAxgb
zK0mpT9ds|$a=YqLgwNgNw^xIjU%krJ<n%YWJboYaC|L;Fv=uI&r`q7QS2!#3oF(Nm
zoMqRRPjd>@%l)%FULO?kxu@HqZ*jiV*gpemx$4{zg0CpQr0m-2h$UEU-^^-neWR<%
z%@n<&GdwFBydcPWH~C5ydPIspf0nKdRQEKj)GH8Ab7AQcl!l^wtc24J4Hq=J8|u9^
zY@v$9^U*kHiB{JNPfI?{S<Rl{WD~$7wO%w-q#dW`Im?~y`bPg6XOp`T{m9$oTIps}
zu54{}HoBX9-Ue5l2gxmYGMn!vS0n6)j)k(I!DVK}S2Wiq)*3m={nM+x&Hj;j(OwDi
z>gsyyd?hqh!%g)fb5td#<~yA&^`7_@^%kJd$*xetmiJa*puhsdjaxibwupCHW3#`6
zja5;Bs{Kc85=SkxgD<XD<$f4mcJn%S!%F`ur@N`i+hlZ<Vbw&xLvypo)YO%=D3}21
z^|6oG%Ia#Tk66CqT-0!rr=bRe)7M<@CR?@6*68(lTAlUodbH!p1_c#F(>XaeV9Ka<
zdFqr#bUA;m9KIOmtscMP^Q;uz!sqrY$SXm0;~Hh9+wb?(yS=qFt~F$HRr@{O24#h-
zdNsyZwVUkq-WK;T#pzp9zXA=25~!A(t?FH#2Bo&i?N&U##^#1<zcR0=$T=-HPnqYi
z&na?F&CQ=4iB8SEOpjMqf@OMcfl^diSvbe;%+H+`l@QIApKngY6q*b#I9c2YF}Wk?
zu@v~pMvOG3CF9vg=N>E47}sTt6}gj@bdjEn=d&^Rmn<%F>9H|N68??Dnr8ToWf!hF
zMM17NKx0S(XO2wI1Ag!Keu&44lazNQt@%vPNz`UA!k>mbx;*9&+@1rK#wuS({^uWB
zjmNby%E`F1%ubP>rko~m*U*o^pQ@ZMaevPD@d%ZyOa;y`bbFP6e6~TskzlN&WTT!&
zFxExV=@S~kSf@$nGlUV0b&>ST3?iXi|46^oL^sxH((_GpW4$DOripH>-=t49(I>T4
z32ZFxa!D4r)kJ688X?<6XHPIfo{27-L?q2J(OI?;Y$m$-c~Lddsgn`POmySfmT8qH
zy18AdOmuXtNLXc}kBd;@f0*dwP4re1J;OwAGtn(3`dubE`<W3onCKG>BH%_7-Ry^M
zGSR<g;=kWSpKhXWHqlQr(YKiBr<>^ACOV&mjj-25Kf@pb_L=DLDkEXPiGEguswn*?
zy0Le{WX(jMVd5V&(Z6A$51HuanCQy7DuJ<&@EOwx=_b0>AOdEZ=;xZ~lT7sUO?0b?
zo@JtEo9Gvq=y@jkg(mte6aAYey3ItNWumJldbWvPW}@eq=#?h=MJ9Tci9W?dUuB|S
zY@#=s=$Dx2YWs)jYA|h4omEjgyZvd}uaT&>?@HfkVnQ^0*&MKhFDS;{GTDY0@l|9B
z^@kBIxSTkrm{6a<rxH&l-YxKph^G+WEb#M)b1DgK68M?KIfaBa2>evyoH{~n0za8J
zyF;i^;K{@}RfMVperz>xP7$Fpfqz2$Wa2h~e?Xj5MJP|;ZxQDd5wZ&WRpR4`XA1l|
z;+zsfiol;F&Z!_YcmjkAb`a+j5b78Bqr^G%Lwy2&m^g=gs9WF<5a&=2Z5H^ih;w@?
zv`OGUA<m&4+92@n6X%c)wF&&Y#J^6wQQ&tF=g<yS3H)Z_9MYjOfj1H7P!8Dyegkn1
z;ZUByR}kmW4Os<#9dQoLP^Q2a66cT%DFVNSIEP|r@Hp!~hq#q^zrZgi&Y>6T6Zll(
z=MnD~_(jAy1VftzejafSz0f9spGllUF0?`5rxNE-3$+RSWa1oRp+<ox6Xy^LRSEpq
z4Zt~cLS+K~g!o0oZ36#*_!Q!Kz{|GRbXF<a1*qC~KFe++`e++QNu{t)+4i<Bq^CvF
ziHdY>0%(?RcUvZ;J=B6x)0_4P0K@jaDhRqO!l<Dfo$=>7<FBI_HyMnTp>5!-&9zMK
z5OO%a%C>J|`JG76^&=|vqkZxZ+5`KpR|9XU?e7ehRoHvGyWKV@(^DouuWgb|(Y_7F
z!?l*l#fq}!9#n0SddXb?YWq-z8hH1n<mEfJus{y{!^^ibDco<F+(z0KVg`TDPzw0f
zzz((NgR9k~K6T$vGcrF)Mz!-fWL5)v^!#th{B5&6^;Sh`p0EfxwYBIl%lD)$1}W*&
zovfeQK6}3#4E}Q>Dok9oeegDD7kD?^w>)$UvREc7A)gKx6@I&Lap9uEigI<`hm$Z3
z1Rl>C2cQNX$$}fGs9h58(sAu6RGLH$QwTTtlt+_s(|!ZzNVC8Lvj_IrgKN|6Q?ANC
zs0RM6_WV1!)^e!3)^dACRv+@Hy&YNofDHcA)WCkVN4v_>IrvTFcfzh}V7D50P-g|)
z&n#WVKZ8PgK6`M6J-9es&G|$PJWuKVbTx2D>-Yj$4L)YtPS0?gt=6VIT7#mraxw(d
z9`slhHE7e+;Ksi8Dp>h7OUG>&DoWcevlM^YcZ-S!&QJpfw32_5I|ZD1+fGJy(87I|
zZG%Fp_C93=dbLXd^>Y53<#14I10rWyah!4{k1XenW22RGyDX;{GGpu0u9x#Vri$45
z^kB5d)TepWa=tFhaYoA7#rkm3(9Vj`_K_wUdu)XEWQ6Zil%VU_M;aF-?LdU~0%;kf
zJtZ_r%Zj7-*GB67kIzT55A|g`{4P??+brjDmQ$^l^Z3e<_W5hp2>UD=wVb1rg>tS2
zEaZ-y;^_KKRu$}H{|6B903gR_q_1yAA7o$e24vXIB~80cND@6<Fm#n<vSX8}w$ReB
zr`FQgt$b>^y+ri)Eg(Ar@2ePlI`_R|*r{FX#rV*kMMc!=j;u<S)!ALRodx+XQG5DR
z)vk%%+uDF~tY%^$Yo`u{!=2rZppJ^!!C@&GDm>C_v2=ch1;!EhP1Y>bNv*z5N_<W2
z>c~<-Vy&PRDr6JRxu~^2PmJDpW&QRj%Tj|6ilX(I$Wo$lU{jphmHL`cG6^K8PZD4m
zpzX+Rs8b7dU^^QK4Ad->OwpQIL5kI^@Ns4sMYU6eYK|Zz!YYOt`dI|!8LByw`l(VL
zNr2gag-3QlwWq8?wMif_Xg{5!D4$p+6lYBe>SnVCG<&<&W^ezntt9`6&et3P2;Z@d
z%v*W0+`SHi!&sAQE#5h8wZ#_iiVcXa*o+P^v<bJq&n(%twqi@Rx&hZFM9MZ7TXHJ0
zs>mEJwk#P!q;c@i$n|QWWyzbL6kGcK&9p-$mL(x{rp8y8_K9W5r%d|DvgC7!{rJz8
zB`<wq={v^EyI9aWD5wwOUq!b4g_i6+mL<EGf$<OoYnJStBw6}2OHN&u6(#prmb^!n
zT~xjwGWsc~`!mau7mzymNuj0hC90YS?4OWI-lw3vBdbgh1|jSjqBg2|coXj*en#Do
zK#5O?zV&Am|0Lw^mg)?aSSpW_MeXMOM5Kbdk>zvClASDM2u#PBT-L|*=OAPW4DbQU
z`%GqE(an@EHlcEs>|Koi6M7#)rC7s4OXWa`NZJYA1r=hoVoN1z4q?YEOP&PZ=inP+
z<p)vky_--XYoA@5RRuX<>j%#+OJ#9Zo+SrFaIQcLE_nh;hA}E(4BaG1*oTBZR^U05
zRFP#94Q+$$7id^0m~HE4wVnj;UY6F+rhNsuUKh4SwVwf@YLHCNk>Rjl0Q(DI5`E)&
zFue?>Nle~}Zty%(_CnEpmdckvzorjGeawyqD`vkA<cZIa{{*gAJ}E)pf@;T+`Z^ls
zBM{$2FJs|vK<*ppWK!D4w6n!~GV+J0{1dS5Zm|E0y+YUoHhLF?k4bpXQdz@J@dgMz
zB<#0TF3Ym1fum|q*s6A28CHAVPc61gam-e5ub7?5n-$SpXQOvbSvK2dnc|;aW|>kv
zyUH@9W_F`x%6+rjFv{d$4EH#aK2q21*Z`86#(Tjlmi6DkdRG|kDO~rt#-Z5xZQ$B-
zSIL1~1vjD7W!0C}-Zl+>1XnY$xD^MEXn{|-8tmQ{Sr6%d9R4&MMg=Z<26F-Xc{ch}
zWj8t>pmAl6&n;#d2EHWiLg37O;bC3mIV-VzVKNuQUs9wc6GQ$J%d(2BNr<PtTSNiZ
z_dy7*2_h|X8Oz&=<nhmlWFa3o28DhZI;8zg&%T3YJ`C|As&R>4rcr|>Y3qk2N-5W)
zM>BsDSnYvb_F!rj9gu5+4_XJQdFMe($8Qm7yA2*y=66eq1}>-__Q9D``xqq$vBV8~
zSBHdmg|$u#3blH*=V@guR$@}_4C4v<TuBitNN3nsf{`*7DW%{^cLc)PF-WJN=gx;R
z5Fl1gA;5-|!uCDMh3zNKu&iH*^>DzV4Gf0E;>k92G9Kb$y{vDDGSga|R=qozyv8+n
z#*6W?ETUqGyv9FM7QBeh!KIgT0h?QRjeiw6(=y_ebJoameg~Zsd#83;&i5ka412+)
zdO5$neq??Af$GQ9=ju_*`4?rOJ~@$chP}pjkyEc@WRz(8BD7(z@gEV|u-Ewe2yNJF
z`~hefmLDx2skddM-rqSintiA*^}aDu&fB$y*LaCu&Yf|}`SY10>@#`Pa(XBW<(ve#
zg-)XyfS2HaGq)*)4nP}7O2l8}hCKa6MLHI`zgQ|{2!By9NPm$`YS582sn*iagIU((
zGY&zH1M??}(R0N3j6X#+=re9)d0*x;zIkvIpRwobEa{P~R@8=`nJ!WIjCX<bAA4pu
z%EID!W|SA>nQbHqp4kRK-80LFJ}_V#2n<xMj7-%JO4Sb1ovi|CglgNEfli|q-*<lL
zL){LKL^+LvEQ0dz;CWDCPDj!pb!sFHklS*FNA|#L`~fzcpj#yf44QkETx>GoBhzqn
z7wzpIwdFri%MjVK=sf5Mpq#fJV>aR6Tkh_Jl&}29ul&ca{Ktg;<C|*F3HXoGkElKG
zrIuKxjGKvD#mvk?-mQq=Ix`RVWixG{_-B^kUOcl3_nMiFxZgLk4Z}?i&2X<HX&~B%
zJY-ql4xcgHV|b8@=s|A4!IF3stin=N^lhu|GQ@b0YS5oGNh{H~HpKXln?qc-uto%S
zS{`~0|5xukOWlZhC8buiG<4TmR-}!`-bSruS=wm=-U=EXFv{puz0J&!80A7<4<h_R
zhj1YsS($Vpe^nq{NVxuV|KY4w@yVqOUS$vE{8Y+`tlK1>4x(r09@qr?uAf*g!ezLD
ztC5t*4ZI&2gd2Dg-N19i{v`ds_gT+G{+}ox{vRy__f5>Ny8kCBVlklqr!S=R|By25
z|3NwyK<zo~*;9a61$F;VlneW+?cK>@F>W7%Be+I)1mAiOj^KPWYG?sY8)6;7iK6hd
z`bDx;O^)ENa2OHeIE@h{C2|Dsrz|)IgW%H3DW)TMKCGoZC~~Hy#VMzFWH~>F&WRmC
zn=I$9NIAofV6k4#&niaN=RH(Eram)9EoVPvp+43~Im3?N>&O}F2=0y0h8@92BD7&g
za8raf><F$04WsMb@{xMa8maf%cSo}i^`+jcBIUeYYdC^gdO5enDd$yefyLP8^ij*%
zLRlzB0o1n-)vo)p`jaq}vFRiC#`CbK)Bm$c7eV@eBCGKK1cUVd$b`L%*!{v;lo2>U
z-%sa`@%?@Q&FTB?hd27AzTaDKkK+67#(|z$(&Om633nbhN=Y$}o$j^W5-mw=ktN!~
zTp$Fg!S%y#m})86s|Mf+AL)Ucl?!3oNqFBwOv9PMM?_Ui1IJ1OecY5Q4ZK|%c*Wix
zYEy%^avnI;mVYqX-#VfOh8%$>9f2O<a#`+PO<PC&G3;%8wO@x_v#<8+zS^&g?AKu@
z&fbRGw>IFy7UGmOn{ivxZJBaopJhs6zh#PR&@$x=IUK_~)pb4Gii`HCKDDg>1%0aB
z>yB#}a+b~;ft3bcFAcn`&A|C`#Ip*OTH|_F+?zZ2Ca2mMzpAo794-!gn#iZxny33z
zJp!72s(X-egim!ch_OD^d}M`7l*p%=C*?%>R67nN^r^lMetTfIUH6f=wU#|ddvR;+
zhlmKDDvdtXne@&!veNPWsYBWoqU>I+9mVOBp74cUKpq|~b>h@iUyB@p<8Y-Or>xF{
z`l0GW!kP!P?_{EIxKd(K5!pX`0~$uQ(q`iH3|ncNKol>HgqEg{^q`&-TBc2L$ksD?
zP~Sgk#EQBLvJ!hXwX6(Q)IUL{?p>^+2X!87qurpFv-8@K<xC%0&azR<IUx+ttIds+
zGweYn>*XwsQ_f4MevF53`1R51^FvwAo6ttD&#(velL#%+eY86xv|$gbDMB0epstV5
zhCQfjK!XSM>uW~ZCwrvcSB|PT^<_I;7%AuNhzIqS$eDIUoN}JT$y<zlUU+RZ`?Sb%
zwn1j$cCnY%RhQL>_ZTAnQ#afx`cHj;47&d`US!;<2JXue3|&Q;+?XK~{ij}I?+ou@
zbY$(Nh`=GbPdamq`&0$B>GbezEV@slIz7LBRd=7_Iz4xC^UWklKTl0?aJxiGG99Mk
z;8rUXC#6?XQeat@)g*3IAL<k#Ev^rhCw!<6F!^e`FOU=NBuv_(zIZOpb9C9P`UX|R
zAL=~f2<&tO4$7^mM`&2{{*@YdRSkT!1v0T^RgCR0OZG9GA`M!yU*K_KnI*?B4#>u!
zCB!k}UYvH_&<{7ONt|Z!MDax)S-rs%Egr6klfFKjM)5Rj7n43>63zhkAoVy-w0bQ$
zI7Gr39gfqzXOiPnoOivARGiNA{R=8QYpHw-X>0SC7~TZRJ}C7x4!+)o9^ub$ruZxh
zexC<l|3pzwBBdW3C?WerB=%aCyv+4~2M)h>f$`sveS!oW%vR!@)jPima`tV)jR#;K
zfaMh)Zw>Gu5$9cC#pxLcIF>~HElXYnL7a%asULu~@5xl#cjT$zy_U`=PGD$srQ#^8
zpIaFdQ@L4@%B_r=RBmNBQn{5;oNB}U)>Lj~)TMGO!=Ea)GE&=Mdo|o^G~~M78%S8U
zlc3<JW&IM!AKuBB0!-U;zG=N(GuN76y*;7b_A)1co#wN!1Ddv<>+NyNLkIDH_3qd`
z4112g-hNBK;q~@_Hu)7@+GI}VY19(0;`{{COzW*U275Dml=b$2_A;`J&}t!wp56N)
zHD(Xvfc85iiT9Sq=qFkC<>|{U-VLIW#8fqg?*^T@ff{AA(s5o6;;zF#MbW+5L_$1c
z<Y9%yvg$^b($0fwkRz2-7tg8=Y4Zj79U|{*g~FDHv@@_u;k_g-!3VVUFL9ym#hFzY
zTNN32;ROrrOb|o0XtmH47;joS);%uIeK<MPi(3o^QuvaVPALW@zgu0mi|=?XEnHT(
zyzu%$=klG~gRu3s2WjwjRCVjKhza?y2PF(2zx_(el*ezV)CN!@+b-=)#=9m}Xg6&!
z781Vw!Lzr}3CNYwhBtF6q%^si18H;2(k_$IHgPgam(sYC6WVM@<MSJKfdcO$?-;Uk
ztNh=(<8tIh;?jg6BG8Nv+}h)u!=V{nnSqr;ExnOM(?HZ?;6_X7?eiyr$oBcAkaKHR
zqjnR8@XfIdt^{bM_g+9-@GJb#=TL88T<8;mw9tT5mX6z7h|^npl)Ntj2|cZ+4m_gO
zlC}oDUt7l5%Zx2#>@v*On#$PQj9tZ8A(l#QI%Butd04xMvDX;0GFI>oVy83qK4aq;
zE5Y`JmdscM&RDde=Mj6Dv5y$b{0Omk7~96!tBjqElP#@}v6>@@?O|*OV~;atJ&M@F
zj6K8H{fu2aB*sv5KYRu$(fyG5IZirYOzmpiLSyv7H}JApA58BfwtqOu9(Yj=e8$%Z
zyKcd7ct#t4F6R#MGE#d#%!ikf+WREzmXJM!^i2|S2orD9VNoNHqAK3X)ZX?%P#MTE
z3oS^PE#V{y(*eP%0x+n(Dcv3$x>0kwI;B~`u<AL1{Y0NC7T!YfPLcmCsJu#Bj;`wn
z7O84b6t4z@cR_%otGdk5^_UQw)1&r$p6tMA%Xy~dRCFe8N@y3zd;{Zsr(DyO+NK8D
znGs?(3J}~x5R<p-F(zsu^iGy($}70L9;0O-LI*+W87%_7(XCmT+Wm%rGkj186sZ9-
z45SH)8ietJ*H%X7&)_t6gOU4WU%H;LsM68(>%Dr}ItccSC|l&sHS!Kuo1BiWzx3<b
zbr95uMb#eJk7|c7-Ne+KoLG;)k%|=!P5aX!;CL;ow6ojNaSGbgf%>I8f~ByFZ<QMS
zLmSddgNq<$X{H*yrOXlh1C5G?qt#0C!}M-U=W|8D)U9&@soUlRivOVic@)U@Il<Iz
zmWSFE$PbA7oFGy`Nfi{{1?9&@N3VRWh&g{e=jfH&k##GWsA53egY91+KF9J<>h^$$
zmS8ctgl}BT307?V0tppczL*o3xWyhE1Il(2W#Ak7`^kd2H+2h&RLEsS!@Z>iy_P#>
zV-YD0^f<6m;d6y{(IbOFeq<am?=$R}r}NE<4Vc;;!E068^O1AW7xY%T3!RfW99@5~
zGGnk^2f^%kVh(9`lbHHwG0cI+<~W2H9RzJ7#GJ*(PL^UE@JutQ(z%p%szAk{p<c{e
z+W7<?n-Q@GGxOoD?p53S`S#ZHmX4pmDvrQ2YWsUbYS-E!wdYU@-dYa#`X{S_&(-#S
zB{iRp+2%Pgr|q<gndddc_YR!?DI7~&3p4+b4iTX3#Py`w9!~aQfb1D4R=XSz<f%Cy
zf$3X)xGm|&%{7QyPOW8@Z3k~p@rIY^X4#%r+k0$k`<}t(XOSDC_T$>elvgR+a%wl@
zD5G(25@UO7Eo+~(7ku2D8W<XARl6FJ>;)k;FkrXLdqoYrqUL-I<x|w0PyJ&tJ9R$c
z&sN(%Px7CEdvY@r{A?iA^3atJ<9a|1V|V!@VrqC#U`TEMcWTpxfg$638GR3I#SeqO
z_~CTk#zD_iN5L~q@4&GBLVGZ;G&r_2I5Eo}Tn-awY=@<l{DboSh94ij@=t|<EAJ0Z
z{8ORjp@|O!CjKFi`g;`jClu@`c)ICLd=xXu<d9&$g)8br2;pkE>{;_0{Iwg(xfo#n
zU1dWLc}(;W;b7G~&6Vac(enhMTqlSgEeLaTLO%ryLcUHAGm;=!b%K~41;L^d!~`G+
zC;ll)X^Y4>C<wQ#(j^|yu`(Uor(@Cs+^J)kOdGQ_^a#+UJ8L(H^{(6zoTti3={1<o
z5xj>=ID$W-a*p7P{3E4-o%Z&((QSELa42l)_!>KO*NvI>uIq8#oLL&^wYR@N%pkvc
zzyppKPq|+9!IkWOlX2a`$P`@j8JSUQx%~7Y^sCEHe<JC}M?Vjs<E@3DwNL-}=yAMY
zg1&nBc|9V12=DTPD;ZoQ?hxouiXf&KM1g98|Ko@}4e~zI<sD<z&qcZ>1bl4BJ0?&~
z2>95LcMPZ;CjYLroY;Zq_7mu)d#q~vPF%fKxW_}5&aYu6a|GT52dlf+a_SR^Vw}uM
zegZkxK4mW$Y)&r@yb`*Zjn}mnBW4JF2_t4dUt`W0<Y*b6{{$w7Bo2|I0}~5_sek)%
z;n6Gc8tFq5A1+LK9@#NIlAA6^MD}m+Y$MOsz)Rqo{%F!Og~8%&sUY3DH8p@2J0Eg9
znwo^+*qiJq=xyo?>|Lrm-ZFoF_s-$N1?*>h86Mah%|Glf;S~5&d*FodOGBumJvcYr
z9-N;kCvp`pKDKkmLfsBaF<f05yb9l*IC74uJ;&g`oKSOKQVZU+-0>*63SNct<*>{d
zuqVA}`Th<^!6D1{d(!t|=iyLEK8F55ztvuF-140tf%%%QEB^?fXg`07PT(#GX6cR`
zSa`SEb3ED4)!l&-0z0%<p9D|Wl@H<#^`9|Q{t<276Bt%WZ`ixeelYSp@n4{o1}?M*
zu1~+VEA>HpU~cBXX==~g$@w^;>BY8!HUS*=r0df$fux%zknP{W+@L+O7yjh|6lJz9
z_8(043cM%`hNr<o%Y-rZ-np4n2Tu(*TJHKL_Lv;i14hNvz?E5A4;2semIkg`(3SO2
zX`oiDCYW_5W@()$4wQ6z`00RV_7B>JlGa`xeyXT9^&yB<0=s(-PEOjr?>HDcUkhlr
z9|-Z6pv($D5w0otLI(+f14j?Ehfn;2?e!>cmIW^-y{d+H*tf#p4ub=;<UYr_Pvjr5
zpS#b}!DcAM3+Op}k;}5+P|_Qm-`n5CfZ2`Z?O+P57vAkZNezU!y!j_tveQCmC`u9D
zV1MYaxWBWF#jEYdlbe=Z({=G552GBlYusN!gj4XeBk4FqZ`E~qHt8u9+xeK7FnS>B
z=MWX`A6X`pD7Xf3_2cS;9^DwYgU#=i26l7lE%?OJ@g0aR4up>GYybR&tmVErU5<;A
z15dWUm4a0!6(2)QX+M_aKdb#%vi}rXlU0Wi<44;+Pi?yF+OBaA6?QF3Qn!(p7sgkw
z_^q_w!Y_2w-FVpEzWWQTzJu*MzUX}J_Wr=`_Tx#-r{bln*lwlXfCuIIxIK#VMP=(!
z#I%fF^s+T*dpKJipz`}Jd{1T?qFeYinh7slT#ipeC*j*Aj#a#;7kE<*JgJ?A4;a;g
z{r>auYC<|AXR5(GwW}g4ldjH66vM6lUab^mk(|Em`C1!Xm!sOXNYk~&DtbM3I(xNj
zFd6z_z5G=BJKh|CP|FXxvE)Rz&+vG__g{CxDE7c%oPgSc<Ff1pj;wTlnjNRB>1x-q
ztjtFzp@Muu3GC7QyHRY{ty$w@<RKa3T)PZfsX@4ucI`-Ux;4$xnGe}&(q2mkeLQ;*
zRj^-y>iN&%hi+rw!;K3hW6LYm9zas44<(7`!P4NxS!$rqk%Kq$hA>I(-YOc-QXEo~
zc3ZyR&5qm?y2|o{p7b7d3wtHP)i@Q_upb~!aQzpX=moni-x15XF=KtW2Z{!EvKXOM
zY2bt-=L37-WqVH;#i76Uf1xJrhlUPln9k|x6j8^Iv<b*Mzw6e=QX|_12X=9WI6JG9
z4;9$5`-2^Cd}3wB+n6vPvS|llHtv@rE1y7~;np6=wIGLjxbt=~QbO0G4xxS;JD7IG
zR4Z%_F=}wUmvMf)7I35jPg^O|tZHv@R`%v~RY)z)YE-*gwXOW&0=3}T<_eM4*8vR8
zRi&N93J22;=R=efJIHuMEJU%L1$(Qu44Prrh&{tJ$D)@GiHBYKmS>2`V9?T&9$25%
z0Y;q96AX+ErgwcS9LR+KeNPLj#EtbfBxoyPr%(+kft}E$(xl7!C|!bS9Wrf{D*CVf
zPu017d_<MkcR~wVZ`eYgLpx~oT;8ySo?$$Jg)S#M+VEjO>Cfq3Q|6(nYTzmKrNCzx
z?D~Pkt>D5$*$n^IG66EL5FfJQ_#s8@vZ0F<VVeVI6vErwumh(-nP>`auAb{^Gz@at
zgIDW|)>wOBEVJ1I^U`t2Y)g=>H*LW*d=tuM#F@w@z)JQ9?1JgN|K6+(pl$m;C$W=g
zg6iF+U0Hk7u6qRoCKgBFJ?-`k)?l|9#OuNoW$IU!&MUEl<|ugA(!p&kN^TRdBCAyw
z`!Lj?Jd&lnOCiru(C?qZS&otO#4PuC)?Fg&y;*&L+8SgWI1gPT^t*IjvXQ(O$ztNu
z775{xi<+@5$)Q!TBG6T5A07jiP8GT`r44M+wOGa!D8Gp*fj6}ek5OxE@^O}3wQb`K
zFp$<mT18f)tOmbk7d<y-tXVo2(`_ht+tRT`6m)A=CG(Z(Vty&bR7x@DLJW<7XEWUz
zlZ7=fNNF}cxM&T?GBBB@I6a+u$}~2tZiiAK)35`M611hX3}yicF<anQ(2Qg#WF~JV
zSyrF(_QB8f>Z{b-w(S?JK7VI`s4DgL;NyN-eXyCObwV#u{UfM8J9bVnw!1O9@_-BV
zGbt!tm-{^_mrERb+XTpsYmZ5^#}FiF*CUH=kDIYmM0<#|L8R&Sm@Q-u+hdY;EiHtK
zgVC(&M^KeyC}bw5ne3r<`LnFrYw)ww;H{_!xU*HbwCzKwmQHSS_ojVtsT^cty!LM+
zt3KSK#d4nkvnXc-&Z1)H-8^YoI`}(;=6UpD;Fw1VLjR3hH1{7z;odlcd($Y~!4cg5
zH469I5!`o<!hPci?lq%u*Not<8-@GZBe<QTa9=Zmd*LYDvqx~3jKa-EH7#@@F@Zdi
zMK?!V3&%b5GAJg7*ZI*hrYym*oD5HbVq$po3lqbYWVjy`6T`#DO$=v|fxq-%Vwhkq
z;@IP02uJI1vYFu(G7LpC>@zcLA;b1)CHYBuq!vFS!(XEr_%2U`p@j@xpqP~Wftg_u
z8Td+dXc3JMw*V_@6{;qqi&)FcWm3H5sN+da<R-Vo(!n8~Xh^++w9p3}mGnSj=kbo)
zMhM$6iZG7G5T9k^&Wau4t(hpzxEYy9vvo)tfY-h=^n1jF#5gNj2cH_FtGi+pp{2U0
z(0_A!gXq16=-73l<4+^x(v?f3tQTnm*m&3(`e0aM^i12)H$viXN0InFT~6pj!0rEy
z<wt+ys)Kd6KFZbTr_hB326cOfj!Wo422<~PjB+r__9Iz4iaf%cCWfC#hD<>1B{JL<
zErD|tC3I85U&*l2%%J;_{Y>0QmEw699hXstsp+GAWlrvLj*s${!w|}nH449iB_BjA
zqPt%5BxKgk5hcg2>K*)?=|3;_E?KOVl6T=*YG6%Fne=M32aqyi<s8iik6%r_7Mm%t
zQb^<yE0)n8|6ModCZW&p>ioYg<9?QLm_jk9jDyAS$QpTn#r*Rp^giu@BlbYp9z3VC
z`kT_fz*FkBQ&!s)tsgSxcU}B8eHc8!<3aZ$js7Ecyl9Z<e&6|fNGmP)&|gR~d0I6q
z$p?+HDjPoqW6e`>@zF4Q`zeU!KSu{Nh(fl*xsd3sK!>%#L*>M*ZD&!LawJ^SHC-r!
zXNCjXa~#`H13bevYo2EhUR4;J`6#y~wsJ2*yY5pcvTheE8+oF`7LMtGZaQof4vPmm
zM^-i-e6t7WqYgju#bbzc70W$i<g?yh6rtVnH}v1bc(gS!tTHoPM~20q3|t{*yI4bt
zLz<}K7mU=e2w4aGGJli>{MSe2oi;M>HzS(eJ~cAncl}lGFRw$p;q+5=pYV4waXS)e
z`o9Cp{apTx&L-5Gjh?ABkO7UJE*pIVQjA7VH#PbXA<)$5-=X|<yC_F)*hKnk?EcM{
z;bOB1x};;9we1nSNuydW;)g&0Pz&Dxr5!to@6uHE@Em66hDGJvM-EQgei1rZez$h(
zU!Y3Rm%05)9j^l(^rdev)6wNXx3}TZavPsXw7Y=@b_)W=Mit6KlXM<jdq!k`3{Sk;
z{gm0hGd=ut^I(2h+X~c_ezlkdLjMIS&PO8SlUp(zn_rX$J}b?6*WPnH4f`%@o9$h2
z;)lnx=mI``F&Wm*Me*EZ;rK@t@K%3V8!u^MwGK`b-;1LlvaxNzhF5XI+Iy0A2Wa;8
zT~@WL$coLz+kXkQ@wELqzsbz$$BtAAo(7Wq({WG6mR<W$N^_+>I6aFpekEl1uSI_B
z!|OEK2R{lm(@eWrr=4FKI6QC;w?-j<;52d!Sa8oAI0^Sj11U(yPl*f<S^a!h*r%er
z4BK{?w@SNiO4qjrup!VpZxYnUA(1m1`^)pPL6}7X9=Ei2(OPy+7WOIld>=b&*gmk?
zgA20l!EfiOIZx^*0(d9kr5608)dB4X?{gZOXB)twJ~o5b*#ism3VX*wic;7+&q}d)
zBhsIP^S}J(%=2semETKhK1U5s>rLgG%XkrOhgz_sNi%j=@hG%&`wDcXZGPOuPXulK
zDPr1R{w#Le(O>jcp`ADmEW{Gg0)#fc3bu>h&-}9S@SmoaZGT{;p_~=^Z+(l(osRN#
z`<v5qm_D@*=|?ulOTUEazh(MMk@WvmpI(|AHs>At${ToIWVW|yt6A)y4)5=BT-yW3
z9R&lH4!)A*K=%*cl3t1tm4lx3F3)M~!Q0R+@S8+<%;k1#!S0rY`LFSPaIAOOf#olC
zy}-P9&*Nd(0Hdw6;HaNn#1X*J`diwEqM2@!N3730#710>*^qmoC7GpN?(~6Hdsj0?
zL<v@{9cs=ooL}$_7_}heJC~c!1w;OP2-6;hjiioJt@MGjPzP!=RhD))csN`xP;qk2
zS7=V)DE|vb(h+;kQ9pLmdL4mdng>h+DPU83-c7cr)MLHA7$Y>&1?8J?+MmJ4`6du9
zT#Tuva1jo`3M(947gSBM;wRg>E@=fUcjUYai}W1F$p0{1JMB*}0u;ixjr#t-Bk%!q
zD~62l#Md#JlhlHN<_|<~Uv9j0hZc*>=b!GxnBZ?kV!R`YSrp5-5p4I^^k}+-Rvq&6
zI}-AW@C&*n5mZD-V~C|we)Ki_|3Xk^Tj#hNs#n#!npWet$gSmzrnTa-T3csZTQ6Fp
zSgj7LRavpdj~|ym>7<jCd2=rHoph3Q-rPEG<E1|3lK*vpWUCbd6P16nUVftF`sO-+
z_RSZ!Uc4qJp@@1{wZW2|o!yMzy`C}M>Cd^)>8)}3T}$(p=UkZAIyc8Udv>0Zo6Ent
za%r6lzjs~jzI0xbr^YD}@ssems}#~Mzucj7T&ZK8@@iL|t0_KK>ca24x5~7ZBK$sl
zox64J#nx$8S;fDIuxwoz?{o0;^VW-UtV>(GD{ipPw*G%EX1UVhuEsCDUw*mx5&v2j
z4Ciz;;y2`LL^q6cYs3q5494FtNCdm*>3o72xB1hl?JT1jG5y|1pPFBUzhB_?d0Y)J
zNquvJr`iSmz`fy)Dj_OUmaogp&!fN!@%IUIkp`bl@)_kf;vj`#&@==&V)G<la=J)X
zDh*smN8B@SunAhXU59-V_DeV@Vf^~%Pout(I-0DIY3S$C<r;Pth2ci^%opix9l3=K
z4YOzCx6L6{{2_(h>T0Fh+u-wC@y8)9zg%zLE77~JlD&nT#ct8k<(~SQY$RQ5#VrTJ
zwZ74V`$}*CphqgW^uPUppO*LB?9R?%SuI{qjY(LUtI6kfxYppu{ImJD6|5IsBm`bO
zQiBWQ(O7VedqwlgOYjr&_!9%1Hze+_;a><Say749h5Cq8>jgC@V|KtV%UfMFHBD}x
z&w4>!T`OraRO7E72rg%1leZcS-X`DDn&s3Se}y5^rSws3rZ^k$*Dz>5n&|S&BaLM4
zhzZEqBaD4vj@eELWR+u_Rd^S=t6SzadF!)<VGZL&A)<XucK;vAn}fge(GaB>YvNnz
zzOfm9uEU8Qf%b?hCJOmqm5CBCkNMrB7fXLD>8zA1xEGd(c=vC^;r)nj_<cCsmZB(|
z0sGRB_CPqi3Fif6fcGQa_J?rTib2cwn=1haUjZGk`_*u`oA7lU#4;UFft-H8Nq~cZ
z+{{znz@iVB_hvY}8E`{C_yIQo^6_pn;3PonzrtY~pbfARaKk%rFo+L^!+VM2hnf`V
zHwb72R6Y%d`2&oZfMtMIz*T@{fNg+%fE!UxKj0?d%4aASF!Kob$ABM@zwDd`*am0=
z+z40($T#t;0QUm60rmrK0vrPD2IP++_5o%C_5<1g2LUSqm80QsBVZ=rU4VT3dlK@M
z4dI84fOi9mNd>HYZ;v7A=7ma9Yi80}<Hn|MNE(|-`q{WP-5(BLL<Sk+^D9ag(#hZd
z8?+(XLdnRSn=$Dc%PBXdw<%YjdBsK3vd*Ij@-U>_T;>oZj!og8xJl8=Rg~#SqpV5j
z3&CtWeV^TIQwlRO=VVNplVP2ckv%6PuP|fQyo|n#_Oa{7bfw*$dQZxo$s*q}<ZIj)
z4o^f~Q(If&<txr8i^>+Me@aySb-Pek+sonbGl0~g9rdDa)Nu&(&(SxBbt}xsZa1{i
z^HZNb^i_0$NSKJ2AJ^VE^!1>3$D#iQ^er)T>hL(O&7hwfrANd@g&DT!x`JKQ_b^~O
z#-5b~)Qf<6jf0##^z{r9WMDa2ptoYIeVgq%vMmypvkdt*VGN=<B7x=jLEjid7ySqF
zH-NqhJgL8Fu_<#hG7B>%6=&qNr`*l55tyGJt?dUNem^48F4S`wcsTaw!{DhOBL7HT
z=Vf%KL>Yzo?nS<#1L5#*PC-7~P0{%ZGxFwT^e4BESwGyk$isa9M83uWJkbEA-rp+p
zjr3EIuTA9ZO1(Sf9`G=*Px%qQ>tHy%kOb&BFJp5O`xe7QT-0YT(oxMwApJqmyW`Mz
zfxaaU{dLecgU<J#l;R940n0xu<=;dC%7={$Gpy|#M#{w?O@#es;nR{csQ}u*TqoMW
zTp#LH4F3JU4u_i%r#u46`4IF>e3+69IjMtY`(UiVKJ97iBbK2(*MndA82Yl@k!_c-
zpY1}vEhs0}Uay0`IS&0W==aB=k4r^=jYH1@ePbLtzvbKj`k7GIs5k9VA^BtJby9vT
zy#w@k`rIe^W9#*Zlpjm~C+O72JbtOqJD}eMI?ZbY()l-I+Tzq}BG#z8eic1G6q6tH
z4VWL?6oP(nPjX4dq`Ol}GOS&xB^lZ4)6|T-JIAOQv)ad&X4vX7W}%JI(qi7m?2Ueu
zF8X&dCWI}JgYFg;!YMD_(blG<aAzEQ7DgH)Dy}=R94PnVD#0bfE4b4kC}oNV)x#Sa
z%7Sm=_DdjEzzQNZ=)D}*85aUkFvt;CP6kvgxJ3l~7`Z86G4s5Lm=$m-krs=hd=r^g
zQou>4=>on6s7%6bSO9*|jsli5&pselz!gNE!_MAW5b#$Dxb_@fz#>59I@}@xUa{%|
zZegCxb9Di2L{2|XQRbpfCn(^N3v>b70F|A%bs9EMVTkD<if}6sl&RY-oRX75L5f(+
zj7SwkCXE4>i>~lq63GbRl$~gWLPj}^MPP(=yMkB&u9F#ID8hyJ6_qMn$|78NHBt}Z
zls+Y8KI3JQA|ihWvX)KE?Z_G<g?~_QvXOER5UT!TK;>TC__~D%;gq-TfN}?tIf$r3
z$ynSdpGX-I?iDk2DUofsPRCV6<R~sQi&Bz~8@~mls`H3U1;X~Uvz%|?Mgi`VaHC#K
zSwJ?fJe=O>Z7I8ea2Wl82>n<}+f3vX)Q%}XVV=oAnCDkSY(QwN3f7<N71gO{IX;>4
z0NMTmBvQ^!Ae1&l(JupGImd~dh2~?*C(OgY{K!0qiIf3hIW>Sv9d0bAndP+0lrPA(
zQzFGoc@fC76w4s{0GA7vz}HSh&tlnKL>|CJgh$g#4=(nOr-`tK-NFb%IORSi<;-b_
zaWDG~l2+sT5hGuPuPpGD1-`Pt|BDvLoMaPfZ#Z3tn<dQqx{eP@D41|F=EGQ+jD?rM
zpwkC6<2e!^r<0Y9U$coAXIq9{ZkvE-h&!g^+p0wDEQwns<a)-S$#hQa3|s{mK06Wi
z5^hYk5s1~7(3Wl^Q0OEuXvsE$$nt*`muUogQw(r)MBtKX4p>|C&{icNmN5~mvVL4-
z7>xCY>lH(*#U`LZ-;|-_hTi){>hRwxG|I;d45B`R<8^+#uOM(k54fKKUn|q4nE&5F
z<nP$Wl=PBp*Y_kmBH@^cy8fq2c%g(dB%CYZVhL*{Y?1Iz34bWzZzcSjgnJ}>Ny7Ie
zJR;$k(`5M)UMS%V3Fk_<Si)KfTO_<w!XHZbTM7Rr;T{QJlJGqVk4QL1j;qrpyimd!
z63&%yv4ph}wn%uVgfZskpBlZ~n!RX6bA!LxIz4w*?$k^23z`L*pQ95LnTo%ji$ky$
z{3%@gnN#bu+{<$FE}5>gPF?G7YWDfpOtohFT`PT;E-Z1B6qc7br(KqJX{`sl+Bw#2
z{_)@%D{?v<i%a$FIdHh~CsF;b6@dOG9j-EB*nxF7H7dCcUcWmRf8z8Kh?dyOhUVNA
z&7QiNOR!g;OKg?Pw@S&aS<?V{I`lW`DN^Am#EG;fcb$tFWUR5yujKM?F)O+F!>_nQ
zZx7PMd6<&xUgfN9a@D(?t7?$fz#tTVc-Gxo?QX=5viM8VLWa}T)Pz3@D`Xh)8>*2J
z<+<uT)yU^X5xC`wa(U*a<gW1flw5AfyBqwm?exFJIsb7^HRhRF$u?l{LM6eN{{#^?
zWBxVp^AX{kW8|+$wgHolJD`<_DadIoLdi$WoZp{p6KUN0VaU=!<Q|*97%AA!^%~`G
zOtuN$jgn9%u;F>ym=7gc;bRx)WFx;ZZ*!lVfq%n?f%7`+Zsa%S&&8l|t~Bx+>wrx%
z8uLC_M2NND^`LUjHS!zt>{_{SiM=B5NHrqm82o%l;2dt?#ymbK^KTZV>iSbpLw`wD
zZbpPQHS!zl#-<FNW4~lH$~W>G_3wzuZ>&SD7Cob}ZZSXC=GgpqgGQSe>!z`8Z{#x*
zg0UX!n69f4Ck)^+=kJ#5LAT5=_Sj7s_~gf6z<Uul=QsS64W|n>W$FSQ8h0a~f&T;%
zbAH3m*yYwa_`qlmvHf=w5}Dtyf1le14Ey0SKYuyI9AeAAABpDtO1({Iuh#+3kjx>r
z{J$a5oPR@uO|Wlh&~vRf8b(JW?I%&C8xeDU9z!sk-7J8@Jw$VeE&n+rva1<vhF`Y#
zHa(YB2a&tce+<H#h_I^}`3=8Kllf!$WAP7=m--v|`4<Hl2LFfo4}q5^GV&XK%aF`(
zG^}ApqkN+shmhA?zVUaZhi*p#g2At2!?-~|js$alrL9V5ZPP(36HUR$XTUK?H|K9%
zS0%Vx*XfxIJq-~?J_F}j2x&(BH_7~W$^2HsA(t7fMn(xIAV2<#b53)Tkzbl2l0;qx
z15QBPC|e}R+t@qwh}fGqB^db(5^-bf88kxeyE@j$0uau@bu})d{>J>r@g1-H%KMOj
z@FI7YOd+;>Q__fN3Tink*%pVMjE%8)^c0+##G|K1o>vr8NWrS2#pNFpd7d$`qn=|G
zV||LIrz^&~6HPx!G1il4`pJ>!m1z1Yk>|E(`Z&dy*Q4p<6=S}Rre{Q+_oC?*%&Rst
zNyJmQ8W{7YDK$lzpcwO^i3~fOiuuiECW&|oHu{YD%bb{ktwCd6il(22d8W!t67dx5
z#2WL6IWYyhlg9iIO-FAu#=R*aML9z;#<PhG{m)d4aT-lGo`#31?60XfY#OHMG2Oiw
zX@|)o`Ba6+%H}{nFByBEMtfOhq607q`qXq*35qd3mI8}c?kd4`s={MWbD$r?ty~^w
zn!|d5oT~7+(j0y+={$}!hs{8!r?F3A_+S47oqF<90|sOKhY%;7-ytv<bfaEvl5W(C
zd-?I|W!RI)f98-2e(GcFmsn+@A)kNdm%)(F?+4?_=T}ro*r1Q?Z~W4U^4nteVfb~F
zutN;}N0J`9Z}UeX|4gMSrX3#_^pXAlZ7Ijt9}<<o?U3LfIgS$5D+7~v62|GInDKbF
zppUFC%ZX>7Y{5Tr99<<Ahp&wozY8QicD%0;az>8d)u3DBjK>zxW0nYEieC!)c%`h`
zCUBTVgqITVe<1n0WB7SS9Z!CJ0(wURI^WMZ7oL{!exK3L|03zedx3_(@RX1<Ug2Mv
zW`KGkyp}-zCz8J`roVkI<PXnRZBouGIFj+|wLAfRji8TLHc9)ODdp?}UDV5C6BK>^
zQk3^Ww@Mx}IFf*VG9GZ^>2o&d=VE@a-ccpQlu2RNC*W^NK<`LE|3v~i-w!(#<9blm
zH$#>?299VvJ*|R1a^1X8(8nuQDJMzFS(-pjTLStoK&PEIud)dVv!t9yr2Nbpoo>kI
z%P`d6I$x(tH%#GsT%_BU>vX<T!Z1hB^|<-I4D|EBZ@hns?j*tn$<M##ZVrD1-D)Pt
z_&(67zwth-(a#S{`F%E<;N&}14Bv#KI05{IzkzNe!VJ)<zp}z6aK7`wuv*adxcUBF
z&{?kWUanz>UnJoFGw4>c02zM@^vpQ(&we2X^MAEXa2n$x8;)(feo`Umu)pM&>MQF&
zr=5-Wdks7PS@QSC@V}Qpe(K4(eX3&gnGHI}QQI9=LeENZR}`=0&x?_NpWsJ-ljGXx
zZ-1BkeKGv+N&1Fw*@R@i<HT?#AnRqkuPokW1AQ*&Y!~A_>WP9|QLYC)p54}hJ|X!G
z<*s#Al2ZrYO~C&E`BC3~Szmt6%g~d6f4`L979;;%!H;#sZ4>$3;*N6+Nw-$%^h+f@
zeVlH8e%Qxwlel9)5OmQGJvKr4x}=v&{z0csmuf0apbNX%b-K|X?i2jO@5f6q%43q>
zcwgM$|5VD~RI1C7!%NA)5Q=B#)d}crpo{)FSC{jQEcl0#-Y4s8%##lZIq+9yy%tLT
z=Rv0(jQ7utex*q{RymJ}ce9Wu4FcJYjh$6Of-w(dfgVr)s{|e8#?<#Z&@&+?w*S;7
zkh2bSju)jvFV`3^e@nps66my#a*a*sd9KlrmiY6@X`si`{}Mrm9BEHO%zT3`?PIj>
zM$oBeRg9j`NPgRVT_3)y$dHDKmv-pxsuH+yI~jD!>6UWZWI{3MoNx24wF!!NHxn_J
zpi3qbXh}f-Iq0-czpR(BukbqPY+rsWDgx$HhT{q3oR0~d{5HvN72Jxl5cGKVTn#$s
z?bvp2lKj@wZBl{`{!7ruE15C&|6>CA-I8Bfq02GWqt6oXkB5@+^qeB-D7RXdW3+o^
z0{$Bl(7O`QH%a-MOLh6H6g@mC>9$Eao$o?3JS*sN_7lXXAim~l;!B32qNu}}JRWyN
zQSue%Jja4Lg$}3F*Sx}Mcbc<1uPgUEW;vZ6uQ(SHNsPr4J$;1e86zmOMo<dk7H^c{
zyiWhh$B{onlwG5Lx8lTiDkaKr;>@T-ewbDAKmF7rUgdcJWqf!P`6i^g$?x+w<NLpA
z#p%4JXra?#FRySql^S;wKC|%Qpw8*9cUITo6B3^ihvanDc%3WjyenLF&Ki8t<a4^3
zTdAe^T%sm-M#0RfkQ|4}$#XdU49$siv8FYOehydDTwlKiSxlJt0w{{r>72W;u(ZTk
zGQSwg>YB^AZ1x3n=a!dLI4cV0I7-Mt+ebvyL=${|e4M2?i<itVEVUOYJhXF48hXmq
zJf{<13wdvH*0~#2`d2yKO-<gWvTLhtws{WwoTAD~=d|4Ek#<Y)saO7G^MsYko9h>g
zZ_`T3YCK9~6AnXbr6A|j+^M;jeVL${b8rUfaW|E;kh9#MKMM!2o@yuzSu1gOyXx)L
zpy%_eC8xj1<?;LM)g|RF+)9e_OBP;R?nCl4r`_vwF2FZ&_4u-AL|w|gb-of$G#Otf
z`P}|kZ7b~Zo3+?)z?|1blA=Cmb2{<WVYRc`*DU&gvcgrp8fsP>N*H!9nxJAa)V<7H
z%N5PFiK`f`Gsk-V45I~CxN4l$E}y?8nyJz2^RzlqJy#vR#)_?=UcChEI(2IF*G|Tl
zLM02!{WJ8>)a)ME-c!P%T~R)&Ft9jR)_WWD-p8JOO^b7(9Ee4AE}svhu(8?ii&e3r
z0=4{)IzqGooY&RVmBkpv-Rg;LYbP~m@;hr?p1N2S=3iUxzpUE7rqPXlhTc``yvbAJ
zZV&^nBwBQCKBi|m2PQCU{?zhSo?5@VW}&NLH9Deap{G3GsjPJ4Q`LI6x3<Q$CPo2q
ztp8<FM-IYTk3NiMc`zFID#LU;W(o|GXqC}0qSJ~#>8y8oM)Z=3#d_14J_Y>}ZKmdz
z8DH`hdA&{iR*tnT@x<*qa*!2pQ7EdfS<vXm+ZrXUn4^6qWo6Z{4hMF`u7XcDewn8X
z45&eMtIO%aOAsU4RklUUSDaI)6|HhtuXZ=h@w%Guai!0XH$FtEp14LU9%(dtMX7Um
zDXAze;zCj(-lrIz9TIeb;iWJ}ucE1O23PrXC;lr=`vMG}8c&0>8E>UTtdXFIxRQqF
zBPZTDS%G;lrl&<lv8kzRo7}O}7wc}Un9((NMy)WT&ItJh(bJ3g_%TlP%UV3;mpNs-
z_@-mhXw1(OA3tM7@UG#c&-p)g)Q?(rL1e`$S$IRllrfEwKhxYv#Gu7!)fEw+Mkg?R
z{`5IbURMn#<8OPL>S{_nsDc=Zo*9^fy-n_z5%}dRWB&ALTb0l#`MOasI$;ltfL7NE
zPfLFOOvE{#IyqrsnACb<8!Y8`9m$!WJ1tV%1lpTT!(L_#{utv7ci}{GB1VYRUl<ME
z0*1Ni6J!qYIP(J9Jfdo(H_q_d8*1FGae4_gl`By67v`9$FnP?R-{yklM>;o7^kBo=
zg4f~01PP`Dy{g<ly&7LQM|U=S=j^KH>=84XoHahLbCs*1rp|3lvh)n-uSKdh6N4FF
zf7j?fkKV1O3h&qlPfdJ4Ek!iJjs>8wMV<aK`bd!)kMvLEoGL8mys_Ei_Eo##8rQge
zrjlaXG=I9e!@JRO<;>dH+~jt)c$)mpE-a35`ckZ*sb)bX@!SQepQy0>xV;q52&|Rv
z)p&J(57y$k9M30WFUX&!J5Q3^4!1bZSyDd3S$1vtG^eqW>wau*fdW6M-rEv8oyj#K
z(#wjc>NXmow~?SXlQ25g0DK5<G?Bh4iKdaRV??MJFe4<o$~k{w`8O6+bLX8!=iKe8
z@#%Rz(;MCH)iHC&a7Hn%q$&BU7OaC4@8?DQ{rTnI%WCVIeXB-_h?zt9T;TJ3=`>RA
zkGK}JO!S~Yg+|mCa}yU$^N&`P)I&@8lLgMz?lsP8)Yy;lg6y6qU&+E4N8PZ~RN;3!
zn;R1K?F7@()O<Q3bV#B+lms(M^wc^v-|1|r|8l#SMsdWQu{ZeT!_mTWUpy~}Kj6XH
z7JYj7fdCIY2_6U%tatkCDtiHcEdnc+;_)>$H&h#wmhlV%1A7b59&&)yl)whw3TIR+
zV#shj;voY*XpNoDm=BMh<x}~8;k6iRc!-?QgelHj?{ro3w;P-uJX!M>K=5+u%1G~=
znlI+G$n%gB)vCplo)FI^OMeXILyW7g&RZR)GwDk|x@Ni4jUTyK0}rGTecRjQTIr6l
z24+dYGV+nmEnZv2qeyG3vk|Wad-0;L2eMl7Mw!=Gd9dA#bsPFN+*BWBBU2sY{>THL
zOGk_Z#kr{ACYTi7jIX(#6B2fLrsd{Q7?@!JSDibK*E#y5g?a6+6{5V;@Jv-Q@_D}8
vH{ExWs}Y(RE}(qY#|&BBXnFw2H$DD~As6Y<Wv(@KaC6KaIv*YseSP~s+Z!4&

literal 0
HcmV?d00001

diff --git a/benchmarks/Grid_wilson b/benchmarks/Grid_wilson
new file mode 100755
index 0000000000000000000000000000000000000000..ca7a454ade9a37f0dfd0ce94956c1e82ce1df942
GIT binary patch
literal 121651
zcmeFaePC3@xj(*35{QC5iz3DfvU=Cvtf)-{B@xhFve9#7gHfUejV2g^qykCFMo|%h
zn^ev{tj$%t#TMJNwY|PjL2Fw;g$-{Bs2H$Tlq&JHXN^cef$$>veLgd1H)kQCz2DpZ
z{<;e}GxN+d&ph+Y%ri63%$c(wKX`3Fr_&Mt^>bY9V5saop8(0H!%-dX!SOkA9H%*a
zjuDPwj{d+(!(SZFXZhl8mz6B;rURG16#S*)FQ*?vbbWLZ?l2K@AP4!2=T4XT_56I6
zSGqY5NX)qFaJWT0q2zVdLo$BVLsJDj?q-k~{OVlNji)1fOxI<l>$1|tU5CYg+>Muy
z<oH))#TWJSA;hqcjw2wRJ+76#qwurPH2TNgCixt9n_pnY-I){|esMRR?na~|pa1AD
zUd)@U{Q8oMYVjX;C07o|_h(k#b>-;q&zyGY%*xrdb1$8nbLFL1j=rp>>axp4xk;b6
z*G;-vuoRbL214X{2>w_U#xGy6@2P_+u30y|d+^-%4S!YDz0;2{?h5>o2Fs;Sb(KwR
z4mb%LgTJfr*ShfZ9Z!|Nw&~GX-K}4_uK9Unif`YCw?F32%zr01Wa*$h$BeY!{(M;V
zs{6Hj{s{D5e|LbHGkb~I-v^xa*&F_uXq3Id$MgYDL&18(Z|fsp%G2EopCkH!@9u-n
z<9)!Leen5dANVVgZ*Tm&`=Ar)qr4CHLI3tX=p60?e^eiG(E6Y=s}J}MebAZTM|rR6
zqdlbe0YA`(94_sHpP%)C|9l_t;eEjUebC?DN4XC50dGUc?5*CP?xS8R`;e#BhkWMr
z!RMd)$aila@DKao|GYl(?cYbfoBP0jq>p_6)Q22C?W0~k>4VNaec&s7;Gf$E|4rck
z0>^h9(TAr(A`VBfcsd-T!6)Iq`egX|pig-FLJ6<4@FNzy=>+_ppg$P+bqx~VZQ*xY
z_+?hUg%<n`r))o2=lCp$bP9n_I_|Kf;{qOkPvMX4t?dN-+bsGHE8jZc@OQ+juV{u(
zgcA>pS2dw&<s(OOjJncMT6*`as@bJAp{a91rKOJ2>+dbSsp9U+noz}@yqQyLYAR|R
zrCL!z>9mSD7PzP&Z)Vl(ilV7^&8(1dy~mVRhi1&Fm^!WWo{IZR%i#+F{kBPyYepl7
z>g=pi60Hf%sk%S^riqiMYUTN0rhM+y((2kd6{Yu9&I#2{oyimvCx@;m58Yo~0hH`f
zNOoW4w2IlKm9r~D`S%Kvm7}XGD((@1$)W5qrKQNZ>b}yukonY^GpouaR@D{LXV%ut
zh@)L;M=PIN6CxIp16~ETDyGgt7Bv-fLRvYX+2u2)&M6JenOYgD0U!A{0XFKcsWp}5
zacr%qv?!}IfAW>3g%c-_Diy2~Y0lKy)2e2ntoK%ySBwGsf^ifXk`?8hQz6(Dx#vRe
zvnpppR+8rv)8``n-4&sGr_QvB9?Ge#szFI>Dn@G%NFX~7zY^u1I<q1HsL9LDFPvDO
z5H;mmGp4+1R`t|56^s#*JTX7=I~@#LR4cEnynA*P@GZH|smZ^o(n^>kr2)}Z&c0g;
zus6wxn$8C^lmO*Zs-XZkR9DQNRW*$*r0AAOXdIBt+^Kg}-kZ(pEN4qBWfOq#rdP!q
z%3WxYvrEgThRSCM9%Re9JganaX~nGS(EX)zDyq@4tL99-yMk%%o;$a+x?)aE)$FM=
zD-nHfmX)mLzNyuyQB(+W2Txbni*i@(^uEi16q7@v%d2WbC#OXV>LV?fxo!4+vqX}l
z!d{+TT6*uS%3cbLj#I_Tx0>WMw!o?)bXY|6w1L3oR?Sq6s;&*?qfSK&tC&44A*P}t
z6!3pkSG}-_`l;!a%gG^#-|BudD`wvvno(LYXHM0eMEgo^u4o2qNz-N)-s^y3nk9A8
z%&NO*GI)1+d1(zn*^bhiXWv(eQruKgQ#-4Ik>wJ%x~isfZt1LwS*X&xXFE`8b}K3_
z=(FilD`z^Y(H=w7ElHqb&8-YMYAWv*Eu^L*<Um@>DX+fYfzB1GoK;aZecIIfi8i%7
zR9Q9Kao5!Hd(ipID~LX;>fVYTNNLUev+k;z$!y7%rAlW_t(@(cKBuC>QCU-6JG(sO
zxGpcRbkt>8j_ZQjguK$rFT4D*D-!<f(b8L743yEA<tBx)bCQBrIP!{%{S&m(?8`<a
zMPz5&gHZVWu(IHvd~yVchg0E8v*HNFAc9bT{K+%}9O(j=;nOTi=_17dE9L0|cL1IP
z0Vgf~$$v7${J3s$I-bQC8=toqe((N)4#(+^2QW^@<9}+4ul;T%GCa-ka3a2+W3v^X
zg#=6f^S`hR8t8Zw@uVs7Q*8MCk)w<weSLu!BVoGZNeh2Q*G~{0;P`JV+#urnJN{^e
z-(GPKd})r2R(PtwPj$Ryg+t!Qu(a*x_z+?Kx)#WBnrpN8nwRkSyn$$3r^SEqc>v*D
zyTyO;c?scM+r@wJc>v+#;y}3Z`3m7z+Tihd3E??5czk|9_|-Oee7-_>t__}1C(AVe
zFl!1TV%;`41ey3{+Td(W@n4n=j%rN&a%^zs9sl`kaQix5wZX|#{8wm$+vQ(ugP$Hp
zhFfNXqah}KGi>ld2`KD6HuxDf_*@(OJ2rTo4X)VWi)?VV+xV~11|J*;!d+^E+t(RQ
zHuzaK{AX<Nu{QWh8~kh=e3cD8)CP~*;IvW1f2}t7xp5%eHXHoAHuz2({5%`H-3E`_
zv>0vL;KObBoi_OSHh7l}KHdg*So3wZ5!wXezjPbi9S6d7+29x0;2Adfg*Ld`2LCr3
zJktjE*x*?<`1fq^92@-KZE&9ro@s-tHn`UYFSNlgw!w>S@R2rnnGJr44L-vLztje=
zw!yEl!ROlGm)YQTHu(2#@I^K_^?3Z(XoF|RfpC}F;FsIrO*S}f9r52YHu&f`5Uv{j
zGF>&&&i=s#4z(c~N;4l9z>Vn}Y)J5SU(pXJ-4~_esf_T!$M6gy#oD{^yXZ(I!c>K^
zHWA*>FjZhID#BkdOw|`#DZ(E!OqCaF65;n5rmBlIitrAGsp4XFBD|GhHrrUW2(M?D
zDlJwf!fP1jU>qwH;g=YuYK!?q_&J8D(qdU6{40j3%3^L2{wc#$VKJ8oKgKXsSIi;8
z|G_X-R;=?l02e*XFjZBoU4$QCm?|pPCc<+VrfP~sMfe8{QzgY#itt?w)4UmL65-nz
zrV5HRittSgQ}x8^MEH7!&tbS)geNde6%;EI;j0*?>WLMK@Z}6s<-~j<d@;jRHL)xa
zzK~(6n3!9HzsoRHOUxz0XEIEc5_5>~=?qh4#5%ua`TH^KX1HC1kIX=rDk9b<!uuJf
zYKTQe_zQ-q5@IVw_+y5t3Svzn{651MG2AG^I~b-)h}DVkR)(nxV$~wNo?)thSeXc~
zVVGS%Rw%+RG0ZL>^NH|t4701pvPAe-46}>J+#>u_hS{}aE)jl=VRqe^Lxlf>VRqSA
z=P^<L4701o+C}&QhS^19Z6Z8}VRp?}6yd^kPs0LX{sE-c(WbGM!8NPUD~pAE3fDy=
zh<_px&!C7me+ig!Yg8GWc1;Mqv?VP7hre}YO(4{oK%AREyi6jFN<wUpBNoRd0dx9g
z%7}VFhy8i-y39wB|JjNB8I<{7kRZFaM&$nu9Fae3&wL53L)FvXW?V$|sQMW@;cXEZ
zExuH0gccvooNxxQ|F`(W0N>&pBI7b-+%OsE28IJf3=ok8?27Yk{se-Ql6Z)@wWK`^
zhre}#lT69URK`h8ro*dbrd3_+bSQ(7z?Z?U3|DmV7N47mk=d2NlbN~V#lAh}MQ*CT
zK{du7nsb4p<8<jyA@zoCXiEJk<9D0QkWXXbx=Sd(cL0<1%b-=i?En_8%VPW=67dYi
z>mR`EJ(|A#PF4R{4S(8MSfsT?I|F`j+*~J|mcn5^hxrH)znQL#2w;EeT{NAW)l1vp
zsNt^fsQPF3^}A!kDp4RP@^9`~$EfahWkeles~C#oH+Q7~U)9&E&0k!jI@{FkUA0KO
zMkd~Y#HzkorvKPVUpM|q^mj+?;G2=soDV79u{rHW06D+j!1kkt$G1)eLg@Si3qx*R
z*NI&);CHv(5ql6xlo5_tjdXAJ-|D}`f3v@6vbylgRZ#QVT14=7jsB|VON;}m{<5c?
zZo~>lWZ3khK<46VVxV$HsmA1VRex9Y9?H-DRMo##o4@VHM&hmaM47y$-qQw`zfeXn
z_ki-qPB#b~g)B9FqeUR7KP|B7wgOv=q%K!=nNs!cSM|5m=A->W=|TN(=0``6GLpKC
z@x&#A&%j-qiY(T!F`JLh!sBoi=;$q)u^?R=c}?~nRX<817H^GA3ErCE@)sH$%>7z<
z#P3GRQt+ngn^b+dN1(x7O7uwJDHK|!(Tp*gacjDrIaS|l2EbebPbOOK@qsfH;uG?S
zED_wQk;NAAz2*pHx{jUF{ETJOThvufgjf_!AE2&c&t%~={kn9~|LQ#}0aFHp6}YOX
za)+by8Zs9p1O;nlD*OoJZouYxPdh<O)dde-6E5S^$sp3Wm<xV{ai<k&G6_;6Rw4A|
zo-6>2dQT1nqO7R*jDrj896-_RU91d2wY<~JybH>v-m{ds3L>uxzgc*Z!x6M720^i!
zT$C>pB(t}u4ZEx|))5W@jTV8>#j3tjZEo+cMuzy9Q77_VTL;o+!69_vEe%ms2`m<1
zRbT88=_9GPizLog7K)_kE`ocoSy38}V?;(4bFD0bs3nS;)X6t%jqC;W{pPdR(OvFQ
z3Gap7V!}ad23-D5!S2rw_QI}6pUf^vk=-)5g54V=yM<O3Dk_5PgjaPj$N4vB@1X3U
z4;1ghm4Lu$l1`&^>V)I3R=oZ){1xKQhrcZRx$);xyw2|R>cZwI@WwWh>b#d(Xg4{D
z7y4jN9c2-YLi;R^PEqJ*51`NoMIF6t)zQlykv>uAP8OD==;pxn9o&pcnu2MRss}ti
zWRz)TWGoTwrD4bX^6V%It?K)xPq&S!|L^6uD0fFKG32(*dr{Rx9+h(FLZ?*A-?v(5
zI#~r%85UD$g1b&3f&c!dmueR@5)u%tl;kJ@7hF;8#h6zj8;6w*8EF%YoZM;J2?uT`
zT>ehMPDXF+i1f+qkQCW5;R<%PWC(V+r^H`RRTPV=_+Pfa-sH~`KSMSWqF==j%Gqi$
zaiZv>$qM3<uX_=_pb<&kWm6C&$FfG@in5-MvT{_Sl?mCcw6e1+2r@<m@eB-L!Sbgq
zJyy{0Y3Smh-VWlaP*^AMvuytf{D{o1$NGYrs}oWU>hnCA>zd#*-$F%ku+>XEg%-tN
zWOOvB?^Bx(r)ukRAN?eU|AWek!)l~HT8&{`UHxMT{6szUk*dF=Ht!jrc5h<OtzRU_
zk8H`~^35UIa-K_V$#XN9556ZQ36Ev_a^DI3s`^AI`@9Ue851Kzy4B{7Q&s0yRiEru
zBSm{0iLk3<81a<Bd9I*-SZ)3Y(ipS_$=uNOI!5$+GF$Spn6qRg2aKq3My@-Fk@8*`
z$pJOr#7HWbfUK$M^HdDNWI}{H&PMJzLA^`XMQW4?*?jttePK&paZmPT$iBMo>_6>1
zSq^2$enxWk)hA_NmYjXH$bJU0pW70yV}AGv0fbpQYD;*j0Ll8e^&~NrkdGZgnJ%%E
z>Ej|{Buq`O>W|Zz7#ReaK;q8rs&0^?8VOP?0@N`=mDtK)k*U`Fw^WfrqM~|?Wuo1y
z2F6-JHC;}<*CO}*%Ec-8o0z&^nbHbRUOGI@gbb39!9h<}_AYDw%8^QRVDv3^CAIvS
zE>djR<6j5Y5z10^Gu~Cim>w^(`NO^4-3`$odBAwTir7IVzl*aoGNtJs_z$*VlsE&6
z`JnN#$i^RDOY)BS{)v&9EbwQkYRYmCJ3MBoLH+0&DW%=2{*~E?()hbKstY&y5U(!W
zz_|Reza@ove<A0h@Da2dw5Oo6E2wv%4gV}?O!E|~P**IFrVn|;-x7F3bScf4_l66#
z=~o5^-f(FEyFtrQpaZPwfj2Vva7$UvLGje?4dJ7y%HqFZ1r>f(C=iEg6nWJ7)2zX5
zI|jFZ()gAb*v86P+`Qkbk%w1;Jr1!soB>hJLCk>W3UTeBJ`d`jtmd4QwNghZSremx
z{)M>`Rt!psLy|QUQ63$Ov1(!@_&X2^yzXKPYGg7<OO7yA?(AK*@me#|)yQaQ;?Gm3
zD^-KC@hHl6;aQUH#&cSB)^w%1xoNtxK(-%**2@s<2B8fy)Gk9CWvESrs%B>UfU!x!
zq7oL$E(C0|gt7F<WnOj}U@a2XL|8zn3S?JL_bb&=32n5Xi?i$KLxjbY2qJU<ECxK)
zptD|jNpcsw)t~aQhmV7NbRU?W53=I9<pA2@YBwmxZ34`BdV4_0yrmP5^NIT9i%O;}
zykD8JQOOK?s^NKaztYyMWXdrE;jKj757Z)08G?U7^j4*~l}Xz{AD|7yIQk;W`!+)F
z0%5&U+>G_wpf=Py5h*rAZ3N02a1&WXC44L3?*MM6O$h?D0a1T}Yp2RZWI&nnw5JI;
z@a)~6ue5yz&k_%%nXgRgB;+soO7WioUFNBi(8Gl82K4=a;$1d~jZzIaXQ|=MS?a>W
z?Lbl*-l55%{B<%d;v<7JD7XH~$kfw?u&}UIMg~rIG2pTSZU)>|Ad7)4E8t_mX9Wrw
zD6|4)43t@cY6hxBp#F3+%W$0)W|0_fw8BkRxXB7rQ0VSZHc|M-E~S!D7Jmy{N5@&V
z{t)jlR5ildF|h+BHG&?WIx@(q=>7uzn5uuM)c=aw6pC7OtLDS~)m6_wCx5E4qE$6!
zdUCR(jhxd^O&O_wh{6(qnG8EORytg~&4X9d-@+<h(+>vq&oxXJ^!Lry;iN3n1sU-L
z25^>*qGpqGuuPdtq+~OfNug#wE#;YsC8LOcSxPbURUtsMW+^;)utWm}JZK5hv&<vp
zRGgv|Pjo58lif=3?O953xlbv+r%)-bDN~9ctPUu}V#tJRP~31IZ<6lMC|;$wp%615
zr8q*%v{L*;wQ%mJQ;I7ZmEu`VN^$K<rTC#J66UwzQP7UZ&FE;wcVNXN6MJQ1uT1Pk
zVlNVVk=TpGUMBW3v6qRxOzgb_6^liKs9205QLv)swki+Tv5`iG{ZTan94}ILs^xX8
zb{PD{*r(}#(ewkM%%tHjs2?)#Gdc2!(H9b5g(MCYBcD$UcjuqM;SLmG3)!g}>EYF+
z>3}rP*gX{XwnzVpkQNE<bY<@l18UOvp&H(N3{d+6N?YKHI(VCvTem<J1g}`>hfle6
z2LeS`v?H*6y0Ui{LM2x;BJ>_YpCUBviYP)KA+!gfnOAfo6hr7BLZK^~5Zb5Qib+D-
zyerxeILyqL7qrB!%@9^#3?v#H1CbVuflN!rK&aEkL>2GMF>Q)BG^Sng&KuLIcvHuy
zMkbe}n_1OkpOF1!JZdyZ^V12Vj##D|i$!P1-eK=gk<mNigUc8wsiHAra2ex<f7+P;
z)!;J52W@`B;4-F{!DS3bf)fXqF})5hW9m@Wf8yX08GRE9-(O96Sr|oLbCY~hd*FC8
z?J_jw%%aQCmNSQ4_W#R}bIHGC$hqX7Gvr(%hn!6qZoX}Y#FPfk&(!k$aUF;m%e|B<
zhMR1{ggzV@$hHqR*>1+TsR<EdA8xXJjPY49_TeVGj4@?ajD5HfvkpeqiOBkFb_#~A
z;U>Gu3O8Bt)Q4*KdTGj079YSkQ=oq%hm@{QySrDTBSAU6>M7*x43j!e;y9_>g{?CT
zKImY@fMAf=NdyE1BEm$ZYLe4A%q|cX3m7p8FGDCu3$m4J23@2wi3Q6e<4LXBf_Kuz
zsKhXTlEYiTAxAJ(&0Q|+K`O_fztMQ80x_t6MT;P&t;I_H(_$Pf!DzPyc5dq0Vk5P|
z8tjN|4R%3zHT_$Rpql;-Mm_T*Z1Q1X!?+jJ_iBmJGhfN<kH1MSJZ^F0ac5RQDUpL8
zJaY7dM-G4R$nnpwWKQ=fC2|l%V0NKWB1b|5?k!VF<Zy_<{A#5{j)@5Ts1AiDdCn^k
zepFN}a$!Xy{9jEXUJjH9|C|-eJXz5gxWXLq;2IqA;6BbV5AN+4^vW^l-Gf1|27}&%
zQKdMq4Wny2M%PZI_&lg$E3MZ`>$TE)k=Bc}UZnLRtruy%Nb5yfFVcF?Q@fivp{Jov
z)!S4r7I0ntHDv_`K{5Ji3pYo>rkV>h*3rY$*?;sRk5AK&W2_8!rHVCFe66ABo0P?m
z!BWEpnZf!*gM0?{wO?_i-G-yaX2VQa%H83qj<t*P{f;%<UN)OQW+O>EQh<#i!)L0-
zQWp+C3q!}BAe_1fa~z?x|H#YhIG4Lc;g5)&0?e$nOb%Fgo3gf3P&J?UgbcQrPl2YH
z^8@CL#>$CP_8zW`$P~G)W|1Sygz;7E2VuAlqD+`Iz4Sxu)^A$)tqGKQrulWHMWhp%
zN!EWt*11z|_Lr<*(+lhU!8(c^@MK{%omQbZ5sJHm0X?Bu%iB*?thLMs#oCO*OCEkn
z9=Q8%K1_k2NUEsC<K{~)i%-*J3<p_C3vyka!7^aUuGI7Rf89euu=#l%-{sE7#e#!;
z)<Imddxc~v_3;)1`3VMINHE~E7!YH(e%#DXFmQL@3>-yoK7oN}cT)s?Fz~7%)naZ0
zmr??+Bp9dz45F4xsQsdAqnhKwzq463jrUo;-lZ}3WM+n)ika_Q%!~oaAZ9CFXySfr
z{-!)So(&&Rn#N%zf<=QS;&m$_driy%5%1vQ8Nb9HGHS<D3^WYg^|?|PWG>Dv6G|X+
zNv_lenI*Y()M4W#LdoLs5}{<tc!^L_GQLj9oR-_DluR4nsASH}ZBj~Rj&D*jL%A!J
zlF;~-O6I)Ws8TX-d{oH{<hCg#f$?ovisrT}CBgCSm_6inDkVkZJ28XERX`5?3%i_0
z+^~RR3i5IP0LIFpcm#*K@GKhY#<OH-mNIhMP@gh#=FmcAWN2uaGIHL~YGq{o&^i=V
z8M%09qi~lDZ4&OXp%g~<hM*IgrX>o$f9#WTV)zX-jI7_*MEy2>C|fwitfGOnVdgCB
zv_EU9_d0z6$R{)~^@CG1u-gTx7PB0gnO{{PCyZuP7mbtrsLoeCS<GE5<?2WWJ+2tB
zX3CYU?N>cM1Q4wnC7w)dZkGTDRwR#0CMmFjIWl;)6)cm%?~CB89$^3!Squk+256X9
z|7Mo`m3+S{tY0$0aSI|wXOZAb3nF$S1mt}SB6es5q}773ok==>v><f}$kP^t+E3zq
zyi4ZZlz_Amf{hYv8}BwZFaTlr(1*-ha6X8fHbT9~)K6Qfxxp(Tk6940nJW0I0|Xsq
z;YRLb-|jE*-{HU0UwX%ebvIn+cdV_(V@()sWcB0lnKeMx^v$vJAn(}Q0H27p{d!;(
zSXgJF(un;xVAFlqa(`~Q^>9xl@UR8~gDe84NkkP^y-sB22%W+5(QDohlpv}&GrqeT
z<BAQ2wpa<et@(rZySro89EPuqzE1jXp|Ac3d_nq-(l>#=`A6Z)p-=smE2B@qY~KyA
z8;FSp`0ywxbE^TO?X@F1i80l9&}xU9n9_LAYKDK3p)y%9&x+71J&kZF0r7g?^&V5&
z8{z!|-<m**ioDx`^fbboEl5uz%(EapjWFAS^fbZ|7Nn;U-eKk5(+DR5Qh;31L%PkY
z8I3XQFR<Pww?B`Su&4c<X+e70p92ss*K<&z=G(~I+=fT&56}pxhK{JJcLeosStBTa
z4^LZmN{RIF^d~?xeXpv25Sxl7(4*a_H`>l?gujIO7zP^H43Daj2hoeRnZNrG3p!<Q
z^lN27xE-yCZky#sEtfiz@FvS`wA?!B=2asyuZ)jEY(KDx^2Zc8XvB0gEthJI;dHn_
zRpIcbwxp~KQrXp}t0QYUb~PXGr|LE8YR3A-*cCbtEYC0t+-T_rc4-V@$x#i1%_bNr
zFAPRr5oEp1*xTwCgwVX(?mZKlLJjY&Gtaitb(~Q%@cKw<ovO?25<_g4fJ_s}s3)Y2
zys{Kf^F17%W1hCO0z2{wB?C^RW022e5-t^caV~RdoWQv?U<eS4119K56A;ya@QjJY
z$?3mCEqq%%_31U~GGShEF!JkGnYVO|wv+M}Y0r<R?I|`fgOT62%jD8AOooz5kK~8U
z-RGZFa$>^tG5pUsAcm^^Sr@|j*;@)4qDuWh)F*cz(}P9<<We(3HJ-0Se1UN@1b91)
z`41Kbjpr#;NSsp3&+Zn}k)9TtXQZy4pr@{#pa))aAg~63brX!#waSVxr_;K4PB0J)
z2v4c$;wd0c<{cXMN*+@_IpNT_bx67zNMunLPa}K`{t3#8)OEV>=A$EDsu`)C2}aTC
zV~8kPb!>t@WR+(02V|WM(lK0~-xA0zsjJv@i52&Dw-n^IC=ZWE_bbqwsRh%~GDM3s
z48H7`5briN%%RjCb+D5LjfpDdS)bx`SPg6u8Yfc(BhR~;&<IP%$n1^GUNhH5W_&Uk
zra)!0AA*c@456^&l`XhAkCnZDnB<1GLE5;6@9PJa&J;n>X~-AT7Bp;8>i-KK&2VMI
zy4$LT+q;n9ZKeLFkV;VBriMT7QX})RZMruF=cl?`LL*dc;f6nQ)((ZX*#YFV$3Y$b
zUGTSd4E<WE`Y!%%a=n%g5`b|-V5=JL?pK2jvbiIG{m19BRPR?ny0Hz9DeZVn?Zm@7
zUCHsS=i_xgu=$YVdlLqK9}NDTwJ=U^2C1F+gFQ0G_daPWXGZBq9@VW*`dX(e^WV~P
z_tmEAT^)1P$ZV&U8-x1Ql<VG8_4icoJ}{r6dcO`0fOcxw0-J02urqWnp8aaU;DL@*
zWyQFc@b?_|%Jr`pQoA=}=NS8QbH1l{#rJ2>_J9_63HX7R()kz!K2w9a+va=<!G`>r
zkyT&}C@_Y2G~*74_&e($X{;ix{m>^5jeEhbk9)=#@`7JkG2}UY$n$#Yv&ilRWE{+W
zbIylqcw>fbG9%8T?KVe?IqY8ZjXx)k4?M=V@+>qDjtQJLMv6S?Ya<ge6}j<kj5B~`
zsF4t~*m|VoavUgg&DDIrM$LVz78CRkFfbu|3}<o$x|u=Dg&Rrs_6=${+9mWj#dhv$
z^pEaNr8ymk6_z6&%N05u1B~>@jorGdVatLSxOi^@&}`a)&5d8E`i72qc$}1XXnGRu
zoW5v(kBM0ha{)@!feqsp^8kAU+&yr6vOZhb4Vd*s$eM+EiZX!v1YEIRm*Xi;KZd__
z`SrK_G{JuZ;ZZNBdWy9@&C_Umgi)*Kd;|^xiJfZe6dDwSu&$e1+Y%=Of>0Xbctl_p
zww`fbO&m#Z<Geuzj(OBhu?}8n#yF2^48Vff<u^i}^niY(yG_qS+TAcl1@!#^{jhl!
z7U?u{g*Up`8pOnwZ6PrwyL>x~w%1&PqMkSv6?55zV61U68wSQ_GzisrMQ9K)<q9AY
z0HK)#V3Y(1jVAyXN<ceWubk%%mH;tg2@Hn>h~ZNJ_P!;EiUC3ZK9qpE1dTTZ;K3P^
z;yM{Bl%W@8C`*QZEkk1d+5h&~5(IDGFuk$R{3*(9K7p)l^(oG?hQG|jHWo~LEdfsk
za&*9jP6YeqXIET(0S^GE4LBzU?Kh3Ri3eBEtlY91Q;-3MGafjY=5ZG&fn!G6_mK(%
z<D-`XJ)+7J72#%AxXZc5MY_rj+o?di&Hvig-E9na5Pg^O=y5d4e0d@&^Nfk#(&V04
zuWK+=7%Rn)*7@Y2qk6Nx+xu?7xmjlj8lg3yulF9A-_N<7+ceA^=^f5i&A8d68TY!>
z@-1p)TtpNrGH!v2qZ?{u=GExNC0D^%d2959@fcf$u$`{MA`3>^J=3wa{{Sv_WM;RE
zeVgh7SmZx|1^i7|x?iYFIm8`m>=9$bnme{i+X1Dx(@FssJ3ZCkHYrn%z}*o?`vPd+
zD#gd;=Fw2Gb5#8Wu)YM=r^MO;Vt)qK7fSJ7rMN>4ub11o-ES!kKY&Jo!EzIgib;ED
zxy{OAj>kb=oWID&p3nfTJ8BGHI_?68KX<3Hcm)>y6u2C=KG)xalFyD}<uL%Y40Uxi
zoa`ty{Y93qUuY0~ex+CuGXPb`Z1ba3ga;|y!%GfQ^<Qeiu8#zBvtVCViZxhKHCRzK
zSW!=l6sAIn;jh`3`~JL-u{W#fhdR!Ji)=auqCs`^hYw*$p?MPb+s{Jl)Qqdua0^a8
zK=;a%%c*}x1>J@yu1yAG-iXqA%t>!T+bfx=9y~C{oxKI>ATw!j*kG+?{$^eXT&lX~
z&=z~z8~T46RDdxFJRTmXJaW|l7*ITKK69hbu_lT}ffN{F%}q!YI~Q?ab5SGUFH-x7
z%|Hyh)r*=Wh}(hK2&`VjJ`WUb2!5n^&r-a%yI?bRBT6j;oSTSDr&c47NK}`UXmS=(
z`H-p*0s_9u&A2UA;zAz+JFLXT+*I3zcC=t(86K0X@%R~uVgA9C=^`N~CI}Q_?t_%@
z`P{LyU_B58*`kz)Vu3ZKq_GKU!Ix69l;u#o17X(&|F^r9OtnnOEUZ>Cm)41x+Y1o|
zCMRYg3V-CQiu_0h5Lu1LGL%3m!4Yi`FLWs-$T<@^gApMOn;=gmkc1SaJOsa@WKy1j
zYf&~SQo%dIQZQK*447MRJDBTJDx9!EGR5vhx-zD#j+X~uH4B0kk`xu_V<=vOYT=8q
zJD>}dOdrIgLQIWPOxUmYV&C4IqGZ;w^(dL5#UNe^crwH|5n{}P2ob5eAW0;7T&A7~
z8RmhY53#BbQH8Plp|FxrZ8`LyLHs?0*<Mi#s6bH-)v|)E9Li)(iB==fPHYV&b>uYm
zLl~>5Me(-<7SnXNJKz>#Gw9Y`@D#hz=#`SPET}}KWQI?<^`KHxjTYRkl*~oWFp>I@
zGn<>pSu{5x5z*Y5kO!!Uys!Z?5Py?l4`s{qmS<td1v08ZksbuUd60@Bh<L@BKE?Yp
zncOQDtg(B*s%+b+(f?6n8OS<!v8|vTh_<4l8KCc36<>(<Ahf%vc+nFQ?Inw8eN0>U
z_Zv$>K}<$@Zbx~3&GH~!x?M-OY&t@9v*+SalD){-sS=fQzi4{BH!uiLHo#Lh1BiGs
zM2uc7nhI)>J=TXxgV529MRyjK30vcfeTnfrA!$ogvQJu)vO1(yIME@KWNfKaTfdSD
z)z-C6(5OAi@$X2hcM+p_5Myr|6<{B0F8X4u0d*mZ)km-CQH!EM%Vyl0{)Be4%Qc}0
z+2+w>5|Zude$b^oqDN~I=Xv0Y^E_~6g)g#kcoNZM?ve;1M~kDp!jp8>2}-Be=JYt*
z8@gYUaIBNHJ8$2xZs`<QmfztH!~19ytM|cYp1TS2?Ax#R0T=J(TkeF3GW@FLJ`t4R
zqn2A!Aj9WRlCJ9p=`OO|(n1+_VxZ<PzWz+CUlx8@2e4n?=GXtM>R)R*rt@R5sxWHO
zG572)%-~wV0{F#bC!l{E(DxxGcT?@Zapqg>Ukx3C1wZ#M1Nuk#dZ#%L02S7c8682Q
zWckJE&m@#77K%?!)Sugkc?VF_{j1SGlc`;1p=Mw?Sw}x6H)q6`GqGx<j0f;C8xb5y
z_D(gR2R)<x`WraMj(HVrQ#kU?RW6p+JQbf=Q2vc-421P<#sp*ZO4vwPaWFTn#MyvB
zerIPuk3M<mICh>rFivVZ-UZv;eWNR&f9A(4Rai@@uxk~<O!a%<yH{~>ivzv`uu=%*
z?o%G!jylGK+pmA2clZzOQ1v5NG`Kaf942lM=iZyYaIdA``4@fvp&kC*9e$-?N2KU-
zCr)2ztGXBimQ?*O!3#FyWy1^F(CJ}o`w4;GeUZ@v)bfK`WajU`5maDv!0Ek1LFYEL
z`71Y;1&0p3>(4y`nnxn@#(-vv8vYxMHG89&hn%hKlXD>3`T}4~(5h%%DWG*FpXxSx
z9v)lpm<~J3F8oc$QO;fXI}U5h@z>Sxrglt6yK2`5^-UeApuPp*W{_!B&ZHMp&zyen
z%zsPG{i61CzkUps>X%UhtoeJ&95~&nZw`N+g4e+?@%^;nAgnPTbX*fY?$kdBT@*gv
zFXZxv#~DuL7weTj?r^@Vc5nH+Vf`4+3d{*2N2`6<QUshI`@{2$V;~q+{up((G1L_&
zzn-CQ7#C3f*zR}!8FTbwxv>Rnah7rrEMSkm0{+AEJfr(5OPCxDU|4?kLAL!&bR*gZ
z&}8&4fbI#vY%)ymK-UTA$Mxn3k<@IoD!<-_^zMH1ChDzhWc$3^*RZa=>;1Xglwe!K
zj!<?$_h6}yX6tA3{fgaeUjQUWn`Z$P>--L{Sg<A5hXPqYkc@jbCX;&1-@O^FQvX0{
zxRq^BP&L?!OFS-E4>VX5^c2mw59`G1VA0v6dJo{8%>LTi__<J=(D;h)UaZH<i)=!$
z|Mr+SunGuWz<%tL>*250*~}Wg9<q4*18;u7%tRB%$1rioS3hz$humtK4~(haPVLa=
z=obUjNGkSz`P$y%^e6-)5(<-sYFoqftW|j4JgNWB$k*QtpkiE51B0OCM3<)m$f9a1
z{{r=-f7R`3h_0d@^DFtAfINif7I1pjSVU__c-)gX68S0O1IDYKCURjcY86fq$(*Sg
zk1+^4%phgO=%3<!uR%C~?}BaaNvI7M<rVZ~z<68_ERmWf5|YLFBo`Z#zD?T4fTt>t
z-hr%a{gE$_`BbBhh8APcQceFW?O<6!0~0BBVw@Sru1mmEM&bp=3#;gie1U3AYu?i@
zzhO`4EUo+iE2dD@UpEI0pd8=RXL`nwPzEdpz?l!V81%Raf&Ox8h9JPUR7eUJ2Xr&s
z?yS8q`(VeY6WW_h2SjWhc2qR7DL>@U&ynt0aP5FOT>r!${<hzoYc+kRAFV1<@+Vpy
z{*XFd;vdr|9}@|V>CGbfSu!~+qy5zIhIA;Hbbojgx=!s|IK8PElT?_MX@<T68Gxv&
z!&dyU&}$R`bk<%(*1@Mr^RyZ%hBdk)1t+LOLuu0pe-6Vh_0-<zYd~{^z5``90Dm_~
z6Ey6${pMVQ1@;P9oz0#$28>&1f4-9%aym8SWl(TXiN|4m7U~eT=gqJ^M>}iZ#@ZRc
z-V`Ak{p+fCL#<1H2L<Yw-P1n8pP==@D!(~2h-KVWn??lKn~yrB_iVBj>Yp;?5m+1$
zCq43s?Y;2v)Hx%#*$@dmGDv?XW%f5Iv%6AeA4|Cme7UH^YCFQ8Ipx?p|5ezX(Bze0
zn`n(_5ol~mcm?IgRz?dW&yGW(Y4AW<MTT6hmhV8v?Pnd#Ks^RrZs%UL{M}?|aIhN(
zX&e3FZfEVq7(V>%SRO3!wbR=vyq)kmU+l<cgG%Tx?8qn$>?$J3q@E<gE0WNM6a7(`
zfSR+q*N2a!&U;d{Txcd7_(iW%83fZP#Ul2$?#<ns#twN>S^UB==ofU)42hRvb#f7R
z7vsVkXV_GWoUZM=7Spv2U}<$K4+8w)72wev%^oGJlsN*j+pK=AyE{q$6zA*Pg493k
zT`<q~D4-lm0ihqF*PWn+HghZyr>86x^ej{mFYdId9x9k;SyoX&Aa|N-=yAP-eFYhr
zF4a#~k9;kOgvUMWkS|avkMf{gPy3UqCSY9hjA#M@Wraykj$PyUsBsD*5ony>N{#bd
zsd2u9#`!dKhK6YshXV5{&<13FW?m#X9=jO{z2YO@g#o>tdgZ&&D<kFp8pZ1C_(6i7
zUiFDTd<@FO??onYO%m6ymKym-^vOn{PlPhrC6ozw%X-r!p$F9ejTXs`YY~o5|2+Ni
zrs~}U{qc6k5AFR&4SyEb9|N*ctUVZ3!{0hXXX2UA8c8bSZJ{zQNvMpsllo)Io!_L~
z*_CqVv6S;|`r{Mo4<4{}<Jd<o=Dlbk!<1j+bpp((Yx3A&(*2Rt0Rg9}E``0>2Rn0r
z(TvoTxoENI61M2kh~^>l8kaCBnse9iy}B0J1K589W5+$KES`!JD(t-I^KAH8iO!3-
zix?N)SZwRORv)!i8B6}eUc1|T1G_{$y}^Dy0>dL#LC3}U&lsAvW?YLo(Il6(ajWLO
zr-WYxT*bsj{H`g06;@v#^kP<j1m_iwtKPq;xgRPIuOU|M`#5Y7R3>z2&Uci@)(3O<
zDvvd%Z^s*mdoj5!V8v*;-ztxAB7c2k-0$Ex%=5t=F1G+-<{tERaT<nuF+$wU7mKL%
z=64}(Ei!I76!23_lzq^=^^Yi=^8+n1Y<c2n%C7+{(7&hYccxFor5Kuit*hf~wfU2N
zn4{vN2>pHY0$^y)J2AR$Ot;Bx-Mi3t=AZw7U6N~%RgbVv-Y-#kzXm2K(4&tm1uI-c
zj_t2B6hc=8%bSh?6x65TZO%P7Q)Y;FSD@NFVu<GH>rAEoR#Y!kfwJfdrV(#Fr=h__
z&9jiV*t;z8_y`ZTr|@9fK)n2d*;#@9Stx}2Q=FFndiYTV0`6w3SF#VT5t~!k;mSc-
z3L-N-Serd1H>`Hq;&+=Fh}XK`#Np%KVvoFRkJZA*Qk91nqhyx8!fruNUT9wyw!V|+
zb|l3lgy)mUVR9B(>w}WC*fkd?N8_BpAskgKgK**)KLCbB7)&b-51=ukhCjtjIZZ`>
zqg>e!eJTbQ7giV^j2`7wZQhxPi8M&Dpod~B8^}3H{T+3I{T{!V6QRM)e}J*J@tHl(
zry4(Eb;}yXrV}g$K~IjeIavOYUtH;f^|rO%((d{5Ts7qubgsSFT*J)E`~g_OTVb5j
zY29zteyCx?9KG77E?iH}xP6wDpZi(u6IijU3$dD_c$9|Mks%dFM$o8siAztYe_Un<
zDo@o>Vll$1>4FtZVz=;pKD@^C9lXY5=idmO4J9ACu~Ri(K3j?+bYmB0K4)Wq9(^OB
zavTWVh@P?Dr7k=|tlG`IW(_DUruKW}2u^aujt|_iXw%B#_1H{e%fs|?FS?X%i4r@9
zDUv`N)=gAl2`a~9%`%j1_5JXf3h^+tXn>tzZ)lXc`z3~Vo9Cd6>jvR?;M#NWSo0kw
z@bk3bhd?l&M1zPmbM}Iyz>XoD7(eaT_0%DqrvDjog`RsaZR$r3$F6Z6^EBY^!l?UR
z+GGoI2z>0C>G6HI@4d7<z}{=#`OybK&)oN#@vN4|vOud&9Yx@l-LVgLs>%dLMf{_)
zcc_POT%ybAe+_cOE7#YVJci%V&XXg1luQ$ryU@J^?F2}M@xqAJUT|(6FCOCMF82Dr
znGb;k4hMUD(^X}5bh>g^+C4}#T`5Vc5$?Pw2%+uaj5}T-FK$Efpl8Mh0ndz`XDkH9
z2PK{|V9fNCsok&F{t682_baY3d@M*;>R)t-L-sfufP^8B+m^eq$-2@;gNTxeX}q<5
z14fZ26VDPX!oc8PL4myO1aHIN+AVk+{;qK6VcP@e@PRyKCwM~wB*W_r;I4S*8J_?L
z+!X?+2=C%`zghb<xf;O74e3he$D-JRca+X0m~(;Aj=@UiW+n42$=^jwKxV&k=R5nA
zlH<te2r`oUsJG!&+B^RWa4W#C1KbK*?SADptVZUeIP<x2=~ho`)%5<BaV;?hZ0OPX
zjqy8zoamT0RBPA+ob!OT@Nfg_1?Mk}TW4qsn;%5W=YBQsw^Ac>aFFEs0cyiRRnfKu
zjUjKT#(iaK%IJ5s$e?~&?*4*vcj2iid0VwG<`Ld~LA|t0)d#&xYA|}K;Ufca?xs!E
zLA<-U_IGN_b=8nEkC`DH)NiN`8rPKx?BvW;W6)bcW6--dM24i}=tMAgyL#?HK9#&?
zF8G4p)}R+U5A5iv$nFMY*No<hLaEBNxar{585@%3?>Xo#Vtm7)0L^Hi&J*KXG0qDI
z%U5!I!wzjh!@=IiwwEBJ{Dy<dB0ur<PNAVO88XMbh#^I9!K*}AwPIL17*cSuK)g%T
zsnq``vk4z@%5bO@ew=5-dq$n1OF6u|QIWVH;JQLSe0*f_;pW4I#{eyIEoKNel<~yP
zcWI#?jQe^9`+fM!!k-&|E@d!`fvs3U{exH<)|JY`ZD0)q?fMjXoCAV=Wx$2``m&Hm
z8N84y1UOC7X_QW#aQxLEfi4P$fP68qcmNKLLpq%jxm`Mfi@^({2-%ZVOWZQ)6iUY@
zoh<24lQRn@)3+-PSIe;zFXOeR;eG}&c8U{>*06bIDk{#4F-VJ`#bK~?y~4>0EA>&{
zw+wI=+LDLIv$$vlbqu;$XcVe3%)|M<)oc<&Ws4qjHml{$>OJOMoXIFuonNSt2itIN
z$Gq_fT15i1lc1}UK<xyLOahq%Wh8+*2}(->;o^zU&Haaa$cuwQX1fi7^T0%UGYRA(
zXiX9*gP{LT0=WrVLQu-ALKG$wVay>oJZG8n0EY>u4D<Q?+~Z2ammDQIsz)I#aZD0#
zM?#WfXjhG5IkcN+zrb@FSk?x)c*dw63DRDO5#2aP)rSnk6>6xDqJfy9<1RILV6=qx
zfCpzIZbg!!fo^!J>B)pAgPtsS=F*b`kDDH!X8fd1)LPL%6~Iga3jthY0gC}d4Hpe8
z1CULY{uuz~&{GYdPzvy)HgEN*`jUa92Y8b}UNVrB5T!DT1Q7L%6~&~8Dr8iXj9O_$
z5d~4jj3NxF*zgz>ZhS8C+H(*mj2kmK1F6FzC{~Il4m)(BJ;r6NG*q$ia|WdvGb}N`
zusm7JnxUxr;^8W~g=Rp3z+Wi+5c1;T#nKNsFCNY}fPe#0FCIQa`XTMb!>gqq0$)6Q
zuJl9Z!Qpj4cGI&+Gk#nr$`Tyj2q0u09KIAlDf8g)CIBJx;P7Vvlrj$vUkM;&9vr?3
zKq>R!@F;+gd2o0ufK3+7HUJ^*;P9OQN>K-gw*v?{2Zx&gN+Ab_cLE3r2Zwh7D8(Bb
z?nun@rasH2x*-a>G@XwPQO3*BeyN0cfEQD`oW!7cM=YE8NuZNvc~@VLNt+jV9T*Sf
zRIeNJyhG0*C1!5lf4+}--a#OkFQTqve_(%-^E@Z8<@x@}7!XzK5cV$Aa!{5rX#AuV
z%m<Atvkw-)Y#RPVSV&>>>{jZzf1pKv=+Yv0;_m@h0d71@Yz+ay(r0|Xy9GNrw0Yi5
zbLR;BJxI?;{7s_g%IV5gL%U#3y=v$d=aYvH>)SECKu+`DeDct@)0NpUj$U<Pvxx8N
zC$Rbfi@^0F*p(uHDRH0(ngV}cf}RO_2P8e59lPqn!y?`k1op-09T7oO5ZD)|cLYJ}
zSnyF{s`V?!aV&Q;>>?ZR*W%W&b)__%g|-^hKLiGgo4j%5SLVO2<#yKMq~&|D2iVRc
z^KnAB3kRbxQ`yO5mflXD4n70+HX5{WB=}Ir5WkW7$0z-V#=Xdsx-a>iZzH*|D_;dq
zd_M|D+ljMU$Mq*GMz3*h^BaM+sQ}GeovLpKEVw4siTQL(zhG|5oHo7nc6pXFo*vh6
zPR4XzGk(q%Twtt?vdVLhhAx!zF>G`QBjw}2W$WCKq2=zb{ldadykv7PA^~fX6xeA<
z!7bj0w$e{NAw6*u>93UO*@M9W`|L^S&pIJJaTDne{)YL_ua>NuGE^)>@5_)+hTf2&
zOc`1$LRT|zx{yn3IUvHfL9Wkj^SOmz&Sjg|#Z|Okyh=`e$XCbV@~257hOtz<RZxPB
zH=K%}2e$G0c?oeNc;^UsJS5Dp*@X9iNq|!g{^)O$Ug6&fKU$SIV(7&R9vjY{>?mIc
zX4ctTU<c~>w-eUa#Nqqxf;pUyo&ikF_{CC^H=Y+l8~J=4L)hI>%ioDxr}V$@wE`Uc
zJ7jwIK);-AwU3%X*hOpP8}y`vot}BB@iUPg+gZ$nwi2<svlc|LqxeziY*=NVJTz86
z$8wv+E)Mr^eB8U4$Gw}9-(#DX^L^S|A3UbtD)w*S!Tt?W5;_)r47)ctrHCCIxp}ig
zGuDZcKm>2F43Q<CR^cx8Y=m2Y0|fVAGC-qIhE*K_LtOX;JM?ma6To#`Oe60--NOc;
zZ<I6QRBT@1f<qF-5<Id@)|NQHYEHwxmmf1V#z@Fb0uu0~_JFRjKp5=eSeIC!L2=em
zpzMRIUC1P;e{7C_9%hhwQ5cLNEZbOrCX6p0rC~1u8cx&Wygcri75JZ<v#=8@4%3>W
z1m~Nz$StY<7OW2_jX2&xvlh+8Z=BDi0p?|g%y)i;X1L7rG-;M^(u~qVP9SMxhehXQ
z0XGMyg9e%)x!_N6;l5J3g99VWxDuU9Um<-n=<|uysaQU94XP@x61;0axMU<22}L8_
zSV@n>G9h&&jQu04Nl-<TVL{N{SVZ2Yz!p@Xe+4V>C~%l~bM$)%(2$dX>(TtVf3@th
zR*Unu3_vPU2@S{6lbzBwigx@G#$OBL#K;XDcz+5fR<XmtwRaS#*wW)(jBLwNDLe}>
zN7IjfEr-e;u0tEm#Z2>B$JY4xX44PhFN-z(O$}EP(-4dUxKZ3T1G`jwuY(~sLs>!?
zID6~k+Y*@mHk<FDUbP5c`XSi7<yrKMlmbXe#n8*E6Te46p{1-uTFqE4nr0AYFE`eY
z`iENh^W$O@coj*9YJ*rsV5TkJ$9M=0bfza=i`<r~MP{TH=*<QC2U_^E<JmZy-00U^
z{CcaAcgz@tom2Eb9t3{~??&BlOj!Z@dvgHG8mkiulN)+s?N&4?&b?hh%+WC&1o;BJ
z74|z0WL-FRv906&puS7>9u1+fb3_ZBgTZVK2dY*2W@Sa_H3v?ZbhqFlN$!&dHSf`a
zNN^PxELp>y-Pb@u>IZT*5b5qerne|79y;bc1l{02luIhL4(D+#+$<D?<ZZ+8TAW9!
zg@n*k4WFCRgna+j-R8%A#^Bg3t9bJ-P_+nJgTkK}?A7{j;r~}M{v!;cg{(tk=&^O2
zVBanj*DK+skM4rpZ<P99q5{!bj|IKB>U%v-Lw|g1t%1s7Lwxt=X#3vJHN3W1V4PE6
zl;SMYKvwj^<Fld8?i=Y}4K?Dx8anU35&qR^EA)TwzO&>3D){@~H=rOgsax9~EAYOp
z;kIclGSL~#-KQxNwu!wd_WT0<P=WU?ZTsGdkzu)<;zR$!e4*pK1YTQs2lqLR$=)IS
zZF$Be^?Am)<=N1K*W$uQK>ee}Ez!H0&GWLmDx+siKNlVFV8|8B{qWxNV2W_=?YMyJ
zReE=K;4Lj{{^vJ}UP9F&UUk0f+?1zZQjY;;l%(h!w<J%;P6Q4e;l#?rocd3~*_L2#
zb4_glZW{A$)wX|8;KUt-iTz#j+X`-Qpb7k*F6iL?Fu!rhtA1nH%T8R7=(qSy8MZ{<
zNq%=$Mk|csJM<0Sud_`k@mG=ZG|55suqA$d*vqivAk*4kp%K`(GBAHh>faf>&y<}>
zY{jDe3**QD7?`Nfv1ijE=f&9vIXGhu?8B*#QCjX>A#toeC0`$d$zp0yUyvSaLjul^
zurr9cUvpPK3`64hPEZMatT{I-k3}&+;26u*$`k9;*Q?mh5Wj1%1Jn)`CQa<2%Xss7
zB2LTQs65gD%GeS54Oub={elA=EabINzXiB9e4nbn$t@wdC3!69G(#x?9WTA?;jw>d
zWZ!BYc#5;xg9VX!ze{C3qr`{I#$~eCU|IitmI2o+{sslO0pYKN62M5Es^;Q=P=B!6
zG040I?K<{D`V~kv`K>Z2-G|VGc%gm{>7qMT1N}sK<25$@0Qyk8j2uONfY*F4R&zJi
zd=E=knu~$Zg2P37Ax{*+T*NEua?P5rS;vPu%I>{I)4$T1zsd?mrg_p&*pGld#<<6R
z7=k|2QR=TaFo(c3M~3E4vPkj@l63CG790rQx3zvt?dA>)tb5I2j4%e_78B<QMU?tW
z*w2e5Lm9jUeIQ4M0xY6cu{${t6=Hwi7^~;q=Cim9Mxb=`ys?{(cSW(8cp^H7_6JTx
z$5vfB-lq*^i>u;spEyfrz~u3r>|KpU>d)~C;4jb%p0vFBkSB4fC78Qm&L{EakC(mT
z=OK;$YGo@XDVQ(D+;}x;>-I04TbsW)zj7;9j|&LEdUx$l-oDg{8{+n>mD?s^J3?Fi
zS+&FQ98<Bg;u>Hya(r#T<TnSJ&GXgDC{wfS4F~;em;|kQyH>fMUZIeX1}mc8(rgY^
zZWd?@Kkx1)U!S{!r$Mq~@SYCx;@djCq|>}nt+G{;|2r&I?vvy{lH@-E`DV0obO9*J
zQ}AUK{_hnzSlN66ef2bSO7^0EDgRJj|5E<l|6=|#|Hb^HpZ`ncpYt!~-}f)(U;P*J
zU--Y`UuqO6l!YgB+hTdVR@GsEZ@F#;bdtlqKh-ndKt*G+b{s>iQXfu*5;N|E*1Z<z
zH}LHTd^lyz-ml~j2c&Tvqg@Fvr-EB`q5SpJj4Q2Iy_*lG00&3(aL{#YF!y8S;R&D@
z%>4+<g92s3mzwiU3?#u^6X!b8w?gxow7u;K4byTrDUZ0B_dVavHE;Y0hoENAB`Y`?
zf{`0{sAFXpj7(gldG~3n^hNTAm*}-a?U*d|=LVSG#?ubi5eoUCoNLEHBgHPm^fRb`
z)<~1d*-~F)s96hQ@q5QHc+XX_jrf6f&&Ii$^V;;-LyW~nEUwi6g<rC?I3+>jRBOjX
zZVp|o8rQ;Ti`^I3wPoCafy<2?tP!vE!Nm%(zv7zYQe3yV71y0u7f!+7jaV9LSQ%o#
z+>9lte=aSiAO-V{KG!0J_qzlsycZ)#`A#NP3{r(4g_pVTi64>*NB^m$a4=ALxIgz6
zFmK#p*+Y2t#rZBSsyAj11-~<0U^sw_qXq|DU^#$$q~HV70bDBuAJ|SEiY4+;7naCF
z-B=>y&Z)sAE<B1{U_R&q_o+Bm*)kWW#Ej)bO-fl`?xKqY_|Tnn2iLoVi*TzI7B0eE
z@G)Mv$aOC66C1inBwjLfsc@GKeMTTHb4hoJOTY&&b_qOi3DLF8RqE@2MjnLi2t3dw
z4{|H4-X&aweS{-V;Ub)6g@uc7C$j?^4v~JzP?vC*4Rs5=WjOB1^h;a<4mJcH=rJ9x
zY*-ZM<kg`&fvM`Ry6BRVR*|yaC0vA8Sz+NK{EQVAF2XZJy2V513U|rSMZ#S+bg95w
z=92Cbmw<y4fd@{Q4(CbQ>EynR`7EsVwmx>xdbHb_>2tB`41*934gD>#-bwCXcd|q+
z@?L-TFCJJC{Y!Pi2qWx{IREyaatf*G-_i8h>DoP;=3@7m2d?7tvG~1G;$+nWjj-i^
z_L%klxqYg2s^J|gZ86)5Pu2GQgR_<Yo1lqf4*xenv(4H2m>~b(1Wm4K`kLVX-vq6D
z!T)%ImL0Xt*08?+dttpb8Q(U))87#XbfvAozt(cC3)&eMW`usIEc$ou8kBd$57MeS
z&b4ii&~(yLpkIAM#IvG6pDu1I!fJnr$GphE8eN}`6Nqo>ru}>)rc3j$>V7@1C3OWz
zI`mD=dqy}nZT}XC4ZC!6!S2{%3@SKflVAXhq?3_Y4I=d2hjxd%kG}@%L-!g!lpEf4
zzRy?rR(Bx;42*}{gBNVcKB!%=U8(1>-vZs#ysb#3+_2aA0q+kAe+HZFCcLn+Ck5u$
z?oFZ7R6WM)lR_EzD&iDait{izS@Df{e!7;~^LY21lItUv{O%j%qece(0YJ=r-V8dw
z1?kn2m$#g+^BV%#2*#WUq`m`E$@Uawa6ZQN{7(F};}0{@O;L<|owc7A=$m-8Q|^AH
z{t=K4=&?iF!-tPsrQALt61=#dz7^-0FtJR<+;t?5Fon*;*`?4xN|Qy05PSEA52wz#
z0-ums;g8(xRK<Q1@zGvdob<H#k4e$}@Ea|>=@^bJb%xg;YuK@%UEhS$?X_p}bXBtL
zZ(U9S&P!g6$C~eh!s?0enJ?%Vdn-}*iSd3FKf`Hja;x~YO<NedtGMb1Ywx#tHXJ8X
zxp}nJJo_vNH+N^~-!N&#)P85^yQ-1J-2hkt4w>it2TW$T+_uHM19{0Sm2Sj!o{T#R
z79_W@r`?GIn=NJzkmB~%$uKdQf5IKwAgVkO#g#S5_0!WIaB&l-J>vQy9AFsa(Q<JF
zHk77`cTFOAsc%gN3J!}DFX-?rLblkT7<2+XMB|vRc{Vu4w*c^-k9jbV?oLx0vOycu
z7^OawnP8_PT^ozyh0f=?q(3ZRgLLfftxwf(AUJLfg+02!xCB$GwxE}HI>Si0Nq+rB
z3B)jEf=!eSxj6=#XmfgV+$Pw44Qx*)ESP+y(h<7E2Iqo`vt1ksmnY<pL44c;LN<b_
z0{wW<`vo>_wB~MPhxXcj42C{%$T#a?N9)FsahCBb^CO60Qe@sMsa&fG<KKcus3C@V
z@O>Tb+@wCQ@rW19?AYMsxuf0YYe<6{>W^F-@Yd}%%p(aN3UI6$0dp8)Vt1l<#oEcH
zk=C)_4Vi<CYMe3RYxsEuvL3`5!ALuYVZ@1Dq|8oyPlM-iaB9|ep;i%hm@-|wUJKP#
zW{zZmjkKq2WSi(ukOeyHS)k7~F9B!RRAbBN*;G5lC0zAD#5OY>+a+AekM9*)<YiAC
zBFu9jr`R+?xYw4JZsTQK5-&#DXe;g%MGReoPSeMoV7;@m;0$hZ3RV{X2wWsf=rk7D
zNPBpn6fOlnpZ-21bUF&dI$RGo{{Dyfnkkn*cr6cYi5BbQa)Ws>u%IYE`#qODpgC4-
z8DP^fMUCL)IC1M1UV;^N@T>+L-{GKS-Xc@|0u36eG~*gMXbs?3IgyOt<HVnp>~MW|
zcS~9~+6EZFa--hU1b3-u0n+-*eey=Y);>%nwi?-}<(mp3p3RJ4_uzgW4&g1p$#A|i
zfqh2VX?og~(lCyOhuqJU`Ztk)d(CyiE%MBjWbYzb&?DHQ_kGZVr6e?xT8^Fzg_g4}
z3<ho+a7`}~c3g<25&Lh9M5N4hX?$;gEft@cCWu=v(4sDhVb-xO!K*~K^QzKvftWf$
z$R1O|7-Z4J82v-@OB{4$M?ohl^2}Jfhyq#)fihTa6@y;~7cKXMzNR$X!f8Y9CrbS)
zkrA#mV7fv{=0=N5u|?(rkf9JV<I)%*q=8P#vZkZ*%=3{1Uq^!|h6+(wajaM=2aLsF
zEiQ);`uBsB4731<*vt7V+ARPELv}E)NQkGW-rmFGU?{#yzI|(d%;G;O^1z0^#dqbP
zMDc-Ur`7<zMDY)z_-xqT<Q;D8kdbb>q>EQMg;+Vnv9%2b-CpI9L3wn6r1>2rk@9!|
z8=z!V#C0N0%45&N$@0iB@1fLDa3I=Q{R*rS42JCBG@CrcX0f>o^DNciPBE}Ev7HT@
zSE|xbhE3O&w4ZibolNxCx`jlQ-7T3cF9d@YrAD9@m6zw)l$KJ@tv)fSvMw{c1Yzh=
zI<ep2VOMdO=A8nYHjHHKrc+=WCt*K+3henOVgKk9*grf8`~Fj4Pdf>F<|(lGo>y`y
zc~G@430{8^_D!e29)A*c{wc6osfjBel^@510(sfPX^weW9R|900kI+c6<1_)_z?(a
z5Me7IHiR{N^(Kz+<pLl)1BeabCBB>(M|hnGO@P=C2HUfEjtJe!Je+Pv(23BMjIiB~
zFoOu|l9_C=BV0!We$w2=!!PWm_#P1=$)$M0j*v!#Fd(s;DSS)|FrrpNIF`49a=OZj
z>TNix-ztpLJ@Wvio?X1J2Js%^VqdUVat3mu9<P^{ZacB{ry$1O7<;ORSh7~E7k7py
zb#`8M+ZSgG5!ZoP`-a%F@Co*M8PVz&Gct+jU8f*gAW6l3L-hvIt#Q&PjuZ7SoJ5yX
zsJ@u3LR?27DSpvIF<H~rx1B`sx2K@^pOTKa<7(X#7=GlEYbK<-`Z<n9W9NxR5i_K`
zW8YeCB@)`2-z%hpKC%<h=6@rN5T^~{rxt<>uK6P(EJ~(8T}28}Qm7!p-F5^yhiqr0
zN|t(?U5M)>?6s79VXaNaC(V_yi!E|CNxb7h<h%#I1n)BE4D!7Vk!-%jNyAM&k6!-I
zvz61PmpyJ$4uNvV{U_wfS&ey{m1ok(d8#>h@5Pj*SV@ZMf@0z@R_+5*oJ}l+ZFK&B
z&Epy7@id7-rwl_~Vab{}KV*NOk+V;6V=K-MpI=b^@7DYRtJJke+4LGgV^ZXjPdJCc
z3SKM%Ie)}S9acheyKlgeJx4+Am)LW|G$qR{XCb-JX#a&ze83se*x5q~OYnK+f$;Gt
z(}(6*WU>yE3oCdDf<4?|Q8~o3cHks1lY)rrBclZ~*pJ(7?qJ^r8^GzA(bQgtZ3VwE
zW(~~}t7(fcZ~GdIEZoS#CRTLG3%b!g-5qyzVBwp|`$BrOVW>BRL{m|I%+o&XIjzXT
z9C<(5FTaLjLzrPlIGqT$0Map5XuA^yDJ}s|%KnCv`O8Dnj*yi;NdloelhTelIqknE
zI9qUeqQi&g%J%Xlxa)~OQ_cx*wIbg_Bt_o>*V+%{{nt$>H!Iy`&L#pXJ>9DGM#RJ`
zJ>6F6*Mgv}(#Meg!i}V38Eg{m^+fy4-wO+46L`V7OVfQHzAvOI7vmi$I$#U0a;Dw(
zJHGK*j5iuLW^8D@c{0@b*T$%I-$f(Kj+&QX;f|~KUF$B9;b92lO1E`B89W2Qbx|xW
z*K$o_UWTB)NdRCVaUnlc2@Z*$o3M{z#cQghi9_7jT<*;7Hm4wHi@!!Jeq*Bm#r?{O
z{z;PzZ&N;2pdTpkex@~ln+Chf{8}x7NqkR#7EPe0lb0ZG2L&2g*#A)kytLMBzGlJ1
zXl;xl=s63-x&aytIk-RBf;|kF7T)MqBYAEZ8yEZ%Y-6?kR~X!aUhJ>q7yt~sq0!t2
zhsiG7l~P-*8KXU<ag(4Cnuzo`Hy~kGOBhc5^}sHcuz$nn@H%iO#)eMN?>L(n9SWYV
zj??hW=tx04?k(u)vJe^UV?F1|a0HS!Yx6TdcrTduZ}o>cAtPA6ic2S+C=f3y49?5M
z4rC5Mj?jhjTi~!E@U!saJo2V39cSI}75k{F1h}0!F%!1|@QdaH<n`e=x0082uo)k^
z=2a6dd2W&lMjrG*cp2hzV%b~#tGb|S{HtA$PQ}{4+dQZf<<}guguNq7AHLYZIWT{c
zS6n9jdk)9qfFaMpj<_FzNR8CB`^_?7#;vkEl7<@*#MR|u@%4Lf&8)`a+To62@&a<T
z5Z2l-Qmr0+D^#WfDB|lU4_dE#ED&i9nFCM-R{u@1hb|>+;-jdl0qY7ZZ8J`~kUij;
zOA`wRPyRv|+rA&pl{K4{lxucPLavJi7-l@IDDjx9+CCD`cPmJZZXH}9>J=Dfphv0^
z#yvaYf`?tU>lS+2zc}PA#L1=iKlE{Pa%e7@!_LBwn={CP-uk#X4s^@UnqR_^&ghSq
zOssEf<6APk04;0&INNwC?gV#!XnodvulXtpja1^Z<{QX2?$l-ufw4_&mh<p7w@~=0
zQM53A)10qpx+#<y@$%wbO}uIVYr<f>!aNVcvILV56jtX}g}#7N@J4fs!})pQ&Dohq
zgQN4fF58C^JmgZHyDk2{GE3Mdkb?KaEZk5B<(yzp^qYCxR%2+waK^*oI6Xc?HHthr
zW&=79kGT4R<-cJ$a7suRo5d;kP?o%S1Si*rL`B`l8-jT$PQRjO8OwUa3|sZigB#GD
zUPsdpzeZ}flQ;8>U;P4|<x!5I;@W+)j*=-{Q%8VvJ1(IR{hyZu^W(YxI{2F9J*(PE
zeB0LgbncDq@NbtF!cJ*~|2=s_>(nUxAIXc<D?8zj;p$*q#`?o1`1i>#=HA~1zxBaf
zC%pVzuCGq<s*Q@buu1V2uT;EcF!Ic3Q@qvfig#|O;&tG>Q5q{)UD!+~{-N9=2oZXO
zw~C~~p3YO#Jr#@!L>EZZGF#QjwJdu1n>Z6rYfPcEkgi3LwU@oes&U?Dht!!t{pkGB
zC)PKOOf#((f?Y9L-P*#9f=}^Q(7ZM_$HYY(qSXM82-?CHIyC+){>d@Ph1KeE$kf7(
zOA(z=&+)H*^{626)vt%DfXXE&ODsxsR6y`ONVE*>PZgSY35-gz7$<I_-;?wazaA>|
z5iZ>LR*O93D~S9?Xwt}H5BnAmD(}K~+#nK{B@pG&%dX%HygPyss1vlXAaL8^TKW#}
zle39w#uM#uaK+j6<%1ekqaW@pZRFq?1S%rgDsegC1Y^i!_+&x-l>*wZ5>O)x*E?b@
zAU}AHTh{8+DJR^a%qxF!QzrmT62R-waa9BWOlaM*yc2ibrz^!p=`LJpk?vN!7JJ?$
zo=&pqUFLD1Xh;@F=T+`;s)2Og<=#WGE{S9?J&QDbq*+LsF3?aw(=FU(;wyLs`s=vw
zj@P;L=uf;)#|<vLPFW9v%(rR)ZVpXH{?SPtN+WI$RT}n;5)4Io?VJ<W^y3>0I5CCq
zFr+}A@y3Y_SsMPZS2=G*>|&InhdyOWP4Q0`8taw?Y_x>d#BBeULR)Fkj9N6?;C9?0
zUZ#=toraGuQoeXkkthiG7!dfGskn6{h-%F&LbYa=phi%aagt<@_0lL@*)-W-Z-@-r
zp*4Rp0D3xe`UIsabp)Tm5iUAKBiz%kRjNuxWZ^a9>Sz<6Z3j@weNCu|m&LaW5sCq{
zK)w=%(EE&ACf>?K=m_KFhwu0*vfKsY7vRYV)b4VR5L~`ez3oMO___*ea>{0UcEGb8
z4S2mWWgqBle^Hsj?@aJ>6lf!#AoeqUD0>syMAcqpiix<-`047SO$2vHma<sd;?EWD
z)8b7x(%lXG14#WL{`Mi!hq8n|rurJJR~;vnPY!_F1y6fiBrxT0K$-IKe!tRoOesD}
z!IX?B1W%t3o!_|F#oW+ufDEd8#C<Vs9sJgXcez;JGtImaWt87;B&yym6SSdW<||Xa
z2Km?b=PPad$V+g9_%!%oQI7W{VibtVl(&>AZ-e8l_*}&{Qd}~k4sej8$nlL0FtJ3u
zpayy!%9QmSr%ZfWBZhHgxp>PJ{4N$v0{G4BP11M{e2g_y+Q`FtR?`yEq$r4wU*uP&
zeX?ZhmCPlccI1gt@^c>R!AX?2>mZ42l1$Pnl2pqiQPlCv;=6|^=swDDvr@dCjg|<A
z>_o<gnTMz=&~E~Li}^0(BR;y#YJh(u^2FF#d<gz_rm-ZBtApC`MGjHVWECXeV>IP!
zk=ltQM+l1oMpBj{MoQ#Tj8Csj*~w^XO87|;nsPV0kYEg>K6!`SP(1NTdt9Ad-Ddgm
zvU2rK%P-&k1)$vmY=Ykueo$Z_`Xcr=Gp;XBZPbiNJ2J$!eZ~Z1(0}qq)QR9fF#y~I
zup9wK11#z*3Ve9vd$RC6Eh+?_WkSmEED>UX2U=JVi?C3}{i(Pv0Ru;>4|9V4_^5~z
z>cUBN;dJBMu+Byx;TyQlCgCn|G7;;k9W<ypI)w}3)aaQ3CqC8?bo!XQP$>>L%ar1v
zvsx)$;;ge0${}E}H3TdXLjX>o1oidV(Zu=_cO2qqB(EDw=znKJw@v?dQ~zHonz_*b
z`S|`T4z*waz=wE_rp(*MTfD6A$zWAD?}eCGh$=x3Z9EE&Adi1sW~_(?B>I1oSe+%Y
zMoVH%bpH{Ft+WzGt%PlGssFJol+{N44{`VCe^fw^{<l>qI(msSOSB3}6#dE~F1eB;
zz;v`gsrTzgu>Pr47LRDMd`m{K4Nxa79}%Sw>Z`qpO`WO+CoJHFn!<3vaL7|gg#8>P
zFg7$<3Emf)w8T>;bdL}Xx*3^3=ZE@%hsm}J?f+-!eDuEe(fi&8KJ_{yTX6$}P}W~5
z#a~c~Q+e;C#%>eIzJ{7Wk`JkUJ`j5I06BVDO!}4L55z!7evtGN(EUo8@)0O-Jo$`8
zUOuAD;`RVjR~<x|_K?DIPaPtZDNs67V!+1e^9>U$8_^CpurLHb_a(-F5>GYMoXHYE
z_jW+{ekIX6iH?$Bbh3sET<`~?Z$VYPP0g}Z;yEmFylunF0^WMyy-vJlSdJvBi>ONV
z>o!a`u$oB_nIlDU6lKMTAQ9iEz8~fR3a%!D4M&->fq3n}kuwMo>%=JAg}H;L5CbI2
z3$<8$j3s<pOgJd<ze2lX&_;Q{)|aTmpr<U(Tns}V{O>E7a#@SmK^~wTW=6Xa)5(}R
zq*bOIC+twZQv5X#KE>yyJ=MyTqsa6BWA9zyqpYs|?+FPQ5qY9bC0dlBO>H6wK~qZr
zZ3ZSVqlu=9Hny>VNf6pV(qw|zQjI3Gj*~&!!|9=YIW6a3PD@))rLDE-v6^uA4ydSA
zYgE*U@q!kGDrDa8Z$Hm7mk=&id-{Lh&v_uT@3*!0UTd$l*WUY8%6x^mU~9I-3EZKW
z@8h#w^)6z4Adl@NeUi|J<mpvoSpAap0SfJKh)dPOB0d6?RrRiDSE;c+u63KS(h@=3
zs(M|aMaH^Ap~hOU#-LQI>IH?)Rw!9(PFPhtG<dsZt?sf4Rv9W!>CdTw8caT``k-Q0
zk_!KA!l1ItuhS0S=8zzWK~Xv}C~}FxPe_p@26)@V;Csr543&{pXcuTuwQ)POE_B*f
z=s*BU<~8Z;rm!Zk!2}IblOoq4!AjBx1_L-9s+Cy(XQZwF4VEWZPXmdi$YA*oF{!G~
z@G9eOr=&f`Ot2I-l}X99-H;nR5{!kRJBf^KtXV5MQxD<;*3~as)xB!0O<vYG<mEPp
zyxeb-mxTOQ^yQIs`ZDAY`V#$|nM(cW%a(rhrK`xMFZa`?1pN}F$T5xl*t+9m>x5RP
z6P7!ju-a6-@#E}-btdaIHV49p+m5?utr><N8$3Qd^#g;kJYm4w)jJIBGyHy%{%LXu
zv#uCb<k(h?DwS<I2y#@VY#{JGLbox|hI;599`DP;ti4e@!YX{s%4;<+Q1D}h{LM7v
zZ<P@U2yrYYH(?6XBgA^ikg->+E5+Bm&Xf{wlS2N$(I9`91VU~ac34+Fsr=tEDHZrG
z$##hXK4mE2Hr3IVds<ab2&ZY5D_dBDULj86_o%^ljGC>+!GTeAy|UW)o=rVlNdVS$
z3f@IFioRx4qXUaghFI<)!EPm}6$aPR{AF_9YgN5tL$lS1y)y++LhFA9qrWJU_N<+&
zU_w@pE1HuZxL$<sA&2m7v?CgzL5NZ!g8x4t;_HgYON3lr)o`0j_g)s=;|&iW<j}oM
zq90Tns@CaLZuf5qKcMhRs%9eWhrsj!VtCf=iUnjJF`Uh6z(S-g5E||!h2Hn5*v(-1
zt$6JxBd7p<Of=kMMF2Oe#HS3gV&P?{7c(P*^&mNLAt=ITB6cZaC!|~a*c0*<s7&(h
z@LclHazJV6m#UrE{~}MXNo!Cs)YoP~lEjMlv}`X!8@;rK5*|^!O6=5I;j#kyt`zLN
zB?U2W#lD+jyAvd}`fUwik8-FstW_gcpRP9Ll|wR=Q)QGIZx}`VWS3u&Ohj&Y8Ee69
zHmR-FQYu*m)*GEEb=ZjACOBl#4H#7t1XTnV$}cM2X{?VWNoydTv<6(#dKooHl2*KJ
z(mF*IW3rfhAY>+MYT2!PL2!GF=``CRt%j5estHocZ9_>c-jCH9kXPz9#U6#pa(J{V
zhew0eT4!qG(!&0%mO~4rG5+D`;di`dC>$d_OyjryZrB|I%QYb6S{=wOcRFsh%>K;Q
z;UTxqN#5xs-;UY)mA{AOIwqo%!f&Z$8$+sW*;b8G={CLTmJR&=q38|#g%B}Xt01>@
zYQxN5brf>5kM+5V+y=sIcnyS-;CP@4I3E)V>{TF3bc5s{m(TV?$<3}*a`U21)kJPy
zP%G~kwMT6`26FSbqTe>kh@+|3W<zek?KW>#ZxE&gH+s$%tpC0ZO|1)w10R~8%cWpR
z@M)6T+%`&bX(BwKSTEbuYanWK+o&QtB5Y+MqL;jQ3=(pKyzDtp=~KYupGj{Z6a&y3
z2nNj2A@t^ff$5Dy9&K_1dF)SZtST-z{P4skhrR|vJf;b9_S%joMM6B-k>sg@l~^HW
zh8oB&vLZZ--v<hBG~qT~DOlqmet0rAVeFKKyOQ04jtFCuGN$k%`BgnQAYquTofqgt
zH}PS?X1hBuzamYNFyL(y2C3;Lk(*tNRp`_P9?T{TcDoEg$T11*I^bcq-ldC_E8HDf
zobypPCXCp>R$~8p#Qv!aQn#r9+@LYx2JMmjA+~kcUqH%N%O(Sve+cgD!_u1*PcxJ8
z80n34Nt65+`#YzZ2D&u+qj$|m+g>|-hqO8!vfSyA)utgIW}oc{9kR~Jywk~iJ7&*s
z7-JvjpygJ#f5b*q%6LigW2G9d5P%h<DjnOZQL|;6ZZ+I9s!=wVK6cv(-0)+D&3Hf_
z?B0X3C2?Z|YBi9+^;ZROYy*US2z_+6qR5mNT;Y>;LGnO(FF{=%mm5OXouH3Bgor-A
zY*oETW|BlbA!;aYqv1_l;^K_YRzn{jRlzPhW&}{0|H`)8rjK0lZ_~%yMzyMKR^!9e
zyn%am(?yZ&R+xezcn56yxZIBZ@@gvApm)ZN=o0b>qCaSA0#9)eBpg@#dq%Cc<L@K;
z#wS%R#B4T!{IeLDXKbLavty;Qw-87;oGTwttS<1i9kJ7nnB0S!nJ)r~@DqXyfkZ}F
z#qGaI0{Mg(vfNaL%<NQ&A#?Vf=z>3sA=@d|MhuyeR7&EvOKm&?Lsn}TvKLgtO%MUN
z$Aax-v{__iIRM9<FNVx;c%8IG0j&g>LTrW%rtHdBRI<&GnXRQ4Qpls(g);I<n;Dbj
zalcj7N3#CB*z-y9mE81_U#Vtk)h3rEqFOd11gthPC>l4w>9~@&O#^d4u3$6Cg;l|Y
zt?4JHgiN-(!b!vCc~vrgm6$+iv#KDwNT;D;RlSUhG!{&(pDmxGW}E*G9&B+s54PCl
z!7icPB&o&QCbbh)oe$50Ijm%%4T1zOmU|>#?C^|1x2eRDIkIYPdsS=OtNLi$UJfs|
z{FvmC2TwK#;27zl6z7xl5P9VNjOUL5euw8^4-d6g2WrcmE?h0MA2&82)Ydr(JDr5v
zF?-O+`ZyZ$4-z2ulOJvu9yIi-+p(=0wMVw;Lb{$zY%aY@l^>>x#YC@C<%dV+=hh_n
zIh$TN;=^81B@vd_61!($Rxo0k_z<Vz-d-2IlF%+8J{*8vrHK#O-IDlF^vV$*=BRDQ
zz|R@+;T@yAcJwz5z2bs-Tih5j^y+}A3G}KDOz1FlI?qF_R7wQB(sojkUac8bYR5Xd
z_)w3VkzJbjki1^v&oFc^Q*t*zxx*D8z9M>M1c>j6D&#oSBVB;_tmu`I&wVI*)lYyZ
zdWFV@<ber#l`DE>C?xbMm7#NYyhN|wOr=*bR6f{XlUJ!pEBcfa9xB~-d-JS6y+S4n
z*VM?|gVsKbV};p0E3>Mfz_H4ZEx==TxjdSw+T>#EMb@fbRoak{>WxY{TfStcg^t-U
zyCc8bSfN;n4KpYtdy9BOi(XaiN>X8eNV`a~%dgW8e<&f<Yh8{{8vkZ|To&$+q)X71
zA1JO%TK-)*lCyHOpcE?ELX6jKOsD#Ab84Jn+Hy<+r-|$+c<lfi6hg9UrQ!D)u_V|!
zZa+`sB0db$B|WYW!_c*VXa&Gy($*^v|K0M#^Ep|}6gx)x6$Wq#`t=X&9H|0sqM!c}
zUDHY%G@h0l)8Qok9^K4Cx@Mip(Owv9OKUss-YfrMI)O?4OK%q1^sLdbts2!T8}tlq
z5!nW!XQ^nqKRqM<|0njtR;zKxO3>DfTIayx*!IIZo!F^_Uue4je%N-q5=SR!-FCz@
zg7&}rVF?|+BkhNYcnPDUrz7H3X{;Y=H_Q;s9MVA+?@#Q9i4#kgTf${EDtEj~F`;ms
zF7qe9B2#k8eweK@X`yb^j~P4iewdU|oK36h<bGInazCt9oAYVxSXe$TM^-^Sl9ba9
zv6-@SRl~zFWlnXIl+)Hdv<hwbCYZ62M_|SV<O{k@u}5LU9DZZ~<;5Hp;36qrW4bzs
zHeoaC_haJ677RA{;27!Ie5#zFXWG{4GK&jWKIwE11ESmb9~E?m<--O9m`#aXMQ=y-
zp+5(9cz~^Q;N59VZivC`(ZREi(D(l{{MQ}_ukK&X;@uYRF?T*NqP#~9sd`sj(H2-2
z<XEhMx7pN92SgRTLFrg^(_kZ(lV3MW|AcY};y=)|^qyZtCCpBoqr_I@1EftVeIV*u
zZRgVXMj`|I1|Twc7)c>{Jm9Qx*8y8QFRxP(2!o`J$&zQoouKZaf<=5gMinV0vKZ9x
z5r-MI$NP<*Nk@Jw*}*_Oh3xMkN#C9ZFBt`kj!~s{U0*PX9wK|vNJ=U>A(eVqst5P8
z?tf6#=@?b%ROe~h=0+8rdPe$usPoyeX4{z>HBcu%M@;yF5yClQ!j4gmcEr@&yi^or
zlpYoQ2KAJb5DjdB&fbNjayi72O2yl1jH8aBu_;;WE5Eh3vJmuAuZNUP7xmvNpr91z
zbit!2Z7>ah3z67*y3N)KHA2)Uz{8RcqFD$h1}=xaBUVgV5q2dEpREFlT>@PW4pfPg
zhEqYlnvej$LKbL`+w7J(9`+QG3<Fk#MFK;_wnoVue#fXQx(eoo$C^!5MFlh#fCDvx
zLCV3?Z&j1;9j*ZHZ35O9-UN=!Y=Bh~iJZl%fg?#y07{Pzu`%*b8_$pv+WWUTf=ARH
z9j(D?lwQ_2<mV*;s*_AC5sc|MT<CYH-fZ++V{)^tF(EdY#+0+kG{(=i(3pC*g~rro
zR}xyC4eYQt^4pQ!N)A}s4O?Q3xhH$IvK+I*oI(n6l_IC5{F9gKh?CXv`UB=Hmrh!x
zp5<D3B{xyb)p62|M7t|%2FFQ%Tf%#Jm9ejoB6(h`=OUab`v*Of)7?iquYL6Jgz7}f
z`7rJqYkNJ%7k<>I)2e&*j$DMj5iS+8Pk{1_98V?d2qkxmZJp+&4W1C;T&l-$JahV&
zBjWZk)mtVej;V70ALoD3(f7aMu@CNT<M1!1GpFUf*JZU_xpi8$ax#Z-BfO_mn&XRv
z@_E(n{rGlslNnbeyy6eP;}8EW_7b-ynuCm+as*iW3MAbv_HF0dvUez#cd}yNn#7Uv
zO91e_{m#z$h+w}@=)2r!T0AijHYaG!qw!wO#07X2V{)0-S{kPR0+Gx>#C=g?aNzD=
z|H}Tji+xA1CMs52x|oH#Lf%C~b&P-A6j+(uK*^3~%iUJ8k30i~>ij&;qZX^VzP9g-
z<YZ<aL!INL(q>8NZ=D}3@P{YDw|E&?1;I!~>4h~T*9RiwIYT@zQoItOz+D+Us?<kc
zDYiG#m>s`7$2Ykw&sw^kOqG<!gt>R_%axJhAL+!|PrKTt{Foz{Ki$;UH^}N(q*>;D
zh&*{FIlLwOXMgzZ_~m(F?vPf}<9v~pJfE`i*g=(fC7WBy%G$Zf{x-E*bqs_#|4+~6
z7YD*w`M&i~E6xwT6`%vXCRI7dP4i+)=;59rWaXR}G?}b8TqY`#O!QQjm##PY=$LF?
zFx(THKvn}5O<$Q*P6EXptYU9HTrmUJwfCw4b3bC8{xss)>-2r%26=XN@^?mB9l5%>
zvESpIPk)I)d$W+l5^8?ekFdo$KF??lVY36~B?6iW;Jjzb`Odc2CsuHxJ^YZl>fl{7
ziOU$G>ZktGSZAggj|F6xn8-tz$edALl3!#`<gVhBiCm<q!VM+<1)9l2%(!+&fmd_6
ziQICOo0&D7y62iU<^=a##<xRZP6C@~I+ZeHnBgXbNv@C~I?S49A7&j#qxwO`Iqm91
zI8hOpYX5rPp_6>ag}|$tN+-3$ockZ#$<Cdr+yJ5T<OV9`HRcStt~S_hkXt%b>_gc7
zq4Mi<4}l8jl4}P73NHyWcwE2)4Tb}RK!%G3cnvAO-S#UmFu^Yn{<G_sh{v5St{+oE
z$xJMunK+?)CZ$a1yi3$JxL$prA?v)yW$+_M&hwzC?N%kHs6|?>*|uec#nEH=R9mf3
z)b^ovgzAo=c7*B`L+uFF)k9m6;#teB>e`{pSuI$rt?K%rtF3~Nwa%&z4PD2H2dmSn
zUNW?k6@j(gs`d|M6;f_>Th--5yE*${?Xjw74&7rFjI^xkks!<sfv)$xj5%oXi=xR{
zBlsyFk;CuI5ng_)M?jWqM?jYAM^r+uN6fayEE&-VDIU>E3!%*;mK$@$h}FhiH9{~5
zL_QJUQl8Pt3Gz;ylYeJZ=65vXbT;=dv46W7t2^o}ZoA0JkNy4}E;r$Ig?xAX>4u2(
z!MB_!8hm+-vHlds48G?lqz*nqU!D6kAg>Uck7N9ep$zACjq-eZR(_cAS2V!*W70Dx
z)5l*TUds5J_`l=t;~9S=m|gwH-?>Fc9e+7Q(fC`*5EyX$%^H8C@#miWKOn4QQ-zh9
z8d}^vGXVb6YD=g-ohtl#+#%yH5Y77PK=`lank4QM{x1zUgYQ+Y3KddwJ<aID^PM>N
zCnSi$Xj*gl$(X57fDr*G0|MhxD(c>+Yu0-=z8(NE`orz{?X+5}r*J5Lad~8xx1xsk
ziZ9DyWpy{9TH<gkkF_k$b+h<aPHmfV_aJNO@fNqc-*4S~8YYj3I2YY=Tr3!Ao=4hy
zNqaMCOSXlbO`>JlM6Ia#^#Oc?k>3kBIq^DepOs0UwOD=Y!N!oMX@Lu)#U9^1!sC8k
zsb97qTcGB>GA)JeqlmtJ-1fCK(DkY}z@6##yS{;pM+04NdCg_-83)5}?R(N!^1jci
zc;6oljn4>z6vOWZ?%Pih$~Z9k;Dl!?!fyvhqm%O%dkOy{&%!sSMyF5~H<nc2z4t!V
zhj%qu^6qzX_h)Pk@87rGSHfZGiVe{v6EhT^c&WSLlNWwf{O()#BQo;K*6=ic-Mq_v
zksjyHgca`mlsIsH=5Cm4J;UGJ2Qc~T;|y}&gIs^m&2<R-TQ-)5AME8_U**-;MK0;D
zU|ZKe8Z7xs3v9~)f5e7L&qK6<2PTvCt1ik2KiBqhCfD+x#MLuP-)I@?M(!=wRqaXV
z7_O`jS#48($9lQby0bH5YasrJ8@Zo%KPOda`yi|7OPKxE^T~+*jPAB2zdJzY8?8G(
z$k>X5L1ve<AIDyW_^mr%@?~t|`Jn^;lJ{=t;@+EJq%E8N3PvK*BSElv*Hr8FVe11m
z!1b+8Bp$Z3RhI~rs$co^!0R`#FTJ0Ucn2QV&36pbpw)ziUz<`I^PjEhvze!X46_`F
zw&yE*xs4ee#*}`6t5$jf<e`j3JKJ968K`Ic-15^&uH0n4luX`aK@tk_&`yS9zYsfj
zpBsN3Jbfed3Dxwk)(vG|{p9F}72;)#v6doT)^xHHK3(C{70$a;4*-Otg#vWzrr%N(
zWnjjevKy{k(zD_U^|q}=Axj~Z`YF{<ktqfLlW~2wm3-vesGr$xP@aM`b7ih}Of5T<
zp(2I*T5jnrasWMGceucv@WzDuUDyS}*%La;fX!+ASU2Z$f!~GEtT)IP$k0D)#;AWd
z$Z9<71Ds5F-5Y=E8N|_*@#1#N3%O6N-E{I&27@yoBx})t@M7w-+Uv0`?Yz#4eQ*}{
zTXEa>mY!c6Po$I~F~JmRhJ;DakjOLrew%Hk4+&+R91{PR4256NRs)2DkmH{;9!8si
z(7zir5Cl>UgbB~k@r;5{3FdX)SPdQ7d#bJ%V?4B<kS*xG6<hE*E;sDsuIjYG@jc>d
zY09<MrU?5g$&8VtamPh537v8Ajm{zlQU{>XP;0->wv>ls*CD}a&lkrKd%1wpKPfRq
zBULDm+-AIc^Z<DoL?b$OGqI?V9bGlcqz<aBl1zAyE4TyU*J58g6PoJ|!aYjCAoS2;
zeWWVqF^LB#DiG-~1+cbwFu;Qi^MZQD!d6$1sbxtZI&xbeyvii!)e(pt1swb=_56H!
z*sh<drt%M4q<zKNyfPlp<=pHllW*HShW-!Y9Y3&bX8<UP3<n~My@)<orSJeXS4l;_
zi5%)R>Z!R;y^%=KScG0>OXFqXH(d}}%T8o7MS_EQdf1O;)jV^pAC^jgIOf0kg~#^O
zAW1y-ZO0<aQ5G$W2jiXP#9>LT=T-W+0irLk;|)Ew60YY+T(m5MP_&b$<tfhsJfOo|
z4AIS_AF!#AGBw&4%_=5SRq;pr?0fG0CGT4uGZ9X^D+M~e9HC65wX9WBc71~WM@xs4
zkM1<92Z#7Gc!oPKyJ(bnLZ1pmYzb;0GBgmml)CVE#bw#|N#bI^@PeT0hD4<3um^O#
z8T@SS`w|4@1q%E6(ZfKbBoLWTM&F7B!e8GLh+J2MJAiwCAbb{=V9e(73$N&3%j0^i
zu!I*swozL2n;F13p9f;!;YpXmJ?2T5FY%;H3l+P=pvhidcp2n6LWtmUgMmmct31Y7
zysN*HbyTNXksgcdxtglgo3x>LTOgW677hfWQ#Sa+ZxlXk8kEE@T1!ulA)BL3(`nBP
zc(Uo-LN|jp4WT`}%)N#7Y@N+JF)cfmGuljV`S!)bEArP+1A{5Ht18TuZcKd@kg==u
zEK67Hd~4~y0B?89Yk@p3wv>4g6#1^suOhAjm4(cwFfWQ$zsr1EW%3K1@5tq<)H(BF
zztkKmG)dh|6Lq*SI`UnW@1`xA0FGK^@>IdxvWBuU;Q(c|j1JQTKdsW5x*Ye-46@jH
zC#f#A`K5`cP1bx`v))D@D6TBaC7R~Y9+FAI*T{%A?^}nvSzTzljAo0WY-?t6bZdni
z-lcJ_6?%9FFUt#x8(Hm?Jzru2(^{H`aL*s^{)|1<j^D`ZKi@9nQ5JI5G!I5@P;X>A
z3&!vR0q!%?0Mq?uT0O>2J1ZG0Py>?Xf*IP?Nmw9K+(pY3oyV_vi`!1(Cs^n;c1fWE
z$~+f^{V#(z_w}B-{E!^7x=d&=&LVF!sKy3~kB7E9JS?`$!UwpF?jA$X4S+OhedMW|
z0cX3bQC$G2)?8nb5u4nm4zZ#W86<&~T0FoN3=U>~QGF%wf;QjFg+th^Y$o4kvR<;4
zI%{Q^vCS7&<}N~gRd?1gP(XC$OoNH)n<B|$OAWq>Zz?M}V0E;Azz`bYj-J>r=-i&K
za)*4+Zn6vjYSXRee!5?K>3CM(*<#jyf&+XMnvf)CBTG&9yZ7ZJU-52ILV>7lW2}V<
zwj{sN77||RWHL5AliQpwa$f;Ir(b$|&1z?Hio$N(Bz`bumqz4Ft?%WHaL0BUw!|3j
z9;Gp?2$?2AYBM(x=0Y*i+}rW&*W9r**HA3yJ|FJ0wX4*o%-c57+|99;o|NW(-N|HZ
zu<PZ_i=y=iVCnr%0Im6wNI$J#`b2ZB_NR&4^E^NI{o!&~kX%I5?+VOBVtN({@7s_F
zn4SnYNdb+Qxhd48c8dLH5b*OvinkL1&nsX_BE>t2fGrAGnh5x7BH+IW5G)z-op^e9
z+QnfmVoe0~j*MMISM^Ng&S|3co*bJZ=gz%$D!4hsR{LV#Q+Ki^`65Z=z81G~T?)~E
z-fI8VdpwNv8Nzy2V<VvF(mt(I#Gj_f&w0O!8o+b<p|E4^%`j1XZ-84vE7=+fOrF8k
zkT(nbQ1eJ2Qd!wH<rZ#pD!>C*P9*mhmNHYIN!GHn@nfN2-F&_sIL}`)<rZsMz_^yL
zqa1%E{}w+ea#$d;T>M=i(kjL)GWKTc_OZ8=*O=|LLO<`YAVg^tY$n{Vv2+#-H@LXL
z5U$cd^q%}yLTq{cNv9_yGxol@5!qkli0lU=vY)V%9OOG&i-<<ES~)_7og>L`r@6@2
zeuhU&wbF7HC^>CQ^2a7uP^@+OEuXw*`GmA0lwZ#79daECEGk3JzU&h4l(2Cp6<0Qy
z*1!7mPhg2=^KT#>mOsOWgtvj+5BzA~lKd&Y@DtSAYCoQows&~D)@V6!sK7xtNA?W%
z&woK1bhnh}pWu7!faZ@^V$Io*yz+3DpKQm9JPCKlUL_+DdAru42Bw;?=?>AV@~A(*
zI+mMHU(6IHbfQIO6qIbR+V{{$%B<Q~g2?0pTxH*Qh&9Vr-YV0Ahia9^W_(6C`-@yJ
z?Va!*eRgIz*4CZTGHOC+Z+*Y@P`J+zavj!FpdnDRH4r($5FB+d#g|$IQ@I{XcVrbz
zEtLhIK~}+qIaa}?UaR0M+zMi@2AP@RwF-6nSz%TN*R`)FH&1w<$r`DKb(4-z04YcV
z5X+>F(aMZ)ANxlcTaYUWiA0MR1-f40LN1>Q`~n5~RNxmVFvBVcQXuav7x3P4Vf7dI
znfV13_yraC1q#gA)jyjXHOQTuZd~oy$;>yK+#0Qd@3+dY;77|%&>?w77xS81=OD6n
z#Q1F+29+GNmX^nnUcKoHf58{uQdaVc)$t%?N4;HYqtVw-vz^X<(Q|V0{1}C#AbRmm
z482lMh{Ypj>`q0)C(rT8i>;9l6l*MV;J#%E%=I7{-F4ihpv}LYO2E2AfR+w+z@HWH
zUScb6O(LL60do>5w1}}&tW`h~(f(H=;8q1B@u^Nf+9?((Ac;?JPXt^=Ko59;t-*FY
z`{B3sc$72!v&%q^m^p!PF+AW0S_>9v5Y)UT_QMMw#J*k5(oVCh%Q9>Wvv*{9WDNS?
zLQ-v0*0O;;o0V2VgPqKfO06}yIu}cu`Ah_1H?ZAWX1~v{71pxjNowO?!X-(Fj9jak
zHqpABv*(TwV>Pwanpgu#Uq$2(YCy%9KgcK8_2!^5R>?U*#xHnqHGZ*sXN%yzcYWd1
z=&;Gb$bTuK2#Nv~d9**gM>Fx%3GeluW#b!~UgSpUy?QzLW#4x^;SsiJUVaRB2|h(W
zw$7V;zCuc?{v$%P`afhbe@wgs;mZy<mM)Y06LxZz{*2;{+0gWCU)z?xa97KtZ0HAI
zMwi|%rmVnfmmVT7I0o8Y<R*Z(I@LmnFi^9*5gSjV!}IMAM5}mxvp18+GDD}ZBNJ$Q
zB}0*v;!jE;nODPc{9Tgsr);Hb`(RKL^dWS6QJ|0qMHBoAQLfm|&)2&7nY)LdaR^tD
z!S+4KpX9F(Q|nUZ-kuiJBDUi(kmMDy*TaFr%`G|M$9lgBN=xJqGs|OiXhm!Z9j9i#
z5n{Q0Fe7xb*u~J0%-rw6#-{#;9}BegWi^e}bJEe!_lAWZ6NXIWm-)55nP1zV`LzR?
z+-<)_w<I2D>1}%{!=CLIpMsQwp=tf>DXZ;Y)ou+Uzjb?eS;m224bN(zs~0H@L|J8a
zy#z{?K2$UdE}y-#jBQ#jLq*ws3}#+|nu@Rpy^<ZfldPn5WFsg4>0m?qSS1MUbXLL*
zs7CQ?PUJSsG{<kvqV9(3(~IDxW}fclxBT=Re*LHGZt#l8IA<nkwd*z0q2yJo<Dj_-
zx;%d(3@Q^6wGT7FCJ$R{b#y80#!VU#{cx8vqHP#l_(>a6-84e`9hEFo+V&r3eeW`+
zSUi7ocb1Rg)Od?p!b!(Y`{d&ed9Aj+bO8;b<i=YxS%{)babB>hPPVGP!u@AggXTwA
zg|k5P*U+fO3uilrCzZv8x<W0HzE$aa=0aOjX;k%128h+BIIH>xD#a>1&MLfwEWhl9
zt`$MgN}*?!R@KzmR@H@#R@D!bSI-S(<rbMz@c~LbTCLsY>(uAr+uqZux~0(9^!gmO
z)^-c4>=ss;7WQmpS_`Nj7VZT@t%9#ZY@56m3eVgOSH0INw5-AgpmpuBbO0?xXZk{b
zdNn}(dZo$cC966H#Ea;}Qu3<o(egj7r319#aP6(8m#XQdYP*-jU3|D+I%*qTs2#bx
zOiQaSEn=XSGIS~#PBd^94V+QgdueLx(rbL!b`3aEXOQP5<oV^wKzRRGHr{76`5uRg
z-#v+j-}M%@*sCXr;I`uSgfe_GJTE(D(IDIXACCFDV@5J<|6PtbpLwb;jUN%^6ur*z
zOVlI%8`jgv?%Bbek{_t~v*?mn#QHA@Vr`XehE|WY-=lSi^-75Ke`KW*>pY0{N|U9C
zwY_hdO0OeXE0@}o#aiYkX+r;-xLQ~^WZNH^vQi=pt*^kkhJqPc(jGigWxq$J-HII>
zbfrzZ&t~;MKFAx?N%F05k?&7C<oh(p_ZWNqAkwYi-mlwkoACCd--dKg??<}r_0fl>
z-w13IB-}C&bWzuV33w~@`bY#E5spK^6ZG)or{6TFcVVJ`Q|R|FL%&-FD+2E~u~%xW
zpG>9XgbhH<M<t2*k?Hrc;f8*XHT3&FM!P6<UerI><L4!h6zV)v4{h3)u!=(G#B%=z
z)q{$|cKNacV=^FmIo4A1iX7uHW+1PpRd94Wjz&42avbyI*zP!H%dyRI!16iUZcHRN
zM0d->&!eU|);ZBwDYWOWbsQ7rSnW89g5gKY!!MOb)|mB!q(UndL=-bm5It(_F<<uh
zGTyWw=>3CbZ=A5n$lS65k(0En?RtfECVW|R$~l2BWdySW8C)(MK<JTwj-INB$XaL~
z%AP4@pf3TA0Dn9&Vmpkz6nl;v5U?F40F?HnJU5`qc9?+K*u8GRJlkOc8nG9+0gH`e
zX{Tg+TVorB!FFtR5RWa<06T$5cr&|WGXg_NQ3U`kb!=N=wX@RdXj6<EH!0zvs~yd7
zd$m(dDdNjkFmmp@D1dP4RRmG?BRxO30$lu6A|#iO==)xAS@;R`{){^0^z7gmt&bQg
zzb8u@_9bLx)L&r09}Y%N2}I5@Ti+5aM5nMJ0^~2v#){6`VxS<e$rw1ljdRIxnvW84
zWXVujfaj9oIaq+-lA*EyuO-9tuoTHh7BXBi+>3=lv}C9(0Y>%kf<U;L0ard0%SBZI
z06Pr8@+$jPJ#{C<<Faf_X|u$)ckQS3v1<rD<o%&aTi=Iv>*c2QZ?l*R{>jy0X7Jef
zQmT*bmnsa^=g9T6;2`VzK-a;npmqBjfx@jo#;9)wU0oitVZ_eR`YSFs-{1fp-1avr
z6MM^aPFFC@<L{6wU}md}qDHVS$ZX1~hzzNST%28@U7e>4aobj2*a@?-J-F}XK-X)7
z_4)<EowjcD40f`IgYZc=tZ%j2HD%~IbPVJUU5*_c17klEZYp%D&&uKGKu&-I4>^@y
z>m<Y#Cn2^tDI*wNj5gahjV{3{R?cx&4olA*mYyeF!QaJ1VMuH(g1tDVC~fbfb8wK_
zl*6?ZzCYoO-pR3_(*ZpP5i0h2G!?eu>^<J*FA?bmOP-}EwxIW|*pH7VFcy1B#+c;+
z_INU4p^<1Fj<<5k$5XSJ<Ehrnar~5zqjFi}fKO5A3@gV!zDQ9@6{S*9_$eP(YMtaC
zN9~D|uQ;<6r_st;F}{^bS|_a-XPf2Y8WnRkfZHA-qxSL36`s>Eezh@Ij2CKCeB108
zXZ%lUAE$V{YSbyC<TYz)fQcS#>)KvX*i|0h=Js4P3q?+8mvH3Q^nS2yQ(iFssI_#Z
zL^QldmOigzbB5Ye`F>+=jlG@J&qrbR#@AaZ&li5g7iJ^yuc++`f|28RVGX7AcqQ8E
z2@ju<clpCFN{`(rtA9q!T$S%Gm>~LzJyTTNUnmRjiAl4XNtZvbw_H(*e0MPn)1zGJ
zL~$faKhCC_M9uc;vFr4x<D95@V6Il#+44A1&D8OPp4p^y9lXY)2&wD4w$}rGW=XLx
z{16v92la^c65;J7(7MP~3?zgy6*YS$5{yhk!K`>)AkwD3L5FM|vuT(wV~;=F$w>gj
zviXq~7KR9SpO1fu<>Jdeo^(g|mrb0nR!)m#{hs(<qg{p+@HqC`{3UN&%XZNAK+XNW
z@N3~-qfqufys$TDW;lHuFVA>@N$I^;_KaQOclPb{mFz_AYiD%kIjDVY2qrg^!@J5y
zZ<+8$a5NGFDU<1lU7+yoXz_6f27}T1-|scK1mauJCEHh?@kpTSO>f4Q@V<nST6D=5
z`m2(~i?#AZ%#lRsaNVvxS%st*W#T=Ks!+DMn;E{pWrLP7M^q?de76imi!1zGkI-xM
zOYvlz{X;XtTiRa91TQ15-p#8%S>fk;zsUKD@N=QhA{`3l_}ZpKP~+NQ-MKU4$w2)6
zqq@->W5>Ci5}omplER<yqOWa9L>gm89c{Z@QW-PZZ;(qmW1VO*bo(+MmuRG<=Z1S(
z6b(a7aoK-Ue?*$a64RsSzyKJDkX>URf^?j=HyAQ#IFxf0h-4_?UZ##ejF4(-G;4x6
zU$UL_-a$()(jvJp{C44vyERz~H~2~(vC6lDpeOjlC@GNTM4yK5o94iVpDc@GpCBrd
z|K380bCUgMB8T5ikGmBWif|7ST}Ea2Icw>c8E@Jqn;4nKAg<5P3726aS1M!fHZwb<
z7171;3UvJ~D|m0ovX>_6f4O!0+bFZ5SYG&!Mj?x-GL9>id8sthX-jCd=5eXLTk=w7
zVz&fwcrx1xNOncuMbTWAUoS$qkaS({4U}x*VZOoAg%1?&3EX|FUEoWBeJ?UWc>geK
zr;YnM%po!5pu<sZn*u2~&(MPu{;uuysxlEUozH+`g5+#|4g^3C8DZ=|a@bsRkm6=r
z4*N@3yBS%uE^_lWHIkKIwsm{%FSSzx$aK0Xa(o~XwgYqX?XcW@feC(c?s5`p$<;RH
zmn4Q{haU_?hUGIj?h(+eWsiVC)B4Zgs7R|<-5VLZRtxrm^2kKNThcB-$wHx0i7O(1
zl<cZv%pZlcTzTz@MqVqS`$-q^*k?88hqUHrH5`-VcKozx`Da;-{8VW)SW3nC6*`RH
zXc)huL&j4nyDF}{TSIlS&Hhbxq_>9qyAUqTFo5-v-B$k|*=;K&82RnD4}>o_@>?v(
zZzqQ@H{O$N@$F{Bw;NsY?RGE-N4{+duJexB;n=?$)q?-~<+s9#|1|#3WB9*Z_&>O!
z<p7LX&wv1-SZ$JwgIvuBgXLh_L&MO|V(~8dEV@1UzG!Z)KjSqcwDm=ci~SjI25NR{
zD=J!S*e8fXR?%;f@-jbP1oPjJbrN3wqMYAqu%3i~7Khh%c#WgTb|A(jq||nl8b_t=
zs5Fk*wqv$&G}?|v9PLG|)J^X8lVs~{)NoDWzp2sO@v5RzBUVc%d3_nr1#6y2diis_
z6Lvb|6Do5Z*_L)+OqRZGS@#(09;_bOGW1|vmstR|z0dnWdMeX~6d*E_JH9o((LXu=
zmsZCGM%>J4AT6^oh#)*NIhU|^#DKMJs!G9Fwy>+;vDNqc`i%0S=BCY)Z8JX4$2=MF
z_<;Bglqzn(MsP4UTISdY?(n-!{juB`AR?;kVmWr3Y%IsHKyIFj<x)|gjpvGz@OMBW
zeCKXPtu<oigg!<ZYAX_65;!N8M;H<V;JBaLIPT{*j(eRwsof!P++IU!VrNg6P;2+~
z1)#O_L2JKBL2J7u&+688RGK^suo_{O7#<rx9kL(gYdbKg33(QUB)6hc;vd_7)Fx*?
z>S7n06)+SAl4OOx75FG<ti+XKNnhpTmtXzU_Mgm7PH1rK{tx=xO(-o|3Yw>Bz+`Km
z$)6E3Dp{SVWM!FQ6v$j`5SiJAa)OHqM$(hnoMWDqjb$^R^q#B^OqvPopFD*LI>+pE
zBj>RvZxMls$@^KG#Y&yL+T>2o(R?Z?xk)u#YH~KDypOE|;D*oYLv6-t&tfcaW}Y{6
zF3yoOuj_AvgZCt)2iENm1xlW_+Ey__gX^7h(Svv@-YmZ0BHka2_~7w!wwVie1xo&6
z-L!$2fs#LSG_~BC`av+`N$Uq2DoXZSKj_NdQDK$sN6`>5MQE)3qO_ykk~mX--ur!V
zV(`GQ<*!Eq9=Sku3zTdPi8)>!DBK_1$J4N{aahT4p@T1s`U__Fj)2+GF$U?r1jEy_
zdndr1OMfOh5^wecGd9{7>`wBp2#*Vfzm|PbH0uWG*OX_(NRk6aH0RB)5=*!7z9*+6
zZVCLYFR{O#d_Q|HENVsc+H6~3G9x><bw-|Ma4<Tn013*B0z4+lLmEmi8}Kqpf@8As
zQ5YT+49_TKVZnLtFSBC{+l4Btf!-pdP0z<(_Q7&TW|a00ZAPjx?8adDk|N*M<H*J1
z+d9LmY$Y!$3`T~hzen0O<@*^eX9ObWZq53kiN)Y4+0dl5so}>>tM+^uIohBq<Y=6D
z^4wL3E%pf-8AR#Y7F!ICy|f=U9c+~PcH!VKU~D*}DxJO^{r-Uu7r)Y`L8fwERjyH$
zTSWP`{3<mZ?Mu&HoeIBV!0@vbUd&AlvGWq)AGW?*1#)1nUk~rgPv|r`B;@lOlas68
zdcV@lu7l+zy;gf4iOYFDKJxAC3SOiy<YjimUk8~%-(l9lv=p-4k9@0S^P)@bU39}w
zbj2>PmOieuJZ^py9mE{3DA^~a?dLfdtmk_lQn!(d-B088sS}YUM6-S!4436pMC-D9
z7YC!r3L>yn;*{PGgm*BzDXXV>wC;W=*&mufX0fU2W4n${wX%CZB^u72{y9!*XAnov
z&3~4?F*96IWPjyAdB%Ie!hIpo36t&p*x#sxG$x-7biF(%n7M!_h{vLnl^B9%YF)K4
zHiyj8?`N0?qZ%lgksZnkAkMf;Vh9wgy;?w^lUf-lXAKj86|dmdiqVhwj3$KnaY%dn
zmGl-`)P!?{VX3-ILmExNGE?!*I=Jwb&K3OUy8`B0v9pc;Ci%ZA|0w>bnM&LT5IL7K
zj0l{Ys?bTio#o_<K%i{t&WdoKj=&&g?xDQULUbCCR2utq?963jKSRJf$>mMiAuG!<
zmEy=+APUcj95crW%?NDunZQ7N8|%9xhar<}x<(0?D~n+;zG=zBsF!TW3!R9+k8}UN
z0@EDj>-rYqTY89Z0N*O(OGMQpyrbnOAjn1K(IuN%D%49GuvUi{>hmjsg4O;5vr0pS
zk99h0^;Ru+CZx)5BDpvF!ZV{DMShS2Nbv{4yLP;JJi?`}*9IN`PIy<>n_0Fi+jiv~
zzbn4Q7cTBVhu>eaqv>9$L6W$DlFHc*&C%Hq@ww8~r&^3^qxv^f6+|_ICDLX_BKL0~
z{ea#@vO411{MOI*SZ${=Is%aqfyti;0p6XVLE&?N2xDzT&jIRf%Wo;7N|bwcAi~OZ
zd!P{QOQGvcw&2k6zYxsWWc{EsV1_3n^ox>8qnQR}>ngX(UU8px51kV*0ysqpqVD~@
zUs49*_6ws^{!=iH{nm^@i1y{w$|<n$wIHHFFoP`>Kfu%HA)~!g70#AkGRHBV=cHZi
zU-$OKFOJs#IxBGi=-j_DP)#;rmid6BJUpI}{4*_4Kqi^x$W<z(*v;Bm{fOisAU=x5
z0!={9|F*t7tK$KQwew8O*u50fBk^y7-?n?jx@j@7=ocw9<yh?zK{g5=L%5N*UStpp
z^Ol0N%jXGqJ@zWygx~<3$q8?y@JIB|ZSieO5%`}$#9-u#{lUqngicNAo?!Tj{XKCq
zN%oH!gu%$7J*I;=G@-45ji!SFmj7wUe7_EQG*I%q>7W2d2Uzde(vHyQlRXqD*=T+5
zbT|WpW0?B?PKln_i2#7US6Ql?ftrb7rT}uM^zJIp=uYddlMPxyD>b&8>j(ty*_TA?
zZ==`V(<a)!*mvI6LS!x&YvKN2G;38bJS5ps$v&b#=^{=G1>3gd`AeR#ZrTL=>1}Vk
zvu!^U&-AFeqN9xfBObq7fTxShI7PdRw!tzF4UB#Sq3nJ1DFyEcw!Ltic&2yC3m<41
zj<9v`w?|0^+pw*-B8YY@%hoBAIPhCu_^1NE<^(2*9=<#Hdx~|wPx9I0=UNj>c}{p@
zDLAOhYs9c<WfNy&X;QIlVyPZ_??R`|_+HN%<&l=-!rP4yR+5`&Q$}=tMtEQ0TO2_x
zd^2?Vgic;P<OIUz-V0FZPLAWAQOFWGifZpw=W!4<l<k986#5{xYjw@=b}65IDHsiP
z4D*MdEPND=H+>0uZrwh8FMs>_J7C?;Mv*ztlhd~KK+EplQ|(!?mQL$EQ3~gRd(Xck
zk8^=-wF}7@=pw6_6d3(JVsim_S9#$hHh@p#IHv({c`#a$<Comn1#kRxJ1C_-#|YoI
zj|5+zb$w@`<bjsKepKqe<%?dL!Raa*zfZ7zCo}||<#?z7AcjT=e#e(txBH_W0C=c{
z97KCQAB@h7+5p5S9ez!~uL<}y0YC7o;J3ksU-%tiXfLQ?wwb}A<kIpe=c0_RR|aXF
z{O^0rd^7Zz`vGDH-;n#y$fak4BtyKO+WtKTa$r4zjW`s6bCScB2S#_9av%#;Li(xo
zF|=bJ5p5C=)~lkp*nisj+8ap|I+Oj+V!IUd`EWLP^I>*0I8G2IHFpdJy5fU^(c*R;
zeEyT>R=}F^me4P&v7SKN3mNuCX-?z>pY^jxnloF6XV@4CMaX8h?H}%)f&OS@Y=q!9
zn~yhHJofVO;{puy$jnF<-n*Jlh<BNx^w^8XpQ6KoB?rm8=>es-_ZL+IUu5P|2<e3T
zOHR2q(~n?0-svyd?h9|TmYxq-mfr6`>O5*`juM4mD`TW)Nw#6#J~hh+#vf+geqpu`
z&2YrW+sZO|<uBvmK;(+-z~n1(LZ=D+k~8P?L>{lmHaMy06*W3?-rtLK*OS~68ru)&
z{)aMk(d|#(6qZC+GZ2{&h@jn2$dGwLD`l-(+g^YkcBx{8f;*Q)Z8pm}^lD@6bg0&t
zj1iUw^=pM<O|8*HWxzqo_r-Evw@GAyT|}4_+kCjItd&r34)MXKyM$3&6O%#K)=l@o
zXmj$3kjmIP%IK*C!a7o7YoEfDqF^BUGZ|!TEj<O0vj7Ou->=42_fn^lgCcqpLn8y>
zUiv@5U$qga^FP6o=UZN5DQYj^7vQToLHd2M*~alk=oGcGZPUD@q^_sTq)zqPXwV|n
zOcXzwI0j)+2MT8kKm6SC<aw(>CX3Bf;4RJ{v8*th-s`Wa0!{;-Hx5poQ-Rn{9IDk&
zL1FRxsAF{KlDi-CbN46d#Jlbir}YrAR1+e;0JyRC{z*T-zVDSp?CM19pV+Z`E(fSK
zf0rUpbkFCM=g-roNh8yHrZ4=YpM?av4zS;SX6fwX_JH8l>F&aso<?&g7@b}!ODUGB
ziki1LPr^IRH%!-S%{NRBnBR%!7dCzRJSxPQ`Egpzl8AFFoaFh^3BoR8rJaxOTd_f=
zIn%2+GPxGP!Mwy7<tuyC&aWXFW(!K$HwL$Uw^hmblyV!84n{9-)Vg&xLWV{=i=VU-
zp2+1B$|d@noAIM~@>}fY_g(ps<w^{FR{LY1KqC3WA6PfvJ5+_#=P!0o$Cbmjh*!T<
zrH1Xq(Phg?Xx7bm$!;Gkiw;BRu}m44hhIY_J=2_hWA>MadqUR)INBu3R4icUp3v!%
zIk*f9+)Swh^a@6DCrj!90@)NjT3E2{U{=$7wN~?kOvKM5a#sfMtuE)nY3;404nITA
z4nI>0KT}#tGoV^2zVI`pRmRuwGi0>A<tM?2fusY61dvw~4<N5obUr~a$BfSJuEt61
zwHQzb0Zfs6;pc-jPeRzm5ei>s3**>e>vO@@_YVXcoq~vwj&E9E!>*ND{m;Np?eA&h
zCo!tidQ4O3kcq%FTdWE?V_f0=2X-)OhNi*SreD%;5HU$sD~n_|Nt+cA1f$E=6JREe
z>8M1v%A5GxF!WjNSAe03cipM67hYu<2X#<-Bo~0Ppy%2l_B3IZr(Di)T+B=*lUH<E
zC;4b9c^MS4Ulc80ppmhSk)d+_*;8quJsj-)f&Y>Dq*nKiVhkju5L%1TpPhuLKfMz=
z9cYz>uOA7=;}wr;7|_Ppw$~-+(e@IQa$f8%oOIcClIjyF+4Y9F&A7E}A!}_!!QMbk
z>wdxls3*5#A}e<G&&+Y`Xf+)YnfVlpxM5GZ(p5Gb6%_&XqSIJ#JQcffETsLjS}mrm
zx0`cOVpGBucI@b|?+3z2&kPngQl6)nJkM61qmP=O7<Cx$BUbX0Cc|gPkfArZJ{`Dz
z%CC|9F0RR6EBid`(<8ZS(eMM%v+N}oE1k><@A*r*t!1w;2?8rok=nNdRVkwcWKL>9
zhEW1So}Z=!RI<aW*x`?spTnI}FjbEP?n>M#G5V1TG{Q!2<Nnb~4Ld0$EVkN3H2tjg
zCkDeiQK0bsw=$j@E&i0*stx^iFB(hxo(M)S%RoEmF{ea-#-GEF?R%``IjiEiX!R#h
z4&v0_OOB4Y&OSAK_Afa`J0S(WDg4pJ;y;JLS0>eiIN$UPe{;Dl`WgQEU`+eqOShmN
zw9(u*wbM_9C-$Nt)k*a!W5Cl-k5rU&S?#)aC>Y*>+S$IH(RuNGPoNuD@tPm)<qUs#
z1H`io+BIatGthz(C}-$1P)*%0b)&1ovkQ^hk)7ecO8tMa4<(`SJK?`H`Ig)9xk2i$
z8N1q++$iNBqYm^A4(SFZLZljGhd^(+OsYG)>xK=|_b`$r&(v`7TAIoN>I%d|h?FkL
z;hMAmV7I+A@A|LkTG(AipN{4|!PO@l?U6&g(mY>ek#jpNF~U8B_LOXFS>#>;yU!nf
zlYrO;R!=?RfzswvTF#Hf;Zn99HtVO!`e|Ru)|LzV=RJf4Pw#B)C)ykvIKeJ8{9x=t
z%Jr=$f-r^S`*f@r-`;V^|BNgo8S-FE{MDiH*WeqBZw6YGz7jMIAqrN<?Y55c&E%b2
zd$Z7Hm&DB90x=Ryuu}si&l&m=Ivepm;mP@$TyOv2c`f)i=LJf7TY7?#@fAp%;bIU2
zy%k$T7@NhbDJU5E$yy9@42?GYf-Bp<G(EkEoXFtw3Zzntky;vg6?*{WFfwJme{y>I
zvH6L_`~PAmo*jrxw7Gy+9G{|u&*P3yX~JiN;}b~utiva<W%M%~T0}k5P3EGpJrY^P
z>T&k`r#a8Rwf8rGs`uYwRq~touK230J<VVW2g2OdCz{N~LrM8#3V+b{I_hiw{9Jef
z$%S>UHS%q5ux1~71|YzH(<&u0b{%K$ewWcBJ)whOq+#Yvy2QQ2szk1ki@sb}+mv+~
zsMYmY%hcDts6RhW&M2O~*ZawI`T%FBQkX)KW{Q7btaMp8CjC@fkB&p^;U_3f<`&yz
z*S~PPkp&dK?JL=0l|K@e`q%SNJ5)E{M~T(P%|Yap8<PamU&5iEmLK+hr7RlCkE3sV
zQVPC9CNOaE88K7j_N`g@Gy%%!W{;v>xYN#EvRGfqn=N<4P3EPH=#>1^;ae8o1k7>(
zZ>|)@bz3y2ka!xZZOTe+J&;nEfbHj+YJbMtrnOvCy`xOmCXOT!T`ue9u_o57>JH8i
z{U{K*)jlooZ`#d{Wc|pxJ?qC>BXDamAV3^m70$XvDi3rFozf*(x8zCH@P5GT3vV_n
z0rB5N;x+*ykoYk+q%bBt)33jx<)=%#%eSwOo07g$WUTRU5=?K=3%Ov%98gcVvn>2V
z>__C>dycc7@JFIrh?pYS;m-&!HO_EKva)b@>|NF+@r^!s`DOiJYajGBT=JjE=&r$E
zflL1RGkwlMZxirrA|R>&bI{uaJeLRv5g`1R)<5&4PweNgn7O>-U2{-On<45Wb4aX$
z#(9Ggvn;KMEHeNf2EgI>*;Yp$^tbZxl>DWgzR1`gakvSbsrw>0XU67$wd^(A!5TKr
z*dIq3Q>!?pG7@~2L$(s(ax?}XV8|Nq=$tK)CI%u|`PjpM(;~UZ)^A3{f)IvfnidIC
z#p0w5dV{lRA9{oLYGqw9=3WhvVAosd)Sy<aiH!PYW$bBgo5AOLqc~WERC=(Gq~dDP
z$UanD`hxqVSTnG`U}Z1drK9gd{~xL^7$4L$!&gXguCibfe=Gj6=?iX^dJUJHqF(a`
zyK+{0p~k(o;`*KR4t@J0sn<+Os@M2+OX0_LzWc#I;YQAPKha;m=B4EMZrb&X&Ub%c
z&Ub$>@cHg1`kn87B5gg8xo~gh!u^>G4`iaru?2MtbH4j!=Y03Vr;KV1+jX(#%QUDa
zqMMA4gnfKSi_3xwqq%2^Pw<+PQjuX3!dc%Bx~H)sSw&m3mIHT2h;xn$FOsjuE6j80
zI<RD>gc-mh-vR$*u3+h?m7MQ5)=Nm2av0x2ZA=`^7!?bV!#{Hh^px=*-!i%Q`&P#u
zgwQ>k$yt-;DQnr?SQ*Tk8WlC%zBKZCWfOJ=nLA|@o^i+D(r1ubUVq(12x)?mRpxwD
z5Y9yPQ!_OYR9^Pf=-873k=%Ur2Tzv%VAqRWbvCRdkg<bHsV)d`|3+r+<O#a+Pf3&G
zHClbXCp_}*ty#7*Y}Q@uE=-GzyuIZ4CA+6Z#{QBvscF+QbARa*!s<n6YIV$o#HLFa
zH#+J>mG*+*iCv=N&&N)D+ExHb?3WyhkH(59m)aO9g=nxN3lI%fBBSz1i#LInekdq|
zjHcYB6A{@$N4a-f?LMKSof4Za8!EOBOeC#q8xw&_bb|~N47J7tK3Wex`g0N=4GD94
zPTGQIePD1<5t|MS$VC1B4tx|mBLyE#3`U1#rRez?baa-3jy}!$;oNk3WKdQw5}Oo!
zG=RPz7%3PGKV^_oQss~NE`0QzW7GFD2#J%4tmP9p>GMFEd<Q~HR(mTW=_8@0f9UyD
z7eCRUH2viSei|mc`;0aMH|qTA2WC%jBez^~H->u?1z~CDS8Zf9Dsg`Gi6i5uM;<fy
zsZ{vsOS-htpIM(jHV}zu*{FG21bX^(8hY}F$9|s(t4)Mq6Sii39~*Ha_Eb;!u&&YA
z2yjtcd8C=iVK5YGyz8xwFAF&hP3(Ukf}dDNW)`nk`6&8M&b`a(cv+}PnnTV6x*w~P
zWkt=ykzwmWV`DWfN_mnb*!6-?lxD_qog`UGWZupR1|q*#yMpNYME-j<HZ2VzKUT%r
z_IA(|o189s8hJt><FP<wtQE+30`xSzEc}cxY$=yppUT50rvA}k7leo1z19CGGU2U+
zES*|GXuFkrj$2C}UBbbZ52i(OZz>OinBWxeH0KU7r`(H{v#;cFtE2j8m}<;U8&lOh
zt_70^eZ6qpcLfJReV0>d-^}vpXT2p}&*g#W*#XSU$_rmc<EramCJK?+vCr-xHC)*l
z?)*fUhwhV~JmR^}A~nbm`-Z3-ZMv&C_hqXIrr?vmZlpQnq!?H|86I(_X<sX@)W5Vq
z<|k>Ae;!ZKH2%ucA{jsV6qWt>NBZviq6uHUqSTnBx7wy>5B{xR#!v29)8^FI&tG_r
z*MF@?*1B5R7B)03@HW*ox73Hcu4As3&$D1|eM95;X7BlOm}s^0=PJlEZru3#`B#ts
zW=-w*8BOzRXJMIDJHKg8&A54<&~=S<3+F7T^EQWSCr!e7b$#8e`E#c%m=mg*r`R*=
z7B)9D1sB%VEuJ#}in^LmL(`N+N;TGdaq|UZy_Yr4UuZ%vm^5k8ocb304~^y}P-(o%
zanhtOP4iEh)X-ShG>6>IDyV5#*c?g?nlWd=f;rwZ$9nCM0rESu(EC+W3z}=^=|-|$
zi>5U!XsoYWJbkS9+zYPqEHd?;G-+1FtlGM{a{%6~IgK>D)`g8y&1#;%pcarUYM5W^
zWIBU7E@%nW6)aL%vPKE#{DmRa#DuyvNtAt+r=Xyqg(&A0&k7ZuIjf;|PH4_oi>@j>
zvuN@3Lht$K7rDjP&k2R**VNhF<2YtrJ*T;j>evk`w_{9k8hcsIoccLU1G_5bgqr3r
zcES>s5g@$g-=s-SW(9U;`khKy-m}z{hO57xgz4X3%vGMc#UX}RsIGQg^Zaks6%?L-
z{>1{5+hmnG%ZZZ?^aXWI*8p)<gdt$YrQLs15`$HNlO{DAAk4x}_g>&NV<N5k&Tt(N
zW=De3SZ^_drhXoi>3na~yk@t~19lT*bS{%p6DPeLDFJCiJPGfy?l4RCz`))8Q5vGT
zcg!9BQ3_(3C1=jZ*NN$4Fuj6luC$@9F7(ogcHbXy_M~NU1T<o2`Jur1(dzD?DrmtW
z-|TXq@!LM4^oVBn0z2G9;^*RG0c4^(1pCn`H~gFl)8;gV>YC@zSqL6o(6VrTjW9K_
zx8F3!&dMp@&Tm4IGMqWDscufKouwPE$cg8czpmX*-+hr?fBB2zdYyQKvhA1w*KIow
z{jIRmx871}!j?PcYR6pXm;<+e!nvw%qKzp~sC4V+l<&gXtw1)G)Ia&RwvY2ITzLNZ
z6U<<nHGkp!(72i!k4?K$(HE59f|MEWw<*zOAjJZLpqmgA&)H`$Xq?aQHPjFTGUnkg
zWMJEzK&2B>=b9Q*Ij5<)u6)jQ4K1O9nt5}Yyl0(dGE7IHrUw0xs_mzOYwNCVx#sNV
z`PVF*Q{S9)hib``B}mIP^Fm%b)ce`mp&qYyLEQrHoZ4EJBhB8=*4HnVpYzo;)Yh3;
zF65GC23GYs{Gta&(KDDo_HjMO@#lUym)QL~`xm9gE9W`etbD;Xy^g!o^>@r7$DHk$
zjjs8koo>%=+iZN*Htq3yE(?+g=X`m7eRIP?n=ZIR!W|ELF1G7a#8YpnpOe05+Pu1&
zZ`3tSZJ5(kTUUEoqk6;6Nt88s@ny5V;NT!*pKl`u+iv3@?B~vsM@a&Ym6gvwx2%lc
zVt(x}wQSj)`FGwwx<q$AuXFmvt%nPrBF8_^fe$OrBP?mg*$W*l2Oycv4J}PIb>n7f
z#Gik@F+6UJP~F0s`Ss~^H8F25vqL9M3e2f*C|k_1U&zA2o_8+B*Ghq<LHabP!E>y4
zPCX;NuGUvy-(b+&6nSffMC=;ooQ8En(*VngkAl^e_2E}o$4-jQp`vNkbXT8dZEX)_
zt%2=ET;-{;2k6xrpe$-;rMf*Vaniw`LdXw=_KZ29d38;fg&+~44$X&yxQo*Uz(Ysx
z022TIvL7_&XVn`#EsDW7TL3S%K=$T&^XG;xfGIGRDRXN*ix<of%?dTlYMQg~8{X0M
znx@Fv7tEYDzxh(U3JN_<bv27-&24H}aD+H%eBZSO$<3;5XcUcF#LSs14sKdQJ+sgr
zeh!Hma3Cewwt_;Z-xnQ{{>;Ku>U8u0WGLWartSX@qDhBN=Cna{j;DFC_>_fLH_d8T
z>}i>0##F=Hxy^N<G4rR)V)Zt6eqHUXt6S#It!skG{Z^feUOz_m%X^l=n{?Mvn8*)@
z;?j5ur@HRA6CUk{*pI{@{=;gh<yV5ma#`~9W~Nh&BeR|p!R(3bj?SWgl!C7E%xY|E
zs4;Y|`Kz^8xfsb^@Y`fs<H_}zwXk7foiK;p=)~L>A#meNnlzwy!68fG!{CYZye>mZ
zGqd4R@Z5CJ9AW{4Gw#wh&7q+;Al^ArepBnNnZGctW{TZ>Y27zl>YBm5W>BXUljiul
z%A^EDQwWt(0dh#Og&xR4^H&$NTqVp8QLq;~&dSMln5o{d&|vZ;>z#mpVtrIzcXeI!
zH7#|QusCgkYRqyyF0E^jm%~;Fi!5rW9}tL%yi7$C{E9={4IDB)l{tAHi~4#NhnxPp
zFMbF1to!=n)yI20`Ixt%p}hw4p&XBAcV}OG&B=s!^~Hzf5{}tU{M-$F@dKEfUh0b%
z!yN2J&AJ{l7wzZ{%q+CJpOru6Ub#7dG8P8naSpL^Ev4rU&VPnv9-utTV(z@yjk!B3
z9v=ylkvk+FpNYAN>)kpqvxddv_h9bDd>nJlC*tv8Fd}QXwYeB`6E|4T#N5q&^BtIb
zN5<ou<j<zk0nB2avM7cj%o-h!*JEzNT!FcUm$q)htVJJiH>RIwFb`nP#LNXAwU}cu
zLzsTdcFYdU-MnGD33D&zvzS>tXS^3PggFw`@fDcGn7hxTJh`XF<7?#h6CcxGMmd->
zF>`@;E#_Fv5T;-7$E?Q8=i-YH`4(d?!98;Z<tm)hgL@TT5s%jnBOG%H=HoZT<9efT
zH|9f_dog!oX0=fsW-ey#iKN3Ei&>26$Mj>az?><6;`!StA9wEefREzcOg@-*+(LRz
zw--lACvzF~!<_jO;>rCp;$i0h0ytRI<5$1|vv?hFz|8+G@RR@VX$R(}J7|YY&UB7t
zysh~Y?ZKSc6_2mNT(Kb@e_j3?>0e$$x&yNwa}#De=Hr;FFn43#fw>p+A<W!O)C02^
z^8n^d%v{ECE#_Fv5T+lq9kUv91!f3y4Q2=C9hf6Yw|g_=4)@+2)DtuJFQoqj>7S>b
zn7J=dKIWR28JC#;*W&Sf=E)tH^_cuOU&gmC^<*s0$@tW;<Fc1$9G4^ik^HUvQ(ycn
zJBs-oI6iQ5;%Be#i<dI*Ji~LQ56`>MI^o*vR?intn|#)}`JYw>%2QwFw<^m`iN|Gj
z=9PLf?Q($!VahA#Ugq^^#>J+EK96sB&eY*~Q-^z}4lkHGyvR3v;*8;YPHa1_V@P!H
zvaDM(zdOjJt0vveXuIWVK5l9lD4l<JWm+;*pB1EQe7P?^CWCZ)>~u4RcYdx7;ezUO
z2Y;=v^u^!8%qnk8&(}A+$UnTaZAgb(6C&~D+6L`Rg{$o^7<f@ae?)O;n_G@Fr(*b@
zB79fhuN=GG{^7IJ>#uU^`P<H!g=@egZhm92w)1y0=TD!;OoJhHo-uq~Mrx(qc249t
z@jL3jLt*Y8-kOnKIMI~uaq8^>Z?+1Lt~^Y=SEr{{x?#XK@B#g-_HH^%y5)zatERlp
zzP|Wqm3wGCeRN0a&>)gs&n1+%KHeAqh05D=nDY9yQE+^Sa*8tJ@z0aqGc{!_rob_p
zwJh^i!qp!Kh+jH59^ZrA{nXwAgy%swZrAwqe=~g?x&7OiUU$ufa?&k_uKX>ny}scE
zZOJjGF%=@-JTAFCP4Vn{`i6VkOwCmO652hF3yAYfKMhzu!y+@S0!k<PTlopDh9v*2
z*9T1JA3i`oWs$A``uA{Jy%Tfzu>F)z{W_tSjik#O)-*7j`psk2Gemj&c^%;a>;sI8
zL*~As3BI?HZh2ljZVqxL`WN`pziop%4E%D5_qf_~YCQe~^>aU!`#S!6PLId`TmHv5
z9+jWkx%kuZ_^y=v<Ua*>>F4Zot~kvnf4%lNaR7g<Yl;T&Z^yr20RL6^ul_?{+^d|a
zuMFjX2mWiP#Pu9@;w#4X5dM2uhkTj-|1j;#V!f1y7B9m+@u{4V_~-QV&zC<Dv+<vY
zJu5)J(-*$sd475!bD7HKQ~GMsR}v5MXX@kP&uY?F;eRpxk+polaRXXEbbPNMT_XoL
z*05g5+TAamnd9AqQgNk<HE<za><g$ZYhILxOnnn{%A{*G>0I2PbegAXr)N?;eyKcw
zzrn>R?VCb)DdCzn?x%KZy*Y6J|9bq32JmmkzX1R7%EUk1D?{b4a`L}c9+VG^g(pNt
zJ;HbQ5WbFdDR{)Dy_ydYy43clUau3s@%Mf4CMQ1`%I9_H(an>AkLK};RQRONEyWi+
zN3yQf`1=myEvu9Ght8!L!?zzw4`!0idu`mN&#(7ON1YExpVv^o^`-GR-EQC@L-n`^
z|8@BP^z*iVqEFqmN9*7;9I14I!(P(mElST%{==wW4*m?=#3%oJ{Idt}pMt*!|3ms8
z|NSk8`io!KGk||P{;_`kYTqjSyUG7yjk!Z%>>u8b1`*4&?{U(t9U#BQS^v+zj`gd-
z*DV9}+aY;sd<=uX2z)CZ|EA{8Skk59UafQ-+^g~xe+uzeE^*?QA3eVKn%MR6xNc#}
zsy}jk0~g}2zkzWNM$B42U^`|EZ#@)$JWjgRx5nd7Y20pFG*J7&JBN<jVGO*$ig^4@
z;j2A|Nw+6GoyJWy<#|@d<JXZ+{l~8vx8ldv{y!)wGSIR{UVE5&3`!fj!UwmJU*~ER
z$CQyh9-MW<cLxbBcaZP$|KQ-a`YAAAzK76q^;Z_@^47$UdHq>T{B6Ylz-d>%d3V_L
z=geW$Z*4r@E%>h)uzmxsPdZ39@xQ?Pg8%koOQ&`{OFh@YFG_tiIc8bcBymjwMGC%q
zDQEj{`r=>I{5W8jBbud8`HA0M4c|FP;S(39&jaDylzE{3s3yPV@TcQd{-N{tkiH<6
znZGMY7x>+w_~SPGO9$}Zg#W|={GY|YXaIk;vtR)KVYJVS|7og-GhQ{0^YK^w{{B<&
zSN{F|XW~Cled?X~0*Cu4|8^(;{{E}*SA9~~If{P={(1Oo__?3_AHqKe|Ni;x#(yIH
z>Q&^Fc_{gUm;M<FJhH0j&qMpjKYRdwNb-pN_r~KRly1s(1Em`fkJM9M_XFI?I`DWH
zaNS!>mDi9i`yu25A2OZfB8?Bn<87L6-2>Kh!1f&=-8SSDx;Q6m*tZ61-y!J)|8mmp
zxh);ee*8PYe0jU#@jC39hqn$Y8=ki;vuwCGnpHNupkr`gc+q!<1cpy+JFa4QY5nkt
z%q3PHW-Ws>0hvO!P1ne$lzf~isocyL&wT&JR!`<W)9(+uzOszq?=z$Ld6B=V{9QAe
zpL`;gVVZ9~jsk3B6z%&s7%ZN1<*1<?Ftn#$X_|2;WU@j=G4k>-r^!*O)dXg(96!b(
zvtB=U;kXtUny-k#@-77U;2Ai7LW0vVf23&tjbjn!b8=Lkibw`?upIMnG_q(jAy476
zl=_+PD(a-VnfAmp3o}0_??+6SW0_4QGf3<iM-_CS*nIKKU9G_CVo;-j>Hm?h%A)T)
zzmUVnpIW+Gj*Iyd75jj{j8`4UU*&k4pGo|g&(1xbxyqAy+6sKAx5sD@{}&d)-F%69
z{s~~3PZky2Cy$5txzP6Ku$J4V5pK<85ec0uh<*;omG1mc#68}PD~qaqi=uo2$M@Y5
z<C)uO{XD!`BzaaTypg|C+;FX}Riily`BxkVG0mr|^aOH~=HGCrGUl`Mti`T8*9hh}
z60QndBS(xH%Q0V$cW|tA>-l4Rl&4AaQyl&BoI;rLJVkAOi@z=WJ)){TO&02<7v-2q
zJ->*#T@F1?uMmSCSL0ahmS*=)31JO*Jtc3|<VsBQ#WU}0Wo#7RhfAe?SuwWq*TSFq
zrr@LcPFIMJpYgW8O4~?MpPXY|KUq|h$?{OK0)qLb@rw}DGh439`9tDqe>N;-uXKEj
z<&QY7r!Ai2D74lM`34S+T@!LE4%O{Pn4T5<+%1c~c;<^YQMu;_iBPEyWykTO{N<>W
z<KOvH=vU>qm%mf^n=Qva{)9%#1d-unsjAM9;~X5H#1z})`4T_MppGf-QN0v$iK5+s
zL)|Jo;n{>ksPflxs0WnWIyp|Dbqe{h(u~5PH2+7AQXB%7df8LWkE*jk<p?P$<X4LJ
zdmM>!HsVlj`<3;RI8@F-IX=Z0QOG+=GaiT1?3JSuhjOdM^wjgCa#~bQn-g+C(Kb2`
z?P7SI!11WE)#r=>Eu~a~E*&)MU6mb`<5&F2u}zLH{xmv_)Jd4@+w#?C=JTysKE+>=
ze4pk|wK>3_XAgfVW7ffaC(}st$#N>T8qJw8IP{v5`HqgCgunicxS!+i1*OqYxIz|v
zcKYvQyO~N|plna&r%GY^jxBr{F+a_pAyn(I2*;W)fya&sKYs5fPiFg|5p;lh<pCkB
zKhk7Pfv4rT2ZsRoGdXhbR-q#lycR@Lj4Ao1=UIN#Cx274Gw8~xm}jd?HT=h5nlFMw
zG~;9qmmlKzx$O_PNAa313)PwTGov)}zdW3uzu>6HG~b1!F2Yu-kUWT?x^S!<AK*~o
z=gFbHB4uQ9>%?0XHjR7KahS$Ej$=9@<}+~nDqW({`<S}@NgOv}z6*3bR}S)c^mM8D
z5-@(*@$kr^5_c%>JN!xfYraHW(K_?VqPUi7)yJRS`Y@kMSC`;Yf80n7GFtdi%p2tR
z7C+-mTz-Ui74i#}^(%hl_#<Y<?feMF<|7tgq8@{7H?gq@cWqJe77kPSw7ADa3vGO_
zbQ%xllSNq@x~t(Y$7fXVHvX>V&wTb+KNtH%{+^THH~7=!F<(5hJ0tVCj2avoHjnFP
zA%B9N`M%AsJfGLkclpztHQ!J96$p1L<bA*)$GHqs&mg+!hnVJDi9>a}TOmII5Q=uE
z()?Tw(wXl!1Sg`MKpe#w#E|oxf@7Zx(G!WIKF~1roQz`yrulx3LmB)*KkNAUikl{$
z`O_OcnX58XB^BPIv_IkR9RAF=onM8#sgS4m69~*F3aXHU3c1;7<X~0n>(qWc<~P;1
zJ=AFe<`atcb((TAzCTk<mhe;K)<i&5?M)5->Nrf+iLvtr!RC`i8O&Gy`}ph7z}I*9
zxZ)s8+@HH~O@jaZ`Ckb9F9iN65b%yHHRYC`YMZkiv)eJfpK`(-)8oW9iTH6bM}PCj
zn&~OwzwGBW6o+tODt%9VvecL^Uf=8Y8Mlkg*MDb@aevDE+DiagwN};_bHcS~)F+&%
zkGr=UWUd@kD#KtmnX&9r8Rq<e%-En(83v=td`)POPxw+_o^6;#iDNZJ0*|%Y^>fL^
zHmAM0PGl{F^o_6$PtHlD#^(A@%(2~Wy@llUX>qL2Ex#{rx2MV~HU1ovHSKeX^Xzfl
zJBQohCer`F&wdB~{fj!~6uTa0I_7zfIo&a@aLl=mxyUiU>zF@u%%3~vosPN1F?TuU
zD~|b|V-6W%mwSq1p6Qt9Ip%c7yuvZ(I_4tB{H|mE&@q4Rn0Gqn7RTJ>n6EhIdyY9I
z&ne$A&veZ59CNy3Ug4N?9dnUme%CR7=$Jou%sU-(i(~F`%vT)qJ;xm4%->TS^GwG)
z&oTS=ljg6+uEw**(u`@-CV2}kzq(~%sKr}6ZsNFe&Yn=xVw@8SZ4ZwVT2s&Q3Gbr1
zrsnw#3%%!#`~0|~vy1Jh=Ug9ZYH1E#caFCpH0PS;@t2mBm-#L$n|1Ezi^k8L&uO7T
zZ-Gv$)_O@<UVcS|oxG40Zu4~uLUXRh3^m#2JlBTKU|mzAXWYVuP~AA+)Zp1<?YOR4
z*fQ?wmihIyXLAN>oLuwfG|%&ltG#X^`PpWu$qsQUoaUHCSW{j793^mUjrAeVI33OM
zj0>Sh>={SB=M!em<a)-{%|nZxW6-na)sohAk!ej0hg56o8qo#S2~RtZS#z42=3Hmz
z;o84mLqf`%vtWJ==^7}4pK+#Ko%{8SySkZ!q&29Q)-4RtH6P3ODROc9Tp3OrR4N10
z7zw>C+m*!guKP1sgu7gPH*-)aE=Rb4>6IhpbMcewrNwA!`p}?K6C|A$efhSL`0bzG
zPsn@VI=B3lgGx=@m5xWHaV7D+i&q_Q!z>H8yXjs0A9Fp#4e-ejoZR#-9u_mNx#C85
z=}F9q=+Y0Om`{16eqX~O-0h}!@%2tIuY9fMH(7r-o@;)CaM1$S?b3rP7+QVZ=GU&j
zVCUA~@%DTRi~Qa6F8!J4@QhD65gjs^Nbk14y<d8lp5>fmCv@qb(u>Bs<-6&Z;U}0l
zU{28gr#^`%-)Xjy_*Gp~zc7x}^d1NA8xaCwr3u1pr*c_g*Idyry}SM?bru}IlX76n
z=caSLevBoxe0ROF*KenD>9!lcfBjcG>0S7H%6xXzGCTbg2M%uf{^hUjmww7u?DSI{
zu2OQh)USW}clJxa_9{F5+N<pHrE8k{x&7z*bz)IWxBgPw)VFD&oj&9)oQ$tM)*P!b
zcJiyC?pD)X->v_SoxUhFGWP!E{{=q{bvM1c{);*3`^WF^evPy$$xXk<S-<Q##`JHI
zR%6#q@2-FLJLz4(xZ~R`-|ff!{mS2e!yJ4Nv`WUq=f*dV6u0ZI?S$0&dz|&Br`1ls
z&NkfSiTvF3LkLezzwX94Chj^3ocO%vH~Hh%+w~bnH09;CzuQT_-AP~NF6^BIUN@m*
zo<w?;)W4tI^f&a2uekbLQ^V6O+XOg2m)vB>IL(dd`1f;~eB7~{n!d<Lo57zBBXI_S
zHT=17bMcqv_dw<6NFvAg8Huz~zW(K>1RY|{q;?Yrl@8!P2;u!e{+Z~44&<Me;CnnN
zUnX2oY{2+K5`0lgblQ2G$E8o{{@EUv?xg!4?{Vo#y8qAw-;?ftLIVG%`!k4Kyq*$}
ziS*CK*D2oAbGXOF!|DED3Kzeo1Y{y0b@673H~F08aq(fg|H<GtU#gd}XCnA^@mFeK
zCfgS-UP|}>q{qcK>Hep9Ts)HQKLWfkC)LZ?GsP9U^FAds6B(#GpHsYP*J&Pio~HY2
zHJJWsyk>dcj;H&&BIQHj@M#kf|Aut<6g&P>6HonSxbge>f5q{ax_0Wf!10$Fb?VpZ
z_^<5e|1XZe)caGv|8)GN)|~pScl@P(occZJ`0E%z>SrX?Y!pasKlKw`7?d(Ty-whX
z8NywsdDbkQV|;L$ZzPt1;9O`Toa~WWZ0hHhD|Oe@Z?Y3#>a3|>h2t-E)zohm{;H=d
z?`d)ZwRxs`c1PzJ9~a-Pbo{sV^LN|T>G%`Pe0O3Qs9o+jk$PzA=l1WgmURCeCc~3G
zQWs49c02x33rzi5je~ya=kJaOshOsJ`<#3_`o(w0gVX|3zoCX`FwRlmOZ{}AuQraP
z)|dJfIsQ`5OZ}!f{!*_?{jPBQrPh}E&3F8zwwC(ojDyORnpx`SjvuMVrhc~)U;XaN
z_r926hr0b(uxyU;aJ)T#ApSu8Xs-uO_DFp#^?TOI-<2DBoj^DLvHkKN3=ut2{!*t(
z{XT6RCwrtWmHOoyVtlek>Pi1^bJrUq*LBsOgQB8R(j<<cX`5;tN&}@?&wI0fW+{ZR
zcVFDGch}RMac~UM_3X^-`q?u(llkY}HdR&ygc7OXe4r3DQk^QOQiG5Il>$Ynp}>bq
z2`(Ry3JP)~0fi5ZKO|~YM8)r%d+wX}?t5?6No<QJdv@kGch0-_+;h)4_uTvLiC$*B
z(DtL~HHY#%P9?56C109)=B#@DdPY4rf$u|9(AMYlYy#20VB)4i|4G3wTR80oq4IWo
z^-r1p`<;uH-uN5FllJ9%LdV2)O`Bim@<}_D)vo=RFzkaJylmOK1B@rzOZQ3Z#|OE5
z(oQY%gk*ohKGHJ}w2*Au(@mz6w2xij_Y>dt>;ApUzsGVQ{!XRX!T*->cdixbM!X|{
z{#G$jy&H7Qc`;`C^0&Z={?$&AiaExw{4SSgDE9ImIQ&Fs>O;Cd`{dh0z`34}6&ayt
za_HKY@)w>f@@L)zf_MMHI76|QADZ$O{SPCUa(Aqx{_vR${Au6^Xa=>S>uL1-9|5O&
z<~EB=fY<i%?w8>JQ+ac~l9>;j08Z__Ch>X<f4-WG%ZJFH$ARAkI_8`nb6?u7LUc}B
z<Kj8s2O@sUuF$%gLH~9*=4pI5!1zx(UE!OFFSLMDdy7YlOqrjl!#mom&iq-{{GshZ
zbR7gv<*$kS@UwHkp9MZe1+43Z4E!6w)9U$c;MCsJpRF;ave4TPflMp^0OR1>D@883
zPs%@&QT~Mt{2wy#9|--+!Z!&|IS;}$CqA713C$ncD^Axpfv45;72w2AbIy;EpCt_4
zyHou3ql|;k_i6g3T~9Es?-hN0CIf#FIMwsQ2Q;065c(xFgm7~%)IEZK6*!OIDP5lS
z^wRZR#`XKyD~1r2H|IqO8D|8X<oxkfO-F{i^LxOF4;PPVydcEBB6O}=<N6JubH<``
z4;V=G@4j2p`7fbEdt|9Tmp)fy#DsCaD)^OWHBQek(e;mp&Zmm}`CI(zIP3V~U68{!
zEr0JR;A#E)Jn;K(gZ@0C=|F7w`a0pLk2zlrW{9uXnErd6ZL3{(fH=|5$vTP@ck%#<
zpXNL*Gk!k_e99Dx{A~kg{`_Z=(I3foZvZF$n{(sPe7@edUzc|tD)MJ~rkJi{!0Enb
z{%HE~l;B;f{yV^l|5rY!`{lNMn$90HuHQ#rUjk0`G3VNWEquM2QT}z{Q&J}S=Z<@5
zZkC+)eE>Msr)=rHbBsqR==Z-b<?poQ@b7>VKg~I4^b8MOM-ZnZd}^u4pH0|wDEOtP
zG;aF+4}nvCY<su=r!;=ay6-0#XVS5k4d8U&)B7}^`Pmqh_@>~y|EBND&pZMD11!AK
zeRCEcz73rE_q5fI*MO5AzWmu5-}fZH;+`;AM8~n}c^o*8Yq1vvDSs9?(H~g!p9D_z
zFK!gM<TWY(@4$)vMT`DD_i8#Vi%uRm^ZyRbhwliTXMhu(si!sl=Vd^?B;|AGHU82*
z{e$~4pooq+_mH1G##gJrsa@tgMu-7l8&ZBs#ydU3PS?Kz#}G?g{{#H4+wOI)NSx3V
zco#wsY2_aPPWor+H#Gn6lzKjtQT{{*-Xl7=?_XH=eS*tlTzp*fjdv!YQZER;EB;JV
zD*AQ7FMYAd$U(th7W}H<=LP?fp>NscF5-&RFP9edeV>(feI7XL0g)ewA76hWxH%{9
zF~L7_pQf`b_Z2dZ&$twd{22kK{@u04<3%ZN&I2_+`G%CgV3q#?m&dsKa*>M~J-?1X
zDa-SxHU5`E=+A&tdvjusM6Eh6GcJW9f9UavG<hfjXZjCo`Y+3G9ufTXMUBHu^VJdj
z+9Mj@6~+Ck;LeW5yTZ7GQ+j;u%J}Ls6}ZR1>Auc;b${`*YWVKYf%EwKViBLPIhbD^
z=U=9DKbrGDjUIRnIMM0;Z;?yVGk|pc^#P4vlkt8~>hm4oB>$IW-e>I2o50ifbNBsH
zAB)e^!0!Ydd))mZm)CWOzMcS1dawAKHHJ^hmrnub`hTy;2t6}NSL<h}p7%PJWqcWX
z+6A6g&kMkb&Xg+h?+-|O|4Q%+uW0-={ucXQKAcYHqri#Ig*SD1dKQzeAOBg6m;EAt
zHgwJcpJF8b@&s_!Z{<V$1=KHJuLDo3|9d{7`F!<Pba{HViLQ?WCq3V=^z$hxe?jC#
zn(KTTIMv^26#4gj)&$<&_kgB<Mdabf_**;xBDguX)X@27!E+ze^`U3U=z9OpY5Es`
ztH__tnFF5$PW^k)>fhf6PV#(3<cFTAr|U(bGj+SBPtTgt_2kcMzL|49>l${h0VjDY
z{7y{}{pStftOuljg^aWLQC%NL<a}3t^&D_2fBC5*BYYMC-n|H%<?vOF8$aZsU!d~I
zs{)+LU)|N^`OF3s`=^ZZKa}#_i@Lm-?-xFnPQMDA#_46LzgZ|6W|aRz2L3JJT+g4@
ze5Pkq>B>E*>F)~tDMoR~KXBsD)h`tpp=apn`aa|Oef0HR)4Kew@PD6t*=IajQony%
z%9jheJ}2Y{Ujxqk|Ft3)H1#}iP}A>zS>yaHJxcs8aPmuS{quP)?>Lv=r^_=jynGEf
z(RUu#xH)^{b)kRJ8b^2J^nIr+x%v#_hU)n51u1X)=YJ2J?mIQq_uY4!{>{5xqMy9(
z&%ld}qhG9fYa2MxFQ3-*=}a2BZuipZe<%YlXW*?2{0ZPBpLfc5H2Uo;Tpo7W(i`6w
zy!g2y*Y6C!LJ#LPpUWTA`11$!7xyuqvVKSB{3(5CC?ET`xER|Hm7^VIF$AYz+ek{G
zlv4TR59gB5LIU!VNb-hWvQ+YZbB;Q)eEjgCWu^MVbyZSU<JGCkV7Z`F(BX|_{H;0d
z+ivtsqE>|jvXFW|Ls6a5hdVF3b9^V0xB|z6`_A~w`Pi3q>ay=lIPul4r#qjhmYb}p
zm+UVu$v-dIXfH|LOLk#CrFW9#^Rub^qHHGdQ*CLFojHth!V#8fogMvYc-G|}M=#ne
zO|Pa9i5%Hj?+y9`9LhRfca%E1xTcm%m8wz>4rIislqla*gSM)-aN=ek2mK}>g;RXh
zW~;MaYpDi~=kKf9aGOpP=F=q`({l^+9w?_2QndL|Z(LN^jM&?8^ybILu-)FlPa>cA
zIQv*pr5;{8wBjp&bqST#HJA4?dX7K*aK*2x>Y>BSK9wMe4oNiPK%b70cGS|NtA|!f
zi=kY*g~d*HM|H6$wW%Zti*<6k%crH!2l_PC)RWWondzgo=<(}S<@=RGDxk}ujhRZS
zAj8e|7Ww>TMW=$dtb3|?$qxbC+v~LUjQfUdOrAPFv(l(-`fJCc^`-ae$HmX=Re`n2
z0@!cbHD{j-y5#vId|erMIi-+bpmScenp>NLGpgC^b$aDv^<weJa_R8m$&)HSJrh#O
z$PVGo9${Im4BPrlTEE-~!XM#VSQ2C@piD-swbdgtm1eV1stnvbPPcC00P99`gcLZw
zKU%AfYn|lLqU*={llh}>nCz^HvSH5EZ#dl^HY0CHYn3<cO=AoT-!fV|e;9j!gJzFe
zxQb>JumL)#LzqBo6R*u$yHp46QZ@#KR5-z)U#k0+8eV*A#IZ`BTCYl-zB-OvL~UeD
zh%+4N6>x^LU9D<K4*j>2SeP0oS-laoCQ0%hc?TdMvrwm$<VK-cP&QCby<mMfz((;>
z1A{Cn>1)*-bz)f8Sb`9xkdCUZ>ir=rB4@o;KS#NOO!LDYFh5U90W6>yw$(?##@Q(M
zanf}*>*FjY@ptEp_^;O*s$T04M)5-3PCwXI=nc*oVE2dajSMmIy!iRWi=v(VwaQ>_
z#966I0c0ZZ$2Of>+m`+s3w9zcgzrP_7Q&2OrIY=hmush_?b5ocs<gZ>Lrf|;wY(4}
z?jkKY%c;Bk%DnPt=n!pILOC_dnYc(+=eav86B|2j!i`g+QB_E)n{=91->@T_?ar2#
z1F}R%M`}&1<YKGV@55?#hl9S|U)3tQ{4I^vxHPC!jaJ!`5Hs|Wf?8PVbehej8?M%y
z+kvea6t%fGP#d+NWjA{DSY<F<r=ErSfznxl%hYIYv1#_>jhS}cy_BEpRVY_Mvq4!1
z`hF0s1r?W-Hgtuj=*W4P`~bt{Q#g2@w2#^<EwAT%-<>tz8n`?j2wJ!z)-E5lx3db?
zVqUv9+og>#If_T}OL`Sz;rs2%B3?)EK5}a9SQWL}Y$CT>yV==j)OKvP7zN|r%mgZ0
z$r%yJP!uWFa|6~>)wgR<x@?_`MOb1yD5KozJoQjzPzYc<=)s$r68H_H67NY}$f%hF
zcXLthvNw}UOFpCdOFF}yvs0#&Ds$kr@6IO@>-p_ah;AZtXUSn&A^lWPYxKWSSz<Zl
zG*@oI<#F@o<a0g>os74(MQlt|yB2KOBBI%r9N4bQBfGRUy77bC$V&3AUbRhHw)%*!
zRwR*^`Avp=d2ZR9^uO5Y^yn;o8oYk6harlmMe<H{$4Y*dbG#*;uFz&>ddOO&w5u1n
ze|Z&V#?ANPcf(~udI#7?9@vK=WD!?#fn3SOcH?-LJP>~y?nvJ+m+R0$<W+@&$K|w&
zD+gs1&;!DoG2RTkzPK~f_UkQZ87~JHr;bcHY8|tHbC@F`OF)ypm_uNI(;7m5!cj(<
z-j4aYz1*HD_{;5C{CCmB<#xWXK<`~rP;I!i<k@2w<&Yk#)>dq1!^$T6+C?pp--X)P
z2iI8@)bQ|HQt|Vnd#DWbEM|>uJdrnOhJ>wTXI?&<t7-{Z4yZWs5%VD<M9Io{r&#b3
zN@poW=~Tr{NK!#FN>$3iX{NK)l|`E0R{7bw@b0%Noz}uiw~o<^*Rcv|V1gR@tr>%9
zJO(W0E_x}BU1dh$A6kAKzDgt5QbRl`If>OLHLF>NWz-<`Mq@`>;W5eePAlpANY6y{
z#zwDc``y$<=^dm}Cd?T%bQj{DFK60F;ZAuJRAw<9Cx%q|GjQR$ZjQ|fCK;U_>KK|b
z**=)oe;M~)2xp@H+SxIuZ<SchX|Wp1R9J<?)C{%}(_m;~%z&5l1eoUJx^<Rtx%1JH
z%N`iU?8H$^ViNVB>HRo@VFebhbFyq|m%)2?=5Vi5YtS_6S3A8{!w=BI+_}LV@~w4x
zP0LG(%(Yoz%-Q*l>Md}RYp&DiXp0kCnLTE&V{RtB0WX?3Zh|4(;}#MNLMce{a$p=L
zmEIiKf>Augqvti@d=Q>1TASu<mB`szBd2!HPKXaapGKK0qm1=RQa5`i*|x*A!1#l)
z*{-bzBNwh6LXenQu(^UU-RNK}#vRc~iYhu3S+k1gV#)Scqo!2sQDb^;yj_NDjmKMt
zL@$H_9m;YR3#$?LXyP3Bodz#M0SGmU$^J=>P>NFr$>c#eDD<IlRKNLHVY9hK8Mu+1
zasgB6EhG-x-lp!R+~J7w1aKuN>uafec05Hp3uSzk+!+XQRl(y9D-xQNxeliTPf_(K
zYg~hP##kfGt`}^fPViJ4!}?&tfXwtPO{UDqOtB%_!{p!O0I;4Vz**!8WerLx8SYeY
zPE3@WY!l{JD3oE3DubCi(xk@C6q2PP12%jsds2bGP3WlEb2ioJcN9`tHd+xME}^>?
zC_Y}M@HpE5vI66%8qL1Lj+uwMf#iCPsGxhiXw*mSo0GlD?Rt>!1t2uY)u22c7|AiI
z!y1_USj^2<)9*A|J&Vv;n_>-Y@IpCC20_hN`CDb~$GXeqL7*NR2F-rGiGkE;_BGoP
za*D^f7vSO`GJ&w0Z4R;g7LJE_>)e^B1JHzNm03Y|*lVg$&>IYEi2S4|XG_~cHAV8(
z`s3l97K)6|09~4GQ!kt0bfbDa)F1g7?Y+@7wc3fzT9;`_hVT^S0v8)YGe1^HE-PtC
zvb=&tQ<7&L?d|CJJggZ7%+g}TbOSf9gUV96#4#lw5sxz6$AsAQ;>h%ZgZOZ}GqQbY
zapl6+$)XG9Q6?@UJU3r(t*>MGE9%JYsSc9?`fv#hvW7sCF$#;`p4K2pRGqAWF$N|w
zkyIIxVKQln?Yb>vR0!?Juvxp|@q_6m@>&}mCm~O=MY50*_D!UQ3p&ZIEI=aPCe~En
zzCBun?}cBZn;@k((?R?MIf&4Z#z#zaRjq>IE3sNFX_1wg0<8ifg*!zXAQiz(x7j>r
zsi(1S;sre!UZf0Yd?}o96E4iGRywmAtzrL6l7!`C(gI{ZxIRBK8PxM~c1B`KC_ACI
zZh<>beK2PJULL*fuOsat^L-rFC7lrWD0sA5nqzYGbxtp5TkLp8qYM_f)!8%(FkG#Z
zNo&GY!TuF1W3*M7s4Dw^+Be1`r*kvffIW3_n%-jft*b|$I@jD$b&Nxf;Ta`$YrUCD
z(ChnaK|=Pl%1HLGhne{R9&N-S4HZSw5_peBrb!sRw8d<=mEtvthN1-|nfmXw7JYMK
zw<C<@$jTE%a`pQ5#3+>KQiPo%?J*}N#dn#&t65%UxJf=&QkT*k2Ld`aiO^$MI<z#7
zUJZud@Rs7?2u9JOkgfY&rAF=RRQ?`MT*A<QX=|XvK;UUVqHk?C)>sHJ=aNTyf)>gg
z#3r^Av*&u!3AVAQC*oH$SC0G45FV^phFmSLR#TTfj?%J6Xw1#L<|Zzu!R{m5@&sut
ztTeHleKQuktX&+~J(IDEBO?;5;~?UWko^be#tpeuQ^9;}DZvT)-QiZ<1O>*1a@_Sd
zLtoKScpjpY=@(QwH2ajv5BBe@RXSA_)Ag9n&|C={W3W@OQKGfAKDqcF{ll_)PSM!!
z>5W^pI%N<K<3%_qaZ~1{5GJfj4eDo7C#9atnm^qABN^(+`NZT4fhfHvjW*?AWvf$9
zbA_x~HLXL_-bTx9nvTw?bK{+bmIk<PrzmqeMG_~A81QK;s@d)icCePxg;U(=)i#?M
z(wnrAuDjG3;=suYeuv#SQKy04zuh2yf#zy#O<2WI$hsgKgRlb$u^+B+%C0kQ>5}#I
zmd@X3<HQS{>)6S&H9>firfZg?113Nu&S-XLM|qg3?oBvxVu!PrlH|<H(wR>8M)Ri0
zM~Na_yS+`@V*4E&z8XNpMmc+ky=CKEIy<7pg-8HJ&z>=!cq30Fewb|REE|R!-P%)I
z=i4#86l(O0P(AId94lAb4G6s&Xdd3GszXfVrAf$|7D@0g#pQ9-iLLX{g;<~L58D*w
z!BY<T=^T&v!5q|6t<_9h44>#CMps8R!p4O${cx}Bcmrp+>;?IaE{9#m;vFR7?e62m
zR6dWrbbfM2Fons)4m5cEl2kCopCJnsv;HhGGOdN7AN%KP7#Y^r6hNzYBU`lG@s{JH
yz`1#B(ZKfbdXtXR)e(z~*$Mp4>-TAyhMtpzZf3c*)54Nwbc+e?6mqzo&i?^Xs&=jb

literal 0
HcmV?d00001

diff --git a/configure b/configure
index ee90f6bc..f6bc2da7 100755
--- a/configure
+++ b/configure
@@ -2309,7 +2309,7 @@ ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
 
-am__api_version='1.15'
+am__api_version='1.14'
 
 ac_aux_dir=
 for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
@@ -2510,8 +2510,8 @@ test "$program_suffix" != NONE &&
 ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
 program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
 
-# Expand $ac_aux_dir to an absolute path.
-am_aux_dir=`cd "$ac_aux_dir" && pwd`
+# expand $ac_aux_dir to an absolute path
+am_aux_dir=`cd $ac_aux_dir && pwd`
 
 if test x"${MISSING+set}" != xset; then
   case $am_aux_dir in
@@ -2530,7 +2530,7 @@ else
 $as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
 fi
 
-if test x"${install_sh+set}" != xset; then
+if test x"${install_sh}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -2858,8 +2858,8 @@ MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 mkdir_p='$(MKDIR_P)'
 
-# We need awk for the "check" target (and possibly the TAP driver).  The
-# system "awk" is bad on some platforms.
+# We need awk for the "check" target.  The system "awk" is bad on
+# some platforms.
 # Always define AMTAR for backward compatibility.  Yes, it's still used
 # in the wild :-(  We should find a proper way to deprecate it ...
 AMTAR='$${TAR-tar}'
@@ -2918,7 +2918,6 @@ END
 fi
 
 
-
 ac_config_headers="$ac_config_headers lib/Grid_config.h"
 
 
diff --git a/lib/Grid_config.h b/lib/Grid_config.h
new file mode 100644
index 00000000..e1674850
--- /dev/null
+++ b/lib/Grid_config.h
@@ -0,0 +1,121 @@
+/* lib/Grid_config.h.  Generated from Grid_config.h.in by configure.  */
+/* lib/Grid_config.h.in.  Generated from configure.ac by autoheader.  */
+
+/* AVX */
+/* #undef AVX1 */
+
+/* AVX2 */
+/* #undef AVX2 */
+
+/* AVX512 */
+/* #undef AVX512 */
+
+/* GRID_COMMS_MPI */
+/* #undef GRID_COMMS_MPI */
+
+/* GRID_COMMS_NONE */
+#define GRID_COMMS_NONE 1
+
+/* Define to 1 if you have the declaration of `be64toh', and to 0 if you
+   don't. */
+#define HAVE_DECL_BE64TOH 1
+
+/* Define to 1 if you have the declaration of `ntohll', and to 0 if you don't.
+   */
+#define HAVE_DECL_NTOHLL 0
+
+/* Define to 1 if you have the <endian.h> header file. */
+#define HAVE_ENDIAN_H 1
+
+/* Define to 1 if you have the `gettimeofday' function. */
+#define HAVE_GETTIMEOFDAY 1
+
+/* Define to 1 if you have the <gmp.h> header file. */
+#define HAVE_GMP_H 1
+
+/* Define to 1 if you have the <inttypes.h> header file. */
+#define HAVE_INTTYPES_H 1
+
+/* Define to 1 if you have the <malloc.h> header file. */
+#define HAVE_MALLOC_H 1
+
+/* Define to 1 if you have the <malloc/malloc.h> header file. */
+/* #undef HAVE_MALLOC_MALLOC_H */
+
+/* Define to 1 if you have the <memory.h> header file. */
+#define HAVE_MEMORY_H 1
+
+/* Define to 1 if you have the <mm_malloc.h> header file. */
+#define HAVE_MM_MALLOC_H 1
+
+/* Define to 1 if you have the <stdint.h> header file. */
+#define HAVE_STDINT_H 1
+
+/* Define to 1 if you have the <stdlib.h> header file. */
+#define HAVE_STDLIB_H 1
+
+/* Define to 1 if you have the <strings.h> header file. */
+#define HAVE_STRINGS_H 1
+
+/* Define to 1 if you have the <string.h> header file. */
+#define HAVE_STRING_H 1
+
+/* Define to 1 if you have the <sys/stat.h> header file. */
+#define HAVE_SYS_STAT_H 1
+
+/* Define to 1 if you have the <sys/types.h> header file. */
+#define HAVE_SYS_TYPES_H 1
+
+/* Define to 1 if you have the <unistd.h> header file. */
+#define HAVE_UNISTD_H 1
+
+/* Name of package */
+#define PACKAGE "grid"
+
+/* Define to the address where bug reports for this package should be sent. */
+#define PACKAGE_BUGREPORT "paboyle@ph.ed.ac.uk"
+
+/* Define to the full name of this package. */
+#define PACKAGE_NAME "Grid"
+
+/* Define to the full name and version of this package. */
+#define PACKAGE_STRING "Grid 1.0"
+
+/* Define to the one symbol short name of this package. */
+#define PACKAGE_TARNAME "grid"
+
+/* Define to the home page for this package. */
+#define PACKAGE_URL ""
+
+/* Define to the version of this package. */
+#define PACKAGE_VERSION "1.0"
+
+/* SSE4 */
+#define SSE4 1
+
+/* Define to 1 if you have the ANSI C header files. */
+#define STDC_HEADERS 1
+
+/* Version number of package */
+#define VERSION "1.0"
+
+/* Define for Solaris 2.5.1 so the uint32_t typedef from <sys/synch.h>,
+   <pthread.h>, or <semaphore.h> is not used. If the typedef were allowed, the
+   #define below would cause a syntax error. */
+/* #undef _UINT32_T */
+
+/* Define for Solaris 2.5.1 so the uint64_t typedef from <sys/synch.h>,
+   <pthread.h>, or <semaphore.h> is not used. If the typedef were allowed, the
+   #define below would cause a syntax error. */
+/* #undef _UINT64_T */
+
+/* Define to `unsigned int' if <sys/types.h> does not define. */
+/* #undef size_t */
+
+/* Define to the type of an unsigned integer type of width exactly 32 bits if
+   such a type exists and the standard includes do not define it. */
+/* #undef uint32_t */
+
+/* Define to the type of an unsigned integer type of width exactly 64 bits if
+   such a type exists and the standard includes do not define it. */
+/* #undef uint64_t */
diff --git a/lib/algorithms/approx/.dirstamp b/lib/algorithms/approx/.dirstamp
new file mode 100644
index 00000000..e69de29b
diff --git a/lib/algorithms/approx/Zolotarev.cc b/lib/algorithms/approx/Zolotarev.cc
index dea7a439..7629bbde 100755
--- a/lib/algorithms/approx/Zolotarev.cc
+++ b/lib/algorithms/approx/Zolotarev.cc
@@ -1,5 +1,5 @@
 /* -*- Mode: C; comment-column: 22; fill-column: 79; compile-command: "gcc -o zolotarev zolotarev.c -ansi -pedantic -lm -DTEST"; -*- */
-#define VERSION Source Time-stamp: <19-OCT-2004 09:33:22.00 adk@MISSCONTRARY>
+#define VERSION Source Time-stamp: <2015-05-18 16:32:08 neo>
 
 /* These C routines evalute the optimal rational approximation to the signum
  * function for epsilon < |x| < 1 using Zolotarev's theorem.
@@ -26,13 +26,13 @@
 #define INTERNAL_PRECISION double
 #endif
 
-#include "zolotarev.h"
+#include "Zolotarev.h"
 #define ZOLOTAREV_INTERNAL
 #undef ZOLOTAREV_DATA
 #define ZOLOTAREV_DATA izd
 #undef ZPRECISION
 #define ZPRECISION INTERNAL_PRECISION
-#include "zolotarev.h"
+#include "Zolotarev.h"
 #undef ZOLOTAREV_INTERNAL
 
 /* The ANSI standard appears not to know what pi is */
diff --git a/lib/algorithms/approx/zolotarev.h b/lib/algorithms/approx/Zolotarev.h
similarity index 100%
rename from lib/algorithms/approx/zolotarev.h
rename to lib/algorithms/approx/Zolotarev.h
diff --git a/lib/communicator/.dirstamp b/lib/communicator/.dirstamp
new file mode 100644
index 00000000..e69de29b
diff --git a/lib/qcd/.dirstamp b/lib/qcd/.dirstamp
new file mode 100644
index 00000000..e69de29b
diff --git a/lib/simd/Grid_vComplexD.h b/lib/simd/Grid_vComplexD.h
index f0bc2da4..41c8509f 100644
--- a/lib/simd/Grid_vComplexD.h
+++ b/lib/simd/Grid_vComplexD.h
@@ -345,6 +345,9 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
 // REDUCE FIXME must be a cleaner implementation
        friend inline ComplexD Reduce(const vComplexD & in)
        { 
+#if defined SSE4
+         return ComplexD(in.v[0], in.v[1]); // inefficient
+#endif
 #if defined (AVX1) || defined(AVX2)
 	 //            return std::complex<double>(_mm256_mask_reduce_add_pd(0x55, in.v),_mm256_mask_reduce_add_pd(0xAA, in.v));
 	 __attribute__ ((aligned(32))) double c_[4];
diff --git a/lib/simd/Grid_vComplexF.h b/lib/simd/Grid_vComplexF.h
index ac15d19f..8a892d24 100644
--- a/lib/simd/Grid_vComplexF.h
+++ b/lib/simd/Grid_vComplexF.h
@@ -219,7 +219,7 @@ namespace Grid {
             ret.v = _mm256_set_ps(b,a,b,a,b,a,b,a);
 #endif
 #ifdef SSE4
-            ret.v = _mm_set_ps(a,b,a,b);
+            ret.v = _mm_set_ps(b,a,b,a);
 #endif
 #ifdef AVX512
             ret.v = _mm512_set_ps(b,a,b,a,b,a,b,a,b,a,b,a,b,a,b,a);
@@ -354,9 +354,7 @@ namespace Grid {
 
 #endif
 #ifdef SSE4
-	    cvec tmp;
-	    tmp = _mm_addsub_ps(ret.v,_mm_shuffle_ps(in.v,in.v,_MM_SHUFFLE(2,3,0,1))); // ymm1 <- br,bi
-	    ret.v=_mm_shuffle_ps(tmp,tmp,_MM_SHUFFLE(2,3,0,1));
+	    ret.v = _mm_xor_ps(_mm_addsub_ps(ret.v,in.v), _mm_set1_ps(-0.f));
 #endif
 #ifdef AVX512
             ret.v = _mm512_mask_sub_ps(in.v,0xaaaa,ret.v,in.v); // Zero out 0+real 0-imag 
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
new file mode 100644
index 00000000..14b2ddf0
--- /dev/null
+++ b/lib/simd/Grid_vector_types.h
@@ -0,0 +1,299 @@
+#ifndef GRID_VECTOR_TYPES
+#define GRID_VECTOR_TYPES
+
+
+namespace Grid {
+
+  // To take the floating point type of real/complex type
+  template <typename T> 
+    struct RealPart {
+      typedef T type;
+    };
+  template <typename T> 
+    struct RealPart< std::complex<T> >{
+    typedef T type;
+  };
+  ////////////////////////////////////////////////////////
+
+  // Check for complexity with type traits
+  template <typename T> 
+    struct is_complex : std::false_type {};
+  template < typename T > 
+    struct is_complex< std::complex<T> >: std::true_type {};
+  ////////////////////////////////////////////////////////
+
+
+  // Define the operation templates functors
+  template < class SIMD, class Operation > 
+    SIMD binary(SIMD src_1, SIMD src_2, Operation op){
+    return op(src_1, src_2);
+  }
+
+  template < class SIMD, class Operation > 
+    SIMD unary(SIMD src, Operation op){
+    return op(src);
+  }
+  ///////////////////////////////////////////////
+
+  /*
+    @brief Grid_simd class for the SIMD vector type operations
+
+   */
+  template < class Scalar_type, class Vector_type > 
+    class Grid_simd {
+    
+  public:
+    typedef typename RealPart < Scalar_type >::type Real; 
+    Vector_type v;
+    
+
+    static inline int Nsimd(void) { return sizeof(Vector_type)/sizeof(Scalar_type);}
+
+    // Constructors
+    Grid_simd & operator = ( Zero & z){
+      vzero(*this);
+      return (*this);
+    }
+    Grid_simd(){};
+    
+    
+    //Enable if complex type
+    template < class S = Scalar_type > 
+    Grid_simd(typename std::enable_if< is_complex < S >::value, S>::type a){
+      vsplat(*this,a);
+    };
+    
+
+    Grid_simd(Real a){
+      vsplat(*this,Scalar_type(a));
+    };
+
+
+
+       
+    ///////////////////////////////////////////////
+    // mac, mult, sub, add, adj
+    // Should do an AVX2 version with mac.
+    ///////////////////////////////////////////////
+    friend inline void mac (Grid_simd * __restrict__ y,const Grid_simd * __restrict__ a,const Grid_simd *__restrict__ x){ *y = (*a)*(*x)+(*y); };
+    friend inline void mult(Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) * (*r); }
+    friend inline void sub (Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) - (*r); }
+    friend inline void add (Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) + (*r); }
+    friend inline Grid_simd adj(const Grid_simd &in){ return conj(in); }
+        
+    //////////////////////////////////
+    // Initialise to 1,0,i
+    //////////////////////////////////
+    // if not complex overload here 
+    friend inline void vone(Grid_simd &ret)      { vsplat(ret,1.0); }
+    friend inline void vzero(Grid_simd &ret)     { vsplat(ret,0.0); }
+
+    // overload for complex type
+    template <  class S = Scalar_type,typename std::enable_if < is_complex < S >::value, int >::type = 0 > 
+    friend inline void vone(Grid_simd &ret)      { vsplat(ret,1.0,0.0); }
+    template < class S = Scalar_type,typename std::enable_if < is_complex < S >::value, int >::type = 0 > 
+    friend inline void vzero(Grid_simd &ret)     { vsplat(ret,0.0,0.0); }
+
+    // For integral type
+    template <  class S = Scalar_type,typename std::enable_if < std::is_integral < S >::value, int >::type = 0 > 
+    friend inline void vone(Grid_simd &ret)      { vsplat(ret,1); }
+    template <  class S = Scalar_type,typename std::enable_if < std::is_integral < S >::value, int >::type = 0 > 
+    friend inline void vzero(Grid_simd &ret)      { vsplat(ret,0); }
+
+
+    
+    // do not compile if real or integer, send an error message from the compiler
+    template < class S = Scalar_type,typename std::enable_if < is_complex < S >::value, int >::type = 0 > 
+    friend inline void vcomplex_i(Grid_simd &ret){ vsplat(ret,0.0,1.0);}
+       
+
+    
+   
+    ////////////////////////////////////
+    // Arithmetic operator overloads +,-,*
+    ////////////////////////////////////
+    friend inline Grid_simd operator + (Grid_simd a, Grid_simd b)
+    {
+      vComplexF ret;
+      // FIXME call the binary op
+      return ret;
+    };
+        
+    friend inline Grid_simd operator - (Grid_simd a, Grid_simd b)
+    {
+      vComplexF ret;
+      // FIXME call the binary op
+      return ret;
+    };
+        
+    friend inline Grid_simd operator * (Grid_simd a, Grid_simd b)
+      {
+	vComplexF ret;
+	// FIXME call the binary op
+	return ret;
+      };
+      
+
+    ////////////////////////////////////////////////////////////////////////
+    // FIXME:  gonna remove these load/store, get, set, prefetch
+    ////////////////////////////////////////////////////////////////////////
+    friend inline void vset(Grid_simd &ret, Scalar_type *a){
+      // FIXME set 
+    }
+        
+    ///////////////////////
+    // Splat
+    ///////////////////////
+    // overload if complex
+    template < class S = Scalar_type > 
+    friend inline void vsplat(Grid_simd &ret, typename std::enable_if< is_complex < S >::value, S>::type c){
+      Real a= real(c);
+      Real b= imag(c);
+      vsplat(ret,a,b);
+    }
+
+    // this only for the complex version
+    template < class S = Scalar_type, typename std::enable_if < is_complex < S >::value, int >::type = 0 > 
+    friend inline void vsplat(Grid_simd &ret,Real a, Real b){
+      // FIXME add operator
+    }    
+
+    //if real fill with a, if complex fill with a in the real part
+    friend inline void vsplat(Grid_simd &ret,Real a){
+      // FIXME add operator
+    }    
+
+
+
+    friend inline void vstore(const Grid_simd &ret, Scalar_type *a){
+      //FIXME
+    }
+    friend inline void vprefetch(const Grid_simd &v)
+    {
+      _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
+    }
+
+
+
+    friend inline Scalar_type Reduce(const Grid_simd & in)
+    {
+      // FIXME add operator
+    }
+
+    friend inline Grid_simd operator * (const Scalar_type &a, Grid_simd b){
+      Grid_simd va;
+      vsplat(va,a);
+      return va*b;
+    }
+    friend inline Grid_simd operator * (Grid_simd b,const Scalar_type &a){
+      return a*b;
+    }
+
+    ///////////////////////
+    // Conjugate
+    ///////////////////////
+								     
+    friend inline Grid_simd  conj(const Grid_simd  &in){
+      Grid_simd  ret ; vzero(ret);
+      // FIXME add operator
+      return ret;
+    }
+    friend inline Grid_simd timesMinusI(const Grid_simd &in){
+      Grid_simd ret; 
+      vzero(ret);
+      // FIXME add operator
+      return ret;
+    }
+    friend inline Grid_simd timesI(const Grid_simd &in){
+      Grid_simd ret; vzero(ret);
+      // FIXME add operator
+      return ret;
+    }
+        
+    // Unary negation
+    friend inline Grid_simd operator -(const Grid_simd &r) {
+      vComplexF ret;
+      vzero(ret);
+      ret = ret - r;
+      return ret;
+    }
+    // *=,+=,-= operators
+    inline Grid_simd &operator *=(const Grid_simd &r) {
+      *this = (*this)*r;
+      return *this;
+    }
+    inline Grid_simd &operator +=(const Grid_simd &r) {
+      *this = *this+r;
+      return *this;
+    }
+    inline Grid_simd &operator -=(const Grid_simd &r) {
+      *this = *this-r;
+      return *this;
+    }
+
+
+
+    friend inline void permute(Grid_simd &y,Grid_simd b,int perm)
+    {
+      Gpermute<Grid_simd>(y,b,perm);
+    }
+    friend inline void merge(Grid_simd &y,std::vector<Scalar_type *> &extracted)
+    {
+      Gmerge<Grid_simd,Scalar_type >(y,extracted);
+    }
+    friend inline void extract(const Grid_simd &y,std::vector<Scalar_type *> &extracted)
+    {
+      Gextract<Grid_simd,Scalar_type>(y,extracted);
+    }
+    friend inline void merge(Grid_simd &y,std::vector<Scalar_type > &extracted)
+    {
+      Gmerge<Grid_simd,Scalar_type >(y,extracted);
+    }
+    friend inline void extract(const Grid_simd &y,std::vector<Scalar_type > &extracted)
+    {
+      Gextract<Grid_simd,Scalar_type>(y,extracted);
+    }
+
+
+  };// end of Grid_simd class definition 
+
+
+
+
+
+
+  template<class scalar_type, class vector_type > 
+    inline Grid_simd< scalar_type, vector_type>  innerProduct(const Grid_simd< scalar_type, vector_type> & l, const Grid_simd< scalar_type, vector_type> & r) 
+  {
+    return conj(l)*r; 
+  }
+
+  template<class scalar_type, class vector_type >
+  inline void zeroit(Grid_simd< scalar_type, vector_type> &z){ vzero(z);}
+
+
+  template<class scalar_type, class vector_type >
+  inline Grid_simd< scalar_type, vector_type> outerProduct(const Grid_simd< scalar_type, vector_type> &l, const Grid_simd< scalar_type, vector_type>& r)
+  {
+    return l*r;
+  }
+
+
+  template<class scalar_type, class vector_type >
+  inline Grid_simd< scalar_type, vector_type> trace(const Grid_simd< scalar_type, vector_type> &arg){
+    return arg;
+  }
+
+
+  // Define available types (now change names to avoid clashing)
+  typedef __m128 SIMD_type;// decided at compilation time
+  typedef Grid_simd< float                 , SIMD_type > MyRealF;
+  typedef Grid_simd< double                , SIMD_type > MyRealD;
+  typedef Grid_simd< std::complex< float > , SIMD_type > MyComplexF;
+  typedef Grid_simd< std::complex< double >, SIMD_type > MyComplexD;
+
+
+
+}
+
+#endif
diff --git a/lib/stamp-h1 b/lib/stamp-h1
new file mode 100644
index 00000000..753f890f
--- /dev/null
+++ b/lib/stamp-h1
@@ -0,0 +1 @@
+timestamp for lib/Grid_config.h
diff --git a/lib/stencil/.dirstamp b/lib/stencil/.dirstamp
new file mode 100644
index 00000000..e69de29b
diff --git a/tests/Grid_cshift b/tests/Grid_cshift
new file mode 100755
index 0000000000000000000000000000000000000000..54d76474cbf499d5a7c8eede1e31c62f6efade12
GIT binary patch
literal 60539
zcmeFa4R{pQ^*=tFED$v^i$z4Fy4t9TASQxJ0CksaU=}uz3Tjj^33)M*ki=xeiy}lf
zu}qg$wAf;+Ep4r(ibB;|K!t<=iKrN~DoWLeZ?i-tf|7v9{yv{OGn?6vX#4$s`#k^W
z|8(-~oH^&7d+xb!=brmAlckxyNpUutWL|O7RT8n<VJ<<^?~;`Fj6e<ASm`9mC7mM;
zm->U!4?mX(A=IBt*ky<|VS>RI4`|14Y@ALQ8|M-aCLFGlk%E4i8WJ+S?#pF7>M&`I
zj+#)CoPxl3)V1YFfdb}<2NQNtAzWhz>ClWv^%$?qh}UJrGofVYZ$dMF%o~3*jqsUq
zE?~r61d>PGyCHdbWq;x?6weRpxb&b)e=y(}L!JrEct1rv>iMr;4ukO;BfY-LrMrIg
z2xdK8QdT<W@-df`6<%CcT3)m8;)P=`zxeVoqpB-LT`KZT`Q)8E?OGAhEXSe1sPka_
zSmX?E{P4EPKm9}HGS?M%esp8rZ4>QF{&t*Um*K~F1Ms`6a8Q}kc2Zn|&F+eG)!G;~
z7QgZMCCy9BXu0d;*B>c(ar?dVA}wD!uKsQ42G>{b-}HbpIrB~5;8lqkQc1tZemlH!
z!y@n8=MbgO>k%k-S}&oa(dc@kuYiKR(VLKg-sm&>kh8E4dOs9mZ*m6p!OwDc_M-ne
zebAkK$a$lWa+uf$|M`8;*CXBD^v~~u|M@=BeYg+)aedH7_E9gR`zVK>_aW!0KIqf?
zkh7$Z{Lb$~pQC-$?`M6`%lptL)Cd2^eU#@TeenOd5BlAGq`SM1{C4(1AJT_DrG1oV
zeIMoaOdtAu*GIp5zYl#*?L+?LKIF&sA^+7r=v{r}_pLt4Ii(N&xqa|Y?IYbA`_TVZ
z$R8vPlU6^ShZ2z_2juelC(7Ym&>Nn~6L^leOOhY_q`UqWOTQ863jV+9^jd@e4THYA
z2mfHunQrZKI=|E4zs!_x=#y>GHyim~e}>DT=wP`a$N6lYAej0;0*rR1a;Qtt&D$>x
zeIz5@S|i|c(8=#Q(`8T$fN`zbUsy0=1jn+=rQF<G=2evER{Qg-{JFVO?v(kt*A(4S
zTJ0~Y$|%dPt}d#Ua=kONatn*94D!sZjIxUIqM7+~%8K-`-a~RL{UudJ`GvW2ix%Y;
z;K>j68>h{v9)lDr(^7LOwAx=)u_*JJsWbAu1({H$U}1i4WldF4?)=g!e@%WFV@#dl
zzpTK&sImwwX`>NsL1|%8d2VTWsXueR5Lr5=vZ!dTKxX*UuE@<r$`uQ8=OFR?va*T-
zovY%q;<B3R5|iz6D_cQ+wVzyw4tfRDD$1XSB&v(5{N4gk(+W!Rt8)ES`KA79=#hC1
zXrt%kSC<x;+}@eFGgEUjXI!3}J$1(DT%kIds`ASVE9N1u^Ggegu7LVNaby`Ki}cb_
z((_8oQ98PoJ!7OJ#w|tu`T1o=X8mJJE2@#V>Y_1T6oDtrq+gDl=9d*k3Dp^Cnb}hd
zqJ_(ntiGb4VqRr_RS`o(+4W418WuxEqnOI7y~Y1({!?>jrilVTl75j;S<yli3n~`%
zUV3@yE#(zR*{HLs>db3O4Hd@fC5d7$Ex$!C{N74kw319@i>1f1&Gn)vrdJk~&#Ne8
z>zaApG&C%<qJ{Z$O6RAsDHO03=d$^r;EOBFMm7hnxjeTZ-(OH7bkJMnrK!0ya*O6w
z`WNL^6;+~FR8-~PQp7m7EL@mdSyWYBQJ!B`is1RFMzrb$`ITr5tjw9$L8r^CS(#H)
z+;=_@V}^fBK}C)K#JFgEeZ<uZv!bjzlZ98lV4jE^lkH2>a&za;EA1ugsD=tQD5GH)
zvK?2<M7Kvkcf0ajYSdB1=*k*@Cfmu(naJ9|RZED(+!a+-^`ao^sJd7EX82JBM)xf%
zD!;{Fk_!nHRngWO+xXEG*qRE<vgb=MZ1Z%RR**lZx|sMa1qHd)q@_u@*Oo6REiZ)W
zsjit<#J~cbxw4|VbYbqiqIsy-TgoM57*&}|%Ljv4oL^cdRiaV)iw)z1e!8&KFIAV`
zBAQ8ckzYbw0~Ay)lF(QErSpm^iVO1>kuAT#Us_Qv&B-sA3+qu(MD}?V^NYGExz&s2
z%|V4C1*&D(;(7U{<x+7~QIS+yU0G9J;Fl(6WaN$>l`2j4c_(J%UOMX1QI|!Z)5hqJ
zIXPe%Gb%j>OB*X?<m9*~dUMl8jgAR11Q@Z>(yW0TFXAwVi4%813@Nzur+CbA%qI-3
zcxGrE=B_$_9A-5Mm__UH$-($1;hq54AMj+snSlE#`1Kbw()B++alU86-+qjtrr)1^
z!J<<n#C;y)KQD&AYYDFzR(cX=R87%vXo6?PPsMNL{lA3RQ>51nn$90@;qMQLIzND<
zE=_^*CrR%b{3Tt#0-hjg2L8snxp?X?bsBiR;72JS6{eXl`0bLz!2Kx?U<w^4ox{xG
z>ROj4AW<kDy&4R^17gKI&<t=Hegx^98=9Bt?~u;9qj{PB4(XShL_pKOApJ@U-SmS<
zzrsQ{{T<TNEp*fGApL3!-SmS<A7`NtU7sf~&Up>jOyEunovp>Zk}Y(-{plg87CNdw
zdX2TvnYMYkEOevjMUY~lQz!Gvw$RNr4~FGf=+^qmv(V8%qgRQAo*1RVkF(HEwa^z@
z=mRbES_@sa(3e^0Y;WeZ!a^Tp5&>6P==2ZGYqf>`Ba;ZY#zLQ9p|7{lxz=G`8!Yr8
zCK0f~LO;VoZ?VwNw9s2E^kEkI9t-^}3%$)kH`mk{tXb$kw(xgY=x1B#T^2gm6rz{W
zAK6BnQ6emXg?^QV?y%6$wa|xJ=;vAJP7D1f7J9OUo@Ak?TIlCn=wmJP3oUe)g`RAo
zD;9c+g`RDpUu2=@Sm+}x^gIiFq=jB$p<isFS6b*U3w@!5KFUI`wa_oI(3e^0wD0D%
z!a`3oiGZst^h+)D)fW0_3w@1+KE^^<0-q-+!F~@u<dl^927f>8j|fx(+Y*{AOn8i3
zb~jie=hxvbpX0(4@e(qH+akEmznwTuVYpS`Rm5om!wmwzl{igbc)h^q5U0rtuNL@?
z#A)inD+GQGahkYrt-z-cXR{4g3Vb4QnzV49z{e4%DGO%{{8Hj46L$&xBH}b@;Z%X2
zN1UcC>=gJg;xu7lhrmxGPSX{Z1b#AcnyhfgaS+arBTiElZWH*?RlsSY!mR>7Nc=S7
z4Fca!{71ys3w$4Onw;=zfxk=qbmA)n{yK4*o^Y+ecM_+`30DeyEAgSk^8~($_?g7B
z1^zT~nw+pp;ExlhsR^eF{13!wV!}>=|C%^WOV}as2Z+<8ge8IhoH$KMxZ^wK|8C+=
z;%x%Ioj6TIxK-d)#AzDB4FbQFI88!$y};)Xrzr@p7Wj?CX#&D41bz+i3y9YWd<t=z
zf^en4ClaR#2<HiW9C3F2aJIlNCC)A%b_x6<;_T|-RDqvIoLxNZ6!<XW?Al?6z)vI2
zt{avFell@(*>J}(QUAo*Rl{upKl&@+?4sdTfgdE!t{H9sp1tK!tZZn1gQzWBE7?qR
z(+2d)98o^mTN*Ya{K05AQ4y~F9yIy-26<4wEsvs?Huu{Gz-vqPG6-skGA@iVKBY50
z6T{eMGUkM*fwOp&d`_K^!~UGTr3=B@>CyCw8v3z((!TJjyKh$1eM;ctj_jG<<_1^D
z1!WprmlHHsp5>CX&ydlti{*1XlC+^2S-V!bxB);3bPZJ0PZq?@ZrZ>EX=uKly@f%M
zHu;=d(l!t?`5U|9!LO)WmB#&7E4Ef;cUKJ}Zz7{o|2iTo>JB~rJ|lkZm5YAkl%$$L
z*CM921Qj%UN55i_Y=@edKP7PGng_t(KRc?z;A^*ZJPGa8Pa>_e!*?Kxe2x^Z*5S48
z>)qG6uXWFyq1^TPDwr+rW(4rMT74mjt6PftY*H;@$iPE7t_>=cByH9Wh@v)ogI9Qi
z*C#0I+e%7jX4)5u`mNIVZ5*pJr7kIt(VOd%Dgk-<hx*bF$v<m@Rncxjp?lTsh5(<s
zR&W#2h<ZbwN1%ZoRY^IhsBciNKf$LS(DG0XQ3i5TMFns(&Fv=IWNp@UxHab(N-j<C
zj<`BA?c-SWQW+KMxd3_j2MP5iQN0(KhPutCKAJ=}!m)tfko{q*#VBUpW3>BB+xjiF
z)7coshjMGfxS;@w0tu}Y$||AdhUiZ1Y$PeVKwVNcSmi-Tfl)+VN>azwOf!!Zs4E5`
zs4I#Gv>kOxwE!a4Dm*auGV&tM8bZWbPX=71ZNM|)w2<UOCAZ=sSeLX1<tDGIOX3PF
zxb}dGcI;r;3TXu$+J#dk3BtxQCN2i?SzCI#E{ThvzJf=EsLZrTR{h6Drkn_%8mUH1
z|AjvEw8njQidca{+FT2HT4z@zf_U`}KG~zweQ1n~>{Hv6#x}dp*2Nr93}G^$`*5=?
z*I$o;4DyE>@|i7-8Wub(lu3I@so!hp(nUIW6AWJe1&aFaiMq^%E@>~-Hz=}axlmD2
zmnVt1jUU+++slgjun<8gLDb<!sL_h?6{0(k1g~iF5hUM7I`K>gSw=c~mR%x4_dt@O
z&Pqx(YNozn$t6DZi;xmg8b7ovfksE5%eD#4MpESIyVDLU>f)pX?Vip^#E8(2P6-m?
zurL~gHhYXBXe9<c+W>hXpM9xU2#Gz4!Iz}?)WW2(KFD{mRD`(hQfY%kmdYMLQ7W6@
zg&?&S;Jn-~u;gJZ*fSdN=-q-F%TQ8+{-gw9UlsL`Pwn)mUuxIDhX}kNsu5z}UsB{z
z+YTx6MJ>q8A=%MbELXM+l7}>;H7G%-rl<$g_7<-wmMaD!<lWWs4SN7z|3<!H_d&V!
zT{&l%qBbjy5vS7lnOz<+{HjP3TUB7IgM~U?Z!e1*CdumxpOn0j?e4&Lc6s?c7)B+8
z5`&g(d2H#(lG)yva;l=ftgsY<o}{tqjq|TlRAeUYi~5&*$d$`~UQAm@o6`~wbKDVC
zzg9G{top;WaGlyNgedA-(e#y&=f@&Vw)%Ew{Yy%)4B}2x)B_>~N}7gRYVW6`!Nn*9
z+k1@8{-M-I;Pu`Kfza2mCsh3fmN_n8(6qc|NkbJh9vW0%il#Jbaf<3mO7<0Wcte?X
zK0DI(x&w<J!Dz;qq>PpCJq<o-mfDFbN`U2%<obtj-@<i^%_}g3vhpSE5u`6@tl_cR
zuCHlbcWi!y4D?Bv=|$iHRNvRu>A~J$e<kSBrUf5OWsjmcsrxlNVt4K0q_L8Jfcxj3
zq>T1qKJ_#0$KOC;(4GYT)Xj8?L5#G@n>#4jiPAx(Yl<jU^eIsjV4_K8O``v!0>^5f
z!H%H>Rv`#=%rsdbD+qxLgA#l}Hv(u2U7(8&CAd}_T%<DA8Uts$rbQOvpszRRG(W;J
z1LwthO3>a!&M7m+lT^vjS_a|T;=oP2P-I}cqL#6pmw_de>7Z)02yM<p(NJb4fvBB@
z&6zBHDXqaBxHA#8bX>l7G^QBv0}O7^6Qqsgk?-aC1~@mm&(xgBu~d&~N8oPEREmqc
zeJUdr#Anps>!Fr#7BO21i&zv~Xl7ESHXjAd5Ei<k9@i$FX=csG3RIhAN2cVK2Murs
z?Y|RsVrzNG-H9gWRtNvi9YkPox?Nt^;+EHC+TFqH6Y+3uf;;~D1bJPnJ6^W!rfFzA
zBoFc=l{9;j%9siE$Pz_uR2sw1V!6DbSRQv?qwSEqWVujLp4F(mdMFYRkz7XUXDf)E
z$Jlz&uSLbc+IW&&hvXX@4pP389Uxl?DX0QpQl8Ni@R)PRBad7wWP1b0`&IqO8=8Mi
z2`x^_)@q#S!>6JA5fH!Nq$wf$dP>Pr&})<HjTGd2QK-<mEGat^7Qx|BI|6OC%~Zyt
zhW%%GRP7-a&d|<u^;(d$hw#wIol40aI38CuF{M)pmHkavB8=a|*CRc(<)O|yj;Yrk
zGaecr>Ku>A#KtRu9dSxvdx8>ZN(eL`R~u_~%cJb)&qU{x@2Q5^EnKzS`~*8ICz#r6
z2hp2J(|fb@75i7y=D(AHjZjfP(4GJfAURUC_d9fs26E6!kV7R0szwRIbbg?1GI*NE
z!@Q7K6y>Re1^PgHus0@GHxmZE546hOn9k~EN+i=%gK0aHV_PTju*k^=Y_x2=+66|=
zc3N|`+hp=5m1uT@V~;gw>vWETY%fl25K~s0lwdrqnG#$(R9Ke!FXXzbVQXvege@EP
z^URF)ODz4z$J!6?Ls`MjYgbWOwEn1h15ubaD(x2Xo3&D;s@K#TR0h3S(#|(&97%Hy
z=(X?;gUu3-2Zu<^9jp;$aH+u+QxES!0WgR<7}OiX3p#@+|Bf&9bYdj@MrROZ|C-Lw
zTZvqO9E%d^#}dI<eL!<Fy}mN<5k=jsl|ekaT_Jip+Ek7h!Mfsn>R}~i$NWKh@9V^J
zK>Hx!UsRtwHRSmlGThzymfSA79PF^Hc6E2S582_`Vme$~cZUl^{!jXwUT=G&^)?TA
z7Br9gskZuyUg~WsIAZGUc7rRX)XxW(S#LM=#$eYOM7@pejo~BI6(opyJ3}yRH!1^7
z`Qc=}Desl*mZ7=T-hrmv@8=mA?Ps!KE+e~W_4=3$n$c(vXv>i<$K+=rKCAu<ePVYZ
zB4lwK^r;bTn23#$4pkp%P&47Fz(aA5XD`4QE-!!R2%Me^wWmLiM8bc^bWK~sr!AOQ
zYQN=EIefP0coxjGQ-~e$;m%!!sKKm+NTZsS;2xiqATLiOB{PvciQq|;m%j_gynDP;
zs3Tir0HTG97ika<kb694s`B!q7@FHp*3M_bv*>R`ngZV<rB^`L*3CjUhJ{biov{uf
z!f+&nd-lFTaiUl1Jg*p(Wl_q>2E`SnbhRTLjHYx-wGmjFQ{bJ5Nv#n24ue7A%vcxN
z3`(@HHvxgqeAhP4{%>;6a?f_(?9QFtgjp|j%|;Ya;_MU(T*#k`$PxdkteiW6w8xXx
zESI^USn`;7LLQF_27NlGtD_l|?mS*#P`dLtm=rUQli>x$=JA{TJ@fbxDA7E=1%&S5
zacBc&!xPXX!e`+|%>!_hCI87(TG3`99j}V@ix0x%5sqmS`A6V!%aixHq|Li<+Y~?>
z-}n`tv_~MA3@7Wd+CxP0btIe&sEt69+6B0U*Q2R3cqYiYT;`!W(`k@JN_VEu5z@a1
z$0J7gHr$F=Sk`L-pAYq_Zz}3HT+ImGfmXRoyAmy|SU!LW(b7O0v?OdbV1ohI8*sG&
zR~WEXhZ&WKl#$09oS#4ksJtkNlYGKt19I9!JOL1_3IH#qIlke37iy>`K^aj4-%)8i
zjyXiNA{<ONM^wpw7E~_L=3$}N7tBzUV4#*UgF)I4U#KA47kXZZO@X_11onG}lCo<)
zCOaP}fdjSLWk$UAQ>#yb^Hi&-0Y-$FRl1nf1U=%0o@bzT1dGi~(-K#3hn{E2K!gs0
z)H7BD9K6Lz4sC-e;LK_$1d5>nBXITwiV{TO1*hi3#vjNwc)J<<<mv=HVn&WH^v4!G
zZ5;$#Pe@zDy~>Q+oo#aZLeI46(RC2ih{fa{(L1#W*JomLPEO2`fBf}qQI)^x0-l-n
zQdWI~Tz3lUlau=dUoZ>hQjHPyiCTnb1+Rsen~=#nvVFlPSg5EtmKqjqFgdW!4>E%G
zjT2S-=83B3MG43zAX_E|?VIIw0SWR|ai17OC@6M8;ayN3%IKW%d<J7aG_iBS7DU|$
zCaS24doXYe&lBZ!_ARP-%!I+anDf7h!I>M6Az<c)V-wZE8!&?eWs8N<{$qXpRxmf)
zHy}wY<n^CuK1Q0e(wpVGuSBQIQX74+4GCx&qD2OSV%lOB?=CjXAvA`ou|nYsPE}Z*
zkD?<Q4Qv@2Cu8_RPdEV}MhC%UVm-wi&~C7ZnH?*JF;Lm;hY+KKps*J)V$Nb^pJs^h
z!3}ayr9qT+nm|RTp<axeRsWJ)_e(r@gO0SrKD9*&v~?lE8<=IIRD9|#CGcsN5?a!w
zG#-e@+HR!Te~zLaQ34;@YKFktybk7m&)g&|&t5#Yv=2EXR}91N8pn$X5CK{femfO-
zN!92eJK8-;$oF`vlJX^(e%gxL4Q;sPci@&%ERS_<<?UtOw(<5V)?{3mo^;f_hS(5=
zwVaCG40)Hb<<l^i8|y+El`S?twG_)sUh}4ZRbyAX+7~LJa+^0jtf=i?dGb4o`i_$F
z6_k%xQV#hCz&h2x<WE)tM{NExagVEkg5R`bayj8?{2o^#m}h@UOo{AJu|V{#z3KwB
z%bXvh?SaknH26JFWAf>FIuUx>ed)WZK1RX%-QHkoR&YR8aBz}0I2%Pga0^OWO8dfC
z&wr?M!c%T_!kXaVr!a*b{J1*!3Dy1=B=-~&_NBjC^*%N!hgybZMTI^eXdbM#&<z>+
zsnt6vUcqJj_1JwAV*(8k=GzII|F{BURpZATLKkCoC*+3}I}b7Gz<_0;wvp#Ql=RnX
zoD3Yw6eeB>DP-3cege+Bwo&X&O-i7lOW1K1+nF2DKO!A+V*-}{)*?KpgZ2TzgwRhT
zs-ym;rB87dwHjfX0}8bNPEniM7vpw9+Wz7g*<<^Xjdj{4U9k@+5R=Actp@W^!g+w*
zNuSQ7*D&b^KSCCesqa_@fZqTL|379t%~1~t>+QpCM!{3GEih>CY>c%47uN!^LDkp2
z<hrGJMCh`|=vjp(e8C{?=Maq*3_3)zqH5?&HXk~Rs`&@grybyJ=8O#dnb;nfj>-0S
zaaET^{Zp3e0Hv92L{cIxii%FNH32SXvzt8$`#W0`1buhSKd`%Sz#Z63dx*7{Cd`<8
zDd?jbHkGhmhxNpQCi&j;#MIG)70A855Ecx3*k{CNKBX6V^92i&5|$`d`|S?SPv&CW
zyJ|Q~)wGL$3>DC5TrlCQxx~_N2+K53quFynEEJ*Dy2p#6TfB|y)<ZWSeJSmAu2Up;
z9)VIXa+j6)lkc4d;0d`R-bjN_ZA6zm&J`4=FZ9#n-rf7Xk*%JPV#9V#8Z3;l9+g$_
zvAP=zKfZ#nH#p(#EVapdZo6oY{=6(~nPHsUO~WKsN&NB@b@At^yp7#VJ-mUVDC{Yr
z#0!+rEUcY{5?|&d!U;=Y^hZTqq~X_rAGR3MVRxKZ4mnoynNNMko8BVV{S0D4H`qMt
zu9Wu!pT=jZFS}Fr`t4Ysxj1m#=KnDkUi^tuLSs_UlD<w=B8}E(ukBT#5y);Dk}K66
zDZ8eGhM((I+p$zbjb^<Ei~C+QzlwSX^tl82+yQ;CMAL{x|0ZW($FZ7^mB?1N4Nf{r
zHhrgD{~-!i?^Hn-#%7c~O&xX#<YW)#Nhx+ToF@19-?F`krXVjrj!9DZ=fKgW#6o<b
z(LkbmrZMef$(1XN_TW+Xvq7|J!?DvY8br{(>Id6{ur^T6r-ozn!t#3N_@}7X;`^fW
zzVkj{cDS_=*p|bIg4}Hd!~GE0jW)G21JRh;`({ji{zDXf-u59p@rT=x<7~_l5uyEE
z)1SufXW(;(TYb%~zU5WFhVc}O$ZEV&u!d@J-$+=2K^YFEIl~E(<aa9Seto|P&Vjyh
zbrq5_c1e727}Ti_rSTK&61}RZbHP4$DE2{WZfMRBT^a2cliC4T%W=4a*dp+#M<cCj
z2BLl7Q$O{n2R-T$ZSkqBoINN52lFdtW*h>7cR2LP(-R_h2eh9cW4%sMj=pCdA!6)Z
ziI_hiX+@oo<O^PqRq#R*twC^70xZa2Q3gR5LOn@Cy;v6V2CoIeE#9=fUiG*aZv6?%
zr2ZfK)D9S-tK|AxbUEz~RwH_8H2N$|DPb6L;QKy9$$4U1_f4hoJ8bK+o>b;2WPgHz
zIUJ6%xrjeAX=plhlIy!*G}taUF~xNB+kb+415669bA0Dwgzk!y>)XkNxIPYm=vKZY
z52Vn9@l6y%YbOnQDk~F?qg?lA<6+q=h&&}U;X8(Cn?wN{%e=|39q5PH<zZ_}q+y4#
z0l<6)yQBk^X2!t`NC4RP2*$JoFIR_L`hM1lTuS669s!`4N8@toX5NIt+#|fT26av)
z(A!h7a-70vQ~R_AG*Bp&tQ63Ou*n56>?`rNI&}Xsp6<<p!NLZ@UFGQM)bg=d-!3C0
zzUm}s>0o&$>#N*5&}MYQsKQ&AifU2d{$diKo}vW#$aXM@IR<Rj7VHIkfhWnM1nn5t
z2PY|NoEK-S(5ihKnIF!3s-~k{u)iqkIFm5~+e$2b7>$J0>g0x{crq$@7+KLQCPJqz
zA7N!|v*TXO!52X6#w+looe735r(rg(2P}*RNC?oLK>N@h!!3N#aSCdpti!-zVm*>H
zgN_pBt8kE%;Pb+x7R~C>cflNdUU=$)@C%(FW&nb4w@zq7a(YX=O((QO33)n!6C<5t
zx=vUXC0wNwYNLdU1>ueoo&OviGYVpmj-~2h@jB)p*8isP*YCg}HktNJoaYF9p6pc*
zd)4pIOok!_32-=}QgC|aA}4&Ck8#LEz8>EHv+ONm8y*#)o3d&M)c4@w6gabF&#_=X
z8L==iy!TG9_qhYjnNWX34qn{05bR&*!+hFaIl15O&;)w;A6~5L1=n{$E4AaH&hOMl
z^+3wow)X`huX|a2H|6M(INNS$iJk8oY%NN`ZYm0M;!tdFdE>8hAjq;_E?@sTws(if
z*SE^a!!RG%hmGHl0QdgY*!_J2!TaQ#SKy^%r+2F((wN@37`<l?&lR}kl*9oTC4(0y
zLHG1_dHF0@73Ser&=6Kx4P)Mr)&Spb09AU)e-%Jl11AsX8&4Si@#L0s<P9h$bQvsQ
zd<e-oiL%1A!Wa~r%2gK`v?(E9da9h`ZN;w*zYaO)q);}7`v&gwyVCbB-j}5w^QxWg
zKLSJ=?WdrE+WX@PII`C`P0neQe<Y0cRZ5^aLHP3eIT#O)Y{)kvh<0lb`Xl^-B&~c8
z`$4ij2{+7{PJan~GC8(S8l&oG+C=cO?`%TdcCUwKYqtZ}>JiwBKkZdt^<unG4|%Du
z*v{;BqGN3sb#>8LJK+jy55k%B!b*uOxU^zuuf}J=r%}=lXW<A(-32JyEMpth8(f>z
z#`=NLbK<y(`o1^t>2cu{ZU9;G*Z8nWI+RNpnC#q%&5N?61aIg@yEjx~$Lu~!ea{>C
z<ain?a)n!McB?JHjAOyk4|U#o4DAuk@=mNxOg|>C8{n}udTbbBP_)=1<6fCJu>E+=
zPI$;%B5*_(N)xiw7F(kattzDplh9r5i+t){CFNUx0*+{4M&LgKJDZ!Zr8!}Px<g*)
ze^IjS!n|6=y!6|+DxWvy+pLgp0~DOK31?%<UW5qUUQx4C5O2{jY?7|K^O&s@TX$~k
z`$0;LWIOH+G+K0PI$pEOt#aTNTjEX@GRhZQ;%*g=2RgfMOrCp_){efGFl~P(3a=}+
zC0O|HiRE8F6iwrIZ(Ey&Xb;ijWnT3gxC~g4mFs>1zrYJ0!yBAtj!(YyupC%}y7ESc
za!i#m*XL-}cs$;R2~|p`Fa3agcP?W4(!Ye8mL*U8#B2Lneqd`>dYk+}Q^Ffra%LO)
z1WpL}omuJI<eyCdGe+`H*bKCex5Ug0f|<IKj`}$&*WHSi<O{jFyeTbA1xp;L2t_@r
z1op+FIOMwf5toLtei<GRCHxv7mzFWt2y^(a>4%5*-W`A}!vVk~?`m;A?|(l2dmBHA
z&$5*#gl9lwC2&-vd$c_bk65&TVz3Jw43yv?uX+Yd2HH4|<Y7_-QktS(N|KmA_Cg9o
z>{T~nf7r?%T7;qaEf`A9KZ7>_xe@W%_6-VLh8tl-yD45wIi?IXDM~tAB^=p1Qu7xt
zmXxN3W{yo@xye3#TCAwM@kGxF^ERa!RfT-%04Gi015QErLcvY4*mus4!Ar{uULA0n
zj5R7K1^>yICYumcF<t(uW;^Ct`@O-d^d`RTbuR2(<yCJ^m>RNw=2a&N|61&QG;O&I
z-h^fYi&pjzYm+WsA=>lI8QEU-H809WShg%UW{TM0f~RCI#EJE4EC9ftg{p$3y?ezM
zZNER7e@}L<+uhic9lD{|8vJSEpy3PRSc)%r3Yu}+-W5UnZ?S~*JGW|o$aqu-KZFs?
zm)=zM5&Cd~F>Ao#&{_zVy8VB6;V{d`c_6MXjF8wK8K<PbBL`jpUD-m<8rRiXC_Htm
zFXgDxcoggT$CZ?~l=S!IyEl<5{as8ZeDcJ0ukB6rNOsi+8WVQ=<jezbG+6CUZ~Axg
z&p42xM?MK4X*a(iwnQM9sdM!LQyWeMxZ3SQ3hGwvc2t8mG-0h08vF<&r5)D(+zNYR
zd(RshzSbC?BHAI)veXNV3SG+zZ9g67u;O?o4*mIE?KW_DZ8xL&>ZM}rbLo{m7=hYS
zB-K5%S;jMQI<^*xyH(iJe~6NU!5ok2Td+DIGp!+#J(QC&gWwh;zenBYQNKb+`u3Wi
z(B;l?Z^S|oObGJzu1EbaQ|-_egFwSt(vF$R;-pmfMz-k~mKzKf!#HnZ<~cOFH?qgX
zu<{r;r47Y$T`iws4^A~&!0d1>Vw%W_?Ey1j#e+%27`OV0qP__~g9GMRV>@Fu9GO{3
z$yo(oD?!WwGD;=}$E;WMy%MYit;gIU(QWJSs0|Nw9#>~3C1LYN-$#r56bDy6ajTPI
zL+((pZ;*{`fM>w(6s8_)Fayv)Jn3KIa1jk#!Aou|F0{KlU#Agn+~@R)k(@2iXWNOB
z($39%w!Nhu?0nsw{<>SvdObArQX5u7yc@b0!i=kL`Oe*e!`$b#ayv|}W3$7uZ)nT_
zrQonPRQAl*LJHP&6?t8!52J76mrmObwX^eWcluGtJQ`Yj1!OiWfzPlQegF&16{o+3
z2?L(3{wB!p;K$YLEUsQ>f}yhEEZkngtr+Xrd+{sY35efutY#m78Gb_>Y-Cr>R-d}P
z-7X}`$8mtMqD4OKC2+tM8ykm*C9f&z`)f{itH;#$!p|ZHSX1rJv*%=+x&up4Fb!}7
zKdwIvN8`QrtFa2DzUM!m>rf7NU_#I)|8A@N-0QZtG0*%#Ts;=(vRC<#qG2yAF+8??
zIQ17i27wLoa}BmuVqK>At;D`&m`8rD&29S!Mw?^l;ia1~VNJvi!@bW#e{8jmiIZ0{
zIvT+6O!Q5()?{=euDzhisQba5<bgpQuC}A=cvKuini#UDp;funR>XJ4EuPBLO2^f&
zQg(~eD=Ayu>APfKD@KttkD7!v6*OBno9~zGW?y+Aa*Xy7lE;Eddy1HVMAx?jv-(Xe
z@1Nm)s$q9z2U?~29!?dr?O`%I)=j?w```{WlNMaTX#{U@z5~V!1K(CGWN;qN6Y8wg
z25gPuSYC6c6G=<{bJ=NK`s8ZJE0%R4SGR~PjHp>d#pzL=ES~LD(sy7`3p8{nDIM?|
zJFt+Ep!0AD-KJPPG~GY!w(scAIhpD%9|{^V#Bvl?w{c>EJ$fu{Rl$CG!Z_?XK>ae}
zsBhRn>*SU*cOVK%F<uES<AgXc;V)RL%|IXsVs;BZ+kwrBzbNSFSZTzONhNra5>(AV
z?1Sim3C6law^4-%PN6B&Esp!j_r3}<2IcEmN=TxXb8b&?<QlGbBzxp_8Xv~`f~k5p
zT#D7_=t6u}@G-jhStB0fEZf`od0ZxZzf-+1Q!H)m)FYV1)_=StIV0q|5_R>In2e)=
zxX5=N4)nkouNjthxP63Peryk}_FJR+u&LcCsb}a=oUK!_c6$t7i8}>}%=$b1^S{|m
z{RQ7Kb%r=!$N7VZ{v$m)d=wn8(<ct@z&4(+dktQIlC~Faja(=81G(V8Pgoo=ccF3c
z->&GJa!M!%>-+6+P@DWia0VLta@g*IJwjs>3;6zlFhK+Go33xXiP=Oty#96AYt!g|
zBJ4Wux7|#ORZNRD3I+!z>Ns}Wgn7bV+YYR`Hgwdyfn_HUdoxTs74=mmWqXZ7eWU$0
z>wbX}_ypSz*c07>tu<r|!#;L6Y&fP2PtM;ze(=vQ3AWzRT>{7LRU_#*h5SEDRNr6+
zjrPy@o4<~~xhwwWWASH0FLZ^*V>Ru8k8HXZcf1Ab3}~$KANO!~+JiwcWXCK`Q&t(x
z(u?+qM(Pd?KEq?%uM}V-0tXQlbXqPWz~tJ81Xqfqz4foxoJQL{t^IUM`gF7XB>$D`
z1I6-EyUmS30ZBm!JA_ye;)x{B_z8cJmruZoI#W!JY-R6A1_%4l_`wL{`&GS9tBj%j
z5@d%S<MlhSqV)<_w5H>>sR>Pc;~cuAqrlKzHmuJd?(fMx`n$8&{yw~Ce?Q}Y(ce?M
zb?l|T|NeXWd+PtPzw`X}fBXCYKl?ieUME)kPueaz{cv=8;XPDO=&R?W$DYvp7j$?0
z?%s^G5YdtKK3rn-;Z(g3qoa5C;gLVohv#n7`|vMzi1+Mb`e$;Mx?SI7a*ItSB{)~%
z0;m^f5RHi=&KmGM3FbcswCsblC36+d8?hm>#tW|_ItIeAfC~-F04M(o|8x)D#7WFj
zKj)6K*nBC#His`XQp~d`><~(gEBx>UunMSzFur2&cZ%7OPu;K0XyW|o6AxB<6Wj61
z42Ie6B=|xfJq4aJ!x9}F`cJieS75+Z-^~h6QN#(7bx_3{d|(wdf=d9W2uHslvW0Vh
z{U}M6ial$0WJ+l9y*LLy@@^&lOSk;XCMEqS?;-nLaChWiHYn*wK;D`b@zm|Dnd5P`
zW~uvXX5#<~R_|`W?^bLew(yV&)+;c8WpWhD;?oR1!R;RF!nifo^$r}c>p@z24!ms)
zLrwT?$FHHQhD$x|IPwQQ+h;+%H~olzrV^^q&xvR_C(_8Rf|PHue&;_$!GObKA>Xy1
z92|}#NW(X&!#8uEsmdATVGaA^s{N^;=Px`2^0-~~JcR|@?QuN4(dG@_n&TaDYqrw(
zMVwNADI)f<loU*3ID^I-!CN@f22TY!dPB?W;B5Ha_EV})5{1Evs{dlio!;P-Y_@cq
z&OucVP&F<f4)Lg6JbrwJ4)s(2YBSeMd@0{RCLNPmf3*vDx7!{IZoB|+{W0@Nya6Em
z8{-84eQ^`(npg|1c`aJMt$O|H4ip+kkNVw;;4JlRZ^#?1-=FuW-^67K9Q9u|VqGX&
zzjFOOi0E<d&ca$;{he$)qMmb%dd7<Mhh3KO_y_8{Vltb6Y`Qe4dlHsnm&@7!8oO%#
z&IT}uhQpB+`U5l7UI-P8XEujxl~6941J4VgIl!-n!}Fbza)iy{6VV*LMRORrL0&g<
zlWioL!;!eEb5{ftSIO%Ve{D<rZP2$0?ZNkJyUq7o?%AU~yj<0m72KpdtP_&FjR)eq
z1+N)R0`CB`Nx*Y@Ykq39No3(zG@C?<u_wmu9=UEMJPIwYK{qrPZH+`kgIOkeHg_`b
zy#x921{1|p2(5<4;;_BU!|ZG~SkV^khR|p?JKWr&^tr_wBSy;+>mPz!yf9)u_4USx
z8?TJG)g3jj;s_)gt2cO~A{Lh#gfdxJxWTdqom+pd63W0D#>k(6lpbE7xAyer`ID8<
zZK(GO*eL9HpV+^zW96QW%B5S%-c7<%I)>dndl#21(+~Nxl+blHC7nL3B2UJy-*yGv
zBBi6d`VO}D503o1ZRB4?Pd3g22JOGH+3yc}er30L?nkw4#X{J-Rbk!Tybo&zaNw*v
zB-&I|L--d*Sr=k!@i)M`yPJMqcZM%GSuqY(?nT}Cg7>k-`GUXVjL8?g9Ge>0A`g6I
ztTY@z9sdY?UKEEnbTfXpJF@VSb#y5L490w^*dtdo!;Ni4$=t%?IS0Qx_%H&$X?(aG
z?>P<W!bD-*ke6%^bskZ1aBc}wTJkD3-0&V0oPWLtCAi`QR~)!V+$zwnctMOei2~IG
z|5s6Y8svSW%R9=bM?|<L1bk)6J1S632>8mBcNC~`%<o&#Z{6~7Y_YIPX|S7}UM!W!
z^*`bqTzwxL%x>&LiCbRsvNye>25tPE@a-s4*bsDGtR{YouDge=3Daks_B$2xQBL?^
z%RAc#V+-avZo&MOC-0ti+ulHQF=HBs2Xp-cM^(wQQGE+sV>a1#xr3g~c90ftw5z*8
z^KG)*ux-<f{@q;Fs<zyuAJRBsy%h)Z7P3xN{l&hz=zH9ipz`ttf5yQxxJsOYorNhl
zR)c*nMk08zZx(oz5RUkRfrr|NYYhoN@N!_bA%M@c(!BV*2N0ze!3yuaDRAd8`A<q~
z;M=x&h>`XZ7LD3}(qnue4_{j(hl105C4k=G@AOgqiDi^L;)z-UF`&H}bz9z|+k%b7
zPVJ#EPUW3$<i9!*L)i)iS)e40W{Z{JZ$$jy3nCGWXtEg5jQ83;^q<bxZ17eZMm7QU
zH`+L`xw;3oXD}Dz{yitwVmLD%`X0?N*?!#{+#+&<B6x*)2(1)vBP*fhN!x&DVL^60
zP73n1W*m4C)!_+>6F``ktQ9<-__ZRAeo*3(7igciVQ-Vi;F6TkTKz3(yptPxSg$Qi
z)8Sstggb^sdekZ2))f?fx!8>%9@~*7=@NsK&`la{kP=PRvD=b%c%uW_;?;TW$5`=2
zZ80<6;C;;7cy}_kNt5LI1EkSeW4l=+_&?QVqZw?m1JOQuo|^CQhOV=_n{g6^rD5_N
z){0PB?%>(c7m**_h-Ub3(xYJW7T{wIx!KBGZ6SxLWHv|AwKPa!v!fQgWCy-KKsd-V
zB9E`YE#y-+pGx@T;#0Eh!19soz#$#kfm1>9pjjhQ<v}w?IORdU5e^jU2spYUDk)Gw
zlVL_+jG09r3Y*O|OZ^hFf<stM=Dr`^8Tw576bKJ94s}PixzqnCFF#+n^m>bPHxEE8
zu{Qo=tdp}-b}yrapFmWTqi0}KJI?##oEcV_m`W{Ju^0opZ?t87lqdz6v-tZr>4QSI
z*8Xjpk8xgX-)Y+K?)d?>lz8?G_8}|%Yq{?CPz#lF%$KqcqYy@-eaAKjp*9o#^WTxz
zluvp7Dl2$KRxlTP0XX<7dG9)2j>+|c5$=rxhz5>lEI7x#5lx8C7cBUZz5#*f{tE_V
zg{DQkyTe&2Z+P)V4R2_w&6oa_SDv^_tVOX#@iuo>%4^=;2d0LGr>7kj{S7BM)edhb
zj+MNL<EK}oy)K?xGlC=QGJ+G<rlH+WiU_KE^n{gaSEF`TT32bqsl{ia!0`^NFa7=b
zXQ|CT+ky6TxqiXN5882x+!)W?!I@8EYa8K;?QPrk40U8349I9*qHV&;3>7;;ShR;D
zd3Pmj+cc~V`_db$YqHdXSt&blY&6TZ+u|=$-xlZyuK@a<Eacz_n>#r21$S`xvo;*N
zbQ}7{4_~S7p}u=c8;XKCv(=`QLund1?F)!`lCDA8@Re?L__H{_i9~Du&|@nJ^O5fQ
z#QG~1h4Fq44jW*@NbIBV48$ND@8t_jXn{CjBF6c&!!*Vi(_L8j9PLei%`c8=;Be^`
z82s!$b!kGFhMoJfnE7BVYwU`H*%8a%KH0O+YuhG2&>+60vHvRh!L12fqetSd#tvK?
znymXpCg~rASxk4BmwV*(pkqhwPYl;yLDRzLSn%!+x*KPuU(xZ!mK+lfv>o%=G=IED
z#UZ3Gc#Px#P8i{gvi9b4w3>&rLW`fYv!0RTPA%_m=m2yDeJZ|Lu@oDvzCR%Y(}91<
z$vk%An(RtC7TEivCfgIW2&9PjoOmp)$EkO8vO7he?WZMBE;?8$bnRK*IA+g-2GJvd
zv>LbKHKdD_^zGFb;B+z?L?;-Uz4E#PC{Ip}#i*99RUcv<cedxvzs{?E>23Tn)fXyE
zN)Q8NbUl)@?gW;U4_|daMv&_-X^8#E0WCq}N2wHf88{T%9&CXs>Dq#wbuZQIXoqDw
zpz#2fLeD(j)+39?`tt?~=T|U%!ReLWz>!m`Zuj1Gq#o7)=i`Idmw4}LyaNZ>u;(cH
zqY|pZ%U@FlpwJcByUQ0Gj0wepJSBb%recY4-t>c6XYR$_D`)J4gK0yB`qbP!MNNF0
z(y(cU;{&I_OKMeA2#++@Jc;Hh%q?C60`60%SHgYG6Wp<hAs?^#f{AZW4-HOGY%RX@
z-O8DVd6zSAnEO+k4<{;88Wa`ksCFbb9mzGK{E;a|p0t|@HenN`wB-q8m@nn4TT^!K
z-f!F4xc}_Zolis}OG&^f!_B{X<6@g4XC72acTIa3#%$v!(C~4_V(0!)bP9YuN#i$^
z8nv^P(gwz+Z&H8Qy@^rK3wC=;5AsnMFvNjJ*jt<(zS13n?Jm3^LA^e8`c8UFQLpf+
z4M+?3-r{*v8oZ?~y7(VZsq`ydJdPt%{D%;a#fItVGBB*)qnDBSAIh??w6TYL<s@`s
z_PYP5{xEj`QT?6&Vf~Z;!}`N6{71`w?0;B)*MC@l<v*-{_Ww$M-J-yt+|{Go=IGnm
ziV73iJh_CnyQ$~=UwmH;{vDh<oPfjEYXWe=y}_HYhdjxlq~Pc&J_NVvz?b@k0|_`j
z(<TSj(!gyvPru^4!OM+T*BXz+g9lr1*m2(JOW!BoJrVMJ=^x^_CC<csj<-xO@c7a-
zoWn`j3CpK(?codmaBuo{`Dae1J@>nX+K1~nWHmxARl!LfjP!VaA2S4BXzDU=%2(bE
z>N5SpdaJjy4QDU<bHKogN}vsA<6if>VVr9wz#@f5VSw_fpRC{s@6&aAjS+e?gyDS&
zF`QtqU#LW0t$EKociTd*ZBjz`PKM(1Rd`Jp5`NFzVojBk^Op9hJm__m5}Z`2sP0PH
zF)2?T<i;#wQZ^vw5R+Vhm_hWH9n&1L<2tA8xEb??8}R!nrVd`r95BG{z!b&3kmuAP
zY6(OwG(@=!Q7%K2>qJpG5S0y43PeFMhzd0PpNa|`waa(+hrf=MiL}>^`7+Nc+1~a~
zgMTyx`j$DMuxAJ|?{Pq7&yZ|BL21tr6oAJ8we3SNg&N|(6l#bQQ>Y<OdzJ&YnGPuL
zb3l7LwoRKCVi!5Mc8JDOR@XWRF+m%$hj37xLqOmy1}-3Qx+4r15UDPd9m|I-6OmUA
zSta1ZL)HkUhaEax=@9fm%N>FbTB7Kh7s_?DV51IVZ2>wEQU|VIFbx5Ly9``F;Hd^K
zAn*<*2Q?%Sf8`K|fDaFG3ciOOI$Y@xbf_WtAfNH@3hFXblNW~UK@3HG!9hq(T7-l;
zhk(F07`TAI*BH2fz)M8D<wF(<xN^uc0UsW+O7K1G(BVplphFYE2Td4Hxoc}1LHZl;
zf8o(t`q<p9Xt!kv3*n_<!Js`${VX~^itS%FGe?&G<&C_i_b<hUNxE^=@twcW4Nrt)
zJkYC_CwS*>UkGoF2jqH8U-Xl3Cxcz9f&V1@ORKB(eZxCWE(>qVbk)B47rmALI%seJ
z|LdT|`0RZ+$p3ZF^f^sm4*q`~w8+wb+d)ffuy|`&5BMQmZHeXYUVq3|-xLR`^|j#s
z-e!J37E9iH{kO@>F2peJD`+<lJ}c^kBrX3jF10k{6#Dc~(z-0QSbTi}rh0Iaw*8OX
zjW15{MqX7l>v<@+OH0N@Ue0K?uY*WQ-QM`cIkxS)zXN0aUR7KAL3lX^6)Z+a6@VfM
zR3ywVxTzm>eh`Qp$EHFgg5^mV`RQ-l-sRERja|UNf$?y+@7$Nt4tvktE!Xk*K$fcE
z-3Y{zrys!gB=J3jz$e(C*^Z-VU&LbvFS6Z#lA^x};U9{(y#8{gBxS%kS@*TLf4rH=
zi`U#|O~HW#d_WI-e~Ets5uVShKHGN?jpb$P@)}M(M(~M&kGl9qc5=U)Au4wLMIMw1
zUm>#tzc&2fiEeLzXW3EnDb7~#Qx)k4^>?)Kg}lz)fg{I_T<)G2@?8|C?!-$Sa4hZc
zT}R+(m;WrhJK;ZtrOB+L5PN?N9I;njHZ_#E&K<herieXL@}s>p+xXDzKBgDlZC~Ss
zpkp|t))Clxtp4?-ZR&QMeycf+@5BaTkDV(I;C$CK+%{bT2^(+1lQ!)S7<-SP?xW-V
zhL+8F)(OQ1e%;#A1l`y1)n&}R-{4t(wNq>^?9@)jM-P<rJ^r7-X@zUQ$3Ki;*W+#g
zcWC~OWcW?BX6*~4MRLOC{9m}>z;|lDL71*BzB+L;wltfy7BHFXvoo-isr~&lAzgm3
z0k0g!)=zhTz<w9q9`QPL7dNB5={SVy@8?x{7Bw_0$+2lDGLD5j9E8xGU5#Y1osif=
z9)fYgUHcofQi2Ie(5oHxBslxY^=Xg|H%31->BX^%1n+oc&wsXfg9ZEgi7L)oZnb-H
z3eQ|U^`J2aN5Z9QH9iZ6U14$FUXeXvxJ+0>WkYTZV-2-2p|N|N6l+f!Ry=uxS@Msx
z&^e)^w~Mboh?6$jaw>{F0wg1p!nZbkDf{sfoVPK8HB_|MwqsZr^!LZnlSCEobwscs
z%{+do-T!+oXD@!<&bgXcEIbE=;#G$yW#PmwPXImP<pCZn%ZuXU66id{K@B|yz%Taf
zf*(opbBGvj1fs1)Ncd)Su5cUG4EEbQ&WSRI7$tbBJRZ;d7NP3H93j|m0C9vlxyqT`
ziZ7M(fJ9}?2SR3wFb6*ug6hiCo<{<Ry3!(c1)rm0A&T`Z*cWR5gl5=uVaw>=bU}~r
z>!TudV8k{gf#stj@_qA#6?s;C@ml-)?~uPjQn(k>uPYtt8(AC~qf4+~lA-7KWYK^1
zzpKvDQ@IIN3mvtQP^Cu+J;W>r`&C)AD(7<_CA5hWP>0tbn)yDjc>jdcAADOFZArXJ
zr*A+G2Nw+GW%$?+r>Vj5Vv7NrQt?U%Ux5&x6TmkM(9HU2t&iYXDF-F(Q>1}dTTn5G
z<qck~4_X7fcq<Umym(myKO@>hh}PUszS7Bg6s8;cH*8ml7NEb4#C_Bi2;2Msjo1yp
zMFtACXN8h>Fklq|U|8_-thN+i=N2C?&^HOK?Mkknz=enOPvp8+nF{tIYXzK{v``lt
zN3oDcvLWvh$iqsBe*~=@AI=k24j)w$QCEt+SnVa~-+msNM)=Pdi743&ZULJxeA=JD
z!;sY?;<JK7bFe0`yQ;Hye<Vk)7oXBYNUdPEhRouchABvWU;CUVL(oysiDo91Y+l6z
z8Y>dRcW{kt@V{}0mfNGR$@SOKZAkw}uG=6I!Yf1+ldX&S>S2nWiEmu$rF$;Kun>~X
z(l|$y20AH=ErG3bGorMg!=e}>ioytE#p>noh>&TPgCG6-6c#(I0EAfc`3J6BfDDDK
zWWJJQW}kNZ3!J}a_7y7X+Wb3aKi$w<G4w7?jAkEfR%zVz5ZOPB?6YB~7`7=DIMTt^
z66Gf7azFfyULG9c*xCj`Zm;Drl;zO{3EESLqL;_**f6C)5!Qh)y*xrfW_NiE)xM?n
z$T%3Ss(uMoNrpmJaz9IXh?j4*z3{V?Am6+NcQTF_V$I7g*XLoAwz=Qq=|(3L{nhzv
zvg+L}R_w2t{kyi&B1B*4vKlS9j$4DmsWLv!G!i)MC_(s7xW#fm`90iR!-(Zx{XN{l
z6S#lzJ={x9;J)X3xNkdwd(roB7oNaf_C4I!pTM2_J={}H;J)U2xUW2cJM(+ES*g)a
zSjqQ=g#me1e8XMawF(2>+n`t&{`oDNy}|G#8FqqVVc2xc!mykSYe2CuJpG-8VJ;a~
zgJNMAWKCiU86vSdoNQ%CAwyRz!)_}>0vWc%D%osh`0Q6;_**Q)@2t7lL55IlE*`Wp
z{D}+!P{P-;@ZlC<M3o*wKJ~|MsKPiS2wjYWR7@AE+bI^OyWJAGj$OR3F7*z=!uu(N
z9!O6+UZ<~gTZL`?9%1Z_;YYfK#ahKW@nu(w!=!(_uP@E+g0OaY?M>mo;7O?0%ZOIT
zSz~N3p94x?LbG&H;Xl#5L3E2Ly5~4i_tXh;>B{vbw*q19*_61yTVkxGt!q6&;$z<<
z@qS%S_;bK5f5PyiPr1rax*IofG}^mTG>ULgFYoYo1}sGcOBZ;aa?s26AXuCHAX*QL
z)57p;gTVo)T|fr<xE2Q5DoSXegnneW#mb=jkZlaqs8Vm<McrkTVX6Aq5Ae~r8#X@1
zSH?E8NLi!sAcTj%z*AH;J>{W@td)zDdyeXLJev3KC#$<n&x(Iup$&%Q_C-C?+(3nn
z{(`yfF>-#N58it=^<1o{#5F=<^y3!qJfIinYUaW+I{#18Si>|tOanG$E%JpaYxI0b
z^m`YW{<eZqe1Ql5lH%;Ff(s4*0<+Z3r(m|Nr9#HE(8!PI!(aw4CV})Hak7U665H<U
z{{m@Q>7QfI4Q@)Rc8CgyY1+?R{68MdId=AtL%gQF<rF;e=+pb!UgUlYoC^be1!%W6
zm{bl<+I$w3`6B}GsXCzy_TxU#UT5Ei8sO>nXlf>7TfrT?ViT7oHgYXOyYUbdxoaCU
z8=cX`PUyyTcXxcF3KQStcKWE@pONg1A=*^r`*=EtJ*NdpXfNN7_WSkmnDME^%J4WD
zt^=igys&mX3sQX4G$#4!C+e4hsO^3ueoP7YZ;pvO`oy>wMpa*WX|%%+hE^=q#VF?P
zaJ>7053*>TEA+rz1d{1`Kx03U-}79Jd<$n+J5DJ)fz+y3`U-@Ym7ZX!^iLtsQt59~
z{$1NB$5^n5w%4BPH$Q}n#U?aB975F^?u$eY;v0~7iG~1b9h7hAcRT*=%0YaHh{v0n
zR$My+=KN4_^p=az$kH0L7w&~B!D`2r2|B(OI6k4UC0j=y0lK9DQ_IbqlW4nus@nwt
z3nc529;yU~h|i3!k735EX_SfAYO%T8k%sj%pqB8f#pE~q2B`QxR<wU|Nrqp?I+mq=
zla=y`xAD7vSa(@c;|;+Sb@yk{1gbmuQLEdKJQrEm|B;0*(q<bpF<J*llQhL3v2MVI
zg`9|Xxk0-dG<+D%sf03|SZrMSd#H`s_CK-U=1al;I*tKgf!9BV``}oz3v|WT<amQ)
zk|^UKGJuWjsfdqr13K+}orY6_-L#!L?I&64*X?*UqNxM&+fOG)yNtV|{UqFnw#OqJ
z|4pmA%c97zKGywx6LAtXv>-uW8o+`;^W>pWABR$Nld--$IT?hpBw)IweS+HZ(pY#$
z;Ec04ojBR$4NgzSe}zkpecuUR_4z=nd7fryvP*xpE%7d|Iz81mb?a`P?4(%k+Nba%
zC$C%WYhc3twwkk*;OJ)iF9jFQBcyMw(#+LWd?2xD%N#VO&3@cAVSzx}_!~TFhn9=g
zcE0y6CRa_w%`MXBfUt~Np0VqhKhz)o-SBMvy<VZ5lgjO*a@Qe!z5K1=DGa}KCBiT7
zHT+_RKgRGO(eQs)AN><LZvA_O=pWqwu$H%Fsadp-?)6>vYkXYJ$3v{gp@|R8KX^w1
zK1G(5f|m6O4~=?pW)m+s;*1M63%c+s&Gz}%=x>E%yfa=@{0HOWQ1dk*(Yv$KJMn=Z
z4u<#?^*-%kP$@;Pw1GiZBECf;)<QG!>Q+%g`$BIB-x>*KV${Ncn4{Rg;1}cZVo^0d
ztNJxwHuI-Jn06`aSl7`|D*@kZ!ogguwq9zaHV{1QF6S$1mv}wzIL<NR%<ExqN+-U<
zfOE;}Q7sH5_&*Q+i4QHtdE@6{ydH@j8f}8ccV)CR2&27GBW@fhJ?y>~XQtgVeWCN)
zoKE};$k4@IfHQn4pP)n<zeCUeJVCowMIk_;v(5E(yqXE!JdhC?{7>{|ywcQOvtP9K
z+2;RhK#fK1`JpEC9R9Cc^mjbMieec5o(uh!mNxpsW3BZ!qolOhUs`^PbIu~C<SZ?B
z7M4~O75Gal%B4weug_V{@Ms7gZjweyvz=HhbQUeF1gBG)T;X@(k}?gvY*ax>QNi4z
zsyP+;RfShx=`0(SdrMVm;l)>F==6!CCs&vHi&CUfqb?~copVW9zTaP3P;|*;@Z}mv
zZe>x?+{%i%MWafj`4y#wP6mz}N9qj3O|Gt(bF1@$k<NmO@@l`+r%RYHAJnTzuJ(g@
zJ~XVTnowHq2Q9^U6GNfAE~#`zL4H|&)r23$=4VvQt1K&8IO!@U{vi9`6(f`>(1lDW
zF006gKH_h-R9$*|QAM%HP0B^ns!Qh;I#Ff@mHMVgdM?I|%gxFyEGo{gDf8#%S5_94
z7n;Q<ij7fYi)wOZQPsQ}6xDnoHd<Cu6y?G~(GxXFC7RdSQgU)~4OlK8lj~2pAh)70
z-=BX|>g<#YQWs82fs(0a^1asHocwCmoL;zQ?fh_2$5u*I8KbzyjWZ-A>ymh%h=iOM
z{a;n?Y{@8M^b*TedB00wl|#QDw&v*Jf*H4IW2oX-Q_oD3ZbCi$(zJ~HDt}RRX?{6c
z+Ps?b(t>=nTX45NlP6dmM*2E`S}Fz3EU7BWFVsbvd@h5}l)wIOdi)N20M)!`F;We_
zxCEV=W8x-L59oNnB`B3uI;=I|3Inb-VDI@)8%=$qd9;*5wyB?yzFE#@@>I|IBfYk6
zl)JqA$}7`E<H{{9FZGWqD3EkJa6-3d%HziI?$P_s4D^#^H0zNtcm5RV;)~~1mg0U3
zH~`Q`CH(xNY^z7HA10?TEsF`Fu<ZP*>LOqMqKX=SazROcmGh#DguvbmPj56PTv#-x
z=9Y`AOK&O9FRSjx{e=)JW#re~QsRdpV5sx_!jmPZb6(LrXMSN}RZ(@d^Zc^1g`^o*
zK}BJa;L5G6swe<MMOF1pg|n$S>`t^v=@!;X$t|xaFJk%W1r@C%YeO72u2*vM359(D
zj8(U(?lsc5aWi1KW>#ENR4{*1RmHqy5y33H7(~?1zbkKI(JiIrF`6+a)z=i=R#Q~%
z&*eZc(ixLX4Dvrp6DeSL;Fx3?bWTW?{+{iu`0-=V0oes=?}|kD4ef@yNF*7JybTal
zD&^I~Nx<etEnqgnEANR!S^%9`)ReG(;KG&-|Ko5a_OEgXv478huABEL_=x{G5*dnV
zqianhk_y=HcqGFAP~W;X5?KwndR-*a3YfP(66pe5@mKKmL%gRW5$;pC@S4slz|>9P
z18jXU5=p?zf2~cCh>H-Pn&5x5?|2FF02^M8MD_rtz8Z-*_+r{F=nvQdm<QPU2J!{C
zVoxNp9<cH~=m}U0*hV^F7vO3@2g+eRU=E<84fz6e0<Hl}1#AFZ4X6P&07@r;?_-n)
zV8ek(q>}hS@B>x?wgA=wwgIjH>;ha3=s-EQ0wx1yV}-*7d3jvv0ABkw(gj@c4fp|9
z13IwcvK}xQumR8o*b0~f*alb$C>@SOmH|2eR|C2LHvnb>wgTn>YJio166+UmDBudf
zRKV4M4M?xG6Xgcn)rEWj^1}YFxNNsyBiR-@Y-c47NLXPT;2?cCe#wDI<RYC#+<WJj
zq$GpCYYFmzwjvF5Od2?JihRm~gj(t9VdF0vopc@rP$sYV+p~Ql#R2iIGhI@=o~|T~
zK^SGNM!HW8!YakK-@7FDK*z*^LnjV&P8^s#abT)@;MmCnTL%ONEbkxccdz}v_<Q0+
zyjh4h^ub7EFxrfL+aG$1=NXtC6HVx|67i(pL?R>L(Aqonc#{V<To^##pgzyy=Xf|0
zIgWhUeShp-?h0FsV8;6#zdh(np90!f_K4>mnCcnm3iMxY<^W8%7%vHWtXLh1oXqr|
z?GaDUzh~g;*z^XA*eq9TJ?iz#GS5a|-bEgirI|0T-6x=L7bBncq(^(tS6)oEgnru)
zFB^x@Y>b!PN4y^8F%n5XieEBp$2P>1CR*wgWx;|7+3$_Nk72mj2L6Fx3v7(|j$@SH
zJusR4{gw-Eme(=-@^J9@T<FzHTXCb9txW|?5+)lcYw+j5cD+M_wT(gk<bmsL*2-Zz
z+!;)GG7{kyPYl0VC$+J4La~hZHS)9?HgO;8v+`dnFUA`Q{+wqckw+Ns;eS0|A^02q
z9*KMd8*6XrBVLbsWqEs`=du?gks~qXi1H3}w^8b~4f6B0=wpFB`A@yI8x&uBOrbO0
zqmZw_*1sAfKia2zttZACV}~u=23v@D_Qk!XcS65ryur|O=nm8uo_npY9=3pNgYokr
z-g?CAAi=muXSptGjzr#sKIX-A7K5JsN+j|;>DG3nw~Jo$MS0I6-k#P-<SmOl(q936
z^_!6hK=dO0V-QyLLO%xjieBi0A$M6X^pT*iSR9EsDINL}P`(HBg!V{;i(lq71`o49
z&xV}q**E`<{H;X1g>Oe9{Y_zjOlK|Vl|AT5pbrOQ3Fswwwr@Z^qtCks4)vh7$KT7e
zaWVeqh`%0u7#@T^raha5_!~f<hVrmWkGXo$6=wa!cDWMqTHlRC9>o}HpV=dxumx>#
zR$IViSuU?2p7JSHexRHEg+B{@qvMo_S1aOK?HX<ITJ+02EfTqgf>55q)>!iQ3c^<-
zoZZB{nD384U)2l!80af{p>rHu20G7)N}hpE0;WF_bjn{q0@6nry9YW097cs*^da04
zjzor0fvDX@Kc(I6VW+9rV({nP9*I;L@(5^OKZoz7K~5a|t$mH9d}3`l`B|RZAZHKS
z-z7}<-)LvY5O2l)6U%Eb(qGmKeI)3$z0f_NFYJXr3-rof=zh>kKp(~oKwkou$8v+e
zC;d@Feoy)fp!cTFKMek!?SSRln?82<n$*YIf2q%K`k<iqv?pUg&*>#!KF}SDV(o(m
za*n~rEkXSh;hF8?zPQYRL+_2x9Ow+$GY2Lw@23n*y{Er2aBN^e)<9R;z_DmAsA=K1
z!Fz)bm!R7j_@XNy$7u4DCay^Q(YtD;_)d}ES4}D(&cJUfyZguZW#Csq<U%kK5!bCi
zlJS(y5YB|qDWnBNIINutm_y{JK&G1sueQTa1u?>^dkV;3&7#glliCF0IgGOo<$W4p
z29Z1z4!&R~Wirl1C=p~vI!NR;AZH=fVMJbnTpU7|juN>}MhM_vDBw0AUz_O^fkH(^
zej@QH@fQM-cJaoG4|%9I8`P0TLBUVWh>`fWYEf3xV6sG+{sB-r2R}%c{!GM;AG7^2
zk!ko*(QomyeP$qsh<t_HIQ+z=OOM1iLA3j)AR)8T5w;upq!hh_$Qaz1FHQ<24{lEY
zit7Ln=10Vd#6K*>4+{ciNjyu2NAa7;2VRl*{merNpV#xoiw}wT(OHmQz^?{Baiu{8
z$`<X02jmt3`8C;=;3kY2`Mv=XWjhLlnu;r0zNa!0FFr6=(?}SO+Z0lH>AI)miDygp
zuwd5Ub_2tBMP;4DAYOc65wByR@GD%enW^;68!&l)-pxBZ{w9j6=ZzO1SPlD`sSti1
z{KT~z_Zmo0?0A@j8_$x6s|~zl7jZtplcfw=(P}0jj53EX<^lY6;J1s)@IVq<4SiK<
zCbYU5u$2h^`vyaV8Cea4FC~gA68~H+>T%2#V3f);aZ)pW(W;vO3iX-95I1g@0E)|K
zYkcOAAuc{JrSZgA1uRx^W#G=9m`Chd{C>m-UXgg_i%;3S@#4eZ@nh?d1|ZQwKyl3l
z!oDIx?gK*I9s-p3g}hCC;1!AAcQ^9&K^zS>b;ulm8^sdICc^z&hTcSk$6bct$F&Km
z6TdU?%VdcRgh;A7naHI;P6zZdouA@H0Yxw#67^!pbh14Qgw^JvG)}o_%AX*@)<|hw
z^N>zK?iliG#yJNF<NSe$3kVB~wJ6QPjq1!}I@LzV<79gVNHiV(Ge}C~vWfII5T^4T
zk+aYS7;=ztE&;+gUlYj&!gS~!OJ%q*of@XY=FN~}WNR`I{s&U&O(3sPEH9%C&^2KS
zJPIoO0j3=y@;H7(cw9_s#E-2*%pzzY?;w?z-e*VSiTlg%ku(>-U-99;*MCdkza{YB
z68LWk{C~RyTxYq2f(>Wtu+4yZ!*twfgbOCzOxxcRzDv)>6289&K7d0$F1nh$a)!7B
z<jBIy6LJZ7rnqD1c`{Eton_!o19I%<r5WKI#CUPU<Mqw)NCn}lI2VENGzc9hxd<dW
z_Pn~{Tm(^9XM_=OilAa0%YbwbtQW>n{aT+VAckylwHW#3fXd4pYdJRaS~18aph;it
z&~a1m4iO5k)_`XEI6x`#(;(~oI4>!1BRQ$Vz$X~tM)dzU=rYQ`XF|U;>h&=LK4-uk
z27JqapBnJ60sEh!=WmDsFEHTc2ApKT>kL?I!1)He$AG^y;9~}S&VV}%_?7`bHQ-?b
z_8)4bZ@>!-c)0;58Spv-78`KB0q-&3FAeyZ0iQGA4g<brz)uZ$*ns_wemBH`7Z~tz
z15PsFbp|Xp;Cut#W58b;@G%2EXTYBIG&v(<oHO~_IW^_}8t0f%V@F+jaawwfK+{rm
zf@Fjil;z`G(fp#SYMdr=jvjT{sML$c=&YA6@mJMU`xjm6O!nvBQhmuanZ8W-jLh88
zm!)1(T#Ehf6lXHej}$r))91S`OOKv{$#wOjdH(!4fc`2SmY7f2xh<-yltz_T_=`rl
zCwec2Xal>Yyk^v#n$ohui?PQ)idad0b%`{pa8Wtr>Cj)LhZqXSAh`&uDk{ro1mmf)
z%rA}N(F19ezi1)u(7P02;*^Xus;DHlxGH~MQEo{g;+hzQ78K+bEi5Ri#7=bnoU$Ta
zMs9vpRsJGfhWUJJ0U{#3{CTAXh*yCma2qAk<w+cA)ST*SX;c9=w~NaCJ?rUzil2pu
zw9%%08XM;V22b4}n08YTaWn0?iT?x-v{h#OnQ<;)#^eUjNyKuQ_Q)iq;lUc;ALkN5
zenViAz`Ace<9iT<ZPW@g{Z(-;!Mw^K%o13)eK+l>K^6-zwB=^}t~i$n>M{sJ4SP?(
zyqob&yL%mIwDo3ubNp}^2HG4iz#^`m<##ivwAE&O(>_m-b^g0`5FMXPJ`-{xMcZ%U
zru}a);<p*`sXxok)ZZXWw;OabzB%5k<}?zQIo=q<ax}h~|2ptct{LAPzg7%lY>dwg
z#Krg=)6MkF`1gXwGBJudI^O4S!j5a04x)EMTohOFWQ|{Ij1#p+d{=ZJ)}t8N60>Ad
zR`!T*`Zo=RgK>AY9?!g+@lE{KJ>r}Gi6^9UnB%m`-&6n9MtrmUvO_Mxnr*~?%5($-
zRiE7(j{>a0o%uK8OAopPU3yRlTyeEtJ=1><f!6qAf8`SFV~yGQ^5`0mF2$@r9m9Sm
z9<1?sEP&Ux-wPn+nH8&}qOBN@0=$m<K>W;VGX1%hCv`5T4x;y->A#N<Hgz+;>F;Sq
z{GR+h@%@NP{mu9thQHGBFUCKBxXhv%-}HC7jQD27n8jeGZ?>bZ9_e?j%L4^|=Q;tR
z{YA&R`;RH-I|NwumkfW?T)17&K+9#uGhu&(TjQ_)Yo6d<|5rVdsi!HzjA!DBh)kN9
z{{|y|gAw0#s>!KCXDliWGSW}&K{w+s?Lnumyi7<lW2P+vjN8>u>kneD%o1S6GfBjI
zbP*Bcd|u~JPS@et08{WY^KaTO_V3=(FKN}|{iw$P+B1EftM_|6ayd56bwXe~G&aXs
zYhXMyF~`qXx;;A2iKX|8&TnGr{iE}sSo#3T9G_z836eSP#L`ca%<&|aesXl46H7lO
zYX4*DiIQp8W9g?#roE1(4~)(eW9hPF+Sgb*TDfUAEdlY;Aj!0c7BWitG}t$nl_Z|y
zu^4LFFKb{tmQPK)6iYu{GVM()eTZb*ky!c}unT!sl6a2iEY$4xme6?Y@0k7BLWcgs
zB(tB!(&-JxUTm*+>8nUA)f|UTEQj+1Bj|r{#Y7on(0R<pdQCRyJVs-^ZZzosuePg?
zk)){NeHS8ugxrT9aKvQ6e@N7Id-ijp#@^fOy9~FtH%<@7aS>`~XL|SgZf1L!p4r<q
z{BZ^&3MQV$PeQ`RpwSrhKtvED7eh2~3E>}&0!HI^@FN$E2_!21-m9vf?&|J4HcAD0
zr+@wWb@i+FUcIWWdPmD-lBJ_?THcZ@r}<>Jb>Yhn|I&xg@Q>XVEl)|7uPB_BjU~(X
z6#iKQe^%j-8Td;Ir)65n@>hk^GOT1FL2kE(dOM;nzI_|{v_00tk1sNUQ!cbQHc$S#
zp0pezS;~ClZi|*#B+EgC(=v)=(e<Qd63KE0=&7E1o~NfSZUCox(pFI_+7EtS;k1u5
ziR<?|r*N(Y2<R(H-h0XZX17Jl5|V}fnH{Q6Jl~`)YW}Yp{4ax3&*Q%X_%?jZG3|~%
zG7|r@d(zLf75-}jr;pb}Z_axiWBz-rO`{zH#<PCljN;=M^Jfn-eb%28z1LGrzsovj
z_@!qV&%Q5-FHb%{W%{f?`7?0(*kk&!e<|Gbdppo6Zo^kM(~n&Td?d#&y#qM3HdAQL
z&~C{2h5H=7ruBzDVckUdF^SXqXDZ(>;6oo#iH~!?!*rU@s|EBQ1i!rc(7Nh8{BGbk
zA`;d6CLlpBUr@N-Zz1&?U!$21>Z9U)79;MUf6OJ1KdnWh`kyxR@R!Vg=+95f$6EC7
z%;fQzEx_+!9Qv%}qsLue1J3XDX@@z}=UFP}fsgR7^x?$<{BO)ZP0MF*A^v+T=g~zz
zttyhe!1L;VqyVS&>3Q_`6yT3BewEc#?@MD4Dz6mKUv@e1&&Cy30w+B@v)|$Sf5At5
zH)i^)tW862A7ES&CH_1Od>7=fspLRuC#d|afc~Grsa?*VahPs~kC>-MfJ%CD#?bRy
zfm1!5gOZ-t5#XLD75(Nw;uKF)`HjNQebnJ|T1!WTJ{c3Au_cF2t>^e?S)aNhU;bYQ
z&URXJn3C4MQu!0%+u?6N;PAPgPk&v}>wQKrPcGwEN<Q5ghtK~YVe4bSsol?BD{%!|
z-%#`~%u4(jRsZdHP@bOL4tzx4ocKEcPIA`!r?mZk5jefq$h#$fTGL16Jn&th*KsDy
zlgo?1*?vFa@Hwr$qp|~&BE+XV;qW=FWv6lycwRm415WkP`@3{|Jybyd2=Ec+m_D2b
zzB9)U|D5??(00GWH)ua_GoZZs)EP%SbDyMFuVsA%IK7uMF7bD&3*T0_-an?>_g9L~
zqI$08b2SVz@3|$$5ud2%>Uy38PX6TFcNV$+NBM{aA&UO2QP1a?9(JVso3^9h1E>01
zO<Dg7iqGs0*^X7Co?V6S+>rFNW|hjrz`1=7IDF1))bVRO6p8Q)cS@XBO#r`5;d)=)
zYf2FJ0H=1>`~9wFZrC0HJWoDP0pGR#`b?bvlLGob5<TQL^4KDC*8b-|1@zZI@u>cb
zWryi$PXv{Rfalfc2f(R5Pdu{7bo%gY0sYI2qksRT!=%2t;V%l`yjSAyQ8+C+Ci|j9
z8dOg45i4R)8MYVIzSk=J7T`qRRrFfl0>%|l;?D!XxgC2BqgrpDQS^GBsJ`X5ieB%V
z)buyJRr1-~C;2?7ZfFCi`iv+)tlWupFXNi)@bB*m@YfXo?)xPF=hZDcpfFVb>c=Gx
z_2zQ3!p|*AoYwMF833pH>{RlpN!a=laFVCqx2f&rTZ+%7vI|}-jC;-lC%rnW;$iI%
zUt#`=Y2wdi*Qn<jau@-=lkEDD#RN4Lr!jpB;*$fwIj;Dwq}P7n?gIJ;fwMll4&Q&H
zlEaUIk1&!xyi~yFAB-bTGxX|e%tw<vPb>LoeYg`iJ@-t<;RonF2cH8@_1`q=e_ru9
zr{qKNF_rC5Y?6=l^dg_@!{xw<kM+33Ppb>Fz^OiGl|J(x0$jV3aYdB)69K1szHr=O
zl=oEN*WU~1_gtsssq{_H!yN@q`an5hxFDXOvH=_>lPM1s;Lj^Qrxl+Ob?3he=--C<
z{5<{>z^NZJ_3FKf-nrgM)#*dPNe<5YB~E*^sQgIb7n%;A>%AziGat0Cp%2%iN%HFR
zF2<oJMx4+9PR~83;-zisxnC{d^TPuCWySxD(GOm6Lw<d3D!}Unc((w*hjGM>11X>5
zEG5hOroy{7NSyW-QhAi|?0n!8z)22chaIM)Jy=|5H?wGR`U>9MJU5s6?mA}SOLhvu
zZgLZu%tBLHXgUj(Gf*juWFI)5A)3e_**ayPU$(@7`NMl>=Y@y{OQIqSVTfa1e||=Y
zAmk6&d{ytFH-(UW;h9|S`hIatrYM-Vr<ku!Sb<IUt~I<kow`T~B-}T;Pd)eD5?kEk
z4eiZdKk5&VJf&#~am!py%vZd+5J>RQ!yfSns|7^ABbsgOe~+vjNQhP_R@&iGqb*uU
z#}J9eV4Z5od%atuQ|0LrIOh-vS_~jR<O_V9?`>FeF+gk3>1^PZ#2KeANRbM$zczcw
z74E`5R94np{Z_4dc>jLSt&95X-g%cONcJ&{RvL)<$Sh%reYY>n9;(b)v~oZw3|dec
z69TCY!c(H{udeiu3%}P3d)0$Y$2l-x**kais2Creh}*3oPlY{sfTiROI!8F+id$_3
z@lCiY^CdLeZNFVxI56S)eyig3?eSh?wH0=d)MO3GUtD*sRu@*chlL!=>iMEHS{j{<
zSrt5gdN0;b1b)w5BT6uy!QzpiiT4Dr6&(FWr_uy&Q>F&d?=^ycRB5_i0|$4`c54T{
zi0UIMVI&SCvq}fq6fz?4!gk~aX&4z9BEN6ewq8+rI?TKm;#hnsmBK>^WWi~|UkzAw
ztffZtB&ylemC$m~EmuE+>P{x#a%r$!{EF#1llj|Iy5W}^@RN<Gzm_KIhEcFC@Oq6l
zGEkT=c+s>arBXV>gU*iN*1Z0d%w<yvAp5{2k5Kn6$_yl8r4z17r6q&7WlhwS{m!);
zQ3T`d4*HQ<#d;lYxg|wN3qTxewX23u{Pn<WYe6;W^~G`{Xq#17IOz2!oBfTh4?Tl+
z9TKO4mcPpO>ZVzb+V-Wi+%9plCcjl{te%7-2DQMmg|*@%by5c@gj$UaqXN7nN&8&8
zG{?Q6XopC9R9W!Ew92)E-?q>PQI2!&xVo&>h+1Xdsnmt*dSwv|Exf=ibKJr*-{kdY
z0vH;t@S3Q=cfkVF(ts$u{zMbb$Y>L3CTs<w(+DzpSwA9`DUonTzkfj`OLkS~OPLG9
z9%UpU`MN=_vMRHlnG3rcqKgG?KC)DCj-~&HK!b!#?J{}IxlZeFm$GrW>*%l|w_0tY
zt;k))+QM{FrK&*{1at+SeF9m8mcw2Lm-?cGyL-XXfLK*JZnF*dTN(q}L@F+^1eG|6
zjtYrWe7Z1&sDZpsANq<u0W`f0eYrE=nV50sJCpd|F1a|4&y?x6P1@3d2d^Dm@WzBa
z)@@(`oFMHd8*Z*QL<1}3G6tmLg}B8^Hp#!T+Va=yq3l<vVW981O@jU#%M4M+c3YrF
z?_l*{-;af3Bpxb_CsUB^Lp~THC-EjEKPByCOATt;N6nj224d>^z8qagb7{XQ6)eXx
zz5cX}Jk~4A$}ZtTy(t)@J)KDm=62YG+EHaWr{5z-Y>&&{BUIgEeN5DC;esKwYxpb}
zyJ3QrI%Of#;Zpt39K}y{UduS_-3pA2{Jk=vbk8_f%IO@u34#ACQ8^6f-wIa604qXc
z4h4_M<rLp!YcEDB1(wj2Ewyvj69&V+T3{{d5@H0y3&%ZNtl}=hLEr4!Nv`=)6{(<2
z%CF$<?Q**7<zz@XqOGn4-lW#hXaXH-*B)anf;xm76cABtWHh~MyMkwz;~>zjoft~o
zjQ;E?3jbi^i2U}+z>mW=PLyWYt392xSML7E4_vzEnb_M48!ZZJZVP+umK)%OsY+fj
zg=i-1`9_Cx=@`>S)S66N<&p?(DMIu*kRDvzdSfYAvtf_qm7<89AN=of2uUKA#-e0l
zkB-N(c>{tZeb=3uAsjZQH^YrL+cJ>m7AdM@EZ|tSvH^8->*Q$EW#u?-XkRRb{~<ET
z^Sd0I=r|*Su8GLMYi7k?^?PjYWdxL~18Oa5C@2F$T!Dm=xEl5<)e}Q!d;&&rLOR>E
zx`0cMs~ns1RLDJ5j)_sjsn9l^t`BO6{+qpd1B#d(Yza73J^F*nMhK5URA7=6Cm~&2
z+ES4;4!t+}HPaPI=P{#mXl}vxC1XTHi-|5mcSk&a8{_Uj9W3TXQeM9tch_d{Ct6V`
zj>AJ>eSEXA6q%#J;b4hIzHy^WqkljWz*bUwOFYjIVa&^U^b}V_6V);hGELF1lK+X@
zD%~+9dMs{pan~U5qoxnNY57q+;7pHX%kZQK%hwt(q*TL<+L%1a?*Z+Jq^tAcvQ*I2
z9rS#$7WDdq1_l^8Uc_(<u%<+sRd=Y(!y%W_0us$$#Ngu-!!WBJro!QQ8Y^)4m2E`y
z&Ctp5p;6fE0aTX7LzwR&xyqrYiZb8}X|R+xP|b}=1fmiONviZ5J-m{05~F1cgVRp9
zX7;0IJh`u=TS>=QN)W>fGulhcl`0FDa1TL7pyVLZ-l4=ggeVE=9ks^E2%R-eO%B99
zL^*P!dJT_yVGQtMg-Og9%+Rb2l7B|>-e7_(+EBLPjqH<EyU-|(1R}6|B%TOb@WRRE
z_8>Z*WnuJfG+`74m+amWQ7?%(Wdu~Oj1UWASfLVeij)3^XyT1!Z0`lV$gLSuF<MR~
zmXjE)=9ucmx->cPpaEdY1oB-w-e60^B!s0DlAF@Bg;?ucTAZ@l?91X2VP&<iCfjOW
zl(kL^12!2vzygOeMVJXIm?<g<_T=ncB6TVam4c`{SZ(S~M^8>AoAu3zOK7|^ev$DG
zw3s}v@-ZzJEf>~BO84M3$R%O&3v(l_3Js?tGTsvXS33xIL+#p=(~^xQ<x!1818mq@
ziEXN6bH@_T(+Nm*88eW~p5h8Qt4fn=7@3C6JQFbHKWP+4Gd6~~s6NA!Q9A6;6F&0O
zZeXy`g#m}X#)@yW9eOq<$<9pjFnozJq+-O}Zyrx~C_e8iLwK~lPSYAuh@E%=p0qZW
zrqexGDOg2N1#4*yn*A-r5|}?4jxbwrb7mOl&9U3tvW-<vtfAF0iL8Br>>Ej}c#{t!
z>c^GUQ=ODf#KZ-mF3k^T{8fBY?8TuGJPA&{C9>P<ISjZdIB_{?6OZ!auK3o~Q_w>U
zzoJ2hIx>8L8Xp}a=6E7c)cLs+1{d>V$#K{+vkJ#CLF#7b^l27A>3WMtVx)o;Qt^a%
zN=QR3j7~MW5QrY-sxJiSi<{lV{G*-tXu$qB_FL7)2J)Y`l2chUlV|aRW3k#}!zs2c
NX;yZGF2@6b^<S|_o~r-=

literal 0
HcmV?d00001

diff --git a/tests/Grid_cshift.cc b/tests/Grid_cshift.cc
index d0c83e30..76125977 100644
--- a/tests/Grid_cshift.cc
+++ b/tests/Grid_cshift.cc
@@ -34,6 +34,8 @@ int main (int argc, char ** argv)
     U=lex;
   }
 
+
+
   TComplex cm;
   
   for(int dir=0;dir<4;dir++){
diff --git a/tests/Grid_gamma b/tests/Grid_gamma
new file mode 100755
index 0000000000000000000000000000000000000000..641d5c8872bf525f0e5288531be47faab67cb9e5
GIT binary patch
literal 72670
zcmeFa3w%`7wLgCHf~d$DD#fU%106LH#6(dOiaLP+XK(^3M2Q+rLJ~->A(<u<jE_Pz
zp>jH;O?&ArwzSP{?Y&;5m22%SphChc0R@9kd?4|Go*^Rf1<@k&`+oP?XXZ>^V*Bg=
z|GB@92j;A^)?Rz<wfA0Y?Z<g!`3&#O*q9i_I%1XU6k<K+xCO~PD@yM>0+qCM<ut{u
zT&RpyhJliRpIf95npYO=m7*<}DEZ<5<MB(6H3{jlZgFA3u_hTMn1`hyA<K&#Zh2|K
zo>xuOf{Nl41m>fztp@}OSST(mxQ5E$Xmyy-%18B>uUF>lmH8~FNc}Bn)sN!%%a!T5
zv2I|*-2_UOI7gr?UdcJ+FX6R&m&xGX<rWtf94qBn(8~8c<fERadRWz*C(9eGUEH!f
z3tIhfc}4lct0!MxQF2*Dd1Y<gWp(LSUv~B6Ni|iIt`PO6eDcnkbE8m+VyIsde$;t1
zezXCmzg#kU`729vhTrzm<!k1wy5Udn<+-0=+GPB!e$$H_g+*;yF{E9MUj}|1_nrRy
zuZy4F@#7_-jt?Ez|EB3?_eXEu`ZH(pjF-Hl*Nm91lqLMfZ^rsI+~v9Bf51NAaSxRF
z)&MEPVPpf*??*)jqE8xxfBqo!1XOGwIczuQ0Qy`w2>sJR^#AWc=;sfj&*y{iuSdB9
z={a=}IS&pZ=gvX+zdZ>3%0cQ|Itc&BLFjV_k+W=&dX){*zh4Z}U+)h>_Y9&>(;)m?
z2Whuo55j-`AoSY?DR<u>^*ufa{UjW4pnmTdM9=a;+WF2w+H>L{bbS#04-KNvn}g_o
z<{<LN4<i55LHPeNNV%^M(he^V(l4V1;lE=L{%Z#*_o_kk-wXZ6Dd#HgcM8!WijpVJ
ziZT!7lHU4mp`ciFFXWS6__j%}7paQ!q@+6!nRJVP80e$Gue@i{os$22DZfj~$&vKU
zvR*x7+>&B~Wl~PdJB5Pa03VNE<41dIIopkM<X9>7sh9fH%cNw`$={shmJ|svF4g!;
ziYH9qSah{gP_TGORb@eqzo^<@P@ojdURp4(ba8o&zqERKMNv&nX^m3g$;~b(DXo^|
z-0bNURh6Z=MGGrR&9s423Vi;u>e8Z;f;&p@Dk#R4AMCfxnO`#*CHT@(3n;Y4UtM+A
zjCnWAFY*-6fHK8(MFqav>e7Ox<<<V$q6+4?VZQ&WV*g#fQm~|5iEMY4my}i(lvkGf
zXDk&W%P0FvOYacKe1F<C1qCR%>du0ND7>hmqN>>Bs=8`XMQu%)#dfuwt+=ShPcCE!
zy$WiT7A-*$HKo;lPcf)z#brg+1^()ya(@l<m@yBuD;E~klowmvp4@`m)PfoFuP(^B
zVg8i`LUl4#7gd&2EkRwEmKT>^1NDXCs4`j><z=9ymy}ncbxbY$=Ey*f#ijnGMHRBL
z{`B&y8q}?(bg~Cckd<c9uSQLaDoVqIn(1jXa&9ONH!fSU=9=QFCBCBSQl^Nu>suZ*
zTm%&@sTpP^QRaZ{a8+?dX;JkIRP>4h5A?#%Z}oKf)#ZyTt5BBgrRtg)^U7tR>84mT
zba~}sv$+S7FYIClD#G?+Yl=eV`bsO8RF%-4a&MXgQ-a;p6)h}Zn#RsArrj3MV9?Zy
zsw}fw2>Y!pC@%6BmkAwAJG>&bV17aA5}*IBg6dKq{Gh73XmKg?EUv37@Re5AR8<yL
zlp}d*s?1h%XOR#6&+f~;2|8V6ugb#OMT6G^Ip+H(7gyE#`{#wd4U!ki+){bx5|JgU
zvR9-P6f9j*K0t-xW>qvCX}~44*Q#8&HxeTDlXZpcnW`&&wf-5jhumCL=xg=O0BWH>
zY6jG5z8}pkowK5}a<RXxptQQWsyb|o(WZ@Vr{$DX<SbPXww9O?tGH-k%_60uYH<aL
zi;Ig3YM79w6x>*OXL)4_qDxKfl2RrXn>kQ9bXEbQ8G_uRqVfvG2eb1pQeb${L&hDL
zZC$xvsVQG9?4hRAuPiBDB0;h5E(LDrFJDqxwWy@%F0vIB`^&2;m4!vccOWnnmy&%+
z)zZ=krJ&}nB@3%6SUS~_VRlJTd8JZb<EyPK_A9fdPcOJ~QmQh`>zOva;EG9COu8z3
zoi^FL%*zAI<VhJ(SX#O=JufeFnx`Ob(v?vuQh>~rmS#`nC=iR41%EvAu(0^>6=@hi
zke+~_$v<34#HyCO{5M>rGUYU2iGafZPZw#4IG+KW6#mSACKiJg%N~r0);vDv;=9gJ
zkng7$w|OvqgDt)G-15^Dl=fgaJytm^(^J8}`ltVfWz!kTACOL&CV!lbe;6d1X#k48
zJO!0LO<6Db%X)tSJW<&!@fRMu16RY8JrZ9b_!E@<5-$?`@ydG=_ow^}OWIiFGvGXW
z4;Bi@wH({XI;?pF*|<Kl4r^XOI@fI0;eMx3T$0W;opo6A0@A;05dp3F3+Y$c=+-=i
z^lNN%Yra7GwKlpne<3~HMo;Q06d3Cyrz8S*+UTY=39^!GqtlYCBh^MnH-?XN8=Yla
zhucPHn_7ovqubj#$3}<YgpWKM{q!(ZQ3`Ez7-IM+v(ZO{sfyyW(a*He>umIqHhR5{
zuG;7;ZFJhLbu`=Pqbwre8XJ9zjoxCTpJk&zY@?61(bwDPV{G&dHu~8%daI3|WTSW3
z=;zqzoi;jV!{KAUjecI3swiDHy0w?ZWWz>3-^Sl#qffQbdu{X!Y;@(VLV?jnxJO_e
zi8i{^A_6*W^l>(Nl8rvzMt9oi-?7n?ZS;$5^i&)DVjDf(M!&>HciZU6Ho9h`r`YH@
zHu|MDdY+9w!A38%(I?vIWj6YCHoDJ7zsyFjv(YEn==C=G<u>|C8=Z07I+|_tG>Zth
z#zw!wMsKmvxhG;BTEhp4S|H)7-A+YY(dthy8jzwjY){-~ahNc4)p)RkE^^_lUg*XZ
z@iHa`yFxfFI+r*@Ua(W(-y$AMyj9?*6OSXlUf{9B8QOv^0{>zcaE7#Cv%o(l&QKPt
z7x-b~G|!+<;BOO8Bwi@+*N8Ke1#<-cJaLAwpj+U(iJw6{Rp48RGh_vw0^dZOp(^MQ
z_~XPGqJoORA0f`r6zn+(!bQI$&X5%B68JBPGZY0o1^zSQTv`TO1^y%A3_Zd10>7U)
zLr$<o;CB=MHt}YGR}*LG3DyhzUx+j01bqTuNSvW2SSavYh%>|la|Aw*I73U&E%4dI
z8B&6&0-r{lp(N-O_!Qy{Awh@0uOQA45mW?zDRG8|V9%GV|9Ij~;#~qimpDT~uv6gQ
zB0ipYtH4huPS+2v7kDgj7x5N>f6)P)At2Z+@Q;bp^@H^SKTMo1AM^?QZQ{wq3kCif
z@f6}Y0)L)3T|4L&_-^78h^GpCD{;DV&?)du#ObO*hrk~vP8SU-0)K=!T{GBoLi9gz
zx@53R;J+kJR}6Ln&)MR7uTU}mfUH}%cd?o1rVa4IJW>CgE$$DIz9O7XRHPd}1x;Ph
zs*XxH`XaotJ>gRT9$VZ`LXZ|_EDJNfX)?YO#aL)D<^`_<=b}mKg^fZE{WoVz9m`J$
zmrqpc$NJ}NxtHmOP!&^uqB5QJPdnn#_ua1PZ)*+j_T=Px+FMgw-B7lzbqhgz*ST)R
z_$wGcU8G)^r6?N;-Y!&b)Go^d&>DJ2YWkr&V{hBGfeJD3e0tj!CWX4x3+qYSK+NKA
z>x~1yrf=2S4qvaubZYy0Yms>q8MPJ9BeSOOH1oeL^VeUiKjXwSchrr@X)HsB-L^Af
zDM&G&Y-9bkhHKNefx&-4ScTCyZs|Dy?es&T&f9|bAd7mT609-djhPEFZ_2zeGk3mr
z-v`E-isIRf1RmGxPq|7!(DWx<d4x?8Z!&S?r)w3(cw-f^=<S}sHJ-qNL`{E1OF5qQ
zx~6}rwS5`O-cD(BxtY7Y(Uk+p!=L2M_)LAk(2##B`b5*Wh4Zg9x%IU!O_Uk`sHU4a
zwUm!F{UwU^qnv}rcyxl+WTu<EPq~tjXES5(rdXVegE~&LEe)3^dL~?-c0|+vrnUXe
z(#;`tb6UEM@@9OZKJY#2*1{6>9h$z*CD;Iygi=|*MyoWsn8#9_%Dq6jRPLy;;%`(=
z@|l^|x>^yXHME{+kEYc^Tdir86n@Z9P^fT_Mpr6W)ln#dS)OL2;g;j|)Wl5+8dC%0
zL0S&3cQ(2TNn)-#To7_tVERfz<Y^`Y4${`(8hIWjNken5$3>vgwE-iC`dFjuDL_<l
z1E{cK2RmO#EABDws#X*TOJYuDMiQ@$qHGiC#NZ^W(Zx$BLJ|w@Aw=F@K#><T1$m)}
zgW|X=su)RT0UZ_-3gESoI>!_YlUfg)rqv>0vxM~$`XnrrFh@eSgsBobC3HxrNZ50l
zDZfj?P6=B9HFTW;K31H1u}cMAf62JEJffn^(jsBAg!K~oBrKFLM?$xRsS-LRbV#U3
z*pnz#l(196RteWj*dk%Gg!K~oBrKFLM?$xRsS-K?y{IB_EwEPfr59BfbR7=zp{BoP
zj4m}BB$-?$DF!63eq3Uol$Z?Z5r!_1(n6b@=yR`r(s=7P;kI*tiQTIJvMl1%LEvGL
zEmCA+L}U|4-mp^3B|}n#p~7C~F=GkJ{6uvm+}wf!&qAnT{{xmu*eu}&z(B%dZBAN#
zFJ^D?P#XSm0+%w%`7np0&V>@@Na&U@RYIqP4ha=N;a(!NAUdFbX<{tJIB*MuIvQR3
zkx1d95a1mW7li@eAaPMB@P{QX3I|?B+_B14C*W$=N&(lp)(F0}E>pSHE>k&#B%5-w
zem#=ZC1Iz8trD)6utmaV3F`?Lw4#+)xfEP~qTbvAWVI{aL^{>vS3gnnU;9MO+qX!a
zGVV2pde&N($OX8PxdDYVz%?QhL^<uETtkU!ks=<JP$&vf8$>3Ea+|qLt+GsHV69rE
zhz<!mCEPDEK~$l*U<Ge(bd@kQc^@j*iORLHa-FE$YFC%YhA1LokA%H|T2rGd5pW|z
z;i6_u->tR1o$yH(tYCC2?3h=pMsJr_t4DXptF@z_l~)gq?&Ou~SU37b^Wuro`^}4|
zM!$gzOgPex$@v7>V)2_DpQ$E8%bPm^(H%FlOJ2bBeqIB<1=xjJyu(_&=vbtw%U;4j
zr?nk*E>bIB0(JZ=F`uZ*)=sWxW%6D_r-@7CeO}6a-6j{^E#$s)%5qmm<nj`7U!+_$
z`G8IA8@yJN51q19VOOTM;xhSVQ|LiUDBxcF+9N&Kr|vsg_jcBO)nr<yn%AM8WnO9x
zJDpm?R;PB~$$Ipry5h%JacH47t)VqhYxq1~U2#4LYI37XSOENcJ-Uauun^!a67Qs7
z;LQ?mm3Y0xMWZLLGTkJ3wX22c$r#lI-&&WCd?~A4g#xZ}<p_v)2N+0r<tY)@xGI%#
zts)ZFx_F7AJ0)zDaJ_^r5;jX%FQE@`pt#0TjD(vY6miLlYoZX~R$LQ>0k`6sC=|FA
z*F@pKzbdYo%B^;p%B^zQl@skOm9^Sklxu5uQLe4sMY+JOcK0zq+Fi&+yNhzr?q<1P
z)$RpPigr(j5ZUfiDbJiW=SbDufcADIH^&Djy3T0IG21b!9a7N0w#=+2g-}nEj`%M7
zUja-zlKIqiyWH6R$bvD%|FIQ|IgD}-Cxo?HhfArIQwnxSs7TmD7Bozkgq;$$O33|o
zdo3uPwcvnj44;L}=uOKSL?OT}YY>G2x2!=F3f!^=Q8@6ZR=^0evd<`ID!1BYD!0mI
zSB|Y@RnD>otCB5jEu>qOZI-ww*Jcf@BwJa?g*AwBU=3!uLtBHlX`L$za%ZF+TeK*$
zrk%0kn16y6IDh*_Czgb-`{#1$bLWRxy|%nlh)Dh>g1IsLXeboC-yAOjV~_Np1Iass
zyf^gWeI0oB%~-rPf**YlNZ5!Om=<swTGIx~46JX(xnXaN;e^=kN#OX!U>qkQ4BFcC
z6>4oZTHVSfF_*6Zhga{Okrp!7%NT2>pWql0=P;Z|Qy1(~C$-fk-+i%})7yreu~5vo
z0;$0-aTO(B3rs(u>1|qDkmF7(V{co`B6WF|%ctfg{ORA2ao3Ts*z@i_#}s>*3JI~V
zBQ^LR5UVDed49w^Kab=|x_gY7=SVoue<HQJ*2+-M3?&0(=m}?7$_!bkYk~r1HQ7t;
zI@IJgV<PiSioZw$Z|(TIon}#=gmX?sy1hL!8$xlLk0IHkW_>D{&2}9PK2NA!YuJ_u
zmUwm5?>V#t5<b`<J2SAR`CZctvg_|jSNy|%G<|yaxtLEJG&;J;8izhl-3<4G6&(z9
zs+)U+c;hul*7SDc?GISOPEW%QXJ*4ml<<8l)w@;WQsnMlXS1KX`WJblDDo^8c?r8O
zV$UpJ8?yXcWcdqGz82ci@Z|_~)gmmXyN4T(9R_&FY_eW98Mh6M)5sV|`1r4VYpyn)
z#4RtON&;0mhlM`_8G~2;0A-=_hoB;CR)bSAsL(l^5uz70PYc99EdwMUiN${|1J(k{
z0EsRafl!gKhe-^K5_U@1D&cwwTL3ws!YYS?!?CeD6ZsL_tk@{z1MgsLgIrM#a4R;7
z@_;`q(?z+!S&gAq0H$86U8Y{ETz0)|m2QS^_|HnO29%X<mX)oS&?jM`ggFwrB}|pj
z2`DSg`g`@mMt62}eT8USeMJ$r`ii1#^%X^)LVZn@FaQZvR=MmdG1iN6(Z1GLAj$=9
zjRm4y;MQ0m$^~98#|9_mVk{7HF&2n&Fc##9@`fG@WUJP*fg#c-VWEUM61pW!mCz}n
z15j&k?ZPHrvldttUkH`NI^r;6Cf6b3w~F~{d?A;&`E6M7##0EHssF&t^$})jKxQ_^
zz|1WXW}jqk9hA8>!rV+|OpnzS)1W>(!fFMpBPfoo!m^oewTjgd)Q#z^$Rf20J4+;>
zCq(n0CsGBDjYh#Za8F=FF(=1q7-EjLrtPxg9b)>bcsE4jVs$(RY;(QDwasq)u-3_Y
z0c)pS9OL#m7OBg&M%rA>12X08Ht03wY<7X8F<vuet%~;wSzLp0z0`)^F35t+e5f#G
zTnA)XeP=EvqvhmSa&rGMIeC_x{J%?%uQdkU=hUvPN03lg?7^snHfYt_x}5qh?X)&5
zo+?qdTUtL>M`8S(9T@wiHC=HrJg+?zJ++zm9zy=5w(%p7aN6U}f#)ke@S20e@7k^T
z#|M_Yk5vcCGb+I2(YJd7@h;AJ<^-N__CV8YOta=@)-S_Ad1mI1W=`*R%Jnfeu#7eD
zB2Qpgc)SEnjF(>hePbQualm*2<zPI<SUgN!^~&dBxKfa>`w>$k2j?}s>tx&lP7GJm
z|6A4~JW#8RH{&)-pW$jOWDGuBO_HTw4VjoYdYc{!4_B?&M&j04T`Xr%;d>+q<$sam
z7atvk3ow+g!lev4B+-3(NCNDVuv5ZT3D-;5B4INi8pMmeKTPW>+i+?SflzD>T|zW)
zYv>Y10JnxNQ50}%=n_Q&|Ei(ORB5%#RB4sVu9BFx%PLtBM3jrFSrJ5(3*3qzqFmsN
zAYv$B6%jy$Tm%qN4nnzE?!W<rO~)|7_TK-NSWvMiWP9I>OWEE%yaaQXgq;$$O1NIa
z773drtOqpP+pB+J{ADJjpsB2;7E*y*O)aDYx0+g%0^Dk9Q5x{CYHCv(>@Nr{u-{<Q
zMpnot4WdxO90}bLrb_6P&>^7$C`T=uy$(Ho2{puiOFYF5J${Kgn#|N?=<%zI6*HL&
zC37pfbKt>_s!Qg2GDnSHR(n|1M{(?M%lfSLv8>N6(`kKT{1Vb(exjjZeoivL{89zu
zz~=YO_!a;E)%f*U)cEzro6+M}#le2#m*-LEz~k3?@b(?QCR#K(eyszJ`=#cPU}%AL
zNgTek6-WHvtN#H;FZD+=rgys{@t!v$_aERGGJ}O$3%;Xb0ONq$Ed;h3pFnXe_psE%
z0QRyufL)J_-4B~8UEYQPjE_ke4}uc|*!_>lYS;#_-(9No9l&lCg%7saTp~-ah0H+)
zu>B}AQXkv+WnhCr#?udQDZ@{pj7T{Wx+P4N&?%uqLPf$Jh(J8$`ip|O6!GD~OM6fi
zqHRG{6k!Xhq9|KX6-AyxP&HLr?J`xuT@ss0VumYmD}sn}Z4pG23*3qzqFmsNAk5FI
zar_c;5kN#a2q0#;0|$@+#xHh+Z0|R4Dcd`T!cmx8!c+;J5;`R0Q={agxe|5(it+0t
zn>qo{ZrPh!NVPS!kZx;gQHrgpMQNwd)TTD8U8XjxqO_3}>XeGMO1NIa773drte4Oy
zVIknx9lsRjHAfkW8+!b5MwnY9GslpDha`7|xm7Z^4$52@VeTSx)c9q!hh=>f#|~uK
zv@WZCEbD8L>9jsEehKL?KhaPyzj`vj{F(*hz~=YO`1SwO_%(Ka^!T+7c%Su`{*ZIv
z@vFzmyE(G{s<3Es{Q3a+*B!r-UKwKin(!(Gwi{PMaNqH3`O6%?A`dM@9w5Meo_K`A
zd{SXS7K|QiCEJ4pS(4-sP9q+1NRuSn(*zfQ6n=u?m~r7Uu`kf*8cW|u$#OZk#CI{n
zCtU81@CKv`0kbfXJX$iR3XBB`W~avYAQU(WTpYLG2%#fq=p3X>9FYwC#t*M%re$*e
zj)Hvx^NBscELWkx>LivUuzHEPF^3RB*i38yFK{a14Ll??H;4bZ9R4xkRE%GuQ9a_}
zvJeh+!Mjj9*EK>-=Jv<4+$2Y>^LYn+JN7TQ;eoviuim4z8O}wSYUNH)#<#{`4+9zV
zT~0Om88c(93oVdODY&tl_X0ESM#f$%<15VAiHvwm!1r2AmYTd3H^Wg%mYTN>xp40{
z|FuO~$hHl8B|am@yn79D9&ZtUfMkP04>&a45JJ_;LzLMC&I<0(f1v66WdiSWiv+)`
zP|f?)D|0~qaF%;rIcnYsuij-T_^y@5>vF4k&+<O8!P7@PeZkXSSgO2bPF2Pqje$;@
z-XRNpOA5$;Thluw?RnBVV>0m&fhHR<(uVolP|-b5pB3a5R1iOI8=*<_k09r`Jq~rj
z>uTPsEPX!+XTOS|?0Xhjc&nUQ-m+z34aPm|kVU`BqW7c0`{dq@+`NC!+<3YnTp(3N
zm&x3Gm_v3K%I*Yq_CA)~f!v+QHOMZP%q?~)g_SU4t3v(TyyM#JTS&q(jbFYvNJLQ^
z&v#-v%2{y&XT|4oF8u8CL&TDULR7o){tKayaK^#>xUbI$=2P?Ijy`^D;?j>d<I(mg
zKXym?vFW*%ADdon`LQn&Th>WGen|Q;p3j3nn|?e$?8nceB~H<MBmN>Dc0jk}pOW)N
z+=<>yW5_psR|X3hDSe+F`WhI<J=4c;^QGVBNx!v4k@qMNj+<ZNa@(AfHa{MopI_n1
zv3VvQ)}1oG2V{7&Euvs+a?1E#kneT*XkcAD1$p|Ir;~rTx1yf^mz-7G40{m|>TBbA
zzS?v3t*7C6Ap}~rk}cZLdrcu<;lMT_GWsy)qCun+n?k<Ak8MIY2AB~a7nUz;<38%|
z%X>Qq_2pZ~_VZ=sxgmV{tbG*KZk$Jv7}_{aa%kfaC!C5t!7&;G5bhC3#Agn=Sa69!
zx0TQ~=$g26)=_w9KFV25hqMOWXlHc`PRm(Sy_f({BL)(4(6yZP37fMn2s`UVtZI00
zdWH>(_RsgaoHpaPoYNf444a|1%N|J1<!FY1#2Wv?S@JoU?YFtib}I`UOwH@Ext=wE
zC;MHgYF>|N7_bg8O1w?<Sj2I<DCT{r`c`C^v&NPD1O0|MH6KrvnNx+xD91<V#E^v{
z{N^M0Swq$y1Q0o7A(E+iZ?SNec=lULU&na{!ww%d!*qoYw~0aPS>%5m><BzIbN%1w
zI6vP*`xszbNNt>WPCv(Ke}+CW$cE5sLKN=*KWlfKhx$8?&+a&>(T>CUq1|yzTsqGF
z34I*L88yn8u40XHrW;v~Q)rF6mMhbd%-O2lac&Je&Ihb&*l{pj5|JqEA>p~3W$yV|
zuGENo>>+=DM-z`bax}~3;1iyp9wayBOZjjno4?qGszMrQ2TvlNzQ(KfBF@4I+W3PI
z0LQTAgw`~Q(?4rYxLc}~*U{fu^5N3HcVO(@FmgHz=Xo&tJUkO=8v8rul<*A}E~kWj
zo#i#;f7{G=Gwx<`rk4jf`C^9mn%!B#1H=my7d1fagAjTer-=jH8nt;`xIVinqyx1(
zdzbRh8)eFN59%zZjp^qszw8*oS^g+QwHtrgC1;G*B+wzIgyF^9h{zCRTlrbj%T&P}
zUM9+wA6Hvk`K|BcOC~OTX;pF`U-Cuyk~t&|>`UR5A7_q(ulx$azSM%2IOQ3mwUDx0
z?(bgii7DfLF85mdE%)|EmV1X`ZCsSlOs#3-VOY{o%e?`d6ocCTYwNwf&ixgOKbv3v
zZyX5&I|p^{tMS%qpXK1w+lO%OP9dt@*uUc|ojZC9C2H+Bv~xe)$GJ^hI`?C)KF)0p
zjRVgxO;@)D%2Qkp7KfdC8(QL&oxAV!@$X#@ime=sn91BO;%v-X4su*FmxDIH4sYW~
zzaGGSzgZ5Ndp~kk{J-3nw`?8Mm;HEyy^k+{)H;MO$8V#kc4GuZ_E`?v=dq?2S!+YN
z9OOPMN8*+~&T8V)S!2iban_uuk=S%yYa~Wcz;3POtj<$g4lWKm>zFpyrtc7oSuMsc
ztouy&#9n21c~`>iO)TQVbJg&Q(94A%=8=6}8cRj2`#J|&5$1E<hqa=)PP1+4V(@gz
z6=9!Au5Cp)<RtgS{*&DNJp)W~F<t*=k~`#9O2ZbK$bc)tX?WGPkMI2DX}ZNAE5Z&T
zs@>?^Z1){=MQHaOr!Z>EciiTR(DI!%eSF8nrSI%KzmM;j{&7kJ-Knhz?+E+OYpiNN
z-@&>MTPfk?;Indtg&h`m#5Zi~zEqo+9Uyaf-4}MT_enOlC<|@Q@*&CQ7Nyv9?Btp!
zvUh@8P1Y@f@ZL#RWH}h#J9#zYDd9EP0Cxz!**oE#1e>RL@cOXrF;w#&!c9*N>MR#x
zm#&YqtlTt&v;0bkYBzrWB<54tS>!Gsw;5Bdtv=H&66g*XQkwg?g^649zjOPzg)eGI
zF~=%vNHK>SYe>Q3E_%<CxAXdMlvIV?VjC(2ORRi}Z>yw=b$Gw`Y;eym;tbZ_NZ#&<
zH(A^K$@v^_UbFd#<vPhF9HZX0x!Q~L2W+Nb^ukDV*ybrOTPc1HMPf^SKNS<(e1~~8
z(6F*U;w<646YR$K**obsru-cPig#Qua{YsQC8m&lF?dwsAJ{B0rF2kAC#KSa+*`9v
z2M64f_m2BK8(;3|=N{K=9MnBdhavZIkMh3`;U4!2QSC<K2Gc#Pjd$T596F-iV`U%r
zFmdS~&3OGsID~j=6+ULpu}*Q+u-wBO#|GX<tO>ivL#PzpV~>i7k#G+L+di+M;cj=t
zJuHLIgY8)pHQT;D_6@^>%^S4yzkB=;Gp(V<kBC3W!K0ru;C6jKe^B#bBEA=CQP#FQ
zcLe!<!%JAGi3P@s*h$2;yV&!W+wR_RxL4<Uf!VCxcJ9hB8}90x8;~#9HxR>n3}Q1>
z?njzZ{=p4IQ^-Gfo60*b*xw&6`JX}kAs)Am`uM}mPYmG?6+%?I;r~l0WT<(te{5)X
z;(ZuxMF?LlJMw4dmET?wl-DIid>KYic9DXcRIQlS86Pk!_r&>_h)8}!W=IZG)<}vw
zO!*-wX|39Q+YOLz&CJidE%Wxwg4?!n0*UVt2z>(Sro3BZZt*!7QGzBZ5tXhc#Zu|!
zekz?UGelHMkd%l@pF9o`cs_Lk8Ybw+lP19`p8kcWQl1KNf_~jF-EG~!#VNSRq*yH&
z^a3~Tft+A(JFdRZt6%TM)nZ=dzJ#mWc=hT*T+Qdzuin9xhgZAtW`%J*uZ{$9brr8#
zkKk$oukQZ{SL1l~#HYAAn^&(K$JLp<`rr#(#qsLnFLC<oNucpC0waW42LHhD2yOoc
zGU0l_$n&kX?=dxekmSJ!kMz&@#iFKr5;c9VabN)-syo{HiMqU@3$c-~Q^HmWxj#&L
zi-ep~5U)4kbW9?aPcP&tM{95B0hI?i(($knVX}lt5+(wIRRiFmwa3-(aKp@M6SWDo
z3Jj<1WUQvwXd9WD$$LGD|2(K%W*oVWcU`AzT0j)91p<uV-lpOlZ_|@PYzl(t=doUW
zr<SsJ>9^41TEo$L<1aE__n9?k%x;RW*YpMv`U4GO-4F=0K#bNWZ_|@ZH16u3S8z8y
z$(Dg8CJ0c^Xb}!gzt`n3(k%h!)IcFnqz25;oggT94LL^(+>jTYe<U7gHNLpktaA71
zHHl`%>3QC!-*=d0n;_8HuWXTbvz0eeZE|{>{@i6|H$lK47FBy>KW;4VpPihD#piFo
zS#<5MdV$YK+m*edRc$;2{pm&h61{<Jv`bBy7I?HC>DhrB(P_6jw7@+%-oT@5RCF9$
zjjwS>zt1~85WjJn9=~~-p7pc>WD}4rcu0G*`d9;>UDL&RS^%k_#0v_~g7V<><5QoU
z&YTZUJ3e&_vTg(uRn)~f&~O6R)6~b}x9H+>2I9?Se0OVFAa~;lB;;;5F-;%60dHr4
zvc*Q}KHq%bTrjuCZ$OdwR@bm;)@7(YJEL8_|5}8IY`x7JKr6s9ghd7d4seH?cP|Zd
z3_dDkIH$5_Z_wDD{yf-%*-I;7oXp{Edeq5`fd&%<k_QrV)Ohu~5iyrVi(w9QHvJG{
zOb~GQ6LTIr`)5*&H<0DRb2C-vzlGS){VjnCr=ecVo4sO}+W2p{@B|!qM+|os8@e!Y
zt$IOi{3Tk&tMAns-s{zxmSN?3G!7p~3$^<%)b!7_hPPsB#~|8ZeOZ;zU`8I?F0bn7
z9`lJ>buNDM98V`g1ZcR;Q?*-b2*uXGL3Va$X-(coQni#1!Sp@MNUQkqn5v>4oKhC4
z>2Ay|(%qO2rMt0{lkVQ5HMF_$wUnOP=a3tsFb}B0>^<F$=|R;t^;@mH=EH8B;-<TC
zi*VU<o{W!b<MrO|I<2WP#*-1$^lpzj3+v)nwUm#be4Li@iGMhv(~4dGWPENV#(xgZ
zv9(a}v+j8Hv8j*a_Xu7w#}}%IX`!9iA8YtqeD%e8ul2qsY!7T%kApw!@kE|RK+kw@
z#@_08(XjqZParirFg!aj+T{t{h9(}l1uczlYRLBoA3Q$wPnr7EhXbSkl&L;8`VoEf
zqk8;*p}0SxU~k5r>NmB9?MYEXF7!EQxZ+>}cwTMdpX{T<qGo=kZ#}-YDaHhZCao#g
zmALta?QX^R_jMRhn(lQaX-$4avGK^M(KfWEI%6Zh>#SuwSL<YA5=)qeZ$JumqwbI3
zyl*?j-ntDBfA@+w&SpCYv-MD@M{P^gSGm?9Jz#?F;eo`a?}c>7ie1b9#HUePkY@Be
zipOm^5_jK=Q@^tPi=t$w4<`F!W@bH%gQ!t=Dz|OFeLkz)Zk!1iDf-(i`e7Dbi=tVn
zFWClwp8={DW??kCi7l3l?~!`+la!jMHH~(e4d!DekkCSYSEBI|^P<7P5os`WMH{+P
za2fUB$C7<4gt1^kJp9WIF5{wkbg$cy^(oO7R=QDf0hsYM6$hAk@OfbVaVK@u^g>?u
zKzm&0@Vd*qcJrDI%GdSTDg0PgtAP0Wn&7!LRUI|osgAnAp^l2v+Nb*{I<tKkP0!Tz
zA=d4_?X38%nYRzot1fV$R|#yrs{Z6ObUbUZ1_c^p9>K)rxH(}l<!3>DcQw1@?(<M4
z`n**3c~tY^dnI;z8fWiwd&w##r9x7l&UHd9W6gg-6?CeTbplKRwES~E9Rc$ZIDi&-
zQbbHKCw<~!@CKd~kysFZZxTd=7lfaggf2FjAlzpX#0V`2K9j(K&*UgD31UDL9CJ)U
zeVF4~LAa;P<iFg+a!hQTiKUv@StjNnHtg15JkVRW;d@^VA0+cdlu@&M_!Z)pgP$8e
zbonfF`7Ct#@W9M?d~Ea#^??xxzaiI-j-_r`$uws10Ww=5AhW`<rZKx4ANs;~z8tD!
zo>O()0%-;KEr7(i4oD1c8jf0;9}PhH7(5$MDb5)jt77ZTi`B8DnGbc~yI}le+s%mW
zk<Yf}4|js!8(WBTZfwbroNmF%b8al|?Q`z~GV`QN-t;#=1(!dZ(!8-w!RZ=;bM0{I
z346oG17^T_gYOTc{>x?;s>um2=0NiW7&l&4K|0;MoD0#*W|v7M5zT{^nv3R{&1P{k
zF;OO#5%EdHK?G}Giks^rQYVol<VCB`O$9{r&dp(BQg9n~No<X4d%soG4QhO4dpW@~
zSWgVpYS8|aTFq)-^4oi&fA98d-v>}D)4GOBqeIihnzl9#Ef}^z7@e?9w8}IXo3Kf=
z&MaRr3nJJ=Zgf%FT&s)BrZ1FjTBtfsS4XL`SI`d2W@|D@BVzTR=%O5%A{yT6FCX%v
zTo}69E2+$uCpq(o)CHZ0<G7+NUiP2(&?u>iIL&k5bW0LW3t9BPB;!;~0e7e4ag=4#
z+&s_Lc%GNT^DTKiFD&G_x-3gg!C)+|eYl?N5|?$E>Vh5k?ZNLQ{0`uE2)`rveTLtc
zGmwto9{gUy?*M*>pc%$0Tzv^axB@=aMHIi!V2Q(Njfprdz;y8Cf+U<4J8`N?#_7&f
zoR+8Kd6}1Ixp|(e@w_01=Yl+*7sF4)RPyCnWyF{Jc)qhP_*>3JBB3Gbdp=*&D$#@k
z#D~G{*Zq}NY1uC&=f+0j!ztKrJN9bOk;3?%@P>^Kl9nL=nBMJ%$rQr+NVBYtCNMkM
zGC^UTW+btUPuN$gWr>jBqwz79WpwF8(k%;=Zs7}F1&tyt&F?v%a(l7<gT;sbZIM6R
zEhID7(5+^*FZu29p?{zCiw}!29brA`+_^Vc?JJtn@@<qK4#p@U((Z*stOlnahTUAW
zexlX-OtHG6jEL0-qPeXAAUechbqLw4-tiH!dWBvd4&|bYlB_Obiq&6r!6q>vLi6N=
zduB&f%(5F+EIfD!{Vgq%VGkT;q-B|wW9cTewsaF(TN()6EDh3OF*!1j!nS7lG_G8>
zCvV%5jUDUFcjC0^5w!lsf5ny2@!L@7kxDmoR+J+WwpN+=Y@Z38-#1~gghwPytTE|b
z5?1?7++AzJb20PaVa*qE%*hL)ZDf-23&amhQhM)pE7l}rlsW0BLkjF_HY(?sH<nqg
zi9mmoy^xG(AIH4WLvR(KtsM;9j(H=(Z2fJufPzjZSA4MA3<J}W)Z{b&sIfQO5G#jk
zfoa$enuTqFY1n$3#h!@`{+r+p>+#hJ%bHo$8bA~3YHIRT`pRkrKVy4~krd-D$6H76
za_A+lheJ(_vB_t?Gs~)pP<>ib@HNvNW0APiOa$*dcoN>>T6PO85Oc6)3t?|>vHJKe
z;9yd~!DMk#aF576P*?S@(?DI-pKHLb>fi6St_n9$ge%yN#y}TgNgJ-ZF*96sH14u#
zs3LoG7B;l7O*tzm$iL@|R=2Fl){FhM6m{v-i;?23S*oZ@pI#h*i&l3E#+lT;kb1Kl
z$A;@7s)ybYRbjpG3460z_BtkYUm}u{&wP7^)e0h~)d-m7ia8i;yZ07*yr6k(yk{%o
z#hPD*LUZqQqhS?gp@i>CcvhWBUnAk85<Vy4nRl7#e~|E13G3keJpL@96)#S4y)#X&
zcP=s4JNp~lF!z}v^Z{ak=3H~842d%lUb)^GraI<e%6F4fb=;oXcfIox);npfT<Y|l
z&v=KiyK{$)4Vznks~C$mb$ExdKXZpg;!%Ib{NeO(?=YTohebkCf8)p>PH7{C3C8$v
z*#j{qVUZL5#MW#sb%vuh==V8@VF|(M?0BwCR4p)5#;}=I3>zs&G}Lj~9BVvrf-x1Y
zkqnp&n94+JC`%`jC$qW1kpm75Py@_iqT4n~&Z(0~Avk>CC<815^pUerEQQQ=7yC1!
zaai%8n^CrR-;NomaoEzQ)QT*%f5!~eh5g(67h>r%lfxYCjH`0TOxPG!=#H6Az*K3D
zFhgOEIe<RU3jt}azF@8pfxL82*lBV?>NKn4#AFejY0g2hs6=OAVM<2(<RILCm=B;6
z%q8m_48iPtvGxr2TOK%2u9$7i@ktKRC3Qi55D%h<VL@@(x=hoM(2C1$LJKXsorbcA
zMOP-{v@{i`Wn$?wQ!IVvh^5a>V(D|cSo$myOP{5-rO(pH(r4-auBFdon4k!2wc?eq
znXfj(LE^(;_Ihe&NQ3QX*{5c&?_+)dKa``Su(qw&hpmmBJqH7XFdK}L!{Dw-U{42(
zGCEqooMg$E3JxbY(gE`T-AuHMPa~pPmIw|vIPw7N01KID8DSZbGRq>tQ3#GYu@o9C
zKC>b0@uGjT4iYa0YAjU#)LNCL_V1|vW#<p?A1~xmNHn|$w-?Roaw!z{a_F(l3eG4h
z(jj!fu-l97u)=mSQ>-wZO2q1$bRt$?h>o!WxW*JKVCN9AdMS^H)hnWltWaLYltQbE
zh*<p<UJ8vw{PJBJRk1{w0W%KNdz2Z|ENw+SEN!tQLItsFM$wjTLTgJm4IDY(5DHtG
z2pug=%4C4WkkfA|bULhmWAQdjZJz#_yf5&}d(8RB0SO<y*Tir9feFuBX2RPgG$f3x
zH|ftv_=EdQ{D%8YcoAkMJgj(O|MN=qpnLQe@dP67MZB7D<hzRU`c#+k;J=1Kuixu(
zznXARQXU|s!sUMRqgNB21MSte{cj!cy6RqS!`WxvW{0eL?Mf1N9Sk1Wuc_0R)RZ|n
z?Ro7uZh!U0WIi1OCa*fnz_(Wt&YH?Qfk)J2LtXH^e<?`{NG6Uibj$ly_$+6ID_u<y
zZ#KSf`~V{O#FLxPK4DjQVZyJV&muKH;r9aGi|=?kysq?*@x{`Q@whm$dtGG*vRq~R
z&pZ<x2l8Eo;J}0FTIiYD-$8+oGgZemJS|UD8~+;$A`9;I;EadIZKdP071S{wLh!wa
z@P0Hna$U(d;~{k3ZGQt2s2gq=E<8UpMd)_^a8ox5$RhL2`=J{$AX}Cz58CDJKeGzF
z&@Kl&xvm^fXix1!)M_|Sa})9Uqo}sf9kp`^<s4A7dz6~IQ%!!()bFCFAm(HB_Ln|Z
z^G~9nFHq1)H1NmjEw6z52FM*CKLc_H=9(a5%Ffp$m*KI8x18E(9Ugrc?$gwx8sVFO
z+aiyHG;14>f+GfZp6B{J4WFM;eYfYn&sU&da6c`upv-e$+da53GXt@b_q(*FYCJ45
zd$_jZn5KI6dIO`M(*k!EYH^ca@idKy^<;dUea`DRd(`Q>Jq_=~;8v+uFF?p0@d~AB
zp{-iO7iZu;YNw_{c&M%Rk6QaIaeoPq4FLD*bMbD`EX?k}9bK3f81bApFyfWDO`{XF
zm=15oKJA=iJgd|3CokTM@uqZmQ(840>ge$(ZZ3*zgSnzonmQBD$t@_`Mt_;_(O>3c
zy126wn4gF{?_T|w@n`%oE>F{PEXPSYX8hj7F<P1UznVCPEfa4rai@jXm^fy*W_q!S
zV-Pd(`6iCR%*3agxZA?7G;z#(&Ghe>c#egiW#Sm(%=9=DM?0DLC-{Up^v7)2#NRP-
zpM}3*;&m3j4LI&D)}!z7$B}RkFS8)H<to^B_*1T;zjKV&eu5{9cKlezHSzq=d~}c|
zAM$t|mf{Wkgg3FgfnV_EmN#&9+Ocf>&9#Pi@EE{Rc=gec+ISZDJWbzsc$#j<?`}sn
z{-|5S`w<3tKL}gGRd{!VPnlLN?!gJ`<EndjF#*3hyto>R;xWB3JnxEm@c8FC?&U0l
zpk;d=JpLs<Od%dU<wyI$6)U)6!G(+QTLs!1Cx~$tQJ{w4|0pcafV|I4d0#N==OWz@
z0zR_jeIZao2>8g7_XSY-*umcxsVBGMn?xruV(>4uY{ReJ>1k;1RacyaWtCTd6CA8=
zhx#pi(fK4MNa?Y=kYm|1o{XN_#4P>Q;N56aJUoFX5PD&;@SOd8;4`I%Pam9_whIqi
z#PE@W<K3e(1M&a!VCM0u|IH@>9?y(<0on1qL2UIDT*&wfp3UUhsJ{%Z$(v&KW(Kl0
z$Afh5#&~@nXx>fnF^KQ&vEGdK>Q22wn8t1R<A3!~M8VvxyyH=-gujrWZ%x^~?{LiS
zw!;^c@5W8j<s{&*5^ny*3zx-c>Wq)I^1XA`qTe?jf`>oCTzZG+xcF2<3jP|Q!M{n^
zW?Z0^w=y>$iCJ+ha}%?8)fxLd<sb7BkB*8wsPRB~c6!Tq3by<3NFDWh&*?qwWlg`v
ztG6N_&I9H1q_ld<J52Fkp;Gxrruer^@ozyq9#fx-1@F*$$;y9KmA&O{edKGWeFXWA
zf3p5b|7880|788KkMK`ze+>WsRQpN)C+qM2C+m;hnSW~g=lnzZn-K*8<-R^{o2RaB
zRsW}zt3*sZ${5q^^P$M|u{j?7Q_Q2YSG=n>Hed&XkKcH>-<5*L?eT|IHy!=ZJT&#O
zt=^0-wP76t+=lVy5$g$DEguzY`#cUjc#UPNma^NM@wR&ZG|2O2yyemNWUJFY@Wkv<
zf7a&BFw~!IP27#hXK-vi;vehD*r7ha8^@kIzN|C8zmkuqv_Y<T5-0<InrDyN$S<LL
zn{HU?N%_dLL0@SecnHsP9MAtB9>xh5-c4!fisK)A%|vjnor;JQoWx;UKh(^><MVAZ
zUgOD^%@Afi?t|f~PQ$eIfai|wb)J}+iNPN*6%&6vmJ130&e}Hgis#)rx^Lj)T-RZB
zjrCEc&)kH{<Yp9BRhitJ!uC-nHmmyVncj?L+4r34D33!Og%A7h#Itspbx@L`mO)gV
z6y=tp+)|Xgzo<Nj%7G{iqM#T=HMIU;71i)Xyn6pIZgix4lJ>kD5Aj@)7W0aKjuxmG
z1AQwTP&jK$Ar=A-sGKz>hgVQKYYZA7%K^3H$6!S>#(`zQ7$-hZI0kCxJ8;T%KzXkN
z+Q&OE{MPyJ#v5H@47Req&OwOhw#Muy9M$L$5I84X#03P-={#`(QEDCT+KyQ%GOr%9
zM!>aW9u`b%9VT4u5cE;29D)y8qUqY})W&+SQHMg<3Uwf)4mpG<LqOnei3<okRpJ5y
z?_qIJLlOB`k8udNc8pW-t#z1iwL{RMhTwyI=EEbOD=kf)8nYicH2o<DAvNg`5>R`I
zZ;-fvNPk%30s=1+e5=OP3AlR9N&(l7StIz?I!w6QA?VOV@Ie#i)9%~aMUe4=x?&hs
zL^dD0V=L^oBC!su9Q=`q?ri=2@Om=ZziwxZ<l}VoFHh(>)4#Ns9iG5-=961D-_J+O
zMz}osNRM8b=(%G@9acLY9igXTdt!JMciXlt^S5Km;gS0=KTl|s^K<)DYpC$7e`L<K
ztf|^Zr#M^rW`ZV`)!$6eY;*QOCdl7R(9AW>U=#drCTOAMUweWUnV;dYYldf!j?c=p
zswwzGO^nWX6xs*K!xMOb16yE?cq=5|h0uK57`se7W1UZ4JWJgL20n#M+;HNpD+<Vw
z0(jjC1|GcL4~SKJXvOR5k5e!g(jhagm47#yMW^kWk@jx)cf{@9$R-MTH2mFjp&Xur
z=1AfR{FaL0Z5R%?+63`ZjMjD}Rx5r<z80eCFY~pKQG5~OG`!!|{cSnls2QO(#W(ZK
zZc18lWZAu1;8*oR+Ebzs%o|nRz^1@v5S1+bE&sQ(^sWbwe^)<S!Zz_&&>Oytt#+q;
zhNrFn8m~3w|2ZB*7WVh_jt7t5b3$KmLSD3qoJ78hIaMqs2KdXe+Y%dgoYdQDpZ5f|
zh?<}Yo}~`>%RC(du5xV$%+`C1J1}AL>Lsou*&SJdhkL1*`q(<blNGm4<k{i?)oA(w
zcCqyB=+R8fgIxUkc<cf3H{8~my~RICdD!6JKF-v^p&u20r1G#aPf)U4@e$JXl7#uH
z#WhiqMp&w&LTSe~vT(2dw(-aDFqlSB8BgFRtlM`plLP2Ob;VI2o~B$r&Ht2(GbgY9
zp0NZsS+`(<t{5-eOU-wBnr@26gYuvV`!e~C8MQE%%)kZwGf>!FJZ{w94KrNpdI)Tu
zVhsHSIocftCK30#l4*{l8$bQfG)HQ|gOBagU@(uIP{^I<I$m*Z=_=!un^(Bif$ehc
zIpdZ_zAF!t%L$xZPH<v!If0YQ30iyB1Rn)z0ka~QcCd>4BZM^5Z2d!jg7Ly#cynpR
zyHxjmV<!-A0G*f_+Mb#5hPvt^ym-j$<j-s$j<a}#{d{aS(<vj9lwbf^Esnm4+q&`o
z+cuVRCQB&<Yj>P@%-@u3&WAi8^RfK<?jgi`A7QQi+l`6L^K|!a>wbtWJ~VughXs~L
ze?`+j@#rT+l)yXOsJ<(Q)fXRUC6-Ez?`XLBwu{rNw;3;?Up?pzyleonvC|39h|5MR
zY1rC?|2JODhJvN6M^;Z@oiI&r0J+ofzSo<chWAd2ms~bbw7=G?SGbb+DjbaG2QbhI
zSE8rsmUvH7S$wwMmaQM~G#omawkxf*Ia6=X)I0D;NCQ_sc>D(^^ftW*5bw6lJ)u4}
zJS(OxD+Zg1rW0xnI}-8dGf&p;hDq@iGlw^Ta<CL}C1&d#G1yDTMqw{rV%pn%msfvX
zOZl50#?Bo>|Jm3G+{BH*4f;;?G5^y_%wD{gg;^o@3%#C{zhyUhH$cJsO^6N^PeX(m
zAE?<G$amKXy<L6m2Pa~VBQ|6n&!CiACFZ23p-lt`Q@3p=YxidAwP+!DYQXJeYodMM
z4|QhhcmVERm(%RxMR(GiaF5DF7Z7dOo)<AA@uqNu|LSP|ONhck3SbP8&@19u#2&Do
z$F4t?^r7sGPu0eMM+d@LPk2+_#-Gr}<3Deo*c?D-(GVY43foV4&x6-avjb;m2MQ81
zH=e<czVBou!r7e@GB+ZOD4vFs)9<`6b0chp*B9S;R^~?dIj@J^IXt^*PRO$_n4R*1
zr|oaCo~9dOycr*P)M<OMQdS%3;o17}?3Cv``;OkwG&X~OF+B^T$n#=b<rv<~^)$s_
zllHvlIIcUV2PQU74@_N`hCkprGbE^)S5956_qG}5rS+D#p1J58IN&kA!<+Hu()09o
zZ_Lr|aeTg)mj}D?mYv*R&kW=~j;=#GmaQ>6rt1?MF`!&&N{pGhdb&P!9bQz^GWM$X
z`!LqcX^KztX0+ARX6qkkr|kCZJDeS}&-Q)+^~GKTK9un?^gUh3(ezg`0~4Rh42*pu
zrW*@-sc+oa)%t$wyT81(G>~_jz76YW13~;L<UGyPAZ_gGOnvMVxNn0(Yk${g!J}TN
zMEVm?c3}$>?=QpJ;o;)7S^D^>;n>%4pn0)W>o)H@q#fhnjCXq6N><=XPsVfpq-?wv
z6E{P@2AlEmUVV9DuoD>|AGP7$dRuQSh9U8%aJ_2Q+n$*1>d#s+KwwMdI`!vU6Sr#k
zzJxfg$9_#0LRjY0EYd$o!v`+HX`YPj>H{l4pWQU|4@@`K!L&U34!qlp@SB~;HwpDU
zTHBX+aFFj!p76#P{<th1FYz4lWAs1j(b2TKji1+Y!FepZ>E1uav!7Ap<3=i8Tc_7x
zr}%Q#;f5yfA5a1A9DgW+07mL~E#m;zt5CIjgmDm3g5T%0ik6LjD-dz@2O@?i82f8X
zcXB}2zU__I5cSXCLslI*iu?;+)4o*8*imya-Z|v$VKB6N)W?pZJy8W?4c>a!GK?DR
zns|;M?=D<=lSlv1)AnJix2eRH82x-4<8dNe%B%ZXG1U4)4$OQ*pTWQFom7gPL6I^0
zv26yyckXW7RlBnrBNg6fWC9k?S{$~MqN-@dFLA#-cRqr_bBG7#$dHYf@^IfR4li7z
zL+tO1<F&Dah7N56i|vhT>;w+;1m^T*=g@vjUv{kA6Y)N@Kg|=k#)H9&<GD9*243Y%
zd%Zaj{~Nr*`CE9wgYr@z{h$`w=FQkv{f=e+xHo6re~I)*j!^z2HW<{#AVgws5?al{
z{-mgZmhr0E@D%9U7Bn4?>$9=Lpl|i2e4({{0ed^CrM#?Vys6&5iCh`4A>??~Y2BWf
zm#`1%%{Yq2P230TK8krQw<DdNj4#y(egS4!^B(~e<4sr!`~-qox;Ld)gK5K<{e0UT
zw-x!v!(XcgPZOd_)97C_Q`#}(64;;?bHLLyb{+i?9Tzf+LCe-J_UO06j^o#HyQBNt
z=<ZkySlDCJUo$>JJo3cg9ms)pK^{V)5kgUY;(uDJZFd*;_*J;x5zTLFZ~Z0CBXIYf
zbJJm-_L(BcVE^>{>dH$v%PsD<8b;Ic+(jEJuD56F*UfEmJ(jI65^qPtI!C*Vi=Zwj
ziFhtRH{b!`o37YiT{b23%=Gs7#~@PCceEY3FlNWTFTuFtb=_EgFt`c>mt6rUA{-`I
zg9QEH@q-PalTWk1@xdi*3e9~b<~3{rdp7n00|x@cKJU0)X~#U{_Nk5R#%$g2q;w#c
zI`?SI0lu%?a0qjZ9e7>)NE})}w8MXzrU&`Dwx7>`zIu+LOvi-tu}{VM-OVgs<9Z^0
zcGJW^;)4}<v2(;<KvcApJ>HlvA$p^!%X2Z$z_vBq7{(qdMEw||qWxbTH3Jje89n%Q
z;fI~;9jypsJ+<#;>pS==dB(?T;{y<#r3a7iYxw-6tmVFGP2Njm_1z6`$6+Hn9(&yr
z8or3}pV#n3tp5zQCaaD{9Qd-~^Z4qkZfF|uSZ33WF`8H(kstQb&S$xm_RJGz(|!L_
zOt?<qZStOmttVDIzr0Jw9}thJ{T9C})yFIocj8TnkvMI-91=F3hb!YZm9WyA(f8r;
zc!Ty~q5THLhQer^?-z@K?ghSI%!d}R{Ml`M8*j~O8T<X;;p>RJI9Cg#a`uXldE7XT
zV)*As+Kp_KWxm2auHB88dfa#&X{NS_|9Ej1Ovzxf<_CD+^%LW-_>e7xsy}bVE2Gi<
z6Y&SEzqw6{DHWcG@^HU0!|O`)CwO!KtSDWMO-ZOYA}C(DH(X0lEGE<=`p82v;)Ica
zkFRKfL@nSkj%6h}6Vw%Hkd19mwK186U}l}@`7WyGzkv5Kh9P2)(6Jib8t>7OBu8(|
zpRxlJu_o;Frr>*%y$EtUHj0i@vx3-S#TZUQZVO@zZ%b^m#<0-!xL0jp#C;O_D*lN!
zI&Z)*UTapw`vA?yH4BOnN@eROy(x#W$ndm<P#o;F>jVZSXy}E8i2EV`8KRCK8vD=$
zbDHjbGM>Ry3|3VYj1Xg8+5AWbreTkIbi5#kX`yI7jy-=w9`w*Kl*%t^>=my_A2g03
zW3UZ~F#;*U+u>coE~*(w2&6mF<`AO=&fLOl{1swN_u`&HAmOhv&B;+V+4puKQt`K(
zG~4^=wYeg#`4H03U4_PZtZ*P9%O=(xB^ILC&w{<q7z>552&84$XYclxw4+}7Zv-Q@
zMF^r^569F)ydGYU1mpM}s9y;wT>d3Id=084tM~@5ZB=|xrVUX=|MhU1LB2sSusX}n
z<f2@tQUO)^w9x113}`jprqu;eT1B_e7nFdmeH_^qHJkpF%Z~`8qN<wy3~Y%rtdr)3
z$i3h~pe)6E>b$KL_^z1ZVv!uDHMt#{J{>c`>Eb2t1jC7s7n!+cnz?>*j>zQ+TyG9q
z!#!AaA)5!U9pNXl1<OOB_JrSNI29-$Ho95>lin*tmzuK`?&~yzws{_d*n=oWE8daa
z<l4yux(C;v*bP}ek~P5Owoy>+cBQVEilvb^<B;082N@{2Ucg*eohf!2#X=s*Qr_i|
zhdZnO35;^Q7%!sSQ(_y0yqvTfry*nac$h};KQR(f@=B1I+yS_mfzLREIx?k2q&y)w
z3?F*}x|;0vZ!&1ml_~Xt-JX)q6ibl)ronG9(NW+;xvsL!d2FCMQCP0aC#%8F#|g{r
z<7?`Qn>cOAct>sAAPTzIl}9l-rkFG-CQphP2Qh4fWUDnU6s-X#WnB~LC@0I%6c@57
z#)zhnY3x|D9S(pYtUeawg~lChc0>UPvDfo=(2Qg#WG8PVSyrEM`^e{J^)>44-dw=y
z-zCbx29(shd_=hVV6#i(Q--4Y$54G5b_%|dO7BVqj&j6)3BtQ6cbt^VA&%BI3UUW*
zk0iE7FC-WrF1Fg^ZY<a-P^9%B&1{c7V6f`L&50!AR<=4S4o17GA3{}<v4LYJC)nCU
zY;G8@W1gi2xU&K7WNg3S9(laFqA&roh=eIuODDsykYM=mK`~lci>Rz!>@P{VZ#yFb
zqp10<G+WS88@V(UlPdY~l!?F*M+t&|z{wusFwGl+oA*4Uxm$+d4)o*x*CDu<_2d5G
z5ZvGI$9>lj+$H_ED~8}+(2u)d2=3YaxaSSQeQiJP8AEWhQ^Q{&QGXH?0ptl6r#Z$*
ze0eVT3Me*)H}Ie)habUk1fyVZHz+oSO?VPC%<wE39tOq6@HjVXEeVg3p#>Bh!zg<Z
zKO#dYT8Go^4CQ3#jb_+qXVA#7C0faLJHsVp_)9dyZ|$`hMuw*7TKwG3aHtpz4WI;X
zWaDF6fDyG2ttT&UWJ6AoNdpZ>jk_7hqg;wEKA20rinQQidL?HdedBne`D9FFQlU0(
z9YPqrG03&B6c!y78~JzFqc!8lwFcAdKBRSH*1j$HU$_zy2N=;BxoV6Sx^M`g*`}!A
z9~j;sy2BFPcbsVaQ$M+!LJg*D71Fw~aj`9UI3h7R(l&PXllUJ)Nc^cOC-?#2mQoBq
z=8~%dt-JAVjz$xTgi!>!x66XRl&~BbY%cI5<-kYwBiVQWc|>#C7=9@k9Dv3mGOUc2
zz*t2It&}i>416gkEWw;Zb}=!TDh)KdXuOItY(^RVfh-1h8RMho%6RHklx$G=Gx%<0
z@CdH@l$?aj#-ET#^UaGI#Tt35@aq?APMe-^IVm|8lDqHfQzmCM#uQm*)W|v19DLwv
zn(eoq5{&}ZB7E~8MWFqM(Z9Ehhgn7=g(9YmKwn|W8s49=zi-3YC-!Ht8|uNWz2Zyc
z`~s`g&Ck&Ee2_7xY2rJa8)2&&3*8S5&L6R3#!iT~`xXBMY1tVc;2AwkQ&NrjY&I^_
zS~s~B+<a!u<MHFtFnJ5_MsVx>P2)TivIUb1iH-mo(FTi3e2naMD)SB&iL;v~3uUlo
zIA}ai--a6CiA+6eXW=oV%)m99xGk}fdlAMhpFolOwzIO~6<zfAz9vW9-Rnxm!Z*2_
zbJWNZYaoWOsi;5U?le8814S5TK+oW(C!@wEO#C7Yhi?PJO`vptS46wM4Jkf47*+h-
z{`yTv)^5MdAJqc>+oSSc*+1_kVbzyk5q9`oXvJ221L{Z8zh%w|uaSvckVvM#0?PaO
zT==!1-eR(A+)f5`dZO&~W~5l1o@ndziy_d~>1R{^ecLHVZrFtFweSATSK(r_2?mJ!
zgoYzO6#7_GFZH?{1W*gF8YRqz?csGJym7NVX<PG+^SKWlxN^&37+G4YG4WQY5~y)(
zIbq^sfd^_5w>Yp0L-LtGx47RcR5o)>VoU<6?+^s|Mi2Aq+wf4sIpO^=tay!d((%v%
z9!Bj+!}oK6+S0EV_Y8yI0Ttgj3;QRxWayhGvh~lhQx19BzD&Tr%d%Qe6DIKye-<WC
z)59I;|Aqu^ve5rgg<jHrEoox34qQpnk0gnG12(aFy;Rcf2hG#4-KjNAcVe?~`M*PL
zthV34hMPA9&*S2;VQldFC-b>kY}qyR#?|I|0+U^odxenUzXAENk8jd0Flo3;6``GO
z(!P_ef7*Qkw?-kq``hH`R&jQ8pN4Z%cO2634e*G|E^6leugLumaVNCt&O~!-02>1B
zvyz}bZi*HpV}E&8G6?A;V7X-+LT`B(EIbo%Cs*7Zo8|Td<|car3sUh^vw24W-`IV5
z=}21mjqfuI&2saW2{wc8^XPL^<x>Ee?X#Q|%cp!(coXD#dwgw}`qP-&3$(zM?eYI6
zxbXaP#@1@X+Fj+XkS!0vm^S|pPMfenplm#VD`Wc2Vz(XkV%|eOi@5bj<uD*@<Ll6N
z(fgTS<quysJ;(e)j!@1?<y;+5?mzN`$z~~vJw1i#ey0C$!1T+QZZLgOIQ{GD(~geg
zBN{wopShX$KkV%tb+-fhMfP{i`%7PXc@y;{GJ9eEfqN3O;ZZ5DtV6ud<H0>cPhbY_
zIpJmjwp%lHES;D3I`3U#yyIW|xlUd2BJ<*=`s3&Wcw2VHaX(E2cQbik?}X?k>(1lt
z3)qRbAsTWobcQ3lsWh>>&eK#2kC=f`3lGqLf%_NyqO+C}tQp76=Zs!|8eiC}MjJ^T
zrCN#I=OH#xn=52#bHGD)xk$rHy!<5eNxV0Rd#=YkDaZZTO>6h+Ul=K1>W%}O)^;e?
z6SoB8^+b4R*aYRb%Z;CckMTwiW#WeBvCJEB11&Sx+jNm}zEinTYq~7)0;e+Hn{o&(
z()J}h|ARzh$Bk$ND0G3f-|p4%7b%WqK}Klw8}Q~B+_bJeEUf)D>pSV_v2Z;9WE(sO
zlE{HeL{SXmO8xaX5#l4diW|QI{H&uM^eq-*LW`f6NUJyLlo$OYL-a3ox}v1c0L^xg
z#-lIYq$*0!^~s8}T)C`Vai%Gkr707Y+|n9<dF5i~tfD1Lif+Hcd6_fKd95=wOi#`(
zudJ;Rsm=+`S<aN~sO0Q^=}wWJ=W{A~OPtEhfHwo)0(c8xE?_R;Re)CkPMUOiMft+Z
zzh7K(`K;>lk^)>5l$2K&6;CQt{CD|ED~pztI&1tTQ>Nf@VMS>{`J$;yiu}c8<jyUv
ztf{K@RF;(1O)bwZ@>iGFO<l@d6P<HvuAAtb?<=nqsn<=Jvb3n87WDpXP9!pqlM<&)
z`QG%bDO0L^rPW3Ls_ILVi>oSY{PrZXB`%)mG*bqUcX5jIR-pwN!jxG)zqqKPsCufX
zM{(5>Uj-CiR8dvrzs`BxZC|sVH<cDcv#+dYKem6cp8e(hlk54<?9YLE&FWE`8OXl=
zy8T&Fx~QnO!mlh9JvC)YL3V*86%_ek(j}HR3QyhP7HG@aL&Nne&2+(x)11jvG@Yku
zMKz^;;((QYMcVYDYJX`>c~Rx`swGQmE6a;%*Wk{1r%;IM-0l|8l$VxDfw^VXr9~yC
zNQ+OAdJ1NoQgLI7$9kEc*NlKY5|2$J1-CC`KBf(HbjtL?Ke$D*Pr`Z$n<X5${%KcI
z-*6pmt<d>yp^&$JhbcEzLP5ld>e+v!*EdeetgO8D+BAhy3(70Y{gaA|71Ps&^O{{c
zwY(CZOP(oHc+N6Ae17?ol4K+y%2fJOl*=w#;zOvtY%w?h;PwiBe$loylWNNEE=^8h
zSxc+ROKie&imGc$y+wCb)%uf*%ZjR<mtHCaPBg=yT?0g+ep+43f=fyl)-JxRrhIW_
zQAJGz_m@DZGQFsFahcy~raCVwIbCr&my|AX7L}A#m)6ucFRG}hBTXK~RVAf@%TrQX
z=`Z)+<y3@U<RRU>x;cDxOZX}`e03GCF0(Q#msy$R6*9{!WR_RREU%DRL4Na43VhX7
z#Za-Ty5`oB+t_vpreUKt2M{}@pt7p6ls#y6RoG1It}tcFfaK(X8rIfR17?~sWj?$+
zw`yK#@zR;qRZEgZ2CMs`5Md!-SKhSJ#pRVznz1G|^Gd&8i&4A)HVKP}Dkci~yUIif
z7_R)KLl>K(#GDuR-LY`;Y4D>KZzxoT>(05M(E4GB34qFJisH)&g^~c90gclUI{?$c
z*YZ8^0jB;6bikZ{4~3j@@W<bULT<pqheDxxz|=p4LTdnf{s?|R=i{MJ2cYs~DAWb$
zemWF##Di~hD3lE7+k#0RVOuEV1I*bO3atcmJQE7F0`@!?3hgKUBJw3b-pk+z?0E&{
z0T#X*3T*&PeGT8nCmk>e^>Mxt3grN%0+s=~0qX&C0M`H(0(Jqmz6t)}Xa~Slz%D=y
zFy|oj0rUZ`1Z;g9`Vjvx6zV131AYW2z6X&5m<m`1=mx9@%mG{jSP!@!Q27|`fbtxG
zoxoGEQ<RE!cH<GCHGnyQ>j4V^I{<xvU4Zp~y@1Vt4)zCNGGHg58?Xm34^a6G`T{xt
zR{}Z#TL4o5HvqZ;`KN<&01GMi82TH#6Fq<$cgS$yzPUJJ?w+T_)H!0#8!<ewIcB){
zT>e<3_~wK{mr5Rjf%z51g*5V~&PE;C_ahxMM<&fy&$u(OUb+6<?_PSPYdolehcNE$
z9Iq&GcwBRmTZuEvRg}p{qbvjE{?LQnorhk;qI{%d+Q_77Bc0PmCQlognmICk*2tEz
z4Z~LrYfAWW{7>S37%TGSBcFRsC^Q;v9^doQK>4yp=0s%^`m9F2<llrs6Jn5W?8_qG
ztdXskG{EMm%M<ve{}wjM^g5(Z8|la#nFN;PhPWSt3CC!0X1(m~Xwum(`&&Yxz2re%
zD3|<`@oR=p<<VBMMyBE{=w86|N2Cvd-=de`BIi$`P>qd$B53#Gm-N?Ah+i+JZsE4d
z9GN<6<ocL~VXG3F;(r|X6It#$uqAJV?*Li{b=*#5Qz$eJ^^|flM>-osPPW6l_&MRz
zpEEt&Cz&IY8?0V1^&f%s-lsyLm*T-E57yNMdMkYXNz%jl!+n}L(jDD5%*u8WFC!sF
zyxd2g7xiJxNCG|n#g|3Dh3jtF$@;h`q12-h`Fi)_X+OaD<W~mDmo>6EreD5yp?_y*
zD8x^(o~r&V?^Bdh_B?zQ`Qo)f%Ijm7%$JLNEiZ&ZyKy~0dHrmeY%cs(Az%87q0mMW
z!uEiEf;}`OtP=f6`A>ka6MUk-zeaz2fP5V<g+g?5>llrTco13#ppON8!vOTjpsyc*
z?gc%`gRzFf?e-6PYIi8aO)cwSJ?;g)4ssSiuL0U=P`i2p`C2fhAY6nG*5g^w*Yu&Y
zU7v!THG^J;>-a|WqZtR_x#2J%>~#eA_JeN}r9yxDJM}yPdKc((AV2=u0qp=WrN6yS
z#u(R&G43J6?0DyXcT^pPk9Wn=$7LvFsiM3J<nz6U9UYV(zw%X~Z`eOXzIu_*iVe)i
zIPfg=Txo<t_?s$1e-T62wm8^6A0WL0>2w3@AUz)CwGKcZ3;Kou=&aXz(E0o);-8a%
z<$FP={5wfN`ytjc4l&GQXGBq0uk>IjbS@PL+c)9}?b{aXn4Nm91-~&f6sjh{W=~Im
zo--@j&M5yO7(_4R$I@Rr?d=xrLlh_amGWVims20e54W)yg9f#~$tb7gaQ}Ang1%+|
zdI{*w1JLgUedPf3)u7i8Kwk%X9q8w>0=D+pF8TY?-;nbA(mw@#Abmz)t~8K7F3<<k
zXDaB_$L`ncw_MPDpu=6l2k8}{mkm&_M$lVkME7SF_@BUdS%>~9r4-o9Phw|`O!{%$
zjFHZ!_!%RUS0!j8Q-3&28=2lPJbR?OVq`k(2t6(48<=&VKNE!=oq(Jc$dRr<Y2paQ
zb;QQ)5L6MWN#w=Z_-)p3dJDg4_$|eRM)?k`WCox(%77%}Y8%<EfC!k9vW#rgfSd`~
z!4yt1P6zBH@_QhK_$?&cM({ie_y?x^2rYO8Bwj$IM^%(bR!SNs5fm-*_<%5FBiYsg
zsj^a5kAbHGd4wsqLjQ0|?)l&c@iC^1y}*_-+GVCZ!IU?E{AIi>=clt2WiHeaM;DTR
zfGeTFZBQNRnMVZ>>Lywz6qgcr3F&*yMDwB$CJ+N_Qzjwwyl06F#l2i#uf)yaRF-z}
zJ7&2MKZ^cOBANKH5#J>;2R}CA-|&lhUm~9n`3R>e_=&@mZpmhDJWH-DBH4)_W#2($
zCVm&;Cu;aHE-3q9q{Q%9C?dBL;pBxAG;!b>M<}inT2IAQB7S2~CFLl7JMr5~Tj#?x
zEcb09oCI+9x|7Ji0%3}9i54JrrbJTKDsks#AdSzEaHcABxfxI#p}7BHTTSLhH%sO7
zUP?QDwfKo+Dk$uRnM}#V>2i}!+1rpclPJTZICw!#zDtb7vclq+jx(N*R|<*Uh##9z
z9A>?FnPU>ni@yR{hPo)jQD}*k!Zv1Kij<!Kp>7WXDywng<9*@?#l3w$lsg#9;6xo}
z49AJ`iR2LBy-B9tN`#Mcz{r(CBFFKglg?m^jD$$4I*SOs=-YrEmh(NFD4_H-oTwL5
z=929RAn+f>O=;VK&<r0X!XBWs^+e7<?U?dQ=D83E^Zbs88weYVy{zQpM0J+1oEn+(
z2-*G&BwWrmAe6=>sqzXCmh&Z%^I#)P`Ivbw2f{p`63GF=a!LS|3Y=I@Ez4<;DGYAP
zHi=|0<s~4`Q7jK>100@NLIBkhV;{?IBJv1+L^_DH;YaHbGi7@6J*4t5?et1q@!shH
zlJ3Co7rgl9_@)KEX@UQ@T44RTZc!n}IVMc~Kh0f#j2zcppWL->`ol?lBu?TAx(QWC
zRAs%p_rrHBKgK?f%_isba(i)&EA{Q%?%dtjySLY!+4EgfrACTe3KBIU4QU0`K$AuY
z#08Z!lu`~9f=Wb*q7WjXHcmo}RRlGEG+0$>_<X<b`(|fmXYCXvh^O;@XFl)yet+Nh
zeczk+o`r4-eMaaR;WMQ-`XWuApwSY};D4Wb|Bk%BN3&T^++gz)X(MuDTQ<|1`4{t*
zZ`Jtez4Cs$P}2A0E{H!VK60cb$$jsKueXUlG-MOu(TV6YSKCA^GFHf47_x~(!y{WU
zcqf~M+)bMBkRp#vpnJ7DS7Qn_&u&x7M@pES(a)sM$z8p}W@_-lb^6`p`vvy3EK=~~
zO!_~3U6-eFt<CuB@-E3)So-AhK6{P!XQuyypG#8z>7-thcHOp9=VzzT+k{RCJs|Xm
z&_$tZLQe~QQ0OB<9~b(R(C3Ao7y6peZ8u8#LT?i~A@qRIBSII2t_eLY^g*GI2z^}W
zQ$n8?dS2*jLbu%{=?lG0=!DP%LXQYt6uKt#w9p5IJ|gsSp-%~YUg&wDuL<2I{Wsk{
z@ptO0B)=rnBu`|?QU4$rCAlP7Bx{ZO#I1j8;}?de&<UZo(7S}r3;n3jWuYCRpAdRl
z=oz7(7kY4d`h&UK?^)=sgx%a&ellOWvpCh|*Ww)-u*A3Cbg>a*&FchydnGrTAI}%=
z9Mi0&`@&8)2scW(+e3FL*gfaWID4y(8XYg}Ui9&m?~dH<3-}D&$RXs+%#lMn_#K$Y
z1RJf;T>uR`nl76sd=K(Et5$xc9eVk_`^tABYI$2)>E;)@ezS2WzFp?&ZP^W$t$br+
z1@UPbcC?RVIPs$3*YTPz1(2szd=bynCsivSdh7T_zJ2($8*b=Yd2d-QcCgt)EjJL>
zydlzhU3u$uZx!FSXak6jL%E%fyP@MS&&TTsi1gf+Uq`q$lE6=%)75D#1c8;O4}|!T
z$&LRe{E$93a!1ik4%tLNnMB_q^jmK{G4DT#2a->tcjt#}yosR>nxmJPGkVfMMLfjA
zheI}dg%aWW`K|Yp4!<8>_)OC_=|41NGxI|NRQRno?il?pFi&==Kbi1m9HQ@d<Sx8j
zQ|gCQPbRz>4~_t%{%XRTagDxXku&3%e(SHlkHUk-5EI^v8_!;&L)<3;B^yyXhMy({
zG$xsMGcHxG)d9~*c*;N3oyotz*2e`m;mtT$peYgD3z|gZwh3>_e?R;wUK8Got5@@)
zF6MRQ2}k)STO%F**ML#I8au$)Ay3=@5N?+y(XT`oy?+`{@$lzlygMi1`5Rlrt!H}l
zD#f!NNQF1{*L7m}-1?<h;8-|=em)i6*iB<wn#0%+hCiMEv+$t&oBFf2Y`j^Py~U(w
zfXJ)Kx4c@9;vo_K8#a5{-_RWNO*`(==|2w7c=+1WHnZ2B)}eTYs~`E-w4Z^t@DLAA
zpCQS8@i`{EMy;RpZ#w;F0h6dxA0lV$%q^bs>P0{8qkq%se;Y7~x(RRW&<hei&7Xe%
z4#HCYP56sqM_qh_;lGQpB)cZOvBNG&c+;>(zD@e39WSNQzw}}a2o^0~WD@lk{nqP0
zCZ7L>M?C-5xteA@rwJ{z#hnRf=r;Jr!*9M+WA@FLbf`z;f$?neZD9I_NpYF-*Iur%
z*U^_XN6t(z^sAK<YRC?R$DdU&fKU_ulPR2d$r(y&#-z<2^7BuUK7ad&d6;kpqW4r6
zVXr6tO>>;RUelidIf#Fz-i-dE{+*TnQOV(HGhx!vMERE2%aUPmPY&6AJj2NGd?r3@
zg_-!<qWM=mvSDcD3z_`eqWM{p{~eYYpOX01mKk@F_&Y5#o+R;iMf0yD{_bcVn#A8@
z8NHsww_8SEC-G~d`D+rt7JAi=L;O4pskTLL#yy9v9hT9DNqhwQ&5lF-JdA}sqrc*w
z!&nM5dMSzDU>SXr#CKXok0kLMp%-d#h@XdPc53>4%y-zj*)sh(218zVS*D*R@szN_
zllsTotp9vH&PUJMyy&loTbOagqHQH{M=P1Mv`X6#;_d*yj99e2Anx>n{D?){0OGF6
zFC!LhKZ!eSCyiLN?IiBb;AIGFtVT|9;`v9QSD_x>f4;_u(F<S0TUL4Xa5`eq=j^!q
zXMP#6=reWP{k!0~7m~QH$PtS^Gshk2LCR<0W{r2m^G@KDPdY+J&gdOFSxz{euZ!cR
zT(;n*T=W4gt6X%lp3>cvbaNb=Wtn)^Q}KLZKs-hs=reHKJtgr!Ga&v?q~ibXfcU=;
z{3@*er2ES|pu8#nxn0S0$?hS%lEQZjo?b_DfKz?uQu06_9IndfzxPT!X5G$|`BTi_
zuLlN|>k;PPVVzCslP@#gUtUTls~xwPzh4hL51c+Yr1j6g3ZB+i-{*Mx_0KQ{xm<=m
zxd!+MC6RJdjBmHL{=#M-G#|Sa;J1>Tf6nH2+LJ`?;Q{f`MVKSR|8<-BOgyg);QwjF
zmzAFi@LQm@%zkatk4^|~_IVrohW@7xPIpVE%hiP({KkOz|3LUJikx$1@$(qR-_y6J
z#N*<Ccy7Q%iPF7J%4_C7HQ-clt1C9gz<c5FGz7zv%k`wj-^afA;tD(`&v8O~DYEdN
z2cA`L$AI60esS*Q8b?!+$bM@8Kkb*zis!Ef;NKd6UxP&fs)w4?gG`65{lKXnYWLcl
z6W-&8r}NA&nTrz`Y$KRqn0cA;+Q#^H>r6^MKg77<>e1Q&{FA_`{#T#0nR!xzJRtF(
zO3Ceuf}eR*^UE~Zx*CQgm3Q+i8mGNT<ZcB{<vsg*HofwG38H_>{PdJ`|3Lgh;QwWt
zJ>D(e=OrF<E&<(#-7YLBQhQz8X=^}!e-3yqf#{cy0H^Yra}uV%+y|WHvtcvi^=Mae
zj{)BSeq(n*?AdJrr}D0T&E|L7LqzU-jO*w4@AW7MrEAV>n0mVdc$U2V4Deh$0(t&0
z@ar=4ScBu)Ze0|4F#Yb!1L8j?{F^EH*^c=R)x(+NHmCRjDc6S#{;<YP{ht6%^-xL4
z$yvrF(=qaO;eRHT&;OG63nz8_yssSLKDS->U;C9B-Yn}V|6115z$x9csr+AJe#rCR
z*vx6<`Nz-!sa&T{Yy6_b(*{oQ*s1*gj^NfGYJS?2M~>FDD4yCEY<}l2-+1~d7!ZV;
zb27Yl7I+mnwYxdLb3KQ`$8bPd?f$F4cMRQZ9eueboSOXR0RFEKKlIz#S2RCe%R%m&
z1NdJZfDfS~P`XdNRO9gG=UU)d`Mi~JjE{e5Gpi>7N`jyHs>XK<?ihZ-PYT`z&iz8#
z>qf!P2!BoZrC8P<GcHUqat=7R%fmLKMn2z#jz;-Ed!xqL6F+wVCw_C@$?&_tcc6c5
zNj(%K!2O2*_jNpfCip*c{21>Z)wrk)>muXA6eEX_nXLL;07&^c+tTrTQ^KqZZvVN)
zA%5(BUGT#1YMk~QllwMsl4ncmZBE115cX_PJ)3h_lY-v}oYFlb^1oZ+83RsoRXAT`
zPNVm#z>!3sYcY;=Q}ys^;MXCZv^;-yKs=kkX}sS0ml~&WR8o3%0RP+3p(y`_FWSs`
zi{yVd@T`391)h^Q<K)Ny{uRa{Cn-7n3~)-<zTf6(cuzi_iZ|%`UzPeaa&<3o;;%fQ
z`OO~pKLJkV&AnRVcuaXONIaWTpR|{l-1~NBmund~)$^(Us4<`UX#uDBPf7fz#N$gG
z59Cey3-76h-;=;8pR1p;nbXv#jgFaBpZ5SK{xhG^{JiHMA)gz-{{!K#h&-G1tq<O$
z<&EydU<W&c+(&?CrTa?*@J)%wl6Z0)i)FnyfdA3}eCN%2oK4H^j{~Rt&|#Li`&r;r
z51YTDadRf(UgpQR^Hq(Dcv_zo{;d-l=be0r@UJ9(JSN<85>MLRcqayWmaE^kng1%W
zyJmrB)!X6#{C>uf&(wJO*Z}^&8-V{{0DjYZv-9%-;G}OaioRWzQvU+;V_q!sZ}58s
zKetPlm(C86JH>d$`Wu}+R7-Tao=+l9PbWTS7A$-qP9QjZnnou2p*{W3R3B98Lq;ui
zaOUv7y)#M$-33)v$=s=9)o^A~DZkAhYT47Af=%aYtS|EZY)<y^Ol4-kq@j-KW5_eb
z2~OJ_o5n#}$Le|RNq9EhFm$Yp8wBe0%*1+E41j^Y2AfmSmpMS1RcrB}(Mi*KCk(<a
z?ukI!>aOWIHB+w6D`hpj4i0Yy_~5R>mZ~>#Fg36;Af*~@wbX1cxJ}i-N$o(n-F3<(
zAHr<pC#Lq45OW5TqVMVYTe!ma?aqd!zo$34t=0yD#NPO1bAnZ=19N*1Im(&ckId?v
z%d^PK;R6S%&b*r6yKlxJ4r;O}ibfI$LL54`)c#+b-Fv7!ZPCYfB^U&(R8R`1pxY-^
z(_2{zmzCG)v^$l%>$ZJxro3<Z=utJAAB&oMz!CQ1_(86vYPY4YCU7bZKMDk2bUPfS
zcAHJFIX8Q7tm=7<ay2ZDcHEUlyM+_^Yku8xoVB@mWvzCw7h+M;SEYO@KORLjApJf2
zum!>QI?ft#BH~HxzwqlQC*oSdujjVPb>Kxh#I3@P>xV(P?o?g;IMYRE?(S+p`BCL|
zpbq0&fEEr~C#wLbvI8$nXKTJJXZ6!(-_^0mD{+ipDO|BpSM{LFGGZ;b^%KZX-K1s;
zY?^5P2=X-^FU~@Dad26Zc_tN7oG>lC;5Jm<4Z^iFlP<!5x-FHG=cuPe{GGXKIH3=~
zmwi;E@AS07C@$`3HOsSAwMVYoKx^;O9cR6~=#WD*O-0v;l#d3TQX%Uu5gm!Xf1*6E
zX8bXPouO7PsF|tgQW1)AoUaves#D4tqf124s>C(Uhuo=&r_fi9%~vm5Toghrwc0DX
zk5K=+Yfa6Gx|nXdK>%H`+6{v~Ntj1n|44F|ZiYs)l4=@n-A_xLq8xNWwdnfIbQWgs
zu7=}vDmug(!hA@b^l?uBs|P0;b-q}<Ql~1Gs>}Xj=r!itl@o#E`*VJ^NHwVI@Bd93
zqgfLTZ7q3W=(oJ~V#D1?RS|D<G9?%w@EOW%k(F6Bob;g{XlvJ485$RQBUz~s9q!!R
zcU8;vS5nCzn?Ir}FLqVP6>?K5R?IyM)9rSLZZe?uaQw@3zv2`r-04>1@M?(NYtA~x
zy1=Pa>S#aGbWs~JpB9gbUjYO8U{DytD7@J2wBQxu1|U+S3tft;+;Zwo^yN|kXx;5J
z)IzgeKY=k0mmzrMF#%JC^jQem!*Bwg&W7=B&9ufQotf4+{uN8;#;wuGDS9qa_iaJ>
z>xL{ARxwz#6nA1&YJJ^RF7~PQcLv_2heVo9h0(l)&3B;}l`eH3NojO?*{h%MI{Vtl
z_<p`>$>sX-Gn;pRfAf~->Gl|7Da;?5CevY_cf$4b{Q$A*X~$H)rb`*Bv^u8nXQ}dG
zD58eHqPo~m7?o^55}7geMi_<Nf(z|tzvdE+7!yjgIP8Ix^it`w%dQeu#bxztaVj~S
z@HHA46;N68tK&*qCK!VXUM&{*@*d=^y+P`n25-sY8#7a;A}ht2J04YLpLtT8Ao~Qp
z45rC`ZJDrL!I%7`8n3t~E>x^Ap(%B?nSVf27032<+HQk};}5qx&4%Nnz})-%3EULZ
z?szHPa%Cf^2(v1F;fX_4Sd=Q{yu9fuTN`_C&aixB7VV3!zV0sgYsKOoJkw~)N&tFj
zv5mHdIf|{Zs+h;rNrs^T^&4+$>NiI5rkamC<y)c7>P>BnUQPz4A`ysMUyGQrm(A5f
zq`EWoJ^BYrOgI@pURB_IY;wt4q07>|^6V7Mn3XQoCsic`s)b364<$)gPL(Uid&b*w
zbfn{?fMI{ntLkKLh>sM)M4NAJ%jl+5wW2<f*Hxi?Fw(Hi=R%nKXA}zR(@b$N=mZ!|
z!8D*-SHrP7u1iYxc3hf-aR#GON>9P)or*?-z7e_+wAC{DHg+%9-Gv}M(de1yG^vWj
zY!CD=uAXYdTeZY9-vHAozS~J+?VvHIfRM<lutJ(6s-gZ#g4UHqu4(nLuI~kP53SYk
zf@m(2oFq*_2g1OCshd_1#mOkQDJd_G#g&u?Wg~;?YPaL5HNO*fT^N%Yy(=A2DISs2
zlcB5$fVR6411-0(0gvmVdZS<@T%%)JEt7dO-%eONGi?m*gl9+|hDQj?Xir*;+<D~s
zuR~a*psLsNpSGR?LXD8^oHcu*1d5}2nkC#&Yx4!=R3}u0(lOSy9%}Ma7AAzP_F8)0
zCTb}v-Tsnk$kb%1=Mi<^>CzSZt>0JY)NhmBw!~rvLk0Ja8KKlh9t)$)Sfd<@4oqdw
zt<n%1v-jiOdy>`+a4Vf|&Z<AQ>Uk$p^4JqbR%g;KGqgU0Y$$C1kCq{3tL^c{W;a;w
zj{<r-J}l63Mc`j?WQbYnRLx1X!6t61w9t|1djXx6)od@BRyWgfC3nWD6W)fZqYkva
zS@k=CGw1guq}4w!0I2RtM*r3dzp&^xoBCc#vZ2R~`HlFo@)Wnf8|ZXzh4H-x`HxM|
z$rOk1M{PJ@lr0r$E<$sXM7Pc8trXLbI6Htf+d{Ck1QeB8Yh76cbj`DhjwV9ol~9(a
z=Bh#V{0T#cRsxvmW$KCXded_|XoX^l7mKi96ha76El>AKQr5xNTf4y`)PS{{0cNbW
zOk)=s!nBN<M62ei*YRCmQzH>dthn&%i!7YXR*~C83uE@hODPTs)MW)uFUzqw_`zy-
zrEZ2{(<i89O@DeTq!acD^snP&cTy6pHjRp++#3$WuRX64Edq5-V*!@JumaF1(^NPh
zty!YKxw|3iB%B?+8tK;QUc<;(!cnKCsMbqG9`vJCJB1cq#KI#-PnMQmOr@nZOiJ7J
zEZvh@_9WFq>k=tFmDl5WRnL8(_^Qgojg=dif~`Vxwma^Umy#P;9n8|dPRUc~1-h{i
zlVx1kn;h&tK9L23aeZCkvfZGKGi<n9Qmqw|eE*9sTt2>r_Q0wxnn~+AJvgDjNaN#}
z<)nNffgUgmTRGWEv}_c7fQq8U-JW=+?UZN`M7#0|wl+dYtBJ)+Qz*S`63KWJ$vrD4
zAx)T?2Hh5E0DLMM%@-&~ND+<iHoc5Rgf~8MS>&kNqgZZq`WNKMU=fLJ@GO-CiHuen
zxXmPAtd&TXf|KqlDkSMmrc8Y`wpLc`t@zWNQ7pNMo+Y<Fi_?KIW3I$QT%>3#?glPY
PZp4?IXf@2@>{<T<6(d&a

literal 0
HcmV?d00001

diff --git a/tests/Grid_main b/tests/Grid_main
new file mode 100755
index 0000000000000000000000000000000000000000..97e4f39ecb73a6d05e11032820ead9e17027ee72
GIT binary patch
literal 143261
zcmeFa4PcbTwKu+jMN^1An}QTutHEyDP4Oe~i$rM6LW0lgLZe1WG&aEiLW?9N!D8j&
zV#uveH?e8UZRx$VH}{tIax2x?a^E7NYzQR54-DW3g31bhYz!X}6oMjof4?*HY@URy
zQp^9|-v8@Do|!pwX6DS9nKNg;@5>8L8kCgevVVhI-*VBF|D_xOsjthmU~ex@jw{Rc
zIaiMB0@r!2Gk`S&e>Pr@w%KO44rZIlz~yf+{@nP>8blY(92#-+JaM^@g8JF1(`0(R
zzZ`8<=6l-|rfs@hY2r^PX>EH|`M13~P2jfaAu;%6d6a2~BY6zht;2QeaJK2v^xLML
zKSsyD0_|TgC<iXOb7;77<=boO<X!o1R!ygU$7)6Aj@5p_v`sG=ho5cQ;ch@U(s`;s
zJDWG^^b+x<w|rF0iYk}u(mAv5xax{a=S;tN&g{AK7hJp`>#B>dy5f?`@=Gq0d^3LH
zUOVMx325_i6kMcv1pb&6`ft2C{L^o3zj4!Jk3{}Y;I~(Yb{zkG3;nY2$8cxjZ}Ifh
zS+1ndospd6&KZ;wN}}I5{Cx|5ZHtF)TVD3Wrt~k@p7-EgJ#7a(*ZiigDCgbRZu?o9
zFYkrmh-E1gU9*P#{x|1Uthp!fz5jzC34cEaai1T+Hw%?+Ab33p90=Zw3=9M>Oc2kK
z1n?nX;z03?On^_mrwu^=1qt9^O%TuK1bnzQ0sc1<z#Eb7K=j|00RPbh>8?zGe@z1T
zMG4B~@&tVNMS^(NCV+oCK|G5R<oE6b^n8$@eBVw0|6u}p>Js4pIsyF43E&SVh<|H>
zd>u^yADN&W_a=bPPQcHa1bkbcfX<H+w7b_5(0NvZ_=5@JACw^eXA{7WB*^cJ3Hbk7
z0{rv@@b4vv=lTTn%uJB4oeAPM65ziAzKwQ$(G}S<4V-nk3gzo^eHZZ%p7TaucnRW3
z1-#)+1-JXjTR|uBqpvG?NXN5X!yEd=GXikN6VmadY5Xg7x}}<)d<}mxNy(r5Q8^l-
zj9MK}+Fq5e2YCGb8h;|M&ds4S{`-Zd$EDK^X`ibACw{aaz6<?hW!3buF=N;#Ugat&
zx$~~_xh0iV)8<u`l(<T+ySwDZ8F$XEteP=z;+$!fl`|?`C4qv;CDUij)8K;16X%rA
zol!9DjyW@w-@ra46;-q5&6qa5<a;yjDJg@k3h1{^xvBCBq)?HWQNl<otLBy8lXv6w
zH%$wa<$;*81=C6@=Fgi^a`)_cRr9CKVTkK*s=BhQ>Yj=jK*_uu!G17%`i!|Hv**sP
z%DY>loP9;bj2Yh($4yn4V@pbqa`_KR?m*(x=FBNCQ&{C!&YUy9a+ZyDl@qOOT4fcn
z5FGGwkTql4T}YyG#=NRP8K9YEv!=}}shT%!c2y<l$h#4+%kP+0IlIip4iuCWWR&FH
zbX7_I^*3E!BBT>(-n6;X%kM&7@19*YV=TxQ!jWZg7U^Y!(|66D3+^Z?`-jLzh&yLg
z-92rN&TLiI?D9(Ft#ZZ{0k9x9(}rJ#oKBlFBMzvXn3<P<eOa8j<YeX8vhuqsrp=o{
zAK_j9^hn`Mkf_--r!p|}RP&!vQjjSOK$2CG(3}|yz!nrN%6;}#v+tZ+j+Awo&8y72
zakeHQOK}oxpFQ_Z#r%P|F0~{N*&_GI+mZlS@$HHkbMGpj&e~OQ%M?^B)S?B`?wEad
zCaXdjYjFvy4;VkQ+^%GIpf=AfDVtVRHcMzwweqrzlAB6q+*MI^PszL)6=)UZ^QPT7
zgJJGmu%M)3#=Oe%xzpy%hWFhWI#}ford6OeurLd50ZmsrvvSA$nThj(5I0p_QC2>`
z>g2GfeF?%UW|q&X%p>#W{@^YN+$Y<YWtNoOeb?*(vW{XXXNA%gdphfJc>$U|JbLR@
z?q#}+$}g{&UzNvtQc!@b{Z*wj09h!R$^ogmsS1qOjdsqAxp!90Dw#2FUirLu9qn7?
zQ2|(orq9X0o8s&)CCTQL-|4FS{=6y|EOQ9Iv#hM7l75-4lAGuLVD{YU5ImLh@0vl+
zG8Jg1hy+(fdFAW{C3nrZ3x#{<To-bU@+_hFgH)V3ZT1{j1!`v1Of8|%SQpH$a#ha0
zQ|e0Pj4Btx(tv1XdBRmzagPh_wrciWGs<U9pLP$ir<GOBE}!eVV_MnwAW6z*5bv(?
zyJz%5N-FQU>kbqy5+adWF5fk6_FUJ@c{66XW>;3spIcVtx_08klFKj2a9tY=Oqf`5
z*(H}<a%J3}d4;kT76RppOS1d8GP7J03k&@d0wtN3T;9h=N1#JxW;#9DsSLunXpnqq
zqI<zl*#=|eWZTgF!cN~o7!ND_K^Q3|3kKbv6TiWL5Iz{A*g>wL0;hYBYZx#HJwrm!
zSN)^y42<29Tru>i_IM)yYxfLuA$%+PasKGv?(m=g#o0*kbFSy&{-6tXhR?v?(x3hU
z!_#4|T^gqF2RrbYroxBevhGVq7L#3XY5ZB;zl8ft*L&K%{gLm%2HFuT)2I>rA+EEv
zdz#?8UFT_cRr=4Matv~fft$bX-KyWGlbo`DwjP4WWqfJFZM}kUjydh0t^W|tF{%Bt
z^$NmsY#{AgpCLTk0k`!W!pA${w!T651P9#Ke+d7k1KyzZ6T+`{z(++@emR!bs*d=k
zIpAP${O5DPRUK4b84fs#KmN;dz?ruF%W=Rp>&44-z)6$+%Xh$S9g%*84!E<tN*!=i
z(D-kb1D+Cxx?B|w_*oA40tftT2R!6}8xHsb4mj(d{i}DtQ*9v3We&J=Jlx=b4|m|N
zbil81z#AQK&Y{@9H4gYl8wfMvfPcXOZ*#!Ub->#l@Gm;xI~?#YIp9$T+@7nWcgz9*
zvID=%0YBdX?{>gBR}uf|aRKXyKMsUsaKO_XaE}8%+5sQsfM4i<r#awXalm~JxYq&C
zaKOLnfM+@2Uw6QB9B`ilZaU!U4tTx;{tXAb&;cLgfR{Sp7dhaw9Po=B@CpZfq65Cb
z0l&ln4>{nMI^YjD;FRz7uigRAw1F^}IpCK$;0+Eq=Vt8RN(cN38wj(}0l(4#U*mvZ
z<$y;V@UafKS-mgW3=es~ZM4g*iBt`V{T7~P^~U682NJA3SMC8y&sX2T*SH`DHo9jK
zsUzBh-&c3iO_AEsF79n~Q<Qc@#QhZAgXwM*_XfHtIy)M~y_Rl@%#M0-|A}sj%8rn@
z|3EjZe@BJ5|DA4%#*R{P|1;ebi5>ak{u$jv>CO@NkLjjp?8p%J{d7|#cBF~>hjddE
zc6h`+k8X;<4wtyUOE*PdN7o4ezIq4U6nPy{ao<WeMO{a`xNoGJBCaDM?(68LXzOSc
z_XN5rx;h%feKp<Z&|NR?%jl*E>j;Va8+222bySG^Lb^xMT`KM`(oGT8kuUDg(@oLU
zkt6P*bW>z?WQcnZ-4s<FY2yB%6>f^C4v)AG(M{3R;S%>dbW<dCbbZA9@1Z-5?x?tT
z(oIp*(Jt<7bW_B1M8y3R-Cv=*QQRBorby{%5cgWTDM~u(#r-F`zeaaR+<%~(BBi53
z-2YBDMM+1gxc`}Mija<casP~NijIyPasQZZij0m7ao<ljMMX!NxPM4DMMQ^3-1F$B
z2<UK$`@3|r>34J;m-45ZO}-;4?px_*Q}1XO_l<P3iFZWAeI4Cw+8vGJo<KL7bVq}@
zucn(#xuYKL{PnBeoaTzvAnEm-vw4!vwe4udg~Ffw^^yJXpAq+`Q~YD!2h5lnF;a&d
z+=f=&I^;bV{H>2J^ShpGizAwG#Mcz!-ad#?8?mtCd|=MJ#JFII#KZQVzrLI4zX?(m
zeROL2$e+yp0c-1bOlyx>{YF=QL7+9#{%b#oX(|<?HRrq>SL`o9d~c?4L9WZSW;e2S
zvw86j7-n_%*{1d84+h=dyoL!<FuiyCdV2LljSE7At)a`tZ|WWleAC)sHobF=nbdA>
z?VgXoYl&#qY(rqvYEj|$=<uO$KDrrPnV)(yLdF)MKyPmu@*9AX4mUG@X7x9BGy$RN
z{5T0CZeHJoRWIQ8wBO!wFM=2sxH>A8d9#12{}%tv{(_s##rxKPPu4o`s(DD>TJ3G1
zS*P7~${pLa00WD!%}xSFvl+fO*&H)2FY}0LePlL$G-#%AFf!A)Z;3a=psh>1^)UIX
z`jWW@3G#Q1nZ3#QQ71APyA;I}uv&shivegp05eeMe%Q2BU^9HV8NMmmOh05=+s&pA
z2UR5pt)sE8q6C2U8=^4?$(#lANjDN(OX(fEa3%<?gJcg_t%2};$$>G~WbW!uXFkJ%
z&Qh5CokXWeI)8-Jm|bhRR|3E+ltkSRkpKn`gvSQLQx%;rkQG(QrnNiv4<90AokEi!
zVMT8+y~}Ls9)ub*_y=DQqRsRcW`cZ(T{|hs)%kgN4w;t{sBZDtj4=_K2!?OSd-o-|
zj8s<;#p3!t^RHNF)-4r2n02*6DtYzTWAtC{Z7ffWM_M0UPyZ!ekKnEI7Q&SFnBzf8
zrqm~wtLqwOIfFRhTNX_I2+M)JWr;VRAS3l!1Ui%81sXNvKz%@=E+dGUYk&oL>qbtD
zN0xYx!$ioHfT9T{n`K?Gs+*HsNH3Wo@FS~ej|Dd=qPym*n4&PvvQ-jQ=4P{IzvKY4
z=7^SG^$pXCnN3l*SvMlWNE+`&r3oR%*p1z2{hMndrjc7KVAHDgGFr1PcfU)*B(*U^
zE~HDd8-7Qc45Q{rtVoh^Z2*AA7!3ZHu{8|;smN(QIsiFLp+7l02m?9oFom2;H8~9c
zfSd>no1A*$elj`9Cv)eDfyiMBHaW?r;!Y!BEIgP(&cXj866yf}AZHm2o19t1{Z!=q
z?#KY#VG8}pd4MpG6M`w^lxlMF0RTA_Fl=&ii2KRp3{ZDQ4nz)9u*q?2?vxS+atdJz
zIrpz7(h{!^0FYzCu*n%k+)qW$kKP}EJ4~TJIa!2(oD7&k&fz~Ra$Ep_oHQ8xQSx=&
z1ry|~tZ=#hx|Y@b-T>q<h5qDt2m?9EFom4ynw%X-8{~Auu*qrr)a1+_h#aQSpPU$B
zASVh_$a&#UL|Wo)1OVi;!?4L|AnvE)&d$RFaEB@MCnrJ}$XNqZ$eE<cc>n;Avl50)
z&I00=Esq!0u9L;!j)BGC^#hT^6wEsJE{7CcMi8i}hbh#od6YQl4FLc(AsGBo3g#0R
z)KI>&>YO??PaGP6IZUA+H5CMbnprS~ns0y_P*Dgse`cBLU6^{L>g&kI05UHB<ivvB
zoOpE1To&SEj(ki{=O{08|H)YzV_G(YbNco(@4Va7gCI4Lppm;uI+~!hN+L#{ZX<_-
zNipWdWEwDYVET`2LPzsKBOsu{0|+vg0fRph<RWsMpq`Tn`nqZDG@JIE(U+Lpg_s^8
zMs<ckD1eOLY}$K9FzG|K{4_dYj)Lhw(h5R;tO<$D5kivTVH2_gAe)enDQnO%q90?n
zb`4P0OhNPI3@t;u2?AfbU`pYfqp67i0JcP7*wi!<*RFX36cf||yO<(<IWv*epp>8v
zs8TYmRbHk5YIHl0Qfem%)U?6Oky3i(VIrZY0|ex(fnk&L0PMlC)v6)vnp|_FU$gvL
z&^m-dLfv#~CTk9YlwCp!v^Xd%EDXjTXnHF-kn|eMe<fpv*#PrMGm7ZkM@Yp|Zv_CL
zq8<i+Buo@WkX*_zHZ_}VY6h9D6Ffw%`bN-#mO4;N`5GzeuPZU7>+9S{B#2T5LI9CD
z3t&o7?fD~;tXttV0RUCAVDRUpDhsgwRQZimw>Qw5>s5qlHlZVSvgxM>1|W(Fflcl}
z(iXC(kded8hbaX4KoAJYfm;YETN;;A+p4~Tvgnt{MJFc`^rlHv7a%J0lfv)Ye-e>F
z?<fhsdlh-J+FJqrlJz1T{V&L(>CvW3n_b^i@TfN1wHeW7BTO=fDl&%#QFN%NTCpux
zV%&yGL0#PGzXtxX?_WyK-az#_(0`8U3!kOoKkWx^1aVAcwj|OU&jT7h`s;n;Db?^m
zzj!EH8PB6%>5EShC-JBCh40YuG<^YMy(d|UN%%9v-wdrC?TX!osv1@;4$bXB?sGG%
z_lE*?d1;xOGq+)YG!wu09KlSZyfA9`%&uI+mpn7qsAzr^pqoSZ%&*61Mgu<o&^Xhm
z*wA1UHN)KYzTYTnKV-CTfa3sw9SpV=!75&)*EV>)V-&{Ve!c<jtuy^b#k=(0_`Xrp
z0^fF{@C7(qb%bv-R21-yOb_GfihUBRQSk<S_P}SK#8;lz4$p2zdJL9`j;`A%{0JfQ
zqJTvaa<BLSDT0symgjW=vI9OZ+ddoMv%@HSg`O?|+Zg9&@l_<Y8HG=q))uqrM4DN5
z#oZ=al;0RraHR|1g;#p;4PKe1+1#_iOgd&R-c-+0sE%Z0A2MoxhpD)Lb%G_Z5_75M
z;^r>+2I{WOU?EJl-nZvyK*0j!s=Snh=68XW<!R*S?nn83UcOH~Ix`R4W5N6j=R9FP
z#JNzIf6h>5m~*F~zZtn~g}IBI+l5@R#9jpuS*_Sl;5z`mY838(y9ID5M8Z1&-((aL
z|6Ote^F0-t>Cpy{ca1{uu)I27rP79g&%r-J|7Q5N(X&)}lF>{}JT;@i`10zC-oQ`U
z{!I)FUc-N42*4doIv$z@h+<;siS_UuFx`m_@QESOeRiZwG-ecbFj%7w)<4B(=^2rb
zNHHFiDcV6%ED2IXS)m#lMaN;rvx#bmT3J33jFMpeQv}1x2g9$OS$y6O$pq0Jw%V1g
z>8&$Rxvm(CPw`mRvpHka@GTgd(X%0#)Pah`){bhHGfqivqxLb(Dg`l{K)K+rUek?=
z<}x1q3Z~oxiz9TI;Z@!kJo7T&2*S#K%d|T4GC411cPT;Z02HRlR!i(UY%QB?bp@>`
zW^B?7AJl(#!FBE&baZDJV{(UJJ_f^V0O|11(XJhW9@UNv`gCbW4jrzADoj2d9_=Wl
zBTYLh=*SSqk|7~x1nwN|W&@);U%MN$yHvXy;r53!{5`E|@`hqDkhBpzi_o@s6Iagt
z*+-4qGf)+KMW@|CcP0m+El_E%Il?CT;BG8ynBZ)L91GV96V19C(lU==GE~j6nswu$
z0KDx+;ihEaQ;_Asw>T>e-|1NyDCQ>APs|XOf*QCsC|AC6H7i$+a&1(u3~`mu$;tu#
zCIw4Vu&S(lz*-cHTL6*D!mLujS{00ygEAyHtAb)TqM(#Lgw|$-XoH6)DoQ6@?+c0T
z%C$kcBFeQ3E(l_L_VaUupPu(AkRtNk{Jv?u^h6qvWqz2++!;k@GL=2b1;pF;C<~4G
zN5$W0_-egO2=-@D_nYV@H{gDX*oT0<)EiM4F9PE!qp(eCfkJ8~(ub6V+Y#g?AZ#!S
zo1ipYRL`o$>Z4Nv$}`SX$U{lRD1QN_Gj6gHAPB$m7hyVMZ9uHKM$rl{Sqsa-LwQE~
zo3JePlC610Q5PXE<{5?01G?NBRnTLE?gn&cu93d{B56a_n=;Jm)(ms;M?^Aeo<lET
zmYpEc;$JuZKGY?&F?Tj$x`|~wn^<-leRLDcb~dr>9Qx=cmhEg}*`@T+O)T5l#Ii%A
z1D?8xW!G!BZerOD+O3;db|c(oPm4dCWiH+jfg}5bQQM8_;+h`g!AACUXfDf~%>^A|
z=C)pa1HI{u=uOW6r`?&3hEX2OM8mmH)dhGKWTwHn)SCmY!&#h(rgYzODO*&)2+ALA
ziX{qHlX9u{P_aR|SX>C0o0%i1EGvl?4G7Ujl{1)G3a6@&@ST%c0Vjq%%6CO(h!)i7
z6`5oVZObzov~6i7nStbYF+mJeluMH)SZMBw7zrIGG^8*$(J$AFB&BbM6~vT3qnZrA
zho#^D;88A$q#+E~JAg5%9iN#|1nX+RC-OcTTCKD%)W0o;Z<X-HD2g4*HQJwrCnTn-
zyDju2o3!U9qiD5o*eF8H0UWKV3sD|tyl=cu><qy^$Viw$G+^Mw{ODZ?>Jm^ZTWo_^
zpJ#gndAD={C1Mn<6Hc@FkOlBOZWNwC<cFBTHl(mZgn?1?mf>6O4M`k3fwtEue3QvX
z1aiRe6?-d;BKmY9_Vb|*8b$9KMFrkcfDRdjM-d%`X~p*UA>PVgF^YE3{KzQ!00_IO
z>cP1edM3>i3j8(_L2KX8+u~c$RL6@PNnbIZEgy1aylT2=w$&{;=Hi|xNvf%mrh7>E
z0z9oGvgu|Lrkif2v*~6!n{KAF>1H~cZl<&8W;&a0rnBiv-DFa_=}HZyTQ^;)r*!M4
zE49^xt~!JT?H}hcYAd+R1BF0M?);jn0++UGc2zlMb0W5epjHjei*h#Sh)kfY7OCYw
z(t_dlAAW-sjsId4)OpWG5|jjL`siJ=x+4UYB`tGXJxVVp%JGKb^PoLJaYLa+-)GZ`
zK$dPItQ`%bd>=~X9oFDYd4><Q80tB^UbekDb>%L4hu}Sr0{abusL35;oE_+>%DarB
zqkhA;$0&Le)!;qD_qtK^HUPVfqE`VxefR(Y_rZ!0M8!^qdk3S6W+eB*AwRp8qNibh
zmG-BNqV2H1NjoHcx<7<ZpdO#u4MzI+8;$g77{lJv4xcDIx{UNx#Srj(aTBY*JviuQ
z-O;lu!K5xuv~Voxm^(4U>qP0mu$U`b)IW^WBo5?76~PQc8)YoacA7OkRl{O;?8d-P
zO29084U-kSs=lVC!1|5oZ+)%bh+GEao?p9M)labepgTT6Vs-JUa^v{ich-H=F%q@5
zILD8MkHg1O7N;?FKN#6*OoifI@jle!!!Rivp*a{uo$z}f!w6$4hb6BVQ#lyx1Qtu|
z4PxyA)+-W2;Q@uUQ(^4|)*c{iv$5h5BK>P%Zgq=?oE`U1q~56gE_6Fg=3xZ_BREVn
zD$eG>;E$Z;90@WclPp~HqMK7p`PwX`S%VeO&Z@Iw!@w>xyj<c5S2GDt1Q`!rj?OM~
z8#F=r#mCcRo5b&PCqn{4H!_=g(x@RqG-UOSsHY{E^p08G<jHPX*s5ZE6-&7iYh1r{
z##j7T$5)pqJ~R$@GI+`%6-IB;=FpWu8lnWS?B^gm<zvWhRAhfml9BazC%xza#<;O<
z$;mwy7;?N};R`BgCqxAlgarBh<nRv-9RB0;XN*ZnMAnmvtO6Bny<MJ&FNOTt3Vv;c
zNkN41G2{dIwbdxx@G+D=uA)qmC{>L)UG)W1aHp}pEShH5=#%LCcz&LGeF>IzXKv1e
zz`~s72PaQAKJoHE!l`l@8}3qRgPD=)wIZe#<)Yf4d{#?)BYjXR6VzuTJtc{l6b7t-
z5zn|^Fp`Cl09s}s={Yw4Q<M~3FrcKMsU-!(#2X;ouIRzC$CaFzoG_m+>=-DY;a?7v
zPmBeI3@J%MIdi<O^J`4qjEa|XhB37*Vdl1<p3Du2or%op&SB9tY!U3&$wf|P^_PFD
zJPe=vY$IJU8=6y-M~UZ!yMqwG<aD6yUlo(6Zw%a1HNG?ulV<*6ASPW*?&Q)Zp8gp|
zx_jVK$HI{G4=ln{EcF5U2bB5&Q21A5X!nZ)W$2&&X`l>U#0=&9H|OX4zb-$<fce?9
z^zS4;Lt;}vpjjrGt*p%@sA$)|Fi^RC^z*n%v?wDpA_9o@o(fQ0J{rD3M*0~&5v7G}
z$QV~6as_*CPJ;##unwY<S)lH(sANBUexS@`ATuY6(6PGt_U7lmv;7Q-Jq%h;p?v;h
z`#@Q`{b%v=F&@OWHRLZYnbOdDL}h)ZNU5y0e)0cfeHEL+zXL{{0zap28;GB0{qz+0
zxv=p6C4Qa_Msb8(iViD2I^qNh=h2r@2|$&rpow)vPFrX0;$$*s4D|k0v#gn09L&0l
zpu4|@nXlwv(gExRz>@kNST_W%w_|PZ_eNwkz3C3rt@4FXpP?b0JidTtZgs5TwomEr
z4BxdOV~jrpvmRzW%m$bZFdJbu!h{m;kHBn)*$y)bGYYc{CdO@N8e@{|0qWvb8K?HN
z&HoR~@?lLY2Z}vr`oyR{;hrBdd$0>uF=U*`RipM?47k|Xux?xbU62Jcq|FL#mcr!r
z$p)DGaZVvRgVPN?nMc-!kek@gpNs1l{&mFRAQWTKFABH&ae2SZ<$AMhlj84r7%3NE
z-=g&*7wdvan3az+_}YU`2JZ!fmyZe|_Fe|Z<vBH3>*fD0{x%Q5-%S)_tZ?I2vMW*i
z6`q56!SD+20x{PKv&qQPLYcR*8F}3n)LCx&jHS&q=0*-1scP~aj!oj=WT33>I1q9G
z!_3&+je@~Rzo6AEQREtPqnM1Gdys1*4<M-^CL}6gs{P<W>#3mC!lk4oAQmOdV9_C%
z-r&rIo3ve5cM&Q$3w;jCr*4UxC2!W9=R2Hhcv5_3_0#UdhG!W5pjabi!}ro*%*t#9
zKv8b^b{xjy?O~&MJFW10+3>v$(Cded!o4$%t4BZYF@}#wXMDN7WP~g5rC{E(=P+iJ
zx(#0xsC(d9e4KU!N9{id*gFWm57S3ikN(J$YYbnQp3E2ud>+Iw-8TX;9Aylc<m&?V
zdyL@-?TFzJ0LKyhgakjz;8p1ir)Nxk?-T1?0G6u>R^NvRaST~FYQPG=V~GA$M2URm
z8il*T{HsSFMmDO_M;R(K+9m^i;0T1wpbU#PAZ4ubUOl=?0;Ca&C^s0s!|;6%0Wi~h
zSm-o-9|5FDMSxZW$W8ZQ3QR&Y!2)y=THM0CKEpuI;!}v#@T*5R33`T&jwuu0GeCa|
z=uJf5Nc0zojydV8M{f}HEE^pYtAdUo3jKMaze#jV=UhE{qoC)I=|SHp8U?=dr1aA0
z$n|-pAbj}F@m*xLM!2QkwBGGJw^v^Q;-Y)H`OP%z0;7Ps+TDl-0?Ru0YFI!+6G(1>
z39^fxBQ3CRrX4ejI}x<#HGHtj%pz#^%ruQf*!upe*|eVx2x~%DU$`mhu(2qBrRX+e
zI1C>OKw@mOW7Lr&Vpi4HP(GO6N*o(F&}s~)6h?z9LSa2?M+6gvRrEe$z?}BgqgxWt
zgBV&6Lky#^QAW)X%qYrMq`DN$eZ7w6SDV6ChJ1N)n=)two35Zy(3Az%6gQ2>cY0F}
zzH^##O{@Je4_YY3mjJ9zp}2-+UCMODzzB)7nWh;iSqBWwwCF`pi%AhyfP~z`j4!!o
zVqrkB&+t8spmupk*%-d<ry3hp+Zh`g#HZyTN)$76h$I=q(^2H3S$(V!n+`cUizVd6
zEol%&=={W}?r|G67vOY5pr<8J9WevdP53(j;0GT7U-xI+dqk0^y0pfpwa-m`1oVHl
z_4I(dkZr<dh-aoC5b|3jFm()p%(5Yri~^ZsLnyNa!d^Bmw*^vVLtHivI}O1pB?J{=
zwRa&wuq~DDnzmKmC4vUa1In`6TdOQfrMp24YHb#&EcY%IuT*T6M%@8qsdt%zsP#3W
zWVwPYk<}xCPzMl78dRty-W4{a9T0Xi$pxyLiJ=Myv3;ux3=HB#M;F}v=36IQyC+-E
zPqv;7T1_$^B>g`c0G8$JR*#P&<aKrD>Hg=uT_~vW(l88<`vQ<te1V%8S4DuD*^c_x
zPDrh+lj%B-idwuC!eR$}igsyW8|4@_u3SBOE9BUS^MwH{B?xwrDv$)86<YdJYCet7
z#pt<M)t@$e&r4ElIP8ITvI{9bpPkoIS#(1z?6ZoZq-LG0&LLqm7Zne>zM`EJr>$sv
z%@ReFQA|Idy-@B2l!VhSIM2?KrkRp>E9?|fuNc1L8jQBwo0Ed`85XnQ$ck#C@Q!DM
zRKJ8+o>Fb~CFHd?E8EG9r@)Qi`Dqd+L*iiO+mIB(sjx7Q+q6qR)+S+m2$Oq32EGd~
zxB~c67@9=N37hiW1Cz~wpGY>6-aR0}SrS?iOu^KjVCl5UYzM(P*fsMw)&ZdPJl8o&
z^l_BdqObE)(rwn=2a(c3Zk~a^Db#Y7%Q~&%T{~*j2^R4&@S>Q^c^Mgd3uY9_XuXE=
zwB59z<VtVEZuY_#*c+9gH`-|oFWG@G+YqK5=BtbYjM@(SP95KyrxG7(@^)kRRP;wL
zFe<}`9!$jqlYLVQC~9Kqwu^*!P!u6Wo01U2Mq~K3F|+4sqvj|?XwW)=v5jztHF&u<
z!sZU!3MDS2R!22ViwD4Jy$yu1%2OKD$C-yLv2$QHU$KdUeHZO8-+*~aY(qChwy_%`
z%NZYAgUz<xIzGg8YVn~XVtn0*hg~6fCsD8zSjDMlu-1Gm^H7|IbVUZXhe(|fHoB<p
z$m$_G2<%@{#ztkd(lDz(Ko@jo|9)u$`le>xKke+-KT#yA{>igR^-rbW=D^h+KMdVK
zD8_1l24p0Lbf6m8`u)pPFi<Q@Qn>3DmfDon+<t2{t<42uKlu#*r6ZP86+s;s(Ey9q
z#|>WxSknnUv4)a$D2BqFtYdp24W>mT<Z3(Qno|#nc#!5ZcF6Wq57{LlSJ)v}o_a`%
z$fYCJ*&!pR9ugG|A)mKH?l|?3-4b$@9kTJ%Lqf13&5d@*wo?xolaR02A=#ClT6vB(
zt2dZXv`lmH2UtbM4vIb4w#eCpA*vpKh<g05+{uXH<%m0#9#DZQ=wHj~7hfNoY&B#1
zL9hO=w(O(7U*eYPI00QqyZbPVly>L_7@lTsMwN_@?}IqFYi)m$LR-me!*kO<Wa>7I
zw%DicF?`5lyk3G+6b*L0q%@A2R;$@`kYfXp#xGGC->$^U?W{pJVMOJ*4VvlS5}y()
zx3i3I!hpc@rN1pYmq@AGAupOCUp9?0d$t%gO1?C)WCGQ#D4k;{gI$o_)NvO#i<l7=
z_m=|O5Ln!#V3awPRFj=*FUm@ZmB%95=Q^iY`STM>tfXB}84)iJvn~ZwVIokLt04<|
zh?F;2!PrBj9M`thDc!2ltx6d&Q>4l26pym3OWCK9mZVT)1h<x?xWx~;bt$S3qpD4D
zDfA^N=h=Q#aRhsm4au<~?@2@mPmMrusC<gvN*)L##SZ1MA;~txWkb9+MC-^)QqpWl
z6m<|%Q>NhK{E)u#d8x`Edg=2NtxyH`a<sh{mVLCmgHQ3B9QvS#X{M3}6KZ8)n<#>>
zQxTlT!Fr1z!1A2Nkg=~go(A+qLZLj^gEtH)F}PKMI3`uG7B^`-l-2=4uqTa3I?iOY
zovE~+oC=kzCJ6&ZF4a2GQ;zh+&cItw4lEk6QmlcXcp%ZRIH_Luk{ypmW*++Hl#>Qf
zMp&ACqhX1kdNeXxP`g9>M#D;R>e0j$HLMeTqhY;qMgu%3{-^}c<5Rd@CM`Jcl0;2`
z@>Mi~|2kQ?{tUalC|t8p|2W9%Li6<Vx>0u7@%lH}+A!IAE-!PJp1%%S?G&JKf1Dq|
zW){fXb>nsS?}lpb2|nk>g#kgU=Lt$y2R{6!QYe;S^Ho&IS7*PEzM6eM@_~MTkLZ?f
znLHEr4(02M)GfDBzCt`x&AN?IO{GvQk-0pP*PjsI9+?t8!1%Du<mwAwl=zf<y^T?Q
zfzi27mAnC?`;t)iC8fPV!uC9C)C>~s4bnBDC2NyN(dNKn(cA)RVj)n?j_(&*=*A}%
z=J%7XQ52SvuD3m|q-*IUr*yp;Q)5)xC|tN>tO355lgIuD>nUy@#SDAAznc<4hHu%Z
zh6rfy=%$MIq8y;Z6yngr)uUfWb-IH{Q&2pfTQE@MU?L$Vk*2U<Zb2D;EGChrP*u9+
znNPqZ(iGGe&n<6!0w$5Buo!PajriEqi8Q61n9qL#CXuE<4e;Eu?-MYIbT$!De?Bhp
z3;Rtc^*m?Pe4k|u1xP9g8!l@?uwTKmm{q~w@t7*&=r>gnm)?#Fu+8wDb8I}W#_)Z4
z&^MD@EblGQPo9XEH}=I$3h8|@>O=~93~uo<(!ULO`J~1Ko`9E6iYD*`ynIr90#Crp
zCq)u?0$x6;D}g8A<&zo`cmgh4pB;y5db83FPkZ}Mzib-j;#RfH(jo@i=3Byg?1imI
zyRa9(L~yRN`Cm*fV>S=mUdwYMu|0nQ|Mk*L{FKc&tdCKca_!ImSO~EN2V;2zt2Yp#
zlP%1WoOA0qDV{PK7hv0qUJo1};ikY~(sMy;D_2m!7}>vKJ!@Kr*E-8DFElVmt_Qq4
zuQ7oq;N@~8jywGNvE=$FJ?j&M1iU;C<P5~<>j78i=?03ni=GV$LISQ1@C_8Qou00N
zLk`0M9#w#9IkM*|qvlO1zzAAZpAzgw3t9`WpcTp7togaYv`#!8=cnPDfc4TzA(&V%
zom3AK>!p(#U}C*=QX@>Pmrjbn#Cqu@4Cg0cy>wC(Ce}+QVU#e)7;{EE-3^j_&(mBl
zZL%xc5TO>@^!&vEE|)?HMTLOCXFbXtx2&7Oxb_gcXe!WIPC;J#C)LVrL7s+7Hz~+_
z$KRjHbKOJzuaxI3Z6(it_GeYnr5XL?dHPrp7?;8LOrD=Cy#6+M9=+6-=bvXw`xNs0
z$2jhd?)K+2K9lDIi|p(FAC~8(W1RAQ-KxIwyaB$S$@2k69RDljIZIo~^9_GgB^|x6
zpFA(h6oGLEjL+ox>6YiE;I`7wC!nQ4E`NwyJ)m{J!OK1ROZSUENn8I+o_{9KPtwo7
z?3CyGe&1J~r=9<qJWnu#^uJP`M^V~Jo*()TRnn#B^pocn?(c!1`5BDQ<oW5A=g|?i
zJfDY_#$yj0JNv8GbI^?K!1Wl(f82yiE@Xc!J&9{3U+#kW>E~OxiU4WYd&iBD?woNM
z$h(oV7*P?S19`*8o^=_>yPn<{>0{nyAaD5CGcN;qchS2cVQBc+vo8aMcIfK^cn>rK
zvpw+)%*x&w7%Xu8PtU)^>rnVdtV4Pl#-H76)b7R<%;dVi5X>XJ*MFjVkn>r6Qgg@G
z|G!(GbOp*AcfL`302Qe73%UwX^xXATC3@z=<Z;kO2(M~64GP&c7mIDwNh1Ja;g`1}
zoOIq6%`_0gXEvq5G;;ZdMl9;$!X7O7@v<(tlj~_-b<jlkQ-o7a$+c|T35Tm3_gvsE
z49XF#dG+1~hGjO~1_mhh*{JKlcr5=_1o(agkOTC32bLqe*y}L74#4Xkh?a_@^g0Af
z54}Ew7q8qnPUr`)$W3i~0mYSA<<lTdDmK!K1joI2z1wI2o}mv(j{7`gpC7IpkL}BE
zyK{L~9v5I`d2riWRvMhatPEU*m6d}Vr?T>K9adJUF{Uc3!Wg5jG~!A6J$fF0+@IN=
zjoLp^Kw*A$4SE#L<Uate$G(Z>bROm8I=>v`6%*6^Fj+Fd#ia`MEFd@~DBKLnivxJ#
z2ZMJmw22kBgySqs8xz1y0|YTtySTA|irCmeApU^w(D0SS!(kY@`Q!Bsc+cb4AQyK`
zF)bX6seVj%co$(oRc<rKnWBJo0C#wChgEz%b}e@r;5KvSko(>P*0Zs!e^ohe#MS9I
zbi_LZhx`Rro{=Lua^4}wJM{HlM8mb0y;pky(0jF)cBJtxk*IcL(9xwGIdr%xRhWD_
zJlau8N1Aq2(2*ey$JJii&1rP{=WBO^c9&{*qjpzlcSO5G+TE_*_1YcP?gs7df;+sJ
zEcN$n3?#i2SlmKR__H_S{xnn+xzfaVa6MoFYrkBLwgkwEs5BNNPBlqRm_6<GFmkwE
z+U?TrE>tq&3xnX#g>qieBezxJs+@LlFZZh3EMbz7>41v~JTVatCTp;|8w%wMcffi+
zXuV@vPsjE=!h*zd^vl=-MKS~pSGM&nOCCl0`@5!%^j2Wa%owv|NJO(aq}}nd)b6PE
zZ_sX*CGL0&B)!Nh-B4n=<D{_|wcn=VffG`tEXK0)yeLP<Z8ihkfIQ__EPCg4!0Nu=
zF@@y3>`cNTy<`6B@m|R9`G}VHg`~j@;+#2``FJM_cX|Y>4}^Fzavkn+K{3VGYq&oR
z>+y0nP|ojLE!Zrq)}wr9U2*}&5PjNn9^?y!_@q#-F|`S<&2&v^gln@gm3M325;dl7
zRrjNOrydui7*pQ>bY{eu`ntM-WiIYm=`f}q0rc($W9qxc)E-zDwHs59F><3Fmj*Q9
z+L~*TMEbYujr6GvMtTYEuemc~q?hCVnjb`s^!sps&6!xxst2>N+ND-bv9-52z9&^c
zY5`SNPA`Pd^>xKK;?=YlO#4t~WlZa!NG8<6*uVXAPmc!fHmw7(pTZsJ*~F%Zb+Np0
z7kQ1^XDPHH1<7JcT(Y(kS9Pq@%*H88@ljV=nUKw;C<F-8Supt1;+aLm=OS>xpH`OI
z0&Rw9;&vBE?i|Dg(d%1=M4hL^6W@~j<zEhTa`ulmBg*(S$!+nIqA!D>%x$DTf<)y4
zB`66)vhbL-y`E>6aByr0);3kd1!Eii##9s-ugWPN8--(sD(2~9>)}8R;8i<w#zx?X
zsiLkLi;Y-QQS>bAg=5jjPVH93ojbN24%831F#<PBPHjWM2Qx!PdO;?Nzc>@+KRpu_
zU`{4#KvgEHz`{(_f!s_~f|N||K$c}z`-UA8C9MjD*&5*bA}ItM6oIdiI*80yi0BI?
zK<~9hLt+p80;gfH6ALAY?h9KfYQ5?djOSv{dEnF3XJt=YlLuZ)<;;pP^#uk%i6Ve_
zK1EM;9>ti-Lv_Hcw>`JhbA^}3W{j!WYYapls^LbLSCkhIy(lkr6oy{UD=!{jQC_R%
za0|WOQ(in{qr6bZbcDy17thQnFX#^1YoqewSr_GnO*-1^8Sd9p`G=)e$b$uUSrd8q
za>*#&A#p>M5mTK2Wk?>oVL(R68<n#>)FIx`JG%KM%9Ts060DL#x7@;o=6C3~dZq(y
zG-w@(4MNoxeQ6jftv-}{;^zv+u0>`=i1;!}Ou8k=IUDf>Aj+mPz_`h*i`@?QGpO^S
zz59>M@OLA@@VvHQ_{MfKeBF*<_`A_y_|}*izO~B?Pv{PYuQsi@5!0H2ONhSH9<=V>
z5wzw;O>0stXx)NaNx#z_w7!L771&6uPG<EyH~$X-bn1&E6dLMW)^Ydzf27j126++r
z1G0pEzWZ{nrk|UaHSo2VTLZJsB<qgY`TFQ*y$rBn9e_z*kw@m0OZ5(U&-1Xo`9<E4
z#~Ycq2=d#8d`7_o<nxE^!SEfiU>L$?QUv*GL;i5CV#*HW8~L4!{4PR%XCl9sNPfA2
z8kzm0?9Nr0wKwGAWVYm&)Z6klp5OF;QhDf;-w_1bmv`^t<4Xfv{(UKO+UFv6+_WBe
zf?^h&hJ403a9E6*?fGz(-Hh6hAfmT^iimJ&94?nvm&WBB;iYl^f^gOUz?jsTC*#4T
zjCb*dE?iy<Sp-$vdbe-?X!w5YIf{vX9Na7XalY7oT(1GXpGA++tKBI4SsG3~z$g`C
zD?;%xFBOW<ULi1F7sG}(!*^pyg4`G}3O|pnu(-F4*LgH~KnFXfae1zVmT6UaJ=7(V
zS%Z!uJL@CA!6RMV?U)&f{l4A?U<!eNe~JUhDFBuLz`+5VC=~+pPMDTIhl)@sU>F-=
z-w@j|6hvv~zUAsTeQHsLk$OYgVdK64o*2UBMTp7@gqFR?Az0A+ErG!w5{Tfoa&(J@
zyr~3E-gM4xGGE-Q<;@;;u<ZhWO5T~``%P|hefEQUv-saM9yv($rc_5!0aLx-<c3!(
zp0m392D~hl{chC-wsxxS26Q2heX0wt*jLY2wJ`S>c1e`I&kLcdBQJ=)a~v;Hn8k#!
zi!}YPyi(AulZ#cJH95Ph>Ku_<JJInVWqB0KgJK9;he<^8nPh7&l@DCDkR%Z^I<kP5
z!G}ZC{c&p{q$8gLFqRyIn~dlpBNjK+!pZZ4hmCJ+sQXh4<qF#t@OW_*dOMm6lxZzg
zW*`Khe+fSi;1j6j2f<VQOZizw=W>1;=v=|iN;+5Z(@5uPe%8>rj-Lpf8~JIY6Nlzf
z1MPI;$X%*`2c0<hlp2WAiQ`VGer!HW4dC!XYM_fw96Lz$W70h}fCB@mpHnA?sJ~@z
zFlxM@8WklCd{S)};|1^`%(~(w^T4QO1ek8)k!1APJ^+nJOj<93m3nUX8Njn2=z+}G
z&*#rjFV8evuSJgoT?+R&y>s%VUa`7<^@3GFN(`4>m`SfvgBV2`V0vK6dk({{lbVeu
zl;sEVcrmjm>x;foXY!b{o2|6EJ83NwH;r#^g)g6G+Z~&P6xDt@(%tT$8{MKNoTK<C
ze2{_O=d$o>$N(AmI-)~Bx%6^{jI)B)3JJ{XI)X`f(r%Tvi>bhj!So;52=QhHBnvwL
z;E(l`q~uyBVnc3+<Xb0MaxJ=T<^*Hzl65Q_%!))HE`rwUxVsVOTsRB#xtT_m4|nvP
z7%C@E)NO*eIyPk-r#)ggzJdD>alw|l^nc;AogT;hz~q`Y*jV0<O++37J*|EoWUiI{
zk6<aKgWk)y=w{rI1UcLW5X)8X|3*6Zhd?~c2^IX5@{`X`4nG<Er19gyXAn2R$)31z
z53s>4jG)dfM4pH0{D)v?2PIVrCwu7^Yx_BY%av>blU8aZg^a_*Y@(2v@@m7d5)pHa
zz(Liev}4tkAaIMggT8I&0);!6vTO~|SrrH;euPN?pJEU=2uJ`t+O)?GB7%6j`BRfL
zC&HBCP`rpG(>D76A-j-y2y%iEs6q!?&x#q#2Iuu_dV)|l(Zi&P)H;z5qaqCZ5g}-n
zKvTqt{xt-}C7hQ+L6r~dua4f6!kOw+wBbxBAs8S5pO>L9<3T8ROu;!nq*DmzB_nJy
z`(Vs+Aa;BF(K?Y^Bawz1{_dIx{V>$6;g|ZGqAZ+;q_b-xi*AqSHHQf?(;2oU&9;yx
zvLPQPe>fHpwBEwBMeND%AgRlHTCB?vV{fLiaRKJ<!}q#-TCIECnw0F;h1KE8WZ+gN
zYwELGjM^aNSoq!)Vy1vGWJq?aQF{)+@$Px01qV9?2#najoP|Z8H6+89d+ZA18+dnt
z^XSY_@!2;?KbG0!Xg^eP*hQcn=3W`XXrp}kmGy8;zp@c+wYveIcFc(Sa<9Vp%@@21
z!!vv_DeEh^3WIjc>QF^uxE>tb=!T`D3s*;$@5CJFYaH5tNWX)a#O=UQ4t2lnA<n(R
z>i~`hBmQ?ERhN~2z`z@sBClp{$4GJ~g7N%s3$JBH0NOhcam?^F>zkU5^i?vLM&eDl
zWtmu<ht0h*ig==ojF*1$=JGCHZ2KZ9<`5fzXAn8@aRY9;x|dsET^L_(AtZ+x1I_O>
z@Vig$@I3;G4$;XQd3~IT-lS>k!fBPGxJafvcvXa1J`v9@F8D1v#;x2)s~cD!5v$;;
zb|P%!#lGMNU#-CHu}2wl`jruS9dAHpasG$)XOM@3$RV&|IEmZ|_QQ+U_zDye$8f`k
z;shf60Q1j4JqFZnVMZ7D(AGfyf*aJxkTpaH)F|*a8inc{n31l$5N7(74Tya+Fl-(b
zDjvm|1#cW!$4e<leH&6&cbQ9-5YF*=6eRnMsy$KW>?wrd5w40Ts|b!Y<33y#0GP4`
z*({%al{^r#jVSL43IaS2G^t73f%X#7&bx{T&8Rlp;m7woa6u`~(v%0UY)6Xkf^r`i
z?Ry1|&A2_h=mh>==6&@ag3gcN`w9y00D!OYT@X<4Dx!WBwNgii=Rh#~kLX;8p@X^L
zO{kj?;B{6a;g@0|FUo(Hu(yoD73xZXqE`WHA`B<k)l59*FwLf8X=Yu@W_IhC?@t+u
zPr*>k{}&8RgLUE1419}+=3v@?C>8*6hn5;+rVqse!1AFXwC$eOU{bR^O(9QU7`1-r
z`H<-Ho&%)b=TN3x3v8gIM5|r^(+tn?=J0tO-rJ={CNkXUonP1YcLnRbxGMAwTot-c
zy`UwNiGjK+zJM!yIlwM}tf59B6MdT<;VoS8^;YkA5c|kO*he0MSr4-wW&=#r<x|^#
z4z>@PX}o>X4)o{d{|-CQ1B*ARX*qibx>0j4C1X#sW&~F~(*fa)v{mWmGN{XZnJm&*
zksrH6VPLpctEQE$pf~FX-~NnEL_VF5%LSyz0B>FL%p7hy>mOT3^#HMDNNiOT9BR-v
z&2X+aD_GY^os}|qjyI!zEggD>BCQ6L<NTDGjZ_crj8m`dEqn>uln-U#nt$FQ!!smr
zXcBIl_nbv@{j(UUH4cg;fesEcDF1ukNqGSs0DB^82E61^jeFLINW*^Y@4zCSE&s=b
z*hr_36Zt>RA^#!%vqNY$!EBUgFq`eGAp7rG`j3mE&HndGe{e6D{fW{aSOwXMrT@6o
zE&a!R3h6%%Wqf+1KkLpX3x8C*>`#{c5dGPSME|(cD*98#oi53bx4=Ju<R7t(<3>vU
zk(B%+Dfvf=<R2-Le<UUUNJ{>Zl>8$p`A4#m_G}6!b?b!+zkOxiB#8dF2KcMNco_S=
z9zH*T)mW}5up6MloQu}Z5d^yfwto%|(cY;E+WK)kmYWbuFngU0MvR*DaR{BgTHZc3
zYOa7QGh%O2dW>Y@o&n>*Cdiil{RuQpTb>M6@?@yUlc5fIg5EdxEYT;=5`FS4(I?OP
zdzL6e|7Ro$^~tk7QKC?vJS(w88G5=U%Fs_CQHGv2i6XV<lSK;k$+JFLrU0$@ELS3_
zGW7IHm7%9csz8DShlMCdhC#_H7zWmkSvU;MXH4<12;|7HcF2)oQOJ>DUB;O6hH;R`
zNe0aOs+T&{EzU;m6vz>581PpgpKjE?i(7VLt3K3ng!*mo^*C`caS`N9TooyX3Q{YB
zi9<%Qx>1bE$x6HgaLD)$Mk|m<jA#8`&LvZaTFd!>#-!NIKg5LcIse)him(blb``wM
zajQg#88yqVk|~iGaxhL!i5NBi23UL@vRQN-OlAN>&GN(tyPVCi<<VFrkH(5T8tafp
z5Jwqcr!Sa6{U!tK^c7_MJximp|1;8v`c1|sN+aqw8HuIQ*wZbI#(oNEG!|ui+Vq=@
zPZmayMH!zgi=f|RB$7m9Pp>2zd-^2NWoSBM3NE80x(sX`v+y#JM3+$#T}DZC870wW
zlth=YIf^90JgQoA{|V<&joJc8qLT+NJ3emkBJ+2jjbGvuGJpH*)o0@uP7n5-z4~nY
z^0$v)q8~ecdHD#Yv9CC+^$r)*1ne`@!xYY1#fD(Gf$46JXzJsN$Zru!u1>?XAzUUs
ziVGQIXJ60B4z(m|LRVl=?CH7$`U74!rT%~gQGrbZEIvt2{eg2Mwcc)ct_@Kl+4lir
z2xaRJBbELzQuK$Bef5V_2ti*kmHI;}grKh=_3v4K82LY^Kct?L{*anje;9eX^@ow4
zLVp;EGCpnkL+UB%52>f5KcptoA4Z;D{bA(k(;tST>5M5DPW@px*g9t6aM2%zQ-2sv
z{b4xuhvC#8hI9VX9?7V6*njNRAE;cg0jrhE%gP}aHdTLwkxeT{Hroz3W=P^<r1%i0
zNBU2iMn6vP;EYFnfBI+o$7lM-0IO%8&7Yng`M~3K{~zfe_AG``^CR^4GWWo_Q_N)G
z#DiG$Y$~QQhP>H@S&ZC%^A)k5%<jQ%WY3Vlz!z9*@^$_F{n2J=w3U6(*7Zm811-Mv
zwBN)6*4JgX*GUUb3xJrpUv?W}H~zg`Kbqt9<s8|B^{YWXI3Gub@QYr<jYv^SM5<3>
zD)E`Q?J>^A#9s3=FphC)oClmZyMRUADz*k#YeK-ccl2Pfks1?jpi+F3Brx&YhTw;c
zTd}bbS5bP{;XIXUjKm48^5Q`ue5Vhp$2WIS1HN+xHNyWe&e!tP72igE84J0GjdUzf
zfi}SS#47z>b^7ItAL641!{=3boXZZ~$=v{2ecz}3UeTe&k1PK|O$4F#0t4@K#6#~j
z(pSh{0>NN!iv-Zk_ojfA;!c9Q2qtOVThPP><SxlqJ3YJU$unbdPp-kc2hlG|Kc*S?
zi%P{4{)2(bPGaC5&A?n?U{DxH)*-STjEo1}7@681LGTL%gXa~SY%mG}OsA58O#tx3
zwuRm>vL4?=Fb}ge*<rvSgdv-{!6rL9d>mVL6bwR!!G>Ng&_^+$0X{L!d_@^{Cm^=q
zAkb0qGJHrg7u26Y7(63Vo`Mx$900V>|0$0qo+x5Xz}@lKp@H3^EPC7vi}&VzYLEs8
zcD}BV{T#rG5}pPmTsDuF2(Z3ijK%+Atoav5<)z@>br{T_p5ua2agGQ2M%6iK_!gg&
z0R`lo94a8^<l|d#PAR@~&Z)q6#W}40Jx#%+z3NUqj3c=2Zqy#<N(}T4mYmW%*3MKA
zHU<S*2?#f7akCT7&~Pl%fF&4iYgz!4TaGYQn(Hl`SUvt>m+a*+tz2)u@!%HRzJPsD
z)(I1P$*A|l&btPJ2zkT;%v*8bnE>{RvtPh>ecIk>t|t#e#H{@yoU#7`2?8_2Ro?vA
zk?Ek(YJT$U1|VP$IfTIO*iT8gwL2Gk=U-I*W_Yj}&cWXGRihxfkemI+k_$mn=pV9h
zuH?sg6Fa|%rR$hEgE{7gr<{x@Whb0<u&3R4vWv0ab|+(Pjg^rJkW(gQQ7L(Yp&hf7
ziyevKAc(r*n=;Ki3x3!>X2Od@IYEUxN8ETpi6%B8&?M(%dn52F7$xw;AMOR(T?jnc
z0VfFX2x*T}Muc5lbn~am>Ns-_3bOzO-{Zus9xue@0E|T1QEViRSWL7{-0QqMXo5JJ
zPU1|BC;kYRuia7L32}k|k3#JcV?-d1?q1?bG;tn4QT}n_zMzO31u%%|W>~~UEE3lx
z?sZ<5<{V8YacLS){1MKl-N`oR1OXly+QXwcN4HJf=VBYm)J{PrJqj51zRSsKM&1Vy
z`U`BSyFY|LFhA9POO_2|XYC=^WA}cGj$R;h8=DCrvnU(*y08O~ymV=D)vm>A@G@X+
z<VIH_1nP3J0Uq2UmcKRED;pOBt!}U6kZc7OJ!J2qKRl`5rbQ-aJUAJxEgl+3vMZ5$
z5=UqcL;{TPB)uKccyQOb455<ohtH60h{l7z160p%E|rvVf;gD|wrPE6Hhl<jiG7Lf
zh>F>YN|gFNIlOGoN2K_>#>8frA6<)xV!uSu;EfF)9TaAur(tfmn^$6ixr6Sy++RzL
zVU)pi9KGDm)T)Nb{*<?3p*z?%G!6qy2nYX&ss>S&)HE^demJR#<1^XFIpOH|hpiT?
z?N>)nOt1?6=u%8STL@{<weJiH&RGzO_D)I3|21HN--_gd<rM$9))|Uj_X*4VR#Psv
zY`er~id(VxK4G)p+K_AAk_-^GP2MM4C#@!HN(#?5l*%{_B!>rwE8V~HAMLi5N{50}
z8c-ru7huQW2+^@rUh@f0Oa|}+Na6}W>fsTdh**zV&3@~p@R>l)0TtnE-LO%yfCLQB
zcmFCk{DWi_e;R=B^U$#4kE-}7PPyw7@#k6311X~5?E;rnHu<d=q@Wu?wpbc$OTBFg
z!NNKafytksmD;}fwk5~5WWYlF3Yc+xkL~NSEnU#=Bq|vP13%4)AF+KKZA*h~sfUH}
zx5JFbUt#-}+LnCVk^>9zLonm`X|}J&wzzCd7mlA0Kf{UN4xmsQu`P|Zr9tCIVXF9F
z#4!Nzg{9uMJYZWwwq=2Bsjw}xY)h$aDHID@VvcReuq|n}#RCfo&WFh#B~=u}Q+(x!
z7{hi6FgLd3GUZvXJ)_#Qot}4U&m8Tk1fULio%Rd?)9zEL8A!Aj!!yr%*Ka)X^59}`
zMJ}Z{sw3Px1~2!9KoOcCtFiIOZU7np$VJCuJaQs;Ft&lyf2nG0#v>mP)Ckb*W}Y7z
zj9eigCQVCWAjq-Yq;Aw?0@(*Jn@~A%wv8_};ti+^qj;?va#X|B3yy@x1TNJO_+4@Q
zE`c*G#?u~$w+md=*175xN5s-C&{5z5U#<{jKoxG6j-xFeN1I?LD{KV@wt}*@BMwCf
zpj}B&Ay;@=G`OmLm3Jj9fb7Q~_Se|ZWeWO&Q4B8Gz&2Qn;unC4(~CB6jR2oz=yq6f
z6mpf9HP0Bn&dZ9&qmHb1@L26-z2n(PRy(?-)?rKVz9eYQco4}mWb^CrnYsfTUN@4>
zeBuxOuO<N|Zk(5Z`?a|Hum>BED|RMvEA1{7kXx^>9{r&QdiZX^vt2uF*LI2IWh8nZ
zhKr~K2#Adsuk*m@UWR+w#(bJCl^XmGL&F}vLfRZG5ThzzxCcU6ARd(kM`V;`wU@02
z$k^}*3(X8MmwL0r#M203;*gn`=y7PO7h1be`;2tdK5e#ZGXhf`!(9esIb^mR7G#0%
zoMA7xypjZd7V@<voKPxymN(|Z3rCTPC7Z9{+IT$kq$64-SwTeWyh~L?>TOAkPI%t8
z=p%dzu>)J4K71A?4mTCy>7?S9HLoRyZH$eN)sY>r)Ou@GWW;_AalOm9-e+8V`tkr>
zuj5nvCd!aUbrB0{5n@3K>LCi|1Z3mks6uSQS3tZgDtXz1^fCP8$zbMXhYH6BE{kx}
z`fg`90V3Sn1ojIj%++sY7`~U|!Q#(ZfXdeZEtY2wc1l_?2aOoD;kJF8G(tT>u8`M4
zc~MT@y5PGR#Ss#@Ljtlykq`Brr+U;1C&mi-E=AEkCN0m4VORGf2O!ou2Cwo)5G|}R
zSfjAwxF2jDM3W1%9T4EYLwlF9cg)_T1zH-Bl46izrLd8Pmp}uwCAjp%Ax-KXEJXqz
zBLxL`1gm%pvH4OS<R!~ej(mr-t=79l+m=cTM@g^pE+mL$U!g4Pyj8GdMmjHa%r7IJ
zMXTLp9&PK~WW8yjetMR>*<fH<;XV%*uqXmvHp2utc<5BU1In8G3}AeK3h#^AFQM?E
zpC-PYpdO-V0`V7a;KLEM?g;bBcO#V*Vfh*;USZ)oX?C=H;}30yuc>`vtT=^$$3U03
zS=Y%#iHr~ZNn|kjBpg~sA@=LrPl}#Q8)CtOpiA7W>t>zF$1F{w=2hIMpjbVN%#dUA
zbF@v#OmfOQU2{sLCm^f5#el5wp#1yoUxepsH$@UG>)az?(Tr`;jO9apd|;J~eGVoW
z+wXnQlZh9`Huf@>R)m#jR!{Yv&;+rr>5Wyar}$oIg4p^MzIrvZ8?T1G0Y>xWh(d6G
zH3K`kSc%wBztDtZsp#X4xl)fJNjy)&iu$n#Y$Z;FwI~d}*wkZ=ZZ|*I1h2VG#wG5r
zHX<dCsYSoRtx~YEC+5sBtkg1|WJu#5U&mL?dwh}88YAJ1J#jPQgVI4|D4Zd_v12z8
zhJ|3lzK?fZT4QhZ0pSdEYixTT5C##gu?RsHF1$Z^R^%N_qV$^;xvYT6B$yS+yM<A;
z#wL@#kIkHVA-#9K35WbB$~aU_6H}dMGBGI1fg49e1J*ME>je;MT#y4zY%s6lYU*I|
zu&^#{N`lB*G#P41lTidnMH|{Z+WRFb!fN9mZ$&hUkD27-SIEctl9|4IeDp?Uun!-9
z(+4Dc4EF&E9~Tg0ZF=%1MvughnS-(6OaZs4Sg*wjh0p!`4+{`CIUI;(4dgH7#hV`9
zijC6#H#2??*BSqxRd|d~OOa&ky&Dt_cMu0de!=-k94DVR?FvU1BCax6?PHI@Q{M&t
z&?3x%o#FrSq=}v9@_@`lWGWEGkXFxu0jB4`4#uWR-dbZNjK8`$IZ(X`e;bo=%MZ@0
z4mN6ULF?)q5&H=e?EGE7^+jFYc&;AxSc<$FPx(SE&j5`$91FWDdK{7+Reb`n?ZC<Z
z#FZ)sEhh6we1F5K)~A5g!urH4{+0Eq5Ok{gG>gps8dw<ns?>_U^=Vo@$nIO8rt|?x
zeHzyXB=yNh5N0zS^=WskSqg7AFZ4(jYd`<r+WrS(v4O%@+x~CG+O)r;KJAW`DSx{@
zt@yU0!BL-f$F7OvNPP+k&L+D)VM;>k6VG+t7`j)cR41Y~an|&GqKcfcXH<8`Yt!eX
z`LxDT5qqGz<>cB_$l5eA=0_Zz|Ix31;{=V!kH5M;K@Eq%<Op+(bKv?k<O4^wT@dH=
z<msdM@bh7KYQ?+(wSbCwy*5MItboZ8awAOs(C;c)56g1z5+uo99kpHCmV2*&4V@=+
z(xNC#Jh7>sGlcye+F?>x#+?C$?^C0A3rg7vFVkU6E4*2ZiI(xOK*et3YhzGm80bBq
z6mLNcxNM;4Fpd$zz-<GCCq!v~j_Nk<<zZA*{Bu-HtG&4@rU*h8wW<z%3u9_Ra6XsH
z=VjF+#t}w=45}zF8KI)6^)816>B=KZFriiP;jm(=@O=IjM=d3%F~o!gsjEl7D$4N#
zhzCB<M)@);2MzW^uVUwWdN?kNFr5tZf~cob0_t(tz7OG2Cv!Z6OI3E6O=qJ`=W0bK
zX~MNj>U~<!xt&?x%FhnI+qM&f%d1Ckm-JZj1JFp^*AWW`CWUX1yypddCD8TTwSCDu
z24ohhyc_?b;Ih&n9#jvL^ue~=8`QQH-h4&#DsO?ZVDPLg>%7IV;2HuP!81F*;@r=H
zI2J8&OSWOFb(2|2-(KQ=K!vV#*TRAvox}scSbiKq^1-nvh;=-3ytP-~=40~e;kV^(
zR!lGq<1w-Ja?1JIu`bO49y>lcR<Ls`%FfMd$`=j$J=$v&;o6V>Z}vK3(PII`5~04$
znmUlaT}ozv;keF9+m?D~(Z({Eqbz83%CgE^1xrGGo2nffl$*^R42Xchzg^!h6~@Xy
zhE~LTI`r*Xy|Ie*Z_~F`;R}tl_SLu9;WrWrT`vmpO8r{`_-aLfaJ{ph<X~}>jm$`2
z?rkR?HnB#;#OYQsG06>6_>cAG0&ijbJ|}dz?t1t^p>T6%gzu|0$zy_24_efiP*N`j
zM3luMmJbG*P#UMgdMgfk6yg@-ty;^(^77ReZf3xx?iiCV=7RH90OYl3v>MdUsR(~7
z4yhO7#;961sq1YWx&}%F>{g6oE={*H0;bFeuu<SvTozckwU=X;Vd#KmRAkN!FT{ku
zTwn1IxXX>J4DF~`3i3Tb?1d12_zG`Ni{HA!YcJ1Y1%fJwx{CtWHpr?@?GN5UPkK(w
zbzM(S$0M-E-ntgHl=pgi>iyv=ubXkeu$Ogsk3%}ps4vkUmpss)D-*I%LI18dd;L?C
ze~MbUz=aR+|BLpwmv!e?Di!J{RG}_MKN;JDj0g-RinZg|jR7ubwFbgxa?SR-@Tx2-
zt5~zGTG(%7vJBJ@I5NRS7O{sA4M!$_M)dK^Rj!f{x!!!7AmYmO#BuJ2d4Ac~e=Xan
zo%}j@b#83qByi|?*kfyH!+9RCBNyvzJ)8X1AK{7?5Bx|Rl)8hlf0XzadNWiR{Xjdj
zlyjCi^(9DBthZ<cTv^4tWv|uFQOcR6ooULMu7jI%>^krT4Uwx8C59myB6r*h<S@u)
z&5@h^1oFCu#NzQhry(BdJc9FthGfMd4+&%;j^PW!a*e<PN(%C04S66AS)d_}amY*!
zk@aJVr$9qw(N-W6G=vMSD&`CgktK1#IbTB-#BowIq#+LZAWvnyEe?rkNLL*4qJ|Wf
z+i8u^5KkO3h!7}-jex}Rw3f)bsbX1h#4g{2px`f?nj&{=u1Z)|1clu7fW+_E`V$=3
zq`i3Knk6pRZT@2a?f&ohOKxxO#GSDi8>uF5hKs#E)QB1#S8uzo)R3;YB$!5sedU>3
z=#SUH_XT^Qmo?C+sYXNU;D;$8#JyC}F9b^G_dD<J46PsiZ#c-7h0j{tr{;R>4%lK}
zf%p3H|89G5TZHl$3J=87`BNbq<aS_G*t32bQSZlREfado`aR;<JyB5O&9FqMayEp?
z0$Iqr+t?7krJcGVT&kUE$~R31?`;TYYe;WHIF2M)EWHik9S!Mi2s<>Sw;{A>NN+<}
zts%V);ok)kZwNov2)zwqv4-?EgmMk(Z3wq(NN+>9UPF2tLbitVHiWNhNN+<pM?-oW
z!e9;QZ3yp9P^{=}2)i|;w;?>MA-xS@n1=K=gkyfhV>g8HX{hdSNJF@kjy`SRS>giH
z+Xf!j5N;k)#rU6qaK>RPDDF%~E$Wzv0e`$UeF2XCwds~eP@8^=aB-}`17IaT1g!5V
ztPZ}g%(&mi2B-ef{qw^>oB+iB9>srvM>;$@aOIUts(8mWsNIZcxHeB<blR~U-u(mb
z`6B}F%Q-pLi|Da^0FR1Q@K`&5$6Ds(-{2wnRrgtSWFjXdWG&1NUc}p<fZw6@#lDA7
z9WTIF2W`XTkmCF6@G-5<jyCQle-D|A9b`@)hU8$hc`8iDe$BYAFrT&f^llc%{+FBm
zQ~kI2Z}u15WLEDRCG&Leb5}>*z32$H#Ln8lWe3^SR2>D6usNu@U7HbYHfpm$oAufZ
zDRW{40!=LCC*N$XrV_v(W6FY(NYkgyQQAy~2~-mXf4Huz@li~=hUX`nW9IW@Ow$S6
za$9LWMqj^dP^<b9NS+nDD~z2Y;fbaht`0F|IQ#&NU|m^$u<mh*HXW0X#|GifP&0kY
z-JgfxG^-DWVpr;Loo7`JyROb1GOcO`L^R7(Gz~QRg{^y>p0Q&;R%trI3U=M&<P4%v
zMwoQ=jRG5dASz>P><G@S1VMmk3K*bsh(OFRm>0gjuy6RYdC}Yt?a)IjlU2Zpg~7Vt
zwyCt05pF*zZ3+7=J8W;Zi5aZ>b5sRaMmR=SpWGw((O8dlQqGBqh2Sr{H%ig|OE=tk
znVTnL&(V@$D9<4BmmCaF245;?nc;^*@ShyM8HIM6#|+<_9}GW4rlR1;HOwCOn=it)
zp2sFw?k6W$xlg#@SPRGc31Rn>#v|1(#Ba%WLKwb)xCO$mKz=px==jGcGUTtYA80*-
zJ_aO`XvsHReH``)#v|_amRR%9i(MQDyS)>_1&<wvN5Puo6RZ(yFzX1&dIzNQ%j*6^
zL2h-gL6V%CoM2mfS|{V)wEMpap)=WP3PL?dM$M2K84i1Z9cS+rR?Lx9eH$Jid#^Xi
z&(&y%x`I7`%E=JHx`)yjFkG#Suy3Gf4#tWc(cINH8iqh&vmGKDWrTB1isnl!?9c0H
zf?@1TBT1tf*XPA4jfQkF?Btrw#*$ya5(s-Tj|8nYvpUL6y4#H<{|v4KF`}z}tJ|zw
zgsl<>2jgsAPb+RVwT_w9uP4pNIi7>qEL=XMxk)6?F4)^TM;<oHzlgsZJx?Sf3c#B2
z_mo-PGpG^`q@^?0tPB1j!%RN_q#H08E9YcK`7}(=l&9l5#~cjUvT~m02e+|h<vfeI
zY1~8A(l!4%gho_5@P|o|tQ_nGE^jtIk9CNw9HfEqC2hD3a?x{v?04t8t?teRX5HMR
zKz0XYZ@{?rWnQM0{w|0gY^EQsIup{VW^<Jf%YjK%=i)nPJ_vlj(``I5{!jS(gW0nQ
zw?Wfo;;y}JzCu3lYgV^C|Bbo_wA?=dKle|`{G@<RcQAX)yf?tuDt{oHF*$tZ<nRb@
zAbdMme75Yq&D^i=kNg$(x%sW}E5jom#**`hKUgCkvfTfP<Q_)C!R%+}y=GQ#9OY0~
zvG76d@rqHX0vP$EjCXbrGwb{G4?fc>@OlE4vxn%!=KjBj9wGxhL;yWR;N%`cR`PTQ
z5v*ZVzv-k7qOxBHVOlTt=^#evaBPESoxA?yI*52!-9hZY0D);b!Zx55n;+9b=qRLv
z7!~`a9f9s3MrjJxAQsg@V9}*d2cg4@lrFMEqk~WZKfZ(L4f}TN%XZk_Y)e;C_fV8M
z2v;k^*+JOBS({$N*r8v}V+e;f_T{}#^ywgQk+repVhRnul>+G);8G>V_GyleA{f32
zoy7I%38m*XDT#vNQRp-5&Z0hGy*Sy5(kBhQ!WxKp)m5BA7Xe{E-$1I?hNgEeNZPRV
z1MBAAZbSN+Ys22hCt2Rd5ZQWcl{_w8NrE2aG1Y@iwl=a8@v&c+7%q6i(Pad}N&UK&
z&QbM!`;R2M|G=9vxJ)*X{gQG27e#N~dTL$B?f3U~A>Uv3sk@Moj%dEoH=4hu3wds_
zqK19OC{}{kAUsqTGV0_m<hKKMA^(B;bBbL^NQtIX>_Tp^yO2KXouBn5r_i7HPVP_M
zNJYzs1Xy=v-~Qw#bMd~$;q=28gYyJ+1>WQ*&u+v>=Y}gdE?ww_bgP0^ya6Glg`0H?
zVvmL1oaQpKpPQdX&r$5L{)9Ry=&@vCF5bv!Z)?T`Y`2UH*bSd62gbULrethKSOtH~
zrPAm;Gn`y^LyzUD*?ix_TqSG(FxGH62C)BXTFspc@i{5&s+oP_&Po(FF6C}kgq0&l
zof(|H{GG6)HTG+my-A<Lq*pTOuOew?s*XH>Sp!q%CpkRh&<wF04&dBPcyPcvBUmO^
zZc?r8!IZb(NS)*fv`$Jx({NR#1+5Qow;M$LLa*=15J0h)GUGQw2ED$3HN_KXdOJCg
z^xR}?Q5yb6$>oq6(BAy5lc-=pXb+CxM(}IEGb3xe0LEh~op{;Gu^$Z5*`r=MsU7Ko
zWa67D=|OKChNWu2_>NA&NW6VA!i!f|-rxzf=_A`=-B0VPk!iq!>j+(|N0usA4_ymK
z;x0D4%K~fh$SADxK2Pn)jCj0DM^=bk-fkJ$sNZcFiI-XUX3MIP7$t)OzRpuMGNhkt
z85u!L^6rlt*WdY|)bO6z)OZGJQ%RBGOYus6YEzjUo&<^lZAB_$gacRHrmsrPSB@??
zR;QA3_;kauE|p}#u>+1`UdN!Gfx(pwcxfiRAhiKr^Z~_DBYk>mgip()&q;0Pt1{_T
zsZl;ClfEz&F;9&k4tJ^<zMO9)pg4LW<McW6g&5HD&7<m#8Mkr%vAO>wf=oSp4fLz=
zNgsW#&~M@TcC^A^`0IhPRbGyMa1p4wXdXLjb1_Sd6DIkr5q|X7s9hJ~%+&)s2SEnc
zjsGJd9H_LSexcF_>n?go?t@1Sj9q@IJoV(qf&m9uxZ`>fLIu)a4K(e?@EddWfwBWU
zD4L8x9!{ZY{gwD6!{wJ<REzxm#(!MD@L|}CvqrhryTLk;7+aRbZUr-W{;>NI^!R_q
ziye8mo}KY5+_J_EgRcUi>K{FZvB`tA;9zcDA+Bd{4nldae!bhoeXFB_bvJYen%)^4
z=-E(bCIRBj+>9p*<QL!X6fA4v@tG?Tg|*dqq%BzXZup{~2d&r4$GTD80K3f5TSNd=
zRbq?NCX68>7!qx(GJ;r<!Vt%=JVzX!00ytlP)3gbF}Axv^#_CIWd|`pOxn!$TWZ#&
z{DgDKqk`$ri9En`i67&e`}`Q++~>#m=Ds8~mTB#pzZb7}1hSttmi&@YSufWWC4pOB
z>uCD!>bD1{zgG1*XgEWH*3+GrRQDvotojlLP*s@l7@uj)S(kweZ5`G?(v}I<oYjO@
zpK#BcWVNJkfp-!jHtUML<ITG9k0XwL{ISa4v&y@z^5fQ}pbIYsw0f#rj?a(T;~Tt^
z;IB)8d_gg0A2MoIq5djK8_x0Iojl%049y_dn}Pc0PR8;~9`0RZ4dVJvvQfJW(^>IZ
z@=Gq}gnUmGhs$UJg}wby$8dP{1vNvkQZ&oV^m`XE-T3V9n{-0%o`qT)TX_j)%Fbgt
z=n8f^r^|92^JVz~YfHd-DPX;az6`5efpKRTHGc-7{_wYx{o|%28#S!LfpKoHQS&U}
z<ATUQ%`WD}8f=!WWUlyz8JZL}i>Kn6`qqhQfVuECjD_O}cv(ZvT95ZCG4278z?Wwc
zF7gJg@1gh1-7imY{9pFoJ}|24+WXIt01?qMX%T75)oGjB8KD{~ZHZ8ukpO3O0`Y<x
zT2zpiLPe9pM5&?#l2p$f$I?oxz1OzfKDWL0)>dq_1#2ZCKmgx@Rz>k8zMv;WCHTUd
zko-R1ea<8^A)?;iKELOWi)QvYd#|(h+uCcdz4qE`>s|t9e2BkcQFHG47SQ;<=&2W0
zXpG6Yv)5gEx^mw^%jEYoR#Trl275=KlI&~8V22WV?J@lDSIpq+`}o2n{8|aUb`*>b
zI7I@jT&9h;IO4(yem~UHdxqY#_|<%)pJeCKcE+(q`qi@8zt$LM8<T~?5YO6;G=xi5
zf~*EF<mG8QvVYXhn9~(&?CPg>j=<2w)n@W#PU4@;#f<JbR{i)~X38Tb|E8YLX8`QX
zD6BxU_7`EE*?ogcKksgu_f>;b-fxvyA>pJvuJ!Sru4#qCiYb*<zZSkHQY-f;@7mM5
zy1=$e+4Et8{q>bRS@r)VFuspVB;r(Rys#8BWsd9Gr;zTIJfxBr>C*}IuUd7t!f*6&
zQ<9I@3y(Y2)IjSdFoadA@cF4BuR=*JTD=#+J8igu_HYJMASlDC`#v9{OMxZ4mbh3-
z6+MJ?;{VfzGE2A-rzFH57vi9eEhzz>6N-sKMp%RWA#ipV`ks#p6{|g*XRX){dS!fS
z399af#L48`R|$mKX@<qviO&GCgunUxRq$uq%{4cvl>9RV0OpHUK}`ei1Z%utwHK(}
zfNvqtzf5>3=z!NqCD7WHV26O3Yox8#KNY|f<$zUR#vIUol4z`_r%XpM0XYTs>{9jY
zI272g=ZWj(+qp{^#EHU?yO>T^da1Sy={yJ{8E-1FQS{i(c_r+rH_-;?7U?0xLr4$w
z$GJf(nAfCh-!S5YtH#&aDidSrvv(18?<a;K)7p*mGX}kQA!Hfih9B?AC(viUs4@AX
z8-P&P&-y=X1Kz^V{nvnY-F-Z{I{;Y!<Nrt}Ea^^IPlEfM2tRZAX}iK>p~7_^?-}FS
z>^20W&SO#MsWb!4qn>H5$PwM6nIh>KOc51<#}yBqmLc3ymGgVJq8-?h=88zT)zf{3
z7)y>Tw2fIp(3nG%I)}Piobh}sMp^2-UFsZka|fsu#VCVnGqbfTU0L!ve{b`0(@cBZ
zXKC3!dD}e{EL(q8Vf1G@|2Yh)Yp`<zI^fg0y@nppU`7zSxB(ra!>l3u1d&onw-9D0
zf3{Y=JK|@eV7sxsAN_lRJ)1k3n9AHr_T=h!-Y~*W{!Zy}$e{#)5zda<)<AXPsC}hQ
zTl-fuWB152f+5THd7gY&>bO7Me|=8W8R|bdjmA+bAhqxJTT~KxCze|JCnlza_vLAP
z>O}4Jhn<TL`wxx!+Bw<0F)}tomd*Jh&Dy3;r?I+4RFA?YBt3_ntY9l(X`x*f!CTr9
zZpG@CJ#rFJ)8*pM534?inlseu^>2!=g$#4EQzP$`7+I9~9LxaoS_z-?X!`a~sC*v6
z#$D_e0b?AKhy!ZKbv@w^0FN)fT-en$c1Gdg>a*?S@Irg`P6q1E^gtb^fqH%_q1pB|
zGjxM?ZdK5D%Xi06rY@w5;f3AC0Y3&AvWqX-giFW%&ae=8+RdjU@w8j+c#7D@2xF*i
zY9Dcwq{887I;NfdAdoOfy)vEy+U`6?@uG1)9ZcQ!Q(5yR^kx>zT4yIu?Y<XVQmW~P
zWJx~lIGFqQdWfzUz5UjsH=|&)k=&W6p||%udh5cJqW(?M+rN4A#_bLOjBwm;dfVGR
zx);3xBYK;Yrniot?jIXP)mF-x<o#na$n8kKMc;_L2f5t`xy|?^<i>aqVO^J|Hy)t3
zu^zdN?V+}&{Mt>THiUkt%}$18T@+6)f!gMX+TMseujnLsFKQcau9^_FJ!+`!b@web
zasAFrY8!sL<bk3D+z4uWiyML61D~fyhj{hz>FpKh4JiGD=_5mL(XOYnI`g}qgWwiI
zaK-1R#_dZ}8g#{!kfy-VP@qR|P+*$gwos}xz4>y}7r@1;$EUYprT(Sd8T9r9g=~XW
zN!;p#-kM~4YUs_I6g+x+^s}NjKaWOU#LL^j)J<<wrO0W{Ekc2VR@Uqw8TopZfRleQ
zJ5wg2){D@D>fv7GMYPh?Uc@XfVoP^Kxffwrox$-9FXHxYh%>y1n(l~yG7;C!_b^)C
zn?J7wytg7R?HyM5X=Xm)TWzoM`mLsUE6d$yX-an&u-Ws~@;8)A)Oj@O{LScM_<Ag}
z2zMq;7}d~KK1_Y<M?}|3G$?gmW{FBl@uup}Yrs@Q*CAuV#yWBTQtG@k!Rc_95}^S;
zza6;`bDyYjgp>s`-3KZ03cgzf2Q)vr4xuj_+ZjISFw|i+dZyV<NKapFYgEVdT;{L0
zv-pKw%BpvC)Opx;o{Y(Zr3{XwAF$bKGPGPa`&~PU|L*bg&rObAV|#nssKeKYk3<6<
zrA`AL7LYRwwH4#Qr(f*4AXw_`iaL{+?5?xf=PhU3_K)J<#xn#f+=1K!P+Gj#x@EhF
zWA@glv)gHp?nh#hA}bU#3uY*{D;_Wh1RvCM;3?<r{f|eBACFpPkEf=6Il#&?wzhi>
z?mQL$?B-E##y;C(OeW2em~Fb$=z;d^eX-QSKfY_Qum{H-i~HjNV`UoH?CjtFbhP*o
zI1i<kehr+6m#?#VDc2p9C!P5#v)8pXWBw#sE`zk&Sf<^^32bguHGGDjt^CYm-*G2@
z^RPJG$zLbCkDU)O9kwzp9;oK3noaF2clT^18Y|PvE!HRW$adqHemvapm|eWPdQj9k
z>^$FgFa3$mvZv0ag96Ux+ShWCkdN1GdHdL%e7^k~wVeUy`NS7$JNqSqT%?^0Sijg{
z{ps<*({|U^4}EtI*B;1Q6eq`SeM~oo(!k5n+NH_EAZ)Py)DURV+krIt4SIW5X{FYm
zTBCu#vHU+=+_vmN^e%&?bwL@QQ@gZqbU$mAK8FB?Gx3P_Vb2343qmr^ZomsmVcWx<
zc7_gj_c@K{rt(HXRZ*vf??e5Tp07!u)7cx@Zcb}OHbjdbwc;)4SVomPg@_f9ZB*p@
zmWJ8hQc#X|KM8Fi*R@BC!lGOLF~H7y*|3xPGtsm~yEa3W&hu9N#n79?@-evJkQyTB
zOd!m)oe7NH#f*;e!*mFVUEs!pd9er67kkF7yZ0i=IvZFs`dg$ESmRF!(Gm=&&56&N
zY19*A&l;W~#+LM;EeqOC*)W!H%n&bgd*e3SxiDlGZ%&+U_m_szj&#`f|G>LR^c8#m
z%bcJc$m)UXJ)}OkIe(?etBeA2bKyoiL;la;Z@&G{{0Yt@rF2XXHo!?Ht_3yIm#h$B
zZL{%<G|+jgD_Gwkj{z94PuK+T5L!YbJ$aM3|GBl}euJ*>J^)IS4-~Fe?#Y`BPfxP?
zw3DB(lRwfc;>}<TDn|boBFbQrg1ij*jrNz-WT@l3`IL_}Wf-&~i=fEKdB`Xuty@Ns
zO|Sl>$R(;n+)mbL_Bl5>{LV7x4b6;2@g!O$<Ao<v2T+#$y@Z^y5x>_=BlYA2NjWi3
z$|-k0^=;JwK>{Zgq}(tvHF$&sNmwH*wOG=8CORtDcV{3sOZqu&)V}twai38)pn)iJ
zcC}Y!%HN)#oEDgksO>ajk&In~U`yG|j7AW*X8hTcxefebXKJ>*{YejvCjm7Xc{}Qf
zK&bdP#gdb4shKd#klqrNf^6GsUp9PARo4bSi8Vagp1Fpds=zv<J(ruT5<@uyiNywn
z5aLsN8cf?KPK4PF<Zp^+l7Srq;u><!-B@9`J?je~E<#^fd8s($JaNb~U~sU+POLEu
zJH!@aa|s%Pj_N0D2X5UbBZTT7vLj8^LFY*-)P7aB{9)I=!Z^fcW^-b&>bSF-cH7jB
z{l)tM@9`uREioW>$c=Va18Lra_sg+V={8T5@l7gm<6!5>+^gQrz3M>jRfltR3A8Ty
zJY3ygyDQ+$y$=rKiZdy7{*NnSgvOvYs}|b;C?sAOO648T9bc)DEbw=kRYH-SJ3m;8
zAEMc0FqmjAQRTz1oofe2J&ygI^yS!~bn8c%1O{p&dZ4C}$?eB;9y3t=J-u-M+_D^|
ztJ(%ig;upQcj@E0bGCL<5v44uJ(Roj4`NPZyk!SWRU=P>r>iO2)o||Ek-xESejZyx
zo6$Cm4;gF7#@?tMJA}lQO3c;~tgrbc%lsx8`8QF?bx-l+?l>8^ES15pP0Wk#qo2y$
z?bG@zOTMU>;5-@^f6)NX&tx!Td<Mg42_3~{HuK4F@msI39)~%Fp-)7bvjoLGyyZ0=
z0lMStq?1m>j>b#O_jpTD4}JPgB%IOp)DYfvCt_6(eR*7@G|4KHjw`}dgR6%znY^TD
zfnJvUwZ}_-?eUT~;3cn$m%M9Oz5B@JMqzhDczi1Ubu^USyrV>LpDT4*#XCL??--Hh
ze=O|Vztdfw58xMv`z@LnjgU#2Po(wjAA?_P)RO5j!z|bXf**vToxV6l;#&Lf;S$AZ
zE+P8(pW_dY*pW@}hbP*<o5^48+Ffb>FwksmB|>)XkwD@T{C0DNESB(uVF@F<S;7;V
zV7ldn+)Llhz4SotrH6A*$>0yKh(BoXg)n;m=@7#thFkw<hxkKy@putTel(Rg5MLhR
z4?PONG2UvAA(4?h8a~0IFL$x8U)V8Jx8gK=BR~VXbnG~W>ny3)G^u)HnYq=BWiyn&
zptP(FktTn{`m2a{F@M%#ZaBDmYL{j2wfqg`WjnZPWn9FZ1F>4S23yXo_BcP4#~qTb
z1K?YE12~jg86OZ-P&yz))Qka4xK%ByVO^fbGTxr>-fF-J#GC;$iPO!T%cd1p5wUPU
z6=yZ8wtF!oRwf3N0P>U<QO@I&-eZa23~t&0xTOP>__UW8B4X75FO#Zmiepon<OYu5
zF3T(AsYy?zg&jO@^d4J<u+e*NA$qg-*udi^?{SSF42sf1a8@;Y(JOh};62vForZPF
z)|a30zdxwP3jc@kUBy=OMoyU`=~p%0!^>7HJiL(Oxw<rz4emyEc}oDEHfSY@&jBgV
zq}v3EH=HLp&nhVVjYcZ)>px95ScTsON}gYL-@^%H&J5$6N5}9~0#bMD1KknZW`$=I
zc9`@I#l2;P4}qs@HQr(lndrT2kax0eUD#?qjm%%eRi_?8Qp`QIEu>hHB@I?Mf$J{L
zZ3<~C?o)(u3+QexY%n0@Eu=nWO@BxU+euhexS9kWY`|aROf)EW$-Qld7hT>#VV4eC
z1HwU|XPd<ebFnP9$%f|?a*+68wfoatu&QfhsmlS=Xa}i2Re#27O78Hb26TIT5ZQ<M
zip3U}`BD^$5PpdxoGK27ge$nOcTQnBksIlrLNaWmiJI~EDgewWD)7-7e$I=2#;+g#
ztSX)&`du&jT_2sU+2tbEFv=5f91%Vm&U6$F%zdk^)KHz(%6U&+aJz}~xFp~ti3CcF
zBO4`c^pf_H#2GJ=Xe_TUfZ42uE+Q8k&*Ci?zqyB0=Y@rIU-%K-^tzfa)Sf~LrpHtf
zs4FpuK7U2-Pt?#>#1E&xCLFJE*KZh4P}3hK(?`hWC9C3f+Hd3_#S7w1i0XZ61&q@1
z9#sLcAoP+su9!*-S8`gNBC!VdvfNH<sPU<k;x`MqibDHVD9)PxypM}B$rUfCwSu&k
zP}Kl`;B9WHeOg!dwyL<(X$>v}ChpyIgF(r}xr%#J7YzIKW7k-bnV6R>SxvK*ui>Yp
zfl-2;MR_Y!jy*-#ziG^KmuoLSXbz;-UxLUJ(><;YCH>%WwFmRKYmoC4+y*4nm|M&w
z-l%i@`Piu|tWgSnDps1%qiwx6@IJw_80elau!bDfkTHLppBl{4?4<#AF88bJd+qSD
zu#2aL7tnCy20@x!!y7KuHN5;5#ZwCh1>>p2pin%;t-SHnoIxd|OdC|fWY5*SJe8x3
z=YHNWKNv)CNiY3g6;ItfsD=!zuy<oGwgEN^vIe5=M|4W_C%XDE&_UH0oA~~zB3v;7
z1f=k7F*&Zlh8DQ;IlQ2ixbMmnbtNV)(~l_r`Fj|fl`qk)uc)UFLrYwuUD>L%w@lhr
zhP>bFFPS63;HLLq1L{+PdWUh{#Nc^Uy$TNuZ^J<oiBrUbw~NHdEX=LdVGgf*7btHa
z_m^~mGVHXXnUt4-cnV0|Zs9p>`au=tX^8G`-kr>kLcMq_khiHy3nT|sFS&f)1N<|}
zw_ViKs&<*apSU)~aRIj15;#pD&fKf~g2r?UaFW3#ciUattolEp)1P2P#zw8kT<R7X
zz{^1Xh7?+nPg;>H6<5UHr&xmc7fsnHTC93=IEVRrIi8F_>Ji}~)(9R|S@jQ?N#3V9
zK2G{%=iOEXed-ermnRR@d1SSGqbBdH)qV{b^peM8&LfyG7<J}hl=4zu_sFt(%iP6e
zKSEp6A<R?UY7Q2nnMa6cC#A%F?FT4Ka?6l*>{0v5>UUaor;&S^&Wn?0Ea^3E9APCi
zit`UI#A;vbG+I5g`b<>I9%z`<jqKfBpQkl*ek}FPyjW^}9;YSBoabY;yE?JPX;>L`
znxoE^<oLtMv&_a*qjn9h=d8z+!`9u%SB*G*^tPP5WiYO7I;*#FenA^e*g27pF(fH-
zwgejE?C(Vmpf-E7{d;j|ryY4ek<Wq*5j=4+wnX=1<2!CGerE1Y{51y+VOKL_NbB5q
zEb@L?D!!Hi&bVLe^1p(@;}0s?@qG8a!%nkx_w|PZ`;ow-`-_F8Iw#N>t8HX?o-SKc
zXEpj{nJVO=wkl!UmwhSrr@C6A4#%du_kAV~lBmZ@z*~lXO|N}QcOeg8Xg2&ny&~+L
zIXF|;Gx{H_3o&G25sC@RfLQXPd^@5W4sz{vX)PA|QIrJ-1J``wK}anJt9YFTDWref
z82|@!7Y*`wYRs9K?|Ic-dpy}Si`uBziM9igTwvQyIXbgXY8O0d^E8$9ACG!sId}n~
ziA7JPXGMp2M7L(5!``!^i+H3DO*ASPJuA9|N0yEzn#}a9=yD#zesqQRtmv6MA}|~1
z`QEdlt9TsaM=$Z7iN3Xln($WQrQCW``W4Nq#WoOKficacZL9N5x9AyPS9L(a=+ejQ
zI5mO&CYvc%-4M2xW67&Aai18J`mXUnu6=(-b8m<jyH@Re%4O{>o@OsMmOMusJaP~f
zW3K5EsTH@yi+5Wq9_`OIKye#>w#uw?cf|q^TR&=y7q?nJ+K|63ZcS*#z)t2Kg|Xs|
z){Vaf_`GiOlW`61Ac(`9(?3ULvWwB19<b`3VUCEWCa;Y}4#w7|)r*|viS2Ji!#F^S
z2d9ORl8fPfViZ;y)g$B1!M1NeB5`RNTd^?FRv`%(_PA}oQEh4u1gbyD8%JpT$AfLp
zLP)&zt3Fu?#d@^rzeS|b{fiQDZ{?YxOvC_UBDN0`F?$$bH%}}9CtPR(V_`E^D8~e|
zyS99EsF*wgtk?*4n4*?DJu4(-gFQ;0MNuB&5RJwjWg_+{6R}5`h&{?g>`}five7?N
z6kxDN`F;cM1mcnetG!?)0q~k-^e+=$OdY@jL=(7PutPx2-ycSr{;2?_D35yfDChe|
zV!5%v#Qe6~=vbM|oo**5mBgG$Se6X5f>W>}xfqj@%b2AK%+JLko~8u(83;eMW2q@L
z@z%s?`pkBsy+5-GrqeqYgsg%Xy@O3+vu(=p-}t2V5RE@kJs;^j%KlQChd@h$u{+~r
z$}yvWg5;fS1>+}EgJ~yIg&8MPkQoL1Oe>%`@dC<&?QCs>onUC+ak4jh?<i1+4b926
zKCaNm^Yt;kHlu)_X$ACMynxOtEuboVU9PXyDcn#}pb*bcC#$b;I&<=B6RtXWjS25K
zxj|vU9R*&vs=&k-++48Q#Mc$*J7|5au-2WduCtxG0)@h=9u-#8!NFB0oBj^oakAI_
zcNBQxssfJ=ZZ7Z$p{~GC!txDf*R}X5_OX#mG6pT%0{E40w%P4Kp=(B{1wpezf#s!D
z#~E14)EV&J6VrKS8(V4I{PJPOU}bDdmO~<s`-2rrB^fEnJ_*~T=5tFx4k>TNL)=G%
z@<yR-C?%St@4q7h#@`2G=hzm^K=xqg(uwENhv(9T=Q93QsKfQo6apB=---s_3B2A}
zvD$kxebc<6g*W}ng!dYLh1c#Yyk=ivTD^Hi&-jBIQBeo;R@sA{pNeBPk{Z56JlyPK
zU(-*gK+lOem*%5Cc{P^GYi9N|c1FgDMAQ1USXa4g0_;S8CNFc&UTNl&X_~Vi<EWD-
zxs{<_vW<4=e5UWr^UcgoFS^qwOB4g7eNguaVdfrY`^mk*HM5`I8(b4R=XVpMV#%+?
zSVG)Ro5YjHGqsP}xiXpebB-MTBI@M*)PHn_{FGgTcyZIBSG<YQ7*8=F)~GSF^ubKA
zPngq$M^yT;0jvJ{f$R)T%^!OKhO+;{b8PGcOe25FQ*`Ww#NEl0ZQ`c#WTIyC5TCtT
zhM>%HR-JrZOTQd8hLI~totzr|E9I@)4bL^RokUCoPHCIj#4*QI87ObN*XOYB?+L;)
zWp!1h@6yv{JT<yOICju|Q&J=UQ)VGE^)vtc)bLkP>+R5Hc@;Kf-smzGme?b`oVKv+
z#KQ7rSy*;rVfpe$u&|u=5iKkc&_05NrGfG+EZ0&k&21}>ZC}Z@3mXpJcshoCC5XN8
z^bzeVd*HqC^kMdu8c!6Z!y8ZZtJ-<K-%lmo)^S$dJMd4E<IXPYraww55<N4DB&6|4
ziOvfk9Aa|D7U_%XItCKfK<N%MhXj!6#1oI0j=P11@_3@&6W{W9qInK$Fnf~8B$(YK
znB63p-9Wo}QVCxnnW+0xm9V=vVMtI_!35wY05<`+3C!6O44^7kyXm8XsRN}~4q7Wt
z*K}bdlpcKpc?~;OMq>S5hfaV&PJls9fI&`xLDu&$$nFlRZ=fPpLA}Qy>sNbk9)qm!
zVURuWUWe9u46@#1ko6vetj}hUbm)6leH;eaKDTE$`kmS{_|&`e$rcXE;xv1O>YW^E
zw@i|06VrjFscGGkpuRJGJ_=VoF((H$(|xSzr$$uWWmrX7<cRitW2s3jK2gTYBF^Ji
zQ&NTdVvVoT;@1V^0b@8g(`|kml7gd#?c|blY>9ghu{s<<0*eowX22fcr93N~m#<<M
z`3F&=`;V|s9CLQZQ=Fo(ZfT&5=AemLkJWRyWVK2?yU}UJP8+_<sSQnnIf;qRQ`X%b
z_UyOV6ZB$E!&cjSIw6_3t0?M>zRheQ!;8*HE`5NAakt6DCGcKysgli`EIdk_O_<gN
zI<P#e&601Em%STxhToRF?0^>>Ho>X-g>Jm_9GebT?_dGITAS5J{dx4@g!;!h3c{^2
zl4IC^W2M9$C&VUBD8*`{HDI>Z5~HHE-^h^&IkChOAqQB&$;fRNg{<JE42>)i(wu9L
zav;W9p&e&6E82mPYZ<vm)EtBj6P%W4<Upo;UVHuF97%`M29U+HKnc<znXoB=fH#0_
z*pWcM3-B{7z(9-#7>RkZ^Qpa_7Dhl^H9*6_c2)%xA|MWsfH*(`;s6PV0~j?k0{l!1
zNI(oQcuE8CCsu)*^3%GV<&7F5(b|SitNuH*Pn?Zr%0DcgI}@@nR8-=QqOpw4UT*2w
zHD<uve2dZcbg%QL9O^X7{+XWGP{M1*X_qBE+MTv92(or#Mil!!#&XnG8RTW7dheiz
zinUpn>+&K)(%|9Fqt100-%+mv(C0~zN0rdmNs#;c_`;k7xq}3E()*oVZS%0zQW?qk
zHr{z`Q&YL8D{=fGlh}syw~Xy6_RBN%6%#q^KOlX4;u#v;Q$G|7TQjdTZ!h&jf$eZ*
ztCSI_ACf2<933dX042dx)Cg!B9JCI}3s4kHWo)Bba8NfSFJO|N3TdEIaL_X(FPKk=
zz=dMLLBWu`0JXu?3c@17nS{tSQ|A+oA*^B_x`v`)n7+iQBw{7DlQc0#{a{`V$@S7E
z{PO|RrAC{u{&88`@8;+JJHWLrN2hHr03h4G+&1TtIDK-&J3Y@7Gy8T?7Up4?MtD<D
zw*?aBN6xdxs0c+~iD!lMyZxOUF?!V;F`C1C>C6q<jj!bCo;l|ds(%T`i#Bqt<8jb#
z^Mrg*Q!zv-<=4$we^2bFHw_P^2PR<I6V59#ED%-kRRfJ)(d(*Ppa%9OO~G=#sOyMA
zIg2p`d&A4}QP}oh!q6l%yL@|43h%)_&REo)@N<*G)?N{?$4i!>OPk%3sYUn6he5n(
zgP~|VXDCOjpg1WGc5NDK&R<zKe-jcjgOl--X&!|SrS_nlE~Sr9H15-wc~u9MUU%n!
zRno<CbV6_I|1t8F_D7(5`&SBMe&QW^h?-MlcJd0_6ewAimc;Q5BEJ(jZrJ0lK9|-7
zL~=Xa?#>uyXEMMs`<8Sg^Ni^Lru-?T&YZ%*vU5WD(kK(=*%E(JBmFa)d)&{9oHV!T
zXIR>5J|z1S?3j1EweL|oXICi}(u3OzdS!uSje6OZS16<!3_4$P`(4ZUa-J<qo@X16
zFLzUxSn@}=6KDzu>I{{Xy<1|qfAL(ulsSKkrJ|gET6)W2O0dE%=8BA;G}*<6^qb0C
zj$NDelLot(ou;2`7}Zr;x3l{4(ojp8^K$hiIE;b+Pv`GD@KCflU_Dj#0~2IbRHmn?
zdcy2-ObJvnrUci+%PuBr*{P@_LTFw2+r(ePf$A-Br@4KQDQEi((8r1oCZ-{)d4647
z{JL-<K@$pL^uydYR{ILZL=3C;LCN8`3>tpFGyFkmsuqQka;}p1Uryfdlcm3r)0Y3r
zb4q{DJh`c#ysfmxlHaL_jrdNvZ0GISaAX<XL@;AL#L^lcR8Qf9is46%o^aQcm=<k6
ze(`apG1O7mn_VAEo(Go!Xw*KsdZ4i{OK?8p05_94Cwrlp_EC;w)vzcogR!MD7@KiV
zo@Vz9{gr_>2K4Et54NBOUG-z#{VQpY{#|4G7dq<IzZ-}z<2Yn0*4@AV(yM<5-(+K`
z`IAPw_&|66TJ_)O!_v_9GUpJ>KOr~$T;caK`hu63;JfA959#m9Ng@HZcsvs``d$&s
ztZV{kJW%~-5divUDpHpEr7G1vha&jTB8Lm@)Mb!^{QE!-Eax#WIf4RK<gm+-!~2lK
z$hFqpBkvE4gd7g`TlCqL$-%c<cMty8z~G-J<F`W(@qf(=#D5MwG(is!Eov=G-oI8Y
zHDX+0tZ`4j*zCuAl91LNlE9MfsjG{+N#bYccq9?=tv7V`$f{cf_jSMgwg&Wz?l<4<
z>@oy%lNs3<k6E`|M;5W<VD=OY)yM-$KK-qY4d{j)6GJx&eY)9<of(7^jqGPP!>1f0
zj~Zy`<-)`1Cqpk6nu8erkqhJ=b~K{3&6wzJu6_iMG|D!X{3c5pwt^Z=VaklD+p{t|
zOk8HC#<MaSc_ZAXxa}H`+KZd7o@A%KOTSmbqd0hV^!UC=oqGzM%kb{h=vKrs8RGPe
zY>oszWf#AbD6>-+1?*zk6x!A#BMmVuA{{;5_jdb$<j6k<M&4savhTkunRi<t?}y3K
z+wuaXKcw3>^s|efUDW0YVCy?*^4g8z9>J^0y^d~ZyVZy*Wb&f>5br(YYV<Yo-{5bu
zjM{g<oJ+eWfr<TGT%kEkBm3MR6DeY}ZD%|(YbghXgK;cQ@r@it=qwy#6?_Ysbt)q3
z`ADp}vDBZW<dgLF&7Mz<&$<uz8i{o(0_*w6tK-q;{z?0W`usB)pKFNs{rSJC6Rm<<
zHBhXA<Z2Vtt|S1}Of)V+rOU(oO4NFq%)aXJZ{8B|RwMBi7|%ggLG2EM>S|kfZpy$j
zlUja2k;;`+aBGPPl0_ykuFxpbeBOlW>lvu-6hxm*`){|Kk8i!n1j!{Ps1*)S=_V|p
zdYdp^yADD0jjGjrd~1sdk_{%PT|)pWUDPI2|Cxbm49_2cDy*b}TXn4xZJuue_Du{b
zHN8;j-k@};Y6Q`DsjAG!x6U*{vfKo<A^}RJ!U<IYuW3|U1kpDt9cQLgcbXu1y9sJ<
zA^=qjflzG*Wg1nGN`C;V93>UpYJ6%JBwIaH!U3udzpbvzKvnU<sB90_7!xGJCa4`o
z04mXzN>!ABYPBHxuHZ`Z@vSu`NLHDkRx}POwZ2g80%f{_t|0nGwZnXT>lPCv*P5XA
zP6AM=a6<L38K{aFj30n1q@;pdgC<Dkn4ngT6kEP|3~!(kBZ+5VTOy#o3ntI_V5=}e
zQkNAl28G5u@>=fs5<$K)qIs5j3&;4_eg`kg<EQ`g-{s4TVaI2Gb{jV!Sj0^prqtmX
zu~YIDU8LyInbAQ-EA67enbG=wa!ApInbC@wJVw#uGNZ$a7N(1aWJW8u$x6FuWM;I!
zpIoA7ysC5;dkRak18AsAssHjfv>3zJyQ~?jToxWx4HsJB(>+tEM|j2J#ahPYl`KhG
zFfYNhWSbSSFhDt%-LmkLn7Xv+Lc=Y9Rc_&JtBg<Z_#*Aq#hCV7%g$Q(No)Gk%HYK`
z)I({gAW*XXxx5IQm^WB4RXwZ><JiPamvziu*|x~{OeJ-^VWO_GtZGRYQ@LX+_>eK&
z+v1@dQb>MWjyPR+cVLX8=x6XeQ&_}pdPF@$(#u}-OIBn~VTb}y)XU1P$hT)whQE>S
zna+{v#Z_c_Q3=hk4b!4c+=#*B%QO&8v7M{)!2JaFI33`{%x(H!qA+e#k-3~9+(q;&
zL?4u0m#l?A{jkglsoQ2g-C{*<tD*VH=+EMCgzHxmT9F@U<{|5OGs*8cRrl2ku_M&N
zu1dGfn<kmw1n!4r7bH_klY%85g~FKUeJjEV4c=biRu!&&vm$plfD*VHTTBk?TM1OC
zR^tA%Lvi6R%62PDepWxOu_6m)H(Ht4Ktn7`6xOH@9=(TOb(5~_C>@XwmW72Y)qf3O
z`WxV~czQt*9Tc5hJd3B7XcyP2z<5&yrqqyS<<br1!ugnw6O%3s42gf5&np`tuSyK)
zrb9$(wHB-5NsqojL9dk-%KECZbi+)OyQqhFvXKca-^;S{P4814E80}hAjO?i*r4y`
zY^c(A8@=yViwe2uL@mP=_7&8LZYUiP=7VX}&rqKg9)d}2CEY<+(l-qklErR)B~#Oi
z-Bx5|A(m>!Om~lppj!&F(m9+hfRHJlDQY>i!bP#Y!9t=)0A|Se(K9`5p;#2f6w9_E
zaZ1JTqtO=@uAx*|D$anw!ta=Z88TJ5o}$&r;w@EqzY1POk3g@cru#to9;tiO)T~y*
zGDU=HO=2gBSb8ojtf2&4W+>~mu*UBetAfl;Ip&jDhsYG>G5sxWp&=O?Gw8;e-l+mX
zwTgS+s#uDJE3~QXH=Bx;^Rz=$3^`ZOnY6U&Kbg^m(<OY3G&+pcEYkOZVOIU+=nk#O
z7ez*l-bpYT);2tZ_`H(F#i2li-F{+=cqrEcBYo0Lk}s}0_PF!Miar6-8!tSU^Q=Jw
z1~Xn@qMSPw`^0l<dcSu08@CfdH((-O<*w#u2bP5Wh=6)qL_Ryj0>y-U7U)?oTalA#
zjB|qmB+anY18K;|XgR}DDbB?~;Ql$}!#yt2ecQlkeEDFM)bP|rMWA|1AdFTG){wEr
z4Zmc1hh_rH<PysG%xaYk3qkR<Cunw+ubYvB#Te~A1~>8J4d*N&9s}7&t?BRk(FfJD
zGkJLdvVB;52+k|+`fZA|Y$jZ`iY#xbF-?S7B2{5X_82<uk)0zW?+qxdm4w$Jjo1CS
zS9ynPU}sItc53i4`9PZt^H4f_y5?>nR~Lz)60A1Cdht`T{z-!gjfs5K<L{X%X8N-#
zsCX%CbowDtVlxl0_qSCRte}l$?lvXtg$9Iz3=CaRPB}cgt<{VBa|a<zw0r$7QdVD0
zTR}5p@^2_$o6qLNO*lCa6hkt`_#LZt$4m`4)MqZmQ|6DTc-t+R_j<HKhFG`j<M@hl
z&ArbUgs=!&1DfSym?l3@XgFihycym}gKdEOHK;_4J(yWg7^bZ-T$3V+KGqln<@%O0
zYq$so0*mn8-!oz(qYIUoSYepwW_@027{or=D;s0dL}9BH;YZaUK>yEx-*WM>S4c)^
zT&C`gVnr*(mn_pCO>jUy-;^$DkttD0ca@-!-XJ9gMAe|s{InkpVn`COrKX5alfkoA
zg((lGCRWnkqFa^j&}yhm2JYBhhd1-=nsU(#WNeNXaP4%3Xm+F8R+E|G%;9!Gpi2X?
z3y@YpP#7pS!3-LPCz@m(1W=-E<2(6NSq_OQk~?K-QQvdCVYS~J6a_PV`oBS}F(hgr
zZ%$Bb6KQTGqj(`U`9}D7(hNnDu$=_yEzR)=h>LF+ye0|GUNO25K8iKtfg-E^o8qJ4
zZYHEzsY+Z%4|Ez<Qih$PbG9N=;HREkSFVSdB=dJ!736S?Vb?Qcexd3Cd|Rs*n&INH
zQE^TD5g?D|v*r+K4b{(NL!?N3R^?Siv}e`Jw9K+O&^Y|+O>wPd3pE8-Ud>O{W+lR?
zdB3zu17oj&@D3jkc|!?6%^r{*o4iMZxWRi40X7)~RgK<zkoTX|P&lGCD2l^0Ab$Kk
zW$hX6lj5J3>EO{nMboOnW*n~ZkIpmz;h)4=BZ!Q{H9qi4W#k{Ov1@nN#GHe%+BOa%
zVG~rsvIm9F_O+~BuxLkqx5BX+wNv&v8xh_%iMY^+M8k<S(CMs|0{olg3FZ7ShaJ4~
zi<<)YIDi*15~owkN*R(<kXP>E!<|+*Z7VMMyz&Snt2v7EDCUq2<4q&}h)!^ru@v5K
zz(B`nntqs@^JHP}L#$EJBUi7X9LP_kyrIk)Rgt?2SDUmO8iW`gGab1e;^h#DKsaH+
ziJKZgc(|&CWJKFXl)@;COcsrE9EksSk=47nrz~vnVCQvEiP#!`zSp3FX>Qt^W<P3*
zgZFS8Mu<ApLOV%*kpZfdJ*1e-chDx$7P{X%QUWgTDr-g_&D2UWO?O!aNcaX%JEr&@
z%Ao|nM)72-#EF{eji43)5)@a?Y1D>WXd;>0KdDjvEQ515ksJ|G1)1yqCh*6JuJhX#
zd4`%wEK&w*Wc+Wn?3bCoY1#1iTKVvRIXJ?>lGQYUFmR?N%;}<4=A;AS1#_yb{cA{8
zORXjZ4F{l@OvxX{GW{_cqpAxknNzsh9N4LNLw%%ND6(m<EmruuYFuz}<rxW@p*jfp
z9O=L6DZbta1EaQ&suZm}aHyd&Ak-bAEiIa~R(nm!&yoDRiV!4)V^7H5Rv2bFeGd??
z3j(>Ig0-O9RPoG9FDqKdA~<jJ46$i+U}Fk-Rvqe7d3bQ4q1icwHGFzl&E2KbN2^tC
zNy2}l1-R4fUaujIFz~LRHm46137N-BL~u1!1V+T;yp{Z1x|*M>)=)SY5|0xbh-<MT
zH?%_S9h{NiM2<92?3tUOYl~HXjyaCQ5s3{eDMZZMMvs}35}(xJWFugTA9mYUPbDtn
zXyVGa!*RrykToj@pGbp?X(7hNou+j)V4O$$O&g?w?(h!RJq|rfbb8i`e5*!jYx+)G
zqqNq((^{1Vm+A|r$*-aBG={*mK^=XkbtsLp_k|PUn>L7{*H>B#E3rX+CVeJ-0BU)Y
z+P3=yR68c3y-z^>nK)gGLO(RnL!LVu?|cK?S4W(><+-@7+2c4<U@vW~tzt4Ce0)#K
z$DAkVlD~BGXNKL>KHI-ydW{vCUZeCr+p=Q%O43&<z0Y>3n7*2H_-lCTn!XyKV)`1=
z*C@Tu{-~G^eNJysdY^4u0gs57D@Lcb_KhF9HW$E7?dUrlnxEbQ{!jLq{3)Fd@(JGW
zqOOPVduF<414BWB1plxc8oJ5EX_R<zI!H8!IQcJ{3UU@#O!_Ltc|jiIM&i^8$;jKx
zv7T^BGjI&mz^)Ph*?mtu_46QuizB|>{fpfy1L_qGs62LFD&wbsvU1ufV5`~zCC<@;
z_Bmb}-eaW0*WpYJ45%G;DlaA<Byo;C8*)fbUA$}+EW8?6t*IJeRw}^D;7b&?z>LHz
zx3T{2p3W*-;UI+|sw_RF$P}Z*6r&tKis6?aop(GHA4oA7{mfP-zs#H@)FNy4WWmWs
zt3tk(^${7&G#OO!yXq9fHTuqD+CVkJOW@k2yi{Cno@{9~EFIYhcAS|9wcN;BV_=9=
zM}H}22pV~XOy(PbrOHSnC98=jH{J1wTKChfG}Ncz2E#v1vplR;I!u>Uy{X*xh}A#B
z4Dhfdu&3a+)3rd+>W8U%pFrLb$Wwr97f7o>UJ=MU2E-M}TLRey$jji}O{w1?d|W9K
z#dlj3dsL(bBHt!)Z#q$WPbLQhv&!-oBGsnvrxTwd5h0#7t}JgQvM;nOBq~cfNNnND
z?Su`=vxzUI&ms|92h)SPhGnbLw4A3Eae2{Hls?D#eydut{yxZ`^L8Q5*cJ2F52wYk
z<xOT@F=slzBEu@@z_*&R?2pHo!*96Ai^*c9ULv<W7B9YsU|0C(l}WK&6xF|(G#08(
zX`wk>hmsuIKdbJKj1{@VepTSIyJ+o?;a!Ql_Nh*DTbHw&-A)e3yeW%P4CMC8&qv)4
zl}qg*Oo8>zUMMDR{_~gQ+b@L5T^`-b^;4psBK=S`4!-a+z~;nDEHxRUu&uuNX|&ub
zh?elj7GX5RA6tajv|%IB_)($Ei5K7RYNow_GQr;d19OlgnGG`&g;v4xHR_3i<*U6w
zlMHV)CWuynVx}n(z^4;0K0}b>1sxu6s~5C*K?6bNCsiKoOfM++f)Xz%@`4b7%43ed
zY)77uZBMH7${h7gJkVh}F76}>gLslEl%FD)YCM83>R4*xOxwvTB*cn~eM_vMy!FML
z!lL>&qYqFRc6?^AF?)-0N$a<KvhfIn8+SI!rliO?e2HPobA`JEH^k+3U^liiEx2`a
zcZ!X7N9|CQHrsllS{0>o^+dUf`ejyBhoXLv6?H&SOEaS6&a~70PDTv&JBqqED+=3_
zPPa5G3e)&bcT`psY^&2fMNzp|mQ$H9?>0uP&U<b?v5d)?(XNf9#k;I}m%%7ci$(Tw
zoO`Jot@^9gGctj+lNCW@=_Lo3LHFh-WJiUO9mWcd&2fmqb0km1l0$802xkl~qF1IJ
z&v6A#Ch~w)!A0}+fW_HGm<dcfKEy+n9>P3eus7}aA|9~7n|Ayd9zuF3i6zaFRnHh)
zT@=<sIg!Thj^_#@>Go;I&m>ZmtLOPdV!=1<_$nd|Z{V5kyr{%>Rvljh6!EJK@T%j5
zi6r*#R~;_|B;DyJ=~I%*m9*MRTH_}Pgro{3DUPq!`bj2C8?;Sqfuqp6W9JkWS~t&!
znTb>2^?X0(2xEH85!(I<!0f1|-nN6q|A}?lNrIE#ZH_1GPn_lMddxE>IpJ~4*&YVM
zyFT<p3bXEMeefy(?hm$O>+lI3k4M(S*5O6odqf2@pLL&iL~&808e#I~?C0d;GyUgY
zl#{b_T%mg$U+u)=@R|M>`!V||%}zErUfBCg|4GC>)41cM=i`M-o@wN_%Dd(dTs8Jt
ziro5Q+w(hY>s%$JqNDMwyq`mN2LjQ*LSs2`YZ8m^VE-pB)OnCMwYPHd<B6*k)j*W@
z9(US(`)+P%#$}QfL5m}^zUwu#(d3lOv#qO`^DggyJ|E4qX7rzG!llSI7-o!l+syS9
z%zy>)!uii*qQ2vVdLW)>rvV67Ja*SZ)xV?wESQ2{)6cSetL`2O$S3&o=a-|t8M*6{
z3(7d$l#vP5*PjD{2?)Pz*QtQdfk9<s{eAgA-962e?VtW^!lgpM#6;QHSwNdt#Ai4q
zr*suNPIw&{N>>gHdaPL2Bh~+*QVrD4lzgK}rrM_L)GlZW7o?!=1==h_TV~4lg{7c*
z+qx9(>5NV?Iz&dD=6;WFA}w6X_TP*xM5D<!mvO(%W!$(iq&hcnEa|0?ecTqVFY%pI
zc9U&W4-KufMfdy*aKvaz6%HAWUvzD3`N4Ynh5J{M7tco$zD|?9)IX&R)YKv_k~wdn
zj%--5N8`yM7*t&|)6N}@DcazEvEsMOKDCqIm^FS|3^P8Q^u?XaU@C*37B=HR>(KES
zc(vFL$h#V=f5-WaDir}&H@wH4DOK_0q?rafJ2RNpWAWtRr>CTb<l|~OUcBA@)IR-M
z;~V8LEgsntk2Kg0#c}e;ZVK5oLatQGwkC4L{l)Y9mPdk^8rm81$P+Nw`0O>3M>KgA
zjoQh;?OP_<qk`A(vu=`^hWJC{>liM=hTYF`7?5$xDJMxK;vX2y;$!m}t`E_<mP4HT
zGtZpCl_i>aK3UeUgNzZtc!Ub<jwUjpo*$A};6-Q18XpiFlZ&+rg#&$^J{*`Kp&0og
zl*lK_2ru4*??sdoS&|vyNfEUhLf-sd{RcZ>928=jTIL+qNv?c<UZ1Gx1_gT<OS6#X
zb&iS{dZBH+n@nuydAtB?ni^)^Qb()#>)T|VIpEekp>M)zq?yj+S9HMq89VX7+Ze6m
z6wkF_h22r9>zScmaNoxiEF+Gte8GL4xUyWX2C!4tGex~XA87U~avzo57&fU<uDK|F
zt9nJudCsN=;S#I<_olOPioI)q=1trJ&-dm{BZRoqSQy$4v#B?4$|<W;#tci>I2aQr
zISJ_CIx=}aa~;kflXyjU%0h%igxI#9PdUn%cF$*qE8|va-*Wo-t(s_>OUEJ}SvSM;
zjU6>}@r&8BO542vW=9=0(G0tjdE+j@VdXC;$}cj}EH&?Y-2Zs!116g50B}zuwe1`u
zvzr3~SM%=mPjC7CDY>|&ZtvSk@2XbG?`;xD((=39h4(#HhC?Kh-!;5guD(f&6C%Lx
z-zEW`BV*kHyoL$~(j~CZ_k?%ze6A<Fo9E#=nqhgu`@mT9q@agRvicP6zuuZS(F~)2
z-5m9jjC_8q6lC&6SMpJBdHi@u_Ze8s7#fxs%y1b-eAr|x>>IquA|gwAM{?jRT45pq
z^T*RKjjXKWs76rMfI<?)O0mp&*B@AC(Pr@oB+6MSNJs<g&CK%YOr_*yb1EtByl*^i
z+_0#7KI<(nZ_bpL_hib;#u>kR*M1GBv;>Xx`WpV}iSmR{MAz4h5PHO|B^DucnDN^B
zUiEFtxX279_H_P}ikeXrPgbR+Ppajf2(fWISL=h9aON=tv->f2lHDMu3X$_`w)5p!
z@*1kSh_{RLW6rb_5kM~v>0uZT91oawVu**Z9ykx^kD}W)YZPrN#k)77sEtvSacn13
zT)PUu4;e*wY81^AoWwO`<t|3iSIzs2?xkBkU=)1?0JpW7QS^B;ij25c1a#K;L0ZK=
zmi4clyzzFLh+9(3`nOf4Y2K0AxPu%&>P^l)^gV|17(3>m?IvT+_B!*;KySYJ1GzKu
zIW_6cH?Nviu<dZYW2;qXgkv-3=sj_jH|Ln=(ONNo-VBxraQq1~`C}@WP{Lmke<A*Y
z{9!~_$^lF63}QP-1MG9=8*i=2Sm2w!7sVeJ+B-6P5|r-^#!PxwzNeB=6@L`;12bie
zF5gf?8z(U3UEaY1L8}+EctL{~tnq@?Ua-;&Y6vt)GjM#w<#6^|yBCfQ)roR=L5mkO
zc)=PkSnUNXy`Y9bn{Xb(u<DO!h6r$(m(GWz7uDY6Ya(46pMhpQtvfAnn`sX3?)Ii&
zR**MbLt|#Or-Kw~C%-Rp^eB!+HOD59x$)OC`=zK^JK~?*m)~Po{B4PytHsqG<HqI%
z_u6-K-VQH5o47sdvFaDhH4H*ZvCXP<Lh-KZ+gX&D)oG6EWiDzGf8b^cpFbEk&+c0r
zMZ-*&#$Bnr#mKj4<4gImRQyC7xUxT+Oq`U%7`mVX*GwmwZsO&DqM3n5U!d5V)Op0_
z!;ypb@<W2E=3J4_Ef#Db6EYto=TK^+vFMqsn^}oz>XQ52J@AN(qe*RF(-&D$M@_!P
zzkzL>DL9sDx-Vl7Y@@4D?=QMvq*{Hh@;U)<f71lp(B<^{6zGGzck}O$#6QYEI{&zs
z_BZ_FD~5mIT&5NG)wrcuOs*0fn1fiEwR#?9UX{#BBeio<DUr~k&woCTXJl3rj~I%^
z&9}6t;_jUd9BoQ1)M4zO$@*S0ZBP1(?oFh#AhMS?^_BXatW=K|+)NL8@<?x7<9ujm
zhmMYU7_&0_OcuJtCu6Axb8x1WJ{WL#eP&s_CI0f}niF~;o&~~3Orfp%YteDZi-c*p
zfa}fND5^>_hvaOH+lYo_s@_ymDlUIWW{tnz<MUcGJbM;s=tDkae2QZoYkWSt=L5zk
zyp7K39-=?)9-=glRsWFC_Ax}yH#FUQh^7W>8?bwrW|TjByowr*9<P3E_ju(!W4x|=
zkPbL{yfW54XuRI2FOEH4w`#mD6I{l4onqc!bgzMC`W&y50dNmO5N(Y;<243!-{ZBy
z%=e=UbsL&D<D2Pzsu`}`%gbSzi%X2E*%!m4FvH8j@64EGjd>Oy;vC~Wx<-O?gZrRp
z6VggMTHrtn8ohmXYVa1FvoALaD_uUePr?A8n^_8_9nW+znDs8lGhI3zr{@oMJaZAl
zx{8xVAL9NAJ@rR$5@n#YV6|4fK&ASc?~mpG37-aN<h=XHj^Y1_T)g|pzTcosKGgqH
z54`riKivOQv*#~^!{_EcvXiqr&~u2;AXnHhftaH;^AL35{lRU7`Rw_4VPS|xGk$=K
z!#?X4>98`!6Wz+>k7=(@QQo9&-QDQ4c@y?u8NNuCPkH<@cEuY@V!4&f*Dva*tMjZo
zh-}dlHM}Eb)wrJcxB1MAN8vUO4bqluEfS}AncyEmU&EvvMzf-WbV&RgxMKv7ma7|#
zLuRXfy4eI+`X)lYZe7bOOkcWI;{LvoG0Rclg}TLn#;|V2Lms2JL9?#>x>a;5sV-@2
zyx)xCj3}l@?i*$yVLS(DLGd%eFlTOh-~&eN9RRqKNNsEC8L_wVezboa?<4{(hocNe
zY_w+CupF-R*gR=ij#1hL+fGCHey0BkQBSXW&tKAd)vgjR<`;_b^{RV_GkR6eWXPyi
zmt?9|{rDd3>XK~j>I3wXw5zSg%g*0;V|1V?JYfNtnUB5xMdir;hfzE-0qt=|zQF;0
ztMY_AfYLfuRD$6Xaus3C;Mv!y@{#sWvFa-^NUG3VR_!?9OyJO4&L<}{t?KYQP&5jV
z%q#l8`whwge`42+ONN8}pW_|9PB;HEB~f`aZBd>(+Et!#|2p6Zp{dIL^Yx5Q6*cAM
z{LNr9a;cDESwxvy$cB#%qM4Lw=8hBojSnbOg@LJ%ruTNrBtSl^IB6!y+6xcyb%Pat
zmH^dk1xndq|8XXbY|nUx=A+rH8(D)=q%($p;m)5&bAUNv7%-$%=Ud6&k=72fCZnM?
zhc?`2AR$op4**8Cl+wMLDD9}1NO!GlTcD7Fiw&^^c-_M5^vU`BR-A~jkZ+~kL)$&}
z+U=b^V_W*VY<UlDzF9B!+AnSFnLsk<4vz1bxdZ(JotncxG~m#U(NF%m_uR2!oi}$B
zaRL=t7~V7{bM8P0gth)ZKX-UM)l48StHK}8f8hU{^T%)vu%1pd{o~1w=01foUF)OG
z)W!5Iejm_2-%`Y*;iz^Bucq{6aMVsuUynQ=Mj30?SDY{Ftq*E;p7FG;&O4sM^+Ka>
zO<mtA)z8m&dD<YUU{lw#E&7_9%!=wz)RL?ywnbla=VnAXdlWM@BL>y2qDr!&q+4>o
zlof?C>^1k(M3v=MOP3qs9U&#g{U`|BVYlm5CJ#J0;jDi`F|+TfX6EU2i=ceG7P+JU
zqM4_O3kO;1y{=I{1}D!_?Hhr+$NokluiOmT0d{luxrguPJI_AI%v*&^pVw(?MAePR
zKqw%426bET-cEDhm+>igNmDzbZm;R;tv`Cx)leK5_bFh$34nV$))j5#J=2xTdwPGW
zhd(f-n;YtN<hu>h$Ga(7v&nckH6H2g;%BT{?V{W3X%+qZMwxt?Ziq*?<x3k5og96)
zi(jx-+%Kr&XSwV*Zk^j63p|M@(|GY7yqe~3$DipQw&vwrG!!d7V%?~_uFv!DB6rWB
zWHDzWh?RRhazHCcu9Zxn*n8b<SZ{F8qg7(5ad+9NAwT0=R$lInf1nux&&N{3?~<Yp
zV(4<8Ag;_g&G*rFm;Cd!pM)h;KjSe&g3AR!!~&P9>2cCBBkDJ^=ezISduX`J$SRYa
zmiqqhrm~@AkJ^e;J*#dfBPE_R$J^q`+qO_3E;h!AveU$N)dA1l#~#l7e-gMD$KzwE
zOZmGtSmrdxy00b!z~3*#xTjLivs*YR|LqQbPUY`9J&fRQvL3$5Y0aSr0{Yz=`04(G
z4)fg&<aEO$Kiz+XE6X|PaQ5dKP5J>$$$;tylp;5n*8{mGGB+J*UR?vfw;P`ezV|$Q
zhxF+|lkOS>d(-$1nOD~!*qg?8h}SVC`0gK5f;|y_9HLg|0q=-^Q%HUsVqBeO)pLIZ
zY93A>cB;DB7pFz78y<=kcU0r^;F-2-X;OSj;K~7Nc%PlV9XjKR+ug`Ji4zclb>RwI
zWc%BP;PUxTa`}v3{JV$zAsTps&yBz5vw864GhiMR%sK~m<9Cm~Kk#TYS^8ieQA^k5
zIrtdAJAQv&fc@y^e(~bwMJ>)2bCmWH3|4C>NalRH{0$v6N3t{1WL@D*@yzDB!YV@j
zOAFMu=Me$hL@T}DodI3pf_{czQ{DvucIsa9?zRl{5S04}giN|c{nPUfLJd4d=q(YZ
zU*R0p5a)AcJ!&iCs{Ql5*83)3_QXu#k~bHcL-EO9NIhpKe`_>bBYs<>H{;;s$@FMP
zYQ>FB0h6}-47+69iR{&X@nCjZR@ziSnDB10l*>cS_js06Ceqy)4<}{^fm;$2Cvhu|
zx#iOjy$fmO=%l2_(a9e%JpK6hgT|8UO-*P5T`+5>RuyUl+NqlhHxiaP9qvQF6rHh4
z=yykH()bPH9-zAnOlj_2<{Qs@(O=gX21Jh{f3ea?c~<dAo_5N+3xe%Gb}Za!{jsZc
zfG|6WiaFQBlosaoM4jnJp-I$!qo9WSQG?S(uWiO<`MMybMbqwn>C321>r7>0$see0
zW20f_rb4S;JHv8G7!=*yY2aUTFM$l!qu$7Ib4|%N$5I#NMVki@W9Z9++vm=uXQIhd
zv~Y(z?sxA)Fm>)I{13olv$<gGvT|+R$ZZK4rh84i8>OfsJC#EJaBN#Bf?=IEVy2{P
zTvC*dDAyB?DN09_80!UNULhL=+^!*FWB4!vrT?@MAr=-RQ2XbNK$;v;C4@F4qXIJT
zsfuwLTNG60yp`zhev%S)dutlNi`nYML@8SQoOQFgAIGP-X!Ah65|+e2AsK5>_OSD|
zhwkd^N<;KcY-;CzUUp@;;Yqd`s{w7#)g<d<7RvsvL$ua}_xxU=N%Tvd*yt|QH-By4
zmge6X^C>Go?#6~cloh{g)%`ogqH_+%BkZSaz%$9qhaXH*Y-Qa1Tk0BlE!NoCuPk|T
zS@JUO20fm%*z(S+SR!6CBDxONfc0?a_-js$uA^V{eA+di^e!ajdB8OT%TklOV%yuw
zB2UD)8aI|YKM*h88?(-Rlr0XcPCZ@b>@SNv7TdlDcc;bjkXQ=Wz=EJ7wtpX!HT#BN
z8}+z(ZW*5(SvNj8?ygbL{lqR48$D~>D(66>`<YP(7BrkN?^87RzC<uy{KD0r!2?xb
zPy1)J-q7QV?L)iQZ_(tmKj2EAbUW~LVAFVKWE~9TEDvK~+^X>oGdUg(aV)e#*C9^k
zuE=<C<KpTv=k2n{w%GRFWr6J({)1FroS37}c#854GI(s~>1cA~1JUI0djst(dj0a|
z4qxT$P<eMOXqcO<xWs9SyfeyW&_2MI$9W|fHGCDW5${F&K&I8d>?H~y-}dw;PGE+i
z7nN3)@%)Z9$%5ki#$l+&FEUTNH_wmSCuYo?Ucx1cXT^#iOQ4hMO5{#(zQ*L27k8HB
zw~4WXU;eO}%Nh^#L+&xxVaBb}mt%pA){h#r_13ui9P7s$@;BJ#vQ--B8}bMb$$DEb
zi8Ews(w$~J#MgrviRxD>lKTdv6+?)^nU`(6lwq{`h(s!7T-$m$9&i)6r5NzRx8e5Q
zCd@W>*MG^ZZ(mtz>F@H?&(wIoTk{(RfLsH`oikMjc3yu)1*Y)+mevDEQcZbwF&AzQ
zprGx8-BZcD?Ynx$fz!reJ)_bnUXs_FG27n_3oiq#9Pr-lJp^T^v?u90ifeo)-yU(g
zUA$@WY3P1rJ_bN@%({CI?J2)&OLih-7rTq!Q6KMbpLz8~m~zG%-_qqPh55bYbC$OG
zYAIt?YMgMqB}SZ})EexKuM`R87lbmf0}&R)*KDiXTD`d)M!LrxtOOZx<l-|utBT%y
zmK<O&*z#NO+l_4@XG3J$_T7PPjk`}-u<f^9UCR`Kw(7y#p7>J0wkEu7FL-qF9SrVu
zyI_$2rLRQnFBu|fvA4UpPRBdkDfWT}eXY3~bC~<}2~@E?w%~0&ngPu>NJxEsvpK$C
zvjJQFT346K^;#%?96GA6aq|^<@!QvTu}DK~!4?nwhZJeSUJw0C9{QI+-w2gJLNJ<R
zl*_ODhgCVgps^Qz`#6YNRQL}qKZE}tTK>>KxcuRNaQPW0|Iqdy^A9e6$v?RK_CL7%
z<$qWCJ&poHS>9{dR(RJv+YTm_&6DPfRc6|I_V8;wq#^8ZLkd3JYY|Cf$xG26=!TRX
z-VG^x-tzu&`w8lRR;%_dF}Ss#^ZxaVCBKT%qIU^%)(t5yTPw~5U%dDwt~=s}ls965
zN6ZZ=Zu*84S8MS%6T@T0o2(l{%KM5VOWeKxp=qTNyebNs31&X4?k4FvQ|I3li|mc9
zb#C(hW!$llbumg5=3LF5#+J4-5Z`c1j=%MtiIY5j&5ZRR$n56vo-|wAw8Tce{e0|-
zjZ4f`9oH+BJ??5)-<Gz2S8a7s3X9b(?dQwz<QzOHSK&#y$_h@LX$(+;6J_;+Og*uL
z5Q+L&9+p{DOUJ)N`yHUV0aQzTR3!-LLRI3UDmfZe1!8$QQn`&#PQgG`+web%%G?7o
z0Qr@b`Y7(ot)T&!oSG2?6O)k_hEm>zK?+<tl**R|DRSvh%3K<x(4|9ZfYKnv&Kt^-
zVrY;h#n2E-ilG#HMv$LrK?)xaQu;hj{x>h-R@3BNLtT@{T|tE`ScdLUSWp)<A@5uK
zcN6ly)_*r4xh~<_j-fZ1k5>)7-Gp}xz0&~Q5%j`UK@(qabI`z1N}8^DiB(qvn95+*
z+Rj};g(`#A^D2W0c`xzbO~`wZ|87FwJCq&8$T8oq8X7d=9YaF~?v9`rt_qrXieccu
zukYA;y(wLi2Zrw83)^`hs8A)@VldPNP00IN|J{VV-|4@bkoWoKyPJnDG2yDAH<|E`
zp|=~jJAz)gDrn*<iGiae`VL#CRt3dRAP=G-%@|`>Y=CYT<}a}wS-)cR@p|`qA$xpX
zsv2dCFBt@x@nr{CwEDMkwEbLS>0kuK6JyR*`LQcDEy0>w3cp_Byr<dwG~jOiYN<e3
z|740<$C}<b>}8~_w5YcCBT6g(E1)&L6#8EQElXzq00H@b1vGC>^Fads{|abb%RaV%
z#%A8C2J24^Y5ldGJTV!~%MIj4^K+wv(Vb@qbLM<Ua?<?V;93}WPNZYshJaPNIRhf=
zerfa}6P!)XmZ;8$9j<?z<C;wNF|0Y}Cyxtk(ar?NfQ`Fz^rWn_gCON5qkGLE+_Dj#
zUkcXc+_AndF*r6GJ0p&leTV%yvj^z_9YbZkhR{xqD;%{?aO??m_<oSFBsYGL75aXV
zH$}OdXA$dtHyQz5+7)uoCQhjXwC@kud-`C1$eZzpys7$;xU;DpyAvEfQ#D(ckfTo<
zNehvSwo$R}c{#n&LyFn2Fv^t!cB<ll-PoRs`}MA-M1fSm94n$twyIqcxZ$iE$lsLU
zU-MN*T}?TT?Zvx`?1<~jOC(>Uy`W25`I)EfpEJy+iq-iLA-gP{k0861AiH%`%U3cH
zjYsXLSU0dSZSGeeD0dE<+A-JezE*v(xmLZpzv9EXTits+Df5DL()ze#J*+7)mAG-R
zgKHgs|LPd`tD{Oi=eEwS(!l%sngWUA9JkRuHQ=@nN=*pxlq&5X*v>vtmJY?JTgMf8
zde{ITOHCB{X>a>D_Tz5cLJP)|Cq|RUIa`Ey%k~5D<i*$${aN(G{?0jOcPAeg4E+{;
zCHfapIL`sA?m`~RQrC8AJFZMV!)xECkFKH__q<vC2@kTZ03helwx6lnQ<n~jrM|~s
z)eyl*q13(KtNlA%7MT4z(_N>~WtHl;ukm-j9>!|>&D(G5*}v0%+tK@X-hNxJ{X1{J
zZEJeJZG+i<>!8SI4?Wnuf7b!NBi{bqZtdT-ne+~`VRtaSf4AGbb{GT))A)Ax?BBH;
zQe*$lMY9*u-ktXGdiL)~R$a3<@2nfPmK8tiJeo+vQrEF-cgWkddzM|hc!d9VvTJAV
zXce8k*#3#}$-JNbbo~BtKOfJo-R*(xIE2zb_^R`$_!0Yc>$G3@py|EAYXf`PuUlvK
z>mJNAdw7YpdED)4_UoQzzpf*Dziw1R2Kz!k_+jtPmTdTp^{w-y*&z4!h6co%CmMdn
zdhol}O<#ob#b>vtH?nO9M)uvGNcGyxMR8n0ba$CE&%36~P7Wz_pS)GO)AM8i=eU{l
z!*cP4v%4M|-<)?hD07@mjc=YB*tGo!fb}~aciD?=H$#GQBakitp}0p*o6I+y7x%wd
z+tvA3w3l628sG_S%F}^oIVBle$8{}TK(HRz9{=ptQTt+_-EP(Wj)*eHjYYQbl{IBg
z;CWQ4Fb4+WCRo>-xpY<6ro?gPl*UT|uBHn2Z~IhE&UhAQcfV_XUwzPHUhL=1IFAb}
z-X$LlM(<OZ?8qbWz!6Zd^UCsA;2}66*D)MmO$e%=gDP7et8*dzb@129Ukl}EfcQG9
zU*ql#ISVVs-t<OLmpX0xx7Qx*^lQ2O+*JJZe$F<IMzXNZ<L06f4DQ5d810GU)tah0
zjo9}iJ|-4@1#6|dqp1r6wwX%=54|)8^w1nV>^0qY-{nH$!}w+BsNKMnx2)CKB#!wB
z&S3UlY=r*6DeTkux&KTstUHS*_tGD+_Bd`;2LD>S<;_`{3YE33+3(z!<83@}(OxG?
zc(aYT%{}RpG;Z;Z#OICwP#!*QCo#O5Cf)pgx0!6(Kh09xJp!F6I{(4qKSoht?00`l
znpax%@t2~AZ+3S8l;+>~V|vH^^Ct{;>&Feaq{;4|p79`0%QOmFAGvVS`m2~z7K#2b
zEK0d$WJY1|{$W%cj*Tg>yKyDivb{LC7e11)>u@(wD(`k3cVB6KsJ~UO+m%?xS#@D$
zf@x@eY%JAFoT4i#2Vhk**y*Q@G)`_6rgu|V&y^)dvZQQ@M{wt%69t>rnU1qc+gLKP
zry@dbY-3NQF~2drr_l8cbnh-ZAb+8}If;=O@lq&hwU1<?p_1d?p@On!NH(TWWlm>2
zvRn7|G|J`~dTl+-Mgb*^Q$p@t?Bd9S>iCxX!yk6Cy|wf&dD2yx&4V*(l$bNTu#CMV
znk?_PF<Dx%O=p@<bKRAELl6ClK*lDz4^+{c`LGSc`7ZYXQra$MbhWjrn92TEd^1Fw
zgT_vt@C!XBOI&QX(UBzk-{z-<B#nhTUYbQJMw)hNiMy_57jMOnRZlZ%LA}vkGu?;C
zfKLlE(1x<mf=c~tz?ZnsQ8Jk)((0MpJ>1!1%#Wq@`jlY?9adw$b8}%GfXH|X+6N}{
zQx|kOLFB6+Ty40>y@fR-xXnMHl5-TJ?brS{P4B7V1A5M=;=O*_F{+sO27~5f6=&HA
z+7vfM8Anliwa{-=<z)YU8Kvs1$ShTM3*9da^x@@vmX@cdO=UC{QB~V{2wE~%O?w6j
zBY}cZRu7R1iJFczTaIi}=VF-`<hBWPi{o{x2+~z+3M(*8O71@L;Hy1w3BHOYzu`?<
z17qAu&SzX3jNSMo|FhrlS#$qo-wH{Ogu11!a5drWh623XLS<L6lDG%IBPO=s7Ai1%
zQ(3BTvl4D60mdTJx@GvAkj0jN20N3ktom_Myy9M@?h!ta4MU9yrxh;o(0)Q_!KY{+
z-<jY;Q=AwfE~kgj87}vLF+&h+3$X4pl)wFRkVe~oG7|~&O++hs3!!0r?tQ>0WwlAU
z*Fd;M>Iur~f%l97tKOF8mXsO;o|!U3DQQH`3+@|-G~40*SzLDhgH>uE+hm5@QNJ4V
zPv9F-Zm+Rs)n6pBq4*W6Zmr2^X<>!XlzV6bKAH+2&1XTQMnFjSTH{pH8Vpj^HJ{FN
zAHA8TI8{wC)HH>krjGU6;dcOZ7ax1YneGTRJ0$~2W<BMtM$r@~GkTq({pyQvzj@HB
zzO8bXJa~<&A2WGi0OFT>!Qk%d1D1(RULj2N_fdTjHg>9J;1=;tK5R96lb6A}H>t9F
zQ=F)+0K9#+$1t_W0Wi1^@`=|T*UGjEgeI+nG_O5=W^ktIf@25nXVmIc9Kg(?zC}?L
zO@T6_`)9O=3`0ZiPF;(aTuMEFW`%?#CI4<SG1_SE|7rLFi2Qx;qJL3V-f+uC`>%po
zzEK=OT=dfDWtSi|^)M}~PNo8xd{paE_^5)mU-6R-trI^V1T7tIHuUOaKqrrazV#T;
zHyj21{bN9X_bBM^9RqsKQP2yI0j+b2*|ogv7|`b(1^v&*fd2YX&=ZaUtxnCD=-Lbe
zxwlYaj{5?x0o$G?CIjF(%<nY)7=XVDU>h+R0QbwXHw|!y0PZ9v1K<xjLY@Y=P5`Tk
z$p9$G%;K8@=*lj`piF>b0UXE%*q#Y6OaSY%3)!3ru>VE??#l-FMP@CY5<n`u7C+7e
zxK9AJ#I$MkF6PRN8gv*>-eZ3r+s^ir`kIdFwwcYj%ps*A{y~G}8PeKzYgG2d$LqYE
zzD#5rj)6?0vF&F)$g;U&UC~h``^hnoEhnuVslBP~H$0i*^)aK>Y1Nob|Cb*FX_<$r
z?N?%Npx%;3-Fu#>`~6XPCCYtJGggw;UM`HgdoX5m+PapbF#h@&7=P&DX?ufk{q;;g
z-jZt}t-J0=yt?gmhA7&SUVFD4@xujtkTC?L57HRfL9%;sJ=CM-%mDaTA0SBRepLWB
zWdn#;2}6T0oFss6X99RKWUCSns7QTD7j<6|jtoAZEh~$;`#e5NR>q3Q<m?Lhzk$5%
zO`f{T?&Um;kKG$g&b?>#I+^Nz{A@k3>E6PSFc*Tk{d>LglvLvm^7G7^IgcfS_gzh|
z7HfoYZLQi*Q>+nt+g5nZxmvZzn4SN-dEBWyrYR42$^(?1C2O~SFVjDoq)(g{V4W7j
zw08Cvefa_j<-u1(dL`hPoErIxgchua%py?oBbI)BNEp8XTUqfNiKx&Nxw&dKEz@S+
z9?d}}SIx0(poW}OS${lFGKGA>ZKw9@5ib1K=XmWYO+JSt3Lkt&h4~{1=cPuQ!mws|
z(S2NF8yN+d9*L%U65FcL<k#+(T4J4a5$-qf$AB!N%66~lvV@Vu=$_%uNlq5NVJVV(
zxM80F-K1juK-Pj9Ia|oW-MJk4eV3(e2EhDGfcph-5i#v!4Y%tps?{0cDKq;iN0)Cr
zpSCCb@3UGUacNf3&pP_sFLsw~*_XQqd?BS$tKLjA_oQ<-sAR%{eqtqw0{S+guiw|g
zZ#DHc!mfLw0O<65ztdNelJ4~Uj85NIlhx_3^GX*9kFT)lrq|y3n-6<8$|eX9qja}r
z83z?@>va5=E1+09wzvVE9j77jkhXDH)5;6EK-qaGdDi-V=$A${xDVezQId;;>xX#n
z5AdE`oWDNoz5b5Z^$jd7ACw&IzQC)q$wZ(aX&^tkghK_N>fTRa#p@mrPHw2gxTa$i
z`ZK^X(!XK2K-((-;Xk>1e44A28Jma8ocGEiyJC$;`lENbp;}I7f<5Eekig;&ofujU
zUMX2L{;2}z&br(y{WvpQC(lyUH~lE}4PYp8y4;C=+zR4iwHre=UdvE6F8g<i&1(BO
z6mIcIt1*K@!J8N@J0MhcwFh#mD`NO75{`ZbN8)_G$8O4tdy7h8+=JC{Pu#O!+~@J>
z)XvqJO&#EGKS>bn7Qeyv<M<uco=ZA@H+qIFO@@BVw9mAu<-TQl>Y98{HGo2(nS;qH
zf9kR@`sGPsBE~3!Wt_W<-inD?#73}}XRMVcmBf-$!Z=T2w<_~K+!ICNzuKyUev?YP
ztB(gSk2zC{d@KBD^Q4f_%4#XX1>bmK$>6WG<9-;ZKE+O+)tvVe1BE?S@rFfgKzN6=
zSPC_*KO16tFwW2YC<t=aE#}G1z20csp%-sfcw(s-p>HCxnHROO+A~|9`C;SX<EEFd
zulZ%Va3K}$eid#W`Fri3nI2L4pnFN5*k}5elzxNKPwP(qxaDc4<LLEf{pvmv?SEvp
zch+tP^wFc=)wt%Ia9l?1oqQaJ{FB$=O6+i11j^baBk~wF+`1uL%Ur2ji#J{U&t^L@
z!F=amWc@dN%OQu$=^#d1S@HgaNCfM0*;zlJZtJ!LKLyv2G3OGvp>&}WxY~Dae*2PG
zs+tiofmv&V9br2OkM#%1t8MXTrF<?vkQk+tm0wmLdu8;CmEZmu8S0~L{j#6iiNI*M
ze8F}Opp~{GoovWr1HUg8*`Gk0#!cFX+%^i)o-0>=jl25csEqk~BqOw&1pRY@ZWLtV
zHyrI}!+BryLe5Y`r^QoWC@%@+aQpC=<`YhhM|RO7jYo3ftoiQT@6iYp=#;d+ovXVk
zTPZlYhTuD}i7Rj0t9Kh}za)Joksj;j=kGKza=;`A9^s-);~D?*dCvUfwI=(ohGXaY
zmku+ZGvf@9xOCQEPEOHy{w8>->0ia^pC3Q$$5HQngda{XG4HGX-GA@!%h0jjgUe~~
zLqDjwn1>SnE}MNuC_eY<xeG&Im@_En3-dzr7FJe;W?eNW#Pj9fOUzw7DCf%Rg^AGQ
z>MN(sUo>|X33EatL-ARO#Ddv#LyH$&JJ*W=zDysV`LVt@SK#9p&tEVv5h^2}vp-Ts
z)r&(_a~FkX&sw<fzwvc<F+X;`Wu~}ORoVqteLHk{WnzA)YSy9!i3M{P(^{dN(3}N}
z=FUzmsJsfwG?k*s7KRdwX3<vjW-VBljyBI>Q}4NRMxE(DeECa@=FW?NX5ri=p}9+{
zK;^%rv0d(ANq>1(?=Ss+8DF`ua#7i=#G(aDLZ_!YD>L${MORYRdGkVF59OS3#+eHj
zTz=-HMGNL!ws^snbIzEbb9Lo{IUz-yefA_;`pW9W-0;;(3!OGHG`sSu#feb&>w>Ei
zk<hoi*RT5@T|K_?$|@=`@sgZ{^B13S*|$Lz`uf*Hi_$OYOc#1|<Yn%^Tt)`-Di>Wj
z>+*$jv&)i^5cW#p*LPgy<=+`~PEX-7qQ6Wu<r4sv_3E@>Dx@=O(YQX}yk_&$x+^j=
zbk;c`{u>zr)0h7zFnLWgd;Z+nSIk{>dF8A{-7S<EmlF<$t7+n|j=n4rIqkB_IkOV8
zzEyNd<g}tC6C*Tf(J@={@>z@3gkC+<jpEn(BB-nKztILGL!)Vjv(NU?g*|loJ(t`<
zr;8Y9qn;-EdorSt&d6*IXlS8FPG34_yw`N+s_DXu7bNCJj@cGfbLU=BRe41(l2Ytc
zz6~!{nY*bw-BzlJp{vCS#=_NJlhGX>QDxWj?;$Fa`^O<qQ=Zx0C>sagot1zzJrcbn
zM_jpbo~csg^vP!2(r|vP3LVMus&e+(mz7;MXYRaNu-?mNRaMQsYEI7FC1O{JxpU6Y
z2jR%qziu$3+g2TYnGcgaI<K6&=-aRfza29hEb%=s&$@DMNQq~kz1WPt%Xp?e|7TnB
zl1$e1{~o)zI(AiJ?ziVIx}?t`nNIx}gDd?>-w2P%=d#E(jmPk~k2Sip&>SnntSqwn
z=^sTn*^DRaEsL*v9L?}!f08;j74_zLBmHS5g9m0J_`g%VtbXXO9!x>p`u|FEXO&>V
z2RFB-(f=z2%9_cuN%Q}g2Ffl^)?0UZ#KhBm=FwStD)Qd4t8uKitN$6%bK*a8K=)+u
zce9_gy!ekSZdM*2sl-`bl-<_d9p?Ae-@!9KMlK(z)LA3%7~{|*v%jO@A1|Z-M%lAC
z%Ku=CXBYgKxqPHjXJI%-i;MI99qs*bviV5m&ZfMqx9-;NVMu>R$+J)|$a?D|qmNYf
z?8f+Dl>eh`e(Y>6$?-*LU(1piw65`0GoO;$g?}Fqme2L!JgA(>qs8~fq(zHfdZ8RY
z{mY}q&sv1CX2Gnh;Br@1U$tQNEEab_x4b;dV5;%+_wbD>5<*X*<DpE$tx1>D&wtIU
z-uE*z;-wK=?Z5YvGK_y)((m5uQNK0!`1QKM3xj@G^#8T@?cs40SH7(WAYQTn1HwE!
zfD;7RLL<rY2%Kc}7~8Ui88iwJ&(>&WBn?O!F*B0okg(bqjM*4af&&f<M4RO!-h|xs
zZdl?CWY>Gc1}7olu!J?6O;89){Dr+iBs`Q$mcGAVJzCv8J+@)PKlgKe({s8`ojP^u
zRMn|d)h`933eHR_I>kZCTOXvnU8!X2P&{#iT~~-6)$%F+ww}Eb;2NC>n^gG<1{ADQ
z&|m+-2A1zESF_jU&-j!-c}s1}$jC=N5)|IPnI1{!YZD1W?p!$6M`A(A60qHZ|8}_`
zi($W|21!c;_&eWNwrqGbjsF2q0AT57;K++jxtzUL3pzbdGK<Euxm0I-V`eO0lNgL=
z0~cH%7%mMINPyLEA&vznQ+;Ct%W~;~k@!$<KAum4snHf68yL)Eb;(q*jTi_Fr-lRZ
zWHOt|<pLiX8k#u92<VW=B)MaFjA4#uGYJr6vbn31*HTMZ&e~m9?g%+C=168FMg2)d
z*^S}s@E2d~i>~pjmSd00H`B!z$FLmk&UB>`<EyfnVXSPiiKUh2iiq}kul!n51L+Z$
zWUNW9D|O>oDwj98FkTvP72}G$uQE{rw&l3n;n>b7D3;evN=`4kWJ#G}Byhf_?_FyY
z=ettxS^-=U+6OpVWf(pE?^<Ep%05Ae^ud95tzCc>=UUb*V9oiKRfEf`6Bk)lFW`<k
z%i0b&S8rLn0YgoeHAnmk%c?s9beCAxAmHFi%bEt<_fgB*2Z;AYT2*-ER!uAT0amnG
zRuAE7%bEflTxVH(0IM&<8&Pl{;vBAChX5<Cw5(CU!K*+AxT_cC0#3wnV>jbz%L>4L
z?YZ8v!hnNAmNh|`wXB_hH5*VqVD$v*L;R;LtKw+L`x)c|+;bbs0o?Z&mK6n@y94qA
z_S_Bm0k?m_vi1UQxd-I|?)oC~9Rs@iAwOW~uOJuT{)Zqx;PxGmA9o(q;N>$Rz%_ph
z`2i<(TGkH2zXw0S?T<kY+?+Xx_Y#Ex_u(4)1mKQuL2ke;{|tVBYn}u@+|*iwi-pyI
zA;3Do9>6t#6M(&d6;GpG0BfFsegH#&`vBJf&I0xTR-oMn0i%F>_FC2!_5;9a!2N)`
z0DFF9Su=nWfOCL*o=15L5TAiw0QbBMdcc}rpd7#u;6A`LfU|%-fEDOhTL1%q6?k1}
z2z;sm!-$9Q<l+{<HF&YkEMN~{#R-rLFaS6K7y{e^7zNx8I0(24a1wAI;C8?{z@2~<
zzlNRws{v;KYXIi}Lx5F$L<6t}um^A#%G<IZ?S}Xq;4~lu?i&}O<da=S$wXDj>Bk*a
zF;#L@74c`_s7t(ST_8!szdyZUEK&4(uajfVsH|F5S-p0_u^TETjgKw5<bsAJ=a2z;
zGL(HX+9^srs%!?XM42qtFdC6YUOQ0k-H+q$hHX!^7%i1mt(DcSm4VjEn%2s?mdeoT
z%GujXHyyS0$UBa>v;3~I&85tD4UVeCmbDOLsr-%a`Q>Y`j5>44^0p!0+!D)bXTHGq
z{qyy@^RXUJv%Cu}>%tP`>y-IcSMEM<6C4b!r!tpb=k-*v)Uwvn5or~5YpD!SwDLR4
z_{~oQ)`vGb)zw;7J#10=jv{hRmAC|BKHm8n3R>0;lxx;EUrS}!l}(g)Kj=qc+kU|E
zs(#?Fhv~xd_^$8@*uvLPcKM_yUrS|OI|@5;t7<6FFi_6pP}SYYHvm{Z?a3$GxxLb-
ze*Bi{Tu9}6<IwW4J+?v4I@suEC;<8t+i59I9|3+2_!kx4?rV^Jb>+?yr|^{fKS93(
zHvS#x$)|6&S5CThM7GTLN0c`jwyZA98Rcu9K4kqiqMYsRmh}+ZC0`_;kDhKpc~xsr
zFU%D_a-yrch7iRRa<ZIVn2Sn~zkK@8%Hw!=1N2ju!^S{A?AI-o)#%ro%I*Z7dR_>=
zd%N)THpZFi_i{`_&(!f9^W9(AfOgQ&_E^?Gf_$F7rLu<fM{E_;)N42Nxdn4&1LnB$
z-9_X@Gdj$LutTip6z0+eKJt?9Bj7vPXIXrxw|wOf{oBja*38H6Mek2o)|;67%Lj|(
z^VpLm$k&K`JJXi+Z9u187<Jf%g-E&v>1#%CgB#9uAU+R#AY)mpQLeEXBL{%^t-$x-
zg>@g5_-T!Q2>8j3m|GP+nn}<<0sNMmEvpjx(c^@Co+tgEW3F=G8;M8$Ew@<K<$&dT
ziqv=1rGC*LmVkce?UuEi1xq_<_Xj=hbe~}Pe5dN}&tmSxxehG98~thfR?E5oa+Oy-
z<KIr6@|kZd@=e`^xtaNT4n5zKEbsG{wF3GmpDvQmr(gUL`KIAZeTC)iKlFT!7{Ar`
zS=Qe$-^yqG>*qKAS-%G?>oe4^-yFKzJtnAA>g6GnxBpA%7p&hcMauK3-y0}z?_XQi
z+3aA2<rU01EN=;pr6_ODLzeXiS3ZpMP4jaG<ywjHJ^iqfYoD)N=r(1pjwad}{I`77
zvVQ2|Z~GFyaxvc{C@=5`d|2cwU%J;{?>_#-A5mV<*DdS&;Nhc(0w08A3vo1}yjkS?
zG4!qj<>&-{7kuZ(i9bZWlHV=J7yhPYz2M+S{QbaJJYiXI>Fq%Lqre-$E7wq@5&yKJ
z=eZ7zIC&Zjsz84m`d9hjq3MV1eFpLk{*z^$=IWR3_GVe_pzGajS?Kn5ApRQQdwlSD
z;G;hHt-!DG!9N82)F-eeBzNeQfc&2Te$S6#|A<h5^?e@rD)8aw)!$1y9EY`0=zm)l
zhKC(k&Jy73Jb3D73D!t8zz^cQd<*m{=hAk}dSWgm6XwSzgfR*_xcovM`ZJU>$+Z^n
z>uE3c`nC(^{=#u`Kk`leEBp+=@+CiV_jln7&6GNQAr>q20{RN9nL5}Hwmv81b^B10
zBA>S9%twEL^$h$Q{2?Nse_`{fTLjjx9qGMD=djg*_-lag@xkYTkNV)Z0>1|Mi^!zC
zGC;ucSx@rcKm^Ly{>o;^K!4>)r0+%gA`0M{pS=DG<$8nsKK`zi#kuZJ<nzXH@E5U`
zETtWJ-l-qAFT-rq=St9Te*yBc-1jiv^T-!PzxC>CEAVT4@DBkW_Q5{^e8>m?Jn(fs
z`2PaF2KYs+fMXn$W9?0PFa8X!(S7iZ!28M5sp!4+N-F+d{HK7YJkIgU_Spt}0C=xG
zc?9@s;JtjF1iq@<Z6DgfryOhj8njP}^+mt9t29zseP>yuGH^$Eq_Sq~5o;>zHXpgB
zGPLQaODn@el_88Bw6ySTu!_Z6wnFspPVk9>kLvQ|CW2M=;w_VtMwztRPiZtyPQ`KO
zxA6Bf9P4q6V~s2pioJj$Y(+I7xUrcj|BOft;62RqYZZBrk*}as_*2HsB>N^J%=0i)
z{sU)k0!A2l7D`-v8BQ2}d#z!d3@B2{u7i~kHWQeeMxZLdAAOUNjA5J!cpDi^Bl01@
z1uO@9zM`DPjO1@L3^vhq%(EX6N+9yQiO8EUwj%uICc`)#XP1&&-KQkk>s%9jY!bD^
z+$2(Fwiw1aKt4=vr`_SmbIRxCaF4LiixD{sa4jRxeE}U2y<ss~pZi70dMhIf?}JGL
ze2wLtfe6Z#VLKue^lL2VOStcUA*^tgk<~j4<0`25caI_R#NSBujxy!mufkv<<qf7B
z^N1s*2a$_`6b!EZhD`Yb3r(qrVDKM^<bXVnQs4HdB)gN5&#8#WvjhCurlRS;j)-{m
zKM=+cXqh6qi&b`h*?Gi1!#@U|^g<COoT6&W4dYjkRs@^&S%olo@)^(uNqaeHe?L$A
z0=w;cm_gCMzsxL)aV)^`6-HWcP<gK~vK|L?XS{=><ku?ldq!TzAGIQa<ZhLH2drBz
z2a4>MaSUIBBaK6Zf5rbvIJn3*_TgB^6NYm^%esp^R@wbV*&?XZH~~kD%%8*|qysdp
zvfr={gFOFV{A1t=>yTop-^MY5Lj-R9unxPK@(YwmnXYE!WBA(uD8gfiPy|8tI3jx7
zpghX64QZ$0AaCRAI9hqaV3qBKMh9^Ae{eLhPTeR|-5+fjcOwGxYlNBdB}6s?ick#<
z%RGlEXX0<I#5;Mu3Y6g4g`6!qojln}g6A|M6FTKUJXN1M6Z9hRgsKUZ<L^8~E~Hu*
z>^^n0rsWA~|AHhZA@Vs2%^>yDjI#{T`<P(|{;p&igH`r6nrn>1ll=mg6$1nZbBM4T
z|3sd~KYlHNWEV2R!4U=G`lAq8fIk{N5z3KF)d_A#BT}srjI&jWU?A$T3rSQKN51g`
z9IS`Q$d7UC!6AZGc5T_!JiArK<q5bMzhOyt;Aqz=Kf=ivG_i)7TZO;-$c|wKv?LdK
zeu*<jKl%>GM*@jtx{6e{f(N_pv5ag(gqjm!B``z^GJeOy%ZI~?d=?{@u?$KgpI6u@
zEAjUtpa>@;nS2D<B1F7=l1L;Ue*M>IQ+)m}`E)t?uxpTyAmj7RUOrl$rBn$AH061n
zOu0Ea3n;=lNG2aab{--QJ{(7bNTWFSGv_FdQz!}peLnJxGvy12{5P=-p9h9KMV@;R
zaqtv6A4Vbx9%R;P{ILgI$H+PO6V4#cMFZGmUqp(74`pK&nSMO0*rnKTE7sVGKT7>W
zM%a!SKoQDReMO#%!uqZR6?2|Kp0DBfE{<o&h7YW>+rG?5H|k4^vX_y&5n+mOlco`w
z(7nbgdu$S&tdS38kSpJrU_6Cm42K9Sfngn1F{K567fHMx$JZiyG!BkKqaVlFJYi7w
zou^Cr$G{UxatR|G&{V7lZTLSCM=xU^$3e?4g0z=BjY@<(;VgVJ%Q^}|C3Oln$T>fW
zl)Dh2>|X>FTgBhv34>Mk@@G)*mr55QLKz}37J@$`QAQrX!PKi6`6nEw;OJ%KZ5*fK
zh)`pdU`bI|Gg6Pp$$%X!=W_g!L5k|3Tuix)WRD`kt`a7<ClR45`!z;r4asdMBgdk4
zOu3JF79+wuUuJ}6oXTP=8`t2E;taE#oJ#pB$-aSzUCwt9A-6fQ{&z%J&VMs<I{FAx
ze#bl)A;LUwFcL+C<s<=(A^fqNF_yDQrTmd(-%$~+?u{QK@+{dhs6KE8PFccdQ9ZF9
zV%c{v@>Lv+>|x|zaIklXU7creaucx(a?WVLSqR5ZiMj#Dy*xP_4r}1B1`ccBum%om
z;IIY`Yv8a34r}252Mtu65f;^$S|s641?x_i@mU205&pEV?1g)|Qgo#6_r#Cj5($BO
zP7HIWgaxE0&hY#LEdrh<{^6?sG%n6gSMh*?T<b9GSLt;68Ms1Vc=KILta7)Ph8Yzr
zGs2#u!vqF59~tIK!vyxyaD__XE)PSs1Y$X*P6fcaLxveCAC@>GRH^!L0m7iyCS1!f
z>^~tapvKQt$+(t#R;0>RE>Cs&cwv#KPqm`Q`-wzcsn?iQ@u>xppJMZYgE^&tZ$V9`
zN;y_3c!h%f3XUtdS;2c0d|1K96#TA&FDUpc1>aKe$ZA>6DGGjA!6pS)DR_l~{R)mN
zxLLt_6nt30#}xdof-fleD+S+D@W|6t`3inm!6pS)DR_l~{R)mNxLLt_6nt30#}xdo
zf-fleD+S+D@W@4~d<8$OV3UHY6ud&geg(%B+^nGC39H-ME)LXuyl-qIKNe`L4b|2!
z3pS65Xz+ZAFjQ({D2|swjHj|WzW1S_c6n{xvPMZ-e{()NmdkIf57gx21G$U3BAt<z
zSj22tUUyM{8u!beAE@Co`pE!tc6MHIsmy-9Vbta}4(H>2fcdP1gZd12ey6gdM(s!@
zpQ>$X?N|oZDmE}OR@*n09!f65GXk}Y4aReWMs0HA2>3~u&&m`fuq$FBEt?vOGlM!C
z9m*TEe00I6&Ev6eqZV?fktW_mVAQ4t&HgMN?KTIK$g5*unn>U&@kDAAce3+E4w8o%
z&t~HrB@cameF7O#UVJ#6K)wu$z+bH>mv1{TYWs3Iqc(w?<5MGfPcyym=;_Ew+pX<Z
zs5Fcic03e9+f#wWA8DQ8TtsNYbpGzrFk)(x2r$5iBWOFNQ9+zI^XE&$A}OyJE)lW$
zKX3kaB)tWjr_0}78WzOc72+BZo45PgzACi1ctjho^UsxrMbex?^s0G)fOXgTs|A%|
zTmi^AK<C%<N)LCX5cE7EV}<&|#ZQAaT<6zzduoBqu|;K6(}`V<rsswf=L8+s^FlSZ
z(-3CFzntf&J1xIL8=u4p@jAbrhvrm%JrC*fb$(s{TRi#oytbd0#Srv7$o!n^z4`A1
zj(WA{;Zwqh(H{^%>}AR``A=D$;dVrv`2%V`2&jp6)SjSaE>$hD+epkdPk!z9Of8c1
zt((YwKIr@!b01Ee<!is?dEV@apt^xL=bu;pY2Zky^*^Kho*9+@Eu{yLh~O=M2e8ik
z`|uhshJD=bMOfmgzqkCyK;z7B?2`G7T{8bR)ei+T{OR`7@!dFa=I3qp4BMWP`Fphk
zC9s9{e-8Lc99m56=k58a%pXwsH1L-HGbB(%onQNb`&E7~y*K_F<Yh@Z|E%(3W)Cs{
zYskyKtMh9=XinwV4XgXNE?>9foTvP`m*T*n|Gy;bujvI6e>$!qFE2UeH<TZ0%*gy=
zi(6;*$$uosocZ_sJT9pB{9NYOa%vX3Tpd3SWW?$E&#3(SRQ`Z=@KuI@&ZyuC$j_4M
zJb0b|&pdc3mCVB=ni*ZTlk^iWOUgw$gG%&7McI1nIrE2Jm6S7T0df`+KY>HnU)w*9
zZ@==Z`NnpHTQ3m6{_~b^C-|I{LAp?BxFDqrol4KM&crfEqUTRHzTDVe8g`<@c^Oti
zbH4Z^?e&o}QyFGzJwLhe6^5R7-1wsnJ)gMo$Jpy1H~v_A9q7g%XK1_b#vgBJd+o+o
z+UrL*eu1Iws~cYhd)eYdiSsh{E^QATsb#o4rR|&p4eg&~uT$Olh4%W?jX&AY_Qs7r
z#n5)djXxE3A?`$p^D?du^|*JWmf>CyJ)RwC$hpYS<J66(H|P%RujR(;mK!T=HZRk8
z^$-w*CmMWg${Ff49*@~LLqy^E7>hIXC_EoiaR&N3ESHayIK#~%a-zY<uAE^?;rTd+
zGyJu}^KlGk;Cx8_eB8nrz9%9l8hkv<8GfShd|bg9eyi|&jKCR6N*Fs4&l0-v3l*M^
z0XPG1cp*J+^LK`e6`r^8JHw?4&)e;tp;zH~ySp>w6rQ)GJHuv$=i_3|uwCJKo3=B2
zP2u^Nm@|A^;rSStGd!#Ce9X%k<g|hI_27RG`chn{)#E*&GA)3aEkS>N<d<=Q5o~dG
z7Vv)k_CjHnPc(Qtv@>Y=d7HB{XgPTsvomNpd7H8`d<6U{r`|Wqs>CamoP5<a14gU}
z;|kCBU^8gEuGembFA=Hmt3dBpFZxjIC%i4#86H>i=>5BZO4R)4Jp6fYjvxQmgq(%;
zinqzJzLTDP!4ohk3O#u6^hb&J?jv3dJlk=HryV;4zR-TsuO}b9U#aW-DM4Rohl|zg
zF9rPx#+33~H6K1BQ-a+-Kl|*bpMMbag?3onc!POC`=y%ye}bMjA!vVB<6jee3hn<K
z@aLeuGw*~2DWCmg7yu0T*!e8*CrT36{{z62Pq<$86Y(q;DER#OiNN>%rNo!0T+6U<
z6Z8utUOaP$v?+y;4oEzo@nLuhc)xOgBJjr>dp-P*fMe`Oe_av$81R&5a)l}vd>LLT
zLO%!m*<2y~y`q=o#!1+8AfLUh63=JO82HQ*>GgX&wEz1A@YK&<wLU|)7vX8er|JsH
zhtFg&Tnk6huU>h9Ki-IX#>Gv*v)tallYDqP5yOK5KR;hj3M}qMDZ;-L4yj+cX93Up
zMo*Fa`HUmOJ%S$gO0|nx_8LD|^x^X)y?B-o^ylGLCCaJaqoU<p3w(eKJ>kkC_+jAv
z<oq+>Dd*IBS#DVA^Ej*{&xU>IeK{^lh^lgUldvD3z9M)&+u=ulR}uVffyY&e`=vb6
z&SK9Ac*?o^N{OevVE8oFJAVA{75GBG^Z|iC-l+48i|2~)c^i0^J9E4&x2HswI|LE^
z%Ka?xY_GshNgtBr#<ZY6-Uz6ADK^Hd3SYNb(yMJ8V<9Te{vTDZT;?-647I?Ees!~?
z=QCUk{R+QxMB=|M(Z(MH9~E(clQ2ZdCw#V~=QFPiUBs7SzIZDv;!wK?X+^K!Tcgx(
zJfQG<e;pQdyseXAPV;{~EaH4toFRb$PQ68|!;IR2?^p2C=S#o`Y^wQ_cYvp!_4|Mj
z72zx>P_);-hee$46kxa<_!B^{*CWb&8oj_%&sFzHJ(QM6`Y#B)Ja_&-3cRozs(ld^
z;d@2se*!$^nfzr~(1~{k;A}Z68xR=x-(|p8`OGiZ08crmJa*#|fp_y@Y@ec!diwbh
zxDQYCpI5>n(}k+O4GM4kOyc=Y1%?rY*Y7FP?fxL}lxNn%|KAk-Y?;)jc!nK(>KFRA
zZ&cuqH%2}6x*d4RIlWTy5zjP(Pua=RF719fPVtN$@eh3hPkF)~dHMuB`iZjhdOUs_
zc=8`zEBVhVJ})XhGafz_*cT+d@pf2HzNqL&fM>bWw@W^JHjUxSihlG4iT`1V#J{cZ
z`h8D-Q22APf69KP--Ctfi*O0>Xv%_cHSi~to@VS*`#+jspa}hKq=!8Z{30$$t5mrU
z7NP%#BKYUX2j%Ydl>3sPhd-m{>y#4cHHEL+-Xbs;DSRnhH|lfhEQw#B<jDh1J?Qs#
zAu7VXIH8{T0TqS~;vfExqSx=iLUAIztmq9zuh<(Ur}^u%9(d8dn<RhT&#%$+>bjp|
zZ``Elt334IQ242nC7*v(8OkxyQ~sT|O1v6g#$tijT<77Zir_y7Jmn0$6&A$Ls4QPm
z_!(7Sn0*m;D|~oZmOEM^@yBE9fqZ6Ee||?*cpP}qpYN6QU5d|rz_VTSd(J`%|F?=y
zotoD!Qv9C=p8aj7T3={;`>NotGCRN#xbP*)_2}np;H$vL+s~H?dItk_x(0aqE#X(<
z0@I`LpDsfG0PsSeNm<{smHbaBd~`M}IJ~C#eCSO7{<8x3fTD4N&LZ?l;Hl4m$1dF_
z=rJF8`t1(jDQDdpDW`b%0t#A-jZ*5ZO6gzs+ik#;es-y(7w?t;{X#g9qP`m?UO!{L
z4tVNq+M~BW2cG<+%0J<=$qc&`AN}4u@k}`IzXP7_W&AcS==9&)ioc=y!=y?G1|<LR
zCMl12MjL7Cfv221e;-B+>n`O#d@~^NdoP#te3t`5``P~OawYKO)0>y{;#qvq?<+zN
zVnIKv^uMS?us1#mJo_`haU_EB=S|@K?DMTf@Q*6~VMEHpcbPCOI#=?aRQv-1i-!z=
z_v7COJnJ?2UtvKh*S&ZsPvNUxmG~Q!{MjY`e5QaWAN@W=J?}iN@J6qcQ?EC#{E+ld
zyyLMKc=DNEEa@Lm<-Pzs<>$+yoZ)RnuiwY0>$u{>s$Ln%N54}c4?O!}z%wrH6nH0R
zdHz?5-s{)Bp!oN$ll=Ma5r)KhvfRDLN_<JF#P0xJ%O)}Me|Hi5&jcQFdi;lDYyA0~
zSp?ry1V2~=e@hYky}+|wc79COE1>Fo#`zL&sC5LEpCYUWo_gM`?C_w%PYAq<IKb_S
z&*-(1zj$W}(oelW_UCEUpDR?J3xOy9@EKu&x%+XMP$%$(*CUz)US)QGwZOAphO%$3
z>NLKk(HxGaN5m@|+uB@rgVY(gzs7~ZJ2Jf(AEM=jR857bPywpG5aWYtETk$du%Upc
z!3R|+Rl|zHjD><WG#5}c7V4p~Q253|NgE4GEF@Z9h$@u5kWr!J&4#(U^Rm{KPSebd
z^_d-}s{@!<#qymY(@bZ?twbVCzYo>h@QuElG_`q^d?jpWut{*yU$4N+Q6t8D;tE&d
z5I!>#In?)y`qhJXAL+O1+AktaWb?WF7(O1AFidl8TbJ3{5$iThyahOmcYEe=mzJ3y
zHWNd5Z)wi(L78SUV-5^u`r<=o60hscnenj+N-5sanXGMUUQrLuK13#O!jd;onYbw`
zyU~z0VI{|ghc_aNBPL!u>moJHRb4HYM$E|ib_gpaSLdn)FI%-L7U?#-TUt9Kq@eC?
z7D+cEm&coj4YU2q^(~inwAn2X47FuOH=3ikH!fwWPY89akC`jfYt(TbG$lg5RGH_^
z_RF9no#}>%4&z<SsNRF$(QQT|u@*DUVDJVpQ;C2sR$phDcq49RgE^EM8ORTsscbfr
zjjl_C!>c<xTHAVh%!XR@FV%;dTX7Fwb|XGK<q-{H+EPg|9il}bT%(Y!YhA3#j5PL-
zjU*zmkcpqoy2u-ugUeTo_KuAW%a@QxqRBKf#qv$#;_EUUvU_{=R5veDEWbkbk%^9e
zhn`~bj@Sz5Cb**Dn=X-ITP_G7UkZ$GXQhU^)~{}irBca`SU%X0#b>`V!}uJ`IKJ!^
ziHvu3o5pAscN_LA?V0tp^|i}wRz<3~q7^s!rBm6+xU5$&ggYG53A7S;4d6c&AMQw?
zz#v~oY~o`x>3psO3XJ0~(iV(Bt2xSJc4TtqW%zQ;FkTs4(15YbP%e@#K#SL9JKD8t
z-TUfv%qAMX-t34qn!3RqZQW)Xc*}7vmG>&ATeV=pJEL{4D{O-P^pH;vlpO>;6PoZF
z99k%J`{l4r{A8BN&yAT$*q&^<Z;am|>qw%f77n1U?mG5WTUOWmz)ch11522R+?cR4
zc7q?P;P%k@pak#3OTfi9)jYb_hM+rkkk^Kz=2xJLEqC_JzOnw|-O}67khw{ln!b3_
zOvH2faW~OuCYPQt(MRG#cr~_1Qe}XaZy3Z};W0jkXcy5XLi(FJ{%wpt95Or5(lD5<
zCVqBH78$zr_3l?*x2dTp(lu|M(io@d$oy2>5Kmq&E1;U)$2`fNDn?Yd*&_tvC~p#7
z&~7W=w{W8%b6_|#B8@!ftF_~1mzrMNhT^##=GM`%e9m9Nhty>;#)Xlv2b}*z^gkJA
z-RLmy$qu^vl6h5fDC+57sfo1L_Lx*hHgEFlf(80m7t1eC@Jn1UvoNlgnj6x|)QFhd
zBW~8U!QjDqO0Zsh-Y}KyijUj?JDl!H!(Q6C-jmvGb5Ts~!R8Tu4)D+o+dv9mryRy7
z(30_u9zBS=*R>@sG7KQScarW`lXNJZ$(g)kzp(>85hg3Q=4=J5P+v%Ue?yxVdtJBb
z7<KD0wFet=&JU+?!o-;7P?YvsSU<7k^3)l|PTE7qXCw}yg8E=ozn#A=lgaY^_iX7%
z+98wI5(Jy+TC@!(FB|2|6Pds?l#4{83G@uk>9%qNo$PqgbQA=3N|En81Xnn}^(T5B
zx*Xg;;W*Vd>e&yzihJ@KjwFWQ+tk-#N=)E0fo9)OCUFB?X1sbo#e^oTRmA7+FdE=4
zf+o9BCl7ZHH-;je!^?35>*4hcH-wsb9;EX(4EKm$LY*0`!#6<V!zRZT8zV6hH{*C#
zqF{I`qi?rDeULis7)hokx-+uNQQ54TpPI3fz=I}`acIcI+F}^KYd&SG&f{*>H#k++
zjxP$3oXsNDCj6wJcB)xyQt<QZuyR6jPBv74pIaeXM6zxgP2)=)j;~w#ah`W-{Tv)s
zO1i-PM{jBt#wM0`56E_(0U65=<vRL(98S(^!3OC9WRzcuxv4v7Mldy@jCnsBYYvY=
z?&#)+6ZN95`_eY9^18*7F7s|x5&MEpqPl{o(c0>LW?HzP_!(x$WtaoeNah%x7qU4N
zDZ-bNVf#%co(AfJ16^P!)MRXT(FhinDeZiF{QvoxUw(^6Hn45Y53p{EL^$6%7j&vc
zsa1cSny!_ecJF*$hG4VXixD5%@KJ1ZJRMt(b)>K&xkh*uqrp0{c0y;!Y@~<6C5flu
zRm&C8d$T=rM7plG8>wJA3^sAqh0%`%SG!zqiPe%AIz^12Z<FfI;oqMQWH(h&{u*p-
z&1T|BI<=q3WQUTGG<0Rx<^ZcJ=XjQ0m1Es3m-P0kB~-X_j-{L6Drd4O&$N2*MQ0Gp
zSm(!DB5d5CYFts(LzMaez1F!VTp#1YumKB0Zd2$wIhKd*v`?6K(W5XSOvL-r<3X&<
zX~T0lF;!p!?9ZTgVN(JR@tMKe23y02(79*<^tA4=BEr0U5R+bFNUrG*)ODKep5yK`
z!Fu_%Z4(o4+`HI4#1)U<j3ah}Y{l6FO@H!;?uXB0igC|8m_BkChh{EyV`v~XlFDL2
z?O5N8701XJTa68-Tu*!=7^#_TNA&vnwb1o25Z9|&s@t{wV;j(>jF{hj%nut-nOpbg
zxF8t3C9%g7$51HTy)rS~=-oS1A_q4VL`BS1EZ>;GXPVqr3?FO4Cw}1#cqf2ZzS(vy
zysoX8%w^0$*epEF>DVIR=AhjGD=<0X`&LOuM_DeufBL>>f4W)o8ZYJtw+KJj-p?pv
zLgy#m`g*ZHoWtg!_->V|7ksa1GcX*QW<E-rY`2m36JYinF523>aSR`UPo&UalBt}%
z_vPMHYsTz?waIuK1Ig7OwZV|BLeFRpHaZ=-6l}3_F-OO;DRVrX&5y;g*z>U>UPkp!
zMv-})di{92?>02fJEh$j*MgvvqgXhrj>b`pFejleHi!`>Ohr)qtFM|}ihX+ORqmiD
zzYDgI3s=AW7hf;U<Q@W?v?$BbOI^9STieWix#7%s5wFFrU3<N5CB5^i<eyKltwUS=
zLCr$CK%q1YZmaO8F@goY6P#hWA#V4fdeK?opyIo2x@)VIKD)Hqp|n+@W(<!W;xW2F
z&Q-l*moUVA_EbEXlX=sP{E)h*SKH!9v#dG*-9BpG#ADs|*8TceW_kb6SZ=V8g=Z4w
z7Jn{%(0SCcGGKeUMjPcSXsrWpn>WeX>-}0Wzb1p@Ig}aDomczTruw3vc|&TWnSh4n
z5-*m{<|17M+uhm&IJmtNO=pgc_?Rg<M(el<tsMD-?R0$*&IKoT$BL^zLc@<?Zg-+0
zPLW(hbA!TVp!_FXxo&W{f^3-iD^gDP6+VeeoOv(L?FSyfVN~uQ2GI&LCLBU^yZ5x^
z0H2?iZoESvXqw~02Ok5{PZKkpy(`l(l2;d=x?;IP+wIUcC&+>?ZXU>;fdejNg=L@L
zSjWYlA{*;P7FI4>!B1eiHY}Fog3F&|Q=#~tf*cM7m&yFE(b%iqg4ND^e>vD_@^foN
z&rjz@ab;1vZrVaPZPy1e5V@_`aG>jmu?+nfHDlxQQdca~ZMqCB=3q|nxI%*M)?^2_
z0CSuQ>-odZW}V4#kd@aP;|YER)YJ|(UKcXp>4n&ERo8n=^re1WS`_T*oy&{8yaIz=
zu}s3x&U-Fw^6Drzfjx7A%LA*Mzjpn8Eq>u8H+9XLb&JW}225o!Glg%sZp5a?DCWXU
zHa?K@j0<?9g2d-aquSKv%o~IH-k+%II;SvWT+_~PdO0sHvyY<LX!<-GHw6=^-FSb(
zi(3VyIxcN>L9ww_m?y->YahR=iapYNV$jzqE;3Ud+IH!o7zJqv9QO|tyQ=YiFWRQW
zOL6SgM}23Em+9i!bPiX0p{R!Op}eOn2$`f{-d$W=%7Y(`f2WGY_{a3?)bcv)i$cc(
z6BE3h17+c|COYnT-Fs0p9ALR4+?v5|y#E^HJx`Gx*N?*p;bsL}C(`@HD2I%nObIuC
zWW%t_MA^he+zQ@@P`qo~`waG76#mnLeMM0!-?16?Czu{|b!|5ueOp|&;<^mqE!Ni-
z;5WjHmlt8|yG-mw-^WKbz(!;9GdDKOB@^z#YpAUgRw~~lHfvITyXptrfpzYJ_8Yky
zvp#cJrU6%NBZW8Sa8F*%3`wywN`uWk2611As|}UShBeJ?h>xPZ^nSLw%?Ini#Hix}
zV9@u@A8FJ4oxp--bMN=ro**1YyQ!NGyfMfds_Z>vUmweF#jkS+H{N#UqVbJG*r#<~
zW5bd{>B#9X6uW}@fn6O-7P&fxD;V>eQA$GIBx;BzHu}XHu%H6MB@>q$e5NNFE`7I_
K^X>v{@%=w@XkmT;

literal 0
HcmV?d00001

diff --git a/tests/Grid_nersc_io b/tests/Grid_nersc_io
new file mode 100755
index 0000000000000000000000000000000000000000..2ab855983b08f387e5ff7915f0292c972ce6a088
GIT binary patch
literal 97496
zcmeFadwf*Y)jvGB01?p>1R+{4(T<vk+62)if;9sZJOdMrfSOn}i6Iy?kVqyP6)`x8
zavTP!T8pjLJhqQTYYSc;3#cI>Kmv*eyderk5IsXc!aZCh@Atd+nVEAYnLPH_-}}$|
zc?UB4thLu(d+oK?Ui*6XJeutu9~%>63w~m4zp)YOydX=U<ToBig=w;E8MaewS+?_S
z=h*rqsvrJ>bXmqzFzhmt1;Yf;@)L)@c>HC=5@MLeAQYY>ARBU!pJ46`nP2#mWjx9-
zB}GyOLz^vG#1l(dTQ8UKt(O-Gd@x)@Ver|UD8pbnipO+aM!GH|T`;s6@(06U`N)oc
zQ;hg2u~~o!XECrj#dkO~vbW{#^i1XR(t%RWr2`#;G8jH<@C$~)bT=U#<@~vyU@<2d
z`9;%9xPByODV5FktFq~LT|MeoWm7LNn?AFu`ts_Gt1rKL)D@NGSBw<pCV$dif5Xip
zVNj1rfGG1&{Luy&|Inc9d#-ruqNg95J@oMpU*4UXZ2!YKjC0|S=?36$?$p7hwwP05
z6N=)qVzX*u7&j4rH{!2#?!Zki7Qeq{!Hn)!cgu%uKPF!HXMcXyzHPsKIyoi#GxyL%
zgT~oP`@Qn#b1IfS=(_hEB#HWIffCQ?CH7(n-5b6H!u5v#34-^Ae+<Uk8~#sG__RmC
z_k(hK<1;P_J?kU6mvWsS1wSbYpHrjYpNxXPH41(y^6gFjby4VdMyZ#7MA5^6DD?4Y
zpuOe$WEA>^QR@BMDD}7^3ZFNl@Odmsc^`{nKb2ANsZsD(MycQZQSu!YML%q3$@q)>
z%mJb|eeQ@N&#)-?xG4C~qsaMglzM*&j-xku9*)B2+$el@MWKH=3Vvf0J=`0G|B)#4
z??thz>7Y->U*u<h6uo^Mh0j}2<o_&+eU?R`|3wr#DUYK6bE341Z75K0^;I9GTu()j
z=Zz@&&X1yp52Dn|rBV3rk3!!Rg+4J#c^`_R=Ydh=ToHx8CkoyLJzQWr+m>}@kua4!
z!(fUa5AltotoXB#FY$#VCBDX>Z#MWOUnTL$2L5S-KFh!d`R@h&VDL%4+Dd<!k#D=f
zKiA;1B}UeJ&3Rb{Mg|kXpM2^K`4d6M&x`n@on)P#h41j^W<#EOL!KHVYOtZ_n&hl7
zF`qDQsq{=O9x;Mrm8)$91$WOVpIK1pDVpUeD6kbwxW8ah$=%Z{Jtecol@(Q1mQ>mb
zTvKuirk2by&{J~8m6gvdnNoCDS&5A6J*J?-Q#z}pXllW|B@Y%9<H-Z^+isX#ISM&c
zq^A{-X{BdY`GeV$CQdGL6=y@3;_9M;imF*91@}*%<*6zvV~UBBJy#Wb9;_$<N&1yY
z_Q3S1B{K`A&z$bbzF)AMKB}Um<X!<x_N4#1pa40SKTvQNGA}ACD=(H*<yTEBtEwyw
zl3i^kD=w<^kP69xFNatqMKh2^Wyvg$s~Fhy;?km71)f<&(>;}tBYP5XSKd`rIlVYY
z?V3_BC9NQP^3?^o6DMC;K+z|AMrIaFx(mrlN(^QNMKh-s%qqF7sH|vaamih|V3QFd
znOQ!y1X5R)OinA9G#Rxa^f9YM6i4vOM7_+IJ`=i>WvU#PPVH6Bm?0G5$(UYVDL9OB
zK|RiNLqVRap@pKdk`RLFawist6i+Qz{<^q)Mn%!A62^$!$|_yces1B@3Z|rs(u1c*
za4IXQhE7ncsPyThXbRc6#fDnag32$i@)VR$E0|R@^X?Lv?F~@LD5=@$(_}^TS_4MO
zSq}v+l={Ysl9@B0K$+jw)9;>Hj-nU_GOIFs(sZe!^bE-ytf$YsJ9-AfB(wX-Wy&o#
zz=}~t)kSwrzdxOJSIj0;KwCw1Pb+8HEfqX{CK|$wl9`?;^-c*<5u>I}p9y76EM~O|
z8@LONZ)QPpk*By6G*hPhOdW|5%LX?xtzdFN$&3omgD}4eH2CsaMR%95&D~vHT~JXn
ztFnA%QQ34v-=8ML{`K7@9?Wgf>X2E4`AglH+5j8|TTHMG{2JzamuvE<g2`7&vls<q
zX4_96iS}DoQgm+tn$uLHDUKW!w1(+3?}rR%wX^Q75^55L2=lqB5)J(B?8%-{(<;g<
zUFqos1vB8s&<+Yb(!wgQDw&0rlYK)bN;I>I)xsKq5zHtmE0~UUQ-mVBa*HQdW@k?x
zX~;9G@`0iXG-i0Y+os%tf?j1dj=QR+MK8Tj(x_q>aL=?*MU=Es*z&T<Y?|ZD2iR^S
z3~3|^cmIs(y%`hQMyQ25M;gXmc{Pj<<}Mro*!|oF6k<?*Wkr=I8%<uAS_vGK#nAq(
z_R>pn&_*hIEsdpVOfC7}o5tjzyF)JUtsy5;Rx<N$PiaBPtXbu=f=*=mRi2Wvvb%#0
zFr+r&7}2Wf7^arx-p>wkhU^RRba!!aK_#JdTfxmUA3!riXH^MXBw4W}g8@#jE|^gQ
zyMoqj&>gI&fPE#p&}l`}%WM_!ES_n$>6MigMa3mHv<*-BR2jhk2`$5NtEYQxs5j9j
zD@!~!aAZ)xa?C6*pvU%1FP}+#@r())R6HoTRF&D#Iz7{8l$1}KS_I_uyNiki@m)p5
z_oA0Vu1Gec{QeSK2vbn`;EcOqdB}l<G5Y)&Mbl^6rlEs^xD{11i#@jM$Biqv@`^Ou
z^={YLaRno<NFNn?9C^i6@;NULB%`iKztRF`TA-_J<MQ$xV_gO5S6Io?)6H}ol*Hm+
zobk;NgDHIEQykVy_|z{%L+T(shRKK}eJs|g!uU9>nh-B(NfnHfSaM51Ji}P5=7sqU
z6j+Iu-=yFZhFVzDiLsr7v1f2SA@|}3PqiWaFO7w-7|>-J@oAuc_IJ-=1US$(0`W{!
zb@udADDl-H`u^Z3;}F=2yZ!)S6KrD*dhJopYzNqGGT=G^_qP=o@Fz>|g;@P;GYq&$
zq>r~%8?Yz!X*|W+p1?OhUB8m!QwRmA!k^&ymu2L-Qt%TTw-V1eaPSixUlY%@sNg3!
zZYBPvAQEA4+)exh6FxY;Cf;qr2gldM=a}%paW3(bP59uroA?_{_@p#hUQ~(Db1*8|
zgr{``KPe`>v_ct^X2N4=9r|RL@XR~-$ui-o>)=N<;VDz_lWW2oRVAYGOn4Yh=u>FI
zo9nUEgdY^5w%IC7_|r`IY7_o+6TZfTS4{YMCOqvq_*rPe4-O&`E;8ZG^W1t9eu#;F
zu?f#L(%@&Q34dk~iEx<-Kg@)0G~tJv@U131*J6X8HWU7=AQItb6aH)yzTJfXg$b{l
z@WHivMt7R<=bGrdO!!<A-j*J=+w)BL1QR~lgikc#FEHVgO!!}#@X02;-Gomu;V&}b
z(@gk_P52BG{t^>D%Y;uc;Z+kp)r8MA;V(7e^Gx_jCVZg@Kf;7BHQ_Ha;VVq|%T4%d
z6aESlzQ%<Al?gx3gr|QGeioYW=|LpIMJ7DgB7>iL6Mj?>iEyzAf0YTp)P%p<gkNUD
z|JsCaG~qK$_*N4>(}Zs`;jb~_H=FRIP55>bevApPoAB3~@SP_7Z%p_u6Ml*bZ!^|Q
z=p(N);S)^wEE7J_gm;+mNhbVQ6F%95A7{d+nD9;$KFx&BHsLc&`0*xumI;5o39p*)
zsu}OyouK;ql?=JSrq(rj`sr^WQuVG*Xf_eyvHPk?AnCqn3ceMr1$AFUxRgYJ_HKMG
znm{;)aGQX~5@r_{XcX`@gxR$PmI`<zVRmVOdI4Wbm|a<5p@1(W%q}cYBjB?MvswZb
z0zQK<yR1N=fCmz0R~5(=a4cbVQGqN0A3Fmu#!R+Ant=BcW>*zR7Vs{@?4kmR0^UxT
zT~oj&;4OsNB?UT90&&qM!t9Cy?E+p;m|akyO~9)Nv+D^o3U~!!b~%Bi0)CS)yP80~
zfL|lbE+(*0z%LVK*Al1^@E-}YO9@m6_-VrIN&<xfev&Y|kU*}0A0y1JBakKFhX}LF
z2&4&k7GZW3fn)*SLzrDeAW^_~5oQ+<unG7!!rZnBbpFKhPa>R5xLv>#2(t?av<Y}D
z;R^{j3iukr><R))1w4{4yMREwfG;IX*B@9Y;0p=U<p*j6d^Taa`ap$%&mc?}A1D;?
zK*DtGfm{K{5~fQJWC{4#V8C?QfiwZ{CrnozNEYxe!gSGrL;-InOxGN+33v-(y5vCT
z2~q!q>52pG0$xv;E;!I8;8lc25^fam3c_^7fu#a|lQ3OypkBbQ5vJ=6EEMp|gz0hv
zH3I%4VY=Etg@B(XOcxs{6!4RT=~@H10)C7zU1}gpzz-3oD-EOxcot#0&_J?)?;%Xr
z8AufHU4-c}12zHQMwl)!(0N?cKjAFG?E;=am@Y8TCg8Dz>G}eV0=|YYU0z_RfJYLh
zs|(Z%_)@}jae;*bzK}3oTc8GT?n)<Cxb^>rC@Z;k-9T_f9=uwfsGr=Gg}*_3T_~QQ
zh}XXbPPw&F8QgF0Iq-5V{SG4FXJz4GFe(iZE(;NElY~1ggf&6JyubobPP;-mzgF;}
zKg?aZ2+?}KQ2qoB`B*>cdtBPaJ5_DF>iw=WcZ#c}G52^DglTF&#-OGB!YrHq1qym_
znsPq0vg~dY?Pm4zLIkR}>vUDy`9SO)&C8e}9reLGRx+x)T{*voxMhTb^i5rHpjWk3
zYSXUk)R;DPV^<XtuOOjXw+V?=ZLLhd-AG?^?W^#YwyME5Bc(nE6@16qe$N3FbD){!
zQ@z(_oC*TZc_9&o-n_CCr!qj_-F8RdVI)z`w*@L?c(dbH$1RSV9aAQ&b9X1ejcf1O
zzk_#IwYTk?8Tt*_FJb-Nk0r7Zkqz+>bOrkk{n_{N)x8pvJ#DSaHz&b0;y3AgRPDIh
zbUc=onp$hGC%cwf`%;Abc+OO{&1#eW8>Q}i_#gcOG+$L)qiU~9QiO|z1kM*Bj!ftB
zUG4JSlAxyUSG7;b+moPbd-ZovA*$b?V<IEV8Y9bvMwTnt59qWXBO^iJ&rT$N618ND
zS$Io!JSkhU6J#(b84zjag?SemyellcZx_6qRqX}4An=b9yh+p&M|HJ~i<d!y3<j}l
zBy0AmYVnLgZfH6YaVkP7-m~^b#`)(9y|}d_`u(V3)=VRGPdb^El}^7&5Y<X%$Y2qP
zi(jxOfLZ!FweBm?*lO*IiATLGL&%TkQc*AM{Z;?aM&`l;J`FX9(XZ%23t3m!s4C9+
z0<CKEWqN=7S85ZI#I&+3WW{hP!oy99Qs+X4ge>ZfEZoKKNin|4sq6M!>ejlDIKF3o
z4M&jQS0cX`M1F^x!Av=q+%L1Qtfx%7j<Gh^8a&+o%NpHUKy5l2?`p_gup@*2-O7@E
zs=v080}^%lQxdsDJ@T!peWo_;8K8ErQR@yXwa<Yc=nReZZyC?J5w<jVe4^SiKAB)P
zB%Tsfi%UBwG~?EG>4PDxTl+pjKQ|<*{_#nwc0<xc|Iou~)Ao2ZX1%K2lC1it?6HO5
z#Ex@FrywJ@woh&P7TOrJ4#|>HRaycoUl@gULrTkd_GX4y84ybiiuGHX-BLh7@l@?V
zP&}9Rp<aHNs@fJQVFt3vn&^**bf^T@;CL0SiP8z!t$o{ZZjd!3%y4U+vcBRQNt@2^
zEHKG!8J`yxtq`JB1Vx)1MWn9}^(s;!xRqK(s)!WnOo&ow6{SLmQi{aYEsxhQZ+wKA
zKRypnYRlt`1WMLofk`G+yWO4?)RjISQl)PacK57d=g$ZRe(yr^(Vk&2(LV?(LrukO
zR5c$ttA01NCQu!A)&Drjl))l%SJU_LB89d<kZK-#66)6Xo~$v{F$}ApJ{onjECZe9
zG*y||s0>4YSZhx}yJ47QglR^YXM`NTGrrmg=OJX_I%60<Yu6DrG@>KZ&!Gt*rd`Cm
zEhzNsSqVTc63O1S^DrWlkx0K3V?otdYow^wmvbpi&HPxYJCcMoIp5ndnePSR2(r_^
zn}$!Q&vh05vx|zU+v7mz0OyjXYI?A-3w;MGLEm_wySuK@4TaF*EF-mB$?jr*uoTaZ
z!!2n0LojmAPT%A3Hju5&t^Gv9tjFUV<e6n#A?s%^%v`TWu^edHbDOhRu(@j(mpxht
zH2ZUwiF{Q*tQdywj_E@6{8jZ$u_w8;QR`cr>qYZdeY4jm3LA2+hZAtFPevkS46Ld-
z*Q03zCF}0dgWcV(?$s)EQ4bnq%p4uw6Y<LY0~nl#eG6P9bzJ<qaDyli6lA#LRz|>A
zZbrzD=Mq)hCH#cN75`;_cQ?{OVT$v0p)ofUB@#F7L}AeRzAkt$Yy&UH;Wg0TLg)|7
z8x7{r0lR~`#yJ=G)ox+A-ljxvSImk;WUVSUZcIPyMhDwZ<aSuOb)DOg0u@btPxM~~
zmAUOnsxLmE7B7^Qy-C(}iV$1-z-U+9`j7iK#Dhco5^c+<XqS?GGE3C$!6anq2OCZB
zOYklibv$RH%)svsB&KhYHlzJyjE7X;SxSoY${HoFNy&3w*{I~jU#X;K!(x-VSE=4L
ziR#?7tc20ekk{<C^gUkL^`Hik(Xj?#ZR)%Hrb_0Fq4k>VS4oLc@?wR}b|uqH9Lfl1
zKU?=&HD;^oU6rUpU8CO?N}T-(tsp^v5V}GkVF<G3zl^B)g~%w>52NCP{ct_;sGFrm
z-FPlmwJkj>{<ThyQiX4l^)3cXOxv^ruBbyeYJzANixFm_-f4IoE1>={GesVYj675h
zJV=aj!7@zx&EtY<;z2vlpp7EW*_0=8Od!$<c|M>VWWlfop%@RGBE_gMSpOU|<T!86
zUiKDc_LTRTP?$wTeK8ijrSl)Kq@hi!HpQOk_D^j}O#e!b*D*f$(+&PHf#re$Y&i=L
zZtW!c@3|>9TL_&5wCeNN6SXZmXRquZl2za9b`eorC+)n^A?(L<dG8*r`$y@~-nMwO
zPRHT3Xo)VYDR|!(7VcD1PK(=s>RGT3#bTG{)oWDWe0vI{SW&~G#%xUA#P*gAFR|aL
zv^i?<)C4ozue7<AI`E*}x(UxWwc~lCa_d)k&R>Y<FO^%r!*fw1p1)OY-Gk@pop=r?
zw;smxef4<Wr`+0&=Lg&He3Tg|Z83Okaw@4=HA<?wP)W_LS5otqDyfAS%9pk&sTJ)?
zYIUcQYC|jQ#|Ea(Z4&M5_bibze+q_LF5dvPxW>?w^axk=uBgxVbazt;CqJMs3VMcR
z-~=C(-6h>a5*B5^(z{IPOj&@52&%|wPnzI&>WS&=q@sLjK@~ZNnRMhHW>V6WVJ0oz
zKFp-1sl!ZqDjQ}}lxLVpQ?rMeR24tW(3NvIl;s`{ZA}>tb=^K3`kFc%3M(59jd_Mc
zWwVDvXYs>bzF&7YsdHBejV+KG8w-sEeUq}F7JW}PMl&#i+@KM-eY5Q;uvN@ams2&s
zLm}tbrzz#mAvM#SN_p9kg#bK5mQGVDHrFdDWrJ&!lqrK3Dk)P3*8|shNJ()IhFJ4O
z*T*Esol5>sh}!T5Dqsy#cPaTt8Sw!k*5jdKKV!Ba24&9woW|F!<bRDMC*Dv}b}RWU
zL<!38naAgcvXzv!Y$g9wCFNUK(q<)pJ1IVOC@DLD{u<$0C4Uuhn_*$_3MdCM+o|M#
zLF9H(o_4TmlQJW2m6G2IK6vUR2`$rUuTb)LDfwN~vXzRh_4^%4+xJT9U+wkaHqD__
ztT}{;@&ihWEGv@kV`-Y+0Q*+RwO2`f-OdU?sanZ&A0nm>ZbZJHz5xd1$4RK<pFrtR
z%JoQcG+W6#;!yIA3d%N6ehA&<f32h(RPvjQ0_DNepgzclRUij1;PDg6%|bOELXzW3
zemm(;fa5XfYlD))!gUzv-A*O%b$bnjQ}VYV?dN0<bqUU$VA3u%qvV4!zma9vm6Sse
zjoJ+Wt>m{M!7e12GNc|EtOwnPMnTspc^)V)^=)YcO2sO2+XT5%oP*n0_wCHSp4q3C
z*>h28?No9bRkt4M0<*mNLU*W#`SuzRvM}Ewdg@?^_oz_|Q1^##poHI(C-CKz;vg8T
z6Z!_1BS6{*Blm-EDEX_EloPZtrEL{C4G{`Pt4o35rJyJk-yKr&H==N$_zY!j7DfQ6
z<!}o`9r$Zf@}Yp$pwU6|g+R0@`QHPv8i+kCQlUtUDL*7fR_6=>^Bq*eSFD`X`1@pk
zHpGC0a*znV%`)NxD!U0|4STLw)Bp$Ah?0K*{7))*@7ZgJNtGeBwwvmoJ*<W%1z$tc
zJSa?aC-nLQ!H=oeA0akLc2YGvAxa0aCzbpiXrJIs%JSL6>S@Ct)9Us@kNZgcBT-xG
z8Q+X_n+_ozgv>{w_aWh0YAMZ#E#I%?qacTZ_$)B`2DBT=XbXrcI>|^HDKZxg<0y<!
z$=8u)E4&)YosUK!46>7K;aXYFkAwsP3WGKV#Nk7XMN<!;iZ&tz`xTuy=deZy+A3Pv
zQ6ga9T}lc(#YT`ELoMxA@=h>f$}k$iDyZU7SrL$zM%2mb`3MDd+6zH|<^T)(4o{7T
z_|ys2ZpWV>`yLj!f!Krm(vkWA{sQEH+RmG5PXn=%FZm<?b|5|WEJ(2f8DP-@6^fJ<
zy3qY*w6c^{D9sx1Z>BLcg0h85E(CTf{yqScFOkJAC6C&Iha(Fh<m8+gcL>od$aZr*
z@)LCk*->QAuyzJtG47;LQ6uyZwk=@0nNiMRXiIF3@Nb_G(F)#ebWy0pjY{5DF#Lk}
zLBpUo!__Ml0cx?#UIAiCw}x`iqsW%b?3zRs>R`T(Mud9*5&3dxuopV~k}SJGSH2F_
z1rF<3`lGBZ0Kw`>nQE^<q7Nxc6XH%nnAIYe8bgok*wEHNjrseK?kJfYp~k7cCTRj_
zc==LpHb81G$a0;iOry1afbMJxI^C4p(dp)+->BFGH?k4ohI&*3!mm)~ial@}9Vo<h
zQrF{Q1CT$0f<9#vDYk)Pi<06Qz7(+?C}+hc(G0c#|2g6hgDo5Q=Sm)WHT3DgeFFyU
z1-~PMBOp8of^B5dMqSY;;dKyxL<01@qQ$_o?j-+a!Jh>v!?++H9dE@RWD)?6HA>zp
zbkoDxxX=a40<EJZROsMBlFyWsLBo|iR({2opkQ{AB4AK#MOT2i&_y+v^phW)WnO=`
zhM8_?8WTUDdrb~jd%SBgNs>bmIUQBzzk;Q#j?+2b#FR!1JFqhGUpu?IbLtLz61W2S
zEC7zvI6{?^-^YN#tk|AHG{-DJt3Hm#eHc0;eG!v%H)nQ;al11AKFpS}^5*>sv*T<$
z8<PNKNv(Yl=A;ahunxLkxfG5g@w78m@%c`4-GgP91J4e)U#!F%;cblvxDb+v{4sSI
zih!S$Ja*LZ33Rc7JCFMP6ou=oh!=7CU%vxJYgT=6sxM2&KtHJygq)Svo`KECnnyBh
zo_<eekLx&_^N`1PfZE4dW>#7Q_b!3#Zc`dMRqY%7)$hrwMU<Cw`qWyx;01Zakd5RR
zwPD^@%S~K_^U{%p>L0TepiMtZq&E^_ekKOc{z03$)=CDLF?t4J3axs-jX~h~1@{?F
zS6i4E6E+$6Cw8me#st-SJa*Q2(|oo!+HOq9F;`xd;KBMvmR$4*hQO+fZ>w6v1rz4q
zF|4;Dapf&2>wETvloJYO!B=2jY14;b3zLgQ#JDk!VySXc54oE5#9{K1Etfx_s-?gy
z3sxe%8(R?dgl=s-;7~4IBPw<w5LkN1PCuNT&V39DMWO>^l{xVkU`qzFzPGZzSsYPc
z0_W7e*Hgb0Ro(ER&0$-Cd{BAc=u=Un@KSQLhSszu=75rjHCM#r;aOoxh~IiZG`8|~
zgxe1&w||YW34fm*Q1aSD3JM6z1g;r>>oNO6@F_@FAp|*H!!;K)Cn`f~?YhjIisGYw
z{`ROw_%!0PPdVah`Y{=eer5~4uDNh+OmKV>Eci5+GG{H~=Cr8GkV5Q;yDnHO%y%fw
zH?_xnd9S{#y$@^tilz>0eg?8(Nh4e9&d%Je%&*0wM#n(?G$aj7#mogNlMXwEm0XXk
zIWlA@>*9OX1(x+}`Z1^<nv!i)sEn;tN%^a+`V{C2of*c>Osu2`jYEB(0IlQiTc{Aj
zqiZhs6e`3LC5Wg$hO#AZVQG4A$$Z%|q9wogO&?|dM$|`({s&qN)(U9V`aO1*&1`3T
zST>#j(9Z+me_%zsP&V|w*Icj_WqZNC*(e)BS+<juw)e8tLAeoSYv0yK+2V|{y-RhW
zY|B`-1;30eTl|S&*~Wqpez?|N1932n?Dzb8!Xfw5M0gwe#By1Ou!Rp}Flt9_eTtO`
z1Z)l;pb|GC+<`*2hbyoR+*pAOWg*XsUdY$bFGLjb_|`rO$@-x2eomGsWGM>C%tS>N
zT|z)2^=B`PEcq{3ayTuUt|ImSy;^NG3d~RzIED4pd#yI31B@u}z`hE6xiJ0~eJ5Dz
z?Qg@y20QG(d`&&QXBR7TVy9uFVDz57o=Opzl?J9Yg!zZSpf3`NGcW#MATaNVy#l@e
z2(aK%jOwdSB(GCZ9=(EzvBm4F&IsXd58(<!I5mh{HjlaqCQmoe=-sf%+oFH_l`MhS
zuN3ksq(BXFX;Xg_XFdpUn@s<20w}H67!|aCF;a^yRe|}PfvJa4G1GqfqXwoigqZ;h
zdbYW%xi$V<$L)?g9Ctbj?r28e+O-_QBH=nu*79zAI+dmb_Rt{2q78$ND=cK!n911O
zK*pdW#h2i}sas2|Tc<qrIy-;%2M|HO4s{mzBQPd{d2qZT6F!dw_XD#sxf@3$9Qaz{
zg;dLjp>NaMwt!Sy(}8_<bl#ys84CJ2isknIRp=^5``&67%Y%Ah4GX>e&A^21;`aj6
z(=I&MS&Dp?xyWYC3<#Cz(bHI(E*3YJOi_bf$XI_Xg*6yyQs<H+!lX_K;R-{xb4U~&
zre`&&Z3a4MQVYHWJ7^|sQty!>Xi^22QwF({&q42vgfaSpIew#0GSY?(D$Brx4eAO5
z6E>)Gh`|VQ3Y(+0CeV%Aimbe;s78H=AY9Y&PuAcc_*F;xDp*=1{YR|q9h<{d3POkp
zA(guC!5=jo7>0B}2M-qno~0R?;9mg_S_G40Sk=z@_nX`ND+aSz5bI|k!@zeKw+9ye
z1Tc=^OI>(6`URfuJBp{hd~zSd(@s8Jd0ga4-kspY@nStgRDcLE)d<_9Qag=ocQfJb
ze0mc#6ZnEH9%}d-C8#H64J8;xf*5FwMIFJSncL^jvBEQHg`Kp*^(YA0oC}@|&qNsB
z&orHv@$OD?;rxJhh{s&~55p)v((ibK3rc%233=4pj<PXi&&IIP2$vdRy%8=n!WtQl
zs{rx1LVo3{EnaGhA9Bb*jWFb-i(!%xCLjc<ihv)rC9YzC1A47WP)AhR;J=zqVhgKM
z6&}kW28Nh5_#ZCS&&9zyw{M)P`n)wv>GRPAyZyzvZvVT2ZE6$FQpCEowQA}|_n!fq
zQoVa?^qp|{OxJN*<*5_=@inUE6~5c&T_oAmGl)p*f0vQ^t3A^SYX7^`3?z|(k8)aB
zV8eB`JyGZBK#}=bl@JIRLjoq~=qE6$56bgR%(JFH-L3tgUm8q3urfg=9GB<z|D{#t
zEdyU$kGw_NA8}4GqS&N#``>Dp$z|Zv30X>y<VW-+J(H6XV~ca`m<pmw|I`IIJAGYF
zU87QaD(aIn>IAnh2YRV2RegW2L41zyW=QedMAi3juG{x_Disw+tzk|SF}@$?i(fug
zi*FdKIp4PdSOH+=SYLdDvc${DoF=}<`Vb3Dyuk2VV4fd$WX!wcnDY6tN5-r~(&Zqc
zh?@BJc~9VZtg<A2r6wMEHs*4dFWx@ZH)Z(=L`+$BVyrfFnakH7n3X0>$GLJpP7t@m
zvmeA>M*p$F$L^M#%ogRbYvHzXv?eY=B)~F+Mf!Z4m4(#%5e;*gTkF<yajMkqo2XKs
zpQ0fe#yJnh$rNt?-;<fp=aqpkr8hQv^;1l2{8lzhfy!n(1RELnvU;%j1uJ`1ij>BU
zPArikImqh_0g6ULxtKNw{eJCpcyRd=x!aBvnf5Lu_(ZAwBedeyK2p8kcd7n4I3%?<
z4o3yLTRi8h+ELE`tFXeicN2*F?LhtGnVS!-9m5VN<!9q>QsVmwU;!LfhRfHh-tO2+
zG?29&PMlPEElo}R5kxm(pkAKej<2Fle5FoPGP1A$pOJ+Hw2UmQ-DPBLz&3Xl7K%En
zK1OP=;=18R#&BP#SvjMT&s=?Ojp0-4G-b}muFQQ^@mg0$wd$W4<I2QtdxuN8{!6Z0
zr0#?8acb%T&j5HQ453m~@6i~~S@@2vf`ErQ;*}+1-o)Q)Je#HcNJ#Boi*sV$<MFdD
z*1AGxI+w(Q%=sqhoo^=aYY^m&cV~Vy>pLjc<8b-Xa(n}Fd_(On-yKl#>0DZ{rSCD$
z4?KTl%s(C4n8m)K|8yuzhQ6i^{ks<bDzf`0GInQfn6(Xu?32tRKXi`h{hb%V1kjJq
z|J2+&Oj~H`A8?+Mm8yN_(hj>cj4H4nkF1f}t^KHfvXL94zZTGbz27xxQ&@e%X*`lx
zdfhE?;NpKrG;@qh>Or9Y9;rD@k;rGoMn=PvF&GDo-qhWVz3Ov9y{hk4onshDd9SXH
zo~&2Fk#5stKH=`PLwR;}M;w+^mdSHHYm~Y_AVV^rCq#MHUW2gVSu)?oY=hmk!B|om
zjE9i%Vu%3$^=zYt;af*S3%<d@7S3R*$Y8b|mNCU%(eNj-_V!?XJLAoKb4`2`1m6TJ
z-)juMrQ}=K@ZaP+0(_m?RQn*S+_&37WY)$3(3!O1*2?UOVTn&SIOmabZo}*3ys?eG
z=_j0aEmVarZnq=a%=4un&lYO9JIrnmCM?;t-Ae73C;<G!AfBK}Xj0y3!qFSI*5QP|
zLk&WNsxhj6wp~RN(EsaW*0~1&Pl6&LY{#h`)4#kGk6;44d9ap>c@VG(js$Fby|WsO
zmAYaWB>82L-%SR;tO$OAJ0w5gP5f#vVOrC$N~z644$I-7gHs>&8<^8D2#i1+bIPE?
z(i)aiVfDx`tgSM8hTH!L?D1y3N})XWL7z06=8o9(uY|5#=mP3^gHJ8iO8a4O_A^wK
zLRR;GBvq9XQMB`lWzm2)6|F86wQE(CRC_XcU25=3j^MR_y5t4CnO8h5#aNWI##~GV
zVkU{e>`yEzO+#cfmHvY1z0F<AX!KaeRce0;%rc__;xL*P6(EDa={=~=>Vujot-O^!
zsE<(qizNN}pziEJeH@r*a&y9BH8aW`O(&wn;o0+Nm}_t#zZp#Qm<^a2Gy~&{!`U|N
zRA}s}i#P=_2<X1Ieh8XGAWbxPMbIoKO^%i3xd@uylV+}!=EDe@`$+RIE6tGznjF$>
zC(^qnRMDQ%L1kszthL;6X=om;JXMd0O0aBtJQ*zj2D=n2eGbmmC{Mk8B22;&)3y&l
z^%9X5E3AR~qG-HhrkT|9P&Beq3z;Vx*<B|2*BZigA@`1aQ^eJVlG+iWkD$nQYB5~x
zd~t>)=U7}HNP;VXF%jHr5Hyk?upR{Y#6;<FYq{8NmS3rTun!xYinxw}9SI%%JK{TH
zJK&Z@M_x}RCfCr{(ugo)={Y+hmqs|Q*RO)i9Cn<GYF+upNv!p(heutJ4BxnX7k%q@
ztwwnw`+?dDI0$QDPU)jC2PriR;6x{3l3uAz>A`0^XrokYU_+RX)!EjzM@-L^TAsd)
z#)nG|ri$!hV$`$_bl`geQdX-IuALo`hi7&a{vG$2n2$G@+n^u4843-I0>(sekwHL@
zroT*rAy$HTgP@fJb4jqDO)zY!JB&6;AFdaYU?VW$<ppE>h_`XPyAvJS#}4gtykdX_
z2F{aIElw?7%;v#kn)19bUN31Gmkg8*1B39%YIeN|;;5|t(Av=XOgAP`+C;pN5I)<l
zYU9w$jY~S*q*Ud%j1vPYmv4++#o0YyR-(h_u_s`V(cPwvLt2bnzjtc;o!U|T!1*+k
z%@`dgvNw~fqFgw7SS$ed(`U7?cJ}J`tn0aAC6*`CA((Z%%vu#CKEg?Sh1+*YPThA(
ztqVJ>Vx_#H8kWD}0#NC1gS+ple{M@Q*3hS{VQy<0{zwY$@iObDD1zLk<rp|c^hDfu
zO$5i9hsBoexN#lB*s|Ut17Do6oBK_7ZkpK6-K%dwEphqzX)w#iTO#u|gFUC^X>Po`
z<;zx`nVrgluYtr#aEvO8U*x3NRs1;*pJHT~jcL1o5hW^)Ur3THck!s-slGZkY&URX
zw(q&jt>HB=yd8ppcRON4<>esX6{|VS><S&RWh75X=i&4><*EGvh#YOTOZ!HMcP}KF
z;2(1cSew29Lc6fcr|nn0+vBhq;MR`kXzf^^uiJzJ&^T)9fvFwCm{UBD!ZX%-#PgXs
zb$dJsF1!)NkIO&4+vU%}OA6~e1Jo9@5J=IDscKViD3|wS{H#gx{D75S!z2_N%?WCY
zL)e<&VIUGiObtYmK!`2qR#n?r^(IH<80Kcf6fvnbHrE>|3O3i<NR1CGcm+b}gnR3}
z`6Rfv4jt($V0K)3Yy=6zR)iD$gYcfge8B>5BpFKjw>H@way{ce*k5=1{wf6M7|!iQ
zA)cvJn4$4(b21Mi;}Gayh1@FU32CCK-iUYNxKfAX3pl&PC6P1T*n)ZFEY3hT<K-eO
znlA5evth*ya}Bum_*OooshT4pn@4VE+w<m}<?!d?B>)&9PPTI?qFEobiUzdVjki@Y
z)D;|S;&UClhf-BThTkTu#hZ<asHa9DG7Xy_td!r7gN4J)Pn0K)B5#+rfreY^YWnFs
zzEqGV+N`k%jG9=CG^+oyMLf_>9_feC?p#_6U#_Zc!hZ7wA~{%6LG`3I<%S6s-#Ele
zgNWcNOR~c_D(pr<Q<Md_z-Yy5K4>^O+Ws8vc*iLVd8C^OT)sv0$*vKLn6ui{5vvwA
zyHmHSO+R7j6pNof-jDZr<m?*)xU~&A+86pZ?E2<t$6Q(<N9*hujPnSUgIxZ2FYSt4
z>%N+Eh0C{?EZxRSAyCQ)ZL_yMrplgv*r^3NuHuSa=zW<WS;sVKPfksL{>UTnibISi
zt*L>HkLNnZr5<qkXTJrPQT9La#(IX<3i1=$ttX6!CZ^Vg3kYMHwXv$VB>{uW1aI?6
zt*NSwHCgB?hR9G-h3a!9iWAx}qC6!f9w&yxt0y=~G#KZW`@6c=WRDg-kTQQ2`#K!$
zN$Th?wQ0P`fU2p1fUKH@2>J2MaKmcxRt@IPIod~VbQ5AkD@-Y~S$SeVW{Yl2cSELe
zW;4wWHUFNy1`HU^L+J9K^Ag`9AY<AmTIw!+`byI3Ioduq-rdRVfNdc&nn(@E%yQ5~
zuq}f35fG7Gw@FzrDHar%tu>O8?wiM0aI&~VG4#ns*@{TB>?m}Fc-bNjGk*7osTBJ}
zoXdM6=7Hn}FtoW-zo)H`seb%Sj1H4gnYDmW0`G~qS%YQ!@it-|Mmiw*V(R&5Y^o2h
zae|P&2bvf<>k067YJ%#+p6Wo=$EV4_45p_+*rKm%U<a`;yKbH1J!U~gs!)i~hO!_Z
z<`TBA9A8CV&WMU!93+W#6|d#_aw@}}dX&rj!qPvw|58d{Lo~aN4`}HJarj<X`j_C`
zF+5ZrVee0{7*eN)=UndA_PBg`SeceQ&?$LRomwYN{sNl(w;gF=d4g0Or9pddfBwj5
zhc?`)@xicmhxa7R-I012_Wm!}`|WR;?fv1Pz2gD)K3bal=&-dnCpf%oY3>_c+8$x?
z4qyBeSnpdUtUjhmEpBr8+-sop^>`H<wKcN|CloGR6~p*ao-G>mt&-R3P=<hrr(iNa
zROY`S8UTz%?UqL4%xqKUe}M5ER=ldAJsZN=44GvXUv2ZyTVqv6S->eq`d4^0#a?Y`
z|Jvr~>A^FOoOJnA5Mu~2I{tT6DLL92L>)P)77LkG)fbPqTaXWI%j`cu7n!?wK^wy%
zjNSCDf1s=D&tg@p8qSn;%b|W|bB?CZevEJ9;&@-TUiBU=aW*#ed5G_JyUO!>0w>K{
z?hD^YOmIJRtdS&7##O$LxVfu!l+(Q00fRE^0#hdF8z8I|`Bq`PiTpZFWA=yN4QG#>
ze-NNK=g)u<;}iBHtp>G`^HB-{^ggr;WS}CydPpHM>iFl%KmUZ;)8Fy6Vk)r9Kf#m!
z3OMKve69Q|#*Z?#EEGGB5pRVeW`rWjne1$26Yz;QYvuN=o<NLtMC2a01x`*s3}pm<
z>Jk_sN8oazzacvQco;2I5Ey{CpuaG!Cwq5ixU`cwzAJKy-#-Bv)Z#P$`AIUCvN4uJ
z4=fu$Uj(W>$(~M(B{4jZvj7kJZVVyV)0bi7syKK0`c)t`Rs|NYL%|wBw736$&<7x~
z?_3ATzfu-F1#&n69KXA@aiPQaKV6VT>wNymaji+)n)+GHNdYKJKGu$<?wS)HvtGmd
zNO(a^S@0q#V_N;wV={FmN7r_xuE(G>b%V=yf1+B9aU#h!d9Fo(A*tsQ5ypB3i&nWk
z9%o*Mp%JQhbB4_nh`d3=dO$c%IxE8<z}n-mG&Ya0zftlg;CSu?9K)T^spJjx--oeI
z6URGQnQLcnSLKW74kdL^JhoYU=jLomwq<^!%s&%Vg0kZ@=Yl`6xL;!XBE1o&FJxZl
zxel_XH*yc-V&f^+a~hs-0D6!(<N;IXXbG<m<_#Jo9EJttuEt1%WP;y)L0Yi>l{^>Z
zbwOSi<UPfo3t1Z}b5`b-+1oJ#!6>(5FhXPo@jCh=<h%q<AO(Wrq#6dwgpvtie@?08
zXaohc6PTBQ{tB~&pBSsZ1be1aLAQi8dpRg}+mq2cp?kY$kXvp#A?Wqcsmp&OR~iyr
z*k+pbpO1&>H52Knk~_|%<}D>jZ%kjuLU2Tw6dd|O{|^imJun8qoB}86vvD&}hk|{j
zxLKz_Hd!*++r9y-W@E$)_bFm(gI*S69y`2zHe5D<Ga8j4F`<68=wF-wb}I}14qY8k
z=0L;gIJgRr>D0Q=yLnRa%Jv*|1*q8o`w;AcU31~O7#v=mlZ`{zP0A1?S&r6&SAwCT
zyIp?Ih6G3I$7cI9Snx&=vS^gu5OTYr3$~B`$DTrxR(WYiD3|pyAn6~>1LlEo-r7Wm
zx8-EjKU}`wh(RB=4pQW09bfzoojGTB^qx<?Ls$A8#}a$>_fbY!G)(L>ERGu;1r$o>
zK2*O2X*y2R&&R4tpaC5oR$)Rc9s&zasAnP4vLEK%U8-Ulk>=7)sJ^TDsPgSK4$0AA
zztNj8DOY{JRmBWPUO|y84AEWO!HHio-iO0EVb8C_gGX298s&+5#o!Td)4Q5ZV(fQB
z-;BW}96zTI==ApM>i>uVA_n8tv?=tROW6kaoT2)b@fovgv}G|a_nhYI|NN2b9GZL?
zlml%Cwp|ur*fIO5c$dFyUVJKcMvd{DZ~Tepk34!p%RV7eX|Nord<?Qf1MR9W-E37{
z!|{ptA@RQTFI`k;RJW@?^EL`WX+!Tj4&ly16Z**WWXtep#Zx^AI13&2Z>-jbBNX%j
z+IWuk13iBdD|s1eNX)I6JPU7^>lY(4qlw|n31$RP2|U(b-49Ziw%Uyiv|N7WP4u7h
zIxDD<!|xxMcGp|5g_NK^kWUXr=y&9>I*X%k?vk-wZr@_gxZESe^a$s`F&<sxPTj#<
zUBoza8;+Zx4d?3BSkdShEXRkHr(k|nOTqJC3pv^eWx<n_Ro1IuMLW%<p>WE?QRk@~
zb?)o<1=k}m5S25c;Dek_IlMo`&T=|Z4<SbmFfpN$`Tgwql*7A*Bg~IET1u(=ijF0j
z#}=~YR8sM|u&+VFzQiFM&k!t2VWj7_CmzXkXaj^-5NGw_6rPG_aT2g-F%OdXX>OtU
zp)Xt7+xQb?aO66zf3NmN20`0aeKl0Hnu_xSdTd>z5MWLb7nD4MwN>8jG2#?_188iX
zfexSI8!o4qU9qz=Qago4VD|sQkUPufyW@niWJ}Bk%92}8;8cpTWPQw+%97#Xu{LIw
zjVG%%qd#{+2KaAm|Ea!7m{xv^9y}f{2X|(CkG?y8BxaPD%%HO5h%x3(IwE5LdhiVx
z9^SHNy*fuP<rqVme?DAIhoUcrwFX8=x6^TpbW7gVrR1W7T~^=?un2>=lBc;Nm0vhB
zBz&2X?UNu=y{IS@p1iX3Y6FvKU=sCd#K1-U89D2z%-$zPZD9J1t^p4D1F>++17XHV
zDDx6}-Q^sY`Q#NMmWVMjf>S8VqI=y$8Bx<C;x-X64m;%L&Bro{#Z@V0>>P<LwPF`a
zY@rqVvczJ6$jsqwiN$8F8M{_uQ>@s{5}Rzr?vmIfEB1uMVv=j-dpi7;SXZ_jvsG%(
z13tWt?-sk}g93`64j;!=YKQhp*RKcNz+UN&QzpxHn5|d3eLd(71JhH!A9~R3?v-w9
z54!EW(tXl{jvDWo@46mzAN5Md^Dx%(uI`oYe|pf-ZS>6d?>*?2_Dc6(J?LKPmF^Ec
z=w9fR&JV0<r{&obXtEgcB<M%~iT271=S;Crg<`*AY>hSc$x!V3jD6S|OIDKeON_nW
z8v95nb}nOQT4Sq1vBiv?W{sT@ip^&1oz~crQ0yg)y~P@PM=16b#@=9!y*U)y@ejm~
zx5nm#VmC1MH`dtfQ0x-MUTuxNHWd3i#zt(sUKNV1WbDNN!+jb1lgadzIIr5D=T+Yo
zC+gtF`{@V&4ki2E6({usVuwWV^n~m#Hc3P~U8A6Al8Dw2VyQ%kv7VrKNg@`7C>BUW
zO$hObKs+q^l74!*gp7`&NJ42cx5*MpB-H=60VlxUHcwlKWv}a1_vrDeQujx2#nSp5
zjC5{Hbo*|>6&2GkUyj!tqvM~f8m0QK&-KO6Q+>DO624aTJx)o1a{FH5)ANoaV|;a+
za1^Dx$(h+Sdv#v(%CGTKO2fl^Gle&vE`J$Mdf+uIPn^nC4H2P)BH9=+CKQnviue~J
zCWTlfgd+Zo2y6X>*B^3mnHbg|xXN1_Knj;{e1gk|>&?WzL#&$lrQ&@Lcy%S_%KTiH
zgNcW>%AI;lZ90Y_-AOg|b2W3D@)$Q~)XXh78Ru5UcDQ0bQ=VSs&fKd!-ITBqH$Lsf
zc#*@{WLM@-$`e?y;so^{2yFVv*TrxYjG4O|BO0vlV1?ep)48}A1KH`Xp(b4ZF|XtP
zmlv5R{jh%ht1zaRtyn#NJv6|02DlvUVwZMj!bE@k>n?44B3_he+7WAvjJN26LE(zI
z69f6x38rzx$_y&~uZuAfI3MDe)#bF=a`kI~)IWnF72XdSx?a(x9pdT@UnqSB{fSk3
zaqi53;#~}BUBQL%^Kn#BF2~~ZcAQ&lQB#jlNskpEwL9}iYzyZoV|Th@4nli5neEEc
z%?Y36DB0~;<mOx~IVW?q^28VrLwh@^2)&V-!H$aWbmp#RHo3K9O6@&pQEq=$mn*fE
zx!{N?Rz7eu5;qKC+)6Dbi)dk+lsXPikR<RiLRJuz9uS96`{ALZx7ScHtu|H!FucA_
z?g#dKSM_JUf9h)58J80?Ga)dUJ^(93Rmk^PM>_W`<>L_6E*3WjyEKeQn-X09Td_O2
ziKui{8%Y#*AA3W2yW?lw#`j50<o*XSC;l8mTkco-@&Vjt%<2pbRugW+4N#otNsJ5g
zs|iU#jG77i+>evGRj;~mq<5l!N=5?prns?DkBJGkH1Wixludkp7gdFF$-w2$MjtZ)
zODRy?c+>h7_g_H^4Ukp_*oNg9VTgd09tT@9qKpt&Zn0|ns@6>KpR<c=0m2nN`wGqf
zH-`EDK=TjTc(XL|rvHW^k3k~{n|N?OFK!F1K|T)cBZu~>@C<O*qp=m~tHgw~u{$@3
zE3XE^v5bN`we6UOAtrN8)h{`-%X2Kplolq1cDn`p0@+%pJ{t&*!b>~cBuR6KnURGg
z-yq47B^mfPcE>pG;8;!#T1a^<Fu5gRCAx;sDEnzFCyy%x+Mt6VfXh=!U<>;@oWM|>
zSX~<B&_2XHkuL2BH~lk=*9-2z^cWMpoZ^GrnDLD(9qSvl6k~fjZXB;)iV5l<M@*+v
zYs8{GwsP!v8H1Y-+Be-dB|5d84()o_)Whf-ew&M9`EP+fSu}msKWG3PpfhuyvS1?(
zy?C8N+og3lj%;G1Y}%eIFaM#3b>k*UtP^3mxw|{&b8Y{TO^(b>4kc%kf6B-hwR@Fo
z8J3*8se9P5`+~K&IOqc0r?D4D2kBOE4BI~nS2ST)rR=SPf(y=AtICoi?wF5MylEV>
zRy%U!3r8ky<j6VZpZ#m_Y*D>G;Bfk0EKHT3S^k(fbH2`GpFraUXTBd@p2O4R*)|*s
zZ~O(m*5PX!&Xs?KziI0c;_oC*m!JFq+mY>PGhJ1yu;0{yiTB=hKw}36*YBLcM=Yde
z#NuJj$7<%Ts(}vegtj$+mm6((Z7y79=fD_kt@rylOlwc&Qukq;KHu7Lo%dvnw$*bH
zc91++5F6u*QC?c5yt66h3%u9<zrxxHZ&&;*H*z%W!<51qv)$pH?K=U6jmkTXF>Qo!
z%2u*pMd%>ooXR`xj+js3eNJQs9&Ny!e-KXmFW{*sOgl%#;-EB=!vM~~N)D|x1<i<O
ze_%4&E|A-um<6Asb)e}uaZ_2-SbuyvtjeLaA$@Y}?1@~YIH~PR-6$sVsjD2BA1UrO
zc$IXgX2($;n5~26`y-p#J||d?(ucq-aE`}#Aw0CcAc(a?R@#5{!;*Kv(Y+Q{sclv2
zZiQ8Gog>4CLqS;FNYrp512-pd^~+a@IsZ8L!6u&a?{N()PA=h`OX_}_5Z-smhR4cI
zh9+R4YdsgR9cRf6i{}3_9b3u@v(16i&9UC4dULXxxfZw9_7g|luwt<ye8uZ_G>`$R
zKYo+Ecx-;cLXlSv1cVP>t%jW59q<6?uPyCAFPodyVm(|OojSlapGFUWwUyCo@h+ao
zNJ4FCKXxb9HSz*ITvU?17D=G>IMp|g16uExS8=03T0F|=o7c+hu_ls;mknMO*VAzk
z2<%Q#eOfRQ?=i^81geWmzNY$_$00Cz$_q7>1-xG&wElp_9*&uvzRMPK?IgviEYZbc
zklU9go9&}q^TQfnE+oOLM6YmdC}+egbX=~c-LcsiJDlc%TVqw|Kj8{ZVCJNZ^SiIb
zpyi(=qm5*d?m}$Ra)Isa^urw^<a(Ki>$oQ*k34t~Jb)n9m9!?@i-=y$k%~-iH&)w%
z>xagA+hptcg&~rSTxDuTce@5Vx~s){8-}1NmKFI9O$rLmogeRZ;h|VCC}!_E&qJ!#
z%9m<Xe;(GrIxxU#_6)<D3V5$3MtYmQjqj7%b2?ns0Q}u3Uz`!wXJlfsw*>FX=p4Tw
z?iRi;b0<CAG<vuz;Gp2rPGU`1xHG&fvo;G8>dvZ9aBT*Vy*V=60K7p>T~n2)ebRBC
zdHrAY?nFDnJ56ixY7B~nVX*4`DaLaKzGFT8<Ay$g%_H+SuEFMEtIIeH_j{fgq<sQ*
zVS9<Y^I+VaU2%7wh&vB*@w&<rRUO_P*wx@HYr&r}D~7Quf7z^fZ=(7DB{=-?1Dr9s
zTHGouhvxv%E3FgzgZ?pS=`df1f9P;$%r3Qfe|Uw@KT0g^;Rw%29Cl$t^B#|>dIY{d
z+lD)E9vy%u=g34pIur0p6o6z2Bnu!-0<crWWJw@P0J###6+oc`3I$LhfeHY%BWsZ5
z8v%ZY><_b*x@`ghA#Ml@!N-sg5-^02fFXng3?U?72q6JO2nmFRAY7Y3K^*8Yp;Ih@
zLbsynie-4`=rOM<^Kazr7}`w1C5jj$q!<ok^iaeNVVrk$p2<NO>z2m2XVVKH!1Ck3
zdQ9fTk>FL>qF&LAOqYL#Cw(Ib%;zh;yL<I#DG~lGxmSONw^6MA>^yj#|5JZvI{#rP
z#q7`cGIpr`BK%p}f9TIH-}nFfv;VL9Gd3*VGqwDA;ndE7QxlspmE$a4Z4vfW@J93x
zVv{99jl+c%XrORt(&-_36r#V6SK}%Z#ck4h!i&5|8ts@ZfW5D5y+z!C;P!@Wy=da1
zyBDqZBF6RAdWS+twq7?HY1msE^Q*~RL^56i{tEfIeNVC9b^HFn5rW%yb^75Pt=Z+>
zVT{E0VwTSFoXdY-qRW3L{vJxi(P39;s15>SzmnItwP0k;KJ0FGU+3fRVLpt&-wk}Y
z8hg&ey3jFRGi+VV^GA+q8!_@mPIERqe*`y$&O~Q-&4o=OzAIKx#e#~+RRZja6UewA
zQh>Uk-xuPigWn;^?--LF74f=YurJ8(m;iObU|*2mF@VNvsmDdv>rhVOwa~TbC!6ut
zlI+4%0vCm2jOy05fr7=gP9(4)|A8yBvkDvcUj`n6O7U_n-cZGC{W#uG-ORmbj3V)N
z>S-8%ae#wYQjc^D#oFCFT$Fo*Z;HO@i1`G`#R%aVJOt+lc(s}|%eBuzHEKo7M-HE}
zAs(pN%i}fNg1E%JB0dJMzqZ7>Gh1f0X|2DNXXks)Kj-f5;+T<F3}6{s#nBW$u2_QG
z<pwM6IBVo1LZW!eu?-<VJjl#AhOI_iJwje8uln&OC@e)>iv|0`s9T>-Q^Xla;^pOM
zFBq4fVWsbuB=br6vuLu~bEHRG2;F`54evf{<J}U12SVn@t$6?MLF>l>EWmnh4l)2B
zcAostid&%@?u^IVE06zyEG3zImlER#ii4L`mAb_^d&u`na~Fvh1nU+F7}E~b|FCYn
zREZY_GfyS^7i8Z4*TrSiZe&*tMyCEGT+j)=*V8(wFfPiy4k~fe&S|_R>%w4r;U@SZ
z7DB#8V7cwB4J8!syFT09-LM7aNZ%w@E2yOJpd@UZW0b5c_%rfw6@QqH6Zhg?dKKkx
zHGP*XujoR)HdoAQ93e%|i(}2yTxOxJJZ7f?T5#@}h74!1&8=<M2O;hId{dW4H@YWM
z%ia1BgjDQeq4gj2zfsXVJ$q6gBs5zqul>O;o0_?oOV71JJKT07VnV~$QhJ=gK*DeI
zi&@3qF1VEsPh!~~FRt6nTSV3yZ#rS6I7Z5#uN!a7Wab-h$~l>Y$>>P0=RnngJAg6m
z)f5;6?X;oGNRR`J{`50=ngH=uI<a}6zlBN&YygN+Qefy{sSui@eiVepEH#Z<s-RwB
zex(Q-exUgcFszjNqbQ`%pBC2oNCun8T44dTMpE-$<$f%;k!hj9G>=RVkm;KcC|D_E
zN(%j2$+VD6?=zUnq-Lfc7))~wrYf19Ak#Y{n36)*-IU>xEHXXEU@DWEnf}#anq@G}
zAk$O(foVbnQ&Q;9kf|?8zZa4CO&>=K4)9CubU{lif|fFfX)|B26RFhxI20pZ?{6+t
zY7gK^EC<wz8`iBGyw@9|q#2@c;?^*nqCCR7GAx-&BZZzIMKOtBgRO)_(3l1y5UJFD
z9TMSNAp*MjTEWID!u|oW-jWRwk|@G2DZ;U*ts;;@uVV}IC7DFv&-a8x=x>NXq*A*+
zB*Mo+1k9Lf?J^^Cg}i8pkZ6d&kwwF=D8d5~B9KCl?IA*<Sp+ZBKnp}FwaY>xye~w+
z+wZmFGDE8tREoeUpX#kv<CVHyAckKr#o4;x`KF=sMo;nMhXZ`4A*#Cx2kGW?u!d}g
z)LpQOhS`+5>33GekwSM*h*Vq!1f-8yW$27BHW@+!KZhsW@tLhsWSNUuYt67j<}VyK
zTCvXjpJ4uui2O;Rf9Q(LzuL%utdT!qrS`E<{!bYB=Q4kpi#h-MjQrcdonB-!^Z%VF
z|3Yi|Nuj%$zyA>|TCLS<C?J?J!ded#45n=&=2`K^LBk7VS{A{S6#5ZWFvS|yT787U
zR3<e``b&dpqrsHxAq{^c)8P?JNuj?inI@Cz=OlonjIc_20!x8Z|56e2ta!_%;Y~7K
z^^~>vq|kFDQ_L~f>T?aIGO1b8&kUyZ2Gd0d8{Q|=84*lLp?6&m87z%)De6%EVhIfE
zENTQTfk>tH8VnYsOS%qET6P}GMK;!QiCer-u7x57V-ti8YbnC61=ezrLa(9-{zpn#
zu2(4|n98K)a^)N4su4`5*sB{hk>jEWj*QlSDLGb<<E;iqna<4dD!~!&M)F#P(uQxz
z&ShbTo89#<#z!isjS33->^vrbf{4T&GiL-9^b(%5?Az%Oz~7whZcuVTu#nm2HSA-y
zoBU?Ic<hCGA+z;AQq3Yf%RImcDa?#E8jNzuNNqSuMp^`;TzxnhVXTCkrpQY*cp;IQ
z*INd!Eb_`|7|2n;wGq6s^bfKr)cdSY@;X5R$c02^UcWbZrIA-kLlSxIn{O>(n*ONd
zB`O-cmKwZ}$js{jgI6+nB{f_`UjH4zD_Or%@}j9zuBiquBr@~L2QS|p$N*Qpz@0Ym
z=g{?eJ}DS~u1P;tB$VSuYubsF1DCOKS(*$uy3uzdTTE}}uIAjrytZVy&fzzRT!ujw
zZ}#BD5}g!71_9nHVjc`Cwdufw&jHDipfMl}oYRBa-UqePgL-%$)UiFN2lhdIh-$E$
z>-q_pXmaa4s5|?h{;CJ{jy|YA>OuW^AJk1fsMqyD{dN!P_xhmzTMz0beNg|g2ldN+
zQ2TpO|Gp3EnjX~i`k=1tLH%$a)YE!U&+LObzXx?uAJpz1)KmJPzOD!L^?gu}>_N>7
zj-pxYMLno5?}Iw22X%5E)Tj2K=3#WRB=k;b^9%wf;g?MJsg!ooUExv|B+<*Bz%=a^
zU`!Ld_x4KyF$0`Pg4cjC5p<dfE+xS>D}nuhiNh%*I0G0HhXrN={c#Y?wi0xj30g?7
z%}Oxopec)2NRS1LDT}ww1a%~M*-F446{H?SF{hGX6fh<Z|7|81M}lXp1Yep7E+j#V
zmEg3)rYvGeaNJ7ZFcY-bf#6rbm}ICj6Radbv6bKpGr^xpu*OPo>JgI+kC5P~m0*mS
z;C2#R3XDmHr_BVTNl;@YSYakOn*@uk1Ut<HM{7awiIpI+%OvDx5}dRWj5ZVeiv+2_
zn1q~WCU}knQ>+AY4FdL^dKC%EiNWfG(hoh6*32<uHQI~0*Y@0D?xH-uHa9WR2^;Gy
z7p<Ixp_9A>=1mMCmixIn@Xcc)gN~$-obq;Ec4qp1FCvj;O+=PT?V}O>sK;_)syW;6
z-89NmE~bJxvtQJI59Wbi1GDm82qNWrIwaTiQZD9R(IfYmh}>}(VuIDTm?Q|yMVz6#
z!XB*NfhcDEDYbX^!D>>NRc;Sf_x50w(FdzvhFP_9KxZDq)zYzAr8=b#RtGsHm&Vc7
zgVks7tX5WAKpw?N0&Bvo8hfyMvIncb_rdD9Fsr3KSY6kH)$Bf4-5qAtW5!f_kQHGW
zq~-L%DkIFQriWCo_F#2hAFNIZvWlFt)E4#NvmX=dXgc4DI8l8#U|}_bvwHAZ+XtUF
z!hEu<d^WPA78U)m4?cBaKGtbUt=5Cj%s%+!hxsI%<?&=&Wf|WGk95gH{JL$&D1GB2
z@EP?#2$yA@;QVb4keu?su=Go7q~EG%Gcq--zV6d9d5j+SX^HN4Cv%9L8?|My?!M8N
zF-Q36ySea;u)qj!HNqQ>@Op$RujBOYdln_GBM4ps?tzoLL~1&{ELOcqbftC>Sy(DD
z6`evfqZ%qZjKJqu-j*3wF~c)M8NNd>5uo8Y(t2e0CnU9I_$mBuv<z3$C!^#gp$y}h
z!zG~%vk6`p&M>)0hL`rpkljqQ4DT^AJor<vgwM=I4w<10KP5OSoFSJ)ttwrPq}CE%
z+DC?GB5vjAP#;RCCi~DoJuI{sxQCKl!^%B{Sr<|g0Zt(J+pr|I9+Ld3ha~^&91OZX
zdLo#^B8U(u4`uiz!F$3PcJ^>}4M=J&`rT+O(R4kD4G9^}V=9Dv`GsM%5mp#sDZ-Vb
zsDS~L;9t9gpi1yffJjC2uJ0kx<z!(oZ{Fe=O`xspKOlg(3=!ep<5Hj>5K41@{2}Cc
z>z*LkLU0m5AwcA6RqYa{x8%OCkK7+Na(~ds^*$q<ZiGcfn2#_tu2P<wKs1IwNid|q
z3F`>=3L`F=aqZT)GmN+t#;vx-9X=*`wlZ$9HExR$*T}f}*0=^EZW-e$tZ^?Jaf=x@
z#Txgx5x0nO8P>RajJSo2<FAIB%6@|pSHn1)HEyI4SIW5Wth0|{MqDA|TCH)%j>@v<
zG46G1+%_XlW!yq*+$tk3jd31p+$%<02IKOqaZeg?S&SQFjhku2<uWeW8aK&^YiC@%
zHSTI7ZZqR_>+J7rBd(rtA6ny18dHUNj9ZGh@QD!9dDOYP8Sm0Q)IPysbez%>FUPyI
zpV(DaAWCd%B9s?FkF3VVosT#9@dD`A1%=yRB`@5?J<UJrwg=^fyF7b5RNSYGtq<d#
zW>@A~rS4|I6q}m{bJmT#2>P=JtjtMaW&Rgrk2^oaCD4h2d7_p1B?faUnUb?|<yFwt
z5zI+pWj@}-e2`#1$jbbv7%aSJ&oY?P0_1(rzldN?3b((+q?Q0Ef|5sGGws4lHFX~d
z=C9k+;5k+>W2{wkuRR%3sl^A<aZrD+9(O-qu=o*|?HDic&l!>a6%OG0<|L{y-=rU2
zk%iRyKRnc9|Im-r%geQ9f0?5l!ikQ{K*EoBqaOQr0C}wy-$!NGiJpgHyAie-VWSZ)
zHNtu$Txf(fMp$8lg$TuQU%se+cttWmeJBLuIyp`^XO()Z2jK+s7T_1Af90)C0MN*^
zI1lTsjtBHNKt5hXu&{2OI7q83ctQ|J(eZ8rPaLPQEb{J+t5^#>p@0MIF8}q|9z(Q4
z`&G8~K)gdMNXXXi#_fAJzbs_%R;OSGG6o!1G=c@~ow}#{rR;hOgoUg_YvsG`o?nLB
zkFRz<Nt9MiKJdO6@6oiXA3&d2+sHjJo97}E{rp-%zfM`F$wyw+;_XVS+KCi{f+_65
z{ZU+?wVDZ31t+zQN0A_@>N_Mz2qw75INxKU@0nsdO9q98O2!?8EFSxzCFq?OGX`g`
zaXOzd$@)p&KZldN;?Aa~Sr}B}r5mHtd<mHwl=-{~l?V4nSKwC`3jlOCkIqX_<~s#`
zwR3dgAZ7j~9IF{hm0JuKRT>zWy1tmUFn2Y*k9o&{cYfe?8XiCykF0U58p8n#mv4X;
z7;?FbTrj%6tqP&(c4MXXZ>WQ1kc_(x;)df|<b-hrgTOPaa0@w)(?lhIvIqI^`y!v)
zgM5A;<Q)&=6j1m!<!?>hAP+k&L~#Qj0%@U2GCKSkl!jFIkSY#H3)PJgR8CS|6G`>B
zh3W_WJW?407V7pSs|~e)#{=O<J^Z*8kAW$u?x6iz`j-sMGoth_$((|~OD7?e{mZ8N
ztX%*ptX{ev(ghu9ul-9usD$>BXfRJ8bMZq1RT0cdVP!tf#Jt!3<qM+=uo=v|keTdX
zk|LOs!pi)6Py{9KwSRflVBRT8fKv!$E`D#|-3qH(NMY??G!ye)`<I6W^VjX|X!Pd(
zr4b23{mWm<e@_3BGqbP$<!fZX59=TgPE7j$8a(Mkk&wP7BiJFt5fSt)vP1Y1g#VZN
zgNs-t|0n$cg{An^75f7gL;n|Kg{V>b1CogTfDf`iDD1005cIM?kSW+76!y^{2rB8_
z!wJ%&_Xh@j&lI9RND=lcOGdAs!Tx|Xr+1#qGNX<{{lSkIV}<&IH_HCg{(wrA{lS?t
zXbw;_J^3P<!rayL7E$_xtC4l6KL~UA80Zj}N#p`YH35cfb?`m<gHgx{{Q-lY(;t}0
ze@1^`CZEgJVCijo^am#Li2fkm@Xl$Flb*f?#SNSj>BCDSsBR<G<VdQQBB(}@>d8o|
z&n;9x=%<saiAdQA{K#sPk2}JTo5PPA@EE;6Xoc^>rM>bJa{Ppi%b0Vp4wo68Zz(;W
z_~n_Vd+BQ)!I=X&;gA$}m`2u5LA;<_jqDIq+yh&8CNxLpiw)-WWG*^{sS(U2g_-$C
z6Z6OpK`{S8FvY5(!F(Z^i{G9Z5W!qhn3;bDilF3?9fDxK++aS>V9wrNb_lQCV^xc!
zFf*?<F^}vJ1aqHY`hvY0?a}BEzTs`2ODK2oVNN){(Vw2qd<Fz(8<%soF>RsC_o#Wk
zac!v}>Sexhl6t2n;w8DpBtQYRr4u-bCDDjZ;1Faoy>EstL{1RefXDmva(?g*0L~Bg
zV1B^&UNL2OPR1LTg+<N}&P2(0w}ZIx+K;zHIX}?uz#FSG;+<Lv4(>nFV}6hbj$-n{
zl2BEl{@`yY$G_DdsQ-3<p#Ho4f&Jg^5A0Rj`SZK1BV{nI_aeYE%mh>d)drA9V4eY#
za)oXY0IIN>G4&#*)?UMU04%17c#mMA0XG`)^#<H#z)Jy>S-W5+4+k>6Zp615^j!wL
z*?<#R)8MZgaFPLc8T?XMRXAkC^RMKWX2d5NaFzil8*r`xrx|de0cQcGd=-GnFPE_F
z9uLdu(#NL&gI6l?GR0o~=@L$cLe`Aw(g(`?8Cbe?bgyxaF2!^yR{)$NVZwB!09u@*
zt1)5X?@b1+Om>h5!A)S;K|a7{GIw<=nys~iWag(a^D{vn>L6*WlFL8MT++z}4Lg;s
z(9$RM=pa`kCzPDQ&*>n|<UgZ>G?UL|rCEgO(LtKXBRa?nj0saC{TWJDfZ_%QMNXLV
zBB(M+H7=6ssR*heq<SclYHbA7o~fXEgGkvcZkQH&Tor!&SNQRDJQ^Kj4T^4lKOFDp
zpu37PUuqXU8IA`dl-=XY$lAOfM+)n#qXI<;b~2IOqgam%&X?K@=B;Edy2tDY=A^JP
zr+$LWBfCezJUCx!G?*_VbJ0EO#a78lVP*afvJWzk>>dU4;CyMR!F(~9i}{i-f;lOy
z^QC+f^T_T|tj9fU%$K-BXr3=EM8eQ~sq(I$GhaHlh((RI9`^z=;3wQwa=ncHN&l`8
zd70-+pAhl?+I;B{wf289U!t%SU+$(F^Q9+HYIFa1v1vW5cE0FvxPHb5&X?vMWmh5k
zuh0@27Q||(8sGd{L2sQeT`LmoKpvqLv<vY{mJ|H2ekS-v&VJN?yMI)x#C%B?CKfSY
zV$JF7IIQ!f2B1UprMU(F>3oSwm5zKTcnLMb)9Iy9FHhV;Fl%=ZT0iR#@=$jW=JJx6
z%Z20ur+GFE+3JCNbO#5K6S@NiKc_n|lmCqFz)b!#x&sq=M0XIJFLgpr&X;mg-01VA
zAl0R$iaK8kQpJ-h>U>F3{h)8V6I7=8QUe>U>>1t)KQ0bGzKBQ5d}$eM*gXE-j}Z)a
zC%X45x8Z`5!OqzCfO3wn3Bip5o)>~w3wU7&ZWZvN5d5Kl>qBsxfES10PX)X*1aB7b
zvJm{WfRU>qPrHDTivfQpU~n~HUBKXCz`F&ER0iBBU{D$GK>-76z+D2yw*j9396#Ow
z<8d!_qpDsz6|O|7`xpz6z8KzLf~i6+-G?pC`P@M*2H#!c`vkt@pVMM+ZCAWH_ZSz0
ztG;dyyC>`Y2BqTnxW&t|xbq3`N#_*5aso+k+ZhJGL*dxnzG_@rSB0BQZpI++3*4#3
z-;6oVJBNE*TkJVkTtcn?Myb1w&3G=11#^;ca@^|62}+*ocfKfw*M+#)M#Y63;3mhI
z=iPxM!C26Fc#ATAFMO|~`-@CSq}08I!mGZ}9k@T;#+N<Yaam12)gRxeUMmK_b+3>L
zsc|P3UYk(;Zo3mq@RJSt*COVTg!IG9IX+f>c;!I<+tpC0yfHmIvEr+!wI4tme3&;2
z=~jxK>ov$Iej<t&7Na;RH~PnX=W6<)Urr2mwQvO&<?*)T$D_XE&qg)k=SlIi=pGLi
z!T3|9e4c=3{A`qXP8K(u88b+}q`E#I9WP|ZrF<_65pff;V#L#*y@q&K_ebj7=H6oG
zTVnC+De;vl{MC$<0H-6o6Fyu*f#<JcrFRE-_|T>8UdW%PFy}VL;HPa>=dNLfrr5pu
zdBNOd)DYGGE_D=~4|nGoqRQxDkhu_gG!<qmC=A9dlh&csJp^Zf8~KHBIocRIIzjj7
z>=dPrb2oP}Zb9cTAwl)!CFZ2=bv1SOQ<r5f*pb2iE@cTW-0BM7YQL8KSF(N>xpoXj
zOx&zfU31q^#vJW3yWCs*^0&y4cOi&-?f)UPs+HoF!$G*7<y+i^H}q7Ob~8*R7nyoa
zMLQS?i#jzieG~7$^Wh$YF*o5(kYR|ez~AhfKtafhoAc%#OM=u@UxqHlceB97PS~To
zRiU3dX0M!t`#iA34UmP`p1V`^pe>&!%+x5cl=u=#tfi=CDWybMzwn%iw&l60)8%{n
zOj#DsO<hDotWh@+n?jBtp@@kPkyKS{Ln6L<D-4jcK3d}o0B|uwskI|!3ryyfIQQxg
zlK^J<7Fr^^f?(UcNiQ;Do=42ev9UPjT8^(3H^b7FKZhs15Ckh9iVH^E5{kGU5&C}E
zmc9pHfm(UMhPmXB*NaT@Klobl0+W1zCw(ZAaDUabKZE;Quo&lp*WyK-CyWN36}rpG
z+8F3Avc$C4tu^WW2B0~>bP@zwe`h4#0*cGQ6<#W?lHiS=7ei+`7&EX{xo}#b9U*@}
zC>gg&YTI1i?@x+r!{vTy75xN$&WXPV;C}?S(3jb9*Wzus9KSR^M{CN_wz|AKPp0D+
zF%~+s7KhgA8+XEYrTAg0CVBt5vgAhGBaT~!nw&AXyn#0;%4?lmxJJEdJw^lnPkY}2
zA60ShfARn$A{!}Md@qQasDx}1AVI**W5cd&AU6q!6?<5cP1w50#_Vnk)oOTXO=Fby
zT5DgowEx!H*EP58?L}%a2nf}xXsu#>rS;K`K}m~+_&opbcg~#MbM}GKzn^>W=l|(I
zcFz3fH^2GKZ)Se;n>ll4cy>BYO%>;BfqY%xzAT(qiGvx&aje_pqwBqW&$|oXi$L-5
zgoencaX7<v9=xy(w}agm`Hg}T?6kdoIA{Uq_j^5s@6`=>w}HXc+j+YCZ$Lt~Kgroi
zNVop|zP+~lzVLq5Q5-c@d9;F5qDt0>9&u)*!P~A6qd4!7!$&UmRXFXGCc4>QwEZ~4
zJ>G~W#OAF)<3R!BQ9TzjnD#Uf(dZ_`CR*{2HtB6}4mww2tV=zgtN@3AVp$fN*|s?g
zMpRw@br{{2<a=>?#m?~w?(K#i!I^$IVRhKH=_ypJdz-FYUQxvB_qzT$UR<43ID*64
zCIAsfQ{jNNN4yn(wr%`1#yxMvpWJ<qdTdoMxN+?JS9ZE94&cPmyhq$N*8vCap?n^f
zxAP9KuLD=VwE;ud!l77$=q@{{-0a@|0xy*bj!&CjSAm02pRjEngVMsa=DZHnTlkjg
z%(lKU_fg!$@(1_MH*t0))&OH1ez<gUDf)C&+#G?k6mb(hPR}0^=c6P>sFR6;p8m?n
zJuE78XY_1@M!%pxila2*X>Xw7z|a+)L6n7jwB3A)u1(OQjVwyoH*kxUF^<qyeY*jQ
zZD2J<wG9tb-z3LBjd-{AXolEdSz<Y=**4w^ep|OB)45H@q}7mSyz0<nji^s~26mXo
ztL4_{)4+{sykaqC6GJCVZu{QS$E&3Sxv$a&@nkK@8~GupP2va=oIQG)uH5^c;5(<j
zM`H)sa&Zg_>jEPYMd2L~rU)lD`2rg|1b`je2~>&V*|1j^S`x~?rJ&&0Gsx!akWF3T
zTaM?qc(L6n>&3`a&p_mCx!Mb@-byZ{e9rI;Ja8_ldj@`aHhsZEtU4Kmf+7li9mOtC
zK46QC2r3j7QRwR^AFyX`*u~X-4~{av8Lv;oEI-eEf8xx+KjF+lC|Ysm-~=A;mycse
z@&rpfPtamb;R1QIJ>8m;&%&CL&#<O&sX^SD(qcE%%XU{caGx$hoF^NPKNMSyc+u`e
zRV3SDi!HHwyW`Y6p=6R%Ox9E2PB%loMs8fjd<$Elk(&{P_V*ie`+LF~jrO<4(*6#`
zent#a`+GCF{o&N5)fk#_+9CuoA@(Bhy?v$LzUyI`aH9*2WHeN7)|k5wcj0c@U0@p7
z^>b|BS%!f(f`-oVRGi1$)L4(s?ZMf;VkE5tCAT=yW~sbI6UHpAe0;nc_tL){d*}PG
zABZwKDV%rtKyEK`f|{L(lLwR0*`*#yM)&m9;@IDzh{MxA-_wuxPY}4b|0Ep4jP~50
z`)^-Ay6`)deYszkkA2g=j`In<6}!4#L2>d}9Gq5boKM@I^&lE!;yjz2njK=kv_s67
zc8K}X4$hbM;*uz>Ir=hAMp||MB}lT(iJvW)jMc{Hj*~cdtQ|C_YLhVO%dg!Q4|3*M
z8=p8%;>58wK5v}Fd1LMVI5MY=wIHC+`X+JKSR0=-P8wq{K4+Z7Ib$tg5*d@l{963T
zAZLsdWhi$N3a|#p7P_(Q<VA1BIX~!ghih#46ZT)sdzkWTviITFoBcZio3djFtj<1I
zX)Ac@u<gqIwt`W3BZqCP|A32_u6zW6U59Z|5)Pf3u|J=8DLpA@Jk8YJY#i$d8n5E_
zhV1>IfrM8cvK9Od9?r&E{T4uf<+H7Tl%4=37!nI+jNlEGI5HRKW8t9M1v4MdvaNm$
zhZ2rtRobq=VOI-gJeH5!o3cS9za{$seuLRB*a|@4%HP=v-qy)qxd&OGFjsy6x*syp
z6N1Cv0=*_1%L!m=H#lI{`)mb=;T=Va$HDrQ&mwRDC0zAKrhC-H56<IVF!LqwgNM#X
z$qz54n(-X?S)Dx&av)D~e}wlk$URDOWNbgl;qb2oGafOqA%$%#c$z4-f*0XEhh)Ep
z|0n!G{;Li$(H~4aP~0;g1`iK`hkbD#c7g|ENnnkr+)JQ*&GY-?Ywx9K@3`HtFWb!n
z&iksMBx=D$6^?kT#c3H;*v(jLhYy8Pi|IyH0sLb4=>4@g=%)&2e$+bQ<B*72H~cm5
z>*241zY2aQ{59^rjntWVT!%;YW;}M{ajX7_^A9%OD;{s@rPsHSMzy>D7LM=;(S}Ay
z(CFKEKo1EreH(}M&{leV8~Mfzn1Wc}#xXr4i1lq8*F%C>-$n((6CIqbc*1`ba6zRP
z82{`a|DX4d#QYFL8iuL)m78uH$6~<SJu|s6bGDabmk#U^TQ$!VkPAn1faAf~8mk+)
z<SY#DUfg}fyN+*u^G3fHYYDjUI|u6lSQ)Cc?fAk6vBQ^(S3Ebf*%I<$>`xN%MG4Vj
zPY6i#M#kejAU)LW+4L|bUanlX_Sj8>xIq}ZqP_i>%)-(4oTtD-j}+7Lj+}wt9NW!2
znH&Tb$2REc7vmZz5x!Or7a3un9zJJu9VU$*fa<6XQKOTPU-oD&esh>jUBz3GC9oPj
zSubEy_2`*A9oMv9;Y&G{x4Lkp>mCngx!S|9FAsYvUWiP!?YK0iAOr6}I14k1*yuB!
zzC)gV+otv7O3z6Y?Nf*az@6iX=?1Pidf0qb#K?UfSU0$#_b=bp{g<WxEd12-{x%)N
zd`=uP50!<fx_2N1{e`{JV$IFA`34B4uKz+0=EO0S7K;;Nz2<s5H{4)Zz*Bh0ZCmmO
zZVb5q8H4LnJsU7qYB$|fhzF$Z#YKfT<la2A{&d{4hpSzmj}DE_u*w5xcQZpsLrlP<
zUmwL}>aac;uNaLi@bq7%@9zmC+u9hXX@z@`?f%%$-6GFv?hPM^JWKJwJaMlqvqXJ!
zjRi}WcJBZzx9zJ=;FT|2lEBe^bFn_b)s{`W3OI+fnsEv1)VP0OMI~+^_ypy31fTN#
zcgt&|p65R%uk)erk6&IG0^?V>>|r+(Ckg%rPnfi|%*q?RmY+>=rak9=HXQ87&cnoR
zGNv$-N6$0j3e0iS5I1e~toX-4H!WxTyTyg~S*`?dDHlreV<Cbsv-Ce4?2d(`qcO~@
zuxNQb2I^{YpBS$?xfweIF}tcptLcdez<L5;l0Q?;B|Xdp-iz))W0`KU`4f#Lb-e`T
ze@qv_gT*Lw?7mub;l4eb^UTI_S|v^$%jXZ;rAw(B@S}-i6`&3)%;NA&&t6vn578XW
zWgNJcXEK(fGVn5%EB{yx>GTXNgT1c}W<{6bQk;l~Q;mM(JDikmb3J5&rg!!>6zwLX
z@7QGD;ppCLtJ||Pb}lYmwf$;b%)0*=J5mG2wc3KZzC)vzc=}yJWcTw}NO#2(Q9!Uf
zW%}MUdMTGazgxw*6IR<S@?x!ur9Iv(Qf97Nz1j;b2(!%ydfm@RRPz^E&3Um~Q9C0&
zC{H}bcE|OsInf9{$5`yug9faz?5Egta5}XZ-;ccK9;nXq;+Eje6GTH_jnx<^9xga~
z``co5{ZYH8zXN*r6?Y--2K}?H_m0GhAQuP|*L}Q4ao+T4JWzyv2W+?5KnZ&kZao1$
z8V=NXUEyDF-0MMBBo?cobl-@*3FbiHOnj)Lt4`Y;6|CKa`kdr=6IA-nBVU*lg3EPw
zXiMAXiv`_7u{S`1Ti3uNHKgEim)Evrn944nXLwRRnYhdX6>|9w*cRFm_a%<}gn>=h
z7by7Jy8jxG{Q~}=^ZVyB2h3wVg}Al@I=Dzu_H2L|jq;y{w;_0@2yox!4c8)g^H9&{
z(1q~5(&!aN0j5WS;FxXX<VyC4hw+t?$N^0nH*!KqPYVgf9!C=b=g?WYe9=y&d~cW!
z!lhUf;D=CGAVc&k3=SCNA=acO2Ww;Wb^L&v(U)))5k{gR1U<Hzx4gER*ytl@(E?_3
zy@^#uF!m;>jsC*A9u9lw2uvOiF4tny{X$*;1;U=Oj%Qpk7d5dGjJ)VB{Da$8`+Ka|
z#&YzcoNIKsGZMoUY=roI3n<lO&p_RD$I$2`=8xz6L0k)eBJzxWZ}U)H-@(WkR0tt#
zZ+x}x($Mrs9R<1jG754xFL{-9HBTQefA22bCF*J;L=9P<hTWSft@Vb9j;AnYP6SDB
z-y34d$ZS}cCP)eU591;AyBiB+T-d-G`p(!gwp~=`z)Cx+h6C+wc=rHZ<>`Bx@AEpm
z1MN~*dFm=&i_{nukM}F&n*)5D;U}GfVM9|{TY%J5yP&Crn!@d{f5xp)F)tFiH5WA1
zQEDp8y3Bfs5TV`Ux(k}>Xzb{s|0YfK(&A%jsv*WwQze?imdce}J9<VRKSC@VtQKhf
zbND*ul<IujDLEcoZai?kJ@!|W!o9az3>msngT<WB;^UmXvC~Atmt#|a5(^NoZ&z$5
znhXcG3K6+C_8<R>^ZOk`rpkuB_fg@#eOCNL`>5~xgBbnwfo`H_Z*4xhDQ<L+Tx+}a
zLM(#fnq}jnHFw{_S+OsRCcGE7L@ghfbzfayYd-J%#eh6*R_yPWVgT-I&GTrF^~KEg
zG|c%M7if=G@6EjrB$d8BJKsDvYtNG(0P*JM`(ih}JaQY_BnH_y17IYNjEsa4(f9Jv
zmp5r2{suzVw)4lf{MoEO@+zbI#}NVs%$_H`Gk)th;+gTJZS&6nz-w^aU4~S)<%hCf
z<Qs!cuVIwigR4H?%)#p+ZBOJxw|-MuWV)?j!u_W!N;P(7-1oNleRT)3=k*$0t1ln8
z<Y#XqA6$3wp8zWE!pFQ>AAqzN_H4_a$a)lOb8Z~Gj?=+Gssp5~`i9m6{>Jb(jKBTh
zV+dC5SoHNeYS_o!6@RmB=8UPPZ{+Bco8I|Qmh#D}0q;fGeUISMH7K)O+}(3AZa9ma
ziM!1rCsCR#I)s@1ann1wU32k*_rA)36<Ka^%L?()UiNZ8QTA5yrd5dU+P5)RdLPSa
zW1Aj&|K_J|8t&VJ+eM;Ny?E&&7aOnE&m(_u9QOH+5Am~o4wBw~QXbw)qH1yfE!u1R
zJne1k7W))aliSv74Wqy4iV5E=<C-s;OwdiSQ*jn2UL-|mRN^MO1DO2b#WOZHV1PUt
zyAG{u^bB^C+3Xtt=j}M<#7Yd?M`KqZPUjX@pWd7DfMo0+X@m-l{nu1M-S*WXT*+vy
zpG1Ff<9lid1+UQDeR!Wdd6uUFm;6K~cwq5j^?voN{O!|)%wFdCjIWj>Tf9J>oJ3C_
zN@HIIS1=ac{T|q6dG-mmn;oF-&f2HnO5?$s)I86nC|=|oal;%ehsju_!mbt%F8;_H
z*^iurJyeG&E3S><`|_RRxHNH(exkUoX2hMf$M%&WHsqZni)~-snYS}>%{NBL7|ixx
zgNhQl#02O4=y$Ue#tn)OkWqZOiEIQ@b$uUt3*VsKwNpcOXs^TXyR!~NCer5b<Jf>r
zA2b)<iv0^Jf7!qdzs`-nOkWP6c>2zoRmbTnmg0Zn>BCpGQ1wE5)Mbm12G#g5JXkfG
zABR~&aCZeX>rMn>FCb>*v*>ao!(_96!Zmevv{R6A_n-V7KKDB@2zhf6p71ppC$@YP
zbe6!}g&;dELrN3hhZhXuoJ?+&SQC2>86fE_6Ir{3EYMlc0=*~p0662fuBbiJq{3nY
zTTaK?3`CrJN6&Q9jO`A5{f&nEOL5v(>^jJ4q=gU+7X1@$^(Kn=0G~}o{H2ULMiGy7
z@J)Jl^i#lk!DH;;d~jiv&~q$u|AcH4S064lb7hs#H&7VX;a8B%n2(G50J$HZZ;Jb#
z6<1yATRXhKf}(WcWC7bLDCwnQoW!TXIqm^mD}*;G;=PS{xueX(W(W>SvD5Tas$Hal
z>8}r3CwTCd9mzbnJP&`c`}j`iw{taZ?}YCxvvUzt;Y|oz;p?{@=x<y2;p!H^c2rS`
zolX%n@2MM@h21oF>aSzC?(%x<CS0m2Om}=SK~g4D#dh;Tym#?dyk^_{n235{R<H1R
zs;?mX5V$8jf}yBLZw~0;wR_}ZYB@g46k6^9ak?;Z+3jO}NH{tZO=IMT7>V2iiRgWZ
z7K$(C7sRn`7Z5h9TbXbTYXW;V7_jo{2Ct0iO^I2}7-X@PF|Y>@zDus-2MW6>pm7M4
zO1D!MW8KNb6`x92`kL+Ll{6bFUcq_eA|qVs$aM8P1eyPkWL8NsGeCwy;EYm8<6I#P
zajNj0;_HvtjYu+jE=6&g5QU6m#p-gnTF^A)z(a?xVr@bTtg%Yqd*F;<Fl0VYmsdm@
z#pg`r&3E+TyUDk6$8r|`oie@nZpn9Ga=iFJGgG@6e2L;8LGjtJ3(ecnC|2x`;ov3c
zepk}v5XaVW3h1Vl2i|JlIcppwV;xAM%j5H0AOk@WH-<P}9(96dLLSp&OW9ITa3Gpl
zeG9A-42I0$2_|`n%`CC!$C+NtTA)X<-3F_&I2nEowwLUkumE}igzym!J6?SWQT6U-
zCHoEHY}ieWAm3kJ4XY`!&|uqay*nHS8~_!1l-|e>@MFdP86fvVf9fkzR_wb|u(uuq
zyFUf{Uyp%(V+!^y$H2Ze1$+H5u(<`+TFT%tu&+qLR*!*wc?x#JF|Ze<V7rci%}O;b
z?2HHn^2=G=gcrNrfr0K>KuicPyvOb&5!#9H2p}ee?c5G$Ab5#zFCZp_pMPLNxR?lA
z0Wl$*V$Na`5i~0glgtPs#XuOhB0OnE*hhp1txWDUBm9sE|7k_|j=2=K5MjVtim#dx
z{6yFU$jAx`AIcRYs+0Xmf1HcQkp(geT`Xld+WZI=xur|7i)ZR2{8WGB4fe`(dVKR2
zQiwft3}Wn!Bi~LCvuef7#VN$@ItH-~h#Q5~zH3Ar*hu!$jA)xVYqS=3?J<bf5gR0b
zK=lTHpB_#fCpQ0k3SF8|nZy`E+-N;1zLB74)wG-Ur%=4-7!<#((-}Dk|G^a)e)K6<
z5YoN>a~zFMnk^c|NWU)ckq@LFKmt=2_%-RE6CFTw?1Un;9*Wb1@C}KO4?p%gh>p<4
zH6c(}k-`uu>>)y%89}!phZ%V<SxRSJY@SORrbcN!NtVi8>UfK-%vV#)IY#13ne&@?
zinFHY%zekPB9U|QsJ?jz2^@E}y4mzgaWhNo=}R!uu20I7R!wX>Vp2xVW7**8i>a4l
zkQCPmiX39aFnZ#}x;VG86sFPnAI;-l=8?-hpi>S)VwkeV*E8ey(PR2c>(f|7J*Teu
zLTO)Mmb&BDY<gp0Y}vpiuh53U(xR9I(tgCcW*Q6QN1#?$aWGOzGR3j;SXG>84FBD!
zz$RzOv1<+`#I@NEo`ffUw*FG=o5<ur7#A{l8-kvGgBwgLr_I{&D4A(S#N`8Jf*H&i
zUXDG*z72~8I0=g;x&$|zSN6}}&Si=FxfT)o%waIHVK)mKpV3*25$juz0!(}hMy+o>
zEU!f*LcFQi?wBa&di#(?tp6gk-?uTfH6h>wUNOQ6gwGLSB_N}hLXo5ja<ed=%&mEO
zLLGrO!Cy6!jz(m9iv%K{we*D2)U+4IIlBoP|7C}Nx=?Q~4?xU`_^G;0_+K*eJBXy{
zzXo5vzivc3+KO@uV>ebr1XOyStn@92F)BUJRO$DCpsCWoM*16ela5@liMQ9}^_!0h
zi^V21K<o>Tja`V(u-vwbumO`E*!r98$_Zy+ap-T@0LGURySA)oggQUmU;5x5(a0P_
zv4a<Y72IL{;6Xk70>Ze`{y_!HPl$dD!3WpiJ=+f2!Lbh!?As#%SRm;vpm=s+2gT{}
z^)bwNW7EM0_qS`0MaLXiKSR(If0@t%BRPQXYTXj=pW;;*UU0mRT?=)EuX%QUFahf>
zH%2`JFya&aSu}z07&jz1kUbY!*#A)kyergVXG@qEt^2WWQj1NJAl41g@(4SE%$al}
zV4h98?e2kUI~E&n`YzbUZ2JW)xOoflRnmBl*soER*sl>?<>@cOyL~P8b3r4r6zTD@
zPKW)Q4x4X=eMyI%hy5p`xRZF-80e3lN(}7Z*t9!;^hEql7yDE2;v>;zAu_x#Ozhtj
zJKqPc%hQ(zupqE^$#k%fH#BMi)|ZzQ08mZ<rdzSsP+J}<3(v)PHHwurEFU=W>Zrie
ze?>8NySsP3i`6ge`Fy5(3S0Pu>#2s8IJxB$i@_T_eank0_wv=DQn`1Don-OSDpJTz
zoKKnc>!yWY&WfJn?l0Y&dzZk%<)9T0bzuv8Vk_r^Q_+}q%)-xhED$L7FU3>ryz|A2
znfOlE!?Vc2Vtt5s9RbSt9Mqk4J@cdb!|}$~>sJX2=ImtdD`f5#<gd%$9AC)zKQq2H
zZTxJ;ubhPV+Ial&^0SvBhs^nlb!PEC#4K;q(z9qEiS=FfYaEX0tr)d!<{MBint%WG
zd3ESfg>U17EbeyVaW#0^n9pf0ms>0LbT{a?ZezT|3M{|<dzxwS!f+!s16EEejz-u-
z@Lrbha;Y!Gp*LbEUwj@b@oH#8u7$etu}>gxw8t|LMUQY{)WU}q@8bOhzO+DIBjFib
ze6AReI6y4+H`cMvqhu>@^i00c!|sF<ncL&QVRt#-jraI$t{>tG^Y^`3M?8f`BUnwt
zfr{_Oegve^9H6;(zLxFDX~%dic3T-uP%iMtd}wjhH+@>=N|*taEAaJXWs`T{{QV_%
zWrcfSb_{-_x9~McWakIy`3Li2(HRf|7&^yTZx^p8kJNyMHthxU<}7!`Xp}E}^RBuY
zJC{XXsPni!KfDV)2b73`M`%$D<5t&;p!v@d%jRJI(<8?&)BWjA-OrQ${>zHyG+)yh
z3Uox&pf6HXR$N@H)CYanHa7c$ekB|XH2YiZS1FND#22(H05(T~2UsVNqW(z4AGUyl
zfsSkJkuG1eAK;L$E9~#EB8Q{x_DIOCbo-klp{@lB)H=1r-|CA7BdV{n)8Ek|lg(bN
zcC-ZA{T<;zsKdT!k-at0;kN*lqM|v$z}h+BCjvI-Ea?igun-KWV4)-2>hCIAr*ww`
zEp|pOSP+h`Z3%P-TKwSx9l+lzU)XQI=n{#NfD|S;>Ff$M`@`W-SNN)yt4+*U;0g3_
zpq-qo4@DzZd~1YKAj{MhFQ}kD$F55brM1QsUX&?j(W11Zls1%a#v)bihz5hllvkI(
zCE6T;sFMW72GAI7FK7;Rgd=vZF5QLQp|zh|Y`@?VdpH8A>Q%WgvL4mvYxmn33t6MF
z__|a)qoxHi8d;(JD#qFuNofbz&U{@9_1ZXAsJlAU-Wl}w)GoH;-_^<$l?}^0%a+(@
z*c&B}_CVMk?TD=F^7~rQ<`ldBi1r!kimVH@LCE&Si|oaw;6nSd7DFs#L1hJ5C9tH^
z-_;(C_zSuj$1>}phzDh<M66M)II~wP1qB6Bpv)^%BZU{Jp%!1ncUAG#g%=d})E1&8
z6&u;7ZN+QRG{KH89YgpZEi0=W<ID)j*%vQZAStEvflsb>j+^5;U!*J0Bilio<4f$N
zpDfeI&gSa$l3poPRr)r1EP|KV%fS6{i2MH~qpKB<%jGI9sq!q5Pr{p5W&q}uDP=Wk
z<BC#sN#%+qE=xcyT~%r{#7<wAFBtR(J<EkgUFPo!H+z;_I|aL#h^P*Av<BLuT|TOg
z0@<#X_@Zt8+JHYOx{KZv1a=%I)J7OJ-qBX51dE#2`J1otcdf;E(E?2$EK=Lh8D}r9
z*5Orz7lZ>5f1$x-+TPOX_hTr#hCRll^9gB0TB*XqR6pp&MsG^1%Ftn<u~NFj|4aQq
zS6UR`$Dk#o)U2z$g{{GmFS1zg5NudiE8)QB{h`(Zk}bSwnXZ2p*(H?4DpdSE9Eveq
z7ZF@gxM-2UGo;9>Q6-L5Xzl*4HmE!)tK-n}pQy~PR;pY}Jj+zKtFp$`puo29FIr@3
zI|Vb%@iPk*)SDbw;w{w8(V(lqlqz+kHz4LlTBs~iT+3>Vyxlbo>PlBbqi6XtrKYl}
zvgyitm(tj@yrFW5ORZhrP*>Tc)OhM#%ZQ;mOwT2z=Tg&inHZcc77Z<nIJnhf6mWn?
zhboE#VOS_$&$6XzQ$uC7OR4u({>utiQ<F=nTE1dgO=ZKCU?Toll6WjlJeDcdZddiv
z#uaraX``#I%Ii`+HHl|$WtGdTG`c`|S+z@DwxX^IwNTyQszkXBUv*bD)L<}n=>V1<
z@^LAZ4b^VXN|zp}^)9b$!f375FZV2KQo@l?G|1l;{sa{abAe`6*9XE#!=rAeCVuTF
zREmJ}T-|id<ach?<BgQd%Gk=w`9nj`Wxx$zk1ugl`?_Gu27Dc`Guopafo31fQ()JZ
z`UL9GL_L2U-%(70O~TgDNg8-#63?JtkBt^|TTO6U^uscoohPD|JXyXJYhRC6hGcy2
zuv0*`NPnyJ2c@6B{Ekxc8!w|t3j50>f0Di-XPK=+_RN2T_iiq#?C4mu$RWl{HP8`=
z6g4-){-up*Ru)2!GUo*g_*<hJ`i(H83J`UP9Y2N2?Ah&|0sOWB10M99f<I}z#TA#s
ztkX%Xo|6l&FtQ7p*ZI2a7r}-XR8u%eGE@au7ck=%|JrEV>~Ns1!xs!E!jTqO?n*U=
zv2`eqh_#>JGD)%9+x_i!UrS3DCIj~KgTWrcq-zef_yv|~nH5k<(BbhWrB74L5Va%J
z;ivp`LB%V{+z=NmNDJ20TDNYW7{xM;eQ6UdSkQ=RR#T|K-`ri>6>5i?#{|g`o+Tja
z=XmK=`P%{=7S33baDyN7#c)K`=K_{&EP;P0Pvn5f0S0+<-KS*B^_Z+Pb1uCzTT%Xn
z4SA}j4dc1@G))_yggy%2jZGN#PixwAo1!@34<f#Jx~3IiU4PA`nzjahFLoy14S#5<
zrtOD6=+(45EaUFKLere^x2)E*9{9aJP2(AtgRR)e#c-RZDHEVm)`33!EdfodhrcGM
zX}$2B?Z}V*wZMm;->qr(iO6TYrmdonMYmq~`@gAa+u#>}4=?=TJMRU5*#EHRKJW*>
z^FP5K{IOqw9{l~k20i%0zX3gTX8SJCgFg(v7ykZ-fe(M|x4?%#_$crh|2yEH2tFPM
zKK&=KAOe4IKgtDv_$ibJavS@-rj^5A^9N1qgufsD-SCIu4>BJ9KKRNrkT?8%_yu^w
z(Rl#;!taIO1AhzrEqwFw9LfjZ4u3!V&KDqGYzDE%AYb^MZvY?ue)xC8ABH~&e+>RU
z_{u2qfnN-N9R48ue9#$!KOOZud=UA<AA1w{@RhfK4?iFNID9+&e5^bc!!LmEgztpE
z2L3Aez3@BXZ-sv={6Y9z;Sa&z27f>N{qTq3$Ka2_SJ0l6Lnt5U7msOLF~VEn_Ye*z
zZkB<c_4x)Rt0zC}%*iL@ZOJ+zpYXHrck47wy9mDF(&H=2EFchnx(#)UHm6L<ubnde
za@$GQ<@G9;opI?!rL$&|265<eK40$@Ii8TyYj-L+dcKNMhB(qvPSvyv&cj=Znlmbu
z$|?C(Q>Ir<u~$tgsG3q-Ii-Bbl%dI+PPlF2z=Yd#@5s3&Tclf!WaHR&G!1Ptck`L)
z($!3<w<Htk?nJs_>{XqMHj+D-HXZWIwj?B<zr$Z_iKZQbpZoo@toc<=DXy8~+%)kv
zgF_&}F<ovJ>cEXQfwqymJ1Lz}zOAYGEkwFo8#QeQ<eR(gY-@gcx|%7S7B)qG5v0ps
zrD=a*I^~@7>9%HB5>6A`vYyQCLa#s8;UIL)Oy~;MP0bYhCa5dnm~J8dhM<EUCj{|~
zB)kcKJ!>`XAql5fhQA>GikmfU2KY=<*Od^XxvGJrD3{>x`}iww(X`#nyvo!@AbSda
zAoup1J0xSoe;R)Wpc9{iZ^;MvOQsBFnJbHYv7T;)?tBlr)`D-yy4NZ*l4ZJAk!Sri
zn)U?ceRl?VZ?UFjy0ef^ETn1Qho8Iq<E5L0bUy{3Yr>lLjzxZzQwlaE+8*iO0J`N-
zO`B*3{js$4A%;m7VKW`?F3<m}J{IJb&rH{T&_7O0Nq3O>Va&*5f83oR-B@aVvmm$p
z?`qoV7&~(7&r6?QZ)&>LNH_jNncoW;(hXYm#x#`gAnI%X&onJc`ANre{}lAf@6)sw
z;ODNJm5z^^DM|f?>0Sl?@GtOH1nA}}=ciAXrv4+4*S7!Ew1u)?CF#7HDQheu7M#u^
zpGY^fUDNI-B;F5D|Lg}7ZWFzL<$M}=t1u?NCi?`@Mfm|Q-l1uy!Z(~KKlmu$3A&I=
zu03rzQu-#-EkyZ;9@4Z~mi8Fe-D%6gbP=T62HW6x)3KfgcWc^T$$aWqk)rd7#{5`M
zJ$p3mo0<4b8Y_%+XF)E*uwBG8V8@YP6Ve^nhyIs7zZ6*-`Q3(egRqhO>C>ex-z+2!
zqW{Dm(X<erbJw8$&^9WkOs|Bx&beK-foX_42>Qj3YMQC~MZ3sFzb!}uKMQbs8h9Dt
z)6>AcfcKsRn+|!ge&{jZ7QpWw)-;$ZamRYO0q|k)eFb?;m2X-(k=`$nF7GkOH%?8@
z=XV4o!6~2HU~7)U)?A0@+y~<FsX?!?$Y(O-=tTNcNEiHYJR_gKLwYyh%OJSi+zZn6
zbLhn!OQ&VJT-dN{V9S1=<NvCpbYd(T&Nh!l%$DgEBHe-K@um^^=MG&c>NBn*M7my)
z&OGjMto|MNAA3>L8b}EIi?PmJ|A;SsN&kwAD0Gx(F3MMs27VUc_B8M^z^4N~hXi00
z&}05yz)Al)0+7Ga24S{|aXg6lt%yH^?K`f&pf9MulE!iJ^)2AnpQveKu?F-=mgD<s
z*l#^AYg#tw<mMHa<YU!^#HT!81)X(&0>A0|RmOI97V5!{cFSajBVT2JPfr8)0-m1+
z-U4`D8u$%>D{0_&0zQuNoxx0HxtZ@zC4Mq|x1^s8e*y4x{Jbsklgl+3a!tq2EWpW+
zx&M-%g@BI%o~%!r0FMElOg~8ac~<>U15j`>=tB;EJhNTgk?oo?{q`K!6#GD~Yf8aw
z6WmjZZ<*+xQoiYgx+%`!lybBe)U=o<V77ocL!M~IzXTmS>RC2<QWH+g`SnI6XPW?f
z4Im$C@%M|F`1uO{R^v~&xA4nZ+qY-n=W-N+f$uT!6$A?4|BwOhpg$k}PZ=O_s*T4O
z_#^Zz>*62-JbwFB`0q1tA~e+5@Fz3V3lMk;Lz8e^ANVw$hR?;14GIHaHY%?%@cL<r
zvL1f5U=%nYsO({&5&LbJPbUM<Aiz!PH!z>Y0DlkFB4UP-<CR7}6~H9{LCuZ8c?N+W
z$rvG!U(0~VN6RV9xe#cIF!YUwHK3nGjavlTuOnySw466L^!6$_ZwNrKgerWX7=H>C
zgq3e3!k=*0;FpCs%9wpJ@D2loAi-w#9s?U>-~$GD<|Jb@1{$(;(pl)`ii!YtA>F_L
zzd>Q9B8H#CGG;epCWChp$YFq+uj6Q6LV#$K+1Gw+j(JALbTZ~G1eUPybXv~0l$<l}
z#1kv+Hw66*e^q>-v&!VNLNWmJf%!kmZqWhu7<iF^t3Vx1Kv@T0X~hrQr*Pv45bY4r
zV00=!6=3|(IAU{A8S^QyMgkN0a}EOA;R|;FfeJjm&r;k4`7zBQ2Ac6h75gCrVf^sp
zb>Vh`B-@N&^!G9#7=29!1Uv7`fXGzK*@ae6$@D=KO*lS~&OKmNS%8w9iyUrc;9K~K
z$9xw7b`s&VoUd#Ep*ylV5oBp-q$}6s?_`5L?xQB-Ut$cuMmZ0@aNLy5G<PuOVFY?~
zI64Se0lUhbjA_IVbHAE_RrvWleBszV86z@%76IlXT)a=52(rTQfh1WL<x>befD#Iq
z8JDe1;C|VV5#)A@k~1_>MFiX7Fn_*-ze@bQEg*Pchq;xFZ42^Q1%DXd^C~tFp{l85
z(=J5dV)!w}bR$5ZKr2M1ER>*DiU7H;A}5#Qhdn{Ws0c8hEzD;<0yE%0M=G2rGUjf^
zJT7BqG3OIecE)UF%y|efO^7j{l`;1+hKrX3KF+|1j5(bNbHEs5E@L2w0R1oV=N1_=
z$iN{P^Bx0Ks!TKEO#J)+0RnGk;6WMl69&#lV`t0}1};N@Y0jeE@&g&u#=wIJkn8P?
zepkkX7&8qWk1-RVkd@^KklJ;O3CNgw#(ZDK+{74u+s-uKWXwA<=8KH^6nYr5_!$GV
z1W0Y1fx8f38Y+I}Nd!19h#nD>F%Ck`L$$Fs^4M4ABQO<y5d+_lG2IOOI|Apxzkz99
zkuf(i<|qQpXE$T!pr4T1EsUu~fH9NV-maH1w=?Dr8Pmj=y)x!5#yl@$zQvd~@PjJS
z8`2>Jb|N6$AgY|TE~<V8%FKYM?XSpysNft_5@ST24#<G0#t&sc)Z87A7tw?e^T7rK
zLIzsS&0-wdm^BL^%JxDu24#Z`lrZp;3@l>c(kva#%>X|N`Ye3met`f)r8F|;Hwa9_
zpKwp2Td}`T_bN7Y1a|vxFfbE8^AIasDlgAcI2-Us#{SHJ!ONQr7`*VtlJvaf7`*HT
zhHx7sFRPjTg9wn9%NckIjHTh_d_*z(7)foEfx`^EECU}hun=U3b|SihvJ!!~Tz-iF
z<uZ*iI}q3fUpRx8cJfbNX5#NE2L1$w$`KImf0tb3I77}7gDL{Fdd0}F9S@X~a1;xx
zej6q1LEwk@6OQL<Ux=q_#(WyGzgLGt5(rRWB8IlbYevk!1DE0v)P8|Lp%J6o78MBc
z+ouD}<X-$-nTUTt$tj~sgK|;6i>R#GGVmV^bmNCHKVslU{9KBFaCnB(a-QkM__~bG
z7x9OV4`inbe_5R6orgc+8t}`QD;WBW479<|;(!$&9Sz#M3>)AQ#0tj;%1yArVo@N%
zRpXZ`Yz;#z@W+8mI4x(t<gnfV@!`MlcO%=<3COO+h`9!VbB&lg5Foo>CqsAQheo?_
zTF#$0g1MKo*|y1pi>;Vs8K`IA+xTPbRSfLFANBDX29Dy7O4CKjP60`>x`Y89RZgwW
z7NPtLen`L%wXKjZ#&F<J*mWo@C#mg5fU5T=46uetjqlZzlTbRwe1mDuMSy9($AA+7
z3ag3bUyUEK)6RUtGUmrb`vn5=e0Cu~YU3pRECS3&*wbgCEi&eBOfv@org@uzdIXqH
z3w$MrALbKfKAU9B`$XF%12v5K0|HMl4?1ZFa$I8${V1N8lQZuD27Zh`2KF(q6Mt+S
zk1@d6#`S!l)B8awp33p}B0<;S?`wScr~9V_{waZfO5mRo_}?pm!I@6MT=5Luub2LQ
z>2Hzo0tr8c-cR<Y%7MjzpOqA@km1Ra(5+`V#S=#rx~BJ>!arU7V%VAO6Hm0D=<MQG
zAGi4!lkptD=s4QZ{ryAnx<$D=+sUA?GU$&@a?(?1I@673JL$#EoPU)BFk2E1b0gts
z8kil%P~8>l{9(8j?p{flgDIUc&T?F)8$Q)3d;>l<O%EG<kBL~A^U^o+|4@H@q!deh
zO@<9Vu$6?bmz2$)a>|tdJ>IyKe{yaQ%X)oA`mam>i1a6(t@C-B^e>S9Jn7d;f2H(W
zrQa?6Tcm%N^zV`WucZI5^q-Ob>(W0W{fXzu{H1?^^yf*xR{ATY-zxoX>E9y#yQF`Q
z^nWG&ho%3F^k0|$5$R7nSLQGM3#30!`nA$uDg9RIcT4{k>E9*&d!+v>=|3#}XQcnS
z^p8k?qFv@M{R^Z&Px`geUn%`o>32*27U|z5{d=VUE9pNh{b!{Ay7Z4of1=bYr%C?;
z=_^TYNp<xCd%=pe__8NzFDoi9Dw*x5h>D=2PzNY7wmInQXtQ_Yg9UyvR$4T-sCagn
zj#_eKq$?VZtS_+_M0{=GISnqatFqChmd-7ngD<Ci!9sfhKelPHBc<27vQAH4h{cNV
z`u2!#E&NEA?yobRuv5|B)u|M9gd+Z;$|}!nkd~pgj%d-^Xdu`!8+%%d7+U8GuTzRz
z)^~uO?qiR$h>;9i0xIIV{6QZR$fwR=L@DA{Tcs$1J^M-#_zoaWd{U$o`PZqfUHJM+
zUDtxNMhHZko3X{e+24tsm%g?5rccmNeO+C?^*RmX`E$)ki2Qu*fo7x&Aq)HziG2Cl
zl2Wub99D{&v6<7~5lO11e>h0J&aom}QYg=MiYH?c;w%X_JmQ+&c?eLK8tI#|oe0T^
z7koPdCWpDZ#W@f%r;lViMSMgOoFzhupXBs4h&lq@Yvg}-wo@SAEdi@VC^7yRdR~GR
zPV70>80p8eod|JU5}rO8#3P3lmfc8ij0-CP%fX+K-WcD;<h<P&&wwIaviv>^D8~{b
zy)k}l=9~|X(?2@S(&-F5!{;QAW0Mg!#;0OA01hz`9Qh}o2LFW7eI9?5sgd3o565JB
zV>~qSH_{vB-<*`*7+;6wfNhM&Oi!C5IsNT`Q6`2xVAvzK%7(}F0kcc4|2{;T(+|q=
zZBVA~O-jI-JDuU*iDz?q!~QCsA<&eWf7E^B*GOlCzkvsHdc%HddQZnN><9xti5+FD
zOmE14jkLek$n?u>MoQgJ&i~$|^n>`Eoo;Xvxx%83L0E+Jx|WZ^`xSnPX`s1J*3-M?
z>!bG#Ls2F&>Q4rhAv~DVbDJt%jo8G1y#<D0DWIwOKLz*{{IPA)8TRJB*`%r{cJZ6|
zNzVTzL{UUWdcz)#$@Izi$>BGUmN^;e$D}<q_A%2RLRzD5;ZZT{v2mH+5UkO@jr@&z
z97h^+{^N6eB%_Rr%@~RPB7pcY(i{E<fSJ=P(jHcd3B>}HQL!SFntmeC%<1b(e0n;r
z)WI2iiU59$e2wsApb=)sZ;ed9MyBU}akI0V19(0K>6ue;65L3CQxaU~N>9V6Q}Acx
zZAN`r<}aIJJjzIK02odZk%$`NI|jG|Qqi4*=X(4Z<u~*n`**th?{3o5I}3nda>@Cd
zqEa4nAer)PXBv1m7B<ttbCgJ0cy4?iW+s+{QDr<W{>1n^&5D0Q+WAcmq+*OaR{Rqc
zV?43KC&lMsR`^MZq5rM$$%>)Zt?-i-Ltk6rQ{wYjE8M0S`qc{0hrX;dgT!+VT9=^@
z&5=1+@-_6F75=HX{<6ZS#plUZ_^FDaZ>;ds;(Eji|1|W1&kPdJIeZgt^m|im4qo>e
z{n-RYdp$!j`l%Ie%<>aZw%1(cuUZ14KXK8UOW(<Im?h!-?7-~iNjSHGo1I6(xy{?`
zbh~Y;!tL2+Cnnj*m)oz+ZlefHRruM0*>QX)pWJqBc0Uw>sS3AMo85K^=eB9HdrZQ)
z{n_k>C7j!v&F-*-^K%HZn~=rORE67!&2GAc^D_vuyGX*hZPn}+N%%H={$O^?C7ho(
zn4J&sY`i-#`iosgZU8(BBPee>qC-%)ZwdGrN)67EqBGz>!9%+CVDQiFqh@FD!)>8v
zw@nb7s&Lz<**zlR+}3G!F91&d>t{FxxJyQk1D=KU%%jxk4843Rlse%&XUz;Z%C$zq
zkzBY^;HN8>(ayPz((FD1eDbqj@?)2g2K~E}=yR`ZI{MsyoP`%($$H?MlD>VWQ&QGF
zeh(+)kOb$;E5ei437-@6&rqC6^%xWIRK3Kv^rSPGM4zv~vhX6xu+I#pGL?(lm$INg
zoo86(#*?4Zm6!C=$)`!+r|P%Q0nQg=hP}rS-SrZl%;)WbPO4u08sIZg-m&YQB0axV
zp_>Tbo<@J30yy!v>~jh{@r?qWdI8tliU7xdev$z`855$Zpffz~6u4P3MWukplgA%c
zNc{SDoC0Rx-;@FWKn6H>2BqVZ8zyEz4;bG$8~uoz;|Mptk2dVVd4QAudO2S~a}usr
zz-7SXT`lp)W;jhyeee4W(|FV?WB*SEI{PxfUj>}?lD}Q2pNBBrJ}fSf&tfUJGjy=>
zDHw9ZFYa;*$U_po0B|9PiF&&|SHiEB_*-An@%^#_954V$r}s%6ZhSlP0N{3+*6cl+
z0sb=J>FVXLfKxuhhFva+m12D%9sZIG@GCRGBN^a75O9=B%8he0y5}?C|6RaS?XLGE
zox!AbUW`STbo?*P0QYBre+h8P-`VCA{PEj9x<6&W=ORS9d=~-EdS4^!y<8`wG-SZv
zAmAq}Ta)_Z?<9Qi6{n!U`3c?M0VkhZcREE_?3l#Uw=q9X$Io*Dj`8JhIzGR3pgWua
z|727U=?^{U6nNj1?Q<CLboBEvX-)^fC<A;2;4`z)Um`kvh*!8^2K-xz&+-oI_)xRL
zJs|OW&(Yx$to%vhJ0<<yvb^O`IO+Jj18}>M*yR5NaORsY^9{+kCR7CBTb^|a$Sf(J
zF9ObbAC~ow<ih<};^)^oMVQ~x(!B`yDQqWF@5$z-90Z(tX3J5hz~i?UbThzwI{uvi
zGT*H-UnCc9MF#vf!0i&p>}>!%Kg~FGyPyL(B<a1EGSHud^+xi)KdGP22b^*nOsdB=
ziQk#j{=X&QW)Ah|?Gk@X(l_+*VM)L97N<a*XxH)QOrv~Yx5)E9R_S2H2RQj0O5*c&
zfscBTcC68Ve<bnOB=LCwaF%P$6;45!-yqX9VBJ!*pX;0={G9m3&NB(`y+?=hJOH|H
zoTl?*|BF+E`7IOO<A9S+=Vqq}za!hjyMVLajdM!ChHz8yf|B^fL8l0dZ;J7BG2p1O
z6t@)cQ?gG_na5p`0slJUL!MiuJo6<#w`airK?e9kq=S4{$??cw?1>EcZ)JdIPuJx(
z{H{}w@Jm8dC4BJjI(&|VpKstx_;nJ#0&vlfWW6Ih;W{OLuf&(dRqg<sa_&uP&-V*_
zMRCgabCQf${~`XCB>eo-C7rcSfy!^u={hBR&GSwX7T-qV>0Jrm|CkQHU&JcPQgj4i
zpS+^OWwTOR1zaMTyj>aK2LLCZYp&JlJTB9`Wx$W<a0~~+op7d(KPKDxds0Dn0#1JH
zQf>_r|4xZ-A94za@s02dtg}=8#<?}+5`P}x<Y!R!JGq>xGy<NkJzonr+h@I;rx<#5
zO9uQO3pn^plH0?8=Yvl2`o>ck=*&1<%Iz7wT*kPt0`PQvb^=blG$haMStZN+14(E2
zr#c<O{`?Xa4n_NUQHOKAmF^JW>FDE1G=Z;#odR!$#IKO>&Xqd+kch<wKfvws+3fuk
zaPn!F=N}n3{A4O!z8_?OpM`|vf9&T@L7v~D(+%562mK<ce`U>(a9QuhxW5!|A&1AE
zOs*K;UOy-B1<3OAW(N42nR>aLvRsDz{eY97{jWF$9e#^S_btHL9=6K)xDLfmU`c1I
zq|+-So|klnhMfX(MYbL>F5nW$<ehw8dU?*w09OI096DuxG}foTodN%Tz}YW&rnhjI
zhtT~o13nK8A^o9Vr@-SG3UsZ2lMe6TrL&7)Mfob=>Dogq13c^e^!Voj&V08%?G%*n
z&(bqk3^?tR<o@WAbb9SNzK#t(1srse+EuT_H}DMqUjY~8J)qP1tBiP8(n+?*=3JoT
z_dcxSXUY1F0G>_`TQb1!%>aK`z`_5xq%Zk7oB==YLY<#tua3{NSLkj7JRLtj$N=A$
z0sd+R_)!7Jyy+FE$ZwtaMc*n&pYL1&$9ly~9iL~?&@B+~H0$;}O-t?I>6L0AB#zgr
zuD0xiC|0ny!~(&2X33Dbsi1kOpz>5uMGB}S6_UzANh%K|X%eQgRWd(?d}%5ql~h>@
zQdufjsoBj_)Fs~KRh3><#i7coM@<p8`kBUv*HO_K@HhLf3xxfuI3BJGrwl71s$50A
zi2W=gsd0#0at)-4wjyoTMtb!#`U!nr$2>vTI9wWM%(;-U5jo!y$>T&_ANQEHbOq<8
z7~{~u_<3&4U6F7kinG0&6;-{wx<U1N8k<yA!FiN8A}x%~Lu#a5Z4To2x3H21q^d0;
zwJjK0>kFzaIE^r@`l3DLQXF;GQZ%n(ehDb2K~lLlNZ;3^Vy{ludPUzG)DmrPUymfF
zkT|{1f~u;u4V85+m8V~-l5_bit9*HFZKJD6ZK~vnk-(tr<0M+FfN%r{D=KQumCGvY
zJk{|UaFkbvI@hZ>MZCkW$|IIN%No`B^6X+fJ5(K@AF<2liul36E^nX|$KC3#yMYPi
zfv%n=)#YleR0DJh_buUINx{NrmNiD^>1XA6^ky(Wpr67D+QsIa8IZX)M&|1R@A0(8
zGxY|_kaMHY(^!to9P?AoFn6_!HeOP!syN3mbe$UXceF*;ss666P*?rZW~XzB*Hcx!
zYL!}AR3=d8xA`MH-_+CG7;!jwo*|Dg4@APAW@gmv!r7w{N2z{NCfW|pH*H5zn~GIe
z<9xNDajmDZOl_R2HaP$&Q<F2$%bVepOUK+LqLJa8Z(Sa)`j&vwiPL!_t&)&hQdCkj
zH(q-g2+gm;wz7b~%hgRx@LO&^*;6u|v4GVMSHsfAFpH^rLSc0|juUUkp^7PGX$%F!
zu7DMcgBipANHVuLb5tDvnR5K4`k8o?<s?s4#fkFGYI8U$x)4q#ZN5gFW}igaXi`lp
z!R}mhDc43@GuNS&XR1NRJfn%M^|h$YzHp@5iqshj2YOT#&lki&lu2wzb>^5GKkjz9
zIBs3<IPu19Zrh2boxVfqZCm!f5_Un<#xZ(9GG@8zX-VoR$jC8YoUMzKKcTzoq8)Ai
zE>GhUwXqug$K^`Xfx}HDYM`Mp{83G=q{ModWwq3f>@G%U51<1DTnX*(C{El1;Hn3A
ziPK@BbdY0H<1zb-pwt*CR~y!%lluK0HrldIU!cpw_8WGgT8q_2e-P(#I>Hf8V+s0X
zq%4R-LxC#v1|Y7I66g&-1cJ*F)OF}=Dn<~Tk?FY{-5^rpNHcP(k;cf}NE9*z5_p9Y
zs!=_c3pI#Lp$K%2RGfE>Gr2<$vdFunG#(>HVQYL%=kg{NTlSSQQB|cqB^c3IPY{qv
zQ3|b$pEOOv8z=!MmIjo*<)KL#`di0#H3v~)^PtD7)P~Pt0BLIvb?ADVL(t{jYJ(gx
ztAoC97{g^}G!jnk3{6dt!Evb^Ygbc0(-N#t>VE#7K(ba*$w5~{ZS@6$$t)~G&CPA*
zO$bog&|`J#b+~P!LyV9vE9oMKBQEm{`qWX<xGvC&<GUMt9oInR1sVcupcpD5xEUkV
z+TvTE#DLfnY}8~Xs;TO}H9%Gx19Kz(V6ZKTQB+5Bk5BFLi=oWv#IS|U&Z?uRR3wU&
z2SQ<$o2Sb>umhmQt>ugtvI19sR0TuL*M#*ZrC<GkXQN4>p_xW_(WuiZ2Xv2enjJz_
z3r(!H`vNJtk&PkV>|6ZJU4HIcuD5_YFun-hlUqX5O1(EER0GO^URh?S>p(|0G;#~f
zp0=o<n>wl=Q;C!~>Wxc9szaeJ9>mL9c0uW7R$LXdTC3YzmUr^JeOC_*fv~H-z8NCq
z&>0tXxpC-yQs+(I9j$V=_RM7Up4l9t8ea-~B*hp@Um#RNnY3nFGg}6(nv#soDL!p*
z)d%V!1koD9(uzo~Pgo=I#_&I*ua`shf3oTt{WXj$5861mw8BpCI2@|lz7{<qf(vk5
zNg@!P7%`~|jqN&asp8~oHPou&`kXd@s-=qA3L;>q1cNGun09<R5%CMb2g+!Jxayk?
zoftQ6<*;HsfxM*De7Qpn4u&QvHfmVy!g#c*m6~*FCDBv#Jd!65$-|VcQfRl&JEq+p
z40~GBn5kwHJY<;Q#k$Fju?wm{K3BjbBuV_9CSG)8%<`J*syXdz5?^HK#WhsbKiE1Q
zWf*Lr<5>69YX6`WLG_q076c)h5OEWsF;d#@>m-k~dCVq_#R77<0zEFYcvPifXlEFW
zOlFz2g;bZMvEew$DTD6!EQfW_66jE)_-2RgMVR>+vPerQVcM(sB4;h86tcqNQys$!
zGz|f*u=bMBSg+}8Rkcn&kHw**!fM5eBh=FrzrH(wIiBnx;WAi7ov<|7u+SVs>p6eo
zTrsKX9a}mT@yVZyrk7y_S}Vgb-`rh=Rf3MHGjhy=l~r|xd@Y>UToLLDwzvW)g6Q9Y
zd6>?Gy8KD%{i7$DjxwvPToj5!3PqpoC8k6A?H{hju+_}x809EVT=&67h!G#MgqG~_
ztqpWLU`$ilg~KYXIT&!QA;{Hg`y8v}X0ft`8U{t0spv5qZ$(^yaoLG%%f;Pnn6e+I
zF|`y$x2%<zhUaBiN^q1~F76Vu70%K`t?A+|sSeXDRbk8dg6XpVs5vI}aGE(E7i|&^
zFj0yBQ*&CwE^Vxk(-rFkM|C96lp1S~XGV)<7wa_^wB*v5f9UaXsl;{BypX4(#ov=g
zN3gBQ`K$Gol6coPHGG;^Bo5<Ij9or-$TXuqhAU$^5Yq(OSU}34Y%nY|!-%lzAY7t>
zdx$W+CuvWZ9u;vrB6;4RwuD3KI$uW%zFO3IEW`W;HpbHC_Vy(D_&76USRRH3Ppsu|
zfif<``4}@W3vSU%uJ_>*VN8axRw!=!khylD%T)lv&aX!5DXF+<t@_b4iis*xuZ;%$
zVa)3}f`&~wC3SyCnOU#;p&#XJt25f=SGxmUk*E*FN~3m?36+=$!6MAG_)&^kSh_wd
z8hg6=*RrM*ozP%WBEp!WbUy(Ty8?q5rngYHVv22A6;WZ}qFBdX$>5quyq@EIrL;^R
z;yBN03R$LIUaBy{({v;l8ffCB34#vMR^%l%b$Fk^1_65sOL)>8lzgXK0AvQn`fc2J
z)Yot4srAgqXui7PTvVZ8(XTz!oosN(0XkkfqYI~0xe@URD(%#<qe9i!+T&6$5gh{-
z32xRgx{+M|h~uPN4J7Hra(w!Us>#4kU1JmtIKE(6rlzdd>x)d71!C1vSUQGf8W*ML
z#VK<1HPWtP!{YjVM@Wo&>2#T#=$ICi%enlIX;E0G94O-zbx9UaLMmcBkj;y4GuXaV
zd~+Rtfq~{<VN4T@E&}xk`xCE-d>zgHwHa17@Oq=6G4-Xu@fh<BvL`f#=C%f-;dQB{
zNHU-}r3?o?o(=WCU;#Tm*qauxKXSbPxGUJ7fM(?M&ZHUjrDCBJ4W)-njFq^wt0pN8
zJJX9Ry&ccft-N$&#mBMl(y0oXcdESEOTEUwUd~t11H{tp$5{`u4*MK*0$pKOL(<zI
z!o%}mykj&MFO^b~=}Pf`(8h!kf{4;-Lc|=;dl8{-@aiCw#gSoBT;kwzBbOa517ezS
zFrlE5?Om(Rj`yI%GKit2tsO}>HB5^&iS~rGh75X}F9(F)#;Dlb-kEGiqQ2+il?N{P
zTULSB2OUvP<T!`J922k35-&WkSju&{k`!&2tiG_8*4_Rw^;e0oAL9$No{orok%~81
zrukL;MHH{O=9Ca7LE_~h-som{d6;2BuTLQ4fWb@kxONr@hdc2?$ts^fxHH<(Y}n+6
zeN5XWz9@?E<oH&|p-!-o0QR_t3(#Tg2Y^uVUn~M+q?(5%#FkLI>TBlJ+3^a%DI{&u
z#)TBHk{D@Tm)5{7aR{3{{svZsaB!hFM^f;On$_Q|^R+2fBSX!}vj<}#qd2kA;2+gx
z8Rn|`Y_(Behrw{y$hUcs^%#yjF=U6jd~N>Jngj12N6&q%m*GW@QrrUDCUQ-E`z;oH
z*YYyHV@TOC53faFWL~ijuM9=Yfd;^`onM^zf}YgV!|fU25YEU6plsd6X)8;gjj%ZY
z?ekdFK`nJ$mo!QHM5YMk7zg!^o?)pnuC6jDK1w!aD2ykH+k$9*VpD`ZX`rr@6K*ax
z;Xh2mbaN~*IYwMKd4{U4=(rA@1xt3AfU$dGr&(!HF*61sv6SLZyU49iP}AQ}YF=b&
z#a*hH1V@c&M7)>F<)xDN9)W7}rOhtO`x>i8Pj7_-$Ml-c`nm|)R5+@cRv29RYpKRi
zDPGySQg*&TZ#(!NJm~jbgXuySX5}ekHF{Y1I$tLnrZtX3L8s{zsDpcHWO2o?7}v=4
bzV*028A}I=cV$9aEml3r*NBRkttkH&hjv>s

literal 0
HcmV?d00001

diff --git a/tests/Grid_rng b/tests/Grid_rng
new file mode 100755
index 0000000000000000000000000000000000000000..039db0665ed14e8079e9c886f88661c6e82127bb
GIT binary patch
literal 60784
zcmeFa4R{pQ^*=s&gQ&nPh!_=h)zv0~m<Vcu2)iVK*}B0LqNWv1LJ~-7NMf?V;A=EN
znJz2TQj0BZ(c0E3R(xp-sF27@B2^4p73&M}eb$HsREUb~|MR(XXEVD=X#4$t{r#Tb
z^K|m;oH^&7d+xdC-h1xrOm55d&4`bSQ>-gqnW_*gKG`iu=3P-Lo)oC0Wh%!gZsl}k
ztTF_YB>dbWgwTAlV6PNy!DPvo0GNnhX1qzrjCYF%3yw9(NWr`;4GEcE-{qE%CQSLG
ziCR!moPxl3)V1SLfdUqb2MY>iK*@BN;Z{7V$9TOmUayR2K}G6sK`VdE8-E2dydd5U
zjJTUX$r1NHNM2s~ANtGqyyj;D1$2uC3+7097PR910P(2jzk6BPoGsHE%r1T9V@kHl
z;ryz~c^6GOzpCuKs><qy#`798FFNm{NfYX8CR`x$P5I=VIqM1$(XwL-FzP%SKh^<;
zcb$6M@lAK69J=)DRi{6gb=7^x-+9sf47&(FtK9SwM{!A8P8?~O_+5fu$L+_y{Oi(Z
zx81!k-0_)X+HZnCbnk!X>U*52xv%<0uNs!ER3!c3H)CtpFZN#dSFjIw-3Vny4G@xo
zN;VLE1u`-aeG3#Gh@Lr!oNET5Cn3WF$vHTPp0u-b0R2xNgq}1=y6Xm!|Gh!@j~#@*
z4(Sf0|BONSR}GTx?St^2J_vpMAmuV~5IfWiBIjp=&@UfE&aH#wchMmFd_73{emn?$
z&LH{(2jTzqAoNoQp)VLj{;omtb!ZU!w}aIG9fOo_<skN4GKk$)4Wj2egXlAS5IN%q
zkrO|N{x1$fe|wO8y*7xQw++JY7=(V^AaX7nM4xGc<g0BE`ES8KXDBBttxpugf{Ico
z?uzmQ@RRO-s#s7g`j3z<>0OVT^d=FiD6dHRni&4kppOK9(_c(}r{o_i<rm9z^Cf+g
z%x}|)Zb>o0GAYOTWU(MPz{l&?_|ea#r?_$NyY7|xC^Fq98FUfo!Y`fV?qlW?`muU{
zS?RcO?7J>fii#F2tf?-l_m|Z9i;9$@?=LEvUA~~Q-d|pqT~$(FUtX^ic?<H2%F62`
zxgamQs;0WUpk!WExfwQaNKvi7qOQE8tmwM(#YLrf@`L?Hv*y%KLJGAR=|vP;@2{&_
zoICrnIVIlGTqsl8SW;BmP*+~GsIt!AP*TMhm(B5CSn6L~TMm|tiHLSXWm$Q3QDt?d
zKX;K3Svje;y!<+W%<*SjTvUXVYi=l-hr~;&s%lD2u9^$yS2fgESZo*B*-A_5{p3P)
z&}*PpdC5W~QD0u?_m+a1QCd+_SLClNsr1)FkKEayO`KO!Us-B#dkcyR(u;EETvU{Q
z*_??*LUl6Ll~kA2EJR)xRhE`t4E2TL$TBR8^s->-g_YH?j;UpAj4Z@hQ0`w;QYAC%
z&#bJeN8akoCwXCloD7S85pr5mRURSKXJ_Q*Usf71E-hJqacRxM+LF3*h6vlmrbi9u
zLq%z(s(SDIe>eZ>MFknc07%j=5~?b1gjrCqDEG>XDi>7OAZ1x*b@jQkE2RpVrX^wa
z%IXEC@dvWHs3p0`7Oh9y7I|Tc%WKQ47uJ-qb`@MX3l$5ssIg>T<)RE$g;LhyB32(5
zetwNr$>yOpR~MC*_)9B<4rZ;qAiZc#QTf7J|Kg&$@>;Ztn!1t&<&3kSv9YMOyso~c
zx}>TS!Hd#mwE7!LYEc_lm<3ltrwi>_nb$CX@O&V~9RH-!ng)OWxTt-D#5IjsQ&pc!
z<5l0VP(+T(_5~S5MT-_z4v=*eLtpjDxj+_0&BWRUe=e&-K>>2`9~H&`siOeu2TXj9
zANG}Pv#Pv$fxn`tysoaME>abvYc|T6Ri&&de-XXbLepVY)htlzudnkfc&H+ML1}4G
zJ;O4TqARLzsH`r7f2nU+SkAywGg56$ePv_O!t#YEyam+?vWxO6qI-kan_p5{rPQM4
z`R7XygvQud=~wD27l^u0U+z~BSAx>o#R?jvzj9%D&HS>G#bhff^;gzZEAvW9uY-#y
zEhqcJnnmS(l%o2@3+JKKkOI|`4tHTmWwkQDuDo2StgmgTF7+!jv$KmPPDoc~`n=P#
zi!PXO!GsGV&l!`<$HGFeOq!4tg=J(a*@cCk>E5D@2@|72qyQN!Bf}nu4j~@%l6Y|^
zM7M#<d`iH~#d<>biD!n!V?Ju~$16jXWWhoF7!gMPp`Z-`JXTPN$1BGne2Abj#Qfuv
zHHVAC%;zQamDcz^|D46gDUh%R{X8#*|IHTOaB?LQJVtpa5+1L7D#O#kzw)Q|LiBOU
zQwXO_lRv@6KLirZFaSlrEe%;rRyIriir!xUAF8}4@mC(Z4o^drw<Nwq@FyuBOT0wz
zCn{e`+@E$2rpEC~qRe3LW5rBH<d>GRE^B;&Xkwmc(XH_U>74snmo<JNopVF$vc?Of
zUuqEnt#J+M7u)F8_=WT=8{Ha5kUq^ux5hQ3Pqxuh))fnEDDHBoB5<dT&bnh=sW!S<
z*UXS~8y&?MxiW2Zrfpqr8{Miy4AX3M>SSH{HoCQD!LULb-Ci!mHaaR|<f^dIhefE0
zQfs3Rx6vDI^bt0ClZ~$0=u2&M)?4dZVWW?<h=8kXbo)GXjg5Z1jsGDVeTt2~&PL~2
zg>|jB(Z^Usz*ZamL>s-sMnB0$@3hfRw$XRn=%?7|T{gP4CdOdHMnBcY-(#bnW~2Ao
z=v*U+Tyk8`Ix;mvga@$Eoi@6|MnA(wPqER@w9%b5`dK!5s*Ucl(bH}8vu*TD8~uAW
zy4yxiwb3;jJ<UeXx6#kF(F<+#aW;CfjXvH+udvbGHhQg%ex8lqXroWC(VJ}a^KJB{
zHah*db*-?`Gb|$DDjWR*8-0z9KG8;h$VQ)Jqp!2kFSOCM=1-HgK+?1KITdY5t3S#3
z6#}*9t;t(VrtpPNASitHQ@E?AyYWQ4f`Or~Fs`#7CQjEF>J<0`#OV@4tpfigak|3L
zI)UFqoGvi5M&LgoPS+P&A@Dnh)8&Pl1b!oNR^w2u!0U+9#f6Fm{$t{FZJ~UD&m(>;
zaks#KM4T=zlrHeu#Oc~XPJw@)I9*!EA@J$M>B>Tiz$X)@3k&re0paWmh|_h2x&(eM
zak{Kfr@+r7K8kp&z)vQAJn?k`A4QxlD6~f4#}Yq*_zHo?6Q?T*H3|IdRlw<jLbU=v
zKs<$bvA{nePFEDl7x;U`>4HLTf&Y^@T~8=o;4c%W%LzFJzJoYjO~@he=ZMqAgcN~q
zBu>{7>iL%We}cG^c$dH*CQg?T>J<0`#OW$Rtpfigak_}mI)UFqoUS3XM&LgoPL~i`
zA@DnhpF_M!;5QPdYY5c}ypA|sLa12aKPFCB5Xu+$JmPc#A-BMPM4U}OlrHeu#M$IS
zPJw@)IGc3HA@J$M*_1<yz$X)D6Atwp7UfTzO*hmf@N<c?$%Z-wekSn?h_?#-Wa4b1
zp>@FXH$RA#4CD6@y_stxn}}{~Me8gS_Q~Je`YghiM8b)RaO0<-saLhCBa`<04z0C4
z=>q^>o0TOH)DdB<h%oLn8DEKF?6MdOLsP*ye}Z~?i;%<ioxi!4=_f_fCo1)$eKPiX
z^<CF$`g>aQ2R-=(-uBiW-3?{hngnRiKgF#WuOp*h%vVp(QIz#BA!}D?=XC&R&AlTu
z{lgpLui3Jm3DV(wam{81g}c<#n@C$v%;InBO#r{9Kc}^QGEIx?)OPhYAo4~sYD->5
zWKG|0#(z)7Z@TopUpg@y9(e^~8n>druGyY+CrELhZ(;tl=1V*70fYavhzg^x*xd6d
zw9`KfcU}{^2~pJ3l~BD2ukc*uxzclmr(lkD`==V*ly?&XcumuvaV>)%)AXlYO@u*-
z2Tj~~w_H(-@mG>#ix!xu1?D7c`s-TSq1=oQH2qtx?b~=3XIhJ^n9<u?T(y9_{3*Vy
zFV(x2!7CY;!029ms}$hV9~9h#bfVs1;#y5NV`^y!H2oE=?VEUivQPim@SqeBzMA3W
zrGgsZCZ@769yepc9Ng}^0&?_rZ{W6M@3?6hdo}$V3X!^{GbVI%GZrrYNMF|H>Rn&L
zgBrg@3Yxx6)7QEL8(=2cgE_xOrp-9sz{TFcRmpbcG<~1(;5UeB@tKjHadkl{4dxT}
z2(A>`YQbefxbYlP+)Te~q%vbjc0GF?=EJM!D%yJXF;o&*GV{6$L8861xE|u;$QD8J
zZUzsoo@qXm;Cd#B?6p!3F48*iyuHP>n<TDdD#kgQqJW8;MZ74T&&I4#L{_^rATp^!
zJ|ZcSD69fdlr11H(i-uM1eTH{643NL#&a{lsLypJdxO_FXmA-pB!+zX^v^W?pT=?1
zO*Y47=G7QWJHi&hD-zptDc@9%rJ#RijBqPT;2N}*y?)faea1Vz;jq5im4kXrI}>Jv
zHwr<oR3eGMQ$lYh*+CJK$M^_SB4&cSU%XUDiz^k+9(Cl5Vj|pLBa%-fSt4#Cl%qP9
zxpoV<(q#y^+SLn)7}uubcBK=ySq|KWc-v>xG6iV>T3-vm+ThlfgfTuj1zC`$UWZVc
zdJQ1W+6qW5I~WTWX`Oh6sk=xbFZZnp4M2t`!^(`~kk%$G1<#0+N)k&~R7;pG9fSbt
z1uB~QSc~f|KveQfP~pfNG#aI9r9H;MT#<uSOjQU$ZN#&21mgssUgmPj93+D%v;vbz
znem~GQ${2$70=KnlO*!y<AKl($nau7#;G6!F4Ai8j5tjsY0!Kr9{Ov(0#r+LN~7je
zOwD_QD3Fk;hhTeDKA&FYO3?xZE{Aa!X3h*D7wZT)j0e!iv8afcMOAyyEUGlvADr}b
zof5W6xK6?~60VT2Nx&RetuOeXYaJ}i<k;l2pdUp2JtH9<?trTTo0@*&RyLl62*$KS
z=v4(R7yu3BqZs2r6OlxKu?#)^rk^s@v`xElEExhx@Bh>Z_1%URSmnUphqmNRwWS@y
z(59O*6@SuCva`ESVk>>&OL8ZGGkp^WnV^L`)lEI7^qrKYw;LZJ=NcSXvDBPxStw!n
zFp@A;e2{88sp3*_$ExiY5|CF@@T(ZGM5{fA41uKa1E_s+f3;6Is@f+?wa<r4*37gV
zS21t-TCmDh>_efh5*e5aioVCEf8^7@Hr|8DwcyRJLW#djT;C_LtunO2tA8s=kAnnn
z!z^q1JI1_n%M#ln1$ZmejJ3p%sX9tM-?YR{7}`Z!!YlHYbmj_YzXCitsb2ve_XX`X
zSOE^FRZ)Ovi$u-*YQeQ;L9z`TFg`g`<{J){_&+6HEb$${efmC$Jp;^^>4jxhrbCh9
z*P&5trW>ebU#9QG_%=Gz1tJ?s=PmD_>GRP5V`}Y~&ySYr6lNBgP8W%;bk)W)htG%}
zNGM^PbGD`Sv1~fXlVZFKnXzhbVmf_lzmyuQ_OFE^NsI38ul5}Y{nRcxa5S|`s4moA
zEE0u}q1t+mD^)99Eu3_2#z8h?4w-XYnOb{}OM~Kyf1IZ8*4nxfwcuzs^Vtf6ZE9kk
z?%p2`Bi52upPKWOp!;+`q6QPMW&~ec4-?RMb9)YMSJfpCKx+twGosB_x)f<Ux_eq8
z)wmIJE7MNr(RQ3^822(`3^nWYi5TdKK_aUqJkZ*d9*SGCS8chPR=GvEpxo^4li7)T
zp`;eTh^Nm7?GrwT{>evUz$j%Rg_W*Cp@-$Fsm~{v`<eP&O?^01GKNI!a~FnMgXr@r
z@)x7ehMuG7^9&_JpMOAU<G0r+O5T!#9G)IB;zc%4%c!B~+TB9XGp=5dd3N>_B*lpE
zjFpr%5z4WWqS5j~PVc^OI1f)qCTw&A(E>N4>k?sa5k;A^vKad?&1aV-2CqxNHrM4u
zs_R7*bFgg2`U5FT-sB)^k8vN6j8^UTt(*>A?V0Pj#&fNw=$b8}gA~2Hu$?(*2^~ac
zEMn+K&^m&o*%1{;H^R`zMi^cfQEe3J%4~+7j139h2TJG$8ikXMzFv5W@yTc5aQB7A
z?*KyQe}z)Gk56y$X}Ng%8c$33)Q+*FaTA~XVqh->^%c+k2%+e{;$eJJ`ikckN$D$|
z>jWjog;l0Z)0jV$Jfcq#g))<rW|WcQKUS2LaUm&IF==cbNJbVDN@$D}p+&|TiBg19
zif|eS_Cf@Ti}hpkw7cEPrefSSHp5UGn(<`p2ZON;x6rSUUE>Zuoq^dFEQm^~wj9Co
z2P-P|4ebYF6rtpeAK?~?Lo<t%Qypf5(Ixk>h|2Mjz9sZmAeaodAYnzZ%_o~bP4Vil
zYWkO)a0PEd-P&m^oU17F)qSm>tG6|Gft|2Z!d3~_Nw`MB6%sa?FuN9!vWt1k*V>!u
zNqA8b$B=}n5~fI)%zD$_tO4-S+7pUbxMA{!WNlo7g1X*zBwo|&wG9mQh>5fR6sTNb
zEX2Z;FOaQif#xR03<Q<}_=2VRzTneBY+9Sv_Eo%3->#+YTr>&=qBZYpGA@+yx`)>v
z_x)gElcqN_BE+mR#jGKSi5q;HfyP&3%rtFr1$XdiS_UFa5TKsXA{?52GlsA0EdeLh
zLm^P428__1Bq&+{#tU3l7#)8Er`0!Fv5&1!HX~*i`hvgfFw-_cptE1vBJNZxZeO;^
z=?gyDWkxqaz#tZtdqh8Egiq*~b8^mCYfc?~tti@G^#ae$cs_4QtJ-oL%F~DZCHn$-
zuuFY~7I>ry;dy~8P-s^>w7^aIzQ7|iDhiHP%gqSS7guCWC_9k2VY;5UX}X^CtO8^s
zkj>KriJR2Nnia^`#eI4Jp`auR3h#n)U-qFXPiHgcebWz3*^H<gz(f^waSt>f#`ARb
zvBb@~c+7=6IL{kMbWINwY&eX7g7t@|>!a6u14BUBY@>9aYOXH{=Jv$(NKzq}^%!o?
z%W7BexD-t@PjB;~1Aqf+7Bw;uaDY2v-kq$NW7t)!!K$n;aG6GXZbd_s6>KRgCu8`6
zk2o1I&}@Q0>Of-l89%g%xg}Z*W1z6v4k5+_0e3$!r?9X`Nin`aj>|!nGAL`5Kt-dW
zUW}W!<axE_UOadMj*Npoy+dp6>P3WC)RteuDn5Ou*8EYg7QD4rYulHA^__6L|8!0N
zN^5>Mu3-$k&C6g;dgW^2dEUfxNB5Y|)tZy>o9%cu86rU2g5M6UIUHY)2C}_7M+^EM
zPS?^t1Je&War<ExZY4dqrOj6}-Ous%ciy)0_M+C@=GL0G_cXkO*bucFzg-OZCuOUn
zTKSALYCGci)G=SZ^(AlC{)R-ox4TgbR>yg>LYm&~RcF4T>2GLh`=NY-miD=SD7@2>
z=l!W#^H*{HlW>o3fP!CkC#sK4c>=$OwJ_GFJ|m`ux9h!H^EZig=jgqbKStdHTh0^U
z&v_!5w_(sT(U-Ne?gJRs@9_rG^8!Qj0;64CtfY7YBR0d*O2%Hf-gw`kDS!6pQyvP8
z{<BAYZ1ltW=tuO#KOwn4BVk|Gi*@f{GbP0~N)YR1dkjD3XdKK{h3qS^^H+GT<U(UA
z%HFFV(SoeD!O^Zw*~`&kISh7+@K+o<CL7lys)ot|jz(hxH?Xy=m((R~P)~3Pek@Ro
zg)q+239Z$h53gc%xmK)`tvS9g+&G*HgzF(-=EWr(9`060W{ZWM9x&rstn+l4&tjQ}
zLb%McPQaC}RsmPLIt9<w>FUTiPIcsE4s~RL)}CEU(Vq4pG^MHQXy)zqt<3n<o@=*g
z&8@wnw_)L)w1MlbJ!)IBzRcwUN5BN#Lj%ddAB1%W8o=^nz`|UvLW0I$KSCdU2>q^x
z(GV?WJoPa~L3JF7Zp2;?$lc`(i#4DB@~D*07%XL#l4>C-qQPw^Fm)%iGcu7nM;<GX
zCgD;*tG=;xX(Q3+i=r%WC7Z>Y!Dvu6g~jVI?qO`UeS7gXI2njx#U7&$D+z+#*v*Ky
zZ>89)k7BvC2#4heY?iz2&(xd6XEc2X8~v8HxmnsKDsBJ#DBEP(ZDW^x4kI%k#xy9_
zolaBbVy%SmQotCKJ;bDwkTf%OjWpdAfMWk3YP_Ur`XrZEx7H}qIi+*6Dsi#GLCi_a
z7ntB_ewMwEqTbz!@p8tSnAE{hVA_@()4zEGk1hqDcif{*09xDLc&&7sH|-s-81iiQ
zr5#)}jG0L{?!aWB`?&gJFd1y7q|7CIZ@tYM__c_K{v{oYtjFs{b64CZNP>SyVtSh&
zOJ+Vibli9768%IIw23v}=5OQcrloxe;eW$SbncUhBERIg>9Bs)VfoP37Y}|ZUTbbo
z)|$5@H*Y(lw>7*>&Crcca%sT=sJIV#-E=2sb<8`vwDUf|&@@oyO7{gYkImD+%uD<5
zS*qvLKle|;K8GhTX)PHG-s2dyn+yM1FgvVeZS#-XwA3!@z4uW>aHd&r*X{7>?{E4E
z1CgLeXBz@HMHbyC@X3CZ`$_qkJ=StkFu@ffUVazIy;iVjUkHXMvpGBo1)~-EL0@o+
zv9e$A1%1Izqk+MZWk2e7JkyS%#Py@>Lo9D#C@1B!0&6pSplKdT`NtkH10L=938v(!
z(c>ko{1^`*Ctmvm9YirfXS$Cd6=rS=WQ(d1Zqszs5DlXqUm(?UNstv+T~>;jNcS<u
zr(FQcyT|1%`M})OIR(Of?8mlhwqB+CSTs6O(B0TF!~8{Z??rUBt0bf<>en1P9*pnN
zpJ;)n#Rx=9!M2hy@U$4&2*Q&lL5!FL;Q^D-1tXaAxSyJYjtHU2Byb>Ua@3dvF~Afu
zt~Ciw5so}TxT(VApJrnDCYE7h=_YoTi8+W3xjHl&=+#^1uZZ#wSZu_63VoxdPk}L~
z=9T`91I)mTWVF}OW=SF}$Ccs@G@$>x0th$cGT!v+hrMlw%{?!!d%~3zBc|d%)u(r9
zZN^k}i5C^scof!$efy$|CBnI`Ld*nyN6E!vqw7^}?wP9vPgCz;;v)<!WL3`3(+^`F
z5iW40WI-!+$tE-+&U$6GC+`V|A?-^hkBAsH^`@&!o*)xq``FSojJ{ks5R#`G#y=4v
z7j2!?JDm}8Ie|SH6^Rc^e+HJ#q#I^^fS;r~sh0(ZEz*L+I#>bIv6iVdzn37^XRx{I
z)A##JwC1nl_%Ia@n8WSItPdLoBgB##h6!#5)A9yqVfNmN1(tq6TKM^+hj^QhB-T|@
zo}3?3$Z_h(h;T~KpVOM3OYsj=N4isxC`Kr-1tm)&`T@5)mZ*VPD2QMaQaT$Rpi_3T
z)unfjVxQ4xu#DCWB;@1bA?*uqiCNPk^nc@LYi%DU<^`urFlP)2+Qvs7HZM(mY!@wb
z!)O*EW)!c{A3F`6cIj3Tl2CUHvX2fBwl=pe7ou>*GLxpGo4*Vlxt8n=U^H=@7T^;$
zO7)Ri08a%IFkAfr+yco36TnGCJfrY_0IzQbqzVL9L(F#L+qWRUG{=SUIciS8w*`Eo
zU7G%Vhw4b&qK=IBHg8Qv=6oAcpfRVjf6s&F*3#A{MsMC)Ve>4g?W~;6Zpar;vYP&k
ztne(4yo2SH`{1@WZKv<P0b)1o#glOl7&gDS%nE3X1T02?F&fbf6}J!_9YsS{5per)
zgq)c5l0vWkq8BYz{~YaA)_FJUd_IJWK|o)Dvaz_?Se!n+%}By<+l#`%fEQ$A2r{iF
z<YB!}%Q%?Ft)6qxR`Re#whn^4fwitKKp!@{oEexuzT<8F=!h7Ot_NB1H^|i<F@C!x
zQNyYoc5Q!@=nYmRVz!y5zwK@Q@JPn<8LcZkdb>yO2xK1)OuX;VEr<0sy$29;<;xGN
zj}6U<!z+;R<z}a@HE&DyqGxYF|9Xg1R);UZH6+}U^Yo6mHXqil(|R!}+S$F>r@yJC
zedABY8aIX<{u4czo8RX-G-bWMU46{|tP-~q^E@5%q;KNueBQKg@`AqgP;l->3^uBs
zg$Q&0OwEo%yv2w0cJ;Aa4#yqBz{7JWi&7etxFg=?HZkBZb=z{JVW&r@>r8Z^S9H>v
zu<ys=PLGam64k_X){Fj$G6m{3(FH`Cw-yditxSd)(fk(^MSZv&H6^A#<QvDb`keh{
zzwu!{rhS|<an*b;TA(+u2(4v?Lrd$?aFE3NtkrD2>SGwi7;5t)C`)a<Y|O`c15-8F
zEA1<+S0#YwMf#nz9a`2W>K$t#Ps<8<^%s2V^bfspFR1sl`Lepyd!9?);Zt+Fux`RG
z*X7OHs@|o884mnIRuwHc#ceb~S#iK(Ds|Gb4x#^6TP}jn@C9cW-n3qy-s(N{Ar84D
zYT@TJy%Qyx!1FB`@P-ZJefn2&W!EqR_WFl201eUzEny<fy>Si41M4m}9~t9ZRA^xF
z4JR_-JU*&R-T^lQ4=@%M;!RIJL~EZRJh>LkMLBJUC+{B=?)VoW&EF>0)u2&EkB2A<
z{=6><$YHH)6aIV#{P_&&&mmtD;m>Edp&^`n9})h1hV<uc@aK@S6aGAyvk-&!t^{w~
zjO5TEwqXnpR4p*m;niot=MBY~f}v<%Dc<1a$ky}z5e(CG55v@vaSUR*M+<8HI<aoC
zmUblLWsH|KEU(0M`bWWup9nc~N7AvLsE$k=$&SuHn)FoCQ&mUKi7D858i@p~5H~|G
zU(PuZ6XKka=_nP0adJ*9L?<z_kRcjFG<0soTGnd~%WZWdm^j-T%z?r<3DW!#YRF#y
zaj-AzsUGq#(3+3L$!8Cqv0^Le19jP2Fb{*0d}RA&SqI{^_UUFFFgw<xgs`CJi!K}f
znS3bUsOb7@tSaoUr=xtQ_x0D3i1M8-%NL!vB%*w$%ko79WUqgK<?BQF2FG@=af%V-
z*w?tAH`S|On|xU?u?**%+Pyj&Il48QciG&7e#gjv6`tae*X=%Jg*eVtFel%uzvP9U
zq9@72r~qAl0v3F6PRYJrjgEPbp`#`WmhQe4UwD=w-oUI_cCOp~D3%@iC1X9-pCOJy
zVJ{sFzQA!<O3HY1MIiAvSVH=(M^C&@KI)_I)52SPSzGGfw{nH;B^+LBLOk340sI}h
zLHIlLULl0Q*FmdU@OL5yTGktC^E04po3S#&Yg!&X+;hIPueG+XQ9F)sz>@WjddEg`
zW&IO_XLz@6Z``Z!Yrd>~@Nvnz;NAA&yc(xHPH)z?>RrD8^Y?>O9tBX0-=H*bI1PfC
zx-YGlt6LoUa<$us6!hnejVJ|gaLQV&WBr<uG7cJtUV(p)d)pfvyEd{<m;DbG>^ZVv
z*RoK%Pte-lk4FcmxAWvv8)A9mux2$-S(hQu*o>rN`irRfL8d5jEUs`8B)1wa8KN!;
z!4+$z4{@#lPmXpKXzIl?hHb9(R9Ot*)RwXEFX&yCzC!k3;$9I%2mcFZkxG8T5wvWV
z7(ssp*Xyq^2hrkrx;c&(&trwFLOY_1^0v>6c;JYuVjF{tYWK*7DV(A3R0D^wVJ<`v
zJaA<Fp6qC!QBrg)<eEcg589`v&;Lpy+NZ}YGDrU_iD;i5**-a*hX36q{V&^~un<qL
zVfPZFtUUb-Z0EVK{4x<X^QGPFO7#Y3!f_?Kax?bY_2WniD<>YYX5!Po5hI>k;a}u@
zz#CY?_F@(YR#>--(R*B*R@&ypA&cvbMoi=KwYX2vQgP12@F6{mf;}~ovPeo;n6CwI
zasWpu`Sfp%vp~g4N{T1EJtynShF>sY_+w~_+14MCB<BhkfYmz0$X)Cp7=K1ByG;0`
zws$d@8jWh6DVAUKTC}~3HT?o_;09!~4%k)6UcF$10}oC<V9I$FI;DaUPCTUYA(eBT
zWRBWr?n3<uZ_Bx)%H>j*g^rlp*W8mpUC-?EYoXJ?7){!;`Pd+a*k|IlaW%BwkclT_
z4EeYW!1WXVMB}4Qb3k?EWsJ$6XM}$+0#>w?4M!N^LogUk2o0UpHyrN-Za@Co1a`0f
z53ha*e&>y(ce54c%_%P9B(S}SvECa=uSm*JQmS0;clN)Lv<0*`+IGMDw$Ii0MjP(6
z<~AZ^HE9z`e11>p?cJI>ok2m*q>PufLl_|R#(ADa5h*@(rg1=ZBsD`0ypWn|s8_x0
zUqn(ZNG6VzE9oB(s3WUfnQB^r%gx!%EfArm0-3L=)!1a1m-Nka?q|$RN<ifK>dmbX
z<a1>nP=DBVKrQS<bf2r@?HpId?%{FZczdp^7#uhSrG@|A@BkDz;87hH^LAUZ+VVIQ
zL=>!b;*O!PEp_#(H>u+bNKQ@djMQgv6u44xpUWBd$HrTbK;5t*GVjzPQ@2wW2zWCE
zVCD$jc0)HrK(riJA+#&pJ-ie=&@LZ51+IK=_{D|?snt;4E>Bid-xJvux+8ZEp_~JX
zc8^q3x2vfynfjeQ9%2rt*S>l{oqGfceT{^Uz`zI8AH5FpTOfCU{CAK$;8{R!!`2M4
zb1OQTcb(cX9bSD1)}EV?jYvJ+(&v6wXu^T=-8o<2@>*~6SI5=e=)L``B`6oHCkL*o
z@ZR2b6DoXe29^Lgo(R@qGyMBQIj&K?JAHxCFKK}rinWAEuX}^T;=Nf1@=kgacdweg
z!`u9R9JYdddQq{a4||=`IDT&a`Z(+pb!s|<huaz+)!Jv)LOpJa0{7{c*ZKl8iv@Rd
zVp?F>OTNIc*Dnu_PS)Z&d|A7+lMeE(X18I$73WLq@TIkCI@Hk<k=*4-t_{@{nbOo5
zyO`h>YkYx|*k1i0)~fRY3(&@eZ%Rj|VZqkVS&W!9lE{uNB|=A4OoVN@{bjRnp9pe)
zf~*WA<>d%{Kq(l5L^Pq8-$mG?l?g=X1B-Fw;heb89liTMjkyu_h!{afQA#Zb7Ly_m
z#rbs1U(K-_MmeFpOVJ!6ofpR2w*LF}PnSZd=)y743wxSw!+8;hXZy?)MO1o6=$v@?
znj|kyr_wLo{-WiVE|HWPQlwuR2b%Cp%Un$mD_m2f-8D&ipLfDOq}@9`gHoe+>cY%V
zZTSZ<(Lfx--<S$drGIG7uhC?&+9VpToBMsA9>x$wtr_kIljW1pq*901lc3JUv-C-b
zBzzLKIE0W|wENoK!><D$+I<DuG&~jC{c1E+I<HpzlFAS)R<=YVb|m?LW3}Gyf3QRd
zf$)jcXRgpEHR%f)UQI>Kt4Z3;XZR#E^LhBq#YXMT@Cfil)TtB6AhO6(F?1^4J^Xb<
zfKG*ir;wg<6?7saq7|OdiC{Gqt8Iwgs;2%E0IsbNPR)_DR)ith32(rBFq?YU^Kb{i
z$DZBz6So4M;Y-1TkcMP6@F>E%A#98A@M;>IqH*5~;V?udT@N-S3oVU{KvGkEE;pSZ
zlkG$e8F`hcEO9_Pfc|h+GVX;V)YLXOFjlRI1G@%0(Q^FJeZ*cJ-3s<+F?Ya_^qan|
z45NJq_7V~y0qw##_IV&t`xp-~II=lHo_j$NZRnu!CfOtV5F8CGCVOP-h3pIZ3Ah3j
zZxD+K2)(bLgp2!0IJ=*O<4p+|3hyzl>L=k#QlfNtpOmOf>;lD$>6^^RsD47$#ZdMb
zzX64ZFGY*L7IPYf{&J2q=q1bLj*_v6F)f2`Xa^EyQ0j|;I^XBnw4AGsY>LdtRK|-^
za=b_pp7x+|eFp-gEUhBY(;QAvgIuhM&hO{sjuQ1QD3;C7?k8&tDNz=Cyq|>Mni7P~
z-)F9(w7I>XgqnU5t}!Lh=EnW~ButBy5aQUG3o;Rx8l9yPrZ`cJX7>}<LrRnh-T?(B
z5E(eLAAf5<{y&@iLfr$~V>7&h{8L%zR#D$d{%|{LmDAfCPE?niiuP`|(jMb#u%HUz
zkN_sdI4@z=pVu%b6_dI-JoL%Z&i<0BqN@s!bkLY+;#dzi@v}@k#lnv_aqJG5;Rz<5
zYT=*bv@ztRTlo7Xo@wE)m^c>Q&G=hP9P0`u{-lZLTlnuxywJjbX5tu^nEcHqUSZ+&
zCSGgdr6%5J;d6kaPi;bhW8y8$6AE95(vK8<-+uZW41hFwNbF6t9baJiI>rn9qC<d-
zuqcjAuIBe~jA&mgeqpucc<_0H*E_tyYw^3$k%u?Nnm_Jikp6Ohj#|@>PK0NI7VtRG
z>G<8mhjIAL;=@Jr)yZRean5J*nCIi}JM@)~{^eE(y7k5T4t+act;X4*$!E5S@ZNaA
z6%Q^FpA%?rf*>YXM1dNDe}6=t0eN4V^1f!&uSB>Z1njrueJxN!2-t7Q`x>Y`w)D+>
z^~iJcJ?arG(QkKRwuoOlYG!+{y5xAQsQC1Ez`^WxsH2|8Bl@w-_~#Mh*1vnRdK!{*
z^fy8`!lXESg%eD@sG%5{?xqLCqBKq~4acSvR_}Ra=}`A*EaCr^OZboTB+(O|xK|Ke
ztlv+@gPh-hXA^lg=&ylm(#E)*o<Po~M38RYkf`qh&9^Zz4)dq>cwbg~U8mk5s=_t>
z*N<=zrIBT-A4JEirnN*~B&wx1rP6`B=svL3k`q|ZO=<Iuvxm2`0ova<!>SjSB#rxz
z!93mevf1^|iI>ghBBmC^iED}sKII~A{2VQw$Bgo=wD8;s241A^2IQJm4ptM!;#8n|
z_n8S;I%@SnKr!NEJdbsRJpJSD)9Kx8{!POk4jD9@bg2OJ27b#-;N&BFcx~JxO$6eY
z$E(&#8P)-D1n>kd-u?Phj4w{I$*afdkHi(4-Xf&o$j8lE;8#LMz<l)z3u9{EsgE2i
z#Ga6&8{*W4gnh?7-1g%h=J@bGJMZBQY!<nJ6<%OI@WKIGtcG_aw*uzrJ;vX8;sc8+
zSY{B#mlJrX7eP7dV`~LZPQqFd$J~#68)bwAJk$V;kFckMS6x^P7#V7O@l8CwgB25F
zG=|n#q{LR@KD@j`O1z+miHt?s2ZMXCYKH#ABE2O^!z|UY*OhT_g9DQJ7FQl#oJNT;
zGv2^*=Is(ss=B0`IQH(a|B9pRg7+h1B<h!01-~b>-tG-vnTSJV2o@F0<U44bG!$9{
zPK&%~5ZWy2;e(Wkvps>Ld@dVtt2UMOk#r*onPI)72E2IejTHtnwQ<E9`8M(?pHCHh
z;u^jSOZ-?m;QU)t9T@&%Qa=vs`UT@~(#<yxN8J*~VRLd^Ed^?*E>sY!*ghDURWwik
z%%5aj2JIrpHKqdLLX^{Et_F=&myP3kjI3*(_MwPJ8UE8SW@Yp38?%Pu5ZTk7=n^ck
zp<B9f)LEPh8O~*gS2v-a^Q4F++FV`V1Tu#fueB12K&iBUt8qW$Jlnm)-lN#(XE@KU
z={P3IZMd+uJ}fTg{Kq&(elAWF)t`g&LtNegLpyfZ_F??T^D?{nc14y^|9QsC-b07F
zix*w#)j#vLeU|PEmdOKKkxq=ey~#8vU(CUB9_spMtVn*OMvdFv$Fj;kV+lpZ?Zy%|
z1mCcu<@tu~-53Voyd?ty!(3Vd_I{!cW}59WFhr~D5F3MSA*cRa+KyeH#O-MN<h05i
zkA%awk$_dZO}}{Myf{tGJ)l+YoP||4?}iUi$Pcq<^bYT#cezQFhPTxXzO&wDoTgQ_
zGB*9ql7pU&jDmIbUEazAeAM(F5eGfOKykMFDz^)^+i@li+-)B@eaB$u;$pEg^e*lL
z#q*}MdMi6j@&7`l%KfHz9BrZacOf1}k}gLhKDu5q^ZzQ#zRI>3`Pwo2As>Z@UC-}M
ze^kcrRDb7ptbgivtUubxcWVF4?^u8LcdWnm9qXU}|4V<<qrjou9@A_K)s?O4Ut2XD
zOPuX9E9ffi>#=?7X;{X<tmzB*>b$Hk)Rz0vZJ<Lv>`QwOdp?QS$U3}<Yk}-7X2O~0
zrG3ORv3Y?L^8!W5o(;z#%zOKhYIK}8jPq<jpR3?7SN093dp0DQ&*$85yk`SOZKMym
zVQ3zXiFtR0^3q=M;*hB~cv+k;Yrj{Wz7tziY72YeJpE8!+DqPD`z{NP&B{0^<}2tf
z*o?7~`YQIBE@lsb=g#cF_?GOzl(iWcr_2Zos%PSqm3nWRaY{ySW$W<yCy4_N4qw(g
zi%!woeR2D`&)|9iANO?Qtp_<@@dOH<z|j|kYjLl~ZOhijw_w0G(UcfBWo5RGoy+ic
zEo-NGM=ko8S;52%UshXvL!N#hFKvf+*C%;#yKLh{>I;911)|rW@3BG-4jFm^<Dc;a
z#y%C-jYUMMZ^GD>`flpGyRx-BP<V~LCGGPJ1B0w*5c3#QgN(5&J^I+Eu+EG`vESAl
zwap#9J}UZOoPWe=A#4qc*O)l6$N2|qb1|#ioX6oXG{(<j{Uj$a(VO*>KZTo-3Ay^k
znA{}#^xKj{90o$Z=*oD2CLY5?^H{N(^PV?ut9nl>`dyq^nX3N$x#Z`p{ysd7=V36Q
z@jMFwihlyXbRx$B;c4Q043Df}2l@95H|8RY-}1nDOf8Uy{d9B<FLDc+2{(5g_Qe_g
zgd81*K=<OyBl<qERJ6k={~JfS2lIkAKbpuRCdlz2L-{NFb{?D3&u1R6Q}=shfIZD;
z;t&;PaEUm-tS%XXlI$L4EQgfP^?b%@y{N%}82sEKMp{Y6Lx17eVvo@Z_P+KLnLnAw
z(fiMz<N!RdigPDyjbiM4$ha9gVlpw{q^Ri-()l+gMcDDbI}HtsGNY$NA9SA;GanM^
z{*UKGu=pWE-va}kAm^X|@qFlA&WF0>d`Q6mv-6>+UuR7z<B{KgV?OjEiu&K04<%xG
zD0)7001Ggj{9t3pn*8jMB-Boe^nxS}G9S9sHNbob#}r-al6`0{{pLj9;+0)<PISD`
ze6TrD($`SVHYa-ZiGMvO+7DI#PtA###prp_L}(b97o7uyr`xexztxlVmb&chfoDah
z_Ma7XKh9Z^IdA$GlcH9z4lpVD6Uc1-ZBG)$Op2B<4kkrbf70h4(7&Q{JtB68niElW
zc-;!8WN<zQA0NPntv2ra%)D@n7{_h8)aJD)VQu}H<`wS^TqIw0!+AvT?BIE>v>m>z
z_tZP4L!K||T^xtaQ>TCGjeAj?M>MSShz8g8_WH+qv$m;sIhpo#-!>ZiVMI-DgIw|E
zR~BaRFRCp|;YfVJ%a(f6_IuasOU(-}5qb}G!MP8KK9K0)xx*=N&7leKX+HhK6+Ak8
zf;k_8iTE;y^?DbaZKH<mskgn?ZEY0i^KM}%&nlulbcF6_Zh1fvi(ehxm+_#=R4p*0
z7Kig{RR=yRG}42aH@+w|(t~;T3^yQV-9uEzEF8JJ(y2PG#a#P``27IWZ7*iqn9yzK
z9$h2PRY26O5Y;F}xuqz#6y@$Ost}^`AxeWNC<al@t^cc{aE?&DV+a=-(9h#6VG|61
z?x`IE5C1ImPh+5Ol>-Xrj6vpe98fuDOg^8Wbj}zUAjbi<6USiLVT=RI4r82Hb{GS-
z=Q?mJa6oyV1KKCzI9z+9|3-XBVvHfuSnD9fqQ;orgd<xV0s`-lxPZXdOI$#t+K89>
z#w-<)SB_aF;Oa3C38vK!6Rva!`p9Jt!3Qm2y7or3r3q}*L9EB=YaN8tfs20BK|tVc
zi3<okUE%@)?_qLKLlN;;j&TUMdW=)>t#+7jr9;r6hTwyI#=|SCOD#>F8M7NPH2oO|
zAvNg`5?UMr0$(q20f9dxaRGr>h<MA!Gzz$K%u)eYk69)7Ry$0%(jn;3MDRfq#?x+p
zu8Sb+6^uWytzm0p*FA@NTb101Qw$g+bYopJGCqxNU)M56^7To!FK_rIvwdlCSlE(B
zB)?jVc8(<uuRg-7S0{U~+t!HF41=DMJQwW7l0VDy)b^!}qfMWAQktt7)>6p+e{(h9
zzY8GZ*vx+yKx}JbgDgO$e#aB=|1N;YHOWEjuYJe%|G%;TlF@2g4Z-~OU*UR3G{0?p
zr@tx=V4Hg<{BWTI-dVwT=z4YO_t4GzO1rIt(wdI*NZ0>~bBlJIUcWr(dMr<$FTU0Q
zS3TNgeE0{9cJ=wmIM1US_Wdjjmu{UL{(E+N;$sl0=-b-%o*uVt*SBC?@}_RwwkNa<
zoeDN@A__o}WGWJ>M}WTP(4OY-k!LZ-4R7R4z5MmKe_~MQ-OvjR9Ow^s`ObJg<DmD9
zU24mtAmF7!Z(0Xpsh97Idz-KPHh+i>&~5m9#NGs)#tv`uAA=VT`1uHbikg~q-${y+
zjls!dUx@n$o0z=D*PG{lKREtT%p&lv!LYx8h>_2WzPN88TD&{1Uj9<t-*MJm!$vd4
zL=ZI#qQnF?vfi(b%*9wCw+Fv2{5qjWD|)`3hL7^}ZTz%F)&aHUE{Mj?+o4^}UmcOT
z+%-MuJ2zh6f%DiHSSI2vg>iUO&wmQuWAq<KYclIF;=qs1UnSOEh_@pj^8~Mm)5PoA
z<VSsJkK;qT=dfwI>%YJYTZi!mNl){0hnKv3TbI7A`P;aLQ9ivVX02wkflume`V6;?
z=R?AVZ}4PP{~mqsF@NNFC%GQAzT??q`-0v2u4tU!hY$-rI9`Z@cG!5uFn)(|0$!rm
zvUdB=!k`sz8143-tOe5f_8Znd4jHc_8Gbtq?+2ouk(|6~I?i1%>}iCV+Ttq}`0xUl
zc7w^%2k#4iZv1VOkgoo`6|dYzmrq}Nz<LOWJ(%@qdasw`k1V{_>`(IItxCLhH`nFZ
zn1YOBCWTLc82|VklEoI>uo!s=#_<K?RcM8`9kqbhIGB^{Oj4I*KsJUkYD+2;!JB!>
z-b;`@Y=og~hv2OCFg>0NmpEDNvR1EhP?-bcF{JABr5Rkq!oKweQE+Nb2ty{UVX-2&
zg|LRzmfY62PKCW411nBXLtn)|-bUwyilbe;*Ttry7^|r$Hr9}gP%2M9;!FDkFDZK4
z!blGFwd*i8qM)G<8sa^WFy7f=9<g8kAeR$wemaqJHL*E04My?mV_kVXLyu+1N4!{1
zz)6ck!hD>A-hwzNq3r<t!p}~OBT0S@5kqZ2jAsxMx)x0?)I~J|NtaD^!sZa81%}_o
zXMSN$_hF6@NV-9WIk~Hr+VglZLUG72>Vp#nBFw?Bi=eoQjlUxSL`}4bRieZ~6w6t#
zHyWQpGaj#l)%&(8kMfa<7BFJ#p1}5z3U&D+;YFSjpENaIe*pO_BZbrPq>4+CEt$pD
ze70qAr3^bt7X8!EXa@P($iV8{Hk=#BP0-QsL6sN_{fSu)B)vDKPpfzM9MLM;LOUn{
zWjGJftoPZ)*QPoB(e%HgE{PW#%?<nU;DVz(TYQoQGA|KJyx7T2(1LD<re|Yro$WAp
zb9>fcafqFgaS&-B)?+7g;)T{T&Fr*>dhymKqTyY+WWFzFM%#dB?MX$`oSa8tx`7Yu
z0<ID@z<hg&=MGkYw&{YO!ig=u9~mg!mKStwX8@ZA91GvPxD78D@TdsZAI(N%|4z*m
zE<9v?sJ6U_2;Apy60pG4Xo@|PVj+)YDerv9!%7K0;0Q0rhcktjdq!;4leZJR2LG=>
z_nD{~p+BN0qGZlX8QcLV9G|fbJPcVQLY@*FMmkFZx|-~ET{W&;Af!pK+e79u#1f>x
zV|<E*HZ~MAq5@aNrc4^BP$X8s@74;l@V^a+nj6#B)FoGP*pT(U+Ol3GbhE3FV)9Kf
zBc&L8a8-yo17c`|RLdHt3u~Z}GOx)fJma0;!W5^|6k~)bWEcz9w8OJth^UX3d5ev4
zG&{TigxK@>0M{)*hC+7o29jm=>9_ZOWoBQa-tJA)nEfA$G;l~l>RmZ3l6|n*rSYu*
zk^O_nJ}Y*b{S|$h6$e7tc?r33QZBnV*0zz5JD@#MXpdfqG<N>Rvd4|su%bW_)`Kw9
z9*=^-$`4;|PBBiQ)sb;9+Ex7ws*((a>|}9{#HtQ_Iofy=<18(}H>1Iwir2TX=7oce
z#n>ckPx|^2*~rAZjqYEQ)ogCjVt>i%aQjwz1bR{Nxk<azQd_t+WB(w}c;N7(1fk#K
z7R~+WQMkE=5zW2kDBOX5+&?=C_pSZ7?>q|k_5HXPABDTDA9vMJxUcHRU33)g@Au=L
zeH89X`*G(Ug`0&M`6QCMJR}^*Q{uDV#%I4mNB24?Hioyb<j?Mh;~?WjGVB1w#;_4b
z`XdZ$$?y;;Hijp-cy37ukYNodHinT97-9G^8N$&z9BXIrlA$-6VV9jDl?<Cfu_eK8
zjYg6fN`}8gGl>5!U@7_0FToIu&c)B|+1y5kW>7*`(D)b@phtDAMjGa05*|a7Wzay~
zQOgdoINdi6P+Qo<2WwJqAS}c;aR&6`E#^wMUD$I+5ysXS`gNbMXs_6k-cK`r>~Ap5
zZbw)*M(tZdf5MYcZ-5@Hg|o)!RzB}2Li0>fq2JTpLv)8FI<}u^`Ex(H96}AIYz@M?
z^C|I@K8ewuwxzS5#6KKG;!jOEp-%xf=c4;Dr(9LA?uH++H~QumqEdu-wt)$MD`6!f
z*y`ugl!NNG8^OkcRj55Qr;Xv4lEDFJ%pe2DxHbm*DoSXjgtN)8z|LTfA-foujKBd$
z7cCc3#-M{!I(O;gqsGc%2xZC!g+GG)hW6qqqMDg<3L+bK(-{4F^_EQ(@E<2@4x645
z|GdDM1j*ftW76cP#yDA~8P#(hZ45qeHqBhDqr{GTnTyEBN51;GY0fpwg{^n~pQiB;
z(<oya@F~wgdrVm)>l60(V>$Z7`2?J(@nWmB^m}rAfm!ONM_BdpA!Al>{QDfk;B=ap
z1aka{4d($y7=MDaysS_C9*Rjf5@<F~)0$e`{O=gdId=At!@2ml5KrUq=*c_AH^@Ez
z$)iL+1=^<#CYAV@<kM7U3j)3$oFtUNoMDggvK--or*A~lFcaGfp1{Q$JFpC~fol=Q
zk3NSYw{K-;BQv__3EiZ==8g|4Vd9(G%`s}<6l)-cNL5jn^K=I5ZwHbvx>ut9esLtK
z^Qf>h>>$IHpmbj%yj|>2O?(73D*4O%>z9qF-F_KA$^!mtqvB5NANP9^)o;5X(%>gS
zD_V6K%-k0~N_;TKI8g@HAdpN80OfulzYn<v`4)p+V+0vc=*hCsS0KbH^kiG14}m~i
zp&$AgqTRlga^!+dq`t<k-~20FEH<G6;?TCybq`*&)6{c)E(ZbB;%j$5#;_Z2^WhEI
zttnepTrmgk{PV!X%{x%ZGFpwdehO6r^^VPNnfMF91NF(9drb6MpqpDUwcNxxiSZdw
zeVZU)fus}Zp-6D({-ntI7-qaiGU@mTA~u(MGQ!45KyBgE#4I-SIYi0#gT%jY;*t#8
z=3(<Ah~Bnuld$e`YlAn4L403(7FD3W2M3FdpF%DdS=j!O1$?C^Y}_JgqPGrAB<aVJ
z#IgY!7IMPIm6CP`Xx`?nPA!=2#A4%Z_d{*Ww%@{nn=cLf>o^901z!Io?t^2=uDLg%
zq0k$c<f4pGLWch`#K*Y-lXe7|!sa$kY4y=QF==Pvn+x5iacLCtyH6lTw~D)?`xx9)
zx)TtNZ`t)VS(ps#W7hl1@+4sJhGcVT01E=`c(IoH2a8g%zC1G(giI1J-7-EzX?f9c
z;6Fy;od9t<Y^K{AxI7gfH%RAKlf)NB@Xhbn7U5rh?J;tnrW=~+<|kCJ7<{`|zdT)@
z%J8($bW$vL?bG;Cl$Y)PwRy@<;~GxGf4^!^yjO7H1@Elq>I`djm2ZJ<j=u-}{Ak?F
zFB)8eC*$}PVznLh#az{Re>pXfTcoc6p^aJJqSrJ3ReShP!}B*^d$*+=v$^38Dz^yf
zoA$Sdr!o9xhF>{g_<0PU@iT;96bb)N^=U`Jq14-X$2#*0?tj?r9kttm`q8((%XZCw
z=E6gSM-bVE>L0iXpZYkQmxh{!&sTkmGb^{CTIAx43pNX|+?us*(d>*jd3F!|9adoZ
z9mRhzE)JE>Ljj?+<z*ehN150e;?vdd8PBlTAjK@S33sy)@lKIg3(a-p1<RAW8@<5>
zw1`~vS~w8%HTEy~g?W6QtNsiwK4<m%Gx);abe6HHqf{%o`xG9Ggf<t*)J_8ro6Fgn
z-iuXfeB=r5`QyytL2ue2{6{C83&lUW`3g#OCx8tf0*uFpM9^Q4M+=QqLHX|=3=f3y
zeMG#PekELu=L(!z^%VGmXQ#WI_?LFU^EALYzO)Zvk+yHq@;^;BZdnE+K%vvD^>!b>
zPI52@GQy+ZLTirGvbr1aSs9%ByvF)h6DYBWKmU9SS`Pn*Cphp3FN$v5cConaKSfE;
z#xK{#f=AML#nMS@!l$AYH-1I<$*X6Y`Ovdj5?UoxW|vf#)hu+*sjaMb=9S?8-8VW@
z>jbT+tb9>rX?fbQ1M<wTY%DKx)|Z!;#V}l1UW)%xZwYY@D8%B45;C)-p}xMdq}tSF
zKo+U2uLS??ytLezT~k%lQ0KH*O||=_I0$!C+E**%m1~Ys<~VOsZc(nBqD;L>shUt)
zQC@modELC4lDe`>FLhQ;C|XcgS$5vkY?D5n^wj!Fe|efRVZ!<NFY4!4nTnl16MRJy
zDXJ|mzpl3Cy7CDX%A%UeGA9EkPbPH^;-=Qu%=@wPobk@mn(BJLQz|=U5vWs1uJ?m^
zk+A6${6~1u(wtW_)Oo3tR5_=#q^hKDiYYe+Gi@Wo7G>8gtgR|<oH5=xajFx45dWV_
z68e;ya;D6$swwe9iYe_H{O9=_@jvKAmeS6hRbRQV%n8FuDwR%C%&bhFT$G3UHNT{x
z%3oAcTU%aTrj$4GU)B4|%O)^FYTBij3JI2Ng-s<-v?&*s*DWYd6*h@6n13-Omn<xI
zGH~+bdSR#{Jj0&<yOz90sj8@FX6q{IEsY1P4JhXMHFXP1=2gX12wOm^Sr^O#<$Zcd
zeYx}8@v^}CC}~P+YH9<j#YK~f{AuSD)s&U^ORi48CheT`#u;f)C;h0p%!62%(oDyy
zc_tQ9jiRgA(JPg6>g35%SgI+^rttq>-MvQ1o1)xu>@mkEvL0C;&4dnY*YZkFIXZOC
zHp9uwYt|%cn`x;W<9Ds_3o^1x>ip$&S7=iU8>%ZyOW^gu-T6eZkh5;NTR>A@Mmhx+
zRMeH1l$jzezAnipm~mV8RIzwmVWYE$=#+STvItgsE!;%=-Pg(dJ8Rq`EL}pkg!vK<
zod1l8)Hjkxn-%gc{iJ-$&XxjH&;BF5sbzwvy86;fGelJ@s;sW`Pbe)_%v#>h!7=5@
zlX=fE9mE{?kyKRj@lM>*l=IG8SX+tv0&oDp6)E`ng>CC2z9Yl#Y$+_iq^`c)SF*UK
z!Jk@MQBvnT7Y#~C?WaME7XcHU1=1=jpVzS9y!y%o)g@K+eYn32LY3^2h6NRVcsqtV
z&n`PwaXJ^4FLaiamDQEk*E`Rys%j)nUZpi<<$|lIwyvfW3^jH2SC?HwE5W@+s+8H6
z*eONTHPz*`pJ}K_A=xY9<jDh)Q~PHX9=Wem8Zgr2$#dYI3u<PUmoA!7SF;eF7QL8d
zyeLGJ&wna!dijFN>L|^alls}^*Ef{c`-^D%@o?b(t7#$ybcF1EZQX0XWI4W%J0;<g
z95le0cwTo028?WscVZBTO6~+)g@2H+29W=uycMt*;p^@Shj)_?>qm|R@Nr!mP{C$Z
zEnqq}!d3t_VgIrVFduuwy`<w{3IAtSJ`NsL05&}u4zB@p<FIE3VC`dg<1-2Jp9qKf
zpIv*N0v}-4U&7%g!1TX`!w&(rV$Guq(1A&uG6eFT4Tqh8O`F1DH(@K{0d{VMe84r^
zpfBn8#7qxh{tMx7D)!r2@%g1<z^+}$C*Ydb!r@N9rZ;hL0dNH%8~Yl-6u@<W>42?(
zjkM1@;qW@ZbifWkH((dwI>6*(kUpRj(D`0CTu6K`>_WT;{D4h>2H*-ng?0u^0bB=|
z4%h{#0T%Cvo{(D$$p0k2;sDYGT!WRb6s)wZ155{O1=IjL0gC~<02={&09ODyzJUDz
z`F|B#0rLTO0~Q1J0M-H~qnw)noq#I<GXd8C<^!$+?4n*@qC9}-A4EO?d13!wTyZzf
zR^l2Rai<I$n!F-zsQ9-nV-Zr=5)Pj$c?bsPR}>e*$e+@LG*M@i5sn!nQogSqcSCZM
zGVSC`&YkEw6I8)Nm~dmhPoy|BK{?5-B$(+c$|QtQRwvS(iEaACmNjn0Gr}=_M9TCL
z&gmmkr;kYYjL4ihqUpHiq05E@lkQGjo^WTph&LDU46JsHMqNo<fB!)7az^AwMH6YQ
zWO`UT-;D`Q;s=N~eT2g^B4y@?*3{;NyV+n&y`KSGiVdQ1<a0kD<b&yZ&0wM4Z{fG<
zSNO6c@}Icc43E(}XGBw6lmeDL6Qb;I+L`uT^_y^bCm?NLrIUm7^U+p!F`b@4(peLo
z4)wbk@p{p=<DhS1@$Uw<&x-!>9z@)(2gBjB(bfkj3(p95RF=SGmLcu2`=M}HeDNGM
zHErP;k?I+d-kh{7@$Q7>@zg63dey?WoB{an%4a+u@=@_Hd<X5&G)R7GqjejPcz&h_
zA9f*&Hli2m7WA4qVqIMGkY!0hp`BR{4}!l5KCTDs*2Oyd4CsaH!r^SdJ~{pK_W|QS
z9uD8Z^aiTSt<g4T8Z7&<;NSfu4s=TSsFR)%PP0lfd<w#o;bW)v<%8iubB50aJr%xJ
z*wxHuB%J!+jPT6$;qX_KAJGf?H(RaBtk+M$zK)H^C(1vu<9F^pUR!%ymk?!Wmd{&E
ze^WTT1J47*>DLy?Hkv=A*NS+rvcAcS^2UR{bW1q=VwAoS9R{q2jOT|OC;F0F#7i9e
z;6VEK(}nTYB3|Kk_-f|6aKLz|F8%ZU59sMepY=5LkJ$C-dNSX`z~8Ym9Da@Q<VCs*
z_=;D;VTg@fq)!2T=>YTs(3=LJSApI*0KEnDb+@4JpmgX<K>4dd?=`|<xZ%h(2~SUf
z?uMMJSnvLW{C$de6|aWFaL<v8=_F!IP#i;dfj%CLg`iiEXD#e+`sEyW56dkJJ^AnF
z*MM&%r9vOePgWuRL!i%s;S#?<{QkPaefKZF2NADzcR2hY`qo74_tEi$kLZf0kC0st
zQ#JiR;$?n-l|a}faorz;z7hW~;x&nQmTzG^w$nt!%l{}Go=pOD6h4Ty#YOwL5WWK8
ztoGJL`V`QY4nQvey=eeC?a>H2&kZX%Bb)?GzeUQwfdr&)`5APpSbtI!+C%v`96p%}
z#Q00QuSBTO_a99Erf|591jr+xoWmHK?S`Cqw!>b#eWHCP`B~4#g1;jKeVK0mvKz3y
z(=G*w*Mxc-Yp*KM8wa4bfL=QQeKqJ61JIuWy?6ln3!oQ*j?<Bmd{dtfB!4WO_MrS&
z`e=?B2arD=^nvur0ev8S=7LUr?CqEO_(9JHJ=UKr16>;+-3LKWzcJcB<Ur0aj42CI
zKIM33y;vTfJ0j)ogxnF%U}Elw)MZK9i1a&$Xd^P4hvtoNSB=O-8$d~maU4c-7!M?i
z`ceQnO^_q2Jf(>%obcy6nv{e`1pV(8l@BN4mpcr%_wbvJ-zm(a3k9AFC@!AEI2%v1
z$#yr8;ebU%6v#u0N)?e4fH0(*NQp!^NLJWe9}oCTB0mRG2sn<&g^7v+m!qUI(Oe)D
zu$CeFWXNQOgn_VVzh=luh;Rbn<3yGMIUS)QWD;~a75aTnDIAuc2Pi_m1o8<GaqU5{
ziYLML8IZGNfHD$^Q5|7}a6($b_ZYI%3g$yGiaL(1cmh)W2#JX+obXx`>^qAIGpF+z
z;KYyVTt{REerMx%IgtbS(QXd`iQ~C7BG(e(R|swfA}&1RvSh=BiZXtH#MAMk>^~Cm
z;K!6dATkR-iu(q?xQ`|BIg$OiO~y}Lmh4Vw-2w`?4fq8Er3@jcrONk+aGW~|a3YcW
zfaKsO*jUAAsv!(X#BC;^xWWma&=wU${%n%W2Vo1cKZ{?L6_N&t)OR~WI7RpyU<r|2
z+=L^=v+@iOrYG3ifCyj6kZ{6kCE?^OJn`%jXCA_?KV%rMaKaX-<0-=<^K~IbXCob|
z^EIF{8@D$A#WhZ*BiPcT)1j@VfQz>FB;s}o659dz6=fa+F$@{bkQN}90Di-e%`)U%
zhP(iT{}o(Z;e@|7LAgn^K9%DcNTnUW2K>Y|1r+9c217i!oo~{aFH0ZB97Ear@Y{~x
zPNvEOan$EMA_Yiw8ek_8{_`=02)DHcNF#;wlD6XGc;0yNf!xfFLNl|p#FdRZYezA$
zEAV607ni9UAM;Iu`M{#R73nE_`M=BxVf$cx79q=lFu$y9%1YdLj8<IXg!kyf_QcZ@
zP>0;1xKTcld?Gvpz|gCSY{G90e#Jx%;ddf_xwO>?h@`4BiCh4L?!?P<et;VVlw(ao
zv|EN;PPV5I!Yb#cw5>qs{vRR28b@jCh#ZI9G31wwb2<>ld4Px;2#v*(5c3JDvykc3
z%aDi3_N0s>(%Axp(t0WSbs$XVTOy~R92s(ean1+AbiN?MB?G2Y2B=iw#&jB(PO}U-
zOtvi&$zjN=KwhF)Ua}5wIKvcpoLGz@n0AoJ!}t;5kwc{oKh_R0lc0mViBw)@yPSw8
zo>6$4r0eke1t0!<{nrBjwZMNZ@Lvo3*8=~qSipITTd2AQi|o8wC3K%`;))Cxfw)=y
zU@ZI_M{2h4mtt_P?eM~sMO^7)+yb(f<@JT`7Vso-M>qUvv3NQ~;!X)Uj^Sm<aCYdt
z*!%PP@<_Oba8<mUK#aP?!!d3FILYB<#JdS1!#;-$;CzHviV4IZN<IpJF@<@pD;5yL
z7jbpT{Bdx>%NloZoWd)Aq+39X-smuKOYhwx6yt3Pt@MwCWqwqXALoxn{$z3rwvq(y
zkl`}=cMf``{bLiFFUxh2gm+4KuY`Y)@UIeXm+&<SKa%jEghNg=^F2nwb0oY-!Wj}?
zDdBtx7fE=hg!fAL2MPZw;dTjMlkg)64@x*BMW!#|ITBtZ;S33{lyJU;izK{L!h0qB
zgM@#TaJz)BN%)b32PGUL+v6As&ynyV31>)nrG)b(TqNP065cCeY<bPh&YtW{y<%QN
zwZFkRX+q|N3(m{PY7l5fnn_S(XlYeR^#bRj^1AxUnri372^UUCKW~!Bdcm#!x`ulH
z;tQOq{*ndt=g-dd<$C7i7EQb`{rvfrI9HJ7Oy#MaGACmCd{^d~(bKTtP``MgzhoYu
zzs`ge))RK+%j;^D3Dq_J@(G^l-t!<@VhgGpCd_N7tSUPXXDKESt0<|jP$rZuu7*4l
z`s>UPsc;lhgs{5usuD(!PqkHkWde_QC=>iRuBA+X-jxUwXOol(<rPKq>u@BisG<yU
zEet|SON+`IOUrAq<IlrUri`MJy1J6ZrVQ)($EApf^hy?1mLgsalE7_(NSEiIlnL|d
zaR#dthX~56{jufrKgCZ$MEYvWPi4ltfnoL{36}p9MBFUDZsBL)fj-NMUl8vGCMSG=
zP9nC;@=F#e0}uB2{&=?t@=Jj(f%V;D<L4j<+tD>v`m5sIf_arBaHh<Qvuj?KKQ%FP
z;fOxnif{RSLlRP?-zQ++t@xJzy%IF~d@H^p`w`1gTm1%D#1(75YeA*Yw&GiU+M$~K
zcbFiezr|<4>k!U9z`|EaKfg-G?~?JUKW%5}FUiV{l5WMf`kh8jt#R!(K_q=EzLozL
zgj22+-|ENm#b%14Sp6RD&%Ql2{@tL_CeqA`)$gZs5d+uBCWzdbmi^L!*yHERej#7R
z7iS4<5oER^3u!B3;#=dKHFBUayVe#Nk5+t(`pcO3);Q%0Du&DIuT9LfmGw-x20!X=
z*-!b28yK#inBX@t0%FsD2yFKFYkuYy5!d|8jL$PZ_AAzYe?_1@zVUOnU^jkl##+r~
zCfj9|pG9iLgFQZvTkvXGBY?8fDp(U0bwxf3@G|Zr@UzBE*0`?YQIpGQg2+8K{dW+;
zs&2)%#(jp2AIl$$e}cHw--_QO$1OeoX8e7K%Pd;)t#MDUjBgdJWd<vKs~me{((iq&
z7!)jiJtjb;y_i^E`?2JFivYX+iX3+;>&zJI8ECt#corOjaC`jqj~5H>^^cp8EIlm|
zRy+$IhRCE@`R|hPJ7j#PHK3OfoK{2$N6PfmW9U}=+hXX{m6rwSW~{VDfV|Co!h8^W
zj<x_Ro<$-a(?mp2>eD8Nc7h2{1JLlZ@^AS+HqU|5ukAGBDW^&(L2UZbq0v|Za+w+L
zjwS+2fW}roYY$97h}C~a(-R}}ooISeWS$dEM{TwGF<U?adTOhGvXPN$vSRf+(ez^!
zt3QdR9~+tPMAMIp`2T47Fvarg(e&Yp<*%dZBO>#~XgWN_PuzBrcus&<xBR9(FhLoK
z{}40~|0wvkf#}Ca=B3ecMn~qO(ex7(%il!P$0(K`iKd?jzff!^iRT2)My+;l3&qJe
z#cIzsGRo^@#cHR~bdCn2FV@#YWq&xDYOa2B$vxH%+|nSO$LH)<hCoIsJZ5FTawMI{
z=Ij^8Rg}ZyLiVdtAfpr>!?9mCOFEC)*stZ1?tZ{+CoA_$`sx_^lak&OL*FLpJoaY4
zUI!A7y#T8oI%VJiz&P0X!M_v>N}Qy};b9<qTKe#qiv6<m;qepu#oZj*hsR6oms8St
ze8hfn!-slW>o;{Wa2Dv)ldn|sviusir%31f)^@s;uU1JH$$>$~!$A2OE@F>TcwEDN
ztph*xvDT5CGSHIW8zZ0ldIQP-0Q5Ml!o;?t{Zjsllil`Y6@K?v*dc~~lBCD3XPpl^
z?XV`M9C>Cg4qHXmc*9DhT*`6AtczVQ`1|{Z!ScoP*m3YbO>3g{|Bayc&o9#-$UYAX
z{{H@<6?ATB#p=IH(yehAxp;jb<n;GTdqH;&;AeV4A7w_3x}Sy4aumwF=`OdRqnN~1
z2>O|z7nisN&NFkomJ5FK*?#|>l+zO<=fy$z&w;#w^tp5p`VT-q10zf8eG{u)@C+R3
z)_W?}_=soVn6Dn)Ez}fd#M8<_<Ub<$i(~5H6GDEUe{K@e6lMP)a>gVLoUaQ6eYnyo
z?PJZOZUJ57Yq?uUHvKi`A)q^DT6XZ*AoMMu4`k0*L1#HSvfV<A8|A@k-VmmHvQqp+
zvB2}i-8>sMkev14ABetV5c-FM(8pkLf#u>{>J}+-uEEO(I`!<lz%6i|x#smp!EZj>
z?`?wKe_YWn<>%k!7VK7i=Gm-)^gjaw$bsl?&{;0|vRpD{;#UZM_(`e1C4Y^i=hvC(
zavaR-c}cIm-7RpQCFJ!!=*(|zp<CcP#9dJqVE{XjK73!{%=qC-*A$cgILZG<(jmus
z-%6TIc~jEY%`*AvA9<a0tSP@qu9uorMVTt;7H+|xgYL8t8+;se+Q)id%xZW40G;}*
zNplNsp55m)6bg?7zct=N^AXn=&{-eS?{o{CX9;;-Cg|p~{r+RnnXdKTnq}wP2I2oH
z=uW!;`TPjzjsg7SdLd^x-mP&9POF{reT;$1Ws;zy9AoNR3FyKO8K#_ZGGEIj-Fh$1
zDzD8_j$^6GFWHqZ1YM{SbsL3_oOW9gqvvIyvt75!c3miA-6HvmWAuDj@S~ouFw?c_
z`8Lq0{|e2d_eeQspusX-B}UJD(8=E;>!*0P0CAp_{H+b99G=1D^|9o4Uh5XPIJtqR
zDX1cpW4$l<l{7>J=qz{ZJ;u=@mZCI)K2W)=0exirNlM4##gfwmYX{+ff&8$W^GTDR
zpKs#z{viBc4?;f?4T0&l#-!^4opx)|-9m1;xZ_NVr02Jo^z$Wsro}JmH%R&opo{u1
z%j-l*zen;rCBIdF9v5`UWCQPkF6><67F4Sq4nu>YJ+1dyttjI_Cx3bjf2pK9J*J!u
zGQv|r4*dD`CSAG#<s~WKl54^7NR;WuULFvIo)@`A&`!y4m82_6OggHQxN0r_0+Y^n
z9C$q_={-emf%Dw~Ui(33xwJOB1wLEcaTx1()6Uj=r&hZ-7j){AFYCWJ%ZxZPL1+D0
zC&#6hf13|FlIV9e2|Cif-Yvo``>Y1t0Xebt_Mt)KYyq9^-ThRtkYly?-a+_N&~T`K
zah{oOp0vZ2lHMiD@eE0C0{#E1<!X8yN6K&z_Z26^hlKVApeA<G>14U|OxiPeCfyUo
z*<p7i^sd`>((bY2jO}E)_cYpHfcOP_<A{WWxB)i~>|u`_kl-h901iki@w`>#ia)z&
zBqr*KGOx?!s<+;6S3Pax{}t#TiYHdTe%;ajkAe>QsmbR*Ko|UcB*!Dp@Sw{tQ0L)w
z=_@`L^7$3$-2U`hwlB_BeEYkS4`0^!Fae$W*`~5RpB>=O7Ifxw?+>4e{}*37)lctt
zlKw_r_>Jo4GbR7xtOTxo1v-!CpYv1Evf$g#pqOoZ`vvIS{!Jj;i!&v-^KXj&9VO32
zo$LqSlXAsaNaBOIfj>V5y{&(@gZ@j^&l}axhoTqk3hikB*ADuBKxe;Mm(OpC_O<tx
zKSJGJ@Mm-?p2%n1`15;3KmGJn{J%IG^e@nPy-sU#@&T5L`}xyPWP7p07I%IGdK(|k
z1-%(pe4+aJ`XkxTw^Us8=Z=2<uG+t`WqWZp4A1=-rJ1(pW`d6K*Zl049qoVHLH|n!
z{hJQ@_wiz|Z9G2~bi|FX<oK^JcK$r?Xuk%X`TX^hQ_)VGal@~FtA6G`lypAJ$)EoU
zdP|*<PmNQ$S&I|o(`n_UYrlur*cA$n?AIw~gVaK4-e(>+?;ABpg9c^Tpfv9~>Cxla
zvxg66MA`0=Y${KZUO4$|OcW;Kt+cpQKoHa?H*P#^(%N-#KfRFn4l`pU`Yj+$fKXud
z4i`@<7a14CdfP+(Z4={H+Xdu=Mg#EDJj-`L8uEHXPo@i++0K#(=ol#u*k{-RKzU4F
z1n_{Y*Fqxl6Iw^fr5lkCEQ^fX-3^aY9F+HuMw1g0J-0L>e#0-{&J*6ar?)-%hTq@C
z@hxsC{ZC*&RGJb!T0A(n$htVgVCBfwZ?&Y)9zAlbB`qI3oLSt0d9S<htAs2EEJu&d
zKD~HwZclr>9e@-KQegIpfD@6trYP8~^D7F{G)d=Az0=djGyCE6lTT=HG}P<X0rFuS
zKNehacCiFPvF3iLZ^BgxHsMB55G^hq4;^46*-mZ@DzHkHwO=d=?hm^&R_gg=9+^kS
zx~q=opFG5-h%iX4J-0;1W9$eCJ<Jn*t??bWvF(9wFw6x3R42?c+p`=OU)I#H7Ehgw
z$3u3K(K8_A#DJn`u)s;8%nGYyz*%HLULWn!Rv;&J&=7i|FIDhQhya%4k(cd+==3gK
z?<HgZ6oV-EP^{eYIfi>&p5^6k)jh@Pm|6XeQL*rsu1}tu<@;)*ZIXpIgz32v;5+IQ
zl+){24%4i{V*o`$XpFUR@{vS4vP1AbwAdmnojW^26I#c~MhYz}#FITOl>Sa5H_M>B
z+g+Zu53wnHt1yb}3x`e=pbvhVBneV@`S(NvLs_$MV95{E<rqeW!HNhX1)~5Xl7^SN
z9J`imAC2}4?pQ(n*h{jk#j=kfzgRB8@^_NT>T1vnKbqH8D7Xphf=A4bG^dptMh$*E
zb@F46@t+{{5cP9<9s0pW7!Rx3^^sxRr*ekrT!kw@QvrbNCA1=3gpR>8m5Ui=H5MCh
z$6_$dNt_c}Jgqabw+;Yt76VMncW-NR6Z@%(M|TKii;Y?d8|CEVFv*DbkPU4ZQka7(
zHz0sQIfd>~?1l|(SU#7tE@33yGaA#F7Z5koBuN=Wh8byvttujWHzsUHr?LNR%K~TJ
zK>9LkKKH<QHej0hqMcmVc{oP{z5?ML7QzbyTb13|@*)^yvk!^(fZ;}$QR2OX>jDl<
zz!ydcA_Nv0ga%#$ZPMG~ay*NNV`~;4<DX$#_#TWW{M%r)iDA=uHTz`rw=VX%5sM&e
zwRhu^iw$v2DHhuS@sKDYIO!O>{PIB0t_H+p4$9zjXTZnd=FFe);DTVcAy-nULW${D
zkxwT-k#7CQUSZ6$aENcowW@HOHH8Dl6w<zGaW8D|jDa-LiSj})2TOx2XT(c(cq&Xq
zP2ue&!@(6%$?|;4VaihM13Wm94p|pN7pPq|W=+i&8|(}Te?79F!N&UGhIZIZsoCA}
zh}NDCZi}#2@)GW{rl{K3mo#OS26ZpTb8nwHRpOUB9fvWgdLaR-)H40~K6H*r3X78s
z;aj#wUtsZJ$&y=+4LD4wEswHu9o(7dc*t5@YgRUVkul;(2#OE9$uj{hDI=J+r87>-
zaxRQ7B)lZ!9YZmO57Wf;*~9-lNh9A1F<~Lv;RudBNrRfXyFc7A5Wtn;uPkPvAz27h
z2<}0*Z`{jpZx|=|%}!cqJ#5k{0VfewV$&@dM+429j?tGH$ue13+L~M|UH$*tFE1*s
zNY@xOgAcsbagG$xQmi2yVIR?l3N}hcU{#JaTpMoCvcSMPETk1{u4#!XaNO7(wD=GR
zxLERh%<!a?1+T{IU=yUUtM<hN0oG>6OUA1Oi9pClz~RKzB(>+)hoJ2mnsCiV9g)tG
zobkcmYkeLi_QnrxTJ(%pPlb9_>_0dscrRgqT}Wdy`5;)oA{&Au%iy8Ws6#KEeCPpt
zuBxs;paYI3oI>3mIr&6;n1<K)vxKgooY;C@iqARO(hG0`TM5uUzbqxDNNqEuSpxS?
zb&*0iCZMniP(J5OMYE%sB`A+C*h;T<VUT$Nmc|b<`6PBMmE>sxF+qg1cOkHqxgQy0
zJ=&Vq7{jtj31AwOSKjW@fc9aU?_7kDE!tY|$Sikc9R%hO<;w|aF&PY{h2-F8nN(t-
zSpvt;^cER{pTe>a4svG*Q>jo9&rxDvzRK)Id0f~W4vE`xBYr@`KNCDOyXQGSrwF2D
z8hkYAAs>p9ecfLvXP{?yW)`bdOLvM3(!jUp#hiTzBwqFntZAGSivUh5@6^0kYs1DO
zu+~<U))aDwC<73Vg_NpvfTdIAuzFU?$kwLhI_x1!p?_Jq$6Q<hz-K1!4Tp^JTNBWO
zbP9f`wd5R%nMgRfpz9SEPIA18cG*?4i<;lzx<(e>>v~GDKQB#1arC$<!Q&xC)Hc0h
z!K=YzjaTi)JWAGuoE2$4Dez8u8QhWw4$An~2~z~JwF-pJG(<sxcAFM+A?2;lsjtMX
z<(U4?T=xAW1AAy(DbnWDK<3AHMslXXxe@0`6+_WtAEchVQ?`OYmi=j6ehi|0e7^t*
z#+t(7>%`37<Wfp7z^pBwV`yi7BCOA$go6@QM+v7R^pw881PEJBc%*@Lf;4nR#jU02
zfe=JzLMVV#7N(G53%2}jIg$2BtYVmLcN?$pyoD%p=2@u0`w%LGBrw$7ffAmMdTY4x
zZO)W4Hi0JpEuKR1Frblgqprt@vpVj<k*l?~0Hz^ax-QBQR%(TEBN|{05j4pXDp%5w
zvnF2Ckg%hv_-hX9xNK1~I?7wCJOozEp#Vs}w@8P!FoQ|zu7jF9Af^yanq^3lVP7(q
zEOP1GyQ;eKK%5Li?00hmK=w3CaF#B_llJ{;yTZ8@@a&bu`2v<-5)Ov3wRWZ&iq>^$
zJ5<<&H1$QjpjJjZ2vI4f;N^RV!UWvr^`1*|rT9U*K&%8N|9K$YwzQVwiq@cxRBiJ*
zZg(0TxdPXkTc;wPsa-7uP88H-c*NFEMb(^V;SU~HE0yNQ$D7v>bL6ktF6IaYZ>a`H
zeJ#XqXA<->))%R45{!w;5?TtN2a}E=LKoa9Xe*lBY4FQg->R3rN?VLdGC<MSYQCpv
z)ijRTMUlU<SO;axUb|bi)|JZ~?0MT#t<fmG2ocU!oBX+Z3*3LdT#SQPSG|=jVHfA#
STlvlJD8F$TvnPVR-v0m&!q0F3

literal 0
HcmV?d00001

diff --git a/tests/Grid_simd b/tests/Grid_simd
new file mode 100755
index 0000000000000000000000000000000000000000..381b4a3e8b15d9f0db589c9223f0590bad1d1e13
GIT binary patch
literal 98268
zcmeEvd3+Q_`u^k?hzjXJkqw73=;$U2m;h>mpgExjCJ-S?P%wm$kVr^kGQp?_!33op
zqIj$)x~#g2ipsj~E|-vSC7@!!3sfYcXb<5MP$Jj-p11m#o}Qj0#IXDO<2(6es^7QX
zdh4yXs_S{Xs=H@yiep4@P>@Oa1e<O)5tY_Ep0T8Flc_w(D-&;uHC<qeH}y63F?E8J
z8UKo0ynLjj&2lj%4TCIwLhv7o|JY!nD2=B?P5Us)gc_ueQahzopZ60lA4+Ncb&{l#
znoL%fPh7IKez=@JJd5$`<<vsP;1g?+Ql%W3N9CI3a?P^5lA2`ul~idTsiXgNIX^u(
z9u(2>l$es)w>OomH>K@!PGiSiBs)!A5+q4Eb;|mbR4I2W%8{LI{V2^GC)XFqFW&x<
zG|N_*t}Do&dc%<G3Z@Mx$S*7_A5b29!+;xx3@j}gIEb~I^pot!F}JaTiXS6Dk<H!l
zPh)`c&peaR#Z!3IWB+>L^JBxVyz=i;A6UXuUM&7e2enJcvP{XUO%5X7Q2Y<afBn1*
zcf6GS;ikuCdg>2ZZh5(SV*D@P-T9<7GUW?L_vgAKnexo9zuc#I#eMb}|Ax5XXE%(w
z*dXUY3^F7BA~eK^-!*`qmjdWnAAoO0gN^jx89<)=Y&EdIZvg(?0rXrFfWI&RzbF8I
zIqEgC|Lp+srvmuxrvUO*0r>p`^vmD?e)v3qp5_4ju>thV3DDkI0qnUZfc}#K`u*ns
z{1pN8R0oi^1(5HK6V0f<UJ9V+vH<*#1L)r#z&|Gf@TUYA57!0IpC6!K<_GZqnE-l5
z2B`PD0Csi_p#O;g`hx@b`A-4rO$?yt%K-i_4WMU60QtWJu;;Y^^?r-?_A~V|)g)%Y
zBPLS@`!<<AK)u9|Pxj@%2K~fuOpy2$vixPT{1Tn~U6Ahv`HB=t-YUz_mF1_%^`^=E
zwL#LjO^b+^Ia2batjC%p>9IhLJ}=>)#(8wlczk<5jk294x!wvns{(wIxAcmaIdTHr
zD|JrGzV=$ex*JTHnbT($6=s$?vr3$qnWoH9vogo!OwTWM=9DBAWR;fYl$tW_>8Y91
za!O=&dTLTZQDIJc*3^O=DbF}3v)Gwel9M$pb4Jd6nb|mULi~;~<4cF2hT@p$Oj26v
zEGfD#W!&iTS@!G{7?WL|m04U?l9M?rzr<OVRX`<1k9S_5?Yyrz2O=?pQS9FQX*q?N
z`Gxt;lvzw={*dCFoEeN5?~J)IGZQrz-J3ZTm1h+c6lF_NMc3yRl$GWwVmD~Tva?E^
zB!%MO7s0HYteL2yG^fOA&jvRpJ1?sw(^-<0?<|EKDdWH!JT<E{KU<Nur)Q=|XQqt5
zAv10C_`#XXbP_7bDx6j{6K$Q9pPh3f%x8w9$?z=di-V_U<`=>{l9l=raVRl8$2lvj
zKyIuvHovG8Z7a<gVuuToV-)@kXlYhKj>;%aib+Wuovk{TJXv~UcG1k@tdbncVZPJX
zM;7M7MA=ORrS{ymwm&*EJ%%{|RXSNk1v%w#3py74o_|CB^ui+4EcaPSY09{K*@Rfh
zlW==};dIIQMqX#5Bn8bP?~!ja?Qq4|;+(>nMbl_>rQbdV0}G?5JZozHtQZ;!*))nX
zY52hLxkbt#n~Kp~n3<jB%+5o(@lLdJTF10g8sCGWGskD<%q(`^msyfijPt9gBx`yO
z)jqwvJhM2bq_n6os~{iQv!a<3N{5u*n^lZ)MZ+TfcC_w#ty89!<p%Z$Yr&B0qB5s{
zS&ZWVWu<N@Dkx2%ZYaEWCM)dYz(FyYnX_i*8#oaiT0}!rK1HU{8Bvsu2?QD55t=+m
z?(L$%#bwSE8gc3AXxcgTw}Dyc<5EMb#yjD7`SdTyDV*-i%giY$DJoG%yYI<^p+uu`
zT0z<@6DG!)(wvx`HMKO?MBk=@qUi;cGd(*yvy^gUOqsV8-kV=I4U<@D*~}cu%$AB3
z7nSChXU@!-iGH14XhMV0m6<e$VrtFJ$}cb#V|Y7r<!KIQe0jdpRGL4X4U^IwrwL`{
zB)j-N6V7^P{>+@B+-X_&kyuu?Gry?NG&L)G2Iisc91@>dG%Lr;$t=BZ=2UbjY9O=Z
zS$$?!exWJ1Bqzs|Us_yNnC&!;OiIceJTTfc(qT_b${aLs(7@}}<Cr1RVMYc-h764J
zp<-f9Nf{XliT2ExfrEW=WCe1mm>6v)Ax|)_mV((gr2r><q@xgAPbo(LX&h5-Fs_n2
z%O!%@Q5de-B^^Xd8ZUrE7`UB~UdW`w@O=?z;?TeJCs8Pb>#-oy5X1<|bw}FO_g!Q{
zS+jicN%_4r`DMNGFEF9Lu4;a;=~g*E8v34o;u+*zWa^1;(kaP@XyiLVqm+llB+QLM
zBQG%7Wcj@2XF-RV2Fdhi@6CYrPNoEzp3mgXCWlOCG5Ju_9Ww2VdJ>oH!KQrB^l83D
ziW`W8;p#_;W1yT}e<~SDTta-hCRIL4oJ4%OW>r2)TtfU?6edz7ej<LHhOfj)#J@?y
zSK<}oCusOe{6zd=8h%8))Gq3I@~85#YWOsIlux9FFO4=SCtAZtXr_K*HGHaD`NV7Z
zvWr<3ui=wT$|p_3SJqM}FGIuE_RACv9|KYS<Z1X_RIbTXtl@Xn@XIy)a1Fmg!{;>o
z1sXn$TjjGz!|$drkv^y4Yp=JKX!w_C<d<srH*5IIHGEn_Q9dg){2mGuX^n>eCk?+|
z!|$o#H)#01H2j?!es2xGQNvf(>L^>(@GsNIAJOnH*YKM)d|JCuKl1ehjS;KL#2K&Q
z_tWq#8vYd;euRb}ui;xY{3|v5NDben;YVxuS84dM8vfN9e!PbNXAPg%@FO++G!6e6
z4L?J}zgEMaqT%<~@bfhM0UCa>hCfilFW2y|)9@=ae44M7&jJlUQDGuoq~XVC_|Iwh
zgEagl8a}PnD4(So{t$(Ubh(Cqy@tO+!@ohpuhH;t)bRPr17W<&{A*-C6F<MkX%=5a
zCSSQRY=cG!ho0;ALd0{`kND>L#^Z?SJQ8v@dhogGYocjNbvH122hlW@x@#D{m1vqm
z-OCxho@kmn-Afp~ifEcL-HRCgKG8H)x+@s{7SS~P-NlT4g=m@@-BTF-0?{-jy3-i_
zB+(ZV9na`TiKeO19nI*6h^8sgZDsWRMAKC0wlKPcXqp1uCPv>)G);Z(BWJ<5YAVq*
z<+&RfeFxDr)wvrOJ&tIa;@mZi9z`@wZSLiaP9&P9F82~f4<-6iq8BlG5YaS+xhoib
z4be1pxr-Tn1<?^iPhoT~qG<|qr!o3sqG{@K$20muqG`%<M>9H@Xqu|rRz{!vGiaKk
z+!jV3C7Pxtw~5jFiKZ#ZedG+ae=pHiq8l0gHPJK$xf>Y0gJ_z1+%=5eN;FL^?&XYL
zPc%&_?j?*~MKnz%?nR7#pJ<vw+!c&|i)flU+{KK3g=m^G+*26+0?|=Kr!o3TqG{@I
z$20m-qG`%-M>F~%qG_sdTN!;n(KJQ4EsQQ9nx+P~iP3ixO;dvV$Z6L9MAKB@Ze;Ww
zMAND6Zea8{qUjWO*D!h%(R6CNmoqw%Xga0cOBg+rXgZbMi$JHXdo3l)B+f_G>u7Cf
zEs?7la297U|D>&}Nkx8+nolIl7w-X&n^41bGanv=v%1cF90`5aMF%CA*4C?nyejye
zB>01mV51_K;l3P_xdXYrRZI_^|7q)*ss4R1RkDvr**@}5%t5=b{VrbE%UAw*BrV-u
zS7Ysz0Ap&Gn<=S_?-Oql{{z9}xm@35lWD~sH0?Hiz)mE5Wpg+$?7KI3@`e>uAx%uj
zC$FO{Pb1g2f_N*4Qsir!Lm<x!>-pOKx9~v?{PyND6kbJw{QMm#%nO^P@_Xg-6~o_*
zf>+AA-G-9l9CYa9&E{vo3Ocrd+Q(N8-`NiW&dXI3y5F|$2zI?7?`fFqegH+dz9x66
zl-`ywA>sCf+Y-{p^Yad5VA`{<Lk4|r5!Tpf_lp-ku<?{u%XGCwi~YVwm{71CMTE_E
z*9~^p?P0v|6(4mXCFVz7IK|hV3a0Lis<K5>@wzHoJW~2NBOGz3xJObk3yOE5i|oQi
zS%E`%jY(5VTUYk#(8atUmE@z2^1^4N)*0pyeim;<FDUsWO9n<DT}zYmDym8>+=j1R
z4YG-I!tB@HOdWj2UVBEd4c89LHVYNN$Js44?hN<HzcFKr>rewPY~qDwHYSEN3}U0I
zliz_`s%%t7X}QKYUN|g%a0*4MlV4(TREVrw4*fjb$NW+KG_#hkUdWV-Q&HnOnwLeM
z8YXP!R|J6~v+P1Aeg#1gwahMz2wO*&A#1<I$XrD;lXwHw<}!0v^)vCVc#(HKXTka`
zKmQx9su!TB;(=I`)BI>sQd3V}*dx~ePSPQej9yDX1D>aWTYH3Q6;F~bp-#LQsZ@`t
zGwLBk5VxIV^}J#8uV)P)P*=~C7VCLZuIDZ2)VC*Bs^=DidhYVCr@6VM_FN;^lSn$z
z2}`LVmjXbESVF`Cpc@g-5pfF8L&PE??z%!E77$U3Nt$vhh;U-Y<f~`f%Ehl^(Qmr`
zc%8M}Joid}KP^Qc>iiVjqWyQ1-RO^ppi`>n616{Wmg@PFK|Q1V>uEUAlAnU*dU~n#
zc*ob5%tG^fSNON*0LG=RJ$WtG^8(q7_PkFz;lW}t3>VULlqU{Mk&dQ_>7%8i3~|)0
z(ovciKS4U;#aWZ3qj+&)w!ilyC2NBWtfhTvowcieYso7INEfUPMJjp!YVtnCQ(_Hk
zr}=rzxB9bgKTca+J>y!er%JA8F?8xjQ-)Mej6ppM{p;EPYfJ6vBiA#8bi(Jc;@x?Y
zFQdicd!(aCan}9PkyV^s>F?iq))MpItp4rp<ZsU%E!wk??0|n?gHEa4H>m!dBGnUT
zP|s3KjXM7xJ>F8gZE`)s)Ox(5yO~*N-rm>0J)`{Vd9=lP){xC;&*!8Q{*4fax+Gs(
z#KxzjqcHK|#dI{xCJy*J#?jJmO{UcOzY(gP5KkSWf)ZAU@4f>$*BS|Z=6i6N^DSkA
zLqNwf&Q@?z=l>{S;)ft&YWR5@Gmw90!lZ=B33nxAPTpV_>fBo}W+{D1PT8d9Kmqr+
zI3>kIIvR>eK#ZlM=jdo49gV>RD%$9186EYaqYD91qJ@s$rK3)C)CEW`o<54ByXojC
z9UY+~k&eb77#F{zqh&aP9cqg&Kw-7TYnJI-ydPzJTf7q-wZ$KSa0njpN;q6xi@p(8
z;>#V6agIuwHfh7ZWRpn~?^GU$unS-C!mkvpR6l^=Wt-S_x5<>t9j-aX&8=*NpD0E6
zz|tByB@`$A5;-NLAi6?IlZrtkO`)$ezOM2Jxbz{3Sd1o0Bjq$gPQ#Ewlt)4zzAmID
zECC%~7RFy&W<r0~o(<-OQhp`nCa}QH*&8P3iIX>)Ob%BP&%0Rlyvs$s>Zs06b5yTp
zYNKlT+TVj62q&Vp&AJ#+&sQF<5U-NUHFYh$XjFA*1us-mL8u`RgqkIk=*m{FW(eut
zAl0cU%cQGUlV_kpN?c^8uL=t<%(hv?rHX={r7#E_*#auiWM&-R1?Ra&XZV&6r?BsC
zrSyfRVN$`Q3`g~=^-|qZ;%e}#o0T1?l=U{7BpucNY?O*iiAyBPr}Zd)LOi_QujM3}
z%N1Sr!FQ}{Uu*`Q60;?BehpW35&F}C_JuiIsqjl_9`Aa)0{N+~+t6uuT6ot3X%5%h
z<WzW^yp|H<$z>n)nkC5<x-wA+U7IK*e`o@+3dFiZSLj;qy-E}G3+y}5g<Nn#8Hc_Z
z=lP@)!&WCz$>$SK3|ohyD<MQC3hdiec^b!w+<T$x1a_E$8D)Uo6>3X#rLR1VjPw<!
z6NT<8?5<AWtkZCsE|ac*m~dSv3sp%<IrP+}#?^5T4abE<s!;22!3!7}Y(%<Tw5hMU
zcN-1N9&`#XswMZ1=E<Maa3abBYyk!*m2gzQZKZ;)N-1$g8mT!f+BIq}^HoD7(Ajhz
zLXDKT;{DY0rp^wQ)i_+qHVc`w9n*Bv#f)Sp4cSFyQ|E8ts-D4t-DQdS%^}qDm5t3P
z@EKS20^I8mw(*reHS^VTn)%wpAu$JePo1+bFZ|9|ejilU1GCKz2%Afx9mn6`xW1{!
zF|Mc={>NE941)^rHsF6NU+D=h#R;;xDVeW!ycNwy9fHuU4fvYah_9?8_=?KqV&m7-
z*GKfVk-k3RD{JHV%FRd0K1FG$+KK<|l=C&|<}R+GW7JVxA4Es>x!jyj?Qy@9g$m70
z<$QHvkUh@L3r%)z<St&=#Yg=D<3sqUW6sW)o#t<GM)H-v2RVD<JGcx6{@N7Ey*KQA
z{J+I}aQ$?MDBiPKXyz+Vg_c|`G^^L-?}b7v`F+SIzaK_lU0`RZBW_#Ck8rFr!S0Gq
zb#+d4b+_4Fli}j<b?~$)=Ae8%{QQYw?<NStmb$vXo4~!-{Vk#U+d}A@sP0`<?1=lM
z<U1@1Mre?bUC-?i(_hCBz;)VzL)sOB+(jCEn0H_)Y9bJiXUh`FwluC=vMrvcc?zE+
zx9~!;Z3<tPY|A5>sN&64whAKX{uYxs=UtNt*|EeSMX};C)GJ-(O-0STuz?pQ*&=bJ
z=Io4VpUKrFQJ>3F!yv`L(&OxGufruGhFRT6jX-avW)naEpfilGL;Q>nQjNs#2dN^|
zViyk5bY!J<3DhQw#xqeeK7vqq0oZFR@U<!$U*fPPj|bQD^J{ol70a`8DH};R%u(&M
zWzdNoAM=eDWO<w|k$iRk2sqav;A}W4#sRisJgcZ0j0Ew~uaH8@>MM1LuhcY1*{cgK
zqfV0b`~tsg^dyp=$_-%&o{b5Wr-NtD=Uq-z?drt4(lOZx!Mtk(l0?iq4pim}!ZD`Y
z1mZ<tqzC_lo3Q05*HF(zjj|r&ES>N_CX}17gNxjGjGOrNF>cZ>q+5@188{t=_S=QR
z1-1z8lF7hu%ZM-%?-EcJVyHOL>JXZE0hii|k-U&3U$-Yyge#2?c$daSG>_GQ=is2V
zFC$Mpv7bU7C)lprMHUr{>5ydWENEW5ILfsw`Uv@R{z0y47luv6gX{`FC8?<wjq0D=
zq#y-;Rxj}nvrBxM*m@qI@lYqmQ03?dR9-Hwq>V>fxQY8TVI4G^SggSd4x7|$2RYZ7
z#aGreOF1|<d#<E)q9a^wn6QwI4VRQOb#{eS-|7)8G;p>^>ex9$<Hj<#4=sT#K^5Ys
zFcUIBJ<KAMm%C_`D|tBj96^iQn#&bpUK)BuZP2kqZcdk*xJk8Q7q|xf#tp3H9)1`;
zv<thX%Z3r;R5NLfV%{?cclQtHr=tWlIbuxpkl4}H1;h0h5$}lPlR!%&dq^`pcLH`x
zwK~+U{HMZ87Jd>p|4f1#){dcUMB-@C9`OlC03!Vxm&DC+3^!5h214W*cL$uYN8F2J
z(xHwPVx}_Y9-I;4R5DecKS0W-^DI%yms%t8u4NJOh;h!+jg}Q;t<h+q8k?BLIx&P&
zxX$8{x{3`58r^hK*emOl^N?E@hxoq9reulxe?mC^2zV-ICvnO#bQ}UZW1)0@P1)U&
zM{gyUg;GCevA!@5y;$l;hif@4YT>ehs~U@bOsJTHG2bcS(GghIqH+BT)RKRdF%O<*
zS?fx<9XBzp6ppZ~r(x7~Hh1E*nF|e=im6FfX~1K$*dyjKHqEb)xf6tX=`x0LXbO~+
z-9qHNjZ0y@265c~1nmYAn3{r{Fkm;kkK<ULkvF?{YKgU4;x!QB+xy_G%h*EXZdw%d
zIbm_OI&V3*b8`Ncxn1c5Ttv#rewxCqO_zy@RM8vgCtjc)6{t3r=MWB38)b)GBsV!i
zYO({RRR7%qA{B1Mg{(&$L^&u2aH9*?D`ir4Rd<loX$A1V7d<)}f6oGKeet*dUlnBC
zOAToif7d{&Mf`n{sc8v+?_(0}#NRbP{9*iE`41o5ys$x<19kZOl&{o}pZekNJ-!m_
zA<=I9ML-4U4F(L?{3Dl}n`jKc88QG*Y6Af4W!nb8XW(960Dcnlk2s6iz5x6gSOEZd
zr#}Ga>?H*O0Qk8dY2?+3%ULbw1i(|y?PNCwU@RTT06Yd}`T%gN_`47#TEX8pS=-GI
zb!i9w4)~t@76^Yc$zgEZy{JKtzk?MX;qQn>64c>uH%PRCzwe@Y1%IbfGxYfS3E3ww
z{M{?p;e*fbNcn`n>!tiQ;BPER8Vwc#pBc=II<b;@rXBcu_Z~g|W)P=!{QaJpWnOh*
z8}auZs11z2uQJcniL04b6;9TmkcuCk?}+fjUy-q8{QU-mioe?^$BVz~i0okey%sv%
zm!Bv6CHocpUBA0c`0J9JEaUITD5d&u9S9YF-=`ce{=O=cGX6ddlJNHc{D1!NcMj0j
z7k`&MtKjcc>iJgjcNh&SS}YED{may}gulO&$J%}U`|Tgb-?+DY@b}SrL;StOS85uh
z^w+;{`shg{J-*k!6l}RV6L29^MIby939f;0J&fyLtRRoTwH&U8g%J^4WG60TgoTS5
zft=3%mn7Vr3vt;iU+~i6C8^S1lyP%Lg+YqDBo%dxP%m{yMDhlgx=+E8zDT=(khTl4
zeUUaFtN=*+oj=l^`kE93K-z&c=m8ibSS{y-w7DIDwEa}1{SjvRAg#Q9)!OxLpKn^l
z+c=0|^Z#OVJMi|yT~sy@-X12+a9?Mndc56$1~DGt?Uy@AP=~jFgG4KMdox57ygiCz
zKfImHm<(@A$W%QZ50~-@Z^udbZNS@lveszG5Z<!!gf?G7n%i@|d-5wi-ipL&9dGYp
zePJFI(nh=uLNfy6?O^7aI&n19s^YDK+OAyhJ`TeS@V1<>WxSmMLdDy=D94MpV~Olw
zyj}ezoE32zirE!t$6oJ}{R-ZWql#L;-t8?nS;pH*D5d&uEC>~E6Dh}ww}WL;##<Xm
z!rR&K|M|n)aI7Kv;_Wb8=S%C&N2wvL;_b(fY7uYWW@=i(TY*Wm^LqE=&;KyqZh6HA
zCwtWz;_a)xQpaoj@OGZB#12UOA-tW`3f@ktmGPFqEFj)as^u<8X#?I~`31V$XKDE%
z!rR60uMTg22dnLP>-vlo1i;(1R4FFUFIg@BOT2Z$%=X~zGL&oO`gRNRfccd(R`gT6
zsJ+>))%E4n9pt@0csrNe2KPOS8uWNOUf~hm=58lJ9p1)3q7}T|hUyi(eTbT&$J2vk
zpLBg2%no(y&!0&7gtvPo8`^-k86;^mWC(BBctV?BCC%-@+eaJpcw0`K*6}uo)oA|u
zl+=%{t}j0UwSn<=1M^Is_zi2UinrgQkct<58R2(*8>W>x^{EDL4}s9&?H(dK7;i^I
z=Xt|hvR}d5z1!M^w{Oc$mhtugN~!+a146~y9hBq6+YK@)<Lw6^32(cgJI){8z6^Br
z#oLb;D0ur2^=hklI|))P;_XdLO-p#&lS#A_Zzp~7hw*mei#~Y!7WN0}MAoe@5B8O+
zf|Tz1wwtfSRMO*neS0qZyG6lV<S5*?QQWYEn}}N-inm~)xwz(6ZelG~o*THRao#m&
zY~^;t|HM!(axb=g8@b3PByQ{g??bvB|JV`Ez^xKP`~9Ho0_SLUfl>BW+7F(H{ovLW
zaF^r&zgycH&PK=h;`eOA?{7b*QSY-g90$p^<M&4&k%9pDeU>W4WZZ*NpPeRPY0F!~
z9elm~0nBU<ejh-&R`B}*=0fw{U)zD-)mzCif$)1BX@>j0f)RTBeoEmHe!sqj1a<g*
z4<uT_?{J7H_`M#-e)v6rG1>KUBAKek_pVYt;rA6%ejD)nIkMJh$Pj+B@q{-2z&z8A
z_35qkdi-8NoYwI>fz@d4b-a!Ey&Y-;<97)2Or6-9X;tyN54BynUe1IDgX?9UvE}RK
zn?R`7%L6ILd%fJ3$PUJDSKWEUZ?a#(?<=UH)~}bFAVy7=@w-2fivRk8Q1Sa>Sy0{|
z4v|S2zmIR0@Y@0ZpFjNm4ruL*-(BV@_`RMwt5y7745=3J`vIn=CH$VqB-)AJi#PpY
z{C*y{rEBi1+K+8s1N<)amD&g?9e&^HEAa~HX)k`u`@`G>^9<ZQ!9{5{h=H@0-cHO3
zTM3YB6+lkW0J54VT`RYSwI|?2U#zVrtSu(CFV=nnRsgKs>%TUgxsenEz}hD<!U-(i
zVzr#pwesB^u{M2`x;D+jOdqVx^`3F9txwNJ*;erOdFBOk$)R@O?PVLtgMsjNEV&LY
zoQ@jwczdP7BfPy~JqhaYHV6`};O)z(UcuXO)C@hIen|F7*ULNPI(+tnUy||(Z{L^l
z+km%`Bxy8k2ygFXX4HwL%rou4+dFFYc*_%~b-dlq%rd`qppAH&1+{_k_9^C>I`M6$
zRmIz7D5T=W;Tj_Ru9tT*wv4wQfl%>wE#-Le_B|py7;k$+r@Q-kTAwES6})|aU7PUs
z0lCRC-mXC@)qn4SQ1SK^%JE(=KP8hg-d2I6>*a6ZfA87f(f4B&0$qLacJX}*-j1W5
zZxwI*)1Yc;Klmc1rX{?k_d2wDKe+$eKa96g3w`kRer)_2;BAPnR63+|c>6sbEYslj
zAky>4@%Gf)26+1uF*^=#AAu8n@%DDYTNAN;@pdv;ZO7ZgYe+!=yuE@3JiuZotL1-*
zw>x2Gd-1l*hppmm6hyEe9Q<=T@b;b6R5lRa?jg-^;R!TAkGKC+c!al`SCOC&Z=Zlf
zD|kB)A`0H_#jzjWrZOhGzMV>@>hXAhluvk@DCM^SZ&#4DM#F~iwux!P{m+!PXMKAA
zN<H50Bu?vio5lLVJak_h@%9&}4UD%}%I&y`X;txd7`0uwKAQ^-26&st*fQSU2|~r&
zF_h!Q+gph2V7z^M#d*YAvR}d5M5?Iu>)VUuCd+v1Kq=LKw}4Rbb`a%w@%BoYl<~GZ
zNW$9!_&*@t;_8$=He{8bKZ>U3kF37WAH~!2M?3^^M>(EQ+KeZX<cE)f@DMlhB3K@F
z0wcuhN))i&N38(Z;<=+vOnHd3N&E@pGxY4t_$l)4@dwDT3!74Ds30SO>!#$qz={;m
z{oysTGMEx-qX|Aw^CZ#Hf8){QE%R&0-+2x$<#)ciz!pol*;{d*;f!?%N5z8=F-)_t
z=6omK^(s9PI1LXQ2|vpcA9*Dlm}!QQgm75Sd{>)!NbHMD&JqG+aZ?l?S>h}v9FMf*
zCUB9RbGgkmc-U>imRvjo1WwdgEC?5c*oE(~PF#QJC@wQI@pSevZVncHC+;F~Jcw6%
zG7MBhE*FVM>?Yxfxs1;oE}ISIFA1gZKjGP<1kN(jDtU;#7ec)v5Fh&w+z!83`08|9
zx%9Blg~Wpwa2=~W>@%HMc<+cbCt-(iapV#52))lEnl3Ya-{;Z%FVf&YY188h{b|@?
z%1&mr;BrBI*ayqG-2B^vO{Sx;H`GSo4qF(KNh!pdlo}?EMVCn)8dy7Ps5IkypR=h&
zE^y|BMrnT1?t}gSwamy7^wN<#q>|nzJ06irPBAF?Pq`#l)r=OSwEKb}8rplrivzTh
zDVN%Zc20t7Jv2oqyqMp_{vR;cYM;zGES`Ysz+SLZ#c<`VWVIe}*04+f(of<%cBtDh
zU999E60_K$aj$MhBbmdlCx_Xoy;qQ>2E8h^H=P+%Cl)|yQ-6{_`cGqdGW}{<`esP0
z6I22!P#>eOLTQ}b4&fL*y>VD{GHohM<PeF;0(SxK5i{gGtMp`toyd6`X|mEh_2RQc
zaS7%K`AIQ{Ef3G*<*Cnz;o;fG?>_?#)gRa@ph8>)b?!Xk2%Fri!ND0xhs$wD575x@
zK`;9Y9MTgjVX_!jhJ5H}K#RHWL+9qxOeX_z?OEWQ*mhgz4&d?FW3hNCjBM&BhLhe)
zV5=BPM@e*a@*g;&SCzp~)z@Djo39*ZGdVBdU3l-qrJ&hdO1=(bneesAc_AT2nD`rX
zs37YGL2TLGUlWN^gV24t>$5+G7u!qqc;T6$;KjeFj}7qRCB`?v3mQ+xcyTu?(mGzO
zU)~11_;9ulUJQK64=-NyO3+E#g(OtGn4`@+BpzPuix;^749w|-7{ELnHOQy|v>54!
z7IC=QxkJ%n3iY%XE!dMO-@+?8w1{H^!gyXWK#Q&Kkp^S5ILlOF|Bq4%cD3h8YiQ9A
zrUgcer<md9g0IgREhb7O2N{$Mj21t>Yurj{_GIk?S`370JzDICe;5yFarPZ(5jOdv
z#TEzzMvLC$UYwp+lGS>&c#LH-v>4A0b!btj<R20TvqR%vH9(8wWT`=~O6|Rx8B-_T
z45boUYz!bzrZ=;y>qHAubq2cx4pq_Oc9hnl#R#TNMT;AWRM6rYIZvLxdK1}kXt5CL
zIvOovmqBMoqQ%cp2()N?8+sX9d`U+PEk33rJzBg0c`aJ}3?UUQz6QZpYSE&eWy|jm
zT20wrv{*(Y`ZN^XNACypdB4FQ!;8Po)Z@jcISO9Pr(QO|i<yjXfEVe=Fvg3atVrv4
z@#I@=z>DfxK6r5!(Sh#zVun|O@ZwvNQ1N1nHuI3U4w-?kFA}&6cBuiV0FX!*78YDR
zWYo)74Bfie0$(a*Q^$EHfA+A%9yC!m`440LYdlR1_P<p4<W16GJoyLFsKcb#lZ_z1
z$$w&tmkKtR7I^Z1h8b=y`rLo=|Ec9mg-KG$>kUf&JK&|l&r6M4DNTl~eQ0M4RO=`I
zgYb_s`3I2}{p9~K1OiX~my>&Owp~qD>nHywSSFkNZ)b<P$^RZD|B!eCJ2dXqE&i7Z
zC*EkOy;01VIxzuCrOCf8fIOLgidBv0ACRh(e-F4qo%|=Fv^4qC{mNm|^+GbKRMwGh
zWQ2URa1E<W4PSc``G@BJd$m^!3!%<UZ<D3Vf)2V`h<zP8H+10D!p~4h^Z(zWmtN8<
zeo064dSCHlIvSw6T6hEUKJ!0>yz@T@-Tcq8mHD5tHS<4_H2<RqJ9_?~?l=Eus`Eef
zveEp{_(t<TG6Ky1tVkQ?|JT|w{}=nt|BL+Rf3F0a|4G6-|7$Zb|0A=d`G1n({67h=
z!(uB&-9lTM|Ec4=^FO<Je-E0dpZ{6^8qX9)^Z%=)A;A1kqmJf(HiFJ|{)cHT%>T@A
zbJ4bQpZ}$j*Bg{<Vg7%`xRvt!&s>K2AFBQ4fA~k8|4EDA{11T^=6`Z8&HrSz-~7)q
zrTL#7>gRtYAM-yuH11WS`TuV%wKs|xgZUpy<@rB=Je&Vn)tLX0dgp(*!aM(?wEz51
zD%JU)5z73}D)Y|&M0Vi(4|N?r|1W{g4xRs@kmmotLa#Lc(~&g)(~;l&4|$*YA41;w
zAB1lHXW7d9PuZIJpGcbj(Ssd5|7ZEl{}a{upL*G7{%3rn`5zep=6_bC4fFrYZJGb?
z_MQJ9^`HN}5^Vk_3Ge)`&BXkV%mDL$QD-a{MUG`RpeJL;0C%8o-h%Ch%{3vQYO&|g
z0D-6p$|@4J9&l+;ynPJ5b{>KVHXSz8h6Bk);{7L+@HW4U4a#0a7q-`cEsekz5LZyo
zdjW*qJ^nI$qyvys#0s$2U;rQsUm^|008+!oTAjF))pCvi@&gnH29VpBO7p<_a|V#!
zQpw|W{v`tg$m@%ZTPXnuYaf8}2u#)k$TEe8)uWmhnQMJlk6aK43?RFyVw|q~Ams-j
zX^hDL<Wi&jXeIxU_$$-XW&rs&S!&R$QhP<Fw@x%6l>p@70P<w|Dp`65Oi%&j3lvZR
zB!UsN*Pyw82qBdUfc*9X5i)?diSW*7UlZAJ05Tf&b~J#Tcpf@C5<p%@n*ktyCA|zl
z{z69#KwNaB2asIyk!HK$bqJ{d@;nG%0C|LE%K&meWqaK{gGhAi2KWiT8_|L5M_G63
z@#FFF3Vx)}XfVK!0gP{eAKj5*j2|b-7qmgt%Kr4Y=h}cDsndM$V{Mfmeq8I7ApBTN
z5-NU#YcmgtbCDSUKlJ<43_aK;H8!S0@IRV1yf9&Qz2jlB!A2XnxS*hKM@z795%s4R
z7TAr>kD%!~Soo1x0bs#kQ=#lx(qIe=Z?X|oC)ToB{9vKwE$EM-I4~^S%v75HwBCPm
zZ)FQQR4Tc<*1u%nO@+swF>a+aO|td@7Iwj8JuLi1;Q<!j{0nofFDy7A5EvFVQ^hz5
zw?WDe7UCF_!NOT4r@OA`spKCLzhn8vuwbyM@K>_bpjV~#ZfAPyL^qU5uuv30o=ksL
zmR=2M6&BW_fC>vijPPzj|ME1|u5K#qB~rnJFXTLVQ(-fae+UzNwxDl>B^|w~@WWHk
z*^!$HPomAh1Q+RLn+o&jh;1s|OGo-mg>mE~4JJGZAr%wmgYd<KQkE@aLM~-{F<}Cc
z=+pP$C-2Qm?Dvp55FZk5)8j)?s)7$!p=Kj|=*sv8_;4JpH^zsr$uDidhe1!a0Urk3
z<%16|&+)^Di@Xwq53@-^#fL+a)#47^!kCH7Ka3AUE6xcYn$S!QKCl~erxVZj7Nw2E
zYCAsM_5^7##s`6oraJL&td?`chh<P47#}WYD$OTW`_J;N;KPs1VDpCtB?IF_(IVqk
zO7kUaAMjx%RO|7fOyL0^gvXg{eeoe30)g@2b*dOA<hx|Ge%+}jV={dBj>+lp;djPF
z{vmNKJ8UyP%q2?=dR1!gKUlqW;ua{C@L_BKc{06HmVOb^Dn7i70xCXy{}`LF)LWFc
zgVx}~dLk8k_?Mg~<HPGjb{szZ31)RPK5VFl&W^-~60{lkFoX0me8{3Bh7S|yNRJPL
z$VVD{D1nfQ4>=%w@nJm6mhoW(Wqa}AMk3LtkHb$b;RF5t312P!)(yMcw8~av!j#vo
zO8Tj^6(I1T7wuQs8gN8EBDa(Aa134}j`7C*BgB&4xZh<gK%<KN$QLn5`%<LhkdX>>
z?5B1L8(4kvU9Vp;7M{RlOlZbkw$lBuA0h|OAApazx|76vlkql6Ji5<yjg9VA)ZFUY
zc_H24e^7d9KIP%1`S=wCdc(eJ8Ee`u^Cb_MOyA&6u#e!YZ)V%lcA3pGXC-k8Y-!*9
zvdjGI{U+0{+MVC;UdUb#zpEAp*mN*SKLvw>3VyJhwPlZc_fDQmlm)62hQ#dPPvD0d
znu8M9Zzednk<GYpsIt|g+)=ufI2PnW=Gdd$q<XL%wrJh&kKk0qEwYd^kAo==;gs`U
zl2Cu2gz1Urq8zp~>hTZZW-by$F3(=xKGpmY9`M2HmH9K4&c<thE$O!Squj*xN3j!*
z;!m?z$M2SYU0`R|Q;^%e*jB-8gt2tMdp;_AhVDm-992DJ=~3=3l$!`!p}}E`MMoUv
z?x;P64#MLwp^z7^zL$QML$Ulhs3gm?aQ&*Yx2vDMy=8l^3}i2E+})jQE8p4mR<d-G
ztr#uB4?mENuaJ$_Hrn_Q?j`Wrc)VP)ai-4yI(sLN(EE1=3X*?IS!&?lbXy+m&D+`a
zZ;*h!Q<%L|n7yN`hrHVccYaldA?(?dP<bYlTe$xWo$KfeJcx@2m(pz!@O&_c_)r`q
z;>!|`mtk~Eq&H&3ccG&$>?asN1`%b!K?0NQPDe}@IxGoHwhtE>%w;6v)s%@=RKQs-
zz%aqd*Tn`AJO)hXAsyN1Xav=k7|9N)y2SqMu!{t*-ute5JPMm}o-U!mgr8cV-=SH{
zZmC!(P-~>uA$xDY{)xB&qMI51cU9cGfRq?VH*ZtbK>ZD@qAJ-0`3D#9V=&6^EjVxU
zxeL|jw-uUUHhx=y67{}WOfT|JE6io@bEaR&5$C{=n1l2yLdx%}z#o>-W%!;H`mX$(
zZq>aIl|1CVnfo|vg87s8t>{zVms;HpadvO=C3J7{pV<Ia&wsifxcm|0=1XC~Tg;br
z;(F8|)$0Ax1hSX$5PLlNFgZ&1%+O>A_%tEi`~n7}@{G`0vRWVSM0v}|Kc(=jkyx-I
z_O7C>7#v(9*yY>qM|p3%cV;pNTuctQnjGNUezM%4`w`cR%UIQQ;zsx=!BdAHrC}#g
z&RKwSH*wp10PzwWiyLv&^a6f=sVq-&lJ{=;iOjv`LI3dYnk1!byg%;Q1Vu*1NS$$+
zY|iyer}~zL8_)tZ&bkj8aPu+m0wB`O9Vf);OsBeqmqDZwX^m!i@^726H~q&EXWm8v
zcg_9a?!RI5x^aW8dK33<9OEQ&6CAOV4fVP4^KV|_Nf6>&Rq&DPG<4DlsXKw_2N1@b
z+lhl8Kw!T>VuF1lCSXrfS8)W@@lW)FVh8V1ebWmfs&6JkP`xob4Fn0wL1qGxid%-T
zJgr-<APz>v{qVJS{P<iipNDwpbc%62WV>03hrXun@{Nb;7|SRgdIvfBc<6b?YbhQ&
z@lfmWPzat)qIhUE+$6<AzuqHp-XxA35A6f*|C4wq5vH^n54{67`^H17izU3?z)}Oe
z*2P0jko&{&P>(`wtg!$p1H?mThU@LUNVfNqK=!KfPy$)n{&=Vax08C~p&yId6c45P
z*?T7nQaqH+QUm|0@z4{HXkR>3SMXnshbH;NLz8L|`h@<Gc&O7nob&p4h}~i|y8=0W
z@z9sVZF@X4eJ&|6j)z{Psu2(UlU1af&O16DDuf~Ji-&$>O)!7>ra$Pm8V}{oF>byD
z!7ngh;`fhHg9P>7c!=y}Jj6o}JxGqy#Y3qO(8WWKVlXNZ%)4Z@J|61nEhEQ6j~Iy^
zh1E>88V^l@SbO53JhI%N`zan;%&M*vKR|8%@zC1YN<8!djsnC(cIIC56-)iQ#vmU0
z28xV~kvgNRY|d3or;7PkqXjDF=R<>UJT!snRDqu-QUU&(Se_R6qlnXS@z4tooX2=*
zIBZ2cbOY59ARf99BC2mvA*jYf@gOwuPz;fZTWl;(>z0d&gAs8zeEr{xhYk!gj)%Hm
zuf#)Jsk?mRq5m+JQ9Se#a`f?#z<4diLqFfwdOUP2Pl|rthMS~#XrIh^o;Y$m^aXhT
zpTt9hU`ng;&`WT$Z#?ui?jk0<{wGTf@LCrSeFM2a91nHEgDT!wqZ}#&#6t&%=<Pj~
zE7=<o$X+!biXlte9}i8%y#ihvKS#A>V=L=J!~N_XE!#Vur3U_0<Dq$wXkR?^pX~o~
zJT%cK9-8=v*N1+)7w5b_9%6UQ6;O^}JhYj(ZI6d0%pxVm@sO*WRN?*CtfKQA58VYr
z+7}Of&YEC;XNf=Pwi*v*lo>Z)g5XD)FYCltP=h`mB72p1sKiN*(#1o=A)t$g?#EzM
z;-QzxYJEI(fwznt58Z1dwij0W#zSckYfn6sL6#eIKgB}}S=DvoYpBgX9(uP_iHBap
zQGj?Tj=9%-@n8MB#vmU02#So1kviiH>B4Bdl<8FCp&n>~iun_v!8aZn$#kmmP%M!O
z@L$XFw7}n!I2{)cJz8=e<Dmht74gv3R7Ze#=p^z~-wbDjyek$BLK6@5BT{iocb2Dh
zOE7UTBJO~%|9kP!w}Xx2q0s(HJhY0s%Qqf+gRzX_p=#vl<DvT)ucdhCtK!z<p&#&U
z5v>os2scUb&=)dCAdValZ36HAlX&O~n9^!IR1G)##zQaOCB+x7veW>tb@9+gko&{&
z&~LaE(;I8#L1ln=X!o^xdz&Up_8w!YfxT)x)Q>D}e>^nK$HtAQt_|_f06%+glI=|j
zY_Ggub`K=l7Z1I0$A38<lJ?5}=&I0>LY(RPD2UwvluJ2&QBVzW14KbN{<~zi&Lkzq
zQPBJ%QdK9u#42hb3L1NEi~X8ivP*fqVTSFJWxx;?1z|tfCwP#4SKM%a>~m(k`JHF|
zLALeyr@*-3a{R-*i1-IJ_{Bd859>iCGss2y_y+<$O-MK2kHM$LKV-FE{NpX7#6L!2
zdttRM{()G#;vcfypyOHm!>UI7gW3$@pS#ug2S<V8ALd^3#n1S6jZypqMMlQR@ek>u
z_=oBA#y@C*it!VnK^Olpo!<C|NCoJxWqDel?@62vjDPasqbE+2-5tL_HUPF#{6lpF
zihq!&`eryIl=ugPHvS<}aZ7iWr*%s(aVY+QuiFs+;HJNtT)I=J2AhQJU^#6auD};Q
z=W@%?u4{}#pr9+25a>VDV*+0G#;TNW8hV8>5d_I>?QKI3BM)H^%b^<<q#K8d8F!<?
zHN17`<LRx3LOXAlLZL<Qm=p?al{s^WBZoq(!29opLiBv0H;T9d^|v1Xz{k4y=QcV1
zVW~kpp^twc*JAu5>ThttEj8itdvnDz<FtWK5p<EsdEW7-yZ>hQ6?%J{(&hMvr3UtT
z;~%orIR0tT#)&>QeuR3+#%Z|4BU!$W%X|OTa6kJ-%l3^AY@ZzO%!7n~yn`1Z`rdA(
zb>G@sl|Uz$TNrefPIufGgd2>c8-pMk3?e=Z2l(Aty!XutGKeS(4iXqY0?#+%hv0F`
zPcGf_!```tyME|BDfI5GE;dLeTIso#PIMGWM<c={r1ak5Gb9(4>th#ohtFF%WS0I8
zpHPwy5W7tDk6r#Ujg%P2E?bzYI`Ml}Q46ul_;$xGJ=NG{F$|GomkOHb*ge4q@I5IM
zPmR*cEC<$dRXMn~LGqCI4xfvedFH0a{9!T~;_L>OZghjoRr>o|p363FyabtBNiQIC
zFD#Sl^zLW9tnd)Oyg!xXbn(lB5b$k6x_K7{72Q_!J@Wj57l&GaMAjOKL@6SNNaRIO
z0fLr0Sev_2oBJvEOHq-fWT`=CQ@rvmtFcZz4xcFT3dVjJ#y;-nIhzHXd>_Z+aU3=M
z)$o2Efw|K>wc5WU4C0mE0gREHpDLS^!*r_gN-heinEWP8@r_rW1xEwS3y4&J`2m)v
z1?GI>bX>gh^%VH%rSlZ8JP2D6ugs!40>mq$Afo!F8iMNnRRsu*TkauJamzH8r*+FX
z;$TF)1z&sdBH;Zza+J^k@k$itc%xgKZ|$YTE1hY~=;M{6)Q$8r=&~i+c;!pv`Nu0Y
zjN5X&(kHX^cqMX-6t6VGV^X}*OXh4NjvTLC2;ToPUMa!DMXkmwyWwNsc%{i9<3CFc
z@Lw0N^hP!9k5^JhY2%f3sI%pGWq_Z(H_7%U1+rI-SMGsCd*YQh?OGdq%QhOvE7d*q
z{(Nz?9RIP@z@O@V)pkgnW4sa$)1(`j|8Ts5TXJ;q%A{I~SMUgKTjG_f$v7MJ@d~>G
z=j0^h_{A&fB;WRUW#65o#5i8*c^9d|`)^34{)S5J9XK5kuN;6O?Tc5YGC!Dade|Q(
zTZvc1JB%AIL1r)J$vSZm^y}l5L$H?d5U&K27Ja<(2?X@<N)!eag$*%euzs;+v08vc
z!i_}kQ$$$2(gf-p<CW%##_gqer9Z0?@BfBc|9E8#N=orcGL8boE4!FG&F@$FcZ5N_
zG9!R7lJj4J88r6(#dNCi%D+)a#pK^J^zq8i;Ant(Cy@#;f5h^%z`T+;9T%@$19j&u
zUik#d5wC2X5IA0W0V1kzc0*8&R~kTQ+)_)V;+79sp4Kg|5C<dTBz*l}jaTl&9B&k_
z{H>c3uS}pZqmNe-8Phmkxf*%?@rs3UTaH%>Zf`waDaSq2wAB<1&r9*j-7@EL;>hvJ
z9pL>h<CXQe`foK}8E~;4UvEM&y5A;=r3Uz_i&thquKn@KO9|RofRjxQyuR`v#)NNt
z<w}z5eTt<9_NwtpEhO3#ulydbwQ&Zt1YTd+ZPDA;lqko4EH$)GUSH`a*{6H{s>St{
zyI@)e#VfdFMi;M4JeT#A&(m>=>f;r5hs-F-@rzepBW~N{mFvfo661I!|29%pC(dIP
zo#*wHVKAh9@yh$G31-)W{xI1}yb?RkxbYHX-pxE&Cq4xI`grALg@<@$+^ys$-TKO9
z5YWdf_hC>`Ffxw}*2gOcV2f0MMD8*Y*`kQBc;zODoMXJglcffoP4UWXR%4y`G%E3r
zS6)t2;+3ay6d+!SV(v5t&-U*KgLq{%6d4&KIsYK(!q_u2ooc+&2`x}D*$xf9>nk@h
zo$7v2B#{a*_hxxoV7`Dj9T%_MKlVJvE0@7m#4A0hjsWW``;n*mW&k7P{U9p{O}x^L
zNX0FoEKlo}qhpYb5ittB{;$R>TQJ8P#VfyrDe=l1)MNU1<w?dgj#p+O&p%$7!niHR
zD{E3)k5}q(Pcy|U74V@HudI?eg~XBLmAAnAU&bpJqyARol?N}-<E!f?8UI;ofUmlE
zWi{m5AFmvY)y4vu(ADz#$|o2TzVXU8LnV8EV5xz<YP`~oEN#brkln||)u<<MyfVPg
zzMEwGk^<YO-2VcJb6j6}_r?y0SLgw7%?&MWT28s%fiqGcp0Hb2-X4t{zwpFN@&Uq=
zJ3{2$5br~^&x|4^#^K3UrV9HXtfCgelk|3P|MXI~e_nzia(GgO-5=lmA9edD*I@hS
z5@w$HfYTopQ<UwW0M9o+XE$!W1d&@wF(7g;Ow-3FFDpFS|KUka7oR)`0eyV33xkM)
zkMEJ^7oRxP0wl85NF+)Tk@kN;onw5mlq@yqZMOfzYQ+8zeBvLUoE-^V{0H0r!BK$t
zL}2bTPc8ND34{2gcK~A~=cmf%<S?CTe3FYoDh9s^Q+(r-XTi|`@d6?hAbx=5X@NMO
zI2{+Cd_4j_T6`Kh>HO23FMsrT?gwEj?f+060pgQU5K(<o4MBB1r2>S;E%y+qxMdp4
z)4F9GaS)%p1z&sPlkvQ8imwZ)>70Nq@YS(6v!owMv$684tt#UZi9v6okR*a7Mk7vY
ziSbJcnX|fHWi(0*cApD-(*9G4@ti8LR$^4BjCUC0fjmj$UnMF{qMnebXo;$nC<{@Y
z?sPjr-nk)nQNr!x`FRH-yJKD#&eD6q!uaa$wzYSlePS9}EWJw7i61T7&PER2N$mTh
zQMF_YNr}af$4h(qKp9PFlqxT5DT_e`uDehaFX_0%n=eiZ27zeI0%7{l`)f=%=DH<K
zq2nVE#&H@QH%iCxbWBR<<u=P%`n0WvrHgG1Oy*AflID0T*KM?g>lVV-B^8tEgt|`X
z4B6HyYTLYx)bKkK?%EJ|>b5}w@8+OHeK!aFrriY|a?S^~=TM>eQ!>Wm3Cs(a&m>>a
zcxWo2dGmg%Q`pdyLwUG;!pD71-}@c=t%O$35w12&kg8N_=15aCG`Yg6Z}kWkyySr0
z?`VZtVjgOz+0sO9LfVYJRrb5n#8sfgze9xGPomsF6@+(+jxDxV{(e!({q}jk&xdE6
zxx8yao_${J187J}OpPR8U2=-A9@Ux03nMuDHixVGr@ZUlDSXI~uk6)bg6(lfQ+s}c
zZ#$Q?)n55S5Fho6L&%)M3the<ZM<haUwQH(J2$d{7oglzTlNlLH<I;7O3W6}4q<Gu
z!!>dWllHBQcXj#H;p+0$*y`?Kd{Dh3Zad%eH~Qw1YWd2af*euxj;I=5fH^`asvC>y
zYVlGeG>Yd&Y^Mr0xH|Fq^>0%b)Vi(0`lzkj_XlmQ-G6!h*0(*Lxx~P0W7j_W*?=IP
zOF7EtZyU21BX#9ISojv}xO)4E?|Gq~k80qzi+oT6Un^eD=hskaKB^XONmxZi@Pnt@
z?fFOPkQa`yGDnQdY<A>tW@7Vx@_5LupRA4xzT|})9YPJt;oGR3J*vi@UoWX|#iaaS
zB=z4*>c5BjT3#57pLT57F1hh&Y>Q2H<k#x-^B4RA{g(4=f5dsV-+G?yk37%zN1tc=
zW6!hw@#ooo{yf{C_J{13W)w^)^K_?e2KRIg_wO2Bz!X_GG7oc-X@fT$S>(MRCCx4z
z2Zkf`;i@V_CAhV_Mp*c$oqTO`2*2v^A?d?&@2z*lHFA~9Xo6dDh4cxwyKdlxU6{Z5
z+TTMUvy~Ut^HEzJaeKLk5~0r#_q|>CB$Z1%U=R9)d$QIMCvs1&58H~FPmI|@ZxZZd
zkK4pOVx_ugoGBNxZXr0Tg<gjIarl+jPq?ZDxcYKbk6vJp`o+FNSRj2IH0Lz(wfj2}
zU<eI-Wn&0Ge}^*xlXKZH%t-Em04RsBZxID-mrC<BT^~T10`Gj~hH~EXN!f1ujE&{?
zpb=s22Pqe?$1B6sSI^#01S+`94611~KYxq!X5KZTm=_X?Im?JCT(<;793#?@BH9=c
zj}(zcC(bg)!dY&&a+bRgR7}MGtq3ga2rPp5%FX3`bwWA4cpR$cKvlV{DqdC<FRP08
zSCs)(X;8&O70iUH%9;*URe3U$d#Dq|S5e1ec5r<wXzU=4;Df%xJC$4oJz#Hv1qLVg
zK;x4wFgdwL8Xdvt<Q{N9vIS;`_P`HE_pso#em$(5Wl|5AJ;{QvbPJ4kSYUmqg|DnD
zcixYeANCMg9m_0~B2MeElhSTg7M6mhXqRY~f?grhEJdy52<mz)V1=LV@f=GR_gKn=
z7F(qBX$#|bTWDc&uoAATE9a^zAVxM!!C2viWfn@whBQi1Crd%c%QQ<tN6R!zK_8*&
zV1|j6f4YZ-rHgx5ncQNFls;`?e3-%Hpr6X|m7WEPC2M-@L<wG4W1*BRsb?C{dYN7!
z(=0{)QkiBc=sYI3utzydpYE}MrHgw!$K)1Ur1WVE<HHgr2TQ0NKW}{_C2^l|^E)A;
z)0|^7)??fjgq0&ov++Vxs_?V=J~`j>3r}?UoL_d&r_%Yw2jMkpH%nI*QE@oWU0rN;
zA>1w$hS_IqDo3<t7cliyZVc1BB5d7Mx~jM%7V+q>L*;l>8)~%_o(R84(Uuab{nEy0
z<^KY-+V8FZ7ohpX>;VGg{|nHhYnng-{{IDNp1J27pe0O5xIN*vgmhYEslaS%7e2QO
zClIIYGXHXgw5{{z^=w-wewX<Nne!}h3T*N8hTzY@W7|3oTRC12?7*97S3(bc@NQsY
zDlZ16ET6Xme~9TH7p5ce^ohk>3OxwYxJZ$k@S}4Uv8I3}(GH<Wjt}tcMgd;A8#T>_
z?*3W)<S^#xND%7#e!uA_?|-nA?SJ4My?9@5E#B7Kh~f_AeYD-6Ufy3Quu1Qug#ol0
zd<%Wu8^%?=3j<LE>zep>($gbq?c9VNWaAoC!$s~?@d*;xg2Q*Z4HrB=i~V6B+18ND
zO}+FM$+k;JO16;#_PEN#2H1uID3)x?z}s>&c6Kd=46I9o48H9M9A6`=I@8zKFxF;d
z!58>umRVqBQ#UTMj*HwT8FyJ<=)mJM`|x(<6Y#)s)PnS`&!o34PuvQ+9{Rf7Kwo3R
zu-;UG2CIm;0newlFg~_hrSTyQR>sE$nR733<ni%O@H%3AjE5d&eC&|MM?6?;eC+Ta
zA9%+zx&NJmG(JG6<70JSX?$#8scw9rAH3t^7f2Y7j~l5I=3;#O!(==@P$4(r8}y8M
zF_dtT=>5L(ehG~X8XuFS@zDhal5H44Q!nc)*>>5VCEG}WG(L8sr%*sUKE_Md(fCL=
z8XtREvyp2sL^gAgpGx*!b^3BPM83d~IEx{25*3|gqvwvVFjBsiN6JSuQsnns=W=t<
zoH?~T18&4oo&3g){`zk`7SMRQ2CLz!z6ruP?|OVWJm+vdThEdkVtz{%Heh+KIhQ+(
zrz#J7xT;GaXRp4;Vz0gn|My!`g*toX&t3u9pSj6gQC%)q)PNtbn|=geeewSQ9bAk5
zF?4W4E;qDCGk*JSXpb#H&!6~Rz;EHrfucE|Jb&U$E?2l7Iak!O{AN4|2B~035pz8w
zoAHZzV1_76Mv6@S7ge7KeZNZjPEyg|S-!{={G#YP$w-kY_(joo5@c*VcHv{=PvvrF
z*W)Jcv(32Pvf_df|8-V-WnD8j{}L>`I_MXd&Qf#hxr_0r)mi*5Uu^Iel$i68J?=<Z
zShBFo&DL|P=LFf~nz0vvmCc>Ju#=C%YTq$u*O)E1bPb|Q%o9!B6I`MHem>#Euzw_Q
z?{$AaA?Py{=POSImkh;$Jl{fQEy=7DzJ%0}RYBVlT*+%g!J54?RM-xlV^wGn-tt}-
z?1-x?X%OnMT%C|P8FNPk`qmUvqgf5We9{^%)ceY7)(?2$80`Vf|Awn_Lo?P6!tAaw
zSU<o}{5_bD+r?F`0iRz7PtoU=R9pcF>m5-i`P!4gJc>qr$;W-iJ+z9X;=Z;EUpTnL
zCVS8qSSN7A9mbkL*mkTI98QV(k(LXr_P8_LBhNy3RQ0fTkeI}Q*U@I)MkuD*9Z~S`
zCcgGeurnN!jRQ3Z>&2l1AyYkU8TKh&qC)sZjgkFv1roH|UfpLIo=>HB8hga2z)Kab
zwhMQKjjj$|MwcT^m-4kg1jqbF*TBNpVmC<GgRm!Ibd~bKwYW+di>kcmn(p^+(u6OV
z!*JSScJE}b8(|@Hus(4Qx8TpgIQ_Dll!oyFZr`|-3>WHBg`3A#+ulnRaxF2x!BXDU
z-6qaQb>M^nhXhgkd>)oB<_z_GlvEe`9#ooyO|=L625s7Y27>dy5yZKB+zW9cYYl)Q
zVPuHA6dA&v6MHH>XFsI=T18*n*sp@V#v+S-Wiu#9;QDmCqu-X8-|YRib5+z0se)*a
zsz)hq?BSr@c%wC6xeq7Ire?nOU<kbD+2p){7u<a1_d(7GF4Fuw?*B~!vc7kmegC+Y
zs^=@u1eZ)2Ro(xc<EV$P?(#1%u@do#Bj^lNvwaS3?59B=@g8ix{LqZE2dc(E6}^62
z<MeJRxOPuDg8xSRV`XSl4Nlx6Wk00~n;gP6KJF-2^$1ia3+@x!D}O&Lw{m-8wd0y#
zVQb~y5G?eBVlC;~%9BCP-jyeVofnZesp)W{@nPlfp(WRkuI}<)LiKGyJp1JdlE?W~
z7eohj38y93-E-Vtx#=`tS$m{%{ptBT<~9m=|94Q?#SY<!epvArB$(EY!Plzmpkd{m
zI1<CJ!6<X1?^XO-QJ<QnUEo{M=u6XnUFCrYUigg8RpA6LU|DUecqw+F__&?UE3t?a
zMhCrkS2SOpZo{hl3GoqB)6|Q$kanXkVu!8Gx-h{+c{7kFSsTGN!XUH)LJFQ@zw?;*
zAKYaOrQBa?ux04mKi>HR0WK{=U=@-Vn(ee65QiVCbDHe}625wp&9W*24aWu*FKiZ_
zk*F5ChFx^}kd5<P{5!1TU3jF}F8-DrW;Jv3W1t($3S3nrRf3--2(#aW<~c8?b;3^A
zSL-4K^OfsE?E<pg4Xn44F_>NbvBc5fh!U{~ib-nIO4e~)vKvbn+><plkZavHbAPD~
ztL0bF8G_F(*b5TzW0k9+#pLX-;nM;qtv>O}j5V7K6*f>c%&1i1tRrf_UHHmg>p^uG
zuZ^eqpq;SL0Sm)udb)_U@sM~=l*cosdiLs2nhY^#9TxZenP!MSwp4-b!-l?X$8U-N
znNKkHp_rJ8GU%aakPz5!V_S}U#98pByB37F7&-2{&<1xSndvh39%qHmp@w&L9ZSdb
z_J6^FH9MDiq?~8v>ktPWNytJj{nSOZ{PP9atuISwd6pF9p}VGtZ=(Vf{rpz1+6FpS
z)j}2ZGZQZt*T6EwFEldLxOZ|3+pSd)#HJ@DKHIH$jT+l-tw4r&FZ|@jld<G^*b=L;
zrNGA)mwB3;*FqDWw*=n(H8y8%SK9eyrX08$R*?6+9ula@F7x|oUJt!X#{oR_0vbbI
zn}}k$@_IhIT%_elUib)ON%$2fyY!BUE|9`Rc@_43Y1!Cy6X1vTxP=hDI^MzyNm%Pn
zVw?PCanV({W5<H75WhqnD7B{tD`j`xB7s(CyU>}6*@cl|_?L@KM6o*ab+{ma0a*US
zW;@d7*a%p|ex`<gVq+0_YoDY^EQ(x^y(zWYwwW^M^uWYIKYlhhoLa!{0+CJ{@3`XT
z52KBMxP4sJCn!Mo1yryUx6esxcfbtNM{HT&b<l^M3+J^o%hACVY?fQYeqn}W8z3t_
zkAh8CU}(7C01}br1z=NlJyJIDiDgW=@`E8CFbUB_J%Kw`Byp{+JnbYPr-CWc=1ig-
zx~YM3gzv-yC&}vBbQ0k~#<kCpgTh#4>9%6I8F^hvmfZ~CbFOjo>9;Z+aX)ZXD_BLd
zZ5gB{O;S^9Bh~4)3|UP-s39jrD&FYJyn&OH+8Tz=6CXp7roQBg9?TVT9(AnbhnY;L
z;s+<r?^9$fW&tSCw(}KOMr;_QWv?W*+<cn15B@GSpC?1(*Z!57e}QZ_FWa5pMQuLB
zwA$z{1lIiD(0m%$QTVwXI=iAlqaNI3LVCj_z5lpU@&`d2jka#kYv_*%@<%f?h;vXx
z^2hz1C50^S2=XL<OlCU0{)iBp$a*v!f?87#!Bk?yAT8Uh@dsaxbp!Dm#96$H?iGi0
zB#)4`vN@ESKgEnF!n}2?e3G%-eH#0bsC2sds(pc)KW`(=2sDeb3jpn;#Z|3BmiFe6
z{-80BQsVwQzI>(MX+fIy6nv$Zv>@&BlYXiN={bJV54RwFkDv5?El5xElP+jMdV-&H
zW((4z{G`XVAU)hqI;90^>Qwcv9`12Bn?OFWQJ5n>W(ClF1&&7G+fy{`Wr3L_uoWDQ
zz$&`;P7!dBz*2BD0`H&E2wY17OTf_xbkkOGAqjYVZMaY?;O+~7W?zBrT7h~JSm$eG
zomSut68M*|z+bhkc$fsLeOvJttw0V5RD$Eajhv6k6^MESmdl6xs1ZZuESzF~a8$LG
zZ5G!wv&B_x4s%s>iU&HSb|KHbpUz67dAv&6dDbdh--0qa8{Kp*D=YJz6|3myn0-&k
zXIfA;4|z?9+Bdl0#1YeIfM`{8)#$5qY70tJB~|Xf)AR<_^@?gePE@_?r<X#gKyEBT
zUQ-%r-0#)sJJVJ*_-TB-1&vQgdfW$)uKPRSN4n%HfOl7}A~brj4;vJ2m*j8v89B{I
z0nI6}n)Ki#+KFtj^JN%4<W7yi3$lO(srVC=x+%uh2+&+b8fr+xCK8yg6_8@cM#@}D
zCK*K+Ro9b_O7A#w`uuDdP44m>?-MI~pp<GBN%?51`5=x|)1;asP*@zuYSyEA)ml<;
z&ecj`(+4&yY5u%7koG=ZofOrGtB~V|oGr!R#?6#kv79vKF^%fuSkL#8+_{8Wp+V>N
z*0Gf82&FnOr!0oYwCfYv^?Qmw?ZR(%!DDw_o|^q<3Rp0hah1CEq6;zfOwci=y8jQv
z!WBF&bPtIXKVrqpAi{VP+EU{V;2s#Li57n#1L#6yxha7jU!s=dP8+lk`yuNt!V&I#
zL;(6WTE7m#g-k96>9q#e;kdcCHyKlk4E&NIGX~cTd&C`dZX-s4j5new8;NcH1lNtL
zXjx(<two4;9D^bAHd15ND>~l`x*^`v-C>Kwg>PgN#i;7k>NFBihYI(2Cl6}C_^n43
zqN^vy?{Qq(Y6SAM0(X-Dp8EDQ-Na@)eW+H&f+ubDWBu(*LeVCtT;9h6&bxd}AM9WD
z&#Gl}2dO9g=MmC)DS(^3`4>wu;eX`JvB)IXk3=f3=h8mm611BIyW%0VjXFI{?({{-
zQ93<L)9L?)LQSW?LHg%yBt7zijrx0b`t_UEXtvmd0T_m5)`~yjVGf?V#$mHi0<*BA
zVSXWub~rL{NAAXm4U29YkLmoFYw)^jFvwzR#MwPyimTMJZkR+DfOeIJtxJ>0$spI&
z;L>s}U6Y89ffP0|2F{IQ)Q2vae{fDub$tw1yy8;QS-Bw$dv-@+JmN}_n*3YXT`BIT
zz@_`!mHk&*lA&|+bgJ-cYScb^?HM!HUFMY8s}aO|*BjXoC_O?e$O)*P7Fp>0M-!Tf
zH$dhwv~~?9Rxg=FeFHJJ_}oe6Jp`V;a-)^6PO@ULaqiz>Hm<h6#e$n73irF=wmdBG
zI)~7GX;`wWYz`^Qu)BuXNbh<y05_J6MtQ8mOT2d_-i=z`3li_jRN;8j<+L;k{Y{sW
zL=%T^OVb7Tj%W%&zC&pAp0aQm*2kEC(#NN<n+H9q)!3C~>)BWks2dpp^RY>k8Hx4f
zk&$4;5(Af8;y(12ohA$WwRp`XR@Sh55N~&ljkLQaMB}bOX-8l)b_8aH(+K~|&om8<
zjBmQI6lYhLd3IrJbV6NcXu)myBdw&C?xKsbyTW63Xy@0;VNV2=UCz4(*M&aAq;P*=
z-1-txSzX1Y(1vwWF__jm@wEyI1g4cQ;z;~wcedJ&@ghI}K(unp(ku|<<8APr?|Noy
z=fk<?r>)zo8fPWrek9|1pnl2!+WaWWA94}$f5UE+fBpc<UrPC3s`=+?PaU}&KBsT2
zGcTh3BCWrDc84%Nyz9GkUfYE;j<_bS>MRO7F#KH)gr(w)io(d+NBcB(*Mk@qDcCc@
zW&xI4<2KD27xN8PQk}p%tiXylbMrr^ve-IKyp)~Ysc|QqG(;Q%HbeG`3&ABVQm1XZ
zm^yJXW<y#FO|hg_=Y%zt+pEiPMx+3>aD(be>|aRxQ|{7!wD=s??2Lgj@nY&@$wt|%
zu%_N<16eajuI&)~NT<tHynwAH4A--GuHdvI=r?=R2`5(5>KwvJaSMc+LLkQ1?hCeu
z%miNd#~G>)!O7C~xY!RB(tIO$5^l#-lW-e0z!K6O)mNEft)|=f>H!f*$2+3-!6UV2
zaONKf6EE!sC%~Y~mGyRquphQ1Lx-pPw>X=F__(IB{cN;P#+5&-3q7XJ=f^hS%z-A7
zz#(Q)z_{i^-L3rxS)uy}$#`-aon$ga<J~9P5Ao;|odTSltNoYw-f!g5XD<6jD&-ux
zMNZeDstGub$A9MV)<357oKk0g;dJY)q@tO{1v%v-tbVLy4QpIZR)JQ8Xyji*AJd(t
z$ySqf?C5@`glTu1rsd>jl@&NmvvRVXMI}RrW~OG!tjw(9;+(>1ru6)oIi;r5{KB$Q
zQ(6JOl8OrNHaUsQn0D#dF15|~yErU?F`|3Mx0qT*DW3`9Ys`?5CHd20=_@g-G)LDD
zO8%gjq^uHWPHBEtA^bA4tS~=23ypwuLt++FWs&Qb^u<Jz!t}h7oUCb*N<}VSmQ(6q
zo-CC=(pj>f_%x!TWw~IWJ$*71S|a_vEx6XAk9B`MOQYp9UQW~G)VTdIgUMdCjT$ec
zDfY?rD}H8`;44h`PyC9ifeD3$!-vP1NNZ+(VZL)<cD5<IsIb&&Ep<*CI#e11!}1HA
zH(N<&=urAjmPW|<{F&1tk=5UduPD=i0W*v9@jV?9NH8o-_;)hjmJTe<zdt83it3tG
zls`?QEG?^~G{=#3Us0JeGCMD;#Cpv&Oksb23$8ZeQN`18rj|_~P?|rzFsq={OFO4Q
zsVOO|Y<eEr!*Z=xO}o%!wa(0$Y0a88tt6+k)OuAxK{@f{Pj=C?943`nTvC(`fufSq
zJEu)1%d<5@>N@5!Ehn?Es4$28Cpk*(B<(;PI@FLI>F4aLqfF6SPZ<^)I&?fv$Mm9c
zIoY#DloZX3WUf`*=R>d_(s;g2`Vw=d=NJ0eMr|n_mvc`UW}i%q3G|3hH9kbU>O>8g
z&YU?dR~uyt>K$^E6DMFQ#<BGxkEaI5i^4sg_zO*@<wzG{(L0*+c%~roNkh5_`SBKy
zCoBYfTs_1iP3!CNlp}4xM$K}faU+6=v<9~sT0<c>+~dhWdL$X;kxs!A@AXI<#^VVx
zGb+Br<B3HYJ;mcGrZm^%SxRYv$J2<kxWwa$=mfjUJ)ShA%jbAJ3y{`40(+4*Eb@4o
zkv2Z>@kC?Yz3~-~ClBeOw>_RENaH{7cy=N^vc}`F-~?W@3GG2TWi#wXnpO`vr14uk
zo)t)~+dQ5Iq>Ve!&I?e_S7-;)#$Bi%sr6frr-IV&U?<XOq%}z6k?usAhV%&1DM-WM
z$3=V4E~H2B=*?23roE6yYC*aaX$4XfuCSLNjX-*2AL=3c7qpY;V~|Ihb{zgdIt6Jn
z(qg0*^iKuSNTkb=##4X&3V%ayJklwkr~HO`krtnTJT`SIkVYb1gft%M5~LYOmm@7k
zT7z@}(nh39keZs|7o=9C4M?Mrib&&;nrK`gjX*jDX*AMeq&(6Jq&vy3-{EJ_){|%l
zQu<(jk$r;hA7=_Gw*>X>(m8BVP-pg-ZXe{V=;HBQBg;@?EN?Q|kVo<vVXz%z&J=DL
z5gsv$yXfAq3ezpUZn|c$?Fw+245cCWr#V=SokJ|W;!Pn^y(ZHT<dLrEt{%_#==;!b
zdnK3>!Yzs65sBf}#PG<(@aTl_*pcCj%$1!NcB(c%7W#O|!@;cFB$N{oS9Zr349)0m
zR4zF@&8HYE_cY2iVA;iqu^C!rSZ-u^V~|flvh!p7M_?m~l}qkpRG(qF&>-{??kAW=
z^@U#6=5oVO?ztGb+{U(-bD~^BtjF^KYv1K<u5TI2WyH(%CHHM}xgSw((MY9z);5>x
zgE_5!jK?#O`fp=^ay9<_kdAUS=^jsiDz~(saeW5m7NXn=%z39Vw}mdeqRr(tq8z<r
znw2ZQvd!gAqa2SpbvW6XZfkS7AvBj`4(>+fVy|j*xdN1n#hlH`^|`vu<z7QMYq`hM
z33F$t>CeXH8iM`Khi_4?xk4^i)An-R34;W=Tt#G?>vN#o(kDHh-)NqQk7{$dDwHdJ
z(c^iY%9*ZdbGbDrC%!6|t7&_=<0v;}nOv^o+BVnMALDZ82OiHetbP64R8E<nLwx3Q
zY2K#!Ir2k~XB$$QQ!6nZZ8&@y^)!ON6X$KHX+VN0G2D_69x*b!CbBZ*F$m#9dTD-L
zx(=914EaNRC;lsHJf1JmPUSNM2Mh7f1LIc{-+vBI2#@y}LaZq1+lc=L!~kDv^z{et
zTl_zVnBX!p2{x*jkPsd{GJJVZWv7Ma>d?nR9!EaOcR_m=G<ZDMw?W<pc{aZy*7zOe
zy!Ix9M^-8*Bl%$(@{2$Bcre`657|8%{5%8x)8J1r;IsKv$0xsSgnSzK{g8U?P_Zwd
z-%dlG4@HcGVpNV2va37V->~1~ks+Vh4EOqjAMu69GlBeb4*fF?^6_8cd}W%XFV=qW
zV|9G;>v7;tH28Tq4xOZqujKG(b$Bs5*Pwhk<hqecd3=&ThG6_xfIkKf4}A^g{cXiq
z4e=RFRIUK!p8MM4c?~gIXtS;y8#9f;G-l)wf@)=cc@5<(KViiP_JmHkj-3y_W37Uf
zQ^p#VqjPIE?28mVo^iy0jciOp2Ytvti;<s)eCk%^L;MfGpJKrO82k(a{*U0Nflv3B
znv%n<lu-SrW&QUO1NGw^P6)SFQh3D9hdwx;cA{J_GC+6!XwMxgPIl4!8WHUAln?{@
zD53LeBXD84+v5p_p3tRQ|KMDKUn|WE<+D)o=hM)$qzU#?z5aa{a6EsEav2!6dVl>0
zewqRQH2Ay$zdPo)cmqDoudxRFWbmWG??nxk+fDwMB+KjhPFcU6zYu&QdtQ^}_3c_C
z>(}!?2cPWGp1)+z0q`TiN4KjV;)mkIv>LRl5BT*a-}xaK@}J}Uh(`b9kQR)K$AeSC
zBOVJ$3Aa{<ri4c>H1pxn4|n3jV=FtShQ}9#$D)tX(<~lGRE~K#jE$E*UC>|9BM*7f
z#y*~q6BVYAGCKZ9;z$R70`W8E2GidE*WT4Y$x&75ni>8S0!fqv17h>1pfXJN{A31K
z>B*#n14A-Q4-xUFQqw<4lb-3ZyJs>ZyKHtr&WZ<#pa{w?*<D1@)y3nUu<Kdfcu+)9
zj3|h#Kd8GbKQ2UFF&u^L_uaa;r>eR-sAqBYIPY}Tt9##l_q})DefRxzy;r!_;Q9_4
zo^vvapMfu~1mF}rJxQ{61K>317X)?z_*WEY8{<@Q+XP?Sy#OE?pWFV9(>h*3v~+s*
zun8h-qxtHD$~+8C;`Vp^%U^6qPItAv9xx_u&Ra-xkpeD(JTAN??RW&P#KjDi2z*R|
zSpr|hMFy7<;5$fM484%RTOplS!S7%pt^x2C_#a>hr|gp;%@P<!32uZhWIKRiiG=KF
z1tjEI0J8`Y_dChI8_fT;&vX9IqbN-N8wi;E&nIB=e<J~tf02O6zed31|9S#6>XiS_
z2>hAm|4IToRsKT+22}n+wjCJCU&tO+Kti4XZ~*!Lq38dvDGHPSO$1E-e@npRe=7l#
z|1JV1|2qhn{I4f)gUWx1`u|mx|AP$KtMb2=z&lm`LbeAO%3sI^5Rrg{ybi$cR{oiw
z{>=a3>Cf|;|A*Q-zT1{T<<RyW&sw70Bak5A;iA1fk-)qHuOx7d0=)$O0T<(B2z0<A
zu80330`~wo75?1>I4+>^bP0SAHNn2`a02%!AY?0`R$sA@Lk@z?2OuGR063nJI2Rxy
z;|jNr<LM<vb|-)>0L1O@2$k_TMzJzVfvt>EU@POPB(pMh6R<K45wJ2YBk&b1<Ch3L
zqh)*x0X7|$@dX6Fp@5LRSj#9O2Wc54q({prAz4IZT;cXHJh6;I_IUuU%J^fjXq52@
z0#?So1gwnv2v`|An2eQCs?N&z1CrgKWxRo8U)D07PN1m8`ZmdSDj;M(0Kj^bkgub_
ztVaoH11;-OLQVn~G7@f=;E81vGEQ^<n9C><F<m`_GO~J160mw4AYk>#p>U%f#|c<H
zN*jJq>+wN`9F7WNdz>QhVXeAV1Xy)cTp>GAbyY%^qv}|X62jA`SdS8tM?}UIZX6l2
z9);{a0Ilk=jZv(OqJ&n)gBfCFTtdLgcnks4RcWbK#;2%+pVTsbf@D7gz%s5OaH$qc
z+QdIAAY@Mipsq^DqpGVC@)9khgz(Q6G7@e(@We6-*{uNn$jWGAZyRoO90p-jaIwcY
zn82SZa43PRaj}y)iomV7NVbx|XB0S@0EgReLZv;+;YLxl?Iwoo1i-3%4S~Ocytz#N
zG|86Wb}4)zOCp$cBxLVXKtlLJTp1za7RKI-`G-JQ8UN4n{~4>&^ylXUOn+qlX!^6C
zA*Mf~S*AZPrec`>JVCNItNvWgkn2=`UPWL|_2*HNJ%iyY^+(7~Krr=3$ljrVgj@+A
zMt`W2RP4Vet12$a@<Rmpk_~|m6L>E!lKmrr+i)@Db^>=Q@Oc7U+dCJE{Wk=T(fpsq
zERSja|D7RcB8w9syMe&RHUC2Pbzqo(A?wxrOURjv`G;J7Z{>e2tJ3u6eFRK@-cP{v
z=OzNCKcZQtKm5C}p+D~?aH#6f8PuOa)t|c=avD@52mN^`0nQq!|3daTV5mPrc8cnc
zguF7WKlg9n?sTlD7QYBr5zL$fuA^{mC9n?HDY!Ne_$e-G|3?6{agP>(w-R_WZfxk{
z@C>)V;|{PM<m0P(qvHe9dJTaATx1|_2>0c<HW9i27qcU7f5#r=kdLE2!hg6O0EYt3
zp-3)oA<Xm3EaYkc6avS+4*w4LG#}{pcl_`#k-6`;uL8h4WGF%MCon4F!~;X$MBuZy
zR^Zx1;1{@7;>xf-js#0)^=twu07t_gW;*BNMg|3}+lnk1vVmm(gpj51)8xj#9XLw+
z7y<qlL~gec;K+_4?D3qJ0brcJBaj9_Vev0mhi7~<JCjVOrXe3C*{1+B(zycwx$PtC
zZvkLB&k#5kBPE9XlyS}gz&QUyU=#q;$-{T}zb4a}VLF#-$gfCthXQLEavy-Nku9C-
zz;SqF3hzSkWUZfRUq#@fxCq=u;B&aB9bYAI53Wo2K)1i+MyF#{H=g*8^mmEcitGJ+
zSagdLSd_q`1QsQ*D1k)@EJ|Qe0*ew@l)$0{7A5e<BY~yIrKPYt@KQV7ZR&R&>%)81
zmq6TXUK#b}KafCp50?)D@c9ZK;iH{>E7J1BSu5SihYX6(e-!BMdaWn^v2yp9R`|$u
z0lG&toD*a^&Zg;p{_K9Pp*q*Mr|HQm2K}xVr0F?W>Jj%yds=*3Mw!uo!v%|_NA<b%
z5OkQM`!4P056c1KmTLJpS*NpkJm>v%k1R`zZ}_DOTfQHVQ2bY_zNP={e!o0jOa1uR
zzbNIyVwSkQ3g2>+AFk-<-P@<~k0#W&($B{_^*5?tRDV|eE7ZSE{ae)ErT(4j->?3|
z>OZCaA;<XXtWf{u>i4O?PW_GQ7uBCt{|fc5Q~wtAcd37;`uD5<u=-D_e~9j%T%rEU
z)$db(o%$QqFRDMQ{uSz9r~WPK?^6Fx_3u~zVfCL<|BzK$e)V6jexLg5)ZeInQT<u<
zuTVc)e`gO3otZfGf{B@_`b?syv%fQSdU9Y!z~pH@!qL!N+2aP-LbX<^OeMNIdpo;M
z@9{}fm)5H@wfcN2acbS$TsvbdGm;t1X58-Ht}}`y>=HaJaq0xNk>(SKIWn^GJU{wr
z_;a;3KUw!C;Mc3Zzr~)gpRG`xb~>jj^+M<1n&H#ITA|HTGo2GNrE>ms?0xJcw8g7!
zaXRz!Q{d<O^{OAD84dw1!m5R`#|Zi~jV-vH+>+^Z)(dmEBi|*2Rq|flb2<xK++x+6
zEVx_ph-(m-=5lUfE?1bw{ycA@T=03gUbX7Y`#kLV;v6C(J#Vs<L%a%-z^zl#_0yQB
z)tpXlaV$*LqxJOL!p9;aZMWI2{`NE=tPU!|?5QAeBW=)~1b{Zo;*Ynd0qKG&`~-o3
zGdpFdBp!nC>+NX?;$A|!lYOsokH%k%5Nu1USo+tur-k-<MT`oxVE4_wDq3F6p^dlr
z)(<?Q2zloY;p`4rc8h-up6NE?Vjp1ftzYR<!)^TtY3ZW!dn<6X;TGTQ_AUJ10PYIk
zYxG-2XFeBi*(Vrm{X!R)p5Y#nyPto`&hoEl=k0hP-r`$7v{YVV$2z(aYIoL%Z{@!{
z65sl@M>M|mgN)C<-qN@D?*@)Cv3}Ub0sHtT0o*FzYuuUFpnE5PVElyk2MO(HWiMYK
zf{H~}=81l9B)*M%c4)%aJRXRQM~iRR4@BbIxaH(S{F=7@+vubD->LCU{#`om>C*Un
zOz_Gen*K*3@tq?EeM)W~fwOkxBmKIzXZpKvk<!wiJJOFo$G>CX@-{#dXrcZTbavw*
z7@yl6>AoW`>Ej*m7Yo?9MazFLLXN~`<0u>F-NnNR;SzFRxS9SK=6iTxQ?d9q4tzx8
zN9m*RV~ER?EdE{{$L#$z<3EYG)Ln~j<Dh*S-vn#=ZRuM*?u(?q?<@}({JC_NUw)$(
zByI+q{|s=!{5v`hb<%!(`8&uW$Hb?a1>@f}=n3^*gMNI=r?Ief4W^nAXXSrL<3FJB
z6A}GMSVZ-gAwE;;ir_6i|MLwxKUaPn;%R0q?I7J7*7%gFhK(SLZwSJs`-Y?6&zBD1
zE>t^mJf3Omt^8)c*uLY^-@&&L;6`35UgH)`zY!4g&;iNxx2KyzI*?=QXM=$q$cgoz
zVf;aj`A-;su=7ZJI*5|z4rsLXW5K`<^wicrh4F_u*6)PzFL10s3FBYbnE!<Fhd1Vd
zVf+z}+4V4fiDULUj6bq5e+=V~a?HMl@z8R!n}L81XqVZ;02)$W4*ND3M9Fgp{(v?6
z6%6daU$$nK!uX>dvo~S<3dih77{3yB!3(0~xr4J)YxjZB4!n<N?Kyx({#QBHPQ!SP
z2E&f}dXV#z{b8(cv-y{HR1OItT<&oDdeEg5&+W}Ymr*>oF9+R)is$y^pyPOl>2mvV
z&|NCPa);ZCgYG?w=l0>CyIJwv9vpPHE1uhbgYL_U=l0&9lYh{WA8y|by1fD{ceout
z=-S%|;Z*=^zYV&h70+$CL3gU+x$QRSUZeQ*tHb#9isv@ipz{>ZZT>;Wk8dzP+{Pbt
z?@&Cq=?C5W6@U5hVf?L%=QjJG<N2h*KM}^un+Zsl+vJ1p`=D>fy9?H?6B_tS;M<_L
zH)lM-02KE^81Fd!v~qE~X3$ywxeYVuET7zF8FZFUZleskl}MNQwDqy726h9_eDVlN
zI_oD!70>f2gLo^~ZpF7rC|(<GQ7(>^C{J!%47&Geer%mCp@GJKUxfc>Tk!vq<g?i>
z-LL$&N7nxySNv@e{IiOWt{3u6f68ZPq#jQae6yWA2YAZa)+MdX8->2v-nJ^&jL<K0
z9*Ed&zJAk&SB>KAaGYLUE%eRy_Cvt)1)Qjz-=+9yzwjmD(`@JO1wIjDx4#QKtVWYN
z3<K)r)WZ&6A54E<EBGdI=X|A492BO%1o#u#?k-PDB%ZlI_o)_qj=;PmE<fD95Qkq4
z{0SJT+j(2oF2)pZ=XBY4^J3w%#7XFU2<jy6?JfAfN9ji+dU(C?U+C|)E1z$*;Pa~%
zc%B<XIXfyho0sN*r`$GOm=?}F6NT=Bitj3?1twp*ZvroJILwznU$LY61@H-Z3f~R^
zKk}cxG%W#EFJ}Rt2$Kqw1wKyx1?BV5htfjFGbiZyUgQaAkDJzb0*}fa`|Vrs`48o@
zcQh@OR^FF_UtGF;Q!);JP7C}cf?wi{s+@g$hg}RU=<f!e^4a|(KR<G|0&>Q8_?Ba4
zY4dT(EN(TyFLAba3<|1E19<)t`RutdEf}61LiY`!_n(9JeJ$`S@dpC=jJ_%@<UCu4
zF01%GZdzcT5kq%1@YILf4)ycX@1vcYfv3JrN80I=ir@Z3S}1jy#o>8Rq~H0BkLNZ~
zx|HDk=iq$=_!HZgIO%Vu1r1PKj(G67?aH*kCWosOzkATfTfKi#@#)8WJj|51Ujom1
zA3Y}RBlOPsV4OZ|13nR=^nq)Er+n;OKeG?JfM-6tPDu+P&rG6w6!>MNSGxi8C2k+^
zl+UJiU(WtG4BM>Gq~r240zA{T^94=LUJLpf@Ci*T=v@W;(inUG0pYX6c}Vrp?8##-
z_^&`ikor~nHcey*c+t0r{A)^|j_A)v1Ru<y|9q#??~de$oAzlRcCYb-@IsB6#G(t+
zb&6@h@a!VGD)7wbjz~Ul0$%i7?VjoRr<KpB>Y+bw#yi5OX!7Zb<Ub>LKds>X4NAZF
zCuu?R%nZ7B0ndEe`HfGj;BE(=`LS~)ZJFw8E%-kP{Id3!I1dc^a`3r3&M#Zg9}9(9
z2Dv4Me0m<gPnQBdPTz)G;EUvgboWHkt+k-PLhxw62M<bm1-WClhvK(8KK=~F-)i)V
z-=_F)x8UD}%uBngriIe_wTpq5_M-Kx%O=h{fM-39Dt{G&bF<(zv4D4<@=rwcdB4)z
zIi(im_!nt@BJ>-9XS&W4X(3hhblxU-i@o6AqkJCvgip`2mFV`hpkD%op&WL6JuP(i
z$lY;HQhZmNFK09#aeo3l^Rst<THrAs?Q8;`a<lV6tzFD0pY5vuXQ<$=0iJqwTgDR(
zX5T(3{FQ0IyIb%`_rKGEH97nj@Jqobs?R@b!RN@Mna@>Ddaa+X*?}S8DTh5NpTG5Y
zmjjO~YjQU$eb=y0FPk(#^*zOJ(t12W%X`=gpMQF{PjB_o4SXE`4Dbo?i@J+i&{qYI
zej%ct*8<OSJ@lVxVI^lU;i(sj&3fFV^<w?qp93%S-}CA1?5AD8Q*JvWa{ICJ*{S-;
zvw`SNI>zU-``Wa?*8i>tp7rId_vJ5vxMhJSe@FFUy9QhZJoCBZ@w8y%j3hjLRPag^
z@cs>W=0EXZTCj5F6`o#)7ck=FTm+u<Y1MOEXT2GC>L<_27pLv$3oZCO-U7c0MIxW=
z$|vD-bB2Lux?OW=!Ms)#`ZD0-@_D1u@4d^X=UHTQUuZ%9BjDLzN880?O22efTH@R7
za4eaL9Pab+c81vNfv5bZBYHKV^rL6`^fr$ApyD^(>*Kk9h3<>O$A1pqA5{L)@yt>9
z!!}M1uV{hK2p;+D7!>xJtW#}4|E?DJ&$PgQy9NHoz*BBJZcj_P)~;7#y_|Aes_kNn
z7J5+e>EnEQo~=o@PVh17#&RBT4M!r&5x_%3p?~4K9K6>M!r;v1C?<v|hAPz@vAY@7
z(~RnEM)ftLRyU*in^6NzsOD^?n#+@FE>9{ZVsm*?&6#f|YA$ngtmYCo7r42+1CD$4
z$c8n8Bd%MUnQ({QP-VLpW$PpTu3M_e>lzYfXKBa9YlL0oA3{Eo?2|Zl$~unG&Nyx(
zaCIoK?A7a;U-u;HxYYSRhn<VwII=odt=H-^I6gn;xb8VaWA4ascHDJw7JL<FK-ciz
zi(8*`b7h=xU2|ekuA8s8o6D66uk7Y=;CjvVX6BenIm<cU**CB{1<o-<E?;Z$UrBNC
znn`uu@n37n&rD9vBT4|01Ik0BuDfn*@VtzhS-%#U^>ePzT3;L1t;=S{-SNRSBN<Xq
z_6-*KFrrq++3}9M_KoWY&l?_c_zsM#IMi^e%XM*>d1aehE=+B%Z*dFNYNa}QZZ4fZ
zdt`Xc(1jPe-JLy+dTVhKeX{p#ky3VMa-*EDpBc@Ua8NYfEh;J_H`STy?0qgqtJmO_
zo>HNjnFVDZzXnjK4rlADeVNY<=QF-?4cG?))@D{0XQpyE;XgAwcJ6o*(zs}Sww~<A
zdr+ku3W_{z#=YQ84(EVR@-TL{UiC`#+HfwD^>E7!B{O5^W@`xUc84o9cLTl$Fo^@x
zn`Dr!lxvw%7>z^MYlV8Wl;gvC6o2$+d-tM7Q0SC?*DaSOa&E3RBf8^Ec)6{}X3jEU
z^0TTN--zt?21_|HQ*2$taGu$=lYLh66JFlUdA0g%m}t6EE6uqmo>#^p^U(tO#cQDt
zsZ{s~_aSXEnXzoW&p&;CxCGgkGBgP>^<0eB^SYRNj)@neAN|x1e$Qb+LB_KO%sukt
zZl0`6`OSn)FfNO&lHyk*8cCfq>yD|Z87h0V8tlyUOuZHrrj?}V=c{j4K5P+F<neJ-
z+;61Y;W~9M%9lqY^&DHem<kX<21{3%D@FC*Wf`k=x9FA1(L%02H(T$`)#s-RXme<3
z=egTT`NEVKuS_^koyjEIv3wk(g^DFp*)64Fy^tUCrnaI{mc~k1wEoS7I=|RaDdxTT
zrVI$bWLL3NE@yLI*`sMm^`hG-R;rUOYy`|ax`k?KVy2FlFJWzHAc4wx)x0}VuH?4D
zn&ayM1xD;cr^yytM>$|PK~tT#;K`B6p8m|pWG}8{Dua7>{{Wwp8Ms+AgMdjJ8+XUX
zyP#(BF648SyV=whO92gysL%klU%qV8&`H|I{u0a_Ulr{cM%MxR7nVO7fLJ3jHtwWX
zYHI$bv4Mibs0?In$qz529}_pGF~!Fn-{@;a;FBrOA+S_()V_EzRH;<?MG2}*rW9+7
zeD28sj!uRq^BbnA2$?x(MJ+QrnuCh5r)Vgr(D@oXS{elbB$u1>Tn{g|Ha8A=wHJy~
zF1zU0Ch^8@z0kVqVhmS|<YJ^QbuOQ*S&+GwE8H8Ux`(zDa$5`4H5DlMTKV>lCj~B1
zrJD3)ZL^-BBPrL0i!og*ZGE!a@1QE$14GNvMQ5^ol;T1Y7<V%fGc`QU&-2)jZ2Y_-
zj?u>D&EJI%w1s6@7-dGjD>Iae>3_04E`E+XyaDDuUz&1f@cwk8ep@6F%W0v@cJX%f
z1cnMx17!V9KzWPRLbQix+2c#@h6Zm3Hje}09#}ql5>x!q*;2OG)#|SGptqV%cFDJq
zP-T^Qj*B=%id4^mP4hrwFqj#;xFO0&Wh7Szbx&+LjJThX1KRzbHPwoj=Xm7}m1;Sk
zDWM3Wtff8-Q!3R$#FjnxpflMMmQ{vAN%}&8y@5Q?rsuqg(rhxh8qe&=UG`&W_{9oj
zgHbYGoOhF*-N;Bx8;!GWy$#Fge_SfsR9k3XP_|T4vxgDsN*j1SI<H(=hUN#<y@5sz
z`IVFNCYqXp+M{0D*}%7wXk%jxGgP#pM~%kQR~eql7v^G=nsU)0Uied3uxYT~()D^!
zl;&xU3sYnBF4(^sdPWenXaw1MPY$1A3bze>kO@NqbQ;mdpRErxjDAyZo3B;eE#6cf
zI~ypi=%i_3ZorvdLlmoFtMFN?e6UJVGOw;-jwfG+Qm$3h!~if<ShqgP*_9aX{QtG#
z{x#J<u8V3^S-6u5m~*Lm_iZz!LM>OoEI(hspq|oTPg8~XDwF6nW=pxk08AK$g0mjf
zHB>*H$^M4;B5fww6SPPLSQhQfr)R1KceYfm&v+Qrh8lZpx0*}^8DR|BJmt}NV(9d{
zW9MOg#IG05%(TaZh2eNQZeADLzq;sQ*}aaQo9%XO1n$q_JF(o;sGYo@0e_;gHq$iJ
zz^rewG8^rg)a*A};M$alKT<*q@C~`tCn^SCGc$G6K|gY2v*89)HC+*oUY1>vKfTvs
zJQdyl#$7*`Fz~Irnjg2+GhHZbja2+Xt&7c3p^tFg2~u~lDX6h3vOZhsEn>pBrI|&f
zJK*X`t@K;;2r;9!<N<nwz<`~i4QX_sRs%z-m;_o(1tNk*4;TG1VrDXsMnrR(nU*V?
zP4R5a!qxBPxmyeKZVoji!`}7TQni*Di!7%!M)R^d;?7LPG<Q{i->-RGG1HjG9e8+@
zN^<_ec}Qqt65o%(wEadCO^i%gTTwkty6)`cb1JHD6jIWKG$ud8Q*~X_8_U*W=S()@
ziJN+cry=~zCNl?ny|8qJrMVVMS1pE}SZzep4}7i>14pM+o5obrI%Cs3b`7TB&&O~G
zgJT%lOr}+<fhl2Eq*gqiK_yql-B9}#yNq07QMfdfALb;phPB;nia#uAcDYmm?cKZ_
zzuoATFp}UGK(QHeGX$#jmXxl)v2f@@szofNf&lWyz|~*p<!T>hpp{&c8ENiDX$!b0
z5ivR1iDv!z0*qkRE#RA&^O&|w!-`g_-sVD6-I<mIj8x3HxY?dy)FRAd7FmsD#3nVZ
z2+Ndwf}efF<h!2R67IwcvMm%#esgo~G`8JVu#vWe!pwGs>8MM;VOt!&cotPYzA$+8
zha8kC(s(<jDySSQML;ky*1M^TwrclTW((DlR~}n`cCs&9kVP>JhSthhND`UaHW|{o
zfY!u*G?Po>3r(l7R(rwJHYh1(iM5$Yjt=l=Uw3C0nSmHO<&_I@iv*^iu}KA|G(}-m
z>Bb^yrui?3rePlSlOtkTz7LgeK}BoZylIrhW+3{91_n5}>A(tMlK*j7O+@r5Zm_B)
Zl1{2&Do4HfGUleig(9w<Ig&@`e*w$#_>2Gm

literal 0
HcmV?d00001

diff --git a/tests/Grid_stencil b/tests/Grid_stencil
new file mode 100755
index 0000000000000000000000000000000000000000..92d844eedd06f93a24b40886982bad57a5f3fb02
GIT binary patch
literal 97307
zcmeEv4R}=5wf0OhK-B1&D2Ax0u^nw93W=g-3StH(a1P8!5YS-JfB{S@LL?JJMF>n%
zIXxMgw)9q7+S1#fYZVK%whgE-AwU9F3|JMVNW{N0j6gso0b}NS-+j(ZP6Dy^ySLAG
zzx&*Q%szYVwf9;3Z|$|$T6@pZ0^isqyWJN5B-uvU7~613z5p5DHrtFFyK(YudA4(G
z`L-)<gKhnR)enDhynOR19(I`7;$aGK`ANo~1Alo*j4{k-kO&8hm<?|*KJm9R<omim
z`R1b$4!+93jE6Q`rbs80w1%%Y(!<x!6!3T`hyf(iDZqRZ4<wKII?Q|>X1;i6GwF|q
z@%piD{F`E?Pf5y0jPZO1Hm~^Z4vEW8;Rk_Pd@ghtbQU^12HXrCY&d)pp%JxBLO#;@
zRX_1+-eJD4uXf2d>5qp7RW{qOxn=j=Jbc*PS=Y}kn^(E$`bBv+Uw`xPp%wFo-XQAD
z@`*d{_NjuTcsphw#`+D!AFIOjZM#0I7?yY28zbio_|w=IQu8lf{?c7cy9s|RgY{Bp
zI%mvm@Y)Hx1%KbfU-Pna_x`NprR^&pj5Qy1-uCm*wEW}meCr39Sp~2A20ocO##Y+z
zw?7|TUjK;lz@LGB)=xhWbKzN1e$@y31PD4C{?#D(Z1BcD%F+9P_v@qlWqshYoioou
z|CN2f>-s2XL?7_y`hfqa5BLVW_iXL3rw{xG`l#>M`@m1d<Ud>aPxpa;V;}8sx)1y-
z`Y3;5ALT6RqrUt4sMq#B;Lr5|f2R*RAL#@C@;=}v`Y2~lAN05P0bkU|IG@@_`DJ~y
z+edxS`Sm{9zqpV0{<shLJAKgqmp<s61qOAtaW$-u@=N<D=a+rdtFRAzt&j4b?Sr0O
zeU#(t13n7nTxGk|w)TdZXc3#ONPOFDf5&?X&m7(p{-TK=&NbjwCjLz(epN60g}_fk
zIWulD@H0*P|1k0M&G#0XaMf<Kb4z-@2{D4BW;wPSjdGm8<L77iW53K8l#lQ3Plrio
zt4U9lne-&!#Lv&jHz8(#^{OH;t7OO!u2DDJii_`maQ?jFionbTf#PCY@%V+scg()O
ztRgUb!I-%-D=KDJ*ou`Y{^D7)7ntBF{xNgs&zn7E=6!Q#8);`xDJ~C`E|@)YR`CO~
zA1N-uQvm39-#)owINng6n^Vj}D*_AVKT>eVgvm3Nk^&G@vS?;;dF6uH#S6<81S)6F
zWsV7x12>ff9x0y<l-wJU?cuUnv*#6;%_|EOEEGkS4KJTP`vDP|9LT+;xEODq|8Vhr
zc=^n^bLW>BSo3e1Gq<v$G>&$&6|H1uMSxhy4)}bKHGAfRctyqR1p%c5(A<*JnG1>o
z3ucxDDnLiU9e~|<-^_}#k~p?9rFcqCalz!9iwh@AzOfjji`p!hEr=0iyV0r-md!&;
z7!(Ea%H~&q%!=8=71YL?Yt|=lGl-r!cXk3$F($X5a6(C<o!BT9x0K9(uzco%*-SB-
z;pVdY=gkMeY?TES1$UH*lt6BtK`Sa&Ht&9;dS|Pb=!XKl4X>(T6^j*6II(>8ya(sc
zV*gFK^LBI{YPM+RePs)C+1Mo<C&la_G}xT^@ou~i<7HlP$;?1Wsi48=_ZxDGCl}9t
zusrZc@q*dq7<cm*%)Ea#^W48^QE~a~1r_t>&750?<b^qAwu*;mmZQBf5b<`<bd$9z
z_f^j6yB^3fIWW9rer4dyyy)LP@`4x))A@5N3fOq_9)3_{?osU<a*K->K3H~^sxzk>
zW7Hfvvp7=bPr<}OLU(t2Z!r64{*C38fdX`&7%Q{q%}O-blqsm{Kh<kzArn1SaaJlP
z2hb7bG?_bl-u;2n;@Jxp%wG^6O5LQ2k%<ApF*R#$;X)f2$b$y&m^=S|Tg5{Q0yaF%
zCH(%9lHv-c<=Tp;&U?5F)wyGKMdgFDnOI_EDxY6bwy5~Q*$<+Z@1JKw_0h4#<TPMF
zb7q#!wUuLt2IiPeg1NG&EMTiByI%~NirE1h@|r<O`6IUbX9ogh56+%HXV%O|h&Hn%
zP&R*_?Y@~M4`8;I%qIGS^B2zUh7?yk^5A{wUc7;1nY{18nPu~Aa~8~=Z7ZuNubfvB
zu#FotrufF8Iks^=W%QWh8;0I6^rpmf?r`I=s0b*-hq`;jZnBLjD)Nk0igSnF*drk~
z*BS*wPJ-;8B)%D9sldm0O2+OW89O{e2{*ozum>>klWYS3<8zXYdoAJ(F!7n|906zQ
zxk&FPQV2Kx@yXad+Oe~^8*64<elEP`k@IZGJ;X>xs@*o)l3saf**P|pcw-_x$@Z|B
zo&!2oegEIEO*z+g8`4>(fuC%_?}rkNGz2zvX*O1>0k-ibevOFtw-uT3*VaFPr+&6M
zW_)JHkAd&7J!HlM**`$_lWbLp^V2cRSU0&>0-=ddd>tnm_f7Fnd|f7-d#U&*E+-Jq
zy;kC5uFHgvOaN^O`GN48E%3OUKzN=79$&8sA7z2Z<p;vu7I;REQLh1jnIeHm&9uPL
zFo{o=1#XNhBPquMN4F(Dc@{X|9slH8;MRS$Y=M)e_@~eUkLv)KR%C%&`(=g&j)9l>
zlv?1a38>9hZh@a~fiJSaFR;L?EO5yJUtxiBoW?(O7I<15i10}Z+$tZhwZJd3;6G)7
zbB`bYY_Pyb#DNIwE%0;;JZymvvcQ`y@JlT477P4R3w)miewhW{YJtafOib1-@XIau
zZ5H?y7I=pReyas;Gv#NF5$?s~pA-u`GY&-Pw7{>jz%wlHuUX)k7WmgK@GJ}5Wr62d
z;MZ8-c^3FLEbx2_Jj()?E%0m$ywC!_)&eiGz=v4iGc54yEbvkb{5A``+ycMe0$*f-
z54FInEbw6#_zDZ0d^`TBv%quXK!i_P;5S&{Yb|i<7~-F&Eb!rRAVN9#af+<<tGV<l
zn_LqP^wWPOfZI|UEl7Bb-LwEGv8x04mafdl6XT^sinPY?x%%6TlNCl<MEpL+$pRx`
z5x<*pvcAX$5x;|Rvb@Mz5g*SuSzV+~#78qu78j`!@ozHDVH+tI@f#Q?ON-19@oO0;
zD~lA0_}3Uem+^cNzm##Zv`CJKU&uIFStL`$&t;q}EaDXLB*w|QA~q2}eF@@ZS&_Cb
z0Ioj4I9XMsRm2Z7P8Jnu5%EKelQl)cBK|hxWJ!??BEFY#vZBaZ5#Po5#f;a9_*TZr
zdLmUKzKL<NoJhHdKgT%LmXR4E{xsucJ&{5Y|98g8aw7R6{u9Q@Y9cuz{sYFzVj`I$
z{$0k&S|UynU(PsLO2j7Ok1|e{5o!B^^<ThvCgZIl{%yv|A|fp!ejnpx4Uw>j-_1B#
zLS%!8-@!OpL1e9nk7xX9#_L3UG~;9mktz}YCgWrUk#Z5gfpM~c$P5v`mT^w~NTG;-
zjd4!-NWO?)$~dQbBuB(AWSmnxk}2ZnGR`R+af)~n<D9w?n~0yzK%7%H($*>ZpK(sr
zNUMk+W}H(r(jwxA80XZCgb^>?QVmUnz8ytvp^jlQqnqk6D~m*X6>eF3E7E_HNM}@}
z>wf@Dx+^TD_4}R+v$Uz-Dg^v&soIHxo=PCDNFXjT5Q9As!*Rr-$i2XvGgP{=T9m{2
zT)1T)-+y)D{fwIQvAqhn{GI9j66uVZ>1?muBZ|7`URgaP2j6cioT4;^TTbMI*oGM*
zXj*$|zD<7<4e{w5=}Irw*Ctebs(gJIfgJ3(Kvq9^IO(3odJ;rN^yxiYm=tT3uB;-g
zp0PN7Lq{_3Wp%6EaQHUa-Xibms6^&XM3igxBD1XSF!CQV^H<%PeHmJ<GHohy>Pygp
z_w4AG4v_tmM%G^r-n!@_AOx;RkT7uSmNr;bfFEnQC-NAwNLSh-6-GGKbC>5%&s5Kp
z$?~#~Gr*RVElA+!HuXi<r{G(%`n*eL7&7A_Bd(`CgXMhsPmx94p=dWN+MOw~`lg)S
zo_j=AKbIRmPnshg3(t|3R=Zl5zp2{Qijbc`hR^-E^h7_f3H@5Mx~y(blrYzoC#&m>
zEC~0BS9^a3I*j~s_6b>ijl~D>zGM2O=nOezA{f{&y6TZ@Gnw)xa&G<3C-K$r6O+^>
zDaw#h1-b9{rgE)G<%T$wX@bgGU~YOf-lC}6eCn@VL}U0ALM7x_L&BIPUIPn#KJ_F0
z(a(|7L}L~?w8mA4^+85EigpRD66DIET2ZvD9@NL;m27XCa{9v4Jz$D#;ZNAY<yeu#
z!wfvgqVH|Cw%bTs6(MN+6$tqutPamRs$FXd64~nUBp#l^12LKzB3BDh@F8p;o{>u@
zC{Yb{jjIg+t=c7{cGCK4mk%MZI{?KDaIzwz)RH#+$P+dj3N2y@_-x_2qi3Oo<c)Sb
zva@70Mc&ACcAu=uDRQV#l+nS)Q+LR!*Oj3U{|-rbQIy0MVxm6rQ2@0UMMSQNK?-=O
zBNp=^nGNp4K+ZwIa>*J&YVJ<CW`Cjq^O*wD3Qf`i*UIWXxuMk|hX#iECMPDx<|@#n
z=b_Ntoi$-u^41Eltk$|j{*Yt8$YO71bFnCfGQyJ$l2o%E%PL-xXTE|2r4T1h&=D&^
z8B76omWll(1pP-l8J8evok`F-m&l(WNGBmIieV<g{FB>3(d7n3PV*h4s1==%pr}U^
zwJ`-0+5Ry_7n7piO(gOsC=yM?qDT?d&7z5FF(U8^og)Ds5>($OLCq#X`%LUFX`<?r
zXCX-B?@17gB0(((1wo%fRFc)Xt}IX$Hs2#-%7#g~WdL$y>GS3jxJpbKYa)ADPrILq
zm|5Z#UYAqGoT=x)4E@?whvA>QUsk8PGJ_+Zw7ta2ODk7GuG>PD{pRvd@@<<a_2`$P
zX$4GuN1yi@xl+^STBdTo_|%ww3ROj@9@Qrr$%-~W*7Eh+wRL%I!131{k*Y67$)2jm
z^K60hJ>M-D6TQk@Uwq)G^~|OfU@GP3Y^LN1V5~*j+{V&h!0TnTNpHkk68wEFtAqyH
zU^dVe(xMj%RB#!lhGYk)2{9D)fLG08>?y<&g;Ws@h14PB$0+1Hvk;ML7D5bJjj9<e
zV+A6c$UHEg$!YLNh9G7<TAMHTfweNfKJ-tP_p!<OzO~GZkHL(7ih9Xv14|YNA&-;e
zfk-B=urvN7;=vaM9}#eA<qd!o<i>pHRA|#|(es6Tz&<9fPs3aX*q^;aX*imsl)NEl
zH_HtjNwSKk9Sa9>sdyD)8eb9}1mU7u5LJ;XhaBIJzp<d^q_lE165?->HG@nf1e{*A
zE!b)oL&vLj1qON5j@6#_JT)!;u;V3AV)F#Mk`~;rsBd|)kIJFxzjG*|#ZNiR{#V;p
zw=eBfCwH0;J65;fre4IzZE|pXk{oPIk%LVsp5Trybw}kMpZaoi@SHi_GFY6TIDw-z
z$M5X>R*z{P`P682;O{)KChrKN*`(T5$m1OGIb=7B*r(*jT+Bcwd!~xy;v<%`K#qkW
z3q~C0%KCqth{eGD(9~qo8<|Cx3ebmUQK3|@<#!#jU34o;W%n*(Xc;k_!JNF}?HDoK
z1nSpeCn9UpU0KA;7dgyTM4RHulr^6#N8g8eiFeCL#T&^>Ns2&5TyS^{MDk;E1c~%Q
z7KXK~oMZ}SjZ8%XXmDt;91AQY%rFxK3rEv1F6vpcKX55Uh9KI2YoEv<*1m@T;jn8(
z%X)&FJBTi={2DU)fY-)(p)PhU@@ebD;st>Sav>?tVOMDQU+a>^<2j?X)L)J-ajdP`
zDOLXj#mL$!*9?(#;b?8(zt`-Qs)GVt>&gIFJ4X%;56cZ7I(+t{KK1k5BZ={x8<thb
zZwKS!dyX_esb*(o${eXYe0x-?3!bt|HO<)m67UC%h5@lK2Aoq4ZgmDbk{7=!$F^e-
z=E%VvIl(WI7q5@ris9Ei{w9P*zAra?<dBn5wKx?uI|KbdNi;=HeqMC%raDloJ3j%R
zi)~YuH8EN+;<s!fsrD3yC3>6fCaYa~88%lZq%<$e(<q!GN%Sd_ZjUFGa=h0ggC|%o
z8rr7?T;+-witH(A<pF%NMMb-#SJ>lV_;})au%!ygx#2l=UMb7#u9C8R?m8)JihHe;
zHQl|TK$`9yRppVUPam~G%9`bF#pA3|{Pv9k+FW-Sk8?+brL2Iv4Ud6Q{GKujxfi=z
z@VIzXi<ISX<H<2fR^OHzPG>@_*(*03a(Ja7j&x~=FWo5(nUbC<4Vj*vBMq6Go-Ykq
zoL(pmnUy|68WKn^mxffQlgL=uXMb5Xn#?nz%4o8&*e|eM*aB>_o3TF?O;&EKUf<{X
zL>-4dy)70yAvrPZPe^w|biF$)F!XJJb0ynPOtu);nqhI|MR8<8SDy{l5l1bxqRul=
z&$a;45ye;lQ^f+9Y3%33`giop);|;}*1uz1Wdp9P#QK#9vdOh_NIpKsdNRQv!H3-x
z5tPt4v7Qjg(=?84VpE#Hwb*cw&<rpZXUw94R1O$!H&(Yll4|e{48zh~FuO!-QR1nb
zPtwXa&<EY~A^r}nAO(e_)Py&)Xldn7fyMnI(z#aSWe%h}ShTcqwn(fZ++)xNX<&7`
zn`vYYmiZ?5wGC8@2Z|Tsp*8-{f(JH=6Da(M)D7{6webf|K#^312jYhj^21Rj%D|E?
zCL|a%+fknk6%rGx1d=I8#}jWBiPbKyp<K+te*cEY0`5$`u6$YbxnxECE4H1v`^i4K
z+e_98I+>#(k~v437oH=1^J@+EPdw6+>8?&`dZRQApz;>`CsI)h(wn8B1Fz1<kX?8`
zG#ePrTRqw>9_=oNN1K>Zpp8ofXy=5G_d$?ay^Zzog#!B=QuY*A;U|)FDz;pA@5SGK
zb+Bv8Pf*SVM}ZpAAC6*=o@L<pJMnfW-X0ym{!-MVjQ9UUnzj>vcxO=qO7gkN@znf@
zl*T6@I9y2GZ(=kGj713yJYfcwBEA(l@Fd3?vAU-y!EIRGKTnowJ_8R_z-cwMRxq5w
zg-(`(4H*iUu2l19B%`LW9Uk=-!4!qG7$2){_YN@>YY$>;@6nbzVoe@(slyy%?k1`B
zw<6);6c+Gsia9pjJEYnNjU1^WM=E9jM?P~5M}omi{z?wvL{sX{kGZ)sWaoH|;`}6l
zG)3KR<tNAVgJ@TQh{n1*%&G1_8(MuF4eey0T>~^(ZIrcS3es3HGul9a5Tsl2VDOQF
zf{$eN98Y{n>ES;33D@lhMA1!p1zzgMx}8E6*B?2!n3rRuV<SreBon*YUzp%J0}RN4
zh{1R%_3F+HU?W!*rl_&~-OOY_)+Qk91O!q`tbt2Cw;v-?pOn>Cq}m^1IgDPZj|WEN
zT|DdgeA;&!PdD=^9ZbFsBPIB`L#n-l2nV73VewqSj2C_(iZ`SS0}?eMLc$Sp$y+8Q
zpMsKrGzrLJSB4QyF%h0hAe_{Vddx^b-X{dppIQR_7FHY)%KZAZW-AD7jwrs~%uS_`
z@rK`;5TSn&II94GY_e<{bztA}O!wU5xz|&CPa||&=oLA%&d}MYQKN%qn+2F1gq}z?
ziSF(j4{<=1=$imE;63~1Yi1GMed81v%nhQJXOQ}giAK4>fUGtl-7Wfr3F&UpN&!i<
z=$}6_+ONAse{Dj#Tl5En#9Q<Ol8e!Kz$Cl7F&{89b~olU6Vlz7;{frg?RqiVWXlCf
z9^2;IKf|PL#8lZh6HoeBAfRO}%^LYVIZjW|vj9aNI|ELp<c#b;11{_ny^8qhnOF<Y
z1l<WxWG;rc$Zgr7M^|jw{RMQ>$MCf&Nb)c20{KsT2>g-2j~oD7LZ(QkQK&GPFdjDJ
zPau@l0{}%YGC~dlp~<R=RZhYtHZm0``Uhwby%k@PAAk?AgeMa@nWMYoq8}Jt+1+t(
z5rTqlV?lp4BQX%h^}?dx4St-VsISZFXViO!9!rtc-TKeAaNQL;#2`5t_8dVjQw+mq
zxWNq9nqi$8RvF<KFw>=DX7H;}ZVHky^1~AHz?B$Ms$rO6hA9YvDkI=WZc3hE^I&>Z
zrpQAoAqLA0UDy{_$U=Pf)DtIg8AvYGpM~MZr;U+iEm#GptZ6F{_(COxzR(Mz*leub
zr;>aaN!hy>UWhR)2ai?hm1e%^`4#7l4>_u2HOP!8=1HTNwG4XY4ZXlb-F;?WfgO5*
zErTMAKqH+!i-4rJ*yYqaV0|Ld(Mu{o5FjQ6%n<D-AhL$W(<T)4%zpuu0-NHw&#g!?
zGL9+og?`m+yxR!0mNVWh@-B(z?XEU4eW5?L8rh9N(;4egdt`6dZ#y$PF|lX6+!YpG
z`qK`?3vze*AsSYnhyLW2GR3F)(JmFGvi3|B(*4?0)c;%1cRg0<)1G0YqT|?V(3gni
zvRnQdqd7K?Rvnv1tKOGvh-^Y+3zP_(rS(CInX336tsxZ<9^z2Nw}4R1@WL48Ts^vd
z<Q8Pz2qcoIif=90iRaPMddC)3JQjfWU$1Bm*Jy3Z#!e(msqciEqF&MZ1G2>eiC%8(
z4+L_Pqn<;TSn<bLlOJNr@>{_a{Az<wLn~lph!Lr2PGBdRcQ*%SI#>BtSS@|p1exvm
zx6M=;iNU!7gOfRYp=T)Rq8KC4vd&h_F@3eAm{8ARm;;^7d5B_+K+8X)n9JDNcbLWa
z$X-d3hh<$Tq6xZ~*H10Lzu`gAoVh1`YO@?{?Ldat4BdcF-7N<{?2toCI^>39$q>0?
zO@S+A^^_cZ*It<pZnGE2{oZ>?a2_ap<~K*vKau8NioZLYFQuSi0c*tHE;$%Ws=x%<
z5%tO;-_tpA_E8{BYQfjER(#ED!&mkkDKCF3zh36oHh#SVQ5ACQj<(8IksC$r!`~jJ
zyv?$u3n9hl<>MXY&2~OD&ykkAs<@9=I@FFR)CTkHiaP?Hsz~GBlGV55?BgIlS<e0>
zFaX>M>VzyAii*G`_)e+>fuBVk()y9l;qPhcC)A^i$*~=32gGT|f@{={xV{5p4`|-!
zfbV@Sg<q+l)8TXPUhqB|HsDdT9KSZeuMKo5+C6CE3$~!8ZMjEG{oLyIk-zt-BcIX+
z{@x?4ANaI7@EO(dJG}1qc(Kp@%7S;~;I<5l0E0aa7xIBZjeSi2>$_Hq>72SA>jQMM
zQ(R7^#|rTnnDBpx6{0nU%@0-x<;)dg6B;PKK-3C--Q+VCh>BhdgiyO$7KkZkKF&co
z<f!|y1tO8xTp%*e%)1s8$zQTSm_>*MB13;OgY$E8Vu8poDX2#&#sYy2SdRt5%rCg~
z>UeG}5Jtu?Um&{kex%<M&)Z#Xv6O_KX=M#GArydFTNa3TcG_OAI5T^Efk;NLV9k#&
z5HM#;)jZ>;s1dP72m=E)<C&I)!lzBfA~69gp;&o^py$&vux7*;i#kPp!>_h7B@?Sc
zJy^W4RD6Xc0?fWr!bVly+Hm+v`_?_D)v30`vLV))ahmIev8rn$3fltBd8b%P`dC3W
z8Y_rj-Nr>Ei|fJ|ZOTiQB}38dy_S_|MqSVK!yaEheCi1$)}*-Kl$KvA<j_4|Z3+1q
z<n#Cv@_h`K|H~!hPD?T4dlvK0EFqFv4A+bdFxtzz{t{n8GR|B=es$I*<UcU(zTy(H
z7NvtVeZ?hYQhW*Nq3`+M_2et8Cs}8%C)==2lLIW?|0U}Q965ZL;PBR<qEu18khT8G
zh=NS1M&)jn+LQ??hG687H@R27m&gCIw0@6L5)ym}vPAY9fy{;Hat*7KF`5R-?mZQf
z97^43m~@(DbquyL(_I-&Ue|C89;o0S9}Ep=7lzlF@};6Zo(a~2`RdOfUoyB})`rR2
zxKeHSlbG}y?XYuod0>&l0M63NQ({w{o)HBZQcYNl)yKiCCW1rVRj3VL9qh78zi0@K
z{E6Ko{bCzzYEtbIV0kkCiZTmw!|D$8ZLfMfHZ_CJEzCN=qg7^jgy?7_6=t}%SNaR8
z-<PVN7wneo5IhlV9CeXZhs#qBjR)q#F&XxRERo9xRa}O?%>{?73-G>P^0O#xRbPbx
zpN09fS*|Rf`k^R~`a5h4Bl<%qQZUxuB%rTEE;9ph3;6o)_4S&w(;U6$tfhY`O$X4U
zXU%mMn*IP>Fwue$j}Mk=2Xgxy+?Hn<xmzZ+V=Et~CSoxt-}nYK5ih<06~#I{840C{
zgkK>6_Ao;}f_GH#W2Z-b6=OqDKg9-`d|XzO<&vj5kcF-!M}%+N7^mDcCKDiA0O@+S
z0|_<+!f2_Vem&ud;6qQV6X4w9Jz`7vjf}w%7?T0tgUV@5V=&7oXODDY^g#9K^k5jI
zc-7Of7BT7%q7%L939oueZ-g%u4Wz+sPS#l{uPFOzk2wvs(S4{8XY8&wQRA~|HU0p?
zddh>PLSKM4%IZj$PrKP)^4m_(qm50$ARUP94@g29m^~yjkU13*;Q+Q@QDaI&EQw8$
z?R7yf;C0PlRRWj$AQLy}qokTK7$y4kAVyXjyR%ORGA45ocDgUH<Qc-<>vg%|3*qi{
zP#?l-gdEQ>u}BUDT!lW$tH)wfTp4aqCDjNc6gFUmM!h%vEfzy*AMo|a7?y9X!;@6=
zG=>oK`{d9R=(a;XmlwtORb8)lAl;xmla%K$Ye5ZW=?^hj`LIGm<}EBoe?T-VCoW3J
zB2`%vLIdB2d98)gbPl%ga`2;Mf6e=WR1ky?ld8XijM&@Q5eeMIO%5Nj=q`Av7)1Fw
zQd)i{{+v=8v?KZXz=!21;KGm&%l`mRj$!}xRk4&{r$bu42gUQn&=;}rL3s1oC}7EI
z6Rfdv;4+{r!e1%=its1PO;syMB|d@x;Cw?$s|q8IV1p5?H3GI9@LLe@W5wr&8Tg!2
z2y`RZN2I2z3~~xSf&f8Ehg9Q-uoS(RW30QUL`UTAmDM!4#LJ!?3560Y!r<vJ058IH
z3%D}%_t42q_X_2@7#NzbF<-{UOIF9{!$8uAHW;152Xw{gOg^ALMmu3#X_V673b8!}
zv(dLvCQX=FSN$F#%<-562X*gW!GB6E{0HM)#<0wU8QOR<2k0e4MV4F))GY^K-PnBY
zr{GZ!qu7nPc+xX~uqBN=N+c{wB*gYW<M;Gy`1%dLA`d|4$D&cz2Z#&*6S{X!R@5#<
z`*}U``L)eq1n_GNe9iPv8DUJ5AEGQ6)-x3M!OFuXcA_>5kqFpYqlBl-5;BFqfSG@C
zul&SK<iFX-&k+I&*#D%UNB)a?<tJ_;|H&4<e@VGPs&2%JjM&>oEZ>Oz)res-(fx0;
zh>c?8+;5R5M%Do$p0N3+|L$b)a_AphyYjKlydX3rB4;)501%oN0jMwl9H2(o_Zfgz
zK__1}$pADb0Qm-hgWt$9%m6%@09;`JsuF<n1>mt#13&gEiFnK$C&!Igj*<4B5py!u
z|67s2A^NSxUi<kXMSVp9M^-;k)J~J5<&&co;)S-}vnBvjA6Q|gPi@d^;DG_RAW9+t
z>kHxPt0wzlhLUqn`fK(})z_dM{4gM~a*DRj)r!zZmzZ4Wzur-TA9g`+5ck}v^ud8H
zgWEn#zQ-MOr%O>ncf;|h)B$ao9}4o|2VJ>4bHjBW6}G}=ZA_<j<LdUuJJkld9z70S
z-^5O7{Q$2WSCl}f(gZUl9g%3V+}>5W3;HYWmz+MzG8AZ1{A#nk!H4xQy8}v{-O)#U
z>V7%<^FT`ONvNyrfk7VV7O}@0S+DMp)(2j)*>^)vr9#j1dC~%(lKr_q<f{jP(>K8Z
zbnZ(i!Wb{4>^$Uqq*HB@)<529Z&#ZUb+eR8o4rd3HdttD?5f=DQMoiiTF~;bMu7ET
zMToU{RIKM<21aM5?`B6YMhA$oHh|IKwxY8z4+G)DS9;*T$tY(1F${ukRjM)n@0?ty
zs3(QQ(qFh}XT#h4bh5nBB5p&`cFO+ng`2UcK~K4x+Q8xH-zdCpP9-lmSMJ9dtEgjB
z43~yQ9C}UcUQ`So?-(SLWLYhQ#uG-4G+Best~ME54v&0-mSYk^0foyRp9zMXGYP`9
z0B0G`1n1xpeJH@#K=7I1d^~2FV7!da1VdVwlx2d8jAw#p;4#Mpmm1Fmm*X+d1TQk4
z0sdhXYJwP;Ki~R8!Fm|QS}2JXZ(9m@Mx^|&qFeY3U;Ap--&UxPM`!7?(E|y7E8OX!
zMFZOvCgOpxJ1W}36h#~Bl(XC5^_SdZx3ksf)}<g^6XCdR?tzjO4URg;<?K_iW+Y=r
zw2P+%vUmC1ho$AaVdnI?BXD5wOQSzf?0=Ph(BN~oN<Y||vdbqGv|`soal@s!w@FVt
z1LSV;XJlC41XAH?gBOq{+1(C_Osakh(wHxFd%co<N~!;zu@QR@x?QEc6^9f0%dxGp
z+Jc#r%yS93Xqn3EeCnylZ5T2>h`C5WRRiZU0TygM@;lZh*kP}{2r;Ns&Bs%bx5+?*
z9Z8k&c+{hAK1wy;0+7Xhmx*v?G{m5>Sb)c3?H!9{x4+ypHXmhzD-~dF?0^*XrD$@j
z`Kx2WsR2^ZHk1w~+0%Z{*FYXRD=B;jIYxt!hsHu48Vh-7EaaiFkcYm*VZ@Ih1vrq0
zz7s|q0l6lEwMM|%1^gBS{8;e~dFVSG00Un)f_+44`cAhz)Bt%1r0h22A^4k!?L@L-
zADa^41upa}xE9;od_^4x$!37$oCqoAE{HV6IUXrZeBG6ar-@E{4WNLfgeIbicLuIu
zHd*!b&L*WzL+?yN?<nXUyO7Dy5dSw>f~V2=W7+fBUAcRE?Pq0}7wj$Y{nVxp0v*%S
zK!{@y99{;Y22%#13eyLHkm+gonvw=`d}*M^kwy_XfT_M_kTILrq%njHH%MmYB4#dS
zX7t+hG<;1-L*Mz*(0Sf8S>!HcZgz^Su1RAkIjaY;uOw&fptT}gHE4qf*9;0XOk0y?
zgsaj7Jgqiut$<gjF(1ZyeE_9a4`SEJYIPbzmdYMwsiK3Os|JbwcCHy@bpM((BV3hc
zjDy-VV<1$giIK2utB_LMZ%8$}vFo)=+K^+f67n|5KJ_R}GGc~G7|>uPJ~bj$Yl5aO
z_%kyem?lGsVLpF@yzfglBYy0{3en$tO*BZ#6p|?)LpE+2osYsXly|~$^DAN*vk*on
zV=rRxSH1xAug}t<s{u3Meq=N{aWwjHG`es!=HGI5I6s02;9&kO4<nAi=$++jjhN`0
zrsXY&@ngj|8h*La?#qp4UoKj`X?gej12;lN)f4gg=fnI94Q?h67wh>gJ7^UgrKtC&
zOb9vNS3-^^tfpe0F5FBSw+x4Lua|GfO0vJnx`x+@<5^R<hQA8cw;{>-v6GQ|XwTyI
zZmd=Zbrp3GDA98h7GtbESp9D|m2C-huJm{JM(08gdu||C(QZ-Td+;O>eA;<f>T~zk
zX^x-61oaD#>R4?)ssmTUDahw;T<{)b7A%fJ`G5to3i&Ml2Uw{ghIT=wm#QO3!D<Mq
zZpVr!EcUYdEh+dS;PMvi>G-+L&oywXFZ(nE77WZTr~+m8JJNF6dt~?9P)_)y(NV?z
zI#xcP`xsWils#Dgj=__ei&3WH{z7`<$3Pw*8u=^&o4y$xfm2*4m~Z!GcTht@)dH>M
zKD<HQs_#NCD4~(-u!H{@Gv%Jt2W`W`Z$F@f2CqxFXq5owSFbVqbRGLMda>N_UJ?XM
zY?9Tt^>-Ulko{hWgGQ^E9z{l9S0hnBhFA5H*LvFHAh|(n3V}U?lpq|;P%gRnruQ<T
z+<c!_mAhb#^@7RMnOBC$o@uC=eux)95U2@L*_Y{;yq#LV*r80E$dXo`5A={66AIK|
z1N^G(!l6!<Chnm@H>O{M%-Z!WL_Z>}d<TnsZma-nBfOK(|K?gK9TR6RX~xkuaDGrw
z{{r%QH2XDsmxxH~UsX?Mzr7^Mz6Y82^38|9@a&ivQ=Je#oN`G7!ZjDG4?y~=YpYnf
z(DOqXG6(jTsNM;wWqcK^33pa`5W~Z~I(#i$3u5lwFvlyEA72YY?p8R9m!J9!7+z`G
z3FfIY(97RlD@}uax_mn^_7Fpw_5l&rng|gfw6LHL|G@7EP{Pu*qd*zG0VoZ~*0#1l
zn)b0rn)b;Yk5v9LOA3E>!Yj43OVb)m8pFtNh($_ScZPv*YApb5(zMff7S!$htUxMi
zL%jgMqXn3Iq-o8l1UyO0w}OVo6CSAruJYykS^nP7q-n1K^A!HJnweTfrmTBgk-iH#
zUiL^uANI%~6@3i+T_%2;iQfjCL%@GeD*CJ!dKb_S8R!>EMNLMAD}Z!(EgVXvX;C~j
zn)DRCi-rRJVZ23NwpH-WWjh$e8r%(Dspxg7c7pI*f{w4%Q6Oax9DrSwcKvr+Gi{il
zwY#CA;D}m{EYNwW5m?o8!<efB&}@)4Mj^-z)3koI`IHn$#S^M2)nn*j(FC)2*MWGd
zB;#a2&nU#W1=pFd@gbi(XHK0|q_p6#6@P6~(K(?)2iA7{HRQVwFFxc~I~BD(dJ#gr
zC3+qvOte3q5J$BeZijQNbdgYRjuM+rY~M1B6Q$K~eU$D-5*b1oxQCSOaOquPOlWx8
zyYK~{f{U^GZOiJpi!&L>NBUmil3jmutCa~P^v%mYu0nl1D)x_m6Pyzp#1YT}YZWOl
zF2o99RY(8~n}Sz8M0<lUDtw)spvWW6Q3}2WQ^dOkYMZ{8jfkCQX%u?T#jYHWICt3t
zWtxd%Zbq*qoj^_TY~=p12P%)1v$PC2PFC?5jd8xY0^Ad6+d^H7sf#{iJ%C1C_QF0i
z+(Sp~*U=AD5MakgEdg8~s0PmRmwYN~=fU`0I$9gPK{kA?V5!*vJ!`7R-sV-qu-HJ)
z>Y{6a3is}Vu}L@x^?^qnhppUWGW?DT;YRdF;AhrzRe|Mb063@DeOy|(2SZ9O+38Ua
zt5HuoG=^Ad4rLmrHi+x9@51IMb2FdqZ>T5Q_j=rWJ(7QKXvz(CY>AY5oCJ*-<MIvP
zRXcEa{3=Xlc;Q!b*@4?gX!roR<fIas`^Qg3DewuwX0F`_Z9~J+O#2SCz5Pv(`!vct
z9a?+~%7l9LBQPtTo1TC1mY)a%=T3`%I>2w?kKQkSdcPFdFi^vn;cF+p=D-hTKmO+I
zLWsXEc*AtP3_VFJhHpnD#JufM2TBJQ?E+hy-z;6oM-0`xBs?s6Rdye)JlCUkss|#^
zqXwZ#vF<v1&$X*NAVxwoz(je!<|Gs=2coyZ4yGOmTn*VV;PeDXYOsoJmHxcf{w8Fi
ze;8LggB^|qKD^N!AC??m`yo$ovDS$K!_uF__7=uE;>B-e>{F(Br9ZcN?0>_Y?sP|%
zZiaa*6&^|}UjY5V#jfE=(kf=h02o|=wHc!|i>5f(axi4n!$5a=Ay5rgqnJ84H>Wm?
z4mom#9k2!YGm{oifQko$?0EJbaWW-)tH-@t^0jay>{VS*>|oe>IDC%^gPIpb4%aWj
zuz+1An%!$ZNo4i2p19+%M^_`B*ba<J^?+1!7e*D;(s>$8o6u%C)dGaLE1&?gtpc-S
zOa{3eypkGDIh4(DO^#1me;k%5Dhf_;2*LfL0OqKIOtb{1(~iJZoX7cwd{FU%#S8__
zN4!|SQ&#;MFkze&0tIG(IyF;v@4yDKUpeW50%KELK_Isu!UP!r?K#d8@u+ldidYm-
z0*zn6ZL)>=3H?Y9`KGi$eZ`B8ae@t`g2{;~PX>@f)>d*D{iGg^Icvhg3hI#xwgY(}
zT7olOTakw*9vvL{I~aTa1c}D0!O@*H)mAi%0M-{-yHM7Cz$jGpsc`Tc{xnR{sZdlp
z5gGV9`a7}7U{W>od&o?gUlhw>;pA{qUxu?9`2EUUyrx%wf=lJI4nA3{;y!G&Hux#O
z`Z28yIX*4Nn9577HZZ^T`+Ah-AM*PR2$Y5+Nd;7$C?%h<dkXcPpR)I{v`FjsA!W&R
zq0}K%nSwP=cV#M?$i_ie$b3V+st$!_<u$n{qqoed>qZn9e)WTBIVeuZYaGgt+tfiu
zxETc7)CN!R3;5X)*9+$wPc{gd{uk;!|Kt#O;yfrbLLV!#Uu0y**2_E*x*aD%w`1y5
zzUrs7`*s;_e{>VP2`_+2O;({|JS6xUs}7!Z42e~E;{G1)gL0?{B11G8JBdKLtip-y
zLpbJ<PaO-3whvqYW;XzT6Ac$Zp{KnK`qcIC3e>yn2aG%EA9yc0<Q#Iyp<r-eiCypx
z6zmYbfjjac1hiGcEdkv6tc(z&{$)A4vC^r&jtWKRVQ`rH(cpWShww_+5lCen_gA9b
zwnwu-kKK5@n4|^zCl7oA0t(WM2jd&=*TF8wf-AY_2?d@=RbNk@_i6IHj^ufr$wNUG
z6s`@OmC@h_c4On%^%uzDG;01z=U~Jb!)TL&HEcKx%*>GEJo!mEbR7&_kf1r_<m6H(
zwB&LLUXThfPQ3!0B`3?yT*}kjoxmqjkJkr_zyLZ%4d60zd-OaH%#xLlFanuxsSV3j
zFbt}ioWZW-#e1m+3~tW>l{vxF$%~&ObILQ034y6{=5g}G8a=V_=n>cbPO3e?%Q-OH
zGN?a901Po}M+`(_9TS7wN-aF%tfvJs&<Z_GW*`2Ukb`;`mh|qvKf~X>s6d4>8<Ut5
zJjKdjpcxVj2Gc6@3@)<+Sq6*!wIJ5uFjrvegZemUaguuocS&{xiaSTv)OaHNI*r5>
zgSX^#(`Ysk%k*qPH-f1=1+l7od>LHCt6le$=ufY-Ugtxem6vqeaQxbD4PNqFgO_{^
zUh+P9$)|$V>yNi#7g+QDlB_Wy->ukkCwNCb@xJC&Tgf{v1@9OV=YLosqTf!?(+9uk
zOj<D3lZ}_e`9yp?{S*9Ro8T9MS?m|gA`4^ZY@8zSsQjPd67D#c;Q08j;}5UM+1tS%
zUW-0t<*#z^gE)T}faASXL&(7|?12mMo!|;RSi)<9C0v(a39oU1N$~6Bdp}LSw<G!9
z&g3gB{NX+F2Tneo3VZ91f=LXPezlMMA**5x2TY15<bb_QCx7TRfaeJVIFg-~WaFr|
zoLu2F&P8|p3CH#ziPAqyOla?jk(jPCX<o)$&N1dP)>bh+#8l>eOPI=Mu~?bAh~XfT
zGuDT)yKC3OUb*aJ83VwJQ?O*U^uoeYjpOOXnf?;kuO~Yp_isY#I;28nA+0<JpJu=d
zu~=_{5g~#o)|?fmg{!lvx5C7qq2MS2@@_`n70Byu#^HZX7ipyqRDa1coroyf9|<$@
z!Up8YZpc%3^oSjVf}<ScL&I&J7eTJg>+X_Pc5vjDoGhR#&hxIdMCEp(34RBe#wiv^
zPEaAe3Xk~Am-uY|LOqrc{k@^UFuM}lK~I(~i7mxdU*zuhtH+hAPH`24^9hQD2mI*J
zNW2UCnCY87-NK@{X^_@;n5BE|O=?Gbi^shcn~JTW#d*Cq72jTIZYqR4X4y|n;A)xt
zRkjn2#&)8Sw?BO8c0zH#Qwg`JPOoOPO60GoLuhQQo4)Be*gas*z;7arn~Ecq=f;tv
zK1>`-bO>jV6dd_nd_&L#FOWELFL19YNT477B!rw@mig6<=(V>77eB|{Mx*rS&+Sc!
zaI+z{9vz5quhEFT#(N(7HcZ7%ulwZE1`n<e(8MthMbpH}rloF#WSLqomtdb}UVoLt
z;ekWV`5Z9#Pz9>vxn?IuBKKL+idE+WCFD53WKLN9qUSha;Jd$094MDVxHpK!ATC1Q
zX%X_h0s!tF5(x+CX9S6aa&Q>Mz;+2H0oA|ywnTM>jV?z<JTQx1N9ObOf=iKDZWwa-
zl_|Ck`~uwt`N#{RCS%N1LYIFsaHWYnmdIm?Jl2V?gpE#)LEJM;bYB&fhf0_8IFD$w
zZVtiMUB(xD$*a_Mnx6t!5)DUomyX5B=A~nc_|?-62daQK1PR{>qe=`p=NYaw!#adL
zY;l%6lx>LHVdCm$xX;Lwfn<L861ITaCDq)D`5DbIds!TC?>!dc9h-Y3cW1okWHlu>
zOyT}gd`T;wI8T^xkBGuUBlGq4^&A_<A$i_dBFa>ou&LBP=2-;fgtT)^-|{S0hH5dw
zj2i<41w32C(a7BjKT`GI@l+FjMAGj?!BOZ@c0|s_y+1w1V^F+Pj7L!b#v?1t@hC#;
zcx1_7_U3r}|8E?!r8y4i-}FCV{9Sd{Zs)*b&vD@8?#DpGs0+9eK9Z^5iGdTn1ml2i
zt<_xv*qz7p@@L>=b{rA-6Lb&8Zv&UoImV$F7%H4nkLt4qAy0f9-hzzgID8<?fm@3P
z=FqOD2#g1~0z3k@JDf*HT;I2jPJP_%mV*S|PeX9z2E4fdD`e=p?TCUR1D%w8h#aJr
zS!ls%m(|1i%%`c0HCb>Qb70WhP-A|O%cp*#KZ}eqF4^D<VDNdF5<R?L-(le*3&IYO
zHW&u>H6pud;5n?@@Pg<~KGmR)r1AZIzLHnp;I0KIp?J~P8c{B;kfN|d9Tk?AFpuOv
z$=7L%xGjOp5Oz#H*COp9IQ<?5on%*_8}7fTEIf>{bt_Ph#+yi5$=lZ{{guku`^4LX
zvk6{X2sMFs#k;R{phBxqNP7z&$m|N>Podya7vSVUJqY==3tPM{c@NI*VYbhOia1nY
zN2*H>QSF4yBmfXP!q~UNU<N$})@@cA66J6dN)lJMzbHg4uluxA%L_*0ViEfqnwUaq
zWd(UexBfJg-`{stSon;Y{;?Q5@GXk))S%cX;RG;l9SWm#SvyzO?f@<4$?D{kE!0X=
zx?V65Ti&G!q)_TEN~>Au*3<N!&wzUY^dp*E*2<9gQ79$8-6m^~K#nbw5$}-IYq7DN
z0fTiW*Y`@O-5V5lvsBG(UFab@n$@Ll%{~!4l$_lba6tVxDR>$JiVWd~WJQMrVWu=A
zV0xi}3cN`n#bQ|DwuN3%$cfC<3rub3bao)f9^Nd694Nx9bZF#OuX-f=m7ekoM~ONX
zTf}jw(^%AJTn@qxgs4-0)CabX9jH&!jNtap$~|=$V4{OO?J;$g>(`(_FozbBg6YwU
zkRLzT-aS4QzDoM7Qq2$1#PMNHwNj~G`8!THj1RbK1DA;`g5f@d6GGEJ!xUR1$_w0T
zo?j?pf0XL0kvEzxvc?Nd&@etU_%psgo>$!1vPP827cYW!P`n6@vKDBV5v+n2wc<4L
zA-)3RO8;{zX_@89ryUsnNXPW&f6Iw(a10tE2Lg@LrVZJ+kt)X^=Kc^nVu(+iAy;)k
z;n<YiJ)pS*P~YIt<_<*N0~;q>u@TZ-!9lmtm{JXIWk4ZcH)hv4jj(DxR-ktA$}P@p
zMO(_=NHGOs=*2>{E)Fms*fFJJ{q(YKEV%@H)sasQ6-Jt@4R1iri7w-aHLJ}mUL@Bd
z)Q2G}F0ippP`w|3RklYjsXL<tM}tVky;qp#;4Vg!c;G5l%ln}V)Xv-^;7XKAMFUl#
zp*$#){*^Q0(l$I}L~WM1?fM9DVSp6M!yxJ2Y%0l6&OI-h7p(Y=3!oL+&SFNa6Gbm>
z0?mAfz1ah;cAS9~z*!z!SPT^KHl8BOW8XlJimntaCbq4>-6_?6A2)_Z;Xr0^Uj0AN
zxsh*S3mSPJMapPu(XGPw;)P$M<Ktf0zefZtD3IH|f(_0W-5=Lq$lBvrx9HBhoEh0f
z!A89j6KpYVY7+@%W+B)0EM%S8Me+Qjd*wHTa-)cAn6_*i_q!J57gsRWKI%$5U5cl~
z7;gI&$1o9~00*kIntwxDycHAG-2z~*>RyNHibX1}&x9UK)ar*U9RA;^vR)anRQFM{
zkn5C&50WQ@=01o9UL)!%&I`ozPwtgJQC;{JqEjLV@wH{x=kbF!6j?VuaVp=Ye-jzE
zj7%iZ5zI)qiV0r_ig7fDU1}txFd;LM<9H(B*uNuTSR%)XM8X?P$V()Al1O+732aM@
z_%MLBzz{YoANA!(iVoH%psOPHoe*fz3-ytV&WDDcxv*OWb7|KfW_N<gOhgvO??+su
zW1T-JRsZEAd_8VNTI5+gAt7=HjO2#C&0-H(_AwcRUwxJ41TU<jP=R={$Pea9pGQA+
za9byi(8(jl;0PT}9!kk>St~3Ij{Gie&3g#|#GKl|??Q^1g0)FnxfL*RU}dk7Sn74V
zzKhG~M67ND-e+K6Cu?=w-oS{Fhd^_zz(Muzf{s+VL^nOh%eg>QAqzja-1P{rTeX~U
zdQr#|hIQ%UUnZ<exS$gF)WGJ*mV-{Ydns|^wkjDiLzTe54SQcw<kqz)5>6zt<XvSJ
zIR%`88eEv^yxk(yW0|;M)FM6Q3F!%%nRjC?lvl`HuH!|TbVgy;Wa&p<*Lgf*pN|Fr
zLde(8Mnz`V>h{I1DKNmZ39EkxbOD7UES2v<xh~EjMjT~t(FpflpX<EWKIh!YgU-=a
zj-y#>0~PmPu1s7Q_1YwIcbo>q8sJq=!e)=Rxk`N@uWPOzdkjra0Tc+@xGeWpsrqMl
zDNC5^aYF_46k}yCiv~pjtO>*ypQ}h8j;x&jEqI9sYXeeJYfdi7gm=ZsK(-2@#a{U}
zHr9;XBhm4e{WU>qGb~mSjs}4tj0c9!5NurN1Yw&)iADMW^sD*~8M(MhRBeFS7kL*v
z?4utKIE2HOJl9p^4OQCJX11qQ0oh#`cL+Z~QE<X{dF^dImJKEJC)UqUMLsCr7xMG8
zO%kz{W!t&C^uY4u-d*`idmB8_V86vIq3^#F4;ad*Qf#~XGpTkcol(#o=24O2Ul>vI
z2t^^I%eHaWS-3FwG8h#_wg4nMgnBm70Ld{SIr<<%pv+u~A-TnqU@I@9pp8r0!PDma
z90?_f1TNA>f}aWVfMW183NJ>&a3%~)<XD$TxQq$iT=3aMLJ||YnceRb36Y;6p_|#Q
zPb9P;A#!=ls(|fz-VgQ?h;%;MOT?i+!Ev&_%v3^yEL#5S*l#HiCCEF1BkQ19<qesH
z7_thAgifqZlx=K-@=SX_amaYQ+Q>R=W_afa6|?IwO@w}u>FjWAa2=;+9&DoK28%b-
z!6UAa8T=Ue^$drs6XI=eqiVh$z(GSSg-eEw4yE4611n;*qkr?cKxUgp)T>|!xuJ)C
zkUSStj7E{!9M5C`XV^~#5{;+}JUxu3&>sl=f5v{QU^>gR9Kb;O4OINB29ZbpQw9+-
zCt)@PegA{(A&}X2|D=~GWSkf@Frsx7*xQwoCaWz3hXEm88m17OWX1YqZzd<|Z4TM^
zQ<_7dhxmGYuyP-hDMl=m|FzQ~crtq)HhGvldBoyG7?D~h)edIjml;Mb1IgqiH?Faq
zH#UEdfh&f$k16RZIZc15Oe}P8T1hxb7(No5l5sL%9jV8Rx}^7nG_=&5?TH!bHRdem
z<5+EYJua2@yu|4xt$eBXe1w}r{g+NiTm~y9<csEnJO;}ooWqoocEd9H9!@9%vScVG
z<d9wy68CA~%4%9BGch4^t@`?ZW<JKme8l^@jg<e)d~7o9lkxe;9Z;rJ&3nGjHYxwR
z`!oDMw^w3*4*hYj`RR|(&$a^lQFAhS`#}opYc|3u1$-Z}_T<%^H{xCd2vR3>-X`L~
z6-eE)R19HC0dgM`bo7YSda3qYrY_sYxz<YxAlElGM+^VWn4~M8v@V`p52oUHEu9zP
zR)ISK;L$z=N|z3&RP!myGmgtC$V3quN4ZO0$B%@_I5XF9C>ffcy<kUavI;JQgIJ-F
z>9V>AG95wi^kfv5^l6LJQt;r!LpmQa2*CpV7-s(BbUt8#UYrK)&tqJt()oY_7pG<7
zA)gRF$b@9!p^y+h6cLhxhZ%(Mp%ks2hJD6_ay+E-VUajvr=Y#HDpbcf9w(wJ7=<Gs
zytbNf9PXaTlpl$6wf8h`naPdWBlz0H{r$%8;Yq&}sqy_zAMJGs`p;;uOE5^{?e&VH
zrOVn45Vg@>xbh=35;{Rt?~W9io@wTh!+{qVr2P*$bC_t<*bmQa)F2!68=YuW`>G!I
zHQ<N{?xuj|`*XM$OVxB$Q(lTY>M82qjrPaEK&jdg&Tt@P2}aZx@MFLkh@K}xBXV{V
z`T(cFAaX;>a^kHv&nA9D4;w!qm@K=uOTi)NIeiv-SW&n7)qyTQcApr|IE1Ez99tDg
zh)%!O->==C;-@zCJ6K9^M#+~QhHFv1(*D7j=L=!D4ExkKd^loTy%Dd(jKW<BjKdWK
z2L{dO5q{`khbme=&b5us$EC+M{K-yy6=veAC<o-<&96HUO&gDdv?L|82xr?C2s5U6
zwry;FuOo3%+Pyf;bSDlo-HyXdn9s21ndZ!YHj6mh_9z<22Tfo0Y3#j5>8H@Ek(axm
zU9e&H=;xvvqp1n~lQ@F^cc_mI#mX`L^!Jgj>tlX^;I=v$I$68U^Kialn3ukxAl@+y
zN7{z*JlwEM9BCT{;-?SA*OZ~Cf^R75;DB?PaUQOAn3%e1?NEj|4>yeG;fC=%+%TSp
z8-`O<(}&_~%21w%8;ZkJ-l3>Ia)Wi$4r^g|AU8vtTpLy=!rEbLMYw9%1`)0q7G{{X
zW~dRa8Y<vvwL{kmc=b@`OPq%rMmrSr07DtFR5D|nha0AgaMiF55w00V$%LgE;i{nm
zj#33)TJ=!olZ_oBJeiI2a5#}&@6X0raGYf|#u~g;{h=vaWie2d;I<4U_Nr8K$nZuJ
z2bj#GI@k*8Bhc=Y&wePvCjGo^)+HcuKFe~rDGfySJly1v6F$1|)YvVKI`M$4Qs4B@
zVE(s@<G%3Q<xwXGa4^<!fvIua=@nLr$DQ89ai`A}`<poK<ac+7<4%65paYs~%=o~y
z;%1fj?-h-78H$+N??KFhUq#p_!F(YO{f@<<-=@H=ad+Mhd1IVQ$nEc1-Fe4{sx#S7
z(Lvz7<}YY8^K@7=7X#OaYuG}A0e%lZtkGz!lU;aQ_n{~pFlxmwK(z8~6l7r0vbcl6
zJngo8B^t8V{!?5PH5}?s-1hLqldSiM{!-0kurgoC272l6#(f;QIH`#%Lcm%En#!r}
zy`WV6EINR{l(OPRWE>xI7#0OORcx+fXJY3_eTyuVK@&o$CsDm-ite~d2Cf9WtmUL>
zw1!L0iTFWDI!`dpgOYTf5YPYkpyZu!n<y9j=1WdBTRbNIF_r=!Qfc`+Xbx0fn1d|(
z9FX4kd1Wk_hOY$9OMc0DWy9y;tDjfy#y2MBcZ_-2=XquDQ`0}f4{bzr5O1RB?m*Eu
zxna6adT~#LngZR+K<q*Kp<D4wBI1+qj}!VobT+)+xl#b_5BRpU2f7vF4!qXP5oL@V
zX?=ql!37Z|?|NQ32N|$-8tbF-z{iV}<oSh|@`ihieks-~<2NO6Q~t<qcv1q6+xJJ}
z@B}Xr<>{nM`5erW{z`C1iV}OJ@>vSAS+oFR*9h0ytuSn05X04<iYh2_r=bn7Cpv;-
zaJ;3)f!_-=E~&}a;i`q>sM|-(rPn(eL#j<BG6XLOhd3%?b-0V{(U8;w)M)CmQfxd5
zjUznzv%~>6+a7|BsOSqwY9;vE1oMn-V+c?BIw0tGAvMxSerVh#^JB#0=QAw#Nf^?u
zd7aW*D8i86rUAQ;>meZ7UL<S0ECQmMhKnFnUidf(tvxvdp%Wp*+DSzSA=TpQ2Nl;r
zXp?3jL<J^gB7{ggsT3h3T3q~~;sOY5QaQpRggFQy&Q4l{5YjBJdQfrYgQixo8NZ4+
zYrXAZ`FBNgf@|Gt6>8V8shTK%Edzp;e=Yl%OR6}fi>n`Eo~or8HXMoBF)qV~gMB!1
zAGroY=B<9$-*2<+AL-J6iKgC<Eyi2@t~4RvBLw@KcaFc+PXg?%hJEiIsC8AL*tZ%E
zzz@LiB}BAJzR`s5JQz8!PnJeAHRKtNdym?2B|(SX@J{qe<64Et_cyMbcj1Vl^Zbo-
zXl3}o81ML%7HqY|Z?)ueTScEn_ynFm3rKTh@e3ID;iOR3`6UEM)A78RrHD+^8cs+>
zn7X2n@&ou$mwo3?1m1z^u2O+h$_cRMgmf<vk*%no7s=w{8FA#$+3z~Ued>E4h6Y^o
zS7S)e6l0rrouDVH-}|%qdLF8q->;d^pbgIZ_u+5)9?*m{mH0^uWaI(GB4id-ngN<-
z>^uJjWB^S?0;fohy<GVuXd)uA6<xdm$@tkFS0-|YrL4CR><4@w(EHVjG`x1`TWCHs
zoAks^h;E1vzPc25H38@RQs5w^GDXsOWqlNdHS#Au@}+Dj5A}^0Cn&PtB%tB7pi-X$
z5Psta7+KIH38EpK-X#W$9noeCm}$4C;JfGoDXRh7Qw}Z5_B7F6(XMV@C+rvXq8Z`l
z;?pLUE5TFeEqGK}cB%%=7?^{b3rm${4ZKnxX9$IURSqq{W$@z%@LB*#+3nK?zA9@E
z&ybUczo~>$lN9#}|0VnJtw>{bDM9!hXCL>e#h8$(Z?ZITAmH?QxF@+qR#AAY0sO9M
zTseurwcv>R)QROjZ5*a0uzS8t)>2>fX{m2c3=PC~uG#0_BVTfo-_n={9su!Wi(P^W
za#ROiHxaLEzyL?3WNGXkzOd0^Kf`%Kg=ZzfTybbx(P|O`L_@a#VU-<RrtFYIJ#s!t
zma6}Nh7z*FZ+puQ%U6&e@`wdg5a29OMq6cv{Qo~>2V+eY0>mL!_@BG)<iDr<FqjOb
zyAwtK7&=XH{=!cqp9_)!U5;{(uy<F|{47>fIfPr8(3j5K{l<KNzCR1cfVnn1p&vJg
zks&qU#_cBcEpfUAPFTJN+7{s?d!f&M0=HVV;ykmy62*3dbb=o10cs<tv<Ikzps5xR
zJ<~qa$5<et$6;pvP$$vdr8^0_yaxzZY<;Mo3y?p#vJf<7BF>wI;m-!6f0lj>Jq|;f
z4Cnp=Xs&Cn6q3hC$neJXco}iYBS#+yH*_4v8Zq7Fl4?I9SIfgs7rAoI{JkVNfWckO
zDcj-qcT)d*JMt}S<hDV`Ew05d#qu7p;5Nv>v4GUg#3GqN<9x8W+v0XunBKa@m9FKS
za|iYGnm)vpFMyywRSj{auv=Ww5s$AY)}J@70#V=)P?F#tinsdxrr21AF9Eu-4*%GM
z3?Rf@ho1mUtixDf#Ufm8U4#kjUV+PduE4@XiWOMrLNW&Bz>`DDf;oL}U?D*{02y(g
z8a3rBEL-3%V)ekGa{YSz`VQBK{P<ej@8x^NI^6Fy5t?gp9@Y#lu~^cuoR?zNE8Tbg
zc3@!Dn}JusAj(zm-$4LHkz4q+GzHuaErKjKz>RNQO=fw!bQjh@tbZs(%EBt!y*|UI
z4tQAENMP+LEQVN%VU@e@@?i#Tm)~O0#sc8U2HLQc79j(&d0j=Iu4v!+-!P~v1P*@a
zigLoQN!0*;O-zxp4vA_D+EF{FAkGOwqiIss4k_zZgTAXjzDF#oub+^n<0gY?r}3gL
zAzIx1Cf3@&1KbSo%K$eE@#k&`FiTJ!tE7(Ql|v&jG8?;>R71X-j0H?Kf8=66rk79q
z-UejwX+Lfj!Ogf750{t*-xDVsVV{Kljpu)r&_hlobT9rMb^3ATR^n(k5X|<(g%k6e
zAk?(L5O;qYzTm|^|1mzmO@03De7G4uLYCfP=l4$g>h@DA<eepW(~?(k;Rb$$42K)}
zBV__BNnj-bi@>cS+JQ42$eJ7nil{E|k0;90QQl`pdAP6Qn_oL6(sfb5@pyTsMN}6B
z9FLcG8d39S7d{t9Ej>~fuIQuFVI%&UG8G&>mufBoukoqx0E5--dGH)RulBOyZmWbM
z|E<WQXj1%m1b%^`gF1C-+&*4wo887=UpOC|LYjs8VILbE2!DV-^9sK|@Rtyt^VnZQ
zc5yEBn|O%tS8(o!I2+YBfHi!ReYZ#RZgv2)c%wtz1DJ1<!w!kLDaq$<TF|04L*C~P
zKbiS0^2M@`7qL&(lR|8$Hhz*5E-~eR@*^863tYA6)#`ba!T15&@m1tu+vU(yM^k}7
zq@(}`h+ksk$$JjyqDav@Z3%oU*0`o4ryP3G)yg76&x^SJBL^aQS7H2Jcy1Ae8Ss60
zMgzyZFbNKZSq;DQ3Hc>A%-_DAfjqf8;YsONKZ;)2d;bh$B@yUiXsk=W5Mq<|3?)u^
z$TL+8a7G!ni}+1FIGKf?Fu9nXPM6nXWw^w0o<{s$Z5@4jSQ-rOi)HPnq73*|@kOYu
zBu*@EM!~RdMgkXmReg2)2z8JdHi%Qnw5I20^GGyLB2WJ#&KdE(PTJES>s0URG#_vt
z895mlfiVX+5qQM}8&izK$a@uSi>M7+fj$^=$aJ0z)w;GJ?pNFNm*D}<>nSr${eV{!
zH#);)yiVYFlh+9+24ipgIQmFYw_yZ8UdK*7WO&@r5EU}KU3qvJ9*+aD)uU?)Qw#|8
z1s+Hmxc)K@I)?Tjh}~~u@hApCsd3U#q1=-jouJmI9@1aB3uDIc`c$;<v2G(gSyD|D
z4@wAU**S2Y61X4gX&7I|5Nvu}xSc7XI~|^;{zw)B%)mRTUorr%)UKe?OL()XkBM>k
zYnEAp!`#J%@&h_Hmy4mrKtmF)_d{hk9!UW|P}&NIZ!q{FGsyL!LG}(Q<P+4$-XZy7
zCE&I(kNh4wXKcr?hzy~cdI;6jL#U=60!e&GISZ6ATu2~f1|J%kBgd~Eg$XhGC{&C%
zy9!c2(%(UZem|M8{CS}z!NuQ0gt5;UYxwa><YO6uE5PJAeG^Rnp2#ne+0*2336$Qc
zjZyqmQzKt=K3`M`T%1mad}0n`L*IEXz?4rGE@mhaO`-$abPw~q6y4Rmnn0ek@DuzV
zmke`>xF|9vV<PdxntvNVgLN%TD;3v3_rnc6)Co<BwEh_K)0a5|-iWeWulOW)uhQNg
zow4vv{3;}VRXWEPn&nFAr60rACWQ@3%_&4ijP;`&NCCWy6gRwwW&D`F{Wzlbeb6bR
z;D>ir@2uPrMGGI(*D*m$g-!~0{Pp-@8Dl=e8?mhU88$M0F%#bLyY?Km?`k-FMcJ-r
zVzH$JKwGu>$FE&)$1gIUkjr-84h4#`@dI@7)67M!{JZpK$!?MN=(y?+9_LrcWntzf
zzkv(FCT79CUVD_X6MTdTMC8GWaJD=<d}TWX+OiKtb2faK={x5QS-r)lhLI27XUnH#
zhn2EsqxgS7QrU5%_;-!s-$n6o{FsPI{N;3+mH$Uo_LVjCDqlY5ILb%o{Y&Y`VEmWT
zkIl%xlzu1!|E2U}p8QMc&-)kCpZ_nWU;Y=<U-+M*-{2@<D9d_H+alw7AX$Zqs%cy)
zS>+k>zqruXu&;j#&92}5sZ{+-EE`x*JAK)Qpwq`Gn?s$OHDu;$@lgfVsO%3F{EVSr
z8|2rDad*~vNK=+|&BMa^@DR^NthqMWamGA+rDtP5<N2D0FETFd0=)mj1N@=eW6GY0
zKl?SM;d9unC)j=N<BBwTw{Q*QD(qL={n@W7dyY*A4R+_A6!H=8y~i}bwSlk0`hE*n
z2t2op(XOi=qm5jb3pr&h+^7Nd+&FTT+R>n2mfKMlK7Y<7SSC*foIdwE3onCl&wea=
zmGF1O<H0C?wawhGc(f_c;p%Io%l0?z+sCNaRm&mAjYf&~k*mh2xH%EGhTw*z<-9KF
zcHE`pb2n5}`qdNu>|M&9!+!f7i@ZpB!Ea%|c?0yGE6Tx-NO-jCUi4^#pSMS`r8enJ
z9=uB3M|$^_g=cF;_o$88pXBNgvR*{aa|{Y{2e0y|gP(_<5HGF#cPv>}d+u4*-`8nC
z0_(=zL%4U0l0ELfZ5l^;3r5$dv5&)vJ;=}4v*p82^+v`0YCv53mRz9Tg3XP?r!Gy2
zv>*e@7fTthcus<tX#9}5<UOR=w@E(;W8H;$b(Hj@ttngM>wWAt-q;K|!|=2V45dSc
zDGS7I6E`^5pk&-&`5Tg?cVcMq;1zyv14n>Tcn6gF3a_x_i-WD5J~#&?dsTQiAK|rh
z$Ao9nF8#9wlyXn{LyMnvu%A)mc75$bT-#A?pE`_nm=N;)1{L5+179pQOR&))yKx~^
zfAnNDRlf}{k04VlD)kZugzytT`69DiNJTS%-YxH7{$TRpMr=5s1RWUt+P%2{6P96|
zR1`O2A3JJ%VE*39>#T!xf|l#QW_*(9Ic5A5U#}an4@t{MqdcGcU0ipM8?irD?5~I$
zvGw?k*gE&_M*@QtxT`#o$#*~S#UlL=XcHMn{Ir`%3T_tVuOL^ipsqMHVTF=?T&Y)A
z7$3`>y1e|N1uZQu!uCM8sf10iIWiRNz^8st$FnyV8}bfXU^5EC@4*-nZLo2X3^m6A
z<$-OB#MR7?GZm&9{Ol=8_$6!0i?Ok}ZH`XBEDwy5wXv{mdCDc{*cnoq2a@&JLWGd3
z;k1(GfmGdJa>7Z)31<~2omP5X%DjU%p+&S8qA2`)g1X2oD&H(B-z+Ns%%X}=Tp=WL
z8G<<oLs7x-e`!&|H>N)~5}3!ho2Uv6fZ^8!#uB(4<1Zcb&UJ!dZ#pXPb%JDXI;!t=
zf@p6#8o=uW*^YE<E7G0VR-|WQTaga3r#taA#R=klPEhZF_d(MlxDILS(sl8UbxwxZ
zw50E2m{#o+A>z$uT!e_%n{g51t&8B6lD<M@UX}i&2-l=PC6LxQjc}Dyz|(4-0uNN8
z>6#Wv)m1<v4Pw8ou5&UZ4cyd|1`#5jZ^lK4c#attA>wU(9mueW{HxNPB3zT6De%@f
zjc}Dyz(Iz<L;1`H*M${vN?uIgha9r{qLU#hX%;0^J4J|iy%`rF;!l}z5h7kH^3|p<
z65*=!6(U@d{-nTL<21rmP5}oc0uPihA1)AWW#E1d`~?bT%N%=PE5_~Iltr*IVFMk7
z-XgL7^qgPNA@-PGaO5%Omu$zz)x3Rq<6L74qo@}s>bw-?f$fW6ZL&J0^thS#@B>gO
zPn|~<vKO4fvQ_yq@e=pIi``0mt8@G-O8fu2!D)Cm^M5xuJ@&kPY%sI_#RB&KZg5PU
zLm%xg|BKuI|H%di#!zuRYFyt9=XA~ENVX??Qj$HX(6w7OL&CoRKbDy6tOw(^Ww)K&
zYL~n_EEv7~`7&^xOvM3j*xNdBQ+drOEcQI#RH~h0hsO4l`nq|R3wqMHld&Hs<GI-S
zhdghd965@UBPRo?N(ub*;5PF#jsyxDt_my#YU>SL1hkQ^+>^wE4WSKW2hxQ$*==f-
z-9o4AQMY?=U~vKTWx6LD#myDE<v%#ORdga36wFgQ<4;1vJXI*3O%?NYp<?y~ci^WY
zw^zO*F4ZLg)}veWTftjshK6^XarGrCi$h()zR~kOJN`y?V*t)kJ?_0c>y?)yyU%V_
zq2B-1=eGu^udzDwEZ5kjifimD`w^bSyX=fdTyxodp&ioBa;IIE<xac!{_#uhvx9Q?
zg4r-(*grqnXb+sD>J9ppc0GD7t{=fu$lK4p2X>_%I!LkYz>gzB!P`&YGt1a^MCQ;x
z$K0=q{RixKPc);keA)#b?Hsk41vl^M@M(9!7w<0|Kd=Xl64oI6pw#K41vhzK;0VXF
zU8<&S-XD52M)RFNRBTsp)hEnu4d~7zl@}VA5jy><yCXcV3ya@CCG-gX$_EmM3fY9e
zBJF+KO?#iXp7aV>_~z5hcMJZ+6>KADt~2a&-S$4(=gzeE8TPrIar>NM?=$Rkz3hE$
z!k%|3Vee~0d0!azzQeTlMMQd=Fz}s<mv>l1+l0OERNUToxZB=`yWw%QD(roF+}20C
zU$?yv$*gNnW1qBSr{Ddyx;qe1LXW}9ciOP>y$vg$FB|{&!^$Uqql@G0VDv)V-o1MD
znD&uB9|J4jllDFQd3pRs*Uk938|=qn1Kdda-e%EzsrB~bu<vaY_PxyxVe<>X7Kbdf
z@4X58URzK5UT)aJzOaA)M{%{eC%$EWtG+IN;KHyC^;eq4Ix&QB*IVErX~j3d`FthO
z_zzLYDi~QV#w!&2_+gNVA=i4pI>#w~d;xdHxb)x5q$BklI*_Wmb^Ktvxa6C$m&Y_Y
z)}u(9y1n7ZmG<p>z5rs)epO$3Fj9*l2;)?O01%QwLLzj3Q4h8s492=%!X7@hiC@yh
zH|=lpD)WtS_%N?B-{ZS#XYNVmsy%R}Mn>G<t7JDLmo)L1{Qz$T3VwizxE&Yv9!W-5
z#kL2|F)!)~WJp>4&KYI1jlt$@{iovl{mp#6xS(hH_|SFFev0?N;rdUcNzQ)7Xa534
zZ!~Cm)sCxj-~;kr2lj+0YV)mdcIYvGq_hI;l?vMM*NQ*v#J7hr_}VHz#8tce-Do%b
zwVy!Iuq?Lk37+aQYq@81$aig$x(k>0U}Me8dvIHiJ#ZOjd*D2_CK?C79{bAKAA_eH
z3vPn9*LqKAs$CXq3Gp#rn(TaN@^mtg)rUUC57Kqw+MTxGRxEi-Th;C4m>2RYou0a=
zExOHP+k6yXn}#9l#?SGj51oO%$6M(81plgk>eiW7g=BrWXB@}RaFITA_pkA{RcKg-
z8e^A!@kOA<y)W=}Y}%pP*cZ4|*5J4{g}R=0{l|Ds^ipiK^_TE2f>Soz|4)1G9vEeD
z?vKAA0fJ%@En>XY)uJYfVG{x*+}tHu*wqAM5|QJjn`N_*)g+tlZep<3MhTSd5{%!f
zr9IZBt@YT}qny@SFD(X9KwCASQg2mi)w3a9sI@{Z`u;wjnR$2Ly(Egqp5OVy7qaut
zGtWHp%ri63JTvcH#@R7}aUk|}gh_4-x$>1rn}B5K(*{hH#eQ>)pq_v8Ag&xtE}uku
zaAG|ygd(oXJ?O$GBv&O4;SG;>VMR8!2jM7-2MdMFUd2-S*iBzWviQU}IfWkPk#OTc
zY%REwo6%w)ROed8=dUdVZ4Be`dyAMz6j^Z1L-z2Y4`n+6i=`(=@HP0692a*Zso6P<
z=W3(VF{B)-DaOGS2Qg{(n7IGInO{AOAtUDKSdq65V~%cX?$*RS9KKz}Fr)Sm_|?MG
zO>j;qakLN1giS?@T}DRn;Sk9PrfPBYPw_tN-Qab&a^~m|KD2>Do-2-fcJwe#u|gi-
ziCyUCbm=t@<-`ws(J!ZlI0-wdxPl6o-pb8XC}oO|Q*7rT4oavC4o)@d<^4|t-vVh4
zZ-pne0U^Wi>BD=;W_0{7R$EX{vEa=aJ?d4yM{P0eeZU+^bo?Jxn1$9@(NL)ep;*-G
z?CXv_&@>Ca%BMRUMcDpw_)uI;u_zKi)MgXerX;c;RY=hgy*qX*c;RwZ6k6g_EL!~F
zL9u+FvtGvwWw1YgYi?x2v|b>>#-lSbF`A1D2_gm98h=9<vc#{^JqU<>7jhbgHl^gj
z%g*jZ77y{=l*RW|*ce$1pL@9MeDkO{E0e)WiiG^ka&-J66ITn9xJs7L)uezjJPpwd
zdm8soU-uTLe{d{0>XNuxP%cP#3)oPUH&<{92|1l7rm*p)8v7+}s3N#PAc8xog@LsB
zQmmq5rxd$S#+tEM#BxQ?m7Uf^7jC6QG+Yyxi$4`@HKOep|M1lo&Ld&ErFYr@+-}wf
ztTxz07}izv!`)Jr1tnH_88YB~thR63b_NguiiJxPzlnVYTnd8~3^H1xUgfWy&xzN{
zH}ZR*6cm3utw*?x)4C<u*T4qp5v=IV06okU<Io;xIli1Gw46M9j<`d>jr|M}M@~c4
z82%6RMC5rrzzp6AS17*NT0#7`B4;d1V#M|`Bj8n{n|XC|<`yA6Kr^?F{I!=c1W61q
zB=Sn^tq;jH&yIqYZP`*m0TqhG8mD!rY_uFjT-4l@wwAwk8O?^uzvTDcFA~COBqUQO
z$?Up>q;Uq0B6BLpPzW&6r8G_v(m*3+UUN}+vD*-3<P?fxYA*RyVJuiFhi?d)h8)6Z
z-!G9_a0Vh~3H%tG5e$aRU|tbpWS@HbZ7dIDGEOpT+j1_mf4t(`srdF!j%OceW@=ou
zF0%gtvd@ZLj79ovT~_!7`&jV8&P&kU`fpMm?BZD4jt1R~@+hP{4uV2#Bce!oT#XNM
zWK@LhN0^jHo1mGHM`7&u<Q^FZqM6lqz$(FD$P5<iWQ^*7^|O}POR%$?QE_ksbS$o?
z#hiam{@Nyd1l%$HlB-lB6YW)7PE^_4lF5Fb*}v{lY6NOgdD**3hvxV4)6)ach`Q3F
z9FIbe(i#3Ke#|OP?wA`5n{&9y*aLu!h8<1A{>~WK*QQ~w8w2}mY1pgBz-~^%ZXE;r
z;xugc7})2fVb_m=y&w(SF$OjZ)zr}qf5=*W?=+e@vAdU{qx&NuCWM#&X+r2D!V`d)
z5FX^pXoEs45$*)Ugz(FcObF)_VE_;l!qMg=&LV=I%)=38gh@m=n2fN?j1apB2oEPS
zxx<XGg9yJ)M);vQ7k3b$FF6-Cn-SI$VGSU|9KBPyqDLJ9W9s!W@;FxoW$KQ4pAaAC
z5@tz$FPr#i)&2#94ew*C%%sPA<(ED)u}8)r#{H|qKS&Tu){4E>G|p}rgV=Ql8-dln
zZJ6hm3HCDdXuTXYCb#kxV-T&Cq=tV=bq~@z4brLoMDN{cbZN?sCdL55M(Rj$UxH$?
zrtKX{qj<*{6mOJthTnqwumjzX9CEcnx({5!-squvQ7MMGc7q9jq+CBDn3@3RgV>0k
zM{w-S2Gkyk(}eJSg-`%DHh~DVaZL!+RirRT3U46Q;Z<e?X+!R1;7+oX$+{5N>6l9?
z*;b}<S2{k)R>q=2k#da0uVl(^<1NmbOt}z|V@pNKsl9qHm+BsVveIn2R~+9Jd;dc8
zw5wCnq*W7p5g}<k=U6s)=4{GbY$nC^f+D+E(TzS(C&f9yT$p<2&zi=aOk*0;fKIsv
z62p)+KEG$)Z$j%6>o2fk$c5##-ZPbbfl=y~U$E*GfX33k>3^XOgAu$K1k!%Q%8?8f
z#?L^lw(_m8on)-BZDfGMH0^Pl2AiBY$I1re5IcPyJ`!(STJuWmA>{sH7#GTW7rY5>
zFsPg|ZOcPsW(@*xtA$_&V}@5_yOo6poP<Twxdfk8?a?_8a$4d6&PBv7`5PFy?onno
zKB7w=(3K~eJ9a{2;9JCr<b*a1XF`ZqmHZpH`h+cKCz6Q$;sVs)_dibRpIXca-zEao
zyFM~cXuDKXtydUN=HxF(=dTJ;N5U$8k_5tEO=5j!dfYSPtbemC-r(O~D(g!P#GD8}
zhQ~x&Vt-VDGZ0A8p9xp3hv2|(0QqL27sQ@I-dO0ls?hrpVibC=snEBBpsCP*L;BY}
zN;+!7CSG4t=WjmAFD9G73l^ov25Paf)tP@5ju4{*wy@P^{JuHdk+U2-I3F$C)_)N;
zXh;4Qo%!$+sAQ#sv0H1v3ND*__+{ze06(s@e0aa~egocz2QakU!ZAtgZg?Y)2>>QY
z`jH-r1e*$ukI#=`#2ed5nppIY#Wnj&F@FZHDg0ca1%`hG=(d&eybd0(#wi)L%|o@3
zzt<MO;oADqc+9(8+v)0q5ua$!q6&oe^HSUkkvu0^*#40P?Az003lvQB*3p>+ouxo5
z8=zqowH}+QVAlhN14k^*zA6hQ8^8G@u#M67%b0MhDc&oVL1BV7T+Vesn6g`QFuQZP
zD_TB{G<G8cSjcle;$z*Egl&;9tU^n`elB5Osg1lpf;)q@?FaplV~H`6kKclkN%$=s
z$woMi@JKXShz#>%run-yZ($GpL}%ra+%WbmEXfr$9S4hBxKs{jGS^`Mho$^Hegup6
zPJK$QVEY}iXk2a8`)?>j<PG6CwSO%`FyW2O4zXDf2$r#N!W%mu_QigE9*g$f*vSI^
zDB=H2_zwWb2_Ec}(FAby#UbPlQwUg1SRL7obAWiSxnu2aOtWDgjmJO?BfT)bU-4RV
z@xEAIYcAd~If!L({OE{#JRF$Ia1;opuqM*khdT=~`&6Jpvz$BZBG9St#C-19VOYu(
zT*Zi3|MhryGKeQd=O@5N3d2VMZZ8DgB9R@^HK5qNYPY39aeIk<xqt$Zd_;D2-gbUp
zP8ViMAA_}mEs8wlyn_2za2eq4{NDdZ+YrN0E9Sf`ILvo5Vwall4~<S)Z;v$gX+-`#
zHi+m?ME2}@XA&+q-MVkwr1v6ww!V`ieYw(CFlmpz-5x35fHjrXmAisluoRM!i;$Ao
zQ^4yGc)$hEss~%>JJ|E~HIl(bP8(eI80Mia<23;vp#4R2Ht0L6^MA5Gf6eiz9nR=+
z&Us%7L%h4g<0568v2{j{8$JZSMSbLUZXE4EyE0nD`BOP@i!Sm3#KyLO>OafHl}=GT
zq4v+k^?WszZ}NDYtyo+6CKepZdE2tYyscPQ$qp5(E#pfoe<bQ4reQ7;A`}U7>h?M4
z2J_S{u4TZ9yPUgyTg=^ZFW1d3cIB<Vz~V}^o2$x+iYoYyV1Fz`f<>07$v@#vTy*x?
z+_qB3gUu_r)Y4WHMN@J{MP?yQh#dddd|a5n;R*D0BZUGpwgt%ygRn`ziBieGz8hF5
z7alEAfCFnN***v4j^6M7LSz&#Xtzk{VTvuX^_l-hF%^lENO-17ygxws+@<eai-Whm
z3`AFS*+JL5so~>I#p8-BJ2*^}GEqLF65>Md{i1-x>R0GZQ9zhke*wn>ac13AK)7f2
zRZ&2gSbr5uy6nFNJuN&dQ9{njZTa6i877F(F=+bKuv+XRWFBSDw8)!~n!pfE0Qa6I
z7Z(m63Y&`y-(jFuIIZ>i5*82MF6e8$E3qUNih$qUpt0HyVpZ|2IL9SXP>C`^d7_B$
z<qL<r;3Svl@~?lC771jm53V_AQ6<Wv*sulxV!4D^42&WYV#x2k4tcZpl^@|e^(m~E
zei&tngm<~t{COe|zIm^v_=(Oud^;Wg&}rQ78`wEg>*_;CYww#s7Z-lj6hF<t+f-np
z`~b!F(7^Nfm~7`)<w@L}xs5;2L0c{O4tlT?r;-Dw!iF!{>f3CE4JTZ5a1W_~Ui%2v
zZ+A|N42esCxmcnvnAPXciX13@7sn$NzY{(g=Su6HID!AMk%c(WJW-E_Pl2t;J0!(=
zfkGUV6wc-LOsWrH8?S~N8RC7I7rOew8zxsr{#g7pZsg<>uaEinEPfw<2l00({~nyh
zCN8unShM3$=j$WK%T{qG3T@;l9=OhyH2ectM24{Qd#PlEbnSgp<~mRP5Z^i};5{|P
zPe}n^a;g$=jjOK~cW2_>Vne*=svr|TpHjYm9ZSC0?mFnKe4=xFb!7X<)%L#nEZi=F
z%pag^-wRJD{Q%>ZA_y^j9OXC3k$+EhAI<^Vn*IU`?0aTm^IxA7ARx)|3qpQD$S(-_
zLBCRdk4X7N-lK-LKs7Xp7Ay`D`yhN;4L&AreG{kb9f;wk7q&OiALAr}*ilvVzlZ}B
znxG_c9soYJq9Qx*$5e}s9svv0am@3nF#eDZx`pLb`euGa-*EtlvWth-!}Rqtv0pC|
zbnT+0lny4!A76@XP@hlcLT^5a@5MMyKZ4jUex}p0E26v?D_XODheFnz`AgoV`ZaEi
zc>pjRz!N>vp8u1lLfJieSyCfm-3qO;Cr6gx`)zc36v6L=CJ7SHyfLZ7iHaU=jN*>@
zpX>=8sc#ik>6zCiP34#G%8y`G1fO8+NkadT1MG=di;lGP@ye-JVeS1^+>KLRIb_EH
z=;HpbwT~-F%?6j^DviiK2Wn~#@Ak~UXHkxw-vaaRIWO0qjZdTb_w00JPoHMbdeRwP
zmg}6itN@lEdPh(`OLmXTa)nMBev_Hzs(AE@pF!zz^qulG6FT=tBx<PJjVNyRb}%PO
zah?*k8=%TOC$}5y!p%JRl)06R6;tjwbp>DGVa|nd)I=`|6t_At=P(Ug@1%+qT;nE8
zR5*0y2V&T@DZ0oanM4E@n;&DXoK3u7muJSsJ4RlOJrBw{|N7rzW)e4lAfzs~8EFh-
zsDv81efVY}jK$bGIYE5cod;!{zji7_&P|*s@BfZ%wt=V0kJ#x|gmJ3^&Rl@}jQcMT
z2y3~}RrzY?KJLdAa|ho5G`=ymLwMc}Ph~FGnD4C}zDUG2gSH8^2sY!%mjXv9EbxHD
zagJu>PyLDeMvf6A9s>jK;{FXBKPDJBbS@cC8F*KCU|@iO*lX~xiGBzv4tHa^_2F^j
zFuhfPp9hbHFdmqJx7f?TA}2uH1w_1#{Ek1rc=)n7_GNMGyCwE8W>0BiB##H=`BpV(
z)h#tB!VbLPZiGK_%XyvL+aXM6J1zn=7QeXq7NexAr{T!W#Wud!@Upzt`z{V${~Hi-
zNgscm;<rNl79nm67{XaRB{XIc$;u7w>o3Hd2G@@;j6;R@ATKQ9UXCk+?sP@I+7iE(
z%fFBTUu`nDS=`~;v95>F@G$26jM=xeg9@q%BswIC+j<Zlr}7n2>AQP9U`QVHx(#~k
z2m{GYaQNi+J_8j9r0U2&^Ve_T#d$Eax?${bnhl#t=G!=aTO|0Ek^44^(IL_=az)nh
zU0$rxH#z2tY$V~D$UZC^#7TrSCt=rNpZqE(R;=)45niC3hr=gxXNNIh!Dz-N1}x~t
zP?HB?QfCNs>Nr7R>@e4wk8*;3=9cXPL^z&_QzwDQBk_@o4dWS`8qa{T8qe4;p0U}G
z4fI+jD8@6k<pL_kGaxk7d7CRLB&p;O7xDn`AY=|;_)vnHsGaKu;05-baL_wYU_`P<
zUUhl@h>o2jLh4J14WmdQwvP?5ef%#J8wv%d_o3i}ZBlkU;Oc)ze$4;y8vMj-byA73
z35m94ZU|^*-DFVJafYW>Y*(MCnnGWTa!I}!R74pdi9|7Ry(>iE>f3lf0z|_R1r;w=
zf`Pvi4xjwqWzf*TTYY@&&-0<$(K|3cb{ImGMZ;I&v;tvA7_CW60vrG)YJi~Cw{Z}3
zh+_eW@7TuH#+tDcHG}E=|J{^uc|4m^FIf3sst?Td2(Lp;%tAmu%Tb<#WC8U7_rfS?
zIU?7dgb|O0V^lE!?T9Vs@-BF7?7drJE_hL7Lx{?UUCExeIoiZ2Bki!Yv5C*(^oj#n
zosr&YJ;1?XBgTF%Tor@c;7}mZ#^+(gO~z3!PvDSSaZw&oamsDi2`JU)W4l5q|I=wL
zn(Xb+6lY&GBg%yzDApUD5zK!HEie|%se<M=Nb}TBq{o3eW*Nq0@>oG|Itf}5_GxDR
zNUsC*mU^e%$@dn#FOE*R6DMw4oKd$N=EBm!^2uuKblmtR8i8{YZYVmi3%B*+ba;#F
zVaP^9H>}rW;ZWdxmMwhG!aU7^-IcrYYj;)m)s*4RXbh^JawgVpp88ZRw$q-v6Nly1
zvD!gGIEq6o<E)0QU*U@EhOKCSh{P84m7j=9t1$)rU+-5(-Z+2_!wop4;htwyiq%;!
zMxHtF4EBH4{;BV>FJ-BfnIwz9$1xH*{mWBQ*Jql>N8<3J2}$=!Ke{siCH$@Y16=%l
zjDhLL7?*C}3(M<ZC+?HpT@8jSMkeBTLYc<Li9zbe7^7B<L%P>KRa?1L9$w{&?7|gF
z2X^<h=m(y|iO;qBs<}|SI`RlQPY3$038gQgFR0{N&#!Q;C+?5F4i`B;l7HXstUZ`g
z8{3OR-q3Km?O37_c`x!x(B3ccaYOXqvi7XG<~m*%Eoi+553h@khw!>+8G`;6mqqW5
z?D^&+ynj+mWoe5d<#!^hFrb!UdI)nA7Zl*4jvr<5%Fk;rLG~lK`|Gu8X74%N_PI^g
z9N^gs?a{76IZy&dWEi2tmD@VI?DxZ5V(YAqyn}!kul^e57$~VdB~gAx$0%jTsH9&2
z=@;ykJ37xxrOjQNBTZucwHUF*Q;qySb{^8T-w%WYFOe5w<$#WqVE!)+pW~}^{M!L_
z>5oBLb@ocE7(=I!zu_J(5=c7>lT*sTL;Ee_KmrZycx+l0{YUtm+Q=b<C)#Vm-u`hp
z4gSXpvCF%2*cF|DU54?=yLAY2MeiO!CS1|m?t}x1VQh$phA?(HuSkwBkmCd#3<Oi%
zU`ota;hZN>;3Y;z{L_--V_B*(b(pInH#wsfauo2vOi9ifhrF&Jwm4+Bf;i)lK|tcW
zAb$l9ji}$x0x9;f!<@2;wZc36)A!k*?ijh7Ki`clC#>S{0yZ3irce^@mVI(DAo1@j
zN#6?!tI@Sg!I%I;K+HYjCN$i|h6`V?XUG}-F0GX4^v%`LX}`=G=7~)oK_gA()oL3#
z-3zQbp7e~1+_tWnzc~x{;ST3-T#pbtKFR2yoQuhXkz<ql2aF7v3Yt(fgycWuP}&iR
z>2-yXt&4E>0QLqP!ovKW!rd-iT+dlNB<2E&|7x$?o?r7+g!@ciMQ>LC64jBv+HqMf
zC{2lR2C}*`JT15LmXS*w_*kdo2>+u^{R!p-GAExE6G;x?fFWc7ovuMXGHj0wO6r`8
zwO78=`C!Vn78Cts`8pzxLS_XJ?-ZUBz<cDpFG6qyYvymlZGRSaBM^3Vw0u)_)?Y<#
zu}5i_11o0NpM*f}n_AEB=T3Qo9qust9$vXN5Cv2L2b(Q$^WXPrBjh^05CNvcmq&8G
zZC`VU1%pDtIo1=SorRGXt0Rv?%-8}YECG(cfrPvCfPl!)927KC&x^G8)tt<|RQ3ZO
z<8t6{SjDRzNr>rPl)@Bf#sa9G$e<(g=h*i^d8AC)6V=f^8X_VITsWK+Su4B|Q#40p
zZ|v`|OZ07ajO90`$X4E^CPv9Gg3tqB=P3C;D~8YFE;SMGay;M%3=nsziGV-F19}la
z{g+feE!F5Ps&QkJJ@T}ji*ygUIP)*cXkz3Q9rs}1bFV8Z%+lKEMj`OY5IDyBIBo&+
zLI0?U%%8S)5XWD9FTZydbS5uJ#MT%dnYD2r{5X*dL)5J}en41pXv#e9HoA^QC~sbS
z0##Osr!NOr16T9=Le4b2NB%)0*~-r|1-Rx4AvovsZP}ZD6Ka;<m79=1ERo+r%ewaL
zEo`x_t;shx)y0ah5XRrN<xItjS?TdqlEkIai0@E3?^6@kH}O6-Yf}FsP)0DiRQ52+
zKYoDQ7UTh+ChVB>3@))3W0kwypNF&G;`7187u)mH%jmoF#Pv->uIN%Hw%74CwPM!t
z#BMz15i{%W;(lIvcE&+x-*W6}!7*Rhj5jr9Gv4oropDFr+{BK&>d$jO-tV2o+d2y(
z&y9S|Y!72^42u3KXYuy%MCR+IFy{3CnT1(3Y>Em`Fz>>X@0hOR6#Mk}UcBd$$|Jk|
z{p|LG+3kn2OJIlLpoFc5I!D&LktO?`YtIp1HSoDE7HVK|S3sv0anmLBGL6fk^ZKTo
z#c_fK7pHZ>zAugB+~~sMr2&DEW8IO{&tfHr5=UjUB057Oll#%bt0o~TFuQMO^}Olp
z^EWheo^K-TC019|Z>lyl9JUxHi%t&k?D^<VQUCSr^U80`->{$S{T>5lHk#-2H$I40
zRAx4fS}$%NKIvNyoDze!<KSVxy$I#@EX-P7yZU@g(zv3xifp-}_X_@)P02Er7%%FZ
zehhbJE$W+cjMKXtCOSqilPjHByPVPTg-+O#*;D3~@|ODnP)*d=zRMP9k&_<Wks~X2
z&I9<mP!&Dtp2}CRc^!xH{0w<w))r?^`I(&xYXp;~`5RWEi%p(k!~#ZqnMCuF1@p$)
z?8RS=eS`u^D$jI%G@YHYO==`lr-3iB5TejZAXF3Ww8Zt&EtE2p(fkK^!(1-a(Ub@C
zd+k(5yE$#T@d&6_Y9h!%X(Ff+o<M{KhP6y_eZ-6BUQFnt2{<JII;et5OX(mhK027|
z_h-;Y-@~R+C>^PfDsXCf4otTsb##_eN2fs@&7`;z<s<bGW^7FQDD8enp^;p!$n!!e
zC2o^Ld#4)u#;3dAQQjw4Lt8$slV(BEpqoi8<@ffWCjF08)47IfO5_)x1eXor=c{}1
zBQ~IxU<~#M{(g@?jLklN5_-UfG0XnWXD~kpK&BBG4gxpy6EZY{DGp<PBcY!r3;o0o
zf!ny=>!0GA;I<KL98PE|gqc;AQdXxVl-1b!>6Id(pKR1m7uvB=xEi<5;sV&Hp`NTz
zPuLV<Qcu;9={EvlfU<?Q+L3c3-tZcgO+5~QUBkNJ#6{^f(GVJk&`{O$rrn>v;bJPM
zBjVq`)ASQ+cXs*x)Ki7}BlD&_kiX##s-|a=kZM4W<5iKY^*$M$d_P#6&Sqi5Ia4m&
zb_Ny2meG%nYvWoVWAmZQ8T}RW<%%41Mjsl$8%=}g_sKXudowwxDSYHK)YC~vI<uZ}
zMyKaHvz~)`%5y|sqK36$ncwlHYJR_O^1{gE2X|D9{!QDFBbG`)M=gp@?x}qGn!Wa{
ze^y1OT#o~)ar>4ldLNZlb@u!%cKuO%<+J%4F8d@+HSbKRsTweBfPP5mr#h?)!#Vvm
zJY?%*WT6Jjr=nla#el807%P)-Rntc}^XVH96V@?cn*_J(RX)~LFSes&!3eLOjxEez
zOR4hf&TFU(P?EKk2lxdLizzYt`&w)|k|h<XHt#-zgBgs)#9lf_L2aKecLcMQotRN_
zVC9Oqfhn@fx%D4NYdCZ&H!oHm443j`^Isw(ibus(?g&2N{Efr5GL_m!;BdLs=}wdp
zZ*ck;v(Y${>jPKufiOlyxHRd*m>)>I!Tn|D);GqvvfEI2)3L%U28C}vtSq(*c=7wC
z#0M6=w(qK3xm61rw07WP5f*))y9oPzm*B(6k_rMiL3cC8b}pLXuCsBrf;f$AiOm(g
z0DbqxR$~TETrt%}VZ|<LBK{KF$PpoYft!4tD{_HVot|pnvBW~M{1Q;?isqH>HtWwd
z^KZ=R{IWATb4SiC0t@G3RX!4oXGYxj^h?;-TOP%ayl7?wZ?US=DJqQI)%UkA(Qh2t
zGXWkBR=*4pB<t%>TAz5h;dKw^#hGI^A1KQPb9YQLzJHm+iy3}B!`EdDKZoHTFnnb^
z{BZf%f%b)g)DHe)-)}$nqnqWOwEG<W57ryws+nb2%4oj`O2FP&(|5Xao<+OJ*>_G0
z+=iOsH_&2uzu%g-a%0b(hDL%4`(;C4V{#mIw;D9`DKBFdE{oG4`1FLTcsX_!iIb7Q
zLEQfS+cwk$P$C9iwJeP@J)=L&j`<JXp5#Akm8Mx8fSDe`_!#gW9DA(U@aM)Kp8e-4
zSIm;)eNg(belg^4Uu6k}eeGU<t0f$;gj)P7!<J@$n?<v%SZ&dkmNolVuCy%Bsse$a
zrRel#i}AnMGRYG3!VAD8%jvB?gtq#o<FA;$Beam$)9MLY3<?&lwZz-v^Iqu-t_XO7
z%?lP-T1(ukg8t@n7FJ34BEpM8{;;oDD=C@L>R&OV)e{c;y}lVsfag{ocZbh+Wk=vj
zUrCGB74SD(7&vz>p$&*z6bh`k!g9uRi#O073R`L<h5228E+jY<26C6r8x92L``g2S
z6<aQ4sAYko<ZlqH&zE#le6^l%(BG}Xx~c+g9j(6Z#nUY_7h3QS<PVpU;KwWJ%wO3W
z@Pt7{QoCFW`LFf`R*EbYpS3jPZ)>(da0*JMi#3^*xpUpM?q=UgPiJe`?dj<7wKp3w
z6tZK~WO*&=@CDmCA+Ih$HZHKZ7s8=rWTHwm@eG5tKxfeFlSQ~-fhfu36lJYSPPEDs
z4EsWUPkVz5ojcd*X$?5Ky)B;hRlcH6QDk0Zdb-8a>R;9FYqqzx2D~1YB*82Jk&U*F
zpf40ch5dYsmT7!Gm4wBuSTrhKgP%2t^+^U~u6@ZhbL_fkS$MgY(&F4^zZPr>B^RHm
zHKiBS5>L3r7i<Wl1i@unXtcsYwV7^7{Fs~DCzl=KeNlu7b;2TK`-AFR9q0@%SWv2=
z>a@9`p!qJe^}+>K`7(cHv)0|_54*zwchJ**rRCHG7NuYtTl}GVfQpK>pwHXoUKtFu
zje&Dm^#B#O`HQ;nztfzP>Oz?o>Sk}yz9Q%jbZeb%Q8fZ9SB8Ayv;Fhko$daWeqXbD
zMd!+uzMz|(hUV)I2R+`fuepTYqT&S$M5mIZxD`&a;%@T=p`S%JnbL=bSEIA>wD~Ly
zoI6)``)<6Wi?y63G_vYRJ}32gxw+R@%&H3be~-V<Cb!SmD7jo~Z3(faK;auDozbvD
z_q#F>Z1b#WO|h*^0ZP)+HsbFhPsnFEYq~NJ5|Co8sHmtDML)aT9WFk@9ccE1J(pT9
zFFwQCy|@_cSjXgL1<0Zb%5)4PPWn<Tvcu~W=gw8ciX<@;!=IlSc)4L<E4N>_iMOWX
z#&!p9mf->!KTFHWUWLJ9irsI7mzBnQYxc}&_rfLu_U1B=AZ1nQOM0bN5^QV<`aI2&
zq=DC@@Qn2Pac~Wv0u!7gm^$Sjmn(v`Zo@CVX}?w+*TiSwZkup7D|bk_`<2`CUou=E
z;YadYR!V;Bea(wnJ>Dzhc?IiMgFi*zkh2lra5o$2|H>vno0U7L+#$oCQU0YfnZ7w)
zgRYT&|0wBGJmx>f>*+1Ax3{zVCRMk;-5)OTdeN_Qv|_fF&~z4>T2-|)?nQTtAYo$_
zYv-KP*5SwRD$s%h^HIZJSd?ujuH;M2DwE09d4eHdjW8;TV0#BGXQ2fPs%adgsC&WH
z8BDm@x1w{^IUyM0p4L#pA8v-FqE&f1SG6ENA{3*<BQ%Sp&DUn}G&j?}#SpQzn=tiw
z1I<2x1<NLMX<O&zG+toW$WekE{wa^><^t`M*_d`pjc<i7w5rp0L9p2ugyG{h0QJ5A
zK}tB(i(P@%v1v(z*N}R`yam;cV8AN|8=*^^FE<P{qX|mJ*)JOqcYC1SN8VWqbLYk@
zmAT=L0xn7~8jSuD1K?3;HDD;%7^sKYw-{C*x^Rq`jlxdyAbW>TZ;@}6zdeaFA**`d
z*E)S6n4+TpPf8}q^O@2_3K%MJq;DF^rzOkzr>qmR=jCL}i4fb-x?YF(o+-LsF&VlD
z?s^=FWI0yXiwZDd0(bCeO>2T{0bboyUEc)PI$hU?=r7atTzttHsL=K0a69H<!VzxM
zd|lrO*NPR(+IZk$;kX5^1?%lA;99G7y#wxm1Nd<FpO19lZpPls7~H+s#A(68&Ar&7
zz8vl#_Q<cNi(T1+a4nbW`a!ruU)6Oh&fOUb=z0s>y0ESfz}<V5u0IdA|LeM50A(|H
z4e;O=T#Nj|?Y|E6;0|4{>zm>BtV6nle+&7U1p43BbsOA*DAI>(y#ai}?YI%?)BO(c
z;SPQmd}6tF|0dwU?f3!m1=sdNUB46V<^f&bOZZQ9y%4L<>wXSC;5NZ^!tH_E1a|;#
zH(cv~f*-hb_ke%6O>p<a?SPw$^67zVf!hza1@8X;LVn?Dzr+?XxCL;B;P${h2zLN(
z!BN2bHPVASuvOQ4=-&o>xOI;r9k@+!_rvXgn~U=Afop-g8Eyq!>tm1`=-J>dhrek%
z(uLcx8}Z@xzXJOCp!+K5!|new=)<+V$>W0{r+wgq;UkbM>Xr63<OH_>?s~WuxC3yl
zaPNm}gF6Jb4sHx?2V4#Hupe$A+|6*Ua0lT!;SRxVg1Z-PH{AVj`{8Qupd8>9z<nNW
z4_pftqw>MRL-Ax?U9V+z7i66{d17vV*2DtBPr_ftF}i-1L=nH4@ilFl!jBz=yrE5K
zc?FB}3eU?w@~Ye(?c5XQoi%gXX(T|Je6p{us}U(q%(fNUv}~EKrj;X%v^FE%m$CgK
zXTwyxX3r~FlvlVY&$1}5Xi=Wko>#FXZzylg#0?Yr#&68IA$#385$`g@TaK?6Q_v=I
zCKqOkSDjav6ivjt3Gr+tI5C&;Dn^OdlpN2HLw1tfB>&8#tz6fiq(PL<dv#t%GA&>U
zKDe^z`PsVuF8n#oDSX)Ttkrq8H4`=%Ic6*wuMzS3=ji(HA@`i#lz38J)p-M1Nr{rj
zNAXt#U3CKDYm0=8?Rl0pIUBQY7{~m*&T#0kr{FRlYmgt7&!fQK3wRd%Iq#2>UVnOe
z<%oCxLbM%~o;|OS>1A&Oo#~*<eB?q`I#3RZA6e$eJ2T!VI71j7))Ah7Q?yvu#RfIW
zKk)5&MZ_P!L12^LyMQ+cyi=KOdU@C(R&!MXlFa8_;OqV+y8Z#>I_3Dx^(m!ZFdru&
zUJrEn3nXMt4}2}j+ni-CMDkYyd@FSRKiO`~_(s|FB+HUyMLggiyg=9MbHJZ%6#i1m
zo$-1R&(f&tw=-X@qr^)oKgRnl@Hbto>mNcF<!s6rZ%JNHc9Jj|?|sB;YSHyh@_G9S
zndFG}oSa9-n-2UtTXnsUd=6!dSDlxE&t}BaI&}TVjCUwwyd`=4<C6Gfdw3T3-B_o2
zR-&G$_ONao!<mou=vyzy5KewyM!NTRBA-mR=)_F<Osi|t5U+;(uGaO1q&r4Eo)4R#
z@M>NE8R_?Cq@N}a@_iTR_kRPQjTL=!yCdI!KzI*qg6mNaGt|E&d4tJi11!;R<ba>&
zd(gKLpdQkDoi+??i$c)N`NJrDrqpTC{t<7}T3w$-KH_qK&NsCiLkCO)N-yy5gst+A
zM0v>ezb2t0$xj&iq$i^5RpiHVQl@f2vrMnI%*UoaU4H^JGRP&ZFCp3#{GEjKtQ&Rx
zIizPi<f9z$gZ~fxLK466J}s?oFs_J48}2693W%2z&d7fnFN}9P(%X5nu0KRR()CKl
z^vM6qh*$T0UEgEkpYZnqFN948lP~@dJ{kNK0Ip055k~klz=_Xy<0;2W4s1*d;N0<?
zv-#8IXBy&#L9hEqx_*38J%zqEl{eG66?h##k^88PbP2x~@RkhlKLFm80scDR%QL_a
z0dC8M4MpORD;?6Gf_y!Hzpld+j6clpbmkXyE>7Zm>~bf)X2f&CW`@|~58>AUZcBkv
zK9`|jDgbZ6d(IH#iZ+C{UX9*R^sywu_`gN`9^lb5B>zUgek<ay2Ye~@OX10x+sR;h
zJ9!rI2JS?E0vkQ2H6@;CyL-o(`#dH~xlG2GpzvNSS71GO_RE6bxSc2B^@w;zn`b=g
zvulu^^*;1_1b~0hr&6|fm_O#D1L16D#zXkMfVX6T{{ir(4Di<hUk>;T5~<F!&|&(A
z6#c6RK>CJ0g&C@e{vY|;i+CoxYE(OheANKIV1lj(6+JqX&veN1{@>{O5tAXGlTOK0
z9%*(R<255*|2ADej`1{0rg&*Oo$;OpA9ucl{2r!$nvC>pPvefd%<<B66Ot17t46$*
zAzi=qaN{9gL%VdHx8LR5o-uzJ<#G+;746Y=9&C{_^y%WUeBJ_|`*V`p`*gtn0D6?y
zIPjly2=wB8R<aGs{E|*N@YlbbUN36^@5umf2E021ocZX;0KW<FmJIN_0B-{P1TtdM
zSC1<ERQSt^ek%NZz%%hPnPcV*=}rSY6F>7g)&{;=Pf|XOfG-Ce>L>mX-U@hK2Kv2#
zcV{P;OEvIs2VpDleWZdixnZ0muW)0wBhS*8<H#%8Fy5JGT{pp*SFvVdZJw<)uL8OT
zB_`S^Mwu9E=8E!q3v|ejQZ1w=9zFZexa@xj=rab4FUR4p3S(AnAO0NpyO|!grYgAN
z3Bpr^xAPhDeRxiX+dvO5G=;sP{eYfp;Xz~6E~V!`;Gyf~&u^9IN_rL`LK)mVdVUPg
zzr+10MYarm1_}I*o_|-KA$qpMGac?<=s6NS8HA{PKu-i7l(5!EkB&aV1@~%toKSdA
zgIVz$ht_ZgxDX+iA~p$3A_trcnG4r~OB1xa&@KL%KoU3_rU5f27(G#WDjDZ=<ylD2
zd^4II9`1n?PZYr<AS9}17iXVI@KYj8y)*$gi(O?2nm0FsiYJcyv;t^+xgNNs#Jv!>
ze@o!rs%4*W6Trvg?|1mq{u6(T_`*le-q+KkWwU$L?&c34zU1T2M^+i99e?6E8+2I)
zgADl#Bz76xOX)cmKce%%yLKNu%$h)Z2p(ZiFhtKLKb3e3fYkG3^G^Ky8^id-^Y#N0
z<-->;^9liX;paR;`NVNgCK*0_A$t`Nfp#W5XB$-FxRVTAz7Y4@MEMdtHzGUY(X+pG
zy_VfOt`H%pGTP6{eL3>OEImL^CEiYiyM~@Vc)kMnCI;V(3Re!7_L}xMepqi_Ali{2
zMX{eilC}6b8?Jb|;Xwt|-Xq!?c<zvJT%(0bIu)L!dHDGma9E?}P#Rm{X@x7EvrxsY
zc(X8svx5}pWO`mgMW#EIKl4F_6<bi_q#PL*nY%}MMCR^=XE8#=BjmOU(qY67iuD<I
zR>S3Sa$=ZYoRFKU!>=JkJba;yL&SXxe?LmY{R^Pt;R|tj$*T4?{(han-NQP%^jiQ@
zcw4|h)-Ulx{(no)Z}3wJk9cA#<Z=>X3nE$@+$`2-WKKT7!lP$z18I93UWs-a<FMSx
z$yez)4L?D+;<*W)INJB%iBr?Fhrpp9SPVakf~rS2$#4ow+khYP{v<sl6M!q8Y~aS@
zaLPS}=6qmL_;%7f5nMk3_a7uP2?An>NUay1dGLq{mxoozS&Z`sc&?JUWq$ALL59n@
zIFkHa%zz#E>%^aU<^w{07Bj?-pBWO)xNN(Z!;5Mul)JVPFQ<aOc$6&pK3ziO3wb1K
z8k;>OBAzPzLLF*N^j(BMwiNM5S@N|`0_4kY;9)1CO$1ZTM#z=$oMMFB01tWp9$f7v
z{5;4PK6>_R*CXGrj-vrU9vo<!f<8TU^!xyS484>d9&5~&*F?_&{IMf+P-1x?NmiH8
zQw9&)q>JfXh#wO0Q9R^}Ar}zsUU*onHd5nMBRh~^&_fMKYMbdf61ii@_ZjCDco^r$
z^w{8`uo{{F%kV>X+L%sAh5Vdozk(;8&Ng^R?I2125gw-V5j`iO_Aul<#+d;R<GfD~
zKN&HdX1H1_ewa=t(^;cJ4iRme@>DbAIe4BXSw5-`a7e)vc-WxOhfKSVo}c57o}Kh;
z#UE>jm_2w3FJC8=kDk3ZEBjAbUc6E@KFc5WLc|u&*YQi>tNdAqzvJ;Io*&|uOuo*L
z50P(rW=_D*IB+UfqT^lT*DjX*gABP14TWg;G0q+IAf9;cMsOUBI-D5epkp;&lX1Y{
z_bA}7=AFmjW8k?7u6XW%hXj7jpUwEW#E7G3|L{63`_?S7#Keag_cr{M;ZHn6_+`jD
z40#@Z6oGiyIvMg2L)NQIjwf4JfO|ICuff%Z!4nnZb40rWSvdyKpOBMl@Z&W&p%BTv
z;NWk{BS^>l&9@OO9=?!(pY+)>sM^FcIxa^w0{4ytE_KOFfVnT>PmF&h{=UnXFP<+X
z@P!1vkiZub_(B3-NZ<<zd?A4^B=Cg<zL3EGH3@W_U=u726-sx%ayyQf{(j}wDLjFU
zA7j*&>Mr6a$rOG`ivKF*=O~X)%ds}$8e`&T-;;i0Bs_SIM}(g!e&r-RUpYs?C#J$V
zSm(p>B%i;3tc%&HP2+6zijg7RSgwtZ#sM>**f<*<F&d=1fkQ|<9Bc9^l#ZANQLn-=
z#(<ht@@EXSxvrECCl2@+a|)b;;4@HQ6RrW@nJ@hY-_MIsIXlHyBmIwcnV*As5?@z-
zm7KO;`NJw)k@=T92bKI&6S_l{>wU`IuG~G!eN(v~D0jkCnXjqJJwv&(mAhEE%aprP
zxn0U#r`%hVdxvuGQ|@-<?osZW%KbpO6OL2qEB6fL&Q|VX<t|h1O67Jbcb#%?QSKee
zy-&H@mAgl|Zz}f#<xWuTZmM$6Q0{EyE>`X`<*rn2mvYxB_ZH>eq1^kFyIr}d<+P-#
zYObZ|q7|L(;Z93=NkvK7Ii;1I!dqG_0h$W+wqnPxrOOu#`2+2inI*GItml+V)Us>C
z!Ol>4b(y6o>{%6>QSYd6*c%+~nX{}jR{F8qpx9Ex?Lo~J#H^`VRx6_y<D*b$bz9i8
z0&X}c-4^2wE6shu4y~j;5cZYW7rD*>Y2{nh-dVDu)8E>B4)#!#(AVM#wP+>HtJ^_O
zy5XP<Q4A+}+z1Q$T0M-Q-a1;tS_!vyXeD86QPWDmw;y3*6O>lsYrz&w?3r`7G$XFz
z15vNn?d$gXI<Rt{yRak;w<j3%td=y4_ba@Ji1a*delOw$kOY28M7rE)rIoA*VY8PP
zTMvBg;nZ^atbXcE>S#khRgAO2hZ!pcF!ZJX;>XbEhW{&gp{_FGH;%KxrzUaXTIex7
zhCVVtoVYW`50A5nps*r1P52T&sqw23^Z|66k^ZJ}Hi5iJ0qTS=q2CQXs=(S(yik`L
z@eO?+GXSb1pu@Zy@eRGZ3^3|?1KsF9xE`O6(O*bky8OP10P1QZzM-GPc``<iim1As
zcsd3?Clsms4ZorP532Z_UFAdmDLaFI!uVW`Kf;apMt`$2pP_O>madV$5#PvvFT%|6
zjs9z(fUz(I&Okhj&ptgh{zkwkSEIi*`umn+0mO5HbmG5?T-+bQn>l{J>M#0Ld|SLD
zmQfU2;%fjQH>Jck?3^K$!ErCjc*d_0-|&AwCB9*wJo}!+;Ku{=lgj@90?EH2|Gn?o
z;M26dDt^0I>?J_rCpG;$0X4_3%drc*x*S4LHYxc}P5(aNnBy;>DB~~Z2Piz{svZar
zevI-n{DXKg$LF$tJ{x%R1fG6Qqnn=8^mikSRn=hAu;&KPk@1U6arv5>{wsiyMI*jp
z@5NO7RQy!`KE!28M*RKCUfKUC<G+izth+{h!`?Zl;u{5P)NdnwqZ|*Wq<?UR2M~PF
z;_?$b2EOo^{096Zz|8y`_NHc)@i$Ayh#seB#Gin0bNtPv9)Y{LR7N!TGzb{+4F6=H
z5oY9nP{kiq@wwl}{8-E$ydRDDOv#!8H{yRY1unUgaTvtrVWe$FJ(piC@rbp2rT`<p
z0ixe%dj^cIWuC;esDZ|p0lOG~M*a=`$M&5meH%AL;Cbmx;lzJZ(~k#ac*%xjD#qE;
zLbAcJ(a)L#v%!hce<s6ov`yn|W{`N#hE|ScginZ%gUpe#(Ni1!Q!+ePGy0um_$1Bf
zPm<wB#K$?w@FU~ny=3@g&Cu(~@S`+CUnj%!G(!(3!}Fn6ZDx>o&qi-x=uLBAHfool
z50l}?K)>0|AVud(nxVhUf!P@C8G0!heynEbn`HP@&Cnys@Z+EtJZ6x1&*msljWf-G
z+4x>;wC7~_37XMPli|iFIss*U&C&j<Cm`}oc1!7U_=*TVM&mXv^D|4qxgE#+I2D}R
zV9d{@3eIgV=BHi3xsApAT&LjN24;S~qu|`WVt(#WaBfdAKMyE4x1X4wKPWi2hnSyN
z6`b2T%#ZB8kI}e|!u*WSqVE`u+ZoJHp@MUpfcfEKG~#nxnfY0u;9Q1pe(DsQ+sVw&
z3I*r3FY^-yd>mFB812ZS0>1-z7Obz{Ssnp_S3Ez)OQ!l@@X6)t=I0&}@feNE(#?;-
z50{;rpREd?%gW6US7eb-WBw_q0^bFkeDW%0K89YN45dyu?`bx}jeHF#IHHT^EZ}F#
zmr>8TjNAMe^_<JJ&5t2}E^jtJh8(zj+5B(~J@aMEr&&~>!RPZSeA3V#i%%|RHb1v0
z`o=t6D*cyJ=yO$7Ci*;8I}5X1srrg*Zpi=M<89_-wYL<!;Dls27v&N^b-wUebfPS;
zo|O7FO~BLj=lOsOxuwv*Sm3AY;nDKdCGd~d)~D$6UI9<fFEg1*KK%keT@TB}AYV$;
z|Gxn~7YU{I7f%Q}>H7Z}z)yqR3NE!VqQ<?0d^n$K$)M+J0VjS>wN2oikR=?g74T!^
zYtrv`1^g(j|F1RySLkAo8R_hINx0bK35*#S&}E`iCE%E?RsF0%r)?Dc?W4f|Gz$E$
zfS(GB-#Gsx-riwg6K<SaV%V+I06zx&4=H|7{lvpPU77elU*T^~sn7L_et~8e@lR27
zIt4t@pY|yDy`#{7VHEh^1sv^2$=?|F+Aul7d>QAx80mTex2UwtPH+_XI>0mKcaxxh
zf@Ul82r_kw*o&j!zcUJ)hH)nPCyoNI5%8llt11^YU7=kMIP06`TQ<Q7_e}Dc1Vfwf
zrt59=YTT>I=idR(l<uzuJlzg_NWeiSrJnB}h0YNezh<I818~v4oHoJ7a#hi97x+hM
zo1c(y?g8R6Gzy(}6`g@@o51T+`T7q`O0nFBK9=wb5sIb2fU`c=DY>a+wTVba)bsl!
zzA<fJRq*|fNjUo(KIZ{0+V$Nw;pd({KG!RF!L2soe?t6XnJ1cbMmYd}+Bno(r60ha
zcoq{5K8^EuRJOFM6`i4}Hiapj9|CSkf=JIpfKv{}`9uW@=M_c2$7&OhDQG`@jz&c~
z8u*5t549#9o(Dns*!J6mf1DEBMFK8y%)gv}W4Z;p#5d#_83q5}0k@b5sP|t3UXVdQ
zKO*QrJ}K?}7))kllJi0VhdxiKw^ssAIS<U1eDW*=K6fa%aSoCp|2>M1C8hox3BhJc
zw-|8Jw}fqiF86}-;W~cGb0CHPvmvNc(T@)(eX?AlYnKB~{Eig<zb)`lzDlne_57!R
zGu@7jHbG&(qLT$CNvGgF2{-D$6>#EPmr3{w3jY>`Z+lR}Z&wxc357rSBME;}5jy2d
zl8$k1*FO~ee85>BjPtanDEeN&K`QN81NhP7j!zqB-7pINJ;aCHw3!}3t3ai@eH8pZ
zjROCWbdc`-Y97X5Y$6z@-1fd<6Ckl#6>l>Yyh+K^ry^7-_|Q&?KSRNn0Zus&eOtow
z6`wl+r+loc919ixh@zuCF7dBY@DnjVOMYyxOSsreh%h_gq|>kH9H;OvQ}_dlAC;{3
zb->Bbz^^17qn`g%;A>jbaT2b`Xpbs<<Ge5f|B!<3T`Tbyssy^GO8T}s376wTY$pJm
z{MX$l;i}nbuL-!3TEaaS8Oc;n8UYvlJ0zW_RFo?eyhq6m>Owqg6nyZzHsP;Q?eRAX
z-v1K`?@;{bpdd)U;9?0ETPF~+4R9fUCC_?64I6|27xGtnuR_7^FzCE1>C8~{w*k(2
z+pp|)L*Kpzc&7Xw5^$7zirfm1r=CpMQ7Tz&8sM4eoC`ShjHTQoP?xI+;Zg8!0-SQt
zE|mPAs`%fc;DhH%_`8bENob0h(k%hpqHxSk)hPIv0#11jrO5MIfe-z*&nC!lZ!4ew
zRPfCy_53No$$!sxYywZ5wScz{OxcKXRPEQO&kq7ld~3bL=UFIx&c+8l;t%ez2|xE5
z^9cY>{K22ugx@$r;10kk=bjWfzpUu&SLMjP=X}0&vf}f<Y{D=0{Nt?_aF*|0HQp3~
z_^DU;7Ozc!ct#1IQ!wF5ehRL$3I7GEylw%U{Oncy^oUT{SAa8LdzBs#d-f6bwxYB5
zn>K+d&IrMq2ZnT}^6dhg>1scf_~Hx(;QxIT{F5ybzhAXeW1Mv<;F;)OrSSL9w`qLD
zSwnE=oGRPVfNDp^MDudMGts#gaPryVwvmi>yORIwicXWFV-cY!m(w!SnGQJVY_{11
zrr5}Zx5q}o|BJ#m&Uu`o=%4ZxNvG)sNrz{M@VN_cQ7*69=+(qtbGW+!r(Kjf5BY+?
z*ECz9jp*^`Lq*>>XVQ?#+-Wjh>oqok`{QgGWxs;&y;Z`EGd<2ZUGg)S(!Q1mxQuK5
zy+q-s+Ko2=F66&h($5+vV?8?xowr7TA904{r|z$cj!JxvfGdy5ag73BF$(<JQQ$WU
zc)A{V4shnnIM)-yBJoT-Q}WrQ+Ix$FmkPM@n4Cp`lg{3#O+c?#@)=Te22PN0o)yOD
zIRVcw|1XZzR0l<hQx>bLk``20H7rp{g5Vg=R7iO`Xihq)A{|tj1}aO3q_a?#&O=#-
zh_ljZrL#3Noz~3s{LD;eJ{_q%9hA;Oh2~yTbHO5ejoTgST;X=PQ%k^oNkh1%!tM44
z#BvxBY@B(TD*4%IIjqb`-iYtML>>@ZQ#xDFHSRyak*E$W5jZC)kS8BIKIvr9F*0R3
zf0V0hjDtPLIO;PKF>}=TnV{ZaI26Wd-6dYl?LM!n-d*ErXmq<ZoK}isHc_pE?r@vi
z+lnJZLpaDi4diw=2i&V#11mhO?q=MI5ORAuyLn=~I4-lfWOn78GEmNd<mUP$xwgoS
z^+Lhbnp~gM+}YN)8c|F>abj%}s@uJ|-d^i)JC;_1S;@J2S9Rlp#fuvpjqXPKq8bM=
zSb%X7&B=feZy3<r)&I8CUhArg7h`EfRiI<Fy93L{e7FHbTwUQ>+Tfm}PJ_mKsapc%
zwFrE#jNez`sPWHq7*AI{BdT+tU5##sqrvX>^U=6=%B@&XwX1Y?{AASzc^++bnWGDg
zHH6EoZa2=_4P51J^|h}Gx43=5U?5m`zSm}3QsY`wwS2jIW=VNG52GB^TsmusD6@vn
zwq@d85=ULLpOG5Ev*qm`E?Li0^J+!zRc0H)b7VPpyH=X=+Td|D%t8K2=cHW^;%L)4
zf>;%|Qt|38D=90P^+~+WS%k%CeqT@s01TE^U?ra43z31=D*XC9Z7wh1r94*Hjnn4+
z;gHMg03%+9qpH+Ve||#<!86^iK*)Uoj*f4`(Z6ZpY6!H39R6f5jt|Dk_Nm-9D%qqR
z((Jw@9+Y(6wcCx0WxQ^0s8dvWZ3WJh=Arlo6NVg&O3>Kw38zpap=HR?EVCeR=KknX
zNX{1PQR!@>9<A^+yS<)JxGNc{BM|a;yCDH_@ky!-3<l%%w92HI%F2>Y07jTl2aftg
z_41>5{0{2OL~b(@1B!c9TcBMw5w`a8y4>|j6IHc(LLq3Tj?Qo>mBq$J<nplUO0odl
zmo&H5r3l5>?N6-^ZgLO|yH|Srt*I<5J-;D5%ggHuP!&-DYu#7*n|<v<899<kmz0)%
zQbQ;$Yv5f;zGmL5fx7Ci_d@~2W1&sUW41l5t-jX!rAx|J`deF56js@%Q&^B(WjoL1
zhgL@~5cIF;3}ex(s~P$%U6s{0S_K=>PGxRh0s&p?N42>cS3!^mce|@*g%v7cmJw=z
z>k<_rP2ZH0)tR44-`JsV+8V0x+Xa<xb=RNY2*IrK;U=dxTxQhlS)EccVr{I^U4MEV
zxe_|k?d|rsgFc~yZ8qpztn77{mds>1G=wYs&@o&oT<(GppnQ@`IbK+E)GZPz#d8?_
z$@z_LQ*&Ahom)CH=_U*+z=lPTNj%Z?Zl!dlaYVT|vz#hc>MEg%+-)9zdy3?Y9;#Aa
z&ZBxq+!BHo{@F6qFx!p3v2-?$nFkH(3V9uo3jyptyXwzR?K@2wM`?@Lsn;{P8WuA)
zI@o&F1!_!L!%SBjd_n~%E3Gq5TIXG4Jj<Ld&*3*EpHO3xdTAvaVO3l61s!ZKj&8J`
zkfW~7i<ZiMC|(u<Pbz{se;p9wA*0+qP&&taXPM9h%$@4B%gWL0c#{T5hLhWDo5S1c
zqTzs+?8S9w?iIMf>PpyLI7Z#afY~sk#T_lsPB18d6I^Wsx7D<jS2$|gX5p{2%z@vT
z6_tE1Wzn?NqY|QYtnN~4hX>2c-PCo|*lZ*otV2&%mtqw@5`kKPb=ii+^<iH;?<rks
z*-Ueut8u(L(J5CW-^uN*I=#KQ8hJ^XVYxQeR?(Dg6f4~m-SsFMz|?Y*?_;Pc%g~iI
zl)Ld?bGt4;SJmuqcXwjHKs;xoB$1I)!a#6iH^2%QCaI0k=rc`~d1cU-YELllnRD*W
zV04bcVQFP@XD*K4&yfAPE`P%;n6%7sL#P~GbVsRGn1JX%0;_3-(I8Vf`E+J#W!y@2
z)L&r|CEHlq<s2DBX%vc~wvx+ntTrX?K^aXZ<%@y=PcyBiivz*dW``fS7a8}@#`P0{
zpf5%5o2=XnN_ckaMQ16B^wW0Qb4F97w8t|lQmV&97d*NyT|xr&hAJJ?_$w_>t}hN&
zj8a)Kpo38-pd7kAEBsxh7$RV33HwQ?4QQ^F0o0G=E_JM8Fjo(@G(*LwXc5tj{}l?*
z+bYK-WAYfUDe+1o&`Zmz#9eA`UMHApr5%PDklHnhVP(9j#BD<3>J@fH<|0@97h0~U
zzTt|H4@O(a_qB>uzII;_y4kg~5(aoX?)vKPW`QZQ5Z7KJG!S&vU6H6jSD=Ys!Kx+<
zv4#pkRAe`<sAdOT?ajXK3<`t=s0KdC*9AeLpyw2WO%GaD`dHTuwQfuWQ2D^@0@5f7
zry*SK#dR^sdI*=s;EpnMEUAjNAzT^nrBY2fcXKG<#x0V~*fnLC66vkGl9LE^oI?=Z
zH5FuHI7mx59(N8p4BW}nY^qeV#ARM7>fkVH$EUIpX+2Z;YFImp{gLPl%gV&;RtU2e
z;;tPfgCHt+B_M3Z?r<HaE0j%?Hk&bvqeYHjM0XJ{^<?FoNU&7Y9{1Ne{l1XbhZ@-I
z3z>wJ(%ehS%|?R{%30YX9i2g+yUQO8cX}{V%b;db36+@%ITF>}Wb`F~GOuIIm(Wbg
zI6EZlr`hQStT2BwPoo&3jW?^A<+7`!q1qTovRiAYU<l99=FmN&naoVI(^An|)x}S>
znADWBPMj?W$f*{nX^wX@%{4HA(wTBnLSbX73%!a1rhFaKF-FL;`!A`~Fu%|i=t?zQ
zRK<zsPR)K{d}Yi|sUk1}q)JU205?FHvGG){^mqH3>7!<X(aO4s!J47$s>(i5%lXsN
zrj~1To0Z(>qtjD0dVNEsTTT|nTU427DX?pC9hTAh<XD2mWi;ft_(V%e<KNRjo6J0c
zV4Ae5;3Ok1YYWM^{&L=RoMP0(*^#}Bsx+MAWUX;yH)vv7vmr2RWou`sC7ncyO~Coh
zkpC0S0@FyFbs$-ri^5H*DPy;z*|IbKtco^MkwLY!238pr-k9QZtBcaySNc}Fy^ydR
z{51H3A=p=`*)R;6PgKe2-3}&;yeJ7@(A4$B3#?3(G-hf>RjF3a+sJDUIYgXgj4r`(
z+LX}0!n9&DO`2flNE4bQk%i1R5SEp4h|ZyUk`BzE`r?#I%lz#yb%5X8=?#C@Dqv!^
z8SN^yMl@6>SWho38Z3yBXs|HLF?E<yg)zlLrVn$(j$u(7qNO&oQn$OS?UUP?P^_}U
zk56j2+QaIjL48BWG<-01i!xf~OBBbl0C|VF(YK|G3Dq{`MrdA5m{MZ@HwR0f_k16!
zOXdkbPJ(hMn(~1F-$+LJKrqUPJV~ch%xcO-l<enfmX4Yg&RzO7f2gCg-D}wEh8|@z
zFq+4|LjQ2?bSBatwcBV2RAO>N*+MWY>Kg)$?j-#U>xLg%U^!v7x!J`zzz_|FGXBR5
zwbjj*9K`qmH)DIeywcHajH+-J!zyL9qDf%3EbMK`Y_g`=)}>|U`|bbN8iz>_e9|<6
zoR2jo5yV&`{zXh(9ZjnPwj_p|@>?grQDMd~;7#x8rRir)e5y^!@?ngY(v-IRL`#jN
zX0%|PG`M}Z?Q}IJ);eI`1cIJbzLX|}F_1vX@aa%yx`M`31Iqfdv8p~hi;O|rnR*`c
zcUWQQ@>57^u1qBr*Vtmdb_Fli9HXv{VugItB&{r<G~4`t@^NZxdxSYp%e2{cNbgx@
zi9wPYkov@FCYbc-?&fFzPypw@_)+3r){ISX6tQ41(Bku6=?g9jVA4JF=uvi)s#jB=
zQvOFDh7R{z3WjDT&gzynX8efcq#}-gNI851#C1g%6r>z=$=V>tFVKtPE1?)<wqMnj
zq(0f|q<k`4di=1T=mX+4Id$lW(h(~i<iME5p_+f<q@?^P6YogkL@sK-3Mv~@*rCog
zj!m&#Yi5a6=z;KTF=6M+@Tp;Je$A7WE49$qf*HEBz)XC|cBC)mfc6Pp<>^3%jV}Re
z{v5++Z-?pQW@%#OM|vfPCHU#BOeOk^GwqpXSd7ezkwx58tn;jH#YDOJ0~iKHs+`Sz
Y`DmY;|7D-9nn(F`#ZR7^7)xvaAM|vw`Tzg`

literal 0
HcmV?d00001


From 17e4e478cd539d9efef7a2e3544da601dc71e8ac Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Mon, 18 May 2015 17:15:14 +0900
Subject: [PATCH 194/429] Minor modification to the configure.ac

Enables silent rules (use make V=1 to override)
Prints a summary after configure is completed
---
 Makefile.in                      |  17 ++-
 benchmarks/Grid_comms            | Bin 45225 -> 0 bytes
 benchmarks/Grid_memory_bandwidth | Bin 45026 -> 0 bytes
 benchmarks/Grid_su3              | Bin 44845 -> 0 bytes
 benchmarks/Grid_wilson           | Bin 121651 -> 0 bytes
 config.guess                     |   1 +
 config.sub                       |   1 +
 configure                        | 208 ++++++++++++++++++++++++++++++-
 configure.ac                     |  40 ++++++
 lib/algorithms/approx/bigfloat.h |   2 +-
 tests/Grid_cshift                | Bin 60539 -> 0 bytes
 tests/Grid_gamma                 | Bin 72670 -> 0 bytes
 tests/Grid_main                  | Bin 143261 -> 0 bytes
 tests/Grid_nersc_io              | Bin 97496 -> 0 bytes
 tests/Grid_rng                   | Bin 60784 -> 0 bytes
 tests/Grid_simd                  | Bin 98268 -> 0 bytes
 tests/Grid_stencil               | Bin 97307 -> 0 bytes
 17 files changed, 265 insertions(+), 4 deletions(-)
 delete mode 100755 benchmarks/Grid_comms
 delete mode 100755 benchmarks/Grid_memory_bandwidth
 delete mode 100755 benchmarks/Grid_su3
 delete mode 100755 benchmarks/Grid_wilson
 create mode 120000 config.guess
 create mode 120000 config.sub
 delete mode 100755 tests/Grid_cshift
 delete mode 100755 tests/Grid_gamma
 delete mode 100755 tests/Grid_main
 delete mode 100755 tests/Grid_nersc_io
 delete mode 100755 tests/Grid_rng
 delete mode 100755 tests/Grid_simd
 delete mode 100755 tests/Grid_stencil

diff --git a/Makefile.in b/Makefile.in
index c905f84b..b6894ef6 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -75,11 +75,14 @@ POST_INSTALL = :
 NORMAL_UNINSTALL = :
 PRE_UNINSTALL = :
 POST_UNINSTALL = :
+build_triplet = @build@
+host_triplet = @host@
+target_triplet = @target@
 subdir = .
 DIST_COMMON = INSTALL NEWS README AUTHORS ChangeLog \
 	$(srcdir)/Makefile.in $(srcdir)/Makefile.am \
 	$(top_srcdir)/configure $(am__configure_deps) COPYING TODO \
-	compile depcomp install-sh missing
+	compile config.guess config.sub depcomp install-sh missing
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/configure.ac
 am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
@@ -246,14 +249,22 @@ am__quote = @am__quote@
 am__tar = @am__tar@
 am__untar = @am__untar@
 bindir = @bindir@
+build = @build@
 build_alias = @build_alias@
+build_cpu = @build_cpu@
+build_os = @build_os@
+build_vendor = @build_vendor@
 builddir = @builddir@
 datadir = @datadir@
 datarootdir = @datarootdir@
 docdir = @docdir@
 dvidir = @dvidir@
 exec_prefix = @exec_prefix@
+host = @host@
 host_alias = @host_alias@
+host_cpu = @host_cpu@
+host_os = @host_os@
+host_vendor = @host_vendor@
 htmldir = @htmldir@
 includedir = @includedir@
 infodir = @infodir@
@@ -273,7 +284,11 @@ sbindir = @sbindir@
 sharedstatedir = @sharedstatedir@
 srcdir = @srcdir@
 sysconfdir = @sysconfdir@
+target = @target@
 target_alias = @target_alias@
+target_cpu = @target_cpu@
+target_os = @target_os@
+target_vendor = @target_vendor@
 top_build_prefix = @top_build_prefix@
 top_builddir = @top_builddir@
 top_srcdir = @top_srcdir@
diff --git a/benchmarks/Grid_comms b/benchmarks/Grid_comms
deleted file mode 100755
index 9013b3da11e6d41d2bd1fdf5976cc5dc80387c21..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 45225
zcmeHwdwf*Y_3z0GqGBdc#Hg)|HrfOwB;jF#sFTOY8J$2%qNP5VOp*yq$;-?H!3PE>
zX_*ew(*CHmz5Lqr<<F`u)LM&ZArT>{zbe>jsalCv>lx!Cz93p*?)SUTK9e((8LGeg
zx%d9l4Rh96Yp=cb+H0@<Jo}v7W_Mnmo|dK<Azist5m=s0ko2pflm!e9MX@Pm%1OA+
zR6egv0yzsmgU=@K3^*o58!%gn$pFm6uPj|Bl%?CmjR8NelaYfS3=IjH-*~XeTOF49
zb<}{0Vig3YqptgJ6)0eXxG~@XDugg*(V>x!>M`AzOgARe8Bme>8_*~p%f?^5jIU3(
z0V8fBP^!gs9Fmuz_BDSq@3*%I6woGa47f!a!hlA)FC!iG{InrgGA@((rLxO-{pbl~
z_R5@AZ{vb_b6T5cw|d*V*UnyBwqW*xc}2dCqEb<A$|vvQC0B`rh8^>OQRnIS(d3M;
zdbi~3vz~szf7ymxep3A1(eoO6UcP{F#rPTZ7Ifz}xCg7#NGru}K7RW*oczpxHvM7m
zx7won-_E`4N8zv9-hJuX?^+A&&pM~yIi*Tjm37yTKHs@xon!TbU{4ydpv<%+F<(zX
ze-8>J<3A4yC!^n$Le7&Z=vgV`)1KBOdd^Hi_oa}3aSHnPQqb3?pl?UM$;$gc3jWn8
z%Jsz*{O_cohf~n!rm(}`Qt&TL!M{F5xele!e{Tx<{VC{!DfC&Fg8vIC=+CB*b1;Se
zM^n((rtn*D3j4gCLZ38r%E|n}l|ugaQ_x>Zq2~`%l(!ijOR{|Dq>z723OPSXQLa!5
zIloGg@2^tGzZCrEC}%1?w>O}Kin3f>73D?9Ctc}n5EO&{h~)3zpwoLqEc!jk-;lt+
z2K-YY$JVFwTP6R0Nq(!$w^q`Fvb;T~*(5~=Z%aADl0O%G4F8Fru=5NX(O4LhdJaf^
zdSuL<pp*YV0>6TB!{={qy5Iti6$=!XYh_zUyUXWyclli|#dXCR*D}vaug~x4s%mxn
zd>)_Ta@5zjnmt{TTwhbw+R^T*cQ>|r^tj|Pu1^1|E|0s}wc4}J)r31g*sobq=bMKd
zI!lUO6zcPLb*!^5yRy#hXtG0@rnPQYXLpy!wZ_}!?{>E`#g%pb`Az<HogT21%tf*r
zz0IC>m$%*Px33W*z4JOfp49@W^Os!Yav|rA8(oda+}+yR(WG;A%x`J!_N_A57MR(Z
z+&(|KkR0?5sO53DAq$_U%kO9cwWMj4yUXS8a(n$g=wV+5+T2FB&)Z~hJL+Ba#V&i@
z0$1&ob#q-%U6iKFBa{)c%VE_vZ#yiZYveEUcKD#C&oj?~(o~no^7t1(X?Lq9M(|aY
z*lVwBirI-q@m<u^(bnni@-Rj>!vgQh_6`uFRl0okWnK~EFDcV?MZvu7EA`?fE0?GT
zJ90x-J{HmCfQCyuJ?(8B&8)xrtCyhaP_nh|M(>&u8oP;p;$j8CU@aX+H8#Ru+Fea<
zf72?VgI@2Y#jZM+r>)b!&ei4Vgx_^^xmS9aX64$ou1-&vucO`F>P7ULVwuc$qq`IK
zh9e?(=rrG4l*aCs)a5{mI{&<;j&A?Nw5Z<{X`u|<w4>E$r}5fvY!iv&id|aba;<6e
zCMh~omcvJ->on6-I_lA|5HVie)upnII_7qE`|YSc;VYi@=9t0i>rvFtRBK7pL{0gU
zYFXz;MaX8->S<r;U*+<2b#-(Zt~9Qy@Ju)aeX6;&b`3j_Hofn_-O8pWmycM9;<~E+
zMlTAm%;W2B^N_7cXF~~5cP@4&=wn*k-d3d(Ug&R;rABjD>-8%>?@HkpK965%^R!9O
z)VWSs>GAu$ZJv&nX7@U>xtsjnj&`Nd-Lx7FvdKgCwvIKPaf-{guB{O@h#aVg?B3ek
z-gd?7>+EiC@+*t0s$6r6ij~DqM`e|(w5ViW?6$OMzJ9-aIauZumB(T8m8#{-D=HnX
zlA^hB5hW#N6n$?xX4vWCO2|N%fq$9!WyyPe9-fX>j?SB|Ocs2^bQzN+FH>X-u5_&b
zvILbh{U7hLFkeqooEQU*`D*R?>rPRS>Lj^nO9Piz#ur2WmhaqwWy?v*RKzn)_nF?4
zsL|(R{FJH30Vu)S3Zd3yr9krc34D^WK;nPewi<U?O0~q@W8VjVrm|Gx{=)BKiI%Rc
z0M0OWTZ4dHI~i*?5{z+|Y^5Z~V2q=ra}8nyV_YYl<uZaXj*@<nK_rypCFvz5x-qVk
z&N;0SjPaH9a)StHjQ6C^Gtu*U8U!{OS2=_W+-johJ|W0Tfr*YOLo5`V=%}_>C^ONS
zw-Ib6x_LdSn&{NY2(>1<+5XE-^pg!{zy=c?jui{5O!O%+s-kq7=%<?KYfbbV6TQbo
zk8K4Yexr#_KQ+Q;6Md>d1iaHkH_uP|P4v&1`0qB+7ntbVP4v@D^c^Pp=_dMsiO#jQ
z5%!ztXBb4l1136Vp|NntL_aG=Rg_^9-PoI9v}U50nfOOd^e>p`V<!69Cc3hrL16R|
zu5FEwZK7KZB4DnGevXNrXQF@6M7Nsg`6haSiGH4mUTmVDZ=#o(=wC9?Z6<nwiLRRH
zg(iBfiN4T8Uv8q$GSM4M^b1V%RVMm{CVHodKHEfJYoZsK=shO-91~p)yqT?rvi7dF
zDr(<=KTG=$M5=+uviF#na2uU}KUkvYJ%FoarVV$*SCJ_)97Q<q9^&khA_oNiQ{w5w
z2L%2@;u*xZ3;esp+4V&F1^!Lq>~bQT1%5kmb~TY6f!{=&?i1-0co%VYEs+L+e~mcC
z7$s6G@J8Y%6SoQc8sh9)BE<q<Mx0$r#47MBh_fq+<O;l!IJ=ODBJhR8*>yxlkArYt
zDRFigkzs+)BF?TNazNnc5@#0?84&oH#Mw1OwhMe3ac-AI`UQS6adrif%>qv+K7)9V
zz(4*PaCQliPJzEioLxbrLEvu?XBQBu75FQ}+4Lhefj>{2O+HdA@Mnm#sYk2=e}Xuh
zawJ#aj}m7Sjwk}(Nt{hLGJ1^Vzn{33_^`n5A<m{6IUw+#5<i#tfWUu9oJ}^eUEtp(
z&ZZja7x*`cvx!DF3;cHCY?_fCf!{=&O)}Cc@GjzPijf9^e~mbsV5C;yjl^dWw+Z|j
z;ujDv7WgvaY*G=cz^@?ArWDBqUc0-wuR+nWQ1sngOYb7Ob2D1ea$%p^-92H%zXn!4
zo~Ve|-UZF_l>y7ttY|k{)lgRE8Y{#q%5Lj*5VSGIxIf1Dw$AuIz&K;S!MHr~BXG79
zS!VW%XtuxF-8-1SCzd}^sUPiA@|GibaD^ItMGd?<T3hcJ8pw6qpv+)@1HsT}o=wp%
z0^<iQmYLOxvg5}n+Ewc8I|0<dSdJQe?Z))$_v~PX>=-_{emA3{!<Lynr0pPP@DGk<
zfL{$hst&$!nVNP$Jvi2l#5>8T_C15dYH**P{uP<N=aPq7t(X>0y$UI{^>C=`_hsD$
zQri1_SUxpyNr49p{<C8$OuuUP=&jH$_*(S9^^u#A#4=Nf_;h$x#aAk>uDGhAzE0in
z=8jBEC?Cn^o~0UmApdT{u*AbUuC2aICvHaMF8q|8?7OsTc!0JIJ)|SJ&k<Um?YQ95
zk{8tAF?H}*dW+?qffmbcz4@IKIMkcp!*mG#)6}4*4r-TL`ffrC(p<2s8r-7>x9Y5b
z+;3IGnfIzeJ((l4z!ADSTP=J~4L(cZ{%kdPSo3^>l#-1}0vD%MSbn@OaC0W%V}YBq
z35NnVPf4>`e!O=#r=h!QQL{Unp@%ku$wOI((Js}HO;ba69tbp`3SO}E{sv9G!t&rg
z{Qf&_5Z9r!{Wi;kZ>8-Gw&oZ2++3#kv%XnXHF5^kc<E!x$$*UFT_-~@Xwd_fU85pD
zZ98)b4rzz}3Fj9Y{Eiy1ZE0H!TJUWxn^J;%cD>BlQ*<$!4-C3!sCQr9D9L?qS$fZc
zem%DeE$mezXR5(>wB<qzJBwnd@s3b~DqbL!s53G&nZY9R$U3xgq8WCf2TUmJyB`@E
z8M&AXyN9w+*bykB7xpa{hXa-NUoqPIq@|PgA2HgI7~gF%+Bo0p810i7-&bO^alWb;
zExi$WaKO{%#AxGuXT@lzkdFhQHW@Uy!-1BGb{N3OkYI<%*oo}$FlE6Ge}}ay{y@jK
zmmG1TRb^)=isdVhJHtQE$NCOVeqVmAaQ3bFY+KHtKVPkekJVQPM=QchqN-)vel@sX
zJ*asHn4rS@+yI(h+M^ZW%a0>&@a0K?mwUq1Q*CG}8<9_SP_t|ssIY9?8?4T6ZK<%d
z?`^Sk4k&1f)%l%hhFieI#%n3uq-PCU12sh0%@JCA3QD6vJJA_XH%IvLTsA~9wOBeJ
z<^2lFZIeh^E>wP94Zf`gpI5`1L<82|hP6V-+_Fv2O>NR*oxw1wot?obs~>^Rd>rcZ
z4M1Ek2$qdEpf*GgF`T7_7umGS=y7`gAX1}$&?qHb-KMC6U`jhgT_`cMN!Y-$ZIdXb
z+Vrj(JgNp>&rti0PzAq5?Sri6)nLTZOGBuEXqt}uUARw|_p{Im_`mSJ<+iF?UTnM+
z^J3*C7nck*T!X%+u7N!^IV;IoNzO`gR+6)loat)xaW(MhXJ8B*&$Qfr5{!buh9-fU
zpDL5!b828;gMSt>y?r}U5%x>CS;8J2R(4Xo2B8{~P~+RV$xTS_AY^AlI4t1-2?qe<
z5_ZZ24HDK$Xp^v5LMtFOWa8j^mcH{b7>(p}{E~GqhKocGDS4q}fc48d-_5+WqNDWB
z$9A!GJA<#Q;Vt=0q^-exixnrvhBH2lMr+Wzm?E@QxK`f_F*h6lKi%*EAcOx_^!gh!
z&`&voKhN(6#q5E9I3hA+6cZBervSqPooU0=kisT-;6XKbPz}5c4}3;=V71!!c=u(_
z;LDL`k&7JPjZB3B5_Pkeyjn64`41Rx+#doTrFCaP$jH~!z(?u&?Sh)VSN(--o7MTX
zI<V=WSO-=e<Pv1&tI<C?!p?m7<>SJED{4{M6*gYz4i#2ja&bumry3ai42*$~;lQtm
z;u-$?Y4Qj@!FD42cZ2lb4Xg_2utECoh8`WS=%jY=Bxr^tRCt5*-wo1#H%R~8Ff2t#
z|J^VEJT9U1-wo1#H%R~8ApLiP^xqA-|Dq)$*Wf|je_8MJ1>G~Ze88@1ukcKa$@EOt
zruHI+Yj*7o&hYO<_QCt|?}i$zv23`ow)(wjRQm_E4BoR$t<Enf8K_?U_XZU``0|*`
z_CpX|b~_*g+QnOX%eWI*3qx$qtA_MxS76cI>6T3_xR|dwgHiO<=%%+qhBMro&*xLx
zWs4vf3HrgNhNkDig&9v0b##U@^Qu83Z*?#~A3{`23#pE_<zk(WQ@qghyQ5De&*5*9
zP0yh|uVF$K)p-r6@+g*xS<0yB33#VQ_o{(Cd4Xe@mQ6Y6N1VZbVxALI>?|h2yl1XR
z{D{yNT|ZXrY45TvA679yEsqWa#!j|u+Ja?^=DkQ7f4S@}#PlWP&#%|oNA+5bO&>Vj
z`YYrBh8_?jjdn#cp5N&_)xm#gvy#PskMS@w3uM`L$h)U!GL@J$058y~?^$x4`ODw4
zT<6R$08!kp7-asB3)y>j#VJ({B9a(smqXABl#~2FKomMnM+hw+lie#q_gUFM#BB0I
z^etjGY568*llfX0ypi|C6y_aGq2{3^^Gf>dpM(}e+CK%;R^c#e=wjW}cf7-tH|GI!
zhMMzH!<g5g3fAlNVnN4z2c*?HN!1fxuVd(OSYu0dY^{z}2<+j_LU%?N>jb*0&M}h&
zV;GnwVh!gm(>ad4t=qkyX+mqMAOyW0Po;XG9*n0-4IhrDN=<(ePn8;fJC3?zT=O;Y
z5#uGeUPQoaV`aLM6igpE1=N0xeT`VpQGyt_keJnR4#k_m*i_6h)$mrq?+Ct_IL)+d
zI*#FSWV*KJE!1=x{r?9rRHP3BT}5DTW<_*gMc~*}%cfSa>GSyUWz%8ih*c2g%0Iu>
z3IoP3ojP;jOjsaUdhKv4ERrHu13H33&d|?AiJVychRjW}9ZINZbu@s!zK*$29i!0B
z9px`|zrj@FMk+&;6YaBM<c4_RBfkYcl8XzRo0EZev?(Ho`$YD%-0Ae)qi<4xuG7fu
z$TjRm4D@V@I3-?gT>pBjbhi5;L=EhPf^T(4VwJL$RHM40*ruxKJHn3ssCGMJP!9H(
zx`TB}3a9@%Wd$hcNVThBXtCYKo@||9OSQ<Pe*{Ok;<y1t)objuMyzbaGIO>;3%;cN
zWCRK%E;DueP<Z?@^Vd1DW5{C1AeNc$GJUe{fJ#0_J41;#p=_et+${=_b;DODy3CiE
z4P3^wrv9eLCRK5-llk<<@)=)d+G4a=*Qi|>qm3^!^J27dzRVbHe3>~4P4u#i^Sv0O
zrPIjSa9nMFj5f~qK#X<@`NT4_A2c)_@A8Rus5^Cn9d3Vv+^OtP4}EFxTOd<x%Wy(a
zo!@||M!g!k1vg{ir4!bW-oFf>MW#I(UOZt1*^80{N3npkY}*Gxtu2=JeaV-Pv3W@k
z=L(pYYzAb|*N~$Af*7gv6&aym3|EM?B$?$}k`qkyTN^0`O@@;#4&yra7O^Uhtt_`-
z2im-{L^?4UFfA~1nI7tJvc|v#W;SF89ucfqU;a5^eR-HB*4LLJG1ixPLW%hGB?V&M
zLJK8bU&gI1`@v;eTdp*%EsHodokR0s()#Jvmj?HzT4R2C=GZ!u%gaxfOfE5tr23z6
znK{o$MmwYQpJ}NXOWY|FH%M43p-sYK30Yz0<>m}7H@^_`Tk2qT6l0yqy4Pp(`!qU+
zJ`2{T^u_>$c|DVBnRtQ`n`->mKiE{|q&eI_02eTvE|#g7a=|+lq-OAU8Z4HL=VJ9~
zUUe=O*^7Nqkqmx0h9s)7>XiE)6g9ywO{-4-9IiNLslmTSb|ZO8RapMWcOVQaN-QqH
zKjP6Bm%N?*DZLg;qwtjtTvHkj^{LjCpPpH)E#IIWK6M^)U1>P&XIxtve)m6GU&a!5
z!mwPVHb_`2p-sYK39W!)w@9ur>9q7g*1YbO#dI@J`ay|#B@wPR=_pz|hEqBsiO%r$
z0f3m3roW20-mQ4hBNm#fK3*)ZUVWeF9UJT=QLZbQnlqHZqC2sEtcOl9w>dy*@Ze!U
z2C=YY$6W@l;7jbP-vpVfOASza8IK5zR&hmWdKp}m@rHg&LYZY;`H{CecwxRFi9-Oh
z(n^?KijNT&389Y&4t>rv>+gEw(^sN*#?;-;yrAv?AVc&CLs$}B=f~+<U&eK6yt370
zVx5{yTjrz%E{xY!5WfW?maFJIgf?PXanmc%H@e3W7|IL$6KmLSVh(O-rrk;@`T|w#
z0f<$qo3r#i1;#2>Oi9G@^y!!B5PQrk!Nb}@+9P<_Sf1XSjAw?<Q;mlYp=9w#pd%cC
zBP_O)y#ps&ovP;5DV{e7(Fr=;1$jwVr-!wFA_}u@M`*IXINe${id^Bfc+`QxTTBiu
zeBkm$Z4|7LKa0ibVXcXJCz&2``h1blVMuc!f#GgyNG<_C9ySw=CE%;^RJjCvDxNBr
zfIH%;=<~Q-{AoN@E&(^iQB6xgUwp*fqR^Cc4I?l%^n;@KPZbL$+-a3Ef#5U>GdqrG
z6z24Jsw_-eJXIEE<gjiIUF=#}m}lduvM{^jsHVc)6(2EPm@p$y7*b?mc&l}jF6VsO
zp=?w*m^O|x@m8w^6TEQ30(a9*`Vv+_eB=c@L)4x_(IV$!35%vv9~j6~L)hMoVh##`
zC2??{Wz*lWL8rB!$L3{TgT46@i~oo|*%bBoJSgZKH$29tqt{kkS8;vCiVD~Dd$_6u
z;{j+U)`vvKTlLR{oUNkP@VqX5JLDUTjICvjy0!Ekf-H*ZXK5S36xoNn2uBFy78w*9
zH~NAW3rwtB;T{)L7F)+1Vgu%IBnYT|_%|fB!|k;Z-o3!Pmw8vlUjHw=Tkx7_Xpt4<
z{{>BVi{A7iYxR60+(U}|1R5aok+WD{LsgR-J`7p00%AUUI#+*pfjDyN2ouKQLdHTl
z$P{9s{APC~nKP^+gW(KMPBJ-zrS}Q3y`=kwb{QBWJ3ygzzI7aV9?|~!B4}U77KZjN
z?^KrVb>97gcYovE&#=YH6zh?4<OuFYe#DDWLOP_`H~)$>!osVfxCtJPTnotb%O$J@
zG@j3xo~H-i%yZxwb?`$KJCfm>vHkFr_Ex?=fMNm=7)CaP^dQ0k3F$<{`z72gVUG^0
zxZz#Zz)P(<6c`1Sfhjof5*A3v2`uq!K(ML+4C+uuXOj({)}5_h(5=9^2an@<mrs3&
zu@yUb^`8ZmS810_;kIv;s)hnRpsJw|TcR`ERO<{sEW{QLs)HY;JA?bw!l%|ugL%}z
zJ3U&KOgD0>@02UTnLTPSz=ROPF$7}z2@=wVA7-TXI39U3Pg7dK9e$XWfe0OhsAs$g
zEGTcu&($^>0?zP3AyA|SOfZ7!EGTLS#tU7!JU)F6H-oM)QlIS0))Q7OcZPqqU(Z_y
zp#u~07HM;gwByAlr!)N9VLiDHLK?BS(j)m%?P;7DNGLfuu{{35&(9ZC`M<`1+e;p=
z=^L>0o`U*xqBPmgPz~(jTcw8X?LmA^=qiZ07KOaI))~5&Muo*`HB9Qn<AL$#1685S
zhbn`ayDEd#e^7wz1hTs_l)1~YEx`3eP+TiRhy^86P<R!TA5<M(^l%kZ{-E;cqTNXP
z5SXZ9P+UWSPjFvp*_OFGC~ocOQD<}GxiVD$&?ktf-|<OhaQY5MXc8#9O%yKAc)mt3
z4`uE^mbjl(Y1~GKYRZQ!w_k#$R}&m`hF}GFhVaNxC>Pu@^FBq#oWW^8Czkup(3L9f
zS&yDyI@m@yCsR1X_lh}pD4>H-K{7G#XoV&*r^bt63RE`RA+*s!$TmUDS*+|=vUP2o
z7^8Bj(jfYVk!b=IjfQ$LZB5_f=-qMS2<77F0FD3zhR2ZLPnO;v!Ya<-Q)=KJV`_N)
zm^%1Q2Bz%MA^%J@_z|8prghK2{1=ZKJF=Seo(B&;JNA#vc;C`-CVtCu{{SuMV0-Yx
zL$_$U4-I7BNVOVv-czg=z73`?AHe0S!??IdaVc!El-VBT<w;&1<AqNJZCGKCc0Y~O
z5OoN@gN%8evMtjFc#k|f_owl0e~V@P(~k0YyEB7hBWu-gdzzy>q6SACmc@U?8dEKN
zS3D>!eBVDAy%Uyq1zd{y&oEY@AC6>Nwk^6JzkB%jC-^ooHM%c2reY4=b$)Qn=#Sxh
zV5`0#{MGko^D+f`W;)BC>UtH1^;bAT#WkVHHKFPGj?ncman5d7S}A!;K9Bvu(M7+m
z2rjxiH2v2&NiqGN;PiWgnZHDKzedK+@;`RHghQWsrfC5l`f&uD4k6rc-g?M<0<C3p
z9R~ySzl}C{$iu=z#4N-A9LAQ=!@~0fp-CqQj~0YwI$@ZC1))MG?2i$Ob%N*~1;<RC
zAUXg+$k7Qsu_VWyWwtl3(j~s5W3@W=vX05w;-7RZmvNJ>jXVbQ+C43sYqf9w1>@@P
zAXv25@$s+L5q!>pGZ(>s*7QX!y?k-d@rQfw(F2}D=<OTx7)x&><3f{oZj4993isk*
zSC+%F?Jc!%2qyw7EK@V{9Ybg)&|49M6o#iAQdYWJzSkngHf<(UQG<hy;FNp^rh3`x
z_y);-2R1QtYeJK1LQAr1%J*A(HLNDkn4E<JYG9z>ar8AD!keT<A60_~(7^bVp8I&%
zMSaQ{9J2KOi8c$2VTV%3H|jXXP2&Cvdg;6hORlZfGPR<XH#XeZaD(GxxoR#hwN_k~
z7h9%&6~8ZAEmIx2mZ@ouFecgI8!*v6;xDq~E=KyxYXOn`a+@W0vL$zEt|j*?R!go6
z$yVTZHIgnt(n)Gys6p&&{U0P1yC}Qh^G<Az6n<Rt*2p65`>5Q=uQ{KB4-aW~106Y8
zK93R8))!$K{n_3-+Ri^AR?{;64B*(T21nI2&A-r+TRjIQsV+iMs^_3A)kP>w<{Ui5
zoP*-j&p~;vn}fn!SA<J_5sKt2LYXp)P#X>YdOW(ELr2Er&msYV_e)$rNNu37PD}2V
zIX#TY-8yHp;MiKE!!1RE4#AzwAEtt2B<h{R?w*nb1g=P2K;Q#H-ljPR1l%%bSir4w
zMhT~GEz;qZB0-<JsYvia-T>m^B>j3Wol>fRklG+|0fAeI=Wd!)EZ~+oHUYQJsTF)%
zi*&f9NYIgs;6pA<r*7EWLs0&PrEeo@$`KgsuPGdaLmQqP&RizWM1`H%4WczT&`~*}
zPlNvxG!q<EkL$JeXx~9avg?WnIHa8ew#}kF*EvIrRXOQ+0l7Lu-`<XrIz!)Q=i>}5
zC^=FS#16*aF)Z<@=R0Wcp97yGd_%4yyaK<Qa%-@r54=9kAkQ<kR9iZRFh+2US;^sL
zCVn^b<^ued@Mb}aW#No5j2jDQJf8N0qaOt^u&jrm^?&@q(PMZlfw5@exq~8p3};Qj
zl@2Zv9~J0Wh9G7bM1g98|J|58oJ(8(p)T)ZCjCgnYeK-ghP;mjstEz_8uC5{Du=;;
zwpfloii38?F%0dqVugp_kQF;SW0t<pVW#2?z61^ycfV!Y<G95*SeE`cQmlW{Q9jz8
zU5%~To9IK~^%xPy(05`)#F@53YT>Ag(eTs~93UGOd$LDIrdNbAAN)bZ(M7-E$aH^2
z+Mkdd<6nB$Lfpvl3p~5X^HA_PaLwD9_EbfvdRHb$w>*>?JP4X|XJ#6P+o5!4`B2w^
z;Qnj%<*iI#GO%ZSvWRoT{|XQ8kLNdy2Ws$rNAS2<NkvdbN9giwM`%f|oT#bgf3*Z2
zVcB=XQVf^Xgf7L|XlLQa>fp!dXOF9e&#C1vS#IA+uJY%xxN=%5M;vL-TE6?Jv-};)
zcL%c%IxY5ha7?NO&wH(o@?(~7d>_nLgctn^K+%pL;L7bW2xjiiLRgq%O}Zb`@7JBk
zA^50vCV9e(?o-3l|C5PIj%eR_3N0h;Z;tTi@55Ne$G6yj1Fa@_z9YCI`^s?UeU9Me
zxg)2kgMUviIf6bDrw6q!gTs-wB0Hg}?XGCFDcT2r#Hcw3S(&Yi`L4-c!DofR^Z{^^
zW9af+s)Hr=4VH~x;^5UZViZgbF3Q)=5rrG730}H1oWHFm*dk_&7~ZGnYkQF$DA|ta
zAA_3td=N&o^$Vj<Rt;rtgGeQ~ckr#5X?qVI17qI{LG8A~kxgjG7=dF7K#^=J67eA-
zc=+hyK=k+@SYJDNu`K;d+VdDi9S`ADY!n<AK@K|4dA#I^<D7$*URFa*P;(USM=HzG
zchdeQj_<sN2D2B_*S9iYz35*5NjR3t>8n4_Qjqn_GZdu?9r(5n#PwBut2=Nkz3aLw
z!WaGu`#9JwnDRU7q!#|snRX1KAJTPsI_*hpt*Ypl(0U+h1w@GjQ0#n^WvU%Lx_uPC
zVf--k?;SwH9qs-{O>i%#*5&V6dcOhD)xpTogMp8Z%TgY!3_E9~2cHPMl7WFg6O)At
z0w1UO&kB5;?mvarWYJ;7<hOy3GP~wq8J@DOB79Yv%F$Qh5AAD6e0yl9;uGC;H+<j-
z?EM6@>(RiYpY%O*+i-Ai;8<GsG$)4H_#S-sSQB27IF8FsoWfTgI;9z2cIkh^%N|GF
zWAN?;74IMLH40M`-NCmdOnAlBb$GQYPYwQwZI#ovm(<`B+G%(JM=d|(|02fFY~Gxy
zhKkj2eSR*NGT%owJQFaa{R(-JoW1Kqw+&N@quR}g)3wDcc?F(f4QU&}WaxwW@cY{D
z@wyL$TD~`cDP?^9jJF58pRf@|aRfh5gJX^mrg-I;;`y`WOz*n<+?{zSAVx4OL9_>+
zK(^sq@~0%oLp0jC)(x%H5SEM%?MQXDHOtah0@-RB=7n6tIYKCc<6;!ge>NwAldwjc
z5=_VAm`q0yQ4ulVbA4J9!jp>N0cRoJV;Dmxwf7+;S5!-NL`~ak`R)K6c`$OR<$HtK
zgX#`?CBkK@#jfE<%EPP?|Ai(xrn27<({-cAa#GRY9%dtystF!<7QW#K{>3rKGs>K*
zeuBq;(9j7Dv)MhJBFgx-_T?v|(Iw$q9?oQEh(7Bb%>tu1f}hW?;iiF#*Cq5xF={ys
z^C6iw8);BOEdbnMcnUKMlJ7vm$RH5y7Q{qWpbU{=mNS$!vcL+PV{Wa6PMvuX?s;+~
z=*$G3bvkjxSyfEl3$|})Kr9Z8tKqfULwr$JEq}VZUc?>f1%~Qs&~8QsJWpg2+t2%$
zScqah3--0zR%nL3S$c+fIujTe6YB)d%Fhs;!Kh_08%Ga%!H8!$1S69}+2JoogSl8=
ze|rszBn}%OLi_&XC|@%vI1CWV$~Nh;C{CA9);t+^f+G4aL!<dj>fCl}OqD56CBZ^p
zWsyT!x6g-Fq~X`|K81yDq6F08G$d=$=X3hoKAcXf!6)HM!4J{c_0t+Jf*T#>e5_wB
zQz7$W@kUmL8pJVo8(gFcE3_)H2+Yz3hHxH+gOYZap6bUlL@Gz<GJVjR><CU~GDmQ6
zHhwZ$5XpwJqVNqUfXzt#@#J#)12%ni@5gBm#O<nNCsw{61!&q^6VBhq2b*L-$Ku3M
z%G+{S0z6NXhd)eyW$9alMWwU+HB0XwnF}QM2w0!LRu}sU)Sx_)rMx+i=PV!gU%+0D
zH|L68PTy@NFPpvgOC%gQ7p@WcFARi~yb)wZb2cnGKCMp(*LU(+m*mg}SP<x{vyZm{
zOP>l|8PfwcbL+75UB?(?(a#th+Io|ZyU<Y3i0boK?NVu=Jds&_ey1!3-^LW4o6y!Q
zeOGhZQ2uvI?+%gCEjSPgF}1pwgZn7DK7Y9sa}LDN2-thktua$r1C5kr<;H~eBP1D_
zNmI<ork*m66|39fRw2`{1CHov4^vt60uW-hz)zu>c+eS7eu!jQeD>RK@eL4Uqf$}Z
zuB9yg8B%Xm>g~nevn)Q?%$f8-FH!s>C_Wv#Q1(s5z=is`l$))~9o(zigF_s>Z7TP)
zli4GW_85Z%?ORBq+v6sj-=(4=ZWM94J=%oKaeL%x_fmV<1B_->--fCrLm@Lc%VZBV
z?9aDqFJPXfhHgP03+@6HQ`*2-rlqeTOHqchemGAyGO_!U`xsgE<`yrO`-td8*(0zQ
z70<WKotCAS5BJUe=!L-1j}k<FflEC1uTH>y=LGKl6L5znaP#$tRGO@x!2R_TaNjV2
zn_nGECA)b7ck2nbzcPW_bpq}yCU7r10rw>nxa}w4W~G`2x`^mN9?0i3NBi}o80h{2
ziizPxzPN`jOE84V@B}C(hMn9#Fc@0Na5pF>hWn417_KBkKPV=Kspc$Zks%td!^vib
zlgKa@&v4Mppbdgycf68AW`-d${4Sp1$L3PpMTT&EDZXcB=p#b_l*m;yKF<gq)&`)M
zy!{<@TqvWG4M)9Cup>9k11!C4;;EeNuZWAh!B)u`NMb+Ud+P*YkDfpnTVsT4*`&Q<
zZ}9|SeAtjmvkizF!K{5x<d?V;5|fN*y<9cM7q{^QLThwUkzcTTgXsN+=)`fN_tz8T
zaw?Zf*$%{w)KcOb;}YY0+TH^bB;IubiQ-rs<h===-B)4w(U)AUu<k=&<7o8cF5wiB
zkZ$kDF$uj$U}^#nGgmaCLx|SCv>V<-bD9`_C>e4AwMk^y7|+1IiV_AW;WgwMS!rg_
z=a9pU+)tH~%`ST9Q-;YY<LAok++`mhH&>2AD09{*d^K}^3wJTy^_=sNSX&};=6fgR
zQLJ|t1$^dg^=Z=s;ykbR0k#}Q)+OZ0S&jBQVkV57Cz^vNFQ#6K?UcAtNaPSJhSA3#
z(aqV<Qkcs1Kh5KA=8?}l(5JM*Vpy`q_9x7*=X3Vy2;#s()Db$nrs+#^et}i$u7~M*
zxsb6WeBs|YhrtRS3*EOh&L6SkMT5lqecvx3t)~1<Jh{L$rC9qd72q<h|5}@ZIXO$7
zi657S1-tQFTPb;l6VQiI`rVjZNc29S<Jw?RIX!>ZlT;>vh%3VLgfdt&9M+y;+lCt8
z8K2Q~FLs13tq5JTliLyxaW6u<=6xu#;V~9AwxYupj%fuoZ@jtVWhg9s3r09c9baN4
zV~Dwm<=d0wT5mtH(C&R0{`<l4xbaCG^nim4hHsMLY9txCSoC&@h7=EB<Fa2mQNJoA
z9r4TbaTf5eh)X+nV%jgoRKKk>*5E(bsr$<!n0Y*Ynm#A|i;SF!NSgk1K)IjG&ph;_
z+^qCmZ5PVMO3#*+z8Nt_rDvNeeIEpxD*gAAzu_^;ksCHKe@)!K`B%8uY=SP?*k;|1
zvcIQVW;ye7382<{mak@Alg)#U%Ln<Kb<gIj>d>9PADX-Sc{o|gfc6OXRB-emclVn*
zejo6VFMGG*5z+q%bazi*gR+Zj674CV!M%cjw$YFLP$hkDtv@5SKZX^rc8D?qd$OZ{
z>>e$NY7wBO_{&5u5ZMDNo`1&LC%0tSHb1EeepplZngdTTvG20J+Y!bje!M*k7x0a8
z2l>m$o|`Oe|0u#3X_rWv7_CEdNtz`|tQ)Y2&F2}Cb~|WztYcNfRaR^^-u4ryjn(#x
z*l=?e4r50ugO4oz^KebamR(>hqkFj{G>>m?YR{km_&~{(NRNGZowiG-U1X-+qtm`v
z6Z~N0Y;KK0{>W+M7_s1*J8}}Pc_SH!cLs;Yn=DL*{V`#m`np<d+o9j83E!BlZw+8W
zU}$k3)W<_2R{{2y7Z-p~Mgo>w+H0sS2RjSL1$avlJ8RfJusK3Y3ml=Z;8?;F`V#@1
zIeKmlKDlyO``+v9h8EjKaA*gc!5bXGrNtFPlOaW^7+P$lSbmMA5YP8Zo-y~Yfkod*
z>poiz%^k|TLvY~%aQUNMnz6fzrO=+;)8I_I@^RUT4Fct%D{-g&{Xy=f;dq;T#qqbq
z@xVe15qANhjei8&#qVeStM>3&<7;<+H>RAG%59-?mm`1O{^s~X#=pw=^OMHUX8iIN
z#7~OFe^z~lXmZ$`SL`cq=JO)6y-i!qiRT5x^PywT@)3Lo0*Re)|Ip3ZHE2<V@GLx@
zKbG#mAwoyUj%P-ASb*);^1W-8;WIPX-^6$)za;Z#ro}m~X}=e(yQci8pDyAI;^@pP
zniEt?(d+5>1FXdB(Hn9v)Sg=t_GFK&b%eXoBJ3Eo9>o)JzO|1NZfbeNcMdn7%g6jB
z5T?ENJJhkRqf{$<<Sdkd+LX%No&pb>%XunJrtpZ|aU9k8#F=))QFzpk-LxTR@MG<l
zU>eB)n>zSfx+9|v<MoATp)nVf=S{UVR0{i>V6@_DbTt)M;Q?4hy)%4X?iZ~1q)>Qv
z0brf8@HJRu@EBVDo7vh`zl9N?(Ama*yEFI(bgPDp==2xSn$y(s5xh@{)_m>t#@SZX
zSgb#Pe-By?zjX!<+@cr7Fn)r->No!``F{|UORSZi_NG;B?ylANny$5}qrIuStIN~w
zx2|%xc33@Y@v-3cl^(0N-Ktod@#)fbA3j*>Q><2})vBCyl45oGaMQTXkIyUDRL=2P
zyW73`O{2Hjn<(GsxuF}vz3$c|^4EBp@D1IC3&rP@Tio5PewVuw-|KCTH=4D{rW0nI
z80J>Vc%55WGH-F0x4Dd$N_?l-F3B;V$Cs8=;rr1ZpV!@9)zQ|5(%`$l9bMqwe0zi7
z9cZ)(sPmT;Q((RLKC~{<;2V{Ef*F_2-Ue~o-)Iw%7kb2SqYYQwW3*<-mgP(3Edw$>
z*KQO1RtbwGv`Lt}{3Ua#Z>)?aE9CYz2zmWde$38-s(;gzDjGkMdwPp1+S@O=1mBB=
zRPiC}qNXMVZshY@@p0~j3-wA}gpcH2Y9-IYg}hdKME%xz+nNgybs_xAU#QHU-PY;F
zbtO0e@J(n1KmFU=_%yHgCQm^j^IFs4Z8iz3b$9tZPWL)|9JzqswzkfiB?MkLQG@f7
z(U@_wr?GqGY<!UypKSJx<NjuTak<Lfy>b=GBVw)RHJ=Rk$A@~Y?&jt$kI!d4ueEh8
zX)-k7`_h8T)!Eh21cr_--?h!xQ*(UhI_6SygvrblS3AB&PW$PGidB-?5f?5@N-mgS
z?DGrp^;naok|tZYunz63zGIoEY0c$b9c=|7fnmNlMAXn{l~>6xXUFNrviO#<8}RvE
z@CnpNTsCpY|0qx7(Anzodrq8fp^|o1#>M>GC;Vn^|F5FaVcc)t9*quU;BTe?=1#)*
zcYg!=WJRe3+>UtL{n2Ov8s_GM(P#tUb{w1Iw?sA{ibnSXHoO>(DoEe(68Hc+0d0Ui
zfXe|l19k%T18xL7a5x$r1uT9A{8&EN0BwM^fc!4h0ALT`FyNhld~fXl@pq%q>@47T
zgl_}f49I=!e!#VW+W|KN4glT_I0`rb*!f{JdI<UT01gB1KZ1Nw&h1CR4>$mLH{b!l
z0l;CvLx7`zqkzg7@`XLDfL1^oU>RToU@c%LAitQ|1K0z&8SqZPe!%U3+X43j4gfBv
z+>fHsG2jg!qa3pT!}Bu{(r#L&q^-?OJ8R11?9FMDb4mX^es^w*MrY|P;+mXaQSv4K
zj(edy>PE@Qy*wxH3d<=sX7?zUoq6%Bx%uZ(0A(^{+*IonIZn>tpBl)}^VPqFOj)_`
z2^C*Q%AC?+Qz~+DD|7NHbF7s)1(i9)6**;#b4E`JOx`ppob|2DZ)bcxU3^3NI;1Oj
zIvSmhvSwD_kStwwPHkK=#VAikT>16Cb3<Ln{v3@y1xOtNDAzpPK8D}Gv(f0&sN2lW
zr1@0j*y8hHnYj73{IAjINAT_OGF9Xh1dLMX`EEgU@e9!?#K!{lybr&EBy{RyO+tSW
z^t=Q*_5T3BT+q*n(<5fv3A%$_*ohlZ`|pD<k^p-UP_JsxH=~Vk(P0GUa~<f1kk41>
z4-@M!Wj^dD6tr130V6P<`#>N2TRfe1t%v+k&{yF;^S@Cal{vW;IeFDN#es}(F>eH>
ze-r7;AXj|v*;Jofz+(e_3F~uumo15|i*p7t;wqHo`){OM{&F<>bL={0u1ZK(kyE@l
zXE;4DY14S+A`R1JBA)|qV#ba1nFoAA-<W@kbUh+nIP+T>-<JB*7w^NphR<d!!##Ce
zoU=WRKFPrHQJ(_5QAG0@f%I2Fw<e)~0(xE&`gG`*3p&r|DAhSu0_J}q=#+mW3CJHd
zuE?<l*!+|WL2@GgPQ;%{1yBd(GSLp^@=&i^$bU~X+J$?{BcPo8O!RG|(P%p4Wajxz
z_CZ^LeF9mVVwRyjAA_6*yel<_`9e-CU%jtL>1UrHT^{^4(O%P$e{K@`g`j6Ap;v>h
zB%xmi`WVWQDBlnIDClP*nNe=qW0T}hr2kyXPozHrdNO_fEcp}jeN)O$q-S!>f<ET<
zOMUn)e+_h)DHced2l{Z5ayda)e;MCDRFj`$5A5Ti6!^ur)9pEV-^#G(Si_n2oPteR
zYEJRjC#gAQfyp&Fw$_|7)G=yW^!MoXQJ>kuzi)w@Jk&GXEiObe?(gkUGPbeUiqnYV
zjf!9H*|==MuMs~H_*DHy2pMBc9#Y`|f%3C2;Nk=_1MqGl_W|K3d7Q`&<W08iMD_wX
z4e)nFiZG&*?MoJ1mYt-t9i}Z^K*;tgk?#TF?)1|{-UG4=u#3nuCo9TXfS;p)Y2eWS
z1>4zE^go(+E@KLSaOXqBWFsN0t=uEhNF<NtnGJ-MBnnpnq}{0QZQwo6DE4(Y%7%RP
zkF|dv$a%O=$AGI$`J7&!G9p&!k4R-Qky0S^!6`(34ur*C!Au`KU1vLw$bKMXn?>X}
z5Ek1?<VtL$tpi+0yL|#=4d7xTci_My&AX3mJAe!WiZl;H6ia#?WA*~M(1>{weG+4y
zV$8D=u~D~UKt2H!HkgEHvI##p0OX5COf;h~<4Zu4r+8uDO#|FL4Hc{uq5l727%!Uf
zTn~I~$t{RrvH3JE?J-=021FjmWe#FQ7?Ux-XWQ9{3)zM!c{eVirz0P@5TY4-Ai9FL
zESw0up+WF9IQ<Vth{_^Tg&(V=0YBv`{7_$dh-MsshP<zp6mj!=AnVy?xXsvX#PAQh
z(N-en+d!z{4*(SozdS}ELNw!*+o9aybk-<!u*2PjSR$N-ia#;M*lUUG!jCPefyhz(
zQ2&aZ7R!N1s=Am+DG)f2;$S{s#)SerXcY?eV$4#qJphEpvQZkXP0zlU$RQw<#(A!C
z3QEVAA2Q8MAWZX9A~qm27N-cxb+}NSHs<4#G53(|w?Ja~>;WQL5Jmq52=h5c<Sf)3
zW8PyL{-rXe`G80*5a!bisI=n3e7c!WK*oGRwmlN5X3Vodo~Bp^=>u?FkpqvDh`9pu
z4imWtKO%hOtPJ8u@AxATI_=H8VTfjIRWi<;i#zU7|Ba;8_<f%@{|^7Qz`rf<Zwvg}
z0{^zazb)`@3;f#x|Nkv8kY^K0*iP4BgM`Bp_RDxfQ)8Z&2rIEzVKC@i$(nKg?LG!9
zB1I^eZWEBRNQUWdn}Aq)iD3MFbuLU9tm3NA(|N1i+8`*LurhFV%kbgxXb0h)={5qf
zsvzvlwh<^?STShnHUfPO3ivfHKM=T9WWeH31TF>40dr<O^fU;F>A48CvV5GLGZ^z~
z&chiFS!@CtbVcHZ-iJgi|DvHDjQo#B_43r_*hB=*SBdf&dSDww;Lo3`$Lp#8?_Xno
z$37ve7Fn-rB>cLBcSv}bgbzx%Pr~OU{D*``B%E}bUalDuo-g472``uMY6)8;TqEJv
zCA>qzyCi&2!hI4xC*eONJR;$w88Uwf&zEq4gqKTrwS+Aau95KT65b)<T@pSh;XVnU
zlkguB9+7a8Y=<)>JYT{E5+<0Re;#z9wcx78?sk8-bzV_fQR(cG@@|2a6zT*;#^Uea
z;t;?Z{OMXe<hRZ(nqO2rd!Ei(y58T_?eniIwHEl@D}8g8*`4-^I=gG`{Ngz+_(Q<0
zh1LT8`QK(MQaYVi*XYR$v1x(7VCr`_0{XjjxXQT0^F2>jr&843;rA5b50TD>Xo;<C
z?=EWW_O>?9#<SoeVyoP~RZ3Cwx^~FZp}$LykqXBlF2r?tTHQ<_?>bxkN|6_TomeTt
z-)Y4adV3Mq(d_oSl_Jk7S4)?>&Es0tjI;&@q4>kFI0xbB#Pdq=m!gFXm%FPAf5TPC
zFz&zBgoMb?-R5mVx(;N4OOeP|&!f@jQ;PV=7l$Ac>*;@rvp-}XYxFy1={8_kZA*gD
z9||HaMt^MJU&IajAR~Q!x(%4zpa8TIF$LMJ#VCA!YEJJ@w~08v6qqlt@k?U*YDD2N
zWQUReo#{5g%)gbuP^*Jjzisr7lB{rhiG8$@-ss2ql@f-HvL9zZ&$1iojsEm%(Aakx
z>5XxrQ!*Ok09ZswwBHI)*#{fxjehreIgqW9>7*L5d<=eWhOkdJaHC(Z&CwHXm+7fL
z^)&RCWaTE@(56OuV;oT|I>#Z&Xyk9CH_G3eklq-V4$1V!IL7oGs}s|I3pCoq7)On9
zykI(r2;4_Ahs631BGR0`R`&C?GQD_qYZAa6E(QZ`!M!=XF>kS*F4&Z396B_vMmhuk
zA#Tj+jd{m(x6WaVzXpGz{{1q&VSlUJCRnX9J)gsyLt_4SBa-<W>9@Do1pD?D0mSB~
zey3j(^M4Rzb9!}^O)#pfbpFSTiqX+n{Yg|Az>PUQA8#{E5wi$v04M05nEx~2rK{36
z8H{<_{#$k2s)N|o@E?Ql5^m^fMtWl&r^)n*{E7G*NK5^V^!$r`45Obi{X5{LiH!8d
zJZ4O$H!8+3qmjQ+k7Gz{&foYu&||kD0>R+dv2om>A47yWz0%X5v-aqKFRz$`k<NgV
z5N}SuaYKV(-?%|fW$0<hFwz<L6eK3iD1X09f2T}uH74jXfz?PT;Z&r@f1Eao4@P=v
zhFBzd84Sp0-$vddLS9b4UB3~}ZcPzJI)g;qXgdataO|5priq8^igGqk8-7OljsB1A
zJ6Zn2TM>cqIQO_sAu)eb)P&m%l(H<{mV};;2LQ?F8HztCJu|j`F$?40Ue}WHPl~N?
z;`t{l#`qLZ&sL0aC!P-L$aP6P{p8sCC7ynYV)Xy<^eKwbugB9*RgC^Ro}Lq1|HacS
ziqXHu({mN0-;Af@O?=rO#?z;vf3ulM;ywcpG>rbs9GQVfAV$9wPd^R)OoN#u?lbVj
z$>>MSkr{XjWAqR4ba<oD?oAOH${C8$o=s%<>zT2?D<4lcR>$L1`fDZ*1B_GjJAHVc
zXoqt|^fZOXmCS)paQH}x$BWFt4kR7VevEpw%E(5*H0asf*B~fHySV{aGJOpBJT7An
z>jg4R;V~F<*evNh=3)*%m2@6sF^30$P|pUrPU@182LP!jU!h_!`h~Y}PdeYPVle1N
zxq2ksDAy^tNmeey&OA0^4t(H9eTJnzRvBr?=bzVQFyvpGLVgSAY05-D<VS`mzc^vP
zhU1B_LjwJONqT<*{Wn7XnMy-KJq`%^#CGwzlrxYZ=Y7FHv3;c~*R*t&YpRl+&`zy_
zKC!&CRWkcuDEKF~uS(GQ05h>&*GYO}KhPrNOl;SjxLcF7)0;r&1K7mCH+83>Uiw$q
zL=?;-!t*Kk-<JHjE}P&p`13GmC6j+c3Odi5Cga}<`Z<_^8RyCkKj%RK(zBcNd~-$a
zct9-VoT~7zJ2OB%5e^HwL`>j_<W~~>;TY)WpnVzhjUJh15hj?)^mGgQsmiFd+nJJ|
zhhl{OjW$8y5dwy%LAQ##_{&Qv=<f;n@w~)%7KwUVZ*LHhwc@HMmxG>6&udc9TT{@#
zk%GQM(9y5Tde4#hzMX<U8xz!Nn28(f7s;;V2>Pi?XM(>~3A({G4zEu^zdZ&0C!o{*
z4UIM-q)du>Q1GL_P4L?^Y!=Y}I~_W|oSrGOBz^O8ozAm}4Ap|J-<z*1KtDGf{o7?W
zK{objyGduiXV>X)GZB6%<@CsXJV{ob0^MpLCin*EEZ1<OO;8R0KNS;5%D0v2be`>H
zuz@}m{KotQ?j!<_jj+76*V_cnGlUGig0A12uRjBw`Sv&31lh394$!TF5`TFH^xP!<
z&tW0|ROLWIdzphNT{6GDR?tsXHYfPWM$koh6YRM|@)sx6?^`M4pK`KphwTaaRD;g;
z)g#;2a#_4JlD{WGpPvbS_*;`ra2o!`KO0UvTNCw>^kJvY&$F=%=K!)?#(Og23^eF9
zpws@NjW%HV;TJ*==*jHX2l~|XGn6|wG>|P8zLSFgH{?fohh=&BnF5CWDfkbkpnpO+
z$aix>z9(V5CdQWrn~>`fS4BBP(mSu!>2oCgLW5t@Z<O@wKo@=}?JT!blp7_#P4Y|i
zl{-Ke_Dt}T0l^QUGxU6=c;&F<H{M?|+Ue=1N`I5>n&S<_uRv$HMpc`@WwTTE3%Vf1
zUCu*?nXJARf=)dfF4g5cB~x*~jdbgEIvuD8UfeL>;cImIGD#1EPWu?|Ng00lQz@rY
z`r#aDpxvOe-aGpmgao61cuB~YLQP;y(2;LKd8cDSnF~3I{$>R|S$Qu8o$X8M)$=vl
zS7!?T&7ji`wXz=1k^1ih-D=X$xP3Z>oYw^%auWO`8yzX_Su5>h)cZ2fsb@oio*O0q
zW`|8=$usc`{C*nkFc5AKxN&(*${CPydSt|LDW~Uho1lny0C0B}I@n}(D^5XQ4m#~q
zEd9jTpSU9h|NWq|-uW)Q2<Vp?{*;3MBgt=*{8qt@)3#7DnVuCX=q}JXE+qO%i{$4^
zF6QtJK}WktsIQ--kpF<>PaM}GDfrVqmt6muf{uI>`gvyxeoqSeh7|NIQobVnXO%4U
zE=eEF)9t`_dl>czdXjyN*!y}u@s3_qRor8tVm@e#qu`~;1j;<cwb;3|vcl<d`MMij
z4p+QL*EMy1XPL|8?GR6?L?rK$Ww&60@UjV%@}$`r5?$BmZ?HN`7Kogh^>;E|cn8z2
zj7MG+7uo9e`|Y3hHO5qhHD0QXz3<r6<@fo!@$Ph!;&NS4wan#o)YZFOO0%a6FA)0h
zkkI9Cb2YW%^+=zRgyeEHceqxzb~L(MUCnq~*XMF~ucemaMZo5w1?3l&LUIx&7oP#@
z&+lA#CfBu2(Vqb}cek~zLlP4v-foR!b-6BIR#9Vj*_Tv9SzU8^FROX!<(Jpl>s|E~
zl}<Z3X#1Fm=6Hh7j~9;>SM^tyRMa@C6h1I?NgDj3wAkgso3I@>x>`N$EB&inp02Kr
zuG%Y`Y_`QtM`hLW<*vC!@Gn`ZDPChPnZH<Asjj>2YVo$Uy%sNNcXr`HY>O1+DlIB4
zn*Xl^T~vwZpI%RweGNJ5{3T_0`0H(g!jQERSC6~R(FA%4-^g|OyWC#C&(UPBbK_#K
zDzPuSvd)L-xh_YC&$Se9dAH%M*$HK->uB}az42tcdF%7|6Sb{($oH`m-*0wZ6N`#_
zRomri^)@!Snta{D2b4y4(`u;MWGG?S!Ki}ztD)|Eb156UTT&M>UT3!Kk_AQ$HoBW#
zO>UomO(N3?UqLS|jel3yc+b_otj@ncf6dzAg?+ttHsAWX6AA;1Yh_zUyY6B1-YeF)
zmdVCe)#~>7(9$}){k{oyu185eQ)MR>vbnW3q2Qji-o*NJQG+hOtHteYO;lmYm3994
zP5yPA9{3VGsK#}px7pJ!8kap@bWuskzw+dg(z;dN7Qd%?nY(>8oXxw;TUX*zR(kw?
zeD0v5rP;kMK>=}W;IlWh7B6hxwAStF@`#3Pv!T)A(1WX_Xs+If%DiX`JVY_ifqnvw
zB)%|6E?ic4q^r&CZBJ0_n)<8tN;iGv;9uxjT2gDgqh8g~(ZzSbX$QMEsZ+@&R?Z=z
zs;zlxryu7;>}%0O`|P!~O|ToAb<E;|Pp_6*Z!H+40o*u#F|o$Q;S;or)>aof{5G5^
z@q1#T5<Xi{I=5<-r)jmPtFpu0g|7+t{A8yxURaqOQoJpzC)(0cU*j4dU+QbBIJ(q}
z10LgDMv889e0)sMtEx0fYpk2+!hgl(Sc=xx>}_{-<CsjWWGS*pDrvlva^Vn6BRa<k
z9w3{&*#lToW6+E*rz>udIbnxUQXb#?h&TI_6t;Gaw{E^m)`D*y`jpO+V(~=}3<@3V
z*za=)ka~Z*QL8*QXxW#2EoQg`)0bRib_vm@;S9PW;>#Z?3|%s>va7@0%>MW*9bK)>
zb}xz`e8{^1AAIQO@+36Fe?2&s%!{{`okl6qje;f$d!Qd!>u&U}DJi)K_v~(6?9b3v
zS~_5x_>SR3+M7*7pE7!Y1mlc5Yzixs$BYmwzsMF72DEk47dF_=lXMQK^H{~id~ZRA
zqrKU)Hi?%|Q|YPkA5cNtMGtZjr*3Xoe&YPb1rIhRTxjP$^papo;Z=41c}@6$N4&G)
zGal|H_7#b3r_Ns<>v$&gyv@E27e0m4+=@PFeCELUT&&y|p<Uz4C(ZhtNUyb0F?;r5
zCMP}~B1Lq;$mO80x?TQS&Y@!Fm^goty{<5;>xOQx$JgY+)VbN?Gv$;};U)9Tt-%AY
zmVIw$cbCVt#@prZc4J^o;%JG2O3i}o$)+0UZA(jCF3jp-4Sd=M-)U*aoC^JTyTjR7
zY`2%pH)0KP?YJDhbQgK+(DS>y982ovx-RnSlT#q`U5+}xaEDgUT1T_p>78r0<64TV
zowLch3YXp9RtKN3ufasS7$Z<UhHqooD-qQoKQ~i@l?`pJ&&4|QSt{BbXBx?7a&+91
zx%!k(ayvvG_PPbG+AE={F*nisv7&MXlhd}2HHmXbIVq1djOtRo=EkicCvHZB?mWU3
zFi7FkG=}5INe&%GPIzNNMH`<W(Ot(b*K85fEVztI3Xi+lr>FJK>-2b5Cv<n?rW9R)
zG^La4K-kiS6OFM|%96T{`7N#8zEu-NBy@XR&G@|kvgb1m<FrhCI~K!y!o(NdDMxkl
z$D{0R&{BRx$hF$D&eenx>obcwZ<o)$EMa+JcuuLX6TCWAi%ZdymX>e=#fenh%qm4E
z7(W`6mbhGN+Wxg&Onpmi8s=#C%O%~iI$yHc8$YDQ0RlcfzKp>#EyXe>#SpCz2+|Aq
z{U*-1yuQxv_9mmBHC9gOo;%7>AF_cp+hK!_dRJU6qRFs5Vrhg~b+dz$NFNr=b*21|
zrd;eo=dz#+9dAdQ%iYAUDY?AprTI-R9LHTLdo`2|Ge>{Zs-#_7X^H5CV=Futirj*=
ztH_!ou)eV7vJA7&j;17zrq5*HwskHKK1sC>6RJ-5eMgsjr6<8s=+_0y#6_x@{WtQH
zNXeA-h$F&SmaSdu>ckQ34jiiXBBwRQNeiX-(=?mtaYcvHeq&plbz^Neg|BiQKB31@
zTvxT<h|0j!&DY(=t_-_abBl^849u{PyVa9qLVe<$2lGUwMTqK{i}j^_;{HXQZ=UZ)
scPBJ6X18*dh`zO{)3g{XF>QB<#u{r2weEGTnA4fpBwQ{jqW4t(H>p34HUIzs

diff --git a/benchmarks/Grid_memory_bandwidth b/benchmarks/Grid_memory_bandwidth
deleted file mode 100755
index 632fb5eb4123b0f247951342fbac8c2b5e1a68c9..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 45026
zcmeHweSB2K)&I>KM1|adVDXhzqb33|8v-VXx*=J(D;tO;U<CAL*=)#al3jN<5PX5)
zCYIYZ%44lnAKTK}_OX@LJl4koDkQuG^-;k-TIw57i@j?^;u}N-_V+z==Wgz9c4Pbe
zKEK~z4(y#dXU?2Cb7tn8nYk}nXSL5wO-T{-kSbgw5LEdC3qxsFLC6g<kmHJl(*%n!
zSvXG^1xyBh7M4S?meS)v&RUN%Ib9m!bo`1_HAHc$g(dX(JPnN+w4h5!G1WH|EIg^j
z0j~z?u^^Zjg36Js`|e^O;#!u_<8BfNVNld!y&Q>0<pz1VL0(Re1unlH>+Peq(XX24
zSEpJ45p1DIC}-E9m}n_f9uCw|y3ff##1@v&;|HlB2zsoSyB_68&VM#!bHYWuzGQkC
z>L0BDuU@#K(Y<13(G`t#(;D5)Ev?g9i)T)oS(NYh<QKAb6Mv$ew_q_VsOxbOAd-0k
zex!2BzvaHm`yW|)Z~bdu&VTa0U+vjF@$CmMqr80lhzGSxab(vzddgD>SAgGi{PwLs
z{q>(YU)XbBQ)J(#+1LKO^G3^O?=5@4lxuy%KH=dpWkN&7Z-0KCcgq^vt<QoyVc3I;
z#wW;G3?UQY{|5pk!heMZCc?KQ;qzz`d`1%f)05DXo=pklpPU3gCkdZ#ljwm4W1@D=
zOM>r0y@}|9N$5XJQt$R8^vjdrX*`(liw~QV(C<%zUkC=KcqV|?;+Lph+miIxe<Z=v
zyiLSsV-k7JPeT9eB=iT8$ay>oUP_{WH|S0H#fKM@$n!=Ld9Fyp|2;5Hqz88rK9MBz
z>zpL?!6b4nOCrx*N%-HHgue;&7YgSHZFko)71eM>JqG@S7usX-_j3A8>ot5Er_bQ@
zwQ=-wK|c<BEFBuXiPNv-^d??!CCA^&+dDA9!ZBL3o%8AA^x2@J@H6~K|K5ofT!+Fw
zE@wBFr;TSd08jL$vn(9LBlrP-pw2mE3fcadf-J9W@-)l-fWsG%WkJ4fmAuHc((Mnp
zd}WOezu)B-WLtHGT<7v}bah2pqo>(b?O4(1(()4L$lgGM&*i9-Z*{Gaok#^hzH~vA
zzX&yW%>^<s^#^>OHP%J*s~k3`6~Z`M9kRE@=aN^seSsE7BbAt66`0`+tns=)VxEp-
ztKD_3X4&2B4p>()mhK|2%XKRQssiRKWf?VlR?90;xudbs<J71;GwK^#{0%zUOe2}o
z;SUfMiUaR~ST08ss_?sf0h<$8v$MhBlLJ17JK%>L)<wWgU*Yh(ojSFxTCOgTtyMGS
z%K25(WhQ!6pxEv4LlD2K$Od`K&0ONZOo-)ZbVU*VGPAXEzB8%`YPSDMr>Dv5@VO{Q
zYv@e(%4QD`yxl&(b&;Fp1kA;nG!Wh0yi#jGBFUMit*8xE`Kb}I4H7Q&x|*9jb<{)E
zOBSGW&=akW74B7LQm2zff=pcimDGFCQPF-ZERd^YSCcoeM)tY97)KtTW2K9_a%F3)
z>~;D4o@Pg*8`-N0crpKKhZp+8z(QS+V1`lf6)pA2rACRWK#|ka5*S_<{gtGwCa$N^
zZzWYVuWn+6V;Wg#mgQAV?!*oClwi#9p-@LX@2SQ_fd+;;x4e+|h-bRDC19n_s;)-U
z{-r)iAQt+;pHQl*02I%Md84a&WuQTJ`FtLqK2U~4W8)6pO(UkRv2v9F1JI<+eWX@8
zowA=Gvmh^SUhQtK!{qe0G`Wb(sgZd-es`<f<Z41cu51?2Ky;l<wgD!i-r;T(yclwU
zdOm$I&s*IA!S7zlhK1i15Kxv!PVX9Fr7I9{H@Q6Zb&fSe=5Pkwo@QZ%!+9&_m(xY$
zO`cV*A&l%_)3gE|h#E*NZmpUe?q;Fh=W+>dzqh5?84%``mC4id3xs)g+nh4FFyCAh
zO%~?Q(9$(EASuc(5z1<6O6S;QbN=+0j94<W*_a6vnu=9$D!Wn)<Bp)E(y&g}Q-q_u
zRP22;>d`EZ(i)G^taPf7$?{XNKO2RAsq9BoT56P#f`4P+)AV&?<)v%R5K!t};3%Yk
zGLPpM0Dkb_9>VhbG~qDEwVdNVjiecf(r2KI#*fM?>+&GdXyF8>|I1UiV!Jj<I2~82
zvxDVl2opK%82l;d(}jyU9LReBTc}i_5HN+ob+s&}Jpd|>2HlSm8Of;!-7gZJ)`WV{
z{WRfe51|L$FA{!+j-;6TKf+&O!0UdR@MZ&E_m_mf(ty|fH{lBn_(^TG3>u9qcgYMk
z8SvD%ddM~4X(Z^Oz<}pn!m^4Dc&b|u76YDist3t{Cz<q6X~64yTgt03;EnxKYrtb*
zMMHxDKPHOB`on-9YrwY}@L2|Yn*kr)sw01c0Z-#h4;u~maXJ$5!v?%@9lFVYKhr?}
zgaKb<z;_w&6Akz+2K-qDe76Bldtp87GvLqGk%;#j@K{ww!$AZ7+$a|31_r!-c0$>z
z0YB3~KVZP0Z@>>4@D~{H!unbU(HNmUrXDg4c$1DqoNd5gXuwZ0;4e1dO$L090iSEY
zUt+))81R=G@WlrFWd^*(fG;-SB?CU!fUh**^9=YJ1O9RYzSe-BV!$^T@KX(VuK_>J
zfR}=wWJ;lo#f>IG>gWz+sE;F43hvI_WgtQ_GGh)%BA1loDo(Z_MQ{TVh5I51m&_uV
zOiXw`g9`~xCAgcxmlK>ua2JCwBA84`coT!qA(%`^cq4<y6HKNf+{WP338vu?_A)q?
zU@{fqS_Xe}D_}Aa;YtR7PVngjTNwN?!DK4J1q}XxU@{S56NBF(cr3x$41S$pG6`XU
z!7mX^rXW0U3W!T~5=<r_+{fT;1k=<H?`QDS1k;occQg3+1k+RxcQN=^1k>?UcoTyk
zC77mmcq4-!Aeg3fxQ)R-BKQXcdl`H;!8EnQwG6(UV4Bk5N(TD~rl}mZF!&aNX$pr6
z7`%dDnz~^VgO?IaQ!|{+;6((}lne_DzK&p;is6Bi)c!dHn+Wb>@GOF9>V@|+xRBtB
z2<~R^<pk3d40kd3B7$k^g*P$y9D-@eg*P&IJi#>8!fgybonV?`VK0MI38pC&u4V8y
zw*aQ86Ru?N=LBC)u!X@N6Fi0B0>G6!>N;u#^%68~2km9I6S%Dnv!sUUr*cPIC-O6*
z`2@0jbsTWw4c+3njHgy%*7RmPgNVY8>>4oI5G7Pmk4E@&jqvdp!c971O?W#f>+{9Q
z?Tio2ugV=;sQwOQXz~%r<s<!=kJyw0H%rO~Qt<BsmDRT1Zs8^igz4G1oT6UKB#WTl
z0KzZp#mVJ@u;qR<ZLu_M10pFnm?bHPSEt^zYYSCK6aUMbc2HKNPn_IFxGe<f^gV-V
zpqG@LQqRZNN-6uL1A{Fnyp0H@j@MCGQub=)Kj7utu68$>1fgZzVw6<ZV!+(AH)9D<
zDPQcO_DR9354u1QxF9ORgvC1s?t*m6;mH1*!grvEI9Uk$wRmyq4W&y;7nfF7N$Wq^
z0JESxpK~W7NqHv6OK~TMJ2hB6W44B@K_(sY3ESwR{v9JkT?or)Q})_IYcp+At}!2x
zly9Y;Z&T~Vquur5y7n9kG3;&6sYFa6Fi}ztN<Hc|V#i~cF{%u`O3EHd*{qQwmY59X
zKZ8J8Ia_F^Ewm(4%KKbW{zBXXnUZo;Z9jpcot%ye?Vy$6_EvP>wmMX$RuMrcW3$^N
zNFj?Vg&y7?tc8}}7Tc#|_O{(oECe!sR94o1wxk?Vt=|%L8Ym04pN`_dMfQu^2N+lN
zAH+-PRj)v-)$=aZLz7VLXC*Vr6V&qttVUct?|w5<J?nTqz2F(&o}gCGJ1`aT?O8p%
zo|X~o`2nv-j@FZMGulH=Lp?W&%OYGV;YLSsqoZ_RpaxCGv4kUsq#lam#u1JUl$WEp
zA%1_3;?5#Eaz^Uoz@fiRtQ_9HU&0#4>G3CDkEF*dNmuBvb<uh*pn7IeJy&V<JeZ)K
zvf=ei8?m0(h!^Vl0pe0NIqaA|bW{yv(T>_E!RLaB1~%qG0b&X}$mOXwF%G*V<*^(_
z&{>vE1D}ZMOU35B^`f_1_(EI<yV0JL3#?tiuz!XVKbS+aQtG^yO&_%u@=MMKbM~Pl
zVNOb4pn3v%QqKpYx3?ilbGr=%)Z@n^k&bSAXfp$u;joKVwe;8?QS8_VUuod7H5rqq
z851dV?+_D7wC<A>dyaIh2aM(sBV~Dx#lWrf`DE;zT&7<Sv4c!~rN?%I)d~1>)q(!|
zTpU-PQw+hWjTN0%Z$+m)$18=(b0n#=(rZ3eF6L$oUMpa2rRl!|M(MA1<wd*F{Tx*-
z-uG)<4L%}w-o{e##%?ip;U>iU#M~RY5bqar<!;0W#oU$q5%-C?o<78?n7eua@qn0n
zEuulOrgWoNW7{OwT;C<ufK82KzgXk$6Kif85Nk@rDVB}m6ls$<rLs$$QqwI?sogJ5
zY3LKDcn8EOdOIS!q@da+1;h9qZ7b;)@2!Wij_h53O2xDiJBk6Q&Jy1GO3acs>r9#z
z=xtMB_Gt34xmd0oQ{VcE2IZdZ(dkC!(hBXCD_@9Dy^a63?wTY$EcNuI)=OeDEcJ?v
zH?N_Qx;*1O7T*b6UU`m*3f_r=#O+3!K+rjQ^r(9LSS0efm~GFg{9Iga&nZSa<1Z-s
zxp?!w&!I&067s}c@BpfZiWG4^F;YDRK0P0%!?%AfUcL*Y<vEo^JQ~C{rPmf3O{VU;
z(B|R+s=4Ec*#0OIZFj+TkNT0dtpD7NM81Jq9I}!|bnAVTS#L}hbqcGvSM{T=ijKbr
zGVRJg)V-|4oftG7C`m)Kb2FQ_Hsz>YIfCW;uZg1Lh~{j53V?V`K98h$n~;ap559zu
z!M*9F!BZEC_fCLG?HAQfAj8YiU*Q#4qQ=|s9jw}nNlxB-2A^~QE7&+2_n+XE+$U9`
zJ!E_S9Xz#qWa=Ul^rPOO)zh_Ncs>5%_1H$Nr=NJCp6StghJ5T%Rx(3MP|w8S^}PSZ
zNbL#odR_)ktvy3N>4#BVboi+LC~n9n$x+;pPnsXa4f&)iqqre{7e{eJK1l=)v#r4~
zT#whHd*k{mWyJP9PQ0MUeo$%c8}dp2!AfQb3F>(tJD<3E9{GGEJzmA@34-Sqcp^+#
znpGIzwJ;O5?ZB`DEWW;s`ll|?l7H%^G6<S~dVwTUn15mf+&>YK6v7;@7n^%xd=yoQ
zIY~ZBD-h?So=1DgAJh=tNIt6Svk`pMC)ht4xg_-}V>y?EUgfvM1rDm%z$&S`+ciM)
zBy@Kc=I(01O?_cHx6&r~Mb<{RCX2n(;<a^_=Wzd215+OLPcLFSU{`vXujU?#(t5s_
zJBRzH>$rbf#QoD!?w?k0|8xuYPd@ISW-<S?eBMTJ`NB=&@*BFu<#M;UeC2*|xu;KD
zzIs4heyya0rJk=%QqTLNOT{VWV+CC8W3zFs9&5sN`Pc$n>&9AeZ5&&PYhY|Gu6K_0
z^7ch`*;BqEw<QHXZYz0Nymx+#>pGuY*As<YDH0~rxi^~<x-Kac$f;D%KT6{*&UtNs
z>dD1dD!Y<7u!+T*13RC^+<{>!byR(h3dcLJnLsiJCNT#FFO{7!9z|mum_!aNTX$f7
z&L?Uyi20`wH_m}cgr5uIBo6E`bS`sX8RWpuA?kgfQu~tluUlEYz3QJxGod@-syZU@
zPw4>R$+-V2*5@~JJ_ff6@?3VXewSDyc3}Y(^<25EKC%kI?sR4zf`f2jpJCMu7xvN-
zc(0}8!Vcq9C*FmfU}ZBFFXz2zaAB6LVW#R+@JjB&HnPs{RS$xvR!<wbunV9e^+Bzk
zBg=-@^8pT*;!NT_Bi574>+weG8FFD$wR$=d)YCq^o(o2-r<-`e#Ain98FFE}QBrGT
zbO5PcQQVLVdn}3@a$y^yxFHu7h~kF$)kbkcF3b)bX5DX=4%g!!(Y<kcyylaU`m2w4
zp?$fCV_nNY3Qa#cABp-C3%m{6`y-~H`I9Ebg836h!2L-j5hwH~^t=lOF!>X$K%75m
zL?g(2tJtWLKS}Bd()u-jlF$=8c3LbKW{nf=oL=~N1h2G&xab=mE>Kc`;%A%^`jcYj
zPcp$xZJHMKC!1LtZP@deU&JQoOn4fc>(KsYC-WOSNXz1V58+xrw7<cAXwJrRw!yh>
z6WiY`>SEiQrQK|IvtmEn+}zT~_BOr&wzZjsosFctC-t0yYxroiRh*Ju1gmc^VxFOh
zd4?k98H$)^C}N(Wh<S!0FLr0t_(-ok<qPH&j*9oTkXPtg|FyQEnMYpX0URE&?HqZ9
zj2hDjUSV{Y2Hq~?mgbP^g90_Lut$9A2>vT-PhWc6#^xmM`O(HE11_ju9P$f?)W@kf
zcp5e~htvyzjQ0!QfJIWja7bOo`9!B7?G4_9z&(f4w2jRkwy}X<IHcx-*ytA?;~N_s
zN<EH#W3zEA9ZFrO`+q9+gi0+PemF&25<9EDSACFF1ou#^xrf6j0{d^=-q0L{dw3Z+
z(S6W%;!OUERzmv5<}_B_F!ykgw!?YyAMgxS<Q~T0+$P>VyvxdF%v{2|)8HO#xW^D@
zZe9bgL=9tH)!SIF_o`dLQ>*6}<Q~q4qSO|xo>vzSucsek#nrQN#ClHS^;{pVXUIK_
z)#|BFP*3OZdLkc<)Sh1wFPP6Tq|w?l<Q{$##YM-F`fpL(kbAf*iW_nd?kH}^JuHsm
zhWM36aYOE58gQ6?_f!wp<DYR-6sO0@Bet)Hc%gl#Me7;b;p|~0GcHI_&yB<D`SanC
z^mrq$=a=AFx&!+N!QKh0&f><LntBS+djL^?v>(Ht{81lb3YtH9m*mn;pcz4@m7Tv4
z5uLyF>U$e3&Ts&C5NKL-!5L}Pan9%v4oJuuwNhPhMr^Z{$c?P}yXHooi8<^VC_T0p
zhg~9Eo?WTSsgazIaaM2Rs5_^D+Jdt{G{eGfnQOjfs-(O%{D^M?t0AF#fhDr%G)O%k
zj+RnhCV$mR7cX-<KZfD9D*>!BmBt)XZ+T8F>FS?q)2}`pp-oqTula5D=a+JuU##0V
z7%98*srr*M;4pTr|Cap2o&D-6osJH>!q=M;`i64llzQX?>c2S8P=mIYD<?It(3yUO
zH6ixE>o9I$kzY8%VxwOOp)he=O16du7g^v@Kr+8zIPm%aMPvK|JMiNBg+po?=M$ZO
z#AiEn)N`00cxiqC2VRHNvp{T9cH2VfIogTWV<zpy3y$G;kb%v^3FH`Z+3+43-&A@7
zDIuYM_=h@`iP5VDP;*7cG1?UzS9@6je*D$>SWX+Nuqp3IA@n&r4uhX~oM?2H@DKnx
zH@kpVIQpsB`|r~MST7F1BG^6rnB2p7yYhvq03W`NJVXKJf%<3;%7mWR?(nd^%M&0V
z3_HALn1_HdcT4Mc(;ca0rOQiiD!sW>zG)ZOh5I4gju+4y+ux%u2?E@jB3lG?J=L%5
zg)`G+W+!`}Fy30zw3EEI2&2pV3IY6#Uaf>j3Xg(};G%D>h*hy&ISdW_(pd080MT95
z4>=b)QBhAPF5OInj0<^SCXiPBG9#A-j0-=6EahAVNK3onP5SRbdr21DB%`ASI+uh?
zmjYwStd1vgfDM|5Ph))fS>&-ktyyglLtO(_5*A_1MA4aPI+-JQ`2tXdkAoz96qo)n
zVU?nc@cW3UTDmuO5NR4y=|uZ7fbcFYxBnTn>kx3eFrd`mQtCojY4tHmt)tWflv)gX
zs&1gvpD4AKQkP)mtp+F+q|_~xdY@9Yl=?Ha9O@EEO~l?nwNvU@oZ6^!C^Z3RU1~9<
ze3Z(kl!P-WHHT7vz}c>P4yDc*L@Jw7Zc2@!)U~*Gqn>yVsh=Um`q(%MF9J>MD5UZ<
zJr1Jn>W@hQOEGIuu>Y^o`#FaDr@>Dq*_1aV<tw`F*?9*Jg<n;Fb}`u%W~PIEP#ny4
za6gZ`dECY0O+4Pn<2EfW^P*5$EnO<5-XPfl3dE@x9Z7L6k0<ds6ER37L=>dnv`w`Z
zjDeO+X-bQLiPdu|RZ{#Ca~P#th$e6@L~c-z;AW>iRE8yZu#HNFLK_g-JDrvG&gU50
zydInorP?t(@?Kpv9z%l;d({8p<@(3^&$zBLy-iYrR1j<)*4S*KD6VYhbCjvB8D5rA
zcRoj&0She(k({wC;G6Hv$ySSX250*r5HMT<D$t+7Fj5H03(c>IEuTd;>>JoHYI6VS
z{!Fc4S&hB(@qJp|S`^wptZr6zyIyvv*+gmY{8OJ+T#G^~K{2gI@#E^c;l+uvUi6&b
zehKT^e;)*FHNRNV(Ji*0f&R3keVO)91@z)?kU~$kA-^KD7%f<qErsr=w1=J~rK00V
zYgTh4&Oele(zni0(znl1%3lxwYy+@kPAGl5_*9V2ffaV06GAR9WGNJOWtd0Gj?aFs
zj7mN-=lJX$D7qCyB%;Euq2LLm=ZH_G?@(CM3Tr-%yx5#j_0|)}sNQm7jxu43Ei?+4
z9R^JQ`Pz8_BkoP#LPM9R^kk&BqNG>6`|1eHuF_+N-OI$tU?VaV$_8~*y|2>1Jd5^1
zrcCPD`4Z`~2wsN|&J7rxRKni*q=^cKf?5>HO~mG?`Zw55oy`%<X9F85fzGCR2sT<2
zvJ7K$E_L<;oQ)kPm)RuJFNxQ92C_**a#7ieju*xDhmf#^vdzcr%04OBH;4j%(e8HH
zl~<+UM}t!5+Ci!3Xd0ffi1Y>~OUl<$@copQvtVsr2XRK3W_jL5dSCxpUx=P_@LQDq
z0(3$HY!`knNx?{}9}{G6f4S6Y|9yd!_bG_3-;c|UeYiLVaLKC|i!D3p@-kg^)8!Q@
z*kh4`dk0!xLus%&h~ELqd5d_9<GU%1I=uT*D7CL%T>F}><g=D^Ww5_h>TFK2m4qdw
z-zLuco22|r%KHq$r%8EV1V+O;b-WnJm4aWV1kT1awFLrx)t@dtHT!A&elJDve8{H+
zNs+zEpcMQz-FK-nsIMO}_CQwtH0aBp&ZNs2$eC_0dDZuKC^k@P3l&s^MpuL;<k&(t
zLB&}+plQK;gx~*o<oN7AmMXKK2u=87srb}{-zyWIRMLNk>i&p|?Io}H-orzDlMG7|
zwp%!);`SEvshi(2K5<mPv5K4l>@OB5Qs_B0Lf8uDnS<nQo?|1AAs*KdY(z7}0~(@_
z7&F9r4Y4nZ@M;KV9vOwKA(#PRhy@y=En4JihPZ>T8LrTvN{#=88dRV`XKGM3L8F$1
z(*Z8qRlhOb-yO3lCne{^!;%Tc=I_|#h&Q}0w$odd*n^}?ov_B+se1Lte~m<7j_5Ii
zx}59^vHV0R<0+JaVY}~VAn%kSxb2XRft4K`Oxc!AB8v+T&;*F6m!fcJS{soc5$|gT
zx%q8+0&Frm&?<zwm<bwq<oHRYM>(4J*Od1Z0>r0YR^H0{W^HQ90r7?xsiF>WP1$D)
zEzY)uR%J`huce-UjFvj9XL+U0z%^b_iMfk5in&WSiMcDf#N1nOD7FMQ^l>P5Gw$iX
zCEi$%+xj1hH#XzW{wJVC#uD7%e*;BuVs<m`^S=cwPR*9!cK=_*+;hYmFCbja%XIQ3
z1wU<z^py16`3QUNsV}`$%o{UWitGwa`}I7NKzI75!oX1*e+o(M%125V^TOPX<?3iC
z7UTCCMCNXq9GCD^YG4dfrDDwwn9-CJIOzAmxn_bWMY`j_;K_8~<s5fir@f>A4=O<Z
z>mm5{ka%2YC3*li0WiHwKECsV3gv`NIo^LJV${)p20E~R6jBWC{z9yQUGTn?f&^x`
zcW0V7{&l1e;0GHrt~zIoEp&|(?9GIBbH=Ef&Ly9gU7j=M3vp?81qK%O2mipCUwV#O
z{wCW|rRSgzhRv{DM9Y@*=Yx!vqgD(EcD~HUW0gHLPvTDUZ43^3=zf|M_Rvphve-j2
z&BrQ~UAEwdFr)PN&C!V1ekSN_owsG%I&a4B_UsBgUl9Dq5P|-@PPyp8uHQp8aV43<
z$@twti7EIkpv262an@ObuyeD{dNJjZ<6kQ<gloZQ?JF3AxFHH1&AO<E<qzUnI8dd6
zijX@QIGDzeX*!aDDx?1_%1;HquQYz&P|>ehzRDPUrt|xTfhuG0na=MUfZS|<TQ8p4
ziF1roFq?Z#Qg9c3y(aA02E~puG4Slld!V4^?i0tqh$MQoIQ2!8So^ZAWS}LpT=`r0
zcBmB22NoP>!r#JzAEd#YH$c{VtQnoxmqI4{c>jdbQ2Mivlpde`pJYj&E=~Chio;%~
z`eq@)?Ja1w6U|oTO;8nWOL?_4RK7hOs5`f&D+hqHZ%a>sd+1HIm-PDfEBlsdTR~pl
z+`VgP^9y_MU9hq*mVU^dVdIU>)+x5@38Rg+(A-R0XhAk#SV$#*6NArF+jl@y6t1nn
z;Kt3`yl<qQZ?K*|CFQ*-mAoh3y^W|!-oj?UF3#z<rMw|Nu+v_0RD7T(^8lV<ItoKc
zTSt?v<fQmRn&{Vc&i(@;LB08ZXn(OAjHz~e9yGj1>N%Mjpw*@wH7GmPNB;tv&e@x#
z&Iv!GLgr)Y=+`kTQr@+7p0}CLl!#gdT!nI}O}RO9erNh-n=&`Me<BWfQq9M(d&C)$
zdf)&UrQDoJPo*0i+73T$rUS@Se}Sru+AS9ZgT9nEn8IL(OT}@cY`t^IW(znXxJ}$}
z8J!zB`}KxN%Iq9<JqfP#Rw&mj?96$pLaAq~8+gG9IjR-a0h4KqyrQVa^E&u~wX-5G
zm-VJU1(t%cr{~DzlsyMdg0SOlMO}9&d@m-lQ2+=+YaBe`iVWq@@k7DLsTZifw$Vjg
z_}7%Tq{vR&)<Hm^fY&}?zwkx#G24X)#CGb23PrW$?L#SX;n9?LX}u9Vj0v*`%dI16
z&|YLu;552Xj^$Qh67IbG?QB6P!*gg)eaWtW-%iy_!IP=J<=1si{R5uAz!Afk{{j+=
z)K}~&C&7BFCd+F$aHFTaJ{-jC0jvAKiY?fq=au3(tAO7Cetr1uhaBCQxC1R8RVaIC
zDOmEk*#1MXE?2_G4+OtH#anq`PN)6yROO}M2Wjvc>3EoUO7NSMz`4P1QUhm@HmT`Q
zV&Y`*>vZ3Y`JH2)D(zgHB5fyH`eA<cvTu#_mY&d5ciWe?;GPq37X!haCpunV*Qe|W
zo=j<hT_1=aR=Exw+2rH0?Fz8kdKFUYtFK{{m7wpV`{^z8n73hyZlR~S4RGm_D!f`T
z3D5r0Tvd+KZ7Jm?bs`?WkxC8*E{6BYq{KN=r~r;VC!3bum+i+{6mG5es$)?Xp)<E<
z<M4>`&On|fEmn0m<4$C+ssg32C$WC}Lj5y6q*8KF{7E;KK(YNZG#}{sX(+{}d?_h|
zw$PXyTZug<Gmv4!=_(xH@|^5#lhAP5F)Mr3qE}HZR!3vv_#qqfT>T1)kwVzw*wka?
znWhY}!wlY1%0983_8hhlnqa#M%?n(BHCJF1_7r23R4KSK-KHQbyr1<J=4(Z0YL2Ar
zx96!+&mfG{o~^9o#PYC|vPXOX>jOOT5xz$JNl#{vw1q|`!nJsCQ^gq>yN?){YQUH5
z5r4>*m%7E;OoA%As2V0zg>uTC_pwd+tF0%3>M&mWPDm*SA)y@-W|Db2gSGKf^#D|`
zp!3e>(xc~Szj%c##CbUtw8@n^)1S2AStWSp<4k=hrv3qC&_f>~q8;(8cx8almjVv=
z08l@n)Xiu^xQ~<_%J_Dc2|5QGDKz$qtB|ILzZH8r;Ec-%MxIH+$#!nn`dZ}TsT`@Z
zRozN2{z)aTwN$gb{q2CzUA5}lr~pNuF|h5Tbd)VvQ9m>CR`on0$2lgnZm_W!rGxD7
zh4wmUGh;9y_GBve=Cp$lkJ3=oKRT4zd3{95#(wbrRcI33XJ|u)dKvT-t|JT{#R_Fi
zGRR_&ktLK-&htiSV&K|;Rh)HWqoO?a-;jc|H>9DhkPcEkPy(%R#otHiBpUjLIG_)G
zD5mc(*xLvNXsV>VjIpGAg~_hnpqULS7)lrRTjDtIyo$B~<I*H0P3pAZJg5xEZg^%I
z!z@Fsd>J>^$SJ9#v{HwDz)IOd*J@5{v`xY4-;!-p=4Il?i=B;Py&2+c6SH}Ak7yHk
z)X;e7d_G4T{SW4B1a7;JjM$5ymz;YlI&=0)oewes7#1wv)pd9RhCFzvofjJ#S7OKP
zG`Pr;!(#g@;6$9;SX`acs<HJETks<^=XVA8*-QEYQ^?9u;v#0{p3m99iawY_(_Sq>
z!TyUdG{XM@Peh#c?0v{)hEJWym_N^2Mtw;Xs)>36vTEc;S=mz;$Y}$aadwCu%h3v|
zViV;k@2Q`BLsH*Klc+kUVS6hn=!usISDoYK&7ilz*vO5WYhuR|S~iq?D7J5572Sz@
zUMN?ov2k!VHJr_bU_(m4S%IdF$xIuVq@*_5NUQlM(m$D0aaJbD$@8dVH9d@FJas(;
zFu%_w&A<wPiBSW;f@Fk-Kt}XdLi6U6-9CboWvT{GpI7Htwy&e+|L#St`4X4gO^>M1
z#AOeu8@Y8rF4p{GXg&?>JZ_r`05kEkiEpOH_j=BkTpW$Iap0R!ACpKQgW#ZEfFhbc
zZpUd93Ci*Ykf-V6<UW#hNFS5bGe|=a6@*4nKZU4-hCoJihCv^+=`*QsW1S_1?xY(@
zuo4oMw86o2v7;8ZMtd`UU(6>No3ERCiBy|zv1}iu=C9vP7J;lNyBB0MT4FnW$H!<#
zrvio@r6~LxTw<yJFaq_%!>BinKph%J{o@g+*AAn;X9ViohEcB>fx2!Ob>j%sHw>ee
zN1(oL81<qNsIMMIZ5@G{I@RFl!pwj?lS9iK^_snKbbkfLK=2N|%mb6f2+k*hmw+)4
zY{Q+$D8Y$6Ab0{81HscL4Fqo!0lg_{AQ)$?VjB@eVr4kpNbm>|48{^1FcP#9!H!rV
z>6NaikT(#)f5j5~!q|$LM9>-Aik}z>CK5pqnDAm!K0Tp*NHsw@JUNl{G>c~?a!2hi
zks&uMOT>1X;>o7e-;fvnm}aGB$YRz^V*6dgnC%>a8O_Ep?PU{M#rA?>%;<1DnPlsc
z*N;{EuJG@WVjL5Aw07Dx#x{4w2uv$9R^i`}c?0WxI_r2p(f-F_d}%3{Ojr-{`f>2E
zEBx^g$5>0-zJD0U-;TiX-!wkqPZ00ufcw$5T#eA~)<*%V%XYFs5e{kk4xi+)8wCtg
z;5p)hS#}WF>Yq`DsnbC4C@089tUg2p8)7+-ts)NH#G!=<RvHPkHDn)U){sbvRu}Cv
zh=*ZN#;%pgxN{pHvsR9PDb=hJ^KW}l^AV(?vTHR@LSgkiWRhR1iE%gW^qkJWUahul
z!n=6HxgDJQ*TmIHs~YtdUT2Kw9BB=nxS3ijx`^W>#*thsbE7wQYwFxYtuT1!|7{&l
zP#w<@Q`nUMMqSvlM$adVuMgAe)285#K*Schpu%|>UteICy1k!<-X`!^&^h%(TEk!m
zkB#o9Dy<)J;+0^9@jGy<DETB%N^A<$i>Z;c(b#x{MZlVzTAq#{Z4GmGoPm`2by|SV
zM<qM3xZvPyfJ4$?Q#m1L`?n;_naH@Vvxo_UJ;NdOb(-543!oWV(X`CN33h4d%58Kk
zv6apu)TLiQkoCK%vC$nJj&KYcx}u@!ZqLcZ#y7X0)~G`}zC;Akp(5TtitqLIp$he?
z9T>k~Vry$4XfP6-Lj+5J>A#9uyZEIVdqpm$`h~;gD?`!#052b-fxyi%Wv34>ds$Sn
zb%oIh-@aWNFONaGq5SdMns6o0T!&0j{c6PgJeOWW--LEkr)R7AM1W4u<ek0|IeMpO
z8ajOm7#cc#9`RqloA~epo9K9rKfn1dm>o7jmP{P8T5yc`xg=h0$6FwXAQoP3%y{%P
zdhV;HhwfVJ+PJui&XGgYcbvY$g2QNa_VW-W<j>x5o(4|^9P(%GDA2(30q$tS)^a=T
zNz?@Zl|2lBxnV+m=#q{jYtN3Jk738FHW1I?uFS|QEd%C=x(1*j|5|1R!d1ZHeqD5a
z(vb|!%@Y;MR~31OZ9OM5aPG3Uh2BHT9-7Z=2>1u+rrZ1AK!+?e|Ivg&!o9+A%v*=1
z6KWerQQv?Jw`uS^5XapOoGrN9Bz2aVaM-x+mk=Af?RRkCX3y)xiBuYP11W(bTvKsm
z7aUA$sj-EMa)`$Y#v?Ev<#7(L;YdIDsKS*-+$;@uafR|_{{?h33jX~QiK1V`HM{>b
zTqpIXA>Xd_4NY0?c>|`O>Z(c{+re&Cbgs_Sjs|cb&^vDu#K%n{ITz>4^KyYGCIq%y
z>S6Sjjf{nD3LeqJ$r_FiEVj_XTwCad0x9n$?T!GRk$H0!J_&J1{Ry5?wT0$c=phpv
z2Cug%3kynnM}vz{+KXF%2=ujxJlu~qzizaz!P)<o(sF?mn%<jEPx1>mx|T|I`c(aN
z6<eWQJLar_pRdMc8x9DBt#=@$`nR#uc8nLzE0ht88y3t(%mhF>KJ^^4<_3HIUGw2z
z&9B_?4h~v1;Y=jlP7>}-SP@PC#{4|WAMZr|vkCL3QT{!Y|4=mlUzMjf1x%oGy5d}U
zBi%1D>f3PCJam7NeIG-+KXlTLyR_}6P}q*)AG#wGw?ZrOFtQHQeHvTnE({AiGmLF6
z9k-V3S+&UgHa%?!e}@xT`qIV!P+2^8<U=1|wpElI56}>?E4b7Bfx4M>(_Q?Q|9v=-
zwJA5j8q!&)HM^qImD%5F>ukY{u)=HYl=8m8{R?^`ODYNbFQmiglEHu(%v2wA#AW1S
zW%i#7+eFe7^4b=FhNjCU5}u@>N4ieo(UKGPlw-EM;{lwe_1cwh)I1RNr_uAEJ%>|m
zX-)9gQ!zuMLy$k4son-UvNuYkbO}sN>0;afE3LM7UShh?gioGyPAf)SWzRbdjr5$v
z%>N`)edAdu0Rmm1pSRnUk0Dz*ctj?=gV~%SmGrlK%trf7`g5S@v8X-&Vi)EFeQOEx
z9Z6VGxbfIv0ciLYl;MkW)>sn4QHaMAzt|rmv)pgF`u{@^t~SkaH9H%c9KKue4Np^j
zi{DjeDy^x!(X_(RT!+t|1{wsD$!;<UE7k<?W!BS9J5883=L)|tufEaaMLg|)9l+NF
zZgCPgE^jqWF|9FFA4gDCR%*8<&t7CLEjQ%m=H|BGOOZ2+<Urn~vZu}wa4ajhDeuyP
z*1376tFJB)^7H9C<ySO1@P$aH>xy|kcb&`u_7T<m27z$1X4y50*&5_t;&S4{qa0eb
z$mM99JJnQtjp>>wWv)g^*E#qatLgGQ)3Q~b6}OnKHjT{VCSet0K5G{HU~Rplr7<8o
zy!c*f9UB|`Qdw(ZDgFDr$ztJJjgA)+78FtS#kyQ^<5b}Jh2}DRpV#GgJDSTpO-(J$
zZl?nc2KC0fYZ<EnS>vnmGZzrUYWDqIjipX!;&STsdvT+K!hT+!(ln#8IbCWdVes95
z9oE2M*8!bxsfFcb^Vr1W0v;!Bzj->z8*QUO3)#ASoWHJTT>ui#_#=E<dwywi^VL`5
z+nwObJ`|kqbP7&Svp-<M$B$>t(z<ju#^N=+`-!sL#d^2O-Bg#0tf?kk@-U~Gylz}q
zf&vk~$SdHdecKqHPIcez%FUy?R(afY24<BGpWkJ7tii`>bLrc`rpqs945tp4;L=1m
zs<_UzqGjbYd?FQ}(51=E;XoaI0k_Q2va$i~VY#MD>Q2WL#+Oq~j=DOZ%kMW`(%9Ha
zI3Aq%x-X-Wy*`f<1RkG%S=~(}IleL+9a7qyGh*aseC?U^M=F{%E80oMftUoHJ51S^
z<{9;rgjW^jR<&o5%eiW<&(oC46ss#Q24MYT(EC5dZw|g=-5etswZy;3bz2KQM=fJS
zpg&@&i2?psb)p7uqbuMVxmsEerJS2~6@7CaFC7YfPemeqNOwIQiFBu7jv~$;B?y&&
z0v_l2Uc_C<uYD#G$xRi6O}N=zi?|yPOLZaMcrX&#huHLXBqE@^>0QtvE<kKSY(ZRu
zxDv4!aV_Evh&R3$i3}hf=mR}%zupHu;%vn9O@lVXZHPA^ei-opUJ2MwFkV&9%mDm3
z=n+?9=Uj`p7I7<LFXD}e+Ymp2xC?PN;sQJ}aS-)c5cdJD#iu~%TSwkwphw(>_zA=t
z5qBfrg!mxhF2n<fyAfwXAAN{Th=t>kNHJm);!4B?i0O-m7Q}6cD-l18xE662VlU!-
zh}#g;J1Jebg)s=&^flVS?H@hwNg?I-MM6qzcFMV9MrUqJ8J$h|^YGj9pOMJrM8HF0
zdO^rRKG6^U0Xjq92wB;4vnE|9p0PT!O}O@)t1h3Oa}lv18VYH*SK3*Pqtob9?rB=R
zf>4A!;#Gh#apn2gCC_fO2&GxsbFwDQ$uiB!%AJ!{P?}XdFROcOaP+;SIy3G|zd!As
zR90>|$`#{2{sgo&y{jovx$>;am}2bjf1nuTF}7d(yC5WM<{Od7t3&`9f@l}Xw;R81
z;6s4Z-}&e2p*GRs<s^*VuQ9fV+EkjA8`N8&)%zf_2mTg`VCY1{1f(|OrzXIYJbel9
z?*M-=4o~ubiQj(UFN~2Rs@u{mORSDS&h$hFc<LW976}1;P(*T-17C?bLYp={P(90m
zfB3CP<OUiK!}~C4J!B`k--$%fjnP2$YzBTy9G>)A4gOugHz1v!*=!N!WM!9TO)AeS
z2&Uafbt6#uPf&gkbo3EUV}CwKG_Vm1pz!oclpiiD##>s9QmB3ZjdGJ<OMZ=0yYyej
zl`G9En3vU;8XR@+Q0Jlyl}krG8~d?iLHYDCUM6pJe6w<GtXyaMeQEb|`DrX}M*4{(
zk;o#XNyd3uT`4psDNy@J9u@g$W;77~@4(YH4&w1AfIpZ3KLPUX2cDj963Vko6jA+C
zfhYc}34!{d<I*fskfxt76(~FMYd(!c&Y_CX2gWv$9*pfFx$Xr0md%lf4{72@5%I}M
z$NWDUiKK#0dbLqMm@CjvFyr2+W=NmA!N>F&<fVFt_g&I)b^_%NqMmrYOhEnn6X2%;
zzb^s49Qf`8c={VLTN2;{z;^+E4#d^lP5QW()5qiKZ>$jic>MFgCz9uXIDLG(KH>c1
z@#*BVz~4B3NuKk7e*$=@DH;f01pFr8<N4TuZ+J4+K9mzZ`5xE@7jeP3xIfjJHR--I
zYnG`q-I|qqZ-$graL*_yt2j8iBFoa4Rg6AHPcwTDtB-M%$;LN*&+{PoV7SGENZQf6
zg|tr?+OFqNLc*^QPDFSRKP!GL{0-MO5PwZ$T0!IR{DqJ~KsOB}#HSNLFUgjH>@`yR
z&2XD&o%Vs@Ci1B|gFL*nMGVGJ87x*AgK-4h46`7EY~r~9E`)N4!4y(*0WOnt25T_x
ziNTpv<{1Dc#FGg4BMenO$d(ZU3p7s*uA?4Uh)Yz0)9^G9F)&k^QUE5zGYPm02Y+-&
z;pfC)BM1IS?fDG=x^>QSWMF7H&`3GA;X+1*Y3ECDDFTv(TxgOcU>c32<)vLl=vP@D
zPt-zzDP*?up}dd4pM@yxOz>ra5>S=kCxiu;`5GQCOQ-E>6VgfvBvZfwC8&W{5mbU7
z6=k6e*YWt(60{gUJU*#~NZNiOZ2_e#Iff<v3}7u)N=KG;dJg^3n#p?3{Q#($k054$
z-i#i}U?GzB!QBw<P%7z(WUyi&Ft!BHij4g^8_Hcq0NsAWw<3jF0*>Q{z7edX=`64$
zQRfj*2w);&8`X0?F2ulv(|Cd8qMU_9_6z_rkrv{%8vqT2CkZ$RfVk0`PdEduqnt;n
z%wzym=2rxeo=I8cE`{Z|kT^|L4|N0O{GQ1E1Rz?^E&#-hmW{$+0Z=_B2{;#BO*x-a
z8T#u=ROU+p=%Y8ptq!r!hzr%zLiGfB&Iuyh#es6lc>};}#Fhdd12lrE26{AwjY6us
zlYrmjM*!Wv7JBfb(eVlaG#c-q1cgZ2V?x?F(~&C1?_EOOir-Hu@qPGS1K(@ldkuWA
zf$uf&y#~J5!1o&XUIX82;Cl^xuYvzN4P;NUFa{gX(&8>27yLkj2YAegaMAZ_@o_2+
z5Gd&QF>x^cZ9ocRv`BdPObbiVPLM+SO%@iP&92xg-c`#|bP__r#ID+2o02Ner%fjX
z+M!bT>Qux-@x!SWirC2j#g<G9MFJgmP*77X6h)6FuAw6Y1Ui19fb$j>XghBV*!OCo
zt(L{uVzXf4?V}Aa1$}Q!dua--qJ_meep8kP>vA7txkG=eP_O?~L~GB$SdBiyVO<U!
zud(*7<@ub=|MMvNckJV8dXx9-M?5~p<53ed`Oo6<r97U=<GDOu!sB`#uj27NJbsAB
zzvc0>Jl@OWH+lRKkB{+q)I?rCk1yr%OdikW@e&@_^LQ1H@8R)7JpL_@pXKph9>2-s
zk9d5H$D{bXI*Z4b@^~hX=kj<7kL!87ipTfxI8NR4r$%R)au=^?X%4iQit>x|3#XY&
zS{P`~(+~pB#b3|G!#u0-r*QFSPEFJEXXF=5E7C{{*9Lqo{=k|-Q*OYq(tpJwtKC{!
zWtFGTD7d2Djl2GNrd;~t!F48-wA+_dXvOnzaEL#N8gQ&Y9PnvzgPy`2YM0L|<TrZ)
zu6+E7(`jJMp_R=o`72u7jdj!TEJHp)4Gw>UkYBf^8T_<3;L~!rz%hV~JfEx4K?Qiq
z+ZYh?>2EO$`S`=HxI%6>^4QZoLcXg(uJ<{bTyjGl%IXl9;t$WdTAeO0?zpqRB+Yop
z4xbNy6qfPO)3-QL5cN5l+)kA9pbA{_S-tcGk&wT_?-%mvkp)+CAikgew=mg1vZ=bA
zDNeNj!U-8i==P5xanbFs4quD}*&Mxmb*cprKU73)BES&18H-}fNEyoqQY|bmz!~N+
zXy_7Oz8qP&9q!TVe>l~`h#%&NN(K$tY26-jv_MC~WRvysy4|KPJ5#9PcAM-vwOuc-
z+s`Gyk*(Cr>wZALpRU_|kgyQ1zng(2o2!@C?d)3axO}`E7b9AaPEQAzWW#k>x8no6
zd>1cI@{^pp{2VRZjs)pcFR%NJO<5YnK~AXGub0=`-yT<9_d~6sR#5j_RGxfueEIu;
zBc14eQ}^2s(|IleoqZcaeE%!RG?wq?{-B$eXU`rO7|^LS1wDQcX=8bPokD*dih^&h
z7VB5NoDM&Vgt5H7&PcyWqoBiVV~Cf36Edm1uK#SlPRZuwU)2@EnZ(!sL|pkkr-f1W
zIkf_G@7x&T>wgxR#`;@b7DnFc(n{@)t6%RwEf=@QkT8~~$1^C*Ze)>A99Ll+D7+3b
z8fr8)Dd_96eRpZFNsFS_`1;>N4h=QEyuNNzdHHzyc=%(KCHeL8^cMvw4E&SjkAjv|
zq?gy%ErYzg-m$tG_4@UG97I`T{rcaP9$beE1f5=khF~3k5*fzwLR+my+NMRVlxYZh
zIXxbQd}I06^|g$;b-h+dms4k<m($@fC`>rL{hN6C4ZOTbUy$<xCcPk!$Dus_$Na5q
z%F@eoHE?Q;mH;Y4VH|#X-3-Amqwm%d?AZfDhF(rb5v<QW9Y?YHBMsU_6(C%Ip9Md?
z{kr|5`JJf#4UZuMVJWvm8Z;!g4v!S4p_Rp{mSH6H*VlDFYs{p-#jg9$SbTbPzY>ej
zi0)Hj@uQ;qnOOX2LHAFw_)J0fJF)oF1l^y+;!lt6S7PyJME7m6_%VWR*JJTx1>Ii9
z;<KXry;!^`==L=hpDpNiGZsHi(CuL?emv}(#fW0*G~9&L?Uyk#4Yw_IyA+F`2s=}2
zM6q-l?h@;E#F&|eJEyvRh{a<x>hs=^ktUoi==0ft#wa;Q(C29^Uf&H5VQGG(3!g=X
zFj`9U%ft0Bm1U0?=y5w^xQ65D@i=3s1dxjRwR%6AcxEf&6v$cEQOhv;y!|1dMDmay
zr`yc*xR){fmI31hdW_2$ws1T>rezE-aXdYSWeo2CAUXB(3w>QS2t3J2uTWAzPp~i%
z0u!FzucV;kTahMw8^`PInvO)GcIkSi$DWK~Dd<Ta{T#)_Gj;y-=YJ{a{DVpO-v@jO
z9x8|*hyTv`x5b^$&~vy<4{`Wi93Ow4Lod`*du!wRQDyky^Fo9%CjR=l5o-i4=K`OS
zV16ZQ*DOXqPAG1)=qy+)G5qlMQf4Ck)58%d)O&IB>lWbYp@aB&eFw+K+W~s+FeSmf
z{weUL1oM=>%QBv-hzqYUJm$+%3(J8TSxAKwN<@Da!(%hB%);n&dP@@eN0Q*XlHhj(
ze<4<21GO4|Xo`hH9IwAmp|2y3Gd_@quXE8H7Di)Xm`MH$7=El!88_Z80scaock63e
z`8IaNLncZ1C`s_Y;r#o!KK1>^0OOB#$ri?WGUtCLHZmrjHbxgE!OsLfQM)AINq*Da
zwTunfK?*-gLjUt5_^u@Qmy_U6Fg)xA9}ihtwL&2lsEOpXGyHIWx`^S&3bk?L@P|qG
zY)*oI8F(5$8@Zl~xjcQGe`Vabor1%H@vslIRtux#%QK;x<J+n=JUvTGp_$>ewDGzD
z_=~_te{ToFl7;6vJ^hV$V|X8UlM%tw^z0MKpUua=KF&>8C=&m+85TxK&!AJN0)8CP
z^YM>i#=;8Vsa=*EEexh-T`ByO;kC5!`YiBN?|!Z~T@NoOp??>6laT>W9|t}=f&EFx
z!j|Oej++<N3_qMv8>erJ8xMZ~p7fx<7p3d@T~2R`>)-LGCmx3d3_n)b6eo`xc;fG!
zZDB&uGx!vC0#Ea+jnA(d4J~}h`FP{xyby;#Y#j1_EavpHfhRp=$H{ps$M2UcjFO)D
zrtk~IB#-|77kjoJ_?Lkv{wB_U0;3j$4;dd07^1W>7$>P`V;@v&z+$fpe4=@J1<|9u
zeZ0N&`4S3qlhEJ9`M1UKzlG7mzbv<~@-B82gxfj3_67}q1;=;l^c=sM<6i)tjZfa*
zA8`CZPG7+3xtPLn;F;b^HU9cIp^u?ad-eAWIUB(OJkjg#-|6(NjDD=pSFZ8l(@yv?
z=cAX`V>%~Gq|Xc(M&iHeDvi&pyui5}pIxKjb^BR_1dWsZOEml<PG18&>7kPAK_6F5
zoKGzuSL|7Jl=%_x)Q{edTE;=Q55HslIa5Qli{VjkTzd}zpAA0o<EK9fpHU*o3Gdgg
z)vM2!tAJ0WpJl+4ek{DdF615A0ldi|pPqa)37@AK9(>})!(Vhf*N@)6=fl8~oV9Us
zRsm1?Z?jrhmGtfgg&zS=dg$(~Ww3s^pY!SFeA;-%i=2;lj)h^^yDCV1#PAFibNMz2
z{$dzp(vOLc6a5^*0X+3DeH()X*k1~F1D{CGPjGq*r#CTbJbRmjegJr~=kfi1oYR}m
zvTzPsGzpcG-YVntU&!z=qztff`glKedlLEwlHj{IpUOBp{B9EZV@dGialV4AVc`;n
zA1e&-dC|Zr7IM60l7(UDoiYm544>fqfxTLb_f0*%HF!0fzn#ur6D=!?xwBJ1_sU{0
zcrP~&GhLA9*%!_!wac=<Wrb{$lX91rRt4<Evh4P-drpjwQi*CV8dlHDVVL4!n39CD
z=yk|TwKuHo=9#SSI_;fd8Se;Ng`vzVV>0O#cI!X;nr$LY(9JFVrSItbxK3ZdA85h1
z5uAc7UstwBw%e+zWm%|m`S6OXA2$KzK$Glj#Ou3$ApuI3>pb$xM$ZaIqg;o#$^EjU
zrIn;)ud3GN&n&sJ5S$Yb$#joTyFVx6o||utpxxuEYiVj)gCYiqy?q@+D$8>hl~!0~
z>w<C!t4Yq&d<-m{JGaVOEmxP$v0I6PbRT6=7mM%*@M63mm*2Rcw8B;<&<#PE<1kVR
z3uGB@YI|18jjraEfd<*-^Lc!g^PLvUJiBd9Sxt>RJ-;a0Z%JOWH_w>Iv{cp7q<v_>
zT3P27yguA$t>=v7!u-Pg8Q;a|$~m~d>2~?7tBA5HU@peZUAGeggV#!2U5+N36L>Sd
zF)jyu4tKzBb6Tq$xLC`~)<yHH{K%dz+dO`GA-+M-gtxkfwWZ3_=(oCK(U_@zS0G;6
zY8(GvK>Yjn^3rHl%%=oo8Q*7c%1(a^8w0`$hx1m5>C}bL^`Lh_^%96X!`R9dE%nKp
z7%MZ)b@NQU2Uj@iWT(R)SQSg;_4wVbGMeXT#GCN(4b+;Kq+bgQW8XB_-_o`&stU~1
zJ`rGZL-%egP3`Kc5t)HRUfJYn)<z$V?CVy^i}*w=YjpVin1$Y!fInWu>T0y|Um6Ip
z3Xqr9HCDzc#ntML?`xSP@CD?0hr2Oegaz}f0yCU}HC`9S8AexyjBky&n%Ts&#<I>g
zo4;$6n+vNN-1Pxh-6BWxtr&>zMeZuIEUa|lqaIBzPko(ZO`HJikpbO!Ch4pMld#^c
zO`~ErW&=GYQDlRqz%+>!86Cq0EgO?^lfyl1lvFR#y4LU+j_;7B&|Im1MWM{&@zHk*
zsI^vi!lA<_SqV9XvZlI)-T<Biv9`iS`>mCgPH2ZFc2uv7PE&rRyAlLwfwR>iJMe(T
zuzuzKDDDdyh11I#T+UlvzBwL;4<EMh2k<lrtJR%QY30L}W~;7{hg?Z@MHxAgYWA$g
zkabAXSwpTcPOh@T1e2?(NXCCbwk^cusdG2W^xd1NHj-44kkgPolJPvt3fRH8krtiB
zhOVynx#G<iwO#j^u`QQlJj{p|!CVq+Uf72x5;VVZmAh(&%=^V(1f$_K7qE}bz$19p
zkkKdmA3y6yth*%Yv8;=3iK;TLGt5^S2ML?Bn5~*1?6WjURBtYt<MTM`$Qa+?@io?2
z-Dm=vitd@P!5*J0ZU%n0XEYbZ>dHz=F>|FbS<Hp*=4O`<)2^<?8L%y=lCR{S{h{<s
z?V~?dySvy*p(=2twvcJH)j@$o%|97rtCo>nis7~t(;kY()M|CCaIZ3(uSA**wM>=}
z^RM0mrDKhNXQ*UzKGvvwt|d{6QRg&#b*mDm^`W_uj89a%QN?N+m_hc#@Ld_Q9SIB$
z22)hj=rEt@u{GDZS`!REl9ao)*v|%`xv)x1j}CO{@KuwHsi809u$=M3@`5PIaIFdy
zIq}(=*inp+*EpPHmg1(VT<7=54cJvQx^(MJOAcBHMw@metOP#kQ>QJawDDTVmfe0V
z0ohk^I13;2SON_7s2r%I#aFb&!xui>SToJZx3#!key0m-?mCy>P*Yr=nv0Ax$%Prg
zO}4ki=aN^seSsDSyl#TA6wj#8$OyxD_&S8k#~RGPSX6<{2enFDE_t+tB(y~iKha`~
zMuB;{wkqS)m=@IqvbAcaTsgmLx~wk|v;}g0iGT%GlV?@Do#YcDI?BomH60C;ThGwC
zi76e<1z$JPyGV0)tZVq1G|H4sm|+|pRb&{kuFz+taqt$?s^8_P^J`_@MP8Td);Oay
zRFKUpu1c~J)CL(36-GA_3#vRb>Kj}94Z~T)*$~=r_}$-WMydkQH4AAbc2b~0!&(d5
zMBdo=^)YJ`q~zbBl5cgbk)3F90P_XK-9Eo{QQQtd*Ha-|(912&$;NgPvs7rN6%MU*
zVwOHhEK00tEi}vWs;2MOi(wW=*GabK0N-^js`4jVUeTA>$l78|5A8ItlS;DFNa9Gb
zzJzt<qkujE23IJ!{oa;lr*5?L%?%XnDM5em307x?4m{OzOfO=}(0s)11&iA{8?EyE
z*wa=O(*H&CG1ss=o#}&#^EAm0Cw-7hc4Ny<pLN3%u`8pav(U_JT6Ci&qgnOXLNfN`
zvb6mz?T#FcjUH!$!KAGzFf^-V7rtD!21^?+#<s`jSm}z>2CO6_8NOF^v1e_u$7yYq
zy?B<{gQt_-;I*n?1iMC!2ifgd5J2AM)lD%vGPE(_m%C`gJ8UKh^5W*zP$`y4{+1>(
zNH|NHo?k%BKnx8y8eIvNv;VyBu4k+~)3I5z4nHWU@)!A6JG_ugUt97+1XxI?*RZoO
b8_pHjl#7n&O2?W;te=caf7+D@nt%I0!DXRr

diff --git a/benchmarks/Grid_su3 b/benchmarks/Grid_su3
deleted file mode 100755
index 72d8a8d0170e828d287c6b87996116ca96fb2d3a..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 44845
zcmeIb3wTu3)i-`}gQ$=R7A;;Xqeh#E#sm<EicUxZXLJH72{Z^e8IlQ%B$=3*#NY*k
zNvzW$O8eTDwzj30x3$(*YHbT@A>kI(s=+H>NxhyiB2f_{73TZ>_St80W-=4szW?X{
zJ<s<%Za8P{wf5R;uf6tdoxRVQ++O0Co1B!S7$I4?N+DMEb(<jRcSXr=7pSDoQchBA
z%4FqiWeg~3xNIVYP>&gKNQyRKy5vg%OvN=TStra&wuy)V&(_H(K@Wz8ge-43*kn|P
znQL^^fQn)j1m>fzdv6mcV3mj%u$u}Y3|Vw&<fD4bHze~7$$SP>r2Ymp>c_hAs*vdw
z$u?lbZ3IfOxDP|}GL*gRuMzQj9arjYdPKtY(hvqT@_h^WsOSGQWJ<<`vb;og8Ezjv
zgDhUTw9d0)#`H_;YA&hsG&Hwf(mHF#B{Qby`n<VQMZGDXyz}NS5*ZCUP69@qC*q>X
znSTB|xBPSR1-EZq`dZqQ-B0wrSv>jeFPN5#%V@WNE3?YgQ=CLv9<FJ)_O3hmxt~@4
zWB1+l;k_SaUj6f~CAN>>T>5=$cFBv5i5tfiDXY?c`}4CKx2&<R{u|iihTSOW)Ho@N
zpkzGy3s5K?{WDZB9=$n%oS!71rzMa-Edf96Y>lJ;<OK9N3FQ1cfgRWx<JD_k0{Ui@
z8;`#|0sq?x%H5WLe^~-L`-2r%Y<MsMe_sOn0tm3i3IV<rSG;;XnxMUYmw<k60{Vso
z`g|h+|8ElT?@OTP(FAlgf&D$;x8jNoPbARi#RU3Xnn3=W5FF2*o&<8j3C7o%3HaL+
z=(#k3KDQ;1e`^ByR`8#zoT;?kQ6)@NDGhZW<dd#+MAO$x{!Qz2dYj}=ll)aN{ByxS
z0dj1eI=@x&uax{&S#Fu6ua@;4oM@929c+_w`XzrR_!xeMi}r6k&4&AM*emtymin~G
zq*b7k-+H=DQY64S;Pcm1Pnp8Ge}>|8uB`VqIDLLsli%r7oY%BC7rIw^e13ORQJu@@
zbNdvhy`t1v<8G4ViqfJwZ-cwSwW7|gr^QckHu_gJxm`8R)$TRUYQ+3tzixiHZ#qh7
z%+GUDsL$WzT~o5~+H#k@x&+Epx4N8-%}s7+i>Jxo?5bmqYs>wYRr}X8y1|k^4cTt;
z)VLd*o(7M<q(zAIOmB3%R|}-vpFh*-M9JQpoGVbctFF#lt#f%VtF3GHtuokVnAxgb
zK0mpT9ds|$a=YqLgwNgNw^xIjU%krJ<n%YWJboYaC|L;Fv=uI&r`q7QS2!#3oF(Nm
zoMqRRPjd>@%l)%FULO?kxu@HqZ*jiV*gpemx$4{zg0CpQr0m-2h$UEU-^^-neWR<%
z%@n<&GdwFBydcPWH~C5ydPIspf0nKdRQEKj)GH8Ab7AQcl!l^wtc24J4Hq=J8|u9^
zY@v$9^U*kHiB{JNPfI?{S<Rl{WD~$7wO%w-q#dW`Im?~y`bPg6XOp`T{m9$oTIps}
zu54{}HoBX9-Ue5l2gxmYGMn!vS0n6)j)k(I!DVK}S2Wiq)*3m={nM+x&Hj;j(OwDi
z>gsyyd?hqh!%g)fb5td#<~yA&^`7_@^%kJd$*xetmiJa*puhsdjaxibwupCHW3#`6
zja5;Bs{Kc85=SkxgD<XD<$f4mcJn%S!%F`ur@N`i+hlZ<Vbw&xLvypo)YO%=D3}21
z^|6oG%Ia#Tk66CqT-0!rr=bRe)7M<@CR?@6*68(lTAlUodbH!p1_c#F(>XaeV9Ka<
zdFqr#bUA;m9KIOmtscMP^Q;uz!sqrY$SXm0;~Hh9+wb?(yS=qFt~F$HRr@{O24#h-
zdNsyZwVUkq-WK;T#pzp9zXA=25~!A(t?FH#2Bo&i?N&U##^#1<zcR0=$T=-HPnqYi
z&na?F&CQ=4iB8SEOpjMqf@OMcfl^diSvbe;%+H+`l@QIApKngY6q*b#I9c2YF}Wk?
zu@v~pMvOG3CF9vg=N>E47}sTt6}gj@bdjEn=d&^Rmn<%F>9H|N68??Dnr8ToWf!hF
zMM17NKx0S(XO2wI1Ag!Keu&44lazNQt@%vPNz`UA!k>mbx;*9&+@1rK#wuS({^uWB
zjmNby%E`F1%ubP>rko~m*U*o^pQ@ZMaevPD@d%ZyOa;y`bbFP6e6~TskzlN&WTT!&
zFxExV=@S~kSf@$nGlUV0b&>ST3?iXi|46^oL^sxH((_GpW4$DOripH>-=t49(I>T4
z32ZFxa!D4r)kJ688X?<6XHPIfo{27-L?q2J(OI?;Y$m$-c~Lddsgn`POmySfmT8qH
zy18AdOmuXtNLXc}kBd;@f0*dwP4re1J;OwAGtn(3`dubE`<W3onCKG>BH%_7-Ry^M
zGSR<g;=kWSpKhXWHqlQr(YKiBr<>^ACOV&mjj-25Kf@pb_L=DLDkEXPiGEguswn*?
zy0Le{WX(jMVd5V&(Z6A$51HuanCQy7DuJ<&@EOwx=_b0>AOdEZ=;xZ~lT7sUO?0b?
zo@JtEo9Gvq=y@jkg(mte6aAYey3ItNWumJldbWvPW}@eq=#?h=MJ9Tci9W?dUuB|S
zY@#=s=$Dx2YWs)jYA|h4omEjgyZvd}uaT&>?@HfkVnQ^0*&MKhFDS;{GTDY0@l|9B
z^@kBIxSTkrm{6a<rxH&l-YxKph^G+WEb#M)b1DgK68M?KIfaBa2>evyoH{~n0za8J
zyF;i^;K{@}RfMVperz>xP7$Fpfqz2$Wa2h~e?Xj5MJP|;ZxQDd5wZ&WRpR4`XA1l|
z;+zsfiol;F&Z!_YcmjkAb`a+j5b78Bqr^G%Lwy2&m^g=gs9WF<5a&=2Z5H^ih;w@?
zv`OGUA<m&4+92@n6X%c)wF&&Y#J^6wQQ&tF=g<yS3H)Z_9MYjOfj1H7P!8Dyegkn1
z;ZUByR}kmW4Os<#9dQoLP^Q2a66cT%DFVNSIEP|r@Hp!~hq#q^zrZgi&Y>6T6Zll(
z=MnD~_(jAy1VftzejafSz0f9spGllUF0?`5rxNE-3$+RSWa1oRp+<ox6Xy^LRSEpq
z4Zt~cLS+K~g!o0oZ36#*_!Q!Kz{|GRbXF<a1*qC~KFe++`e++QNu{t)+4i<Bq^CvF
ziHdY>0%(?RcUvZ;J=B6x)0_4P0K@jaDhRqO!l<Dfo$=>7<FBI_HyMnTp>5!-&9zMK
z5OO%a%C>J|`JG76^&=|vqkZxZ+5`KpR|9XU?e7ehRoHvGyWKV@(^DouuWgb|(Y_7F
z!?l*l#fq}!9#n0SddXb?YWq-z8hH1n<mEfJus{y{!^^ibDco<F+(z0KVg`TDPzw0f
zzz((NgR9k~K6T$vGcrF)Mz!-fWL5)v^!#th{B5&6^;Sh`p0EfxwYBIl%lD)$1}W*&
zovfeQK6}3#4E}Q>Dok9oeegDD7kD?^w>)$UvREc7A)gKx6@I&Lap9uEigI<`hm$Z3
z1Rl>C2cQNX$$}fGs9h58(sAu6RGLH$QwTTtlt+_s(|!ZzNVC8Lvj_IrgKN|6Q?ANC
zs0RM6_WV1!)^e!3)^dACRv+@Hy&YNofDHcA)WCkVN4v_>IrvTFcfzh}V7D50P-g|)
z&n#WVKZ8PgK6`M6J-9es&G|$PJWuKVbTx2D>-Yj$4L)YtPS0?gt=6VIT7#mraxw(d
z9`slhHE7e+;Ksi8Dp>h7OUG>&DoWcevlM^YcZ-S!&QJpfw32_5I|ZD1+fGJy(87I|
zZG%Fp_C93=dbLXd^>Y53<#14I10rWyah!4{k1XenW22RGyDX;{GGpu0u9x#Vri$45
z^kB5d)TepWa=tFhaYoA7#rkm3(9Vj`_K_wUdu)XEWQ6Zil%VU_M;aF-?LdU~0%;kf
zJtZ_r%Zj7-*GB67kIzT55A|g`{4P??+brjDmQ$^l^Z3e<_W5hp2>UD=wVb1rg>tS2
zEaZ-y;^_KKRu$}H{|6B903gR_q_1yAA7o$e24vXIB~80cND@6<Fm#n<vSX8}w$ReB
zr`FQgt$b>^y+ri)Eg(Ar@2ePlI`_R|*r{FX#rV*kMMc!=j;u<S)!ALRodx+XQG5DR
z)vk%%+uDF~tY%^$Yo`u{!=2rZppJ^!!C@&GDm>C_v2=ch1;!EhP1Y>bNv*z5N_<W2
z>c~<-Vy&PRDr6JRxu~^2PmJDpW&QRj%Tj|6ilX(I$Wo$lU{jphmHL`cG6^K8PZD4m
zpzX+Rs8b7dU^^QK4Ad->OwpQIL5kI^@Ns4sMYU6eYK|Zz!YYOt`dI|!8LByw`l(VL
zNr2gag-3QlwWq8?wMif_Xg{5!D4$p+6lYBe>SnVCG<&<&W^ezntt9`6&et3P2;Z@d
z%v*W0+`SHi!&sAQE#5h8wZ#_iiVcXa*o+P^v<bJq&n(%twqi@Rx&hZFM9MZ7TXHJ0
zs>mEJwk#P!q;c@i$n|QWWyzbL6kGcK&9p-$mL(x{rp8y8_K9W5r%d|DvgC7!{rJz8
zB`<wq={v^EyI9aWD5wwOUq!b4g_i6+mL<EGf$<OoYnJStBw6}2OHN&u6(#prmb^!n
zT~xjwGWsc~`!mau7mzymNuj0hC90YS?4OWI-lw3vBdbgh1|jSjqBg2|coXj*en#Do
zK#5O?zV&Am|0Lw^mg)?aSSpW_MeXMOM5Kbdk>zvClASDM2u#PBT-L|*=OAPW4DbQU
z`%GqE(an@EHlcEs>|Koi6M7#)rC7s4OXWa`NZJYA1r=hoVoN1z4q?YEOP&PZ=inP+
z<p)vky_--XYoA@5RRuX<>j%#+OJ#9Zo+SrFaIQcLE_nh;hA}E(4BaG1*oTBZR^U05
zRFP#94Q+$$7id^0m~HE4wVnj;UY6F+rhNsuUKh4SwVwf@YLHCNk>Rjl0Q(DI5`E)&
zFue?>Nle~}Zty%(_CnEpmdckvzorjGeawyqD`vkA<cZIa{{*gAJ}E)pf@;T+`Z^ls
zBM{$2FJs|vK<*ppWK!D4w6n!~GV+J0{1dS5Zm|E0y+YUoHhLF?k4bpXQdz@J@dgMz
zB<#0TF3Ym1fum|q*s6A28CHAVPc61gam-e5ub7?5n-$SpXQOvbSvK2dnc|;aW|>kv
zyUH@9W_F`x%6+rjFv{d$4EH#aK2q21*Z`86#(Tjlmi6DkdRG|kDO~rt#-Z5xZQ$B-
zSIL1~1vjD7W!0C}-Zl+>1XnY$xD^MEXn{|-8tmQ{Sr6%d9R4&MMg=Z<26F-Xc{ch}
zWj8t>pmAl6&n;#d2EHWiLg37O;bC3mIV-VzVKNuQUs9wc6GQ$J%d(2BNr<PtTSNiZ
z_dy7*2_h|X8Oz&=<nhmlWFa3o28DhZI;8zg&%T3YJ`C|As&R>4rcr|>Y3qk2N-5W)
zM>BsDSnYvb_F!rj9gu5+4_XJQdFMe($8Qm7yA2*y=66eq1}>-__Q9D``xqq$vBV8~
zSBHdmg|$u#3blH*=V@guR$@}_4C4v<TuBitNN3nsf{`*7DW%{^cLc)PF-WJN=gx;R
z5Fl1gA;5-|!uCDMh3zNKu&iH*^>DzV4Gf0E;>k92G9Kb$y{vDDGSga|R=qozyv8+n
z#*6W?ETUqGyv9FM7QBeh!KIgT0h?QRjeiw6(=y_ebJoameg~Zsd#83;&i5ka412+)
zdO5$neq??Af$GQ9=ju_*`4?rOJ~@$chP}pjkyEc@WRz(8BD7(z@gEV|u-Ewe2yNJF
z`~hefmLDx2skddM-rqSintiA*^}aDu&fB$y*LaCu&Yf|}`SY10>@#`Pa(XBW<(ve#
zg-)XyfS2HaGq)*)4nP}7O2l8}hCKa6MLHI`zgQ|{2!By9NPm$`YS582sn*iagIU((
zGY&zH1M??}(R0N3j6X#+=re9)d0*x;zIkvIpRwobEa{P~R@8=`nJ!WIjCX<bAA4pu
z%EID!W|SA>nQbHqp4kRK-80LFJ}_V#2n<xMj7-%JO4Sb1ovi|CglgNEfli|q-*<lL
zL){LKL^+LvEQ0dz;CWDCPDj!pb!sFHklS*FNA|#L`~fzcpj#yf44QkETx>GoBhzqn
z7wzpIwdFri%MjVK=sf5Mpq#fJV>aR6Tkh_Jl&}29ul&ca{Ktg;<C|*F3HXoGkElKG
zrIuKxjGKvD#mvk?-mQq=Ix`RVWixG{_-B^kUOcl3_nMiFxZgLk4Z}?i&2X<HX&~B%
zJY-ql4xcgHV|b8@=s|A4!IF3stin=N^lhu|GQ@b0YS5oGNh{H~HpKXln?qc-uto%S
zS{`~0|5xukOWlZhC8buiG<4TmR-}!`-bSruS=wm=-U=EXFv{puz0J&!80A7<4<h_R
zhj1YsS($Vpe^nq{NVxuV|KY4w@yVqOUS$vE{8Y+`tlK1>4x(r09@qr?uAf*g!ezLD
ztC5t*4ZI&2gd2Dg-N19i{v`ds_gT+G{+}ox{vRy__f5>Ny8kCBVlklqr!S=R|By25
z|3NwyK<zo~*;9a61$F;VlneW+?cK>@F>W7%Be+I)1mAiOj^KPWYG?sY8)6;7iK6hd
z`bDx;O^)ENa2OHeIE@h{C2|Dsrz|)IgW%H3DW)TMKCGoZC~~Hy#VMzFWH~>F&WRmC
zn=I$9NIAofV6k4#&niaN=RH(Eram)9EoVPvp+43~Im3?N>&O}F2=0y0h8@92BD7&g
za8raf><F$04WsMb@{xMa8maf%cSo}i^`+jcBIUeYYdC^gdO5enDd$yefyLP8^ij*%
zLRlzB0o1n-)vo)p`jaq}vFRiC#`CbK)Bm$c7eV@eBCGKK1cUVd$b`L%*!{v;lo2>U
z-%sa`@%?@Q&FTB?hd27AzTaDKkK+67#(|z$(&Om633nbhN=Y$}o$j^W5-mw=ktN!~
zTp$Fg!S%y#m})86s|Mf+AL)Ucl?!3oNqFBwOv9PMM?_Ui1IJ1OecY5Q4ZK|%c*Wix
zYEy%^avnI;mVYqX-#VfOh8%$>9f2O<a#`+PO<PC&G3;%8wO@x_v#<8+zS^&g?AKu@
z&fbRGw>IFy7UGmOn{ivxZJBaopJhs6zh#PR&@$x=IUK_~)pb4Gii`HCKDDg>1%0aB
z>yB#}a+b~;ft3bcFAcn`&A|C`#Ip*OTH|_F+?zZ2Ca2mMzpAo794-!gn#iZxny33z
zJp!72s(X-egim!ch_OD^d}M`7l*p%=C*?%>R67nN^r^lMetTfIUH6f=wU#|ddvR;+
zhlmKDDvdtXne@&!veNPWsYBWoqU>I+9mVOBp74cUKpq|~b>h@iUyB@p<8Y-Or>xF{
z`l0GW!kP!P?_{EIxKd(K5!pX`0~$uQ(q`iH3|ncNKol>HgqEg{^q`&-TBc2L$ksD?
zP~Sgk#EQBLvJ!hXwX6(Q)IUL{?p>^+2X!87qurpFv-8@K<xC%0&azR<IUx+ttIds+
zGweYn>*XwsQ_f4MevF53`1R51^FvwAo6ttD&#(velL#%+eY86xv|$gbDMB0epstV5
zhCQfjK!XSM>uW~ZCwrvcSB|PT^<_I;7%AuNhzIqS$eDIUoN}JT$y<zlUU+RZ`?Sb%
zwn1j$cCnY%RhQL>_ZTAnQ#afx`cHj;47&d`US!;<2JXue3|&Q;+?XK~{ij}I?+ou@
zbY$(Nh`=GbPdamq`&0$B>GbezEV@slIz7LBRd=7_Iz4xC^UWklKTl0?aJxiGG99Mk
z;8rUXC#6?XQeat@)g*3IAL<k#Ev^rhCw!<6F!^e`FOU=NBuv_(zIZOpb9C9P`UX|R
zAL=~f2<&tO4$7^mM`&2{{*@YdRSkT!1v0T^RgCR0OZG9GA`M!yU*K_KnI*?B4#>u!
zCB!k}UYvH_&<{7ONt|Z!MDax)S-rs%Egr6klfFKjM)5Rj7n43>63zhkAoVy-w0bQ$
zI7Gr39gfqzXOiPnoOivARGiNA{R=8QYpHw-X>0SC7~TZRJ}C7x4!+)o9^ub$ruZxh
zexC<l|3pzwBBdW3C?WerB=%aCyv+4~2M)h>f$`sveS!oW%vR!@)jPima`tV)jR#;K
zfaMh)Zw>Gu5$9cC#pxLcIF>~HElXYnL7a%asULu~@5xl#cjT$zy_U`=PGD$srQ#^8
zpIaFdQ@L4@%B_r=RBmNBQn{5;oNB}U)>Lj~)TMGO!=Ea)GE&=Mdo|o^G~~M78%S8U
zlc3<JW&IM!AKuBB0!-U;zG=N(GuN76y*;7b_A)1co#wN!1Ddv<>+NyNLkIDH_3qd`
z4112g-hNBK;q~@_Hu)7@+GI}VY19(0;`{{COzW*U275Dml=b$2_A;`J&}t!wp56N)
zHD(Xvfc85iiT9Sq=qFkC<>|{U-VLIW#8fqg?*^T@ff{AA(s5o6;;zF#MbW+5L_$1c
z<Y9%yvg$^b($0fwkRz2-7tg8=Y4Zj79U|{*g~FDHv@@_u;k_g-!3VVUFL9ym#hFzY
zTNN32;ROrrOb|o0XtmH47;joS);%uIeK<MPi(3o^QuvaVPALW@zgu0mi|=?XEnHT(
zyzu%$=klG~gRu3s2WjwjRCVjKhza?y2PF(2zx_(el*ezV)CN!@+b-=)#=9m}Xg6&!
z781Vw!Lzr}3CNYwhBtF6q%^si18H;2(k_$IHgPgam(sYC6WVM@<MSJKfdcO$?-;Uk
ztNh=(<8tIh;?jg6BG8Nv+}h)u!=V{nnSqr;ExnOM(?HZ?;6_X7?eiyr$oBcAkaKHR
zqjnR8@XfIdt^{bM_g+9-@GJb#=TL88T<8;mw9tT5mX6z7h|^npl)Ntj2|cZ+4m_gO
zlC}oDUt7l5%Zx2#>@v*On#$PQj9tZ8A(l#QI%Butd04xMvDX;0GFI>oVy83qK4aq;
zE5Y`JmdscM&RDde=Mj6Dv5y$b{0Omk7~96!tBjqElP#@}v6>@@?O|*OV~;atJ&M@F
zj6K8H{fu2aB*sv5KYRu$(fyG5IZirYOzmpiLSyv7H}JApA58BfwtqOu9(Yj=e8$%Z
zyKcd7ct#t4F6R#MGE#d#%!ikf+WREzmXJM!^i2|S2orD9VNoNHqAK3X)ZX?%P#MTE
z3oS^PE#V{y(*eP%0x+n(Dcv3$x>0kwI;B~`u<AL1{Y0NC7T!YfPLcmCsJu#Bj;`wn
z7O84b6t4z@cR_%otGdk5^_UQw)1&r$p6tMA%Xy~dRCFe8N@y3zd;{Zsr(DyO+NK8D
znGs?(3J}~x5R<p-F(zsu^iGy($}70L9;0O-LI*+W87%_7(XCmT+Wm%rGkj186sZ9-
z45SH)8ietJ*H%X7&)_t6gOU4WU%H;LsM68(>%Dr}ItccSC|l&sHS!Kuo1BiWzx3<b
zbr95uMb#eJk7|c7-Ne+KoLG;)k%|=!P5aX!;CL;ow6ojNaSGbgf%>I8f~ByFZ<QMS
zLmSddgNq<$X{H*yrOXlh1C5G?qt#0C!}M-U=W|8D)U9&@soUlRivOVic@)U@Il<Iz
zmWSFE$PbA7oFGy`Nfi{{1?9&@N3VRWh&g{e=jfH&k##GWsA53egY91+KF9J<>h^$$
zmS8ctgl}BT307?V0tppczL*o3xWyhE1Il(2W#Ak7`^kd2H+2h&RLEsS!@Z>iy_P#>
zV-YD0^f<6m;d6y{(IbOFeq<am?=$R}r}NE<4Vc;;!E068^O1AW7xY%T3!RfW99@5~
zGGnk^2f^%kVh(9`lbHHwG0cI+<~W2H9RzJ7#GJ*(PL^UE@JutQ(z%p%szAk{p<c{e
z+W7<?n-Q@GGxOoD?p53S`S#ZHmX4pmDvrQ2YWsUbYS-E!wdYU@-dYa#`X{S_&(-#S
zB{iRp+2%Pgr|q<gndddc_YR!?DI7~&3p4+b4iTX3#Py`w9!~aQfb1D4R=XSz<f%Cy
zf$3X)xGm|&%{7QyPOW8@Z3k~p@rIY^X4#%r+k0$k`<}t(XOSDC_T$>elvgR+a%wl@
zD5G(25@UO7Eo+~(7ku2D8W<XARl6FJ>;)k;FkrXLdqoYrqUL-I<x|w0PyJ&tJ9R$c
z&sN(%Px7CEdvY@r{A?iA^3atJ<9a|1V|V!@VrqC#U`TEMcWTpxfg$638GR3I#SeqO
z_~CTk#zD_iN5L~q@4&GBLVGZ;G&r_2I5Eo}Tn-awY=@<l{DboSh94ij@=t|<EAJ0Z
z{8ORjp@|O!CjKFi`g;`jClu@`c)ICLd=xXu<d9&$g)8br2;pkE>{;_0{Iwg(xfo#n
zU1dWLc}(;W;b7G~&6Vac(enhMTqlSgEeLaTLO%ryLcUHAGm;=!b%K~41;L^d!~`G+
zC;ll)X^Y4>C<wQ#(j^|yu`(Uor(@Cs+^J)kOdGQ_^a#+UJ8L(H^{(6zoTti3={1<o
z5xj>=ID$W-a*p7P{3E4-o%Z&((QSELa42l)_!>KO*NvI>uIq8#oLL&^wYR@N%pkvc
zzyppKPq|+9!IkWOlX2a`$P`@j8JSUQx%~7Y^sCEHe<JC}M?Vjs<E@3DwNL-}=yAMY
zg1&nBc|9V12=DTPD;ZoQ?hxouiXf&KM1g98|Ko@}4e~zI<sD<z&qcZ>1bl4BJ0?&~
z2>95LcMPZ;CjYLroY;Zq_7mu)d#q~vPF%fKxW_}5&aYu6a|GT52dlf+a_SR^Vw}uM
zegZkxK4mW$Y)&r@yb`*Zjn}mnBW4JF2_t4dUt`W0<Y*b6{{$w7Bo2|I0}~5_sek)%
z;n6Gc8tFq5A1+LK9@#NIlAA6^MD}m+Y$MOsz)Rqo{%F!Og~8%&sUY3DH8p@2J0Eg9
znwo^+*qiJq=xyo?>|Lrm-ZFoF_s-$N1?*>h86Mah%|Glf;S~5&d*FodOGBumJvcYr
z9-N;kCvp`pKDKkmLfsBaF<f05yb9l*IC74uJ;&g`oKSOKQVZU+-0>*63SNct<*>{d
zuqVA}`Th<^!6D1{d(!t|=iyLEK8F55ztvuF-140tf%%%QEB^?fXg`07PT(#GX6cR`
zSa`SEb3ED4)!l&-0z0%<p9D|Wl@H<#^`9|Q{t<276Bt%WZ`ixeelYSp@n4{o1}?M*
zu1~+VEA>HpU~cBXX==~g$@w^;>BY8!HUS*=r0df$fux%zknP{W+@L+O7yjh|6lJz9
z_8(043cM%`hNr<o%Y-rZ-np4n2Tu(*TJHKL_Lv;i14hNvz?E5A4;2semIkg`(3SO2
zX`oiDCYW_5W@()$4wQ6z`00RV_7B>JlGa`xeyXT9^&yB<0=s(-PEOjr?>HDcUkhlr
z9|-Z6pv($D5w0otLI(+f14j?Ehfn;2?e!>cmIW^-y{d+H*tf#p4ub=;<UYr_Pvjr5
zpS#b}!DcAM3+Op}k;}5+P|_Qm-`n5CfZ2`Z?O+P57vAkZNezU!y!j_tveQCmC`u9D
zV1MYaxWBWF#jEYdlbe=Z({=G552GBlYusN!gj4XeBk4FqZ`E~qHt8u9+xeK7FnS>B
z=MWX`A6X`pD7Xf3_2cS;9^DwYgU#=i26l7lE%?OJ@g0aR4up>GYybR&tmVErU5<;A
z15dWUm4a0!6(2)QX+M_aKdb#%vi}rXlU0Wi<44;+Pi?yF+OBaA6?QF3Qn!(p7sgkw
z_^q_w!Y_2w-FVpEzWWQTzJu*MzUX}J_Wr=`_Tx#-r{bln*lwlXfCuIIxIK#VMP=(!
z#I%fF^s+T*dpKJipz`}Jd{1T?qFeYinh7slT#ipeC*j*Aj#a#;7kE<*JgJ?A4;a;g
z{r>auYC<|AXR5(GwW}g4ldjH66vM6lUab^mk(|Em`C1!Xm!sOXNYk~&DtbM3I(xNj
zFd6z_z5G=BJKh|CP|FXxvE)Rz&+vG__g{CxDE7c%oPgSc<Ff1pj;wTlnjNRB>1x-q
ztjtFzp@Muu3GC7QyHRY{ty$w@<RKa3T)PZfsX@4ucI`-Ux;4$xnGe}&(q2mkeLQ;*
zRj^-y>iN&%hi+rw!;K3hW6LYm9zas44<(7`!P4NxS!$rqk%Kq$hA>I(-YOc-QXEo~
zc3ZyR&5qm?y2|o{p7b7d3wtHP)i@Q_upb~!aQzpX=moni-x15XF=KtW2Z{!EvKXOM
zY2bt-=L37-WqVH;#i76Uf1xJrhlUPln9k|x6j8^Iv<b*Mzw6e=QX|_12X=9WI6JG9
z4;9$5`-2^Cd}3wB+n6vPvS|llHtv@rE1y7~;np6=wIGLjxbt=~QbO0G4xxS;JD7IG
zR4Z%_F=}wUmvMf)7I35jPg^O|tZHv@R`%v~RY)z)YE-*gwXOW&0=3}T<_eM4*8vR8
zRi&N93J22;=R=efJIHuMEJU%L1$(Qu44Prrh&{tJ$D)@GiHBYKmS>2`V9?T&9$25%
z0Y;q96AX+ErgwcS9LR+KeNPLj#EtbfBxoyPr%(+kft}E$(xl7!C|!bS9Wrf{D*CVf
zPu017d_<MkcR~wVZ`eYgLpx~oT;8ySo?$$Jg)S#M+VEjO>Cfq3Q|6(nYTzmKrNCzx
z?D~Pkt>D5$*$n^IG66EL5FfJQ_#s8@vZ0F<VVeVI6vErwumh(-nP>`auAb{^Gz@at
zgIDW|)>wOBEVJ1I^U`t2Y)g=>H*LW*d=tuM#F@w@z)JQ9?1JgN|K6+(pl$m;C$W=g
zg6iF+U0Hk7u6qRoCKgBFJ?-`k)?l|9#OuNoW$IU!&MUEl<|ugA(!p&kN^TRdBCAyw
z`!Lj?Jd&lnOCiru(C?qZS&otO#4PuC)?Fg&y;*&L+8SgWI1gPT^t*IjvXQ(O$ztNu
z775{xi<+@5$)Q!TBG6T5A07jiP8GT`r44M+wOGa!D8Gp*fj6}ek5OxE@^O}3wQb`K
zFp$<mT18f)tOmbk7d<y-tXVo2(`_ht+tRT`6m)A=CG(Z(Vty&bR7x@DLJW<7XEWUz
zlZ7=fNNF}cxM&T?GBBB@I6a+u$}~2tZiiAK)35`M611hX3}yicF<anQ(2Qg#WF~JV
zSyrF(_QB8f>Z{b-w(S?JK7VI`s4DgL;NyN-eXyCObwV#u{UfM8J9bVnw!1O9@_-BV
zGbt!tm-{^_mrERb+XTpsYmZ5^#}FiF*CUH=kDIYmM0<#|L8R&Sm@Q-u+hdY;EiHtK
zgVC(&M^KeyC}bw5ne3r<`LnFrYw)ww;H{_!xU*HbwCzKwmQHSS_ojVtsT^cty!LM+
zt3KSK#d4nkvnXc-&Z1)H-8^YoI`}(;=6UpD;Fw1VLjR3hH1{7z;odlcd($Y~!4cg5
zH469I5!`o<!hPci?lq%u*Not<8-@GZBe<QTa9=Zmd*LYDvqx~3jKa-EH7#@@F@Zdi
zMK?!V3&%b5GAJg7*ZI*hrYym*oD5HbVq$po3lqbYWVjy`6T`#DO$=v|fxq-%Vwhkq
z;@IP02uJI1vYFu(G7LpC>@zcLA;b1)CHYBuq!vFS!(XEr_%2U`p@j@xpqP~Wftg_u
z8Td+dXc3JMw*V_@6{;qqi&)FcWm3H5sN+da<R-Vo(!n8~Xh^++w9p3}mGnSj=kbo)
zMhM$6iZG7G5T9k^&Wau4t(hpzxEYy9vvo)tfY-h=^n1jF#5gNj2cH_FtGi+pp{2U0
z(0_A!gXq16=-73l<4+^x(v?f3tQTnm*m&3(`e0aM^i12)H$viXN0InFT~6pj!0rEy
z<wt+ys)Kd6KFZbTr_hB326cOfj!Wo422<~PjB+r__9Iz4iaf%cCWfC#hD<>1B{JL<
zErD|tC3I85U&*l2%%J;_{Y>0QmEw699hXstsp+GAWlrvLj*s${!w|}nH449iB_BjA
zqPt%5BxKgk5hcg2>K*)?=|3;_E?KOVl6T=*YG6%Fne=M32aqyi<s8iik6%r_7Mm%t
zQb^<yE0)n8|6ModCZW&p>ioYg<9?QLm_jk9jDyAS$QpTn#r*Rp^giu@BlbYp9z3VC
z`kT_fz*FkBQ&!s)tsgSxcU}B8eHc8!<3aZ$js7Ecyl9Z<e&6|fNGmP)&|gR~d0I6q
z$p?+HDjPoqW6e`>@zF4Q`zeU!KSu{Nh(fl*xsd3sK!>%#L*>M*ZD&!LawJ^SHC-r!
zXNCjXa~#`H13bevYo2EhUR4;J`6#y~wsJ2*yY5pcvTheE8+oF`7LMtGZaQof4vPmm
zM^-i-e6t7WqYgju#bbzc70W$i<g?yh6rtVnH}v1bc(gS!tTHoPM~20q3|t{*yI4bt
zLz<}K7mU=e2w4aGGJli>{MSe2oi;M>HzS(eJ~cAncl}lGFRw$p;q+5=pYV4waXS)e
z`o9Cp{apTx&L-5Gjh?ABkO7UJE*pIVQjA7VH#PbXA<)$5-=X|<yC_F)*hKnk?EcM{
z;bOB1x};;9we1nSNuydW;)g&0Pz&Dxr5!to@6uHE@Em66hDGJvM-EQgei1rZez$h(
zU!Y3Rm%05)9j^l(^rdev)6wNXx3}TZavPsXw7Y=@b_)W=Mit6KlXM<jdq!k`3{Sk;
z{gm0hGd=ut^I(2h+X~c_ezlkdLjMIS&PO8SlUp(zn_rX$J}b?6*WPnH4f`%@o9$h2
z;)lnx=mI``F&Wm*Me*EZ;rK@t@K%3V8!u^MwGK`b-;1LlvaxNzhF5XI+Iy0A2Wa;8
zT~@WL$coLz+kXkQ@wELqzsbz$$BtAAo(7Wq({WG6mR<W$N^_+>I6aFpekEl1uSI_B
z!|OEK2R{lm(@eWrr=4FKI6QC;w?-j<;52d!Sa8oAI0^Sj11U(yPl*f<S^a!h*r%er
z4BK{?w@SNiO4qjrup!VpZxYnUA(1m1`^)pPL6}7X9=Ei2(OPy+7WOIld>=b&*gmk?
zgA20l!EfiOIZx^*0(d9kr5608)dB4X?{gZOXB)twJ~o5b*#ism3VX*wic;7+&q}d)
zBhsIP^S}J(%=2semETKhK1U5s>rLgG%XkrOhgz_sNi%j=@hG%&`wDcXZGPOuPXulK
zDPr1R{w#Le(O>jcp`ADmEW{Gg0)#fc3bu>h&-}9S@SmoaZGT{;p_~=^Z+(l(osRN#
z`<v5qm_D@*=|?ulOTUEazh(MMk@WvmpI(|AHs>At${ToIWVW|yt6A)y4)5=BT-yW3
z9R&lH4!)A*K=%*cl3t1tm4lx3F3)M~!Q0R+@S8+<%;k1#!S0rY`LFSPaIAOOf#olC
zy}-P9&*Nd(0Hdw6;HaNn#1X*J`diwEqM2@!N3730#710>*^qmoC7GpN?(~6Hdsj0?
zL<v@{9cs=ooL}$_7_}heJC~c!1w;OP2-6;hjiioJt@MGjPzP!=RhD))csN`xP;qk2
zS7=V)DE|vb(h+;kQ9pLmdL4mdng>h+DPU83-c7cr)MLHA7$Y>&1?8J?+MmJ4`6du9
zT#Tuva1jo`3M(947gSBM;wRg>E@=fUcjUYai}W1F$p0{1JMB*}0u;ixjr#t-Bk%!q
zD~62l#Md#JlhlHN<_|<~Uv9j0hZc*>=b!GxnBZ?kV!R`YSrp5-5p4I^^k}+-Rvq&6
zI}-AW@C&*n5mZD-V~C|we)Ki_|3Xk^Tj#hNs#n#!npWet$gSmzrnTa-T3csZTQ6Fp
zSgj7LRavpdj~|ym>7<jCd2=rHoph3Q-rPEG<E1|3lK*vpWUCbd6P16nUVftF`sO-+
z_RSZ!Uc4qJp@@1{wZW2|o!yMzy`C}M>Cd^)>8)}3T}$(p=UkZAIyc8Udv>0Zo6Ent
za%r6lzjs~jzI0xbr^YD}@ssems}#~Mzucj7T&ZK8@@iL|t0_KK>ca24x5~7ZBK$sl
zox64J#nx$8S;fDIuxwoz?{o0;^VW-UtV>(GD{ipPw*G%EX1UVhuEsCDUw*mx5&v2j
z4Ciz;;y2`LL^q6cYs3q5494FtNCdm*>3o72xB1hl?JT1jG5y|1pPFBUzhB_?d0Y)J
zNquvJr`iSmz`fy)Dj_OUmaogp&!fN!@%IUIkp`bl@)_kf;vj`#&@==&V)G<la=J)X
zDh*smN8B@SunAhXU59-V_DeV@Vf^~%Pout(I-0DIY3S$C<r;Pth2ci^%opix9l3=K
z4YOzCx6L6{{2_(h>T0Fh+u-wC@y8)9zg%zLE77~JlD&nT#ct8k<(~SQY$RQ5#VrTJ
zwZ74V`$}*CphqgW^uPUppO*LB?9R?%SuI{qjY(LUtI6kfxYppu{ImJD6|5IsBm`bO
zQiBWQ(O7VedqwlgOYjr&_!9%1Hze+_;a><Say749h5Cq8>jgC@V|KtV%UfMFHBD}x
z&w4>!T`OraRO7E72rg%1leZcS-X`DDn&s3Se}y5^rSws3rZ^k$*Dz>5n&|S&BaLM4
zhzZEqBaD4vj@eELWR+u_Rd^S=t6SzadF!)<VGZL&A)<XucK;vAn}fge(GaB>YvNnz
zzOfm9uEU8Qf%b?hCJOmqm5CBCkNMrB7fXLD>8zA1xEGd(c=vC^;r)nj_<cCsmZB(|
z0sGRB_CPqi3Fif6fcGQa_J?rTib2cwn=1haUjZGk`_*u`oA7lU#4;UFft-H8Nq~cZ
z+{{znz@iVB_hvY}8E`{C_yIQo^6_pn;3PonzrtY~pbfARaKk%rFo+L^!+VM2hnf`V
zHwb72R6Y%d`2&oZfMtMIz*T@{fNg+%fE!UxKj0?d%4aASF!Kob$ABM@zwDd`*am0=
z+z40($T#t;0QUm60rmrK0vrPD2IP++_5o%C_5<1g2LUSqm80QsBVZ=rU4VT3dlK@M
z4dI84fOi9mNd>HYZ;v7A=7ma9Yi80}<Hn|MNE(|-`q{WP-5(BLL<Sk+^D9ag(#hZd
z8?+(XLdnRSn=$Dc%PBXdw<%YjdBsK3vd*Ij@-U>_T;>oZj!og8xJl8=Rg~#SqpV5j
z3&CtWeV^TIQwlRO=VVNplVP2ckv%6PuP|fQyo|n#_Oa{7bfw*$dQZxo$s*q}<ZIj)
z4o^f~Q(If&<txr8i^>+Me@aySb-Pek+sonbGl0~g9rdDa)Nu&(&(SxBbt}xsZa1{i
z^HZNb^i_0$NSKJ2AJ^VE^!1>3$D#iQ^er)T>hL(O&7hwfrANd@g&DT!x`JKQ_b^~O
z#-5b~)Qf<6jf0##^z{r9WMDa2ptoYIeVgq%vMmypvkdt*VGN=<B7x=jLEjid7ySqF
zH-NqhJgL8Fu_<#hG7B>%6=&qNr`*l55tyGJt?dUNem^48F4S`wcsTaw!{DhOBL7HT
z=Vf%KL>Yzo?nS<#1L5#*PC-7~P0{%ZGxFwT^e4BESwGyk$isa9M83uWJkbEA-rp+p
zjr3EIuTA9ZO1(Sf9`G=*Px%qQ>tHy%kOb&BFJp5O`xe7QT-0YT(oxMwApJqmyW`Mz
zfxaaU{dLecgU<J#l;R940n0xu<=;dC%7={$Gpy|#M#{w?O@#es;nR{csQ}u*TqoMW
zTp#LH4F3JU4u_i%r#u46`4IF>e3+69IjMtY`(UiVKJ97iBbK2(*MndA82Yl@k!_c-
zpY1}vEhs0}Uay0`IS&0W==aB=k4r^=jYH1@ePbLtzvbKj`k7GIs5k9VA^BtJby9vT
zy#w@k`rIe^W9#*Zlpjm~C+O72JbtOqJD}eMI?ZbY()l-I+Tzq}BG#z8eic1G6q6tH
z4VWL?6oP(nPjX4dq`Ol}GOS&xB^lZ4)6|T-JIAOQv)ad&X4vX7W}%JI(qi7m?2Ueu
zF8X&dCWI}JgYFg;!YMD_(blG<aAzEQ7DgH)Dy}=R94PnVD#0bfE4b4kC}oNV)x#Sa
z%7Sm=_DdjEzzQNZ=)D}*85aUkFvt;CP6kvgxJ3l~7`Z86G4s5Lm=$m-krs=hd=r^g
zQou>4=>on6s7%6bSO9*|jsli5&pselz!gNE!_MAW5b#$Dxb_@fz#>59I@}@xUa{%|
zZegCxb9Di2L{2|XQRbpfCn(^N3v>b70F|A%bs9EMVTkD<if}6sl&RY-oRX75L5f(+
zj7SwkCXE4>i>~lq63GbRl$~gWLPj}^MPP(=yMkB&u9F#ID8hyJ6_qMn$|78NHBt}Z
zls+Y8KI3JQA|ihWvX)KE?Z_G<g?~_QvXOER5UT!TK;>TC__~D%;gq-TfN}?tIf$r3
z$ynSdpGX-I?iDk2DUofsPRCV6<R~sQi&Bz~8@~mls`H3U1;X~Uvz%|?Mgi`VaHC#K
zSwJ?fJe=O>Z7I8ea2Wl82>n<}+f3vX)Q%}XVV=oAnCDkSY(QwN3f7<N71gO{IX;>4
z0NMTmBvQ^!Ae1&l(JupGImd~dh2~?*C(OgY{K!0qiIf3hIW>Sv9d0bAndP+0lrPA(
zQzFGoc@fC76w4s{0GA7vz}HSh&tlnKL>|CJgh$g#4=(nOr-`tK-NFb%IORSi<;-b_
zaWDG~l2+sT5hGuPuPpGD1-`Pt|BDvLoMaPfZ#Z3tn<dQqx{eP@D41|F=EGQ+jD?rM
zpwkC6<2e!^r<0Y9U$coAXIq9{ZkvE-h&!g^+p0wDEQwns<a)-S$#hQa3|s{mK06Wi
z5^hYk5s1~7(3Wl^Q0OEuXvsE$$nt*`muUogQw(r)MBtKX4p>|C&{icNmN5~mvVL4-
z7>xCY>lH(*#U`LZ-;|-_hTi){>hRwxG|I;d45B`R<8^+#uOM(k54fKKUn|q4nE&5F
z<nP$Wl=PBp*Y_kmBH@^cy8fq2c%g(dB%CYZVhL*{Y?1Iz34bWzZzcSjgnJ}>Ny7Ie
zJR;$k(`5M)UMS%V3Fk_<Si)KfTO_<w!XHZbTM7Rr;T{QJlJGqVk4QL1j;qrpyimd!
z63&%yv4ph}wn%uVgfZskpBlZ~n!RX6bA!LxIz4w*?$k^23z`L*pQ95LnTo%ji$ky$
z{3%@gnN#bu+{<$FE}5>gPF?G7YWDfpOtohFT`PT;E-Z1B6qc7br(KqJX{`sl+Bw#2
z{_)@%D{?v<i%a$FIdHh~CsF;b6@dOG9j-EB*nxF7H7dCcUcWmRf8z8Kh?dyOhUVNA
z&7QiNOR!g;OKg?Pw@S&aS<?V{I`lW`DN^Am#EG;fcb$tFWUR5yujKM?F)O+F!>_nQ
zZx7PMd6<&xUgfN9a@D(?t7?$fz#tTVc-Gxo?QX=5viM8VLWa}T)Pz3@D`Xh)8>*2J
z<+<uT)yU^X5xC`wa(U*a<gW1flw5AfyBqwm?exFJIsb7^HRhRF$u?l{LM6eN{{#^?
zWBxVp^AX{kW8|+$wgHolJD`<_DadIoLdi$WoZp{p6KUN0VaU=!<Q|*97%AA!^%~`G
zOtuN$jgn9%u;F>ym=7gc;bRx)WFx;ZZ*!lVfq%n?f%7`+Zsa%S&&8l|t~Bx+>wrx%
z8uLC_M2NND^`LUjHS!zt>{_{SiM=B5NHrqm82o%l;2dt?#ymbK^KTZV>iSbpLw`wD
zZbpPQHS!zl#-<FNW4~lH$~W>G_3wzuZ>&SD7Cob}ZZSXC=GgpqgGQSe>!z`8Z{#x*
zg0UX!n69f4Ck)^+=kJ#5LAT5=_Sj7s_~gf6z<Uul=QsS64W|n>W$FSQ8h0a~f&T;%
zbAH3m*yYwa_`qlmvHf=w5}Dtyf1le14Ey0SKYuyI9AeAAABpDtO1({Iuh#+3kjx>r
z{J$a5oPR@uO|Wlh&~vRf8b(JW?I%&C8xeDU9z!sk-7J8@Jw$VeE&n+rva1<vhF`Y#
zHa(YB2a&tce+<H#h_I^}`3=8Kllf!$WAP7=m--v|`4<Hl2LFfo4}q5^GV&XK%aF`(
zG^}ApqkN+shmhA?zVUaZhi*p#g2At2!?-~|js$alrL9V5ZPP(36HUR$XTUK?H|K9%
zS0%Vx*XfxIJq-~?J_F}j2x&(BH_7~W$^2HsA(t7fMn(xIAV2<#b53)Tkzbl2l0;qx
z15QBPC|e}R+t@qwh}fGqB^db(5^-bf88kxeyE@j$0uau@bu})d{>J>r@g1-H%KMOj
z@FI7YOd+;>Q__fN3Tink*%pVMjE%8)^c0+##G|K1o>vr8NWrS2#pNFpd7d$`qn=|G
zV||LIrz^&~6HPx!G1il4`pJ>!m1z1Yk>|E(`Z&dy*Q4p<6=S}Rre{Q+_oC?*%&Rst
zNyJmQ8W{7YDK$lzpcwO^i3~fOiuuiECW&|oHu{YD%bb{ktwCd6il(22d8W!t67dx5
z#2WL6IWYyhlg9iIO-FAu#=R*aML9z;#<PhG{m)d4aT-lGo`#31?60XfY#OHMG2Oiw
zX@|)o`Ba6+%H}{nFByBEMtfOhq607q`qXq*35qd3mI8}c?kd4`s={MWbD$r?ty~^w
zn!|d5oT~7+(j0y+={$}!hs{8!r?F3A_+S47oqF<90|sOKhY%;7-ytv<bfaEvl5W(C
zd-?I|W!RI)f98-2e(GcFmsn+@A)kNdm%)(F?+4?_=T}ro*r1Q?Z~W4U^4nteVfb~F
zutN;}N0J`9Z}UeX|4gMSrX3#_^pXAlZ7Ijt9}<<o?U3LfIgS$5D+7~v62|GInDKbF
zppUFC%ZX>7Y{5Tr99<<Ahp&wozY8QicD%0;az>8d)u3DBjK>zxW0nYEieC!)c%`h`
zCUBTVgqITVe<1n0WB7SS9Z!CJ0(wURI^WMZ7oL{!exK3L|03zedx3_(@RX1<Ug2Mv
zW`KGkyp}-zCz8J`roVkI<PXnRZBouGIFj+|wLAfRji8TLHc9)ODdp?}UDV5C6BK>^
zQk3^Ww@Mx}IFf*VG9GZ^>2o&d=VE@a-ccpQlu2RNC*W^NK<`LE|3v~i-w!(#<9blm
zH$#>?299VvJ*|R1a^1X8(8nuQDJMzFS(-pjTLStoK&PEIud)dVv!t9yr2Nbpoo>kI
z%P`d6I$x(tH%#GsT%_BU>vX<T!Z1hB^|<-I4D|EBZ@hns?j*tn$<M##ZVrD1-D)Pt
z_&(67zwth-(a#S{`F%E<;N&}14Bv#KI05{IzkzNe!VJ)<zp}z6aK7`wuv*adxcUBF
z&{?kWUanz>UnJoFGw4>c02zM@^vpQ(&we2X^MAEXa2n$x8;)(feo`Umu)pM&>MQF&
zr=5-Wdks7PS@QSC@V}Qpe(K4(eX3&gnGHI}QQI9=LeENZR}`=0&x?_NpWsJ-ljGXx
zZ-1BkeKGv+N&1Fw*@R@i<HT?#AnRqkuPokW1AQ*&Y!~A_>WP9|QLYC)p54}hJ|X!G
z<*s#Al2ZrYO~C&E`BC3~Szmt6%g~d6f4`L979;;%!H;#sZ4>$3;*N6+Nw-$%^h+f@
zeVlH8e%Qxwlel9)5OmQGJvKr4x}=v&{z0csmuf0apbNX%b-K|X?i2jO@5f6q%43q>
zcwgM$|5VD~RI1C7!%NA)5Q=B#)d}crpo{)FSC{jQEcl0#-Y4s8%##lZIq+9yy%tLT
z=Rv0(jQ7utex*q{RymJ}ce9Wu4FcJYjh$6Of-w(dfgVr)s{|e8#?<#Z&@&+?w*S;7
zkh2bSju)jvFV`3^e@nps66my#a*a*sd9KlrmiY6@X`si`{}Mrm9BEHO%zT3`?PIj>
zM$oBeRg9j`NPgRVT_3)y$dHDKmv-pxsuH+yI~jD!>6UWZWI{3MoNx24wF!!NHxn_J
zpi3qbXh}f-Iq0-czpR(BukbqPY+rsWDgx$HhT{q3oR0~d{5HvN72Jxl5cGKVTn#$s
z?bvp2lKj@wZBl{`{!7ruE15C&|6>CA-I8Bfq02GWqt6oXkB5@+^qeB-D7RXdW3+o^
z0{$Bl(7O`QH%a-MOLh6H6g@mC>9$Eao$o?3JS*sN_7lXXAim~l;!B32qNu}}JRWyN
zQSue%Jja4Lg$}3F*Sx}Mcbc<1uPgUEW;vZ6uQ(SHNsPr4J$;1e86zmOMo<dk7H^c{
zyiWhh$B{onlwG5Lx8lTiDkaKr;>@T-ewbDAKmF7rUgdcJWqf!P`6i^g$?x+w<NLpA
z#p%4JXra?#FRySql^S;wKC|%Qpw8*9cUITo6B3^ihvanDc%3WjyenLF&Ki8t<a4^3
zTdAe^T%sm-M#0RfkQ|4}$#XdU49$siv8FYOehydDTwlKiSxlJt0w{{r>72W;u(ZTk
zGQSwg>YB^AZ1x3n=a!dLI4cV0I7-Mt+ebvyL=${|e4M2?i<itVEVUOYJhXF48hXmq
zJf{<13wdvH*0~#2`d2yKO-<gWvTLhtws{WwoTAD~=d|4Ek#<Y)saO7G^MsYko9h>g
zZ_`T3YCK9~6AnXbr6A|j+^M;jeVL${b8rUfaW|E;kh9#MKMM!2o@yuzSu1gOyXx)L
zpy%_eC8xj1<?;LM)g|RF+)9e_OBP;R?nCl4r`_vwF2FZ&_4u-AL|w|gb-of$G#Otf
z`P}|kZ7b~Zo3+?)z?|1blA=Cmb2{<WVYRc`*DU&gvcgrp8fsP>N*H!9nxJAa)V<7H
z%N5PFiK`f`Gsk-V45I~CxN4l$E}y?8nyJz2^RzlqJy#vR#)_?=UcChEI(2IF*G|Tl
zLM02!{WJ8>)a)ME-c!P%T~R)&Ft9jR)_WWD-p8JOO^b7(9Ee4AE}svhu(8?ii&e3r
z0=4{)IzqGooY&RVmBkpv-Rg;LYbP~m@;hr?p1N2S=3iUxzpUE7rqPXlhTc``yvbAJ
zZV&^nBwBQCKBi|m2PQCU{?zhSo?5@VW}&NLH9Deap{G3GsjPJ4Q`LI6x3<Q$CPo2q
ztp8<FM-IYTk3NiMc`zFID#LU;W(o|GXqC}0qSJ~#>8y8oM)Z=3#d_14J_Y>}ZKmdz
z8DH`hdA&{iR*tnT@x<*qa*!2pQ7EdfS<vXm+ZrXUn4^6qWo6Z{4hMF`u7XcDewn8X
z45&eMtIO%aOAsU4RklUUSDaI)6|HhtuXZ=h@w%Guai!0XH$FtEp14LU9%(dtMX7Um
zDXAze;zCj(-lrIz9TIeb;iWJ}ucE1O23PrXC;lr=`vMG}8c&0>8E>UTtdXFIxRQqF
zBPZTDS%G;lrl&<lv8kzRo7}O}7wc}Un9((NMy)WT&ItJh(bJ3g_%TlP%UV3;mpNs-
z_@-mhXw1(OA3tM7@UG#c&-p)g)Q?(rL1e`$S$IRllrfEwKhxYv#Gu7!)fEw+Mkg?R
z{`5IbURMn#<8OPL>S{_nsDc=Zo*9^fy-n_z5%}dRWB&ALTb0l#`MOasI$;ltfL7NE
zPfLFOOvE{#IyqrsnACb<8!Y8`9m$!WJ1tV%1lpTT!(L_#{utv7ci}{GB1VYRUl<ME
z0*1Ni6J!qYIP(J9Jfdo(H_q_d8*1FGae4_gl`By67v`9$FnP?R-{yklM>;o7^kBo=
zg4f~01PP`Dy{g<ly&7LQM|U=S=j^KH>=84XoHahLbCs*1rp|3lvh)n-uSKdh6N4FF
zf7j?fkKV1O3h&qlPfdJ4Ek!iJjs>8wMV<aK`bd!)kMvLEoGL8mys_Ei_Eo##8rQge
zrjlaXG=I9e!@JRO<;>dH+~jt)c$)mpE-a35`ckZ*sb)bX@!SQepQy0>xV;q52&|Rv
z)p&J(57y$k9M30WFUX&!J5Q3^4!1bZSyDd3S$1vtG^eqW>wau*fdW6M-rEv8oyj#K
z(#wjc>NXmow~?SXlQ25g0DK5<G?Bh4iKdaRV??MJFe4<o$~k{w`8O6+bLX8!=iKe8
z@#%Rz(;MCH)iHC&a7Hn%q$&BU7OaC4@8?DQ{rTnI%WCVIeXB-_h?zt9T;TJ3=`>RA
zkGK}JO!S~Yg+|mCa}yU$^N&`P)I&@8lLgMz?lsP8)Yy;lg6y6qU&+E4N8PZ~RN;3!
zn;R1K?F7@()O<Q3bV#B+lms(M^wc^v-|1|r|8l#SMsdWQu{ZeT!_mTWUpy~}Kj6XH
z7JYj7fdCIY2_6U%tatkCDtiHcEdnc+;_)>$H&h#wmhlV%1A7b59&&)yl)whw3TIR+
zV#shj;voY*XpNoDm=BMh<x}~8;k6iRc!-?QgelHj?{ro3w;P-uJX!M>K=5+u%1G~=
znlI+G$n%gB)vCplo)FI^OMeXILyW7g&RZR)GwDk|x@Ni4jUTyK0}rGTecRjQTIr6l
z24+dYGV+nmEnZv2qeyG3vk|Wad-0;L2eMl7Mw!=Gd9dA#bsPFN+*BWBBU2sY{>THL
zOGk_Z#kr{ACYTi7jIX(#6B2fLrsd{Q7?@!JSDibK*E#y5g?a6+6{5V;@Jv-Q@_D}8
vH{ExWs}Y(RE}(qY#|&BBXnFw2H$DD~As6Y<Wv(@KaC6KaIv*YseSP~s+Z!4&

diff --git a/benchmarks/Grid_wilson b/benchmarks/Grid_wilson
deleted file mode 100755
index ca7a454ade9a37f0dfd0ce94956c1e82ce1df942..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 121651
zcmeFaePC3@xj(*35{QC5iz3DfvU=Cvtf)-{B@xhFve9#7gHfUejV2g^qykCFMo|%h
zn^ev{tj$%t#TMJNwY|PjL2Fw;g$-{Bs2H$Tlq&JHXN^cef$$>veLgd1H)kQCz2DpZ
z{<;e}GxN+d&ph+Y%ri63%$c(wKX`3Fr_&Mt^>bY9V5saop8(0H!%-dX!SOkA9H%*a
zjuDPwj{d+(!(SZFXZhl8mz6B;rURG16#S*)FQ*?vbbWLZ?l2K@AP4!2=T4XT_56I6
zSGqY5NX)qFaJWT0q2zVdLo$BVLsJDj?q-k~{OVlNji)1fOxI<l>$1|tU5CYg+>Muy
z<oH))#TWJSA;hqcjw2wRJ+76#qwurPH2TNgCixt9n_pnY-I){|esMRR?na~|pa1AD
zUd)@U{Q8oMYVjX;C07o|_h(k#b>-;q&zyGY%*xrdb1$8nbLFL1j=rp>>axp4xk;b6
z*G;-vuoRbL214X{2>w_U#xGy6@2P_+u30y|d+^-%4S!YDz0;2{?h5>o2Fs;Sb(KwR
z4mb%LgTJfr*ShfZ9Z!|Nw&~GX-K}4_uK9Unif`YCw?F32%zr01Wa*$h$BeY!{(M;V
zs{6Hj{s{D5e|LbHGkb~I-v^xa*&F_uXq3Id$MgYDL&18(Z|fsp%G2EopCkH!@9u-n
z<9)!Leen5dANVVgZ*Tm&`=Ar)qr4CHLI3tX=p60?e^eiG(E6Y=s}J}MebAZTM|rR6
zqdlbe0YA`(94_sHpP%)C|9l_t;eEjUebC?DN4XC50dGUc?5*CP?xS8R`;e#BhkWMr
z!RMd)$aila@DKao|GYl(?cYbfoBP0jq>p_6)Q22C?W0~k>4VNaec&s7;Gf$E|4rck
z0>^h9(TAr(A`VBfcsd-T!6)Iq`egX|pig-FLJ6<4@FNzy=>+_ppg$P+bqx~VZQ*xY
z_+?hUg%<n`r))o2=lCp$bP9n_I_|Kf;{qOkPvMX4t?dN-+bsGHE8jZc@OQ+juV{u(
zgcA>pS2dw&<s(OOjJncMT6*`as@bJAp{a91rKOJ2>+dbSsp9U+noz}@yqQyLYAR|R
zrCL!z>9mSD7PzP&Z)Vl(ilV7^&8(1dy~mVRhi1&Fm^!WWo{IZR%i#+F{kBPyYepl7
z>g=pi60Hf%sk%S^riqiMYUTN0rhM+y((2kd6{Yu9&I#2{oyimvCx@;m58Yo~0hH`f
zNOoW4w2IlKm9r~D`S%Kvm7}XGD((@1$)W5qrKQNZ>b}yukonY^GpouaR@D{LXV%ut
zh@)L;M=PIN6CxIp16~ETDyGgt7Bv-fLRvYX+2u2)&M6JenOYgD0U!A{0XFKcsWp}5
zacr%qv?!}IfAW>3g%c-_Diy2~Y0lKy)2e2ntoK%ySBwGsf^ifXk`?8hQz6(Dx#vRe
zvnpppR+8rv)8``n-4&sGr_QvB9?Ge#szFI>Dn@G%NFX~7zY^u1I<q1HsL9LDFPvDO
z5H;mmGp4+1R`t|56^s#*JTX7=I~@#LR4cEnynA*P@GZH|smZ^o(n^>kr2)}Z&c0g;
zus6wxn$8C^lmO*Zs-XZkR9DQNRW*$*r0AAOXdIBt+^Kg}-kZ(pEN4qBWfOq#rdP!q
z%3WxYvrEgThRSCM9%Re9JganaX~nGS(EX)zDyq@4tL99-yMk%%o;$a+x?)aE)$FM=
zD-nHfmX)mLzNyuyQB(+W2Txbni*i@(^uEi16q7@v%d2WbC#OXV>LV?fxo!4+vqX}l
z!d{+TT6*uS%3cbLj#I_Tx0>WMw!o?)bXY|6w1L3oR?Sq6s;&*?qfSK&tC&44A*P}t
z6!3pkSG}-_`l;!a%gG^#-|BudD`wvvno(LYXHM0eMEgo^u4o2qNz-N)-s^y3nk9A8
z%&NO*GI)1+d1(zn*^bhiXWv(eQruKgQ#-4Ik>wJ%x~isfZt1LwS*X&xXFE`8b}K3_
z=(FilD`z^Y(H=w7ElHqb&8-YMYAWv*Eu^L*<Um@>DX+fYfzB1GoK;aZecIIfi8i%7
zR9Q9Kao5!Hd(ipID~LX;>fVYTNNLUev+k;z$!y7%rAlW_t(@(cKBuC>QCU-6JG(sO
zxGpcRbkt>8j_ZQjguK$rFT4D*D-!<f(b8L743yEA<tBx)bCQBrIP!{%{S&m(?8`<a
zMPz5&gHZVWu(IHvd~yVchg0E8v*HNFAc9bT{K+%}9O(j=;nOTi=_17dE9L0|cL1IP
z0Vgf~$$v7${J3s$I-bQC8=toqe((N)4#(+^2QW^@<9}+4ul;T%GCa-ka3a2+W3v^X
zg#=6f^S`hR8t8Zw@uVs7Q*8MCk)w<weSLu!BVoGZNeh2Q*G~{0;P`JV+#urnJN{^e
z-(GPKd})r2R(PtwPj$Ryg+t!Qu(a*x_z+?Kx)#WBnrpN8nwRkSyn$$3r^SEqc>v*D
zyTyO;c?scM+r@wJc>v+#;y}3Z`3m7z+Tihd3E??5czk|9_|-Oee7-_>t__}1C(AVe
zFl!1TV%;`41ey3{+Td(W@n4n=j%rN&a%^zs9sl`kaQix5wZX|#{8wm$+vQ(ugP$Hp
zhFfNXqah}KGi>ld2`KD6HuxDf_*@(OJ2rTo4X)VWi)?VV+xV~11|J*;!d+^E+t(RQ
zHuzaK{AX<Nu{QWh8~kh=e3cD8)CP~*;IvW1f2}t7xp5%eHXHoAHuz2({5%`H-3E`_
zv>0vL;KObBoi_OSHh7l}KHdg*So3wZ5!wXezjPbi9S6d7+29x0;2Adfg*Ld`2LCr3
zJktjE*x*?<`1fq^92@-KZE&9ro@s-tHn`UYFSNlgw!w>S@R2rnnGJr44L-vLztje=
zw!yEl!ROlGm)YQTHu(2#@I^K_^?3Z(XoF|RfpC}F;FsIrO*S}f9r52YHu&f`5Uv{j
zGF>&&&i=s#4z(c~N;4l9z>Vn}Y)J5SU(pXJ-4~_esf_T!$M6gy#oD{^yXZ(I!c>K^
zHWA*>FjZhID#BkdOw|`#DZ(E!OqCaF65;n5rmBlIitrAGsp4XFBD|GhHrrUW2(M?D
zDlJwf!fP1jU>qwH;g=YuYK!?q_&J8D(qdU6{40j3%3^L2{wc#$VKJ8oKgKXsSIi;8
z|G_X-R;=?l02e*XFjZBoU4$QCm?|pPCc<+VrfP~sMfe8{QzgY#itt?w)4UmL65-nz
zrV5HRittSgQ}x8^MEH7!&tbS)geNde6%;EI;j0*?>WLMK@Z}6s<-~j<d@;jRHL)xa
zzK~(6n3!9HzsoRHOUxz0XEIEc5_5>~=?qh4#5%ua`TH^KX1HC1kIX=rDk9b<!uuJf
zYKTQe_zQ-q5@IVw_+y5t3Svzn{651MG2AG^I~b-)h}DVkR)(nxV$~wNo?)thSeXc~
zVVGS%Rw%+RG0ZL>^NH|t4701pvPAe-46}>J+#>u_hS{}aE)jl=VRqe^Lxlf>VRqSA
z=P^<L4701o+C}&QhS^19Z6Z8}VRp?}6yd^kPs0LX{sE-c(WbGM!8NPUD~pAE3fDy=
zh<_px&!C7me+ig!Yg8GWc1;Mqv?VP7hre}YO(4{oK%AREyi6jFN<wUpBNoRd0dx9g
z%7}VFhy8i-y39wB|JjNB8I<{7kRZFaM&$nu9Fae3&wL53L)FvXW?V$|sQMW@;cXEZ
zExuH0gccvooNxxQ|F`(W0N>&pBI7b-+%OsE28IJf3=ok8?27Yk{se-Ql6Z)@wWK`^
zhre}#lT69URK`h8ro*dbrd3_+bSQ(7z?Z?U3|DmV7N47mk=d2NlbN~V#lAh}MQ*CT
zK{du7nsb4p<8<jyA@zoCXiEJk<9D0QkWXXbx=Sd(cL0<1%b-=i?En_8%VPW=67dYi
z>mR`EJ(|A#PF4R{4S(8MSfsT?I|F`j+*~J|mcn5^hxrH)znQL#2w;EeT{NAW)l1vp
zsNt^fsQPF3^}A!kDp4RP@^9`~$EfahWkeles~C#oH+Q7~U)9&E&0k!jI@{FkUA0KO
zMkd~Y#HzkorvKPVUpM|q^mj+?;G2=soDV79u{rHW06D+j!1kkt$G1)eLg@Si3qx*R
z*NI&);CHv(5ql6xlo5_tjdXAJ-|D}`f3v@6vbylgRZ#QVT14=7jsB|VON;}m{<5c?
zZo~>lWZ3khK<46VVxV$HsmA1VRex9Y9?H-DRMo##o4@VHM&hmaM47y$-qQw`zfeXn
z_ki-qPB#b~g)B9FqeUR7KP|B7wgOv=q%K!=nNs!cSM|5m=A->W=|TN(=0``6GLpKC
z@x&#A&%j-qiY(T!F`JLh!sBoi=;$q)u^?R=c}?~nRX<817H^GA3ErCE@)sH$%>7z<
z#P3GRQt+ngn^b+dN1(x7O7uwJDHK|!(Tp*gacjDrIaS|l2EbebPbOOK@qsfH;uG?S
zED_wQk;NAAz2*pHx{jUF{ETJOThvufgjf_!AE2&c&t%~={kn9~|LQ#}0aFHp6}YOX
za)+by8Zs9p1O;nlD*OoJZouYxPdh<O)dde-6E5S^$sp3Wm<xV{ai<k&G6_;6Rw4A|
zo-6>2dQT1nqO7R*jDrj896-_RU91d2wY<~JybH>v-m{ds3L>uxzgc*Z!x6M720^i!
zT$C>pB(t}u4ZEx|))5W@jTV8>#j3tjZEo+cMuzy9Q77_VTL;o+!69_vEe%ms2`m<1
zRbT88=_9GPizLog7K)_kE`ocoSy38}V?;(4bFD0bs3nS;)X6t%jqC;W{pPdR(OvFQ
z3Gap7V!}ad23-D5!S2rw_QI}6pUf^vk=-)5g54V=yM<O3Dk_5PgjaPj$N4vB@1X3U
z4;1ghm4Lu$l1`&^>V)I3R=oZ){1xKQhrcZRx$);xyw2|R>cZwI@WwWh>b#d(Xg4{D
z7y4jN9c2-YLi;R^PEqJ*51`NoMIF6t)zQlykv>uAP8OD==;pxn9o&pcnu2MRss}ti
zWRz)TWGoTwrD4bX^6V%It?K)xPq&S!|L^6uD0fFKG32(*dr{Rx9+h(FLZ?*A-?v(5
zI#~r%85UD$g1b&3f&c!dmueR@5)u%tl;kJ@7hF;8#h6zj8;6w*8EF%YoZM;J2?uT`
zT>ehMPDXF+i1f+qkQCW5;R<%PWC(V+r^H`RRTPV=_+Pfa-sH~`KSMSWqF==j%Gqi$
zaiZv>$qM3<uX_=_pb<&kWm6C&$FfG@in5-MvT{_Sl?mCcw6e1+2r@<m@eB-L!Sbgq
zJyy{0Y3Smh-VWlaP*^AMvuytf{D{o1$NGYrs}oWU>hnCA>zd#*-$F%ku+>XEg%-tN
zWOOvB?^Bx(r)ukRAN?eU|AWek!)l~HT8&{`UHxMT{6szUk*dF=Ht!jrc5h<OtzRU_
zk8H`~^35UIa-K_V$#XN9556ZQ36Ev_a^DI3s`^AI`@9Ue851Kzy4B{7Q&s0yRiEru
zBSm{0iLk3<81a<Bd9I*-SZ)3Y(ipS_$=uNOI!5$+GF$Spn6qRg2aKq3My@-Fk@8*`
z$pJOr#7HWbfUK$M^HdDNWI}{H&PMJzLA^`XMQW4?*?jttePK&paZmPT$iBMo>_6>1
zSq^2$enxWk)hA_NmYjXH$bJU0pW70yV}AGv0fbpQYD;*j0Ll8e^&~NrkdGZgnJ%%E
z>Ej|{Buq`O>W|Zz7#ReaK;q8rs&0^?8VOP?0@N`=mDtK)k*U`Fw^WfrqM~|?Wuo1y
z2F6-JHC;}<*CO}*%Ec-8o0z&^nbHbRUOGI@gbb39!9h<}_AYDw%8^QRVDv3^CAIvS
zE>djR<6j5Y5z10^Gu~Cim>w^(`NO^4-3`$odBAwTir7IVzl*aoGNtJs_z$*VlsE&6
z`JnN#$i^RDOY)BS{)v&9EbwQkYRYmCJ3MBoLH+0&DW%=2{*~E?()hbKstY&y5U(!W
zz_|Reza@ove<A0h@Da2dw5Oo6E2wv%4gV}?O!E|~P**IFrVn|;-x7F3bScf4_l66#
z=~o5^-f(FEyFtrQpaZPwfj2Vva7$UvLGje?4dJ7y%HqFZ1r>f(C=iEg6nWJ7)2zX5
zI|jFZ()gAb*v86P+`Qkbk%w1;Jr1!soB>hJLCk>W3UTeBJ`d`jtmd4QwNghZSremx
z{)M>`Rt!psLy|QUQ63$Ov1(!@_&X2^yzXKPYGg7<OO7yA?(AK*@me#|)yQaQ;?Gm3
zD^-KC@hHl6;aQUH#&cSB)^w%1xoNtxK(-%**2@s<2B8fy)Gk9CWvESrs%B>UfU!x!
zq7oL$E(C0|gt7F<WnOj}U@a2XL|8zn3S?JL_bb&=32n5Xi?i$KLxjbY2qJU<ECxK)
zptD|jNpcsw)t~aQhmV7NbRU?W53=I9<pA2@YBwmxZ34`BdV4_0yrmP5^NIT9i%O;}
zykD8JQOOK?s^NKaztYyMWXdrE;jKj757Z)08G?U7^j4*~l}Xz{AD|7yIQk;W`!+)F
z0%5&U+>G_wpf=Py5h*rAZ3N02a1&WXC44L3?*MM6O$h?D0a1T}Yp2RZWI&nnw5JI;
z@a)~6ue5yz&k_%%nXgRgB;+soO7WioUFNBi(8Gl82K4=a;$1d~jZzIaXQ|=MS?a>W
z?Lbl*-l55%{B<%d;v<7JD7XH~$kfw?u&}UIMg~rIG2pTSZU)>|Ad7)4E8t_mX9Wrw
zD6|4)43t@cY6hxBp#F3+%W$0)W|0_fw8BkRxXB7rQ0VSZHc|M-E~S!D7Jmy{N5@&V
z{t)jlR5ildF|h+BHG&?WIx@(q=>7uzn5uuM)c=aw6pC7OtLDS~)m6_wCx5E4qE$6!
zdUCR(jhxd^O&O_wh{6(qnG8EORytg~&4X9d-@+<h(+>vq&oxXJ^!Lry;iN3n1sU-L
z25^>*qGpqGuuPdtq+~OfNug#wE#;YsC8LOcSxPbURUtsMW+^;)utWm}JZK5hv&<vp
zRGgv|Pjo58lif=3?O953xlbv+r%)-bDN~9ctPUu}V#tJRP~31IZ<6lMC|;$wp%615
zr8q*%v{L*;wQ%mJQ;I7ZmEu`VN^$K<rTC#J66UwzQP7UZ&FE;wcVNXN6MJQ1uT1Pk
zVlNVVk=TpGUMBW3v6qRxOzgb_6^liKs9205QLv)swki+Tv5`iG{ZTan94}ILs^xX8
zb{PD{*r(}#(ewkM%%tHjs2?)#Gdc2!(H9b5g(MCYBcD$UcjuqM;SLmG3)!g}>EYF+
z>3}rP*gX{XwnzVpkQNE<bY<@l18UOvp&H(N3{d+6N?YKHI(VCvTem<J1g}`>hfle6
z2LeS`v?H*6y0Ui{LM2x;BJ>_YpCUBviYP)KA+!gfnOAfo6hr7BLZK^~5Zb5Qib+D-
zyerxeILyqL7qrB!%@9^#3?v#H1CbVuflN!rK&aEkL>2GMF>Q)BG^Sng&KuLIcvHuy
zMkbe}n_1OkpOF1!JZdyZ^V12Vj##D|i$!P1-eK=gk<mNigUc8wsiHAra2ex<f7+P;
z)!;J52W@`B;4-F{!DS3bf)fXqF})5hW9m@Wf8yX08GRE9-(O96Sr|oLbCY~hd*FC8
z?J_jw%%aQCmNSQ4_W#R}bIHGC$hqX7Gvr(%hn!6qZoX}Y#FPfk&(!k$aUF;m%e|B<
zhMR1{ggzV@$hHqR*>1+TsR<EdA8xXJjPY49_TeVGj4@?ajD5HfvkpeqiOBkFb_#~A
z;U>Gu3O8Bt)Q4*KdTGj079YSkQ=oq%hm@{QySrDTBSAU6>M7*x43j!e;y9_>g{?CT
zKImY@fMAf=NdyE1BEm$ZYLe4A%q|cX3m7p8FGDCu3$m4J23@2wi3Q6e<4LXBf_Kuz
zsKhXTlEYiTAxAJ(&0Q|+K`O_fztMQ80x_t6MT;P&t;I_H(_$Pf!DzPyc5dq0Vk5P|
z8tjN|4R%3zHT_$Rpql;-Mm_T*Z1Q1X!?+jJ_iBmJGhfN<kH1MSJZ^F0ac5RQDUpL8
zJaY7dM-G4R$nnpwWKQ=fC2|l%V0NKWB1b|5?k!VF<Zy_<{A#5{j)@5Ts1AiDdCn^k
zepFN}a$!Xy{9jEXUJjH9|C|-eJXz5gxWXLq;2IqA;6BbV5AN+4^vW^l-Gf1|27}&%
zQKdMq4Wny2M%PZI_&lg$E3MZ`>$TE)k=Bc}UZnLRtruy%Nb5yfFVcF?Q@fivp{Jov
z)!S4r7I0ntHDv_`K{5Ji3pYo>rkV>h*3rY$*?;sRk5AK&W2_8!rHVCFe66ABo0P?m
z!BWEpnZf!*gM0?{wO?_i-G-yaX2VQa%H83qj<t*P{f;%<UN)OQW+O>EQh<#i!)L0-
zQWp+C3q!}BAe_1fa~z?x|H#YhIG4Lc;g5)&0?e$nOb%Fgo3gf3P&J?UgbcQrPl2YH
z^8@CL#>$CP_8zW`$P~G)W|1Sygz;7E2VuAlqD+`Iz4Sxu)^A$)tqGKQrulWHMWhp%
zN!EWt*11z|_Lr<*(+lhU!8(c^@MK{%omQbZ5sJHm0X?Bu%iB*?thLMs#oCO*OCEkn
z9=Q8%K1_k2NUEsC<K{~)i%-*J3<p_C3vyka!7^aUuGI7Rf89euu=#l%-{sE7#e#!;
z)<Imddxc~v_3;)1`3VMINHE~E7!YH(e%#DXFmQL@3>-yoK7oN}cT)s?Fz~7%)naZ0
zmr??+Bp9dz45F4xsQsdAqnhKwzq463jrUo;-lZ}3WM+n)ika_Q%!~oaAZ9CFXySfr
z{-!)So(&&Rn#N%zf<=QS;&m$_driy%5%1vQ8Nb9HGHS<D3^WYg^|?|PWG>Dv6G|X+
zNv_lenI*Y()M4W#LdoLs5}{<tc!^L_GQLj9oR-_DluR4nsASH}ZBj~Rj&D*jL%A!J
zlF;~-O6I)Ws8TX-d{oH{<hCg#f$?ovisrT}CBgCSm_6inDkVkZJ28XERX`5?3%i_0
z+^~RR3i5IP0LIFpcm#*K@GKhY#<OH-mNIhMP@gh#=FmcAWN2uaGIHL~YGq{o&^i=V
z8M%09qi~lDZ4&OXp%g~<hM*IgrX>o$f9#WTV)zX-jI7_*MEy2>C|fwitfGOnVdgCB
zv_EU9_d0z6$R{)~^@CG1u-gTx7PB0gnO{{PCyZuP7mbtrsLoeCS<GE5<?2WWJ+2tB
zX3CYU?N>cM1Q4wnC7w)dZkGTDRwR#0CMmFjIWl;)6)cm%?~CB89$^3!Squk+256X9
z|7Mo`m3+S{tY0$0aSI|wXOZAb3nF$S1mt}SB6es5q}773ok==>v><f}$kP^t+E3zq
zyi4ZZlz_Amf{hYv8}BwZFaTlr(1*-ha6X8fHbT9~)K6Qfxxp(Tk6940nJW0I0|Xsq
z;YRLb-|jE*-{HU0UwX%ebvIn+cdV_(V@()sWcB0lnKeMx^v$vJAn(}Q0H27p{d!;(
zSXgJF(un;xVAFlqa(`~Q^>9xl@UR8~gDe84NkkP^y-sB22%W+5(QDohlpv}&GrqeT
z<BAQ2wpa<et@(rZySro89EPuqzE1jXp|Ac3d_nq-(l>#=`A6Z)p-=smE2B@qY~KyA
z8;FSp`0ywxbE^TO?X@F1i80l9&}xU9n9_LAYKDK3p)y%9&x+71J&kZF0r7g?^&V5&
z8{z!|-<m**ioDx`^fbboEl5uz%(EapjWFAS^fbZ|7Nn;U-eKk5(+DR5Qh;31L%PkY
z8I3XQFR<Pww?B`Su&4c<X+e70p92ss*K<&z=G(~I+=fT&56}pxhK{JJcLeosStBTa
z4^LZmN{RIF^d~?xeXpv25Sxl7(4*a_H`>l?gujIO7zP^H43Daj2hoeRnZNrG3p!<Q
z^lN27xE-yCZky#sEtfiz@FvS`wA?!B=2asyuZ)jEY(KDx^2Zc8XvB0gEthJI;dHn_
zRpIcbwxp~KQrXp}t0QYUb~PXGr|LE8YR3A-*cCbtEYC0t+-T_rc4-V@$x#i1%_bNr
zFAPRr5oEp1*xTwCgwVX(?mZKlLJjY&Gtaitb(~Q%@cKw<ovO?25<_g4fJ_s}s3)Y2
zys{Kf^F17%W1hCO0z2{wB?C^RW022e5-t^caV~RdoWQv?U<eS4119K56A;ya@QjJY
z$?3mCEqq%%_31U~GGShEF!JkGnYVO|wv+M}Y0r<R?I|`fgOT62%jD8AOooz5kK~8U
z-RGZFa$>^tG5pUsAcm^^Sr@|j*;@)4qDuWh)F*cz(}P9<<We(3HJ-0Se1UN@1b91)
z`41Kbjpr#;NSsp3&+Zn}k)9TtXQZy4pr@{#pa))aAg~63brX!#waSVxr_;K4PB0J)
z2v4c$;wd0c<{cXMN*+@_IpNT_bx67zNMunLPa}K`{t3#8)OEV>=A$EDsu`)C2}aTC
zV~8kPb!>t@WR+(02V|WM(lK0~-xA0zsjJv@i52&Dw-n^IC=ZWE_bbqwsRh%~GDM3s
z48H7`5briN%%RjCb+D5LjfpDdS)bx`SPg6u8Yfc(BhR~;&<IP%$n1^GUNhH5W_&Uk
zra)!0AA*c@456^&l`XhAkCnZDnB<1GLE5;6@9PJa&J;n>X~-AT7Bp;8>i-KK&2VMI
zy4$LT+q;n9ZKeLFkV;VBriMT7QX})RZMruF=cl?`LL*dc;f6nQ)((ZX*#YFV$3Y$b
zUGTSd4E<WE`Y!%%a=n%g5`b|-V5=JL?pK2jvbiIG{m19BRPR?ny0Hz9DeZVn?Zm@7
zUCHsS=i_xgu=$YVdlLqK9}NDTwJ=U^2C1F+gFQ0G_daPWXGZBq9@VW*`dX(e^WV~P
z_tmEAT^)1P$ZV&U8-x1Ql<VG8_4icoJ}{r6dcO`0fOcxw0-J02urqWnp8aaU;DL@*
zWyQFc@b?_|%Jr`pQoA=}=NS8QbH1l{#rJ2>_J9_63HX7R()kz!K2w9a+va=<!G`>r
zkyT&}C@_Y2G~*74_&e($X{;ix{m>^5jeEhbk9)=#@`7JkG2}UY$n$#Yv&ilRWE{+W
zbIylqcw>fbG9%8T?KVe?IqY8ZjXx)k4?M=V@+>qDjtQJLMv6S?Ya<ge6}j<kj5B~`
zsF4t~*m|VoavUgg&DDIrM$LVz78CRkFfbu|3}<o$x|u=Dg&Rrs_6=${+9mWj#dhv$
z^pEaNr8ymk6_z6&%N05u1B~>@jorGdVatLSxOi^@&}`a)&5d8E`i72qc$}1XXnGRu
zoW5v(kBM0ha{)@!feqsp^8kAU+&yr6vOZhb4Vd*s$eM+EiZX!v1YEIRm*Xi;KZd__
z`SrK_G{JuZ;ZZNBdWy9@&C_Umgi)*Kd;|^xiJfZe6dDwSu&$e1+Y%=Of>0Xbctl_p
zww`fbO&m#Z<Geuzj(OBhu?}8n#yF2^48Vff<u^i}^niY(yG_qS+TAcl1@!#^{jhl!
z7U?u{g*Up`8pOnwZ6PrwyL>x~w%1&PqMkSv6?55zV61U68wSQ_GzisrMQ9K)<q9AY
z0HK)#V3Y(1jVAyXN<ceWubk%%mH;tg2@Hn>h~ZNJ_P!;EiUC3ZK9qpE1dTTZ;K3P^
z;yM{Bl%W@8C`*QZEkk1d+5h&~5(IDGFuk$R{3*(9K7p)l^(oG?hQG|jHWo~LEdfsk
za&*9jP6YeqXIET(0S^GE4LBzU?Kh3Ri3eBEtlY91Q;-3MGafjY=5ZG&fn!G6_mK(%
z<D-`XJ)+7J72#%AxXZc5MY_rj+o?di&Hvig-E9na5Pg^O=y5d4e0d@&^Nfk#(&V04
zuWK+=7%Rn)*7@Y2qk6Nx+xu?7xmjlj8lg3yulF9A-_N<7+ceA^=^f5i&A8d68TY!>
z@-1p)TtpNrGH!v2qZ?{u=GExNC0D^%d2959@fcf$u$`{MA`3>^J=3wa{{Sv_WM;RE
zeVgh7SmZx|1^i7|x?iYFIm8`m>=9$bnme{i+X1Dx(@FssJ3ZCkHYrn%z}*o?`vPd+
zD#gd;=Fw2Gb5#8Wu)YM=r^MO;Vt)qK7fSJ7rMN>4ub11o-ES!kKY&Jo!EzIgib;ED
zxy{OAj>kb=oWID&p3nfTJ8BGHI_?68KX<3Hcm)>y6u2C=KG)xalFyD}<uL%Y40Uxi
zoa`ty{Y93qUuY0~ex+CuGXPb`Z1ba3ga;|y!%GfQ^<Qeiu8#zBvtVCViZxhKHCRzK
zSW!=l6sAIn;jh`3`~JL-u{W#fhdR!Ji)=auqCs`^hYw*$p?MPb+s{Jl)Qqdua0^a8
zK=;a%%c*}x1>J@yu1yAG-iXqA%t>!T+bfx=9y~C{oxKI>ATw!j*kG+?{$^eXT&lX~
z&=z~z8~T46RDdxFJRTmXJaW|l7*ITKK69hbu_lT}ffN{F%}q!YI~Q?ab5SGUFH-x7
z%|Hyh)r*=Wh}(hK2&`VjJ`WUb2!5n^&r-a%yI?bRBT6j;oSTSDr&c47NK}`UXmS=(
z`H-p*0s_9u&A2UA;zAz+JFLXT+*I3zcC=t(86K0X@%R~uVgA9C=^`N~CI}Q_?t_%@
z`P{LyU_B58*`kz)Vu3ZKq_GKU!Ix69l;u#o17X(&|F^r9OtnnOEUZ>Cm)41x+Y1o|
zCMRYg3V-CQiu_0h5Lu1LGL%3m!4Yi`FLWs-$T<@^gApMOn;=gmkc1SaJOsa@WKy1j
zYf&~SQo%dIQZQK*447MRJDBTJDx9!EGR5vhx-zD#j+X~uH4B0kk`xu_V<=vOYT=8q
zJD>}dOdrIgLQIWPOxUmYV&C4IqGZ;w^(dL5#UNe^crwH|5n{}P2ob5eAW0;7T&A7~
z8RmhY53#BbQH8Plp|FxrZ8`LyLHs?0*<Mi#s6bH-)v|)E9Li)(iB==fPHYV&b>uYm
zLl~>5Me(-<7SnXNJKz>#Gw9Y`@D#hz=#`SPET}}KWQI?<^`KHxjTYRkl*~oWFp>I@
zGn<>pSu{5x5z*Y5kO!!Uys!Z?5Py?l4`s{qmS<td1v08ZksbuUd60@Bh<L@BKE?Yp
zncOQDtg(B*s%+b+(f?6n8OS<!v8|vTh_<4l8KCc36<>(<Ahf%vc+nFQ?Inw8eN0>U
z_Zv$>K}<$@Zbx~3&GH~!x?M-OY&t@9v*+SalD){-sS=fQzi4{BH!uiLHo#Lh1BiGs
zM2uc7nhI)>J=TXxgV529MRyjK30vcfeTnfrA!$ogvQJu)vO1(yIME@KWNfKaTfdSD
z)z-C6(5OAi@$X2hcM+p_5Myr|6<{B0F8X4u0d*mZ)km-CQH!EM%Vyl0{)Be4%Qc}0
z+2+w>5|Zude$b^oqDN~I=Xv0Y^E_~6g)g#kcoNZM?ve;1M~kDp!jp8>2}-Be=JYt*
z8@gYUaIBNHJ8$2xZs`<QmfztH!~19ytM|cYp1TS2?Ax#R0T=J(TkeF3GW@FLJ`t4R
zqn2A!Aj9WRlCJ9p=`OO|(n1+_VxZ<PzWz+CUlx8@2e4n?=GXtM>R)R*rt@R5sxWHO
zG572)%-~wV0{F#bC!l{E(DxxGcT?@Zapqg>Ukx3C1wZ#M1Nuk#dZ#%L02S7c8682Q
zWckJE&m@#77K%?!)Sugkc?VF_{j1SGlc`;1p=Mw?Sw}x6H)q6`GqGx<j0f;C8xb5y
z_D(gR2R)<x`WraMj(HVrQ#kU?RW6p+JQbf=Q2vc-421P<#sp*ZO4vwPaWFTn#MyvB
zerIPuk3M<mICh>rFivVZ-UZv;eWNR&f9A(4Rai@@uxk~<O!a%<yH{~>ivzv`uu=%*
z?o%G!jylGK+pmA2clZzOQ1v5NG`Kaf942lM=iZyYaIdA``4@fvp&kC*9e$-?N2KU-
zCr)2ztGXBimQ?*O!3#FyWy1^F(CJ}o`w4;GeUZ@v)bfK`WajU`5maDv!0Ek1LFYEL
z`71Y;1&0p3>(4y`nnxn@#(-vv8vYxMHG89&hn%hKlXD>3`T}4~(5h%%DWG*FpXxSx
z9v)lpm<~J3F8oc$QO;fXI}U5h@z>Sxrglt6yK2`5^-UeApuPp*W{_!B&ZHMp&zyen
z%zsPG{i61CzkUps>X%UhtoeJ&95~&nZw`N+g4e+?@%^;nAgnPTbX*fY?$kdBT@*gv
zFXZxv#~DuL7weTj?r^@Vc5nH+Vf`4+3d{*2N2`6<QUshI`@{2$V;~q+{up((G1L_&
zzn-CQ7#C3f*zR}!8FTbwxv>Rnah7rrEMSkm0{+AEJfr(5OPCxDU|4?kLAL!&bR*gZ
z&}8&4fbI#vY%)ymK-UTA$Mxn3k<@IoD!<-_^zMH1ChDzhWc$3^*RZa=>;1Xglwe!K
zj!<?$_h6}yX6tA3{fgaeUjQUWn`Z$P>--L{Sg<A5hXPqYkc@jbCX;&1-@O^FQvX0{
zxRq^BP&L?!OFS-E4>VX5^c2mw59`G1VA0v6dJo{8%>LTi__<J=(D;h)UaZH<i)=!$
z|Mr+SunGuWz<%tL>*250*~}Wg9<q4*18;u7%tRB%$1rioS3hz$humtK4~(haPVLa=
z=obUjNGkSz`P$y%^e6-)5(<-sYFoqftW|j4JgNWB$k*QtpkiE51B0OCM3<)m$f9a1
z{{r=-f7R`3h_0d@^DFtAfINif7I1pjSVU__c-)gX68S0O1IDYKCURjcY86fq$(*Sg
zk1+^4%phgO=%3<!uR%C~?}BaaNvI7M<rVZ~z<68_ERmWf5|YLFBo`Z#zD?T4fTt>t
z-hr%a{gE$_`BbBhh8APcQceFW?O<6!0~0BBVw@Sru1mmEM&bp=3#;gie1U3AYu?i@
zzhO`4EUo+iE2dD@UpEI0pd8=RXL`nwPzEdpz?l!V81%Raf&Ox8h9JPUR7eUJ2Xr&s
z?yS8q`(VeY6WW_h2SjWhc2qR7DL>@U&ynt0aP5FOT>r!${<hzoYc+kRAFV1<@+Vpy
z{*XFd;vdr|9}@|V>CGbfSu!~+qy5zIhIA;Hbbojgx=!s|IK8PElT?_MX@<T68Gxv&
z!&dyU&}$R`bk<%(*1@Mr^RyZ%hBdk)1t+LOLuu0pe-6Vh_0-<zYd~{^z5``90Dm_~
z6Ey6${pMVQ1@;P9oz0#$28>&1f4-9%aym8SWl(TXiN|4m7U~eT=gqJ^M>}iZ#@ZRc
z-V`Ak{p+fCL#<1H2L<Yw-P1n8pP==@D!(~2h-KVWn??lKn~yrB_iVBj>Yp;?5m+1$
zCq43s?Y;2v)Hx%#*$@dmGDv?XW%f5Iv%6AeA4|Cme7UH^YCFQ8Ipx?p|5ezX(Bze0
zn`n(_5ol~mcm?IgRz?dW&yGW(Y4AW<MTT6hmhV8v?Pnd#Ks^RrZs%UL{M}?|aIhN(
zX&e3FZfEVq7(V>%SRO3!wbR=vyq)kmU+l<cgG%Tx?8qn$>?$J3q@E<gE0WNM6a7(`
zfSR+q*N2a!&U;d{Txcd7_(iW%83fZP#Ul2$?#<ns#twN>S^UB==ofU)42hRvb#f7R
z7vsVkXV_GWoUZM=7Spv2U}<$K4+8w)72wev%^oGJlsN*j+pK=AyE{q$6zA*Pg493k
zT`<q~D4-lm0ihqF*PWn+HghZyr>86x^ej{mFYdId9x9k;SyoX&Aa|N-=yAP-eFYhr
zF4a#~k9;kOgvUMWkS|avkMf{gPy3UqCSY9hjA#M@Wraykj$PyUsBsD*5ony>N{#bd
zsd2u9#`!dKhK6YshXV5{&<13FW?m#X9=jO{z2YO@g#o>tdgZ&&D<kFp8pZ1C_(6i7
zUiFDTd<@FO??onYO%m6ymKym-^vOn{PlPhrC6ozw%X-r!p$F9ejTXs`YY~o5|2+Ni
zrs~}U{qc6k5AFR&4SyEb9|N*ctUVZ3!{0hXXX2UA8c8bSZJ{zQNvMpsllo)Io!_L~
z*_CqVv6S;|`r{Mo4<4{}<Jd<o=Dlbk!<1j+bpp((Yx3A&(*2Rt0Rg9}E``0>2Rn0r
z(TvoTxoENI61M2kh~^>l8kaCBnse9iy}B0J1K589W5+$KES`!JD(t-I^KAH8iO!3-
zix?N)SZwRORv)!i8B6}eUc1|T1G_{$y}^Dy0>dL#LC3}U&lsAvW?YLo(Il6(ajWLO
zr-WYxT*bsj{H`g06;@v#^kP<j1m_iwtKPq;xgRPIuOU|M`#5Y7R3>z2&Uci@)(3O<
zDvvd%Z^s*mdoj5!V8v*;-ztxAB7c2k-0$Ex%=5t=F1G+-<{tERaT<nuF+$wU7mKL%
z=64}(Ei!I76!23_lzq^=^^Yi=^8+n1Y<c2n%C7+{(7&hYccxFor5Kuit*hf~wfU2N
zn4{vN2>pHY0$^y)J2AR$Ot;Bx-Mi3t=AZw7U6N~%RgbVv-Y-#kzXm2K(4&tm1uI-c
zj_t2B6hc=8%bSh?6x65TZO%P7Q)Y;FSD@NFVu<GH>rAEoR#Y!kfwJfdrV(#Fr=h__
z&9jiV*t;z8_y`ZTr|@9fK)n2d*;#@9Stx}2Q=FFndiYTV0`6w3SF#VT5t~!k;mSc-
z3L-N-Serd1H>`Hq;&+=Fh}XK`#Np%KVvoFRkJZA*Qk91nqhyx8!fruNUT9wyw!V|+
zb|l3lgy)mUVR9B(>w}WC*fkd?N8_BpAskgKgK**)KLCbB7)&b-51=ukhCjtjIZZ`>
zqg>e!eJTbQ7giV^j2`7wZQhxPi8M&Dpod~B8^}3H{T+3I{T{!V6QRM)e}J*J@tHl(
zry4(Eb;}yXrV}g$K~IjeIavOYUtH;f^|rO%((d{5Ts7qubgsSFT*J)E`~g_OTVb5j
zY29zteyCx?9KG77E?iH}xP6wDpZi(u6IijU3$dD_c$9|Mks%dFM$o8siAztYe_Un<
zDo@o>Vll$1>4FtZVz=;pKD@^C9lXY5=idmO4J9ACu~Ri(K3j?+bYmB0K4)Wq9(^OB
zavTWVh@P?Dr7k=|tlG`IW(_DUruKW}2u^aujt|_iXw%B#_1H{e%fs|?FS?X%i4r@9
zDUv`N)=gAl2`a~9%`%j1_5JXf3h^+tXn>tzZ)lXc`z3~Vo9Cd6>jvR?;M#NWSo0kw
z@bk3bhd?l&M1zPmbM}Iyz>XoD7(eaT_0%DqrvDjog`RsaZR$r3$F6Z6^EBY^!l?UR
z+GGoI2z>0C>G6HI@4d7<z}{=#`OybK&)oN#@vN4|vOud&9Yx@l-LVgLs>%dLMf{_)
zcc_POT%ybAe+_cOE7#YVJci%V&XXg1luQ$ryU@J^?F2}M@xqAJUT|(6FCOCMF82Dr
znGb;k4hMUD(^X}5bh>g^+C4}#T`5Vc5$?Pw2%+uaj5}T-FK$Efpl8Mh0ndz`XDkH9
z2PK{|V9fNCsok&F{t682_baY3d@M*;>R)t-L-sfufP^8B+m^eq$-2@;gNTxeX}q<5
z14fZ26VDPX!oc8PL4myO1aHIN+AVk+{;qK6VcP@e@PRyKCwM~wB*W_r;I4S*8J_?L
z+!X?+2=C%`zghb<xf;O74e3he$D-JRca+X0m~(;Aj=@UiW+n42$=^jwKxV&k=R5nA
zlH<te2r`oUsJG!&+B^RWa4W#C1KbK*?SADptVZUeIP<x2=~ho`)%5<BaV;?hZ0OPX
zjqy8zoamT0RBPA+ob!OT@Nfg_1?Mk}TW4qsn;%5W=YBQsw^Ac>aFFEs0cyiRRnfKu
zjUjKT#(iaK%IJ5s$e?~&?*4*vcj2iid0VwG<`Ld~LA|t0)d#&xYA|}K;Ufca?xs!E
zLA<-U_IGN_b=8nEkC`DH)NiN`8rPKx?BvW;W6)bcW6--dM24i}=tMAgyL#?HK9#&?
zF8G4p)}R+U5A5iv$nFMY*No<hLaEBNxar{585@%3?>Xo#Vtm7)0L^Hi&J*KXG0qDI
z%U5!I!wzjh!@=IiwwEBJ{Dy<dB0ur<PNAVO88XMbh#^I9!K*}AwPIL17*cSuK)g%T
zsnq``vk4z@%5bO@ew=5-dq$n1OF6u|QIWVH;JQLSe0*f_;pW4I#{eyIEoKNel<~yP
zcWI#?jQe^9`+fM!!k-&|E@d!`fvs3U{exH<)|JY`ZD0)q?fMjXoCAV=Wx$2``m&Hm
z8N84y1UOC7X_QW#aQxLEfi4P$fP68qcmNKLLpq%jxm`Mfi@^({2-%ZVOWZQ)6iUY@
zoh<24lQRn@)3+-PSIe;zFXOeR;eG}&c8U{>*06bIDk{#4F-VJ`#bK~?y~4>0EA>&{
zw+wI=+LDLIv$$vlbqu;$XcVe3%)|M<)oc<&Ws4qjHml{$>OJOMoXIFuonNSt2itIN
z$Gq_fT15i1lc1}UK<xyLOahq%Wh8+*2}(->;o^zU&Haaa$cuwQX1fi7^T0%UGYRA(
zXiX9*gP{LT0=WrVLQu-ALKG$wVay>oJZG8n0EY>u4D<Q?+~Z2ammDQIsz)I#aZD0#
zM?#WfXjhG5IkcN+zrb@FSk?x)c*dw63DRDO5#2aP)rSnk6>6xDqJfy9<1RILV6=qx
zfCpzIZbg!!fo^!J>B)pAgPtsS=F*b`kDDH!X8fd1)LPL%6~Iga3jthY0gC}d4Hpe8
z1CULY{uuz~&{GYdPzvy)HgEN*`jUa92Y8b}UNVrB5T!DT1Q7L%6~&~8Dr8iXj9O_$
z5d~4jj3NxF*zgz>ZhS8C+H(*mj2kmK1F6FzC{~Il4m)(BJ;r6NG*q$ia|WdvGb}N`
zusm7JnxUxr;^8W~g=Rp3z+Wi+5c1;T#nKNsFCNY}fPe#0FCIQa`XTMb!>gqq0$)6Q
zuJl9Z!Qpj4cGI&+Gk#nr$`Tyj2q0u09KIAlDf8g)CIBJx;P7Vvlrj$vUkM;&9vr?3
zKq>R!@F;+gd2o0ufK3+7HUJ^*;P9OQN>K-gw*v?{2Zx&gN+Ab_cLE3r2Zwh7D8(Bb
z?nun@rasH2x*-a>G@XwPQO3*BeyN0cfEQD`oW!7cM=YE8NuZNvc~@VLNt+jV9T*Sf
zRIeNJyhG0*C1!5lf4+}--a#OkFQTqve_(%-^E@Z8<@x@}7!XzK5cV$Aa!{5rX#AuV
z%m<Atvkw-)Y#RPVSV&>>>{jZzf1pKv=+Yv0;_m@h0d71@Yz+ay(r0|Xy9GNrw0Yi5
zbLR;BJxI?;{7s_g%IV5gL%U#3y=v$d=aYvH>)SECKu+`DeDct@)0NpUj$U<Pvxx8N
zC$Rbfi@^0F*p(uHDRH0(ngV}cf}RO_2P8e59lPqn!y?`k1op-09T7oO5ZD)|cLYJ}
zSnyF{s`V?!aV&Q;>>?ZR*W%W&b)__%g|-^hKLiGgo4j%5SLVO2<#yKMq~&|D2iVRc
z^KnAB3kRbxQ`yO5mflXD4n70+HX5{WB=}Ir5WkW7$0z-V#=Xdsx-a>iZzH*|D_;dq
zd_M|D+ljMU$Mq*GMz3*h^BaM+sQ}GeovLpKEVw4siTQL(zhG|5oHo7nc6pXFo*vh6
zPR4XzGk(q%Twtt?vdVLhhAx!zF>G`QBjw}2W$WCKq2=zb{ldadykv7PA^~fX6xeA<
z!7bj0w$e{NAw6*u>93UO*@M9W`|L^S&pIJJaTDne{)YL_ua>NuGE^)>@5_)+hTf2&
zOc`1$LRT|zx{yn3IUvHfL9Wkj^SOmz&Sjg|#Z|Okyh=`e$XCbV@~257hOtz<RZxPB
zH=K%}2e$G0c?oeNc;^UsJS5Dp*@X9iNq|!g{^)O$Ug6&fKU$SIV(7&R9vjY{>?mIc
zX4ctTU<c~>w-eUa#Nqqxf;pUyo&ikF_{CC^H=Y+l8~J=4L)hI>%ioDxr}V$@wE`Uc
zJ7jwIK);-AwU3%X*hOpP8}y`vot}BB@iUPg+gZ$nwi2<svlc|LqxeziY*=NVJTz86
z$8wv+E)Mr^eB8U4$Gw}9-(#DX^L^S|A3UbtD)w*S!Tt?W5;_)r47)ctrHCCIxp}ig
zGuDZcKm>2F43Q<CR^cx8Y=m2Y0|fVAGC-qIhE*K_LtOX;JM?ma6To#`Oe60--NOc;
zZ<I6QRBT@1f<qF-5<Id@)|NQHYEHwxmmf1V#z@Fb0uu0~_JFRjKp5=eSeIC!L2=em
zpzMRIUC1P;e{7C_9%hhwQ5cLNEZbOrCX6p0rC~1u8cx&Wygcri75JZ<v#=8@4%3>W
z1m~Nz$StY<7OW2_jX2&xvlh+8Z=BDi0p?|g%y)i;X1L7rG-;M^(u~qVP9SMxhehXQ
z0XGMyg9e%)x!_N6;l5J3g99VWxDuU9Um<-n=<|uysaQU94XP@x61;0axMU<22}L8_
zSV@n>G9h&&jQu04Nl-<TVL{N{SVZ2Yz!p@Xe+4V>C~%l~bM$)%(2$dX>(TtVf3@th
zR*Unu3_vPU2@S{6lbzBwigx@G#$OBL#K;XDcz+5fR<XmtwRaS#*wW)(jBLwNDLe}>
zN7IjfEr-e;u0tEm#Z2>B$JY4xX44PhFN-z(O$}EP(-4dUxKZ3T1G`jwuY(~sLs>!?
zID6~k+Y*@mHk<FDUbP5c`XSi7<yrKMlmbXe#n8*E6Te46p{1-uTFqE4nr0AYFE`eY
z`iENh^W$O@coj*9YJ*rsV5TkJ$9M=0bfza=i`<r~MP{TH=*<QC2U_^E<JmZy-00U^
z{CcaAcgz@tom2Eb9t3{~??&BlOj!Z@dvgHG8mkiulN)+s?N&4?&b?hh%+WC&1o;BJ
z74|z0WL-FRv906&puS7>9u1+fb3_ZBgTZVK2dY*2W@Sa_H3v?ZbhqFlN$!&dHSf`a
zNN^PxELp>y-Pb@u>IZT*5b5qerne|79y;bc1l{02luIhL4(D+#+$<D?<ZZ+8TAW9!
zg@n*k4WFCRgna+j-R8%A#^Bg3t9bJ-P_+nJgTkK}?A7{j;r~}M{v!;cg{(tk=&^O2
zVBanj*DK+skM4rpZ<P99q5{!bj|IKB>U%v-Lw|g1t%1s7Lwxt=X#3vJHN3W1V4PE6
zl;SMYKvwj^<Fld8?i=Y}4K?Dx8anU35&qR^EA)TwzO&>3D){@~H=rOgsax9~EAYOp
z;kIclGSL~#-KQxNwu!wd_WT0<P=WU?ZTsGdkzu)<;zR$!e4*pK1YTQs2lqLR$=)IS
zZF$Be^?Am)<=N1K*W$uQK>ee}Ez!H0&GWLmDx+siKNlVFV8|8B{qWxNV2W_=?YMyJ
zReE=K;4Lj{{^vJ}UP9F&UUk0f+?1zZQjY;;l%(h!w<J%;P6Q4e;l#?rocd3~*_L2#
zb4_glZW{A$)wX|8;KUt-iTz#j+X`-Qpb7k*F6iL?Fu!rhtA1nH%T8R7=(qSy8MZ{<
zNq%=$Mk|csJM<0Sud_`k@mG=ZG|55suqA$d*vqivAk*4kp%K`(GBAHh>faf>&y<}>
zY{jDe3**QD7?`Nfv1ijE=f&9vIXGhu?8B*#QCjX>A#toeC0`$d$zp0yUyvSaLjul^
zurr9cUvpPK3`64hPEZMatT{I-k3}&+;26u*$`k9;*Q?mh5Wj1%1Jn)`CQa<2%Xss7
zB2LTQs65gD%GeS54Oub={elA=EabINzXiB9e4nbn$t@wdC3!69G(#x?9WTA?;jw>d
zWZ!BYc#5;xg9VX!ze{C3qr`{I#$~eCU|IitmI2o+{sslO0pYKN62M5Es^;Q=P=B!6
zG040I?K<{D`V~kv`K>Z2-G|VGc%gm{>7qMT1N}sK<25$@0Qyk8j2uONfY*F4R&zJi
zd=E=knu~$Zg2P37Ax{*+T*NEua?P5rS;vPu%I>{I)4$T1zsd?mrg_p&*pGld#<<6R
z7=k|2QR=TaFo(c3M~3E4vPkj@l63CG790rQx3zvt?dA>)tb5I2j4%e_78B<QMU?tW
z*w2e5Lm9jUeIQ4M0xY6cu{${t6=Hwi7^~;q=Cim9Mxb=`ys?{(cSW(8cp^H7_6JTx
z$5vfB-lq*^i>u;spEyfrz~u3r>|KpU>d)~C;4jb%p0vFBkSB4fC78Qm&L{EakC(mT
z=OK;$YGo@XDVQ(D+;}x;>-I04TbsW)zj7;9j|&LEdUx$l-oDg{8{+n>mD?s^J3?Fi
zS+&FQ98<Bg;u>Hya(r#T<TnSJ&GXgDC{wfS4F~;em;|kQyH>fMUZIeX1}mc8(rgY^
zZWd?@Kkx1)U!S{!r$Mq~@SYCx;@djCq|>}nt+G{;|2r&I?vvy{lH@-E`DV0obO9*J
zQ}AUK{_hnzSlN66ef2bSO7^0EDgRJj|5E<l|6=|#|Hb^HpZ`ncpYt!~-}f)(U;P*J
zU--Y`UuqO6l!YgB+hTdVR@GsEZ@F#;bdtlqKh-ndKt*G+b{s>iQXfu*5;N|E*1Z<z
zH}LHTd^lyz-ml~j2c&Tvqg@Fvr-EB`q5SpJj4Q2Iy_*lG00&3(aL{#YF!y8S;R&D@
z%>4+<g92s3mzwiU3?#u^6X!b8w?gxow7u;K4byTrDUZ0B_dVavHE;Y0hoENAB`Y`?
zf{`0{sAFXpj7(gldG~3n^hNTAm*}-a?U*d|=LVSG#?ubi5eoUCoNLEHBgHPm^fRb`
z)<~1d*-~F)s96hQ@q5QHc+XX_jrf6f&&Ii$^V;;-LyW~nEUwi6g<rC?I3+>jRBOjX
zZVp|o8rQ;Ti`^I3wPoCafy<2?tP!vE!Nm%(zv7zYQe3yV71y0u7f!+7jaV9LSQ%o#
z+>9lte=aSiAO-V{KG!0J_qzlsycZ)#`A#NP3{r(4g_pVTi64>*NB^m$a4=ALxIgz6
zFmK#p*+Y2t#rZBSsyAj11-~<0U^sw_qXq|DU^#$$q~HV70bDBuAJ|SEiY4+;7naCF
z-B=>y&Z)sAE<B1{U_R&q_o+Bm*)kWW#Ej)bO-fl`?xKqY_|Tnn2iLoVi*TzI7B0eE
z@G)Mv$aOC66C1inBwjLfsc@GKeMTTHb4hoJOTY&&b_qOi3DLF8RqE@2MjnLi2t3dw
z4{|H4-X&aweS{-V;Ub)6g@uc7C$j?^4v~JzP?vC*4Rs5=WjOB1^h;a<4mJcH=rJ9x
zY*-ZM<kg`&fvM`Ry6BRVR*|yaC0vA8Sz+NK{EQVAF2XZJy2V513U|rSMZ#S+bg95w
z=92Cbmw<y4fd@{Q4(CbQ>EynR`7EsVwmx>xdbHb_>2tB`41*934gD>#-bwCXcd|q+
z@?L-TFCJJC{Y!Pi2qWx{IREyaatf*G-_i8h>DoP;=3@7m2d?7tvG~1G;$+nWjj-i^
z_L%klxqYg2s^J|gZ86)5Pu2GQgR_<Yo1lqf4*xenv(4H2m>~b(1Wm4K`kLVX-vq6D
z!T)%ImL0Xt*08?+dttpb8Q(U))87#XbfvAozt(cC3)&eMW`usIEc$ou8kBd$57MeS
z&b4ii&~(yLpkIAM#IvG6pDu1I!fJnr$GphE8eN}`6Nqo>ru}>)rc3j$>V7@1C3OWz
zI`mD=dqy}nZT}XC4ZC!6!S2{%3@SKflVAXhq?3_Y4I=d2hjxd%kG}@%L-!g!lpEf4
zzRy?rR(Bx;42*}{gBNVcKB!%=U8(1>-vZs#ysb#3+_2aA0q+kAe+HZFCcLn+Ck5u$
z?oFZ7R6WM)lR_EzD&iDait{izS@Df{e!7;~^LY21lItUv{O%j%qece(0YJ=r-V8dw
z1?kn2m$#g+^BV%#2*#WUq`m`E$@Uawa6ZQN{7(F};}0{@O;L<|owc7A=$m-8Q|^AH
z{t=K4=&?iF!-tPsrQALt61=#dz7^-0FtJR<+;t?5Fon*;*`?4xN|Qy05PSEA52wz#
z0-ums;g8(xRK<Q1@zGvdob<H#k4e$}@Ea|>=@^bJb%xg;YuK@%UEhS$?X_p}bXBtL
zZ(U9S&P!g6$C~eh!s?0enJ?%Vdn-}*iSd3FKf`Hja;x~YO<NedtGMb1Ywx#tHXJ8X
zxp}nJJo_vNH+N^~-!N&#)P85^yQ-1J-2hkt4w>it2TW$T+_uHM19{0Sm2Sj!o{T#R
z79_W@r`?GIn=NJzkmB~%$uKdQf5IKwAgVkO#g#S5_0!WIaB&l-J>vQy9AFsa(Q<JF
zHk77`cTFOAsc%gN3J!}DFX-?rLblkT7<2+XMB|vRc{Vu4w*c^-k9jbV?oLx0vOycu
z7^OawnP8_PT^ozyh0f=?q(3ZRgLLfftxwf(AUJLfg+02!xCB$GwxE}HI>Si0Nq+rB
z3B)jEf=!eSxj6=#XmfgV+$Pw44Qx*)ESP+y(h<7E2Iqo`vt1ksmnY<pL44c;LN<b_
z0{wW<`vo>_wB~MPhxXcj42C{%$T#a?N9)FsahCBb^CO60Qe@sMsa&fG<KKcus3C@V
z@O>Tb+@wCQ@rW19?AYMsxuf0YYe<6{>W^F-@Yd}%%p(aN3UI6$0dp8)Vt1l<#oEcH
zk=C)_4Vi<CYMe3RYxsEuvL3`5!ALuYVZ@1Dq|8oyPlM-iaB9|ep;i%hm@-|wUJKP#
zW{zZmjkKq2WSi(ukOeyHS)k7~F9B!RRAbBN*;G5lC0zAD#5OY>+a+AekM9*)<YiAC
zBFu9jr`R+?xYw4JZsTQK5-&#DXe;g%MGReoPSeMoV7;@m;0$hZ3RV{X2wWsf=rk7D
zNPBpn6fOlnpZ-21bUF&dI$RGo{{Dyfnkkn*cr6cYi5BbQa)Ws>u%IYE`#qODpgC4-
z8DP^fMUCL)IC1M1UV;^N@T>+L-{GKS-Xc@|0u36eG~*gMXbs?3IgyOt<HVnp>~MW|
zcS~9~+6EZFa--hU1b3-u0n+-*eey=Y);>%nwi?-}<(mp3p3RJ4_uzgW4&g1p$#A|i
zfqh2VX?og~(lCyOhuqJU`Ztk)d(CyiE%MBjWbYzb&?DHQ_kGZVr6e?xT8^Fzg_g4}
z3<ho+a7`}~c3g<25&Lh9M5N4hX?$;gEft@cCWu=v(4sDhVb-xO!K*~K^QzKvftWf$
z$R1O|7-Z4J82v-@OB{4$M?ohl^2}Jfhyq#)fihTa6@y;~7cKXMzNR$X!f8Y9CrbS)
zkrA#mV7fv{=0=N5u|?(rkf9JV<I)%*q=8P#vZkZ*%=3{1Uq^!|h6+(wajaM=2aLsF
zEiQ);`uBsB4731<*vt7V+ARPELv}E)NQkGW-rmFGU?{#yzI|(d%;G;O^1z0^#dqbP
zMDc-Ur`7<zMDY)z_-xqT<Q;D8kdbb>q>EQMg;+Vnv9%2b-CpI9L3wn6r1>2rk@9!|
z8=z!V#C0N0%45&N$@0iB@1fLDa3I=Q{R*rS42JCBG@CrcX0f>o^DNciPBE}Ev7HT@
zSE|xbhE3O&w4ZibolNxCx`jlQ-7T3cF9d@YrAD9@m6zw)l$KJ@tv)fSvMw{c1Yzh=
zI<ep2VOMdO=A8nYHjHHKrc+=WCt*K+3henOVgKk9*grf8`~Fj4Pdf>F<|(lGo>y`y
zc~G@430{8^_D!e29)A*c{wc6osfjBel^@510(sfPX^weW9R|900kI+c6<1_)_z?(a
z5Me7IHiR{N^(Kz+<pLl)1BeabCBB>(M|hnGO@P=C2HUfEjtJe!Je+Pv(23BMjIiB~
zFoOu|l9_C=BV0!We$w2=!!PWm_#P1=$)$M0j*v!#Fd(s;DSS)|FrrpNIF`49a=OZj
z>TNix-ztpLJ@Wvio?X1J2Js%^VqdUVat3mu9<P^{ZacB{ry$1O7<;ORSh7~E7k7py
zb#`8M+ZSgG5!ZoP`-a%F@Co*M8PVz&Gct+jU8f*gAW6l3L-hvIt#Q&PjuZ7SoJ5yX
zsJ@u3LR?27DSpvIF<H~rx1B`sx2K@^pOTKa<7(X#7=GlEYbK<-`Z<n9W9NxR5i_K`
zW8YeCB@)`2-z%hpKC%<h=6@rN5T^~{rxt<>uK6P(EJ~(8T}28}Qm7!p-F5^yhiqr0
zN|t(?U5M)>?6s79VXaNaC(V_yi!E|CNxb7h<h%#I1n)BE4D!7Vk!-%jNyAM&k6!-I
zvz61PmpyJ$4uNvV{U_wfS&ey{m1ok(d8#>h@5Pj*SV@ZMf@0z@R_+5*oJ}l+ZFK&B
z&Epy7@id7-rwl_~Vab{}KV*NOk+V;6V=K-MpI=b^@7DYRtJJke+4LGgV^ZXjPdJCc
z3SKM%Ie)}S9acheyKlgeJx4+Am)LW|G$qR{XCb-JX#a&ze83se*x5q~OYnK+f$;Gt
z(}(6*WU>yE3oCdDf<4?|Q8~o3cHks1lY)rrBclZ~*pJ(7?qJ^r8^GzA(bQgtZ3VwE
zW(~~}t7(fcZ~GdIEZoS#CRTLG3%b!g-5qyzVBwp|`$BrOVW>BRL{m|I%+o&XIjzXT
z9C<(5FTaLjLzrPlIGqT$0Map5XuA^yDJ}s|%KnCv`O8Dnj*yi;NdloelhTelIqknE
zI9qUeqQi&g%J%Xlxa)~OQ_cx*wIbg_Bt_o>*V+%{{nt$>H!Iy`&L#pXJ>9DGM#RJ`
zJ>6F6*Mgv}(#Meg!i}V38Eg{m^+fy4-wO+46L`V7OVfQHzAvOI7vmi$I$#U0a;Dw(
zJHGK*j5iuLW^8D@c{0@b*T$%I-$f(Kj+&QX;f|~KUF$B9;b92lO1E`B89W2Qbx|xW
z*K$o_UWTB)NdRCVaUnlc2@Z*$o3M{z#cQghi9_7jT<*;7Hm4wHi@!!Jeq*Bm#r?{O
z{z;PzZ&N;2pdTpkex@~ln+Chf{8}x7NqkR#7EPe0lb0ZG2L&2g*#A)kytLMBzGlJ1
zXl;xl=s63-x&aytIk-RBf;|kF7T)MqBYAEZ8yEZ%Y-6?kR~X!aUhJ>q7yt~sq0!t2
zhsiG7l~P-*8KXU<ag(4Cnuzo`Hy~kGOBhc5^}sHcuz$nn@H%iO#)eMN?>L(n9SWYV
zj??hW=tx04?k(u)vJe^UV?F1|a0HS!Yx6TdcrTduZ}o>cAtPA6ic2S+C=f3y49?5M
z4rC5Mj?jhjTi~!E@U!saJo2V39cSI}75k{F1h}0!F%!1|@QdaH<n`e=x0082uo)k^
z=2a6dd2W&lMjrG*cp2hzV%b~#tGb|S{HtA$PQ}{4+dQZf<<}guguNq7AHLYZIWT{c
zS6n9jdk)9qfFaMpj<_FzNR8CB`^_?7#;vkEl7<@*#MR|u@%4Lf&8)`a+To62@&a<T
z5Z2l-Qmr0+D^#WfDB|lU4_dE#ED&i9nFCM-R{u@1hb|>+;-jdl0qY7ZZ8J`~kUij;
zOA`wRPyRv|+rA&pl{K4{lxucPLavJi7-l@IDDjx9+CCD`cPmJZZXH}9>J=Dfphv0^
z#yvaYf`?tU>lS+2zc}PA#L1=iKlE{Pa%e7@!_LBwn={CP-uk#X4s^@UnqR_^&ghSq
zOssEf<6APk04;0&INNwC?gV#!XnodvulXtpja1^Z<{QX2?$l-ufw4_&mh<p7w@~=0
zQM53A)10qpx+#<y@$%wbO}uIVYr<f>!aNVcvILV56jtX}g}#7N@J4fs!})pQ&Dohq
zgQN4fF58C^JmgZHyDk2{GE3Mdkb?KaEZk5B<(yzp^qYCxR%2+waK^*oI6Xc?HHthr
zW&=79kGT4R<-cJ$a7suRo5d;kP?o%S1Si*rL`B`l8-jT$PQRjO8OwUa3|sZigB#GD
zUPsdpzeZ}flQ;8>U;P4|<x!5I;@W+)j*=-{Q%8VvJ1(IR{hyZu^W(YxI{2F9J*(PE
zeB0LgbncDq@NbtF!cJ*~|2=s_>(nUxAIXc<D?8zj;p$*q#`?o1`1i>#=HA~1zxBaf
zC%pVzuCGq<s*Q@buu1V2uT;EcF!Ic3Q@qvfig#|O;&tG>Q5q{)UD!+~{-N9=2oZXO
zw~C~~p3YO#Jr#@!L>EZZGF#QjwJdu1n>Z6rYfPcEkgi3LwU@oes&U?Dht!!t{pkGB
zC)PKOOf#((f?Y9L-P*#9f=}^Q(7ZM_$HYY(qSXM82-?CHIyC+){>d@Ph1KeE$kf7(
zOA(z=&+)H*^{626)vt%DfXXE&ODsxsR6y`ONVE*>PZgSY35-gz7$<I_-;?wazaA>|
z5iZ>LR*O93D~S9?Xwt}H5BnAmD(}K~+#nK{B@pG&%dX%HygPyss1vlXAaL8^TKW#}
zle39w#uM#uaK+j6<%1ekqaW@pZRFq?1S%rgDsegC1Y^i!_+&x-l>*wZ5>O)x*E?b@
zAU}AHTh{8+DJR^a%qxF!QzrmT62R-waa9BWOlaM*yc2ibrz^!p=`LJpk?vN!7JJ?$
zo=&pqUFLD1Xh;@F=T+`;s)2Og<=#WGE{S9?J&QDbq*+LsF3?aw(=FU(;wyLs`s=vw
zj@P;L=uf;)#|<vLPFW9v%(rR)ZVpXH{?SPtN+WI$RT}n;5)4Io?VJ<W^y3>0I5CCq
zFr+}A@y3Y_SsMPZS2=G*>|&InhdyOWP4Q0`8taw?Y_x>d#BBeULR)Fkj9N6?;C9?0
zUZ#=toraGuQoeXkkthiG7!dfGskn6{h-%F&LbYa=phi%aagt<@_0lL@*)-W-Z-@-r
zp*4Rp0D3xe`UIsabp)Tm5iUAKBiz%kRjNuxWZ^a9>Sz<6Z3j@weNCu|m&LaW5sCq{
zK)w=%(EE&ACf>?K=m_KFhwu0*vfKsY7vRYV)b4VR5L~`ez3oMO___*ea>{0UcEGb8
z4S2mWWgqBle^Hsj?@aJ>6lf!#AoeqUD0>syMAcqpiix<-`047SO$2vHma<sd;?EWD
z)8b7x(%lXG14#WL{`Mi!hq8n|rurJJR~;vnPY!_F1y6fiBrxT0K$-IKe!tRoOesD}
z!IX?B1W%t3o!_|F#oW+ufDEd8#C<Vs9sJgXcez;JGtImaWt87;B&yym6SSdW<||Xa
z2Km?b=PPad$V+g9_%!%oQI7W{VibtVl(&>AZ-e8l_*}&{Qd}~k4sej8$nlL0FtJ3u
zpayy!%9QmSr%ZfWBZhHgxp>PJ{4N$v0{G4BP11M{e2g_y+Q`FtR?`yEq$r4wU*uP&
zeX?ZhmCPlccI1gt@^c>R!AX?2>mZ42l1$Pnl2pqiQPlCv;=6|^=swDDvr@dCjg|<A
z>_o<gnTMz=&~E~Li}^0(BR;y#YJh(u^2FF#d<gz_rm-ZBtApC`MGjHVWECXeV>IP!
zk=ltQM+l1oMpBj{MoQ#Tj8Csj*~w^XO87|;nsPV0kYEg>K6!`SP(1NTdt9Ad-Ddgm
zvU2rK%P-&k1)$vmY=Ykueo$Z_`Xcr=Gp;XBZPbiNJ2J$!eZ~Z1(0}qq)QR9fF#y~I
zup9wK11#z*3Ve9vd$RC6Eh+?_WkSmEED>UX2U=JVi?C3}{i(Pv0Ru;>4|9V4_^5~z
z>cUBN;dJBMu+Byx;TyQlCgCn|G7;;k9W<ypI)w}3)aaQ3CqC8?bo!XQP$>>L%ar1v
zvsx)$;;ge0${}E}H3TdXLjX>o1oidV(Zu=_cO2qqB(EDw=znKJw@v?dQ~zHonz_*b
z`S|`T4z*waz=wE_rp(*MTfD6A$zWAD?}eCGh$=x3Z9EE&Adi1sW~_(?B>I1oSe+%Y
zMoVH%bpH{Ft+WzGt%PlGssFJol+{N44{`VCe^fw^{<l>qI(msSOSB3}6#dE~F1eB;
zz;v`gsrTzgu>Pr47LRDMd`m{K4Nxa79}%Sw>Z`qpO`WO+CoJHFn!<3vaL7|gg#8>P
zFg7$<3Emf)w8T>;bdL}Xx*3^3=ZE@%hsm}J?f+-!eDuEe(fi&8KJ_{yTX6$}P}W~5
z#a~c~Q+e;C#%>eIzJ{7Wk`JkUJ`j5I06BVDO!}4L55z!7evtGN(EUo8@)0O-Jo$`8
zUOuAD;`RVjR~<x|_K?DIPaPtZDNs67V!+1e^9>U$8_^CpurLHb_a(-F5>GYMoXHYE
z_jW+{ekIX6iH?$Bbh3sET<`~?Z$VYPP0g}Z;yEmFylunF0^WMyy-vJlSdJvBi>ONV
z>o!a`u$oB_nIlDU6lKMTAQ9iEz8~fR3a%!D4M&->fq3n}kuwMo>%=JAg}H;L5CbI2
z3$<8$j3s<pOgJd<ze2lX&_;Q{)|aTmpr<U(Tns}V{O>E7a#@SmK^~wTW=6Xa)5(}R
zq*bOIC+twZQv5X#KE>yyJ=MyTqsa6BWA9zyqpYs|?+FPQ5qY9bC0dlBO>H6wK~qZr
zZ3ZSVqlu=9Hny>VNf6pV(qw|zQjI3Gj*~&!!|9=YIW6a3PD@))rLDE-v6^uA4ydSA
zYgE*U@q!kGDrDa8Z$Hm7mk=&id-{Lh&v_uT@3*!0UTd$l*WUY8%6x^mU~9I-3EZKW
z@8h#w^)6z4Adl@NeUi|J<mpvoSpAap0SfJKh)dPOB0d6?RrRiDSE;c+u63KS(h@=3
zs(M|aMaH^Ap~hOU#-LQI>IH?)Rw!9(PFPhtG<dsZt?sf4Rv9W!>CdTw8caT``k-Q0
zk_!KA!l1ItuhS0S=8zzWK~Xv}C~}FxPe_p@26)@V;Csr543&{pXcuTuwQ)POE_B*f
z=s*BU<~8Z;rm!Zk!2}IblOoq4!AjBx1_L-9s+Cy(XQZwF4VEWZPXmdi$YA*oF{!G~
z@G9eOr=&f`Ot2I-l}X99-H;nR5{!kRJBf^KtXV5MQxD<;*3~as)xB!0O<vYG<mEPp
zyxeb-mxTOQ^yQIs`ZDAY`V#$|nM(cW%a(rhrK`xMFZa`?1pN}F$T5xl*t+9m>x5RP
z6P7!ju-a6-@#E}-btdaIHV49p+m5?utr><N8$3Qd^#g;kJYm4w)jJIBGyHy%{%LXu
zv#uCb<k(h?DwS<I2y#@VY#{JGLbox|hI;599`DP;ti4e@!YX{s%4;<+Q1D}h{LM7v
zZ<P@U2yrYYH(?6XBgA^ikg->+E5+Bm&Xf{wlS2N$(I9`91VU~ac34+Fsr=tEDHZrG
z$##hXK4mE2Hr3IVds<ab2&ZY5D_dBDULj86_o%^ljGC>+!GTeAy|UW)o=rVlNdVS$
z3f@IFioRx4qXUaghFI<)!EPm}6$aPR{AF_9YgN5tL$lS1y)y++LhFA9qrWJU_N<+&
zU_w@pE1HuZxL$<sA&2m7v?CgzL5NZ!g8x4t;_HgYON3lr)o`0j_g)s=;|&iW<j}oM
zq90Tns@CaLZuf5qKcMhRs%9eWhrsj!VtCf=iUnjJF`Uh6z(S-g5E||!h2Hn5*v(-1
zt$6JxBd7p<Of=kMMF2Oe#HS3gV&P?{7c(P*^&mNLAt=ITB6cZaC!|~a*c0*<s7&(h
z@LclHazJV6m#UrE{~}MXNo!Cs)YoP~lEjMlv}`X!8@;rK5*|^!O6=5I;j#kyt`zLN
zB?U2W#lD+jyAvd}`fUwik8-FstW_gcpRP9Ll|wR=Q)QGIZx}`VWS3u&Ohj&Y8Ee69
zHmR-FQYu*m)*GEEb=ZjACOBl#4H#7t1XTnV$}cM2X{?VWNoydTv<6(#dKooHl2*KJ
z(mF*IW3rfhAY>+MYT2!PL2!GF=``CRt%j5estHocZ9_>c-jCH9kXPz9#U6#pa(J{V
zhew0eT4!qG(!&0%mO~4rG5+D`;di`dC>$d_OyjryZrB|I%QYb6S{=wOcRFsh%>K;Q
z;UTxqN#5xs-;UY)mA{AOIwqo%!f&Z$8$+sW*;b8G={CLTmJR&=q38|#g%B}Xt01>@
zYQxN5brf>5kM+5V+y=sIcnyS-;CP@4I3E)V>{TF3bc5s{m(TV?$<3}*a`U21)kJPy
zP%G~kwMT6`26FSbqTe>kh@+|3W<zek?KW>#ZxE&gH+s$%tpC0ZO|1)w10R~8%cWpR
z@M)6T+%`&bX(BwKSTEbuYanWK+o&QtB5Y+MqL;jQ3=(pKyzDtp=~KYupGj{Z6a&y3
z2nNj2A@t^ff$5Dy9&K_1dF)SZtST-z{P4skhrR|vJf;b9_S%joMM6B-k>sg@l~^HW
zh8oB&vLZZ--v<hBG~qT~DOlqmet0rAVeFKKyOQ04jtFCuGN$k%`BgnQAYquTofqgt
zH}PS?X1hBuzamYNFyL(y2C3;Lk(*tNRp`_P9?T{TcDoEg$T11*I^bcq-ldC_E8HDf
zobypPCXCp>R$~8p#Qv!aQn#r9+@LYx2JMmjA+~kcUqH%N%O(Sve+cgD!_u1*PcxJ8
z80n34Nt65+`#YzZ2D&u+qj$|m+g>|-hqO8!vfSyA)utgIW}oc{9kR~Jywk~iJ7&*s
z7-JvjpygJ#f5b*q%6LigW2G9d5P%h<DjnOZQL|;6ZZ+I9s!=wVK6cv(-0)+D&3Hf_
z?B0X3C2?Z|YBi9+^;ZROYy*US2z_+6qR5mNT;Y>;LGnO(FF{=%mm5OXouH3Bgor-A
zY*oETW|BlbA!;aYqv1_l;^K_YRzn{jRlzPhW&}{0|H`)8rjK0lZ_~%yMzyMKR^!9e
zyn%am(?yZ&R+xezcn56yxZIBZ@@gvApm)ZN=o0b>qCaSA0#9)eBpg@#dq%Cc<L@K;
z#wS%R#B4T!{IeLDXKbLavty;Qw-87;oGTwttS<1i9kJ7nnB0S!nJ)r~@DqXyfkZ}F
z#qGaI0{Mg(vfNaL%<NQ&A#?Vf=z>3sA=@d|MhuyeR7&EvOKm&?Lsn}TvKLgtO%MUN
z$Aax-v{__iIRM9<FNVx;c%8IG0j&g>LTrW%rtHdBRI<&GnXRQ4Qpls(g);I<n;Dbj
zalcj7N3#CB*z-y9mE81_U#Vtk)h3rEqFOd11gthPC>l4w>9~@&O#^d4u3$6Cg;l|Y
zt?4JHgiN-(!b!vCc~vrgm6$+iv#KDwNT;D;RlSUhG!{&(pDmxGW}E*G9&B+s54PCl
z!7icPB&o&QCbbh)oe$50Ijm%%4T1zOmU|>#?C^|1x2eRDIkIYPdsS=OtNLi$UJfs|
z{FvmC2TwK#;27zl6z7xl5P9VNjOUL5euw8^4-d6g2WrcmE?h0MA2&82)Ydr(JDr5v
zF?-O+`ZyZ$4-z2ulOJvu9yIi-+p(=0wMVw;Lb{$zY%aY@l^>>x#YC@C<%dV+=hh_n
zIh$TN;=^81B@vd_61!($Rxo0k_z<Vz-d-2IlF%+8J{*8vrHK#O-IDlF^vV$*=BRDQ
zz|R@+;T@yAcJwz5z2bs-Tih5j^y+}A3G}KDOz1FlI?qF_R7wQB(sojkUac8bYR5Xd
z_)w3VkzJbjki1^v&oFc^Q*t*zxx*D8z9M>M1c>j6D&#oSBVB;_tmu`I&wVI*)lYyZ
zdWFV@<ber#l`DE>C?xbMm7#NYyhN|wOr=*bR6f{XlUJ!pEBcfa9xB~-d-JS6y+S4n
z*VM?|gVsKbV};p0E3>Mfz_H4ZEx==TxjdSw+T>#EMb@fbRoak{>WxY{TfStcg^t-U
zyCc8bSfN;n4KpYtdy9BOi(XaiN>X8eNV`a~%dgW8e<&f<Yh8{{8vkZ|To&$+q)X71
zA1JO%TK-)*lCyHOpcE?ELX6jKOsD#Ab84Jn+Hy<+r-|$+c<lfi6hg9UrQ!D)u_V|!
zZa+`sB0db$B|WYW!_c*VXa&Gy($*^v|K0M#^Ep|}6gx)x6$Wq#`t=X&9H|0sqM!c}
zUDHY%G@h0l)8Qok9^K4Cx@Mip(Owv9OKUss-YfrMI)O?4OK%q1^sLdbts2!T8}tlq
z5!nW!XQ^nqKRqM<|0njtR;zKxO3>DfTIayx*!IIZo!F^_Uue4je%N-q5=SR!-FCz@
zg7&}rVF?|+BkhNYcnPDUrz7H3X{;Y=H_Q;s9MVA+?@#Q9i4#kgTf${EDtEj~F`;ms
zF7qe9B2#k8eweK@X`yb^j~P4iewdU|oK36h<bGInazCt9oAYVxSXe$TM^-^Sl9ba9
zv6-@SRl~zFWlnXIl+)Hdv<hwbCYZ62M_|SV<O{k@u}5LU9DZZ~<;5Hp;36qrW4bzs
zHeoaC_haJ677RA{;27!Ie5#zFXWG{4GK&jWKIwE11ESmb9~E?m<--O9m`#aXMQ=y-
zp+5(9cz~^Q;N59VZivC`(ZREi(D(l{{MQ}_ukK&X;@uYRF?T*NqP#~9sd`sj(H2-2
z<XEhMx7pN92SgRTLFrg^(_kZ(lV3MW|AcY};y=)|^qyZtCCpBoqr_I@1EftVeIV*u
zZRgVXMj`|I1|Twc7)c>{Jm9Qx*8y8QFRxP(2!o`J$&zQoouKZaf<=5gMinV0vKZ9x
z5r-MI$NP<*Nk@Jw*}*_Oh3xMkN#C9ZFBt`kj!~s{U0*PX9wK|vNJ=U>A(eVqst5P8
z?tf6#=@?b%ROe~h=0+8rdPe$usPoyeX4{z>HBcu%M@;yF5yClQ!j4gmcEr@&yi^or
zlpYoQ2KAJb5DjdB&fbNjayi72O2yl1jH8aBu_;;WE5Eh3vJmuAuZNUP7xmvNpr91z
zbit!2Z7>ah3z67*y3N)KHA2)Uz{8RcqFD$h1}=xaBUVgV5q2dEpREFlT>@PW4pfPg
zhEqYlnvej$LKbL`+w7J(9`+QG3<Fk#MFK;_wnoVue#fXQx(eoo$C^!5MFlh#fCDvx
zLCV3?Z&j1;9j*ZHZ35O9-UN=!Y=Bh~iJZl%fg?#y07{Pzu`%*b8_$pv+WWUTf=ARH
z9j(D?lwQ_2<mV*;s*_AC5sc|MT<CYH-fZ++V{)^tF(EdY#+0+kG{(=i(3pC*g~rro
zR}xyC4eYQt^4pQ!N)A}s4O?Q3xhH$IvK+I*oI(n6l_IC5{F9gKh?CXv`UB=Hmrh!x
zp5<D3B{xyb)p62|M7t|%2FFQ%Tf%#Jm9ejoB6(h`=OUab`v*Of)7?iquYL6Jgz7}f
z`7rJqYkNJ%7k<>I)2e&*j$DMj5iS+8Pk{1_98V?d2qkxmZJp+&4W1C;T&l-$JahV&
zBjWZk)mtVej;V70ALoD3(f7aMu@CNT<M1!1GpFUf*JZU_xpi8$ax#Z-BfO_mn&XRv
z@_E(n{rGlslNnbeyy6eP;}8EW_7b-ynuCm+as*iW3MAbv_HF0dvUez#cd}yNn#7Uv
zO91e_{m#z$h+w}@=)2r!T0AijHYaG!qw!wO#07X2V{)0-S{kPR0+Gx>#C=g?aNzD=
z|H}Tji+xA1CMs52x|oH#Lf%C~b&P-A6j+(uK*^3~%iUJ8k30i~>ij&;qZX^VzP9g-
z<YZ<aL!INL(q>8NZ=D}3@P{YDw|E&?1;I!~>4h~T*9RiwIYT@zQoItOz+D+Us?<kc
zDYiG#m>s`7$2Ykw&sw^kOqG<!gt>R_%axJhAL+!|PrKTt{Foz{Ki$;UH^}N(q*>;D
zh&*{FIlLwOXMgzZ_~m(F?vPf}<9v~pJfE`i*g=(fC7WBy%G$Zf{x-E*bqs_#|4+~6
z7YD*w`M&i~E6xwT6`%vXCRI7dP4i+)=;59rWaXR}G?}b8TqY`#O!QQjm##PY=$LF?
zFx(THKvn}5O<$Q*P6EXptYU9HTrmUJwfCw4b3bC8{xss)>-2r%26=XN@^?mB9l5%>
zvESpIPk)I)d$W+l5^8?ekFdo$KF??lVY36~B?6iW;Jjzb`Odc2CsuHxJ^YZl>fl{7
ziOU$G>ZktGSZAggj|F6xn8-tz$edALl3!#`<gVhBiCm<q!VM+<1)9l2%(!+&fmd_6
ziQICOo0&D7y62iU<^=a##<xRZP6C@~I+ZeHnBgXbNv@C~I?S49A7&j#qxwO`Iqm91
zI8hOpYX5rPp_6>ag}|$tN+-3$ockZ#$<Cdr+yJ5T<OV9`HRcStt~S_hkXt%b>_gc7
zq4Mi<4}l8jl4}P73NHyWcwE2)4Tb}RK!%G3cnvAO-S#UmFu^Yn{<G_sh{v5St{+oE
z$xJMunK+?)CZ$a1yi3$JxL$prA?v)yW$+_M&hwzC?N%kHs6|?>*|uec#nEH=R9mf3
z)b^ovgzAo=c7*B`L+uFF)k9m6;#teB>e`{pSuI$rt?K%rtF3~Nwa%&z4PD2H2dmSn
zUNW?k6@j(gs`d|M6;f_>Th--5yE*${?Xjw74&7rFjI^xkks!<sfv)$xj5%oXi=xR{
zBlsyFk;CuI5ng_)M?jWqM?jYAM^r+uN6fayEE&-VDIU>E3!%*;mK$@$h}FhiH9{~5
zL_QJUQl8Pt3Gz;ylYeJZ=65vXbT;=dv46W7t2^o}ZoA0JkNy4}E;r$Ig?xAX>4u2(
z!MB_!8hm+-vHlds48G?lqz*nqU!D6kAg>Uck7N9ep$zACjq-eZR(_cAS2V!*W70Dx
z)5l*TUds5J_`l=t;~9S=m|gwH-?>Fc9e+7Q(fC`*5EyX$%^H8C@#miWKOn4QQ-zh9
z8d}^vGXVb6YD=g-ohtl#+#%yH5Y77PK=`lank4QM{x1zUgYQ+Y3KddwJ<aID^PM>N
zCnSi$Xj*gl$(X57fDr*G0|MhxD(c>+Yu0-=z8(NE`orz{?X+5}r*J5Lad~8xx1xsk
ziZ9DyWpy{9TH<gkkF_k$b+h<aPHmfV_aJNO@fNqc-*4S~8YYj3I2YY=Tr3!Ao=4hy
zNqaMCOSXlbO`>JlM6Ia#^#Oc?k>3kBIq^DepOs0UwOD=Y!N!oMX@Lu)#U9^1!sC8k
zsb97qTcGB>GA)JeqlmtJ-1fCK(DkY}z@6##yS{;pM+04NdCg_-83)5}?R(N!^1jci
zc;6oljn4>z6vOWZ?%Pih$~Z9k;Dl!?!fyvhqm%O%dkOy{&%!sSMyF5~H<nc2z4t!V
zhj%qu^6qzX_h)Pk@87rGSHfZGiVe{v6EhT^c&WSLlNWwf{O()#BQo;K*6=ic-Mq_v
zksjyHgca`mlsIsH=5Cm4J;UGJ2Qc~T;|y}&gIs^m&2<R-TQ-)5AME8_U**-;MK0;D
zU|ZKe8Z7xs3v9~)f5e7L&qK6<2PTvCt1ik2KiBqhCfD+x#MLuP-)I@?M(!=wRqaXV
z7_O`jS#48($9lQby0bH5YasrJ8@Zo%KPOda`yi|7OPKxE^T~+*jPAB2zdJzY8?8G(
z$k>X5L1ve<AIDyW_^mr%@?~t|`Jn^;lJ{=t;@+EJq%E8N3PvK*BSElv*Hr8FVe11m
z!1b+8Bp$Z3RhI~rs$co^!0R`#FTJ0Ucn2QV&36pbpw)ziUz<`I^PjEhvze!X46_`F
zw&yE*xs4ee#*}`6t5$jf<e`j3JKJ968K`Ic-15^&uH0n4luX`aK@tk_&`yS9zYsfj
zpBsN3Jbfed3Dxwk)(vG|{p9F}72;)#v6doT)^xHHK3(C{70$a;4*-Otg#vWzrr%N(
zWnjjevKy{k(zD_U^|q}=Axj~Z`YF{<ktqfLlW~2wm3-vesGr$xP@aM`b7ih}Of5T<
zp(2I*T5jnrasWMGceucv@WzDuUDyS}*%La;fX!+ASU2Z$f!~GEtT)IP$k0D)#;AWd
z$Z9<71Ds5F-5Y=E8N|_*@#1#N3%O6N-E{I&27@yoBx})t@M7w-+Uv0`?Yz#4eQ*}{
zTXEa>mY!c6Po$I~F~JmRhJ;DakjOLrew%Hk4+&+R91{PR4256NRs)2DkmH{;9!8si
z(7zir5Cl>UgbB~k@r;5{3FdX)SPdQ7d#bJ%V?4B<kS*xG6<hE*E;sDsuIjYG@jc>d
zY09<MrU?5g$&8VtamPh537v8Ajm{zlQU{>XP;0->wv>ls*CD}a&lkrKd%1wpKPfRq
zBULDm+-AIc^Z<DoL?b$OGqI?V9bGlcqz<aBl1zAyE4TyU*J58g6PoJ|!aYjCAoS2;
zeWWVqF^LB#DiG-~1+cbwFu;Qi^MZQD!d6$1sbxtZI&xbeyvii!)e(pt1swb=_56H!
z*sh<drt%M4q<zKNyfPlp<=pHllW*HShW-!Y9Y3&bX8<UP3<n~My@)<orSJeXS4l;_
zi5%)R>Z!R;y^%=KScG0>OXFqXH(d}}%T8o7MS_EQdf1O;)jV^pAC^jgIOf0kg~#^O
zAW1y-ZO0<aQ5G$W2jiXP#9>LT=T-W+0irLk;|)Ew60YY+T(m5MP_&b$<tfhsJfOo|
z4AIS_AF!#AGBw&4%_=5SRq;pr?0fG0CGT4uGZ9X^D+M~e9HC65wX9WBc71~WM@xs4
zkM1<92Z#7Gc!oPKyJ(bnLZ1pmYzb;0GBgmml)CVE#bw#|N#bI^@PeT0hD4<3um^O#
z8T@SS`w|4@1q%E6(ZfKbBoLWTM&F7B!e8GLh+J2MJAiwCAbb{=V9e(73$N&3%j0^i
zu!I*swozL2n;F13p9f;!;YpXmJ?2T5FY%;H3l+P=pvhidcp2n6LWtmUgMmmct31Y7
zysN*HbyTNXksgcdxtglgo3x>LTOgW677hfWQ#Sa+ZxlXk8kEE@T1!ulA)BL3(`nBP
zc(Uo-LN|jp4WT`}%)N#7Y@N+JF)cfmGuljV`S!)bEArP+1A{5Ht18TuZcKd@kg==u
zEK67Hd~4~y0B?89Yk@p3wv>4g6#1^suOhAjm4(cwFfWQ$zsr1EW%3K1@5tq<)H(BF
zztkKmG)dh|6Lq*SI`UnW@1`xA0FGK^@>IdxvWBuU;Q(c|j1JQTKdsW5x*Ye-46@jH
zC#f#A`K5`cP1bx`v))D@D6TBaC7R~Y9+FAI*T{%A?^}nvSzTzljAo0WY-?t6bZdni
z-lcJ_6?%9FFUt#x8(Hm?Jzru2(^{H`aL*s^{)|1<j^D`ZKi@9nQ5JI5G!I5@P;X>A
z3&!vR0q!%?0Mq?uT0O>2J1ZG0Py>?Xf*IP?Nmw9K+(pY3oyV_vi`!1(Cs^n;c1fWE
z$~+f^{V#(z_w}B-{E!^7x=d&=&LVF!sKy3~kB7E9JS?`$!UwpF?jA$X4S+OhedMW|
z0cX3bQC$G2)?8nb5u4nm4zZ#W86<&~T0FoN3=U>~QGF%wf;QjFg+th^Y$o4kvR<;4
zI%{Q^vCS7&<}N~gRd?1gP(XC$OoNH)n<B|$OAWq>Zz?M}V0E;Azz`bYj-J>r=-i&K
za)*4+Zn6vjYSXRee!5?K>3CM(*<#jyf&+XMnvf)CBTG&9yZ7ZJU-52ILV>7lW2}V<
zwj{sN77||RWHL5AliQpwa$f;Ir(b$|&1z?Hio$N(Bz`bumqz4Ft?%WHaL0BUw!|3j
z9;Gp?2$?2AYBM(x=0Y*i+}rW&*W9r**HA3yJ|FJ0wX4*o%-c57+|99;o|NW(-N|HZ
zu<PZ_i=y=iVCnr%0Im6wNI$J#`b2ZB_NR&4^E^NI{o!&~kX%I5?+VOBVtN({@7s_F
zn4SnYNdb+Qxhd48c8dLH5b*OvinkL1&nsX_BE>t2fGrAGnh5x7BH+IW5G)z-op^e9
z+QnfmVoe0~j*MMISM^Ng&S|3co*bJZ=gz%$D!4hsR{LV#Q+Ki^`65Z=z81G~T?)~E
z-fI8VdpwNv8Nzy2V<VvF(mt(I#Gj_f&w0O!8o+b<p|E4^%`j1XZ-84vE7=+fOrF8k
zkT(nbQ1eJ2Qd!wH<rZ#pD!>C*P9*mhmNHYIN!GHn@nfN2-F&_sIL}`)<rZsMz_^yL
zqa1%E{}w+ea#$d;T>M=i(kjL)GWKTc_OZ8=*O=|LLO<`YAVg^tY$n{Vv2+#-H@LXL
z5U$cd^q%}yLTq{cNv9_yGxol@5!qkli0lU=vY)V%9OOG&i-<<ES~)_7og>L`r@6@2
zeuhU&wbF7HC^>CQ^2a7uP^@+OEuXw*`GmA0lwZ#79daECEGk3JzU&h4l(2Cp6<0Qy
z*1!7mPhg2=^KT#>mOsOWgtvj+5BzA~lKd&Y@DtSAYCoQows&~D)@V6!sK7xtNA?W%
z&woK1bhnh}pWu7!faZ@^V$Io*yz+3DpKQm9JPCKlUL_+DdAru42Bw;?=?>AV@~A(*
zI+mMHU(6IHbfQIO6qIbR+V{{$%B<Q~g2?0pTxH*Qh&9Vr-YV0Ahia9^W_(6C`-@yJ
z?Va!*eRgIz*4CZTGHOC+Z+*Y@P`J+zavj!FpdnDRH4r($5FB+d#g|$IQ@I{XcVrbz
zEtLhIK~}+qIaa}?UaR0M+zMi@2AP@RwF-6nSz%TN*R`)FH&1w<$r`DKb(4-z04YcV
z5X+>F(aMZ)ANxlcTaYUWiA0MR1-f40LN1>Q`~n5~RNxmVFvBVcQXuav7x3P4Vf7dI
znfV13_yraC1q#gA)jyjXHOQTuZd~oy$;>yK+#0Qd@3+dY;77|%&>?w77xS81=OD6n
z#Q1F+29+GNmX^nnUcKoHf58{uQdaVc)$t%?N4;HYqtVw-vz^X<(Q|V0{1}C#AbRmm
z482lMh{Ypj>`q0)C(rT8i>;9l6l*MV;J#%E%=I7{-F4ihpv}LYO2E2AfR+w+z@HWH
zUScb6O(LL60do>5w1}}&tW`h~(f(H=;8q1B@u^Nf+9?((Ac;?JPXt^=Ko59;t-*FY
z`{B3sc$72!v&%q^m^p!PF+AW0S_>9v5Y)UT_QMMw#J*k5(oVCh%Q9>Wvv*{9WDNS?
zLQ-v0*0O;;o0V2VgPqKfO06}yIu}cu`Ah_1H?ZAWX1~v{71pxjNowO?!X-(Fj9jak
zHqpABv*(TwV>Pwanpgu#Uq$2(YCy%9KgcK8_2!^5R>?U*#xHnqHGZ*sXN%yzcYWd1
z=&;Gb$bTuK2#Nv~d9**gM>Fx%3GeluW#b!~UgSpUy?QzLW#4x^;SsiJUVaRB2|h(W
zw$7V;zCuc?{v$%P`afhbe@wgs;mZy<mM)Y06LxZz{*2;{+0gWCU)z?xa97KtZ0HAI
zMwi|%rmVnfmmVT7I0o8Y<R*Z(I@LmnFi^9*5gSjV!}IMAM5}mxvp18+GDD}ZBNJ$Q
zB}0*v;!jE;nODPc{9Tgsr);Hb`(RKL^dWS6QJ|0qMHBoAQLfm|&)2&7nY)LdaR^tD
z!S+4KpX9F(Q|nUZ-kuiJBDUi(kmMDy*TaFr%`G|M$9lgBN=xJqGs|OiXhm!Z9j9i#
z5n{Q0Fe7xb*u~J0%-rw6#-{#;9}BegWi^e}bJEe!_lAWZ6NXIWm-)55nP1zV`LzR?
z+-<)_w<I2D>1}%{!=CLIpMsQwp=tf>DXZ;Y)ou+Uzjb?eS;m224bN(zs~0H@L|J8a
zy#z{?K2$UdE}y-#jBQ#jLq*ws3}#+|nu@Rpy^<ZfldPn5WFsg4>0m?qSS1MUbXLL*
zs7CQ?PUJSsG{<kvqV9(3(~IDxW}fclxBT=Re*LHGZt#l8IA<nkwd*z0q2yJo<Dj_-
zx;%d(3@Q^6wGT7FCJ$R{b#y80#!VU#{cx8vqHP#l_(>a6-84e`9hEFo+V&r3eeW`+
zSUi7ocb1Rg)Od?p!b!(Y`{d&ed9Aj+bO8;b<i=YxS%{)babB>hPPVGP!u@AggXTwA
zg|k5P*U+fO3uilrCzZv8x<W0HzE$aa=0aOjX;k%128h+BIIH>xD#a>1&MLfwEWhl9
zt`$MgN}*?!R@KzmR@H@#R@D!bSI-S(<rbMz@c~LbTCLsY>(uAr+uqZux~0(9^!gmO
z)^-c4>=ss;7WQmpS_`Nj7VZT@t%9#ZY@56m3eVgOSH0INw5-AgpmpuBbO0?xXZk{b
zdNn}(dZo$cC966H#Ea;}Qu3<o(egj7r319#aP6(8m#XQdYP*-jU3|D+I%*qTs2#bx
zOiQaSEn=XSGIS~#PBd^94V+QgdueLx(rbL!b`3aEXOQP5<oV^wKzRRGHr{76`5uRg
z-#v+j-}M%@*sCXr;I`uSgfe_GJTE(D(IDIXACCFDV@5J<|6PtbpLwb;jUN%^6ur*z
zOVlI%8`jgv?%Bbek{_t~v*?mn#QHA@Vr`XehE|WY-=lSi^-75Ke`KW*>pY0{N|U9C
zwY_hdO0OeXE0@}o#aiYkX+r;-xLQ~^WZNH^vQi=pt*^kkhJqPc(jGigWxq$J-HII>
zbfrzZ&t~;MKFAx?N%F05k?&7C<oh(p_ZWNqAkwYi-mlwkoACCd--dKg??<}r_0fl>
z-w13IB-}C&bWzuV33w~@`bY#E5spK^6ZG)or{6TFcVVJ`Q|R|FL%&-FD+2E~u~%xW
zpG>9XgbhH<M<t2*k?Hrc;f8*XHT3&FM!P6<UerI><L4!h6zV)v4{h3)u!=(G#B%=z
z)q{$|cKNacV=^FmIo4A1iX7uHW+1PpRd94Wjz&42avbyI*zP!H%dyRI!16iUZcHRN
zM0d->&!eU|);ZBwDYWOWbsQ7rSnW89g5gKY!!MOb)|mB!q(UndL=-bm5It(_F<<uh
zGTyWw=>3CbZ=A5n$lS65k(0En?RtfECVW|R$~l2BWdySW8C)(MK<JTwj-INB$XaL~
z%AP4@pf3TA0Dn9&Vmpkz6nl;v5U?F40F?HnJU5`qc9?+K*u8GRJlkOc8nG9+0gH`e
zX{Tg+TVorB!FFtR5RWa<06T$5cr&|WGXg_NQ3U`kb!=N=wX@RdXj6<EH!0zvs~yd7
zd$m(dDdNjkFmmp@D1dP4RRmG?BRxO30$lu6A|#iO==)xAS@;R`{){^0^z7gmt&bQg
zzb8u@_9bLx)L&r09}Y%N2}I5@Ti+5aM5nMJ0^~2v#){6`VxS<e$rw1ljdRIxnvW84
zWXVujfaj9oIaq+-lA*EyuO-9tuoTHh7BXBi+>3=lv}C9(0Y>%kf<U;L0ard0%SBZI
z06Pr8@+$jPJ#{C<<Faf_X|u$)ckQS3v1<rD<o%&aTi=Iv>*c2QZ?l*R{>jy0X7Jef
zQmT*bmnsa^=g9T6;2`VzK-a;npmqBjfx@jo#;9)wU0oitVZ_eR`YSFs-{1fp-1avr
z6MM^aPFFC@<L{6wU}md}qDHVS$ZX1~hzzNST%28@U7e>4aobj2*a@?-J-F}XK-X)7
z_4)<EowjcD40f`IgYZc=tZ%j2HD%~IbPVJUU5*_c17klEZYp%D&&uKGKu&-I4>^@y
z>m<Y#Cn2^tDI*wNj5gahjV{3{R?cx&4olA*mYyeF!QaJ1VMuH(g1tDVC~fbfb8wK_
zl*6?ZzCYoO-pR3_(*ZpP5i0h2G!?eu>^<J*FA?bmOP-}EwxIW|*pH7VFcy1B#+c;+
z_INU4p^<1Fj<<5k$5XSJ<Ehrnar~5zqjFi}fKO5A3@gV!zDQ9@6{S*9_$eP(YMtaC
zN9~D|uQ;<6r_st;F}{^bS|_a-XPf2Y8WnRkfZHA-qxSL36`s>Eezh@Ij2CKCeB108
zXZ%lUAE$V{YSbyC<TYz)fQcS#>)KvX*i|0h=Js4P3q?+8mvH3Q^nS2yQ(iFssI_#Z
zL^QldmOigzbB5Ye`F>+=jlG@J&qrbR#@AaZ&li5g7iJ^yuc++`f|28RVGX7AcqQ8E
z2@ju<clpCFN{`(rtA9q!T$S%Gm>~LzJyTTNUnmRjiAl4XNtZvbw_H(*e0MPn)1zGJ
zL~$faKhCC_M9uc;vFr4x<D95@V6Il#+44A1&D8OPp4p^y9lXY)2&wD4w$}rGW=XLx
z{16v92la^c65;J7(7MP~3?zgy6*YS$5{yhk!K`>)AkwD3L5FM|vuT(wV~;=F$w>gj
zviXq~7KR9SpO1fu<>Jdeo^(g|mrb0nR!)m#{hs(<qg{p+@HqC`{3UN&%XZNAK+XNW
z@N3~-qfqufys$TDW;lHuFVA>@N$I^;_KaQOclPb{mFz_AYiD%kIjDVY2qrg^!@J5y
zZ<+8$a5NGFDU<1lU7+yoXz_6f27}T1-|scK1mauJCEHh?@kpTSO>f4Q@V<nST6D=5
z`m2(~i?#AZ%#lRsaNVvxS%st*W#T=Ks!+DMn;E{pWrLP7M^q?de76imi!1zGkI-xM
zOYvlz{X;XtTiRa91TQ15-p#8%S>fk;zsUKD@N=QhA{`3l_}ZpKP~+NQ-MKU4$w2)6
zqq@->W5>Ci5}omplER<yqOWa9L>gm89c{Z@QW-PZZ;(qmW1VO*bo(+MmuRG<=Z1S(
z6b(a7aoK-Ue?*$a64RsSzyKJDkX>URf^?j=HyAQ#IFxf0h-4_?UZ##ejF4(-G;4x6
zU$UL_-a$()(jvJp{C44vyERz~H~2~(vC6lDpeOjlC@GNTM4yK5o94iVpDc@GpCBrd
z|K380bCUgMB8T5ikGmBWif|7ST}Ea2Icw>c8E@Jqn;4nKAg<5P3726aS1M!fHZwb<
z7171;3UvJ~D|m0ovX>_6f4O!0+bFZ5SYG&!Mj?x-GL9>id8sthX-jCd=5eXLTk=w7
zVz&fwcrx1xNOncuMbTWAUoS$qkaS({4U}x*VZOoAg%1?&3EX|FUEoWBeJ?UWc>geK
zr;YnM%po!5pu<sZn*u2~&(MPu{;uuysxlEUozH+`g5+#|4g^3C8DZ=|a@bsRkm6=r
z4*N@3yBS%uE^_lWHIkKIwsm{%FSSzx$aK0Xa(o~XwgYqX?XcW@feC(c?s5`p$<;RH
zmn4Q{haU_?hUGIj?h(+eWsiVC)B4Zgs7R|<-5VLZRtxrm^2kKNThcB-$wHx0i7O(1
zl<cZv%pZlcTzTz@MqVqS`$-q^*k?88hqUHrH5`-VcKozx`Da;-{8VW)SW3nC6*`RH
zXc)huL&j4nyDF}{TSIlS&Hhbxq_>9qyAUqTFo5-v-B$k|*=;K&82RnD4}>o_@>?v(
zZzqQ@H{O$N@$F{Bw;NsY?RGE-N4{+duJexB;n=?$)q?-~<+s9#|1|#3WB9*Z_&>O!
z<p7LX&wv1-SZ$JwgIvuBgXLh_L&MO|V(~8dEV@1UzG!Z)KjSqcwDm=ci~SjI25NR{
zD=J!S*e8fXR?%;f@-jbP1oPjJbrN3wqMYAqu%3i~7Khh%c#WgTb|A(jq||nl8b_t=
zs5Fk*wqv$&G}?|v9PLG|)J^X8lVs~{)NoDWzp2sO@v5RzBUVc%d3_nr1#6y2diis_
z6Lvb|6Do5Z*_L)+OqRZGS@#(09;_bOGW1|vmstR|z0dnWdMeX~6d*E_JH9o((LXu=
zmsZCGM%>J4AT6^oh#)*NIhU|^#DKMJs!G9Fwy>+;vDNqc`i%0S=BCY)Z8JX4$2=MF
z_<;Bglqzn(MsP4UTISdY?(n-!{juB`AR?;kVmWr3Y%IsHKyIFj<x)|gjpvGz@OMBW
zeCKXPtu<oigg!<ZYAX_65;!N8M;H<V;JBaLIPT{*j(eRwsof!P++IU!VrNg6P;2+~
z1)#O_L2JKBL2J7u&+688RGK^suo_{O7#<rx9kL(gYdbKg33(QUB)6hc;vd_7)Fx*?
z>S7n06)+SAl4OOx75FG<ti+XKNnhpTmtXzU_Mgm7PH1rK{tx=xO(-o|3Yw>Bz+`Km
z$)6E3Dp{SVWM!FQ6v$j`5SiJAa)OHqM$(hnoMWDqjb$^R^q#B^OqvPopFD*LI>+pE
zBj>RvZxMls$@^KG#Y&yL+T>2o(R?Z?xk)u#YH~KDypOE|;D*oYLv6-t&tfcaW}Y{6
zF3yoOuj_AvgZCt)2iENm1xlW_+Ey__gX^7h(Svv@-YmZ0BHka2_~7w!wwVie1xo&6
z-L!$2fs#LSG_~BC`av+`N$Uq2DoXZSKj_NdQDK$sN6`>5MQE)3qO_ykk~mX--ur!V
zV(`GQ<*!Eq9=Sku3zTdPi8)>!DBK_1$J4N{aahT4p@T1s`U__Fj)2+GF$U?r1jEy_
zdndr1OMfOh5^wecGd9{7>`wBp2#*Vfzm|PbH0uWG*OX_(NRk6aH0RB)5=*!7z9*+6
zZVCLYFR{O#d_Q|HENVsc+H6~3G9x><bw-|Ma4<Tn013*B0z4+lLmEmi8}Kqpf@8As
zQ5YT+49_TKVZnLtFSBC{+l4Btf!-pdP0z<(_Q7&TW|a00ZAPjx?8adDk|N*M<H*J1
z+d9LmY$Y!$3`T~hzen0O<@*^eX9ObWZq53kiN)Y4+0dl5so}>>tM+^uIohBq<Y=6D
z^4wL3E%pf-8AR#Y7F!ICy|f=U9c+~PcH!VKU~D*}DxJO^{r-Uu7r)Y`L8fwERjyH$
zTSWP`{3<mZ?Mu&HoeIBV!0@vbUd&AlvGWq)AGW?*1#)1nUk~rgPv|r`B;@lOlas68
zdcV@lu7l+zy;gf4iOYFDKJxAC3SOiy<YjimUk8~%-(l9lv=p-4k9@0S^P)@bU39}w
zbj2>PmOieuJZ^py9mE{3DA^~a?dLfdtmk_lQn!(d-B088sS}YUM6-S!4436pMC-D9
z7YC!r3L>yn;*{PGgm*BzDXXV>wC;W=*&mufX0fU2W4n${wX%CZB^u72{y9!*XAnov
z&3~4?F*96IWPjyAdB%Ie!hIpo36t&p*x#sxG$x-7biF(%n7M!_h{vLnl^B9%YF)K4
zHiyj8?`N0?qZ%lgksZnkAkMf;Vh9wgy;?w^lUf-lXAKj86|dmdiqVhwj3$KnaY%dn
zmGl-`)P!?{VX3-ILmExNGE?!*I=Jwb&K3OUy8`B0v9pc;Ci%ZA|0w>bnM&LT5IL7K
zj0l{Ys?bTio#o_<K%i{t&WdoKj=&&g?xDQULUbCCR2utq?963jKSRJf$>mMiAuG!<
zmEy=+APUcj95crW%?NDunZQ7N8|%9xhar<}x<(0?D~n+;zG=zBsF!TW3!R9+k8}UN
z0@EDj>-rYqTY89Z0N*O(OGMQpyrbnOAjn1K(IuN%D%49GuvUi{>hmjsg4O;5vr0pS
zk99h0^;Ru+CZx)5BDpvF!ZV{DMShS2Nbv{4yLP;JJi?`}*9IN`PIy<>n_0Fi+jiv~
zzbn4Q7cTBVhu>eaqv>9$L6W$DlFHc*&C%Hq@ww8~r&^3^qxv^f6+|_ICDLX_BKL0~
z{ea#@vO411{MOI*SZ${=Is%aqfyti;0p6XVLE&?N2xDzT&jIRf%Wo;7N|bwcAi~OZ
zd!P{QOQGvcw&2k6zYxsWWc{EsV1_3n^ox>8qnQR}>ngX(UU8px51kV*0ysqpqVD~@
zUs49*_6ws^{!=iH{nm^@i1y{w$|<n$wIHHFFoP`>Kfu%HA)~!g70#AkGRHBV=cHZi
zU-$OKFOJs#IxBGi=-j_DP)#;rmid6BJUpI}{4*_4Kqi^x$W<z(*v;Bm{fOisAU=x5
z0!={9|F*t7tK$KQwew8O*u50fBk^y7-?n?jx@j@7=ocw9<yh?zK{g5=L%5N*UStpp
z^Ol0N%jXGqJ@zWygx~<3$q8?y@JIB|ZSieO5%`}$#9-u#{lUqngicNAo?!Tj{XKCq
zN%oH!gu%$7J*I;=G@-45ji!SFmj7wUe7_EQG*I%q>7W2d2Uzde(vHyQlRXqD*=T+5
zbT|WpW0?B?PKln_i2#7US6Ql?ftrb7rT}uM^zJIp=uYddlMPxyD>b&8>j(ty*_TA?
zZ==`V(<a)!*mvI6LS!x&YvKN2G;38bJS5ps$v&b#=^{=G1>3gd`AeR#ZrTL=>1}Vk
zvu!^U&-AFeqN9xfBObq7fTxShI7PdRw!tzF4UB#Sq3nJ1DFyEcw!Ltic&2yC3m<41
zj<9v`w?|0^+pw*-B8YY@%hoBAIPhCu_^1NE<^(2*9=<#Hdx~|wPx9I0=UNj>c}{p@
zDLAOhYs9c<WfNy&X;QIlVyPZ_??R`|_+HN%<&l=-!rP4yR+5`&Q$}=tMtEQ0TO2_x
zd^2?Vgic;P<OIUz-V0FZPLAWAQOFWGifZpw=W!4<l<k986#5{xYjw@=b}65IDHsiP
z4D*MdEPND=H+>0uZrwh8FMs>_J7C?;Mv*ztlhd~KK+EplQ|(!?mQL$EQ3~gRd(Xck
zk8^=-wF}7@=pw6_6d3(JVsim_S9#$hHh@p#IHv({c`#a$<Comn1#kRxJ1C_-#|YoI
zj|5+zb$w@`<bjsKepKqe<%?dL!Raa*zfZ7zCo}||<#?z7AcjT=e#e(txBH_W0C=c{
z97KCQAB@h7+5p5S9ez!~uL<}y0YC7o;J3ksU-%tiXfLQ?wwb}A<kIpe=c0_RR|aXF
z{O^0rd^7Zz`vGDH-;n#y$fak4BtyKO+WtKTa$r4zjW`s6bCScB2S#_9av%#;Li(xo
zF|=bJ5p5C=)~lkp*nisj+8ap|I+Oj+V!IUd`EWLP^I>*0I8G2IHFpdJy5fU^(c*R;
zeEyT>R=}F^me4P&v7SKN3mNuCX-?z>pY^jxnloF6XV@4CMaX8h?H}%)f&OS@Y=q!9
zn~yhHJofVO;{puy$jnF<-n*Jlh<BNx^w^8XpQ6KoB?rm8=>es-_ZL+IUu5P|2<e3T
zOHR2q(~n?0-svyd?h9|TmYxq-mfr6`>O5*`juM4mD`TW)Nw#6#J~hh+#vf+geqpu`
z&2YrW+sZO|<uBvmK;(+-z~n1(LZ=D+k~8P?L>{lmHaMy06*W3?-rtLK*OS~68ru)&
z{)aMk(d|#(6qZC+GZ2{&h@jn2$dGwLD`l-(+g^YkcBx{8f;*Q)Z8pm}^lD@6bg0&t
zj1iUw^=pM<O|8*HWxzqo_r-Evw@GAyT|}4_+kCjItd&r34)MXKyM$3&6O%#K)=l@o
zXmj$3kjmIP%IK*C!a7o7YoEfDqF^BUGZ|!TEj<O0vj7Ou->=42_fn^lgCcqpLn8y>
zUiv@5U$qga^FP6o=UZN5DQYj^7vQToLHd2M*~alk=oGcGZPUD@q^_sTq)zqPXwV|n
zOcXzwI0j)+2MT8kKm6SC<aw(>CX3Bf;4RJ{v8*th-s`Wa0!{;-Hx5poQ-Rn{9IDk&
zL1FRxsAF{KlDi-CbN46d#Jlbir}YrAR1+e;0JyRC{z*T-zVDSp?CM19pV+Z`E(fSK
zf0rUpbkFCM=g-roNh8yHrZ4=YpM?av4zS;SX6fwX_JH8l>F&aso<?&g7@b}!ODUGB
ziki1LPr^IRH%!-S%{NRBnBR%!7dCzRJSxPQ`Egpzl8AFFoaFh^3BoR8rJaxOTd_f=
zIn%2+GPxGP!Mwy7<tuyC&aWXFW(!K$HwL$Uw^hmblyV!84n{9-)Vg&xLWV{=i=VU-
zp2+1B$|d@noAIM~@>}fY_g(ps<w^{FR{LY1KqC3WA6PfvJ5+_#=P!0o$Cbmjh*!T<
zrH1Xq(Phg?Xx7bm$!;Gkiw;BRu}m44hhIY_J=2_hWA>MadqUR)INBu3R4icUp3v!%
zIk*f9+)Swh^a@6DCrj!90@)NjT3E2{U{=$7wN~?kOvKM5a#sfMtuE)nY3;404nITA
z4nI>0KT}#tGoV^2zVI`pRmRuwGi0>A<tM?2fusY61dvw~4<N5obUr~a$BfSJuEt61
zwHQzb0Zfs6;pc-jPeRzm5ei>s3**>e>vO@@_YVXcoq~vwj&E9E!>*ND{m;Np?eA&h
zCo!tidQ4O3kcq%FTdWE?V_f0=2X-)OhNi*SreD%;5HU$sD~n_|Nt+cA1f$E=6JREe
z>8M1v%A5GxF!WjNSAe03cipM67hYu<2X#<-Bo~0Ppy%2l_B3IZr(Di)T+B=*lUH<E
zC;4b9c^MS4Ulc80ppmhSk)d+_*;8quJsj-)f&Y>Dq*nKiVhkju5L%1TpPhuLKfMz=
z9cYz>uOA7=;}wr;7|_Ppw$~-+(e@IQa$f8%oOIcClIjyF+4Y9F&A7E}A!}_!!QMbk
z>wdxls3*5#A}e<G&&+Y`Xf+)YnfVlpxM5GZ(p5Gb6%_&XqSIJ#JQcffETsLjS}mrm
zx0`cOVpGBucI@b|?+3z2&kPngQl6)nJkM61qmP=O7<Cx$BUbX0Cc|gPkfArZJ{`Dz
z%CC|9F0RR6EBid`(<8ZS(eMM%v+N}oE1k><@A*r*t!1w;2?8rok=nNdRVkwcWKL>9
zhEW1So}Z=!RI<aW*x`?spTnI}FjbEP?n>M#G5V1TG{Q!2<Nnb~4Ld0$EVkN3H2tjg
zCkDeiQK0bsw=$j@E&i0*stx^iFB(hxo(M)S%RoEmF{ea-#-GEF?R%``IjiEiX!R#h
z4&v0_OOB4Y&OSAK_Afa`J0S(WDg4pJ;y;JLS0>eiIN$UPe{;Dl`WgQEU`+eqOShmN
zw9(u*wbM_9C-$Nt)k*a!W5Cl-k5rU&S?#)aC>Y*>+S$IH(RuNGPoNuD@tPm)<qUs#
z1H`io+BIatGthz(C}-$1P)*%0b)&1ovkQ^hk)7ecO8tMa4<(`SJK?`H`Ig)9xk2i$
z8N1q++$iNBqYm^A4(SFZLZljGhd^(+OsYG)>xK=|_b`$r&(v`7TAIoN>I%d|h?FkL
z;hMAmV7I+A@A|LkTG(AipN{4|!PO@l?U6&g(mY>ek#jpNF~U8B_LOXFS>#>;yU!nf
zlYrO;R!=?RfzswvTF#Hf;Zn99HtVO!`e|Ru)|LzV=RJf4Pw#B)C)ykvIKeJ8{9x=t
z%Jr=$f-r^S`*f@r-`;V^|BNgo8S-FE{MDiH*WeqBZw6YGz7jMIAqrN<?Y55c&E%b2
zd$Z7Hm&DB90x=Ryuu}si&l&m=Ivepm;mP@$TyOv2c`f)i=LJf7TY7?#@fAp%;bIU2
zy%k$T7@NhbDJU5E$yy9@42?GYf-Bp<G(EkEoXFtw3Zzntky;vg6?*{WFfwJme{y>I
zvH6L_`~PAmo*jrxw7Gy+9G{|u&*P3yX~JiN;}b~utiva<W%M%~T0}k5P3EGpJrY^P
z>T&k`r#a8Rwf8rGs`uYwRq~touK230J<VVW2g2OdCz{N~LrM8#3V+b{I_hiw{9Jef
z$%S>UHS%q5ux1~71|YzH(<&u0b{%K$ewWcBJ)whOq+#Yvy2QQ2szk1ki@sb}+mv+~
zsMYmY%hcDts6RhW&M2O~*ZawI`T%FBQkX)KW{Q7btaMp8CjC@fkB&p^;U_3f<`&yz
z*S~PPkp&dK?JL=0l|K@e`q%SNJ5)E{M~T(P%|Yap8<PamU&5iEmLK+hr7RlCkE3sV
zQVPC9CNOaE88K7j_N`g@Gy%%!W{;v>xYN#EvRGfqn=N<4P3EPH=#>1^;ae8o1k7>(
zZ>|)@bz3y2ka!xZZOTe+J&;nEfbHj+YJbMtrnOvCy`xOmCXOT!T`ue9u_o57>JH8i
z{U{K*)jlooZ`#d{Wc|pxJ?qC>BXDamAV3^m70$XvDi3rFozf*(x8zCH@P5GT3vV_n
z0rB5N;x+*ykoYk+q%bBt)33jx<)=%#%eSwOo07g$WUTRU5=?K=3%Ov%98gcVvn>2V
z>__C>dycc7@JFIrh?pYS;m-&!HO_EKva)b@>|NF+@r^!s`DOiJYajGBT=JjE=&r$E
zflL1RGkwlMZxirrA|R>&bI{uaJeLRv5g`1R)<5&4PweNgn7O>-U2{-On<45Wb4aX$
z#(9Ggvn;KMEHeNf2EgI>*;Yp$^tbZxl>DWgzR1`gakvSbsrw>0XU67$wd^(A!5TKr
z*dIq3Q>!?pG7@~2L$(s(ax?}XV8|Nq=$tK)CI%u|`PjpM(;~UZ)^A3{f)IvfnidIC
z#p0w5dV{lRA9{oLYGqw9=3WhvVAosd)Sy<aiH!PYW$bBgo5AOLqc~WERC=(Gq~dDP
z$UanD`hxqVSTnG`U}Z1drK9gd{~xL^7$4L$!&gXguCibfe=Gj6=?iX^dJUJHqF(a`
zyK+{0p~k(o;`*KR4t@J0sn<+Os@M2+OX0_LzWc#I;YQAPKha;m=B4EMZrb&X&Ub%c
z&Ub$>@cHg1`kn87B5gg8xo~gh!u^>G4`iaru?2MtbH4j!=Y03Vr;KV1+jX(#%QUDa
zqMMA4gnfKSi_3xwqq%2^Pw<+PQjuX3!dc%Bx~H)sSw&m3mIHT2h;xn$FOsjuE6j80
zI<RD>gc-mh-vR$*u3+h?m7MQ5)=Nm2av0x2ZA=`^7!?bV!#{Hh^px=*-!i%Q`&P#u
zgwQ>k$yt-;DQnr?SQ*Tk8WlC%zBKZCWfOJ=nLA|@o^i+D(r1ubUVq(12x)?mRpxwD
z5Y9yPQ!_OYR9^Pf=-873k=%Ur2Tzv%VAqRWbvCRdkg<bHsV)d`|3+r+<O#a+Pf3&G
zHClbXCp_}*ty#7*Y}Q@uE=-GzyuIZ4CA+6Z#{QBvscF+QbARa*!s<n6YIV$o#HLFa
zH#+J>mG*+*iCv=N&&N)D+ExHb?3WyhkH(59m)aO9g=nxN3lI%fBBSz1i#LInekdq|
zjHcYB6A{@$N4a-f?LMKSof4Za8!EOBOeC#q8xw&_bb|~N47J7tK3Wex`g0N=4GD94
zPTGQIePD1<5t|MS$VC1B4tx|mBLyE#3`U1#rRez?baa-3jy}!$;oNk3WKdQw5}Oo!
zG=RPz7%3PGKV^_oQss~NE`0QzW7GFD2#J%4tmP9p>GMFEd<Q~HR(mTW=_8@0f9UyD
z7eCRUH2viSei|mc`;0aMH|qTA2WC%jBez^~H->u?1z~CDS8Zf9Dsg`Gi6i5uM;<fy
zsZ{vsOS-htpIM(jHV}zu*{FG21bX^(8hY}F$9|s(t4)Mq6Sii39~*Ha_Eb;!u&&YA
z2yjtcd8C=iVK5YGyz8xwFAF&hP3(Ukf}dDNW)`nk`6&8M&b`a(cv+}PnnTV6x*w~P
zWkt=ykzwmWV`DWfN_mnb*!6-?lxD_qog`UGWZupR1|q*#yMpNYME-j<HZ2VzKUT%r
z_IA(|o189s8hJt><FP<wtQE+30`xSzEc}cxY$=yppUT50rvA}k7leo1z19CGGU2U+
zES*|GXuFkrj$2C}UBbbZ52i(OZz>OinBWxeH0KU7r`(H{v#;cFtE2j8m}<;U8&lOh
zt_70^eZ6qpcLfJReV0>d-^}vpXT2p}&*g#W*#XSU$_rmc<EramCJK?+vCr-xHC)*l
z?)*fUhwhV~JmR^}A~nbm`-Z3-ZMv&C_hqXIrr?vmZlpQnq!?H|86I(_X<sX@)W5Vq
z<|k>Ae;!ZKH2%ucA{jsV6qWt>NBZviq6uHUqSTnBx7wy>5B{xR#!v29)8^FI&tG_r
z*MF@?*1B5R7B)03@HW*ox73Hcu4As3&$D1|eM95;X7BlOm}s^0=PJlEZru3#`B#ts
zW=-w*8BOzRXJMIDJHKg8&A54<&~=S<3+F7T^EQWSCr!e7b$#8e`E#c%m=mg*r`R*=
z7B)9D1sB%VEuJ#}in^LmL(`N+N;TGdaq|UZy_Yr4UuZ%vm^5k8ocb304~^y}P-(o%
zanhtOP4iEh)X-ShG>6>IDyV5#*c?g?nlWd=f;rwZ$9nCM0rESu(EC+W3z}=^=|-|$
zi>5U!XsoYWJbkS9+zYPqEHd?;G-+1FtlGM{a{%6~IgK>D)`g8y&1#;%pcarUYM5W^
zWIBU7E@%nW6)aL%vPKE#{DmRa#DuyvNtAt+r=Xyqg(&A0&k7ZuIjf;|PH4_oi>@j>
zvuN@3Lht$K7rDjP&k2R**VNhF<2YtrJ*T;j>evk`w_{9k8hcsIoccLU1G_5bgqr3r
zcES>s5g@$g-=s-SW(9U;`khKy-m}z{hO57xgz4X3%vGMc#UX}RsIGQg^Zaks6%?L-
z{>1{5+hmnG%ZZZ?^aXWI*8p)<gdt$YrQLs15`$HNlO{DAAk4x}_g>&NV<N5k&Tt(N
zW=De3SZ^_drhXoi>3na~yk@t~19lT*bS{%p6DPeLDFJCiJPGfy?l4RCz`))8Q5vGT
zcg!9BQ3_(3C1=jZ*NN$4Fuj6luC$@9F7(ogcHbXy_M~NU1T<o2`Jur1(dzD?DrmtW
z-|TXq@!LM4^oVBn0z2G9;^*RG0c4^(1pCn`H~gFl)8;gV>YC@zSqL6o(6VrTjW9K_
zx8F3!&dMp@&Tm4IGMqWDscufKouwPE$cg8czpmX*-+hr?fBB2zdYyQKvhA1w*KIow
z{jIRmx871}!j?PcYR6pXm;<+e!nvw%qKzp~sC4V+l<&gXtw1)G)Ia&RwvY2ITzLNZ
z6U<<nHGkp!(72i!k4?K$(HE59f|MEWw<*zOAjJZLpqmgA&)H`$Xq?aQHPjFTGUnkg
zWMJEzK&2B>=b9Q*Ij5<)u6)jQ4K1O9nt5}Yyl0(dGE7IHrUw0xs_mzOYwNCVx#sNV
z`PVF*Q{S9)hib``B}mIP^Fm%b)ce`mp&qYyLEQrHoZ4EJBhB8=*4HnVpYzo;)Yh3;
zF65GC23GYs{Gta&(KDDo_HjMO@#lUym)QL~`xm9gE9W`etbD;Xy^g!o^>@r7$DHk$
zjjs8koo>%=+iZN*Htq3yE(?+g=X`m7eRIP?n=ZIR!W|ELF1G7a#8YpnpOe05+Pu1&
zZ`3tSZJ5(kTUUEoqk6;6Nt88s@ny5V;NT!*pKl`u+iv3@?B~vsM@a&Ym6gvwx2%lc
zVt(x}wQSj)`FGwwx<q$AuXFmvt%nPrBF8_^fe$OrBP?mg*$W*l2Oycv4J}PIb>n7f
z#Gik@F+6UJP~F0s`Ss~^H8F25vqL9M3e2f*C|k_1U&zA2o_8+B*Ghq<LHabP!E>y4
zPCX;NuGUvy-(b+&6nSffMC=;ooQ8En(*VngkAl^e_2E}o$4-jQp`vNkbXT8dZEX)_
zt%2=ET;-{;2k6xrpe$-;rMf*Vaniw`LdXw=_KZ29d38;fg&+~44$X&yxQo*Uz(Ysx
z022TIvL7_&XVn`#EsDW7TL3S%K=$T&^XG;xfGIGRDRXN*ix<of%?dTlYMQg~8{X0M
znx@Fv7tEYDzxh(U3JN_<bv27-&24H}aD+H%eBZSO$<3;5XcUcF#LSs14sKdQJ+sgr
zeh!Hma3Cewwt_;Z-xnQ{{>;Ku>U8u0WGLWartSX@qDhBN=Cna{j;DFC_>_fLH_d8T
z>}i>0##F=Hxy^N<G4rR)V)Zt6eqHUXt6S#It!skG{Z^feUOz_m%X^l=n{?Mvn8*)@
z;?j5ur@HRA6CUk{*pI{@{=;gh<yV5ma#`~9W~Nh&BeR|p!R(3bj?SWgl!C7E%xY|E
zs4;Y|`Kz^8xfsb^@Y`fs<H_}zwXk7foiK;p=)~L>A#meNnlzwy!68fG!{CYZye>mZ
zGqd4R@Z5CJ9AW{4Gw#wh&7q+;Al^ArepBnNnZGctW{TZ>Y27zl>YBm5W>BXUljiul
z%A^EDQwWt(0dh#Og&xR4^H&$NTqVp8QLq;~&dSMln5o{d&|vZ;>z#mpVtrIzcXeI!
zH7#|QusCgkYRqyyF0E^jm%~;Fi!5rW9}tL%yi7$C{E9={4IDB)l{tAHi~4#NhnxPp
zFMbF1to!=n)yI20`Ixt%p}hw4p&XBAcV}OG&B=s!^~Hzf5{}tU{M-$F@dKEfUh0b%
z!yN2J&AJ{l7wzZ{%q+CJpOru6Ub#7dG8P8naSpL^Ev4rU&VPnv9-utTV(z@yjk!B3
z9v=ylkvk+FpNYAN>)kpqvxddv_h9bDd>nJlC*tv8Fd}QXwYeB`6E|4T#N5q&^BtIb
zN5<ou<j<zk0nB2avM7cj%o-h!*JEzNT!FcUm$q)htVJJiH>RIwFb`nP#LNXAwU}cu
zLzsTdcFYdU-MnGD33D&zvzS>tXS^3PggFw`@fDcGn7hxTJh`XF<7?#h6CcxGMmd->
zF>`@;E#_Fv5T;-7$E?Q8=i-YH`4(d?!98;Z<tm)hgL@TT5s%jnBOG%H=HoZT<9efT
zH|9f_dog!oX0=fsW-ey#iKN3Ei&>26$Mj>az?><6;`!StA9wEefREzcOg@-*+(LRz
zw--lACvzF~!<_jO;>rCp;$i0h0ytRI<5$1|vv?hFz|8+G@RR@VX$R(}J7|YY&UB7t
zysh~Y?ZKSc6_2mNT(Kb@e_j3?>0e$$x&yNwa}#De=Hr;FFn43#fw>p+A<W!O)C02^
z^8n^d%v{ECE#_Fv5T+lq9kUv91!f3y4Q2=C9hf6Yw|g_=4)@+2)DtuJFQoqj>7S>b
zn7J=dKIWR28JC#;*W&Sf=E)tH^_cuOU&gmC^<*s0$@tW;<Fc1$9G4^ik^HUvQ(ycn
zJBs-oI6iQ5;%Be#i<dI*Ji~LQ56`>MI^o*vR?intn|#)}`JYw>%2QwFw<^m`iN|Gj
z=9PLf?Q($!VahA#Ugq^^#>J+EK96sB&eY*~Q-^z}4lkHGyvR3v;*8;YPHa1_V@P!H
zvaDM(zdOjJt0vveXuIWVK5l9lD4l<JWm+;*pB1EQe7P?^CWCZ)>~u4RcYdx7;ezUO
z2Y;=v^u^!8%qnk8&(}A+$UnTaZAgb(6C&~D+6L`Rg{$o^7<f@ae?)O;n_G@Fr(*b@
zB79fhuN=GG{^7IJ>#uU^`P<H!g=@egZhm92w)1y0=TD!;OoJhHo-uq~Mrx(qc249t
z@jL3jLt*Y8-kOnKIMI~uaq8^>Z?+1Lt~^Y=SEr{{x?#XK@B#g-_HH^%y5)zatERlp
zzP|Wqm3wGCeRN0a&>)gs&n1+%KHeAqh05D=nDY9yQE+^Sa*8tJ@z0aqGc{!_rob_p
zwJh^i!qp!Kh+jH59^ZrA{nXwAgy%swZrAwqe=~g?x&7OiUU$ufa?&k_uKX>ny}scE
zZOJjGF%=@-JTAFCP4Vn{`i6VkOwCmO652hF3yAYfKMhzu!y+@S0!k<PTlopDh9v*2
z*9T1JA3i`oWs$A``uA{Jy%Tfzu>F)z{W_tSjik#O)-*7j`psk2Gemj&c^%;a>;sI8
zL*~As3BI?HZh2ljZVqxL`WN`pziop%4E%D5_qf_~YCQe~^>aU!`#S!6PLId`TmHv5
z9+jWkx%kuZ_^y=v<Ua*>>F4Zot~kvnf4%lNaR7g<Yl;T&Z^yr20RL6^ul_?{+^d|a
zuMFjX2mWiP#Pu9@;w#4X5dM2uhkTj-|1j;#V!f1y7B9m+@u{4V_~-QV&zC<Dv+<vY
zJu5)J(-*$sd475!bD7HKQ~GMsR}v5MXX@kP&uY?F;eRpxk+polaRXXEbbPNMT_XoL
z*05g5+TAamnd9AqQgNk<HE<za><g$ZYhILxOnnn{%A{*G>0I2PbegAXr)N?;eyKcw
zzrn>R?VCb)DdCzn?x%KZy*Y6J|9bq32JmmkzX1R7%EUk1D?{b4a`L}c9+VG^g(pNt
zJ;HbQ5WbFdDR{)Dy_ydYy43clUau3s@%Mf4CMQ1`%I9_H(an>AkLK};RQRONEyWi+
zN3yQf`1=myEvu9Ght8!L!?zzw4`!0idu`mN&#(7ON1YExpVv^o^`-GR-EQC@L-n`^
z|8@BP^z*iVqEFqmN9*7;9I14I!(P(mElST%{==wW4*m?=#3%oJ{Idt}pMt*!|3ms8
z|NSk8`io!KGk||P{;_`kYTqjSyUG7yjk!Z%>>u8b1`*4&?{U(t9U#BQS^v+zj`gd-
z*DV9}+aY;sd<=uX2z)CZ|EA{8Skk59UafQ-+^g~xe+uzeE^*?QA3eVKn%MR6xNc#}
zsy}jk0~g}2zkzWNM$B42U^`|EZ#@)$JWjgRx5nd7Y20pFG*J7&JBN<jVGO*$ig^4@
z;j2A|Nw+6GoyJWy<#|@d<JXZ+{l~8vx8ldv{y!)wGSIR{UVE5&3`!fj!UwmJU*~ER
z$CQyh9-MW<cLxbBcaZP$|KQ-a`YAAAzK76q^;Z_@^47$UdHq>T{B6Ylz-d>%d3V_L
z=geW$Z*4r@E%>h)uzmxsPdZ39@xQ?Pg8%koOQ&`{OFh@YFG_tiIc8bcBymjwMGC%q
zDQEj{`r=>I{5W8jBbud8`HA0M4c|FP;S(39&jaDylzE{3s3yPV@TcQd{-N{tkiH<6
znZGMY7x>+w_~SPGO9$}Zg#W|={GY|YXaIk;vtR)KVYJVS|7og-GhQ{0^YK^w{{B<&
zSN{F|XW~Cled?X~0*Cu4|8^(;{{E}*SA9~~If{P={(1Oo__?3_AHqKe|Ni;x#(yIH
z>Q&^Fc_{gUm;M<FJhH0j&qMpjKYRdwNb-pN_r~KRly1s(1Em`fkJM9M_XFI?I`DWH
zaNS!>mDi9i`yu25A2OZfB8?Bn<87L6-2>Kh!1f&=-8SSDx;Q6m*tZ61-y!J)|8mmp
zxh);ee*8PYe0jU#@jC39hqn$Y8=ki;vuwCGnpHNupkr`gc+q!<1cpy+JFa4QY5nkt
z%q3PHW-Ws>0hvO!P1ne$lzf~isocyL&wT&JR!`<W)9(+uzOszq?=z$Ld6B=V{9QAe
zpL`;gVVZ9~jsk3B6z%&s7%ZN1<*1<?Ftn#$X_|2;WU@j=G4k>-r^!*O)dXg(96!b(
zvtB=U;kXtUny-k#@-77U;2Ai7LW0vVf23&tjbjn!b8=Lkibw`?upIMnG_q(jAy476
zl=_+PD(a-VnfAmp3o}0_??+6SW0_4QGf3<iM-_CS*nIKKU9G_CVo;-j>Hm?h%A)T)
zzmUVnpIW+Gj*Iyd75jj{j8`4UU*&k4pGo|g&(1xbxyqAy+6sKAx5sD@{}&d)-F%69
z{s~~3PZky2Cy$5txzP6Ku$J4V5pK<85ec0uh<*;omG1mc#68}PD~qaqi=uo2$M@Y5
z<C)uO{XD!`BzaaTypg|C+;FX}Riily`BxkVG0mr|^aOH~=HGCrGUl`Mti`T8*9hh}
z60QndBS(xH%Q0V$cW|tA>-l4Rl&4AaQyl&BoI;rLJVkAOi@z=WJ)){TO&02<7v-2q
zJ->*#T@F1?uMmSCSL0ahmS*=)31JO*Jtc3|<VsBQ#WU}0Wo#7RhfAe?SuwWq*TSFq
zrr@LcPFIMJpYgW8O4~?MpPXY|KUq|h$?{OK0)qLb@rw}DGh439`9tDqe>N;-uXKEj
z<&QY7r!Ai2D74lM`34S+T@!LE4%O{Pn4T5<+%1c~c;<^YQMu;_iBPEyWykTO{N<>W
z<KOvH=vU>qm%mf^n=Qva{)9%#1d-unsjAM9;~X5H#1z})`4T_MppGf-QN0v$iK5+s
zL)|Jo;n{>ksPflxs0WnWIyp|Dbqe{h(u~5PH2+7AQXB%7df8LWkE*jk<p?P$<X4LJ
zdmM>!HsVlj`<3;RI8@F-IX=Z0QOG+=GaiT1?3JSuhjOdM^wjgCa#~bQn-g+C(Kb2`
z?P7SI!11WE)#r=>Eu~a~E*&)MU6mb`<5&F2u}zLH{xmv_)Jd4@+w#?C=JTysKE+>=
ze4pk|wK>3_XAgfVW7ffaC(}st$#N>T8qJw8IP{v5`HqgCgunicxS!+i1*OqYxIz|v
zcKYvQyO~N|plna&r%GY^jxBr{F+a_pAyn(I2*;W)fya&sKYs5fPiFg|5p;lh<pCkB
zKhk7Pfv4rT2ZsRoGdXhbR-q#lycR@Lj4Ao1=UIN#Cx274Gw8~xm}jd?HT=h5nlFMw
zG~;9qmmlKzx$O_PNAa313)PwTGov)}zdW3uzu>6HG~b1!F2Yu-kUWT?x^S!<AK*~o
z=gFbHB4uQ9>%?0XHjR7KahS$Ej$=9@<}+~nDqW({`<S}@NgOv}z6*3bR}S)c^mM8D
z5-@(*@$kr^5_c%>JN!xfYraHW(K_?VqPUi7)yJRS`Y@kMSC`;Yf80n7GFtdi%p2tR
z7C+-mTz-Ui74i#}^(%hl_#<Y<?feMF<|7tgq8@{7H?gq@cWqJe77kPSw7ADa3vGO_
zbQ%xllSNq@x~t(Y$7fXVHvX>V&wTb+KNtH%{+^THH~7=!F<(5hJ0tVCj2avoHjnFP
zA%B9N`M%AsJfGLkclpztHQ!J96$p1L<bA*)$GHqs&mg+!hnVJDi9>a}TOmII5Q=uE
z()?Tw(wXl!1Sg`MKpe#w#E|oxf@7Zx(G!WIKF~1roQz`yrulx3LmB)*KkNAUikl{$
z`O_OcnX58XB^BPIv_IkR9RAF=onM8#sgS4m69~*F3aXHU3c1;7<X~0n>(qWc<~P;1
zJ=AFe<`atcb((TAzCTk<mhe;K)<i&5?M)5->Nrf+iLvtr!RC`i8O&Gy`}ph7z}I*9
zxZ)s8+@HH~O@jaZ`Ckb9F9iN65b%yHHRYC`YMZkiv)eJfpK`(-)8oW9iTH6bM}PCj
zn&~OwzwGBW6o+tODt%9VvecL^Uf=8Y8Mlkg*MDb@aevDE+DiagwN};_bHcS~)F+&%
zkGr=UWUd@kD#KtmnX&9r8Rq<e%-En(83v=td`)POPxw+_o^6;#iDNZJ0*|%Y^>fL^
zHmAM0PGl{F^o_6$PtHlD#^(A@%(2~Wy@llUX>qL2Ex#{rx2MV~HU1ovHSKeX^Xzfl
zJBQohCer`F&wdB~{fj!~6uTa0I_7zfIo&a@aLl=mxyUiU>zF@u%%3~vosPN1F?TuU
zD~|b|V-6W%mwSq1p6Qt9Ip%c7yuvZ(I_4tB{H|mE&@q4Rn0Gqn7RTJ>n6EhIdyY9I
z&ne$A&veZ59CNy3Ug4N?9dnUme%CR7=$Jou%sU-(i(~F`%vT)qJ;xm4%->TS^GwG)
z&oTS=ljg6+uEw**(u`@-CV2}kzq(~%sKr}6ZsNFe&Yn=xVw@8SZ4ZwVT2s&Q3Gbr1
zrsnw#3%%!#`~0|~vy1Jh=Ug9ZYH1E#caFCpH0PS;@t2mBm-#L$n|1Ezi^k8L&uO7T
zZ-Gv$)_O@<UVcS|oxG40Zu4~uLUXRh3^m#2JlBTKU|mzAXWYVuP~AA+)Zp1<?YOR4
z*fQ?wmihIyXLAN>oLuwfG|%&ltG#X^`PpWu$qsQUoaUHCSW{j793^mUjrAeVI33OM
zj0>Sh>={SB=M!em<a)-{%|nZxW6-na)sohAk!ej0hg56o8qo#S2~RtZS#z42=3Hmz
z;o84mLqf`%vtWJ==^7}4pK+#Ko%{8SySkZ!q&29Q)-4RtH6P3ODROc9Tp3OrR4N10
z7zw>C+m*!guKP1sgu7gPH*-)aE=Rb4>6IhpbMcewrNwA!`p}?K6C|A$efhSL`0bzG
zPsn@VI=B3lgGx=@m5xWHaV7D+i&q_Q!z>H8yXjs0A9Fp#4e-ejoZR#-9u_mNx#C85
z=}F9q=+Y0Om`{16eqX~O-0h}!@%2tIuY9fMH(7r-o@;)CaM1$S?b3rP7+QVZ=GU&j
zVCUA~@%DTRi~Qa6F8!J4@QhD65gjs^Nbk14y<d8lp5>fmCv@qb(u>Bs<-6&Z;U}0l
zU{28gr#^`%-)Xjy_*Gp~zc7x}^d1NA8xaCwr3u1pr*c_g*Idyry}SM?bru}IlX76n
z=caSLevBoxe0ROF*KenD>9!lcfBjcG>0S7H%6xXzGCTbg2M%uf{^hUjmww7u?DSI{
zu2OQh)USW}clJxa_9{F5+N<pHrE8k{x&7z*bz)IWxBgPw)VFD&oj&9)oQ$tM)*P!b
zcJiyC?pD)X->v_SoxUhFGWP!E{{=q{bvM1c{);*3`^WF^evPy$$xXk<S-<Q##`JHI
zR%6#q@2-FLJLz4(xZ~R`-|ff!{mS2e!yJ4Nv`WUq=f*dV6u0ZI?S$0&dz|&Br`1ls
z&NkfSiTvF3LkLezzwX94Chj^3ocO%vH~Hh%+w~bnH09;CzuQT_-AP~NF6^BIUN@m*
zo<w?;)W4tI^f&a2uekbLQ^V6O+XOg2m)vB>IL(dd`1f;~eB7~{n!d<Lo57zBBXI_S
zHT=17bMcqv_dw<6NFvAg8Huz~zW(K>1RY|{q;?Yrl@8!P2;u!e{+Z~44&<Me;CnnN
zUnX2oY{2+K5`0lgblQ2G$E8o{{@EUv?xg!4?{Vo#y8qAw-;?ftLIVG%`!k4Kyq*$}
ziS*CK*D2oAbGXOF!|DED3Kzeo1Y{y0b@673H~F08aq(fg|H<GtU#gd}XCnA^@mFeK
zCfgS-UP|}>q{qcK>Hep9Ts)HQKLWfkC)LZ?GsP9U^FAds6B(#GpHsYP*J&Pio~HY2
zHJJWsyk>dcj;H&&BIQHj@M#kf|Aut<6g&P>6HonSxbge>f5q{ax_0Wf!10$Fb?VpZ
z_^<5e|1XZe)caGv|8)GN)|~pScl@P(occZJ`0E%z>SrX?Y!pasKlKw`7?d(Ty-whX
z8NywsdDbkQV|;L$ZzPt1;9O`Toa~WWZ0hHhD|Oe@Z?Y3#>a3|>h2t-E)zohm{;H=d
z?`d)ZwRxs`c1PzJ9~a-Pbo{sV^LN|T>G%`Pe0O3Qs9o+jk$PzA=l1WgmURCeCc~3G
zQWs49c02x33rzi5je~ya=kJaOshOsJ`<#3_`o(w0gVX|3zoCX`FwRlmOZ{}AuQraP
z)|dJfIsQ`5OZ}!f{!*_?{jPBQrPh}E&3F8zwwC(ojDyORnpx`SjvuMVrhc~)U;XaN
z_r926hr0b(uxyU;aJ)T#ApSu8Xs-uO_DFp#^?TOI-<2DBoj^DLvHkKN3=ut2{!*t(
z{XT6RCwrtWmHOoyVtlek>Pi1^bJrUq*LBsOgQB8R(j<<cX`5;tN&}@?&wI0fW+{ZR
zcVFDGch}RMac~UM_3X^-`q?u(llkY}HdR&ygc7OXe4r3DQk^QOQiG5Il>$Ynp}>bq
z2`(Ry3JP)~0fi5ZKO|~YM8)r%d+wX}?t5?6No<QJdv@kGch0-_+;h)4_uTvLiC$*B
z(DtL~HHY#%P9?56C109)=B#@DdPY4rf$u|9(AMYlYy#20VB)4i|4G3wTR80oq4IWo
z^-r1p`<;uH-uN5FllJ9%LdV2)O`Bim@<}_D)vo=RFzkaJylmOK1B@rzOZQ3Z#|OE5
z(oQY%gk*ohKGHJ}w2*Au(@mz6w2xij_Y>dt>;ApUzsGVQ{!XRX!T*->cdixbM!X|{
z{#G$jy&H7Qc`;`C^0&Z={?$&AiaExw{4SSgDE9ImIQ&Fs>O;Cd`{dh0z`34}6&ayt
za_HKY@)w>f@@L)zf_MMHI76|QADZ$O{SPCUa(Aqx{_vR${Au6^Xa=>S>uL1-9|5O&
z<~EB=fY<i%?w8>JQ+ac~l9>;j08Z__Ch>X<f4-WG%ZJFH$ARAkI_8`nb6?u7LUc}B
z<Kj8s2O@sUuF$%gLH~9*=4pI5!1zx(UE!OFFSLMDdy7YlOqrjl!#mom&iq-{{GshZ
zbR7gv<*$kS@UwHkp9MZe1+43Z4E!6w)9U$c;MCsJpRF;ave4TPflMp^0OR1>D@883
zPs%@&QT~Mt{2wy#9|--+!Z!&|IS;}$CqA713C$ncD^Axpfv45;72w2AbIy;EpCt_4
zyHou3ql|;k_i6g3T~9Es?-hN0CIf#FIMwsQ2Q;065c(xFgm7~%)IEZK6*!OIDP5lS
z^wRZR#`XKyD~1r2H|IqO8D|8X<oxkfO-F{i^LxOF4;PPVydcEBB6O}=<N6JubH<``
z4;V=G@4j2p`7fbEdt|9Tmp)fy#DsCaD)^OWHBQek(e;mp&Zmm}`CI(zIP3V~U68{!
zEr0JR;A#E)Jn;K(gZ@0C=|F7w`a0pLk2zlrW{9uXnErd6ZL3{(fH=|5$vTP@ck%#<
zpXNL*Gk!k_e99Dx{A~kg{`_Z=(I3foZvZF$n{(sPe7@edUzc|tD)MJ~rkJi{!0Enb
z{%HE~l;B;f{yV^l|5rY!`{lNMn$90HuHQ#rUjk0`G3VNWEquM2QT}z{Q&J}S=Z<@5
zZkC+)eE>Msr)=rHbBsqR==Z-b<?poQ@b7>VKg~I4^b8MOM-ZnZd}^u4pH0|wDEOtP
zG;aF+4}nvCY<su=r!;=ay6-0#XVS5k4d8U&)B7}^`Pmqh_@>~y|EBND&pZMD11!AK
zeRCEcz73rE_q5fI*MO5AzWmu5-}fZH;+`;AM8~n}c^o*8Yq1vvDSs9?(H~g!p9D_z
zFK!gM<TWY(@4$)vMT`DD_i8#Vi%uRm^ZyRbhwliTXMhu(si!sl=Vd^?B;|AGHU82*
z{e$~4pooq+_mH1G##gJrsa@tgMu-7l8&ZBs#ydU3PS?Kz#}G?g{{#H4+wOI)NSx3V
zco#wsY2_aPPWor+H#Gn6lzKjtQT{{*-Xl7=?_XH=eS*tlTzp*fjdv!YQZER;EB;JV
zD*AQ7FMYAd$U(th7W}H<=LP?fp>NscF5-&RFP9edeV>(feI7XL0g)ewA76hWxH%{9
zF~L7_pQf`b_Z2dZ&$twd{22kK{@u04<3%ZN&I2_+`G%CgV3q#?m&dsKa*>M~J-?1X
zDa-SxHU5`E=+A&tdvjusM6Eh6GcJW9f9UavG<hfjXZjCo`Y+3G9ufTXMUBHu^VJdj
z+9Mj@6~+Ck;LeW5yTZ7GQ+j;u%J}Ls6}ZR1>Auc;b${`*YWVKYf%EwKViBLPIhbD^
z=U=9DKbrGDjUIRnIMM0;Z;?yVGk|pc^#P4vlkt8~>hm4oB>$IW-e>I2o50ifbNBsH
zAB)e^!0!Ydd))mZm)CWOzMcS1dawAKHHJ^hmrnub`hTy;2t6}NSL<h}p7%PJWqcWX
z+6A6g&kMkb&Xg+h?+-|O|4Q%+uW0-={ucXQKAcYHqri#Ig*SD1dKQzeAOBg6m;EAt
zHgwJcpJF8b@&s_!Z{<V$1=KHJuLDo3|9d{7`F!<Pba{HViLQ?WCq3V=^z$hxe?jC#
zn(KTTIMv^26#4gj)&$<&_kgB<Mdabf_**;xBDguX)X@27!E+ze^`U3U=z9OpY5Es`
ztH__tnFF5$PW^k)>fhf6PV#(3<cFTAr|U(bGj+SBPtTgt_2kcMzL|49>l${h0VjDY
z{7y{}{pStftOuljg^aWLQC%NL<a}3t^&D_2fBC5*BYYMC-n|H%<?vOF8$aZsU!d~I
zs{)+LU)|N^`OF3s`=^ZZKa}#_i@Lm-?-xFnPQMDA#_46LzgZ|6W|aRz2L3JJT+g4@
ze5Pkq>B>E*>F)~tDMoR~KXBsD)h`tpp=apn`aa|Oef0HR)4Kew@PD6t*=IajQony%
z%9jheJ}2Y{Ujxqk|Ft3)H1#}iP}A>zS>yaHJxcs8aPmuS{quP)?>Lv=r^_=jynGEf
z(RUu#xH)^{b)kRJ8b^2J^nIr+x%v#_hU)n51u1X)=YJ2J?mIQq_uY4!{>{5xqMy9(
z&%ld}qhG9fYa2MxFQ3-*=}a2BZuipZe<%YlXW*?2{0ZPBpLfc5H2Uo;Tpo7W(i`6w
zy!g2y*Y6C!LJ#LPpUWTA`11$!7xyuqvVKSB{3(5CC?ET`xER|Hm7^VIF$AYz+ek{G
zlv4TR59gB5LIU!VNb-hWvQ+YZbB;Q)eEjgCWu^MVbyZSU<JGCkV7Z`F(BX|_{H;0d
z+ivtsqE>|jvXFW|Ls6a5hdVF3b9^V0xB|z6`_A~w`Pi3q>ay=lIPul4r#qjhmYb}p
zm+UVu$v-dIXfH|LOLk#CrFW9#^Rub^qHHGdQ*CLFojHth!V#8fogMvYc-G|}M=#ne
zO|Pa9i5%Hj?+y9`9LhRfca%E1xTcm%m8wz>4rIislqla*gSM)-aN=ek2mK}>g;RXh
zW~;MaYpDi~=kKf9aGOpP=F=q`({l^+9w?_2QndL|Z(LN^jM&?8^ybILu-)FlPa>cA
zIQv*pr5;{8wBjp&bqST#HJA4?dX7K*aK*2x>Y>BSK9wMe4oNiPK%b70cGS|NtA|!f
zi=kY*g~d*HM|H6$wW%Zti*<6k%crH!2l_PC)RWWondzgo=<(}S<@=RGDxk}ujhRZS
zAj8e|7Ww>TMW=$dtb3|?$qxbC+v~LUjQfUdOrAPFv(l(-`fJCc^`-ae$HmX=Re`n2
z0@!cbHD{j-y5#vId|erMIi-+bpmScenp>NLGpgC^b$aDv^<weJa_R8m$&)HSJrh#O
z$PVGo9${Im4BPrlTEE-~!XM#VSQ2C@piD-swbdgtm1eV1stnvbPPcC00P99`gcLZw
zKU%AfYn|lLqU*={llh}>nCz^HvSH5EZ#dl^HY0CHYn3<cO=AoT-!fV|e;9j!gJzFe
zxQb>JumL)#LzqBo6R*u$yHp46QZ@#KR5-z)U#k0+8eV*A#IZ`BTCYl-zB-OvL~UeD
zh%+4N6>x^LU9D<K4*j>2SeP0oS-laoCQ0%hc?TdMvrwm$<VK-cP&QCby<mMfz((;>
z1A{Cn>1)*-bz)f8Sb`9xkdCUZ>ir=rB4@o;KS#NOO!LDYFh5U90W6>yw$(?##@Q(M
zanf}*>*FjY@ptEp_^;O*s$T04M)5-3PCwXI=nc*oVE2dajSMmIy!iRWi=v(VwaQ>_
z#966I0c0ZZ$2Of>+m`+s3w9zcgzrP_7Q&2OrIY=hmush_?b5ocs<gZ>Lrf|;wY(4}
z?jkKY%c;Bk%DnPt=n!pILOC_dnYc(+=eav86B|2j!i`g+QB_E)n{=91->@T_?ar2#
z1F}R%M`}&1<YKGV@55?#hl9S|U)3tQ{4I^vxHPC!jaJ!`5Hs|Wf?8PVbehej8?M%y
z+kvea6t%fGP#d+NWjA{DSY<F<r=ErSfznxl%hYIYv1#_>jhS}cy_BEpRVY_Mvq4!1
z`hF0s1r?W-Hgtuj=*W4P`~bt{Q#g2@w2#^<EwAT%-<>tz8n`?j2wJ!z)-E5lx3db?
zVqUv9+og>#If_T}OL`Sz;rs2%B3?)EK5}a9SQWL}Y$CT>yV==j)OKvP7zN|r%mgZ0
z$r%yJP!uWFa|6~>)wgR<x@?_`MOb1yD5KozJoQjzPzYc<=)s$r68H_H67NY}$f%hF
zcXLthvNw}UOFpCdOFF}yvs0#&Ds$kr@6IO@>-p_ah;AZtXUSn&A^lWPYxKWSSz<Zl
zG*@oI<#F@o<a0g>os74(MQlt|yB2KOBBI%r9N4bQBfGRUy77bC$V&3AUbRhHw)%*!
zRwR*^`Avp=d2ZR9^uO5Y^yn;o8oYk6harlmMe<H{$4Y*dbG#*;uFz&>ddOO&w5u1n
ze|Z&V#?ANPcf(~udI#7?9@vK=WD!?#fn3SOcH?-LJP>~y?nvJ+m+R0$<W+@&$K|w&
zD+gs1&;!DoG2RTkzPK~f_UkQZ87~JHr;bcHY8|tHbC@F`OF)ypm_uNI(;7m5!cj(<
z-j4aYz1*HD_{;5C{CCmB<#xWXK<`~rP;I!i<k@2w<&Yk#)>dq1!^$T6+C?pp--X)P
z2iI8@)bQ|HQt|Vnd#DWbEM|>uJdrnOhJ>wTXI?&<t7-{Z4yZWs5%VD<M9Io{r&#b3
zN@poW=~Tr{NK!#FN>$3iX{NK)l|`E0R{7bw@b0%Noz}uiw~o<^*Rcv|V1gR@tr>%9
zJO(W0E_x}BU1dh$A6kAKzDgt5QbRl`If>OLHLF>NWz-<`Mq@`>;W5eePAlpANY6y{
z#zwDc``y$<=^dm}Cd?T%bQj{DFK60F;ZAuJRAw<9Cx%q|GjQR$ZjQ|fCK;U_>KK|b
z**=)oe;M~)2xp@H+SxIuZ<SchX|Wp1R9J<?)C{%}(_m;~%z&5l1eoUJx^<Rtx%1JH
z%N`iU?8H$^ViNVB>HRo@VFebhbFyq|m%)2?=5Vi5YtS_6S3A8{!w=BI+_}LV@~w4x
zP0LG(%(Yoz%-Q*l>Md}RYp&DiXp0kCnLTE&V{RtB0WX?3Zh|4(;}#MNLMce{a$p=L
zmEIiKf>Augqvti@d=Q>1TASu<mB`szBd2!HPKXaapGKK0qm1=RQa5`i*|x*A!1#l)
z*{-bzBNwh6LXenQu(^UU-RNK}#vRc~iYhu3S+k1gV#)Scqo!2sQDb^;yj_NDjmKMt
zL@$H_9m;YR3#$?LXyP3Bodz#M0SGmU$^J=>P>NFr$>c#eDD<IlRKNLHVY9hK8Mu+1
zasgB6EhG-x-lp!R+~J7w1aKuN>uafec05Hp3uSzk+!+XQRl(y9D-xQNxeliTPf_(K
zYg~hP##kfGt`}^fPViJ4!}?&tfXwtPO{UDqOtB%_!{p!O0I;4Vz**!8WerLx8SYeY
zPE3@WY!l{JD3oE3DubCi(xk@C6q2PP12%jsds2bGP3WlEb2ioJcN9`tHd+xME}^>?
zC_Y}M@HpE5vI66%8qL1Lj+uwMf#iCPsGxhiXw*mSo0GlD?Rt>!1t2uY)u22c7|AiI
z!y1_USj^2<)9*A|J&Vv;n_>-Y@IpCC20_hN`CDb~$GXeqL7*NR2F-rGiGkE;_BGoP
za*D^f7vSO`GJ&w0Z4R;g7LJE_>)e^B1JHzNm03Y|*lVg$&>IYEi2S4|XG_~cHAV8(
z`s3l97K)6|09~4GQ!kt0bfbDa)F1g7?Y+@7wc3fzT9;`_hVT^S0v8)YGe1^HE-PtC
zvb=&tQ<7&L?d|CJJggZ7%+g}TbOSf9gUV96#4#lw5sxz6$AsAQ;>h%ZgZOZ}GqQbY
zapl6+$)XG9Q6?@UJU3r(t*>MGE9%JYsSc9?`fv#hvW7sCF$#;`p4K2pRGqAWF$N|w
zkyIIxVKQln?Yb>vR0!?Juvxp|@q_6m@>&}mCm~O=MY50*_D!UQ3p&ZIEI=aPCe~En
zzCBun?}cBZn;@k((?R?MIf&4Z#z#zaRjq>IE3sNFX_1wg0<8ifg*!zXAQiz(x7j>r
zsi(1S;sre!UZf0Yd?}o96E4iGRywmAtzrL6l7!`C(gI{ZxIRBK8PxM~c1B`KC_ACI
zZh<>beK2PJULL*fuOsat^L-rFC7lrWD0sA5nqzYGbxtp5TkLp8qYM_f)!8%(FkG#Z
zNo&GY!TuF1W3*M7s4Dw^+Be1`r*kvffIW3_n%-jft*b|$I@jD$b&Nxf;Ta`$YrUCD
z(ChnaK|=Pl%1HLGhne{R9&N-S4HZSw5_peBrb!sRw8d<=mEtvthN1-|nfmXw7JYMK
zw<C<@$jTE%a`pQ5#3+>KQiPo%?J*}N#dn#&t65%UxJf=&QkT*k2Ld`aiO^$MI<z#7
zUJZud@Rs7?2u9JOkgfY&rAF=RRQ?`MT*A<QX=|XvK;UUVqHk?C)>sHJ=aNTyf)>gg
z#3r^Av*&u!3AVAQC*oH$SC0G45FV^phFmSLR#TTfj?%J6Xw1#L<|Zzu!R{m5@&sut
ztTeHleKQuktX&+~J(IDEBO?;5;~?UWko^be#tpeuQ^9;}DZvT)-QiZ<1O>*1a@_Sd
zLtoKScpjpY=@(QwH2ajv5BBe@RXSA_)Ag9n&|C={W3W@OQKGfAKDqcF{ll_)PSM!!
z>5W^pI%N<K<3%_qaZ~1{5GJfj4eDo7C#9atnm^qABN^(+`NZT4fhfHvjW*?AWvf$9
zbA_x~HLXL_-bTx9nvTw?bK{+bmIk<PrzmqeMG_~A81QK;s@d)icCePxg;U(=)i#?M
z(wnrAuDjG3;=suYeuv#SQKy04zuh2yf#zy#O<2WI$hsgKgRlb$u^+B+%C0kQ>5}#I
zmd@X3<HQS{>)6S&H9>firfZg?113Nu&S-XLM|qg3?oBvxVu!PrlH|<H(wR>8M)Ri0
zM~Na_yS+`@V*4E&z8XNpMmc+ky=CKEIy<7pg-8HJ&z>=!cq30Fewb|REE|R!-P%)I
z=i4#86l(O0P(AId94lAb4G6s&Xdd3GszXfVrAf$|7D@0g#pQ9-iLLX{g;<~L58D*w
z!BY<T=^T&v!5q|6t<_9h44>#CMps8R!p4O${cx}Bcmrp+>;?IaE{9#m;vFR7?e62m
zR6dWrbbfM2Fons)4m5cEl2kCopCJnsv;HhGGOdN7AN%KP7#Y^r6hNzYBU`lG@s{JH
yz`1#B(ZKfbdXtXR)e(z~*$Mp4>-TAyhMtpzZf3c*)54Nwbc+e?6mqzo&i?^Xs&=jb

diff --git a/config.guess b/config.guess
new file mode 120000
index 00000000..5f6aa02d
--- /dev/null
+++ b/config.guess
@@ -0,0 +1 @@
+/usr/share/automake-1.14/config.guess
\ No newline at end of file
diff --git a/config.sub b/config.sub
new file mode 120000
index 00000000..0abfe18c
--- /dev/null
+++ b/config.sub
@@ -0,0 +1 @@
+/usr/share/automake-1.14/config.sub
\ No newline at end of file
diff --git a/configure b/configure
index f6bc2da7..e7d9f32f 100755
--- a/configure
+++ b/configure
@@ -679,6 +679,18 @@ am__isrc
 INSTALL_DATA
 INSTALL_SCRIPT
 INSTALL_PROGRAM
+target_os
+target_vendor
+target_cpu
+target
+host_os
+host_vendor
+host_cpu
+host
+build_os
+build_vendor
+build_cpu
+build
 target_alias
 host_alias
 build_alias
@@ -1337,6 +1349,11 @@ Program names:
   --program-prefix=PREFIX            prepend PREFIX to installed program names
   --program-suffix=SUFFIX            append SUFFIX to installed program names
   --program-transform-name=PROGRAM   run sed PROGRAM on installed program names
+
+System types:
+  --build=BUILD     configure for building on BUILD [guessed]
+  --host=HOST       cross-compile to build programs to run on HOST [BUILD]
+  --target=TARGET   configure for building compilers for TARGET [HOST]
 _ACEOF
 fi
 
@@ -2309,8 +2326,6 @@ ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $
 ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
 
-am__api_version='1.14'
-
 ac_aux_dir=
 for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
   if test -f "$ac_dir/install-sh"; then
@@ -2340,6 +2355,119 @@ ac_config_sub="$SHELL $ac_aux_dir/config.sub"  # Please don't use this var.
 ac_configure="$SHELL $ac_aux_dir/configure"  # Please don't use this var.
 
 
+# Make sure we can run config.sub.
+$SHELL "$ac_aux_dir/config.sub" sun4 >/dev/null 2>&1 ||
+  as_fn_error $? "cannot run $SHELL $ac_aux_dir/config.sub" "$LINENO" 5
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking build system type" >&5
+$as_echo_n "checking build system type... " >&6; }
+if ${ac_cv_build+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_build_alias=$build_alias
+test "x$ac_build_alias" = x &&
+  ac_build_alias=`$SHELL "$ac_aux_dir/config.guess"`
+test "x$ac_build_alias" = x &&
+  as_fn_error $? "cannot guess build type; you must specify one" "$LINENO" 5
+ac_cv_build=`$SHELL "$ac_aux_dir/config.sub" $ac_build_alias` ||
+  as_fn_error $? "$SHELL $ac_aux_dir/config.sub $ac_build_alias failed" "$LINENO" 5
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_build" >&5
+$as_echo "$ac_cv_build" >&6; }
+case $ac_cv_build in
+*-*-*) ;;
+*) as_fn_error $? "invalid value of canonical build" "$LINENO" 5;;
+esac
+build=$ac_cv_build
+ac_save_IFS=$IFS; IFS='-'
+set x $ac_cv_build
+shift
+build_cpu=$1
+build_vendor=$2
+shift; shift
+# Remember, the first character of IFS is used to create $*,
+# except with old shells:
+build_os=$*
+IFS=$ac_save_IFS
+case $build_os in *\ *) build_os=`echo "$build_os" | sed 's/ /-/g'`;; esac
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking host system type" >&5
+$as_echo_n "checking host system type... " >&6; }
+if ${ac_cv_host+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test "x$host_alias" = x; then
+  ac_cv_host=$ac_cv_build
+else
+  ac_cv_host=`$SHELL "$ac_aux_dir/config.sub" $host_alias` ||
+    as_fn_error $? "$SHELL $ac_aux_dir/config.sub $host_alias failed" "$LINENO" 5
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_host" >&5
+$as_echo "$ac_cv_host" >&6; }
+case $ac_cv_host in
+*-*-*) ;;
+*) as_fn_error $? "invalid value of canonical host" "$LINENO" 5;;
+esac
+host=$ac_cv_host
+ac_save_IFS=$IFS; IFS='-'
+set x $ac_cv_host
+shift
+host_cpu=$1
+host_vendor=$2
+shift; shift
+# Remember, the first character of IFS is used to create $*,
+# except with old shells:
+host_os=$*
+IFS=$ac_save_IFS
+case $host_os in *\ *) host_os=`echo "$host_os" | sed 's/ /-/g'`;; esac
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking target system type" >&5
+$as_echo_n "checking target system type... " >&6; }
+if ${ac_cv_target+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test "x$target_alias" = x; then
+  ac_cv_target=$ac_cv_host
+else
+  ac_cv_target=`$SHELL "$ac_aux_dir/config.sub" $target_alias` ||
+    as_fn_error $? "$SHELL $ac_aux_dir/config.sub $target_alias failed" "$LINENO" 5
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_target" >&5
+$as_echo "$ac_cv_target" >&6; }
+case $ac_cv_target in
+*-*-*) ;;
+*) as_fn_error $? "invalid value of canonical target" "$LINENO" 5;;
+esac
+target=$ac_cv_target
+ac_save_IFS=$IFS; IFS='-'
+set x $ac_cv_target
+shift
+target_cpu=$1
+target_vendor=$2
+shift; shift
+# Remember, the first character of IFS is used to create $*,
+# except with old shells:
+target_os=$*
+IFS=$ac_save_IFS
+case $target_os in *\ *) target_os=`echo "$target_os" | sed 's/ /-/g'`;; esac
+
+
+# The aliases save the names the user supplied, while $host etc.
+# will get canonicalized.
+test -n "$target_alias" &&
+  test "$program_prefix$program_suffix$program_transform_name" = \
+    NONENONEs,x,x, &&
+  program_prefix=${target_alias}-
+
+am__api_version='1.14'
+
 # Find a good install program.  We prefer a C program (faster),
 # so one script is as good as another.  But avoid the broken or
 # incompatible versions:
@@ -2920,6 +3048,58 @@ fi
 
 ac_config_headers="$ac_config_headers lib/Grid_config.h"
 
+# Check whether --enable-silent-rules was given.
+if test "${enable_silent_rules+set}" = set; then :
+  enableval=$enable_silent_rules;
+fi
+
+case $enable_silent_rules in # (((
+  yes) AM_DEFAULT_VERBOSITY=0;;
+   no) AM_DEFAULT_VERBOSITY=1;;
+    *) AM_DEFAULT_VERBOSITY=0;;
+esac
+am_make=${MAKE-make}
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $am_make supports nested variables" >&5
+$as_echo_n "checking whether $am_make supports nested variables... " >&6; }
+if ${am_cv_make_support_nested_variables+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if $as_echo 'TRUE=$(BAR$(V))
+BAR0=false
+BAR1=true
+V=1
+am__doit:
+	@$(TRUE)
+.PHONY: am__doit' | $am_make -f - >/dev/null 2>&1; then
+  am_cv_make_support_nested_variables=yes
+else
+  am_cv_make_support_nested_variables=no
+fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_make_support_nested_variables" >&5
+$as_echo "$am_cv_make_support_nested_variables" >&6; }
+if test $am_cv_make_support_nested_variables = yes; then
+    AM_V='$(V)'
+  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
+else
+  AM_V=$AM_DEFAULT_VERBOSITY
+  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
+fi
+AM_BACKSLASH='\'
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}:
+
+:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
+Configuring $PACKAGE v$VERSION  for $host
+:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
+" >&5
+$as_echo "$as_me:
+
+:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
+Configuring $PACKAGE v$VERSION  for $host
+:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
+" >&6;}
 
 # Checks for programs.
 ac_ext=cpp
@@ -4334,6 +4514,11 @@ fi
 done
 
 
+
+
+
+
+
 # Check whether --enable-simd was given.
 if test "${enable_simd+set}" = set; then :
   enableval=$enable_simd; \
@@ -5891,3 +6076,22 @@ if test -n "$ac_unrecognized_opts" && test "$enable_option_checking" != no; then
 $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2;}
 fi
 
+
+
+echo "
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+Summary of configuration for $PACKAGE v$VERSION
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+The following features are enabled:
+
+- architecture (build)          : $build_cpu
+- os (build)                    : $build_os
+- architecture (target)         : $target_cpu
+- os (target)                   : $target_os
+----------------------------------------------------------
+- enabled simd support          : ${ac_SIMD}
+- communications type           : ${ac_COMMS}
+
+
+"
diff --git a/configure.ac b/configure.ac
index 14fd45bb..14170f4e 100644
--- a/configure.ac
+++ b/configure.ac
@@ -1,9 +1,25 @@
+#                         -*- Autoconf -*-
 # Process this file with autoconf to produce a configure script.
+#
+# Project Grid package  
+# 
+# Time-stamp: <2015-05-18 17:14:20 neo>
+
+AC_PREREQ([2.69])
 AC_INIT([Grid], [1.0], [paboyle@ph.ed.ac.uk])
+AC_CANONICAL_SYSTEM
 AM_INIT_AUTOMAKE(subdir-objects)
 AC_CONFIG_MACRO_DIR([m4])
 AC_CONFIG_SRCDIR([lib/Grid.h])
 AC_CONFIG_HEADERS([lib/Grid_config.h])
+m4_ifdef([AM_SILENT_RULES], [AM_SILENT_RULES([yes])])
+
+AC_MSG_NOTICE([
+
+:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
+Configuring $PACKAGE v$VERSION  for $host
+:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
+])
 
 # Checks for programs.
 AC_LANG(C++)
@@ -32,6 +48,11 @@ AC_TYPE_UINT64_T
 # Checks for library functions.
 AC_CHECK_FUNCS([gettimeofday])
 
+
+
+
+
+
 AC_ARG_ENABLE([simd],[AC_HELP_STRING([--enable-simd=SSE|AVX|AVX2|AVX512|MIC],\
 	[Select instructions to be SSE4.0, AVX 1.0, AVX 2.0+FMA, AVX 512, MIC])],\
 	[ac_SIMD=${enable_simd}],[ac_SIMD=AVX2])
@@ -84,3 +105,22 @@ AC_CONFIG_FILES(lib/Makefile)
 AC_CONFIG_FILES(tests/Makefile)
 AC_CONFIG_FILES(benchmarks/Makefile)
 AC_OUTPUT
+
+
+echo "
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+Summary of configuration for $PACKAGE v$VERSION
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+The following features are enabled:
+
+- architecture (build)          : $build_cpu
+- os (build)                    : $build_os
+- architecture (target)         : $target_cpu
+- os (target)                   : $target_os
+----------------------------------------------------------
+- enabled simd support          : ${ac_SIMD}
+- communications type           : ${ac_COMMS}
+
+
+"
diff --git a/lib/algorithms/approx/bigfloat.h b/lib/algorithms/approx/bigfloat.h
index 7f59ea33..32561a73 100755
--- a/lib/algorithms/approx/bigfloat.h
+++ b/lib/algorithms/approx/bigfloat.h
@@ -322,7 +322,7 @@ public:
 
   friend bigfloat abs_bf(const bigfloat& x){
     bigfloat a;
-    a.x=abs(x.x);
+    a.x=fabs(x.x);
     return a;
   }
 
diff --git a/tests/Grid_cshift b/tests/Grid_cshift
deleted file mode 100755
index 54d76474cbf499d5a7c8eede1e31c62f6efade12..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 60539
zcmeFa4R{pQ^*=tFED$v^i$z4Fy4t9TASQxJ0CksaU=}uz3Tjj^33)M*ki=xeiy}lf
zu}qg$wAf;+Ep4r(ibB;|K!t<=iKrN~DoWLeZ?i-tf|7v9{yv{OGn?6vX#4$s`#k^W
z|8(-~oH^&7d+xb!=brmAlckxyNpUutWL|O7RT8n<VJ<<^?~;`Fj6e<ASm`9mC7mM;
zm->U!4?mX(A=IBt*ky<|VS>RI4`|14Y@ALQ8|M-aCLFGlk%E4i8WJ+S?#pF7>M&`I
zj+#)CoPxl3)V1YFfdb}<2NQNtAzWhz>ClWv^%$?qh}UJrGofVYZ$dMF%o~3*jqsUq
zE?~r61d>PGyCHdbWq;x?6weRpxb&b)e=y(}L!JrEct1rv>iMr;4ukO;BfY-LrMrIg
z2xdK8QdT<W@-df`6<%CcT3)m8;)P=`zxeVoqpB-LT`KZT`Q)8E?OGAhEXSe1sPka_
zSmX?E{P4EPKm9}HGS?M%esp8rZ4>QF{&t*Um*K~F1Ms`6a8Q}kc2Zn|&F+eG)!G;~
z7QgZMCCy9BXu0d;*B>c(ar?dVA}wD!uKsQ42G>{b-}HbpIrB~5;8lqkQc1tZemlH!
z!y@n8=MbgO>k%k-S}&oa(dc@kuYiKR(VLKg-sm&>kh8E4dOs9mZ*m6p!OwDc_M-ne
zebAkK$a$lWa+uf$|M`8;*CXBD^v~~u|M@=BeYg+)aedH7_E9gR`zVK>_aW!0KIqf?
zkh7$Z{Lb$~pQC-$?`M6`%lptL)Cd2^eU#@TeenOd5BlAGq`SM1{C4(1AJT_DrG1oV
zeIMoaOdtAu*GIp5zYl#*?L+?LKIF&sA^+7r=v{r}_pLt4Ii(N&xqa|Y?IYbA`_TVZ
z$R8vPlU6^ShZ2z_2juelC(7Ym&>Nn~6L^leOOhY_q`UqWOTQ863jV+9^jd@e4THYA
z2mfHunQrZKI=|E4zs!_x=#y>GHyim~e}>DT=wP`a$N6lYAej0;0*rR1a;Qtt&D$>x
zeIz5@S|i|c(8=#Q(`8T$fN`zbUsy0=1jn+=rQF<G=2evER{Qg-{JFVO?v(kt*A(4S
zTJ0~Y$|%dPt}d#Ua=kONatn*94D!sZjIxUIqM7+~%8K-`-a~RL{UudJ`GvW2ix%Y;
z;K>j68>h{v9)lDr(^7LOwAx=)u_*JJsWbAu1({H$U}1i4WldF4?)=g!e@%WFV@#dl
zzpTK&sImwwX`>NsL1|%8d2VTWsXueR5Lr5=vZ!dTKxX*UuE@<r$`uQ8=OFR?va*T-
zovY%q;<B3R5|iz6D_cQ+wVzyw4tfRDD$1XSB&v(5{N4gk(+W!Rt8)ES`KA79=#hC1
zXrt%kSC<x;+}@eFGgEUjXI!3}J$1(DT%kIds`ASVE9N1u^Ggegu7LVNaby`Ki}cb_
z((_8oQ98PoJ!7OJ#w|tu`T1o=X8mJJE2@#V>Y_1T6oDtrq+gDl=9d*k3Dp^Cnb}hd
zqJ_(ntiGb4VqRr_RS`o(+4W418WuxEqnOI7y~Y1({!?>jrilVTl75j;S<yli3n~`%
zUV3@yE#(zR*{HLs>db3O4Hd@fC5d7$Ex$!C{N74kw319@i>1f1&Gn)vrdJk~&#Ne8
z>zaApG&C%<qJ{Z$O6RAsDHO03=d$^r;EOBFMm7hnxjeTZ-(OH7bkJMnrK!0ya*O6w
z`WNL^6;+~FR8-~PQp7m7EL@mdSyWYBQJ!B`is1RFMzrb$`ITr5tjw9$L8r^CS(#H)
z+;=_@V}^fBK}C)K#JFgEeZ<uZv!bjzlZ98lV4jE^lkH2>a&za;EA1ugsD=tQD5GH)
zvK?2<M7Kvkcf0ajYSdB1=*k*@Cfmu(naJ9|RZED(+!a+-^`ao^sJd7EX82JBM)xf%
zD!;{Fk_!nHRngWO+xXEG*qRE<vgb=MZ1Z%RR**lZx|sMa1qHd)q@_u@*Oo6REiZ)W
zsjit<#J~cbxw4|VbYbqiqIsy-TgoM57*&}|%Ljv4oL^cdRiaV)iw)z1e!8&KFIAV`
zBAQ8ckzYbw0~Ay)lF(QErSpm^iVO1>kuAT#Us_Qv&B-sA3+qu(MD}?V^NYGExz&s2
z%|V4C1*&D(;(7U{<x+7~QIS+yU0G9J;Fl(6WaN$>l`2j4c_(J%UOMX1QI|!Z)5hqJ
zIXPe%Gb%j>OB*X?<m9*~dUMl8jgAR11Q@Z>(yW0TFXAwVi4%813@Nzur+CbA%qI-3
zcxGrE=B_$_9A-5Mm__UH$-($1;hq54AMj+snSlE#`1Kbw()B++alU86-+qjtrr)1^
z!J<<n#C;y)KQD&AYYDFzR(cX=R87%vXo6?PPsMNL{lA3RQ>51nn$90@;qMQLIzND<
zE=_^*CrR%b{3Tt#0-hjg2L8snxp?X?bsBiR;72JS6{eXl`0bLz!2Kx?U<w^4ox{xG
z>ROj4AW<kDy&4R^17gKI&<t=Hegx^98=9Bt?~u;9qj{PB4(XShL_pKOApJ@U-SmS<
zzrsQ{{T<TNEp*fGApL3!-SmS<A7`NtU7sf~&Up>jOyEunovp>Zk}Y(-{plg87CNdw
zdX2TvnYMYkEOevjMUY~lQz!Gvw$RNr4~FGf=+^qmv(V8%qgRQAo*1RVkF(HEwa^z@
z=mRbES_@sa(3e^0Y;WeZ!a^Tp5&>6P==2ZGYqf>`Ba;ZY#zLQ9p|7{lxz=G`8!Yr8
zCK0f~LO;VoZ?VwNw9s2E^kEkI9t-^}3%$)kH`mk{tXb$kw(xgY=x1B#T^2gm6rz{W
zAK6BnQ6emXg?^QV?y%6$wa|xJ=;vAJP7D1f7J9OUo@Ak?TIlCn=wmJP3oUe)g`RAo
zD;9c+g`RDpUu2=@Sm+}x^gIiFq=jB$p<isFS6b*U3w@!5KFUI`wa_oI(3e^0wD0D%
z!a`3oiGZst^h+)D)fW0_3w@1+KE^^<0-q-+!F~@u<dl^927f>8j|fx(+Y*{AOn8i3
zb~jie=hxvbpX0(4@e(qH+akEmznwTuVYpS`Rm5om!wmwzl{igbc)h^q5U0rtuNL@?
z#A)inD+GQGahkYrt-z-cXR{4g3Vb4QnzV49z{e4%DGO%{{8Hj46L$&xBH}b@;Z%X2
zN1UcC>=gJg;xu7lhrmxGPSX{Z1b#AcnyhfgaS+arBTiElZWH*?RlsSY!mR>7Nc=S7
z4Fca!{71ys3w$4Onw;=zfxk=qbmA)n{yK4*o^Y+ecM_+`30DeyEAgSk^8~($_?g7B
z1^zT~nw+pp;ExlhsR^eF{13!wV!}>=|C%^WOV}as2Z+<8ge8IhoH$KMxZ^wK|8C+=
z;%x%Ioj6TIxK-d)#AzDB4FbQFI88!$y};)Xrzr@p7Wj?CX#&D41bz+i3y9YWd<t=z
zf^en4ClaR#2<HiW9C3F2aJIlNCC)A%b_x6<;_T|-RDqvIoLxNZ6!<XW?Al?6z)vI2
zt{avFell@(*>J}(QUAo*Rl{upKl&@+?4sdTfgdE!t{H9sp1tK!tZZn1gQzWBE7?qR
z(+2d)98o^mTN*Ya{K05AQ4y~F9yIy-26<4wEsvs?Huu{Gz-vqPG6-skGA@iVKBY50
z6T{eMGUkM*fwOp&d`_K^!~UGTr3=B@>CyCw8v3z((!TJjyKh$1eM;ctj_jG<<_1^D
z1!WprmlHHsp5>CX&ydlti{*1XlC+^2S-V!bxB);3bPZJ0PZq?@ZrZ>EX=uKly@f%M
zHu;=d(l!t?`5U|9!LO)WmB#&7E4Ef;cUKJ}Zz7{o|2iTo>JB~rJ|lkZm5YAkl%$$L
z*CM921Qj%UN55i_Y=@edKP7PGng_t(KRc?z;A^*ZJPGa8Pa>_e!*?Kxe2x^Z*5S48
z>)qG6uXWFyq1^TPDwr+rW(4rMT74mjt6PftY*H;@$iPE7t_>=cByH9Wh@v)ogI9Qi
z*C#0I+e%7jX4)5u`mNIVZ5*pJr7kIt(VOd%Dgk-<hx*bF$v<m@Rncxjp?lTsh5(<s
zR&W#2h<ZbwN1%ZoRY^IhsBciNKf$LS(DG0XQ3i5TMFns(&Fv=IWNp@UxHab(N-j<C
zj<`BA?c-SWQW+KMxd3_j2MP5iQN0(KhPutCKAJ=}!m)tfko{q*#VBUpW3>BB+xjiF
z)7coshjMGfxS;@w0tu}Y$||AdhUiZ1Y$PeVKwVNcSmi-Tfl)+VN>azwOf!!Zs4E5`
zs4I#Gv>kOxwE!a4Dm*auGV&tM8bZWbPX=71ZNM|)w2<UOCAZ=sSeLX1<tDGIOX3PF
zxb}dGcI;r;3TXu$+J#dk3BtxQCN2i?SzCI#E{ThvzJf=EsLZrTR{h6Drkn_%8mUH1
z|AjvEw8njQidca{+FT2HT4z@zf_U`}KG~zweQ1n~>{Hv6#x}dp*2Nr93}G^$`*5=?
z*I$o;4DyE>@|i7-8Wub(lu3I@so!hp(nUIW6AWJe1&aFaiMq^%E@>~-Hz=}axlmD2
zmnVt1jUU+++slgjun<8gLDb<!sL_h?6{0(k1g~iF5hUM7I`K>gSw=c~mR%x4_dt@O
z&Pqx(YNozn$t6DZi;xmg8b7ovfksE5%eD#4MpESIyVDLU>f)pX?Vip^#E8(2P6-m?
zurL~gHhYXBXe9<c+W>hXpM9xU2#Gz4!Iz}?)WW2(KFD{mRD`(hQfY%kmdYMLQ7W6@
zg&?&S;Jn-~u;gJZ*fSdN=-q-F%TQ8+{-gw9UlsL`Pwn)mUuxIDhX}kNsu5z}UsB{z
z+YTx6MJ>q8A=%MbELXM+l7}>;H7G%-rl<$g_7<-wmMaD!<lWWs4SN7z|3<!H_d&V!
zT{&l%qBbjy5vS7lnOz<+{HjP3TUB7IgM~U?Z!e1*CdumxpOn0j?e4&Lc6s?c7)B+8
z5`&g(d2H#(lG)yva;l=ftgsY<o}{tqjq|TlRAeUYi~5&*$d$`~UQAm@o6`~wbKDVC
zzg9G{top;WaGlyNgedA-(e#y&=f@&Vw)%Ew{Yy%)4B}2x)B_>~N}7gRYVW6`!Nn*9
z+k1@8{-M-I;Pu`Kfza2mCsh3fmN_n8(6qc|NkbJh9vW0%il#Jbaf<3mO7<0Wcte?X
zK0DI(x&w<J!Dz;qq>PpCJq<o-mfDFbN`U2%<obtj-@<i^%_}g3vhpSE5u`6@tl_cR
zuCHlbcWi!y4D?Bv=|$iHRNvRu>A~J$e<kSBrUf5OWsjmcsrxlNVt4K0q_L8Jfcxj3
zq>T1qKJ_#0$KOC;(4GYT)Xj8?L5#G@n>#4jiPAx(Yl<jU^eIsjV4_K8O``v!0>^5f
z!H%H>Rv`#=%rsdbD+qxLgA#l}Hv(u2U7(8&CAd}_T%<DA8Uts$rbQOvpszRRG(W;J
z1LwthO3>a!&M7m+lT^vjS_a|T;=oP2P-I}cqL#6pmw_de>7Z)02yM<p(NJb4fvBB@
z&6zBHDXqaBxHA#8bX>l7G^QBv0}O7^6Qqsgk?-aC1~@mm&(xgBu~d&~N8oPEREmqc
zeJUdr#Anps>!Fr#7BO21i&zv~Xl7ESHXjAd5Ei<k9@i$FX=csG3RIhAN2cVK2Murs
z?Y|RsVrzNG-H9gWRtNvi9YkPox?Nt^;+EHC+TFqH6Y+3uf;;~D1bJPnJ6^W!rfFzA
zBoFc=l{9;j%9siE$Pz_uR2sw1V!6DbSRQv?qwSEqWVujLp4F(mdMFYRkz7XUXDf)E
z$Jlz&uSLbc+IW&&hvXX@4pP389Uxl?DX0QpQl8Ni@R)PRBad7wWP1b0`&IqO8=8Mi
z2`x^_)@q#S!>6JA5fH!Nq$wf$dP>Pr&})<HjTGd2QK-<mEGat^7Qx|BI|6OC%~Zyt
zhW%%GRP7-a&d|<u^;(d$hw#wIol40aI38CuF{M)pmHkavB8=a|*CRc(<)O|yj;Yrk
zGaecr>Ku>A#KtRu9dSxvdx8>ZN(eL`R~u_~%cJb)&qU{x@2Q5^EnKzS`~*8ICz#r6
z2hp2J(|fb@75i7y=D(AHjZjfP(4GJfAURUC_d9fs26E6!kV7R0szwRIbbg?1GI*NE
z!@Q7K6y>Re1^PgHus0@GHxmZE546hOn9k~EN+i=%gK0aHV_PTju*k^=Y_x2=+66|=
zc3N|`+hp=5m1uT@V~;gw>vWETY%fl25K~s0lwdrqnG#$(R9Ke!FXXzbVQXvege@EP
z^URF)ODz4z$J!6?Ls`MjYgbWOwEn1h15ubaD(x2Xo3&D;s@K#TR0h3S(#|(&97%Hy
z=(X?;gUu3-2Zu<^9jp;$aH+u+QxES!0WgR<7}OiX3p#@+|Bf&9bYdj@MrROZ|C-Lw
zTZvqO9E%d^#}dI<eL!<Fy}mN<5k=jsl|ekaT_Jip+Ek7h!Mfsn>R}~i$NWKh@9V^J
zK>Hx!UsRtwHRSmlGThzymfSA79PF^Hc6E2S582_`Vme$~cZUl^{!jXwUT=G&^)?TA
z7Br9gskZuyUg~WsIAZGUc7rRX)XxW(S#LM=#$eYOM7@pejo~BI6(opyJ3}yRH!1^7
z`Qc=}Desl*mZ7=T-hrmv@8=mA?Ps!KE+e~W_4=3$n$c(vXv>i<$K+=rKCAu<ePVYZ
zB4lwK^r;bTn23#$4pkp%P&47Fz(aA5XD`4QE-!!R2%Me^wWmLiM8bc^bWK~sr!AOQ
zYQN=EIefP0coxjGQ-~e$;m%!!sKKm+NTZsS;2xiqATLiOB{PvciQq|;m%j_gynDP;
zs3Tir0HTG97ika<kb694s`B!q7@FHp*3M_bv*>R`ngZV<rB^`L*3CjUhJ{biov{uf
z!f+&nd-lFTaiUl1Jg*p(Wl_q>2E`SnbhRTLjHYx-wGmjFQ{bJ5Nv#n24ue7A%vcxN
z3`(@HHvxgqeAhP4{%>;6a?f_(?9QFtgjp|j%|;Ya;_MU(T*#k`$PxdkteiW6w8xXx
zESI^USn`;7LLQF_27NlGtD_l|?mS*#P`dLtm=rUQli>x$=JA{TJ@fbxDA7E=1%&S5
zacBc&!xPXX!e`+|%>!_hCI87(TG3`99j}V@ix0x%5sqmS`A6V!%aixHq|Li<+Y~?>
z-}n`tv_~MA3@7Wd+CxP0btIe&sEt69+6B0U*Q2R3cqYiYT;`!W(`k@JN_VEu5z@a1
z$0J7gHr$F=Sk`L-pAYq_Zz}3HT+ImGfmXRoyAmy|SU!LW(b7O0v?OdbV1ohI8*sG&
zR~WEXhZ&WKl#$09oS#4ksJtkNlYGKt19I9!JOL1_3IH#qIlke37iy>`K^aj4-%)8i
zjyXiNA{<ONM^wpw7E~_L=3$}N7tBzUV4#*UgF)I4U#KA47kXZZO@X_11onG}lCo<)
zCOaP}fdjSLWk$UAQ>#yb^Hi&-0Y-$FRl1nf1U=%0o@bzT1dGi~(-K#3hn{E2K!gs0
z)H7BD9K6Lz4sC-e;LK_$1d5>nBXITwiV{TO1*hi3#vjNwc)J<<<mv=HVn&WH^v4!G
zZ5;$#Pe@zDy~>Q+oo#aZLeI46(RC2ih{fa{(L1#W*JomLPEO2`fBf}qQI)^x0-l-n
zQdWI~Tz3lUlau=dUoZ>hQjHPyiCTnb1+Rsen~=#nvVFlPSg5EtmKqjqFgdW!4>E%G
zjT2S-=83B3MG43zAX_E|?VIIw0SWR|ai17OC@6M8;ayN3%IKW%d<J7aG_iBS7DU|$
zCaS24doXYe&lBZ!_ARP-%!I+anDf7h!I>M6Az<c)V-wZE8!&?eWs8N<{$qXpRxmf)
zHy}wY<n^CuK1Q0e(wpVGuSBQIQX74+4GCx&qD2OSV%lOB?=CjXAvA`ou|nYsPE}Z*
zkD?<Q4Qv@2Cu8_RPdEV}MhC%UVm-wi&~C7ZnH?*JF;Lm;hY+KKps*J)V$Nb^pJs^h
z!3}ayr9qT+nm|RTp<axeRsWJ)_e(r@gO0SrKD9*&v~?lE8<=IIRD9|#CGcsN5?a!w
zG#-e@+HR!Te~zLaQ34;@YKFktybk7m&)g&|&t5#Yv=2EXR}91N8pn$X5CK{femfO-
zN!92eJK8-;$oF`vlJX^(e%gxL4Q;sPci@&%ERS_<<?UtOw(<5V)?{3mo^;f_hS(5=
zwVaCG40)Hb<<l^i8|y+El`S?twG_)sUh}4ZRbyAX+7~LJa+^0jtf=i?dGb4o`i_$F
z6_k%xQV#hCz&h2x<WE)tM{NExagVEkg5R`bayj8?{2o^#m}h@UOo{AJu|V{#z3KwB
z%bXvh?SaknH26JFWAf>FIuUx>ed)WZK1RX%-QHkoR&YR8aBz}0I2%Pga0^OWO8dfC
z&wr?M!c%T_!kXaVr!a*b{J1*!3Dy1=B=-~&_NBjC^*%N!hgybZMTI^eXdbM#&<z>+
zsnt6vUcqJj_1JwAV*(8k=GzII|F{BURpZATLKkCoC*+3}I}b7Gz<_0;wvp#Ql=RnX
zoD3Yw6eeB>DP-3cege+Bwo&X&O-i7lOW1K1+nF2DKO!A+V*-}{)*?KpgZ2TzgwRhT
zs-ym;rB87dwHjfX0}8bNPEniM7vpw9+Wz7g*<<^Xjdj{4U9k@+5R=Actp@W^!g+w*
zNuSQ7*D&b^KSCCesqa_@fZqTL|379t%~1~t>+QpCM!{3GEih>CY>c%47uN!^LDkp2
z<hrGJMCh`|=vjp(e8C{?=Maq*3_3)zqH5?&HXk~Rs`&@grybyJ=8O#dnb;nfj>-0S
zaaET^{Zp3e0Hv92L{cIxii%FNH32SXvzt8$`#W0`1buhSKd`%Sz#Z63dx*7{Cd`<8
zDd?jbHkGhmhxNpQCi&j;#MIG)70A855Ecx3*k{CNKBX6V^92i&5|$`d`|S?SPv&CW
zyJ|Q~)wGL$3>DC5TrlCQxx~_N2+K53quFynEEJ*Dy2p#6TfB|y)<ZWSeJSmAu2Up;
z9)VIXa+j6)lkc4d;0d`R-bjN_ZA6zm&J`4=FZ9#n-rf7Xk*%JPV#9V#8Z3;l9+g$_
zvAP=zKfZ#nH#p(#EVapdZo6oY{=6(~nPHsUO~WKsN&NB@b@At^yp7#VJ-mUVDC{Yr
z#0!+rEUcY{5?|&d!U;=Y^hZTqq~X_rAGR3MVRxKZ4mnoynNNMko8BVV{S0D4H`qMt
zu9Wu!pT=jZFS}Fr`t4Ysxj1m#=KnDkUi^tuLSs_UlD<w=B8}E(ukBT#5y);Dk}K66
zDZ8eGhM((I+p$zbjb^<Ei~C+QzlwSX^tl82+yQ;CMAL{x|0ZW($FZ7^mB?1N4Nf{r
zHhrgD{~-!i?^Hn-#%7c~O&xX#<YW)#Nhx+ToF@19-?F`krXVjrj!9DZ=fKgW#6o<b
z(LkbmrZMef$(1XN_TW+Xvq7|J!?DvY8br{(>Id6{ur^T6r-ozn!t#3N_@}7X;`^fW
zzVkj{cDS_=*p|bIg4}Hd!~GE0jW)G21JRh;`({ji{zDXf-u59p@rT=x<7~_l5uyEE
z)1SufXW(;(TYb%~zU5WFhVc}O$ZEV&u!d@J-$+=2K^YFEIl~E(<aa9Seto|P&Vjyh
zbrq5_c1e727}Ti_rSTK&61}RZbHP4$DE2{WZfMRBT^a2cliC4T%W=4a*dp+#M<cCj
z2BLl7Q$O{n2R-T$ZSkqBoINN52lFdtW*h>7cR2LP(-R_h2eh9cW4%sMj=pCdA!6)Z
ziI_hiX+@oo<O^PqRq#R*twC^70xZa2Q3gR5LOn@Cy;v6V2CoIeE#9=fUiG*aZv6?%
zr2ZfK)D9S-tK|AxbUEz~RwH_8H2N$|DPb6L;QKy9$$4U1_f4hoJ8bK+o>b;2WPgHz
zIUJ6%xrjeAX=plhlIy!*G}taUF~xNB+kb+415669bA0Dwgzk!y>)XkNxIPYm=vKZY
z52Vn9@l6y%YbOnQDk~F?qg?lA<6+q=h&&}U;X8(Cn?wN{%e=|39q5PH<zZ_}q+y4#
z0l<6)yQBk^X2!t`NC4RP2*$JoFIR_L`hM1lTuS669s!`4N8@toX5NIt+#|fT26av)
z(A!h7a-70vQ~R_AG*Bp&tQ63Ou*n56>?`rNI&}Xsp6<<p!NLZ@UFGQM)bg=d-!3C0
zzUm}s>0o&$>#N*5&}MYQsKQ&AifU2d{$diKo}vW#$aXM@IR<Rj7VHIkfhWnM1nn5t
z2PY|NoEK-S(5ihKnIF!3s-~k{u)iqkIFm5~+e$2b7>$J0>g0x{crq$@7+KLQCPJqz
zA7N!|v*TXO!52X6#w+looe735r(rg(2P}*RNC?oLK>N@h!!3N#aSCdpti!-zVm*>H
zgN_pBt8kE%;Pb+x7R~C>cflNdUU=$)@C%(FW&nb4w@zq7a(YX=O((QO33)n!6C<5t
zx=vUXC0wNwYNLdU1>ueoo&OviGYVpmj-~2h@jB)p*8isP*YCg}HktNJoaYF9p6pc*
zd)4pIOok!_32-=}QgC|aA}4&Ck8#LEz8>EHv+ONm8y*#)o3d&M)c4@w6gabF&#_=X
z8L==iy!TG9_qhYjnNWX34qn{05bR&*!+hFaIl15O&;)w;A6~5L1=n{$E4AaH&hOMl
z^+3wow)X`huX|a2H|6M(INNS$iJk8oY%NN`ZYm0M;!tdFdE>8hAjq;_E?@sTws(if
z*SE^a!!RG%hmGHl0QdgY*!_J2!TaQ#SKy^%r+2F((wN@37`<l?&lR}kl*9oTC4(0y
zLHG1_dHF0@73Ser&=6Kx4P)Mr)&Spb09AU)e-%Jl11AsX8&4Si@#L0s<P9h$bQvsQ
zd<e-oiL%1A!Wa~r%2gK`v?(E9da9h`ZN;w*zYaO)q);}7`v&gwyVCbB-j}5w^QxWg
zKLSJ=?WdrE+WX@PII`C`P0neQe<Y0cRZ5^aLHP3eIT#O)Y{)kvh<0lb`Xl^-B&~c8
z`$4ij2{+7{PJan~GC8(S8l&oG+C=cO?`%TdcCUwKYqtZ}>JiwBKkZdt^<unG4|%Du
z*v{;BqGN3sb#>8LJK+jy55k%B!b*uOxU^zuuf}J=r%}=lXW<A(-32JyEMpth8(f>z
z#`=NLbK<y(`o1^t>2cu{ZU9;G*Z8nWI+RNpnC#q%&5N?61aIg@yEjx~$Lu~!ea{>C
z<ain?a)n!McB?JHjAOyk4|U#o4DAuk@=mNxOg|>C8{n}udTbbBP_)=1<6fCJu>E+=
zPI$;%B5*_(N)xiw7F(kattzDplh9r5i+t){CFNUx0*+{4M&LgKJDZ!Zr8!}Px<g*)
ze^IjS!n|6=y!6|+DxWvy+pLgp0~DOK31?%<UW5qUUQx4C5O2{jY?7|K^O&s@TX$~k
z`$0;LWIOH+G+K0PI$pEOt#aTNTjEX@GRhZQ;%*g=2RgfMOrCp_){efGFl~P(3a=}+
zC0O|HiRE8F6iwrIZ(Ey&Xb;ijWnT3gxC~g4mFs>1zrYJ0!yBAtj!(YyupC%}y7ESc
za!i#m*XL-}cs$;R2~|p`Fa3agcP?W4(!Ye8mL*U8#B2Lneqd`>dYk+}Q^Ffra%LO)
z1WpL}omuJI<eyCdGe+`H*bKCex5Ug0f|<IKj`}$&*WHSi<O{jFyeTbA1xp;L2t_@r
z1op+FIOMwf5toLtei<GRCHxv7mzFWt2y^(a>4%5*-W`A}!vVk~?`m;A?|(l2dmBHA
z&$5*#gl9lwC2&-vd$c_bk65&TVz3Jw43yv?uX+Yd2HH4|<Y7_-QktS(N|KmA_Cg9o
z>{T~nf7r?%T7;qaEf`A9KZ7>_xe@W%_6-VLh8tl-yD45wIi?IXDM~tAB^=p1Qu7xt
zmXxN3W{yo@xye3#TCAwM@kGxF^ERa!RfT-%04Gi015QErLcvY4*mus4!Ar{uULA0n
zj5R7K1^>yICYumcF<t(uW;^Ct`@O-d^d`RTbuR2(<yCJ^m>RNw=2a&N|61&QG;O&I
z-h^fYi&pjzYm+WsA=>lI8QEU-H809WShg%UW{TM0f~RCI#EJE4EC9ftg{p$3y?ezM
zZNER7e@}L<+uhic9lD{|8vJSEpy3PRSc)%r3Yu}+-W5UnZ?S~*JGW|o$aqu-KZFs?
zm)=zM5&Cd~F>Ao#&{_zVy8VB6;V{d`c_6MXjF8wK8K<PbBL`jpUD-m<8rRiXC_Htm
zFXgDxcoggT$CZ?~l=S!IyEl<5{as8ZeDcJ0ukB6rNOsi+8WVQ=<jezbG+6CUZ~Axg
z&p42xM?MK4X*a(iwnQM9sdM!LQyWeMxZ3SQ3hGwvc2t8mG-0h08vF<&r5)D(+zNYR
zd(RshzSbC?BHAI)veXNV3SG+zZ9g67u;O?o4*mIE?KW_DZ8xL&>ZM}rbLo{m7=hYS
zB-K5%S;jMQI<^*xyH(iJe~6NU!5ok2Td+DIGp!+#J(QC&gWwh;zenBYQNKb+`u3Wi
z(B;l?Z^S|oObGJzu1EbaQ|-_egFwSt(vF$R;-pmfMz-k~mKzKf!#HnZ<~cOFH?qgX
zu<{r;r47Y$T`iws4^A~&!0d1>Vw%W_?Ey1j#e+%27`OV0qP__~g9GMRV>@Fu9GO{3
z$yo(oD?!WwGD;=}$E;WMy%MYit;gIU(QWJSs0|Nw9#>~3C1LYN-$#r56bDy6ajTPI
zL+((pZ;*{`fM>w(6s8_)Fayv)Jn3KIa1jk#!Aou|F0{KlU#Agn+~@R)k(@2iXWNOB
z($39%w!Nhu?0nsw{<>SvdObArQX5u7yc@b0!i=kL`Oe*e!`$b#ayv|}W3$7uZ)nT_
zrQonPRQAl*LJHP&6?t8!52J76mrmObwX^eWcluGtJQ`Yj1!OiWfzPlQegF&16{o+3
z2?L(3{wB!p;K$YLEUsQ>f}yhEEZkngtr+Xrd+{sY35efutY#m78Gb_>Y-Cr>R-d}P
z-7X}`$8mtMqD4OKC2+tM8ykm*C9f&z`)f{itH;#$!p|ZHSX1rJv*%=+x&up4Fb!}7
zKdwIvN8`QrtFa2DzUM!m>rf7NU_#I)|8A@N-0QZtG0*%#Ts;=(vRC<#qG2yAF+8??
zIQ17i27wLoa}BmuVqK>At;D`&m`8rD&29S!Mw?^l;ia1~VNJvi!@bW#e{8jmiIZ0{
zIvT+6O!Q5()?{=euDzhisQba5<bgpQuC}A=cvKuini#UDp;funR>XJ4EuPBLO2^f&
zQg(~eD=Ayu>APfKD@KttkD7!v6*OBno9~zGW?y+Aa*Xy7lE;Eddy1HVMAx?jv-(Xe
z@1Nm)s$q9z2U?~29!?dr?O`%I)=j?w```{WlNMaTX#{U@z5~V!1K(CGWN;qN6Y8wg
z25gPuSYC6c6G=<{bJ=NK`s8ZJE0%R4SGR~PjHp>d#pzL=ES~LD(sy7`3p8{nDIM?|
zJFt+Ep!0AD-KJPPG~GY!w(scAIhpD%9|{^V#Bvl?w{c>EJ$fu{Rl$CG!Z_?XK>ae}
zsBhRn>*SU*cOVK%F<uES<AgXc;V)RL%|IXsVs;BZ+kwrBzbNSFSZTzONhNra5>(AV
z?1Sim3C6law^4-%PN6B&Esp!j_r3}<2IcEmN=TxXb8b&?<QlGbBzxp_8Xv~`f~k5p
zT#D7_=t6u}@G-jhStB0fEZf`od0ZxZzf-+1Q!H)m)FYV1)_=StIV0q|5_R>In2e)=
zxX5=N4)nkouNjthxP63Peryk}_FJR+u&LcCsb}a=oUK!_c6$t7i8}>}%=$b1^S{|m
z{RQ7Kb%r=!$N7VZ{v$m)d=wn8(<ct@z&4(+dktQIlC~Faja(=81G(V8Pgoo=ccF3c
z->&GJa!M!%>-+6+P@DWia0VLta@g*IJwjs>3;6zlFhK+Go33xXiP=Oty#96AYt!g|
zBJ4Wux7|#ORZNRD3I+!z>Ns}Wgn7bV+YYR`Hgwdyfn_HUdoxTs74=mmWqXZ7eWU$0
z>wbX}_ypSz*c07>tu<r|!#;L6Y&fP2PtM;ze(=vQ3AWzRT>{7LRU_#*h5SEDRNr6+
zjrPy@o4<~~xhwwWWASH0FLZ^*V>Ru8k8HXZcf1Ab3}~$KANO!~+JiwcWXCK`Q&t(x
z(u?+qM(Pd?KEq?%uM}V-0tXQlbXqPWz~tJ81Xqfqz4foxoJQL{t^IUM`gF7XB>$D`
z1I6-EyUmS30ZBm!JA_ye;)x{B_z8cJmruZoI#W!JY-R6A1_%4l_`wL{`&GS9tBj%j
z5@d%S<MlhSqV)<_w5H>>sR>Pc;~cuAqrlKzHmuJd?(fMx`n$8&{yw~Ce?Q}Y(ce?M
zb?l|T|NeXWd+PtPzw`X}fBXCYKl?ieUME)kPueaz{cv=8;XPDO=&R?W$DYvp7j$?0
z?%s^G5YdtKK3rn-;Z(g3qoa5C;gLVohv#n7`|vMzi1+Mb`e$;Mx?SI7a*ItSB{)~%
z0;m^f5RHi=&KmGM3FbcswCsblC36+d8?hm>#tW|_ItIeAfC~-F04M(o|8x)D#7WFj
zKj)6K*nBC#His`XQp~d`><~(gEBx>UunMSzFur2&cZ%7OPu;K0XyW|o6AxB<6Wj61
z42Ie6B=|xfJq4aJ!x9}F`cJieS75+Z-^~h6QN#(7bx_3{d|(wdf=d9W2uHslvW0Vh
z{U}M6ial$0WJ+l9y*LLy@@^&lOSk;XCMEqS?;-nLaChWiHYn*wK;D`b@zm|Dnd5P`
zW~uvXX5#<~R_|`W?^bLew(yV&)+;c8WpWhD;?oR1!R;RF!nifo^$r}c>p@z24!ms)
zLrwT?$FHHQhD$x|IPwQQ+h;+%H~olzrV^^q&xvR_C(_8Rf|PHue&;_$!GObKA>Xy1
z92|}#NW(X&!#8uEsmdATVGaA^s{N^;=Px`2^0-~~JcR|@?QuN4(dG@_n&TaDYqrw(
zMVwNADI)f<loU*3ID^I-!CN@f22TY!dPB?W;B5Ha_EV})5{1Evs{dlio!;P-Y_@cq
z&OucVP&F<f4)Lg6JbrwJ4)s(2YBSeMd@0{RCLNPmf3*vDx7!{IZoB|+{W0@Nya6Em
z8{-84eQ^`(npg|1c`aJMt$O|H4ip+kkNVw;;4JlRZ^#?1-=FuW-^67K9Q9u|VqGX&
zzjFOOi0E<d&ca$;{he$)qMmb%dd7<Mhh3KO_y_8{Vltb6Y`Qe4dlHsnm&@7!8oO%#
z&IT}uhQpB+`U5l7UI-P8XEujxl~6941J4VgIl!-n!}Fbza)iy{6VV*LMRORrL0&g<
zlWioL!;!eEb5{ftSIO%Ve{D<rZP2$0?ZNkJyUq7o?%AU~yj<0m72KpdtP_&FjR)eq
z1+N)R0`CB`Nx*Y@Ykq39No3(zG@C?<u_wmu9=UEMJPIwYK{qrPZH+`kgIOkeHg_`b
zy#x921{1|p2(5<4;;_BU!|ZG~SkV^khR|p?JKWr&^tr_wBSy;+>mPz!yf9)u_4USx
z8?TJG)g3jj;s_)gt2cO~A{Lh#gfdxJxWTdqom+pd63W0D#>k(6lpbE7xAyer`ID8<
zZK(GO*eL9HpV+^zW96QW%B5S%-c7<%I)>dndl#21(+~Nxl+blHC7nL3B2UJy-*yGv
zBBi6d`VO}D503o1ZRB4?Pd3g22JOGH+3yc}er30L?nkw4#X{J-Rbk!Tybo&zaNw*v
zB-&I|L--d*Sr=k!@i)M`yPJMqcZM%GSuqY(?nT}Cg7>k-`GUXVjL8?g9Ge>0A`g6I
ztTY@z9sdY?UKEEnbTfXpJF@VSb#y5L490w^*dtdo!;Ni4$=t%?IS0Qx_%H&$X?(aG
z?>P<W!bD-*ke6%^bskZ1aBc}wTJkD3-0&V0oPWLtCAi`QR~)!V+$zwnctMOei2~IG
z|5s6Y8svSW%R9=bM?|<L1bk)6J1S632>8mBcNC~`%<o&#Z{6~7Y_YIPX|S7}UM!W!
z^*`bqTzwxL%x>&LiCbRsvNye>25tPE@a-s4*bsDGtR{YouDge=3Daks_B$2xQBL?^
z%RAc#V+-avZo&MOC-0ti+ulHQF=HBs2Xp-cM^(wQQGE+sV>a1#xr3g~c90ftw5z*8
z^KG)*ux-<f{@q;Fs<zyuAJRBsy%h)Z7P3xN{l&hz=zH9ipz`ttf5yQxxJsOYorNhl
zR)c*nMk08zZx(oz5RUkRfrr|NYYhoN@N!_bA%M@c(!BV*2N0ze!3yuaDRAd8`A<q~
z;M=x&h>`XZ7LD3}(qnue4_{j(hl105C4k=G@AOgqiDi^L;)z-UF`&H}bz9z|+k%b7
zPVJ#EPUW3$<i9!*L)i)iS)e40W{Z{JZ$$jy3nCGWXtEg5jQ83;^q<bxZ17eZMm7QU
zH`+L`xw;3oXD}Dz{yitwVmLD%`X0?N*?!#{+#+&<B6x*)2(1)vBP*fhN!x&DVL^60
zP73n1W*m4C)!_+>6F``ktQ9<-__ZRAeo*3(7igciVQ-Vi;F6TkTKz3(yptPxSg$Qi
z)8Sstggb^sdekZ2))f?fx!8>%9@~*7=@NsK&`la{kP=PRvD=b%c%uW_;?;TW$5`=2
zZ80<6;C;;7cy}_kNt5LI1EkSeW4l=+_&?QVqZw?m1JOQuo|^CQhOV=_n{g6^rD5_N
z){0PB?%>(c7m**_h-Ub3(xYJW7T{wIx!KBGZ6SxLWHv|AwKPa!v!fQgWCy-KKsd-V
zB9E`YE#y-+pGx@T;#0Eh!19soz#$#kfm1>9pjjhQ<v}w?IORdU5e^jU2spYUDk)Gw
zlVL_+jG09r3Y*O|OZ^hFf<stM=Dr`^8Tw576bKJ94s}PixzqnCFF#+n^m>bPHxEE8
zu{Qo=tdp}-b}yrapFmWTqi0}KJI?##oEcV_m`W{Ju^0opZ?t87lqdz6v-tZr>4QSI
z*8Xjpk8xgX-)Y+K?)d?>lz8?G_8}|%Yq{?CPz#lF%$KqcqYy@-eaAKjp*9o#^WTxz
zluvp7Dl2$KRxlTP0XX<7dG9)2j>+|c5$=rxhz5>lEI7x#5lx8C7cBUZz5#*f{tE_V
zg{DQkyTe&2Z+P)V4R2_w&6oa_SDv^_tVOX#@iuo>%4^=;2d0LGr>7kj{S7BM)edhb
zj+MNL<EK}oy)K?xGlC=QGJ+G<rlH+WiU_KE^n{gaSEF`TT32bqsl{ia!0`^NFa7=b
zXQ|CT+ky6TxqiXN5882x+!)W?!I@8EYa8K;?QPrk40U8349I9*qHV&;3>7;;ShR;D
zd3Pmj+cc~V`_db$YqHdXSt&blY&6TZ+u|=$-xlZyuK@a<Eacz_n>#r21$S`xvo;*N
zbQ}7{4_~S7p}u=c8;XKCv(=`QLund1?F)!`lCDA8@Re?L__H{_i9~Du&|@nJ^O5fQ
z#QG~1h4Fq44jW*@NbIBV48$ND@8t_jXn{CjBF6c&!!*Vi(_L8j9PLei%`c8=;Be^`
z82s!$b!kGFhMoJfnE7BVYwU`H*%8a%KH0O+YuhG2&>+60vHvRh!L12fqetSd#tvK?
znymXpCg~rASxk4BmwV*(pkqhwPYl;yLDRzLSn%!+x*KPuU(xZ!mK+lfv>o%=G=IED
z#UZ3Gc#Px#P8i{gvi9b4w3>&rLW`fYv!0RTPA%_m=m2yDeJZ|Lu@oDvzCR%Y(}91<
z$vk%An(RtC7TEivCfgIW2&9PjoOmp)$EkO8vO7he?WZMBE;?8$bnRK*IA+g-2GJvd
zv>LbKHKdD_^zGFb;B+z?L?;-Uz4E#PC{Ip}#i*99RUcv<cedxvzs{?E>23Tn)fXyE
zN)Q8NbUl)@?gW;U4_|daMv&_-X^8#E0WCq}N2wHf88{T%9&CXs>Dq#wbuZQIXoqDw
zpz#2fLeD(j)+39?`tt?~=T|U%!ReLWz>!m`Zuj1Gq#o7)=i`Idmw4}LyaNZ>u;(cH
zqY|pZ%U@FlpwJcByUQ0Gj0wepJSBb%recY4-t>c6XYR$_D`)J4gK0yB`qbP!MNNF0
z(y(cU;{&I_OKMeA2#++@Jc;Hh%q?C60`60%SHgYG6Wp<hAs?^#f{AZW4-HOGY%RX@
z-O8DVd6zSAnEO+k4<{;88Wa`ksCFbb9mzGK{E;a|p0t|@HenN`wB-q8m@nn4TT^!K
z-f!F4xc}_Zolis}OG&^f!_B{X<6@g4XC72acTIa3#%$v!(C~4_V(0!)bP9YuN#i$^
z8nv^P(gwz+Z&H8Qy@^rK3wC=;5AsnMFvNjJ*jt<(zS13n?Jm3^LA^e8`c8UFQLpf+
z4M+?3-r{*v8oZ?~y7(VZsq`ydJdPt%{D%;a#fItVGBB*)qnDBSAIh??w6TYL<s@`s
z_PYP5{xEj`QT?6&Vf~Z;!}`N6{71`w?0;B)*MC@l<v*-{_Ww$M-J-yt+|{Go=IGnm
ziV73iJh_CnyQ$~=UwmH;{vDh<oPfjEYXWe=y}_HYhdjxlq~Pc&J_NVvz?b@k0|_`j
z(<TSj(!gyvPru^4!OM+T*BXz+g9lr1*m2(JOW!BoJrVMJ=^x^_CC<csj<-xO@c7a-
zoWn`j3CpK(?codmaBuo{`Dae1J@>nX+K1~nWHmxARl!LfjP!VaA2S4BXzDU=%2(bE
z>N5SpdaJjy4QDU<bHKogN}vsA<6if>VVr9wz#@f5VSw_fpRC{s@6&aAjS+e?gyDS&
zF`QtqU#LW0t$EKociTd*ZBjz`PKM(1Rd`Jp5`NFzVojBk^Op9hJm__m5}Z`2sP0PH
zF)2?T<i;#wQZ^vw5R+Vhm_hWH9n&1L<2tA8xEb??8}R!nrVd`r95BG{z!b&3kmuAP
zY6(OwG(@=!Q7%K2>qJpG5S0y43PeFMhzd0PpNa|`waa(+hrf=MiL}>^`7+Nc+1~a~
zgMTyx`j$DMuxAJ|?{Pq7&yZ|BL21tr6oAJ8we3SNg&N|(6l#bQQ>Y<OdzJ&YnGPuL
zb3l7LwoRKCVi!5Mc8JDOR@XWRF+m%$hj37xLqOmy1}-3Qx+4r15UDPd9m|I-6OmUA
zSta1ZL)HkUhaEax=@9fm%N>FbTB7Kh7s_?DV51IVZ2>wEQU|VIFbx5Ly9``F;Hd^K
zAn*<*2Q?%Sf8`K|fDaFG3ciOOI$Y@xbf_WtAfNH@3hFXblNW~UK@3HG!9hq(T7-l;
zhk(F07`TAI*BH2fz)M8D<wF(<xN^uc0UsW+O7K1G(BVplphFYE2Td4Hxoc}1LHZl;
zf8o(t`q<p9Xt!kv3*n_<!Js`${VX~^itS%FGe?&G<&C_i_b<hUNxE^=@twcW4Nrt)
zJkYC_CwS*>UkGoF2jqH8U-Xl3Cxcz9f&V1@ORKB(eZxCWE(>qVbk)B47rmALI%seJ
z|LdT|`0RZ+$p3ZF^f^sm4*q`~w8+wb+d)ffuy|`&5BMQmZHeXYUVq3|-xLR`^|j#s
z-e!J37E9iH{kO@>F2peJD`+<lJ}c^kBrX3jF10k{6#Dc~(z-0QSbTi}rh0Iaw*8OX
zjW15{MqX7l>v<@+OH0N@Ue0K?uY*WQ-QM`cIkxS)zXN0aUR7KAL3lX^6)Z+a6@VfM
zR3ywVxTzm>eh`Qp$EHFgg5^mV`RQ-l-sRERja|UNf$?y+@7$Nt4tvktE!Xk*K$fcE
z-3Y{zrys!gB=J3jz$e(C*^Z-VU&LbvFS6Z#lA^x};U9{(y#8{gBxS%kS@*TLf4rH=
zi`U#|O~HW#d_WI-e~Ets5uVShKHGN?jpb$P@)}M(M(~M&kGl9qc5=U)Au4wLMIMw1
zUm>#tzc&2fiEeLzXW3EnDb7~#Qx)k4^>?)Kg}lz)fg{I_T<)G2@?8|C?!-$Sa4hZc
zT}R+(m;WrhJK;ZtrOB+L5PN?N9I;njHZ_#E&K<herieXL@}s>p+xXDzKBgDlZC~Ss
zpkp|t))Clxtp4?-ZR&QMeycf+@5BaTkDV(I;C$CK+%{bT2^(+1lQ!)S7<-SP?xW-V
zhL+8F)(OQ1e%;#A1l`y1)n&}R-{4t(wNq>^?9@)jM-P<rJ^r7-X@zUQ$3Ki;*W+#g
zcWC~OWcW?BX6*~4MRLOC{9m}>z;|lDL71*BzB+L;wltfy7BHFXvoo-isr~&lAzgm3
z0k0g!)=zhTz<w9q9`QPL7dNB5={SVy@8?x{7Bw_0$+2lDGLD5j9E8xGU5#Y1osif=
z9)fYgUHcofQi2Ie(5oHxBslxY^=Xg|H%31->BX^%1n+oc&wsXfg9ZEgi7L)oZnb-H
z3eQ|U^`J2aN5Z9QH9iZ6U14$FUXeXvxJ+0>WkYTZV-2-2p|N|N6l+f!Ry=uxS@Msx
z&^e)^w~Mboh?6$jaw>{F0wg1p!nZbkDf{sfoVPK8HB_|MwqsZr^!LZnlSCEobwscs
z%{+do-T!+oXD@!<&bgXcEIbE=;#G$yW#PmwPXImP<pCZn%ZuXU66id{K@B|yz%Taf
zf*(opbBGvj1fs1)Ncd)Su5cUG4EEbQ&WSRI7$tbBJRZ;d7NP3H93j|m0C9vlxyqT`
ziZ7M(fJ9}?2SR3wFb6*ug6hiCo<{<Ry3!(c1)rm0A&T`Z*cWR5gl5=uVaw>=bU}~r
z>!TudV8k{gf#stj@_qA#6?s;C@ml-)?~uPjQn(k>uPYtt8(AC~qf4+~lA-7KWYK^1
zzpKvDQ@IIN3mvtQP^Cu+J;W>r`&C)AD(7<_CA5hWP>0tbn)yDjc>jdcAADOFZArXJ
zr*A+G2Nw+GW%$?+r>Vj5Vv7NrQt?U%Ux5&x6TmkM(9HU2t&iYXDF-F(Q>1}dTTn5G
z<qck~4_X7fcq<Umym(myKO@>hh}PUszS7Bg6s8;cH*8ml7NEb4#C_Bi2;2Msjo1yp
zMFtACXN8h>Fklq|U|8_-thN+i=N2C?&^HOK?Mkknz=enOPvp8+nF{tIYXzK{v``lt
zN3oDcvLWvh$iqsBe*~=@AI=k24j)w$QCEt+SnVa~-+msNM)=Pdi743&ZULJxeA=JD
z!;sY?;<JK7bFe0`yQ;Hye<Vk)7oXBYNUdPEhRouchABvWU;CUVL(oysiDo91Y+l6z
z8Y>dRcW{kt@V{}0mfNGR$@SOKZAkw}uG=6I!Yf1+ldX&S>S2nWiEmu$rF$;Kun>~X
z(l|$y20AH=ErG3bGorMg!=e}>ioytE#p>noh>&TPgCG6-6c#(I0EAfc`3J6BfDDDK
zWWJJQW}kNZ3!J}a_7y7X+Wb3aKi$w<G4w7?jAkEfR%zVz5ZOPB?6YB~7`7=DIMTt^
z66Gf7azFfyULG9c*xCj`Zm;Drl;zO{3EESLqL;_**f6C)5!Qh)y*xrfW_NiE)xM?n
z$T%3Ss(uMoNrpmJaz9IXh?j4*z3{V?Am6+NcQTF_V$I7g*XLoAwz=Qq=|(3L{nhzv
zvg+L}R_w2t{kyi&B1B*4vKlS9j$4DmsWLv!G!i)MC_(s7xW#fm`90iR!-(Zx{XN{l
z6S#lzJ={x9;J)X3xNkdwd(roB7oNaf_C4I!pTM2_J={}H;J)U2xUW2cJM(+ES*g)a
zSjqQ=g#me1e8XMawF(2>+n`t&{`oDNy}|G#8FqqVVc2xc!mykSYe2CuJpG-8VJ;a~
zgJNMAWKCiU86vSdoNQ%CAwyRz!)_}>0vWc%D%osh`0Q6;_**Q)@2t7lL55IlE*`Wp
z{D}+!P{P-;@ZlC<M3o*wKJ~|MsKPiS2wjYWR7@AE+bI^OyWJAGj$OR3F7*z=!uu(N
z9!O6+UZ<~gTZL`?9%1Z_;YYfK#ahKW@nu(w!=!(_uP@E+g0OaY?M>mo;7O?0%ZOIT
zSz~N3p94x?LbG&H;Xl#5L3E2Ly5~4i_tXh;>B{vbw*q19*_61yTVkxGt!q6&;$z<<
z@qS%S_;bK5f5PyiPr1rax*IofG}^mTG>ULgFYoYo1}sGcOBZ;aa?s26AXuCHAX*QL
z)57p;gTVo)T|fr<xE2Q5DoSXegnneW#mb=jkZlaqs8Vm<McrkTVX6Aq5Ae~r8#X@1
zSH?E8NLi!sAcTj%z*AH;J>{W@td)zDdyeXLJev3KC#$<n&x(Iup$&%Q_C-C?+(3nn
z{(`yfF>-#N58it=^<1o{#5F=<^y3!qJfIinYUaW+I{#18Si>|tOanG$E%JpaYxI0b
z^m`YW{<eZqe1Ql5lH%;Ff(s4*0<+Z3r(m|Nr9#HE(8!PI!(aw4CV})Hak7U665H<U
z{{m@Q>7QfI4Q@)Rc8CgyY1+?R{68MdId=AtL%gQF<rF;e=+pb!UgUlYoC^be1!%W6
zm{bl<+I$w3`6B}GsXCzy_TxU#UT5Ei8sO>nXlf>7TfrT?ViT7oHgYXOyYUbdxoaCU
z8=cX`PUyyTcXxcF3KQStcKWE@pONg1A=*^r`*=EtJ*NdpXfNN7_WSkmnDME^%J4WD
zt^=igys&mX3sQX4G$#4!C+e4hsO^3ueoP7YZ;pvO`oy>wMpa*WX|%%+hE^=q#VF?P
zaJ>7053*>TEA+rz1d{1`Kx03U-}79Jd<$n+J5DJ)fz+y3`U-@Ym7ZX!^iLtsQt59~
z{$1NB$5^n5w%4BPH$Q}n#U?aB975F^?u$eY;v0~7iG~1b9h7hAcRT*=%0YaHh{v0n
zR$My+=KN4_^p=az$kH0L7w&~B!D`2r2|B(OI6k4UC0j=y0lK9DQ_IbqlW4nus@nwt
z3nc529;yU~h|i3!k735EX_SfAYO%T8k%sj%pqB8f#pE~q2B`QxR<wU|Nrqp?I+mq=
zla=y`xAD7vSa(@c;|;+Sb@yk{1gbmuQLEdKJQrEm|B;0*(q<bpF<J*llQhL3v2MVI
zg`9|Xxk0-dG<+D%sf03|SZrMSd#H`s_CK-U=1al;I*tKgf!9BV``}oz3v|WT<amQ)
zk|^UKGJuWjsfdqr13K+}orY6_-L#!L?I&64*X?*UqNxM&+fOG)yNtV|{UqFnw#OqJ
z|4pmA%c97zKGywx6LAtXv>-uW8o+`;^W>pWABR$Nld--$IT?hpBw)IweS+HZ(pY#$
z;Ec04ojBR$4NgzSe}zkpecuUR_4z=nd7fryvP*xpE%7d|Iz81mb?a`P?4(%k+Nba%
zC$C%WYhc3twwkk*;OJ)iF9jFQBcyMw(#+LWd?2xD%N#VO&3@cAVSzx}_!~TFhn9=g
zcE0y6CRa_w%`MXBfUt~Np0VqhKhz)o-SBMvy<VZ5lgjO*a@Qe!z5K1=DGa}KCBiT7
zHT+_RKgRGO(eQs)AN><LZvA_O=pWqwu$H%Fsadp-?)6>vYkXYJ$3v{gp@|R8KX^w1
zK1G(5f|m6O4~=?pW)m+s;*1M63%c+s&Gz}%=x>E%yfa=@{0HOWQ1dk*(Yv$KJMn=Z
z4u<#?^*-%kP$@;Pw1GiZBECf;)<QG!>Q+%g`$BIB-x>*KV${Ncn4{Rg;1}cZVo^0d
ztNJxwHuI-Jn06`aSl7`|D*@kZ!ogguwq9zaHV{1QF6S$1mv}wzIL<NR%<ExqN+-U<
zfOE;}Q7sH5_&*Q+i4QHtdE@6{ydH@j8f}8ccV)CR2&27GBW@fhJ?y>~XQtgVeWCN)
zoKE};$k4@IfHQn4pP)n<zeCUeJVCowMIk_;v(5E(yqXE!JdhC?{7>{|ywcQOvtP9K
z+2;RhK#fK1`JpEC9R9Cc^mjbMieec5o(uh!mNxpsW3BZ!qolOhUs`^PbIu~C<SZ?B
z7M4~O75Gal%B4weug_V{@Ms7gZjweyvz=HhbQUeF1gBG)T;X@(k}?gvY*ax>QNi4z
zsyP+;RfShx=`0(SdrMVm;l)>F==6!CCs&vHi&CUfqb?~copVW9zTaP3P;|*;@Z}mv
zZe>x?+{%i%MWafj`4y#wP6mz}N9qj3O|Gt(bF1@$k<NmO@@l`+r%RYHAJnTzuJ(g@
zJ~XVTnowHq2Q9^U6GNfAE~#`zL4H|&)r23$=4VvQt1K&8IO!@U{vi9`6(f`>(1lDW
zF006gKH_h-R9$*|QAM%HP0B^ns!Qh;I#Ff@mHMVgdM?I|%gxFyEGo{gDf8#%S5_94
z7n;Q<ij7fYi)wOZQPsQ}6xDnoHd<Cu6y?G~(GxXFC7RdSQgU)~4OlK8lj~2pAh)70
z-=BX|>g<#YQWs82fs(0a^1asHocwCmoL;zQ?fh_2$5u*I8KbzyjWZ-A>ymh%h=iOM
z{a;n?Y{@8M^b*TedB00wl|#QDw&v*Jf*H4IW2oX-Q_oD3ZbCi$(zJ~HDt}RRX?{6c
z+Ps?b(t>=nTX45NlP6dmM*2E`S}Fz3EU7BWFVsbvd@h5}l)wIOdi)N20M)!`F;We_
zxCEV=W8x-L59oNnB`B3uI;=I|3Inb-VDI@)8%=$qd9;*5wyB?yzFE#@@>I|IBfYk6
zl)JqA$}7`E<H{{9FZGWqD3EkJa6-3d%HziI?$P_s4D^#^H0zNtcm5RV;)~~1mg0U3
zH~`Q`CH(xNY^z7HA10?TEsF`Fu<ZP*>LOqMqKX=SazROcmGh#DguvbmPj56PTv#-x
z=9Y`AOK&O9FRSjx{e=)JW#re~QsRdpV5sx_!jmPZb6(LrXMSN}RZ(@d^Zc^1g`^o*
zK}BJa;L5G6swe<MMOF1pg|n$S>`t^v=@!;X$t|xaFJk%W1r@C%YeO72u2*vM359(D
zj8(U(?lsc5aWi1KW>#ENR4{*1RmHqy5y33H7(~?1zbkKI(JiIrF`6+a)z=i=R#Q~%
z&*eZc(ixLX4Dvrp6DeSL;Fx3?bWTW?{+{iu`0-=V0oes=?}|kD4ef@yNF*7JybTal
zD&^I~Nx<etEnqgnEANR!S^%9`)ReG(;KG&-|Ko5a_OEgXv478huABEL_=x{G5*dnV
zqianhk_y=HcqGFAP~W;X5?KwndR-*a3YfP(66pe5@mKKmL%gRW5$;pC@S4slz|>9P
z18jXU5=p?zf2~cCh>H-Pn&5x5?|2FF02^M8MD_rtz8Z-*_+r{F=nvQdm<QPU2J!{C
zVoxNp9<cH~=m}U0*hV^F7vO3@2g+eRU=E<84fz6e0<Hl}1#AFZ4X6P&07@r;?_-n)
zV8ek(q>}hS@B>x?wgA=wwgIjH>;ha3=s-EQ0wx1yV}-*7d3jvv0ABkw(gj@c4fp|9
z13IwcvK}xQumR8o*b0~f*alb$C>@SOmH|2eR|C2LHvnb>wgTn>YJio166+UmDBudf
zRKV4M4M?xG6Xgcn)rEWj^1}YFxNNsyBiR-@Y-c47NLXPT;2?cCe#wDI<RYC#+<WJj
zq$GpCYYFmzwjvF5Od2?JihRm~gj(t9VdF0vopc@rP$sYV+p~Ql#R2iIGhI@=o~|T~
zK^SGNM!HW8!YakK-@7FDK*z*^LnjV&P8^s#abT)@;MmCnTL%ONEbkxccdz}v_<Q0+
zyjh4h^ub7EFxrfL+aG$1=NXtC6HVx|67i(pL?R>L(Aqonc#{V<To^##pgzyy=Xf|0
zIgWhUeShp-?h0FsV8;6#zdh(np90!f_K4>mnCcnm3iMxY<^W8%7%vHWtXLh1oXqr|
z?GaDUzh~g;*z^XA*eq9TJ?iz#GS5a|-bEgirI|0T-6x=L7bBncq(^(tS6)oEgnru)
zFB^x@Y>b!PN4y^8F%n5XieEBp$2P>1CR*wgWx;|7+3$_Nk72mj2L6Fx3v7(|j$@SH
zJusR4{gw-Eme(=-@^J9@T<FzHTXCb9txW|?5+)lcYw+j5cD+M_wT(gk<bmsL*2-Zz
z+!;)GG7{kyPYl0VC$+J4La~hZHS)9?HgO;8v+`dnFUA`Q{+wqckw+Ns;eS0|A^02q
z9*KMd8*6XrBVLbsWqEs`=du?gks~qXi1H3}w^8b~4f6B0=wpFB`A@yI8x&uBOrbO0
zqmZw_*1sAfKia2zttZACV}~u=23v@D_Qk!XcS65ryur|O=nm8uo_npY9=3pNgYokr
z-g?CAAi=muXSptGjzr#sKIX-A7K5JsN+j|;>DG3nw~Jo$MS0I6-k#P-<SmOl(q936
z^_!6hK=dO0V-QyLLO%xjieBi0A$M6X^pT*iSR9EsDINL}P`(HBg!V{;i(lq71`o49
z&xV}q**E`<{H;X1g>Oe9{Y_zjOlK|Vl|AT5pbrOQ3Fswwwr@Z^qtCks4)vh7$KT7e
zaWVeqh`%0u7#@T^raha5_!~f<hVrmWkGXo$6=wa!cDWMqTHlRC9>o}HpV=dxumx>#
zR$IViSuU?2p7JSHexRHEg+B{@qvMo_S1aOK?HX<ITJ+02EfTqgf>55q)>!iQ3c^<-
zoZZB{nD384U)2l!80af{p>rHu20G7)N}hpE0;WF_bjn{q0@6nry9YW097cs*^da04
zjzor0fvDX@Kc(I6VW+9rV({nP9*I;L@(5^OKZoz7K~5a|t$mH9d}3`l`B|RZAZHKS
z-z7}<-)LvY5O2l)6U%Eb(qGmKeI)3$z0f_NFYJXr3-rof=zh>kKp(~oKwkou$8v+e
zC;d@Feoy)fp!cTFKMek!?SSRln?82<n$*YIf2q%K`k<iqv?pUg&*>#!KF}SDV(o(m
za*n~rEkXSh;hF8?zPQYRL+_2x9Ow+$GY2Lw@23n*y{Er2aBN^e)<9R;z_DmAsA=K1
z!Fz)bm!R7j_@XNy$7u4DCay^Q(YtD;_)d}ES4}D(&cJUfyZguZW#Csq<U%kK5!bCi
zlJS(y5YB|qDWnBNIINutm_y{JK&G1sueQTa1u?>^dkV;3&7#glliCF0IgGOo<$W4p
z29Z1z4!&R~Wirl1C=p~vI!NR;AZH=fVMJbnTpU7|juN>}MhM_vDBw0AUz_O^fkH(^
zej@QH@fQM-cJaoG4|%9I8`P0TLBUVWh>`fWYEf3xV6sG+{sB-r2R}%c{!GM;AG7^2
zk!ko*(QomyeP$qsh<t_HIQ+z=OOM1iLA3j)AR)8T5w;upq!hh_$Qaz1FHQ<24{lEY
zit7Ln=10Vd#6K*>4+{ciNjyu2NAa7;2VRl*{merNpV#xoiw}wT(OHmQz^?{Baiu{8
z$`<X02jmt3`8C;=;3kY2`Mv=XWjhLlnu;r0zNa!0FFr6=(?}SO+Z0lH>AI)miDygp
zuwd5Ub_2tBMP;4DAYOc65wByR@GD%enW^;68!&l)-pxBZ{w9j6=ZzO1SPlD`sSti1
z{KT~z_Zmo0?0A@j8_$x6s|~zl7jZtplcfw=(P}0jj53EX<^lY6;J1s)@IVq<4SiK<
zCbYU5u$2h^`vyaV8Cea4FC~gA68~H+>T%2#V3f);aZ)pW(W;vO3iX-95I1g@0E)|K
zYkcOAAuc{JrSZgA1uRx^W#G=9m`Chd{C>m-UXgg_i%;3S@#4eZ@nh?d1|ZQwKyl3l
z!oDIx?gK*I9s-p3g}hCC;1!AAcQ^9&K^zS>b;ulm8^sdICc^z&hTcSk$6bct$F&Km
z6TdU?%VdcRgh;A7naHI;P6zZdouA@H0Yxw#67^!pbh14Qgw^JvG)}o_%AX*@)<|hw
z^N>zK?iliG#yJNF<NSe$3kVB~wJ6QPjq1!}I@LzV<79gVNHiV(Ge}C~vWfII5T^4T
zk+aYS7;=ztE&;+gUlYj&!gS~!OJ%q*of@XY=FN~}WNR`I{s&U&O(3sPEH9%C&^2KS
zJPIoO0j3=y@;H7(cw9_s#E-2*%pzzY?;w?z-e*VSiTlg%ku(>-U-99;*MCdkza{YB
z68LWk{C~RyTxYq2f(>Wtu+4yZ!*twfgbOCzOxxcRzDv)>6289&K7d0$F1nh$a)!7B
z<jBIy6LJZ7rnqD1c`{Eton_!o19I%<r5WKI#CUPU<Mqw)NCn}lI2VENGzc9hxd<dW
z_Pn~{Tm(^9XM_=OilAa0%YbwbtQW>n{aT+VAckylwHW#3fXd4pYdJRaS~18aph;it
z&~a1m4iO5k)_`XEI6x`#(;(~oI4>!1BRQ$Vz$X~tM)dzU=rYQ`XF|U;>h&=LK4-uk
z27JqapBnJ60sEh!=WmDsFEHTc2ApKT>kL?I!1)He$AG^y;9~}S&VV}%_?7`bHQ-?b
z_8)4bZ@>!-c)0;58Spv-78`KB0q-&3FAeyZ0iQGA4g<brz)uZ$*ns_wemBH`7Z~tz
z15PsFbp|Xp;Cut#W58b;@G%2EXTYBIG&v(<oHO~_IW^_}8t0f%V@F+jaawwfK+{rm
zf@Fjil;z`G(fp#SYMdr=jvjT{sML$c=&YA6@mJMU`xjm6O!nvBQhmuanZ8W-jLh88
zm!)1(T#Ehf6lXHej}$r))91S`OOKv{$#wOjdH(!4fc`2SmY7f2xh<-yltz_T_=`rl
zCwec2Xal>Yyk^v#n$ohui?PQ)idad0b%`{pa8Wtr>Cj)LhZqXSAh`&uDk{ro1mmf)
z%rA}N(F19ezi1)u(7P02;*^Xus;DHlxGH~MQEo{g;+hzQ78K+bEi5Ri#7=bnoU$Ta
zMs9vpRsJGfhWUJJ0U{#3{CTAXh*yCma2qAk<w+cA)ST*SX;c9=w~NaCJ?rUzil2pu
zw9%%08XM;V22b4}n08YTaWn0?iT?x-v{h#OnQ<;)#^eUjNyKuQ_Q)iq;lUc;ALkN5
zenViAz`Ace<9iT<ZPW@g{Z(-;!Mw^K%o13)eK+l>K^6-zwB=^}t~i$n>M{sJ4SP?(
zyqob&yL%mIwDo3ubNp}^2HG4iz#^`m<##ivwAE&O(>_m-b^g0`5FMXPJ`-{xMcZ%U
zru}a);<p*`sXxok)ZZXWw;OabzB%5k<}?zQIo=q<ax}h~|2ptct{LAPzg7%lY>dwg
z#Krg=)6MkF`1gXwGBJudI^O4S!j5a04x)EMTohOFWQ|{Ij1#p+d{=ZJ)}t8N60>Ad
zR`!T*`Zo=RgK>AY9?!g+@lE{KJ>r}Gi6^9UnB%m`-&6n9MtrmUvO_Mxnr*~?%5($-
zRiE7(j{>a0o%uK8OAopPU3yRlTyeEtJ=1><f!6qAf8`SFV~yGQ^5`0mF2$@r9m9Sm
z9<1?sEP&Ux-wPn+nH8&}qOBN@0=$m<K>W;VGX1%hCv`5T4x;y->A#N<Hgz+;>F;Sq
z{GR+h@%@NP{mu9thQHGBFUCKBxXhv%-}HC7jQD27n8jeGZ?>bZ9_e?j%L4^|=Q;tR
z{YA&R`;RH-I|NwumkfW?T)17&K+9#uGhu&(TjQ_)Yo6d<|5rVdsi!HzjA!DBh)kN9
z{{|y|gAw0#s>!KCXDliWGSW}&K{w+s?Lnumyi7<lW2P+vjN8>u>kneD%o1S6GfBjI
zbP*Bcd|u~JPS@et08{WY^KaTO_V3=(FKN}|{iw$P+B1EftM_|6ayd56bwXe~G&aXs
zYhXMyF~`qXx;;A2iKX|8&TnGr{iE}sSo#3T9G_z836eSP#L`ca%<&|aesXl46H7lO
zYX4*DiIQp8W9g?#roE1(4~)(eW9hPF+Sgb*TDfUAEdlY;Aj!0c7BWitG}t$nl_Z|y
zu^4LFFKb{tmQPK)6iYu{GVM()eTZb*ky!c}unT!sl6a2iEY$4xme6?Y@0k7BLWcgs
zB(tB!(&-JxUTm*+>8nUA)f|UTEQj+1Bj|r{#Y7on(0R<pdQCRyJVs-^ZZzosuePg?
zk)){NeHS8ugxrT9aKvQ6e@N7Id-ijp#@^fOy9~FtH%<@7aS>`~XL|SgZf1L!p4r<q
z{BZ^&3MQV$PeQ`RpwSrhKtvED7eh2~3E>}&0!HI^@FN$E2_!21-m9vf?&|J4HcAD0
zr+@wWb@i+FUcIWWdPmD-lBJ_?THcZ@r}<>Jb>Yhn|I&xg@Q>XVEl)|7uPB_BjU~(X
z6#iKQe^%j-8Td;Ir)65n@>hk^GOT1FL2kE(dOM;nzI_|{v_00tk1sNUQ!cbQHc$S#
zp0pezS;~ClZi|*#B+EgC(=v)=(e<Qd63KE0=&7E1o~NfSZUCox(pFI_+7EtS;k1u5
ziR<?|r*N(Y2<R(H-h0XZX17Jl5|V}fnH{Q6Jl~`)YW}Yp{4ax3&*Q%X_%?jZG3|~%
zG7|r@d(zLf75-}jr;pb}Z_axiWBz-rO`{zH#<PCljN;=M^Jfn-eb%28z1LGrzsovj
z_@!qV&%Q5-FHb%{W%{f?`7?0(*kk&!e<|Gbdppo6Zo^kM(~n&Td?d#&y#qM3HdAQL
z&~C{2h5H=7ruBzDVckUdF^SXqXDZ(>;6oo#iH~!?!*rU@s|EBQ1i!rc(7Nh8{BGbk
zA`;d6CLlpBUr@N-Zz1&?U!$21>Z9U)79;MUf6OJ1KdnWh`kyxR@R!Vg=+95f$6EC7
z%;fQzEx_+!9Qv%}qsLue1J3XDX@@z}=UFP}fsgR7^x?$<{BO)ZP0MF*A^v+T=g~zz
zttyhe!1L;VqyVS&>3Q_`6yT3BewEc#?@MD4Dz6mKUv@e1&&Cy30w+B@v)|$Sf5At5
zH)i^)tW862A7ES&CH_1Od>7=fspLRuC#d|afc~Grsa?*VahPs~kC>-MfJ%CD#?bRy
zfm1!5gOZ-t5#XLD75(Nw;uKF)`HjNQebnJ|T1!WTJ{c3Au_cF2t>^e?S)aNhU;bYQ
z&URXJn3C4MQu!0%+u?6N;PAPgPk&v}>wQKrPcGwEN<Q5ghtK~YVe4bSsol?BD{%!|
z-%#`~%u4(jRsZdHP@bOL4tzx4ocKEcPIA`!r?mZk5jefq$h#$fTGL16Jn&th*KsDy
zlgo?1*?vFa@Hwr$qp|~&BE+XV;qW=FWv6lycwRm415WkP`@3{|Jybyd2=Ec+m_D2b
zzB9)U|D5??(00GWH)ua_GoZZs)EP%SbDyMFuVsA%IK7uMF7bD&3*T0_-an?>_g9L~
zqI$08b2SVz@3|$$5ud2%>Uy38PX6TFcNV$+NBM{aA&UO2QP1a?9(JVso3^9h1E>01
zO<Dg7iqGs0*^X7Co?V6S+>rFNW|hjrz`1=7IDF1))bVRO6p8Q)cS@XBO#r`5;d)=)
zYf2FJ0H=1>`~9wFZrC0HJWoDP0pGR#`b?bvlLGob5<TQL^4KDC*8b-|1@zZI@u>cb
zWryi$PXv{Rfalfc2f(R5Pdu{7bo%gY0sYI2qksRT!=%2t;V%l`yjSAyQ8+C+Ci|j9
z8dOg45i4R)8MYVIzSk=J7T`qRRrFfl0>%|l;?D!XxgC2BqgrpDQS^GBsJ`X5ieB%V
z)buyJRr1-~C;2?7ZfFCi`iv+)tlWupFXNi)@bB*m@YfXo?)xPF=hZDcpfFVb>c=Gx
z_2zQ3!p|*AoYwMF833pH>{RlpN!a=laFVCqx2f&rTZ+%7vI|}-jC;-lC%rnW;$iI%
zUt#`=Y2wdi*Qn<jau@-=lkEDD#RN4Lr!jpB;*$fwIj;Dwq}P7n?gIJ;fwMll4&Q&H
zlEaUIk1&!xyi~yFAB-bTGxX|e%tw<vPb>LoeYg`iJ@-t<;RonF2cH8@_1`q=e_ru9
zr{qKNF_rC5Y?6=l^dg_@!{xw<kM+33Ppb>Fz^OiGl|J(x0$jV3aYdB)69K1szHr=O
zl=oEN*WU~1_gtsssq{_H!yN@q`an5hxFDXOvH=_>lPM1s;Lj^Qrxl+Ob?3he=--C<
z{5<{>z^NZJ_3FKf-nrgM)#*dPNe<5YB~E*^sQgIb7n%;A>%AziGat0Cp%2%iN%HFR
zF2<oJMx4+9PR~83;-zisxnC{d^TPuCWySxD(GOm6Lw<d3D!}Unc((w*hjGM>11X>5
zEG5hOroy{7NSyW-QhAi|?0n!8z)22chaIM)Jy=|5H?wGR`U>9MJU5s6?mA}SOLhvu
zZgLZu%tBLHXgUj(Gf*juWFI)5A)3e_**ayPU$(@7`NMl>=Y@y{OQIqSVTfa1e||=Y
zAmk6&d{ytFH-(UW;h9|S`hIatrYM-Vr<ku!Sb<IUt~I<kow`T~B-}T;Pd)eD5?kEk
z4eiZdKk5&VJf&#~am!py%vZd+5J>RQ!yfSns|7^ABbsgOe~+vjNQhP_R@&iGqb*uU
z#}J9eV4Z5od%atuQ|0LrIOh-vS_~jR<O_V9?`>FeF+gk3>1^PZ#2KeANRbM$zczcw
z74E`5R94np{Z_4dc>jLSt&95X-g%cONcJ&{RvL)<$Sh%reYY>n9;(b)v~oZw3|dec
z69TCY!c(H{udeiu3%}P3d)0$Y$2l-x**kais2Creh}*3oPlY{sfTiROI!8F+id$_3
z@lCiY^CdLeZNFVxI56S)eyig3?eSh?wH0=d)MO3GUtD*sRu@*chlL!=>iMEHS{j{<
zSrt5gdN0;b1b)w5BT6uy!QzpiiT4Dr6&(FWr_uy&Q>F&d?=^ycRB5_i0|$4`c54T{
zi0UIMVI&SCvq}fq6fz?4!gk~aX&4z9BEN6ewq8+rI?TKm;#hnsmBK>^WWi~|UkzAw
ztffZtB&ylemC$m~EmuE+>P{x#a%r$!{EF#1llj|Iy5W}^@RN<Gzm_KIhEcFC@Oq6l
zGEkT=c+s>arBXV>gU*iN*1Z0d%w<yvAp5{2k5Kn6$_yl8r4z17r6q&7WlhwS{m!);
zQ3T`d4*HQ<#d;lYxg|wN3qTxewX23u{Pn<WYe6;W^~G`{Xq#17IOz2!oBfTh4?Tl+
z9TKO4mcPpO>ZVzb+V-Wi+%9plCcjl{te%7-2DQMmg|*@%by5c@gj$UaqXN7nN&8&8
zG{?Q6XopC9R9W!Ew92)E-?q>PQI2!&xVo&>h+1Xdsnmt*dSwv|Exf=ibKJr*-{kdY
z0vH;t@S3Q=cfkVF(ts$u{zMbb$Y>L3CTs<w(+DzpSwA9`DUonTzkfj`OLkS~OPLG9
z9%UpU`MN=_vMRHlnG3rcqKgG?KC)DCj-~&HK!b!#?J{}IxlZeFm$GrW>*%l|w_0tY
zt;k))+QM{FrK&*{1at+SeF9m8mcw2Lm-?cGyL-XXfLK*JZnF*dTN(q}L@F+^1eG|6
zjtYrWe7Z1&sDZpsANq<u0W`f0eYrE=nV50sJCpd|F1a|4&y?x6P1@3d2d^Dm@WzBa
z)@@(`oFMHd8*Z*QL<1}3G6tmLg}B8^Hp#!T+Va=yq3l<vVW981O@jU#%M4M+c3YrF
z?_l*{-;af3Bpxb_CsUB^Lp~THC-EjEKPByCOATt;N6nj224d>^z8qagb7{XQ6)eXx
zz5cX}Jk~4A$}ZtTy(t)@J)KDm=62YG+EHaWr{5z-Y>&&{BUIgEeN5DC;esKwYxpb}
zyJ3QrI%Of#;Zpt39K}y{UduS_-3pA2{Jk=vbk8_f%IO@u34#ACQ8^6f-wIa604qXc
z4h4_M<rLp!YcEDB1(wj2Ewyvj69&V+T3{{d5@H0y3&%ZNtl}=hLEr4!Nv`=)6{(<2
z%CF$<?Q**7<zz@XqOGn4-lW#hXaXH-*B)anf;xm76cABtWHh~MyMkwz;~>zjoft~o
zjQ;E?3jbi^i2U}+z>mW=PLyWYt392xSML7E4_vzEnb_M48!ZZJZVP+umK)%OsY+fj
zg=i-1`9_Cx=@`>S)S66N<&p?(DMIu*kRDvzdSfYAvtf_qm7<89AN=of2uUKA#-e0l
zkB-N(c>{tZeb=3uAsjZQH^YrL+cJ>m7AdM@EZ|tSvH^8->*Q$EW#u?-XkRRb{~<ET
z^Sd0I=r|*Su8GLMYi7k?^?PjYWdxL~18Oa5C@2F$T!Dm=xEl5<)e}Q!d;&&rLOR>E
zx`0cMs~ns1RLDJ5j)_sjsn9l^t`BO6{+qpd1B#d(Yza73J^F*nMhK5URA7=6Cm~&2
z+ES4;4!t+}HPaPI=P{#mXl}vxC1XTHi-|5mcSk&a8{_Uj9W3TXQeM9tch_d{Ct6V`
zj>AJ>eSEXA6q%#J;b4hIzHy^WqkljWz*bUwOFYjIVa&^U^b}V_6V);hGELF1lK+X@
zD%~+9dMs{pan~U5qoxnNY57q+;7pHX%kZQK%hwt(q*TL<+L%1a?*Z+Jq^tAcvQ*I2
z9rS#$7WDdq1_l^8Uc_(<u%<+sRd=Y(!y%W_0us$$#Ngu-!!WBJro!QQ8Y^)4m2E`y
z&Ctp5p;6fE0aTX7LzwR&xyqrYiZb8}X|R+xP|b}=1fmiONviZ5J-m{05~F1cgVRp9
zX7;0IJh`u=TS>=QN)W>fGulhcl`0FDa1TL7pyVLZ-l4=ggeVE=9ks^E2%R-eO%B99
zL^*P!dJT_yVGQtMg-Og9%+Rb2l7B|>-e7_(+EBLPjqH<EyU-|(1R}6|B%TOb@WRRE
z_8>Z*WnuJfG+`74m+amWQ7?%(Wdu~Oj1UWASfLVeij)3^XyT1!Z0`lV$gLSuF<MR~
zmXjE)=9ucmx->cPpaEdY1oB-w-e60^B!s0DlAF@Bg;?ucTAZ@l?91X2VP&<iCfjOW
zl(kL^12!2vzygOeMVJXIm?<g<_T=ncB6TVam4c`{SZ(S~M^8>AoAu3zOK7|^ev$DG
zw3s}v@-ZzJEf>~BO84M3$R%O&3v(l_3Js?tGTsvXS33xIL+#p=(~^xQ<x!1818mq@
ziEXN6bH@_T(+Nm*88eW~p5h8Qt4fn=7@3C6JQFbHKWP+4Gd6~~s6NA!Q9A6;6F&0O
zZeXy`g#m}X#)@yW9eOq<$<9pjFnozJq+-O}Zyrx~C_e8iLwK~lPSYAuh@E%=p0qZW
zrqexGDOg2N1#4*yn*A-r5|}?4jxbwrb7mOl&9U3tvW-<vtfAF0iL8Br>>Ej}c#{t!
z>c^GUQ=ODf#KZ-mF3k^T{8fBY?8TuGJPA&{C9>P<ISjZdIB_{?6OZ!auK3o~Q_w>U
zzoJ2hIx>8L8Xp}a=6E7c)cLs+1{d>V$#K{+vkJ#CLF#7b^l27A>3WMtVx)o;Qt^a%
zN=QR3j7~MW5QrY-sxJiSi<{lV{G*-tXu$qB_FL7)2J)Y`l2chUlV|aRW3k#}!zs2c
NX;yZGF2@6b^<S|_o~r-=

diff --git a/tests/Grid_gamma b/tests/Grid_gamma
deleted file mode 100755
index 641d5c8872bf525f0e5288531be47faab67cb9e5..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 72670
zcmeFa3w%`7wLgCHf~d$DD#fU%106LH#6(dOiaLP+XK(^3M2Q+rLJ~->A(<u<jE_Pz
zp>jH;O?&ArwzSP{?Y&;5m22%SphChc0R@9kd?4|Go*^Rf1<@k&`+oP?XXZ>^V*Bg=
z|GB@92j;A^)?Rz<wfA0Y?Z<g!`3&#O*q9i_I%1XU6k<K+xCO~PD@yM>0+qCM<ut{u
zT&RpyhJliRpIf95npYO=m7*<}DEZ<5<MB(6H3{jlZgFA3u_hTMn1`hyA<K&#Zh2|K
zo>xuOf{Nl41m>fztp@}OSST(mxQ5E$Xmyy-%18B>uUF>lmH8~FNc}Bn)sN!%%a!T5
zv2I|*-2_UOI7gr?UdcJ+FX6R&m&xGX<rWtf94qBn(8~8c<fERadRWz*C(9eGUEH!f
z3tIhfc}4lct0!MxQF2*Dd1Y<gWp(LSUv~B6Ni|iIt`PO6eDcnkbE8m+VyIsde$;t1
zezXCmzg#kU`729vhTrzm<!k1wy5Udn<+-0=+GPB!e$$H_g+*;yF{E9MUj}|1_nrRy
zuZy4F@#7_-jt?Ez|EB3?_eXEu`ZH(pjF-Hl*Nm91lqLMfZ^rsI+~v9Bf51NAaSxRF
z)&MEPVPpf*??*)jqE8xxfBqo!1XOGwIczuQ0Qy`w2>sJR^#AWc=;sfj&*y{iuSdB9
z={a=}IS&pZ=gvX+zdZ>3%0cQ|Itc&BLFjV_k+W=&dX){*zh4Z}U+)h>_Y9&>(;)m?
z2Whuo55j-`AoSY?DR<u>^*ufa{UjW4pnmTdM9=a;+WF2w+H>L{bbS#04-KNvn}g_o
z<{<LN4<i55LHPeNNV%^M(he^V(l4V1;lE=L{%Z#*_o_kk-wXZ6Dd#HgcM8!WijpVJ
ziZT!7lHU4mp`ciFFXWS6__j%}7paQ!q@+6!nRJVP80e$Gue@i{os$22DZfj~$&vKU
zvR*x7+>&B~Wl~PdJB5Pa03VNE<41dIIopkM<X9>7sh9fH%cNw`$={shmJ|svF4g!;
ziYH9qSah{gP_TGORb@eqzo^<@P@ojdURp4(ba8o&zqERKMNv&nX^m3g$;~b(DXo^|
z-0bNURh6Z=MGGrR&9s423Vi;u>e8Z;f;&p@Dk#R4AMCfxnO`#*CHT@(3n;Y4UtM+A
zjCnWAFY*-6fHK8(MFqav>e7Ox<<<V$q6+4?VZQ&WV*g#fQm~|5iEMY4my}i(lvkGf
zXDk&W%P0FvOYacKe1F<C1qCR%>du0ND7>hmqN>>Bs=8`XMQu%)#dfuwt+=ShPcCE!
zy$WiT7A-*$HKo;lPcf)z#brg+1^()ya(@l<m@yBuD;E~klowmvp4@`m)PfoFuP(^B
zVg8i`LUl4#7gd&2EkRwEmKT>^1NDXCs4`j><z=9ymy}ncbxbY$=Ey*f#ijnGMHRBL
z{`B&y8q}?(bg~Cckd<c9uSQLaDoVqIn(1jXa&9ONH!fSU=9=QFCBCBSQl^Nu>suZ*
zTm%&@sTpP^QRaZ{a8+?dX;JkIRP>4h5A?#%Z}oKf)#ZyTt5BBgrRtg)^U7tR>84mT
zba~}sv$+S7FYIClD#G?+Yl=eV`bsO8RF%-4a&MXgQ-a;p6)h}Zn#RsArrj3MV9?Zy
zsw}fw2>Y!pC@%6BmkAwAJG>&bV17aA5}*IBg6dKq{Gh73XmKg?EUv37@Re5AR8<yL
zlp}d*s?1h%XOR#6&+f~;2|8V6ugb#OMT6G^Ip+H(7gyE#`{#wd4U!ki+){bx5|JgU
zvR9-P6f9j*K0t-xW>qvCX}~44*Q#8&HxeTDlXZpcnW`&&wf-5jhumCL=xg=O0BWH>
zY6jG5z8}pkowK5}a<RXxptQQWsyb|o(WZ@Vr{$DX<SbPXww9O?tGH-k%_60uYH<aL
zi;Ig3YM79w6x>*OXL)4_qDxKfl2RrXn>kQ9bXEbQ8G_uRqVfvG2eb1pQeb${L&hDL
zZC$xvsVQG9?4hRAuPiBDB0;h5E(LDrFJDqxwWy@%F0vIB`^&2;m4!vccOWnnmy&%+
z)zZ=krJ&}nB@3%6SUS~_VRlJTd8JZb<EyPK_A9fdPcOJ~QmQh`>zOva;EG9COu8z3
zoi^FL%*zAI<VhJ(SX#O=JufeFnx`Ob(v?vuQh>~rmS#`nC=iR41%EvAu(0^>6=@hi
zke+~_$v<34#HyCO{5M>rGUYU2iGafZPZw#4IG+KW6#mSACKiJg%N~r0);vDv;=9gJ
zkng7$w|OvqgDt)G-15^Dl=fgaJytm^(^J8}`ltVfWz!kTACOL&CV!lbe;6d1X#k48
zJO!0LO<6Db%X)tSJW<&!@fRMu16RY8JrZ9b_!E@<5-$?`@ydG=_ow^}OWIiFGvGXW
z4;Bi@wH({XI;?pF*|<Kl4r^XOI@fI0;eMx3T$0W;opo6A0@A;05dp3F3+Y$c=+-=i
z^lNN%Yra7GwKlpne<3~HMo;Q06d3Cyrz8S*+UTY=39^!GqtlYCBh^MnH-?XN8=Yla
zhucPHn_7ovqubj#$3}<YgpWKM{q!(ZQ3`Ez7-IM+v(ZO{sfyyW(a*He>umIqHhR5{
zuG;7;ZFJhLbu`=Pqbwre8XJ9zjoxCTpJk&zY@?61(bwDPV{G&dHu~8%daI3|WTSW3
z=;zqzoi;jV!{KAUjecI3swiDHy0w?ZWWz>3-^Sl#qffQbdu{X!Y;@(VLV?jnxJO_e
zi8i{^A_6*W^l>(Nl8rvzMt9oi-?7n?ZS;$5^i&)DVjDf(M!&>HciZU6Ho9h`r`YH@
zHu|MDdY+9w!A38%(I?vIWj6YCHoDJ7zsyFjv(YEn==C=G<u>|C8=Z07I+|_tG>Zth
z#zw!wMsKmvxhG;BTEhp4S|H)7-A+YY(dthy8jzwjY){-~ahNc4)p)RkE^^_lUg*XZ
z@iHa`yFxfFI+r*@Ua(W(-y$AMyj9?*6OSXlUf{9B8QOv^0{>zcaE7#Cv%o(l&QKPt
z7x-b~G|!+<;BOO8Bwi@+*N8Ke1#<-cJaLAwpj+U(iJw6{Rp48RGh_vw0^dZOp(^MQ
z_~XPGqJoORA0f`r6zn+(!bQI$&X5%B68JBPGZY0o1^zSQTv`TO1^y%A3_Zd10>7U)
zLr$<o;CB=MHt}YGR}*LG3DyhzUx+j01bqTuNSvW2SSavYh%>|la|Aw*I73U&E%4dI
z8B&6&0-r{lp(N-O_!Qy{Awh@0uOQA45mW?zDRG8|V9%GV|9Ij~;#~qimpDT~uv6gQ
zB0ipYtH4huPS+2v7kDgj7x5N>f6)P)At2Z+@Q;bp^@H^SKTMo1AM^?QZQ{wq3kCif
z@f6}Y0)L)3T|4L&_-^78h^GpCD{;DV&?)du#ObO*hrk~vP8SU-0)K=!T{GBoLi9gz
zx@53R;J+kJR}6Ln&)MR7uTU}mfUH}%cd?o1rVa4IJW>CgE$$DIz9O7XRHPd}1x;Ph
zs*XxH`XaotJ>gRT9$VZ`LXZ|_EDJNfX)?YO#aL)D<^`_<=b}mKg^fZE{WoVz9m`J$
zmrqpc$NJ}NxtHmOP!&^uqB5QJPdnn#_ua1PZ)*+j_T=Px+FMgw-B7lzbqhgz*ST)R
z_$wGcU8G)^r6?N;-Y!&b)Go^d&>DJ2YWkr&V{hBGfeJD3e0tj!CWX4x3+qYSK+NKA
z>x~1yrf=2S4qvaubZYy0Yms>q8MPJ9BeSOOH1oeL^VeUiKjXwSchrr@X)HsB-L^Af
zDM&G&Y-9bkhHKNefx&-4ScTCyZs|Dy?es&T&f9|bAd7mT609-djhPEFZ_2zeGk3mr
z-v`E-isIRf1RmGxPq|7!(DWx<d4x?8Z!&S?r)w3(cw-f^=<S}sHJ-qNL`{E1OF5qQ
zx~6}rwS5`O-cD(BxtY7Y(Uk+p!=L2M_)LAk(2##B`b5*Wh4Zg9x%IU!O_Uk`sHU4a
zwUm!F{UwU^qnv}rcyxl+WTu<EPq~tjXES5(rdXVegE~&LEe)3^dL~?-c0|+vrnUXe
z(#;`tb6UEM@@9OZKJY#2*1{6>9h$z*CD;Iygi=|*MyoWsn8#9_%Dq6jRPLy;;%`(=
z@|l^|x>^yXHME{+kEYc^Tdir86n@Z9P^fT_Mpr6W)ln#dS)OL2;g;j|)Wl5+8dC%0
zL0S&3cQ(2TNn)-#To7_tVERfz<Y^`Y4${`(8hIWjNken5$3>vgwE-iC`dFjuDL_<l
z1E{cK2RmO#EABDws#X*TOJYuDMiQ@$qHGiC#NZ^W(Zx$BLJ|w@Aw=F@K#><T1$m)}
zgW|X=su)RT0UZ_-3gESoI>!_YlUfg)rqv>0vxM~$`XnrrFh@eSgsBobC3HxrNZ50l
zDZfj?P6=B9HFTW;K31H1u}cMAf62JEJffn^(jsBAg!K~oBrKFLM?$xRsS-LRbV#U3
z*pnz#l(196RteWj*dk%Gg!K~oBrKFLM?$xRsS-K?y{IB_EwEPfr59BfbR7=zp{BoP
zj4m}BB$-?$DF!63eq3Uol$Z?Z5r!_1(n6b@=yR`r(s=7P;kI*tiQTIJvMl1%LEvGL
zEmCA+L}U|4-mp^3B|}n#p~7C~F=GkJ{6uvm+}wf!&qAnT{{xmu*eu}&z(B%dZBAN#
zFJ^D?P#XSm0+%w%`7np0&V>@@Na&U@RYIqP4ha=N;a(!NAUdFbX<{tJIB*MuIvQR3
zkx1d95a1mW7li@eAaPMB@P{QX3I|?B+_B14C*W$=N&(lp)(F0}E>pSHE>k&#B%5-w
zem#=ZC1Iz8trD)6utmaV3F`?Lw4#+)xfEP~qTbvAWVI{aL^{>vS3gnnU;9MO+qX!a
zGVV2pde&N($OX8PxdDYVz%?QhL^<uETtkU!ks=<JP$&vf8$>3Ea+|qLt+GsHV69rE
zhz<!mCEPDEK~$l*U<Ge(bd@kQc^@j*iORLHa-FE$YFC%YhA1LokA%H|T2rGd5pW|z
z;i6_u->tR1o$yH(tYCC2?3h=pMsJr_t4DXptF@z_l~)gq?&Ou~SU37b^Wuro`^}4|
zM!$gzOgPex$@v7>V)2_DpQ$E8%bPm^(H%FlOJ2bBeqIB<1=xjJyu(_&=vbtw%U;4j
zr?nk*E>bIB0(JZ=F`uZ*)=sWxW%6D_r-@7CeO}6a-6j{^E#$s)%5qmm<nj`7U!+_$
z`G8IA8@yJN51q19VOOTM;xhSVQ|LiUDBxcF+9N&Kr|vsg_jcBO)nr<yn%AM8WnO9x
zJDpm?R;PB~$$Ipry5h%JacH47t)VqhYxq1~U2#4LYI37XSOENcJ-Uauun^!a67Qs7
z;LQ?mm3Y0xMWZLLGTkJ3wX22c$r#lI-&&WCd?~A4g#xZ}<p_v)2N+0r<tY)@xGI%#
zts)ZFx_F7AJ0)zDaJ_^r5;jX%FQE@`pt#0TjD(vY6miLlYoZX~R$LQ>0k`6sC=|FA
z*F@pKzbdYo%B^;p%B^zQl@skOm9^Sklxu5uQLe4sMY+JOcK0zq+Fi&+yNhzr?q<1P
z)$RpPigr(j5ZUfiDbJiW=SbDufcADIH^&Djy3T0IG21b!9a7N0w#=+2g-}nEj`%M7
zUja-zlKIqiyWH6R$bvD%|FIQ|IgD}-Cxo?HhfArIQwnxSs7TmD7Bozkgq;$$O33|o
zdo3uPwcvnj44;L}=uOKSL?OT}YY>G2x2!=F3f!^=Q8@6ZR=^0evd<`ID!1BYD!0mI
zSB|Y@RnD>otCB5jEu>qOZI-ww*Jcf@BwJa?g*AwBU=3!uLtBHlX`L$za%ZF+TeK*$
zrk%0kn16y6IDh*_Czgb-`{#1$bLWRxy|%nlh)Dh>g1IsLXeboC-yAOjV~_Np1Iass
zyf^gWeI0oB%~-rPf**YlNZ5!Om=<swTGIx~46JX(xnXaN;e^=kN#OX!U>qkQ4BFcC
z6>4oZTHVSfF_*6Zhga{Okrp!7%NT2>pWql0=P;Z|Qy1(~C$-fk-+i%})7yreu~5vo
z0;$0-aTO(B3rs(u>1|qDkmF7(V{co`B6WF|%ctfg{ORA2ao3Ts*z@i_#}s>*3JI~V
zBQ^LR5UVDed49w^Kab=|x_gY7=SVoue<HQJ*2+-M3?&0(=m}?7$_!bkYk~r1HQ7t;
zI@IJgV<PiSioZw$Z|(TIon}#=gmX?sy1hL!8$xlLk0IHkW_>D{&2}9PK2NA!YuJ_u
zmUwm5?>V#t5<b`<J2SAR`CZctvg_|jSNy|%G<|yaxtLEJG&;J;8izhl-3<4G6&(z9
zs+)U+c;hul*7SDc?GISOPEW%QXJ*4ml<<8l)w@;WQsnMlXS1KX`WJblDDo^8c?r8O
zV$UpJ8?yXcWcdqGz82ci@Z|_~)gmmXyN4T(9R_&FY_eW98Mh6M)5sV|`1r4VYpyn)
z#4RtON&;0mhlM`_8G~2;0A-=_hoB;CR)bSAsL(l^5uz70PYc99EdwMUiN${|1J(k{
z0EsRafl!gKhe-^K5_U@1D&cwwTL3ws!YYS?!?CeD6ZsL_tk@{z1MgsLgIrM#a4R;7
z@_;`q(?z+!S&gAq0H$86U8Y{ETz0)|m2QS^_|HnO29%X<mX)oS&?jM`ggFwrB}|pj
z2`DSg`g`@mMt62}eT8USeMJ$r`ii1#^%X^)LVZn@FaQZvR=MmdG1iN6(Z1GLAj$=9
zjRm4y;MQ0m$^~98#|9_mVk{7HF&2n&Fc##9@`fG@WUJP*fg#c-VWEUM61pW!mCz}n
z15j&k?ZPHrvldttUkH`NI^r;6Cf6b3w~F~{d?A;&`E6M7##0EHssF&t^$})jKxQ_^
zz|1WXW}jqk9hA8>!rV+|OpnzS)1W>(!fFMpBPfoo!m^oewTjgd)Q#z^$Rf20J4+;>
zCq(n0CsGBDjYh#Za8F=FF(=1q7-EjLrtPxg9b)>bcsE4jVs$(RY;(QDwasq)u-3_Y
z0c)pS9OL#m7OBg&M%rA>12X08Ht03wY<7X8F<vuet%~;wSzLp0z0`)^F35t+e5f#G
zTnA)XeP=EvqvhmSa&rGMIeC_x{J%?%uQdkU=hUvPN03lg?7^snHfYt_x}5qh?X)&5
zo+?qdTUtL>M`8S(9T@wiHC=HrJg+?zJ++zm9zy=5w(%p7aN6U}f#)ke@S20e@7k^T
z#|M_Yk5vcCGb+I2(YJd7@h;AJ<^-N__CV8YOta=@)-S_Ad1mI1W=`*R%Jnfeu#7eD
zB2Qpgc)SEnjF(>hePbQualm*2<zPI<SUgN!^~&dBxKfa>`w>$k2j?}s>tx&lP7GJm
z|6A4~JW#8RH{&)-pW$jOWDGuBO_HTw4VjoYdYc{!4_B?&M&j04T`Xr%;d>+q<$sam
z7atvk3ow+g!lev4B+-3(NCNDVuv5ZT3D-;5B4INi8pMmeKTPW>+i+?SflzD>T|zW)
zYv>Y10JnxNQ50}%=n_Q&|Ei(ORB5%#RB4sVu9BFx%PLtBM3jrFSrJ5(3*3qzqFmsN
zAYv$B6%jy$Tm%qN4nnzE?!W<rO~)|7_TK-NSWvMiWP9I>OWEE%yaaQXgq;$$O1NIa
z773drtOqpP+pB+J{ADJjpsB2;7E*y*O)aDYx0+g%0^Dk9Q5x{CYHCv(>@Nr{u-{<Q
zMpnot4WdxO90}bLrb_6P&>^7$C`T=uy$(Ho2{puiOFYF5J${Kgn#|N?=<%zI6*HL&
zC37pfbKt>_s!Qg2GDnSHR(n|1M{(?M%lfSLv8>N6(`kKT{1Vb(exjjZeoivL{89zu
zz~=YO_!a;E)%f*U)cEzro6+M}#le2#m*-LEz~k3?@b(?QCR#K(eyszJ`=#cPU}%AL
zNgTek6-WHvtN#H;FZD+=rgys{@t!v$_aERGGJ}O$3%;Xb0ONq$Ed;h3pFnXe_psE%
z0QRyufL)J_-4B~8UEYQPjE_ke4}uc|*!_>lYS;#_-(9No9l&lCg%7saTp~-ah0H+)
zu>B}AQXkv+WnhCr#?udQDZ@{pj7T{Wx+P4N&?%uqLPf$Jh(J8$`ip|O6!GD~OM6fi
zqHRG{6k!Xhq9|KX6-AyxP&HLr?J`xuT@ss0VumYmD}sn}Z4pG23*3qzqFmsNAk5FI
zar_c;5kN#a2q0#;0|$@+#xHh+Z0|R4Dcd`T!cmx8!c+;J5;`R0Q={agxe|5(it+0t
zn>qo{ZrPh!NVPS!kZx;gQHrgpMQNwd)TTD8U8XjxqO_3}>XeGMO1NIa773drte4Oy
zVIknx9lsRjHAfkW8+!b5MwnY9GslpDha`7|xm7Z^4$52@VeTSx)c9q!hh=>f#|~uK
zv@WZCEbD8L>9jsEehKL?KhaPyzj`vj{F(*hz~=YO`1SwO_%(Ka^!T+7c%Su`{*ZIv
z@vFzmyE(G{s<3Es{Q3a+*B!r-UKwKin(!(Gwi{PMaNqH3`O6%?A`dM@9w5Meo_K`A
zd{SXS7K|QiCEJ4pS(4-sP9q+1NRuSn(*zfQ6n=u?m~r7Uu`kf*8cW|u$#OZk#CI{n
zCtU81@CKv`0kbfXJX$iR3XBB`W~avYAQU(WTpYLG2%#fq=p3X>9FYwC#t*M%re$*e
zj)Hvx^NBscELWkx>LivUuzHEPF^3RB*i38yFK{a14Ll??H;4bZ9R4xkRE%GuQ9a_}
zvJeh+!Mjj9*EK>-=Jv<4+$2Y>^LYn+JN7TQ;eoviuim4z8O}wSYUNH)#<#{`4+9zV
zT~0Om88c(93oVdODY&tl_X0ESM#f$%<15VAiHvwm!1r2AmYTd3H^Wg%mYTN>xp40{
z|FuO~$hHl8B|am@yn79D9&ZtUfMkP04>&a45JJ_;LzLMC&I<0(f1v66WdiSWiv+)`
zP|f?)D|0~qaF%;rIcnYsuij-T_^y@5>vF4k&+<O8!P7@PeZkXSSgO2bPF2Pqje$;@
z-XRNpOA5$;Thluw?RnBVV>0m&fhHR<(uVolP|-b5pB3a5R1iOI8=*<_k09r`Jq~rj
z>uTPsEPX!+XTOS|?0Xhjc&nUQ-m+z34aPm|kVU`BqW7c0`{dq@+`NC!+<3YnTp(3N
zm&x3Gm_v3K%I*Yq_CA)~f!v+QHOMZP%q?~)g_SU4t3v(TyyM#JTS&q(jbFYvNJLQ^
z&v#-v%2{y&XT|4oF8u8CL&TDULR7o){tKayaK^#>xUbI$=2P?Ijy`^D;?j>d<I(mg
zKXym?vFW*%ADdon`LQn&Th>WGen|Q;p3j3nn|?e$?8nceB~H<MBmN>Dc0jk}pOW)N
z+=<>yW5_psR|X3hDSe+F`WhI<J=4c;^QGVBNx!v4k@qMNj+<ZNa@(AfHa{MopI_n1
zv3VvQ)}1oG2V{7&Euvs+a?1E#kneT*XkcAD1$p|Ir;~rTx1yf^mz-7G40{m|>TBbA
zzS?v3t*7C6Ap}~rk}cZLdrcu<;lMT_GWsy)qCun+n?k<Ak8MIY2AB~a7nUz;<38%|
z%X>Qq_2pZ~_VZ=sxgmV{tbG*KZk$Jv7}_{aa%kfaC!C5t!7&;G5bhC3#Agn=Sa69!
zx0TQ~=$g26)=_w9KFV25hqMOWXlHc`PRm(Sy_f({BL)(4(6yZP37fMn2s`UVtZI00
zdWH>(_RsgaoHpaPoYNf444a|1%N|J1<!FY1#2Wv?S@JoU?YFtib}I`UOwH@Ext=wE
zC;MHgYF>|N7_bg8O1w?<Sj2I<DCT{r`c`C^v&NPD1O0|MH6KrvnNx+xD91<V#E^v{
z{N^M0Swq$y1Q0o7A(E+iZ?SNec=lULU&na{!ww%d!*qoYw~0aPS>%5m><BzIbN%1w
zI6vP*`xszbNNt>WPCv(Ke}+CW$cE5sLKN=*KWlfKhx$8?&+a&>(T>CUq1|yzTsqGF
z34I*L88yn8u40XHrW;v~Q)rF6mMhbd%-O2lac&Je&Ihb&*l{pj5|JqEA>p~3W$yV|
zuGENo>>+=DM-z`bax}~3;1iyp9wayBOZjjno4?qGszMrQ2TvlNzQ(KfBF@4I+W3PI
z0LQTAgw`~Q(?4rYxLc}~*U{fu^5N3HcVO(@FmgHz=Xo&tJUkO=8v8rul<*A}E~kWj
zo#i#;f7{G=Gwx<`rk4jf`C^9mn%!B#1H=my7d1fagAjTer-=jH8nt;`xIVinqyx1(
zdzbRh8)eFN59%zZjp^qszw8*oS^g+QwHtrgC1;G*B+wzIgyF^9h{zCRTlrbj%T&P}
zUM9+wA6Hvk`K|BcOC~OTX;pF`U-Cuyk~t&|>`UR5A7_q(ulx$azSM%2IOQ3mwUDx0
z?(bgii7DfLF85mdE%)|EmV1X`ZCsSlOs#3-VOY{o%e?`d6ocCTYwNwf&ixgOKbv3v
zZyX5&I|p^{tMS%qpXK1w+lO%OP9dt@*uUc|ojZC9C2H+Bv~xe)$GJ^hI`?C)KF)0p
zjRVgxO;@)D%2Qkp7KfdC8(QL&oxAV!@$X#@ime=sn91BO;%v-X4su*FmxDIH4sYW~
zzaGGSzgZ5Ndp~kk{J-3nw`?8Mm;HEyy^k+{)H;MO$8V#kc4GuZ_E`?v=dq?2S!+YN
z9OOPMN8*+~&T8V)S!2iban_uuk=S%yYa~Wcz;3POtj<$g4lWKm>zFpyrtc7oSuMsc
ztouy&#9n21c~`>iO)TQVbJg&Q(94A%=8=6}8cRj2`#J|&5$1E<hqa=)PP1+4V(@gz
z6=9!Au5Cp)<RtgS{*&DNJp)W~F<t*=k~`#9O2ZbK$bc)tX?WGPkMI2DX}ZNAE5Z&T
zs@>?^Z1){=MQHaOr!Z>EciiTR(DI!%eSF8nrSI%KzmM;j{&7kJ-Knhz?+E+OYpiNN
z-@&>MTPfk?;Indtg&h`m#5Zi~zEqo+9Uyaf-4}MT_enOlC<|@Q@*&CQ7Nyv9?Btp!
zvUh@8P1Y@f@ZL#RWH}h#J9#zYDd9EP0Cxz!**oE#1e>RL@cOXrF;w#&!c9*N>MR#x
zm#&YqtlTt&v;0bkYBzrWB<54tS>!Gsw;5Bdtv=H&66g*XQkwg?g^649zjOPzg)eGI
zF~=%vNHK>SYe>Q3E_%<CxAXdMlvIV?VjC(2ORRi}Z>yw=b$Gw`Y;eym;tbZ_NZ#&<
zH(A^K$@v^_UbFd#<vPhF9HZX0x!Q~L2W+Nb^ukDV*ybrOTPc1HMPf^SKNS<(e1~~8
z(6F*U;w<646YR$K**obsru-cPig#Qua{YsQC8m&lF?dwsAJ{B0rF2kAC#KSa+*`9v
z2M64f_m2BK8(;3|=N{K=9MnBdhavZIkMh3`;U4!2QSC<K2Gc#Pjd$T596F-iV`U%r
zFmdS~&3OGsID~j=6+ULpu}*Q+u-wBO#|GX<tO>ivL#PzpV~>i7k#G+L+di+M;cj=t
zJuHLIgY8)pHQT;D_6@^>%^S4yzkB=;Gp(V<kBC3W!K0ru;C6jKe^B#bBEA=CQP#FQ
zcLe!<!%JAGi3P@s*h$2;yV&!W+wR_RxL4<Uf!VCxcJ9hB8}90x8;~#9HxR>n3}Q1>
z?njzZ{=p4IQ^-Gfo60*b*xw&6`JX}kAs)Am`uM}mPYmG?6+%?I;r~l0WT<(te{5)X
z;(ZuxMF?LlJMw4dmET?wl-DIid>KYic9DXcRIQlS86Pk!_r&>_h)8}!W=IZG)<}vw
zO!*-wX|39Q+YOLz&CJidE%Wxwg4?!n0*UVt2z>(Sro3BZZt*!7QGzBZ5tXhc#Zu|!
zekz?UGelHMkd%l@pF9o`cs_Lk8Ybw+lP19`p8kcWQl1KNf_~jF-EG~!#VNSRq*yH&
z^a3~Tft+A(JFdRZt6%TM)nZ=dzJ#mWc=hT*T+Qdzuin9xhgZAtW`%J*uZ{$9brr8#
zkKk$oukQZ{SL1l~#HYAAn^&(K$JLp<`rr#(#qsLnFLC<oNucpC0waW42LHhD2yOoc
zGU0l_$n&kX?=dxekmSJ!kMz&@#iFKr5;c9VabN)-syo{HiMqU@3$c-~Q^HmWxj#&L
zi-ep~5U)4kbW9?aPcP&tM{95B0hI?i(($knVX}lt5+(wIRRiFmwa3-(aKp@M6SWDo
z3Jj<1WUQvwXd9WD$$LGD|2(K%W*oVWcU`AzT0j)91p<uV-lpOlZ_|@PYzl(t=doUW
zr<SsJ>9^41TEo$L<1aE__n9?k%x;RW*YpMv`U4GO-4F=0K#bNWZ_|@ZH16u3S8z8y
z$(Dg8CJ0c^Xb}!gzt`n3(k%h!)IcFnqz25;oggT94LL^(+>jTYe<U7gHNLpktaA71
zHHl`%>3QC!-*=d0n;_8HuWXTbvz0eeZE|{>{@i6|H$lK47FBy>KW;4VpPihD#piFo
zS#<5MdV$YK+m*edRc$;2{pm&h61{<Jv`bBy7I?HC>DhrB(P_6jw7@+%-oT@5RCF9$
zjjwS>zt1~85WjJn9=~~-p7pc>WD}4rcu0G*`d9;>UDL&RS^%k_#0v_~g7V<><5QoU
z&YTZUJ3e&_vTg(uRn)~f&~O6R)6~b}x9H+>2I9?Se0OVFAa~;lB;;;5F-;%60dHr4
zvc*Q}KHq%bTrjuCZ$OdwR@bm;)@7(YJEL8_|5}8IY`x7JKr6s9ghd7d4seH?cP|Zd
z3_dDkIH$5_Z_wDD{yf-%*-I;7oXp{Edeq5`fd&%<k_QrV)Ohu~5iyrVi(w9QHvJG{
zOb~GQ6LTIr`)5*&H<0DRb2C-vzlGS){VjnCr=ecVo4sO}+W2p{@B|!qM+|os8@e!Y
zt$IOi{3Tk&tMAns-s{zxmSN?3G!7p~3$^<%)b!7_hPPsB#~|8ZeOZ;zU`8I?F0bn7
z9`lJ>buNDM98V`g1ZcR;Q?*-b2*uXGL3Va$X-(coQni#1!Sp@MNUQkqn5v>4oKhC4
z>2Ay|(%qO2rMt0{lkVQ5HMF_$wUnOP=a3tsFb}B0>^<F$=|R;t^;@mH=EH8B;-<TC
zi*VU<o{W!b<MrO|I<2WP#*-1$^lpzj3+v)nwUm#be4Li@iGMhv(~4dGWPENV#(xgZ
zv9(a}v+j8Hv8j*a_Xu7w#}}%IX`!9iA8YtqeD%e8ul2qsY!7T%kApw!@kE|RK+kw@
z#@_08(XjqZParirFg!aj+T{t{h9(}l1uczlYRLBoA3Q$wPnr7EhXbSkl&L;8`VoEf
zqk8;*p}0SxU~k5r>NmB9?MYEXF7!EQxZ+>}cwTMdpX{T<qGo=kZ#}-YDaHhZCao#g
zmALta?QX^R_jMRhn(lQaX-$4avGK^M(KfWEI%6Zh>#SuwSL<YA5=)qeZ$JumqwbI3
zyl*?j-ntDBfA@+w&SpCYv-MD@M{P^gSGm?9Jz#?F;eo`a?}c>7ie1b9#HUePkY@Be
zipOm^5_jK=Q@^tPi=t$w4<`F!W@bH%gQ!t=Dz|OFeLkz)Zk!1iDf-(i`e7Dbi=tVn
zFWClwp8={DW??kCi7l3l?~!`+la!jMHH~(e4d!DekkCSYSEBI|^P<7P5os`WMH{+P
za2fUB$C7<4gt1^kJp9WIF5{wkbg$cy^(oO7R=QDf0hsYM6$hAk@OfbVaVK@u^g>?u
zKzm&0@Vd*qcJrDI%GdSTDg0PgtAP0Wn&7!LRUI|osgAnAp^l2v+Nb*{I<tKkP0!Tz
zA=d4_?X38%nYRzot1fV$R|#yrs{Z6ObUbUZ1_c^p9>K)rxH(}l<!3>DcQw1@?(<M4
z`n**3c~tY^dnI;z8fWiwd&w##r9x7l&UHd9W6gg-6?CeTbplKRwES~E9Rc$ZIDi&-
zQbbHKCw<~!@CKd~kysFZZxTd=7lfaggf2FjAlzpX#0V`2K9j(K&*UgD31UDL9CJ)U
zeVF4~LAa;P<iFg+a!hQTiKUv@StjNnHtg15JkVRW;d@^VA0+cdlu@&M_!Z)pgP$8e
zbonfF`7Ct#@W9M?d~Ea#^??xxzaiI-j-_r`$uws10Ww=5AhW`<rZKx4ANs;~z8tD!
zo>O()0%-;KEr7(i4oD1c8jf0;9}PhH7(5$MDb5)jt77ZTi`B8DnGbc~yI}le+s%mW
zk<Yf}4|js!8(WBTZfwbroNmF%b8al|?Q`z~GV`QN-t;#=1(!dZ(!8-w!RZ=;bM0{I
z346oG17^T_gYOTc{>x?;s>um2=0NiW7&l&4K|0;MoD0#*W|v7M5zT{^nv3R{&1P{k
zF;OO#5%EdHK?G}Giks^rQYVol<VCB`O$9{r&dp(BQg9n~No<X4d%soG4QhO4dpW@~
zSWgVpYS8|aTFq)-^4oi&fA98d-v>}D)4GOBqeIihnzl9#Ef}^z7@e?9w8}IXo3Kf=
z&MaRr3nJJ=Zgf%FT&s)BrZ1FjTBtfsS4XL`SI`d2W@|D@BVzTR=%O5%A{yT6FCX%v
zTo}69E2+$uCpq(o)CHZ0<G7+NUiP2(&?u>iIL&k5bW0LW3t9BPB;!;~0e7e4ag=4#
z+&s_Lc%GNT^DTKiFD&G_x-3gg!C)+|eYl?N5|?$E>Vh5k?ZNLQ{0`uE2)`rveTLtc
zGmwto9{gUy?*M*>pc%$0Tzv^axB@=aMHIi!V2Q(Njfprdz;y8Cf+U<4J8`N?#_7&f
zoR+8Kd6}1Ixp|(e@w_01=Yl+*7sF4)RPyCnWyF{Jc)qhP_*>3JBB3Gbdp=*&D$#@k
z#D~G{*Zq}NY1uC&=f+0j!ztKrJN9bOk;3?%@P>^Kl9nL=nBMJ%$rQr+NVBYtCNMkM
zGC^UTW+btUPuN$gWr>jBqwz79WpwF8(k%;=Zs7}F1&tyt&F?v%a(l7<gT;sbZIM6R
zEhID7(5+^*FZu29p?{zCiw}!29brA`+_^Vc?JJtn@@<qK4#p@U((Z*stOlnahTUAW
zexlX-OtHG6jEL0-qPeXAAUechbqLw4-tiH!dWBvd4&|bYlB_Obiq&6r!6q>vLi6N=
zduB&f%(5F+EIfD!{Vgq%VGkT;q-B|wW9cTewsaF(TN()6EDh3OF*!1j!nS7lG_G8>
zCvV%5jUDUFcjC0^5w!lsf5ny2@!L@7kxDmoR+J+WwpN+=Y@Z38-#1~gghwPytTE|b
z5?1?7++AzJb20PaVa*qE%*hL)ZDf-23&amhQhM)pE7l}rlsW0BLkjF_HY(?sH<nqg
zi9mmoy^xG(AIH4WLvR(KtsM;9j(H=(Z2fJufPzjZSA4MA3<J}W)Z{b&sIfQO5G#jk
zfoa$enuTqFY1n$3#h!@`{+r+p>+#hJ%bHo$8bA~3YHIRT`pRkrKVy4~krd-D$6H76
za_A+lheJ(_vB_t?Gs~)pP<>ib@HNvNW0APiOa$*dcoN>>T6PO85Oc6)3t?|>vHJKe
z;9yd~!DMk#aF576P*?S@(?DI-pKHLb>fi6St_n9$ge%yN#y}TgNgJ-ZF*96sH14u#
zs3LoG7B;l7O*tzm$iL@|R=2Fl){FhM6m{v-i;?23S*oZ@pI#h*i&l3E#+lT;kb1Kl
z$A;@7s)ybYRbjpG3460z_BtkYUm}u{&wP7^)e0h~)d-m7ia8i;yZ07*yr6k(yk{%o
z#hPD*LUZqQqhS?gp@i>CcvhWBUnAk85<Vy4nRl7#e~|E13G3keJpL@96)#S4y)#X&
zcP=s4JNp~lF!z}v^Z{ak=3H~842d%lUb)^GraI<e%6F4fb=;oXcfIox);npfT<Y|l
z&v=KiyK{$)4Vznks~C$mb$ExdKXZpg;!%Ib{NeO(?=YTohebkCf8)p>PH7{C3C8$v
z*#j{qVUZL5#MW#sb%vuh==V8@VF|(M?0BwCR4p)5#;}=I3>zs&G}Lj~9BVvrf-x1Y
zkqnp&n94+JC`%`jC$qW1kpm75Py@_iqT4n~&Z(0~Avk>CC<815^pUerEQQQ=7yC1!
zaai%8n^CrR-;NomaoEzQ)QT*%f5!~eh5g(67h>r%lfxYCjH`0TOxPG!=#H6Az*K3D
zFhgOEIe<RU3jt}azF@8pfxL82*lBV?>NKn4#AFejY0g2hs6=OAVM<2(<RILCm=B;6
z%q8m_48iPtvGxr2TOK%2u9$7i@ktKRC3Qi55D%h<VL@@(x=hoM(2C1$LJKXsorbcA
zMOP-{v@{i`Wn$?wQ!IVvh^5a>V(D|cSo$myOP{5-rO(pH(r4-auBFdon4k!2wc?eq
znXfj(LE^(;_Ihe&NQ3QX*{5c&?_+)dKa``Su(qw&hpmmBJqH7XFdK}L!{Dw-U{42(
zGCEqooMg$E3JxbY(gE`T-AuHMPa~pPmIw|vIPw7N01KID8DSZbGRq>tQ3#GYu@o9C
zKC>b0@uGjT4iYa0YAjU#)LNCL_V1|vW#<p?A1~xmNHn|$w-?Roaw!z{a_F(l3eG4h
z(jj!fu-l97u)=mSQ>-wZO2q1$bRt$?h>o!WxW*JKVCN9AdMS^H)hnWltWaLYltQbE
zh*<p<UJ8vw{PJBJRk1{w0W%KNdz2Z|ENw+SEN!tQLItsFM$wjTLTgJm4IDY(5DHtG
z2pug=%4C4WkkfA|bULhmWAQdjZJz#_yf5&}d(8RB0SO<y*Tir9feFuBX2RPgG$f3x
zH|ftv_=EdQ{D%8YcoAkMJgj(O|MN=qpnLQe@dP67MZB7D<hzRU`c#+k;J=1Kuixu(
zznXARQXU|s!sUMRqgNB21MSte{cj!cy6RqS!`WxvW{0eL?Mf1N9Sk1Wuc_0R)RZ|n
z?Ro7uZh!U0WIi1OCa*fnz_(Wt&YH?Qfk)J2LtXH^e<?`{NG6Uibj$ly_$+6ID_u<y
zZ#KSf`~V{O#FLxPK4DjQVZyJV&muKH;r9aGi|=?kysq?*@x{`Q@whm$dtGG*vRq~R
z&pZ<x2l8Eo;J}0FTIiYD-$8+oGgZemJS|UD8~+;$A`9;I;EadIZKdP071S{wLh!wa
z@P0Hna$U(d;~{k3ZGQt2s2gq=E<8UpMd)_^a8ox5$RhL2`=J{$AX}Cz58CDJKeGzF
z&@Kl&xvm^fXix1!)M_|Sa})9Uqo}sf9kp`^<s4A7dz6~IQ%!!()bFCFAm(HB_Ln|Z
z^G~9nFHq1)H1NmjEw6z52FM*CKLc_H=9(a5%Ffp$m*KI8x18E(9Ugrc?$gwx8sVFO
z+aiyHG;14>f+GfZp6B{J4WFM;eYfYn&sU&da6c`upv-e$+da53GXt@b_q(*FYCJ45
zd$_jZn5KI6dIO`M(*k!EYH^ca@idKy^<;dUea`DRd(`Q>Jq_=~;8v+uFF?p0@d~AB
zp{-iO7iZu;YNw_{c&M%Rk6QaIaeoPq4FLD*bMbD`EX?k}9bK3f81bApFyfWDO`{XF
zm=15oKJA=iJgd|3CokTM@uqZmQ(840>ge$(ZZ3*zgSnzonmQBD$t@_`Mt_;_(O>3c
zy126wn4gF{?_T|w@n`%oE>F{PEXPSYX8hj7F<P1UznVCPEfa4rai@jXm^fy*W_q!S
zV-Pd(`6iCR%*3agxZA?7G;z#(&Ghe>c#egiW#Sm(%=9=DM?0DLC-{Up^v7)2#NRP-
zpM}3*;&m3j4LI&D)}!z7$B}RkFS8)H<to^B_*1T;zjKV&eu5{9cKlezHSzq=d~}c|
zAM$t|mf{Wkgg3FgfnV_EmN#&9+Ocf>&9#Pi@EE{Rc=gec+ISZDJWbzsc$#j<?`}sn
z{-|5S`w<3tKL}gGRd{!VPnlLN?!gJ`<EndjF#*3hyto>R;xWB3JnxEm@c8FC?&U0l
zpk;d=JpLs<Od%dU<wyI$6)U)6!G(+QTLs!1Cx~$tQJ{w4|0pcafV|I4d0#N==OWz@
z0zR_jeIZao2>8g7_XSY-*umcxsVBGMn?xruV(>4uY{ReJ>1k;1RacyaWtCTd6CA8=
zhx#pi(fK4MNa?Y=kYm|1o{XN_#4P>Q;N56aJUoFX5PD&;@SOd8;4`I%Pam9_whIqi
z#PE@W<K3e(1M&a!VCM0u|IH@>9?y(<0on1qL2UIDT*&wfp3UUhsJ{%Z$(v&KW(Kl0
z$Afh5#&~@nXx>fnF^KQ&vEGdK>Q22wn8t1R<A3!~M8VvxyyH=-gujrWZ%x^~?{LiS
zw!;^c@5W8j<s{&*5^ny*3zx-c>Wq)I^1XA`qTe?jf`>oCTzZG+xcF2<3jP|Q!M{n^
zW?Z0^w=y>$iCJ+ha}%?8)fxLd<sb7BkB*8wsPRB~c6!Tq3by<3NFDWh&*?qwWlg`v
ztG6N_&I9H1q_ld<J52Fkp;Gxrruer^@ozyq9#fx-1@F*$$;y9KmA&O{edKGWeFXWA
zf3p5b|7880|788KkMK`ze+>WsRQpN)C+qM2C+m;hnSW~g=lnzZn-K*8<-R^{o2RaB
zRsW}zt3*sZ${5q^^P$M|u{j?7Q_Q2YSG=n>Hed&XkKcH>-<5*L?eT|IHy!=ZJT&#O
zt=^0-wP76t+=lVy5$g$DEguzY`#cUjc#UPNma^NM@wR&ZG|2O2yyemNWUJFY@Wkv<
zf7a&BFw~!IP27#hXK-vi;vehD*r7ha8^@kIzN|C8zmkuqv_Y<T5-0<InrDyN$S<LL
zn{HU?N%_dLL0@SecnHsP9MAtB9>xh5-c4!fisK)A%|vjnor;JQoWx;UKh(^><MVAZ
zUgOD^%@Afi?t|f~PQ$eIfai|wb)J}+iNPN*6%&6vmJ130&e}Hgis#)rx^Lj)T-RZB
zjrCEc&)kH{<Yp9BRhitJ!uC-nHmmyVncj?L+4r34D33!Og%A7h#Itspbx@L`mO)gV
z6y=tp+)|Xgzo<Nj%7G{iqM#T=HMIU;71i)Xyn6pIZgix4lJ>kD5Aj@)7W0aKjuxmG
z1AQwTP&jK$Ar=A-sGKz>hgVQKYYZA7%K^3H$6!S>#(`zQ7$-hZI0kCxJ8;T%KzXkN
z+Q&OE{MPyJ#v5H@47Req&OwOhw#Muy9M$L$5I84X#03P-={#`(QEDCT+KyQ%GOr%9
zM!>aW9u`b%9VT4u5cE;29D)y8qUqY})W&+SQHMg<3Uwf)4mpG<LqOnei3<okRpJ5y
z?_qIJLlOB`k8udNc8pW-t#z1iwL{RMhTwyI=EEbOD=kf)8nYicH2o<DAvNg`5>R`I
zZ;-fvNPk%30s=1+e5=OP3AlR9N&(l7StIz?I!w6QA?VOV@Ie#i)9%~aMUe4=x?&hs
zL^dD0V=L^oBC!su9Q=`q?ri=2@Om=ZziwxZ<l}VoFHh(>)4#Ns9iG5-=961D-_J+O
zMz}osNRM8b=(%G@9acLY9igXTdt!JMciXlt^S5Km;gS0=KTl|s^K<)DYpC$7e`L<K
ztf|^Zr#M^rW`ZV`)!$6eY;*QOCdl7R(9AW>U=#drCTOAMUweWUnV;dYYldf!j?c=p
zswwzGO^nWX6xs*K!xMOb16yE?cq=5|h0uK57`se7W1UZ4JWJgL20n#M+;HNpD+<Vw
z0(jjC1|GcL4~SKJXvOR5k5e!g(jhagm47#yMW^kWk@jx)cf{@9$R-MTH2mFjp&Xur
z=1AfR{FaL0Z5R%?+63`ZjMjD}Rx5r<z80eCFY~pKQG5~OG`!!|{cSnls2QO(#W(ZK
zZc18lWZAu1;8*oR+Ebzs%o|nRz^1@v5S1+bE&sQ(^sWbwe^)<S!Zz_&&>Oytt#+q;
zhNrFn8m~3w|2ZB*7WVh_jt7t5b3$KmLSD3qoJ78hIaMqs2KdXe+Y%dgoYdQDpZ5f|
zh?<}Yo}~`>%RC(du5xV$%+`C1J1}AL>Lsou*&SJdhkL1*`q(<blNGm4<k{i?)oA(w
zcCqyB=+R8fgIxUkc<cf3H{8~my~RICdD!6JKF-v^p&u20r1G#aPf)U4@e$JXl7#uH
z#WhiqMp&w&LTSe~vT(2dw(-aDFqlSB8BgFRtlM`plLP2Ob;VI2o~B$r&Ht2(GbgY9
zp0NZsS+`(<t{5-eOU-wBnr@26gYuvV`!e~C8MQE%%)kZwGf>!FJZ{w94KrNpdI)Tu
zVhsHSIocftCK30#l4*{l8$bQfG)HQ|gOBagU@(uIP{^I<I$m*Z=_=!un^(Bif$ehc
zIpdZ_zAF!t%L$xZPH<v!If0YQ30iyB1Rn)z0ka~QcCd>4BZM^5Z2d!jg7Ly#cynpR
zyHxjmV<!-A0G*f_+Mb#5hPvt^ym-j$<j-s$j<a}#{d{aS(<vj9lwbf^Esnm4+q&`o
z+cuVRCQB&<Yj>P@%-@u3&WAi8^RfK<?jgi`A7QQi+l`6L^K|!a>wbtWJ~VughXs~L
ze?`+j@#rT+l)yXOsJ<(Q)fXRUC6-Ez?`XLBwu{rNw;3;?Up?pzyleonvC|39h|5MR
zY1rC?|2JODhJvN6M^;Z@oiI&r0J+ofzSo<chWAd2ms~bbw7=G?SGbb+DjbaG2QbhI
zSE8rsmUvH7S$wwMmaQM~G#omawkxf*Ia6=X)I0D;NCQ_sc>D(^^ftW*5bw6lJ)u4}
zJS(OxD+Zg1rW0xnI}-8dGf&p;hDq@iGlw^Ta<CL}C1&d#G1yDTMqw{rV%pn%msfvX
zOZl50#?Bo>|Jm3G+{BH*4f;;?G5^y_%wD{gg;^o@3%#C{zhyUhH$cJsO^6N^PeX(m
zAE?<G$amKXy<L6m2Pa~VBQ|6n&!CiACFZ23p-lt`Q@3p=YxidAwP+!DYQXJeYodMM
z4|QhhcmVERm(%RxMR(GiaF5DF7Z7dOo)<AA@uqNu|LSP|ONhck3SbP8&@19u#2&Do
z$F4t?^r7sGPu0eMM+d@LPk2+_#-Gr}<3Deo*c?D-(GVY43foV4&x6-avjb;m2MQ81
zH=e<czVBou!r7e@GB+ZOD4vFs)9<`6b0chp*B9S;R^~?dIj@J^IXt^*PRO$_n4R*1
zr|oaCo~9dOycr*P)M<OMQdS%3;o17}?3Cv``;OkwG&X~OF+B^T$n#=b<rv<~^)$s_
zllHvlIIcUV2PQU74@_N`hCkprGbE^)S5956_qG}5rS+D#p1J58IN&kA!<+Hu()09o
zZ_Lr|aeTg)mj}D?mYv*R&kW=~j;=#GmaQ>6rt1?MF`!&&N{pGhdb&P!9bQz^GWM$X
z`!LqcX^KztX0+ARX6qkkr|kCZJDeS}&-Q)+^~GKTK9un?^gUh3(ezg`0~4Rh42*pu
zrW*@-sc+oa)%t$wyT81(G>~_jz76YW13~;L<UGyPAZ_gGOnvMVxNn0(Yk${g!J}TN
zMEVm?c3}$>?=QpJ;o;)7S^D^>;n>%4pn0)W>o)H@q#fhnjCXq6N><=XPsVfpq-?wv
z6E{P@2AlEmUVV9DuoD>|AGP7$dRuQSh9U8%aJ_2Q+n$*1>d#s+KwwMdI`!vU6Sr#k
zzJxfg$9_#0LRjY0EYd$o!v`+HX`YPj>H{l4pWQU|4@@`K!L&U34!qlp@SB~;HwpDU
zTHBX+aFFj!p76#P{<th1FYz4lWAs1j(b2TKji1+Y!FepZ>E1uav!7Ap<3=i8Tc_7x
zr}%Q#;f5yfA5a1A9DgW+07mL~E#m;zt5CIjgmDm3g5T%0ik6LjD-dz@2O@?i82f8X
zcXB}2zU__I5cSXCLslI*iu?;+)4o*8*imya-Z|v$VKB6N)W?pZJy8W?4c>a!GK?DR
zns|;M?=D<=lSlv1)AnJix2eRH82x-4<8dNe%B%ZXG1U4)4$OQ*pTWQFom7gPL6I^0
zv26yyckXW7RlBnrBNg6fWC9k?S{$~MqN-@dFLA#-cRqr_bBG7#$dHYf@^IfR4li7z
zL+tO1<F&Dah7N56i|vhT>;w+;1m^T*=g@vjUv{kA6Y)N@Kg|=k#)H9&<GD9*243Y%
zd%Zaj{~Nr*`CE9wgYr@z{h$`w=FQkv{f=e+xHo6re~I)*j!^z2HW<{#AVgws5?al{
z{-mgZmhr0E@D%9U7Bn4?>$9=Lpl|i2e4({{0ed^CrM#?Vys6&5iCh`4A>??~Y2BWf
zm#`1%%{Yq2P230TK8krQw<DdNj4#y(egS4!^B(~e<4sr!`~-qox;Ld)gK5K<{e0UT
zw-x!v!(XcgPZOd_)97C_Q`#}(64;;?bHLLyb{+i?9Tzf+LCe-J_UO06j^o#HyQBNt
z=<ZkySlDCJUo$>JJo3cg9ms)pK^{V)5kgUY;(uDJZFd*;_*J;x5zTLFZ~Z0CBXIYf
zbJJm-_L(BcVE^>{>dH$v%PsD<8b;Ic+(jEJuD56F*UfEmJ(jI65^qPtI!C*Vi=Zwj
ziFhtRH{b!`o37YiT{b23%=Gs7#~@PCceEY3FlNWTFTuFtb=_EgFt`c>mt6rUA{-`I
zg9QEH@q-PalTWk1@xdi*3e9~b<~3{rdp7n00|x@cKJU0)X~#U{_Nk5R#%$g2q;w#c
zI`?SI0lu%?a0qjZ9e7>)NE})}w8MXzrU&`Dwx7>`zIu+LOvi-tu}{VM-OVgs<9Z^0
zcGJW^;)4}<v2(;<KvcApJ>HlvA$p^!%X2Z$z_vBq7{(qdMEw||qWxbTH3Jje89n%Q
z;fI~;9jypsJ+<#;>pS==dB(?T;{y<#r3a7iYxw-6tmVFGP2Njm_1z6`$6+Hn9(&yr
z8or3}pV#n3tp5zQCaaD{9Qd-~^Z4qkZfF|uSZ33WF`8H(kstQb&S$xm_RJGz(|!L_
zOt?<qZStOmttVDIzr0Jw9}thJ{T9C})yFIocj8TnkvMI-91=F3hb!YZm9WyA(f8r;
zc!Ty~q5THLhQer^?-z@K?ghSI%!d}R{Ml`M8*j~O8T<X;;p>RJI9Cg#a`uXldE7XT
zV)*As+Kp_KWxm2auHB88dfa#&X{NS_|9Ej1Ovzxf<_CD+^%LW-_>e7xsy}bVE2Gi<
z6Y&SEzqw6{DHWcG@^HU0!|O`)CwO!KtSDWMO-ZOYA}C(DH(X0lEGE<=`p82v;)Ica
zkFRKfL@nSkj%6h}6Vw%Hkd19mwK186U}l}@`7WyGzkv5Kh9P2)(6Jib8t>7OBu8(|
zpRxlJu_o;Frr>*%y$EtUHj0i@vx3-S#TZUQZVO@zZ%b^m#<0-!xL0jp#C;O_D*lN!
zI&Z)*UTapw`vA?yH4BOnN@eROy(x#W$ndm<P#o;F>jVZSXy}E8i2EV`8KRCK8vD=$
zbDHjbGM>Ry3|3VYj1Xg8+5AWbreTkIbi5#kX`yI7jy-=w9`w*Kl*%t^>=my_A2g03
zW3UZ~F#;*U+u>coE~*(w2&6mF<`AO=&fLOl{1swN_u`&HAmOhv&B;+V+4puKQt`K(
zG~4^=wYeg#`4H03U4_PZtZ*P9%O=(xB^ILC&w{<q7z>552&84$XYclxw4+}7Zv-Q@
zMF^r^569F)ydGYU1mpM}s9y;wT>d3Id=084tM~@5ZB=|xrVUX=|MhU1LB2sSusX}n
z<f2@tQUO)^w9x113}`jprqu;eT1B_e7nFdmeH_^qHJkpF%Z~`8qN<wy3~Y%rtdr)3
z$i3h~pe)6E>b$KL_^z1ZVv!uDHMt#{J{>c`>Eb2t1jC7s7n!+cnz?>*j>zQ+TyG9q
z!#!AaA)5!U9pNXl1<OOB_JrSNI29-$Ho95>lin*tmzuK`?&~yzws{_d*n=oWE8daa
z<l4yux(C;v*bP}ek~P5Owoy>+cBQVEilvb^<B;082N@{2Ucg*eohf!2#X=s*Qr_i|
zhdZnO35;^Q7%!sSQ(_y0yqvTfry*nac$h};KQR(f@=B1I+yS_mfzLREIx?k2q&y)w
z3?F*}x|;0vZ!&1ml_~Xt-JX)q6ibl)ronG9(NW+;xvsL!d2FCMQCP0aC#%8F#|g{r
z<7?`Qn>cOAct>sAAPTzIl}9l-rkFG-CQphP2Qh4fWUDnU6s-X#WnB~LC@0I%6c@57
z#)zhnY3x|D9S(pYtUeawg~lChc0>UPvDfo=(2Qg#WG8PVSyrEM`^e{J^)>44-dw=y
z-zCbx29(shd_=hVV6#i(Q--4Y$54G5b_%|dO7BVqj&j6)3BtQ6cbt^VA&%BI3UUW*
zk0iE7FC-WrF1Fg^ZY<a-P^9%B&1{c7V6f`L&50!AR<=4S4o17GA3{}<v4LYJC)nCU
zY;G8@W1gi2xU&K7WNg3S9(laFqA&roh=eIuODDsykYM=mK`~lci>Rz!>@P{VZ#yFb
zqp10<G+WS88@V(UlPdY~l!?F*M+t&|z{wusFwGl+oA*4Uxm$+d4)o*x*CDu<_2d5G
z5ZvGI$9>lj+$H_ED~8}+(2u)d2=3YaxaSSQeQiJP8AEWhQ^Q{&QGXH?0ptl6r#Z$*
ze0eVT3Me*)H}Ie)habUk1fyVZHz+oSO?VPC%<wE39tOq6@HjVXEeVg3p#>Bh!zg<Z
zKO#dYT8Go^4CQ3#jb_+qXVA#7C0faLJHsVp_)9dyZ|$`hMuw*7TKwG3aHtpz4WI;X
zWaDF6fDyG2ttT&UWJ6AoNdpZ>jk_7hqg;wEKA20rinQQidL?HdedBne`D9FFQlU0(
z9YPqrG03&B6c!y78~JzFqc!8lwFcAdKBRSH*1j$HU$_zy2N=;BxoV6Sx^M`g*`}!A
z9~j;sy2BFPcbsVaQ$M+!LJg*D71Fw~aj`9UI3h7R(l&PXllUJ)Nc^cOC-?#2mQoBq
z=8~%dt-JAVjz$xTgi!>!x66XRl&~BbY%cI5<-kYwBiVQWc|>#C7=9@k9Dv3mGOUc2
zz*t2It&}i>416gkEWw;Zb}=!TDh)KdXuOItY(^RVfh-1h8RMho%6RHklx$G=Gx%<0
z@CdH@l$?aj#-ET#^UaGI#Tt35@aq?APMe-^IVm|8lDqHfQzmCM#uQm*)W|v19DLwv
zn(eoq5{&}ZB7E~8MWFqM(Z9Ehhgn7=g(9YmKwn|W8s49=zi-3YC-!Ht8|uNWz2Zyc
z`~s`g&Ck&Ee2_7xY2rJa8)2&&3*8S5&L6R3#!iT~`xXBMY1tVc;2AwkQ&NrjY&I^_
zS~s~B+<a!u<MHFtFnJ5_MsVx>P2)TivIUb1iH-mo(FTi3e2naMD)SB&iL;v~3uUlo
zIA}ai--a6CiA+6eXW=oV%)m99xGk}fdlAMhpFolOwzIO~6<zfAz9vW9-Rnxm!Z*2_
zbJWNZYaoWOsi;5U?le8814S5TK+oW(C!@wEO#C7Yhi?PJO`vptS46wM4Jkf47*+h-
z{`yTv)^5MdAJqc>+oSSc*+1_kVbzyk5q9`oXvJ221L{Z8zh%w|uaSvckVvM#0?PaO
zT==!1-eR(A+)f5`dZO&~W~5l1o@ndziy_d~>1R{^ecLHVZrFtFweSATSK(r_2?mJ!
zgoYzO6#7_GFZH?{1W*gF8YRqz?csGJym7NVX<PG+^SKWlxN^&37+G4YG4WQY5~y)(
zIbq^sfd^_5w>Yp0L-LtGx47RcR5o)>VoU<6?+^s|Mi2Aq+wf4sIpO^=tay!d((%v%
z9!Bj+!}oK6+S0EV_Y8yI0Ttgj3;QRxWayhGvh~lhQx19BzD&Tr%d%Qe6DIKye-<WC
z)59I;|Aqu^ve5rgg<jHrEoox34qQpnk0gnG12(aFy;Rcf2hG#4-KjNAcVe?~`M*PL
zthV34hMPA9&*S2;VQldFC-b>kY}qyR#?|I|0+U^odxenUzXAENk8jd0Flo3;6``GO
z(!P_ef7*Qkw?-kq``hH`R&jQ8pN4Z%cO2634e*G|E^6leugLumaVNCt&O~!-02>1B
zvyz}bZi*HpV}E&8G6?A;V7X-+LT`B(EIbo%Cs*7Zo8|Td<|car3sUh^vw24W-`IV5
z=}21mjqfuI&2saW2{wc8^XPL^<x>Ee?X#Q|%cp!(coXD#dwgw}`qP-&3$(zM?eYI6
zxbXaP#@1@X+Fj+XkS!0vm^S|pPMfenplm#VD`Wc2Vz(XkV%|eOi@5bj<uD*@<Ll6N
z(fgTS<quysJ;(e)j!@1?<y;+5?mzN`$z~~vJw1i#ey0C$!1T+QZZLgOIQ{GD(~geg
zBN{wopShX$KkV%tb+-fhMfP{i`%7PXc@y;{GJ9eEfqN3O;ZZ5DtV6ud<H0>cPhbY_
zIpJmjwp%lHES;D3I`3U#yyIW|xlUd2BJ<*=`s3&Wcw2VHaX(E2cQbik?}X?k>(1lt
z3)qRbAsTWobcQ3lsWh>>&eK#2kC=f`3lGqLf%_NyqO+C}tQp76=Zs!|8eiC}MjJ^T
zrCN#I=OH#xn=52#bHGD)xk$rHy!<5eNxV0Rd#=YkDaZZTO>6h+Ul=K1>W%}O)^;e?
z6SoB8^+b4R*aYRb%Z;CckMTwiW#WeBvCJEB11&Sx+jNm}zEinTYq~7)0;e+Hn{o&(
z()J}h|ARzh$Bk$ND0G3f-|p4%7b%WqK}Klw8}Q~B+_bJeEUf)D>pSV_v2Z;9WE(sO
zlE{HeL{SXmO8xaX5#l4diW|QI{H&uM^eq-*LW`f6NUJyLlo$OYL-a3ox}v1c0L^xg
z#-lIYq$*0!^~s8}T)C`Vai%Gkr707Y+|n9<dF5i~tfD1Lif+Hcd6_fKd95=wOi#`(
zudJ;Rsm=+`S<aN~sO0Q^=}wWJ=W{A~OPtEhfHwo)0(c8xE?_R;Re)CkPMUOiMft+Z
zzh7K(`K;>lk^)>5l$2K&6;CQt{CD|ED~pztI&1tTQ>Nf@VMS>{`J$;yiu}c8<jyUv
ztf{K@RF;(1O)bwZ@>iGFO<l@d6P<HvuAAtb?<=nqsn<=Jvb3n87WDpXP9!pqlM<&)
z`QG%bDO0L^rPW3Ls_ILVi>oSY{PrZXB`%)mG*bqUcX5jIR-pwN!jxG)zqqKPsCufX
zM{(5>Uj-CiR8dvrzs`BxZC|sVH<cDcv#+dYKem6cp8e(hlk54<?9YLE&FWE`8OXl=
zy8T&Fx~QnO!mlh9JvC)YL3V*86%_ek(j}HR3QyhP7HG@aL&Nne&2+(x)11jvG@Yku
zMKz^;;((QYMcVYDYJX`>c~Rx`swGQmE6a;%*Wk{1r%;IM-0l|8l$VxDfw^VXr9~yC
zNQ+OAdJ1NoQgLI7$9kEc*NlKY5|2$J1-CC`KBf(HbjtL?Ke$D*Pr`Z$n<X5${%KcI
z-*6pmt<d>yp^&$JhbcEzLP5ld>e+v!*EdeetgO8D+BAhy3(70Y{gaA|71Ps&^O{{c
zwY(CZOP(oHc+N6Ae17?ol4K+y%2fJOl*=w#;zOvtY%w?h;PwiBe$loylWNNEE=^8h
zSxc+ROKie&imGc$y+wCb)%uf*%ZjR<mtHCaPBg=yT?0g+ep+43f=fyl)-JxRrhIW_
zQAJGz_m@DZGQFsFahcy~raCVwIbCr&my|AX7L}A#m)6ucFRG}hBTXK~RVAf@%TrQX
z=`Z)+<y3@U<RRU>x;cDxOZX}`e03GCF0(Q#msy$R6*9{!WR_RREU%DRL4Na43VhX7
z#Za-Ty5`oB+t_vpreUKt2M{}@pt7p6ls#y6RoG1It}tcFfaK(X8rIfR17?~sWj?$+
zw`yK#@zR;qRZEgZ2CMs`5Md!-SKhSJ#pRVznz1G|^Gd&8i&4A)HVKP}Dkci~yUIif
z7_R)KLl>K(#GDuR-LY`;Y4D>KZzxoT>(05M(E4GB34qFJisH)&g^~c90gclUI{?$c
z*YZ8^0jB;6bikZ{4~3j@@W<bULT<pqheDxxz|=p4LTdnf{s?|R=i{MJ2cYs~DAWb$
zemWF##Di~hD3lE7+k#0RVOuEV1I*bO3atcmJQE7F0`@!?3hgKUBJw3b-pk+z?0E&{
z0T#X*3T*&PeGT8nCmk>e^>Mxt3grN%0+s=~0qX&C0M`H(0(Jqmz6t)}Xa~Slz%D=y
zFy|oj0rUZ`1Z;g9`Vjvx6zV131AYW2z6X&5m<m`1=mx9@%mG{jSP!@!Q27|`fbtxG
zoxoGEQ<RE!cH<GCHGnyQ>j4V^I{<xvU4Zp~y@1Vt4)zCNGGHg58?Xm34^a6G`T{xt
zR{}Z#TL4o5HvqZ;`KN<&01GMi82TH#6Fq<$cgS$yzPUJJ?w+T_)H!0#8!<ewIcB){
zT>e<3_~wK{mr5Rjf%z51g*5V~&PE;C_ahxMM<&fy&$u(OUb+6<?_PSPYdolehcNE$
z9Iq&GcwBRmTZuEvRg}p{qbvjE{?LQnorhk;qI{%d+Q_77Bc0PmCQlognmICk*2tEz
z4Z~LrYfAWW{7>S37%TGSBcFRsC^Q;v9^doQK>4yp=0s%^`m9F2<llrs6Jn5W?8_qG
ztdXskG{EMm%M<ve{}wjM^g5(Z8|la#nFN;PhPWSt3CC!0X1(m~Xwum(`&&Yxz2re%
zD3|<`@oR=p<<VBMMyBE{=w86|N2Cvd-=de`BIi$`P>qd$B53#Gm-N?Ah+i+JZsE4d
z9GN<6<ocL~VXG3F;(r|X6It#$uqAJV?*Li{b=*#5Qz$eJ^^|flM>-osPPW6l_&MRz
zpEEt&Cz&IY8?0V1^&f%s-lsyLm*T-E57yNMdMkYXNz%jl!+n}L(jDD5%*u8WFC!sF
zyxd2g7xiJxNCG|n#g|3Dh3jtF$@;h`q12-h`Fi)_X+OaD<W~mDmo>6EreD5yp?_y*
zD8x^(o~r&V?^Bdh_B?zQ`Qo)f%Ijm7%$JLNEiZ&ZyKy~0dHrmeY%cs(Az%87q0mMW
z!uEiEf;}`OtP=f6`A>ka6MUk-zeaz2fP5V<g+g?5>llrTco13#ppON8!vOTjpsyc*
z?gc%`gRzFf?e-6PYIi8aO)cwSJ?;g)4ssSiuL0U=P`i2p`C2fhAY6nG*5g^w*Yu&Y
zU7v!THG^J;>-a|WqZtR_x#2J%>~#eA_JeN}r9yxDJM}yPdKc((AV2=u0qp=WrN6yS
z#u(R&G43J6?0DyXcT^pPk9Wn=$7LvFsiM3J<nz6U9UYV(zw%X~Z`eOXzIu_*iVe)i
zIPfg=Txo<t_?s$1e-T62wm8^6A0WL0>2w3@AUz)CwGKcZ3;Kou=&aXz(E0o);-8a%
z<$FP={5wfN`ytjc4l&GQXGBq0uk>IjbS@PL+c)9}?b{aXn4Nm91-~&f6sjh{W=~Im
zo--@j&M5yO7(_4R$I@Rr?d=xrLlh_amGWVims20e54W)yg9f#~$tb7gaQ}Ang1%+|
zdI{*w1JLgUedPf3)u7i8Kwk%X9q8w>0=D+pF8TY?-;nbA(mw@#Abmz)t~8K7F3<<k
zXDaB_$L`ncw_MPDpu=6l2k8}{mkm&_M$lVkME7SF_@BUdS%>~9r4-o9Phw|`O!{%$
zjFHZ!_!%RUS0!j8Q-3&28=2lPJbR?OVq`k(2t6(48<=&VKNE!=oq(Jc$dRr<Y2paQ
zb;QQ)5L6MWN#w=Z_-)p3dJDg4_$|eRM)?k`WCox(%77%}Y8%<EfC!k9vW#rgfSd`~
z!4yt1P6zBH@_QhK_$?&cM({ie_y?x^2rYO8Bwj$IM^%(bR!SNs5fm-*_<%5FBiYsg
zsj^a5kAbHGd4wsqLjQ0|?)l&c@iC^1y}*_-+GVCZ!IU?E{AIi>=clt2WiHeaM;DTR
zfGeTFZBQNRnMVZ>>Lywz6qgcr3F&*yMDwB$CJ+N_Qzjwwyl06F#l2i#uf)yaRF-z}
zJ7&2MKZ^cOBANKH5#J>;2R}CA-|&lhUm~9n`3R>e_=&@mZpmhDJWH-DBH4)_W#2($
zCVm&;Cu;aHE-3q9q{Q%9C?dBL;pBxAG;!b>M<}inT2IAQB7S2~CFLl7JMr5~Tj#?x
zEcb09oCI+9x|7Ji0%3}9i54JrrbJTKDsks#AdSzEaHcABxfxI#p}7BHTTSLhH%sO7
zUP?QDwfKo+Dk$uRnM}#V>2i}!+1rpclPJTZICw!#zDtb7vclq+jx(N*R|<*Uh##9z
z9A>?FnPU>ni@yR{hPo)jQD}*k!Zv1Kij<!Kp>7WXDywng<9*@?#l3w$lsg#9;6xo}
z49AJ`iR2LBy-B9tN`#Mcz{r(CBFFKglg?m^jD$$4I*SOs=-YrEmh(NFD4_H-oTwL5
z=929RAn+f>O=;VK&<r0X!XBWs^+e7<?U?dQ=D83E^Zbs88weYVy{zQpM0J+1oEn+(
z2-*G&BwWrmAe6=>sqzXCmh&Z%^I#)P`Ivbw2f{p`63GF=a!LS|3Y=I@Ez4<;DGYAP
zHi=|0<s~4`Q7jK>100@NLIBkhV;{?IBJv1+L^_DH;YaHbGi7@6J*4t5?et1q@!shH
zlJ3Co7rgl9_@)KEX@UQ@T44RTZc!n}IVMc~Kh0f#j2zcppWL->`ol?lBu?TAx(QWC
zRAs%p_rrHBKgK?f%_isba(i)&EA{Q%?%dtjySLY!+4EgfrACTe3KBIU4QU0`K$AuY
z#08Z!lu`~9f=Wb*q7WjXHcmo}RRlGEG+0$>_<X<b`(|fmXYCXvh^O;@XFl)yet+Nh
zeczk+o`r4-eMaaR;WMQ-`XWuApwSY};D4Wb|Bk%BN3&T^++gz)X(MuDTQ<|1`4{t*
zZ`Jtez4Cs$P}2A0E{H!VK60cb$$jsKueXUlG-MOu(TV6YSKCA^GFHf47_x~(!y{WU
zcqf~M+)bMBkRp#vpnJ7DS7Qn_&u&x7M@pES(a)sM$z8p}W@_-lb^6`p`vvy3EK=~~
zO!_~3U6-eFt<CuB@-E3)So-AhK6{P!XQuyypG#8z>7-thcHOp9=VzzT+k{RCJs|Xm
z&_$tZLQe~QQ0OB<9~b(R(C3Ao7y6peZ8u8#LT?i~A@qRIBSII2t_eLY^g*GI2z^}W
zQ$n8?dS2*jLbu%{=?lG0=!DP%LXQYt6uKt#w9p5IJ|gsSp-%~YUg&wDuL<2I{Wsk{
z@ptO0B)=rnBu`|?QU4$rCAlP7Bx{ZO#I1j8;}?de&<UZo(7S}r3;n3jWuYCRpAdRl
z=oz7(7kY4d`h&UK?^)=sgx%a&ellOWvpCh|*Ww)-u*A3Cbg>a*&FchydnGrTAI}%=
z9Mi0&`@&8)2scW(+e3FL*gfaWID4y(8XYg}Ui9&m?~dH<3-}D&$RXs+%#lMn_#K$Y
z1RJf;T>uR`nl76sd=K(Et5$xc9eVk_`^tABYI$2)>E;)@ezS2WzFp?&ZP^W$t$br+
z1@UPbcC?RVIPs$3*YTPz1(2szd=bynCsivSdh7T_zJ2($8*b=Yd2d-QcCgt)EjJL>
zydlzhU3u$uZx!FSXak6jL%E%fyP@MS&&TTsi1gf+Uq`q$lE6=%)75D#1c8;O4}|!T
z$&LRe{E$93a!1ik4%tLNnMB_q^jmK{G4DT#2a->tcjt#}yosR>nxmJPGkVfMMLfjA
zheI}dg%aWW`K|Yp4!<8>_)OC_=|41NGxI|NRQRno?il?pFi&==Kbi1m9HQ@d<Sx8j
zQ|gCQPbRz>4~_t%{%XRTagDxXku&3%e(SHlkHUk-5EI^v8_!;&L)<3;B^yyXhMy({
zG$xsMGcHxG)d9~*c*;N3oyotz*2e`m;mtT$peYgD3z|gZwh3>_e?R;wUK8Got5@@)
zF6MRQ2}k)STO%F**ML#I8au$)Ay3=@5N?+y(XT`oy?+`{@$lzlygMi1`5Rlrt!H}l
zD#f!NNQF1{*L7m}-1?<h;8-|=em)i6*iB<wn#0%+hCiMEv+$t&oBFf2Y`j^Py~U(w
zfXJ)Kx4c@9;vo_K8#a5{-_RWNO*`(==|2w7c=+1WHnZ2B)}eTYs~`E-w4Z^t@DLAA
zpCQS8@i`{EMy;RpZ#w;F0h6dxA0lV$%q^bs>P0{8qkq%se;Y7~x(RRW&<hei&7Xe%
z4#HCYP56sqM_qh_;lGQpB)cZOvBNG&c+;>(zD@e39WSNQzw}}a2o^0~WD@lk{nqP0
zCZ7L>M?C-5xteA@rwJ{z#hnRf=r;Jr!*9M+WA@FLbf`z;f$?neZD9I_NpYF-*Iur%
z*U^_XN6t(z^sAK<YRC?R$DdU&fKU_ulPR2d$r(y&#-z<2^7BuUK7ad&d6;kpqW4r6
zVXr6tO>>;RUelidIf#Fz-i-dE{+*TnQOV(HGhx!vMERE2%aUPmPY&6AJj2NGd?r3@
zg_-!<qWM=mvSDcD3z_`eqWM{p{~eYYpOX01mKk@F_&Y5#o+R;iMf0yD{_bcVn#A8@
z8NHsww_8SEC-G~d`D+rt7JAi=L;O4pskTLL#yy9v9hT9DNqhwQ&5lF-JdA}sqrc*w
z!&nM5dMSzDU>SXr#CKXok0kLMp%-d#h@XdPc53>4%y-zj*)sh(218zVS*D*R@szN_
zllsTotp9vH&PUJMyy&loTbOagqHQH{M=P1Mv`X6#;_d*yj99e2Anx>n{D?){0OGF6
zFC!LhKZ!eSCyiLN?IiBb;AIGFtVT|9;`v9QSD_x>f4;_u(F<S0TUL4Xa5`eq=j^!q
zXMP#6=reWP{k!0~7m~QH$PtS^Gshk2LCR<0W{r2m^G@KDPdY+J&gdOFSxz{euZ!cR
zT(;n*T=W4gt6X%lp3>cvbaNb=Wtn)^Q}KLZKs-hs=reHKJtgr!Ga&v?q~ibXfcU=;
z{3@*er2ES|pu8#nxn0S0$?hS%lEQZjo?b_DfKz?uQu06_9IndfzxPT!X5G$|`BTi_
zuLlN|>k;PPVVzCslP@#gUtUTls~xwPzh4hL51c+Yr1j6g3ZB+i-{*Mx_0KQ{xm<=m
zxd!+MC6RJdjBmHL{=#M-G#|Sa;J1>Tf6nH2+LJ`?;Q{f`MVKSR|8<-BOgyg);QwjF
zmzAFi@LQm@%zkatk4^|~_IVrohW@7xPIpVE%hiP({KkOz|3LUJikx$1@$(qR-_y6J
z#N*<Ccy7Q%iPF7J%4_C7HQ-clt1C9gz<c5FGz7zv%k`wj-^afA;tD(`&v8O~DYEdN
z2cA`L$AI60esS*Q8b?!+$bM@8Kkb*zis!Ef;NKd6UxP&fs)w4?gG`65{lKXnYWLcl
z6W-&8r}NA&nTrz`Y$KRqn0cA;+Q#^H>r6^MKg77<>e1Q&{FA_`{#T#0nR!xzJRtF(
zO3Ceuf}eR*^UE~Zx*CQgm3Q+i8mGNT<ZcB{<vsg*HofwG38H_>{PdJ`|3Lgh;QwWt
zJ>D(e=OrF<E&<(#-7YLBQhQz8X=^}!e-3yqf#{cy0H^Yra}uV%+y|WHvtcvi^=Mae
zj{)BSeq(n*?AdJrr}D0T&E|L7LqzU-jO*w4@AW7MrEAV>n0mVdc$U2V4Deh$0(t&0
z@ar=4ScBu)Ze0|4F#Yb!1L8j?{F^EH*^c=R)x(+NHmCRjDc6S#{;<YP{ht6%^-xL4
z$yvrF(=qaO;eRHT&;OG63nz8_yssSLKDS->U;C9B-Yn}V|6115z$x9csr+AJe#rCR
z*vx6<`Nz-!sa&T{Yy6_b(*{oQ*s1*gj^NfGYJS?2M~>FDD4yCEY<}l2-+1~d7!ZV;
zb27Yl7I+mnwYxdLb3KQ`$8bPd?f$F4cMRQZ9eueboSOXR0RFEKKlIz#S2RCe%R%m&
z1NdJZfDfS~P`XdNRO9gG=UU)d`Mi~JjE{e5Gpi>7N`jyHs>XK<?ihZ-PYT`z&iz8#
z>qf!P2!BoZrC8P<GcHUqat=7R%fmLKMn2z#jz;-Ed!xqL6F+wVCw_C@$?&_tcc6c5
zNj(%K!2O2*_jNpfCip*c{21>Z)wrk)>muXA6eEX_nXLL;07&^c+tTrTQ^KqZZvVN)
zA%5(BUGT#1YMk~QllwMsl4ncmZBE115cX_PJ)3h_lY-v}oYFlb^1oZ+83RsoRXAT`
zPNVm#z>!3sYcY;=Q}ys^;MXCZv^;-yKs=kkX}sS0ml~&WR8o3%0RP+3p(y`_FWSs`
zi{yVd@T`391)h^Q<K)Ny{uRa{Cn-7n3~)-<zTf6(cuzi_iZ|%`UzPeaa&<3o;;%fQ
z`OO~pKLJkV&AnRVcuaXONIaWTpR|{l-1~NBmund~)$^(Us4<`UX#uDBPf7fz#N$gG
z59Cey3-76h-;=;8pR1p;nbXv#jgFaBpZ5SK{xhG^{JiHMA)gz-{{!K#h&-G1tq<O$
z<&EydU<W&c+(&?CrTa?*@J)%wl6Z0)i)FnyfdA3}eCN%2oK4H^j{~Rt&|#Li`&r;r
z51YTDadRf(UgpQR^Hq(Dcv_zo{;d-l=be0r@UJ9(JSN<85>MLRcqayWmaE^kng1%W
zyJmrB)!X6#{C>uf&(wJO*Z}^&8-V{{0DjYZv-9%-;G}OaioRWzQvU+;V_q!sZ}58s
zKetPlm(C86JH>d$`Wu}+R7-Tao=+l9PbWTS7A$-qP9QjZnnou2p*{W3R3B98Lq;ui
zaOUv7y)#M$-33)v$=s=9)o^A~DZkAhYT47Af=%aYtS|EZY)<y^Ol4-kq@j-KW5_eb
z2~OJ_o5n#}$Le|RNq9EhFm$Yp8wBe0%*1+E41j^Y2AfmSmpMS1RcrB}(Mi*KCk(<a
z?ukI!>aOWIHB+w6D`hpj4i0Yy_~5R>mZ~>#Fg36;Af*~@wbX1cxJ}i-N$o(n-F3<(
zAHr<pC#Lq45OW5TqVMVYTe!ma?aqd!zo$34t=0yD#NPO1bAnZ=19N*1Im(&ckId?v
z%d^PK;R6S%&b*r6yKlxJ4r;O}ibfI$LL54`)c#+b-Fv7!ZPCYfB^U&(R8R`1pxY-^
z(_2{zmzCG)v^$l%>$ZJxro3<Z=utJAAB&oMz!CQ1_(86vYPY4YCU7bZKMDk2bUPfS
zcAHJFIX8Q7tm=7<ay2ZDcHEUlyM+_^Yku8xoVB@mWvzCw7h+M;SEYO@KORLjApJf2
zum!>QI?ft#BH~HxzwqlQC*oSdujjVPb>Kxh#I3@P>xV(P?o?g;IMYRE?(S+p`BCL|
zpbq0&fEEr~C#wLbvI8$nXKTJJXZ6!(-_^0mD{+ipDO|BpSM{LFGGZ;b^%KZX-K1s;
zY?^5P2=X-^FU~@Dad26Zc_tN7oG>lC;5Jm<4Z^iFlP<!5x-FHG=cuPe{GGXKIH3=~
zmwi;E@AS07C@$`3HOsSAwMVYoKx^;O9cR6~=#WD*O-0v;l#d3TQX%Uu5gm!Xf1*6E
zX8bXPouO7PsF|tgQW1)AoUaves#D4tqf124s>C(Uhuo=&r_fi9%~vm5Toghrwc0DX
zk5K=+Yfa6Gx|nXdK>%H`+6{v~Ntj1n|44F|ZiYs)l4=@n-A_xLq8xNWwdnfIbQWgs
zu7=}vDmug(!hA@b^l?uBs|P0;b-q}<Ql~1Gs>}Xj=r!itl@o#E`*VJ^NHwVI@Bd93
zqgfLTZ7q3W=(oJ~V#D1?RS|D<G9?%w@EOW%k(F6Bob;g{XlvJ485$RQBUz~s9q!!R
zcU8;vS5nCzn?Ir}FLqVP6>?K5R?IyM)9rSLZZe?uaQw@3zv2`r-04>1@M?(NYtA~x
zy1=Pa>S#aGbWs~JpB9gbUjYO8U{DytD7@J2wBQxu1|U+S3tft;+;Zwo^yN|kXx;5J
z)IzgeKY=k0mmzrMF#%JC^jQem!*Bwg&W7=B&9ufQotf4+{uN8;#;wuGDS9qa_iaJ>
z>xL{ARxwz#6nA1&YJJ^RF7~PQcLv_2heVo9h0(l)&3B;}l`eH3NojO?*{h%MI{Vtl
z_<p`>$>sX-Gn;pRfAf~->Gl|7Da;?5CevY_cf$4b{Q$A*X~$H)rb`*Bv^u8nXQ}dG
zD58eHqPo~m7?o^55}7geMi_<Nf(z|tzvdE+7!yjgIP8Ix^it`w%dQeu#bxztaVj~S
z@HHA46;N68tK&*qCK!VXUM&{*@*d=^y+P`n25-sY8#7a;A}ht2J04YLpLtT8Ao~Qp
z45rC`ZJDrL!I%7`8n3t~E>x^Ap(%B?nSVf27032<+HQk};}5qx&4%Nnz})-%3EULZ
z?szHPa%Cf^2(v1F;fX_4Sd=Q{yu9fuTN`_C&aixB7VV3!zV0sgYsKOoJkw~)N&tFj
zv5mHdIf|{Zs+h;rNrs^T^&4+$>NiI5rkamC<y)c7>P>BnUQPz4A`ysMUyGQrm(A5f
zq`EWoJ^BYrOgI@pURB_IY;wt4q07>|^6V7Mn3XQoCsic`s)b364<$)gPL(Uid&b*w
zbfn{?fMI{ntLkKLh>sM)M4NAJ%jl+5wW2<f*Hxi?Fw(Hi=R%nKXA}zR(@b$N=mZ!|
z!8D*-SHrP7u1iYxc3hf-aR#GON>9P)or*?-z7e_+wAC{DHg+%9-Gv}M(de1yG^vWj
zY!CD=uAXYdTeZY9-vHAozS~J+?VvHIfRM<lutJ(6s-gZ#g4UHqu4(nLuI~kP53SYk
zf@m(2oFq*_2g1OCshd_1#mOkQDJd_G#g&u?Wg~;?YPaL5HNO*fT^N%Yy(=A2DISs2
zlcB5$fVR6411-0(0gvmVdZS<@T%%)JEt7dO-%eONGi?m*gl9+|hDQj?Xir*;+<D~s
zuR~a*psLsNpSGR?LXD8^oHcu*1d5}2nkC#&Yx4!=R3}u0(lOSy9%}Ma7AAzP_F8)0
zCTb}v-Tsnk$kb%1=Mi<^>CzSZt>0JY)NhmBw!~rvLk0Ja8KKlh9t)$)Sfd<@4oqdw
zt<n%1v-jiOdy>`+a4Vf|&Z<AQ>Uk$p^4JqbR%g;KGqgU0Y$$C1kCq{3tL^c{W;a;w
zj{<r-J}l63Mc`j?WQbYnRLx1X!6t61w9t|1djXx6)od@BRyWgfC3nWD6W)fZqYkva
zS@k=CGw1guq}4w!0I2RtM*r3dzp&^xoBCc#vZ2R~`HlFo@)Wnf8|ZXzh4H-x`HxM|
z$rOk1M{PJ@lr0r$E<$sXM7Pc8trXLbI6Htf+d{Ck1QeB8Yh76cbj`DhjwV9ol~9(a
z=Bh#V{0T#cRsxvmW$KCXded_|XoX^l7mKi96ha76El>AKQr5xNTf4y`)PS{{0cNbW
zOk)=s!nBN<M62ei*YRCmQzH>dthn&%i!7YXR*~C83uE@hODPTs)MW)uFUzqw_`zy-
zrEZ2{(<i89O@DeTq!acD^snP&cTy6pHjRp++#3$WuRX64Edq5-V*!@JumaF1(^NPh
zty!YKxw|3iB%B?+8tK;QUc<;(!cnKCsMbqG9`vJCJB1cq#KI#-PnMQmOr@nZOiJ7J
zEZvh@_9WFq>k=tFmDl5WRnL8(_^Qgojg=dif~`Vxwma^Umy#P;9n8|dPRUc~1-h{i
zlVx1kn;h&tK9L23aeZCkvfZGKGi<n9Qmqw|eE*9sTt2>r_Q0wxnn~+AJvgDjNaN#}
z<)nNffgUgmTRGWEv}_c7fQq8U-JW=+?UZN`M7#0|wl+dYtBJ)+Qz*S`63KWJ$vrD4
zAx)T?2Hh5E0DLMM%@-&~ND+<iHoc5Rgf~8MS>&kNqgZZq`WNKMU=fLJ@GO-CiHuen
zxXmPAtd&TXf|KqlDkSMmrc8Y`wpLc`t@zWNQ7pNMo+Y<Fi_?KIW3I$QT%>3#?glPY
PZp4?IXf@2@>{<T<6(d&a

diff --git a/tests/Grid_main b/tests/Grid_main
deleted file mode 100755
index 97e4f39ecb73a6d05e11032820ead9e17027ee72..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 143261
zcmeFa4PcbTwKu+jMN^1An}QTutHEyDP4Oe~i$rM6LW0lgLZe1WG&aEiLW?9N!D8j&
zV#uveH?e8UZRx$VH}{tIax2x?a^E7NYzQR54-DW3g31bhYz!X}6oMjof4?*HY@URy
zQp^9|-v8@Do|!pwX6DS9nKNg;@5>8L8kCgevVVhI-*VBF|D_xOsjthmU~ex@jw{Rc
zIaiMB0@r!2Gk`S&e>Pr@w%KO44rZIlz~yf+{@nP>8blY(92#-+JaM^@g8JF1(`0(R
zzZ`8<=6l-|rfs@hY2r^PX>EH|`M13~P2jfaAu;%6d6a2~BY6zht;2QeaJK2v^xLML
zKSsyD0_|TgC<iXOb7;77<=boO<X!o1R!ygU$7)6Aj@5p_v`sG=ho5cQ;ch@U(s`;s
zJDWG^^b+x<w|rF0iYk}u(mAv5xax{a=S;tN&g{AK7hJp`>#B>dy5f?`@=Gq0d^3LH
zUOVMx325_i6kMcv1pb&6`ft2C{L^o3zj4!Jk3{}Y;I~(Yb{zkG3;nY2$8cxjZ}Ifh
zS+1ndospd6&KZ;wN}}I5{Cx|5ZHtF)TVD3Wrt~k@p7-EgJ#7a(*ZiigDCgbRZu?o9
zFYkrmh-E1gU9*P#{x|1Uthp!fz5jzC34cEaai1T+Hw%?+Ab33p90=Zw3=9M>Oc2kK
z1n?nX;z03?On^_mrwu^=1qt9^O%TuK1bnzQ0sc1<z#Eb7K=j|00RPbh>8?zGe@z1T
zMG4B~@&tVNMS^(NCV+oCK|G5R<oE6b^n8$@eBVw0|6u}p>Js4pIsyF43E&SVh<|H>
zd>u^yADN&W_a=bPPQcHa1bkbcfX<H+w7b_5(0NvZ_=5@JACw^eXA{7WB*^cJ3Hbk7
z0{rv@@b4vv=lTTn%uJB4oeAPM65ziAzKwQ$(G}S<4V-nk3gzo^eHZZ%p7TaucnRW3
z1-#)+1-JXjTR|uBqpvG?NXN5X!yEd=GXikN6VmadY5Xg7x}}<)d<}mxNy(r5Q8^l-
zj9MK}+Fq5e2YCGb8h;|M&ds4S{`-Zd$EDK^X`ibACw{aaz6<?hW!3buF=N;#Ugat&
zx$~~_xh0iV)8<u`l(<T+ySwDZ8F$XEteP=z;+$!fl`|?`C4qv;CDUij)8K;16X%rA
zol!9DjyW@w-@ra46;-q5&6qa5<a;yjDJg@k3h1{^xvBCBq)?HWQNl<otLBy8lXv6w
zH%$wa<$;*81=C6@=Fgi^a`)_cRr9CKVTkK*s=BhQ>Yj=jK*_uu!G17%`i!|Hv**sP
z%DY>loP9;bj2Yh($4yn4V@pbqa`_KR?m*(x=FBNCQ&{C!&YUy9a+ZyDl@qOOT4fcn
z5FGGwkTql4T}YyG#=NRP8K9YEv!=}}shT%!c2y<l$h#4+%kP+0IlIip4iuCWWR&FH
zbX7_I^*3E!BBT>(-n6;X%kM&7@19*YV=TxQ!jWZg7U^Y!(|66D3+^Z?`-jLzh&yLg
z-92rN&TLiI?D9(Ft#ZZ{0k9x9(}rJ#oKBlFBMzvXn3<P<eOa8j<YeX8vhuqsrp=o{
zAK_j9^hn`Mkf_--r!p|}RP&!vQjjSOK$2CG(3}|yz!nrN%6;}#v+tZ+j+Awo&8y72
zakeHQOK}oxpFQ_Z#r%P|F0~{N*&_GI+mZlS@$HHkbMGpj&e~OQ%M?^B)S?B`?wEad
zCaXdjYjFvy4;VkQ+^%GIpf=AfDVtVRHcMzwweqrzlAB6q+*MI^PszL)6=)UZ^QPT7
zgJJGmu%M)3#=Oe%xzpy%hWFhWI#}ford6OeurLd50ZmsrvvSA$nThj(5I0p_QC2>`
z>g2GfeF?%UW|q&X%p>#W{@^YN+$Y<YWtNoOeb?*(vW{XXXNA%gdphfJc>$U|JbLR@
z?q#}+$}g{&UzNvtQc!@b{Z*wj09h!R$^ogmsS1qOjdsqAxp!90Dw#2FUirLu9qn7?
zQ2|(orq9X0o8s&)CCTQL-|4FS{=6y|EOQ9Iv#hM7l75-4lAGuLVD{YU5ImLh@0vl+
zG8Jg1hy+(fdFAW{C3nrZ3x#{<To-bU@+_hFgH)V3ZT1{j1!`v1Of8|%SQpH$a#ha0
zQ|e0Pj4Btx(tv1XdBRmzagPh_wrciWGs<U9pLP$ir<GOBE}!eVV_MnwAW6z*5bv(?
zyJz%5N-FQU>kbqy5+adWF5fk6_FUJ@c{66XW>;3spIcVtx_08klFKj2a9tY=Oqf`5
z*(H}<a%J3}d4;kT76RppOS1d8GP7J03k&@d0wtN3T;9h=N1#JxW;#9DsSLunXpnqq
zqI<zl*#=|eWZTgF!cN~o7!ND_K^Q3|3kKbv6TiWL5Iz{A*g>wL0;hYBYZx#HJwrm!
zSN)^y42<29Tru>i_IM)yYxfLuA$%+PasKGv?(m=g#o0*kbFSy&{-6tXhR?v?(x3hU
z!_#4|T^gqF2RrbYroxBevhGVq7L#3XY5ZB;zl8ft*L&K%{gLm%2HFuT)2I>rA+EEv
zdz#?8UFT_cRr=4Matv~fft$bX-KyWGlbo`DwjP4WWqfJFZM}kUjydh0t^W|tF{%Bt
z^$NmsY#{AgpCLTk0k`!W!pA${w!T651P9#Ke+d7k1KyzZ6T+`{z(++@emR!bs*d=k
zIpAP${O5DPRUK4b84fs#KmN;dz?ruF%W=Rp>&44-z)6$+%Xh$S9g%*84!E<tN*!=i
z(D-kb1D+Cxx?B|w_*oA40tftT2R!6}8xHsb4mj(d{i}DtQ*9v3We&J=Jlx=b4|m|N
zbil81z#AQK&Y{@9H4gYl8wfMvfPcXOZ*#!Ub->#l@Gm;xI~?#YIp9$T+@7nWcgz9*
zvID=%0YBdX?{>gBR}uf|aRKXyKMsUsaKO_XaE}8%+5sQsfM4i<r#awXalm~JxYq&C
zaKOLnfM+@2Uw6QB9B`ilZaU!U4tTx;{tXAb&;cLgfR{Sp7dhaw9Po=B@CpZfq65Cb
z0l&ln4>{nMI^YjD;FRz7uigRAw1F^}IpCK$;0+Eq=Vt8RN(cN38wj(}0l(4#U*mvZ
z<$y;V@UafKS-mgW3=es~ZM4g*iBt`V{T7~P^~U682NJA3SMC8y&sX2T*SH`DHo9jK
zsUzBh-&c3iO_AEsF79n~Q<Qc@#QhZAgXwM*_XfHtIy)M~y_Rl@%#M0-|A}sj%8rn@
z|3EjZe@BJ5|DA4%#*R{P|1;ebi5>ak{u$jv>CO@NkLjjp?8p%J{d7|#cBF~>hjddE
zc6h`+k8X;<4wtyUOE*PdN7o4ezIq4U6nPy{ao<WeMO{a`xNoGJBCaDM?(68LXzOSc
z_XN5rx;h%feKp<Z&|NR?%jl*E>j;Va8+222bySG^Lb^xMT`KM`(oGT8kuUDg(@oLU
zkt6P*bW>z?WQcnZ-4s<FY2yB%6>f^C4v)AG(M{3R;S%>dbW<dCbbZA9@1Z-5?x?tT
z(oIp*(Jt<7bW_B1M8y3R-Cv=*QQRBorby{%5cgWTDM~u(#r-F`zeaaR+<%~(BBi53
z-2YBDMM+1gxc`}Mija<casP~NijIyPasQZZij0m7ao<ljMMX!NxPM4DMMQ^3-1F$B
z2<UK$`@3|r>34J;m-45ZO}-;4?px_*Q}1XO_l<P3iFZWAeI4Cw+8vGJo<KL7bVq}@
zucn(#xuYKL{PnBeoaTzvAnEm-vw4!vwe4udg~Ffw^^yJXpAq+`Q~YD!2h5lnF;a&d
z+=f=&I^;bV{H>2J^ShpGizAwG#Mcz!-ad#?8?mtCd|=MJ#JFII#KZQVzrLI4zX?(m
zeROL2$e+yp0c-1bOlyx>{YF=QL7+9#{%b#oX(|<?HRrq>SL`o9d~c?4L9WZSW;e2S
zvw86j7-n_%*{1d84+h=dyoL!<FuiyCdV2LljSE7At)a`tZ|WWleAC)sHobF=nbdA>
z?VgXoYl&#qY(rqvYEj|$=<uO$KDrrPnV)(yLdF)MKyPmu@*9AX4mUG@X7x9BGy$RN
z{5T0CZeHJoRWIQ8wBO!wFM=2sxH>A8d9#12{}%tv{(_s##rxKPPu4o`s(DD>TJ3G1
zS*P7~${pLa00WD!%}xSFvl+fO*&H)2FY}0LePlL$G-#%AFf!A)Z;3a=psh>1^)UIX
z`jWW@3G#Q1nZ3#QQ71APyA;I}uv&shivegp05eeMe%Q2BU^9HV8NMmmOh05=+s&pA
z2UR5pt)sE8q6C2U8=^4?$(#lANjDN(OX(fEa3%<?gJcg_t%2};$$>G~WbW!uXFkJ%
z&Qh5CokXWeI)8-Jm|bhRR|3E+ltkSRkpKn`gvSQLQx%;rkQG(QrnNiv4<90AokEi!
zVMT8+y~}Ls9)ub*_y=DQqRsRcW`cZ(T{|hs)%kgN4w;t{sBZDtj4=_K2!?OSd-o-|
zj8s<;#p3!t^RHNF)-4r2n02*6DtYzTWAtC{Z7ffWM_M0UPyZ!ekKnEI7Q&SFnBzf8
zrqm~wtLqwOIfFRhTNX_I2+M)JWr;VRAS3l!1Ui%81sXNvKz%@=E+dGUYk&oL>qbtD
zN0xYx!$ioHfT9T{n`K?Gs+*HsNH3Wo@FS~ej|Dd=qPym*n4&PvvQ-jQ=4P{IzvKY4
z=7^SG^$pXCnN3l*SvMlWNE+`&r3oR%*p1z2{hMndrjc7KVAHDgGFr1PcfU)*B(*U^
zE~HDd8-7Qc45Q{rtVoh^Z2*AA7!3ZHu{8|;smN(QIsiFLp+7l02m?9oFom2;H8~9c
zfSd>no1A*$elj`9Cv)eDfyiMBHaW?r;!Y!BEIgP(&cXj866yf}AZHm2o19t1{Z!=q
z?#KY#VG8}pd4MpG6M`w^lxlMF0RTA_Fl=&ii2KRp3{ZDQ4nz)9u*q?2?vxS+atdJz
zIrpz7(h{!^0FYzCu*n%k+)qW$kKP}EJ4~TJIa!2(oD7&k&fz~Ra$Ep_oHQ8xQSx=&
z1ry|~tZ=#hx|Y@b-T>q<h5qDt2m?9EFom4ynw%X-8{~Auu*qrr)a1+_h#aQSpPU$B
zASVh_$a&#UL|Wo)1OVi;!?4L|AnvE)&d$RFaEB@MCnrJ}$XNqZ$eE<cc>n;Avl50)
z&I00=Esq!0u9L;!j)BGC^#hT^6wEsJE{7CcMi8i}hbh#od6YQl4FLc(AsGBo3g#0R
z)KI>&>YO??PaGP6IZUA+H5CMbnprS~ns0y_P*Dgse`cBLU6^{L>g&kI05UHB<ivvB
zoOpE1To&SEj(ki{=O{08|H)YzV_G(YbNco(@4Va7gCI4Lppm;uI+~!hN+L#{ZX<_-
zNipWdWEwDYVET`2LPzsKBOsu{0|+vg0fRph<RWsMpq`Tn`nqZDG@JIE(U+Lpg_s^8
zMs<ckD1eOLY}$K9FzG|K{4_dYj)Lhw(h5R;tO<$D5kivTVH2_gAe)enDQnO%q90?n
zb`4P0OhNPI3@t;u2?AfbU`pYfqp67i0JcP7*wi!<*RFX36cf||yO<(<IWv*epp>8v
zs8TYmRbHk5YIHl0Qfem%)U?6Oky3i(VIrZY0|ex(fnk&L0PMlC)v6)vnp|_FU$gvL
z&^m-dLfv#~CTk9YlwCp!v^Xd%EDXjTXnHF-kn|eMe<fpv*#PrMGm7ZkM@Yp|Zv_CL
zq8<i+Buo@WkX*_zHZ_}VY6h9D6Ffw%`bN-#mO4;N`5GzeuPZU7>+9S{B#2T5LI9CD
z3t&o7?fD~;tXttV0RUCAVDRUpDhsgwRQZimw>Qw5>s5qlHlZVSvgxM>1|W(Fflcl}
z(iXC(kded8hbaX4KoAJYfm;YETN;;A+p4~Tvgnt{MJFc`^rlHv7a%J0lfv)Ye-e>F
z?<fhsdlh-J+FJqrlJz1T{V&L(>CvW3n_b^i@TfN1wHeW7BTO=fDl&%#QFN%NTCpux
zV%&yGL0#PGzXtxX?_WyK-az#_(0`8U3!kOoKkWx^1aVAcwj|OU&jT7h`s;n;Db?^m
zzj!EH8PB6%>5EShC-JBCh40YuG<^YMy(d|UN%%9v-wdrC?TX!osv1@;4$bXB?sGG%
z_lE*?d1;xOGq+)YG!wu09KlSZyfA9`%&uI+mpn7qsAzr^pqoSZ%&*61Mgu<o&^Xhm
z*wA1UHN)KYzTYTnKV-CTfa3sw9SpV=!75&)*EV>)V-&{Ve!c<jtuy^b#k=(0_`Xrp
z0^fF{@C7(qb%bv-R21-yOb_GfihUBRQSk<S_P}SK#8;lz4$p2zdJL9`j;`A%{0JfQ
zqJTvaa<BLSDT0symgjW=vI9OZ+ddoMv%@HSg`O?|+Zg9&@l_<Y8HG=q))uqrM4DN5
z#oZ=al;0RraHR|1g;#p;4PKe1+1#_iOgd&R-c-+0sE%Z0A2MoxhpD)Lb%G_Z5_75M
z;^r>+2I{WOU?EJl-nZvyK*0j!s=Snh=68XW<!R*S?nn83UcOH~Ix`R4W5N6j=R9FP
z#JNzIf6h>5m~*F~zZtn~g}IBI+l5@R#9jpuS*_Sl;5z`mY838(y9ID5M8Z1&-((aL
z|6Ote^F0-t>Cpy{ca1{uu)I27rP79g&%r-J|7Q5N(X&)}lF>{}JT;@i`10zC-oQ`U
z{!I)FUc-N42*4doIv$z@h+<;siS_UuFx`m_@QESOeRiZwG-ecbFj%7w)<4B(=^2rb
zNHHFiDcV6%ED2IXS)m#lMaN;rvx#bmT3J33jFMpeQv}1x2g9$OS$y6O$pq0Jw%V1g
z>8&$Rxvm(CPw`mRvpHka@GTgd(X%0#)Pah`){bhHGfqivqxLb(Dg`l{K)K+rUek?=
z<}x1q3Z~oxiz9TI;Z@!kJo7T&2*S#K%d|T4GC411cPT;Z02HRlR!i(UY%QB?bp@>`
zW^B?7AJl(#!FBE&baZDJV{(UJJ_f^V0O|11(XJhW9@UNv`gCbW4jrzADoj2d9_=Wl
zBTYLh=*SSqk|7~x1nwN|W&@);U%MN$yHvXy;r53!{5`E|@`hqDkhBpzi_o@s6Iagt
z*+-4qGf)+KMW@|CcP0m+El_E%Il?CT;BG8ynBZ)L91GV96V19C(lU==GE~j6nswu$
z0KDx+;ihEaQ;_Asw>T>e-|1NyDCQ>APs|XOf*QCsC|AC6H7i$+a&1(u3~`mu$;tu#
zCIw4Vu&S(lz*-cHTL6*D!mLujS{00ygEAyHtAb)TqM(#Lgw|$-XoH6)DoQ6@?+c0T
z%C$kcBFeQ3E(l_L_VaUupPu(AkRtNk{Jv?u^h6qvWqz2++!;k@GL=2b1;pF;C<~4G
zN5$W0_-egO2=-@D_nYV@H{gDX*oT0<)EiM4F9PE!qp(eCfkJ8~(ub6V+Y#g?AZ#!S
zo1ipYRL`o$>Z4Nv$}`SX$U{lRD1QN_Gj6gHAPB$m7hyVMZ9uHKM$rl{Sqsa-LwQE~
zo3JePlC610Q5PXE<{5?01G?NBRnTLE?gn&cu93d{B56a_n=;Jm)(ms;M?^Aeo<lET
zmYpEc;$JuZKGY?&F?Tj$x`|~wn^<-leRLDcb~dr>9Qx=cmhEg}*`@T+O)T5l#Ii%A
z1D?8xW!G!BZerOD+O3;db|c(oPm4dCWiH+jfg}5bQQM8_;+h`g!AACUXfDf~%>^A|
z=C)pa1HI{u=uOW6r`?&3hEX2OM8mmH)dhGKWTwHn)SCmY!&#h(rgYzODO*&)2+ALA
ziX{qHlX9u{P_aR|SX>C0o0%i1EGvl?4G7Ujl{1)G3a6@&@ST%c0Vjq%%6CO(h!)i7
z6`5oVZObzov~6i7nStbYF+mJeluMH)SZMBw7zrIGG^8*$(J$AFB&BbM6~vT3qnZrA
zho#^D;88A$q#+E~JAg5%9iN#|1nX+RC-OcTTCKD%)W0o;Z<X-HD2g4*HQJwrCnTn-
zyDju2o3!U9qiD5o*eF8H0UWKV3sD|tyl=cu><qy^$Viw$G+^Mw{ODZ?>Jm^ZTWo_^
zpJ#gndAD={C1Mn<6Hc@FkOlBOZWNwC<cFBTHl(mZgn?1?mf>6O4M`k3fwtEue3QvX
z1aiRe6?-d;BKmY9_Vb|*8b$9KMFrkcfDRdjM-d%`X~p*UA>PVgF^YE3{KzQ!00_IO
z>cP1edM3>i3j8(_L2KX8+u~c$RL6@PNnbIZEgy1aylT2=w$&{;=Hi|xNvf%mrh7>E
z0z9oGvgu|Lrkif2v*~6!n{KAF>1H~cZl<&8W;&a0rnBiv-DFa_=}HZyTQ^;)r*!M4
zE49^xt~!JT?H}hcYAd+R1BF0M?);jn0++UGc2zlMb0W5epjHjei*h#Sh)kfY7OCYw
z(t_dlAAW-sjsId4)OpWG5|jjL`siJ=x+4UYB`tGXJxVVp%JGKb^PoLJaYLa+-)GZ`
zK$dPItQ`%bd>=~X9oFDYd4><Q80tB^UbekDb>%L4hu}Sr0{abusL35;oE_+>%DarB
zqkhA;$0&Le)!;qD_qtK^HUPVfqE`VxefR(Y_rZ!0M8!^qdk3S6W+eB*AwRp8qNibh
zmG-BNqV2H1NjoHcx<7<ZpdO#u4MzI+8;$g77{lJv4xcDIx{UNx#Srj(aTBY*JviuQ
z-O;lu!K5xuv~Voxm^(4U>qP0mu$U`b)IW^WBo5?76~PQc8)YoacA7OkRl{O;?8d-P
zO29084U-kSs=lVC!1|5oZ+)%bh+GEao?p9M)labepgTT6Vs-JUa^v{ich-H=F%q@5
zILD8MkHg1O7N;?FKN#6*OoifI@jle!!!Rivp*a{uo$z}f!w6$4hb6BVQ#lyx1Qtu|
z4PxyA)+-W2;Q@uUQ(^4|)*c{iv$5h5BK>P%Zgq=?oE`U1q~56gE_6Fg=3xZ_BREVn
zD$eG>;E$Z;90@WclPp~HqMK7p`PwX`S%VeO&Z@Iw!@w>xyj<c5S2GDt1Q`!rj?OM~
z8#F=r#mCcRo5b&PCqn{4H!_=g(x@RqG-UOSsHY{E^p08G<jHPX*s5ZE6-&7iYh1r{
z##j7T$5)pqJ~R$@GI+`%6-IB;=FpWu8lnWS?B^gm<zvWhRAhfml9BazC%xza#<;O<
z$;mwy7;?N};R`BgCqxAlgarBh<nRv-9RB0;XN*ZnMAnmvtO6Bny<MJ&FNOTt3Vv;c
zNkN41G2{dIwbdxx@G+D=uA)qmC{>L)UG)W1aHp}pEShH5=#%LCcz&LGeF>IzXKv1e
zz`~s72PaQAKJoHE!l`l@8}3qRgPD=)wIZe#<)Yf4d{#?)BYjXR6VzuTJtc{l6b7t-
z5zn|^Fp`Cl09s}s={Yw4Q<M~3FrcKMsU-!(#2X;ouIRzC$CaFzoG_m+>=-DY;a?7v
zPmBeI3@J%MIdi<O^J`4qjEa|XhB37*Vdl1<p3Du2or%op&SB9tY!U3&$wf|P^_PFD
zJPe=vY$IJU8=6y-M~UZ!yMqwG<aD6yUlo(6Zw%a1HNG?ulV<*6ASPW*?&Q)Zp8gp|
zx_jVK$HI{G4=ln{EcF5U2bB5&Q21A5X!nZ)W$2&&X`l>U#0=&9H|OX4zb-$<fce?9
z^zS4;Lt;}vpjjrGt*p%@sA$)|Fi^RC^z*n%v?wDpA_9o@o(fQ0J{rD3M*0~&5v7G}
z$QV~6as_*CPJ;##unwY<S)lH(sANBUexS@`ATuY6(6PGt_U7lmv;7Q-Jq%h;p?v;h
z`#@Q`{b%v=F&@OWHRLZYnbOdDL}h)ZNU5y0e)0cfeHEL+zXL{{0zap28;GB0{qz+0
zxv=p6C4Qa_Msb8(iViD2I^qNh=h2r@2|$&rpow)vPFrX0;$$*s4D|k0v#gn09L&0l
zpu4|@nXlwv(gExRz>@kNST_W%w_|PZ_eNwkz3C3rt@4FXpP?b0JidTtZgs5TwomEr
z4BxdOV~jrpvmRzW%m$bZFdJbu!h{m;kHBn)*$y)bGYYc{CdO@N8e@{|0qWvb8K?HN
z&HoR~@?lLY2Z}vr`oyR{;hrBdd$0>uF=U*`RipM?47k|Xux?xbU62Jcq|FL#mcr!r
z$p)DGaZVvRgVPN?nMc-!kek@gpNs1l{&mFRAQWTKFABH&ae2SZ<$AMhlj84r7%3NE
z-=g&*7wdvan3az+_}YU`2JZ!fmyZe|_Fe|Z<vBH3>*fD0{x%Q5-%S)_tZ?I2vMW*i
z6`q56!SD+20x{PKv&qQPLYcR*8F}3n)LCx&jHS&q=0*-1scP~aj!oj=WT33>I1q9G
z!_3&+je@~Rzo6AEQREtPqnM1Gdys1*4<M-^CL}6gs{P<W>#3mC!lk4oAQmOdV9_C%
z-r&rIo3ve5cM&Q$3w;jCr*4UxC2!W9=R2Hhcv5_3_0#UdhG!W5pjabi!}ro*%*t#9
zKv8b^b{xjy?O~&MJFW10+3>v$(Cded!o4$%t4BZYF@}#wXMDN7WP~g5rC{E(=P+iJ
zx(#0xsC(d9e4KU!N9{id*gFWm57S3ikN(J$YYbnQp3E2ud>+Iw-8TX;9Aylc<m&?V
zdyL@-?TFzJ0LKyhgakjz;8p1ir)Nxk?-T1?0G6u>R^NvRaST~FYQPG=V~GA$M2URm
z8il*T{HsSFMmDO_M;R(K+9m^i;0T1wpbU#PAZ4ubUOl=?0;Ca&C^s0s!|;6%0Wi~h
zSm-o-9|5FDMSxZW$W8ZQ3QR&Y!2)y=THM0CKEpuI;!}v#@T*5R33`T&jwuu0GeCa|
z=uJf5Nc0zojydV8M{f}HEE^pYtAdUo3jKMaze#jV=UhE{qoC)I=|SHp8U?=dr1aA0
z$n|-pAbj}F@m*xLM!2QkwBGGJw^v^Q;-Y)H`OP%z0;7Ps+TDl-0?Ru0YFI!+6G(1>
z39^fxBQ3CRrX4ejI}x<#HGHtj%pz#^%ruQf*!upe*|eVx2x~%DU$`mhu(2qBrRX+e
zI1C>OKw@mOW7Lr&Vpi4HP(GO6N*o(F&}s~)6h?z9LSa2?M+6gvRrEe$z?}BgqgxWt
zgBV&6Lky#^QAW)X%qYrMq`DN$eZ7w6SDV6ChJ1N)n=)two35Zy(3Az%6gQ2>cY0F}
zzH^##O{@Je4_YY3mjJ9zp}2-+UCMODzzB)7nWh;iSqBWwwCF`pi%AhyfP~z`j4!!o
zVqrkB&+t8spmupk*%-d<ry3hp+Zh`g#HZyTN)$76h$I=q(^2H3S$(V!n+`cUizVd6
zEol%&=={W}?r|G67vOY5pr<8J9WevdP53(j;0GT7U-xI+dqk0^y0pfpwa-m`1oVHl
z_4I(dkZr<dh-aoC5b|3jFm()p%(5Yri~^ZsLnyNa!d^Bmw*^vVLtHivI}O1pB?J{=
zwRa&wuq~DDnzmKmC4vUa1In`6TdOQfrMp24YHb#&EcY%IuT*T6M%@8qsdt%zsP#3W
zWVwPYk<}xCPzMl78dRty-W4{a9T0Xi$pxyLiJ=Myv3;ux3=HB#M;F}v=36IQyC+-E
zPqv;7T1_$^B>g`c0G8$JR*#P&<aKrD>Hg=uT_~vW(l88<`vQ<te1V%8S4DuD*^c_x
zPDrh+lj%B-idwuC!eR$}igsyW8|4@_u3SBOE9BUS^MwH{B?xwrDv$)86<YdJYCet7
z#pt<M)t@$e&r4ElIP8ITvI{9bpPkoIS#(1z?6ZoZq-LG0&LLqm7Zne>zM`EJr>$sv
z%@ReFQA|Idy-@B2l!VhSIM2?KrkRp>E9?|fuNc1L8jQBwo0Ed`85XnQ$ck#C@Q!DM
zRKJ8+o>Fb~CFHd?E8EG9r@)Qi`Dqd+L*iiO+mIB(sjx7Q+q6qR)+S+m2$Oq32EGd~
zxB~c67@9=N37hiW1Cz~wpGY>6-aR0}SrS?iOu^KjVCl5UYzM(P*fsMw)&ZdPJl8o&
z^l_BdqObE)(rwn=2a(c3Zk~a^Db#Y7%Q~&%T{~*j2^R4&@S>Q^c^Mgd3uY9_XuXE=
zwB59z<VtVEZuY_#*c+9gH`-|oFWG@G+YqK5=BtbYjM@(SP95KyrxG7(@^)kRRP;wL
zFe<}`9!$jqlYLVQC~9Kqwu^*!P!u6Wo01U2Mq~K3F|+4sqvj|?XwW)=v5jztHF&u<
z!sZU!3MDS2R!22ViwD4Jy$yu1%2OKD$C-yLv2$QHU$KdUeHZO8-+*~aY(qChwy_%`
z%NZYAgUz<xIzGg8YVn~XVtn0*hg~6fCsD8zSjDMlu-1Gm^H7|IbVUZXhe(|fHoB<p
z$m$_G2<%@{#ztkd(lDz(Ko@jo|9)u$`le>xKke+-KT#yA{>igR^-rbW=D^h+KMdVK
zD8_1l24p0Lbf6m8`u)pPFi<Q@Qn>3DmfDon+<t2{t<42uKlu#*r6ZP86+s;s(Ey9q
z#|>WxSknnUv4)a$D2BqFtYdp24W>mT<Z3(Qno|#nc#!5ZcF6Wq57{LlSJ)v}o_a`%
z$fYCJ*&!pR9ugG|A)mKH?l|?3-4b$@9kTJ%Lqf13&5d@*wo?xolaR02A=#ClT6vB(
zt2dZXv`lmH2UtbM4vIb4w#eCpA*vpKh<g05+{uXH<%m0#9#DZQ=wHj~7hfNoY&B#1
zL9hO=w(O(7U*eYPI00QqyZbPVly>L_7@lTsMwN_@?}IqFYi)m$LR-me!*kO<Wa>7I
zw%DicF?`5lyk3G+6b*L0q%@A2R;$@`kYfXp#xGGC->$^U?W{pJVMOJ*4VvlS5}y()
zx3i3I!hpc@rN1pYmq@AGAupOCUp9?0d$t%gO1?C)WCGQ#D4k;{gI$o_)NvO#i<l7=
z_m=|O5Ln!#V3awPRFj=*FUm@ZmB%95=Q^iY`STM>tfXB}84)iJvn~ZwVIokLt04<|
zh?F;2!PrBj9M`thDc!2ltx6d&Q>4l26pym3OWCK9mZVT)1h<x?xWx~;bt$S3qpD4D
zDfA^N=h=Q#aRhsm4au<~?@2@mPmMrusC<gvN*)L##SZ1MA;~txWkb9+MC-^)QqpWl
z6m<|%Q>NhK{E)u#d8x`Edg=2NtxyH`a<sh{mVLCmgHQ3B9QvS#X{M3}6KZ8)n<#>>
zQxTlT!Fr1z!1A2Nkg=~go(A+qLZLj^gEtH)F}PKMI3`uG7B^`-l-2=4uqTa3I?iOY
zovE~+oC=kzCJ6&ZF4a2GQ;zh+&cItw4lEk6QmlcXcp%ZRIH_Luk{ypmW*++Hl#>Qf
zMp&ACqhX1kdNeXxP`g9>M#D;R>e0j$HLMeTqhY;qMgu%3{-^}c<5Rd@CM`Jcl0;2`
z@>Mi~|2kQ?{tUalC|t8p|2W9%Li6<Vx>0u7@%lH}+A!IAE-!PJp1%%S?G&JKf1Dq|
zW){fXb>nsS?}lpb2|nk>g#kgU=Lt$y2R{6!QYe;S^Ho&IS7*PEzM6eM@_~MTkLZ?f
znLHEr4(02M)GfDBzCt`x&AN?IO{GvQk-0pP*PjsI9+?t8!1%Du<mwAwl=zf<y^T?Q
zfzi27mAnC?`;t)iC8fPV!uC9C)C>~s4bnBDC2NyN(dNKn(cA)RVj)n?j_(&*=*A}%
z=J%7XQ52SvuD3m|q-*IUr*yp;Q)5)xC|tN>tO355lgIuD>nUy@#SDAAznc<4hHu%Z
zh6rfy=%$MIq8y;Z6yngr)uUfWb-IH{Q&2pfTQE@MU?L$Vk*2U<Zb2D;EGChrP*u9+
znNPqZ(iGGe&n<6!0w$5Buo!PajriEqi8Q61n9qL#CXuE<4e;Eu?-MYIbT$!De?Bhp
z3;Rtc^*m?Pe4k|u1xP9g8!l@?uwTKmm{q~w@t7*&=r>gnm)?#Fu+8wDb8I}W#_)Z4
z&^MD@EblGQPo9XEH}=I$3h8|@>O=~93~uo<(!ULO`J~1Ko`9E6iYD*`ynIr90#Crp
zCq)u?0$x6;D}g8A<&zo`cmgh4pB;y5db83FPkZ}Mzib-j;#RfH(jo@i=3Byg?1imI
zyRa9(L~yRN`Cm*fV>S=mUdwYMu|0nQ|Mk*L{FKc&tdCKca_!ImSO~EN2V;2zt2Yp#
zlP%1WoOA0qDV{PK7hv0qUJo1};ikY~(sMy;D_2m!7}>vKJ!@Kr*E-8DFElVmt_Qq4
zuQ7oq;N@~8jywGNvE=$FJ?j&M1iU;C<P5~<>j78i=?03ni=GV$LISQ1@C_8Qou00N
zLk`0M9#w#9IkM*|qvlO1zzAAZpAzgw3t9`WpcTp7togaYv`#!8=cnPDfc4TzA(&V%
zom3AK>!p(#U}C*=QX@>Pmrjbn#Cqu@4Cg0cy>wC(Ce}+QVU#e)7;{EE-3^j_&(mBl
zZL%xc5TO>@^!&vEE|)?HMTLOCXFbXtx2&7Oxb_gcXe!WIPC;J#C)LVrL7s+7Hz~+_
z$KRjHbKOJzuaxI3Z6(it_GeYnr5XL?dHPrp7?;8LOrD=Cy#6+M9=+6-=bvXw`xNs0
z$2jhd?)K+2K9lDIi|p(FAC~8(W1RAQ-KxIwyaB$S$@2k69RDljIZIo~^9_GgB^|x6
zpFA(h6oGLEjL+ox>6YiE;I`7wC!nQ4E`NwyJ)m{J!OK1ROZSUENn8I+o_{9KPtwo7
z?3CyGe&1J~r=9<qJWnu#^uJP`M^V~Jo*()TRnn#B^pocn?(c!1`5BDQ<oW5A=g|?i
zJfDY_#$yj0JNv8GbI^?K!1Wl(f82yiE@Xc!J&9{3U+#kW>E~OxiU4WYd&iBD?woNM
z$h(oV7*P?S19`*8o^=_>yPn<{>0{nyAaD5CGcN;qchS2cVQBc+vo8aMcIfK^cn>rK
zvpw+)%*x&w7%Xu8PtU)^>rnVdtV4Pl#-H76)b7R<%;dVi5X>XJ*MFjVkn>r6Qgg@G
z|G!(GbOp*AcfL`302Qe73%UwX^xXATC3@z=<Z;kO2(M~64GP&c7mIDwNh1Ja;g`1}
zoOIq6%`_0gXEvq5G;;ZdMl9;$!X7O7@v<(tlj~_-b<jlkQ-o7a$+c|T35Tm3_gvsE
z49XF#dG+1~hGjO~1_mhh*{JKlcr5=_1o(agkOTC32bLqe*y}L74#4Xkh?a_@^g0Af
z54}Ew7q8qnPUr`)$W3i~0mYSA<<lTdDmK!K1joI2z1wI2o}mv(j{7`gpC7IpkL}BE
zyK{L~9v5I`d2riWRvMhatPEU*m6d}Vr?T>K9adJUF{Uc3!Wg5jG~!A6J$fF0+@IN=
zjoLp^Kw*A$4SE#L<Uate$G(Z>bROm8I=>v`6%*6^Fj+Fd#ia`MEFd@~DBKLnivxJ#
z2ZMJmw22kBgySqs8xz1y0|YTtySTA|irCmeApU^w(D0SS!(kY@`Q!Bsc+cb4AQyK`
zF)bX6seVj%co$(oRc<rKnWBJo0C#wChgEz%b}e@r;5KvSko(>P*0Zs!e^ohe#MS9I
zbi_LZhx`Rro{=Lua^4}wJM{HlM8mb0y;pky(0jF)cBJtxk*IcL(9xwGIdr%xRhWD_
zJlau8N1Aq2(2*ey$JJii&1rP{=WBO^c9&{*qjpzlcSO5G+TE_*_1YcP?gs7df;+sJ
zEcN$n3?#i2SlmKR__H_S{xnn+xzfaVa6MoFYrkBLwgkwEs5BNNPBlqRm_6<GFmkwE
z+U?TrE>tq&3xnX#g>qieBezxJs+@LlFZZh3EMbz7>41v~JTVatCTp;|8w%wMcffi+
zXuV@vPsjE=!h*zd^vl=-MKS~pSGM&nOCCl0`@5!%^j2Wa%owv|NJO(aq}}nd)b6PE
zZ_sX*CGL0&B)!Nh-B4n=<D{_|wcn=VffG`tEXK0)yeLP<Z8ihkfIQ__EPCg4!0Nu=
zF@@y3>`cNTy<`6B@m|R9`G}VHg`~j@;+#2``FJM_cX|Y>4}^Fzavkn+K{3VGYq&oR
z>+y0nP|ojLE!Zrq)}wr9U2*}&5PjNn9^?y!_@q#-F|`S<&2&v^gln@gm3M325;dl7
zRrjNOrydui7*pQ>bY{eu`ntM-WiIYm=`f}q0rc($W9qxc)E-zDwHs59F><3Fmj*Q9
z+L~*TMEbYujr6GvMtTYEuemc~q?hCVnjb`s^!sps&6!xxst2>N+ND-bv9-52z9&^c
zY5`SNPA`Pd^>xKK;?=YlO#4t~WlZa!NG8<6*uVXAPmc!fHmw7(pTZsJ*~F%Zb+Np0
z7kQ1^XDPHH1<7JcT(Y(kS9Pq@%*H88@ljV=nUKw;C<F-8Supt1;+aLm=OS>xpH`OI
z0&Rw9;&vBE?i|Dg(d%1=M4hL^6W@~j<zEhTa`ulmBg*(S$!+nIqA!D>%x$DTf<)y4
zB`66)vhbL-y`E>6aByr0);3kd1!Eii##9s-ugWPN8--(sD(2~9>)}8R;8i<w#zx?X
zsiLkLi;Y-QQS>bAg=5jjPVH93ojbN24%831F#<PBPHjWM2Qx!PdO;?Nzc>@+KRpu_
zU`{4#KvgEHz`{(_f!s_~f|N||K$c}z`-UA8C9MjD*&5*bA}ItM6oIdiI*80yi0BI?
zK<~9hLt+p80;gfH6ALAY?h9KfYQ5?djOSv{dEnF3XJt=YlLuZ)<;;pP^#uk%i6Ve_
zK1EM;9>ti-Lv_Hcw>`JhbA^}3W{j!WYYapls^LbLSCkhIy(lkr6oy{UD=!{jQC_R%
za0|WOQ(in{qr6bZbcDy17thQnFX#^1YoqewSr_GnO*-1^8Sd9p`G=)e$b$uUSrd8q
za>*#&A#p>M5mTK2Wk?>oVL(R68<n#>)FIx`JG%KM%9Ts060DL#x7@;o=6C3~dZq(y
zG-w@(4MNoxeQ6jftv-}{;^zv+u0>`=i1;!}Ou8k=IUDf>Aj+mPz_`h*i`@?QGpO^S
zz59>M@OLA@@VvHQ_{MfKeBF*<_`A_y_|}*izO~B?Pv{PYuQsi@5!0H2ONhSH9<=V>
z5wzw;O>0stXx)NaNx#z_w7!L771&6uPG<EyH~$X-bn1&E6dLMW)^Ydzf27j126++r
z1G0pEzWZ{nrk|UaHSo2VTLZJsB<qgY`TFQ*y$rBn9e_z*kw@m0OZ5(U&-1Xo`9<E4
z#~Ycq2=d#8d`7_o<nxE^!SEfiU>L$?QUv*GL;i5CV#*HW8~L4!{4PR%XCl9sNPfA2
z8kzm0?9Nr0wKwGAWVYm&)Z6klp5OF;QhDf;-w_1bmv`^t<4Xfv{(UKO+UFv6+_WBe
zf?^h&hJ403a9E6*?fGz(-Hh6hAfmT^iimJ&94?nvm&WBB;iYl^f^gOUz?jsTC*#4T
zjCb*dE?iy<Sp-$vdbe-?X!w5YIf{vX9Na7XalY7oT(1GXpGA++tKBI4SsG3~z$g`C
zD?;%xFBOW<ULi1F7sG}(!*^pyg4`G}3O|pnu(-F4*LgH~KnFXfae1zVmT6UaJ=7(V
zS%Z!uJL@CA!6RMV?U)&f{l4A?U<!eNe~JUhDFBuLz`+5VC=~+pPMDTIhl)@sU>F-=
z-w@j|6hvv~zUAsTeQHsLk$OYgVdK64o*2UBMTp7@gqFR?Az0A+ErG!w5{Tfoa&(J@
zyr~3E-gM4xGGE-Q<;@;;u<ZhWO5T~``%P|hefEQUv-saM9yv($rc_5!0aLx-<c3!(
zp0m392D~hl{chC-wsxxS26Q2heX0wt*jLY2wJ`S>c1e`I&kLcdBQJ=)a~v;Hn8k#!
zi!}YPyi(AulZ#cJH95Ph>Ku_<JJInVWqB0KgJK9;he<^8nPh7&l@DCDkR%Z^I<kP5
z!G}ZC{c&p{q$8gLFqRyIn~dlpBNjK+!pZZ4hmCJ+sQXh4<qF#t@OW_*dOMm6lxZzg
zW*`Khe+fSi;1j6j2f<VQOZizw=W>1;=v=|iN;+5Z(@5uPe%8>rj-Lpf8~JIY6Nlzf
z1MPI;$X%*`2c0<hlp2WAiQ`VGer!HW4dC!XYM_fw96Lz$W70h}fCB@mpHnA?sJ~@z
zFlxM@8WklCd{S)};|1^`%(~(w^T4QO1ek8)k!1APJ^+nJOj<93m3nUX8Njn2=z+}G
z&*#rjFV8evuSJgoT?+R&y>s%VUa`7<^@3GFN(`4>m`SfvgBV2`V0vK6dk({{lbVeu
zl;sEVcrmjm>x;foXY!b{o2|6EJ83NwH;r#^g)g6G+Z~&P6xDt@(%tT$8{MKNoTK<C
ze2{_O=d$o>$N(AmI-)~Bx%6^{jI)B)3JJ{XI)X`f(r%Tvi>bhj!So;52=QhHBnvwL
z;E(l`q~uyBVnc3+<Xb0MaxJ=T<^*Hzl65Q_%!))HE`rwUxVsVOTsRB#xtT_m4|nvP
z7%C@E)NO*eIyPk-r#)ggzJdD>alw|l^nc;AogT;hz~q`Y*jV0<O++37J*|EoWUiI{
zk6<aKgWk)y=w{rI1UcLW5X)8X|3*6Zhd?~c2^IX5@{`X`4nG<Er19gyXAn2R$)31z
z53s>4jG)dfM4pH0{D)v?2PIVrCwu7^Yx_BY%av>blU8aZg^a_*Y@(2v@@m7d5)pHa
zz(Liev}4tkAaIMggT8I&0);!6vTO~|SrrH;euPN?pJEU=2uJ`t+O)?GB7%6j`BRfL
zC&HBCP`rpG(>D76A-j-y2y%iEs6q!?&x#q#2Iuu_dV)|l(Zi&P)H;z5qaqCZ5g}-n
zKvTqt{xt-}C7hQ+L6r~dua4f6!kOw+wBbxBAs8S5pO>L9<3T8ROu;!nq*DmzB_nJy
z`(Vs+Aa;BF(K?Y^Bawz1{_dIx{V>$6;g|ZGqAZ+;q_b-xi*AqSHHQf?(;2oU&9;yx
zvLPQPe>fHpwBEwBMeND%AgRlHTCB?vV{fLiaRKJ<!}q#-TCIECnw0F;h1KE8WZ+gN
zYwELGjM^aNSoq!)Vy1vGWJq?aQF{)+@$Px01qV9?2#najoP|Z8H6+89d+ZA18+dnt
z^XSY_@!2;?KbG0!Xg^eP*hQcn=3W`XXrp}kmGy8;zp@c+wYveIcFc(Sa<9Vp%@@21
z!!vv_DeEh^3WIjc>QF^uxE>tb=!T`D3s*;$@5CJFYaH5tNWX)a#O=UQ4t2lnA<n(R
z>i~`hBmQ?ERhN~2z`z@sBClp{$4GJ~g7N%s3$JBH0NOhcam?^F>zkU5^i?vLM&eDl
zWtmu<ht0h*ig==ojF*1$=JGCHZ2KZ9<`5fzXAn8@aRY9;x|dsET^L_(AtZ+x1I_O>
z@Vig$@I3;G4$;XQd3~IT-lS>k!fBPGxJafvcvXa1J`v9@F8D1v#;x2)s~cD!5v$;;
zb|P%!#lGMNU#-CHu}2wl`jruS9dAHpasG$)XOM@3$RV&|IEmZ|_QQ+U_zDye$8f`k
z;shf60Q1j4JqFZnVMZ7D(AGfyf*aJxkTpaH)F|*a8inc{n31l$5N7(74Tya+Fl-(b
zDjvm|1#cW!$4e<leH&6&cbQ9-5YF*=6eRnMsy$KW>?wrd5w40Ts|b!Y<33y#0GP4`
z*({%al{^r#jVSL43IaS2G^t73f%X#7&bx{T&8Rlp;m7woa6u`~(v%0UY)6Xkf^r`i
z?Ry1|&A2_h=mh>==6&@ag3gcN`w9y00D!OYT@X<4Dx!WBwNgii=Rh#~kLX;8p@X^L
zO{kj?;B{6a;g@0|FUo(Hu(yoD73xZXqE`WHA`B<k)l59*FwLf8X=Yu@W_IhC?@t+u
zPr*>k{}&8RgLUE1419}+=3v@?C>8*6hn5;+rVqse!1AFXwC$eOU{bR^O(9QU7`1-r
z`H<-Ho&%)b=TN3x3v8gIM5|r^(+tn?=J0tO-rJ={CNkXUonP1YcLnRbxGMAwTot-c
zy`UwNiGjK+zJM!yIlwM}tf59B6MdT<;VoS8^;YkA5c|kO*he0MSr4-wW&=#r<x|^#
z4z>@PX}o>X4)o{d{|-CQ1B*ARX*qibx>0j4C1X#sW&~F~(*fa)v{mWmGN{XZnJm&*
zksrH6VPLpctEQE$pf~FX-~NnEL_VF5%LSyz0B>FL%p7hy>mOT3^#HMDNNiOT9BR-v
z&2X+aD_GY^os}|qjyI!zEggD>BCQ6L<NTDGjZ_crj8m`dEqn>uln-U#nt$FQ!!smr
zXcBIl_nbv@{j(UUH4cg;fesEcDF1ukNqGSs0DB^82E61^jeFLINW*^Y@4zCSE&s=b
z*hr_36Zt>RA^#!%vqNY$!EBUgFq`eGAp7rG`j3mE&HndGe{e6D{fW{aSOwXMrT@6o
zE&a!R3h6%%Wqf+1KkLpX3x8C*>`#{c5dGPSME|(cD*98#oi53bx4=Ju<R7t(<3>vU
zk(B%+Dfvf=<R2-Le<UUUNJ{>Zl>8$p`A4#m_G}6!b?b!+zkOxiB#8dF2KcMNco_S=
z9zH*T)mW}5up6MloQu}Z5d^yfwto%|(cY;E+WK)kmYWbuFngU0MvR*DaR{BgTHZc3
zYOa7QGh%O2dW>Y@o&n>*Cdiil{RuQpTb>M6@?@yUlc5fIg5EdxEYT;=5`FS4(I?OP
zdzL6e|7Ro$^~tk7QKC?vJS(w88G5=U%Fs_CQHGv2i6XV<lSK;k$+JFLrU0$@ELS3_
zGW7IHm7%9csz8DShlMCdhC#_H7zWmkSvU;MXH4<12;|7HcF2)oQOJ>DUB;O6hH;R`
zNe0aOs+T&{EzU;m6vz>581PpgpKjE?i(7VLt3K3ng!*mo^*C`caS`N9TooyX3Q{YB
zi9<%Qx>1bE$x6HgaLD)$Mk|m<jA#8`&LvZaTFd!>#-!NIKg5LcIse)him(blb``wM
zajQg#88yqVk|~iGaxhL!i5NBi23UL@vRQN-OlAN>&GN(tyPVCi<<VFrkH(5T8tafp
z5Jwqcr!Sa6{U!tK^c7_MJximp|1;8v`c1|sN+aqw8HuIQ*wZbI#(oNEG!|ui+Vq=@
zPZmayMH!zgi=f|RB$7m9Pp>2zd-^2NWoSBM3NE80x(sX`v+y#JM3+$#T}DZC870wW
zlth=YIf^90JgQoA{|V<&joJc8qLT+NJ3emkBJ+2jjbGvuGJpH*)o0@uP7n5-z4~nY
z^0$v)q8~ecdHD#Yv9CC+^$r)*1ne`@!xYY1#fD(Gf$46JXzJsN$Zru!u1>?XAzUUs
ziVGQIXJ60B4z(m|LRVl=?CH7$`U74!rT%~gQGrbZEIvt2{eg2Mwcc)ct_@Kl+4lir
z2xaRJBbELzQuK$Bef5V_2ti*kmHI;}grKh=_3v4K82LY^Kct?L{*anje;9eX^@ow4
zLVp;EGCpnkL+UB%52>f5KcptoA4Z;D{bA(k(;tST>5M5DPW@px*g9t6aM2%zQ-2sv
z{b4xuhvC#8hI9VX9?7V6*njNRAE;cg0jrhE%gP}aHdTLwkxeT{Hroz3W=P^<r1%i0
zNBU2iMn6vP;EYFnfBI+o$7lM-0IO%8&7Yng`M~3K{~zfe_AG``^CR^4GWWo_Q_N)G
z#DiG$Y$~QQhP>H@S&ZC%^A)k5%<jQ%WY3Vlz!z9*@^$_F{n2J=w3U6(*7Zm811-Mv
zwBN)6*4JgX*GUUb3xJrpUv?W}H~zg`Kbqt9<s8|B^{YWXI3Gub@QYr<jYv^SM5<3>
zD)E`Q?J>^A#9s3=FphC)oClmZyMRUADz*k#YeK-ccl2Pfks1?jpi+F3Brx&YhTw;c
zTd}bbS5bP{;XIXUjKm48^5Q`ue5Vhp$2WIS1HN+xHNyWe&e!tP72igE84J0GjdUzf
zfi}SS#47z>b^7ItAL641!{=3boXZZ~$=v{2ecz}3UeTe&k1PK|O$4F#0t4@K#6#~j
z(pSh{0>NN!iv-Zk_ojfA;!c9Q2qtOVThPP><SxlqJ3YJU$unbdPp-kc2hlG|Kc*S?
zi%P{4{)2(bPGaC5&A?n?U{DxH)*-STjEo1}7@681LGTL%gXa~SY%mG}OsA58O#tx3
zwuRm>vL4?=Fb}ge*<rvSgdv-{!6rL9d>mVL6bwR!!G>Ng&_^+$0X{L!d_@^{Cm^=q
zAkb0qGJHrg7u26Y7(63Vo`Mx$900V>|0$0qo+x5Xz}@lKp@H3^EPC7vi}&VzYLEs8
zcD}BV{T#rG5}pPmTsDuF2(Z3ijK%+Atoav5<)z@>br{T_p5ua2agGQ2M%6iK_!gg&
z0R`lo94a8^<l|d#PAR@~&Z)q6#W}40Jx#%+z3NUqj3c=2Zqy#<N(}T4mYmW%*3MKA
zHU<S*2?#f7akCT7&~Pl%fF&4iYgz!4TaGYQn(Hl`SUvt>m+a*+tz2)u@!%HRzJPsD
z)(I1P$*A|l&btPJ2zkT;%v*8bnE>{RvtPh>ecIk>t|t#e#H{@yoU#7`2?8_2Ro?vA
zk?Ek(YJT$U1|VP$IfTIO*iT8gwL2Gk=U-I*W_Yj}&cWXGRihxfkemI+k_$mn=pV9h
zuH?sg6Fa|%rR$hEgE{7gr<{x@Whb0<u&3R4vWv0ab|+(Pjg^rJkW(gQQ7L(Yp&hf7
ziyevKAc(r*n=;Ki3x3!>X2Od@IYEUxN8ETpi6%B8&?M(%dn52F7$xw;AMOR(T?jnc
z0VfFX2x*T}Muc5lbn~am>Ns-_3bOzO-{Zus9xue@0E|T1QEViRSWL7{-0QqMXo5JJ
zPU1|BC;kYRuia7L32}k|k3#JcV?-d1?q1?bG;tn4QT}n_zMzO31u%%|W>~~UEE3lx
z?sZ<5<{V8YacLS){1MKl-N`oR1OXly+QXwcN4HJf=VBYm)J{PrJqj51zRSsKM&1Vy
z`U`BSyFY|LFhA9POO_2|XYC=^WA}cGj$R;h8=DCrvnU(*y08O~ymV=D)vm>A@G@X+
z<VIH_1nP3J0Uq2UmcKRED;pOBt!}U6kZc7OJ!J2qKRl`5rbQ-aJUAJxEgl+3vMZ5$
z5=UqcL;{TPB)uKccyQOb455<ohtH60h{l7z160p%E|rvVf;gD|wrPE6Hhl<jiG7Lf
zh>F>YN|gFNIlOGoN2K_>#>8frA6<)xV!uSu;EfF)9TaAur(tfmn^$6ixr6Sy++RzL
zVU)pi9KGDm)T)Nb{*<?3p*z?%G!6qy2nYX&ss>S&)HE^demJR#<1^XFIpOH|hpiT?
z?N>)nOt1?6=u%8STL@{<weJiH&RGzO_D)I3|21HN--_gd<rM$9))|Uj_X*4VR#Psv
zY`er~id(VxK4G)p+K_AAk_-^GP2MM4C#@!HN(#?5l*%{_B!>rwE8V~HAMLi5N{50}
z8c-ru7huQW2+^@rUh@f0Oa|}+Na6}W>fsTdh**zV&3@~p@R>l)0TtnE-LO%yfCLQB
zcmFCk{DWi_e;R=B^U$#4kE-}7PPyw7@#k6311X~5?E;rnHu<d=q@Wu?wpbc$OTBFg
z!NNKafytksmD;}fwk5~5WWYlF3Yc+xkL~NSEnU#=Bq|vP13%4)AF+KKZA*h~sfUH}
zx5JFbUt#-}+LnCVk^>9zLonm`X|}J&wzzCd7mlA0Kf{UN4xmsQu`P|Zr9tCIVXF9F
z#4!Nzg{9uMJYZWwwq=2Bsjw}xY)h$aDHID@VvcReuq|n}#RCfo&WFh#B~=u}Q+(x!
z7{hi6FgLd3GUZvXJ)_#Qot}4U&m8Tk1fULio%Rd?)9zEL8A!Aj!!yr%*Ka)X^59}`
zMJ}Z{sw3Px1~2!9KoOcCtFiIOZU7np$VJCuJaQs;Ft&lyf2nG0#v>mP)Ckb*W}Y7z
zj9eigCQVCWAjq-Yq;Aw?0@(*Jn@~A%wv8_};ti+^qj;?va#X|B3yy@x1TNJO_+4@Q
zE`c*G#?u~$w+md=*175xN5s-C&{5z5U#<{jKoxG6j-xFeN1I?LD{KV@wt}*@BMwCf
zpj}B&Ay;@=G`OmLm3Jj9fb7Q~_Se|ZWeWO&Q4B8Gz&2Qn;unC4(~CB6jR2oz=yq6f
z6mpf9HP0Bn&dZ9&qmHb1@L26-z2n(PRy(?-)?rKVz9eYQco4}mWb^CrnYsfTUN@4>
zeBuxOuO<N|Zk(5Z`?a|Hum>BED|RMvEA1{7kXx^>9{r&QdiZX^vt2uF*LI2IWh8nZ
zhKr~K2#Adsuk*m@UWR+w#(bJCl^XmGL&F}vLfRZG5ThzzxCcU6ARd(kM`V;`wU@02
z$k^}*3(X8MmwL0r#M203;*gn`=y7PO7h1be`;2tdK5e#ZGXhf`!(9esIb^mR7G#0%
zoMA7xypjZd7V@<voKPxymN(|Z3rCTPC7Z9{+IT$kq$64-SwTeWyh~L?>TOAkPI%t8
z=p%dzu>)J4K71A?4mTCy>7?S9HLoRyZH$eN)sY>r)Ou@GWW;_AalOm9-e+8V`tkr>
zuj5nvCd!aUbrB0{5n@3K>LCi|1Z3mks6uSQS3tZgDtXz1^fCP8$zbMXhYH6BE{kx}
z`fg`90V3Sn1ojIj%++sY7`~U|!Q#(ZfXdeZEtY2wc1l_?2aOoD;kJF8G(tT>u8`M4
zc~MT@y5PGR#Ss#@Ljtlykq`Brr+U;1C&mi-E=AEkCN0m4VORGf2O!ou2Cwo)5G|}R
zSfjAwxF2jDM3W1%9T4EYLwlF9cg)_T1zH-Bl46izrLd8Pmp}uwCAjp%Ax-KXEJXqz
zBLxL`1gm%pvH4OS<R!~ej(mr-t=79l+m=cTM@g^pE+mL$U!g4Pyj8GdMmjHa%r7IJ
zMXTLp9&PK~WW8yjetMR>*<fH<;XV%*uqXmvHp2utc<5BU1In8G3}AeK3h#^AFQM?E
zpC-PYpdO-V0`V7a;KLEM?g;bBcO#V*Vfh*;USZ)oX?C=H;}30yuc>`vtT=^$$3U03
zS=Y%#iHr~ZNn|kjBpg~sA@=LrPl}#Q8)CtOpiA7W>t>zF$1F{w=2hIMpjbVN%#dUA
zbF@v#OmfOQU2{sLCm^f5#el5wp#1yoUxepsH$@UG>)az?(Tr`;jO9apd|;J~eGVoW
z+wXnQlZh9`Huf@>R)m#jR!{Yv&;+rr>5Wyar}$oIg4p^MzIrvZ8?T1G0Y>xWh(d6G
zH3K`kSc%wBztDtZsp#X4xl)fJNjy)&iu$n#Y$Z;FwI~d}*wkZ=ZZ|*I1h2VG#wG5r
zHX<dCsYSoRtx~YEC+5sBtkg1|WJu#5U&mL?dwh}88YAJ1J#jPQgVI4|D4Zd_v12z8
zhJ|3lzK?fZT4QhZ0pSdEYixTT5C##gu?RsHF1$Z^R^%N_qV$^;xvYT6B$yS+yM<A;
z#wL@#kIkHVA-#9K35WbB$~aU_6H}dMGBGI1fg49e1J*ME>je;MT#y4zY%s6lYU*I|
zu&^#{N`lB*G#P41lTidnMH|{Z+WRFb!fN9mZ$&hUkD27-SIEctl9|4IeDp?Uun!-9
z(+4Dc4EF&E9~Tg0ZF=%1MvughnS-(6OaZs4Sg*wjh0p!`4+{`CIUI;(4dgH7#hV`9
zijC6#H#2??*BSqxRd|d~OOa&ky&Dt_cMu0de!=-k94DVR?FvU1BCax6?PHI@Q{M&t
z&?3x%o#FrSq=}v9@_@`lWGWEGkXFxu0jB4`4#uWR-dbZNjK8`$IZ(X`e;bo=%MZ@0
z4mN6ULF?)q5&H=e?EGE7^+jFYc&;AxSc<$FPx(SE&j5`$91FWDdK{7+Reb`n?ZC<Z
z#FZ)sEhh6we1F5K)~A5g!urH4{+0Eq5Ok{gG>gps8dw<ns?>_U^=Vo@$nIO8rt|?x
zeHzyXB=yNh5N0zS^=WskSqg7AFZ4(jYd`<r+WrS(v4O%@+x~CG+O)r;KJAW`DSx{@
zt@yU0!BL-f$F7OvNPP+k&L+D)VM;>k6VG+t7`j)cR41Y~an|&GqKcfcXH<8`Yt!eX
z`LxDT5qqGz<>cB_$l5eA=0_Zz|Ix31;{=V!kH5M;K@Eq%<Op+(bKv?k<O4^wT@dH=
z<msdM@bh7KYQ?+(wSbCwy*5MItboZ8awAOs(C;c)56g1z5+uo99kpHCmV2*&4V@=+
z(xNC#Jh7>sGlcye+F?>x#+?C$?^C0A3rg7vFVkU6E4*2ZiI(xOK*et3YhzGm80bBq
z6mLNcxNM;4Fpd$zz-<GCCq!v~j_Nk<<zZA*{Bu-HtG&4@rU*h8wW<z%3u9_Ra6XsH
z=VjF+#t}w=45}zF8KI)6^)816>B=KZFriiP;jm(=@O=IjM=d3%F~o!gsjEl7D$4N#
zhzCB<M)@);2MzW^uVUwWdN?kNFr5tZf~cob0_t(tz7OG2Cv!Z6OI3E6O=qJ`=W0bK
zX~MNj>U~<!xt&?x%FhnI+qM&f%d1Ckm-JZj1JFp^*AWW`CWUX1yypddCD8TTwSCDu
z24ohhyc_?b;Ih&n9#jvL^ue~=8`QQH-h4&#DsO?ZVDPLg>%7IV;2HuP!81F*;@r=H
zI2J8&OSWOFb(2|2-(KQ=K!vV#*TRAvox}scSbiKq^1-nvh;=-3ytP-~=40~e;kV^(
zR!lGq<1w-Ja?1JIu`bO49y>lcR<Ls`%FfMd$`=j$J=$v&;o6V>Z}vK3(PII`5~04$
znmUlaT}ozv;keF9+m?D~(Z({Eqbz83%CgE^1xrGGo2nffl$*^R42Xchzg^!h6~@Xy
zhE~LTI`r*Xy|Ie*Z_~F`;R}tl_SLu9;WrWrT`vmpO8r{`_-aLfaJ{ph<X~}>jm$`2
z?rkR?HnB#;#OYQsG06>6_>cAG0&ijbJ|}dz?t1t^p>T6%gzu|0$zy_24_efiP*N`j
zM3luMmJbG*P#UMgdMgfk6yg@-ty;^(^77ReZf3xx?iiCV=7RH90OYl3v>MdUsR(~7
z4yhO7#;961sq1YWx&}%F>{g6oE={*H0;bFeuu<SvTozckwU=X;Vd#KmRAkN!FT{ku
zTwn1IxXX>J4DF~`3i3Tb?1d12_zG`Ni{HA!YcJ1Y1%fJwx{CtWHpr?@?GN5UPkK(w
zbzM(S$0M-E-ntgHl=pgi>iyv=ubXkeu$Ogsk3%}ps4vkUmpss)D-*I%LI18dd;L?C
ze~MbUz=aR+|BLpwmv!e?Di!J{RG}_MKN;JDj0g-RinZg|jR7ubwFbgxa?SR-@Tx2-
zt5~zGTG(%7vJBJ@I5NRS7O{sA4M!$_M)dK^Rj!f{x!!!7AmYmO#BuJ2d4Ac~e=Xan
zo%}j@b#83qByi|?*kfyH!+9RCBNyvzJ)8X1AK{7?5Bx|Rl)8hlf0XzadNWiR{Xjdj
zlyjCi^(9DBthZ<cTv^4tWv|uFQOcR6ooULMu7jI%>^krT4Uwx8C59myB6r*h<S@u)
z&5@h^1oFCu#NzQhry(BdJc9FthGfMd4+&%;j^PW!a*e<PN(%C04S66AS)d_}amY*!
zk@aJVr$9qw(N-W6G=vMSD&`CgktK1#IbTB-#BowIq#+LZAWvnyEe?rkNLL*4qJ|Wf
z+i8u^5KkO3h!7}-jex}Rw3f)bsbX1h#4g{2px`f?nj&{=u1Z)|1clu7fW+_E`V$=3
zq`i3Knk6pRZT@2a?f&ohOKxxO#GSDi8>uF5hKs#E)QB1#S8uzo)R3;YB$!5sedU>3
z=#SUH_XT^Qmo?C+sYXNU;D;$8#JyC}F9b^G_dD<J46PsiZ#c-7h0j{tr{;R>4%lK}
zf%p3H|89G5TZHl$3J=87`BNbq<aS_G*t32bQSZlREfado`aR;<JyB5O&9FqMayEp?
z0$Iqr+t?7krJcGVT&kUE$~R31?`;TYYe;WHIF2M)EWHik9S!Mi2s<>Sw;{A>NN+<}
zts%V);ok)kZwNov2)zwqv4-?EgmMk(Z3wq(NN+>9UPF2tLbitVHiWNhNN+<pM?-oW
z!e9;QZ3yp9P^{=}2)i|;w;?>MA-xS@n1=K=gkyfhV>g8HX{hdSNJF@kjy`SRS>giH
z+Xf!j5N;k)#rU6qaK>RPDDF%~E$Wzv0e`$UeF2XCwds~eP@8^=aB-}`17IaT1g!5V
ztPZ}g%(&mi2B-ef{qw^>oB+iB9>srvM>;$@aOIUts(8mWsNIZcxHeB<blR~U-u(mb
z`6B}F%Q-pLi|Da^0FR1Q@K`&5$6Ds(-{2wnRrgtSWFjXdWG&1NUc}p<fZw6@#lDA7
z9WTIF2W`XTkmCF6@G-5<jyCQle-D|A9b`@)hU8$hc`8iDe$BYAFrT&f^llc%{+FBm
zQ~kI2Z}u15WLEDRCG&Leb5}>*z32$H#Ln8lWe3^SR2>D6usNu@U7HbYHfpm$oAufZ
zDRW{40!=LCC*N$XrV_v(W6FY(NYkgyQQAy~2~-mXf4Huz@li~=hUX`nW9IW@Ow$S6
za$9LWMqj^dP^<b9NS+nDD~z2Y;fbaht`0F|IQ#&NU|m^$u<mh*HXW0X#|GifP&0kY
z-JgfxG^-DWVpr;Loo7`JyROb1GOcO`L^R7(Gz~QRg{^y>p0Q&;R%trI3U=M&<P4%v
zMwoQ=jRG5dASz>P><G@S1VMmk3K*bsh(OFRm>0gjuy6RYdC}Yt?a)IjlU2Zpg~7Vt
zwyCt05pF*zZ3+7=J8W;Zi5aZ>b5sRaMmR=SpWGw((O8dlQqGBqh2Sr{H%ig|OE=tk
znVTnL&(V@$D9<4BmmCaF245;?nc;^*@ShyM8HIM6#|+<_9}GW4rlR1;HOwCOn=it)
zp2sFw?k6W$xlg#@SPRGc31Rn>#v|1(#Ba%WLKwb)xCO$mKz=px==jGcGUTtYA80*-
zJ_aO`XvsHReH``)#v|_amRR%9i(MQDyS)>_1&<wvN5Puo6RZ(yFzX1&dIzNQ%j*6^
zL2h-gL6V%CoM2mfS|{V)wEMpap)=WP3PL?dM$M2K84i1Z9cS+rR?Lx9eH$Jid#^Xi
z&(&y%x`I7`%E=JHx`)yjFkG#Suy3Gf4#tWc(cINH8iqh&vmGKDWrTB1isnl!?9c0H
zf?@1TBT1tf*XPA4jfQkF?Btrw#*$ya5(s-Tj|8nYvpUL6y4#H<{|v4KF`}z}tJ|zw
zgsl<>2jgsAPb+RVwT_w9uP4pNIi7>qEL=XMxk)6?F4)^TM;<oHzlgsZJx?Sf3c#B2
z_mo-PGpG^`q@^?0tPB1j!%RN_q#H08E9YcK`7}(=l&9l5#~cjUvT~m02e+|h<vfeI
zY1~8A(l!4%gho_5@P|o|tQ_nGE^jtIk9CNw9HfEqC2hD3a?x{v?04t8t?teRX5HMR
zKz0XYZ@{?rWnQM0{w|0gY^EQsIup{VW^<Jf%YjK%=i)nPJ_vlj(``I5{!jS(gW0nQ
zw?Wfo;;y}JzCu3lYgV^C|Bbo_wA?=dKle|`{G@<RcQAX)yf?tuDt{oHF*$tZ<nRb@
zAbdMme75Yq&D^i=kNg$(x%sW}E5jom#**`hKUgCkvfTfP<Q_)C!R%+}y=GQ#9OY0~
zvG76d@rqHX0vP$EjCXbrGwb{G4?fc>@OlE4vxn%!=KjBj9wGxhL;yWR;N%`cR`PTQ
z5v*ZVzv-k7qOxBHVOlTt=^#evaBPESoxA?yI*52!-9hZY0D);b!Zx55n;+9b=qRLv
z7!~`a9f9s3MrjJxAQsg@V9}*d2cg4@lrFMEqk~WZKfZ(L4f}TN%XZk_Y)e;C_fV8M
z2v;k^*+JOBS({$N*r8v}V+e;f_T{}#^ywgQk+repVhRnul>+G);8G>V_GyleA{f32
zoy7I%38m*XDT#vNQRp-5&Z0hGy*Sy5(kBhQ!WxKp)m5BA7Xe{E-$1I?hNgEeNZPRV
z1MBAAZbSN+Ys22hCt2Rd5ZQWcl{_w8NrE2aG1Y@iwl=a8@v&c+7%q6i(Pad}N&UK&
z&QbM!`;R2M|G=9vxJ)*X{gQG27e#N~dTL$B?f3U~A>Uv3sk@Moj%dEoH=4hu3wds_
zqK19OC{}{kAUsqTGV0_m<hKKMA^(B;bBbL^NQtIX>_Tp^yO2KXouBn5r_i7HPVP_M
zNJYzs1Xy=v-~Qw#bMd~$;q=28gYyJ+1>WQ*&u+v>=Y}gdE?ww_bgP0^ya6Glg`0H?
zVvmL1oaQpKpPQdX&r$5L{)9Ry=&@vCF5bv!Z)?T`Y`2UH*bSd62gbULrethKSOtH~
zrPAm;Gn`y^LyzUD*?ix_TqSG(FxGH62C)BXTFspc@i{5&s+oP_&Po(FF6C}kgq0&l
zof(|H{GG6)HTG+my-A<Lq*pTOuOew?s*XH>Sp!q%CpkRh&<wF04&dBPcyPcvBUmO^
zZc?r8!IZb(NS)*fv`$Jx({NR#1+5Qow;M$LLa*=15J0h)GUGQw2ED$3HN_KXdOJCg
z^xR}?Q5yb6$>oq6(BAy5lc-=pXb+CxM(}IEGb3xe0LEh~op{;Gu^$Z5*`r=MsU7Ko
zWa67D=|OKChNWu2_>NA&NW6VA!i!f|-rxzf=_A`=-B0VPk!iq!>j+(|N0usA4_ymK
z;x0D4%K~fh$SADxK2Pn)jCj0DM^=bk-fkJ$sNZcFiI-XUX3MIP7$t)OzRpuMGNhkt
z85u!L^6rlt*WdY|)bO6z)OZGJQ%RBGOYus6YEzjUo&<^lZAB_$gacRHrmsrPSB@??
zR;QA3_;kauE|p}#u>+1`UdN!Gfx(pwcxfiRAhiKr^Z~_DBYk>mgip()&q;0Pt1{_T
zsZl;ClfEz&F;9&k4tJ^<zMO9)pg4LW<McW6g&5HD&7<m#8Mkr%vAO>wf=oSp4fLz=
zNgsW#&~M@TcC^A^`0IhPRbGyMa1p4wXdXLjb1_Sd6DIkr5q|X7s9hJ~%+&)s2SEnc
zjsGJd9H_LSexcF_>n?go?t@1Sj9q@IJoV(qf&m9uxZ`>fLIu)a4K(e?@EddWfwBWU
zD4L8x9!{ZY{gwD6!{wJ<REzxm#(!MD@L|}CvqrhryTLk;7+aRbZUr-W{;>NI^!R_q
ziye8mo}KY5+_J_EgRcUi>K{FZvB`tA;9zcDA+Bd{4nldae!bhoeXFB_bvJYen%)^4
z=-E(bCIRBj+>9p*<QL!X6fA4v@tG?Tg|*dqq%BzXZup{~2d&r4$GTD80K3f5TSNd=
zRbq?NCX68>7!qx(GJ;r<!Vt%=JVzX!00ytlP)3gbF}Axv^#_CIWd|`pOxn!$TWZ#&
z{DgDKqk`$ri9En`i67&e`}`Q++~>#m=Ds8~mTB#pzZb7}1hSttmi&@YSufWWC4pOB
z>uCD!>bD1{zgG1*XgEWH*3+GrRQDvotojlLP*s@l7@uj)S(kweZ5`G?(v}I<oYjO@
zpK#BcWVNJkfp-!jHtUML<ITG9k0XwL{ISa4v&y@z^5fQ}pbIYsw0f#rj?a(T;~Tt^
z;IB)8d_gg0A2MoIq5djK8_x0Iojl%049y_dn}Pc0PR8;~9`0RZ4dVJvvQfJW(^>IZ
z@=Gq}gnUmGhs$UJg}wby$8dP{1vNvkQZ&oV^m`XE-T3V9n{-0%o`qT)TX_j)%Fbgt
z=n8f^r^|92^JVz~YfHd-DPX;az6`5efpKRTHGc-7{_wYx{o|%28#S!LfpKoHQS&U}
z<ATUQ%`WD}8f=!WWUlyz8JZL}i>Kn6`qqhQfVuECjD_O}cv(ZvT95ZCG4278z?Wwc
zF7gJg@1gh1-7imY{9pFoJ}|24+WXIt01?qMX%T75)oGjB8KD{~ZHZ8ukpO3O0`Y<x
zT2zpiLPe9pM5&?#l2p$f$I?oxz1OzfKDWL0)>dq_1#2ZCKmgx@Rz>k8zMv;WCHTUd
zko-R1ea<8^A)?;iKELOWi)QvYd#|(h+uCcdz4qE`>s|t9e2BkcQFHG47SQ;<=&2W0
zXpG6Yv)5gEx^mw^%jEYoR#Trl275=KlI&~8V22WV?J@lDSIpq+`}o2n{8|aUb`*>b
zI7I@jT&9h;IO4(yem~UHdxqY#_|<%)pJeCKcE+(q`qi@8zt$LM8<T~?5YO6;G=xi5
zf~*EF<mG8QvVYXhn9~(&?CPg>j=<2w)n@W#PU4@;#f<JbR{i)~X38Tb|E8YLX8`QX
zD6BxU_7`EE*?ogcKksgu_f>;b-fxvyA>pJvuJ!Sru4#qCiYb*<zZSkHQY-f;@7mM5
zy1=$e+4Et8{q>bRS@r)VFuspVB;r(Rys#8BWsd9Gr;zTIJfxBr>C*}IuUd7t!f*6&
zQ<9I@3y(Y2)IjSdFoadA@cF4BuR=*JTD=#+J8igu_HYJMASlDC`#v9{OMxZ4mbh3-
z6+MJ?;{VfzGE2A-rzFH57vi9eEhzz>6N-sKMp%RWA#ipV`ks#p6{|g*XRX){dS!fS
z399af#L48`R|$mKX@<qviO&GCgunUxRq$uq%{4cvl>9RV0OpHUK}`ei1Z%utwHK(}
zfNvqtzf5>3=z!NqCD7WHV26O3Yox8#KNY|f<$zUR#vIUol4z`_r%XpM0XYTs>{9jY
zI272g=ZWj(+qp{^#EHU?yO>T^da1Sy={yJ{8E-1FQS{i(c_r+rH_-;?7U?0xLr4$w
z$GJf(nAfCh-!S5YtH#&aDidSrvv(18?<a;K)7p*mGX}kQA!Hfih9B?AC(viUs4@AX
z8-P&P&-y=X1Kz^V{nvnY-F-Z{I{;Y!<Nrt}Ea^^IPlEfM2tRZAX}iK>p~7_^?-}FS
z>^20W&SO#MsWb!4qn>H5$PwM6nIh>KOc51<#}yBqmLc3ymGgVJq8-?h=88zT)zf{3
z7)y>Tw2fIp(3nG%I)}Piobh}sMp^2-UFsZka|fsu#VCVnGqbfTU0L!ve{b`0(@cBZ
zXKC3!dD}e{EL(q8Vf1G@|2Yh)Yp`<zI^fg0y@nppU`7zSxB(ra!>l3u1d&onw-9D0
zf3{Y=JK|@eV7sxsAN_lRJ)1k3n9AHr_T=h!-Y~*W{!Zy}$e{#)5zda<)<AXPsC}hQ
zTl-fuWB152f+5THd7gY&>bO7Me|=8W8R|bdjmA+bAhqxJTT~KxCze|JCnlza_vLAP
z>O}4Jhn<TL`wxx!+Bw<0F)}tomd*Jh&Dy3;r?I+4RFA?YBt3_ntY9l(X`x*f!CTr9
zZpG@CJ#rFJ)8*pM534?inlseu^>2!=g$#4EQzP$`7+I9~9LxaoS_z-?X!`a~sC*v6
z#$D_e0b?AKhy!ZKbv@w^0FN)fT-en$c1Gdg>a*?S@Irg`P6q1E^gtb^fqH%_q1pB|
zGjxM?ZdK5D%Xi06rY@w5;f3AC0Y3&AvWqX-giFW%&ae=8+RdjU@w8j+c#7D@2xF*i
zY9Dcwq{887I;NfdAdoOfy)vEy+U`6?@uG1)9ZcQ!Q(5yR^kx>zT4yIu?Y<XVQmW~P
zWJx~lIGFqQdWfzUz5UjsH=|&)k=&W6p||%udh5cJqW(?M+rN4A#_bLOjBwm;dfVGR
zx);3xBYK;Yrniot?jIXP)mF-x<o#na$n8kKMc;_L2f5t`xy|?^<i>aqVO^J|Hy)t3
zu^zdN?V+}&{Mt>THiUkt%}$18T@+6)f!gMX+TMseujnLsFKQcau9^_FJ!+`!b@web
zasAFrY8!sL<bk3D+z4uWiyML61D~fyhj{hz>FpKh4JiGD=_5mL(XOYnI`g}qgWwiI
zaK-1R#_dZ}8g#{!kfy-VP@qR|P+*$gwos}xz4>y}7r@1;$EUYprT(Sd8T9r9g=~XW
zN!;p#-kM~4YUs_I6g+x+^s}NjKaWOU#LL^j)J<<wrO0W{Ekc2VR@Uqw8TopZfRleQ
zJ5wg2){D@D>fv7GMYPh?Uc@XfVoP^Kxffwrox$-9FXHxYh%>y1n(l~yG7;C!_b^)C
zn?J7wytg7R?HyM5X=Xm)TWzoM`mLsUE6d$yX-an&u-Ws~@;8)A)Oj@O{LScM_<Ag}
z2zMq;7}d~KK1_Y<M?}|3G$?gmW{FBl@uup}Yrs@Q*CAuV#yWBTQtG@k!Rc_95}^S;
zza6;`bDyYjgp>s`-3KZ03cgzf2Q)vr4xuj_+ZjISFw|i+dZyV<NKapFYgEVdT;{L0
zv-pKw%BpvC)Opx;o{Y(Zr3{XwAF$bKGPGPa`&~PU|L*bg&rObAV|#nssKeKYk3<6<
zrA`AL7LYRwwH4#Qr(f*4AXw_`iaL{+?5?xf=PhU3_K)J<#xn#f+=1K!P+Gj#x@EhF
zWA@glv)gHp?nh#hA}bU#3uY*{D;_Wh1RvCM;3?<r{f|eBACFpPkEf=6Il#&?wzhi>
z?mQL$?B-E##y;C(OeW2em~Fb$=z;d^eX-QSKfY_Qum{H-i~HjNV`UoH?CjtFbhP*o
zI1i<kehr+6m#?#VDc2p9C!P5#v)8pXWBw#sE`zk&Sf<^^32bguHGGDjt^CYm-*G2@
z^RPJG$zLbCkDU)O9kwzp9;oK3noaF2clT^18Y|PvE!HRW$adqHemvapm|eWPdQj9k
z>^$FgFa3$mvZv0ag96Ux+ShWCkdN1GdHdL%e7^k~wVeUy`NS7$JNqSqT%?^0Sijg{
z{ps<*({|U^4}EtI*B;1Q6eq`SeM~oo(!k5n+NH_EAZ)Py)DURV+krIt4SIW5X{FYm
zTBCu#vHU+=+_vmN^e%&?bwL@QQ@gZqbU$mAK8FB?Gx3P_Vb2343qmr^ZomsmVcWx<
zc7_gj_c@K{rt(HXRZ*vf??e5Tp07!u)7cx@Zcb}OHbjdbwc;)4SVomPg@_f9ZB*p@
zmWJ8hQc#X|KM8Fi*R@BC!lGOLF~H7y*|3xPGtsm~yEa3W&hu9N#n79?@-evJkQyTB
zOd!m)oe7NH#f*;e!*mFVUEs!pd9er67kkF7yZ0i=IvZFs`dg$ESmRF!(Gm=&&56&N
zY19*A&l;W~#+LM;EeqOC*)W!H%n&bgd*e3SxiDlGZ%&+U_m_szj&#`f|G>LR^c8#m
z%bcJc$m)UXJ)}OkIe(?etBeA2bKyoiL;la;Z@&G{{0Yt@rF2XXHo!?Ht_3yIm#h$B
zZL{%<G|+jgD_Gwkj{z94PuK+T5L!YbJ$aM3|GBl}euJ*>J^)IS4-~Fe?#Y`BPfxP?
zw3DB(lRwfc;>}<TDn|boBFbQrg1ij*jrNz-WT@l3`IL_}Wf-&~i=fEKdB`Xuty@Ns
zO|Sl>$R(;n+)mbL_Bl5>{LV7x4b6;2@g!O$<Ao<v2T+#$y@Z^y5x>_=BlYA2NjWi3
z$|-k0^=;JwK>{Zgq}(tvHF$&sNmwH*wOG=8CORtDcV{3sOZqu&)V}twai38)pn)iJ
zcC}Y!%HN)#oEDgksO>ajk&In~U`yG|j7AW*X8hTcxefebXKJ>*{YejvCjm7Xc{}Qf
zK&bdP#gdb4shKd#klqrNf^6GsUp9PARo4bSi8Vagp1Fpds=zv<J(ruT5<@uyiNywn
z5aLsN8cf?KPK4PF<Zp^+l7Srq;u><!-B@9`J?je~E<#^fd8s($JaNb~U~sU+POLEu
zJH!@aa|s%Pj_N0D2X5UbBZTT7vLj8^LFY*-)P7aB{9)I=!Z^fcW^-b&>bSF-cH7jB
z{l)tM@9`uREioW>$c=Va18Lra_sg+V={8T5@l7gm<6!5>+^gQrz3M>jRfltR3A8Ty
zJY3ygyDQ+$y$=rKiZdy7{*NnSgvOvYs}|b;C?sAOO648T9bc)DEbw=kRYH-SJ3m;8
zAEMc0FqmjAQRTz1oofe2J&ygI^yS!~bn8c%1O{p&dZ4C}$?eB;9y3t=J-u-M+_D^|
ztJ(%ig;upQcj@E0bGCL<5v44uJ(Roj4`NPZyk!SWRU=P>r>iO2)o||Ek-xESejZyx
zo6$Cm4;gF7#@?tMJA}lQO3c;~tgrbc%lsx8`8QF?bx-l+?l>8^ES15pP0Wk#qo2y$
z?bG@zOTMU>;5-@^f6)NX&tx!Td<Mg42_3~{HuK4F@msI39)~%Fp-)7bvjoLGyyZ0=
z0lMStq?1m>j>b#O_jpTD4}JPgB%IOp)DYfvCt_6(eR*7@G|4KHjw`}dgR6%znY^TD
zfnJvUwZ}_-?eUT~;3cn$m%M9Oz5B@JMqzhDczi1Ubu^USyrV>LpDT4*#XCL??--Hh
ze=O|Vztdfw58xMv`z@LnjgU#2Po(wjAA?_P)RO5j!z|bXf**vToxV6l;#&Lf;S$AZ
zE+P8(pW_dY*pW@}hbP*<o5^48+Ffb>FwksmB|>)XkwD@T{C0DNESB(uVF@F<S;7;V
zV7ldn+)Llhz4SotrH6A*$>0yKh(BoXg)n;m=@7#thFkw<hxkKy@putTel(Rg5MLhR
z4?PONG2UvAA(4?h8a~0IFL$x8U)V8Jx8gK=BR~VXbnG~W>ny3)G^u)HnYq=BWiyn&
zptP(FktTn{`m2a{F@M%#ZaBDmYL{j2wfqg`WjnZPWn9FZ1F>4S23yXo_BcP4#~qTb
z1K?YE12~jg86OZ-P&yz))Qka4xK%ByVO^fbGTxr>-fF-J#GC;$iPO!T%cd1p5wUPU
z6=yZ8wtF!oRwf3N0P>U<QO@I&-eZa23~t&0xTOP>__UW8B4X75FO#Zmiepon<OYu5
zF3T(AsYy?zg&jO@^d4J<u+e*NA$qg-*udi^?{SSF42sf1a8@;Y(JOh};62vForZPF
z)|a30zdxwP3jc@kUBy=OMoyU`=~p%0!^>7HJiL(Oxw<rz4emyEc}oDEHfSY@&jBgV
zq}v3EH=HLp&nhVVjYcZ)>px95ScTsON}gYL-@^%H&J5$6N5}9~0#bMD1KknZW`$=I
zc9`@I#l2;P4}qs@HQr(lndrT2kax0eUD#?qjm%%eRi_?8Qp`QIEu>hHB@I?Mf$J{L
zZ3<~C?o)(u3+QexY%n0@Eu=nWO@BxU+euhexS9kWY`|aROf)EW$-Qld7hT>#VV4eC
z1HwU|XPd<ebFnP9$%f|?a*+68wfoatu&QfhsmlS=Xa}i2Re#27O78Hb26TIT5ZQ<M
zip3U}`BD^$5PpdxoGK27ge$nOcTQnBksIlrLNaWmiJI~EDgewWD)7-7e$I=2#;+g#
ztSX)&`du&jT_2sU+2tbEFv=5f91%Vm&U6$F%zdk^)KHz(%6U&+aJz}~xFp~ti3CcF
zBO4`c^pf_H#2GJ=Xe_TUfZ42uE+Q8k&*Ci?zqyB0=Y@rIU-%K-^tzfa)Sf~LrpHtf
zs4FpuK7U2-Pt?#>#1E&xCLFJE*KZh4P}3hK(?`hWC9C3f+Hd3_#S7w1i0XZ61&q@1
z9#sLcAoP+su9!*-S8`gNBC!VdvfNH<sPU<k;x`MqibDHVD9)PxypM}B$rUfCwSu&k
zP}Kl`;B9WHeOg!dwyL<(X$>v}ChpyIgF(r}xr%#J7YzIKW7k-bnV6R>SxvK*ui>Yp
zfl-2;MR_Y!jy*-#ziG^KmuoLSXbz;-UxLUJ(><;YCH>%WwFmRKYmoC4+y*4nm|M&w
z-l%i@`Piu|tWgSnDps1%qiwx6@IJw_80elau!bDfkTHLppBl{4?4<#AF88bJd+qSD
zu#2aL7tnCy20@x!!y7KuHN5;5#ZwCh1>>p2pin%;t-SHnoIxd|OdC|fWY5*SJe8x3
z=YHNWKNv)CNiY3g6;ItfsD=!zuy<oGwgEN^vIe5=M|4W_C%XDE&_UH0oA~~zB3v;7
z1f=k7F*&Zlh8DQ;IlQ2ixbMmnbtNV)(~l_r`Fj|fl`qk)uc)UFLrYwuUD>L%w@lhr
zhP>bFFPS63;HLLq1L{+PdWUh{#Nc^Uy$TNuZ^J<oiBrUbw~NHdEX=LdVGgf*7btHa
z_m^~mGVHXXnUt4-cnV0|Zs9p>`au=tX^8G`-kr>kLcMq_khiHy3nT|sFS&f)1N<|}
zw_ViKs&<*apSU)~aRIj15;#pD&fKf~g2r?UaFW3#ciUattolEp)1P2P#zw8kT<R7X
zz{^1Xh7?+nPg;>H6<5UHr&xmc7fsnHTC93=IEVRrIi8F_>Ji}~)(9R|S@jQ?N#3V9
zK2G{%=iOEXed-ermnRR@d1SSGqbBdH)qV{b^peM8&LfyG7<J}hl=4zu_sFt(%iP6e
zKSEp6A<R?UY7Q2nnMa6cC#A%F?FT4Ka?6l*>{0v5>UUaor;&S^&Wn?0Ea^3E9APCi
zit`UI#A;vbG+I5g`b<>I9%z`<jqKfBpQkl*ek}FPyjW^}9;YSBoabY;yE?JPX;>L`
znxoE^<oLtMv&_a*qjn9h=d8z+!`9u%SB*G*^tPP5WiYO7I;*#FenA^e*g27pF(fH-
zwgejE?C(Vmpf-E7{d;j|ryY4ek<Wq*5j=4+wnX=1<2!CGerE1Y{51y+VOKL_NbB5q
zEb@L?D!!Hi&bVLe^1p(@;}0s?@qG8a!%nkx_w|PZ`;ow-`-_F8Iw#N>t8HX?o-SKc
zXEpj{nJVO=wkl!UmwhSrr@C6A4#%du_kAV~lBmZ@z*~lXO|N}QcOeg8Xg2&ny&~+L
zIXF|;Gx{H_3o&G25sC@RfLQXPd^@5W4sz{vX)PA|QIrJ-1J``wK}anJt9YFTDWref
z82|@!7Y*`wYRs9K?|Ic-dpy}Si`uBziM9igTwvQyIXbgXY8O0d^E8$9ACG!sId}n~
ziA7JPXGMp2M7L(5!``!^i+H3DO*ASPJuA9|N0yEzn#}a9=yD#zesqQRtmv6MA}|~1
z`QEdlt9TsaM=$Z7iN3Xln($WQrQCW``W4Nq#WoOKficacZL9N5x9AyPS9L(a=+ejQ
zI5mO&CYvc%-4M2xW67&Aai18J`mXUnu6=(-b8m<jyH@Re%4O{>o@OsMmOMusJaP~f
zW3K5EsTH@yi+5Wq9_`OIKye#>w#uw?cf|q^TR&=y7q?nJ+K|63ZcS*#z)t2Kg|Xs|
z){Vaf_`GiOlW`61Ac(`9(?3ULvWwB19<b`3VUCEWCa;Y}4#w7|)r*|viS2Ji!#F^S
z2d9ORl8fPfViZ;y)g$B1!M1NeB5`RNTd^?FRv`%(_PA}oQEh4u1gbyD8%JpT$AfLp
zLP)&zt3Fu?#d@^rzeS|b{fiQDZ{?YxOvC_UBDN0`F?$$bH%}}9CtPR(V_`E^D8~e|
zyS99EsF*wgtk?*4n4*?DJu4(-gFQ;0MNuB&5RJwjWg_+{6R}5`h&{?g>`}five7?N
z6kxDN`F;cM1mcnetG!?)0q~k-^e+=$OdY@jL=(7PutPx2-ycSr{;2?_D35yfDChe|
zV!5%v#Qe6~=vbM|oo**5mBgG$Se6X5f>W>}xfqj@%b2AK%+JLko~8u(83;eMW2q@L
z@z%s?`pkBsy+5-GrqeqYgsg%Xy@O3+vu(=p-}t2V5RE@kJs;^j%KlQChd@h$u{+~r
z$}yvWg5;fS1>+}EgJ~yIg&8MPkQoL1Oe>%`@dC<&?QCs>onUC+ak4jh?<i1+4b926
zKCaNm^Yt;kHlu)_X$ACMynxOtEuboVU9PXyDcn#}pb*bcC#$b;I&<=B6RtXWjS25K
zxj|vU9R*&vs=&k-++48Q#Mc$*J7|5au-2WduCtxG0)@h=9u-#8!NFB0oBj^oakAI_
zcNBQxssfJ=ZZ7Z$p{~GC!txDf*R}X5_OX#mG6pT%0{E40w%P4Kp=(B{1wpezf#s!D
z#~E14)EV&J6VrKS8(V4I{PJPOU}bDdmO~<s`-2rrB^fEnJ_*~T=5tFx4k>TNL)=G%
z@<yR-C?%St@4q7h#@`2G=hzm^K=xqg(uwENhv(9T=Q93QsKfQo6apB=---s_3B2A}
zvD$kxebc<6g*W}ng!dYLh1c#Yyk=ivTD^Hi&-jBIQBeo;R@sA{pNeBPk{Z56JlyPK
zU(-*gK+lOem*%5Cc{P^GYi9N|c1FgDMAQ1USXa4g0_;S8CNFc&UTNl&X_~Vi<EWD-
zxs{<_vW<4=e5UWr^UcgoFS^qwOB4g7eNguaVdfrY`^mk*HM5`I8(b4R=XVpMV#%+?
zSVG)Ro5YjHGqsP}xiXpebB-MTBI@M*)PHn_{FGgTcyZIBSG<YQ7*8=F)~GSF^ubKA
zPngq$M^yT;0jvJ{f$R)T%^!OKhO+;{b8PGcOe25FQ*`Ww#NEl0ZQ`c#WTIyC5TCtT
zhM>%HR-JrZOTQd8hLI~totzr|E9I@)4bL^RokUCoPHCIj#4*QI87ObN*XOYB?+L;)
zWp!1h@6yv{JT<yOICju|Q&J=UQ)VGE^)vtc)bLkP>+R5Hc@;Kf-smzGme?b`oVKv+
z#KQ7rSy*;rVfpe$u&|u=5iKkc&_05NrGfG+EZ0&k&21}>ZC}Z@3mXpJcshoCC5XN8
z^bzeVd*HqC^kMdu8c!6Z!y8ZZtJ-<K-%lmo)^S$dJMd4E<IXPYraww55<N4DB&6|4
ziOvfk9Aa|D7U_%XItCKfK<N%MhXj!6#1oI0j=P11@_3@&6W{W9qInK$Fnf~8B$(YK
znB63p-9Wo}QVCxnnW+0xm9V=vVMtI_!35wY05<`+3C!6O44^7kyXm8XsRN}~4q7Wt
z*K}bdlpcKpc?~;OMq>S5hfaV&PJls9fI&`xLDu&$$nFlRZ=fPpLA}Qy>sNbk9)qm!
zVURuWUWe9u46@#1ko6vetj}hUbm)6leH;eaKDTE$`kmS{_|&`e$rcXE;xv1O>YW^E
zw@i|06VrjFscGGkpuRJGJ_=VoF((H$(|xSzr$$uWWmrX7<cRitW2s3jK2gTYBF^Ji
zQ&NTdVvVoT;@1V^0b@8g(`|kml7gd#?c|blY>9ghu{s<<0*eowX22fcr93N~m#<<M
z`3F&=`;V|s9CLQZQ=Fo(ZfT&5=AemLkJWRyWVK2?yU}UJP8+_<sSQnnIf;qRQ`X%b
z_UyOV6ZB$E!&cjSIw6_3t0?M>zRheQ!;8*HE`5NAakt6DCGcKysgli`EIdk_O_<gN
zI<P#e&601Em%STxhToRF?0^>>Ho>X-g>Jm_9GebT?_dGITAS5J{dx4@g!;!h3c{^2
zl4IC^W2M9$C&VUBD8*`{HDI>Z5~HHE-^h^&IkChOAqQB&$;fRNg{<JE42>)i(wu9L
zav;W9p&e&6E82mPYZ<vm)EtBj6P%W4<Upo;UVHuF97%`M29U+HKnc<znXoB=fH#0_
z*pWcM3-B{7z(9-#7>RkZ^Qpa_7Dhl^H9*6_c2)%xA|MWsfH*(`;s6PV0~j?k0{l!1
zNI(oQcuE8CCsu)*^3%GV<&7F5(b|SitNuH*Pn?Zr%0DcgI}@@nR8-=QqOpw4UT*2w
zHD<uve2dZcbg%QL9O^X7{+XWGP{M1*X_qBE+MTv92(or#Mil!!#&XnG8RTW7dheiz
zinUpn>+&K)(%|9Fqt100-%+mv(C0~zN0rdmNs#;c_`;k7xq}3E()*oVZS%0zQW?qk
zHr{z`Q&YL8D{=fGlh}syw~Xy6_RBN%6%#q^KOlX4;u#v;Q$G|7TQjdTZ!h&jf$eZ*
ztCSI_ACf2<933dX042dx)Cg!B9JCI}3s4kHWo)Bba8NfSFJO|N3TdEIaL_X(FPKk=
zz=dMLLBWu`0JXu?3c@17nS{tSQ|A+oA*^B_x`v`)n7+iQBw{7DlQc0#{a{`V$@S7E
z{PO|RrAC{u{&88`@8;+JJHWLrN2hHr03h4G+&1TtIDK-&J3Y@7Gy8T?7Up4?MtD<D
zw*?aBN6xdxs0c+~iD!lMyZxOUF?!V;F`C1C>C6q<jj!bCo;l|ds(%T`i#Bqt<8jb#
z^Mrg*Q!zv-<=4$we^2bFHw_P^2PR<I6V59#ED%-kRRfJ)(d(*Ppa%9OO~G=#sOyMA
zIg2p`d&A4}QP}oh!q6l%yL@|43h%)_&REo)@N<*G)?N{?$4i!>OPk%3sYUn6he5n(
zgP~|VXDCOjpg1WGc5NDK&R<zKe-jcjgOl--X&!|SrS_nlE~Sr9H15-wc~u9MUU%n!
zRno<CbV6_I|1t8F_D7(5`&SBMe&QW^h?-MlcJd0_6ewAimc;Q5BEJ(jZrJ0lK9|-7
zL~=Xa?#>uyXEMMs`<8Sg^Ni^Lru-?T&YZ%*vU5WD(kK(=*%E(JBmFa)d)&{9oHV!T
zXIR>5J|z1S?3j1EweL|oXICi}(u3OzdS!uSje6OZS16<!3_4$P`(4ZUa-J<qo@X16
zFLzUxSn@}=6KDzu>I{{Xy<1|qfAL(ulsSKkrJ|gET6)W2O0dE%=8BA;G}*<6^qb0C
zj$NDelLot(ou;2`7}Zr;x3l{4(ojp8^K$hiIE;b+Pv`GD@KCflU_Dj#0~2IbRHmn?
zdcy2-ObJvnrUci+%PuBr*{P@_LTFw2+r(ePf$A-Br@4KQDQEi((8r1oCZ-{)d4647
z{JL-<K@$pL^uydYR{ILZL=3C;LCN8`3>tpFGyFkmsuqQka;}p1Uryfdlcm3r)0Y3r
zb4q{DJh`c#ysfmxlHaL_jrdNvZ0GISaAX<XL@;AL#L^lcR8Qf9is46%o^aQcm=<k6
ze(`apG1O7mn_VAEo(Go!Xw*KsdZ4i{OK?8p05_94Cwrlp_EC;w)vzcogR!MD7@KiV
zo@Vz9{gr_>2K4Et54NBOUG-z#{VQpY{#|4G7dq<IzZ-}z<2Yn0*4@AV(yM<5-(+K`
z`IAPw_&|66TJ_)O!_v_9GUpJ>KOr~$T;caK`hu63;JfA959#m9Ng@HZcsvs``d$&s
ztZV{kJW%~-5divUDpHpEr7G1vha&jTB8Lm@)Mb!^{QE!-Eax#WIf4RK<gm+-!~2lK
z$hFqpBkvE4gd7g`TlCqL$-%c<cMty8z~G-J<F`W(@qf(=#D5MwG(is!Eov=G-oI8Y
zHDX+0tZ`4j*zCuAl91LNlE9MfsjG{+N#bYccq9?=tv7V`$f{cf_jSMgwg&Wz?l<4<
z>@oy%lNs3<k6E`|M;5W<VD=OY)yM-$KK-qY4d{j)6GJx&eY)9<of(7^jqGPP!>1f0
zj~Zy`<-)`1Cqpk6nu8erkqhJ=b~K{3&6wzJu6_iMG|D!X{3c5pwt^Z=VaklD+p{t|
zOk8HC#<MaSc_ZAXxa}H`+KZd7o@A%KOTSmbqd0hV^!UC=oqGzM%kb{h=vKrs8RGPe
zY>oszWf#AbD6>-+1?*zk6x!A#BMmVuA{{;5_jdb$<j6k<M&4savhTkunRi<t?}y3K
z+wuaXKcw3>^s|efUDW0YVCy?*^4g8z9>J^0y^d~ZyVZy*Wb&f>5br(YYV<Yo-{5bu
zjM{g<oJ+eWfr<TGT%kEkBm3MR6DeY}ZD%|(YbghXgK;cQ@r@it=qwy#6?_Ysbt)q3
z`ADp}vDBZW<dgLF&7Mz<&$<uz8i{o(0_*w6tK-q;{z?0W`usB)pKFNs{rSJC6Rm<<
zHBhXA<Z2Vtt|S1}Of)V+rOU(oO4NFq%)aXJZ{8B|RwMBi7|%ggLG2EM>S|kfZpy$j
zlUja2k;;`+aBGPPl0_ykuFxpbeBOlW>lvu-6hxm*`){|Kk8i!n1j!{Ps1*)S=_V|p
zdYdp^yADD0jjGjrd~1sdk_{%PT|)pWUDPI2|Cxbm49_2cDy*b}TXn4xZJuue_Du{b
zHN8;j-k@};Y6Q`DsjAG!x6U*{vfKo<A^}RJ!U<IYuW3|U1kpDt9cQLgcbXu1y9sJ<
zA^=qjflzG*Wg1nGN`C;V93>UpYJ6%JBwIaH!U3udzpbvzKvnU<sB90_7!xGJCa4`o
z04mXzN>!ABYPBHxuHZ`Z@vSu`NLHDkRx}POwZ2g80%f{_t|0nGwZnXT>lPCv*P5XA
zP6AM=a6<L38K{aFj30n1q@;pdgC<Dkn4ngT6kEP|3~!(kBZ+5VTOy#o3ntI_V5=}e
zQkNAl28G5u@>=fs5<$K)qIs5j3&;4_eg`kg<EQ`g-{s4TVaI2Gb{jV!Sj0^prqtmX
zu~YIDU8LyInbAQ-EA67enbG=wa!ApInbC@wJVw#uGNZ$a7N(1aWJW8u$x6FuWM;I!
zpIoA7ysC5;dkRak18AsAssHjfv>3zJyQ~?jToxWx4HsJB(>+tEM|j2J#ahPYl`KhG
zFfYNhWSbSSFhDt%-LmkLn7Xv+Lc=Y9Rc_&JtBg<Z_#*Aq#hCV7%g$Q(No)Gk%HYK`
z)I({gAW*XXxx5IQm^WB4RXwZ><JiPamvziu*|x~{OeJ-^VWO_GtZGRYQ@LX+_>eK&
z+v1@dQb>MWjyPR+cVLX8=x6XeQ&_}pdPF@$(#u}-OIBn~VTb}y)XU1P$hT)whQE>S
zna+{v#Z_c_Q3=hk4b!4c+=#*B%QO&8v7M{)!2JaFI33`{%x(H!qA+e#k-3~9+(q;&
zL?4u0m#l?A{jkglsoQ2g-C{*<tD*VH=+EMCgzHxmT9F@U<{|5OGs*8cRrl2ku_M&N
zu1dGfn<kmw1n!4r7bH_klY%85g~FKUeJjEV4c=biRu!&&vm$plfD*VHTTBk?TM1OC
zR^tA%Lvi6R%62PDepWxOu_6m)H(Ht4Ktn7`6xOH@9=(TOb(5~_C>@XwmW72Y)qf3O
z`WxV~czQt*9Tc5hJd3B7XcyP2z<5&yrqqyS<<br1!ugnw6O%3s42gf5&np`tuSyK)
zrb9$(wHB-5NsqojL9dk-%KECZbi+)OyQqhFvXKca-^;S{P4814E80}hAjO?i*r4y`
zY^c(A8@=yViwe2uL@mP=_7&8LZYUiP=7VX}&rqKg9)d}2CEY<+(l-qklErR)B~#Oi
z-Bx5|A(m>!Om~lppj!&F(m9+hfRHJlDQY>i!bP#Y!9t=)0A|Se(K9`5p;#2f6w9_E
zaZ1JTqtO=@uAx*|D$anw!ta=Z88TJ5o}$&r;w@EqzY1POk3g@cru#to9;tiO)T~y*
zGDU=HO=2gBSb8ojtf2&4W+>~mu*UBetAfl;Ip&jDhsYG>G5sxWp&=O?Gw8;e-l+mX
zwTgS+s#uDJE3~QXH=Bx;^Rz=$3^`ZOnY6U&Kbg^m(<OY3G&+pcEYkOZVOIU+=nk#O
z7ez*l-bpYT);2tZ_`H(F#i2li-F{+=cqrEcBYo0Lk}s}0_PF!Miar6-8!tSU^Q=Jw
z1~Xn@qMSPw`^0l<dcSu08@CfdH((-O<*w#u2bP5Wh=6)qL_Ryj0>y-U7U)?oTalA#
zjB|qmB+anY18K;|XgR}DDbB?~;Ql$}!#yt2ecQlkeEDFM)bP|rMWA|1AdFTG){wEr
z4Zmc1hh_rH<PysG%xaYk3qkR<Cunw+ubYvB#Te~A1~>8J4d*N&9s}7&t?BRk(FfJD
zGkJLdvVB;52+k|+`fZA|Y$jZ`iY#xbF-?S7B2{5X_82<uk)0zW?+qxdm4w$Jjo1CS
zS9ynPU}sItc53i4`9PZt^H4f_y5?>nR~Lz)60A1Cdht`T{z-!gjfs5K<L{X%X8N-#
zsCX%CbowDtVlxl0_qSCRte}l$?lvXtg$9Iz3=CaRPB}cgt<{VBa|a<zw0r$7QdVD0
zTR}5p@^2_$o6qLNO*lCa6hkt`_#LZt$4m`4)MqZmQ|6DTc-t+R_j<HKhFG`j<M@hl
z&ArbUgs=!&1DfSym?l3@XgFihycym}gKdEOHK;_4J(yWg7^bZ-T$3V+KGqln<@%O0
zYq$so0*mn8-!oz(qYIUoSYepwW_@027{or=D;s0dL}9BH;YZaUK>yEx-*WM>S4c)^
zT&C`gVnr*(mn_pCO>jUy-;^$DkttD0ca@-!-XJ9gMAe|s{InkpVn`COrKX5alfkoA
zg((lGCRWnkqFa^j&}yhm2JYBhhd1-=nsU(#WNeNXaP4%3Xm+F8R+E|G%;9!Gpi2X?
z3y@YpP#7pS!3-LPCz@m(1W=-E<2(6NSq_OQk~?K-QQvdCVYS~J6a_PV`oBS}F(hgr
zZ%$Bb6KQTGqj(`U`9}D7(hNnDu$=_yEzR)=h>LF+ye0|GUNO25K8iKtfg-E^o8qJ4
zZYHEzsY+Z%4|Ez<Qih$PbG9N=;HREkSFVSdB=dJ!736S?Vb?Qcexd3Cd|Rs*n&INH
zQE^TD5g?D|v*r+K4b{(NL!?N3R^?Siv}e`Jw9K+O&^Y|+O>wPd3pE8-Ud>O{W+lR?
zdB3zu17oj&@D3jkc|!?6%^r{*o4iMZxWRi40X7)~RgK<zkoTX|P&lGCD2l^0Ab$Kk
zW$hX6lj5J3>EO{nMboOnW*n~ZkIpmz;h)4=BZ!Q{H9qi4W#k{Ov1@nN#GHe%+BOa%
zVG~rsvIm9F_O+~BuxLkqx5BX+wNv&v8xh_%iMY^+M8k<S(CMs|0{olg3FZ7ShaJ4~
zi<<)YIDi*15~owkN*R(<kXP>E!<|+*Z7VMMyz&Snt2v7EDCUq2<4q&}h)!^ru@v5K
zz(B`nntqs@^JHP}L#$EJBUi7X9LP_kyrIk)Rgt?2SDUmO8iW`gGab1e;^h#DKsaH+
ziJKZgc(|&CWJKFXl)@;COcsrE9EksSk=47nrz~vnVCQvEiP#!`zSp3FX>Qt^W<P3*
zgZFS8Mu<ApLOV%*kpZfdJ*1e-chDx$7P{X%QUWgTDr-g_&D2UWO?O!aNcaX%JEr&@
z%Ao|nM)72-#EF{eji43)5)@a?Y1D>WXd;>0KdDjvEQ515ksJ|G1)1yqCh*6JuJhX#
zd4`%wEK&w*Wc+Wn?3bCoY1#1iTKVvRIXJ?>lGQYUFmR?N%;}<4=A;AS1#_yb{cA{8
zORXjZ4F{l@OvxX{GW{_cqpAxknNzsh9N4LNLw%%ND6(m<EmruuYFuz}<rxW@p*jfp
z9O=L6DZbta1EaQ&suZm}aHyd&Ak-bAEiIa~R(nm!&yoDRiV!4)V^7H5Rv2bFeGd??
z3j(>Ig0-O9RPoG9FDqKdA~<jJ46$i+U}Fk-Rvqe7d3bQ4q1icwHGFzl&E2KbN2^tC
zNy2}l1-R4fUaujIFz~LRHm46137N-BL~u1!1V+T;yp{Z1x|*M>)=)SY5|0xbh-<MT
zH?%_S9h{NiM2<92?3tUOYl~HXjyaCQ5s3{eDMZZMMvs}35}(xJWFugTA9mYUPbDtn
zXyVGa!*RrykToj@pGbp?X(7hNou+j)V4O$$O&g?w?(h!RJq|rfbb8i`e5*!jYx+)G
zqqNq((^{1Vm+A|r$*-aBG={*mK^=XkbtsLp_k|PUn>L7{*H>B#E3rX+CVeJ-0BU)Y
z+P3=yR68c3y-z^>nK)gGLO(RnL!LVu?|cK?S4W(><+-@7+2c4<U@vW~tzt4Ce0)#K
z$DAkVlD~BGXNKL>KHI-ydW{vCUZeCr+p=Q%O43&<z0Y>3n7*2H_-lCTn!XyKV)`1=
z*C@Tu{-~G^eNJysdY^4u0gs57D@Lcb_KhF9HW$E7?dUrlnxEbQ{!jLq{3)Fd@(JGW
zqOOPVduF<414BWB1plxc8oJ5EX_R<zI!H8!IQcJ{3UU@#O!_Ltc|jiIM&i^8$;jKx
zv7T^BGjI&mz^)Ph*?mtu_46QuizB|>{fpfy1L_qGs62LFD&wbsvU1ufV5`~zCC<@;
z_Bmb}-eaW0*WpYJ45%G;DlaA<Byo;C8*)fbUA$}+EW8?6t*IJeRw}^D;7b&?z>LHz
zx3T{2p3W*-;UI+|sw_RF$P}Z*6r&tKis6?aop(GHA4oA7{mfP-zs#H@)FNy4WWmWs
zt3tk(^${7&G#OO!yXq9fHTuqD+CVkJOW@k2yi{Cno@{9~EFIYhcAS|9wcN;BV_=9=
zM}H}22pV~XOy(PbrOHSnC98=jH{J1wTKChfG}Ncz2E#v1vplR;I!u>Uy{X*xh}A#B
z4Dhfdu&3a+)3rd+>W8U%pFrLb$Wwr97f7o>UJ=MU2E-M}TLRey$jji}O{w1?d|W9K
z#dlj3dsL(bBHt!)Z#q$WPbLQhv&!-oBGsnvrxTwd5h0#7t}JgQvM;nOBq~cfNNnND
z?Su`=vxzUI&ms|92h)SPhGnbLw4A3Eae2{Hls?D#eydut{yxZ`^L8Q5*cJ2F52wYk
z<xOT@F=slzBEu@@z_*&R?2pHo!*96Ai^*c9ULv<W7B9YsU|0C(l}WK&6xF|(G#08(
zX`wk>hmsuIKdbJKj1{@VepTSIyJ+o?;a!Ql_Nh*DTbHw&-A)e3yeW%P4CMC8&qv)4
zl}qg*Oo8>zUMMDR{_~gQ+b@L5T^`-b^;4psBK=S`4!-a+z~;nDEHxRUu&uuNX|&ub
zh?elj7GX5RA6tajv|%IB_)($Ei5K7RYNow_GQr;d19OlgnGG`&g;v4xHR_3i<*U6w
zlMHV)CWuynVx}n(z^4;0K0}b>1sxu6s~5C*K?6bNCsiKoOfM++f)Xz%@`4b7%43ed
zY)77uZBMH7${h7gJkVh}F76}>gLslEl%FD)YCM83>R4*xOxwvTB*cn~eM_vMy!FML
z!lL>&qYqFRc6?^AF?)-0N$a<KvhfIn8+SI!rliO?e2HPobA`JEH^k+3U^liiEx2`a
zcZ!X7N9|CQHrsllS{0>o^+dUf`ejyBhoXLv6?H&SOEaS6&a~70PDTv&JBqqED+=3_
zPPa5G3e)&bcT`psY^&2fMNzp|mQ$H9?>0uP&U<b?v5d)?(XNf9#k;I}m%%7ci$(Tw
zoO`Jot@^9gGctj+lNCW@=_Lo3LHFh-WJiUO9mWcd&2fmqb0km1l0$802xkl~qF1IJ
z&v6A#Ch~w)!A0}+fW_HGm<dcfKEy+n9>P3eus7}aA|9~7n|Ayd9zuF3i6zaFRnHh)
zT@=<sIg!Thj^_#@>Go;I&m>ZmtLOPdV!=1<_$nd|Z{V5kyr{%>Rvljh6!EJK@T%j5
zi6r*#R~;_|B;DyJ=~I%*m9*MRTH_}Pgro{3DUPq!`bj2C8?;Sqfuqp6W9JkWS~t&!
znTb>2^?X0(2xEH85!(I<!0f1|-nN6q|A}?lNrIE#ZH_1GPn_lMddxE>IpJ~4*&YVM
zyFT<p3bXEMeefy(?hm$O>+lI3k4M(S*5O6odqf2@pLL&iL~&808e#I~?C0d;GyUgY
zl#{b_T%mg$U+u)=@R|M>`!V||%}zErUfBCg|4GC>)41cM=i`M-o@wN_%Dd(dTs8Jt
ziro5Q+w(hY>s%$JqNDMwyq`mN2LjQ*LSs2`YZ8m^VE-pB)OnCMwYPHd<B6*k)j*W@
z9(US(`)+P%#$}QfL5m}^zUwu#(d3lOv#qO`^DggyJ|E4qX7rzG!llSI7-o!l+syS9
z%zy>)!uii*qQ2vVdLW)>rvV67Ja*SZ)xV?wESQ2{)6cSetL`2O$S3&o=a-|t8M*6{
z3(7d$l#vP5*PjD{2?)Pz*QtQdfk9<s{eAgA-962e?VtW^!lgpM#6;QHSwNdt#Ai4q
zr*suNPIw&{N>>gHdaPL2Bh~+*QVrD4lzgK}rrM_L)GlZW7o?!=1==h_TV~4lg{7c*
z+qx9(>5NV?Iz&dD=6;WFA}w6X_TP*xM5D<!mvO(%W!$(iq&hcnEa|0?ecTqVFY%pI
zc9U&W4-KufMfdy*aKvaz6%HAWUvzD3`N4Ynh5J{M7tco$zD|?9)IX&R)YKv_k~wdn
zj%--5N8`yM7*t&|)6N}@DcazEvEsMOKDCqIm^FS|3^P8Q^u?XaU@C*37B=HR>(KES
zc(vFL$h#V=f5-WaDir}&H@wH4DOK_0q?rafJ2RNpWAWtRr>CTb<l|~OUcBA@)IR-M
z;~V8LEgsntk2Kg0#c}e;ZVK5oLatQGwkC4L{l)Y9mPdk^8rm81$P+Nw`0O>3M>KgA
zjoQh;?OP_<qk`A(vu=`^hWJC{>liM=hTYF`7?5$xDJMxK;vX2y;$!m}t`E_<mP4HT
zGtZpCl_i>aK3UeUgNzZtc!Ub<jwUjpo*$A};6-Q18XpiFlZ&+rg#&$^J{*`Kp&0og
zl*lK_2ru4*??sdoS&|vyNfEUhLf-sd{RcZ>928=jTIL+qNv?c<UZ1Gx1_gT<OS6#X
zb&iS{dZBH+n@nuydAtB?ni^)^Qb()#>)T|VIpEekp>M)zq?yj+S9HMq89VX7+Ze6m
z6wkF_h22r9>zScmaNoxiEF+Gte8GL4xUyWX2C!4tGex~XA87U~avzo57&fU<uDK|F
zt9nJudCsN=;S#I<_olOPioI)q=1trJ&-dm{BZRoqSQy$4v#B?4$|<W;#tci>I2aQr
zISJ_CIx=}aa~;kflXyjU%0h%igxI#9PdUn%cF$*qE8|va-*Wo-t(s_>OUEJ}SvSM;
zjU6>}@r&8BO542vW=9=0(G0tjdE+j@VdXC;$}cj}EH&?Y-2Zs!116g50B}zuwe1`u
zvzr3~SM%=mPjC7CDY>|&ZtvSk@2XbG?`;xD((=39h4(#HhC?Kh-!;5guD(f&6C%Lx
z-zEW`BV*kHyoL$~(j~CZ_k?%ze6A<Fo9E#=nqhgu`@mT9q@agRvicP6zuuZS(F~)2
z-5m9jjC_8q6lC&6SMpJBdHi@u_Ze8s7#fxs%y1b-eAr|x>>IquA|gwAM{?jRT45pq
z^T*RKjjXKWs76rMfI<?)O0mp&*B@AC(Pr@oB+6MSNJs<g&CK%YOr_*yb1EtByl*^i
z+_0#7KI<(nZ_bpL_hib;#u>kR*M1GBv;>Xx`WpV}iSmR{MAz4h5PHO|B^DucnDN^B
zUiEFtxX279_H_P}ikeXrPgbR+Ppajf2(fWISL=h9aON=tv->f2lHDMu3X$_`w)5p!
z@*1kSh_{RLW6rb_5kM~v>0uZT91oawVu**Z9ykx^kD}W)YZPrN#k)77sEtvSacn13
zT)PUu4;e*wY81^AoWwO`<t|3iSIzs2?xkBkU=)1?0JpW7QS^B;ij25c1a#K;L0ZK=
zmi4clyzzFLh+9(3`nOf4Y2K0AxPu%&>P^l)^gV|17(3>m?IvT+_B!*;KySYJ1GzKu
zIW_6cH?Nviu<dZYW2;qXgkv-3=sj_jH|Ln=(ONNo-VBxraQq1~`C}@WP{Lmke<A*Y
z{9!~_$^lF63}QP-1MG9=8*i=2Sm2w!7sVeJ+B-6P5|r-^#!PxwzNeB=6@L`;12bie
zF5gf?8z(U3UEaY1L8}+EctL{~tnq@?Ua-;&Y6vt)GjM#w<#6^|yBCfQ)roR=L5mkO
zc)=PkSnUNXy`Y9bn{Xb(u<DO!h6r$(m(GWz7uDY6Ya(46pMhpQtvfAnn`sX3?)Ii&
zR**MbLt|#Or-Kw~C%-Rp^eB!+HOD59x$)OC`=zK^JK~?*m)~Po{B4PytHsqG<HqI%
z_u6-K-VQH5o47sdvFaDhH4H*ZvCXP<Lh-KZ+gX&D)oG6EWiDzGf8b^cpFbEk&+c0r
zMZ-*&#$Bnr#mKj4<4gImRQyC7xUxT+Oq`U%7`mVX*GwmwZsO&DqM3n5U!d5V)Op0_
z!;ypb@<W2E=3J4_Ef#Db6EYto=TK^+vFMqsn^}oz>XQ52J@AN(qe*RF(-&D$M@_!P
zzkzL>DL9sDx-Vl7Y@@4D?=QMvq*{Hh@;U)<f71lp(B<^{6zGGzck}O$#6QYEI{&zs
z_BZ_FD~5mIT&5NG)wrcuOs*0fn1fiEwR#?9UX{#BBeio<DUr~k&woCTXJl3rj~I%^
z&9}6t;_jUd9BoQ1)M4zO$@*S0ZBP1(?oFh#AhMS?^_BXatW=K|+)NL8@<?x7<9ujm
zhmMYU7_&0_OcuJtCu6Axb8x1WJ{WL#eP&s_CI0f}niF~;o&~~3Orfp%YteDZi-c*p
zfa}fND5^>_hvaOH+lYo_s@_ymDlUIWW{tnz<MUcGJbM;s=tDkae2QZoYkWSt=L5zk
zyp7K39-=?)9-=glRsWFC_Ax}yH#FUQh^7W>8?bwrW|TjByowr*9<P3E_ju(!W4x|=
zkPbL{yfW54XuRI2FOEH4w`#mD6I{l4onqc!bgzMC`W&y50dNmO5N(Y;<243!-{ZBy
z%=e=UbsL&D<D2Pzsu`}`%gbSzi%X2E*%!m4FvH8j@64EGjd>Oy;vC~Wx<-O?gZrRp
z6VggMTHrtn8ohmXYVa1FvoALaD_uUePr?A8n^_8_9nW+znDs8lGhI3zr{@oMJaZAl
zx{8xVAL9NAJ@rR$5@n#YV6|4fK&ASc?~mpG37-aN<h=XHj^Y1_T)g|pzTcosKGgqH
z54`riKivOQv*#~^!{_EcvXiqr&~u2;AXnHhftaH;^AL35{lRU7`Rw_4VPS|xGk$=K
z!#?X4>98`!6Wz+>k7=(@QQo9&-QDQ4c@y?u8NNuCPkH<@cEuY@V!4&f*Dva*tMjZo
zh-}dlHM}Eb)wrJcxB1MAN8vUO4bqluEfS}AncyEmU&EvvMzf-WbV&RgxMKv7ma7|#
zLuRXfy4eI+`X)lYZe7bOOkcWI;{LvoG0Rclg}TLn#;|V2Lms2JL9?#>x>a;5sV-@2
zyx)xCj3}l@?i*$yVLS(DLGd%eFlTOh-~&eN9RRqKNNsEC8L_wVezboa?<4{(hocNe
zY_w+CupF-R*gR=ij#1hL+fGCHey0BkQBSXW&tKAd)vgjR<`;_b^{RV_GkR6eWXPyi
zmt?9|{rDd3>XK~j>I3wXw5zSg%g*0;V|1V?JYfNtnUB5xMdir;hfzE-0qt=|zQF;0
ztMY_AfYLfuRD$6Xaus3C;Mv!y@{#sWvFa-^NUG3VR_!?9OyJO4&L<}{t?KYQP&5jV
z%q#l8`whwge`42+ONN8}pW_|9PB;HEB~f`aZBd>(+Et!#|2p6Zp{dIL^Yx5Q6*cAM
z{LNr9a;cDESwxvy$cB#%qM4Lw=8hBojSnbOg@LJ%ruTNrBtSl^IB6!y+6xcyb%Pat
zmH^dk1xndq|8XXbY|nUx=A+rH8(D)=q%($p;m)5&bAUNv7%-$%=Ud6&k=72fCZnM?
zhc?`2AR$op4**8Cl+wMLDD9}1NO!GlTcD7Fiw&^^c-_M5^vU`BR-A~jkZ+~kL)$&}
z+U=b^V_W*VY<UlDzF9B!+AnSFnLsk<4vz1bxdZ(JotncxG~m#U(NF%m_uR2!oi}$B
zaRL=t7~V7{bM8P0gth)ZKX-UM)l48StHK}8f8hU{^T%)vu%1pd{o~1w=01foUF)OG
z)W!5Iejm_2-%`Y*;iz^Bucq{6aMVsuUynQ=Mj30?SDY{Ftq*E;p7FG;&O4sM^+Ka>
zO<mtA)z8m&dD<YUU{lw#E&7_9%!=wz)RL?ywnbla=VnAXdlWM@BL>y2qDr!&q+4>o
zlof?C>^1k(M3v=MOP3qs9U&#g{U`|BVYlm5CJ#J0;jDi`F|+TfX6EU2i=ceG7P+JU
zqM4_O3kO;1y{=I{1}D!_?Hhr+$NokluiOmT0d{luxrguPJI_AI%v*&^pVw(?MAePR
zKqw%426bET-cEDhm+>igNmDzbZm;R;tv`Cx)leK5_bFh$34nV$))j5#J=2xTdwPGW
zhd(f-n;YtN<hu>h$Ga(7v&nckH6H2g;%BT{?V{W3X%+qZMwxt?Ziq*?<x3k5og96)
zi(jx-+%Kr&XSwV*Zk^j63p|M@(|GY7yqe~3$DipQw&vwrG!!d7V%?~_uFv!DB6rWB
zWHDzWh?RRhazHCcu9Zxn*n8b<SZ{F8qg7(5ad+9NAwT0=R$lInf1nux&&N{3?~<Yp
zV(4<8Ag;_g&G*rFm;Cd!pM)h;KjSe&g3AR!!~&P9>2cCBBkDJ^=ezISduX`J$SRYa
zmiqqhrm~@AkJ^e;J*#dfBPE_R$J^q`+qO_3E;h!AveU$N)dA1l#~#l7e-gMD$KzwE
zOZmGtSmrdxy00b!z~3*#xTjLivs*YR|LqQbPUY`9J&fRQvL3$5Y0aSr0{Yz=`04(G
z4)fg&<aEO$Kiz+XE6X|PaQ5dKP5J>$$$;tylp;5n*8{mGGB+J*UR?vfw;P`ezV|$Q
zhxF+|lkOS>d(-$1nOD~!*qg?8h}SVC`0gK5f;|y_9HLg|0q=-^Q%HUsVqBeO)pLIZ
zY93A>cB;DB7pFz78y<=kcU0r^;F-2-X;OSj;K~7Nc%PlV9XjKR+ug`Ji4zclb>RwI
zWc%BP;PUxTa`}v3{JV$zAsTps&yBz5vw864GhiMR%sK~m<9Cm~Kk#TYS^8ieQA^k5
zIrtdAJAQv&fc@y^e(~bwMJ>)2bCmWH3|4C>NalRH{0$v6N3t{1WL@D*@yzDB!YV@j
zOAFMu=Me$hL@T}DodI3pf_{czQ{DvucIsa9?zRl{5S04}giN|c{nPUfLJd4d=q(YZ
zU*R0p5a)AcJ!&iCs{Ql5*83)3_QXu#k~bHcL-EO9NIhpKe`_>bBYs<>H{;;s$@FMP
zYQ>FB0h6}-47+69iR{&X@nCjZR@ziSnDB10l*>cS_js06Ceqy)4<}{^fm;$2Cvhu|
zx#iOjy$fmO=%l2_(a9e%JpK6hgT|8UO-*P5T`+5>RuyUl+NqlhHxiaP9qvQF6rHh4
z=yykH()bPH9-zAnOlj_2<{Qs@(O=gX21Jh{f3ea?c~<dAo_5N+3xe%Gb}Za!{jsZc
zfG|6WiaFQBlosaoM4jnJp-I$!qo9WSQG?S(uWiO<`MMybMbqwn>C321>r7>0$see0
zW20f_rb4S;JHv8G7!=*yY2aUTFM$l!qu$7Ib4|%N$5I#NMVki@W9Z9++vm=uXQIhd
zv~Y(z?sxA)Fm>)I{13olv$<gGvT|+R$ZZK4rh84i8>OfsJC#EJaBN#Bf?=IEVy2{P
zTvC*dDAyB?DN09_80!UNULhL=+^!*FWB4!vrT?@MAr=-RQ2XbNK$;v;C4@F4qXIJT
zsfuwLTNG60yp`zhev%S)dutlNi`nYML@8SQoOQFgAIGP-X!Ah65|+e2AsK5>_OSD|
zhwkd^N<;KcY-;CzUUp@;;Yqd`s{w7#)g<d<7RvsvL$ua}_xxU=N%Tvd*yt|QH-By4
zmge6X^C>Go?#6~cloh{g)%`ogqH_+%BkZSaz%$9qhaXH*Y-Qa1Tk0BlE!NoCuPk|T
zS@JUO20fm%*z(S+SR!6CBDxONfc0?a_-js$uA^V{eA+di^e!ajdB8OT%TklOV%yuw
zB2UD)8aI|YKM*h88?(-Rlr0XcPCZ@b>@SNv7TdlDcc;bjkXQ=Wz=EJ7wtpX!HT#BN
z8}+z(ZW*5(SvNj8?ygbL{lqR48$D~>D(66>`<YP(7BrkN?^87RzC<uy{KD0r!2?xb
zPy1)J-q7QV?L)iQZ_(tmKj2EAbUW~LVAFVKWE~9TEDvK~+^X>oGdUg(aV)e#*C9^k
zuE=<C<KpTv=k2n{w%GRFWr6J({)1FroS37}c#854GI(s~>1cA~1JUI0djst(dj0a|
z4qxT$P<eMOXqcO<xWs9SyfeyW&_2MI$9W|fHGCDW5${F&K&I8d>?H~y-}dw;PGE+i
z7nN3)@%)Z9$%5ki#$l+&FEUTNH_wmSCuYo?Ucx1cXT^#iOQ4hMO5{#(zQ*L27k8HB
zw~4WXU;eO}%Nh^#L+&xxVaBb}mt%pA){h#r_13ui9P7s$@;BJ#vQ--B8}bMb$$DEb
zi8Ews(w$~J#MgrviRxD>lKTdv6+?)^nU`(6lwq{`h(s!7T-$m$9&i)6r5NzRx8e5Q
zCd@W>*MG^ZZ(mtz>F@H?&(wIoTk{(RfLsH`oikMjc3yu)1*Y)+mevDEQcZbwF&AzQ
zprGx8-BZcD?Ynx$fz!reJ)_bnUXs_FG27n_3oiq#9Pr-lJp^T^v?u90ifeo)-yU(g
zUA$@WY3P1rJ_bN@%({CI?J2)&OLih-7rTq!Q6KMbpLz8~m~zG%-_qqPh55bYbC$OG
zYAIt?YMgMqB}SZ})EexKuM`R87lbmf0}&R)*KDiXTD`d)M!LrxtOOZx<l-|utBT%y
zmK<O&*z#NO+l_4@XG3J$_T7PPjk`}-u<f^9UCR`Kw(7y#p7>J0wkEu7FL-qF9SrVu
zyI_$2rLRQnFBu|fvA4UpPRBdkDfWT}eXY3~bC~<}2~@E?w%~0&ngPu>NJxEsvpK$C
zvjJQFT346K^;#%?96GA6aq|^<@!QvTu}DK~!4?nwhZJeSUJw0C9{QI+-w2gJLNJ<R
zl*_ODhgCVgps^Qz`#6YNRQL}qKZE}tTK>>KxcuRNaQPW0|Iqdy^A9e6$v?RK_CL7%
z<$qWCJ&poHS>9{dR(RJv+YTm_&6DPfRc6|I_V8;wq#^8ZLkd3JYY|Cf$xG26=!TRX
z-VG^x-tzu&`w8lRR;%_dF}Ss#^ZxaVCBKT%qIU^%)(t5yTPw~5U%dDwt~=s}ls965
zN6ZZ=Zu*84S8MS%6T@T0o2(l{%KM5VOWeKxp=qTNyebNs31&X4?k4FvQ|I3li|mc9
zb#C(hW!$llbumg5=3LF5#+J4-5Z`c1j=%MtiIY5j&5ZRR$n56vo-|wAw8Tce{e0|-
zjZ4f`9oH+BJ??5)-<Gz2S8a7s3X9b(?dQwz<QzOHSK&#y$_h@LX$(+;6J_;+Og*uL
z5Q+L&9+p{DOUJ)N`yHUV0aQzTR3!-LLRI3UDmfZe1!8$QQn`&#PQgG`+web%%G?7o
z0Qr@b`Y7(ot)T&!oSG2?6O)k_hEm>zK?+<tl**R|DRSvh%3K<x(4|9ZfYKnv&Kt^-
zVrY;h#n2E-ilG#HMv$LrK?)xaQu;hj{x>h-R@3BNLtT@{T|tE`ScdLUSWp)<A@5uK
zcN6ly)_*r4xh~<_j-fZ1k5>)7-Gp}xz0&~Q5%j`UK@(qabI`z1N}8^DiB(qvn95+*
z+Rj};g(`#A^D2W0c`xzbO~`wZ|87FwJCq&8$T8oq8X7d=9YaF~?v9`rt_qrXieccu
zukYA;y(wLi2Zrw83)^`hs8A)@VldPNP00IN|J{VV-|4@bkoWoKyPJnDG2yDAH<|E`
zp|=~jJAz)gDrn*<iGiae`VL#CRt3dRAP=G-%@|`>Y=CYT<}a}wS-)cR@p|`qA$xpX
zsv2dCFBt@x@nr{CwEDMkwEbLS>0kuK6JyR*`LQcDEy0>w3cp_Byr<dwG~jOiYN<e3
z|740<$C}<b>}8~_w5YcCBT6g(E1)&L6#8EQElXzq00H@b1vGC>^Fads{|abb%RaV%
z#%A8C2J24^Y5ldGJTV!~%MIj4^K+wv(Vb@qbLM<Ua?<?V;93}WPNZYshJaPNIRhf=
zerfa}6P!)XmZ;8$9j<?z<C;wNF|0Y}Cyxtk(ar?NfQ`Fz^rWn_gCON5qkGLE+_Dj#
zUkcXc+_AndF*r6GJ0p&leTV%yvj^z_9YbZkhR{xqD;%{?aO??m_<oSFBsYGL75aXV
zH$}OdXA$dtHyQz5+7)uoCQhjXwC@kud-`C1$eZzpys7$;xU;DpyAvEfQ#D(ckfTo<
zNehvSwo$R}c{#n&LyFn2Fv^t!cB<ll-PoRs`}MA-M1fSm94n$twyIqcxZ$iE$lsLU
zU-MN*T}?TT?Zvx`?1<~jOC(>Uy`W25`I)EfpEJy+iq-iLA-gP{k0861AiH%`%U3cH
zjYsXLSU0dSZSGeeD0dE<+A-JezE*v(xmLZpzv9EXTits+Df5DL()ze#J*+7)mAG-R
zgKHgs|LPd`tD{Oi=eEwS(!l%sngWUA9JkRuHQ=@nN=*pxlq&5X*v>vtmJY?JTgMf8
zde{ITOHCB{X>a>D_Tz5cLJP)|Cq|RUIa`Ey%k~5D<i*$${aN(G{?0jOcPAeg4E+{;
zCHfapIL`sA?m`~RQrC8AJFZMV!)xECkFKH__q<vC2@kTZ03helwx6lnQ<n~jrM|~s
z)eyl*q13(KtNlA%7MT4z(_N>~WtHl;ukm-j9>!|>&D(G5*}v0%+tK@X-hNxJ{X1{J
zZEJeJZG+i<>!8SI4?Wnuf7b!NBi{bqZtdT-ne+~`VRtaSf4AGbb{GT))A)Ax?BBH;
zQe*$lMY9*u-ktXGdiL)~R$a3<@2nfPmK8tiJeo+vQrEF-cgWkddzM|hc!d9VvTJAV
zXce8k*#3#}$-JNbbo~BtKOfJo-R*(xIE2zb_^R`$_!0Yc>$G3@py|EAYXf`PuUlvK
z>mJNAdw7YpdED)4_UoQzzpf*Dziw1R2Kz!k_+jtPmTdTp^{w-y*&z4!h6co%CmMdn
zdhol}O<#ob#b>vtH?nO9M)uvGNcGyxMR8n0ba$CE&%36~P7Wz_pS)GO)AM8i=eU{l
z!*cP4v%4M|-<)?hD07@mjc=YB*tGo!fb}~aciD?=H$#GQBakitp}0p*o6I+y7x%wd
z+tvA3w3l628sG_S%F}^oIVBle$8{}TK(HRz9{=ptQTt+_-EP(Wj)*eHjYYQbl{IBg
z;CWQ4Fb4+WCRo>-xpY<6ro?gPl*UT|uBHn2Z~IhE&UhAQcfV_XUwzPHUhL=1IFAb}
z-X$LlM(<OZ?8qbWz!6Zd^UCsA;2}66*D)MmO$e%=gDP7et8*dzb@129Ukl}EfcQG9
zU*ql#ISVVs-t<OLmpX0xx7Qx*^lQ2O+*JJZe$F<IMzXNZ<L06f4DQ5d810GU)tah0
zjo9}iJ|-4@1#6|dqp1r6wwX%=54|)8^w1nV>^0qY-{nH$!}w+BsNKMnx2)CKB#!wB
z&S3UlY=r*6DeTkux&KTstUHS*_tGD+_Bd`;2LD>S<;_`{3YE33+3(z!<83@}(OxG?
zc(aYT%{}RpG;Z;Z#OICwP#!*QCo#O5Cf)pgx0!6(Kh09xJp!F6I{(4qKSoht?00`l
znpax%@t2~AZ+3S8l;+>~V|vH^^Ct{;>&Feaq{;4|p79`0%QOmFAGvVS`m2~z7K#2b
zEK0d$WJY1|{$W%cj*Tg>yKyDivb{LC7e11)>u@(wD(`k3cVB6KsJ~UO+m%?xS#@D$
zf@x@eY%JAFoT4i#2Vhk**y*Q@G)`_6rgu|V&y^)dvZQQ@M{wt%69t>rnU1qc+gLKP
zry@dbY-3NQF~2drr_l8cbnh-ZAb+8}If;=O@lq&hwU1<?p_1d?p@On!NH(TWWlm>2
zvRn7|G|J`~dTl+-Mgb*^Q$p@t?Bd9S>iCxX!yk6Cy|wf&dD2yx&4V*(l$bNTu#CMV
znk?_PF<Dx%O=p@<bKRAELl6ClK*lDz4^+{c`LGSc`7ZYXQra$MbhWjrn92TEd^1Fw
zgT_vt@C!XBOI&QX(UBzk-{z-<B#nhTUYbQJMw)hNiMy_57jMOnRZlZ%LA}vkGu?;C
zfKLlE(1x<mf=c~tz?ZnsQ8Jk)((0MpJ>1!1%#Wq@`jlY?9adw$b8}%GfXH|X+6N}{
zQx|kOLFB6+Ty40>y@fR-xXnMHl5-TJ?brS{P4B7V1A5M=;=O*_F{+sO27~5f6=&HA
z+7vfM8Anliwa{-=<z)YU8Kvs1$ShTM3*9da^x@@vmX@cdO=UC{QB~V{2wE~%O?w6j
zBY}cZRu7R1iJFczTaIi}=VF-`<hBWPi{o{x2+~z+3M(*8O71@L;Hy1w3BHOYzu`?<
z17qAu&SzX3jNSMo|FhrlS#$qo-wH{Ogu11!a5drWh623XLS<L6lDG%IBPO=s7Ai1%
zQ(3BTvl4D60mdTJx@GvAkj0jN20N3ktom_Myy9M@?h!ta4MU9yrxh;o(0)Q_!KY{+
z-<jY;Q=AwfE~kgj87}vLF+&h+3$X4pl)wFRkVe~oG7|~&O++hs3!!0r?tQ>0WwlAU
z*Fd;M>Iur~f%l97tKOF8mXsO;o|!U3DQQH`3+@|-G~40*SzLDhgH>uE+hm5@QNJ4V
zPv9F-Zm+Rs)n6pBq4*W6Zmr2^X<>!XlzV6bKAH+2&1XTQMnFjSTH{pH8Vpj^HJ{FN
zAHA8TI8{wC)HH>krjGU6;dcOZ7ax1YneGTRJ0$~2W<BMtM$r@~GkTq({pyQvzj@HB
zzO8bXJa~<&A2WGi0OFT>!Qk%d1D1(RULj2N_fdTjHg>9J;1=;tK5R96lb6A}H>t9F
zQ=F)+0K9#+$1t_W0Wi1^@`=|T*UGjEgeI+nG_O5=W^ktIf@25nXVmIc9Kg(?zC}?L
zO@T6_`)9O=3`0ZiPF;(aTuMEFW`%?#CI4<SG1_SE|7rLFi2Qx;qJL3V-f+uC`>%po
zzEK=OT=dfDWtSi|^)M}~PNo8xd{paE_^5)mU-6R-trI^V1T7tIHuUOaKqrrazV#T;
zHyj21{bN9X_bBM^9RqsKQP2yI0j+b2*|ogv7|`b(1^v&*fd2YX&=ZaUtxnCD=-Lbe
zxwlYaj{5?x0o$G?CIjF(%<nY)7=XVDU>h+R0QbwXHw|!y0PZ9v1K<xjLY@Y=P5`Tk
z$p9$G%;K8@=*lj`piF>b0UXE%*q#Y6OaSY%3)!3ru>VE??#l-FMP@CY5<n`u7C+7e
zxK9AJ#I$MkF6PRN8gv*>-eZ3r+s^ir`kIdFwwcYj%ps*A{y~G}8PeKzYgG2d$LqYE
zzD#5rj)6?0vF&F)$g;U&UC~h``^hnoEhnuVslBP~H$0i*^)aK>Y1Nob|Cb*FX_<$r
z?N?%Npx%;3-Fu#>`~6XPCCYtJGggw;UM`HgdoX5m+PapbF#h@&7=P&DX?ufk{q;;g
z-jZt}t-J0=yt?gmhA7&SUVFD4@xujtkTC?L57HRfL9%;sJ=CM-%mDaTA0SBRepLWB
zWdn#;2}6T0oFss6X99RKWUCSns7QTD7j<6|jtoAZEh~$;`#e5NR>q3Q<m?Lhzk$5%
zO`f{T?&Um;kKG$g&b?>#I+^Nz{A@k3>E6PSFc*Tk{d>LglvLvm^7G7^IgcfS_gzh|
z7HfoYZLQi*Q>+nt+g5nZxmvZzn4SN-dEBWyrYR42$^(?1C2O~SFVjDoq)(g{V4W7j
zw08Cvefa_j<-u1(dL`hPoErIxgchua%py?oBbI)BNEp8XTUqfNiKx&Nxw&dKEz@S+
z9?d}}SIx0(poW}OS${lFGKGA>ZKw9@5ib1K=XmWYO+JSt3Lkt&h4~{1=cPuQ!mws|
z(S2NF8yN+d9*L%U65FcL<k#+(T4J4a5$-qf$AB!N%66~lvV@Vu=$_%uNlq5NVJVV(
zxM80F-K1juK-Pj9Ia|oW-MJk4eV3(e2EhDGfcph-5i#v!4Y%tps?{0cDKq;iN0)Cr
zpSCCb@3UGUacNf3&pP_sFLsw~*_XQqd?BS$tKLjA_oQ<-sAR%{eqtqw0{S+guiw|g
zZ#DHc!mfLw0O<65ztdNelJ4~Uj85NIlhx_3^GX*9kFT)lrq|y3n-6<8$|eX9qja}r
z83z?@>va5=E1+09wzvVE9j77jkhXDH)5;6EK-qaGdDi-V=$A${xDVezQId;;>xX#n
z5AdE`oWDNoz5b5Z^$jd7ACw&IzQC)q$wZ(aX&^tkghK_N>fTRa#p@mrPHw2gxTa$i
z`ZK^X(!XK2K-((-;Xk>1e44A28Jma8ocGEiyJC$;`lENbp;}I7f<5Eekig;&ofujU
zUMX2L{;2}z&br(y{WvpQC(lyUH~lE}4PYp8y4;C=+zR4iwHre=UdvE6F8g<i&1(BO
z6mIcIt1*K@!J8N@J0MhcwFh#mD`NO75{`ZbN8)_G$8O4tdy7h8+=JC{Pu#O!+~@J>
z)XvqJO&#EGKS>bn7Qeyv<M<uco=ZA@H+qIFO@@BVw9mAu<-TQl>Y98{HGo2(nS;qH
zf9kR@`sGPsBE~3!Wt_W<-inD?#73}}XRMVcmBf-$!Z=T2w<_~K+!ICNzuKyUev?YP
ztB(gSk2zC{d@KBD^Q4f_%4#XX1>bmK$>6WG<9-;ZKE+O+)tvVe1BE?S@rFfgKzN6=
zSPC_*KO16tFwW2YC<t=aE#}G1z20csp%-sfcw(s-p>HCxnHROO+A~|9`C;SX<EEFd
zulZ%Va3K}$eid#W`Fri3nI2L4pnFN5*k}5elzxNKPwP(qxaDc4<LLEf{pvmv?SEvp
zch+tP^wFc=)wt%Ia9l?1oqQaJ{FB$=O6+i11j^baBk~wF+`1uL%Ur2ji#J{U&t^L@
z!F=amWc@dN%OQu$=^#d1S@HgaNCfM0*;zlJZtJ!LKLyv2G3OGvp>&}WxY~Dae*2PG
zs+tiofmv&V9br2OkM#%1t8MXTrF<?vkQk+tm0wmLdu8;CmEZmu8S0~L{j#6iiNI*M
ze8F}Opp~{GoovWr1HUg8*`Gk0#!cFX+%^i)o-0>=jl25csEqk~BqOw&1pRY@ZWLtV
zHyrI}!+BryLe5Y`r^QoWC@%@+aQpC=<`YhhM|RO7jYo3ftoiQT@6iYp=#;d+ovXVk
zTPZlYhTuD}i7Rj0t9Kh}za)Joksj;j=kGKza=;`A9^s-);~D?*dCvUfwI=(ohGXaY
zmku+ZGvf@9xOCQEPEOHy{w8>->0ia^pC3Q$$5HQngda{XG4HGX-GA@!%h0jjgUe~~
zLqDjwn1>SnE}MNuC_eY<xeG&Im@_En3-dzr7FJe;W?eNW#Pj9fOUzw7DCf%Rg^AGQ
z>MN(sUo>|X33EatL-ARO#Ddv#LyH$&JJ*W=zDysV`LVt@SK#9p&tEVv5h^2}vp-Ts
z)r&(_a~FkX&sw<fzwvc<F+X;`Wu~}ORoVqteLHk{WnzA)YSy9!i3M{P(^{dN(3}N}
z=FUzmsJsfwG?k*s7KRdwX3<vjW-VBljyBI>Q}4NRMxE(DeECa@=FW?NX5ri=p}9+{
zK;^%rv0d(ANq>1(?=Ss+8DF`ua#7i=#G(aDLZ_!YD>L${MORYRdGkVF59OS3#+eHj
zTz=-HMGNL!ws^snbIzEbb9Lo{IUz-yefA_;`pW9W-0;;(3!OGHG`sSu#feb&>w>Ei
zk<hoi*RT5@T|K_?$|@=`@sgZ{^B13S*|$Lz`uf*Hi_$OYOc#1|<Yn%^Tt)`-Di>Wj
z>+*$jv&)i^5cW#p*LPgy<=+`~PEX-7qQ6Wu<r4sv_3E@>Dx@=O(YQX}yk_&$x+^j=
zbk;c`{u>zr)0h7zFnLWgd;Z+nSIk{>dF8A{-7S<EmlF<$t7+n|j=n4rIqkB_IkOV8
zzEyNd<g}tC6C*Tf(J@={@>z@3gkC+<jpEn(BB-nKztILGL!)Vjv(NU?g*|loJ(t`<
zr;8Y9qn;-EdorSt&d6*IXlS8FPG34_yw`N+s_DXu7bNCJj@cGfbLU=BRe41(l2Ytc
zz6~!{nY*bw-BzlJp{vCS#=_NJlhGX>QDxWj?;$Fa`^O<qQ=Zx0C>sagot1zzJrcbn
zM_jpbo~csg^vP!2(r|vP3LVMus&e+(mz7;MXYRaNu-?mNRaMQsYEI7FC1O{JxpU6Y
z2jR%qziu$3+g2TYnGcgaI<K6&=-aRfza29hEb%=s&$@DMNQq~kz1WPt%Xp?e|7TnB
zl1$e1{~o)zI(AiJ?ziVIx}?t`nNIx}gDd?>-w2P%=d#E(jmPk~k2Sip&>SnntSqwn
z=^sTn*^DRaEsL*v9L?}!f08;j74_zLBmHS5g9m0J_`g%VtbXXO9!x>p`u|FEXO&>V
z2RFB-(f=z2%9_cuN%Q}g2Ffl^)?0UZ#KhBm=FwStD)Qd4t8uKitN$6%bK*a8K=)+u
zce9_gy!ekSZdM*2sl-`bl-<_d9p?Ae-@!9KMlK(z)LA3%7~{|*v%jO@A1|Z-M%lAC
z%Ku=CXBYgKxqPHjXJI%-i;MI99qs*bviV5m&ZfMqx9-;NVMu>R$+J)|$a?D|qmNYf
z?8f+Dl>eh`e(Y>6$?-*LU(1piw65`0GoO;$g?}Fqme2L!JgA(>qs8~fq(zHfdZ8RY
z{mY}q&sv1CX2Gnh;Br@1U$tQNEEab_x4b;dV5;%+_wbD>5<*X*<DpE$tx1>D&wtIU
z-uE*z;-wK=?Z5YvGK_y)((m5uQNK0!`1QKM3xj@G^#8T@?cs40SH7(WAYQTn1HwE!
zfD;7RLL<rY2%Kc}7~8Ui88iwJ&(>&WBn?O!F*B0okg(bqjM*4af&&f<M4RO!-h|xs
zZdl?CWY>Gc1}7olu!J?6O;89){Dr+iBs`Q$mcGAVJzCv8J+@)PKlgKe({s8`ojP^u
zRMn|d)h`933eHR_I>kZCTOXvnU8!X2P&{#iT~~-6)$%F+ww}Eb;2NC>n^gG<1{ADQ
z&|m+-2A1zESF_jU&-j!-c}s1}$jC=N5)|IPnI1{!YZD1W?p!$6M`A(A60qHZ|8}_`
zi($W|21!c;_&eWNwrqGbjsF2q0AT57;K++jxtzUL3pzbdGK<Euxm0I-V`eO0lNgL=
z0~cH%7%mMINPyLEA&vznQ+;Ct%W~;~k@!$<KAum4snHf68yL)Eb;(q*jTi_Fr-lRZ
zWHOt|<pLiX8k#u92<VW=B)MaFjA4#uGYJr6vbn31*HTMZ&e~m9?g%+C=168FMg2)d
z*^S}s@E2d~i>~pjmSd00H`B!z$FLmk&UB>`<EyfnVXSPiiKUh2iiq}kul!n51L+Z$
zWUNW9D|O>oDwj98FkTvP72}G$uQE{rw&l3n;n>b7D3;evN=`4kWJ#G}Byhf_?_FyY
z=ettxS^-=U+6OpVWf(pE?^<Ep%05Ae^ud95tzCc>=UUb*V9oiKRfEf`6Bk)lFW`<k
z%i0b&S8rLn0YgoeHAnmk%c?s9beCAxAmHFi%bEt<_fgB*2Z;AYT2*-ER!uAT0amnG
zRuAE7%bEflTxVH(0IM&<8&Pl{;vBAChX5<Cw5(CU!K*+AxT_cC0#3wnV>jbz%L>4L
z?YZ8v!hnNAmNh|`wXB_hH5*VqVD$v*L;R;LtKw+L`x)c|+;bbs0o?Z&mK6n@y94qA
z_S_Bm0k?m_vi1UQxd-I|?)oC~9Rs@iAwOW~uOJuT{)Zqx;PxGmA9o(q;N>$Rz%_ph
z`2i<(TGkH2zXw0S?T<kY+?+Xx_Y#Ex_u(4)1mKQuL2ke;{|tVBYn}u@+|*iwi-pyI
zA;3Do9>6t#6M(&d6;GpG0BfFsegH#&`vBJf&I0xTR-oMn0i%F>_FC2!_5;9a!2N)`
z0DFF9Su=nWfOCL*o=15L5TAiw0QbBMdcc}rpd7#u;6A`LfU|%-fEDOhTL1%q6?k1}
z2z;sm!-$9Q<l+{<HF&YkEMN~{#R-rLFaS6K7y{e^7zNx8I0(24a1wAI;C8?{z@2~<
zzlNRws{v;KYXIi}Lx5F$L<6t}um^A#%G<IZ?S}Xq;4~lu?i&}O<da=S$wXDj>Bk*a
zF;#L@74c`_s7t(ST_8!szdyZUEK&4(uajfVsH|F5S-p0_u^TETjgKw5<bsAJ=a2z;
zGL(HX+9^srs%!?XM42qtFdC6YUOQ0k-H+q$hHX!^7%i1mt(DcSm4VjEn%2s?mdeoT
z%GujXHyyS0$UBa>v;3~I&85tD4UVeCmbDOLsr-%a`Q>Y`j5>44^0p!0+!D)bXTHGq
z{qyy@^RXUJv%Cu}>%tP`>y-IcSMEM<6C4b!r!tpb=k-*v)Uwvn5or~5YpD!SwDLR4
z_{~oQ)`vGb)zw;7J#10=jv{hRmAC|BKHm8n3R>0;lxx;EUrS}!l}(g)Kj=qc+kU|E
zs(#?Fhv~xd_^$8@*uvLPcKM_yUrS|OI|@5;t7<6FFi_6pP}SYYHvm{Z?a3$GxxLb-
ze*Bi{Tu9}6<IwW4J+?v4I@suEC;<8t+i59I9|3+2_!kx4?rV^Jb>+?yr|^{fKS93(
zHvS#x$)|6&S5CThM7GTLN0c`jwyZA98Rcu9K4kqiqMYsRmh}+ZC0`_;kDhKpc~xsr
zFU%D_a-yrch7iRRa<ZIVn2Sn~zkK@8%Hw!=1N2ju!^S{A?AI-o)#%ro%I*Z7dR_>=
zd%N)THpZFi_i{`_&(!f9^W9(AfOgQ&_E^?Gf_$F7rLu<fM{E_;)N42Nxdn4&1LnB$
z-9_X@Gdj$LutTip6z0+eKJt?9Bj7vPXIXrxw|wOf{oBja*38H6Mek2o)|;67%Lj|(
z^VpLm$k&K`JJXi+Z9u187<Jf%g-E&v>1#%CgB#9uAU+R#AY)mpQLeEXBL{%^t-$x-
zg>@g5_-T!Q2>8j3m|GP+nn}<<0sNMmEvpjx(c^@Co+tgEW3F=G8;M8$Ew@<K<$&dT
ziqv=1rGC*LmVkce?UuEi1xq_<_Xj=hbe~}Pe5dN}&tmSxxehG98~thfR?E5oa+Oy-
z<KIr6@|kZd@=e`^xtaNT4n5zKEbsG{wF3GmpDvQmr(gUL`KIAZeTC)iKlFT!7{Ar`
zS=Qe$-^yqG>*qKAS-%G?>oe4^-yFKzJtnAA>g6GnxBpA%7p&hcMauK3-y0}z?_XQi
z+3aA2<rU01EN=;pr6_ODLzeXiS3ZpMP4jaG<ywjHJ^iqfYoD)N=r(1pjwad}{I`77
zvVQ2|Z~GFyaxvc{C@=5`d|2cwU%J;{?>_#-A5mV<*DdS&;Nhc(0w08A3vo1}yjkS?
zG4!qj<>&-{7kuZ(i9bZWlHV=J7yhPYz2M+S{QbaJJYiXI>Fq%Lqre-$E7wq@5&yKJ
z=eZ7zIC&Zjsz84m`d9hjq3MV1eFpLk{*z^$=IWR3_GVe_pzGajS?Kn5ApRQQdwlSD
z;G;hHt-!DG!9N82)F-eeBzNeQfc&2Te$S6#|A<h5^?e@rD)8aw)!$1y9EY`0=zm)l
zhKC(k&Jy73Jb3D73D!t8zz^cQd<*m{=hAk}dSWgm6XwSzgfR*_xcovM`ZJU>$+Z^n
z>uE3c`nC(^{=#u`Kk`leEBp+=@+CiV_jln7&6GNQAr>q20{RN9nL5}Hwmv81b^B10
zBA>S9%twEL^$h$Q{2?Nse_`{fTLjjx9qGMD=djg*_-lag@xkYTkNV)Z0>1|Mi^!zC
zGC;ucSx@rcKm^Ly{>o;^K!4>)r0+%gA`0M{pS=DG<$8nsKK`zi#kuZJ<nzXH@E5U`
zETtWJ-l-qAFT-rq=St9Te*yBc-1jiv^T-!PzxC>CEAVT4@DBkW_Q5{^e8>m?Jn(fs
z`2PaF2KYs+fMXn$W9?0PFa8X!(S7iZ!28M5sp!4+N-F+d{HK7YJkIgU_Spt}0C=xG
zc?9@s;JtjF1iq@<Z6DgfryOhj8njP}^+mt9t29zseP>yuGH^$Eq_Sq~5o;>zHXpgB
zGPLQaODn@el_88Bw6ySTu!_Z6wnFspPVk9>kLvQ|CW2M=;w_VtMwztRPiZtyPQ`KO
zxA6Bf9P4q6V~s2pioJj$Y(+I7xUrcj|BOft;62RqYZZBrk*}as_*2HsB>N^J%=0i)
z{sU)k0!A2l7D`-v8BQ2}d#z!d3@B2{u7i~kHWQeeMxZLdAAOUNjA5J!cpDi^Bl01@
z1uO@9zM`DPjO1@L3^vhq%(EX6N+9yQiO8EUwj%uICc`)#XP1&&-KQkk>s%9jY!bD^
z+$2(Fwiw1aKt4=vr`_SmbIRxCaF4LiixD{sa4jRxeE}U2y<ss~pZi70dMhIf?}JGL
ze2wLtfe6Z#VLKue^lL2VOStcUA*^tgk<~j4<0`25caI_R#NSBujxy!mufkv<<qf7B
z^N1s*2a$_`6b!EZhD`Yb3r(qrVDKM^<bXVnQs4HdB)gN5&#8#WvjhCurlRS;j)-{m
zKM=+cXqh6qi&b`h*?Gi1!#@U|^g<COoT6&W4dYjkRs@^&S%olo@)^(uNqaeHe?L$A
z0=w;cm_gCMzsxL)aV)^`6-HWcP<gK~vK|L?XS{=><ku?ldq!TzAGIQa<ZhLH2drBz
z2a4>MaSUIBBaK6Zf5rbvIJn3*_TgB^6NYm^%esp^R@wbV*&?XZH~~kD%%8*|qysdp
zvfr={gFOFV{A1t=>yTop-^MY5Lj-R9unxPK@(YwmnXYE!WBA(uD8gfiPy|8tI3jx7
zpghX64QZ$0AaCRAI9hqaV3qBKMh9^Ae{eLhPTeR|-5+fjcOwGxYlNBdB}6s?ick#<
z%RGlEXX0<I#5;Mu3Y6g4g`6!qojln}g6A|M6FTKUJXN1M6Z9hRgsKUZ<L^8~E~Hu*
z>^^n0rsWA~|AHhZA@Vs2%^>yDjI#{T`<P(|{;p&igH`r6nrn>1ll=mg6$1nZbBM4T
z|3sd~KYlHNWEV2R!4U=G`lAq8fIk{N5z3KF)d_A#BT}srjI&jWU?A$T3rSQKN51g`
z9IS`Q$d7UC!6AZGc5T_!JiArK<q5bMzhOyt;Aqz=Kf=ivG_i)7TZO;-$c|wKv?LdK
zeu*<jKl%>GM*@jtx{6e{f(N_pv5ag(gqjm!B``z^GJeOy%ZI~?d=?{@u?$KgpI6u@
zEAjUtpa>@;nS2D<B1F7=l1L;Ue*M>IQ+)m}`E)t?uxpTyAmj7RUOrl$rBn$AH061n
zOu0Ea3n;=lNG2aab{--QJ{(7bNTWFSGv_FdQz!}peLnJxGvy12{5P=-p9h9KMV@;R
zaqtv6A4Vbx9%R;P{ILgI$H+PO6V4#cMFZGmUqp(74`pK&nSMO0*rnKTE7sVGKT7>W
zM%a!SKoQDReMO#%!uqZR6?2|Kp0DBfE{<o&h7YW>+rG?5H|k4^vX_y&5n+mOlco`w
z(7nbgdu$S&tdS38kSpJrU_6Cm42K9Sfngn1F{K567fHMx$JZiyG!BkKqaVlFJYi7w
zou^Cr$G{UxatR|G&{V7lZTLSCM=xU^$3e?4g0z=BjY@<(;VgVJ%Q^}|C3Oln$T>fW
zl)Dh2>|X>FTgBhv34>Mk@@G)*mr55QLKz}37J@$`QAQrX!PKi6`6nEw;OJ%KZ5*fK
zh)`pdU`bI|Gg6Pp$$%X!=W_g!L5k|3Tuix)WRD`kt`a7<ClR45`!z;r4asdMBgdk4
zOu3JF79+wuUuJ}6oXTP=8`t2E;taE#oJ#pB$-aSzUCwt9A-6fQ{&z%J&VMs<I{FAx
ze#bl)A;LUwFcL+C<s<=(A^fqNF_yDQrTmd(-%$~+?u{QK@+{dhs6KE8PFccdQ9ZF9
zV%c{v@>Lv+>|x|zaIklXU7creaucx(a?WVLSqR5ZiMj#Dy*xP_4r}1B1`ccBum%om
z;IIY`Yv8a34r}252Mtu65f;^$S|s641?x_i@mU205&pEV?1g)|Qgo#6_r#Cj5($BO
zP7HIWgaxE0&hY#LEdrh<{^6?sG%n6gSMh*?T<b9GSLt;68Ms1Vc=KILta7)Ph8Yzr
zGs2#u!vqF59~tIK!vyxyaD__XE)PSs1Y$X*P6fcaLxveCAC@>GRH^!L0m7iyCS1!f
z>^~tapvKQt$+(t#R;0>RE>Cs&cwv#KPqm`Q`-wzcsn?iQ@u>xppJMZYgE^&tZ$V9`
zN;y_3c!h%f3XUtdS;2c0d|1K96#TA&FDUpc1>aKe$ZA>6DGGjA!6pS)DR_l~{R)mN
zxLLt_6nt30#}xdof-fleD+S+D@W|6t`3inm!6pS)DR_l~{R)mNxLLt_6nt30#}xdo
zf-fleD+S+D@W@4~d<8$OV3UHY6ud&geg(%B+^nGC39H-ME)LXuyl-qIKNe`L4b|2!
z3pS65Xz+ZAFjQ({D2|swjHj|WzW1S_c6n{xvPMZ-e{()NmdkIf57gx21G$U3BAt<z
zSj22tUUyM{8u!beAE@Co`pE!tc6MHIsmy-9Vbta}4(H>2fcdP1gZd12ey6gdM(s!@
zpQ>$X?N|oZDmE}OR@*n09!f65GXk}Y4aReWMs0HA2>3~u&&m`fuq$FBEt?vOGlM!C
z9m*TEe00I6&Ev6eqZV?fktW_mVAQ4t&HgMN?KTIK$g5*unn>U&@kDAAce3+E4w8o%
z&t~HrB@cameF7O#UVJ#6K)wu$z+bH>mv1{TYWs3Iqc(w?<5MGfPcyym=;_Ew+pX<Z
zs5Fcic03e9+f#wWA8DQ8TtsNYbpGzrFk)(x2r$5iBWOFNQ9+zI^XE&$A}OyJE)lW$
zKX3kaB)tWjr_0}78WzOc72+BZo45PgzACi1ctjho^UsxrMbex?^s0G)fOXgTs|A%|
zTmi^AK<C%<N)LCX5cE7EV}<&|#ZQAaT<6zzduoBqu|;K6(}`V<rsswf=L8+s^FlSZ
z(-3CFzntf&J1xIL8=u4p@jAbrhvrm%JrC*fb$(s{TRi#oytbd0#Srv7$o!n^z4`A1
zj(WA{;Zwqh(H{^%>}AR``A=D$;dVrv`2%V`2&jp6)SjSaE>$hD+epkdPk!z9Of8c1
zt((YwKIr@!b01Ee<!is?dEV@apt^xL=bu;pY2Zky^*^Kho*9+@Eu{yLh~O=M2e8ik
z`|uhshJD=bMOfmgzqkCyK;z7B?2`G7T{8bR)ei+T{OR`7@!dFa=I3qp4BMWP`Fphk
zC9s9{e-8Lc99m56=k58a%pXwsH1L-HGbB(%onQNb`&E7~y*K_F<Yh@Z|E%(3W)Cs{
zYskyKtMh9=XinwV4XgXNE?>9foTvP`m*T*n|Gy;bujvI6e>$!qFE2UeH<TZ0%*gy=
zi(6;*$$uosocZ_sJT9pB{9NYOa%vX3Tpd3SWW?$E&#3(SRQ`Z=@KuI@&ZyuC$j_4M
zJb0b|&pdc3mCVB=ni*ZTlk^iWOUgw$gG%&7McI1nIrE2Jm6S7T0df`+KY>HnU)w*9
zZ@==Z`NnpHTQ3m6{_~b^C-|I{LAp?BxFDqrol4KM&crfEqUTRHzTDVe8g`<@c^Oti
zbH4Z^?e&o}QyFGzJwLhe6^5R7-1wsnJ)gMo$Jpy1H~v_A9q7g%XK1_b#vgBJd+o+o
z+UrL*eu1Iws~cYhd)eYdiSsh{E^QATsb#o4rR|&p4eg&~uT$Olh4%W?jX&AY_Qs7r
z#n5)djXxE3A?`$p^D?du^|*JWmf>CyJ)RwC$hpYS<J66(H|P%RujR(;mK!T=HZRk8
z^$-w*CmMWg${Ff49*@~LLqy^E7>hIXC_EoiaR&N3ESHayIK#~%a-zY<uAE^?;rTd+
zGyJu}^KlGk;Cx8_eB8nrz9%9l8hkv<8GfShd|bg9eyi|&jKCR6N*Fs4&l0-v3l*M^
z0XPG1cp*J+^LK`e6`r^8JHw?4&)e;tp;zH~ySp>w6rQ)GJHuv$=i_3|uwCJKo3=B2
zP2u^Nm@|A^;rSStGd!#Ce9X%k<g|hI_27RG`chn{)#E*&GA)3aEkS>N<d<=Q5o~dG
z7Vv)k_CjHnPc(Qtv@>Y=d7HB{XgPTsvomNpd7H8`d<6U{r`|Wqs>CamoP5<a14gU}
z;|kCBU^8gEuGembFA=Hmt3dBpFZxjIC%i4#86H>i=>5BZO4R)4Jp6fYjvxQmgq(%;
zinqzJzLTDP!4ohk3O#u6^hb&J?jv3dJlk=HryV;4zR-TsuO}b9U#aW-DM4Rohl|zg
zF9rPx#+33~H6K1BQ-a+-Kl|*bpMMbag?3onc!POC`=y%ye}bMjA!vVB<6jee3hn<K
z@aLeuGw*~2DWCmg7yu0T*!e8*CrT36{{z62Pq<$86Y(q;DER#OiNN>%rNo!0T+6U<
z6Z8utUOaP$v?+y;4oEzo@nLuhc)xOgBJjr>dp-P*fMe`Oe_av$81R&5a)l}vd>LLT
zLO%!m*<2y~y`q=o#!1+8AfLUh63=JO82HQ*>GgX&wEz1A@YK&<wLU|)7vX8er|JsH
zhtFg&Tnk6huU>h9Ki-IX#>Gv*v)tallYDqP5yOK5KR;hj3M}qMDZ;-L4yj+cX93Up
zMo*Fa`HUmOJ%S$gO0|nx_8LD|^x^X)y?B-o^ylGLCCaJaqoU<p3w(eKJ>kkC_+jAv
z<oq+>Dd*IBS#DVA^Ej*{&xU>IeK{^lh^lgUldvD3z9M)&+u=ulR}uVffyY&e`=vb6
z&SK9Ac*?o^N{OevVE8oFJAVA{75GBG^Z|iC-l+48i|2~)c^i0^J9E4&x2HswI|LE^
z%Ka?xY_GshNgtBr#<ZY6-Uz6ADK^Hd3SYNb(yMJ8V<9Te{vTDZT;?-647I?Ees!~?
z=QCUk{R+QxMB=|M(Z(MH9~E(clQ2ZdCw#V~=QFPiUBs7SzIZDv;!wK?X+^K!Tcgx(
zJfQG<e;pQdyseXAPV;{~EaH4toFRb$PQ68|!;IR2?^p2C=S#o`Y^wQ_cYvp!_4|Mj
z72zx>P_);-hee$46kxa<_!B^{*CWb&8oj_%&sFzHJ(QM6`Y#B)Ja_&-3cRozs(ld^
z;d@2se*!$^nfzr~(1~{k;A}Z68xR=x-(|p8`OGiZ08crmJa*#|fp_y@Y@ec!diwbh
zxDQYCpI5>n(}k+O4GM4kOyc=Y1%?rY*Y7FP?fxL}lxNn%|KAk-Y?;)jc!nK(>KFRA
zZ&cuqH%2}6x*d4RIlWTy5zjP(Pua=RF719fPVtN$@eh3hPkF)~dHMuB`iZjhdOUs_
zc=8`zEBVhVJ})XhGafz_*cT+d@pf2HzNqL&fM>bWw@W^JHjUxSihlG4iT`1V#J{cZ
z`h8D-Q22APf69KP--Ctfi*O0>Xv%_cHSi~to@VS*`#+jspa}hKq=!8Z{30$$t5mrU
z7NP%#BKYUX2j%Ydl>3sPhd-m{>y#4cHHEL+-Xbs;DSRnhH|lfhEQw#B<jDh1J?Qs#
zAu7VXIH8{T0TqS~;vfExqSx=iLUAIztmq9zuh<(Ur}^u%9(d8dn<RhT&#%$+>bjp|
zZ``Elt334IQ242nC7*v(8OkxyQ~sT|O1v6g#$tijT<77Zir_y7Jmn0$6&A$Ls4QPm
z_!(7Sn0*m;D|~oZmOEM^@yBE9fqZ6Ee||?*cpP}qpYN6QU5d|rz_VTSd(J`%|F?=y
zotoD!Qv9C=p8aj7T3={;`>NotGCRN#xbP*)_2}np;H$vL+s~H?dItk_x(0aqE#X(<
z0@I`LpDsfG0PsSeNm<{smHbaBd~`M}IJ~C#eCSO7{<8x3fTD4N&LZ?l;Hl4m$1dF_
z=rJF8`t1(jDQDdpDW`b%0t#A-jZ*5ZO6gzs+ik#;es-y(7w?t;{X#g9qP`m?UO!{L
z4tVNq+M~BW2cG<+%0J<=$qc&`AN}4u@k}`IzXP7_W&AcS==9&)ioc=y!=y?G1|<LR
zCMl12MjL7Cfv221e;-B+>n`O#d@~^NdoP#te3t`5``P~OawYKO)0>y{;#qvq?<+zN
zVnIKv^uMS?us1#mJo_`haU_EB=S|@K?DMTf@Q*6~VMEHpcbPCOI#=?aRQv-1i-!z=
z_v7COJnJ?2UtvKh*S&ZsPvNUxmG~Q!{MjY`e5QaWAN@W=J?}iN@J6qcQ?EC#{E+ld
zyyLMKc=DNEEa@Lm<-Pzs<>$+yoZ)RnuiwY0>$u{>s$Ln%N54}c4?O!}z%wrH6nH0R
zdHz?5-s{)Bp!oN$ll=Ma5r)KhvfRDLN_<JF#P0xJ%O)}Me|Hi5&jcQFdi;lDYyA0~
zSp?ry1V2~=e@hYky}+|wc79COE1>Fo#`zL&sC5LEpCYUWo_gM`?C_w%PYAq<IKb_S
z&*-(1zj$W}(oelW_UCEUpDR?J3xOy9@EKu&x%+XMP$%$(*CUz)US)QGwZOAphO%$3
z>NLKk(HxGaN5m@|+uB@rgVY(gzs7~ZJ2Jf(AEM=jR857bPywpG5aWYtETk$du%Upc
z!3R|+Rl|zHjD><WG#5}c7V4p~Q253|NgE4GEF@Z9h$@u5kWr!J&4#(U^Rm{KPSebd
z^_d-}s{@!<#qymY(@bZ?twbVCzYo>h@QuElG_`q^d?jpWut{*yU$4N+Q6t8D;tE&d
z5I!>#In?)y`qhJXAL+O1+AktaWb?WF7(O1AFidl8TbJ3{5$iThyahOmcYEe=mzJ3y
zHWNd5Z)wi(L78SUV-5^u`r<=o60hscnenj+N-5sanXGMUUQrLuK13#O!jd;onYbw`
zyU~z0VI{|ghc_aNBPL!u>moJHRb4HYM$E|ib_gpaSLdn)FI%-L7U?#-TUt9Kq@eC?
z7D+cEm&coj4YU2q^(~inwAn2X47FuOH=3ikH!fwWPY89akC`jfYt(TbG$lg5RGH_^
z_RF9no#}>%4&z<SsNRF$(QQT|u@*DUVDJVpQ;C2sR$phDcq49RgE^EM8ORTsscbfr
zjjl_C!>c<xTHAVh%!XR@FV%;dTX7Fwb|XGK<q-{H+EPg|9il}bT%(Y!YhA3#j5PL-
zjU*zmkcpqoy2u-ugUeTo_KuAW%a@QxqRBKf#qv$#;_EUUvU_{=R5veDEWbkbk%^9e
zhn`~bj@Sz5Cb**Dn=X-ITP_G7UkZ$GXQhU^)~{}irBca`SU%X0#b>`V!}uJ`IKJ!^
ziHvu3o5pAscN_LA?V0tp^|i}wRz<3~q7^s!rBm6+xU5$&ggYG53A7S;4d6c&AMQw?
zz#v~oY~o`x>3psO3XJ0~(iV(Bt2xSJc4TtqW%zQ;FkTs4(15YbP%e@#K#SL9JKD8t
z-TUfv%qAMX-t34qn!3RqZQW)Xc*}7vmG>&ATeV=pJEL{4D{O-P^pH;vlpO>;6PoZF
z99k%J`{l4r{A8BN&yAT$*q&^<Z;am|>qw%f77n1U?mG5WTUOWmz)ch11522R+?cR4
zc7q?P;P%k@pak#3OTfi9)jYb_hM+rkkk^Kz=2xJLEqC_JzOnw|-O}67khw{ln!b3_
zOvH2faW~OuCYPQt(MRG#cr~_1Qe}XaZy3Z};W0jkXcy5XLi(FJ{%wpt95Or5(lD5<
zCVqBH78$zr_3l?*x2dTp(lu|M(io@d$oy2>5Kmq&E1;U)$2`fNDn?Yd*&_tvC~p#7
z&~7W=w{W8%b6_|#B8@!ftF_~1mzrMNhT^##=GM`%e9m9Nhty>;#)Xlv2b}*z^gkJA
z-RLmy$qu^vl6h5fDC+57sfo1L_Lx*hHgEFlf(80m7t1eC@Jn1UvoNlgnj6x|)QFhd
zBW~8U!QjDqO0Zsh-Y}KyijUj?JDl!H!(Q6C-jmvGb5Ts~!R8Tu4)D+o+dv9mryRy7
z(30_u9zBS=*R>@sG7KQScarW`lXNJZ$(g)kzp(>85hg3Q=4=J5P+v%Ue?yxVdtJBb
z7<KD0wFet=&JU+?!o-;7P?YvsSU<7k^3)l|PTE7qXCw}yg8E=ozn#A=lgaY^_iX7%
z+98wI5(Jy+TC@!(FB|2|6Pds?l#4{83G@uk>9%qNo$PqgbQA=3N|En81Xnn}^(T5B
zx*Xg;;W*Vd>e&yzihJ@KjwFWQ+tk-#N=)E0fo9)OCUFB?X1sbo#e^oTRmA7+FdE=4
zf+o9BCl7ZHH-;je!^?35>*4hcH-wsb9;EX(4EKm$LY*0`!#6<V!zRZT8zV6hH{*C#
zqF{I`qi?rDeULis7)hokx-+uNQQ54TpPI3fz=I}`acIcI+F}^KYd&SG&f{*>H#k++
zjxP$3oXsNDCj6wJcB)xyQt<QZuyR6jPBv74pIaeXM6zxgP2)=)j;~w#ah`W-{Tv)s
zO1i-PM{jBt#wM0`56E_(0U65=<vRL(98S(^!3OC9WRzcuxv4v7Mldy@jCnsBYYvY=
z?&#)+6ZN95`_eY9^18*7F7s|x5&MEpqPl{o(c0>LW?HzP_!(x$WtaoeNah%x7qU4N
zDZ-bNVf#%co(AfJ16^P!)MRXT(FhinDeZiF{QvoxUw(^6Hn45Y53p{EL^$6%7j&vc
zsa1cSny!_ecJF*$hG4VXixD5%@KJ1ZJRMt(b)>K&xkh*uqrp0{c0y;!Y@~<6C5flu
zRm&C8d$T=rM7plG8>wJA3^sAqh0%`%SG!zqiPe%AIz^12Z<FfI;oqMQWH(h&{u*p-
z&1T|BI<=q3WQUTGG<0Rx<^ZcJ=XjQ0m1Es3m-P0kB~-X_j-{L6Drd4O&$N2*MQ0Gp
zSm(!DB5d5CYFts(LzMaez1F!VTp#1YumKB0Zd2$wIhKd*v`?6K(W5XSOvL-r<3X&<
zX~T0lF;!p!?9ZTgVN(JR@tMKe23y02(79*<^tA4=BEr0U5R+bFNUrG*)ODKep5yK`
z!Fu_%Z4(o4+`HI4#1)U<j3ah}Y{l6FO@H!;?uXB0igC|8m_BkChh{EyV`v~XlFDL2
z?O5N8701XJTa68-Tu*!=7^#_TNA&vnwb1o25Z9|&s@t{wV;j(>jF{hj%nut-nOpbg
zxF8t3C9%g7$51HTy)rS~=-oS1A_q4VL`BS1EZ>;GXPVqr3?FO4Cw}1#cqf2ZzS(vy
zysoX8%w^0$*epEF>DVIR=AhjGD=<0X`&LOuM_DeufBL>>f4W)o8ZYJtw+KJj-p?pv
zLgy#m`g*ZHoWtg!_->V|7ksa1GcX*QW<E-rY`2m36JYinF523>aSR`UPo&UalBt}%
z_vPMHYsTz?waIuK1Ig7OwZV|BLeFRpHaZ=-6l}3_F-OO;DRVrX&5y;g*z>U>UPkp!
zMv-})di{92?>02fJEh$j*MgvvqgXhrj>b`pFejleHi!`>Ohr)qtFM|}ihX+ORqmiD
zzYDgI3s=AW7hf;U<Q@W?v?$BbOI^9STieWix#7%s5wFFrU3<N5CB5^i<eyKltwUS=
zLCr$CK%q1YZmaO8F@goY6P#hWA#V4fdeK?opyIo2x@)VIKD)Hqp|n+@W(<!W;xW2F
z&Q-l*moUVA_EbEXlX=sP{E)h*SKH!9v#dG*-9BpG#ADs|*8TceW_kb6SZ=V8g=Z4w
z7Jn{%(0SCcGGKeUMjPcSXsrWpn>WeX>-}0Wzb1p@Ig}aDomczTruw3vc|&TWnSh4n
z5-*m{<|17M+uhm&IJmtNO=pgc_?Rg<M(el<tsMD-?R0$*&IKoT$BL^zLc@<?Zg-+0
zPLW(hbA!TVp!_FXxo&W{f^3-iD^gDP6+VeeoOv(L?FSyfVN~uQ2GI&LCLBU^yZ5x^
z0H2?iZoESvXqw~02Ok5{PZKkpy(`l(l2;d=x?;IP+wIUcC&+>?ZXU>;fdejNg=L@L
zSjWYlA{*;P7FI4>!B1eiHY}Fog3F&|Q=#~tf*cM7m&yFE(b%iqg4ND^e>vD_@^foN
z&rjz@ab;1vZrVaPZPy1e5V@_`aG>jmu?+nfHDlxQQdca~ZMqCB=3q|nxI%*M)?^2_
z0CSuQ>-odZW}V4#kd@aP;|YER)YJ|(UKcXp>4n&ERo8n=^re1WS`_T*oy&{8yaIz=
zu}s3x&U-Fw^6Drzfjx7A%LA*Mzjpn8Eq>u8H+9XLb&JW}225o!Glg%sZp5a?DCWXU
zHa?K@j0<?9g2d-aquSKv%o~IH-k+%II;SvWT+_~PdO0sHvyY<LX!<-GHw6=^-FSb(
zi(3VyIxcN>L9ww_m?y->YahR=iapYNV$jzqE;3Ud+IH!o7zJqv9QO|tyQ=YiFWRQW
zOL6SgM}23Em+9i!bPiX0p{R!Op}eOn2$`f{-d$W=%7Y(`f2WGY_{a3?)bcv)i$cc(
z6BE3h17+c|COYnT-Fs0p9ALR4+?v5|y#E^HJx`Gx*N?*p;bsL}C(`@HD2I%nObIuC
zWW%t_MA^he+zQ@@P`qo~`waG76#mnLeMM0!-?16?Czu{|b!|5ueOp|&;<^mqE!Ni-
z;5WjHmlt8|yG-mw-^WKbz(!;9GdDKOB@^z#YpAUgRw~~lHfvITyXptrfpzYJ_8Yky
zvp#cJrU6%NBZW8Sa8F*%3`wywN`uWk2611As|}UShBeJ?h>xPZ^nSLw%?Ini#Hix}
zV9@u@A8FJ4oxp--bMN=ro**1YyQ!NGyfMfds_Z>vUmweF#jkS+H{N#UqVbJG*r#<~
zW5bd{>B#9X6uW}@fn6O-7P&fxD;V>eQA$GIBx;BzHu}XHu%H6MB@>q$e5NNFE`7I_
K^X>v{@%=w@XkmT;

diff --git a/tests/Grid_nersc_io b/tests/Grid_nersc_io
deleted file mode 100755
index 2ab855983b08f387e5ff7915f0292c972ce6a088..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 97496
zcmeFadwf*Y)jvGB01?p>1R+{4(T<vk+62)if;9sZJOdMrfSOn}i6Iy?kVqyP6)`x8
zavTP!T8pjLJhqQTYYSc;3#cI>Kmv*eyderk5IsXc!aZCh@Atd+nVEAYnLPH_-}}$|
zc?UB4thLu(d+oK?Ui*6XJeutu9~%>63w~m4zp)YOydX=U<ToBig=w;E8MaewS+?_S
z=h*rqsvrJ>bXmqzFzhmt1;Yf;@)L)@c>HC=5@MLeAQYY>ARBU!pJ46`nP2#mWjx9-
zB}GyOLz^vG#1l(dTQ8UKt(O-Gd@x)@Ver|UD8pbnipO+aM!GH|T`;s6@(06U`N)oc
zQ;hg2u~~o!XECrj#dkO~vbW{#^i1XR(t%RWr2`#;G8jH<@C$~)bT=U#<@~vyU@<2d
z`9;%9xPByODV5FktFq~LT|MeoWm7LNn?AFu`ts_Gt1rKL)D@NGSBw<pCV$dif5Xip
zVNj1rfGG1&{Luy&|Inc9d#-ruqNg95J@oMpU*4UXZ2!YKjC0|S=?36$?$p7hwwP05
z6N=)qVzX*u7&j4rH{!2#?!Zki7Qeq{!Hn)!cgu%uKPF!HXMcXyzHPsKIyoi#GxyL%
zgT~oP`@Qn#b1IfS=(_hEB#HWIffCQ?CH7(n-5b6H!u5v#34-^Ae+<Uk8~#sG__RmC
z_k(hK<1;P_J?kU6mvWsS1wSbYpHrjYpNxXPH41(y^6gFjby4VdMyZ#7MA5^6DD?4Y
zpuOe$WEA>^QR@BMDD}7^3ZFNl@Odmsc^`{nKb2ANsZsD(MycQZQSu!YML%q3$@q)>
z%mJb|eeQ@N&#)-?xG4C~qsaMglzM*&j-xku9*)B2+$el@MWKH=3Vvf0J=`0G|B)#4
z??thz>7Y->U*u<h6uo^Mh0j}2<o_&+eU?R`|3wr#DUYK6bE341Z75K0^;I9GTu()j
z=Zz@&&X1yp52Dn|rBV3rk3!!Rg+4J#c^`_R=Ydh=ToHx8CkoyLJzQWr+m>}@kua4!
z!(fUa5AltotoXB#FY$#VCBDX>Z#MWOUnTL$2L5S-KFh!d`R@h&VDL%4+Dd<!k#D=f
zKiA;1B}UeJ&3Rb{Mg|kXpM2^K`4d6M&x`n@on)P#h41j^W<#EOL!KHVYOtZ_n&hl7
zF`qDQsq{=O9x;Mrm8)$91$WOVpIK1pDVpUeD6kbwxW8ah$=%Z{Jtecol@(Q1mQ>mb
zTvKuirk2by&{J~8m6gvdnNoCDS&5A6J*J?-Q#z}pXllW|B@Y%9<H-Z^+isX#ISM&c
zq^A{-X{BdY`GeV$CQdGL6=y@3;_9M;imF*91@}*%<*6zvV~UBBJy#Wb9;_$<N&1yY
z_Q3S1B{K`A&z$bbzF)AMKB}Um<X!<x_N4#1pa40SKTvQNGA}ACD=(H*<yTEBtEwyw
zl3i^kD=w<^kP69xFNatqMKh2^Wyvg$s~Fhy;?km71)f<&(>;}tBYP5XSKd`rIlVYY
z?V3_BC9NQP^3?^o6DMC;K+z|AMrIaFx(mrlN(^QNMKh-s%qqF7sH|vaamih|V3QFd
znOQ!y1X5R)OinA9G#Rxa^f9YM6i4vOM7_+IJ`=i>WvU#PPVH6Bm?0G5$(UYVDL9OB
zK|RiNLqVRap@pKdk`RLFawist6i+Qz{<^q)Mn%!A62^$!$|_yces1B@3Z|rs(u1c*
za4IXQhE7ncsPyThXbRc6#fDnag32$i@)VR$E0|R@^X?Lv?F~@LD5=@$(_}^TS_4MO
zSq}v+l={Ysl9@B0K$+jw)9;>Hj-nU_GOIFs(sZe!^bE-ytf$YsJ9-AfB(wX-Wy&o#
zz=}~t)kSwrzdxOJSIj0;KwCw1Pb+8HEfqX{CK|$wl9`?;^-c*<5u>I}p9y76EM~O|
z8@LONZ)QPpk*By6G*hPhOdW|5%LX?xtzdFN$&3omgD}4eH2CsaMR%95&D~vHT~JXn
ztFnA%QQ34v-=8ML{`K7@9?Wgf>X2E4`AglH+5j8|TTHMG{2JzamuvE<g2`7&vls<q
zX4_96iS}DoQgm+tn$uLHDUKW!w1(+3?}rR%wX^Q75^55L2=lqB5)J(B?8%-{(<;g<
zUFqos1vB8s&<+Yb(!wgQDw&0rlYK)bN;I>I)xsKq5zHtmE0~UUQ-mVBa*HQdW@k?x
zX~;9G@`0iXG-i0Y+os%tf?j1dj=QR+MK8Tj(x_q>aL=?*MU=Es*z&T<Y?|ZD2iR^S
z3~3|^cmIs(y%`hQMyQ25M;gXmc{Pj<<}Mro*!|oF6k<?*Wkr=I8%<uAS_vGK#nAq(
z_R>pn&_*hIEsdpVOfC7}o5tjzyF)JUtsy5;Rx<N$PiaBPtXbu=f=*=mRi2Wvvb%#0
zFr+r&7}2Wf7^arx-p>wkhU^RRba!!aK_#JdTfxmUA3!riXH^MXBw4W}g8@#jE|^gQ
zyMoqj&>gI&fPE#p&}l`}%WM_!ES_n$>6MigMa3mHv<*-BR2jhk2`$5NtEYQxs5j9j
zD@!~!aAZ)xa?C6*pvU%1FP}+#@r())R6HoTRF&D#Iz7{8l$1}KS_I_uyNiki@m)p5
z_oA0Vu1Gec{QeSK2vbn`;EcOqdB}l<G5Y)&Mbl^6rlEs^xD{11i#@jM$Biqv@`^Ou
z^={YLaRno<NFNn?9C^i6@;NULB%`iKztRF`TA-_J<MQ$xV_gO5S6Io?)6H}ol*Hm+
zobk;NgDHIEQykVy_|z{%L+T(shRKK}eJs|g!uU9>nh-B(NfnHfSaM51Ji}P5=7sqU
z6j+Iu-=yFZhFVzDiLsr7v1f2SA@|}3PqiWaFO7w-7|>-J@oAuc_IJ-=1US$(0`W{!
zb@udADDl-H`u^Z3;}F=2yZ!)S6KrD*dhJopYzNqGGT=G^_qP=o@Fz>|g;@P;GYq&$
zq>r~%8?Yz!X*|W+p1?OhUB8m!QwRmA!k^&ymu2L-Qt%TTw-V1eaPSixUlY%@sNg3!
zZYBPvAQEA4+)exh6FxY;Cf;qr2gldM=a}%paW3(bP59uroA?_{_@p#hUQ~(Db1*8|
zgr{``KPe`>v_ct^X2N4=9r|RL@XR~-$ui-o>)=N<;VDz_lWW2oRVAYGOn4Yh=u>FI
zo9nUEgdY^5w%IC7_|r`IY7_o+6TZfTS4{YMCOqvq_*rPe4-O&`E;8ZG^W1t9eu#;F
zu?f#L(%@&Q34dk~iEx<-Kg@)0G~tJv@U131*J6X8HWU7=AQItb6aH)yzTJfXg$b{l
z@WHivMt7R<=bGrdO!!<A-j*J=+w)BL1QR~lgikc#FEHVgO!!}#@X02;-Gomu;V&}b
z(@gk_P52BG{t^>D%Y;uc;Z+kp)r8MA;V(7e^Gx_jCVZg@Kf;7BHQ_Ha;VVq|%T4%d
z6aESlzQ%<Al?gx3gr|QGeioYW=|LpIMJ7DgB7>iL6Mj?>iEyzAf0YTp)P%p<gkNUD
z|JsCaG~qK$_*N4>(}Zs`;jb~_H=FRIP55>bevApPoAB3~@SP_7Z%p_u6Ml*bZ!^|Q
z=p(N);S)^wEE7J_gm;+mNhbVQ6F%95A7{d+nD9;$KFx&BHsLc&`0*xumI;5o39p*)
zsu}OyouK;ql?=JSrq(rj`sr^WQuVG*Xf_eyvHPk?AnCqn3ceMr1$AFUxRgYJ_HKMG
znm{;)aGQX~5@r_{XcX`@gxR$PmI`<zVRmVOdI4Wbm|a<5p@1(W%q}cYBjB?MvswZb
z0zQK<yR1N=fCmz0R~5(=a4cbVQGqN0A3Fmu#!R+Ant=BcW>*zR7Vs{@?4kmR0^UxT
zT~oj&;4OsNB?UT90&&qM!t9Cy?E+p;m|akyO~9)Nv+D^o3U~!!b~%Bi0)CS)yP80~
zfL|lbE+(*0z%LVK*Al1^@E-}YO9@m6_-VrIN&<xfev&Y|kU*}0A0y1JBakKFhX}LF
z2&4&k7GZW3fn)*SLzrDeAW^_~5oQ+<unG7!!rZnBbpFKhPa>R5xLv>#2(t?av<Y}D
z;R^{j3iukr><R))1w4{4yMREwfG;IX*B@9Y;0p=U<p*j6d^Taa`ap$%&mc?}A1D;?
zK*DtGfm{K{5~fQJWC{4#V8C?QfiwZ{CrnozNEYxe!gSGrL;-InOxGN+33v-(y5vCT
z2~q!q>52pG0$xv;E;!I8;8lc25^fam3c_^7fu#a|lQ3OypkBbQ5vJ=6EEMp|gz0hv
zH3I%4VY=Etg@B(XOcxs{6!4RT=~@H10)C7zU1}gpzz-3oD-EOxcot#0&_J?)?;%Xr
z8AufHU4-c}12zHQMwl)!(0N?cKjAFG?E;=am@Y8TCg8Dz>G}eV0=|YYU0z_RfJYLh
zs|(Z%_)@}jae;*bzK}3oTc8GT?n)<Cxb^>rC@Z;k-9T_f9=uwfsGr=Gg}*_3T_~QQ
zh}XXbPPw&F8QgF0Iq-5V{SG4FXJz4GFe(iZE(;NElY~1ggf&6JyubobPP;-mzgF;}
zKg?aZ2+?}KQ2qoB`B*>cdtBPaJ5_DF>iw=WcZ#c}G52^DglTF&#-OGB!YrHq1qym_
znsPq0vg~dY?Pm4zLIkR}>vUDy`9SO)&C8e}9reLGRx+x)T{*voxMhTb^i5rHpjWk3
zYSXUk)R;DPV^<XtuOOjXw+V?=ZLLhd-AG?^?W^#YwyME5Bc(nE6@16qe$N3FbD){!
zQ@z(_oC*TZc_9&o-n_CCr!qj_-F8RdVI)z`w*@L?c(dbH$1RSV9aAQ&b9X1ejcf1O
zzk_#IwYTk?8Tt*_FJb-Nk0r7Zkqz+>bOrkk{n_{N)x8pvJ#DSaHz&b0;y3AgRPDIh
zbUc=onp$hGC%cwf`%;Abc+OO{&1#eW8>Q}i_#gcOG+$L)qiU~9QiO|z1kM*Bj!ftB
zUG4JSlAxyUSG7;b+moPbd-ZovA*$b?V<IEV8Y9bvMwTnt59qWXBO^iJ&rT$N618ND
zS$Io!JSkhU6J#(b84zjag?SemyellcZx_6qRqX}4An=b9yh+p&M|HJ~i<d!y3<j}l
zBy0AmYVnLgZfH6YaVkP7-m~^b#`)(9y|}d_`u(V3)=VRGPdb^El}^7&5Y<X%$Y2qP
zi(jxOfLZ!FweBm?*lO*IiATLGL&%TkQc*AM{Z;?aM&`l;J`FX9(XZ%23t3m!s4C9+
z0<CKEWqN=7S85ZI#I&+3WW{hP!oy99Qs+X4ge>ZfEZoKKNin|4sq6M!>ejlDIKF3o
z4M&jQS0cX`M1F^x!Av=q+%L1Qtfx%7j<Gh^8a&+o%NpHUKy5l2?`p_gup@*2-O7@E
zs=v080}^%lQxdsDJ@T!peWo_;8K8ErQR@yXwa<Yc=nReZZyC?J5w<jVe4^SiKAB)P
zB%Tsfi%UBwG~?EG>4PDxTl+pjKQ|<*{_#nwc0<xc|Iou~)Ao2ZX1%K2lC1it?6HO5
z#Ex@FrywJ@woh&P7TOrJ4#|>HRaycoUl@gULrTkd_GX4y84ybiiuGHX-BLh7@l@?V
zP&}9Rp<aHNs@fJQVFt3vn&^**bf^T@;CL0SiP8z!t$o{ZZjd!3%y4U+vcBRQNt@2^
zEHKG!8J`yxtq`JB1Vx)1MWn9}^(s;!xRqK(s)!WnOo&ow6{SLmQi{aYEsxhQZ+wKA
zKRypnYRlt`1WMLofk`G+yWO4?)RjISQl)PacK57d=g$ZRe(yr^(Vk&2(LV?(LrukO
zR5c$ttA01NCQu!A)&Drjl))l%SJU_LB89d<kZK-#66)6Xo~$v{F$}ApJ{onjECZe9
zG*y||s0>4YSZhx}yJ47QglR^YXM`NTGrrmg=OJX_I%60<Yu6DrG@>KZ&!Gt*rd`Cm
zEhzNsSqVTc63O1S^DrWlkx0K3V?otdYow^wmvbpi&HPxYJCcMoIp5ndnePSR2(r_^
zn}$!Q&vh05vx|zU+v7mz0OyjXYI?A-3w;MGLEm_wySuK@4TaF*EF-mB$?jr*uoTaZ
z!!2n0LojmAPT%A3Hju5&t^Gv9tjFUV<e6n#A?s%^%v`TWu^edHbDOhRu(@j(mpxht
zH2ZUwiF{Q*tQdywj_E@6{8jZ$u_w8;QR`cr>qYZdeY4jm3LA2+hZAtFPevkS46Ld-
z*Q03zCF}0dgWcV(?$s)EQ4bnq%p4uw6Y<LY0~nl#eG6P9bzJ<qaDyli6lA#LRz|>A
zZbrzD=Mq)hCH#cN75`;_cQ?{OVT$v0p)ofUB@#F7L}AeRzAkt$Yy&UH;Wg0TLg)|7
z8x7{r0lR~`#yJ=G)ox+A-ljxvSImk;WUVSUZcIPyMhDwZ<aSuOb)DOg0u@btPxM~~
zmAUOnsxLmE7B7^Qy-C(}iV$1-z-U+9`j7iK#Dhco5^c+<XqS?GGE3C$!6anq2OCZB
zOYklibv$RH%)svsB&KhYHlzJyjE7X;SxSoY${HoFNy&3w*{I~jU#X;K!(x-VSE=4L
ziR#?7tc20ekk{<C^gUkL^`Hik(Xj?#ZR)%Hrb_0Fq4k>VS4oLc@?wR}b|uqH9Lfl1
zKU?=&HD;^oU6rUpU8CO?N}T-(tsp^v5V}GkVF<G3zl^B)g~%w>52NCP{ct_;sGFrm
z-FPlmwJkj>{<ThyQiX4l^)3cXOxv^ruBbyeYJzANixFm_-f4IoE1>={GesVYj675h
zJV=aj!7@zx&EtY<;z2vlpp7EW*_0=8Od!$<c|M>VWWlfop%@RGBE_gMSpOU|<T!86
zUiKDc_LTRTP?$wTeK8ijrSl)Kq@hi!HpQOk_D^j}O#e!b*D*f$(+&PHf#re$Y&i=L
zZtW!c@3|>9TL_&5wCeNN6SXZmXRquZl2za9b`eorC+)n^A?(L<dG8*r`$y@~-nMwO
zPRHT3Xo)VYDR|!(7VcD1PK(=s>RGT3#bTG{)oWDWe0vI{SW&~G#%xUA#P*gAFR|aL
zv^i?<)C4ozue7<AI`E*}x(UxWwc~lCa_d)k&R>Y<FO^%r!*fw1p1)OY-Gk@pop=r?
zw;smxef4<Wr`+0&=Lg&He3Tg|Z83Okaw@4=HA<?wP)W_LS5otqDyfAS%9pk&sTJ)?
zYIUcQYC|jQ#|Ea(Z4&M5_bibze+q_LF5dvPxW>?w^axk=uBgxVbazt;CqJMs3VMcR
z-~=C(-6h>a5*B5^(z{IPOj&@52&%|wPnzI&>WS&=q@sLjK@~ZNnRMhHW>V6WVJ0oz
zKFp-1sl!ZqDjQ}}lxLVpQ?rMeR24tW(3NvIl;s`{ZA}>tb=^K3`kFc%3M(59jd_Mc
zWwVDvXYs>bzF&7YsdHBejV+KG8w-sEeUq}F7JW}PMl&#i+@KM-eY5Q;uvN@ams2&s
zLm}tbrzz#mAvM#SN_p9kg#bK5mQGVDHrFdDWrJ&!lqrK3Dk)P3*8|shNJ()IhFJ4O
z*T*Esol5>sh}!T5Dqsy#cPaTt8Sw!k*5jdKKV!Ba24&9woW|F!<bRDMC*Dv}b}RWU
zL<!38naAgcvXzv!Y$g9wCFNUK(q<)pJ1IVOC@DLD{u<$0C4Uuhn_*$_3MdCM+o|M#
zLF9H(o_4TmlQJW2m6G2IK6vUR2`$rUuTb)LDfwN~vXzRh_4^%4+xJT9U+wkaHqD__
ztT}{;@&ihWEGv@kV`-Y+0Q*+RwO2`f-OdU?sanZ&A0nm>ZbZJHz5xd1$4RK<pFrtR
z%JoQcG+W6#;!yIA3d%N6ehA&<f32h(RPvjQ0_DNepgzclRUij1;PDg6%|bOELXzW3
zemm(;fa5XfYlD))!gUzv-A*O%b$bnjQ}VYV?dN0<bqUU$VA3u%qvV4!zma9vm6Sse
zjoJ+Wt>m{M!7e12GNc|EtOwnPMnTspc^)V)^=)YcO2sO2+XT5%oP*n0_wCHSp4q3C
z*>h28?No9bRkt4M0<*mNLU*W#`SuzRvM}Ewdg@?^_oz_|Q1^##poHI(C-CKz;vg8T
z6Z!_1BS6{*Blm-EDEX_EloPZtrEL{C4G{`Pt4o35rJyJk-yKr&H==N$_zY!j7DfQ6
z<!}o`9r$Zf@}Yp$pwU6|g+R0@`QHPv8i+kCQlUtUDL*7fR_6=>^Bq*eSFD`X`1@pk
zHpGC0a*znV%`)NxD!U0|4STLw)Bp$Ah?0K*{7))*@7ZgJNtGeBwwvmoJ*<W%1z$tc
zJSa?aC-nLQ!H=oeA0akLc2YGvAxa0aCzbpiXrJIs%JSL6>S@Ct)9Us@kNZgcBT-xG
z8Q+X_n+_ozgv>{w_aWh0YAMZ#E#I%?qacTZ_$)B`2DBT=XbXrcI>|^HDKZxg<0y<!
z$=8u)E4&)YosUK!46>7K;aXYFkAwsP3WGKV#Nk7XMN<!;iZ&tz`xTuy=deZy+A3Pv
zQ6ga9T}lc(#YT`ELoMxA@=h>f$}k$iDyZU7SrL$zM%2mb`3MDd+6zH|<^T)(4o{7T
z_|ys2ZpWV>`yLj!f!Krm(vkWA{sQEH+RmG5PXn=%FZm<?b|5|WEJ(2f8DP-@6^fJ<
zy3qY*w6c^{D9sx1Z>BLcg0h85E(CTf{yqScFOkJAC6C&Iha(Fh<m8+gcL>od$aZr*
z@)LCk*->QAuyzJtG47;LQ6uyZwk=@0nNiMRXiIF3@Nb_G(F)#ebWy0pjY{5DF#Lk}
zLBpUo!__Ml0cx?#UIAiCw}x`iqsW%b?3zRs>R`T(Mud9*5&3dxuopV~k}SJGSH2F_
z1rF<3`lGBZ0Kw`>nQE^<q7Nxc6XH%nnAIYe8bgok*wEHNjrseK?kJfYp~k7cCTRj_
zc==LpHb81G$a0;iOry1afbMJxI^C4p(dp)+->BFGH?k4ohI&*3!mm)~ial@}9Vo<h
zQrF{Q1CT$0f<9#vDYk)Pi<06Qz7(+?C}+hc(G0c#|2g6hgDo5Q=Sm)WHT3DgeFFyU
z1-~PMBOp8of^B5dMqSY;;dKyxL<01@qQ$_o?j-+a!Jh>v!?++H9dE@RWD)?6HA>zp
zbkoDxxX=a40<EJZROsMBlFyWsLBo|iR({2opkQ{AB4AK#MOT2i&_y+v^phW)WnO=`
zhM8_?8WTUDdrb~jd%SBgNs>bmIUQBzzk;Q#j?+2b#FR!1JFqhGUpu?IbLtLz61W2S
zEC7zvI6{?^-^YN#tk|AHG{-DJt3Hm#eHc0;eG!v%H)nQ;al11AKFpS}^5*>sv*T<$
z8<PNKNv(Yl=A;ahunxLkxfG5g@w78m@%c`4-GgP91J4e)U#!F%;cblvxDb+v{4sSI
zih!S$Ja*LZ33Rc7JCFMP6ou=oh!=7CU%vxJYgT=6sxM2&KtHJygq)Svo`KECnnyBh
zo_<eekLx&_^N`1PfZE4dW>#7Q_b!3#Zc`dMRqY%7)$hrwMU<Cw`qWyx;01Zakd5RR
zwPD^@%S~K_^U{%p>L0TepiMtZq&E^_ekKOc{z03$)=CDLF?t4J3axs-jX~h~1@{?F
zS6i4E6E+$6Cw8me#st-SJa*Q2(|oo!+HOq9F;`xd;KBMvmR$4*hQO+fZ>w6v1rz4q
zF|4;Dapf&2>wETvloJYO!B=2jY14;b3zLgQ#JDk!VySXc54oE5#9{K1Etfx_s-?gy
z3sxe%8(R?dgl=s-;7~4IBPw<w5LkN1PCuNT&V39DMWO>^l{xVkU`qzFzPGZzSsYPc
z0_W7e*Hgb0Ro(ER&0$-Cd{BAc=u=Un@KSQLhSszu=75rjHCM#r;aOoxh~IiZG`8|~
zgxe1&w||YW34fm*Q1aSD3JM6z1g;r>>oNO6@F_@FAp|*H!!;K)Cn`f~?YhjIisGYw
z{`ROw_%!0PPdVah`Y{=eer5~4uDNh+OmKV>Eci5+GG{H~=Cr8GkV5Q;yDnHO%y%fw
zH?_xnd9S{#y$@^tilz>0eg?8(Nh4e9&d%Je%&*0wM#n(?G$aj7#mogNlMXwEm0XXk
zIWlA@>*9OX1(x+}`Z1^<nv!i)sEn;tN%^a+`V{C2of*c>Osu2`jYEB(0IlQiTc{Aj
zqiZhs6e`3LC5Wg$hO#AZVQG4A$$Z%|q9wogO&?|dM$|`({s&qN)(U9V`aO1*&1`3T
zST>#j(9Z+me_%zsP&V|w*Icj_WqZNC*(e)BS+<juw)e8tLAeoSYv0yK+2V|{y-RhW
zY|B`-1;30eTl|S&*~Wqpez?|N1932n?Dzb8!Xfw5M0gwe#By1Ou!Rp}Flt9_eTtO`
z1Z)l;pb|GC+<`*2hbyoR+*pAOWg*XsUdY$bFGLjb_|`rO$@-x2eomGsWGM>C%tS>N
zT|z)2^=B`PEcq{3ayTuUt|ImSy;^NG3d~RzIED4pd#yI31B@u}z`hE6xiJ0~eJ5Dz
z?Qg@y20QG(d`&&QXBR7TVy9uFVDz57o=Opzl?J9Yg!zZSpf3`NGcW#MATaNVy#l@e
z2(aK%jOwdSB(GCZ9=(EzvBm4F&IsXd58(<!I5mh{HjlaqCQmoe=-sf%+oFH_l`MhS
zuN3ksq(BXFX;Xg_XFdpUn@s<20w}H67!|aCF;a^yRe|}PfvJa4G1GqfqXwoigqZ;h
zdbYW%xi$V<$L)?g9Ctbj?r28e+O-_QBH=nu*79zAI+dmb_Rt{2q78$ND=cK!n911O
zK*pdW#h2i}sas2|Tc<qrIy-;%2M|HO4s{mzBQPd{d2qZT6F!dw_XD#sxf@3$9Qaz{
zg;dLjp>NaMwt!Sy(}8_<bl#ys84CJ2isknIRp=^5``&67%Y%Ah4GX>e&A^21;`aj6
z(=I&MS&Dp?xyWYC3<#Cz(bHI(E*3YJOi_bf$XI_Xg*6yyQs<H+!lX_K;R-{xb4U~&
zre`&&Z3a4MQVYHWJ7^|sQty!>Xi^22QwF({&q42vgfaSpIew#0GSY?(D$Brx4eAO5
z6E>)Gh`|VQ3Y(+0CeV%Aimbe;s78H=AY9Y&PuAcc_*F;xDp*=1{YR|q9h<{d3POkp
zA(guC!5=jo7>0B}2M-qno~0R?;9mg_S_G40Sk=z@_nX`ND+aSz5bI|k!@zeKw+9ye
z1Tc=^OI>(6`URfuJBp{hd~zSd(@s8Jd0ga4-kspY@nStgRDcLE)d<_9Qag=ocQfJb
ze0mc#6ZnEH9%}d-C8#H64J8;xf*5FwMIFJSncL^jvBEQHg`Kp*^(YA0oC}@|&qNsB
z&orHv@$OD?;rxJhh{s&~55p)v((ibK3rc%233=4pj<PXi&&IIP2$vdRy%8=n!WtQl
zs{rx1LVo3{EnaGhA9Bb*jWFb-i(!%xCLjc<ihv)rC9YzC1A47WP)AhR;J=zqVhgKM
z6&}kW28Nh5_#ZCS&&9zyw{M)P`n)wv>GRPAyZyzvZvVT2ZE6$FQpCEowQA}|_n!fq
zQoVa?^qp|{OxJN*<*5_=@inUE6~5c&T_oAmGl)p*f0vQ^t3A^SYX7^`3?z|(k8)aB
zV8eB`JyGZBK#}=bl@JIRLjoq~=qE6$56bgR%(JFH-L3tgUm8q3urfg=9GB<z|D{#t
zEdyU$kGw_NA8}4GqS&N#``>Dp$z|Zv30X>y<VW-+J(H6XV~ca`m<pmw|I`IIJAGYF
zU87QaD(aIn>IAnh2YRV2RegW2L41zyW=QedMAi3juG{x_Disw+tzk|SF}@$?i(fug
zi*FdKIp4PdSOH+=SYLdDvc${DoF=}<`Vb3Dyuk2VV4fd$WX!wcnDY6tN5-r~(&Zqc
zh?@BJc~9VZtg<A2r6wMEHs*4dFWx@ZH)Z(=L`+$BVyrfFnakH7n3X0>$GLJpP7t@m
zvmeA>M*p$F$L^M#%ogRbYvHzXv?eY=B)~F+Mf!Z4m4(#%5e;*gTkF<yajMkqo2XKs
zpQ0fe#yJnh$rNt?-;<fp=aqpkr8hQv^;1l2{8lzhfy!n(1RELnvU;%j1uJ`1ij>BU
zPArikImqh_0g6ULxtKNw{eJCpcyRd=x!aBvnf5Lu_(ZAwBedeyK2p8kcd7n4I3%?<
z4o3yLTRi8h+ELE`tFXeicN2*F?LhtGnVS!-9m5VN<!9q>QsVmwU;!LfhRfHh-tO2+
zG?29&PMlPEElo}R5kxm(pkAKej<2Fle5FoPGP1A$pOJ+Hw2UmQ-DPBLz&3Xl7K%En
zK1OP=;=18R#&BP#SvjMT&s=?Ojp0-4G-b}muFQQ^@mg0$wd$W4<I2QtdxuN8{!6Z0
zr0#?8acb%T&j5HQ453m~@6i~~S@@2vf`ErQ;*}+1-o)Q)Je#HcNJ#Boi*sV$<MFdD
z*1AGxI+w(Q%=sqhoo^=aYY^m&cV~Vy>pLjc<8b-Xa(n}Fd_(On-yKl#>0DZ{rSCD$
z4?KTl%s(C4n8m)K|8yuzhQ6i^{ks<bDzf`0GInQfn6(Xu?32tRKXi`h{hb%V1kjJq
z|J2+&Oj~H`A8?+Mm8yN_(hj>cj4H4nkF1f}t^KHfvXL94zZTGbz27xxQ&@e%X*`lx
zdfhE?;NpKrG;@qh>Or9Y9;rD@k;rGoMn=PvF&GDo-qhWVz3Ov9y{hk4onshDd9SXH
zo~&2Fk#5stKH=`PLwR;}M;w+^mdSHHYm~Y_AVV^rCq#MHUW2gVSu)?oY=hmk!B|om
zjE9i%Vu%3$^=zYt;af*S3%<d@7S3R*$Y8b|mNCU%(eNj-_V!?XJLAoKb4`2`1m6TJ
z-)juMrQ}=K@ZaP+0(_m?RQn*S+_&37WY)$3(3!O1*2?UOVTn&SIOmabZo}*3ys?eG
z=_j0aEmVarZnq=a%=4un&lYO9JIrnmCM?;t-Ae73C;<G!AfBK}Xj0y3!qFSI*5QP|
zLk&WNsxhj6wp~RN(EsaW*0~1&Pl6&LY{#h`)4#kGk6;44d9ap>c@VG(js$Fby|WsO
zmAYaWB>82L-%SR;tO$OAJ0w5gP5f#vVOrC$N~z644$I-7gHs>&8<^8D2#i1+bIPE?
z(i)aiVfDx`tgSM8hTH!L?D1y3N})XWL7z06=8o9(uY|5#=mP3^gHJ8iO8a4O_A^wK
zLRR;GBvq9XQMB`lWzm2)6|F86wQE(CRC_XcU25=3j^MR_y5t4CnO8h5#aNWI##~GV
zVkU{e>`yEzO+#cfmHvY1z0F<AX!KaeRce0;%rc__;xL*P6(EDa={=~=>Vujot-O^!
zsE<(qizNN}pziEJeH@r*a&y9BH8aW`O(&wn;o0+Nm}_t#zZp#Qm<^a2Gy~&{!`U|N
zRA}s}i#P=_2<X1Ieh8XGAWbxPMbIoKO^%i3xd@uylV+}!=EDe@`$+RIE6tGznjF$>
zC(^qnRMDQ%L1kszthL;6X=om;JXMd0O0aBtJQ*zj2D=n2eGbmmC{Mk8B22;&)3y&l
z^%9X5E3AR~qG-HhrkT|9P&Beq3z;Vx*<B|2*BZigA@`1aQ^eJVlG+iWkD$nQYB5~x
zd~t>)=U7}HNP;VXF%jHr5Hyk?upR{Y#6;<FYq{8NmS3rTun!xYinxw}9SI%%JK{TH
zJK&Z@M_x}RCfCr{(ugo)={Y+hmqs|Q*RO)i9Cn<GYF+upNv!p(heutJ4BxnX7k%q@
ztwwnw`+?dDI0$QDPU)jC2PriR;6x{3l3uAz>A`0^XrokYU_+RX)!EjzM@-L^TAsd)
z#)nG|ri$!hV$`$_bl`geQdX-IuALo`hi7&a{vG$2n2$G@+n^u4843-I0>(sekwHL@
zroT*rAy$HTgP@fJb4jqDO)zY!JB&6;AFdaYU?VW$<ppE>h_`XPyAvJS#}4gtykdX_
z2F{aIElw?7%;v#kn)19bUN31Gmkg8*1B39%YIeN|;;5|t(Av=XOgAP`+C;pN5I)<l
zYU9w$jY~S*q*Ud%j1vPYmv4++#o0YyR-(h_u_s`V(cPwvLt2bnzjtc;o!U|T!1*+k
z%@`dgvNw~fqFgw7SS$ed(`U7?cJ}J`tn0aAC6*`CA((Z%%vu#CKEg?Sh1+*YPThA(
ztqVJ>Vx_#H8kWD}0#NC1gS+ple{M@Q*3hS{VQy<0{zwY$@iObDD1zLk<rp|c^hDfu
zO$5i9hsBoexN#lB*s|Ut17Do6oBK_7ZkpK6-K%dwEphqzX)w#iTO#u|gFUC^X>Po`
z<;zx`nVrgluYtr#aEvO8U*x3NRs1;*pJHT~jcL1o5hW^)Ur3THck!s-slGZkY&URX
zw(q&jt>HB=yd8ppcRON4<>esX6{|VS><S&RWh75X=i&4><*EGvh#YOTOZ!HMcP}KF
z;2(1cSew29Lc6fcr|nn0+vBhq;MR`kXzf^^uiJzJ&^T)9fvFwCm{UBD!ZX%-#PgXs
zb$dJsF1!)NkIO&4+vU%}OA6~e1Jo9@5J=IDscKViD3|wS{H#gx{D75S!z2_N%?WCY
zL)e<&VIUGiObtYmK!`2qR#n?r^(IH<80Kcf6fvnbHrE>|3O3i<NR1CGcm+b}gnR3}
z`6Rfv4jt($V0K)3Yy=6zR)iD$gYcfge8B>5BpFKjw>H@way{ce*k5=1{wf6M7|!iQ
zA)cvJn4$4(b21Mi;}Gayh1@FU32CCK-iUYNxKfAX3pl&PC6P1T*n)ZFEY3hT<K-eO
znlA5evth*ya}Bum_*OooshT4pn@4VE+w<m}<?!d?B>)&9PPTI?qFEobiUzdVjki@Y
z)D;|S;&UClhf-BThTkTu#hZ<asHa9DG7Xy_td!r7gN4J)Pn0K)B5#+rfreY^YWnFs
zzEqGV+N`k%jG9=CG^+oyMLf_>9_feC?p#_6U#_Zc!hZ7wA~{%6LG`3I<%S6s-#Ele
zgNWcNOR~c_D(pr<Q<Md_z-Yy5K4>^O+Ws8vc*iLVd8C^OT)sv0$*vKLn6ui{5vvwA
zyHmHSO+R7j6pNof-jDZr<m?*)xU~&A+86pZ?E2<t$6Q(<N9*hujPnSUgIxZ2FYSt4
z>%N+Eh0C{?EZxRSAyCQ)ZL_yMrplgv*r^3NuHuSa=zW<WS;sVKPfksL{>UTnibISi
zt*L>HkLNnZr5<qkXTJrPQT9La#(IX<3i1=$ttX6!CZ^Vg3kYMHwXv$VB>{uW1aI?6
zt*NSwHCgB?hR9G-h3a!9iWAx}qC6!f9w&yxt0y=~G#KZW`@6c=WRDg-kTQQ2`#K!$
zN$Th?wQ0P`fU2p1fUKH@2>J2MaKmcxRt@IPIod~VbQ5AkD@-Y~S$SeVW{Yl2cSELe
zW;4wWHUFNy1`HU^L+J9K^Ag`9AY<AmTIw!+`byI3Ioduq-rdRVfNdc&nn(@E%yQ5~
zuq}f35fG7Gw@FzrDHar%tu>O8?wiM0aI&~VG4#ns*@{TB>?m}Fc-bNjGk*7osTBJ}
zoXdM6=7Hn}FtoW-zo)H`seb%Sj1H4gnYDmW0`G~qS%YQ!@it-|Mmiw*V(R&5Y^o2h
zae|P&2bvf<>k067YJ%#+p6Wo=$EV4_45p_+*rKm%U<a`;yKbH1J!U~gs!)i~hO!_Z
z<`TBA9A8CV&WMU!93+W#6|d#_aw@}}dX&rj!qPvw|58d{Lo~aN4`}HJarj<X`j_C`
zF+5ZrVee0{7*eN)=UndA_PBg`SeceQ&?$LRomwYN{sNl(w;gF=d4g0Or9pddfBwj5
zhc?`)@xicmhxa7R-I012_Wm!}`|WR;?fv1Pz2gD)K3bal=&-dnCpf%oY3>_c+8$x?
z4qyBeSnpdUtUjhmEpBr8+-sop^>`H<wKcN|CloGR6~p*ao-G>mt&-R3P=<hrr(iNa
zROY`S8UTz%?UqL4%xqKUe}M5ER=ldAJsZN=44GvXUv2ZyTVqv6S->eq`d4^0#a?Y`
z|Jvr~>A^FOoOJnA5Mu~2I{tT6DLL92L>)P)77LkG)fbPqTaXWI%j`cu7n!?wK^wy%
zjNSCDf1s=D&tg@p8qSn;%b|W|bB?CZevEJ9;&@-TUiBU=aW*#ed5G_JyUO!>0w>K{
z?hD^YOmIJRtdS&7##O$LxVfu!l+(Q00fRE^0#hdF8z8I|`Bq`PiTpZFWA=yN4QG#>
ze-NNK=g)u<;}iBHtp>G`^HB-{^ggr;WS}CydPpHM>iFl%KmUZ;)8Fy6Vk)r9Kf#m!
z3OMKve69Q|#*Z?#EEGGB5pRVeW`rWjne1$26Yz;QYvuN=o<NLtMC2a01x`*s3}pm<
z>Jk_sN8oazzacvQco;2I5Ey{CpuaG!Cwq5ixU`cwzAJKy-#-Bv)Z#P$`AIUCvN4uJ
z4=fu$Uj(W>$(~M(B{4jZvj7kJZVVyV)0bi7syKK0`c)t`Rs|NYL%|wBw736$&<7x~
z?_3ATzfu-F1#&n69KXA@aiPQaKV6VT>wNymaji+)n)+GHNdYKJKGu$<?wS)HvtGmd
zNO(a^S@0q#V_N;wV={FmN7r_xuE(G>b%V=yf1+B9aU#h!d9Fo(A*tsQ5ypB3i&nWk
z9%o*Mp%JQhbB4_nh`d3=dO$c%IxE8<z}n-mG&Ya0zftlg;CSu?9K)T^spJjx--oeI
z6URGQnQLcnSLKW74kdL^JhoYU=jLomwq<^!%s&%Vg0kZ@=Yl`6xL;!XBE1o&FJxZl
zxel_XH*yc-V&f^+a~hs-0D6!(<N;IXXbG<m<_#Jo9EJttuEt1%WP;y)L0Yi>l{^>Z
zbwOSi<UPfo3t1Z}b5`b-+1oJ#!6>(5FhXPo@jCh=<h%q<AO(Wrq#6dwgpvtie@?08
zXaohc6PTBQ{tB~&pBSsZ1be1aLAQi8dpRg}+mq2cp?kY$kXvp#A?Wqcsmp&OR~iyr
z*k+pbpO1&>H52Knk~_|%<}D>jZ%kjuLU2Tw6dd|O{|^imJun8qoB}86vvD&}hk|{j
zxLKz_Hd!*++r9y-W@E$)_bFm(gI*S69y`2zHe5D<Ga8j4F`<68=wF-wb}I}14qY8k
z=0L;gIJgRr>D0Q=yLnRa%Jv*|1*q8o`w;AcU31~O7#v=mlZ`{zP0A1?S&r6&SAwCT
zyIp?Ih6G3I$7cI9Snx&=vS^gu5OTYr3$~B`$DTrxR(WYiD3|pyAn6~>1LlEo-r7Wm
zx8-EjKU}`wh(RB=4pQW09bfzoojGTB^qx<?Ls$A8#}a$>_fbY!G)(L>ERGu;1r$o>
zK2*O2X*y2R&&R4tpaC5oR$)Rc9s&zasAnP4vLEK%U8-Ulk>=7)sJ^TDsPgSK4$0AA
zztNj8DOY{JRmBWPUO|y84AEWO!HHio-iO0EVb8C_gGX298s&+5#o!Td)4Q5ZV(fQB
z-;BW}96zTI==ApM>i>uVA_n8tv?=tROW6kaoT2)b@fovgv}G|a_nhYI|NN2b9GZL?
zlml%Cwp|ur*fIO5c$dFyUVJKcMvd{DZ~Tepk34!p%RV7eX|Nord<?Qf1MR9W-E37{
z!|{ptA@RQTFI`k;RJW@?^EL`WX+!Tj4&ly16Z**WWXtep#Zx^AI13&2Z>-jbBNX%j
z+IWuk13iBdD|s1eNX)I6JPU7^>lY(4qlw|n31$RP2|U(b-49Ziw%Uyiv|N7WP4u7h
zIxDD<!|xxMcGp|5g_NK^kWUXr=y&9>I*X%k?vk-wZr@_gxZESe^a$s`F&<sxPTj#<
zUBoza8;+Zx4d?3BSkdShEXRkHr(k|nOTqJC3pv^eWx<n_Ro1IuMLW%<p>WE?QRk@~
zb?)o<1=k}m5S25c;Dek_IlMo`&T=|Z4<SbmFfpN$`Tgwql*7A*Bg~IET1u(=ijF0j
z#}=~YR8sM|u&+VFzQiFM&k!t2VWj7_CmzXkXaj^-5NGw_6rPG_aT2g-F%OdXX>OtU
zp)Xt7+xQb?aO66zf3NmN20`0aeKl0Hnu_xSdTd>z5MWLb7nD4MwN>8jG2#?_188iX
zfexSI8!o4qU9qz=Qago4VD|sQkUPufyW@niWJ}Bk%92}8;8cpTWPQw+%97#Xu{LIw
zjVG%%qd#{+2KaAm|Ea!7m{xv^9y}f{2X|(CkG?y8BxaPD%%HO5h%x3(IwE5LdhiVx
z9^SHNy*fuP<rqVme?DAIhoUcrwFX8=x6^TpbW7gVrR1W7T~^=?un2>=lBc;Nm0vhB
zBz&2X?UNu=y{IS@p1iX3Y6FvKU=sCd#K1-U89D2z%-$zPZD9J1t^p4D1F>++17XHV
zDDx6}-Q^sY`Q#NMmWVMjf>S8VqI=y$8Bx<C;x-X64m;%L&Bro{#Z@V0>>P<LwPF`a
zY@rqVvczJ6$jsqwiN$8F8M{_uQ>@s{5}Rzr?vmIfEB1uMVv=j-dpi7;SXZ_jvsG%(
z13tWt?-sk}g93`64j;!=YKQhp*RKcNz+UN&QzpxHn5|d3eLd(71JhH!A9~R3?v-w9
z54!EW(tXl{jvDWo@46mzAN5Md^Dx%(uI`oYe|pf-ZS>6d?>*?2_Dc6(J?LKPmF^Ec
z=w9fR&JV0<r{&obXtEgcB<M%~iT271=S;Crg<`*AY>hSc$x!V3jD6S|OIDKeON_nW
z8v95nb}nOQT4Sq1vBiv?W{sT@ip^&1oz~crQ0yg)y~P@PM=16b#@=9!y*U)y@ejm~
zx5nm#VmC1MH`dtfQ0x-MUTuxNHWd3i#zt(sUKNV1WbDNN!+jb1lgadzIIr5D=T+Yo
zC+gtF`{@V&4ki2E6({usVuwWV^n~m#Hc3P~U8A6Al8Dw2VyQ%kv7VrKNg@`7C>BUW
zO$hObKs+q^l74!*gp7`&NJ42cx5*MpB-H=60VlxUHcwlKWv}a1_vrDeQujx2#nSp5
zjC5{Hbo*|>6&2GkUyj!tqvM~f8m0QK&-KO6Q+>DO624aTJx)o1a{FH5)ANoaV|;a+
za1^Dx$(h+Sdv#v(%CGTKO2fl^Gle&vE`J$Mdf+uIPn^nC4H2P)BH9=+CKQnviue~J
zCWTlfgd+Zo2y6X>*B^3mnHbg|xXN1_Knj;{e1gk|>&?WzL#&$lrQ&@Lcy%S_%KTiH
zgNcW>%AI;lZ90Y_-AOg|b2W3D@)$Q~)XXh78Ru5UcDQ0bQ=VSs&fKd!-ITBqH$Lsf
zc#*@{WLM@-$`e?y;so^{2yFVv*TrxYjG4O|BO0vlV1?ep)48}A1KH`Xp(b4ZF|XtP
zmlv5R{jh%ht1zaRtyn#NJv6|02DlvUVwZMj!bE@k>n?44B3_he+7WAvjJN26LE(zI
z69f6x38rzx$_y&~uZuAfI3MDe)#bF=a`kI~)IWnF72XdSx?a(x9pdT@UnqSB{fSk3
zaqi53;#~}BUBQL%^Kn#BF2~~ZcAQ&lQB#jlNskpEwL9}iYzyZoV|Th@4nli5neEEc
z%?Y36DB0~;<mOx~IVW?q^28VrLwh@^2)&V-!H$aWbmp#RHo3K9O6@&pQEq=$mn*fE
zx!{N?Rz7eu5;qKC+)6Dbi)dk+lsXPikR<RiLRJuz9uS96`{ALZx7ScHtu|H!FucA_
z?g#dKSM_JUf9h)58J80?Ga)dUJ^(93Rmk^PM>_W`<>L_6E*3WjyEKeQn-X09Td_O2
ziKui{8%Y#*AA3W2yW?lw#`j50<o*XSC;l8mTkco-@&Vjt%<2pbRugW+4N#otNsJ5g
zs|iU#jG77i+>evGRj;~mq<5l!N=5?prns?DkBJGkH1Wixludkp7gdFF$-w2$MjtZ)
zODRy?c+>h7_g_H^4Ukp_*oNg9VTgd09tT@9qKpt&Zn0|ns@6>KpR<c=0m2nN`wGqf
zH-`EDK=TjTc(XL|rvHW^k3k~{n|N?OFK!F1K|T)cBZu~>@C<O*qp=m~tHgw~u{$@3
zE3XE^v5bN`we6UOAtrN8)h{`-%X2Kplolq1cDn`p0@+%pJ{t&*!b>~cBuR6KnURGg
z-yq47B^mfPcE>pG;8;!#T1a^<Fu5gRCAx;sDEnzFCyy%x+Mt6VfXh=!U<>;@oWM|>
zSX~<B&_2XHkuL2BH~lk=*9-2z^cWMpoZ^GrnDLD(9qSvl6k~fjZXB;)iV5l<M@*+v
zYs8{GwsP!v8H1Y-+Be-dB|5d84()o_)Whf-ew&M9`EP+fSu}msKWG3PpfhuyvS1?(
zy?C8N+og3lj%;G1Y}%eIFaM#3b>k*UtP^3mxw|{&b8Y{TO^(b>4kc%kf6B-hwR@Fo
z8J3*8se9P5`+~K&IOqc0r?D4D2kBOE4BI~nS2ST)rR=SPf(y=AtICoi?wF5MylEV>
zRy%U!3r8ky<j6VZpZ#m_Y*D>G;Bfk0EKHT3S^k(fbH2`GpFraUXTBd@p2O4R*)|*s
zZ~O(m*5PX!&Xs?KziI0c;_oC*m!JFq+mY>PGhJ1yu;0{yiTB=hKw}36*YBLcM=Yde
z#NuJj$7<%Ts(}vegtj$+mm6((Z7y79=fD_kt@rylOlwc&Qukq;KHu7Lo%dvnw$*bH
zc91++5F6u*QC?c5yt66h3%u9<zrxxHZ&&;*H*z%W!<51qv)$pH?K=U6jmkTXF>Qo!
z%2u*pMd%>ooXR`xj+js3eNJQs9&Ny!e-KXmFW{*sOgl%#;-EB=!vM~~N)D|x1<i<O
ze_%4&E|A-um<6Asb)e}uaZ_2-SbuyvtjeLaA$@Y}?1@~YIH~PR-6$sVsjD2BA1UrO
zc$IXgX2($;n5~26`y-p#J||d?(ucq-aE`}#Aw0CcAc(a?R@#5{!;*Kv(Y+Q{sclv2
zZiQ8Gog>4CLqS;FNYrp512-pd^~+a@IsZ8L!6u&a?{N()PA=h`OX_}_5Z-smhR4cI
zh9+R4YdsgR9cRf6i{}3_9b3u@v(16i&9UC4dULXxxfZw9_7g|luwt<ye8uZ_G>`$R
zKYo+Ecx-;cLXlSv1cVP>t%jW59q<6?uPyCAFPodyVm(|OojSlapGFUWwUyCo@h+ao
zNJ4FCKXxb9HSz*ITvU?17D=G>IMp|g16uExS8=03T0F|=o7c+hu_ls;mknMO*VAzk
z2<%Q#eOfRQ?=i^81geWmzNY$_$00Cz$_q7>1-xG&wElp_9*&uvzRMPK?IgviEYZbc
zklU9go9&}q^TQfnE+oOLM6YmdC}+egbX=~c-LcsiJDlc%TVqw|Kj8{ZVCJNZ^SiIb
zpyi(=qm5*d?m}$Ra)Isa^urw^<a(Ki>$oQ*k34t~Jb)n9m9!?@i-=y$k%~-iH&)w%
z>xagA+hptcg&~rSTxDuTce@5Vx~s){8-}1NmKFI9O$rLmogeRZ;h|VCC}!_E&qJ!#
z%9m<Xe;(GrIxxU#_6)<D3V5$3MtYmQjqj7%b2?ns0Q}u3Uz`!wXJlfsw*>FX=p4Tw
z?iRi;b0<CAG<vuz;Gp2rPGU`1xHG&fvo;G8>dvZ9aBT*Vy*V=60K7p>T~n2)ebRBC
zdHrAY?nFDnJ56ixY7B~nVX*4`DaLaKzGFT8<Ay$g%_H+SuEFMEtIIeH_j{fgq<sQ*
zVS9<Y^I+VaU2%7wh&vB*@w&<rRUO_P*wx@HYr&r}D~7Quf7z^fZ=(7DB{=-?1Dr9s
zTHGouhvxv%E3FgzgZ?pS=`df1f9P;$%r3Qfe|Uw@KT0g^;Rw%29Cl$t^B#|>dIY{d
z+lD)E9vy%u=g34pIur0p6o6z2Bnu!-0<crWWJw@P0J###6+oc`3I$LhfeHY%BWsZ5
z8v%ZY><_b*x@`ghA#Ml@!N-sg5-^02fFXng3?U?72q6JO2nmFRAY7Y3K^*8Yp;Ih@
zLbsynie-4`=rOM<^Kazr7}`w1C5jj$q!<ok^iaeNVVrk$p2<NO>z2m2XVVKH!1Ck3
zdQ9fTk>FL>qF&LAOqYL#Cw(Ib%;zh;yL<I#DG~lGxmSONw^6MA>^yj#|5JZvI{#rP
z#q7`cGIpr`BK%p}f9TIH-}nFfv;VL9Gd3*VGqwDA;ndE7QxlspmE$a4Z4vfW@J93x
zVv{99jl+c%XrORt(&-_36r#V6SK}%Z#ck4h!i&5|8ts@ZfW5D5y+z!C;P!@Wy=da1
zyBDqZBF6RAdWS+twq7?HY1msE^Q*~RL^56i{tEfIeNVC9b^HFn5rW%yb^75Pt=Z+>
zVT{E0VwTSFoXdY-qRW3L{vJxi(P39;s15>SzmnItwP0k;KJ0FGU+3fRVLpt&-wk}Y
z8hg&ey3jFRGi+VV^GA+q8!_@mPIERqe*`y$&O~Q-&4o=OzAIKx#e#~+RRZja6UewA
zQh>Uk-xuPigWn;^?--LF74f=YurJ8(m;iObU|*2mF@VNvsmDdv>rhVOwa~TbC!6ut
zlI+4%0vCm2jOy05fr7=gP9(4)|A8yBvkDvcUj`n6O7U_n-cZGC{W#uG-ORmbj3V)N
z>S-8%ae#wYQjc^D#oFCFT$Fo*Z;HO@i1`G`#R%aVJOt+lc(s}|%eBuzHEKo7M-HE}
zAs(pN%i}fNg1E%JB0dJMzqZ7>Gh1f0X|2DNXXks)Kj-f5;+T<F3}6{s#nBW$u2_QG
z<pwM6IBVo1LZW!eu?-<VJjl#AhOI_iJwje8uln&OC@e)>iv|0`s9T>-Q^Xla;^pOM
zFBq4fVWsbuB=br6vuLu~bEHRG2;F`54evf{<J}U12SVn@t$6?MLF>l>EWmnh4l)2B
zcAostid&%@?u^IVE06zyEG3zImlER#ii4L`mAb_^d&u`na~Fvh1nU+F7}E~b|FCYn
zREZY_GfyS^7i8Z4*TrSiZe&*tMyCEGT+j)=*V8(wFfPiy4k~fe&S|_R>%w4r;U@SZ
z7DB#8V7cwB4J8!syFT09-LM7aNZ%w@E2yOJpd@UZW0b5c_%rfw6@QqH6Zhg?dKKkx
zHGP*XujoR)HdoAQ93e%|i(}2yTxOxJJZ7f?T5#@}h74!1&8=<M2O;hId{dW4H@YWM
z%ia1BgjDQeq4gj2zfsXVJ$q6gBs5zqul>O;o0_?oOV71JJKT07VnV~$QhJ=gK*DeI
zi&@3qF1VEsPh!~~FRt6nTSV3yZ#rS6I7Z5#uN!a7Wab-h$~l>Y$>>P0=RnngJAg6m
z)f5;6?X;oGNRR`J{`50=ngH=uI<a}6zlBN&YygN+Qefy{sSui@eiVepEH#Z<s-RwB
zex(Q-exUgcFszjNqbQ`%pBC2oNCun8T44dTMpE-$<$f%;k!hj9G>=RVkm;KcC|D_E
zN(%j2$+VD6?=zUnq-Lfc7))~wrYf19Ak#Y{n36)*-IU>xEHXXEU@DWEnf}#anq@G}
zAk$O(foVbnQ&Q;9kf|?8zZa4CO&>=K4)9CubU{lif|fFfX)|B26RFhxI20pZ?{6+t
zY7gK^EC<wz8`iBGyw@9|q#2@c;?^*nqCCR7GAx-&BZZzIMKOtBgRO)_(3l1y5UJFD
z9TMSNAp*MjTEWID!u|oW-jWRwk|@G2DZ;U*ts;;@uVV}IC7DFv&-a8x=x>NXq*A*+
zB*Mo+1k9Lf?J^^Cg}i8pkZ6d&kwwF=D8d5~B9KCl?IA*<Sp+ZBKnp}FwaY>xye~w+
z+wZmFGDE8tREoeUpX#kv<CVHyAckKr#o4;x`KF=sMo;nMhXZ`4A*#Cx2kGW?u!d}g
z)LpQOhS`+5>33GekwSM*h*Vq!1f-8yW$27BHW@+!KZhsW@tLhsWSNUuYt67j<}VyK
zTCvXjpJ4uui2O;Rf9Q(LzuL%utdT!qrS`E<{!bYB=Q4kpi#h-MjQrcdonB-!^Z%VF
z|3Yi|Nuj%$zyA>|TCLS<C?J?J!ded#45n=&=2`K^LBk7VS{A{S6#5ZWFvS|yT787U
zR3<e``b&dpqrsHxAq{^c)8P?JNuj?inI@Cz=OlonjIc_20!x8Z|56e2ta!_%;Y~7K
z^^~>vq|kFDQ_L~f>T?aIGO1b8&kUyZ2Gd0d8{Q|=84*lLp?6&m87z%)De6%EVhIfE
zENTQTfk>tH8VnYsOS%qET6P}GMK;!QiCer-u7x57V-ti8YbnC61=ezrLa(9-{zpn#
zu2(4|n98K)a^)N4su4`5*sB{hk>jEWj*QlSDLGb<<E;iqna<4dD!~!&M)F#P(uQxz
z&ShbTo89#<#z!isjS33->^vrbf{4T&GiL-9^b(%5?Az%Oz~7whZcuVTu#nm2HSA-y
zoBU?Ic<hCGA+z;AQq3Yf%RImcDa?#E8jNzuNNqSuMp^`;TzxnhVXTCkrpQY*cp;IQ
z*INd!Eb_`|7|2n;wGq6s^bfKr)cdSY@;X5R$c02^UcWbZrIA-kLlSxIn{O>(n*ONd
zB`O-cmKwZ}$js{jgI6+nB{f_`UjH4zD_Or%@}j9zuBiquBr@~L2QS|p$N*Qpz@0Ym
z=g{?eJ}DS~u1P;tB$VSuYubsF1DCOKS(*$uy3uzdTTE}}uIAjrytZVy&fzzRT!ujw
zZ}#BD5}g!71_9nHVjc`Cwdufw&jHDipfMl}oYRBa-UqePgL-%$)UiFN2lhdIh-$E$
z>-q_pXmaa4s5|?h{;CJ{jy|YA>OuW^AJk1fsMqyD{dN!P_xhmzTMz0beNg|g2ldN+
zQ2TpO|Gp3EnjX~i`k=1tLH%$a)YE!U&+LObzXx?uAJpz1)KmJPzOD!L^?gu}>_N>7
zj-pxYMLno5?}Iw22X%5E)Tj2K=3#WRB=k;b^9%wf;g?MJsg!ooUExv|B+<*Bz%=a^
zU`!Ld_x4KyF$0`Pg4cjC5p<dfE+xS>D}nuhiNh%*I0G0HhXrN={c#Y?wi0xj30g?7
z%}Oxopec)2NRS1LDT}ww1a%~M*-F446{H?SF{hGX6fh<Z|7|81M}lXp1Yep7E+j#V
zmEg3)rYvGeaNJ7ZFcY-bf#6rbm}ICj6Radbv6bKpGr^xpu*OPo>JgI+kC5P~m0*mS
z;C2#R3XDmHr_BVTNl;@YSYakOn*@uk1Ut<HM{7awiIpI+%OvDx5}dRWj5ZVeiv+2_
zn1q~WCU}knQ>+AY4FdL^dKC%EiNWfG(hoh6*32<uHQI~0*Y@0D?xH-uHa9WR2^;Gy
z7p<Ixp_9A>=1mMCmixIn@Xcc)gN~$-obq;Ec4qp1FCvj;O+=PT?V}O>sK;_)syW;6
z-89NmE~bJxvtQJI59Wbi1GDm82qNWrIwaTiQZD9R(IfYmh}>}(VuIDTm?Q|yMVz6#
z!XB*NfhcDEDYbX^!D>>NRc;Sf_x50w(FdzvhFP_9KxZDq)zYzAr8=b#RtGsHm&Vc7
zgVks7tX5WAKpw?N0&Bvo8hfyMvIncb_rdD9Fsr3KSY6kH)$Bf4-5qAtW5!f_kQHGW
zq~-L%DkIFQriWCo_F#2hAFNIZvWlFt)E4#NvmX=dXgc4DI8l8#U|}_bvwHAZ+XtUF
z!hEu<d^WPA78U)m4?cBaKGtbUt=5Cj%s%+!hxsI%<?&=&Wf|WGk95gH{JL$&D1GB2
z@EP?#2$yA@;QVb4keu?su=Go7q~EG%Gcq--zV6d9d5j+SX^HN4Cv%9L8?|My?!M8N
zF-Q36ySea;u)qj!HNqQ>@Op$RujBOYdln_GBM4ps?tzoLL~1&{ELOcqbftC>Sy(DD
z6`evfqZ%qZjKJqu-j*3wF~c)M8NNd>5uo8Y(t2e0CnU9I_$mBuv<z3$C!^#gp$y}h
z!zG~%vk6`p&M>)0hL`rpkljqQ4DT^AJor<vgwM=I4w<10KP5OSoFSJ)ttwrPq}CE%
z+DC?GB5vjAP#;RCCi~DoJuI{sxQCKl!^%B{Sr<|g0Zt(J+pr|I9+Ld3ha~^&91OZX
zdLo#^B8U(u4`uiz!F$3PcJ^>}4M=J&`rT+O(R4kD4G9^}V=9Dv`GsM%5mp#sDZ-Vb
zsDS~L;9t9gpi1yffJjC2uJ0kx<z!(oZ{Fe=O`xspKOlg(3=!ep<5Hj>5K41@{2}Cc
z>z*LkLU0m5AwcA6RqYa{x8%OCkK7+Na(~ds^*$q<ZiGcfn2#_tu2P<wKs1IwNid|q
z3F`>=3L`F=aqZT)GmN+t#;vx-9X=*`wlZ$9HExR$*T}f}*0=^EZW-e$tZ^?Jaf=x@
z#Txgx5x0nO8P>RajJSo2<FAIB%6@|pSHn1)HEyI4SIW5Wth0|{MqDA|TCH)%j>@v<
zG46G1+%_XlW!yq*+$tk3jd31p+$%<02IKOqaZeg?S&SQFjhku2<uWeW8aK&^YiC@%
zHSTI7ZZqR_>+J7rBd(rtA6ny18dHUNj9ZGh@QD!9dDOYP8Sm0Q)IPysbez%>FUPyI
zpV(DaAWCd%B9s?FkF3VVosT#9@dD`A1%=yRB`@5?J<UJrwg=^fyF7b5RNSYGtq<d#
zW>@A~rS4|I6q}m{bJmT#2>P=JtjtMaW&Rgrk2^oaCD4h2d7_p1B?faUnUb?|<yFwt
z5zI+pWj@}-e2`#1$jbbv7%aSJ&oY?P0_1(rzldN?3b((+q?Q0Ef|5sGGws4lHFX~d
z=C9k+;5k+>W2{wkuRR%3sl^A<aZrD+9(O-qu=o*|?HDic&l!>a6%OG0<|L{y-=rU2
zk%iRyKRnc9|Im-r%geQ9f0?5l!ikQ{K*EoBqaOQr0C}wy-$!NGiJpgHyAie-VWSZ)
zHNtu$Txf(fMp$8lg$TuQU%se+cttWmeJBLuIyp`^XO()Z2jK+s7T_1Af90)C0MN*^
zI1lTsjtBHNKt5hXu&{2OI7q83ctQ|J(eZ8rPaLPQEb{J+t5^#>p@0MIF8}q|9z(Q4
z`&G8~K)gdMNXXXi#_fAJzbs_%R;OSGG6o!1G=c@~ow}#{rR;hOgoUg_YvsG`o?nLB
zkFRz<Nt9MiKJdO6@6oiXA3&d2+sHjJo97}E{rp-%zfM`F$wyw+;_XVS+KCi{f+_65
z{ZU+?wVDZ31t+zQN0A_@>N_Mz2qw75INxKU@0nsdO9q98O2!?8EFSxzCFq?OGX`g`
zaXOzd$@)p&KZldN;?Aa~Sr}B}r5mHtd<mHwl=-{~l?V4nSKwC`3jlOCkIqX_<~s#`
zwR3dgAZ7j~9IF{hm0JuKRT>zWy1tmUFn2Y*k9o&{cYfe?8XiCykF0U58p8n#mv4X;
z7;?FbTrj%6tqP&(c4MXXZ>WQ1kc_(x;)df|<b-hrgTOPaa0@w)(?lhIvIqI^`y!v)
zgM5A;<Q)&=6j1m!<!?>hAP+k&L~#Qj0%@U2GCKSkl!jFIkSY#H3)PJgR8CS|6G`>B
zh3W_WJW?407V7pSs|~e)#{=O<J^Z*8kAW$u?x6iz`j-sMGoth_$((|~OD7?e{mZ8N
ztX%*ptX{ev(ghu9ul-9usD$>BXfRJ8bMZq1RT0cdVP!tf#Jt!3<qM+=uo=v|keTdX
zk|LOs!pi)6Py{9KwSRflVBRT8fKv!$E`D#|-3qH(NMY??G!ye)`<I6W^VjX|X!Pd(
zr4b23{mWm<e@_3BGqbP$<!fZX59=TgPE7j$8a(Mkk&wP7BiJFt5fSt)vP1Y1g#VZN
zgNs-t|0n$cg{An^75f7gL;n|Kg{V>b1CogTfDf`iDD1005cIM?kSW+76!y^{2rB8_
z!wJ%&_Xh@j&lI9RND=lcOGdAs!Tx|Xr+1#qGNX<{{lSkIV}<&IH_HCg{(wrA{lS?t
zXbw;_J^3P<!rayL7E$_xtC4l6KL~UA80Zj}N#p`YH35cfb?`m<gHgx{{Q-lY(;t}0
ze@1^`CZEgJVCijo^am#Li2fkm@Xl$Flb*f?#SNSj>BCDSsBR<G<VdQQBB(}@>d8o|
z&n;9x=%<saiAdQA{K#sPk2}JTo5PPA@EE;6Xoc^>rM>bJa{Ppi%b0Vp4wo68Zz(;W
z_~n_Vd+BQ)!I=X&;gA$}m`2u5LA;<_jqDIq+yh&8CNxLpiw)-WWG*^{sS(U2g_-$C
z6Z6OpK`{S8FvY5(!F(Z^i{G9Z5W!qhn3;bDilF3?9fDxK++aS>V9wrNb_lQCV^xc!
zFf*?<F^}vJ1aqHY`hvY0?a}BEzTs`2ODK2oVNN){(Vw2qd<Fz(8<%soF>RsC_o#Wk
zac!v}>Sexhl6t2n;w8DpBtQYRr4u-bCDDjZ;1Faoy>EstL{1RefXDmva(?g*0L~Bg
zV1B^&UNL2OPR1LTg+<N}&P2(0w}ZIx+K;zHIX}?uz#FSG;+<Lv4(>nFV}6hbj$-n{
zl2BEl{@`yY$G_DdsQ-3<p#Ho4f&Jg^5A0Rj`SZK1BV{nI_aeYE%mh>d)drA9V4eY#
za)oXY0IIN>G4&#*)?UMU04%17c#mMA0XG`)^#<H#z)Jy>S-W5+4+k>6Zp615^j!wL
z*?<#R)8MZgaFPLc8T?XMRXAkC^RMKWX2d5NaFzil8*r`xrx|de0cQcGd=-GnFPE_F
z9uLdu(#NL&gI6l?GR0o~=@L$cLe`Aw(g(`?8Cbe?bgyxaF2!^yR{)$NVZwB!09u@*
zt1)5X?@b1+Om>h5!A)S;K|a7{GIw<=nys~iWag(a^D{vn>L6*WlFL8MT++z}4Lg;s
z(9$RM=pa`kCzPDQ&*>n|<UgZ>G?UL|rCEgO(LtKXBRa?nj0saC{TWJDfZ_%QMNXLV
zBB(M+H7=6ssR*heq<SclYHbA7o~fXEgGkvcZkQH&Tor!&SNQRDJQ^Kj4T^4lKOFDp
zpu37PUuqXU8IA`dl-=XY$lAOfM+)n#qXI<;b~2IOqgam%&X?K@=B;Edy2tDY=A^JP
zr+$LWBfCezJUCx!G?*_VbJ0EO#a78lVP*afvJWzk>>dU4;CyMR!F(~9i}{i-f;lOy
z^QC+f^T_T|tj9fU%$K-BXr3=EM8eQ~sq(I$GhaHlh((RI9`^z=;3wQwa=ncHN&l`8
zd70-+pAhl?+I;B{wf289U!t%SU+$(F^Q9+HYIFa1v1vW5cE0FvxPHb5&X?vMWmh5k
zuh0@27Q||(8sGd{L2sQeT`LmoKpvqLv<vY{mJ|H2ekS-v&VJN?yMI)x#C%B?CKfSY
zV$JF7IIQ!f2B1UprMU(F>3oSwm5zKTcnLMb)9Iy9FHhV;Fl%=ZT0iR#@=$jW=JJx6
z%Z20ur+GFE+3JCNbO#5K6S@NiKc_n|lmCqFz)b!#x&sq=M0XIJFLgpr&X;mg-01VA
zAl0R$iaK8kQpJ-h>U>F3{h)8V6I7=8QUe>U>>1t)KQ0bGzKBQ5d}$eM*gXE-j}Z)a
zC%X45x8Z`5!OqzCfO3wn3Bip5o)>~w3wU7&ZWZvN5d5Kl>qBsxfES10PX)X*1aB7b
zvJm{WfRU>qPrHDTivfQpU~n~HUBKXCz`F&ER0iBBU{D$GK>-76z+D2yw*j9396#Ow
z<8d!_qpDsz6|O|7`xpz6z8KzLf~i6+-G?pC`P@M*2H#!c`vkt@pVMM+ZCAWH_ZSz0
ztG;dyyC>`Y2BqTnxW&t|xbq3`N#_*5aso+k+ZhJGL*dxnzG_@rSB0BQZpI++3*4#3
z-;6oVJBNE*TkJVkTtcn?Myb1w&3G=11#^;ca@^|62}+*ocfKfw*M+#)M#Y63;3mhI
z=iPxM!C26Fc#ATAFMO|~`-@CSq}08I!mGZ}9k@T;#+N<Yaam12)gRxeUMmK_b+3>L
zsc|P3UYk(;Zo3mq@RJSt*COVTg!IG9IX+f>c;!I<+tpC0yfHmIvEr+!wI4tme3&;2
z=~jxK>ov$Iej<t&7Na;RH~PnX=W6<)Urr2mwQvO&<?*)T$D_XE&qg)k=SlIi=pGLi
z!T3|9e4c=3{A`qXP8K(u88b+}q`E#I9WP|ZrF<_65pff;V#L#*y@q&K_ebj7=H6oG
zTVnC+De;vl{MC$<0H-6o6Fyu*f#<JcrFRE-_|T>8UdW%PFy}VL;HPa>=dNLfrr5pu
zdBNOd)DYGGE_D=~4|nGoqRQxDkhu_gG!<qmC=A9dlh&csJp^Zf8~KHBIocRIIzjj7
z>=dPrb2oP}Zb9cTAwl)!CFZ2=bv1SOQ<r5f*pb2iE@cTW-0BM7YQL8KSF(N>xpoXj
zOx&zfU31q^#vJW3yWCs*^0&y4cOi&-?f)UPs+HoF!$G*7<y+i^H}q7Ob~8*R7nyoa
zMLQS?i#jzieG~7$^Wh$YF*o5(kYR|ez~AhfKtafhoAc%#OM=u@UxqHlceB97PS~To
zRiU3dX0M!t`#iA34UmP`p1V`^pe>&!%+x5cl=u=#tfi=CDWybMzwn%iw&l60)8%{n
zOj#DsO<hDotWh@+n?jBtp@@kPkyKS{Ln6L<D-4jcK3d}o0B|uwskI|!3ryyfIQQxg
zlK^J<7Fr^^f?(UcNiQ;Do=42ev9UPjT8^(3H^b7FKZhs15Ckh9iVH^E5{kGU5&C}E
zmc9pHfm(UMhPmXB*NaT@Klobl0+W1zCw(ZAaDUabKZE;Quo&lp*WyK-CyWN36}rpG
z+8F3Avc$C4tu^WW2B0~>bP@zwe`h4#0*cGQ6<#W?lHiS=7ei+`7&EX{xo}#b9U*@}
zC>gg&YTI1i?@x+r!{vTy75xN$&WXPV;C}?S(3jb9*Wzus9KSR^M{CN_wz|AKPp0D+
zF%~+s7KhgA8+XEYrTAg0CVBt5vgAhGBaT~!nw&AXyn#0;%4?lmxJJEdJw^lnPkY}2
zA60ShfARn$A{!}Md@qQasDx}1AVI**W5cd&AU6q!6?<5cP1w50#_Vnk)oOTXO=Fby
zT5DgowEx!H*EP58?L}%a2nf}xXsu#>rS;K`K}m~+_&opbcg~#MbM}GKzn^>W=l|(I
zcFz3fH^2GKZ)Se;n>ll4cy>BYO%>;BfqY%xzAT(qiGvx&aje_pqwBqW&$|oXi$L-5
zgoencaX7<v9=xy(w}agm`Hg}T?6kdoIA{Uq_j^5s@6`=>w}HXc+j+YCZ$Lt~Kgroi
zNVop|zP+~lzVLq5Q5-c@d9;F5qDt0>9&u)*!P~A6qd4!7!$&UmRXFXGCc4>QwEZ~4
zJ>G~W#OAF)<3R!BQ9TzjnD#Uf(dZ_`CR*{2HtB6}4mww2tV=zgtN@3AVp$fN*|s?g
zMpRw@br{{2<a=>?#m?~w?(K#i!I^$IVRhKH=_ypJdz-FYUQxvB_qzT$UR<43ID*64
zCIAsfQ{jNNN4yn(wr%`1#yxMvpWJ<qdTdoMxN+?JS9ZE94&cPmyhq$N*8vCap?n^f
zxAP9KuLD=VwE;ud!l77$=q@{{-0a@|0xy*bj!&CjSAm02pRjEngVMsa=DZHnTlkjg
z%(lKU_fg!$@(1_MH*t0))&OH1ez<gUDf)C&+#G?k6mb(hPR}0^=c6P>sFR6;p8m?n
zJuE78XY_1@M!%pxila2*X>Xw7z|a+)L6n7jwB3A)u1(OQjVwyoH*kxUF^<qyeY*jQ
zZD2J<wG9tb-z3LBjd-{AXolEdSz<Y=**4w^ep|OB)45H@q}7mSyz0<nji^s~26mXo
ztL4_{)4+{sykaqC6GJCVZu{QS$E&3Sxv$a&@nkK@8~GupP2va=oIQG)uH5^c;5(<j
zM`H)sa&Zg_>jEPYMd2L~rU)lD`2rg|1b`je2~>&V*|1j^S`x~?rJ&&0Gsx!akWF3T
zTaM?qc(L6n>&3`a&p_mCx!Mb@-byZ{e9rI;Ja8_ldj@`aHhsZEtU4Kmf+7li9mOtC
zK46QC2r3j7QRwR^AFyX`*u~X-4~{av8Lv;oEI-eEf8xx+KjF+lC|Ysm-~=A;mycse
z@&rpfPtamb;R1QIJ>8m;&%&CL&#<O&sX^SD(qcE%%XU{caGx$hoF^NPKNMSyc+u`e
zRV3SDi!HHwyW`Y6p=6R%Ox9E2PB%loMs8fjd<$Elk(&{P_V*ie`+LF~jrO<4(*6#`
zent#a`+GCF{o&N5)fk#_+9CuoA@(Bhy?v$LzUyI`aH9*2WHeN7)|k5wcj0c@U0@p7
z^>b|BS%!f(f`-oVRGi1$)L4(s?ZMf;VkE5tCAT=yW~sbI6UHpAe0;nc_tL){d*}PG
zABZwKDV%rtKyEK`f|{L(lLwR0*`*#yM)&m9;@IDzh{MxA-_wuxPY}4b|0Ep4jP~50
z`)^-Ay6`)deYszkkA2g=j`In<6}!4#L2>d}9Gq5boKM@I^&lE!;yjz2njK=kv_s67
zc8K}X4$hbM;*uz>Ir=hAMp||MB}lT(iJvW)jMc{Hj*~cdtQ|C_YLhVO%dg!Q4|3*M
z8=p8%;>58wK5v}Fd1LMVI5MY=wIHC+`X+JKSR0=-P8wq{K4+Z7Ib$tg5*d@l{963T
zAZLsdWhi$N3a|#p7P_(Q<VA1BIX~!ghih#46ZT)sdzkWTviITFoBcZio3djFtj<1I
zX)Ac@u<gqIwt`W3BZqCP|A32_u6zW6U59Z|5)Pf3u|J=8DLpA@Jk8YJY#i$d8n5E_
zhV1>IfrM8cvK9Od9?r&E{T4uf<+H7Tl%4=37!nI+jNlEGI5HRKW8t9M1v4MdvaNm$
zhZ2rtRobq=VOI-gJeH5!o3cS9za{$seuLRB*a|@4%HP=v-qy)qxd&OGFjsy6x*syp
z6N1Cv0=*_1%L!m=H#lI{`)mb=;T=Va$HDrQ&mwRDC0zAKrhC-H56<IVF!LqwgNM#X
z$qz54n(-X?S)Dx&av)D~e}wlk$URDOWNbgl;qb2oGafOqA%$%#c$z4-f*0XEhh)Ep
z|0n!G{;Li$(H~4aP~0;g1`iK`hkbD#c7g|ENnnkr+)JQ*&GY-?Ywx9K@3`HtFWb!n
z&iksMBx=D$6^?kT#c3H;*v(jLhYy8Pi|IyH0sLb4=>4@g=%)&2e$+bQ<B*72H~cm5
z>*241zY2aQ{59^rjntWVT!%;YW;}M{ajX7_^A9%OD;{s@rPsHSMzy>D7LM=;(S}Ay
z(CFKEKo1EreH(}M&{leV8~Mfzn1Wc}#xXr4i1lq8*F%C>-$n((6CIqbc*1`ba6zRP
z82{`a|DX4d#QYFL8iuL)m78uH$6~<SJu|s6bGDabmk#U^TQ$!VkPAn1faAf~8mk+)
z<SY#DUfg}fyN+*u^G3fHYYDjUI|u6lSQ)Cc?fAk6vBQ^(S3Ebf*%I<$>`xN%MG4Vj
zPY6i#M#kejAU)LW+4L|bUanlX_Sj8>xIq}ZqP_i>%)-(4oTtD-j}+7Lj+}wt9NW!2
znH&Tb$2REc7vmZz5x!Or7a3un9zJJu9VU$*fa<6XQKOTPU-oD&esh>jUBz3GC9oPj
zSubEy_2`*A9oMv9;Y&G{x4Lkp>mCngx!S|9FAsYvUWiP!?YK0iAOr6}I14k1*yuB!
zzC)gV+otv7O3z6Y?Nf*az@6iX=?1Pidf0qb#K?UfSU0$#_b=bp{g<WxEd12-{x%)N
zd`=uP50!<fx_2N1{e`{JV$IFA`34B4uKz+0=EO0S7K;;Nz2<s5H{4)Zz*Bh0ZCmmO
zZVb5q8H4LnJsU7qYB$|fhzF$Z#YKfT<la2A{&d{4hpSzmj}DE_u*w5xcQZpsLrlP<
zUmwL}>aac;uNaLi@bq7%@9zmC+u9hXX@z@`?f%%$-6GFv?hPM^JWKJwJaMlqvqXJ!
zjRi}WcJBZzx9zJ=;FT|2lEBe^bFn_b)s{`W3OI+fnsEv1)VP0OMI~+^_ypy31fTN#
zcgt&|p65R%uk)erk6&IG0^?V>>|r+(Ckg%rPnfi|%*q?RmY+>=rak9=HXQ87&cnoR
zGNv$-N6$0j3e0iS5I1e~toX-4H!WxTyTyg~S*`?dDHlreV<Cbsv-Ce4?2d(`qcO~@
zuxNQb2I^{YpBS$?xfweIF}tcptLcdez<L5;l0Q?;B|Xdp-iz))W0`KU`4f#Lb-e`T
ze@qv_gT*Lw?7mub;l4eb^UTI_S|v^$%jXZ;rAw(B@S}-i6`&3)%;NA&&t6vn578XW
zWgNJcXEK(fGVn5%EB{yx>GTXNgT1c}W<{6bQk;l~Q;mM(JDikmb3J5&rg!!>6zwLX
z@7QGD;ppCLtJ||Pb}lYmwf$;b%)0*=J5mG2wc3KZzC)vzc=}yJWcTw}NO#2(Q9!Uf
zW%}MUdMTGazgxw*6IR<S@?x!ur9Iv(Qf97Nz1j;b2(!%ydfm@RRPz^E&3Um~Q9C0&
zC{H}bcE|OsInf9{$5`yug9faz?5Egta5}XZ-;ccK9;nXq;+Eje6GTH_jnx<^9xga~
z``co5{ZYH8zXN*r6?Y--2K}?H_m0GhAQuP|*L}Q4ao+T4JWzyv2W+?5KnZ&kZao1$
z8V=NXUEyDF-0MMBBo?cobl-@*3FbiHOnj)Lt4`Y;6|CKa`kdr=6IA-nBVU*lg3EPw
zXiMAXiv`_7u{S`1Ti3uNHKgEim)Evrn944nXLwRRnYhdX6>|9w*cRFm_a%<}gn>=h
z7by7Jy8jxG{Q~}=^ZVyB2h3wVg}Al@I=Dzu_H2L|jq;y{w;_0@2yox!4c8)g^H9&{
z(1q~5(&!aN0j5WS;FxXX<VyC4hw+t?$N^0nH*!KqPYVgf9!C=b=g?WYe9=y&d~cW!
z!lhUf;D=CGAVc&k3=SCNA=acO2Ww;Wb^L&v(U)))5k{gR1U<Hzx4gER*ytl@(E?_3
zy@^#uF!m;>jsC*A9u9lw2uvOiF4tny{X$*;1;U=Oj%Qpk7d5dGjJ)VB{Da$8`+Ka|
z#&YzcoNIKsGZMoUY=roI3n<lO&p_RD$I$2`=8xz6L0k)eBJzxWZ}U)H-@(WkR0tt#
zZ+x}x($Mrs9R<1jG754xFL{-9HBTQefA22bCF*J;L=9P<hTWSft@Vb9j;AnYP6SDB
z-y34d$ZS}cCP)eU591;AyBiB+T-d-G`p(!gwp~=`z)Cx+h6C+wc=rHZ<>`Bx@AEpm
z1MN~*dFm=&i_{nukM}F&n*)5D;U}GfVM9|{TY%J5yP&Crn!@d{f5xp)F)tFiH5WA1
zQEDp8y3Bfs5TV`Ux(k}>Xzb{s|0YfK(&A%jsv*WwQze?imdce}J9<VRKSC@VtQKhf
zbND*ul<IujDLEcoZai?kJ@!|W!o9az3>msngT<WB;^UmXvC~Atmt#|a5(^NoZ&z$5
znhXcG3K6+C_8<R>^ZOk`rpkuB_fg@#eOCNL`>5~xgBbnwfo`H_Z*4xhDQ<L+Tx+}a
zLM(#fnq}jnHFw{_S+OsRCcGE7L@ghfbzfayYd-J%#eh6*R_yPWVgT-I&GTrF^~KEg
zG|c%M7if=G@6EjrB$d8BJKsDvYtNG(0P*JM`(ih}JaQY_BnH_y17IYNjEsa4(f9Jv
zmp5r2{suzVw)4lf{MoEO@+zbI#}NVs%$_H`Gk)th;+gTJZS&6nz-w^aU4~S)<%hCf
z<Qs!cuVIwigR4H?%)#p+ZBOJxw|-MuWV)?j!u_W!N;P(7-1oNleRT)3=k*$0t1ln8
z<Y#XqA6$3wp8zWE!pFQ>AAqzN_H4_a$a)lOb8Z~Gj?=+Gssp5~`i9m6{>Jb(jKBTh
zV+dC5SoHNeYS_o!6@RmB=8UPPZ{+Bco8I|Qmh#D}0q;fGeUISMH7K)O+}(3AZa9ma
ziM!1rCsCR#I)s@1ann1wU32k*_rA)36<Ka^%L?()UiNZ8QTA5yrd5dU+P5)RdLPSa
zW1Aj&|K_J|8t&VJ+eM;Ny?E&&7aOnE&m(_u9QOH+5Am~o4wBw~QXbw)qH1yfE!u1R
zJne1k7W))aliSv74Wqy4iV5E=<C-s;OwdiSQ*jn2UL-|mRN^MO1DO2b#WOZHV1PUt
zyAG{u^bB^C+3Xtt=j}M<#7Yd?M`KqZPUjX@pWd7DfMo0+X@m-l{nu1M-S*WXT*+vy
zpG1Ff<9lid1+UQDeR!Wdd6uUFm;6K~cwq5j^?voN{O!|)%wFdCjIWj>Tf9J>oJ3C_
zN@HIIS1=ac{T|q6dG-mmn;oF-&f2HnO5?$s)I86nC|=|oal;%ehsju_!mbt%F8;_H
z*^iurJyeG&E3S><`|_RRxHNH(exkUoX2hMf$M%&WHsqZni)~-snYS}>%{NBL7|ixx
zgNhQl#02O4=y$Ue#tn)OkWqZOiEIQ@b$uUt3*VsKwNpcOXs^TXyR!~NCer5b<Jf>r
zA2b)<iv0^Jf7!qdzs`-nOkWP6c>2zoRmbTnmg0Zn>BCpGQ1wE5)Mbm12G#g5JXkfG
zABR~&aCZeX>rMn>FCb>*v*>ao!(_96!Zmevv{R6A_n-V7KKDB@2zhf6p71ppC$@YP
zbe6!}g&;dELrN3hhZhXuoJ?+&SQC2>86fE_6Ir{3EYMlc0=*~p0662fuBbiJq{3nY
zTTaK?3`CrJN6&Q9jO`A5{f&nEOL5v(>^jJ4q=gU+7X1@$^(Kn=0G~}o{H2ULMiGy7
z@J)Jl^i#lk!DH;;d~jiv&~q$u|AcH4S064lb7hs#H&7VX;a8B%n2(G50J$HZZ;Jb#
z6<1yATRXhKf}(WcWC7bLDCwnQoW!TXIqm^mD}*;G;=PS{xueX(W(W>SvD5Tas$Hal
z>8}r3CwTCd9mzbnJP&`c`}j`iw{taZ?}YCxvvUzt;Y|oz;p?{@=x<y2;p!H^c2rS`
zolX%n@2MM@h21oF>aSzC?(%x<CS0m2Om}=SK~g4D#dh;Tym#?dyk^_{n235{R<H1R
zs;?mX5V$8jf}yBLZw~0;wR_}ZYB@g46k6^9ak?;Z+3jO}NH{tZO=IMT7>V2iiRgWZ
z7K$(C7sRn`7Z5h9TbXbTYXW;V7_jo{2Ct0iO^I2}7-X@PF|Y>@zDus-2MW6>pm7M4
zO1D!MW8KNb6`x92`kL+Ll{6bFUcq_eA|qVs$aM8P1eyPkWL8NsGeCwy;EYm8<6I#P
zajNj0;_HvtjYu+jE=6&g5QU6m#p-gnTF^A)z(a?xVr@bTtg%Yqd*F;<Fl0VYmsdm@
z#pg`r&3E+TyUDk6$8r|`oie@nZpn9Ga=iFJGgG@6e2L;8LGjtJ3(ecnC|2x`;ov3c
zepk}v5XaVW3h1Vl2i|JlIcppwV;xAM%j5H0AOk@WH-<P}9(96dLLSp&OW9ITa3Gpl
zeG9A-42I0$2_|`n%`CC!$C+NtTA)X<-3F_&I2nEowwLUkumE}igzym!J6?SWQT6U-
zCHoEHY}ieWAm3kJ4XY`!&|uqay*nHS8~_!1l-|e>@MFdP86fvVf9fkzR_wb|u(uuq
zyFUf{Uyp%(V+!^y$H2Ze1$+H5u(<`+TFT%tu&+qLR*!*wc?x#JF|Ze<V7rci%}O;b
z?2HHn^2=G=gcrNrfr0K>KuicPyvOb&5!#9H2p}ee?c5G$Ab5#zFCZp_pMPLNxR?lA
z0Wl$*V$Na`5i~0glgtPs#XuOhB0OnE*hhp1txWDUBm9sE|7k_|j=2=K5MjVtim#dx
z{6yFU$jAx`AIcRYs+0Xmf1HcQkp(geT`Xld+WZI=xur|7i)ZR2{8WGB4fe`(dVKR2
zQiwft3}Wn!Bi~LCvuef7#VN$@ItH-~h#Q5~zH3Ar*hu!$jA)xVYqS=3?J<bf5gR0b
zK=lTHpB_#fCpQ0k3SF8|nZy`E+-N;1zLB74)wG-Ur%=4-7!<#((-}Dk|G^a)e)K6<
z5YoN>a~zFMnk^c|NWU)ckq@LFKmt=2_%-RE6CFTw?1Un;9*Wb1@C}KO4?p%gh>p<4
zH6c(}k-`uu>>)y%89}!phZ%V<SxRSJY@SORrbcN!NtVi8>UfK-%vV#)IY#13ne&@?
zinFHY%zekPB9U|QsJ?jz2^@E}y4mzgaWhNo=}R!uu20I7R!wX>Vp2xVW7**8i>a4l
zkQCPmiX39aFnZ#}x;VG86sFPnAI;-l=8?-hpi>S)VwkeV*E8ey(PR2c>(f|7J*Teu
zLTO)Mmb&BDY<gp0Y}vpiuh53U(xR9I(tgCcW*Q6QN1#?$aWGOzGR3j;SXG>84FBD!
zz$RzOv1<+`#I@NEo`ffUw*FG=o5<ur7#A{l8-kvGgBwgLr_I{&D4A(S#N`8Jf*H&i
zUXDG*z72~8I0=g;x&$|zSN6}}&Si=FxfT)o%waIHVK)mKpV3*25$juz0!(}hMy+o>
zEU!f*LcFQi?wBa&di#(?tp6gk-?uTfH6h>wUNOQ6gwGLSB_N}hLXo5ja<ed=%&mEO
zLLGrO!Cy6!jz(m9iv%K{we*D2)U+4IIlBoP|7C}Nx=?Q~4?xU`_^G;0_+K*eJBXy{
zzXo5vzivc3+KO@uV>ebr1XOyStn@92F)BUJRO$DCpsCWoM*16ela5@liMQ9}^_!0h
zi^V21K<o>Tja`V(u-vwbumO`E*!r98$_Zy+ap-T@0LGURySA)oggQUmU;5x5(a0P_
zv4a<Y72IL{;6Xk70>Ze`{y_!HPl$dD!3WpiJ=+f2!Lbh!?As#%SRm;vpm=s+2gT{}
z^)bwNW7EM0_qS`0MaLXiKSR(If0@t%BRPQXYTXj=pW;;*UU0mRT?=)EuX%QUFahf>
zH%2`JFya&aSu}z07&jz1kUbY!*#A)kyergVXG@qEt^2WWQj1NJAl41g@(4SE%$al}
zV4h98?e2kUI~E&n`YzbUZ2JW)xOoflRnmBl*soER*sl>?<>@cOyL~P8b3r4r6zTD@
zPKW)Q4x4X=eMyI%hy5p`xRZF-80e3lN(}7Z*t9!;^hEql7yDE2;v>;zAu_x#Ozhtj
zJKqPc%hQ(zupqE^$#k%fH#BMi)|ZzQ08mZ<rdzSsP+J}<3(v)PHHwurEFU=W>Zrie
ze?>8NySsP3i`6ge`Fy5(3S0Pu>#2s8IJxB$i@_T_eank0_wv=DQn`1Don-OSDpJTz
zoKKnc>!yWY&WfJn?l0Y&dzZk%<)9T0bzuv8Vk_r^Q_+}q%)-xhED$L7FU3>ryz|A2
znfOlE!?Vc2Vtt5s9RbSt9Mqk4J@cdb!|}$~>sJX2=ImtdD`f5#<gd%$9AC)zKQq2H
zZTxJ;ubhPV+Ial&^0SvBhs^nlb!PEC#4K;q(z9qEiS=FfYaEX0tr)d!<{MBint%WG
zd3ESfg>U17EbeyVaW#0^n9pf0ms>0LbT{a?ZezT|3M{|<dzxwS!f+!s16EEejz-u-
z@Lrbha;Y!Gp*LbEUwj@b@oH#8u7$etu}>gxw8t|LMUQY{)WU}q@8bOhzO+DIBjFib
ze6AReI6y4+H`cMvqhu>@^i00c!|sF<ncL&QVRt#-jraI$t{>tG^Y^`3M?8f`BUnwt
zfr{_Oegve^9H6;(zLxFDX~%dic3T-uP%iMtd}wjhH+@>=N|*taEAaJXWs`T{{QV_%
zWrcfSb_{-_x9~McWakIy`3Li2(HRf|7&^yTZx^p8kJNyMHthxU<}7!`Xp}E}^RBuY
zJC{XXsPni!KfDV)2b73`M`%$D<5t&;p!v@d%jRJI(<8?&)BWjA-OrQ${>zHyG+)yh
z3Uox&pf6HXR$N@H)CYanHa7c$ekB|XH2YiZS1FND#22(H05(T~2UsVNqW(z4AGUyl
zfsSkJkuG1eAK;L$E9~#EB8Q{x_DIOCbo-klp{@lB)H=1r-|CA7BdV{n)8Ek|lg(bN
zcC-ZA{T<;zsKdT!k-at0;kN*lqM|v$z}h+BCjvI-Ea?igun-KWV4)-2>hCIAr*ww`
zEp|pOSP+h`Z3%P-TKwSx9l+lzU)XQI=n{#NfD|S;>Ff$M`@`W-SNN)yt4+*U;0g3_
zpq-qo4@DzZd~1YKAj{MhFQ}kD$F55brM1QsUX&?j(W11Zls1%a#v)bihz5hllvkI(
zCE6T;sFMW72GAI7FK7;Rgd=vZF5QLQp|zh|Y`@?VdpH8A>Q%WgvL4mvYxmn33t6MF
z__|a)qoxHi8d;(JD#qFuNofbz&U{@9_1ZXAsJlAU-Wl}w)GoH;-_^<$l?}^0%a+(@
z*c&B}_CVMk?TD=F^7~rQ<`ldBi1r!kimVH@LCE&Si|oaw;6nSd7DFs#L1hJ5C9tH^
z-_;(C_zSuj$1>}phzDh<M66M)II~wP1qB6Bpv)^%BZU{Jp%!1ncUAG#g%=d})E1&8
z6&u;7ZN+QRG{KH89YgpZEi0=W<ID)j*%vQZAStEvflsb>j+^5;U!*J0Bilio<4f$N
zpDfeI&gSa$l3poPRr)r1EP|KV%fS6{i2MH~qpKB<%jGI9sq!q5Pr{p5W&q}uDP=Wk
z<BC#sN#%+qE=xcyT~%r{#7<wAFBtR(J<EkgUFPo!H+z;_I|aL#h^P*Av<BLuT|TOg
z0@<#X_@Zt8+JHYOx{KZv1a=%I)J7OJ-qBX51dE#2`J1otcdf;E(E?2$EK=Lh8D}r9
z*5Orz7lZ>5f1$x-+TPOX_hTr#hCRll^9gB0TB*XqR6pp&MsG^1%Ftn<u~NFj|4aQq
zS6UR`$Dk#o)U2z$g{{GmFS1zg5NudiE8)QB{h`(Zk}bSwnXZ2p*(H?4DpdSE9Eveq
z7ZF@gxM-2UGo;9>Q6-L5Xzl*4HmE!)tK-n}pQy~PR;pY}Jj+zKtFp$`puo29FIr@3
zI|Vb%@iPk*)SDbw;w{w8(V(lqlqz+kHz4LlTBs~iT+3>Vyxlbo>PlBbqi6XtrKYl}
zvgyitm(tj@yrFW5ORZhrP*>Tc)OhM#%ZQ;mOwT2z=Tg&inHZcc77Z<nIJnhf6mWn?
zhboE#VOS_$&$6XzQ$uC7OR4u({>utiQ<F=nTE1dgO=ZKCU?Toll6WjlJeDcdZddiv
z#uaraX``#I%Ii`+HHl|$WtGdTG`c`|S+z@DwxX^IwNTyQszkXBUv*bD)L<}n=>V1<
z@^LAZ4b^VXN|zp}^)9b$!f375FZV2KQo@l?G|1l;{sa{abAe`6*9XE#!=rAeCVuTF
zREmJ}T-|id<ach?<BgQd%Gk=w`9nj`Wxx$zk1ugl`?_Gu27Dc`Guopafo31fQ()JZ
z`UL9GL_L2U-%(70O~TgDNg8-#63?JtkBt^|TTO6U^uscoohPD|JXyXJYhRC6hGcy2
zuv0*`NPnyJ2c@6B{Ekxc8!w|t3j50>f0Di-XPK=+_RN2T_iiq#?C4mu$RWl{HP8`=
z6g4-){-up*Ru)2!GUo*g_*<hJ`i(H83J`UP9Y2N2?Ah&|0sOWB10M99f<I}z#TA#s
ztkX%Xo|6l&FtQ7p*ZI2a7r}-XR8u%eGE@au7ck=%|JrEV>~Ns1!xs!E!jTqO?n*U=
zv2`eqh_#>JGD)%9+x_i!UrS3DCIj~KgTWrcq-zef_yv|~nH5k<(BbhWrB74L5Va%J
z;ivp`LB%V{+z=NmNDJ20TDNYW7{xM;eQ6UdSkQ=RR#T|K-`ri>6>5i?#{|g`o+Tja
z=XmK=`P%{=7S33baDyN7#c)K`=K_{&EP;P0Pvn5f0S0+<-KS*B^_Z+Pb1uCzTT%Xn
z4SA}j4dc1@G))_yggy%2jZGN#PixwAo1!@34<f#Jx~3IiU4PA`nzjahFLoy14S#5<
zrtOD6=+(45EaUFKLere^x2)E*9{9aJP2(AtgRR)e#c-RZDHEVm)`33!EdfodhrcGM
zX}$2B?Z}V*wZMm;->qr(iO6TYrmdonMYmq~`@gAa+u#>}4=?=TJMRU5*#EHRKJW*>
z^FP5K{IOqw9{l~k20i%0zX3gTX8SJCgFg(v7ykZ-fe(M|x4?%#_$crh|2yEH2tFPM
zKK&=KAOe4IKgtDv_$ibJavS@-rj^5A^9N1qgufsD-SCIu4>BJ9KKRNrkT?8%_yu^w
z(Rl#;!taIO1AhzrEqwFw9LfjZ4u3!V&KDqGYzDE%AYb^MZvY?ue)xC8ABH~&e+>RU
z_{u2qfnN-N9R48ue9#$!KOOZud=UA<AA1w{@RhfK4?iFNID9+&e5^bc!!LmEgztpE
z2L3Aez3@BXZ-sv={6Y9z;Sa&z27f>N{qTq3$Ka2_SJ0l6Lnt5U7msOLF~VEn_Ye*z
zZkB<c_4x)Rt0zC}%*iL@ZOJ+zpYXHrck47wy9mDF(&H=2EFchnx(#)UHm6L<ubnde
za@$GQ<@G9;opI?!rL$&|265<eK40$@Ii8TyYj-L+dcKNMhB(qvPSvyv&cj=Znlmbu
z$|?C(Q>Ir<u~$tgsG3q-Ii-Bbl%dI+PPlF2z=Yd#@5s3&Tclf!WaHR&G!1Ptck`L)
z($!3<w<Htk?nJs_>{XqMHj+D-HXZWIwj?B<zr$Z_iKZQbpZoo@toc<=DXy8~+%)kv
zgF_&}F<ovJ>cEXQfwqymJ1Lz}zOAYGEkwFo8#QeQ<eR(gY-@gcx|%7S7B)qG5v0ps
zrD=a*I^~@7>9%HB5>6A`vYyQCLa#s8;UIL)Oy~;MP0bYhCa5dnm~J8dhM<EUCj{|~
zB)kcKJ!>`XAql5fhQA>GikmfU2KY=<*Od^XxvGJrD3{>x`}iww(X`#nyvo!@AbSda
zAoup1J0xSoe;R)Wpc9{iZ^;MvOQsBFnJbHYv7T;)?tBlr)`D-yy4NZ*l4ZJAk!Sri
zn)U?ceRl?VZ?UFjy0ef^ETn1Qho8Iq<E5L0bUy{3Yr>lLjzxZzQwlaE+8*iO0J`N-
zO`B*3{js$4A%;m7VKW`?F3<m}J{IJb&rH{T&_7O0Nq3O>Va&*5f83oR-B@aVvmm$p
z?`qoV7&~(7&r6?QZ)&>LNH_jNncoW;(hXYm#x#`gAnI%X&onJc`ANre{}lAf@6)sw
z;ODNJm5z^^DM|f?>0Sl?@GtOH1nA}}=ciAXrv4+4*S7!Ew1u)?CF#7HDQheu7M#u^
zpGY^fUDNI-B;F5D|Lg}7ZWFzL<$M}=t1u?NCi?`@Mfm|Q-l1uy!Z(~KKlmu$3A&I=
zu03rzQu-#-EkyZ;9@4Z~mi8Fe-D%6gbP=T62HW6x)3KfgcWc^T$$aWqk)rd7#{5`M
zJ$p3mo0<4b8Y_%+XF)E*uwBG8V8@YP6Ve^nhyIs7zZ6*-`Q3(egRqhO>C>ex-z+2!
zqW{Dm(X<erbJw8$&^9WkOs|Bx&beK-foX_42>Qj3YMQC~MZ3sFzb!}uKMQbs8h9Dt
z)6>AcfcKsRn+|!ge&{jZ7QpWw)-;$ZamRYO0q|k)eFb?;m2X-(k=`$nF7GkOH%?8@
z=XV4o!6~2HU~7)U)?A0@+y~<FsX?!?$Y(O-=tTNcNEiHYJR_gKLwYyh%OJSi+zZn6
zbLhn!OQ&VJT-dN{V9S1=<NvCpbYd(T&Nh!l%$DgEBHe-K@um^^=MG&c>NBn*M7my)
z&OGjMto|MNAA3>L8b}EIi?PmJ|A;SsN&kwAD0Gx(F3MMs27VUc_B8M^z^4N~hXi00
z&}05yz)Al)0+7Ga24S{|aXg6lt%yH^?K`f&pf9MulE!iJ^)2AnpQveKu?F-=mgD<s
z*l#^AYg#tw<mMHa<YU!^#HT!81)X(&0>A0|RmOI97V5!{cFSajBVT2JPfr8)0-m1+
z-U4`D8u$%>D{0_&0zQuNoxx0HxtZ@zC4Mq|x1^s8e*y4x{Jbsklgl+3a!tq2EWpW+
zx&M-%g@BI%o~%!r0FMElOg~8ac~<>U15j`>=tB;EJhNTgk?oo?{q`K!6#GD~Yf8aw
z6WmjZZ<*+xQoiYgx+%`!lybBe)U=o<V77ocL!M~IzXTmS>RC2<QWH+g`SnI6XPW?f
z4Im$C@%M|F`1uO{R^v~&xA4nZ+qY-n=W-N+f$uT!6$A?4|BwOhpg$k}PZ=O_s*T4O
z_#^Zz>*62-JbwFB`0q1tA~e+5@Fz3V3lMk;Lz8e^ANVw$hR?;14GIHaHY%?%@cL<r
zvL1f5U=%nYsO({&5&LbJPbUM<Aiz!PH!z>Y0DlkFB4UP-<CR7}6~H9{LCuZ8c?N+W
z$rvG!U(0~VN6RV9xe#cIF!YUwHK3nGjavlTuOnySw466L^!6$_ZwNrKgerWX7=H>C
zgq3e3!k=*0;FpCs%9wpJ@D2loAi-w#9s?U>-~$GD<|Jb@1{$(;(pl)`ii!YtA>F_L
zzd>Q9B8H#CGG;epCWChp$YFq+uj6Q6LV#$K+1Gw+j(JALbTZ~G1eUPybXv~0l$<l}
z#1kv+Hw66*e^q>-v&!VNLNWmJf%!kmZqWhu7<iF^t3Vx1Kv@T0X~hrQr*Pv45bY4r
zV00=!6=3|(IAU{A8S^QyMgkN0a}EOA;R|;FfeJjm&r;k4`7zBQ2Ac6h75gCrVf^sp
zb>Vh`B-@N&^!G9#7=29!1Uv7`fXGzK*@ae6$@D=KO*lS~&OKmNS%8w9iyUrc;9K~K
z$9xw7b`s&VoUd#Ep*ylV5oBp-q$}6s?_`5L?xQB-Ut$cuMmZ0@aNLy5G<PuOVFY?~
zI64Se0lUhbjA_IVbHAE_RrvWleBszV86z@%76IlXT)a=52(rTQfh1WL<x>befD#Iq
z8JDe1;C|VV5#)A@k~1_>MFiX7Fn_*-ze@bQEg*Pchq;xFZ42^Q1%DXd^C~tFp{l85
z(=J5dV)!w}bR$5ZKr2M1ER>*DiU7H;A}5#Qhdn{Ws0c8hEzD;<0yE%0M=G2rGUjf^
zJT7BqG3OIecE)UF%y|efO^7j{l`;1+hKrX3KF+|1j5(bNbHEs5E@L2w0R1oV=N1_=
z$iN{P^Bx0Ks!TKEO#J)+0RnGk;6WMl69&#lV`t0}1};N@Y0jeE@&g&u#=wIJkn8P?
zepkkX7&8qWk1-RVkd@^KklJ;O3CNgw#(ZDK+{74u+s-uKWXwA<=8KH^6nYr5_!$GV
z1W0Y1fx8f38Y+I}Nd!19h#nD>F%Ck`L$$Fs^4M4ABQO<y5d+_lG2IOOI|Apxzkz99
zkuf(i<|qQpXE$T!pr4T1EsUu~fH9NV-maH1w=?Dr8Pmj=y)x!5#yl@$zQvd~@PjJS
z8`2>Jb|N6$AgY|TE~<V8%FKYM?XSpysNft_5@ST24#<G0#t&sc)Z87A7tw?e^T7rK
zLIzsS&0-wdm^BL^%JxDu24#Z`lrZp;3@l>c(kva#%>X|N`Ye3met`f)r8F|;Hwa9_
zpKwp2Td}`T_bN7Y1a|vxFfbE8^AIasDlgAcI2-Us#{SHJ!ONQr7`*VtlJvaf7`*HT
zhHx7sFRPjTg9wn9%NckIjHTh_d_*z(7)foEfx`^EECU}hun=U3b|SihvJ!!~Tz-iF
z<uZ*iI}q3fUpRx8cJfbNX5#NE2L1$w$`KImf0tb3I77}7gDL{Fdd0}F9S@X~a1;xx
zej6q1LEwk@6OQL<Ux=q_#(WyGzgLGt5(rRWB8IlbYevk!1DE0v)P8|Lp%J6o78MBc
z+ouD}<X-$-nTUTt$tj~sgK|;6i>R#GGVmV^bmNCHKVslU{9KBFaCnB(a-QkM__~bG
z7x9OV4`inbe_5R6orgc+8t}`QD;WBW479<|;(!$&9Sz#M3>)AQ#0tj;%1yArVo@N%
zRpXZ`Yz;#z@W+8mI4x(t<gnfV@!`MlcO%=<3COO+h`9!VbB&lg5Foo>CqsAQheo?_
zTF#$0g1MKo*|y1pi>;Vs8K`IA+xTPbRSfLFANBDX29Dy7O4CKjP60`>x`Y89RZgwW
z7NPtLen`L%wXKjZ#&F<J*mWo@C#mg5fU5T=46uetjqlZzlTbRwe1mDuMSy9($AA+7
z3ag3bUyUEK)6RUtGUmrb`vn5=e0Cu~YU3pRECS3&*wbgCEi&eBOfv@org@uzdIXqH
z3w$MrALbKfKAU9B`$XF%12v5K0|HMl4?1ZFa$I8${V1N8lQZuD27Zh`2KF(q6Mt+S
zk1@d6#`S!l)B8awp33p}B0<;S?`wScr~9V_{waZfO5mRo_}?pm!I@6MT=5Luub2LQ
z>2Hzo0tr8c-cR<Y%7MjzpOqA@km1Ra(5+`V#S=#rx~BJ>!arU7V%VAO6Hm0D=<MQG
zAGi4!lkptD=s4QZ{ryAnx<$D=+sUA?GU$&@a?(?1I@673JL$#EoPU)BFk2E1b0gts
z8kil%P~8>l{9(8j?p{flgDIUc&T?F)8$Q)3d;>l<O%EG<kBL~A^U^o+|4@H@q!deh
zO@<9Vu$6?bmz2$)a>|tdJ>IyKe{yaQ%X)oA`mam>i1a6(t@C-B^e>S9Jn7d;f2H(W
zrQa?6Tcm%N^zV`WucZI5^q-Ob>(W0W{fXzu{H1?^^yf*xR{ATY-zxoX>E9y#yQF`Q
z^nWG&ho%3F^k0|$5$R7nSLQGM3#30!`nA$uDg9RIcT4{k>E9*&d!+v>=|3#}XQcnS
z^p8k?qFv@M{R^Z&Px`geUn%`o>32*27U|z5{d=VUE9pNh{b!{Ay7Z4of1=bYr%C?;
z=_^TYNp<xCd%=pe__8NzFDoi9Dw*x5h>D=2PzNY7wmInQXtQ_Yg9UyvR$4T-sCagn
zj#_eKq$?VZtS_+_M0{=GISnqatFqChmd-7ngD<Ci!9sfhKelPHBc<27vQAH4h{cNV
z`u2!#E&NEA?yobRuv5|B)u|M9gd+Z;$|}!nkd~pgj%d-^Xdu`!8+%%d7+U8GuTzRz
z)^~uO?qiR$h>;9i0xIIV{6QZR$fwR=L@DA{Tcs$1J^M-#_zoaWd{U$o`PZqfUHJM+
zUDtxNMhHZko3X{e+24tsm%g?5rccmNeO+C?^*RmX`E$)ki2Qu*fo7x&Aq)HziG2Cl
zl2Wub99D{&v6<7~5lO11e>h0J&aom}QYg=MiYH?c;w%X_JmQ+&c?eLK8tI#|oe0T^
z7koPdCWpDZ#W@f%r;lViMSMgOoFzhupXBs4h&lq@Yvg}-wo@SAEdi@VC^7yRdR~GR
zPV70>80p8eod|JU5}rO8#3P3lmfc8ij0-CP%fX+K-WcD;<h<P&&wwIaviv>^D8~{b
zy)k}l=9~|X(?2@S(&-F5!{;QAW0Mg!#;0OA01hz`9Qh}o2LFW7eI9?5sgd3o565JB
zV>~qSH_{vB-<*`*7+;6wfNhM&Oi!C5IsNT`Q6`2xVAvzK%7(}F0kcc4|2{;T(+|q=
zZBVA~O-jI-JDuU*iDz?q!~QCsA<&eWf7E^B*GOlCzkvsHdc%HddQZnN><9xti5+FD
zOmE14jkLek$n?u>MoQgJ&i~$|^n>`Eoo;Xvxx%83L0E+Jx|WZ^`xSnPX`s1J*3-M?
z>!bG#Ls2F&>Q4rhAv~DVbDJt%jo8G1y#<D0DWIwOKLz*{{IPA)8TRJB*`%r{cJZ6|
zNzVTzL{UUWdcz)#$@Izi$>BGUmN^;e$D}<q_A%2RLRzD5;ZZT{v2mH+5UkO@jr@&z
z97h^+{^N6eB%_Rr%@~RPB7pcY(i{E<fSJ=P(jHcd3B>}HQL!SFntmeC%<1b(e0n;r
z)WI2iiU59$e2wsApb=)sZ;ed9MyBU}akI0V19(0K>6ue;65L3CQxaU~N>9V6Q}Acx
zZAN`r<}aIJJjzIK02odZk%$`NI|jG|Qqi4*=X(4Z<u~*n`**th?{3o5I}3nda>@Cd
zqEa4nAer)PXBv1m7B<ttbCgJ0cy4?iW+s+{QDr<W{>1n^&5D0Q+WAcmq+*OaR{Rqc
zV?43KC&lMsR`^MZq5rM$$%>)Zt?-i-Ltk6rQ{wYjE8M0S`qc{0hrX;dgT!+VT9=^@
z&5=1+@-_6F75=HX{<6ZS#plUZ_^FDaZ>;ds;(Eji|1|W1&kPdJIeZgt^m|im4qo>e
z{n-RYdp$!j`l%Ie%<>aZw%1(cuUZ14KXK8UOW(<Im?h!-?7-~iNjSHGo1I6(xy{?`
zbh~Y;!tL2+Cnnj*m)oz+ZlefHRruM0*>QX)pWJqBc0Uw>sS3AMo85K^=eB9HdrZQ)
z{n_k>C7j!v&F-*-^K%HZn~=rORE67!&2GAc^D_vuyGX*hZPn}+N%%H={$O^?C7ho(
zn4J&sY`i-#`iosgZU8(BBPee>qC-%)ZwdGrN)67EqBGz>!9%+CVDQiFqh@FD!)>8v
zw@nb7s&Lz<**zlR+}3G!F91&d>t{FxxJyQk1D=KU%%jxk4843Rlse%&XUz;Z%C$zq
zkzBY^;HN8>(ayPz((FD1eDbqj@?)2g2K~E}=yR`ZI{MsyoP`%($$H?MlD>VWQ&QGF
zeh(+)kOb$;E5ei437-@6&rqC6^%xWIRK3Kv^rSPGM4zv~vhX6xu+I#pGL?(lm$INg
zoo86(#*?4Zm6!C=$)`!+r|P%Q0nQg=hP}rS-SrZl%;)WbPO4u08sIZg-m&YQB0axV
zp_>Tbo<@J30yy!v>~jh{@r?qWdI8tliU7xdev$z`855$Zpffz~6u4P3MWukplgA%c
zNc{SDoC0Rx-;@FWKn6H>2BqVZ8zyEz4;bG$8~uoz;|Mptk2dVVd4QAudO2S~a}usr
zz-7SXT`lp)W;jhyeee4W(|FV?WB*SEI{PxfUj>}?lD}Q2pNBBrJ}fSf&tfUJGjy=>
zDHw9ZFYa;*$U_po0B|9PiF&&|SHiEB_*-An@%^#_954V$r}s%6ZhSlP0N{3+*6cl+
z0sb=J>FVXLfKxuhhFva+m12D%9sZIG@GCRGBN^a75O9=B%8he0y5}?C|6RaS?XLGE
zox!AbUW`STbo?*P0QYBre+h8P-`VCA{PEj9x<6&W=ORS9d=~-EdS4^!y<8`wG-SZv
zAmAq}Ta)_Z?<9Qi6{n!U`3c?M0VkhZcREE_?3l#Uw=q9X$Io*Dj`8JhIzGR3pgWua
z|727U=?^{U6nNj1?Q<CLboBEvX-)^fC<A;2;4`z)Um`kvh*!8^2K-xz&+-oI_)xRL
zJs|OW&(Yx$to%vhJ0<<yvb^O`IO+Jj18}>M*yR5NaORsY^9{+kCR7CBTb^|a$Sf(J
zF9ObbAC~ow<ih<};^)^oMVQ~x(!B`yDQqWF@5$z-90Z(tX3J5hz~i?UbThzwI{uvi
zGT*H-UnCc9MF#vf!0i&p>}>!%Kg~FGyPyL(B<a1EGSHud^+xi)KdGP22b^*nOsdB=
ziQk#j{=X&QW)Ah|?Gk@X(l_+*VM)L97N<a*XxH)QOrv~Yx5)E9R_S2H2RQj0O5*c&
zfscBTcC68Ve<bnOB=LCwaF%P$6;45!-yqX9VBJ!*pX;0={G9m3&NB(`y+?=hJOH|H
zoTl?*|BF+E`7IOO<A9S+=Vqq}za!hjyMVLajdM!ChHz8yf|B^fL8l0dZ;J7BG2p1O
z6t@)cQ?gG_na5p`0slJUL!MiuJo6<#w`airK?e9kq=S4{$??cw?1>EcZ)JdIPuJx(
z{H{}w@Jm8dC4BJjI(&|VpKstx_;nJ#0&vlfWW6Ih;W{OLuf&(dRqg<sa_&uP&-V*_
zMRCgabCQf${~`XCB>eo-C7rcSfy!^u={hBR&GSwX7T-qV>0Jrm|CkQHU&JcPQgj4i
zpS+^OWwTOR1zaMTyj>aK2LLCZYp&JlJTB9`Wx$W<a0~~+op7d(KPKDxds0Dn0#1JH
zQf>_r|4xZ-A94za@s02dtg}=8#<?}+5`P}x<Y!R!JGq>xGy<NkJzonr+h@I;rx<#5
zO9uQO3pn^plH0?8=Yvl2`o>ck=*&1<%Iz7wT*kPt0`PQvb^=blG$haMStZN+14(E2
zr#c<O{`?Xa4n_NUQHOKAmF^JW>FDE1G=Z;#odR!$#IKO>&Xqd+kch<wKfvws+3fuk
zaPn!F=N}n3{A4O!z8_?OpM`|vf9&T@L7v~D(+%562mK<ce`U>(a9QuhxW5!|A&1AE
zOs*K;UOy-B1<3OAW(N42nR>aLvRsDz{eY97{jWF$9e#^S_btHL9=6K)xDLfmU`c1I
zq|+-So|klnhMfX(MYbL>F5nW$<ehw8dU?*w09OI096DuxG}foTodN%Tz}YW&rnhjI
zhtT~o13nK8A^o9Vr@-SG3UsZ2lMe6TrL&7)Mfob=>Dogq13c^e^!Voj&V08%?G%*n
z&(bqk3^?tR<o@WAbb9SNzK#t(1srse+EuT_H}DMqUjY~8J)qP1tBiP8(n+?*=3JoT
z_dcxSXUY1F0G>_`TQb1!%>aK`z`_5xq%Zk7oB==YLY<#tua3{NSLkj7JRLtj$N=A$
z0sd+R_)!7Jyy+FE$ZwtaMc*n&pYL1&$9ly~9iL~?&@B+~H0$;}O-t?I>6L0AB#zgr
zuD0xiC|0ny!~(&2X33Dbsi1kOpz>5uMGB}S6_UzANh%K|X%eQgRWd(?d}%5ql~h>@
zQdufjsoBj_)Fs~KRh3><#i7coM@<p8`kBUv*HO_K@HhLf3xxfuI3BJGrwl71s$50A
zi2W=gsd0#0at)-4wjyoTMtb!#`U!nr$2>vTI9wWM%(;-U5jo!y$>T&_ANQEHbOq<8
z7~{~u_<3&4U6F7kinG0&6;-{wx<U1N8k<yA!FiN8A}x%~Lu#a5Z4To2x3H21q^d0;
zwJjK0>kFzaIE^r@`l3DLQXF;GQZ%n(ehDb2K~lLlNZ;3^Vy{ludPUzG)DmrPUymfF
zkT|{1f~u;u4V85+m8V~-l5_bit9*HFZKJD6ZK~vnk-(tr<0M+FfN%r{D=KQumCGvY
zJk{|UaFkbvI@hZ>MZCkW$|IIN%No`B^6X+fJ5(K@AF<2liul36E^nX|$KC3#yMYPi
zfv%n=)#YleR0DJh_buUINx{NrmNiD^>1XA6^ky(Wpr67D+QsIa8IZX)M&|1R@A0(8
zGxY|_kaMHY(^!to9P?AoFn6_!HeOP!syN3mbe$UXceF*;ss666P*?rZW~XzB*Hcx!
zYL!}AR3=d8xA`MH-_+CG7;!jwo*|Dg4@APAW@gmv!r7w{N2z{NCfW|pH*H5zn~GIe
z<9xNDajmDZOl_R2HaP$&Q<F2$%bVepOUK+LqLJa8Z(Sa)`j&vwiPL!_t&)&hQdCkj
zH(q-g2+gm;wz7b~%hgRx@LO&^*;6u|v4GVMSHsfAFpH^rLSc0|juUUkp^7PGX$%F!
zu7DMcgBipANHVuLb5tDvnR5K4`k8o?<s?s4#fkFGYI8U$x)4q#ZN5gFW}igaXi`lp
z!R}mhDc43@GuNS&XR1NRJfn%M^|h$YzHp@5iqshj2YOT#&lki&lu2wzb>^5GKkjz9
zIBs3<IPu19Zrh2boxVfqZCm!f5_Un<#xZ(9GG@8zX-VoR$jC8YoUMzKKcTzoq8)Ai
zE>GhUwXqug$K^`Xfx}HDYM`Mp{83G=q{ModWwq3f>@G%U51<1DTnX*(C{El1;Hn3A
ziPK@BbdY0H<1zb-pwt*CR~y!%lluK0HrldIU!cpw_8WGgT8q_2e-P(#I>Hf8V+s0X
zq%4R-LxC#v1|Y7I66g&-1cJ*F)OF}=Dn<~Tk?FY{-5^rpNHcP(k;cf}NE9*z5_p9Y
zs!=_c3pI#Lp$K%2RGfE>Gr2<$vdFunG#(>HVQYL%=kg{NTlSSQQB|cqB^c3IPY{qv
zQ3|b$pEOOv8z=!MmIjo*<)KL#`di0#H3v~)^PtD7)P~Pt0BLIvb?ADVL(t{jYJ(gx
ztAoC97{g^}G!jnk3{6dt!Evb^Ygbc0(-N#t>VE#7K(ba*$w5~{ZS@6$$t)~G&CPA*
zO$bog&|`J#b+~P!LyV9vE9oMKBQEm{`qWX<xGvC&<GUMt9oInR1sVcupcpD5xEUkV
z+TvTE#DLfnY}8~Xs;TO}H9%Gx19Kz(V6ZKTQB+5Bk5BFLi=oWv#IS|U&Z?uRR3wU&
z2SQ<$o2Sb>umhmQt>ugtvI19sR0TuL*M#*ZrC<GkXQN4>p_xW_(WuiZ2Xv2enjJz_
z3r(!H`vNJtk&PkV>|6ZJU4HIcuD5_YFun-hlUqX5O1(EER0GO^URh?S>p(|0G;#~f
zp0=o<n>wl=Q;C!~>Wxc9szaeJ9>mL9c0uW7R$LXdTC3YzmUr^JeOC_*fv~H-z8NCq
z&>0tXxpC-yQs+(I9j$V=_RM7Up4l9t8ea-~B*hp@Um#RNnY3nFGg}6(nv#soDL!p*
z)d%V!1koD9(uzo~Pgo=I#_&I*ua`shf3oTt{WXj$5861mw8BpCI2@|lz7{<qf(vk5
zNg@!P7%`~|jqN&asp8~oHPou&`kXd@s-=qA3L;>q1cNGun09<R5%CMb2g+!Jxayk?
zoftQ6<*;HsfxM*De7Qpn4u&QvHfmVy!g#c*m6~*FCDBv#Jd!65$-|VcQfRl&JEq+p
z40~GBn5kwHJY<;Q#k$Fju?wm{K3BjbBuV_9CSG)8%<`J*syXdz5?^HK#WhsbKiE1Q
zWf*Lr<5>69YX6`WLG_q076c)h5OEWsF;d#@>m-k~dCVq_#R77<0zEFYcvPifXlEFW
zOlFz2g;bZMvEew$DTD6!EQfW_66jE)_-2RgMVR>+vPerQVcM(sB4;h86tcqNQys$!
zGz|f*u=bMBSg+}8Rkcn&kHw**!fM5eBh=FrzrH(wIiBnx;WAi7ov<|7u+SVs>p6eo
zTrsKX9a}mT@yVZyrk7y_S}Vgb-`rh=Rf3MHGjhy=l~r|xd@Y>UToLLDwzvW)g6Q9Y
zd6>?Gy8KD%{i7$DjxwvPToj5!3PqpoC8k6A?H{hju+_}x809EVT=&67h!G#MgqG~_
ztqpWLU`$ilg~KYXIT&!QA;{Hg`y8v}X0ft`8U{t0spv5qZ$(^yaoLG%%f;Pnn6e+I
zF|`y$x2%<zhUaBiN^q1~F76Vu70%K`t?A+|sSeXDRbk8dg6XpVs5vI}aGE(E7i|&^
zFj0yBQ*&CwE^Vxk(-rFkM|C96lp1S~XGV)<7wa_^wB*v5f9UaXsl;{BypX4(#ov=g
zN3gBQ`K$Gol6coPHGG;^Bo5<Ij9or-$TXuqhAU$^5Yq(OSU}34Y%nY|!-%lzAY7t>
zdx$W+CuvWZ9u;vrB6;4RwuD3KI$uW%zFO3IEW`W;HpbHC_Vy(D_&76USRRH3Ppsu|
zfif<``4}@W3vSU%uJ_>*VN8axRw!=!khylD%T)lv&aX!5DXF+<t@_b4iis*xuZ;%$
zVa)3}f`&~wC3SyCnOU#;p&#XJt25f=SGxmUk*E*FN~3m?36+=$!6MAG_)&^kSh_wd
z8hg6=*RrM*ozP%WBEp!WbUy(Ty8?q5rngYHVv22A6;WZ}qFBdX$>5quyq@EIrL;^R
z;yBN03R$LIUaBy{({v;l8ffCB34#vMR^%l%b$Fk^1_65sOL)>8lzgXK0AvQn`fc2J
z)Yot4srAgqXui7PTvVZ8(XTz!oosN(0XkkfqYI~0xe@URD(%#<qe9i!+T&6$5gh{-
z32xRgx{+M|h~uPN4J7Hra(w!Us>#4kU1JmtIKE(6rlzdd>x)d71!C1vSUQGf8W*ML
z#VK<1HPWtP!{YjVM@Wo&>2#T#=$ICi%enlIX;E0G94O-zbx9UaLMmcBkj;y4GuXaV
zd~+Rtfq~{<VN4T@E&}xk`xCE-d>zgHwHa17@Oq=6G4-Xu@fh<BvL`f#=C%f-;dQB{
zNHU-}r3?o?o(=WCU;#Tm*qauxKXSbPxGUJ7fM(?M&ZHUjrDCBJ4W)-njFq^wt0pN8
zJJX9Ry&ccft-N$&#mBMl(y0oXcdESEOTEUwUd~t11H{tp$5{`u4*MK*0$pKOL(<zI
z!o%}mykj&MFO^b~=}Pf`(8h!kf{4;-Lc|=;dl8{-@aiCw#gSoBT;kwzBbOa517ezS
zFrlE5?Om(Rj`yI%GKit2tsO}>HB5^&iS~rGh75X}F9(F)#;Dlb-kEGiqQ2+il?N{P
zTULSB2OUvP<T!`J922k35-&WkSju&{k`!&2tiG_8*4_Rw^;e0oAL9$No{orok%~81
zrukL;MHH{O=9Ca7LE_~h-som{d6;2BuTLQ4fWb@kxONr@hdc2?$ts^fxHH<(Y}n+6
zeN5XWz9@?E<oH&|p-!-o0QR_t3(#Tg2Y^uVUn~M+q?(5%#FkLI>TBlJ+3^a%DI{&u
z#)TBHk{D@Tm)5{7aR{3{{svZsaB!hFM^f;On$_Q|^R+2fBSX!}vj<}#qd2kA;2+gx
z8Rn|`Y_(Behrw{y$hUcs^%#yjF=U6jd~N>Jngj12N6&q%m*GW@QrrUDCUQ-E`z;oH
z*YYyHV@TOC53faFWL~ijuM9=Yfd;^`onM^zf}YgV!|fU25YEU6plsd6X)8;gjj%ZY
z?ekdFK`nJ$mo!QHM5YMk7zg!^o?)pnuC6jDK1w!aD2ykH+k$9*VpD`ZX`rr@6K*ax
z;Xh2mbaN~*IYwMKd4{U4=(rA@1xt3AfU$dGr&(!HF*61sv6SLZyU49iP}AQ}YF=b&
z#a*hH1V@c&M7)>F<)xDN9)W7}rOhtO`x>i8Pj7_-$Ml-c`nm|)R5+@cRv29RYpKRi
zDPGySQg*&TZ#(!NJm~jbgXuySX5}ekHF{Y1I$tLnrZtX3L8s{zsDpcHWO2o?7}v=4
bzV*028A}I=cV$9aEml3r*NBRkttkH&hjv>s

diff --git a/tests/Grid_rng b/tests/Grid_rng
deleted file mode 100755
index 039db0665ed14e8079e9c886f88661c6e82127bb..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 60784
zcmeFa4R{pQ^*=s&gQ&nPh!_=h)zv0~m<Vcu2)iVK*}B0LqNWv1LJ~-7NMf?V;A=EN
znJz2TQj0BZ(c0E3R(xp-sF27@B2^4p73&M}eb$HsREUb~|MR(XXEVD=X#4$t{r#Tb
z^K|m;oH^&7d+xdC-h1xrOm55d&4`bSQ>-gqnW_*gKG`iu=3P-Lo)oC0Wh%!gZsl}k
ztTF_YB>dbWgwTAlV6PNy!DPvo0GNnhX1qzrjCYF%3yw9(NWr`;4GEcE-{qE%CQSLG
ziCR!moPxl3)V1SLfdUqb2MY>iK*@BN;Z{7V$9TOmUayR2K}G6sK`VdE8-E2dydd5U
zjJTUX$r1NHNM2s~ANtGqyyj;D1$2uC3+7097PR910P(2jzk6BPoGsHE%r1T9V@kHl
z;ryz~c^6GOzpCuKs><qy#`798FFNm{NfYX8CR`x$P5I=VIqM1$(XwL-FzP%SKh^<;
zcb$6M@lAK69J=)DRi{6gb=7^x-+9sf47&(FtK9SwM{!A8P8?~O_+5fu$L+_y{Oi(Z
zx81!k-0_)X+HZnCbnk!X>U*52xv%<0uNs!ER3!c3H)CtpFZN#dSFjIw-3Vny4G@xo
zN;VLE1u`-aeG3#Gh@Lr!oNET5Cn3WF$vHTPp0u-b0R2xNgq}1=y6Xm!|Gh!@j~#@*
z4(Sf0|BONSR}GTx?St^2J_vpMAmuV~5IfWiBIjp=&@UfE&aH#wchMmFd_73{emn?$
z&LH{(2jTzqAoNoQp)VLj{;omtb!ZU!w}aIG9fOo_<skN4GKk$)4Wj2egXlAS5IN%q
zkrO|N{x1$fe|wO8y*7xQw++JY7=(V^AaX7nM4xGc<g0BE`ES8KXDBBttxpugf{Ico
z?uzmQ@RRO-s#s7g`j3z<>0OVT^d=FiD6dHRni&4kppOK9(_c(}r{o_i<rm9z^Cf+g
z%x}|)Zb>o0GAYOTWU(MPz{l&?_|ea#r?_$NyY7|xC^Fq98FUfo!Y`fV?qlW?`muU{
zS?RcO?7J>fii#F2tf?-l_m|Z9i;9$@?=LEvUA~~Q-d|pqT~$(FUtX^ic?<H2%F62`
zxgamQs;0WUpk!WExfwQaNKvi7qOQE8tmwM(#YLrf@`L?Hv*y%KLJGAR=|vP;@2{&_
zoICrnIVIlGTqsl8SW;BmP*+~GsIt!AP*TMhm(B5CSn6L~TMm|tiHLSXWm$Q3QDt?d
zKX;K3Svje;y!<+W%<*SjTvUXVYi=l-hr~;&s%lD2u9^$yS2fgESZo*B*-A_5{p3P)
z&}*PpdC5W~QD0u?_m+a1QCd+_SLClNsr1)FkKEayO`KO!Us-B#dkcyR(u;EETvU{Q
z*_??*LUl6Ll~kA2EJR)xRhE`t4E2TL$TBR8^s->-g_YH?j;UpAj4Z@hQ0`w;QYAC%
z&#bJeN8akoCwXCloD7S85pr5mRURSKXJ_Q*Usf71E-hJqacRxM+LF3*h6vlmrbi9u
zLq%z(s(SDIe>eZ>MFknc07%j=5~?b1gjrCqDEG>XDi>7OAZ1x*b@jQkE2RpVrX^wa
z%IXEC@dvWHs3p0`7Oh9y7I|Tc%WKQ47uJ-qb`@MX3l$5ssIg>T<)RE$g;LhyB32(5
zetwNr$>yOpR~MC*_)9B<4rZ;qAiZc#QTf7J|Kg&$@>;Ztn!1t&<&3kSv9YMOyso~c
zx}>TS!Hd#mwE7!LYEc_lm<3ltrwi>_nb$CX@O&V~9RH-!ng)OWxTt-D#5IjsQ&pc!
z<5l0VP(+T(_5~S5MT-_z4v=*eLtpjDxj+_0&BWRUe=e&-K>>2`9~H&`siOeu2TXj9
zANG}Pv#Pv$fxn`tysoaME>abvYc|T6Ri&&de-XXbLepVY)htlzudnkfc&H+ML1}4G
zJ;O4TqARLzsH`r7f2nU+SkAywGg56$ePv_O!t#YEyam+?vWxO6qI-kan_p5{rPQM4
z`R7XygvQud=~wD27l^u0U+z~BSAx>o#R?jvzj9%D&HS>G#bhff^;gzZEAvW9uY-#y
zEhqcJnnmS(l%o2@3+JKKkOI|`4tHTmWwkQDuDo2StgmgTF7+!jv$KmPPDoc~`n=P#
zi!PXO!GsGV&l!`<$HGFeOq!4tg=J(a*@cCk>E5D@2@|72qyQN!Bf}nu4j~@%l6Y|^
zM7M#<d`iH~#d<>biD!n!V?Ju~$16jXWWhoF7!gMPp`Z-`JXTPN$1BGne2Abj#Qfuv
zHHVAC%;zQamDcz^|D46gDUh%R{X8#*|IHTOaB?LQJVtpa5+1L7D#O#kzw)Q|LiBOU
zQwXO_lRv@6KLirZFaSlrEe%;rRyIriir!xUAF8}4@mC(Z4o^drw<Nwq@FyuBOT0wz
zCn{e`+@E$2rpEC~qRe3LW5rBH<d>GRE^B;&Xkwmc(XH_U>74snmo<JNopVF$vc?Of
zUuqEnt#J+M7u)F8_=WT=8{Ha5kUq^ux5hQ3Pqxuh))fnEDDHBoB5<dT&bnh=sW!S<
z*UXS~8y&?MxiW2Zrfpqr8{Miy4AX3M>SSH{HoCQD!LULb-Ci!mHaaR|<f^dIhefE0
zQfs3Rx6vDI^bt0ClZ~$0=u2&M)?4dZVWW?<h=8kXbo)GXjg5Z1jsGDVeTt2~&PL~2
zg>|jB(Z^Usz*ZamL>s-sMnB0$@3hfRw$XRn=%?7|T{gP4CdOdHMnBcY-(#bnW~2Ao
z=v*U+Tyk8`Ix;mvga@$Eoi@6|MnA(wPqER@w9%b5`dK!5s*Ucl(bH}8vu*TD8~uAW
zy4yxiwb3;jJ<UeXx6#kF(F<+#aW;CfjXvH+udvbGHhQg%ex8lqXroWC(VJ}a^KJB{
zHah*db*-?`Gb|$DDjWR*8-0z9KG8;h$VQ)Jqp!2kFSOCM=1-HgK+?1KITdY5t3S#3
z6#}*9t;t(VrtpPNASitHQ@E?AyYWQ4f`Or~Fs`#7CQjEF>J<0`#OV@4tpfigak|3L
zI)UFqoGvi5M&LgoPS+P&A@Dnh)8&Pl1b!oNR^w2u!0U+9#f6Fm{$t{FZJ~UD&m(>;
zaks#KM4T=zlrHeu#Oc~XPJw@)I9*!EA@J$M>B>Tiz$X)@3k&re0paWmh|_h2x&(eM
zak{Kfr@+r7K8kp&z)vQAJn?k`A4QxlD6~f4#}Yq*_zHo?6Q?T*H3|IdRlw<jLbU=v
zKs<$bvA{nePFEDl7x;U`>4HLTf&Y^@T~8=o;4c%W%LzFJzJoYjO~@he=ZMqAgcN~q
zBu>{7>iL%We}cG^c$dH*CQg?T>J<0`#OW$Rtpfigak_}mI)UFqoUS3XM&LgoPL~i`
zA@DnhpF_M!;5QPdYY5c}ypA|sLa12aKPFCB5Xu+$JmPc#A-BMPM4U}OlrHeu#M$IS
zPJw@)IGc3HA@J$M*_1<yz$X)D6Atwp7UfTzO*hmf@N<c?$%Z-wekSn?h_?#-Wa4b1
zp>@FXH$RA#4CD6@y_stxn}}{~Me8gS_Q~Je`YghiM8b)RaO0<-saLhCBa`<04z0C4
z=>q^>o0TOH)DdB<h%oLn8DEKF?6MdOLsP*ye}Z~?i;%<ioxi!4=_f_fCo1)$eKPiX
z^<CF$`g>aQ2R-=(-uBiW-3?{hngnRiKgF#WuOp*h%vVp(QIz#BA!}D?=XC&R&AlTu
z{lgpLui3Jm3DV(wam{81g}c<#n@C$v%;InBO#r{9Kc}^QGEIx?)OPhYAo4~sYD->5
zWKG|0#(z)7Z@TopUpg@y9(e^~8n>druGyY+CrELhZ(;tl=1V*70fYavhzg^x*xd6d
zw9`KfcU}{^2~pJ3l~BD2ukc*uxzclmr(lkD`==V*ly?&XcumuvaV>)%)AXlYO@u*-
z2Tj~~w_H(-@mG>#ix!xu1?D7c`s-TSq1=oQH2qtx?b~=3XIhJ^n9<u?T(y9_{3*Vy
zFV(x2!7CY;!029ms}$hV9~9h#bfVs1;#y5NV`^y!H2oE=?VEUivQPim@SqeBzMA3W
zrGgsZCZ@769yepc9Ng}^0&?_rZ{W6M@3?6hdo}$V3X!^{GbVI%GZrrYNMF|H>Rn&L
zgBrg@3Yxx6)7QEL8(=2cgE_xOrp-9sz{TFcRmpbcG<~1(;5UeB@tKjHadkl{4dxT}
z2(A>`YQbefxbYlP+)Te~q%vbjc0GF?=EJM!D%yJXF;o&*GV{6$L8861xE|u;$QD8J
zZUzsoo@qXm;Cd#B?6p!3F48*iyuHP>n<TDdD#kgQqJW8;MZ74T&&I4#L{_^rATp^!
zJ|ZcSD69fdlr11H(i-uM1eTH{643NL#&a{lsLypJdxO_FXmA-pB!+zX^v^W?pT=?1
zO*Y47=G7QWJHi&hD-zptDc@9%rJ#RijBqPT;2N}*y?)faea1Vz;jq5im4kXrI}>Jv
zHwr<oR3eGMQ$lYh*+CJK$M^_SB4&cSU%XUDiz^k+9(Cl5Vj|pLBa%-fSt4#Cl%qP9
zxpoV<(q#y^+SLn)7}uubcBK=ySq|KWc-v>xG6iV>T3-vm+ThlfgfTuj1zC`$UWZVc
zdJQ1W+6qW5I~WTWX`Oh6sk=xbFZZnp4M2t`!^(`~kk%$G1<#0+N)k&~R7;pG9fSbt
z1uB~QSc~f|KveQfP~pfNG#aI9r9H;MT#<uSOjQU$ZN#&21mgssUgmPj93+D%v;vbz
znem~GQ${2$70=KnlO*!y<AKl($nau7#;G6!F4Ai8j5tjsY0!Kr9{Ov(0#r+LN~7je
zOwD_QD3Fk;hhTeDKA&FYO3?xZE{Aa!X3h*D7wZT)j0e!iv8afcMOAyyEUGlvADr}b
zof5W6xK6?~60VT2Nx&RetuOeXYaJ}i<k;l2pdUp2JtH9<?trTTo0@*&RyLl62*$KS
z=v4(R7yu3BqZs2r6OlxKu?#)^rk^s@v`xElEExhx@Bh>Z_1%URSmnUphqmNRwWS@y
z(59O*6@SuCva`ESVk>>&OL8ZGGkp^WnV^L`)lEI7^qrKYw;LZJ=NcSXvDBPxStw!n
zFp@A;e2{88sp3*_$ExiY5|CF@@T(ZGM5{fA41uKa1E_s+f3;6Is@f+?wa<r4*37gV
zS21t-TCmDh>_efh5*e5aioVCEf8^7@Hr|8DwcyRJLW#djT;C_LtunO2tA8s=kAnnn
z!z^q1JI1_n%M#ln1$ZmejJ3p%sX9tM-?YR{7}`Z!!YlHYbmj_YzXCitsb2ve_XX`X
zSOE^FRZ)Ovi$u-*YQeQ;L9z`TFg`g`<{J){_&+6HEb$${efmC$Jp;^^>4jxhrbCh9
z*P&5trW>ebU#9QG_%=Gz1tJ?s=PmD_>GRP5V`}Y~&ySYr6lNBgP8W%;bk)W)htG%}
zNGM^PbGD`Sv1~fXlVZFKnXzhbVmf_lzmyuQ_OFE^NsI38ul5}Y{nRcxa5S|`s4moA
zEE0u}q1t+mD^)99Eu3_2#z8h?4w-XYnOb{}OM~Kyf1IZ8*4nxfwcuzs^Vtf6ZE9kk
z?%p2`Bi52upPKWOp!;+`q6QPMW&~ec4-?RMb9)YMSJfpCKx+twGosB_x)f<Ux_eq8
z)wmIJE7MNr(RQ3^822(`3^nWYi5TdKK_aUqJkZ*d9*SGCS8chPR=GvEpxo^4li7)T
zp`;eTh^Nm7?GrwT{>evUz$j%Rg_W*Cp@-$Fsm~{v`<eP&O?^01GKNI!a~FnMgXr@r
z@)x7ehMuG7^9&_JpMOAU<G0r+O5T!#9G)IB;zc%4%c!B~+TB9XGp=5dd3N>_B*lpE
zjFpr%5z4WWqS5j~PVc^OI1f)qCTw&A(E>N4>k?sa5k;A^vKad?&1aV-2CqxNHrM4u
zs_R7*bFgg2`U5FT-sB)^k8vN6j8^UTt(*>A?V0Pj#&fNw=$b8}gA~2Hu$?(*2^~ac
zEMn+K&^m&o*%1{;H^R`zMi^cfQEe3J%4~+7j139h2TJG$8ikXMzFv5W@yTc5aQB7A
z?*KyQe}z)Gk56y$X}Ng%8c$33)Q+*FaTA~XVqh->^%c+k2%+e{;$eJJ`ikckN$D$|
z>jWjog;l0Z)0jV$Jfcq#g))<rW|WcQKUS2LaUm&IF==cbNJbVDN@$D}p+&|TiBg19
zif|eS_Cf@Ti}hpkw7cEPrefSSHp5UGn(<`p2ZON;x6rSUUE>Zuoq^dFEQm^~wj9Co
z2P-P|4ebYF6rtpeAK?~?Lo<t%Qypf5(Ixk>h|2Mjz9sZmAeaodAYnzZ%_o~bP4Vil
zYWkO)a0PEd-P&m^oU17F)qSm>tG6|Gft|2Z!d3~_Nw`MB6%sa?FuN9!vWt1k*V>!u
zNqA8b$B=}n5~fI)%zD$_tO4-S+7pUbxMA{!WNlo7g1X*zBwo|&wG9mQh>5fR6sTNb
zEX2Z;FOaQif#xR03<Q<}_=2VRzTneBY+9Sv_Eo%3->#+YTr>&=qBZYpGA@+yx`)>v
z_x)gElcqN_BE+mR#jGKSi5q;HfyP&3%rtFr1$XdiS_UFa5TKsXA{?52GlsA0EdeLh
zLm^P428__1Bq&+{#tU3l7#)8Er`0!Fv5&1!HX~*i`hvgfFw-_cptE1vBJNZxZeO;^
z=?gyDWkxqaz#tZtdqh8Egiq*~b8^mCYfc?~tti@G^#ae$cs_4QtJ-oL%F~DZCHn$-
zuuFY~7I>ry;dy~8P-s^>w7^aIzQ7|iDhiHP%gqSS7guCWC_9k2VY;5UX}X^CtO8^s
zkj>KriJR2Nnia^`#eI4Jp`auR3h#n)U-qFXPiHgcebWz3*^H<gz(f^waSt>f#`ARb
zvBb@~c+7=6IL{kMbWINwY&eX7g7t@|>!a6u14BUBY@>9aYOXH{=Jv$(NKzq}^%!o?
z%W7BexD-t@PjB;~1Aqf+7Bw;uaDY2v-kq$NW7t)!!K$n;aG6GXZbd_s6>KRgCu8`6
zk2o1I&}@Q0>Of-l89%g%xg}Z*W1z6v4k5+_0e3$!r?9X`Nin`aj>|!nGAL`5Kt-dW
zUW}W!<axE_UOadMj*Npoy+dp6>P3WC)RteuDn5Ou*8EYg7QD4rYulHA^__6L|8!0N
zN^5>Mu3-$k&C6g;dgW^2dEUfxNB5Y|)tZy>o9%cu86rU2g5M6UIUHY)2C}_7M+^EM
zPS?^t1Je&War<ExZY4dqrOj6}-Ous%ciy)0_M+C@=GL0G_cXkO*bucFzg-OZCuOUn
zTKSALYCGci)G=SZ^(AlC{)R-ox4TgbR>yg>LYm&~RcF4T>2GLh`=NY-miD=SD7@2>
z=l!W#^H*{HlW>o3fP!CkC#sK4c>=$OwJ_GFJ|m`ux9h!H^EZig=jgqbKStdHTh0^U
z&v_!5w_(sT(U-Ne?gJRs@9_rG^8!Qj0;64CtfY7YBR0d*O2%Hf-gw`kDS!6pQyvP8
z{<BAYZ1ltW=tuO#KOwn4BVk|Gi*@f{GbP0~N)YR1dkjD3XdKK{h3qS^^H+GT<U(UA
z%HFFV(SoeD!O^Zw*~`&kISh7+@K+o<CL7lys)ot|jz(hxH?Xy=m((R~P)~3Pek@Ro
zg)q+239Z$h53gc%xmK)`tvS9g+&G*HgzF(-=EWr(9`060W{ZWM9x&rstn+l4&tjQ}
zLb%McPQaC}RsmPLIt9<w>FUTiPIcsE4s~RL)}CEU(Vq4pG^MHQXy)zqt<3n<o@=*g
z&8@wnw_)L)w1MlbJ!)IBzRcwUN5BN#Lj%ddAB1%W8o=^nz`|UvLW0I$KSCdU2>q^x
z(GV?WJoPa~L3JF7Zp2;?$lc`(i#4DB@~D*07%XL#l4>C-qQPw^Fm)%iGcu7nM;<GX
zCgD;*tG=;xX(Q3+i=r%WC7Z>Y!Dvu6g~jVI?qO`UeS7gXI2njx#U7&$D+z+#*v*Ky
zZ>89)k7BvC2#4heY?iz2&(xd6XEc2X8~v8HxmnsKDsBJ#DBEP(ZDW^x4kI%k#xy9_
zolaBbVy%SmQotCKJ;bDwkTf%OjWpdAfMWk3YP_Ur`XrZEx7H}qIi+*6Dsi#GLCi_a
z7ntB_ewMwEqTbz!@p8tSnAE{hVA_@()4zEGk1hqDcif{*09xDLc&&7sH|-s-81iiQ
zr5#)}jG0L{?!aWB`?&gJFd1y7q|7CIZ@tYM__c_K{v{oYtjFs{b64CZNP>SyVtSh&
zOJ+Vibli9768%IIw23v}=5OQcrloxe;eW$SbncUhBERIg>9Bs)VfoP37Y}|ZUTbbo
z)|$5@H*Y(lw>7*>&Crcca%sT=sJIV#-E=2sb<8`vwDUf|&@@oyO7{gYkImD+%uD<5
zS*qvLKle|;K8GhTX)PHG-s2dyn+yM1FgvVeZS#-XwA3!@z4uW>aHd&r*X{7>?{E4E
z1CgLeXBz@HMHbyC@X3CZ`$_qkJ=StkFu@ffUVazIy;iVjUkHXMvpGBo1)~-EL0@o+
zv9e$A1%1Izqk+MZWk2e7JkyS%#Py@>Lo9D#C@1B!0&6pSplKdT`NtkH10L=938v(!
z(c>ko{1^`*Ctmvm9YirfXS$Cd6=rS=WQ(d1Zqszs5DlXqUm(?UNstv+T~>;jNcS<u
zr(FQcyT|1%`M})OIR(Of?8mlhwqB+CSTs6O(B0TF!~8{Z??rUBt0bf<>en1P9*pnN
zpJ;)n#Rx=9!M2hy@U$4&2*Q&lL5!FL;Q^D-1tXaAxSyJYjtHU2Byb>Ua@3dvF~Afu
zt~Ciw5so}TxT(VApJrnDCYE7h=_YoTi8+W3xjHl&=+#^1uZZ#wSZu_63VoxdPk}L~
z=9T`91I)mTWVF}OW=SF}$Ccs@G@$>x0th$cGT!v+hrMlw%{?!!d%~3zBc|d%)u(r9
zZN^k}i5C^scof!$efy$|CBnI`Ld*nyN6E!vqw7^}?wP9vPgCz;;v)<!WL3`3(+^`F
z5iW40WI-!+$tE-+&U$6GC+`V|A?-^hkBAsH^`@&!o*)xq``FSojJ{ks5R#`G#y=4v
z7j2!?JDm}8Ie|SH6^Rc^e+HJ#q#I^^fS;r~sh0(ZEz*L+I#>bIv6iVdzn37^XRx{I
z)A##JwC1nl_%Ia@n8WSItPdLoBgB##h6!#5)A9yqVfNmN1(tq6TKM^+hj^QhB-T|@
zo}3?3$Z_h(h;T~KpVOM3OYsj=N4isxC`Kr-1tm)&`T@5)mZ*VPD2QMaQaT$Rpi_3T
z)unfjVxQ4xu#DCWB;@1bA?*uqiCNPk^nc@LYi%DU<^`urFlP)2+Qvs7HZM(mY!@wb
z!)O*EW)!c{A3F`6cIj3Tl2CUHvX2fBwl=pe7ou>*GLxpGo4*Vlxt8n=U^H=@7T^;$
zO7)Ri08a%IFkAfr+yco36TnGCJfrY_0IzQbqzVL9L(F#L+qWRUG{=SUIciS8w*`Eo
zU7G%Vhw4b&qK=IBHg8Qv=6oAcpfRVjf6s&F*3#A{MsMC)Ve>4g?W~;6Zpar;vYP&k
ztne(4yo2SH`{1@WZKv<P0b)1o#glOl7&gDS%nE3X1T02?F&fbf6}J!_9YsS{5per)
zgq)c5l0vWkq8BYz{~YaA)_FJUd_IJWK|o)Dvaz_?Se!n+%}By<+l#`%fEQ$A2r{iF
z<YB!}%Q%?Ft)6qxR`Re#whn^4fwitKKp!@{oEexuzT<8F=!h7Ot_NB1H^|i<F@C!x
zQNyYoc5Q!@=nYmRVz!y5zwK@Q@JPn<8LcZkdb>yO2xK1)OuX;VEr<0sy$29;<;xGN
zj}6U<!z+;R<z}a@HE&DyqGxYF|9Xg1R);UZH6+}U^Yo6mHXqil(|R!}+S$F>r@yJC
zedABY8aIX<{u4czo8RX-G-bWMU46{|tP-~q^E@5%q;KNueBQKg@`AqgP;l->3^uBs
zg$Q&0OwEo%yv2w0cJ;Aa4#yqBz{7JWi&7etxFg=?HZkBZb=z{JVW&r@>r8Z^S9H>v
zu<ys=PLGam64k_X){Fj$G6m{3(FH`Cw-yditxSd)(fk(^MSZv&H6^A#<QvDb`keh{
zzwu!{rhS|<an*b;TA(+u2(4v?Lrd$?aFE3NtkrD2>SGwi7;5t)C`)a<Y|O`c15-8F
zEA1<+S0#YwMf#nz9a`2W>K$t#Ps<8<^%s2V^bfspFR1sl`Lepyd!9?);Zt+Fux`RG
z*X7OHs@|o884mnIRuwHc#ceb~S#iK(Ds|Gb4x#^6TP}jn@C9cW-n3qy-s(N{Ar84D
zYT@TJy%Qyx!1FB`@P-ZJefn2&W!EqR_WFl201eUzEny<fy>Si41M4m}9~t9ZRA^xF
z4JR_-JU*&R-T^lQ4=@%M;!RIJL~EZRJh>LkMLBJUC+{B=?)VoW&EF>0)u2&EkB2A<
z{=6><$YHH)6aIV#{P_&&&mmtD;m>Edp&^`n9})h1hV<uc@aK@S6aGAyvk-&!t^{w~
zjO5TEwqXnpR4p*m;niot=MBY~f}v<%Dc<1a$ky}z5e(CG55v@vaSUR*M+<8HI<aoC
zmUblLWsH|KEU(0M`bWWup9nc~N7AvLsE$k=$&SuHn)FoCQ&mUKi7D858i@p~5H~|G
zU(PuZ6XKka=_nP0adJ*9L?<z_kRcjFG<0soTGnd~%WZWdm^j-T%z?r<3DW!#YRF#y
zaj-AzsUGq#(3+3L$!8Cqv0^Le19jP2Fb{*0d}RA&SqI{^_UUFFFgw<xgs`CJi!K}f
znS3bUsOb7@tSaoUr=xtQ_x0D3i1M8-%NL!vB%*w$%ko79WUqgK<?BQF2FG@=af%V-
z*w?tAH`S|On|xU?u?**%+Pyj&Il48QciG&7e#gjv6`tae*X=%Jg*eVtFel%uzvP9U
zq9@72r~qAl0v3F6PRYJrjgEPbp`#`WmhQe4UwD=w-oUI_cCOp~D3%@iC1X9-pCOJy
zVJ{sFzQA!<O3HY1MIiAvSVH=(M^C&@KI)_I)52SPSzGGfw{nH;B^+LBLOk340sI}h
zLHIlLULl0Q*FmdU@OL5yTGktC^E04po3S#&Yg!&X+;hIPueG+XQ9F)sz>@WjddEg`
zW&IO_XLz@6Z``Z!Yrd>~@Nvnz;NAA&yc(xHPH)z?>RrD8^Y?>O9tBX0-=H*bI1PfC
zx-YGlt6LoUa<$us6!hnejVJ|gaLQV&WBr<uG7cJtUV(p)d)pfvyEd{<m;DbG>^ZVv
z*RoK%Pte-lk4FcmxAWvv8)A9mux2$-S(hQu*o>rN`irRfL8d5jEUs`8B)1wa8KN!;
z!4+$z4{@#lPmXpKXzIl?hHb9(R9Ot*)RwXEFX&yCzC!k3;$9I%2mcFZkxG8T5wvWV
z7(ssp*Xyq^2hrkrx;c&(&trwFLOY_1^0v>6c;JYuVjF{tYWK*7DV(A3R0D^wVJ<`v
zJaA<Fp6qC!QBrg)<eEcg589`v&;Lpy+NZ}YGDrU_iD;i5**-a*hX36q{V&^~un<qL
zVfPZFtUUb-Z0EVK{4x<X^QGPFO7#Y3!f_?Kax?bY_2WniD<>YYX5!Po5hI>k;a}u@
zz#CY?_F@(YR#>--(R*B*R@&ypA&cvbMoi=KwYX2vQgP12@F6{mf;}~ovPeo;n6CwI
zasWpu`Sfp%vp~g4N{T1EJtynShF>sY_+w~_+14MCB<BhkfYmz0$X)Cp7=K1ByG;0`
zws$d@8jWh6DVAUKTC}~3HT?o_;09!~4%k)6UcF$10}oC<V9I$FI;DaUPCTUYA(eBT
zWRBWr?n3<uZ_Bx)%H>j*g^rlp*W8mpUC-?EYoXJ?7){!;`Pd+a*k|IlaW%BwkclT_
z4EeYW!1WXVMB}4Qb3k?EWsJ$6XM}$+0#>w?4M!N^LogUk2o0UpHyrN-Za@Co1a`0f
z53ha*e&>y(ce54c%_%P9B(S}SvECa=uSm*JQmS0;clN)Lv<0*`+IGMDw$Ii0MjP(6
z<~AZ^HE9z`e11>p?cJI>ok2m*q>PufLl_|R#(ADa5h*@(rg1=ZBsD`0ypWn|s8_x0
zUqn(ZNG6VzE9oB(s3WUfnQB^r%gx!%EfArm0-3L=)!1a1m-Nka?q|$RN<ifK>dmbX
z<a1>nP=DBVKrQS<bf2r@?HpId?%{FZczdp^7#uhSrG@|A@BkDz;87hH^LAUZ+VVIQ
zL=>!b;*O!PEp_#(H>u+bNKQ@djMQgv6u44xpUWBd$HrTbK;5t*GVjzPQ@2wW2zWCE
zVCD$jc0)HrK(riJA+#&pJ-ie=&@LZ51+IK=_{D|?snt;4E>Bid-xJvux+8ZEp_~JX
zc8^q3x2vfynfjeQ9%2rt*S>l{oqGfceT{^Uz`zI8AH5FpTOfCU{CAK$;8{R!!`2M4
zb1OQTcb(cX9bSD1)}EV?jYvJ+(&v6wXu^T=-8o<2@>*~6SI5=e=)L``B`6oHCkL*o
z@ZR2b6DoXe29^Lgo(R@qGyMBQIj&K?JAHxCFKK}rinWAEuX}^T;=Nf1@=kgacdweg
z!`u9R9JYdddQq{a4||=`IDT&a`Z(+pb!s|<huaz+)!Jv)LOpJa0{7{c*ZKl8iv@Rd
zVp?F>OTNIc*Dnu_PS)Z&d|A7+lMeE(X18I$73WLq@TIkCI@Hk<k=*4-t_{@{nbOo5
zyO`h>YkYx|*k1i0)~fRY3(&@eZ%Rj|VZqkVS&W!9lE{uNB|=A4OoVN@{bjRnp9pe)
zf~*WA<>d%{Kq(l5L^Pq8-$mG?l?g=X1B-Fw;heb89liTMjkyu_h!{afQA#Zb7Ly_m
z#rbs1U(K-_MmeFpOVJ!6ofpR2w*LF}PnSZd=)y743wxSw!+8;hXZy?)MO1o6=$v@?
znj|kyr_wLo{-WiVE|HWPQlwuR2b%Cp%Un$mD_m2f-8D&ipLfDOq}@9`gHoe+>cY%V
zZTSZ<(Lfx--<S$drGIG7uhC?&+9VpToBMsA9>x$wtr_kIljW1pq*901lc3JUv-C-b
zBzzLKIE0W|wENoK!><D$+I<DuG&~jC{c1E+I<HpzlFAS)R<=YVb|m?LW3}Gyf3QRd
zf$)jcXRgpEHR%f)UQI>Kt4Z3;XZR#E^LhBq#YXMT@Cfil)TtB6AhO6(F?1^4J^Xb<
zfKG*ir;wg<6?7saq7|OdiC{Gqt8Iwgs;2%E0IsbNPR)_DR)ith32(rBFq?YU^Kb{i
z$DZBz6So4M;Y-1TkcMP6@F>E%A#98A@M;>IqH*5~;V?udT@N-S3oVU{KvGkEE;pSZ
zlkG$e8F`hcEO9_Pfc|h+GVX;V)YLXOFjlRI1G@%0(Q^FJeZ*cJ-3s<+F?Ya_^qan|
z45NJq_7V~y0qw##_IV&t`xp-~II=lHo_j$NZRnu!CfOtV5F8CGCVOP-h3pIZ3Ah3j
zZxD+K2)(bLgp2!0IJ=*O<4p+|3hyzl>L=k#QlfNtpOmOf>;lD$>6^^RsD47$#ZdMb
zzX64ZFGY*L7IPYf{&J2q=q1bLj*_v6F)f2`Xa^EyQ0j|;I^XBnw4AGsY>LdtRK|-^
za=b_pp7x+|eFp-gEUhBY(;QAvgIuhM&hO{sjuQ1QD3;C7?k8&tDNz=Cyq|>Mni7P~
z-)F9(w7I>XgqnU5t}!Lh=EnW~ButBy5aQUG3o;Rx8l9yPrZ`cJX7>}<LrRnh-T?(B
z5E(eLAAf5<{y&@iLfr$~V>7&h{8L%zR#D$d{%|{LmDAfCPE?niiuP`|(jMb#u%HUz
zkN_sdI4@z=pVu%b6_dI-JoL%Z&i<0BqN@s!bkLY+;#dzi@v}@k#lnv_aqJG5;Rz<5
zYT=*bv@ztRTlo7Xo@wE)m^c>Q&G=hP9P0`u{-lZLTlnuxywJjbX5tu^nEcHqUSZ+&
zCSGgdr6%5J;d6kaPi;bhW8y8$6AE95(vK8<-+uZW41hFwNbF6t9baJiI>rn9qC<d-
zuqcjAuIBe~jA&mgeqpucc<_0H*E_tyYw^3$k%u?Nnm_Jikp6Ohj#|@>PK0NI7VtRG
z>G<8mhjIAL;=@Jr)yZRean5J*nCIi}JM@)~{^eE(y7k5T4t+act;X4*$!E5S@ZNaA
z6%Q^FpA%?rf*>YXM1dNDe}6=t0eN4V^1f!&uSB>Z1njrueJxN!2-t7Q`x>Y`w)D+>
z^~iJcJ?arG(QkKRwuoOlYG!+{y5xAQsQC1Ez`^WxsH2|8Bl@w-_~#Mh*1vnRdK!{*
z^fy8`!lXESg%eD@sG%5{?xqLCqBKq~4acSvR_}Ra=}`A*EaCr^OZboTB+(O|xK|Ke
ztlv+@gPh-hXA^lg=&ylm(#E)*o<Po~M38RYkf`qh&9^Zz4)dq>cwbg~U8mk5s=_t>
z*N<=zrIBT-A4JEirnN*~B&wx1rP6`B=svL3k`q|ZO=<Iuvxm2`0ova<!>SjSB#rxz
z!93mevf1^|iI>ghBBmC^iED}sKII~A{2VQw$Bgo=wD8;s241A^2IQJm4ptM!;#8n|
z_n8S;I%@SnKr!NEJdbsRJpJSD)9Kx8{!POk4jD9@bg2OJ27b#-;N&BFcx~JxO$6eY
z$E(&#8P)-D1n>kd-u?Phj4w{I$*afdkHi(4-Xf&o$j8lE;8#LMz<l)z3u9{EsgE2i
z#Ga6&8{*W4gnh?7-1g%h=J@bGJMZBQY!<nJ6<%OI@WKIGtcG_aw*uzrJ;vX8;sc8+
zSY{B#mlJrX7eP7dV`~LZPQqFd$J~#68)bwAJk$V;kFckMS6x^P7#V7O@l8CwgB25F
zG=|n#q{LR@KD@j`O1z+miHt?s2ZMXCYKH#ABE2O^!z|UY*OhT_g9DQJ7FQl#oJNT;
zGv2^*=Is(ss=B0`IQH(a|B9pRg7+h1B<h!01-~b>-tG-vnTSJV2o@F0<U44bG!$9{
zPK&%~5ZWy2;e(Wkvps>Ld@dVtt2UMOk#r*onPI)72E2IejTHtnwQ<E9`8M(?pHCHh
z;u^jSOZ-?m;QU)t9T@&%Qa=vs`UT@~(#<yxN8J*~VRLd^Ed^?*E>sY!*ghDURWwik
z%%5aj2JIrpHKqdLLX^{Et_F=&myP3kjI3*(_MwPJ8UE8SW@Yp38?%Pu5ZTk7=n^ck
zp<B9f)LEPh8O~*gS2v-a^Q4F++FV`V1Tu#fueB12K&iBUt8qW$Jlnm)-lN#(XE@KU
z={P3IZMd+uJ}fTg{Kq&(elAWF)t`g&LtNegLpyfZ_F??T^D?{nc14y^|9QsC-b07F
zix*w#)j#vLeU|PEmdOKKkxq=ey~#8vU(CUB9_spMtVn*OMvdFv$Fj;kV+lpZ?Zy%|
z1mCcu<@tu~-53Voyd?ty!(3Vd_I{!cW}59WFhr~D5F3MSA*cRa+KyeH#O-MN<h05i
zkA%awk$_dZO}}{Myf{tGJ)l+YoP||4?}iUi$Pcq<^bYT#cezQFhPTxXzO&wDoTgQ_
zGB*9ql7pU&jDmIbUEazAeAM(F5eGfOKykMFDz^)^+i@li+-)B@eaB$u;$pEg^e*lL
z#q*}MdMi6j@&7`l%KfHz9BrZacOf1}k}gLhKDu5q^ZzQ#zRI>3`Pwo2As>Z@UC-}M
ze^kcrRDb7ptbgivtUubxcWVF4?^u8LcdWnm9qXU}|4V<<qrjou9@A_K)s?O4Ut2XD
zOPuX9E9ffi>#=?7X;{X<tmzB*>b$Hk)Rz0vZJ<Lv>`QwOdp?QS$U3}<Yk}-7X2O~0
zrG3ORv3Y?L^8!W5o(;z#%zOKhYIK}8jPq<jpR3?7SN093dp0DQ&*$85yk`SOZKMym
zVQ3zXiFtR0^3q=M;*hB~cv+k;Yrj{Wz7tziY72YeJpE8!+DqPD`z{NP&B{0^<}2tf
z*o?7~`YQIBE@lsb=g#cF_?GOzl(iWcr_2Zos%PSqm3nWRaY{ySW$W<yCy4_N4qw(g
zi%!woeR2D`&)|9iANO?Qtp_<@@dOH<z|j|kYjLl~ZOhijw_w0G(UcfBWo5RGoy+ic
zEo-NGM=ko8S;52%UshXvL!N#hFKvf+*C%;#yKLh{>I;911)|rW@3BG-4jFm^<Dc;a
z#y%C-jYUMMZ^GD>`flpGyRx-BP<V~LCGGPJ1B0w*5c3#QgN(5&J^I+Eu+EG`vESAl
zwap#9J}UZOoPWe=A#4qc*O)l6$N2|qb1|#ioX6oXG{(<j{Uj$a(VO*>KZTo-3Ay^k
znA{}#^xKj{90o$Z=*oD2CLY5?^H{N(^PV?ut9nl>`dyq^nX3N$x#Z`p{ysd7=V36Q
z@jMFwihlyXbRx$B;c4Q043Df}2l@95H|8RY-}1nDOf8Uy{d9B<FLDc+2{(5g_Qe_g
zgd81*K=<OyBl<qERJ6k={~JfS2lIkAKbpuRCdlz2L-{NFb{?D3&u1R6Q}=shfIZD;
z;t&;PaEUm-tS%XXlI$L4EQgfP^?b%@y{N%}82sEKMp{Y6Lx17eVvo@Z_P+KLnLnAw
z(fiMz<N!RdigPDyjbiM4$ha9gVlpw{q^Ri-()l+gMcDDbI}HtsGNY$NA9SA;GanM^
z{*UKGu=pWE-va}kAm^X|@qFlA&WF0>d`Q6mv-6>+UuR7z<B{KgV?OjEiu&K04<%xG
zD0)7001Ggj{9t3pn*8jMB-Boe^nxS}G9S9sHNbob#}r-al6`0{{pLj9;+0)<PISD`
ze6TrD($`SVHYa-ZiGMvO+7DI#PtA###prp_L}(b97o7uyr`xexztxlVmb&chfoDah
z_Ma7XKh9Z^IdA$GlcH9z4lpVD6Uc1-ZBG)$Op2B<4kkrbf70h4(7&Q{JtB68niElW
zc-;!8WN<zQA0NPntv2ra%)D@n7{_h8)aJD)VQu}H<`wS^TqIw0!+AvT?BIE>v>m>z
z_tZP4L!K||T^xtaQ>TCGjeAj?M>MSShz8g8_WH+qv$m;sIhpo#-!>ZiVMI-DgIw|E
zR~BaRFRCp|;YfVJ%a(f6_IuasOU(-}5qb}G!MP8KK9K0)xx*=N&7leKX+HhK6+Ak8
zf;k_8iTE;y^?DbaZKH<mskgn?ZEY0i^KM}%&nlulbcF6_Zh1fvi(ehxm+_#=R4p*0
z7Kig{RR=yRG}42aH@+w|(t~;T3^yQV-9uEzEF8JJ(y2PG#a#P``27IWZ7*iqn9yzK
z9$h2PRY26O5Y;F}xuqz#6y@$Ost}^`AxeWNC<al@t^cc{aE?&DV+a=-(9h#6VG|61
z?x`IE5C1ImPh+5Ol>-Xrj6vpe98fuDOg^8Wbj}zUAjbi<6USiLVT=RI4r82Hb{GS-
z=Q?mJa6oyV1KKCzI9z+9|3-XBVvHfuSnD9fqQ;orgd<xV0s`-lxPZXdOI$#t+K89>
z#w-<)SB_aF;Oa3C38vK!6Rva!`p9Jt!3Qm2y7or3r3q}*L9EB=YaN8tfs20BK|tVc
zi3<okUE%@)?_qLKLlN;;j&TUMdW=)>t#+7jr9;r6hTwyI#=|SCOD#>F8M7NPH2oO|
zAvNg`5?UMr0$(q20f9dxaRGr>h<MA!Gzz$K%u)eYk69)7Ry$0%(jn;3MDRfq#?x+p
zu8Sb+6^uWytzm0p*FA@NTb101Qw$g+bYopJGCqxNU)M56^7To!FK_rIvwdlCSlE(B
zB)?jVc8(<uuRg-7S0{U~+t!HF41=DMJQwW7l0VDy)b^!}qfMWAQktt7)>6p+e{(h9
zzY8GZ*vx+yKx}JbgDgO$e#aB=|1N;YHOWEjuYJe%|G%;TlF@2g4Z-~OU*UR3G{0?p
zr@tx=V4Hg<{BWTI-dVwT=z4YO_t4GzO1rIt(wdI*NZ0>~bBlJIUcWr(dMr<$FTU0Q
zS3TNgeE0{9cJ=wmIM1US_Wdjjmu{UL{(E+N;$sl0=-b-%o*uVt*SBC?@}_RwwkNa<
zoeDN@A__o}WGWJ>M}WTP(4OY-k!LZ-4R7R4z5MmKe_~MQ-OvjR9Ow^s`ObJg<DmD9
zU24mtAmF7!Z(0Xpsh97Idz-KPHh+i>&~5m9#NGs)#tv`uAA=VT`1uHbikg~q-${y+
zjls!dUx@n$o0z=D*PG{lKREtT%p&lv!LYx8h>_2WzPN88TD&{1Uj9<t-*MJm!$vd4
zL=ZI#qQnF?vfi(b%*9wCw+Fv2{5qjWD|)`3hL7^}ZTz%F)&aHUE{Mj?+o4^}UmcOT
z+%-MuJ2zh6f%DiHSSI2vg>iUO&wmQuWAq<KYclIF;=qs1UnSOEh_@pj^8~Mm)5PoA
z<VSsJkK;qT=dfwI>%YJYTZi!mNl){0hnKv3TbI7A`P;aLQ9ivVX02wkflume`V6;?
z=R?AVZ}4PP{~mqsF@NNFC%GQAzT??q`-0v2u4tU!hY$-rI9`Z@cG!5uFn)(|0$!rm
zvUdB=!k`sz8143-tOe5f_8Znd4jHc_8Gbtq?+2ouk(|6~I?i1%>}iCV+Ttq}`0xUl
zc7w^%2k#4iZv1VOkgoo`6|dYzmrq}Nz<LOWJ(%@qdasw`k1V{_>`(IItxCLhH`nFZ
zn1YOBCWTLc82|VklEoI>uo!s=#_<K?RcM8`9kqbhIGB^{Oj4I*KsJUkYD+2;!JB!>
z-b;`@Y=og~hv2OCFg>0NmpEDNvR1EhP?-bcF{JABr5Rkq!oKweQE+Nb2ty{UVX-2&
zg|LRzmfY62PKCW411nBXLtn)|-bUwyilbe;*Ttry7^|r$Hr9}gP%2M9;!FDkFDZK4
z!blGFwd*i8qM)G<8sa^WFy7f=9<g8kAeR$wemaqJHL*E04My?mV_kVXLyu+1N4!{1
zz)6ck!hD>A-hwzNq3r<t!p}~OBT0S@5kqZ2jAsxMx)x0?)I~J|NtaD^!sZa81%}_o
zXMSN$_hF6@NV-9WIk~Hr+VglZLUG72>Vp#nBFw?Bi=eoQjlUxSL`}4bRieZ~6w6t#
zHyWQpGaj#l)%&(8kMfa<7BFJ#p1}5z3U&D+;YFSjpENaIe*pO_BZbrPq>4+CEt$pD
ze70qAr3^bt7X8!EXa@P($iV8{Hk=#BP0-QsL6sN_{fSu)B)vDKPpfzM9MLM;LOUn{
zWjGJftoPZ)*QPoB(e%HgE{PW#%?<nU;DVz(TYQoQGA|KJyx7T2(1LD<re|Yro$WAp
zb9>fcafqFgaS&-B)?+7g;)T{T&Fr*>dhymKqTyY+WWFzFM%#dB?MX$`oSa8tx`7Yu
z0<ID@z<hg&=MGkYw&{YO!ig=u9~mg!mKStwX8@ZA91GvPxD78D@TdsZAI(N%|4z*m
zE<9v?sJ6U_2;Apy60pG4Xo@|PVj+)YDerv9!%7K0;0Q0rhcktjdq!;4leZJR2LG=>
z_nD{~p+BN0qGZlX8QcLV9G|fbJPcVQLY@*FMmkFZx|-~ET{W&;Af!pK+e79u#1f>x
zV|<E*HZ~MAq5@aNrc4^BP$X8s@74;l@V^a+nj6#B)FoGP*pT(U+Ol3GbhE3FV)9Kf
zBc&L8a8-yo17c`|RLdHt3u~Z}GOx)fJma0;!W5^|6k~)bWEcz9w8OJth^UX3d5ev4
zG&{TigxK@>0M{)*hC+7o29jm=>9_ZOWoBQa-tJA)nEfA$G;l~l>RmZ3l6|n*rSYu*
zk^O_nJ}Y*b{S|$h6$e7tc?r33QZBnV*0zz5JD@#MXpdfqG<N>Rvd4|su%bW_)`Kw9
z9*=^-$`4;|PBBiQ)sb;9+Ex7ws*((a>|}9{#HtQ_Iofy=<18(}H>1Iwir2TX=7oce
z#n>ckPx|^2*~rAZjqYEQ)ogCjVt>i%aQjwz1bR{Nxk<azQd_t+WB(w}c;N7(1fk#K
z7R~+WQMkE=5zW2kDBOX5+&?=C_pSZ7?>q|k_5HXPABDTDA9vMJxUcHRU33)g@Au=L
zeH89X`*G(Ug`0&M`6QCMJR}^*Q{uDV#%I4mNB24?Hioyb<j?Mh;~?WjGVB1w#;_4b
z`XdZ$$?y;;Hijp-cy37ukYNodHinT97-9G^8N$&z9BXIrlA$-6VV9jDl?<Cfu_eK8
zjYg6fN`}8gGl>5!U@7_0FToIu&c)B|+1y5kW>7*`(D)b@phtDAMjGa05*|a7Wzay~
zQOgdoINdi6P+Qo<2WwJqAS}c;aR&6`E#^wMUD$I+5ysXS`gNbMXs_6k-cK`r>~Ap5
zZbw)*M(tZdf5MYcZ-5@Hg|o)!RzB}2Li0>fq2JTpLv)8FI<}u^`Ex(H96}AIYz@M?
z^C|I@K8ewuwxzS5#6KKG;!jOEp-%xf=c4;Dr(9LA?uH++H~QumqEdu-wt)$MD`6!f
z*y`ugl!NNG8^OkcRj55Qr;Xv4lEDFJ%pe2DxHbm*DoSXjgtN)8z|LTfA-foujKBd$
z7cCc3#-M{!I(O;gqsGc%2xZC!g+GG)hW6qqqMDg<3L+bK(-{4F^_EQ(@E<2@4x645
z|GdDM1j*ftW76cP#yDA~8P#(hZ45qeHqBhDqr{GTnTyEBN51;GY0fpwg{^n~pQiB;
z(<oya@F~wgdrVm)>l60(V>$Z7`2?J(@nWmB^m}rAfm!ONM_BdpA!Al>{QDfk;B=ap
z1aka{4d($y7=MDaysS_C9*Rjf5@<F~)0$e`{O=gdId=At!@2ml5KrUq=*c_AH^@Ez
z$)iL+1=^<#CYAV@<kM7U3j)3$oFtUNoMDggvK--or*A~lFcaGfp1{Q$JFpC~fol=Q
zk3NSYw{K-;BQv__3EiZ==8g|4Vd9(G%`s}<6l)-cNL5jn^K=I5ZwHbvx>ut9esLtK
z^Qf>h>>$IHpmbj%yj|>2O?(73D*4O%>z9qF-F_KA$^!mtqvB5NANP9^)o;5X(%>gS
zD_V6K%-k0~N_;TKI8g@HAdpN80OfulzYn<v`4)p+V+0vc=*hCsS0KbH^kiG14}m~i
zp&$AgqTRlga^!+dq`t<k-~20FEH<G6;?TCybq`*&)6{c)E(ZbB;%j$5#;_Z2^WhEI
zttnepTrmgk{PV!X%{x%ZGFpwdehO6r^^VPNnfMF91NF(9drb6MpqpDUwcNxxiSZdw
zeVZU)fus}Zp-6D({-ntI7-qaiGU@mTA~u(MGQ!45KyBgE#4I-SIYi0#gT%jY;*t#8
z=3(<Ah~Bnuld$e`YlAn4L403(7FD3W2M3FdpF%DdS=j!O1$?C^Y}_JgqPGrAB<aVJ
z#IgY!7IMPIm6CP`Xx`?nPA!=2#A4%Z_d{*Ww%@{nn=cLf>o^901z!Io?t^2=uDLg%
zq0k$c<f4pGLWch`#K*Y-lXe7|!sa$kY4y=QF==Pvn+x5iacLCtyH6lTw~D)?`xx9)
zx)TtNZ`t)VS(ps#W7hl1@+4sJhGcVT01E=`c(IoH2a8g%zC1G(giI1J-7-EzX?f9c
z;6Fy;od9t<Y^K{AxI7gfH%RAKlf)NB@Xhbn7U5rh?J;tnrW=~+<|kCJ7<{`|zdT)@
z%J8($bW$vL?bG;Cl$Y)PwRy@<;~GxGf4^!^yjO7H1@Elq>I`djm2ZJ<j=u-}{Ak?F
zFB)8eC*$}PVznLh#az{Re>pXfTcoc6p^aJJqSrJ3ReShP!}B*^d$*+=v$^38Dz^yf
zoA$Sdr!o9xhF>{g_<0PU@iT;96bb)N^=U`Jq14-X$2#*0?tj?r9kttm`q8((%XZCw
z=E6gSM-bVE>L0iXpZYkQmxh{!&sTkmGb^{CTIAx43pNX|+?us*(d>*jd3F!|9adoZ
z9mRhzE)JE>Ljj?+<z*ehN150e;?vdd8PBlTAjK@S33sy)@lKIg3(a-p1<RAW8@<5>
zw1`~vS~w8%HTEy~g?W6QtNsiwK4<m%Gx);abe6HHqf{%o`xG9Ggf<t*)J_8ro6Fgn
z-iuXfeB=r5`QyytL2ue2{6{C83&lUW`3g#OCx8tf0*uFpM9^Q4M+=QqLHX|=3=f3y
zeMG#PekELu=L(!z^%VGmXQ#WI_?LFU^EALYzO)Zvk+yHq@;^;BZdnE+K%vvD^>!b>
zPI52@GQy+ZLTirGvbr1aSs9%ByvF)h6DYBWKmU9SS`Pn*Cphp3FN$v5cConaKSfE;
z#xK{#f=AML#nMS@!l$AYH-1I<$*X6Y`Ovdj5?UoxW|vf#)hu+*sjaMb=9S?8-8VW@
z>jbT+tb9>rX?fbQ1M<wTY%DKx)|Z!;#V}l1UW)%xZwYY@D8%B45;C)-p}xMdq}tSF
zKo+U2uLS??ytLezT~k%lQ0KH*O||=_I0$!C+E**%m1~Ys<~VOsZc(nBqD;L>shUt)
zQC@modELC4lDe`>FLhQ;C|XcgS$5vkY?D5n^wj!Fe|efRVZ!<NFY4!4nTnl16MRJy
zDXJ|mzpl3Cy7CDX%A%UeGA9EkPbPH^;-=Qu%=@wPobk@mn(BJLQz|=U5vWs1uJ?m^
zk+A6${6~1u(wtW_)Oo3tR5_=#q^hKDiYYe+Gi@Wo7G>8gtgR|<oH5=xajFx45dWV_
z68e;ya;D6$swwe9iYe_H{O9=_@jvKAmeS6hRbRQV%n8FuDwR%C%&bhFT$G3UHNT{x
z%3oAcTU%aTrj$4GU)B4|%O)^FYTBij3JI2Ng-s<-v?&*s*DWYd6*h@6n13-Omn<xI
zGH~+bdSR#{Jj0&<yOz90sj8@FX6q{IEsY1P4JhXMHFXP1=2gX12wOm^Sr^O#<$Zcd
zeYx}8@v^}CC}~P+YH9<j#YK~f{AuSD)s&U^ORi48CheT`#u;f)C;h0p%!62%(oDyy
zc_tQ9jiRgA(JPg6>g35%SgI+^rttq>-MvQ1o1)xu>@mkEvL0C;&4dnY*YZkFIXZOC
zHp9uwYt|%cn`x;W<9Ds_3o^1x>ip$&S7=iU8>%ZyOW^gu-T6eZkh5;NTR>A@Mmhx+
zRMeH1l$jzezAnipm~mV8RIzwmVWYE$=#+STvItgsE!;%=-Pg(dJ8Rq`EL}pkg!vK<
zod1l8)Hjkxn-%gc{iJ-$&XxjH&;BF5sbzwvy86;fGelJ@s;sW`Pbe)_%v#>h!7=5@
zlX=fE9mE{?kyKRj@lM>*l=IG8SX+tv0&oDp6)E`ng>CC2z9Yl#Y$+_iq^`c)SF*UK
z!Jk@MQBvnT7Y#~C?WaME7XcHU1=1=jpVzS9y!y%o)g@K+eYn32LY3^2h6NRVcsqtV
z&n`PwaXJ^4FLaiamDQEk*E`Rys%j)nUZpi<<$|lIwyvfW3^jH2SC?HwE5W@+s+8H6
z*eONTHPz*`pJ}K_A=xY9<jDh)Q~PHX9=Wem8Zgr2$#dYI3u<PUmoA!7SF;eF7QL8d
zyeLGJ&wna!dijFN>L|^alls}^*Ef{c`-^D%@o?b(t7#$ybcF1EZQX0XWI4W%J0;<g
z95le0cwTo028?WscVZBTO6~+)g@2H+29W=uycMt*;p^@Shj)_?>qm|R@Nr!mP{C$Z
zEnqq}!d3t_VgIrVFduuwy`<w{3IAtSJ`NsL05&}u4zB@p<FIE3VC`dg<1-2Jp9qKf
zpIv*N0v}-4U&7%g!1TX`!w&(rV$Guq(1A&uG6eFT4Tqh8O`F1DH(@K{0d{VMe84r^
zpfBn8#7qxh{tMx7D)!r2@%g1<z^+}$C*Ydb!r@N9rZ;hL0dNH%8~Yl-6u@<W>42?(
zjkM1@;qW@ZbifWkH((dwI>6*(kUpRj(D`0CTu6K`>_WT;{D4h>2H*-ng?0u^0bB=|
z4%h{#0T%Cvo{(D$$p0k2;sDYGT!WRb6s)wZ155{O1=IjL0gC~<02={&09ODyzJUDz
z`F|B#0rLTO0~Q1J0M-H~qnw)noq#I<GXd8C<^!$+?4n*@qC9}-A4EO?d13!wTyZzf
zR^l2Rai<I$n!F-zsQ9-nV-Zr=5)Pj$c?bsPR}>e*$e+@LG*M@i5sn!nQogSqcSCZM
zGVSC`&YkEw6I8)Nm~dmhPoy|BK{?5-B$(+c$|QtQRwvS(iEaACmNjn0Gr}=_M9TCL
z&gmmkr;kYYjL4ihqUpHiq05E@lkQGjo^WTph&LDU46JsHMqNo<fB!)7az^AwMH6YQ
zWO`UT-;D`Q;s=N~eT2g^B4y@?*3{;NyV+n&y`KSGiVdQ1<a0kD<b&yZ&0wM4Z{fG<
zSNO6c@}Icc43E(}XGBw6lmeDL6Qb;I+L`uT^_y^bCm?NLrIUm7^U+p!F`b@4(peLo
z4)wbk@p{p=<DhS1@$Uw<&x-!>9z@)(2gBjB(bfkj3(p95RF=SGmLcu2`=M}HeDNGM
zHErP;k?I+d-kh{7@$Q7>@zg63dey?WoB{an%4a+u@=@_Hd<X5&G)R7GqjejPcz&h_
zA9f*&Hli2m7WA4qVqIMGkY!0hp`BR{4}!l5KCTDs*2Oyd4CsaH!r^SdJ~{pK_W|QS
z9uD8Z^aiTSt<g4T8Z7&<;NSfu4s=TSsFR)%PP0lfd<w#o;bW)v<%8iubB50aJr%xJ
z*wxHuB%J!+jPT6$;qX_KAJGf?H(RaBtk+M$zK)H^C(1vu<9F^pUR!%ymk?!Wmd{&E
ze^WTT1J47*>DLy?Hkv=A*NS+rvcAcS^2UR{bW1q=VwAoS9R{q2jOT|OC;F0F#7i9e
z;6VEK(}nTYB3|Kk_-f|6aKLz|F8%ZU59sMepY=5LkJ$C-dNSX`z~8Ym9Da@Q<VCs*
z_=;D;VTg@fq)!2T=>YTs(3=LJSApI*0KEnDb+@4JpmgX<K>4dd?=`|<xZ%h(2~SUf
z?uMMJSnvLW{C$de6|aWFaL<v8=_F!IP#i;dfj%CLg`iiEXD#e+`sEyW56dkJJ^AnF
z*MM&%r9vOePgWuRL!i%s;S#?<{QkPaefKZF2NADzcR2hY`qo74_tEi$kLZf0kC0st
zQ#JiR;$?n-l|a}faorz;z7hW~;x&nQmTzG^w$nt!%l{}Go=pOD6h4Ty#YOwL5WWK8
ztoGJL`V`QY4nQvey=eeC?a>H2&kZX%Bb)?GzeUQwfdr&)`5APpSbtI!+C%v`96p%}
z#Q00QuSBTO_a99Erf|591jr+xoWmHK?S`Cqw!>b#eWHCP`B~4#g1;jKeVK0mvKz3y
z(=G*w*Mxc-Yp*KM8wa4bfL=QQeKqJ61JIuWy?6ln3!oQ*j?<Bmd{dtfB!4WO_MrS&
z`e=?B2arD=^nvur0ev8S=7LUr?CqEO_(9JHJ=UKr16>;+-3LKWzcJcB<Ur0aj42CI
zKIM33y;vTfJ0j)ogxnF%U}Elw)MZK9i1a&$Xd^P4hvtoNSB=O-8$d~maU4c-7!M?i
z`ceQnO^_q2Jf(>%obcy6nv{e`1pV(8l@BN4mpcr%_wbvJ-zm(a3k9AFC@!AEI2%v1
z$#yr8;ebU%6v#u0N)?e4fH0(*NQp!^NLJWe9}oCTB0mRG2sn<&g^7v+m!qUI(Oe)D
zu$CeFWXNQOgn_VVzh=luh;Rbn<3yGMIUS)QWD;~a75aTnDIAuc2Pi_m1o8<GaqU5{
ziYLML8IZGNfHD$^Q5|7}a6($b_ZYI%3g$yGiaL(1cmh)W2#JX+obXx`>^qAIGpF+z
z;KYyVTt{REerMx%IgtbS(QXd`iQ~C7BG(e(R|swfA}&1RvSh=BiZXtH#MAMk>^~Cm
z;K!6dATkR-iu(q?xQ`|BIg$OiO~y}Lmh4Vw-2w`?4fq8Er3@jcrONk+aGW~|a3YcW
zfaKsO*jUAAsv!(X#BC;^xWWma&=wU${%n%W2Vo1cKZ{?L6_N&t)OR~WI7RpyU<r|2
z+=L^=v+@iOrYG3ifCyj6kZ{6kCE?^OJn`%jXCA_?KV%rMaKaX-<0-=<^K~IbXCob|
z^EIF{8@D$A#WhZ*BiPcT)1j@VfQz>FB;s}o659dz6=fa+F$@{bkQN}90Di-e%`)U%
zhP(iT{}o(Z;e@|7LAgn^K9%DcNTnUW2K>Y|1r+9c217i!oo~{aFH0ZB97Ear@Y{~x
zPNvEOan$EMA_Yiw8ek_8{_`=02)DHcNF#;wlD6XGc;0yNf!xfFLNl|p#FdRZYezA$
zEAV607ni9UAM;Iu`M{#R73nE_`M=BxVf$cx79q=lFu$y9%1YdLj8<IXg!kyf_QcZ@
zP>0;1xKTcld?Gvpz|gCSY{G90e#Jx%;ddf_xwO>?h@`4BiCh4L?!?P<et;VVlw(ao
zv|EN;PPV5I!Yb#cw5>qs{vRR28b@jCh#ZI9G31wwb2<>ld4Px;2#v*(5c3JDvykc3
z%aDi3_N0s>(%Axp(t0WSbs$XVTOy~R92s(ean1+AbiN?MB?G2Y2B=iw#&jB(PO}U-
zOtvi&$zjN=KwhF)Ua}5wIKvcpoLGz@n0AoJ!}t;5kwc{oKh_R0lc0mViBw)@yPSw8
zo>6$4r0eke1t0!<{nrBjwZMNZ@Lvo3*8=~qSipITTd2AQi|o8wC3K%`;))Cxfw)=y
zU@ZI_M{2h4mtt_P?eM~sMO^7)+yb(f<@JT`7Vso-M>qUvv3NQ~;!X)Uj^Sm<aCYdt
z*!%PP@<_Oba8<mUK#aP?!!d3FILYB<#JdS1!#;-$;CzHviV4IZN<IpJF@<@pD;5yL
z7jbpT{Bdx>%NloZoWd)Aq+39X-smuKOYhwx6yt3Pt@MwCWqwqXALoxn{$z3rwvq(y
zkl`}=cMf``{bLiFFUxh2gm+4KuY`Y)@UIeXm+&<SKa%jEghNg=^F2nwb0oY-!Wj}?
zDdBtx7fE=hg!fAL2MPZw;dTjMlkg)64@x*BMW!#|ITBtZ;S33{lyJU;izK{L!h0qB
zgM@#TaJz)BN%)b32PGUL+v6As&ynyV31>)nrG)b(TqNP065cCeY<bPh&YtW{y<%QN
zwZFkRX+q|N3(m{PY7l5fnn_S(XlYeR^#bRj^1AxUnri372^UUCKW~!Bdcm#!x`ulH
z;tQOq{*ndt=g-dd<$C7i7EQb`{rvfrI9HJ7Oy#MaGACmCd{^d~(bKTtP``MgzhoYu
zzs`ge))RK+%j;^D3Dq_J@(G^l-t!<@VhgGpCd_N7tSUPXXDKESt0<|jP$rZuu7*4l
z`s>UPsc;lhgs{5usuD(!PqkHkWde_QC=>iRuBA+X-jxUwXOol(<rPKq>u@BisG<yU
zEet|SON+`IOUrAq<IlrUri`MJy1J6ZrVQ)($EApf^hy?1mLgsalE7_(NSEiIlnL|d
zaR#dthX~56{jufrKgCZ$MEYvWPi4ltfnoL{36}p9MBFUDZsBL)fj-NMUl8vGCMSG=
zP9nC;@=F#e0}uB2{&=?t@=Jj(f%V;D<L4j<+tD>v`m5sIf_arBaHh<Qvuj?KKQ%FP
z;fOxnif{RSLlRP?-zQ++t@xJzy%IF~d@H^p`w`1gTm1%D#1(75YeA*Yw&GiU+M$~K
zcbFiezr|<4>k!U9z`|EaKfg-G?~?JUKW%5}FUiV{l5WMf`kh8jt#R!(K_q=EzLozL
zgj22+-|ENm#b%14Sp6RD&%Ql2{@tL_CeqA`)$gZs5d+uBCWzdbmi^L!*yHERej#7R
z7iS4<5oER^3u!B3;#=dKHFBUayVe#Nk5+t(`pcO3);Q%0Du&DIuT9LfmGw-x20!X=
z*-!b28yK#inBX@t0%FsD2yFKFYkuYy5!d|8jL$PZ_AAzYe?_1@zVUOnU^jkl##+r~
zCfj9|pG9iLgFQZvTkvXGBY?8fDp(U0bwxf3@G|Zr@UzBE*0`?YQIpGQg2+8K{dW+;
zs&2)%#(jp2AIl$$e}cHw--_QO$1OeoX8e7K%Pd;)t#MDUjBgdJWd<vKs~me{((iq&
z7!)jiJtjb;y_i^E`?2JFivYX+iX3+;>&zJI8ECt#corOjaC`jqj~5H>^^cp8EIlm|
zRy+$IhRCE@`R|hPJ7j#PHK3OfoK{2$N6PfmW9U}=+hXX{m6rwSW~{VDfV|Co!h8^W
zj<x_Ro<$-a(?mp2>eD8Nc7h2{1JLlZ@^AS+HqU|5ukAGBDW^&(L2UZbq0v|Za+w+L
zjwS+2fW}roYY$97h}C~a(-R}}ooISeWS$dEM{TwGF<U?adTOhGvXPN$vSRf+(ez^!
zt3QdR9~+tPMAMIp`2T47Fvarg(e&Yp<*%dZBO>#~XgWN_PuzBrcus&<xBR9(FhLoK
z{}40~|0wvkf#}Ca=B3ecMn~qO(ex7(%il!P$0(K`iKd?jzff!^iRT2)My+;l3&qJe
z#cIzsGRo^@#cHR~bdCn2FV@#YWq&xDYOa2B$vxH%+|nSO$LH)<hCoIsJZ5FTawMI{
z=Ij^8Rg}ZyLiVdtAfpr>!?9mCOFEC)*stZ1?tZ{+CoA_$`sx_^lak&OL*FLpJoaY4
zUI!A7y#T8oI%VJiz&P0X!M_v>N}Qy};b9<qTKe#qiv6<m;qepu#oZj*hsR6oms8St
ze8hfn!-slW>o;{Wa2Dv)ldn|sviusir%31f)^@s;uU1JH$$>$~!$A2OE@F>TcwEDN
ztph*xvDT5CGSHIW8zZ0ldIQP-0Q5Ml!o;?t{Zjsllil`Y6@K?v*dc~~lBCD3XPpl^
z?XV`M9C>Cg4qHXmc*9DhT*`6AtczVQ`1|{Z!ScoP*m3YbO>3g{|Bayc&o9#-$UYAX
z{{H@<6?ATB#p=IH(yehAxp;jb<n;GTdqH;&;AeV4A7w_3x}Sy4aumwF=`OdRqnN~1
z2>O|z7nisN&NFkomJ5FK*?#|>l+zO<=fy$z&w;#w^tp5p`VT-q10zf8eG{u)@C+R3
z)_W?}_=soVn6Dn)Ez}fd#M8<_<Ub<$i(~5H6GDEUe{K@e6lMP)a>gVLoUaQ6eYnyo
z?PJZOZUJ57Yq?uUHvKi`A)q^DT6XZ*AoMMu4`k0*L1#HSvfV<A8|A@k-VmmHvQqp+
zvB2}i-8>sMkev14ABetV5c-FM(8pkLf#u>{>J}+-uEEO(I`!<lz%6i|x#smp!EZj>
z?`?wKe_YWn<>%k!7VK7i=Gm-)^gjaw$bsl?&{;0|vRpD{;#UZM_(`e1C4Y^i=hvC(
zavaR-c}cIm-7RpQCFJ!!=*(|zp<CcP#9dJqVE{XjK73!{%=qC-*A$cgILZG<(jmus
z-%6TIc~jEY%`*AvA9<a0tSP@qu9uorMVTt;7H+|xgYL8t8+;se+Q)id%xZW40G;}*
zNplNsp55m)6bg?7zct=N^AXn=&{-eS?{o{CX9;;-Cg|p~{r+RnnXdKTnq}wP2I2oH
z=uW!;`TPjzjsg7SdLd^x-mP&9POF{reT;$1Ws;zy9AoNR3FyKO8K#_ZGGEIj-Fh$1
zDzD8_j$^6GFWHqZ1YM{SbsL3_oOW9gqvvIyvt75!c3miA-6HvmWAuDj@S~ouFw?c_
z`8Lq0{|e2d_eeQspusX-B}UJD(8=E;>!*0P0CAp_{H+b99G=1D^|9o4Uh5XPIJtqR
zDX1cpW4$l<l{7>J=qz{ZJ;u=@mZCI)K2W)=0exirNlM4##gfwmYX{+ff&8$W^GTDR
zpKs#z{viBc4?;f?4T0&l#-!^4opx)|-9m1;xZ_NVr02Jo^z$Wsro}JmH%R&opo{u1
z%j-l*zen;rCBIdF9v5`UWCQPkF6><67F4Sq4nu>YJ+1dyttjI_Cx3bjf2pK9J*J!u
zGQv|r4*dD`CSAG#<s~WKl54^7NR;WuULFvIo)@`A&`!y4m82_6OggHQxN0r_0+Y^n
z9C$q_={-emf%Dw~Ui(33xwJOB1wLEcaTx1()6Uj=r&hZ-7j){AFYCWJ%ZxZPL1+D0
zC&#6hf13|FlIV9e2|Cif-Yvo``>Y1t0Xebt_Mt)KYyq9^-ThRtkYly?-a+_N&~T`K
zah{oOp0vZ2lHMiD@eE0C0{#E1<!X8yN6K&z_Z26^hlKVApeA<G>14U|OxiPeCfyUo
z*<p7i^sd`>((bY2jO}E)_cYpHfcOP_<A{WWxB)i~>|u`_kl-h901iki@w`>#ia)z&
zBqr*KGOx?!s<+;6S3Pax{}t#TiYHdTe%;ajkAe>QsmbR*Ko|UcB*!Dp@Sw{tQ0L)w
z=_@`L^7$3$-2U`hwlB_BeEYkS4`0^!Fae$W*`~5RpB>=O7Ifxw?+>4e{}*37)lctt
zlKw_r_>Jo4GbR7xtOTxo1v-!CpYv1Evf$g#pqOoZ`vvIS{!Jj;i!&v-^KXj&9VO32
zo$LqSlXAsaNaBOIfj>V5y{&(@gZ@j^&l}axhoTqk3hikB*ADuBKxe;Mm(OpC_O<tx
zKSJGJ@Mm-?p2%n1`15;3KmGJn{J%IG^e@nPy-sU#@&T5L`}xyPWP7p07I%IGdK(|k
z1-%(pe4+aJ`XkxTw^Us8=Z=2<uG+t`WqWZp4A1=-rJ1(pW`d6K*Zl049qoVHLH|n!
z{hJQ@_wiz|Z9G2~bi|FX<oK^JcK$r?Xuk%X`TX^hQ_)VGal@~FtA6G`lypAJ$)EoU
zdP|*<PmNQ$S&I|o(`n_UYrlur*cA$n?AIw~gVaK4-e(>+?;ABpg9c^Tpfv9~>Cxla
zvxg66MA`0=Y${KZUO4$|OcW;Kt+cpQKoHa?H*P#^(%N-#KfRFn4l`pU`Yj+$fKXud
z4i`@<7a14CdfP+(Z4={H+Xdu=Mg#EDJj-`L8uEHXPo@i++0K#(=ol#u*k{-RKzU4F
z1n_{Y*Fqxl6Iw^fr5lkCEQ^fX-3^aY9F+HuMw1g0J-0L>e#0-{&J*6ar?)-%hTq@C
z@hxsC{ZC*&RGJb!T0A(n$htVgVCBfwZ?&Y)9zAlbB`qI3oLSt0d9S<htAs2EEJu&d
zKD~HwZclr>9e@-KQegIpfD@6trYP8~^D7F{G)d=Az0=djGyCE6lTT=HG}P<X0rFuS
zKNehacCiFPvF3iLZ^BgxHsMB55G^hq4;^46*-mZ@DzHkHwO=d=?hm^&R_gg=9+^kS
zx~q=opFG5-h%iX4J-0;1W9$eCJ<Jn*t??bWvF(9wFw6x3R42?c+p`=OU)I#H7Ehgw
z$3u3K(K8_A#DJn`u)s;8%nGYyz*%HLULWn!Rv;&J&=7i|FIDhQhya%4k(cd+==3gK
z?<HgZ6oV-EP^{eYIfi>&p5^6k)jh@Pm|6XeQL*rsu1}tu<@;)*ZIXpIgz32v;5+IQ
zl+){24%4i{V*o`$XpFUR@{vS4vP1AbwAdmnojW^26I#c~MhYz}#FITOl>Sa5H_M>B
z+g+Zu53wnHt1yb}3x`e=pbvhVBneV@`S(NvLs_$MV95{E<rqeW!HNhX1)~5Xl7^SN
z9J`imAC2}4?pQ(n*h{jk#j=kfzgRB8@^_NT>T1vnKbqH8D7Xphf=A4bG^dptMh$*E
zb@F46@t+{{5cP9<9s0pW7!Rx3^^sxRr*ekrT!kw@QvrbNCA1=3gpR>8m5Ui=H5MCh
z$6_$dNt_c}Jgqabw+;Yt76VMncW-NR6Z@%(M|TKii;Y?d8|CEVFv*DbkPU4ZQka7(
zHz0sQIfd>~?1l|(SU#7tE@33yGaA#F7Z5koBuN=Wh8byvttujWHzsUHr?LNR%K~TJ
zK>9LkKKH<QHej0hqMcmVc{oP{z5?ML7QzbyTb13|@*)^yvk!^(fZ;}$QR2OX>jDl<
zz!ydcA_Nv0ga%#$ZPMG~ay*NNV`~;4<DX$#_#TWW{M%r)iDA=uHTz`rw=VX%5sM&e
zwRhu^iw$v2DHhuS@sKDYIO!O>{PIB0t_H+p4$9zjXTZnd=FFe);DTVcAy-nULW${D
zkxwT-k#7CQUSZ6$aENcowW@HOHH8Dl6w<zGaW8D|jDa-LiSj})2TOx2XT(c(cq&Xq
zP2ue&!@(6%$?|;4VaihM13Wm94p|pN7pPq|W=+i&8|(}Te?79F!N&UGhIZIZsoCA}
zh}NDCZi}#2@)GW{rl{K3mo#OS26ZpTb8nwHRpOUB9fvWgdLaR-)H40~K6H*r3X78s
z;aj#wUtsZJ$&y=+4LD4wEswHu9o(7dc*t5@YgRUVkul;(2#OE9$uj{hDI=J+r87>-
zaxRQ7B)lZ!9YZmO57Wf;*~9-lNh9A1F<~Lv;RudBNrRfXyFc7A5Wtn;uPkPvAz27h
z2<}0*Z`{jpZx|=|%}!cqJ#5k{0VfewV$&@dM+429j?tGH$ue13+L~M|UH$*tFE1*s
zNY@xOgAcsbagG$xQmi2yVIR?l3N}hcU{#JaTpMoCvcSMPETk1{u4#!XaNO7(wD=GR
zxLERh%<!a?1+T{IU=yUUtM<hN0oG>6OUA1Oi9pClz~RKzB(>+)hoJ2mnsCiV9g)tG
zobkcmYkeLi_QnrxTJ(%pPlb9_>_0dscrRgqT}Wdy`5;)oA{&Au%iy8Ws6#KEeCPpt
zuBxs;paYI3oI>3mIr&6;n1<K)vxKgooY;C@iqARO(hG0`TM5uUzbqxDNNqEuSpxS?
zb&*0iCZMniP(J5OMYE%sB`A+C*h;T<VUT$Nmc|b<`6PBMmE>sxF+qg1cOkHqxgQy0
zJ=&Vq7{jtj31AwOSKjW@fc9aU?_7kDE!tY|$Sikc9R%hO<;w|aF&PY{h2-F8nN(t-
zSpvt;^cER{pTe>a4svG*Q>jo9&rxDvzRK)Id0f~W4vE`xBYr@`KNCDOyXQGSrwF2D
z8hkYAAs>p9ecfLvXP{?yW)`bdOLvM3(!jUp#hiTzBwqFntZAGSivUh5@6^0kYs1DO
zu+~<U))aDwC<73Vg_NpvfTdIAuzFU?$kwLhI_x1!p?_Jq$6Q<hz-K1!4Tp^JTNBWO
zbP9f`wd5R%nMgRfpz9SEPIA18cG*?4i<;lzx<(e>>v~GDKQB#1arC$<!Q&xC)Hc0h
z!K=YzjaTi)JWAGuoE2$4Dez8u8QhWw4$An~2~z~JwF-pJG(<sxcAFM+A?2;lsjtMX
z<(U4?T=xAW1AAy(DbnWDK<3AHMslXXxe@0`6+_WtAEchVQ?`OYmi=j6ehi|0e7^t*
z#+t(7>%`37<Wfp7z^pBwV`yi7BCOA$go6@QM+v7R^pw881PEJBc%*@Lf;4nR#jU02
zfe=JzLMVV#7N(G53%2}jIg$2BtYVmLcN?$pyoD%p=2@u0`w%LGBrw$7ffAmMdTY4x
zZO)W4Hi0JpEuKR1Frblgqprt@vpVj<k*l?~0Hz^ax-QBQR%(TEBN|{05j4pXDp%5w
zvnF2Ckg%hv_-hX9xNK1~I?7wCJOozEp#Vs}w@8P!FoQ|zu7jF9Af^yanq^3lVP7(q
zEOP1GyQ;eKK%5Li?00hmK=w3CaF#B_llJ{;yTZ8@@a&bu`2v<-5)Ov3wRWZ&iq>^$
zJ5<<&H1$QjpjJjZ2vI4f;N^RV!UWvr^`1*|rT9U*K&%8N|9K$YwzQVwiq@cxRBiJ*
zZg(0TxdPXkTc;wPsa-7uP88H-c*NFEMb(^V;SU~HE0yNQ$D7v>bL6ktF6IaYZ>a`H
zeJ#XqXA<->))%R45{!w;5?TtN2a}E=LKoa9Xe*lBY4FQg->R3rN?VLdGC<MSYQCpv
z)ijRTMUlU<SO;axUb|bi)|JZ~?0MT#t<fmG2ocU!oBX+Z3*3LdT#SQPSG|=jVHfA#
STlvlJD8F$TvnPVR-v0m&!q0F3

diff --git a/tests/Grid_simd b/tests/Grid_simd
deleted file mode 100755
index 381b4a3e8b15d9f0db589c9223f0590bad1d1e13..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 98268
zcmeEvd3+Q_`u^k?hzjXJkqw73=;$U2m;h>mpgExjCJ-S?P%wm$kVr^kGQp?_!33op
zqIj$)x~#g2ipsj~E|-vSC7@!!3sfYcXb<5MP$Jj-p11m#o}Qj0#IXDO<2(6es^7QX
zdh4yXs_S{Xs=H@yiep4@P>@Oa1e<O)5tY_Ep0T8Flc_w(D-&;uHC<qeH}y63F?E8J
z8UKo0ynLjj&2lj%4TCIwLhv7o|JY!nD2=B?P5Us)gc_ueQahzopZ60lA4+Ncb&{l#
znoL%fPh7IKez=@JJd5$`<<vsP;1g?+Ql%W3N9CI3a?P^5lA2`ul~idTsiXgNIX^u(
z9u(2>l$es)w>OomH>K@!PGiSiBs)!A5+q4Eb;|mbR4I2W%8{LI{V2^GC)XFqFW&x<
zG|N_*t}Do&dc%<G3Z@Mx$S*7_A5b29!+;xx3@j}gIEb~I^pot!F}JaTiXS6Dk<H!l
zPh)`c&peaR#Z!3IWB+>L^JBxVyz=i;A6UXuUM&7e2enJcvP{XUO%5X7Q2Y<afBn1*
zcf6GS;ikuCdg>2ZZh5(SV*D@P-T9<7GUW?L_vgAKnexo9zuc#I#eMb}|Ax5XXE%(w
z*dXUY3^F7BA~eK^-!*`qmjdWnAAoO0gN^jx89<)=Y&EdIZvg(?0rXrFfWI&RzbF8I
zIqEgC|Lp+srvmuxrvUO*0r>p`^vmD?e)v3qp5_4ju>thV3DDkI0qnUZfc}#K`u*ns
z{1pN8R0oi^1(5HK6V0f<UJ9V+vH<*#1L)r#z&|Gf@TUYA57!0IpC6!K<_GZqnE-l5
z2B`PD0Csi_p#O;g`hx@b`A-4rO$?yt%K-i_4WMU60QtWJu;;Y^^?r-?_A~V|)g)%Y
zBPLS@`!<<AK)u9|Pxj@%2K~fuOpy2$vixPT{1Tn~U6Ahv`HB=t-YUz_mF1_%^`^=E
zwL#LjO^b+^Ia2batjC%p>9IhLJ}=>)#(8wlczk<5jk294x!wvns{(wIxAcmaIdTHr
zD|JrGzV=$ex*JTHnbT($6=s$?vr3$qnWoH9vogo!OwTWM=9DBAWR;fYl$tW_>8Y91
za!O=&dTLTZQDIJc*3^O=DbF}3v)Gwel9M$pb4Jd6nb|mULi~;~<4cF2hT@p$Oj26v
zEGfD#W!&iTS@!G{7?WL|m04U?l9M?rzr<OVRX`<1k9S_5?Yyrz2O=?pQS9FQX*q?N
z`Gxt;lvzw={*dCFoEeN5?~J)IGZQrz-J3ZTm1h+c6lF_NMc3yRl$GWwVmD~Tva?E^
zB!%MO7s0HYteL2yG^fOA&jvRpJ1?sw(^-<0?<|EKDdWH!JT<E{KU<Nur)Q=|XQqt5
zAv10C_`#XXbP_7bDx6j{6K$Q9pPh3f%x8w9$?z=di-V_U<`=>{l9l=raVRl8$2lvj
zKyIuvHovG8Z7a<gVuuToV-)@kXlYhKj>;%aib+Wuovk{TJXv~UcG1k@tdbncVZPJX
zM;7M7MA=ORrS{ymwm&*EJ%%{|RXSNk1v%w#3py74o_|CB^ui+4EcaPSY09{K*@Rfh
zlW==};dIIQMqX#5Bn8bP?~!ja?Qq4|;+(>nMbl_>rQbdV0}G?5JZozHtQZ;!*))nX
zY52hLxkbt#n~Kp~n3<jB%+5o(@lLdJTF10g8sCGWGskD<%q(`^msyfijPt9gBx`yO
z)jqwvJhM2bq_n6os~{iQv!a<3N{5u*n^lZ)MZ+TfcC_w#ty89!<p%Z$Yr&B0qB5s{
zS&ZWVWu<N@Dkx2%ZYaEWCM)dYz(FyYnX_i*8#oaiT0}!rK1HU{8Bvsu2?QD55t=+m
z?(L$%#bwSE8gc3AXxcgTw}Dyc<5EMb#yjD7`SdTyDV*-i%giY$DJoG%yYI<^p+uu`
zT0z<@6DG!)(wvx`HMKO?MBk=@qUi;cGd(*yvy^gUOqsV8-kV=I4U<@D*~}cu%$AB3
z7nSChXU@!-iGH14XhMV0m6<e$VrtFJ$}cb#V|Y7r<!KIQe0jdpRGL4X4U^IwrwL`{
zB)j-N6V7^P{>+@B+-X_&kyuu?Gry?NG&L)G2Iisc91@>dG%Lr;$t=BZ=2UbjY9O=Z
zS$$?!exWJ1Bqzs|Us_yNnC&!;OiIceJTTfc(qT_b${aLs(7@}}<Cr1RVMYc-h764J
zp<-f9Nf{XliT2ExfrEW=WCe1mm>6v)Ax|)_mV((gr2r><q@xgAPbo(LX&h5-Fs_n2
z%O!%@Q5de-B^^Xd8ZUrE7`UB~UdW`w@O=?z;?TeJCs8Pb>#-oy5X1<|bw}FO_g!Q{
zS+jicN%_4r`DMNGFEF9Lu4;a;=~g*E8v34o;u+*zWa^1;(kaP@XyiLVqm+llB+QLM
zBQG%7Wcj@2XF-RV2Fdhi@6CYrPNoEzp3mgXCWlOCG5Ju_9Ww2VdJ>oH!KQrB^l83D
ziW`W8;p#_;W1yT}e<~SDTta-hCRIL4oJ4%OW>r2)TtfU?6edz7ej<LHhOfj)#J@?y
zSK<}oCusOe{6zd=8h%8))Gq3I@~85#YWOsIlux9FFO4=SCtAZtXr_K*HGHaD`NV7Z
zvWr<3ui=wT$|p_3SJqM}FGIuE_RACv9|KYS<Z1X_RIbTXtl@Xn@XIy)a1Fmg!{;>o
z1sXn$TjjGz!|$drkv^y4Yp=JKX!w_C<d<srH*5IIHGEn_Q9dg){2mGuX^n>eCk?+|
z!|$o#H)#01H2j?!es2xGQNvf(>L^>(@GsNIAJOnH*YKM)d|JCuKl1ehjS;KL#2K&Q
z_tWq#8vYd;euRb}ui;xY{3|v5NDben;YVxuS84dM8vfN9e!PbNXAPg%@FO++G!6e6
z4L?J}zgEMaqT%<~@bfhM0UCa>hCfilFW2y|)9@=ae44M7&jJlUQDGuoq~XVC_|Iwh
zgEagl8a}PnD4(So{t$(Ubh(Cqy@tO+!@ohpuhH;t)bRPr17W<&{A*-C6F<MkX%=5a
zCSSQRY=cG!ho0;ALd0{`kND>L#^Z?SJQ8v@dhogGYocjNbvH122hlW@x@#D{m1vqm
z-OCxho@kmn-Afp~ifEcL-HRCgKG8H)x+@s{7SS~P-NlT4g=m@@-BTF-0?{-jy3-i_
zB+(ZV9na`TiKeO19nI*6h^8sgZDsWRMAKC0wlKPcXqp1uCPv>)G);Z(BWJ<5YAVq*
z<+&RfeFxDr)wvrOJ&tIa;@mZi9z`@wZSLiaP9&P9F82~f4<-6iq8BlG5YaS+xhoib
z4be1pxr-Tn1<?^iPhoT~qG<|qr!o3sqG{@K$20muqG`%<M>9H@Xqu|rRz{!vGiaKk
z+!jV3C7Pxtw~5jFiKZ#ZedG+ae=pHiq8l0gHPJK$xf>Y0gJ_z1+%=5eN;FL^?&XYL
zPc%&_?j?*~MKnz%?nR7#pJ<vw+!c&|i)flU+{KK3g=m^G+*26+0?|=Kr!o3TqG{@I
z$20m-qG`%-M>F~%qG_sdTN!;n(KJQ4EsQQ9nx+P~iP3ixO;dvV$Z6L9MAKB@Ze;Ww
zMAND6Zea8{qUjWO*D!h%(R6CNmoqw%Xga0cOBg+rXgZbMi$JHXdo3l)B+f_G>u7Cf
zEs?7la297U|D>&}Nkx8+nolIl7w-X&n^41bGanv=v%1cF90`5aMF%CA*4C?nyejye
zB>01mV51_K;l3P_xdXYrRZI_^|7q)*ss4R1RkDvr**@}5%t5=b{VrbE%UAw*BrV-u
zS7Ysz0Ap&Gn<=S_?-Oql{{z9}xm@35lWD~sH0?Hiz)mE5Wpg+$?7KI3@`e>uAx%uj
zC$FO{Pb1g2f_N*4Qsir!Lm<x!>-pOKx9~v?{PyND6kbJw{QMm#%nO^P@_Xg-6~o_*
zf>+AA-G-9l9CYa9&E{vo3Ocrd+Q(N8-`NiW&dXI3y5F|$2zI?7?`fFqegH+dz9x66
zl-`ywA>sCf+Y-{p^Yad5VA`{<Lk4|r5!Tpf_lp-ku<?{u%XGCwi~YVwm{71CMTE_E
z*9~^p?P0v|6(4mXCFVz7IK|hV3a0Lis<K5>@wzHoJW~2NBOGz3xJObk3yOE5i|oQi
zS%E`%jY(5VTUYk#(8atUmE@z2^1^4N)*0pyeim;<FDUsWO9n<DT}zYmDym8>+=j1R
z4YG-I!tB@HOdWj2UVBEd4c89LHVYNN$Js44?hN<HzcFKr>rewPY~qDwHYSEN3}U0I
zliz_`s%%t7X}QKYUN|g%a0*4MlV4(TREVrw4*fjb$NW+KG_#hkUdWV-Q&HnOnwLeM
z8YXP!R|J6~v+P1Aeg#1gwahMz2wO*&A#1<I$XrD;lXwHw<}!0v^)vCVc#(HKXTka`
zKmQx9su!TB;(=I`)BI>sQd3V}*dx~ePSPQej9yDX1D>aWTYH3Q6;F~bp-#LQsZ@`t
zGwLBk5VxIV^}J#8uV)P)P*=~C7VCLZuIDZ2)VC*Bs^=DidhYVCr@6VM_FN;^lSn$z
z2}`LVmjXbESVF`Cpc@g-5pfF8L&PE??z%!E77$U3Nt$vhh;U-Y<f~`f%Ehl^(Qmr`
zc%8M}Joid}KP^Qc>iiVjqWyQ1-RO^ppi`>n616{Wmg@PFK|Q1V>uEUAlAnU*dU~n#
zc*ob5%tG^fSNON*0LG=RJ$WtG^8(q7_PkFz;lW}t3>VULlqU{Mk&dQ_>7%8i3~|)0
z(ovciKS4U;#aWZ3qj+&)w!ilyC2NBWtfhTvowcieYso7INEfUPMJjp!YVtnCQ(_Hk
zr}=rzxB9bgKTca+J>y!er%JA8F?8xjQ-)Mej6ppM{p;EPYfJ6vBiA#8bi(Jc;@x?Y
zFQdicd!(aCan}9PkyV^s>F?iq))MpItp4rp<ZsU%E!wk??0|n?gHEa4H>m!dBGnUT
zP|s3KjXM7xJ>F8gZE`)s)Ox(5yO~*N-rm>0J)`{Vd9=lP){xC;&*!8Q{*4fax+Gs(
z#KxzjqcHK|#dI{xCJy*J#?jJmO{UcOzY(gP5KkSWf)ZAU@4f>$*BS|Z=6i6N^DSkA
zLqNwf&Q@?z=l>{S;)ft&YWR5@Gmw90!lZ=B33nxAPTpV_>fBo}W+{D1PT8d9Kmqr+
zI3>kIIvR>eK#ZlM=jdo49gV>RD%$9186EYaqYD91qJ@s$rK3)C)CEW`o<54ByXojC
z9UY+~k&eb77#F{zqh&aP9cqg&Kw-7TYnJI-ydPzJTf7q-wZ$KSa0njpN;q6xi@p(8
z;>#V6agIuwHfh7ZWRpn~?^GU$unS-C!mkvpR6l^=Wt-S_x5<>t9j-aX&8=*NpD0E6
zz|tByB@`$A5;-NLAi6?IlZrtkO`)$ezOM2Jxbz{3Sd1o0Bjq$gPQ#Ewlt)4zzAmID
zECC%~7RFy&W<r0~o(<-OQhp`nCa}QH*&8P3iIX>)Ob%BP&%0Rlyvs$s>Zs06b5yTp
zYNKlT+TVj62q&Vp&AJ#+&sQF<5U-NUHFYh$XjFA*1us-mL8u`RgqkIk=*m{FW(eut
zAl0cU%cQGUlV_kpN?c^8uL=t<%(hv?rHX={r7#E_*#auiWM&-R1?Ra&XZV&6r?BsC
zrSyfRVN$`Q3`g~=^-|qZ;%e}#o0T1?l=U{7BpucNY?O*iiAyBPr}Zd)LOi_QujM3}
z%N1Sr!FQ}{Uu*`Q60;?BehpW35&F}C_JuiIsqjl_9`Aa)0{N+~+t6uuT6ot3X%5%h
z<WzW^yp|H<$z>n)nkC5<x-wA+U7IK*e`o@+3dFiZSLj;qy-E}G3+y}5g<Nn#8Hc_Z
z=lP@)!&WCz$>$SK3|ohyD<MQC3hdiec^b!w+<T$x1a_E$8D)Uo6>3X#rLR1VjPw<!
z6NT<8?5<AWtkZCsE|ac*m~dSv3sp%<IrP+}#?^5T4abE<s!;22!3!7}Y(%<Tw5hMU
zcN-1N9&`#XswMZ1=E<Maa3abBYyk!*m2gzQZKZ;)N-1$g8mT!f+BIq}^HoD7(Ajhz
zLXDKT;{DY0rp^wQ)i_+qHVc`w9n*Bv#f)Sp4cSFyQ|E8ts-D4t-DQdS%^}qDm5t3P
z@EKS20^I8mw(*reHS^VTn)%wpAu$JePo1+bFZ|9|ejilU1GCKz2%Afx9mn6`xW1{!
zF|Mc={>NE941)^rHsF6NU+D=h#R;;xDVeW!ycNwy9fHuU4fvYah_9?8_=?KqV&m7-
z*GKfVk-k3RD{JHV%FRd0K1FG$+KK<|l=C&|<}R+GW7JVxA4Es>x!jyj?Qy@9g$m70
z<$QHvkUh@L3r%)z<St&=#Yg=D<3sqUW6sW)o#t<GM)H-v2RVD<JGcx6{@N7Ey*KQA
z{J+I}aQ$?MDBiPKXyz+Vg_c|`G^^L-?}b7v`F+SIzaK_lU0`RZBW_#Ck8rFr!S0Gq
zb#+d4b+_4Fli}j<b?~$)=Ae8%{QQYw?<NStmb$vXo4~!-{Vk#U+d}A@sP0`<?1=lM
z<U1@1Mre?bUC-?i(_hCBz;)VzL)sOB+(jCEn0H_)Y9bJiXUh`FwluC=vMrvcc?zE+
zx9~!;Z3<tPY|A5>sN&64whAKX{uYxs=UtNt*|EeSMX};C)GJ-(O-0STuz?pQ*&=bJ
z=Io4VpUKrFQJ>3F!yv`L(&OxGufruGhFRT6jX-avW)naEpfilGL;Q>nQjNs#2dN^|
zViyk5bY!J<3DhQw#xqeeK7vqq0oZFR@U<!$U*fPPj|bQD^J{ol70a`8DH};R%u(&M
zWzdNoAM=eDWO<w|k$iRk2sqav;A}W4#sRisJgcZ0j0Ew~uaH8@>MM1LuhcY1*{cgK
zqfV0b`~tsg^dyp=$_-%&o{b5Wr-NtD=Uq-z?drt4(lOZx!Mtk(l0?iq4pim}!ZD`Y
z1mZ<tqzC_lo3Q05*HF(zjj|r&ES>N_CX}17gNxjGjGOrNF>cZ>q+5@188{t=_S=QR
z1-1z8lF7hu%ZM-%?-EcJVyHOL>JXZE0hii|k-U&3U$-Yyge#2?c$daSG>_GQ=is2V
zFC$Mpv7bU7C)lprMHUr{>5ydWENEW5ILfsw`Uv@R{z0y47luv6gX{`FC8?<wjq0D=
zq#y-;Rxj}nvrBxM*m@qI@lYqmQ03?dR9-Hwq>V>fxQY8TVI4G^SggSd4x7|$2RYZ7
z#aGreOF1|<d#<E)q9a^wn6QwI4VRQOb#{eS-|7)8G;p>^>ex9$<Hj<#4=sT#K^5Ys
zFcUIBJ<KAMm%C_`D|tBj96^iQn#&bpUK)BuZP2kqZcdk*xJk8Q7q|xf#tp3H9)1`;
zv<thX%Z3r;R5NLfV%{?cclQtHr=tWlIbuxpkl4}H1;h0h5$}lPlR!%&dq^`pcLH`x
zwK~+U{HMZ87Jd>p|4f1#){dcUMB-@C9`OlC03!Vxm&DC+3^!5h214W*cL$uYN8F2J
z(xHwPVx}_Y9-I;4R5DecKS0W-^DI%yms%t8u4NJOh;h!+jg}Q;t<h+q8k?BLIx&P&
zxX$8{x{3`58r^hK*emOl^N?E@hxoq9reulxe?mC^2zV-ICvnO#bQ}UZW1)0@P1)U&
zM{gyUg;GCevA!@5y;$l;hif@4YT>ehs~U@bOsJTHG2bcS(GghIqH+BT)RKRdF%O<*
zS?fx<9XBzp6ppZ~r(x7~Hh1E*nF|e=im6FfX~1K$*dyjKHqEb)xf6tX=`x0LXbO~+
z-9qHNjZ0y@265c~1nmYAn3{r{Fkm;kkK<ULkvF?{YKgU4;x!QB+xy_G%h*EXZdw%d
zIbm_OI&V3*b8`Ncxn1c5Ttv#rewxCqO_zy@RM8vgCtjc)6{t3r=MWB38)b)GBsV!i
zYO({RRR7%qA{B1Mg{(&$L^&u2aH9*?D`ir4Rd<loX$A1V7d<)}f6oGKeet*dUlnBC
zOAToif7d{&Mf`n{sc8v+?_(0}#NRbP{9*iE`41o5ys$x<19kZOl&{o}pZekNJ-!m_
zA<=I9ML-4U4F(L?{3Dl}n`jKc88QG*Y6Af4W!nb8XW(960Dcnlk2s6iz5x6gSOEZd
zr#}Ga>?H*O0Qk8dY2?+3%ULbw1i(|y?PNCwU@RTT06Yd}`T%gN_`47#TEX8pS=-GI
zb!i9w4)~t@76^Yc$zgEZy{JKtzk?MX;qQn>64c>uH%PRCzwe@Y1%IbfGxYfS3E3ww
z{M{?p;e*fbNcn`n>!tiQ;BPER8Vwc#pBc=II<b;@rXBcu_Z~g|W)P=!{QaJpWnOh*
z8}auZs11z2uQJcniL04b6;9TmkcuCk?}+fjUy-q8{QU-mioe?^$BVz~i0okey%sv%
zm!Bv6CHocpUBA0c`0J9JEaUITD5d&u9S9YF-=`ce{=O=cGX6ddlJNHc{D1!NcMj0j
z7k`&MtKjcc>iJgjcNh&SS}YED{may}gulO&$J%}U`|Tgb-?+DY@b}SrL;StOS85uh
z^w+;{`shg{J-*k!6l}RV6L29^MIby939f;0J&fyLtRRoTwH&U8g%J^4WG60TgoTS5
zft=3%mn7Vr3vt;iU+~i6C8^S1lyP%Lg+YqDBo%dxP%m{yMDhlgx=+E8zDT=(khTl4
zeUUaFtN=*+oj=l^`kE93K-z&c=m8ibSS{y-w7DIDwEa}1{SjvRAg#Q9)!OxLpKn^l
z+c=0|^Z#OVJMi|yT~sy@-X12+a9?Mndc56$1~DGt?Uy@AP=~jFgG4KMdox57ygiCz
zKfImHm<(@A$W%QZ50~-@Z^udbZNS@lveszG5Z<!!gf?G7n%i@|d-5wi-ipL&9dGYp
zePJFI(nh=uLNfy6?O^7aI&n19s^YDK+OAyhJ`TeS@V1<>WxSmMLdDy=D94MpV~Olw
zyj}ezoE32zirE!t$6oJ}{R-ZWql#L;-t8?nS;pH*D5d&uEC>~E6Dh}ww}WL;##<Xm
z!rR&K|M|n)aI7Kv;_Wb8=S%C&N2wvL;_b(fY7uYWW@=i(TY*Wm^LqE=&;KyqZh6HA
zCwtWz;_a)xQpaoj@OGZB#12UOA-tW`3f@ktmGPFqEFj)as^u<8X#?I~`31V$XKDE%
z!rR60uMTg22dnLP>-vlo1i;(1R4FFUFIg@BOT2Z$%=X~zGL&oO`gRNRfccd(R`gT6
zsJ+>))%E4n9pt@0csrNe2KPOS8uWNOUf~hm=58lJ9p1)3q7}T|hUyi(eTbT&$J2vk
zpLBg2%no(y&!0&7gtvPo8`^-k86;^mWC(BBctV?BCC%-@+eaJpcw0`K*6}uo)oA|u
zl+=%{t}j0UwSn<=1M^Is_zi2UinrgQkct<58R2(*8>W>x^{EDL4}s9&?H(dK7;i^I
z=Xt|hvR}d5z1!M^w{Oc$mhtugN~!+a146~y9hBq6+YK@)<Lw6^32(cgJI){8z6^Br
z#oLb;D0ur2^=hklI|))P;_XdLO-p#&lS#A_Zzp~7hw*mei#~Y!7WN0}MAoe@5B8O+
zf|Tz1wwtfSRMO*neS0qZyG6lV<S5*?QQWYEn}}N-inm~)xwz(6ZelG~o*THRao#m&
zY~^;t|HM!(axb=g8@b3PByQ{g??bvB|JV`Ez^xKP`~9Ho0_SLUfl>BW+7F(H{ovLW
zaF^r&zgycH&PK=h;`eOA?{7b*QSY-g90$p^<M&4&k%9pDeU>W4WZZ*NpPeRPY0F!~
z9elm~0nBU<ejh-&R`B}*=0fw{U)zD-)mzCif$)1BX@>j0f)RTBeoEmHe!sqj1a<g*
z4<uT_?{J7H_`M#-e)v6rG1>KUBAKek_pVYt;rA6%ejD)nIkMJh$Pj+B@q{-2z&z8A
z_35qkdi-8NoYwI>fz@d4b-a!Ey&Y-;<97)2Or6-9X;tyN54BynUe1IDgX?9UvE}RK
zn?R`7%L6ILd%fJ3$PUJDSKWEUZ?a#(?<=UH)~}bFAVy7=@w-2fivRk8Q1Sa>Sy0{|
z4v|S2zmIR0@Y@0ZpFjNm4ruL*-(BV@_`RMwt5y7745=3J`vIn=CH$VqB-)AJi#PpY
z{C*y{rEBi1+K+8s1N<)amD&g?9e&^HEAa~HX)k`u`@`G>^9<ZQ!9{5{h=H@0-cHO3
zTM3YB6+lkW0J54VT`RYSwI|?2U#zVrtSu(CFV=nnRsgKs>%TUgxsenEz}hD<!U-(i
zVzr#pwesB^u{M2`x;D+jOdqVx^`3F9txwNJ*;erOdFBOk$)R@O?PVLtgMsjNEV&LY
zoQ@jwczdP7BfPy~JqhaYHV6`};O)z(UcuXO)C@hIen|F7*ULNPI(+tnUy||(Z{L^l
z+km%`Bxy8k2ygFXX4HwL%rou4+dFFYc*_%~b-dlq%rd`qppAH&1+{_k_9^C>I`M6$
zRmIz7D5T=W;Tj_Ru9tT*wv4wQfl%>wE#-Le_B|py7;k$+r@Q-kTAwES6})|aU7PUs
z0lCRC-mXC@)qn4SQ1SK^%JE(=KP8hg-d2I6>*a6ZfA87f(f4B&0$qLacJX}*-j1W5
zZxwI*)1Yc;Klmc1rX{?k_d2wDKe+$eKa96g3w`kRer)_2;BAPnR63+|c>6sbEYslj
zAky>4@%Gf)26+1uF*^=#AAu8n@%DDYTNAN;@pdv;ZO7ZgYe+!=yuE@3JiuZotL1-*
zw>x2Gd-1l*hppmm6hyEe9Q<=T@b;b6R5lRa?jg-^;R!TAkGKC+c!al`SCOC&Z=Zlf
zD|kB)A`0H_#jzjWrZOhGzMV>@>hXAhluvk@DCM^SZ&#4DM#F~iwux!P{m+!PXMKAA
zN<H50Bu?vio5lLVJak_h@%9&}4UD%}%I&y`X;txd7`0uwKAQ^-26&st*fQSU2|~r&
zF_h!Q+gph2V7z^M#d*YAvR}d5M5?Iu>)VUuCd+v1Kq=LKw}4Rbb`a%w@%BoYl<~GZ
zNW$9!_&*@t;_8$=He{8bKZ>U3kF37WAH~!2M?3^^M>(EQ+KeZX<cE)f@DMlhB3K@F
z0wcuhN))i&N38(Z;<=+vOnHd3N&E@pGxY4t_$l)4@dwDT3!74Ds30SO>!#$qz={;m
z{oysTGMEx-qX|Aw^CZ#Hf8){QE%R&0-+2x$<#)ciz!pol*;{d*;f!?%N5z8=F-)_t
z=6omK^(s9PI1LXQ2|vpcA9*Dlm}!QQgm75Sd{>)!NbHMD&JqG+aZ?l?S>h}v9FMf*
zCUB9RbGgkmc-U>imRvjo1WwdgEC?5c*oE(~PF#QJC@wQI@pSevZVncHC+;F~Jcw6%
zG7MBhE*FVM>?Yxfxs1;oE}ISIFA1gZKjGP<1kN(jDtU;#7ec)v5Fh&w+z!83`08|9
zx%9Blg~Wpwa2=~W>@%HMc<+cbCt-(iapV#52))lEnl3Ya-{;Z%FVf&YY188h{b|@?
z%1&mr;BrBI*ayqG-2B^vO{Sx;H`GSo4qF(KNh!pdlo}?EMVCn)8dy7Ps5IkypR=h&
zE^y|BMrnT1?t}gSwamy7^wN<#q>|nzJ06irPBAF?Pq`#l)r=OSwEKb}8rplrivzTh
zDVN%Zc20t7Jv2oqyqMp_{vR;cYM;zGES`Ysz+SLZ#c<`VWVIe}*04+f(of<%cBtDh
zU999E60_K$aj$MhBbmdlCx_Xoy;qQ>2E8h^H=P+%Cl)|yQ-6{_`cGqdGW}{<`esP0
z6I22!P#>eOLTQ}b4&fL*y>VD{GHohM<PeF;0(SxK5i{gGtMp`toyd6`X|mEh_2RQc
zaS7%K`AIQ{Ef3G*<*Cnz;o;fG?>_?#)gRa@ph8>)b?!Xk2%Fri!ND0xhs$wD575x@
zK`;9Y9MTgjVX_!jhJ5H}K#RHWL+9qxOeX_z?OEWQ*mhgz4&d?FW3hNCjBM&BhLhe)
zV5=BPM@e*a@*g;&SCzp~)z@Djo39*ZGdVBdU3l-qrJ&hdO1=(bneesAc_AT2nD`rX
zs37YGL2TLGUlWN^gV24t>$5+G7u!qqc;T6$;KjeFj}7qRCB`?v3mQ+xcyTu?(mGzO
zU)~11_;9ulUJQK64=-NyO3+E#g(OtGn4`@+BpzPuix;^749w|-7{ELnHOQy|v>54!
z7IC=QxkJ%n3iY%XE!dMO-@+?8w1{H^!gyXWK#Q&Kkp^S5ILlOF|Bq4%cD3h8YiQ9A
zrUgcer<md9g0IgREhb7O2N{$Mj21t>Yurj{_GIk?S`370JzDICe;5yFarPZ(5jOdv
z#TEzzMvLC$UYwp+lGS>&c#LH-v>4A0b!btj<R20TvqR%vH9(8wWT`=~O6|Rx8B-_T
z45boUYz!bzrZ=;y>qHAubq2cx4pq_Oc9hnl#R#TNMT;AWRM6rYIZvLxdK1}kXt5CL
zIvOovmqBMoqQ%cp2()N?8+sX9d`U+PEk33rJzBg0c`aJ}3?UUQz6QZpYSE&eWy|jm
zT20wrv{*(Y`ZN^XNACypdB4FQ!;8Po)Z@jcISO9Pr(QO|i<yjXfEVe=Fvg3atVrv4
z@#I@=z>DfxK6r5!(Sh#zVun|O@ZwvNQ1N1nHuI3U4w-?kFA}&6cBuiV0FX!*78YDR
zWYo)74Bfie0$(a*Q^$EHfA+A%9yC!m`440LYdlR1_P<p4<W16GJoyLFsKcb#lZ_z1
z$$w&tmkKtR7I^Z1h8b=y`rLo=|Ec9mg-KG$>kUf&JK&|l&r6M4DNTl~eQ0M4RO=`I
zgYb_s`3I2}{p9~K1OiX~my>&Owp~qD>nHywSSFkNZ)b<P$^RZD|B!eCJ2dXqE&i7Z
zC*EkOy;01VIxzuCrOCf8fIOLgidBv0ACRh(e-F4qo%|=Fv^4qC{mNm|^+GbKRMwGh
zWQ2URa1E<W4PSc``G@BJd$m^!3!%<UZ<D3Vf)2V`h<zP8H+10D!p~4h^Z(zWmtN8<
zeo064dSCHlIvSw6T6hEUKJ!0>yz@T@-Tcq8mHD5tHS<4_H2<RqJ9_?~?l=Eus`Eef
zveEp{_(t<TG6Ky1tVkQ?|JT|w{}=nt|BL+Rf3F0a|4G6-|7$Zb|0A=d`G1n({67h=
z!(uB&-9lTM|Ec4=^FO<Je-E0dpZ{6^8qX9)^Z%=)A;A1kqmJf(HiFJ|{)cHT%>T@A
zbJ4bQpZ}$j*Bg{<Vg7%`xRvt!&s>K2AFBQ4fA~k8|4EDA{11T^=6`Z8&HrSz-~7)q
zrTL#7>gRtYAM-yuH11WS`TuV%wKs|xgZUpy<@rB=Je&Vn)tLX0dgp(*!aM(?wEz51
zD%JU)5z73}D)Y|&M0Vi(4|N?r|1W{g4xRs@kmmotLa#Lc(~&g)(~;l&4|$*YA41;w
zAB1lHXW7d9PuZIJpGcbj(Ssd5|7ZEl{}a{upL*G7{%3rn`5zep=6_bC4fFrYZJGb?
z_MQJ9^`HN}5^Vk_3Ge)`&BXkV%mDL$QD-a{MUG`RpeJL;0C%8o-h%Ch%{3vQYO&|g
z0D-6p$|@4J9&l+;ynPJ5b{>KVHXSz8h6Bk);{7L+@HW4U4a#0a7q-`cEsekz5LZyo
zdjW*qJ^nI$qyvys#0s$2U;rQsUm^|008+!oTAjF))pCvi@&gnH29VpBO7p<_a|V#!
zQpw|W{v`tg$m@%ZTPXnuYaf8}2u#)k$TEe8)uWmhnQMJlk6aK43?RFyVw|q~Ams-j
zX^hDL<Wi&jXeIxU_$$-XW&rs&S!&R$QhP<Fw@x%6l>p@70P<w|Dp`65Oi%&j3lvZR
zB!UsN*Pyw82qBdUfc*9X5i)?diSW*7UlZAJ05Tf&b~J#Tcpf@C5<p%@n*ktyCA|zl
z{z69#KwNaB2asIyk!HK$bqJ{d@;nG%0C|LE%K&meWqaK{gGhAi2KWiT8_|L5M_G63
z@#FFF3Vx)}XfVK!0gP{eAKj5*j2|b-7qmgt%Kr4Y=h}cDsndM$V{Mfmeq8I7ApBTN
z5-NU#YcmgtbCDSUKlJ<43_aK;H8!S0@IRV1yf9&Qz2jlB!A2XnxS*hKM@z795%s4R
z7TAr>kD%!~Soo1x0bs#kQ=#lx(qIe=Z?X|oC)ToB{9vKwE$EM-I4~^S%v75HwBCPm
zZ)FQQR4Tc<*1u%nO@+swF>a+aO|td@7Iwj8JuLi1;Q<!j{0nofFDy7A5EvFVQ^hz5
zw?WDe7UCF_!NOT4r@OA`spKCLzhn8vuwbyM@K>_bpjV~#ZfAPyL^qU5uuv30o=ksL
zmR=2M6&BW_fC>vijPPzj|ME1|u5K#qB~rnJFXTLVQ(-fae+UzNwxDl>B^|w~@WWHk
z*^!$HPomAh1Q+RLn+o&jh;1s|OGo-mg>mE~4JJGZAr%wmgYd<KQkE@aLM~-{F<}Cc
z=+pP$C-2Qm?Dvp55FZk5)8j)?s)7$!p=Kj|=*sv8_;4JpH^zsr$uDidhe1!a0Urk3
z<%16|&+)^Di@Xwq53@-^#fL+a)#47^!kCH7Ka3AUE6xcYn$S!QKCl~erxVZj7Nw2E
zYCAsM_5^7##s`6oraJL&td?`chh<P47#}WYD$OTW`_J;N;KPs1VDpCtB?IF_(IVqk
zO7kUaAMjx%RO|7fOyL0^gvXg{eeoe30)g@2b*dOA<hx|Ge%+}jV={dBj>+lp;djPF
z{vmNKJ8UyP%q2?=dR1!gKUlqW;ua{C@L_BKc{06HmVOb^Dn7i70xCXy{}`LF)LWFc
zgVx}~dLk8k_?Mg~<HPGjb{szZ31)RPK5VFl&W^-~60{lkFoX0me8{3Bh7S|yNRJPL
z$VVD{D1nfQ4>=%w@nJm6mhoW(Wqa}AMk3LtkHb$b;RF5t312P!)(yMcw8~av!j#vo
zO8Tj^6(I1T7wuQs8gN8EBDa(Aa134}j`7C*BgB&4xZh<gK%<KN$QLn5`%<LhkdX>>
z?5B1L8(4kvU9Vp;7M{RlOlZbkw$lBuA0h|OAApazx|76vlkql6Ji5<yjg9VA)ZFUY
zc_H24e^7d9KIP%1`S=wCdc(eJ8Ee`u^Cb_MOyA&6u#e!YZ)V%lcA3pGXC-k8Y-!*9
zvdjGI{U+0{+MVC;UdUb#zpEAp*mN*SKLvw>3VyJhwPlZc_fDQmlm)62hQ#dPPvD0d
znu8M9Zzednk<GYpsIt|g+)=ufI2PnW=Gdd$q<XL%wrJh&kKk0qEwYd^kAo==;gs`U
zl2Cu2gz1Urq8zp~>hTZZW-by$F3(=xKGpmY9`M2HmH9K4&c<thE$O!Squj*xN3j!*
z;!m?z$M2SYU0`R|Q;^%e*jB-8gt2tMdp;_AhVDm-992DJ=~3=3l$!`!p}}E`MMoUv
z?x;P64#MLwp^z7^zL$QML$Ulhs3gm?aQ&*Yx2vDMy=8l^3}i2E+})jQE8p4mR<d-G
ztr#uB4?mENuaJ$_Hrn_Q?j`Wrc)VP)ai-4yI(sLN(EE1=3X*?IS!&?lbXy+m&D+`a
zZ;*h!Q<%L|n7yN`hrHVccYaldA?(?dP<bYlTe$xWo$KfeJcx@2m(pz!@O&_c_)r`q
z;>!|`mtk~Eq&H&3ccG&$>?asN1`%b!K?0NQPDe}@IxGoHwhtE>%w;6v)s%@=RKQs-
zz%aqd*Tn`AJO)hXAsyN1Xav=k7|9N)y2SqMu!{t*-ute5JPMm}o-U!mgr8cV-=SH{
zZmC!(P-~>uA$xDY{)xB&qMI51cU9cGfRq?VH*ZtbK>ZD@qAJ-0`3D#9V=&6^EjVxU
zxeL|jw-uUUHhx=y67{}WOfT|JE6io@bEaR&5$C{=n1l2yLdx%}z#o>-W%!;H`mX$(
zZq>aIl|1CVnfo|vg87s8t>{zVms;HpadvO=C3J7{pV<Ia&wsifxcm|0=1XC~Tg;br
z;(F8|)$0Ax1hSX$5PLlNFgZ&1%+O>A_%tEi`~n7}@{G`0vRWVSM0v}|Kc(=jkyx-I
z_O7C>7#v(9*yY>qM|p3%cV;pNTuctQnjGNUezM%4`w`cR%UIQQ;zsx=!BdAHrC}#g
z&RKwSH*wp10PzwWiyLv&^a6f=sVq-&lJ{=;iOjv`LI3dYnk1!byg%;Q1Vu*1NS$$+
zY|iyer}~zL8_)tZ&bkj8aPu+m0wB`O9Vf);OsBeqmqDZwX^m!i@^726H~q&EXWm8v
zcg_9a?!RI5x^aW8dK33<9OEQ&6CAOV4fVP4^KV|_Nf6>&Rq&DPG<4DlsXKw_2N1@b
z+lhl8Kw!T>VuF1lCSXrfS8)W@@lW)FVh8V1ebWmfs&6JkP`xob4Fn0wL1qGxid%-T
zJgr-<APz>v{qVJS{P<iipNDwpbc%62WV>03hrXun@{Nb;7|SRgdIvfBc<6b?YbhQ&
z@lfmWPzat)qIhUE+$6<AzuqHp-XxA35A6f*|C4wq5vH^n54{67`^H17izU3?z)}Oe
z*2P0jko&{&P>(`wtg!$p1H?mThU@LUNVfNqK=!KfPy$)n{&=Vax08C~p&yId6c45P
z*?T7nQaqH+QUm|0@z4{HXkR>3SMXnshbH;NLz8L|`h@<Gc&O7nob&p4h}~i|y8=0W
z@z9sVZF@X4eJ&|6j)z{Psu2(UlU1af&O16DDuf~Ji-&$>O)!7>ra$Pm8V}{oF>byD
z!7ngh;`fhHg9P>7c!=y}Jj6o}JxGqy#Y3qO(8WWKVlXNZ%)4Z@J|61nEhEQ6j~Iy^
zh1E>88V^l@SbO53JhI%N`zan;%&M*vKR|8%@zC1YN<8!djsnC(cIIC56-)iQ#vmU0
z28xV~kvgNRY|d3or;7PkqXjDF=R<>UJT!snRDqu-QUU&(Se_R6qlnXS@z4tooX2=*
zIBZ2cbOY59ARf99BC2mvA*jYf@gOwuPz;fZTWl;(>z0d&gAs8zeEr{xhYk!gj)%Hm
zuf#)Jsk?mRq5m+JQ9Se#a`f?#z<4diLqFfwdOUP2Pl|rthMS~#XrIh^o;Y$m^aXhT
zpTt9hU`ng;&`WT$Z#?ui?jk0<{wGTf@LCrSeFM2a91nHEgDT!wqZ}#&#6t&%=<Pj~
zE7=<o$X+!biXlte9}i8%y#ihvKS#A>V=L=J!~N_XE!#Vur3U_0<Dq$wXkR?^pX~o~
zJT%cK9-8=v*N1+)7w5b_9%6UQ6;O^}JhYj(ZI6d0%pxVm@sO*WRN?*CtfKQA58VYr
z+7}Of&YEC;XNf=Pwi*v*lo>Z)g5XD)FYCltP=h`mB72p1sKiN*(#1o=A)t$g?#EzM
z;-QzxYJEI(fwznt58Z1dwij0W#zSckYfn6sL6#eIKgB}}S=DvoYpBgX9(uP_iHBap
zQGj?Tj=9%-@n8MB#vmU02#So1kviiH>B4Bdl<8FCp&n>~iun_v!8aZn$#kmmP%M!O
z@L$XFw7}n!I2{)cJz8=e<Dmht74gv3R7Ze#=p^z~-wbDjyek$BLK6@5BT{iocb2Dh
zOE7UTBJO~%|9kP!w}Xx2q0s(HJhY0s%Qqf+gRzX_p=#vl<DvT)ucdhCtK!z<p&#&U
z5v>os2scUb&=)dCAdValZ36HAlX&O~n9^!IR1G)##zQaOCB+x7veW>tb@9+gko&{&
z&~LaE(;I8#L1ln=X!o^xdz&Up_8w!YfxT)x)Q>D}e>^nK$HtAQt_|_f06%+glI=|j
zY_Ggub`K=l7Z1I0$A38<lJ?5}=&I0>LY(RPD2UwvluJ2&QBVzW14KbN{<~zi&Lkzq
zQPBJ%QdK9u#42hb3L1NEi~X8ivP*fqVTSFJWxx;?1z|tfCwP#4SKM%a>~m(k`JHF|
zLALeyr@*-3a{R-*i1-IJ_{Bd859>iCGss2y_y+<$O-MK2kHM$LKV-FE{NpX7#6L!2
zdttRM{()G#;vcfypyOHm!>UI7gW3$@pS#ug2S<V8ALd^3#n1S6jZypqMMlQR@ek>u
z_=oBA#y@C*it!VnK^Olpo!<C|NCoJxWqDel?@62vjDPasqbE+2-5tL_HUPF#{6lpF
zihq!&`eryIl=ugPHvS<}aZ7iWr*%s(aVY+QuiFs+;HJNtT)I=J2AhQJU^#6auD};Q
z=W@%?u4{}#pr9+25a>VDV*+0G#;TNW8hV8>5d_I>?QKI3BM)H^%b^<<q#K8d8F!<?
zHN17`<LRx3LOXAlLZL<Qm=p?al{s^WBZoq(!29opLiBv0H;T9d^|v1Xz{k4y=QcV1
zVW~kpp^twc*JAu5>ThttEj8itdvnDz<FtWK5p<EsdEW7-yZ>hQ6?%J{(&hMvr3UtT
z;~%orIR0tT#)&>QeuR3+#%Z|4BU!$W%X|OTa6kJ-%l3^AY@ZzO%!7n~yn`1Z`rdA(
zb>G@sl|Uz$TNrefPIufGgd2>c8-pMk3?e=Z2l(Aty!XutGKeS(4iXqY0?#+%hv0F`
zPcGf_!```tyME|BDfI5GE;dLeTIso#PIMGWM<c={r1ak5Gb9(4>th#ohtFF%WS0I8
zpHPwy5W7tDk6r#Ujg%P2E?bzYI`Ml}Q46ul_;$xGJ=NG{F$|GomkOHb*ge4q@I5IM
zPmR*cEC<$dRXMn~LGqCI4xfvedFH0a{9!T~;_L>OZghjoRr>o|p363FyabtBNiQIC
zFD#Sl^zLW9tnd)Oyg!xXbn(lB5b$k6x_K7{72Q_!J@Wj57l&GaMAjOKL@6SNNaRIO
z0fLr0Sev_2oBJvEOHq-fWT`=CQ@rvmtFcZz4xcFT3dVjJ#y;-nIhzHXd>_Z+aU3=M
z)$o2Efw|K>wc5WU4C0mE0gREHpDLS^!*r_gN-heinEWP8@r_rW1xEwS3y4&J`2m)v
z1?GI>bX>gh^%VH%rSlZ8JP2D6ugs!40>mq$Afo!F8iMNnRRsu*TkauJamzH8r*+FX
z;$TF)1z&sdBH;Zza+J^k@k$itc%xgKZ|$YTE1hY~=;M{6)Q$8r=&~i+c;!pv`Nu0Y
zjN5X&(kHX^cqMX-6t6VGV^X}*OXh4NjvTLC2;ToPUMa!DMXkmwyWwNsc%{i9<3CFc
z@Lw0N^hP!9k5^JhY2%f3sI%pGWq_Z(H_7%U1+rI-SMGsCd*YQh?OGdq%QhOvE7d*q
z{(Nz?9RIP@z@O@V)pkgnW4sa$)1(`j|8Ts5TXJ;q%A{I~SMUgKTjG_f$v7MJ@d~>G
z=j0^h_{A&fB;WRUW#65o#5i8*c^9d|`)^34{)S5J9XK5kuN;6O?Tc5YGC!Dade|Q(
zTZvc1JB%AIL1r)J$vSZm^y}l5L$H?d5U&K27Ja<(2?X@<N)!eag$*%euzs;+v08vc
z!i_}kQ$$$2(gf-p<CW%##_gqer9Z0?@BfBc|9E8#N=orcGL8boE4!FG&F@$FcZ5N_
zG9!R7lJj4J88r6(#dNCi%D+)a#pK^J^zq8i;Ant(Cy@#;f5h^%z`T+;9T%@$19j&u
zUik#d5wC2X5IA0W0V1kzc0*8&R~kTQ+)_)V;+79sp4Kg|5C<dTBz*l}jaTl&9B&k_
z{H>c3uS}pZqmNe-8Phmkxf*%?@rs3UTaH%>Zf`waDaSq2wAB<1&r9*j-7@EL;>hvJ
z9pL>h<CXQe`foK}8E~;4UvEM&y5A;=r3Uz_i&thquKn@KO9|RofRjxQyuR`v#)NNt
z<w}z5eTt<9_NwtpEhO3#ulydbwQ&Zt1YTd+ZPDA;lqko4EH$)GUSH`a*{6H{s>St{
zyI@)e#VfdFMi;M4JeT#A&(m>=>f;r5hs-F-@rzepBW~N{mFvfo661I!|29%pC(dIP
zo#*wHVKAh9@yh$G31-)W{xI1}yb?RkxbYHX-pxE&Cq4xI`grALg@<@$+^ys$-TKO9
z5YWdf_hC>`Ffxw}*2gOcV2f0MMD8*Y*`kQBc;zODoMXJglcffoP4UWXR%4y`G%E3r
zS6)t2;+3ay6d+!SV(v5t&-U*KgLq{%6d4&KIsYK(!q_u2ooc+&2`x}D*$xf9>nk@h
zo$7v2B#{a*_hxxoV7`Dj9T%_MKlVJvE0@7m#4A0hjsWW``;n*mW&k7P{U9p{O}x^L
zNX0FoEKlo}qhpYb5ittB{;$R>TQJ8P#VfyrDe=l1)MNU1<w?dgj#p+O&p%$7!niHR
zD{E3)k5}q(Pcy|U74V@HudI?eg~XBLmAAnAU&bpJqyARol?N}-<E!f?8UI;ofUmlE
zWi{m5AFmvY)y4vu(ADz#$|o2TzVXU8LnV8EV5xz<YP`~oEN#brkln||)u<<MyfVPg
zzMEwGk^<YO-2VcJb6j6}_r?y0SLgw7%?&MWT28s%fiqGcp0Hb2-X4t{zwpFN@&Uq=
zJ3{2$5br~^&x|4^#^K3UrV9HXtfCgelk|3P|MXI~e_nzia(GgO-5=lmA9edD*I@hS
z5@w$HfYTopQ<UwW0M9o+XE$!W1d&@wF(7g;Ow-3FFDpFS|KUka7oR)`0eyV33xkM)
zkMEJ^7oRxP0wl85NF+)Tk@kN;onw5mlq@yqZMOfzYQ+8zeBvLUoE-^V{0H0r!BK$t
zL}2bTPc8ND34{2gcK~A~=cmf%<S?CTe3FYoDh9s^Q+(r-XTi|`@d6?hAbx=5X@NMO
zI2{+Cd_4j_T6`Kh>HO23FMsrT?gwEj?f+060pgQU5K(<o4MBB1r2>S;E%y+qxMdp4
z)4F9GaS)%p1z&sPlkvQ8imwZ)>70Nq@YS(6v!owMv$684tt#UZi9v6okR*a7Mk7vY
ziSbJcnX|fHWi(0*cApD-(*9G4@ti8LR$^4BjCUC0fjmj$UnMF{qMnebXo;$nC<{@Y
z?sPjr-nk)nQNr!x`FRH-yJKD#&eD6q!uaa$wzYSlePS9}EWJw7i61T7&PER2N$mTh
zQMF_YNr}af$4h(qKp9PFlqxT5DT_e`uDehaFX_0%n=eiZ27zeI0%7{l`)f=%=DH<K
zq2nVE#&H@QH%iCxbWBR<<u=P%`n0WvrHgG1Oy*AflID0T*KM?g>lVV-B^8tEgt|`X
z4B6HyYTLYx)bKkK?%EJ|>b5}w@8+OHeK!aFrriY|a?S^~=TM>eQ!>Wm3Cs(a&m>>a
zcxWo2dGmg%Q`pdyLwUG;!pD71-}@c=t%O$35w12&kg8N_=15aCG`Yg6Z}kWkyySr0
z?`VZtVjgOz+0sO9LfVYJRrb5n#8sfgze9xGPomsF6@+(+jxDxV{(e!({q}jk&xdE6
zxx8yao_${J187J}OpPR8U2=-A9@Ux03nMuDHixVGr@ZUlDSXI~uk6)bg6(lfQ+s}c
zZ#$Q?)n55S5Fho6L&%)M3the<ZM<haUwQH(J2$d{7oglzTlNlLH<I;7O3W6}4q<Gu
z!!>dWllHBQcXj#H;p+0$*y`?Kd{Dh3Zad%eH~Qw1YWd2af*euxj;I=5fH^`asvC>y
zYVlGeG>Yd&Y^Mr0xH|Fq^>0%b)Vi(0`lzkj_XlmQ-G6!h*0(*Lxx~P0W7j_W*?=IP
zOF7EtZyU21BX#9ISojv}xO)4E?|Gq~k80qzi+oT6Un^eD=hskaKB^XONmxZi@Pnt@
z?fFOPkQa`yGDnQdY<A>tW@7Vx@_5LupRA4xzT|})9YPJt;oGR3J*vi@UoWX|#iaaS
zB=z4*>c5BjT3#57pLT57F1hh&Y>Q2H<k#x-^B4RA{g(4=f5dsV-+G?yk37%zN1tc=
zW6!hw@#ooo{yf{C_J{13W)w^)^K_?e2KRIg_wO2Bz!X_GG7oc-X@fT$S>(MRCCx4z
z2Zkf`;i@V_CAhV_Mp*c$oqTO`2*2v^A?d?&@2z*lHFA~9Xo6dDh4cxwyKdlxU6{Z5
z+TTMUvy~Ut^HEzJaeKLk5~0r#_q|>CB$Z1%U=R9)d$QIMCvs1&58H~FPmI|@ZxZZd
zkK4pOVx_ugoGBNxZXr0Tg<gjIarl+jPq?ZDxcYKbk6vJp`o+FNSRj2IH0Lz(wfj2}
zU<eI-Wn&0Ge}^*xlXKZH%t-Em04RsBZxID-mrC<BT^~T10`Gj~hH~EXN!f1ujE&{?
zpb=s22Pqe?$1B6sSI^#01S+`94611~KYxq!X5KZTm=_X?Im?JCT(<;793#?@BH9=c
zj}(zcC(bg)!dY&&a+bRgR7}MGtq3ga2rPp5%FX3`bwWA4cpR$cKvlV{DqdC<FRP08
zSCs)(X;8&O70iUH%9;*URe3U$d#Dq|S5e1ec5r<wXzU=4;Df%xJC$4oJz#Hv1qLVg
zK;x4wFgdwL8Xdvt<Q{N9vIS;`_P`HE_pso#em$(5Wl|5AJ;{QvbPJ4kSYUmqg|DnD
zcixYeANCMg9m_0~B2MeElhSTg7M6mhXqRY~f?grhEJdy52<mz)V1=LV@f=GR_gKn=
z7F(qBX$#|bTWDc&uoAATE9a^zAVxM!!C2viWfn@whBQi1Crd%c%QQ<tN6R!zK_8*&
zV1|j6f4YZ-rHgx5ncQNFls;`?e3-%Hpr6X|m7WEPC2M-@L<wG4W1*BRsb?C{dYN7!
z(=0{)QkiBc=sYI3utzydpYE}MrHgw!$K)1Ur1WVE<HHgr2TQ0NKW}{_C2^l|^E)A;
z)0|^7)??fjgq0&ov++Vxs_?V=J~`j>3r}?UoL_d&r_%Yw2jMkpH%nI*QE@oWU0rN;
zA>1w$hS_IqDo3<t7cliyZVc1BB5d7Mx~jM%7V+q>L*;l>8)~%_o(R84(Uuab{nEy0
z<^KY-+V8FZ7ohpX>;VGg{|nHhYnng-{{IDNp1J27pe0O5xIN*vgmhYEslaS%7e2QO
zClIIYGXHXgw5{{z^=w-wewX<Nne!}h3T*N8hTzY@W7|3oTRC12?7*97S3(bc@NQsY
zDlZ16ET6Xme~9TH7p5ce^ohk>3OxwYxJZ$k@S}4Uv8I3}(GH<Wjt}tcMgd;A8#T>_
z?*3W)<S^#xND%7#e!uA_?|-nA?SJ4My?9@5E#B7Kh~f_AeYD-6Ufy3Quu1Qug#ol0
zd<%Wu8^%?=3j<LE>zep>($gbq?c9VNWaAoC!$s~?@d*;xg2Q*Z4HrB=i~V6B+18ND
zO}+FM$+k;JO16;#_PEN#2H1uID3)x?z}s>&c6Kd=46I9o48H9M9A6`=I@8zKFxF;d
z!58>umRVqBQ#UTMj*HwT8FyJ<=)mJM`|x(<6Y#)s)PnS`&!o34PuvQ+9{Rf7Kwo3R
zu-;UG2CIm;0newlFg~_hrSTyQR>sE$nR733<ni%O@H%3AjE5d&eC&|MM?6?;eC+Ta
zA9%+zx&NJmG(JG6<70JSX?$#8scw9rAH3t^7f2Y7j~l5I=3;#O!(==@P$4(r8}y8M
zF_dtT=>5L(ehG~X8XuFS@zDhal5H44Q!nc)*>>5VCEG}WG(L8sr%*sUKE_Md(fCL=
z8XtREvyp2sL^gAgpGx*!b^3BPM83d~IEx{25*3|gqvwvVFjBsiN6JSuQsnns=W=t<
zoH?~T18&4oo&3g){`zk`7SMRQ2CLz!z6ruP?|OVWJm+vdThEdkVtz{%Heh+KIhQ+(
zrz#J7xT;GaXRp4;Vz0gn|My!`g*toX&t3u9pSj6gQC%)q)PNtbn|=geeewSQ9bAk5
zF?4W4E;qDCGk*JSXpb#H&!6~Rz;EHrfucE|Jb&U$E?2l7Iak!O{AN4|2B~035pz8w
zoAHZzV1_76Mv6@S7ge7KeZNZjPEyg|S-!{={G#YP$w-kY_(joo5@c*VcHv{=PvvrF
z*W)Jcv(32Pvf_df|8-V-WnD8j{}L>`I_MXd&Qf#hxr_0r)mi*5Uu^Iel$i68J?=<Z
zShBFo&DL|P=LFf~nz0vvmCc>Ju#=C%YTq$u*O)E1bPb|Q%o9!B6I`MHem>#Euzw_Q
z?{$AaA?Py{=POSImkh;$Jl{fQEy=7DzJ%0}RYBVlT*+%g!J54?RM-xlV^wGn-tt}-
z?1-x?X%OnMT%C|P8FNPk`qmUvqgf5We9{^%)ceY7)(?2$80`Vf|Awn_Lo?P6!tAaw
zSU<o}{5_bD+r?F`0iRz7PtoU=R9pcF>m5-i`P!4gJc>qr$;W-iJ+z9X;=Z;EUpTnL
zCVS8qSSN7A9mbkL*mkTI98QV(k(LXr_P8_LBhNy3RQ0fTkeI}Q*U@I)MkuD*9Z~S`
zCcgGeurnN!jRQ3Z>&2l1AyYkU8TKh&qC)sZjgkFv1roH|UfpLIo=>HB8hga2z)Kab
zwhMQKjjj$|MwcT^m-4kg1jqbF*TBNpVmC<GgRm!Ibd~bKwYW+di>kcmn(p^+(u6OV
z!*JSScJE}b8(|@Hus(4Qx8TpgIQ_Dll!oyFZr`|-3>WHBg`3A#+ulnRaxF2x!BXDU
z-6qaQb>M^nhXhgkd>)oB<_z_GlvEe`9#ooyO|=L625s7Y27>dy5yZKB+zW9cYYl)Q
zVPuHA6dA&v6MHH>XFsI=T18*n*sp@V#v+S-Wiu#9;QDmCqu-X8-|YRib5+z0se)*a
zsz)hq?BSr@c%wC6xeq7Ire?nOU<kbD+2p){7u<a1_d(7GF4Fuw?*B~!vc7kmegC+Y
zs^=@u1eZ)2Ro(xc<EV$P?(#1%u@do#Bj^lNvwaS3?59B=@g8ix{LqZE2dc(E6}^62
z<MeJRxOPuDg8xSRV`XSl4Nlx6Wk00~n;gP6KJF-2^$1ia3+@x!D}O&Lw{m-8wd0y#
zVQb~y5G?eBVlC;~%9BCP-jyeVofnZesp)W{@nPlfp(WRkuI}<)LiKGyJp1JdlE?W~
z7eohj38y93-E-Vtx#=`tS$m{%{ptBT<~9m=|94Q?#SY<!epvArB$(EY!Plzmpkd{m
zI1<CJ!6<X1?^XO-QJ<QnUEo{M=u6XnUFCrYUigg8RpA6LU|DUecqw+F__&?UE3t?a
zMhCrkS2SOpZo{hl3GoqB)6|Q$kanXkVu!8Gx-h{+c{7kFSsTGN!XUH)LJFQ@zw?;*
zAKYaOrQBa?ux04mKi>HR0WK{=U=@-Vn(ee65QiVCbDHe}625wp&9W*24aWu*FKiZ_
zk*F5ChFx^}kd5<P{5!1TU3jF}F8-DrW;Jv3W1t($3S3nrRf3--2(#aW<~c8?b;3^A
zSL-4K^OfsE?E<pg4Xn44F_>NbvBc5fh!U{~ib-nIO4e~)vKvbn+><plkZavHbAPD~
ztL0bF8G_F(*b5TzW0k9+#pLX-;nM;qtv>O}j5V7K6*f>c%&1i1tRrf_UHHmg>p^uG
zuZ^eqpq;SL0Sm)udb)_U@sM~=l*cosdiLs2nhY^#9TxZenP!MSwp4-b!-l?X$8U-N
znNKkHp_rJ8GU%aakPz5!V_S}U#98pByB37F7&-2{&<1xSndvh39%qHmp@w&L9ZSdb
z_J6^FH9MDiq?~8v>ktPWNytJj{nSOZ{PP9atuISwd6pF9p}VGtZ=(Vf{rpz1+6FpS
z)j}2ZGZQZt*T6EwFEldLxOZ|3+pSd)#HJ@DKHIH$jT+l-tw4r&FZ|@jld<G^*b=L;
zrNGA)mwB3;*FqDWw*=n(H8y8%SK9eyrX08$R*?6+9ula@F7x|oUJt!X#{oR_0vbbI
zn}}k$@_IhIT%_elUib)ON%$2fyY!BUE|9`Rc@_43Y1!Cy6X1vTxP=hDI^MzyNm%Pn
zVw?PCanV({W5<H75WhqnD7B{tD`j`xB7s(CyU>}6*@cl|_?L@KM6o*ab+{ma0a*US
zW;@d7*a%p|ex`<gVq+0_YoDY^EQ(x^y(zWYwwW^M^uWYIKYlhhoLa!{0+CJ{@3`XT
z52KBMxP4sJCn!Mo1yryUx6esxcfbtNM{HT&b<l^M3+J^o%hACVY?fQYeqn}W8z3t_
zkAh8CU}(7C01}br1z=NlJyJIDiDgW=@`E8CFbUB_J%Kw`Byp{+JnbYPr-CWc=1ig-
zx~YM3gzv-yC&}vBbQ0k~#<kCpgTh#4>9%6I8F^hvmfZ~CbFOjo>9;Z+aX)ZXD_BLd
zZ5gB{O;S^9Bh~4)3|UP-s39jrD&FYJyn&OH+8Tz=6CXp7roQBg9?TVT9(AnbhnY;L
z;s+<r?^9$fW&tSCw(}KOMr;_QWv?W*+<cn15B@GSpC?1(*Z!57e}QZ_FWa5pMQuLB
zwA$z{1lIiD(0m%$QTVwXI=iAlqaNI3LVCj_z5lpU@&`d2jka#kYv_*%@<%f?h;vXx
z^2hz1C50^S2=XL<OlCU0{)iBp$a*v!f?87#!Bk?yAT8Uh@dsaxbp!Dm#96$H?iGi0
zB#)4`vN@ESKgEnF!n}2?e3G%-eH#0bsC2sds(pc)KW`(=2sDeb3jpn;#Z|3BmiFe6
z{-80BQsVwQzI>(MX+fIy6nv$Zv>@&BlYXiN={bJV54RwFkDv5?El5xElP+jMdV-&H
zW((4z{G`XVAU)hqI;90^>Qwcv9`12Bn?OFWQJ5n>W(ClF1&&7G+fy{`Wr3L_uoWDQ
zz$&`;P7!dBz*2BD0`H&E2wY17OTf_xbkkOGAqjYVZMaY?;O+~7W?zBrT7h~JSm$eG
zomSut68M*|z+bhkc$fsLeOvJttw0V5RD$Eajhv6k6^MESmdl6xs1ZZuESzF~a8$LG
zZ5G!wv&B_x4s%s>iU&HSb|KHbpUz67dAv&6dDbdh--0qa8{Kp*D=YJz6|3myn0-&k
zXIfA;4|z?9+Bdl0#1YeIfM`{8)#$5qY70tJB~|Xf)AR<_^@?gePE@_?r<X#gKyEBT
zUQ-%r-0#)sJJVJ*_-TB-1&vQgdfW$)uKPRSN4n%HfOl7}A~brj4;vJ2m*j8v89B{I
z0nI6}n)Ki#+KFtj^JN%4<W7yi3$lO(srVC=x+%uh2+&+b8fr+xCK8yg6_8@cM#@}D
zCK*K+Ro9b_O7A#w`uuDdP44m>?-MI~pp<GBN%?51`5=x|)1;asP*@zuYSyEA)ml<;
z&ecj`(+4&yY5u%7koG=ZofOrGtB~V|oGr!R#?6#kv79vKF^%fuSkL#8+_{8Wp+V>N
z*0Gf82&FnOr!0oYwCfYv^?Qmw?ZR(%!DDw_o|^q<3Rp0hah1CEq6;zfOwci=y8jQv
z!WBF&bPtIXKVrqpAi{VP+EU{V;2s#Li57n#1L#6yxha7jU!s=dP8+lk`yuNt!V&I#
zL;(6WTE7m#g-k96>9q#e;kdcCHyKlk4E&NIGX~cTd&C`dZX-s4j5new8;NcH1lNtL
zXjx(<two4;9D^bAHd15ND>~l`x*^`v-C>Kwg>PgN#i;7k>NFBihYI(2Cl6}C_^n43
zqN^vy?{Qq(Y6SAM0(X-Dp8EDQ-Na@)eW+H&f+ubDWBu(*LeVCtT;9h6&bxd}AM9WD
z&#Gl}2dO9g=MmC)DS(^3`4>wu;eX`JvB)IXk3=f3=h8mm611BIyW%0VjXFI{?({{-
zQ93<L)9L?)LQSW?LHg%yBt7zijrx0b`t_UEXtvmd0T_m5)`~yjVGf?V#$mHi0<*BA
zVSXWub~rL{NAAXm4U29YkLmoFYw)^jFvwzR#MwPyimTMJZkR+DfOeIJtxJ>0$spI&
z;L>s}U6Y89ffP0|2F{IQ)Q2vae{fDub$tw1yy8;QS-Bw$dv-@+JmN}_n*3YXT`BIT
zz@_`!mHk&*lA&|+bgJ-cYScb^?HM!HUFMY8s}aO|*BjXoC_O?e$O)*P7Fp>0M-!Tf
zH$dhwv~~?9Rxg=FeFHJJ_}oe6Jp`V;a-)^6PO@ULaqiz>Hm<h6#e$n73irF=wmdBG
zI)~7GX;`wWYz`^Qu)BuXNbh<y05_J6MtQ8mOT2d_-i=z`3li_jRN;8j<+L;k{Y{sW
zL=%T^OVb7Tj%W%&zC&pAp0aQm*2kEC(#NN<n+H9q)!3C~>)BWks2dpp^RY>k8Hx4f
zk&$4;5(Af8;y(12ohA$WwRp`XR@Sh55N~&ljkLQaMB}bOX-8l)b_8aH(+K~|&om8<
zjBmQI6lYhLd3IrJbV6NcXu)myBdw&C?xKsbyTW63Xy@0;VNV2=UCz4(*M&aAq;P*=
z-1-txSzX1Y(1vwWF__jm@wEyI1g4cQ;z;~wcedJ&@ghI}K(unp(ku|<<8APr?|Noy
z=fk<?r>)zo8fPWrek9|1pnl2!+WaWWA94}$f5UE+fBpc<UrPC3s`=+?PaU}&KBsT2
zGcTh3BCWrDc84%Nyz9GkUfYE;j<_bS>MRO7F#KH)gr(w)io(d+NBcB(*Mk@qDcCc@
zW&xI4<2KD27xN8PQk}p%tiXylbMrr^ve-IKyp)~Ysc|QqG(;Q%HbeG`3&ABVQm1XZ
zm^yJXW<y#FO|hg_=Y%zt+pEiPMx+3>aD(be>|aRxQ|{7!wD=s??2Lgj@nY&@$wt|%
zu%_N<16eajuI&)~NT<tHynwAH4A--GuHdvI=r?=R2`5(5>KwvJaSMc+LLkQ1?hCeu
z%miNd#~G>)!O7C~xY!RB(tIO$5^l#-lW-e0z!K6O)mNEft)|=f>H!f*$2+3-!6UV2
zaONKf6EE!sC%~Y~mGyRquphQ1Lx-pPw>X=F__(IB{cN;P#+5&-3q7XJ=f^hS%z-A7
zz#(Q)z_{i^-L3rxS)uy}$#`-aon$ga<J~9P5Ao;|odTSltNoYw-f!g5XD<6jD&-ux
zMNZeDstGub$A9MV)<357oKk0g;dJY)q@tO{1v%v-tbVLy4QpIZR)JQ8Xyji*AJd(t
z$ySqf?C5@`glTu1rsd>jl@&NmvvRVXMI}RrW~OG!tjw(9;+(>1ru6)oIi;r5{KB$Q
zQ(6JOl8OrNHaUsQn0D#dF15|~yErU?F`|3Mx0qT*DW3`9Ys`?5CHd20=_@g-G)LDD
zO8%gjq^uHWPHBEtA^bA4tS~=23ypwuLt++FWs&Qb^u<Jz!t}h7oUCb*N<}VSmQ(6q
zo-CC=(pj>f_%x!TWw~IWJ$*71S|a_vEx6XAk9B`MOQYp9UQW~G)VTdIgUMdCjT$ec
zDfY?rD}H8`;44h`PyC9ifeD3$!-vP1NNZ+(VZL)<cD5<IsIb&&Ep<*CI#e11!}1HA
zH(N<&=urAjmPW|<{F&1tk=5UduPD=i0W*v9@jV?9NH8o-_;)hjmJTe<zdt83it3tG
zls`?QEG?^~G{=#3Us0JeGCMD;#Cpv&Oksb23$8ZeQN`18rj|_~P?|rzFsq={OFO4Q
zsVOO|Y<eEr!*Z=xO}o%!wa(0$Y0a88tt6+k)OuAxK{@f{Pj=C?943`nTvC(`fufSq
zJEu)1%d<5@>N@5!Ehn?Es4$28Cpk*(B<(;PI@FLI>F4aLqfF6SPZ<^)I&?fv$Mm9c
zIoY#DloZX3WUf`*=R>d_(s;g2`Vw=d=NJ0eMr|n_mvc`UW}i%q3G|3hH9kbU>O>8g
z&YU?dR~uyt>K$^E6DMFQ#<BGxkEaI5i^4sg_zO*@<wzG{(L0*+c%~roNkh5_`SBKy
zCoBYfTs_1iP3!CNlp}4xM$K}faU+6=v<9~sT0<c>+~dhWdL$X;kxs!A@AXI<#^VVx
zGb+Br<B3HYJ;mcGrZm^%SxRYv$J2<kxWwa$=mfjUJ)ShA%jbAJ3y{`40(+4*Eb@4o
zkv2Z>@kC?Yz3~-~ClBeOw>_RENaH{7cy=N^vc}`F-~?W@3GG2TWi#wXnpO`vr14uk
zo)t)~+dQ5Iq>Ve!&I?e_S7-;)#$Bi%sr6frr-IV&U?<XOq%}z6k?usAhV%&1DM-WM
z$3=V4E~H2B=*?23roE6yYC*aaX$4XfuCSLNjX-*2AL=3c7qpY;V~|Ihb{zgdIt6Jn
z(qg0*^iKuSNTkb=##4X&3V%ayJklwkr~HO`krtnTJT`SIkVYb1gft%M5~LYOmm@7k
zT7z@}(nh39keZs|7o=9C4M?Mrib&&;nrK`gjX*jDX*AMeq&(6Jq&vy3-{EJ_){|%l
zQu<(jk$r;hA7=_Gw*>X>(m8BVP-pg-ZXe{V=;HBQBg;@?EN?Q|kVo<vVXz%z&J=DL
z5gsv$yXfAq3ezpUZn|c$?Fw+245cCWr#V=SokJ|W;!Pn^y(ZHT<dLrEt{%_#==;!b
zdnK3>!Yzs65sBf}#PG<(@aTl_*pcCj%$1!NcB(c%7W#O|!@;cFB$N{oS9Zr349)0m
zR4zF@&8HYE_cY2iVA;iqu^C!rSZ-u^V~|flvh!p7M_?m~l}qkpRG(qF&>-{??kAW=
z^@U#6=5oVO?ztGb+{U(-bD~^BtjF^KYv1K<u5TI2WyH(%CHHM}xgSw((MY9z);5>x
zgE_5!jK?#O`fp=^ay9<_kdAUS=^jsiDz~(saeW5m7NXn=%z39Vw}mdeqRr(tq8z<r
znw2ZQvd!gAqa2SpbvW6XZfkS7AvBj`4(>+fVy|j*xdN1n#hlH`^|`vu<z7QMYq`hM
z33F$t>CeXH8iM`Khi_4?xk4^i)An-R34;W=Tt#G?>vN#o(kDHh-)NqQk7{$dDwHdJ
z(c^iY%9*ZdbGbDrC%!6|t7&_=<0v;}nOv^o+BVnMALDZ82OiHetbP64R8E<nLwx3Q
zY2K#!Ir2k~XB$$QQ!6nZZ8&@y^)!ON6X$KHX+VN0G2D_69x*b!CbBZ*F$m#9dTD-L
zx(=914EaNRC;lsHJf1JmPUSNM2Mh7f1LIc{-+vBI2#@y}LaZq1+lc=L!~kDv^z{et
zTl_zVnBX!p2{x*jkPsd{GJJVZWv7Ma>d?nR9!EaOcR_m=G<ZDMw?W<pc{aZy*7zOe
zy!Ix9M^-8*Bl%$(@{2$Bcre`657|8%{5%8x)8J1r;IsKv$0xsSgnSzK{g8U?P_Zwd
z-%dlG4@HcGVpNV2va37V->~1~ks+Vh4EOqjAMu69GlBeb4*fF?^6_8cd}W%XFV=qW
zV|9G;>v7;tH28Tq4xOZqujKG(b$Bs5*Pwhk<hqecd3=&ThG6_xfIkKf4}A^g{cXiq
z4e=RFRIUK!p8MM4c?~gIXtS;y8#9f;G-l)wf@)=cc@5<(KViiP_JmHkj-3y_W37Uf
zQ^p#VqjPIE?28mVo^iy0jciOp2Ytvti;<s)eCk%^L;MfGpJKrO82k(a{*U0Nflv3B
znv%n<lu-SrW&QUO1NGw^P6)SFQh3D9hdwx;cA{J_GC+6!XwMxgPIl4!8WHUAln?{@
zD53LeBXD84+v5p_p3tRQ|KMDKUn|WE<+D)o=hM)$qzU#?z5aa{a6EsEav2!6dVl>0
zewqRQH2Ay$zdPo)cmqDoudxRFWbmWG??nxk+fDwMB+KjhPFcU6zYu&QdtQ^}_3c_C
z>(}!?2cPWGp1)+z0q`TiN4KjV;)mkIv>LRl5BT*a-}xaK@}J}Uh(`b9kQR)K$AeSC
zBOVJ$3Aa{<ri4c>H1pxn4|n3jV=FtShQ}9#$D)tX(<~lGRE~K#jE$E*UC>|9BM*7f
z#y*~q6BVYAGCKZ9;z$R70`W8E2GidE*WT4Y$x&75ni>8S0!fqv17h>1pfXJN{A31K
z>B*#n14A-Q4-xUFQqw<4lb-3ZyJs>ZyKHtr&WZ<#pa{w?*<D1@)y3nUu<Kdfcu+)9
zj3|h#Kd8GbKQ2UFF&u^L_uaa;r>eR-sAqBYIPY}Tt9##l_q})DefRxzy;r!_;Q9_4
zo^vvapMfu~1mF}rJxQ{61K>317X)?z_*WEY8{<@Q+XP?Sy#OE?pWFV9(>h*3v~+s*
zun8h-qxtHD$~+8C;`Vp^%U^6qPItAv9xx_u&Ra-xkpeD(JTAN??RW&P#KjDi2z*R|
zSpr|hMFy7<;5$fM484%RTOplS!S7%pt^x2C_#a>hr|gp;%@P<!32uZhWIKRiiG=KF
z1tjEI0J8`Y_dChI8_fT;&vX9IqbN-N8wi;E&nIB=e<J~tf02O6zed31|9S#6>XiS_
z2>hAm|4IToRsKT+22}n+wjCJCU&tO+Kti4XZ~*!Lq38dvDGHPSO$1E-e@npRe=7l#
z|1JV1|2qhn{I4f)gUWx1`u|mx|AP$KtMb2=z&lm`LbeAO%3sI^5Rrg{ybi$cR{oiw
z{>=a3>Cf|;|A*Q-zT1{T<<RyW&sw70Bak5A;iA1fk-)qHuOx7d0=)$O0T<(B2z0<A
zu80330`~wo75?1>I4+>^bP0SAHNn2`a02%!AY?0`R$sA@Lk@z?2OuGR063nJI2Rxy
z;|jNr<LM<vb|-)>0L1O@2$k_TMzJzVfvt>EU@POPB(pMh6R<K45wJ2YBk&b1<Ch3L
zqh)*x0X7|$@dX6Fp@5LRSj#9O2Wc54q({prAz4IZT;cXHJh6;I_IUuU%J^fjXq52@
z0#?So1gwnv2v`|An2eQCs?N&z1CrgKWxRo8U)D07PN1m8`ZmdSDj;M(0Kj^bkgub_
ztVaoH11;-OLQVn~G7@f=;E81vGEQ^<n9C><F<m`_GO~J160mw4AYk>#p>U%f#|c<H
zN*jJq>+wN`9F7WNdz>QhVXeAV1Xy)cTp>GAbyY%^qv}|X62jA`SdS8tM?}UIZX6l2
z9);{a0Ilk=jZv(OqJ&n)gBfCFTtdLgcnks4RcWbK#;2%+pVTsbf@D7gz%s5OaH$qc
z+QdIAAY@Mipsq^DqpGVC@)9khgz(Q6G7@e(@We6-*{uNn$jWGAZyRoO90p-jaIwcY
zn82SZa43PRaj}y)iomV7NVbx|XB0S@0EgReLZv;+;YLxl?Iwoo1i-3%4S~Ocytz#N
zG|86Wb}4)zOCp$cBxLVXKtlLJTp1za7RKI-`G-JQ8UN4n{~4>&^ylXUOn+qlX!^6C
zA*Mf~S*AZPrec`>JVCNItNvWgkn2=`UPWL|_2*HNJ%iyY^+(7~Krr=3$ljrVgj@+A
zMt`W2RP4Vet12$a@<Rmpk_~|m6L>E!lKmrr+i)@Db^>=Q@Oc7U+dCJE{Wk=T(fpsq
zERSja|D7RcB8w9syMe&RHUC2Pbzqo(A?wxrOURjv`G;J7Z{>e2tJ3u6eFRK@-cP{v
z=OzNCKcZQtKm5C}p+D~?aH#6f8PuOa)t|c=avD@52mN^`0nQq!|3daTV5mPrc8cnc
zguF7WKlg9n?sTlD7QYBr5zL$fuA^{mC9n?HDY!Ne_$e-G|3?6{agP>(w-R_WZfxk{
z@C>)V;|{PM<m0P(qvHe9dJTaATx1|_2>0c<HW9i27qcU7f5#r=kdLE2!hg6O0EYt3
zp-3)oA<Xm3EaYkc6avS+4*w4LG#}{pcl_`#k-6`;uL8h4WGF%MCon4F!~;X$MBuZy
zR^Zx1;1{@7;>xf-js#0)^=twu07t_gW;*BNMg|3}+lnk1vVmm(gpj51)8xj#9XLw+
z7y<qlL~gec;K+_4?D3qJ0brcJBaj9_Vev0mhi7~<JCjVOrXe3C*{1+B(zycwx$PtC
zZvkLB&k#5kBPE9XlyS}gz&QUyU=#q;$-{T}zb4a}VLF#-$gfCthXQLEavy-Nku9C-
zz;SqF3hzSkWUZfRUq#@fxCq=u;B&aB9bYAI53Wo2K)1i+MyF#{H=g*8^mmEcitGJ+
zSagdLSd_q`1QsQ*D1k)@EJ|Qe0*ew@l)$0{7A5e<BY~yIrKPYt@KQV7ZR&R&>%)81
zmq6TXUK#b}KafCp50?)D@c9ZK;iH{>E7J1BSu5SihYX6(e-!BMdaWn^v2yp9R`|$u
z0lG&toD*a^&Zg;p{_K9Pp*q*Mr|HQm2K}xVr0F?W>Jj%yds=*3Mw!uo!v%|_NA<b%
z5OkQM`!4P056c1KmTLJpS*NpkJm>v%k1R`zZ}_DOTfQHVQ2bY_zNP={e!o0jOa1uR
zzbNIyVwSkQ3g2>+AFk-<-P@<~k0#W&($B{_^*5?tRDV|eE7ZSE{ae)ErT(4j->?3|
z>OZCaA;<XXtWf{u>i4O?PW_GQ7uBCt{|fc5Q~wtAcd37;`uD5<u=-D_e~9j%T%rEU
z)$db(o%$QqFRDMQ{uSz9r~WPK?^6Fx_3u~zVfCL<|BzK$e)V6jexLg5)ZeInQT<u<
zuTVc)e`gO3otZfGf{B@_`b?syv%fQSdU9Y!z~pH@!qL!N+2aP-LbX<^OeMNIdpo;M
z@9{}fm)5H@wfcN2acbS$TsvbdGm;t1X58-Ht}}`y>=HaJaq0xNk>(SKIWn^GJU{wr
z_;a;3KUw!C;Mc3Zzr~)gpRG`xb~>jj^+M<1n&H#ITA|HTGo2GNrE>ms?0xJcw8g7!
zaXRz!Q{d<O^{OAD84dw1!m5R`#|Zi~jV-vH+>+^Z)(dmEBi|*2Rq|flb2<xK++x+6
zEVx_ph-(m-=5lUfE?1bw{ycA@T=03gUbX7Y`#kLV;v6C(J#Vs<L%a%-z^zl#_0yQB
z)tpXlaV$*LqxJOL!p9;aZMWI2{`NE=tPU!|?5QAeBW=)~1b{Zo;*Ynd0qKG&`~-o3
zGdpFdBp!nC>+NX?;$A|!lYOsokH%k%5Nu1USo+tur-k-<MT`oxVE4_wDq3F6p^dlr
z)(<?Q2zloY;p`4rc8h-up6NE?Vjp1ftzYR<!)^TtY3ZW!dn<6X;TGTQ_AUJ10PYIk
zYxG-2XFeBi*(Vrm{X!R)p5Y#nyPto`&hoEl=k0hP-r`$7v{YVV$2z(aYIoL%Z{@!{
z65sl@M>M|mgN)C<-qN@D?*@)Cv3}Ub0sHtT0o*FzYuuUFpnE5PVElyk2MO(HWiMYK
zf{H~}=81l9B)*M%c4)%aJRXRQM~iRR4@BbIxaH(S{F=7@+vubD->LCU{#`om>C*Un
zOz_Gen*K*3@tq?EeM)W~fwOkxBmKIzXZpKvk<!wiJJOFo$G>CX@-{#dXrcZTbavw*
z7@yl6>AoW`>Ej*m7Yo?9MazFLLXN~`<0u>F-NnNR;SzFRxS9SK=6iTxQ?d9q4tzx8
zN9m*RV~ER?EdE{{$L#$z<3EYG)Ln~j<Dh*S-vn#=ZRuM*?u(?q?<@}({JC_NUw)$(
zByI+q{|s=!{5v`hb<%!(`8&uW$Hb?a1>@f}=n3^*gMNI=r?Ief4W^nAXXSrL<3FJB
z6A}GMSVZ-gAwE;;ir_6i|MLwxKUaPn;%R0q?I7J7*7%gFhK(SLZwSJs`-Y?6&zBD1
zE>t^mJf3Omt^8)c*uLY^-@&&L;6`35UgH)`zY!4g&;iNxx2KyzI*?=QXM=$q$cgoz
zVf;aj`A-;su=7ZJI*5|z4rsLXW5K`<^wicrh4F_u*6)PzFL10s3FBYbnE!<Fhd1Vd
zVf+z}+4V4fiDULUj6bq5e+=V~a?HMl@z8R!n}L81XqVZ;02)$W4*ND3M9Fgp{(v?6
z6%6daU$$nK!uX>dvo~S<3dih77{3yB!3(0~xr4J)YxjZB4!n<N?Kyx({#QBHPQ!SP
z2E&f}dXV#z{b8(cv-y{HR1OItT<&oDdeEg5&+W}Ymr*>oF9+R)is$y^pyPOl>2mvV
z&|NCPa);ZCgYG?w=l0>CyIJwv9vpPHE1uhbgYL_U=l0&9lYh{WA8y|by1fD{ceout
z=-S%|;Z*=^zYV&h70+$CL3gU+x$QRSUZeQ*tHb#9isv@ipz{>ZZT>;Wk8dzP+{Pbt
z?@&Cq=?C5W6@U5hVf?L%=QjJG<N2h*KM}^un+Zsl+vJ1p`=D>fy9?H?6B_tS;M<_L
zH)lM-02KE^81Fd!v~qE~X3$ywxeYVuET7zF8FZFUZleskl}MNQwDqy726h9_eDVlN
zI_oD!70>f2gLo^~ZpF7rC|(<GQ7(>^C{J!%47&Geer%mCp@GJKUxfc>Tk!vq<g?i>
z-LL$&N7nxySNv@e{IiOWt{3u6f68ZPq#jQae6yWA2YAZa)+MdX8->2v-nJ^&jL<K0
z9*Ed&zJAk&SB>KAaGYLUE%eRy_Cvt)1)Qjz-=+9yzwjmD(`@JO1wIjDx4#QKtVWYN
z3<K)r)WZ&6A54E<EBGdI=X|A492BO%1o#u#?k-PDB%ZlI_o)_qj=;PmE<fD95Qkq4
z{0SJT+j(2oF2)pZ=XBY4^J3w%#7XFU2<jy6?JfAfN9ji+dU(C?U+C|)E1z$*;Pa~%
zc%B<XIXfyho0sN*r`$GOm=?}F6NT=Bitj3?1twp*ZvroJILwznU$LY61@H-Z3f~R^
zKk}cxG%W#EFJ}Rt2$Kqw1wKyx1?BV5htfjFGbiZyUgQaAkDJzb0*}fa`|Vrs`48o@
zcQh@OR^FF_UtGF;Q!);JP7C}cf?wi{s+@g$hg}RU=<f!e^4a|(KR<G|0&>Q8_?Ba4
zY4dT(EN(TyFLAba3<|1E19<)t`RutdEf}61LiY`!_n(9JeJ$`S@dpC=jJ_%@<UCu4
zF01%GZdzcT5kq%1@YILf4)ycX@1vcYfv3JrN80I=ir@Z3S}1jy#o>8Rq~H0BkLNZ~
zx|HDk=iq$=_!HZgIO%Vu1r1PKj(G67?aH*kCWosOzkATfTfKi#@#)8WJj|51Ujom1
zA3Y}RBlOPsV4OZ|13nR=^nq)Er+n;OKeG?JfM-6tPDu+P&rG6w6!>MNSGxi8C2k+^
zl+UJiU(WtG4BM>Gq~r240zA{T^94=LUJLpf@Ci*T=v@W;(inUG0pYX6c}Vrp?8##-
z_^&`ikor~nHcey*c+t0r{A)^|j_A)v1Ru<y|9q#??~de$oAzlRcCYb-@IsB6#G(t+
zb&6@h@a!VGD)7wbjz~Ul0$%i7?VjoRr<KpB>Y+bw#yi5OX!7Zb<Ub>LKds>X4NAZF
zCuu?R%nZ7B0ndEe`HfGj;BE(=`LS~)ZJFw8E%-kP{Id3!I1dc^a`3r3&M#Zg9}9(9
z2Dv4Me0m<gPnQBdPTz)G;EUvgboWHkt+k-PLhxw62M<bm1-WClhvK(8KK=~F-)i)V
z-=_F)x8UD}%uBngriIe_wTpq5_M-Kx%O=h{fM-39Dt{G&bF<(zv4D4<@=rwcdB4)z
zIi(im_!nt@BJ>-9XS&W4X(3hhblxU-i@o6AqkJCvgip`2mFV`hpkD%op&WL6JuP(i
z$lY;HQhZmNFK09#aeo3l^Rst<THrAs?Q8;`a<lV6tzFD0pY5vuXQ<$=0iJqwTgDR(
zX5T(3{FQ0IyIb%`_rKGEH97nj@Jqobs?R@b!RN@Mna@>Ddaa+X*?}S8DTh5NpTG5Y
zmjjO~YjQU$eb=y0FPk(#^*zOJ(t12W%X`=gpMQF{PjB_o4SXE`4Dbo?i@J+i&{qYI
zej%ct*8<OSJ@lVxVI^lU;i(sj&3fFV^<w?qp93%S-}CA1?5AD8Q*JvWa{ICJ*{S-;
zvw`SNI>zU-``Wa?*8i>tp7rId_vJ5vxMhJSe@FFUy9QhZJoCBZ@w8y%j3hjLRPag^
z@cs>W=0EXZTCj5F6`o#)7ck=FTm+u<Y1MOEXT2GC>L<_27pLv$3oZCO-U7c0MIxW=
z$|vD-bB2Lux?OW=!Ms)#`ZD0-@_D1u@4d^X=UHTQUuZ%9BjDLzN880?O22efTH@R7
za4eaL9Pab+c81vNfv5bZBYHKV^rL6`^fr$ApyD^(>*Kk9h3<>O$A1pqA5{L)@yt>9
z!!}M1uV{hK2p;+D7!>xJtW#}4|E?DJ&$PgQy9NHoz*BBJZcj_P)~;7#y_|Aes_kNn
z7J5+e>EnEQo~=o@PVh17#&RBT4M!r&5x_%3p?~4K9K6>M!r;v1C?<v|hAPz@vAY@7
z(~RnEM)ftLRyU*in^6NzsOD^?n#+@FE>9{ZVsm*?&6#f|YA$ngtmYCo7r42+1CD$4
z$c8n8Bd%MUnQ({QP-VLpW$PpTu3M_e>lzYfXKBa9YlL0oA3{Eo?2|Zl$~unG&Nyx(
zaCIoK?A7a;U-u;HxYYSRhn<VwII=odt=H-^I6gn;xb8VaWA4ascHDJw7JL<FK-ciz
zi(8*`b7h=xU2|ekuA8s8o6D66uk7Y=;CjvVX6BenIm<cU**CB{1<o-<E?;Z$UrBNC
znn`uu@n37n&rD9vBT4|01Ik0BuDfn*@VtzhS-%#U^>ePzT3;L1t;=S{-SNRSBN<Xq
z_6-*KFrrq++3}9M_KoWY&l?_c_zsM#IMi^e%XM*>d1aehE=+B%Z*dFNYNa}QZZ4fZ
zdt`Xc(1jPe-JLy+dTVhKeX{p#ky3VMa-*EDpBc@Ua8NYfEh;J_H`STy?0qgqtJmO_
zo>HNjnFVDZzXnjK4rlADeVNY<=QF-?4cG?))@D{0XQpyE;XgAwcJ6o*(zs}Sww~<A
zdr+ku3W_{z#=YQ84(EVR@-TL{UiC`#+HfwD^>E7!B{O5^W@`xUc84o9cLTl$Fo^@x
zn`Dr!lxvw%7>z^MYlV8Wl;gvC6o2$+d-tM7Q0SC?*DaSOa&E3RBf8^Ec)6{}X3jEU
z^0TTN--zt?21_|HQ*2$taGu$=lYLh66JFlUdA0g%m}t6EE6uqmo>#^p^U(tO#cQDt
zsZ{s~_aSXEnXzoW&p&;CxCGgkGBgP>^<0eB^SYRNj)@neAN|x1e$Qb+LB_KO%sukt
zZl0`6`OSn)FfNO&lHyk*8cCfq>yD|Z87h0V8tlyUOuZHrrj?}V=c{j4K5P+F<neJ-
z+;61Y;W~9M%9lqY^&DHem<kX<21{3%D@FC*Wf`k=x9FA1(L%02H(T$`)#s-RXme<3
z=egTT`NEVKuS_^koyjEIv3wk(g^DFp*)64Fy^tUCrnaI{mc~k1wEoS7I=|RaDdxTT
zrVI$bWLL3NE@yLI*`sMm^`hG-R;rUOYy`|ax`k?KVy2FlFJWzHAc4wx)x0}VuH?4D
zn&ayM1xD;cr^yytM>$|PK~tT#;K`B6p8m|pWG}8{Dua7>{{Wwp8Ms+AgMdjJ8+XUX
zyP#(BF648SyV=whO92gysL%klU%qV8&`H|I{u0a_Ulr{cM%MxR7nVO7fLJ3jHtwWX
zYHI$bv4Mibs0?In$qz529}_pGF~!Fn-{@;a;FBrOA+S_()V_EzRH;<?MG2}*rW9+7
zeD28sj!uRq^BbnA2$?x(MJ+QrnuCh5r)Vgr(D@oXS{elbB$u1>Tn{g|Ha8A=wHJy~
zF1zU0Ch^8@z0kVqVhmS|<YJ^QbuOQ*S&+GwE8H8Ux`(zDa$5`4H5DlMTKV>lCj~B1
zrJD3)ZL^-BBPrL0i!og*ZGE!a@1QE$14GNvMQ5^ol;T1Y7<V%fGc`QU&-2)jZ2Y_-
zj?u>D&EJI%w1s6@7-dGjD>Iae>3_04E`E+XyaDDuUz&1f@cwk8ep@6F%W0v@cJX%f
z1cnMx17!V9KzWPRLbQix+2c#@h6Zm3Hje}09#}ql5>x!q*;2OG)#|SGptqV%cFDJq
zP-T^Qj*B=%id4^mP4hrwFqj#;xFO0&Wh7Szbx&+LjJThX1KRzbHPwoj=Xm7}m1;Sk
zDWM3Wtff8-Q!3R$#FjnxpflMMmQ{vAN%}&8y@5Q?rsuqg(rhxh8qe&=UG`&W_{9oj
zgHbYGoOhF*-N;Bx8;!GWy$#Fge_SfsR9k3XP_|T4vxgDsN*j1SI<H(=hUN#<y@5sz
z`IVFNCYqXp+M{0D*}%7wXk%jxGgP#pM~%kQR~eql7v^G=nsU)0Uied3uxYT~()D^!
zl;&xU3sYnBF4(^sdPWenXaw1MPY$1A3bze>kO@NqbQ;mdpRErxjDAyZo3B;eE#6cf
zI~ypi=%i_3ZorvdLlmoFtMFN?e6UJVGOw;-jwfG+Qm$3h!~if<ShqgP*_9aX{QtG#
z{x#J<u8V3^S-6u5m~*Lm_iZz!LM>OoEI(hspq|oTPg8~XDwF6nW=pxk08AK$g0mjf
zHB>*H$^M4;B5fww6SPPLSQhQfr)R1KceYfm&v+Qrh8lZpx0*}^8DR|BJmt}NV(9d{
zW9MOg#IG05%(TaZh2eNQZeADLzq;sQ*}aaQo9%XO1n$q_JF(o;sGYo@0e_;gHq$iJ
zz^rewG8^rg)a*A};M$alKT<*q@C~`tCn^SCGc$G6K|gY2v*89)HC+*oUY1>vKfTvs
zJQdyl#$7*`Fz~Irnjg2+GhHZbja2+Xt&7c3p^tFg2~u~lDX6h3vOZhsEn>pBrI|&f
zJK*X`t@K;;2r;9!<N<nwz<`~i4QX_sRs%z-m;_o(1tNk*4;TG1VrDXsMnrR(nU*V?
zP4R5a!qxBPxmyeKZVoji!`}7TQni*Di!7%!M)R^d;?7LPG<Q{i->-RGG1HjG9e8+@
zN^<_ec}Qqt65o%(wEadCO^i%gTTwkty6)`cb1JHD6jIWKG$ud8Q*~X_8_U*W=S()@
ziJN+cry=~zCNl?ny|8qJrMVVMS1pE}SZzep4}7i>14pM+o5obrI%Cs3b`7TB&&O~G
zgJT%lOr}+<fhl2Eq*gqiK_yql-B9}#yNq07QMfdfALb;phPB;nia#uAcDYmm?cKZ_
zzuoATFp}UGK(QHeGX$#jmXxl)v2f@@szofNf&lWyz|~*p<!T>hpp{&c8ENiDX$!b0
z5ivR1iDv!z0*qkRE#RA&^O&|w!-`g_-sVD6-I<mIj8x3HxY?dy)FRAd7FmsD#3nVZ
z2+Ndwf}efF<h!2R67IwcvMm%#esgo~G`8JVu#vWe!pwGs>8MM;VOt!&cotPYzA$+8
zha8kC(s(<jDySSQML;ky*1M^TwrclTW((DlR~}n`cCs&9kVP>JhSthhND`UaHW|{o
zfY!u*G?Po>3r(l7R(rwJHYh1(iM5$Yjt=l=Uw3C0nSmHO<&_I@iv*^iu}KA|G(}-m
z>Bb^yrui?3rePlSlOtkTz7LgeK}BoZylIrhW+3{91_n5}>A(tMlK*j7O+@r5Zm_B)
Zl1{2&Do4HfGUleig(9w<Ig&@`e*w$#_>2Gm

diff --git a/tests/Grid_stencil b/tests/Grid_stencil
deleted file mode 100755
index 92d844eedd06f93a24b40886982bad57a5f3fb02..0000000000000000000000000000000000000000
GIT binary patch
literal 0
HcmV?d00001

literal 97307
zcmeEv4R}=5wf0OhK-B1&D2Ax0u^nw93W=g-3StH(a1P8!5YS-JfB{S@LL?JJMF>n%
zIXxMgw)9q7+S1#fYZVK%whgE-AwU9F3|JMVNW{N0j6gso0b}NS-+j(ZP6Dy^ySLAG
zzx&*Q%szYVwf9;3Z|$|$T6@pZ0^isqyWJN5B-uvU7~613z5p5DHrtFFyK(YudA4(G
z`L-)<gKhnR)enDhynOR19(I`7;$aGK`ANo~1Alo*j4{k-kO&8hm<?|*KJm9R<omim
z`R1b$4!+93jE6Q`rbs80w1%%Y(!<x!6!3T`hyf(iDZqRZ4<wKII?Q|>X1;i6GwF|q
z@%piD{F`E?Pf5y0jPZO1Hm~^Z4vEW8;Rk_Pd@ghtbQU^12HXrCY&d)pp%JxBLO#;@
zRX_1+-eJD4uXf2d>5qp7RW{qOxn=j=Jbc*PS=Y}kn^(E$`bBv+Uw`xPp%wFo-XQAD
z@`*d{_NjuTcsphw#`+D!AFIOjZM#0I7?yY28zbio_|w=IQu8lf{?c7cy9s|RgY{Bp
zI%mvm@Y)Hx1%KbfU-Pna_x`NprR^&pj5Qy1-uCm*wEW}meCr39Sp~2A20ocO##Y+z
zw?7|TUjK;lz@LGB)=xhWbKzN1e$@y31PD4C{?#D(Z1BcD%F+9P_v@qlWqshYoioou
z|CN2f>-s2XL?7_y`hfqa5BLVW_iXL3rw{xG`l#>M`@m1d<Ud>aPxpa;V;}8sx)1y-
z`Y3;5ALT6RqrUt4sMq#B;Lr5|f2R*RAL#@C@;=}v`Y2~lAN05P0bkU|IG@@_`DJ~y
z+edxS`Sm{9zqpV0{<shLJAKgqmp<s61qOAtaW$-u@=N<D=a+rdtFRAzt&j4b?Sr0O
zeU#(t13n7nTxGk|w)TdZXc3#ONPOFDf5&?X&m7(p{-TK=&NbjwCjLz(epN60g}_fk
zIWulD@H0*P|1k0M&G#0XaMf<Kb4z-@2{D4BW;wPSjdGm8<L77iW53K8l#lQ3Plrio
zt4U9lne-&!#Lv&jHz8(#^{OH;t7OO!u2DDJii_`maQ?jFionbTf#PCY@%V+scg()O
ztRgUb!I-%-D=KDJ*ou`Y{^D7)7ntBF{xNgs&zn7E=6!Q#8);`xDJ~C`E|@)YR`CO~
zA1N-uQvm39-#)owINng6n^Vj}D*_AVKT>eVgvm3Nk^&G@vS?;;dF6uH#S6<81S)6F
zWsV7x12>ff9x0y<l-wJU?cuUnv*#6;%_|EOEEGkS4KJTP`vDP|9LT+;xEODq|8Vhr
zc=^n^bLW>BSo3e1Gq<v$G>&$&6|H1uMSxhy4)}bKHGAfRctyqR1p%c5(A<*JnG1>o
z3ucxDDnLiU9e~|<-^_}#k~p?9rFcqCalz!9iwh@AzOfjji`p!hEr=0iyV0r-md!&;
z7!(Ea%H~&q%!=8=71YL?Yt|=lGl-r!cXk3$F($X5a6(C<o!BT9x0K9(uzco%*-SB-
z;pVdY=gkMeY?TES1$UH*lt6BtK`Sa&Ht&9;dS|Pb=!XKl4X>(T6^j*6II(>8ya(sc
zV*gFK^LBI{YPM+RePs)C+1Mo<C&la_G}xT^@ou~i<7HlP$;?1Wsi48=_ZxDGCl}9t
zusrZc@q*dq7<cm*%)Ea#^W48^QE~a~1r_t>&750?<b^qAwu*;mmZQBf5b<`<bd$9z
z_f^j6yB^3fIWW9rer4dyyy)LP@`4x))A@5N3fOq_9)3_{?osU<a*K->K3H~^sxzk>
zW7Hfvvp7=bPr<}OLU(t2Z!r64{*C38fdX`&7%Q{q%}O-blqsm{Kh<kzArn1SaaJlP
z2hb7bG?_bl-u;2n;@Jxp%wG^6O5LQ2k%<ApF*R#$;X)f2$b$y&m^=S|Tg5{Q0yaF%
zCH(%9lHv-c<=Tp;&U?5F)wyGKMdgFDnOI_EDxY6bwy5~Q*$<+Z@1JKw_0h4#<TPMF
zb7q#!wUuLt2IiPeg1NG&EMTiByI%~NirE1h@|r<O`6IUbX9ogh56+%HXV%O|h&Hn%
zP&R*_?Y@~M4`8;I%qIGS^B2zUh7?yk^5A{wUc7;1nY{18nPu~Aa~8~=Z7ZuNubfvB
zu#FotrufF8Iks^=W%QWh8;0I6^rpmf?r`I=s0b*-hq`;jZnBLjD)Nk0igSnF*drk~
z*BS*wPJ-;8B)%D9sldm0O2+OW89O{e2{*ozum>>klWYS3<8zXYdoAJ(F!7n|906zQ
zxk&FPQV2Kx@yXad+Oe~^8*64<elEP`k@IZGJ;X>xs@*o)l3saf**P|pcw-_x$@Z|B
zo&!2oegEIEO*z+g8`4>(fuC%_?}rkNGz2zvX*O1>0k-ibevOFtw-uT3*VaFPr+&6M
zW_)JHkAd&7J!HlM**`$_lWbLp^V2cRSU0&>0-=ddd>tnm_f7Fnd|f7-d#U&*E+-Jq
zy;kC5uFHgvOaN^O`GN48E%3OUKzN=79$&8sA7z2Z<p;vu7I;REQLh1jnIeHm&9uPL
zFo{o=1#XNhBPquMN4F(Dc@{X|9slH8;MRS$Y=M)e_@~eUkLv)KR%C%&`(=g&j)9l>
zlv?1a38>9hZh@a~fiJSaFR;L?EO5yJUtxiBoW?(O7I<15i10}Z+$tZhwZJd3;6G)7
zbB`bYY_Pyb#DNIwE%0;;JZymvvcQ`y@JlT477P4R3w)miewhW{YJtafOib1-@XIau
zZ5H?y7I=pReyas;Gv#NF5$?s~pA-u`GY&-Pw7{>jz%wlHuUX)k7WmgK@GJ}5Wr62d
z;MZ8-c^3FLEbx2_Jj()?E%0m$ywC!_)&eiGz=v4iGc54yEbvkb{5A``+ycMe0$*f-
z54FInEbw6#_zDZ0d^`TBv%quXK!i_P;5S&{Yb|i<7~-F&Eb!rRAVN9#af+<<tGV<l
zn_LqP^wWPOfZI|UEl7Bb-LwEGv8x04mafdl6XT^sinPY?x%%6TlNCl<MEpL+$pRx`
z5x<*pvcAX$5x;|Rvb@Mz5g*SuSzV+~#78qu78j`!@ozHDVH+tI@f#Q?ON-19@oO0;
zD~lA0_}3Uem+^cNzm##Zv`CJKU&uIFStL`$&t;q}EaDXLB*w|QA~q2}eF@@ZS&_Cb
z0Ioj4I9XMsRm2Z7P8Jnu5%EKelQl)cBK|hxWJ!??BEFY#vZBaZ5#Po5#f;a9_*TZr
zdLmUKzKL<NoJhHdKgT%LmXR4E{xsucJ&{5Y|98g8aw7R6{u9Q@Y9cuz{sYFzVj`I$
z{$0k&S|UynU(PsLO2j7Ok1|e{5o!B^^<ThvCgZIl{%yv|A|fp!ejnpx4Uw>j-_1B#
zLS%!8-@!OpL1e9nk7xX9#_L3UG~;9mktz}YCgWrUk#Z5gfpM~c$P5v`mT^w~NTG;-
zjd4!-NWO?)$~dQbBuB(AWSmnxk}2ZnGR`R+af)~n<D9w?n~0yzK%7%H($*>ZpK(sr
zNUMk+W}H(r(jwxA80XZCgb^>?QVmUnz8ytvp^jlQqnqk6D~m*X6>eF3E7E_HNM}@}
z>wf@Dx+^TD_4}R+v$Uz-Dg^v&soIHxo=PCDNFXjT5Q9As!*Rr-$i2XvGgP{=T9m{2
zT)1T)-+y)D{fwIQvAqhn{GI9j66uVZ>1?muBZ|7`URgaP2j6cioT4;^TTbMI*oGM*
zXj*$|zD<7<4e{w5=}Irw*Ctebs(gJIfgJ3(Kvq9^IO(3odJ;rN^yxiYm=tT3uB;-g
zp0PN7Lq{_3Wp%6EaQHUa-Xibms6^&XM3igxBD1XSF!CQV^H<%PeHmJ<GHohy>Pygp
z_w4AG4v_tmM%G^r-n!@_AOx;RkT7uSmNr;bfFEnQC-NAwNLSh-6-GGKbC>5%&s5Kp
z$?~#~Gr*RVElA+!HuXi<r{G(%`n*eL7&7A_Bd(`CgXMhsPmx94p=dWN+MOw~`lg)S
zo_j=AKbIRmPnshg3(t|3R=Zl5zp2{Qijbc`hR^-E^h7_f3H@5Mx~y(blrYzoC#&m>
zEC~0BS9^a3I*j~s_6b>ijl~D>zGM2O=nOezA{f{&y6TZ@Gnw)xa&G<3C-K$r6O+^>
zDaw#h1-b9{rgE)G<%T$wX@bgGU~YOf-lC}6eCn@VL}U0ALM7x_L&BIPUIPn#KJ_F0
z(a(|7L}L~?w8mA4^+85EigpRD66DIET2ZvD9@NL;m27XCa{9v4Jz$D#;ZNAY<yeu#
z!wfvgqVH|Cw%bTs6(MN+6$tqutPamRs$FXd64~nUBp#l^12LKzB3BDh@F8p;o{>u@
zC{Yb{jjIg+t=c7{cGCK4mk%MZI{?KDaIzwz)RH#+$P+dj3N2y@_-x_2qi3Oo<c)Sb
zva@70Mc&ACcAu=uDRQV#l+nS)Q+LR!*Oj3U{|-rbQIy0MVxm6rQ2@0UMMSQNK?-=O
zBNp=^nGNp4K+ZwIa>*J&YVJ<CW`Cjq^O*wD3Qf`i*UIWXxuMk|hX#iECMPDx<|@#n
z=b_Ntoi$-u^41Eltk$|j{*Yt8$YO71bFnCfGQyJ$l2o%E%PL-xXTE|2r4T1h&=D&^
z8B76omWll(1pP-l8J8evok`F-m&l(WNGBmIieV<g{FB>3(d7n3PV*h4s1==%pr}U^
zwJ`-0+5Ry_7n7piO(gOsC=yM?qDT?d&7z5FF(U8^og)Ds5>($OLCq#X`%LUFX`<?r
zXCX-B?@17gB0(((1wo%fRFc)Xt}IX$Hs2#-%7#g~WdL$y>GS3jxJpbKYa)ADPrILq
zm|5Z#UYAqGoT=x)4E@?whvA>QUsk8PGJ_+Zw7ta2ODk7GuG>PD{pRvd@@<<a_2`$P
zX$4GuN1yi@xl+^STBdTo_|%ww3ROj@9@Qrr$%-~W*7Eh+wRL%I!131{k*Y67$)2jm
z^K60hJ>M-D6TQk@Uwq)G^~|OfU@GP3Y^LN1V5~*j+{V&h!0TnTNpHkk68wEFtAqyH
zU^dVe(xMj%RB#!lhGYk)2{9D)fLG08>?y<&g;Ws@h14PB$0+1Hvk;ML7D5bJjj9<e
zV+A6c$UHEg$!YLNh9G7<TAMHTfweNfKJ-tP_p!<OzO~GZkHL(7ih9Xv14|YNA&-;e
zfk-B=urvN7;=vaM9}#eA<qd!o<i>pHRA|#|(es6Tz&<9fPs3aX*q^;aX*imsl)NEl
zH_HtjNwSKk9Sa9>sdyD)8eb9}1mU7u5LJ;XhaBIJzp<d^q_lE165?->HG@nf1e{*A
zE!b)oL&vLj1qON5j@6#_JT)!;u;V3AV)F#Mk`~;rsBd|)kIJFxzjG*|#ZNiR{#V;p
zw=eBfCwH0;J65;fre4IzZE|pXk{oPIk%LVsp5Trybw}kMpZaoi@SHi_GFY6TIDw-z
z$M5X>R*z{P`P682;O{)KChrKN*`(T5$m1OGIb=7B*r(*jT+Bcwd!~xy;v<%`K#qkW
z3q~C0%KCqth{eGD(9~qo8<|Cx3ebmUQK3|@<#!#jU34o;W%n*(Xc;k_!JNF}?HDoK
z1nSpeCn9UpU0KA;7dgyTM4RHulr^6#N8g8eiFeCL#T&^>Ns2&5TyS^{MDk;E1c~%Q
z7KXK~oMZ}SjZ8%XXmDt;91AQY%rFxK3rEv1F6vpcKX55Uh9KI2YoEv<*1m@T;jn8(
z%X)&FJBTi={2DU)fY-)(p)PhU@@ebD;st>Sav>?tVOMDQU+a>^<2j?X)L)J-ajdP`
zDOLXj#mL$!*9?(#;b?8(zt`-Qs)GVt>&gIFJ4X%;56cZ7I(+t{KK1k5BZ={x8<thb
zZwKS!dyX_esb*(o${eXYe0x-?3!bt|HO<)m67UC%h5@lK2Aoq4ZgmDbk{7=!$F^e-
z=E%VvIl(WI7q5@ris9Ei{w9P*zAra?<dBn5wKx?uI|KbdNi;=HeqMC%raDloJ3j%R
zi)~YuH8EN+;<s!fsrD3yC3>6fCaYa~88%lZq%<$e(<q!GN%Sd_ZjUFGa=h0ggC|%o
z8rr7?T;+-witH(A<pF%NMMb-#SJ>lV_;})au%!ygx#2l=UMb7#u9C8R?m8)JihHe;
zHQl|TK$`9yRppVUPam~G%9`bF#pA3|{Pv9k+FW-Sk8?+brL2Iv4Ud6Q{GKujxfi=z
z@VIzXi<ISX<H<2fR^OHzPG>@_*(*03a(Ja7j&x~=FWo5(nUbC<4Vj*vBMq6Go-Ykq
zoL(pmnUy|68WKn^mxffQlgL=uXMb5Xn#?nz%4o8&*e|eM*aB>_o3TF?O;&EKUf<{X
zL>-4dy)70yAvrPZPe^w|biF$)F!XJJb0ynPOtu);nqhI|MR8<8SDy{l5l1bxqRul=
z&$a;45ye;lQ^f+9Y3%33`giop);|;}*1uz1Wdp9P#QK#9vdOh_NIpKsdNRQv!H3-x
z5tPt4v7Qjg(=?84VpE#Hwb*cw&<rpZXUw94R1O$!H&(Yll4|e{48zh~FuO!-QR1nb
zPtwXa&<EY~A^r}nAO(e_)Py&)Xldn7fyMnI(z#aSWe%h}ShTcqwn(fZ++)xNX<&7`
zn`vYYmiZ?5wGC8@2Z|Tsp*8-{f(JH=6Da(M)D7{6webf|K#^312jYhj^21Rj%D|E?
zCL|a%+fknk6%rGx1d=I8#}jWBiPbKyp<K+te*cEY0`5$`u6$YbxnxECE4H1v`^i4K
z+e_98I+>#(k~v437oH=1^J@+EPdw6+>8?&`dZRQApz;>`CsI)h(wn8B1Fz1<kX?8`
zG#ePrTRqw>9_=oNN1K>Zpp8ofXy=5G_d$?ay^Zzog#!B=QuY*A;U|)FDz;pA@5SGK
zb+Bv8Pf*SVM}ZpAAC6*=o@L<pJMnfW-X0ym{!-MVjQ9UUnzj>vcxO=qO7gkN@znf@
zl*T6@I9y2GZ(=kGj713yJYfcwBEA(l@Fd3?vAU-y!EIRGKTnowJ_8R_z-cwMRxq5w
zg-(`(4H*iUu2l19B%`LW9Uk=-!4!qG7$2){_YN@>YY$>;@6nbzVoe@(slyy%?k1`B
zw<6);6c+Gsia9pjJEYnNjU1^WM=E9jM?P~5M}omi{z?wvL{sX{kGZ)sWaoH|;`}6l
zG)3KR<tNAVgJ@TQh{n1*%&G1_8(MuF4eey0T>~^(ZIrcS3es3HGul9a5Tsl2VDOQF
zf{$eN98Y{n>ES;33D@lhMA1!p1zzgMx}8E6*B?2!n3rRuV<SreBon*YUzp%J0}RN4
zh{1R%_3F+HU?W!*rl_&~-OOY_)+Qk91O!q`tbt2Cw;v-?pOn>Cq}m^1IgDPZj|WEN
zT|DdgeA;&!PdD=^9ZbFsBPIB`L#n-l2nV73VewqSj2C_(iZ`SS0}?eMLc$Sp$y+8Q
zpMsKrGzrLJSB4QyF%h0hAe_{Vddx^b-X{dppIQR_7FHY)%KZAZW-AD7jwrs~%uS_`
z@rK`;5TSn&II94GY_e<{bztA}O!wU5xz|&CPa||&=oLA%&d}MYQKN%qn+2F1gq}z?
ziSF(j4{<=1=$imE;63~1Yi1GMed81v%nhQJXOQ}giAK4>fUGtl-7Wfr3F&UpN&!i<
z=$}6_+ONAse{Dj#Tl5En#9Q<Ol8e!Kz$Cl7F&{89b~olU6Vlz7;{frg?RqiVWXlCf
z9^2;IKf|PL#8lZh6HoeBAfRO}%^LYVIZjW|vj9aNI|ELp<c#b;11{_ny^8qhnOF<Y
z1l<WxWG;rc$Zgr7M^|jw{RMQ>$MCf&Nb)c20{KsT2>g-2j~oD7LZ(QkQK&GPFdjDJ
zPau@l0{}%YGC~dlp~<R=RZhYtHZm0``Uhwby%k@PAAk?AgeMa@nWMYoq8}Jt+1+t(
z5rTqlV?lp4BQX%h^}?dx4St-VsISZFXViO!9!rtc-TKeAaNQL;#2`5t_8dVjQw+mq
zxWNq9nqi$8RvF<KFw>=DX7H;}ZVHky^1~AHz?B$Ms$rO6hA9YvDkI=WZc3hE^I&>Z
zrpQAoAqLA0UDy{_$U=Pf)DtIg8AvYGpM~MZr;U+iEm#GptZ6F{_(COxzR(Mz*leub
zr;>aaN!hy>UWhR)2ai?hm1e%^`4#7l4>_u2HOP!8=1HTNwG4XY4ZXlb-F;?WfgO5*
zErTMAKqH+!i-4rJ*yYqaV0|Ld(Mu{o5FjQ6%n<D-AhL$W(<T)4%zpuu0-NHw&#g!?
zGL9+og?`m+yxR!0mNVWh@-B(z?XEU4eW5?L8rh9N(;4egdt`6dZ#y$PF|lX6+!YpG
z`qK`?3vze*AsSYnhyLW2GR3F)(JmFGvi3|B(*4?0)c;%1cRg0<)1G0YqT|?V(3gni
zvRnQdqd7K?Rvnv1tKOGvh-^Y+3zP_(rS(CInX336tsxZ<9^z2Nw}4R1@WL48Ts^vd
z<Q8Pz2qcoIif=90iRaPMddC)3JQjfWU$1Bm*Jy3Z#!e(msqciEqF&MZ1G2>eiC%8(
z4+L_Pqn<;TSn<bLlOJNr@>{_a{Az<wLn~lph!Lr2PGBdRcQ*%SI#>BtSS@|p1exvm
zx6M=;iNU!7gOfRYp=T)Rq8KC4vd&h_F@3eAm{8ARm;;^7d5B_+K+8X)n9JDNcbLWa
z$X-d3hh<$Tq6xZ~*H10Lzu`gAoVh1`YO@?{?Ldat4BdcF-7N<{?2toCI^>39$q>0?
zO@S+A^^_cZ*It<pZnGE2{oZ>?a2_ap<~K*vKau8NioZLYFQuSi0c*tHE;$%Ws=x%<
z5%tO;-_tpA_E8{BYQfjER(#ED!&mkkDKCF3zh36oHh#SVQ5ACQj<(8IksC$r!`~jJ
zyv?$u3n9hl<>MXY&2~OD&ykkAs<@9=I@FFR)CTkHiaP?Hsz~GBlGV55?BgIlS<e0>
zFaX>M>VzyAii*G`_)e+>fuBVk()y9l;qPhcC)A^i$*~=32gGT|f@{={xV{5p4`|-!
zfbV@Sg<q+l)8TXPUhqB|HsDdT9KSZeuMKo5+C6CE3$~!8ZMjEG{oLyIk-zt-BcIX+
z{@x?4ANaI7@EO(dJG}1qc(Kp@%7S;~;I<5l0E0aa7xIBZjeSi2>$_Hq>72SA>jQMM
zQ(R7^#|rTnnDBpx6{0nU%@0-x<;)dg6B;PKK-3C--Q+VCh>BhdgiyO$7KkZkKF&co
z<f!|y1tO8xTp%*e%)1s8$zQTSm_>*MB13;OgY$E8Vu8poDX2#&#sYy2SdRt5%rCg~
z>UeG}5Jtu?Um&{kex%<M&)Z#Xv6O_KX=M#GArydFTNa3TcG_OAI5T^Efk;NLV9k#&
z5HM#;)jZ>;s1dP72m=E)<C&I)!lzBfA~69gp;&o^py$&vux7*;i#kPp!>_h7B@?Sc
zJy^W4RD6Xc0?fWr!bVly+Hm+v`_?_D)v30`vLV))ahmIev8rn$3fltBd8b%P`dC3W
z8Y_rj-Nr>Ei|fJ|ZOTiQB}38dy_S_|MqSVK!yaEheCi1$)}*-Kl$KvA<j_4|Z3+1q
z<n#Cv@_h`K|H~!hPD?T4dlvK0EFqFv4A+bdFxtzz{t{n8GR|B=es$I*<UcU(zTy(H
z7NvtVeZ?hYQhW*Nq3`+M_2et8Cs}8%C)==2lLIW?|0U}Q965ZL;PBR<qEu18khT8G
zh=NS1M&)jn+LQ??hG687H@R27m&gCIw0@6L5)ym}vPAY9fy{;Hat*7KF`5R-?mZQf
z97^43m~@(DbquyL(_I-&Ue|C89;o0S9}Ep=7lzlF@};6Zo(a~2`RdOfUoyB})`rR2
zxKeHSlbG}y?XYuod0>&l0M63NQ({w{o)HBZQcYNl)yKiCCW1rVRj3VL9qh78zi0@K
z{E6Ko{bCzzYEtbIV0kkCiZTmw!|D$8ZLfMfHZ_CJEzCN=qg7^jgy?7_6=t}%SNaR8
z-<PVN7wneo5IhlV9CeXZhs#qBjR)q#F&XxRERo9xRa}O?%>{?73-G>P^0O#xRbPbx
zpN09fS*|Rf`k^R~`a5h4Bl<%qQZUxuB%rTEE;9ph3;6o)_4S&w(;U6$tfhY`O$X4U
zXU%mMn*IP>Fwue$j}Mk=2Xgxy+?Hn<xmzZ+V=Et~CSoxt-}nYK5ih<06~#I{840C{
zgkK>6_Ao;}f_GH#W2Z-b6=OqDKg9-`d|XzO<&vj5kcF-!M}%+N7^mDcCKDiA0O@+S
z0|_<+!f2_Vem&ud;6qQV6X4w9Jz`7vjf}w%7?T0tgUV@5V=&7oXODDY^g#9K^k5jI
zc-7Of7BT7%q7%L939oueZ-g%u4Wz+sPS#l{uPFOzk2wvs(S4{8XY8&wQRA~|HU0p?
zddh>PLSKM4%IZj$PrKP)^4m_(qm50$ARUP94@g29m^~yjkU13*;Q+Q@QDaI&EQw8$
z?R7yf;C0PlRRWj$AQLy}qokTK7$y4kAVyXjyR%ORGA45ocDgUH<Qc-<>vg%|3*qi{
zP#?l-gdEQ>u}BUDT!lW$tH)wfTp4aqCDjNc6gFUmM!h%vEfzy*AMo|a7?y9X!;@6=
zG=>oK`{d9R=(a;XmlwtORb8)lAl;xmla%K$Ye5ZW=?^hj`LIGm<}EBoe?T-VCoW3J
zB2`%vLIdB2d98)gbPl%ga`2;Mf6e=WR1ky?ld8XijM&@Q5eeMIO%5Nj=q`Av7)1Fw
zQd)i{{+v=8v?KZXz=!21;KGm&%l`mRj$!}xRk4&{r$bu42gUQn&=;}rL3s1oC}7EI
z6Rfdv;4+{r!e1%=its1PO;syMB|d@x;Cw?$s|q8IV1p5?H3GI9@LLe@W5wr&8Tg!2
z2y`RZN2I2z3~~xSf&f8Ehg9Q-uoS(RW30QUL`UTAmDM!4#LJ!?3560Y!r<vJ058IH
z3%D}%_t42q_X_2@7#NzbF<-{UOIF9{!$8uAHW;152Xw{gOg^ALMmu3#X_V673b8!}
zv(dLvCQX=FSN$F#%<-562X*gW!GB6E{0HM)#<0wU8QOR<2k0e4MV4F))GY^K-PnBY
zr{GZ!qu7nPc+xX~uqBN=N+c{wB*gYW<M;Gy`1%dLA`d|4$D&cz2Z#&*6S{X!R@5#<
z`*}U``L)eq1n_GNe9iPv8DUJ5AEGQ6)-x3M!OFuXcA_>5kqFpYqlBl-5;BFqfSG@C
zul&SK<iFX-&k+I&*#D%UNB)a?<tJ_;|H&4<e@VGPs&2%JjM&>oEZ>Oz)res-(fx0;
zh>c?8+;5R5M%Do$p0N3+|L$b)a_AphyYjKlydX3rB4;)501%oN0jMwl9H2(o_Zfgz
zK__1}$pADb0Qm-hgWt$9%m6%@09;`JsuF<n1>mt#13&gEiFnK$C&!Igj*<4B5py!u
z|67s2A^NSxUi<kXMSVp9M^-;k)J~J5<&&co;)S-}vnBvjA6Q|gPi@d^;DG_RAW9+t
z>kHxPt0wzlhLUqn`fK(})z_dM{4gM~a*DRj)r!zZmzZ4Wzur-TA9g`+5ck}v^ud8H
zgWEn#zQ-MOr%O>ncf;|h)B$ao9}4o|2VJ>4bHjBW6}G}=ZA_<j<LdUuJJkld9z70S
z-^5O7{Q$2WSCl}f(gZUl9g%3V+}>5W3;HYWmz+MzG8AZ1{A#nk!H4xQy8}v{-O)#U
z>V7%<^FT`ONvNyrfk7VV7O}@0S+DMp)(2j)*>^)vr9#j1dC~%(lKr_q<f{jP(>K8Z
zbnZ(i!Wb{4>^$Uqq*HB@)<529Z&#ZUb+eR8o4rd3HdttD?5f=DQMoiiTF~;bMu7ET
zMToU{RIKM<21aM5?`B6YMhA$oHh|IKwxY8z4+G)DS9;*T$tY(1F${ukRjM)n@0?ty
zs3(QQ(qFh}XT#h4bh5nBB5p&`cFO+ng`2UcK~K4x+Q8xH-zdCpP9-lmSMJ9dtEgjB
z43~yQ9C}UcUQ`So?-(SLWLYhQ#uG-4G+Best~ME54v&0-mSYk^0foyRp9zMXGYP`9
z0B0G`1n1xpeJH@#K=7I1d^~2FV7!da1VdVwlx2d8jAw#p;4#Mpmm1Fmm*X+d1TQk4
z0sdhXYJwP;Ki~R8!Fm|QS}2JXZ(9m@Mx^|&qFeY3U;Ap--&UxPM`!7?(E|y7E8OX!
zMFZOvCgOpxJ1W}36h#~Bl(XC5^_SdZx3ksf)}<g^6XCdR?tzjO4URg;<?K_iW+Y=r
zw2P+%vUmC1ho$AaVdnI?BXD5wOQSzf?0=Ph(BN~oN<Y||vdbqGv|`soal@s!w@FVt
z1LSV;XJlC41XAH?gBOq{+1(C_Osakh(wHxFd%co<N~!;zu@QR@x?QEc6^9f0%dxGp
z+Jc#r%yS93Xqn3EeCnylZ5T2>h`C5WRRiZU0TygM@;lZh*kP}{2r;Ns&Bs%bx5+?*
z9Z8k&c+{hAK1wy;0+7Xhmx*v?G{m5>Sb)c3?H!9{x4+ypHXmhzD-~dF?0^*XrD$@j
z`Kx2WsR2^ZHk1w~+0%Z{*FYXRD=B;jIYxt!hsHu48Vh-7EaaiFkcYm*VZ@Ih1vrq0
zz7s|q0l6lEwMM|%1^gBS{8;e~dFVSG00Un)f_+44`cAhz)Bt%1r0h22A^4k!?L@L-
zADa^41upa}xE9;od_^4x$!37$oCqoAE{HV6IUXrZeBG6ar-@E{4WNLfgeIbicLuIu
zHd*!b&L*WzL+?yN?<nXUyO7Dy5dSw>f~V2=W7+fBUAcRE?Pq0}7wj$Y{nVxp0v*%S
zK!{@y99{;Y22%#13eyLHkm+gonvw=`d}*M^kwy_XfT_M_kTILrq%njHH%MmYB4#dS
zX7t+hG<;1-L*Mz*(0Sf8S>!HcZgz^Su1RAkIjaY;uOw&fptT}gHE4qf*9;0XOk0y?
zgsaj7Jgqiut$<gjF(1ZyeE_9a4`SEJYIPbzmdYMwsiK3Os|JbwcCHy@bpM((BV3hc
zjDy-VV<1$giIK2utB_LMZ%8$}vFo)=+K^+f67n|5KJ_R}GGc~G7|>uPJ~bj$Yl5aO
z_%kyem?lGsVLpF@yzfglBYy0{3en$tO*BZ#6p|?)LpE+2osYsXly|~$^DAN*vk*on
zV=rRxSH1xAug}t<s{u3Meq=N{aWwjHG`es!=HGI5I6s02;9&kO4<nAi=$++jjhN`0
zrsXY&@ngj|8h*La?#qp4UoKj`X?gej12;lN)f4gg=fnI94Q?h67wh>gJ7^UgrKtC&
zOb9vNS3-^^tfpe0F5FBSw+x4Lua|GfO0vJnx`x+@<5^R<hQA8cw;{>-v6GQ|XwTyI
zZmd=Zbrp3GDA98h7GtbESp9D|m2C-huJm{JM(08gdu||C(QZ-Td+;O>eA;<f>T~zk
zX^x-61oaD#>R4?)ssmTUDahw;T<{)b7A%fJ`G5to3i&Ml2Uw{ghIT=wm#QO3!D<Mq
zZpVr!EcUYdEh+dS;PMvi>G-+L&oywXFZ(nE77WZTr~+m8JJNF6dt~?9P)_)y(NV?z
zI#xcP`xsWils#Dgj=__ei&3WH{z7`<$3Pw*8u=^&o4y$xfm2*4m~Z!GcTht@)dH>M
zKD<HQs_#NCD4~(-u!H{@Gv%Jt2W`W`Z$F@f2CqxFXq5owSFbVqbRGLMda>N_UJ?XM
zY?9Tt^>-Ulko{hWgGQ^E9z{l9S0hnBhFA5H*LvFHAh|(n3V}U?lpq|;P%gRnruQ<T
z+<c!_mAhb#^@7RMnOBC$o@uC=eux)95U2@L*_Y{;yq#LV*r80E$dXo`5A={66AIK|
z1N^G(!l6!<Chnm@H>O{M%-Z!WL_Z>}d<TnsZma-nBfOK(|K?gK9TR6RX~xkuaDGrw
z{{r%QH2XDsmxxH~UsX?Mzr7^Mz6Y82^38|9@a&ivQ=Je#oN`G7!ZjDG4?y~=YpYnf
z(DOqXG6(jTsNM;wWqcK^33pa`5W~Z~I(#i$3u5lwFvlyEA72YY?p8R9m!J9!7+z`G
z3FfIY(97RlD@}uax_mn^_7Fpw_5l&rng|gfw6LHL|G@7EP{Pu*qd*zG0VoZ~*0#1l
zn)b0rn)b;Yk5v9LOA3E>!Yj43OVb)m8pFtNh($_ScZPv*YApb5(zMff7S!$htUxMi
zL%jgMqXn3Iq-o8l1UyO0w}OVo6CSAruJYykS^nP7q-n1K^A!HJnweTfrmTBgk-iH#
zUiL^uANI%~6@3i+T_%2;iQfjCL%@GeD*CJ!dKb_S8R!>EMNLMAD}Z!(EgVXvX;C~j
zn)DRCi-rRJVZ23NwpH-WWjh$e8r%(Dspxg7c7pI*f{w4%Q6Oax9DrSwcKvr+Gi{il
zwY#CA;D}m{EYNwW5m?o8!<efB&}@)4Mj^-z)3koI`IHn$#S^M2)nn*j(FC)2*MWGd
zB;#a2&nU#W1=pFd@gbi(XHK0|q_p6#6@P6~(K(?)2iA7{HRQVwFFxc~I~BD(dJ#gr
zC3+qvOte3q5J$BeZijQNbdgYRjuM+rY~M1B6Q$K~eU$D-5*b1oxQCSOaOquPOlWx8
zyYK~{f{U^GZOiJpi!&L>NBUmil3jmutCa~P^v%mYu0nl1D)x_m6Pyzp#1YT}YZWOl
zF2o99RY(8~n}Sz8M0<lUDtw)spvWW6Q3}2WQ^dOkYMZ{8jfkCQX%u?T#jYHWICt3t
zWtxd%Zbq*qoj^_TY~=p12P%)1v$PC2PFC?5jd8xY0^Ad6+d^H7sf#{iJ%C1C_QF0i
z+(Sp~*U=AD5MakgEdg8~s0PmRmwYN~=fU`0I$9gPK{kA?V5!*vJ!`7R-sV-qu-HJ)
z>Y{6a3is}Vu}L@x^?^qnhppUWGW?DT;YRdF;AhrzRe|Mb063@DeOy|(2SZ9O+38Ua
zt5HuoG=^Ad4rLmrHi+x9@51IMb2FdqZ>T5Q_j=rWJ(7QKXvz(CY>AY5oCJ*-<MIvP
zRXcEa{3=Xlc;Q!b*@4?gX!roR<fIas`^Qg3DewuwX0F`_Z9~J+O#2SCz5Pv(`!vct
z9a?+~%7l9LBQPtTo1TC1mY)a%=T3`%I>2w?kKQkSdcPFdFi^vn;cF+p=D-hTKmO+I
zLWsXEc*AtP3_VFJhHpnD#JufM2TBJQ?E+hy-z;6oM-0`xBs?s6Rdye)JlCUkss|#^
zqXwZ#vF<v1&$X*NAVxwoz(je!<|Gs=2coyZ4yGOmTn*VV;PeDXYOsoJmHxcf{w8Fi
ze;8LggB^|qKD^N!AC??m`yo$ovDS$K!_uF__7=uE;>B-e>{F(Br9ZcN?0>_Y?sP|%
zZiaa*6&^|}UjY5V#jfE=(kf=h02o|=wHc!|i>5f(axi4n!$5a=Ay5rgqnJ84H>Wm?
z4mom#9k2!YGm{oifQko$?0EJbaWW-)tH-@t^0jay>{VS*>|oe>IDC%^gPIpb4%aWj
zuz+1An%!$ZNo4i2p19+%M^_`B*ba<J^?+1!7e*D;(s>$8o6u%C)dGaLE1&?gtpc-S
zOa{3eypkGDIh4(DO^#1me;k%5Dhf_;2*LfL0OqKIOtb{1(~iJZoX7cwd{FU%#S8__
zN4!|SQ&#;MFkze&0tIG(IyF;v@4yDKUpeW50%KELK_Isu!UP!r?K#d8@u+ldidYm-
z0*zn6ZL)>=3H?Y9`KGi$eZ`B8ae@t`g2{;~PX>@f)>d*D{iGg^Icvhg3hI#xwgY(}
zT7olOTakw*9vvL{I~aTa1c}D0!O@*H)mAi%0M-{-yHM7Cz$jGpsc`Tc{xnR{sZdlp
z5gGV9`a7}7U{W>od&o?gUlhw>;pA{qUxu?9`2EUUyrx%wf=lJI4nA3{;y!G&Hux#O
z`Z28yIX*4Nn9577HZZ^T`+Ah-AM*PR2$Y5+Nd;7$C?%h<dkXcPpR)I{v`FjsA!W&R
zq0}K%nSwP=cV#M?$i_ie$b3V+st$!_<u$n{qqoed>qZn9e)WTBIVeuZYaGgt+tfiu
zxETc7)CN!R3;5X)*9+$wPc{gd{uk;!|Kt#O;yfrbLLV!#Uu0y**2_E*x*aD%w`1y5
zzUrs7`*s;_e{>VP2`_+2O;({|JS6xUs}7!Z42e~E;{G1)gL0?{B11G8JBdKLtip-y
zLpbJ<PaO-3whvqYW;XzT6Ac$Zp{KnK`qcIC3e>yn2aG%EA9yc0<Q#Iyp<r-eiCypx
z6zmYbfjjac1hiGcEdkv6tc(z&{$)A4vC^r&jtWKRVQ`rH(cpWShww_+5lCen_gA9b
zwnwu-kKK5@n4|^zCl7oA0t(WM2jd&=*TF8wf-AY_2?d@=RbNk@_i6IHj^ufr$wNUG
z6s`@OmC@h_c4On%^%uzDG;01z=U~Jb!)TL&HEcKx%*>GEJo!mEbR7&_kf1r_<m6H(
zwB&LLUXThfPQ3!0B`3?yT*}kjoxmqjkJkr_zyLZ%4d60zd-OaH%#xLlFanuxsSV3j
zFbt}ioWZW-#e1m+3~tW>l{vxF$%~&ObILQ034y6{=5g}G8a=V_=n>cbPO3e?%Q-OH
zGN?a901Po}M+`(_9TS7wN-aF%tfvJs&<Z_GW*`2Ukb`;`mh|qvKf~X>s6d4>8<Ut5
zJjKdjpcxVj2Gc6@3@)<+Sq6*!wIJ5uFjrvegZemUaguuocS&{xiaSTv)OaHNI*r5>
zgSX^#(`Ysk%k*qPH-f1=1+l7od>LHCt6le$=ufY-Ugtxem6vqeaQxbD4PNqFgO_{^
zUh+P9$)|$V>yNi#7g+QDlB_Wy->ukkCwNCb@xJC&Tgf{v1@9OV=YLosqTf!?(+9uk
zOj<D3lZ}_e`9yp?{S*9Ro8T9MS?m|gA`4^ZY@8zSsQjPd67D#c;Q08j;}5UM+1tS%
zUW-0t<*#z^gE)T}faASXL&(7|?12mMo!|;RSi)<9C0v(a39oU1N$~6Bdp}LSw<G!9
z&g3gB{NX+F2Tneo3VZ91f=LXPezlMMA**5x2TY15<bb_QCx7TRfaeJVIFg-~WaFr|
zoLu2F&P8|p3CH#ziPAqyOla?jk(jPCX<o)$&N1dP)>bh+#8l>eOPI=Mu~?bAh~XfT
zGuDT)yKC3OUb*aJ83VwJQ?O*U^uoeYjpOOXnf?;kuO~Yp_isY#I;28nA+0<JpJu=d
zu~=_{5g~#o)|?fmg{!lvx5C7qq2MS2@@_`n70Byu#^HZX7ipyqRDa1coroyf9|<$@
z!Up8YZpc%3^oSjVf}<ScL&I&J7eTJg>+X_Pc5vjDoGhR#&hxIdMCEp(34RBe#wiv^
zPEaAe3Xk~Am-uY|LOqrc{k@^UFuM}lK~I(~i7mxdU*zuhtH+hAPH`24^9hQD2mI*J
zNW2UCnCY87-NK@{X^_@;n5BE|O=?Gbi^shcn~JTW#d*Cq72jTIZYqR4X4y|n;A)xt
zRkjn2#&)8Sw?BO8c0zH#Qwg`JPOoOPO60GoLuhQQo4)Be*gas*z;7arn~Ecq=f;tv
zK1>`-bO>jV6dd_nd_&L#FOWELFL19YNT477B!rw@mig6<=(V>77eB|{Mx*rS&+Sc!
zaI+z{9vz5quhEFT#(N(7HcZ7%ulwZE1`n<e(8MthMbpH}rloF#WSLqomtdb}UVoLt
z;ekWV`5Z9#Pz9>vxn?IuBKKL+idE+WCFD53WKLN9qUSha;Jd$094MDVxHpK!ATC1Q
zX%X_h0s!tF5(x+CX9S6aa&Q>Mz;+2H0oA|ywnTM>jV?z<JTQx1N9ObOf=iKDZWwa-
zl_|Ck`~uwt`N#{RCS%N1LYIFsaHWYnmdIm?Jl2V?gpE#)LEJM;bYB&fhf0_8IFD$w
zZVtiMUB(xD$*a_Mnx6t!5)DUomyX5B=A~nc_|?-62daQK1PR{>qe=`p=NYaw!#adL
zY;l%6lx>LHVdCm$xX;Lwfn<L861ITaCDq)D`5DbIds!TC?>!dc9h-Y3cW1okWHlu>
zOyT}gd`T;wI8T^xkBGuUBlGq4^&A_<A$i_dBFa>ou&LBP=2-;fgtT)^-|{S0hH5dw
zj2i<41w32C(a7BjKT`GI@l+FjMAGj?!BOZ@c0|s_y+1w1V^F+Pj7L!b#v?1t@hC#;
zcx1_7_U3r}|8E?!r8y4i-}FCV{9Sd{Zs)*b&vD@8?#DpGs0+9eK9Z^5iGdTn1ml2i
zt<_xv*qz7p@@L>=b{rA-6Lb&8Zv&UoImV$F7%H4nkLt4qAy0f9-hzzgID8<?fm@3P
z=FqOD2#g1~0z3k@JDf*HT;I2jPJP_%mV*S|PeX9z2E4fdD`e=p?TCUR1D%w8h#aJr
zS!ls%m(|1i%%`c0HCb>Qb70WhP-A|O%cp*#KZ}eqF4^D<VDNdF5<R?L-(le*3&IYO
zHW&u>H6pud;5n?@@Pg<~KGmR)r1AZIzLHnp;I0KIp?J~P8c{B;kfN|d9Tk?AFpuOv
z$=7L%xGjOp5Oz#H*COp9IQ<?5on%*_8}7fTEIf>{bt_Ph#+yi5$=lZ{{guku`^4LX
zvk6{X2sMFs#k;R{phBxqNP7z&$m|N>Podya7vSVUJqY==3tPM{c@NI*VYbhOia1nY
zN2*H>QSF4yBmfXP!q~UNU<N$})@@cA66J6dN)lJMzbHg4uluxA%L_*0ViEfqnwUaq
zWd(UexBfJg-`{stSon;Y{;?Q5@GXk))S%cX;RG;l9SWm#SvyzO?f@<4$?D{kE!0X=
zx?V65Ti&G!q)_TEN~>Au*3<N!&wzUY^dp*E*2<9gQ79$8-6m^~K#nbw5$}-IYq7DN
z0fTiW*Y`@O-5V5lvsBG(UFab@n$@Ll%{~!4l$_lba6tVxDR>$JiVWd~WJQMrVWu=A
zV0xi}3cN`n#bQ|DwuN3%$cfC<3rub3bao)f9^Nd694Nx9bZF#OuX-f=m7ekoM~ONX
zTf}jw(^%AJTn@qxgs4-0)CabX9jH&!jNtap$~|=$V4{OO?J;$g>(`(_FozbBg6YwU
zkRLzT-aS4QzDoM7Qq2$1#PMNHwNj~G`8!THj1RbK1DA;`g5f@d6GGEJ!xUR1$_w0T
zo?j?pf0XL0kvEzxvc?Nd&@etU_%psgo>$!1vPP827cYW!P`n6@vKDBV5v+n2wc<4L
zA-)3RO8;{zX_@89ryUsnNXPW&f6Iw(a10tE2Lg@LrVZJ+kt)X^=Kc^nVu(+iAy;)k
z;n<YiJ)pS*P~YIt<_<*N0~;q>u@TZ-!9lmtm{JXIWk4ZcH)hv4jj(DxR-ktA$}P@p
zMO(_=NHGOs=*2>{E)Fms*fFJJ{q(YKEV%@H)sasQ6-Jt@4R1iri7w-aHLJ}mUL@Bd
z)Q2G}F0ippP`w|3RklYjsXL<tM}tVky;qp#;4Vg!c;G5l%ln}V)Xv-^;7XKAMFUl#
zp*$#){*^Q0(l$I}L~WM1?fM9DVSp6M!yxJ2Y%0l6&OI-h7p(Y=3!oL+&SFNa6Gbm>
z0?mAfz1ah;cAS9~z*!z!SPT^KHl8BOW8XlJimntaCbq4>-6_?6A2)_Z;Xr0^Uj0AN
zxsh*S3mSPJMapPu(XGPw;)P$M<Ktf0zefZtD3IH|f(_0W-5=Lq$lBvrx9HBhoEh0f
z!A89j6KpYVY7+@%W+B)0EM%S8Me+Qjd*wHTa-)cAn6_*i_q!J57gsRWKI%$5U5cl~
z7;gI&$1o9~00*kIntwxDycHAG-2z~*>RyNHibX1}&x9UK)ar*U9RA;^vR)anRQFM{
zkn5C&50WQ@=01o9UL)!%&I`ozPwtgJQC;{JqEjLV@wH{x=kbF!6j?VuaVp=Ye-jzE
zj7%iZ5zI)qiV0r_ig7fDU1}txFd;LM<9H(B*uNuTSR%)XM8X?P$V()Al1O+732aM@
z_%MLBzz{YoANA!(iVoH%psOPHoe*fz3-ytV&WDDcxv*OWb7|KfW_N<gOhgvO??+su
zW1T-JRsZEAd_8VNTI5+gAt7=HjO2#C&0-H(_AwcRUwxJ41TU<jP=R={$Pea9pGQA+
za9byi(8(jl;0PT}9!kk>St~3Ij{Gie&3g#|#GKl|??Q^1g0)FnxfL*RU}dk7Sn74V
zzKhG~M67ND-e+K6Cu?=w-oS{Fhd^_zz(Muzf{s+VL^nOh%eg>QAqzja-1P{rTeX~U
zdQr#|hIQ%UUnZ<exS$gF)WGJ*mV-{Ydns|^wkjDiLzTe54SQcw<kqz)5>6zt<XvSJ
zIR%`88eEv^yxk(yW0|;M)FM6Q3F!%%nRjC?lvl`HuH!|TbVgy;Wa&p<*Lgf*pN|Fr
zLde(8Mnz`V>h{I1DKNmZ39EkxbOD7UES2v<xh~EjMjT~t(FpflpX<EWKIh!YgU-=a
zj-y#>0~PmPu1s7Q_1YwIcbo>q8sJq=!e)=Rxk`N@uWPOzdkjra0Tc+@xGeWpsrqMl
zDNC5^aYF_46k}yCiv~pjtO>*ypQ}h8j;x&jEqI9sYXeeJYfdi7gm=ZsK(-2@#a{U}
zHr9;XBhm4e{WU>qGb~mSjs}4tj0c9!5NurN1Yw&)iADMW^sD*~8M(MhRBeFS7kL*v
z?4utKIE2HOJl9p^4OQCJX11qQ0oh#`cL+Z~QE<X{dF^dImJKEJC)UqUMLsCr7xMG8
zO%kz{W!t&C^uY4u-d*`idmB8_V86vIq3^#F4;ad*Qf#~XGpTkcol(#o=24O2Ul>vI
z2t^^I%eHaWS-3FwG8h#_wg4nMgnBm70Ld{SIr<<%pv+u~A-TnqU@I@9pp8r0!PDma
z90?_f1TNA>f}aWVfMW183NJ>&a3%~)<XD$TxQq$iT=3aMLJ||YnceRb36Y;6p_|#Q
zPb9P;A#!=ls(|fz-VgQ?h;%;MOT?i+!Ev&_%v3^yEL#5S*l#HiCCEF1BkQ19<qesH
z7_thAgifqZlx=K-@=SX_amaYQ+Q>R=W_afa6|?IwO@w}u>FjWAa2=;+9&DoK28%b-
z!6UAa8T=Ue^$drs6XI=eqiVh$z(GSSg-eEw4yE4611n;*qkr?cKxUgp)T>|!xuJ)C
zkUSStj7E{!9M5C`XV^~#5{;+}JUxu3&>sl=f5v{QU^>gR9Kb;O4OINB29ZbpQw9+-
zCt)@PegA{(A&}X2|D=~GWSkf@Frsx7*xQwoCaWz3hXEm88m17OWX1YqZzd<|Z4TM^
zQ<_7dhxmGYuyP-hDMl=m|FzQ~crtq)HhGvldBoyG7?D~h)edIjml;Mb1IgqiH?Faq
zH#UEdfh&f$k16RZIZc15Oe}P8T1hxb7(No5l5sL%9jV8Rx}^7nG_=&5?TH!bHRdem
z<5+EYJua2@yu|4xt$eBXe1w}r{g+NiTm~y9<csEnJO;}ooWqoocEd9H9!@9%vScVG
z<d9wy68CA~%4%9BGch4^t@`?ZW<JKme8l^@jg<e)d~7o9lkxe;9Z;rJ&3nGjHYxwR
z`!oDMw^w3*4*hYj`RR|(&$a^lQFAhS`#}opYc|3u1$-Z}_T<%^H{xCd2vR3>-X`L~
z6-eE)R19HC0dgM`bo7YSda3qYrY_sYxz<YxAlElGM+^VWn4~M8v@V`p52oUHEu9zP
zR)ISK;L$z=N|z3&RP!myGmgtC$V3quN4ZO0$B%@_I5XF9C>ffcy<kUavI;JQgIJ-F
z>9V>AG95wi^kfv5^l6LJQt;r!LpmQa2*CpV7-s(BbUt8#UYrK)&tqJt()oY_7pG<7
zA)gRF$b@9!p^y+h6cLhxhZ%(Mp%ks2hJD6_ay+E-VUajvr=Y#HDpbcf9w(wJ7=<Gs
zytbNf9PXaTlpl$6wf8h`naPdWBlz0H{r$%8;Yq&}sqy_zAMJGs`p;;uOE5^{?e&VH
zrOVn45Vg@>xbh=35;{Rt?~W9io@wTh!+{qVr2P*$bC_t<*bmQa)F2!68=YuW`>G!I
zHQ<N{?xuj|`*XM$OVxB$Q(lTY>M82qjrPaEK&jdg&Tt@P2}aZx@MFLkh@K}xBXV{V
z`T(cFAaX;>a^kHv&nA9D4;w!qm@K=uOTi)NIeiv-SW&n7)qyTQcApr|IE1Ez99tDg
zh)%!O->==C;-@zCJ6K9^M#+~QhHFv1(*D7j=L=!D4ExkKd^loTy%Dd(jKW<BjKdWK
z2L{dO5q{`khbme=&b5us$EC+M{K-yy6=veAC<o-<&96HUO&gDdv?L|82xr?C2s5U6
zwry;FuOo3%+Pyf;bSDlo-HyXdn9s21ndZ!YHj6mh_9z<22Tfo0Y3#j5>8H@Ek(axm
zU9e&H=;xvvqp1n~lQ@F^cc_mI#mX`L^!Jgj>tlX^;I=v$I$68U^Kialn3ukxAl@+y
zN7{z*JlwEM9BCT{;-?SA*OZ~Cf^R75;DB?PaUQOAn3%e1?NEj|4>yeG;fC=%+%TSp
z8-`O<(}&_~%21w%8;ZkJ-l3>Ia)Wi$4r^g|AU8vtTpLy=!rEbLMYw9%1`)0q7G{{X
zW~dRa8Y<vvwL{kmc=b@`OPq%rMmrSr07DtFR5D|nha0AgaMiF55w00V$%LgE;i{nm
zj#33)TJ=!olZ_oBJeiI2a5#}&@6X0raGYf|#u~g;{h=vaWie2d;I<4U_Nr8K$nZuJ
z2bj#GI@k*8Bhc=Y&wePvCjGo^)+HcuKFe~rDGfySJly1v6F$1|)YvVKI`M$4Qs4B@
zVE(s@<G%3Q<xwXGa4^<!fvIua=@nLr$DQ89ai`A}`<poK<ac+7<4%65paYs~%=o~y
z;%1fj?-h-78H$+N??KFhUq#p_!F(YO{f@<<-=@H=ad+Mhd1IVQ$nEc1-Fe4{sx#S7
z(Lvz7<}YY8^K@7=7X#OaYuG}A0e%lZtkGz!lU;aQ_n{~pFlxmwK(z8~6l7r0vbcl6
zJngo8B^t8V{!?5PH5}?s-1hLqldSiM{!-0kurgoC272l6#(f;QIH`#%Lcm%En#!r}
zy`WV6EINR{l(OPRWE>xI7#0OORcx+fXJY3_eTyuVK@&o$CsDm-ite~d2Cf9WtmUL>
zw1!L0iTFWDI!`dpgOYTf5YPYkpyZu!n<y9j=1WdBTRbNIF_r=!Qfc`+Xbx0fn1d|(
z9FX4kd1Wk_hOY$9OMc0DWy9y;tDjfy#y2MBcZ_-2=XquDQ`0}f4{bzr5O1RB?m*Eu
zxna6adT~#LngZR+K<q*Kp<D4wBI1+qj}!VobT+)+xl#b_5BRpU2f7vF4!qXP5oL@V
zX?=ql!37Z|?|NQ32N|$-8tbF-z{iV}<oSh|@`ihieks-~<2NO6Q~t<qcv1q6+xJJ}
z@B}Xr<>{nM`5erW{z`C1iV}OJ@>vSAS+oFR*9h0ytuSn05X04<iYh2_r=bn7Cpv;-
zaJ;3)f!_-=E~&}a;i`q>sM|-(rPn(eL#j<BG6XLOhd3%?b-0V{(U8;w)M)CmQfxd5
zjUznzv%~>6+a7|BsOSqwY9;vE1oMn-V+c?BIw0tGAvMxSerVh#^JB#0=QAw#Nf^?u
zd7aW*D8i86rUAQ;>meZ7UL<S0ECQmMhKnFnUidf(tvxvdp%Wp*+DSzSA=TpQ2Nl;r
zXp?3jL<J^gB7{ggsT3h3T3q~~;sOY5QaQpRggFQy&Q4l{5YjBJdQfrYgQixo8NZ4+
zYrXAZ`FBNgf@|Gt6>8V8shTK%Edzp;e=Yl%OR6}fi>n`Eo~or8HXMoBF)qV~gMB!1
zAGroY=B<9$-*2<+AL-J6iKgC<Eyi2@t~4RvBLw@KcaFc+PXg?%hJEiIsC8AL*tZ%E
zzz@LiB}BAJzR`s5JQz8!PnJeAHRKtNdym?2B|(SX@J{qe<64Et_cyMbcj1Vl^Zbo-
zXl3}o81ML%7HqY|Z?)ueTScEn_ynFm3rKTh@e3ID;iOR3`6UEM)A78RrHD+^8cs+>
zn7X2n@&ou$mwo3?1m1z^u2O+h$_cRMgmf<vk*%no7s=w{8FA#$+3z~Ued>E4h6Y^o
zS7S)e6l0rrouDVH-}|%qdLF8q->;d^pbgIZ_u+5)9?*m{mH0^uWaI(GB4id-ngN<-
z>^uJjWB^S?0;fohy<GVuXd)uA6<xdm$@tkFS0-|YrL4CR><4@w(EHVjG`x1`TWCHs
zoAks^h;E1vzPc25H38@RQs5w^GDXsOWqlNdHS#Au@}+Dj5A}^0Cn&PtB%tB7pi-X$
z5Psta7+KIH38EpK-X#W$9noeCm}$4C;JfGoDXRh7Qw}Z5_B7F6(XMV@C+rvXq8Z`l
z;?pLUE5TFeEqGK}cB%%=7?^{b3rm${4ZKnxX9$IURSqq{W$@z%@LB*#+3nK?zA9@E
z&ybUczo~>$lN9#}|0VnJtw>{bDM9!hXCL>e#h8$(Z?ZITAmH?QxF@+qR#AAY0sO9M
zTseurwcv>R)QROjZ5*a0uzS8t)>2>fX{m2c3=PC~uG#0_BVTfo-_n={9su!Wi(P^W
za#ROiHxaLEzyL?3WNGXkzOd0^Kf`%Kg=ZzfTybbx(P|O`L_@a#VU-<RrtFYIJ#s!t
zma6}Nh7z*FZ+puQ%U6&e@`wdg5a29OMq6cv{Qo~>2V+eY0>mL!_@BG)<iDr<FqjOb
zyAwtK7&=XH{=!cqp9_)!U5;{(uy<F|{47>fIfPr8(3j5K{l<KNzCR1cfVnn1p&vJg
zks&qU#_cBcEpfUAPFTJN+7{s?d!f&M0=HVV;ykmy62*3dbb=o10cs<tv<Ikzps5xR
zJ<~qa$5<et$6;pvP$$vdr8^0_yaxzZY<;Mo3y?p#vJf<7BF>wI;m-!6f0lj>Jq|;f
z4Cnp=Xs&Cn6q3hC$neJXco}iYBS#+yH*_4v8Zq7Fl4?I9SIfgs7rAoI{JkVNfWckO
zDcj-qcT)d*JMt}S<hDV`Ew05d#qu7p;5Nv>v4GUg#3GqN<9x8W+v0XunBKa@m9FKS
za|iYGnm)vpFMyywRSj{auv=Ww5s$AY)}J@70#V=)P?F#tinsdxrr21AF9Eu-4*%GM
z3?Rf@ho1mUtixDf#Ufm8U4#kjUV+PduE4@XiWOMrLNW&Bz>`DDf;oL}U?D*{02y(g
z8a3rBEL-3%V)ekGa{YSz`VQBK{P<ej@8x^NI^6Fy5t?gp9@Y#lu~^cuoR?zNE8Tbg
zc3@!Dn}JusAj(zm-$4LHkz4q+GzHuaErKjKz>RNQO=fw!bQjh@tbZs(%EBt!y*|UI
z4tQAENMP+LEQVN%VU@e@@?i#Tm)~O0#sc8U2HLQc79j(&d0j=Iu4v!+-!P~v1P*@a
zigLoQN!0*;O-zxp4vA_D+EF{FAkGOwqiIss4k_zZgTAXjzDF#oub+^n<0gY?r}3gL
zAzIx1Cf3@&1KbSo%K$eE@#k&`FiTJ!tE7(Ql|v&jG8?;>R71X-j0H?Kf8=66rk79q
z-UejwX+Lfj!Ogf750{t*-xDVsVV{Kljpu)r&_hlobT9rMb^3ATR^n(k5X|<(g%k6e
zAk?(L5O;qYzTm|^|1mzmO@03De7G4uLYCfP=l4$g>h@DA<eepW(~?(k;Rb$$42K)}
zBV__BNnj-bi@>cS+JQ42$eJ7nil{E|k0;90QQl`pdAP6Qn_oL6(sfb5@pyTsMN}6B
z9FLcG8d39S7d{t9Ej>~fuIQuFVI%&UG8G&>mufBoukoqx0E5--dGH)RulBOyZmWbM
z|E<WQXj1%m1b%^`gF1C-+&*4wo887=UpOC|LYjs8VILbE2!DV-^9sK|@Rtyt^VnZQ
zc5yEBn|O%tS8(o!I2+YBfHi!ReYZ#RZgv2)c%wtz1DJ1<!w!kLDaq$<TF|04L*C~P
zKbiS0^2M@`7qL&(lR|8$Hhz*5E-~eR@*^863tYA6)#`ba!T15&@m1tu+vU(yM^k}7
zq@(}`h+ksk$$JjyqDav@Z3%oU*0`o4ryP3G)yg76&x^SJBL^aQS7H2Jcy1Ae8Ss60
zMgzyZFbNKZSq;DQ3Hc>A%-_DAfjqf8;YsONKZ;)2d;bh$B@yUiXsk=W5Mq<|3?)u^
z$TL+8a7G!ni}+1FIGKf?Fu9nXPM6nXWw^w0o<{s$Z5@4jSQ-rOi)HPnq73*|@kOYu
zBu*@EM!~RdMgkXmReg2)2z8JdHi%Qnw5I20^GGyLB2WJ#&KdE(PTJES>s0URG#_vt
z895mlfiVX+5qQM}8&izK$a@uSi>M7+fj$^=$aJ0z)w;GJ?pNFNm*D}<>nSr${eV{!
zH#);)yiVYFlh+9+24ipgIQmFYw_yZ8UdK*7WO&@r5EU}KU3qvJ9*+aD)uU?)Qw#|8
z1s+Hmxc)K@I)?Tjh}~~u@hApCsd3U#q1=-jouJmI9@1aB3uDIc`c$;<v2G(gSyD|D
z4@wAU**S2Y61X4gX&7I|5Nvu}xSc7XI~|^;{zw)B%)mRTUorr%)UKe?OL()XkBM>k
zYnEAp!`#J%@&h_Hmy4mrKtmF)_d{hk9!UW|P}&NIZ!q{FGsyL!LG}(Q<P+4$-XZy7
zCE&I(kNh4wXKcr?hzy~cdI;6jL#U=60!e&GISZ6ATu2~f1|J%kBgd~Eg$XhGC{&C%
zy9!c2(%(UZem|M8{CS}z!NuQ0gt5;UYxwa><YO6uE5PJAeG^Rnp2#ne+0*2336$Qc
zjZyqmQzKt=K3`M`T%1mad}0n`L*IEXz?4rGE@mhaO`-$abPw~q6y4Rmnn0ek@DuzV
zmke`>xF|9vV<PdxntvNVgLN%TD;3v3_rnc6)Co<BwEh_K)0a5|-iWeWulOW)uhQNg
zow4vv{3;}VRXWEPn&nFAr60rACWQ@3%_&4ijP;`&NCCWy6gRwwW&D`F{Wzlbeb6bR
z;D>ir@2uPrMGGI(*D*m$g-!~0{Pp-@8Dl=e8?mhU88$M0F%#bLyY?Km?`k-FMcJ-r
zVzH$JKwGu>$FE&)$1gIUkjr-84h4#`@dI@7)67M!{JZpK$!?MN=(y?+9_LrcWntzf
zzkv(FCT79CUVD_X6MTdTMC8GWaJD=<d}TWX+OiKtb2faK={x5QS-r)lhLI27XUnH#
zhn2EsqxgS7QrU5%_;-!s-$n6o{FsPI{N;3+mH$Uo_LVjCDqlY5ILb%o{Y&Y`VEmWT
zkIl%xlzu1!|E2U}p8QMc&-)kCpZ_nWU;Y=<U-+M*-{2@<D9d_H+alw7AX$Zqs%cy)
zS>+k>zqruXu&;j#&92}5sZ{+-EE`x*JAK)Qpwq`Gn?s$OHDu;$@lgfVsO%3F{EVSr
z8|2rDad*~vNK=+|&BMa^@DR^NthqMWamGA+rDtP5<N2D0FETFd0=)mj1N@=eW6GY0
zKl?SM;d9unC)j=N<BBwTw{Q*QD(qL={n@W7dyY*A4R+_A6!H=8y~i}bwSlk0`hE*n
z2t2op(XOi=qm5jb3pr&h+^7Nd+&FTT+R>n2mfKMlK7Y<7SSC*foIdwE3onCl&wea=
zmGF1O<H0C?wawhGc(f_c;p%Io%l0?z+sCNaRm&mAjYf&~k*mh2xH%EGhTw*z<-9KF
zcHE`pb2n5}`qdNu>|M&9!+!f7i@ZpB!Ea%|c?0yGE6Tx-NO-jCUi4^#pSMS`r8enJ
z9=uB3M|$^_g=cF;_o$88pXBNgvR*{aa|{Y{2e0y|gP(_<5HGF#cPv>}d+u4*-`8nC
z0_(=zL%4U0l0ELfZ5l^;3r5$dv5&)vJ;=}4v*p82^+v`0YCv53mRz9Tg3XP?r!Gy2
zv>*e@7fTthcus<tX#9}5<UOR=w@E(;W8H;$b(Hj@ttngM>wWAt-q;K|!|=2V45dSc
zDGS7I6E`^5pk&-&`5Tg?cVcMq;1zyv14n>Tcn6gF3a_x_i-WD5J~#&?dsTQiAK|rh
z$Ao9nF8#9wlyXn{LyMnvu%A)mc75$bT-#A?pE`_nm=N;)1{L5+179pQOR&))yKx~^
zfAnNDRlf}{k04VlD)kZugzytT`69DiNJTS%-YxH7{$TRpMr=5s1RWUt+P%2{6P96|
zR1`O2A3JJ%VE*39>#T!xf|l#QW_*(9Ic5A5U#}an4@t{MqdcGcU0ipM8?irD?5~I$
zvGw?k*gE&_M*@QtxT`#o$#*~S#UlL=XcHMn{Ir`%3T_tVuOL^ipsqMHVTF=?T&Y)A
z7$3`>y1e|N1uZQu!uCM8sf10iIWiRNz^8st$FnyV8}bfXU^5EC@4*-nZLo2X3^m6A
z<$-OB#MR7?GZm&9{Ol=8_$6!0i?Ok}ZH`XBEDwy5wXv{mdCDc{*cnoq2a@&JLWGd3
z;k1(GfmGdJa>7Z)31<~2omP5X%DjU%p+&S8qA2`)g1X2oD&H(B-z+Ns%%X}=Tp=WL
z8G<<oLs7x-e`!&|H>N)~5}3!ho2Uv6fZ^8!#uB(4<1Zcb&UJ!dZ#pXPb%JDXI;!t=
zf@p6#8o=uW*^YE<E7G0VR-|WQTaga3r#taA#R=klPEhZF_d(MlxDILS(sl8UbxwxZ
zw50E2m{#o+A>z$uT!e_%n{g51t&8B6lD<M@UX}i&2-l=PC6LxQjc}Dyz|(4-0uNN8
z>6#Wv)m1<v4Pw8ou5&UZ4cyd|1`#5jZ^lK4c#attA>wU(9mueW{HxNPB3zT6De%@f
zjc}Dyz(Iz<L;1`H*M${vN?uIgha9r{qLU#hX%;0^J4J|iy%`rF;!l}z5h7kH^3|p<
z65*=!6(U@d{-nTL<21rmP5}oc0uPihA1)AWW#E1d`~?bT%N%=PE5_~Iltr*IVFMk7
z-XgL7^qgPNA@-PGaO5%Omu$zz)x3Rq<6L74qo@}s>bw-?f$fW6ZL&J0^thS#@B>gO
zPn|~<vKO4fvQ_yq@e=pIi``0mt8@G-O8fu2!D)Cm^M5xuJ@&kPY%sI_#RB&KZg5PU
zLm%xg|BKuI|H%di#!zuRYFyt9=XA~ENVX??Qj$HX(6w7OL&CoRKbDy6tOw(^Ww)K&
zYL~n_EEv7~`7&^xOvM3j*xNdBQ+drOEcQI#RH~h0hsO4l`nq|R3wqMHld&Hs<GI-S
zhdghd965@UBPRo?N(ub*;5PF#jsyxDt_my#YU>SL1hkQ^+>^wE4WSKW2hxQ$*==f-
z-9o4AQMY?=U~vKTWx6LD#myDE<v%#ORdga36wFgQ<4;1vJXI*3O%?NYp<?y~ci^WY
zw^zO*F4ZLg)}veWTftjshK6^XarGrCi$h()zR~kOJN`y?V*t)kJ?_0c>y?)yyU%V_
zq2B-1=eGu^udzDwEZ5kjifimD`w^bSyX=fdTyxodp&ioBa;IIE<xac!{_#uhvx9Q?
zg4r-(*grqnXb+sD>J9ppc0GD7t{=fu$lK4p2X>_%I!LkYz>gzB!P`&YGt1a^MCQ;x
z$K0=q{RixKPc);keA)#b?Hsk41vl^M@M(9!7w<0|Kd=Xl64oI6pw#K41vhzK;0VXF
zU8<&S-XD52M)RFNRBTsp)hEnu4d~7zl@}VA5jy><yCXcV3ya@CCG-gX$_EmM3fY9e
zBJF+KO?#iXp7aV>_~z5hcMJZ+6>KADt~2a&-S$4(=gzeE8TPrIar>NM?=$Rkz3hE$
z!k%|3Vee~0d0!azzQeTlMMQd=Fz}s<mv>l1+l0OERNUToxZB=`yWw%QD(roF+}20C
zU$?yv$*gNnW1qBSr{Ddyx;qe1LXW}9ciOP>y$vg$FB|{&!^$Uqql@G0VDv)V-o1MD
znD&uB9|J4jllDFQd3pRs*Uk938|=qn1Kdda-e%EzsrB~bu<vaY_PxyxVe<>X7Kbdf
z@4X58URzK5UT)aJzOaA)M{%{eC%$EWtG+IN;KHyC^;eq4Ix&QB*IVErX~j3d`FthO
z_zzLYDi~QV#w!&2_+gNVA=i4pI>#w~d;xdHxb)x5q$BklI*_Wmb^Ktvxa6C$m&Y_Y
z)}u(9y1n7ZmG<p>z5rs)epO$3Fj9*l2;)?O01%QwLLzj3Q4h8s492=%!X7@hiC@yh
zH|=lpD)WtS_%N?B-{ZS#XYNVmsy%R}Mn>G<t7JDLmo)L1{Qz$T3VwizxE&Yv9!W-5
z#kL2|F)!)~WJp>4&KYI1jlt$@{iovl{mp#6xS(hH_|SFFev0?N;rdUcNzQ)7Xa534
zZ!~Cm)sCxj-~;kr2lj+0YV)mdcIYvGq_hI;l?vMM*NQ*v#J7hr_}VHz#8tce-Do%b
zwVy!Iuq?Lk37+aQYq@81$aig$x(k>0U}Me8dvIHiJ#ZOjd*D2_CK?C79{bAKAA_eH
z3vPn9*LqKAs$CXq3Gp#rn(TaN@^mtg)rUUC57Kqw+MTxGRxEi-Th;C4m>2RYou0a=
zExOHP+k6yXn}#9l#?SGj51oO%$6M(81plgk>eiW7g=BrWXB@}RaFITA_pkA{RcKg-
z8e^A!@kOA<y)W=}Y}%pP*cZ4|*5J4{g}R=0{l|Ds^ipiK^_TE2f>Soz|4)1G9vEeD
z?vKAA0fJ%@En>XY)uJYfVG{x*+}tHu*wqAM5|QJjn`N_*)g+tlZep<3MhTSd5{%!f
zr9IZBt@YT}qny@SFD(X9KwCASQg2mi)w3a9sI@{Z`u;wjnR$2Ly(Egqp5OVy7qaut
zGtWHp%ri63JTvcH#@R7}aUk|}gh_4-x$>1rn}B5K(*{hH#eQ>)pq_v8Ag&xtE}uku
zaAG|ygd(oXJ?O$GBv&O4;SG;>VMR8!2jM7-2MdMFUd2-S*iBzWviQU}IfWkPk#OTc
zY%REwo6%w)ROed8=dUdVZ4Be`dyAMz6j^Z1L-z2Y4`n+6i=`(=@HP0692a*Zso6P<
z=W3(VF{B)-DaOGS2Qg{(n7IGInO{AOAtUDKSdq65V~%cX?$*RS9KKz}Fr)Sm_|?MG
zO>j;qakLN1giS?@T}DRn;Sk9PrfPBYPw_tN-Qab&a^~m|KD2>Do-2-fcJwe#u|gi-
ziCyUCbm=t@<-`ws(J!ZlI0-wdxPl6o-pb8XC}oO|Q*7rT4oavC4o)@d<^4|t-vVh4
zZ-pne0U^Wi>BD=;W_0{7R$EX{vEa=aJ?d4yM{P0eeZU+^bo?Jxn1$9@(NL)ep;*-G
z?CXv_&@>Ca%BMRUMcDpw_)uI;u_zKi)MgXerX;c;RY=hgy*qX*c;RwZ6k6g_EL!~F
zL9u+FvtGvwWw1YgYi?x2v|b>>#-lSbF`A1D2_gm98h=9<vc#{^JqU<>7jhbgHl^gj
z%g*jZ77y{=l*RW|*ce$1pL@9MeDkO{E0e)WiiG^ka&-J66ITn9xJs7L)uezjJPpwd
zdm8soU-uTLe{d{0>XNuxP%cP#3)oPUH&<{92|1l7rm*p)8v7+}s3N#PAc8xog@LsB
zQmmq5rxd$S#+tEM#BxQ?m7Uf^7jC6QG+Yyxi$4`@HKOep|M1lo&Ld&ErFYr@+-}wf
ztTxz07}izv!`)Jr1tnH_88YB~thR63b_NguiiJxPzlnVYTnd8~3^H1xUgfWy&xzN{
zH}ZR*6cm3utw*?x)4C<u*T4qp5v=IV06okU<Io;xIli1Gw46M9j<`d>jr|M}M@~c4
z82%6RMC5rrzzp6AS17*NT0#7`B4;d1V#M|`Bj8n{n|XC|<`yA6Kr^?F{I!=c1W61q
zB=Sn^tq;jH&yIqYZP`*m0TqhG8mD!rY_uFjT-4l@wwAwk8O?^uzvTDcFA~COBqUQO
z$?Up>q;Uq0B6BLpPzW&6r8G_v(m*3+UUN}+vD*-3<P?fxYA*RyVJuiFhi?d)h8)6Z
z-!G9_a0Vh~3H%tG5e$aRU|tbpWS@HbZ7dIDGEOpT+j1_mf4t(`srdF!j%OceW@=ou
zF0%gtvd@ZLj79ovT~_!7`&jV8&P&kU`fpMm?BZD4jt1R~@+hP{4uV2#Bce!oT#XNM
zWK@LhN0^jHo1mGHM`7&u<Q^FZqM6lqz$(FD$P5<iWQ^*7^|O}POR%$?QE_ksbS$o?
z#hiam{@Nyd1l%$HlB-lB6YW)7PE^_4lF5Fb*}v{lY6NOgdD**3hvxV4)6)ach`Q3F
z9FIbe(i#3Ke#|OP?wA`5n{&9y*aLu!h8<1A{>~WK*QQ~w8w2}mY1pgBz-~^%ZXE;r
z;xugc7})2fVb_m=y&w(SF$OjZ)zr}qf5=*W?=+e@vAdU{qx&NuCWM#&X+r2D!V`d)
z5FX^pXoEs45$*)Ugz(FcObF)_VE_;l!qMg=&LV=I%)=38gh@m=n2fN?j1apB2oEPS
zxx<XGg9yJ)M);vQ7k3b$FF6-Cn-SI$VGSU|9KBPyqDLJ9W9s!W@;FxoW$KQ4pAaAC
z5@tz$FPr#i)&2#94ew*C%%sPA<(ED)u}8)r#{H|qKS&Tu){4E>G|p}rgV=Ql8-dln
zZJ6hm3HCDdXuTXYCb#kxV-T&Cq=tV=bq~@z4brLoMDN{cbZN?sCdL55M(Rj$UxH$?
zrtKX{qj<*{6mOJthTnqwumjzX9CEcnx({5!-squvQ7MMGc7q9jq+CBDn3@3RgV>0k
zM{w-S2Gkyk(}eJSg-`%DHh~DVaZL!+RirRT3U46Q;Z<e?X+!R1;7+oX$+{5N>6l9?
z*;b}<S2{k)R>q=2k#da0uVl(^<1NmbOt}z|V@pNKsl9qHm+BsVveIn2R~+9Jd;dc8
zw5wCnq*W7p5g}<k=U6s)=4{GbY$nC^f+D+E(TzS(C&f9yT$p<2&zi=aOk*0;fKIsv
z62p)+KEG$)Z$j%6>o2fk$c5##-ZPbbfl=y~U$E*GfX33k>3^XOgAu$K1k!%Q%8?8f
z#?L^lw(_m8on)-BZDfGMH0^Pl2AiBY$I1re5IcPyJ`!(STJuWmA>{sH7#GTW7rY5>
zFsPg|ZOcPsW(@*xtA$_&V}@5_yOo6poP<Twxdfk8?a?_8a$4d6&PBv7`5PFy?onno
zKB7w=(3K~eJ9a{2;9JCr<b*a1XF`ZqmHZpH`h+cKCz6Q$;sVs)_dibRpIXca-zEao
zyFM~cXuDKXtydUN=HxF(=dTJ;N5U$8k_5tEO=5j!dfYSPtbemC-r(O~D(g!P#GD8}
zhQ~x&Vt-VDGZ0A8p9xp3hv2|(0QqL27sQ@I-dO0ls?hrpVibC=snEBBpsCP*L;BY}
zN;+!7CSG4t=WjmAFD9G73l^ov25Paf)tP@5ju4{*wy@P^{JuHdk+U2-I3F$C)_)N;
zXh;4Qo%!$+sAQ#sv0H1v3ND*__+{ze06(s@e0aa~egocz2QakU!ZAtgZg?Y)2>>QY
z`jH-r1e*$ukI#=`#2ed5nppIY#Wnj&F@FZHDg0ca1%`hG=(d&eybd0(#wi)L%|o@3
zzt<MO;oADqc+9(8+v)0q5ua$!q6&oe^HSUkkvu0^*#40P?Az003lvQB*3p>+ouxo5
z8=zqowH}+QVAlhN14k^*zA6hQ8^8G@u#M67%b0MhDc&oVL1BV7T+Vesn6g`QFuQZP
zD_TB{G<G8cSjcle;$z*Egl&;9tU^n`elB5Osg1lpf;)q@?FaplV~H`6kKclkN%$=s
z$woMi@JKXShz#>%run-yZ($GpL}%ra+%WbmEXfr$9S4hBxKs{jGS^`Mho$^Hegup6
zPJK$QVEY}iXk2a8`)?>j<PG6CwSO%`FyW2O4zXDf2$r#N!W%mu_QigE9*g$f*vSI^
zDB=H2_zwWb2_Ec}(FAby#UbPlQwUg1SRL7obAWiSxnu2aOtWDgjmJO?BfT)bU-4RV
z@xEAIYcAd~If!L({OE{#JRF$Ia1;opuqM*khdT=~`&6Jpvz$BZBG9St#C-19VOYu(
zT*Zi3|MhryGKeQd=O@5N3d2VMZZ8DgB9R@^HK5qNYPY39aeIk<xqt$Zd_;D2-gbUp
zP8ViMAA_}mEs8wlyn_2za2eq4{NDdZ+YrN0E9Sf`ILvo5Vwall4~<S)Z;v$gX+-`#
zHi+m?ME2}@XA&+q-MVkwr1v6ww!V`ieYw(CFlmpz-5x35fHjrXmAisluoRM!i;$Ao
zQ^4yGc)$hEss~%>JJ|E~HIl(bP8(eI80Mia<23;vp#4R2Ht0L6^MA5Gf6eiz9nR=+
z&Us%7L%h4g<0568v2{j{8$JZSMSbLUZXE4EyE0nD`BOP@i!Sm3#KyLO>OafHl}=GT
zq4v+k^?WszZ}NDYtyo+6CKepZdE2tYyscPQ$qp5(E#pfoe<bQ4reQ7;A`}U7>h?M4
z2J_S{u4TZ9yPUgyTg=^ZFW1d3cIB<Vz~V}^o2$x+iYoYyV1Fz`f<>07$v@#vTy*x?
z+_qB3gUu_r)Y4WHMN@J{MP?yQh#dddd|a5n;R*D0BZUGpwgt%ygRn`ziBieGz8hF5
z7alEAfCFnN***v4j^6M7LSz&#Xtzk{VTvuX^_l-hF%^lENO-17ygxws+@<eai-Whm
z3`AFS*+JL5so~>I#p8-BJ2*^}GEqLF65>Md{i1-x>R0GZQ9zhke*wn>ac13AK)7f2
zRZ&2gSbr5uy6nFNJuN&dQ9{njZTa6i877F(F=+bKuv+XRWFBSDw8)!~n!pfE0Qa6I
z7Z(m63Y&`y-(jFuIIZ>i5*82MF6e8$E3qUNih$qUpt0HyVpZ|2IL9SXP>C`^d7_B$
z<qL<r;3Svl@~?lC771jm53V_AQ6<Wv*sulxV!4D^42&WYV#x2k4tcZpl^@|e^(m~E
zei&tngm<~t{COe|zIm^v_=(Oud^;Wg&}rQ78`wEg>*_;CYww#s7Z-lj6hF<t+f-np
z`~b!F(7^Nfm~7`)<w@L}xs5;2L0c{O4tlT?r;-Dw!iF!{>f3CE4JTZ5a1W_~Ui%2v
zZ+A|N42esCxmcnvnAPXciX13@7sn$NzY{(g=Su6HID!AMk%c(WJW-E_Pl2t;J0!(=
zfkGUV6wc-LOsWrH8?S~N8RC7I7rOew8zxsr{#g7pZsg<>uaEinEPfw<2l00({~nyh
zCN8unShM3$=j$WK%T{qG3T@;l9=OhyH2ectM24{Qd#PlEbnSgp<~mRP5Z^i};5{|P
zPe}n^a;g$=jjOK~cW2_>Vne*=svr|TpHjYm9ZSC0?mFnKe4=xFb!7X<)%L#nEZi=F
z%pag^-wRJD{Q%>ZA_y^j9OXC3k$+EhAI<^Vn*IU`?0aTm^IxA7ARx)|3qpQD$S(-_
zLBCRdk4X7N-lK-LKs7Xp7Ay`D`yhN;4L&AreG{kb9f;wk7q&OiALAr}*ilvVzlZ}B
znxG_c9soYJq9Qx*$5e}s9svv0am@3nF#eDZx`pLb`euGa-*EtlvWth-!}Rqtv0pC|
zbnT+0lny4!A76@XP@hlcLT^5a@5MMyKZ4jUex}p0E26v?D_XODheFnz`AgoV`ZaEi
zc>pjRz!N>vp8u1lLfJieSyCfm-3qO;Cr6gx`)zc36v6L=CJ7SHyfLZ7iHaU=jN*>@
zpX>=8sc#ik>6zCiP34#G%8y`G1fO8+NkadT1MG=di;lGP@ye-JVeS1^+>KLRIb_EH
z=;HpbwT~-F%?6j^DviiK2Wn~#@Ak~UXHkxw-vaaRIWO0qjZdTb_w00JPoHMbdeRwP
zmg}6itN@lEdPh(`OLmXTa)nMBev_Hzs(AE@pF!zz^qulG6FT=tBx<PJjVNyRb}%PO
zah?*k8=%TOC$}5y!p%JRl)06R6;tjwbp>DGVa|nd)I=`|6t_At=P(Ug@1%+qT;nE8
zR5*0y2V&T@DZ0oanM4E@n;&DXoK3u7muJSsJ4RlOJrBw{|N7rzW)e4lAfzs~8EFh-
zsDv81efVY}jK$bGIYE5cod;!{zji7_&P|*s@BfZ%wt=V0kJ#x|gmJ3^&Rl@}jQcMT
z2y3~}RrzY?KJLdAa|ho5G`=ymLwMc}Ph~FGnD4C}zDUG2gSH8^2sY!%mjXv9EbxHD
zagJu>PyLDeMvf6A9s>jK;{FXBKPDJBbS@cC8F*KCU|@iO*lX~xiGBzv4tHa^_2F^j
zFuhfPp9hbHFdmqJx7f?TA}2uH1w_1#{Ek1rc=)n7_GNMGyCwE8W>0BiB##H=`BpV(
z)h#tB!VbLPZiGK_%XyvL+aXM6J1zn=7QeXq7NexAr{T!W#Wud!@Upzt`z{V${~Hi-
zNgscm;<rNl79nm67{XaRB{XIc$;u7w>o3Hd2G@@;j6;R@ATKQ9UXCk+?sP@I+7iE(
z%fFBTUu`nDS=`~;v95>F@G$26jM=xeg9@q%BswIC+j<Zlr}7n2>AQP9U`QVHx(#~k
z2m{GYaQNi+J_8j9r0U2&^Ve_T#d$Eax?${bnhl#t=G!=aTO|0Ek^44^(IL_=az)nh
zU0$rxH#z2tY$V~D$UZC^#7TrSCt=rNpZqE(R;=)45niC3hr=gxXNNIh!Dz-N1}x~t
zP?HB?QfCNs>Nr7R>@e4wk8*;3=9cXPL^z&_QzwDQBk_@o4dWS`8qa{T8qe4;p0U}G
z4fI+jD8@6k<pL_kGaxk7d7CRLB&p;O7xDn`AY=|;_)vnHsGaKu;05-baL_wYU_`P<
zUUhl@h>o2jLh4J14WmdQwvP?5ef%#J8wv%d_o3i}ZBlkU;Oc)ze$4;y8vMj-byA73
z35m94ZU|^*-DFVJafYW>Y*(MCnnGWTa!I}!R74pdi9|7Ry(>iE>f3lf0z|_R1r;w=
zf`Pvi4xjwqWzf*TTYY@&&-0<$(K|3cb{ImGMZ;I&v;tvA7_CW60vrG)YJi~Cw{Z}3
zh+_eW@7TuH#+tDcHG}E=|J{^uc|4m^FIf3sst?Td2(Lp;%tAmu%Tb<#WC8U7_rfS?
zIU?7dgb|O0V^lE!?T9Vs@-BF7?7drJE_hL7Lx{?UUCExeIoiZ2Bki!Yv5C*(^oj#n
zosr&YJ;1?XBgTF%Tor@c;7}mZ#^+(gO~z3!PvDSSaZw&oamsDi2`JU)W4l5q|I=wL
zn(Xb+6lY&GBg%yzDApUD5zK!HEie|%se<M=Nb}TBq{o3eW*Nq0@>oG|Itf}5_GxDR
zNUsC*mU^e%$@dn#FOE*R6DMw4oKd$N=EBm!^2uuKblmtR8i8{YZYVmi3%B*+ba;#F
zVaP^9H>}rW;ZWdxmMwhG!aU7^-IcrYYj;)m)s*4RXbh^JawgVpp88ZRw$q-v6Nly1
zvD!gGIEq6o<E)0QU*U@EhOKCSh{P84m7j=9t1$)rU+-5(-Z+2_!wop4;htwyiq%;!
zMxHtF4EBH4{;BV>FJ-BfnIwz9$1xH*{mWBQ*Jql>N8<3J2}$=!Ke{siCH$@Y16=%l
zjDhLL7?*C}3(M<ZC+?HpT@8jSMkeBTLYc<Li9zbe7^7B<L%P>KRa?1L9$w{&?7|gF
z2X^<h=m(y|iO;qBs<}|SI`RlQPY3$038gQgFR0{N&#!Q;C+?5F4i`B;l7HXstUZ`g
z8{3OR-q3Km?O37_c`x!x(B3ccaYOXqvi7XG<~m*%Eoi+553h@khw!>+8G`;6mqqW5
z?D^&+ynj+mWoe5d<#!^hFrb!UdI)nA7Zl*4jvr<5%Fk;rLG~lK`|Gu8X74%N_PI^g
z9N^gs?a{76IZy&dWEi2tmD@VI?DxZ5V(YAqyn}!kul^e57$~VdB~gAx$0%jTsH9&2
z=@;ykJ37xxrOjQNBTZucwHUF*Q;qySb{^8T-w%WYFOe5w<$#WqVE!)+pW~}^{M!L_
z>5oBLb@ocE7(=I!zu_J(5=c7>lT*sTL;Ee_KmrZycx+l0{YUtm+Q=b<C)#Vm-u`hp
z4gSXpvCF%2*cF|DU54?=yLAY2MeiO!CS1|m?t}x1VQh$phA?(HuSkwBkmCd#3<Oi%
zU`ota;hZN>;3Y;z{L_--V_B*(b(pInH#wsfauo2vOi9ifhrF&Jwm4+Bf;i)lK|tcW
zAb$l9ji}$x0x9;f!<@2;wZc36)A!k*?ijh7Ki`clC#>S{0yZ3irce^@mVI(DAo1@j
zN#6?!tI@Sg!I%I;K+HYjCN$i|h6`V?XUG}-F0GX4^v%`LX}`=G=7~)oK_gA()oL3#
z-3zQbp7e~1+_tWnzc~x{;ST3-T#pbtKFR2yoQuhXkz<ql2aF7v3Yt(fgycWuP}&iR
z>2-yXt&4E>0QLqP!ovKW!rd-iT+dlNB<2E&|7x$?o?r7+g!@ciMQ>LC64jBv+HqMf
zC{2lR2C}*`JT15LmXS*w_*kdo2>+u^{R!p-GAExE6G;x?fFWc7ovuMXGHj0wO6r`8
zwO78=`C!Vn78Cts`8pzxLS_XJ?-ZUBz<cDpFG6qyYvymlZGRSaBM^3Vw0u)_)?Y<#
zu}5i_11o0NpM*f}n_AEB=T3Qo9qust9$vXN5Cv2L2b(Q$^WXPrBjh^05CNvcmq&8G
zZC`VU1%pDtIo1=SorRGXt0Rv?%-8}YECG(cfrPvCfPl!)927KC&x^G8)tt<|RQ3ZO
z<8t6{SjDRzNr>rPl)@Bf#sa9G$e<(g=h*i^d8AC)6V=f^8X_VITsWK+Su4B|Q#40p
zZ|v`|OZ07ajO90`$X4E^CPv9Gg3tqB=P3C;D~8YFE;SMGay;M%3=nsziGV-F19}la
z{g+feE!F5Ps&QkJJ@T}ji*ygUIP)*cXkz3Q9rs}1bFV8Z%+lKEMj`OY5IDyBIBo&+
zLI0?U%%8S)5XWD9FTZydbS5uJ#MT%dnYD2r{5X*dL)5J}en41pXv#e9HoA^QC~sbS
z0##Osr!NOr16T9=Le4b2NB%)0*~-r|1-Rx4AvovsZP}ZD6Ka;<m79=1ERo+r%ewaL
zEo`x_t;shx)y0ah5XRrN<xItjS?TdqlEkIai0@E3?^6@kH}O6-Yf}FsP)0DiRQ52+
zKYoDQ7UTh+ChVB>3@))3W0kwypNF&G;`7187u)mH%jmoF#Pv->uIN%Hw%74CwPM!t
z#BMz15i{%W;(lIvcE&+x-*W6}!7*Rhj5jr9Gv4oropDFr+{BK&>d$jO-tV2o+d2y(
z&y9S|Y!72^42u3KXYuy%MCR+IFy{3CnT1(3Y>Em`Fz>>X@0hOR6#Mk}UcBd$$|Jk|
z{p|LG+3kn2OJIlLpoFc5I!D&LktO?`YtIp1HSoDE7HVK|S3sv0anmLBGL6fk^ZKTo
z#c_fK7pHZ>zAugB+~~sMr2&DEW8IO{&tfHr5=UjUB057Oll#%bt0o~TFuQMO^}Olp
z^EWheo^K-TC019|Z>lyl9JUxHi%t&k?D^<VQUCSr^U80`->{$S{T>5lHk#-2H$I40
zRAx4fS}$%NKIvNyoDze!<KSVxy$I#@EX-P7yZU@g(zv3xifp-}_X_@)P02Er7%%FZ
zehhbJE$W+cjMKXtCOSqilPjHByPVPTg-+O#*;D3~@|ODnP)*d=zRMP9k&_<Wks~X2
z&I9<mP!&Dtp2}CRc^!xH{0w<w))r?^`I(&xYXp;~`5RWEi%p(k!~#ZqnMCuF1@p$)
z?8RS=eS`u^D$jI%G@YHYO==`lr-3iB5TejZAXF3Ww8Zt&EtE2p(fkK^!(1-a(Ub@C
zd+k(5yE$#T@d&6_Y9h!%X(Ff+o<M{KhP6y_eZ-6BUQFnt2{<JII;et5OX(mhK027|
z_h-;Y-@~R+C>^PfDsXCf4otTsb##_eN2fs@&7`;z<s<bGW^7FQDD8enp^;p!$n!!e
zC2o^Ld#4)u#;3dAQQjw4Lt8$slV(BEpqoi8<@ffWCjF08)47IfO5_)x1eXor=c{}1
zBQ~IxU<~#M{(g@?jLklN5_-UfG0XnWXD~kpK&BBG4gxpy6EZY{DGp<PBcY!r3;o0o
zf!ny=>!0GA;I<KL98PE|gqc;AQdXxVl-1b!>6Id(pKR1m7uvB=xEi<5;sV&Hp`NTz
zPuLV<Qcu;9={EvlfU<?Q+L3c3-tZcgO+5~QUBkNJ#6{^f(GVJk&`{O$rrn>v;bJPM
zBjVq`)ASQ+cXs*x)Ki7}BlD&_kiX##s-|a=kZM4W<5iKY^*$M$d_P#6&Sqi5Ia4m&
zb_Ny2meG%nYvWoVWAmZQ8T}RW<%%41Mjsl$8%=}g_sKXudowwxDSYHK)YC~vI<uZ}
zMyKaHvz~)`%5y|sqK36$ncwlHYJR_O^1{gE2X|D9{!QDFBbG`)M=gp@?x}qGn!Wa{
ze^y1OT#o~)ar>4ldLNZlb@u!%cKuO%<+J%4F8d@+HSbKRsTweBfPP5mr#h?)!#Vvm
zJY?%*WT6Jjr=nla#el807%P)-Rntc}^XVH96V@?cn*_J(RX)~LFSes&!3eLOjxEez
zOR4hf&TFU(P?EKk2lxdLizzYt`&w)|k|h<XHt#-zgBgs)#9lf_L2aKecLcMQotRN_
zVC9Oqfhn@fx%D4NYdCZ&H!oHm443j`^Isw(ibus(?g&2N{Efr5GL_m!;BdLs=}wdp
zZ*ck;v(Y${>jPKufiOlyxHRd*m>)>I!Tn|D);GqvvfEI2)3L%U28C}vtSq(*c=7wC
z#0M6=w(qK3xm61rw07WP5f*))y9oPzm*B(6k_rMiL3cC8b}pLXuCsBrf;f$AiOm(g
z0DbqxR$~TETrt%}VZ|<LBK{KF$PpoYft!4tD{_HVot|pnvBW~M{1Q;?isqH>HtWwd
z^KZ=R{IWATb4SiC0t@G3RX!4oXGYxj^h?;-TOP%ayl7?wZ?US=DJqQI)%UkA(Qh2t
zGXWkBR=*4pB<t%>TAz5h;dKw^#hGI^A1KQPb9YQLzJHm+iy3}B!`EdDKZoHTFnnb^
z{BZf%f%b)g)DHe)-)}$nqnqWOwEG<W57ryws+nb2%4oj`O2FP&(|5Xao<+OJ*>_G0
z+=iOsH_&2uzu%g-a%0b(hDL%4`(;C4V{#mIw;D9`DKBFdE{oG4`1FLTcsX_!iIb7Q
zLEQfS+cwk$P$C9iwJeP@J)=L&j`<JXp5#Akm8Mx8fSDe`_!#gW9DA(U@aM)Kp8e-4
zSIm;)eNg(belg^4Uu6k}eeGU<t0f$;gj)P7!<J@$n?<v%SZ&dkmNolVuCy%Bsse$a
zrRel#i}AnMGRYG3!VAD8%jvB?gtq#o<FA;$Beam$)9MLY3<?&lwZz-v^Iqu-t_XO7
z%?lP-T1(ukg8t@n7FJ34BEpM8{;;oDD=C@L>R&OV)e{c;y}lVsfag{ocZbh+Wk=vj
zUrCGB74SD(7&vz>p$&*z6bh`k!g9uRi#O073R`L<h5228E+jY<26C6r8x92L``g2S
z6<aQ4sAYko<ZlqH&zE#le6^l%(BG}Xx~c+g9j(6Z#nUY_7h3QS<PVpU;KwWJ%wO3W
z@Pt7{QoCFW`LFf`R*EbYpS3jPZ)>(da0*JMi#3^*xpUpM?q=UgPiJe`?dj<7wKp3w
z6tZK~WO*&=@CDmCA+Ih$HZHKZ7s8=rWTHwm@eG5tKxfeFlSQ~-fhfu36lJYSPPEDs
z4EsWUPkVz5ojcd*X$?5Ky)B;hRlcH6QDk0Zdb-8a>R;9FYqqzx2D~1YB*82Jk&U*F
zpf40ch5dYsmT7!Gm4wBuSTrhKgP%2t^+^U~u6@ZhbL_fkS$MgY(&F4^zZPr>B^RHm
zHKiBS5>L3r7i<Wl1i@unXtcsYwV7^7{Fs~DCzl=KeNlu7b;2TK`-AFR9q0@%SWv2=
z>a@9`p!qJe^}+>K`7(cHv)0|_54*zwchJ**rRCHG7NuYtTl}GVfQpK>pwHXoUKtFu
zje&Dm^#B#O`HQ;nztfzP>Oz?o>Sk}yz9Q%jbZeb%Q8fZ9SB8Ayv;Fhko$daWeqXbD
zMd!+uzMz|(hUV)I2R+`fuepTYqT&S$M5mIZxD`&a;%@T=p`S%JnbL=bSEIA>wD~Ly
zoI6)``)<6Wi?y63G_vYRJ}32gxw+R@%&H3be~-V<Cb!SmD7jo~Z3(faK;auDozbvD
z_q#F>Z1b#WO|h*^0ZP)+HsbFhPsnFEYq~NJ5|Co8sHmtDML)aT9WFk@9ccE1J(pT9
zFFwQCy|@_cSjXgL1<0Zb%5)4PPWn<Tvcu~W=gw8ciX<@;!=IlSc)4L<E4N>_iMOWX
z#&!p9mf->!KTFHWUWLJ9irsI7mzBnQYxc}&_rfLu_U1B=AZ1nQOM0bN5^QV<`aI2&
zq=DC@@Qn2Pac~Wv0u!7gm^$Sjmn(v`Zo@CVX}?w+*TiSwZkup7D|bk_`<2`CUou=E
z;YadYR!V;Bea(wnJ>Dzhc?IiMgFi*zkh2lra5o$2|H>vno0U7L+#$oCQU0YfnZ7w)
zgRYT&|0wBGJmx>f>*+1Ax3{zVCRMk;-5)OTdeN_Qv|_fF&~z4>T2-|)?nQTtAYo$_
zYv-KP*5SwRD$s%h^HIZJSd?ujuH;M2DwE09d4eHdjW8;TV0#BGXQ2fPs%adgsC&WH
z8BDm@x1w{^IUyM0p4L#pA8v-FqE&f1SG6ENA{3*<BQ%Sp&DUn}G&j?}#SpQzn=tiw
z1I<2x1<NLMX<O&zG+toW$WekE{wa^><^t`M*_d`pjc<i7w5rp0L9p2ugyG{h0QJ5A
zK}tB(i(P@%v1v(z*N}R`yam;cV8AN|8=*^^FE<P{qX|mJ*)JOqcYC1SN8VWqbLYk@
zmAT=L0xn7~8jSuD1K?3;HDD;%7^sKYw-{C*x^Rq`jlxdyAbW>TZ;@}6zdeaFA**`d
z*E)S6n4+TpPf8}q^O@2_3K%MJq;DF^rzOkzr>qmR=jCL}i4fb-x?YF(o+-LsF&VlD
z?s^=FWI0yXiwZDd0(bCeO>2T{0bboyUEc)PI$hU?=r7atTzttHsL=K0a69H<!VzxM
zd|lrO*NPR(+IZk$;kX5^1?%lA;99G7y#wxm1Nd<FpO19lZpPls7~H+s#A(68&Ar&7
zz8vl#_Q<cNi(T1+a4nbW`a!ruU)6Oh&fOUb=z0s>y0ESfz}<V5u0IdA|LeM50A(|H
z4e;O=T#Nj|?Y|E6;0|4{>zm>BtV6nle+&7U1p43BbsOA*DAI>(y#ai}?YI%?)BO(c
z;SPQmd}6tF|0dwU?f3!m1=sdNUB46V<^f&bOZZQ9y%4L<>wXSC;5NZ^!tH_E1a|;#
zH(cv~f*-hb_ke%6O>p<a?SPw$^67zVf!hza1@8X;LVn?Dzr+?XxCL;B;P${h2zLN(
z!BN2bHPVASuvOQ4=-&o>xOI;r9k@+!_rvXgn~U=Afop-g8Eyq!>tm1`=-J>dhrek%
z(uLcx8}Z@xzXJOCp!+K5!|new=)<+V$>W0{r+wgq;UkbM>Xr63<OH_>?s~WuxC3yl
zaPNm}gF6Jb4sHx?2V4#Hupe$A+|6*Ua0lT!;SRxVg1Z-PH{AVj`{8Qupd8>9z<nNW
z4_pftqw>MRL-Ax?U9V+z7i66{d17vV*2DtBPr_ftF}i-1L=nH4@ilFl!jBz=yrE5K
zc?FB}3eU?w@~Ye(?c5XQoi%gXX(T|Je6p{us}U(q%(fNUv}~EKrj;X%v^FE%m$CgK
zXTwyxX3r~FlvlVY&$1}5Xi=Wko>#FXZzylg#0?Yr#&68IA$#385$`g@TaK?6Q_v=I
zCKqOkSDjav6ivjt3Gr+tI5C&;Dn^OdlpN2HLw1tfB>&8#tz6fiq(PL<dv#t%GA&>U
zKDe^z`PsVuF8n#oDSX)Ttkrq8H4`=%Ic6*wuMzS3=ji(HA@`i#lz38J)p-M1Nr{rj
zNAXt#U3CKDYm0=8?Rl0pIUBQY7{~m*&T#0kr{FRlYmgt7&!fQK3wRd%Iq#2>UVnOe
z<%oCxLbM%~o;|OS>1A&Oo#~*<eB?q`I#3RZA6e$eJ2T!VI71j7))Ah7Q?yvu#RfIW
zKk)5&MZ_P!L12^LyMQ+cyi=KOdU@C(R&!MXlFa8_;OqV+y8Z#>I_3Dx^(m!ZFdru&
zUJrEn3nXMt4}2}j+ni-CMDkYyd@FSRKiO`~_(s|FB+HUyMLggiyg=9MbHJZ%6#i1m
zo$-1R&(f&tw=-X@qr^)oKgRnl@Hbto>mNcF<!s6rZ%JNHc9Jj|?|sB;YSHyh@_G9S
zndFG}oSa9-n-2UtTXnsUd=6!dSDlxE&t}BaI&}TVjCUwwyd`=4<C6Gfdw3T3-B_o2
zR-&G$_ONao!<mou=vyzy5KewyM!NTRBA-mR=)_F<Osi|t5U+;(uGaO1q&r4Eo)4R#
z@M>NE8R_?Cq@N}a@_iTR_kRPQjTL=!yCdI!KzI*qg6mNaGt|E&d4tJi11!;R<ba>&
zd(gKLpdQkDoi+??i$c)N`NJrDrqpTC{t<7}T3w$-KH_qK&NsCiLkCO)N-yy5gst+A
zM0v>ezb2t0$xj&iq$i^5RpiHVQl@f2vrMnI%*UoaU4H^JGRP&ZFCp3#{GEjKtQ&Rx
zIizPi<f9z$gZ~fxLK466J}s?oFs_J48}2693W%2z&d7fnFN}9P(%X5nu0KRR()CKl
z^vM6qh*$T0UEgEkpYZnqFN948lP~@dJ{kNK0Ip055k~klz=_Xy<0;2W4s1*d;N0<?
zv-#8IXBy&#L9hEqx_*38J%zqEl{eG66?h##k^88PbP2x~@RkhlKLFm80scDR%QL_a
z0dC8M4MpORD;?6Gf_y!Hzpld+j6clpbmkXyE>7Zm>~bf)X2f&CW`@|~58>AUZcBkv
zK9`|jDgbZ6d(IH#iZ+C{UX9*R^sywu_`gN`9^lb5B>zUgek<ay2Ye~@OX10x+sR;h
zJ9!rI2JS?E0vkQ2H6@;CyL-o(`#dH~xlG2GpzvNSS71GO_RE6bxSc2B^@w;zn`b=g
zvulu^^*;1_1b~0hr&6|fm_O#D1L16D#zXkMfVX6T{{ir(4Di<hUk>;T5~<F!&|&(A
z6#c6RK>CJ0g&C@e{vY|;i+CoxYE(OheANKIV1lj(6+JqX&veN1{@>{O5tAXGlTOK0
z9%*(R<255*|2ADej`1{0rg&*Oo$;OpA9ucl{2r!$nvC>pPvefd%<<B66Ot17t46$*
zAzi=qaN{9gL%VdHx8LR5o-uzJ<#G+;746Y=9&C{_^y%WUeBJ_|`*V`p`*gtn0D6?y
zIPjly2=wB8R<aGs{E|*N@YlbbUN36^@5umf2E021ocZX;0KW<FmJIN_0B-{P1TtdM
zSC1<ERQSt^ek%NZz%%hPnPcV*=}rSY6F>7g)&{;=Pf|XOfG-Ce>L>mX-U@hK2Kv2#
zcV{P;OEvIs2VpDleWZdixnZ0muW)0wBhS*8<H#%8Fy5JGT{pp*SFvVdZJw<)uL8OT
zB_`S^Mwu9E=8E!q3v|ejQZ1w=9zFZexa@xj=rab4FUR4p3S(AnAO0NpyO|!grYgAN
z3Bpr^xAPhDeRxiX+dvO5G=;sP{eYfp;Xz~6E~V!`;Gyf~&u^9IN_rL`LK)mVdVUPg
zzr+10MYarm1_}I*o_|-KA$qpMGac?<=s6NS8HA{PKu-i7l(5!EkB&aV1@~%toKSdA
zgIVz$ht_ZgxDX+iA~p$3A_trcnG4r~OB1xa&@KL%KoU3_rU5f27(G#WDjDZ=<ylD2
zd^4II9`1n?PZYr<AS9}17iXVI@KYj8y)*$gi(O?2nm0FsiYJcyv;t^+xgNNs#Jv!>
ze@o!rs%4*W6Trvg?|1mq{u6(T_`*le-q+KkWwU$L?&c34zU1T2M^+i99e?6E8+2I)
zgADl#Bz76xOX)cmKce%%yLKNu%$h)Z2p(ZiFhtKLKb3e3fYkG3^G^Ky8^id-^Y#N0
z<-->;^9liX;paR;`NVNgCK*0_A$t`Nfp#W5XB$-FxRVTAz7Y4@MEMdtHzGUY(X+pG
zy_VfOt`H%pGTP6{eL3>OEImL^CEiYiyM~@Vc)kMnCI;V(3Re!7_L}xMepqi_Ali{2
zMX{eilC}6b8?Jb|;Xwt|-Xq!?c<zvJT%(0bIu)L!dHDGma9E?}P#Rm{X@x7EvrxsY
zc(X8svx5}pWO`mgMW#EIKl4F_6<bi_q#PL*nY%}MMCR^=XE8#=BjmOU(qY67iuD<I
zR>S3Sa$=ZYoRFKU!>=JkJba;yL&SXxe?LmY{R^Pt;R|tj$*T4?{(han-NQP%^jiQ@
zcw4|h)-Ulx{(no)Z}3wJk9cA#<Z=>X3nE$@+$`2-WKKT7!lP$z18I93UWs-a<FMSx
z$yez)4L?D+;<*W)INJB%iBr?Fhrpp9SPVakf~rS2$#4ow+khYP{v<sl6M!q8Y~aS@
zaLPS}=6qmL_;%7f5nMk3_a7uP2?An>NUay1dGLq{mxoozS&Z`sc&?JUWq$ALL59n@
zIFkHa%zz#E>%^aU<^w{07Bj?-pBWO)xNN(Z!;5Mul)JVPFQ<aOc$6&pK3ziO3wb1K
z8k;>OBAzPzLLF*N^j(BMwiNM5S@N|`0_4kY;9)1CO$1ZTM#z=$oMMFB01tWp9$f7v
z{5;4PK6>_R*CXGrj-vrU9vo<!f<8TU^!xyS484>d9&5~&*F?_&{IMf+P-1x?NmiH8
zQw9&)q>JfXh#wO0Q9R^}Ar}zsUU*onHd5nMBRh~^&_fMKYMbdf61ii@_ZjCDco^r$
z^w{8`uo{{F%kV>X+L%sAh5Vdozk(;8&Ng^R?I2125gw-V5j`iO_Aul<#+d;R<GfD~
zKN&HdX1H1_ewa=t(^;cJ4iRme@>DbAIe4BXSw5-`a7e)vc-WxOhfKSVo}c57o}Kh;
z#UE>jm_2w3FJC8=kDk3ZEBjAbUc6E@KFc5WLc|u&*YQi>tNdAqzvJ;Io*&|uOuo*L
z50P(rW=_D*IB+UfqT^lT*DjX*gABP14TWg;G0q+IAf9;cMsOUBI-D5epkp;&lX1Y{
z_bA}7=AFmjW8k?7u6XW%hXj7jpUwEW#E7G3|L{63`_?S7#Keag_cr{M;ZHn6_+`jD
z40#@Z6oGiyIvMg2L)NQIjwf4JfO|ICuff%Z!4nnZb40rWSvdyKpOBMl@Z&W&p%BTv
z;NWk{BS^>l&9@OO9=?!(pY+)>sM^FcIxa^w0{4ytE_KOFfVnT>PmF&h{=UnXFP<+X
z@P!1vkiZub_(B3-NZ<<zd?A4^B=Cg<zL3EGH3@W_U=u726-sx%ayyQf{(j}wDLjFU
zA7j*&>Mr6a$rOG`ivKF*=O~X)%ds}$8e`&T-;;i0Bs_SIM}(g!e&r-RUpYs?C#J$V
zSm(p>B%i;3tc%&HP2+6zijg7RSgwtZ#sM>**f<*<F&d=1fkQ|<9Bc9^l#ZANQLn-=
z#(<ht@@EXSxvrECCl2@+a|)b;;4@HQ6RrW@nJ@hY-_MIsIXlHyBmIwcnV*As5?@z-
zm7KO;`NJw)k@=T92bKI&6S_l{>wU`IuG~G!eN(v~D0jkCnXjqJJwv&(mAhEE%aprP
zxn0U#r`%hVdxvuGQ|@-<?osZW%KbpO6OL2qEB6fL&Q|VX<t|h1O67Jbcb#%?QSKee
zy-&H@mAgl|Zz}f#<xWuTZmM$6Q0{EyE>`X`<*rn2mvYxB_ZH>eq1^kFyIr}d<+P-#
zYObZ|q7|L(;Z93=NkvK7Ii;1I!dqG_0h$W+wqnPxrOOu#`2+2inI*GItml+V)Us>C
z!Ol>4b(y6o>{%6>QSYd6*c%+~nX{}jR{F8qpx9Ex?Lo~J#H^`VRx6_y<D*b$bz9i8
z0&X}c-4^2wE6shu4y~j;5cZYW7rD*>Y2{nh-dVDu)8E>B4)#!#(AVM#wP+>HtJ^_O
zy5XP<Q4A+}+z1Q$T0M-Q-a1;tS_!vyXeD86QPWDmw;y3*6O>lsYrz&w?3r`7G$XFz
z15vNn?d$gXI<Rt{yRak;w<j3%td=y4_ba@Ji1a*delOw$kOY28M7rE)rIoA*VY8PP
zTMvBg;nZ^atbXcE>S#khRgAO2hZ!pcF!ZJX;>XbEhW{&gp{_FGH;%KxrzUaXTIex7
zhCVVtoVYW`50A5nps*r1P52T&sqw23^Z|66k^ZJ}Hi5iJ0qTS=q2CQXs=(S(yik`L
z@eO?+GXSb1pu@Zy@eRGZ3^3|?1KsF9xE`O6(O*bky8OP10P1QZzM-GPc``<iim1As
zcsd3?Clsms4ZorP532Z_UFAdmDLaFI!uVW`Kf;apMt`$2pP_O>madV$5#PvvFT%|6
zjs9z(fUz(I&Okhj&ptgh{zkwkSEIi*`umn+0mO5HbmG5?T-+bQn>l{J>M#0Ld|SLD
zmQfU2;%fjQH>Jck?3^K$!ErCjc*d_0-|&AwCB9*wJo}!+;Ku{=lgj@90?EH2|Gn?o
z;M26dDt^0I>?J_rCpG;$0X4_3%drc*x*S4LHYxc}P5(aNnBy;>DB~~Z2Piz{svZar
zevI-n{DXKg$LF$tJ{x%R1fG6Qqnn=8^mikSRn=hAu;&KPk@1U6arv5>{wsiyMI*jp
z@5NO7RQy!`KE!28M*RKCUfKUC<G+izth+{h!`?Zl;u{5P)NdnwqZ|*Wq<?UR2M~PF
z;_?$b2EOo^{096Zz|8y`_NHc)@i$Ayh#seB#Gin0bNtPv9)Y{LR7N!TGzb{+4F6=H
z5oY9nP{kiq@wwl}{8-E$ydRDDOv#!8H{yRY1unUgaTvtrVWe$FJ(piC@rbp2rT`<p
z0ixe%dj^cIWuC;esDZ|p0lOG~M*a=`$M&5meH%AL;Cbmx;lzJZ(~k#ac*%xjD#qE;
zLbAcJ(a)L#v%!hce<s6ov`yn|W{`N#hE|ScginZ%gUpe#(Ni1!Q!+ePGy0um_$1Bf
zPm<wB#K$?w@FU~ny=3@g&Cu(~@S`+CUnj%!G(!(3!}Fn6ZDx>o&qi-x=uLBAHfool
z50l}?K)>0|AVud(nxVhUf!P@C8G0!heynEbn`HP@&Cnys@Z+EtJZ6x1&*msljWf-G
z+4x>;wC7~_37XMPli|iFIss*U&C&j<Cm`}oc1!7U_=*TVM&mXv^D|4qxgE#+I2D}R
zV9d{@3eIgV=BHi3xsApAT&LjN24;S~qu|`WVt(#WaBfdAKMyE4x1X4wKPWi2hnSyN
z6`b2T%#ZB8kI}e|!u*WSqVE`u+ZoJHp@MUpfcfEKG~#nxnfY0u;9Q1pe(DsQ+sVw&
z3I*r3FY^-yd>mFB812ZS0>1-z7Obz{Ssnp_S3Ez)OQ!l@@X6)t=I0&}@feNE(#?;-
z50{;rpREd?%gW6US7eb-WBw_q0^bFkeDW%0K89YN45dyu?`bx}jeHF#IHHT^EZ}F#
zmr>8TjNAMe^_<JJ&5t2}E^jtJh8(zj+5B(~J@aMEr&&~>!RPZSeA3V#i%%|RHb1v0
z`o=t6D*cyJ=yO$7Ci*;8I}5X1srrg*Zpi=M<89_-wYL<!;Dls27v&N^b-wUebfPS;
zo|O7FO~BLj=lOsOxuwv*Sm3AY;nDKdCGd~d)~D$6UI9<fFEg1*KK%keT@TB}AYV$;
z|Gxn~7YU{I7f%Q}>H7Z}z)yqR3NE!VqQ<?0d^n$K$)M+J0VjS>wN2oikR=?g74T!^
zYtrv`1^g(j|F1RySLkAo8R_hINx0bK35*#S&}E`iCE%E?RsF0%r)?Dc?W4f|Gz$E$
zfS(GB-#Gsx-riwg6K<SaV%V+I06zx&4=H|7{lvpPU77elU*T^~sn7L_et~8e@lR27
zIt4t@pY|yDy`#{7VHEh^1sv^2$=?|F+Aul7d>QAx80mTex2UwtPH+_XI>0mKcaxxh
zf@Ul82r_kw*o&j!zcUJ)hH)nPCyoNI5%8llt11^YU7=kMIP06`TQ<Q7_e}Dc1Vfwf
zrt59=YTT>I=idR(l<uzuJlzg_NWeiSrJnB}h0YNezh<I818~v4oHoJ7a#hi97x+hM
zo1c(y?g8R6Gzy(}6`g@@o51T+`T7q`O0nFBK9=wb5sIb2fU`c=DY>a+wTVba)bsl!
zzA<fJRq*|fNjUo(KIZ{0+V$Nw;pd({KG!RF!L2soe?t6XnJ1cbMmYd}+Bno(r60ha
zcoq{5K8^EuRJOFM6`i4}Hiapj9|CSkf=JIpfKv{}`9uW@=M_c2$7&OhDQG`@jz&c~
z8u*5t549#9o(Dns*!J6mf1DEBMFK8y%)gv}W4Z;p#5d#_83q5}0k@b5sP|t3UXVdQ
zKO*QrJ}K?}7))kllJi0VhdxiKw^ssAIS<U1eDW*=K6fa%aSoCp|2>M1C8hox3BhJc
zw-|8Jw}fqiF86}-;W~cGb0CHPvmvNc(T@)(eX?AlYnKB~{Eig<zb)`lzDlne_57!R
zGu@7jHbG&(qLT$CNvGgF2{-D$6>#EPmr3{w3jY>`Z+lR}Z&wxc357rSBME;}5jy2d
zl8$k1*FO~ee85>BjPtanDEeN&K`QN81NhP7j!zqB-7pINJ;aCHw3!}3t3ai@eH8pZ
zjROCWbdc`-Y97X5Y$6z@-1fd<6Ckl#6>l>Yyh+K^ry^7-_|Q&?KSRNn0Zus&eOtow
z6`wl+r+loc919ixh@zuCF7dBY@DnjVOMYyxOSsreh%h_gq|>kH9H;OvQ}_dlAC;{3
zb->Bbz^^17qn`g%;A>jbaT2b`Xpbs<<Ge5f|B!<3T`Tbyssy^GO8T}s376wTY$pJm
z{MX$l;i}nbuL-!3TEaaS8Oc;n8UYvlJ0zW_RFo?eyhq6m>Owqg6nyZzHsP;Q?eRAX
z-v1K`?@;{bpdd)U;9?0ETPF~+4R9fUCC_?64I6|27xGtnuR_7^FzCE1>C8~{w*k(2
z+pp|)L*Kpzc&7Xw5^$7zirfm1r=CpMQ7Tz&8sM4eoC`ShjHTQoP?xI+;Zg8!0-SQt
zE|mPAs`%fc;DhH%_`8bENob0h(k%hpqHxSk)hPIv0#11jrO5MIfe-z*&nC!lZ!4ew
zRPfCy_53No$$!sxYywZ5wScz{OxcKXRPEQO&kq7ld~3bL=UFIx&c+8l;t%ez2|xE5
z^9cY>{K22ugx@$r;10kk=bjWfzpUu&SLMjP=X}0&vf}f<Y{D=0{Nt?_aF*|0HQp3~
z_^DU;7Ozc!ct#1IQ!wF5ehRL$3I7GEylw%U{Oncy^oUT{SAa8LdzBs#d-f6bwxYB5
zn>K+d&IrMq2ZnT}^6dhg>1scf_~Hx(;QxIT{F5ybzhAXeW1Mv<;F;)OrSSL9w`qLD
zSwnE=oGRPVfNDp^MDudMGts#gaPryVwvmi>yORIwicXWFV-cY!m(w!SnGQJVY_{11
zrr5}Zx5q}o|BJ#m&Uu`o=%4ZxNvG)sNrz{M@VN_cQ7*69=+(qtbGW+!r(Kjf5BY+?
z*ECz9jp*^`Lq*>>XVQ?#+-Wjh>oqok`{QgGWxs;&y;Z`EGd<2ZUGg)S(!Q1mxQuK5
zy+q-s+Ko2=F66&h($5+vV?8?xowr7TA904{r|z$cj!JxvfGdy5ag73BF$(<JQQ$WU
zc)A{V4shnnIM)-yBJoT-Q}WrQ+Ix$FmkPM@n4Cp`lg{3#O+c?#@)=Te22PN0o)yOD
zIRVcw|1XZzR0l<hQx>bLk``20H7rp{g5Vg=R7iO`Xihq)A{|tj1}aO3q_a?#&O=#-
zh_ljZrL#3Noz~3s{LD;eJ{_q%9hA;Oh2~yTbHO5ejoTgST;X=PQ%k^oNkh1%!tM44
z#BvxBY@B(TD*4%IIjqb`-iYtML>>@ZQ#xDFHSRyak*E$W5jZC)kS8BIKIvr9F*0R3
zf0V0hjDtPLIO;PKF>}=TnV{ZaI26Wd-6dYl?LM!n-d*ErXmq<ZoK}isHc_pE?r@vi
z+lnJZLpaDi4diw=2i&V#11mhO?q=MI5ORAuyLn=~I4-lfWOn78GEmNd<mUP$xwgoS
z^+Lhbnp~gM+}YN)8c|F>abj%}s@uJ|-d^i)JC;_1S;@J2S9Rlp#fuvpjqXPKq8bM=
zSb%X7&B=feZy3<r)&I8CUhArg7h`EfRiI<Fy93L{e7FHbTwUQ>+Tfm}PJ_mKsapc%
zwFrE#jNez`sPWHq7*AI{BdT+tU5##sqrvX>^U=6=%B@&XwX1Y?{AASzc^++bnWGDg
zHH6EoZa2=_4P51J^|h}Gx43=5U?5m`zSm}3QsY`wwS2jIW=VNG52GB^TsmusD6@vn
zwq@d85=ULLpOG5Ev*qm`E?Li0^J+!zRc0H)b7VPpyH=X=+Td|D%t8K2=cHW^;%L)4
zf>;%|Qt|38D=90P^+~+WS%k%CeqT@s01TE^U?ra43z31=D*XC9Z7wh1r94*Hjnn4+
z;gHMg03%+9qpH+Ve||#<!86^iK*)Uoj*f4`(Z6ZpY6!H39R6f5jt|Dk_Nm-9D%qqR
z((Jw@9+Y(6wcCx0WxQ^0s8dvWZ3WJh=Arlo6NVg&O3>Kw38zpap=HR?EVCeR=KknX
zNX{1PQR!@>9<A^+yS<)JxGNc{BM|a;yCDH_@ky!-3<l%%w92HI%F2>Y07jTl2aftg
z_41>5{0{2OL~b(@1B!c9TcBMw5w`a8y4>|j6IHc(LLq3Tj?Qo>mBq$J<nplUO0odl
zmo&H5r3l5>?N6-^ZgLO|yH|Srt*I<5J-;D5%ggHuP!&-DYu#7*n|<v<899<kmz0)%
zQbQ;$Yv5f;zGmL5fx7Ci_d@~2W1&sUW41l5t-jX!rAx|J`deF56js@%Q&^B(WjoL1
zhgL@~5cIF;3}ex(s~P$%U6s{0S_K=>PGxRh0s&p?N42>cS3!^mce|@*g%v7cmJw=z
z>k<_rP2ZH0)tR44-`JsV+8V0x+Xa<xb=RNY2*IrK;U=dxTxQhlS)EccVr{I^U4MEV
zxe_|k?d|rsgFc~yZ8qpztn77{mds>1G=wYs&@o&oT<(GppnQ@`IbK+E)GZPz#d8?_
z$@z_LQ*&Ahom)CH=_U*+z=lPTNj%Z?Zl!dlaYVT|vz#hc>MEg%+-)9zdy3?Y9;#Aa
z&ZBxq+!BHo{@F6qFx!p3v2-?$nFkH(3V9uo3jyptyXwzR?K@2wM`?@Lsn;{P8WuA)
zI@o&F1!_!L!%SBjd_n~%E3Gq5TIXG4Jj<Ld&*3*EpHO3xdTAvaVO3l61s!ZKj&8J`
zkfW~7i<ZiMC|(u<Pbz{se;p9wA*0+qP&&taXPM9h%$@4B%gWL0c#{T5hLhWDo5S1c
zqTzs+?8S9w?iIMf>PpyLI7Z#afY~sk#T_lsPB18d6I^Wsx7D<jS2$|gX5p{2%z@vT
z6_tE1Wzn?NqY|QYtnN~4hX>2c-PCo|*lZ*otV2&%mtqw@5`kKPb=ii+^<iH;?<rks
z*-Ueut8u(L(J5CW-^uN*I=#KQ8hJ^XVYxQeR?(Dg6f4~m-SsFMz|?Y*?_;Pc%g~iI
zl)Ld?bGt4;SJmuqcXwjHKs;xoB$1I)!a#6iH^2%QCaI0k=rc`~d1cU-YELllnRD*W
zV04bcVQFP@XD*K4&yfAPE`P%;n6%7sL#P~GbVsRGn1JX%0;_3-(I8Vf`E+J#W!y@2
z)L&r|CEHlq<s2DBX%vc~wvx+ntTrX?K^aXZ<%@y=PcyBiivz*dW``fS7a8}@#`P0{
zpf5%5o2=XnN_ckaMQ16B^wW0Qb4F97w8t|lQmV&97d*NyT|xr&hAJJ?_$w_>t}hN&
zj8a)Kpo38-pd7kAEBsxh7$RV33HwQ?4QQ^F0o0G=E_JM8Fjo(@G(*LwXc5tj{}l?*
z+bYK-WAYfUDe+1o&`Zmz#9eA`UMHApr5%PDklHnhVP(9j#BD<3>J@fH<|0@97h0~U
zzTt|H4@O(a_qB>uzII;_y4kg~5(aoX?)vKPW`QZQ5Z7KJG!S&vU6H6jSD=Ys!Kx+<
zv4#pkRAe`<sAdOT?ajXK3<`t=s0KdC*9AeLpyw2WO%GaD`dHTuwQfuWQ2D^@0@5f7
zry*SK#dR^sdI*=s;EpnMEUAjNAzT^nrBY2fcXKG<#x0V~*fnLC66vkGl9LE^oI?=Z
zH5FuHI7mx59(N8p4BW}nY^qeV#ARM7>fkVH$EUIpX+2Z;YFImp{gLPl%gV&;RtU2e
z;;tPfgCHt+B_M3Z?r<HaE0j%?Hk&bvqeYHjM0XJ{^<?FoNU&7Y9{1Ne{l1XbhZ@-I
z3z>wJ(%ehS%|?R{%30YX9i2g+yUQO8cX}{V%b;db36+@%ITF>}Wb`F~GOuIIm(Wbg
zI6EZlr`hQStT2BwPoo&3jW?^A<+7`!q1qTovRiAYU<l99=FmN&naoVI(^An|)x}S>
znADWBPMj?W$f*{nX^wX@%{4HA(wTBnLSbX73%!a1rhFaKF-FL;`!A`~Fu%|i=t?zQ
zRK<zsPR)K{d}Yi|sUk1}q)JU205?FHvGG){^mqH3>7!<X(aO4s!J47$s>(i5%lXsN
zrj~1To0Z(>qtjD0dVNEsTTT|nTU427DX?pC9hTAh<XD2mWi;ft_(V%e<KNRjo6J0c
zV4Ae5;3Ok1YYWM^{&L=RoMP0(*^#}Bsx+MAWUX;yH)vv7vmr2RWou`sC7ncyO~Coh
zkpC0S0@FyFbs$-ri^5H*DPy;z*|IbKtco^MkwLY!238pr-k9QZtBcaySNc}Fy^ydR
z{51H3A=p=`*)R;6PgKe2-3}&;yeJ7@(A4$B3#?3(G-hf>RjF3a+sJDUIYgXgj4r`(
z+LX}0!n9&DO`2flNE4bQk%i1R5SEp4h|ZyUk`BzE`r?#I%lz#yb%5X8=?#C@Dqv!^
z8SN^yMl@6>SWho38Z3yBXs|HLF?E<yg)zlLrVn$(j$u(7qNO&oQn$OS?UUP?P^_}U
zk56j2+QaIjL48BWG<-01i!xf~OBBbl0C|VF(YK|G3Dq{`MrdA5m{MZ@HwR0f_k16!
zOXdkbPJ(hMn(~1F-$+LJKrqUPJV~ch%xcO-l<enfmX4Yg&RzO7f2gCg-D}wEh8|@z
zFq+4|LjQ2?bSBatwcBV2RAO>N*+MWY>Kg)$?j-#U>xLg%U^!v7x!J`zzz_|FGXBR5
zwbjj*9K`qmH)DIeywcHajH+-J!zyL9qDf%3EbMK`Y_g`=)}>|U`|bbN8iz>_e9|<6
zoR2jo5yV&`{zXh(9ZjnPwj_p|@>?grQDMd~;7#x8rRir)e5y^!@?ngY(v-IRL`#jN
zX0%|PG`M}Z?Q}IJ);eI`1cIJbzLX|}F_1vX@aa%yx`M`31Iqfdv8p~hi;O|rnR*`c
zcUWQQ@>57^u1qBr*Vtmdb_Fli9HXv{VugItB&{r<G~4`t@^NZxdxSYp%e2{cNbgx@
zi9wPYkov@FCYbc-?&fFzPypw@_)+3r){ISX6tQ41(Bku6=?g9jVA4JF=uvi)s#jB=
zQvOFDh7R{z3WjDT&gzynX8efcq#}-gNI851#C1g%6r>z=$=V>tFVKtPE1?)<wqMnj
zq(0f|q<k`4di=1T=mX+4Id$lW(h(~i<iME5p_+f<q@?^P6YogkL@sK-3Mv~@*rCog
zj!m&#Yi5a6=z;KTF=6M+@Tp;Je$A7WE49$qf*HEBz)XC|cBC)mfc6Pp<>^3%jV}Re
z{v5++Z-?pQW@%#OM|vfPCHU#BOeOk^GwqpXSd7ezkwx58tn;jH#YDOJ0~iKHs+`Sz
Y`DmY;|7D-9nn(F`#ZR7^7)xvaAM|vw`Tzg`


From 2843264bd859404c09d8415b438ffb8b53c0ed02 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 18 May 2015 12:09:25 +0100
Subject: [PATCH 195/429] Remez tested

---
 lib/algorithms/approx/Remez.cc          |   6 +-
 lib/algorithms/approx/bigfloat.h        | 164 +-----------------------
 lib/algorithms/approx/bigfloat_double.h | 162 +++++++++++++++++++++++
 lib/parallelIO/GridNerscIO.h            |   2 +-
 scripts/configure-commands              |  28 ++--
 tests/Grid_remez.cc                     | 103 +++++++++++++++
 tests/Makefile.am                       |   5 +-
 7 files changed, 290 insertions(+), 180 deletions(-)
 create mode 100644 lib/algorithms/approx/bigfloat_double.h
 create mode 100644 tests/Grid_remez.cc

diff --git a/lib/algorithms/approx/Remez.cc b/lib/algorithms/approx/Remez.cc
index a787e766..4ec2270d 100755
--- a/lib/algorithms/approx/Remez.cc
+++ b/lib/algorithms/approx/Remez.cc
@@ -116,7 +116,7 @@ double AlgRemez::generateApprox(int num_degree, int den_degree,
 				int a_len, double *a_param, int *a_pow)
 {
   std::cout<<"Degree of the approximation is ("<<num_degree<<","<<den_degree<<")\n";
-  std::cout<<"Approximating the function x^("<<pnum<<"/"<<pden<<"\n";
+  std::cout<<"Approximating the function x^("<<pnum<<"/"<<pden<<")\n";
 
   // Reallocate arrays, since degree has changed
   if (num_degree != n || den_degree != d) allocate(num_degree,den_degree);
@@ -149,6 +149,10 @@ double AlgRemez::generateApprox(int num_degree, int den_degree,
       std::cout<<"Iteration " <<iter-1<<" spread "<<(double)spread<<" delta "<<(double)delta<<std::endl; 
 
     equations();
+    if (delta < tolerance) {
+      std::cout<<"Delta too small, try increasing precision\n";
+      assert(0);
+    };    
     assert( delta>= tolerance);
 
     search(step);
diff --git a/lib/algorithms/approx/bigfloat.h b/lib/algorithms/approx/bigfloat.h
index 7f59ea33..aee71d03 100755
--- a/lib/algorithms/approx/bigfloat.h
+++ b/lib/algorithms/approx/bigfloat.h
@@ -10,7 +10,7 @@
 #ifndef INCLUDED_BIGFLOAT_H
 #define INCLUDED_BIGFLOAT_H
 
-#ifdef HAVE_GMP
+
 #include <gmp.h>
 #include <mpf2mpfr.h>
 #include <mpfr.h>
@@ -202,167 +202,5 @@ public:
   friend int sgn(const bigfloat&);
 
 };
-#else
-
-typedef double mfloat; 
-class bigfloat {
-private:
-
-  mfloat x;
-
-public:
-
-  bigfloat() { }
-  bigfloat(const bigfloat& y) { x=y.x; }
-  bigfloat(const unsigned long u) { x=u; }
-  bigfloat(const long i) { x=i; }
-  bigfloat(const int i) { x=i;}
-  bigfloat(const float d) { x=d;}
-  bigfloat(const double d) {  x=d;}
-  bigfloat(const char *str) { x=std::stod(std::string(str));}
-  ~bigfloat(void) { }
-  operator double (void) const { return (double)x; }
-  static void setDefaultPrecision(unsigned long dprec) {
-  }
-
-  void setPrecision(unsigned long dprec) {
-  }
-  
-  unsigned long getPrecision(void) const { return 64; }
-  unsigned long getDefaultPrecision(void) const { return 64; }
-
-  bigfloat& operator=(const bigfloat& y)     { x=y.x;    return *this;  }
-  bigfloat& operator=(const unsigned long y) { x=y; return *this; }
-  bigfloat& operator=(const signed long y)   { x=y; return *this; }
-  bigfloat& operator=(const float y)    { x=y; return *this; }
-  bigfloat& operator=(const double y)   { x=y; return *this; }
-
-  size_t write(void);
-  size_t read(void);
-
-  /* Arithmetic Functions */
-
-  bigfloat& operator+=(const bigfloat& y) { return *this = *this + y; }
-  bigfloat& operator-=(const bigfloat& y) { return *this = *this - y; }
-  bigfloat& operator*=(const bigfloat& y) { return *this = *this * y; }
-  bigfloat& operator/=(const bigfloat& y) { return *this = *this / y; }
-
-  friend bigfloat operator+(const bigfloat& x, const bigfloat& y) { 
-    bigfloat a;
-    a.x=x.x+y.x;
-    return a;
-  }
-
-  friend bigfloat operator+(const bigfloat& x, const unsigned long y) {
-    bigfloat a;
-    a.x=x.x+y;
-    return a;
-  }
-
-  friend bigfloat operator-(const bigfloat& x, const bigfloat& y) {
-    bigfloat a;
-    a.x=x.x-y.x;
-    return a;
-  }
-  
-  friend bigfloat operator-(const unsigned long x, const bigfloat& y) {
-    bigfloat bx(x);
-    return bx-y;
-  }
-  
-  friend bigfloat operator-(const bigfloat& x, const unsigned long y) {
-    bigfloat by(y);
-    return x-by;
-  }
-
-  friend bigfloat operator-(const bigfloat& x) {
-    bigfloat a;
-    a.x=-x.x;
-    return a;
-  }
-
-  friend bigfloat operator*(const bigfloat& x, const bigfloat& y) {
-    bigfloat a;
-    a.x=x.x*y.x;
-    return a;
-  }
-
-  friend bigfloat operator*(const bigfloat& x, const unsigned long y) {
-    bigfloat a;
-    a.x=x.x*y;
-    return a;
-  }
-
-  friend bigfloat operator/(const bigfloat& x, const bigfloat& y){
-    bigfloat a;
-    a.x=x.x/y.x;
-    return a;
-  }
-
-  friend bigfloat operator/(const unsigned long x, const bigfloat& y){
-    bigfloat bx(x);
-    return bx/y;
-  }
-
-  friend bigfloat operator/(const bigfloat& x, const unsigned long y){
-    bigfloat by(y);
-    return x/by;
-  }
-
-  friend bigfloat sqrt_bf(const bigfloat& x){
-    bigfloat a;
-    a.x= sqrt(x.x);
-    return a;
-  }
-
-  friend bigfloat sqrt_bf(const unsigned long x){
-    bigfloat a(x);
-    return sqrt_bf(a);
-  }
-
-  friend bigfloat abs_bf(const bigfloat& x){
-    bigfloat a;
-    a.x=abs(x.x);
-    return a;
-  }
-
-  friend bigfloat pow_bf(const bigfloat& a, long power) {
-    bigfloat b;
-    b.x=pow(a.x,power);
-    return b;
-  }
-
-  friend bigfloat pow_bf(const bigfloat& a, bigfloat &power) {
-    bigfloat b;
-    b.x=pow(a.x,power.x);
-    return b;
-  }
-
-  friend bigfloat exp_bf(const bigfloat& a) {
-    bigfloat b;
-    b.x=exp(a.x);
-    return b;
-  }
-
-  /* Comparison Functions */
-  friend int operator>(const bigfloat& x, const bigfloat& y) {
-    return x.x>y.x;
-  }
-
-  friend int operator<(const bigfloat& x, const bigfloat& y) {
-    return x.x<y.x;
-  }
-
-  friend int sgn(const bigfloat& x) {
-    if ( x.x>=0 )  return 1;   
-    else return 0;
-  }
-
-  /* Miscellaneous Functions */
-
-  //  friend bigfloat& random(void);
-};
-
-#endif
 
 #endif
diff --git a/lib/algorithms/approx/bigfloat_double.h b/lib/algorithms/approx/bigfloat_double.h
new file mode 100644
index 00000000..d928b4a2
--- /dev/null
+++ b/lib/algorithms/approx/bigfloat_double.h
@@ -0,0 +1,162 @@
+#include <math.h>
+
+typedef double mfloat; 
+class bigfloat {
+private:
+
+  mfloat x;
+
+public:
+
+  bigfloat() { }
+  bigfloat(const bigfloat& y) { x=y.x; }
+  bigfloat(const unsigned long u) { x=u; }
+  bigfloat(const long i) { x=i; }
+  bigfloat(const int i) { x=i;}
+  bigfloat(const float d) { x=d;}
+  bigfloat(const double d) {  x=d;}
+  bigfloat(const char *str) { x=std::stod(std::string(str));}
+  ~bigfloat(void) { }
+  operator double (void) const { return (double)x; }
+  static void setDefaultPrecision(unsigned long dprec) {
+  }
+
+  void setPrecision(unsigned long dprec) {
+  }
+  
+  unsigned long getPrecision(void) const { return 64; }
+  unsigned long getDefaultPrecision(void) const { return 64; }
+
+  bigfloat& operator=(const bigfloat& y)     { x=y.x;    return *this;  }
+  bigfloat& operator=(const unsigned long y) { x=y; return *this; }
+  bigfloat& operator=(const signed long y)   { x=y; return *this; }
+  bigfloat& operator=(const float y)    { x=y; return *this; }
+  bigfloat& operator=(const double y)   { x=y; return *this; }
+
+  size_t write(void);
+  size_t read(void);
+
+  /* Arithmetic Functions */
+
+  bigfloat& operator+=(const bigfloat& y) { return *this = *this + y; }
+  bigfloat& operator-=(const bigfloat& y) { return *this = *this - y; }
+  bigfloat& operator*=(const bigfloat& y) { return *this = *this * y; }
+  bigfloat& operator/=(const bigfloat& y) { return *this = *this / y; }
+
+  friend bigfloat operator+(const bigfloat& x, const bigfloat& y) { 
+    bigfloat a;
+    a.x=x.x+y.x;
+    return a;
+  }
+
+  friend bigfloat operator+(const bigfloat& x, const unsigned long y) {
+    bigfloat a;
+    a.x=x.x+y;
+    return a;
+  }
+
+  friend bigfloat operator-(const bigfloat& x, const bigfloat& y) {
+    bigfloat a;
+    a.x=x.x-y.x;
+    return a;
+  }
+  
+  friend bigfloat operator-(const unsigned long x, const bigfloat& y) {
+    bigfloat bx(x);
+    return bx-y;
+  }
+  
+  friend bigfloat operator-(const bigfloat& x, const unsigned long y) {
+    bigfloat by(y);
+    return x-by;
+  }
+
+  friend bigfloat operator-(const bigfloat& x) {
+    bigfloat a;
+    a.x=-x.x;
+    return a;
+  }
+
+  friend bigfloat operator*(const bigfloat& x, const bigfloat& y) {
+    bigfloat a;
+    a.x=x.x*y.x;
+    return a;
+  }
+
+  friend bigfloat operator*(const bigfloat& x, const unsigned long y) {
+    bigfloat a;
+    a.x=x.x*y;
+    return a;
+  }
+
+  friend bigfloat operator/(const bigfloat& x, const bigfloat& y){
+    bigfloat a;
+    a.x=x.x/y.x;
+    return a;
+  }
+
+  friend bigfloat operator/(const unsigned long x, const bigfloat& y){
+    bigfloat bx(x);
+    return bx/y;
+  }
+
+  friend bigfloat operator/(const bigfloat& x, const unsigned long y){
+    bigfloat by(y);
+    return x/by;
+  }
+
+  friend bigfloat sqrt_bf(const bigfloat& x){
+    bigfloat a;
+    a.x= sqrt(x.x);
+    return a;
+  }
+
+  friend bigfloat sqrt_bf(const unsigned long x){
+    bigfloat a(x);
+    return sqrt_bf(a);
+  }
+
+  friend bigfloat abs_bf(const bigfloat& x){
+    bigfloat a;
+    a.x=fabs(x.x);
+    return a;
+  }
+
+  friend bigfloat pow_bf(const bigfloat& a, long power) {
+    bigfloat b;
+    b.x=pow(a.x,power);
+    return b;
+  }
+
+  friend bigfloat pow_bf(const bigfloat& a, bigfloat &power) {
+    bigfloat b;
+    b.x=pow(a.x,power.x);
+    return b;
+  }
+
+  friend bigfloat exp_bf(const bigfloat& a) {
+    bigfloat b;
+    b.x=exp(a.x);
+    return b;
+  }
+
+  /* Comparison Functions */
+  friend int operator>(const bigfloat& x, const bigfloat& y) {
+    return x.x>y.x;
+  }
+
+  friend int operator<(const bigfloat& x, const bigfloat& y) {
+    return x.x<y.x;
+  }
+
+  friend int sgn(const bigfloat& x) {
+    if ( x.x>=0 )  return 1;   
+    else return 0;
+  }
+
+  /* Miscellaneous Functions */
+
+  //  friend bigfloat& random(void);
+};
+
+
diff --git a/lib/parallelIO/GridNerscIO.h b/lib/parallelIO/GridNerscIO.h
index 4094ab70..8a1775f1 100644
--- a/lib/parallelIO/GridNerscIO.h
+++ b/lib/parallelIO/GridNerscIO.h
@@ -481,7 +481,7 @@ inline void writeNerscObject(Lattice<vobj> &Umu,std::string file,munger munge,in
   // Now write the body
   //////////////////////////////////////////////////
   {
-    std::ofstream fout(file,std::ios::binary|std::ios::in);
+    std::ofstream fout(file,std::ios::binary|std::ios::out);
     fout.seekp(offset);
 
     Umu = zero;
diff --git a/scripts/configure-commands b/scripts/configure-commands
index 5e83fa89..89b72294 100755
--- a/scripts/configure-commands
+++ b/scripts/configure-commands
@@ -17,46 +17,46 @@ echo -e $YELLOW
 
 case $WD in
 g++-sse4)
-  CXX=g++-5 ../../configure --enable-simd=SSE4 CXXFLAGS="-msse4 -O3 -std=c++11" --enable-comms=none
+  CXX=g++-5 ../../configure --enable-simd=SSE4 CXXFLAGS="-msse4 -O3 -std=c++11" LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;
 g++-avx)
-  CXX=g++-5 ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11" --enable-comms=none
+  CXX=g++-5 ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11" LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;
 icpc-avx)
-  CXX=icpc ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11" --enable-comms=none
+  CXX=icpc ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11" LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;
 icpc-avx2)
-  CXX=icpc ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -std=c++11" --enable-comms=none
+  CXX=icpc ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -std=c++11" LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;
 icpc-avx512)
-  CXX=icpc ../../configure --enable-simd=AVX512 CXXFLAGS="-xCOMMON-AVX512 -O3  -std=c++11" --host=none  --enable-comms=none
+  CXX=icpc ../../configure --enable-simd=AVX512 CXXFLAGS="-xCOMMON-AVX512 -O3  -std=c++11" --host=none  LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;
 icpc-mic)
-  CXX=icpc ../../configure --host=none --enable-simd=AVX512 CXXFLAGS="-mmic -O3  -std=c++11" LDFLAGS=-mmic --enable-comms=none
+  CXX=icpc ../../configure --host=none --enable-simd=AVX512 CXXFLAGS="-mmic -O3  -std=c++11" LDFLAGS=-mmic LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;
 clang-avx)
-CXX=clang++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11"  --enable-comms=none
+CXX=clang++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11"  LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;
 clang-avx2)
-CXX=clang++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -std=c++11"  --enable-comms=none
+CXX=clang++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -std=c++11"  LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;
 clang-avx-openmp)
-CXX=clang-omp++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -std=c++11" LDFLAGS="-fopenmp"  --enable-comms=none
+CXX=clang-omp++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -std=c++11" LDFLAGS="-fopenmp"  LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;
 clang-avx2-openmp)
-CXX=clang-omp++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -fopenmp -O3 -std=c++11" LDFLAGS="-fopenmp"  --enable-comms=none
+CXX=clang-omp++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -fopenmp -O3 -std=c++11" LDFLAGS="-fopenmp"  LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;
 clang-avx-openmp-mpi)
-CXX=clang-omp++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp"  --enable-comms=mpi
+CXX=clang-omp++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp"  LIBS="-lgmp -lmpfr" --enable-comms=mpi
 ;;
 clang-avx2-openmp-mpi)
-CXX=clang-omp++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp"  --enable-comms=mpi
+CXX=clang-omp++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp"  LIBS="-lgmp -lmpfr" --enable-comms=mpi
 ;;
 clang-avx-mpi)
-CXX=clang++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx "  --enable-comms=mpi
+CXX=clang++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx "  LIBS="-lgmp -lmpfr" --enable-comms=mpi
 ;;
 clang-avx2-mpi)
-CXX=clang++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx "  --enable-comms=mpi
+CXX=clang++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx "  LIBS="-lgmp -lmpfr" --enable-comms=mpi
 ;;
 esac
 echo -e $NORMAL
diff --git a/tests/Grid_remez.cc b/tests/Grid_remez.cc
new file mode 100644
index 00000000..b5993917
--- /dev/null
+++ b/tests/Grid_remez.cc
@@ -0,0 +1,103 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+class MultiShiftFunction {
+public:
+  std::vector<double> poles;
+  std::vector<double> residues;
+  double norm;
+  double lo,hi;
+  MultiShiftFunction(int n,double _lo,double _hi): poles(n), residues(n), lo(_lo), hi(_hi) {;};
+  double approx(double x);
+  void csv(std::ostream &out);
+  void gnuplot(std::ostream &out);
+};
+double MultiShiftFunction::approx(double x)
+{
+  double a = norm;
+  for(int n=0;n<poles.size();n++){
+    a = a + residues[n]/(x+poles[n]);
+  }
+  return a;
+}
+void MultiShiftFunction::gnuplot(std::ostream &out)
+{
+  out<<"f(x) = "<<norm<<"";
+  for(int n=0;n<poles.size();n++){
+    out<<"+("<<residues[n]<<"/(x+"<<poles[n]<<"))";
+  }
+  out<<";"<<std::endl;
+}
+void MultiShiftFunction::csv(std::ostream &out)
+{
+  for (double x=lo; x<hi; x*=1.05) {
+    double f = approx(x);
+    double r = sqrt(x);
+    out<< x<<","<<r<<","<<f<<","<<r-f<<std::endl;
+  }
+  return;
+}
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::cout << "Testing Remez"<<std::endl;
+
+  double     lo=0.01;
+  double     hi=1.0;
+  int precision=64;
+  int    degree=10;
+  AlgRemez remez(0.001,1.0,precision);
+
+  ////////////////////////////////////////
+  // sqrt and inverse sqrt
+  ////////////////////////////////////////
+  MultiShiftFunction Sqrt(degree,lo,hi);
+  MultiShiftFunction InvSqrt(degree,lo,hi);
+
+  MultiShiftFunction SqrtSqrt(degree,lo,hi);
+  MultiShiftFunction InvSqrtSqrt(degree,lo,hi);
+
+
+  std::cout << "Generating degree "<<degree<<" for x^(1/2)"<<std::endl;
+  remez.generateApprox(degree,1,2);
+  remez.getPFE (&   Sqrt.residues[0],&   Sqrt.poles[0],&   Sqrt.norm);
+  remez.getIPFE(&InvSqrt.residues[0],&InvSqrt.poles[0],&InvSqrt.norm);
+
+  std::cout << "Generating degree "<<degree<<" for x^(1/4)"<<std::endl;
+  remez.generateApprox(degree,1,4);
+  remez.getPFE (&SqrtSqrt.residues[0],&SqrtSqrt.poles[0],&SqrtSqrt.norm);
+  remez.getIPFE(&InvSqrtSqrt.residues[0],&InvSqrtSqrt.poles[0],&InvSqrtSqrt.norm);
+  
+  ofstream gnuplot(std::string("Sqrt.gnu"),std::ios::out|std::ios::trunc);
+  Sqrt.gnuplot(gnuplot);
+
+  ofstream gnuplot_inv(std::string("InvSqrt.gnu"),std::ios::out|std::ios::trunc);
+  InvSqrt.gnuplot(gnuplot);
+
+  double x=0.6789;
+  double sx=sqrt(x);
+  double ssx=sqrt(sx);
+  double isx=1.0/sx;
+  double issx=1.0/ssx;
+
+  double asx  =Sqrt.approx(x);
+  double assx =SqrtSqrt.approx(x);
+  double aisx =InvSqrt.approx(x);
+  double aissx=InvSqrtSqrt.approx(x);
+
+  std::cout << "x^(1/2) : "<<sx<<" "<<asx<<std::endl;
+  std::cout << "x^(1/4) : "<<ssx<<" "<<assx<<std::endl;
+  std::cout << "x^(-1/2): "<<isx<<" "<<aisx<<std::endl;
+  std::cout << "x^(-1/4): "<<issx<<" "<<aissx<<std::endl;
+  assert(fabs(sx-asx)<1.0e-6);
+  assert(fabs(ssx-assx)<1.0e-6);
+  assert(fabs(isx-aisx)<1.0e-6);
+  assert(fabs(issx-aissx)<1.0e-6);
+
+  Grid_finalize();
+}
diff --git a/tests/Makefile.am b/tests/Makefile.am
index 655bb9c5..82d1b3be 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -5,7 +5,7 @@ AM_LDFLAGS = -L$(top_builddir)/lib
 #
 # Test code
 #
-bin_PROGRAMS = Grid_main Grid_stencil Grid_nersc_io Grid_cshift Grid_gamma  Grid_simd Grid_rng
+bin_PROGRAMS = Grid_main Grid_stencil Grid_nersc_io Grid_cshift Grid_gamma  Grid_simd Grid_rng Grid_remez
 
 Grid_main_SOURCES = Grid_main.cc
 Grid_main_LDADD = -lGrid
@@ -13,6 +13,9 @@ Grid_main_LDADD = -lGrid
 Grid_rng_SOURCES = Grid_rng.cc
 Grid_rng_LDADD = -lGrid
 
+Grid_remez_SOURCES = Grid_remez.cc
+Grid_remez_LDADD = -lGrid
+
 Grid_nersc_io_SOURCES = Grid_nersc_io.cc
 Grid_nersc_io_LDADD = -lGrid
 

From a19deba26cb04dcb6f8cb54343ce3ae258500dbb Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 18 May 2015 16:28:29 +0100
Subject: [PATCH 196/429] Convience function

---
 scripts/linecount | 3 +++
 1 file changed, 3 insertions(+)
 create mode 100755 scripts/linecount

diff --git a/scripts/linecount b/scripts/linecount
new file mode 100755
index 00000000..c0731b23
--- /dev/null
+++ b/scripts/linecount
@@ -0,0 +1,3 @@
+#!/bin/sh 
+
+wc -l */*.h */*/*.h */*/*/*.h */*.cc */*/*.cc */*/*/*.cc
\ No newline at end of file

From f9a8377fe6bfebf50cd107d3123b9f27fb4e2ecc Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 18 May 2015 16:35:48 +0100
Subject: [PATCH 197/429] lib/algorithms/approx/bigfloat.h

---
 scripts/cleanup    | 3 +++
 scripts/substitute | 6 ++++++
 2 files changed, 9 insertions(+)
 create mode 100755 scripts/cleanup
 create mode 100755 scripts/substitute

diff --git a/scripts/cleanup b/scripts/cleanup
new file mode 100755
index 00000000..956a6884
--- /dev/null
+++ b/scripts/cleanup
@@ -0,0 +1,3 @@
+find . -name "*~" -print -exec rm -f {} \;
+find . -name "errs" -print -exec rm -f {} \;
+find . -name "log" -print -exec rm -f {} \;
diff --git a/scripts/substitute b/scripts/substitute
new file mode 100755
index 00000000..3066c4ef
--- /dev/null
+++ b/scripts/substitute
@@ -0,0 +1,6 @@
+#!/bin/bash
+
+export LANG=C
+find . -name "*.cc" -print -exec sed -e "s/${1}/${2}/g" -i .bak {} \;
+find . -name "*.h" -print -exec sed -e "s/${1}/${2}/g" -i .bak {} \;
+find . -name "*.bak" -exec rm -f {} \;

From b5af3fbe45c7270baec80d4309362edcdc84237d Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Tue, 19 May 2015 13:36:03 +0900
Subject: [PATCH 198/429] Merging with upstream

---
 README.md                    | 2 ++
 lib/simd/Grid_vector_types.h | 6 ------
 2 files changed, 2 insertions(+), 6 deletions(-)

diff --git a/README.md b/README.md
index 80714d76..e18ca474 100644
--- a/README.md
+++ b/README.md
@@ -37,6 +37,8 @@ MPI, OpenMP, and SIMD parallelism are present in the library.
 by setting variables in the command line or in the environment.  Here
 are examples:
 
+     ./configure CXX=clang++ CXXFLAGS="-std=c++11 -O3 -msse4" --enable-simd=SSE4
+
      ./configure CXX=clang++ CXXFLAGS="-std=c++11 -O3 -mavx" --enable-simd=AVX1
 
      ./configure CXX=clang++ CXXFLAGS="-std=c++11 -O3 -mavx2" --enable-simd=AVX2
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index 14b2ddf0..360e9f1b 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -37,7 +37,6 @@ namespace Grid {
 
   /*
     @brief Grid_simd class for the SIMD vector type operations
-
    */
   template < class Scalar_type, class Vector_type > 
     class Grid_simd {
@@ -100,14 +99,9 @@ namespace Grid {
     template <  class S = Scalar_type,typename std::enable_if < std::is_integral < S >::value, int >::type = 0 > 
     friend inline void vzero(Grid_simd &ret)      { vsplat(ret,0); }
 
-
-    
     // do not compile if real or integer, send an error message from the compiler
     template < class S = Scalar_type,typename std::enable_if < is_complex < S >::value, int >::type = 0 > 
     friend inline void vcomplex_i(Grid_simd &ret){ vsplat(ret,0.0,1.0);}
-       
-
-    
    
     ////////////////////////////////////
     // Arithmetic operator overloads +,-,*

From 4cadf11d1d4f57f28f564d3d779e82353c8c8d38 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Tue, 19 May 2015 13:54:55 +0900
Subject: [PATCH 199/429] Added check of mpfr and gmp at configure time

It generates automatically the linker flags or complains if not found.
---
 README                             |   2 +
 configure                          | 104 +++++++++++++++++++++++++
 configure.ac                       |  16 +++-
 lib/Grid_config.h                  | 121 -----------------------------
 lib/Grid_config.h.in               |   6 ++
 lib/Makefile.am                    |   3 +-
 lib/algorithms/approx/.dirstamp    |   0
 lib/algorithms/approx/Remez.cc     |   0
 lib/algorithms/approx/Remez.h      |   0
 lib/algorithms/approx/Zolotarev.cc |   0
 lib/algorithms/approx/Zolotarev.h  |   0
 lib/algorithms/approx/bigfloat.h   |   0
 lib/communicator/.dirstamp         |   0
 lib/qcd/.dirstamp                  |   0
 lib/simd/Grid_sse4.cpp             |  19 +++++
 lib/simd/Grid_vector_types.h       |   8 +-
 lib/stamp-h1                       |   1 -
 lib/stencil/.dirstamp              |   0
 18 files changed, 155 insertions(+), 125 deletions(-)
 delete mode 100644 lib/Grid_config.h
 delete mode 100644 lib/algorithms/approx/.dirstamp
 mode change 100755 => 100644 lib/algorithms/approx/Remez.cc
 mode change 100755 => 100644 lib/algorithms/approx/Remez.h
 mode change 100755 => 100644 lib/algorithms/approx/Zolotarev.cc
 mode change 100755 => 100644 lib/algorithms/approx/Zolotarev.h
 mode change 100755 => 100644 lib/algorithms/approx/bigfloat.h
 delete mode 100644 lib/communicator/.dirstamp
 delete mode 100644 lib/qcd/.dirstamp
 create mode 100644 lib/simd/Grid_sse4.cpp
 delete mode 100644 lib/stamp-h1
 delete mode 100644 lib/stencil/.dirstamp

diff --git a/README b/README
index 41b66b6b..17e92fa0 100644
--- a/README
+++ b/README
@@ -33,6 +33,8 @@ MPI parallelism is UNIMPLEMENTED and for now only OpenMP and SIMD parallelism is
 by setting variables in the command line or in the environment.  Here
 is are examples:
 
+     ./configure CXX=clang++ CXXFLAGS="-std=c++11 -O3 -msse4" --enable-simd=SSE4
+
      ./configure CXX=clang++ CXXFLAGS="-std=c++11 -O3 -mavx" --enable-simd=AVX1
 
      ./configure CXX=clang++ CXXFLAGS="-std=c++11 -O3 -mavx2" --enable-simd=AVX2
diff --git a/configure b/configure
index e7d9f32f..8c9e8c59 100755
--- a/configure
+++ b/configure
@@ -4502,6 +4502,11 @@ _ACEOF
 
 
 # Checks for library functions.
+echo
+echo Checking libraries
+echo :::::::::::::::::::::::::::::::::::::::::::
+
+
 for ac_func in gettimeofday
 do :
   ac_fn_cxx_check_func "$LINENO" "gettimeofday" "ac_cv_func_gettimeofday"
@@ -4514,6 +4519,105 @@ fi
 done
 
 
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for __gmpf_init in -lgmp" >&5
+$as_echo_n "checking for __gmpf_init in -lgmp... " >&6; }
+if ${ac_cv_lib_gmp___gmpf_init+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_check_lib_save_LIBS=$LIBS
+LIBS="-lgmp  $LIBS"
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char __gmpf_init ();
+int
+main ()
+{
+return __gmpf_init ();
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_link "$LINENO"; then :
+  ac_cv_lib_gmp___gmpf_init=yes
+else
+  ac_cv_lib_gmp___gmpf_init=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+LIBS=$ac_check_lib_save_LIBS
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_gmp___gmpf_init" >&5
+$as_echo "$ac_cv_lib_gmp___gmpf_init" >&6; }
+if test "x$ac_cv_lib_gmp___gmpf_init" = xyes; then :
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_LIBGMP 1
+_ACEOF
+
+  LIBS="-lgmp $LIBS"
+
+else
+  as_fn_error $? "GNU Multiple Precision GMP library was not found in your system.
+Please install or provide the correct path to your installation
+Info at: http://www.gmplib.org" "$LINENO" 5
+fi
+
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for mpfr_init in -lmpfr" >&5
+$as_echo_n "checking for mpfr_init in -lmpfr... " >&6; }
+if ${ac_cv_lib_mpfr_mpfr_init+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_check_lib_save_LIBS=$LIBS
+LIBS="-lmpfr  $LIBS"
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char mpfr_init ();
+int
+main ()
+{
+return mpfr_init ();
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_link "$LINENO"; then :
+  ac_cv_lib_mpfr_mpfr_init=yes
+else
+  ac_cv_lib_mpfr_mpfr_init=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+LIBS=$ac_check_lib_save_LIBS
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_mpfr_mpfr_init" >&5
+$as_echo "$ac_cv_lib_mpfr_mpfr_init" >&6; }
+if test "x$ac_cv_lib_mpfr_mpfr_init" = xyes; then :
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_LIBMPFR 1
+_ACEOF
+
+  LIBS="-lmpfr $LIBS"
+
+else
+  as_fn_error $? "GNU Multiple Precision MPFR library was not found in your system.
+Please install or provide the correct path to your installation
+Info at: http://www.mpfr.org/" "$LINENO" 5
+fi
+
 
 
 
diff --git a/configure.ac b/configure.ac
index 14170f4e..93e2b574 100644
--- a/configure.ac
+++ b/configure.ac
@@ -3,7 +3,7 @@
 #
 # Project Grid package  
 # 
-# Time-stamp: <2015-05-18 17:14:20 neo>
+# Time-stamp: <2015-05-19 13:51:08 neo>
 
 AC_PREREQ([2.69])
 AC_INIT([Grid], [1.0], [paboyle@ph.ed.ac.uk])
@@ -46,8 +46,22 @@ AC_TYPE_UINT32_T
 AC_TYPE_UINT64_T
 
 # Checks for library functions.
+echo
+echo Checking libraries 
+echo :::::::::::::::::::::::::::::::::::::::::::
+
+
 AC_CHECK_FUNCS([gettimeofday])
 
+AC_CHECK_LIB([gmp],[__gmpf_init],,
+        [AC_MSG_ERROR(GNU Multiple Precision GMP library was not found in your system.
+Please install or provide the correct path to your installation
+Info at: http://www.gmplib.org)])
+
+AC_CHECK_LIB([mpfr],[mpfr_init],,
+        [AC_MSG_ERROR(GNU Multiple Precision MPFR library was not found in your system.
+Please install or provide the correct path to your installation
+Info at: http://www.mpfr.org/)])
 
 
 
diff --git a/lib/Grid_config.h b/lib/Grid_config.h
deleted file mode 100644
index e1674850..00000000
--- a/lib/Grid_config.h
+++ /dev/null
@@ -1,121 +0,0 @@
-/* lib/Grid_config.h.  Generated from Grid_config.h.in by configure.  */
-/* lib/Grid_config.h.in.  Generated from configure.ac by autoheader.  */
-
-/* AVX */
-/* #undef AVX1 */
-
-/* AVX2 */
-/* #undef AVX2 */
-
-/* AVX512 */
-/* #undef AVX512 */
-
-/* GRID_COMMS_MPI */
-/* #undef GRID_COMMS_MPI */
-
-/* GRID_COMMS_NONE */
-#define GRID_COMMS_NONE 1
-
-/* Define to 1 if you have the declaration of `be64toh', and to 0 if you
-   don't. */
-#define HAVE_DECL_BE64TOH 1
-
-/* Define to 1 if you have the declaration of `ntohll', and to 0 if you don't.
-   */
-#define HAVE_DECL_NTOHLL 0
-
-/* Define to 1 if you have the <endian.h> header file. */
-#define HAVE_ENDIAN_H 1
-
-/* Define to 1 if you have the `gettimeofday' function. */
-#define HAVE_GETTIMEOFDAY 1
-
-/* Define to 1 if you have the <gmp.h> header file. */
-#define HAVE_GMP_H 1
-
-/* Define to 1 if you have the <inttypes.h> header file. */
-#define HAVE_INTTYPES_H 1
-
-/* Define to 1 if you have the <malloc.h> header file. */
-#define HAVE_MALLOC_H 1
-
-/* Define to 1 if you have the <malloc/malloc.h> header file. */
-/* #undef HAVE_MALLOC_MALLOC_H */
-
-/* Define to 1 if you have the <memory.h> header file. */
-#define HAVE_MEMORY_H 1
-
-/* Define to 1 if you have the <mm_malloc.h> header file. */
-#define HAVE_MM_MALLOC_H 1
-
-/* Define to 1 if you have the <stdint.h> header file. */
-#define HAVE_STDINT_H 1
-
-/* Define to 1 if you have the <stdlib.h> header file. */
-#define HAVE_STDLIB_H 1
-
-/* Define to 1 if you have the <strings.h> header file. */
-#define HAVE_STRINGS_H 1
-
-/* Define to 1 if you have the <string.h> header file. */
-#define HAVE_STRING_H 1
-
-/* Define to 1 if you have the <sys/stat.h> header file. */
-#define HAVE_SYS_STAT_H 1
-
-/* Define to 1 if you have the <sys/types.h> header file. */
-#define HAVE_SYS_TYPES_H 1
-
-/* Define to 1 if you have the <unistd.h> header file. */
-#define HAVE_UNISTD_H 1
-
-/* Name of package */
-#define PACKAGE "grid"
-
-/* Define to the address where bug reports for this package should be sent. */
-#define PACKAGE_BUGREPORT "paboyle@ph.ed.ac.uk"
-
-/* Define to the full name of this package. */
-#define PACKAGE_NAME "Grid"
-
-/* Define to the full name and version of this package. */
-#define PACKAGE_STRING "Grid 1.0"
-
-/* Define to the one symbol short name of this package. */
-#define PACKAGE_TARNAME "grid"
-
-/* Define to the home page for this package. */
-#define PACKAGE_URL ""
-
-/* Define to the version of this package. */
-#define PACKAGE_VERSION "1.0"
-
-/* SSE4 */
-#define SSE4 1
-
-/* Define to 1 if you have the ANSI C header files. */
-#define STDC_HEADERS 1
-
-/* Version number of package */
-#define VERSION "1.0"
-
-/* Define for Solaris 2.5.1 so the uint32_t typedef from <sys/synch.h>,
-   <pthread.h>, or <semaphore.h> is not used. If the typedef were allowed, the
-   #define below would cause a syntax error. */
-/* #undef _UINT32_T */
-
-/* Define for Solaris 2.5.1 so the uint64_t typedef from <sys/synch.h>,
-   <pthread.h>, or <semaphore.h> is not used. If the typedef were allowed, the
-   #define below would cause a syntax error. */
-/* #undef _UINT64_T */
-
-/* Define to `unsigned int' if <sys/types.h> does not define. */
-/* #undef size_t */
-
-/* Define to the type of an unsigned integer type of width exactly 32 bits if
-   such a type exists and the standard includes do not define it. */
-/* #undef uint32_t */
-
-/* Define to the type of an unsigned integer type of width exactly 64 bits if
-   such a type exists and the standard includes do not define it. */
-/* #undef uint64_t */
diff --git a/lib/Grid_config.h.in b/lib/Grid_config.h.in
index 0ce09cf5..b7f56d5b 100644
--- a/lib/Grid_config.h.in
+++ b/lib/Grid_config.h.in
@@ -35,6 +35,12 @@
 /* Define to 1 if you have the <inttypes.h> header file. */
 #undef HAVE_INTTYPES_H
 
+/* Define to 1 if you have the `gmp' library (-lgmp). */
+#undef HAVE_LIBGMP
+
+/* Define to 1 if you have the `mpfr' library (-lmpfr). */
+#undef HAVE_LIBMPFR
+
 /* Define to 1 if you have the <malloc.h> header file. */
 #undef HAVE_MALLOC_H
 
diff --git a/lib/Makefile.am b/lib/Makefile.am
index 938f7ca1..557295d5 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -95,5 +95,6 @@ nobase_include_HEADERS = algorithms/approx/bigfloat.h\
 	simd/Grid_vComplexF.h\
 	simd/Grid_vInteger.h\
 	simd/Grid_vRealD.h\
-	simd/Grid_vRealF.h
+	simd/Grid_vRealF.h\
+	simd/Grid_vector_types.h
 
diff --git a/lib/algorithms/approx/.dirstamp b/lib/algorithms/approx/.dirstamp
deleted file mode 100644
index e69de29b..00000000
diff --git a/lib/algorithms/approx/Remez.cc b/lib/algorithms/approx/Remez.cc
old mode 100755
new mode 100644
diff --git a/lib/algorithms/approx/Remez.h b/lib/algorithms/approx/Remez.h
old mode 100755
new mode 100644
diff --git a/lib/algorithms/approx/Zolotarev.cc b/lib/algorithms/approx/Zolotarev.cc
old mode 100755
new mode 100644
diff --git a/lib/algorithms/approx/Zolotarev.h b/lib/algorithms/approx/Zolotarev.h
old mode 100755
new mode 100644
diff --git a/lib/algorithms/approx/bigfloat.h b/lib/algorithms/approx/bigfloat.h
old mode 100755
new mode 100644
diff --git a/lib/communicator/.dirstamp b/lib/communicator/.dirstamp
deleted file mode 100644
index e69de29b..00000000
diff --git a/lib/qcd/.dirstamp b/lib/qcd/.dirstamp
deleted file mode 100644
index e69de29b..00000000
diff --git a/lib/simd/Grid_sse4.cpp b/lib/simd/Grid_sse4.cpp
new file mode 100644
index 00000000..b0be164b
--- /dev/null
+++ b/lib/simd/Grid_sse4.cpp
@@ -0,0 +1,19 @@
+//----------------------------------------------------------------------
+/*! @file Grid_vector_types.h
+  @brief Defines templated class to deal with inner vector types
+*/
+// Time-stamp: <2015-05-19 13:53:47 neo>
+//----------------------------------------------------------------------
+
+namespace Optimization {
+
+
+
+
+}
+
+// Here assign types 
+namespace Grid {
+
+
+}
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index 360e9f1b..7c2b1bed 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -1,3 +1,9 @@
+//----------------------------------------------------------------------
+/*! @file Grid_vector_types.h
+  @brief Defines templated class to deal with inner vector types
+*/
+// Time-stamp: <2015-05-19 13:41:47 neo>
+//----------------------------------------------------------------------
 #ifndef GRID_VECTOR_TYPES
 #define GRID_VECTOR_TYPES
 
@@ -13,8 +19,8 @@ namespace Grid {
     struct RealPart< std::complex<T> >{
     typedef T type;
   };
-  ////////////////////////////////////////////////////////
 
+  ////////////////////////////////////////////////////////
   // Check for complexity with type traits
   template <typename T> 
     struct is_complex : std::false_type {};
diff --git a/lib/stamp-h1 b/lib/stamp-h1
deleted file mode 100644
index 753f890f..00000000
--- a/lib/stamp-h1
+++ /dev/null
@@ -1 +0,0 @@
-timestamp for lib/Grid_config.h
diff --git a/lib/stencil/.dirstamp b/lib/stencil/.dirstamp
deleted file mode 100644
index e69de29b..00000000

From b29caead322d1ead527c3f78316a4b67ed9fe17d Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Tue, 19 May 2015 17:21:17 +0900
Subject: [PATCH 200/429] Partial implementation of the vector types SIMD

Implementing SSE4 now
A systematic series of tests must be written.
---
 lib/Grid_config.h               | 127 +++++++++++++++++++++
 lib/Makefile.am                 | 158 +++++++++++++-------------
 lib/algorithms/approx/.dirstamp |   0
 lib/communicator/.dirstamp      |   0
 lib/qcd/.dirstamp               |   0
 lib/simd/.dirstamp              |   0
 lib/simd/Grid_sse4.cpp          |  19 ----
 lib/simd/Grid_sse4.h            | 194 ++++++++++++++++++++++++++++++++
 lib/simd/Grid_vector_types.h    | 125 ++++++++++++--------
 lib/stamp-h1                    |   1 +
 lib/stencil/.dirstamp           |   0
 tests/Grid_main.cc              |  37 ++++++
 12 files changed, 516 insertions(+), 145 deletions(-)
 create mode 100644 lib/Grid_config.h
 create mode 100644 lib/algorithms/approx/.dirstamp
 create mode 100644 lib/communicator/.dirstamp
 create mode 100644 lib/qcd/.dirstamp
 create mode 100644 lib/simd/.dirstamp
 delete mode 100644 lib/simd/Grid_sse4.cpp
 create mode 100644 lib/simd/Grid_sse4.h
 create mode 100644 lib/stamp-h1
 create mode 100644 lib/stencil/.dirstamp

diff --git a/lib/Grid_config.h b/lib/Grid_config.h
new file mode 100644
index 00000000..78582f3e
--- /dev/null
+++ b/lib/Grid_config.h
@@ -0,0 +1,127 @@
+/* lib/Grid_config.h.  Generated from Grid_config.h.in by configure.  */
+/* lib/Grid_config.h.in.  Generated from configure.ac by autoheader.  */
+
+/* AVX */
+/* #undef AVX1 */
+
+/* AVX2 */
+/* #undef AVX2 */
+
+/* AVX512 */
+/* #undef AVX512 */
+
+/* GRID_COMMS_MPI */
+/* #undef GRID_COMMS_MPI */
+
+/* GRID_COMMS_NONE */
+#define GRID_COMMS_NONE 1
+
+/* Define to 1 if you have the declaration of `be64toh', and to 0 if you
+   don't. */
+#define HAVE_DECL_BE64TOH 1
+
+/* Define to 1 if you have the declaration of `ntohll', and to 0 if you don't.
+   */
+#define HAVE_DECL_NTOHLL 0
+
+/* Define to 1 if you have the <endian.h> header file. */
+#define HAVE_ENDIAN_H 1
+
+/* Define to 1 if you have the `gettimeofday' function. */
+#define HAVE_GETTIMEOFDAY 1
+
+/* Define to 1 if you have the <gmp.h> header file. */
+#define HAVE_GMP_H 1
+
+/* Define to 1 if you have the <inttypes.h> header file. */
+#define HAVE_INTTYPES_H 1
+
+/* Define to 1 if you have the `gmp' library (-lgmp). */
+#define HAVE_LIBGMP 1
+
+/* Define to 1 if you have the `mpfr' library (-lmpfr). */
+#define HAVE_LIBMPFR 1
+
+/* Define to 1 if you have the <malloc.h> header file. */
+#define HAVE_MALLOC_H 1
+
+/* Define to 1 if you have the <malloc/malloc.h> header file. */
+/* #undef HAVE_MALLOC_MALLOC_H */
+
+/* Define to 1 if you have the <memory.h> header file. */
+#define HAVE_MEMORY_H 1
+
+/* Define to 1 if you have the <mm_malloc.h> header file. */
+#define HAVE_MM_MALLOC_H 1
+
+/* Define to 1 if you have the <stdint.h> header file. */
+#define HAVE_STDINT_H 1
+
+/* Define to 1 if you have the <stdlib.h> header file. */
+#define HAVE_STDLIB_H 1
+
+/* Define to 1 if you have the <strings.h> header file. */
+#define HAVE_STRINGS_H 1
+
+/* Define to 1 if you have the <string.h> header file. */
+#define HAVE_STRING_H 1
+
+/* Define to 1 if you have the <sys/stat.h> header file. */
+#define HAVE_SYS_STAT_H 1
+
+/* Define to 1 if you have the <sys/types.h> header file. */
+#define HAVE_SYS_TYPES_H 1
+
+/* Define to 1 if you have the <unistd.h> header file. */
+#define HAVE_UNISTD_H 1
+
+/* Name of package */
+#define PACKAGE "grid"
+
+/* Define to the address where bug reports for this package should be sent. */
+#define PACKAGE_BUGREPORT "paboyle@ph.ed.ac.uk"
+
+/* Define to the full name of this package. */
+#define PACKAGE_NAME "Grid"
+
+/* Define to the full name and version of this package. */
+#define PACKAGE_STRING "Grid 1.0"
+
+/* Define to the one symbol short name of this package. */
+#define PACKAGE_TARNAME "grid"
+
+/* Define to the home page for this package. */
+#define PACKAGE_URL ""
+
+/* Define to the version of this package. */
+#define PACKAGE_VERSION "1.0"
+
+/* SSE4 */
+#define SSE4 1
+
+/* Define to 1 if you have the ANSI C header files. */
+#define STDC_HEADERS 1
+
+/* Version number of package */
+#define VERSION "1.0"
+
+/* Define for Solaris 2.5.1 so the uint32_t typedef from <sys/synch.h>,
+   <pthread.h>, or <semaphore.h> is not used. If the typedef were allowed, the
+   #define below would cause a syntax error. */
+/* #undef _UINT32_T */
+
+/* Define for Solaris 2.5.1 so the uint64_t typedef from <sys/synch.h>,
+   <pthread.h>, or <semaphore.h> is not used. If the typedef were allowed, the
+   #define below would cause a syntax error. */
+/* #undef _UINT64_T */
+
+/* Define to `unsigned int' if <sys/types.h> does not define. */
+/* #undef size_t */
+
+/* Define to the type of an unsigned integer type of width exactly 32 bits if
+   such a type exists and the standard includes do not define it. */
+/* #undef uint32_t */
+
+/* Define to the type of an unsigned integer type of width exactly 64 bits if
+   such a type exists and the standard includes do not define it. */
+/* #undef uint64_t */
diff --git a/lib/Makefile.am b/lib/Makefile.am
index 557295d5..82459763 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -14,87 +14,89 @@ endif
 # Libraries
 #
 lib_LIBRARIES = libGrid.a
-libGrid_a_SOURCES =\
-	 Grid_init.cc\
-	stencil/Grid_stencil_common.cc\
-	qcd/Grid_qcd_dirac.cc\
-	qcd/Grid_qcd_wilson_dop.cc\
-	algorithms/approx/Zolotarev.cc\
-	algorithms/approx/Remez.cc\
+libGrid_a_SOURCES =				\
+	Grid_init.cc				\
+	stencil/Grid_stencil_common.cc		\
+	qcd/Grid_qcd_dirac.cc			\
+	qcd/Grid_qcd_wilson_dop.cc		\
+	algorithms/approx/Zolotarev.cc		\
+	algorithms/approx/Remez.cc		\
 	$(extra_sources)
 
 #
 # Include files
 #
-nobase_include_HEADERS = algorithms/approx/bigfloat.h\
-	algorithms/approx/Chebyshev.h\
-	algorithms/approx/Remez.h\
-	algorithms/approx/Zolotarev.h\
-	algorithms/iterative/ConjugateGradient.h\
-	algorithms/iterative/NormalEquations.h\
-	algorithms/iterative/SchurRedBlack.h\
-	algorithms/LinearOperator.h\
-	algorithms/SparseMatrix.h\
-	cartesian/Grid_cartesian_base.h\
-	cartesian/Grid_cartesian_full.h\
-	cartesian/Grid_cartesian_red_black.h\
-	communicator/Grid_communicator_base.h\
-	cshift/Grid_cshift_common.h\
-	cshift/Grid_cshift_mpi.h\
-	cshift/Grid_cshift_none.h\
-	Grid.h\
-	Grid_algorithms.h\
-	Grid_aligned_allocator.h\
-	Grid_cartesian.h\
-	Grid_communicator.h\
-	Grid_comparison.h\
-	Grid_cshift.h\
-	Grid_extract.h\
-	Grid_lattice.h\
-	Grid_math.h\
-	Grid_simd.h\
-	Grid_stencil.h\
-	Grid_threads.h\
-	lattice/Grid_lattice_arith.h\
-	lattice/Grid_lattice_base.h\
-	lattice/Grid_lattice_comparison.h\
-	lattice/Grid_lattice_conformable.h\
-	lattice/Grid_lattice_coordinate.h\
-	lattice/Grid_lattice_ET.h\
-	lattice/Grid_lattice_local.h\
-	lattice/Grid_lattice_overload.h\
-	lattice/Grid_lattice_peekpoke.h\
-	lattice/Grid_lattice_reality.h\
-	lattice/Grid_lattice_reduction.h\
-	lattice/Grid_lattice_rng.h\
-	lattice/Grid_lattice_trace.h\
-	lattice/Grid_lattice_transfer.h\
-	lattice/Grid_lattice_transpose.h\
-	lattice/Grid_lattice_where.h\
-	math/Grid_math_arith.h\
-	math/Grid_math_arith_add.h\
-	math/Grid_math_arith_mac.h\
-	math/Grid_math_arith_mul.h\
-	math/Grid_math_arith_scalar.h\
-	math/Grid_math_arith_sub.h\
-	math/Grid_math_inner.h\
-	math/Grid_math_outer.h\
-	math/Grid_math_peek.h\
-	math/Grid_math_poke.h\
-	math/Grid_math_reality.h\
-	math/Grid_math_tensors.h\
-	math/Grid_math_trace.h\
-	math/Grid_math_traits.h\
-	math/Grid_math_transpose.h\
-	parallelIO/GridNerscIO.h\
-	qcd/Grid_qcd.h\
-	qcd/Grid_qcd_2spinor.h\
-	qcd/Grid_qcd_dirac.h\
-	qcd/Grid_qcd_wilson_dop.h\
-	simd/Grid_vComplexD.h\
-	simd/Grid_vComplexF.h\
-	simd/Grid_vInteger.h\
-	simd/Grid_vRealD.h\
-	simd/Grid_vRealF.h\
-	simd/Grid_vector_types.h
+nobase_include_HEADERS = algorithms/approx/bigfloat.h		\
+	algorithms/approx/Chebyshev.h				\
+	algorithms/approx/Remez.h				\
+	algorithms/approx/Zolotarev.h				\
+	algorithms/iterative/ConjugateGradient.h		\
+	algorithms/iterative/NormalEquations.h			\
+	algorithms/iterative/SchurRedBlack.h			\
+	algorithms/LinearOperator.h				\
+	algorithms/SparseMatrix.h				\
+	cartesian/Grid_cartesian_base.h				\
+	cartesian/Grid_cartesian_full.h				\
+	cartesian/Grid_cartesian_red_black.h			\
+	communicator/Grid_communicator_base.h			\
+	cshift/Grid_cshift_common.h				\
+	cshift/Grid_cshift_mpi.h				\
+	cshift/Grid_cshift_none.h				\
+	Grid.h							\
+	Grid_algorithms.h					\
+	Grid_aligned_allocator.h				\
+	Grid_cartesian.h					\
+	Grid_communicator.h					\
+	Grid_comparison.h					\
+	Grid_cshift.h						\
+	Grid_extract.h						\
+	Grid_lattice.h						\
+	Grid_math.h						\
+	Grid_simd.h						\
+	Grid_stencil.h						\
+	Grid_threads.h						\
+	lattice/Grid_lattice_arith.h				\
+	lattice/Grid_lattice_base.h				\
+	lattice/Grid_lattice_comparison.h			\
+	lattice/Grid_lattice_conformable.h			\
+	lattice/Grid_lattice_coordinate.h			\
+	lattice/Grid_lattice_ET.h				\
+	lattice/Grid_lattice_local.h				\
+	lattice/Grid_lattice_overload.h				\
+	lattice/Grid_lattice_peekpoke.h				\
+	lattice/Grid_lattice_reality.h				\
+	lattice/Grid_lattice_reduction.h			\
+	lattice/Grid_lattice_rng.h				\
+	lattice/Grid_lattice_trace.h				\
+	lattice/Grid_lattice_transfer.h				\
+	lattice/Grid_lattice_transpose.h			\
+	lattice/Grid_lattice_where.h				\
+	math/Grid_math_arith.h					\
+	math/Grid_math_arith_add.h				\
+	math/Grid_math_arith_mac.h				\
+	math/Grid_math_arith_mul.h				\
+	math/Grid_math_arith_scalar.h				\
+	math/Grid_math_arith_sub.h				\
+	math/Grid_math_inner.h					\
+	math/Grid_math_outer.h					\
+	math/Grid_math_peek.h					\
+	math/Grid_math_poke.h					\
+	math/Grid_math_reality.h				\
+	math/Grid_math_tensors.h				\
+	math/Grid_math_trace.h					\
+	math/Grid_math_traits.h					\
+	math/Grid_math_transpose.h				\
+	parallelIO/GridNerscIO.h				\
+	qcd/Grid_qcd.h						\
+	qcd/Grid_qcd_2spinor.h					\
+	qcd/Grid_qcd_dirac.h					\
+	qcd/Grid_qcd_wilson_dop.h				\
+	simd/Grid_vComplexD.h					\
+	simd/Grid_vComplexF.h					\
+	simd/Grid_vInteger.h					\
+	simd/Grid_vRealD.h					\
+	simd/Grid_vRealF.h					\
+	simd/Grid_vector_types.h				\
+	simd/Grid_sse4.h					
+
 
diff --git a/lib/algorithms/approx/.dirstamp b/lib/algorithms/approx/.dirstamp
new file mode 100644
index 00000000..e69de29b
diff --git a/lib/communicator/.dirstamp b/lib/communicator/.dirstamp
new file mode 100644
index 00000000..e69de29b
diff --git a/lib/qcd/.dirstamp b/lib/qcd/.dirstamp
new file mode 100644
index 00000000..e69de29b
diff --git a/lib/simd/.dirstamp b/lib/simd/.dirstamp
new file mode 100644
index 00000000..e69de29b
diff --git a/lib/simd/Grid_sse4.cpp b/lib/simd/Grid_sse4.cpp
deleted file mode 100644
index b0be164b..00000000
--- a/lib/simd/Grid_sse4.cpp
+++ /dev/null
@@ -1,19 +0,0 @@
-//----------------------------------------------------------------------
-/*! @file Grid_vector_types.h
-  @brief Defines templated class to deal with inner vector types
-*/
-// Time-stamp: <2015-05-19 13:53:47 neo>
-//----------------------------------------------------------------------
-
-namespace Optimization {
-
-
-
-
-}
-
-// Here assign types 
-namespace Grid {
-
-
-}
diff --git a/lib/simd/Grid_sse4.h b/lib/simd/Grid_sse4.h
new file mode 100644
index 00000000..ed4039b7
--- /dev/null
+++ b/lib/simd/Grid_sse4.h
@@ -0,0 +1,194 @@
+//----------------------------------------------------------------------
+/*! @file Grid_sse4.h
+  @brief Optimization libraries
+*/
+// Time-stamp: <2015-05-19 17:06:51 neo>
+//----------------------------------------------------------------------
+
+#include <pmmintrin.h>
+
+namespace Optimization {
+  
+  struct Vsplat{
+    //Complex float
+    inline __m128 operator()(float a, float b){
+      return _mm_set_ps(b,a,b,a);
+    }
+    // Real float
+    inline __m128 operator()(float a){
+      return _mm_set_ps(a,a,a,a);
+    }
+    //Complex double
+    inline __m128d operator()(double a, double b){
+      return _mm_set_pd(b,a);
+    }
+    //Real double
+    inline __m128d operator()(double a){
+      return _mm_set_pd(a,a);
+    }
+    //Integer
+    inline __m128i operator()(Integer a){
+      return _mm_set1_epi32(a);
+    }
+  };
+
+  struct Vstore{
+    //Float 
+    inline void operator()(__m128 a, float* F){
+      _mm_store_ps(F,a);
+    }
+    //Double
+    inline void operator()(__m128d a, double* D){
+      _mm_store_pd(D,a);
+    }
+    //Integer
+    inline void operator()(__m128i a, Integer* I){
+      _mm_store_si128((__m128i *)I,a);
+    }
+
+  };
+
+
+
+  struct Vset{
+    // Complex float 
+    inline __m128 operator()(Grid::ComplexF *a){
+      return _mm_set_ps(a[1].imag(), a[1].real(),a[0].imag(),a[0].real());
+    }
+    // Complex double 
+    inline __m128d operator()(Grid::ComplexD *a){
+      return _mm_set_pd(a[0].imag(),a[0].real());
+    }
+    // Real float 
+    inline __m128 operator()(float *a){
+      return _mm_set_ps(a[3],a[2],a[1],a[0]);
+    }
+    // Real double
+    inline __m128d operator()(double *a){
+      return _mm_set_pd(a[1],a[0]);
+    }
+    // Integer
+    inline __m128i operator()(Integer *a){
+      return _mm_set_epi32(a[0],a[1],a[2],a[3]);
+    }
+
+
+  };
+
+  struct Reduce{
+    //Complex float
+    inline Grid::ComplexF operator()(__m128 in){
+      union {
+	__m128 v1;  
+	float f[4]; 
+      } u128;
+      u128.v1 = _mm_add_ps(in, _mm_shuffle_ps(in,in, 0b01001110)); // FIXME Prefer to use _MM_SHUFFLE macros
+      return Grid::ComplexF(u128.f[0], u128.f[1]);   
+    }
+    //Complex double
+    inline Grid::ComplexD operator()(__m128d in){
+      printf("Missing complex double implementation -> FIX\n");
+      return Grid::ComplexD(0,0); // FIXME wrong
+    }
+
+
+
+  };
+
+
+  /////////////////////////////////////////////////////
+  // Arithmetic operations
+  /////////////////////////////////////////////////////
+  struct Sum{
+    //Complex/Real float
+    inline __m128 operator()(__m128 a, __m128 b){
+      return _mm_add_ps(a,b);
+    }
+    //Complex/Real double
+    inline __m128d operator()(__m128d a, __m128d b){
+      return _mm_add_pd(a,b);
+    }
+    //Integer
+    inline __m128i operator()(__m128i a, __m128i b){
+      return _mm_add_epi32(a,b);
+    }
+  };
+
+  struct Sub{
+    //Complex/Real float
+    inline __m128 operator()(__m128 a, __m128 b){
+      return _mm_sub_ps(a,b);
+    }
+    //Complex/Real double
+    inline __m128d operator()(__m128d a, __m128d b){
+      return _mm_sub_pd(a,b);
+    }
+    //Integer
+    inline __m128i operator()(__m128i a, __m128i b){
+      return _mm_sub_epi32(a,b);
+    }
+  };
+
+  struct MultComplex{
+    // Complex float
+    inline __m128 operator()(__m128 a, __m128 b){
+      __m128 ymm0,ymm1,ymm2;
+      ymm0 = _mm_shuffle_ps(a,a,_MM_SHUFFLE(2,2,0,0)); // ymm0 <- ar ar,
+      ymm0 = _mm_mul_ps(ymm0,b);        // ymm0 <- ar bi, ar br
+      ymm1 = _mm_shuffle_ps(b,b,_MM_SHUFFLE(2,3,0,1)); // ymm1 <- br,bi
+      ymm2 = _mm_shuffle_ps(a,a,_MM_SHUFFLE(3,3,1,1)); // ymm2 <- ai,ai
+      ymm1 = _mm_mul_ps(ymm1,ymm2);       // ymm1 <- br ai, ai bi
+      return _mm_addsub_ps(ymm0,ymm1);    
+    }
+    // Complex double
+    inline __m128d operator()(__m128d a, __m128d b){
+      __m128d ymm0,ymm1,ymm2;
+      ymm0 = _mm_shuffle_pd(a,a,0x0); // ymm0 <- ar ar,
+      ymm0 = _mm_mul_pd(ymm0,b);        // ymm0 <- ar bi, ar br
+      ymm1 = _mm_shuffle_pd(b,b,0x1); // ymm1 <- br,bi   b01
+      ymm2 = _mm_shuffle_pd(a,a,0x3); // ymm2 <- ai,ai   b11
+      ymm1 = _mm_mul_pd(ymm1,ymm2);       // ymm1 <- br ai, ai bi
+      return _mm_addsub_pd(ymm0,ymm1);  
+    }
+  };
+
+  struct Mult{
+    // Real float
+    inline __m128 operator()(__m128 a, __m128 b){
+      return _mm_mul_ps(a,b);
+    }
+    // Real double
+    inline __m128d operator()(__m128d a, __m128d b){
+      return _mm_mul_pd(a,b);
+    }
+    // Integer
+    inline __m128i operator()(__m128i a, __m128i b){
+      return _mm_mul_epi32(a,b);
+    }
+
+  };
+
+
+
+
+}
+
+// Here assign types 
+namespace Grid {
+  typedef __m128 SIMD_Ftype;  // Single precision type
+  typedef __m128d SIMD_Dtype; // Double precision type
+  typedef __m128i SIMD_Itype; // Integer type
+
+
+  // Function names
+  typedef Optimization::Vsplat VsplatSIMD;
+  typedef Optimization::Vstore VstoreSIMD;
+
+  // Arithmetic operations
+  typedef Optimization::Sum         SumSIMD;
+  typedef Optimization::Sub         SubSIMD;
+  typedef Optimization::Mult        MultSIMD;
+  typedef Optimization::MultComplex MultComplexSIMD;
+  typedef Optimization::Vset        VsetSIMD;
+
+}
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index 7c2b1bed..030a8a79 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -1,12 +1,14 @@
-//----------------------------------------------------------------------
+//---------------------------------------------------------------------------
 /*! @file Grid_vector_types.h
-  @brief Defines templated class to deal with inner vector types
+  @brief Defines templated class Grid_simd to deal with inner vector types
 */
-// Time-stamp: <2015-05-19 13:41:47 neo>
-//----------------------------------------------------------------------
+// Time-stamp: <2015-05-19 17:20:36 neo>
+//---------------------------------------------------------------------------
 #ifndef GRID_VECTOR_TYPES
 #define GRID_VECTOR_TYPES
 
+#include "Grid_sse4.h"
+
 
 namespace Grid {
 
@@ -27,18 +29,20 @@ namespace Grid {
   template < typename T > 
     struct is_complex< std::complex<T> >: std::true_type {};
   ////////////////////////////////////////////////////////
-
-
   // Define the operation templates functors
-  template < class SIMD, class Operation > 
-    SIMD binary(SIMD src_1, SIMD src_2, Operation op){
+  // general forms to allow for vsplat syntax
+  // need explicit declaration of types when used since
+  // clang cannot automatically determine the output type sometimes
+  template < class Out, class Input1, class Input2, class Operation > 
+    Out binary(Input1 src_1, Input2 src_2, Operation op){
     return op(src_1, src_2);
-  }
+  } 
 
-  template < class SIMD, class Operation > 
-    SIMD unary(SIMD src, Operation op){
+  template < class SIMDout, class Input, class Operation > 
+    SIMDout unary(Input src, Operation op){
     return op(src);
-  }
+  } 
+
   ///////////////////////////////////////////////
 
   /*
@@ -74,36 +78,42 @@ namespace Grid {
     };
 
 
-
        
     ///////////////////////////////////////////////
     // mac, mult, sub, add, adj
-    // Should do an AVX2 version with mac.
     ///////////////////////////////////////////////
     friend inline void mac (Grid_simd * __restrict__ y,const Grid_simd * __restrict__ a,const Grid_simd *__restrict__ x){ *y = (*a)*(*x)+(*y); };
     friend inline void mult(Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) * (*r); }
     friend inline void sub (Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) - (*r); }
     friend inline void add (Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) + (*r); }
+    //not for integer types... FIXME
     friend inline Grid_simd adj(const Grid_simd &in){ return conj(in); }
         
-    //////////////////////////////////
-    // Initialise to 1,0,i
-    //////////////////////////////////
+    ///////////////////////////////////////////////
+    // Initialise to 1,0,i for the correct types
+    ///////////////////////////////////////////////
     // if not complex overload here 
-    friend inline void vone(Grid_simd &ret)      { vsplat(ret,1.0); }
-    friend inline void vzero(Grid_simd &ret)     { vsplat(ret,0.0); }
-
+    template <  class S = Scalar_type,typename std::enable_if < !is_complex < S >::value, int >::type = 0 > 
+      friend inline void vone(Grid_simd &ret)      { vsplat(ret,1.0); }
+    template <  class S = Scalar_type,typename std::enable_if < !is_complex < S >::value, int >::type = 0 > 
+      friend inline void vzero(Grid_simd &ret)     { vsplat(ret,0.0); }
+    
     // overload for complex type
     template <  class S = Scalar_type,typename std::enable_if < is_complex < S >::value, int >::type = 0 > 
-    friend inline void vone(Grid_simd &ret)      { vsplat(ret,1.0,0.0); }
+      friend inline void vone(Grid_simd &ret)      { vsplat(ret,1.0,0.0); }
     template < class S = Scalar_type,typename std::enable_if < is_complex < S >::value, int >::type = 0 > 
-    friend inline void vzero(Grid_simd &ret)     { vsplat(ret,0.0,0.0); }
-
+      friend inline void vzero(Grid_simd &ret)     { vsplat(ret,0.0,0.0); }// use xor?
+    
     // For integral type
     template <  class S = Scalar_type,typename std::enable_if < std::is_integral < S >::value, int >::type = 0 > 
-    friend inline void vone(Grid_simd &ret)      { vsplat(ret,1); }
+      friend inline void vone(Grid_simd &ret)      { vsplat(ret,1); }
     template <  class S = Scalar_type,typename std::enable_if < std::is_integral < S >::value, int >::type = 0 > 
-    friend inline void vzero(Grid_simd &ret)      { vsplat(ret,0); }
+      friend inline void vzero(Grid_simd &ret)      { vsplat(ret,0); }
+    template <  class S = Scalar_type,typename std::enable_if < std::is_integral < S >::value, int >::type = 0 > 
+      friend inline void vtrue (Grid_simd &ret){vsplat(ret,0xFFFFFFFF);}
+    template <  class S = Scalar_type,typename std::enable_if < std::is_integral < S >::value, int >::type = 0 > 
+      friend inline void vfalse(vInteger &ret){vsplat(ret,0);}
+
 
     // do not compile if real or integer, send an error message from the compiler
     template < class S = Scalar_type,typename std::enable_if < is_complex < S >::value, int >::type = 0 > 
@@ -114,31 +124,44 @@ namespace Grid {
     ////////////////////////////////////
     friend inline Grid_simd operator + (Grid_simd a, Grid_simd b)
     {
-      vComplexF ret;
-      // FIXME call the binary op
+      Grid_simd ret;
+      ret.v = binary<Vector_type>(a.v, b.v, SumSIMD());
       return ret;
     };
         
     friend inline Grid_simd operator - (Grid_simd a, Grid_simd b)
     {
-      vComplexF ret;
-      // FIXME call the binary op
+      Grid_simd ret;
+      ret.v = binary<Vector_type>(a.v, b.v, SubSIMD());
       return ret;
     };
         
-    friend inline Grid_simd operator * (Grid_simd a, Grid_simd b)
+    // Distinguish between complex types and others
+    template < class S = Scalar_type, typename std::enable_if < is_complex < S >::value, int >::type = 0 >
+      friend inline Grid_simd operator * (Grid_simd a, Grid_simd b)
       {
-	vComplexF ret;
-	// FIXME call the binary op
+	Grid_simd ret;
+	ret.v = binary<Vector_type>(a.v,b.v, MultComplexSIMD());
 	return ret;
       };
-      
+
+    // Real/Integer types
+    template <  class S = Scalar_type,typename std::enable_if < !is_complex < S >::value, int >::type = 0 > 
+    friend inline Grid_simd operator * (Grid_simd a, Grid_simd b)
+      {
+	Grid_simd ret;
+	ret.v = binary<Vector_type>(a.v,b.v, MultSIMD());
+	return ret;
+      };
+    
+
+
 
     ////////////////////////////////////////////////////////////////////////
     // FIXME:  gonna remove these load/store, get, set, prefetch
     ////////////////////////////////////////////////////////////////////////
     friend inline void vset(Grid_simd &ret, Scalar_type *a){
-      // FIXME set 
+      ret.v = unary<Vector_type>(a, VsetSIMD());
     }
         
     ///////////////////////
@@ -147,34 +170,33 @@ namespace Grid {
     // overload if complex
     template < class S = Scalar_type > 
     friend inline void vsplat(Grid_simd &ret, typename std::enable_if< is_complex < S >::value, S>::type c){
-      Real a= real(c);
-      Real b= imag(c);
+      Real a = real(c);
+      Real b = imag(c);
       vsplat(ret,a,b);
     }
 
     // this only for the complex version
     template < class S = Scalar_type, typename std::enable_if < is_complex < S >::value, int >::type = 0 > 
     friend inline void vsplat(Grid_simd &ret,Real a, Real b){
-      // FIXME add operator
+      ret.v = binary<Vector_type>(a, b, VsplatSIMD());
     }    
 
-    //if real fill with a, if complex fill with a in the real part
+    //if real fill with a, if complex fill with a in the real part (first function above)
     friend inline void vsplat(Grid_simd &ret,Real a){
-      // FIXME add operator
+      ret.v = unary<Vector_type>(a, VsplatSIMD());
     }    
 
 
-
     friend inline void vstore(const Grid_simd &ret, Scalar_type *a){
-      //FIXME
+      binary<void>(ret.v, (Real*)a, VstoreSIMD());
     }
+
     friend inline void vprefetch(const Grid_simd &v)
     {
       _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
     }
 
 
-
     friend inline Scalar_type Reduce(const Grid_simd & in)
     {
       // FIXME add operator
@@ -221,6 +243,7 @@ namespace Grid {
     inline Grid_simd &operator *=(const Grid_simd &r) {
       *this = (*this)*r;
       return *this;
+      // return (*this)*r; ?
     }
     inline Grid_simd &operator +=(const Grid_simd &r) {
       *this = *this+r;
@@ -233,6 +256,12 @@ namespace Grid {
 
 
 
+     friend inline void permute(Grid_simd &y,Grid_simd b,int perm)
+      {
+        Gpermute<Grid_simd>(y,b,perm);
+      }
+
+     /*
     friend inline void permute(Grid_simd &y,Grid_simd b,int perm)
     {
       Gpermute<Grid_simd>(y,b,perm);
@@ -253,7 +282,7 @@ namespace Grid {
     {
       Gextract<Grid_simd,Scalar_type>(y,extracted);
     }
-
+     */
 
   };// end of Grid_simd class definition 
 
@@ -286,11 +315,11 @@ namespace Grid {
 
 
   // Define available types (now change names to avoid clashing)
-  typedef __m128 SIMD_type;// decided at compilation time
-  typedef Grid_simd< float                 , SIMD_type > MyRealF;
-  typedef Grid_simd< double                , SIMD_type > MyRealD;
-  typedef Grid_simd< std::complex< float > , SIMD_type > MyComplexF;
-  typedef Grid_simd< std::complex< double >, SIMD_type > MyComplexD;
+
+  typedef Grid_simd< float                 , SIMD_Ftype > MyRealF;
+  typedef Grid_simd< double                , SIMD_Dtype > MyRealD;
+  typedef Grid_simd< std::complex< float > , SIMD_Ftype > MyComplexF;
+  typedef Grid_simd< std::complex< double >, SIMD_Dtype > MyComplexD;
 
 
 
diff --git a/lib/stamp-h1 b/lib/stamp-h1
new file mode 100644
index 00000000..753f890f
--- /dev/null
+++ b/lib/stamp-h1
@@ -0,0 +1 @@
+timestamp for lib/Grid_config.h
diff --git a/lib/stencil/.dirstamp b/lib/stencil/.dirstamp
new file mode 100644
index 00000000..e69de29b
diff --git a/tests/Grid_main.cc b/tests/Grid_main.cc
index 45778922..17e36df9 100644
--- a/tests/Grid_main.cc
+++ b/tests/Grid_main.cc
@@ -1,5 +1,9 @@
 #include "Grid.h"
 
+//DEBUG
+#include "simd/Grid_vector_types.h"
+
+
 using namespace std;
 using namespace Grid;
 using namespace Grid::QCD;
@@ -151,6 +155,39 @@ int main (int argc, char ** argv)
     
     scMat = sMat*scMat;  // LatticeSpinColourMatrix = LatticeSpinMatrix       * LatticeSpinColourMatrix
 
+
+
+#ifdef SSE4
+    ///////// Tests the new class Grid_simd 
+    std::complex<double> ctest(3.0,2.0);
+    std::complex<float> ctestf(3.0,2.0);
+    MyComplexF TestMe1(1.0); // fill real part
+    MyComplexD TestMe2(ctest);
+    MyComplexD TestMe3(ctest);// compiler generate conversion of basic types
+    //MyRealF TestMe5(ctest);// Must generate compiler error
+    MyRealD TestMe4(2.0); 
+ 
+    MyComplexF TestMe6(ctestf);
+    MyComplexF TestMe7(ctestf);  
+    
+    MyComplexD TheSum= TestMe2*TestMe3;
+    MyComplexF TheSumF= TestMe6*TestMe7;
+
+    double dsum[2];
+    _mm_store_pd(dsum, TheSum.v);
+    for (int i =0; i< 2; i++)
+      printf("%f\n", dsum[i]);
+
+    float fsum[4];
+    _mm_store_ps(fsum, TheSumF.v);
+    for (int i =0; i< 4; i++)
+      printf("%f\n", fsum[i]);
+
+    vstore(TheSum, &ctest);
+    std::cout << ctest<< std::endl;
+#endif
+    ///////////////////////
+
     // Non-lattice (const objects) * Lattice
     ColourMatrix cm;
     SpinColourMatrix scm;

From ffc00caea3a3308605028b27b24cf2ad07591cff Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 19 May 2015 13:57:35 +0100
Subject: [PATCH 201/429] Got unpreconditioned conjugate gradient to run and
 converge on a random (uniform random, not even SU(3) for now) gauge field.
 Convergence history is correctly indepdendent of decomposition on 1,2,4,8,16
 mpi tasks. Found a couple of simd bugs which required fixed and enhanced the
 Grid_simd.cc test suite. Implemented the Mdag, M, MdagM, Meooe Mooee schur
 type stuff in the wilson dop.

---
 Makefile.in                                  |  44 +++--
 aclocal.m4                                   |  61 +++----
 benchmarks/Grid_wilson.cc                    |  35 +++-
 benchmarks/Grid_wilson_cg_unprec.cc          |  56 +++++++
 benchmarks/Makefile.am                       |   6 +-
 config.guess                                 |   2 +-
 config.sub                                   |   2 +-
 configure                                    |  13 +-
 lib/Grid_simd.h                              |  20 ++-
 lib/algorithms/LinearOperator.h              |  13 +-
 lib/algorithms/iterative/ConjugateGradient.h |  16 +-
 lib/lattice/Grid_lattice_ET.h                |   4 +-
 lib/lattice/Grid_lattice_arith.h             |  40 ++++-
 lib/lattice/Grid_lattice_base.h              |   2 +-
 lib/lattice/Grid_lattice_reality.h           |   4 +-
 lib/math/Grid_math_reality.h                 |  12 +-
 lib/qcd/Grid_qcd_wilson_dop.cc               | 160 ++++++++++---------
 lib/qcd/Grid_qcd_wilson_dop.h                |   5 +-
 lib/simd/Grid_vComplexD.h                    |  26 +--
 lib/simd/Grid_vComplexF.h                    |  66 +++-----
 lib/simd/Grid_vInteger.h                     |   2 +-
 lib/simd/Grid_vRealD.h                       |  55 ++-----
 lib/simd/Grid_vRealF.h                       |  49 +++---
 lib/simd/Grid_vector_types.h                 |   6 +-
 scripts/build-all                            |   4 +-
 scripts/configure-all                        |   2 +-
 scripts/configure-commands                   |  11 +-
 tests/Grid_main.cc                           |   8 +-
 tests/Grid_rng_fixed.cc                      |  52 ++++++
 tests/Grid_simd.cc                           |  93 ++++++++++-
 tests/Grid_stencil.cc                        |   6 +-
 tests/Makefile.am                            |   5 +-
 tests/test_Grid_jacobi.cc                    |   2 +-
 33 files changed, 566 insertions(+), 316 deletions(-)
 create mode 100644 benchmarks/Grid_wilson_cg_unprec.cc
 create mode 100644 tests/Grid_rng_fixed.cc

diff --git a/Makefile.in b/Makefile.in
index b6894ef6..b9fba168 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -1,7 +1,7 @@
-# Makefile.in generated by automake 1.14.1 from Makefile.am.
+# Makefile.in generated by automake 1.15 from Makefile.am.
 # @configure_input@
 
-# Copyright (C) 1994-2013 Free Software Foundation, Inc.
+# Copyright (C) 1994-2014 Free Software Foundation, Inc.
 
 # This Makefile.in is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -14,7 +14,17 @@
 
 @SET_MAKE@
 VPATH = @srcdir@
-am__is_gnu_make = test -n '$(MAKEFILE_LIST)' && test -n '$(MAKELEVEL)'
+am__is_gnu_make = { \
+  if test -z '$(MAKELEVEL)'; then \
+    false; \
+  elif test -n '$(MAKE_HOST)'; then \
+    true; \
+  elif test -n '$(MAKE_VERSION)' && test -n '$(CURDIR)'; then \
+    true; \
+  else \
+    false; \
+  fi; \
+}
 am__make_running_with_option = \
   case $${target_option-} in \
       ?) ;; \
@@ -79,14 +89,12 @@ build_triplet = @build@
 host_triplet = @host@
 target_triplet = @target@
 subdir = .
-DIST_COMMON = INSTALL NEWS README AUTHORS ChangeLog \
-	$(srcdir)/Makefile.in $(srcdir)/Makefile.am \
-	$(top_srcdir)/configure $(am__configure_deps) COPYING TODO \
-	compile config.guess config.sub depcomp install-sh missing
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/configure.ac
 am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
 	$(ACLOCAL_M4)
+DIST_COMMON = $(srcdir)/Makefile.am $(top_srcdir)/configure \
+	$(am__configure_deps) $(am__DIST_COMMON)
 am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \
  configure.lineno config.status.lineno
 mkinstalldirs = $(install_sh) -d
@@ -149,6 +157,9 @@ ETAGS = etags
 CTAGS = ctags
 CSCOPE = cscope
 DIST_SUBDIRS = $(SUBDIRS)
+am__DIST_COMMON = $(srcdir)/Makefile.in AUTHORS COPYING ChangeLog \
+	INSTALL NEWS README TODO compile config.guess config.sub \
+	depcomp install-sh missing
 DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
 distdir = $(PACKAGE)-$(VERSION)
 top_distdir = $(distdir)
@@ -314,7 +325,6 @@ $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu Makefile'; \
 	$(am__cd) $(top_srcdir) && \
 	  $(AUTOMAKE) --gnu Makefile
-.PRECIOUS: Makefile
 Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
 	@case '$?' in \
 	  *config.status*) \
@@ -521,15 +531,15 @@ dist-xz: distdir
 	$(am__post_remove_distdir)
 
 dist-tarZ: distdir
-	@echo WARNING: "Support for shar distribution archives is" \
-	               "deprecated." >&2
+	@echo WARNING: "Support for distribution archives compressed with" \
+		       "legacy program 'compress' is deprecated." >&2
 	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	tardir=$(distdir) && $(am__tar) | compress -c >$(distdir).tar.Z
 	$(am__post_remove_distdir)
 
 dist-shar: distdir
-	@echo WARNING: "Support for distribution archives compressed with" \
-		       "legacy program 'compress' is deprecated." >&2
+	@echo WARNING: "Support for shar distribution archives is" \
+	               "deprecated." >&2
 	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	shar $(distdir) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).shar.gz
 	$(am__post_remove_distdir)
@@ -565,17 +575,17 @@ distcheck: dist
 	esac
 	chmod -R a-w $(distdir)
 	chmod u+w $(distdir)
-	mkdir $(distdir)/_build $(distdir)/_inst
+	mkdir $(distdir)/_build $(distdir)/_build/sub $(distdir)/_inst
 	chmod a-w $(distdir)
 	test -d $(distdir)/_build || exit 0; \
 	dc_install_base=`$(am__cd) $(distdir)/_inst && pwd | sed -e 's,^[^:\\/]:[\\/],/,'` \
 	  && dc_destdir="$${TMPDIR-/tmp}/am-dc-$$$$/" \
 	  && am__cwd=`pwd` \
-	  && $(am__cd) $(distdir)/_build \
-	  && ../configure \
+	  && $(am__cd) $(distdir)/_build/sub \
+	  && ../../configure \
 	    $(AM_DISTCHECK_CONFIGURE_FLAGS) \
 	    $(DISTCHECK_CONFIGURE_FLAGS) \
-	    --srcdir=.. --prefix="$$dc_install_base" \
+	    --srcdir=../.. --prefix="$$dc_install_base" \
 	  && $(MAKE) $(AM_MAKEFLAGS) \
 	  && $(MAKE) $(AM_MAKEFLAGS) dvi \
 	  && $(MAKE) $(AM_MAKEFLAGS) check \
@@ -749,6 +759,8 @@ uninstall-am:
 	maintainer-clean-generic mostlyclean mostlyclean-generic pdf \
 	pdf-am ps ps-am tags tags-am uninstall uninstall-am
 
+.PRECIOUS: Makefile
+
 
 # Tell versions [3.59,3.63) of GNU make to not export all variables.
 # Otherwise a system limit (for SysV at least) may be exceeded.
diff --git a/aclocal.m4 b/aclocal.m4
index 389763bf..bf79d078 100644
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -1,6 +1,6 @@
-# generated automatically by aclocal 1.14.1 -*- Autoconf -*-
+# generated automatically by aclocal 1.15 -*- Autoconf -*-
 
-# Copyright (C) 1996-2013 Free Software Foundation, Inc.
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -20,7 +20,7 @@ You have another version of autoconf.  It may work, but is not guaranteed to.
 If you have problems, you may need to regenerate the build system entirely.
 To do so, use the procedure documented by the package, typically 'autoreconf'.])])
 
-# Copyright (C) 2002-2013 Free Software Foundation, Inc.
+# Copyright (C) 2002-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -32,10 +32,10 @@ To do so, use the procedure documented by the package, typically 'autoreconf'.])
 # generated from the m4 files accompanying Automake X.Y.
 # (This private macro should not be called outside this file.)
 AC_DEFUN([AM_AUTOMAKE_VERSION],
-[am__api_version='1.14'
+[am__api_version='1.15'
 dnl Some users find AM_AUTOMAKE_VERSION and mistake it for a way to
 dnl require some minimum version.  Point them to the right macro.
-m4_if([$1], [1.14.1], [],
+m4_if([$1], [1.15], [],
       [AC_FATAL([Do not call $0, use AM_INIT_AUTOMAKE([$1]).])])dnl
 ])
 
@@ -51,14 +51,14 @@ m4_define([_AM_AUTOCONF_VERSION], [])
 # Call AM_AUTOMAKE_VERSION and AM_AUTOMAKE_VERSION so they can be traced.
 # This function is AC_REQUIREd by AM_INIT_AUTOMAKE.
 AC_DEFUN([AM_SET_CURRENT_AUTOMAKE_VERSION],
-[AM_AUTOMAKE_VERSION([1.14.1])dnl
+[AM_AUTOMAKE_VERSION([1.15])dnl
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
 _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 
 # AM_AUX_DIR_EXPAND                                         -*- Autoconf -*-
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -103,15 +103,14 @@ _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 # configured tree to be moved without reconfiguration.
 
 AC_DEFUN([AM_AUX_DIR_EXPAND],
-[dnl Rely on autoconf to set up CDPATH properly.
-AC_PREREQ([2.50])dnl
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+[AC_REQUIRE([AC_CONFIG_AUX_DIR_DEFAULT])dnl
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 ])
 
 # AM_CONDITIONAL                                            -*- Autoconf -*-
 
-# Copyright (C) 1997-2013 Free Software Foundation, Inc.
+# Copyright (C) 1997-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -142,7 +141,7 @@ AC_CONFIG_COMMANDS_PRE(
 Usually this means the macro was only invoked conditionally.]])
 fi])])
 
-# Copyright (C) 1999-2013 Free Software Foundation, Inc.
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -333,7 +332,7 @@ _AM_SUBST_NOTMAKE([am__nodep])dnl
 
 # Generate code to set up dependency tracking.              -*- Autoconf -*-
 
-# Copyright (C) 1999-2013 Free Software Foundation, Inc.
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -409,7 +408,7 @@ AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 
 # Do all the work for Automake.                             -*- Autoconf -*-
 
-# Copyright (C) 1996-2013 Free Software Foundation, Inc.
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -499,8 +498,8 @@ AC_REQUIRE([AC_PROG_MKDIR_P])dnl
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 AC_SUBST([mkdir_p], ['$(MKDIR_P)'])
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
 AC_REQUIRE([AC_PROG_AWK])dnl
 AC_REQUIRE([AC_PROG_MAKE_SET])dnl
 AC_REQUIRE([AM_SET_LEADING_DOT])dnl
@@ -573,7 +572,11 @@ to "yes", and re-run configure.
 END
     AC_MSG_ERROR([Your 'rm' program is bad, sorry.])
   fi
-fi])
+fi
+dnl The trailing newline in this macro's definition is deliberate, for
+dnl backward compatibility and to allow trailing 'dnl'-style comments
+dnl after the AM_INIT_AUTOMAKE invocation. See automake bug#16841.
+])
 
 dnl Hook into '_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
 dnl add the conditional right here, as _AC_COMPILER_EXEEXT may be further
@@ -602,7 +605,7 @@ for _am_header in $config_headers :; do
 done
 echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_count])
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -613,7 +616,7 @@ echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_co
 # Define $install_sh.
 AC_DEFUN([AM_PROG_INSTALL_SH],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -623,7 +626,7 @@ if test x"${install_sh}" != xset; then
 fi
 AC_SUBST([install_sh])])
 
-# Copyright (C) 2003-2013 Free Software Foundation, Inc.
+# Copyright (C) 2003-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -644,7 +647,7 @@ AC_SUBST([am__leading_dot])])
 
 # Check to see how 'make' treats includes.	            -*- Autoconf -*-
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -694,7 +697,7 @@ rm -f confinc confmf
 
 # Fake the existence of programs that GNU maintainers use.  -*- Autoconf -*-
 
-# Copyright (C) 1997-2013 Free Software Foundation, Inc.
+# Copyright (C) 1997-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -733,7 +736,7 @@ fi
 
 # Helper functions for option handling.                     -*- Autoconf -*-
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -764,7 +767,7 @@ AC_DEFUN([_AM_IF_OPTION],
 
 # Check to make sure that the build environment is sane.    -*- Autoconf -*-
 
-# Copyright (C) 1996-2013 Free Software Foundation, Inc.
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -845,7 +848,7 @@ AC_CONFIG_COMMANDS_PRE(
 rm -f conftest.file
 ])
 
-# Copyright (C) 2009-2013 Free Software Foundation, Inc.
+# Copyright (C) 2009-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -905,7 +908,7 @@ AC_SUBST([AM_BACKSLASH])dnl
 _AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
 ])
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -933,7 +936,7 @@ fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 AC_SUBST([INSTALL_STRIP_PROGRAM])])
 
-# Copyright (C) 2006-2013 Free Software Foundation, Inc.
+# Copyright (C) 2006-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -952,7 +955,7 @@ AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 
 # Check how to create a tarball.                            -*- Autoconf -*-
 
-# Copyright (C) 2004-2013 Free Software Foundation, Inc.
+# Copyright (C) 2004-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
diff --git a/benchmarks/Grid_wilson.cc b/benchmarks/Grid_wilson.cc
index e49d4515..69f5b48f 100644
--- a/benchmarks/Grid_wilson.cc
+++ b/benchmarks/Grid_wilson.cc
@@ -46,6 +46,13 @@ int main (int argc, char ** argv)
     volume=volume*latt_size[mu];
   }  
 
+  // Only one non-zero (y)
+  Umu=zero;
+  for(int nn=0;nn<Nd;nn++){
+    random(pRNG,U[nn]);
+    pokeIndex<LorentzIndex>(Umu,U[nn],nn);
+  }
+
   for(int mu=0;mu<Nd;mu++){
     U[mu] = peekIndex<LorentzIndex>(Umu,mu);
   }
@@ -53,7 +60,6 @@ int main (int argc, char ** argv)
   { // Naive wilson implementation
     ref = zero;
     for(int mu=0;mu<Nd;mu++){
-
       //    ref =  src + Gamma(Gamma::GammaX)* src ; // 1-gamma_x
       tmp = U[mu]*Cshift(src,mu,1);
       for(int i=0;i<ref._odata.size();i++){
@@ -75,7 +81,7 @@ int main (int argc, char ** argv)
   int ncall=1000;
   double t0=usecond();
   for(int i=0;i<ncall;i++){
-    Dw.M(src,result);
+    Dw.Dhop(src,result,0);
   }
   double t1=usecond();
   double flops=1320*volume*ncall;
@@ -99,5 +105,30 @@ int main (int argc, char ** argv)
     }
   }
 
+  { // Naive wilson dag implementation
+    ref = zero;
+    for(int mu=0;mu<Nd;mu++){
+
+      //    ref =  src - Gamma(Gamma::GammaX)* src ; // 1+gamma_x
+      tmp = U[mu]*Cshift(src,mu,1);
+      for(int i=0;i<ref._odata.size();i++){
+	ref._odata[i]+= tmp._odata[i] - Gamma(Gmu[mu])*tmp._odata[i]; ;
+      }
+
+      tmp =adj(U[mu])*src;
+      tmp =Cshift(tmp,mu,-1);
+      for(int i=0;i<ref._odata.size();i++){
+	ref._odata[i]+= tmp._odata[i] + Gamma(Gmu[mu])*tmp._odata[i]; ;
+      }
+    }
+  }
+  Dw.Dhop(src,result,1);
+  std::cout << "Called DwDag"<<std::endl;
+  std::cout << "norm result "<< norm2(result)<<std::endl;
+  std::cout << "norm ref    "<< norm2(ref)<<std::endl;
+  err = ref -result;
+  std::cout << "norm diff   "<< norm2(err)<<std::endl;
+
+
   Grid_finalize();
 }
diff --git a/benchmarks/Grid_wilson_cg_unprec.cc b/benchmarks/Grid_wilson_cg_unprec.cc
new file mode 100644
index 00000000..2a91864b
--- /dev/null
+++ b/benchmarks/Grid_wilson_cg_unprec.cc
@@ -0,0 +1,56 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+  GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+
+  std::vector<int> seeds({1,2,3,4});
+  GridParallelRNG          pRNG(&Grid);  pRNG.SeedFixedIntegers(seeds);
+
+  LatticeFermion src(&Grid); random(pRNG,src);
+  std::cout << "src norm" << norm2(src)<<std::endl;
+  LatticeFermion result(&Grid); result=zero;
+  LatticeGaugeField Umu(&Grid); random(pRNG,Umu);
+
+  std::vector<LatticeColourMatrix> U(4,&Grid);
+
+  double volume=1;
+  for(int mu=0;mu<Nd;mu++){
+    volume=volume*latt_size[mu];
+  }  
+
+  for(int mu=0;mu<Nd;mu++){
+    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+  }
+  
+  RealD mass=0.5;
+  WilsonMatrix Dw(Umu,mass);
+  HermitianOperator<WilsonMatrix,LatticeFermion> HermOp(Dw);
+  ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
+  CG(HermOp,src,result);
+
+  
+  Grid_finalize();
+}
diff --git a/benchmarks/Makefile.am b/benchmarks/Makefile.am
index 2fe40feb..77ce7663 100644
--- a/benchmarks/Makefile.am
+++ b/benchmarks/Makefile.am
@@ -5,12 +5,14 @@ AM_LDFLAGS = -L$(top_builddir)/lib
 #
 # Test code
 #
-bin_PROGRAMS = Grid_wilson Grid_comms Grid_memory_bandwidth Grid_su3
-
+bin_PROGRAMS = Grid_wilson Grid_comms Grid_memory_bandwidth Grid_su3 Grid_wilson_cg_unprec
 
 Grid_wilson_SOURCES = Grid_wilson.cc
 Grid_wilson_LDADD = -lGrid
 
+Grid_wilson_cg_unprec_SOURCES = Grid_wilson_cg_unprec.cc
+Grid_wilson_cg_unprec_LDADD = -lGrid
+
 Grid_comms_SOURCES = Grid_comms.cc
 Grid_comms_LDADD = -lGrid
 
diff --git a/config.guess b/config.guess
index 5f6aa02d..a12faba2 120000
--- a/config.guess
+++ b/config.guess
@@ -1 +1 @@
-/usr/share/automake-1.14/config.guess
\ No newline at end of file
+/opt/local/share/automake-1.15/config.guess
\ No newline at end of file
diff --git a/config.sub b/config.sub
index 0abfe18c..e3c9b5ca 120000
--- a/config.sub
+++ b/config.sub
@@ -1 +1 @@
-/usr/share/automake-1.14/config.sub
\ No newline at end of file
+/opt/local/share/automake-1.15/config.sub
\ No newline at end of file
diff --git a/configure b/configure
index e7d9f32f..b42143e2 100755
--- a/configure
+++ b/configure
@@ -2466,7 +2466,7 @@ test -n "$target_alias" &&
     NONENONEs,x,x, &&
   program_prefix=${target_alias}-
 
-am__api_version='1.14'
+am__api_version='1.15'
 
 # Find a good install program.  We prefer a C program (faster),
 # so one script is as good as another.  But avoid the broken or
@@ -2638,8 +2638,8 @@ test "$program_suffix" != NONE &&
 ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
 program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
 
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 
 if test x"${MISSING+set}" != xset; then
   case $am_aux_dir in
@@ -2658,7 +2658,7 @@ else
 $as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
 fi
 
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -2986,8 +2986,8 @@ MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 mkdir_p='$(MKDIR_P)'
 
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
 # Always define AMTAR for backward compatibility.  Yes, it's still used
 # in the wild :-(  We should find a proper way to deprecate it ...
 AMTAR='$${TAR-tar}'
@@ -3046,6 +3046,7 @@ END
 fi
 
 
+
 ac_config_headers="$ac_config_headers lib/Grid_config.h"
 
 # Check whether --enable-silent-rules was given.
diff --git a/lib/Grid_simd.h b/lib/Grid_simd.h
index af77591c..6d7411d3 100644
--- a/lib/Grid_simd.h
+++ b/lib/Grid_simd.h
@@ -50,15 +50,20 @@ namespace Grid {
   typedef std::complex<Real>  Complex;
 
   inline RealF adj(const RealF  & r){ return r; }
-  inline RealF conj(const RealF  & r){ return r; }
+  inline RealF conjugate(const RealF  & r){ return r; }
   inline RealF real(const RealF  & r){ return r; }
 
   inline RealD adj(const RealD  & r){ return r; }
-  inline RealD conj(const RealD  & r){ return r; }
+  inline RealD conjugate(const RealD  & r){ return r; }
   inline RealD real(const RealD  & r){ return r; }
 
-  inline ComplexD innerProduct(const ComplexD & l, const ComplexD & r) { return conj(l)*r; }
-  inline ComplexF innerProduct(const ComplexF & l, const ComplexF & r) { return conj(l)*r; }
+  inline ComplexD conjugate(const ComplexD& r){ return(conj(r)); }
+  inline ComplexD adj(const ComplexD& r){ return(conjugate(r)); }
+  inline ComplexF conjugate(const ComplexF& r ){ return(conj(r)); }
+  inline ComplexF adj(const ComplexF& r ){ return(conjugate(r)); }
+
+  inline ComplexD innerProduct(const ComplexD & l, const ComplexD & r) { return conjugate(l)*r; }
+  inline ComplexF innerProduct(const ComplexF & l, const ComplexF & r) { return conjugate(l)*r; }
   inline RealD innerProduct(const RealD & l, const RealD & r) { return l*r; }
   inline RealF innerProduct(const RealF & l, const RealF & r) { return l*r; }
 
@@ -70,15 +75,14 @@ namespace Grid {
     inline void mult(ComplexD * __restrict__ y,const ComplexD * __restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) * (*r);}
     inline void sub (ComplexD * __restrict__ y,const ComplexD * __restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) - (*r);}
     inline void add (ComplexD * __restrict__ y,const ComplexD * __restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) + (*r);}
-    inline ComplexD adj(const ComplexD& r){ return(conj(r)); }
-    // conj already supported for complex
+    // conjugate already supported for complex
     
     inline void mac (ComplexF * __restrict__ y,const ComplexF * __restrict__ a,const ComplexF *__restrict__ x){ *y = (*a) * (*x)+(*y); }
     inline void mult(ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) * (*r); }
     inline void sub (ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) - (*r); }
     inline void add (ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) + (*r); }
-    inline ComplexF  adj(const ComplexF& r ){ return(conj(r)); }
-    //conj already supported for complex
+
+    //conjugate already supported for complex
 
     inline ComplexF timesI(const ComplexF &r)     { return(r*ComplexF(0.0,1.0));}
     inline ComplexD timesI(const ComplexD &r)     { return(r*ComplexD(0.0,1.0));}
diff --git a/lib/algorithms/LinearOperator.h b/lib/algorithms/LinearOperator.h
index bb948b95..bdb6c387 100644
--- a/lib/algorithms/LinearOperator.h
+++ b/lib/algorithms/LinearOperator.h
@@ -7,8 +7,8 @@ namespace Grid {
   // LinearOperators Take a something and return a something.
   /////////////////////////////////////////////////////////////////////////////////////////////
   //
-  // Hopefully linearity is satisfied and the AdjOp is indeed the Hermitian conjugate (transpose if real):
-  //
+  // Hopefully linearity is satisfied and the AdjOp is indeed the Hermitian conjugateugate (transpose if real):
+  //SBase
   //   i)  F(a x + b y) = aF(x) + b F(y).
   //  ii)  <x|Op|y> = <y|AdjOp|x>^\ast
   //
@@ -25,12 +25,13 @@ namespace Grid {
   /////////////////////////////////////////////////////////////////////////////////////////////
     template<class Field> class HermitianOperatorBase : public LinearOperatorBase<Field> {
     public:
-      virtual RealD OpAndNorm(const Field &in, Field &out);
+      virtual void OpAndNorm(const Field &in, Field &out,double &n1,double &n2);
       void AdjOp(const Field &in, Field &out) {
 	Op(in,out);
       };
       void Op(const Field &in, Field &out) {
-	OpAndNorm(in,out);
+	double n1,n2;
+	OpAndNorm(in,out,n1,n2);
       };
     };
 
@@ -80,8 +81,8 @@ namespace Grid {
       Matrix &_Mat;
     public:
       HermitianOperator(Matrix &Mat): _Mat(Mat) {};
-      RealD OpAndNorm(const Field &in, Field &out){
-	return _Mat.MdagM(in,out);
+      void OpAndNorm(const Field &in, Field &out,double &n1,double &n2){
+	return _Mat.MdagM(in,out,n1,n2);
       }
     };
 
diff --git a/lib/algorithms/iterative/ConjugateGradient.h b/lib/algorithms/iterative/ConjugateGradient.h
index bcb6ab60..f7f8559e 100644
--- a/lib/algorithms/iterative/ConjugateGradient.h
+++ b/lib/algorithms/iterative/ConjugateGradient.h
@@ -18,6 +18,7 @@ public:
       std::cout << Tolerance<<std::endl;
     };
 
+    void operator() (LinearOperatorBase<Field> &Linop,const Field &src, Field &psi) {assert(0);};
     void operator() (HermitianOperatorBase<Field> &Linop,const Field &src, Field &psi){
 
       RealD cp,c,a,d,b,ssq,qq,b_pred;
@@ -61,21 +62,27 @@ public:
 	
 	Linop.OpAndNorm(p,mmp,d,qq);
 	  
+	//	std::cout <<std::setprecision(4)<< "ConjugateGradient:  d,qq "<<d<< " "<<qq <<std::endl;
+      
 	a      = c/d;
 	b_pred = a*(a*qq-d)/c;
 	
-	cp = axpy_norm(r,mmp,r,-a);
+	//	std::cout <<std::setprecision(4)<< "ConjugateGradient:  a,bp "<<a<< " "<<b_pred <<std::endl;
+
+	cp = axpy_norm(r,-a,mmp,r);
 	b = cp/c;
+	//	std::cout <<std::setprecision(4)<< "ConjugateGradient:  cp,b "<<cp<< " "<<b <<std::endl;
 	
 	// Fuse these loops ; should be really easy
 	psi= a*p+psi;
 	p  = p*b+r;
 	  
-	std::cout << "Iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
+	std::cout<<"ConjugateGradient: Iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
+
 	// Stopping condition
 	if ( cp <= rsq ) { 
 	  
-	  Linop.Op(p,mmp);
+	  Linop.Op(psi,mmp);
 	  p=mmp-src;
 	  
 	  RealD mmpnorm = sqrt(norm2(mmp));
@@ -83,8 +90,11 @@ public:
 	  RealD srcnorm = sqrt(norm2(src));
 	  RealD resnorm = sqrt(norm2(p));
 	  RealD true_residual = resnorm/srcnorm;
+
+	  std::cout<<"ConjugateGradient: Converged on iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
 	  std::cout<<"ConjugateGradient: true   residual  is "<<true_residual<<" sol "<<psinorm<<" src "<<srcnorm<<std::endl;
 	  std::cout<<"ConjugateGradient: target residual was "<<Tolerance<<std::endl;
+	  return;
 	}
       }
       std::cout<<"ConjugateGradient did NOT converge"<<std::endl;
diff --git a/lib/lattice/Grid_lattice_ET.h b/lib/lattice/Grid_lattice_ET.h
index c0b5f97d..b0205f2f 100644
--- a/lib/lattice/Grid_lattice_ET.h
+++ b/lib/lattice/Grid_lattice_ET.h
@@ -102,7 +102,7 @@ template <class arg> struct name\
 
 GridUnopClass(UnarySub,-a);
 GridUnopClass(UnaryAdj,adj(a));
-GridUnopClass(UnaryConj,conj(a));
+GridUnopClass(UnaryConj,conjugate(a));
 GridUnopClass(UnaryTrace,trace(a));
 GridUnopClass(UnaryTranspose,transpose(a));
 
@@ -178,7 +178,7 @@ template <typename T1,typename T2,typename T3> inline auto op(const T1 &pred,con
 
 GRID_DEF_UNOP(operator -,UnarySub);
 GRID_DEF_UNOP(adj,UnaryAdj);
-GRID_DEF_UNOP(conj,UnaryConj);
+GRID_DEF_UNOP(conjugate,UnaryConj);
 GRID_DEF_UNOP(trace,UnaryTrace);
 GRID_DEF_UNOP(transpose,UnaryTranspose);
 
diff --git a/lib/lattice/Grid_lattice_arith.h b/lib/lattice/Grid_lattice_arith.h
index c0bbb2b6..f1e566a2 100644
--- a/lib/lattice/Grid_lattice_arith.h
+++ b/lib/lattice/Grid_lattice_arith.h
@@ -144,14 +144,44 @@ PARALLEL_FOR_LOOP
   }
   
   template<class sobj,class vobj> strong_inline
-  void axpy(Lattice<vobj> &ret,sobj a,const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
-    conformable(lhs,rhs);
+  void axpy(Lattice<vobj> &ret,sobj a,const Lattice<vobj> &x,const Lattice<vobj> &y){
+    conformable(x,y);
 #pragma omp parallel for
-    for(int ss=0;ss<lhs._grid->oSites();ss++){
-      vobj tmp = a*lhs._odata[ss];
-      vstream(ret._odata[ss],tmp+rhs._odata[ss]);
+    for(int ss=0;ss<x._grid->oSites();ss++){
+      vobj tmp = a*x._odata[ss]+y._odata[ss];
+      vstream(ret._odata[ss],tmp);
     }
   }
+  template<class sobj,class vobj> strong_inline
+  void axpby(Lattice<vobj> &ret,sobj a,sobj b,const Lattice<vobj> &x,const Lattice<vobj> &y){
+    conformable(x,y);
+#pragma omp parallel for
+    for(int ss=0;ss<x._grid->oSites();ss++){
+      vobj tmp = a*x._odata[ss]+b*y._odata[ss];
+      vstream(ret._odata[ss],tmp);
+    }
+  }
+
+  template<class sobj,class vobj> strong_inline
+  RealD axpy_norm(Lattice<vobj> &ret,sobj a,const Lattice<vobj> &x,const Lattice<vobj> &y){
+    conformable(x,y);
+#pragma omp parallel for
+    for(int ss=0;ss<x._grid->oSites();ss++){
+      vobj tmp = a*x._odata[ss]+y._odata[ss];
+      vstream(ret._odata[ss],tmp);
+    }
+    return norm2(ret);
+  }
+  template<class sobj,class vobj> strong_inline
+  RealD axpby_norm(Lattice<vobj> &ret,sobj a,sobj b,const Lattice<vobj> &x,const Lattice<vobj> &y){
+    conformable(x,y);
+#pragma omp parallel for
+    for(int ss=0;ss<x._grid->oSites();ss++){
+      vobj tmp = a*x._odata[ss]+b*y._odata[ss];
+      vstream(ret._odata[ss],tmp);
+    }
+    return norm2(ret); // FIXME implement parallel norm in ss loop
+  }
 
 }
 #endif
diff --git a/lib/lattice/Grid_lattice_base.h b/lib/lattice/Grid_lattice_base.h
index ae606ae7..0016bafa 100644
--- a/lib/lattice/Grid_lattice_base.h
+++ b/lib/lattice/Grid_lattice_base.h
@@ -9,7 +9,7 @@ namespace Grid {
 // Functionality:
 //     -=,+=,*=,()
 //     add,+,sub,-,mult,mac,*
-//     adj,conj
+//     adj,conjugate
 //     real,imag
 //     transpose,transposeIndex  
 //     trace,traceIndex
diff --git a/lib/lattice/Grid_lattice_reality.h b/lib/lattice/Grid_lattice_reality.h
index cf86d700..2f52ed22 100644
--- a/lib/lattice/Grid_lattice_reality.h
+++ b/lib/lattice/Grid_lattice_reality.h
@@ -18,11 +18,11 @@ PARALLEL_FOR_LOOP
         return ret;
     };
 
-    template<class vobj> inline Lattice<vobj> conj(const Lattice<vobj> &lhs){
+    template<class vobj> inline Lattice<vobj> conjugate(const Lattice<vobj> &lhs){
         Lattice<vobj> ret(lhs._grid);
 PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){
-            ret._odata[ss] = conj(lhs._odata[ss]);
+            ret._odata[ss] = conjugate(lhs._odata[ss]);
         }
         return ret;
     };
diff --git a/lib/math/Grid_math_reality.h b/lib/math/Grid_math_reality.h
index 18a2f89b..40d4c624 100644
--- a/lib/math/Grid_math_reality.h
+++ b/lib/math/Grid_math_reality.h
@@ -98,26 +98,26 @@ template<class vtype,int N> inline void  timesMinusI(iMatrix<vtype,N> &ret,const
 /////////////////////////////////////////////// 
 // Conj function for scalar, vector, matrix
 /////////////////////////////////////////////// 
-template<class vtype> inline iScalar<vtype> conj(const iScalar<vtype>&r)
+template<class vtype> inline iScalar<vtype> conjugate(const iScalar<vtype>&r)
 {
     iScalar<vtype> ret;
-    ret._internal = conj(r._internal);
+    ret._internal = conjugate(r._internal);
     return ret;
 }
-template<class vtype,int N> inline iVector<vtype,N> conj(const iVector<vtype,N>&r)
+template<class vtype,int N> inline iVector<vtype,N> conjugate(const iVector<vtype,N>&r)
 {
   iVector<vtype,N> ret;
   for(int i=0;i<N;i++){
-    ret._internal[i] = conj(r._internal[i]);
+    ret._internal[i] = conjugate(r._internal[i]);
   }
   return ret;
 }
-template<class vtype,int N> inline iMatrix<vtype,N> conj(const iMatrix<vtype,N>&r)
+template<class vtype,int N> inline iMatrix<vtype,N> conjugate(const iMatrix<vtype,N>&r)
 {
   iMatrix<vtype,N> ret;
   for(int i=0;i<N;i++){
   for(int j=0;j<N;j++){
-    ret._internal[i][j] = conj(r._internal[i][j]);
+    ret._internal[i][j] = conjugate(r._internal[i][j]);
   }}
   return ret;
 }
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index 6a67efe3..5143daf3 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -21,7 +21,13 @@ const int WilsonMatrix::Tm = 7;
   class WilsonCompressor {
   public:
     int mu;
-    
+    int dag;
+
+    WilsonCompressor(int _dag){
+      mu=0;
+      dag=_dag;
+      assert((dag==0)||(dag==1));
+    }
     void Point(int p) { 
       mu=p;
     };
@@ -29,7 +35,11 @@ const int WilsonMatrix::Tm = 7;
     vHalfSpinColourVector operator () (const vSpinColourVector &in)
     {
       vHalfSpinColourVector ret;
-      switch(mu) {
+      int mudag=mu;
+      if (dag) {
+	mudag=(mu+Nd)%(2*Nd);
+      }
+      switch(mudag) {
       case WilsonMatrix::Xp:
 	spProjXp(ret,in);
 	break;
@@ -87,43 +97,52 @@ void WilsonMatrix::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeF
 
 RealD WilsonMatrix::M(const LatticeFermion &in, LatticeFermion &out)
 {
-  Dhop(in,out);
-  return 0.0;
+  Dhop(in,out,0);
+  out = (4+mass)*in - 0.5*out  ; // FIXME : axpby_norm! fusion fun
+  return norm2(out);
 }
 RealD WilsonMatrix::Mdag(const LatticeFermion &in, LatticeFermion &out)
 {
-  Dhop(in,out);
-  return 0.0;
+  Dhop(in,out,1);
+  out = (4+mass)*in - 0.5*out  ; // FIXME : axpby_norm! fusion fun
+  return norm2(out);
 }
 
 void WilsonMatrix::Meooe(const LatticeFermion &in, LatticeFermion &out)
 {
-  Dhop(in,out);
+  Dhop(in,out,0);
+  out = 0.5*out; // FIXME : scale factor in Dhop
 }
 void WilsonMatrix::MeooeDag(const LatticeFermion &in, LatticeFermion &out)
 {
-  Dhop(in,out);
+  Dhop(in,out,1);
 }
 void WilsonMatrix::Mooee(const LatticeFermion &in, LatticeFermion &out)
 {
+  out = (4.0+mass)*in;
   return ;
 }
 void WilsonMatrix::MooeeInv(const LatticeFermion &in, LatticeFermion &out)
 {
+  out = (1.0/(4.0+mass))*in;
   return ;
 }
 void WilsonMatrix::MooeeDag(const LatticeFermion &in, LatticeFermion &out)
 {
+  out = (1.0/(4.0+mass))*in;
   return ;
 }
 void WilsonMatrix::MooeeInvDag(const LatticeFermion &in, LatticeFermion &out)
 {
+  out = (1.0/(4.0+mass))*in;
   return ;
 }
 
-void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out)
+void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out,int dag)
 {
-  WilsonCompressor compressor;
+  assert((dag==0) ||(dag==1));
+
+  WilsonCompressor compressor(dag);
   Stencil.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
 
 PARALLEL_FOR_LOOP
@@ -140,13 +159,13 @@ PARALLEL_FOR_LOOP
     int ssu= ss;
 
     // Xp
-    offset = Stencil._offsets [Xp][ss];
-    local  = Stencil._is_local[Xp][ss];
-    perm   = Stencil._permute[Xp][ss];
-    ptype  = Stencil._permute_type[Xp];
+    int idx = (Xp+dag*4)%8;
+    offset = Stencil._offsets [idx][ss];
+    local  = Stencil._is_local[idx][ss];
+    perm   = Stencil._permute[idx][ss];
 
-    if ( local && perm ) 
-    {
+    ptype  = Stencil._permute_type[idx];
+    if ( local && perm ) {
       spProjXp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
@@ -154,18 +173,17 @@ PARALLEL_FOR_LOOP
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Xp),&chi());
-    //prefetch(Umu._odata[ssu](Yp));
+    mult(&Uchi(),&Umu._odata[ssu](idx),&chi());
     spReconXp(result,Uchi);
 
     // Yp
-    offset = Stencil._offsets [Yp][ss];
-    local  = Stencil._is_local[Yp][ss];
-    perm   = Stencil._permute[Yp][ss];
-    ptype  = Stencil._permute_type[Yp];
+    idx = (Yp+dag*4)%8;
+    offset = Stencil._offsets [idx][ss];
+    local  = Stencil._is_local[idx][ss];
+    perm   = Stencil._permute[idx][ss];
+    ptype  = Stencil._permute_type[idx];
 
-    if ( local && perm ) 
-    {
+    if ( local && perm ) {
       spProjYp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
@@ -173,18 +191,16 @@ PARALLEL_FOR_LOOP
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Yp),&chi());
-    //    prefetch(Umu._odata[ssu](Zp));
+    mult(&Uchi(),&Umu._odata[ssu](idx),&chi());
     accumReconYp(result,Uchi);
 
     // Zp
-    offset = Stencil._offsets [Zp][ss];
-    local  = Stencil._is_local[Zp][ss];
-    perm   = Stencil._permute[Zp][ss];
-    ptype  = Stencil._permute_type[Zp];
-
-    if ( local && perm ) 
-    {
+    idx = (Zp+dag*4)%8;
+    offset = Stencil._offsets [idx][ss];
+    local  = Stencil._is_local[idx][ss];
+    perm   = Stencil._permute[idx][ss];
+    ptype  = Stencil._permute_type[idx];
+    if ( local && perm ) {
       spProjZp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
@@ -192,18 +208,16 @@ PARALLEL_FOR_LOOP
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Zp),&chi());
-    //    prefetch(Umu._odata[ssu](Tp));
+    mult(&Uchi(),&Umu._odata[ssu](idx),&chi());
     accumReconZp(result,Uchi);
 
     // Tp
-    offset = Stencil._offsets [Tp][ss];
-    local  = Stencil._is_local[Tp][ss];
-    perm   = Stencil._permute[Tp][ss];
-    ptype  = Stencil._permute_type[Tp];
-
-    if ( local && perm ) 
-    {
+    idx = (Tp+dag*4)%8;
+    offset = Stencil._offsets [idx][ss];
+    local  = Stencil._is_local[idx][ss];
+    perm   = Stencil._permute[idx][ss];
+    ptype  = Stencil._permute_type[idx];
+    if ( local && perm ) {
       spProjTp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
@@ -211,15 +225,15 @@ PARALLEL_FOR_LOOP
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Tp),&chi());
-    //    prefetch(Umu._odata[ssu](Xm));
+    mult(&Uchi(),&Umu._odata[ssu](idx),&chi());
     accumReconTp(result,Uchi);
 
     // Xm
-    offset = Stencil._offsets [Xm][ss];
-    local  = Stencil._is_local[Xm][ss];
-    perm   = Stencil._permute[Xm][ss];
-    ptype  = Stencil._permute_type[Xm];
+    idx = (Xm+dag*4)%8;
+    offset = Stencil._offsets [idx][ss];
+    local  = Stencil._is_local[idx][ss];
+    perm   = Stencil._permute[idx][ss];
+    ptype  = Stencil._permute_type[idx];
 
     if ( local && perm ) 
     {
@@ -230,18 +244,18 @@ PARALLEL_FOR_LOOP
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Xm),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](idx),&chi());
     accumReconXm(result,Uchi);
 
 
     // Ym
-    offset = Stencil._offsets [Ym][ss];
-    local  = Stencil._is_local[Ym][ss];
-    perm   = Stencil._permute[Ym][ss];
-    ptype  = Stencil._permute_type[Ym];
+    idx = (Ym+dag*4)%8;
+    offset = Stencil._offsets [idx][ss];
+    local  = Stencil._is_local[idx][ss];
+    perm   = Stencil._permute[idx][ss];
+    ptype  = Stencil._permute_type[idx];
 
-    if ( local && perm ) 
-    {
+    if ( local && perm ) {
       spProjYm(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
@@ -249,17 +263,16 @@ PARALLEL_FOR_LOOP
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Ym),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](idx),&chi());
     accumReconYm(result,Uchi);
 
     // Zm
-    offset = Stencil._offsets [Zm][ss];
-    local  = Stencil._is_local[Zm][ss];
-    perm   = Stencil._permute[Zm][ss];
-    ptype  = Stencil._permute_type[Zm];
-
-    if ( local && perm ) 
-    {
+    idx = (Zm+dag*4)%8;
+    offset = Stencil._offsets [idx][ss];
+    local  = Stencil._is_local[idx][ss];
+    perm   = Stencil._permute[idx][ss];
+    ptype  = Stencil._permute_type[idx];
+    if ( local && perm ) {
       spProjZm(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
@@ -267,17 +280,16 @@ PARALLEL_FOR_LOOP
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Zm),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](idx),&chi());
     accumReconZm(result,Uchi);
 
     // Tm
-    offset = Stencil._offsets [Tm][ss];
-    local  = Stencil._is_local[Tm][ss];
-    perm   = Stencil._permute[Tm][ss];
-    ptype  = Stencil._permute_type[Tm];
-
-    if ( local && perm ) 
-    {
+    idx = (Tm+dag*4)%8;
+    offset = Stencil._offsets [idx][ss];
+    local  = Stencil._is_local[idx][ss];
+    perm   = Stencil._permute[idx][ss];
+    ptype  = Stencil._permute_type[idx];
+    if ( local && perm ) {
       spProjTm(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
@@ -285,7 +297,7 @@ PARALLEL_FOR_LOOP
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Tm),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](idx),&chi());
     accumReconTm(result,Uchi);
 
     vstream(out._odata[ss],result);
@@ -294,10 +306,6 @@ PARALLEL_FOR_LOOP
 }
 
 
-void WilsonMatrix::Dw(const LatticeFermion &in, LatticeFermion &out)
-{
-  return;
-}
 
 
 
diff --git a/lib/qcd/Grid_qcd_wilson_dop.h b/lib/qcd/Grid_qcd_wilson_dop.h
index 1098d330..1db95b33 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.h
+++ b/lib/qcd/Grid_qcd_wilson_dop.h
@@ -45,10 +45,7 @@ namespace Grid {
       virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
 
       // non-hermitian hopping term; half cb or both
-      void Dhop(const LatticeFermion &in, LatticeFermion &out);
-
-      // m+4r -1/2 Dhop; both cb's
-      void Dw(const LatticeFermion &in, LatticeFermion &out);
+      void Dhop(const LatticeFermion &in, LatticeFermion &out,int dag);
 
       typedef iScalar<iMatrix<vComplex, Nc> > matrix;
 
diff --git a/lib/simd/Grid_vComplexD.h b/lib/simd/Grid_vComplexD.h
index 41c8509f..2b03c825 100644
--- a/lib/simd/Grid_vComplexD.h
+++ b/lib/simd/Grid_vComplexD.h
@@ -32,7 +32,7 @@ namespace Grid {
         friend inline void mult(vComplexD * __restrict__ y,const vComplexD * __restrict__ l,const vComplexD *__restrict__ r) {*y = (*l) * (*r);}
         friend inline void sub (vComplexD * __restrict__ y,const vComplexD * __restrict__ l,const vComplexD *__restrict__ r) {*y = (*l) - (*r);}
         friend inline void add (vComplexD * __restrict__ y,const vComplexD * __restrict__ l,const vComplexD *__restrict__ r) {*y = (*l) + (*r);}
-        friend inline vComplexD adj(const vComplexD &in){ return conj(in); }
+        friend inline vComplexD adj(const vComplexD &in){ return conjugate(in); }
 
         //////////////////////////////////
         // Initialise to 1,0,i
@@ -166,11 +166,11 @@ namespace Grid {
 	// all subtypes; may not be a good assumption, but could
 	// add the vector width as a template param for BG/Q for example
 	////////////////////////////////////////////////////////////////////
-	/*
 	friend inline void permute(vComplexD &y,vComplexD b,int perm)
 	{
 	  Gpermute<vComplexD>(y,b,perm);
 	}
+	/*
 	friend inline void merge(vComplexD &y,std::vector<ComplexD *> &extracted)
 	{
 	  Gmerge<vComplexD,ComplexD >(y,extracted);
@@ -269,7 +269,7 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
         ////////////////////////
         // Conjugate
         ////////////////////////
-        friend inline vComplexD conj(const vComplexD &in){
+        friend inline vComplexD conjugate(const vComplexD &in){
             vComplexD ret ; vzero(ret);
 #if defined (AVX1)|| defined (AVX2)
 	    //	    addsubps 0, inv=>0+in.v[3] 0-in.v[2], 0+in.v[1], 0-in.v[0], ...
@@ -345,17 +345,17 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
 // REDUCE FIXME must be a cleaner implementation
        friend inline ComplexD Reduce(const vComplexD & in)
        { 
-#if defined SSE4
-         return ComplexD(in.v[0], in.v[1]); // inefficient
+#ifdef SSE4
+	 return ComplexD(in.v[0],in.v[1]);
+#endif
+#if defined(AVX1) || defined (AVX2)
+	 vComplexD v1;
+	 permute(v1,in,0); // sse 128; paired complex single
+	 v1=v1+in;
+	 return ComplexD(v1.v[0],v1.v[1]);
 #endif
-#if defined (AVX1) || defined(AVX2)
-	 //            return std::complex<double>(_mm256_mask_reduce_add_pd(0x55, in.v),_mm256_mask_reduce_add_pd(0xAA, in.v));
-	 __attribute__ ((aligned(32))) double c_[4];
-         _mm256_store_pd(c_,in.v);
-	 return ComplexD(c_[0]+c_[2],c_[1]+c_[3]);
-#endif 
 #ifdef AVX512
-            return ComplexD(_mm512_mask_reduce_add_pd(0x55, in.v),_mm512_mask_reduce_add_pd(0xAA, in.v));
+	 return ComplexD(_mm512_mask_reduce_add_pd(0x55, in.v),_mm512_mask_reduce_add_pd(0xAA, in.v));
 #endif 
 #ifdef QPX
 #endif
@@ -387,7 +387,7 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
     };
 
 
-    inline vComplexD innerProduct(const vComplexD & l, const vComplexD & r) { return conj(l)*r; }
+    inline vComplexD innerProduct(const vComplexD & l, const vComplexD & r) { return conjugate(l)*r; }
 
 
     typedef  vComplexD vDComplex;
diff --git a/lib/simd/Grid_vComplexF.h b/lib/simd/Grid_vComplexF.h
index 8a892d24..713bbdd8 100644
--- a/lib/simd/Grid_vComplexF.h
+++ b/lib/simd/Grid_vComplexF.h
@@ -47,7 +47,7 @@ namespace Grid {
         friend inline void mult(vComplexF * __restrict__ y,const vComplexF * __restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) * (*r); }
         friend inline void sub (vComplexF * __restrict__ y,const vComplexF * __restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) - (*r); }
         friend inline void add (vComplexF * __restrict__ y,const vComplexF * __restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) + (*r); }
-        friend inline vComplexF adj(const vComplexF &in){ return conj(in); }
+        friend inline vComplexF adj(const vComplexF &in){ return conjugate(in); }
         
         //////////////////////////////////
         // Initialise to 1,0,i
@@ -228,42 +228,25 @@ namespace Grid {
             ret.v = {a,b,a,b};
 #endif
         }
-       friend inline ComplexF Reduce(const vComplexF & in)
-       {
+	friend inline void permute(vComplexF &y,vComplexF b,int perm)
+	{
+	  Gpermute<vComplexF>(y,b,perm);
+	}
+	friend inline ComplexF Reduce(const vComplexF & in)
+	{
 #ifdef SSE4
-	   union {
-	     cvec v1;    // SSE 4 x float vector
-	     float f[4];  // scalar array of 4 floats
-	   } u128;
-	   u128.v1= _mm_add_ps(in.v, _mm_shuffle_ps(in.v,in.v, 0b01001110)); // FIXME Prefer to use _MM_SHUFFLE macros
-	   return ComplexF(u128.f[0], u128.f[1]);
+	 vComplexF v1;
+	 permute(v1,in,0); // sse 128; paired complex single
+	 v1=v1+in;
+	 return ComplexF(v1.v[0],v1.v[1]);
 #endif
-#ifdef AVX1
-	   //it would be better passing 2 arguments to saturate the vector lanes
-	   union {
-	     __m256 v1;    
-	     float f[8];  
-	   } u256;
-	   //SWAP lanes
-	   // FIXME .. icc complains with lib/lattice/Grid_lattice_reduction.h (49): (col. 20) warning #13211: Immediate parameter to intrinsic call too large
-	   __m256 t0 = _mm256_permute2f128_ps(in.v, in.v, 1);
-	   __m256 t1 = _mm256_permute_ps(in.v  , 0b11011000);//real (0,2,1,3)
-	   __m256 t2 = _mm256_permute_ps(t0 , 0b10001101);//imag (1,3,0,2)
-	   t0 = _mm256_blend_ps(t1, t2, 0b0101000001010000);// (0,0,1,1,0,0,1,1)
-	   t1 = _mm256_hadd_ps(t0,t0);
-	   u256.v1 = _mm256_hadd_ps(t1, t1);
-	   return ComplexF(u256.f[0], u256.f[4]);
-#endif
-#ifdef AVX2
-	   union {
-	     __m256 v1;    
-	     float f[8];  
-	   } u256;
-	   const __m256i mask= _mm256_set_epi32( 7, 5, 3, 1, 6, 4, 2, 0);
-	   __m256 tmp1 = _mm256_permutevar8x32_ps(in.v, mask);
-	   __m256 tmp2 = _mm256_hadd_ps(tmp1, tmp1);
-	   u256.v1 = _mm256_hadd_ps(tmp2, tmp2);
-	   return ComplexF(u256.f[0], u256.f[4]);
+#if defined(AVX1) || defined (AVX2)
+	 vComplexF v1,v2;
+	 permute(v1,in,0); // sse 128; paired complex single
+	 v1=v1+in;
+	 permute(v2,v1,1); // avx 256; quad complex single
+	 v1=v1+v2;
+	 return ComplexF(v1.v[0],v1.v[1]);
 #endif
 #ifdef AVX512
             return ComplexF(_mm512_mask_reduce_add_ps(0x5555, in.v),_mm512_mask_reduce_add_ps(0xAAAA, in.v));
@@ -345,13 +328,10 @@ namespace Grid {
         // Conjugate
         ///////////////////////
 								     
-        friend inline vComplexF conj(const vComplexF &in){
+        friend inline vComplexF conjugate(const vComplexF &in){
             vComplexF ret ; vzero(ret);
 #if defined (AVX1)|| defined (AVX2)
-	    cvec tmp;
-	    tmp = _mm256_addsub_ps(ret.v,_mm256_shuffle_ps(in.v,in.v,_MM_SHUFFLE(2,3,0,1))); // ymm1 <- br,bi
-	    ret.v=_mm256_shuffle_ps(tmp,tmp,_MM_SHUFFLE(2,3,0,1));
-
+	    ret.v = _mm256_xor_ps(_mm256_addsub_ps(ret.v,in.v), _mm256_set1_ps(-0.f));
 #endif
 #ifdef SSE4
 	    ret.v = _mm_xor_ps(_mm_addsub_ps(ret.v,in.v), _mm_set1_ps(-0.f));
@@ -433,10 +413,6 @@ namespace Grid {
             return *this;
         }
 
-      friend inline void permute(vComplexF &y,vComplexF b,int perm)
-      {
-	Gpermute<vComplexF>(y,b,perm);
-      }
       /*
       friend inline void merge(vComplexF &y,std::vector<ComplexF *> &extracted)
       {
@@ -460,7 +436,7 @@ namespace Grid {
 
     inline vComplexF innerProduct(const vComplexF & l, const vComplexF & r) 
     {
-      return conj(l)*r; 
+      return conjugate(l)*r; 
     }
 
     inline void zeroit(vComplexF &z){ vzero(z);}
diff --git a/lib/simd/Grid_vInteger.h b/lib/simd/Grid_vInteger.h
index 63e8c720..ee4a3633 100644
--- a/lib/simd/Grid_vInteger.h
+++ b/lib/simd/Grid_vInteger.h
@@ -117,7 +117,7 @@ namespace Grid {
         };
         
         ///////////////////////////////////////////////
-        // mult, sub, add, adj,conj, mac functions
+        // mult, sub, add, adj,conjugate, mac functions
         ///////////////////////////////////////////////
         friend inline void mult(vInteger * __restrict__ y,const vInteger * __restrict__ l,const vInteger *__restrict__ r) {*y = (*l) * (*r);}
         friend inline void sub (vInteger * __restrict__ y,const vInteger * __restrict__ l,const vInteger *__restrict__ r) {*y = (*l) - (*r);}
diff --git a/lib/simd/Grid_vRealD.h b/lib/simd/Grid_vRealD.h
index dbfacb0c..83979d14 100644
--- a/lib/simd/Grid_vRealD.h
+++ b/lib/simd/Grid_vRealD.h
@@ -26,7 +26,7 @@ namespace Grid {
         friend inline void sub (vRealD * __restrict__ y,const vRealD * __restrict__ l,const vRealD *__restrict__ r) {*y = (*l) - (*r);}
         friend inline void add (vRealD * __restrict__ y,const vRealD * __restrict__ l,const vRealD *__restrict__ r) {*y = (*l) + (*r);}
         friend inline vRealD adj(const vRealD &in) { return in; }
-        friend inline vRealD conj(const vRealD &in){ return in; }
+        friend inline vRealD conjugate(const vRealD &in){ return in; }
         
         friend inline void mac (vRealD &y,const vRealD a,const vRealD x){
 #if defined (AVX1) || defined (SSE4)
@@ -112,11 +112,12 @@ namespace Grid {
 	// all subtypes; may not be a good assumption, but could
 	// add the vector width as a template param for BG/Q for example
 	////////////////////////////////////////////////////////////////////
-	/*
+
 	friend inline void permute(vRealD &y,vRealD b,int perm)
 	{
 	  Gpermute<vRealD>(y,b,perm);
 	}
+	/*
 	friend inline void merge(vRealD &y,std::vector<RealD *> &extracted)
 	{
 	  Gmerge<vRealD,RealD >(y,extracted);
@@ -209,48 +210,26 @@ namespace Grid {
 
        friend inline RealD Reduce(const vRealD & in)
        {
-#if defined (SSE4)
-	 // FIXME Hack
-	 const RealD * ptr =(const RealD *)  &in;
-	 RealD ret = 0; 
-	 for(int i=0;i<vRealD::Nsimd();i++){
-	   ret = ret+ptr[i];
-	 }
-	 return ret;
+#ifdef SSE4
+	 vRealD v1;
+	 permute(v1,in,0); // sse 128; paired real double
+	 v1=v1+in;
+	 return RealD(v1.v[0]);
 #endif
-#if defined (AVX1) || defined(AVX2)
-	 typedef union  {
-	   uint64_t l;
-	   double   d;
-	 } my_conv_t;
-	 my_conv_t converter;
-// more reduce_add
-/*
-            __attribute__ ((aligned(32))) double c_[16];
-	    __m256d tmp  = _mm256_permute2f128_pd(in.v,in.v,0x01); // tmp 1032; in= 3210
-            __m256d hadd = _mm256_hadd_pd(in.v,tmp);              // hadd = 1+0,3+2,3+2,1+0
-  	             tmp = _mm256_permute2f128_pd(hadd,hadd,0x01);// tmp  = 3+2,1+0,1+0,3+2
-                    hadd = _mm256_hadd_pd(tmp,tmp);               // tmp  = 3+2+1+0,3+2+1+0,1+0+3+2,1+0+3+2
-                    _mm256_store_pd(c_,hadd);ô
-             return c[0]
-*/
-	    __m256d tmp  = _mm256_permute2f128_pd(in.v,in.v,0x01); // tmp 1032; in= 3210
-            __m256d hadd = _mm256_hadd_pd(in.v,tmp);              // hadd = 1+0,3+2,3+2,1+0
-                    hadd = _mm256_hadd_pd(hadd,hadd);             // hadd = 1+0+3+2...
-		    converter.l = _mm256_extract_epi64((ivec)hadd,0);
-            return converter.d;
+#if defined(AVX1) || defined (AVX2)
+	 vRealD v1,v2;
+	 permute(v1,in,0); // avx 256; quad double
+	 v1=v1+in;
+	 permute(v2,v1,1); 
+	 v1=v1+v2;
+	 return v1.v[0];
 #endif
 #ifdef AVX512
             return _mm512_reduce_add_pd(in.v);
-/*
-            __attribute__ ((aligned(32))) double c_[8];
-           _mm512_store_pd(c_,in.v);
-            return c_[0]+c_[1]+c_[2]+c_[3]+c_[4]+c_[5]+c_[6]+c_[7];
-*/
 #endif
 #ifdef QPX
 #endif
-        }
+       }
 
         // *=,+=,-= operators
         inline vRealD &operator *=(const vRealD &r) {
@@ -270,7 +249,7 @@ namespace Grid {
         static int Nsimd(void) { return sizeof(dvec)/sizeof(double);}
     };
 
-    inline vRealD innerProduct(const vRealD & l, const vRealD & r) { return conj(l)*r; }
+    inline vRealD innerProduct(const vRealD & l, const vRealD & r) { return conjugate(l)*r; }
     inline void zeroit(vRealD &z){ vzero(z);}
 
     inline vRealD outerProduct(const vRealD &l, const vRealD& r)
diff --git a/lib/simd/Grid_vRealF.h b/lib/simd/Grid_vRealF.h
index 559a9eda..754c5121 100644
--- a/lib/simd/Grid_vRealF.h
+++ b/lib/simd/Grid_vRealF.h
@@ -92,13 +92,13 @@ namespace Grid {
         };
         
         ///////////////////////////////////////////////
-        // mult, sub, add, adj,conj, mac functions
+        // mult, sub, add, adj,conjugate, mac functions
         ///////////////////////////////////////////////
         friend inline void mult(vRealF * __restrict__ y,const vRealF * __restrict__ l,const vRealF *__restrict__ r) {*y = (*l) * (*r);}
         friend inline void sub (vRealF * __restrict__ y,const vRealF * __restrict__ l,const vRealF *__restrict__ r) {*y = (*l) - (*r);}
         friend inline void add (vRealF * __restrict__ y,const vRealF * __restrict__ l,const vRealF *__restrict__ r) {*y = (*l) + (*r);}
         friend inline vRealF adj(const vRealF &in) { return in; }
-        friend inline vRealF conj(const vRealF &in){ return in; }
+        friend inline vRealF conjugate(const vRealF &in){ return in; }
 
         friend inline void mac (vRealF &y,const vRealF a,const vRealF x){
 #if defined (AVX1) || defined (SSE4)
@@ -133,11 +133,12 @@ namespace Grid {
 	// all subtypes; may not be a good assumption, but could
 	// add the vector width as a template param for BG/Q for example
 	////////////////////////////////////////////////////////////////////
-	/*
+
 	friend inline void permute(vRealF &y,vRealF b,int perm)
 	{
 	  Gpermute<vRealF>(y,b,perm);
 	}
+	/*
 	friend inline void merge(vRealF &y,std::vector<RealF *> &extracted)
 	{
 	  Gmerge<vRealF,RealF >(y,extracted);
@@ -155,7 +156,6 @@ namespace Grid {
 	  Gextract<vRealF,RealF>(y,extracted);
 	}
 	*/
-
         
         /////////////////////////////////////////////////////
         // Broadcast a value across Nsimd copies.
@@ -243,33 +243,26 @@ friend inline void vstore(const vRealF &ret, float *a){
         }
        friend inline RealF Reduce(const vRealF & in)
        {
-#if defined (SSE4)
-	 // FIXME Hack
-	 const RealF * ptr = (const RealF *) &in;
-	 RealF ret = 0; 
-	 for(int i=0;i<vRealF::Nsimd();i++){
-	   ret = ret+ptr[i];
-	 }
-	 return ret;
+#ifdef SSE4
+	 vRealF v1,v2;
+	 permute(v1,in,0); // sse 128; quad single
+	 v1=v1+in;
+	 permute(v2,v1,1); 
+	 v1=v1+v2;
+	 return v1.v[0];
 #endif
-#if defined (AVX1) || defined(AVX2)
-            __attribute__ ((aligned(32))) float c_[16];
-            __m256 tmp = _mm256_permute2f128_ps(in.v,in.v,0x01);
-            __m256 hadd = _mm256_hadd_ps(in.v,tmp);
-                   tmp = _mm256_permute2f128_ps(hadd,hadd,0x01);
-                   hadd = _mm256_hadd_ps(tmp,tmp);
-                  _mm256_store_ps(c_,hadd);
-         return (float)c_[0];
-
+#if defined(AVX1) || defined (AVX2)
+	 vRealF v1,v2;
+	 permute(v1,in,0); // avx 256; octo-double
+	 v1=v1+in;
+	 permute(v2,v1,1); 
+	 v1=v1+v2;
+	 permute(v2,v1,2); 
+	 v1=v1+v2;
+	 return v1.v[0];
 #endif
 #ifdef AVX512
             return _mm512_reduce_add_ps(in.v);
-/*
-             __attribute__ ((aligned(64))) float c_[16];
-             _mm512_store_ps(c_,in.v);
-             return c_[0]+c_[1]+c_[2]+c_[3]+c_[4]+c_[5]+c_[6]+c_[7]
-                    +c_[8]+c_[9]+c_[10]+c_[11]+c_[12]+c_[13]+c_[14]+c_[15];
-*/
 #endif
 #ifdef QPX
 #endif
@@ -291,7 +284,7 @@ friend inline void vstore(const vRealF &ret, float *a){
     public:
         static inline int Nsimd(void) { return sizeof(fvec)/sizeof(float);}
     };
-    inline vRealF innerProduct(const vRealF & l, const vRealF & r) { return conj(l)*r; }
+    inline vRealF innerProduct(const vRealF & l, const vRealF & r) { return conjugate(l)*r; }
     inline void  zeroit(vRealF &z){ vzero(z);}
 
     inline vRealF outerProduct(const vRealF &l, const vRealF& r)
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index 14b2ddf0..4576abe3 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -79,7 +79,7 @@ namespace Grid {
     friend inline void mult(Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) * (*r); }
     friend inline void sub (Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) - (*r); }
     friend inline void add (Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) + (*r); }
-    friend inline Grid_simd adj(const Grid_simd &in){ return conj(in); }
+    friend inline Grid_simd adj(const Grid_simd &in){ return conjugate(in); }
         
     //////////////////////////////////
     // Initialise to 1,0,i
@@ -193,7 +193,7 @@ namespace Grid {
     // Conjugate
     ///////////////////////
 								     
-    friend inline Grid_simd  conj(const Grid_simd  &in){
+    friend inline Grid_simd  conjugate(const Grid_simd  &in){
       Grid_simd  ret ; vzero(ret);
       // FIXME add operator
       return ret;
@@ -265,7 +265,7 @@ namespace Grid {
   template<class scalar_type, class vector_type > 
     inline Grid_simd< scalar_type, vector_type>  innerProduct(const Grid_simd< scalar_type, vector_type> & l, const Grid_simd< scalar_type, vector_type> & r) 
   {
-    return conj(l)*r; 
+    return conjugate(l)*r; 
   }
 
   template<class scalar_type, class vector_type >
diff --git a/scripts/build-all b/scripts/build-all
index d3a978b7..b97dca19 100755
--- a/scripts/build-all
+++ b/scripts/build-all
@@ -1,6 +1,6 @@
-#!/bin/bash 
+#!/bin/bash  -e
 
-DIRS="clang-avx clang-avx-openmp clang-avx-openmp-mpi clang-avx-mpi clang-avx2 clang-avx2-openmp clang-avx2-openmp-mpi clang-avx2-mpi"
+DIRS="clang-avx clang-avx-openmp clang-avx-openmp-mpi clang-avx-mpi clang-avx2 clang-avx2-openmp clang-avx2-openmp-mpi clang-avx2-mpi clang-sse"
 EXTRADIRS="g++-avx g++-sse4 icpc-avx icpc-avx2 icpc-avx512"
 BLACK="\033[30m"
 RED="\033[31m"
diff --git a/scripts/configure-all b/scripts/configure-all
index 2f80b60a..f42b1960 100755
--- a/scripts/configure-all
+++ b/scripts/configure-all
@@ -1,6 +1,6 @@
 #!/bin/bash
 
-DIRS="clang-avx clang-avx-openmp clang-avx-openmp-mpi clang-avx-mpi clang-avx2 clang-avx2-openmp clang-avx2-openmp-mpi clang-avx2-mpi icpc-avx icpc-avx2 icpc-avx512 g++-sse4 g++-avx"
+DIRS="clang-avx clang-avx-openmp clang-avx-openmp-mpi clang-avx-mpi clang-avx2 clang-avx2-openmp clang-avx2-openmp-mpi clang-avx2-mpi icpc-avx icpc-avx2 icpc-avx512 g++-sse4 g++-avx clang-sse"
 
 for D in $DIRS
 do
diff --git a/scripts/configure-commands b/scripts/configure-commands
index 89b72294..c7c35e1c 100755
--- a/scripts/configure-commands
+++ b/scripts/configure-commands
@@ -34,6 +34,9 @@ icpc-avx512)
 icpc-mic)
   CXX=icpc ../../configure --host=none --enable-simd=AVX512 CXXFLAGS="-mmic -O3  -std=c++11" LDFLAGS=-mmic LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;
+clang-sse)
+CXX=clang++ ../../configure --enable-simd=SSE4 CXXFLAGS="-msse4 -O3 -std=c++11"  LIBS="-lgmp -lmpfr" --enable-comms=none
+  ;;
 clang-avx)
 CXX=clang++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11"  LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;
@@ -47,16 +50,16 @@ clang-avx2-openmp)
 CXX=clang-omp++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -fopenmp -O3 -std=c++11" LDFLAGS="-fopenmp"  LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;
 clang-avx-openmp-mpi)
-CXX=clang-omp++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp"  LIBS="-lgmp -lmpfr" --enable-comms=mpi
+CXX=clang-omp++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp -lgmp -lmpfr" --enable-comms=mpi
 ;;
 clang-avx2-openmp-mpi)
-CXX=clang-omp++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp"  LIBS="-lgmp -lmpfr" --enable-comms=mpi
+CXX=clang-omp++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp -lgmp -lmpfr" --enable-comms=mpi
 ;;
 clang-avx-mpi)
-CXX=clang++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx "  LIBS="-lgmp -lmpfr" --enable-comms=mpi
+CXX=clang++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -lgmp -lmpfr" --enable-comms=mpi
 ;;
 clang-avx2-mpi)
-CXX=clang++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx "  LIBS="-lgmp -lmpfr" --enable-comms=mpi
+CXX=clang++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -lgmp -lmpfr" --enable-comms=mpi
 ;;
 esac
 echo -e $NORMAL
diff --git a/tests/Grid_main.cc b/tests/Grid_main.cc
index 45778922..1aa4fc43 100644
--- a/tests/Grid_main.cc
+++ b/tests/Grid_main.cc
@@ -140,7 +140,7 @@ int main (int argc, char ** argv)
 
 //     -=,+=,*=,()
 //     add,+,sub,-,mult,mac,*
-//     adj,conj
+//     adj,conjugate
 //     real,imag
 //     transpose,transposeIndex  
 //     trace,traceIndex
@@ -437,7 +437,7 @@ int main (int argc, char ** argv)
         for(int r=0;r<3;r++){
         for(int c=0;c<3;c++){
             diff =shifted1()()(r,c)-shifted2()()(r,c);
-            nn=real(conj(diff)*diff);
+            nn=real(conjugate(diff)*diff);
             if ( nn > 0 )
                 cout<<"Shift fail (shifted1/shifted2-ref) "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "
                     <<shifted1()()(r,c)<<" "<<shifted2()()(r,c)
@@ -451,7 +451,7 @@ int main (int argc, char ** argv)
         for(int r=0;r<3;r++){
         for(int c=0;c<3;c++){
             diff =shifted3()()(r,c)-shifted2()()(r,c);
-            nn=real(conj(diff)*diff);
+            nn=real(conjugate(diff)*diff);
             if ( nn > 0 )
                 cout<<"Shift rb fail (shifted3/shifted2-ref) "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<" "
                 <<shifted3()()(r,c)<<" "<<shifted2()()(r,c)
@@ -468,7 +468,7 @@ int main (int argc, char ** argv)
         for(int r=0;r<Nc;r++){
         for(int c=0;c<Nc;c++){
             diff =foobar2()()(r,c)-foobar1()()(r,c);
-            nrm = nrm + real(conj(diff)*diff);
+            nrm = nrm + real(conjugate(diff)*diff);
         }}
     }}}}
 	if( Fine.IsBoss() ){
diff --git a/tests/Grid_rng_fixed.cc b/tests/Grid_rng_fixed.cc
new file mode 100644
index 00000000..01e3315b
--- /dev/null
+++ b/tests/Grid_rng_fixed.cc
@@ -0,0 +1,52 @@
+#include <Grid.h>
+#include <parallelIO/GridNerscIO.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+     
+  GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+
+  std::vector<int> seeds({1,2,3,4});
+
+  GridSerialRNG            fsRNG;  fsRNG.SeedFixedIntegers(seeds);
+  GridParallelRNG          fpRNG(&Grid);  fpRNG.SeedFixedIntegers(seeds);
+
+  vComplexF tmp; random(fsRNG,tmp);
+  std::cout<<"Random vComplexF (fixed seed)\n"<< tmp<<std::endl;
+  std::cout<<"conjugate(tmp)\n"<< conjugate(tmp)<<std::endl;
+  std::cout<<"conjugate(tmp)*tmp\n"<< conjugate(tmp)*tmp<<std::endl;
+  std::cout<<"innerProduct"<< innerProduct(tmp,tmp)<<std::endl;
+  std::cout<<"Reduce(innerProduct)"<< Reduce(innerProduct(tmp,tmp))<<std::endl;
+
+  SpinMatrix rnd  ; 
+  random(fsRNG,rnd);
+  std::cout<<"Random Spin Matrix (fixed seed)\n"<< rnd<<std::endl;
+
+  SpinVector rv; 
+  random(fsRNG,rv);
+  std::cout<<"Random Spin Vector (fixed seed)\n"<< rv<<std::endl;
+
+  gaussian(fsRNG,rv);
+  std::cout<<"Gaussian Spin Vector (fixed seed)\n"<< rv<<std::endl;
+
+  LatticeColourVector lcv(&Grid);
+
+  LatticeFermion src(&Grid); random(fpRNG,src);
+  std::cout << "src norm : " << norm2(src)<<std::endl;
+  std::cout << "src " << src<<std::endl;
+
+  random(fpRNG,lcv);
+  std::cout<<"Random Lattice Colour Vector (fixed seed)\n"<< lcv<<std::endl;
+
+
+  Grid_finalize();
+}
diff --git a/tests/Grid_simd.cc b/tests/Grid_simd.cc
index 32f4cdfb..db600fe4 100644
--- a/tests/Grid_simd.cc
+++ b/tests/Grid_simd.cc
@@ -26,7 +26,7 @@ public:
 class funcConj {
 public:
   funcConj() {};
-  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = conj(i1);}
+  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = conjugate(i1);}
   std::string name(void) const { return std::string("Conj"); }
 };
 class funcAdj {
@@ -49,6 +49,34 @@ public:
   template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = timesMinusI(i1);}
   std::string name(void) const { return std::string("timesMinusI"); }
 };
+class funcInnerProduct {
+public:
+  funcInnerProduct() {};
+  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = innerProduct(i1,i2);}
+  std::string name(void) const { return std::string("innerProduct"); }
+};
+
+// FIXME still to test:
+//
+//  innerProduct,
+//  norm2, 
+//  Reduce,
+//
+//  mac,mult,sub,add, vone,vzero,vcomplex_i, =zero,
+//  vset,vsplat,vstore,vstream,vload, scalar*vec, vec*scalar
+//  unary -,
+//  *= , -=, +=
+//  outerproduct, 
+//  zeroit
+//  permute
+
+class funcReduce {
+public:
+  funcReduce() {};
+template<class reduce,class vec>    void vfunc(reduce &rr,vec &i1,vec &i2)   const { rr = Reduce(i1);}
+template<class reduce,class scal>   void sfunc(reduce &rr,scal &i1,scal &i2) const { rr = i1;}
+  std::string name(void) const { return std::string("Reduce"); }
+};
 
 template<class scal, class vec,class functor > 
 void Tester(const functor &func)
@@ -96,8 +124,59 @@ void Tester(const functor &func)
       ok++;
     }
   }
-  if ( ok==0 ) std::cout << " OK!" <<std::endl;
+  if ( ok==0 ) {
+    std::cout << " OK!" <<std::endl;
+  }
+  assert(ok==0);
+}
 
+
+template<class reduced,class scal, class vec,class functor > 
+void ReductionTester(const functor &func)
+{
+  GridSerialRNG          sRNG;
+  sRNG.SeedRandomDevice();
+  
+  int Nsimd = vec::Nsimd();
+
+  std::vector<scal> input1(Nsimd);
+  std::vector<scal> input2(Nsimd);
+  reduced result(0);
+  reduced reference(0);
+  reduced tmp;
+
+  std::vector<vec,alignedAllocator<vec> > buf(3);
+  vec & v_input1 = buf[0];
+  vec & v_input2 = buf[1];
+
+
+  for(int i=0;i<Nsimd;i++){
+    random(sRNG,input1[i]);
+    random(sRNG,input2[i]);
+  }
+
+  merge<vec,scal>(v_input1,input1);
+  merge<vec,scal>(v_input2,input2);
+
+  func.template vfunc<reduced,vec>(result,v_input1,v_input2);
+
+  for(int i=0;i<Nsimd;i++) {
+    func.template sfunc<reduced,scal>(tmp,input1[i],input2[i]);
+    reference+=tmp;
+  }
+
+  std::cout << " " << func.name()<<std::endl;
+
+  int ok=0;
+  if ( abs(reference-result)/abs(reference) > 1.0e-6 ){ // rounding is possible for reduce order
+    std::cout<< "*****" << std::endl;
+    std::cout<< abs(reference-result) << " " <<reference<< " " << result<<std::endl;
+    ok++;
+  }
+  if ( ok==0 ) {
+    std::cout << " OK!" <<std::endl;
+  }
+  assert(ok==0);
 }
 
 
@@ -127,6 +206,8 @@ int main (int argc, char ** argv)
   Tester<ComplexF,vComplexF>(funcTimes());
   Tester<ComplexF,vComplexF>(funcConj());
   Tester<ComplexF,vComplexF>(funcAdj());
+  Tester<ComplexF,vComplexF>(funcInnerProduct());
+  ReductionTester<ComplexF,ComplexF,vComplexF>(funcReduce());
 
   std::cout << "==================================="<<  std::endl;
   std::cout << "Testing vComplexD "<<std::endl;
@@ -140,6 +221,8 @@ int main (int argc, char ** argv)
   Tester<ComplexD,vComplexD>(funcTimes());
   Tester<ComplexD,vComplexD>(funcConj());
   Tester<ComplexD,vComplexD>(funcAdj());
+  Tester<ComplexD,vComplexD>(funcInnerProduct());
+  ReductionTester<ComplexD,ComplexD,vComplexD>(funcReduce());
 
   std::cout << "==================================="<<  std::endl;
   std::cout << "Testing vRealF "<<std::endl;
@@ -150,6 +233,9 @@ int main (int argc, char ** argv)
   Tester<RealF,vRealF>(funcMinus());
   Tester<RealF,vRealF>(funcTimes());
   Tester<RealF,vRealF>(funcAdj());
+  Tester<RealF,vRealF>(funcConj());
+  Tester<RealF,vRealF>(funcInnerProduct());
+  ReductionTester<RealF,RealF,vRealF>(funcReduce());
 
   std::cout << "==================================="<<  std::endl;
   std::cout << "Testing vRealD "<<std::endl;
@@ -159,6 +245,9 @@ int main (int argc, char ** argv)
   Tester<RealD,vRealD>(funcMinus());
   Tester<RealD,vRealD>(funcTimes());
   Tester<RealD,vRealD>(funcAdj());
+  Tester<RealD,vRealD>(funcConj());
+  Tester<RealD,vRealD>(funcInnerProduct());
+  ReductionTester<RealD,RealD,vRealD>(funcReduce());
 
   Grid_finalize();
 }
diff --git a/tests/Grid_stencil.cc b/tests/Grid_stencil.cc
index 1fdb7265..9770ed7c 100644
--- a/tests/Grid_stencil.cc
+++ b/tests/Grid_stencil.cc
@@ -93,7 +93,7 @@ int main (int argc, char ** argv)
 	  for(int r=0;r<3;r++){
 	  for(int c=0;c<3;c++){
             diff =check()()(r,c)-bar()()(r,c);
-            double nn=real(conj(diff)*diff);
+            double nn=real(conjugate(diff)*diff);
             if ( nn > 0){
 	      printf("Coor (%d %d %d %d) \t rc %d%d \t %le (%le,%le) %le\n",
 		     coor[0],coor[1],coor[2],coor[3],r,c,
@@ -103,8 +103,8 @@ int main (int argc, char ** argv)
 		     real(bar()()(r,c))
 		     );
 	    }
-	    snrmC=snrmC+real(conj(check()()(r,c))*check()()(r,c));
-	    snrmB=snrmB+real(conj(bar()()(r,c))*bar()()(r,c));
+	    snrmC=snrmC+real(conjugate(check()()(r,c))*check()()(r,c));
+	    snrmB=snrmB+real(conjugate(bar()()(r,c))*bar()()(r,c));
 	    snrm=snrm+nn;
 	  }}
 	 
diff --git a/tests/Makefile.am b/tests/Makefile.am
index 82d1b3be..80f3a34c 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -5,7 +5,7 @@ AM_LDFLAGS = -L$(top_builddir)/lib
 #
 # Test code
 #
-bin_PROGRAMS = Grid_main Grid_stencil Grid_nersc_io Grid_cshift Grid_gamma  Grid_simd Grid_rng Grid_remez
+bin_PROGRAMS = Grid_main Grid_stencil Grid_nersc_io Grid_cshift Grid_gamma  Grid_simd Grid_rng Grid_remez Grid_rng_fixed
 
 Grid_main_SOURCES = Grid_main.cc
 Grid_main_LDADD = -lGrid
@@ -13,6 +13,9 @@ Grid_main_LDADD = -lGrid
 Grid_rng_SOURCES = Grid_rng.cc
 Grid_rng_LDADD = -lGrid
 
+Grid_rng_fixed_SOURCES = Grid_rng_fixed.cc
+Grid_rng_fixed_LDADD = -lGrid
+
 Grid_remez_SOURCES = Grid_remez.cc
 Grid_remez_LDADD = -lGrid
 
diff --git a/tests/test_Grid_jacobi.cc b/tests/test_Grid_jacobi.cc
index 95e808f1..0cf15a6f 100644
--- a/tests/test_Grid_jacobi.cc
+++ b/tests/test_Grid_jacobi.cc
@@ -186,7 +186,7 @@ int main (int argc, char ** argv)
 	  for(int r=0;r<3;r++){
 	  for(int c=0;c<3;c++){
             diff =check._internal._internal[r][c]-bar._internal._internal[r][c];
-            double nn=real(conj(diff)*diff);
+            double nn=real(conjugate(diff)*diff);
             if ( nn > 0 ){
 	      printf("Coor (%d %d %d %d) \t rc %d%d \t %le %le %le\n",
 		     coor[0],coor[1],coor[2],coor[3],r,c,

From 592cec72e28a268cd448b988e3b2fe3b4a55d88f Mon Sep 17 00:00:00 2001
From: azusayamaguchi <ayamaguc@YAMAKAZE.local>
Date: Tue, 19 May 2015 14:54:42 +0100
Subject: [PATCH 202/429] Add messages to get the number of threads for openmp

---
 Makefile.in                         | 44 +++++++++++++--------
 aclocal.m4                          | 61 +++++++++++++++--------------
 benchmarks/Grid_comms.cc            |  2 +
 benchmarks/Grid_memory_bandwidth.cc |  3 ++
 benchmarks/Grid_su3.cc              |  3 ++
 benchmarks/Grid_wilson.cc           |  3 ++
 config.guess                        |  2 +-
 config.sub                          |  2 +-
 configure                           | 13 +++---
 lib/Grid.h                          |  2 +-
 lib/Grid_init.cc                    | 23 +++++++++++
 11 files changed, 104 insertions(+), 54 deletions(-)

diff --git a/Makefile.in b/Makefile.in
index b6894ef6..b9fba168 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -1,7 +1,7 @@
-# Makefile.in generated by automake 1.14.1 from Makefile.am.
+# Makefile.in generated by automake 1.15 from Makefile.am.
 # @configure_input@
 
-# Copyright (C) 1994-2013 Free Software Foundation, Inc.
+# Copyright (C) 1994-2014 Free Software Foundation, Inc.
 
 # This Makefile.in is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -14,7 +14,17 @@
 
 @SET_MAKE@
 VPATH = @srcdir@
-am__is_gnu_make = test -n '$(MAKEFILE_LIST)' && test -n '$(MAKELEVEL)'
+am__is_gnu_make = { \
+  if test -z '$(MAKELEVEL)'; then \
+    false; \
+  elif test -n '$(MAKE_HOST)'; then \
+    true; \
+  elif test -n '$(MAKE_VERSION)' && test -n '$(CURDIR)'; then \
+    true; \
+  else \
+    false; \
+  fi; \
+}
 am__make_running_with_option = \
   case $${target_option-} in \
       ?) ;; \
@@ -79,14 +89,12 @@ build_triplet = @build@
 host_triplet = @host@
 target_triplet = @target@
 subdir = .
-DIST_COMMON = INSTALL NEWS README AUTHORS ChangeLog \
-	$(srcdir)/Makefile.in $(srcdir)/Makefile.am \
-	$(top_srcdir)/configure $(am__configure_deps) COPYING TODO \
-	compile config.guess config.sub depcomp install-sh missing
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/configure.ac
 am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
 	$(ACLOCAL_M4)
+DIST_COMMON = $(srcdir)/Makefile.am $(top_srcdir)/configure \
+	$(am__configure_deps) $(am__DIST_COMMON)
 am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \
  configure.lineno config.status.lineno
 mkinstalldirs = $(install_sh) -d
@@ -149,6 +157,9 @@ ETAGS = etags
 CTAGS = ctags
 CSCOPE = cscope
 DIST_SUBDIRS = $(SUBDIRS)
+am__DIST_COMMON = $(srcdir)/Makefile.in AUTHORS COPYING ChangeLog \
+	INSTALL NEWS README TODO compile config.guess config.sub \
+	depcomp install-sh missing
 DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
 distdir = $(PACKAGE)-$(VERSION)
 top_distdir = $(distdir)
@@ -314,7 +325,6 @@ $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu Makefile'; \
 	$(am__cd) $(top_srcdir) && \
 	  $(AUTOMAKE) --gnu Makefile
-.PRECIOUS: Makefile
 Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
 	@case '$?' in \
 	  *config.status*) \
@@ -521,15 +531,15 @@ dist-xz: distdir
 	$(am__post_remove_distdir)
 
 dist-tarZ: distdir
-	@echo WARNING: "Support for shar distribution archives is" \
-	               "deprecated." >&2
+	@echo WARNING: "Support for distribution archives compressed with" \
+		       "legacy program 'compress' is deprecated." >&2
 	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	tardir=$(distdir) && $(am__tar) | compress -c >$(distdir).tar.Z
 	$(am__post_remove_distdir)
 
 dist-shar: distdir
-	@echo WARNING: "Support for distribution archives compressed with" \
-		       "legacy program 'compress' is deprecated." >&2
+	@echo WARNING: "Support for shar distribution archives is" \
+	               "deprecated." >&2
 	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	shar $(distdir) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).shar.gz
 	$(am__post_remove_distdir)
@@ -565,17 +575,17 @@ distcheck: dist
 	esac
 	chmod -R a-w $(distdir)
 	chmod u+w $(distdir)
-	mkdir $(distdir)/_build $(distdir)/_inst
+	mkdir $(distdir)/_build $(distdir)/_build/sub $(distdir)/_inst
 	chmod a-w $(distdir)
 	test -d $(distdir)/_build || exit 0; \
 	dc_install_base=`$(am__cd) $(distdir)/_inst && pwd | sed -e 's,^[^:\\/]:[\\/],/,'` \
 	  && dc_destdir="$${TMPDIR-/tmp}/am-dc-$$$$/" \
 	  && am__cwd=`pwd` \
-	  && $(am__cd) $(distdir)/_build \
-	  && ../configure \
+	  && $(am__cd) $(distdir)/_build/sub \
+	  && ../../configure \
 	    $(AM_DISTCHECK_CONFIGURE_FLAGS) \
 	    $(DISTCHECK_CONFIGURE_FLAGS) \
-	    --srcdir=.. --prefix="$$dc_install_base" \
+	    --srcdir=../.. --prefix="$$dc_install_base" \
 	  && $(MAKE) $(AM_MAKEFLAGS) \
 	  && $(MAKE) $(AM_MAKEFLAGS) dvi \
 	  && $(MAKE) $(AM_MAKEFLAGS) check \
@@ -749,6 +759,8 @@ uninstall-am:
 	maintainer-clean-generic mostlyclean mostlyclean-generic pdf \
 	pdf-am ps ps-am tags tags-am uninstall uninstall-am
 
+.PRECIOUS: Makefile
+
 
 # Tell versions [3.59,3.63) of GNU make to not export all variables.
 # Otherwise a system limit (for SysV at least) may be exceeded.
diff --git a/aclocal.m4 b/aclocal.m4
index 389763bf..bf79d078 100644
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -1,6 +1,6 @@
-# generated automatically by aclocal 1.14.1 -*- Autoconf -*-
+# generated automatically by aclocal 1.15 -*- Autoconf -*-
 
-# Copyright (C) 1996-2013 Free Software Foundation, Inc.
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -20,7 +20,7 @@ You have another version of autoconf.  It may work, but is not guaranteed to.
 If you have problems, you may need to regenerate the build system entirely.
 To do so, use the procedure documented by the package, typically 'autoreconf'.])])
 
-# Copyright (C) 2002-2013 Free Software Foundation, Inc.
+# Copyright (C) 2002-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -32,10 +32,10 @@ To do so, use the procedure documented by the package, typically 'autoreconf'.])
 # generated from the m4 files accompanying Automake X.Y.
 # (This private macro should not be called outside this file.)
 AC_DEFUN([AM_AUTOMAKE_VERSION],
-[am__api_version='1.14'
+[am__api_version='1.15'
 dnl Some users find AM_AUTOMAKE_VERSION and mistake it for a way to
 dnl require some minimum version.  Point them to the right macro.
-m4_if([$1], [1.14.1], [],
+m4_if([$1], [1.15], [],
       [AC_FATAL([Do not call $0, use AM_INIT_AUTOMAKE([$1]).])])dnl
 ])
 
@@ -51,14 +51,14 @@ m4_define([_AM_AUTOCONF_VERSION], [])
 # Call AM_AUTOMAKE_VERSION and AM_AUTOMAKE_VERSION so they can be traced.
 # This function is AC_REQUIREd by AM_INIT_AUTOMAKE.
 AC_DEFUN([AM_SET_CURRENT_AUTOMAKE_VERSION],
-[AM_AUTOMAKE_VERSION([1.14.1])dnl
+[AM_AUTOMAKE_VERSION([1.15])dnl
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
 _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 
 # AM_AUX_DIR_EXPAND                                         -*- Autoconf -*-
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -103,15 +103,14 @@ _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 # configured tree to be moved without reconfiguration.
 
 AC_DEFUN([AM_AUX_DIR_EXPAND],
-[dnl Rely on autoconf to set up CDPATH properly.
-AC_PREREQ([2.50])dnl
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+[AC_REQUIRE([AC_CONFIG_AUX_DIR_DEFAULT])dnl
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 ])
 
 # AM_CONDITIONAL                                            -*- Autoconf -*-
 
-# Copyright (C) 1997-2013 Free Software Foundation, Inc.
+# Copyright (C) 1997-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -142,7 +141,7 @@ AC_CONFIG_COMMANDS_PRE(
 Usually this means the macro was only invoked conditionally.]])
 fi])])
 
-# Copyright (C) 1999-2013 Free Software Foundation, Inc.
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -333,7 +332,7 @@ _AM_SUBST_NOTMAKE([am__nodep])dnl
 
 # Generate code to set up dependency tracking.              -*- Autoconf -*-
 
-# Copyright (C) 1999-2013 Free Software Foundation, Inc.
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -409,7 +408,7 @@ AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 
 # Do all the work for Automake.                             -*- Autoconf -*-
 
-# Copyright (C) 1996-2013 Free Software Foundation, Inc.
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -499,8 +498,8 @@ AC_REQUIRE([AC_PROG_MKDIR_P])dnl
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 AC_SUBST([mkdir_p], ['$(MKDIR_P)'])
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
 AC_REQUIRE([AC_PROG_AWK])dnl
 AC_REQUIRE([AC_PROG_MAKE_SET])dnl
 AC_REQUIRE([AM_SET_LEADING_DOT])dnl
@@ -573,7 +572,11 @@ to "yes", and re-run configure.
 END
     AC_MSG_ERROR([Your 'rm' program is bad, sorry.])
   fi
-fi])
+fi
+dnl The trailing newline in this macro's definition is deliberate, for
+dnl backward compatibility and to allow trailing 'dnl'-style comments
+dnl after the AM_INIT_AUTOMAKE invocation. See automake bug#16841.
+])
 
 dnl Hook into '_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
 dnl add the conditional right here, as _AC_COMPILER_EXEEXT may be further
@@ -602,7 +605,7 @@ for _am_header in $config_headers :; do
 done
 echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_count])
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -613,7 +616,7 @@ echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_co
 # Define $install_sh.
 AC_DEFUN([AM_PROG_INSTALL_SH],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -623,7 +626,7 @@ if test x"${install_sh}" != xset; then
 fi
 AC_SUBST([install_sh])])
 
-# Copyright (C) 2003-2013 Free Software Foundation, Inc.
+# Copyright (C) 2003-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -644,7 +647,7 @@ AC_SUBST([am__leading_dot])])
 
 # Check to see how 'make' treats includes.	            -*- Autoconf -*-
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -694,7 +697,7 @@ rm -f confinc confmf
 
 # Fake the existence of programs that GNU maintainers use.  -*- Autoconf -*-
 
-# Copyright (C) 1997-2013 Free Software Foundation, Inc.
+# Copyright (C) 1997-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -733,7 +736,7 @@ fi
 
 # Helper functions for option handling.                     -*- Autoconf -*-
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -764,7 +767,7 @@ AC_DEFUN([_AM_IF_OPTION],
 
 # Check to make sure that the build environment is sane.    -*- Autoconf -*-
 
-# Copyright (C) 1996-2013 Free Software Foundation, Inc.
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -845,7 +848,7 @@ AC_CONFIG_COMMANDS_PRE(
 rm -f conftest.file
 ])
 
-# Copyright (C) 2009-2013 Free Software Foundation, Inc.
+# Copyright (C) 2009-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -905,7 +908,7 @@ AC_SUBST([AM_BACKSLASH])dnl
 _AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
 ])
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -933,7 +936,7 @@ fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 AC_SUBST([INSTALL_STRIP_PROGRAM])])
 
-# Copyright (C) 2006-2013 Free Software Foundation, Inc.
+# Copyright (C) 2006-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -952,7 +955,7 @@ AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 
 # Check how to create a tarball.                            -*- Autoconf -*-
 
-# Copyright (C) 2004-2013 Free Software Foundation, Inc.
+# Copyright (C) 2004-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
diff --git a/benchmarks/Grid_comms.cc b/benchmarks/Grid_comms.cc
index 810cfd06..e844c5d3 100644
--- a/benchmarks/Grid_comms.cc
+++ b/benchmarks/Grid_comms.cc
@@ -10,6 +10,8 @@ int main (int argc, char ** argv)
 
   std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexD::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
+  int threads = GridThread::GetThreads();
+  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
 
   int Nloop=10;
   int nmu=0;
diff --git a/benchmarks/Grid_memory_bandwidth.cc b/benchmarks/Grid_memory_bandwidth.cc
index 8ece09e8..fc959c38 100644
--- a/benchmarks/Grid_memory_bandwidth.cc
+++ b/benchmarks/Grid_memory_bandwidth.cc
@@ -16,6 +16,9 @@ int main (int argc, char ** argv)
   std::vector<int> simd_layout = GridDefaultSimd(Nd,vReal::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
 
+  int threads = GridThread::GetThreads();
+  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+  
   std::cout << "===================================================================================================="<<std::endl;
   std::cout << "= Benchmarking fused AXPY bandwidth"<<std::endl;
   std::cout << "===================================================================================================="<<std::endl;
diff --git a/benchmarks/Grid_su3.cc b/benchmarks/Grid_su3.cc
index eb4dc3a1..d64e8706 100644
--- a/benchmarks/Grid_su3.cc
+++ b/benchmarks/Grid_su3.cc
@@ -13,6 +13,9 @@ int main (int argc, char ** argv)
   std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
 
+  int threads = GridThread::GetThreads();
+  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+
   std::cout << "===================================================================================================="<<std::endl;
   std::cout << "= Benchmarking SU3xSU3  x= x*y"<<std::endl;
   std::cout << "===================================================================================================="<<std::endl;
diff --git a/benchmarks/Grid_wilson.cc b/benchmarks/Grid_wilson.cc
index e49d4515..d1d21cf0 100644
--- a/benchmarks/Grid_wilson.cc
+++ b/benchmarks/Grid_wilson.cc
@@ -26,6 +26,9 @@ int main (int argc, char ** argv)
   std::vector<int> mpi_layout  = GridDefaultMpi();
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
 
+  int threads = GridThread::GetThreads();
+  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+
   std::vector<int> seeds({1,2,3,4});
 
   GridParallelRNG          pRNG(&Grid);
diff --git a/config.guess b/config.guess
index 5f6aa02d..a12faba2 120000
--- a/config.guess
+++ b/config.guess
@@ -1 +1 @@
-/usr/share/automake-1.14/config.guess
\ No newline at end of file
+/opt/local/share/automake-1.15/config.guess
\ No newline at end of file
diff --git a/config.sub b/config.sub
index 0abfe18c..e3c9b5ca 120000
--- a/config.sub
+++ b/config.sub
@@ -1 +1 @@
-/usr/share/automake-1.14/config.sub
\ No newline at end of file
+/opt/local/share/automake-1.15/config.sub
\ No newline at end of file
diff --git a/configure b/configure
index e7d9f32f..b42143e2 100755
--- a/configure
+++ b/configure
@@ -2466,7 +2466,7 @@ test -n "$target_alias" &&
     NONENONEs,x,x, &&
   program_prefix=${target_alias}-
 
-am__api_version='1.14'
+am__api_version='1.15'
 
 # Find a good install program.  We prefer a C program (faster),
 # so one script is as good as another.  But avoid the broken or
@@ -2638,8 +2638,8 @@ test "$program_suffix" != NONE &&
 ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
 program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
 
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 
 if test x"${MISSING+set}" != xset; then
   case $am_aux_dir in
@@ -2658,7 +2658,7 @@ else
 $as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
 fi
 
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -2986,8 +2986,8 @@ MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 mkdir_p='$(MKDIR_P)'
 
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
 # Always define AMTAR for backward compatibility.  Yes, it's still used
 # in the wild :-(  We should find a proper way to deprecate it ...
 AMTAR='$${TAR-tar}'
@@ -3046,6 +3046,7 @@ END
 fi
 
 
+
 ac_config_headers="$ac_config_headers lib/Grid_config.h"
 
 # Check whether --enable-silent-rules was given.
diff --git a/lib/Grid.h b/lib/Grid.h
index edd94769..a1ad74c4 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -86,7 +86,7 @@ namespace Grid {
   // Common parsing chores
   std::string GridCmdOptionPayload(char ** begin, char ** end, const std::string & option);
   bool        GridCmdOptionExists(char** begin, char** end, const std::string& option);
-  void        GridParseIntVector(std::string &str,std::vector<int> & vec);
+  std::string GridCmdVectorIntToString(const std::vector<int> & vec);
 
   void GridParseLayout(char **argv,int argc,
 		       std::vector<int> &latt,
diff --git a/lib/Grid_init.cc b/lib/Grid_init.cc
index c804e810..6af0be84 100644
--- a/lib/Grid_init.cc
+++ b/lib/Grid_init.cc
@@ -111,6 +111,11 @@ void GridParseLayout(char **argv,int argc,
 
 }
 
+std::string GridCmdVectorIntToString(const std::vector<int> & vec){
+  std::ostringstream oss;
+  std::copy(vec.begin(), vec.end(),std::ostream_iterator<int>(oss, " "));
+  return oss.str();
+}
   /////////////////////////////////////////////////////////
   /////////////////////////////////////////////////////////
 void Grid_init(int *argc,char ***argv)
@@ -120,6 +125,15 @@ void Grid_init(int *argc,char ***argv)
 #endif
   // Parse command line args.
 
+  if( GridCmdOptionExists(*argv,*argv+*argc,"--help") ){
+    std::cout<<"--help : this message"<<std::endl;
+    std::cout<<"--debug-signals : catch sigsegv and print a blame report"<<std::endl;
+    std::cout<<"--debug-stdout  : print stdout from EVERY node"<<std::endl;    
+    std::cout<<"--decomposition : report on default omp,mpi and simd decomposition"<<std::endl;    
+    std::cout<<"--mpi n.n.n.n   : default MPI decomposition"<<std::endl;    
+    std::cout<<"--omp n         : default number of OMP threads"<<std::endl;    
+    std::cout<<"--grid n.n.n.n  : default Grid size"<<std::endl;    
+  }
   if( GridCmdOptionExists(*argv,*argv+*argc,"--debug-signals") ){
     Grid_debug_handler_init();
   }
@@ -129,6 +143,15 @@ void Grid_init(int *argc,char ***argv)
   GridParseLayout(*argv,*argc,
 		  Grid_default_latt,
 		  Grid_default_mpi);
+  if( GridCmdOptionExists(*argv,*argv+*argc,"--decomposition") ){
+    std::cout<<"Grid Decomposition\n";
+    std::cout<<"\tOpenMP threads : "<<GridThread::GetThreads()<<std::endl;
+    std::cout<<"\tMPI tasks      : "<<GridCmdVectorIntToString(GridDefaultMpi())<<std::endl;
+    std::cout<<"\tvRealF         : "<<sizeof(vRealF)*8    <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vRealF::Nsimd()))<<std::endl;
+    std::cout<<"\tvRealD         : "<<sizeof(vRealD)*8    <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vRealD::Nsimd()))<<std::endl;
+    std::cout<<"\tvComplexF      : "<<sizeof(vComplexF)*8 <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vComplexF::Nsimd()))<<std::endl;
+    std::cout<<"\tvComplexD      : "<<sizeof(vComplexD)*8 <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vComplexD::Nsimd()))<<std::endl;
+  }
 
 }
 

From 46ab8edf3011cde754d2c98816ed39d09caa5f61 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 19 May 2015 15:50:47 +0100
Subject: [PATCH 203/429] Optimisation...

---
 lib/qcd/Grid_qcd_wilson_dop.cc | 265 +++++++++++++++++++++++++--------
 lib/qcd/Grid_qcd_wilson_dop.h  |   2 +
 scripts/bench_wilson.sh        |   9 ++
 scripts/wilson.gnu             |   7 +
 4 files changed, 223 insertions(+), 60 deletions(-)
 create mode 100755 scripts/bench_wilson.sh
 create mode 100644 scripts/wilson.gnu

diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index 5143daf3..9ef4af0c 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -138,16 +138,8 @@ void WilsonMatrix::MooeeInvDag(const LatticeFermion &in, LatticeFermion &out)
   return ;
 }
 
-void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out,int dag)
+void WilsonMatrix::DhopSite(int ss,const LatticeFermion &in, LatticeFermion &out)
 {
-  assert((dag==0) ||(dag==1));
-
-  WilsonCompressor compressor(dag);
-  Stencil.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
-
-PARALLEL_FOR_LOOP
-  for(int sss=0;sss<grid->oSites();sss++){
-
     vHalfSpinColourVector  tmp;    
     vHalfSpinColourVector  chi;    
     vSpinColourVector result;
@@ -155,16 +147,14 @@ PARALLEL_FOR_LOOP
     int offset,local,perm, ptype;
 
     //    int ss = Stencil._LebesgueReorder[sss];
-    int ss = sss;
     int ssu= ss;
 
     // Xp
-    int idx = (Xp+dag*4)%8;
-    offset = Stencil._offsets [idx][ss];
-    local  = Stencil._is_local[idx][ss];
-    perm   = Stencil._permute[idx][ss];
+    offset = Stencil._offsets [Xp][ss];
+    local  = Stencil._is_local[Xp][ss];
+    perm   = Stencil._permute[Xp][ss];
 
-    ptype  = Stencil._permute_type[idx];
+    ptype  = Stencil._permute_type[Xp];
     if ( local && perm ) {
       spProjXp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
@@ -173,16 +163,14 @@ PARALLEL_FOR_LOOP
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](idx),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Xp),&chi());
     spReconXp(result,Uchi);
 
     // Yp
-    idx = (Yp+dag*4)%8;
-    offset = Stencil._offsets [idx][ss];
-    local  = Stencil._is_local[idx][ss];
-    perm   = Stencil._permute[idx][ss];
-    ptype  = Stencil._permute_type[idx];
-
+    offset = Stencil._offsets [Yp][ss];
+    local  = Stencil._is_local[Yp][ss];
+    perm   = Stencil._permute[Yp][ss];
+    ptype  = Stencil._permute_type[Yp];
     if ( local && perm ) {
       spProjYp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
@@ -191,15 +179,14 @@ PARALLEL_FOR_LOOP
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](idx),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Yp),&chi());
     accumReconYp(result,Uchi);
 
     // Zp
-    idx = (Zp+dag*4)%8;
-    offset = Stencil._offsets [idx][ss];
-    local  = Stencil._is_local[idx][ss];
-    perm   = Stencil._permute[idx][ss];
-    ptype  = Stencil._permute_type[idx];
+    offset = Stencil._offsets [Zp][ss];
+    local  = Stencil._is_local[Zp][ss];
+    perm   = Stencil._permute[Zp][ss];
+    ptype  = Stencil._permute_type[Zp];
     if ( local && perm ) {
       spProjZp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
@@ -208,15 +195,14 @@ PARALLEL_FOR_LOOP
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](idx),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Zp),&chi());
     accumReconZp(result,Uchi);
 
     // Tp
-    idx = (Tp+dag*4)%8;
-    offset = Stencil._offsets [idx][ss];
-    local  = Stencil._is_local[idx][ss];
-    perm   = Stencil._permute[idx][ss];
-    ptype  = Stencil._permute_type[idx];
+    offset = Stencil._offsets [Tp][ss];
+    local  = Stencil._is_local[Tp][ss];
+    perm   = Stencil._permute[Tp][ss];
+    ptype  = Stencil._permute_type[Tp];
     if ( local && perm ) {
       spProjTp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
@@ -225,15 +211,14 @@ PARALLEL_FOR_LOOP
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](idx),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Tp),&chi());
     accumReconTp(result,Uchi);
 
     // Xm
-    idx = (Xm+dag*4)%8;
-    offset = Stencil._offsets [idx][ss];
-    local  = Stencil._is_local[idx][ss];
-    perm   = Stencil._permute[idx][ss];
-    ptype  = Stencil._permute_type[idx];
+    offset = Stencil._offsets [Xm][ss];
+    local  = Stencil._is_local[Xm][ss];
+    perm   = Stencil._permute[Xm][ss];
+    ptype  = Stencil._permute_type[Xm];
 
     if ( local && perm ) 
     {
@@ -244,16 +229,15 @@ PARALLEL_FOR_LOOP
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](idx),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Xm),&chi());
     accumReconXm(result,Uchi);
 
 
     // Ym
-    idx = (Ym+dag*4)%8;
-    offset = Stencil._offsets [idx][ss];
-    local  = Stencil._is_local[idx][ss];
-    perm   = Stencil._permute[idx][ss];
-    ptype  = Stencil._permute_type[idx];
+    offset = Stencil._offsets [Ym][ss];
+    local  = Stencil._is_local[Ym][ss];
+    perm   = Stencil._permute[Ym][ss];
+    ptype  = Stencil._permute_type[Ym];
 
     if ( local && perm ) {
       spProjYm(tmp,in._odata[offset]);
@@ -263,15 +247,14 @@ PARALLEL_FOR_LOOP
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](idx),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Ym),&chi());
     accumReconYm(result,Uchi);
 
     // Zm
-    idx = (Zm+dag*4)%8;
-    offset = Stencil._offsets [idx][ss];
-    local  = Stencil._is_local[idx][ss];
-    perm   = Stencil._permute[idx][ss];
-    ptype  = Stencil._permute_type[idx];
+    offset = Stencil._offsets [Zm][ss];
+    local  = Stencil._is_local[Zm][ss];
+    perm   = Stencil._permute[Zm][ss];
+    ptype  = Stencil._permute_type[Zm];
     if ( local && perm ) {
       spProjZm(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
@@ -280,15 +263,14 @@ PARALLEL_FOR_LOOP
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](idx),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Zm),&chi());
     accumReconZm(result,Uchi);
 
     // Tm
-    idx = (Tm+dag*4)%8;
-    offset = Stencil._offsets [idx][ss];
-    local  = Stencil._is_local[idx][ss];
-    perm   = Stencil._permute[idx][ss];
-    ptype  = Stencil._permute_type[idx];
+    offset = Stencil._offsets [Tm][ss];
+    local  = Stencil._is_local[Tm][ss];
+    perm   = Stencil._permute[Tm][ss];
+    ptype  = Stencil._permute_type[Tm];
     if ( local && perm ) {
       spProjTm(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
@@ -297,12 +279,175 @@ PARALLEL_FOR_LOOP
     } else { 
       chi=comm_buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](idx),&chi());
+    mult(&Uchi(),&Umu._odata[ssu](Tm),&chi());
     accumReconTm(result,Uchi);
 
     vstream(out._odata[ss],result);
-  }
+}
+void WilsonMatrix::DhopSiteDag(int ss,const LatticeFermion &in, LatticeFermion &out)
+{
+    vHalfSpinColourVector  tmp;    
+    vHalfSpinColourVector  chi;    
+    vSpinColourVector result;
+    vHalfSpinColourVector Uchi;
+    int offset,local,perm, ptype;
 
+    int ssu= ss;
+
+    // Xp
+    offset = Stencil._offsets [Xm][ss];
+    local  = Stencil._is_local[Xm][ss];
+    perm   = Stencil._permute[Xm][ss];
+
+    ptype  = Stencil._permute_type[Xm];
+    if ( local && perm ) {
+      spProjXp(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjXp(chi,in._odata[offset]);
+    } else { 
+      chi=comm_buf[offset];
+    }
+    mult(&Uchi(),&Umu._odata[ssu](Xm),&chi());
+    spReconXp(result,Uchi);
+
+    // Yp
+    offset = Stencil._offsets [Ym][ss];
+    local  = Stencil._is_local[Ym][ss];
+    perm   = Stencil._permute[Ym][ss];
+    ptype  = Stencil._permute_type[Ym];
+    if ( local && perm ) {
+      spProjYp(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjYp(chi,in._odata[offset]);
+    } else { 
+      chi=comm_buf[offset];
+    }
+    mult(&Uchi(),&Umu._odata[ssu](Ym),&chi());
+    accumReconYp(result,Uchi);
+
+    // Zp
+    offset = Stencil._offsets [Zm][ss];
+    local  = Stencil._is_local[Zm][ss];
+    perm   = Stencil._permute[Zm][ss];
+    ptype  = Stencil._permute_type[Zm];
+    if ( local && perm ) {
+      spProjZp(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjZp(chi,in._odata[offset]);
+    } else { 
+      chi=comm_buf[offset];
+    }
+    mult(&Uchi(),&Umu._odata[ssu](Zm),&chi());
+    accumReconZp(result,Uchi);
+
+    // Tp
+    offset = Stencil._offsets [Tm][ss];
+    local  = Stencil._is_local[Tm][ss];
+    perm   = Stencil._permute[Tm][ss];
+    ptype  = Stencil._permute_type[Tm];
+    if ( local && perm ) {
+      spProjTp(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjTp(chi,in._odata[offset]);
+    } else { 
+      chi=comm_buf[offset];
+    }
+    mult(&Uchi(),&Umu._odata[ssu](Tm),&chi());
+    accumReconTp(result,Uchi);
+
+    // Xm
+    offset = Stencil._offsets [Xp][ss];
+    local  = Stencil._is_local[Xp][ss];
+    perm   = Stencil._permute[Xp][ss];
+    ptype  = Stencil._permute_type[Xp];
+
+    if ( local && perm ) 
+    {
+      spProjXm(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjXm(chi,in._odata[offset]);
+    } else { 
+      chi=comm_buf[offset];
+    }
+    mult(&Uchi(),&Umu._odata[ssu](Xp),&chi());
+    accumReconXm(result,Uchi);
+
+
+    // Ym
+    offset = Stencil._offsets [Yp][ss];
+    local  = Stencil._is_local[Yp][ss];
+    perm   = Stencil._permute[Yp][ss];
+    ptype  = Stencil._permute_type[Yp];
+
+    if ( local && perm ) {
+      spProjYm(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjYm(chi,in._odata[offset]);
+    } else { 
+      chi=comm_buf[offset];
+    }
+    mult(&Uchi(),&Umu._odata[ssu](Yp),&chi());
+    accumReconYm(result,Uchi);
+
+    // Zm
+    offset = Stencil._offsets [Zp][ss];
+    local  = Stencil._is_local[Zp][ss];
+    perm   = Stencil._permute[Zp][ss];
+    ptype  = Stencil._permute_type[Zp];
+    if ( local && perm ) {
+      spProjZm(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjZm(chi,in._odata[offset]);
+    } else { 
+      chi=comm_buf[offset];
+    }
+    mult(&Uchi(),&Umu._odata[ssu](Zp),&chi());
+    accumReconZm(result,Uchi);
+
+    // Tm
+    offset = Stencil._offsets [Tp][ss];
+    local  = Stencil._is_local[Tp][ss];
+    perm   = Stencil._permute[Tp][ss];
+    ptype  = Stencil._permute_type[Tp];
+    if ( local && perm ) {
+      spProjTm(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjTm(chi,in._odata[offset]);
+    } else { 
+      chi=comm_buf[offset];
+    }
+    mult(&Uchi(),&Umu._odata[ssu](Tp),&chi());
+    accumReconTm(result,Uchi);
+
+    vstream(out._odata[ss],result);
+}
+
+void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out,int dag)
+{
+  assert((dag==0) ||(dag==1));
+
+  WilsonCompressor compressor(dag);
+  Stencil.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
+
+  if ( dag ) {
+PARALLEL_FOR_LOOP
+    for(int sss=0;sss<grid->oSites();sss++){
+      DhopSiteDag(sss,in,out);
+    }
+  } else {
+PARALLEL_FOR_LOOP
+    for(int sss=0;sss<grid->oSites();sss++){
+      DhopSite(sss,in,out);
+    }
+  }
 }
 
 
diff --git a/lib/qcd/Grid_qcd_wilson_dop.h b/lib/qcd/Grid_qcd_wilson_dop.h
index 1db95b33..8d93a926 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.h
+++ b/lib/qcd/Grid_qcd_wilson_dop.h
@@ -46,6 +46,8 @@ namespace Grid {
 
       // non-hermitian hopping term; half cb or both
       void Dhop(const LatticeFermion &in, LatticeFermion &out,int dag);
+      void DhopSite   (int ss,const LatticeFermion &in, LatticeFermion &out);
+      void DhopSiteDag(int ss,const LatticeFermion &in, LatticeFermion &out);
 
       typedef iScalar<iMatrix<vComplex, Nc> > matrix;
 
diff --git a/scripts/bench_wilson.sh b/scripts/bench_wilson.sh
new file mode 100755
index 00000000..af73d591
--- /dev/null
+++ b/scripts/bench_wilson.sh
@@ -0,0 +1,9 @@
+for omp in 1 2 4
+do
+echo > wilson.t$omp
+for vol in 4.4.4.4 4.4.4.8 4.4.8.8  4.8.8.8  8.8.8.8   8.8.8.16 8.8.16.16  8.16.16.16
+do   
+perf=` ./benchmarks/Grid_wilson --grid $vol --omp $omp  | grep mflop | awk '{print $3}'`
+echo $vol $perf >> wilson.t$omp
+done
+done
\ No newline at end of file
diff --git a/scripts/wilson.gnu b/scripts/wilson.gnu
new file mode 100644
index 00000000..69bca5b5
--- /dev/null
+++ b/scripts/wilson.gnu
@@ -0,0 +1,7 @@
+plot 'wilson.t1' u 2 w l t "AVX1-OMP=1"
+replot 'wilson.t2' u 2 w l t "AVX1-OMP=2"
+replot 'wilson.t4' u 2 w l t "AVX1-OMP=4"
+set terminal 'pdf'
+set output 'wilson_clang.pdf'
+replot
+quit

From b562b50196a699587a3477e37fcedc4e3f86861c Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 19 May 2015 21:29:07 +0100
Subject: [PATCH 204/429] Reworking to keep intel compiler happy

---
 benchmarks/Makefile.am           |  2 +-
 lib/algorithms/LinearOperator.h  |  2 +-
 lib/algorithms/approx/bigfloat.h |  2 +-
 lib/simd/Grid_vComplexD.h        | 18 ++++++++++++++----
 lib/simd/Grid_vComplexF.h        | 18 +++++++++++++-----
 lib/simd/Grid_vRealD.h           | 18 +++++++++++++-----
 lib/simd/Grid_vRealF.h           | 20 +++++++++++++++-----
 tests/Grid_nersc_io.cc           |  6 ++++--
 8 files changed, 62 insertions(+), 24 deletions(-)

diff --git a/benchmarks/Makefile.am b/benchmarks/Makefile.am
index 77ce7663..ae09a6db 100644
--- a/benchmarks/Makefile.am
+++ b/benchmarks/Makefile.am
@@ -16,7 +16,7 @@ Grid_wilson_cg_unprec_LDADD = -lGrid
 Grid_comms_SOURCES = Grid_comms.cc
 Grid_comms_LDADD = -lGrid
 
-Grid_su3_SOURCES = Grid_su3.cc
+Grid_su3_SOURCES = Grid_su3.cc Grid_su3_test.cc
 Grid_su3_LDADD = -lGrid
 
 Grid_memory_bandwidth_SOURCES = Grid_memory_bandwidth.cc
diff --git a/lib/algorithms/LinearOperator.h b/lib/algorithms/LinearOperator.h
index bdb6c387..167213ec 100644
--- a/lib/algorithms/LinearOperator.h
+++ b/lib/algorithms/LinearOperator.h
@@ -25,7 +25,7 @@ namespace Grid {
   /////////////////////////////////////////////////////////////////////////////////////////////
     template<class Field> class HermitianOperatorBase : public LinearOperatorBase<Field> {
     public:
-      virtual void OpAndNorm(const Field &in, Field &out,double &n1,double &n2);
+      virtual void OpAndNorm(const Field &in, Field &out,double &n1,double &n2)=0;
       void AdjOp(const Field &in, Field &out) {
 	Op(in,out);
       };
diff --git a/lib/algorithms/approx/bigfloat.h b/lib/algorithms/approx/bigfloat.h
index aee71d03..a749fa81 100755
--- a/lib/algorithms/approx/bigfloat.h
+++ b/lib/algorithms/approx/bigfloat.h
@@ -31,7 +31,7 @@ public:
   bigfloat(const double d) { mpf_init_set_d(x, d); }  
   bigfloat(const char *str) { mpf_init_set_str(x, (char*)str, 10); }
   ~bigfloat(void) { mpf_clear(x); }
-  operator const double (void) const { return (double)mpf_get_d(x); }
+  operator double (void) const { return (double)mpf_get_d(x); }
   static void setDefaultPrecision(unsigned long dprec) {
     unsigned long bprec =  (unsigned long)(3.321928094 * (double)dprec);
     mpf_set_default_prec(bprec);
diff --git a/lib/simd/Grid_vComplexD.h b/lib/simd/Grid_vComplexD.h
index 2b03c825..116c2866 100644
--- a/lib/simd/Grid_vComplexD.h
+++ b/lib/simd/Grid_vComplexD.h
@@ -345,20 +345,30 @@ friend inline void vstore(const vComplexD &ret, ComplexD *a){
 // REDUCE FIXME must be a cleaner implementation
        friend inline ComplexD Reduce(const vComplexD & in)
        { 
+	 vComplexD v1,v2;
+	 union { 
+	   zvec v;
+	   double f[sizeof(zvec)/sizeof(double)];
+	 } conv;
+	   
 #ifdef SSE4
-	 return ComplexD(in.v[0],in.v[1]);
+	 v1=in;
 #endif
 #if defined(AVX1) || defined (AVX2)
-	 vComplexD v1;
 	 permute(v1,in,0); // sse 128; paired complex single
 	 v1=v1+in;
-	 return ComplexD(v1.v[0],v1.v[1]);
 #endif
 #ifdef AVX512
-	 return ComplexD(_mm512_mask_reduce_add_pd(0x55, in.v),_mm512_mask_reduce_add_pd(0xAA, in.v));
+	 permute(v1,in,0); // sse 128; paired complex single
+	 v1=v1+in;
+	 permute(v2,v1,1); // avx 256; quad complex single
+	 v1=v1+v2;
 #endif 
 #ifdef QPX
+#error
 #endif
+	 conv.v = v1.v;
+	 return ComplexD(conv.f[0],conv.f[1]);
         }
         
         // Unary negation
diff --git a/lib/simd/Grid_vComplexF.h b/lib/simd/Grid_vComplexF.h
index 713bbdd8..cf74ff03 100644
--- a/lib/simd/Grid_vComplexF.h
+++ b/lib/simd/Grid_vComplexF.h
@@ -234,26 +234,34 @@ namespace Grid {
 	}
 	friend inline ComplexF Reduce(const vComplexF & in)
 	{
+	 vComplexF v1,v2;
+	 union { 
+	   cvec v;
+	   float f[sizeof(cvec)/sizeof(float)];
+	 } conv;
 #ifdef SSE4
-	 vComplexF v1;
 	 permute(v1,in,0); // sse 128; paired complex single
 	 v1=v1+in;
-	 return ComplexF(v1.v[0],v1.v[1]);
 #endif
 #if defined(AVX1) || defined (AVX2)
-	 vComplexF v1,v2;
 	 permute(v1,in,0); // sse 128; paired complex single
 	 v1=v1+in;
 	 permute(v2,v1,1); // avx 256; quad complex single
 	 v1=v1+v2;
-	 return ComplexF(v1.v[0],v1.v[1]);
 #endif
 #ifdef AVX512
-            return ComplexF(_mm512_mask_reduce_add_ps(0x5555, in.v),_mm512_mask_reduce_add_ps(0xAAAA, in.v));
+	 permute(v1,in,0); // avx512 octo-complex single
+	 v1=v1+in;
+	 permute(v2,v1,1); 
+	 v1=v1+v2;
+	 permute(v2,v1,2); 
+	 v1=v1+v2;
 #endif
 #ifdef QPX
 #error
 #endif
+	 conv.v = v1.v;
+	 return ComplexF(conv.f[0],conv.f[1]);
         }
 
         friend inline vComplexF operator * (const ComplexF &a, vComplexF b){
diff --git a/lib/simd/Grid_vRealD.h b/lib/simd/Grid_vRealD.h
index 83979d14..a32e579e 100644
--- a/lib/simd/Grid_vRealD.h
+++ b/lib/simd/Grid_vRealD.h
@@ -210,25 +210,33 @@ namespace Grid {
 
        friend inline RealD Reduce(const vRealD & in)
        {
+	 vRealD v1,v2;
+	 union { 
+	   dvec v;
+	   double f[sizeof(dvec)/sizeof(double)];
+	 } conv;
 #ifdef SSE4
-	 vRealD v1;
 	 permute(v1,in,0); // sse 128; paired real double
 	 v1=v1+in;
-	 return RealD(v1.v[0]);
 #endif
 #if defined(AVX1) || defined (AVX2)
-	 vRealD v1,v2;
 	 permute(v1,in,0); // avx 256; quad double
 	 v1=v1+in;
 	 permute(v2,v1,1); 
 	 v1=v1+v2;
-	 return v1.v[0];
 #endif
 #ifdef AVX512
-            return _mm512_reduce_add_pd(in.v);
+	 permute(v1,in,0); // avx 512; octo-double
+	 v1=v1+in;
+	 permute(v2,v1,1); 
+	 v1=v1+v2;
+	 permute(v2,v1,2); 
+	 v1=v1+v2;
 #endif
 #ifdef QPX
 #endif
+	 conv.v=v1.v;
+	 return conv.f[0];
        }
 
         // *=,+=,-= operators
diff --git a/lib/simd/Grid_vRealF.h b/lib/simd/Grid_vRealF.h
index 754c5121..05bfa2f5 100644
--- a/lib/simd/Grid_vRealF.h
+++ b/lib/simd/Grid_vRealF.h
@@ -243,29 +243,39 @@ friend inline void vstore(const vRealF &ret, float *a){
         }
        friend inline RealF Reduce(const vRealF & in)
        {
-#ifdef SSE4
 	 vRealF v1,v2;
+	 union { 
+	   fvec v;
+	   float f[sizeof(fvec)/sizeof(double)];
+	 } conv;
+#ifdef SSE4
 	 permute(v1,in,0); // sse 128; quad single
 	 v1=v1+in;
 	 permute(v2,v1,1); 
 	 v1=v1+v2;
-	 return v1.v[0];
 #endif
 #if defined(AVX1) || defined (AVX2)
-	 vRealF v1,v2;
 	 permute(v1,in,0); // avx 256; octo-double
 	 v1=v1+in;
 	 permute(v2,v1,1); 
 	 v1=v1+v2;
 	 permute(v2,v1,2); 
 	 v1=v1+v2;
-	 return v1.v[0];
 #endif
 #ifdef AVX512
-            return _mm512_reduce_add_ps(in.v);
+	 permute(v1,in,0); // avx 256; octo-double
+	 v1=v1+in;
+	 permute(v2,v1,1); 
+	 v1=v1+v2;
+	 permute(v2,v1,2); 
+	 v1=v1+v2;
+	 permute(v2,v1,3); 
+	 v1=v1+v2;
 #endif
 #ifdef QPX
 #endif
+	 conv.v=v1.v;
+	 return conv.f[0];
         }
 
         // *=,+=,-= operators
diff --git a/tests/Grid_nersc_io.cc b/tests/Grid_nersc_io.cc
index fbef3cb1..6fe587a6 100644
--- a/tests/Grid_nersc_io.cc
+++ b/tests/Grid_nersc_io.cc
@@ -13,7 +13,7 @@ int main (int argc, char ** argv)
 
   std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
-  std::vector<int> latt_size  ({16,16,16,32});
+  std::vector<int> latt_size   = GridDefaultLatt();
   std::vector<int> clatt_size  ({4,4,4,8});
   int orthodir=3;
   int orthosz =latt_size[orthodir];
@@ -44,13 +44,15 @@ int main (int argc, char ** argv)
   // (1+2+3)=6 = N(N-1)/2 terms
   LatticeComplex Plaq(&Fine);
   LatticeComplex cPlaq(&Coarse);
+
   Plaq = zero;
+#if 1
   for(int mu=1;mu<Nd;mu++){
     for(int nu=0;nu<mu;nu++){
       Plaq = Plaq + trace(U[mu]*Cshift(U[nu],mu,1)*adj(Cshift(U[mu],nu,1))*adj(U[nu]));
     }
   }
-  
+#endif
   double vol = Fine.gSites();
   Complex PlaqScale(1.0/vol/6.0/3.0);
   std::cout <<"PlaqScale" << PlaqScale<<std::endl;

From 3f57662cd0db1048392f507d7603d50c9d872767 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 19 May 2015 21:29:40 +0100
Subject: [PATCH 205/429] Build a simple kernel to compare intel compiler and
 clang in simple environment

---
 benchmarks/Grid_su3_test.cc | 10 ++++++++++
 1 file changed, 10 insertions(+)
 create mode 100644 benchmarks/Grid_su3_test.cc

diff --git a/benchmarks/Grid_su3_test.cc b/benchmarks/Grid_su3_test.cc
new file mode 100644
index 00000000..e4d4ec0a
--- /dev/null
+++ b/benchmarks/Grid_su3_test.cc
@@ -0,0 +1,10 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+void su3_test_mult(LatticeColourMatrix &z, LatticeColourMatrix &x,LatticeColourMatrix &y)
+{
+  z=x*y;
+}

From 3a3f54932aa4eab5f72d914d3e7b19a09fed3669 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Wed, 20 May 2015 17:22:40 +0900
Subject: [PATCH 206/429] Implemented all SSE4 functions.

A test code Grid_simd_new.cc has been created to test the new class.
Tests are all OK.
---
 lib/simd/Grid_sse4.h         | 170 +++++++++++++++++++++++++-----
 lib/simd/Grid_vector_types.h | 194 ++++++++++++++++++++++++-----------
 tests/Grid_main.cc           |  16 ++-
 tests/Grid_simd_new.cc       | 165 +++++++++++++++++++++++++++++
 tests/Makefile.am            |   5 +-
 5 files changed, 458 insertions(+), 92 deletions(-)
 create mode 100644 tests/Grid_simd_new.cc

diff --git a/lib/simd/Grid_sse4.h b/lib/simd/Grid_sse4.h
index ed4039b7..ddc3490b 100644
--- a/lib/simd/Grid_sse4.h
+++ b/lib/simd/Grid_sse4.h
@@ -1,8 +1,10 @@
 //----------------------------------------------------------------------
 /*! @file Grid_sse4.h
-  @brief Optimization libraries
+  @brief Optimization libraries for SSE4 instructions set
+
+  Using intrinsics
 */
-// Time-stamp: <2015-05-19 17:06:51 neo>
+// Time-stamp: <2015-05-20 16:45:39 neo>
 //----------------------------------------------------------------------
 
 #include <pmmintrin.h>
@@ -49,6 +51,20 @@ namespace Optimization {
   };
 
 
+  struct Vstream{
+    //Float
+    inline void operator()(__m128 a, __m128 b){
+      _mm_stream_ps((float *)&a,b);
+    }
+    //Double
+    inline void operator()(__m128d a, __m128d b){
+      _mm_stream_pd((double *)&a,b);
+    }
+
+
+  };
+
+
 
   struct Vset{
     // Complex float 
@@ -75,27 +91,20 @@ namespace Optimization {
 
   };
 
+  template <typename Out_type, typename In_type>
   struct Reduce{
-    //Complex float
-    inline Grid::ComplexF operator()(__m128 in){
-      union {
-	__m128 v1;  
-	float f[4]; 
-      } u128;
-      u128.v1 = _mm_add_ps(in, _mm_shuffle_ps(in,in, 0b01001110)); // FIXME Prefer to use _MM_SHUFFLE macros
-      return Grid::ComplexF(u128.f[0], u128.f[1]);   
+    //Need templated class to overload output type
+    //General form must generate error if compiled
+    inline Out_type operator()(In_type in){
+      printf("Error, using wrong Reduce function\n");
+      exit(1);
+      return 0;
     }
-    //Complex double
-    inline Grid::ComplexD operator()(__m128d in){
-      printf("Missing complex double implementation -> FIX\n");
-      return Grid::ComplexD(0,0); // FIXME wrong
-    }
-
-
-
   };
 
 
+ 
+
   /////////////////////////////////////////////////////
   // Arithmetic operations
   /////////////////////////////////////////////////////
@@ -129,25 +138,26 @@ namespace Optimization {
     }
   };
 
+
   struct MultComplex{
     // Complex float
     inline __m128 operator()(__m128 a, __m128 b){
       __m128 ymm0,ymm1,ymm2;
       ymm0 = _mm_shuffle_ps(a,a,_MM_SHUFFLE(2,2,0,0)); // ymm0 <- ar ar,
-      ymm0 = _mm_mul_ps(ymm0,b);        // ymm0 <- ar bi, ar br
+      ymm0 = _mm_mul_ps(ymm0,b);                       // ymm0 <- ar bi, ar br
       ymm1 = _mm_shuffle_ps(b,b,_MM_SHUFFLE(2,3,0,1)); // ymm1 <- br,bi
       ymm2 = _mm_shuffle_ps(a,a,_MM_SHUFFLE(3,3,1,1)); // ymm2 <- ai,ai
-      ymm1 = _mm_mul_ps(ymm1,ymm2);       // ymm1 <- br ai, ai bi
+      ymm1 = _mm_mul_ps(ymm1,ymm2);                    // ymm1 <- br ai, ai bi
       return _mm_addsub_ps(ymm0,ymm1);    
     }
     // Complex double
     inline __m128d operator()(__m128d a, __m128d b){
       __m128d ymm0,ymm1,ymm2;
-      ymm0 = _mm_shuffle_pd(a,a,0x0); // ymm0 <- ar ar,
+      ymm0 = _mm_shuffle_pd(a,a,0x0);   // ymm0 <- ar ar,
       ymm0 = _mm_mul_pd(ymm0,b);        // ymm0 <- ar bi, ar br
-      ymm1 = _mm_shuffle_pd(b,b,0x1); // ymm1 <- br,bi   b01
-      ymm2 = _mm_shuffle_pd(a,a,0x3); // ymm2 <- ai,ai   b11
-      ymm1 = _mm_mul_pd(ymm1,ymm2);       // ymm1 <- br ai, ai bi
+      ymm1 = _mm_shuffle_pd(b,b,0x1);   // ymm1 <- br,bi   b01
+      ymm2 = _mm_shuffle_pd(a,a,0x3);   // ymm2 <- ai,ai   b11
+      ymm1 = _mm_mul_pd(ymm1,ymm2);     // ymm1 <- br ai, ai bi
       return _mm_addsub_pd(ymm0,ymm1);  
     }
   };
@@ -165,14 +175,112 @@ namespace Optimization {
     inline __m128i operator()(__m128i a, __m128i b){
       return _mm_mul_epi32(a,b);
     }
+  };
+
+
+  struct Conj{
+    // Complex single
+    inline __m128 operator()(__m128 in){
+      return _mm_xor_ps(_mm_addsub_ps(_mm_setzero_ps(),in), _mm_set1_ps(-0.f));
+    }
+    // Complex double
+    inline __m128d operator()(__m128d in){
+      return _mm_xor_pd(_mm_addsub_pd(_mm_setzero_pd(),in), _mm_set1_pd(-0.f));//untested
+    }
+    // do not define for integer input
+  };
+
+  struct TimesMinusI{
+    //Complex single
+    inline __m128 operator()(__m128 in, __m128 ret){
+      __m128 tmp =_mm_addsub_ps(_mm_setzero_ps(),in); // r,-i
+      return _mm_shuffle_ps(tmp,tmp,_MM_SHUFFLE(2,3,0,1));
+    }
+    //Complex double
+    inline __m128d operator()(__m128d in, __m128d ret){
+      __m128d tmp =_mm_addsub_pd(_mm_setzero_pd(),in); // r,-i
+      return _mm_shuffle_pd(tmp,tmp,0x1);
+    }
+
+
+  };
+
+  struct TimesI{
+    //Complex single
+    inline __m128 operator()(__m128 in, __m128 ret){
+      __m128 tmp =_mm_shuffle_ps(in,in,_MM_SHUFFLE(2,3,0,1));
+      return _mm_addsub_ps(_mm_setzero_ps(),tmp); // r,-i
+    }
+    //Complex double
+    inline __m128d operator()(__m128d in, __m128d ret){
+      __m128d tmp = _mm_shuffle_pd(in,in,0x1);
+      return _mm_addsub_pd(_mm_setzero_pd(),tmp); // r,-i
+    }
+
 
   };
 
 
+  
 
 
+  //////////////////////////////////////////////
+  // Some Template specialization
+  
+  //Complex float Reduce
+  template<>
+  inline Grid::ComplexF Reduce<Grid::ComplexF, __m128>::operator()(__m128 in){
+    union {
+      __m128 v1;  
+      float f[4]; 
+    } u128;
+    u128.v1 = _mm_add_ps(in, _mm_shuffle_ps(in,in, 0b01001110)); // FIXME Prefer to use _MM_SHUFFLE macros
+    return Grid::ComplexF(u128.f[0], u128.f[1]);   
+  }
+  //Real float Reduce
+  template<>
+  inline Grid::RealF Reduce<Grid::RealF, __m128>::operator()(__m128 in){
+    // FIXME Hack
+    const Grid::RealF * ptr = (const Grid::RealF *) &in;
+    Grid::RealF ret = 0; 
+    for(int i=0;i< 4 ;i++){ // 4 number of simd lanes for float
+      ret = ret+ptr[i];
+    }
+    return ret;
+  }
+  
+  
+  //Complex double Reduce
+  template<>
+  inline Grid::ComplexD Reduce<Grid::ComplexD, __m128d>::operator()(__m128d in){
+    printf("Reduce : Missing good complex double implementation -> FIX\n");
+    return Grid::ComplexD(in[0], in[1]); // inefficient
+  }
+  
+  //Real double Reduce
+  template<>
+  inline Grid::RealD Reduce<Grid::RealD, __m128d>::operator()(__m128d in){
+    // FIXME Hack
+    const Grid::RealD * ptr =(const Grid::RealD *)  &in;
+    Grid::RealD ret = 0; 
+    for(int i=0;i< 2 ;i++){// 2 number of simd lanes for float
+      ret = ret+ptr[i];
+    }
+    return ret;
+  }
+
+  //Integer Reduce
+  template<>
+  inline Integer Reduce<Integer, __m128i>::operator()(__m128i in){
+    // FIXME unimplemented
+   printf("Reduce : Missing integer implementation -> FIX\n");
+    assert(0);
+  }
+  
+  
 }
 
+//////////////////////////////////////////////////////////////////////////////////////
 // Here assign types 
 namespace Grid {
   typedef __m128 SIMD_Ftype;  // Single precision type
@@ -180,15 +288,21 @@ namespace Grid {
   typedef __m128i SIMD_Itype; // Integer type
 
 
-  // Function names
-  typedef Optimization::Vsplat VsplatSIMD;
-  typedef Optimization::Vstore VstoreSIMD;
+  // Function name aliases
+  typedef Optimization::Vsplat   VsplatSIMD;
+  typedef Optimization::Vstore   VstoreSIMD;
+  typedef Optimization::Vset     VsetSIMD;
+  typedef Optimization::Vstream  VstreamSIMD;
+  template <typename S, typename T> using ReduceSIMD = Optimization::Reduce<S,T>;
+
 
   // Arithmetic operations
   typedef Optimization::Sum         SumSIMD;
   typedef Optimization::Sub         SubSIMD;
   typedef Optimization::Mult        MultSIMD;
   typedef Optimization::MultComplex MultComplexSIMD;
-  typedef Optimization::Vset        VsetSIMD;
+  typedef Optimization::Conj        ConjSIMD;
+  typedef Optimization::TimesMinusI TimesMinusISIMD;
+  typedef Optimization::TimesI      TimesISIMD;
 
 }
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index 030a8a79..442c5871 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -2,7 +2,7 @@
 /*! @file Grid_vector_types.h
   @brief Defines templated class Grid_simd to deal with inner vector types
 */
-// Time-stamp: <2015-05-19 17:20:36 neo>
+// Time-stamp: <2015-05-20 17:21:52 neo>
 //---------------------------------------------------------------------------
 #ifndef GRID_VECTOR_TYPES
 #define GRID_VECTOR_TYPES
@@ -22,6 +22,16 @@ namespace Grid {
     typedef T type;
   };
 
+  // type alias used to simplify the syntax of std::enable_if
+  template <typename T> using Invoke =
+    typename T::type;
+  template <typename Condition, typename ReturnType> using EnableIf =
+    Invoke<std::enable_if<Condition::value, ReturnType>>;
+  template <typename Condition, typename ReturnType> using NotEnableIf =
+    Invoke<std::enable_if<!Condition::value, ReturnType>>;
+  
+
+
   ////////////////////////////////////////////////////////
   // Check for complexity with type traits
   template <typename T> 
@@ -93,31 +103,32 @@ namespace Grid {
     // Initialise to 1,0,i for the correct types
     ///////////////////////////////////////////////
     // if not complex overload here 
-    template <  class S = Scalar_type,typename std::enable_if < !is_complex < S >::value, int >::type = 0 > 
+    template <  class S = Scalar_type, NotEnableIf<is_complex < S >,int> = 0 > 
       friend inline void vone(Grid_simd &ret)      { vsplat(ret,1.0); }
-    template <  class S = Scalar_type,typename std::enable_if < !is_complex < S >::value, int >::type = 0 > 
+    template <  class S = Scalar_type, NotEnableIf<is_complex < S >,int> = 0 > 
       friend inline void vzero(Grid_simd &ret)     { vsplat(ret,0.0); }
     
-    // overload for complex type
-    template <  class S = Scalar_type,typename std::enable_if < is_complex < S >::value, int >::type = 0 > 
+    // For complex types
+    template <  class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 
       friend inline void vone(Grid_simd &ret)      { vsplat(ret,1.0,0.0); }
-    template < class S = Scalar_type,typename std::enable_if < is_complex < S >::value, int >::type = 0 > 
+    template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 
       friend inline void vzero(Grid_simd &ret)     { vsplat(ret,0.0,0.0); }// use xor?
-    
-    // For integral type
-    template <  class S = Scalar_type,typename std::enable_if < std::is_integral < S >::value, int >::type = 0 > 
+    template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 
+      friend inline void vcomplex_i(Grid_simd &ret){ vsplat(ret,0.0,1.0);} 
+   
+    // For integral types
+    template <  class S = Scalar_type, EnableIf<std::is_integral < S >, int> = 0 > 
       friend inline void vone(Grid_simd &ret)      { vsplat(ret,1); }
-    template <  class S = Scalar_type,typename std::enable_if < std::is_integral < S >::value, int >::type = 0 > 
+    template <  class S = Scalar_type, EnableIf<std::is_integral < S >, int> = 0 > 
       friend inline void vzero(Grid_simd &ret)      { vsplat(ret,0); }
-    template <  class S = Scalar_type,typename std::enable_if < std::is_integral < S >::value, int >::type = 0 > 
+    template <  class S = Scalar_type, EnableIf<std::is_integral < S >, int> = 0 > 
       friend inline void vtrue (Grid_simd &ret){vsplat(ret,0xFFFFFFFF);}
-    template <  class S = Scalar_type,typename std::enable_if < std::is_integral < S >::value, int >::type = 0 > 
+    template <  class S = Scalar_type, EnableIf<std::is_integral < S >, int> = 0 > 
       friend inline void vfalse(vInteger &ret){vsplat(ret,0);}
 
 
-    // do not compile if real or integer, send an error message from the compiler
-    template < class S = Scalar_type,typename std::enable_if < is_complex < S >::value, int >::type = 0 > 
-    friend inline void vcomplex_i(Grid_simd &ret){ vsplat(ret,0.0,1.0);}
+   
+ 
    
     ////////////////////////////////////
     // Arithmetic operator overloads +,-,*
@@ -137,7 +148,7 @@ namespace Grid {
     };
         
     // Distinguish between complex types and others
-    template < class S = Scalar_type, typename std::enable_if < is_complex < S >::value, int >::type = 0 >
+    template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 >
       friend inline Grid_simd operator * (Grid_simd a, Grid_simd b)
       {
 	Grid_simd ret;
@@ -146,7 +157,7 @@ namespace Grid {
       };
 
     // Real/Integer types
-    template <  class S = Scalar_type,typename std::enable_if < !is_complex < S >::value, int >::type = 0 > 
+    template <  class S = Scalar_type, NotEnableIf<is_complex < S >, int> = 0 > 
     friend inline Grid_simd operator * (Grid_simd a, Grid_simd b)
       {
 	Grid_simd ret;
@@ -155,8 +166,6 @@ namespace Grid {
       };
     
 
-
-
     ////////////////////////////////////////////////////////////////////////
     // FIXME:  gonna remove these load/store, get, set, prefetch
     ////////////////////////////////////////////////////////////////////////
@@ -169,14 +178,14 @@ namespace Grid {
     ///////////////////////
     // overload if complex
     template < class S = Scalar_type > 
-    friend inline void vsplat(Grid_simd &ret, typename std::enable_if< is_complex < S >::value, S>::type c){
+    friend inline void vsplat(Grid_simd &ret, EnableIf<is_complex < S >, S> c){
       Real a = real(c);
       Real b = imag(c);
       vsplat(ret,a,b);
     }
 
-    // this only for the complex version
-    template < class S = Scalar_type, typename std::enable_if < is_complex < S >::value, int >::type = 0 > 
+    // this is only for the complex version
+    template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 
     friend inline void vsplat(Grid_simd &ret,Real a, Real b){
       ret.v = binary<Vector_type>(a, b, VsplatSIMD());
     }    
@@ -186,22 +195,45 @@ namespace Grid {
       ret.v = unary<Vector_type>(a, VsplatSIMD());
     }    
 
-
+    ///////////////////////
+    // Vstore
+    ///////////////////////
     friend inline void vstore(const Grid_simd &ret, Scalar_type *a){
       binary<void>(ret.v, (Real*)a, VstoreSIMD());
     }
 
+    ///////////////////////
+    // Vstream
+    ///////////////////////
+    friend inline void vstream(Grid_simd &out,const Grid_simd &in){
+      binary<void>(out.v, in.v, VstreamSIMD());
+    }
+
+    template <  class S = Scalar_type, EnableIf<std::is_integral < S >, int> = 0 > 
+      friend inline void vstream(Grid_simd &out,const Grid_simd &in){
+      out=in;
+    }
+    
+    ///////////////////////
+    // Vprefetch
+    ///////////////////////
     friend inline void vprefetch(const Grid_simd &v)
     {
       _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
     }
 
 
+    ///////////////////////
+    // Reduce
+    ///////////////////////
     friend inline Scalar_type Reduce(const Grid_simd & in)
     {
-      // FIXME add operator
+      return unary<Scalar_type>(in.v, ReduceSIMD<Scalar_type, Vector_type>());
     }
 
+    ////////////////////////////
+    // opreator scalar * simd
+    ////////////////////////////
     friend inline Grid_simd operator * (const Scalar_type &a, Grid_simd b){
       Grid_simd va;
       vsplat(va,a);
@@ -214,25 +246,63 @@ namespace Grid {
     ///////////////////////
     // Conjugate
     ///////////////////////
-								     
+    template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 							     
     friend inline Grid_simd  conj(const Grid_simd  &in){
-      Grid_simd  ret ; vzero(ret);
-      // FIXME add operator
+      Grid_simd  ret ; 
+      ret.v = unary<Vector_type>(in.v, ConjSIMD());
       return ret;
     }
+    template < class S = Scalar_type, NotEnableIf<is_complex < S >, int> = 0 > 
+    friend inline Grid_simd  conj(const Grid_simd  &in){
+      return in; // for real objects
+    }
+
+
+    ///////////////////////
+    // timesMinusI
+    ///////////////////////
+    template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 
+    friend inline void timesMinusI( Grid_simd &ret,const Grid_simd &in){
+      ret.v = binary<Vector_type>(in.v, ret.v, TimesMinusISIMD());
+    }
+
+    template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 
     friend inline Grid_simd timesMinusI(const Grid_simd &in){
       Grid_simd ret; 
-      vzero(ret);
-      // FIXME add operator
+      timesMinusI(ret,in);
       return ret;
     }
-    friend inline Grid_simd timesI(const Grid_simd &in){
-      Grid_simd ret; vzero(ret);
-      // FIXME add operator
-      return ret;
+
+    template < class S = Scalar_type, NotEnableIf<is_complex < S >, int> = 0 > 
+    friend inline Grid_simd timesMinusI(const Grid_simd &in){
+      return in;
+    }
+
+
+    ///////////////////////
+    // timesI
+    ///////////////////////
+    template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 
+    friend inline void timesI(Grid_simd &ret,const Grid_simd &in){
+      ret.v =   binary<Vector_type>(in.v, ret.v, TimesISIMD());     
     }
         
+    template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 
+    friend inline Grid_simd timesI(const Grid_simd &in){
+      Grid_simd ret; 
+      timesI(ret,in);
+      return ret;
+    }
+
+    template < class S = Scalar_type, NotEnableIf<is_complex < S >, int> = 0 > 
+    friend inline Grid_simd timesI(const Grid_simd &in){
+      return in;
+    }
+
+
+    ///////////////////////
     // Unary negation
+    ///////////////////////
     friend inline Grid_simd operator -(const Grid_simd &r) {
       vComplexF ret;
       vzero(ret);
@@ -256,41 +326,22 @@ namespace Grid {
 
 
 
-     friend inline void permute(Grid_simd &y,Grid_simd b,int perm)
-      {
-        Gpermute<Grid_simd>(y,b,perm);
-      }
 
-     /*
+    ////////////////////////////////////////////////////////////////////
+    // General permute; assumes vector length is same across 
+    // all subtypes; may not be a good assumption, but could
+    // add the vector width as a template param for BG/Q for example
+    ////////////////////////////////////////////////////////////////////
     friend inline void permute(Grid_simd &y,Grid_simd b,int perm)
     {
       Gpermute<Grid_simd>(y,b,perm);
     }
-    friend inline void merge(Grid_simd &y,std::vector<Scalar_type *> &extracted)
-    {
-      Gmerge<Grid_simd,Scalar_type >(y,extracted);
-    }
-    friend inline void extract(const Grid_simd &y,std::vector<Scalar_type *> &extracted)
-    {
-      Gextract<Grid_simd,Scalar_type>(y,extracted);
-    }
-    friend inline void merge(Grid_simd &y,std::vector<Scalar_type > &extracted)
-    {
-      Gmerge<Grid_simd,Scalar_type >(y,extracted);
-    }
-    friend inline void extract(const Grid_simd &y,std::vector<Scalar_type > &extracted)
-    {
-      Gextract<Grid_simd,Scalar_type>(y,extracted);
-    }
-     */
-
+    
+    
   };// end of Grid_simd class definition 
 
 
 
-
-
-
   template<class scalar_type, class vector_type > 
     inline Grid_simd< scalar_type, vector_type>  innerProduct(const Grid_simd< scalar_type, vector_type> & l, const Grid_simd< scalar_type, vector_type> & r) 
   {
@@ -314,7 +365,7 @@ namespace Grid {
   }
 
 
-  // Define available types (now change names to avoid clashing)
+  // Define available types (now change names to avoid clashing with the rest of the code)
 
   typedef Grid_simd< float                 , SIMD_Ftype > MyRealF;
   typedef Grid_simd< double                , SIMD_Dtype > MyRealD;
@@ -323,6 +374,29 @@ namespace Grid {
 
 
 
+
+  ////////////////////////////////////////////////////////////////////
+  // Temporary hack to keep independent from the rest of the code
+  template<> struct isGridTensor<MyRealD > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<MyRealF > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<MyComplexD > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+  template<> struct isGridTensor<MyComplexF > {
+    static const bool value = false;
+    static const bool notvalue = true;
+  };
+
+
+
+
 }
 
 #endif
diff --git a/tests/Grid_main.cc b/tests/Grid_main.cc
index 17e36df9..2aa44a0f 100644
--- a/tests/Grid_main.cc
+++ b/tests/Grid_main.cc
@@ -161,30 +161,40 @@ int main (int argc, char ** argv)
     ///////// Tests the new class Grid_simd 
     std::complex<double> ctest(3.0,2.0);
     std::complex<float> ctestf(3.0,2.0);
-    MyComplexF TestMe1(1.0); // fill real part
+    MyComplexF TestMe1(1.0); // fills only real part
     MyComplexD TestMe2(ctest);
     MyComplexD TestMe3(ctest);// compiler generate conversion of basic types
     //MyRealF TestMe5(ctest);// Must generate compiler error
-    MyRealD TestMe4(2.0); 
+    MyRealD TestRe1(2.0); 
+    MyRealF TestRe2(3.0); 
  
+    vone(TestRe2);
+
     MyComplexF TestMe6(ctestf);
     MyComplexF TestMe7(ctestf);  
     
     MyComplexD TheSum= TestMe2*TestMe3;
     MyComplexF TheSumF= TestMe6*TestMe7;
 
+    
+
     double dsum[2];
     _mm_store_pd(dsum, TheSum.v);
     for (int i =0; i< 2; i++)
       printf("%f\n", dsum[i]);
+    MyComplexD TheSumI = timesMinusI(TheSum);
+    MyComplexF TheSumIF = timesMinusI(TheSumF);
 
     float fsum[4];
     _mm_store_ps(fsum, TheSumF.v);
     for (int i =0; i< 4; i++)
       printf("%f\n", fsum[i]);
 
-    vstore(TheSum, &ctest);
+    vstore(TheSumI, &ctest);
+    std::complex<float> sum = Reduce(TheSumF);
     std::cout << ctest<< std::endl;
+    std::cout << sum<< std::endl;
+
 #endif
     ///////////////////////
 
diff --git a/tests/Grid_simd_new.cc b/tests/Grid_simd_new.cc
new file mode 100644
index 00000000..3de12231
--- /dev/null
+++ b/tests/Grid_simd_new.cc
@@ -0,0 +1,165 @@
+#include <Grid.h>
+#include "simd/Grid_vector_types.h"
+#include <parallelIO/GridNerscIO.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+class funcPlus {
+public:
+  funcPlus() {};
+  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = i1+i2;}
+  std::string name(void) const { return std::string("Plus"); }
+};
+class funcMinus {
+public:
+  funcMinus() {};
+  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = i1-i2;}
+  std::string name(void) const { return std::string("Minus"); }
+};
+class funcTimes {
+public:
+  funcTimes() {};
+  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = i1*i2;}
+  std::string name(void) const { return std::string("Times"); }
+};
+class funcConj {
+public:
+  funcConj() {};
+  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = conj(i1);}
+  std::string name(void) const { return std::string("Conj"); }
+};
+class funcAdj {
+public:
+  funcAdj() {};
+  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = adj(i1);}
+  std::string name(void) const { return std::string("Adj"); }
+};
+
+class funcTimesI {
+public:
+  funcTimesI() {};
+  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = timesI(i1);}
+  std::string name(void) const { return std::string("timesI"); }
+};
+
+class funcTimesMinusI {
+public:
+  funcTimesMinusI() {};
+  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = timesMinusI(i1);}
+  std::string name(void) const { return std::string("timesMinusI"); }
+};
+
+template<class scal, class vec,class functor > 
+void Tester(const functor &func)
+{
+  GridSerialRNG          sRNG;
+  sRNG.SeedRandomDevice();
+  
+  int Nsimd = vec::Nsimd();
+
+  std::vector<scal> input1(Nsimd);
+  std::vector<scal> input2(Nsimd);
+  std::vector<scal> result(Nsimd);
+  std::vector<scal> reference(Nsimd);
+
+  std::vector<vec,alignedAllocator<vec> > buf(3);
+  vec & v_input1 = buf[0];
+  vec & v_input2 = buf[1];
+  vec & v_result = buf[2];
+
+
+  for(int i=0;i<Nsimd;i++){
+    random(sRNG,input1[i]);
+    random(sRNG,input2[i]);
+    random(sRNG,result[i]);
+  }
+
+  merge<vec,scal>(v_input1,input1);
+  merge<vec,scal>(v_input2,input2);
+  merge<vec,scal>(v_result,result);
+
+  func(v_result,v_input1,v_input2);
+
+  for(int i=0;i<Nsimd;i++) {
+    func(reference[i],input1[i],input2[i]);
+  }
+
+  extract<vec,scal>(v_result,result);
+  std::cout << " " << func.name()<<std::endl;
+
+  int ok=0;
+  for(int i=0;i<Nsimd;i++){
+    if ( abs(reference[i]-result[i])>0){
+      std::cout<< "*****" << std::endl;
+      std::cout<< "["<<i<<"] "<< abs(reference[i]-result[i]) << " " <<reference[i]<< " " << result[i]<<std::endl;
+      ok++;
+    }
+  }
+  if ( ok==0 ) std::cout << " OK!" <<std::endl;
+
+}
+
+
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(4,MyComplexF::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+    
+  GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+  std::vector<int> seeds({1,2,3,4});
+
+  // Insist that operations on random scalars gives
+  // identical results to on vectors.
+
+  std::cout << "==================================="<<  std::endl;
+  std::cout << "Testing MyComplexF "<<std::endl;
+  std::cout << "==================================="<<  std::endl;
+
+  Tester<ComplexF,MyComplexF>(funcTimesI());
+  Tester<ComplexF,MyComplexF>(funcTimesMinusI());
+  Tester<ComplexF,MyComplexF>(funcPlus());
+  Tester<ComplexF,MyComplexF>(funcMinus());
+  Tester<ComplexF,MyComplexF>(funcTimes());
+  Tester<ComplexF,MyComplexF>(funcConj());
+  Tester<ComplexF,MyComplexF>(funcAdj());
+
+  std::cout << "==================================="<<  std::endl;
+  std::cout << "Testing MyComplexD "<<std::endl;
+  std::cout << "==================================="<<  std::endl;
+
+
+  Tester<ComplexD,MyComplexD>(funcTimesI());
+  Tester<ComplexD,MyComplexD>(funcTimesMinusI());
+  Tester<ComplexD,MyComplexD>(funcPlus());
+  Tester<ComplexD,MyComplexD>(funcMinus());
+  Tester<ComplexD,MyComplexD>(funcTimes());
+  Tester<ComplexD,MyComplexD>(funcConj());
+  Tester<ComplexD,MyComplexD>(funcAdj());
+
+  std::cout << "==================================="<<  std::endl;
+  std::cout << "Testing MyRealF "<<std::endl;
+  std::cout << "==================================="<<  std::endl;
+
+
+  Tester<RealF,MyRealF>(funcPlus());
+  Tester<RealF,MyRealF>(funcMinus());
+  Tester<RealF,MyRealF>(funcTimes());
+  Tester<RealF,MyRealF>(funcAdj());
+
+  std::cout << "==================================="<<  std::endl;
+  std::cout << "Testing MyRealD "<<std::endl;
+  std::cout << "==================================="<<  std::endl;
+
+  Tester<RealD,MyRealD>(funcPlus());
+  Tester<RealD,MyRealD>(funcMinus());
+  Tester<RealD,MyRealD>(funcTimes());
+  Tester<RealD,MyRealD>(funcAdj());
+
+  Grid_finalize();
+}
diff --git a/tests/Makefile.am b/tests/Makefile.am
index 82d1b3be..8b68855d 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -5,7 +5,7 @@ AM_LDFLAGS = -L$(top_builddir)/lib
 #
 # Test code
 #
-bin_PROGRAMS = Grid_main Grid_stencil Grid_nersc_io Grid_cshift Grid_gamma  Grid_simd Grid_rng Grid_remez
+bin_PROGRAMS = Grid_main Grid_stencil Grid_nersc_io Grid_cshift Grid_gamma  Grid_simd Grid_rng Grid_remez Grid_simd_new
 
 Grid_main_SOURCES = Grid_main.cc
 Grid_main_LDADD = -lGrid
@@ -30,3 +30,6 @@ Grid_stencil_LDADD = -lGrid
 
 Grid_simd_SOURCES = Grid_simd.cc
 Grid_simd_LDADD = -lGrid
+
+Grid_simd_new_SOURCES = Grid_simd_new.cc
+Grid_simd_new_LDADD = -lGrid

From 35055ed5c1249c1391a64217a203fe5b3d410b8b Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 21 May 2015 06:34:33 +0100
Subject: [PATCH 207/429] Unroll pragma abstraction

---
 lib/Grid_threads.h | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/lib/Grid_threads.h b/lib/Grid_threads.h
index 89a56d80..24581855 100644
--- a/lib/Grid_threads.h
+++ b/lib/Grid_threads.h
@@ -5,6 +5,8 @@
 #define GRID_OMP
 #endif
 
+#define UNROLL  _Pragma("unroll")
+
 #ifdef GRID_OMP
 #include <omp.h>
 #define PARALLEL_FOR_LOOP _Pragma("omp parallel for")

From d8065816662c56de8741569686947e0b9c03fe06 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 21 May 2015 06:35:46 +0100
Subject: [PATCH 208/429] better comms benchmarking

---
 benchmarks/Grid_comms.cc | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/benchmarks/Grid_comms.cc b/benchmarks/Grid_comms.cc
index e844c5d3..59d709a0 100644
--- a/benchmarks/Grid_comms.cc
+++ b/benchmarks/Grid_comms.cc
@@ -23,7 +23,7 @@ int main (int argc, char ** argv)
 
 
 
-  for(int lat=4;lat<=16;lat+=4){
+  for(int lat=4;lat<=32;lat+=2){
     for(int Ls=1;Ls<=16;Ls*=2){
 
       std::vector<int> latt_size  ({lat,lat,lat,lat});
@@ -94,8 +94,7 @@ int main (int argc, char ** argv)
   std::cout << "  L  "<<"\t\t"<<" Ls  "<<"\t\t"<<"bytes"<<"\t\t"<<"MB/s uni"<<"\t\t"<<"MB/s bidi"<<std::endl;
 
 
-
-  for(int lat=4;lat<=16;lat+=4){
+  for(int lat=4;lat<=32;lat+=2){
     for(int Ls=1;Ls<=16;Ls*=2){
 
       std::vector<int> latt_size  ({lat,lat,lat,lat});

From 57a01e6bbb69f3b2d98fb0e060d58987d283f94a Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 21 May 2015 06:36:15 +0100
Subject: [PATCH 209/429] Didn't like a print statement

---
 lib/algorithms/iterative/ConjugateGradient.h | 1 -
 1 file changed, 1 deletion(-)

diff --git a/lib/algorithms/iterative/ConjugateGradient.h b/lib/algorithms/iterative/ConjugateGradient.h
index f7f8559e..85b196e9 100644
--- a/lib/algorithms/iterative/ConjugateGradient.h
+++ b/lib/algorithms/iterative/ConjugateGradient.h
@@ -15,7 +15,6 @@ public:
     Integer MaxIterations;
 
     ConjugateGradient(RealD tol,Integer maxit) : Tolerance(tol), MaxIterations(maxit) { 
-      std::cout << Tolerance<<std::endl;
     };
 
     void operator() (LinearOperatorBase<Field> &Linop,const Field &src, Field &psi) {assert(0);};

From c96af471eedb11379b1700fd2667aa462526191a Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 21 May 2015 06:36:47 +0100
Subject: [PATCH 210/429] useful to dump assembler

---
 benchmarks/Grid_su3_test.cc | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/benchmarks/Grid_su3_test.cc b/benchmarks/Grid_su3_test.cc
index e4d4ec0a..8a7f6246 100644
--- a/benchmarks/Grid_su3_test.cc
+++ b/benchmarks/Grid_su3_test.cc
@@ -4,7 +4,8 @@ using namespace std;
 using namespace Grid;
 using namespace Grid::QCD;
 
-void su3_test_mult(LatticeColourMatrix &z, LatticeColourMatrix &x,LatticeColourMatrix &y)
+
+void su3_test_mult_routine(LatticeColourMatrix &z, LatticeColourMatrix &x,LatticeColourMatrix &y)
 {
-  z=x*y;
+  mult(z,x,y);
 }

From 3e1d1aff185c7917670893361dfb298e84b7581b Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 21 May 2015 06:37:20 +0100
Subject: [PATCH 211/429] Minor change

---
 benchmarks/Grid_wilson_cg_unprec.cc | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/benchmarks/Grid_wilson_cg_unprec.cc b/benchmarks/Grid_wilson_cg_unprec.cc
index 2a91864b..93023534 100644
--- a/benchmarks/Grid_wilson_cg_unprec.cc
+++ b/benchmarks/Grid_wilson_cg_unprec.cc
@@ -30,7 +30,7 @@ int main (int argc, char ** argv)
   GridParallelRNG          pRNG(&Grid);  pRNG.SeedFixedIntegers(seeds);
 
   LatticeFermion src(&Grid); random(pRNG,src);
-  std::cout << "src norm" << norm2(src)<<std::endl;
+  RealD nrm = norm2(src);
   LatticeFermion result(&Grid); result=zero;
   LatticeGaugeField Umu(&Grid); random(pRNG,Umu);
 

From f1fb92fd010d2560e9b869223860303117c8ac4d Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 21 May 2015 06:37:46 +0100
Subject: [PATCH 212/429] adding two routines containing only a single
 operation so I can easily see the assembly dump

---
 benchmarks/Makefile.am | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/benchmarks/Makefile.am b/benchmarks/Makefile.am
index ae09a6db..fc513e86 100644
--- a/benchmarks/Makefile.am
+++ b/benchmarks/Makefile.am
@@ -16,7 +16,7 @@ Grid_wilson_cg_unprec_LDADD = -lGrid
 Grid_comms_SOURCES = Grid_comms.cc
 Grid_comms_LDADD = -lGrid
 
-Grid_su3_SOURCES = Grid_su3.cc Grid_su3_test.cc
+Grid_su3_SOURCES = Grid_su3.cc Grid_su3_test.cc Grid_su3_expr.cc
 Grid_su3_LDADD = -lGrid
 
 Grid_memory_bandwidth_SOURCES = Grid_memory_bandwidth.cc

From 874b2eb32de3334528e856d59cb6399cb849c7ae Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 21 May 2015 06:39:00 +0100
Subject: [PATCH 213/429] Compile time select if we do the streaming store
 copy. Relies on Clang++ eliminating object copies, and other compliers do not
 necessarily cope.

---
 lib/lattice/Grid_lattice_arith.h | 61 ++++++++++++++++++++++++++------
 1 file changed, 50 insertions(+), 11 deletions(-)

diff --git a/lib/lattice/Grid_lattice_arith.h b/lib/lattice/Grid_lattice_arith.h
index f1e566a2..ff966578 100644
--- a/lib/lattice/Grid_lattice_arith.h
+++ b/lib/lattice/Grid_lattice_arith.h
@@ -12,10 +12,13 @@ namespace Grid {
     conformable(lhs,rhs);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
+#ifdef STREAMING_STORES
       obj1 tmp;
       mult(&tmp,&lhs._odata[ss],&rhs._odata[ss]);
       vstream(ret._odata[ss],tmp);
-      //      mult(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
+#else
+      mult(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
+#endif
     }
   }
   
@@ -24,9 +27,13 @@ PARALLEL_FOR_LOOP
     conformable(lhs,rhs);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
+#ifdef STREAMING_STORES
       obj1 tmp;
       mac(&tmp,&lhs._odata[ss],&rhs._odata[ss]);
       vstream(ret._odata[ss],tmp);
+#else
+      mac(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
+#endif
     }
   }
   
@@ -35,9 +42,13 @@ PARALLEL_FOR_LOOP
     conformable(lhs,rhs);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
+#ifdef STREAMING_STORES
       obj1 tmp;
       sub(&tmp,&lhs._odata[ss],&rhs._odata[ss]);
       vstream(ret._odata[ss],tmp);
+#else
+      sub(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
+#endif
     }
   }
   template<class obj1,class obj2,class obj3> strong_inline
@@ -45,9 +56,13 @@ PARALLEL_FOR_LOOP
     conformable(lhs,rhs);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
+#ifdef STREAMING_STORES
       obj1 tmp;
       add(&tmp,&lhs._odata[ss],&rhs._odata[ss]);
       vstream(ret._odata[ss],tmp);
+#else
+      add(&ret._odata[ss],&lhs._odata[ss],&rhs._odata[ss]);
+#endif
     }
   }
   
@@ -81,9 +96,13 @@ PARALLEL_FOR_LOOP
     conformable(lhs,ret);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
+#ifdef STREAMING_STORES
       obj1 tmp;
       sub(&tmp,&lhs._odata[ss],&rhs);
       vstream(ret._odata[ss],tmp);
+#else 
+      sub(&ret._odata[ss],&lhs._odata[ss],&rhs);
+#endif
     }
   }
   template<class obj1,class obj2,class obj3> strong_inline
@@ -91,9 +110,13 @@ PARALLEL_FOR_LOOP
     conformable(lhs,ret);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
+#ifdef STREAMING_STORES
       obj1 tmp;
       add(&tmp,&lhs._odata[ss],&rhs);
       vstream(ret._odata[ss],tmp);
+#else 
+      add(&ret._odata[ss],&lhs._odata[ss],&rhs);
+#endif
     }
   }
 
@@ -105,9 +128,13 @@ PARALLEL_FOR_LOOP
     conformable(ret,rhs);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<rhs._grid->oSites();ss++){
+#ifdef STREAMING_STORES
       obj1 tmp;
       mult(&tmp,&lhs,&rhs._odata[ss]);
       vstream(ret._odata[ss],tmp);
+#else 
+      mult(&ret._odata[ss],&lhs,&rhs._odata[ss]);
+#endif
     }
   }
   
@@ -116,9 +143,13 @@ PARALLEL_FOR_LOOP
     conformable(ret,rhs);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<rhs._grid->oSites();ss++){
+#ifdef STREAMING_STORES
       obj1 tmp;
       mac(&tmp,&lhs,&rhs._odata[ss]);
       vstream(ret._odata[ss],tmp);
+#else 
+      mac(&ret._odata[ss],&lhs,&rhs._odata[ss]);
+#endif
     }
   }
   
@@ -127,9 +158,13 @@ PARALLEL_FOR_LOOP
     conformable(ret,rhs);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<rhs._grid->oSites();ss++){
+#ifdef STREAMING_STORES
       obj1 tmp;
       sub(&tmp,&lhs,&rhs._odata[ss]);
       vstream(ret._odata[ss],tmp);
+#else 
+      sub(&ret._odata[ss],&lhs,&rhs._odata[ss]);
+#endif
     }
   }
   template<class obj1,class obj2,class obj3> strong_inline
@@ -137,9 +172,13 @@ PARALLEL_FOR_LOOP
     conformable(ret,rhs);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<rhs._grid->oSites();ss++){
+#ifdef STREAMING_STORES
       obj1 tmp;
       add(&tmp,&lhs,&rhs._odata[ss]);
       vstream(ret._odata[ss],tmp);
+#else 
+      add(&ret._odata[ss],&lhs,&rhs._odata[ss]);
+#endif
     }
   }
   
@@ -148,8 +187,12 @@ PARALLEL_FOR_LOOP
     conformable(x,y);
 #pragma omp parallel for
     for(int ss=0;ss<x._grid->oSites();ss++){
+#ifdef STREAMING_STORES
       vobj tmp = a*x._odata[ss]+y._odata[ss];
       vstream(ret._odata[ss],tmp);
+#else
+      ret._odata[ss]=a*x._odata[ss]+y._odata[ss];
+#endif
     }
   }
   template<class sobj,class vobj> strong_inline
@@ -157,29 +200,25 @@ PARALLEL_FOR_LOOP
     conformable(x,y);
 #pragma omp parallel for
     for(int ss=0;ss<x._grid->oSites();ss++){
+#ifdef STREAMING_STORES
       vobj tmp = a*x._odata[ss]+b*y._odata[ss];
       vstream(ret._odata[ss],tmp);
+#else
+      ret._odata[ss]=a*x._odata[ss]+b*y._odata[ss];
+#endif
     }
   }
 
   template<class sobj,class vobj> strong_inline
   RealD axpy_norm(Lattice<vobj> &ret,sobj a,const Lattice<vobj> &x,const Lattice<vobj> &y){
     conformable(x,y);
-#pragma omp parallel for
-    for(int ss=0;ss<x._grid->oSites();ss++){
-      vobj tmp = a*x._odata[ss]+y._odata[ss];
-      vstream(ret._odata[ss],tmp);
-    }
+    axpy(ret,a,x,y);
     return norm2(ret);
   }
   template<class sobj,class vobj> strong_inline
   RealD axpby_norm(Lattice<vobj> &ret,sobj a,sobj b,const Lattice<vobj> &x,const Lattice<vobj> &y){
     conformable(x,y);
-#pragma omp parallel for
-    for(int ss=0;ss<x._grid->oSites();ss++){
-      vobj tmp = a*x._odata[ss]+b*y._odata[ss];
-      vstream(ret._odata[ss],tmp);
-    }
+    axpby(ret,a,b,x,y);
     return norm2(ret); // FIXME implement parallel norm in ss loop
   }
 

From d8061afe2456dc8c46499bd803e1726e6391fc51 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 21 May 2015 06:47:05 +0100
Subject: [PATCH 214/429] Streaming store option ifdef

---
 lib/lattice/Grid_lattice_base.h | 41 ++++++++++++++++++++++++++++-----
 1 file changed, 35 insertions(+), 6 deletions(-)

diff --git a/lib/lattice/Grid_lattice_base.h b/lib/lattice/Grid_lattice_base.h
index 0016bafa..41f349b7 100644
--- a/lib/lattice/Grid_lattice_base.h
+++ b/lib/lattice/Grid_lattice_base.h
@@ -1,6 +1,8 @@
 #ifndef GRID_LATTICE_BASE_H
 #define GRID_LATTICE_BASE_H
 
+#define STREAMING_STORES
+
 namespace Grid {
 
 // TODO: 
@@ -66,8 +68,12 @@ public:
   {
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
-      vobj tmp= eval(ss,expr);
+#ifdef STREAMING_STORES
+      vobj tmp = eval(ss,expr);
       vstream(_odata[ss] ,tmp);
+#else
+      _odata[ss]=eval(ss,expr);
+#endif
     }
     return *this;
   }
@@ -75,8 +81,12 @@ PARALLEL_FOR_LOOP
   {
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
-      vobj tmp= eval(ss,expr);
+#ifdef STREAMING_STORES
+      vobj tmp = eval(ss,expr);
       vstream(_odata[ss] ,tmp);
+#else
+      _odata[ss]=eval(ss,expr);
+#endif
     }
     return *this;
   }
@@ -84,8 +94,12 @@ PARALLEL_FOR_LOOP
   {
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
-      vobj tmp= eval(ss,expr);
+#ifdef STREAMING_STORES
+      vobj tmp = eval(ss,expr);
       vstream(_odata[ss] ,tmp);
+#else
+      _odata[ss] = eval(ss,expr);
+#endif
     }
     return *this;
   }
@@ -97,7 +111,12 @@ PARALLEL_FOR_LOOP
     _odata.resize(_grid->oSites());
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
-      _odata[ss] = eval(ss,expr);
+#ifdef STREAMING_STORES
+      vobj tmp = eval(ss,expr);
+      vstream(_odata[ss] ,tmp);
+#else
+      _odata[ss]=_eval(ss,expr);
+#endif
     }
   };
   template<class Op,class T1, class T2>
@@ -107,7 +126,12 @@ PARALLEL_FOR_LOOP
     _odata.resize(_grid->oSites());
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
-      _odata[ss] = eval(ss,expr);
+#ifdef STREAMING_STORES
+      vobj tmp = eval(tmp,ss,expr);
+      vstream(_odata[ss] ,tmp);
+#else
+      _odata[ss]=eval(ss,expr);
+#endif
     }
   };
   template<class Op,class T1, class T2, class T3>
@@ -117,7 +141,12 @@ PARALLEL_FOR_LOOP
     _odata.resize(_grid->oSites());
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
-      _odata[ss] = eval(ss,expr);
+#ifdef STREAMING_STORES
+      vobj tmp = eval(ss,expr);
+      vstream(_odata[ss] ,tmp);
+#else
+      _odata[ss]=eval(ss,expr);
+#endif
     }
   };
 

From 1c862dc15b934d86d2debfdfb867ad096a992845 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Fri, 22 May 2015 17:33:15 +0900
Subject: [PATCH 215/429] Completed implementation of new Grid_simd classes

Tested performance for SSE4, Ok.
AVX1/2, AVX512 yet untested
---
 configure.ac                        |   4 +-
 lib/Grid_simd.h                     |   6 +-
 lib/Makefile.am                     |  10 +-
 lib/cshift/Grid_cshift_common.h     |   3 +-
 lib/simd/Grid_avx.h                 | 401 ++++++++++++++++++++++++++++
 lib/simd/Grid_knc.h                 | 323 ++++++++++++++++++++++
 lib/simd/Grid_qpx.h                 | 272 +++++++++++++++++++
 lib/simd/Grid_sse4.h                |  10 +-
 lib/simd/Grid_vector_types.h        |  94 ++++---
 lib/simd/{ => Old}/Grid_vComplexD.h |   0
 lib/simd/{ => Old}/Grid_vComplexF.h |   2 +-
 lib/simd/{ => Old}/Grid_vInteger.h  |   0
 lib/simd/{ => Old}/Grid_vRealD.h    |   0
 lib/simd/{ => Old}/Grid_vRealF.h    |   0
 tests/Grid_main.cc                  |  42 ++-
 tests/Makefile.am                   |   6 +-
 16 files changed, 1091 insertions(+), 82 deletions(-)
 create mode 100644 lib/simd/Grid_avx.h
 create mode 100644 lib/simd/Grid_knc.h
 create mode 100644 lib/simd/Grid_qpx.h
 rename lib/simd/{ => Old}/Grid_vComplexD.h (100%)
 rename lib/simd/{ => Old}/Grid_vComplexF.h (99%)
 rename lib/simd/{ => Old}/Grid_vInteger.h (100%)
 rename lib/simd/{ => Old}/Grid_vRealD.h (100%)
 rename lib/simd/{ => Old}/Grid_vRealF.h (100%)

diff --git a/configure.ac b/configure.ac
index 93e2b574..bfcbfcef 100644
--- a/configure.ac
+++ b/configure.ac
@@ -3,9 +3,9 @@
 #
 # Project Grid package  
 # 
-# Time-stamp: <2015-05-19 13:51:08 neo>
+# Time-stamp: <2015-05-22 15:46:09 neo>
 
-AC_PREREQ([2.69])
+AC_PREREQ([2.63])
 AC_INIT([Grid], [1.0], [paboyle@ph.ed.ac.uk])
 AC_CANONICAL_SYSTEM
 AM_INIT_AUTOMAKE(subdir-objects)
diff --git a/lib/Grid_simd.h b/lib/Grid_simd.h
index 6d7411d3..9484c0d6 100644
--- a/lib/Grid_simd.h
+++ b/lib/Grid_simd.h
@@ -203,11 +203,7 @@ inline void Gpermute(vsimd &y,const vsimd &b,int perm){
     };
 };
 
-#include <simd/Grid_vInteger.h>
-#include <simd/Grid_vRealF.h>
-#include <simd/Grid_vRealD.h>
-#include <simd/Grid_vComplexF.h>
-#include <simd/Grid_vComplexD.h>
+#include <simd/Grid_vector_types.h>
 
 
 namespace Grid {
diff --git a/lib/Makefile.am b/lib/Makefile.am
index 82459763..cdb36c71 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -91,12 +91,10 @@ nobase_include_HEADERS = algorithms/approx/bigfloat.h		\
 	qcd/Grid_qcd_2spinor.h					\
 	qcd/Grid_qcd_dirac.h					\
 	qcd/Grid_qcd_wilson_dop.h				\
-	simd/Grid_vComplexD.h					\
-	simd/Grid_vComplexF.h					\
-	simd/Grid_vInteger.h					\
-	simd/Grid_vRealD.h					\
-	simd/Grid_vRealF.h					\
 	simd/Grid_vector_types.h				\
-	simd/Grid_sse4.h					
+	simd/Grid_sse4.h					\
+	simd/Grid_avx.h						\
+	simd/Grid_knc.h					
+
 
 
diff --git a/lib/cshift/Grid_cshift_common.h b/lib/cshift/Grid_cshift_common.h
index cf1f88af..65c0cb87 100644
--- a/lib/cshift/Grid_cshift_common.h
+++ b/lib/cshift/Grid_cshift_common.h
@@ -154,7 +154,6 @@ template<class vobj> void Copy_plane(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int
     cbmask=0x3;
   }
 
-
   int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
   int lo  = lplane*lhs._grid->_ostride[dimension]; // base offset for start of plane 
   
@@ -170,10 +169,12 @@ PARALLEL_NESTED_LOOP2
       
     }
   }
+  
 }
 
 template<class vobj> void Copy_plane_permute(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int dimension,int lplane,int rplane,int cbmask,int permute_type)
 {
+ 
   int rd = rhs._grid->_rdimensions[dimension];
 
   if ( !rhs._grid->CheckerBoarded(dimension) ) {
diff --git a/lib/simd/Grid_avx.h b/lib/simd/Grid_avx.h
new file mode 100644
index 00000000..ea796122
--- /dev/null
+++ b/lib/simd/Grid_avx.h
@@ -0,0 +1,401 @@
+//----------------------------------------------------------------------
+/*! @file Grid_avx.h
+  @brief Optimization libraries for AVX1/2 instructions set
+
+  Using intrinsics
+*/
+// Time-stamp: <2015-05-22 15:51:24 neo>
+//----------------------------------------------------------------------
+
+#include <immintrin.h>
+// _mm256_set_m128i(hi,lo); // not defined in all versions of immintrin.h
+#ifndef _mm256_set_m128i
+#define _mm256_set_m128i(hi,lo) _mm256_insertf128_si256(_mm256_castsi128_si256(lo),(hi),1)
+#endif
+
+namespace Optimization {
+  
+  struct Vsplat{
+    //Complex float
+    inline __m256 operator()(float a, float b){
+      return _mm256_set_ps(b,a,b,a,b,a,b,a);
+    }
+    // Real float
+    inline __m256 operator()(float a){
+      return _mm256_set_ps(a,a,a,a,a,a,a,a);
+    }
+    //Complex double
+    inline __m256d operator()(double a, double b){
+      return _mm256_set_pd(b,a,b,a);
+    }
+    //Real double
+    inline __m256d operator()(double a){
+      return _mm256_set_pd(a,a,a,a);
+    }
+    //Integer
+    inline __m256i operator()(Integer a){
+      return _mm256_set1_epi32(a);
+    }
+  };
+
+  struct Vstore{
+    //Float 
+    inline void operator()(__m256 a, float* F){
+      _mm256_store_ps(F,a);
+    }
+    //Double
+    inline void operator()(__m256d a, double* D){
+      _mm256_store_pd(D,a);
+    }
+    //Integer
+    inline void operator()(__m256i a, Integer* I){
+      _mm256_store_si256((__m256i*)I,a);
+    }
+
+  };
+
+
+  struct Vstream{
+    //Float
+    inline void operator()(float * a, __m256 b){
+      _mm256_stream_ps(a,b);
+    }
+    //Double
+    inline void operator()(double * a, __m256d b){
+      _mm256_stream_pd(a,b);
+    }
+
+
+  };
+
+
+
+  struct Vset{
+    // Complex float 
+    inline __m256 operator()(Grid::ComplexF *a){
+      return _mm256_set_ps(a[3].imag(),a[3].real(),a[2].imag(),a[2].real(),a[1].imag(),a[1].real(),a[0].imag(),a[0].real());
+    }
+    // Complex double 
+    inline __m256d operator()(Grid::ComplexD *a){
+      return _mm256_set_pd(a[1].imag(),a[1].real(),a[0].imag(),a[0].real());
+    }
+    // Real float 
+    inline __m256 operator()(float *a){
+      return _mm256_set_ps(a[7],a[6],a[5],a[4],a[3],a[2],a[1],a[0]);
+    }
+    // Real double
+    inline __m256d operator()(double *a){
+      return _mm256_set_pd(a[3],a[2],a[1],a[0]);
+    }
+    // Integer
+    inline __m256i operator()(Integer *a){
+      return _mm256_set_epi32(a[7],a[6],a[5],a[4],a[3],a[2],a[1],a[0]);
+    }
+
+
+  };
+
+  template <typename Out_type, typename In_type>
+  struct Reduce{
+    //Need templated class to overload output type
+    //General form must generate error if compiled
+    inline Out_type operator()(In_type in){
+      printf("Error, using wrong Reduce function\n");
+      exit(1);
+      return 0;
+    }
+  };
+
+
+ 
+
+  /////////////////////////////////////////////////////
+  // Arithmetic operations
+  /////////////////////////////////////////////////////
+  struct Sum{
+    //Complex/Real float
+    inline __m256 operator()(__m256 a, __m256 b){
+      return _mm256_add_ps(a,b);
+    }
+    //Complex/Real double
+    inline __m256d operator()(__m256d a, __m256d b){
+      return _mm256_add_pd(a,b);
+    }
+    //Integer
+    inline __m256i operator()(__m256i a, __m256i b){
+#if defined (AVX1) 
+          __m128i a0,a1;
+          __m128i b0,b1;
+          a0 = _mm256_extractf128_si256(a,0);
+          b0 = _mm256_extractf128_si256(b,0);
+          a1 = _mm256_extractf128_si256(a,1);
+          b1 = _mm256_extractf128_si256(b,1);
+          a0 = _mm_add_epi32(a0,b0);
+          a1 = _mm_add_epi32(a1,b1);
+          return _mm256_set_m128i(a1,a0);
+#endif
+#if defined (AVX2)
+            return _mm256_add_epi32(a,b);
+#endif
+
+    }
+  };
+
+  struct Sub{
+    //Complex/Real float
+    inline __m256 operator()(__m256 a, __m256 b){
+      return _mm256_sub_ps(a,b);
+    }
+    //Complex/Real double
+    inline __m256d operator()(__m256d a, __m256d b){
+      return _mm256_sub_pd(a,b);
+    }
+    //Integer
+    inline __m256i operator()(__m256i a, __m256i b){
+#if defined (AVX1) 
+          __m128i a0,a1;
+          __m128i b0,b1;
+          a0 = _mm256_extractf128_si256(a,0);
+          b0 = _mm256_extractf128_si256(b,0);
+          a1 = _mm256_extractf128_si256(a,1);
+          b1 = _mm256_extractf128_si256(b,1);
+          a0 = _mm_sub_epi32(a0,b0);
+          a1 = _mm_sub_epi32(a1,b1);
+          return _mm256_set_m128i(a1,a0);
+#endif
+#if defined (AVX2)
+            return _mm256_sub_epi32(a,b);
+#endif
+
+    }
+  };
+
+
+  struct MultComplex{
+    // Complex float
+    inline __m256 operator()(__m256 a, __m256 b){
+      __m256 ymm0,ymm1,ymm2;
+      ymm0 = _mm256_shuffle_ps(a,a,_MM_SHUFFLE(2,2,0,0)); // ymm0 <- ar ar,
+      ymm0 = _mm256_mul_ps(ymm0,b);                       // ymm0 <- ar bi, ar br
+      // FIXME AVX2 could MAC
+      ymm1 = _mm256_shuffle_ps(b,b,_MM_SHUFFLE(2,3,0,1)); // ymm1 <- br,bi
+      ymm2 = _mm256_shuffle_ps(a,a,_MM_SHUFFLE(3,3,1,1)); // ymm2 <- ai,ai
+      ymm1 = _mm256_mul_ps(ymm1,ymm2);                    // ymm1 <- br ai, ai bi
+      return _mm256_addsub_ps(ymm0,ymm1);  
+    }
+    // Complex double
+    inline __m256d operator()(__m256d a, __m256d b){
+      //Multiplication of (ak+ibk)*(ck+idk)
+      // a + i b can be stored as a data structure
+      //From intel optimisation reference guide
+      /*
+	movsldup xmm0, Src1; load real parts into the destination,
+	; a1, a1, a0, a0
+	movaps xmm1, src2; load the 2nd pair of complex values, ; i.e. d1, c1, d0, c0
+	mulps xmm0, xmm1; temporary results, a1d1, a1c1, a0d0, ; a0c0
+	shufps xmm1, xmm1, b1; reorder the real and imaginary ; parts, c1, d1, c0, d0
+	movshdup xmm2, Src1; load the imaginary parts into the ; destination, b1, b1, b0, b0
+	mulps xmm2, xmm1; temporary results, b1c1, b1d1, b0c0, ; b0d0
+	addsubps xmm0, xmm2; b1c1+a1d1, a1c1 -b1d1, b0c0+a0d
+	VSHUFPD (VEX.256 encoded version)
+	IF IMM0[0] = 0
+	THEN DEST[63:0]=SRC1[63:0] ELSE DEST[63:0]=SRC1[127:64] FI;
+	IF IMM0[1] = 0
+	THEN DEST[127:64]=SRC2[63:0] ELSE DEST[127:64]=SRC2[127:64] FI;
+	IF IMM0[2] = 0
+	THEN DEST[191:128]=SRC1[191:128] ELSE DEST[191:128]=SRC1[255:192] FI;
+	IF IMM0[3] = 0
+	THEN DEST[255:192]=SRC2[191:128] ELSE DEST[255:192]=SRC2[255:192] FI; // Ox5 r<->i   ; 0xC unchanged
+      */
+      
+      __m256d ymm0,ymm1,ymm2;
+      ymm0 = _mm256_shuffle_pd(a,a,0x0); // ymm0 <- ar ar, ar,ar b'00,00
+      ymm0 = _mm256_mul_pd(ymm0,b);      // ymm0 <- ar bi, ar br
+      ymm1 = _mm256_shuffle_pd(b,b,0x5); // ymm1 <- br,bi  b'01,01 
+      ymm2 = _mm256_shuffle_pd(a,a,0xF); // ymm2 <- ai,ai  b'11,11
+      ymm1 = _mm256_mul_pd(ymm1,ymm2);   // ymm1 <- br ai, ai bi
+      return _mm256_addsub_pd(ymm0,ymm1);
+    }
+  };
+
+  struct Mult{
+    // Real float
+    inline __m256 operator()(__m256 a, __m256 b){
+      return _mm256_mul_ps(a,b);
+    }
+    // Real double
+    inline __m256d operator()(__m256d a, __m256d b){
+      return _mm256_mul_pd(a,b);
+    }
+    // Integer
+    inline __m256i operator()(__m256i a, __m256i b){
+#if defined (AVX1) 
+      __m128i a0,a1;
+      __m128i b0,b1;
+      a0 = _mm256_extractf128_si256(a,0);
+      b0 = _mm256_extractf128_si256(b,0);
+      a1 = _mm256_extractf128_si256(a,1);
+      b1 = _mm256_extractf128_si256(b,1);
+      a0 = _mm_mul_epi32(a0,b0);
+      a1 = _mm_mul_epi32(a1,b1);
+      return _mm256_set_m128i(a1,a0);
+#endif
+#if defined (AVX2)
+      return _mm256_mul_epi32(a,b);
+#endif
+
+    }
+  };
+
+
+  struct Conj{
+    // Complex single
+    inline __m256 operator()(__m256 in){
+      return _mm256_xor_ps(_mm256_addsub_ps(_mm256_setzero_ps(),in), _mm256_set1_ps(-0.f));
+    }
+    // Complex double
+    inline __m256d operator()(__m256d in){
+      return _mm256_xor_pd(_mm256_addsub_pd(_mm256_setzero_pd(),in), _mm256_set1_pd(-0.f));//untested
+      /*
+	// original 
+	//      addsubps 0, inv=>0+in.v[3] 0-in.v[2], 0+in.v[1], 0-in.v[0], ...
+	__m256d tmp = _mm256_addsub_pd(_mm256_setzero_pd(),_mm256_shuffle_pd(in,in,0x5));
+	return _mm256_shuffle_pd(tmp,tmp,0x5);
+      */
+    }
+    // do not define for integer input
+  };
+
+  struct TimesMinusI{
+    //Complex single
+    inline __m256 operator()(__m256 in, __m256 ret){
+      __m256 tmp =_mm256_addsub_ps(_mm256_setzero_ps(),in);   // r,-i
+      return _mm256_shuffle_ps(tmp,tmp,_MM_SHUFFLE(2,3,0,1)); //-i,r
+    }
+    //Complex double
+    inline __m256d operator()(__m256d in, __m256d ret){
+      __m256d tmp = _mm256_addsub_pd(_mm256_setzero_pd(),in); // r,-i
+      return _mm256_shuffle_pd(tmp,tmp,0x5);
+    }
+  };
+
+  struct TimesI{
+    //Complex single
+    inline __m256 operator()(__m256 in, __m256 ret){
+      __m256 tmp =_mm256_shuffle_ps(in,in,_MM_SHUFFLE(2,3,0,1)); // i,r
+      return _mm256_addsub_ps(_mm256_setzero_ps(),tmp);          // i,-r
+    }
+    //Complex double
+    inline __m256d operator()(__m256d in, __m256d ret){
+      __m256d tmp = _mm256_shuffle_pd(in,in,0x5);
+      return _mm256_addsub_pd(_mm256_setzero_pd(),tmp); // i,-r
+    }
+  };
+
+
+  
+
+
+  //////////////////////////////////////////////
+  // Some Template specialization
+  template < typename vtype > 
+    void permute(vtype a, vtype b, int perm) {
+    union { 
+      __m256 f;
+      vtype v;
+    } conv;
+    conv.v = b;
+    switch (perm){
+      // 8x32 bits=>3 permutes
+    case 2: 
+      conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); 
+      break;
+    case 1: conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2)); break;
+    case 0: conv.f = _mm256_permute2f128_ps(conv.f,conv.f,0x01); break;
+    default: assert(0); break;
+    }
+    a = conv.v;
+    
+  }
+  
+  //Complex float Reduce
+  template<>
+    inline Grid::ComplexF Reduce<Grid::ComplexF, __m256>::operator()(__m256 in){
+    __m256 v1,v2;
+    Optimization::permute(v1,in,0); // sse 128; paired complex single
+    v1 = _mm256_add_ps(v1,in);
+    Optimization::permute(v2,v1,1); // avx 256; quad complex single
+    v1 = _mm256_add_ps(v1,v2);
+    return Grid::ComplexF(v1[0],v1[1]);
+  }
+  //Real float Reduce
+  template<>
+  inline Grid::RealF Reduce<Grid::RealF, __m256>::operator()(__m256 in){
+    __m256 v1,v2;
+    Optimization::permute(v1,in,0); // avx 256; octo-double
+    v1 = _mm256_add_ps(v1,in);
+    Optimization::permute(v2,v1,1); 
+    v1 = _mm256_add_ps(v1,v2);
+    Optimization::permute(v2,v1,2); 
+    v1 = _mm256_add_ps(v1,v2);
+    return v1[0];
+  }
+  
+  
+  //Complex double Reduce
+  template<>
+  inline Grid::ComplexD Reduce<Grid::ComplexD, __m256d>::operator()(__m256d in){
+    __m256d v1;
+    Optimization::permute(v1,in,0); // sse 128; paired complex single
+    v1 = _mm256_add_pd(v1,in);
+    return Grid::ComplexD(v1[0],v1[1]);
+  }
+  
+  //Real double Reduce
+  template<>
+  inline Grid::RealD Reduce<Grid::RealD, __m256d>::operator()(__m256d in){
+    __m256d v1,v2;
+    Optimization::permute(v1,in,0); // avx 256; quad double
+    v1 = _mm256_add_pd(v1,in);
+    Optimization::permute(v2,v1,1); 
+    v1 = _mm256_add_pd(v1,v2);
+    return v1[0];
+  }
+
+  //Integer Reduce
+  template<>
+  inline Integer Reduce<Integer, __m256i>::operator()(__m256i in){
+    // FIXME unimplemented
+    printf("Reduce : Missing integer implementation -> FIX\n");
+    assert(0);
+  }
+  
+  
+}
+
+//////////////////////////////////////////////////////////////////////////////////////
+// Here assign types 
+namespace Grid {
+  typedef __m256  SIMD_Ftype; // Single precision type
+  typedef __m256d SIMD_Dtype; // Double precision type
+  typedef __m256i SIMD_Itype; // Integer type
+
+
+  // Function name aliases
+  typedef Optimization::Vsplat   VsplatSIMD;
+  typedef Optimization::Vstore   VstoreSIMD;
+  typedef Optimization::Vset     VsetSIMD;
+  typedef Optimization::Vstream  VstreamSIMD;
+  template <typename S, typename T> using ReduceSIMD = Optimization::Reduce<S,T>;
+
+
+  // Arithmetic operations
+  typedef Optimization::Sum         SumSIMD;
+  typedef Optimization::Sub         SubSIMD;
+  typedef Optimization::Mult        MultSIMD;
+  typedef Optimization::MultComplex MultComplexSIMD;
+  typedef Optimization::Conj        ConjSIMD;
+  typedef Optimization::TimesMinusI TimesMinusISIMD;
+  typedef Optimization::TimesI      TimesISIMD;
+
+}
diff --git a/lib/simd/Grid_knc.h b/lib/simd/Grid_knc.h
new file mode 100644
index 00000000..daec3973
--- /dev/null
+++ b/lib/simd/Grid_knc.h
@@ -0,0 +1,323 @@
+//----------------------------------------------------------------------
+/*! @file Grid_knc.h
+  @brief Optimization libraries for AVX512 instructions set for KNC
+
+  Using intrinsics
+*/
+// Time-stamp: <2015-05-22 17:12:44 neo>
+//----------------------------------------------------------------------
+
+#include <immintrin.h>
+#ifndef KNC_ONLY_STORES
+#define  _mm512_storenrngo_ps _mm512_store_ps  // not present in AVX512
+#define  _mm512_storenrngo_pd _mm512_store_pd  // not present in AVX512
+#endif
+
+
+namespace Optimization {
+  
+  struct Vsplat{
+    //Complex float
+    inline __m512 operator()(float a, float b){
+      return _mm512_set_ps(b,a,b,a,b,a,b,a,b,a,b,a,b,a,b,a);
+    }
+    // Real float
+    inline __m512 operator()(float a){
+      return _mm512_set1_ps(a);
+    }
+    //Complex double
+    inline __m512d operator()(double a, double b){
+      return _mm512_set_pd(b,a,b,a,b,a,b,a);
+    }
+    //Real double
+    inline __m512d operator()(double a){
+      return _mm512_set1_pd(a);
+    }
+    //Integer
+    inline __m512i operator()(Integer a){
+      return _mm512_set1_epi32(a);
+    }
+  };
+
+  struct Vstore{
+    //Float 
+    inline void operator()(__m512 a, float* F){
+      _mm512_store_ps(F,a);
+    }
+    //Double
+    inline void operator()(__m512d a, double* D){
+      _mm512_store_pd(D,a);
+    }
+    //Integer
+    inline void operator()(__m512i a, Integer* I){
+      _mm512_store_si512((__m512i *)I,a);
+    }
+
+  };
+
+
+  struct Vstream{
+    //Float
+    inline void operator()(float * a, __m512 b){
+      _mm512_storenrngo_ps(a,b);
+    }
+    //Double
+    inline void operator()(double * a, __m512d b){
+      _mm512_storenrngo_pd(a,b);
+    }
+
+
+  };
+
+
+
+  struct Vset{
+    // Complex float 
+    inline __m512 operator()(Grid::ComplexF *a){
+      return _mm512_set_ps(a[7].imag(),a[7].real(),a[6].imag(),a[6].real(),
+			   a[5].imag(),a[5].real(),a[4].imag(),a[4].real(),
+			   a[3].imag(),a[3].real(),a[2].imag(),a[2].real(),
+			   a[1].imag(),a[1].real(),a[0].imag(),a[0].real());
+    }
+    // Complex double 
+    inline __m512d operator()(Grid::ComplexD *a){
+      return _mm512_set_pd(a[3].imag(),a[3].real(),a[2].imag(),a[2].real(),
+			   a[1].imag(),a[1].real(),a[0].imag(),a[0].real());
+    }
+    // Real float 
+    inline __m512 operator()(float *a){
+      return _mm512_set_ps( a[15],a[14],a[13],a[12],a[11],a[10],a[9],a[8],
+			    a[7],a[6],a[5],a[4],a[3],a[2],a[1],a[0]);
+    }
+    // Real double
+    inline __m512d operator()(double *a){
+      return _mm512_set_pd(a[7],a[6],a[5],a[4],a[3],a[2],a[1],a[0]);
+    }
+    // Integer
+    inline __m512i operator()(Integer *a){
+      return _mm512_set_epi32( a[15],a[14],a[13],a[12],a[11],a[10],a[9],a[8],
+			       a[7],a[6],a[5],a[4],a[3],a[2],a[1],a[0]);
+    }
+
+
+  };
+
+  template <typename Out_type, typename In_type>
+  struct Reduce{
+    //Need templated class to overload output type
+    //General form must generate error if compiled
+    inline Out_type operator()(In_type in){
+      printf("Error, using wrong Reduce function\n");
+      exit(1);
+      return 0;
+    }
+  };
+
+
+ 
+
+  /////////////////////////////////////////////////////
+  // Arithmetic operations
+  /////////////////////////////////////////////////////
+  struct Sum{
+    //Complex/Real float
+    inline __m512 operator()(__m512 a, __m512 b){
+      return _mm512_add_ps(a,b);
+    }
+    //Complex/Real double
+    inline __m512d operator()(__m512d a, __m512d b){
+      return _mm512_add_pd(a,b);
+    }
+    //Integer
+    inline __m512i operator()(__m512i a, __m512i b){
+      return _mm512_add_epi32(a,b);
+    }
+  };
+
+  struct Sub{
+    //Complex/Real float
+    inline __m512 operator()(__m512 a, __m512 b){
+      return _mm512_sub_ps(a,b);
+    }
+    //Complex/Real double
+    inline __m512d operator()(__m512d a, __m512d b){
+      return _mm512_sub_pd(a,b);
+    }
+    //Integer
+    inline __m512i operator()(__m512i a, __m512i b){
+      return _mm512_sub_epi32(a,b);
+    }
+  };
+
+
+  struct MultComplex{
+    // Complex float
+    inline __m512 operator()(__m512 a, __m512 b){
+      __m512 vzero,ymm0,ymm1,real, imag;
+      vzero = _mm512_setzero_ps();
+      ymm0  = _mm512_swizzle_ps(a, _MM_SWIZ_REG_CDAB); // 
+      real  = (__m512)_mm512_mask_or_epi32((__m512i)a, 0xAAAA,(__m512i)vzero,(__m512i)ymm0);
+      imag  = _mm512_mask_sub_ps(a, 0x5555,vzero, ymm0);
+      ymm1  = _mm512_mul_ps(real, b);
+      ymm0  = _mm512_swizzle_ps(b, _MM_SWIZ_REG_CDAB); // OK
+      return _mm512_fmadd_ps(ymm0,imag,ymm1);
+    }
+    // Complex double
+    inline __m512d operator()(__m512d a, __m512d b){
+      /* This is from
+       * Automatic SIMD Vectorization of Fast Fourier Transforms for the Larrabee and AVX Instruction Sets 
+       * @inproceedings{McFarlin:2011:ASV:1995896.1995938,
+       * author = {McFarlin, Daniel S. and Arbatov, Volodymyr and Franchetti, Franz and P\"{u}schel, Markus},
+       * title = {Automatic SIMD Vectorization of Fast Fourier Transforms for the Larrabee and AVX Instruction Sets},
+       * booktitle = {Proceedings of the International Conference on Supercomputing},
+       * series = {ICS '11},
+       * year = {2011},
+       * isbn = {978-1-4503-0102-2},
+       * location = {Tucson, Arizona, USA},
+       * pages = {265--274},
+       * numpages = {10},
+       * url = {http://doi.acm.org/10.1145/1995896.1995938},
+       * doi = {10.1145/1995896.1995938},
+       * acmid = {1995938},
+       * publisher = {ACM},
+       * address = {New York, NY, USA},
+       * keywords = {autovectorization, fourier transform, program generation, simd, super-optimization},
+       *                } 
+       */
+      __m512d vzero,ymm0,ymm1,real,imag;
+      vzero =_mm512_setzero_pd();
+      ymm0 =  _mm512_swizzle_pd(a, _MM_SWIZ_REG_CDAB); // 
+      real =(__m512d)_mm512_mask_or_epi64((__m512i)a, 0xAA,(__m512i)vzero,(__m512i) ymm0);
+      imag =  _mm512_mask_sub_pd(a, 0x55,vzero, ymm0);
+      ymm1 =  _mm512_mul_pd(real, b);
+      ymm0 =  _mm512_swizzle_pd(b, _MM_SWIZ_REG_CDAB); // OK
+      return  _mm512_fmadd_pd(ymm0,imag,ymm1);
+    }
+  };
+  
+  struct Mult{
+    // Real float
+    inline __m512 operator()(__m512 a, __m512 b){
+      return _mm512_mul_ps(a,b);
+    }
+    // Real double
+    inline __m512d operator()(__m512d a, __m512d b){
+      return _mm512_mul_pd(a,b);
+    }
+    // Integer
+    inline __m512i operator()(__m512i a, __m512i b){
+      return _mm512_mullo_epi32(a,b);
+    }
+  };
+
+
+  struct Conj{
+    // Complex single
+    inline __m512 operator()(__m512 in){
+      return _mm512_mask_sub_ps(in,0xaaaa,_mm512_setzero_ps(),in); // Zero out 0+real 0-imag  
+    }
+    // Complex double
+    inline __m512d operator()(__m512d in){
+      return _mm512_mask_sub_pd(in, 0xaa,_mm512_setzero_pd(), in);
+    }
+    // do not define for integer input
+  };
+
+  struct TimesMinusI{
+    //Complex single
+    inline __m512 operator()(__m512 in, __m512 ret){
+      __m512 tmp = _mm512_mask_sub_ps(in,0xaaaa,_mm512_setzero_ps(),in); // real -imag 
+      return _mm512_swizzle_ps(tmp, _MM_SWIZ_REG_CDAB);// OK
+    }
+    //Complex double
+    inline __m512d operator()(__m512d in, __m512d ret){
+      __m512d tmp = _mm512_mask_sub_pd(in,0xaa,_mm512_setzero_pd(),in); // real -imag 
+      return  _mm512_swizzle_pd(tmp, _MM_SWIZ_REG_CDAB);// OK
+    }
+
+
+  };
+
+  struct TimesI{
+    //Complex single
+    inline __m512 operator()(__m512 in, __m512 ret){
+      __m512 tmp = _mm512_swizzle_ps(in, _MM_SWIZ_REG_CDAB);// OK
+      return _mm512_mask_sub_ps(tmp,0xaaaa,_mm512_setzero_ps(),tmp); // real -imag
+    }
+    //Complex double
+    inline __m512d operator()(__m512d in, __m512d ret){
+      __m512d tmp = _mm512_swizzle_pd(in, _MM_SWIZ_REG_CDAB);// OK
+      return _mm512_mask_sub_pd(tmp,0xaa,_mm512_setzero_pd(),tmp); // real -imag
+    }
+
+
+  };
+
+
+  
+
+
+  //////////////////////////////////////////////
+  // Some Template specialization
+  
+  //Complex float Reduce
+  template<>
+  inline Grid::ComplexF Reduce<Grid::ComplexF, __m512>::operator()(__m512 in){
+    return Grid::ComplexF(_mm512_mask_reduce_add_ps(0x5555, in),_mm512_mask_reduce_add_ps(0xAAAA, in));
+  }
+  //Real float Reduce
+  template<>
+  inline Grid::RealF Reduce<Grid::RealF, __m512>::operator()(__m512 in){
+    return _mm512_reduce_add_ps(in);
+  }
+  
+  
+  //Complex double Reduce
+  template<>
+  inline Grid::ComplexD Reduce<Grid::ComplexD, __m512d>::operator()(__m512d in){
+    return Grid::ComplexD(_mm512_mask_reduce_add_pd(0x55, in),_mm512_mask_reduce_add_pd(0xAA, in));
+  }
+  
+  //Real double Reduce
+  template<>
+  inline Grid::RealD Reduce<Grid::RealD, __m512d>::operator()(__m512d in){
+    return _mm512_reduce_add_pd(in);
+  }
+
+  //Integer Reduce
+  template<>
+  inline Integer Reduce<Integer, __m512i>::operator()(__m512i in){
+    // FIXME unimplemented
+    printf("Reduce : Missing integer implementation -> FIX\n");
+    assert(0);
+  }
+  
+  
+}
+
+//////////////////////////////////////////////////////////////////////////////////////
+// Here assign types 
+namespace Grid {
+  typedef __m512 SIMD_Ftype;  // Single precision type
+  typedef __m512d SIMD_Dtype; // Double precision type
+  typedef __m512i SIMD_Itype; // Integer type
+
+
+  // Function name aliases
+  typedef Optimization::Vsplat   VsplatSIMD;
+  typedef Optimization::Vstore   VstoreSIMD;
+  typedef Optimization::Vset     VsetSIMD;
+  typedef Optimization::Vstream  VstreamSIMD;
+  template <typename S, typename T> using ReduceSIMD = Optimization::Reduce<S,T>;
+
+
+  // Arithmetic operations
+  typedef Optimization::Sum         SumSIMD;
+  typedef Optimization::Sub         SubSIMD;
+  typedef Optimization::Mult        MultSIMD;
+  typedef Optimization::MultComplex MultComplexSIMD;
+  typedef Optimization::Conj        ConjSIMD;
+  typedef Optimization::TimesMinusI TimesMinusISIMD;
+  typedef Optimization::TimesI      TimesISIMD;
+
+}
diff --git a/lib/simd/Grid_qpx.h b/lib/simd/Grid_qpx.h
new file mode 100644
index 00000000..dc0251f8
--- /dev/null
+++ b/lib/simd/Grid_qpx.h
@@ -0,0 +1,272 @@
+//----------------------------------------------------------------------
+/*! @file Grid_qpx.h
+  @brief Optimization libraries for QPX instructions set for BG/Q
+
+  Using intrinsics
+*/
+// Time-stamp: <2015-05-22 17:29:26 neo>
+//----------------------------------------------------------------------
+
+// lot of undefined functions
+
+namespace Optimization {
+  
+  struct Vsplat{
+    //Complex float
+    inline float operator()(float a, float b){
+      return {a,b,a,b};
+    }
+    // Real float
+    inline float operator()(float a){
+      return {a,a,a,a};
+    }
+    //Complex double
+    inline vector4double operator()(double a, double b){
+      return {a,b,a,b};
+    }
+    //Real double
+    inline vector4double operator()(double a){
+      return {a,a,a,a};
+    }
+    //Integer
+    inline int operator()(Integer a){
+#error
+    }
+  };
+
+  struct Vstore{
+    //Float 
+    inline void operator()(float a, float* F){
+      assert(0);
+    }
+    //Double
+    inline void operator()(vector4double a, double* D){
+      assert(0);
+    }
+    //Integer
+    inline void operator()(int a, Integer* I){
+      assert(0);
+    }
+
+  };
+
+
+  struct Vstream{
+    //Float
+    inline void operator()(float * a, float b){
+      assert(0);
+    }
+    //Double
+    inline void operator()(double * a, vector4double b){
+      assert(0);
+    }
+
+
+  };
+
+
+
+  struct Vset{
+    // Complex float 
+    inline float operator()(Grid::ComplexF *a){
+      return {a[0].real(),a[0].imag(),a[1].real(),a[1].imag(),a[2].real(),a[2].imag(),a[3].real(),a[3].imag()};
+    }
+    // Complex double 
+    inline vector4double operator()(Grid::ComplexD *a){
+      return {a[0].real(),a[0].imag(),a[1].real(),a[1].imag(),a[2].real(),a[2].imag(),a[3].real(),a[3].imag()};
+    }
+    // Real float 
+    inline float operator()(float *a){
+      return {a[0],a[1],a[2],a[3],a[4],a[5],a[6],a[7]};
+    }
+    // Real double
+    inline vector4double operator()(double *a){
+      return {a[0],a[1],a[2],a[3],a[4],a[5],a[6],a[7]};
+    }
+    // Integer
+    inline int operator()(Integer *a){
+#error
+    }
+
+
+  };
+
+  template <typename Out_type, typename In_type>
+    struct Reduce{
+      //Need templated class to overload output type
+      //General form must generate error if compiled
+      inline Out_type operator()(In_type in){
+	printf("Error, using wrong Reduce function\n");
+	exit(1);
+	return 0;
+      }
+    };
+
+
+ 
+
+  /////////////////////////////////////////////////////
+  // Arithmetic operations
+  /////////////////////////////////////////////////////
+  struct Sum{
+    //Complex/Real float
+    inline float operator()(float a, float b){
+#error
+    }
+    //Complex/Real double
+    inline vector4double operator()(vector4double a, vector4double b){
+      return vec_add(a,b);
+    }
+    //Integer
+    inline int operator()(int a, int b){
+#error
+    }
+  };
+
+  struct Sub{
+    //Complex/Real float
+    inline float operator()(float a, float b){
+#error
+    }
+    //Complex/Real double
+    inline vector4double operator()(vector4double a, vector4double b){
+#error
+    }
+    //Integer
+    inline floati operator()(int a, int b){
+#error
+    }
+  };
+
+
+  struct MultComplex{
+    // Complex float
+    inline float operator()(float a, float b){
+#error
+    }
+    // Complex double
+    inline vector4double operator()(vector4double a, vector4double b){
+#error
+    }
+  };
+
+  struct Mult{
+    // Real float
+    inline float operator()(float a, float b){
+#error
+    }
+    // Real double
+    inline vector4double operator()(vector4double a, vector4double b){
+#error
+    }
+    // Integer
+    inline int operator()(int a, int b){
+#error
+    }
+  };
+
+
+  struct Conj{
+    // Complex single
+    inline float operator()(float in){
+      assert(0);
+    }
+    // Complex double
+    inline vector4double operator()(vector4double in){
+      assert(0);
+    }
+    // do not define for integer input
+  };
+
+  struct TimesMinusI{
+    //Complex single
+    inline float operator()(float in, float ret){
+      assert(0);
+    }
+    //Complex double
+    inline vector4double operator()(vector4double in, vector4double ret){
+      assert(0);
+    }
+
+
+  };
+
+  struct TimesI{
+    //Complex single
+    inline float operator()(float in, float ret){
+  
+    }
+    //Complex double
+    inline vector4double operator()(vector4double in, vector4double ret){
+  
+    }
+
+
+  };
+
+
+  
+
+
+  //////////////////////////////////////////////
+  // Some Template specialization
+  
+  //Complex float Reduce
+  template<>
+    inline Grid::ComplexF Reduce<Grid::ComplexF, float>::operator()(float in){
+    assert(0);
+  }
+  //Real float Reduce
+  template<>
+    inline Grid::RealF Reduce<Grid::RealF, float>::operator()(float in){
+    assert(0);
+  }
+  
+  
+  //Complex double Reduce
+  template<>
+    inline Grid::ComplexD Reduce<Grid::ComplexD, vector4double>::operator()(vector4double in){
+    assert(0);
+  }
+  
+  //Real double Reduce
+  template<>
+    inline Grid::RealD Reduce<Grid::RealD, vector4double>::operator()(vector4double in){
+    assert(0);
+  }
+
+  //Integer Reduce
+  template<>
+    inline Integer Reduce<Integer, floati>::operator()(float in){
+    assert(0);
+  }
+  
+  
+}
+
+//////////////////////////////////////////////////////////////////////////////////////
+// Here assign types 
+namespace Grid {
+  typedef float SIMD_Ftype  __attribute__ ((vector_size (16)));         // Single precision type
+  typedef vector4double SIMD_Dtype; // Double precision type
+  typedef int SIMD_Itype;           // Integer type
+
+
+  // Function name aliases
+  typedef Optimization::Vsplat   VsplatSIMD;
+  typedef Optimization::Vstore   VstoreSIMD;
+  typedef Optimization::Vset     VsetSIMD;
+  typedef Optimization::Vstream  VstreamSIMD;
+  template <typename S, typename T> using ReduceSIMD = Optimization::Reduce<S,T>;
+
+
+  // Arithmetic operations
+  typedef Optimization::Sum         SumSIMD;
+  typedef Optimization::Sub         SubSIMD;
+  typedef Optimization::Mult        MultSIMD;
+  typedef Optimization::MultComplex MultComplexSIMD;
+  typedef Optimization::Conj        ConjSIMD;
+  typedef Optimization::TimesMinusI TimesMinusISIMD;
+  typedef Optimization::TimesI      TimesISIMD;
+
+}
diff --git a/lib/simd/Grid_sse4.h b/lib/simd/Grid_sse4.h
index ddc3490b..fa0e3ec3 100644
--- a/lib/simd/Grid_sse4.h
+++ b/lib/simd/Grid_sse4.h
@@ -4,7 +4,7 @@
 
   Using intrinsics
 */
-// Time-stamp: <2015-05-20 16:45:39 neo>
+// Time-stamp: <2015-05-21 18:06:30 neo>
 //----------------------------------------------------------------------
 
 #include <pmmintrin.h>
@@ -53,12 +53,12 @@ namespace Optimization {
 
   struct Vstream{
     //Float
-    inline void operator()(__m128 a, __m128 b){
-      _mm_stream_ps((float *)&a,b);
+    inline void operator()(float * a, __m128 b){
+      _mm_stream_ps(a,b);
     }
     //Double
-    inline void operator()(__m128d a, __m128d b){
-      _mm_stream_pd((double *)&a,b);
+    inline void operator()(double * a, __m128d b){
+      _mm_stream_pd(a,b);
     }
 
 
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index 57480621..34c63964 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -2,12 +2,20 @@
 /*! @file Grid_vector_types.h
   @brief Defines templated class Grid_simd to deal with inner vector types
 */
-// Time-stamp: <2015-05-20 17:31:55 neo>
+// Time-stamp: <2015-05-22 17:08:19 neo>
 //---------------------------------------------------------------------------
 #ifndef GRID_VECTOR_TYPES
 #define GRID_VECTOR_TYPES
 
+#ifdef SSE4
 #include "Grid_sse4.h"
+#endif
+#if defined (AVX1)|| defined (AVX2)
+#include "Grid_avx.h"
+#endif
+#if defined AVX512
+#include "Grid_knc.h"
+#endif
 
 
 namespace Grid {
@@ -43,13 +51,14 @@ namespace Grid {
   // general forms to allow for vsplat syntax
   // need explicit declaration of types when used since
   // clang cannot automatically determine the output type sometimes
+  // use decltype?
   template < class Out, class Input1, class Input2, class Operation > 
     Out binary(Input1 src_1, Input2 src_2, Operation op){
     return op(src_1, src_2);
   } 
 
-  template < class SIMDout, class Input, class Operation > 
-    SIMDout unary(Input src, Operation op){
+  template < class Out, class Input, class Operation > 
+    Out unary(Input src, Operation op){
     return op(src);
   } 
 
@@ -63,27 +72,34 @@ namespace Grid {
     
   public:
     typedef typename RealPart < Scalar_type >::type Real; 
+    typedef Vector_type     vector_type;
+    typedef Scalar_type     scalar_type;
+   
     Vector_type v;
     
-
+    
     static inline int Nsimd(void) { return sizeof(Vector_type)/sizeof(Scalar_type);}
-
+    
     // Constructors
     Grid_simd & operator = ( Zero & z){
       vzero(*this);
       return (*this);
     }
-    Grid_simd(){};
-    
     
+    Grid_simd& operator=(const Grid_simd&& rhs){v=rhs.v;return *this;};
+    Grid_simd& operator=(const Grid_simd& rhs){v=rhs.v;return *this;}; //faster than not declaring it and leaving to the compiler
+    Grid_simd()=default; 
+    Grid_simd(const Grid_simd& rhs):v(rhs.v){};    //compiles in movaps
+    Grid_simd(const Grid_simd&& rhs):v(rhs.v){};  
+  
     //Enable if complex type
     template < class S = Scalar_type > 
-    Grid_simd(typename std::enable_if< is_complex < S >::value, S>::type a){
+    Grid_simd(const typename std::enable_if< is_complex < S >::value, S>::type a){
       vsplat(*this,a);
     };
     
 
-    Grid_simd(Real a){
+    Grid_simd(const Real a){
       vsplat(*this,Scalar_type(a));
     };
 
@@ -97,18 +113,13 @@ namespace Grid {
     friend inline void sub (Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) - (*r); }
     friend inline void add (Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) + (*r); }
 
-    //not for integer types... FIXME
+    //not for integer types... 
+    template <  class S = Scalar_type, NotEnableIf<std::is_integral < S >, int> = 0 > 
     friend inline Grid_simd adj(const Grid_simd &in){ return conjugate(in); }
         
     ///////////////////////////////////////////////
     // Initialise to 1,0,i for the correct types
     ///////////////////////////////////////////////
-    // if not complex overload here 
-    template <  class S = Scalar_type, NotEnableIf<is_complex < S >,int> = 0 > 
-      friend inline void vone(Grid_simd &ret)      { vsplat(ret,1.0); }
-    template <  class S = Scalar_type, NotEnableIf<is_complex < S >,int> = 0 > 
-      friend inline void vzero(Grid_simd &ret)     { vsplat(ret,0.0); }
-    
     // For complex types
     template <  class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 
       friend inline void vone(Grid_simd &ret)      { vsplat(ret,1.0,0.0); }
@@ -116,6 +127,14 @@ namespace Grid {
       friend inline void vzero(Grid_simd &ret)     { vsplat(ret,0.0,0.0); }// use xor?
     template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 
       friend inline void vcomplex_i(Grid_simd &ret){ vsplat(ret,0.0,1.0);} 
+
+    // if not complex overload here 
+    template <  class S = Scalar_type, EnableIf<std::is_floating_point < S >,int> = 0 > 
+      friend inline void vone(Grid_simd &ret)      { vsplat(ret,1.0); }
+    template <  class S = Scalar_type, EnableIf<std::is_floating_point < S >,int> = 0 > 
+      friend inline void vzero(Grid_simd &ret)     { vsplat(ret,0.0); }
+    
+
    
     // For integral types
     template <  class S = Scalar_type, EnableIf<std::is_integral < S >, int> = 0 > 
@@ -125,7 +144,7 @@ namespace Grid {
     template <  class S = Scalar_type, EnableIf<std::is_integral < S >, int> = 0 > 
       friend inline void vtrue (Grid_simd &ret){vsplat(ret,0xFFFFFFFF);}
     template <  class S = Scalar_type, EnableIf<std::is_integral < S >, int> = 0 > 
-      friend inline void vfalse(vInteger &ret){vsplat(ret,0);}
+      friend inline void vfalse(Grid_simd &ret){vsplat(ret,0);}
 
 
    
@@ -206,8 +225,9 @@ namespace Grid {
     ///////////////////////
     // Vstream
     ///////////////////////
+    template <  class S = Scalar_type, NotEnableIf<std::is_integral < S >, int> = 0 > 
     friend inline void vstream(Grid_simd &out,const Grid_simd &in){
-      binary<void>(out.v, in.v, VstreamSIMD());
+      binary<void>((Real*)&out.v, in.v, VstreamSIMD());
     }
 
     template <  class S = Scalar_type, EnableIf<std::is_integral < S >, int> = 0 > 
@@ -305,7 +325,7 @@ namespace Grid {
     // Unary negation
     ///////////////////////
     friend inline Grid_simd operator -(const Grid_simd &r) {
-      vComplexF ret;
+      Grid_simd ret;
       vzero(ret);
       ret = ret - r;
       return ret;
@@ -350,7 +370,7 @@ namespace Grid {
   }
 
   template<class scalar_type, class vector_type >
-  inline void zeroit(Grid_simd< scalar_type, vector_type> &z){ vzero(z);}
+    inline void zeroit(Grid_simd< scalar_type, vector_type> &z){ vzero(z);}
 
 
   template<class scalar_type, class vector_type >
@@ -368,35 +388,11 @@ namespace Grid {
 
   // Define available types (now change names to avoid clashing with the rest of the code)
 
-  typedef Grid_simd< float                 , SIMD_Ftype > MyRealF;
-  typedef Grid_simd< double                , SIMD_Dtype > MyRealD;
-  typedef Grid_simd< std::complex< float > , SIMD_Ftype > MyComplexF;
-  typedef Grid_simd< std::complex< double >, SIMD_Dtype > MyComplexD;
-
-
-
-
-  ////////////////////////////////////////////////////////////////////
-  // Temporary hack to keep independent from the rest of the code
-  template<> struct isGridTensor<MyRealD > {
-    static const bool value = false;
-    static const bool notvalue = true;
-  };
-  template<> struct isGridTensor<MyRealF > {
-    static const bool value = false;
-    static const bool notvalue = true;
-  };
-  template<> struct isGridTensor<MyComplexD > {
-    static const bool value = false;
-    static const bool notvalue = true;
-  };
-  template<> struct isGridTensor<MyComplexF > {
-    static const bool value = false;
-    static const bool notvalue = true;
-  };
-
-
-
+  typedef Grid_simd< float                 , SIMD_Ftype > vRealF;
+  typedef Grid_simd< double                , SIMD_Dtype > vRealD;
+  typedef Grid_simd< std::complex< float > , SIMD_Ftype > vComplexF;
+  typedef Grid_simd< std::complex< double >, SIMD_Dtype > vComplexD;
+  typedef Grid_simd< Integer               , SIMD_Itype > vInteger;
 
 }
 
diff --git a/lib/simd/Grid_vComplexD.h b/lib/simd/Old/Grid_vComplexD.h
similarity index 100%
rename from lib/simd/Grid_vComplexD.h
rename to lib/simd/Old/Grid_vComplexD.h
diff --git a/lib/simd/Grid_vComplexF.h b/lib/simd/Old/Grid_vComplexF.h
similarity index 99%
rename from lib/simd/Grid_vComplexF.h
rename to lib/simd/Old/Grid_vComplexF.h
index 713bbdd8..37760b98 100644
--- a/lib/simd/Grid_vComplexF.h
+++ b/lib/simd/Old/Grid_vComplexF.h
@@ -54,7 +54,7 @@ namespace Grid {
         //////////////////////////////////
         friend inline void vone(vComplexF &ret)      { vsplat(ret,1.0,0.0); }
         friend inline void vzero(vComplexF &ret)     { vsplat(ret,0.0,0.0); }
-        friend inline void vcomplex_i(vComplexF &ret){ vsplat(ret,0.0,1.0);}
+        friend inline void vcomplex_i(vComplexF &ret){ vsplat(ret,0.0,1.0); }
           
         ////////////////////////////////////
         // Arithmetic operator overloads +,-,*
diff --git a/lib/simd/Grid_vInteger.h b/lib/simd/Old/Grid_vInteger.h
similarity index 100%
rename from lib/simd/Grid_vInteger.h
rename to lib/simd/Old/Grid_vInteger.h
diff --git a/lib/simd/Grid_vRealD.h b/lib/simd/Old/Grid_vRealD.h
similarity index 100%
rename from lib/simd/Grid_vRealD.h
rename to lib/simd/Old/Grid_vRealD.h
diff --git a/lib/simd/Grid_vRealF.h b/lib/simd/Old/Grid_vRealF.h
similarity index 100%
rename from lib/simd/Grid_vRealF.h
rename to lib/simd/Old/Grid_vRealF.h
diff --git a/tests/Grid_main.cc b/tests/Grid_main.cc
index 10e74099..c1f1c5d0 100644
--- a/tests/Grid_main.cc
+++ b/tests/Grid_main.cc
@@ -103,6 +103,9 @@ int main (int argc, char ** argv)
     random(FineRNG,scVec);
 
     fflush(stdout);
+    
+
+    /* 
     cVec = cMat * cVec;  // LatticeColourVector     = LatticeColourMatrix     * LatticeColourVector
     sVec = sMat * sVec;  // LatticeSpinVector       = LatticeSpinMatrix       * LatticeSpinVector
     scVec= scMat * scVec;// LatticeSpinColourVector = LatticeSpinColourMatrix * LatticeSpinColourVector
@@ -112,12 +115,14 @@ int main (int argc, char ** argv)
     cMat = outerProduct(cVec,cVec);
     scalar = localInnerProduct(cVec,cVec);
 
+ 
     scalar += scalar;
     scalar -= scalar;
     scalar *= scalar;
     add(scalar,scalar,scalar);
     sub(scalar,scalar,scalar);
     mult(scalar,scalar,scalar);
+
     mac(scalar,scalar,scalar);
     scalar = scalar+scalar;
     scalar = scalar-scalar;
@@ -141,7 +146,7 @@ int main (int argc, char ** argv)
     scalar=trace(scalar);
     scalar=localInnerProduct(cVec,cVec);
     scalar=localNorm2(cVec);
-
+    */
 //     -=,+=,*=,()
 //     add,+,sub,-,mult,mac,*
 //     adj,conjugate
@@ -153,10 +158,11 @@ int main (int argc, char ** argv)
 //     localNorm2
 //     localInnerProduct
     
+  
     scMat = sMat*scMat;  // LatticeSpinColourMatrix = LatticeSpinMatrix       * LatticeSpinColourMatrix
 
-
-
+    
+    /*
 #ifdef SSE4
     ///////// Tests the new class Grid_simd 
     std::complex<double> ctest(3.0,2.0);
@@ -196,8 +202,10 @@ int main (int argc, char ** argv)
     std::cout << sum<< std::endl;
 
 #endif
+    */
     ///////////////////////
-
+    /*
+    printf("DEBUG: calling 3.5 \n");
     // Non-lattice (const objects) * Lattice
     ColourMatrix cm;
     SpinColourMatrix scm;
@@ -217,6 +225,7 @@ int main (int argc, char ** argv)
     vscm = vscm*cplx;
     scMat = scMat*cplx;
 
+    printf("DEBUG: calling 3.7 \n");
     scm = cplx*scm;
     vscm = cplx*vscm;
     scMat = cplx*scMat;
@@ -224,12 +233,14 @@ int main (int argc, char ** argv)
     vscm = myint*vscm;
     scMat = scMat*myint;
     
+    printf("DEBUG: calling 3.9 \n");
     scm = scm*mydouble;
     vscm = vscm*mydouble;
     scMat = scMat*mydouble;
     scMat = mydouble*scMat;
     cMat = mydouble*cMat;
-    
+  
+    printf("DEBUG: calling 4 \n");
     sMat = adj(sMat);       // LatticeSpinMatrix adjoint
     sMat = iGammaFive*sMat; // SpinMatrix * LatticeSpinMatrix
     sMat = GammaFive*sMat;  // SpinMatrix * LatticeSpinMatrix
@@ -240,6 +251,9 @@ int main (int argc, char ** argv)
     scm=transpose(scm);
     scm=transposeIndex<1>(scm);
     
+
+
+
 //    Foo = Foo+scalar; // LatticeColourMatrix+Scalar
 //    Foo = Foo*scalar; // LatticeColourMatrix*Scalar
 //    Foo = Foo-scalar; // LatticeColourMatrix-Scalar
@@ -279,7 +293,8 @@ int main (int argc, char ** argv)
       pokeIndex<1> (c_m,c,0,0);
     }
 
-    
+    */
+
     FooBar = Bar;
  
     /*
@@ -332,14 +347,14 @@ int main (int argc, char ** argv)
     // Lattice SU(3) x SU(3)
     Fine.Barrier();
     FooBar = Foo * Bar;
-    
+
     // Lattice 12x12 GEMM
     scFooBar = scFoo * scBar;
-    
+
     // Benchmark some simple operations LatticeSU3 * Lattice SU3.
     double t0,t1,flops;
     double bytes;
-    int ncall=100;
+    int ncall=5000;
     int Nc = Grid::QCD::Nc;
 
     LatticeGaugeField U(&Fine);
@@ -351,19 +366,21 @@ int main (int argc, char ** argv)
     if ( Fine.IsBoss() ) {
       printf("%f flop and %f bytes\n",flops,bytes/ncall);
     }
-        FooBar = Foo * Bar;
+    FooBar = Foo * Bar;
     Fine.Barrier();
     t0=usecond();
     for(int i=0;i<ncall;i++){
       Fine.Barrier();
       mult(FooBar,Foo,Bar); // this is better
     }
+
     t1=usecond();
     Fine.Barrier();
     if ( Fine.IsBoss() ) {
 #ifdef OMP
       printf("mult NumThread %d , Lattice size %d , %f us per call\n",omp_get_max_threads(),lat,(t1-t0)/ncall);
 #endif
+      printf("mult NumThread %d , Lattice size %d , %f us per call\n",omp,lat,(t1-t0)/ncall);
       printf("mult NumThread %d , Lattice size %d , %f Mflop/s\n",omp,lat,flops/(t1-t0));
       printf("mult NumThread %d , Lattice size %d , %f MB/s\n",omp,lat,bytes/(t1-t0));
     }
@@ -375,6 +392,7 @@ int main (int argc, char ** argv)
     t0=usecond();
     for(int i=0;i<ncall;i++){
       Fine.Barrier();
+      //Cshift(Bar,1,-1);
       mult(FooBar,Foo,Cshift(Bar,1,-1));
       //mult(FooBar,Foo,Bar);
       //FooBar = Foo * Bar; // this is bad
@@ -525,5 +543,9 @@ int main (int argc, char ** argv)
 
    } // loop for lat
  } // loop for omp
+
+
+ std::cout << sizeof(vComplexF) << std::endl;
+ 
  Grid_finalize();
 }
diff --git a/tests/Makefile.am b/tests/Makefile.am
index 95642ed8..3c79191c 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -5,7 +5,7 @@ AM_LDFLAGS = -L$(top_builddir)/lib
 #
 # Test code
 #
-bin_PROGRAMS = Grid_main Grid_stencil Grid_nersc_io Grid_cshift Grid_gamma  Grid_simd Grid_rng Grid_remez Grid_rng_fixed Grid_simd_new
+bin_PROGRAMS = Grid_main Grid_stencil Grid_nersc_io Grid_cshift Grid_gamma  Grid_simd Grid_rng Grid_remez Grid_rng_fixed 
 
 Grid_main_SOURCES = Grid_main.cc
 Grid_main_LDADD = -lGrid
@@ -34,5 +34,5 @@ Grid_stencil_LDADD = -lGrid
 Grid_simd_SOURCES = Grid_simd.cc
 Grid_simd_LDADD = -lGrid
 
-Grid_simd_new_SOURCES = Grid_simd_new.cc
-Grid_simd_new_LDADD = -lGrid
+#Grid_simd_new_SOURCES = Grid_simd_new.cc
+#Grid_simd_new_LDADD = -lGrid

From ae58a9ada21a6fb8e05bd254bd2c99861ae95ef9 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 23 May 2015 09:30:28 +0100
Subject: [PATCH 216/429] Iterator required

---
 lib/Grid_init.cc | 1 +
 1 file changed, 1 insertion(+)

diff --git a/lib/Grid_init.cc b/lib/Grid_init.cc
index 6af0be84..572672eb 100644
--- a/lib/Grid_init.cc
+++ b/lib/Grid_init.cc
@@ -11,6 +11,7 @@
 #include <sys/time.h>
 #include <signal.h>
 #include <iostream>
+#include <iterator>
 #include <Grid.h>
 #include <algorithm>
 

From 764732944f1501368d07f92fafd5f65901764a47 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 23 May 2015 09:31:15 +0100
Subject: [PATCH 217/429] Cosmetic

---
 lib/math/Grid_math_arith_mul.h | 9 ++++-----
 1 file changed, 4 insertions(+), 5 deletions(-)

diff --git a/lib/math/Grid_math_arith_mul.h b/lib/math/Grid_math_arith_mul.h
index 04b05bb4..fcdd60ee 100644
--- a/lib/math/Grid_math_arith_mul.h
+++ b/lib/math/Grid_math_arith_mul.h
@@ -16,19 +16,18 @@ strong_inline void mult(iScalar<rtype> * __restrict__ ret,const iScalar<mtype> *
 template<class rrtype,class ltype,class rtype,int N>
 strong_inline void mult(iMatrix<rrtype,N> * __restrict__ ret,const iMatrix<ltype,N> * __restrict__ lhs,const iMatrix<rtype,N> * __restrict__ rhs){
   for(int c1=0;c1<N;c1++){
-    int c3=0;
     for(int c2=0;c2<N;c2++){
-      mult(&ret->_internal[c1][c2],&lhs->_internal[c1][c3],&rhs->_internal[c3][c2]);
+      mult(&ret->_internal[c1][c2],&lhs->_internal[c1][0],&rhs->_internal[0][c2]);
     }
   }
-  for(int c3=1;c3<N;c3++){
-    for(int c1=0;c1<N;c1++){
+  for(int c1=0;c1<N;c1++){
+    for(int c3=1;c3<N;c3++){
       for(int c2=0;c2<N;c2++){
 	mac(&ret->_internal[c1][c2],&lhs->_internal[c1][c3],&rhs->_internal[c3][c2]);
       }
     }
   }
-    return;
+  return;
 }
 
 template<class rrtype,class ltype,class rtype,int N>

From a2928321b6c9fa23686208fb18225916980698bc Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 23 May 2015 09:32:37 +0100
Subject: [PATCH 218/429] Better pragma use

---
 lib/lattice/Grid_lattice_arith.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/lib/lattice/Grid_lattice_arith.h b/lib/lattice/Grid_lattice_arith.h
index ff966578..a6f59b56 100644
--- a/lib/lattice/Grid_lattice_arith.h
+++ b/lib/lattice/Grid_lattice_arith.h
@@ -185,7 +185,7 @@ PARALLEL_FOR_LOOP
   template<class sobj,class vobj> strong_inline
   void axpy(Lattice<vobj> &ret,sobj a,const Lattice<vobj> &x,const Lattice<vobj> &y){
     conformable(x,y);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<x._grid->oSites();ss++){
 #ifdef STREAMING_STORES
       vobj tmp = a*x._odata[ss]+y._odata[ss];
@@ -198,7 +198,7 @@ PARALLEL_FOR_LOOP
   template<class sobj,class vobj> strong_inline
   void axpby(Lattice<vobj> &ret,sobj a,sobj b,const Lattice<vobj> &x,const Lattice<vobj> &y){
     conformable(x,y);
-#pragma omp parallel for
+PARALLEL_FOR_LOOP
     for(int ss=0;ss<x._grid->oSites();ss++){
 #ifdef STREAMING_STORES
       vobj tmp = a*x._odata[ss]+b*y._odata[ss];

From d07a5c084d803450f2379721854b4e61e1faf79d Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 23 May 2015 09:33:42 +0100
Subject: [PATCH 219/429] Rely on default constructors

---
 lib/math/Grid_math_tensors.h | 120 +++++++++++++++++++----------------
 1 file changed, 66 insertions(+), 54 deletions(-)

diff --git a/lib/math/Grid_math_tensors.h b/lib/math/Grid_math_tensors.h
index a0424576..32ca78e5 100644
--- a/lib/math/Grid_math_tensors.h
+++ b/lib/math/Grid_math_tensors.h
@@ -34,62 +34,64 @@ public:
 
   // Scalar no action
   //  template<int Level> using tensor_reduce_level = typename iScalar<GridTypeMapper<vtype>::tensor_reduce_level<Level> >;
-  iScalar()=default;
+  iScalar() = default;
+  /*
+  iScalar(const iScalar<vtype> &copyme)=default;
+  iScalar(iScalar<vtype> &&copyme)=default;
+  iScalar<vtype> & operator= (const iScalar<vtype> &copyme) = default;
+  iScalar<vtype> & operator= (iScalar<vtype> &&copyme) = default;
+  */
   iScalar(scalar_type s) : _internal(s) {};// recurse down and hit the constructor for vector_type
   iScalar(const Zero &z){ *this = zero; };
 
-    iScalar<vtype> & operator= (const Zero &hero){
-      zeroit(*this);
-      return *this;
-    }
-    friend strong_inline void vstream(iScalar<vtype> &out,const iScalar<vtype> &in){
-      vstream(out._internal,in._internal);
-    }
+  iScalar<vtype> & operator= (const Zero &hero){
+    zeroit(*this);
+    return *this;
+  }
+  friend strong_inline void vstream(iScalar<vtype> &out,const iScalar<vtype> &in){
+    vstream(out._internal,in._internal);
+  }
+  friend strong_inline void zeroit(iScalar<vtype> &that){
+    zeroit(that._internal);
+  }
+  friend strong_inline void prefetch(iScalar<vtype> &that){
+    prefetch(that._internal);
+  }
+  friend strong_inline void permute(iScalar<vtype> &out,const iScalar<vtype> &in,int permutetype){
+    permute(out._internal,in._internal,permutetype);
+  }
 
-
-    friend strong_inline void zeroit(iScalar<vtype> &that){
-        zeroit(that._internal);
-    }
-    friend strong_inline void prefetch(iScalar<vtype> &that){
-      prefetch(that._internal);
-    }
-    friend strong_inline void permute(iScalar<vtype> &out,const iScalar<vtype> &in,int permutetype){
-      permute(out._internal,in._internal,permutetype);
-    }
-
-    // Unary negation
-    friend strong_inline iScalar<vtype> operator -(const iScalar<vtype> &r) {
-        iScalar<vtype> ret;
-        ret._internal= -r._internal;
-        return ret;
-    }
-    // *=,+=,-= operators inherit from corresponding "*,-,+" behaviour
-    strong_inline iScalar<vtype> &operator *=(const iScalar<vtype> &r) {
-        *this = (*this)*r;
-        return *this;
-    }
-    strong_inline iScalar<vtype> &operator -=(const iScalar<vtype> &r) {
-        *this = (*this)-r;
-        return *this;
-    }
-    strong_inline iScalar<vtype> &operator +=(const iScalar<vtype> &r) {
-        *this = (*this)+r;
-        return *this;
-    }
-    
-    strong_inline vtype & operator ()(void) {
-      return _internal;
-    }
-
-    strong_inline const vtype & operator ()(void) const {
-      return _internal;
-    }
-
-    operator ComplexD () const { return(TensorRemove(_internal)); };
-    operator RealD () const { return(real(TensorRemove(_internal))); }
-
-    // convert from a something to a scalar
-    template<class T,typename std::enable_if<!isGridTensor<T>::value, T>::type* = nullptr > strong_inline auto operator = (T arg) -> iScalar<vtype>
+  // Unary negation
+  friend strong_inline iScalar<vtype> operator -(const iScalar<vtype> &r) {
+    iScalar<vtype> ret;
+    ret._internal= -r._internal;
+    return ret;
+  }
+  // *=,+=,-= operators inherit from corresponding "*,-,+" behaviour
+  strong_inline iScalar<vtype> &operator *=(const iScalar<vtype> &r) {
+    *this = (*this)*r;
+    return *this;
+  }
+  strong_inline iScalar<vtype> &operator -=(const iScalar<vtype> &r) {
+    *this = (*this)-r;
+    return *this;
+  }
+  strong_inline iScalar<vtype> &operator +=(const iScalar<vtype> &r) {
+    *this = (*this)+r;
+    return *this;
+  }
+  strong_inline vtype & operator ()(void) {
+    return _internal;
+  }
+  strong_inline const vtype & operator ()(void) const {
+    return _internal;
+  }
+  
+  operator ComplexD () const { return(TensorRemove(_internal)); };
+  operator RealD () const { return(real(TensorRemove(_internal))); }
+  
+  // convert from a something to a scalar
+  template<class T,typename std::enable_if<!isGridTensor<T>::value, T>::type* = nullptr > strong_inline auto operator = (T arg) -> iScalar<vtype>
     { 
       _internal = vtype(arg);
       return *this;
@@ -125,6 +127,12 @@ public:
   enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
   iVector(const Zero &z){ *this = zero; };
   iVector() =default;
+  /*
+  iVector(const iVector<vtype,N> &copyme)=default;
+  iVector(iVector<vtype,N> &&copyme)=default;
+  iVector<vtype,N> & operator= (const iVector<vtype,N> &copyme) = default;
+  iVector<vtype,N> & operator= (iVector<vtype,N> &&copyme) = default;
+  */
 
     iVector<vtype,N> & operator= (const Zero &hero){
         zeroit(*this);
@@ -205,8 +213,12 @@ public:
 
   iMatrix(const Zero &z){ *this = zero; };
   iMatrix() =default;
-
-
+  /*
+  iMatrix(const iMatrix<vtype,N> &copyme)=default;
+  iMatrix(iMatrix<vtype,N> &&copyme)=default;
+  iMatrix<vtype,N> & operator= (const iMatrix<vtype,N> &copyme) = default;
+  iMatrix<vtype,N> & operator= (iMatrix<vtype,N> &&copyme) = default;
+  */
   iMatrix<vtype,N> & operator= (const Zero &hero){
     zeroit(*this);
     return *this;

From 65f2e6b2696394992656f8a6a01c4ffd80507488 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 23 May 2015 09:34:16 +0100
Subject: [PATCH 220/429] Improving even odd sector; lot of work and through
 required cleaning this

---
 lib/qcd/Grid_qcd_wilson_dop.cc | 27 ++++++++++++---------------
 1 file changed, 12 insertions(+), 15 deletions(-)

diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index 9ef4af0c..2020a9ec 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -97,45 +97,44 @@ void WilsonMatrix::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeF
 
 RealD WilsonMatrix::M(const LatticeFermion &in, LatticeFermion &out)
 {
-  Dhop(in,out,0);
+  this->Dhop(in,out,0);
   out = (4+mass)*in - 0.5*out  ; // FIXME : axpby_norm! fusion fun
   return norm2(out);
 }
 RealD WilsonMatrix::Mdag(const LatticeFermion &in, LatticeFermion &out)
 {
-  Dhop(in,out,1);
+  this->Dhop(in,out,1);
   out = (4+mass)*in - 0.5*out  ; // FIXME : axpby_norm! fusion fun
   return norm2(out);
 }
 
 void WilsonMatrix::Meooe(const LatticeFermion &in, LatticeFermion &out)
 {
-  Dhop(in,out,0);
+  this->Dhop(in,out,0);
   out = 0.5*out; // FIXME : scale factor in Dhop
 }
 void WilsonMatrix::MeooeDag(const LatticeFermion &in, LatticeFermion &out)
 {
-  Dhop(in,out,1);
+  this->Dhop(in,out,1);
+  out = 0.5*out; // FIXME : scale factor in Dhop
 }
 void WilsonMatrix::Mooee(const LatticeFermion &in, LatticeFermion &out)
 {
   out = (4.0+mass)*in;
   return ;
 }
-void WilsonMatrix::MooeeInv(const LatticeFermion &in, LatticeFermion &out)
-{
-  out = (1.0/(4.0+mass))*in;
-  return ;
-}
 void WilsonMatrix::MooeeDag(const LatticeFermion &in, LatticeFermion &out)
+{
+  this->Mooee(in,out);
+}
+void WilsonMatrix::MooeeInv(const LatticeFermion &in, LatticeFermion &out)
 {
   out = (1.0/(4.0+mass))*in;
   return ;
 }
 void WilsonMatrix::MooeeInvDag(const LatticeFermion &in, LatticeFermion &out)
 {
-  out = (1.0/(4.0+mass))*in;
-  return ;
+  this->MooeeInv(in,out);
 }
 
 void WilsonMatrix::DhopSite(int ss,const LatticeFermion &in, LatticeFermion &out)
@@ -451,8 +450,6 @@ PARALLEL_FOR_LOOP
 }
 
 
-
-
-
-}}
+}
+}
 

From b8fdb65fbf89273f8cfd7e7b8691bf16a8ea89b1 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 23 May 2015 09:34:50 +0100
Subject: [PATCH 221/429] More targets

---
 scripts/configure-all | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/scripts/configure-all b/scripts/configure-all
index f42b1960..ad91034d 100755
--- a/scripts/configure-all
+++ b/scripts/configure-all
@@ -1,6 +1,6 @@
 #!/bin/bash
 
-DIRS="clang-avx clang-avx-openmp clang-avx-openmp-mpi clang-avx-mpi clang-avx2 clang-avx2-openmp clang-avx2-openmp-mpi clang-avx2-mpi icpc-avx icpc-avx2 icpc-avx512 g++-sse4 g++-avx clang-sse"
+DIRS="clang-avx clang-avx-openmp clang-avx-openmp-mpi clang-avx-mpi clang-avx2 clang-avx2-openmp clang-avx2-openmp-mpi clang-avx2-mpi icpc-avx icpc-avx2 icpc-avx512 g++-sse4 g++-avx clang-sse icpc-avx-openmp-mpi icpc-avx-openmp"
 
 for D in $DIRS
 do

From 73ee36c48d14d02c5be8937b84eb6eda1778b4e6 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 23 May 2015 09:35:37 +0100
Subject: [PATCH 222/429] Extra targets

---
 scripts/configure-commands | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/scripts/configure-commands b/scripts/configure-commands
index c7c35e1c..538f3661 100755
--- a/scripts/configure-commands
+++ b/scripts/configure-commands
@@ -25,6 +25,12 @@ g++-avx)
 icpc-avx)
   CXX=icpc ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11" LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;
+icpc-avx-openmp-mpi)
+CXX=icpc ../../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -fopenmp -lgmp -lmpfr" --enable-comms=mpi
+;;
+icpc-avx-openmp)
+CXX=icpc ../../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -std=c++11" LIBS="-fopenmp -lgmp -lmpfr" --enable-comms=mpi
+;;
 icpc-avx2)
   CXX=icpc ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -std=c++11" LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;

From 280627334076bfafe30a77b30a5d1eab824f98cd Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 23 May 2015 09:36:01 +0100
Subject: [PATCH 223/429] Added

---
 benchmarks/Grid_su3_expr.cc | 11 +++++++++++
 1 file changed, 11 insertions(+)
 create mode 100644 benchmarks/Grid_su3_expr.cc

diff --git a/benchmarks/Grid_su3_expr.cc b/benchmarks/Grid_su3_expr.cc
new file mode 100644
index 00000000..e20bcef4
--- /dev/null
+++ b/benchmarks/Grid_su3_expr.cc
@@ -0,0 +1,11 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+void su3_test_mult_expression(LatticeColourMatrix &z, LatticeColourMatrix &x,LatticeColourMatrix &y)
+{
+  z=x*y;
+}
+

From 17a06af1ff2c79edd662cc436ddf1ba061fe2f31 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 25 May 2015 13:42:12 +0100
Subject: [PATCH 224/429] red black fix

---
 lib/Grid_stencil.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/lib/Grid_stencil.h b/lib/Grid_stencil.h
index ce987746..fa22361b 100644
--- a/lib/Grid_stencil.h
+++ b/lib/Grid_stencil.h
@@ -120,8 +120,8 @@ namespace Grid {
 	  // Map to always positive shift modulo global full dimension.
 	  int shift = (displacement+fd)%fd;
 	  
-     	  int checkerboard = _grid->CheckerBoardDestination(source.checkerboard,shift);
-	  assert (checkerboard== _checkerboard);
+	  //     	  int checkerboard = _grid->CheckerBoardDestination(source.checkerboard,shift);
+	  assert (source.checkerboard== _checkerboard);
 
 	  // the permute type
 	  int simd_layout     = _grid->_simd_layout[dimension];

From 9b5633ff4f6b13bfd07cebffadbe7bd70177efea Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 25 May 2015 13:42:36 +0100
Subject: [PATCH 225/429] Herm op

---
 lib/algorithms/LinearOperator.h | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/lib/algorithms/LinearOperator.h b/lib/algorithms/LinearOperator.h
index 167213ec..cb6240ba 100644
--- a/lib/algorithms/LinearOperator.h
+++ b/lib/algorithms/LinearOperator.h
@@ -106,6 +106,10 @@ namespace Grid {
     public:
       virtual void operator() (LinearOperatorBase<Field> &Linop, const Field &in, Field &out) = 0;
     };
+    template<class Field> class HermitianOperatorFunction {
+    public:
+      virtual void operator() (HermitianOperatorBase<Field> &Linop, const Field &in, Field &out) = 0;
+    };
 
     // FIXME : To think about
 

From 29f72292ba44ec5408b73dc94da2cedc5a03b3c0 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 25 May 2015 13:43:58 +0100
Subject: [PATCH 226/429] Updates now schur red black solver working

---
 lib/algorithms/SparseMatrix.h                | 18 ++---
 lib/algorithms/iterative/ConjugateGradient.h | 49 ++++++++-----
 lib/algorithms/iterative/SchurRedBlack.h     | 73 +++++++++++---------
 3 files changed, 82 insertions(+), 58 deletions(-)

diff --git a/lib/algorithms/SparseMatrix.h b/lib/algorithms/SparseMatrix.h
index 6c54fc92..7bfe959b 100644
--- a/lib/algorithms/SparseMatrix.h
+++ b/lib/algorithms/SparseMatrix.h
@@ -10,6 +10,7 @@ namespace Grid {
   /////////////////////////////////////////////////////////////////////////////////////////////
     template<class Field> class SparseMatrixBase {
     public:
+      GridBase *_grid;
       // Full checkerboar operations
       virtual RealD M    (const Field &in, Field &out)=0;
       virtual RealD Mdag (const Field &in, Field &out)=0;
@@ -18,6 +19,7 @@ namespace Grid {
 	ni=M(in,tmp);
 	no=Mdag(tmp,out);
       }
+      SparseMatrixBase(GridBase *grid) : _grid(grid) {};
     };
 
   /////////////////////////////////////////////////////////////////////////////////////////////
@@ -25,7 +27,7 @@ namespace Grid {
   /////////////////////////////////////////////////////////////////////////////////////////////
     template<class Field> class CheckerBoardedSparseMatrixBase : public SparseMatrixBase<Field> {
     public:
-      
+      GridBase *_cbgrid;
       // half checkerboard operaions
       virtual  void Meooe    (const Field &in, Field &out)=0;
       virtual  void Mooee    (const Field &in, Field &out)=0;
@@ -44,9 +46,7 @@ namespace Grid {
 	Meooe(out,tmp);
 
 	Mooee(in,out);
-	out=out-tmp; // axpy_norm
-	RealD n=norm2(out);
-	return n;
+	return axpy_norm(out,-1.0,tmp,out);
       }
       virtual  RealD MpcDag   (const Field &in, Field &out){
 	Field tmp(in._grid);
@@ -56,15 +56,15 @@ namespace Grid {
 	MeooeDag(out,tmp);
 
 	MooeeDag(in,out);
-	out=out-tmp; // axpy_norm
-	RealD n=norm2(out);
-	return n;
+	return axpy_norm(out,-1.0,tmp,out);
       }
-      virtual void MpcDagMpc(const Field &in, Field &out,RealD ni,RealD no) {
+      virtual void MpcDagMpc(const Field &in, Field &out,RealD &ni,RealD &no) {
 	Field tmp(in._grid);
 	ni=Mpc(in,tmp);
-	no=Mpc(tmp,out);
+	no=MpcDag(tmp,out);
+	//	std::cout<<"MpcDagMpc "<<ni<<" "<<no<<std::endl;
       }
+      CheckerBoardedSparseMatrixBase(GridBase *grid,GridBase *cbgrid) : SparseMatrixBase<Field>(grid), _cbgrid(cbgrid) {};
     };
 
 }
diff --git a/lib/algorithms/iterative/ConjugateGradient.h b/lib/algorithms/iterative/ConjugateGradient.h
index 85b196e9..d4f89662 100644
--- a/lib/algorithms/iterative/ConjugateGradient.h
+++ b/lib/algorithms/iterative/ConjugateGradient.h
@@ -9,17 +9,21 @@ namespace Grid {
     /////////////////////////////////////////////////////////////
 
   template<class Field> 
-    class ConjugateGradient :  public OperatorFunction<Field> {
+    class ConjugateGradient : public HermitianOperatorFunction<Field> {
 public:                                                
     RealD   Tolerance;
     Integer MaxIterations;
-
+    int verbose;
     ConjugateGradient(RealD tol,Integer maxit) : Tolerance(tol), MaxIterations(maxit) { 
+      verbose=0;
     };
 
-    void operator() (LinearOperatorBase<Field> &Linop,const Field &src, Field &psi) {assert(0);};
+
     void operator() (HermitianOperatorBase<Field> &Linop,const Field &src, Field &psi){
 
+      psi.checkerboard = src.checkerboard;
+      conformable(psi,src);
+
       RealD cp,c,a,d,b,ssq,qq,b_pred;
       
       Field   p(src);
@@ -37,14 +41,16 @@ public:
       a  =norm2(p);
       cp =a;
       ssq=norm2(src);
-      
-      std::cout <<std::setprecision(4)<< "ConjugateGradient: guess "<<guess<<std::endl;
-      std::cout <<std::setprecision(4)<< "ConjugateGradient:   src "<<ssq  <<std::endl;
-      std::cout <<std::setprecision(4)<< "ConjugateGradient:    mp "<<d    <<std::endl;
-      std::cout <<std::setprecision(4)<< "ConjugateGradient:   mmp "<<b    <<std::endl;
-      std::cout <<std::setprecision(4)<< "ConjugateGradient:     r "<<cp   <<std::endl;
-      std::cout <<std::setprecision(4)<< "ConjugateGradient:     p "<<a    <<std::endl;
-      
+
+      if ( verbose ) {
+	std::cout <<std::setprecision(4)<< "ConjugateGradient: guess "<<guess<<std::endl;
+	std::cout <<std::setprecision(4)<< "ConjugateGradient:   src "<<ssq  <<std::endl;
+	std::cout <<std::setprecision(4)<< "ConjugateGradient:    mp "<<d    <<std::endl;
+	std::cout <<std::setprecision(4)<< "ConjugateGradient:   mmp "<<b    <<std::endl;
+	std::cout <<std::setprecision(4)<< "ConjugateGradient:  cp,r "<<cp   <<std::endl;
+	std::cout <<std::setprecision(4)<< "ConjugateGradient:     p "<<a    <<std::endl;
+      }
+
       RealD rsq =  Tolerance* Tolerance*ssq;
       
       //Check if guess is really REALLY good :)
@@ -60,14 +66,16 @@ public:
 	c=cp;
 	
 	Linop.OpAndNorm(p,mmp,d,qq);
-	  
-	//	std::cout <<std::setprecision(4)<< "ConjugateGradient:  d,qq "<<d<< " "<<qq <<std::endl;
+
+	RealD    qqck = norm2(mmp);
+	ComplexD dck  = innerProduct(p,mmp);
+	//	if (verbose) std::cout <<std::setprecision(4)<< "ConjugateGradient:  d,qq "<<d<< " "<<qq <<" qqcheck "<< qqck<< " dck "<< dck<<std::endl;
       
 	a      = c/d;
 	b_pred = a*(a*qq-d)/c;
-	
-	//	std::cout <<std::setprecision(4)<< "ConjugateGradient:  a,bp "<<a<< " "<<b_pred <<std::endl;
 
+
+	//	if (verbose) std::cout <<std::setprecision(4)<< "ConjugateGradient:  a,bp "<<a<< " "<<b_pred <<std::endl;
 	cp = axpy_norm(r,-a,mmp,r);
 	b = cp/c;
 	//	std::cout <<std::setprecision(4)<< "ConjugateGradient:  cp,b "<<cp<< " "<<b <<std::endl;
@@ -76,7 +84,16 @@ public:
 	psi= a*p+psi;
 	p  = p*b+r;
 	  
-	std::cout<<"ConjugateGradient: Iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
+	if (verbose) std::cout<<"ConjugateGradient: Iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
+
+	// Hack
+	if (0) {
+	  Field   tt(src);
+      	  Linop.Op(psi,mmp);
+	  tt=mmp-src;
+	  RealD resnorm = norm2(tt);
+	  std::cout<<"ConjugateGradient: Iteration " <<k<<" true residual "<<resnorm << " computed " << cp <<std::endl;
+	}
 
 	// Stopping condition
 	if ( cp <= rsq ) { 
diff --git a/lib/algorithms/iterative/SchurRedBlack.h b/lib/algorithms/iterative/SchurRedBlack.h
index e6a7b59e..d3a76d7f 100644
--- a/lib/algorithms/iterative/SchurRedBlack.h
+++ b/lib/algorithms/iterative/SchurRedBlack.h
@@ -1,6 +1,7 @@
 #ifndef GRID_SCHUR_RED_BLACK_H
 #define GRID_SCHUR_RED_BLACK_H
 
+
   /*
    * Red black Schur decomposition
    *
@@ -25,80 +26,86 @@
    *     M psi = eta
    ***********************
    *Odd
-   * i)   (D_oo)^{\dag} D_oo psi_o = (D_oo)^\dag L^{-1}  eta_o
-   *                        eta_o' = D_oo (eta_o - Moe Mee^{-1} eta_e)
+   * i)   (D_oo)^{\dag} D_oo psi_o = (D_oo)^dag L^{-1}  eta_o
+   *                        eta_o' = (D_oo)^dag (eta_o - Moe Mee^{-1} eta_e)
    *Even
    * ii)  Mee psi_e + Meo psi_o = src_e
    *
    *   => sol_e = M_ee^-1 * ( src_e - Meo sol_o )...
    *
    */
-
 namespace Grid {
 
   ///////////////////////////////////////////////////////////////////////////////////////////////////////
   // Take a matrix and form a Red Black solver calling a Herm solver
   // Use of RB info prevents making SchurRedBlackSolve conform to standard interface
   ///////////////////////////////////////////////////////////////////////////////////////////////////////
-  template<class Field> class SchurRedBlackSolve : public OperatorFunction<Field>{
+  template<class Field> class SchurRedBlackSolve {
   private:
-    SparseMatrixBase<Field> & _Matrix;
-    OperatorFunction<Field> & _HermitianRBSolver;
+    HermitianOperatorFunction<Field> & _HermitianRBSolver;
     int CBfactorise;
   public:
 
     /////////////////////////////////////////////////////
     // Wrap the usual normal equations Schur trick
     /////////////////////////////////////////////////////
-  SchurRedBlackSolve(SparseMatrixBase<Field> &Matrix, OperatorFunction<Field> &HermitianRBSolver) 
-    :  _Matrix(Matrix), _HermitianRBSolver(HermitianRBSolver) { 
+  SchurRedBlackSolve(HermitianOperatorFunction<Field> &HermitianRBSolver)  :
+     _HermitianRBSolver(HermitianRBSolver) 
+    { 
       CBfactorise=0;
     };
 
-    void operator() (const Field &in, Field &out){
+    template<class Matrix>
+      void operator() (Matrix & _Matrix,const Field &in, Field &out){
 
       // FIXME CGdiagonalMee not implemented virtual function
-      // FIXME need to make eo grid from full grid.
       // FIXME use CBfactorise to control schur decomp
-      const int Even=0;
-      const int Odd =1;
-
-      // Make a cartesianRedBlack from full Grid
-      GridRedBlackCartesian grid(in._grid);
+      GridBase *grid = _Matrix._cbgrid;
+      GridBase *fgrid= _Matrix._grid;
  
-      Field src_e(&grid);
-      Field src_o(&grid);
-      Field sol_e(&grid);
-      Field sol_o(&grid);
-      Field   tmp(&grid);
-      Field  Mtmp(&grid);
-      
+      Field src_e(grid);
+      Field src_o(grid);
+      Field sol_e(grid);
+      Field sol_o(grid);
+      Field   tmp(grid);
+      Field  Mtmp(grid);
+      Field resid(fgrid);
+
       pickCheckerboard(Even,src_e,in);
       pickCheckerboard(Odd ,src_o,in);
 
       /////////////////////////////////////////////////////
       // src_o = Mdag * (source_o - Moe MeeInv source_e)
       /////////////////////////////////////////////////////
-      _Matrix.MooeeInv(src_e,tmp);     //    MooeeInv(source[Even],tmp,DaggerNo,Even);
-      _Matrix.Meooe   (tmp,Mtmp);      //    Meo     (tmp,src,Odd,DaggerNo);
-      tmp=src_o-Mtmp;                  //    axpy    (tmp,src,source[Odd],-1.0);
-      _Matrix.MpcDag(tmp,src_o);       //    Mprec(tmp,src,Mtmp,DaggerYes);  
+      _Matrix.MooeeInv(src_e,tmp);     assert(  tmp.checkerboard ==Even);
+      _Matrix.Meooe   (tmp,Mtmp);      assert( Mtmp.checkerboard ==Odd);     
+      tmp=src_o-Mtmp;                  assert(  tmp.checkerboard ==Odd);     
+      _Matrix.MpcDag(tmp,src_o);       assert(src_o.checkerboard ==Odd);       
 
       //////////////////////////////////////////////////////////////
       // Call the red-black solver
       //////////////////////////////////////////////////////////////
-      _HermitianRBSolver(src_o,sol_o); //  CGNE_prec_MdagM(solution[Odd],src);
+      HermitianCheckerBoardedOperator<Matrix,Field> _HermOpEO(_Matrix);
+      std::cout << "SchurRedBlack solver calling the MpcDagMp solver" <<std::endl;
+      _HermitianRBSolver(_HermOpEO,src_o,sol_o);  assert(sol_o.checkerboard==Odd);
 
       ///////////////////////////////////////////////////
       // sol_e = M_ee^-1 * ( src_e - Meo sol_o )...
       ///////////////////////////////////////////////////
-      _Matrix.Meooe(sol_o,tmp);        // Meo(solution[Odd],tmp,Even,DaggerNo);
-      src_e = src_e-tmp;               // axpy(src,tmp,source[Even],-1.0);
-      _Matrix.MooeeInv(src_e,sol_e);   // MooeeInv(src,solution[Even],DaggerNo,Even);
+      _Matrix.Meooe(sol_o,tmp);        assert(  tmp.checkerboard   ==Even);
+      src_e = src_e-tmp;               assert(  src_e.checkerboard ==Even);
+      _Matrix.MooeeInv(src_e,sol_e);   assert(  sol_e.checkerboard ==Even);
      
-      setCheckerboard(out,sol_e);
-      setCheckerboard(out,sol_o);
- 
+      setCheckerboard(out,sol_e); assert(  sol_e.checkerboard ==Even);
+      setCheckerboard(out,sol_o); assert(  sol_o.checkerboard ==Odd );
+
+      // Verify the unprec residual
+      _Matrix.M(out,resid); 
+      resid = resid-in;
+      RealD ns = norm2(in);
+      RealD nr = norm2(resid);
+
+      std::cout << "SchurRedBlack solver true unprec resid "<< sqrt(nr/ns) <<" nr "<< nr <<" ns "<<ns << std::endl;
     }     
   };
 

From bc947477f328aa6f30566f29733687f449115ce8 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 25 May 2015 13:44:35 +0100
Subject: [PATCH 227/429] move constants into red black

---
 lib/cartesian/Grid_cartesian_red_black.h | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/lib/cartesian/Grid_cartesian_red_black.h b/lib/cartesian/Grid_cartesian_red_black.h
index 475b47c2..ff2d3ba8 100644
--- a/lib/cartesian/Grid_cartesian_red_black.h
+++ b/lib/cartesian/Grid_cartesian_red_black.h
@@ -4,6 +4,13 @@
 
 namespace Grid {
 
+    static const int CbRed  =0;
+    static const int CbBlack=1;
+    static const int Even   =CbRed;
+    static const int Odd    =CbBlack;
+    static const int DaggerNo=0;
+    static const int DaggerYes=1;
+
 // Specialise this for red black grids storing half the data like a chess board.
 class GridRedBlackCartesian : public GridBase
 {
@@ -44,6 +51,9 @@ public:
             return source_cb;
         }
     };
+
+    GridRedBlackCartesian(GridBase *base) : GridRedBlackCartesian(base->_fdimensions,base->_simd_layout,base->_processors)  {};
+
     GridRedBlackCartesian(std::vector<int> &dimensions,
 			  std::vector<int> &simd_layout,
 			  std::vector<int> &processor_grid ) : GridBase(processor_grid)

From 3358a77c7a111193b6310ed7b68b2cee48d34775 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 25 May 2015 13:45:08 +0100
Subject: [PATCH 228/429] Better checkerboard tracking.

---
 lib/lattice/Grid_lattice_ET.h    | 41 ++++++++++++++++++++++++++
 lib/lattice/Grid_lattice_arith.h | 28 ++++++++++++++++--
 lib/lattice/Grid_lattice_base.h  | 50 ++++++++++++++++++++++++++++++++
 3 files changed, 117 insertions(+), 2 deletions(-)

diff --git a/lib/lattice/Grid_lattice_ET.h b/lib/lattice/Grid_lattice_ET.h
index b0205f2f..2a7172ea 100644
--- a/lib/lattice/Grid_lattice_ET.h
+++ b/lib/lattice/Grid_lattice_ET.h
@@ -91,6 +91,47 @@ inline void GridFromExpression( GridBase * &grid,const LatticeTrinaryExpression<
   GridFromExpression(grid,std::get<2>(expr.second));
 }
 
+
+//////////////////////////////////////////////////////////////////////////
+// Obtain the CB from an expression, ensuring conformable. This must follow a tree recursion
+//////////////////////////////////////////////////////////////////////////
+template<class T1, typename std::enable_if<is_lattice<T1>::value, T1>::type * =nullptr >
+inline void CBFromExpression(int &cb,const T1& lat)   // Lattice leaf
+{
+  if ( (cb==Odd) || (cb==Even) ) {
+    assert(cb==lat.checkerboard);
+  } 
+  cb=lat.checkerboard;
+  //  std::cout<<"Lattice leaf cb "<<cb<<std::endl;
+}
+template<class T1,typename std::enable_if<!is_lattice<T1>::value, T1>::type * = nullptr >
+inline void CBFromExpression(int &cb,const T1& notlat)   // non-lattice leaf
+{
+  //  std::cout<<"Non lattice leaf cb"<<cb<<std::endl;
+}
+template <typename Op, typename T1>
+inline void CBFromExpression(int &cb,const LatticeUnaryExpression<Op,T1 > &expr)
+{
+  CBFromExpression(cb,std::get<0>(expr.second));// recurse 
+  //  std::cout<<"Unary node cb "<<cb<<std::endl;
+}
+
+template <typename Op, typename T1, typename T2>
+inline void CBFromExpression(int &cb,const LatticeBinaryExpression<Op,T1,T2> &expr) 
+{
+  CBFromExpression(cb,std::get<0>(expr.second));// recurse
+  CBFromExpression(cb,std::get<1>(expr.second));
+  //  std::cout<<"Binary node cb "<<cb<<std::endl;
+}
+template <typename Op, typename T1, typename T2, typename T3>
+inline void CBFromExpression( int &cb,const LatticeTrinaryExpression<Op,T1,T2,T3 > &expr) 
+{
+  CBFromExpression(cb,std::get<0>(expr.second));// recurse
+  CBFromExpression(cb,std::get<1>(expr.second));
+  CBFromExpression(cb,std::get<2>(expr.second));
+  //  std::cout<<"Trinary node cb "<<cb<<std::endl;
+}
+
 ////////////////////////////////////////////
 // Unary operators and funcs
 ////////////////////////////////////////////
diff --git a/lib/lattice/Grid_lattice_arith.h b/lib/lattice/Grid_lattice_arith.h
index a6f59b56..dc5e57e8 100644
--- a/lib/lattice/Grid_lattice_arith.h
+++ b/lib/lattice/Grid_lattice_arith.h
@@ -9,6 +9,8 @@ namespace Grid {
   //////////////////////////////////////////////////////////////////////////////////////////////////////
   template<class obj1,class obj2,class obj3> strong_inline
     void mult(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
+    ret.checkerboard = lhs.checkerboard;
+    conformable(ret,rhs);
     conformable(lhs,rhs);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
@@ -24,6 +26,8 @@ PARALLEL_FOR_LOOP
   
   template<class obj1,class obj2,class obj3> strong_inline
     void mac(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
+    ret.checkerboard = lhs.checkerboard;
+    conformable(ret,rhs);
     conformable(lhs,rhs);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
@@ -39,6 +43,8 @@ PARALLEL_FOR_LOOP
   
   template<class obj1,class obj2,class obj3> strong_inline
     void sub(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
+    ret.checkerboard = lhs.checkerboard;
+    conformable(ret,rhs);
     conformable(lhs,rhs);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
@@ -53,6 +59,8 @@ PARALLEL_FOR_LOOP
   }
   template<class obj1,class obj2,class obj3> strong_inline
     void add(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
+    ret.checkerboard = lhs.checkerboard;
+    conformable(ret,rhs);
     conformable(lhs,rhs);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
@@ -71,6 +79,7 @@ PARALLEL_FOR_LOOP
   //////////////////////////////////////////////////////////////////////////////////////////////////////
   template<class obj1,class obj2,class obj3> strong_inline
     void mult(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
+    ret.checkerboard = lhs.checkerboard;
     conformable(lhs,ret);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
@@ -82,7 +91,8 @@ PARALLEL_FOR_LOOP
   
   template<class obj1,class obj2,class obj3> strong_inline
     void mac(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
-    conformable(lhs,ret);
+    ret.checkerboard = lhs.checkerboard;
+    conformable(ret,lhs);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
       obj1 tmp;
@@ -93,7 +103,8 @@ PARALLEL_FOR_LOOP
   
   template<class obj1,class obj2,class obj3> strong_inline
     void sub(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
-    conformable(lhs,ret);
+    ret.checkerboard = lhs.checkerboard;
+    conformable(ret,lhs);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
 #ifdef STREAMING_STORES
@@ -107,6 +118,7 @@ PARALLEL_FOR_LOOP
   }
   template<class obj1,class obj2,class obj3> strong_inline
     void add(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
+    ret.checkerboard = lhs.checkerboard;
     conformable(lhs,ret);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<lhs._grid->oSites();ss++){
@@ -125,6 +137,7 @@ PARALLEL_FOR_LOOP
   //////////////////////////////////////////////////////////////////////////////////////////////////////
     template<class obj1,class obj2,class obj3> strong_inline
     void mult(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
+    ret.checkerboard = rhs.checkerboard;
     conformable(ret,rhs);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<rhs._grid->oSites();ss++){
@@ -140,6 +153,7 @@ PARALLEL_FOR_LOOP
   
   template<class obj1,class obj2,class obj3> strong_inline
     void mac(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
+    ret.checkerboard = rhs.checkerboard;
     conformable(ret,rhs);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<rhs._grid->oSites();ss++){
@@ -155,6 +169,7 @@ PARALLEL_FOR_LOOP
   
   template<class obj1,class obj2,class obj3> strong_inline
     void sub(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
+    ret.checkerboard = rhs.checkerboard;
     conformable(ret,rhs);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<rhs._grid->oSites();ss++){
@@ -169,6 +184,7 @@ PARALLEL_FOR_LOOP
   }
   template<class obj1,class obj2,class obj3> strong_inline
     void add(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
+    ret.checkerboard = rhs.checkerboard;
     conformable(ret,rhs);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<rhs._grid->oSites();ss++){
@@ -184,6 +200,8 @@ PARALLEL_FOR_LOOP
   
   template<class sobj,class vobj> strong_inline
   void axpy(Lattice<vobj> &ret,sobj a,const Lattice<vobj> &x,const Lattice<vobj> &y){
+    ret.checkerboard = x.checkerboard;
+    conformable(ret,x);
     conformable(x,y);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<x._grid->oSites();ss++){
@@ -197,6 +215,8 @@ PARALLEL_FOR_LOOP
   }
   template<class sobj,class vobj> strong_inline
   void axpby(Lattice<vobj> &ret,sobj a,sobj b,const Lattice<vobj> &x,const Lattice<vobj> &y){
+    ret.checkerboard = x.checkerboard;
+    conformable(ret,x);
     conformable(x,y);
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<x._grid->oSites();ss++){
@@ -211,12 +231,16 @@ PARALLEL_FOR_LOOP
 
   template<class sobj,class vobj> strong_inline
   RealD axpy_norm(Lattice<vobj> &ret,sobj a,const Lattice<vobj> &x,const Lattice<vobj> &y){
+    ret.checkerboard = x.checkerboard;
+    conformable(ret,x);
     conformable(x,y);
     axpy(ret,a,x,y);
     return norm2(ret);
   }
   template<class sobj,class vobj> strong_inline
   RealD axpby_norm(Lattice<vobj> &ret,sobj a,sobj b,const Lattice<vobj> &x,const Lattice<vobj> &y){
+    ret.checkerboard = x.checkerboard;
+    conformable(ret,x);
     conformable(x,y);
     axpby(ret,a,b,x,y);
     return norm2(ret); // FIXME implement parallel norm in ss loop
diff --git a/lib/lattice/Grid_lattice_base.h b/lib/lattice/Grid_lattice_base.h
index 41f349b7..1d3b1efb 100644
--- a/lib/lattice/Grid_lattice_base.h
+++ b/lib/lattice/Grid_lattice_base.h
@@ -66,6 +66,16 @@ public:
   ////////////////////////////////////////////////////////////////////////////////
   template <typename Op, typename T1>                         strong_inline Lattice<vobj> & operator=(const LatticeUnaryExpression<Op,T1> &expr)
   {
+    GridBase *egrid(nullptr);
+    GridFromExpression(egrid,expr);
+    assert(egrid!=nullptr);
+    conformable(_grid,egrid);
+
+    int cb=-1;
+    CBFromExpression(cb,expr);
+    assert( (cb==Odd) || (cb==Even));
+    checkerboard=cb;
+
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
 #ifdef STREAMING_STORES
@@ -79,6 +89,16 @@ PARALLEL_FOR_LOOP
   }
   template <typename Op, typename T1,typename T2>             strong_inline Lattice<vobj> & operator=(const LatticeBinaryExpression<Op,T1,T2> &expr)
   {
+    GridBase *egrid(nullptr);
+    GridFromExpression(egrid,expr);
+    assert(egrid!=nullptr);
+    conformable(_grid,egrid);
+
+    int cb=-1;
+    CBFromExpression(cb,expr);
+    assert( (cb==Odd) || (cb==Even));
+    checkerboard=cb;
+
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
 #ifdef STREAMING_STORES
@@ -92,6 +112,16 @@ PARALLEL_FOR_LOOP
   }
   template <typename Op, typename T1,typename T2,typename T3> strong_inline Lattice<vobj> & operator=(const LatticeTrinaryExpression<Op,T1,T2,T3> &expr)
   {
+    GridBase *egrid(nullptr);
+    GridFromExpression(egrid,expr);
+    assert(egrid!=nullptr);
+    conformable(_grid,egrid);
+
+    int cb=-1;
+    CBFromExpression(cb,expr);
+    assert( (cb==Odd) || (cb==Even));
+    checkerboard=cb;
+
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
 #ifdef STREAMING_STORES
@@ -106,8 +136,15 @@ PARALLEL_FOR_LOOP
   //GridFromExpression is tricky to do
   template<class Op,class T1>
     Lattice(const LatticeUnaryExpression<Op,T1> & expr):    _grid(nullptr){
+
     GridFromExpression(_grid,expr);
     assert(_grid!=nullptr);
+
+    int cb=-1;
+    CBFromExpression(cb,expr);
+    assert( (cb==Odd) || (cb==Even));
+    checkerboard=cb;
+
     _odata.resize(_grid->oSites());
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
@@ -123,6 +160,12 @@ PARALLEL_FOR_LOOP
   Lattice(const LatticeBinaryExpression<Op,T1,T2> & expr):    _grid(nullptr){
     GridFromExpression(_grid,expr);
     assert(_grid!=nullptr);
+
+    int cb=-1;
+    CBFromExpression(cb,expr);
+    assert( (cb==Odd) || (cb==Even));
+    checkerboard=cb;
+
     _odata.resize(_grid->oSites());
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
@@ -138,6 +181,12 @@ PARALLEL_FOR_LOOP
   Lattice(const LatticeTrinaryExpression<Op,T1,T2,T3> & expr):    _grid(nullptr){
     GridFromExpression(_grid,expr);
     assert(_grid!=nullptr);
+
+    int cb=-1;
+    CBFromExpression(cb,expr);
+    assert( (cb==Odd) || (cb==Even));
+    checkerboard=cb;
+
     _odata.resize(_grid->oSites());
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
@@ -169,6 +218,7 @@ PARALLEL_FOR_LOOP
         return *this;
     }
     template<class robj> strong_inline Lattice<vobj> & operator = (const Lattice<robj> & r){
+      this->checkerboard = r.checkerboard;
       conformable(*this,r);
       std::cout<<"Lattice operator ="<<std::endl;
 PARALLEL_FOR_LOOP

From 55685b7cf565f2741ea0506beeca7e0e14f475f6 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 25 May 2015 13:45:32 +0100
Subject: [PATCH 229/429] Most cosmetic

---
 lib/math/Grid_math_tensors.h | 113 ++++++++++++++++++-----------------
 1 file changed, 57 insertions(+), 56 deletions(-)

diff --git a/lib/math/Grid_math_tensors.h b/lib/math/Grid_math_tensors.h
index 32ca78e5..4e6752e9 100644
--- a/lib/math/Grid_math_tensors.h
+++ b/lib/math/Grid_math_tensors.h
@@ -134,66 +134,65 @@ public:
   iVector<vtype,N> & operator= (iVector<vtype,N> &&copyme) = default;
   */
 
-    iVector<vtype,N> & operator= (const Zero &hero){
-        zeroit(*this);
-        return *this;
+  iVector<vtype,N> & operator= (const Zero &hero){
+    zeroit(*this);
+    return *this;
+  }
+  friend strong_inline void zeroit(iVector<vtype,N> &that){
+    for(int i=0;i<N;i++){
+      zeroit(that._internal[i]);
     }
-    friend strong_inline void zeroit(iVector<vtype,N> &that){
-        for(int i=0;i<N;i++){
-            zeroit(that._internal[i]);
-        }
+  }
+  friend strong_inline void prefetch(iVector<vtype,N> &that){
+    for(int i=0;i<N;i++) prefetch(that._internal[i]);
+  }
+  friend strong_inline void vstream(iVector<vtype,N> &out,const iVector<vtype,N> &in){
+    for(int i=0;i<N;i++){
+      vstream(out._internal[i],in._internal[i]);
     }
-    friend strong_inline void prefetch(iVector<vtype,N> &that){
-      for(int i=0;i<N;i++) prefetch(that._internal[i]);
+  }
+  friend strong_inline void permute(iVector<vtype,N> &out,const iVector<vtype,N> &in,int permutetype){
+    for(int i=0;i<N;i++){
+      permute(out._internal[i],in._internal[i],permutetype);
     }
-    friend strong_inline void vstream(iVector<vtype,N> &out,const iVector<vtype,N> &in){
-      for(int i=0;i<N;i++){
-	vstream(out._internal[i],in._internal[i]);
-      }
+  }
+  // Unary negation
+  friend strong_inline iVector<vtype,N> operator -(const iVector<vtype,N> &r) {
+    iVector<vtype,N> ret;
+    for(int i=0;i<N;i++) ret._internal[i]= -r._internal[i];
+    return ret;
+  }
+  // *=,+=,-= operators inherit from corresponding "*,-,+" behaviour
+  strong_inline iVector<vtype,N> &operator *=(const iScalar<vtype> &r) {
+    *this = (*this)*r;
+    return *this;
+  }
+  strong_inline iVector<vtype,N> &operator -=(const iVector<vtype,N> &r) {
+    *this = (*this)-r;
+    return *this;
+  }
+  strong_inline iVector<vtype,N> &operator +=(const iVector<vtype,N> &r) {
+    *this = (*this)+r;
+    return *this;
+  }
+  strong_inline vtype & operator ()(int i) {
+    return _internal[i];
+  }
+  strong_inline const vtype & operator ()(int i) const {
+    return _internal[i];
+  }
+  friend std::ostream& operator<< (std::ostream& stream, const iVector<vtype,N> &o){
+    stream<< "V<"<<N<<">{";
+    for(int i=0;i<N;i++) {
+      stream<<o._internal[i];
+      if (i<N-1)	stream<<",";
     }
-
-    friend strong_inline void permute(iVector<vtype,N> &out,const iVector<vtype,N> &in,int permutetype){
-      for(int i=0;i<N;i++){
-	permute(out._internal[i],in._internal[i],permutetype);
-      }
-    }
-    // Unary negation
-    friend strong_inline iVector<vtype,N> operator -(const iVector<vtype,N> &r) {
-        iVector<vtype,N> ret;
-        for(int i=0;i<N;i++) ret._internal[i]= -r._internal[i];
-        return ret;
-    }
-    // *=,+=,-= operators inherit from corresponding "*,-,+" behaviour
-    strong_inline iVector<vtype,N> &operator *=(const iScalar<vtype> &r) {
-        *this = (*this)*r;
-        return *this;
-    }
-    strong_inline iVector<vtype,N> &operator -=(const iVector<vtype,N> &r) {
-        *this = (*this)-r;
-        return *this;
-    }
-    strong_inline iVector<vtype,N> &operator +=(const iVector<vtype,N> &r) {
-        *this = (*this)+r;
-        return *this;
-    }
-    strong_inline vtype & operator ()(int i) {
-      return _internal[i];
-    }
-    strong_inline const vtype & operator ()(int i) const {
-      return _internal[i];
-    }
-    friend std::ostream& operator<< (std::ostream& stream, const iVector<vtype,N> &o){
-      stream<< "V<"<<N<<">{";
-      for(int i=0;i<N;i++) {
-	stream<<o._internal[i];
-	if (i<N-1)	stream<<",";
-      }
-      stream<<"}";
-      return stream;
-    };
-    //    strong_inline vtype && operator ()(int i) {
-    //      return _internal[i];
-    //    }
+    stream<<"}";
+    return stream;
+  };
+  //    strong_inline vtype && operator ()(int i) {
+  //      return _internal[i];
+  //    }
 };
     
 template<class vtype,int N> class iMatrix 
@@ -213,6 +212,8 @@ public:
 
   iMatrix(const Zero &z){ *this = zero; };
   iMatrix() =default;
+  iMatrix(scalar_type s)  { (*this) = s ;};// recurse down and hit the constructor for vector_type
+
   /*
   iMatrix(const iMatrix<vtype,N> &copyme)=default;
   iMatrix(iMatrix<vtype,N> &&copyme)=default;

From 1a9841a0f1991b0e4d58c0361936bea74bbc6c3f Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 25 May 2015 13:46:28 +0100
Subject: [PATCH 230/429] Better EO support letting Schur solver work

---
 lib/qcd/Grid_qcd.h             |   3 -
 lib/qcd/Grid_qcd_wilson_dop.cc | 317 +++++++++++++++++++--------------
 lib/qcd/Grid_qcd_wilson_dop.h  |  36 ++--
 3 files changed, 210 insertions(+), 146 deletions(-)

diff --git a/lib/qcd/Grid_qcd.h b/lib/qcd/Grid_qcd.h
index 0bf68632..959f3529 100644
--- a/lib/qcd/Grid_qcd.h
+++ b/lib/qcd/Grid_qcd.h
@@ -10,9 +10,6 @@ namespace QCD {
     static const int Nhs=2; // half spinor
     static const int Nds=8; // double stored gauge field
 
-    static const int CbRed  =0;
-    static const int CbBlack=1;
-    
     //////////////////////////////////////////////////////////////////////////////
     // QCD iMatrix types
     // Index conventions:                            Lorentz x Spin x Colour
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index 2020a9ec..318e18df 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -4,8 +4,8 @@
 namespace Grid {
 namespace QCD {
 
-const std::vector<int> WilsonMatrix::directions   ({0,1,2,3, 0, 1, 2, 3,0});
-const std::vector<int> WilsonMatrix::displacements({1,1,1,1,-1,-1,-1,-1,0});
+const std::vector<int> WilsonMatrix::directions   ({0,1,2,3, 0, 1, 2, 3});
+const std::vector<int> WilsonMatrix::displacements({1,1,1,1,-1,-1,-1,-1});
 
   // Should be in header?
 const int WilsonMatrix::Xp = 0;
@@ -16,7 +16,6 @@ const int WilsonMatrix::Xm = 4;
 const int WilsonMatrix::Ym = 5;
 const int WilsonMatrix::Zm = 6;
 const int WilsonMatrix::Tm = 7;
-  //const int WilsonMatrix::X0 = 8;
 
   class WilsonCompressor {
   public:
@@ -72,20 +71,27 @@ const int WilsonMatrix::Tm = 7;
     }
   };
 
-  WilsonMatrix::WilsonMatrix(LatticeGaugeField &_Umu,double _mass)
-  : Stencil(Umu._grid,npoint,0,directions,displacements),
+  WilsonMatrix::WilsonMatrix(LatticeGaugeField &_Umu,GridCartesian &Fgrid,GridRedBlackCartesian &Hgrid, double _mass)  : 
+    CheckerBoardedSparseMatrixBase<LatticeFermion>(&Fgrid,&Hgrid),
+    Stencil    (  _grid,npoint,Even,directions,displacements),
+    StencilEven(_cbgrid,npoint,Even,directions,displacements), // source is Even
+    StencilOdd (_cbgrid,npoint,Odd ,directions,displacements), // source is Odd
     mass(_mass),
-    Umu(_Umu._grid)
-{
-  // Allocate the required comms buffer
-  grid = _Umu._grid;
-  comm_buf.resize(Stencil._unified_buffer_size);
-  DoubleStore(Umu,_Umu);
-}
+    Umu(_grid),
+    UmuEven(_cbgrid),
+    UmuOdd (_cbgrid)
+  {
+    // Allocate the required comms buffer
+    comm_buf.resize(Stencil._unified_buffer_size); // this is always big enough to contain EO
+    
+    DoubleStore(Umu,_Umu);
+    pickCheckerboard(Even,UmuEven,Umu);
+    pickCheckerboard(Odd ,UmuOdd,Umu);
+  }
       
 void WilsonMatrix::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu)
 {
-  LatticeColourMatrix U(grid);
+  LatticeColourMatrix U(_grid);
 
   for(int mu=0;mu<Nd;mu++){
     U = peekIndex<LorentzIndex>(Umu,mu);
@@ -97,47 +103,63 @@ void WilsonMatrix::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeF
 
 RealD WilsonMatrix::M(const LatticeFermion &in, LatticeFermion &out)
 {
-  this->Dhop(in,out,0);
+  out.checkerboard=in.checkerboard;
+  Dhop(in,out,DaggerNo);
   out = (4+mass)*in - 0.5*out  ; // FIXME : axpby_norm! fusion fun
   return norm2(out);
 }
 RealD WilsonMatrix::Mdag(const LatticeFermion &in, LatticeFermion &out)
 {
-  this->Dhop(in,out,1);
+  out.checkerboard=in.checkerboard;
+  Dhop(in,out,DaggerYes);
   out = (4+mass)*in - 0.5*out  ; // FIXME : axpby_norm! fusion fun
   return norm2(out);
 }
 
 void WilsonMatrix::Meooe(const LatticeFermion &in, LatticeFermion &out)
 {
-  this->Dhop(in,out,0);
-  out = 0.5*out; // FIXME : scale factor in Dhop
+  if ( in.checkerboard == Odd ) {
+    DhopEO(in,out,DaggerNo);
+  } else {
+    DhopOE(in,out,DaggerNo);
+  }
+  out = (-0.5)*out; // FIXME : scale factor in Dhop
 }
 void WilsonMatrix::MeooeDag(const LatticeFermion &in, LatticeFermion &out)
 {
-  this->Dhop(in,out,1);
-  out = 0.5*out; // FIXME : scale factor in Dhop
+  if ( in.checkerboard == Odd ) {
+    DhopEO(in,out,DaggerYes);
+  } else {
+    DhopOE(in,out,DaggerYes);
+  }
+  out = (-0.5)*out; // FIXME : scale factor in Dhop
 }
 void WilsonMatrix::Mooee(const LatticeFermion &in, LatticeFermion &out)
 {
+  out.checkerboard = in.checkerboard;
   out = (4.0+mass)*in;
   return ;
 }
 void WilsonMatrix::MooeeDag(const LatticeFermion &in, LatticeFermion &out)
 {
-  this->Mooee(in,out);
+  out.checkerboard = in.checkerboard;
+  Mooee(in,out);
 }
 void WilsonMatrix::MooeeInv(const LatticeFermion &in, LatticeFermion &out)
 {
+  out.checkerboard = in.checkerboard;
   out = (1.0/(4.0+mass))*in;
   return ;
 }
 void WilsonMatrix::MooeeInvDag(const LatticeFermion &in, LatticeFermion &out)
 {
-  this->MooeeInv(in,out);
+  out.checkerboard = in.checkerboard;
+  MooeeInv(in,out);
 }
 
-void WilsonMatrix::DhopSite(int ss,const LatticeFermion &in, LatticeFermion &out)
+void WilsonMatrix::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
+			    std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+			    int ss,const LatticeFermion &in, LatticeFermion &out)
 {
     vHalfSpinColourVector  tmp;    
     vHalfSpinColourVector  chi;    
@@ -145,79 +167,78 @@ void WilsonMatrix::DhopSite(int ss,const LatticeFermion &in, LatticeFermion &out
     vHalfSpinColourVector Uchi;
     int offset,local,perm, ptype;
 
-    //    int ss = Stencil._LebesgueReorder[sss];
-    int ssu= ss;
+    //#define VERBOSE( A)  if ( ss<10 ) { std::cout << "site " <<ss << " " #A " neigh " << offset << " perm "<< perm <<std::endl;}    
 
     // Xp
-    offset = Stencil._offsets [Xp][ss];
-    local  = Stencil._is_local[Xp][ss];
-    perm   = Stencil._permute[Xp][ss];
+    offset = st._offsets [Xp][ss];
+    local  = st._is_local[Xp][ss];
+    perm   = st._permute[Xp][ss];
 
-    ptype  = Stencil._permute_type[Xp];
+    ptype  = st._permute_type[Xp];
     if ( local && perm ) {
       spProjXp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
       spProjXp(chi,in._odata[offset]);
     } else { 
-      chi=comm_buf[offset];
+      chi=buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Xp),&chi());
+    mult(&Uchi(),&U._odata[ss](Xp),&chi());
     spReconXp(result,Uchi);
 
     // Yp
-    offset = Stencil._offsets [Yp][ss];
-    local  = Stencil._is_local[Yp][ss];
-    perm   = Stencil._permute[Yp][ss];
-    ptype  = Stencil._permute_type[Yp];
+    offset = st._offsets [Yp][ss];
+    local  = st._is_local[Yp][ss];
+    perm   = st._permute[Yp][ss];
+    ptype  = st._permute_type[Yp];
     if ( local && perm ) {
       spProjYp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
       spProjYp(chi,in._odata[offset]);
     } else { 
-      chi=comm_buf[offset];
+      chi=buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Yp),&chi());
+    mult(&Uchi(),&U._odata[ss](Yp),&chi());
     accumReconYp(result,Uchi);
 
     // Zp
-    offset = Stencil._offsets [Zp][ss];
-    local  = Stencil._is_local[Zp][ss];
-    perm   = Stencil._permute[Zp][ss];
-    ptype  = Stencil._permute_type[Zp];
+    offset = st._offsets [Zp][ss];
+    local  = st._is_local[Zp][ss];
+    perm   = st._permute[Zp][ss];
+    ptype  = st._permute_type[Zp];
     if ( local && perm ) {
       spProjZp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
       spProjZp(chi,in._odata[offset]);
     } else { 
-      chi=comm_buf[offset];
+      chi=buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Zp),&chi());
+    mult(&Uchi(),&U._odata[ss](Zp),&chi());
     accumReconZp(result,Uchi);
 
     // Tp
-    offset = Stencil._offsets [Tp][ss];
-    local  = Stencil._is_local[Tp][ss];
-    perm   = Stencil._permute[Tp][ss];
-    ptype  = Stencil._permute_type[Tp];
+    offset = st._offsets [Tp][ss];
+    local  = st._is_local[Tp][ss];
+    perm   = st._permute[Tp][ss];
+    ptype  = st._permute_type[Tp];
     if ( local && perm ) {
       spProjTp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
       spProjTp(chi,in._odata[offset]);
     } else { 
-      chi=comm_buf[offset];
+      chi=buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Tp),&chi());
+    mult(&Uchi(),&U._odata[ss](Tp),&chi());
     accumReconTp(result,Uchi);
 
     // Xm
-    offset = Stencil._offsets [Xm][ss];
-    local  = Stencil._is_local[Xm][ss];
-    perm   = Stencil._permute[Xm][ss];
-    ptype  = Stencil._permute_type[Xm];
+    offset = st._offsets [Xm][ss];
+    local  = st._is_local[Xm][ss];
+    perm   = st._permute[Xm][ss];
+    ptype  = st._permute_type[Xm];
 
     if ( local && perm ) 
     {
@@ -226,17 +247,17 @@ void WilsonMatrix::DhopSite(int ss,const LatticeFermion &in, LatticeFermion &out
     } else if ( local ) {
       spProjXm(chi,in._odata[offset]);
     } else { 
-      chi=comm_buf[offset];
+      chi=buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Xm),&chi());
+    mult(&Uchi(),&U._odata[ss](Xm),&chi());
     accumReconXm(result,Uchi);
 
 
     // Ym
-    offset = Stencil._offsets [Ym][ss];
-    local  = Stencil._is_local[Ym][ss];
-    perm   = Stencil._permute[Ym][ss];
-    ptype  = Stencil._permute_type[Ym];
+    offset = st._offsets [Ym][ss];
+    local  = st._is_local[Ym][ss];
+    perm   = st._permute[Ym][ss];
+    ptype  = st._permute_type[Ym];
 
     if ( local && perm ) {
       spProjYm(tmp,in._odata[offset]);
@@ -244,46 +265,49 @@ void WilsonMatrix::DhopSite(int ss,const LatticeFermion &in, LatticeFermion &out
     } else if ( local ) {
       spProjYm(chi,in._odata[offset]);
     } else { 
-      chi=comm_buf[offset];
+      chi=buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Ym),&chi());
+    mult(&Uchi(),&U._odata[ss](Ym),&chi());
     accumReconYm(result,Uchi);
 
     // Zm
-    offset = Stencil._offsets [Zm][ss];
-    local  = Stencil._is_local[Zm][ss];
-    perm   = Stencil._permute[Zm][ss];
-    ptype  = Stencil._permute_type[Zm];
+    offset = st._offsets [Zm][ss];
+    local  = st._is_local[Zm][ss];
+    perm   = st._permute[Zm][ss];
+    ptype  = st._permute_type[Zm];
     if ( local && perm ) {
       spProjZm(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
       spProjZm(chi,in._odata[offset]);
     } else { 
-      chi=comm_buf[offset];
+      chi=buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Zm),&chi());
+    mult(&Uchi(),&U._odata[ss](Zm),&chi());
     accumReconZm(result,Uchi);
 
     // Tm
-    offset = Stencil._offsets [Tm][ss];
-    local  = Stencil._is_local[Tm][ss];
-    perm   = Stencil._permute[Tm][ss];
-    ptype  = Stencil._permute_type[Tm];
+    offset = st._offsets [Tm][ss];
+    local  = st._is_local[Tm][ss];
+    perm   = st._permute[Tm][ss];
+    ptype  = st._permute_type[Tm];
     if ( local && perm ) {
       spProjTm(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
       spProjTm(chi,in._odata[offset]);
     } else { 
-      chi=comm_buf[offset];
+      chi=buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Tm),&chi());
+    mult(&Uchi(),&U._odata[ss](Tm),&chi());
     accumReconTm(result,Uchi);
 
     vstream(out._odata[ss],result);
 }
-void WilsonMatrix::DhopSiteDag(int ss,const LatticeFermion &in, LatticeFermion &out)
+
+void WilsonMatrix::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
+			       std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+			       int ss,const LatticeFermion &in, LatticeFermion &out)
 {
     vHalfSpinColourVector  tmp;    
     vHalfSpinColourVector  chi;    
@@ -291,78 +315,76 @@ void WilsonMatrix::DhopSiteDag(int ss,const LatticeFermion &in, LatticeFermion &
     vHalfSpinColourVector Uchi;
     int offset,local,perm, ptype;
 
-    int ssu= ss;
-
     // Xp
-    offset = Stencil._offsets [Xm][ss];
-    local  = Stencil._is_local[Xm][ss];
-    perm   = Stencil._permute[Xm][ss];
+    offset = st._offsets [Xm][ss];
+    local  = st._is_local[Xm][ss];
+    perm   = st._permute[Xm][ss];
 
-    ptype  = Stencil._permute_type[Xm];
+    ptype  = st._permute_type[Xm];
     if ( local && perm ) {
       spProjXp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
       spProjXp(chi,in._odata[offset]);
     } else { 
-      chi=comm_buf[offset];
+      chi=buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Xm),&chi());
+    mult(&Uchi(),&U._odata[ss](Xm),&chi());
     spReconXp(result,Uchi);
 
     // Yp
-    offset = Stencil._offsets [Ym][ss];
-    local  = Stencil._is_local[Ym][ss];
-    perm   = Stencil._permute[Ym][ss];
-    ptype  = Stencil._permute_type[Ym];
+    offset = st._offsets [Ym][ss];
+    local  = st._is_local[Ym][ss];
+    perm   = st._permute[Ym][ss];
+    ptype  = st._permute_type[Ym];
     if ( local && perm ) {
       spProjYp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
       spProjYp(chi,in._odata[offset]);
     } else { 
-      chi=comm_buf[offset];
+      chi=buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Ym),&chi());
+    mult(&Uchi(),&U._odata[ss](Ym),&chi());
     accumReconYp(result,Uchi);
 
     // Zp
-    offset = Stencil._offsets [Zm][ss];
-    local  = Stencil._is_local[Zm][ss];
-    perm   = Stencil._permute[Zm][ss];
-    ptype  = Stencil._permute_type[Zm];
+    offset = st._offsets [Zm][ss];
+    local  = st._is_local[Zm][ss];
+    perm   = st._permute[Zm][ss];
+    ptype  = st._permute_type[Zm];
     if ( local && perm ) {
       spProjZp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
       spProjZp(chi,in._odata[offset]);
     } else { 
-      chi=comm_buf[offset];
+      chi=buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Zm),&chi());
+    mult(&Uchi(),&U._odata[ss](Zm),&chi());
     accumReconZp(result,Uchi);
 
     // Tp
-    offset = Stencil._offsets [Tm][ss];
-    local  = Stencil._is_local[Tm][ss];
-    perm   = Stencil._permute[Tm][ss];
-    ptype  = Stencil._permute_type[Tm];
+    offset = st._offsets [Tm][ss];
+    local  = st._is_local[Tm][ss];
+    perm   = st._permute[Tm][ss];
+    ptype  = st._permute_type[Tm];
     if ( local && perm ) {
       spProjTp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
       spProjTp(chi,in._odata[offset]);
     } else { 
-      chi=comm_buf[offset];
+      chi=buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Tm),&chi());
+    mult(&Uchi(),&U._odata[ss](Tm),&chi());
     accumReconTp(result,Uchi);
 
     // Xm
-    offset = Stencil._offsets [Xp][ss];
-    local  = Stencil._is_local[Xp][ss];
-    perm   = Stencil._permute[Xp][ss];
-    ptype  = Stencil._permute_type[Xp];
+    offset = st._offsets [Xp][ss];
+    local  = st._is_local[Xp][ss];
+    perm   = st._permute[Xp][ss];
+    ptype  = st._permute_type[Xp];
 
     if ( local && perm ) 
     {
@@ -371,17 +393,16 @@ void WilsonMatrix::DhopSiteDag(int ss,const LatticeFermion &in, LatticeFermion &
     } else if ( local ) {
       spProjXm(chi,in._odata[offset]);
     } else { 
-      chi=comm_buf[offset];
+      chi=buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Xp),&chi());
+    mult(&Uchi(),&U._odata[ss](Xp),&chi());
     accumReconXm(result,Uchi);
 
-
     // Ym
-    offset = Stencil._offsets [Yp][ss];
-    local  = Stencil._is_local[Yp][ss];
-    perm   = Stencil._permute[Yp][ss];
-    ptype  = Stencil._permute_type[Yp];
+    offset = st._offsets [Yp][ss];
+    local  = st._is_local[Yp][ss];
+    perm   = st._permute[Yp][ss];
+    ptype  = st._permute_type[Yp];
 
     if ( local && perm ) {
       spProjYm(tmp,in._odata[offset]);
@@ -389,67 +410,97 @@ void WilsonMatrix::DhopSiteDag(int ss,const LatticeFermion &in, LatticeFermion &
     } else if ( local ) {
       spProjYm(chi,in._odata[offset]);
     } else { 
-      chi=comm_buf[offset];
+      chi=buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Yp),&chi());
+    mult(&Uchi(),&U._odata[ss](Yp),&chi());
     accumReconYm(result,Uchi);
 
     // Zm
-    offset = Stencil._offsets [Zp][ss];
-    local  = Stencil._is_local[Zp][ss];
-    perm   = Stencil._permute[Zp][ss];
-    ptype  = Stencil._permute_type[Zp];
+    offset = st._offsets [Zp][ss];
+    local  = st._is_local[Zp][ss];
+    perm   = st._permute[Zp][ss];
+    ptype  = st._permute_type[Zp];
     if ( local && perm ) {
       spProjZm(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
       spProjZm(chi,in._odata[offset]);
     } else { 
-      chi=comm_buf[offset];
+      chi=buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Zp),&chi());
+    mult(&Uchi(),&U._odata[ss](Zp),&chi());
     accumReconZm(result,Uchi);
 
     // Tm
-    offset = Stencil._offsets [Tp][ss];
-    local  = Stencil._is_local[Tp][ss];
-    perm   = Stencil._permute[Tp][ss];
-    ptype  = Stencil._permute_type[Tp];
+    offset = st._offsets [Tp][ss];
+    local  = st._is_local[Tp][ss];
+    perm   = st._permute[Tp][ss];
+    ptype  = st._permute_type[Tp];
     if ( local && perm ) {
       spProjTm(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
       spProjTm(chi,in._odata[offset]);
     } else { 
-      chi=comm_buf[offset];
+      chi=buf[offset];
     }
-    mult(&Uchi(),&Umu._odata[ssu](Tp),&chi());
+    mult(&Uchi(),&U._odata[ss](Tp),&chi());
     accumReconTm(result,Uchi);
 
     vstream(out._odata[ss],result);
 }
 
-void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out,int dag)
+void WilsonMatrix::DhopInternal(CartesianStencil & st,LatticeDoubledGaugeField & U,
+				const LatticeFermion &in, LatticeFermion &out,int dag)
 {
-  assert((dag==0) ||(dag==1));
-
+  assert((dag==DaggerNo) ||(dag==DaggerYes));
   WilsonCompressor compressor(dag);
-  Stencil.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
 
-  if ( dag ) {
+  st.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
+
+  if ( dag == DaggerYes ) {
 PARALLEL_FOR_LOOP
-    for(int sss=0;sss<grid->oSites();sss++){
-      DhopSiteDag(sss,in,out);
+    for(int sss=0;sss<in._grid->oSites();sss++){
+      DhopSiteDag(st,U,comm_buf,sss,in,out);
     }
   } else {
 PARALLEL_FOR_LOOP
-    for(int sss=0;sss<grid->oSites();sss++){
-      DhopSite(sss,in,out);
+    for(int sss=0;sss<in._grid->oSites();sss++){
+      DhopSite(st,U,comm_buf,sss,in,out);
     }
   }
 }
+void WilsonMatrix::DhopOE(const LatticeFermion &in, LatticeFermion &out,int dag)
+{
+  conformable(in._grid,_cbgrid);    // verifies half grid
+  conformable(in._grid,out._grid); // drops the cb check
 
+  assert(in.checkerboard==Even);
+  out.checkerboard = Odd;
 
+  DhopInternal(StencilEven,UmuOdd,in,out,dag);
 }
+void WilsonMatrix::DhopEO(const LatticeFermion &in, LatticeFermion &out,int dag)
+{
+  conformable(in._grid,_cbgrid);    // verifies half grid
+  conformable(in._grid,out._grid); // drops the cb check
+
+  assert(in.checkerboard==Odd);
+  out.checkerboard = Even;
+
+  DhopInternal(StencilOdd,UmuEven,in,out,dag);
+}
+void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out,int dag)
+{
+  conformable(in._grid,_grid); // verifies full grid
+  conformable(in._grid,out._grid);
+
+  out.checkerboard = in.checkerboard;
+
+  DhopInternal(Stencil,Umu,in,out,dag);
 }
 
+}}
+
+
+
diff --git a/lib/qcd/Grid_qcd_wilson_dop.h b/lib/qcd/Grid_qcd_wilson_dop.h
index 8d93a926..96b29cd0 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.h
+++ b/lib/qcd/Grid_qcd_wilson_dop.h
@@ -11,14 +11,20 @@ namespace Grid {
       //NB r=1;
     public:
       double                        mass;
-      GridBase                     *grid;
+      //      GridBase                     *    grid; // Inherited
+      //      GridBase                     *  cbgrid;
 
-      // Copy of the gauge field 
-      LatticeDoubledGaugeField             Umu;
+      //Defines the stencils for even and odd
+      CartesianStencil Stencil; 
+      CartesianStencil StencilEven; 
+      CartesianStencil StencilOdd; 
 
-      //Defines the stencil
-      CartesianStencil              Stencil; 
-      static const int npoint=9;
+      // Copy of the gauge field , with even and odd subsets
+      LatticeDoubledGaugeField Umu;
+      LatticeDoubledGaugeField UmuEven;
+      LatticeDoubledGaugeField UmuOdd;
+
+      static const int npoint=8;
       static const std::vector<int> directions   ;
       static const std::vector<int> displacements;
       static const int Xp,Xm,Yp,Ym,Zp,Zm,Tp,Tm;
@@ -27,7 +33,7 @@ namespace Grid {
       std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  comm_buf;
 
       // Constructor
-      WilsonMatrix(LatticeGaugeField &Umu,double mass);
+      WilsonMatrix(LatticeGaugeField &_Umu,GridCartesian &Fgrid,GridRedBlackCartesian &Hgrid,double _mass);
 
       // DoubleStore
       void DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu);
@@ -45,9 +51,19 @@ namespace Grid {
       virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
 
       // non-hermitian hopping term; half cb or both
-      void Dhop(const LatticeFermion &in, LatticeFermion &out,int dag);
-      void DhopSite   (int ss,const LatticeFermion &in, LatticeFermion &out);
-      void DhopSiteDag(int ss,const LatticeFermion &in, LatticeFermion &out);
+      void Dhop  (const LatticeFermion &in, LatticeFermion &out,int dag);
+      void DhopOE(const LatticeFermion &in, LatticeFermion &out,int dag);
+      void DhopEO(const LatticeFermion &in, LatticeFermion &out,int dag);
+      void DhopInternal(CartesianStencil & st,LatticeDoubledGaugeField &U,
+			const LatticeFermion &in, LatticeFermion &out,int dag);
+      // These ones will need to be package intelligently. WilsonType base class
+      // for use by DWF etc..
+      void DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
+		    std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+		    int ss,const LatticeFermion &in, LatticeFermion &out);
+      void DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
+		       std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+		       int ss,const LatticeFermion &in, LatticeFermion &out);
 
       typedef iScalar<iMatrix<vComplex, Nc> > matrix;
 

From 2ae6214104f945ebcb4d6876e5052bb4dcb24878 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 25 May 2015 13:47:12 +0100
Subject: [PATCH 231/429] Schur complement based red-black inversion working

---
 benchmarks/Grid_wilson.cc           |  17 ++-
 benchmarks/Grid_wilson_cg_prec.cc   |  61 +++++++++
 benchmarks/Grid_wilson_cg_schur.cc  |  48 +++++++
 benchmarks/Grid_wilson_cg_unprec.cc |   7 +-
 benchmarks/Grid_wilson_evenodd.cc   | 201 ++++++++++++++++++++++++++++
 lib/simd/Grid_sse4.h                |   1 +
 lib/simd/Grid_vector_types.h        |  32 ++---
 tests/Grid_main.cc                  |   3 +-
 8 files changed, 336 insertions(+), 34 deletions(-)
 create mode 100644 benchmarks/Grid_wilson_cg_prec.cc
 create mode 100644 benchmarks/Grid_wilson_cg_schur.cc
 create mode 100644 benchmarks/Grid_wilson_evenodd.cc

diff --git a/benchmarks/Grid_wilson.cc b/benchmarks/Grid_wilson.cc
index adfb6fd4..32255b3e 100644
--- a/benchmarks/Grid_wilson.cc
+++ b/benchmarks/Grid_wilson.cc
@@ -24,7 +24,8 @@ int main (int argc, char ** argv)
   std::vector<int> latt_size   = GridDefaultLatt();
   std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
-  GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+  GridCartesian               Grid(latt_size,simd_layout,mpi_layout);
+  GridRedBlackCartesian     RBGrid(latt_size,simd_layout,mpi_layout);
 
   int threads = GridThread::GetThreads();
   std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
@@ -36,11 +37,11 @@ int main (int argc, char ** argv)
   //  pRNG.SeedFixedIntegers(seeds);
   pRNG.SeedRandomDevice();
 
-  LatticeFermion src(&Grid); random(pRNG,src);
+  LatticeFermion src   (&Grid); random(pRNG,src);
   LatticeFermion result(&Grid); result=zero;
   LatticeFermion    ref(&Grid);    ref=zero;
-  LatticeFermion    err(&Grid);    
   LatticeFermion    tmp(&Grid);    tmp=zero;
+  LatticeFermion    err(&Grid);    tmp=zero;
   LatticeGaugeField Umu(&Grid); random(pRNG,Umu);
   std::vector<LatticeColourMatrix> U(4,&Grid);
 
@@ -51,8 +52,11 @@ int main (int argc, char ** argv)
 
   // Only one non-zero (y)
   Umu=zero;
+  Complex cone(1.0,0.0);
   for(int nn=0;nn<Nd;nn++){
     random(pRNG,U[nn]);
+    if (nn!=0) U[nn]=zero;
+    else U[nn] = cone;
     pokeIndex<LorentzIndex>(Umu,U[nn],nn);
   }
 
@@ -78,7 +82,7 @@ int main (int argc, char ** argv)
   }
 
   RealD mass=0.1;
-  WilsonMatrix Dw(Umu,mass);
+  WilsonMatrix Dw(Umu,Grid,RBGrid,mass);
   
   std::cout << "Calling Dw"<<std::endl;
   int ncall=1000;
@@ -93,7 +97,7 @@ int main (int argc, char ** argv)
   std::cout << "norm result "<< norm2(result)<<std::endl;
   std::cout << "norm ref    "<< norm2(ref)<<std::endl;
   std::cout << "mflop/s =   "<< flops/(t1-t0)<<std::endl;
-  err = ref -result;
+  err = ref-result; 
   std::cout << "norm diff   "<< norm2(err)<<std::endl;
 
 
@@ -129,9 +133,8 @@ int main (int argc, char ** argv)
   std::cout << "Called DwDag"<<std::endl;
   std::cout << "norm result "<< norm2(result)<<std::endl;
   std::cout << "norm ref    "<< norm2(ref)<<std::endl;
-  err = ref -result;
+  err = ref-result; 
   std::cout << "norm diff   "<< norm2(err)<<std::endl;
 
-
   Grid_finalize();
 }
diff --git a/benchmarks/Grid_wilson_cg_prec.cc b/benchmarks/Grid_wilson_cg_prec.cc
new file mode 100644
index 00000000..e86ec820
--- /dev/null
+++ b/benchmarks/Grid_wilson_cg_prec.cc
@@ -0,0 +1,61 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+  GridCartesian               Grid(latt_size,simd_layout,mpi_layout);
+  GridRedBlackCartesian     RBGrid(latt_size,simd_layout,mpi_layout);
+
+  std::vector<int> seeds({1,2,3,4});
+  GridParallelRNG          pRNG(&Grid);  pRNG.SeedFixedIntegers(seeds);
+
+  LatticeFermion src(&Grid); random(pRNG,src);
+  RealD nrm = norm2(src);
+  LatticeFermion result(&Grid); result=zero;
+  LatticeGaugeField Umu(&Grid); random(pRNG,Umu);
+
+  std::vector<LatticeColourMatrix> U(4,&Grid);
+
+  for(int mu=0;mu<Nd;mu++){
+    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+  }
+  
+  RealD mass=0.5;
+  WilsonMatrix Dw(Umu,Grid,RBGrid,mass);
+
+  //  HermitianOperator<WilsonMatrix,LatticeFermion> HermOp(Dw);
+  //  ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
+  //  CG(HermOp,src,result);
+  
+  LatticeFermion    src_o(&RBGrid);
+  LatticeFermion result_o(&RBGrid);
+  pickCheckerboard(Odd,src_o,src);
+  result_o=zero;
+
+  HermitianCheckerBoardedOperator<WilsonMatrix,LatticeFermion> HermOpEO(Dw);
+  ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
+  CG(HermOpEO,src_o,result_o);
+
+  Grid_finalize();
+}
diff --git a/benchmarks/Grid_wilson_cg_schur.cc b/benchmarks/Grid_wilson_cg_schur.cc
new file mode 100644
index 00000000..af630ae1
--- /dev/null
+++ b/benchmarks/Grid_wilson_cg_schur.cc
@@ -0,0 +1,48 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+  GridCartesian               Grid(latt_size,simd_layout,mpi_layout);
+  GridRedBlackCartesian     RBGrid(latt_size,simd_layout,mpi_layout);
+
+  std::vector<int> seeds({1,2,3,4});
+  GridParallelRNG          pRNG(&Grid);  pRNG.SeedFixedIntegers(seeds);
+
+  LatticeGaugeField Umu(&Grid); random(pRNG,Umu);
+
+  LatticeFermion    src(&Grid); random(pRNG,src);
+  LatticeFermion result(&Grid); result=zero;
+  LatticeFermion resid(&Grid); 
+
+  RealD mass=0.5;
+  WilsonMatrix Dw(Umu,Grid,RBGrid,mass);
+
+  ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
+  SchurRedBlackSolve<LatticeFermion> SchurSolver(CG);
+
+  SchurSolver(Dw,src,result);
+  
+  Grid_finalize();
+}
diff --git a/benchmarks/Grid_wilson_cg_unprec.cc b/benchmarks/Grid_wilson_cg_unprec.cc
index 93023534..15302aab 100644
--- a/benchmarks/Grid_wilson_cg_unprec.cc
+++ b/benchmarks/Grid_wilson_cg_unprec.cc
@@ -24,7 +24,8 @@ int main (int argc, char ** argv)
   std::vector<int> latt_size   = GridDefaultLatt();
   std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
-  GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
+  GridCartesian               Grid(latt_size,simd_layout,mpi_layout);
+  GridRedBlackCartesian     RBGrid(latt_size,simd_layout,mpi_layout);
 
   std::vector<int> seeds({1,2,3,4});
   GridParallelRNG          pRNG(&Grid);  pRNG.SeedFixedIntegers(seeds);
@@ -46,11 +47,11 @@ int main (int argc, char ** argv)
   }
   
   RealD mass=0.5;
-  WilsonMatrix Dw(Umu,mass);
+  WilsonMatrix Dw(Umu,Grid,RBGrid,mass);
+
   HermitianOperator<WilsonMatrix,LatticeFermion> HermOp(Dw);
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   CG(HermOp,src,result);
 
-  
   Grid_finalize();
 }
diff --git a/benchmarks/Grid_wilson_evenodd.cc b/benchmarks/Grid_wilson_evenodd.cc
new file mode 100644
index 00000000..4bc8c357
--- /dev/null
+++ b/benchmarks/Grid_wilson_evenodd.cc
@@ -0,0 +1,201 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+  GridCartesian               Grid(latt_size,simd_layout,mpi_layout);
+  GridRedBlackCartesian     RBGrid(latt_size,simd_layout,mpi_layout);
+
+  int threads = GridThread::GetThreads();
+  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+
+  std::vector<int> seeds({1,2,3,4});
+
+  GridParallelRNG          pRNG(&Grid);
+  //  std::vector<int> seeds({1,2,3,4});
+  //  pRNG.SeedFixedIntegers(seeds);
+  pRNG.SeedRandomDevice();
+
+  LatticeFermion src   (&Grid); random(pRNG,src);
+  LatticeFermion phi   (&Grid); random(pRNG,phi);
+  LatticeFermion chi   (&Grid); random(pRNG,chi);
+  LatticeFermion result(&Grid); result=zero;
+  LatticeFermion    ref(&Grid);    ref=zero;
+  LatticeFermion    tmp(&Grid);    tmp=zero;
+  LatticeFermion    err(&Grid);    tmp=zero;
+  LatticeGaugeField Umu(&Grid); random(pRNG,Umu);
+  std::vector<LatticeColourMatrix> U(4,&Grid);
+
+  double volume=1;
+  for(int mu=0;mu<Nd;mu++){
+    volume=volume*latt_size[mu];
+  }  
+
+  // Only one non-zero (y)
+  Umu=zero;
+  for(int nn=0;nn<Nd;nn++){
+    random(pRNG,U[nn]);
+    pokeIndex<LorentzIndex>(Umu,U[nn],nn);
+  }
+
+  RealD mass=0.1;
+
+  WilsonMatrix Dw(Umu,Grid,RBGrid,mass);
+
+  LatticeFermion src_e   (&RBGrid);
+  LatticeFermion src_o   (&RBGrid);
+  LatticeFermion r_e   (&RBGrid);
+  LatticeFermion r_o   (&RBGrid);
+  LatticeFermion r_eo  (&Grid);
+
+  const int Even=0;
+  const int Odd=1;
+  std::cout<<"=========================================================="<<std::endl;
+  std::cout<<"= Testing that Deo + Doe = Dunprec "<<std::endl;
+  std::cout<<"=========================================================="<<std::endl;
+
+  pickCheckerboard(Even,src_e,src);
+  pickCheckerboard(Odd,src_o,src);
+
+  Dw.Meooe(src_e,r_o);  std::cout<<"Applied Meo"<<std::endl;
+  Dw.Meooe(src_o,r_e);  std::cout<<"Applied Moe"<<std::endl;
+  Dw.Dhop (src,ref,0);
+
+  setCheckerboard(r_eo,r_o);
+  setCheckerboard(r_eo,r_e);
+
+  ref = (-0.5)*ref;
+  err= ref - r_eo;
+  std::cout << "EO norm diff   "<< norm2(err)<< " "<<norm2(ref)<< " " << norm2(r_eo) <<std::endl;
+
+  LatticeComplex cerr(&Grid);
+  cerr = localInnerProduct(err,err);
+
+  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<"= Test Ddagger is the dagger of D by requiring                "<<std::endl;
+  std::cout<<"=  < phi | Deo | chi > * = < chi | Deo^dag| phi>  "<<std::endl;
+  std::cout<<"=============================================================="<<std::endl;
+  
+  LatticeFermion chi_e   (&RBGrid);
+  LatticeFermion chi_o   (&RBGrid);
+
+  LatticeFermion dchi_e  (&RBGrid);
+  LatticeFermion dchi_o  (&RBGrid);
+
+  LatticeFermion phi_e   (&RBGrid);
+  LatticeFermion phi_o   (&RBGrid);
+
+  LatticeFermion dphi_e  (&RBGrid);
+  LatticeFermion dphi_o  (&RBGrid);
+
+
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+  pickCheckerboard(Even,phi_e,phi);
+  pickCheckerboard(Odd ,phi_o,phi);
+
+  Dw.Meooe(chi_e,dchi_o);
+  Dw.Meooe(chi_o,dchi_e);
+  Dw.MeooeDag(phi_e,dphi_o);
+  Dw.MeooeDag(phi_o,dphi_e);
+
+  ComplexD pDce = innerProduct(phi_e,dchi_e);
+  ComplexD pDco = innerProduct(phi_o,dchi_o);
+  ComplexD cDpe = innerProduct(chi_e,dphi_e);
+  ComplexD cDpo = innerProduct(chi_o,dphi_o);
+
+  std::cout <<"e "<<pDce<<" "<<cDpe <<std::endl;
+  std::cout <<"o "<<pDco<<" "<<cDpo <<std::endl;
+
+  std::cout <<"pDce - conj(cDpo) "<< pDce-conj(cDpo) <<std::endl;
+  std::cout <<"pDco - conj(cDpe) "<< pDco-conj(cDpe) <<std::endl;
+
+  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<"= Test MeeInv Mee = 1                                         "<<std::endl;
+  std::cout<<"=============================================================="<<std::endl;
+
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+
+  Dw.Mooee(chi_e,src_e);
+  Dw.MooeeInv(src_e,phi_e);
+
+  Dw.Mooee(chi_o,src_o);
+  Dw.MooeeInv(src_o,phi_o);
+  
+  setCheckerboard(phi,phi_e);
+  setCheckerboard(phi,phi_o);
+
+  err = phi-chi;
+  std::cout << "norm diff   "<< norm2(err)<< std::endl;
+
+  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<"= Test MeeInvDag MeeDag = 1                                   "<<std::endl;
+  std::cout<<"=============================================================="<<std::endl;
+
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+
+  Dw.MooeeDag(chi_e,src_e);
+  Dw.MooeeInvDag(src_e,phi_e);
+
+  Dw.MooeeDag(chi_o,src_o);
+  Dw.MooeeInvDag(src_o,phi_o);
+  
+  setCheckerboard(phi,phi_e);
+  setCheckerboard(phi,phi_o);
+
+  err = phi-chi;
+  std::cout << "norm diff   "<< norm2(err)<< std::endl;
+
+  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<"= Test MpcDagMpc is Hermitian              "<<std::endl;
+  std::cout<<"=============================================================="<<std::endl;
+  
+  random(pRNG,phi);
+  random(pRNG,chi);
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+  pickCheckerboard(Even,phi_e,phi);
+  pickCheckerboard(Odd ,phi_o,phi);
+  RealD t1,t2;
+
+  Dw.MpcDagMpc(chi_e,dchi_e,t1,t2);
+  Dw.MpcDagMpc(chi_o,dchi_o,t1,t2);
+
+  Dw.MpcDagMpc(phi_e,dphi_e,t1,t2);
+  Dw.MpcDagMpc(phi_o,dphi_o,t1,t2);
+
+  pDce = innerProduct(phi_e,dchi_e);
+  pDco = innerProduct(phi_o,dchi_o);
+  cDpe = innerProduct(chi_e,dphi_e);
+  cDpo = innerProduct(chi_o,dphi_o);
+
+  std::cout <<"e "<<pDce<<" "<<cDpe <<std::endl;
+  std::cout <<"o "<<pDco<<" "<<cDpo <<std::endl;
+
+  std::cout <<"pDce - conj(cDpo) "<< pDco-conj(cDpo) <<std::endl;
+  std::cout <<"pDco - conj(cDpe) "<< pDce-conj(cDpe) <<std::endl;
+  
+  Grid_finalize();
+}
diff --git a/lib/simd/Grid_sse4.h b/lib/simd/Grid_sse4.h
index ddc3490b..9a467987 100644
--- a/lib/simd/Grid_sse4.h
+++ b/lib/simd/Grid_sse4.h
@@ -283,6 +283,7 @@ namespace Optimization {
 //////////////////////////////////////////////////////////////////////////////////////
 // Here assign types 
 namespace Grid {
+
   typedef __m128 SIMD_Ftype;  // Single precision type
   typedef __m128d SIMD_Dtype; // Double precision type
   typedef __m128i SIMD_Itype; // Integer type
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index 57480621..36665b70 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -13,31 +13,24 @@
 namespace Grid {
 
   // To take the floating point type of real/complex type
-  template <typename T> 
-    struct RealPart {
-      typedef T type;
-    };
-  template <typename T> 
-    struct RealPart< std::complex<T> >{
+  template <typename T> struct RealPart {
+    typedef T type;
+  };
+  template <typename T> struct RealPart< std::complex<T> >{
     typedef T type;
   };
 
   // type alias used to simplify the syntax of std::enable_if
-  template <typename T> using Invoke =
-    typename T::type;
-  template <typename Condition, typename ReturnType> using EnableIf =
-    Invoke<std::enable_if<Condition::value, ReturnType>>;
-  template <typename Condition, typename ReturnType> using NotEnableIf =
-    Invoke<std::enable_if<!Condition::value, ReturnType>>;
+  template <typename T> using Invoke                                  =  typename T::type;
+  template <typename Condition, typename ReturnType> using EnableIf   =    Invoke<std::enable_if<Condition::value, ReturnType>>;
+  template <typename Condition, typename ReturnType> using NotEnableIf=    Invoke<std::enable_if<!Condition::value, ReturnType>>;
   
 
 
   ////////////////////////////////////////////////////////
   // Check for complexity with type traits
-  template <typename T> 
-    struct is_complex : std::false_type {};
-  template < typename T > 
-    struct is_complex< std::complex<T> >: std::true_type {};
+  template <typename T>     struct is_complex : std::false_type {};
+  template < typename T >   struct is_complex< std::complex<T> >: std::true_type {};
   ////////////////////////////////////////////////////////
   // Define the operation templates functors
   // general forms to allow for vsplat syntax
@@ -86,8 +79,6 @@ namespace Grid {
     Grid_simd(Real a){
       vsplat(*this,Scalar_type(a));
     };
-
-
        
     ///////////////////////////////////////////////
     // mac, mult, sub, add, adj
@@ -126,10 +117,6 @@ namespace Grid {
       friend inline void vtrue (Grid_simd &ret){vsplat(ret,0xFFFFFFFF);}
     template <  class S = Scalar_type, EnableIf<std::is_integral < S >, int> = 0 > 
       friend inline void vfalse(vInteger &ret){vsplat(ret,0);}
-
-
-   
- 
    
     ////////////////////////////////////
     // Arithmetic operator overloads +,-,*
@@ -165,7 +152,6 @@ namespace Grid {
 	ret.v = binary<Vector_type>(a.v,b.v, MultSIMD());
 	return ret;
       };
-    
 
     ////////////////////////////////////////////////////////////////////////
     // FIXME:  gonna remove these load/store, get, set, prefetch
diff --git a/tests/Grid_main.cc b/tests/Grid_main.cc
index 10e74099..1a761042 100644
--- a/tests/Grid_main.cc
+++ b/tests/Grid_main.cc
@@ -1,8 +1,9 @@
 #include "Grid.h"
 
 //DEBUG
+#ifdef SSE4
 #include "simd/Grid_vector_types.h"
-
+#endif
 
 using namespace std;
 using namespace Grid;

From 3a6ff2d7b86233ae1da83119c3f97b5cb3177ed7 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 25 May 2015 14:43:08 +0100
Subject: [PATCH 232/429] Makefile update

---
 Makefile.in            | 44 +++++++++++++++++++-----------
 aclocal.m4             | 61 ++++++++++++++++++++++--------------------
 benchmarks/Makefile.am | 11 +++++++-
 config.guess           |  2 +-
 config.sub             |  2 +-
 configure              | 13 ++++-----
 6 files changed, 79 insertions(+), 54 deletions(-)

diff --git a/Makefile.in b/Makefile.in
index b6894ef6..b9fba168 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -1,7 +1,7 @@
-# Makefile.in generated by automake 1.14.1 from Makefile.am.
+# Makefile.in generated by automake 1.15 from Makefile.am.
 # @configure_input@
 
-# Copyright (C) 1994-2013 Free Software Foundation, Inc.
+# Copyright (C) 1994-2014 Free Software Foundation, Inc.
 
 # This Makefile.in is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -14,7 +14,17 @@
 
 @SET_MAKE@
 VPATH = @srcdir@
-am__is_gnu_make = test -n '$(MAKEFILE_LIST)' && test -n '$(MAKELEVEL)'
+am__is_gnu_make = { \
+  if test -z '$(MAKELEVEL)'; then \
+    false; \
+  elif test -n '$(MAKE_HOST)'; then \
+    true; \
+  elif test -n '$(MAKE_VERSION)' && test -n '$(CURDIR)'; then \
+    true; \
+  else \
+    false; \
+  fi; \
+}
 am__make_running_with_option = \
   case $${target_option-} in \
       ?) ;; \
@@ -79,14 +89,12 @@ build_triplet = @build@
 host_triplet = @host@
 target_triplet = @target@
 subdir = .
-DIST_COMMON = INSTALL NEWS README AUTHORS ChangeLog \
-	$(srcdir)/Makefile.in $(srcdir)/Makefile.am \
-	$(top_srcdir)/configure $(am__configure_deps) COPYING TODO \
-	compile config.guess config.sub depcomp install-sh missing
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/configure.ac
 am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
 	$(ACLOCAL_M4)
+DIST_COMMON = $(srcdir)/Makefile.am $(top_srcdir)/configure \
+	$(am__configure_deps) $(am__DIST_COMMON)
 am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \
  configure.lineno config.status.lineno
 mkinstalldirs = $(install_sh) -d
@@ -149,6 +157,9 @@ ETAGS = etags
 CTAGS = ctags
 CSCOPE = cscope
 DIST_SUBDIRS = $(SUBDIRS)
+am__DIST_COMMON = $(srcdir)/Makefile.in AUTHORS COPYING ChangeLog \
+	INSTALL NEWS README TODO compile config.guess config.sub \
+	depcomp install-sh missing
 DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
 distdir = $(PACKAGE)-$(VERSION)
 top_distdir = $(distdir)
@@ -314,7 +325,6 @@ $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu Makefile'; \
 	$(am__cd) $(top_srcdir) && \
 	  $(AUTOMAKE) --gnu Makefile
-.PRECIOUS: Makefile
 Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
 	@case '$?' in \
 	  *config.status*) \
@@ -521,15 +531,15 @@ dist-xz: distdir
 	$(am__post_remove_distdir)
 
 dist-tarZ: distdir
-	@echo WARNING: "Support for shar distribution archives is" \
-	               "deprecated." >&2
+	@echo WARNING: "Support for distribution archives compressed with" \
+		       "legacy program 'compress' is deprecated." >&2
 	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	tardir=$(distdir) && $(am__tar) | compress -c >$(distdir).tar.Z
 	$(am__post_remove_distdir)
 
 dist-shar: distdir
-	@echo WARNING: "Support for distribution archives compressed with" \
-		       "legacy program 'compress' is deprecated." >&2
+	@echo WARNING: "Support for shar distribution archives is" \
+	               "deprecated." >&2
 	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	shar $(distdir) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).shar.gz
 	$(am__post_remove_distdir)
@@ -565,17 +575,17 @@ distcheck: dist
 	esac
 	chmod -R a-w $(distdir)
 	chmod u+w $(distdir)
-	mkdir $(distdir)/_build $(distdir)/_inst
+	mkdir $(distdir)/_build $(distdir)/_build/sub $(distdir)/_inst
 	chmod a-w $(distdir)
 	test -d $(distdir)/_build || exit 0; \
 	dc_install_base=`$(am__cd) $(distdir)/_inst && pwd | sed -e 's,^[^:\\/]:[\\/],/,'` \
 	  && dc_destdir="$${TMPDIR-/tmp}/am-dc-$$$$/" \
 	  && am__cwd=`pwd` \
-	  && $(am__cd) $(distdir)/_build \
-	  && ../configure \
+	  && $(am__cd) $(distdir)/_build/sub \
+	  && ../../configure \
 	    $(AM_DISTCHECK_CONFIGURE_FLAGS) \
 	    $(DISTCHECK_CONFIGURE_FLAGS) \
-	    --srcdir=.. --prefix="$$dc_install_base" \
+	    --srcdir=../.. --prefix="$$dc_install_base" \
 	  && $(MAKE) $(AM_MAKEFLAGS) \
 	  && $(MAKE) $(AM_MAKEFLAGS) dvi \
 	  && $(MAKE) $(AM_MAKEFLAGS) check \
@@ -749,6 +759,8 @@ uninstall-am:
 	maintainer-clean-generic mostlyclean mostlyclean-generic pdf \
 	pdf-am ps ps-am tags tags-am uninstall uninstall-am
 
+.PRECIOUS: Makefile
+
 
 # Tell versions [3.59,3.63) of GNU make to not export all variables.
 # Otherwise a system limit (for SysV at least) may be exceeded.
diff --git a/aclocal.m4 b/aclocal.m4
index 389763bf..bf79d078 100644
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -1,6 +1,6 @@
-# generated automatically by aclocal 1.14.1 -*- Autoconf -*-
+# generated automatically by aclocal 1.15 -*- Autoconf -*-
 
-# Copyright (C) 1996-2013 Free Software Foundation, Inc.
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -20,7 +20,7 @@ You have another version of autoconf.  It may work, but is not guaranteed to.
 If you have problems, you may need to regenerate the build system entirely.
 To do so, use the procedure documented by the package, typically 'autoreconf'.])])
 
-# Copyright (C) 2002-2013 Free Software Foundation, Inc.
+# Copyright (C) 2002-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -32,10 +32,10 @@ To do so, use the procedure documented by the package, typically 'autoreconf'.])
 # generated from the m4 files accompanying Automake X.Y.
 # (This private macro should not be called outside this file.)
 AC_DEFUN([AM_AUTOMAKE_VERSION],
-[am__api_version='1.14'
+[am__api_version='1.15'
 dnl Some users find AM_AUTOMAKE_VERSION and mistake it for a way to
 dnl require some minimum version.  Point them to the right macro.
-m4_if([$1], [1.14.1], [],
+m4_if([$1], [1.15], [],
       [AC_FATAL([Do not call $0, use AM_INIT_AUTOMAKE([$1]).])])dnl
 ])
 
@@ -51,14 +51,14 @@ m4_define([_AM_AUTOCONF_VERSION], [])
 # Call AM_AUTOMAKE_VERSION and AM_AUTOMAKE_VERSION so they can be traced.
 # This function is AC_REQUIREd by AM_INIT_AUTOMAKE.
 AC_DEFUN([AM_SET_CURRENT_AUTOMAKE_VERSION],
-[AM_AUTOMAKE_VERSION([1.14.1])dnl
+[AM_AUTOMAKE_VERSION([1.15])dnl
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
 _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 
 # AM_AUX_DIR_EXPAND                                         -*- Autoconf -*-
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -103,15 +103,14 @@ _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 # configured tree to be moved without reconfiguration.
 
 AC_DEFUN([AM_AUX_DIR_EXPAND],
-[dnl Rely on autoconf to set up CDPATH properly.
-AC_PREREQ([2.50])dnl
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+[AC_REQUIRE([AC_CONFIG_AUX_DIR_DEFAULT])dnl
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 ])
 
 # AM_CONDITIONAL                                            -*- Autoconf -*-
 
-# Copyright (C) 1997-2013 Free Software Foundation, Inc.
+# Copyright (C) 1997-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -142,7 +141,7 @@ AC_CONFIG_COMMANDS_PRE(
 Usually this means the macro was only invoked conditionally.]])
 fi])])
 
-# Copyright (C) 1999-2013 Free Software Foundation, Inc.
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -333,7 +332,7 @@ _AM_SUBST_NOTMAKE([am__nodep])dnl
 
 # Generate code to set up dependency tracking.              -*- Autoconf -*-
 
-# Copyright (C) 1999-2013 Free Software Foundation, Inc.
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -409,7 +408,7 @@ AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 
 # Do all the work for Automake.                             -*- Autoconf -*-
 
-# Copyright (C) 1996-2013 Free Software Foundation, Inc.
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -499,8 +498,8 @@ AC_REQUIRE([AC_PROG_MKDIR_P])dnl
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 AC_SUBST([mkdir_p], ['$(MKDIR_P)'])
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
 AC_REQUIRE([AC_PROG_AWK])dnl
 AC_REQUIRE([AC_PROG_MAKE_SET])dnl
 AC_REQUIRE([AM_SET_LEADING_DOT])dnl
@@ -573,7 +572,11 @@ to "yes", and re-run configure.
 END
     AC_MSG_ERROR([Your 'rm' program is bad, sorry.])
   fi
-fi])
+fi
+dnl The trailing newline in this macro's definition is deliberate, for
+dnl backward compatibility and to allow trailing 'dnl'-style comments
+dnl after the AM_INIT_AUTOMAKE invocation. See automake bug#16841.
+])
 
 dnl Hook into '_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
 dnl add the conditional right here, as _AC_COMPILER_EXEEXT may be further
@@ -602,7 +605,7 @@ for _am_header in $config_headers :; do
 done
 echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_count])
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -613,7 +616,7 @@ echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_co
 # Define $install_sh.
 AC_DEFUN([AM_PROG_INSTALL_SH],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -623,7 +626,7 @@ if test x"${install_sh}" != xset; then
 fi
 AC_SUBST([install_sh])])
 
-# Copyright (C) 2003-2013 Free Software Foundation, Inc.
+# Copyright (C) 2003-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -644,7 +647,7 @@ AC_SUBST([am__leading_dot])])
 
 # Check to see how 'make' treats includes.	            -*- Autoconf -*-
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -694,7 +697,7 @@ rm -f confinc confmf
 
 # Fake the existence of programs that GNU maintainers use.  -*- Autoconf -*-
 
-# Copyright (C) 1997-2013 Free Software Foundation, Inc.
+# Copyright (C) 1997-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -733,7 +736,7 @@ fi
 
 # Helper functions for option handling.                     -*- Autoconf -*-
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -764,7 +767,7 @@ AC_DEFUN([_AM_IF_OPTION],
 
 # Check to make sure that the build environment is sane.    -*- Autoconf -*-
 
-# Copyright (C) 1996-2013 Free Software Foundation, Inc.
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -845,7 +848,7 @@ AC_CONFIG_COMMANDS_PRE(
 rm -f conftest.file
 ])
 
-# Copyright (C) 2009-2013 Free Software Foundation, Inc.
+# Copyright (C) 2009-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -905,7 +908,7 @@ AC_SUBST([AM_BACKSLASH])dnl
 _AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
 ])
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -933,7 +936,7 @@ fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 AC_SUBST([INSTALL_STRIP_PROGRAM])])
 
-# Copyright (C) 2006-2013 Free Software Foundation, Inc.
+# Copyright (C) 2006-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -952,7 +955,7 @@ AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 
 # Check how to create a tarball.                            -*- Autoconf -*-
 
-# Copyright (C) 2004-2013 Free Software Foundation, Inc.
+# Copyright (C) 2004-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
diff --git a/benchmarks/Makefile.am b/benchmarks/Makefile.am
index fc513e86..95ae5eca 100644
--- a/benchmarks/Makefile.am
+++ b/benchmarks/Makefile.am
@@ -5,14 +5,23 @@ AM_LDFLAGS = -L$(top_builddir)/lib
 #
 # Test code
 #
-bin_PROGRAMS = Grid_wilson Grid_comms Grid_memory_bandwidth Grid_su3 Grid_wilson_cg_unprec
+bin_PROGRAMS = Grid_wilson Grid_comms Grid_memory_bandwidth Grid_su3 Grid_wilson_cg_unprec Grid_wilson_evenodd  Grid_wilson_cg_prec Grid_wilson_cg_schur
 
 Grid_wilson_SOURCES = Grid_wilson.cc
 Grid_wilson_LDADD = -lGrid
 
+Grid_wilson_evenodd_SOURCES = Grid_wilson_evenodd.cc
+Grid_wilson_evenodd_LDADD = -lGrid
+
 Grid_wilson_cg_unprec_SOURCES = Grid_wilson_cg_unprec.cc
 Grid_wilson_cg_unprec_LDADD = -lGrid
 
+Grid_wilson_cg_prec_SOURCES = Grid_wilson_cg_prec.cc
+Grid_wilson_cg_prec_LDADD = -lGrid
+
+Grid_wilson_cg_schur_SOURCES = Grid_wilson_cg_schur.cc
+Grid_wilson_cg_schur_LDADD = -lGrid
+
 Grid_comms_SOURCES = Grid_comms.cc
 Grid_comms_LDADD = -lGrid
 
diff --git a/config.guess b/config.guess
index 5f6aa02d..a12faba2 120000
--- a/config.guess
+++ b/config.guess
@@ -1 +1 @@
-/usr/share/automake-1.14/config.guess
\ No newline at end of file
+/opt/local/share/automake-1.15/config.guess
\ No newline at end of file
diff --git a/config.sub b/config.sub
index 0abfe18c..e3c9b5ca 120000
--- a/config.sub
+++ b/config.sub
@@ -1 +1 @@
-/usr/share/automake-1.14/config.sub
\ No newline at end of file
+/opt/local/share/automake-1.15/config.sub
\ No newline at end of file
diff --git a/configure b/configure
index 8c9e8c59..6e27ab11 100755
--- a/configure
+++ b/configure
@@ -2466,7 +2466,7 @@ test -n "$target_alias" &&
     NONENONEs,x,x, &&
   program_prefix=${target_alias}-
 
-am__api_version='1.14'
+am__api_version='1.15'
 
 # Find a good install program.  We prefer a C program (faster),
 # so one script is as good as another.  But avoid the broken or
@@ -2638,8 +2638,8 @@ test "$program_suffix" != NONE &&
 ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
 program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
 
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 
 if test x"${MISSING+set}" != xset; then
   case $am_aux_dir in
@@ -2658,7 +2658,7 @@ else
 $as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
 fi
 
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -2986,8 +2986,8 @@ MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 mkdir_p='$(MKDIR_P)'
 
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
 # Always define AMTAR for backward compatibility.  Yes, it's still used
 # in the wild :-(  We should find a proper way to deprecate it ...
 AMTAR='$${TAR-tar}'
@@ -3046,6 +3046,7 @@ END
 fi
 
 
+
 ac_config_headers="$ac_config_headers lib/Grid_config.h"
 
 # Check whether --enable-silent-rules was given.

From ece86f717bc115afd0eed7593417f7a99da12784 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Tue, 26 May 2015 12:02:54 +0900
Subject: [PATCH 233/429] checked performance of new vector libaries.

Added check for c++11 support on the configure.ac
---
 Makefile.in                     |   4 +-
 aclocal.m4                      |   1 +
 configure                       | 186 ++++++++++++++++++++++++++++++++
 configure.ac                    |   4 +-
 lib/Grid_config.h               |   3 +
 lib/Grid_config.h.in            |   3 +
 lib/Grid_simd.h                 |   4 +
 lib/cshift/Grid_cshift_common.h |  12 ++-
 lib/math/Grid_math_tensors.h    |  14 +++
 lib/simd/Grid_avx.h             |   6 +-
 m4/ax_cxx_compile_stdcxx_11.m4  | 167 ++++++++++++++++++++++++++++
 tests/Grid_main.cc              |  53 +--------
 12 files changed, 398 insertions(+), 59 deletions(-)
 create mode 100644 m4/ax_cxx_compile_stdcxx_11.m4

diff --git a/Makefile.in b/Makefile.in
index b6894ef6..d473c2df 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -84,7 +84,8 @@ DIST_COMMON = INSTALL NEWS README AUTHORS ChangeLog \
 	$(top_srcdir)/configure $(am__configure_deps) COPYING TODO \
 	compile config.guess config.sub depcomp install-sh missing
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
-am__aclocal_m4_deps = $(top_srcdir)/configure.ac
+am__aclocal_m4_deps = $(top_srcdir)/m4/ax_cxx_compile_stdcxx_11.m4 \
+	$(top_srcdir)/configure.ac
 am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
 	$(ACLOCAL_M4)
 am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \
@@ -212,6 +213,7 @@ ECHO_T = @ECHO_T@
 EGREP = @EGREP@
 EXEEXT = @EXEEXT@
 GREP = @GREP@
+HAVE_CXX11 = @HAVE_CXX11@
 INSTALL = @INSTALL@
 INSTALL_DATA = @INSTALL_DATA@
 INSTALL_PROGRAM = @INSTALL_PROGRAM@
diff --git a/aclocal.m4 b/aclocal.m4
index 389763bf..a3d1bc9c 100644
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -1083,3 +1083,4 @@ AC_SUBST([am__tar])
 AC_SUBST([am__untar])
 ]) # _AM_PROG_TAR
 
+m4_include([m4/ax_cxx_compile_stdcxx_11.m4])
diff --git a/configure b/configure
index 8c9e8c59..b7bd49f0 100755
--- a/configure
+++ b/configure
@@ -633,6 +633,7 @@ BUILD_COMMS_MPI_TRUE
 EGREP
 GREP
 CXXCPP
+HAVE_CXX11
 RANLIB
 OPENMP_CXXFLAGS
 am__fastdepCXX_FALSE
@@ -3965,6 +3966,191 @@ else
   RANLIB="$ac_cv_prog_RANLIB"
 fi
 
+    ax_cxx_compile_cxx11_required=true
+  ac_ext=cpp
+ac_cpp='$CXXCPP $CPPFLAGS'
+ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
+  ac_success=no
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX supports C++11 features by default" >&5
+$as_echo_n "checking whether $CXX supports C++11 features by default... " >&6; }
+if ${ax_cv_cxx_compile_cxx11+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+  template <typename T>
+    struct check
+    {
+      static_assert(sizeof(int) <= sizeof(T), "not big enough");
+    };
+
+    struct Base {
+    virtual void f() {}
+    };
+    struct Child : public Base {
+    virtual void f() override {}
+    };
+
+    typedef check<check<bool>> right_angle_brackets;
+
+    int a;
+    decltype(a) b;
+
+    typedef check<int> check_type;
+    check_type c;
+    check_type&& cr = static_cast<check_type&&>(c);
+
+    auto d = a;
+    auto l = [](){};
+    // Prevent Clang error: unused variable 'l' [-Werror,-Wunused-variable]
+    struct use_l { use_l() { l(); } };
+
+    // http://stackoverflow.com/questions/13728184/template-aliases-and-sfinae
+    // Clang 3.1 fails with headers of libstd++ 4.8.3 when using std::function because of this
+    namespace test_template_alias_sfinae {
+        struct foo {};
+
+        template<typename T>
+        using member = typename T::member_type;
+
+        template<typename T>
+        void func(...) {}
+
+        template<typename T>
+        void func(member<T>*) {}
+
+        void test();
+
+        void test() {
+            func<foo>(0);
+        }
+    }
+
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  ax_cv_cxx_compile_cxx11=yes
+else
+  ax_cv_cxx_compile_cxx11=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_cxx_compile_cxx11" >&5
+$as_echo "$ax_cv_cxx_compile_cxx11" >&6; }
+  if test x$ax_cv_cxx_compile_cxx11 = xyes; then
+    ac_success=yes
+  fi
+
+
+
+    if test x$ac_success = xno; then
+            for switch in -std=c++11 -std=c++0x +std=c++11; do
+      cachevar=`$as_echo "ax_cv_cxx_compile_cxx11_$switch" | $as_tr_sh`
+      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX supports C++11 features with $switch" >&5
+$as_echo_n "checking whether $CXX supports C++11 features with $switch... " >&6; }
+if eval \${$cachevar+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_save_CXXFLAGS="$CXXFLAGS"
+         CXXFLAGS="$CXXFLAGS $switch"
+         cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+  template <typename T>
+    struct check
+    {
+      static_assert(sizeof(int) <= sizeof(T), "not big enough");
+    };
+
+    struct Base {
+    virtual void f() {}
+    };
+    struct Child : public Base {
+    virtual void f() override {}
+    };
+
+    typedef check<check<bool>> right_angle_brackets;
+
+    int a;
+    decltype(a) b;
+
+    typedef check<int> check_type;
+    check_type c;
+    check_type&& cr = static_cast<check_type&&>(c);
+
+    auto d = a;
+    auto l = [](){};
+    // Prevent Clang error: unused variable 'l' [-Werror,-Wunused-variable]
+    struct use_l { use_l() { l(); } };
+
+    // http://stackoverflow.com/questions/13728184/template-aliases-and-sfinae
+    // Clang 3.1 fails with headers of libstd++ 4.8.3 when using std::function because of this
+    namespace test_template_alias_sfinae {
+        struct foo {};
+
+        template<typename T>
+        using member = typename T::member_type;
+
+        template<typename T>
+        void func(...) {}
+
+        template<typename T>
+        void func(member<T>*) {}
+
+        void test();
+
+        void test() {
+            func<foo>(0);
+        }
+    }
+
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  eval $cachevar=yes
+else
+  eval $cachevar=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+         CXXFLAGS="$ac_save_CXXFLAGS"
+fi
+eval ac_res=\$$cachevar
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+      if eval test x\$$cachevar = xyes; then
+        CXXFLAGS="$CXXFLAGS $switch"
+        ac_success=yes
+        break
+      fi
+    done
+  fi
+  ac_ext=cpp
+ac_cpp='$CXXCPP $CPPFLAGS'
+ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
+
+  if test x$ax_cxx_compile_cxx11_required = xtrue; then
+    if test x$ac_success = xno; then
+      as_fn_error $? "*** A compiler with support for C++11 language features is required." "$LINENO" 5
+    fi
+  else
+    if test x$ac_success = xno; then
+      HAVE_CXX11=0
+      { $as_echo "$as_me:${as_lineno-$LINENO}: No compiler with C++11 support was found" >&5
+$as_echo "$as_me: No compiler with C++11 support was found" >&6;}
+    else
+      HAVE_CXX11=1
+
+$as_echo "#define HAVE_CXX11 1" >>confdefs.h
+
+    fi
+
+
+  fi
+
+
 
 # Checks for libraries.
 #AX_GCC_VAR_ATTRIBUTE(aligned)
diff --git a/configure.ac b/configure.ac
index bfcbfcef..54a36eb1 100644
--- a/configure.ac
+++ b/configure.ac
@@ -3,7 +3,7 @@
 #
 # Project Grid package  
 # 
-# Time-stamp: <2015-05-22 15:46:09 neo>
+# Time-stamp: <2015-05-25 14:54:34 neo>
 
 AC_PREREQ([2.63])
 AC_INIT([Grid], [1.0], [paboyle@ph.ed.ac.uk])
@@ -26,6 +26,8 @@ AC_LANG(C++)
 AC_PROG_CXX
 AC_OPENMP
 AC_PROG_RANLIB
+AX_CXX_COMPILE_STDCXX_11(noext, mandatory)
+
 
 # Checks for libraries.
 #AX_GCC_VAR_ATTRIBUTE(aligned)
diff --git a/lib/Grid_config.h b/lib/Grid_config.h
index 78582f3e..2397894f 100644
--- a/lib/Grid_config.h
+++ b/lib/Grid_config.h
@@ -16,6 +16,9 @@
 /* GRID_COMMS_NONE */
 #define GRID_COMMS_NONE 1
 
+/* define if the compiler supports basic C++11 syntax */
+/* #undef HAVE_CXX11 */
+
 /* Define to 1 if you have the declaration of `be64toh', and to 0 if you
    don't. */
 #define HAVE_DECL_BE64TOH 1
diff --git a/lib/Grid_config.h.in b/lib/Grid_config.h.in
index b7f56d5b..6f05d6cb 100644
--- a/lib/Grid_config.h.in
+++ b/lib/Grid_config.h.in
@@ -15,6 +15,9 @@
 /* GRID_COMMS_NONE */
 #undef GRID_COMMS_NONE
 
+/* define if the compiler supports basic C++11 syntax */
+#undef HAVE_CXX11
+
 /* Define to 1 if you have the declaration of `be64toh', and to 0 if you
    don't. */
 #undef HAVE_DECL_BE64TOH
diff --git a/lib/Grid_simd.h b/lib/Grid_simd.h
index 9484c0d6..bdcb2d1b 100644
--- a/lib/Grid_simd.h
+++ b/lib/Grid_simd.h
@@ -103,6 +103,10 @@ namespace Grid {
     inline void sub (RealF * __restrict__ y,const RealF * __restrict__ l,const RealF *__restrict__ r){ *y = (*l) - (*r); }
     inline void add (RealF * __restrict__ y,const RealF * __restrict__ l,const RealF *__restrict__ r){ *y = (*l) + (*r); }
     
+    inline void vstream(ComplexF &l, const ComplexF &r){ l=r;}
+    inline void vstream(ComplexD &l, const ComplexD &r){ l=r;}
+    inline void vstream(RealF &l, const RealF &r){ l=r;}
+    inline void vstream(RealD &l, const RealD &r){ l=r;}
 
 
   class Zero{};
diff --git a/lib/cshift/Grid_cshift_common.h b/lib/cshift/Grid_cshift_common.h
index 65c0cb87..90fc10b7 100644
--- a/lib/cshift/Grid_cshift_common.h
+++ b/lib/cshift/Grid_cshift_common.h
@@ -160,13 +160,21 @@ template<class vobj> void Copy_plane(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int
 PARALLEL_NESTED_LOOP2
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-      
+      /*
       int o =n*rhs._grid->_slice_stride[dimension];
       int ocb=1<<lhs._grid->CheckerBoardFromOindex(o+b);
       if ( ocb&cbmask ) {
 	lhs._odata[lo+o+b]=rhs._odata[ro+o+b];
       }
-      
+      */
+
+      int o =n*rhs._grid->_slice_stride[dimension]+b;
+      int ocb=1<<lhs._grid->CheckerBoardFromOindex(o);
+      if ( ocb&cbmask ) {
+	//lhs._odata[lo+o]=rhs._odata[ro+o];
+	vstream(lhs._odata[lo+o],rhs._odata[ro+o]);
+      }
+
     }
   }
   
diff --git a/lib/math/Grid_math_tensors.h b/lib/math/Grid_math_tensors.h
index a0424576..94cd3d58 100644
--- a/lib/math/Grid_math_tensors.h
+++ b/lib/math/Grid_math_tensors.h
@@ -38,6 +38,10 @@ public:
   iScalar(scalar_type s) : _internal(s) {};// recurse down and hit the constructor for vector_type
   iScalar(const Zero &z){ *this = zero; };
 
+
+
+
+
     iScalar<vtype> & operator= (const Zero &hero){
       zeroit(*this);
       return *this;
@@ -206,6 +210,16 @@ public:
   iMatrix(const Zero &z){ *this = zero; };
   iMatrix() =default;
 
+  // No copy constructor...
+  
+  iMatrix& operator=(const iMatrix& rhs){
+    for(int i=0;i<N;i++)
+      for(int j=0;j<N;j++)
+	vstream(_internal[i][j],rhs._internal[i][j]);
+    return *this;
+  }; 
+  
+ 
 
   iMatrix<vtype,N> & operator= (const Zero &hero){
     zeroit(*this);
diff --git a/lib/simd/Grid_avx.h b/lib/simd/Grid_avx.h
index ea796122..303f30c5 100644
--- a/lib/simd/Grid_avx.h
+++ b/lib/simd/Grid_avx.h
@@ -4,7 +4,7 @@
 
   Using intrinsics
 */
-// Time-stamp: <2015-05-22 15:51:24 neo>
+// Time-stamp: <2015-05-22 18:58:27 neo>
 //----------------------------------------------------------------------
 
 #include <immintrin.h>
@@ -307,9 +307,7 @@ namespace Optimization {
     conv.v = b;
     switch (perm){
       // 8x32 bits=>3 permutes
-    case 2: 
-      conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); 
-      break;
+    case 2: conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); break;
     case 1: conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2)); break;
     case 0: conv.f = _mm256_permute2f128_ps(conv.f,conv.f,0x01); break;
     default: assert(0); break;
diff --git a/m4/ax_cxx_compile_stdcxx_11.m4 b/m4/ax_cxx_compile_stdcxx_11.m4
new file mode 100644
index 00000000..395b13d2
--- /dev/null
+++ b/m4/ax_cxx_compile_stdcxx_11.m4
@@ -0,0 +1,167 @@
+# ============================================================================
+#  http://www.gnu.org/software/autoconf-archive/ax_cxx_compile_stdcxx_11.html
+# ============================================================================
+#
+# SYNOPSIS
+#
+#   AX_CXX_COMPILE_STDCXX_11([ext|noext],[mandatory|optional])
+#
+# DESCRIPTION
+#
+#   Check for baseline language coverage in the compiler for the C++11
+#   standard; if necessary, add switches to CXXFLAGS to enable support.
+#
+#   The first argument, if specified, indicates whether you insist on an
+#   extended mode (e.g. -std=gnu++11) or a strict conformance mode (e.g.
+#   -std=c++11).  If neither is specified, you get whatever works, with
+#   preference for an extended mode.
+#
+#   The second argument, if specified 'mandatory' or if left unspecified,
+#   indicates that baseline C++11 support is required and that the macro
+#   should error out if no mode with that support is found.  If specified
+#   'optional', then configuration proceeds regardless, after defining
+#   HAVE_CXX11 if and only if a supporting mode is found.
+#
+# LICENSE
+#
+#   Copyright (c) 2008 Benjamin Kosnik <bkoz@redhat.com>
+#   Copyright (c) 2012 Zack Weinberg <zackw@panix.com>
+#   Copyright (c) 2013 Roy Stogner <roystgnr@ices.utexas.edu>
+#   Copyright (c) 2014, 2015 Google Inc.; contributed by Alexey Sokolov <sokolov@google.com>
+#
+#   Copying and distribution of this file, with or without modification, are
+#   permitted in any medium without royalty provided the copyright notice
+#   and this notice are preserved. This file is offered as-is, without any
+#   warranty.
+
+#serial 11
+
+m4_define([_AX_CXX_COMPILE_STDCXX_11_testbody], [[
+  template <typename T>
+    struct check
+    {
+      static_assert(sizeof(int) <= sizeof(T), "not big enough");
+    };
+
+    struct Base {
+    virtual void f() {}
+    };
+    struct Child : public Base {
+    virtual void f() override {}
+    };
+
+    typedef check<check<bool>> right_angle_brackets;
+
+    int a;
+    decltype(a) b;
+
+    typedef check<int> check_type;
+    check_type c;
+    check_type&& cr = static_cast<check_type&&>(c);
+
+    auto d = a;
+    auto l = [](){};
+    // Prevent Clang error: unused variable 'l' [-Werror,-Wunused-variable]
+    struct use_l { use_l() { l(); } };
+
+    // http://stackoverflow.com/questions/13728184/template-aliases-and-sfinae
+    // Clang 3.1 fails with headers of libstd++ 4.8.3 when using std::function because of this
+    namespace test_template_alias_sfinae {
+        struct foo {};
+
+        template<typename T>
+        using member = typename T::member_type;
+
+        template<typename T>
+        void func(...) {}
+
+        template<typename T>
+        void func(member<T>*) {}
+
+        void test();
+
+        void test() {
+            func<foo>(0);
+        }
+    }
+]])
+
+AC_DEFUN([AX_CXX_COMPILE_STDCXX_11], [dnl
+  m4_if([$1], [], [],
+        [$1], [ext], [],
+        [$1], [noext], [],
+        [m4_fatal([invalid argument `$1' to AX_CXX_COMPILE_STDCXX_11])])dnl
+  m4_if([$2], [], [ax_cxx_compile_cxx11_required=true],
+        [$2], [mandatory], [ax_cxx_compile_cxx11_required=true],
+        [$2], [optional], [ax_cxx_compile_cxx11_required=false],
+        [m4_fatal([invalid second argument `$2' to AX_CXX_COMPILE_STDCXX_11])])
+  AC_LANG_PUSH([C++])dnl
+  ac_success=no
+  AC_CACHE_CHECK(whether $CXX supports C++11 features by default,
+  ax_cv_cxx_compile_cxx11,
+  [AC_COMPILE_IFELSE([AC_LANG_SOURCE([_AX_CXX_COMPILE_STDCXX_11_testbody])],
+    [ax_cv_cxx_compile_cxx11=yes],
+    [ax_cv_cxx_compile_cxx11=no])])
+  if test x$ax_cv_cxx_compile_cxx11 = xyes; then
+    ac_success=yes
+  fi
+
+  m4_if([$1], [noext], [], [dnl
+  if test x$ac_success = xno; then
+    for switch in -std=gnu++11 -std=gnu++0x; do
+      cachevar=AS_TR_SH([ax_cv_cxx_compile_cxx11_$switch])
+      AC_CACHE_CHECK(whether $CXX supports C++11 features with $switch,
+                     $cachevar,
+        [ac_save_CXXFLAGS="$CXXFLAGS"
+         CXXFLAGS="$CXXFLAGS $switch"
+         AC_COMPILE_IFELSE([AC_LANG_SOURCE([_AX_CXX_COMPILE_STDCXX_11_testbody])],
+          [eval $cachevar=yes],
+          [eval $cachevar=no])
+         CXXFLAGS="$ac_save_CXXFLAGS"])
+      if eval test x\$$cachevar = xyes; then
+        CXXFLAGS="$CXXFLAGS $switch"
+        ac_success=yes
+        break
+      fi
+    done
+  fi])
+
+  m4_if([$1], [ext], [], [dnl
+  if test x$ac_success = xno; then
+    dnl HP's aCC needs +std=c++11 according to:
+    dnl http://h21007.www2.hp.com/portal/download/files/unprot/aCxx/PDF_Release_Notes/769149-001.pdf
+    for switch in -std=c++11 -std=c++0x +std=c++11; do
+      cachevar=AS_TR_SH([ax_cv_cxx_compile_cxx11_$switch])
+      AC_CACHE_CHECK(whether $CXX supports C++11 features with $switch,
+                     $cachevar,
+        [ac_save_CXXFLAGS="$CXXFLAGS"
+         CXXFLAGS="$CXXFLAGS $switch"
+         AC_COMPILE_IFELSE([AC_LANG_SOURCE([_AX_CXX_COMPILE_STDCXX_11_testbody])],
+          [eval $cachevar=yes],
+          [eval $cachevar=no])
+         CXXFLAGS="$ac_save_CXXFLAGS"])
+      if eval test x\$$cachevar = xyes; then
+        CXXFLAGS="$CXXFLAGS $switch"
+        ac_success=yes
+        break
+      fi
+    done
+  fi])
+  AC_LANG_POP([C++])
+  if test x$ax_cxx_compile_cxx11_required = xtrue; then
+    if test x$ac_success = xno; then
+      AC_MSG_ERROR([*** A compiler with support for C++11 language features is required.])
+    fi
+  else
+    if test x$ac_success = xno; then
+      HAVE_CXX11=0
+      AC_MSG_NOTICE([No compiler with C++11 support was found])
+    else
+      HAVE_CXX11=1
+      AC_DEFINE(HAVE_CXX11,1,
+                [define if the compiler supports basic C++11 syntax])
+    fi
+
+    AC_SUBST(HAVE_CXX11)
+  fi
+])
diff --git a/tests/Grid_main.cc b/tests/Grid_main.cc
index c1f1c5d0..fd198f30 100644
--- a/tests/Grid_main.cc
+++ b/tests/Grid_main.cc
@@ -105,7 +105,7 @@ int main (int argc, char ** argv)
     fflush(stdout);
     
 
-    /* 
+     
     cVec = cMat * cVec;  // LatticeColourVector     = LatticeColourMatrix     * LatticeColourVector
     sVec = sMat * sVec;  // LatticeSpinVector       = LatticeSpinMatrix       * LatticeSpinVector
     scVec= scMat * scVec;// LatticeSpinColourVector = LatticeSpinColourMatrix * LatticeSpinColourVector
@@ -146,7 +146,7 @@ int main (int argc, char ** argv)
     scalar=trace(scalar);
     scalar=localInnerProduct(cVec,cVec);
     scalar=localNorm2(cVec);
-    */
+    
 //     -=,+=,*=,()
 //     add,+,sub,-,mult,mac,*
 //     adj,conjugate
@@ -162,50 +162,7 @@ int main (int argc, char ** argv)
     scMat = sMat*scMat;  // LatticeSpinColourMatrix = LatticeSpinMatrix       * LatticeSpinColourMatrix
 
     
-    /*
-#ifdef SSE4
-    ///////// Tests the new class Grid_simd 
-    std::complex<double> ctest(3.0,2.0);
-    std::complex<float> ctestf(3.0,2.0);
-    MyComplexF TestMe1(1.0); // fills only real part
-    MyComplexD TestMe2(ctest);
-    MyComplexD TestMe3(ctest);// compiler generate conversion of basic types
-    //MyRealF TestMe5(ctest);// Must generate compiler error
-    MyRealD TestRe1(2.0); 
-    MyRealF TestRe2(3.0); 
- 
-    vone(TestRe2);
-
-    MyComplexF TestMe6(ctestf);
-    MyComplexF TestMe7(ctestf);  
-    
-    MyComplexD TheSum= TestMe2*TestMe3;
-    MyComplexF TheSumF= TestMe6*TestMe7;
-
-    
-
-    double dsum[2];
-    _mm_store_pd(dsum, TheSum.v);
-    for (int i =0; i< 2; i++)
-      printf("%f\n", dsum[i]);
-    MyComplexD TheSumI = timesMinusI(TheSum);
-    MyComplexF TheSumIF = timesMinusI(TheSumF);
-
-    float fsum[4];
-    _mm_store_ps(fsum, TheSumF.v);
-    for (int i =0; i< 4; i++)
-      printf("%f\n", fsum[i]);
-
-    vstore(TheSumI, &ctest);
-    std::complex<float> sum = Reduce(TheSumF);
-    std::cout << ctest<< std::endl;
-    std::cout << sum<< std::endl;
-
-#endif
-    */
     ///////////////////////
-    /*
-    printf("DEBUG: calling 3.5 \n");
     // Non-lattice (const objects) * Lattice
     ColourMatrix cm;
     SpinColourMatrix scm;
@@ -225,7 +182,6 @@ int main (int argc, char ** argv)
     vscm = vscm*cplx;
     scMat = scMat*cplx;
 
-    printf("DEBUG: calling 3.7 \n");
     scm = cplx*scm;
     vscm = cplx*vscm;
     scMat = cplx*scMat;
@@ -233,14 +189,12 @@ int main (int argc, char ** argv)
     vscm = myint*vscm;
     scMat = scMat*myint;
     
-    printf("DEBUG: calling 3.9 \n");
     scm = scm*mydouble;
     vscm = vscm*mydouble;
     scMat = scMat*mydouble;
     scMat = mydouble*scMat;
     cMat = mydouble*cMat;
   
-    printf("DEBUG: calling 4 \n");
     sMat = adj(sMat);       // LatticeSpinMatrix adjoint
     sMat = iGammaFive*sMat; // SpinMatrix * LatticeSpinMatrix
     sMat = GammaFive*sMat;  // SpinMatrix * LatticeSpinMatrix
@@ -293,8 +247,6 @@ int main (int argc, char ** argv)
       pokeIndex<1> (c_m,c,0,0);
     }
 
-    */
-
     FooBar = Bar;
  
     /*
@@ -392,7 +344,6 @@ int main (int argc, char ** argv)
     t0=usecond();
     for(int i=0;i<ncall;i++){
       Fine.Barrier();
-      //Cshift(Bar,1,-1);
       mult(FooBar,Foo,Cshift(Bar,1,-1));
       //mult(FooBar,Foo,Bar);
       //FooBar = Foo * Bar; // this is bad

From 3f576830f91c5411c86fb72d5f58dec635277ca9 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Tue, 26 May 2015 13:31:10 +0900
Subject: [PATCH 234/429] Cleaning up simd files

---
 TODO                         |  4 +-
 lib/Grid_simd.h              | 90 ------------------------------------
 lib/simd/Grid_vector_types.h | 66 ++++++++++++++++++++++++--
 3 files changed, 64 insertions(+), 96 deletions(-)

diff --git a/TODO b/TODO
index ed4dedd4..670c53e3 100644
--- a/TODO
+++ b/TODO
@@ -1,8 +1,8 @@
 ================================================================
 *** Hacks and bug fixes to clean up and Audits
 ================================================================
-* Base class to share common code between vRealF, VComplexF etc... 
-  - Performance check on Guido's reimplementation strategy
+* Base class to share common code between vRealF, VComplexF etc... done
+  - Performance check on Guido's reimplementation strategy  - (GUIDO) tested and no difference was found, merged
 
 * FIXME audit
 
diff --git a/lib/Grid_simd.h b/lib/Grid_simd.h
index 19b30f03..39eb4654 100644
--- a/lib/Grid_simd.h
+++ b/lib/Grid_simd.h
@@ -95,100 +95,10 @@ namespace Grid {
   template<>            inline void zeroit(RealF &arg){ arg=0; };
   template<>            inline void zeroit(RealD &arg){ arg=0; };
 
-  // Eventually delete this part
-#if defined (SSE4)
-    typedef __m128 fvec;
-    typedef __m128d dvec;
-    typedef __m128 cvec;
-    typedef __m128d zvec;
-    typedef __m128i ivec;
-#endif
-#if defined (AVX1) || defined (AVX2)
-    typedef __m256 fvec;
-    typedef __m256d dvec;
-    typedef __m256  cvec;
-    typedef __m256d zvec;
-    typedef __m256i ivec;
-#endif
-#if defined (AVX512)
-    typedef __m512  fvec;
-    typedef __m512d dvec;
-    typedef __m512  cvec;
-    typedef __m512d zvec;
-    typedef __m512i ivec;
-#endif
-#if defined (QPX)
-    typedef float  fvec __attribute__ ((vector_size (16))); // QPX has same SIMD width irrespective of precision
-    typedef float  cvec __attribute__ ((vector_size (16)));
-    
-    typedef vector4double dvec;
-    typedef vector4double zvec;
-#endif
-  
-#if defined (AVX1) || defined (AVX2) || defined (AVX512)
-    inline void v_prefetch0(int size, const char *ptr){
-          for(int i=0;i<size;i+=64){ //  Define L1 linesize above// What about SSE?
-            _mm_prefetch(ptr+i+4096,_MM_HINT_T1);
-            _mm_prefetch(ptr+i+512,_MM_HINT_T0);
-          }
-    }
-#else 
-    inline void v_prefetch0(int size, const char *ptr){};
-#endif
-
-
-//////////////////////////////////////////////////////////
-// Permute
-// Permute 0 every ABCDEFGH -> BA DC FE HG
-// Permute 1 every ABCDEFGH -> CD AB GH EF
-// Permute 2 every ABCDEFGH -> EFGH ABCD
-// Permute 3 possible on longer iVector lengths (512bit = 8 double = 16 single)
-// Permute 4 possible on half precision @512bit vectors.
-//////////////////////////////////////////////////////////
-template<class vsimd>
-inline void Gpermute(vsimd &y,const vsimd &b,int perm){
-	union { 
-	  fvec f;
-	  decltype(vsimd::v) v;
-	} conv;
-	conv.v = b.v;
-      switch (perm){
-#if defined(AVX1)||defined(AVX2)
-      // 8x32 bits=>3 permutes
-      case 2: 
-	conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); 
-	break;
-      case 1: conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2)); break;
-      case 0: conv.f = _mm256_permute2f128_ps(conv.f,conv.f,0x01); break;
-#endif
-#ifdef SSE4
-      case 1: conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); break;
-      case 0: conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2));break;
-#endif
-#ifdef AVX512
-	// 16 floats=> permutes
-        // Permute 0 every abcd efgh ijkl mnop -> badc fehg jilk nmpo 
-        // Permute 1 every abcd efgh ijkl mnop -> cdab ghef jkij opmn 
-        // Permute 2 every abcd efgh ijkl mnop -> efgh abcd mnop ijkl
-        // Permute 3 every abcd efgh ijkl mnop -> ijkl mnop abcd efgh
-      case 3: conv.f = _mm512_swizzle_ps(conv.f,_MM_SWIZ_REG_CDAB); break;
-      case 2: conv.f = _mm512_swizzle_ps(conv.f,_MM_SWIZ_REG_BADC); break;
-      case 1: conv.f = _mm512_permute4f128_ps(conv.f,(_MM_PERM_ENUM)_MM_SHUFFLE(2,3,0,1)); break;
-      case 0: conv.f = _mm512_permute4f128_ps(conv.f,(_MM_PERM_ENUM)_MM_SHUFFLE(1,0,3,2)); break;
-#endif
-#ifdef QPX
-#error not implemented
-#endif
-      default: assert(0); break;
-      }
-      y.v=conv.v;
-
-    };
 };
 
 #include <simd/Grid_vector_types.h>
 
-
 namespace Grid {
 
   // NB: Template the following on "type Complex" and then implement *,+,- for 
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index a77cf963..3664e0f7 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -2,7 +2,7 @@
 /*! @file Grid_vector_types.h
   @brief Defines templated class Grid_simd to deal with inner vector types
 */
-// Time-stamp: <2015-05-26 12:05:39 neo>
+// Time-stamp: <2015-05-26 13:22:36 neo>
 //---------------------------------------------------------------------------
 #ifndef GRID_VECTOR_TYPES
 #define GRID_VECTOR_TYPES
@@ -16,7 +16,9 @@
 #if defined AVX512
 #include "Grid_knc.h"
 #endif
-
+#if defined QPX
+#include "Grid_qpx.h"
+#endif
 
 namespace Grid {
 
@@ -33,8 +35,6 @@ namespace Grid {
   template <typename Condition, typename ReturnType> using EnableIf   =    Invoke<std::enable_if<Condition::value, ReturnType>>;
   template <typename Condition, typename ReturnType> using NotEnableIf=    Invoke<std::enable_if<!Condition::value, ReturnType>>;
   
-
-
   ////////////////////////////////////////////////////////
   // Check for complexity with type traits
   template <typename T>     struct is_complex : std::false_type {};
@@ -57,6 +57,58 @@ namespace Grid {
 
   ///////////////////////////////////////////////
 
+//////////////////////////////////////////////////////////
+// Permute
+// Permute 0 every ABCDEFGH -> BA DC FE HG
+// Permute 1 every ABCDEFGH -> CD AB GH EF
+// Permute 2 every ABCDEFGH -> EFGH ABCD
+// Permute 3 possible on longer iVector lengths (512bit = 8 double = 16 single)
+// Permute 4 possible on half precision @512bit vectors.
+//////////////////////////////////////////////////////////
+template<class vsimd>
+inline void Gpermute(vsimd &y,const vsimd &b,int perm){
+	union { 
+	  SIMD_Ftype f;
+	  decltype(vsimd::v) v;
+	} conv;
+	conv.v = b.v;
+      switch (perm){
+#if defined(AVX1)||defined(AVX2)
+      // 8x32 bits=>3 permutes
+      case 2: 
+	conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); 
+	break;
+      case 1: conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2)); break;
+      case 0: conv.f = _mm256_permute2f128_ps(conv.f,conv.f,0x01); break;
+#endif
+#ifdef SSE4
+      case 1: conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); break;
+      case 0: conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2));break;
+#endif
+#ifdef AVX512
+	// 16 floats=> permutes
+        // Permute 0 every abcd efgh ijkl mnop -> badc fehg jilk nmpo 
+        // Permute 1 every abcd efgh ijkl mnop -> cdab ghef jkij opmn 
+        // Permute 2 every abcd efgh ijkl mnop -> efgh abcd mnop ijkl
+        // Permute 3 every abcd efgh ijkl mnop -> ijkl mnop abcd efgh
+      case 3: conv.f = _mm512_swizzle_ps(conv.f,_MM_SWIZ_REG_CDAB); break;
+      case 2: conv.f = _mm512_swizzle_ps(conv.f,_MM_SWIZ_REG_BADC); break;
+      case 1: conv.f = _mm512_permute4f128_ps(conv.f,(_MM_PERM_ENUM)_MM_SHUFFLE(2,3,0,1)); break;
+      case 0: conv.f = _mm512_permute4f128_ps(conv.f,(_MM_PERM_ENUM)_MM_SHUFFLE(1,0,3,2)); break;
+#endif
+#ifdef QPX
+#error not implemented
+#endif
+      default: assert(0); break;
+      }
+      y.v=conv.v;
+
+ };
+
+///////////////////////////////////////
+
+
+
   /*
     @brief Grid_simd class for the SIMD vector type operations
    */
@@ -380,6 +432,12 @@ namespace Grid {
   typedef Grid_simd< std::complex< double >, SIMD_Dtype > vComplexD;
   typedef Grid_simd< Integer               , SIMD_Itype > vInteger;
 
+
+
+
+
+
+
 }
 
 #endif

From fb5d72973e4e642487b0fa1c8919cd4a674ce0ef Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Tue, 26 May 2015 13:54:34 +0900
Subject: [PATCH 235/429] More cleanup of Grid_simd.h

---
 lib/Grid_simd.h              | 128 ++++++++++++-----------------------
 lib/simd/Grid_vector_types.h |  14 +++-
 2 files changed, 57 insertions(+), 85 deletions(-)

diff --git a/lib/Grid_simd.h b/lib/Grid_simd.h
index 39eb4654..cccc82e0 100644
--- a/lib/Grid_simd.h
+++ b/lib/Grid_simd.h
@@ -44,49 +44,49 @@ namespace Grid {
   inline ComplexF innerProduct(const ComplexF & l, const ComplexF & r) { return conjugate(l)*r; }
   inline RealD innerProduct(const RealD & l, const RealD & r) { return l*r; }
   inline RealF innerProduct(const RealF & l, const RealF & r) { return l*r; }
-
-    ////////////////////////////////////////////////////////////////////////////////
-    //Provide support functions for basic real and complex data types required by Grid
-    //Single and double precision versions. Should be able to template this once only.
-    ////////////////////////////////////////////////////////////////////////////////
-    inline void mac (ComplexD * __restrict__ y,const ComplexD * __restrict__ a,const ComplexD *__restrict__ x){ *y = (*a) * (*x)+(*y); };
-    inline void mult(ComplexD * __restrict__ y,const ComplexD * __restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) * (*r);}
-    inline void sub (ComplexD * __restrict__ y,const ComplexD * __restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) - (*r);}
-    inline void add (ComplexD * __restrict__ y,const ComplexD * __restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) + (*r);}
-    // conjugate already supported for complex
-    
-    inline void mac (ComplexF * __restrict__ y,const ComplexF * __restrict__ a,const ComplexF *__restrict__ x){ *y = (*a) * (*x)+(*y); }
-    inline void mult(ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) * (*r); }
-    inline void sub (ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) - (*r); }
-    inline void add (ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) + (*r); }
-
-    //conjugate already supported for complex
-
-    inline ComplexF timesI(const ComplexF &r)     { return(r*ComplexF(0.0,1.0));}
-    inline ComplexD timesI(const ComplexD &r)     { return(r*ComplexD(0.0,1.0));}
-    inline ComplexF timesMinusI(const ComplexF &r){ return(r*ComplexF(0.0,-1.0));}
-    inline ComplexD timesMinusI(const ComplexD &r){ return(r*ComplexD(0.0,-1.0));}
-    inline void timesI(ComplexF &ret,const ComplexF &r)     { ret = timesI(r);}
-    inline void timesI(ComplexD &ret,const ComplexD &r)     { ret = timesI(r);}
-    inline void timesMinusI(ComplexF &ret,const ComplexF &r){ ret = timesMinusI(r);}
-    inline void timesMinusI(ComplexD &ret,const ComplexD &r){ ret = timesMinusI(r);}
-
-    inline void mac (RealD * __restrict__ y,const RealD * __restrict__ a,const RealD *__restrict__ x){  *y = (*a) * (*x)+(*y);}
-    inline void mult(RealD * __restrict__ y,const RealD * __restrict__ l,const RealD *__restrict__ r){ *y = (*l) * (*r);}
-    inline void sub (RealD * __restrict__ y,const RealD * __restrict__ l,const RealD *__restrict__ r){ *y = (*l) - (*r);}
-    inline void add (RealD * __restrict__ y,const RealD * __restrict__ l,const RealD *__restrict__ r){ *y = (*l) + (*r);}
-    
-    inline void mac (RealF * __restrict__ y,const RealF * __restrict__ a,const RealF *__restrict__ x){  *y = (*a) * (*x)+(*y); }
-    inline void mult(RealF * __restrict__ y,const RealF * __restrict__ l,const RealF *__restrict__ r){ *y = (*l) * (*r); }
-    inline void sub (RealF * __restrict__ y,const RealF * __restrict__ l,const RealF *__restrict__ r){ *y = (*l) - (*r); }
-    inline void add (RealF * __restrict__ y,const RealF * __restrict__ l,const RealF *__restrict__ r){ *y = (*l) + (*r); }
-    
-    inline void vstream(ComplexF &l, const ComplexF &r){ l=r;}
-    inline void vstream(ComplexD &l, const ComplexD &r){ l=r;}
-    inline void vstream(RealF &l, const RealF &r){ l=r;}
-    inline void vstream(RealD &l, const RealD &r){ l=r;}
-
-
+  
+  ////////////////////////////////////////////////////////////////////////////////
+  //Provide support functions for basic real and complex data types required by Grid
+  //Single and double precision versions. Should be able to template this once only.
+  ////////////////////////////////////////////////////////////////////////////////
+  inline void mac (ComplexD * __restrict__ y,const ComplexD * __restrict__ a,const ComplexD *__restrict__ x){ *y = (*a) * (*x)+(*y); };
+  inline void mult(ComplexD * __restrict__ y,const ComplexD * __restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) * (*r);}
+  inline void sub (ComplexD * __restrict__ y,const ComplexD * __restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) - (*r);}
+  inline void add (ComplexD * __restrict__ y,const ComplexD * __restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) + (*r);}
+  // conjugate already supported for complex
+  
+  inline void mac (ComplexF * __restrict__ y,const ComplexF * __restrict__ a,const ComplexF *__restrict__ x){ *y = (*a) * (*x)+(*y); }
+  inline void mult(ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) * (*r); }
+  inline void sub (ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) - (*r); }
+  inline void add (ComplexF * __restrict__ y,const ComplexF * __restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) + (*r); }
+  
+  //conjugate already supported for complex
+  
+  inline ComplexF timesI(const ComplexF &r)     { return(r*ComplexF(0.0,1.0));}
+  inline ComplexD timesI(const ComplexD &r)     { return(r*ComplexD(0.0,1.0));}
+  inline ComplexF timesMinusI(const ComplexF &r){ return(r*ComplexF(0.0,-1.0));}
+  inline ComplexD timesMinusI(const ComplexD &r){ return(r*ComplexD(0.0,-1.0));}
+  inline void timesI(ComplexF &ret,const ComplexF &r)     { ret = timesI(r);}
+  inline void timesI(ComplexD &ret,const ComplexD &r)     { ret = timesI(r);}
+  inline void timesMinusI(ComplexF &ret,const ComplexF &r){ ret = timesMinusI(r);}
+  inline void timesMinusI(ComplexD &ret,const ComplexD &r){ ret = timesMinusI(r);}
+  
+  inline void mac (RealD * __restrict__ y,const RealD * __restrict__ a,const RealD *__restrict__ x){  *y = (*a) * (*x)+(*y);}
+  inline void mult(RealD * __restrict__ y,const RealD * __restrict__ l,const RealD *__restrict__ r){ *y = (*l) * (*r);}
+  inline void sub (RealD * __restrict__ y,const RealD * __restrict__ l,const RealD *__restrict__ r){ *y = (*l) - (*r);}
+  inline void add (RealD * __restrict__ y,const RealD * __restrict__ l,const RealD *__restrict__ r){ *y = (*l) + (*r);}
+  
+  inline void mac (RealF * __restrict__ y,const RealF * __restrict__ a,const RealF *__restrict__ x){  *y = (*a) * (*x)+(*y); }
+  inline void mult(RealF * __restrict__ y,const RealF * __restrict__ l,const RealF *__restrict__ r){ *y = (*l) * (*r); }
+  inline void sub (RealF * __restrict__ y,const RealF * __restrict__ l,const RealF *__restrict__ r){ *y = (*l) - (*r); }
+  inline void add (RealF * __restrict__ y,const RealF * __restrict__ l,const RealF *__restrict__ r){ *y = (*l) + (*r); }
+  
+  inline void vstream(ComplexF &l, const ComplexF &r){ l=r;}
+  inline void vstream(ComplexD &l, const ComplexD &r){ l=r;}
+  inline void vstream(RealF &l, const RealF &r){ l=r;}
+  inline void vstream(RealD &l, const RealD &r){ l=r;}
+  
+  
   class Zero{};
   static Zero zero;
   template<class itype> inline void zeroit(itype &arg){ arg=zero;};
@@ -94,52 +94,12 @@ namespace Grid {
   template<>            inline void zeroit(ComplexD &arg){ arg=0; };
   template<>            inline void zeroit(RealF &arg){ arg=0; };
   template<>            inline void zeroit(RealD &arg){ arg=0; };
-
+  
 };
 
 #include <simd/Grid_vector_types.h>
 
 namespace Grid {
-
-  // NB: Template the following on "type Complex" and then implement *,+,- for 
-  // ComplexF, ComplexD, RealF, RealD above to
-  // get full generality of binops with scalars.
-   inline void mac (vComplexF *__restrict__ y,const ComplexF *__restrict__ a,const vComplexF *__restrict__ x){ *y = (*a)*(*x)+(*y); };
-   inline void mult(vComplexF *__restrict__ y,const ComplexF *__restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) * (*r); }
-   inline void sub (vComplexF *__restrict__ y,const ComplexF *__restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) - (*r); }
-   inline void add (vComplexF *__restrict__ y,const ComplexF *__restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) + (*r); }
-   inline void mac (vComplexF *__restrict__ y,const vComplexF *__restrict__ a,const ComplexF *__restrict__ x){ *y = (*a)*(*x)+(*y); };
-   inline void mult(vComplexF *__restrict__ y,const vComplexF *__restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) * (*r); }
-   inline void sub (vComplexF *__restrict__ y,const vComplexF *__restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) - (*r); }
-   inline void add (vComplexF *__restrict__ y,const vComplexF *__restrict__ l,const ComplexF *__restrict__ r){ *y = (*l) + (*r); }
-
-   inline void mac (vComplexD *__restrict__ y,const ComplexD *__restrict__ a,const vComplexD *__restrict__ x){ *y = (*a)*(*x)+(*y); };
-   inline void mult(vComplexD *__restrict__ y,const ComplexD *__restrict__ l,const vComplexD *__restrict__ r){ *y = (*l) * (*r); }
-   inline void sub (vComplexD *__restrict__ y,const ComplexD *__restrict__ l,const vComplexD *__restrict__ r){ *y = (*l) - (*r); }
-   inline void add (vComplexD *__restrict__ y,const ComplexD *__restrict__ l,const vComplexD *__restrict__ r){ *y = (*l) + (*r); }
-   inline void mac (vComplexD *__restrict__ y,const vComplexD *__restrict__ a,const ComplexD *__restrict__ x){ *y = (*a)*(*x)+(*y); };
-   inline void mult(vComplexD *__restrict__ y,const vComplexD *__restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) * (*r); }
-   inline void sub (vComplexD *__restrict__ y,const vComplexD *__restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) - (*r); }
-   inline void add (vComplexD *__restrict__ y,const vComplexD *__restrict__ l,const ComplexD *__restrict__ r){ *y = (*l) + (*r); }
-
-   inline void mac (vRealF *__restrict__ y,const RealF *__restrict__ a,const vRealF *__restrict__ x){ *y = (*a)*(*x)+(*y); };
-   inline void mult(vRealF *__restrict__ y,const RealF *__restrict__ l,const vRealF *__restrict__ r){ *y = (*l) * (*r); }
-   inline void sub (vRealF *__restrict__ y,const RealF *__restrict__ l,const vRealF *__restrict__ r){ *y = (*l) - (*r); }
-   inline void add (vRealF *__restrict__ y,const RealF *__restrict__ l,const vRealF *__restrict__ r){ *y = (*l) + (*r); }
-   inline void mac (vRealF *__restrict__ y,const vRealF *__restrict__ a,const RealF *__restrict__ x){ *y = (*a)*(*x)+(*y); };
-   inline void mult(vRealF *__restrict__ y,const vRealF *__restrict__ l,const RealF *__restrict__ r){ *y = (*l) * (*r); }
-   inline void sub (vRealF *__restrict__ y,const vRealF *__restrict__ l,const RealF *__restrict__ r){ *y = (*l) - (*r); }
-   inline void add (vRealF *__restrict__ y,const vRealF *__restrict__ l,const RealF *__restrict__ r){ *y = (*l) + (*r); }
-
-   inline void mac (vRealD *__restrict__ y,const RealD *__restrict__ a,const vRealD *__restrict__ x){ *y = (*a)*(*x)+(*y); };
-   inline void mult(vRealD *__restrict__ y,const RealD *__restrict__ l,const vRealD *__restrict__ r){ *y = (*l) * (*r); }
-   inline void sub (vRealD *__restrict__ y,const RealD *__restrict__ l,const vRealD *__restrict__ r){ *y = (*l) - (*r); }
-   inline void add (vRealD *__restrict__ y,const RealD *__restrict__ l,const vRealD *__restrict__ r){ *y = (*l) + (*r); }
-   inline void mac (vRealD *__restrict__ y,const vRealD *__restrict__ a,const RealD *__restrict__ x){ *y = (*a)*(*x)+(*y); };
-   inline void mult(vRealD *__restrict__ y,const vRealD *__restrict__ l,const RealD *__restrict__ r){ *y = (*l) * (*r); }
-   inline void sub (vRealD *__restrict__ y,const vRealD *__restrict__ l,const RealD *__restrict__ r){ *y = (*l) - (*r); }
-   inline void add (vRealD *__restrict__ y,const vRealD *__restrict__ l,const RealD *__restrict__ r){ *y = (*l) + (*r); }
-
   // Default precision
 #ifdef GRID_DEFAULT_PRECISION_DOUBLE
   typedef vRealD vReal;
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index 3664e0f7..ae01269f 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -2,7 +2,7 @@
 /*! @file Grid_vector_types.h
   @brief Defines templated class Grid_simd to deal with inner vector types
 */
-// Time-stamp: <2015-05-26 13:22:36 neo>
+// Time-stamp: <2015-05-26 13:44:54 neo>
 //---------------------------------------------------------------------------
 #ifndef GRID_VECTOR_TYPES
 #define GRID_VECTOR_TYPES
@@ -156,6 +156,18 @@ inline void Gpermute(vsimd &y,const vsimd &b,int perm){
     friend inline void sub (Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) - (*r); }
     friend inline void add (Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) + (*r); }
 
+
+    friend inline void mac (Grid_simd *__restrict__ y,const Scalar_type *__restrict__ a,const Grid_simd   *__restrict__ x){ *y = (*a)*(*x)+(*y); };
+    friend inline void mult(Grid_simd *__restrict__ y,const Scalar_type *__restrict__ l,const Grid_simd   *__restrict__ r){ *y = (*l) * (*r); }
+    friend inline void sub (Grid_simd *__restrict__ y,const Scalar_type *__restrict__ l,const Grid_simd   *__restrict__ r){ *y = (*l) - (*r); }
+    friend inline void add (Grid_simd *__restrict__ y,const Scalar_type *__restrict__ l,const Grid_simd   *__restrict__ r){ *y = (*l) + (*r); }
+    friend inline void mac (Grid_simd *__restrict__ y,const Grid_simd   *__restrict__ a,const Scalar_type *__restrict__ x){ *y = (*a)*(*x)+(*y); };
+    friend inline void mult(Grid_simd *__restrict__ y,const Grid_simd   *__restrict__ l,const Scalar_type *__restrict__ r){ *y = (*l) * (*r); }
+    friend inline void sub (Grid_simd *__restrict__ y,const Grid_simd   *__restrict__ l,const Scalar_type *__restrict__ r){ *y = (*l) - (*r); }
+    friend inline void add (Grid_simd *__restrict__ y,const Grid_simd   *__restrict__ l,const Scalar_type *__restrict__ r){ *y = (*l) + (*r); }
+
+
+
     //not for integer types... 
     template <  class S = Scalar_type, NotEnableIf<std::is_integral < S >, int> = 0 > 
     friend inline Grid_simd adj(const Grid_simd &in){ return conjugate(in); }

From a32ac287bb1eccc88088de51d5fc483f5f8e2f72 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 26 May 2015 19:54:03 +0100
Subject: [PATCH 236/429] Hand unrolled version of dslash in a separate class.
 Useful to compare; raises Intel compiler from 9GFlop/s to 17.5 Gflops.       
             on ivybridge core. Raises Clang form 14.5 to 17.5

---
 benchmarks/Grid_wilson.cc              |  14 +-
 lib/Grid.h                             |   1 -
 lib/Grid_init.cc                       |   3 +
 lib/Grid_simd.h                        | 100 +++++--
 lib/Makefile.am                        |   2 +
 lib/lattice/Grid_lattice_base.h        |  12 +-
 lib/lattice/Grid_lattice_conformable.h |   7 +-
 lib/qcd/Grid_qcd_wilson_dop.cc         | 343 ++-----------------------
 lib/qcd/Grid_qcd_wilson_dop.h          |  45 +++-
 9 files changed, 157 insertions(+), 370 deletions(-)

diff --git a/benchmarks/Grid_wilson.cc b/benchmarks/Grid_wilson.cc
index 32255b3e..3b0d04bc 100644
--- a/benchmarks/Grid_wilson.cc
+++ b/benchmarks/Grid_wilson.cc
@@ -31,11 +31,9 @@ int main (int argc, char ** argv)
   std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
 
   std::vector<int> seeds({1,2,3,4});
-
   GridParallelRNG          pRNG(&Grid);
-  //  std::vector<int> seeds({1,2,3,4});
-  //  pRNG.SeedFixedIntegers(seeds);
-  pRNG.SeedRandomDevice();
+  pRNG.SeedFixedIntegers(seeds);
+  //  pRNG.SeedRandomDevice();
 
   LatticeFermion src   (&Grid); random(pRNG,src);
   LatticeFermion result(&Grid); result=zero;
@@ -55,8 +53,10 @@ int main (int argc, char ** argv)
   Complex cone(1.0,0.0);
   for(int nn=0;nn<Nd;nn++){
     random(pRNG,U[nn]);
-    if (nn!=0) U[nn]=zero;
-    else U[nn] = cone;
+    if(0) {
+      if (nn==-1) { U[nn]=zero; std::cout << "zeroing gauge field in dir "<<nn<<std::endl; }
+      else       { U[nn] = cone;std::cout << "unit gauge field in dir "<<nn<<std::endl; }
+    }
     pokeIndex<LorentzIndex>(Umu,U[nn],nn);
   }
 
@@ -85,7 +85,7 @@ int main (int argc, char ** argv)
   WilsonMatrix Dw(Umu,Grid,RBGrid,mass);
   
   std::cout << "Calling Dw"<<std::endl;
-  int ncall=1000;
+  int ncall=10000;
   double t0=usecond();
   for(int i=0;i<ncall;i++){
     Dw.Dhop(src,result,0);
diff --git a/lib/Grid.h b/lib/Grid.h
index a1ad74c4..1e513a46 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -65,7 +65,6 @@
 
 namespace Grid {
 
-
   void Grid_init(int *argc,char ***argv);
   void Grid_finalize(void);
   // internal, controled with --handle
diff --git a/lib/Grid_init.cc b/lib/Grid_init.cc
index 7ff698c6..7fce4793 100644
--- a/lib/Grid_init.cc
+++ b/lib/Grid_init.cc
@@ -142,6 +142,9 @@ void Grid_init(int *argc,char ***argv)
   if( !GridCmdOptionExists(*argv,*argv+*argc,"--debug-stdout") ){
     Grid_quiesce_nodes();
   }
+  if( GridCmdOptionExists(*argv,*argv+*argc,"--dslash-opt") ){
+    WilsonMatrix::HandOptDslash=1;
+  }
   GridParseLayout(*argv,*argc,
 		  Grid_default_latt,
 		  Grid_default_mpi);
diff --git a/lib/Grid_simd.h b/lib/Grid_simd.h
index 6d7411d3..785c9539 100644
--- a/lib/Grid_simd.h
+++ b/lib/Grid_simd.h
@@ -162,6 +162,74 @@ namespace Grid {
 // Permute 3 possible on longer iVector lengths (512bit = 8 double = 16 single)
 // Permute 4 possible on half precision @512bit vectors.
 //////////////////////////////////////////////////////////
+template<class vsimd>
+inline void Gpermute0(vsimd &y,const vsimd &b) {
+  union { 
+    fvec f;
+    decltype(vsimd::v) v;
+  } conv;
+  conv.v = b.v;
+#ifdef SSE4
+  conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2));
+#endif
+#if defined(AVX1)||defined(AVX2)
+  conv.f = _mm256_permute2f128_ps(conv.f,conv.f,0x01); 
+#endif  
+#ifdef AVX512
+ conv.f = _mm512_permute4f128_ps(conv.f,(_MM_PERM_ENUM)_MM_SHUFFLE(1,0,3,2)); 
+#endif
+ y.v=conv.v;
+};
+template<class vsimd>
+inline void Gpermute1(vsimd &y,const vsimd &b) {
+  union { 
+    fvec f;
+    decltype(vsimd::v) v;
+  } conv;
+  conv.v = b.v;
+#ifdef SSE4
+  conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); 
+#endif
+#if defined(AVX1)||defined(AVX2)
+  conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2)); 
+#endif  
+#ifdef AVX512
+ conv.f = _mm512_permute4f128_ps(conv.f,(_MM_PERM_ENUM)_MM_SHUFFLE(2,3,0,1)); 
+#endif  
+  y.v=conv.v;
+};
+template<class vsimd>
+inline void Gpermute2(vsimd &y,const vsimd &b) {
+  union { 
+    fvec f;
+    decltype(vsimd::v) v;
+  } conv;
+  conv.v = b.v;
+#ifdef SSE4
+#endif
+#if defined(AVX1)||defined(AVX2)
+  conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); 
+#endif  
+#ifdef AVX512
+  conv.f = _mm512_swizzle_ps(conv.f,_MM_SWIZ_REG_BADC); 
+#endif  
+  y.v=conv.v;
+
+};
+template<class vsimd>
+inline void Gpermute3(vsimd &y,const vsimd &b) {
+  union { 
+    fvec f;
+    decltype(vsimd::v) v;
+  } conv;
+  conv.v = b.v;
+#ifdef AVX512
+  conv.f = _mm512_swizzle_ps(conv.f,_MM_SWIZ_REG_CDAB); 
+#endif  
+  y.v=conv.v;
+
+};
+
 template<class vsimd>
 inline void Gpermute(vsimd &y,const vsimd &b,int perm){
 	union { 
@@ -170,36 +238,12 @@ inline void Gpermute(vsimd &y,const vsimd &b,int perm){
 	} conv;
 	conv.v = b.v;
       switch (perm){
-#if defined(AVX1)||defined(AVX2)
-      // 8x32 bits=>3 permutes
-      case 2: 
-	conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); 
-	break;
-      case 1: conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2)); break;
-      case 0: conv.f = _mm256_permute2f128_ps(conv.f,conv.f,0x01); break;
-#endif
-#ifdef SSE4
-      case 1: conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); break;
-      case 0: conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2));break;
-#endif
-#ifdef AVX512
-	// 16 floats=> permutes
-        // Permute 0 every abcd efgh ijkl mnop -> badc fehg jilk nmpo 
-        // Permute 1 every abcd efgh ijkl mnop -> cdab ghef jkij opmn 
-        // Permute 2 every abcd efgh ijkl mnop -> efgh abcd mnop ijkl
-        // Permute 3 every abcd efgh ijkl mnop -> ijkl mnop abcd efgh
-      case 3: conv.f = _mm512_swizzle_ps(conv.f,_MM_SWIZ_REG_CDAB); break;
-      case 2: conv.f = _mm512_swizzle_ps(conv.f,_MM_SWIZ_REG_BADC); break;
-      case 1: conv.f = _mm512_permute4f128_ps(conv.f,(_MM_PERM_ENUM)_MM_SHUFFLE(2,3,0,1)); break;
-      case 0: conv.f = _mm512_permute4f128_ps(conv.f,(_MM_PERM_ENUM)_MM_SHUFFLE(1,0,3,2)); break;
-#endif
-#ifdef QPX
-#error not implemented
-#endif
+      case 3: Gpermute3(y,b); break;
+      case 2: Gpermute2(y,b); break;
+      case 1: Gpermute1(y,b); break;
+      case 0: Gpermute0(y,b); break;
       default: assert(0); break;
       }
-      y.v=conv.v;
-
     };
 };
 
diff --git a/lib/Makefile.am b/lib/Makefile.am
index 82459763..6bb5e187 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -18,6 +18,8 @@ libGrid_a_SOURCES =				\
 	Grid_init.cc				\
 	stencil/Grid_stencil_common.cc		\
 	qcd/Grid_qcd_dirac.cc			\
+	qcd/Grid_qcd_dhop.cc			\
+	qcd/Grid_qcd_dhop_hand.cc		\
 	qcd/Grid_qcd_wilson_dop.cc		\
 	algorithms/approx/Zolotarev.cc		\
 	algorithms/approx/Remez.cc		\
diff --git a/lib/lattice/Grid_lattice_base.h b/lib/lattice/Grid_lattice_base.h
index 1d3b1efb..4a6d3180 100644
--- a/lib/lattice/Grid_lattice_base.h
+++ b/lib/lattice/Grid_lattice_base.h
@@ -47,6 +47,11 @@ class LatticeTrinaryExpression :public std::pair<Op,std::tuple<T1,T2,T3> >, publ
  LatticeTrinaryExpression(const std::pair<Op,std::tuple<T1,T2,T3> > &arg): std::pair<Op,std::tuple<T1,T2,T3> >(arg) {};
 };
 
+void inline conformable(GridBase *lhs,GridBase *rhs)
+{
+  assert(lhs == rhs);
+}
+
 template<class vobj>
 class Lattice : public LatticeBase
 {
@@ -60,7 +65,8 @@ public:
     typedef typename vobj::scalar_type scalar_type;
     typedef typename vobj::vector_type vector_type;
     typedef vobj vector_object;
-    
+ 
+   
   ////////////////////////////////////////////////////////////////////////////////
   // Expression Template closure support
   ////////////////////////////////////////////////////////////////////////////////
@@ -276,17 +282,15 @@ PARALLEL_FOR_LOOP
 }
 
 
-#include <lattice/Grid_lattice_conformable.h>
 
+#include <lattice/Grid_lattice_conformable.h>
 #define GRID_LATTICE_EXPRESSION_TEMPLATES
 #ifdef  GRID_LATTICE_EXPRESSION_TEMPLATES
 #include <lattice/Grid_lattice_ET.h>
 #else 
 #include <lattice/Grid_lattice_overload.h>
 #endif
-
 #include <lattice/Grid_lattice_arith.h>
-
 #include <lattice/Grid_lattice_trace.h>
 #include <lattice/Grid_lattice_transpose.h>
 #include <lattice/Grid_lattice_local.h>
diff --git a/lib/lattice/Grid_lattice_conformable.h b/lib/lattice/Grid_lattice_conformable.h
index faa8c7a7..a77e57af 100644
--- a/lib/lattice/Grid_lattice_conformable.h
+++ b/lib/lattice/Grid_lattice_conformable.h
@@ -3,16 +3,11 @@
 
 namespace Grid {
 
-    template<class obj1,class obj2>
-    void conformable(const Lattice<obj1> &lhs,const Lattice<obj2> &rhs)
+    template<class obj1,class obj2> void conformable(const Lattice<obj1> &lhs,const Lattice<obj2> &rhs)
     {
         assert(lhs._grid == rhs._grid);
         assert(lhs.checkerboard == rhs.checkerboard);
     }
-    void inline conformable(const GridBase *lhs,GridBase *rhs)
-    {
-        assert(lhs == rhs);
-    }
 
 }
 #endif
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
index 318e18df..9a3f5f6a 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ b/lib/qcd/Grid_qcd_wilson_dop.cc
@@ -1,4 +1,3 @@
-
 #include <Grid.h>
 
 namespace Grid {
@@ -7,15 +6,7 @@ namespace QCD {
 const std::vector<int> WilsonMatrix::directions   ({0,1,2,3, 0, 1, 2, 3});
 const std::vector<int> WilsonMatrix::displacements({1,1,1,1,-1,-1,-1,-1});
 
-  // Should be in header?
-const int WilsonMatrix::Xp = 0;
-const int WilsonMatrix::Yp = 1;
-const int WilsonMatrix::Zp = 2;
-const int WilsonMatrix::Tp = 3;
-const int WilsonMatrix::Xm = 4;
-const int WilsonMatrix::Ym = 5;
-const int WilsonMatrix::Zm = 6;
-const int WilsonMatrix::Tm = 7;
+  int WilsonMatrix::HandOptDslash;
 
   class WilsonCompressor {
   public:
@@ -39,28 +30,28 @@ const int WilsonMatrix::Tm = 7;
 	mudag=(mu+Nd)%(2*Nd);
       }
       switch(mudag) {
-      case WilsonMatrix::Xp:
+      case Xp:
 	spProjXp(ret,in);
 	break;
-      case WilsonMatrix::Yp:
+      case Yp:
 	spProjYp(ret,in);
 	break;
-      case WilsonMatrix::Zp:
+      case Zp:
 	spProjZp(ret,in);
 	break;
-      case WilsonMatrix::Tp:
+      case Tp:
 	spProjTp(ret,in);
 	break;
-      case WilsonMatrix::Xm:
+      case Xm:
 	spProjXm(ret,in);
 	break;
-      case WilsonMatrix::Ym:
+      case Ym:
 	spProjYm(ret,in);
 	break;
-      case WilsonMatrix::Zm:
+      case Zm:
 	spProjZm(ret,in);
 	break;
-      case WilsonMatrix::Tm:
+      case Tm:
 	spProjTm(ret,in);
 	break;
       default: 
@@ -157,316 +148,36 @@ void WilsonMatrix::MooeeInvDag(const LatticeFermion &in, LatticeFermion &out)
   MooeeInv(in,out);
 }
 
-void WilsonMatrix::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
-			    std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			    int ss,const LatticeFermion &in, LatticeFermion &out)
-{
-    vHalfSpinColourVector  tmp;    
-    vHalfSpinColourVector  chi;    
-    vSpinColourVector result;
-    vHalfSpinColourVector Uchi;
-    int offset,local,perm, ptype;
-
-    //#define VERBOSE( A)  if ( ss<10 ) { std::cout << "site " <<ss << " " #A " neigh " << offset << " perm "<< perm <<std::endl;}    
-
-    // Xp
-    offset = st._offsets [Xp][ss];
-    local  = st._is_local[Xp][ss];
-    perm   = st._permute[Xp][ss];
-
-    ptype  = st._permute_type[Xp];
-    if ( local && perm ) {
-      spProjXp(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjXp(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[ss](Xp),&chi());
-    spReconXp(result,Uchi);
-
-    // Yp
-    offset = st._offsets [Yp][ss];
-    local  = st._is_local[Yp][ss];
-    perm   = st._permute[Yp][ss];
-    ptype  = st._permute_type[Yp];
-    if ( local && perm ) {
-      spProjYp(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjYp(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[ss](Yp),&chi());
-    accumReconYp(result,Uchi);
-
-    // Zp
-    offset = st._offsets [Zp][ss];
-    local  = st._is_local[Zp][ss];
-    perm   = st._permute[Zp][ss];
-    ptype  = st._permute_type[Zp];
-    if ( local && perm ) {
-      spProjZp(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjZp(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[ss](Zp),&chi());
-    accumReconZp(result,Uchi);
-
-    // Tp
-    offset = st._offsets [Tp][ss];
-    local  = st._is_local[Tp][ss];
-    perm   = st._permute[Tp][ss];
-    ptype  = st._permute_type[Tp];
-    if ( local && perm ) {
-      spProjTp(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjTp(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[ss](Tp),&chi());
-    accumReconTp(result,Uchi);
-
-    // Xm
-    offset = st._offsets [Xm][ss];
-    local  = st._is_local[Xm][ss];
-    perm   = st._permute[Xm][ss];
-    ptype  = st._permute_type[Xm];
-
-    if ( local && perm ) 
-    {
-      spProjXm(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjXm(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[ss](Xm),&chi());
-    accumReconXm(result,Uchi);
-
-
-    // Ym
-    offset = st._offsets [Ym][ss];
-    local  = st._is_local[Ym][ss];
-    perm   = st._permute[Ym][ss];
-    ptype  = st._permute_type[Ym];
-
-    if ( local && perm ) {
-      spProjYm(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjYm(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[ss](Ym),&chi());
-    accumReconYm(result,Uchi);
-
-    // Zm
-    offset = st._offsets [Zm][ss];
-    local  = st._is_local[Zm][ss];
-    perm   = st._permute[Zm][ss];
-    ptype  = st._permute_type[Zm];
-    if ( local && perm ) {
-      spProjZm(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjZm(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[ss](Zm),&chi());
-    accumReconZm(result,Uchi);
-
-    // Tm
-    offset = st._offsets [Tm][ss];
-    local  = st._is_local[Tm][ss];
-    perm   = st._permute[Tm][ss];
-    ptype  = st._permute_type[Tm];
-    if ( local && perm ) {
-      spProjTm(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjTm(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[ss](Tm),&chi());
-    accumReconTm(result,Uchi);
-
-    vstream(out._odata[ss],result);
-}
-
-void WilsonMatrix::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
-			       std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			       int ss,const LatticeFermion &in, LatticeFermion &out)
-{
-    vHalfSpinColourVector  tmp;    
-    vHalfSpinColourVector  chi;    
-    vSpinColourVector result;
-    vHalfSpinColourVector Uchi;
-    int offset,local,perm, ptype;
-
-    // Xp
-    offset = st._offsets [Xm][ss];
-    local  = st._is_local[Xm][ss];
-    perm   = st._permute[Xm][ss];
-
-    ptype  = st._permute_type[Xm];
-    if ( local && perm ) {
-      spProjXp(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjXp(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[ss](Xm),&chi());
-    spReconXp(result,Uchi);
-
-    // Yp
-    offset = st._offsets [Ym][ss];
-    local  = st._is_local[Ym][ss];
-    perm   = st._permute[Ym][ss];
-    ptype  = st._permute_type[Ym];
-    if ( local && perm ) {
-      spProjYp(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjYp(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[ss](Ym),&chi());
-    accumReconYp(result,Uchi);
-
-    // Zp
-    offset = st._offsets [Zm][ss];
-    local  = st._is_local[Zm][ss];
-    perm   = st._permute[Zm][ss];
-    ptype  = st._permute_type[Zm];
-    if ( local && perm ) {
-      spProjZp(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjZp(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[ss](Zm),&chi());
-    accumReconZp(result,Uchi);
-
-    // Tp
-    offset = st._offsets [Tm][ss];
-    local  = st._is_local[Tm][ss];
-    perm   = st._permute[Tm][ss];
-    ptype  = st._permute_type[Tm];
-    if ( local && perm ) {
-      spProjTp(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjTp(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[ss](Tm),&chi());
-    accumReconTp(result,Uchi);
-
-    // Xm
-    offset = st._offsets [Xp][ss];
-    local  = st._is_local[Xp][ss];
-    perm   = st._permute[Xp][ss];
-    ptype  = st._permute_type[Xp];
-
-    if ( local && perm ) 
-    {
-      spProjXm(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjXm(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[ss](Xp),&chi());
-    accumReconXm(result,Uchi);
-
-    // Ym
-    offset = st._offsets [Yp][ss];
-    local  = st._is_local[Yp][ss];
-    perm   = st._permute[Yp][ss];
-    ptype  = st._permute_type[Yp];
-
-    if ( local && perm ) {
-      spProjYm(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjYm(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[ss](Yp),&chi());
-    accumReconYm(result,Uchi);
-
-    // Zm
-    offset = st._offsets [Zp][ss];
-    local  = st._is_local[Zp][ss];
-    perm   = st._permute[Zp][ss];
-    ptype  = st._permute_type[Zp];
-    if ( local && perm ) {
-      spProjZm(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjZm(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[ss](Zp),&chi());
-    accumReconZm(result,Uchi);
-
-    // Tm
-    offset = st._offsets [Tp][ss];
-    local  = st._is_local[Tp][ss];
-    perm   = st._permute[Tp][ss];
-    ptype  = st._permute_type[Tp];
-    if ( local && perm ) {
-      spProjTm(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjTm(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[ss](Tp),&chi());
-    accumReconTm(result,Uchi);
-
-    vstream(out._odata[ss],result);
-}
-
 void WilsonMatrix::DhopInternal(CartesianStencil & st,LatticeDoubledGaugeField & U,
 				const LatticeFermion &in, LatticeFermion &out,int dag)
 {
   assert((dag==DaggerNo) ||(dag==DaggerYes));
   WilsonCompressor compressor(dag);
-
   st.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
 
   if ( dag == DaggerYes ) {
+    if( HandOptDslash ) {
 PARALLEL_FOR_LOOP
-    for(int sss=0;sss<in._grid->oSites();sss++){
-      DhopSiteDag(st,U,comm_buf,sss,in,out);
+      for(int sss=0;sss<in._grid->oSites();sss++){
+        DiracOptHand::DhopSiteDag(st,U,comm_buf,sss,in,out);
+      }
+    } else { 
+PARALLEL_FOR_LOOP
+      for(int sss=0;sss<in._grid->oSites();sss++){
+        DiracOpt::DhopSiteDag(st,U,comm_buf,sss,in,out);
+      }
     }
   } else {
+    if( HandOptDslash ) {
 PARALLEL_FOR_LOOP
-    for(int sss=0;sss<in._grid->oSites();sss++){
-      DhopSite(st,U,comm_buf,sss,in,out);
+      for(int sss=0;sss<in._grid->oSites();sss++){
+        DiracOptHand::DhopSite(st,U,comm_buf,sss,in,out);
+      }
+    } else { 
+PARALLEL_FOR_LOOP
+      for(int sss=0;sss<in._grid->oSites();sss++){
+        DiracOpt::DhopSite(st,U,comm_buf,sss,in,out);
+      }
     }
   }
 }
diff --git a/lib/qcd/Grid_qcd_wilson_dop.h b/lib/qcd/Grid_qcd_wilson_dop.h
index 96b29cd0..87418603 100644
--- a/lib/qcd/Grid_qcd_wilson_dop.h
+++ b/lib/qcd/Grid_qcd_wilson_dop.h
@@ -6,10 +6,22 @@ namespace Grid {
 
   namespace QCD {
 
+  // Should be in header?
+    const int Xp = 0;
+    const int Yp = 1;
+    const int Zp = 2;
+    const int Tp = 3;
+    const int Xm = 4;
+    const int Ym = 5;
+    const int Zm = 6;
+    const int Tm = 7;
+
     class WilsonMatrix : public CheckerBoardedSparseMatrixBase<LatticeFermion>
     {
       //NB r=1;
     public:
+      static int HandOptDslash;
+
       double                        mass;
       //      GridBase                     *    grid; // Inherited
       //      GridBase                     *  cbgrid;
@@ -56,14 +68,6 @@ namespace Grid {
       void DhopEO(const LatticeFermion &in, LatticeFermion &out,int dag);
       void DhopInternal(CartesianStencil & st,LatticeDoubledGaugeField &U,
 			const LatticeFermion &in, LatticeFermion &out,int dag);
-      // These ones will need to be package intelligently. WilsonType base class
-      // for use by DWF etc..
-      void DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
-		    std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-		    int ss,const LatticeFermion &in, LatticeFermion &out);
-      void DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
-		       std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-		       int ss,const LatticeFermion &in, LatticeFermion &out);
 
       typedef iScalar<iMatrix<vComplex, Nc> > matrix;
 
@@ -71,6 +75,31 @@ namespace Grid {
     };
 
 
+    class DiracOpt {
+    public:
+      // These ones will need to be package intelligently. WilsonType base class
+      // for use by DWF etc..
+      static void DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
+		    std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+		    int ss,const LatticeFermion &in, LatticeFermion &out);
+      static void DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
+		       std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+		       int ss,const LatticeFermion &in, LatticeFermion &out);
+
+    };
+    class DiracOptHand {
+    public:
+      // These ones will need to be package intelligently. WilsonType base class
+      // for use by DWF etc..
+      static void DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
+		    std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+		    int ss,const LatticeFermion &in, LatticeFermion &out);
+      static void DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
+		       std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+		       int ss,const LatticeFermion &in, LatticeFermion &out);
+
+    };
+
   }
 }
 #endif

From 5e72e4c0d9690bcf36951d4e29b14b134e80b2d8 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 26 May 2015 19:55:18 +0100
Subject: [PATCH 237/429] Strip out the dslash kernel implementation

---
 lib/qcd/Grid_qcd_dhop.cc      | 309 ++++++++++++++
 lib/qcd/Grid_qcd_dhop_hand.cc | 769 ++++++++++++++++++++++++++++++++++
 2 files changed, 1078 insertions(+)
 create mode 100644 lib/qcd/Grid_qcd_dhop.cc
 create mode 100644 lib/qcd/Grid_qcd_dhop_hand.cc

diff --git a/lib/qcd/Grid_qcd_dhop.cc b/lib/qcd/Grid_qcd_dhop.cc
new file mode 100644
index 00000000..1e5dcd16
--- /dev/null
+++ b/lib/qcd/Grid_qcd_dhop.cc
@@ -0,0 +1,309 @@
+#include <Grid.h>
+
+namespace Grid {
+namespace QCD {
+
+void DiracOpt::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
+			    std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+			    int ss,const LatticeFermion &in, LatticeFermion &out)
+{
+    vHalfSpinColourVector  tmp;    
+    vHalfSpinColourVector  chi;    
+    vSpinColourVector result;
+    vHalfSpinColourVector Uchi;
+    int offset,local,perm, ptype;
+
+    //#define VERBOSE( A)  if ( ss<10 ) { std::cout << "site " <<ss << " " #A " neigh " << offset << " perm "<< perm <<std::endl;}    
+
+    // Xp
+    offset = st._offsets [Xp][ss];
+    local  = st._is_local[Xp][ss];
+    perm   = st._permute[Xp][ss];
+
+    ptype  = st._permute_type[Xp];
+    if ( local && perm ) {
+      spProjXp(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjXp(chi,in._odata[offset]);
+    } else { 
+      chi=buf[offset];
+    }
+    mult(&Uchi(),&U._odata[ss](Xp),&chi());
+    spReconXp(result,Uchi);
+
+    //    std::cout << "XP_RECON"<<std::endl;
+    //    std::cout << result()(0)(0) <<" "<<result()(0)(1) <<" "<<result()(0)(2) <<std::endl;
+    //    std::cout << result()(1)(0) <<" "<<result()(1)(1) <<" "<<result()(1)(2) <<std::endl;
+    //    std::cout << result()(2)(0) <<" "<<result()(2)(1) <<" "<<result()(2)(2) <<std::endl;
+    //    std::cout << result()(3)(0) <<" "<<result()(3)(1) <<" "<<result()(3)(2) <<std::endl;
+
+    // Yp
+    offset = st._offsets [Yp][ss];
+    local  = st._is_local[Yp][ss];
+    perm   = st._permute[Yp][ss];
+    ptype  = st._permute_type[Yp];
+    if ( local && perm ) {
+      spProjYp(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjYp(chi,in._odata[offset]);
+    } else { 
+      chi=buf[offset];
+    }
+    mult(&Uchi(),&U._odata[ss](Yp),&chi());
+    accumReconYp(result,Uchi);
+
+    // Zp
+    offset = st._offsets [Zp][ss];
+    local  = st._is_local[Zp][ss];
+    perm   = st._permute[Zp][ss];
+    ptype  = st._permute_type[Zp];
+    if ( local && perm ) {
+      spProjZp(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjZp(chi,in._odata[offset]);
+    } else { 
+      chi=buf[offset];
+    }
+    mult(&Uchi(),&U._odata[ss](Zp),&chi());
+    accumReconZp(result,Uchi);
+
+    // Tp
+    offset = st._offsets [Tp][ss];
+    local  = st._is_local[Tp][ss];
+    perm   = st._permute[Tp][ss];
+    ptype  = st._permute_type[Tp];
+    if ( local && perm ) {
+      spProjTp(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjTp(chi,in._odata[offset]);
+    } else { 
+      chi=buf[offset];
+    }
+    mult(&Uchi(),&U._odata[ss](Tp),&chi());
+    accumReconTp(result,Uchi);
+
+    // Xm
+    offset = st._offsets [Xm][ss];
+    local  = st._is_local[Xm][ss];
+    perm   = st._permute[Xm][ss];
+    ptype  = st._permute_type[Xm];
+
+    if ( local && perm ) 
+    {
+      spProjXm(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjXm(chi,in._odata[offset]);
+    } else { 
+      chi=buf[offset];
+    }
+    mult(&Uchi(),&U._odata[ss](Xm),&chi());
+    accumReconXm(result,Uchi);
+    //  std::cout << "XM_RECON_ACCUM"<<std::endl;
+    //    std::cout << result()(0)(0) <<" "<<result()(0)(1) <<" "<<result()(0)(2) <<std::endl;
+    //    std::cout << result()(1)(0) <<" "<<result()(1)(1) <<" "<<result()(1)(2) <<std::endl;
+    //    std::cout << result()(2)(0) <<" "<<result()(2)(1) <<" "<<result()(2)(2) <<std::endl;
+    //    std::cout << result()(3)(0) <<" "<<result()(3)(1) <<" "<<result()(3)(2) <<std::endl;
+
+
+    // Ym
+    offset = st._offsets [Ym][ss];
+    local  = st._is_local[Ym][ss];
+    perm   = st._permute[Ym][ss];
+    ptype  = st._permute_type[Ym];
+
+    if ( local && perm ) {
+      spProjYm(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjYm(chi,in._odata[offset]);
+    } else { 
+      chi=buf[offset];
+    }
+    mult(&Uchi(),&U._odata[ss](Ym),&chi());
+    accumReconYm(result,Uchi);
+
+    // Zm
+    offset = st._offsets [Zm][ss];
+    local  = st._is_local[Zm][ss];
+    perm   = st._permute[Zm][ss];
+    ptype  = st._permute_type[Zm];
+    if ( local && perm ) {
+      spProjZm(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjZm(chi,in._odata[offset]);
+    } else { 
+      chi=buf[offset];
+    }
+    mult(&Uchi(),&U._odata[ss](Zm),&chi());
+    accumReconZm(result,Uchi);
+
+    // Tm
+    offset = st._offsets [Tm][ss];
+    local  = st._is_local[Tm][ss];
+    perm   = st._permute[Tm][ss];
+    ptype  = st._permute_type[Tm];
+    if ( local && perm ) {
+      spProjTm(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjTm(chi,in._odata[offset]);
+    } else { 
+      chi=buf[offset];
+    }
+    mult(&Uchi(),&U._odata[ss](Tm),&chi());
+    accumReconTm(result,Uchi);
+
+    vstream(out._odata[ss],result);
+}
+
+void DiracOpt::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
+			       std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+			       int ss,const LatticeFermion &in, LatticeFermion &out)
+{
+    vHalfSpinColourVector  tmp;    
+    vHalfSpinColourVector  chi;    
+    vSpinColourVector result;
+    vHalfSpinColourVector Uchi;
+    int offset,local,perm, ptype;
+
+    // Xp
+    offset = st._offsets [Xm][ss];
+    local  = st._is_local[Xm][ss];
+    perm   = st._permute[Xm][ss];
+
+    ptype  = st._permute_type[Xm];
+    if ( local && perm ) {
+      spProjXp(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjXp(chi,in._odata[offset]);
+    } else { 
+      chi=buf[offset];
+    }
+    mult(&Uchi(),&U._odata[ss](Xm),&chi());
+    spReconXp(result,Uchi);
+
+    // Yp
+    offset = st._offsets [Ym][ss];
+    local  = st._is_local[Ym][ss];
+    perm   = st._permute[Ym][ss];
+    ptype  = st._permute_type[Ym];
+    if ( local && perm ) {
+      spProjYp(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjYp(chi,in._odata[offset]);
+    } else { 
+      chi=buf[offset];
+    }
+    mult(&Uchi(),&U._odata[ss](Ym),&chi());
+    accumReconYp(result,Uchi);
+
+    // Zp
+    offset = st._offsets [Zm][ss];
+    local  = st._is_local[Zm][ss];
+    perm   = st._permute[Zm][ss];
+    ptype  = st._permute_type[Zm];
+    if ( local && perm ) {
+      spProjZp(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjZp(chi,in._odata[offset]);
+    } else { 
+      chi=buf[offset];
+    }
+    mult(&Uchi(),&U._odata[ss](Zm),&chi());
+    accumReconZp(result,Uchi);
+
+    // Tp
+    offset = st._offsets [Tm][ss];
+    local  = st._is_local[Tm][ss];
+    perm   = st._permute[Tm][ss];
+    ptype  = st._permute_type[Tm];
+    if ( local && perm ) {
+      spProjTp(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjTp(chi,in._odata[offset]);
+    } else { 
+      chi=buf[offset];
+    }
+    mult(&Uchi(),&U._odata[ss](Tm),&chi());
+    accumReconTp(result,Uchi);
+
+    // Xm
+    offset = st._offsets [Xp][ss];
+    local  = st._is_local[Xp][ss];
+    perm   = st._permute[Xp][ss];
+    ptype  = st._permute_type[Xp];
+
+    if ( local && perm ) 
+    {
+      spProjXm(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjXm(chi,in._odata[offset]);
+    } else { 
+      chi=buf[offset];
+    }
+    mult(&Uchi(),&U._odata[ss](Xp),&chi());
+    accumReconXm(result,Uchi);
+
+    // Ym
+    offset = st._offsets [Yp][ss];
+    local  = st._is_local[Yp][ss];
+    perm   = st._permute[Yp][ss];
+    ptype  = st._permute_type[Yp];
+
+    if ( local && perm ) {
+      spProjYm(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjYm(chi,in._odata[offset]);
+    } else { 
+      chi=buf[offset];
+    }
+    mult(&Uchi(),&U._odata[ss](Yp),&chi());
+    accumReconYm(result,Uchi);
+
+    // Zm
+    offset = st._offsets [Zp][ss];
+    local  = st._is_local[Zp][ss];
+    perm   = st._permute[Zp][ss];
+    ptype  = st._permute_type[Zp];
+    if ( local && perm ) {
+      spProjZm(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjZm(chi,in._odata[offset]);
+    } else { 
+      chi=buf[offset];
+    }
+    mult(&Uchi(),&U._odata[ss](Zp),&chi());
+    accumReconZm(result,Uchi);
+
+    // Tm
+    offset = st._offsets [Tp][ss];
+    local  = st._is_local[Tp][ss];
+    perm   = st._permute[Tp][ss];
+    ptype  = st._permute_type[Tp];
+    if ( local && perm ) {
+      spProjTm(tmp,in._odata[offset]);
+      permute(chi,tmp,ptype);
+    } else if ( local ) {
+      spProjTm(chi,in._odata[offset]);
+    } else { 
+      chi=buf[offset];
+    }
+    mult(&Uchi(),&U._odata[ss](Tp),&chi());
+    accumReconTm(result,Uchi);
+
+    vstream(out._odata[ss],result);
+}
+}}
diff --git a/lib/qcd/Grid_qcd_dhop_hand.cc b/lib/qcd/Grid_qcd_dhop_hand.cc
new file mode 100644
index 00000000..f8d464fb
--- /dev/null
+++ b/lib/qcd/Grid_qcd_dhop_hand.cc
@@ -0,0 +1,769 @@
+#include <Grid.h>
+
+#define REGISTER
+
+#define LOAD_CHIMU \
+  const vSpinColourVector & ref (in._odata[offset]);	\
+    Chimu_00=ref()(0)(0);\
+    Chimu_01=ref()(0)(1);\
+    Chimu_02=ref()(0)(2);\
+    Chimu_10=ref()(1)(0);\
+    Chimu_11=ref()(1)(1);\
+    Chimu_12=ref()(1)(2);\
+    Chimu_20=ref()(2)(0);\
+    Chimu_21=ref()(2)(1);\
+    Chimu_22=ref()(2)(2);\
+    Chimu_30=ref()(3)(0);\
+    Chimu_31=ref()(3)(1);\
+    Chimu_32=ref()(3)(2);
+
+#define LOAD_CHI\
+  const vHalfSpinColourVector &ref(buf[offset]);	\
+    Chi_00 = ref()(0)(0);\
+    Chi_01 = ref()(0)(1);\
+    Chi_02 = ref()(0)(2);\
+    Chi_10 = ref()(1)(0);\
+    Chi_11 = ref()(1)(1);\
+    Chi_12 = ref()(1)(2);
+
+#define MULT_2SPIN(A)\
+   auto & ref(U._odata[ss](A));	\
+    U_00 = ref()(0,0);\
+    U_10 = ref()(1,0);\
+    U_20 = ref()(2,0);\
+    U_01 = ref()(0,1);\
+    U_11 = ref()(1,1);				\
+    U_21 = ref()(2,1);\
+    UChi_00 = U_00*Chi_00;\
+    UChi_10 = U_00*Chi_10;\
+    UChi_01 = U_10*Chi_00;\
+    UChi_11 = U_10*Chi_10;\
+    UChi_02 = U_20*Chi_00;\
+    UChi_12 = U_20*Chi_10;\
+    UChi_00+= U_01*Chi_01;\
+    UChi_10+= U_01*Chi_11;\
+    UChi_01+= U_11*Chi_01;\
+    UChi_11+= U_11*Chi_11;\
+    UChi_02+= U_21*Chi_01;\
+    UChi_12+= U_21*Chi_11;\
+    U_00 = ref()(0,2);\
+    U_10 = ref()(1,2);\
+    U_20 = ref()(2,2);\
+    UChi_00+= U_00*Chi_02;\
+    UChi_10+= U_00*Chi_12;\
+    UChi_01+= U_10*Chi_02;\
+    UChi_11+= U_10*Chi_12;\
+    UChi_02+= U_20*Chi_02;\
+    UChi_12+= U_20*Chi_12;
+
+#define PERMUTE\
+      permute(Chi_00,Chi_00,ptype);\
+      permute(Chi_01,Chi_01,ptype);\
+      permute(Chi_02,Chi_02,ptype);\
+      permute(Chi_10,Chi_10,ptype);\
+      permute(Chi_11,Chi_11,ptype);\
+      permute(Chi_12,Chi_12,ptype);
+
+//      hspin(0)=fspin(0)+timesI(fspin(3));
+//      hspin(1)=fspin(1)+timesI(fspin(2));
+#define XP_PROJ \
+    Chi_00 = Chimu_00+timesI(Chimu_30);\
+    Chi_01 = Chimu_01+timesI(Chimu_31);\
+    Chi_02 = Chimu_02+timesI(Chimu_32);\
+    Chi_10 = Chimu_10+timesI(Chimu_20);\
+    Chi_11 = Chimu_11+timesI(Chimu_21);\
+    Chi_12 = Chimu_12+timesI(Chimu_22);
+
+#define YP_PROJ \
+    Chi_00 = Chimu_00-Chimu_30;\
+    Chi_01 = Chimu_01-Chimu_31;\
+    Chi_02 = Chimu_02-Chimu_32;\
+    Chi_10 = Chimu_10+Chimu_20;\
+    Chi_11 = Chimu_11+Chimu_21;\
+    Chi_12 = Chimu_12+Chimu_22;
+
+#define ZP_PROJ \
+  Chi_00 = Chimu_00+timesI(Chimu_20);		\
+  Chi_01 = Chimu_01+timesI(Chimu_21);		\
+  Chi_02 = Chimu_02+timesI(Chimu_22);		\
+  Chi_10 = Chimu_10-timesI(Chimu_30);		\
+  Chi_11 = Chimu_11-timesI(Chimu_31);		\
+  Chi_12 = Chimu_12-timesI(Chimu_32);
+
+#define TP_PROJ \
+  Chi_00 = Chimu_00+Chimu_20;		\
+  Chi_01 = Chimu_01+Chimu_21;		\
+  Chi_02 = Chimu_02+Chimu_22;		\
+  Chi_10 = Chimu_10+Chimu_30;		\
+  Chi_11 = Chimu_11+Chimu_31;		\
+  Chi_12 = Chimu_12+Chimu_32;
+
+
+//      hspin(0)=fspin(0)-timesI(fspin(3));
+//      hspin(1)=fspin(1)-timesI(fspin(2));
+#define XM_PROJ \
+    Chi_00 = Chimu_00-timesI(Chimu_30);\
+    Chi_01 = Chimu_01-timesI(Chimu_31);\
+    Chi_02 = Chimu_02-timesI(Chimu_32);\
+    Chi_10 = Chimu_10-timesI(Chimu_20);\
+    Chi_11 = Chimu_11-timesI(Chimu_21);\
+    Chi_12 = Chimu_12-timesI(Chimu_22);
+
+#define YM_PROJ \
+    Chi_00 = Chimu_00+Chimu_30;\
+    Chi_01 = Chimu_01+Chimu_31;\
+    Chi_02 = Chimu_02+Chimu_32;\
+    Chi_10 = Chimu_10-Chimu_20;\
+    Chi_11 = Chimu_11-Chimu_21;\
+    Chi_12 = Chimu_12-Chimu_22;
+
+#define ZM_PROJ \
+  Chi_00 = Chimu_00-timesI(Chimu_20);		\
+  Chi_01 = Chimu_01-timesI(Chimu_21);		\
+  Chi_02 = Chimu_02-timesI(Chimu_22);		\
+  Chi_10 = Chimu_10+timesI(Chimu_30);		\
+  Chi_11 = Chimu_11+timesI(Chimu_31);		\
+  Chi_12 = Chimu_12+timesI(Chimu_32);
+
+#define TM_PROJ \
+  Chi_00 = Chimu_00-Chimu_20;		\
+  Chi_01 = Chimu_01-Chimu_21;		\
+  Chi_02 = Chimu_02-Chimu_22;		\
+  Chi_10 = Chimu_10-Chimu_30;		\
+  Chi_11 = Chimu_11-Chimu_31;		\
+  Chi_12 = Chimu_12-Chimu_32;
+
+//      fspin(0)=hspin(0);
+//      fspin(1)=hspin(1);
+//      fspin(2)=timesMinusI(hspin(1));
+//      fspin(3)=timesMinusI(hspin(0));
+#define XP_RECON\
+  result_00 = UChi_00;\
+  result_01 = UChi_01;\
+  result_02 = UChi_02;\
+  result_10 = UChi_10;\
+  result_11 = UChi_11;\
+  result_12 = UChi_12;\
+  result_20 = timesMinusI(UChi_10);\
+  result_21 = timesMinusI(UChi_11);\
+  result_22 = timesMinusI(UChi_12);\
+  result_30 = timesMinusI(UChi_00);\
+  result_31 = timesMinusI(UChi_01);\
+  result_32 = timesMinusI(UChi_02);
+
+#define XP_RECON_ACCUM\
+  result_00+=UChi_00;\
+  result_01+=UChi_01;\
+  result_02+=UChi_02;\
+  result_10+=UChi_10;\
+  result_11+=UChi_11;\
+  result_12+=UChi_12;\
+  result_20-=timesI(UChi_10);\
+  result_21-=timesI(UChi_11);\
+  result_22-=timesI(UChi_12);\
+  result_30-=timesI(UChi_00);\
+  result_31-=timesI(UChi_01);\
+  result_32-=timesI(UChi_02);
+
+#define XM_RECON\
+  result_00 = UChi_00;\
+  result_01 = UChi_01;\
+  result_02 = UChi_02;\
+  result_10 = UChi_10;\
+  result_11 = UChi_11;\
+  result_12 = UChi_12;\
+  result_20 = timesI(UChi_10);\
+  result_21 = timesI(UChi_11);\
+  result_22 = timesI(UChi_12);\
+  result_30 = timesI(UChi_00);\
+  result_31 = timesI(UChi_01);\
+  result_32 = timesI(UChi_02);
+
+#define XM_RECON_ACCUM\
+  result_00+= UChi_00;\
+  result_01+= UChi_01;\
+  result_02+= UChi_02;\
+  result_10+= UChi_10;\
+  result_11+= UChi_11;\
+  result_12+= UChi_12;\
+  result_20+= timesI(UChi_10);\
+  result_21+= timesI(UChi_11);\
+  result_22+= timesI(UChi_12);\
+  result_30+= timesI(UChi_00);\
+  result_31+= timesI(UChi_01);\
+  result_32+= timesI(UChi_02);
+
+#define YP_RECON_ACCUM\
+  result_00+= UChi_00;\
+  result_01+= UChi_01;\
+  result_02+= UChi_02;\
+  result_10+= UChi_10;\
+  result_11+= UChi_11;\
+  result_12+= UChi_12;\
+  result_20+= UChi_10;\
+  result_21+= UChi_11;\
+  result_22+= UChi_12;\
+  result_30-= UChi_00;\
+  result_31-= UChi_01;\
+  result_32-= UChi_02;
+
+#define YM_RECON_ACCUM\
+  result_00+= UChi_00;\
+  result_01+= UChi_01;\
+  result_02+= UChi_02;\
+  result_10+= UChi_10;\
+  result_11+= UChi_11;\
+  result_12+= UChi_12;\
+  result_20-= UChi_10;\
+  result_21-= UChi_11;\
+  result_22-= UChi_12;\
+  result_30+= UChi_00;\
+  result_31+= UChi_01;\
+  result_32+= UChi_02;
+
+#define ZP_RECON_ACCUM\
+  result_00+= UChi_00;\
+  result_01+= UChi_01;\
+  result_02+= UChi_02;\
+  result_10+= UChi_10;\
+  result_11+= UChi_11;\
+  result_12+= UChi_12;\
+  result_20-= timesI(UChi_00);			\
+  result_21-= timesI(UChi_01);			\
+  result_22-= timesI(UChi_02);			\
+  result_30+= timesI(UChi_10);			\
+  result_31+= timesI(UChi_11);			\
+  result_32+= timesI(UChi_12);
+
+#define ZM_RECON_ACCUM\
+  result_00+= UChi_00;\
+  result_01+= UChi_01;\
+  result_02+= UChi_02;\
+  result_10+= UChi_10;\
+  result_11+= UChi_11;\
+  result_12+= UChi_12;\
+  result_20+= timesI(UChi_00);			\
+  result_21+= timesI(UChi_01);			\
+  result_22+= timesI(UChi_02);			\
+  result_30-= timesI(UChi_10);			\
+  result_31-= timesI(UChi_11);			\
+  result_32-= timesI(UChi_12);
+
+#define TP_RECON_ACCUM\
+  result_00+= UChi_00;\
+  result_01+= UChi_01;\
+  result_02+= UChi_02;\
+  result_10+= UChi_10;\
+  result_11+= UChi_11;\
+  result_12+= UChi_12;\
+  result_20+= UChi_00;			\
+  result_21+= UChi_01;			\
+  result_22+= UChi_02;			\
+  result_30+= UChi_10;			\
+  result_31+= UChi_11;			\
+  result_32+= UChi_12;
+
+#define TM_RECON_ACCUM\
+  result_00+= UChi_00;\
+  result_01+= UChi_01;\
+  result_02+= UChi_02;\
+  result_10+= UChi_10;\
+  result_11+= UChi_11;\
+  result_12+= UChi_12;\
+  result_20-= UChi_00;	\
+  result_21-= UChi_01;	\
+  result_22-= UChi_02;	\
+  result_30-= UChi_10;	\
+  result_31-= UChi_11;	\
+  result_32-= UChi_12;
+
+namespace Grid {
+namespace QCD {
+
+void DiracOptHand::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
+			    std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+			    int ss,const LatticeFermion &in, LatticeFermion &out)
+{
+  REGISTER vComplex result_00; // 12 regs on knc
+  REGISTER vComplex result_01;
+  REGISTER vComplex result_02;
+  
+  REGISTER vComplex result_10;
+  REGISTER vComplex result_11;
+  REGISTER vComplex result_12;
+
+  REGISTER vComplex result_20;
+  REGISTER vComplex result_21;
+  REGISTER vComplex result_22;
+
+  REGISTER vComplex result_30;
+  REGISTER vComplex result_31;
+  REGISTER vComplex result_32; // 20 left
+
+  REGISTER vComplex Chi_00;    // two spinor; 6 regs
+  REGISTER vComplex Chi_01;
+  REGISTER vComplex Chi_02;
+
+  REGISTER vComplex Chi_10;
+  REGISTER vComplex Chi_11;
+  REGISTER vComplex Chi_12;   // 14 left
+
+  REGISTER vComplex UChi_00;  // two spinor; 6 regs
+  REGISTER vComplex UChi_01;
+  REGISTER vComplex UChi_02;
+
+  REGISTER vComplex UChi_10;
+  REGISTER vComplex UChi_11;
+  REGISTER vComplex UChi_12;  // 8 left
+
+  REGISTER vComplex U_00;  // two rows of U matrix
+  REGISTER vComplex U_10;
+  REGISTER vComplex U_20;  
+  REGISTER vComplex U_01;
+  REGISTER vComplex U_11;
+  REGISTER vComplex U_21;  // 2 reg left.
+
+#define Chimu_00 Chi_00
+#define Chimu_01 Chi_01
+#define Chimu_02 Chi_02
+#define Chimu_10 Chi_10
+#define Chimu_11 Chi_11
+#define Chimu_12 Chi_12
+#define Chimu_20 UChi_00
+#define Chimu_21 UChi_01
+#define Chimu_22 UChi_02
+#define Chimu_30 UChi_10
+#define Chimu_31 UChi_11
+#define Chimu_32 UChi_12
+
+
+  int offset,local,perm, ptype;
+
+  // Xp
+  offset = st._offsets [Xp][ss];
+  local  = st._is_local[Xp][ss];
+  perm   = st._permute[Xp][ss];
+  ptype  = st._permute_type[Xp];
+  
+  if ( local ) {
+    LOAD_CHIMU;
+    XP_PROJ;
+    if ( perm) {
+      PERMUTE;
+    }
+  } else { 
+    LOAD_CHI;
+  }
+
+  {
+    MULT_2SPIN(Xp);
+  }
+  XP_RECON;
+  //  std::cout << "XP_RECON"<<std::endl;
+  //  std::cout << result_00 <<" "<<result_01 <<" "<<result_02 <<std::endl;
+  //  std::cout << result_10 <<" "<<result_11 <<" "<<result_12 <<std::endl;
+  //  std::cout << result_20 <<" "<<result_21 <<" "<<result_22 <<std::endl;
+  //  std::cout << result_30 <<" "<<result_31 <<" "<<result_32 <<std::endl;
+
+  // Yp
+  offset = st._offsets [Yp][ss];
+  local  = st._is_local[Yp][ss];
+  perm   = st._permute[Yp][ss];
+  ptype  = st._permute_type[Yp];
+  
+  if ( local ) {
+    LOAD_CHIMU;
+    YP_PROJ;
+    if ( perm) {
+      PERMUTE;
+    }
+  } else { 
+    LOAD_CHI;
+  }
+  {
+    MULT_2SPIN(Yp);
+  }
+  YP_RECON_ACCUM;
+
+
+  // Zp
+  offset = st._offsets [Zp][ss];
+  local  = st._is_local[Zp][ss];
+  perm   = st._permute[Zp][ss];
+  ptype  = st._permute_type[Zp];
+  
+  if ( local ) {
+    LOAD_CHIMU;
+    ZP_PROJ;
+    if ( perm) {
+      PERMUTE;
+    }
+  } else { 
+    LOAD_CHI;
+  }
+  {
+    MULT_2SPIN(Zp);
+  }
+  ZP_RECON_ACCUM;
+
+  // Tp
+  offset = st._offsets [Tp][ss];
+  local  = st._is_local[Tp][ss];
+  perm   = st._permute[Tp][ss];
+  ptype  = st._permute_type[Tp];
+  
+  if ( local ) {
+    LOAD_CHIMU;
+    TP_PROJ;
+    if ( perm) {
+      PERMUTE;
+    }
+  } else { 
+    LOAD_CHI;
+  }
+  {
+    MULT_2SPIN(Tp);
+  }
+  TP_RECON_ACCUM;
+  
+  // Xm
+  offset = st._offsets [Xm][ss];
+  local  = st._is_local[Xm][ss];
+  perm   = st._permute[Xm][ss];
+  ptype  = st._permute_type[Xm];
+  
+  if ( local ) {
+    LOAD_CHIMU;
+    XM_PROJ;
+    if ( perm) {
+      PERMUTE;
+    }
+  } else { 
+    LOAD_CHI;
+  }
+  {
+    MULT_2SPIN(Xm);
+  }
+  XM_RECON_ACCUM;
+  //  std::cout << "XM_RECON_ACCUM"<<std::endl;
+  //  std::cout << result_00 <<" "<<result_01 <<" "<<result_02 <<std::endl;
+  //  std::cout << result_10 <<" "<<result_11 <<" "<<result_12 <<std::endl;
+  //  std::cout << result_20 <<" "<<result_21 <<" "<<result_22 <<std::endl;
+  //  std::cout << result_30 <<" "<<result_31 <<" "<<result_32 <<std::endl;
+  
+  
+  // Ym
+  offset = st._offsets [Ym][ss];
+  local  = st._is_local[Ym][ss];
+  perm   = st._permute[Ym][ss];
+  ptype  = st._permute_type[Ym];
+  
+  if ( local ) {
+    LOAD_CHIMU;
+    YM_PROJ;
+    if ( perm) {
+      PERMUTE;
+    }
+  } else { 
+    LOAD_CHI;
+  }
+  {
+    MULT_2SPIN(Ym);
+  }
+  YM_RECON_ACCUM;
+
+  // Zm
+  offset = st._offsets [Zm][ss];
+  local  = st._is_local[Zm][ss];
+  perm   = st._permute[Zm][ss];
+  ptype  = st._permute_type[Zm];
+
+  if ( local ) {
+    LOAD_CHIMU;
+    ZM_PROJ;
+    if ( perm) {
+      PERMUTE;
+    }
+  } else { 
+    LOAD_CHI;
+  }
+  {
+    MULT_2SPIN(Zm);
+  }
+  ZM_RECON_ACCUM;
+
+  // Tm
+  offset = st._offsets [Tm][ss];
+  local  = st._is_local[Tm][ss];
+  perm   = st._permute[Tm][ss];
+  ptype  = st._permute_type[Tm];
+
+  if ( local ) {
+    LOAD_CHIMU;
+    TM_PROJ;
+    if ( perm) {
+      PERMUTE;
+    }
+  } else { 
+    LOAD_CHI;
+  }
+  {
+    MULT_2SPIN(Tm);
+  }
+  TM_RECON_ACCUM;
+
+  {
+    vSpinColourVector & ref (out._odata[ss]);
+    vstream(ref()(0)(0),result_00);
+    vstream(ref()(0)(1),result_01);
+    vstream(ref()(0)(2),result_02);
+    vstream(ref()(1)(0),result_10);
+    vstream(ref()(1)(1),result_11);
+    vstream(ref()(1)(2),result_12);
+    vstream(ref()(2)(0),result_20);
+    vstream(ref()(2)(1),result_21);
+    vstream(ref()(2)(2),result_22);
+    vstream(ref()(3)(0),result_30);
+    vstream(ref()(3)(1),result_31);
+    vstream(ref()(3)(2),result_32);
+  }
+}
+
+void DiracOptHand::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
+			       std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+			       int ss,const LatticeFermion &in, LatticeFermion &out)
+{
+  REGISTER vComplex result_00; // 12 regs on knc
+  REGISTER vComplex result_01;
+  REGISTER vComplex result_02;
+
+  REGISTER vComplex result_10;
+  REGISTER vComplex result_11;
+  REGISTER vComplex result_12;
+
+  REGISTER vComplex result_20;
+  REGISTER vComplex result_21;
+  REGISTER vComplex result_22;
+
+  REGISTER vComplex result_30;
+  REGISTER vComplex result_31;
+  REGISTER vComplex result_32; // 20 left
+
+  REGISTER vComplex Chi_00;    // two spinor; 6 regs
+  REGISTER vComplex Chi_01;
+  REGISTER vComplex Chi_02;
+
+  REGISTER vComplex Chi_10;
+  REGISTER vComplex Chi_11;
+  REGISTER vComplex Chi_12;   // 14 left
+
+  REGISTER vComplex UChi_00;  // two spinor; 6 regs
+  REGISTER vComplex UChi_01;
+  REGISTER vComplex UChi_02;
+
+  REGISTER vComplex UChi_10;
+  REGISTER vComplex UChi_11;
+  REGISTER vComplex UChi_12;  // 8 left
+
+  REGISTER vComplex U_00;  // two rows of U matrix
+  REGISTER vComplex U_10;
+  REGISTER vComplex U_20;  
+  REGISTER vComplex U_01;
+  REGISTER vComplex U_11;
+  REGISTER vComplex U_21;  // 2 reg left.
+
+#define Chimu_00 Chi_00
+#define Chimu_01 Chi_01
+#define Chimu_02 Chi_02
+#define Chimu_10 Chi_10
+#define Chimu_11 Chi_11
+#define Chimu_12 Chi_12
+#define Chimu_20 UChi_00
+#define Chimu_21 UChi_01
+#define Chimu_22 UChi_02
+#define Chimu_30 UChi_10
+#define Chimu_31 UChi_11
+#define Chimu_32 UChi_12
+
+
+  int offset,local,perm, ptype;
+
+  // Xp
+  offset = st._offsets [Xp][ss];
+  local  = st._is_local[Xp][ss];
+  perm   = st._permute[Xp][ss];
+  ptype  = st._permute_type[Xp];
+  
+  if ( local ) {
+    LOAD_CHIMU;
+    XM_PROJ;
+    if ( perm) {
+      PERMUTE;
+    }
+  } else { 
+    LOAD_CHI;
+  }
+  {
+    MULT_2SPIN(Xp);
+  }
+  XM_RECON;
+  
+  // Yp
+  offset = st._offsets [Yp][ss];
+  local  = st._is_local[Yp][ss];
+  perm   = st._permute[Yp][ss];
+  ptype  = st._permute_type[Yp];
+  
+  if ( local ) {
+    LOAD_CHIMU;
+    YM_PROJ;
+    if ( perm) {
+      PERMUTE;
+    }
+  } else { 
+    LOAD_CHI;
+  }
+  {
+    MULT_2SPIN(Yp);
+  }
+  YM_RECON_ACCUM;
+
+
+  // Zp
+  offset = st._offsets [Zp][ss];
+  local  = st._is_local[Zp][ss];
+  perm   = st._permute[Zp][ss];
+  ptype  = st._permute_type[Zp];
+  
+  if ( local ) {
+    LOAD_CHIMU;
+    ZM_PROJ;
+    if ( perm) {
+      PERMUTE;
+    }
+  } else { 
+    LOAD_CHI;
+  }
+  {
+    MULT_2SPIN(Zp);
+  }
+  ZM_RECON_ACCUM;
+
+  // Tp
+  offset = st._offsets [Tp][ss];
+  local  = st._is_local[Tp][ss];
+  perm   = st._permute[Tp][ss];
+  ptype  = st._permute_type[Tp];
+  
+  if ( local ) {
+    LOAD_CHIMU;
+    TM_PROJ;
+    if ( perm) {
+      PERMUTE;
+    }
+  } else { 
+    LOAD_CHI;
+  }
+  {
+    MULT_2SPIN(Tp);
+  }
+  TM_RECON_ACCUM;
+  
+  // Xm
+  offset = st._offsets [Xm][ss];
+  local  = st._is_local[Xm][ss];
+  perm   = st._permute[Xm][ss];
+  ptype  = st._permute_type[Xm];
+  
+  if ( local ) {
+    LOAD_CHIMU;
+    XP_PROJ;
+    if ( perm) {
+      PERMUTE;
+    }
+  } else { 
+    LOAD_CHI;
+  }
+  {
+    MULT_2SPIN(Xm);
+  }
+  XP_RECON_ACCUM;
+  
+  
+  // Ym
+  offset = st._offsets [Ym][ss];
+  local  = st._is_local[Ym][ss];
+  perm   = st._permute[Ym][ss];
+  ptype  = st._permute_type[Ym];
+  
+  if ( local ) {
+    LOAD_CHIMU;
+    YP_PROJ;
+    if ( perm) {
+      PERMUTE;
+    }
+  } else { 
+    LOAD_CHI;
+  }
+  {
+    MULT_2SPIN(Ym);
+  }
+  YP_RECON_ACCUM;
+
+  // Zm
+  offset = st._offsets [Zm][ss];
+  local  = st._is_local[Zm][ss];
+  perm   = st._permute[Zm][ss];
+  ptype  = st._permute_type[Zm];
+
+  if ( local ) {
+    LOAD_CHIMU;
+    ZP_PROJ;
+    if ( perm) {
+      PERMUTE;
+    }
+  } else { 
+    LOAD_CHI;
+  }
+  {
+    MULT_2SPIN(Zm);
+  }
+  ZP_RECON_ACCUM;
+
+  // Tm
+  offset = st._offsets [Tm][ss];
+  local  = st._is_local[Tm][ss];
+  perm   = st._permute[Tm][ss];
+  ptype  = st._permute_type[Tm];
+
+  if ( local ) {
+    LOAD_CHIMU;
+    TP_PROJ;
+    if ( perm) {
+      PERMUTE;
+    }
+  } else { 
+    LOAD_CHI;
+  }
+  {
+    MULT_2SPIN(Tm);
+  }
+  TP_RECON_ACCUM;
+
+  {
+    vSpinColourVector & ref (out._odata[ss]);
+    vstream(ref()(0)(0),result_00);
+    vstream(ref()(0)(1),result_01);
+    vstream(ref()(0)(2),result_02);
+    vstream(ref()(1)(0),result_10);
+    vstream(ref()(1)(1),result_11);
+    vstream(ref()(1)(2),result_12);
+    vstream(ref()(2)(0),result_20);
+    vstream(ref()(2)(1),result_21);
+    vstream(ref()(2)(2),result_22);
+    vstream(ref()(3)(0),result_30);
+    vstream(ref()(3)(1),result_31);
+    vstream(ref()(3)(2),result_32);
+  }
+}
+}}

From 6d2e056187f620e90919478e50125ed22da590b8 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 26 May 2015 22:20:09 +0100
Subject: [PATCH 238/429] Simd revert to Guido's commit. I edited concurrently
 and things went bad.

---
 lib/Grid_simd.h | 134 +-----------------------------------------------
 1 file changed, 1 insertion(+), 133 deletions(-)

diff --git a/lib/Grid_simd.h b/lib/Grid_simd.h
index 504e1c17..cccc82e0 100644
--- a/lib/Grid_simd.h
+++ b/lib/Grid_simd.h
@@ -94,139 +94,7 @@ namespace Grid {
   template<>            inline void zeroit(ComplexD &arg){ arg=0; };
   template<>            inline void zeroit(RealF &arg){ arg=0; };
   template<>            inline void zeroit(RealD &arg){ arg=0; };
-
-#if defined (SSE4)
-    typedef __m128 fvec;
-    typedef __m128d dvec;
-    typedef __m128 cvec;
-    typedef __m128d zvec;
-    typedef __m128i ivec;
-#endif
-#if defined (AVX1) || defined (AVX2)
-    typedef __m256 fvec;
-    typedef __m256d dvec;
-    typedef __m256  cvec;
-    typedef __m256d zvec;
-    typedef __m256i ivec;
-#endif
-#if defined (AVX512)
-    typedef __m512  fvec;
-    typedef __m512d dvec;
-    typedef __m512  cvec;
-    typedef __m512d zvec;
-    typedef __m512i ivec;
-#endif
-#if defined (QPX)
-    typedef float  fvec __attribute__ ((vector_size (16))); // QPX has same SIMD width irrespective of precision
-    typedef float  cvec __attribute__ ((vector_size (16)));
-    
-    typedef vector4double dvec;
-    typedef vector4double zvec;
-#endif
-#if defined (AVX1) || defined (AVX2) || defined (AVX512)
-    inline void v_prefetch0(int size, const char *ptr){
-          for(int i=0;i<size;i+=64){ //  Define L1 linesize above// What about SSE?
-            _mm_prefetch(ptr+i+4096,_MM_HINT_T1);
-            _mm_prefetch(ptr+i+512,_MM_HINT_T0);
-          }
-    }
-#else 
-    inline void v_prefetch0(int size, const char *ptr){};
-#endif
-
-
-//////////////////////////////////////////////////////////
-// Permute
-// Permute 0 every ABCDEFGH -> BA DC FE HG
-// Permute 1 every ABCDEFGH -> CD AB GH EF
-// Permute 2 every ABCDEFGH -> EFGH ABCD
-// Permute 3 possible on longer iVector lengths (512bit = 8 double = 16 single)
-// Permute 4 possible on half precision @512bit vectors.
-//////////////////////////////////////////////////////////
-template<class vsimd>
-inline void Gpermute0(vsimd &y,const vsimd &b) {
-  union { 
-    fvec f;
-    decltype(vsimd::v) v;
-  } conv;
-  conv.v = b.v;
-#ifdef SSE4
-  conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2));
-#endif
-#if defined(AVX1)||defined(AVX2)
-  conv.f = _mm256_permute2f128_ps(conv.f,conv.f,0x01); 
-#endif  
-#ifdef AVX512
- conv.f = _mm512_permute4f128_ps(conv.f,(_MM_PERM_ENUM)_MM_SHUFFLE(1,0,3,2)); 
-#endif
- y.v=conv.v;
-};
-template<class vsimd>
-inline void Gpermute1(vsimd &y,const vsimd &b) {
-  union { 
-    fvec f;
-    decltype(vsimd::v) v;
-  } conv;
-  conv.v = b.v;
-#ifdef SSE4
-  conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); 
-#endif
-#if defined(AVX1)||defined(AVX2)
-  conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2)); 
-#endif  
-#ifdef AVX512
- conv.f = _mm512_permute4f128_ps(conv.f,(_MM_PERM_ENUM)_MM_SHUFFLE(2,3,0,1)); 
-#endif  
-  y.v=conv.v;
-};
-template<class vsimd>
-inline void Gpermute2(vsimd &y,const vsimd &b) {
-  union { 
-    fvec f;
-    decltype(vsimd::v) v;
-  } conv;
-  conv.v = b.v;
-#ifdef SSE4
-#endif
-#if defined(AVX1)||defined(AVX2)
-  conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); 
-#endif  
-#ifdef AVX512
-  conv.f = _mm512_swizzle_ps(conv.f,_MM_SWIZ_REG_BADC); 
-#endif  
-  y.v=conv.v;
-
-};
-template<class vsimd>
-inline void Gpermute3(vsimd &y,const vsimd &b) {
-  union { 
-    fvec f;
-    decltype(vsimd::v) v;
-  } conv;
-  conv.v = b.v;
-#ifdef AVX512
-  conv.f = _mm512_swizzle_ps(conv.f,_MM_SWIZ_REG_CDAB); 
-#endif  
-  y.v=conv.v;
-
-};
-
-template<class vsimd>
-inline void Gpermute(vsimd &y,const vsimd &b,int perm){
-	union { 
-	  fvec f;
-	  decltype(vsimd::v) v;
-	} conv;
-	conv.v = b.v;
-      switch (perm){
-      case 3: Gpermute3(y,b); break;
-      case 2: Gpermute2(y,b); break;
-      case 1: Gpermute1(y,b); break;
-      case 0: Gpermute0(y,b); break;
-      default: assert(0); break;
-      }
-    };
-
+  
 };
 
 #include <simd/Grid_vector_types.h>

From b6a28f1de786216b9b6fd026e1d7b4f0145ddb77 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 26 May 2015 22:20:40 +0100
Subject: [PATCH 239/429] Auto gen files should never have been committed, but
 making everyone run aclocal, automake, autoconf is a pain in the ass.

---
 Makefile.in  | 44 +++++++++++++++++++++++--------------
 aclocal.m4   | 61 +++++++++++++++++++++++++++-------------------------
 config.guess |  2 +-
 config.sub   |  2 +-
 configure    | 13 +++++------
 5 files changed, 69 insertions(+), 53 deletions(-)

diff --git a/Makefile.in b/Makefile.in
index d473c2df..a6508b45 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -1,7 +1,7 @@
-# Makefile.in generated by automake 1.14.1 from Makefile.am.
+# Makefile.in generated by automake 1.15 from Makefile.am.
 # @configure_input@
 
-# Copyright (C) 1994-2013 Free Software Foundation, Inc.
+# Copyright (C) 1994-2014 Free Software Foundation, Inc.
 
 # This Makefile.in is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -14,7 +14,17 @@
 
 @SET_MAKE@
 VPATH = @srcdir@
-am__is_gnu_make = test -n '$(MAKEFILE_LIST)' && test -n '$(MAKELEVEL)'
+am__is_gnu_make = { \
+  if test -z '$(MAKELEVEL)'; then \
+    false; \
+  elif test -n '$(MAKE_HOST)'; then \
+    true; \
+  elif test -n '$(MAKE_VERSION)' && test -n '$(CURDIR)'; then \
+    true; \
+  else \
+    false; \
+  fi; \
+}
 am__make_running_with_option = \
   case $${target_option-} in \
       ?) ;; \
@@ -79,15 +89,13 @@ build_triplet = @build@
 host_triplet = @host@
 target_triplet = @target@
 subdir = .
-DIST_COMMON = INSTALL NEWS README AUTHORS ChangeLog \
-	$(srcdir)/Makefile.in $(srcdir)/Makefile.am \
-	$(top_srcdir)/configure $(am__configure_deps) COPYING TODO \
-	compile config.guess config.sub depcomp install-sh missing
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/m4/ax_cxx_compile_stdcxx_11.m4 \
 	$(top_srcdir)/configure.ac
 am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
 	$(ACLOCAL_M4)
+DIST_COMMON = $(srcdir)/Makefile.am $(top_srcdir)/configure \
+	$(am__configure_deps) $(am__DIST_COMMON)
 am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \
  configure.lineno config.status.lineno
 mkinstalldirs = $(install_sh) -d
@@ -150,6 +158,9 @@ ETAGS = etags
 CTAGS = ctags
 CSCOPE = cscope
 DIST_SUBDIRS = $(SUBDIRS)
+am__DIST_COMMON = $(srcdir)/Makefile.in AUTHORS COPYING ChangeLog \
+	INSTALL NEWS README TODO compile config.guess config.sub \
+	depcomp install-sh missing
 DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
 distdir = $(PACKAGE)-$(VERSION)
 top_distdir = $(distdir)
@@ -316,7 +327,6 @@ $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu Makefile'; \
 	$(am__cd) $(top_srcdir) && \
 	  $(AUTOMAKE) --gnu Makefile
-.PRECIOUS: Makefile
 Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
 	@case '$?' in \
 	  *config.status*) \
@@ -523,15 +533,15 @@ dist-xz: distdir
 	$(am__post_remove_distdir)
 
 dist-tarZ: distdir
-	@echo WARNING: "Support for shar distribution archives is" \
-	               "deprecated." >&2
+	@echo WARNING: "Support for distribution archives compressed with" \
+		       "legacy program 'compress' is deprecated." >&2
 	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	tardir=$(distdir) && $(am__tar) | compress -c >$(distdir).tar.Z
 	$(am__post_remove_distdir)
 
 dist-shar: distdir
-	@echo WARNING: "Support for distribution archives compressed with" \
-		       "legacy program 'compress' is deprecated." >&2
+	@echo WARNING: "Support for shar distribution archives is" \
+	               "deprecated." >&2
 	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	shar $(distdir) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).shar.gz
 	$(am__post_remove_distdir)
@@ -567,17 +577,17 @@ distcheck: dist
 	esac
 	chmod -R a-w $(distdir)
 	chmod u+w $(distdir)
-	mkdir $(distdir)/_build $(distdir)/_inst
+	mkdir $(distdir)/_build $(distdir)/_build/sub $(distdir)/_inst
 	chmod a-w $(distdir)
 	test -d $(distdir)/_build || exit 0; \
 	dc_install_base=`$(am__cd) $(distdir)/_inst && pwd | sed -e 's,^[^:\\/]:[\\/],/,'` \
 	  && dc_destdir="$${TMPDIR-/tmp}/am-dc-$$$$/" \
 	  && am__cwd=`pwd` \
-	  && $(am__cd) $(distdir)/_build \
-	  && ../configure \
+	  && $(am__cd) $(distdir)/_build/sub \
+	  && ../../configure \
 	    $(AM_DISTCHECK_CONFIGURE_FLAGS) \
 	    $(DISTCHECK_CONFIGURE_FLAGS) \
-	    --srcdir=.. --prefix="$$dc_install_base" \
+	    --srcdir=../.. --prefix="$$dc_install_base" \
 	  && $(MAKE) $(AM_MAKEFLAGS) \
 	  && $(MAKE) $(AM_MAKEFLAGS) dvi \
 	  && $(MAKE) $(AM_MAKEFLAGS) check \
@@ -751,6 +761,8 @@ uninstall-am:
 	maintainer-clean-generic mostlyclean mostlyclean-generic pdf \
 	pdf-am ps ps-am tags tags-am uninstall uninstall-am
 
+.PRECIOUS: Makefile
+
 
 # Tell versions [3.59,3.63) of GNU make to not export all variables.
 # Otherwise a system limit (for SysV at least) may be exceeded.
diff --git a/aclocal.m4 b/aclocal.m4
index a3d1bc9c..a358f21e 100644
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -1,6 +1,6 @@
-# generated automatically by aclocal 1.14.1 -*- Autoconf -*-
+# generated automatically by aclocal 1.15 -*- Autoconf -*-
 
-# Copyright (C) 1996-2013 Free Software Foundation, Inc.
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -20,7 +20,7 @@ You have another version of autoconf.  It may work, but is not guaranteed to.
 If you have problems, you may need to regenerate the build system entirely.
 To do so, use the procedure documented by the package, typically 'autoreconf'.])])
 
-# Copyright (C) 2002-2013 Free Software Foundation, Inc.
+# Copyright (C) 2002-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -32,10 +32,10 @@ To do so, use the procedure documented by the package, typically 'autoreconf'.])
 # generated from the m4 files accompanying Automake X.Y.
 # (This private macro should not be called outside this file.)
 AC_DEFUN([AM_AUTOMAKE_VERSION],
-[am__api_version='1.14'
+[am__api_version='1.15'
 dnl Some users find AM_AUTOMAKE_VERSION and mistake it for a way to
 dnl require some minimum version.  Point them to the right macro.
-m4_if([$1], [1.14.1], [],
+m4_if([$1], [1.15], [],
       [AC_FATAL([Do not call $0, use AM_INIT_AUTOMAKE([$1]).])])dnl
 ])
 
@@ -51,14 +51,14 @@ m4_define([_AM_AUTOCONF_VERSION], [])
 # Call AM_AUTOMAKE_VERSION and AM_AUTOMAKE_VERSION so they can be traced.
 # This function is AC_REQUIREd by AM_INIT_AUTOMAKE.
 AC_DEFUN([AM_SET_CURRENT_AUTOMAKE_VERSION],
-[AM_AUTOMAKE_VERSION([1.14.1])dnl
+[AM_AUTOMAKE_VERSION([1.15])dnl
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
 _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 
 # AM_AUX_DIR_EXPAND                                         -*- Autoconf -*-
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -103,15 +103,14 @@ _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 # configured tree to be moved without reconfiguration.
 
 AC_DEFUN([AM_AUX_DIR_EXPAND],
-[dnl Rely on autoconf to set up CDPATH properly.
-AC_PREREQ([2.50])dnl
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+[AC_REQUIRE([AC_CONFIG_AUX_DIR_DEFAULT])dnl
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 ])
 
 # AM_CONDITIONAL                                            -*- Autoconf -*-
 
-# Copyright (C) 1997-2013 Free Software Foundation, Inc.
+# Copyright (C) 1997-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -142,7 +141,7 @@ AC_CONFIG_COMMANDS_PRE(
 Usually this means the macro was only invoked conditionally.]])
 fi])])
 
-# Copyright (C) 1999-2013 Free Software Foundation, Inc.
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -333,7 +332,7 @@ _AM_SUBST_NOTMAKE([am__nodep])dnl
 
 # Generate code to set up dependency tracking.              -*- Autoconf -*-
 
-# Copyright (C) 1999-2013 Free Software Foundation, Inc.
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -409,7 +408,7 @@ AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 
 # Do all the work for Automake.                             -*- Autoconf -*-
 
-# Copyright (C) 1996-2013 Free Software Foundation, Inc.
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -499,8 +498,8 @@ AC_REQUIRE([AC_PROG_MKDIR_P])dnl
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 AC_SUBST([mkdir_p], ['$(MKDIR_P)'])
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
 AC_REQUIRE([AC_PROG_AWK])dnl
 AC_REQUIRE([AC_PROG_MAKE_SET])dnl
 AC_REQUIRE([AM_SET_LEADING_DOT])dnl
@@ -573,7 +572,11 @@ to "yes", and re-run configure.
 END
     AC_MSG_ERROR([Your 'rm' program is bad, sorry.])
   fi
-fi])
+fi
+dnl The trailing newline in this macro's definition is deliberate, for
+dnl backward compatibility and to allow trailing 'dnl'-style comments
+dnl after the AM_INIT_AUTOMAKE invocation. See automake bug#16841.
+])
 
 dnl Hook into '_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
 dnl add the conditional right here, as _AC_COMPILER_EXEEXT may be further
@@ -602,7 +605,7 @@ for _am_header in $config_headers :; do
 done
 echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_count])
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -613,7 +616,7 @@ echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_co
 # Define $install_sh.
 AC_DEFUN([AM_PROG_INSTALL_SH],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -623,7 +626,7 @@ if test x"${install_sh}" != xset; then
 fi
 AC_SUBST([install_sh])])
 
-# Copyright (C) 2003-2013 Free Software Foundation, Inc.
+# Copyright (C) 2003-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -644,7 +647,7 @@ AC_SUBST([am__leading_dot])])
 
 # Check to see how 'make' treats includes.	            -*- Autoconf -*-
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -694,7 +697,7 @@ rm -f confinc confmf
 
 # Fake the existence of programs that GNU maintainers use.  -*- Autoconf -*-
 
-# Copyright (C) 1997-2013 Free Software Foundation, Inc.
+# Copyright (C) 1997-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -733,7 +736,7 @@ fi
 
 # Helper functions for option handling.                     -*- Autoconf -*-
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -764,7 +767,7 @@ AC_DEFUN([_AM_IF_OPTION],
 
 # Check to make sure that the build environment is sane.    -*- Autoconf -*-
 
-# Copyright (C) 1996-2013 Free Software Foundation, Inc.
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -845,7 +848,7 @@ AC_CONFIG_COMMANDS_PRE(
 rm -f conftest.file
 ])
 
-# Copyright (C) 2009-2013 Free Software Foundation, Inc.
+# Copyright (C) 2009-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -905,7 +908,7 @@ AC_SUBST([AM_BACKSLASH])dnl
 _AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
 ])
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -933,7 +936,7 @@ fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 AC_SUBST([INSTALL_STRIP_PROGRAM])])
 
-# Copyright (C) 2006-2013 Free Software Foundation, Inc.
+# Copyright (C) 2006-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -952,7 +955,7 @@ AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 
 # Check how to create a tarball.                            -*- Autoconf -*-
 
-# Copyright (C) 2004-2013 Free Software Foundation, Inc.
+# Copyright (C) 2004-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
diff --git a/config.guess b/config.guess
index 5f6aa02d..a12faba2 120000
--- a/config.guess
+++ b/config.guess
@@ -1 +1 @@
-/usr/share/automake-1.14/config.guess
\ No newline at end of file
+/opt/local/share/automake-1.15/config.guess
\ No newline at end of file
diff --git a/config.sub b/config.sub
index 0abfe18c..e3c9b5ca 120000
--- a/config.sub
+++ b/config.sub
@@ -1 +1 @@
-/usr/share/automake-1.14/config.sub
\ No newline at end of file
+/opt/local/share/automake-1.15/config.sub
\ No newline at end of file
diff --git a/configure b/configure
index b7bd49f0..6681b765 100755
--- a/configure
+++ b/configure
@@ -2467,7 +2467,7 @@ test -n "$target_alias" &&
     NONENONEs,x,x, &&
   program_prefix=${target_alias}-
 
-am__api_version='1.14'
+am__api_version='1.15'
 
 # Find a good install program.  We prefer a C program (faster),
 # so one script is as good as another.  But avoid the broken or
@@ -2639,8 +2639,8 @@ test "$program_suffix" != NONE &&
 ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
 program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
 
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 
 if test x"${MISSING+set}" != xset; then
   case $am_aux_dir in
@@ -2659,7 +2659,7 @@ else
 $as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
 fi
 
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -2987,8 +2987,8 @@ MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 mkdir_p='$(MKDIR_P)'
 
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
 # Always define AMTAR for backward compatibility.  Yes, it's still used
 # in the wild :-(  We should find a proper way to deprecate it ...
 AMTAR='$${TAR-tar}'
@@ -3047,6 +3047,7 @@ END
 fi
 
 
+
 ac_config_headers="$ac_config_headers lib/Grid_config.h"
 
 # Check whether --enable-silent-rules was given.

From 538bc41bbbf4e0c61f9bfbcbf7ab997e4a99c89f Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Wed, 27 May 2015 10:34:56 +0900
Subject: [PATCH 240/429] Adding support for doxygen generation

---
 Makefile.am                     |    9 +-
 Makefile.in                     |   23 +-
 TODO                            |    3 +
 aclocal.m4                      |    1 +
 configure                       |  183 +++
 configure.ac                    |   19 +-
 docs/doxy.cfg                   | 2305 +++++++++++++++++++++++++++++++
 docs/doxy.cfg.in                | 2305 +++++++++++++++++++++++++++++++
 docs/doxy.cfg.test              | 2305 +++++++++++++++++++++++++++++++
 lib/cshift/Grid_cshift_common.h |   10 +-
 lib/simd/Grid_vector_types.h    |    6 +-
 m4/ac_prog_doxygen.m4           |   54 +
 m4/m4-ax_prog_doxygen.m4        |  533 +++++++
 13 files changed, 7739 insertions(+), 17 deletions(-)
 create mode 100644 docs/doxy.cfg
 create mode 100644 docs/doxy.cfg.in
 create mode 100644 docs/doxy.cfg.test
 create mode 100644 m4/ac_prog_doxygen.m4
 create mode 100644 m4/m4-ax_prog_doxygen.m4

diff --git a/Makefile.am b/Makefile.am
index fae10e5d..5bb858eb 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -1,4 +1,11 @@
 # additional include paths necessary to compile the C++ library
 AM_CXXFLAGS = -I$(top_srcdir)/
-SUBDIRS = lib tests benchmarks
+SUBDIRS = lib tests benchmarks docs
 
+
+if DOXYGEN_DOC
+directory = $(top_srcdir)/docs/
+
+doxyfile:
+        (cd $(directory) && $(MAKE) $(AM_MAKEFLAGS) $@) || exit 1; 
+endif 
diff --git a/Makefile.in b/Makefile.in
index d473c2df..8c277dae 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -81,10 +81,12 @@ target_triplet = @target@
 subdir = .
 DIST_COMMON = INSTALL NEWS README AUTHORS ChangeLog \
 	$(srcdir)/Makefile.in $(srcdir)/Makefile.am \
-	$(top_srcdir)/configure $(am__configure_deps) COPYING TODO \
-	compile config.guess config.sub depcomp install-sh missing
+	$(top_srcdir)/configure $(am__configure_deps) \
+	$(top_srcdir)/docs/doxy.cfg.in COPYING TODO compile \
+	config.guess config.sub depcomp install-sh missing
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
-am__aclocal_m4_deps = $(top_srcdir)/m4/ax_cxx_compile_stdcxx_11.m4 \
+am__aclocal_m4_deps = $(top_srcdir)/m4/ac_prog_doxygen.m4 \
+	$(top_srcdir)/m4/ax_cxx_compile_stdcxx_11.m4 \
 	$(top_srcdir)/configure.ac
 am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
 	$(ACLOCAL_M4)
@@ -92,7 +94,7 @@ am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \
  configure.lineno config.status.lineno
 mkinstalldirs = $(install_sh) -d
 CONFIG_HEADER = $(top_builddir)/lib/Grid_config.h
-CONFIG_CLEAN_FILES =
+CONFIG_CLEAN_FILES = docs/doxy.cfg
 CONFIG_CLEAN_VPATH_FILES =
 AM_V_P = $(am__v_P_@AM_V@)
 am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
@@ -207,6 +209,8 @@ CXXFLAGS = @CXXFLAGS@
 CYGPATH_W = @CYGPATH_W@
 DEFS = @DEFS@
 DEPDIR = @DEPDIR@
+DOT = @DOT@
+DOXYGEN = @DOXYGEN@
 ECHO_C = @ECHO_C@
 ECHO_N = @ECHO_N@
 ECHO_T = @ECHO_T@
@@ -261,6 +265,9 @@ datadir = @datadir@
 datarootdir = @datarootdir@
 docdir = @docdir@
 dvidir = @dvidir@
+enable_dot = @enable_dot@
+enable_html_docs = @enable_html_docs@
+enable_latex_docs = @enable_latex_docs@
 exec_prefix = @exec_prefix@
 host = @host@
 host_alias = @host_alias@
@@ -297,7 +304,8 @@ top_srcdir = @top_srcdir@
 
 # additional include paths necessary to compile the C++ library
 AM_CXXFLAGS = -I$(top_srcdir)/
-SUBDIRS = lib tests benchmarks
+SUBDIRS = lib tests benchmarks docs
+@DOXYGEN_DOC_TRUE@directory = $(top_srcdir)/docs/
 all: all-recursive
 
 .SUFFIXES:
@@ -335,6 +343,8 @@ $(top_srcdir)/configure:  $(am__configure_deps)
 $(ACLOCAL_M4):  $(am__aclocal_m4_deps)
 	$(am__cd) $(srcdir) && $(ACLOCAL) $(ACLOCAL_AMFLAGS)
 $(am__aclocal_m4_deps):
+docs/doxy.cfg: $(top_builddir)/config.status $(top_srcdir)/docs/doxy.cfg.in
+	cd $(top_builddir) && $(SHELL) ./config.status $@
 
 # This directory's subdirectories are mostly independent; you can cd
 # into them and run 'make' without going through this Makefile.
@@ -752,6 +762,9 @@ uninstall-am:
 	pdf-am ps ps-am tags tags-am uninstall uninstall-am
 
 
+@DOXYGEN_DOC_TRUE@doxyfile:
+@DOXYGEN_DOC_TRUE@        (cd $(directory) && $(MAKE) $(AM_MAKEFLAGS) $@) || exit 1; 
+
 # Tell versions [3.59,3.63) of GNU make to not export all variables.
 # Otherwise a system limit (for SysV at least) may be exceeded.
 .NOEXPORT:
diff --git a/TODO b/TODO
index 670c53e3..948fb99f 100644
--- a/TODO
+++ b/TODO
@@ -64,6 +64,9 @@ Insert/Extract
   - optional parallel MPI2 IO
   - move Plaquette and link trace checks into nersc reader from the Grid_nersc_io.cc test.
 
+* Support for ILDG
+
+
 Actions -- coherent framework for implementing actions and their forces.
 
 * Fermion
diff --git a/aclocal.m4 b/aclocal.m4
index a3d1bc9c..d5fa395c 100644
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -1083,4 +1083,5 @@ AC_SUBST([am__tar])
 AC_SUBST([am__untar])
 ]) # _AM_PROG_TAR
 
+m4_include([m4/ac_prog_doxygen.m4])
 m4_include([m4/ax_cxx_compile_stdcxx_11.m4])
diff --git a/configure b/configure
index b7bd49f0..ccfea6b7 100755
--- a/configure
+++ b/configure
@@ -626,6 +626,13 @@ ac_subst_vars='am__EXEEXT_FALSE
 am__EXEEXT_TRUE
 LTLIBOBJS
 LIBOBJS
+enable_latex_docs
+enable_html_docs
+enable_dot
+DOXYGEN_DOC_FALSE
+DOXYGEN_DOC_TRUE
+DOT
+DOXYGEN
 BUILD_COMMS_NONE_FALSE
 BUILD_COMMS_NONE_TRUE
 BUILD_COMMS_MPI_FALSE
@@ -738,6 +745,10 @@ enable_dependency_tracking
 enable_openmp
 enable_simd
 enable_comms
+enable_doxygen
+enable_dot
+enable_html_docs
+enable_latex_docs
 '
       ac_precious_vars='build_alias
 host_alias
@@ -1379,6 +1390,11 @@ Optional Features:
                           Select instructions to be SSE4.0, AVX 1.0, AVX
                           2.0+FMA, AVX 512, MIC
   --enable-comms=none|mpi Select communications
+  --enable-doxygen        enable documentation generation with doxygen (auto)
+  --enable-dot            use 'dot' to generate graphs in doxygen (auto)
+  --enable-html-docs      enable HTML generation with doxygen (yes)
+  --enable-latex-docs     enable LaTeX documentation generation with doxygen
+                          (no)
 
 Some influential environment variables:
   CXX         C++ compiler command
@@ -4893,6 +4909,165 @@ fi
 
 
 
+###################################################################
+# Checks for doxygen support
+# if present enables the "make doxyfile" command
+echo
+echo Checking doxygen support
+echo :::::::::::::::::::::::::::::::::::::::::::
+
+# Check whether --enable-doxygen was given.
+if test "${enable_doxygen+set}" = set; then :
+  enableval=$enable_doxygen;
+fi
+
+
+# Check whether --enable-dot was given.
+if test "${enable_dot+set}" = set; then :
+  enableval=$enable_dot;
+fi
+
+# Check whether --enable-html-docs was given.
+if test "${enable_html_docs+set}" = set; then :
+  enableval=$enable_html_docs;
+else
+   enable_html_docs=yes
+fi
+
+# Check whether --enable-latex-docs was given.
+if test "${enable_latex_docs+set}" = set; then :
+  enableval=$enable_latex_docs;
+else
+  enable_latex_docs=no
+fi
+
+
+if test "x$enable_doxygen" = xno; then
+ enable_doc=no
+else
+ # Extract the first word of "doxygen", so it can be a program name with args.
+set dummy doxygen; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_DOXYGEN+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$DOXYGEN"; then
+  ac_cv_prog_DOXYGEN="$DOXYGEN" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_DOXYGEN="doxygen"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+DOXYGEN=$ac_cv_prog_DOXYGEN
+if test -n "$DOXYGEN"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $DOXYGEN" >&5
+$as_echo "$DOXYGEN" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+ if test x$DOXYGEN = x; then
+  if test "x$enable_doxygen" = xyes; then
+   as_fn_error $? "could not find doxygen" "$LINENO" 5
+  fi
+  enable_doc=no
+ else
+  doxy_ver=`doxygen --version`
+  doxy_major=`expr "$doxy_ver" : '\([0-9]\)\..*'`
+  doxy_minor=`expr "$doxy_ver" : '[0-9]\.\([0-9]\).*'`
+  if test $doxy_major -eq "1" -a $doxy_minor -ge "3" ; then
+   enable_doc=yes
+   # Extract the first word of "dot", so it can be a program name with args.
+set dummy dot; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_DOT+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$DOT"; then
+  ac_cv_prog_DOT="$DOT" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_DOT="dot"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+DOT=$ac_cv_prog_DOT
+if test -n "$DOT"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $DOT" >&5
+$as_echo "$DOT" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+  else
+   { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: doxygen version $doxy_ver too old, doxygen will not be used." >&5
+$as_echo "$as_me: WARNING: doxygen version $doxy_ver too old, doxygen will not be used." >&2;}
+   enable_doc=no
+  fi
+ fi
+fi
+
+ if test x$enable_doc = xyes; then
+  DOXYGEN_DOC_TRUE=
+  DOXYGEN_DOC_FALSE='#'
+else
+  DOXYGEN_DOC_TRUE='#'
+  DOXYGEN_DOC_FALSE=
+fi
+
+
+if test x$DOT = x; then
+ if test "x$enable_dot" = xyes; then
+  as_fn_error $? "could not find dot" "$LINENO" 5
+ fi
+ enable_dot=no
+else
+ enable_dot=yes
+fi
+
+
+
+
+
+
+if test -n "$DOXYGEN"
+then
+ac_config_files="$ac_config_files docs/doxy.cfg"
+
+fi
+
+
 ac_config_files="$ac_config_files Makefile"
 
 ac_config_files="$ac_config_files lib/Makefile"
@@ -5042,6 +5217,10 @@ if test -z "${BUILD_COMMS_NONE_TRUE}" && test -z "${BUILD_COMMS_NONE_FALSE}"; th
   as_fn_error $? "conditional \"BUILD_COMMS_NONE\" was never defined.
 Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
+if test -z "${DOXYGEN_DOC_TRUE}" && test -z "${DOXYGEN_DOC_FALSE}"; then
+  as_fn_error $? "conditional \"DOXYGEN_DOC\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
+fi
 
 : "${CONFIG_STATUS=./config.status}"
 ac_write_fail=0
@@ -5636,6 +5815,7 @@ do
   case $ac_config_target in
     "lib/Grid_config.h") CONFIG_HEADERS="$CONFIG_HEADERS lib/Grid_config.h" ;;
     "depfiles") CONFIG_COMMANDS="$CONFIG_COMMANDS depfiles" ;;
+    "docs/doxy.cfg") CONFIG_FILES="$CONFIG_FILES docs/doxy.cfg" ;;
     "Makefile") CONFIG_FILES="$CONFIG_FILES Makefile" ;;
     "lib/Makefile") CONFIG_FILES="$CONFIG_FILES lib/Makefile" ;;
     "tests/Makefile") CONFIG_FILES="$CONFIG_FILES tests/Makefile" ;;
@@ -6379,6 +6559,9 @@ The following features are enabled:
 - os (build)                    : $build_os
 - architecture (target)         : $target_cpu
 - os (target)                   : $target_os
+- build DOXYGEN documentation   : `if test "x$enable_doc" = xyes; then echo yes; else echo no; fi`
+- graphs and diagrams           : `if test "x$enable_dot" = xyes; then echo yes; else echo no; fi`
+
 ----------------------------------------------------------
 - enabled simd support          : ${ac_SIMD}
 - communications type           : ${ac_COMMS}
diff --git a/configure.ac b/configure.ac
index 54a36eb1..5dcbea36 100644
--- a/configure.ac
+++ b/configure.ac
@@ -3,7 +3,7 @@
 #
 # Project Grid package  
 # 
-# Time-stamp: <2015-05-25 14:54:34 neo>
+# Time-stamp: <2015-05-26 17:18:54 neo>
 
 AC_PREREQ([2.63])
 AC_INIT([Grid], [1.0], [paboyle@ph.ed.ac.uk])
@@ -116,6 +116,20 @@ AM_CONDITIONAL(BUILD_COMMS_MPI,[ test "X${ac_COMMS}X" == "XmpiX" ])
 AM_CONDITIONAL(BUILD_COMMS_NONE,[ test "X${ac_COMMS}X" == "XnoneX" ])
 
 
+###################################################################
+# Checks for doxygen support
+# if present enables the "make doxyfile" command
+echo
+echo Checking doxygen support 
+echo :::::::::::::::::::::::::::::::::::::::::::
+AC_PROG_DOXYGEN
+
+if test -n "$DOXYGEN"
+then
+AC_CONFIG_FILES([docs/doxy.cfg])
+fi
+
+
 AC_CONFIG_FILES(Makefile)
 AC_CONFIG_FILES(lib/Makefile)
 AC_CONFIG_FILES(tests/Makefile)
@@ -134,6 +148,9 @@ The following features are enabled:
 - os (build)                    : $build_os
 - architecture (target)         : $target_cpu
 - os (target)                   : $target_os
+- build DOXYGEN documentation   : `if test "x$enable_doc" = xyes; then echo yes; else echo no; fi`
+- graphs and diagrams           : `if test "x$enable_dot" = xyes; then echo yes; else echo no; fi`
+
 ----------------------------------------------------------
 - enabled simd support          : ${ac_SIMD}
 - communications type           : ${ac_COMMS}
diff --git a/docs/doxy.cfg b/docs/doxy.cfg
new file mode 100644
index 00000000..c4b82094
--- /dev/null
+++ b/docs/doxy.cfg
@@ -0,0 +1,2305 @@
+# Doxyfile 1.8.6
+
+# This file describes the settings to be used by the documentation system
+# doxygen (www.doxygen.org) for a project.
+#
+# All text after a double hash (##) is considered a comment and is placed in
+# front of the TAG it is preceding.
+#
+# All text after a single hash (#) is considered a comment and will be ignored.
+# The format is:
+# TAG = value [value, ...]
+# For lists, items can also be appended using:
+# TAG += value [value, ...]
+# Values that contain spaces should be placed between quotes (\" \").
+
+#---------------------------------------------------------------------------
+# Project related configuration options
+#---------------------------------------------------------------------------
+
+# This tag specifies the encoding used for all characters in the config file
+# that follow. The default is UTF-8 which is also the encoding used for all text
+# before the first occurrence of this tag. Doxygen uses libiconv (or the iconv
+# built into libc) for the transcoding. See http://www.gnu.org/software/libiconv
+# for the list of possible encodings.
+# The default value is: UTF-8.
+
+DOXYFILE_ENCODING      = UTF-8
+
+# The PROJECT_NAME tag is a single word (or a sequence of words surrounded by
+# double-quotes, unless you are using Doxywizard) that should identify the
+# project for which the documentation is generated. This name is used in the
+# title of most generated pages and in a few other places.
+# The default value is: My Project.
+
+PROJECT_NAME           = "Grid"
+
+# The PROJECT_NUMBER tag can be used to enter a project or revision number. This
+# could be handy for archiving the generated documentation or if some version
+# control system is used.
+
+PROJECT_NUMBER         = 1.0
+
+# Using the PROJECT_BRIEF tag one can provide an optional one line description
+# for a project that appears at the top of each page and should give viewer a
+# quick idea about the purpose of the project. Keep the description short.
+
+PROJECT_BRIEF          =
+
+# With the PROJECT_LOGO tag one can specify an logo or icon that is included in
+# the documentation. The maximum height of the logo should not exceed 55 pixels
+# and the maximum width should not exceed 200 pixels. Doxygen will copy the logo
+# to the output directory.
+
+PROJECT_LOGO           =
+
+# The OUTPUT_DIRECTORY tag is used to specify the (relative or absolute) path
+# into which the generated documentation will be written. If a relative path is
+# entered, it will be relative to the location where doxygen was started. If
+# left blank the current directory will be used.
+
+OUTPUT_DIRECTORY       = ./doxy-en/
+
+# If the CREATE_SUBDIRS tag is set to YES, then doxygen will create 4096 sub-
+# directories (in 2 levels) under the output directory of each output format and
+# will distribute the generated files over these directories. Enabling this
+# option can be useful when feeding doxygen a huge amount of source files, where
+# putting all generated files in the same directory would otherwise causes
+# performance problems for the file system.
+# The default value is: NO.
+
+CREATE_SUBDIRS         = NO
+
+# The OUTPUT_LANGUAGE tag is used to specify the language in which all
+# documentation generated by doxygen is written. Doxygen will use this
+# information to generate all constant output in the proper language.
+# Possible values are: Afrikaans, Arabic, Armenian, Brazilian, Catalan, Chinese,
+# Chinese-Traditional, Croatian, Czech, Danish, Dutch, English (United States),
+# Esperanto, Farsi (Persian), Finnish, French, German, Greek, Hungarian,
+# Indonesian, Italian, Japanese, Japanese-en (Japanese with English messages),
+# Korean, Korean-en (Korean with English messages), Latvian, Lithuanian,
+# Macedonian, Norwegian, Persian (Farsi), Polish, Portuguese, Romanian, Russian,
+# Serbian, Serbian-Cyrillic, Slovak, Slovene, Spanish, Swedish, Turkish,
+# Ukrainian and Vietnamese.
+# The default value is: English.
+
+OUTPUT_LANGUAGE        = English
+
+# If the BRIEF_MEMBER_DESC tag is set to YES doxygen will include brief member
+# descriptions after the members that are listed in the file and class
+# documentation (similar to Javadoc). Set to NO to disable this.
+# The default value is: YES.
+
+BRIEF_MEMBER_DESC      = YES
+
+# If the REPEAT_BRIEF tag is set to YES doxygen will prepend the brief
+# description of a member or function before the detailed description
+#
+# Note: If both HIDE_UNDOC_MEMBERS and BRIEF_MEMBER_DESC are set to NO, the
+# brief descriptions will be completely suppressed.
+# The default value is: YES.
+
+REPEAT_BRIEF           = YES
+
+# This tag implements a quasi-intelligent brief description abbreviator that is
+# used to form the text in various listings. Each string in this list, if found
+# as the leading text of the brief description, will be stripped from the text
+# and the result, after processing the whole list, is used as the annotated
+# text. Otherwise, the brief description is used as-is. If left blank, the
+# following values are used ($name is automatically replaced with the name of
+# the entity):The $name class, The $name widget, The $name file, is, provides,
+# specifies, contains, represents, a, an and the.
+
+ABBREVIATE_BRIEF       =
+
+# If the ALWAYS_DETAILED_SEC and REPEAT_BRIEF tags are both set to YES then
+# doxygen will generate a detailed section even if there is only a brief
+# description.
+# The default value is: NO.
+
+ALWAYS_DETAILED_SEC    = NO
+
+# If the INLINE_INHERITED_MEMB tag is set to YES, doxygen will show all
+# inherited members of a class in the documentation of that class as if those
+# members were ordinary class members. Constructors, destructors and assignment
+# operators of the base classes will not be shown.
+# The default value is: NO.
+
+INLINE_INHERITED_MEMB  = NO
+
+# If the FULL_PATH_NAMES tag is set to YES doxygen will prepend the full path
+# before files name in the file list and in the header files. If set to NO the
+# shortest path that makes the file name unique will be used
+# The default value is: YES.
+
+FULL_PATH_NAMES        = YES
+
+# The STRIP_FROM_PATH tag can be used to strip a user-defined part of the path.
+# Stripping is only done if one of the specified strings matches the left-hand
+# part of the path. The tag can be used to show relative paths in the file list.
+# If left blank the directory from which doxygen is run is used as the path to
+# strip.
+#
+# Note that you can specify absolute paths here, but also relative paths, which
+# will be relative from the directory where doxygen is started.
+# This tag requires that the tag FULL_PATH_NAMES is set to YES.
+
+STRIP_FROM_PATH        =
+
+# The STRIP_FROM_INC_PATH tag can be used to strip a user-defined part of the
+# path mentioned in the documentation of a class, which tells the reader which
+# header file to include in order to use a class. If left blank only the name of
+# the header file containing the class definition is used. Otherwise one should
+# specify the list of include paths that are normally passed to the compiler
+# using the -I flag.
+
+STRIP_FROM_INC_PATH    =
+
+# If the SHORT_NAMES tag is set to YES, doxygen will generate much shorter (but
+# less readable) file names. This can be useful is your file systems doesn't
+# support long names like on DOS, Mac, or CD-ROM.
+# The default value is: NO.
+
+SHORT_NAMES            = NO
+
+# If the JAVADOC_AUTOBRIEF tag is set to YES then doxygen will interpret the
+# first line (until the first dot) of a Javadoc-style comment as the brief
+# description. If set to NO, the Javadoc-style will behave just like regular Qt-
+# style comments (thus requiring an explicit @brief command for a brief
+# description.)
+# The default value is: NO.
+
+JAVADOC_AUTOBRIEF      = NO
+
+# If the QT_AUTOBRIEF tag is set to YES then doxygen will interpret the first
+# line (until the first dot) of a Qt-style comment as the brief description. If
+# set to NO, the Qt-style will behave just like regular Qt-style comments (thus
+# requiring an explicit \brief command for a brief description.)
+# The default value is: NO.
+
+QT_AUTOBRIEF           = NO
+
+# The MULTILINE_CPP_IS_BRIEF tag can be set to YES to make doxygen treat a
+# multi-line C++ special comment block (i.e. a block of //! or /// comments) as
+# a brief description. This used to be the default behavior. The new default is
+# to treat a multi-line C++ comment block as a detailed description. Set this
+# tag to YES if you prefer the old behavior instead.
+#
+# Note that setting this tag to YES also means that rational rose comments are
+# not recognized any more.
+# The default value is: NO.
+
+MULTILINE_CPP_IS_BRIEF = NO
+
+# If the INHERIT_DOCS tag is set to YES then an undocumented member inherits the
+# documentation from any documented member that it re-implements.
+# The default value is: YES.
+
+INHERIT_DOCS           = YES
+
+# If the SEPARATE_MEMBER_PAGES tag is set to YES, then doxygen will produce a
+# new page for each member. If set to NO, the documentation of a member will be
+# part of the file/class/namespace that contains it.
+# The default value is: NO.
+
+SEPARATE_MEMBER_PAGES  = NO
+
+# The TAB_SIZE tag can be used to set the number of spaces in a tab. Doxygen
+# uses this value to replace tabs by spaces in code fragments.
+# Minimum value: 1, maximum value: 16, default value: 4.
+
+TAB_SIZE               = 4
+
+# This tag can be used to specify a number of aliases that act as commands in
+# the documentation. An alias has the form:
+# name=value
+# For example adding
+# "sideeffect=@par Side Effects:\n"
+# will allow you to put the command \sideeffect (or @sideeffect) in the
+# documentation, which will result in a user-defined paragraph with heading
+# "Side Effects:". You can put \n's in the value part of an alias to insert
+# newlines.
+
+ALIASES                =
+
+# This tag can be used to specify a number of word-keyword mappings (TCL only).
+# A mapping has the form "name=value". For example adding "class=itcl::class"
+# will allow you to use the command class in the itcl::class meaning.
+
+TCL_SUBST              =
+
+# Set the OPTIMIZE_OUTPUT_FOR_C tag to YES if your project consists of C sources
+# only. Doxygen will then generate output that is more tailored for C. For
+# instance, some of the names that are used will be different. The list of all
+# members will be omitted, etc.
+# The default value is: NO.
+
+OPTIMIZE_OUTPUT_FOR_C  = NO
+
+# Set the OPTIMIZE_OUTPUT_JAVA tag to YES if your project consists of Java or
+# Python sources only. Doxygen will then generate output that is more tailored
+# for that language. For instance, namespaces will be presented as packages,
+# qualified scopes will look different, etc.
+# The default value is: NO.
+
+OPTIMIZE_OUTPUT_JAVA   = NO
+
+# Set the OPTIMIZE_FOR_FORTRAN tag to YES if your project consists of Fortran
+# sources. Doxygen will then generate output that is tailored for Fortran.
+# The default value is: NO.
+
+OPTIMIZE_FOR_FORTRAN   = NO
+
+# Set the OPTIMIZE_OUTPUT_VHDL tag to YES if your project consists of VHDL
+# sources. Doxygen will then generate output that is tailored for VHDL.
+# The default value is: NO.
+
+OPTIMIZE_OUTPUT_VHDL   = NO
+
+# Doxygen selects the parser to use depending on the extension of the files it
+# parses. With this tag you can assign which parser to use for a given
+# extension. Doxygen has a built-in mapping, but you can override or extend it
+# using this tag. The format is ext=language, where ext is a file extension, and
+# language is one of the parsers supported by doxygen: IDL, Java, Javascript,
+# C#, C, C++, D, PHP, Objective-C, Python, Fortran, VHDL. For instance to make
+# doxygen treat .inc files as Fortran files (default is PHP), and .f files as C
+# (default is Fortran), use: inc=Fortran f=C.
+#
+# Note For files without extension you can use no_extension as a placeholder.
+#
+# Note that for custom extensions you also need to set FILE_PATTERNS otherwise
+# the files are not read by doxygen.
+
+EXTENSION_MAPPING      =
+
+# If the MARKDOWN_SUPPORT tag is enabled then doxygen pre-processes all comments
+# according to the Markdown format, which allows for more readable
+# documentation. See http://daringfireball.net/projects/markdown/ for details.
+# The output of markdown processing is further processed by doxygen, so you can
+# mix doxygen, HTML, and XML commands with Markdown formatting. Disable only in
+# case of backward compatibilities issues.
+# The default value is: YES.
+
+MARKDOWN_SUPPORT       = YES
+
+# When enabled doxygen tries to link words that correspond to documented
+# classes, or namespaces to their corresponding documentation. Such a link can
+# be prevented in individual cases by by putting a % sign in front of the word
+# or globally by setting AUTOLINK_SUPPORT to NO.
+# The default value is: YES.
+
+AUTOLINK_SUPPORT       = YES
+
+# If you use STL classes (i.e. std::string, std::vector, etc.) but do not want
+# to include (a tag file for) the STL sources as input, then you should set this
+# tag to YES in order to let doxygen match functions declarations and
+# definitions whose arguments contain STL classes (e.g. func(std::string);
+# versus func(std::string) {}). This also make the inheritance and collaboration
+# diagrams that involve STL classes more complete and accurate.
+# The default value is: NO.
+
+BUILTIN_STL_SUPPORT    = YES
+
+# If you use Microsoft's C++/CLI language, you should set this option to YES to
+# enable parsing support.
+# The default value is: NO.
+
+CPP_CLI_SUPPORT        = NO
+
+# Set the SIP_SUPPORT tag to YES if your project consists of sip (see:
+# http://www.riverbankcomputing.co.uk/software/sip/intro) sources only. Doxygen
+# will parse them like normal C++ but will assume all classes use public instead
+# of private inheritance when no explicit protection keyword is present.
+# The default value is: NO.
+
+SIP_SUPPORT            = NO
+
+# For Microsoft's IDL there are propget and propput attributes to indicate
+# getter and setter methods for a property. Setting this option to YES will make
+# doxygen to replace the get and set methods by a property in the documentation.
+# This will only work if the methods are indeed getting or setting a simple
+# type. If this is not the case, or you want to show the methods anyway, you
+# should set this option to NO.
+# The default value is: YES.
+
+IDL_PROPERTY_SUPPORT   = YES
+
+# If member grouping is used in the documentation and the DISTRIBUTE_GROUP_DOC
+# tag is set to YES, then doxygen will reuse the documentation of the first
+# member in the group (if any) for the other members of the group. By default
+# all members of a group must be documented explicitly.
+# The default value is: NO.
+
+DISTRIBUTE_GROUP_DOC   = NO
+
+# Set the SUBGROUPING tag to YES to allow class member groups of the same type
+# (for instance a group of public functions) to be put as a subgroup of that
+# type (e.g. under the Public Functions section). Set it to NO to prevent
+# subgrouping. Alternatively, this can be done per class using the
+# \nosubgrouping command.
+# The default value is: YES.
+
+SUBGROUPING            = YES
+
+# When the INLINE_GROUPED_CLASSES tag is set to YES, classes, structs and unions
+# are shown inside the group in which they are included (e.g. using \ingroup)
+# instead of on a separate page (for HTML and Man pages) or section (for LaTeX
+# and RTF).
+#
+# Note that this feature does not work in combination with
+# SEPARATE_MEMBER_PAGES.
+# The default value is: NO.
+
+INLINE_GROUPED_CLASSES = NO
+
+# When the INLINE_SIMPLE_STRUCTS tag is set to YES, structs, classes, and unions
+# with only public data fields or simple typedef fields will be shown inline in
+# the documentation of the scope in which they are defined (i.e. file,
+# namespace, or group documentation), provided this scope is documented. If set
+# to NO, structs, classes, and unions are shown on a separate page (for HTML and
+# Man pages) or section (for LaTeX and RTF).
+# The default value is: NO.
+
+INLINE_SIMPLE_STRUCTS  = NO
+
+# When TYPEDEF_HIDES_STRUCT tag is enabled, a typedef of a struct, union, or
+# enum is documented as struct, union, or enum with the name of the typedef. So
+# typedef struct TypeS {} TypeT, will appear in the documentation as a struct
+# with name TypeT. When disabled the typedef will appear as a member of a file,
+# namespace, or class. And the struct will be named TypeS. This can typically be
+# useful for C code in case the coding convention dictates that all compound
+# types are typedef'ed and only the typedef is referenced, never the tag name.
+# The default value is: NO.
+
+TYPEDEF_HIDES_STRUCT   = NO
+
+# The size of the symbol lookup cache can be set using LOOKUP_CACHE_SIZE. This
+# cache is used to resolve symbols given their name and scope. Since this can be
+# an expensive process and often the same symbol appears multiple times in the
+# code, doxygen keeps a cache of pre-resolved symbols. If the cache is too small
+# doxygen will become slower. If the cache is too large, memory is wasted. The
+# cache size is given by this formula: 2^(16+LOOKUP_CACHE_SIZE). The valid range
+# is 0..9, the default is 0, corresponding to a cache size of 2^16=65536
+# symbols. At the end of a run doxygen will report the cache usage and suggest
+# the optimal cache size from a speed point of view.
+# Minimum value: 0, maximum value: 9, default value: 0.
+
+LOOKUP_CACHE_SIZE      = 0
+
+#---------------------------------------------------------------------------
+# Build related configuration options
+#---------------------------------------------------------------------------
+
+# If the EXTRACT_ALL tag is set to YES doxygen will assume all entities in
+# documentation are documented, even if no documentation was available. Private
+# class members and static file members will be hidden unless the
+# EXTRACT_PRIVATE respectively EXTRACT_STATIC tags are set to YES.
+# Note: This will also disable the warnings about undocumented members that are
+# normally produced when WARNINGS is set to YES.
+# The default value is: NO.
+
+EXTRACT_ALL            = YES
+
+# If the EXTRACT_PRIVATE tag is set to YES all private members of a class will
+# be included in the documentation.
+# The default value is: NO.
+
+EXTRACT_PRIVATE        = YES
+
+# If the EXTRACT_PACKAGE tag is set to YES all members with package or internal
+# scope will be included in the documentation.
+# The default value is: NO.
+
+EXTRACT_PACKAGE        = YES
+
+# If the EXTRACT_STATIC tag is set to YES all static members of a file will be
+# included in the documentation.
+# The default value is: NO.
+
+EXTRACT_STATIC         = YES
+
+# If the EXTRACT_LOCAL_CLASSES tag is set to YES classes (and structs) defined
+# locally in source files will be included in the documentation. If set to NO
+# only classes defined in header files are included. Does not have any effect
+# for Java sources.
+# The default value is: YES.
+
+EXTRACT_LOCAL_CLASSES  = YES
+
+# This flag is only useful for Objective-C code. When set to YES local methods,
+# which are defined in the implementation section but not in the interface are
+# included in the documentation. If set to NO only methods in the interface are
+# included.
+# The default value is: NO.
+
+EXTRACT_LOCAL_METHODS  = YES
+
+# If this flag is set to YES, the members of anonymous namespaces will be
+# extracted and appear in the documentation as a namespace called
+# 'anonymous_namespace{file}', where file will be replaced with the base name of
+# the file that contains the anonymous namespace. By default anonymous namespace
+# are hidden.
+# The default value is: NO.
+
+EXTRACT_ANON_NSPACES   = YES
+
+# If the HIDE_UNDOC_MEMBERS tag is set to YES, doxygen will hide all
+# undocumented members inside documented classes or files. If set to NO these
+# members will be included in the various overviews, but no documentation
+# section is generated. This option has no effect if EXTRACT_ALL is enabled.
+# The default value is: NO.
+
+HIDE_UNDOC_MEMBERS     = NO
+
+# If the HIDE_UNDOC_CLASSES tag is set to YES, doxygen will hide all
+# undocumented classes that are normally visible in the class hierarchy. If set
+# to NO these classes will be included in the various overviews. This option has
+# no effect if EXTRACT_ALL is enabled.
+# The default value is: NO.
+
+HIDE_UNDOC_CLASSES     = NO
+
+# If the HIDE_FRIEND_COMPOUNDS tag is set to YES, doxygen will hide all friend
+# (class|struct|union) declarations. If set to NO these declarations will be
+# included in the documentation.
+# The default value is: NO.
+
+HIDE_FRIEND_COMPOUNDS  = NO
+
+# If the HIDE_IN_BODY_DOCS tag is set to YES, doxygen will hide any
+# documentation blocks found inside the body of a function. If set to NO these
+# blocks will be appended to the function's detailed documentation block.
+# The default value is: NO.
+
+HIDE_IN_BODY_DOCS      = NO
+
+# The INTERNAL_DOCS tag determines if documentation that is typed after a
+# \internal command is included. If the tag is set to NO then the documentation
+# will be excluded. Set it to YES to include the internal documentation.
+# The default value is: NO.
+
+INTERNAL_DOCS          = NO
+
+# If the CASE_SENSE_NAMES tag is set to NO then doxygen will only generate file
+# names in lower-case letters. If set to YES upper-case letters are also
+# allowed. This is useful if you have classes or files whose names only differ
+# in case and if your file system supports case sensitive file names. Windows
+# and Mac users are advised to set this option to NO.
+# The default value is: system dependent.
+
+CASE_SENSE_NAMES       = YES
+
+# If the HIDE_SCOPE_NAMES tag is set to NO then doxygen will show members with
+# their full class and namespace scopes in the documentation. If set to YES the
+# scope will be hidden.
+# The default value is: NO.
+
+HIDE_SCOPE_NAMES       = NO
+
+# If the SHOW_INCLUDE_FILES tag is set to YES then doxygen will put a list of
+# the files that are included by a file in the documentation of that file.
+# The default value is: YES.
+
+SHOW_INCLUDE_FILES     = YES
+
+# If the SHOW_GROUPED_MEMB_INC tag is set to YES then Doxygen will add for each
+# grouped member an include statement to the documentation, telling the reader
+# which file to include in order to use the member.
+# The default value is: NO.
+
+SHOW_GROUPED_MEMB_INC  = NO
+
+# If the FORCE_LOCAL_INCLUDES tag is set to YES then doxygen will list include
+# files with double quotes in the documentation rather than with sharp brackets.
+# The default value is: NO.
+
+FORCE_LOCAL_INCLUDES   = NO
+
+# If the INLINE_INFO tag is set to YES then a tag [inline] is inserted in the
+# documentation for inline members.
+# The default value is: YES.
+
+INLINE_INFO            = YES
+
+# If the SORT_MEMBER_DOCS tag is set to YES then doxygen will sort the
+# (detailed) documentation of file and class members alphabetically by member
+# name. If set to NO the members will appear in declaration order.
+# The default value is: YES.
+
+SORT_MEMBER_DOCS       = YES
+
+# If the SORT_BRIEF_DOCS tag is set to YES then doxygen will sort the brief
+# descriptions of file, namespace and class members alphabetically by member
+# name. If set to NO the members will appear in declaration order. Note that
+# this will also influence the order of the classes in the class list.
+# The default value is: NO.
+
+SORT_BRIEF_DOCS        = NO
+
+# If the SORT_MEMBERS_CTORS_1ST tag is set to YES then doxygen will sort the
+# (brief and detailed) documentation of class members so that constructors and
+# destructors are listed first. If set to NO the constructors will appear in the
+# respective orders defined by SORT_BRIEF_DOCS and SORT_MEMBER_DOCS.
+# Note: If SORT_BRIEF_DOCS is set to NO this option is ignored for sorting brief
+# member documentation.
+# Note: If SORT_MEMBER_DOCS is set to NO this option is ignored for sorting
+# detailed member documentation.
+# The default value is: NO.
+
+SORT_MEMBERS_CTORS_1ST = YES
+
+# If the SORT_GROUP_NAMES tag is set to YES then doxygen will sort the hierarchy
+# of group names into alphabetical order. If set to NO the group names will
+# appear in their defined order.
+# The default value is: NO.
+
+SORT_GROUP_NAMES       = NO
+
+# If the SORT_BY_SCOPE_NAME tag is set to YES, the class list will be sorted by
+# fully-qualified names, including namespaces. If set to NO, the class list will
+# be sorted only by class name, not including the namespace part.
+# Note: This option is not very useful if HIDE_SCOPE_NAMES is set to YES.
+# Note: This option applies only to the class list, not to the alphabetical
+# list.
+# The default value is: NO.
+
+SORT_BY_SCOPE_NAME     = NO
+
+# If the STRICT_PROTO_MATCHING option is enabled and doxygen fails to do proper
+# type resolution of all parameters of a function it will reject a match between
+# the prototype and the implementation of a member function even if there is
+# only one candidate or it is obvious which candidate to choose by doing a
+# simple string match. By disabling STRICT_PROTO_MATCHING doxygen will still
+# accept a match between prototype and implementation in such cases.
+# The default value is: NO.
+
+STRICT_PROTO_MATCHING  = NO
+
+# The GENERATE_TODOLIST tag can be used to enable ( YES) or disable ( NO) the
+# todo list. This list is created by putting \todo commands in the
+# documentation.
+# The default value is: YES.
+
+GENERATE_TODOLIST      = YES
+
+# The GENERATE_TESTLIST tag can be used to enable ( YES) or disable ( NO) the
+# test list. This list is created by putting \test commands in the
+# documentation.
+# The default value is: YES.
+
+GENERATE_TESTLIST      = YES
+
+# The GENERATE_BUGLIST tag can be used to enable ( YES) or disable ( NO) the bug
+# list. This list is created by putting \bug commands in the documentation.
+# The default value is: YES.
+
+GENERATE_BUGLIST       = YES
+
+# The GENERATE_DEPRECATEDLIST tag can be used to enable ( YES) or disable ( NO)
+# the deprecated list. This list is created by putting \deprecated commands in
+# the documentation.
+# The default value is: YES.
+
+GENERATE_DEPRECATEDLIST= YES
+
+# The ENABLED_SECTIONS tag can be used to enable conditional documentation
+# sections, marked by \if <section_label> ... \endif and \cond <section_label>
+# ... \endcond blocks.
+
+ENABLED_SECTIONS       =
+
+# The MAX_INITIALIZER_LINES tag determines the maximum number of lines that the
+# initial value of a variable or macro / define can have for it to appear in the
+# documentation. If the initializer consists of more lines than specified here
+# it will be hidden. Use a value of 0 to hide initializers completely. The
+# appearance of the value of individual variables and macros / defines can be
+# controlled using \showinitializer or \hideinitializer command in the
+# documentation regardless of this setting.
+# Minimum value: 0, maximum value: 10000, default value: 30.
+
+MAX_INITIALIZER_LINES  = 30
+
+# Set the SHOW_USED_FILES tag to NO to disable the list of files generated at
+# the bottom of the documentation of classes and structs. If set to YES the list
+# will mention the files that were used to generate the documentation.
+# The default value is: YES.
+
+SHOW_USED_FILES        = YES
+
+# Set the SHOW_FILES tag to NO to disable the generation of the Files page. This
+# will remove the Files entry from the Quick Index and from the Folder Tree View
+# (if specified).
+# The default value is: YES.
+
+SHOW_FILES             = YES
+
+# Set the SHOW_NAMESPACES tag to NO to disable the generation of the Namespaces
+# page. This will remove the Namespaces entry from the Quick Index and from the
+# Folder Tree View (if specified).
+# The default value is: YES.
+
+SHOW_NAMESPACES        = YES
+
+# The FILE_VERSION_FILTER tag can be used to specify a program or script that
+# doxygen should invoke to get the current version for each file (typically from
+# the version control system). Doxygen will invoke the program by executing (via
+# popen()) the command command input-file, where command is the value of the
+# FILE_VERSION_FILTER tag, and input-file is the name of an input file provided
+# by doxygen. Whatever the program writes to standard output is used as the file
+# version. For an example see the documentation.
+
+FILE_VERSION_FILTER    =
+
+# The LAYOUT_FILE tag can be used to specify a layout file which will be parsed
+# by doxygen. The layout file controls the global structure of the generated
+# output files in an output format independent way. To create the layout file
+# that represents doxygen's defaults, run doxygen with the -l option. You can
+# optionally specify a file name after the option, if omitted DoxygenLayout.xml
+# will be used as the name of the layout file.
+#
+# Note that if you run doxygen from a directory containing a file called
+# DoxygenLayout.xml, doxygen will parse it automatically even if the LAYOUT_FILE
+# tag is left empty.
+
+LAYOUT_FILE            =
+
+# The CITE_BIB_FILES tag can be used to specify one or more bib files containing
+# the reference definitions. This must be a list of .bib files. The .bib
+# extension is automatically appended if omitted. This requires the bibtex tool
+# to be installed. See also http://en.wikipedia.org/wiki/BibTeX for more info.
+# For LaTeX the style of the bibliography can be controlled using
+# LATEX_BIB_STYLE. To use this feature you need bibtex and perl available in the
+# search path. Do not use file names with spaces, bibtex cannot handle them. See
+# also \cite for info how to create references.
+
+CITE_BIB_FILES         =
+
+#---------------------------------------------------------------------------
+# Configuration options related to warning and progress messages
+#---------------------------------------------------------------------------
+
+# The QUIET tag can be used to turn on/off the messages that are generated to
+# standard output by doxygen. If QUIET is set to YES this implies that the
+# messages are off.
+# The default value is: NO.
+
+QUIET                  = NO
+
+# The WARNINGS tag can be used to turn on/off the warning messages that are
+# generated to standard error ( stderr) by doxygen. If WARNINGS is set to YES
+# this implies that the warnings are on.
+#
+# Tip: Turn warnings on while writing the documentation.
+# The default value is: YES.
+
+WARNINGS               = YES
+
+# If the WARN_IF_UNDOCUMENTED tag is set to YES, then doxygen will generate
+# warnings for undocumented members. If EXTRACT_ALL is set to YES then this flag
+# will automatically be disabled.
+# The default value is: YES.
+
+WARN_IF_UNDOCUMENTED   = YES
+
+# If the WARN_IF_DOC_ERROR tag is set to YES, doxygen will generate warnings for
+# potential errors in the documentation, such as not documenting some parameters
+# in a documented function, or documenting parameters that don't exist or using
+# markup commands wrongly.
+# The default value is: YES.
+
+WARN_IF_DOC_ERROR      = YES
+
+# This WARN_NO_PARAMDOC option can be enabled to get warnings for functions that
+# are documented, but have no documentation for their parameters or return
+# value. If set to NO doxygen will only warn about wrong or incomplete parameter
+# documentation, but not about the absence of documentation.
+# The default value is: NO.
+
+WARN_NO_PARAMDOC       = NO
+
+# The WARN_FORMAT tag determines the format of the warning messages that doxygen
+# can produce. The string should contain the $file, $line, and $text tags, which
+# will be replaced by the file and line number from which the warning originated
+# and the warning text. Optionally the format may contain $version, which will
+# be replaced by the version of the file (if it could be obtained via
+# FILE_VERSION_FILTER)
+# The default value is: $file:$line: $text.
+
+WARN_FORMAT            = "$file:$line: $text"
+
+# The WARN_LOGFILE tag can be used to specify a file to which warning and error
+# messages should be written. If left blank the output is written to standard
+# error (stderr).
+
+WARN_LOGFILE           =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the input files
+#---------------------------------------------------------------------------
+
+# The INPUT tag is used to specify the files and/or directories that contain
+# documented source files. You may enter file names like myfile.cpp or
+# directories like /usr/src/myproject. Separate the files or directories with
+# spaces.
+# Note: If this tag is empty the current directory is searched.
+
+INPUT                  = ../lib \
+		         ../tests \
+			 ../benchmarks
+
+# This tag can be used to specify the character encoding of the source files
+# that doxygen parses. Internally doxygen uses the UTF-8 encoding. Doxygen uses
+# libiconv (or the iconv built into libc) for the transcoding. See the libiconv
+# documentation (see: http://www.gnu.org/software/libiconv) for the list of
+# possible encodings.
+# The default value is: UTF-8.
+
+INPUT_ENCODING         = UTF-8
+
+# If the value of the INPUT tag contains directories, you can use the
+# FILE_PATTERNS tag to specify one or more wildcard patterns (like *.cpp and
+# *.h) to filter out the source-files in the directories. If left blank the
+# following patterns are tested:*.c, *.cc, *.cxx, *.cpp, *.c++, *.java, *.ii,
+# *.ixx, *.ipp, *.i++, *.inl, *.idl, *.ddl, *.odl, *.h, *.hh, *.hxx, *.hpp,
+# *.h++, *.cs, *.d, *.php, *.php4, *.php5, *.phtml, *.inc, *.m, *.markdown,
+# *.md, *.mm, *.dox, *.py, *.f90, *.f, *.for, *.tcl, *.vhd, *.vhdl, *.ucf,
+# *.qsf, *.as and *.js.
+
+FILE_PATTERNS          =
+
+# The RECURSIVE tag can be used to specify whether or not subdirectories should
+# be searched for input files as well.
+# The default value is: NO.
+
+RECURSIVE              = YES
+
+# The EXCLUDE tag can be used to specify files and/or directories that should be
+# excluded from the INPUT source files. This way you can easily exclude a
+# subdirectory from a directory tree whose root is specified with the INPUT tag.
+#
+# Note that relative paths are relative to the directory from which doxygen is
+# run.
+
+EXCLUDE                =
+
+# The EXCLUDE_SYMLINKS tag can be used to select whether or not files or
+# directories that are symbolic links (a Unix file system feature) are excluded
+# from the input.
+# The default value is: NO.
+
+EXCLUDE_SYMLINKS       = NO
+
+# If the value of the INPUT tag contains directories, you can use the
+# EXCLUDE_PATTERNS tag to specify one or more wildcard patterns to exclude
+# certain files from those directories.
+#
+# Note that the wildcards are matched against the file with absolute path, so to
+# exclude all test directories for example use the pattern */test/*
+
+EXCLUDE_PATTERNS       =
+
+# The EXCLUDE_SYMBOLS tag can be used to specify one or more symbol names
+# (namespaces, classes, functions, etc.) that should be excluded from the
+# output. The symbol name can be a fully qualified name, a word, or if the
+# wildcard * is used, a substring. Examples: ANamespace, AClass,
+# AClass::ANamespace, ANamespace::*Test
+#
+# Note that the wildcards are matched against the file with absolute path, so to
+# exclude all test directories use the pattern */test/*
+
+EXCLUDE_SYMBOLS        =
+
+# The EXAMPLE_PATH tag can be used to specify one or more files or directories
+# that contain example code fragments that are included (see the \include
+# command).
+
+EXAMPLE_PATH           =
+
+# If the value of the EXAMPLE_PATH tag contains directories, you can use the
+# EXAMPLE_PATTERNS tag to specify one or more wildcard pattern (like *.cpp and
+# *.h) to filter out the source-files in the directories. If left blank all
+# files are included.
+
+EXAMPLE_PATTERNS       =
+
+# If the EXAMPLE_RECURSIVE tag is set to YES then subdirectories will be
+# searched for input files to be used with the \include or \dontinclude commands
+# irrespective of the value of the RECURSIVE tag.
+# The default value is: NO.
+
+EXAMPLE_RECURSIVE      = NO
+
+# The IMAGE_PATH tag can be used to specify one or more files or directories
+# that contain images that are to be included in the documentation (see the
+# \image command).
+
+IMAGE_PATH             =
+
+# The INPUT_FILTER tag can be used to specify a program that doxygen should
+# invoke to filter for each input file. Doxygen will invoke the filter program
+# by executing (via popen()) the command:
+#
+# <filter> <input-file>
+#
+# where <filter> is the value of the INPUT_FILTER tag, and <input-file> is the
+# name of an input file. Doxygen will then use the output that the filter
+# program writes to standard output. If FILTER_PATTERNS is specified, this tag
+# will be ignored.
+#
+# Note that the filter must not add or remove lines; it is applied before the
+# code is scanned, but not when the output code is generated. If lines are added
+# or removed, the anchors will not be placed correctly.
+
+INPUT_FILTER           =
+
+# The FILTER_PATTERNS tag can be used to specify filters on a per file pattern
+# basis. Doxygen will compare the file name with each pattern and apply the
+# filter if there is a match. The filters are a list of the form: pattern=filter
+# (like *.cpp=my_cpp_filter). See INPUT_FILTER for further information on how
+# filters are used. If the FILTER_PATTERNS tag is empty or if none of the
+# patterns match the file name, INPUT_FILTER is applied.
+
+FILTER_PATTERNS        =
+
+# If the FILTER_SOURCE_FILES tag is set to YES, the input filter (if set using
+# INPUT_FILTER ) will also be used to filter the input files that are used for
+# producing the source files to browse (i.e. when SOURCE_BROWSER is set to YES).
+# The default value is: NO.
+
+FILTER_SOURCE_FILES    = NO
+
+# The FILTER_SOURCE_PATTERNS tag can be used to specify source filters per file
+# pattern. A pattern will override the setting for FILTER_PATTERN (if any) and
+# it is also possible to disable source filtering for a specific pattern using
+# *.ext= (so without naming a filter).
+# This tag requires that the tag FILTER_SOURCE_FILES is set to YES.
+
+FILTER_SOURCE_PATTERNS =
+
+# If the USE_MDFILE_AS_MAINPAGE tag refers to the name of a markdown file that
+# is part of the input, its contents will be placed on the main page
+# (index.html). This can be useful if you have a project on for instance GitHub
+# and want to reuse the introduction page also for the doxygen output.
+
+USE_MDFILE_AS_MAINPAGE =
+
+#---------------------------------------------------------------------------
+# Configuration options related to source browsing
+#---------------------------------------------------------------------------
+
+# If the SOURCE_BROWSER tag is set to YES then a list of source files will be
+# generated. Documented entities will be cross-referenced with these sources.
+#
+# Note: To get rid of all source code in the generated output, make sure that
+# also VERBATIM_HEADERS is set to NO.
+# The default value is: NO.
+
+SOURCE_BROWSER         = YES
+
+# Setting the INLINE_SOURCES tag to YES will include the body of functions,
+# classes and enums directly into the documentation.
+# The default value is: NO.
+
+INLINE_SOURCES         = NO
+
+# Setting the STRIP_CODE_COMMENTS tag to YES will instruct doxygen to hide any
+# special comment blocks from generated source code fragments. Normal C, C++ and
+# Fortran comments will always remain visible.
+# The default value is: YES.
+
+STRIP_CODE_COMMENTS    = YES
+
+# If the REFERENCED_BY_RELATION tag is set to YES then for each documented
+# function all documented functions referencing it will be listed.
+# The default value is: NO.
+
+REFERENCED_BY_RELATION = NO
+
+# If the REFERENCES_RELATION tag is set to YES then for each documented function
+# all documented entities called/used by that function will be listed.
+# The default value is: NO.
+
+REFERENCES_RELATION    = NO
+
+# If the REFERENCES_LINK_SOURCE tag is set to YES and SOURCE_BROWSER tag is set
+# to YES, then the hyperlinks from functions in REFERENCES_RELATION and
+# REFERENCED_BY_RELATION lists will link to the source code. Otherwise they will
+# link to the documentation.
+# The default value is: YES.
+
+REFERENCES_LINK_SOURCE = YES
+
+# If SOURCE_TOOLTIPS is enabled (the default) then hovering a hyperlink in the
+# source code will show a tooltip with additional information such as prototype,
+# brief description and links to the definition and documentation. Since this
+# will make the HTML file larger and loading of large files a bit slower, you
+# can opt to disable this feature.
+# The default value is: YES.
+# This tag requires that the tag SOURCE_BROWSER is set to YES.
+
+SOURCE_TOOLTIPS        = YES
+
+# If the USE_HTAGS tag is set to YES then the references to source code will
+# point to the HTML generated by the htags(1) tool instead of doxygen built-in
+# source browser. The htags tool is part of GNU's global source tagging system
+# (see http://www.gnu.org/software/global/global.html). You will need version
+# 4.8.6 or higher.
+#
+# To use it do the following:
+# - Install the latest version of global
+# - Enable SOURCE_BROWSER and USE_HTAGS in the config file
+# - Make sure the INPUT points to the root of the source tree
+# - Run doxygen as normal
+#
+# Doxygen will invoke htags (and that will in turn invoke gtags), so these
+# tools must be available from the command line (i.e. in the search path).
+#
+# The result: instead of the source browser generated by doxygen, the links to
+# source code will now point to the output of htags.
+# The default value is: NO.
+# This tag requires that the tag SOURCE_BROWSER is set to YES.
+
+USE_HTAGS              = NO
+
+# If the VERBATIM_HEADERS tag is set the YES then doxygen will generate a
+# verbatim copy of the header file for each class for which an include is
+# specified. Set to NO to disable this.
+# See also: Section \class.
+# The default value is: YES.
+
+VERBATIM_HEADERS       = YES
+
+#---------------------------------------------------------------------------
+# Configuration options related to the alphabetical class index
+#---------------------------------------------------------------------------
+
+# If the ALPHABETICAL_INDEX tag is set to YES, an alphabetical index of all
+# compounds will be generated. Enable this if the project contains a lot of
+# classes, structs, unions or interfaces.
+# The default value is: YES.
+
+ALPHABETICAL_INDEX     = YES
+
+# The COLS_IN_ALPHA_INDEX tag can be used to specify the number of columns in
+# which the alphabetical index list will be split.
+# Minimum value: 1, maximum value: 20, default value: 5.
+# This tag requires that the tag ALPHABETICAL_INDEX is set to YES.
+
+COLS_IN_ALPHA_INDEX    = 5
+
+# In case all classes in a project start with a common prefix, all classes will
+# be put under the same header in the alphabetical index. The IGNORE_PREFIX tag
+# can be used to specify a prefix (or a list of prefixes) that should be ignored
+# while generating the index headers.
+# This tag requires that the tag ALPHABETICAL_INDEX is set to YES.
+
+IGNORE_PREFIX          =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the HTML output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_HTML tag is set to YES doxygen will generate HTML output
+# The default value is: YES.
+
+GENERATE_HTML          = YES
+
+# The HTML_OUTPUT tag is used to specify where the HTML docs will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it.
+# The default directory is: html.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_OUTPUT            = html
+
+# The HTML_FILE_EXTENSION tag can be used to specify the file extension for each
+# generated HTML page (for example: .htm, .php, .asp).
+# The default value is: .html.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_FILE_EXTENSION    = .html
+
+# The HTML_HEADER tag can be used to specify a user-defined HTML header file for
+# each generated HTML page. If the tag is left blank doxygen will generate a
+# standard header.
+#
+# To get valid HTML the header file that includes any scripts and style sheets
+# that doxygen needs, which is dependent on the configuration options used (e.g.
+# the setting GENERATE_TREEVIEW). It is highly recommended to start with a
+# default header using
+# doxygen -w html new_header.html new_footer.html new_stylesheet.css
+# YourConfigFile
+# and then modify the file new_header.html. See also section "Doxygen usage"
+# for information on how to generate the default header that doxygen normally
+# uses.
+# Note: The header is subject to change so you typically have to regenerate the
+# default header when upgrading to a newer version of doxygen. For a description
+# of the possible markers and block names see the documentation.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_HEADER            =
+
+# The HTML_FOOTER tag can be used to specify a user-defined HTML footer for each
+# generated HTML page. If the tag is left blank doxygen will generate a standard
+# footer. See HTML_HEADER for more information on how to generate a default
+# footer and what special commands can be used inside the footer. See also
+# section "Doxygen usage" for information on how to generate the default footer
+# that doxygen normally uses.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_FOOTER            =
+
+# The HTML_STYLESHEET tag can be used to specify a user-defined cascading style
+# sheet that is used by each HTML page. It can be used to fine-tune the look of
+# the HTML output. If left blank doxygen will generate a default style sheet.
+# See also section "Doxygen usage" for information on how to generate the style
+# sheet that doxygen normally uses.
+# Note: It is recommended to use HTML_EXTRA_STYLESHEET instead of this tag, as
+# it is more robust and this tag (HTML_STYLESHEET) will in the future become
+# obsolete.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_STYLESHEET        =
+
+# The HTML_EXTRA_STYLESHEET tag can be used to specify an additional user-
+# defined cascading style sheet that is included after the standard style sheets
+# created by doxygen. Using this option one can overrule certain style aspects.
+# This is preferred over using HTML_STYLESHEET since it does not replace the
+# standard style sheet and is therefor more robust against future updates.
+# Doxygen will copy the style sheet file to the output directory. For an example
+# see the documentation.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_EXTRA_STYLESHEET  =
+
+# The HTML_EXTRA_FILES tag can be used to specify one or more extra images or
+# other source files which should be copied to the HTML output directory. Note
+# that these files will be copied to the base HTML output directory. Use the
+# $relpath^ marker in the HTML_HEADER and/or HTML_FOOTER files to load these
+# files. In the HTML_STYLESHEET file, use the file name only. Also note that the
+# files will be copied as-is; there are no commands or markers available.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_EXTRA_FILES       =
+
+# The HTML_COLORSTYLE_HUE tag controls the color of the HTML output. Doxygen
+# will adjust the colors in the stylesheet and background images according to
+# this color. Hue is specified as an angle on a colorwheel, see
+# http://en.wikipedia.org/wiki/Hue for more information. For instance the value
+# 0 represents red, 60 is yellow, 120 is green, 180 is cyan, 240 is blue, 300
+# purple, and 360 is red again.
+# Minimum value: 0, maximum value: 359, default value: 220.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_COLORSTYLE_HUE    = 220
+
+# The HTML_COLORSTYLE_SAT tag controls the purity (or saturation) of the colors
+# in the HTML output. For a value of 0 the output will use grayscales only. A
+# value of 255 will produce the most vivid colors.
+# Minimum value: 0, maximum value: 255, default value: 100.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_COLORSTYLE_SAT    = 100
+
+# The HTML_COLORSTYLE_GAMMA tag controls the gamma correction applied to the
+# luminance component of the colors in the HTML output. Values below 100
+# gradually make the output lighter, whereas values above 100 make the output
+# darker. The value divided by 100 is the actual gamma applied, so 80 represents
+# a gamma of 0.8, The value 220 represents a gamma of 2.2, and 100 does not
+# change the gamma.
+# Minimum value: 40, maximum value: 240, default value: 80.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_COLORSTYLE_GAMMA  = 80
+
+# If the HTML_TIMESTAMP tag is set to YES then the footer of each generated HTML
+# page will contain the date and time when the page was generated. Setting this
+# to NO can help when comparing the output of multiple runs.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_TIMESTAMP         = YES
+
+# If the HTML_DYNAMIC_SECTIONS tag is set to YES then the generated HTML
+# documentation will contain sections that can be hidden and shown after the
+# page has loaded.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_DYNAMIC_SECTIONS  = NO
+
+# With HTML_INDEX_NUM_ENTRIES one can control the preferred number of entries
+# shown in the various tree structured indices initially; the user can expand
+# and collapse entries dynamically later on. Doxygen will expand the tree to
+# such a level that at most the specified number of entries are visible (unless
+# a fully collapsed tree already exceeds this amount). So setting the number of
+# entries 1 will produce a full collapsed tree by default. 0 is a special value
+# representing an infinite number of entries and will result in a full expanded
+# tree by default.
+# Minimum value: 0, maximum value: 9999, default value: 100.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_INDEX_NUM_ENTRIES = 100
+
+# If the GENERATE_DOCSET tag is set to YES, additional index files will be
+# generated that can be used as input for Apple's Xcode 3 integrated development
+# environment (see: http://developer.apple.com/tools/xcode/), introduced with
+# OSX 10.5 (Leopard). To create a documentation set, doxygen will generate a
+# Makefile in the HTML output directory. Running make will produce the docset in
+# that directory and running make install will install the docset in
+# ~/Library/Developer/Shared/Documentation/DocSets so that Xcode will find it at
+# startup. See http://developer.apple.com/tools/creatingdocsetswithdoxygen.html
+# for more information.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_DOCSET        = NO
+
+# This tag determines the name of the docset feed. A documentation feed provides
+# an umbrella under which multiple documentation sets from a single provider
+# (such as a company or product suite) can be grouped.
+# The default value is: Doxygen generated docs.
+# This tag requires that the tag GENERATE_DOCSET is set to YES.
+
+DOCSET_FEEDNAME        = "Doxygen generated docs"
+
+# This tag specifies a string that should uniquely identify the documentation
+# set bundle. This should be a reverse domain-name style string, e.g.
+# com.mycompany.MyDocSet. Doxygen will append .docset to the name.
+# The default value is: org.doxygen.Project.
+# This tag requires that the tag GENERATE_DOCSET is set to YES.
+
+DOCSET_BUNDLE_ID       = org.doxygen.Project
+
+# The DOCSET_PUBLISHER_ID tag specifies a string that should uniquely identify
+# the documentation publisher. This should be a reverse domain-name style
+# string, e.g. com.mycompany.MyDocSet.documentation.
+# The default value is: org.doxygen.Publisher.
+# This tag requires that the tag GENERATE_DOCSET is set to YES.
+
+DOCSET_PUBLISHER_ID    = org.doxygen.Publisher
+
+# The DOCSET_PUBLISHER_NAME tag identifies the documentation publisher.
+# The default value is: Publisher.
+# This tag requires that the tag GENERATE_DOCSET is set to YES.
+
+DOCSET_PUBLISHER_NAME  = Publisher
+
+# If the GENERATE_HTMLHELP tag is set to YES then doxygen generates three
+# additional HTML index files: index.hhp, index.hhc, and index.hhk. The
+# index.hhp is a project file that can be read by Microsoft's HTML Help Workshop
+# (see: http://www.microsoft.com/en-us/download/details.aspx?id=21138) on
+# Windows.
+#
+# The HTML Help Workshop contains a compiler that can convert all HTML output
+# generated by doxygen into a single compiled HTML file (.chm). Compiled HTML
+# files are now used as the Windows 98 help format, and will replace the old
+# Windows help format (.hlp) on all Windows platforms in the future. Compressed
+# HTML files also contain an index, a table of contents, and you can search for
+# words in the documentation. The HTML workshop also contains a viewer for
+# compressed HTML files.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_HTMLHELP      = NO
+
+# The CHM_FILE tag can be used to specify the file name of the resulting .chm
+# file. You can add a path in front of the file if the result should not be
+# written to the html output directory.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+CHM_FILE               =
+
+# The HHC_LOCATION tag can be used to specify the location (absolute path
+# including file name) of the HTML help compiler ( hhc.exe). If non-empty
+# doxygen will try to run the HTML help compiler on the generated index.hhp.
+# The file has to be specified with full path.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+HHC_LOCATION           =
+
+# The GENERATE_CHI flag controls if a separate .chi index file is generated (
+# YES) or that it should be included in the master .chm file ( NO).
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+GENERATE_CHI           = NO
+
+# The CHM_INDEX_ENCODING is used to encode HtmlHelp index ( hhk), content ( hhc)
+# and project file content.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+CHM_INDEX_ENCODING     =
+
+# The BINARY_TOC flag controls whether a binary table of contents is generated (
+# YES) or a normal table of contents ( NO) in the .chm file.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+BINARY_TOC             = NO
+
+# The TOC_EXPAND flag can be set to YES to add extra items for group members to
+# the table of contents of the HTML help documentation and to the tree view.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+TOC_EXPAND             = NO
+
+# If the GENERATE_QHP tag is set to YES and both QHP_NAMESPACE and
+# QHP_VIRTUAL_FOLDER are set, an additional index file will be generated that
+# can be used as input for Qt's qhelpgenerator to generate a Qt Compressed Help
+# (.qch) of the generated HTML documentation.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_QHP           = NO
+
+# If the QHG_LOCATION tag is specified, the QCH_FILE tag can be used to specify
+# the file name of the resulting .qch file. The path specified is relative to
+# the HTML output folder.
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QCH_FILE               =
+
+# The QHP_NAMESPACE tag specifies the namespace to use when generating Qt Help
+# Project output. For more information please see Qt Help Project / Namespace
+# (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#namespace).
+# The default value is: org.doxygen.Project.
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_NAMESPACE          = org.doxygen.Project
+
+# The QHP_VIRTUAL_FOLDER tag specifies the namespace to use when generating Qt
+# Help Project output. For more information please see Qt Help Project / Virtual
+# Folders (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#virtual-
+# folders).
+# The default value is: doc.
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_VIRTUAL_FOLDER     = doc
+
+# If the QHP_CUST_FILTER_NAME tag is set, it specifies the name of a custom
+# filter to add. For more information please see Qt Help Project / Custom
+# Filters (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#custom-
+# filters).
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_CUST_FILTER_NAME   =
+
+# The QHP_CUST_FILTER_ATTRS tag specifies the list of the attributes of the
+# custom filter to add. For more information please see Qt Help Project / Custom
+# Filters (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#custom-
+# filters).
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_CUST_FILTER_ATTRS  =
+
+# The QHP_SECT_FILTER_ATTRS tag specifies the list of the attributes this
+# project's filter section matches. Qt Help Project / Filter Attributes (see:
+# http://qt-project.org/doc/qt-4.8/qthelpproject.html#filter-attributes).
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_SECT_FILTER_ATTRS  =
+
+# The QHG_LOCATION tag can be used to specify the location of Qt's
+# qhelpgenerator. If non-empty doxygen will try to run qhelpgenerator on the
+# generated .qhp file.
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHG_LOCATION           =
+
+# If the GENERATE_ECLIPSEHELP tag is set to YES, additional index files will be
+# generated, together with the HTML files, they form an Eclipse help plugin. To
+# install this plugin and make it available under the help contents menu in
+# Eclipse, the contents of the directory containing the HTML and XML files needs
+# to be copied into the plugins directory of eclipse. The name of the directory
+# within the plugins directory should be the same as the ECLIPSE_DOC_ID value.
+# After copying Eclipse needs to be restarted before the help appears.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_ECLIPSEHELP   = NO
+
+# A unique identifier for the Eclipse help plugin. When installing the plugin
+# the directory name containing the HTML and XML files should also have this
+# name. Each documentation set should have its own identifier.
+# The default value is: org.doxygen.Project.
+# This tag requires that the tag GENERATE_ECLIPSEHELP is set to YES.
+
+ECLIPSE_DOC_ID         = org.doxygen.Project
+
+# If you want full control over the layout of the generated HTML pages it might
+# be necessary to disable the index and replace it with your own. The
+# DISABLE_INDEX tag can be used to turn on/off the condensed index (tabs) at top
+# of each HTML page. A value of NO enables the index and the value YES disables
+# it. Since the tabs in the index contain the same information as the navigation
+# tree, you can set this option to YES if you also set GENERATE_TREEVIEW to YES.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+DISABLE_INDEX          = NO
+
+# The GENERATE_TREEVIEW tag is used to specify whether a tree-like index
+# structure should be generated to display hierarchical information. If the tag
+# value is set to YES, a side panel will be generated containing a tree-like
+# index structure (just like the one that is generated for HTML Help). For this
+# to work a browser that supports JavaScript, DHTML, CSS and frames is required
+# (i.e. any modern browser). Windows users are probably better off using the
+# HTML help feature. Via custom stylesheets (see HTML_EXTRA_STYLESHEET) one can
+# further fine-tune the look of the index. As an example, the default style
+# sheet generated by doxygen has an example that shows how to put an image at
+# the root of the tree instead of the PROJECT_NAME. Since the tree basically has
+# the same information as the tab index, you could consider setting
+# DISABLE_INDEX to YES when enabling this option.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_TREEVIEW      = YES
+
+# The ENUM_VALUES_PER_LINE tag can be used to set the number of enum values that
+# doxygen will group on one line in the generated HTML documentation.
+#
+# Note that a value of 0 will completely suppress the enum values from appearing
+# in the overview section.
+# Minimum value: 0, maximum value: 20, default value: 4.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+ENUM_VALUES_PER_LINE   = 4
+
+# If the treeview is enabled (see GENERATE_TREEVIEW) then this tag can be used
+# to set the initial width (in pixels) of the frame in which the tree is shown.
+# Minimum value: 0, maximum value: 1500, default value: 250.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+TREEVIEW_WIDTH         = 250
+
+# When the EXT_LINKS_IN_WINDOW option is set to YES doxygen will open links to
+# external symbols imported via tag files in a separate window.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+EXT_LINKS_IN_WINDOW    = NO
+
+# Use this tag to change the font size of LaTeX formulas included as images in
+# the HTML documentation. When you change the font size after a successful
+# doxygen run you need to manually remove any form_*.png images from the HTML
+# output directory to force them to be regenerated.
+# Minimum value: 8, maximum value: 50, default value: 10.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+FORMULA_FONTSIZE       = 10
+
+# Use the FORMULA_TRANPARENT tag to determine whether or not the images
+# generated for formulas are transparent PNGs. Transparent PNGs are not
+# supported properly for IE 6.0, but are supported on all modern browsers.
+#
+# Note that when changing this option you need to delete any form_*.png files in
+# the HTML output directory before the changes have effect.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+FORMULA_TRANSPARENT    = YES
+
+# Enable the USE_MATHJAX option to render LaTeX formulas using MathJax (see
+# http://www.mathjax.org) which uses client side Javascript for the rendering
+# instead of using prerendered bitmaps. Use this if you do not have LaTeX
+# installed or if you want to formulas look prettier in the HTML output. When
+# enabled you may also need to install MathJax separately and configure the path
+# to it using the MATHJAX_RELPATH option.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+USE_MATHJAX            = YES
+
+# When MathJax is enabled you can set the default output format to be used for
+# the MathJax output. See the MathJax site (see:
+# http://docs.mathjax.org/en/latest/output.html) for more details.
+# Possible values are: HTML-CSS (which is slower, but has the best
+# compatibility), NativeMML (i.e. MathML) and SVG.
+# The default value is: HTML-CSS.
+# This tag requires that the tag USE_MATHJAX is set to YES.
+
+MATHJAX_FORMAT         = HTML-CSS
+
+# When MathJax is enabled you need to specify the location relative to the HTML
+# output directory using the MATHJAX_RELPATH option. The destination directory
+# should contain the MathJax.js script. For instance, if the mathjax directory
+# is located at the same level as the HTML output directory, then
+# MATHJAX_RELPATH should be ../mathjax. The default value points to the MathJax
+# Content Delivery Network so you can quickly see the result without installing
+# MathJax. However, it is strongly recommended to install a local copy of
+# MathJax from http://www.mathjax.org before deployment.
+# The default value is: http://cdn.mathjax.org/mathjax/latest.
+# This tag requires that the tag USE_MATHJAX is set to YES.
+
+MATHJAX_RELPATH        = http://cdn.mathjax.org/mathjax/latest
+
+# The MATHJAX_EXTENSIONS tag can be used to specify one or more MathJax
+# extension names that should be enabled during MathJax rendering. For example
+# MATHJAX_EXTENSIONS = TeX/AMSmath TeX/AMSsymbols
+# This tag requires that the tag USE_MATHJAX is set to YES.
+
+MATHJAX_EXTENSIONS     =
+
+# The MATHJAX_CODEFILE tag can be used to specify a file with javascript pieces
+# of code that will be used on startup of the MathJax code. See the MathJax site
+# (see: http://docs.mathjax.org/en/latest/output.html) for more details. For an
+# example see the documentation.
+# This tag requires that the tag USE_MATHJAX is set to YES.
+
+MATHJAX_CODEFILE       =
+
+# When the SEARCHENGINE tag is enabled doxygen will generate a search box for
+# the HTML output. The underlying search engine uses javascript and DHTML and
+# should work on any modern browser. Note that when using HTML help
+# (GENERATE_HTMLHELP), Qt help (GENERATE_QHP), or docsets (GENERATE_DOCSET)
+# there is already a search function so this one should typically be disabled.
+# For large projects the javascript based search engine can be slow, then
+# enabling SERVER_BASED_SEARCH may provide a better solution. It is possible to
+# search using the keyboard; to jump to the search box use <access key> + S
+# (what the <access key> is depends on the OS and browser, but it is typically
+# <CTRL>, <ALT>/<option>, or both). Inside the search box use the <cursor down
+# key> to jump into the search results window, the results can be navigated
+# using the <cursor keys>. Press <Enter> to select an item or <escape> to cancel
+# the search. The filter options can be selected when the cursor is inside the
+# search box by pressing <Shift>+<cursor down>. Also here use the <cursor keys>
+# to select a filter and <Enter> or <escape> to activate or cancel the filter
+# option.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+SEARCHENGINE           = YES
+
+# When the SERVER_BASED_SEARCH tag is enabled the search engine will be
+# implemented using a web server instead of a web client using Javascript. There
+# are two flavours of web server based searching depending on the
+# EXTERNAL_SEARCH setting. When disabled, doxygen will generate a PHP script for
+# searching and an index file used by the script. When EXTERNAL_SEARCH is
+# enabled the indexing and searching needs to be provided by external tools. See
+# the section "External Indexing and Searching" for details.
+# The default value is: NO.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+SERVER_BASED_SEARCH    = NO
+
+# When EXTERNAL_SEARCH tag is enabled doxygen will no longer generate the PHP
+# script for searching. Instead the search results are written to an XML file
+# which needs to be processed by an external indexer. Doxygen will invoke an
+# external search engine pointed to by the SEARCHENGINE_URL option to obtain the
+# search results.
+#
+# Doxygen ships with an example indexer ( doxyindexer) and search engine
+# (doxysearch.cgi) which are based on the open source search engine library
+# Xapian (see: http://xapian.org/).
+#
+# See the section "External Indexing and Searching" for details.
+# The default value is: NO.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+EXTERNAL_SEARCH        = NO
+
+# The SEARCHENGINE_URL should point to a search engine hosted by a web server
+# which will return the search results when EXTERNAL_SEARCH is enabled.
+#
+# Doxygen ships with an example indexer ( doxyindexer) and search engine
+# (doxysearch.cgi) which are based on the open source search engine library
+# Xapian (see: http://xapian.org/). See the section "External Indexing and
+# Searching" for details.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+SEARCHENGINE_URL       =
+
+# When SERVER_BASED_SEARCH and EXTERNAL_SEARCH are both enabled the unindexed
+# search data is written to a file for indexing by an external tool. With the
+# SEARCHDATA_FILE tag the name of this file can be specified.
+# The default file is: searchdata.xml.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+SEARCHDATA_FILE        = searchdata.xml
+
+# When SERVER_BASED_SEARCH and EXTERNAL_SEARCH are both enabled the
+# EXTERNAL_SEARCH_ID tag can be used as an identifier for the project. This is
+# useful in combination with EXTRA_SEARCH_MAPPINGS to search through multiple
+# projects and redirect the results back to the right project.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+EXTERNAL_SEARCH_ID     =
+
+# The EXTRA_SEARCH_MAPPINGS tag can be used to enable searching through doxygen
+# projects other than the one defined by this configuration file, but that are
+# all added to the same external search index. Each project needs to have a
+# unique id set via EXTERNAL_SEARCH_ID. The search mapping then maps the id of
+# to a relative location where the documentation can be found. The format is:
+# EXTRA_SEARCH_MAPPINGS = tagname1=loc1 tagname2=loc2 ...
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+EXTRA_SEARCH_MAPPINGS  =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the LaTeX output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_LATEX tag is set to YES doxygen will generate LaTeX output.
+# The default value is: YES.
+
+GENERATE_LATEX         = YES
+
+# The LATEX_OUTPUT tag is used to specify where the LaTeX docs will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it.
+# The default directory is: latex.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_OUTPUT           = latex
+
+# The LATEX_CMD_NAME tag can be used to specify the LaTeX command name to be
+# invoked.
+#
+# Note that when enabling USE_PDFLATEX this option is only used for generating
+# bitmaps for formulas in the HTML output, but not in the Makefile that is
+# written to the output directory.
+# The default file is: latex.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_CMD_NAME         = latex
+
+# The MAKEINDEX_CMD_NAME tag can be used to specify the command name to generate
+# index for LaTeX.
+# The default file is: makeindex.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+MAKEINDEX_CMD_NAME     = makeindex
+
+# If the COMPACT_LATEX tag is set to YES doxygen generates more compact LaTeX
+# documents. This may be useful for small projects and may help to save some
+# trees in general.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+COMPACT_LATEX          = NO
+
+# The PAPER_TYPE tag can be used to set the paper type that is used by the
+# printer.
+# Possible values are: a4 (210 x 297 mm), letter (8.5 x 11 inches), legal (8.5 x
+# 14 inches) and executive (7.25 x 10.5 inches).
+# The default value is: a4.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+PAPER_TYPE             = a4
+
+# The EXTRA_PACKAGES tag can be used to specify one or more LaTeX package names
+# that should be included in the LaTeX output. To get the times font for
+# instance you can specify
+# EXTRA_PACKAGES=times
+# If left blank no extra packages will be included.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+EXTRA_PACKAGES         =
+
+# The LATEX_HEADER tag can be used to specify a personal LaTeX header for the
+# generated LaTeX document. The header should contain everything until the first
+# chapter. If it is left blank doxygen will generate a standard header. See
+# section "Doxygen usage" for information on how to let doxygen write the
+# default header to a separate file.
+#
+# Note: Only use a user-defined header if you know what you are doing! The
+# following commands have a special meaning inside the header: $title,
+# $datetime, $date, $doxygenversion, $projectname, $projectnumber. Doxygen will
+# replace them by respectively the title of the page, the current date and time,
+# only the current date, the version number of doxygen, the project name (see
+# PROJECT_NAME), or the project number (see PROJECT_NUMBER).
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_HEADER           =
+
+# The LATEX_FOOTER tag can be used to specify a personal LaTeX footer for the
+# generated LaTeX document. The footer should contain everything after the last
+# chapter. If it is left blank doxygen will generate a standard footer.
+#
+# Note: Only use a user-defined footer if you know what you are doing!
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_FOOTER           =
+
+# The LATEX_EXTRA_FILES tag can be used to specify one or more extra images or
+# other source files which should be copied to the LATEX_OUTPUT output
+# directory. Note that the files will be copied as-is; there are no commands or
+# markers available.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_EXTRA_FILES      =
+
+# If the PDF_HYPERLINKS tag is set to YES, the LaTeX that is generated is
+# prepared for conversion to PDF (using ps2pdf or pdflatex). The PDF file will
+# contain links (just like the HTML output) instead of page references. This
+# makes the output suitable for online browsing using a PDF viewer.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+PDF_HYPERLINKS         = YES
+
+# If the LATEX_PDFLATEX tag is set to YES, doxygen will use pdflatex to generate
+# the PDF file directly from the LaTeX files. Set this option to YES to get a
+# higher quality PDF documentation.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+USE_PDFLATEX           = YES
+
+# If the LATEX_BATCHMODE tag is set to YES, doxygen will add the \batchmode
+# command to the generated LaTeX files. This will instruct LaTeX to keep running
+# if errors occur, instead of asking the user for help. This option is also used
+# when generating formulas in HTML.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_BATCHMODE        = NO
+
+# If the LATEX_HIDE_INDICES tag is set to YES then doxygen will not include the
+# index chapters (such as File Index, Compound Index, etc.) in the output.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_HIDE_INDICES     = NO
+
+# If the LATEX_SOURCE_CODE tag is set to YES then doxygen will include source
+# code with syntax highlighting in the LaTeX output.
+#
+# Note that which sources are shown also depends on other settings such as
+# SOURCE_BROWSER.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_SOURCE_CODE      = NO
+
+# The LATEX_BIB_STYLE tag can be used to specify the style to use for the
+# bibliography, e.g. plainnat, or ieeetr. See
+# http://en.wikipedia.org/wiki/BibTeX and \cite for more info.
+# The default value is: plain.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_BIB_STYLE        = plain
+
+#---------------------------------------------------------------------------
+# Configuration options related to the RTF output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_RTF tag is set to YES doxygen will generate RTF output. The
+# RTF output is optimized for Word 97 and may not look too pretty with other RTF
+# readers/editors.
+# The default value is: NO.
+
+GENERATE_RTF           = NO
+
+# The RTF_OUTPUT tag is used to specify where the RTF docs will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it.
+# The default directory is: rtf.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+RTF_OUTPUT             = rtf
+
+# If the COMPACT_RTF tag is set to YES doxygen generates more compact RTF
+# documents. This may be useful for small projects and may help to save some
+# trees in general.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+COMPACT_RTF            = NO
+
+# If the RTF_HYPERLINKS tag is set to YES, the RTF that is generated will
+# contain hyperlink fields. The RTF file will contain links (just like the HTML
+# output) instead of page references. This makes the output suitable for online
+# browsing using Word or some other Word compatible readers that support those
+# fields.
+#
+# Note: WordPad (write) and others do not support links.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+RTF_HYPERLINKS         = NO
+
+# Load stylesheet definitions from file. Syntax is similar to doxygen's config
+# file, i.e. a series of assignments. You only have to provide replacements,
+# missing definitions are set to their default value.
+#
+# See also section "Doxygen usage" for information on how to generate the
+# default style sheet that doxygen normally uses.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+RTF_STYLESHEET_FILE    =
+
+# Set optional variables used in the generation of an RTF document. Syntax is
+# similar to doxygen's config file. A template extensions file can be generated
+# using doxygen -e rtf extensionFile.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+RTF_EXTENSIONS_FILE    =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the man page output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_MAN tag is set to YES doxygen will generate man pages for
+# classes and files.
+# The default value is: NO.
+
+GENERATE_MAN           = NO
+
+# The MAN_OUTPUT tag is used to specify where the man pages will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it. A directory man3 will be created inside the directory specified by
+# MAN_OUTPUT.
+# The default directory is: man.
+# This tag requires that the tag GENERATE_MAN is set to YES.
+
+MAN_OUTPUT             = man
+
+# The MAN_EXTENSION tag determines the extension that is added to the generated
+# man pages. In case the manual section does not start with a number, the number
+# 3 is prepended. The dot (.) at the beginning of the MAN_EXTENSION tag is
+# optional.
+# The default value is: .3.
+# This tag requires that the tag GENERATE_MAN is set to YES.
+
+MAN_EXTENSION          = .3
+
+# If the MAN_LINKS tag is set to YES and doxygen generates man output, then it
+# will generate one additional man file for each entity documented in the real
+# man page(s). These additional files only source the real man page, but without
+# them the man command would be unable to find the correct page.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_MAN is set to YES.
+
+MAN_LINKS              = NO
+
+#---------------------------------------------------------------------------
+# Configuration options related to the XML output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_XML tag is set to YES doxygen will generate an XML file that
+# captures the structure of the code including all documentation.
+# The default value is: NO.
+
+GENERATE_XML           = NO
+
+# The XML_OUTPUT tag is used to specify where the XML pages will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it.
+# The default directory is: xml.
+# This tag requires that the tag GENERATE_XML is set to YES.
+
+XML_OUTPUT             = xml
+
+# The XML_SCHEMA tag can be used to specify a XML schema, which can be used by a
+# validating XML parser to check the syntax of the XML files.
+# This tag requires that the tag GENERATE_XML is set to YES.
+
+XML_SCHEMA             =
+
+# The XML_DTD tag can be used to specify a XML DTD, which can be used by a
+# validating XML parser to check the syntax of the XML files.
+# This tag requires that the tag GENERATE_XML is set to YES.
+
+XML_DTD                =
+
+# If the XML_PROGRAMLISTING tag is set to YES doxygen will dump the program
+# listings (including syntax highlighting and cross-referencing information) to
+# the XML output. Note that enabling this will significantly increase the size
+# of the XML output.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_XML is set to YES.
+
+XML_PROGRAMLISTING     = YES
+
+#---------------------------------------------------------------------------
+# Configuration options related to the DOCBOOK output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_DOCBOOK tag is set to YES doxygen will generate Docbook files
+# that can be used to generate PDF.
+# The default value is: NO.
+
+GENERATE_DOCBOOK       = NO
+
+# The DOCBOOK_OUTPUT tag is used to specify where the Docbook pages will be put.
+# If a relative path is entered the value of OUTPUT_DIRECTORY will be put in
+# front of it.
+# The default directory is: docbook.
+# This tag requires that the tag GENERATE_DOCBOOK is set to YES.
+
+DOCBOOK_OUTPUT         = docbook
+
+#---------------------------------------------------------------------------
+# Configuration options for the AutoGen Definitions output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_AUTOGEN_DEF tag is set to YES doxygen will generate an AutoGen
+# Definitions (see http://autogen.sf.net) file that captures the structure of
+# the code including all documentation. Note that this feature is still
+# experimental and incomplete at the moment.
+# The default value is: NO.
+
+GENERATE_AUTOGEN_DEF   = NO
+
+#---------------------------------------------------------------------------
+# Configuration options related to the Perl module output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_PERLMOD tag is set to YES doxygen will generate a Perl module
+# file that captures the structure of the code including all documentation.
+#
+# Note that this feature is still experimental and incomplete at the moment.
+# The default value is: NO.
+
+GENERATE_PERLMOD       = NO
+
+# If the PERLMOD_LATEX tag is set to YES doxygen will generate the necessary
+# Makefile rules, Perl scripts and LaTeX code to be able to generate PDF and DVI
+# output from the Perl module output.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_PERLMOD is set to YES.
+
+PERLMOD_LATEX          = NO
+
+# If the PERLMOD_PRETTY tag is set to YES the Perl module output will be nicely
+# formatted so it can be parsed by a human reader. This is useful if you want to
+# understand what is going on. On the other hand, if this tag is set to NO the
+# size of the Perl module output will be much smaller and Perl will parse it
+# just the same.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_PERLMOD is set to YES.
+
+PERLMOD_PRETTY         = YES
+
+# The names of the make variables in the generated doxyrules.make file are
+# prefixed with the string contained in PERLMOD_MAKEVAR_PREFIX. This is useful
+# so different doxyrules.make files included by the same Makefile don't
+# overwrite each other's variables.
+# This tag requires that the tag GENERATE_PERLMOD is set to YES.
+
+PERLMOD_MAKEVAR_PREFIX =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the preprocessor
+#---------------------------------------------------------------------------
+
+# If the ENABLE_PREPROCESSING tag is set to YES doxygen will evaluate all
+# C-preprocessor directives found in the sources and include files.
+# The default value is: YES.
+
+ENABLE_PREPROCESSING   = YES
+
+# If the MACRO_EXPANSION tag is set to YES doxygen will expand all macro names
+# in the source code. If set to NO only conditional compilation will be
+# performed. Macro expansion can be done in a controlled way by setting
+# EXPAND_ONLY_PREDEF to YES.
+# The default value is: NO.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+MACRO_EXPANSION        = NO
+
+# If the EXPAND_ONLY_PREDEF and MACRO_EXPANSION tags are both set to YES then
+# the macro expansion is limited to the macros specified with the PREDEFINED and
+# EXPAND_AS_DEFINED tags.
+# The default value is: NO.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+EXPAND_ONLY_PREDEF     = NO
+
+# If the SEARCH_INCLUDES tag is set to YES the includes files in the
+# INCLUDE_PATH will be searched if a #include is found.
+# The default value is: YES.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+SEARCH_INCLUDES        = YES
+
+# The INCLUDE_PATH tag can be used to specify one or more directories that
+# contain include files that are not input files but should be processed by the
+# preprocessor.
+# This tag requires that the tag SEARCH_INCLUDES is set to YES.
+
+INCLUDE_PATH           =
+
+# You can use the INCLUDE_FILE_PATTERNS tag to specify one or more wildcard
+# patterns (like *.h and *.hpp) to filter out the header-files in the
+# directories. If left blank, the patterns specified with FILE_PATTERNS will be
+# used.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+INCLUDE_FILE_PATTERNS  =
+
+# The PREDEFINED tag can be used to specify one or more macro names that are
+# defined before the preprocessor is started (similar to the -D option of e.g.
+# gcc). The argument of the tag is a list of macros of the form: name or
+# name=definition (no spaces). If the definition and the "=" are omitted, "=1"
+# is assumed. To prevent a macro definition from being undefined via #undef or
+# recursively expanded use the := operator instead of the = operator.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+PREDEFINED             =
+
+# If the MACRO_EXPANSION and EXPAND_ONLY_PREDEF tags are set to YES then this
+# tag can be used to specify a list of macro names that should be expanded. The
+# macro definition that is found in the sources will be used. Use the PREDEFINED
+# tag if you want to use a different macro definition that overrules the
+# definition found in the source code.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+EXPAND_AS_DEFINED      =
+
+# If the SKIP_FUNCTION_MACROS tag is set to YES then doxygen's preprocessor will
+# remove all refrences to function-like macros that are alone on a line, have an
+# all uppercase name, and do not end with a semicolon. Such function macros are
+# typically used for boiler-plate code, and will confuse the parser if not
+# removed.
+# The default value is: YES.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+SKIP_FUNCTION_MACROS   = YES
+
+#---------------------------------------------------------------------------
+# Configuration options related to external references
+#---------------------------------------------------------------------------
+
+# The TAGFILES tag can be used to specify one or more tag files. For each tag
+# file the location of the external documentation should be added. The format of
+# a tag file without this location is as follows:
+# TAGFILES = file1 file2 ...
+# Adding location for the tag files is done as follows:
+# TAGFILES = file1=loc1 "file2 = loc2" ...
+# where loc1 and loc2 can be relative or absolute paths or URLs. See the
+# section "Linking to external documentation" for more information about the use
+# of tag files.
+# Note: Each tag file must have an unique name (where the name does NOT include
+# the path). If a tag file is not located in the directory in which doxygen is
+# run, you must also specify the path to the tagfile here.
+
+TAGFILES               =
+
+# When a file name is specified after GENERATE_TAGFILE, doxygen will create a
+# tag file that is based on the input files it reads. See section "Linking to
+# external documentation" for more information about the usage of tag files.
+
+GENERATE_TAGFILE       =
+
+# If the ALLEXTERNALS tag is set to YES all external class will be listed in the
+# class index. If set to NO only the inherited external classes will be listed.
+# The default value is: NO.
+
+ALLEXTERNALS           = NO
+
+# If the EXTERNAL_GROUPS tag is set to YES all external groups will be listed in
+# the modules index. If set to NO, only the current project's groups will be
+# listed.
+# The default value is: YES.
+
+EXTERNAL_GROUPS        = YES
+
+# If the EXTERNAL_PAGES tag is set to YES all external pages will be listed in
+# the related pages index. If set to NO, only the current project's pages will
+# be listed.
+# The default value is: YES.
+
+EXTERNAL_PAGES         = YES
+
+# The PERL_PATH should be the absolute path and name of the perl script
+# interpreter (i.e. the result of 'which perl').
+# The default file (with absolute path) is: /usr/bin/perl.
+
+PERL_PATH              = /usr/bin/perl
+
+#---------------------------------------------------------------------------
+# Configuration options related to the dot tool
+#---------------------------------------------------------------------------
+
+# If the CLASS_DIAGRAMS tag is set to YES doxygen will generate a class diagram
+# (in HTML and LaTeX) for classes with base or super classes. Setting the tag to
+# NO turns the diagrams off. Note that this option also works with HAVE_DOT
+# disabled, but it is recommended to install and use dot, since it yields more
+# powerful graphs.
+# The default value is: YES.
+
+CLASS_DIAGRAMS         = YES
+
+# You can define message sequence charts within doxygen comments using the \msc
+# command. Doxygen will then run the mscgen tool (see:
+# http://www.mcternan.me.uk/mscgen/)) to produce the chart and insert it in the
+# documentation. The MSCGEN_PATH tag allows you to specify the directory where
+# the mscgen tool resides. If left empty the tool is assumed to be found in the
+# default search path.
+
+MSCGEN_PATH            =
+
+# You can include diagrams made with dia in doxygen documentation. Doxygen will
+# then run dia to produce the diagram and insert it in the documentation. The
+# DIA_PATH tag allows you to specify the directory where the dia binary resides.
+# If left empty dia is assumed to be found in the default search path.
+
+DIA_PATH               =
+
+# If set to YES, the inheritance and collaboration graphs will hide inheritance
+# and usage relations if the target is undocumented or is not a class.
+# The default value is: YES.
+
+HIDE_UNDOC_RELATIONS   = YES
+
+# If you set the HAVE_DOT tag to YES then doxygen will assume the dot tool is
+# available from the path. This tool is part of Graphviz (see:
+# http://www.graphviz.org/), a graph visualization toolkit from AT&T and Lucent
+# Bell Labs. The other options in this section have no effect if this option is
+# set to NO
+# The default value is: NO.
+
+HAVE_DOT               = YES
+
+# The DOT_NUM_THREADS specifies the number of dot invocations doxygen is allowed
+# to run in parallel. When set to 0 doxygen will base this on the number of
+# processors available in the system. You can set it explicitly to a value
+# larger than 0 to get control over the balance between CPU load and processing
+# speed.
+# Minimum value: 0, maximum value: 32, default value: 0.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_NUM_THREADS        = 2
+
+# When you want a differently looking font n the dot files that doxygen
+# generates you can specify the font name using DOT_FONTNAME. You need to make
+# sure dot is able to find the font, which can be done by putting it in a
+# standard location or by setting the DOTFONTPATH environment variable or by
+# setting DOT_FONTPATH to the directory containing the font.
+# The default value is: Helvetica.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_FONTNAME           = Helvetica
+
+# The DOT_FONTSIZE tag can be used to set the size (in points) of the font of
+# dot graphs.
+# Minimum value: 4, maximum value: 24, default value: 10.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_FONTSIZE           = 10
+
+# By default doxygen will tell dot to use the default font as specified with
+# DOT_FONTNAME. If you specify a different font using DOT_FONTNAME you can set
+# the path where dot can find it using this tag.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_FONTPATH           =
+
+# If the CLASS_GRAPH tag is set to YES then doxygen will generate a graph for
+# each documented class showing the direct and indirect inheritance relations.
+# Setting this tag to YES will force the CLASS_DIAGRAMS tag to NO.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+CLASS_GRAPH            = YES
+
+# If the COLLABORATION_GRAPH tag is set to YES then doxygen will generate a
+# graph for each documented class showing the direct and indirect implementation
+# dependencies (inheritance, containment, and class references variables) of the
+# class with other documented classes.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+COLLABORATION_GRAPH    = YES
+
+# If the GROUP_GRAPHS tag is set to YES then doxygen will generate a graph for
+# groups, showing the direct groups dependencies.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+GROUP_GRAPHS           = YES
+
+# If the UML_LOOK tag is set to YES doxygen will generate inheritance and
+# collaboration diagrams in a style similar to the OMG's Unified Modeling
+# Language.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+UML_LOOK               = NO
+
+# If the UML_LOOK tag is enabled, the fields and methods are shown inside the
+# class node. If there are many fields or methods and many nodes the graph may
+# become too big to be useful. The UML_LIMIT_NUM_FIELDS threshold limits the
+# number of items for each type to make the size more manageable. Set this to 0
+# for no limit. Note that the threshold may be exceeded by 50% before the limit
+# is enforced. So when you set the threshold to 10, up to 15 fields may appear,
+# but if the number exceeds 15, the total amount of fields shown is limited to
+# 10.
+# Minimum value: 0, maximum value: 100, default value: 10.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+UML_LIMIT_NUM_FIELDS   = 10
+
+# If the TEMPLATE_RELATIONS tag is set to YES then the inheritance and
+# collaboration graphs will show the relations between templates and their
+# instances.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+TEMPLATE_RELATIONS     = NO
+
+# If the INCLUDE_GRAPH, ENABLE_PREPROCESSING and SEARCH_INCLUDES tags are set to
+# YES then doxygen will generate a graph for each documented file showing the
+# direct and indirect include dependencies of the file with other documented
+# files.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+INCLUDE_GRAPH          = YES
+
+# If the INCLUDED_BY_GRAPH, ENABLE_PREPROCESSING and SEARCH_INCLUDES tags are
+# set to YES then doxygen will generate a graph for each documented file showing
+# the direct and indirect include dependencies of the file with other documented
+# files.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+INCLUDED_BY_GRAPH      = YES
+
+# If the CALL_GRAPH tag is set to YES then doxygen will generate a call
+# dependency graph for every global function or class method.
+#
+# Note that enabling this option will significantly increase the time of a run.
+# So in most cases it will be better to enable call graphs for selected
+# functions only using the \callgraph command.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+CALL_GRAPH             = NO
+
+# If the CALLER_GRAPH tag is set to YES then doxygen will generate a caller
+# dependency graph for every global function or class method.
+#
+# Note that enabling this option will significantly increase the time of a run.
+# So in most cases it will be better to enable caller graphs for selected
+# functions only using the \callergraph command.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+CALLER_GRAPH           = NO
+
+# If the GRAPHICAL_HIERARCHY tag is set to YES then doxygen will graphical
+# hierarchy of all classes instead of a textual one.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+GRAPHICAL_HIERARCHY    = YES
+
+# If the DIRECTORY_GRAPH tag is set to YES then doxygen will show the
+# dependencies a directory has on other directories in a graphical way. The
+# dependency relations are determined by the #include relations between the
+# files in the directories.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DIRECTORY_GRAPH        = YES
+
+# The DOT_IMAGE_FORMAT tag can be used to set the image format of the images
+# generated by dot.
+# Note: If you choose svg you need to set HTML_FILE_EXTENSION to xhtml in order
+# to make the SVG files visible in IE 9+ (other browsers do not have this
+# requirement).
+# Possible values are: png, jpg, gif and svg.
+# The default value is: png.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_IMAGE_FORMAT       = png
+
+# If DOT_IMAGE_FORMAT is set to svg, then this option can be set to YES to
+# enable generation of interactive SVG images that allow zooming and panning.
+#
+# Note that this requires a modern browser other than Internet Explorer. Tested
+# and working are Firefox, Chrome, Safari, and Opera.
+# Note: For IE 9+ you need to set HTML_FILE_EXTENSION to xhtml in order to make
+# the SVG files visible. Older versions of IE do not have SVG support.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+INTERACTIVE_SVG        = NO
+
+# The DOT_PATH tag can be used to specify the path where the dot tool can be
+# found. If left blank, it is assumed the dot tool can be found in the path.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_PATH               =
+
+# The DOTFILE_DIRS tag can be used to specify one or more directories that
+# contain dot files that are included in the documentation (see the \dotfile
+# command).
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOTFILE_DIRS           =
+
+# The MSCFILE_DIRS tag can be used to specify one or more directories that
+# contain msc files that are included in the documentation (see the \mscfile
+# command).
+
+MSCFILE_DIRS           =
+
+# The DIAFILE_DIRS tag can be used to specify one or more directories that
+# contain dia files that are included in the documentation (see the \diafile
+# command).
+
+DIAFILE_DIRS           =
+
+# The DOT_GRAPH_MAX_NODES tag can be used to set the maximum number of nodes
+# that will be shown in the graph. If the number of nodes in a graph becomes
+# larger than this value, doxygen will truncate the graph, which is visualized
+# by representing a node as a red box. Note that doxygen if the number of direct
+# children of the root node in a graph is already larger than
+# DOT_GRAPH_MAX_NODES then the graph will not be shown at all. Also note that
+# the size of a graph can be further restricted by MAX_DOT_GRAPH_DEPTH.
+# Minimum value: 0, maximum value: 10000, default value: 50.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_GRAPH_MAX_NODES    = 50
+
+# The MAX_DOT_GRAPH_DEPTH tag can be used to set the maximum depth of the graphs
+# generated by dot. A depth value of 3 means that only nodes reachable from the
+# root by following a path via at most 3 edges will be shown. Nodes that lay
+# further from the root node will be omitted. Note that setting this option to 1
+# or 2 may greatly reduce the computation time needed for large code bases. Also
+# note that the size of a graph can be further restricted by
+# DOT_GRAPH_MAX_NODES. Using a depth of 0 means no depth restriction.
+# Minimum value: 0, maximum value: 1000, default value: 0.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+MAX_DOT_GRAPH_DEPTH    = 0
+
+# Set the DOT_TRANSPARENT tag to YES to generate images with a transparent
+# background. This is disabled by default, because dot on Windows does not seem
+# to support this out of the box.
+#
+# Warning: Depending on the platform used, enabling this option may lead to
+# badly anti-aliased labels on the edges of a graph (i.e. they become hard to
+# read).
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_TRANSPARENT        = YES
+
+# Set the DOT_MULTI_TARGETS tag to YES allow dot to generate multiple output
+# files in one run (i.e. multiple -o and -T options on the command line). This
+# makes dot run faster, but since only newer versions of dot (>1.8.10) support
+# this, this feature is disabled by default.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_MULTI_TARGETS      = YES
+
+# If the GENERATE_LEGEND tag is set to YES doxygen will generate a legend page
+# explaining the meaning of the various boxes and arrows in the dot generated
+# graphs.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+GENERATE_LEGEND        = YES
+
+# If the DOT_CLEANUP tag is set to YES doxygen will remove the intermediate dot
+# files that are used to generate the various graphs.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_CLEANUP            = YES
diff --git a/docs/doxy.cfg.in b/docs/doxy.cfg.in
new file mode 100644
index 00000000..c4b82094
--- /dev/null
+++ b/docs/doxy.cfg.in
@@ -0,0 +1,2305 @@
+# Doxyfile 1.8.6
+
+# This file describes the settings to be used by the documentation system
+# doxygen (www.doxygen.org) for a project.
+#
+# All text after a double hash (##) is considered a comment and is placed in
+# front of the TAG it is preceding.
+#
+# All text after a single hash (#) is considered a comment and will be ignored.
+# The format is:
+# TAG = value [value, ...]
+# For lists, items can also be appended using:
+# TAG += value [value, ...]
+# Values that contain spaces should be placed between quotes (\" \").
+
+#---------------------------------------------------------------------------
+# Project related configuration options
+#---------------------------------------------------------------------------
+
+# This tag specifies the encoding used for all characters in the config file
+# that follow. The default is UTF-8 which is also the encoding used for all text
+# before the first occurrence of this tag. Doxygen uses libiconv (or the iconv
+# built into libc) for the transcoding. See http://www.gnu.org/software/libiconv
+# for the list of possible encodings.
+# The default value is: UTF-8.
+
+DOXYFILE_ENCODING      = UTF-8
+
+# The PROJECT_NAME tag is a single word (or a sequence of words surrounded by
+# double-quotes, unless you are using Doxywizard) that should identify the
+# project for which the documentation is generated. This name is used in the
+# title of most generated pages and in a few other places.
+# The default value is: My Project.
+
+PROJECT_NAME           = "Grid"
+
+# The PROJECT_NUMBER tag can be used to enter a project or revision number. This
+# could be handy for archiving the generated documentation or if some version
+# control system is used.
+
+PROJECT_NUMBER         = 1.0
+
+# Using the PROJECT_BRIEF tag one can provide an optional one line description
+# for a project that appears at the top of each page and should give viewer a
+# quick idea about the purpose of the project. Keep the description short.
+
+PROJECT_BRIEF          =
+
+# With the PROJECT_LOGO tag one can specify an logo or icon that is included in
+# the documentation. The maximum height of the logo should not exceed 55 pixels
+# and the maximum width should not exceed 200 pixels. Doxygen will copy the logo
+# to the output directory.
+
+PROJECT_LOGO           =
+
+# The OUTPUT_DIRECTORY tag is used to specify the (relative or absolute) path
+# into which the generated documentation will be written. If a relative path is
+# entered, it will be relative to the location where doxygen was started. If
+# left blank the current directory will be used.
+
+OUTPUT_DIRECTORY       = ./doxy-en/
+
+# If the CREATE_SUBDIRS tag is set to YES, then doxygen will create 4096 sub-
+# directories (in 2 levels) under the output directory of each output format and
+# will distribute the generated files over these directories. Enabling this
+# option can be useful when feeding doxygen a huge amount of source files, where
+# putting all generated files in the same directory would otherwise causes
+# performance problems for the file system.
+# The default value is: NO.
+
+CREATE_SUBDIRS         = NO
+
+# The OUTPUT_LANGUAGE tag is used to specify the language in which all
+# documentation generated by doxygen is written. Doxygen will use this
+# information to generate all constant output in the proper language.
+# Possible values are: Afrikaans, Arabic, Armenian, Brazilian, Catalan, Chinese,
+# Chinese-Traditional, Croatian, Czech, Danish, Dutch, English (United States),
+# Esperanto, Farsi (Persian), Finnish, French, German, Greek, Hungarian,
+# Indonesian, Italian, Japanese, Japanese-en (Japanese with English messages),
+# Korean, Korean-en (Korean with English messages), Latvian, Lithuanian,
+# Macedonian, Norwegian, Persian (Farsi), Polish, Portuguese, Romanian, Russian,
+# Serbian, Serbian-Cyrillic, Slovak, Slovene, Spanish, Swedish, Turkish,
+# Ukrainian and Vietnamese.
+# The default value is: English.
+
+OUTPUT_LANGUAGE        = English
+
+# If the BRIEF_MEMBER_DESC tag is set to YES doxygen will include brief member
+# descriptions after the members that are listed in the file and class
+# documentation (similar to Javadoc). Set to NO to disable this.
+# The default value is: YES.
+
+BRIEF_MEMBER_DESC      = YES
+
+# If the REPEAT_BRIEF tag is set to YES doxygen will prepend the brief
+# description of a member or function before the detailed description
+#
+# Note: If both HIDE_UNDOC_MEMBERS and BRIEF_MEMBER_DESC are set to NO, the
+# brief descriptions will be completely suppressed.
+# The default value is: YES.
+
+REPEAT_BRIEF           = YES
+
+# This tag implements a quasi-intelligent brief description abbreviator that is
+# used to form the text in various listings. Each string in this list, if found
+# as the leading text of the brief description, will be stripped from the text
+# and the result, after processing the whole list, is used as the annotated
+# text. Otherwise, the brief description is used as-is. If left blank, the
+# following values are used ($name is automatically replaced with the name of
+# the entity):The $name class, The $name widget, The $name file, is, provides,
+# specifies, contains, represents, a, an and the.
+
+ABBREVIATE_BRIEF       =
+
+# If the ALWAYS_DETAILED_SEC and REPEAT_BRIEF tags are both set to YES then
+# doxygen will generate a detailed section even if there is only a brief
+# description.
+# The default value is: NO.
+
+ALWAYS_DETAILED_SEC    = NO
+
+# If the INLINE_INHERITED_MEMB tag is set to YES, doxygen will show all
+# inherited members of a class in the documentation of that class as if those
+# members were ordinary class members. Constructors, destructors and assignment
+# operators of the base classes will not be shown.
+# The default value is: NO.
+
+INLINE_INHERITED_MEMB  = NO
+
+# If the FULL_PATH_NAMES tag is set to YES doxygen will prepend the full path
+# before files name in the file list and in the header files. If set to NO the
+# shortest path that makes the file name unique will be used
+# The default value is: YES.
+
+FULL_PATH_NAMES        = YES
+
+# The STRIP_FROM_PATH tag can be used to strip a user-defined part of the path.
+# Stripping is only done if one of the specified strings matches the left-hand
+# part of the path. The tag can be used to show relative paths in the file list.
+# If left blank the directory from which doxygen is run is used as the path to
+# strip.
+#
+# Note that you can specify absolute paths here, but also relative paths, which
+# will be relative from the directory where doxygen is started.
+# This tag requires that the tag FULL_PATH_NAMES is set to YES.
+
+STRIP_FROM_PATH        =
+
+# The STRIP_FROM_INC_PATH tag can be used to strip a user-defined part of the
+# path mentioned in the documentation of a class, which tells the reader which
+# header file to include in order to use a class. If left blank only the name of
+# the header file containing the class definition is used. Otherwise one should
+# specify the list of include paths that are normally passed to the compiler
+# using the -I flag.
+
+STRIP_FROM_INC_PATH    =
+
+# If the SHORT_NAMES tag is set to YES, doxygen will generate much shorter (but
+# less readable) file names. This can be useful is your file systems doesn't
+# support long names like on DOS, Mac, or CD-ROM.
+# The default value is: NO.
+
+SHORT_NAMES            = NO
+
+# If the JAVADOC_AUTOBRIEF tag is set to YES then doxygen will interpret the
+# first line (until the first dot) of a Javadoc-style comment as the brief
+# description. If set to NO, the Javadoc-style will behave just like regular Qt-
+# style comments (thus requiring an explicit @brief command for a brief
+# description.)
+# The default value is: NO.
+
+JAVADOC_AUTOBRIEF      = NO
+
+# If the QT_AUTOBRIEF tag is set to YES then doxygen will interpret the first
+# line (until the first dot) of a Qt-style comment as the brief description. If
+# set to NO, the Qt-style will behave just like regular Qt-style comments (thus
+# requiring an explicit \brief command for a brief description.)
+# The default value is: NO.
+
+QT_AUTOBRIEF           = NO
+
+# The MULTILINE_CPP_IS_BRIEF tag can be set to YES to make doxygen treat a
+# multi-line C++ special comment block (i.e. a block of //! or /// comments) as
+# a brief description. This used to be the default behavior. The new default is
+# to treat a multi-line C++ comment block as a detailed description. Set this
+# tag to YES if you prefer the old behavior instead.
+#
+# Note that setting this tag to YES also means that rational rose comments are
+# not recognized any more.
+# The default value is: NO.
+
+MULTILINE_CPP_IS_BRIEF = NO
+
+# If the INHERIT_DOCS tag is set to YES then an undocumented member inherits the
+# documentation from any documented member that it re-implements.
+# The default value is: YES.
+
+INHERIT_DOCS           = YES
+
+# If the SEPARATE_MEMBER_PAGES tag is set to YES, then doxygen will produce a
+# new page for each member. If set to NO, the documentation of a member will be
+# part of the file/class/namespace that contains it.
+# The default value is: NO.
+
+SEPARATE_MEMBER_PAGES  = NO
+
+# The TAB_SIZE tag can be used to set the number of spaces in a tab. Doxygen
+# uses this value to replace tabs by spaces in code fragments.
+# Minimum value: 1, maximum value: 16, default value: 4.
+
+TAB_SIZE               = 4
+
+# This tag can be used to specify a number of aliases that act as commands in
+# the documentation. An alias has the form:
+# name=value
+# For example adding
+# "sideeffect=@par Side Effects:\n"
+# will allow you to put the command \sideeffect (or @sideeffect) in the
+# documentation, which will result in a user-defined paragraph with heading
+# "Side Effects:". You can put \n's in the value part of an alias to insert
+# newlines.
+
+ALIASES                =
+
+# This tag can be used to specify a number of word-keyword mappings (TCL only).
+# A mapping has the form "name=value". For example adding "class=itcl::class"
+# will allow you to use the command class in the itcl::class meaning.
+
+TCL_SUBST              =
+
+# Set the OPTIMIZE_OUTPUT_FOR_C tag to YES if your project consists of C sources
+# only. Doxygen will then generate output that is more tailored for C. For
+# instance, some of the names that are used will be different. The list of all
+# members will be omitted, etc.
+# The default value is: NO.
+
+OPTIMIZE_OUTPUT_FOR_C  = NO
+
+# Set the OPTIMIZE_OUTPUT_JAVA tag to YES if your project consists of Java or
+# Python sources only. Doxygen will then generate output that is more tailored
+# for that language. For instance, namespaces will be presented as packages,
+# qualified scopes will look different, etc.
+# The default value is: NO.
+
+OPTIMIZE_OUTPUT_JAVA   = NO
+
+# Set the OPTIMIZE_FOR_FORTRAN tag to YES if your project consists of Fortran
+# sources. Doxygen will then generate output that is tailored for Fortran.
+# The default value is: NO.
+
+OPTIMIZE_FOR_FORTRAN   = NO
+
+# Set the OPTIMIZE_OUTPUT_VHDL tag to YES if your project consists of VHDL
+# sources. Doxygen will then generate output that is tailored for VHDL.
+# The default value is: NO.
+
+OPTIMIZE_OUTPUT_VHDL   = NO
+
+# Doxygen selects the parser to use depending on the extension of the files it
+# parses. With this tag you can assign which parser to use for a given
+# extension. Doxygen has a built-in mapping, but you can override or extend it
+# using this tag. The format is ext=language, where ext is a file extension, and
+# language is one of the parsers supported by doxygen: IDL, Java, Javascript,
+# C#, C, C++, D, PHP, Objective-C, Python, Fortran, VHDL. For instance to make
+# doxygen treat .inc files as Fortran files (default is PHP), and .f files as C
+# (default is Fortran), use: inc=Fortran f=C.
+#
+# Note For files without extension you can use no_extension as a placeholder.
+#
+# Note that for custom extensions you also need to set FILE_PATTERNS otherwise
+# the files are not read by doxygen.
+
+EXTENSION_MAPPING      =
+
+# If the MARKDOWN_SUPPORT tag is enabled then doxygen pre-processes all comments
+# according to the Markdown format, which allows for more readable
+# documentation. See http://daringfireball.net/projects/markdown/ for details.
+# The output of markdown processing is further processed by doxygen, so you can
+# mix doxygen, HTML, and XML commands with Markdown formatting. Disable only in
+# case of backward compatibilities issues.
+# The default value is: YES.
+
+MARKDOWN_SUPPORT       = YES
+
+# When enabled doxygen tries to link words that correspond to documented
+# classes, or namespaces to their corresponding documentation. Such a link can
+# be prevented in individual cases by by putting a % sign in front of the word
+# or globally by setting AUTOLINK_SUPPORT to NO.
+# The default value is: YES.
+
+AUTOLINK_SUPPORT       = YES
+
+# If you use STL classes (i.e. std::string, std::vector, etc.) but do not want
+# to include (a tag file for) the STL sources as input, then you should set this
+# tag to YES in order to let doxygen match functions declarations and
+# definitions whose arguments contain STL classes (e.g. func(std::string);
+# versus func(std::string) {}). This also make the inheritance and collaboration
+# diagrams that involve STL classes more complete and accurate.
+# The default value is: NO.
+
+BUILTIN_STL_SUPPORT    = YES
+
+# If you use Microsoft's C++/CLI language, you should set this option to YES to
+# enable parsing support.
+# The default value is: NO.
+
+CPP_CLI_SUPPORT        = NO
+
+# Set the SIP_SUPPORT tag to YES if your project consists of sip (see:
+# http://www.riverbankcomputing.co.uk/software/sip/intro) sources only. Doxygen
+# will parse them like normal C++ but will assume all classes use public instead
+# of private inheritance when no explicit protection keyword is present.
+# The default value is: NO.
+
+SIP_SUPPORT            = NO
+
+# For Microsoft's IDL there are propget and propput attributes to indicate
+# getter and setter methods for a property. Setting this option to YES will make
+# doxygen to replace the get and set methods by a property in the documentation.
+# This will only work if the methods are indeed getting or setting a simple
+# type. If this is not the case, or you want to show the methods anyway, you
+# should set this option to NO.
+# The default value is: YES.
+
+IDL_PROPERTY_SUPPORT   = YES
+
+# If member grouping is used in the documentation and the DISTRIBUTE_GROUP_DOC
+# tag is set to YES, then doxygen will reuse the documentation of the first
+# member in the group (if any) for the other members of the group. By default
+# all members of a group must be documented explicitly.
+# The default value is: NO.
+
+DISTRIBUTE_GROUP_DOC   = NO
+
+# Set the SUBGROUPING tag to YES to allow class member groups of the same type
+# (for instance a group of public functions) to be put as a subgroup of that
+# type (e.g. under the Public Functions section). Set it to NO to prevent
+# subgrouping. Alternatively, this can be done per class using the
+# \nosubgrouping command.
+# The default value is: YES.
+
+SUBGROUPING            = YES
+
+# When the INLINE_GROUPED_CLASSES tag is set to YES, classes, structs and unions
+# are shown inside the group in which they are included (e.g. using \ingroup)
+# instead of on a separate page (for HTML and Man pages) or section (for LaTeX
+# and RTF).
+#
+# Note that this feature does not work in combination with
+# SEPARATE_MEMBER_PAGES.
+# The default value is: NO.
+
+INLINE_GROUPED_CLASSES = NO
+
+# When the INLINE_SIMPLE_STRUCTS tag is set to YES, structs, classes, and unions
+# with only public data fields or simple typedef fields will be shown inline in
+# the documentation of the scope in which they are defined (i.e. file,
+# namespace, or group documentation), provided this scope is documented. If set
+# to NO, structs, classes, and unions are shown on a separate page (for HTML and
+# Man pages) or section (for LaTeX and RTF).
+# The default value is: NO.
+
+INLINE_SIMPLE_STRUCTS  = NO
+
+# When TYPEDEF_HIDES_STRUCT tag is enabled, a typedef of a struct, union, or
+# enum is documented as struct, union, or enum with the name of the typedef. So
+# typedef struct TypeS {} TypeT, will appear in the documentation as a struct
+# with name TypeT. When disabled the typedef will appear as a member of a file,
+# namespace, or class. And the struct will be named TypeS. This can typically be
+# useful for C code in case the coding convention dictates that all compound
+# types are typedef'ed and only the typedef is referenced, never the tag name.
+# The default value is: NO.
+
+TYPEDEF_HIDES_STRUCT   = NO
+
+# The size of the symbol lookup cache can be set using LOOKUP_CACHE_SIZE. This
+# cache is used to resolve symbols given their name and scope. Since this can be
+# an expensive process and often the same symbol appears multiple times in the
+# code, doxygen keeps a cache of pre-resolved symbols. If the cache is too small
+# doxygen will become slower. If the cache is too large, memory is wasted. The
+# cache size is given by this formula: 2^(16+LOOKUP_CACHE_SIZE). The valid range
+# is 0..9, the default is 0, corresponding to a cache size of 2^16=65536
+# symbols. At the end of a run doxygen will report the cache usage and suggest
+# the optimal cache size from a speed point of view.
+# Minimum value: 0, maximum value: 9, default value: 0.
+
+LOOKUP_CACHE_SIZE      = 0
+
+#---------------------------------------------------------------------------
+# Build related configuration options
+#---------------------------------------------------------------------------
+
+# If the EXTRACT_ALL tag is set to YES doxygen will assume all entities in
+# documentation are documented, even if no documentation was available. Private
+# class members and static file members will be hidden unless the
+# EXTRACT_PRIVATE respectively EXTRACT_STATIC tags are set to YES.
+# Note: This will also disable the warnings about undocumented members that are
+# normally produced when WARNINGS is set to YES.
+# The default value is: NO.
+
+EXTRACT_ALL            = YES
+
+# If the EXTRACT_PRIVATE tag is set to YES all private members of a class will
+# be included in the documentation.
+# The default value is: NO.
+
+EXTRACT_PRIVATE        = YES
+
+# If the EXTRACT_PACKAGE tag is set to YES all members with package or internal
+# scope will be included in the documentation.
+# The default value is: NO.
+
+EXTRACT_PACKAGE        = YES
+
+# If the EXTRACT_STATIC tag is set to YES all static members of a file will be
+# included in the documentation.
+# The default value is: NO.
+
+EXTRACT_STATIC         = YES
+
+# If the EXTRACT_LOCAL_CLASSES tag is set to YES classes (and structs) defined
+# locally in source files will be included in the documentation. If set to NO
+# only classes defined in header files are included. Does not have any effect
+# for Java sources.
+# The default value is: YES.
+
+EXTRACT_LOCAL_CLASSES  = YES
+
+# This flag is only useful for Objective-C code. When set to YES local methods,
+# which are defined in the implementation section but not in the interface are
+# included in the documentation. If set to NO only methods in the interface are
+# included.
+# The default value is: NO.
+
+EXTRACT_LOCAL_METHODS  = YES
+
+# If this flag is set to YES, the members of anonymous namespaces will be
+# extracted and appear in the documentation as a namespace called
+# 'anonymous_namespace{file}', where file will be replaced with the base name of
+# the file that contains the anonymous namespace. By default anonymous namespace
+# are hidden.
+# The default value is: NO.
+
+EXTRACT_ANON_NSPACES   = YES
+
+# If the HIDE_UNDOC_MEMBERS tag is set to YES, doxygen will hide all
+# undocumented members inside documented classes or files. If set to NO these
+# members will be included in the various overviews, but no documentation
+# section is generated. This option has no effect if EXTRACT_ALL is enabled.
+# The default value is: NO.
+
+HIDE_UNDOC_MEMBERS     = NO
+
+# If the HIDE_UNDOC_CLASSES tag is set to YES, doxygen will hide all
+# undocumented classes that are normally visible in the class hierarchy. If set
+# to NO these classes will be included in the various overviews. This option has
+# no effect if EXTRACT_ALL is enabled.
+# The default value is: NO.
+
+HIDE_UNDOC_CLASSES     = NO
+
+# If the HIDE_FRIEND_COMPOUNDS tag is set to YES, doxygen will hide all friend
+# (class|struct|union) declarations. If set to NO these declarations will be
+# included in the documentation.
+# The default value is: NO.
+
+HIDE_FRIEND_COMPOUNDS  = NO
+
+# If the HIDE_IN_BODY_DOCS tag is set to YES, doxygen will hide any
+# documentation blocks found inside the body of a function. If set to NO these
+# blocks will be appended to the function's detailed documentation block.
+# The default value is: NO.
+
+HIDE_IN_BODY_DOCS      = NO
+
+# The INTERNAL_DOCS tag determines if documentation that is typed after a
+# \internal command is included. If the tag is set to NO then the documentation
+# will be excluded. Set it to YES to include the internal documentation.
+# The default value is: NO.
+
+INTERNAL_DOCS          = NO
+
+# If the CASE_SENSE_NAMES tag is set to NO then doxygen will only generate file
+# names in lower-case letters. If set to YES upper-case letters are also
+# allowed. This is useful if you have classes or files whose names only differ
+# in case and if your file system supports case sensitive file names. Windows
+# and Mac users are advised to set this option to NO.
+# The default value is: system dependent.
+
+CASE_SENSE_NAMES       = YES
+
+# If the HIDE_SCOPE_NAMES tag is set to NO then doxygen will show members with
+# their full class and namespace scopes in the documentation. If set to YES the
+# scope will be hidden.
+# The default value is: NO.
+
+HIDE_SCOPE_NAMES       = NO
+
+# If the SHOW_INCLUDE_FILES tag is set to YES then doxygen will put a list of
+# the files that are included by a file in the documentation of that file.
+# The default value is: YES.
+
+SHOW_INCLUDE_FILES     = YES
+
+# If the SHOW_GROUPED_MEMB_INC tag is set to YES then Doxygen will add for each
+# grouped member an include statement to the documentation, telling the reader
+# which file to include in order to use the member.
+# The default value is: NO.
+
+SHOW_GROUPED_MEMB_INC  = NO
+
+# If the FORCE_LOCAL_INCLUDES tag is set to YES then doxygen will list include
+# files with double quotes in the documentation rather than with sharp brackets.
+# The default value is: NO.
+
+FORCE_LOCAL_INCLUDES   = NO
+
+# If the INLINE_INFO tag is set to YES then a tag [inline] is inserted in the
+# documentation for inline members.
+# The default value is: YES.
+
+INLINE_INFO            = YES
+
+# If the SORT_MEMBER_DOCS tag is set to YES then doxygen will sort the
+# (detailed) documentation of file and class members alphabetically by member
+# name. If set to NO the members will appear in declaration order.
+# The default value is: YES.
+
+SORT_MEMBER_DOCS       = YES
+
+# If the SORT_BRIEF_DOCS tag is set to YES then doxygen will sort the brief
+# descriptions of file, namespace and class members alphabetically by member
+# name. If set to NO the members will appear in declaration order. Note that
+# this will also influence the order of the classes in the class list.
+# The default value is: NO.
+
+SORT_BRIEF_DOCS        = NO
+
+# If the SORT_MEMBERS_CTORS_1ST tag is set to YES then doxygen will sort the
+# (brief and detailed) documentation of class members so that constructors and
+# destructors are listed first. If set to NO the constructors will appear in the
+# respective orders defined by SORT_BRIEF_DOCS and SORT_MEMBER_DOCS.
+# Note: If SORT_BRIEF_DOCS is set to NO this option is ignored for sorting brief
+# member documentation.
+# Note: If SORT_MEMBER_DOCS is set to NO this option is ignored for sorting
+# detailed member documentation.
+# The default value is: NO.
+
+SORT_MEMBERS_CTORS_1ST = YES
+
+# If the SORT_GROUP_NAMES tag is set to YES then doxygen will sort the hierarchy
+# of group names into alphabetical order. If set to NO the group names will
+# appear in their defined order.
+# The default value is: NO.
+
+SORT_GROUP_NAMES       = NO
+
+# If the SORT_BY_SCOPE_NAME tag is set to YES, the class list will be sorted by
+# fully-qualified names, including namespaces. If set to NO, the class list will
+# be sorted only by class name, not including the namespace part.
+# Note: This option is not very useful if HIDE_SCOPE_NAMES is set to YES.
+# Note: This option applies only to the class list, not to the alphabetical
+# list.
+# The default value is: NO.
+
+SORT_BY_SCOPE_NAME     = NO
+
+# If the STRICT_PROTO_MATCHING option is enabled and doxygen fails to do proper
+# type resolution of all parameters of a function it will reject a match between
+# the prototype and the implementation of a member function even if there is
+# only one candidate or it is obvious which candidate to choose by doing a
+# simple string match. By disabling STRICT_PROTO_MATCHING doxygen will still
+# accept a match between prototype and implementation in such cases.
+# The default value is: NO.
+
+STRICT_PROTO_MATCHING  = NO
+
+# The GENERATE_TODOLIST tag can be used to enable ( YES) or disable ( NO) the
+# todo list. This list is created by putting \todo commands in the
+# documentation.
+# The default value is: YES.
+
+GENERATE_TODOLIST      = YES
+
+# The GENERATE_TESTLIST tag can be used to enable ( YES) or disable ( NO) the
+# test list. This list is created by putting \test commands in the
+# documentation.
+# The default value is: YES.
+
+GENERATE_TESTLIST      = YES
+
+# The GENERATE_BUGLIST tag can be used to enable ( YES) or disable ( NO) the bug
+# list. This list is created by putting \bug commands in the documentation.
+# The default value is: YES.
+
+GENERATE_BUGLIST       = YES
+
+# The GENERATE_DEPRECATEDLIST tag can be used to enable ( YES) or disable ( NO)
+# the deprecated list. This list is created by putting \deprecated commands in
+# the documentation.
+# The default value is: YES.
+
+GENERATE_DEPRECATEDLIST= YES
+
+# The ENABLED_SECTIONS tag can be used to enable conditional documentation
+# sections, marked by \if <section_label> ... \endif and \cond <section_label>
+# ... \endcond blocks.
+
+ENABLED_SECTIONS       =
+
+# The MAX_INITIALIZER_LINES tag determines the maximum number of lines that the
+# initial value of a variable or macro / define can have for it to appear in the
+# documentation. If the initializer consists of more lines than specified here
+# it will be hidden. Use a value of 0 to hide initializers completely. The
+# appearance of the value of individual variables and macros / defines can be
+# controlled using \showinitializer or \hideinitializer command in the
+# documentation regardless of this setting.
+# Minimum value: 0, maximum value: 10000, default value: 30.
+
+MAX_INITIALIZER_LINES  = 30
+
+# Set the SHOW_USED_FILES tag to NO to disable the list of files generated at
+# the bottom of the documentation of classes and structs. If set to YES the list
+# will mention the files that were used to generate the documentation.
+# The default value is: YES.
+
+SHOW_USED_FILES        = YES
+
+# Set the SHOW_FILES tag to NO to disable the generation of the Files page. This
+# will remove the Files entry from the Quick Index and from the Folder Tree View
+# (if specified).
+# The default value is: YES.
+
+SHOW_FILES             = YES
+
+# Set the SHOW_NAMESPACES tag to NO to disable the generation of the Namespaces
+# page. This will remove the Namespaces entry from the Quick Index and from the
+# Folder Tree View (if specified).
+# The default value is: YES.
+
+SHOW_NAMESPACES        = YES
+
+# The FILE_VERSION_FILTER tag can be used to specify a program or script that
+# doxygen should invoke to get the current version for each file (typically from
+# the version control system). Doxygen will invoke the program by executing (via
+# popen()) the command command input-file, where command is the value of the
+# FILE_VERSION_FILTER tag, and input-file is the name of an input file provided
+# by doxygen. Whatever the program writes to standard output is used as the file
+# version. For an example see the documentation.
+
+FILE_VERSION_FILTER    =
+
+# The LAYOUT_FILE tag can be used to specify a layout file which will be parsed
+# by doxygen. The layout file controls the global structure of the generated
+# output files in an output format independent way. To create the layout file
+# that represents doxygen's defaults, run doxygen with the -l option. You can
+# optionally specify a file name after the option, if omitted DoxygenLayout.xml
+# will be used as the name of the layout file.
+#
+# Note that if you run doxygen from a directory containing a file called
+# DoxygenLayout.xml, doxygen will parse it automatically even if the LAYOUT_FILE
+# tag is left empty.
+
+LAYOUT_FILE            =
+
+# The CITE_BIB_FILES tag can be used to specify one or more bib files containing
+# the reference definitions. This must be a list of .bib files. The .bib
+# extension is automatically appended if omitted. This requires the bibtex tool
+# to be installed. See also http://en.wikipedia.org/wiki/BibTeX for more info.
+# For LaTeX the style of the bibliography can be controlled using
+# LATEX_BIB_STYLE. To use this feature you need bibtex and perl available in the
+# search path. Do not use file names with spaces, bibtex cannot handle them. See
+# also \cite for info how to create references.
+
+CITE_BIB_FILES         =
+
+#---------------------------------------------------------------------------
+# Configuration options related to warning and progress messages
+#---------------------------------------------------------------------------
+
+# The QUIET tag can be used to turn on/off the messages that are generated to
+# standard output by doxygen. If QUIET is set to YES this implies that the
+# messages are off.
+# The default value is: NO.
+
+QUIET                  = NO
+
+# The WARNINGS tag can be used to turn on/off the warning messages that are
+# generated to standard error ( stderr) by doxygen. If WARNINGS is set to YES
+# this implies that the warnings are on.
+#
+# Tip: Turn warnings on while writing the documentation.
+# The default value is: YES.
+
+WARNINGS               = YES
+
+# If the WARN_IF_UNDOCUMENTED tag is set to YES, then doxygen will generate
+# warnings for undocumented members. If EXTRACT_ALL is set to YES then this flag
+# will automatically be disabled.
+# The default value is: YES.
+
+WARN_IF_UNDOCUMENTED   = YES
+
+# If the WARN_IF_DOC_ERROR tag is set to YES, doxygen will generate warnings for
+# potential errors in the documentation, such as not documenting some parameters
+# in a documented function, or documenting parameters that don't exist or using
+# markup commands wrongly.
+# The default value is: YES.
+
+WARN_IF_DOC_ERROR      = YES
+
+# This WARN_NO_PARAMDOC option can be enabled to get warnings for functions that
+# are documented, but have no documentation for their parameters or return
+# value. If set to NO doxygen will only warn about wrong or incomplete parameter
+# documentation, but not about the absence of documentation.
+# The default value is: NO.
+
+WARN_NO_PARAMDOC       = NO
+
+# The WARN_FORMAT tag determines the format of the warning messages that doxygen
+# can produce. The string should contain the $file, $line, and $text tags, which
+# will be replaced by the file and line number from which the warning originated
+# and the warning text. Optionally the format may contain $version, which will
+# be replaced by the version of the file (if it could be obtained via
+# FILE_VERSION_FILTER)
+# The default value is: $file:$line: $text.
+
+WARN_FORMAT            = "$file:$line: $text"
+
+# The WARN_LOGFILE tag can be used to specify a file to which warning and error
+# messages should be written. If left blank the output is written to standard
+# error (stderr).
+
+WARN_LOGFILE           =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the input files
+#---------------------------------------------------------------------------
+
+# The INPUT tag is used to specify the files and/or directories that contain
+# documented source files. You may enter file names like myfile.cpp or
+# directories like /usr/src/myproject. Separate the files or directories with
+# spaces.
+# Note: If this tag is empty the current directory is searched.
+
+INPUT                  = ../lib \
+		         ../tests \
+			 ../benchmarks
+
+# This tag can be used to specify the character encoding of the source files
+# that doxygen parses. Internally doxygen uses the UTF-8 encoding. Doxygen uses
+# libiconv (or the iconv built into libc) for the transcoding. See the libiconv
+# documentation (see: http://www.gnu.org/software/libiconv) for the list of
+# possible encodings.
+# The default value is: UTF-8.
+
+INPUT_ENCODING         = UTF-8
+
+# If the value of the INPUT tag contains directories, you can use the
+# FILE_PATTERNS tag to specify one or more wildcard patterns (like *.cpp and
+# *.h) to filter out the source-files in the directories. If left blank the
+# following patterns are tested:*.c, *.cc, *.cxx, *.cpp, *.c++, *.java, *.ii,
+# *.ixx, *.ipp, *.i++, *.inl, *.idl, *.ddl, *.odl, *.h, *.hh, *.hxx, *.hpp,
+# *.h++, *.cs, *.d, *.php, *.php4, *.php5, *.phtml, *.inc, *.m, *.markdown,
+# *.md, *.mm, *.dox, *.py, *.f90, *.f, *.for, *.tcl, *.vhd, *.vhdl, *.ucf,
+# *.qsf, *.as and *.js.
+
+FILE_PATTERNS          =
+
+# The RECURSIVE tag can be used to specify whether or not subdirectories should
+# be searched for input files as well.
+# The default value is: NO.
+
+RECURSIVE              = YES
+
+# The EXCLUDE tag can be used to specify files and/or directories that should be
+# excluded from the INPUT source files. This way you can easily exclude a
+# subdirectory from a directory tree whose root is specified with the INPUT tag.
+#
+# Note that relative paths are relative to the directory from which doxygen is
+# run.
+
+EXCLUDE                =
+
+# The EXCLUDE_SYMLINKS tag can be used to select whether or not files or
+# directories that are symbolic links (a Unix file system feature) are excluded
+# from the input.
+# The default value is: NO.
+
+EXCLUDE_SYMLINKS       = NO
+
+# If the value of the INPUT tag contains directories, you can use the
+# EXCLUDE_PATTERNS tag to specify one or more wildcard patterns to exclude
+# certain files from those directories.
+#
+# Note that the wildcards are matched against the file with absolute path, so to
+# exclude all test directories for example use the pattern */test/*
+
+EXCLUDE_PATTERNS       =
+
+# The EXCLUDE_SYMBOLS tag can be used to specify one or more symbol names
+# (namespaces, classes, functions, etc.) that should be excluded from the
+# output. The symbol name can be a fully qualified name, a word, or if the
+# wildcard * is used, a substring. Examples: ANamespace, AClass,
+# AClass::ANamespace, ANamespace::*Test
+#
+# Note that the wildcards are matched against the file with absolute path, so to
+# exclude all test directories use the pattern */test/*
+
+EXCLUDE_SYMBOLS        =
+
+# The EXAMPLE_PATH tag can be used to specify one or more files or directories
+# that contain example code fragments that are included (see the \include
+# command).
+
+EXAMPLE_PATH           =
+
+# If the value of the EXAMPLE_PATH tag contains directories, you can use the
+# EXAMPLE_PATTERNS tag to specify one or more wildcard pattern (like *.cpp and
+# *.h) to filter out the source-files in the directories. If left blank all
+# files are included.
+
+EXAMPLE_PATTERNS       =
+
+# If the EXAMPLE_RECURSIVE tag is set to YES then subdirectories will be
+# searched for input files to be used with the \include or \dontinclude commands
+# irrespective of the value of the RECURSIVE tag.
+# The default value is: NO.
+
+EXAMPLE_RECURSIVE      = NO
+
+# The IMAGE_PATH tag can be used to specify one or more files or directories
+# that contain images that are to be included in the documentation (see the
+# \image command).
+
+IMAGE_PATH             =
+
+# The INPUT_FILTER tag can be used to specify a program that doxygen should
+# invoke to filter for each input file. Doxygen will invoke the filter program
+# by executing (via popen()) the command:
+#
+# <filter> <input-file>
+#
+# where <filter> is the value of the INPUT_FILTER tag, and <input-file> is the
+# name of an input file. Doxygen will then use the output that the filter
+# program writes to standard output. If FILTER_PATTERNS is specified, this tag
+# will be ignored.
+#
+# Note that the filter must not add or remove lines; it is applied before the
+# code is scanned, but not when the output code is generated. If lines are added
+# or removed, the anchors will not be placed correctly.
+
+INPUT_FILTER           =
+
+# The FILTER_PATTERNS tag can be used to specify filters on a per file pattern
+# basis. Doxygen will compare the file name with each pattern and apply the
+# filter if there is a match. The filters are a list of the form: pattern=filter
+# (like *.cpp=my_cpp_filter). See INPUT_FILTER for further information on how
+# filters are used. If the FILTER_PATTERNS tag is empty or if none of the
+# patterns match the file name, INPUT_FILTER is applied.
+
+FILTER_PATTERNS        =
+
+# If the FILTER_SOURCE_FILES tag is set to YES, the input filter (if set using
+# INPUT_FILTER ) will also be used to filter the input files that are used for
+# producing the source files to browse (i.e. when SOURCE_BROWSER is set to YES).
+# The default value is: NO.
+
+FILTER_SOURCE_FILES    = NO
+
+# The FILTER_SOURCE_PATTERNS tag can be used to specify source filters per file
+# pattern. A pattern will override the setting for FILTER_PATTERN (if any) and
+# it is also possible to disable source filtering for a specific pattern using
+# *.ext= (so without naming a filter).
+# This tag requires that the tag FILTER_SOURCE_FILES is set to YES.
+
+FILTER_SOURCE_PATTERNS =
+
+# If the USE_MDFILE_AS_MAINPAGE tag refers to the name of a markdown file that
+# is part of the input, its contents will be placed on the main page
+# (index.html). This can be useful if you have a project on for instance GitHub
+# and want to reuse the introduction page also for the doxygen output.
+
+USE_MDFILE_AS_MAINPAGE =
+
+#---------------------------------------------------------------------------
+# Configuration options related to source browsing
+#---------------------------------------------------------------------------
+
+# If the SOURCE_BROWSER tag is set to YES then a list of source files will be
+# generated. Documented entities will be cross-referenced with these sources.
+#
+# Note: To get rid of all source code in the generated output, make sure that
+# also VERBATIM_HEADERS is set to NO.
+# The default value is: NO.
+
+SOURCE_BROWSER         = YES
+
+# Setting the INLINE_SOURCES tag to YES will include the body of functions,
+# classes and enums directly into the documentation.
+# The default value is: NO.
+
+INLINE_SOURCES         = NO
+
+# Setting the STRIP_CODE_COMMENTS tag to YES will instruct doxygen to hide any
+# special comment blocks from generated source code fragments. Normal C, C++ and
+# Fortran comments will always remain visible.
+# The default value is: YES.
+
+STRIP_CODE_COMMENTS    = YES
+
+# If the REFERENCED_BY_RELATION tag is set to YES then for each documented
+# function all documented functions referencing it will be listed.
+# The default value is: NO.
+
+REFERENCED_BY_RELATION = NO
+
+# If the REFERENCES_RELATION tag is set to YES then for each documented function
+# all documented entities called/used by that function will be listed.
+# The default value is: NO.
+
+REFERENCES_RELATION    = NO
+
+# If the REFERENCES_LINK_SOURCE tag is set to YES and SOURCE_BROWSER tag is set
+# to YES, then the hyperlinks from functions in REFERENCES_RELATION and
+# REFERENCED_BY_RELATION lists will link to the source code. Otherwise they will
+# link to the documentation.
+# The default value is: YES.
+
+REFERENCES_LINK_SOURCE = YES
+
+# If SOURCE_TOOLTIPS is enabled (the default) then hovering a hyperlink in the
+# source code will show a tooltip with additional information such as prototype,
+# brief description and links to the definition and documentation. Since this
+# will make the HTML file larger and loading of large files a bit slower, you
+# can opt to disable this feature.
+# The default value is: YES.
+# This tag requires that the tag SOURCE_BROWSER is set to YES.
+
+SOURCE_TOOLTIPS        = YES
+
+# If the USE_HTAGS tag is set to YES then the references to source code will
+# point to the HTML generated by the htags(1) tool instead of doxygen built-in
+# source browser. The htags tool is part of GNU's global source tagging system
+# (see http://www.gnu.org/software/global/global.html). You will need version
+# 4.8.6 or higher.
+#
+# To use it do the following:
+# - Install the latest version of global
+# - Enable SOURCE_BROWSER and USE_HTAGS in the config file
+# - Make sure the INPUT points to the root of the source tree
+# - Run doxygen as normal
+#
+# Doxygen will invoke htags (and that will in turn invoke gtags), so these
+# tools must be available from the command line (i.e. in the search path).
+#
+# The result: instead of the source browser generated by doxygen, the links to
+# source code will now point to the output of htags.
+# The default value is: NO.
+# This tag requires that the tag SOURCE_BROWSER is set to YES.
+
+USE_HTAGS              = NO
+
+# If the VERBATIM_HEADERS tag is set the YES then doxygen will generate a
+# verbatim copy of the header file for each class for which an include is
+# specified. Set to NO to disable this.
+# See also: Section \class.
+# The default value is: YES.
+
+VERBATIM_HEADERS       = YES
+
+#---------------------------------------------------------------------------
+# Configuration options related to the alphabetical class index
+#---------------------------------------------------------------------------
+
+# If the ALPHABETICAL_INDEX tag is set to YES, an alphabetical index of all
+# compounds will be generated. Enable this if the project contains a lot of
+# classes, structs, unions or interfaces.
+# The default value is: YES.
+
+ALPHABETICAL_INDEX     = YES
+
+# The COLS_IN_ALPHA_INDEX tag can be used to specify the number of columns in
+# which the alphabetical index list will be split.
+# Minimum value: 1, maximum value: 20, default value: 5.
+# This tag requires that the tag ALPHABETICAL_INDEX is set to YES.
+
+COLS_IN_ALPHA_INDEX    = 5
+
+# In case all classes in a project start with a common prefix, all classes will
+# be put under the same header in the alphabetical index. The IGNORE_PREFIX tag
+# can be used to specify a prefix (or a list of prefixes) that should be ignored
+# while generating the index headers.
+# This tag requires that the tag ALPHABETICAL_INDEX is set to YES.
+
+IGNORE_PREFIX          =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the HTML output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_HTML tag is set to YES doxygen will generate HTML output
+# The default value is: YES.
+
+GENERATE_HTML          = YES
+
+# The HTML_OUTPUT tag is used to specify where the HTML docs will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it.
+# The default directory is: html.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_OUTPUT            = html
+
+# The HTML_FILE_EXTENSION tag can be used to specify the file extension for each
+# generated HTML page (for example: .htm, .php, .asp).
+# The default value is: .html.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_FILE_EXTENSION    = .html
+
+# The HTML_HEADER tag can be used to specify a user-defined HTML header file for
+# each generated HTML page. If the tag is left blank doxygen will generate a
+# standard header.
+#
+# To get valid HTML the header file that includes any scripts and style sheets
+# that doxygen needs, which is dependent on the configuration options used (e.g.
+# the setting GENERATE_TREEVIEW). It is highly recommended to start with a
+# default header using
+# doxygen -w html new_header.html new_footer.html new_stylesheet.css
+# YourConfigFile
+# and then modify the file new_header.html. See also section "Doxygen usage"
+# for information on how to generate the default header that doxygen normally
+# uses.
+# Note: The header is subject to change so you typically have to regenerate the
+# default header when upgrading to a newer version of doxygen. For a description
+# of the possible markers and block names see the documentation.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_HEADER            =
+
+# The HTML_FOOTER tag can be used to specify a user-defined HTML footer for each
+# generated HTML page. If the tag is left blank doxygen will generate a standard
+# footer. See HTML_HEADER for more information on how to generate a default
+# footer and what special commands can be used inside the footer. See also
+# section "Doxygen usage" for information on how to generate the default footer
+# that doxygen normally uses.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_FOOTER            =
+
+# The HTML_STYLESHEET tag can be used to specify a user-defined cascading style
+# sheet that is used by each HTML page. It can be used to fine-tune the look of
+# the HTML output. If left blank doxygen will generate a default style sheet.
+# See also section "Doxygen usage" for information on how to generate the style
+# sheet that doxygen normally uses.
+# Note: It is recommended to use HTML_EXTRA_STYLESHEET instead of this tag, as
+# it is more robust and this tag (HTML_STYLESHEET) will in the future become
+# obsolete.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_STYLESHEET        =
+
+# The HTML_EXTRA_STYLESHEET tag can be used to specify an additional user-
+# defined cascading style sheet that is included after the standard style sheets
+# created by doxygen. Using this option one can overrule certain style aspects.
+# This is preferred over using HTML_STYLESHEET since it does not replace the
+# standard style sheet and is therefor more robust against future updates.
+# Doxygen will copy the style sheet file to the output directory. For an example
+# see the documentation.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_EXTRA_STYLESHEET  =
+
+# The HTML_EXTRA_FILES tag can be used to specify one or more extra images or
+# other source files which should be copied to the HTML output directory. Note
+# that these files will be copied to the base HTML output directory. Use the
+# $relpath^ marker in the HTML_HEADER and/or HTML_FOOTER files to load these
+# files. In the HTML_STYLESHEET file, use the file name only. Also note that the
+# files will be copied as-is; there are no commands or markers available.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_EXTRA_FILES       =
+
+# The HTML_COLORSTYLE_HUE tag controls the color of the HTML output. Doxygen
+# will adjust the colors in the stylesheet and background images according to
+# this color. Hue is specified as an angle on a colorwheel, see
+# http://en.wikipedia.org/wiki/Hue for more information. For instance the value
+# 0 represents red, 60 is yellow, 120 is green, 180 is cyan, 240 is blue, 300
+# purple, and 360 is red again.
+# Minimum value: 0, maximum value: 359, default value: 220.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_COLORSTYLE_HUE    = 220
+
+# The HTML_COLORSTYLE_SAT tag controls the purity (or saturation) of the colors
+# in the HTML output. For a value of 0 the output will use grayscales only. A
+# value of 255 will produce the most vivid colors.
+# Minimum value: 0, maximum value: 255, default value: 100.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_COLORSTYLE_SAT    = 100
+
+# The HTML_COLORSTYLE_GAMMA tag controls the gamma correction applied to the
+# luminance component of the colors in the HTML output. Values below 100
+# gradually make the output lighter, whereas values above 100 make the output
+# darker. The value divided by 100 is the actual gamma applied, so 80 represents
+# a gamma of 0.8, The value 220 represents a gamma of 2.2, and 100 does not
+# change the gamma.
+# Minimum value: 40, maximum value: 240, default value: 80.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_COLORSTYLE_GAMMA  = 80
+
+# If the HTML_TIMESTAMP tag is set to YES then the footer of each generated HTML
+# page will contain the date and time when the page was generated. Setting this
+# to NO can help when comparing the output of multiple runs.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_TIMESTAMP         = YES
+
+# If the HTML_DYNAMIC_SECTIONS tag is set to YES then the generated HTML
+# documentation will contain sections that can be hidden and shown after the
+# page has loaded.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_DYNAMIC_SECTIONS  = NO
+
+# With HTML_INDEX_NUM_ENTRIES one can control the preferred number of entries
+# shown in the various tree structured indices initially; the user can expand
+# and collapse entries dynamically later on. Doxygen will expand the tree to
+# such a level that at most the specified number of entries are visible (unless
+# a fully collapsed tree already exceeds this amount). So setting the number of
+# entries 1 will produce a full collapsed tree by default. 0 is a special value
+# representing an infinite number of entries and will result in a full expanded
+# tree by default.
+# Minimum value: 0, maximum value: 9999, default value: 100.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_INDEX_NUM_ENTRIES = 100
+
+# If the GENERATE_DOCSET tag is set to YES, additional index files will be
+# generated that can be used as input for Apple's Xcode 3 integrated development
+# environment (see: http://developer.apple.com/tools/xcode/), introduced with
+# OSX 10.5 (Leopard). To create a documentation set, doxygen will generate a
+# Makefile in the HTML output directory. Running make will produce the docset in
+# that directory and running make install will install the docset in
+# ~/Library/Developer/Shared/Documentation/DocSets so that Xcode will find it at
+# startup. See http://developer.apple.com/tools/creatingdocsetswithdoxygen.html
+# for more information.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_DOCSET        = NO
+
+# This tag determines the name of the docset feed. A documentation feed provides
+# an umbrella under which multiple documentation sets from a single provider
+# (such as a company or product suite) can be grouped.
+# The default value is: Doxygen generated docs.
+# This tag requires that the tag GENERATE_DOCSET is set to YES.
+
+DOCSET_FEEDNAME        = "Doxygen generated docs"
+
+# This tag specifies a string that should uniquely identify the documentation
+# set bundle. This should be a reverse domain-name style string, e.g.
+# com.mycompany.MyDocSet. Doxygen will append .docset to the name.
+# The default value is: org.doxygen.Project.
+# This tag requires that the tag GENERATE_DOCSET is set to YES.
+
+DOCSET_BUNDLE_ID       = org.doxygen.Project
+
+# The DOCSET_PUBLISHER_ID tag specifies a string that should uniquely identify
+# the documentation publisher. This should be a reverse domain-name style
+# string, e.g. com.mycompany.MyDocSet.documentation.
+# The default value is: org.doxygen.Publisher.
+# This tag requires that the tag GENERATE_DOCSET is set to YES.
+
+DOCSET_PUBLISHER_ID    = org.doxygen.Publisher
+
+# The DOCSET_PUBLISHER_NAME tag identifies the documentation publisher.
+# The default value is: Publisher.
+# This tag requires that the tag GENERATE_DOCSET is set to YES.
+
+DOCSET_PUBLISHER_NAME  = Publisher
+
+# If the GENERATE_HTMLHELP tag is set to YES then doxygen generates three
+# additional HTML index files: index.hhp, index.hhc, and index.hhk. The
+# index.hhp is a project file that can be read by Microsoft's HTML Help Workshop
+# (see: http://www.microsoft.com/en-us/download/details.aspx?id=21138) on
+# Windows.
+#
+# The HTML Help Workshop contains a compiler that can convert all HTML output
+# generated by doxygen into a single compiled HTML file (.chm). Compiled HTML
+# files are now used as the Windows 98 help format, and will replace the old
+# Windows help format (.hlp) on all Windows platforms in the future. Compressed
+# HTML files also contain an index, a table of contents, and you can search for
+# words in the documentation. The HTML workshop also contains a viewer for
+# compressed HTML files.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_HTMLHELP      = NO
+
+# The CHM_FILE tag can be used to specify the file name of the resulting .chm
+# file. You can add a path in front of the file if the result should not be
+# written to the html output directory.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+CHM_FILE               =
+
+# The HHC_LOCATION tag can be used to specify the location (absolute path
+# including file name) of the HTML help compiler ( hhc.exe). If non-empty
+# doxygen will try to run the HTML help compiler on the generated index.hhp.
+# The file has to be specified with full path.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+HHC_LOCATION           =
+
+# The GENERATE_CHI flag controls if a separate .chi index file is generated (
+# YES) or that it should be included in the master .chm file ( NO).
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+GENERATE_CHI           = NO
+
+# The CHM_INDEX_ENCODING is used to encode HtmlHelp index ( hhk), content ( hhc)
+# and project file content.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+CHM_INDEX_ENCODING     =
+
+# The BINARY_TOC flag controls whether a binary table of contents is generated (
+# YES) or a normal table of contents ( NO) in the .chm file.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+BINARY_TOC             = NO
+
+# The TOC_EXPAND flag can be set to YES to add extra items for group members to
+# the table of contents of the HTML help documentation and to the tree view.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+TOC_EXPAND             = NO
+
+# If the GENERATE_QHP tag is set to YES and both QHP_NAMESPACE and
+# QHP_VIRTUAL_FOLDER are set, an additional index file will be generated that
+# can be used as input for Qt's qhelpgenerator to generate a Qt Compressed Help
+# (.qch) of the generated HTML documentation.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_QHP           = NO
+
+# If the QHG_LOCATION tag is specified, the QCH_FILE tag can be used to specify
+# the file name of the resulting .qch file. The path specified is relative to
+# the HTML output folder.
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QCH_FILE               =
+
+# The QHP_NAMESPACE tag specifies the namespace to use when generating Qt Help
+# Project output. For more information please see Qt Help Project / Namespace
+# (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#namespace).
+# The default value is: org.doxygen.Project.
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_NAMESPACE          = org.doxygen.Project
+
+# The QHP_VIRTUAL_FOLDER tag specifies the namespace to use when generating Qt
+# Help Project output. For more information please see Qt Help Project / Virtual
+# Folders (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#virtual-
+# folders).
+# The default value is: doc.
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_VIRTUAL_FOLDER     = doc
+
+# If the QHP_CUST_FILTER_NAME tag is set, it specifies the name of a custom
+# filter to add. For more information please see Qt Help Project / Custom
+# Filters (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#custom-
+# filters).
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_CUST_FILTER_NAME   =
+
+# The QHP_CUST_FILTER_ATTRS tag specifies the list of the attributes of the
+# custom filter to add. For more information please see Qt Help Project / Custom
+# Filters (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#custom-
+# filters).
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_CUST_FILTER_ATTRS  =
+
+# The QHP_SECT_FILTER_ATTRS tag specifies the list of the attributes this
+# project's filter section matches. Qt Help Project / Filter Attributes (see:
+# http://qt-project.org/doc/qt-4.8/qthelpproject.html#filter-attributes).
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_SECT_FILTER_ATTRS  =
+
+# The QHG_LOCATION tag can be used to specify the location of Qt's
+# qhelpgenerator. If non-empty doxygen will try to run qhelpgenerator on the
+# generated .qhp file.
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHG_LOCATION           =
+
+# If the GENERATE_ECLIPSEHELP tag is set to YES, additional index files will be
+# generated, together with the HTML files, they form an Eclipse help plugin. To
+# install this plugin and make it available under the help contents menu in
+# Eclipse, the contents of the directory containing the HTML and XML files needs
+# to be copied into the plugins directory of eclipse. The name of the directory
+# within the plugins directory should be the same as the ECLIPSE_DOC_ID value.
+# After copying Eclipse needs to be restarted before the help appears.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_ECLIPSEHELP   = NO
+
+# A unique identifier for the Eclipse help plugin. When installing the plugin
+# the directory name containing the HTML and XML files should also have this
+# name. Each documentation set should have its own identifier.
+# The default value is: org.doxygen.Project.
+# This tag requires that the tag GENERATE_ECLIPSEHELP is set to YES.
+
+ECLIPSE_DOC_ID         = org.doxygen.Project
+
+# If you want full control over the layout of the generated HTML pages it might
+# be necessary to disable the index and replace it with your own. The
+# DISABLE_INDEX tag can be used to turn on/off the condensed index (tabs) at top
+# of each HTML page. A value of NO enables the index and the value YES disables
+# it. Since the tabs in the index contain the same information as the navigation
+# tree, you can set this option to YES if you also set GENERATE_TREEVIEW to YES.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+DISABLE_INDEX          = NO
+
+# The GENERATE_TREEVIEW tag is used to specify whether a tree-like index
+# structure should be generated to display hierarchical information. If the tag
+# value is set to YES, a side panel will be generated containing a tree-like
+# index structure (just like the one that is generated for HTML Help). For this
+# to work a browser that supports JavaScript, DHTML, CSS and frames is required
+# (i.e. any modern browser). Windows users are probably better off using the
+# HTML help feature. Via custom stylesheets (see HTML_EXTRA_STYLESHEET) one can
+# further fine-tune the look of the index. As an example, the default style
+# sheet generated by doxygen has an example that shows how to put an image at
+# the root of the tree instead of the PROJECT_NAME. Since the tree basically has
+# the same information as the tab index, you could consider setting
+# DISABLE_INDEX to YES when enabling this option.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_TREEVIEW      = YES
+
+# The ENUM_VALUES_PER_LINE tag can be used to set the number of enum values that
+# doxygen will group on one line in the generated HTML documentation.
+#
+# Note that a value of 0 will completely suppress the enum values from appearing
+# in the overview section.
+# Minimum value: 0, maximum value: 20, default value: 4.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+ENUM_VALUES_PER_LINE   = 4
+
+# If the treeview is enabled (see GENERATE_TREEVIEW) then this tag can be used
+# to set the initial width (in pixels) of the frame in which the tree is shown.
+# Minimum value: 0, maximum value: 1500, default value: 250.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+TREEVIEW_WIDTH         = 250
+
+# When the EXT_LINKS_IN_WINDOW option is set to YES doxygen will open links to
+# external symbols imported via tag files in a separate window.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+EXT_LINKS_IN_WINDOW    = NO
+
+# Use this tag to change the font size of LaTeX formulas included as images in
+# the HTML documentation. When you change the font size after a successful
+# doxygen run you need to manually remove any form_*.png images from the HTML
+# output directory to force them to be regenerated.
+# Minimum value: 8, maximum value: 50, default value: 10.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+FORMULA_FONTSIZE       = 10
+
+# Use the FORMULA_TRANPARENT tag to determine whether or not the images
+# generated for formulas are transparent PNGs. Transparent PNGs are not
+# supported properly for IE 6.0, but are supported on all modern browsers.
+#
+# Note that when changing this option you need to delete any form_*.png files in
+# the HTML output directory before the changes have effect.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+FORMULA_TRANSPARENT    = YES
+
+# Enable the USE_MATHJAX option to render LaTeX formulas using MathJax (see
+# http://www.mathjax.org) which uses client side Javascript for the rendering
+# instead of using prerendered bitmaps. Use this if you do not have LaTeX
+# installed or if you want to formulas look prettier in the HTML output. When
+# enabled you may also need to install MathJax separately and configure the path
+# to it using the MATHJAX_RELPATH option.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+USE_MATHJAX            = YES
+
+# When MathJax is enabled you can set the default output format to be used for
+# the MathJax output. See the MathJax site (see:
+# http://docs.mathjax.org/en/latest/output.html) for more details.
+# Possible values are: HTML-CSS (which is slower, but has the best
+# compatibility), NativeMML (i.e. MathML) and SVG.
+# The default value is: HTML-CSS.
+# This tag requires that the tag USE_MATHJAX is set to YES.
+
+MATHJAX_FORMAT         = HTML-CSS
+
+# When MathJax is enabled you need to specify the location relative to the HTML
+# output directory using the MATHJAX_RELPATH option. The destination directory
+# should contain the MathJax.js script. For instance, if the mathjax directory
+# is located at the same level as the HTML output directory, then
+# MATHJAX_RELPATH should be ../mathjax. The default value points to the MathJax
+# Content Delivery Network so you can quickly see the result without installing
+# MathJax. However, it is strongly recommended to install a local copy of
+# MathJax from http://www.mathjax.org before deployment.
+# The default value is: http://cdn.mathjax.org/mathjax/latest.
+# This tag requires that the tag USE_MATHJAX is set to YES.
+
+MATHJAX_RELPATH        = http://cdn.mathjax.org/mathjax/latest
+
+# The MATHJAX_EXTENSIONS tag can be used to specify one or more MathJax
+# extension names that should be enabled during MathJax rendering. For example
+# MATHJAX_EXTENSIONS = TeX/AMSmath TeX/AMSsymbols
+# This tag requires that the tag USE_MATHJAX is set to YES.
+
+MATHJAX_EXTENSIONS     =
+
+# The MATHJAX_CODEFILE tag can be used to specify a file with javascript pieces
+# of code that will be used on startup of the MathJax code. See the MathJax site
+# (see: http://docs.mathjax.org/en/latest/output.html) for more details. For an
+# example see the documentation.
+# This tag requires that the tag USE_MATHJAX is set to YES.
+
+MATHJAX_CODEFILE       =
+
+# When the SEARCHENGINE tag is enabled doxygen will generate a search box for
+# the HTML output. The underlying search engine uses javascript and DHTML and
+# should work on any modern browser. Note that when using HTML help
+# (GENERATE_HTMLHELP), Qt help (GENERATE_QHP), or docsets (GENERATE_DOCSET)
+# there is already a search function so this one should typically be disabled.
+# For large projects the javascript based search engine can be slow, then
+# enabling SERVER_BASED_SEARCH may provide a better solution. It is possible to
+# search using the keyboard; to jump to the search box use <access key> + S
+# (what the <access key> is depends on the OS and browser, but it is typically
+# <CTRL>, <ALT>/<option>, or both). Inside the search box use the <cursor down
+# key> to jump into the search results window, the results can be navigated
+# using the <cursor keys>. Press <Enter> to select an item or <escape> to cancel
+# the search. The filter options can be selected when the cursor is inside the
+# search box by pressing <Shift>+<cursor down>. Also here use the <cursor keys>
+# to select a filter and <Enter> or <escape> to activate or cancel the filter
+# option.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+SEARCHENGINE           = YES
+
+# When the SERVER_BASED_SEARCH tag is enabled the search engine will be
+# implemented using a web server instead of a web client using Javascript. There
+# are two flavours of web server based searching depending on the
+# EXTERNAL_SEARCH setting. When disabled, doxygen will generate a PHP script for
+# searching and an index file used by the script. When EXTERNAL_SEARCH is
+# enabled the indexing and searching needs to be provided by external tools. See
+# the section "External Indexing and Searching" for details.
+# The default value is: NO.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+SERVER_BASED_SEARCH    = NO
+
+# When EXTERNAL_SEARCH tag is enabled doxygen will no longer generate the PHP
+# script for searching. Instead the search results are written to an XML file
+# which needs to be processed by an external indexer. Doxygen will invoke an
+# external search engine pointed to by the SEARCHENGINE_URL option to obtain the
+# search results.
+#
+# Doxygen ships with an example indexer ( doxyindexer) and search engine
+# (doxysearch.cgi) which are based on the open source search engine library
+# Xapian (see: http://xapian.org/).
+#
+# See the section "External Indexing and Searching" for details.
+# The default value is: NO.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+EXTERNAL_SEARCH        = NO
+
+# The SEARCHENGINE_URL should point to a search engine hosted by a web server
+# which will return the search results when EXTERNAL_SEARCH is enabled.
+#
+# Doxygen ships with an example indexer ( doxyindexer) and search engine
+# (doxysearch.cgi) which are based on the open source search engine library
+# Xapian (see: http://xapian.org/). See the section "External Indexing and
+# Searching" for details.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+SEARCHENGINE_URL       =
+
+# When SERVER_BASED_SEARCH and EXTERNAL_SEARCH are both enabled the unindexed
+# search data is written to a file for indexing by an external tool. With the
+# SEARCHDATA_FILE tag the name of this file can be specified.
+# The default file is: searchdata.xml.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+SEARCHDATA_FILE        = searchdata.xml
+
+# When SERVER_BASED_SEARCH and EXTERNAL_SEARCH are both enabled the
+# EXTERNAL_SEARCH_ID tag can be used as an identifier for the project. This is
+# useful in combination with EXTRA_SEARCH_MAPPINGS to search through multiple
+# projects and redirect the results back to the right project.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+EXTERNAL_SEARCH_ID     =
+
+# The EXTRA_SEARCH_MAPPINGS tag can be used to enable searching through doxygen
+# projects other than the one defined by this configuration file, but that are
+# all added to the same external search index. Each project needs to have a
+# unique id set via EXTERNAL_SEARCH_ID. The search mapping then maps the id of
+# to a relative location where the documentation can be found. The format is:
+# EXTRA_SEARCH_MAPPINGS = tagname1=loc1 tagname2=loc2 ...
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+EXTRA_SEARCH_MAPPINGS  =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the LaTeX output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_LATEX tag is set to YES doxygen will generate LaTeX output.
+# The default value is: YES.
+
+GENERATE_LATEX         = YES
+
+# The LATEX_OUTPUT tag is used to specify where the LaTeX docs will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it.
+# The default directory is: latex.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_OUTPUT           = latex
+
+# The LATEX_CMD_NAME tag can be used to specify the LaTeX command name to be
+# invoked.
+#
+# Note that when enabling USE_PDFLATEX this option is only used for generating
+# bitmaps for formulas in the HTML output, but not in the Makefile that is
+# written to the output directory.
+# The default file is: latex.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_CMD_NAME         = latex
+
+# The MAKEINDEX_CMD_NAME tag can be used to specify the command name to generate
+# index for LaTeX.
+# The default file is: makeindex.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+MAKEINDEX_CMD_NAME     = makeindex
+
+# If the COMPACT_LATEX tag is set to YES doxygen generates more compact LaTeX
+# documents. This may be useful for small projects and may help to save some
+# trees in general.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+COMPACT_LATEX          = NO
+
+# The PAPER_TYPE tag can be used to set the paper type that is used by the
+# printer.
+# Possible values are: a4 (210 x 297 mm), letter (8.5 x 11 inches), legal (8.5 x
+# 14 inches) and executive (7.25 x 10.5 inches).
+# The default value is: a4.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+PAPER_TYPE             = a4
+
+# The EXTRA_PACKAGES tag can be used to specify one or more LaTeX package names
+# that should be included in the LaTeX output. To get the times font for
+# instance you can specify
+# EXTRA_PACKAGES=times
+# If left blank no extra packages will be included.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+EXTRA_PACKAGES         =
+
+# The LATEX_HEADER tag can be used to specify a personal LaTeX header for the
+# generated LaTeX document. The header should contain everything until the first
+# chapter. If it is left blank doxygen will generate a standard header. See
+# section "Doxygen usage" for information on how to let doxygen write the
+# default header to a separate file.
+#
+# Note: Only use a user-defined header if you know what you are doing! The
+# following commands have a special meaning inside the header: $title,
+# $datetime, $date, $doxygenversion, $projectname, $projectnumber. Doxygen will
+# replace them by respectively the title of the page, the current date and time,
+# only the current date, the version number of doxygen, the project name (see
+# PROJECT_NAME), or the project number (see PROJECT_NUMBER).
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_HEADER           =
+
+# The LATEX_FOOTER tag can be used to specify a personal LaTeX footer for the
+# generated LaTeX document. The footer should contain everything after the last
+# chapter. If it is left blank doxygen will generate a standard footer.
+#
+# Note: Only use a user-defined footer if you know what you are doing!
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_FOOTER           =
+
+# The LATEX_EXTRA_FILES tag can be used to specify one or more extra images or
+# other source files which should be copied to the LATEX_OUTPUT output
+# directory. Note that the files will be copied as-is; there are no commands or
+# markers available.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_EXTRA_FILES      =
+
+# If the PDF_HYPERLINKS tag is set to YES, the LaTeX that is generated is
+# prepared for conversion to PDF (using ps2pdf or pdflatex). The PDF file will
+# contain links (just like the HTML output) instead of page references. This
+# makes the output suitable for online browsing using a PDF viewer.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+PDF_HYPERLINKS         = YES
+
+# If the LATEX_PDFLATEX tag is set to YES, doxygen will use pdflatex to generate
+# the PDF file directly from the LaTeX files. Set this option to YES to get a
+# higher quality PDF documentation.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+USE_PDFLATEX           = YES
+
+# If the LATEX_BATCHMODE tag is set to YES, doxygen will add the \batchmode
+# command to the generated LaTeX files. This will instruct LaTeX to keep running
+# if errors occur, instead of asking the user for help. This option is also used
+# when generating formulas in HTML.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_BATCHMODE        = NO
+
+# If the LATEX_HIDE_INDICES tag is set to YES then doxygen will not include the
+# index chapters (such as File Index, Compound Index, etc.) in the output.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_HIDE_INDICES     = NO
+
+# If the LATEX_SOURCE_CODE tag is set to YES then doxygen will include source
+# code with syntax highlighting in the LaTeX output.
+#
+# Note that which sources are shown also depends on other settings such as
+# SOURCE_BROWSER.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_SOURCE_CODE      = NO
+
+# The LATEX_BIB_STYLE tag can be used to specify the style to use for the
+# bibliography, e.g. plainnat, or ieeetr. See
+# http://en.wikipedia.org/wiki/BibTeX and \cite for more info.
+# The default value is: plain.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_BIB_STYLE        = plain
+
+#---------------------------------------------------------------------------
+# Configuration options related to the RTF output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_RTF tag is set to YES doxygen will generate RTF output. The
+# RTF output is optimized for Word 97 and may not look too pretty with other RTF
+# readers/editors.
+# The default value is: NO.
+
+GENERATE_RTF           = NO
+
+# The RTF_OUTPUT tag is used to specify where the RTF docs will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it.
+# The default directory is: rtf.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+RTF_OUTPUT             = rtf
+
+# If the COMPACT_RTF tag is set to YES doxygen generates more compact RTF
+# documents. This may be useful for small projects and may help to save some
+# trees in general.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+COMPACT_RTF            = NO
+
+# If the RTF_HYPERLINKS tag is set to YES, the RTF that is generated will
+# contain hyperlink fields. The RTF file will contain links (just like the HTML
+# output) instead of page references. This makes the output suitable for online
+# browsing using Word or some other Word compatible readers that support those
+# fields.
+#
+# Note: WordPad (write) and others do not support links.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+RTF_HYPERLINKS         = NO
+
+# Load stylesheet definitions from file. Syntax is similar to doxygen's config
+# file, i.e. a series of assignments. You only have to provide replacements,
+# missing definitions are set to their default value.
+#
+# See also section "Doxygen usage" for information on how to generate the
+# default style sheet that doxygen normally uses.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+RTF_STYLESHEET_FILE    =
+
+# Set optional variables used in the generation of an RTF document. Syntax is
+# similar to doxygen's config file. A template extensions file can be generated
+# using doxygen -e rtf extensionFile.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+RTF_EXTENSIONS_FILE    =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the man page output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_MAN tag is set to YES doxygen will generate man pages for
+# classes and files.
+# The default value is: NO.
+
+GENERATE_MAN           = NO
+
+# The MAN_OUTPUT tag is used to specify where the man pages will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it. A directory man3 will be created inside the directory specified by
+# MAN_OUTPUT.
+# The default directory is: man.
+# This tag requires that the tag GENERATE_MAN is set to YES.
+
+MAN_OUTPUT             = man
+
+# The MAN_EXTENSION tag determines the extension that is added to the generated
+# man pages. In case the manual section does not start with a number, the number
+# 3 is prepended. The dot (.) at the beginning of the MAN_EXTENSION tag is
+# optional.
+# The default value is: .3.
+# This tag requires that the tag GENERATE_MAN is set to YES.
+
+MAN_EXTENSION          = .3
+
+# If the MAN_LINKS tag is set to YES and doxygen generates man output, then it
+# will generate one additional man file for each entity documented in the real
+# man page(s). These additional files only source the real man page, but without
+# them the man command would be unable to find the correct page.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_MAN is set to YES.
+
+MAN_LINKS              = NO
+
+#---------------------------------------------------------------------------
+# Configuration options related to the XML output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_XML tag is set to YES doxygen will generate an XML file that
+# captures the structure of the code including all documentation.
+# The default value is: NO.
+
+GENERATE_XML           = NO
+
+# The XML_OUTPUT tag is used to specify where the XML pages will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it.
+# The default directory is: xml.
+# This tag requires that the tag GENERATE_XML is set to YES.
+
+XML_OUTPUT             = xml
+
+# The XML_SCHEMA tag can be used to specify a XML schema, which can be used by a
+# validating XML parser to check the syntax of the XML files.
+# This tag requires that the tag GENERATE_XML is set to YES.
+
+XML_SCHEMA             =
+
+# The XML_DTD tag can be used to specify a XML DTD, which can be used by a
+# validating XML parser to check the syntax of the XML files.
+# This tag requires that the tag GENERATE_XML is set to YES.
+
+XML_DTD                =
+
+# If the XML_PROGRAMLISTING tag is set to YES doxygen will dump the program
+# listings (including syntax highlighting and cross-referencing information) to
+# the XML output. Note that enabling this will significantly increase the size
+# of the XML output.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_XML is set to YES.
+
+XML_PROGRAMLISTING     = YES
+
+#---------------------------------------------------------------------------
+# Configuration options related to the DOCBOOK output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_DOCBOOK tag is set to YES doxygen will generate Docbook files
+# that can be used to generate PDF.
+# The default value is: NO.
+
+GENERATE_DOCBOOK       = NO
+
+# The DOCBOOK_OUTPUT tag is used to specify where the Docbook pages will be put.
+# If a relative path is entered the value of OUTPUT_DIRECTORY will be put in
+# front of it.
+# The default directory is: docbook.
+# This tag requires that the tag GENERATE_DOCBOOK is set to YES.
+
+DOCBOOK_OUTPUT         = docbook
+
+#---------------------------------------------------------------------------
+# Configuration options for the AutoGen Definitions output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_AUTOGEN_DEF tag is set to YES doxygen will generate an AutoGen
+# Definitions (see http://autogen.sf.net) file that captures the structure of
+# the code including all documentation. Note that this feature is still
+# experimental and incomplete at the moment.
+# The default value is: NO.
+
+GENERATE_AUTOGEN_DEF   = NO
+
+#---------------------------------------------------------------------------
+# Configuration options related to the Perl module output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_PERLMOD tag is set to YES doxygen will generate a Perl module
+# file that captures the structure of the code including all documentation.
+#
+# Note that this feature is still experimental and incomplete at the moment.
+# The default value is: NO.
+
+GENERATE_PERLMOD       = NO
+
+# If the PERLMOD_LATEX tag is set to YES doxygen will generate the necessary
+# Makefile rules, Perl scripts and LaTeX code to be able to generate PDF and DVI
+# output from the Perl module output.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_PERLMOD is set to YES.
+
+PERLMOD_LATEX          = NO
+
+# If the PERLMOD_PRETTY tag is set to YES the Perl module output will be nicely
+# formatted so it can be parsed by a human reader. This is useful if you want to
+# understand what is going on. On the other hand, if this tag is set to NO the
+# size of the Perl module output will be much smaller and Perl will parse it
+# just the same.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_PERLMOD is set to YES.
+
+PERLMOD_PRETTY         = YES
+
+# The names of the make variables in the generated doxyrules.make file are
+# prefixed with the string contained in PERLMOD_MAKEVAR_PREFIX. This is useful
+# so different doxyrules.make files included by the same Makefile don't
+# overwrite each other's variables.
+# This tag requires that the tag GENERATE_PERLMOD is set to YES.
+
+PERLMOD_MAKEVAR_PREFIX =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the preprocessor
+#---------------------------------------------------------------------------
+
+# If the ENABLE_PREPROCESSING tag is set to YES doxygen will evaluate all
+# C-preprocessor directives found in the sources and include files.
+# The default value is: YES.
+
+ENABLE_PREPROCESSING   = YES
+
+# If the MACRO_EXPANSION tag is set to YES doxygen will expand all macro names
+# in the source code. If set to NO only conditional compilation will be
+# performed. Macro expansion can be done in a controlled way by setting
+# EXPAND_ONLY_PREDEF to YES.
+# The default value is: NO.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+MACRO_EXPANSION        = NO
+
+# If the EXPAND_ONLY_PREDEF and MACRO_EXPANSION tags are both set to YES then
+# the macro expansion is limited to the macros specified with the PREDEFINED and
+# EXPAND_AS_DEFINED tags.
+# The default value is: NO.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+EXPAND_ONLY_PREDEF     = NO
+
+# If the SEARCH_INCLUDES tag is set to YES the includes files in the
+# INCLUDE_PATH will be searched if a #include is found.
+# The default value is: YES.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+SEARCH_INCLUDES        = YES
+
+# The INCLUDE_PATH tag can be used to specify one or more directories that
+# contain include files that are not input files but should be processed by the
+# preprocessor.
+# This tag requires that the tag SEARCH_INCLUDES is set to YES.
+
+INCLUDE_PATH           =
+
+# You can use the INCLUDE_FILE_PATTERNS tag to specify one or more wildcard
+# patterns (like *.h and *.hpp) to filter out the header-files in the
+# directories. If left blank, the patterns specified with FILE_PATTERNS will be
+# used.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+INCLUDE_FILE_PATTERNS  =
+
+# The PREDEFINED tag can be used to specify one or more macro names that are
+# defined before the preprocessor is started (similar to the -D option of e.g.
+# gcc). The argument of the tag is a list of macros of the form: name or
+# name=definition (no spaces). If the definition and the "=" are omitted, "=1"
+# is assumed. To prevent a macro definition from being undefined via #undef or
+# recursively expanded use the := operator instead of the = operator.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+PREDEFINED             =
+
+# If the MACRO_EXPANSION and EXPAND_ONLY_PREDEF tags are set to YES then this
+# tag can be used to specify a list of macro names that should be expanded. The
+# macro definition that is found in the sources will be used. Use the PREDEFINED
+# tag if you want to use a different macro definition that overrules the
+# definition found in the source code.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+EXPAND_AS_DEFINED      =
+
+# If the SKIP_FUNCTION_MACROS tag is set to YES then doxygen's preprocessor will
+# remove all refrences to function-like macros that are alone on a line, have an
+# all uppercase name, and do not end with a semicolon. Such function macros are
+# typically used for boiler-plate code, and will confuse the parser if not
+# removed.
+# The default value is: YES.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+SKIP_FUNCTION_MACROS   = YES
+
+#---------------------------------------------------------------------------
+# Configuration options related to external references
+#---------------------------------------------------------------------------
+
+# The TAGFILES tag can be used to specify one or more tag files. For each tag
+# file the location of the external documentation should be added. The format of
+# a tag file without this location is as follows:
+# TAGFILES = file1 file2 ...
+# Adding location for the tag files is done as follows:
+# TAGFILES = file1=loc1 "file2 = loc2" ...
+# where loc1 and loc2 can be relative or absolute paths or URLs. See the
+# section "Linking to external documentation" for more information about the use
+# of tag files.
+# Note: Each tag file must have an unique name (where the name does NOT include
+# the path). If a tag file is not located in the directory in which doxygen is
+# run, you must also specify the path to the tagfile here.
+
+TAGFILES               =
+
+# When a file name is specified after GENERATE_TAGFILE, doxygen will create a
+# tag file that is based on the input files it reads. See section "Linking to
+# external documentation" for more information about the usage of tag files.
+
+GENERATE_TAGFILE       =
+
+# If the ALLEXTERNALS tag is set to YES all external class will be listed in the
+# class index. If set to NO only the inherited external classes will be listed.
+# The default value is: NO.
+
+ALLEXTERNALS           = NO
+
+# If the EXTERNAL_GROUPS tag is set to YES all external groups will be listed in
+# the modules index. If set to NO, only the current project's groups will be
+# listed.
+# The default value is: YES.
+
+EXTERNAL_GROUPS        = YES
+
+# If the EXTERNAL_PAGES tag is set to YES all external pages will be listed in
+# the related pages index. If set to NO, only the current project's pages will
+# be listed.
+# The default value is: YES.
+
+EXTERNAL_PAGES         = YES
+
+# The PERL_PATH should be the absolute path and name of the perl script
+# interpreter (i.e. the result of 'which perl').
+# The default file (with absolute path) is: /usr/bin/perl.
+
+PERL_PATH              = /usr/bin/perl
+
+#---------------------------------------------------------------------------
+# Configuration options related to the dot tool
+#---------------------------------------------------------------------------
+
+# If the CLASS_DIAGRAMS tag is set to YES doxygen will generate a class diagram
+# (in HTML and LaTeX) for classes with base or super classes. Setting the tag to
+# NO turns the diagrams off. Note that this option also works with HAVE_DOT
+# disabled, but it is recommended to install and use dot, since it yields more
+# powerful graphs.
+# The default value is: YES.
+
+CLASS_DIAGRAMS         = YES
+
+# You can define message sequence charts within doxygen comments using the \msc
+# command. Doxygen will then run the mscgen tool (see:
+# http://www.mcternan.me.uk/mscgen/)) to produce the chart and insert it in the
+# documentation. The MSCGEN_PATH tag allows you to specify the directory where
+# the mscgen tool resides. If left empty the tool is assumed to be found in the
+# default search path.
+
+MSCGEN_PATH            =
+
+# You can include diagrams made with dia in doxygen documentation. Doxygen will
+# then run dia to produce the diagram and insert it in the documentation. The
+# DIA_PATH tag allows you to specify the directory where the dia binary resides.
+# If left empty dia is assumed to be found in the default search path.
+
+DIA_PATH               =
+
+# If set to YES, the inheritance and collaboration graphs will hide inheritance
+# and usage relations if the target is undocumented or is not a class.
+# The default value is: YES.
+
+HIDE_UNDOC_RELATIONS   = YES
+
+# If you set the HAVE_DOT tag to YES then doxygen will assume the dot tool is
+# available from the path. This tool is part of Graphviz (see:
+# http://www.graphviz.org/), a graph visualization toolkit from AT&T and Lucent
+# Bell Labs. The other options in this section have no effect if this option is
+# set to NO
+# The default value is: NO.
+
+HAVE_DOT               = YES
+
+# The DOT_NUM_THREADS specifies the number of dot invocations doxygen is allowed
+# to run in parallel. When set to 0 doxygen will base this on the number of
+# processors available in the system. You can set it explicitly to a value
+# larger than 0 to get control over the balance between CPU load and processing
+# speed.
+# Minimum value: 0, maximum value: 32, default value: 0.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_NUM_THREADS        = 2
+
+# When you want a differently looking font n the dot files that doxygen
+# generates you can specify the font name using DOT_FONTNAME. You need to make
+# sure dot is able to find the font, which can be done by putting it in a
+# standard location or by setting the DOTFONTPATH environment variable or by
+# setting DOT_FONTPATH to the directory containing the font.
+# The default value is: Helvetica.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_FONTNAME           = Helvetica
+
+# The DOT_FONTSIZE tag can be used to set the size (in points) of the font of
+# dot graphs.
+# Minimum value: 4, maximum value: 24, default value: 10.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_FONTSIZE           = 10
+
+# By default doxygen will tell dot to use the default font as specified with
+# DOT_FONTNAME. If you specify a different font using DOT_FONTNAME you can set
+# the path where dot can find it using this tag.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_FONTPATH           =
+
+# If the CLASS_GRAPH tag is set to YES then doxygen will generate a graph for
+# each documented class showing the direct and indirect inheritance relations.
+# Setting this tag to YES will force the CLASS_DIAGRAMS tag to NO.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+CLASS_GRAPH            = YES
+
+# If the COLLABORATION_GRAPH tag is set to YES then doxygen will generate a
+# graph for each documented class showing the direct and indirect implementation
+# dependencies (inheritance, containment, and class references variables) of the
+# class with other documented classes.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+COLLABORATION_GRAPH    = YES
+
+# If the GROUP_GRAPHS tag is set to YES then doxygen will generate a graph for
+# groups, showing the direct groups dependencies.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+GROUP_GRAPHS           = YES
+
+# If the UML_LOOK tag is set to YES doxygen will generate inheritance and
+# collaboration diagrams in a style similar to the OMG's Unified Modeling
+# Language.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+UML_LOOK               = NO
+
+# If the UML_LOOK tag is enabled, the fields and methods are shown inside the
+# class node. If there are many fields or methods and many nodes the graph may
+# become too big to be useful. The UML_LIMIT_NUM_FIELDS threshold limits the
+# number of items for each type to make the size more manageable. Set this to 0
+# for no limit. Note that the threshold may be exceeded by 50% before the limit
+# is enforced. So when you set the threshold to 10, up to 15 fields may appear,
+# but if the number exceeds 15, the total amount of fields shown is limited to
+# 10.
+# Minimum value: 0, maximum value: 100, default value: 10.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+UML_LIMIT_NUM_FIELDS   = 10
+
+# If the TEMPLATE_RELATIONS tag is set to YES then the inheritance and
+# collaboration graphs will show the relations between templates and their
+# instances.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+TEMPLATE_RELATIONS     = NO
+
+# If the INCLUDE_GRAPH, ENABLE_PREPROCESSING and SEARCH_INCLUDES tags are set to
+# YES then doxygen will generate a graph for each documented file showing the
+# direct and indirect include dependencies of the file with other documented
+# files.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+INCLUDE_GRAPH          = YES
+
+# If the INCLUDED_BY_GRAPH, ENABLE_PREPROCESSING and SEARCH_INCLUDES tags are
+# set to YES then doxygen will generate a graph for each documented file showing
+# the direct and indirect include dependencies of the file with other documented
+# files.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+INCLUDED_BY_GRAPH      = YES
+
+# If the CALL_GRAPH tag is set to YES then doxygen will generate a call
+# dependency graph for every global function or class method.
+#
+# Note that enabling this option will significantly increase the time of a run.
+# So in most cases it will be better to enable call graphs for selected
+# functions only using the \callgraph command.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+CALL_GRAPH             = NO
+
+# If the CALLER_GRAPH tag is set to YES then doxygen will generate a caller
+# dependency graph for every global function or class method.
+#
+# Note that enabling this option will significantly increase the time of a run.
+# So in most cases it will be better to enable caller graphs for selected
+# functions only using the \callergraph command.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+CALLER_GRAPH           = NO
+
+# If the GRAPHICAL_HIERARCHY tag is set to YES then doxygen will graphical
+# hierarchy of all classes instead of a textual one.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+GRAPHICAL_HIERARCHY    = YES
+
+# If the DIRECTORY_GRAPH tag is set to YES then doxygen will show the
+# dependencies a directory has on other directories in a graphical way. The
+# dependency relations are determined by the #include relations between the
+# files in the directories.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DIRECTORY_GRAPH        = YES
+
+# The DOT_IMAGE_FORMAT tag can be used to set the image format of the images
+# generated by dot.
+# Note: If you choose svg you need to set HTML_FILE_EXTENSION to xhtml in order
+# to make the SVG files visible in IE 9+ (other browsers do not have this
+# requirement).
+# Possible values are: png, jpg, gif and svg.
+# The default value is: png.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_IMAGE_FORMAT       = png
+
+# If DOT_IMAGE_FORMAT is set to svg, then this option can be set to YES to
+# enable generation of interactive SVG images that allow zooming and panning.
+#
+# Note that this requires a modern browser other than Internet Explorer. Tested
+# and working are Firefox, Chrome, Safari, and Opera.
+# Note: For IE 9+ you need to set HTML_FILE_EXTENSION to xhtml in order to make
+# the SVG files visible. Older versions of IE do not have SVG support.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+INTERACTIVE_SVG        = NO
+
+# The DOT_PATH tag can be used to specify the path where the dot tool can be
+# found. If left blank, it is assumed the dot tool can be found in the path.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_PATH               =
+
+# The DOTFILE_DIRS tag can be used to specify one or more directories that
+# contain dot files that are included in the documentation (see the \dotfile
+# command).
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOTFILE_DIRS           =
+
+# The MSCFILE_DIRS tag can be used to specify one or more directories that
+# contain msc files that are included in the documentation (see the \mscfile
+# command).
+
+MSCFILE_DIRS           =
+
+# The DIAFILE_DIRS tag can be used to specify one or more directories that
+# contain dia files that are included in the documentation (see the \diafile
+# command).
+
+DIAFILE_DIRS           =
+
+# The DOT_GRAPH_MAX_NODES tag can be used to set the maximum number of nodes
+# that will be shown in the graph. If the number of nodes in a graph becomes
+# larger than this value, doxygen will truncate the graph, which is visualized
+# by representing a node as a red box. Note that doxygen if the number of direct
+# children of the root node in a graph is already larger than
+# DOT_GRAPH_MAX_NODES then the graph will not be shown at all. Also note that
+# the size of a graph can be further restricted by MAX_DOT_GRAPH_DEPTH.
+# Minimum value: 0, maximum value: 10000, default value: 50.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_GRAPH_MAX_NODES    = 50
+
+# The MAX_DOT_GRAPH_DEPTH tag can be used to set the maximum depth of the graphs
+# generated by dot. A depth value of 3 means that only nodes reachable from the
+# root by following a path via at most 3 edges will be shown. Nodes that lay
+# further from the root node will be omitted. Note that setting this option to 1
+# or 2 may greatly reduce the computation time needed for large code bases. Also
+# note that the size of a graph can be further restricted by
+# DOT_GRAPH_MAX_NODES. Using a depth of 0 means no depth restriction.
+# Minimum value: 0, maximum value: 1000, default value: 0.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+MAX_DOT_GRAPH_DEPTH    = 0
+
+# Set the DOT_TRANSPARENT tag to YES to generate images with a transparent
+# background. This is disabled by default, because dot on Windows does not seem
+# to support this out of the box.
+#
+# Warning: Depending on the platform used, enabling this option may lead to
+# badly anti-aliased labels on the edges of a graph (i.e. they become hard to
+# read).
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_TRANSPARENT        = YES
+
+# Set the DOT_MULTI_TARGETS tag to YES allow dot to generate multiple output
+# files in one run (i.e. multiple -o and -T options on the command line). This
+# makes dot run faster, but since only newer versions of dot (>1.8.10) support
+# this, this feature is disabled by default.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_MULTI_TARGETS      = YES
+
+# If the GENERATE_LEGEND tag is set to YES doxygen will generate a legend page
+# explaining the meaning of the various boxes and arrows in the dot generated
+# graphs.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+GENERATE_LEGEND        = YES
+
+# If the DOT_CLEANUP tag is set to YES doxygen will remove the intermediate dot
+# files that are used to generate the various graphs.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_CLEANUP            = YES
diff --git a/docs/doxy.cfg.test b/docs/doxy.cfg.test
new file mode 100644
index 00000000..c4b82094
--- /dev/null
+++ b/docs/doxy.cfg.test
@@ -0,0 +1,2305 @@
+# Doxyfile 1.8.6
+
+# This file describes the settings to be used by the documentation system
+# doxygen (www.doxygen.org) for a project.
+#
+# All text after a double hash (##) is considered a comment and is placed in
+# front of the TAG it is preceding.
+#
+# All text after a single hash (#) is considered a comment and will be ignored.
+# The format is:
+# TAG = value [value, ...]
+# For lists, items can also be appended using:
+# TAG += value [value, ...]
+# Values that contain spaces should be placed between quotes (\" \").
+
+#---------------------------------------------------------------------------
+# Project related configuration options
+#---------------------------------------------------------------------------
+
+# This tag specifies the encoding used for all characters in the config file
+# that follow. The default is UTF-8 which is also the encoding used for all text
+# before the first occurrence of this tag. Doxygen uses libiconv (or the iconv
+# built into libc) for the transcoding. See http://www.gnu.org/software/libiconv
+# for the list of possible encodings.
+# The default value is: UTF-8.
+
+DOXYFILE_ENCODING      = UTF-8
+
+# The PROJECT_NAME tag is a single word (or a sequence of words surrounded by
+# double-quotes, unless you are using Doxywizard) that should identify the
+# project for which the documentation is generated. This name is used in the
+# title of most generated pages and in a few other places.
+# The default value is: My Project.
+
+PROJECT_NAME           = "Grid"
+
+# The PROJECT_NUMBER tag can be used to enter a project or revision number. This
+# could be handy for archiving the generated documentation or if some version
+# control system is used.
+
+PROJECT_NUMBER         = 1.0
+
+# Using the PROJECT_BRIEF tag one can provide an optional one line description
+# for a project that appears at the top of each page and should give viewer a
+# quick idea about the purpose of the project. Keep the description short.
+
+PROJECT_BRIEF          =
+
+# With the PROJECT_LOGO tag one can specify an logo or icon that is included in
+# the documentation. The maximum height of the logo should not exceed 55 pixels
+# and the maximum width should not exceed 200 pixels. Doxygen will copy the logo
+# to the output directory.
+
+PROJECT_LOGO           =
+
+# The OUTPUT_DIRECTORY tag is used to specify the (relative or absolute) path
+# into which the generated documentation will be written. If a relative path is
+# entered, it will be relative to the location where doxygen was started. If
+# left blank the current directory will be used.
+
+OUTPUT_DIRECTORY       = ./doxy-en/
+
+# If the CREATE_SUBDIRS tag is set to YES, then doxygen will create 4096 sub-
+# directories (in 2 levels) under the output directory of each output format and
+# will distribute the generated files over these directories. Enabling this
+# option can be useful when feeding doxygen a huge amount of source files, where
+# putting all generated files in the same directory would otherwise causes
+# performance problems for the file system.
+# The default value is: NO.
+
+CREATE_SUBDIRS         = NO
+
+# The OUTPUT_LANGUAGE tag is used to specify the language in which all
+# documentation generated by doxygen is written. Doxygen will use this
+# information to generate all constant output in the proper language.
+# Possible values are: Afrikaans, Arabic, Armenian, Brazilian, Catalan, Chinese,
+# Chinese-Traditional, Croatian, Czech, Danish, Dutch, English (United States),
+# Esperanto, Farsi (Persian), Finnish, French, German, Greek, Hungarian,
+# Indonesian, Italian, Japanese, Japanese-en (Japanese with English messages),
+# Korean, Korean-en (Korean with English messages), Latvian, Lithuanian,
+# Macedonian, Norwegian, Persian (Farsi), Polish, Portuguese, Romanian, Russian,
+# Serbian, Serbian-Cyrillic, Slovak, Slovene, Spanish, Swedish, Turkish,
+# Ukrainian and Vietnamese.
+# The default value is: English.
+
+OUTPUT_LANGUAGE        = English
+
+# If the BRIEF_MEMBER_DESC tag is set to YES doxygen will include brief member
+# descriptions after the members that are listed in the file and class
+# documentation (similar to Javadoc). Set to NO to disable this.
+# The default value is: YES.
+
+BRIEF_MEMBER_DESC      = YES
+
+# If the REPEAT_BRIEF tag is set to YES doxygen will prepend the brief
+# description of a member or function before the detailed description
+#
+# Note: If both HIDE_UNDOC_MEMBERS and BRIEF_MEMBER_DESC are set to NO, the
+# brief descriptions will be completely suppressed.
+# The default value is: YES.
+
+REPEAT_BRIEF           = YES
+
+# This tag implements a quasi-intelligent brief description abbreviator that is
+# used to form the text in various listings. Each string in this list, if found
+# as the leading text of the brief description, will be stripped from the text
+# and the result, after processing the whole list, is used as the annotated
+# text. Otherwise, the brief description is used as-is. If left blank, the
+# following values are used ($name is automatically replaced with the name of
+# the entity):The $name class, The $name widget, The $name file, is, provides,
+# specifies, contains, represents, a, an and the.
+
+ABBREVIATE_BRIEF       =
+
+# If the ALWAYS_DETAILED_SEC and REPEAT_BRIEF tags are both set to YES then
+# doxygen will generate a detailed section even if there is only a brief
+# description.
+# The default value is: NO.
+
+ALWAYS_DETAILED_SEC    = NO
+
+# If the INLINE_INHERITED_MEMB tag is set to YES, doxygen will show all
+# inherited members of a class in the documentation of that class as if those
+# members were ordinary class members. Constructors, destructors and assignment
+# operators of the base classes will not be shown.
+# The default value is: NO.
+
+INLINE_INHERITED_MEMB  = NO
+
+# If the FULL_PATH_NAMES tag is set to YES doxygen will prepend the full path
+# before files name in the file list and in the header files. If set to NO the
+# shortest path that makes the file name unique will be used
+# The default value is: YES.
+
+FULL_PATH_NAMES        = YES
+
+# The STRIP_FROM_PATH tag can be used to strip a user-defined part of the path.
+# Stripping is only done if one of the specified strings matches the left-hand
+# part of the path. The tag can be used to show relative paths in the file list.
+# If left blank the directory from which doxygen is run is used as the path to
+# strip.
+#
+# Note that you can specify absolute paths here, but also relative paths, which
+# will be relative from the directory where doxygen is started.
+# This tag requires that the tag FULL_PATH_NAMES is set to YES.
+
+STRIP_FROM_PATH        =
+
+# The STRIP_FROM_INC_PATH tag can be used to strip a user-defined part of the
+# path mentioned in the documentation of a class, which tells the reader which
+# header file to include in order to use a class. If left blank only the name of
+# the header file containing the class definition is used. Otherwise one should
+# specify the list of include paths that are normally passed to the compiler
+# using the -I flag.
+
+STRIP_FROM_INC_PATH    =
+
+# If the SHORT_NAMES tag is set to YES, doxygen will generate much shorter (but
+# less readable) file names. This can be useful is your file systems doesn't
+# support long names like on DOS, Mac, or CD-ROM.
+# The default value is: NO.
+
+SHORT_NAMES            = NO
+
+# If the JAVADOC_AUTOBRIEF tag is set to YES then doxygen will interpret the
+# first line (until the first dot) of a Javadoc-style comment as the brief
+# description. If set to NO, the Javadoc-style will behave just like regular Qt-
+# style comments (thus requiring an explicit @brief command for a brief
+# description.)
+# The default value is: NO.
+
+JAVADOC_AUTOBRIEF      = NO
+
+# If the QT_AUTOBRIEF tag is set to YES then doxygen will interpret the first
+# line (until the first dot) of a Qt-style comment as the brief description. If
+# set to NO, the Qt-style will behave just like regular Qt-style comments (thus
+# requiring an explicit \brief command for a brief description.)
+# The default value is: NO.
+
+QT_AUTOBRIEF           = NO
+
+# The MULTILINE_CPP_IS_BRIEF tag can be set to YES to make doxygen treat a
+# multi-line C++ special comment block (i.e. a block of //! or /// comments) as
+# a brief description. This used to be the default behavior. The new default is
+# to treat a multi-line C++ comment block as a detailed description. Set this
+# tag to YES if you prefer the old behavior instead.
+#
+# Note that setting this tag to YES also means that rational rose comments are
+# not recognized any more.
+# The default value is: NO.
+
+MULTILINE_CPP_IS_BRIEF = NO
+
+# If the INHERIT_DOCS tag is set to YES then an undocumented member inherits the
+# documentation from any documented member that it re-implements.
+# The default value is: YES.
+
+INHERIT_DOCS           = YES
+
+# If the SEPARATE_MEMBER_PAGES tag is set to YES, then doxygen will produce a
+# new page for each member. If set to NO, the documentation of a member will be
+# part of the file/class/namespace that contains it.
+# The default value is: NO.
+
+SEPARATE_MEMBER_PAGES  = NO
+
+# The TAB_SIZE tag can be used to set the number of spaces in a tab. Doxygen
+# uses this value to replace tabs by spaces in code fragments.
+# Minimum value: 1, maximum value: 16, default value: 4.
+
+TAB_SIZE               = 4
+
+# This tag can be used to specify a number of aliases that act as commands in
+# the documentation. An alias has the form:
+# name=value
+# For example adding
+# "sideeffect=@par Side Effects:\n"
+# will allow you to put the command \sideeffect (or @sideeffect) in the
+# documentation, which will result in a user-defined paragraph with heading
+# "Side Effects:". You can put \n's in the value part of an alias to insert
+# newlines.
+
+ALIASES                =
+
+# This tag can be used to specify a number of word-keyword mappings (TCL only).
+# A mapping has the form "name=value". For example adding "class=itcl::class"
+# will allow you to use the command class in the itcl::class meaning.
+
+TCL_SUBST              =
+
+# Set the OPTIMIZE_OUTPUT_FOR_C tag to YES if your project consists of C sources
+# only. Doxygen will then generate output that is more tailored for C. For
+# instance, some of the names that are used will be different. The list of all
+# members will be omitted, etc.
+# The default value is: NO.
+
+OPTIMIZE_OUTPUT_FOR_C  = NO
+
+# Set the OPTIMIZE_OUTPUT_JAVA tag to YES if your project consists of Java or
+# Python sources only. Doxygen will then generate output that is more tailored
+# for that language. For instance, namespaces will be presented as packages,
+# qualified scopes will look different, etc.
+# The default value is: NO.
+
+OPTIMIZE_OUTPUT_JAVA   = NO
+
+# Set the OPTIMIZE_FOR_FORTRAN tag to YES if your project consists of Fortran
+# sources. Doxygen will then generate output that is tailored for Fortran.
+# The default value is: NO.
+
+OPTIMIZE_FOR_FORTRAN   = NO
+
+# Set the OPTIMIZE_OUTPUT_VHDL tag to YES if your project consists of VHDL
+# sources. Doxygen will then generate output that is tailored for VHDL.
+# The default value is: NO.
+
+OPTIMIZE_OUTPUT_VHDL   = NO
+
+# Doxygen selects the parser to use depending on the extension of the files it
+# parses. With this tag you can assign which parser to use for a given
+# extension. Doxygen has a built-in mapping, but you can override or extend it
+# using this tag. The format is ext=language, where ext is a file extension, and
+# language is one of the parsers supported by doxygen: IDL, Java, Javascript,
+# C#, C, C++, D, PHP, Objective-C, Python, Fortran, VHDL. For instance to make
+# doxygen treat .inc files as Fortran files (default is PHP), and .f files as C
+# (default is Fortran), use: inc=Fortran f=C.
+#
+# Note For files without extension you can use no_extension as a placeholder.
+#
+# Note that for custom extensions you also need to set FILE_PATTERNS otherwise
+# the files are not read by doxygen.
+
+EXTENSION_MAPPING      =
+
+# If the MARKDOWN_SUPPORT tag is enabled then doxygen pre-processes all comments
+# according to the Markdown format, which allows for more readable
+# documentation. See http://daringfireball.net/projects/markdown/ for details.
+# The output of markdown processing is further processed by doxygen, so you can
+# mix doxygen, HTML, and XML commands with Markdown formatting. Disable only in
+# case of backward compatibilities issues.
+# The default value is: YES.
+
+MARKDOWN_SUPPORT       = YES
+
+# When enabled doxygen tries to link words that correspond to documented
+# classes, or namespaces to their corresponding documentation. Such a link can
+# be prevented in individual cases by by putting a % sign in front of the word
+# or globally by setting AUTOLINK_SUPPORT to NO.
+# The default value is: YES.
+
+AUTOLINK_SUPPORT       = YES
+
+# If you use STL classes (i.e. std::string, std::vector, etc.) but do not want
+# to include (a tag file for) the STL sources as input, then you should set this
+# tag to YES in order to let doxygen match functions declarations and
+# definitions whose arguments contain STL classes (e.g. func(std::string);
+# versus func(std::string) {}). This also make the inheritance and collaboration
+# diagrams that involve STL classes more complete and accurate.
+# The default value is: NO.
+
+BUILTIN_STL_SUPPORT    = YES
+
+# If you use Microsoft's C++/CLI language, you should set this option to YES to
+# enable parsing support.
+# The default value is: NO.
+
+CPP_CLI_SUPPORT        = NO
+
+# Set the SIP_SUPPORT tag to YES if your project consists of sip (see:
+# http://www.riverbankcomputing.co.uk/software/sip/intro) sources only. Doxygen
+# will parse them like normal C++ but will assume all classes use public instead
+# of private inheritance when no explicit protection keyword is present.
+# The default value is: NO.
+
+SIP_SUPPORT            = NO
+
+# For Microsoft's IDL there are propget and propput attributes to indicate
+# getter and setter methods for a property. Setting this option to YES will make
+# doxygen to replace the get and set methods by a property in the documentation.
+# This will only work if the methods are indeed getting or setting a simple
+# type. If this is not the case, or you want to show the methods anyway, you
+# should set this option to NO.
+# The default value is: YES.
+
+IDL_PROPERTY_SUPPORT   = YES
+
+# If member grouping is used in the documentation and the DISTRIBUTE_GROUP_DOC
+# tag is set to YES, then doxygen will reuse the documentation of the first
+# member in the group (if any) for the other members of the group. By default
+# all members of a group must be documented explicitly.
+# The default value is: NO.
+
+DISTRIBUTE_GROUP_DOC   = NO
+
+# Set the SUBGROUPING tag to YES to allow class member groups of the same type
+# (for instance a group of public functions) to be put as a subgroup of that
+# type (e.g. under the Public Functions section). Set it to NO to prevent
+# subgrouping. Alternatively, this can be done per class using the
+# \nosubgrouping command.
+# The default value is: YES.
+
+SUBGROUPING            = YES
+
+# When the INLINE_GROUPED_CLASSES tag is set to YES, classes, structs and unions
+# are shown inside the group in which they are included (e.g. using \ingroup)
+# instead of on a separate page (for HTML and Man pages) or section (for LaTeX
+# and RTF).
+#
+# Note that this feature does not work in combination with
+# SEPARATE_MEMBER_PAGES.
+# The default value is: NO.
+
+INLINE_GROUPED_CLASSES = NO
+
+# When the INLINE_SIMPLE_STRUCTS tag is set to YES, structs, classes, and unions
+# with only public data fields or simple typedef fields will be shown inline in
+# the documentation of the scope in which they are defined (i.e. file,
+# namespace, or group documentation), provided this scope is documented. If set
+# to NO, structs, classes, and unions are shown on a separate page (for HTML and
+# Man pages) or section (for LaTeX and RTF).
+# The default value is: NO.
+
+INLINE_SIMPLE_STRUCTS  = NO
+
+# When TYPEDEF_HIDES_STRUCT tag is enabled, a typedef of a struct, union, or
+# enum is documented as struct, union, or enum with the name of the typedef. So
+# typedef struct TypeS {} TypeT, will appear in the documentation as a struct
+# with name TypeT. When disabled the typedef will appear as a member of a file,
+# namespace, or class. And the struct will be named TypeS. This can typically be
+# useful for C code in case the coding convention dictates that all compound
+# types are typedef'ed and only the typedef is referenced, never the tag name.
+# The default value is: NO.
+
+TYPEDEF_HIDES_STRUCT   = NO
+
+# The size of the symbol lookup cache can be set using LOOKUP_CACHE_SIZE. This
+# cache is used to resolve symbols given their name and scope. Since this can be
+# an expensive process and often the same symbol appears multiple times in the
+# code, doxygen keeps a cache of pre-resolved symbols. If the cache is too small
+# doxygen will become slower. If the cache is too large, memory is wasted. The
+# cache size is given by this formula: 2^(16+LOOKUP_CACHE_SIZE). The valid range
+# is 0..9, the default is 0, corresponding to a cache size of 2^16=65536
+# symbols. At the end of a run doxygen will report the cache usage and suggest
+# the optimal cache size from a speed point of view.
+# Minimum value: 0, maximum value: 9, default value: 0.
+
+LOOKUP_CACHE_SIZE      = 0
+
+#---------------------------------------------------------------------------
+# Build related configuration options
+#---------------------------------------------------------------------------
+
+# If the EXTRACT_ALL tag is set to YES doxygen will assume all entities in
+# documentation are documented, even if no documentation was available. Private
+# class members and static file members will be hidden unless the
+# EXTRACT_PRIVATE respectively EXTRACT_STATIC tags are set to YES.
+# Note: This will also disable the warnings about undocumented members that are
+# normally produced when WARNINGS is set to YES.
+# The default value is: NO.
+
+EXTRACT_ALL            = YES
+
+# If the EXTRACT_PRIVATE tag is set to YES all private members of a class will
+# be included in the documentation.
+# The default value is: NO.
+
+EXTRACT_PRIVATE        = YES
+
+# If the EXTRACT_PACKAGE tag is set to YES all members with package or internal
+# scope will be included in the documentation.
+# The default value is: NO.
+
+EXTRACT_PACKAGE        = YES
+
+# If the EXTRACT_STATIC tag is set to YES all static members of a file will be
+# included in the documentation.
+# The default value is: NO.
+
+EXTRACT_STATIC         = YES
+
+# If the EXTRACT_LOCAL_CLASSES tag is set to YES classes (and structs) defined
+# locally in source files will be included in the documentation. If set to NO
+# only classes defined in header files are included. Does not have any effect
+# for Java sources.
+# The default value is: YES.
+
+EXTRACT_LOCAL_CLASSES  = YES
+
+# This flag is only useful for Objective-C code. When set to YES local methods,
+# which are defined in the implementation section but not in the interface are
+# included in the documentation. If set to NO only methods in the interface are
+# included.
+# The default value is: NO.
+
+EXTRACT_LOCAL_METHODS  = YES
+
+# If this flag is set to YES, the members of anonymous namespaces will be
+# extracted and appear in the documentation as a namespace called
+# 'anonymous_namespace{file}', where file will be replaced with the base name of
+# the file that contains the anonymous namespace. By default anonymous namespace
+# are hidden.
+# The default value is: NO.
+
+EXTRACT_ANON_NSPACES   = YES
+
+# If the HIDE_UNDOC_MEMBERS tag is set to YES, doxygen will hide all
+# undocumented members inside documented classes or files. If set to NO these
+# members will be included in the various overviews, but no documentation
+# section is generated. This option has no effect if EXTRACT_ALL is enabled.
+# The default value is: NO.
+
+HIDE_UNDOC_MEMBERS     = NO
+
+# If the HIDE_UNDOC_CLASSES tag is set to YES, doxygen will hide all
+# undocumented classes that are normally visible in the class hierarchy. If set
+# to NO these classes will be included in the various overviews. This option has
+# no effect if EXTRACT_ALL is enabled.
+# The default value is: NO.
+
+HIDE_UNDOC_CLASSES     = NO
+
+# If the HIDE_FRIEND_COMPOUNDS tag is set to YES, doxygen will hide all friend
+# (class|struct|union) declarations. If set to NO these declarations will be
+# included in the documentation.
+# The default value is: NO.
+
+HIDE_FRIEND_COMPOUNDS  = NO
+
+# If the HIDE_IN_BODY_DOCS tag is set to YES, doxygen will hide any
+# documentation blocks found inside the body of a function. If set to NO these
+# blocks will be appended to the function's detailed documentation block.
+# The default value is: NO.
+
+HIDE_IN_BODY_DOCS      = NO
+
+# The INTERNAL_DOCS tag determines if documentation that is typed after a
+# \internal command is included. If the tag is set to NO then the documentation
+# will be excluded. Set it to YES to include the internal documentation.
+# The default value is: NO.
+
+INTERNAL_DOCS          = NO
+
+# If the CASE_SENSE_NAMES tag is set to NO then doxygen will only generate file
+# names in lower-case letters. If set to YES upper-case letters are also
+# allowed. This is useful if you have classes or files whose names only differ
+# in case and if your file system supports case sensitive file names. Windows
+# and Mac users are advised to set this option to NO.
+# The default value is: system dependent.
+
+CASE_SENSE_NAMES       = YES
+
+# If the HIDE_SCOPE_NAMES tag is set to NO then doxygen will show members with
+# their full class and namespace scopes in the documentation. If set to YES the
+# scope will be hidden.
+# The default value is: NO.
+
+HIDE_SCOPE_NAMES       = NO
+
+# If the SHOW_INCLUDE_FILES tag is set to YES then doxygen will put a list of
+# the files that are included by a file in the documentation of that file.
+# The default value is: YES.
+
+SHOW_INCLUDE_FILES     = YES
+
+# If the SHOW_GROUPED_MEMB_INC tag is set to YES then Doxygen will add for each
+# grouped member an include statement to the documentation, telling the reader
+# which file to include in order to use the member.
+# The default value is: NO.
+
+SHOW_GROUPED_MEMB_INC  = NO
+
+# If the FORCE_LOCAL_INCLUDES tag is set to YES then doxygen will list include
+# files with double quotes in the documentation rather than with sharp brackets.
+# The default value is: NO.
+
+FORCE_LOCAL_INCLUDES   = NO
+
+# If the INLINE_INFO tag is set to YES then a tag [inline] is inserted in the
+# documentation for inline members.
+# The default value is: YES.
+
+INLINE_INFO            = YES
+
+# If the SORT_MEMBER_DOCS tag is set to YES then doxygen will sort the
+# (detailed) documentation of file and class members alphabetically by member
+# name. If set to NO the members will appear in declaration order.
+# The default value is: YES.
+
+SORT_MEMBER_DOCS       = YES
+
+# If the SORT_BRIEF_DOCS tag is set to YES then doxygen will sort the brief
+# descriptions of file, namespace and class members alphabetically by member
+# name. If set to NO the members will appear in declaration order. Note that
+# this will also influence the order of the classes in the class list.
+# The default value is: NO.
+
+SORT_BRIEF_DOCS        = NO
+
+# If the SORT_MEMBERS_CTORS_1ST tag is set to YES then doxygen will sort the
+# (brief and detailed) documentation of class members so that constructors and
+# destructors are listed first. If set to NO the constructors will appear in the
+# respective orders defined by SORT_BRIEF_DOCS and SORT_MEMBER_DOCS.
+# Note: If SORT_BRIEF_DOCS is set to NO this option is ignored for sorting brief
+# member documentation.
+# Note: If SORT_MEMBER_DOCS is set to NO this option is ignored for sorting
+# detailed member documentation.
+# The default value is: NO.
+
+SORT_MEMBERS_CTORS_1ST = YES
+
+# If the SORT_GROUP_NAMES tag is set to YES then doxygen will sort the hierarchy
+# of group names into alphabetical order. If set to NO the group names will
+# appear in their defined order.
+# The default value is: NO.
+
+SORT_GROUP_NAMES       = NO
+
+# If the SORT_BY_SCOPE_NAME tag is set to YES, the class list will be sorted by
+# fully-qualified names, including namespaces. If set to NO, the class list will
+# be sorted only by class name, not including the namespace part.
+# Note: This option is not very useful if HIDE_SCOPE_NAMES is set to YES.
+# Note: This option applies only to the class list, not to the alphabetical
+# list.
+# The default value is: NO.
+
+SORT_BY_SCOPE_NAME     = NO
+
+# If the STRICT_PROTO_MATCHING option is enabled and doxygen fails to do proper
+# type resolution of all parameters of a function it will reject a match between
+# the prototype and the implementation of a member function even if there is
+# only one candidate or it is obvious which candidate to choose by doing a
+# simple string match. By disabling STRICT_PROTO_MATCHING doxygen will still
+# accept a match between prototype and implementation in such cases.
+# The default value is: NO.
+
+STRICT_PROTO_MATCHING  = NO
+
+# The GENERATE_TODOLIST tag can be used to enable ( YES) or disable ( NO) the
+# todo list. This list is created by putting \todo commands in the
+# documentation.
+# The default value is: YES.
+
+GENERATE_TODOLIST      = YES
+
+# The GENERATE_TESTLIST tag can be used to enable ( YES) or disable ( NO) the
+# test list. This list is created by putting \test commands in the
+# documentation.
+# The default value is: YES.
+
+GENERATE_TESTLIST      = YES
+
+# The GENERATE_BUGLIST tag can be used to enable ( YES) or disable ( NO) the bug
+# list. This list is created by putting \bug commands in the documentation.
+# The default value is: YES.
+
+GENERATE_BUGLIST       = YES
+
+# The GENERATE_DEPRECATEDLIST tag can be used to enable ( YES) or disable ( NO)
+# the deprecated list. This list is created by putting \deprecated commands in
+# the documentation.
+# The default value is: YES.
+
+GENERATE_DEPRECATEDLIST= YES
+
+# The ENABLED_SECTIONS tag can be used to enable conditional documentation
+# sections, marked by \if <section_label> ... \endif and \cond <section_label>
+# ... \endcond blocks.
+
+ENABLED_SECTIONS       =
+
+# The MAX_INITIALIZER_LINES tag determines the maximum number of lines that the
+# initial value of a variable or macro / define can have for it to appear in the
+# documentation. If the initializer consists of more lines than specified here
+# it will be hidden. Use a value of 0 to hide initializers completely. The
+# appearance of the value of individual variables and macros / defines can be
+# controlled using \showinitializer or \hideinitializer command in the
+# documentation regardless of this setting.
+# Minimum value: 0, maximum value: 10000, default value: 30.
+
+MAX_INITIALIZER_LINES  = 30
+
+# Set the SHOW_USED_FILES tag to NO to disable the list of files generated at
+# the bottom of the documentation of classes and structs. If set to YES the list
+# will mention the files that were used to generate the documentation.
+# The default value is: YES.
+
+SHOW_USED_FILES        = YES
+
+# Set the SHOW_FILES tag to NO to disable the generation of the Files page. This
+# will remove the Files entry from the Quick Index and from the Folder Tree View
+# (if specified).
+# The default value is: YES.
+
+SHOW_FILES             = YES
+
+# Set the SHOW_NAMESPACES tag to NO to disable the generation of the Namespaces
+# page. This will remove the Namespaces entry from the Quick Index and from the
+# Folder Tree View (if specified).
+# The default value is: YES.
+
+SHOW_NAMESPACES        = YES
+
+# The FILE_VERSION_FILTER tag can be used to specify a program or script that
+# doxygen should invoke to get the current version for each file (typically from
+# the version control system). Doxygen will invoke the program by executing (via
+# popen()) the command command input-file, where command is the value of the
+# FILE_VERSION_FILTER tag, and input-file is the name of an input file provided
+# by doxygen. Whatever the program writes to standard output is used as the file
+# version. For an example see the documentation.
+
+FILE_VERSION_FILTER    =
+
+# The LAYOUT_FILE tag can be used to specify a layout file which will be parsed
+# by doxygen. The layout file controls the global structure of the generated
+# output files in an output format independent way. To create the layout file
+# that represents doxygen's defaults, run doxygen with the -l option. You can
+# optionally specify a file name after the option, if omitted DoxygenLayout.xml
+# will be used as the name of the layout file.
+#
+# Note that if you run doxygen from a directory containing a file called
+# DoxygenLayout.xml, doxygen will parse it automatically even if the LAYOUT_FILE
+# tag is left empty.
+
+LAYOUT_FILE            =
+
+# The CITE_BIB_FILES tag can be used to specify one or more bib files containing
+# the reference definitions. This must be a list of .bib files. The .bib
+# extension is automatically appended if omitted. This requires the bibtex tool
+# to be installed. See also http://en.wikipedia.org/wiki/BibTeX for more info.
+# For LaTeX the style of the bibliography can be controlled using
+# LATEX_BIB_STYLE. To use this feature you need bibtex and perl available in the
+# search path. Do not use file names with spaces, bibtex cannot handle them. See
+# also \cite for info how to create references.
+
+CITE_BIB_FILES         =
+
+#---------------------------------------------------------------------------
+# Configuration options related to warning and progress messages
+#---------------------------------------------------------------------------
+
+# The QUIET tag can be used to turn on/off the messages that are generated to
+# standard output by doxygen. If QUIET is set to YES this implies that the
+# messages are off.
+# The default value is: NO.
+
+QUIET                  = NO
+
+# The WARNINGS tag can be used to turn on/off the warning messages that are
+# generated to standard error ( stderr) by doxygen. If WARNINGS is set to YES
+# this implies that the warnings are on.
+#
+# Tip: Turn warnings on while writing the documentation.
+# The default value is: YES.
+
+WARNINGS               = YES
+
+# If the WARN_IF_UNDOCUMENTED tag is set to YES, then doxygen will generate
+# warnings for undocumented members. If EXTRACT_ALL is set to YES then this flag
+# will automatically be disabled.
+# The default value is: YES.
+
+WARN_IF_UNDOCUMENTED   = YES
+
+# If the WARN_IF_DOC_ERROR tag is set to YES, doxygen will generate warnings for
+# potential errors in the documentation, such as not documenting some parameters
+# in a documented function, or documenting parameters that don't exist or using
+# markup commands wrongly.
+# The default value is: YES.
+
+WARN_IF_DOC_ERROR      = YES
+
+# This WARN_NO_PARAMDOC option can be enabled to get warnings for functions that
+# are documented, but have no documentation for their parameters or return
+# value. If set to NO doxygen will only warn about wrong or incomplete parameter
+# documentation, but not about the absence of documentation.
+# The default value is: NO.
+
+WARN_NO_PARAMDOC       = NO
+
+# The WARN_FORMAT tag determines the format of the warning messages that doxygen
+# can produce. The string should contain the $file, $line, and $text tags, which
+# will be replaced by the file and line number from which the warning originated
+# and the warning text. Optionally the format may contain $version, which will
+# be replaced by the version of the file (if it could be obtained via
+# FILE_VERSION_FILTER)
+# The default value is: $file:$line: $text.
+
+WARN_FORMAT            = "$file:$line: $text"
+
+# The WARN_LOGFILE tag can be used to specify a file to which warning and error
+# messages should be written. If left blank the output is written to standard
+# error (stderr).
+
+WARN_LOGFILE           =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the input files
+#---------------------------------------------------------------------------
+
+# The INPUT tag is used to specify the files and/or directories that contain
+# documented source files. You may enter file names like myfile.cpp or
+# directories like /usr/src/myproject. Separate the files or directories with
+# spaces.
+# Note: If this tag is empty the current directory is searched.
+
+INPUT                  = ../lib \
+		         ../tests \
+			 ../benchmarks
+
+# This tag can be used to specify the character encoding of the source files
+# that doxygen parses. Internally doxygen uses the UTF-8 encoding. Doxygen uses
+# libiconv (or the iconv built into libc) for the transcoding. See the libiconv
+# documentation (see: http://www.gnu.org/software/libiconv) for the list of
+# possible encodings.
+# The default value is: UTF-8.
+
+INPUT_ENCODING         = UTF-8
+
+# If the value of the INPUT tag contains directories, you can use the
+# FILE_PATTERNS tag to specify one or more wildcard patterns (like *.cpp and
+# *.h) to filter out the source-files in the directories. If left blank the
+# following patterns are tested:*.c, *.cc, *.cxx, *.cpp, *.c++, *.java, *.ii,
+# *.ixx, *.ipp, *.i++, *.inl, *.idl, *.ddl, *.odl, *.h, *.hh, *.hxx, *.hpp,
+# *.h++, *.cs, *.d, *.php, *.php4, *.php5, *.phtml, *.inc, *.m, *.markdown,
+# *.md, *.mm, *.dox, *.py, *.f90, *.f, *.for, *.tcl, *.vhd, *.vhdl, *.ucf,
+# *.qsf, *.as and *.js.
+
+FILE_PATTERNS          =
+
+# The RECURSIVE tag can be used to specify whether or not subdirectories should
+# be searched for input files as well.
+# The default value is: NO.
+
+RECURSIVE              = YES
+
+# The EXCLUDE tag can be used to specify files and/or directories that should be
+# excluded from the INPUT source files. This way you can easily exclude a
+# subdirectory from a directory tree whose root is specified with the INPUT tag.
+#
+# Note that relative paths are relative to the directory from which doxygen is
+# run.
+
+EXCLUDE                =
+
+# The EXCLUDE_SYMLINKS tag can be used to select whether or not files or
+# directories that are symbolic links (a Unix file system feature) are excluded
+# from the input.
+# The default value is: NO.
+
+EXCLUDE_SYMLINKS       = NO
+
+# If the value of the INPUT tag contains directories, you can use the
+# EXCLUDE_PATTERNS tag to specify one or more wildcard patterns to exclude
+# certain files from those directories.
+#
+# Note that the wildcards are matched against the file with absolute path, so to
+# exclude all test directories for example use the pattern */test/*
+
+EXCLUDE_PATTERNS       =
+
+# The EXCLUDE_SYMBOLS tag can be used to specify one or more symbol names
+# (namespaces, classes, functions, etc.) that should be excluded from the
+# output. The symbol name can be a fully qualified name, a word, or if the
+# wildcard * is used, a substring. Examples: ANamespace, AClass,
+# AClass::ANamespace, ANamespace::*Test
+#
+# Note that the wildcards are matched against the file with absolute path, so to
+# exclude all test directories use the pattern */test/*
+
+EXCLUDE_SYMBOLS        =
+
+# The EXAMPLE_PATH tag can be used to specify one or more files or directories
+# that contain example code fragments that are included (see the \include
+# command).
+
+EXAMPLE_PATH           =
+
+# If the value of the EXAMPLE_PATH tag contains directories, you can use the
+# EXAMPLE_PATTERNS tag to specify one or more wildcard pattern (like *.cpp and
+# *.h) to filter out the source-files in the directories. If left blank all
+# files are included.
+
+EXAMPLE_PATTERNS       =
+
+# If the EXAMPLE_RECURSIVE tag is set to YES then subdirectories will be
+# searched for input files to be used with the \include or \dontinclude commands
+# irrespective of the value of the RECURSIVE tag.
+# The default value is: NO.
+
+EXAMPLE_RECURSIVE      = NO
+
+# The IMAGE_PATH tag can be used to specify one or more files or directories
+# that contain images that are to be included in the documentation (see the
+# \image command).
+
+IMAGE_PATH             =
+
+# The INPUT_FILTER tag can be used to specify a program that doxygen should
+# invoke to filter for each input file. Doxygen will invoke the filter program
+# by executing (via popen()) the command:
+#
+# <filter> <input-file>
+#
+# where <filter> is the value of the INPUT_FILTER tag, and <input-file> is the
+# name of an input file. Doxygen will then use the output that the filter
+# program writes to standard output. If FILTER_PATTERNS is specified, this tag
+# will be ignored.
+#
+# Note that the filter must not add or remove lines; it is applied before the
+# code is scanned, but not when the output code is generated. If lines are added
+# or removed, the anchors will not be placed correctly.
+
+INPUT_FILTER           =
+
+# The FILTER_PATTERNS tag can be used to specify filters on a per file pattern
+# basis. Doxygen will compare the file name with each pattern and apply the
+# filter if there is a match. The filters are a list of the form: pattern=filter
+# (like *.cpp=my_cpp_filter). See INPUT_FILTER for further information on how
+# filters are used. If the FILTER_PATTERNS tag is empty or if none of the
+# patterns match the file name, INPUT_FILTER is applied.
+
+FILTER_PATTERNS        =
+
+# If the FILTER_SOURCE_FILES tag is set to YES, the input filter (if set using
+# INPUT_FILTER ) will also be used to filter the input files that are used for
+# producing the source files to browse (i.e. when SOURCE_BROWSER is set to YES).
+# The default value is: NO.
+
+FILTER_SOURCE_FILES    = NO
+
+# The FILTER_SOURCE_PATTERNS tag can be used to specify source filters per file
+# pattern. A pattern will override the setting for FILTER_PATTERN (if any) and
+# it is also possible to disable source filtering for a specific pattern using
+# *.ext= (so without naming a filter).
+# This tag requires that the tag FILTER_SOURCE_FILES is set to YES.
+
+FILTER_SOURCE_PATTERNS =
+
+# If the USE_MDFILE_AS_MAINPAGE tag refers to the name of a markdown file that
+# is part of the input, its contents will be placed on the main page
+# (index.html). This can be useful if you have a project on for instance GitHub
+# and want to reuse the introduction page also for the doxygen output.
+
+USE_MDFILE_AS_MAINPAGE =
+
+#---------------------------------------------------------------------------
+# Configuration options related to source browsing
+#---------------------------------------------------------------------------
+
+# If the SOURCE_BROWSER tag is set to YES then a list of source files will be
+# generated. Documented entities will be cross-referenced with these sources.
+#
+# Note: To get rid of all source code in the generated output, make sure that
+# also VERBATIM_HEADERS is set to NO.
+# The default value is: NO.
+
+SOURCE_BROWSER         = YES
+
+# Setting the INLINE_SOURCES tag to YES will include the body of functions,
+# classes and enums directly into the documentation.
+# The default value is: NO.
+
+INLINE_SOURCES         = NO
+
+# Setting the STRIP_CODE_COMMENTS tag to YES will instruct doxygen to hide any
+# special comment blocks from generated source code fragments. Normal C, C++ and
+# Fortran comments will always remain visible.
+# The default value is: YES.
+
+STRIP_CODE_COMMENTS    = YES
+
+# If the REFERENCED_BY_RELATION tag is set to YES then for each documented
+# function all documented functions referencing it will be listed.
+# The default value is: NO.
+
+REFERENCED_BY_RELATION = NO
+
+# If the REFERENCES_RELATION tag is set to YES then for each documented function
+# all documented entities called/used by that function will be listed.
+# The default value is: NO.
+
+REFERENCES_RELATION    = NO
+
+# If the REFERENCES_LINK_SOURCE tag is set to YES and SOURCE_BROWSER tag is set
+# to YES, then the hyperlinks from functions in REFERENCES_RELATION and
+# REFERENCED_BY_RELATION lists will link to the source code. Otherwise they will
+# link to the documentation.
+# The default value is: YES.
+
+REFERENCES_LINK_SOURCE = YES
+
+# If SOURCE_TOOLTIPS is enabled (the default) then hovering a hyperlink in the
+# source code will show a tooltip with additional information such as prototype,
+# brief description and links to the definition and documentation. Since this
+# will make the HTML file larger and loading of large files a bit slower, you
+# can opt to disable this feature.
+# The default value is: YES.
+# This tag requires that the tag SOURCE_BROWSER is set to YES.
+
+SOURCE_TOOLTIPS        = YES
+
+# If the USE_HTAGS tag is set to YES then the references to source code will
+# point to the HTML generated by the htags(1) tool instead of doxygen built-in
+# source browser. The htags tool is part of GNU's global source tagging system
+# (see http://www.gnu.org/software/global/global.html). You will need version
+# 4.8.6 or higher.
+#
+# To use it do the following:
+# - Install the latest version of global
+# - Enable SOURCE_BROWSER and USE_HTAGS in the config file
+# - Make sure the INPUT points to the root of the source tree
+# - Run doxygen as normal
+#
+# Doxygen will invoke htags (and that will in turn invoke gtags), so these
+# tools must be available from the command line (i.e. in the search path).
+#
+# The result: instead of the source browser generated by doxygen, the links to
+# source code will now point to the output of htags.
+# The default value is: NO.
+# This tag requires that the tag SOURCE_BROWSER is set to YES.
+
+USE_HTAGS              = NO
+
+# If the VERBATIM_HEADERS tag is set the YES then doxygen will generate a
+# verbatim copy of the header file for each class for which an include is
+# specified. Set to NO to disable this.
+# See also: Section \class.
+# The default value is: YES.
+
+VERBATIM_HEADERS       = YES
+
+#---------------------------------------------------------------------------
+# Configuration options related to the alphabetical class index
+#---------------------------------------------------------------------------
+
+# If the ALPHABETICAL_INDEX tag is set to YES, an alphabetical index of all
+# compounds will be generated. Enable this if the project contains a lot of
+# classes, structs, unions or interfaces.
+# The default value is: YES.
+
+ALPHABETICAL_INDEX     = YES
+
+# The COLS_IN_ALPHA_INDEX tag can be used to specify the number of columns in
+# which the alphabetical index list will be split.
+# Minimum value: 1, maximum value: 20, default value: 5.
+# This tag requires that the tag ALPHABETICAL_INDEX is set to YES.
+
+COLS_IN_ALPHA_INDEX    = 5
+
+# In case all classes in a project start with a common prefix, all classes will
+# be put under the same header in the alphabetical index. The IGNORE_PREFIX tag
+# can be used to specify a prefix (or a list of prefixes) that should be ignored
+# while generating the index headers.
+# This tag requires that the tag ALPHABETICAL_INDEX is set to YES.
+
+IGNORE_PREFIX          =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the HTML output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_HTML tag is set to YES doxygen will generate HTML output
+# The default value is: YES.
+
+GENERATE_HTML          = YES
+
+# The HTML_OUTPUT tag is used to specify where the HTML docs will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it.
+# The default directory is: html.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_OUTPUT            = html
+
+# The HTML_FILE_EXTENSION tag can be used to specify the file extension for each
+# generated HTML page (for example: .htm, .php, .asp).
+# The default value is: .html.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_FILE_EXTENSION    = .html
+
+# The HTML_HEADER tag can be used to specify a user-defined HTML header file for
+# each generated HTML page. If the tag is left blank doxygen will generate a
+# standard header.
+#
+# To get valid HTML the header file that includes any scripts and style sheets
+# that doxygen needs, which is dependent on the configuration options used (e.g.
+# the setting GENERATE_TREEVIEW). It is highly recommended to start with a
+# default header using
+# doxygen -w html new_header.html new_footer.html new_stylesheet.css
+# YourConfigFile
+# and then modify the file new_header.html. See also section "Doxygen usage"
+# for information on how to generate the default header that doxygen normally
+# uses.
+# Note: The header is subject to change so you typically have to regenerate the
+# default header when upgrading to a newer version of doxygen. For a description
+# of the possible markers and block names see the documentation.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_HEADER            =
+
+# The HTML_FOOTER tag can be used to specify a user-defined HTML footer for each
+# generated HTML page. If the tag is left blank doxygen will generate a standard
+# footer. See HTML_HEADER for more information on how to generate a default
+# footer and what special commands can be used inside the footer. See also
+# section "Doxygen usage" for information on how to generate the default footer
+# that doxygen normally uses.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_FOOTER            =
+
+# The HTML_STYLESHEET tag can be used to specify a user-defined cascading style
+# sheet that is used by each HTML page. It can be used to fine-tune the look of
+# the HTML output. If left blank doxygen will generate a default style sheet.
+# See also section "Doxygen usage" for information on how to generate the style
+# sheet that doxygen normally uses.
+# Note: It is recommended to use HTML_EXTRA_STYLESHEET instead of this tag, as
+# it is more robust and this tag (HTML_STYLESHEET) will in the future become
+# obsolete.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_STYLESHEET        =
+
+# The HTML_EXTRA_STYLESHEET tag can be used to specify an additional user-
+# defined cascading style sheet that is included after the standard style sheets
+# created by doxygen. Using this option one can overrule certain style aspects.
+# This is preferred over using HTML_STYLESHEET since it does not replace the
+# standard style sheet and is therefor more robust against future updates.
+# Doxygen will copy the style sheet file to the output directory. For an example
+# see the documentation.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_EXTRA_STYLESHEET  =
+
+# The HTML_EXTRA_FILES tag can be used to specify one or more extra images or
+# other source files which should be copied to the HTML output directory. Note
+# that these files will be copied to the base HTML output directory. Use the
+# $relpath^ marker in the HTML_HEADER and/or HTML_FOOTER files to load these
+# files. In the HTML_STYLESHEET file, use the file name only. Also note that the
+# files will be copied as-is; there are no commands or markers available.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_EXTRA_FILES       =
+
+# The HTML_COLORSTYLE_HUE tag controls the color of the HTML output. Doxygen
+# will adjust the colors in the stylesheet and background images according to
+# this color. Hue is specified as an angle on a colorwheel, see
+# http://en.wikipedia.org/wiki/Hue for more information. For instance the value
+# 0 represents red, 60 is yellow, 120 is green, 180 is cyan, 240 is blue, 300
+# purple, and 360 is red again.
+# Minimum value: 0, maximum value: 359, default value: 220.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_COLORSTYLE_HUE    = 220
+
+# The HTML_COLORSTYLE_SAT tag controls the purity (or saturation) of the colors
+# in the HTML output. For a value of 0 the output will use grayscales only. A
+# value of 255 will produce the most vivid colors.
+# Minimum value: 0, maximum value: 255, default value: 100.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_COLORSTYLE_SAT    = 100
+
+# The HTML_COLORSTYLE_GAMMA tag controls the gamma correction applied to the
+# luminance component of the colors in the HTML output. Values below 100
+# gradually make the output lighter, whereas values above 100 make the output
+# darker. The value divided by 100 is the actual gamma applied, so 80 represents
+# a gamma of 0.8, The value 220 represents a gamma of 2.2, and 100 does not
+# change the gamma.
+# Minimum value: 40, maximum value: 240, default value: 80.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_COLORSTYLE_GAMMA  = 80
+
+# If the HTML_TIMESTAMP tag is set to YES then the footer of each generated HTML
+# page will contain the date and time when the page was generated. Setting this
+# to NO can help when comparing the output of multiple runs.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_TIMESTAMP         = YES
+
+# If the HTML_DYNAMIC_SECTIONS tag is set to YES then the generated HTML
+# documentation will contain sections that can be hidden and shown after the
+# page has loaded.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_DYNAMIC_SECTIONS  = NO
+
+# With HTML_INDEX_NUM_ENTRIES one can control the preferred number of entries
+# shown in the various tree structured indices initially; the user can expand
+# and collapse entries dynamically later on. Doxygen will expand the tree to
+# such a level that at most the specified number of entries are visible (unless
+# a fully collapsed tree already exceeds this amount). So setting the number of
+# entries 1 will produce a full collapsed tree by default. 0 is a special value
+# representing an infinite number of entries and will result in a full expanded
+# tree by default.
+# Minimum value: 0, maximum value: 9999, default value: 100.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+HTML_INDEX_NUM_ENTRIES = 100
+
+# If the GENERATE_DOCSET tag is set to YES, additional index files will be
+# generated that can be used as input for Apple's Xcode 3 integrated development
+# environment (see: http://developer.apple.com/tools/xcode/), introduced with
+# OSX 10.5 (Leopard). To create a documentation set, doxygen will generate a
+# Makefile in the HTML output directory. Running make will produce the docset in
+# that directory and running make install will install the docset in
+# ~/Library/Developer/Shared/Documentation/DocSets so that Xcode will find it at
+# startup. See http://developer.apple.com/tools/creatingdocsetswithdoxygen.html
+# for more information.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_DOCSET        = NO
+
+# This tag determines the name of the docset feed. A documentation feed provides
+# an umbrella under which multiple documentation sets from a single provider
+# (such as a company or product suite) can be grouped.
+# The default value is: Doxygen generated docs.
+# This tag requires that the tag GENERATE_DOCSET is set to YES.
+
+DOCSET_FEEDNAME        = "Doxygen generated docs"
+
+# This tag specifies a string that should uniquely identify the documentation
+# set bundle. This should be a reverse domain-name style string, e.g.
+# com.mycompany.MyDocSet. Doxygen will append .docset to the name.
+# The default value is: org.doxygen.Project.
+# This tag requires that the tag GENERATE_DOCSET is set to YES.
+
+DOCSET_BUNDLE_ID       = org.doxygen.Project
+
+# The DOCSET_PUBLISHER_ID tag specifies a string that should uniquely identify
+# the documentation publisher. This should be a reverse domain-name style
+# string, e.g. com.mycompany.MyDocSet.documentation.
+# The default value is: org.doxygen.Publisher.
+# This tag requires that the tag GENERATE_DOCSET is set to YES.
+
+DOCSET_PUBLISHER_ID    = org.doxygen.Publisher
+
+# The DOCSET_PUBLISHER_NAME tag identifies the documentation publisher.
+# The default value is: Publisher.
+# This tag requires that the tag GENERATE_DOCSET is set to YES.
+
+DOCSET_PUBLISHER_NAME  = Publisher
+
+# If the GENERATE_HTMLHELP tag is set to YES then doxygen generates three
+# additional HTML index files: index.hhp, index.hhc, and index.hhk. The
+# index.hhp is a project file that can be read by Microsoft's HTML Help Workshop
+# (see: http://www.microsoft.com/en-us/download/details.aspx?id=21138) on
+# Windows.
+#
+# The HTML Help Workshop contains a compiler that can convert all HTML output
+# generated by doxygen into a single compiled HTML file (.chm). Compiled HTML
+# files are now used as the Windows 98 help format, and will replace the old
+# Windows help format (.hlp) on all Windows platforms in the future. Compressed
+# HTML files also contain an index, a table of contents, and you can search for
+# words in the documentation. The HTML workshop also contains a viewer for
+# compressed HTML files.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_HTMLHELP      = NO
+
+# The CHM_FILE tag can be used to specify the file name of the resulting .chm
+# file. You can add a path in front of the file if the result should not be
+# written to the html output directory.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+CHM_FILE               =
+
+# The HHC_LOCATION tag can be used to specify the location (absolute path
+# including file name) of the HTML help compiler ( hhc.exe). If non-empty
+# doxygen will try to run the HTML help compiler on the generated index.hhp.
+# The file has to be specified with full path.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+HHC_LOCATION           =
+
+# The GENERATE_CHI flag controls if a separate .chi index file is generated (
+# YES) or that it should be included in the master .chm file ( NO).
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+GENERATE_CHI           = NO
+
+# The CHM_INDEX_ENCODING is used to encode HtmlHelp index ( hhk), content ( hhc)
+# and project file content.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+CHM_INDEX_ENCODING     =
+
+# The BINARY_TOC flag controls whether a binary table of contents is generated (
+# YES) or a normal table of contents ( NO) in the .chm file.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+BINARY_TOC             = NO
+
+# The TOC_EXPAND flag can be set to YES to add extra items for group members to
+# the table of contents of the HTML help documentation and to the tree view.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
+
+TOC_EXPAND             = NO
+
+# If the GENERATE_QHP tag is set to YES and both QHP_NAMESPACE and
+# QHP_VIRTUAL_FOLDER are set, an additional index file will be generated that
+# can be used as input for Qt's qhelpgenerator to generate a Qt Compressed Help
+# (.qch) of the generated HTML documentation.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_QHP           = NO
+
+# If the QHG_LOCATION tag is specified, the QCH_FILE tag can be used to specify
+# the file name of the resulting .qch file. The path specified is relative to
+# the HTML output folder.
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QCH_FILE               =
+
+# The QHP_NAMESPACE tag specifies the namespace to use when generating Qt Help
+# Project output. For more information please see Qt Help Project / Namespace
+# (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#namespace).
+# The default value is: org.doxygen.Project.
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_NAMESPACE          = org.doxygen.Project
+
+# The QHP_VIRTUAL_FOLDER tag specifies the namespace to use when generating Qt
+# Help Project output. For more information please see Qt Help Project / Virtual
+# Folders (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#virtual-
+# folders).
+# The default value is: doc.
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_VIRTUAL_FOLDER     = doc
+
+# If the QHP_CUST_FILTER_NAME tag is set, it specifies the name of a custom
+# filter to add. For more information please see Qt Help Project / Custom
+# Filters (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#custom-
+# filters).
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_CUST_FILTER_NAME   =
+
+# The QHP_CUST_FILTER_ATTRS tag specifies the list of the attributes of the
+# custom filter to add. For more information please see Qt Help Project / Custom
+# Filters (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#custom-
+# filters).
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_CUST_FILTER_ATTRS  =
+
+# The QHP_SECT_FILTER_ATTRS tag specifies the list of the attributes this
+# project's filter section matches. Qt Help Project / Filter Attributes (see:
+# http://qt-project.org/doc/qt-4.8/qthelpproject.html#filter-attributes).
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHP_SECT_FILTER_ATTRS  =
+
+# The QHG_LOCATION tag can be used to specify the location of Qt's
+# qhelpgenerator. If non-empty doxygen will try to run qhelpgenerator on the
+# generated .qhp file.
+# This tag requires that the tag GENERATE_QHP is set to YES.
+
+QHG_LOCATION           =
+
+# If the GENERATE_ECLIPSEHELP tag is set to YES, additional index files will be
+# generated, together with the HTML files, they form an Eclipse help plugin. To
+# install this plugin and make it available under the help contents menu in
+# Eclipse, the contents of the directory containing the HTML and XML files needs
+# to be copied into the plugins directory of eclipse. The name of the directory
+# within the plugins directory should be the same as the ECLIPSE_DOC_ID value.
+# After copying Eclipse needs to be restarted before the help appears.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_ECLIPSEHELP   = NO
+
+# A unique identifier for the Eclipse help plugin. When installing the plugin
+# the directory name containing the HTML and XML files should also have this
+# name. Each documentation set should have its own identifier.
+# The default value is: org.doxygen.Project.
+# This tag requires that the tag GENERATE_ECLIPSEHELP is set to YES.
+
+ECLIPSE_DOC_ID         = org.doxygen.Project
+
+# If you want full control over the layout of the generated HTML pages it might
+# be necessary to disable the index and replace it with your own. The
+# DISABLE_INDEX tag can be used to turn on/off the condensed index (tabs) at top
+# of each HTML page. A value of NO enables the index and the value YES disables
+# it. Since the tabs in the index contain the same information as the navigation
+# tree, you can set this option to YES if you also set GENERATE_TREEVIEW to YES.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+DISABLE_INDEX          = NO
+
+# The GENERATE_TREEVIEW tag is used to specify whether a tree-like index
+# structure should be generated to display hierarchical information. If the tag
+# value is set to YES, a side panel will be generated containing a tree-like
+# index structure (just like the one that is generated for HTML Help). For this
+# to work a browser that supports JavaScript, DHTML, CSS and frames is required
+# (i.e. any modern browser). Windows users are probably better off using the
+# HTML help feature. Via custom stylesheets (see HTML_EXTRA_STYLESHEET) one can
+# further fine-tune the look of the index. As an example, the default style
+# sheet generated by doxygen has an example that shows how to put an image at
+# the root of the tree instead of the PROJECT_NAME. Since the tree basically has
+# the same information as the tab index, you could consider setting
+# DISABLE_INDEX to YES when enabling this option.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+GENERATE_TREEVIEW      = YES
+
+# The ENUM_VALUES_PER_LINE tag can be used to set the number of enum values that
+# doxygen will group on one line in the generated HTML documentation.
+#
+# Note that a value of 0 will completely suppress the enum values from appearing
+# in the overview section.
+# Minimum value: 0, maximum value: 20, default value: 4.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+ENUM_VALUES_PER_LINE   = 4
+
+# If the treeview is enabled (see GENERATE_TREEVIEW) then this tag can be used
+# to set the initial width (in pixels) of the frame in which the tree is shown.
+# Minimum value: 0, maximum value: 1500, default value: 250.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+TREEVIEW_WIDTH         = 250
+
+# When the EXT_LINKS_IN_WINDOW option is set to YES doxygen will open links to
+# external symbols imported via tag files in a separate window.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+EXT_LINKS_IN_WINDOW    = NO
+
+# Use this tag to change the font size of LaTeX formulas included as images in
+# the HTML documentation. When you change the font size after a successful
+# doxygen run you need to manually remove any form_*.png images from the HTML
+# output directory to force them to be regenerated.
+# Minimum value: 8, maximum value: 50, default value: 10.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+FORMULA_FONTSIZE       = 10
+
+# Use the FORMULA_TRANPARENT tag to determine whether or not the images
+# generated for formulas are transparent PNGs. Transparent PNGs are not
+# supported properly for IE 6.0, but are supported on all modern browsers.
+#
+# Note that when changing this option you need to delete any form_*.png files in
+# the HTML output directory before the changes have effect.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+FORMULA_TRANSPARENT    = YES
+
+# Enable the USE_MATHJAX option to render LaTeX formulas using MathJax (see
+# http://www.mathjax.org) which uses client side Javascript for the rendering
+# instead of using prerendered bitmaps. Use this if you do not have LaTeX
+# installed or if you want to formulas look prettier in the HTML output. When
+# enabled you may also need to install MathJax separately and configure the path
+# to it using the MATHJAX_RELPATH option.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+USE_MATHJAX            = YES
+
+# When MathJax is enabled you can set the default output format to be used for
+# the MathJax output. See the MathJax site (see:
+# http://docs.mathjax.org/en/latest/output.html) for more details.
+# Possible values are: HTML-CSS (which is slower, but has the best
+# compatibility), NativeMML (i.e. MathML) and SVG.
+# The default value is: HTML-CSS.
+# This tag requires that the tag USE_MATHJAX is set to YES.
+
+MATHJAX_FORMAT         = HTML-CSS
+
+# When MathJax is enabled you need to specify the location relative to the HTML
+# output directory using the MATHJAX_RELPATH option. The destination directory
+# should contain the MathJax.js script. For instance, if the mathjax directory
+# is located at the same level as the HTML output directory, then
+# MATHJAX_RELPATH should be ../mathjax. The default value points to the MathJax
+# Content Delivery Network so you can quickly see the result without installing
+# MathJax. However, it is strongly recommended to install a local copy of
+# MathJax from http://www.mathjax.org before deployment.
+# The default value is: http://cdn.mathjax.org/mathjax/latest.
+# This tag requires that the tag USE_MATHJAX is set to YES.
+
+MATHJAX_RELPATH        = http://cdn.mathjax.org/mathjax/latest
+
+# The MATHJAX_EXTENSIONS tag can be used to specify one or more MathJax
+# extension names that should be enabled during MathJax rendering. For example
+# MATHJAX_EXTENSIONS = TeX/AMSmath TeX/AMSsymbols
+# This tag requires that the tag USE_MATHJAX is set to YES.
+
+MATHJAX_EXTENSIONS     =
+
+# The MATHJAX_CODEFILE tag can be used to specify a file with javascript pieces
+# of code that will be used on startup of the MathJax code. See the MathJax site
+# (see: http://docs.mathjax.org/en/latest/output.html) for more details. For an
+# example see the documentation.
+# This tag requires that the tag USE_MATHJAX is set to YES.
+
+MATHJAX_CODEFILE       =
+
+# When the SEARCHENGINE tag is enabled doxygen will generate a search box for
+# the HTML output. The underlying search engine uses javascript and DHTML and
+# should work on any modern browser. Note that when using HTML help
+# (GENERATE_HTMLHELP), Qt help (GENERATE_QHP), or docsets (GENERATE_DOCSET)
+# there is already a search function so this one should typically be disabled.
+# For large projects the javascript based search engine can be slow, then
+# enabling SERVER_BASED_SEARCH may provide a better solution. It is possible to
+# search using the keyboard; to jump to the search box use <access key> + S
+# (what the <access key> is depends on the OS and browser, but it is typically
+# <CTRL>, <ALT>/<option>, or both). Inside the search box use the <cursor down
+# key> to jump into the search results window, the results can be navigated
+# using the <cursor keys>. Press <Enter> to select an item or <escape> to cancel
+# the search. The filter options can be selected when the cursor is inside the
+# search box by pressing <Shift>+<cursor down>. Also here use the <cursor keys>
+# to select a filter and <Enter> or <escape> to activate or cancel the filter
+# option.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_HTML is set to YES.
+
+SEARCHENGINE           = YES
+
+# When the SERVER_BASED_SEARCH tag is enabled the search engine will be
+# implemented using a web server instead of a web client using Javascript. There
+# are two flavours of web server based searching depending on the
+# EXTERNAL_SEARCH setting. When disabled, doxygen will generate a PHP script for
+# searching and an index file used by the script. When EXTERNAL_SEARCH is
+# enabled the indexing and searching needs to be provided by external tools. See
+# the section "External Indexing and Searching" for details.
+# The default value is: NO.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+SERVER_BASED_SEARCH    = NO
+
+# When EXTERNAL_SEARCH tag is enabled doxygen will no longer generate the PHP
+# script for searching. Instead the search results are written to an XML file
+# which needs to be processed by an external indexer. Doxygen will invoke an
+# external search engine pointed to by the SEARCHENGINE_URL option to obtain the
+# search results.
+#
+# Doxygen ships with an example indexer ( doxyindexer) and search engine
+# (doxysearch.cgi) which are based on the open source search engine library
+# Xapian (see: http://xapian.org/).
+#
+# See the section "External Indexing and Searching" for details.
+# The default value is: NO.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+EXTERNAL_SEARCH        = NO
+
+# The SEARCHENGINE_URL should point to a search engine hosted by a web server
+# which will return the search results when EXTERNAL_SEARCH is enabled.
+#
+# Doxygen ships with an example indexer ( doxyindexer) and search engine
+# (doxysearch.cgi) which are based on the open source search engine library
+# Xapian (see: http://xapian.org/). See the section "External Indexing and
+# Searching" for details.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+SEARCHENGINE_URL       =
+
+# When SERVER_BASED_SEARCH and EXTERNAL_SEARCH are both enabled the unindexed
+# search data is written to a file for indexing by an external tool. With the
+# SEARCHDATA_FILE tag the name of this file can be specified.
+# The default file is: searchdata.xml.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+SEARCHDATA_FILE        = searchdata.xml
+
+# When SERVER_BASED_SEARCH and EXTERNAL_SEARCH are both enabled the
+# EXTERNAL_SEARCH_ID tag can be used as an identifier for the project. This is
+# useful in combination with EXTRA_SEARCH_MAPPINGS to search through multiple
+# projects and redirect the results back to the right project.
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+EXTERNAL_SEARCH_ID     =
+
+# The EXTRA_SEARCH_MAPPINGS tag can be used to enable searching through doxygen
+# projects other than the one defined by this configuration file, but that are
+# all added to the same external search index. Each project needs to have a
+# unique id set via EXTERNAL_SEARCH_ID. The search mapping then maps the id of
+# to a relative location where the documentation can be found. The format is:
+# EXTRA_SEARCH_MAPPINGS = tagname1=loc1 tagname2=loc2 ...
+# This tag requires that the tag SEARCHENGINE is set to YES.
+
+EXTRA_SEARCH_MAPPINGS  =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the LaTeX output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_LATEX tag is set to YES doxygen will generate LaTeX output.
+# The default value is: YES.
+
+GENERATE_LATEX         = YES
+
+# The LATEX_OUTPUT tag is used to specify where the LaTeX docs will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it.
+# The default directory is: latex.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_OUTPUT           = latex
+
+# The LATEX_CMD_NAME tag can be used to specify the LaTeX command name to be
+# invoked.
+#
+# Note that when enabling USE_PDFLATEX this option is only used for generating
+# bitmaps for formulas in the HTML output, but not in the Makefile that is
+# written to the output directory.
+# The default file is: latex.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_CMD_NAME         = latex
+
+# The MAKEINDEX_CMD_NAME tag can be used to specify the command name to generate
+# index for LaTeX.
+# The default file is: makeindex.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+MAKEINDEX_CMD_NAME     = makeindex
+
+# If the COMPACT_LATEX tag is set to YES doxygen generates more compact LaTeX
+# documents. This may be useful for small projects and may help to save some
+# trees in general.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+COMPACT_LATEX          = NO
+
+# The PAPER_TYPE tag can be used to set the paper type that is used by the
+# printer.
+# Possible values are: a4 (210 x 297 mm), letter (8.5 x 11 inches), legal (8.5 x
+# 14 inches) and executive (7.25 x 10.5 inches).
+# The default value is: a4.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+PAPER_TYPE             = a4
+
+# The EXTRA_PACKAGES tag can be used to specify one or more LaTeX package names
+# that should be included in the LaTeX output. To get the times font for
+# instance you can specify
+# EXTRA_PACKAGES=times
+# If left blank no extra packages will be included.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+EXTRA_PACKAGES         =
+
+# The LATEX_HEADER tag can be used to specify a personal LaTeX header for the
+# generated LaTeX document. The header should contain everything until the first
+# chapter. If it is left blank doxygen will generate a standard header. See
+# section "Doxygen usage" for information on how to let doxygen write the
+# default header to a separate file.
+#
+# Note: Only use a user-defined header if you know what you are doing! The
+# following commands have a special meaning inside the header: $title,
+# $datetime, $date, $doxygenversion, $projectname, $projectnumber. Doxygen will
+# replace them by respectively the title of the page, the current date and time,
+# only the current date, the version number of doxygen, the project name (see
+# PROJECT_NAME), or the project number (see PROJECT_NUMBER).
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_HEADER           =
+
+# The LATEX_FOOTER tag can be used to specify a personal LaTeX footer for the
+# generated LaTeX document. The footer should contain everything after the last
+# chapter. If it is left blank doxygen will generate a standard footer.
+#
+# Note: Only use a user-defined footer if you know what you are doing!
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_FOOTER           =
+
+# The LATEX_EXTRA_FILES tag can be used to specify one or more extra images or
+# other source files which should be copied to the LATEX_OUTPUT output
+# directory. Note that the files will be copied as-is; there are no commands or
+# markers available.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_EXTRA_FILES      =
+
+# If the PDF_HYPERLINKS tag is set to YES, the LaTeX that is generated is
+# prepared for conversion to PDF (using ps2pdf or pdflatex). The PDF file will
+# contain links (just like the HTML output) instead of page references. This
+# makes the output suitable for online browsing using a PDF viewer.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+PDF_HYPERLINKS         = YES
+
+# If the LATEX_PDFLATEX tag is set to YES, doxygen will use pdflatex to generate
+# the PDF file directly from the LaTeX files. Set this option to YES to get a
+# higher quality PDF documentation.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+USE_PDFLATEX           = YES
+
+# If the LATEX_BATCHMODE tag is set to YES, doxygen will add the \batchmode
+# command to the generated LaTeX files. This will instruct LaTeX to keep running
+# if errors occur, instead of asking the user for help. This option is also used
+# when generating formulas in HTML.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_BATCHMODE        = NO
+
+# If the LATEX_HIDE_INDICES tag is set to YES then doxygen will not include the
+# index chapters (such as File Index, Compound Index, etc.) in the output.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_HIDE_INDICES     = NO
+
+# If the LATEX_SOURCE_CODE tag is set to YES then doxygen will include source
+# code with syntax highlighting in the LaTeX output.
+#
+# Note that which sources are shown also depends on other settings such as
+# SOURCE_BROWSER.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_SOURCE_CODE      = NO
+
+# The LATEX_BIB_STYLE tag can be used to specify the style to use for the
+# bibliography, e.g. plainnat, or ieeetr. See
+# http://en.wikipedia.org/wiki/BibTeX and \cite for more info.
+# The default value is: plain.
+# This tag requires that the tag GENERATE_LATEX is set to YES.
+
+LATEX_BIB_STYLE        = plain
+
+#---------------------------------------------------------------------------
+# Configuration options related to the RTF output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_RTF tag is set to YES doxygen will generate RTF output. The
+# RTF output is optimized for Word 97 and may not look too pretty with other RTF
+# readers/editors.
+# The default value is: NO.
+
+GENERATE_RTF           = NO
+
+# The RTF_OUTPUT tag is used to specify where the RTF docs will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it.
+# The default directory is: rtf.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+RTF_OUTPUT             = rtf
+
+# If the COMPACT_RTF tag is set to YES doxygen generates more compact RTF
+# documents. This may be useful for small projects and may help to save some
+# trees in general.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+COMPACT_RTF            = NO
+
+# If the RTF_HYPERLINKS tag is set to YES, the RTF that is generated will
+# contain hyperlink fields. The RTF file will contain links (just like the HTML
+# output) instead of page references. This makes the output suitable for online
+# browsing using Word or some other Word compatible readers that support those
+# fields.
+#
+# Note: WordPad (write) and others do not support links.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+RTF_HYPERLINKS         = NO
+
+# Load stylesheet definitions from file. Syntax is similar to doxygen's config
+# file, i.e. a series of assignments. You only have to provide replacements,
+# missing definitions are set to their default value.
+#
+# See also section "Doxygen usage" for information on how to generate the
+# default style sheet that doxygen normally uses.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+RTF_STYLESHEET_FILE    =
+
+# Set optional variables used in the generation of an RTF document. Syntax is
+# similar to doxygen's config file. A template extensions file can be generated
+# using doxygen -e rtf extensionFile.
+# This tag requires that the tag GENERATE_RTF is set to YES.
+
+RTF_EXTENSIONS_FILE    =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the man page output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_MAN tag is set to YES doxygen will generate man pages for
+# classes and files.
+# The default value is: NO.
+
+GENERATE_MAN           = NO
+
+# The MAN_OUTPUT tag is used to specify where the man pages will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it. A directory man3 will be created inside the directory specified by
+# MAN_OUTPUT.
+# The default directory is: man.
+# This tag requires that the tag GENERATE_MAN is set to YES.
+
+MAN_OUTPUT             = man
+
+# The MAN_EXTENSION tag determines the extension that is added to the generated
+# man pages. In case the manual section does not start with a number, the number
+# 3 is prepended. The dot (.) at the beginning of the MAN_EXTENSION tag is
+# optional.
+# The default value is: .3.
+# This tag requires that the tag GENERATE_MAN is set to YES.
+
+MAN_EXTENSION          = .3
+
+# If the MAN_LINKS tag is set to YES and doxygen generates man output, then it
+# will generate one additional man file for each entity documented in the real
+# man page(s). These additional files only source the real man page, but without
+# them the man command would be unable to find the correct page.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_MAN is set to YES.
+
+MAN_LINKS              = NO
+
+#---------------------------------------------------------------------------
+# Configuration options related to the XML output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_XML tag is set to YES doxygen will generate an XML file that
+# captures the structure of the code including all documentation.
+# The default value is: NO.
+
+GENERATE_XML           = NO
+
+# The XML_OUTPUT tag is used to specify where the XML pages will be put. If a
+# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
+# it.
+# The default directory is: xml.
+# This tag requires that the tag GENERATE_XML is set to YES.
+
+XML_OUTPUT             = xml
+
+# The XML_SCHEMA tag can be used to specify a XML schema, which can be used by a
+# validating XML parser to check the syntax of the XML files.
+# This tag requires that the tag GENERATE_XML is set to YES.
+
+XML_SCHEMA             =
+
+# The XML_DTD tag can be used to specify a XML DTD, which can be used by a
+# validating XML parser to check the syntax of the XML files.
+# This tag requires that the tag GENERATE_XML is set to YES.
+
+XML_DTD                =
+
+# If the XML_PROGRAMLISTING tag is set to YES doxygen will dump the program
+# listings (including syntax highlighting and cross-referencing information) to
+# the XML output. Note that enabling this will significantly increase the size
+# of the XML output.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_XML is set to YES.
+
+XML_PROGRAMLISTING     = YES
+
+#---------------------------------------------------------------------------
+# Configuration options related to the DOCBOOK output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_DOCBOOK tag is set to YES doxygen will generate Docbook files
+# that can be used to generate PDF.
+# The default value is: NO.
+
+GENERATE_DOCBOOK       = NO
+
+# The DOCBOOK_OUTPUT tag is used to specify where the Docbook pages will be put.
+# If a relative path is entered the value of OUTPUT_DIRECTORY will be put in
+# front of it.
+# The default directory is: docbook.
+# This tag requires that the tag GENERATE_DOCBOOK is set to YES.
+
+DOCBOOK_OUTPUT         = docbook
+
+#---------------------------------------------------------------------------
+# Configuration options for the AutoGen Definitions output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_AUTOGEN_DEF tag is set to YES doxygen will generate an AutoGen
+# Definitions (see http://autogen.sf.net) file that captures the structure of
+# the code including all documentation. Note that this feature is still
+# experimental and incomplete at the moment.
+# The default value is: NO.
+
+GENERATE_AUTOGEN_DEF   = NO
+
+#---------------------------------------------------------------------------
+# Configuration options related to the Perl module output
+#---------------------------------------------------------------------------
+
+# If the GENERATE_PERLMOD tag is set to YES doxygen will generate a Perl module
+# file that captures the structure of the code including all documentation.
+#
+# Note that this feature is still experimental and incomplete at the moment.
+# The default value is: NO.
+
+GENERATE_PERLMOD       = NO
+
+# If the PERLMOD_LATEX tag is set to YES doxygen will generate the necessary
+# Makefile rules, Perl scripts and LaTeX code to be able to generate PDF and DVI
+# output from the Perl module output.
+# The default value is: NO.
+# This tag requires that the tag GENERATE_PERLMOD is set to YES.
+
+PERLMOD_LATEX          = NO
+
+# If the PERLMOD_PRETTY tag is set to YES the Perl module output will be nicely
+# formatted so it can be parsed by a human reader. This is useful if you want to
+# understand what is going on. On the other hand, if this tag is set to NO the
+# size of the Perl module output will be much smaller and Perl will parse it
+# just the same.
+# The default value is: YES.
+# This tag requires that the tag GENERATE_PERLMOD is set to YES.
+
+PERLMOD_PRETTY         = YES
+
+# The names of the make variables in the generated doxyrules.make file are
+# prefixed with the string contained in PERLMOD_MAKEVAR_PREFIX. This is useful
+# so different doxyrules.make files included by the same Makefile don't
+# overwrite each other's variables.
+# This tag requires that the tag GENERATE_PERLMOD is set to YES.
+
+PERLMOD_MAKEVAR_PREFIX =
+
+#---------------------------------------------------------------------------
+# Configuration options related to the preprocessor
+#---------------------------------------------------------------------------
+
+# If the ENABLE_PREPROCESSING tag is set to YES doxygen will evaluate all
+# C-preprocessor directives found in the sources and include files.
+# The default value is: YES.
+
+ENABLE_PREPROCESSING   = YES
+
+# If the MACRO_EXPANSION tag is set to YES doxygen will expand all macro names
+# in the source code. If set to NO only conditional compilation will be
+# performed. Macro expansion can be done in a controlled way by setting
+# EXPAND_ONLY_PREDEF to YES.
+# The default value is: NO.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+MACRO_EXPANSION        = NO
+
+# If the EXPAND_ONLY_PREDEF and MACRO_EXPANSION tags are both set to YES then
+# the macro expansion is limited to the macros specified with the PREDEFINED and
+# EXPAND_AS_DEFINED tags.
+# The default value is: NO.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+EXPAND_ONLY_PREDEF     = NO
+
+# If the SEARCH_INCLUDES tag is set to YES the includes files in the
+# INCLUDE_PATH will be searched if a #include is found.
+# The default value is: YES.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+SEARCH_INCLUDES        = YES
+
+# The INCLUDE_PATH tag can be used to specify one or more directories that
+# contain include files that are not input files but should be processed by the
+# preprocessor.
+# This tag requires that the tag SEARCH_INCLUDES is set to YES.
+
+INCLUDE_PATH           =
+
+# You can use the INCLUDE_FILE_PATTERNS tag to specify one or more wildcard
+# patterns (like *.h and *.hpp) to filter out the header-files in the
+# directories. If left blank, the patterns specified with FILE_PATTERNS will be
+# used.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+INCLUDE_FILE_PATTERNS  =
+
+# The PREDEFINED tag can be used to specify one or more macro names that are
+# defined before the preprocessor is started (similar to the -D option of e.g.
+# gcc). The argument of the tag is a list of macros of the form: name or
+# name=definition (no spaces). If the definition and the "=" are omitted, "=1"
+# is assumed. To prevent a macro definition from being undefined via #undef or
+# recursively expanded use the := operator instead of the = operator.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+PREDEFINED             =
+
+# If the MACRO_EXPANSION and EXPAND_ONLY_PREDEF tags are set to YES then this
+# tag can be used to specify a list of macro names that should be expanded. The
+# macro definition that is found in the sources will be used. Use the PREDEFINED
+# tag if you want to use a different macro definition that overrules the
+# definition found in the source code.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+EXPAND_AS_DEFINED      =
+
+# If the SKIP_FUNCTION_MACROS tag is set to YES then doxygen's preprocessor will
+# remove all refrences to function-like macros that are alone on a line, have an
+# all uppercase name, and do not end with a semicolon. Such function macros are
+# typically used for boiler-plate code, and will confuse the parser if not
+# removed.
+# The default value is: YES.
+# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
+
+SKIP_FUNCTION_MACROS   = YES
+
+#---------------------------------------------------------------------------
+# Configuration options related to external references
+#---------------------------------------------------------------------------
+
+# The TAGFILES tag can be used to specify one or more tag files. For each tag
+# file the location of the external documentation should be added. The format of
+# a tag file without this location is as follows:
+# TAGFILES = file1 file2 ...
+# Adding location for the tag files is done as follows:
+# TAGFILES = file1=loc1 "file2 = loc2" ...
+# where loc1 and loc2 can be relative or absolute paths or URLs. See the
+# section "Linking to external documentation" for more information about the use
+# of tag files.
+# Note: Each tag file must have an unique name (where the name does NOT include
+# the path). If a tag file is not located in the directory in which doxygen is
+# run, you must also specify the path to the tagfile here.
+
+TAGFILES               =
+
+# When a file name is specified after GENERATE_TAGFILE, doxygen will create a
+# tag file that is based on the input files it reads. See section "Linking to
+# external documentation" for more information about the usage of tag files.
+
+GENERATE_TAGFILE       =
+
+# If the ALLEXTERNALS tag is set to YES all external class will be listed in the
+# class index. If set to NO only the inherited external classes will be listed.
+# The default value is: NO.
+
+ALLEXTERNALS           = NO
+
+# If the EXTERNAL_GROUPS tag is set to YES all external groups will be listed in
+# the modules index. If set to NO, only the current project's groups will be
+# listed.
+# The default value is: YES.
+
+EXTERNAL_GROUPS        = YES
+
+# If the EXTERNAL_PAGES tag is set to YES all external pages will be listed in
+# the related pages index. If set to NO, only the current project's pages will
+# be listed.
+# The default value is: YES.
+
+EXTERNAL_PAGES         = YES
+
+# The PERL_PATH should be the absolute path and name of the perl script
+# interpreter (i.e. the result of 'which perl').
+# The default file (with absolute path) is: /usr/bin/perl.
+
+PERL_PATH              = /usr/bin/perl
+
+#---------------------------------------------------------------------------
+# Configuration options related to the dot tool
+#---------------------------------------------------------------------------
+
+# If the CLASS_DIAGRAMS tag is set to YES doxygen will generate a class diagram
+# (in HTML and LaTeX) for classes with base or super classes. Setting the tag to
+# NO turns the diagrams off. Note that this option also works with HAVE_DOT
+# disabled, but it is recommended to install and use dot, since it yields more
+# powerful graphs.
+# The default value is: YES.
+
+CLASS_DIAGRAMS         = YES
+
+# You can define message sequence charts within doxygen comments using the \msc
+# command. Doxygen will then run the mscgen tool (see:
+# http://www.mcternan.me.uk/mscgen/)) to produce the chart and insert it in the
+# documentation. The MSCGEN_PATH tag allows you to specify the directory where
+# the mscgen tool resides. If left empty the tool is assumed to be found in the
+# default search path.
+
+MSCGEN_PATH            =
+
+# You can include diagrams made with dia in doxygen documentation. Doxygen will
+# then run dia to produce the diagram and insert it in the documentation. The
+# DIA_PATH tag allows you to specify the directory where the dia binary resides.
+# If left empty dia is assumed to be found in the default search path.
+
+DIA_PATH               =
+
+# If set to YES, the inheritance and collaboration graphs will hide inheritance
+# and usage relations if the target is undocumented or is not a class.
+# The default value is: YES.
+
+HIDE_UNDOC_RELATIONS   = YES
+
+# If you set the HAVE_DOT tag to YES then doxygen will assume the dot tool is
+# available from the path. This tool is part of Graphviz (see:
+# http://www.graphviz.org/), a graph visualization toolkit from AT&T and Lucent
+# Bell Labs. The other options in this section have no effect if this option is
+# set to NO
+# The default value is: NO.
+
+HAVE_DOT               = YES
+
+# The DOT_NUM_THREADS specifies the number of dot invocations doxygen is allowed
+# to run in parallel. When set to 0 doxygen will base this on the number of
+# processors available in the system. You can set it explicitly to a value
+# larger than 0 to get control over the balance between CPU load and processing
+# speed.
+# Minimum value: 0, maximum value: 32, default value: 0.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_NUM_THREADS        = 2
+
+# When you want a differently looking font n the dot files that doxygen
+# generates you can specify the font name using DOT_FONTNAME. You need to make
+# sure dot is able to find the font, which can be done by putting it in a
+# standard location or by setting the DOTFONTPATH environment variable or by
+# setting DOT_FONTPATH to the directory containing the font.
+# The default value is: Helvetica.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_FONTNAME           = Helvetica
+
+# The DOT_FONTSIZE tag can be used to set the size (in points) of the font of
+# dot graphs.
+# Minimum value: 4, maximum value: 24, default value: 10.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_FONTSIZE           = 10
+
+# By default doxygen will tell dot to use the default font as specified with
+# DOT_FONTNAME. If you specify a different font using DOT_FONTNAME you can set
+# the path where dot can find it using this tag.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_FONTPATH           =
+
+# If the CLASS_GRAPH tag is set to YES then doxygen will generate a graph for
+# each documented class showing the direct and indirect inheritance relations.
+# Setting this tag to YES will force the CLASS_DIAGRAMS tag to NO.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+CLASS_GRAPH            = YES
+
+# If the COLLABORATION_GRAPH tag is set to YES then doxygen will generate a
+# graph for each documented class showing the direct and indirect implementation
+# dependencies (inheritance, containment, and class references variables) of the
+# class with other documented classes.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+COLLABORATION_GRAPH    = YES
+
+# If the GROUP_GRAPHS tag is set to YES then doxygen will generate a graph for
+# groups, showing the direct groups dependencies.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+GROUP_GRAPHS           = YES
+
+# If the UML_LOOK tag is set to YES doxygen will generate inheritance and
+# collaboration diagrams in a style similar to the OMG's Unified Modeling
+# Language.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+UML_LOOK               = NO
+
+# If the UML_LOOK tag is enabled, the fields and methods are shown inside the
+# class node. If there are many fields or methods and many nodes the graph may
+# become too big to be useful. The UML_LIMIT_NUM_FIELDS threshold limits the
+# number of items for each type to make the size more manageable. Set this to 0
+# for no limit. Note that the threshold may be exceeded by 50% before the limit
+# is enforced. So when you set the threshold to 10, up to 15 fields may appear,
+# but if the number exceeds 15, the total amount of fields shown is limited to
+# 10.
+# Minimum value: 0, maximum value: 100, default value: 10.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+UML_LIMIT_NUM_FIELDS   = 10
+
+# If the TEMPLATE_RELATIONS tag is set to YES then the inheritance and
+# collaboration graphs will show the relations between templates and their
+# instances.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+TEMPLATE_RELATIONS     = NO
+
+# If the INCLUDE_GRAPH, ENABLE_PREPROCESSING and SEARCH_INCLUDES tags are set to
+# YES then doxygen will generate a graph for each documented file showing the
+# direct and indirect include dependencies of the file with other documented
+# files.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+INCLUDE_GRAPH          = YES
+
+# If the INCLUDED_BY_GRAPH, ENABLE_PREPROCESSING and SEARCH_INCLUDES tags are
+# set to YES then doxygen will generate a graph for each documented file showing
+# the direct and indirect include dependencies of the file with other documented
+# files.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+INCLUDED_BY_GRAPH      = YES
+
+# If the CALL_GRAPH tag is set to YES then doxygen will generate a call
+# dependency graph for every global function or class method.
+#
+# Note that enabling this option will significantly increase the time of a run.
+# So in most cases it will be better to enable call graphs for selected
+# functions only using the \callgraph command.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+CALL_GRAPH             = NO
+
+# If the CALLER_GRAPH tag is set to YES then doxygen will generate a caller
+# dependency graph for every global function or class method.
+#
+# Note that enabling this option will significantly increase the time of a run.
+# So in most cases it will be better to enable caller graphs for selected
+# functions only using the \callergraph command.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+CALLER_GRAPH           = NO
+
+# If the GRAPHICAL_HIERARCHY tag is set to YES then doxygen will graphical
+# hierarchy of all classes instead of a textual one.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+GRAPHICAL_HIERARCHY    = YES
+
+# If the DIRECTORY_GRAPH tag is set to YES then doxygen will show the
+# dependencies a directory has on other directories in a graphical way. The
+# dependency relations are determined by the #include relations between the
+# files in the directories.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DIRECTORY_GRAPH        = YES
+
+# The DOT_IMAGE_FORMAT tag can be used to set the image format of the images
+# generated by dot.
+# Note: If you choose svg you need to set HTML_FILE_EXTENSION to xhtml in order
+# to make the SVG files visible in IE 9+ (other browsers do not have this
+# requirement).
+# Possible values are: png, jpg, gif and svg.
+# The default value is: png.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_IMAGE_FORMAT       = png
+
+# If DOT_IMAGE_FORMAT is set to svg, then this option can be set to YES to
+# enable generation of interactive SVG images that allow zooming and panning.
+#
+# Note that this requires a modern browser other than Internet Explorer. Tested
+# and working are Firefox, Chrome, Safari, and Opera.
+# Note: For IE 9+ you need to set HTML_FILE_EXTENSION to xhtml in order to make
+# the SVG files visible. Older versions of IE do not have SVG support.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+INTERACTIVE_SVG        = NO
+
+# The DOT_PATH tag can be used to specify the path where the dot tool can be
+# found. If left blank, it is assumed the dot tool can be found in the path.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_PATH               =
+
+# The DOTFILE_DIRS tag can be used to specify one or more directories that
+# contain dot files that are included in the documentation (see the \dotfile
+# command).
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOTFILE_DIRS           =
+
+# The MSCFILE_DIRS tag can be used to specify one or more directories that
+# contain msc files that are included in the documentation (see the \mscfile
+# command).
+
+MSCFILE_DIRS           =
+
+# The DIAFILE_DIRS tag can be used to specify one or more directories that
+# contain dia files that are included in the documentation (see the \diafile
+# command).
+
+DIAFILE_DIRS           =
+
+# The DOT_GRAPH_MAX_NODES tag can be used to set the maximum number of nodes
+# that will be shown in the graph. If the number of nodes in a graph becomes
+# larger than this value, doxygen will truncate the graph, which is visualized
+# by representing a node as a red box. Note that doxygen if the number of direct
+# children of the root node in a graph is already larger than
+# DOT_GRAPH_MAX_NODES then the graph will not be shown at all. Also note that
+# the size of a graph can be further restricted by MAX_DOT_GRAPH_DEPTH.
+# Minimum value: 0, maximum value: 10000, default value: 50.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_GRAPH_MAX_NODES    = 50
+
+# The MAX_DOT_GRAPH_DEPTH tag can be used to set the maximum depth of the graphs
+# generated by dot. A depth value of 3 means that only nodes reachable from the
+# root by following a path via at most 3 edges will be shown. Nodes that lay
+# further from the root node will be omitted. Note that setting this option to 1
+# or 2 may greatly reduce the computation time needed for large code bases. Also
+# note that the size of a graph can be further restricted by
+# DOT_GRAPH_MAX_NODES. Using a depth of 0 means no depth restriction.
+# Minimum value: 0, maximum value: 1000, default value: 0.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+MAX_DOT_GRAPH_DEPTH    = 0
+
+# Set the DOT_TRANSPARENT tag to YES to generate images with a transparent
+# background. This is disabled by default, because dot on Windows does not seem
+# to support this out of the box.
+#
+# Warning: Depending on the platform used, enabling this option may lead to
+# badly anti-aliased labels on the edges of a graph (i.e. they become hard to
+# read).
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_TRANSPARENT        = YES
+
+# Set the DOT_MULTI_TARGETS tag to YES allow dot to generate multiple output
+# files in one run (i.e. multiple -o and -T options on the command line). This
+# makes dot run faster, but since only newer versions of dot (>1.8.10) support
+# this, this feature is disabled by default.
+# The default value is: NO.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_MULTI_TARGETS      = YES
+
+# If the GENERATE_LEGEND tag is set to YES doxygen will generate a legend page
+# explaining the meaning of the various boxes and arrows in the dot generated
+# graphs.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+GENERATE_LEGEND        = YES
+
+# If the DOT_CLEANUP tag is set to YES doxygen will remove the intermediate dot
+# files that are used to generate the various graphs.
+# The default value is: YES.
+# This tag requires that the tag HAVE_DOT is set to YES.
+
+DOT_CLEANUP            = YES
diff --git a/lib/cshift/Grid_cshift_common.h b/lib/cshift/Grid_cshift_common.h
index 90fc10b7..97e66b2c 100644
--- a/lib/cshift/Grid_cshift_common.h
+++ b/lib/cshift/Grid_cshift_common.h
@@ -160,19 +160,13 @@ template<class vobj> void Copy_plane(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int
 PARALLEL_NESTED_LOOP2
   for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
     for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-      /*
-      int o =n*rhs._grid->_slice_stride[dimension];
-      int ocb=1<<lhs._grid->CheckerBoardFromOindex(o+b);
-      if ( ocb&cbmask ) {
-	lhs._odata[lo+o+b]=rhs._odata[ro+o+b];
-      }
-      */
-
+ 
       int o =n*rhs._grid->_slice_stride[dimension]+b;
       int ocb=1<<lhs._grid->CheckerBoardFromOindex(o);
       if ( ocb&cbmask ) {
 	//lhs._odata[lo+o]=rhs._odata[ro+o];
 	vstream(lhs._odata[lo+o],rhs._odata[ro+o]);
+	
       }
 
     }
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index ae01269f..41ce84ea 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -2,7 +2,7 @@
 /*! @file Grid_vector_types.h
   @brief Defines templated class Grid_simd to deal with inner vector types
 */
-// Time-stamp: <2015-05-26 13:44:54 neo>
+// Time-stamp: <2015-05-26 14:08:13 neo>
 //---------------------------------------------------------------------------
 #ifndef GRID_VECTOR_TYPES
 #define GRID_VECTOR_TYPES
@@ -57,6 +57,8 @@ namespace Grid {
 
   ///////////////////////////////////////////////
 
+
+  // Move to the simd files
 //////////////////////////////////////////////////////////
 // Permute
 // Permute 0 every ABCDEFGH -> BA DC FE HG
@@ -436,7 +438,7 @@ inline void Gpermute(vsimd &y,const vsimd &b,int perm){
   }
 
 
-  // Define available types (now change names to avoid clashing with the rest of the code)
+  // Define available types
 
   typedef Grid_simd< float                 , SIMD_Ftype > vRealF;
   typedef Grid_simd< double                , SIMD_Dtype > vRealD;
diff --git a/m4/ac_prog_doxygen.m4 b/m4/ac_prog_doxygen.m4
new file mode 100644
index 00000000..33db5adc
--- /dev/null
+++ b/m4/ac_prog_doxygen.m4
@@ -0,0 +1,54 @@
+dnl Check for doxygen to create API docs
+dnl
+AC_DEFUN([AC_PROG_DOXYGEN],
+[
+AC_ARG_ENABLE(doxygen,
+AS_HELP_STRING([--enable-doxygen],[enable documentation generation with doxygen (auto)]))
+
+AC_ARG_ENABLE(dot,
+AS_HELP_STRING([--enable-dot],[use 'dot' to generate graphs in doxygen (auto)]))              
+AC_ARG_ENABLE(html-docs,
+AS_HELP_STRING([--enable-html-docs],[enable HTML generation with doxygen (yes)]),
+[],[ enable_html_docs=yes])              
+AC_ARG_ENABLE(latex-docs,
+AS_HELP_STRING([--enable-latex-docs],
+[enable LaTeX documentation generation with doxygen (no)]),[],[enable_latex_docs=no])              
+
+if test "x$enable_doxygen" = xno; then
+ enable_doc=no	
+else 
+ AC_CHECK_PROG(DOXYGEN, doxygen, doxygen)
+ if test x$DOXYGEN = x; then
+  if test "x$enable_doxygen" = xyes; then
+   AC_MSG_ERROR([could not find doxygen])
+  fi
+  enable_doc=no
+ else
+  doxy_ver=`doxygen --version`
+  doxy_major=`expr "$doxy_ver" : '\(@<:@0-9@:>@\)\..*'`
+  doxy_minor=`expr "$doxy_ver" : '@<:@0-9@:>@\.\(@<:@0-9@:>@\).*'`
+  if test $doxy_major -eq "1" -a $doxy_minor -ge "3" ; then
+   enable_doc=yes
+   AC_CHECK_PROG(DOT, dot, dot)
+  else
+   AC_MSG_WARN([doxygen version $doxy_ver too old, doxygen will not be used.])
+   enable_doc=no
+  fi
+ fi
+fi
+
+AM_CONDITIONAL(DOXYGEN_DOC, test x$enable_doc = xyes)
+
+if test x$DOT = x; then
+ if test "x$enable_dot" = xyes; then
+  AC_MSG_ERROR([could not find dot])
+ fi
+ enable_dot=no
+else
+ enable_dot=yes
+fi
+
+AC_SUBST(enable_dot)
+AC_SUBST(enable_html_docs)
+AC_SUBST(enable_latex_docs)
+])
diff --git a/m4/m4-ax_prog_doxygen.m4 b/m4/m4-ax_prog_doxygen.m4
new file mode 100644
index 00000000..fd145991
--- /dev/null
+++ b/m4/m4-ax_prog_doxygen.m4
@@ -0,0 +1,533 @@
+# ===========================================================================
+#      http://www.gnu.org/software/autoconf-archive/ax_prog_doxygen.html
+# ===========================================================================
+#
+# SYNOPSIS
+#
+#   DX_INIT_DOXYGEN(PROJECT-NAME, DOXYFILE-PATH, [OUTPUT-DIR])
+#   DX_DOXYGEN_FEATURE(ON|OFF)
+#   DX_DOT_FEATURE(ON|OFF)
+#   DX_HTML_FEATURE(ON|OFF)
+#   DX_CHM_FEATURE(ON|OFF)
+#   DX_CHI_FEATURE(ON|OFF)
+#   DX_MAN_FEATURE(ON|OFF)
+#   DX_RTF_FEATURE(ON|OFF)
+#   DX_XML_FEATURE(ON|OFF)
+#   DX_PDF_FEATURE(ON|OFF)
+#   DX_PS_FEATURE(ON|OFF)
+#
+# DESCRIPTION
+#
+#   The DX_*_FEATURE macros control the default setting for the given
+#   Doxygen feature. Supported features are 'DOXYGEN' itself, 'DOT' for
+#   generating graphics, 'HTML' for plain HTML, 'CHM' for compressed HTML
+#   help (for MS users), 'CHI' for generating a seperate .chi file by the
+#   .chm file, and 'MAN', 'RTF', 'XML', 'PDF' and 'PS' for the appropriate
+#   output formats. The environment variable DOXYGEN_PAPER_SIZE may be
+#   specified to override the default 'a4wide' paper size.
+#
+#   By default, HTML, PDF and PS documentation is generated as this seems to
+#   be the most popular and portable combination. MAN pages created by
+#   Doxygen are usually problematic, though by picking an appropriate subset
+#   and doing some massaging they might be better than nothing. CHM and RTF
+#   are specific for MS (note that you can't generate both HTML and CHM at
+#   the same time). The XML is rather useless unless you apply specialized
+#   post-processing to it.
+#
+#   The macros mainly control the default state of the feature. The use can
+#   override the default by specifying --enable or --disable. The macros
+#   ensure that contradictory flags are not given (e.g.,
+#   --enable-doxygen-html and --enable-doxygen-chm,
+#   --enable-doxygen-anything with --disable-doxygen, etc.) Finally, each
+#   feature will be automatically disabled (with a warning) if the required
+#   programs are missing.
+#
+#   Once all the feature defaults have been specified, call DX_INIT_DOXYGEN
+#   with the following parameters: a one-word name for the project for use
+#   as a filename base etc., an optional configuration file name (the
+#   default is 'Doxyfile', the same as Doxygen's default), and an optional
+#   output directory name (the default is 'doxygen-doc').
+#
+#   Automake Support
+#
+#   The following is a template aminclude.am file for use with Automake.
+#   Make targets and variables values are controlled by the various
+#   DX_COND_* conditionals set by autoconf.
+#
+#   The provided targets are:
+#
+#     doxygen-doc: Generate all doxygen documentation.
+#
+#     doxygen-run: Run doxygen, which will generate some of the
+#                  documentation (HTML, CHM, CHI, MAN, RTF, XML)
+#                  but will not do the post processing required
+#                  for the rest of it (PS, PDF, and some MAN).
+#
+#     doxygen-man: Rename some doxygen generated man pages.
+#
+#     doxygen-ps:  Generate doxygen PostScript documentation.
+#
+#     doxygen-pdf: Generate doxygen PDF documentation.
+#
+#   Note that by default these are not integrated into the automake targets.
+#   If doxygen is used to generate man pages, you can achieve this
+#   integration by setting man3_MANS to the list of man pages generated and
+#   then adding the dependency:
+#
+#     $(man3_MANS): doxygen-doc
+#
+#   This will cause make to run doxygen and generate all the documentation.
+#
+#   The following variable is intended for use in Makefile.am:
+#
+#     DX_CLEANFILES = everything to clean.
+#
+#   Then add this variable to MOSTLYCLEANFILES.
+#
+#     ----- begin aminclude.am -------------------------------------
+#
+#     ## --------------------------------- ##
+#     ## Format-independent Doxygen rules. ##
+#     ## --------------------------------- ##
+#
+#     if DX_COND_doc
+#
+#     ## ------------------------------- ##
+#     ## Rules specific for HTML output. ##
+#     ## ------------------------------- ##
+#
+#     if DX_COND_html
+#
+#     DX_CLEAN_HTML = @DX_DOCDIR@/html
+#
+#     endif DX_COND_html
+#
+#     ## ------------------------------ ##
+#     ## Rules specific for CHM output. ##
+#     ## ------------------------------ ##
+#
+#     if DX_COND_chm
+#
+#     DX_CLEAN_CHM = @DX_DOCDIR@/chm
+#
+#     if DX_COND_chi
+#
+#     DX_CLEAN_CHI = @DX_DOCDIR@/@PACKAGE@.chi
+#
+#     endif DX_COND_chi
+#
+#     endif DX_COND_chm
+#
+#     ## ------------------------------ ##
+#     ## Rules specific for MAN output. ##
+#     ## ------------------------------ ##
+#
+#     if DX_COND_man
+#
+#     DX_CLEAN_MAN = @DX_DOCDIR@/man
+#
+#     endif DX_COND_man
+#
+#     ## ------------------------------ ##
+#     ## Rules specific for RTF output. ##
+#     ## ------------------------------ ##
+#
+#     if DX_COND_rtf
+#
+#     DX_CLEAN_RTF = @DX_DOCDIR@/rtf
+#
+#     endif DX_COND_rtf
+#
+#     ## ------------------------------ ##
+#     ## Rules specific for XML output. ##
+#     ## ------------------------------ ##
+#
+#     if DX_COND_xml
+#
+#     DX_CLEAN_XML = @DX_DOCDIR@/xml
+#
+#     endif DX_COND_xml
+#
+#     ## ----------------------------- ##
+#     ## Rules specific for PS output. ##
+#     ## ----------------------------- ##
+#
+#     if DX_COND_ps
+#
+#     DX_CLEAN_PS = @DX_DOCDIR@/@PACKAGE@.ps
+#
+#     DX_PS_GOAL = doxygen-ps
+#
+#     doxygen-ps: @DX_DOCDIR@/@PACKAGE@.ps
+#
+#     @DX_DOCDIR@/@PACKAGE@.ps: @DX_DOCDIR@/@PACKAGE@.tag
+#         cd @DX_DOCDIR@/latex; \
+#         rm -f *.aux *.toc *.idx *.ind *.ilg *.log *.out; \
+#         $(DX_LATEX) refman.tex; \
+#         $(MAKEINDEX_PATH) refman.idx; \
+#         $(DX_LATEX) refman.tex; \
+#         countdown=5; \
+#         while $(DX_EGREP) 'Rerun (LaTeX|to get cross-references right)' \
+#                           refman.log > /dev/null 2>&1 \
+#            && test $$countdown -gt 0; do \
+#             $(DX_LATEX) refman.tex; \
+#             countdown=`expr $$countdown - 1`; \
+#         done; \
+#         $(DX_DVIPS) -o ../@PACKAGE@.ps refman.dvi
+#
+#     endif DX_COND_ps
+#
+#     ## ------------------------------ ##
+#     ## Rules specific for PDF output. ##
+#     ## ------------------------------ ##
+#
+#     if DX_COND_pdf
+#
+#     DX_CLEAN_PDF = @DX_DOCDIR@/@PACKAGE@.pdf
+#
+#     DX_PDF_GOAL = doxygen-pdf
+#
+#     doxygen-pdf: @DX_DOCDIR@/@PACKAGE@.pdf
+#
+#     @DX_DOCDIR@/@PACKAGE@.pdf: @DX_DOCDIR@/@PACKAGE@.tag
+#         cd @DX_DOCDIR@/latex; \
+#         rm -f *.aux *.toc *.idx *.ind *.ilg *.log *.out; \
+#         $(DX_PDFLATEX) refman.tex; \
+#         $(DX_MAKEINDEX) refman.idx; \
+#         $(DX_PDFLATEX) refman.tex; \
+#         countdown=5; \
+#         while $(DX_EGREP) 'Rerun (LaTeX|to get cross-references right)' \
+#                           refman.log > /dev/null 2>&1 \
+#            && test $$countdown -gt 0; do \
+#             $(DX_PDFLATEX) refman.tex; \
+#             countdown=`expr $$countdown - 1`; \
+#         done; \
+#         mv refman.pdf ../@PACKAGE@.pdf
+#
+#     endif DX_COND_pdf
+#
+#     ## ------------------------------------------------- ##
+#     ## Rules specific for LaTeX (shared for PS and PDF). ##
+#     ## ------------------------------------------------- ##
+#
+#     if DX_COND_latex
+#
+#     DX_CLEAN_LATEX = @DX_DOCDIR@/latex
+#
+#     endif DX_COND_latex
+#
+#     .PHONY: doxygen-run doxygen-doc $(DX_PS_GOAL) $(DX_PDF_GOAL)
+#
+#     .INTERMEDIATE: doxygen-run $(DX_PS_GOAL) $(DX_PDF_GOAL)
+#
+#     doxygen-run: @DX_DOCDIR@/@PACKAGE@.tag
+#
+#     doxygen-doc: doxygen-run $(DX_PS_GOAL) $(DX_PDF_GOAL)
+#
+#     @DX_DOCDIR@/@PACKAGE@.tag: $(DX_CONFIG) $(pkginclude_HEADERS)
+#         rm -rf @DX_DOCDIR@
+#         $(DX_ENV) $(DX_DOXYGEN) $(srcdir)/$(DX_CONFIG)
+#         echo Timestamp >$@
+#
+#     DX_CLEANFILES = \
+#         @DX_DOCDIR@/@PACKAGE@.tag \
+#         -r \
+#         $(DX_CLEAN_HTML) \
+#         $(DX_CLEAN_CHM) \
+#         $(DX_CLEAN_CHI) \
+#         $(DX_CLEAN_MAN) \
+#         $(DX_CLEAN_RTF) \
+#         $(DX_CLEAN_XML) \
+#         $(DX_CLEAN_PS) \
+#         $(DX_CLEAN_PDF) \
+#         $(DX_CLEAN_LATEX)
+#
+#     endif DX_COND_doc
+#
+#     ----- end aminclude.am ---------------------------------------
+#
+# LICENSE
+#
+#   Copyright (c) 2009 Oren Ben-Kiki <oren@ben-kiki.org>
+#
+#   Copying and distribution of this file, with or without modification, are
+#   permitted in any medium without royalty provided the copyright notice
+#   and this notice are preserved. This file is offered as-is, without any
+#   warranty.
+
+#serial 13
+
+## ----------##
+## Defaults. ##
+## ----------##
+
+DX_ENV=""
+AC_DEFUN([DX_FEATURE_doc],  ON)
+AC_DEFUN([DX_FEATURE_dot],  OFF)
+AC_DEFUN([DX_FEATURE_man],  OFF)
+AC_DEFUN([DX_FEATURE_html], ON)
+AC_DEFUN([DX_FEATURE_chm],  OFF)
+AC_DEFUN([DX_FEATURE_chi],  OFF)
+AC_DEFUN([DX_FEATURE_rtf],  OFF)
+AC_DEFUN([DX_FEATURE_xml],  OFF)
+AC_DEFUN([DX_FEATURE_pdf],  ON)
+AC_DEFUN([DX_FEATURE_ps],   ON)
+
+## --------------- ##
+## Private macros. ##
+## --------------- ##
+
+# DX_ENV_APPEND(VARIABLE, VALUE)
+# ------------------------------
+# Append VARIABLE="VALUE" to DX_ENV for invoking doxygen.
+AC_DEFUN([DX_ENV_APPEND], [AC_SUBST([DX_ENV], ["$DX_ENV $1='$2'"])])
+
+# DX_DIRNAME_EXPR
+# ---------------
+# Expand into a shell expression prints the directory part of a path.
+AC_DEFUN([DX_DIRNAME_EXPR],
+         [[expr ".$1" : '\(\.\)[^/]*$' \| "x$1" : 'x\(.*\)/[^/]*$']])
+
+# DX_IF_FEATURE(FEATURE, IF-ON, IF-OFF)
+# -------------------------------------
+# Expands according to the M4 (static) status of the feature.
+AC_DEFUN([DX_IF_FEATURE], [ifelse(DX_FEATURE_$1, ON, [$2], [$3])])
+
+# DX_REQUIRE_PROG(VARIABLE, PROGRAM)
+# ----------------------------------
+# Require the specified program to be found for the DX_CURRENT_FEATURE to work.
+AC_DEFUN([DX_REQUIRE_PROG], [
+AC_PATH_TOOL([$1], [$2])
+if test "$DX_FLAG_[]DX_CURRENT_FEATURE$$1" = 1; then
+    AC_MSG_WARN([$2 not found - will not DX_CURRENT_DESCRIPTION])
+    AC_SUBST(DX_FLAG_[]DX_CURRENT_FEATURE, 0)
+fi
+])
+
+# DX_TEST_FEATURE(FEATURE)
+# ------------------------
+# Expand to a shell expression testing whether the feature is active.
+AC_DEFUN([DX_TEST_FEATURE], [test "$DX_FLAG_$1" = 1])
+
+# DX_CHECK_DEPEND(REQUIRED_FEATURE, REQUIRED_STATE)
+# -------------------------------------------------
+# Verify that a required features has the right state before trying to turn on
+# the DX_CURRENT_FEATURE.
+AC_DEFUN([DX_CHECK_DEPEND], [
+test "$DX_FLAG_$1" = "$2" \
+|| AC_MSG_ERROR([doxygen-DX_CURRENT_FEATURE ifelse([$2], 1,
+                            requires, contradicts) doxygen-DX_CURRENT_FEATURE])
+])
+
+# DX_CLEAR_DEPEND(FEATURE, REQUIRED_FEATURE, REQUIRED_STATE)
+# ----------------------------------------------------------
+# Turn off the DX_CURRENT_FEATURE if the required feature is off.
+AC_DEFUN([DX_CLEAR_DEPEND], [
+test "$DX_FLAG_$1" = "$2" || AC_SUBST(DX_FLAG_[]DX_CURRENT_FEATURE, 0)
+])
+
+# DX_FEATURE_ARG(FEATURE, DESCRIPTION,
+#                CHECK_DEPEND, CLEAR_DEPEND,
+#                REQUIRE, DO-IF-ON, DO-IF-OFF)
+# --------------------------------------------
+# Parse the command-line option controlling a feature. CHECK_DEPEND is called
+# if the user explicitly turns the feature on (and invokes DX_CHECK_DEPEND),
+# otherwise CLEAR_DEPEND is called to turn off the default state if a required
+# feature is disabled (using DX_CLEAR_DEPEND). REQUIRE performs additional
+# requirement tests (DX_REQUIRE_PROG). Finally, an automake flag is set and
+# DO-IF-ON or DO-IF-OFF are called according to the final state of the feature.
+AC_DEFUN([DX_ARG_ABLE], [
+    AC_DEFUN([DX_CURRENT_FEATURE], [$1])
+    AC_DEFUN([DX_CURRENT_DESCRIPTION], [$2])
+    AC_ARG_ENABLE(doxygen-$1,
+                  [AS_HELP_STRING(DX_IF_FEATURE([$1], [--disable-doxygen-$1],
+                                                      [--enable-doxygen-$1]),
+                                  DX_IF_FEATURE([$1], [don't $2], [$2]))],
+                  [
+case "$enableval" in
+#(
+y|Y|yes|Yes|YES)
+    AC_SUBST([DX_FLAG_$1], 1)
+    $3
+;; #(
+n|N|no|No|NO)
+    AC_SUBST([DX_FLAG_$1], 0)
+;; #(
+*)
+    AC_MSG_ERROR([invalid value '$enableval' given to doxygen-$1])
+;;
+esac
+], [
+AC_SUBST([DX_FLAG_$1], [DX_IF_FEATURE([$1], 1, 0)])
+$4
+])
+if DX_TEST_FEATURE([$1]); then
+    $5
+    :
+fi
+AM_CONDITIONAL(DX_COND_$1, DX_TEST_FEATURE([$1]))
+if DX_TEST_FEATURE([$1]); then
+    $6
+    :
+else
+    $7
+    :
+fi
+])
+
+## -------------- ##
+## Public macros. ##
+## -------------- ##
+
+# DX_XXX_FEATURE(DEFAULT_STATE)
+# -----------------------------
+AC_DEFUN([DX_DOXYGEN_FEATURE], [AC_DEFUN([DX_FEATURE_doc],  [$1])])
+AC_DEFUN([DX_DOT_FEATURE],     [AC_DEFUN([DX_FEATURE_dot], [$1])])
+AC_DEFUN([DX_MAN_FEATURE],     [AC_DEFUN([DX_FEATURE_man],  [$1])])
+AC_DEFUN([DX_HTML_FEATURE],    [AC_DEFUN([DX_FEATURE_html], [$1])])
+AC_DEFUN([DX_CHM_FEATURE],     [AC_DEFUN([DX_FEATURE_chm],  [$1])])
+AC_DEFUN([DX_CHI_FEATURE],     [AC_DEFUN([DX_FEATURE_chi],  [$1])])
+AC_DEFUN([DX_RTF_FEATURE],     [AC_DEFUN([DX_FEATURE_rtf],  [$1])])
+AC_DEFUN([DX_XML_FEATURE],     [AC_DEFUN([DX_FEATURE_xml],  [$1])])
+AC_DEFUN([DX_XML_FEATURE],     [AC_DEFUN([DX_FEATURE_xml],  [$1])])
+AC_DEFUN([DX_PDF_FEATURE],     [AC_DEFUN([DX_FEATURE_pdf],  [$1])])
+AC_DEFUN([DX_PS_FEATURE],      [AC_DEFUN([DX_FEATURE_ps],   [$1])])
+
+# DX_INIT_DOXYGEN(PROJECT, [CONFIG-FILE], [OUTPUT-DOC-DIR])
+# ---------------------------------------------------------
+# PROJECT also serves as the base name for the documentation files.
+# The default CONFIG-FILE is "Doxyfile" and OUTPUT-DOC-DIR is "doxygen-doc".
+AC_DEFUN([DX_INIT_DOXYGEN], [
+
+# Files:
+AC_SUBST([DX_PROJECT], [$1])
+AC_SUBST([DX_CONFIG], [ifelse([$2], [], Doxyfile, [$2])])
+AC_SUBST([DX_DOCDIR], [ifelse([$3], [], doxygen-doc, [$3])])
+
+# Environment variables used inside doxygen.cfg:
+DX_ENV_APPEND(SRCDIR, $srcdir)
+DX_ENV_APPEND(PROJECT, $DX_PROJECT)
+DX_ENV_APPEND(DOCDIR, $DX_DOCDIR)
+DX_ENV_APPEND(VERSION, $PACKAGE_VERSION)
+
+# Doxygen itself:
+DX_ARG_ABLE(doc, [generate any doxygen documentation],
+            [],
+            [],
+            [DX_REQUIRE_PROG([DX_DOXYGEN], doxygen)
+             DX_REQUIRE_PROG([DX_PERL], perl)],
+            [DX_ENV_APPEND(PERL_PATH, $DX_PERL)])
+
+# Dot for graphics:
+DX_ARG_ABLE(dot, [generate graphics for doxygen documentation],
+            [DX_CHECK_DEPEND(doc, 1)],
+            [DX_CLEAR_DEPEND(doc, 1)],
+            [DX_REQUIRE_PROG([DX_DOT], dot)],
+            [DX_ENV_APPEND(HAVE_DOT, YES)
+             DX_ENV_APPEND(DOT_PATH, [`DX_DIRNAME_EXPR($DX_DOT)`])],
+            [DX_ENV_APPEND(HAVE_DOT, NO)])
+
+# Man pages generation:
+DX_ARG_ABLE(man, [generate doxygen manual pages],
+            [DX_CHECK_DEPEND(doc, 1)],
+            [DX_CLEAR_DEPEND(doc, 1)],
+            [],
+            [DX_ENV_APPEND(GENERATE_MAN, YES)],
+            [DX_ENV_APPEND(GENERATE_MAN, NO)])
+
+# RTF file generation:
+DX_ARG_ABLE(rtf, [generate doxygen RTF documentation],
+            [DX_CHECK_DEPEND(doc, 1)],
+            [DX_CLEAR_DEPEND(doc, 1)],
+            [],
+            [DX_ENV_APPEND(GENERATE_RTF, YES)],
+            [DX_ENV_APPEND(GENERATE_RTF, NO)])
+
+# XML file generation:
+DX_ARG_ABLE(xml, [generate doxygen XML documentation],
+            [DX_CHECK_DEPEND(doc, 1)],
+            [DX_CLEAR_DEPEND(doc, 1)],
+            [],
+            [DX_ENV_APPEND(GENERATE_XML, YES)],
+            [DX_ENV_APPEND(GENERATE_XML, NO)])
+
+# (Compressed) HTML help generation:
+DX_ARG_ABLE(chm, [generate doxygen compressed HTML help documentation],
+            [DX_CHECK_DEPEND(doc, 1)],
+            [DX_CLEAR_DEPEND(doc, 1)],
+            [DX_REQUIRE_PROG([DX_HHC], hhc)],
+            [DX_ENV_APPEND(HHC_PATH, $DX_HHC)
+             DX_ENV_APPEND(GENERATE_HTML, YES)
+             DX_ENV_APPEND(GENERATE_HTMLHELP, YES)],
+            [DX_ENV_APPEND(GENERATE_HTMLHELP, NO)])
+
+# Seperate CHI file generation.
+DX_ARG_ABLE(chi, [generate doxygen seperate compressed HTML help index file],
+            [DX_CHECK_DEPEND(chm, 1)],
+            [DX_CLEAR_DEPEND(chm, 1)],
+            [],
+            [DX_ENV_APPEND(GENERATE_CHI, YES)],
+            [DX_ENV_APPEND(GENERATE_CHI, NO)])
+
+# Plain HTML pages generation:
+DX_ARG_ABLE(html, [generate doxygen plain HTML documentation],
+            [DX_CHECK_DEPEND(doc, 1) DX_CHECK_DEPEND(chm, 0)],
+            [DX_CLEAR_DEPEND(doc, 1) DX_CLEAR_DEPEND(chm, 0)],
+            [],
+            [DX_ENV_APPEND(GENERATE_HTML, YES)],
+            [DX_TEST_FEATURE(chm) || DX_ENV_APPEND(GENERATE_HTML, NO)])
+
+# PostScript file generation:
+DX_ARG_ABLE(ps, [generate doxygen PostScript documentation],
+            [DX_CHECK_DEPEND(doc, 1)],
+            [DX_CLEAR_DEPEND(doc, 1)],
+            [DX_REQUIRE_PROG([DX_LATEX], latex)
+             DX_REQUIRE_PROG([DX_MAKEINDEX], makeindex)
+             DX_REQUIRE_PROG([DX_DVIPS], dvips)
+             DX_REQUIRE_PROG([DX_EGREP], egrep)])
+
+# PDF file generation:
+DX_ARG_ABLE(pdf, [generate doxygen PDF documentation],
+            [DX_CHECK_DEPEND(doc, 1)],
+            [DX_CLEAR_DEPEND(doc, 1)],
+            [DX_REQUIRE_PROG([DX_PDFLATEX], pdflatex)
+             DX_REQUIRE_PROG([DX_MAKEINDEX], makeindex)
+             DX_REQUIRE_PROG([DX_EGREP], egrep)])
+
+# LaTeX generation for PS and/or PDF:
+AM_CONDITIONAL(DX_COND_latex, DX_TEST_FEATURE(ps) || DX_TEST_FEATURE(pdf))
+if DX_TEST_FEATURE(ps) || DX_TEST_FEATURE(pdf); then
+    DX_ENV_APPEND(GENERATE_LATEX, YES)
+else
+    DX_ENV_APPEND(GENERATE_LATEX, NO)
+fi
+
+# Paper size for PS and/or PDF:
+AC_ARG_VAR(DOXYGEN_PAPER_SIZE,
+           [a4wide (default), a4, letter, legal or executive])
+case "$DOXYGEN_PAPER_SIZE" in
+#(
+"")
+    AC_SUBST(DOXYGEN_PAPER_SIZE, "")
+;; #(
+a4wide|a4|letter|legal|executive)
+    DX_ENV_APPEND(PAPER_SIZE, $DOXYGEN_PAPER_SIZE)
+;; #(
+*)
+    AC_MSG_ERROR([unknown DOXYGEN_PAPER_SIZE='$DOXYGEN_PAPER_SIZE'])
+;;
+esac
+
+#For debugging:
+#echo DX_FLAG_doc=$DX_FLAG_doc
+#echo DX_FLAG_dot=$DX_FLAG_dot
+#echo DX_FLAG_man=$DX_FLAG_man
+#echo DX_FLAG_html=$DX_FLAG_html
+#echo DX_FLAG_chm=$DX_FLAG_chm
+#echo DX_FLAG_chi=$DX_FLAG_chi
+#echo DX_FLAG_rtf=$DX_FLAG_rtf
+#echo DX_FLAG_xml=$DX_FLAG_xml
+#echo DX_FLAG_pdf=$DX_FLAG_pdf
+#echo DX_FLAG_ps=$DX_FLAG_ps
+#echo DX_ENV=$DX_ENV
+])

From 2ccbff6c6cd02c822be23bf1d46e37320312455e Mon Sep 17 00:00:00 2001
From: Guido Cossu <cossu@iroiro-pc.kek.jp>
Date: Wed, 27 May 2015 10:49:33 +0900
Subject: [PATCH 241/429] Corrected AVX regression error. Tested.

---
 INSTALL                      |  301 +-
 Makefile.in                  |  384 +--
 aclocal.m4                   |  623 ++--
 config.guess                 |    2 +-
 config.sub                   |    2 +-
 configure                    | 5168 +++++++++++++++++++++-------------
 depcomp                      |  792 +-----
 install-sh                   |  502 +---
 lib/Grid_config.h            |    7 +-
 lib/Grid_config.h.in         |    3 -
 lib/simd/Grid_avx.h          |    9 +-
 lib/simd/Grid_vector_types.h |    4 +-
 missing                      |  216 +-
 reconfigure_script           |    2 +-
 14 files changed, 3634 insertions(+), 4381 deletions(-)
 mode change 100644 => 120000 INSTALL
 mode change 100755 => 120000 depcomp
 mode change 100755 => 120000 install-sh
 mode change 100755 => 120000 missing

diff --git a/INSTALL b/INSTALL
deleted file mode 100644
index 9996b52c..00000000
--- a/INSTALL
+++ /dev/null
@@ -1,300 +0,0 @@
-Installation Instructions
-*************************
-
-Copyright (C) 1994-1996, 1999-2002, 2004-2013 Free Software Foundation,
-Inc.
-
-   Copying and distribution of this file, with or without modification,
-are permitted in any medium without royalty provided the copyright
-notice and this notice are preserved.  This file is offered as-is,
-without warranty of any kind.
-
-Basic Installation
-==================
-
-   Briefly, the shell command `./configure && make && make install'
-should configure, build, and install this package.  The following
-more-detailed instructions are generic; see the `README' file for
-instructions specific to this package.  Some packages provide this
-`INSTALL' file but do not implement all of the features documented
-below.  The lack of an optional feature in a given package is not
-necessarily a bug.  More recommendations for GNU packages can be found
-in *note Makefile Conventions: (standards)Makefile Conventions.
-
-   The `configure' shell script attempts to guess correct values for
-various system-dependent variables used during compilation.  It uses
-those values to create a `Makefile' in each directory of the package.
-It may also create one or more `.h' files containing system-dependent
-definitions.  Finally, it creates a shell script `config.status' that
-you can run in the future to recreate the current configuration, and a
-file `config.log' containing compiler output (useful mainly for
-debugging `configure').
-
-   It can also use an optional file (typically called `config.cache'
-and enabled with `--cache-file=config.cache' or simply `-C') that saves
-the results of its tests to speed up reconfiguring.  Caching is
-disabled by default to prevent problems with accidental use of stale
-cache files.
-
-   If you need to do unusual things to compile the package, please try
-to figure out how `configure' could check whether to do them, and mail
-diffs or instructions to the address given in the `README' so they can
-be considered for the next release.  If you are using the cache, and at
-some point `config.cache' contains results you don't want to keep, you
-may remove or edit it.
-
-   The file `configure.ac' (or `configure.in') is used to create
-`configure' by a program called `autoconf'.  You need `configure.ac' if
-you want to change it or regenerate `configure' using a newer version
-of `autoconf'.
-
-   The simplest way to compile this package is:
-
-  1. `cd' to the directory containing the package's source code and type
-     `./configure' to configure the package for your system.
-
-     Running `configure' might take a while.  While running, it prints
-     some messages telling which features it is checking for.
-
-  2. Type `make' to compile the package.
-
-  3. Optionally, type `make check' to run any self-tests that come with
-     the package, generally using the just-built uninstalled binaries.
-
-  4. Type `make install' to install the programs and any data files and
-     documentation.  When installing into a prefix owned by root, it is
-     recommended that the package be configured and built as a regular
-     user, and only the `make install' phase executed with root
-     privileges.
-
-  5. Optionally, type `make installcheck' to repeat any self-tests, but
-     this time using the binaries in their final installed location.
-     This target does not install anything.  Running this target as a
-     regular user, particularly if the prior `make install' required
-     root privileges, verifies that the installation completed
-     correctly.
-
-  6. You can remove the program binaries and object files from the
-     source code directory by typing `make clean'.  To also remove the
-     files that `configure' created (so you can compile the package for
-     a different kind of computer), type `make distclean'.  There is
-     also a `make maintainer-clean' target, but that is intended mainly
-     for the package's developers.  If you use it, you may have to get
-     all sorts of other programs in order to regenerate files that came
-     with the distribution.
-
-  7. Often, you can also type `make uninstall' to remove the installed
-     files again.  In practice, not all packages have tested that
-     uninstallation works correctly, even though it is required by the
-     GNU Coding Standards.
-
-  8. Some packages, particularly those that use Automake, provide `make
-     distcheck', which can by used by developers to test that all other
-     targets like `make install' and `make uninstall' work correctly.
-     This target is generally not run by end users.
-
-Compilers and Options
-=====================
-
-   Some systems require unusual options for compilation or linking that
-the `configure' script does not know about.  Run `./configure --help'
-for details on some of the pertinent environment variables.
-
-   You can give `configure' initial values for configuration parameters
-by setting variables in the command line or in the environment.  Here
-is are examples:
-
-     ./configure CXX=clang++ CXXFLAGS="-std=c++11 -O3 -mavx" --enable-simd=AVX1
-
-     ./configure CXX=clang++ CXXFLAGS="-std=c++11 -O3 -mavx2" --enable-simd=AVX2
-
-     ./configure CXX=icpc CXXFLAGS="-std=c++11 -O3 -mmic" --enable-simd=AVX512 --host=none
-
-   *Note Defining Variables::, for more details.
-
-Compiling For Multiple Architectures
-====================================
-
-   You can compile the package for more than one kind of computer at the
-same time, by placing the object files for each architecture in their
-own directory.  To do this, you can use GNU `make'.  `cd' to the
-directory where you want the object files and executables to go and run
-the `configure' script.  `configure' automatically checks for the
-source code in the directory that `configure' is in and in `..'.  This
-is known as a "VPATH" build.
-
-   With a non-GNU `make', it is safer to compile the package for one
-architecture at a time in the source code directory.  After you have
-installed the package for one architecture, use `make distclean' before
-reconfiguring for another architecture.
-
-
-Installation Names
-==================
-
-   By default, `make install' installs the package's commands under
-`/usr/local/bin', include files under `/usr/local/include', etc.  You
-can specify an installation prefix other than `/usr/local' by giving
-`configure' the option `--prefix=PREFIX', where PREFIX must be an
-absolute file name.
-
-   You can specify separate installation prefixes for
-architecture-specific files and architecture-independent files.  If you
-pass the option `--exec-prefix=PREFIX' to `configure', the package uses
-PREFIX as the prefix for installing programs and libraries.
-Documentation and other data files still use the regular prefix.
-
-   In addition, if you use an unusual directory layout you can give
-options like `--bindir=DIR' to specify different values for particular
-kinds of files.  Run `configure --help' for a list of the directories
-you can set and what kinds of files go in them.  In general, the
-default for these options is expressed in terms of `${prefix}', so that
-specifying just `--prefix' will affect all of the other directory
-specifications that were not explicitly provided.
-
-   The most portable way to affect installation locations is to pass the
-correct locations to `configure'; however, many packages provide one or
-both of the following shortcuts of passing variable assignments to the
-`make install' command line to change installation locations without
-having to reconfigure or recompile.
-
-   The first method involves providing an override variable for each
-affected directory.  For example, `make install
-prefix=/alternate/directory' will choose an alternate location for all
-directory configuration variables that were expressed in terms of
-`${prefix}'.  Any directories that were specified during `configure',
-but not in terms of `${prefix}', must each be overridden at install
-time for the entire installation to be relocated.  The approach of
-makefile variable overrides for each directory variable is required by
-the GNU Coding Standards, and ideally causes no recompilation.
-However, some platforms have known limitations with the semantics of
-shared libraries that end up requiring recompilation when using this
-method, particularly noticeable in packages that use GNU Libtool.
-
-   The second method involves providing the `DESTDIR' variable.  For
-example, `make install DESTDIR=/alternate/directory' will prepend
-`/alternate/directory' before all installation names.  The approach of
-`DESTDIR' overrides is not required by the GNU Coding Standards, and
-does not work on platforms that have drive letters.  On the other hand,
-it does better at avoiding recompilation issues, and works well even
-when some directory options were not specified in terms of `${prefix}'
-at `configure' time.
-
-Optional Features
-=================
-
-   If the package supports it, you can cause programs to be installed
-with an extra prefix or suffix on their names by giving `configure' the
-option `--program-prefix=PREFIX' or `--program-suffix=SUFFIX'.
-
-   Some packages pay attention to `--enable-FEATURE' options to
-`configure', where FEATURE indicates an optional part of the package.
-They may also pay attention to `--with-PACKAGE' options, where PACKAGE
-is something like `gnu-as' or `x' (for the X Window System).  The
-`README' should mention any `--enable-' and `--with-' options that the
-package recognizes.
-
-   For packages that use the X Window System, `configure' can usually
-find the X include and library files automatically, but if it doesn't,
-you can use the `configure' options `--x-includes=DIR' and
-`--x-libraries=DIR' to specify their locations.
-
-   Some packages offer the ability to configure how verbose the
-execution of `make' will be.  For these packages, running `./configure
---enable-silent-rules' sets the default to minimal output, which can be
-overridden with `make V=1'; while running `./configure
---disable-silent-rules' sets the default to verbose, which can be
-overridden with `make V=0'.
-
-Particular systems
-==================
-
-Cross compiling for native execution on XeonPhi requires 
-
-   --build=none
-
-Sharing Defaults
-================
-
-   If you want to set default values for `configure' scripts to share,
-you can create a site shell script called `config.site' that gives
-default values for variables like `CC', `cache_file', and `prefix'.
-`configure' looks for `PREFIX/share/config.site' if it exists, then
-`PREFIX/etc/config.site' if it exists.  Or, you can set the
-`CONFIG_SITE' environment variable to the location of the site script.
-A warning: not all `configure' scripts look for a site script.
-
-Defining Variables
-==================
-
-   Variables not defined in a site shell script can be set in the
-environment passed to `configure'.  However, some packages may run
-configure again during the build, and the customized values of these
-variables may be lost.  In order to avoid this problem, you should set
-them in the `configure' command line, using `VAR=value'.  For example:
-
-     ./configure CC=/usr/local2/bin/gcc
-
-causes the specified `gcc' to be used as the C compiler (unless it is
-overridden in the site shell script).
-
-Unfortunately, this technique does not work for `CONFIG_SHELL' due to
-an Autoconf limitation.  Until the limitation is lifted, you can use
-this workaround:
-
-     CONFIG_SHELL=/bin/bash ./configure CONFIG_SHELL=/bin/bash
-
-`configure' Invocation
-======================
-
-   `configure' recognizes the following options to control how it
-operates.
-
-`--help'
-`-h'
-     Print a summary of all of the options to `configure', and exit.
-
-`--help=short'
-`--help=recursive'
-     Print a summary of the options unique to this package's
-     `configure', and exit.  The `short' variant lists options used
-     only in the top level, while the `recursive' variant lists options
-     also present in any nested packages.
-
-`--version'
-`-V'
-     Print the version of Autoconf used to generate the `configure'
-     script, and exit.
-
-`--cache-file=FILE'
-     Enable the cache: use and save the results of the tests in FILE,
-     traditionally `config.cache'.  FILE defaults to `/dev/null' to
-     disable caching.
-
-`--config-cache'
-`-C'
-     Alias for `--cache-file=config.cache'.
-
-`--quiet'
-`--silent'
-`-q'
-     Do not print messages saying which checks are being made.  To
-     suppress all normal output, redirect it to `/dev/null' (any error
-     messages will still be shown).
-
-`--srcdir=DIR'
-     Look for the package's source code in directory DIR.  Usually
-     `configure' can determine that directory automatically.
-
-`--prefix=DIR'
-     Use DIR as the installation prefix.  *note Installation Names::
-     for more details, including other options available for fine-tuning
-     the installation locations.
-
-`--no-create'
-`-n'
-     Run the configure checks, but stop before creating any output
-     files.
-
-`configure' also accepts some other, not widely useful, options.  Run
-`configure --help' for more details.
diff --git a/INSTALL b/INSTALL
new file mode 120000
index 00000000..cbd1c80d
--- /dev/null
+++ b/INSTALL
@@ -0,0 +1 @@
+/usr/share/automake-1.11/INSTALL
\ No newline at end of file
diff --git a/Makefile.in b/Makefile.in
index cc3284b3..cfa68e94 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -1,8 +1,9 @@
-# Makefile.in generated by automake 1.14.1 from Makefile.am.
+# Makefile.in generated by automake 1.11.1 from Makefile.am.
 # @configure_input@
 
-# Copyright (C) 1994-2013 Free Software Foundation, Inc.
-
+# Copyright (C) 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002,
+# 2003, 2004, 2005, 2006, 2007, 2008, 2009  Free Software Foundation,
+# Inc.
 # This Makefile.in is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
@@ -14,51 +15,6 @@
 
 @SET_MAKE@
 VPATH = @srcdir@
-am__is_gnu_make = test -n '$(MAKEFILE_LIST)' && test -n '$(MAKELEVEL)'
-am__make_running_with_option = \
-  case $${target_option-} in \
-      ?) ;; \
-      *) echo "am__make_running_with_option: internal error: invalid" \
-              "target option '$${target_option-}' specified" >&2; \
-         exit 1;; \
-  esac; \
-  has_opt=no; \
-  sane_makeflags=$$MAKEFLAGS; \
-  if $(am__is_gnu_make); then \
-    sane_makeflags=$$MFLAGS; \
-  else \
-    case $$MAKEFLAGS in \
-      *\\[\ \	]*) \
-        bs=\\; \
-        sane_makeflags=`printf '%s\n' "$$MAKEFLAGS" \
-          | sed "s/$$bs$$bs[$$bs $$bs	]*//g"`;; \
-    esac; \
-  fi; \
-  skip_next=no; \
-  strip_trailopt () \
-  { \
-    flg=`printf '%s\n' "$$flg" | sed "s/$$1.*$$//"`; \
-  }; \
-  for flg in $$sane_makeflags; do \
-    test $$skip_next = yes && { skip_next=no; continue; }; \
-    case $$flg in \
-      *=*|--*) continue;; \
-        -*I) strip_trailopt 'I'; skip_next=yes;; \
-      -*I?*) strip_trailopt 'I';; \
-        -*O) strip_trailopt 'O'; skip_next=yes;; \
-      -*O?*) strip_trailopt 'O';; \
-        -*l) strip_trailopt 'l'; skip_next=yes;; \
-      -*l?*) strip_trailopt 'l';; \
-      -[dEDm]) skip_next=yes;; \
-      -[JT]) skip_next=yes;; \
-    esac; \
-    case $$flg in \
-      *$$target_option*) has_opt=yes; break;; \
-    esac; \
-  done; \
-  test $$has_opt = yes
-am__make_dryrun = (target_option=n; $(am__make_running_with_option))
-am__make_keepgoing = (target_option=k; $(am__make_running_with_option))
 pkgdatadir = $(datadir)/@PACKAGE@
 pkgincludedir = $(includedir)/@PACKAGE@
 pkglibdir = $(libdir)/@PACKAGE@
@@ -79,11 +35,11 @@ build_triplet = @build@
 host_triplet = @host@
 target_triplet = @target@
 subdir = .
-DIST_COMMON = INSTALL NEWS README AUTHORS ChangeLog \
-	$(srcdir)/Makefile.in $(srcdir)/Makefile.am \
-	$(top_srcdir)/configure $(am__configure_deps) \
-	$(top_srcdir)/docs/doxy.cfg.in COPYING TODO compile \
-	config.guess config.sub depcomp install-sh missing
+DIST_COMMON = README $(am__configure_deps) $(srcdir)/Makefile.am \
+	$(srcdir)/Makefile.in $(top_srcdir)/configure \
+	$(top_srcdir)/docs/doxy.cfg.in AUTHORS COPYING ChangeLog \
+	INSTALL NEWS TODO compile config.guess config.sub depcomp \
+	install-sh missing
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/m4/ac_prog_doxygen.m4 \
 	$(top_srcdir)/m4/ax_cxx_compile_stdcxx_11.m4 \
@@ -96,72 +52,36 @@ mkinstalldirs = $(install_sh) -d
 CONFIG_HEADER = $(top_builddir)/lib/Grid_config.h
 CONFIG_CLEAN_FILES = docs/doxy.cfg
 CONFIG_CLEAN_VPATH_FILES =
-AM_V_P = $(am__v_P_@AM_V@)
-am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
-am__v_P_0 = false
-am__v_P_1 = :
-AM_V_GEN = $(am__v_GEN_@AM_V@)
-am__v_GEN_ = $(am__v_GEN_@AM_DEFAULT_V@)
-am__v_GEN_0 = @echo "  GEN     " $@;
-am__v_GEN_1 = 
-AM_V_at = $(am__v_at_@AM_V@)
-am__v_at_ = $(am__v_at_@AM_DEFAULT_V@)
+AM_V_GEN = $(am__v_GEN_$(V))
+am__v_GEN_ = $(am__v_GEN_$(AM_DEFAULT_VERBOSITY))
+am__v_GEN_0 = @echo "  GEN   " $@;
+AM_V_at = $(am__v_at_$(V))
+am__v_at_ = $(am__v_at_$(AM_DEFAULT_VERBOSITY))
 am__v_at_0 = @
-am__v_at_1 = 
 SOURCES =
 DIST_SOURCES =
-RECURSIVE_TARGETS = all-recursive check-recursive cscopelist-recursive \
-	ctags-recursive dvi-recursive html-recursive info-recursive \
-	install-data-recursive install-dvi-recursive \
-	install-exec-recursive install-html-recursive \
-	install-info-recursive install-pdf-recursive \
-	install-ps-recursive install-recursive installcheck-recursive \
-	installdirs-recursive pdf-recursive ps-recursive \
-	tags-recursive uninstall-recursive
-am__can_run_installinfo = \
-  case $$AM_UPDATE_INFO_DIR in \
-    n|no|NO) false;; \
-    *) (install-info --version) >/dev/null 2>&1;; \
-  esac
+RECURSIVE_TARGETS = all-recursive check-recursive dvi-recursive \
+	html-recursive info-recursive install-data-recursive \
+	install-dvi-recursive install-exec-recursive \
+	install-html-recursive install-info-recursive \
+	install-pdf-recursive install-ps-recursive install-recursive \
+	installcheck-recursive installdirs-recursive pdf-recursive \
+	ps-recursive uninstall-recursive
 RECURSIVE_CLEAN_TARGETS = mostlyclean-recursive clean-recursive	\
   distclean-recursive maintainer-clean-recursive
-am__recursive_targets = \
-  $(RECURSIVE_TARGETS) \
-  $(RECURSIVE_CLEAN_TARGETS) \
-  $(am__extra_recursive_targets)
-AM_RECURSIVE_TARGETS = $(am__recursive_targets:-recursive=) TAGS CTAGS \
-	cscope distdir dist dist-all distcheck
-am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) $(LISP)
-# Read a list of newline-separated strings from the standard input,
-# and print each of them once, without duplicates.  Input order is
-# *not* preserved.
-am__uniquify_input = $(AWK) '\
-  BEGIN { nonempty = 0; } \
-  { items[$$0] = 1; nonempty = 1; } \
-  END { if (nonempty) { for (i in items) print i; }; } \
-'
-# Make sure the list of sources is unique.  This is necessary because,
-# e.g., the same source file might be shared among _SOURCES variables
-# for different programs/libraries.
-am__define_uniq_tagged_files = \
-  list='$(am__tagged_files)'; \
-  unique=`for i in $$list; do \
-    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
-  done | $(am__uniquify_input)`
+AM_RECURSIVE_TARGETS = $(RECURSIVE_TARGETS:-recursive=) \
+	$(RECURSIVE_CLEAN_TARGETS:-recursive=) tags TAGS ctags CTAGS \
+	distdir dist dist-all distcheck
 ETAGS = etags
 CTAGS = ctags
-CSCOPE = cscope
 DIST_SUBDIRS = $(SUBDIRS)
 DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
 distdir = $(PACKAGE)-$(VERSION)
 top_distdir = $(distdir)
 am__remove_distdir = \
-  if test -d "$(distdir)"; then \
-    find "$(distdir)" -type d ! -perm -200 -exec chmod u+w {} ';' \
-      && rm -rf "$(distdir)" \
-      || { sleep 5 && rm -rf "$(distdir)"; }; \
-  else :; fi
-am__post_remove_distdir = $(am__remove_distdir)
+  { test ! -d "$(distdir)" \
+    || { find "$(distdir)" -type d ! -perm -200 -exec chmod u+w {} ';' \
+         && rm -fr "$(distdir)"; }; }
 am__relativize = \
   dir0=`pwd`; \
   sed_first='s,^\([^/]*\)/.*$$,\1,'; \
@@ -189,10 +109,7 @@ am__relativize = \
   reldir="$$dir2"
 DIST_ARCHIVES = $(distdir).tar.gz
 GZIP_ENV = --best
-DIST_TARGETS = dist-gzip
 distuninstallcheck_listfiles = find . -type f -print
-am__distuninstallcheck_listfiles = $(distuninstallcheck_listfiles) \
-  | sed 's|^\./|$(prefix)/|' | grep -v '$(infodir)/dir$$'
 distcleancheck_listfiles = find . -type f -print
 ACLOCAL = @ACLOCAL@
 AMTAR = @AMTAR@
@@ -236,7 +153,6 @@ PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
 PACKAGE_NAME = @PACKAGE_NAME@
 PACKAGE_STRING = @PACKAGE_STRING@
 PACKAGE_TARNAME = @PACKAGE_TARNAME@
-PACKAGE_URL = @PACKAGE_URL@
 PACKAGE_VERSION = @PACKAGE_VERSION@
 PATH_SEPARATOR = @PATH_SEPARATOR@
 RANLIB = @RANLIB@
@@ -308,7 +224,7 @@ SUBDIRS = lib tests benchmarks
 all: all-recursive
 
 .SUFFIXES:
-am--refresh: Makefile
+am--refresh:
 	@:
 $(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
 	@for dep in $?; do \
@@ -346,25 +262,22 @@ docs/doxy.cfg: $(top_builddir)/config.status $(top_srcdir)/docs/doxy.cfg.in
 	cd $(top_builddir) && $(SHELL) ./config.status $@
 
 # This directory's subdirectories are mostly independent; you can cd
-# into them and run 'make' without going through this Makefile.
-# To change the values of 'make' variables: instead of editing Makefiles,
-# (1) if the variable is set in 'config.status', edit 'config.status'
-#     (which will cause the Makefiles to be regenerated when you run 'make');
-# (2) otherwise, pass the desired values on the 'make' command line.
-$(am__recursive_targets):
-	@fail=; \
-	if $(am__make_keepgoing); then \
-	  failcom='fail=yes'; \
-	else \
-	  failcom='exit 1'; \
-	fi; \
+# into them and run `make' without going through this Makefile.
+# To change the values of `make' variables: instead of editing Makefiles,
+# (1) if the variable is set in `config.status', edit `config.status'
+#     (which will cause the Makefiles to be regenerated when you run `make');
+# (2) otherwise, pass the desired values on the `make' command line.
+$(RECURSIVE_TARGETS):
+	@fail= failcom='exit 1'; \
+	for f in x $$MAKEFLAGS; do \
+	  case $$f in \
+	    *=* | --[!k]*);; \
+	    *k*) failcom='fail=yes';; \
+	  esac; \
+	done; \
 	dot_seen=no; \
 	target=`echo $@ | sed s/-recursive//`; \
-	case "$@" in \
-	  distclean-* | maintainer-clean-*) list='$(DIST_SUBDIRS)' ;; \
-	  *) list='$(SUBDIRS)' ;; \
-	esac; \
-	for subdir in $$list; do \
+	list='$(SUBDIRS)'; for subdir in $$list; do \
 	  echo "Making $$target in $$subdir"; \
 	  if test "$$subdir" = "."; then \
 	    dot_seen=yes; \
@@ -379,12 +292,57 @@ $(am__recursive_targets):
 	  $(MAKE) $(AM_MAKEFLAGS) "$$target-am" || exit 1; \
 	fi; test -z "$$fail"
 
-ID: $(am__tagged_files)
-	$(am__define_uniq_tagged_files); mkid -fID $$unique
-tags: tags-recursive
-TAGS: tags
+$(RECURSIVE_CLEAN_TARGETS):
+	@fail= failcom='exit 1'; \
+	for f in x $$MAKEFLAGS; do \
+	  case $$f in \
+	    *=* | --[!k]*);; \
+	    *k*) failcom='fail=yes';; \
+	  esac; \
+	done; \
+	dot_seen=no; \
+	case "$@" in \
+	  distclean-* | maintainer-clean-*) list='$(DIST_SUBDIRS)' ;; \
+	  *) list='$(SUBDIRS)' ;; \
+	esac; \
+	rev=''; for subdir in $$list; do \
+	  if test "$$subdir" = "."; then :; else \
+	    rev="$$subdir $$rev"; \
+	  fi; \
+	done; \
+	rev="$$rev ."; \
+	target=`echo $@ | sed s/-recursive//`; \
+	for subdir in $$rev; do \
+	  echo "Making $$target in $$subdir"; \
+	  if test "$$subdir" = "."; then \
+	    local_target="$$target-am"; \
+	  else \
+	    local_target="$$target"; \
+	  fi; \
+	  ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) $$local_target) \
+	  || eval $$failcom; \
+	done && test -z "$$fail"
+tags-recursive:
+	list='$(SUBDIRS)'; for subdir in $$list; do \
+	  test "$$subdir" = . || ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) tags); \
+	done
+ctags-recursive:
+	list='$(SUBDIRS)'; for subdir in $$list; do \
+	  test "$$subdir" = . || ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) ctags); \
+	done
 
-tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
+ID: $(HEADERS) $(SOURCES) $(LISP) $(TAGS_FILES)
+	list='$(SOURCES) $(HEADERS) $(LISP) $(TAGS_FILES)'; \
+	unique=`for i in $$list; do \
+	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+	  done | \
+	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
+	      END { if (nonempty) { for (i in files) print i; }; }'`; \
+	mkid -fID $$unique
+tags: TAGS
+
+TAGS: tags-recursive $(HEADERS) $(SOURCES)  $(TAGS_DEPENDENCIES) \
+		$(TAGS_FILES) $(LISP)
 	set x; \
 	here=`pwd`; \
 	if ($(ETAGS) --etags-include --version) >/dev/null 2>&1; then \
@@ -400,7 +358,12 @@ tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
 	      set "$$@" "$$include_option=$$here/$$subdir/TAGS"; \
 	  fi; \
 	done; \
-	$(am__define_uniq_tagged_files); \
+	list='$(SOURCES) $(HEADERS)  $(LISP) $(TAGS_FILES)'; \
+	unique=`for i in $$list; do \
+	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+	  done | \
+	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
+	      END { if (nonempty) { for (i in files) print i; }; }'`; \
 	shift; \
 	if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
 	  test -n "$$unique" || unique=$$empty_fix; \
@@ -412,11 +375,15 @@ tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
 	      $$unique; \
 	  fi; \
 	fi
-ctags: ctags-recursive
-
-CTAGS: ctags
-ctags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
-	$(am__define_uniq_tagged_files); \
+ctags: CTAGS
+CTAGS: ctags-recursive $(HEADERS) $(SOURCES)  $(TAGS_DEPENDENCIES) \
+		$(TAGS_FILES) $(LISP)
+	list='$(SOURCES) $(HEADERS)  $(LISP) $(TAGS_FILES)'; \
+	unique=`for i in $$list; do \
+	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+	  done | \
+	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
+	      END { if (nonempty) { for (i in files) print i; }; }'`; \
 	test -z "$(CTAGS_ARGS)$$unique" \
 	  || $(CTAGS) $(CTAGSFLAGS) $(AM_CTAGSFLAGS) $(CTAGS_ARGS) \
 	     $$unique
@@ -425,31 +392,9 @@ GTAGS:
 	here=`$(am__cd) $(top_builddir) && pwd` \
 	  && $(am__cd) $(top_srcdir) \
 	  && gtags -i $(GTAGS_ARGS) "$$here"
-cscope: cscope.files
-	test ! -s cscope.files \
-	  || $(CSCOPE) -b -q $(AM_CSCOPEFLAGS) $(CSCOPEFLAGS) -i cscope.files $(CSCOPE_ARGS)
-clean-cscope:
-	-rm -f cscope.files
-cscope.files: clean-cscope cscopelist
-cscopelist: cscopelist-recursive
-
-cscopelist-am: $(am__tagged_files)
-	list='$(am__tagged_files)'; \
-	case "$(srcdir)" in \
-	  [\\/]* | ?:[\\/]*) sdir="$(srcdir)" ;; \
-	  *) sdir=$(subdir)/$(srcdir) ;; \
-	esac; \
-	for i in $$list; do \
-	  if test -f "$$i"; then \
-	    echo "$(subdir)/$$i"; \
-	  else \
-	    echo "$$sdir/$$i"; \
-	  fi; \
-	done >> $(top_builddir)/cscope.files
 
 distclean-tags:
 	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
-	-rm -f cscope.out cscope.in.out cscope.po.out cscope.files
 
 distdir: $(DISTFILES)
 	$(am__remove_distdir)
@@ -485,10 +430,13 @@ distdir: $(DISTFILES)
 	done
 	@list='$(DIST_SUBDIRS)'; for subdir in $$list; do \
 	  if test "$$subdir" = .; then :; else \
-	    $(am__make_dryrun) \
-	      || test -d "$(distdir)/$$subdir" \
-	      || $(MKDIR_P) "$(distdir)/$$subdir" \
-	      || exit 1; \
+	    test -d "$(distdir)/$$subdir" \
+	    || $(MKDIR_P) "$(distdir)/$$subdir" \
+	    || exit 1; \
+	  fi; \
+	done
+	@list='$(DIST_SUBDIRS)'; for subdir in $$list; do \
+	  if test "$$subdir" = .; then :; else \
 	    dir1=$$subdir; dir2="$(distdir)/$$subdir"; \
 	    $(am__relativize); \
 	    new_distdir=$$reldir; \
@@ -517,42 +465,36 @@ distdir: $(DISTFILES)
 	|| chmod -R a+r "$(distdir)"
 dist-gzip: distdir
 	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 
 dist-bzip2: distdir
-	tardir=$(distdir) && $(am__tar) | BZIP2=$${BZIP2--9} bzip2 -c >$(distdir).tar.bz2
-	$(am__post_remove_distdir)
+	tardir=$(distdir) && $(am__tar) | bzip2 -9 -c >$(distdir).tar.bz2
+	$(am__remove_distdir)
 
-dist-lzip: distdir
-	tardir=$(distdir) && $(am__tar) | lzip -c $${LZIP_OPT--9} >$(distdir).tar.lz
-	$(am__post_remove_distdir)
+dist-lzma: distdir
+	tardir=$(distdir) && $(am__tar) | lzma -9 -c >$(distdir).tar.lzma
+	$(am__remove_distdir)
 
 dist-xz: distdir
-	tardir=$(distdir) && $(am__tar) | XZ_OPT=$${XZ_OPT--e} xz -c >$(distdir).tar.xz
-	$(am__post_remove_distdir)
+	tardir=$(distdir) && $(am__tar) | xz -c >$(distdir).tar.xz
+	$(am__remove_distdir)
 
 dist-tarZ: distdir
-	@echo WARNING: "Support for shar distribution archives is" \
-	               "deprecated." >&2
-	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	tardir=$(distdir) && $(am__tar) | compress -c >$(distdir).tar.Z
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 
 dist-shar: distdir
-	@echo WARNING: "Support for distribution archives compressed with" \
-		       "legacy program 'compress' is deprecated." >&2
-	@echo WARNING: "It will be removed altogether in Automake 2.0" >&2
 	shar $(distdir) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).shar.gz
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 
 dist-zip: distdir
 	-rm -f $(distdir).zip
 	zip -rq $(distdir).zip $(distdir)
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 
-dist dist-all:
-	$(MAKE) $(AM_MAKEFLAGS) $(DIST_TARGETS) am__post_remove_distdir='@:'
-	$(am__post_remove_distdir)
+dist dist-all: distdir
+	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
+	$(am__remove_distdir)
 
 # This target untars the dist file and tries a VPATH configuration.  Then
 # it guarantees that the distribution is self-contained by making another
@@ -563,8 +505,8 @@ distcheck: dist
 	  GZIP=$(GZIP_ENV) gzip -dc $(distdir).tar.gz | $(am__untar) ;;\
 	*.tar.bz2*) \
 	  bzip2 -dc $(distdir).tar.bz2 | $(am__untar) ;;\
-	*.tar.lz*) \
-	  lzip -dc $(distdir).tar.lz | $(am__untar) ;;\
+	*.tar.lzma*) \
+	  lzma -dc $(distdir).tar.lzma | $(am__untar) ;;\
 	*.tar.xz*) \
 	  xz -dc $(distdir).tar.xz | $(am__untar) ;;\
 	*.tar.Z*) \
@@ -574,19 +516,17 @@ distcheck: dist
 	*.zip*) \
 	  unzip $(distdir).zip ;;\
 	esac
-	chmod -R a-w $(distdir)
-	chmod u+w $(distdir)
-	mkdir $(distdir)/_build $(distdir)/_inst
+	chmod -R a-w $(distdir); chmod u+w $(distdir)
+	mkdir $(distdir)/_build
+	mkdir $(distdir)/_inst
 	chmod a-w $(distdir)
 	test -d $(distdir)/_build || exit 0; \
 	dc_install_base=`$(am__cd) $(distdir)/_inst && pwd | sed -e 's,^[^:\\/]:[\\/],/,'` \
 	  && dc_destdir="$${TMPDIR-/tmp}/am-dc-$$$$/" \
 	  && am__cwd=`pwd` \
 	  && $(am__cd) $(distdir)/_build \
-	  && ../configure \
-	    $(AM_DISTCHECK_CONFIGURE_FLAGS) \
+	  && ../configure --srcdir=.. --prefix="$$dc_install_base" \
 	    $(DISTCHECK_CONFIGURE_FLAGS) \
-	    --srcdir=.. --prefix="$$dc_install_base" \
 	  && $(MAKE) $(AM_MAKEFLAGS) \
 	  && $(MAKE) $(AM_MAKEFLAGS) dvi \
 	  && $(MAKE) $(AM_MAKEFLAGS) check \
@@ -609,21 +549,13 @@ distcheck: dist
 	  && $(MAKE) $(AM_MAKEFLAGS) distcleancheck \
 	  && cd "$$am__cwd" \
 	  || exit 1
-	$(am__post_remove_distdir)
+	$(am__remove_distdir)
 	@(echo "$(distdir) archives ready for distribution: "; \
 	  list='$(DIST_ARCHIVES)'; for i in $$list; do echo $$i; done) | \
 	  sed -e 1h -e 1s/./=/g -e 1p -e 1x -e '$$p' -e '$$x'
 distuninstallcheck:
-	@test -n '$(distuninstallcheck_dir)' || { \
-	  echo 'ERROR: trying to run $@ with an empty' \
-	       '$$(distuninstallcheck_dir)' >&2; \
-	  exit 1; \
-	}; \
-	$(am__cd) '$(distuninstallcheck_dir)' || { \
-	  echo 'ERROR: cannot chdir into $(distuninstallcheck_dir)' >&2; \
-	  exit 1; \
-	}; \
-	test `$(am__distuninstallcheck_listfiles) | wc -l` -eq 0 \
+	@$(am__cd) '$(distuninstallcheck_dir)' \
+	&& test `$(distuninstallcheck_listfiles) | wc -l` -le 1 \
 	   || { echo "ERROR: files left after uninstall:" ; \
 	        if test -n "$(DESTDIR)"; then \
 	          echo "  (check DESTDIR support)"; \
@@ -654,15 +586,10 @@ install-am: all-am
 
 installcheck: installcheck-recursive
 install-strip:
-	if test -z '$(STRIP)'; then \
-	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
-	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
-	      install; \
-	else \
-	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
-	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
-	    "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'" install; \
-	fi
+	$(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
+	  install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
+	  `test -z '$(STRIP)' || \
+	    echo "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'"` install
 mostlyclean-generic:
 
 clean-generic:
@@ -743,22 +670,23 @@ ps-am:
 
 uninstall-am:
 
-.MAKE: $(am__recursive_targets) install-am install-strip
+.MAKE: $(RECURSIVE_CLEAN_TARGETS) $(RECURSIVE_TARGETS) ctags-recursive \
+	install-am install-strip tags-recursive
 
-.PHONY: $(am__recursive_targets) CTAGS GTAGS TAGS all all-am \
-	am--refresh check check-am clean clean-cscope clean-generic \
-	cscope cscopelist-am ctags ctags-am dist dist-all dist-bzip2 \
-	dist-gzip dist-lzip dist-shar dist-tarZ dist-xz dist-zip \
-	distcheck distclean distclean-generic distclean-tags \
-	distcleancheck distdir distuninstallcheck dvi dvi-am html \
-	html-am info info-am install install-am install-data \
-	install-data-am install-dvi install-dvi-am install-exec \
-	install-exec-am install-html install-html-am install-info \
-	install-info-am install-man install-pdf install-pdf-am \
-	install-ps install-ps-am install-strip installcheck \
-	installcheck-am installdirs installdirs-am maintainer-clean \
+.PHONY: $(RECURSIVE_CLEAN_TARGETS) $(RECURSIVE_TARGETS) CTAGS GTAGS \
+	all all-am am--refresh check check-am clean clean-generic \
+	ctags ctags-recursive dist dist-all dist-bzip2 dist-gzip \
+	dist-lzma dist-shar dist-tarZ dist-xz dist-zip distcheck \
+	distclean distclean-generic distclean-tags distcleancheck \
+	distdir distuninstallcheck dvi dvi-am html html-am info \
+	info-am install install-am install-data install-data-am \
+	install-dvi install-dvi-am install-exec install-exec-am \
+	install-html install-html-am install-info install-info-am \
+	install-man install-pdf install-pdf-am install-ps \
+	install-ps-am install-strip installcheck installcheck-am \
+	installdirs installdirs-am maintainer-clean \
 	maintainer-clean-generic mostlyclean mostlyclean-generic pdf \
-	pdf-am ps ps-am tags tags-am uninstall uninstall-am
+	pdf-am ps ps-am tags tags-recursive uninstall uninstall-am
 
 
 # Tell versions [3.59,3.63) of GNU make to not export all variables.
diff --git a/aclocal.m4 b/aclocal.m4
index d5fa395c..e517c84c 100644
--- a/aclocal.m4
+++ b/aclocal.m4
@@ -1,7 +1,7 @@
-# generated automatically by aclocal 1.14.1 -*- Autoconf -*-
-
-# Copyright (C) 1996-2013 Free Software Foundation, Inc.
+# generated automatically by aclocal 1.11.1 -*- Autoconf -*-
 
+# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
+# 2005, 2006, 2007, 2008, 2009  Free Software Foundation, Inc.
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
@@ -11,16 +11,15 @@
 # even the implied warranty of MERCHANTABILITY or FITNESS FOR A
 # PARTICULAR PURPOSE.
 
-m4_ifndef([AC_CONFIG_MACRO_DIRS], [m4_defun([_AM_CONFIG_MACRO_DIRS], [])m4_defun([AC_CONFIG_MACRO_DIRS], [_AM_CONFIG_MACRO_DIRS($@)])])
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
-m4_if(m4_defn([AC_AUTOCONF_VERSION]), [2.69],,
-[m4_warning([this file was generated for autoconf 2.69.
+m4_if(m4_defn([AC_AUTOCONF_VERSION]), [2.63],,
+[m4_warning([this file was generated for autoconf 2.63.
 You have another version of autoconf.  It may work, but is not guaranteed to.
 If you have problems, you may need to regenerate the build system entirely.
-To do so, use the procedure documented by the package, typically 'autoreconf'.])])
+To do so, use the procedure documented by the package, typically `autoreconf'.])])
 
-# Copyright (C) 2002-2013 Free Software Foundation, Inc.
+# Copyright (C) 2002, 2003, 2005, 2006, 2007, 2008  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -32,10 +31,10 @@ To do so, use the procedure documented by the package, typically 'autoreconf'.])
 # generated from the m4 files accompanying Automake X.Y.
 # (This private macro should not be called outside this file.)
 AC_DEFUN([AM_AUTOMAKE_VERSION],
-[am__api_version='1.14'
+[am__api_version='1.11'
 dnl Some users find AM_AUTOMAKE_VERSION and mistake it for a way to
 dnl require some minimum version.  Point them to the right macro.
-m4_if([$1], [1.14.1], [],
+m4_if([$1], [1.11.1], [],
       [AC_FATAL([Do not call $0, use AM_INIT_AUTOMAKE([$1]).])])dnl
 ])
 
@@ -51,22 +50,22 @@ m4_define([_AM_AUTOCONF_VERSION], [])
 # Call AM_AUTOMAKE_VERSION and AM_AUTOMAKE_VERSION so they can be traced.
 # This function is AC_REQUIREd by AM_INIT_AUTOMAKE.
 AC_DEFUN([AM_SET_CURRENT_AUTOMAKE_VERSION],
-[AM_AUTOMAKE_VERSION([1.14.1])dnl
+[AM_AUTOMAKE_VERSION([1.11.1])dnl
 m4_ifndef([AC_AUTOCONF_VERSION],
   [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
 _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 
 # AM_AUX_DIR_EXPAND                                         -*- Autoconf -*-
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001, 2003, 2005  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
 # For projects using AC_CONFIG_AUX_DIR([foo]), Autoconf sets
-# $ac_aux_dir to '$srcdir/foo'.  In other projects, it is set to
-# '$srcdir', '$srcdir/..', or '$srcdir/../..'.
+# $ac_aux_dir to `$srcdir/foo'.  In other projects, it is set to
+# `$srcdir', `$srcdir/..', or `$srcdir/../..'.
 #
 # Of course, Automake must honor this variable whenever it calls a
 # tool from the auxiliary directory.  The problem is that $srcdir (and
@@ -85,7 +84,7 @@ _AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
 #
 # The reason of the latter failure is that $top_srcdir and $ac_aux_dir
 # are both prefixed by $srcdir.  In an in-source build this is usually
-# harmless because $srcdir is '.', but things will broke when you
+# harmless because $srcdir is `.', but things will broke when you
 # start a VPATH build or use an absolute $srcdir.
 #
 # So we could use something similar to $top_srcdir/$ac_aux_dir/missing,
@@ -111,19 +110,22 @@ am_aux_dir=`cd $ac_aux_dir && pwd`
 
 # AM_CONDITIONAL                                            -*- Autoconf -*-
 
-# Copyright (C) 1997-2013 Free Software Foundation, Inc.
+# Copyright (C) 1997, 2000, 2001, 2003, 2004, 2005, 2006, 2008
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 9
+
 # AM_CONDITIONAL(NAME, SHELL-CONDITION)
 # -------------------------------------
 # Define a conditional.
 AC_DEFUN([AM_CONDITIONAL],
-[AC_PREREQ([2.52])dnl
- m4_if([$1], [TRUE],  [AC_FATAL([$0: invalid condition: $1])],
-       [$1], [FALSE], [AC_FATAL([$0: invalid condition: $1])])dnl
+[AC_PREREQ(2.52)dnl
+ ifelse([$1], [TRUE],  [AC_FATAL([$0: invalid condition: $1])],
+	[$1], [FALSE], [AC_FATAL([$0: invalid condition: $1])])dnl
 AC_SUBST([$1_TRUE])dnl
 AC_SUBST([$1_FALSE])dnl
 _AM_SUBST_NOTMAKE([$1_TRUE])dnl
@@ -142,14 +144,16 @@ AC_CONFIG_COMMANDS_PRE(
 Usually this means the macro was only invoked conditionally.]])
 fi])])
 
-# Copyright (C) 1999-2013 Free Software Foundation, Inc.
+# Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2009
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 10
 
-# There are a few dirty hacks below to avoid letting 'AC_PROG_CC' be
+# There are a few dirty hacks below to avoid letting `AC_PROG_CC' be
 # written in clear, in which case automake, when reading aclocal.m4,
 # will think it sees a *use*, and therefore will trigger all it's
 # C support machinery.  Also note that it means that autoscan, seeing
@@ -159,7 +163,7 @@ fi])])
 # _AM_DEPENDENCIES(NAME)
 # ----------------------
 # See how the compiler implements dependency checking.
-# NAME is "CC", "CXX", "OBJC", "OBJCXX", "UPC", or "GJC".
+# NAME is "CC", "CXX", "GCJ", or "OBJC".
 # We try a few techniques and use that to set a single cache variable.
 #
 # We don't AC_REQUIRE the corresponding AC_PROG_CC since the latter was
@@ -172,13 +176,12 @@ AC_REQUIRE([AM_OUTPUT_DEPENDENCY_COMMANDS])dnl
 AC_REQUIRE([AM_MAKE_INCLUDE])dnl
 AC_REQUIRE([AM_DEP_TRACK])dnl
 
-m4_if([$1], [CC],   [depcc="$CC"   am_compiler_list=],
-      [$1], [CXX],  [depcc="$CXX"  am_compiler_list=],
-      [$1], [OBJC], [depcc="$OBJC" am_compiler_list='gcc3 gcc'],
-      [$1], [OBJCXX], [depcc="$OBJCXX" am_compiler_list='gcc3 gcc'],
-      [$1], [UPC],  [depcc="$UPC"  am_compiler_list=],
-      [$1], [GCJ],  [depcc="$GCJ"  am_compiler_list='gcc3 gcc'],
-                    [depcc="$$1"   am_compiler_list=])
+ifelse([$1], CC,   [depcc="$CC"   am_compiler_list=],
+       [$1], CXX,  [depcc="$CXX"  am_compiler_list=],
+       [$1], OBJC, [depcc="$OBJC" am_compiler_list='gcc3 gcc'],
+       [$1], UPC,  [depcc="$UPC"  am_compiler_list=],
+       [$1], GCJ,  [depcc="$GCJ"  am_compiler_list='gcc3 gcc'],
+                   [depcc="$$1"   am_compiler_list=])
 
 AC_CACHE_CHECK([dependency style of $depcc],
                [am_cv_$1_dependencies_compiler_type],
@@ -186,9 +189,8 @@ AC_CACHE_CHECK([dependency style of $depcc],
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named 'D' -- because '-MD' means "put the output
-  # in D".
-  rm -rf conftest.dir
+  # making a dummy file named `D' -- because `-MD' means `put the output
+  # in D'.
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -227,16 +229,16 @@ AC_CACHE_CHECK([dependency style of $depcc],
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
-      # Solaris 10 /bin/sh.
-      echo '/* dummy */' > sub/conftst$i.h
+      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
+      # Solaris 8's {/usr,}/bin/sh.
+      touch sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with '-c' and '-o' for the sake of the "dashmstdout"
+    # We check with `-c' and `-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle '-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs.
+    # handle `-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -245,16 +247,16 @@ AC_CACHE_CHECK([dependency style of $depcc],
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # After this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested.
+      # after this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
-      # This compiler won't grok '-c -o', but also, the minuso test has
+    msvisualcpp | msvcmsys)
+      # This compiler won't grok `-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -302,7 +304,7 @@ AM_CONDITIONAL([am__fastdep$1], [
 # AM_SET_DEPDIR
 # -------------
 # Choose a directory name for dependency files.
-# This macro is AC_REQUIREd in _AM_DEPENDENCIES.
+# This macro is AC_REQUIREd in _AM_DEPENDENCIES
 AC_DEFUN([AM_SET_DEPDIR],
 [AC_REQUIRE([AM_SET_LEADING_DOT])dnl
 AC_SUBST([DEPDIR], ["${am__leading_dot}deps"])dnl
@@ -312,39 +314,34 @@ AC_SUBST([DEPDIR], ["${am__leading_dot}deps"])dnl
 # AM_DEP_TRACK
 # ------------
 AC_DEFUN([AM_DEP_TRACK],
-[AC_ARG_ENABLE([dependency-tracking], [dnl
-AS_HELP_STRING(
-  [--enable-dependency-tracking],
-  [do not reject slow dependency extractors])
-AS_HELP_STRING(
-  [--disable-dependency-tracking],
-  [speeds up one-time build])])
+[AC_ARG_ENABLE(dependency-tracking,
+[  --disable-dependency-tracking  speeds up one-time build
+  --enable-dependency-tracking   do not reject slow dependency extractors])
 if test "x$enable_dependency_tracking" != xno; then
   am_depcomp="$ac_aux_dir/depcomp"
   AMDEPBACKSLASH='\'
-  am__nodep='_no'
 fi
 AM_CONDITIONAL([AMDEP], [test "x$enable_dependency_tracking" != xno])
 AC_SUBST([AMDEPBACKSLASH])dnl
 _AM_SUBST_NOTMAKE([AMDEPBACKSLASH])dnl
-AC_SUBST([am__nodep])dnl
-_AM_SUBST_NOTMAKE([am__nodep])dnl
 ])
 
 # Generate code to set up dependency tracking.              -*- Autoconf -*-
 
-# Copyright (C) 1999-2013 Free Software Foundation, Inc.
+# Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2008
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+#serial 5
 
 # _AM_OUTPUT_DEPENDENCY_COMMANDS
 # ------------------------------
 AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
 [{
-  # Older Autoconf quotes --file arguments for eval, but not when files
+  # Autoconf 2.62 quotes --file arguments for eval, but not when files
   # are listed without --file.  Let's play safe and only enable the eval
   # if we detect the quoting.
   case $CONFIG_FILES in
@@ -357,7 +354,7 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
     # Strip MF so we end up with the name of the file.
     mf=`echo "$mf" | sed -e 's/:.*$//'`
     # Check whether this is an Automake generated Makefile or not.
-    # We used to match only the files named 'Makefile.in', but
+    # We used to match only the files named `Makefile.in', but
     # some people rename them; so instead we look at the file content.
     # Grep'ing the first line is not enough: some people post-process
     # each Makefile.in and add a new line on top of each file to say so.
@@ -369,19 +366,21 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
       continue
     fi
     # Extract the definition of DEPDIR, am__include, and am__quote
-    # from the Makefile without running 'make'.
+    # from the Makefile without running `make'.
     DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
     test -z "$DEPDIR" && continue
     am__include=`sed -n 's/^am__include = //p' < "$mf"`
-    test -z "$am__include" && continue
+    test -z "am__include" && continue
     am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
+    # When using ansi2knr, U may be empty or an underscore; expand it
+    U=`sed -n 's/^U = //p' < "$mf"`
     # Find all dependency output files, they are included files with
     # $(DEPDIR) in their names.  We invoke sed twice because it is the
     # simplest approach to changing $(DEPDIR) to its actual value in the
     # expansion.
     for file in `sed -n "
       s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
-	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
+	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g' -e 's/\$U/'"$U"'/g'`; do
       # Make sure the directory exists.
       test -f "$dirpart/$file" && continue
       fdir=`AS_DIRNAME(["$file"])`
@@ -399,7 +398,7 @@ AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
 # This macro should only be invoked once -- use via AC_REQUIRE.
 #
 # This code is only required when automatic dependency tracking
-# is enabled.  FIXME.  This creates each '.P' file that we will
+# is enabled.  FIXME.  This creates each `.P' file that we will
 # need in order to bootstrap the dependency handling code.
 AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 [AC_CONFIG_COMMANDS([depfiles],
@@ -409,21 +408,18 @@ AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
 
 # Do all the work for Automake.                             -*- Autoconf -*-
 
-# Copyright (C) 1996-2013 Free Software Foundation, Inc.
+# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
+# 2005, 2006, 2008, 2009 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 16
+
 # This macro actually does too much.  Some checks are only needed if
 # your package does certain things.  But this isn't really a big deal.
 
-dnl Redefine AC_PROG_CC to automatically invoke _AM_PROG_CC_C_O.
-m4_define([AC_PROG_CC],
-m4_defn([AC_PROG_CC])
-[_AM_PROG_CC_C_O
-])
-
 # AM_INIT_AUTOMAKE(PACKAGE, VERSION, [NO-DEFINE])
 # AM_INIT_AUTOMAKE([OPTIONS])
 # -----------------------------------------------
@@ -436,7 +432,7 @@ m4_defn([AC_PROG_CC])
 # arguments mandatory, and then we can depend on a new Autoconf
 # release and drop the old call support.
 AC_DEFUN([AM_INIT_AUTOMAKE],
-[AC_PREREQ([2.65])dnl
+[AC_PREREQ([2.62])dnl
 dnl Autoconf wants to disallow AM_ names.  We explicitly allow
 dnl the ones we care about.
 m4_pattern_allow([^AM_[A-Z]+FLAGS$])dnl
@@ -465,40 +461,31 @@ AC_SUBST([CYGPATH_W])
 # Define the identity of the package.
 dnl Distinguish between old-style and new-style calls.
 m4_ifval([$2],
-[AC_DIAGNOSE([obsolete],
-             [$0: two- and three-arguments forms are deprecated.])
-m4_ifval([$3], [_AM_SET_OPTION([no-define])])dnl
+[m4_ifval([$3], [_AM_SET_OPTION([no-define])])dnl
  AC_SUBST([PACKAGE], [$1])dnl
  AC_SUBST([VERSION], [$2])],
 [_AM_SET_OPTIONS([$1])dnl
 dnl Diagnose old-style AC_INIT with new-style AM_AUTOMAKE_INIT.
-m4_if(
-  m4_ifdef([AC_PACKAGE_NAME], [ok]):m4_ifdef([AC_PACKAGE_VERSION], [ok]),
-  [ok:ok],,
+m4_if(m4_ifdef([AC_PACKAGE_NAME], 1)m4_ifdef([AC_PACKAGE_VERSION], 1), 11,,
   [m4_fatal([AC_INIT should be called with package and version arguments])])dnl
  AC_SUBST([PACKAGE], ['AC_PACKAGE_TARNAME'])dnl
  AC_SUBST([VERSION], ['AC_PACKAGE_VERSION'])])dnl
 
 _AM_IF_OPTION([no-define],,
-[AC_DEFINE_UNQUOTED([PACKAGE], ["$PACKAGE"], [Name of package])
- AC_DEFINE_UNQUOTED([VERSION], ["$VERSION"], [Version number of package])])dnl
+[AC_DEFINE_UNQUOTED(PACKAGE, "$PACKAGE", [Name of package])
+ AC_DEFINE_UNQUOTED(VERSION, "$VERSION", [Version number of package])])dnl
 
 # Some tools Automake needs.
 AC_REQUIRE([AM_SANITY_CHECK])dnl
 AC_REQUIRE([AC_ARG_PROGRAM])dnl
-AM_MISSING_PROG([ACLOCAL], [aclocal-${am__api_version}])
-AM_MISSING_PROG([AUTOCONF], [autoconf])
-AM_MISSING_PROG([AUTOMAKE], [automake-${am__api_version}])
-AM_MISSING_PROG([AUTOHEADER], [autoheader])
-AM_MISSING_PROG([MAKEINFO], [makeinfo])
+AM_MISSING_PROG(ACLOCAL, aclocal-${am__api_version})
+AM_MISSING_PROG(AUTOCONF, autoconf)
+AM_MISSING_PROG(AUTOMAKE, automake-${am__api_version})
+AM_MISSING_PROG(AUTOHEADER, autoheader)
+AM_MISSING_PROG(MAKEINFO, makeinfo)
 AC_REQUIRE([AM_PROG_INSTALL_SH])dnl
 AC_REQUIRE([AM_PROG_INSTALL_STRIP])dnl
-AC_REQUIRE([AC_PROG_MKDIR_P])dnl
-# For better backward compatibility.  To be removed once Automake 1.9.x
-# dies out for good.  For more background, see:
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
-AC_SUBST([mkdir_p], ['$(MKDIR_P)'])
+AC_REQUIRE([AM_PROG_MKDIR_P])dnl
 # We need awk for the "check" target.  The system "awk" is bad on
 # some platforms.
 AC_REQUIRE([AC_PROG_AWK])dnl
@@ -509,78 +496,34 @@ _AM_IF_OPTION([tar-ustar], [_AM_PROG_TAR([ustar])],
 			     [_AM_PROG_TAR([v7])])])
 _AM_IF_OPTION([no-dependencies],,
 [AC_PROVIDE_IFELSE([AC_PROG_CC],
-		  [_AM_DEPENDENCIES([CC])],
-		  [m4_define([AC_PROG_CC],
-			     m4_defn([AC_PROG_CC])[_AM_DEPENDENCIES([CC])])])dnl
+		  [_AM_DEPENDENCIES(CC)],
+		  [define([AC_PROG_CC],
+			  defn([AC_PROG_CC])[_AM_DEPENDENCIES(CC)])])dnl
 AC_PROVIDE_IFELSE([AC_PROG_CXX],
-		  [_AM_DEPENDENCIES([CXX])],
-		  [m4_define([AC_PROG_CXX],
-			     m4_defn([AC_PROG_CXX])[_AM_DEPENDENCIES([CXX])])])dnl
+		  [_AM_DEPENDENCIES(CXX)],
+		  [define([AC_PROG_CXX],
+			  defn([AC_PROG_CXX])[_AM_DEPENDENCIES(CXX)])])dnl
 AC_PROVIDE_IFELSE([AC_PROG_OBJC],
-		  [_AM_DEPENDENCIES([OBJC])],
-		  [m4_define([AC_PROG_OBJC],
-			     m4_defn([AC_PROG_OBJC])[_AM_DEPENDENCIES([OBJC])])])dnl
-AC_PROVIDE_IFELSE([AC_PROG_OBJCXX],
-		  [_AM_DEPENDENCIES([OBJCXX])],
-		  [m4_define([AC_PROG_OBJCXX],
-			     m4_defn([AC_PROG_OBJCXX])[_AM_DEPENDENCIES([OBJCXX])])])dnl
+		  [_AM_DEPENDENCIES(OBJC)],
+		  [define([AC_PROG_OBJC],
+			  defn([AC_PROG_OBJC])[_AM_DEPENDENCIES(OBJC)])])dnl
 ])
-AC_REQUIRE([AM_SILENT_RULES])dnl
-dnl The testsuite driver may need to know about EXEEXT, so add the
-dnl 'am__EXEEXT' conditional if _AM_COMPILER_EXEEXT was seen.  This
-dnl macro is hooked onto _AC_COMPILER_EXEEXT early, see below.
+_AM_IF_OPTION([silent-rules], [AC_REQUIRE([AM_SILENT_RULES])])dnl
+dnl The `parallel-tests' driver may need to know about EXEEXT, so add the
+dnl `am__EXEEXT' conditional if _AM_COMPILER_EXEEXT was seen.  This macro
+dnl is hooked onto _AC_COMPILER_EXEEXT early, see below.
 AC_CONFIG_COMMANDS_PRE(dnl
 [m4_provide_if([_AM_COMPILER_EXEEXT],
   [AM_CONDITIONAL([am__EXEEXT], [test -n "$EXEEXT"])])])dnl
+])
 
-# POSIX will say in a future version that running "rm -f" with no argument
-# is OK; and we want to be able to make that assumption in our Makefile
-# recipes.  So use an aggressive probe to check that the usage we want is
-# actually supported "in the wild" to an acceptable degree.
-# See automake bug#10828.
-# To make any issue more visible, cause the running configure to be aborted
-# by default if the 'rm' program in use doesn't match our expectations; the
-# user can still override this though.
-if rm -f && rm -fr && rm -rf; then : OK; else
-  cat >&2 <<'END'
-Oops!
-
-Your 'rm' program seems unable to run without file operands specified
-on the command line, even when the '-f' option is present.  This is contrary
-to the behaviour of most rm programs out there, and not conforming with
-the upcoming POSIX standard: <http://austingroupbugs.net/view.php?id=542>
-
-Please tell bug-automake@gnu.org about your system, including the value
-of your $PATH and any error possibly output before this message.  This
-can help us improve future automake versions.
-
-END
-  if test x"$ACCEPT_INFERIOR_RM_PROGRAM" = x"yes"; then
-    echo 'Configuration will proceed anyway, since you have set the' >&2
-    echo 'ACCEPT_INFERIOR_RM_PROGRAM variable to "yes"' >&2
-    echo >&2
-  else
-    cat >&2 <<'END'
-Aborting the configuration process, to ensure you take notice of the issue.
-
-You can download and install GNU coreutils to get an 'rm' implementation
-that behaves properly: <http://www.gnu.org/software/coreutils/>.
-
-If you want to complete the configuration process using your problematic
-'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
-to "yes", and re-run configure.
-
-END
-    AC_MSG_ERROR([Your 'rm' program is bad, sorry.])
-  fi
-fi])
-
-dnl Hook into '_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
+dnl Hook into `_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
 dnl add the conditional right here, as _AC_COMPILER_EXEEXT may be further
 dnl mangled by Autoconf and run in a shell conditional statement.
 m4_define([_AC_COMPILER_EXEEXT],
 m4_defn([_AC_COMPILER_EXEEXT])[m4_provide([_AM_COMPILER_EXEEXT])])
 
+
 # When config.status generates a header, we must update the stamp-h file.
 # This file resides in the same directory as the config header
 # that is generated.  The stamp files are numbered to have different names.
@@ -602,7 +545,7 @@ for _am_header in $config_headers :; do
 done
 echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_count])
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001, 2003, 2005, 2008  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -621,14 +564,16 @@ if test x"${install_sh}" != xset; then
     install_sh="\${SHELL} $am_aux_dir/install-sh"
   esac
 fi
-AC_SUBST([install_sh])])
+AC_SUBST(install_sh)])
 
-# Copyright (C) 2003-2013 Free Software Foundation, Inc.
+# Copyright (C) 2003, 2005  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 2
+
 # Check whether the underlying file-system supports filenames
 # with a leading dot.  For instance MS-DOS doesn't.
 AC_DEFUN([AM_SET_LEADING_DOT],
@@ -644,12 +589,14 @@ AC_SUBST([am__leading_dot])])
 
 # Check to see how 'make' treats includes.	            -*- Autoconf -*-
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001, 2002, 2003, 2005, 2009  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 4
+
 # AM_MAKE_INCLUDE()
 # -----------------
 # Check to see how make treats includes.
@@ -667,7 +614,7 @@ am__quote=
 _am_result=none
 # First try GNU make style include.
 echo "include confinc" > confmf
-# Ignore all kinds of additional output from 'make'.
+# Ignore all kinds of additional output from `make'.
 case `$am_make -s -f confmf 2> /dev/null` in #(
 *the\ am__doit\ target*)
   am__include=include
@@ -694,12 +641,15 @@ rm -f confinc confmf
 
 # Fake the existence of programs that GNU maintainers use.  -*- Autoconf -*-
 
-# Copyright (C) 1997-2013 Free Software Foundation, Inc.
+# Copyright (C) 1997, 1999, 2000, 2001, 2003, 2004, 2005, 2008
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 6
+
 # AM_MISSING_PROG(NAME, PROGRAM)
 # ------------------------------
 AC_DEFUN([AM_MISSING_PROG],
@@ -707,10 +657,11 @@ AC_DEFUN([AM_MISSING_PROG],
 $1=${$1-"${am_missing_run}$2"}
 AC_SUBST($1)])
 
+
 # AM_MISSING_HAS_RUN
 # ------------------
-# Define MISSING if not defined so far and test if it is modern enough.
-# If it is, set am_missing_run to use it, otherwise, to nothing.
+# Define MISSING if not defined so far and test if it supports --run.
+# If it does, set am_missing_run to use it, otherwise, to nothing.
 AC_DEFUN([AM_MISSING_HAS_RUN],
 [AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
 AC_REQUIRE_AUX_FILE([missing])dnl
@@ -723,35 +674,63 @@ if test x"${MISSING+set}" != xset; then
   esac
 fi
 # Use eval to expand $SHELL
-if eval "$MISSING --is-lightweight"; then
-  am_missing_run="$MISSING "
+if eval "$MISSING --run true"; then
+  am_missing_run="$MISSING --run "
 else
   am_missing_run=
-  AC_MSG_WARN(['missing' script is too old or missing])
+  AC_MSG_WARN([`missing' script is too old or missing])
 fi
 ])
 
-# Helper functions for option handling.                     -*- Autoconf -*-
-
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2003, 2004, 2005, 2006  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# AM_PROG_MKDIR_P
+# ---------------
+# Check for `mkdir -p'.
+AC_DEFUN([AM_PROG_MKDIR_P],
+[AC_PREREQ([2.60])dnl
+AC_REQUIRE([AC_PROG_MKDIR_P])dnl
+dnl Automake 1.8 to 1.9.6 used to define mkdir_p.  We now use MKDIR_P,
+dnl while keeping a definition of mkdir_p for backward compatibility.
+dnl @MKDIR_P@ is magic: AC_OUTPUT adjusts its value for each Makefile.
+dnl However we cannot define mkdir_p as $(MKDIR_P) for the sake of
+dnl Makefile.ins that do not define MKDIR_P, so we do our own
+dnl adjustment using top_builddir (which is defined more often than
+dnl MKDIR_P).
+AC_SUBST([mkdir_p], ["$MKDIR_P"])dnl
+case $mkdir_p in
+  [[\\/$]]* | ?:[[\\/]]*) ;;
+  */*) mkdir_p="\$(top_builddir)/$mkdir_p" ;;
+esac
+])
+
+# Helper functions for option handling.                     -*- Autoconf -*-
+
+# Copyright (C) 2001, 2002, 2003, 2005, 2008  Free Software Foundation, Inc.
+#
+# This file is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# serial 4
+
 # _AM_MANGLE_OPTION(NAME)
 # -----------------------
 AC_DEFUN([_AM_MANGLE_OPTION],
 [[_AM_OPTION_]m4_bpatsubst($1, [[^a-zA-Z0-9_]], [_])])
 
 # _AM_SET_OPTION(NAME)
-# --------------------
+# ------------------------------
 # Set option NAME.  Presently that only means defining a flag for this option.
 AC_DEFUN([_AM_SET_OPTION],
-[m4_define(_AM_MANGLE_OPTION([$1]), [1])])
+[m4_define(_AM_MANGLE_OPTION([$1]), 1)])
 
 # _AM_SET_OPTIONS(OPTIONS)
-# ------------------------
+# ----------------------------------
 # OPTIONS is a space-separated list of Automake options.
 AC_DEFUN([_AM_SET_OPTIONS],
 [m4_foreach_w([_AM_Option], [$1], [_AM_SET_OPTION(_AM_Option)])])
@@ -764,16 +743,22 @@ AC_DEFUN([_AM_IF_OPTION],
 
 # Check to make sure that the build environment is sane.    -*- Autoconf -*-
 
-# Copyright (C) 1996-2013 Free Software Foundation, Inc.
+# Copyright (C) 1996, 1997, 2000, 2001, 2003, 2005, 2008
+# Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 5
+
 # AM_SANITY_CHECK
 # ---------------
 AC_DEFUN([AM_SANITY_CHECK],
 [AC_MSG_CHECKING([whether build environment is sane])
+# Just in case
+sleep 1
+echo timestamp > conftest.file
 # Reject unsafe characters in $srcdir or the absolute working directory
 # name.  Accept space and tab only in the latter.
 am_lf='
@@ -784,40 +769,32 @@ case `pwd` in
 esac
 case $srcdir in
   *[[\\\"\#\$\&\'\`$am_lf\ \	]]*)
-    AC_MSG_ERROR([unsafe srcdir value: '$srcdir']);;
+    AC_MSG_ERROR([unsafe srcdir value: `$srcdir']);;
 esac
 
-# Do 'set' in a subshell so we don't clobber the current shell's
+# Do `set' in a subshell so we don't clobber the current shell's
 # arguments.  Must try -L first in case configure is actually a
 # symlink; some systems play weird games with the mod time of symlinks
 # (eg FreeBSD returns the mod time of the symlink's containing
 # directory).
 if (
-   am_has_slept=no
-   for am_try in 1 2; do
-     echo "timestamp, slept: $am_has_slept" > conftest.file
-     set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
-     if test "$[*]" = "X"; then
-	# -L didn't work.
-	set X `ls -t "$srcdir/configure" conftest.file`
-     fi
-     if test "$[*]" != "X $srcdir/configure conftest.file" \
-	&& test "$[*]" != "X conftest.file $srcdir/configure"; then
+   set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
+   if test "$[*]" = "X"; then
+      # -L didn't work.
+      set X `ls -t "$srcdir/configure" conftest.file`
+   fi
+   rm -f conftest.file
+   if test "$[*]" != "X $srcdir/configure conftest.file" \
+      && test "$[*]" != "X conftest.file $srcdir/configure"; then
+
+      # If neither matched, then we have a broken ls.  This can happen
+      # if, for instance, CONFIG_SHELL is bash and it inherits a
+      # broken ls alias from the environment.  This has actually
+      # happened.  Such a system could not be considered "sane".
+      AC_MSG_ERROR([ls -t appears to fail.  Make sure there is not a broken
+alias in your environment])
+   fi
 
-	# If neither matched, then we have a broken ls.  This can happen
-	# if, for instance, CONFIG_SHELL is bash and it inherits a
-	# broken ls alias from the environment.  This has actually
-	# happened.  Such a system could not be considered "sane".
-	AC_MSG_ERROR([ls -t appears to fail.  Make sure there is not a broken
-  alias in your environment])
-     fi
-     if test "$[2]" = conftest.file || test $am_try -eq 2; then
-       break
-     fi
-     # Just in case.
-     sleep 1
-     am_has_slept=yes
-   done
    test "$[2]" = conftest.file
    )
 then
@@ -827,85 +804,36 @@ else
    AC_MSG_ERROR([newly created file is older than distributed files!
 Check your system clock])
 fi
-AC_MSG_RESULT([yes])
-# If we didn't sleep, we still need to ensure time stamps of config.status and
-# generated files are strictly newer.
-am_sleep_pid=
-if grep 'slept: no' conftest.file >/dev/null 2>&1; then
-  ( sleep 1 ) &
-  am_sleep_pid=$!
-fi
-AC_CONFIG_COMMANDS_PRE(
-  [AC_MSG_CHECKING([that generated files are newer than configure])
-   if test -n "$am_sleep_pid"; then
-     # Hide warnings about reused PIDs.
-     wait $am_sleep_pid 2>/dev/null
-   fi
-   AC_MSG_RESULT([done])])
-rm -f conftest.file
-])
+AC_MSG_RESULT(yes)])
 
-# Copyright (C) 2009-2013 Free Software Foundation, Inc.
+# Copyright (C) 2009  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 1
+
 # AM_SILENT_RULES([DEFAULT])
 # --------------------------
 # Enable less verbose build rules; with the default set to DEFAULT
-# ("yes" being less verbose, "no" or empty being verbose).
+# (`yes' being less verbose, `no' or empty being verbose).
 AC_DEFUN([AM_SILENT_RULES],
-[AC_ARG_ENABLE([silent-rules], [dnl
-AS_HELP_STRING(
-  [--enable-silent-rules],
-  [less verbose build output (undo: "make V=1")])
-AS_HELP_STRING(
-  [--disable-silent-rules],
-  [verbose build output (undo: "make V=0")])dnl
-])
-case $enable_silent_rules in @%:@ (((
-  yes) AM_DEFAULT_VERBOSITY=0;;
-   no) AM_DEFAULT_VERBOSITY=1;;
-    *) AM_DEFAULT_VERBOSITY=m4_if([$1], [yes], [0], [1]);;
+[AC_ARG_ENABLE([silent-rules],
+[  --enable-silent-rules          less verbose build output (undo: `make V=1')
+  --disable-silent-rules         verbose build output (undo: `make V=0')])
+case $enable_silent_rules in
+yes) AM_DEFAULT_VERBOSITY=0;;
+no)  AM_DEFAULT_VERBOSITY=1;;
+*)   AM_DEFAULT_VERBOSITY=m4_if([$1], [yes], [0], [1]);;
 esac
-dnl
-dnl A few 'make' implementations (e.g., NonStop OS and NextStep)
-dnl do not support nested variable expansions.
-dnl See automake bug#9928 and bug#10237.
-am_make=${MAKE-make}
-AC_CACHE_CHECK([whether $am_make supports nested variables],
-   [am_cv_make_support_nested_variables],
-   [if AS_ECHO([['TRUE=$(BAR$(V))
-BAR0=false
-BAR1=true
-V=1
-am__doit:
-	@$(TRUE)
-.PHONY: am__doit']]) | $am_make -f - >/dev/null 2>&1; then
-  am_cv_make_support_nested_variables=yes
-else
-  am_cv_make_support_nested_variables=no
-fi])
-if test $am_cv_make_support_nested_variables = yes; then
-  dnl Using '$V' instead of '$(V)' breaks IRIX make.
-  AM_V='$(V)'
-  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
-else
-  AM_V=$AM_DEFAULT_VERBOSITY
-  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
-fi
-AC_SUBST([AM_V])dnl
-AM_SUBST_NOTMAKE([AM_V])dnl
-AC_SUBST([AM_DEFAULT_V])dnl
-AM_SUBST_NOTMAKE([AM_DEFAULT_V])dnl
 AC_SUBST([AM_DEFAULT_VERBOSITY])dnl
 AM_BACKSLASH='\'
 AC_SUBST([AM_BACKSLASH])dnl
 _AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
 ])
 
-# Copyright (C) 2001-2013 Free Software Foundation, Inc.
+# Copyright (C) 2001, 2003, 2005  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -913,32 +841,34 @@ _AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
 
 # AM_PROG_INSTALL_STRIP
 # ---------------------
-# One issue with vendor 'install' (even GNU) is that you can't
+# One issue with vendor `install' (even GNU) is that you can't
 # specify the program used to strip binaries.  This is especially
 # annoying in cross-compiling environments, where the build's strip
 # is unlikely to handle the host's binaries.
 # Fortunately install-sh will honor a STRIPPROG variable, so we
-# always use install-sh in "make install-strip", and initialize
+# always use install-sh in `make install-strip', and initialize
 # STRIPPROG with the value of the STRIP variable (set by the user).
 AC_DEFUN([AM_PROG_INSTALL_STRIP],
 [AC_REQUIRE([AM_PROG_INSTALL_SH])dnl
-# Installed binaries are usually stripped using 'strip' when the user
-# run "make install-strip".  However 'strip' might not be the right
+# Installed binaries are usually stripped using `strip' when the user
+# run `make install-strip'.  However `strip' might not be the right
 # tool to use in cross-compilation environments, therefore Automake
-# will honor the 'STRIP' environment variable to overrule this program.
-dnl Don't test for $cross_compiling = yes, because it might be 'maybe'.
+# will honor the `STRIP' environment variable to overrule this program.
+dnl Don't test for $cross_compiling = yes, because it might be `maybe'.
 if test "$cross_compiling" != no; then
   AC_CHECK_TOOL([STRIP], [strip], :)
 fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 AC_SUBST([INSTALL_STRIP_PROGRAM])])
 
-# Copyright (C) 2006-2013 Free Software Foundation, Inc.
+# Copyright (C) 2006, 2008  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 2
+
 # _AM_SUBST_NOTMAKE(VARIABLE)
 # ---------------------------
 # Prevent Automake from outputting VARIABLE = @VARIABLE@ in Makefile.in.
@@ -946,22 +876,24 @@ AC_SUBST([INSTALL_STRIP_PROGRAM])])
 AC_DEFUN([_AM_SUBST_NOTMAKE])
 
 # AM_SUBST_NOTMAKE(VARIABLE)
-# --------------------------
+# ---------------------------
 # Public sister of _AM_SUBST_NOTMAKE.
 AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 
 # Check how to create a tarball.                            -*- Autoconf -*-
 
-# Copyright (C) 2004-2013 Free Software Foundation, Inc.
+# Copyright (C) 2004, 2005  Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
 # with or without modifications, as long as this notice is preserved.
 
+# serial 2
+
 # _AM_PROG_TAR(FORMAT)
 # --------------------
 # Check how to create a tarball in format FORMAT.
-# FORMAT should be one of 'v7', 'ustar', or 'pax'.
+# FORMAT should be one of `v7', `ustar', or `pax'.
 #
 # Substitute a variable $(am__tar) that is a command
 # writing to stdout a FORMAT-tarball containing the directory
@@ -971,114 +903,75 @@ AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
 # Substitute a variable $(am__untar) that extract such
 # a tarball read from stdin.
 #     $(am__untar) < result.tar
-#
 AC_DEFUN([_AM_PROG_TAR],
-[# Always define AMTAR for backward compatibility.  Yes, it's still used
-# in the wild :-(  We should find a proper way to deprecate it ...
-AC_SUBST([AMTAR], ['$${TAR-tar}'])
-
-# We'll loop over all known methods to create a tar archive until one works.
-_am_tools='gnutar m4_if([$1], [ustar], [plaintar]) pax cpio none'
-
+[# Always define AMTAR for backward compatibility.
+AM_MISSING_PROG([AMTAR], [tar])
 m4_if([$1], [v7],
-  [am__tar='$${TAR-tar} chof - "$$tardir"' am__untar='$${TAR-tar} xf -'],
+     [am__tar='${AMTAR} chof - "$$tardir"'; am__untar='${AMTAR} xf -'],
+     [m4_case([$1], [ustar],, [pax],,
+              [m4_fatal([Unknown tar format])])
+AC_MSG_CHECKING([how to create a $1 tar archive])
+# Loop over all known methods to create a tar archive until one works.
+_am_tools='gnutar m4_if([$1], [ustar], [plaintar]) pax cpio none'
+_am_tools=${am_cv_prog_tar_$1-$_am_tools}
+# Do not fold the above two line into one, because Tru64 sh and
+# Solaris sh will not grok spaces in the rhs of `-'.
+for _am_tool in $_am_tools
+do
+  case $_am_tool in
+  gnutar)
+    for _am_tar in tar gnutar gtar;
+    do
+      AM_RUN_LOG([$_am_tar --version]) && break
+    done
+    am__tar="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$$tardir"'
+    am__tar_="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$tardir"'
+    am__untar="$_am_tar -xf -"
+    ;;
+  plaintar)
+    # Must skip GNU tar: if it does not support --format= it doesn't create
+    # ustar tarball either.
+    (tar --version) >/dev/null 2>&1 && continue
+    am__tar='tar chf - "$$tardir"'
+    am__tar_='tar chf - "$tardir"'
+    am__untar='tar xf -'
+    ;;
+  pax)
+    am__tar='pax -L -x $1 -w "$$tardir"'
+    am__tar_='pax -L -x $1 -w "$tardir"'
+    am__untar='pax -r'
+    ;;
+  cpio)
+    am__tar='find "$$tardir" -print | cpio -o -H $1 -L'
+    am__tar_='find "$tardir" -print | cpio -o -H $1 -L'
+    am__untar='cpio -i -H $1 -d'
+    ;;
+  none)
+    am__tar=false
+    am__tar_=false
+    am__untar=false
+    ;;
+  esac
 
-  [m4_case([$1],
-    [ustar],
-     [# The POSIX 1988 'ustar' format is defined with fixed-size fields.
-      # There is notably a 21 bits limit for the UID and the GID.  In fact,
-      # the 'pax' utility can hang on bigger UID/GID (see automake bug#8343
-      # and bug#13588).
-      am_max_uid=2097151 # 2^21 - 1
-      am_max_gid=$am_max_uid
-      # The $UID and $GID variables are not portable, so we need to resort
-      # to the POSIX-mandated id(1) utility.  Errors in the 'id' calls
-      # below are definitely unexpected, so allow the users to see them
-      # (that is, avoid stderr redirection).
-      am_uid=`id -u || echo unknown`
-      am_gid=`id -g || echo unknown`
-      AC_MSG_CHECKING([whether UID '$am_uid' is supported by ustar format])
-      if test $am_uid -le $am_max_uid; then
-         AC_MSG_RESULT([yes])
-      else
-         AC_MSG_RESULT([no])
-         _am_tools=none
-      fi
-      AC_MSG_CHECKING([whether GID '$am_gid' is supported by ustar format])
-      if test $am_gid -le $am_max_gid; then
-         AC_MSG_RESULT([yes])
-      else
-        AC_MSG_RESULT([no])
-        _am_tools=none
-      fi],
+  # If the value was cached, stop now.  We just wanted to have am__tar
+  # and am__untar set.
+  test -n "${am_cv_prog_tar_$1}" && break
 
-  [pax],
-    [],
-
-  [m4_fatal([Unknown tar format])])
-
-  AC_MSG_CHECKING([how to create a $1 tar archive])
-
-  # Go ahead even if we have the value already cached.  We do so because we
-  # need to set the values for the 'am__tar' and 'am__untar' variables.
-  _am_tools=${am_cv_prog_tar_$1-$_am_tools}
-
-  for _am_tool in $_am_tools; do
-    case $_am_tool in
-    gnutar)
-      for _am_tar in tar gnutar gtar; do
-        AM_RUN_LOG([$_am_tar --version]) && break
-      done
-      am__tar="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$$tardir"'
-      am__tar_="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$tardir"'
-      am__untar="$_am_tar -xf -"
-      ;;
-    plaintar)
-      # Must skip GNU tar: if it does not support --format= it doesn't create
-      # ustar tarball either.
-      (tar --version) >/dev/null 2>&1 && continue
-      am__tar='tar chf - "$$tardir"'
-      am__tar_='tar chf - "$tardir"'
-      am__untar='tar xf -'
-      ;;
-    pax)
-      am__tar='pax -L -x $1 -w "$$tardir"'
-      am__tar_='pax -L -x $1 -w "$tardir"'
-      am__untar='pax -r'
-      ;;
-    cpio)
-      am__tar='find "$$tardir" -print | cpio -o -H $1 -L'
-      am__tar_='find "$tardir" -print | cpio -o -H $1 -L'
-      am__untar='cpio -i -H $1 -d'
-      ;;
-    none)
-      am__tar=false
-      am__tar_=false
-      am__untar=false
-      ;;
-    esac
-
-    # If the value was cached, stop now.  We just wanted to have am__tar
-    # and am__untar set.
-    test -n "${am_cv_prog_tar_$1}" && break
-
-    # tar/untar a dummy directory, and stop if the command works.
-    rm -rf conftest.dir
-    mkdir conftest.dir
-    echo GrepMe > conftest.dir/file
-    AM_RUN_LOG([tardir=conftest.dir && eval $am__tar_ >conftest.tar])
-    rm -rf conftest.dir
-    if test -s conftest.tar; then
-      AM_RUN_LOG([$am__untar <conftest.tar])
-      AM_RUN_LOG([cat conftest.dir/file])
-      grep GrepMe conftest.dir/file >/dev/null 2>&1 && break
-    fi
-  done
+  # tar/untar a dummy directory, and stop if the command works
   rm -rf conftest.dir
+  mkdir conftest.dir
+  echo GrepMe > conftest.dir/file
+  AM_RUN_LOG([tardir=conftest.dir && eval $am__tar_ >conftest.tar])
+  rm -rf conftest.dir
+  if test -s conftest.tar; then
+    AM_RUN_LOG([$am__untar <conftest.tar])
+    grep GrepMe conftest.dir/file >/dev/null 2>&1 && break
+  fi
+done
+rm -rf conftest.dir
 
-  AC_CACHE_VAL([am_cv_prog_tar_$1], [am_cv_prog_tar_$1=$_am_tool])
-  AC_MSG_RESULT([$am_cv_prog_tar_$1])])
-
+AC_CACHE_VAL([am_cv_prog_tar_$1], [am_cv_prog_tar_$1=$_am_tool])
+AC_MSG_RESULT([$am_cv_prog_tar_$1])])
 AC_SUBST([am__tar])
 AC_SUBST([am__untar])
 ]) # _AM_PROG_TAR
diff --git a/config.guess b/config.guess
index 5f6aa02d..405bc323 120000
--- a/config.guess
+++ b/config.guess
@@ -1 +1 @@
-/usr/share/automake-1.14/config.guess
\ No newline at end of file
+/usr/share/automake-1.11/config.guess
\ No newline at end of file
diff --git a/config.sub b/config.sub
index 0abfe18c..4d47fbcb 120000
--- a/config.sub
+++ b/config.sub
@@ -1 +1 @@
-/usr/share/automake-1.14/config.sub
\ No newline at end of file
+/usr/share/automake-1.11/config.sub
\ No newline at end of file
diff --git a/configure b/configure
index ccfea6b7..e4dd388e 100755
--- a/configure
+++ b/configure
@@ -1,22 +1,20 @@
 #! /bin/sh
 # Guess values for system-dependent variables and create Makefiles.
-# Generated by GNU Autoconf 2.69 for Grid 1.0.
+# Generated by GNU Autoconf 2.63 for Grid 1.0.
 #
 # Report bugs to <paboyle@ph.ed.ac.uk>.
 #
-#
-# Copyright (C) 1992-1996, 1998-2012 Free Software Foundation, Inc.
-#
-#
+# Copyright (C) 1992, 1993, 1994, 1995, 1996, 1998, 1999, 2000, 2001,
+# 2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
 # This configure script is free software; the Free Software Foundation
 # gives unlimited permission to copy, distribute and modify it.
-## -------------------- ##
-## M4sh Initialization. ##
-## -------------------- ##
+## --------------------- ##
+## M4sh Initialization.  ##
+## --------------------- ##
 
 # Be more Bourne compatible
 DUALCASE=1; export DUALCASE # for MKS sh
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
   emulate sh
   NULLCMD=:
   # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
@@ -24,15 +22,23 @@ if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
   alias -g '${1+"$@"}'='"$@"'
   setopt NO_GLOB_SUBST
 else
-  case `(set -o) 2>/dev/null` in #(
-  *posix*) :
-    set -o posix ;; #(
-  *) :
-     ;;
+  case `(set -o) 2>/dev/null` in
+  *posix*) set -o posix ;;
 esac
+
 fi
 
 
+
+
+# PATH needs CR
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
+
 as_nl='
 '
 export as_nl
@@ -40,13 +46,7 @@ export as_nl
 as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
-# Prefer a ksh shell builtin over an external printf program on Solaris,
-# but without wasting forks for bash or zsh.
-if test -z "$BASH_VERSION$ZSH_VERSION" \
-    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
-  as_echo='print -r --'
-  as_echo_n='print -rn --'
-elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+if (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
   as_echo='printf %s\n'
   as_echo_n='printf %s'
 else
@@ -57,7 +57,7 @@ else
     as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
     as_echo_n_body='eval
       arg=$1;
-      case $arg in #(
+      case $arg in
       *"$as_nl"*)
 	expr "X$arg" : "X\\(.*\\)$as_nl";
 	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
@@ -80,6 +80,13 @@ if test "${PATH_SEPARATOR+set}" != set; then
   }
 fi
 
+# Support unset when possible.
+if ( (MAIL=60; unset MAIL) || exit) >/dev/null 2>&1; then
+  as_unset=unset
+else
+  as_unset=false
+fi
+
 
 # IFS
 # We need space, tab and new line, in precisely that order.  Quoting is
@@ -89,16 +96,15 @@ fi
 IFS=" ""	$as_nl"
 
 # Find who we are.  Look in the path if we contain no directory separator.
-as_myself=
-case $0 in #((
+case $0 in
   *[\\/]* ) as_myself=$0 ;;
   *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
 for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
-  done
+  test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+done
 IFS=$as_save_IFS
 
      ;;
@@ -110,16 +116,12 @@ if test "x$as_myself" = x; then
 fi
 if test ! -f "$as_myself"; then
   $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
-  exit 1
+  { (exit 1); exit 1; }
 fi
 
-# Unset variables that we do not need and which cause bugs (e.g. in
-# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
-# suppresses any "Segmentation fault" message there.  '((' could
-# trigger a bug in pdksh 5.2.14.
-for as_var in BASH_ENV ENV MAIL MAILPATH
-do eval test x\${$as_var+set} = xset \
-  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
+# Work around bugs in pre-3.0 UWIN ksh.
+for as_var in ENV MAIL MAILPATH
+do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
 done
 PS1='$ '
 PS2='> '
@@ -131,294 +133,7 @@ export LC_ALL
 LANGUAGE=C
 export LANGUAGE
 
-# CDPATH.
-(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
-
-# Use a proper internal environment variable to ensure we don't fall
-  # into an infinite loop, continuously re-executing ourselves.
-  if test x"${_as_can_reexec}" != xno && test "x$CONFIG_SHELL" != x; then
-    _as_can_reexec=no; export _as_can_reexec;
-    # We cannot yet assume a decent shell, so we have to provide a
-# neutralization value for shells without unset; and this also
-# works around shells that cannot unset nonexistent variables.
-# Preserve -v and -x to the replacement shell.
-BASH_ENV=/dev/null
-ENV=/dev/null
-(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
-case $- in # ((((
-  *v*x* | *x*v* ) as_opts=-vx ;;
-  *v* ) as_opts=-v ;;
-  *x* ) as_opts=-x ;;
-  * ) as_opts= ;;
-esac
-exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
-# Admittedly, this is quite paranoid, since all the known shells bail
-# out after a failed `exec'.
-$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
-as_fn_exit 255
-  fi
-  # We don't want this to propagate to other subprocesses.
-          { _as_can_reexec=; unset _as_can_reexec;}
-if test "x$CONFIG_SHELL" = x; then
-  as_bourne_compatible="if test -n \"\${ZSH_VERSION+set}\" && (emulate sh) >/dev/null 2>&1; then :
-  emulate sh
-  NULLCMD=:
-  # Pre-4.2 versions of Zsh do word splitting on \${1+\"\$@\"}, which
-  # is contrary to our usage.  Disable this feature.
-  alias -g '\${1+\"\$@\"}'='\"\$@\"'
-  setopt NO_GLOB_SUBST
-else
-  case \`(set -o) 2>/dev/null\` in #(
-  *posix*) :
-    set -o posix ;; #(
-  *) :
-     ;;
-esac
-fi
-"
-  as_required="as_fn_return () { (exit \$1); }
-as_fn_success () { as_fn_return 0; }
-as_fn_failure () { as_fn_return 1; }
-as_fn_ret_success () { return 0; }
-as_fn_ret_failure () { return 1; }
-
-exitcode=0
-as_fn_success || { exitcode=1; echo as_fn_success failed.; }
-as_fn_failure && { exitcode=1; echo as_fn_failure succeeded.; }
-as_fn_ret_success || { exitcode=1; echo as_fn_ret_success failed.; }
-as_fn_ret_failure && { exitcode=1; echo as_fn_ret_failure succeeded.; }
-if ( set x; as_fn_ret_success y && test x = \"\$1\" ); then :
-
-else
-  exitcode=1; echo positional parameters were not saved.
-fi
-test x\$exitcode = x0 || exit 1
-test -x / || exit 1"
-  as_suggested="  as_lineno_1=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_1a=\$LINENO
-  as_lineno_2=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_2a=\$LINENO
-  eval 'test \"x\$as_lineno_1'\$as_run'\" != \"x\$as_lineno_2'\$as_run'\" &&
-  test \"x\`expr \$as_lineno_1'\$as_run' + 1\`\" = \"x\$as_lineno_2'\$as_run'\"' || exit 1
-test \$(( 1 + 1 )) = 2 || exit 1"
-  if (eval "$as_required") 2>/dev/null; then :
-  as_have_required=yes
-else
-  as_have_required=no
-fi
-  if test x$as_have_required = xyes && (eval "$as_suggested") 2>/dev/null; then :
-
-else
-  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-as_found=false
-for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-  as_found=:
-  case $as_dir in #(
-	 /*)
-	   for as_base in sh bash ksh sh5; do
-	     # Try only shells that exist, to save several forks.
-	     as_shell=$as_dir/$as_base
-	     if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
-		    { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$as_shell"; } 2>/dev/null; then :
-  CONFIG_SHELL=$as_shell as_have_required=yes
-		   if { $as_echo "$as_bourne_compatible""$as_suggested" | as_run=a "$as_shell"; } 2>/dev/null; then :
-  break 2
-fi
-fi
-	   done;;
-       esac
-  as_found=false
-done
-$as_found || { if { test -f "$SHELL" || test -f "$SHELL.exe"; } &&
-	      { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$SHELL"; } 2>/dev/null; then :
-  CONFIG_SHELL=$SHELL as_have_required=yes
-fi; }
-IFS=$as_save_IFS
-
-
-      if test "x$CONFIG_SHELL" != x; then :
-  export CONFIG_SHELL
-             # We cannot yet assume a decent shell, so we have to provide a
-# neutralization value for shells without unset; and this also
-# works around shells that cannot unset nonexistent variables.
-# Preserve -v and -x to the replacement shell.
-BASH_ENV=/dev/null
-ENV=/dev/null
-(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
-case $- in # ((((
-  *v*x* | *x*v* ) as_opts=-vx ;;
-  *v* ) as_opts=-v ;;
-  *x* ) as_opts=-x ;;
-  * ) as_opts= ;;
-esac
-exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
-# Admittedly, this is quite paranoid, since all the known shells bail
-# out after a failed `exec'.
-$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
-exit 255
-fi
-
-    if test x$as_have_required = xno; then :
-  $as_echo "$0: This script requires a shell more modern than all"
-  $as_echo "$0: the shells that I found on your system."
-  if test x${ZSH_VERSION+set} = xset ; then
-    $as_echo "$0: In particular, zsh $ZSH_VERSION has bugs and should"
-    $as_echo "$0: be upgraded to zsh 4.3.4 or later."
-  else
-    $as_echo "$0: Please tell bug-autoconf@gnu.org and
-$0: paboyle@ph.ed.ac.uk about your system, including any
-$0: error possibly output before this message. Then install
-$0: a modern shell, or manually run the script under such a
-$0: shell if you do have one."
-  fi
-  exit 1
-fi
-fi
-fi
-SHELL=${CONFIG_SHELL-/bin/sh}
-export SHELL
-# Unset more variables known to interfere with behavior of common tools.
-CLICOLOR_FORCE= GREP_OPTIONS=
-unset CLICOLOR_FORCE GREP_OPTIONS
-
-## --------------------- ##
-## M4sh Shell Functions. ##
-## --------------------- ##
-# as_fn_unset VAR
-# ---------------
-# Portably unset VAR.
-as_fn_unset ()
-{
-  { eval $1=; unset $1;}
-}
-as_unset=as_fn_unset
-
-# as_fn_set_status STATUS
-# -----------------------
-# Set $? to STATUS, without forking.
-as_fn_set_status ()
-{
-  return $1
-} # as_fn_set_status
-
-# as_fn_exit STATUS
-# -----------------
-# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
-as_fn_exit ()
-{
-  set +e
-  as_fn_set_status $1
-  exit $1
-} # as_fn_exit
-
-# as_fn_mkdir_p
-# -------------
-# Create "$as_dir" as a directory, including parents if necessary.
-as_fn_mkdir_p ()
-{
-
-  case $as_dir in #(
-  -*) as_dir=./$as_dir;;
-  esac
-  test -d "$as_dir" || eval $as_mkdir_p || {
-    as_dirs=
-    while :; do
-      case $as_dir in #(
-      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
-      *) as_qdir=$as_dir;;
-      esac
-      as_dirs="'$as_qdir' $as_dirs"
-      as_dir=`$as_dirname -- "$as_dir" ||
-$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_dir" : 'X\(//\)[^/]' \| \
-	 X"$as_dir" : 'X\(//\)$' \| \
-	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_dir" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      test -d "$as_dir" && break
-    done
-    test -z "$as_dirs" || eval "mkdir $as_dirs"
-  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
-
-
-} # as_fn_mkdir_p
-
-# as_fn_executable_p FILE
-# -----------------------
-# Test if FILE is an executable regular file.
-as_fn_executable_p ()
-{
-  test -f "$1" && test -x "$1"
-} # as_fn_executable_p
-# as_fn_append VAR VALUE
-# ----------------------
-# Append the text in VALUE to the end of the definition contained in VAR. Take
-# advantage of any shell optimizations that allow amortized linear growth over
-# repeated appends, instead of the typical quadratic growth present in naive
-# implementations.
-if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
-  eval 'as_fn_append ()
-  {
-    eval $1+=\$2
-  }'
-else
-  as_fn_append ()
-  {
-    eval $1=\$$1\$2
-  }
-fi # as_fn_append
-
-# as_fn_arith ARG...
-# ------------------
-# Perform arithmetic evaluation on the ARGs, and store the result in the
-# global $as_val. Take advantage of shells that can avoid forks. The arguments
-# must be portable across $(()) and expr.
-if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
-  eval 'as_fn_arith ()
-  {
-    as_val=$(( $* ))
-  }'
-else
-  as_fn_arith ()
-  {
-    as_val=`expr "$@" || test $? -eq 1`
-  }
-fi # as_fn_arith
-
-
-# as_fn_error STATUS ERROR [LINENO LOG_FD]
-# ----------------------------------------
-# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
-# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
-# script with STATUS, using 1 if that was 0.
-as_fn_error ()
-{
-  as_status=$1; test $as_status -eq 0 && as_status=1
-  if test "$4"; then
-    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
-  fi
-  $as_echo "$as_me: error: $2" >&2
-  as_fn_exit $as_status
-} # as_fn_error
-
+# Required to use basename.
 if expr a : '\(a\)' >/dev/null 2>&1 &&
    test "X`expr 00001 : '.*\(...\)'`" = X001; then
   as_expr=expr
@@ -432,12 +147,8 @@ else
   as_basename=false
 fi
 
-if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
-  as_dirname=dirname
-else
-  as_dirname=false
-fi
 
+# Name of the executable.
 as_me=`$as_basename -- "$0" ||
 $as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
 	 X"$0" : 'X\(//\)$' \| \
@@ -457,19 +168,295 @@ $as_echo X/"$0" |
 	  }
 	  s/.*/./; q'`
 
-# Avoid depending upon Character Ranges.
-as_cr_letters='abcdefghijklmnopqrstuvwxyz'
-as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
-as_cr_Letters=$as_cr_letters$as_cr_LETTERS
-as_cr_digits='0123456789'
-as_cr_alnum=$as_cr_Letters$as_cr_digits
+# CDPATH.
+$as_unset CDPATH
 
 
-  as_lineno_1=$LINENO as_lineno_1a=$LINENO
-  as_lineno_2=$LINENO as_lineno_2a=$LINENO
-  eval 'test "x$as_lineno_1'$as_run'" != "x$as_lineno_2'$as_run'" &&
-  test "x`expr $as_lineno_1'$as_run' + 1`" = "x$as_lineno_2'$as_run'"' || {
-  # Blame Lee E. McMahon (1931-1989) for sed's syntax.  :-)
+if test "x$CONFIG_SHELL" = x; then
+  if (eval ":") 2>/dev/null; then
+  as_have_required=yes
+else
+  as_have_required=no
+fi
+
+  if test $as_have_required = yes &&	 (eval ":
+(as_func_return () {
+  (exit \$1)
+}
+as_func_success () {
+  as_func_return 0
+}
+as_func_failure () {
+  as_func_return 1
+}
+as_func_ret_success () {
+  return 0
+}
+as_func_ret_failure () {
+  return 1
+}
+
+exitcode=0
+if as_func_success; then
+  :
+else
+  exitcode=1
+  echo as_func_success failed.
+fi
+
+if as_func_failure; then
+  exitcode=1
+  echo as_func_failure succeeded.
+fi
+
+if as_func_ret_success; then
+  :
+else
+  exitcode=1
+  echo as_func_ret_success failed.
+fi
+
+if as_func_ret_failure; then
+  exitcode=1
+  echo as_func_ret_failure succeeded.
+fi
+
+if ( set x; as_func_ret_success y && test x = \"\$1\" ); then
+  :
+else
+  exitcode=1
+  echo positional parameters were not saved.
+fi
+
+test \$exitcode = 0) || { (exit 1); exit 1; }
+
+(
+  as_lineno_1=\$LINENO
+  as_lineno_2=\$LINENO
+  test \"x\$as_lineno_1\" != \"x\$as_lineno_2\" &&
+  test \"x\`expr \$as_lineno_1 + 1\`\" = \"x\$as_lineno_2\") || { (exit 1); exit 1; }
+") 2> /dev/null; then
+  :
+else
+  as_candidate_shells=
+    as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+  case $as_dir in
+	 /*)
+	   for as_base in sh bash ksh sh5; do
+	     as_candidate_shells="$as_candidate_shells $as_dir/$as_base"
+	   done;;
+       esac
+done
+IFS=$as_save_IFS
+
+
+      for as_shell in $as_candidate_shells $SHELL; do
+	 # Try only shells that exist, to save several forks.
+	 if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
+		{ ("$as_shell") 2> /dev/null <<\_ASEOF
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
+  emulate sh
+  NULLCMD=:
+  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '${1+"$@"}'='"$@"'
+  setopt NO_GLOB_SUBST
+else
+  case `(set -o) 2>/dev/null` in
+  *posix*) set -o posix ;;
+esac
+
+fi
+
+
+:
+_ASEOF
+}; then
+  CONFIG_SHELL=$as_shell
+	       as_have_required=yes
+	       if { "$as_shell" 2> /dev/null <<\_ASEOF
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
+  emulate sh
+  NULLCMD=:
+  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '${1+"$@"}'='"$@"'
+  setopt NO_GLOB_SUBST
+else
+  case `(set -o) 2>/dev/null` in
+  *posix*) set -o posix ;;
+esac
+
+fi
+
+
+:
+(as_func_return () {
+  (exit $1)
+}
+as_func_success () {
+  as_func_return 0
+}
+as_func_failure () {
+  as_func_return 1
+}
+as_func_ret_success () {
+  return 0
+}
+as_func_ret_failure () {
+  return 1
+}
+
+exitcode=0
+if as_func_success; then
+  :
+else
+  exitcode=1
+  echo as_func_success failed.
+fi
+
+if as_func_failure; then
+  exitcode=1
+  echo as_func_failure succeeded.
+fi
+
+if as_func_ret_success; then
+  :
+else
+  exitcode=1
+  echo as_func_ret_success failed.
+fi
+
+if as_func_ret_failure; then
+  exitcode=1
+  echo as_func_ret_failure succeeded.
+fi
+
+if ( set x; as_func_ret_success y && test x = "$1" ); then
+  :
+else
+  exitcode=1
+  echo positional parameters were not saved.
+fi
+
+test $exitcode = 0) || { (exit 1); exit 1; }
+
+(
+  as_lineno_1=$LINENO
+  as_lineno_2=$LINENO
+  test "x$as_lineno_1" != "x$as_lineno_2" &&
+  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2") || { (exit 1); exit 1; }
+
+_ASEOF
+}; then
+  break
+fi
+
+fi
+
+      done
+
+      if test "x$CONFIG_SHELL" != x; then
+  for as_var in BASH_ENV ENV
+	do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
+	done
+	export CONFIG_SHELL
+	exec "$CONFIG_SHELL" "$as_myself" ${1+"$@"}
+fi
+
+
+    if test $as_have_required = no; then
+  echo This script requires a shell more modern than all the
+      echo shells that I found on your system.  Please install a
+      echo modern shell, or manually run the script under such a
+      echo shell if you do have one.
+      { (exit 1); exit 1; }
+fi
+
+
+fi
+
+fi
+
+
+
+(eval "as_func_return () {
+  (exit \$1)
+}
+as_func_success () {
+  as_func_return 0
+}
+as_func_failure () {
+  as_func_return 1
+}
+as_func_ret_success () {
+  return 0
+}
+as_func_ret_failure () {
+  return 1
+}
+
+exitcode=0
+if as_func_success; then
+  :
+else
+  exitcode=1
+  echo as_func_success failed.
+fi
+
+if as_func_failure; then
+  exitcode=1
+  echo as_func_failure succeeded.
+fi
+
+if as_func_ret_success; then
+  :
+else
+  exitcode=1
+  echo as_func_ret_success failed.
+fi
+
+if as_func_ret_failure; then
+  exitcode=1
+  echo as_func_ret_failure succeeded.
+fi
+
+if ( set x; as_func_ret_success y && test x = \"\$1\" ); then
+  :
+else
+  exitcode=1
+  echo positional parameters were not saved.
+fi
+
+test \$exitcode = 0") || {
+  echo No shell found that supports shell functions.
+  echo Please tell bug-autoconf@gnu.org about your system,
+  echo including any error possibly output before this message.
+  echo This can help us improve future autoconf versions.
+  echo Configuration will now proceed without shell functions.
+}
+
+
+
+  as_lineno_1=$LINENO
+  as_lineno_2=$LINENO
+  test "x$as_lineno_1" != "x$as_lineno_2" &&
+  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2" || {
+
+  # Create $as_me.lineno as a copy of $as_myself, but with $LINENO
+  # uniformly replaced by the line number.  The first 'sed' inserts a
+  # line-number line after each line using $LINENO; the second 'sed'
+  # does the real work.  The second script uses 'N' to pair each
+  # line-number line with the line containing $LINENO, and appends
+  # trailing '-' during substitution so that $LINENO is not a special
+  # case at line end.
+  # (Raja R Harinath suggested sed '=', and Paul Eggert wrote the
+  # scripts with optimization help from Paolo Bonzini.  Blame Lee
+  # E. McMahon (1931-1989) for sed's syntax.  :-)
   sed -n '
     p
     /[$]LINENO/=
@@ -486,12 +473,9 @@ as_cr_alnum=$as_cr_Letters$as_cr_digits
       s/-\n.*//
     ' >$as_me.lineno &&
   chmod +x "$as_me.lineno" ||
-    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2; as_fn_exit 1; }
+    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2
+   { (exit 1); exit 1; }; }
 
-  # If we had to re-execute with $CONFIG_SHELL, we're ensured to have
-  # already done that, so ensure we don't try to do so again and fall
-  # in an infinite loop.  This has already happened in practice.
-  _as_can_reexec=no; export _as_can_reexec
   # Don't try to exec as it changes $[0], causing all sort of problems
   # (the dirname of $[0] is not the place where we might find the
   # original and so on.  Autoconf is especially sensitive to this).
@@ -500,18 +484,29 @@ as_cr_alnum=$as_cr_Letters$as_cr_digits
   exit
 }
 
+
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
+
 ECHO_C= ECHO_N= ECHO_T=
-case `echo -n x` in #(((((
+case `echo -n x` in
 -n*)
-  case `echo 'xy\c'` in
+  case `echo 'x\c'` in
   *c*) ECHO_T='	';;	# ECHO_T is single tab character.
-  xy)  ECHO_C='\c';;
-  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
-       ECHO_T='	';;
+  *)   ECHO_C='\c';;
   esac;;
 *)
   ECHO_N='-n';;
 esac
+if expr a : '\(a\)' >/dev/null 2>&1 &&
+   test "X`expr 00001 : '.*\(...\)'`" = X001; then
+  as_expr=expr
+else
+  as_expr=false
+fi
 
 rm -f conf$$ conf$$.exe conf$$.file
 if test -d conf$$.dir; then
@@ -526,29 +521,49 @@ if (echo >conf$$.file) 2>/dev/null; then
     # ... but there are two gotchas:
     # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
     # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
-    # In both cases, we have to default to `cp -pR'.
+    # In both cases, we have to default to `cp -p'.
     ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
-      as_ln_s='cp -pR'
+      as_ln_s='cp -p'
   elif ln conf$$.file conf$$ 2>/dev/null; then
     as_ln_s=ln
   else
-    as_ln_s='cp -pR'
+    as_ln_s='cp -p'
   fi
 else
-  as_ln_s='cp -pR'
+  as_ln_s='cp -p'
 fi
 rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
 rmdir conf$$.dir 2>/dev/null
 
 if mkdir -p . 2>/dev/null; then
-  as_mkdir_p='mkdir -p "$as_dir"'
+  as_mkdir_p=:
 else
   test -d ./-p && rmdir ./-p
   as_mkdir_p=false
 fi
 
-as_test_x='test -x'
-as_executable_p=as_fn_executable_p
+if test -x / >/dev/null 2>&1; then
+  as_test_x='test -x'
+else
+  if ls -dL / >/dev/null 2>&1; then
+    as_ls_L_option=L
+  else
+    as_ls_L_option=
+  fi
+  as_test_x='
+    eval sh -c '\''
+      if test -d "$1"; then
+	test -d "$1/.";
+      else
+	case $1 in
+	-*)set "./$1";;
+	esac;
+	case `ls -ld'$as_ls_L_option' "$1" 2>/dev/null` in
+	???[sx]*):;;*)false;;esac;fi
+    '\'' sh
+  '
+fi
+as_executable_p=$as_test_x
 
 # Sed expression to map a string onto a valid CPP name.
 as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
@@ -557,11 +572,11 @@ as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
 as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
 
 
-test -n "$DJDIR" || exec 7<&0 </dev/null
-exec 6>&1
+
+exec 7<&0 </dev/null 6>&1
 
 # Name of the host.
-# hostname on some systems (SVR3.2, old GNU/Linux) returns a bogus exit status,
+# hostname on some systems (SVR3.2, Linux) returns a bogus exit status,
 # so uname gets run too.
 ac_hostname=`(hostname || uname -n) 2>/dev/null | sed 1q`
 
@@ -576,6 +591,7 @@ cross_compiling=no
 subdirs=
 MFLAGS=
 MAKEFLAGS=
+SHELL=${CONFIG_SHELL-/bin/sh}
 
 # Identity of this package.
 PACKAGE_NAME='Grid'
@@ -583,7 +599,6 @@ PACKAGE_TARNAME='grid'
 PACKAGE_VERSION='1.0'
 PACKAGE_STRING='Grid 1.0'
 PACKAGE_BUGREPORT='paboyle@ph.ed.ac.uk'
-PACKAGE_URL=''
 
 ac_unique_file="lib/Grid.h"
 # Factoring default headers for most tests.
@@ -646,7 +661,6 @@ OPENMP_CXXFLAGS
 am__fastdepCXX_FALSE
 am__fastdepCXX_TRUE
 CXXDEPMODE
-am__nodep
 AMDEPBACKSLASH
 AMDEP_FALSE
 AMDEP_TRUE
@@ -662,8 +676,6 @@ CXXFLAGS
 CXX
 AM_BACKSLASH
 AM_DEFAULT_VERBOSITY
-AM_DEFAULT_V
-AM_V
 am__untar
 am__tar
 AMTAR
@@ -729,7 +741,6 @@ bindir
 program_transform_name
 prefix
 exec_prefix
-PACKAGE_URL
 PACKAGE_BUGREPORT
 PACKAGE_STRING
 PACKAGE_VERSION
@@ -822,9 +833,8 @@ do
   fi
 
   case $ac_option in
-  *=?*) ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
-  *=)   ac_optarg= ;;
-  *)    ac_optarg=yes ;;
+  *=*)	ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
+  *)	ac_optarg=yes ;;
   esac
 
   # Accept the important Cygnus configure options, so we can diagnose typos.
@@ -869,7 +879,8 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*disable-\(.*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid feature name: $ac_useropt"
+      { $as_echo "$as_me: error: invalid feature name: $ac_useropt" >&2
+   { (exit 1); exit 1; }; }
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -895,7 +906,8 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*enable-\([^=]*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid feature name: $ac_useropt"
+      { $as_echo "$as_me: error: invalid feature name: $ac_useropt" >&2
+   { (exit 1); exit 1; }; }
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1099,7 +1111,8 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*with-\([^=]*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid package name: $ac_useropt"
+      { $as_echo "$as_me: error: invalid package name: $ac_useropt" >&2
+   { (exit 1); exit 1; }; }
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1115,7 +1128,8 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*without-\(.*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid package name: $ac_useropt"
+      { $as_echo "$as_me: error: invalid package name: $ac_useropt" >&2
+   { (exit 1); exit 1; }; }
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1145,17 +1159,17 @@ do
   | --x-librar=* | --x-libra=* | --x-libr=* | --x-lib=* | --x-li=* | --x-l=*)
     x_libraries=$ac_optarg ;;
 
-  -*) as_fn_error $? "unrecognized option: \`$ac_option'
-Try \`$0 --help' for more information"
+  -*) { $as_echo "$as_me: error: unrecognized option: $ac_option
+Try \`$0 --help' for more information." >&2
+   { (exit 1); exit 1; }; }
     ;;
 
   *=*)
     ac_envvar=`expr "x$ac_option" : 'x\([^=]*\)='`
     # Reject names that are not valid shell variable names.
-    case $ac_envvar in #(
-      '' | [0-9]* | *[!_$as_cr_alnum]* )
-      as_fn_error $? "invalid variable name: \`$ac_envvar'" ;;
-    esac
+    expr "x$ac_envvar" : ".*[^_$as_cr_alnum]" >/dev/null &&
+      { $as_echo "$as_me: error: invalid variable name: $ac_envvar" >&2
+   { (exit 1); exit 1; }; }
     eval $ac_envvar=\$ac_optarg
     export $ac_envvar ;;
 
@@ -1164,7 +1178,7 @@ Try \`$0 --help' for more information"
     $as_echo "$as_me: WARNING: you should use --build, --host, --target" >&2
     expr "x$ac_option" : ".*[^-._$as_cr_alnum]" >/dev/null &&
       $as_echo "$as_me: WARNING: invalid host type: $ac_option" >&2
-    : "${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}"
+    : ${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}
     ;;
 
   esac
@@ -1172,13 +1186,15 @@ done
 
 if test -n "$ac_prev"; then
   ac_option=--`echo $ac_prev | sed 's/_/-/g'`
-  as_fn_error $? "missing argument to $ac_option"
+  { $as_echo "$as_me: error: missing argument to $ac_option" >&2
+   { (exit 1); exit 1; }; }
 fi
 
 if test -n "$ac_unrecognized_opts"; then
   case $enable_option_checking in
     no) ;;
-    fatal) as_fn_error $? "unrecognized options: $ac_unrecognized_opts" ;;
+    fatal) { $as_echo "$as_me: error: unrecognized options: $ac_unrecognized_opts" >&2
+   { (exit 1); exit 1; }; } ;;
     *)     $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2 ;;
   esac
 fi
@@ -1201,7 +1217,8 @@ do
     [\\/$]* | ?:[\\/]* )  continue;;
     NONE | '' ) case $ac_var in *prefix ) continue;; esac;;
   esac
-  as_fn_error $? "expected an absolute directory name for --$ac_var: $ac_val"
+  { $as_echo "$as_me: error: expected an absolute directory name for --$ac_var: $ac_val" >&2
+   { (exit 1); exit 1; }; }
 done
 
 # There might be people who depend on the old broken behavior: `$host'
@@ -1215,6 +1232,8 @@ target=$target_alias
 if test "x$host_alias" != x; then
   if test "x$build_alias" = x; then
     cross_compiling=maybe
+    $as_echo "$as_me: WARNING: If you wanted to set the --build type, don't use --host.
+    If a cross compiler is detected then cross compile mode will be used." >&2
   elif test "x$build_alias" != "x$host_alias"; then
     cross_compiling=yes
   fi
@@ -1229,9 +1248,11 @@ test "$silent" = yes && exec 6>/dev/null
 ac_pwd=`pwd` && test -n "$ac_pwd" &&
 ac_ls_di=`ls -di .` &&
 ac_pwd_ls_di=`cd "$ac_pwd" && ls -di .` ||
-  as_fn_error $? "working directory cannot be determined"
+  { $as_echo "$as_me: error: working directory cannot be determined" >&2
+   { (exit 1); exit 1; }; }
 test "X$ac_ls_di" = "X$ac_pwd_ls_di" ||
-  as_fn_error $? "pwd does not report name of working directory"
+  { $as_echo "$as_me: error: pwd does not report name of working directory" >&2
+   { (exit 1); exit 1; }; }
 
 
 # Find the source files, if location was not specified.
@@ -1270,11 +1291,13 @@ else
 fi
 if test ! -r "$srcdir/$ac_unique_file"; then
   test "$ac_srcdir_defaulted" = yes && srcdir="$ac_confdir or .."
-  as_fn_error $? "cannot find sources ($ac_unique_file) in $srcdir"
+  { $as_echo "$as_me: error: cannot find sources ($ac_unique_file) in $srcdir" >&2
+   { (exit 1); exit 1; }; }
 fi
 ac_msg="sources are in $srcdir, but \`cd $srcdir' does not work"
 ac_abs_confdir=`(
-	cd "$srcdir" && test -r "./$ac_unique_file" || as_fn_error $? "$ac_msg"
+	cd "$srcdir" && test -r "./$ac_unique_file" || { $as_echo "$as_me: error: $ac_msg" >&2
+   { (exit 1); exit 1; }; }
 	pwd)`
 # When building in place, set srcdir=.
 if test "$ac_abs_confdir" = "$ac_pwd"; then
@@ -1314,7 +1337,7 @@ Configuration:
       --help=short        display options specific to this package
       --help=recursive    display the short help of all the included packages
   -V, --version           display version information and exit
-  -q, --quiet, --silent   do not print \`checking ...' messages
+  -q, --quiet, --silent   do not print \`checking...' messages
       --cache-file=FILE   cache test results in FILE [disabled]
   -C, --config-cache      alias for \`--cache-file=config.cache'
   -n, --no-create         do not create output files
@@ -1379,12 +1402,10 @@ Optional Features:
   --disable-option-checking  ignore unrecognized --enable/--with options
   --disable-FEATURE       do not include FEATURE (same as --enable-FEATURE=no)
   --enable-FEATURE[=ARG]  include FEATURE [ARG=yes]
-  --enable-silent-rules   less verbose build output (undo: "make V=1")
-  --disable-silent-rules  verbose build output (undo: "make V=0")
-  --enable-dependency-tracking
-                          do not reject slow dependency extractors
-  --disable-dependency-tracking
-                          speeds up one-time build
+  --enable-silent-rules          less verbose build output (undo: `make V=1')
+  --disable-silent-rules         verbose build output (undo: `make V=0')
+  --disable-dependency-tracking  speeds up one-time build
+  --enable-dependency-tracking   do not reject slow dependency extractors
   --disable-openmp        do not use OpenMP
   --enable-simd=SSE|AVX|AVX2|AVX512|MIC
                           Select instructions to be SSE4.0, AVX 1.0, AVX
@@ -1402,7 +1423,7 @@ Some influential environment variables:
   LDFLAGS     linker flags, e.g. -L<lib dir> if you have libraries in a
               nonstandard directory <lib dir>
   LIBS        libraries to pass to the linker, e.g. -l<library>
-  CPPFLAGS    (Objective) C/C++ preprocessor flags, e.g. -I<include dir> if
+  CPPFLAGS    C/C++/Objective C preprocessor flags, e.g. -I<include dir> if
               you have headers in a nonstandard directory <include dir>
   CXXCPP      C++ preprocessor
 
@@ -1473,530 +1494,21 @@ test -n "$ac_init_help" && exit $ac_status
 if $ac_init_version; then
   cat <<\_ACEOF
 Grid configure 1.0
-generated by GNU Autoconf 2.69
+generated by GNU Autoconf 2.63
 
-Copyright (C) 2012 Free Software Foundation, Inc.
+Copyright (C) 1992, 1993, 1994, 1995, 1996, 1998, 1999, 2000, 2001,
+2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
 This configure script is free software; the Free Software Foundation
 gives unlimited permission to copy, distribute and modify it.
 _ACEOF
   exit
 fi
-
-## ------------------------ ##
-## Autoconf initialization. ##
-## ------------------------ ##
-
-# ac_fn_cxx_try_compile LINENO
-# ----------------------------
-# Try to compile conftest.$ac_ext, and return whether this succeeded.
-ac_fn_cxx_try_compile ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  rm -f conftest.$ac_objext
-  if { { ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compile") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_retval=1
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_cxx_try_compile
-
-# ac_fn_cxx_try_link LINENO
-# -------------------------
-# Try to link conftest.$ac_ext, and return whether this succeeded.
-ac_fn_cxx_try_link ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  rm -f conftest.$ac_objext conftest$ac_exeext
-  if { { ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_link") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 test -x conftest$ac_exeext
-       }; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_retval=1
-fi
-  # Delete the IPA/IPO (Inter Procedural Analysis/Optimization) information
-  # created by the PGI compiler (conftest_ipa8_conftest.oo), as it would
-  # interfere with the next link command; also delete a directory that is
-  # left behind by Apple's compiler.  We do this before executing the actions.
-  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_cxx_try_link
-
-# ac_fn_cxx_try_cpp LINENO
-# ------------------------
-# Try to preprocess conftest.$ac_ext, and return whether this succeeded.
-ac_fn_cxx_try_cpp ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  if { { ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } > conftest.i && {
-	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       }; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-    ac_retval=1
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_cxx_try_cpp
-
-# ac_fn_cxx_check_header_mongrel LINENO HEADER VAR INCLUDES
-# ---------------------------------------------------------
-# Tests whether HEADER exists, giving a warning if it cannot be compiled using
-# the include files in INCLUDES and setting the cache variable VAR
-# accordingly.
-ac_fn_cxx_check_header_mongrel ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  if eval \${$3+:} false; then :
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 usability" >&5
-$as_echo_n "checking $2 usability... " >&6; }
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-#include <$2>
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  ac_header_compiler=yes
-else
-  ac_header_compiler=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 presence" >&5
-$as_echo_n "checking $2 presence... " >&6; }
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <$2>
-_ACEOF
-if ac_fn_cxx_try_cpp "$LINENO"; then :
-  ac_header_preproc=yes
-else
-  ac_header_preproc=no
-fi
-rm -f conftest.err conftest.i conftest.$ac_ext
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_cxx_preproc_warn_flag in #((
-  yes:no: )
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $2: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $2:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $2: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
-( $as_echo "## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##"
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  eval "$3=\$ac_header_compiler"
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_cxx_check_header_mongrel
-
-# ac_fn_cxx_try_run LINENO
-# ------------------------
-# Try to link conftest.$ac_ext, and return whether this succeeded. Assumes
-# that executables *can* be run.
-ac_fn_cxx_try_run ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  if { { ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_link") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && { ac_try='./conftest$ac_exeext'
-  { { case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; }; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: program exited with status $ac_status" >&5
-       $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-       ac_retval=$ac_status
-fi
-  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_cxx_try_run
-
-# ac_fn_cxx_check_header_compile LINENO HEADER VAR INCLUDES
-# ---------------------------------------------------------
-# Tests whether HEADER exists and can be compiled using the include files in
-# INCLUDES, setting the cache variable VAR accordingly.
-ac_fn_cxx_check_header_compile ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-#include <$2>
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  eval "$3=yes"
-else
-  eval "$3=no"
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_cxx_check_header_compile
-
-# ac_fn_cxx_check_decl LINENO SYMBOL VAR INCLUDES
-# -----------------------------------------------
-# Tests whether SYMBOL is declared in INCLUDES, setting cache variable VAR
-# accordingly.
-ac_fn_cxx_check_decl ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  as_decl_name=`echo $2|sed 's/ *(.*//'`
-  as_decl_use=`echo $2|sed -e 's/(/((/' -e 's/)/) 0&/' -e 's/,/) 0& (/g'`
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $as_decl_name is declared" >&5
-$as_echo_n "checking whether $as_decl_name is declared... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-int
-main ()
-{
-#ifndef $as_decl_name
-#ifdef __cplusplus
-  (void) $as_decl_use;
-#else
-  (void) $as_decl_name;
-#endif
-#endif
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  eval "$3=yes"
-else
-  eval "$3=no"
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_cxx_check_decl
-
-# ac_fn_cxx_check_type LINENO TYPE VAR INCLUDES
-# ---------------------------------------------
-# Tests whether TYPE exists after having included INCLUDES, setting cache
-# variable VAR accordingly.
-ac_fn_cxx_check_type ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  eval "$3=no"
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-int
-main ()
-{
-if (sizeof ($2))
-	 return 0;
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-int
-main ()
-{
-if (sizeof (($2)))
-	    return 0;
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-
-else
-  eval "$3=yes"
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_cxx_check_type
-
-# ac_fn_c_find_uintX_t LINENO BITS VAR
-# ------------------------------------
-# Finds an unsigned integer type with width BITS, setting cache variable VAR
-# accordingly.
-ac_fn_c_find_uintX_t ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for uint$2_t" >&5
-$as_echo_n "checking for uint$2_t... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  eval "$3=no"
-     # Order is important - never check a type that is potentially smaller
-     # than half of the expected target width.
-     for ac_type in uint$2_t 'unsigned int' 'unsigned long int' \
-	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
-       cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-static int test_array [1 - 2 * !((($ac_type) -1 >> ($2 / 2 - 1)) >> ($2 / 2 - 1) == 3)];
-test_array [0] = 0;
-return test_array [0];
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  case $ac_type in #(
-  uint$2_t) :
-    eval "$3=yes" ;; #(
-  *) :
-    eval "$3=\$ac_type" ;;
-esac
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-       if eval test \"x\$"$3"\" = x"no"; then :
-
-else
-  break
-fi
-     done
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_find_uintX_t
-
-# ac_fn_cxx_check_func LINENO FUNC VAR
-# ------------------------------------
-# Tests whether FUNC exists, setting the cache variable VAR accordingly
-ac_fn_cxx_check_func ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-/* Define $2 to an innocuous variant, in case <limits.h> declares $2.
-   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
-#define $2 innocuous_$2
-
-/* System header to define __stub macros and hopefully few prototypes,
-    which can conflict with char $2 (); below.
-    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
-    <limits.h> exists even on freestanding compilers.  */
-
-#ifdef __STDC__
-# include <limits.h>
-#else
-# include <assert.h>
-#endif
-
-#undef $2
-
-/* Override any GCC internal prototype to avoid an error.
-   Use char because int might match the return type of a GCC
-   builtin and then its argument prototype would still apply.  */
-#ifdef __cplusplus
-extern "C"
-#endif
-char $2 ();
-/* The GNU C library defines this for functions which it implements
-    to always fail with ENOSYS.  Some functions are actually named
-    something starting with __ and the normal name is an alias.  */
-#if defined __stub_$2 || defined __stub___$2
-choke me
-#endif
-
-int
-main ()
-{
-return $2 ();
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_link "$LINENO"; then :
-  eval "$3=yes"
-else
-  eval "$3=no"
-fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_cxx_check_func
 cat >config.log <<_ACEOF
 This file contains any messages produced by compilers while
 running configure, to aid debugging if configure makes a mistake.
 
 It was created by Grid $as_me 1.0, which was
-generated by GNU Autoconf 2.69.  Invocation command line was
+generated by GNU Autoconf 2.63.  Invocation command line was
 
   $ $0 $@
 
@@ -2032,8 +1544,8 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    $as_echo "PATH: $as_dir"
-  done
+  $as_echo "PATH: $as_dir"
+done
 IFS=$as_save_IFS
 
 } >&5
@@ -2070,9 +1582,9 @@ do
       ac_arg=`$as_echo "$ac_arg" | sed "s/'/'\\\\\\\\''/g"` ;;
     esac
     case $ac_pass in
-    1) as_fn_append ac_configure_args0 " '$ac_arg'" ;;
+    1) ac_configure_args0="$ac_configure_args0 '$ac_arg'" ;;
     2)
-      as_fn_append ac_configure_args1 " '$ac_arg'"
+      ac_configure_args1="$ac_configure_args1 '$ac_arg'"
       if test $ac_must_keep_next = true; then
 	ac_must_keep_next=false # Got value, back to normal.
       else
@@ -2088,13 +1600,13 @@ do
 	  -* ) ac_must_keep_next=true ;;
 	esac
       fi
-      as_fn_append ac_configure_args " '$ac_arg'"
+      ac_configure_args="$ac_configure_args '$ac_arg'"
       ;;
     esac
   done
 done
-{ ac_configure_args0=; unset ac_configure_args0;}
-{ ac_configure_args1=; unset ac_configure_args1;}
+$as_unset ac_configure_args0 || test "${ac_configure_args0+set}" != set || { ac_configure_args0=; export ac_configure_args0; }
+$as_unset ac_configure_args1 || test "${ac_configure_args1+set}" != set || { ac_configure_args1=; export ac_configure_args1; }
 
 # When interrupted or exit'd, cleanup temporary files, and complete
 # config.log.  We remove comments because anyway the quotes in there
@@ -2106,9 +1618,11 @@ trap 'exit_status=$?
   {
     echo
 
-    $as_echo "## ---------------- ##
+    cat <<\_ASBOX
+## ---------------- ##
 ## Cache variables. ##
-## ---------------- ##"
+## ---------------- ##
+_ASBOX
     echo
     # The following way of writing the cache mishandles newlines in values,
 (
@@ -2117,13 +1631,13 @@ trap 'exit_status=$?
     case $ac_val in #(
     *${as_nl}*)
       case $ac_var in #(
-      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
+      *_cv_*) { $as_echo "$as_me:$LINENO: WARNING: cache variable $ac_var contains a newline" >&5
 $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
       esac
       case $ac_var in #(
       _ | IFS | as_nl) ;; #(
       BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
-      *) { eval $ac_var=; unset $ac_var;} ;;
+      *) $as_unset $ac_var ;;
       esac ;;
     esac
   done
@@ -2142,9 +1656,11 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
 )
     echo
 
-    $as_echo "## ----------------- ##
+    cat <<\_ASBOX
+## ----------------- ##
 ## Output variables. ##
-## ----------------- ##"
+## ----------------- ##
+_ASBOX
     echo
     for ac_var in $ac_subst_vars
     do
@@ -2157,9 +1673,11 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
     echo
 
     if test -n "$ac_subst_files"; then
-      $as_echo "## ------------------- ##
+      cat <<\_ASBOX
+## ------------------- ##
 ## File substitutions. ##
-## ------------------- ##"
+## ------------------- ##
+_ASBOX
       echo
       for ac_var in $ac_subst_files
       do
@@ -2173,9 +1691,11 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
     fi
 
     if test -s confdefs.h; then
-      $as_echo "## ----------- ##
+      cat <<\_ASBOX
+## ----------- ##
 ## confdefs.h. ##
-## ----------- ##"
+## ----------- ##
+_ASBOX
       echo
       cat confdefs.h
       echo
@@ -2189,39 +1709,37 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
     exit $exit_status
 ' 0
 for ac_signal in 1 2 13 15; do
-  trap 'ac_signal='$ac_signal'; as_fn_exit 1' $ac_signal
+  trap 'ac_signal='$ac_signal'; { (exit 1); exit 1; }' $ac_signal
 done
 ac_signal=0
 
 # confdefs.h avoids OS command line length limits that DEFS can exceed.
 rm -f -r conftest* confdefs.h
 
-$as_echo "/* confdefs.h */" > confdefs.h
-
 # Predefined preprocessor variables.
 
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_NAME "$PACKAGE_NAME"
 _ACEOF
 
+
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_TARNAME "$PACKAGE_TARNAME"
 _ACEOF
 
+
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_VERSION "$PACKAGE_VERSION"
 _ACEOF
 
+
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_STRING "$PACKAGE_STRING"
 _ACEOF
 
-cat >>confdefs.h <<_ACEOF
-#define PACKAGE_BUGREPORT "$PACKAGE_BUGREPORT"
-_ACEOF
 
 cat >>confdefs.h <<_ACEOF
-#define PACKAGE_URL "$PACKAGE_URL"
+#define PACKAGE_BUGREPORT "$PACKAGE_BUGREPORT"
 _ACEOF
 
 
@@ -2230,12 +1748,7 @@ _ACEOF
 ac_site_file1=NONE
 ac_site_file2=NONE
 if test -n "$CONFIG_SITE"; then
-  # We do not want a PATH search for config.site.
-  case $CONFIG_SITE in #((
-    -*)  ac_site_file1=./$CONFIG_SITE;;
-    */*) ac_site_file1=$CONFIG_SITE;;
-    *)   ac_site_file1=./$CONFIG_SITE;;
-  esac
+  ac_site_file1=$CONFIG_SITE
 elif test "x$prefix" != xNONE; then
   ac_site_file1=$prefix/share/config.site
   ac_site_file2=$prefix/etc/config.site
@@ -2246,23 +1759,19 @@ fi
 for ac_site_file in "$ac_site_file1" "$ac_site_file2"
 do
   test "x$ac_site_file" = xNONE && continue
-  if test /dev/null != "$ac_site_file" && test -r "$ac_site_file"; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: loading site script $ac_site_file" >&5
+  if test -r "$ac_site_file"; then
+    { $as_echo "$as_me:$LINENO: loading site script $ac_site_file" >&5
 $as_echo "$as_me: loading site script $ac_site_file" >&6;}
     sed 's/^/| /' "$ac_site_file" >&5
-    . "$ac_site_file" \
-      || { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "failed to load site script $ac_site_file
-See \`config.log' for more details" "$LINENO" 5; }
+    . "$ac_site_file"
   fi
 done
 
 if test -r "$cache_file"; then
-  # Some versions of bash will fail to source /dev/null (special files
-  # actually), so we avoid doing that.  DJGPP emulates it as a regular file.
-  if test /dev/null != "$cache_file" && test -f "$cache_file"; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: loading cache $cache_file" >&5
+  # Some versions of bash will fail to source /dev/null (special
+  # files actually), so we avoid doing that.
+  if test -f "$cache_file"; then
+    { $as_echo "$as_me:$LINENO: loading cache $cache_file" >&5
 $as_echo "$as_me: loading cache $cache_file" >&6;}
     case $cache_file in
       [\\/]* | ?:[\\/]* ) . "$cache_file";;
@@ -2270,7 +1779,7 @@ $as_echo "$as_me: loading cache $cache_file" >&6;}
     esac
   fi
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: creating cache $cache_file" >&5
+  { $as_echo "$as_me:$LINENO: creating cache $cache_file" >&5
 $as_echo "$as_me: creating cache $cache_file" >&6;}
   >$cache_file
 fi
@@ -2285,11 +1794,11 @@ for ac_var in $ac_precious_vars; do
   eval ac_new_val=\$ac_env_${ac_var}_value
   case $ac_old_set,$ac_new_set in
     set,)
-      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
+      { $as_echo "$as_me:$LINENO: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
 $as_echo "$as_me: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&2;}
       ac_cache_corrupted=: ;;
     ,set)
-      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was not set in the previous run" >&5
+      { $as_echo "$as_me:$LINENO: error: \`$ac_var' was not set in the previous run" >&5
 $as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
       ac_cache_corrupted=: ;;
     ,);;
@@ -2299,17 +1808,17 @@ $as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
 	ac_old_val_w=`echo x $ac_old_val`
 	ac_new_val_w=`echo x $ac_new_val`
 	if test "$ac_old_val_w" != "$ac_new_val_w"; then
-	  { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' has changed since the previous run:" >&5
+	  { $as_echo "$as_me:$LINENO: error: \`$ac_var' has changed since the previous run:" >&5
 $as_echo "$as_me: error: \`$ac_var' has changed since the previous run:" >&2;}
 	  ac_cache_corrupted=:
 	else
-	  { $as_echo "$as_me:${as_lineno-$LINENO}: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
+	  { $as_echo "$as_me:$LINENO: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
 $as_echo "$as_me: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&2;}
 	  eval $ac_var=\$ac_old_val
 	fi
-	{ $as_echo "$as_me:${as_lineno-$LINENO}:   former value:  \`$ac_old_val'" >&5
+	{ $as_echo "$as_me:$LINENO:   former value:  \`$ac_old_val'" >&5
 $as_echo "$as_me:   former value:  \`$ac_old_val'" >&2;}
-	{ $as_echo "$as_me:${as_lineno-$LINENO}:   current value: \`$ac_new_val'" >&5
+	{ $as_echo "$as_me:$LINENO:   current value: \`$ac_new_val'" >&5
 $as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
       fi;;
   esac
@@ -2321,20 +1830,43 @@ $as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
     esac
     case " $ac_configure_args " in
       *" '$ac_arg' "*) ;; # Avoid dups.  Use of quotes ensures accuracy.
-      *) as_fn_append ac_configure_args " '$ac_arg'" ;;
+      *) ac_configure_args="$ac_configure_args '$ac_arg'" ;;
     esac
   fi
 done
 if $ac_cache_corrupted; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+  { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-  { $as_echo "$as_me:${as_lineno-$LINENO}: error: changes in the environment can compromise the build" >&5
+  { $as_echo "$as_me:$LINENO: error: changes in the environment can compromise the build" >&5
 $as_echo "$as_me: error: changes in the environment can compromise the build" >&2;}
-  as_fn_error $? "run \`make distclean' and/or \`rm $cache_file' and start over" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: run \`make distclean' and/or \`rm $cache_file' and start over" >&5
+$as_echo "$as_me: error: run \`make distclean' and/or \`rm $cache_file' and start over" >&2;}
+   { (exit 1); exit 1; }; }
 fi
-## -------------------- ##
-## Main body of script. ##
-## -------------------- ##
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
 
 ac_ext=c
 ac_cpp='$CPP $CPPFLAGS'
@@ -2360,7 +1892,9 @@ for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
   fi
 done
 if test -z "$ac_aux_dir"; then
-  as_fn_error $? "cannot find install-sh, install.sh, or shtool in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: cannot find install-sh or install.sh in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" >&5
+$as_echo "$as_me: error: cannot find install-sh or install.sh in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" >&2;}
+   { (exit 1); exit 1; }; }
 fi
 
 # These three variables are undocumented and unsupported,
@@ -2374,27 +1908,35 @@ ac_configure="$SHELL $ac_aux_dir/configure"  # Please don't use this var.
 
 # Make sure we can run config.sub.
 $SHELL "$ac_aux_dir/config.sub" sun4 >/dev/null 2>&1 ||
-  as_fn_error $? "cannot run $SHELL $ac_aux_dir/config.sub" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: cannot run $SHELL $ac_aux_dir/config.sub" >&5
+$as_echo "$as_me: error: cannot run $SHELL $ac_aux_dir/config.sub" >&2;}
+   { (exit 1); exit 1; }; }
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking build system type" >&5
+{ $as_echo "$as_me:$LINENO: checking build system type" >&5
 $as_echo_n "checking build system type... " >&6; }
-if ${ac_cv_build+:} false; then :
+if test "${ac_cv_build+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   ac_build_alias=$build_alias
 test "x$ac_build_alias" = x &&
   ac_build_alias=`$SHELL "$ac_aux_dir/config.guess"`
 test "x$ac_build_alias" = x &&
-  as_fn_error $? "cannot guess build type; you must specify one" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: cannot guess build type; you must specify one" >&5
+$as_echo "$as_me: error: cannot guess build type; you must specify one" >&2;}
+   { (exit 1); exit 1; }; }
 ac_cv_build=`$SHELL "$ac_aux_dir/config.sub" $ac_build_alias` ||
-  as_fn_error $? "$SHELL $ac_aux_dir/config.sub $ac_build_alias failed" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: $SHELL $ac_aux_dir/config.sub $ac_build_alias failed" >&5
+$as_echo "$as_me: error: $SHELL $ac_aux_dir/config.sub $ac_build_alias failed" >&2;}
+   { (exit 1); exit 1; }; }
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_build" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_build" >&5
 $as_echo "$ac_cv_build" >&6; }
 case $ac_cv_build in
 *-*-*) ;;
-*) as_fn_error $? "invalid value of canonical build" "$LINENO" 5;;
+*) { { $as_echo "$as_me:$LINENO: error: invalid value of canonical build" >&5
+$as_echo "$as_me: error: invalid value of canonical build" >&2;}
+   { (exit 1); exit 1; }; };;
 esac
 build=$ac_cv_build
 ac_save_IFS=$IFS; IFS='-'
@@ -2410,24 +1952,28 @@ IFS=$ac_save_IFS
 case $build_os in *\ *) build_os=`echo "$build_os" | sed 's/ /-/g'`;; esac
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking host system type" >&5
+{ $as_echo "$as_me:$LINENO: checking host system type" >&5
 $as_echo_n "checking host system type... " >&6; }
-if ${ac_cv_host+:} false; then :
+if test "${ac_cv_host+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test "x$host_alias" = x; then
   ac_cv_host=$ac_cv_build
 else
   ac_cv_host=`$SHELL "$ac_aux_dir/config.sub" $host_alias` ||
-    as_fn_error $? "$SHELL $ac_aux_dir/config.sub $host_alias failed" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: $SHELL $ac_aux_dir/config.sub $host_alias failed" >&5
+$as_echo "$as_me: error: $SHELL $ac_aux_dir/config.sub $host_alias failed" >&2;}
+   { (exit 1); exit 1; }; }
 fi
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_host" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_host" >&5
 $as_echo "$ac_cv_host" >&6; }
 case $ac_cv_host in
 *-*-*) ;;
-*) as_fn_error $? "invalid value of canonical host" "$LINENO" 5;;
+*) { { $as_echo "$as_me:$LINENO: error: invalid value of canonical host" >&5
+$as_echo "$as_me: error: invalid value of canonical host" >&2;}
+   { (exit 1); exit 1; }; };;
 esac
 host=$ac_cv_host
 ac_save_IFS=$IFS; IFS='-'
@@ -2443,24 +1989,28 @@ IFS=$ac_save_IFS
 case $host_os in *\ *) host_os=`echo "$host_os" | sed 's/ /-/g'`;; esac
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking target system type" >&5
+{ $as_echo "$as_me:$LINENO: checking target system type" >&5
 $as_echo_n "checking target system type... " >&6; }
-if ${ac_cv_target+:} false; then :
+if test "${ac_cv_target+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test "x$target_alias" = x; then
   ac_cv_target=$ac_cv_host
 else
   ac_cv_target=`$SHELL "$ac_aux_dir/config.sub" $target_alias` ||
-    as_fn_error $? "$SHELL $ac_aux_dir/config.sub $target_alias failed" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: $SHELL $ac_aux_dir/config.sub $target_alias failed" >&5
+$as_echo "$as_me: error: $SHELL $ac_aux_dir/config.sub $target_alias failed" >&2;}
+   { (exit 1); exit 1; }; }
 fi
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_target" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_target" >&5
 $as_echo "$ac_cv_target" >&6; }
 case $ac_cv_target in
 *-*-*) ;;
-*) as_fn_error $? "invalid value of canonical target" "$LINENO" 5;;
+*) { { $as_echo "$as_me:$LINENO: error: invalid value of canonical target" >&5
+$as_echo "$as_me: error: invalid value of canonical target" >&2;}
+   { (exit 1); exit 1; }; };;
 esac
 target=$ac_cv_target
 ac_save_IFS=$IFS; IFS='-'
@@ -2482,8 +2032,7 @@ test -n "$target_alias" &&
   test "$program_prefix$program_suffix$program_transform_name" = \
     NONENONEs,x,x, &&
   program_prefix=${target_alias}-
-
-am__api_version='1.14'
+am__api_version='1.11'
 
 # Find a good install program.  We prefer a C program (faster),
 # so one script is as good as another.  But avoid the broken or
@@ -2499,10 +2048,10 @@ am__api_version='1.14'
 # OS/2's system install, which has a completely different semantic
 # ./install, which can be erroneously created by make from ./install.sh.
 # Reject install programs that cannot install multiple files.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a BSD-compatible install" >&5
+{ $as_echo "$as_me:$LINENO: checking for a BSD-compatible install" >&5
 $as_echo_n "checking for a BSD-compatible install... " >&6; }
 if test -z "$INSTALL"; then
-if ${ac_cv_path_install+:} false; then :
+if test "${ac_cv_path_install+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
@@ -2510,11 +2059,11 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    # Account for people who put trailing slashes in PATH elements.
-case $as_dir/ in #((
-  ./ | .// | /[cC]/* | \
+  # Account for people who put trailing slashes in PATH elements.
+case $as_dir/ in
+  ./ | .// | /cC/* | \
   /etc/* | /usr/sbin/* | /usr/etc/* | /sbin/* | /usr/afsws/bin/* | \
-  ?:[\\/]os2[\\/]install[\\/]* | ?:[\\/]OS2[\\/]INSTALL[\\/]* | \
+  ?:\\/os2\\/install\\/* | ?:\\/OS2\\/INSTALL\\/* | \
   /usr/ucb/* ) ;;
   *)
     # OSF1 and SCO ODT 3.0 have their own names for install.
@@ -2522,7 +2071,7 @@ case $as_dir/ in #((
     # by default.
     for ac_prog in ginstall scoinst install; do
       for ac_exec_ext in '' $ac_executable_extensions; do
-	if as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext"; then
+	if { test -f "$as_dir/$ac_prog$ac_exec_ext" && $as_test_x "$as_dir/$ac_prog$ac_exec_ext"; }; then
 	  if test $ac_prog = install &&
 	    grep dspmsg "$as_dir/$ac_prog$ac_exec_ext" >/dev/null 2>&1; then
 	    # AIX install.  It has an incompatible calling convention.
@@ -2551,7 +2100,7 @@ case $as_dir/ in #((
     ;;
 esac
 
-  done
+done
 IFS=$as_save_IFS
 
 rm -rf conftest.one conftest.two conftest.dir
@@ -2567,7 +2116,7 @@ fi
     INSTALL=$ac_install_sh
   fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $INSTALL" >&5
+{ $as_echo "$as_me:$LINENO: result: $INSTALL" >&5
 $as_echo "$INSTALL" >&6; }
 
 # Use test -z because SunOS4 sh mishandles braces in ${var-val}.
@@ -2578,73 +2127,68 @@ test -z "$INSTALL_SCRIPT" && INSTALL_SCRIPT='${INSTALL}'
 
 test -z "$INSTALL_DATA" && INSTALL_DATA='${INSTALL} -m 644'
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether build environment is sane" >&5
+{ $as_echo "$as_me:$LINENO: checking whether build environment is sane" >&5
 $as_echo_n "checking whether build environment is sane... " >&6; }
+# Just in case
+sleep 1
+echo timestamp > conftest.file
 # Reject unsafe characters in $srcdir or the absolute working directory
 # name.  Accept space and tab only in the latter.
 am_lf='
 '
 case `pwd` in
   *[\\\"\#\$\&\'\`$am_lf]*)
-    as_fn_error $? "unsafe absolute working directory name" "$LINENO" 5;;
+    { { $as_echo "$as_me:$LINENO: error: unsafe absolute working directory name" >&5
+$as_echo "$as_me: error: unsafe absolute working directory name" >&2;}
+   { (exit 1); exit 1; }; };;
 esac
 case $srcdir in
   *[\\\"\#\$\&\'\`$am_lf\ \	]*)
-    as_fn_error $? "unsafe srcdir value: '$srcdir'" "$LINENO" 5;;
+    { { $as_echo "$as_me:$LINENO: error: unsafe srcdir value: \`$srcdir'" >&5
+$as_echo "$as_me: error: unsafe srcdir value: \`$srcdir'" >&2;}
+   { (exit 1); exit 1; }; };;
 esac
 
-# Do 'set' in a subshell so we don't clobber the current shell's
+# Do `set' in a subshell so we don't clobber the current shell's
 # arguments.  Must try -L first in case configure is actually a
 # symlink; some systems play weird games with the mod time of symlinks
 # (eg FreeBSD returns the mod time of the symlink's containing
 # directory).
 if (
-   am_has_slept=no
-   for am_try in 1 2; do
-     echo "timestamp, slept: $am_has_slept" > conftest.file
-     set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
-     if test "$*" = "X"; then
-	# -L didn't work.
-	set X `ls -t "$srcdir/configure" conftest.file`
-     fi
-     if test "$*" != "X $srcdir/configure conftest.file" \
-	&& test "$*" != "X conftest.file $srcdir/configure"; then
+   set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
+   if test "$*" = "X"; then
+      # -L didn't work.
+      set X `ls -t "$srcdir/configure" conftest.file`
+   fi
+   rm -f conftest.file
+   if test "$*" != "X $srcdir/configure conftest.file" \
+      && test "$*" != "X conftest.file $srcdir/configure"; then
+
+      # If neither matched, then we have a broken ls.  This can happen
+      # if, for instance, CONFIG_SHELL is bash and it inherits a
+      # broken ls alias from the environment.  This has actually
+      # happened.  Such a system could not be considered "sane".
+      { { $as_echo "$as_me:$LINENO: error: ls -t appears to fail.  Make sure there is not a broken
+alias in your environment" >&5
+$as_echo "$as_me: error: ls -t appears to fail.  Make sure there is not a broken
+alias in your environment" >&2;}
+   { (exit 1); exit 1; }; }
+   fi
 
-	# If neither matched, then we have a broken ls.  This can happen
-	# if, for instance, CONFIG_SHELL is bash and it inherits a
-	# broken ls alias from the environment.  This has actually
-	# happened.  Such a system could not be considered "sane".
-	as_fn_error $? "ls -t appears to fail.  Make sure there is not a broken
-  alias in your environment" "$LINENO" 5
-     fi
-     if test "$2" = conftest.file || test $am_try -eq 2; then
-       break
-     fi
-     # Just in case.
-     sleep 1
-     am_has_slept=yes
-   done
    test "$2" = conftest.file
    )
 then
    # Ok.
    :
 else
-   as_fn_error $? "newly created file is older than distributed files!
-Check your system clock" "$LINENO" 5
+   { { $as_echo "$as_me:$LINENO: error: newly created file is older than distributed files!
+Check your system clock" >&5
+$as_echo "$as_me: error: newly created file is older than distributed files!
+Check your system clock" >&2;}
+   { (exit 1); exit 1; }; }
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+{ $as_echo "$as_me:$LINENO: result: yes" >&5
 $as_echo "yes" >&6; }
-# If we didn't sleep, we still need to ensure time stamps of config.status and
-# generated files are strictly newer.
-am_sleep_pid=
-if grep 'slept: no' conftest.file >/dev/null 2>&1; then
-  ( sleep 1 ) &
-  am_sleep_pid=$!
-fi
-
-rm -f conftest.file
-
 test "$program_prefix" != NONE &&
   program_transform_name="s&^&$program_prefix&;$program_transform_name"
 # Use a double $ so make ignores it.
@@ -2667,12 +2211,12 @@ if test x"${MISSING+set}" != xset; then
   esac
 fi
 # Use eval to expand $SHELL
-if eval "$MISSING --is-lightweight"; then
-  am_missing_run="$MISSING "
+if eval "$MISSING --run true"; then
+  am_missing_run="$MISSING --run "
 else
   am_missing_run=
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: 'missing' script is too old or missing" >&5
-$as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
+  { $as_echo "$as_me:$LINENO: WARNING: \`missing' script is too old or missing" >&5
+$as_echo "$as_me: WARNING: \`missing' script is too old or missing" >&2;}
 fi
 
 if test x"${install_sh}" != xset; then
@@ -2684,17 +2228,17 @@ if test x"${install_sh}" != xset; then
   esac
 fi
 
-# Installed binaries are usually stripped using 'strip' when the user
-# run "make install-strip".  However 'strip' might not be the right
+# Installed binaries are usually stripped using `strip' when the user
+# run `make install-strip'.  However `strip' might not be the right
 # tool to use in cross-compilation environments, therefore Automake
-# will honor the 'STRIP' environment variable to overrule this program.
+# will honor the `STRIP' environment variable to overrule this program.
 if test "$cross_compiling" != no; then
   if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}strip", so it can be a program name with args.
 set dummy ${ac_tool_prefix}strip; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_STRIP+:} false; then :
+if test "${ac_cv_prog_STRIP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$STRIP"; then
@@ -2705,24 +2249,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_STRIP="${ac_tool_prefix}strip"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 STRIP=$ac_cv_prog_STRIP
 if test -n "$STRIP"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $STRIP" >&5
+  { $as_echo "$as_me:$LINENO: result: $STRIP" >&5
 $as_echo "$STRIP" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2732,9 +2276,9 @@ if test -z "$ac_cv_prog_STRIP"; then
   ac_ct_STRIP=$STRIP
   # Extract the first word of "strip", so it can be a program name with args.
 set dummy strip; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_STRIP+:} false; then :
+if test "${ac_cv_prog_ac_ct_STRIP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_STRIP"; then
@@ -2745,24 +2289,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_ac_ct_STRIP="strip"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_STRIP=$ac_cv_prog_ac_ct_STRIP
 if test -n "$ac_ct_STRIP"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_STRIP" >&5
+  { $as_echo "$as_me:$LINENO: result: $ac_ct_STRIP" >&5
 $as_echo "$ac_ct_STRIP" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2771,7 +2315,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -2784,10 +2328,10 @@ fi
 fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a thread-safe mkdir -p" >&5
+{ $as_echo "$as_me:$LINENO: checking for a thread-safe mkdir -p" >&5
 $as_echo_n "checking for a thread-safe mkdir -p... " >&6; }
 if test -z "$MKDIR_P"; then
-  if ${ac_cv_path_mkdir+:} false; then :
+  if test "${ac_cv_path_mkdir+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
@@ -2795,9 +2339,9 @@ for as_dir in $PATH$PATH_SEPARATOR/opt/sfw/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_prog in mkdir gmkdir; do
+  for ac_prog in mkdir gmkdir; do
 	 for ac_exec_ext in '' $ac_executable_extensions; do
-	   as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext" || continue
+	   { test -f "$as_dir/$ac_prog$ac_exec_ext" && $as_test_x "$as_dir/$ac_prog$ac_exec_ext"; } || continue
 	   case `"$as_dir/$ac_prog$ac_exec_ext" --version 2>&1` in #(
 	     'mkdir (GNU coreutils) '* | \
 	     'mkdir (coreutils) '* | \
@@ -2807,12 +2351,11 @@ do
 	   esac
 	 done
        done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 
-  test -d ./--version && rmdir ./--version
   if test "${ac_cv_path_mkdir+set}" = set; then
     MKDIR_P="$ac_cv_path_mkdir -p"
   else
@@ -2820,19 +2363,26 @@ fi
     # value for MKDIR_P within a source directory, because that will
     # break other packages using the cache if that directory is
     # removed, or if the value is a relative name.
+    test -d ./--version && rmdir ./--version
     MKDIR_P="$ac_install_sh -d"
   fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $MKDIR_P" >&5
+{ $as_echo "$as_me:$LINENO: result: $MKDIR_P" >&5
 $as_echo "$MKDIR_P" >&6; }
 
+mkdir_p="$MKDIR_P"
+case $mkdir_p in
+  [\\/$]* | ?:[\\/]*) ;;
+  */*) mkdir_p="\$(top_builddir)/$mkdir_p" ;;
+esac
+
 for ac_prog in gawk mawk nawk awk
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_AWK+:} false; then :
+if test "${ac_cv_prog_AWK+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$AWK"; then
@@ -2843,24 +2393,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_AWK="$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 AWK=$ac_cv_prog_AWK
 if test -n "$AWK"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $AWK" >&5
+  { $as_echo "$as_me:$LINENO: result: $AWK" >&5
 $as_echo "$AWK" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2868,11 +2418,11 @@ fi
   test -n "$AWK" && break
 done
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether ${MAKE-make} sets \$(MAKE)" >&5
+{ $as_echo "$as_me:$LINENO: checking whether ${MAKE-make} sets \$(MAKE)" >&5
 $as_echo_n "checking whether ${MAKE-make} sets \$(MAKE)... " >&6; }
 set x ${MAKE-make}
 ac_make=`$as_echo "$2" | sed 's/+/p/g; s/[^a-zA-Z0-9_]/_/g'`
-if eval \${ac_cv_prog_make_${ac_make}_set+:} false; then :
+if { as_var=ac_cv_prog_make_${ac_make}_set; eval "test \"\${$as_var+set}\" = set"; }; then
   $as_echo_n "(cached) " >&6
 else
   cat >conftest.make <<\_ACEOF
@@ -2880,7 +2430,7 @@ SHELL = /bin/sh
 all:
 	@echo '@@@%%%=$(MAKE)=@@@%%%'
 _ACEOF
-# GNU make sometimes prints "make[1]: Entering ...", which would confuse us.
+# GNU make sometimes prints "make[1]: Entering...", which would confuse us.
 case `${MAKE-make} -f conftest.make 2>/dev/null` in
   *@@@%%%=?*=@@@%%%*)
     eval ac_cv_prog_make_${ac_make}_set=yes;;
@@ -2890,11 +2440,11 @@ esac
 rm -f conftest.make
 fi
 if eval test \$ac_cv_prog_make_${ac_make}_set = yes; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+  { $as_echo "$as_me:$LINENO: result: yes" >&5
 $as_echo "yes" >&6; }
   SET_MAKE=
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
   SET_MAKE="MAKE=${MAKE-make}"
 fi
@@ -2908,52 +2458,15 @@ else
 fi
 rmdir .tst 2>/dev/null
 
-# Check whether --enable-silent-rules was given.
-if test "${enable_silent_rules+set}" = set; then :
-  enableval=$enable_silent_rules;
-fi
-
-case $enable_silent_rules in # (((
-  yes) AM_DEFAULT_VERBOSITY=0;;
-   no) AM_DEFAULT_VERBOSITY=1;;
-    *) AM_DEFAULT_VERBOSITY=1;;
-esac
-am_make=${MAKE-make}
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $am_make supports nested variables" >&5
-$as_echo_n "checking whether $am_make supports nested variables... " >&6; }
-if ${am_cv_make_support_nested_variables+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if $as_echo 'TRUE=$(BAR$(V))
-BAR0=false
-BAR1=true
-V=1
-am__doit:
-	@$(TRUE)
-.PHONY: am__doit' | $am_make -f - >/dev/null 2>&1; then
-  am_cv_make_support_nested_variables=yes
-else
-  am_cv_make_support_nested_variables=no
-fi
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_make_support_nested_variables" >&5
-$as_echo "$am_cv_make_support_nested_variables" >&6; }
-if test $am_cv_make_support_nested_variables = yes; then
-    AM_V='$(V)'
-  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
-else
-  AM_V=$AM_DEFAULT_VERBOSITY
-  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
-fi
-AM_BACKSLASH='\'
-
 if test "`cd $srcdir && pwd`" != "`pwd`"; then
   # Use -I$(srcdir) only when $(srcdir) != ., so that make's output
   # is not polluted with repeated "-I."
   am__isrc=' -I$(srcdir)'
   # test to see if srcdir already configured
   if test -f $srcdir/config.status; then
-    as_fn_error $? "source directory already configured; run \"make distclean\" there first" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: source directory already configured; run \"make distclean\" there first" >&5
+$as_echo "$as_me: error: source directory already configured; run \"make distclean\" there first" >&2;}
+   { (exit 1); exit 1; }; }
   fi
 fi
 
@@ -2997,115 +2510,36 @@ AUTOHEADER=${AUTOHEADER-"${am_missing_run}autoheader"}
 
 MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 
-# For better backward compatibility.  To be removed once Automake 1.9.x
-# dies out for good.  For more background, see:
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
-mkdir_p='$(MKDIR_P)'
-
 # We need awk for the "check" target.  The system "awk" is bad on
 # some platforms.
-# Always define AMTAR for backward compatibility.  Yes, it's still used
-# in the wild :-(  We should find a proper way to deprecate it ...
-AMTAR='$${TAR-tar}'
+# Always define AMTAR for backward compatibility.
 
+AMTAR=${AMTAR-"${am_missing_run}tar"}
 
-# We'll loop over all known methods to create a tar archive until one works.
-_am_tools='gnutar  pax cpio none'
-
-am__tar='$${TAR-tar} chof - "$$tardir"' am__untar='$${TAR-tar} xf -'
+am__tar='${AMTAR} chof - "$$tardir"'; am__untar='${AMTAR} xf -'
 
 
 
 
 
 
-# POSIX will say in a future version that running "rm -f" with no argument
-# is OK; and we want to be able to make that assumption in our Makefile
-# recipes.  So use an aggressive probe to check that the usage we want is
-# actually supported "in the wild" to an acceptable degree.
-# See automake bug#10828.
-# To make any issue more visible, cause the running configure to be aborted
-# by default if the 'rm' program in use doesn't match our expectations; the
-# user can still override this though.
-if rm -f && rm -fr && rm -rf; then : OK; else
-  cat >&2 <<'END'
-Oops!
-
-Your 'rm' program seems unable to run without file operands specified
-on the command line, even when the '-f' option is present.  This is contrary
-to the behaviour of most rm programs out there, and not conforming with
-the upcoming POSIX standard: <http://austingroupbugs.net/view.php?id=542>
-
-Please tell bug-automake@gnu.org about your system, including the value
-of your $PATH and any error possibly output before this message.  This
-can help us improve future automake versions.
-
-END
-  if test x"$ACCEPT_INFERIOR_RM_PROGRAM" = x"yes"; then
-    echo 'Configuration will proceed anyway, since you have set the' >&2
-    echo 'ACCEPT_INFERIOR_RM_PROGRAM variable to "yes"' >&2
-    echo >&2
-  else
-    cat >&2 <<'END'
-Aborting the configuration process, to ensure you take notice of the issue.
-
-You can download and install GNU coreutils to get an 'rm' implementation
-that behaves properly: <http://www.gnu.org/software/coreutils/>.
-
-If you want to complete the configuration process using your problematic
-'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
-to "yes", and re-run configure.
-
-END
-    as_fn_error $? "Your 'rm' program is bad, sorry." "$LINENO" 5
-  fi
-fi
-
 
 ac_config_headers="$ac_config_headers lib/Grid_config.h"
 
 # Check whether --enable-silent-rules was given.
-if test "${enable_silent_rules+set}" = set; then :
+if test "${enable_silent_rules+set}" = set; then
   enableval=$enable_silent_rules;
 fi
 
-case $enable_silent_rules in # (((
-  yes) AM_DEFAULT_VERBOSITY=0;;
-   no) AM_DEFAULT_VERBOSITY=1;;
-    *) AM_DEFAULT_VERBOSITY=0;;
+case $enable_silent_rules in
+yes) AM_DEFAULT_VERBOSITY=0;;
+no)  AM_DEFAULT_VERBOSITY=1;;
+*)   AM_DEFAULT_VERBOSITY=0;;
 esac
-am_make=${MAKE-make}
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $am_make supports nested variables" >&5
-$as_echo_n "checking whether $am_make supports nested variables... " >&6; }
-if ${am_cv_make_support_nested_variables+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if $as_echo 'TRUE=$(BAR$(V))
-BAR0=false
-BAR1=true
-V=1
-am__doit:
-	@$(TRUE)
-.PHONY: am__doit' | $am_make -f - >/dev/null 2>&1; then
-  am_cv_make_support_nested_variables=yes
-else
-  am_cv_make_support_nested_variables=no
-fi
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_make_support_nested_variables" >&5
-$as_echo "$am_cv_make_support_nested_variables" >&6; }
-if test $am_cv_make_support_nested_variables = yes; then
-    AM_V='$(V)'
-  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
-else
-  AM_V=$AM_DEFAULT_VERBOSITY
-  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
-fi
 AM_BACKSLASH='\'
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}:
+{ $as_echo "$as_me:$LINENO:
 
 :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
 Configuring $PACKAGE v$VERSION  for $host
@@ -3139,9 +2573,9 @@ if test -z "$CXX"; then
   do
     # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
 set dummy $ac_tool_prefix$ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CXX+:} false; then :
+if test "${ac_cv_prog_CXX+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CXX"; then
@@ -3152,24 +2586,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_CXX="$ac_tool_prefix$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 CXX=$ac_cv_prog_CXX
 if test -n "$CXX"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CXX" >&5
+  { $as_echo "$as_me:$LINENO: result: $CXX" >&5
 $as_echo "$CXX" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3183,9 +2617,9 @@ if test -z "$CXX"; then
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_CXX+:} false; then :
+if test "${ac_cv_prog_ac_ct_CXX+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CXX"; then
@@ -3196,24 +2630,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_ac_ct_CXX="$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CXX=$ac_cv_prog_ac_ct_CXX
 if test -n "$ac_ct_CXX"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CXX" >&5
+  { $as_echo "$as_me:$LINENO: result: $ac_ct_CXX" >&5
 $as_echo "$ac_ct_CXX" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3226,7 +2660,7 @@ done
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -3237,31 +2671,48 @@ fi
   fi
 fi
 # Provide some information about the compiler.
-$as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler version" >&5
+$as_echo "$as_me:$LINENO: checking for C++ compiler version" >&5
 set X $ac_compile
 ac_compiler=$2
-for ac_option in --version -v -V -qversion; do
-  { { ac_try="$ac_compiler $ac_option >&5"
+{ (ac_try="$ac_compiler --version >&5"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler --version >&5") 2>&5
   ac_status=$?
-  if test -s conftest.err; then
-    sed '10a\
-... rest of stderr output deleted ...
-         10q' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-  fi
-  rm -f conftest.er1 conftest.err
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }
-done
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
+{ (ac_try="$ac_compiler -v >&5"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler -v >&5") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
+{ (ac_try="$ac_compiler -V >&5"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compiler -V >&5") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }
 
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3277,8 +2728,8 @@ ac_clean_files="$ac_clean_files a.out a.out.dSYM a.exe b.out"
 # Try to create an executable without -o first, disregard a.out.
 # It will help us diagnose broken compilers, and finding out an intuition
 # of exeext.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether the C++ compiler works" >&5
-$as_echo_n "checking whether the C++ compiler works... " >&6; }
+{ $as_echo "$as_me:$LINENO: checking for C++ compiler default output file name" >&5
+$as_echo_n "checking for C++ compiler default output file name... " >&6; }
 ac_link_default=`$as_echo "$ac_link" | sed 's/ -o *conftest[^ ]*//'`
 
 # The possible output files:
@@ -3294,17 +2745,17 @@ do
 done
 rm -f $ac_rmfiles
 
-if { { ac_try="$ac_link_default"
+if { (ac_try="$ac_link_default"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
   (eval "$ac_link_default") 2>&5
   ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; then :
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; then
   # Autoconf-2.13 could set the ac_cv_exeext variable to `no'.
 # So ignore a value of `no', otherwise this would lead to `EXEEXT = no'
 # in a Makefile.  We should not override ac_cv_exeext if it was cached,
@@ -3321,7 +2772,7 @@ do
 	# certainly right.
 	break;;
     *.* )
-	if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
+        if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
 	then :; else
 	   ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
 	fi
@@ -3340,41 +2791,84 @@ test "$ac_cv_exeext" = no && ac_cv_exeext=
 else
   ac_file=''
 fi
-if test -z "$ac_file"; then :
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-$as_echo "$as_me: failed program was:" >&5
+
+{ $as_echo "$as_me:$LINENO: result: $ac_file" >&5
+$as_echo "$ac_file" >&6; }
+if test -z "$ac_file"; then
+  $as_echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
 
-{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error 77 "C++ compiler cannot create executables
-See \`config.log' for more details" "$LINENO" 5; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
-$as_echo "yes" >&6; }
+{ { $as_echo "$as_me:$LINENO: error: C++ compiler cannot create executables
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: C++ compiler cannot create executables
+See \`config.log' for more details." >&2;}
+   { (exit 77); exit 77; }; }; }
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler default output file name" >&5
-$as_echo_n "checking for C++ compiler default output file name... " >&6; }
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_file" >&5
-$as_echo "$ac_file" >&6; }
+
 ac_exeext=$ac_cv_exeext
 
+# Check that the compiler produces executables we can run.  If not, either
+# the compiler is broken, or we cross compile.
+{ $as_echo "$as_me:$LINENO: checking whether the C++ compiler works" >&5
+$as_echo_n "checking whether the C++ compiler works... " >&6; }
+# FIXME: These cross compiler hacks should be removed for Autoconf 3.0
+# If not cross compiling, check that we can run a simple program.
+if test "$cross_compiling" != yes; then
+  if { ac_try='./$ac_file'
+  { (case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; }; then
+    cross_compiling=no
+  else
+    if test "$cross_compiling" = maybe; then
+	cross_compiling=yes
+    else
+	{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+{ { $as_echo "$as_me:$LINENO: error: cannot run C++ compiled programs.
+If you meant to cross compile, use \`--host'.
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: cannot run C++ compiled programs.
+If you meant to cross compile, use \`--host'.
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
+    fi
+  fi
+fi
+{ $as_echo "$as_me:$LINENO: result: yes" >&5
+$as_echo "yes" >&6; }
+
 rm -f -r a.out a.out.dSYM a.exe conftest$ac_cv_exeext b.out
 ac_clean_files=$ac_clean_files_save
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of executables" >&5
+# Check that the compiler produces executables we can run.  If not, either
+# the compiler is broken, or we cross compile.
+{ $as_echo "$as_me:$LINENO: checking whether we are cross compiling" >&5
+$as_echo_n "checking whether we are cross compiling... " >&6; }
+{ $as_echo "$as_me:$LINENO: result: $cross_compiling" >&5
+$as_echo "$cross_compiling" >&6; }
+
+{ $as_echo "$as_me:$LINENO: checking for suffix of executables" >&5
 $as_echo_n "checking for suffix of executables... " >&6; }
-if { { ac_try="$ac_link"
+if { (ac_try="$ac_link"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
   (eval "$ac_link") 2>&5
   ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; then :
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; then
   # If both `conftest.exe' and `conftest' are `present' (well, observable)
 # catch `conftest.exe'.  For instance with Cygwin, `ls conftest' will
 # work properly (i.e., refer to `conftest.exe'), while it won't with
@@ -3389,83 +2883,32 @@ for ac_file in conftest.exe conftest conftest.*; do
   esac
 done
 else
-  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+  { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "cannot compute suffix of executables: cannot compile and link
-See \`config.log' for more details" "$LINENO" 5; }
+{ { $as_echo "$as_me:$LINENO: error: cannot compute suffix of executables: cannot compile and link
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: cannot compute suffix of executables: cannot compile and link
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
 fi
-rm -f conftest conftest$ac_cv_exeext
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_exeext" >&5
+
+rm -f conftest$ac_cv_exeext
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_exeext" >&5
 $as_echo "$ac_cv_exeext" >&6; }
 
 rm -f conftest.$ac_ext
 EXEEXT=$ac_cv_exeext
 ac_exeext=$EXEEXT
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <stdio.h>
-int
-main ()
-{
-FILE *f = fopen ("conftest.out", "w");
- return ferror (f) || fclose (f) != 0;
-
-  ;
-  return 0;
-}
-_ACEOF
-ac_clean_files="$ac_clean_files conftest.out"
-# Check that the compiler produces executables we can run.  If not, either
-# the compiler is broken, or we cross compile.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are cross compiling" >&5
-$as_echo_n "checking whether we are cross compiling... " >&6; }
-if test "$cross_compiling" != yes; then
-  { { ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_link") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }
-  if { ac_try='./conftest$ac_cv_exeext'
-  { { case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; }; then
-    cross_compiling=no
-  else
-    if test "$cross_compiling" = maybe; then
-	cross_compiling=yes
-    else
-	{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "cannot run C++ compiled programs.
-If you meant to cross compile, use \`--host'.
-See \`config.log' for more details" "$LINENO" 5; }
-    fi
-  fi
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $cross_compiling" >&5
-$as_echo "$cross_compiling" >&6; }
-
-rm -f conftest.$ac_ext conftest$ac_cv_exeext conftest.out
-ac_clean_files=$ac_clean_files_save
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of object files" >&5
+{ $as_echo "$as_me:$LINENO: checking for suffix of object files" >&5
 $as_echo_n "checking for suffix of object files... " >&6; }
-if ${ac_cv_objext+:} false; then :
+if test "${ac_cv_objext+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3477,17 +2920,17 @@ main ()
 }
 _ACEOF
 rm -f conftest.o conftest.obj
-if { { ac_try="$ac_compile"
+if { (ac_try="$ac_compile"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
   (eval "$ac_compile") 2>&5
   ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; then :
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; then
   for ac_file in conftest.o conftest.obj conftest.*; do
   test -f "$ac_file" || continue;
   case $ac_file in
@@ -3500,23 +2943,31 @@ else
   $as_echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
 
-{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "cannot compute suffix of object files: cannot compile
-See \`config.log' for more details" "$LINENO" 5; }
+{ { $as_echo "$as_me:$LINENO: error: cannot compute suffix of object files: cannot compile
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: cannot compute suffix of object files: cannot compile
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
 fi
+
 rm -f conftest.$ac_cv_objext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_objext" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_objext" >&5
 $as_echo "$ac_cv_objext" >&6; }
 OBJEXT=$ac_cv_objext
 ac_objext=$OBJEXT
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C++ compiler" >&5
+{ $as_echo "$as_me:$LINENO: checking whether we are using the GNU C++ compiler" >&5
 $as_echo_n "checking whether we are using the GNU C++ compiler... " >&6; }
-if ${ac_cv_cxx_compiler_gnu+:} false; then :
+if test "${ac_cv_cxx_compiler_gnu+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3530,16 +2981,37 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_compiler_gnu=yes
 else
-  ac_compiler_gnu=no
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_compiler_gnu=no
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 ac_cv_cxx_compiler_gnu=$ac_compiler_gnu
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_cxx_compiler_gnu" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_cxx_compiler_gnu" >&5
 $as_echo "$ac_cv_cxx_compiler_gnu" >&6; }
 if test $ac_compiler_gnu = yes; then
   GXX=yes
@@ -3548,16 +3020,20 @@ else
 fi
 ac_test_CXXFLAGS=${CXXFLAGS+set}
 ac_save_CXXFLAGS=$CXXFLAGS
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX accepts -g" >&5
+{ $as_echo "$as_me:$LINENO: checking whether $CXX accepts -g" >&5
 $as_echo_n "checking whether $CXX accepts -g... " >&6; }
-if ${ac_cv_prog_cxx_g+:} false; then :
+if test "${ac_cv_prog_cxx_g+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   ac_save_cxx_werror_flag=$ac_cxx_werror_flag
    ac_cxx_werror_flag=yes
    ac_cv_prog_cxx_g=no
    CXXFLAGS="-g"
-   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+   cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3568,11 +3044,35 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_prog_cxx_g=yes
 else
-  CXXFLAGS=""
-      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	CXXFLAGS=""
+      cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3583,12 +3083,36 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  :
 else
-  ac_cxx_werror_flag=$ac_save_cxx_werror_flag
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cxx_werror_flag=$ac_save_cxx_werror_flag
 	 CXXFLAGS="-g"
-	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+	 cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 int
@@ -3599,17 +3123,42 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_prog_cxx_g=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
    ac_cxx_werror_flag=$ac_save_cxx_werror_flag
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cxx_g" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cxx_g" >&5
 $as_echo "$ac_cv_prog_cxx_g" >&6; }
 if test "$ac_test_CXXFLAGS" = set; then
   CXXFLAGS=$ac_save_CXXFLAGS
@@ -3643,14 +3192,14 @@ am__doit:
 .PHONY: am__doit
 END
 # If we don't find an include directive, just comment out the code.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for style of include used by $am_make" >&5
+{ $as_echo "$as_me:$LINENO: checking for style of include used by $am_make" >&5
 $as_echo_n "checking for style of include used by $am_make... " >&6; }
 am__include="#"
 am__quote=
 _am_result=none
 # First try GNU make style include.
 echo "include confinc" > confmf
-# Ignore all kinds of additional output from 'make'.
+# Ignore all kinds of additional output from `make'.
 case `$am_make -s -f confmf 2> /dev/null` in #(
 *the\ am__doit\ target*)
   am__include=include
@@ -3671,19 +3220,18 @@ if test "$am__include" = "#"; then
 fi
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $_am_result" >&5
+{ $as_echo "$as_me:$LINENO: result: $_am_result" >&5
 $as_echo "$_am_result" >&6; }
 rm -f confinc confmf
 
 # Check whether --enable-dependency-tracking was given.
-if test "${enable_dependency_tracking+set}" = set; then :
+if test "${enable_dependency_tracking+set}" = set; then
   enableval=$enable_dependency_tracking;
 fi
 
 if test "x$enable_dependency_tracking" != xno; then
   am_depcomp="$ac_aux_dir/depcomp"
   AMDEPBACKSLASH='\'
-  am__nodep='_no'
 fi
  if test "x$enable_dependency_tracking" != xno; then
   AMDEP_TRUE=
@@ -3697,18 +3245,17 @@ fi
 
 depcc="$CXX"  am_compiler_list=
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
+{ $as_echo "$as_me:$LINENO: checking dependency style of $depcc" >&5
 $as_echo_n "checking dependency style of $depcc... " >&6; }
-if ${am_cv_CXX_dependencies_compiler_type+:} false; then :
+if test "${am_cv_CXX_dependencies_compiler_type+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named 'D' -- because '-MD' means "put the output
-  # in D".
-  rm -rf conftest.dir
+  # making a dummy file named `D' -- because `-MD' means `put the output
+  # in D'.
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -3742,16 +3289,16 @@ else
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
-      # Solaris 10 /bin/sh.
-      echo '/* dummy */' > sub/conftst$i.h
+      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
+      # Solaris 8's {/usr,}/bin/sh.
+      touch sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with '-c' and '-o' for the sake of the "dashmstdout"
+    # We check with `-c' and `-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle '-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs.
+    # handle `-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -3760,16 +3307,16 @@ else
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # After this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested.
+      # after this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
-      # This compiler won't grok '-c -o', but also, the minuso test has
+    msvisualcpp | msvcmsys)
+      # This compiler won't grok `-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -3808,7 +3355,7 @@ else
 fi
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CXX_dependencies_compiler_type" >&5
+{ $as_echo "$as_me:$LINENO: result: $am_cv_CXX_dependencies_compiler_type" >&5
 $as_echo "$am_cv_CXX_dependencies_compiler_type" >&6; }
 CXXDEPMODE=depmode=$am_cv_CXX_dependencies_compiler_type
 
@@ -3827,18 +3374,17 @@ fi
 
   OPENMP_CXXFLAGS=
   # Check whether --enable-openmp was given.
-if test "${enable_openmp+set}" = set; then :
+if test "${enable_openmp+set}" = set; then
   enableval=$enable_openmp;
 fi
 
   if test "$enable_openmp" != no; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CXX option to support OpenMP" >&5
-$as_echo_n "checking for $CXX option to support OpenMP... " >&6; }
-if ${ac_cv_prog_cxx_openmp+:} false; then :
+    { $as_echo "$as_me:$LINENO: checking for $CC option to support OpenMP" >&5
+$as_echo_n "checking for $CC option to support OpenMP... " >&6; }
+if test "${ac_cv_prog_cxx_openmp+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
+  cat >conftest.$ac_ext <<_ACEOF
 
 #ifndef _OPENMP
  choke me
@@ -3847,16 +3393,37 @@ else
 int main () { return omp_get_num_threads (); }
 
 _ACEOF
-if ac_fn_cxx_try_link "$LINENO"; then :
+rm -f conftest.$ac_objext conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_link") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 $as_test_x conftest$ac_exeext
+       }; then
   ac_cv_prog_cxx_openmp='none needed'
 else
-  ac_cv_prog_cxx_openmp='unsupported'
-	  	  	  	  	  	  	                                	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp -homp \
-                           -Popenmp --openmp; do
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cv_prog_cxx_openmp='unsupported'
+	  	  	  	  	  	  	  	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp; do
 	    ac_save_CXXFLAGS=$CXXFLAGS
 	    CXXFLAGS="$CXXFLAGS $ac_option"
-	    cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
+	    cat >conftest.$ac_ext <<_ACEOF
 
 #ifndef _OPENMP
  choke me
@@ -3865,27 +3432,56 @@ else
 int main () { return omp_get_num_threads (); }
 
 _ACEOF
-if ac_fn_cxx_try_link "$LINENO"; then :
+rm -f conftest.$ac_objext conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_link") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 $as_test_x conftest$ac_exeext
+       }; then
   ac_cv_prog_cxx_openmp=$ac_option
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
 fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
+
+rm -rf conftest.dSYM
+rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
+      conftest$ac_exeext conftest.$ac_ext
 	    CXXFLAGS=$ac_save_CXXFLAGS
 	    if test "$ac_cv_prog_cxx_openmp" != unsupported; then
 	      break
 	    fi
 	  done
 fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
+
+rm -rf conftest.dSYM
+rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
+      conftest$ac_exeext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cxx_openmp" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cxx_openmp" >&5
 $as_echo "$ac_cv_prog_cxx_openmp" >&6; }
     case $ac_cv_prog_cxx_openmp in #(
       "none needed" | unsupported)
-	;; #(
+        ;; #(
       *)
-	OPENMP_CXXFLAGS=$ac_cv_prog_cxx_openmp ;;
+        OPENMP_CXXFLAGS=$ac_cv_prog_cxx_openmp ;;
     esac
   fi
 
@@ -3893,9 +3489,9 @@ $as_echo "$ac_cv_prog_cxx_openmp" >&6; }
 if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}ranlib", so it can be a program name with args.
 set dummy ${ac_tool_prefix}ranlib; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_RANLIB+:} false; then :
+if test "${ac_cv_prog_RANLIB+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$RANLIB"; then
@@ -3906,24 +3502,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_RANLIB="${ac_tool_prefix}ranlib"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 RANLIB=$ac_cv_prog_RANLIB
 if test -n "$RANLIB"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $RANLIB" >&5
+  { $as_echo "$as_me:$LINENO: result: $RANLIB" >&5
 $as_echo "$RANLIB" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3933,9 +3529,9 @@ if test -z "$ac_cv_prog_RANLIB"; then
   ac_ct_RANLIB=$RANLIB
   # Extract the first word of "ranlib", so it can be a program name with args.
 set dummy ranlib; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_RANLIB+:} false; then :
+if test "${ac_cv_prog_ac_ct_RANLIB+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_RANLIB"; then
@@ -3946,24 +3542,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_ac_ct_RANLIB="ranlib"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_RANLIB=$ac_cv_prog_ac_ct_RANLIB
 if test -n "$ac_ct_RANLIB"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_RANLIB" >&5
+  { $as_echo "$as_me:$LINENO: result: $ac_ct_RANLIB" >&5
 $as_echo "$ac_ct_RANLIB" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3972,7 +3568,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -3989,12 +3585,16 @@ ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
 ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
 ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
   ac_success=no
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX supports C++11 features by default" >&5
+  { $as_echo "$as_me:$LINENO: checking whether $CXX supports C++11 features by default" >&5
 $as_echo_n "checking whether $CXX supports C++11 features by default... " >&6; }
-if ${ax_cv_cxx_compile_cxx11+:} false; then :
+if test "${ax_cv_cxx_compile_cxx11+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
   template <typename T>
@@ -4046,14 +3646,35 @@ else
     }
 
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ax_cv_cxx_compile_cxx11=yes
 else
-  ax_cv_cxx_compile_cxx11=no
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ax_cv_cxx_compile_cxx11=no
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_cxx_compile_cxx11" >&5
+{ $as_echo "$as_me:$LINENO: result: $ax_cv_cxx_compile_cxx11" >&5
 $as_echo "$ax_cv_cxx_compile_cxx11" >&6; }
   if test x$ax_cv_cxx_compile_cxx11 = xyes; then
     ac_success=yes
@@ -4064,14 +3685,18 @@ $as_echo "$ax_cv_cxx_compile_cxx11" >&6; }
     if test x$ac_success = xno; then
             for switch in -std=c++11 -std=c++0x +std=c++11; do
       cachevar=`$as_echo "ax_cv_cxx_compile_cxx11_$switch" | $as_tr_sh`
-      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX supports C++11 features with $switch" >&5
+      { $as_echo "$as_me:$LINENO: checking whether $CXX supports C++11 features with $switch" >&5
 $as_echo_n "checking whether $CXX supports C++11 features with $switch... " >&6; }
-if eval \${$cachevar+:} false; then :
+if { as_var=$cachevar; eval "test \"\${$as_var+set}\" = set"; }; then
   $as_echo_n "(cached) " >&6
 else
   ac_save_CXXFLAGS="$CXXFLAGS"
          CXXFLAGS="$CXXFLAGS $switch"
-         cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+         cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
   template <typename T>
@@ -4123,16 +3748,38 @@ else
     }
 
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   eval $cachevar=yes
 else
-  eval $cachevar=no
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	eval $cachevar=no
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
          CXXFLAGS="$ac_save_CXXFLAGS"
 fi
-eval ac_res=\$$cachevar
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+ac_res=`eval 'as_val=${'$cachevar'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
 $as_echo "$ac_res" >&6; }
       if eval test x\$$cachevar = xyes; then
         CXXFLAGS="$CXXFLAGS $switch"
@@ -4149,17 +3796,21 @@ ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
 
   if test x$ax_cxx_compile_cxx11_required = xtrue; then
     if test x$ac_success = xno; then
-      as_fn_error $? "*** A compiler with support for C++11 language features is required." "$LINENO" 5
+      { { $as_echo "$as_me:$LINENO: error: *** A compiler with support for C++11 language features is required." >&5
+$as_echo "$as_me: error: *** A compiler with support for C++11 language features is required." >&2;}
+   { (exit 1); exit 1; }; }
     fi
   else
     if test x$ac_success = xno; then
       HAVE_CXX11=0
-      { $as_echo "$as_me:${as_lineno-$LINENO}: No compiler with C++11 support was found" >&5
+      { $as_echo "$as_me:$LINENO: No compiler with C++11 support was found" >&5
 $as_echo "$as_me: No compiler with C++11 support was found" >&6;}
     else
       HAVE_CXX11=1
 
-$as_echo "#define HAVE_CXX11 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define HAVE_CXX11 1
+_ACEOF
 
     fi
 
@@ -4177,10 +3828,10 @@ ac_cpp='$CXXCPP $CPPFLAGS'
 ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
 ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
 ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking how to run the C++ preprocessor" >&5
+{ $as_echo "$as_me:$LINENO: checking how to run the C++ preprocessor" >&5
 $as_echo_n "checking how to run the C++ preprocessor... " >&6; }
 if test -z "$CXXCPP"; then
-  if ${ac_cv_prog_CXXCPP+:} false; then :
+  if test "${ac_cv_prog_CXXCPP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
       # Double quotes because CXXCPP needs to be expanded
@@ -4195,7 +3846,11 @@ do
   # <limits.h> exists even on freestanding compilers.
   # On the NeXT, cc -E runs the code through the compiler's parser,
   # not just through cpp. "Syntax error" is here to catch this case.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #ifdef __STDC__
 # include <limits.h>
@@ -4204,34 +3859,78 @@ do
 #endif
 		     Syntax error
 _ACEOF
-if ac_fn_cxx_try_cpp "$LINENO"; then :
-
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  :
 else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
   # Broken: fails on valid input.
 continue
 fi
-rm -f conftest.err conftest.i conftest.$ac_ext
+
+rm -f conftest.err conftest.$ac_ext
 
   # OK, works on sane cases.  Now check whether nonexistent headers
   # can be detected and how.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <ac_nonexistent.h>
 _ACEOF
-if ac_fn_cxx_try_cpp "$LINENO"; then :
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
   # Broken: success on invalid input.
 continue
 else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
   # Passes both tests.
 ac_preproc_ok=:
 break
 fi
-rm -f conftest.err conftest.i conftest.$ac_ext
+
+rm -f conftest.err conftest.$ac_ext
 
 done
 # Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
-rm -f conftest.i conftest.err conftest.$ac_ext
-if $ac_preproc_ok; then :
+rm -f conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then
   break
 fi
 
@@ -4243,7 +3942,7 @@ fi
 else
   ac_cv_prog_CXXCPP=$CXXCPP
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $CXXCPP" >&5
+{ $as_echo "$as_me:$LINENO: result: $CXXCPP" >&5
 $as_echo "$CXXCPP" >&6; }
 ac_preproc_ok=false
 for ac_cxx_preproc_warn_flag in '' yes
@@ -4254,7 +3953,11 @@ do
   # <limits.h> exists even on freestanding compilers.
   # On the NeXT, cc -E runs the code through the compiler's parser,
   # not just through cpp. "Syntax error" is here to catch this case.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #ifdef __STDC__
 # include <limits.h>
@@ -4263,40 +3966,87 @@ do
 #endif
 		     Syntax error
 _ACEOF
-if ac_fn_cxx_try_cpp "$LINENO"; then :
-
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  :
 else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
   # Broken: fails on valid input.
 continue
 fi
-rm -f conftest.err conftest.i conftest.$ac_ext
+
+rm -f conftest.err conftest.$ac_ext
 
   # OK, works on sane cases.  Now check whether nonexistent headers
   # can be detected and how.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <ac_nonexistent.h>
 _ACEOF
-if ac_fn_cxx_try_cpp "$LINENO"; then :
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
   # Broken: success on invalid input.
 continue
 else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
   # Passes both tests.
 ac_preproc_ok=:
 break
 fi
-rm -f conftest.err conftest.i conftest.$ac_ext
+
+rm -f conftest.err conftest.$ac_ext
 
 done
 # Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
-rm -f conftest.i conftest.err conftest.$ac_ext
-if $ac_preproc_ok; then :
-
+rm -f conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then
+  :
 else
-  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+  { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "C++ preprocessor \"$CXXCPP\" fails sanity check
-See \`config.log' for more details" "$LINENO" 5; }
+{ { $as_echo "$as_me:$LINENO: error: C++ preprocessor \"$CXXCPP\" fails sanity check
+See \`config.log' for more details." >&5
+$as_echo "$as_me: error: C++ preprocessor \"$CXXCPP\" fails sanity check
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }; }
 fi
 
 ac_ext=cpp
@@ -4306,9 +4056,9 @@ ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ex
 ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for grep that handles long lines and -e" >&5
+{ $as_echo "$as_me:$LINENO: checking for grep that handles long lines and -e" >&5
 $as_echo_n "checking for grep that handles long lines and -e... " >&6; }
-if ${ac_cv_path_GREP+:} false; then :
+if test "${ac_cv_path_GREP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -z "$GREP"; then
@@ -4319,10 +4069,10 @@ for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_prog in grep ggrep; do
+  for ac_prog in grep ggrep; do
     for ac_exec_ext in '' $ac_executable_extensions; do
       ac_path_GREP="$as_dir/$ac_prog$ac_exec_ext"
-      as_fn_executable_p "$ac_path_GREP" || continue
+      { test -f "$ac_path_GREP" && $as_test_x "$ac_path_GREP"; } || continue
 # Check for GNU ac_path_GREP and select it if it is found.
   # Check for GNU $ac_path_GREP
 case `"$ac_path_GREP" --version 2>&1` in
@@ -4339,7 +4089,7 @@ case `"$ac_path_GREP" --version 2>&1` in
     $as_echo 'GREP' >> "conftest.nl"
     "$ac_path_GREP" -e 'GREP$' -e '-(cannot match)-' < "conftest.nl" >"conftest.out" 2>/dev/null || break
     diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
-    as_fn_arith $ac_count + 1 && ac_count=$as_val
+    ac_count=`expr $ac_count + 1`
     if test $ac_count -gt ${ac_path_GREP_max-0}; then
       # Best one so far, save it but keep looking for a better one
       ac_cv_path_GREP="$ac_path_GREP"
@@ -4354,24 +4104,26 @@ esac
       $ac_path_GREP_found && break 3
     done
   done
-  done
+done
 IFS=$as_save_IFS
   if test -z "$ac_cv_path_GREP"; then
-    as_fn_error $? "no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&5
+$as_echo "$as_me: error: no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&2;}
+   { (exit 1); exit 1; }; }
   fi
 else
   ac_cv_path_GREP=$GREP
 fi
 
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_GREP" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_path_GREP" >&5
 $as_echo "$ac_cv_path_GREP" >&6; }
  GREP="$ac_cv_path_GREP"
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for egrep" >&5
+{ $as_echo "$as_me:$LINENO: checking for egrep" >&5
 $as_echo_n "checking for egrep... " >&6; }
-if ${ac_cv_path_EGREP+:} false; then :
+if test "${ac_cv_path_EGREP+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if echo a | $GREP -E '(a|b)' >/dev/null 2>&1
@@ -4385,10 +4137,10 @@ for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_prog in egrep; do
+  for ac_prog in egrep; do
     for ac_exec_ext in '' $ac_executable_extensions; do
       ac_path_EGREP="$as_dir/$ac_prog$ac_exec_ext"
-      as_fn_executable_p "$ac_path_EGREP" || continue
+      { test -f "$ac_path_EGREP" && $as_test_x "$ac_path_EGREP"; } || continue
 # Check for GNU ac_path_EGREP and select it if it is found.
   # Check for GNU $ac_path_EGREP
 case `"$ac_path_EGREP" --version 2>&1` in
@@ -4405,7 +4157,7 @@ case `"$ac_path_EGREP" --version 2>&1` in
     $as_echo 'EGREP' >> "conftest.nl"
     "$ac_path_EGREP" 'EGREP$' < "conftest.nl" >"conftest.out" 2>/dev/null || break
     diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
-    as_fn_arith $ac_count + 1 && ac_count=$as_val
+    ac_count=`expr $ac_count + 1`
     if test $ac_count -gt ${ac_path_EGREP_max-0}; then
       # Best one so far, save it but keep looking for a better one
       ac_cv_path_EGREP="$ac_path_EGREP"
@@ -4420,10 +4172,12 @@ esac
       $ac_path_EGREP_found && break 3
     done
   done
-  done
+done
 IFS=$as_save_IFS
   if test -z "$ac_cv_path_EGREP"; then
-    as_fn_error $? "no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&5
+$as_echo "$as_me: error: no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&2;}
+   { (exit 1); exit 1; }; }
   fi
 else
   ac_cv_path_EGREP=$EGREP
@@ -4431,17 +4185,21 @@ fi
 
    fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_EGREP" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_path_EGREP" >&5
 $as_echo "$ac_cv_path_EGREP" >&6; }
  EGREP="$ac_cv_path_EGREP"
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for ANSI C header files" >&5
+{ $as_echo "$as_me:$LINENO: checking for ANSI C header files" >&5
 $as_echo_n "checking for ANSI C header files... " >&6; }
-if ${ac_cv_header_stdc+:} false; then :
+if test "${ac_cv_header_stdc+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <stdlib.h>
 #include <stdarg.h>
@@ -4456,23 +4214,48 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
   ac_cv_header_stdc=yes
 else
-  ac_cv_header_stdc=no
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cv_header_stdc=no
 fi
+
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 
 if test $ac_cv_header_stdc = yes; then
   # SunOS 4.x string.h does not declare mem*, contrary to ANSI.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <string.h>
 
 _ACEOF
 if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
-  $EGREP "memchr" >/dev/null 2>&1; then :
-
+  $EGREP "memchr" >/dev/null 2>&1; then
+  :
 else
   ac_cv_header_stdc=no
 fi
@@ -4482,14 +4265,18 @@ fi
 
 if test $ac_cv_header_stdc = yes; then
   # ISC 2.0.2 stdlib.h does not declare free, contrary to ANSI.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <stdlib.h>
 
 _ACEOF
 if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
-  $EGREP "free" >/dev/null 2>&1; then :
-
+  $EGREP "free" >/dev/null 2>&1; then
+  :
 else
   ac_cv_header_stdc=no
 fi
@@ -4499,10 +4286,14 @@ fi
 
 if test $ac_cv_header_stdc = yes; then
   # /bin/cc in Irix-4.0.5 gets non-ANSI ctype macros unless using -ansi.
-  if test "$cross_compiling" = yes; then :
+  if test "$cross_compiling" = yes; then
   :
 else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 #include <ctype.h>
 #include <stdlib.h>
@@ -4529,33 +4320,118 @@ main ()
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_run "$LINENO"; then :
-
+rm -f conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && { ac_try='./conftest$ac_exeext'
+  { (case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; }; then
+  :
 else
-  ac_cv_header_stdc=no
+  $as_echo "$as_me: program exited with status $ac_status" >&5
+$as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+( exit $ac_status )
+ac_cv_header_stdc=no
 fi
-rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
-  conftest.$ac_objext conftest.beam conftest.$ac_ext
+rm -rf conftest.dSYM
+rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext conftest.$ac_objext conftest.$ac_ext
 fi
 
+
 fi
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_header_stdc" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_header_stdc" >&5
 $as_echo "$ac_cv_header_stdc" >&6; }
 if test $ac_cv_header_stdc = yes; then
 
-$as_echo "#define STDC_HEADERS 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define STDC_HEADERS 1
+_ACEOF
 
 fi
 
 # On IRIX 5.3, sys/types and inttypes.h are conflicting.
+
+
+
+
+
+
+
+
+
 for ac_header in sys/types.h sys/stat.h stdlib.h string.h memory.h strings.h \
 		  inttypes.h stdint.h unistd.h
-do :
-  as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-ac_fn_cxx_check_header_compile "$LINENO" "$ac_header" "$as_ac_Header" "$ac_includes_default
-"
-if eval test \"x\$"$as_ac_Header"\" = x"yes"; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  eval "$as_ac_Header=yes"
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	eval "$as_ac_Header=no"
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
 #define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
@@ -4565,107 +4441,1150 @@ fi
 done
 
 
+
 for ac_header in stdint.h
-do :
-  ac_fn_cxx_check_header_mongrel "$LINENO" "stdint.h" "ac_cv_header_stdint_h" "$ac_includes_default"
-if test "x$ac_cv_header_stdint_h" = xyes; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
+$as_echo_n "checking $ac_header usability... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_header_compiler=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_header_compiler=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
+$as_echo_n "checking $ac_header presence... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <$ac_header>
+_ACEOF
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  ac_header_preproc=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_header_preproc=no
+fi
+
+rm -f conftest.err conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_cxx_preproc_warn_flag in
+  yes:no: )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
+    ac_header_preproc=yes
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
+$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
+    ( cat <<\_ASBOX
+## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##
+_ASBOX
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  eval "$as_ac_Header=\$ac_header_preproc"
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+
+fi
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_STDINT_H 1
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
 
 fi
 
 done
 
+
 for ac_header in mm_malloc.h
-do :
-  ac_fn_cxx_check_header_mongrel "$LINENO" "mm_malloc.h" "ac_cv_header_mm_malloc_h" "$ac_includes_default"
-if test "x$ac_cv_header_mm_malloc_h" = xyes; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
+$as_echo_n "checking $ac_header usability... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_header_compiler=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_header_compiler=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
+$as_echo_n "checking $ac_header presence... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <$ac_header>
+_ACEOF
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  ac_header_preproc=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_header_preproc=no
+fi
+
+rm -f conftest.err conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_cxx_preproc_warn_flag in
+  yes:no: )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
+    ac_header_preproc=yes
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
+$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
+    ( cat <<\_ASBOX
+## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##
+_ASBOX
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  eval "$as_ac_Header=\$ac_header_preproc"
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+
+fi
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_MM_MALLOC_H 1
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
 
 fi
 
 done
 
+
 for ac_header in malloc/malloc.h
-do :
-  ac_fn_cxx_check_header_mongrel "$LINENO" "malloc/malloc.h" "ac_cv_header_malloc_malloc_h" "$ac_includes_default"
-if test "x$ac_cv_header_malloc_malloc_h" = xyes; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
+$as_echo_n "checking $ac_header usability... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_header_compiler=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_header_compiler=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
+$as_echo_n "checking $ac_header presence... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <$ac_header>
+_ACEOF
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  ac_header_preproc=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_header_preproc=no
+fi
+
+rm -f conftest.err conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_cxx_preproc_warn_flag in
+  yes:no: )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
+    ac_header_preproc=yes
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
+$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
+    ( cat <<\_ASBOX
+## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##
+_ASBOX
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  eval "$as_ac_Header=\$ac_header_preproc"
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+
+fi
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_MALLOC_MALLOC_H 1
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
 
 fi
 
 done
 
+
 for ac_header in malloc.h
-do :
-  ac_fn_cxx_check_header_mongrel "$LINENO" "malloc.h" "ac_cv_header_malloc_h" "$ac_includes_default"
-if test "x$ac_cv_header_malloc_h" = xyes; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
+$as_echo_n "checking $ac_header usability... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_header_compiler=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_header_compiler=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
+$as_echo_n "checking $ac_header presence... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <$ac_header>
+_ACEOF
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  ac_header_preproc=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_header_preproc=no
+fi
+
+rm -f conftest.err conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_cxx_preproc_warn_flag in
+  yes:no: )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
+    ac_header_preproc=yes
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
+$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
+    ( cat <<\_ASBOX
+## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##
+_ASBOX
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  eval "$as_ac_Header=\$ac_header_preproc"
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+
+fi
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_MALLOC_H 1
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
 
 fi
 
 done
 
+
 for ac_header in endian.h
-do :
-  ac_fn_cxx_check_header_mongrel "$LINENO" "endian.h" "ac_cv_header_endian_h" "$ac_includes_default"
-if test "x$ac_cv_header_endian_h" = xyes; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
+$as_echo_n "checking $ac_header usability... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_header_compiler=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_header_compiler=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
+$as_echo_n "checking $ac_header presence... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <$ac_header>
+_ACEOF
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  ac_header_preproc=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_header_preproc=no
+fi
+
+rm -f conftest.err conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_cxx_preproc_warn_flag in
+  yes:no: )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
+    ac_header_preproc=yes
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
+$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
+    ( cat <<\_ASBOX
+## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##
+_ASBOX
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  eval "$as_ac_Header=\$ac_header_preproc"
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+
+fi
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_ENDIAN_H 1
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
 
 fi
 
 done
 
+
 for ac_header in gmp.h
-do :
-  ac_fn_cxx_check_header_mongrel "$LINENO" "gmp.h" "ac_cv_header_gmp_h" "$ac_includes_default"
-if test "x$ac_cv_header_gmp_h" = xyes; then :
+do
+as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
+$as_echo_n "checking $ac_header usability... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+#include <$ac_header>
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_header_compiler=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_header_compiler=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
+$as_echo_n "checking $ac_header presence... " >&6; }
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <$ac_header>
+_ACEOF
+if { (ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } >/dev/null && {
+	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       }; then
+  ac_header_preproc=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+  ac_header_preproc=no
+fi
+
+rm -f conftest.err conftest.$ac_ext
+{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_cxx_preproc_warn_flag in
+  yes:no: )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
+    ac_header_preproc=yes
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
+$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
+    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
+$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
+    ( cat <<\_ASBOX
+## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##
+_ASBOX
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
+$as_echo_n "checking for $ac_header... " >&6; }
+if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  eval "$as_ac_Header=\$ac_header_preproc"
+fi
+ac_res=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+
+fi
+as_val=`eval 'as_val=${'$as_ac_Header'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_GMP_H 1
+#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
 
 fi
 
 done
 
-ac_fn_cxx_check_decl "$LINENO" "ntohll" "ac_cv_have_decl_ntohll" "#include <arpa/inet.h>
-"
-if test "x$ac_cv_have_decl_ntohll" = xyes; then :
-  ac_have_decl=1
+{ $as_echo "$as_me:$LINENO: checking whether ntohll is declared" >&5
+$as_echo_n "checking whether ntohll is declared... " >&6; }
+if test "${ac_cv_have_decl_ntohll+set}" = set; then
+  $as_echo_n "(cached) " >&6
 else
-  ac_have_decl=0
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <arpa/inet.h>
+
+int
+main ()
+{
+#ifndef ntohll
+  (void) ntohll;
+#endif
+
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_cv_have_decl_ntohll=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cv_have_decl_ntohll=no
 fi
 
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_have_decl_ntohll" >&5
+$as_echo "$ac_cv_have_decl_ntohll" >&6; }
+if test "x$ac_cv_have_decl_ntohll" = x""yes; then
+
 cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_NTOHLL $ac_have_decl
+#define HAVE_DECL_NTOHLL 1
 _ACEOF
 
-ac_fn_cxx_check_decl "$LINENO" "be64toh" "ac_cv_have_decl_be64toh" "#include <arpa/inet.h>
-"
-if test "x$ac_cv_have_decl_be64toh" = xyes; then :
-  ac_have_decl=1
+
 else
-  ac_have_decl=0
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_DECL_NTOHLL 0
+_ACEOF
+
+
 fi
 
-cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_BE64TOH $ac_have_decl
+
+{ $as_echo "$as_me:$LINENO: checking whether be64toh is declared" >&5
+$as_echo_n "checking whether be64toh is declared... " >&6; }
+if test "${ac_cv_have_decl_be64toh+set}" = set; then
+  $as_echo_n "(cached) " >&6
+else
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
 _ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#include <arpa/inet.h>
+
+int
+main ()
+{
+#ifndef be64toh
+  (void) be64toh;
+#endif
+
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  ac_cv_have_decl_be64toh=yes
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cv_have_decl_be64toh=no
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_have_decl_be64toh" >&5
+$as_echo "$ac_cv_have_decl_be64toh" >&6; }
+if test "x$ac_cv_have_decl_be64toh" = x""yes; then
+
+cat >>confdefs.h <<_ACEOF
+#define HAVE_DECL_BE64TOH 1
+_ACEOF
+
+
+else
+  cat >>confdefs.h <<_ACEOF
+#define HAVE_DECL_BE64TOH 0
+_ACEOF
+
+
+fi
+
 
 
 # Checks for typedefs, structures, and compiler characteristics.
-ac_fn_cxx_check_type "$LINENO" "size_t" "ac_cv_type_size_t" "$ac_includes_default"
-if test "x$ac_cv_type_size_t" = xyes; then :
+{ $as_echo "$as_me:$LINENO: checking for size_t" >&5
+$as_echo_n "checking for size_t... " >&6; }
+if test "${ac_cv_type_size_t+set}" = set; then
+  $as_echo_n "(cached) " >&6
+else
+  ac_cv_type_size_t=no
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+if (sizeof (size_t))
+       return 0;
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+if (sizeof ((size_t)))
+	  return 0;
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  :
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
 
+	ac_cv_type_size_t=yes
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_type_size_t" >&5
+$as_echo "$ac_cv_type_size_t" >&6; }
+if test "x$ac_cv_type_size_t" = x""yes; then
+  :
 else
 
 cat >>confdefs.h <<_ACEOF
@@ -4674,12 +5593,75 @@ _ACEOF
 
 fi
 
-ac_fn_c_find_uintX_t "$LINENO" "32" "ac_cv_c_uint32_t"
-case $ac_cv_c_uint32_t in #(
+
+  { $as_echo "$as_me:$LINENO: checking for uint32_t" >&5
+$as_echo_n "checking for uint32_t... " >&6; }
+if test "${ac_cv_c_uint32_t+set}" = set; then
+  $as_echo_n "(cached) " >&6
+else
+  ac_cv_c_uint32_t=no
+     for ac_type in 'uint32_t' 'unsigned int' 'unsigned long int' \
+	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
+       cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+static int test_array [1 - 2 * !(($ac_type) -1 >> (32 - 1) == 1)];
+test_array [0] = 0
+
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  case $ac_type in
+  uint32_t) ac_cv_c_uint32_t=yes ;;
+  *) ac_cv_c_uint32_t=$ac_type ;;
+esac
+
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+       test "$ac_cv_c_uint32_t" != no && break
+     done
+fi
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_uint32_t" >&5
+$as_echo "$ac_cv_c_uint32_t" >&6; }
+  case $ac_cv_c_uint32_t in #(
   no|yes) ;; #(
   *)
 
-$as_echo "#define _UINT32_T 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define _UINT32_T 1
+_ACEOF
 
 
 cat >>confdefs.h <<_ACEOF
@@ -4688,12 +5670,75 @@ _ACEOF
 ;;
   esac
 
-ac_fn_c_find_uintX_t "$LINENO" "64" "ac_cv_c_uint64_t"
-case $ac_cv_c_uint64_t in #(
+
+  { $as_echo "$as_me:$LINENO: checking for uint64_t" >&5
+$as_echo_n "checking for uint64_t... " >&6; }
+if test "${ac_cv_c_uint64_t+set}" = set; then
+  $as_echo_n "(cached) " >&6
+else
+  ac_cv_c_uint64_t=no
+     for ac_type in 'uint64_t' 'unsigned int' 'unsigned long int' \
+	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
+       cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+static int test_array [1 - 2 * !(($ac_type) -1 >> (64 - 1) == 1)];
+test_array [0] = 0
+
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext
+if { (ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_compile") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then
+  case $ac_type in
+  uint64_t) ac_cv_c_uint64_t=yes ;;
+  *) ac_cv_c_uint64_t=$ac_type ;;
+esac
+
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+
+fi
+
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+       test "$ac_cv_c_uint64_t" != no && break
+     done
+fi
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_uint64_t" >&5
+$as_echo "$ac_cv_c_uint64_t" >&6; }
+  case $ac_cv_c_uint64_t in #(
   no|yes) ;; #(
   *)
 
-$as_echo "#define _UINT64_T 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define _UINT64_T 1
+_ACEOF
 
 
 cat >>confdefs.h <<_ACEOF
@@ -4709,26 +5754,121 @@ echo Checking libraries
 echo :::::::::::::::::::::::::::::::::::::::::::
 
 
+
 for ac_func in gettimeofday
-do :
-  ac_fn_cxx_check_func "$LINENO" "gettimeofday" "ac_cv_func_gettimeofday"
-if test "x$ac_cv_func_gettimeofday" = xyes; then :
+do
+as_ac_var=`$as_echo "ac_cv_func_$ac_func" | $as_tr_sh`
+{ $as_echo "$as_me:$LINENO: checking for $ac_func" >&5
+$as_echo_n "checking for $ac_func... " >&6; }
+if { as_var=$as_ac_var; eval "test \"\${$as_var+set}\" = set"; }; then
+  $as_echo_n "(cached) " >&6
+else
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+/* Define $ac_func to an innocuous variant, in case <limits.h> declares $ac_func.
+   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
+#define $ac_func innocuous_$ac_func
+
+/* System header to define __stub macros and hopefully few prototypes,
+    which can conflict with char $ac_func (); below.
+    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
+    <limits.h> exists even on freestanding compilers.  */
+
+#ifdef __STDC__
+# include <limits.h>
+#else
+# include <assert.h>
+#endif
+
+#undef $ac_func
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char $ac_func ();
+/* The GNU C library defines this for functions which it implements
+    to always fail with ENOSYS.  Some functions are actually named
+    something starting with __ and the normal name is an alias.  */
+#if defined __stub_$ac_func || defined __stub___$ac_func
+choke me
+#endif
+
+int
+main ()
+{
+return $ac_func ();
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_link") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 $as_test_x conftest$ac_exeext
+       }; then
+  eval "$as_ac_var=yes"
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	eval "$as_ac_var=no"
+fi
+
+rm -rf conftest.dSYM
+rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
+      conftest$ac_exeext conftest.$ac_ext
+fi
+ac_res=`eval 'as_val=${'$as_ac_var'}
+		 $as_echo "$as_val"'`
+	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+as_val=`eval 'as_val=${'$as_ac_var'}
+		 $as_echo "$as_val"'`
+   if test "x$as_val" = x""yes; then
   cat >>confdefs.h <<_ACEOF
-#define HAVE_GETTIMEOFDAY 1
+#define `$as_echo "HAVE_$ac_func" | $as_tr_cpp` 1
 _ACEOF
 
 fi
 done
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for __gmpf_init in -lgmp" >&5
+
+{ $as_echo "$as_me:$LINENO: checking for __gmpf_init in -lgmp" >&5
 $as_echo_n "checking for __gmpf_init in -lgmp... " >&6; }
-if ${ac_cv_lib_gmp___gmpf_init+:} false; then :
+if test "${ac_cv_lib_gmp___gmpf_init+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   ac_check_lib_save_LIBS=$LIBS
 LIBS="-lgmp  $LIBS"
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 /* Override any GCC internal prototype to avoid an error.
@@ -4746,18 +5886,43 @@ return __gmpf_init ();
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_link "$LINENO"; then :
+rm -f conftest.$ac_objext conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_link") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 $as_test_x conftest$ac_exeext
+       }; then
   ac_cv_lib_gmp___gmpf_init=yes
 else
-  ac_cv_lib_gmp___gmpf_init=no
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cv_lib_gmp___gmpf_init=no
 fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
+
+rm -rf conftest.dSYM
+rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
+      conftest$ac_exeext conftest.$ac_ext
 LIBS=$ac_check_lib_save_LIBS
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_gmp___gmpf_init" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_lib_gmp___gmpf_init" >&5
 $as_echo "$ac_cv_lib_gmp___gmpf_init" >&6; }
-if test "x$ac_cv_lib_gmp___gmpf_init" = xyes; then :
+if test "x$ac_cv_lib_gmp___gmpf_init" = x""yes; then
   cat >>confdefs.h <<_ACEOF
 #define HAVE_LIBGMP 1
 _ACEOF
@@ -4765,20 +5930,29 @@ _ACEOF
   LIBS="-lgmp $LIBS"
 
 else
-  as_fn_error $? "GNU Multiple Precision GMP library was not found in your system.
+  { { $as_echo "$as_me:$LINENO: error: GNU Multiple Precision GMP library was not found in your system.
 Please install or provide the correct path to your installation
-Info at: http://www.gmplib.org" "$LINENO" 5
+Info at: http://www.gmplib.org" >&5
+$as_echo "$as_me: error: GNU Multiple Precision GMP library was not found in your system.
+Please install or provide the correct path to your installation
+Info at: http://www.gmplib.org" >&2;}
+   { (exit 1); exit 1; }; }
 fi
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for mpfr_init in -lmpfr" >&5
+
+{ $as_echo "$as_me:$LINENO: checking for mpfr_init in -lmpfr" >&5
 $as_echo_n "checking for mpfr_init in -lmpfr... " >&6; }
-if ${ac_cv_lib_mpfr_mpfr_init+:} false; then :
+if test "${ac_cv_lib_mpfr_mpfr_init+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   ac_check_lib_save_LIBS=$LIBS
 LIBS="-lmpfr  $LIBS"
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
 /* end confdefs.h.  */
 
 /* Override any GCC internal prototype to avoid an error.
@@ -4796,18 +5970,43 @@ return mpfr_init ();
   return 0;
 }
 _ACEOF
-if ac_fn_cxx_try_link "$LINENO"; then :
+rm -f conftest.$ac_objext conftest$ac_exeext
+if { (ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
+$as_echo "$ac_try_echo") >&5
+  (eval "$ac_link") 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 $as_test_x conftest$ac_exeext
+       }; then
   ac_cv_lib_mpfr_mpfr_init=yes
 else
-  ac_cv_lib_mpfr_mpfr_init=no
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_cv_lib_mpfr_mpfr_init=no
 fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
+
+rm -rf conftest.dSYM
+rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
+      conftest$ac_exeext conftest.$ac_ext
 LIBS=$ac_check_lib_save_LIBS
 fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_mpfr_mpfr_init" >&5
+{ $as_echo "$as_me:$LINENO: result: $ac_cv_lib_mpfr_mpfr_init" >&5
 $as_echo "$ac_cv_lib_mpfr_mpfr_init" >&6; }
-if test "x$ac_cv_lib_mpfr_mpfr_init" = xyes; then :
+if test "x$ac_cv_lib_mpfr_mpfr_init" = x""yes; then
   cat >>confdefs.h <<_ACEOF
 #define HAVE_LIBMPFR 1
 _ACEOF
@@ -4815,9 +6014,13 @@ _ACEOF
   LIBS="-lmpfr $LIBS"
 
 else
-  as_fn_error $? "GNU Multiple Precision MPFR library was not found in your system.
+  { { $as_echo "$as_me:$LINENO: error: GNU Multiple Precision MPFR library was not found in your system.
 Please install or provide the correct path to your installation
-Info at: http://www.mpfr.org/" "$LINENO" 5
+Info at: http://www.mpfr.org/" >&5
+$as_echo "$as_me: error: GNU Multiple Precision MPFR library was not found in your system.
+Please install or provide the correct path to your installation
+Info at: http://www.mpfr.org/" >&2;}
+   { (exit 1); exit 1; }; }
 fi
 
 
@@ -4826,7 +6029,7 @@ fi
 
 
 # Check whether --enable-simd was given.
-if test "${enable_simd+set}" = set; then :
+if test "${enable_simd+set}" = set; then
   enableval=$enable_simd; \
 	ac_SIMD=${enable_simd}
 else
@@ -4838,35 +6041,45 @@ case ${ac_SIMD} in
      SSE4)
        echo Configuring for SSE4
 
-$as_echo "#define SSE4 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define SSE4 1
+_ACEOF
 
      ;;
      AVX)
        echo Configuring for AVX
 
-$as_echo "#define AVX1 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define AVX1 1
+_ACEOF
 
      ;;
      AVX2)
        echo Configuring for AVX2
 
-$as_echo "#define AVX2 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define AVX2 1
+_ACEOF
 
      ;;
      AVX512|MIC)
        echo Configuring for AVX512 and MIC
 
-$as_echo "#define AVX512 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define AVX512 1
+_ACEOF
 
      ;;
      *)
-     as_fn_error $? "${ac_SIMD} unsupported --enable-simd option" "$LINENO" 5;
+     { { $as_echo "$as_me:$LINENO: error: ${ac_SIMD} unsupported --enable-simd option" >&5
+$as_echo "$as_me: error: ${ac_SIMD} unsupported --enable-simd option" >&2;}
+   { (exit 1); exit 1; }; };
      ;;
 esac
 
 
 # Check whether --enable-comms was given.
-if test "${enable_comms+set}" = set; then :
+if test "${enable_comms+set}" = set; then
   enableval=$enable_comms; ac_COMMS=${enable_comms}
 else
   ac_COMMS=none
@@ -4877,17 +6090,23 @@ case ${ac_COMMS} in
      none)
        echo Configuring for NO communications
 
-$as_echo "#define GRID_COMMS_NONE 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define GRID_COMMS_NONE 1
+_ACEOF
 
      ;;
      mpi)
        echo Configuring for MPI communications
 
-$as_echo "#define GRID_COMMS_MPI 1" >>confdefs.h
+cat >>confdefs.h <<\_ACEOF
+#define GRID_COMMS_MPI 1
+_ACEOF
 
      ;;
      *)
-     as_fn_error $? "${ac_COMMS} unsupported --enable-comms option" "$LINENO" 5;
+     { { $as_echo "$as_me:$LINENO: error: ${ac_COMMS} unsupported --enable-comms option" >&5
+$as_echo "$as_me: error: ${ac_COMMS} unsupported --enable-comms option" >&2;}
+   { (exit 1); exit 1; }; };
      ;;
 esac
 
@@ -4917,25 +6136,25 @@ echo Checking doxygen support
 echo :::::::::::::::::::::::::::::::::::::::::::
 
 # Check whether --enable-doxygen was given.
-if test "${enable_doxygen+set}" = set; then :
+if test "${enable_doxygen+set}" = set; then
   enableval=$enable_doxygen;
 fi
 
 
 # Check whether --enable-dot was given.
-if test "${enable_dot+set}" = set; then :
+if test "${enable_dot+set}" = set; then
   enableval=$enable_dot;
 fi
 
 # Check whether --enable-html-docs was given.
-if test "${enable_html_docs+set}" = set; then :
+if test "${enable_html_docs+set}" = set; then
   enableval=$enable_html_docs;
 else
    enable_html_docs=yes
 fi
 
 # Check whether --enable-latex-docs was given.
-if test "${enable_latex_docs+set}" = set; then :
+if test "${enable_latex_docs+set}" = set; then
   enableval=$enable_latex_docs;
 else
   enable_latex_docs=no
@@ -4947,9 +6166,9 @@ if test "x$enable_doxygen" = xno; then
 else
  # Extract the first word of "doxygen", so it can be a program name with args.
 set dummy doxygen; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_DOXYGEN+:} false; then :
+if test "${ac_cv_prog_DOXYGEN+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$DOXYGEN"; then
@@ -4960,31 +6179,33 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_DOXYGEN="doxygen"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 DOXYGEN=$ac_cv_prog_DOXYGEN
 if test -n "$DOXYGEN"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $DOXYGEN" >&5
+  { $as_echo "$as_me:$LINENO: result: $DOXYGEN" >&5
 $as_echo "$DOXYGEN" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
 
  if test x$DOXYGEN = x; then
   if test "x$enable_doxygen" = xyes; then
-   as_fn_error $? "could not find doxygen" "$LINENO" 5
+   { { $as_echo "$as_me:$LINENO: error: could not find doxygen" >&5
+$as_echo "$as_me: error: could not find doxygen" >&2;}
+   { (exit 1); exit 1; }; }
   fi
   enable_doc=no
  else
@@ -4995,9 +6216,9 @@ fi
    enable_doc=yes
    # Extract the first word of "dot", so it can be a program name with args.
 set dummy dot; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_DOT+:} false; then :
+if test "${ac_cv_prog_DOT+set}" = set; then
   $as_echo_n "(cached) " >&6
 else
   if test -n "$DOT"; then
@@ -5008,30 +6229,30 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
     ac_cv_prog_DOT="dot"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-  done
+done
 IFS=$as_save_IFS
 
 fi
 fi
 DOT=$ac_cv_prog_DOT
 if test -n "$DOT"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $DOT" >&5
+  { $as_echo "$as_me:$LINENO: result: $DOT" >&5
 $as_echo "$DOT" >&6; }
 else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+  { $as_echo "$as_me:$LINENO: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
 
   else
-   { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: doxygen version $doxy_ver too old, doxygen will not be used." >&5
+   { $as_echo "$as_me:$LINENO: WARNING: doxygen version $doxy_ver too old, doxygen will not be used." >&5
 $as_echo "$as_me: WARNING: doxygen version $doxy_ver too old, doxygen will not be used." >&2;}
    enable_doc=no
   fi
@@ -5049,7 +6270,9 @@ fi
 
 if test x$DOT = x; then
  if test "x$enable_dot" = xyes; then
-  as_fn_error $? "could not find dot" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: could not find dot" >&5
+$as_echo "$as_me: error: could not find dot" >&2;}
+   { (exit 1); exit 1; }; }
  fi
  enable_dot=no
 else
@@ -5103,13 +6326,13 @@ _ACEOF
     case $ac_val in #(
     *${as_nl}*)
       case $ac_var in #(
-      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
+      *_cv_*) { $as_echo "$as_me:$LINENO: WARNING: cache variable $ac_var contains a newline" >&5
 $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
       esac
       case $ac_var in #(
       _ | IFS | as_nl) ;; #(
       BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
-      *) { eval $ac_var=; unset $ac_var;} ;;
+      *) $as_unset $ac_var ;;
       esac ;;
     esac
   done
@@ -5117,8 +6340,8 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
   (set) 2>&1 |
     case $as_nl`(ac_space=' '; set) 2>&1` in #(
     *${as_nl}ac_space=\ *)
-      # `set' does not quote correctly, so add quotes: double-quote
-      # substitution turns \\\\ into \\, and sed turns \\ into \.
+      # `set' does not quote correctly, so add quotes (double-quote
+      # substitution turns \\\\ into \\, and sed turns \\ into \).
       sed -n \
 	"s/'/'\\\\''/g;
 	  s/^\\([_$as_cr_alnum]*_cv_[_$as_cr_alnum]*\\)=\\(.*\\)/\\1='\\2'/p"
@@ -5140,23 +6363,12 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
      :end' >>confcache
 if diff "$cache_file" confcache >/dev/null 2>&1; then :; else
   if test -w "$cache_file"; then
-    if test "x$cache_file" != "x/dev/null"; then
-      { $as_echo "$as_me:${as_lineno-$LINENO}: updating cache $cache_file" >&5
+    test "x$cache_file" != "x/dev/null" &&
+      { $as_echo "$as_me:$LINENO: updating cache $cache_file" >&5
 $as_echo "$as_me: updating cache $cache_file" >&6;}
-      if test ! -f "$cache_file" || test -h "$cache_file"; then
-	cat confcache >"$cache_file"
-      else
-        case $cache_file in #(
-        */* | ?:*)
-	  mv -f confcache "$cache_file"$$ &&
-	  mv -f "$cache_file"$$ "$cache_file" ;; #(
-        *)
-	  mv -f confcache "$cache_file" ;;
-	esac
-      fi
-    fi
+    cat confcache >$cache_file
   else
-    { $as_echo "$as_me:${as_lineno-$LINENO}: not updating unwritable cache $cache_file" >&5
+    { $as_echo "$as_me:$LINENO: not updating unwritable cache $cache_file" >&5
 $as_echo "$as_me: not updating unwritable cache $cache_file" >&6;}
   fi
 fi
@@ -5170,29 +6382,20 @@ DEFS=-DHAVE_CONFIG_H
 
 ac_libobjs=
 ac_ltlibobjs=
-U=
 for ac_i in : $LIBOBJS; do test "x$ac_i" = x: && continue
   # 1. Remove the extension, and $U if already installed.
   ac_script='s/\$U\././;s/\.o$//;s/\.obj$//'
   ac_i=`$as_echo "$ac_i" | sed "$ac_script"`
   # 2. Prepend LIBOBJDIR.  When used with automake>=1.10 LIBOBJDIR
   #    will be set to the directory where LIBOBJS objects are built.
-  as_fn_append ac_libobjs " \${LIBOBJDIR}$ac_i\$U.$ac_objext"
-  as_fn_append ac_ltlibobjs " \${LIBOBJDIR}$ac_i"'$U.lo'
+  ac_libobjs="$ac_libobjs \${LIBOBJDIR}$ac_i\$U.$ac_objext"
+  ac_ltlibobjs="$ac_ltlibobjs \${LIBOBJDIR}$ac_i"'$U.lo'
 done
 LIBOBJS=$ac_libobjs
 
 LTLIBOBJS=$ac_ltlibobjs
 
 
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking that generated files are newer than configure" >&5
-$as_echo_n "checking that generated files are newer than configure... " >&6; }
-   if test -n "$am_sleep_pid"; then
-     # Hide warnings about reused PIDs.
-     wait $am_sleep_pid 2>/dev/null
-   fi
-   { $as_echo "$as_me:${as_lineno-$LINENO}: result: done" >&5
-$as_echo "done" >&6; }
  if test -n "$EXEEXT"; then
   am__EXEEXT_TRUE=
   am__EXEEXT_FALSE='#'
@@ -5202,34 +6405,48 @@ else
 fi
 
 if test -z "${AMDEP_TRUE}" && test -z "${AMDEP_FALSE}"; then
-  as_fn_error $? "conditional \"AMDEP\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: conditional \"AMDEP\" was never defined.
+Usually this means the macro was only invoked conditionally." >&5
+$as_echo "$as_me: error: conditional \"AMDEP\" was never defined.
+Usually this means the macro was only invoked conditionally." >&2;}
+   { (exit 1); exit 1; }; }
 fi
 if test -z "${am__fastdepCXX_TRUE}" && test -z "${am__fastdepCXX_FALSE}"; then
-  as_fn_error $? "conditional \"am__fastdepCXX\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: conditional \"am__fastdepCXX\" was never defined.
+Usually this means the macro was only invoked conditionally." >&5
+$as_echo "$as_me: error: conditional \"am__fastdepCXX\" was never defined.
+Usually this means the macro was only invoked conditionally." >&2;}
+   { (exit 1); exit 1; }; }
 fi
 if test -z "${BUILD_COMMS_MPI_TRUE}" && test -z "${BUILD_COMMS_MPI_FALSE}"; then
-  as_fn_error $? "conditional \"BUILD_COMMS_MPI\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: conditional \"BUILD_COMMS_MPI\" was never defined.
+Usually this means the macro was only invoked conditionally." >&5
+$as_echo "$as_me: error: conditional \"BUILD_COMMS_MPI\" was never defined.
+Usually this means the macro was only invoked conditionally." >&2;}
+   { (exit 1); exit 1; }; }
 fi
 if test -z "${BUILD_COMMS_NONE_TRUE}" && test -z "${BUILD_COMMS_NONE_FALSE}"; then
-  as_fn_error $? "conditional \"BUILD_COMMS_NONE\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: conditional \"BUILD_COMMS_NONE\" was never defined.
+Usually this means the macro was only invoked conditionally." >&5
+$as_echo "$as_me: error: conditional \"BUILD_COMMS_NONE\" was never defined.
+Usually this means the macro was only invoked conditionally." >&2;}
+   { (exit 1); exit 1; }; }
 fi
 if test -z "${DOXYGEN_DOC_TRUE}" && test -z "${DOXYGEN_DOC_FALSE}"; then
-  as_fn_error $? "conditional \"DOXYGEN_DOC\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: conditional \"DOXYGEN_DOC\" was never defined.
+Usually this means the macro was only invoked conditionally." >&5
+$as_echo "$as_me: error: conditional \"DOXYGEN_DOC\" was never defined.
+Usually this means the macro was only invoked conditionally." >&2;}
+   { (exit 1); exit 1; }; }
 fi
 
-: "${CONFIG_STATUS=./config.status}"
+: ${CONFIG_STATUS=./config.status}
 ac_write_fail=0
 ac_clean_files_save=$ac_clean_files
 ac_clean_files="$ac_clean_files $CONFIG_STATUS"
-{ $as_echo "$as_me:${as_lineno-$LINENO}: creating $CONFIG_STATUS" >&5
+{ $as_echo "$as_me:$LINENO: creating $CONFIG_STATUS" >&5
 $as_echo "$as_me: creating $CONFIG_STATUS" >&6;}
-as_write_fail=0
-cat >$CONFIG_STATUS <<_ASEOF || as_write_fail=1
+cat >$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 #! $SHELL
 # Generated by $as_me.
 # Run this file to recreate the current configuration.
@@ -5239,18 +6456,17 @@ cat >$CONFIG_STATUS <<_ASEOF || as_write_fail=1
 debug=false
 ac_cs_recheck=false
 ac_cs_silent=false
-
 SHELL=\${CONFIG_SHELL-$SHELL}
-export SHELL
-_ASEOF
-cat >>$CONFIG_STATUS <<\_ASEOF || as_write_fail=1
-## -------------------- ##
-## M4sh Initialization. ##
-## -------------------- ##
+_ACEOF
+
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+## --------------------- ##
+## M4sh Initialization.  ##
+## --------------------- ##
 
 # Be more Bourne compatible
 DUALCASE=1; export DUALCASE # for MKS sh
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
   emulate sh
   NULLCMD=:
   # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
@@ -5258,15 +6474,23 @@ if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
   alias -g '${1+"$@"}'='"$@"'
   setopt NO_GLOB_SUBST
 else
-  case `(set -o) 2>/dev/null` in #(
-  *posix*) :
-    set -o posix ;; #(
-  *) :
-     ;;
+  case `(set -o) 2>/dev/null` in
+  *posix*) set -o posix ;;
 esac
+
 fi
 
 
+
+
+# PATH needs CR
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
+
 as_nl='
 '
 export as_nl
@@ -5274,13 +6498,7 @@ export as_nl
 as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
-# Prefer a ksh shell builtin over an external printf program on Solaris,
-# but without wasting forks for bash or zsh.
-if test -z "$BASH_VERSION$ZSH_VERSION" \
-    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
-  as_echo='print -r --'
-  as_echo_n='print -rn --'
-elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+if (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
   as_echo='printf %s\n'
   as_echo_n='printf %s'
 else
@@ -5291,7 +6509,7 @@ else
     as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
     as_echo_n_body='eval
       arg=$1;
-      case $arg in #(
+      case $arg in
       *"$as_nl"*)
 	expr "X$arg" : "X\\(.*\\)$as_nl";
 	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
@@ -5314,6 +6532,13 @@ if test "${PATH_SEPARATOR+set}" != set; then
   }
 fi
 
+# Support unset when possible.
+if ( (MAIL=60; unset MAIL) || exit) >/dev/null 2>&1; then
+  as_unset=unset
+else
+  as_unset=false
+fi
+
 
 # IFS
 # We need space, tab and new line, in precisely that order.  Quoting is
@@ -5323,16 +6548,15 @@ fi
 IFS=" ""	$as_nl"
 
 # Find who we are.  Look in the path if we contain no directory separator.
-as_myself=
-case $0 in #((
+case $0 in
   *[\\/]* ) as_myself=$0 ;;
   *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
 for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
-  done
+  test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+done
 IFS=$as_save_IFS
 
      ;;
@@ -5344,16 +6568,12 @@ if test "x$as_myself" = x; then
 fi
 if test ! -f "$as_myself"; then
   $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
-  exit 1
+  { (exit 1); exit 1; }
 fi
 
-# Unset variables that we do not need and which cause bugs (e.g. in
-# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
-# suppresses any "Segmentation fault" message there.  '((' could
-# trigger a bug in pdksh 5.2.14.
-for as_var in BASH_ENV ENV MAIL MAILPATH
-do eval test x\${$as_var+set} = xset \
-  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
+# Work around bugs in pre-3.0 UWIN ksh.
+for as_var in ENV MAIL MAILPATH
+do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
 done
 PS1='$ '
 PS2='> '
@@ -5365,89 +6585,7 @@ export LC_ALL
 LANGUAGE=C
 export LANGUAGE
 
-# CDPATH.
-(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
-
-
-# as_fn_error STATUS ERROR [LINENO LOG_FD]
-# ----------------------------------------
-# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
-# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
-# script with STATUS, using 1 if that was 0.
-as_fn_error ()
-{
-  as_status=$1; test $as_status -eq 0 && as_status=1
-  if test "$4"; then
-    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
-  fi
-  $as_echo "$as_me: error: $2" >&2
-  as_fn_exit $as_status
-} # as_fn_error
-
-
-# as_fn_set_status STATUS
-# -----------------------
-# Set $? to STATUS, without forking.
-as_fn_set_status ()
-{
-  return $1
-} # as_fn_set_status
-
-# as_fn_exit STATUS
-# -----------------
-# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
-as_fn_exit ()
-{
-  set +e
-  as_fn_set_status $1
-  exit $1
-} # as_fn_exit
-
-# as_fn_unset VAR
-# ---------------
-# Portably unset VAR.
-as_fn_unset ()
-{
-  { eval $1=; unset $1;}
-}
-as_unset=as_fn_unset
-# as_fn_append VAR VALUE
-# ----------------------
-# Append the text in VALUE to the end of the definition contained in VAR. Take
-# advantage of any shell optimizations that allow amortized linear growth over
-# repeated appends, instead of the typical quadratic growth present in naive
-# implementations.
-if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
-  eval 'as_fn_append ()
-  {
-    eval $1+=\$2
-  }'
-else
-  as_fn_append ()
-  {
-    eval $1=\$$1\$2
-  }
-fi # as_fn_append
-
-# as_fn_arith ARG...
-# ------------------
-# Perform arithmetic evaluation on the ARGs, and store the result in the
-# global $as_val. Take advantage of shells that can avoid forks. The arguments
-# must be portable across $(()) and expr.
-if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
-  eval 'as_fn_arith ()
-  {
-    as_val=$(( $* ))
-  }'
-else
-  as_fn_arith ()
-  {
-    as_val=`expr "$@" || test $? -eq 1`
-  }
-fi # as_fn_arith
-
-
+# Required to use basename.
 if expr a : '\(a\)' >/dev/null 2>&1 &&
    test "X`expr 00001 : '.*\(...\)'`" = X001; then
   as_expr=expr
@@ -5461,12 +6599,8 @@ else
   as_basename=false
 fi
 
-if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
-  as_dirname=dirname
-else
-  as_dirname=false
-fi
 
+# Name of the executable.
 as_me=`$as_basename -- "$0" ||
 $as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
 	 X"$0" : 'X\(//\)$' \| \
@@ -5486,25 +6620,76 @@ $as_echo X/"$0" |
 	  }
 	  s/.*/./; q'`
 
-# Avoid depending upon Character Ranges.
-as_cr_letters='abcdefghijklmnopqrstuvwxyz'
-as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
-as_cr_Letters=$as_cr_letters$as_cr_LETTERS
-as_cr_digits='0123456789'
-as_cr_alnum=$as_cr_Letters$as_cr_digits
+# CDPATH.
+$as_unset CDPATH
+
+
+
+  as_lineno_1=$LINENO
+  as_lineno_2=$LINENO
+  test "x$as_lineno_1" != "x$as_lineno_2" &&
+  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2" || {
+
+  # Create $as_me.lineno as a copy of $as_myself, but with $LINENO
+  # uniformly replaced by the line number.  The first 'sed' inserts a
+  # line-number line after each line using $LINENO; the second 'sed'
+  # does the real work.  The second script uses 'N' to pair each
+  # line-number line with the line containing $LINENO, and appends
+  # trailing '-' during substitution so that $LINENO is not a special
+  # case at line end.
+  # (Raja R Harinath suggested sed '=', and Paul Eggert wrote the
+  # scripts with optimization help from Paolo Bonzini.  Blame Lee
+  # E. McMahon (1931-1989) for sed's syntax.  :-)
+  sed -n '
+    p
+    /[$]LINENO/=
+  ' <$as_myself |
+    sed '
+      s/[$]LINENO.*/&-/
+      t lineno
+      b
+      :lineno
+      N
+      :loop
+      s/[$]LINENO\([^'$as_cr_alnum'_].*\n\)\(.*\)/\2\1\2/
+      t loop
+      s/-\n.*//
+    ' >$as_me.lineno &&
+  chmod +x "$as_me.lineno" ||
+    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2
+   { (exit 1); exit 1; }; }
+
+  # Don't try to exec as it changes $[0], causing all sort of problems
+  # (the dirname of $[0] is not the place where we might find the
+  # original and so on.  Autoconf is especially sensitive to this).
+  . "./$as_me.lineno"
+  # Exit status is that of the last command.
+  exit
+}
+
+
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
 
 ECHO_C= ECHO_N= ECHO_T=
-case `echo -n x` in #(((((
+case `echo -n x` in
 -n*)
-  case `echo 'xy\c'` in
+  case `echo 'x\c'` in
   *c*) ECHO_T='	';;	# ECHO_T is single tab character.
-  xy)  ECHO_C='\c';;
-  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
-       ECHO_T='	';;
+  *)   ECHO_C='\c';;
   esac;;
 *)
   ECHO_N='-n';;
 esac
+if expr a : '\(a\)' >/dev/null 2>&1 &&
+   test "X`expr 00001 : '.*\(...\)'`" = X001; then
+  as_expr=expr
+else
+  as_expr=false
+fi
 
 rm -f conf$$ conf$$.exe conf$$.file
 if test -d conf$$.dir; then
@@ -5519,85 +6704,49 @@ if (echo >conf$$.file) 2>/dev/null; then
     # ... but there are two gotchas:
     # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
     # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
-    # In both cases, we have to default to `cp -pR'.
+    # In both cases, we have to default to `cp -p'.
     ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
-      as_ln_s='cp -pR'
+      as_ln_s='cp -p'
   elif ln conf$$.file conf$$ 2>/dev/null; then
     as_ln_s=ln
   else
-    as_ln_s='cp -pR'
+    as_ln_s='cp -p'
   fi
 else
-  as_ln_s='cp -pR'
+  as_ln_s='cp -p'
 fi
 rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
 rmdir conf$$.dir 2>/dev/null
 
-
-# as_fn_mkdir_p
-# -------------
-# Create "$as_dir" as a directory, including parents if necessary.
-as_fn_mkdir_p ()
-{
-
-  case $as_dir in #(
-  -*) as_dir=./$as_dir;;
-  esac
-  test -d "$as_dir" || eval $as_mkdir_p || {
-    as_dirs=
-    while :; do
-      case $as_dir in #(
-      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
-      *) as_qdir=$as_dir;;
-      esac
-      as_dirs="'$as_qdir' $as_dirs"
-      as_dir=`$as_dirname -- "$as_dir" ||
-$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_dir" : 'X\(//\)[^/]' \| \
-	 X"$as_dir" : 'X\(//\)$' \| \
-	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_dir" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      test -d "$as_dir" && break
-    done
-    test -z "$as_dirs" || eval "mkdir $as_dirs"
-  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
-
-
-} # as_fn_mkdir_p
 if mkdir -p . 2>/dev/null; then
-  as_mkdir_p='mkdir -p "$as_dir"'
+  as_mkdir_p=:
 else
   test -d ./-p && rmdir ./-p
   as_mkdir_p=false
 fi
 
-
-# as_fn_executable_p FILE
-# -----------------------
-# Test if FILE is an executable regular file.
-as_fn_executable_p ()
-{
-  test -f "$1" && test -x "$1"
-} # as_fn_executable_p
-as_test_x='test -x'
-as_executable_p=as_fn_executable_p
+if test -x / >/dev/null 2>&1; then
+  as_test_x='test -x'
+else
+  if ls -dL / >/dev/null 2>&1; then
+    as_ls_L_option=L
+  else
+    as_ls_L_option=
+  fi
+  as_test_x='
+    eval sh -c '\''
+      if test -d "$1"; then
+	test -d "$1/.";
+      else
+	case $1 in
+	-*)set "./$1";;
+	esac;
+	case `ls -ld'$as_ls_L_option' "$1" 2>/dev/null` in
+	???[sx]*):;;*)false;;esac;fi
+    '\'' sh
+  '
+fi
+as_executable_p=$as_test_x
 
 # Sed expression to map a string onto a valid CPP name.
 as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
@@ -5607,19 +6756,13 @@ as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
 
 
 exec 6>&1
-## ----------------------------------- ##
-## Main body of $CONFIG_STATUS script. ##
-## ----------------------------------- ##
-_ASEOF
-test $as_write_fail = 0 && chmod +x $CONFIG_STATUS || ac_write_fail=1
 
-cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-# Save the log message, to keep $0 and so on meaningful, and to
+# Save the log message, to keep $[0] and so on meaningful, and to
 # report actual input values of CONFIG_FILES etc. instead of their
 # values after options handling.
 ac_log="
 This file was extended by Grid $as_me 1.0, which was
-generated by GNU Autoconf 2.69.  Invocation command line was
+generated by GNU Autoconf 2.63.  Invocation command line was
 
   CONFIG_FILES    = $CONFIG_FILES
   CONFIG_HEADERS  = $CONFIG_HEADERS
@@ -5651,15 +6794,13 @@ _ACEOF
 
 cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 ac_cs_usage="\
-\`$as_me' instantiates files and other configuration actions
-from templates according to the current configuration.  Unless the files
-and actions are specified as TAGs, all are instantiated by default.
+\`$as_me' instantiates files from templates according to the
+current configuration.
 
-Usage: $0 [OPTION]... [TAG]...
+Usage: $0 [OPTION]... [FILE]...
 
   -h, --help       print this help, then exit
   -V, --version    print version number and configuration settings, then exit
-      --config     print configuration, then exit
   -q, --quiet, --silent
                    do not print progress messages
   -d, --debug      don't remove temporary files
@@ -5678,17 +6819,16 @@ $config_headers
 Configuration commands:
 $config_commands
 
-Report bugs to <paboyle@ph.ed.ac.uk>."
+Report bugs to <bug-autoconf@gnu.org>."
 
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-ac_cs_config="`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`"
 ac_cs_version="\\
 Grid config.status 1.0
-configured by $0, generated by GNU Autoconf 2.69,
-  with options \\"\$ac_cs_config\\"
+configured by $0, generated by GNU Autoconf 2.63,
+  with options \\"`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`\\"
 
-Copyright (C) 2012 Free Software Foundation, Inc.
+Copyright (C) 2008 Free Software Foundation, Inc.
 This config.status script is free software; the Free Software Foundation
 gives unlimited permission to copy, distribute and modify it."
 
@@ -5706,16 +6846,11 @@ ac_need_defaults=:
 while test $# != 0
 do
   case $1 in
-  --*=?*)
+  --*=*)
     ac_option=`expr "X$1" : 'X\([^=]*\)='`
     ac_optarg=`expr "X$1" : 'X[^=]*=\(.*\)'`
     ac_shift=:
     ;;
-  --*=)
-    ac_option=`expr "X$1" : 'X\([^=]*\)='`
-    ac_optarg=
-    ac_shift=:
-    ;;
   *)
     ac_option=$1
     ac_optarg=$2
@@ -5729,29 +6864,27 @@ do
     ac_cs_recheck=: ;;
   --version | --versio | --versi | --vers | --ver | --ve | --v | -V )
     $as_echo "$ac_cs_version"; exit ;;
-  --config | --confi | --conf | --con | --co | --c )
-    $as_echo "$ac_cs_config"; exit ;;
   --debug | --debu | --deb | --de | --d | -d )
     debug=: ;;
   --file | --fil | --fi | --f )
     $ac_shift
     case $ac_optarg in
     *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
-    '') as_fn_error $? "missing file argument" ;;
     esac
-    as_fn_append CONFIG_FILES " '$ac_optarg'"
+    CONFIG_FILES="$CONFIG_FILES '$ac_optarg'"
     ac_need_defaults=false;;
   --header | --heade | --head | --hea )
     $ac_shift
     case $ac_optarg in
     *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
     esac
-    as_fn_append CONFIG_HEADERS " '$ac_optarg'"
+    CONFIG_HEADERS="$CONFIG_HEADERS '$ac_optarg'"
     ac_need_defaults=false;;
   --he | --h)
     # Conflict between --help and --header
-    as_fn_error $? "ambiguous option: \`$1'
-Try \`$0 --help' for more information.";;
+    { $as_echo "$as_me: error: ambiguous option: $1
+Try \`$0 --help' for more information." >&2
+   { (exit 1); exit 1; }; };;
   --help | --hel | -h )
     $as_echo "$ac_cs_usage"; exit ;;
   -q | -quiet | --quiet | --quie | --qui | --qu | --q \
@@ -5759,10 +6892,11 @@ Try \`$0 --help' for more information.";;
     ac_cs_silent=: ;;
 
   # This is an error.
-  -*) as_fn_error $? "unrecognized option: \`$1'
-Try \`$0 --help' for more information." ;;
+  -*) { $as_echo "$as_me: error: unrecognized option: $1
+Try \`$0 --help' for more information." >&2
+   { (exit 1); exit 1; }; } ;;
 
-  *) as_fn_append ac_config_targets " $1"
+  *) ac_config_targets="$ac_config_targets $1"
      ac_need_defaults=false ;;
 
   esac
@@ -5779,7 +6913,7 @@ fi
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 if \$ac_cs_recheck; then
-  set X $SHELL '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
+  set X '$SHELL' '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
   shift
   \$as_echo "running CONFIG_SHELL=$SHELL \$*" >&6
   CONFIG_SHELL='$SHELL'
@@ -5821,7 +6955,9 @@ do
     "tests/Makefile") CONFIG_FILES="$CONFIG_FILES tests/Makefile" ;;
     "benchmarks/Makefile") CONFIG_FILES="$CONFIG_FILES benchmarks/Makefile" ;;
 
-  *) as_fn_error $? "invalid argument: \`$ac_config_target'" "$LINENO" 5;;
+  *) { { $as_echo "$as_me:$LINENO: error: invalid argument: $ac_config_target" >&5
+$as_echo "$as_me: error: invalid argument: $ac_config_target" >&2;}
+   { (exit 1); exit 1; }; };;
   esac
 done
 
@@ -5844,24 +6980,26 @@ fi
 # after its creation but before its name has been assigned to `$tmp'.
 $debug ||
 {
-  tmp= ac_tmp=
+  tmp=
   trap 'exit_status=$?
-  : "${ac_tmp:=$tmp}"
-  { test ! -d "$ac_tmp" || rm -fr "$ac_tmp"; } && exit $exit_status
+  { test -z "$tmp" || test ! -d "$tmp" || rm -fr "$tmp"; } && exit $exit_status
 ' 0
-  trap 'as_fn_exit 1' 1 2 13 15
+  trap '{ (exit 1); exit 1; }' 1 2 13 15
 }
 # Create a (secure) tmp directory for tmp files.
 
 {
   tmp=`(umask 077 && mktemp -d "./confXXXXXX") 2>/dev/null` &&
-  test -d "$tmp"
+  test -n "$tmp" && test -d "$tmp"
 }  ||
 {
   tmp=./conf$$-$RANDOM
   (umask 077 && mkdir "$tmp")
-} || as_fn_error $? "cannot create a temporary directory in ." "$LINENO" 5
-ac_tmp=$tmp
+} ||
+{
+   $as_echo "$as_me: cannot create a temporary directory in ." >&2
+   { (exit 1); exit 1; }
+}
 
 # Set up the scripts for CONFIG_FILES section.
 # No need to generate them if there are no CONFIG_FILES.
@@ -5869,13 +7007,7 @@ ac_tmp=$tmp
 if test -n "$CONFIG_FILES"; then
 
 
-ac_cr=`echo X | tr X '\015'`
-# On cygwin, bash can eat \r inside `` if the user requested igncr.
-# But we know of no other shell where ac_cr would be empty at this
-# point, so we can use a bashism as a fallback.
-if test "x$ac_cr" = x; then
-  eval ac_cr=\$\'\\r\'
-fi
+ac_cr='
'
 ac_cs_awk_cr=`$AWK 'BEGIN { print "a\rb" }' </dev/null 2>/dev/null`
 if test "$ac_cs_awk_cr" = "a${ac_cr}b"; then
   ac_cs_awk_cr='\\r'
@@ -5883,7 +7015,7 @@ else
   ac_cs_awk_cr=$ac_cr
 fi
 
-echo 'BEGIN {' >"$ac_tmp/subs1.awk" &&
+echo 'BEGIN {' >"$tmp/subs1.awk" &&
 _ACEOF
 
 
@@ -5892,18 +7024,24 @@ _ACEOF
   echo "$ac_subst_vars" | sed 's/.*/&!$&$ac_delim/' &&
   echo "_ACEOF"
 } >conf$$subs.sh ||
-  as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
-ac_delim_num=`echo "$ac_subst_vars" | grep -c '^'`
+  { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
+$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
+   { (exit 1); exit 1; }; }
+ac_delim_num=`echo "$ac_subst_vars" | grep -c '$'`
 ac_delim='%!_!# '
 for ac_last_try in false false false false false :; do
   . ./conf$$subs.sh ||
-    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
+$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
+   { (exit 1); exit 1; }; }
 
   ac_delim_n=`sed -n "s/.*$ac_delim\$/X/p" conf$$subs.awk | grep -c X`
   if test $ac_delim_n = $ac_delim_num; then
     break
   elif $ac_last_try; then
-    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
+$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
+   { (exit 1); exit 1; }; }
   else
     ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
   fi
@@ -5911,7 +7049,7 @@ done
 rm -f conf$$subs.sh
 
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-cat >>"\$ac_tmp/subs1.awk" <<\\_ACAWK &&
+cat >>"\$tmp/subs1.awk" <<\\_ACAWK &&
 _ACEOF
 sed -n '
 h
@@ -5925,7 +7063,7 @@ s/'"$ac_delim"'$//
 t delim
 :nl
 h
-s/\(.\{148\}\)..*/\1/
+s/\(.\{148\}\).*/\1/
 t more1
 s/["\\]/\\&/g; s/^/"/; s/$/\\n"\\/
 p
@@ -5939,7 +7077,7 @@ s/.\{148\}//
 t nl
 :delim
 h
-s/\(.\{148\}\)..*/\1/
+s/\(.\{148\}\).*/\1/
 t more2
 s/["\\]/\\&/g; s/^/"/; s/$/"/
 p
@@ -5959,7 +7097,7 @@ t delim
 rm -f conf$$subs.awk
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 _ACAWK
-cat >>"\$ac_tmp/subs1.awk" <<_ACAWK &&
+cat >>"\$tmp/subs1.awk" <<_ACAWK &&
   for (key in S) S_is_set[key] = 1
   FS = ""
 
@@ -5991,29 +7129,23 @@ if sed "s/$ac_cr//" < /dev/null > /dev/null 2>&1; then
   sed "s/$ac_cr\$//; s/$ac_cr/$ac_cs_awk_cr/g"
 else
   cat
-fi < "$ac_tmp/subs1.awk" > "$ac_tmp/subs.awk" \
-  || as_fn_error $? "could not setup config files machinery" "$LINENO" 5
+fi < "$tmp/subs1.awk" > "$tmp/subs.awk" \
+  || { { $as_echo "$as_me:$LINENO: error: could not setup config files machinery" >&5
+$as_echo "$as_me: error: could not setup config files machinery" >&2;}
+   { (exit 1); exit 1; }; }
 _ACEOF
 
-# VPATH may cause trouble with some makes, so we remove sole $(srcdir),
-# ${srcdir} and @srcdir@ entries from VPATH if srcdir is ".", strip leading and
+# VPATH may cause trouble with some makes, so we remove $(srcdir),
+# ${srcdir} and @srcdir@ from VPATH if srcdir is ".", strip leading and
 # trailing colons and then remove the whole line if VPATH becomes empty
 # (actually we leave an empty line to preserve line numbers).
 if test "x$srcdir" = x.; then
-  ac_vpsub='/^[	 ]*VPATH[	 ]*=[	 ]*/{
-h
-s///
-s/^/:/
-s/[	 ]*$/:/
-s/:\$(srcdir):/:/g
-s/:\${srcdir}:/:/g
-s/:@srcdir@:/:/g
-s/^:*//
+  ac_vpsub='/^[	 ]*VPATH[	 ]*=/{
+s/:*\$(srcdir):*/:/
+s/:*\${srcdir}:*/:/
+s/:*@srcdir@:*/:/
+s/^\([^=]*=[	 ]*\):*/\1/
 s/:*$//
-x
-s/\(=[	 ]*\).*/\1/
-G
-s/\n//
 s/^[^=]*=[	 ]*$//
 }'
 fi
@@ -6025,7 +7157,7 @@ fi # test -n "$CONFIG_FILES"
 # No need to generate them if there are no CONFIG_HEADERS.
 # This happens for instance with `./config.status Makefile'.
 if test -n "$CONFIG_HEADERS"; then
-cat >"$ac_tmp/defines.awk" <<\_ACAWK ||
+cat >"$tmp/defines.awk" <<\_ACAWK ||
 BEGIN {
 _ACEOF
 
@@ -6037,11 +7169,13 @@ _ACEOF
 # handling of long lines.
 ac_delim='%!_!# '
 for ac_last_try in false false :; do
-  ac_tt=`sed -n "/$ac_delim/p" confdefs.h`
-  if test -z "$ac_tt"; then
+  ac_t=`sed -n "/$ac_delim/p" confdefs.h`
+  if test -z "$ac_t"; then
     break
   elif $ac_last_try; then
-    as_fn_error $? "could not make $CONFIG_HEADERS" "$LINENO" 5
+    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_HEADERS" >&5
+$as_echo "$as_me: error: could not make $CONFIG_HEADERS" >&2;}
+   { (exit 1); exit 1; }; }
   else
     ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
   fi
@@ -6126,7 +7260,9 @@ cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 _ACAWK
 _ACEOF
 cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-  as_fn_error $? "could not setup config headers machinery" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: could not setup config headers machinery" >&5
+$as_echo "$as_me: error: could not setup config headers machinery" >&2;}
+   { (exit 1); exit 1; }; }
 fi # test -n "$CONFIG_HEADERS"
 
 
@@ -6139,7 +7275,9 @@ do
   esac
   case $ac_mode$ac_tag in
   :[FHL]*:*);;
-  :L* | :C*:*) as_fn_error $? "invalid tag \`$ac_tag'" "$LINENO" 5;;
+  :L* | :C*:*) { { $as_echo "$as_me:$LINENO: error: invalid tag $ac_tag" >&5
+$as_echo "$as_me: error: invalid tag $ac_tag" >&2;}
+   { (exit 1); exit 1; }; };;
   :[FH]-) ac_tag=-:-;;
   :[FH]*) ac_tag=$ac_tag:$ac_tag.in;;
   esac
@@ -6158,7 +7296,7 @@ do
     for ac_f
     do
       case $ac_f in
-      -) ac_f="$ac_tmp/stdin";;
+      -) ac_f="$tmp/stdin";;
       *) # Look for the file first in the build tree, then in the source tree
 	 # (if the path is not absolute).  The absolute path cannot be DOS-style,
 	 # because $ac_f cannot contain `:'.
@@ -6167,10 +7305,12 @@ do
 	   [\\/$]*) false;;
 	   *) test -f "$srcdir/$ac_f" && ac_f="$srcdir/$ac_f";;
 	   esac ||
-	   as_fn_error 1 "cannot find input file: \`$ac_f'" "$LINENO" 5;;
+	   { { $as_echo "$as_me:$LINENO: error: cannot find input file: $ac_f" >&5
+$as_echo "$as_me: error: cannot find input file: $ac_f" >&2;}
+   { (exit 1); exit 1; }; };;
       esac
       case $ac_f in *\'*) ac_f=`$as_echo "$ac_f" | sed "s/'/'\\\\\\\\''/g"`;; esac
-      as_fn_append ac_file_inputs " '$ac_f'"
+      ac_file_inputs="$ac_file_inputs '$ac_f'"
     done
 
     # Let's still pretend it is `configure' which instantiates (i.e., don't
@@ -6181,7 +7321,7 @@ do
 	`' by configure.'
     if test x"$ac_file" != x-; then
       configure_input="$ac_file.  $configure_input"
-      { $as_echo "$as_me:${as_lineno-$LINENO}: creating $ac_file" >&5
+      { $as_echo "$as_me:$LINENO: creating $ac_file" >&5
 $as_echo "$as_me: creating $ac_file" >&6;}
     fi
     # Neutralize special characters interpreted by sed in replacement strings.
@@ -6193,8 +7333,10 @@ $as_echo "$as_me: creating $ac_file" >&6;}
     esac
 
     case $ac_tag in
-    *:-:* | *:-) cat >"$ac_tmp/stdin" \
-      || as_fn_error $? "could not create $ac_file" "$LINENO" 5 ;;
+    *:-:* | *:-) cat >"$tmp/stdin" \
+      || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; } ;;
     esac
     ;;
   esac
@@ -6222,7 +7364,47 @@ $as_echo X"$ac_file" |
 	    q
 	  }
 	  s/.*/./; q'`
-  as_dir="$ac_dir"; as_fn_mkdir_p
+  { as_dir="$ac_dir"
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || { $as_mkdir_p && mkdir -p "$as_dir"; } || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || { { $as_echo "$as_me:$LINENO: error: cannot create directory $as_dir" >&5
+$as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
+   { (exit 1); exit 1; }; }; }
   ac_builddir=.
 
 case "$ac_dir" in
@@ -6279,6 +7461,7 @@ cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 # If the template does not know about datarootdir, expand it.
 # FIXME: This hack should be removed a few years after 2.60.
 ac_datarootdir_hack=; ac_datarootdir_seen=
+
 ac_sed_dataroot='
 /datarootdir/ {
   p
@@ -6288,11 +7471,12 @@ ac_sed_dataroot='
 /@docdir@/p
 /@infodir@/p
 /@localedir@/p
-/@mandir@/p'
+/@mandir@/p
+'
 case `eval "sed -n \"\$ac_sed_dataroot\" $ac_file_inputs"` in
 *datarootdir*) ac_datarootdir_seen=yes;;
 *@datadir@*|*@docdir@*|*@infodir@*|*@localedir@*|*@mandir@*)
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
+  { $as_echo "$as_me:$LINENO: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
 $as_echo "$as_me: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&2;}
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
@@ -6302,7 +7486,7 @@ cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
   s&@infodir@&$infodir&g
   s&@localedir@&$localedir&g
   s&@mandir@&$mandir&g
-  s&\\\${datarootdir}&$datarootdir&g' ;;
+    s&\\\${datarootdir}&$datarootdir&g' ;;
 esac
 _ACEOF
 
@@ -6330,24 +7514,27 @@ s&@INSTALL@&$ac_INSTALL&;t t
 s&@MKDIR_P@&$ac_MKDIR_P&;t t
 $ac_datarootdir_hack
 "
-eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$ac_tmp/subs.awk" \
-  >$ac_tmp/out || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$tmp/subs.awk" >$tmp/out \
+  || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; }
 
 test -z "$ac_datarootdir_hack$ac_datarootdir_seen" &&
-  { ac_out=`sed -n '/\${datarootdir}/p' "$ac_tmp/out"`; test -n "$ac_out"; } &&
-  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' \
-      "$ac_tmp/out"`; test -z "$ac_out"; } &&
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file contains a reference to the variable \`datarootdir'
-which seems to be undefined.  Please make sure it is defined" >&5
+  { ac_out=`sed -n '/\${datarootdir}/p' "$tmp/out"`; test -n "$ac_out"; } &&
+  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' "$tmp/out"`; test -z "$ac_out"; } &&
+  { $as_echo "$as_me:$LINENO: WARNING: $ac_file contains a reference to the variable \`datarootdir'
+which seems to be undefined.  Please make sure it is defined." >&5
 $as_echo "$as_me: WARNING: $ac_file contains a reference to the variable \`datarootdir'
-which seems to be undefined.  Please make sure it is defined" >&2;}
+which seems to be undefined.  Please make sure it is defined." >&2;}
 
-  rm -f "$ac_tmp/stdin"
+  rm -f "$tmp/stdin"
   case $ac_file in
-  -) cat "$ac_tmp/out" && rm -f "$ac_tmp/out";;
-  *) rm -f "$ac_file" && mv "$ac_tmp/out" "$ac_file";;
+  -) cat "$tmp/out" && rm -f "$tmp/out";;
+  *) rm -f "$ac_file" && mv "$tmp/out" "$ac_file";;
   esac \
-  || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+  || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; }
  ;;
   :H)
   #
@@ -6356,21 +7543,27 @@ which seems to be undefined.  Please make sure it is defined" >&2;}
   if test x"$ac_file" != x-; then
     {
       $as_echo "/* $configure_input  */" \
-      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs"
-    } >"$ac_tmp/config.h" \
-      || as_fn_error $? "could not create $ac_file" "$LINENO" 5
-    if diff "$ac_file" "$ac_tmp/config.h" >/dev/null 2>&1; then
-      { $as_echo "$as_me:${as_lineno-$LINENO}: $ac_file is unchanged" >&5
+      && eval '$AWK -f "$tmp/defines.awk"' "$ac_file_inputs"
+    } >"$tmp/config.h" \
+      || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; }
+    if diff "$ac_file" "$tmp/config.h" >/dev/null 2>&1; then
+      { $as_echo "$as_me:$LINENO: $ac_file is unchanged" >&5
 $as_echo "$as_me: $ac_file is unchanged" >&6;}
     else
       rm -f "$ac_file"
-      mv "$ac_tmp/config.h" "$ac_file" \
-	|| as_fn_error $? "could not create $ac_file" "$LINENO" 5
+      mv "$tmp/config.h" "$ac_file" \
+	|| { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
+$as_echo "$as_me: error: could not create $ac_file" >&2;}
+   { (exit 1); exit 1; }; }
     fi
   else
     $as_echo "/* $configure_input  */" \
-      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs" \
-      || as_fn_error $? "could not create -" "$LINENO" 5
+      && eval '$AWK -f "$tmp/defines.awk"' "$ac_file_inputs" \
+      || { { $as_echo "$as_me:$LINENO: error: could not create -" >&5
+$as_echo "$as_me: error: could not create -" >&2;}
+   { (exit 1); exit 1; }; }
   fi
 # Compute "$ac_file"'s index in $config_headers.
 _am_arg="$ac_file"
@@ -6408,7 +7601,7 @@ $as_echo X"$_am_arg" |
 	  s/.*/./; q'`/stamp-h$_am_stamp_count
  ;;
 
-  :C)  { $as_echo "$as_me:${as_lineno-$LINENO}: executing $ac_file commands" >&5
+  :C)  { $as_echo "$as_me:$LINENO: executing $ac_file commands" >&5
 $as_echo "$as_me: executing $ac_file commands" >&6;}
  ;;
   esac
@@ -6416,7 +7609,7 @@ $as_echo "$as_me: executing $ac_file commands" >&6;}
 
   case $ac_file$ac_mode in
     "depfiles":C) test x"$AMDEP_TRUE" != x"" || {
-  # Older Autoconf quotes --file arguments for eval, but not when files
+  # Autoconf 2.62 quotes --file arguments for eval, but not when files
   # are listed without --file.  Let's play safe and only enable the eval
   # if we detect the quoting.
   case $CONFIG_FILES in
@@ -6429,7 +7622,7 @@ $as_echo "$as_me: executing $ac_file commands" >&6;}
     # Strip MF so we end up with the name of the file.
     mf=`echo "$mf" | sed -e 's/:.*$//'`
     # Check whether this is an Automake generated Makefile or not.
-    # We used to match only the files named 'Makefile.in', but
+    # We used to match only the files named `Makefile.in', but
     # some people rename them; so instead we look at the file content.
     # Grep'ing the first line is not enough: some people post-process
     # each Makefile.in and add a new line on top of each file to say so.
@@ -6463,19 +7656,21 @@ $as_echo X"$mf" |
       continue
     fi
     # Extract the definition of DEPDIR, am__include, and am__quote
-    # from the Makefile without running 'make'.
+    # from the Makefile without running `make'.
     DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
     test -z "$DEPDIR" && continue
     am__include=`sed -n 's/^am__include = //p' < "$mf"`
-    test -z "$am__include" && continue
+    test -z "am__include" && continue
     am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
+    # When using ansi2knr, U may be empty or an underscore; expand it
+    U=`sed -n 's/^U = //p' < "$mf"`
     # Find all dependency output files, they are included files with
     # $(DEPDIR) in their names.  We invoke sed twice because it is the
     # simplest approach to changing $(DEPDIR) to its actual value in the
     # expansion.
     for file in `sed -n "
       s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
-	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
+	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g' -e 's/\$U/'"$U"'/g'`; do
       # Make sure the directory exists.
       test -f "$dirpart/$file" && continue
       fdir=`$as_dirname -- "$file" ||
@@ -6501,7 +7696,47 @@ $as_echo X"$file" |
 	    q
 	  }
 	  s/.*/./; q'`
-      as_dir=$dirpart/$fdir; as_fn_mkdir_p
+      { as_dir=$dirpart/$fdir
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || { $as_mkdir_p && mkdir -p "$as_dir"; } || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || { { $as_echo "$as_me:$LINENO: error: cannot create directory $as_dir" >&5
+$as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
+   { (exit 1); exit 1; }; }; }
       # echo "creating $dirpart/$file"
       echo '# dummy' > "$dirpart/$file"
     done
@@ -6513,12 +7748,15 @@ $as_echo X"$file" |
 done # for ac_tag
 
 
-as_fn_exit 0
+{ (exit 0); exit 0; }
 _ACEOF
+chmod +x $CONFIG_STATUS
 ac_clean_files=$ac_clean_files_save
 
 test $ac_write_fail = 0 ||
-  as_fn_error $? "write failure creating $CONFIG_STATUS" "$LINENO" 5
+  { { $as_echo "$as_me:$LINENO: error: write failure creating $CONFIG_STATUS" >&5
+$as_echo "$as_me: error: write failure creating $CONFIG_STATUS" >&2;}
+   { (exit 1); exit 1; }; }
 
 
 # configure is writing to config.log, and then calls config.status.
@@ -6539,10 +7777,10 @@ if test "$no_create" != yes; then
   exec 5>>config.log
   # Use ||, not &&, to avoid exiting from the if with $? = 1, which
   # would make configure fail if this is the last instruction.
-  $ac_cs_success || as_fn_exit 1
+  $ac_cs_success || { (exit 1); exit 1; }
 fi
 if test -n "$ac_unrecognized_opts" && test "$enable_option_checking" != no; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
+  { $as_echo "$as_me:$LINENO: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
 $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2;}
 fi
 
diff --git a/depcomp b/depcomp
deleted file mode 100755
index fc98710e..00000000
--- a/depcomp
+++ /dev/null
@@ -1,791 +0,0 @@
-#! /bin/sh
-# depcomp - compile a program generating dependencies as side-effects
-
-scriptversion=2013-05-30.07; # UTC
-
-# Copyright (C) 1999-2014 Free Software Foundation, Inc.
-
-# This program is free software; you can redistribute it and/or modify
-# it under the terms of the GNU General Public License as published by
-# the Free Software Foundation; either version 2, or (at your option)
-# any later version.
-
-# This program is distributed in the hope that it will be useful,
-# but WITHOUT ANY WARRANTY; without even the implied warranty of
-# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-# GNU General Public License for more details.
-
-# You should have received a copy of the GNU General Public License
-# along with this program.  If not, see <http://www.gnu.org/licenses/>.
-
-# As a special exception to the GNU General Public License, if you
-# distribute this file as part of a program that contains a
-# configuration script generated by Autoconf, you may include it under
-# the same distribution terms that you use for the rest of that program.
-
-# Originally written by Alexandre Oliva <oliva@dcc.unicamp.br>.
-
-case $1 in
-  '')
-    echo "$0: No command.  Try '$0 --help' for more information." 1>&2
-    exit 1;
-    ;;
-  -h | --h*)
-    cat <<\EOF
-Usage: depcomp [--help] [--version] PROGRAM [ARGS]
-
-Run PROGRAMS ARGS to compile a file, generating dependencies
-as side-effects.
-
-Environment variables:
-  depmode     Dependency tracking mode.
-  source      Source file read by 'PROGRAMS ARGS'.
-  object      Object file output by 'PROGRAMS ARGS'.
-  DEPDIR      directory where to store dependencies.
-  depfile     Dependency file to output.
-  tmpdepfile  Temporary file to use when outputting dependencies.
-  libtool     Whether libtool is used (yes/no).
-
-Report bugs to <bug-automake@gnu.org>.
-EOF
-    exit $?
-    ;;
-  -v | --v*)
-    echo "depcomp $scriptversion"
-    exit $?
-    ;;
-esac
-
-# Get the directory component of the given path, and save it in the
-# global variables '$dir'.  Note that this directory component will
-# be either empty or ending with a '/' character.  This is deliberate.
-set_dir_from ()
-{
-  case $1 in
-    */*) dir=`echo "$1" | sed -e 's|/[^/]*$|/|'`;;
-      *) dir=;;
-  esac
-}
-
-# Get the suffix-stripped basename of the given path, and save it the
-# global variable '$base'.
-set_base_from ()
-{
-  base=`echo "$1" | sed -e 's|^.*/||' -e 's/\.[^.]*$//'`
-}
-
-# If no dependency file was actually created by the compiler invocation,
-# we still have to create a dummy depfile, to avoid errors with the
-# Makefile "include basename.Plo" scheme.
-make_dummy_depfile ()
-{
-  echo "#dummy" > "$depfile"
-}
-
-# Factor out some common post-processing of the generated depfile.
-# Requires the auxiliary global variable '$tmpdepfile' to be set.
-aix_post_process_depfile ()
-{
-  # If the compiler actually managed to produce a dependency file,
-  # post-process it.
-  if test -f "$tmpdepfile"; then
-    # Each line is of the form 'foo.o: dependency.h'.
-    # Do two passes, one to just change these to
-    #   $object: dependency.h
-    # and one to simply output
-    #   dependency.h:
-    # which is needed to avoid the deleted-header problem.
-    { sed -e "s,^.*\.[$lower]*:,$object:," < "$tmpdepfile"
-      sed -e "s,^.*\.[$lower]*:[$tab ]*,," -e 's,$,:,' < "$tmpdepfile"
-    } > "$depfile"
-    rm -f "$tmpdepfile"
-  else
-    make_dummy_depfile
-  fi
-}
-
-# A tabulation character.
-tab='	'
-# A newline character.
-nl='
-'
-# Character ranges might be problematic outside the C locale.
-# These definitions help.
-upper=ABCDEFGHIJKLMNOPQRSTUVWXYZ
-lower=abcdefghijklmnopqrstuvwxyz
-digits=0123456789
-alpha=${upper}${lower}
-
-if test -z "$depmode" || test -z "$source" || test -z "$object"; then
-  echo "depcomp: Variables source, object and depmode must be set" 1>&2
-  exit 1
-fi
-
-# Dependencies for sub/bar.o or sub/bar.obj go into sub/.deps/bar.Po.
-depfile=${depfile-`echo "$object" |
-  sed 's|[^\\/]*$|'${DEPDIR-.deps}'/&|;s|\.\([^.]*\)$|.P\1|;s|Pobj$|Po|'`}
-tmpdepfile=${tmpdepfile-`echo "$depfile" | sed 's/\.\([^.]*\)$/.T\1/'`}
-
-rm -f "$tmpdepfile"
-
-# Avoid interferences from the environment.
-gccflag= dashmflag=
-
-# Some modes work just like other modes, but use different flags.  We
-# parameterize here, but still list the modes in the big case below,
-# to make depend.m4 easier to write.  Note that we *cannot* use a case
-# here, because this file can only contain one case statement.
-if test "$depmode" = hp; then
-  # HP compiler uses -M and no extra arg.
-  gccflag=-M
-  depmode=gcc
-fi
-
-if test "$depmode" = dashXmstdout; then
-  # This is just like dashmstdout with a different argument.
-  dashmflag=-xM
-  depmode=dashmstdout
-fi
-
-cygpath_u="cygpath -u -f -"
-if test "$depmode" = msvcmsys; then
-  # This is just like msvisualcpp but w/o cygpath translation.
-  # Just convert the backslash-escaped backslashes to single forward
-  # slashes to satisfy depend.m4
-  cygpath_u='sed s,\\\\,/,g'
-  depmode=msvisualcpp
-fi
-
-if test "$depmode" = msvc7msys; then
-  # This is just like msvc7 but w/o cygpath translation.
-  # Just convert the backslash-escaped backslashes to single forward
-  # slashes to satisfy depend.m4
-  cygpath_u='sed s,\\\\,/,g'
-  depmode=msvc7
-fi
-
-if test "$depmode" = xlc; then
-  # IBM C/C++ Compilers xlc/xlC can output gcc-like dependency information.
-  gccflag=-qmakedep=gcc,-MF
-  depmode=gcc
-fi
-
-case "$depmode" in
-gcc3)
-## gcc 3 implements dependency tracking that does exactly what
-## we want.  Yay!  Note: for some reason libtool 1.4 doesn't like
-## it if -MD -MP comes after the -MF stuff.  Hmm.
-## Unfortunately, FreeBSD c89 acceptance of flags depends upon
-## the command line argument order; so add the flags where they
-## appear in depend2.am.  Note that the slowdown incurred here
-## affects only configure: in makefiles, %FASTDEP% shortcuts this.
-  for arg
-  do
-    case $arg in
-    -c) set fnord "$@" -MT "$object" -MD -MP -MF "$tmpdepfile" "$arg" ;;
-    *)  set fnord "$@" "$arg" ;;
-    esac
-    shift # fnord
-    shift # $arg
-  done
-  "$@"
-  stat=$?
-  if test $stat -ne 0; then
-    rm -f "$tmpdepfile"
-    exit $stat
-  fi
-  mv "$tmpdepfile" "$depfile"
-  ;;
-
-gcc)
-## Note that this doesn't just cater to obsosete pre-3.x GCC compilers.
-## but also to in-use compilers like IMB xlc/xlC and the HP C compiler.
-## (see the conditional assignment to $gccflag above).
-## There are various ways to get dependency output from gcc.  Here's
-## why we pick this rather obscure method:
-## - Don't want to use -MD because we'd like the dependencies to end
-##   up in a subdir.  Having to rename by hand is ugly.
-##   (We might end up doing this anyway to support other compilers.)
-## - The DEPENDENCIES_OUTPUT environment variable makes gcc act like
-##   -MM, not -M (despite what the docs say).  Also, it might not be
-##   supported by the other compilers which use the 'gcc' depmode.
-## - Using -M directly means running the compiler twice (even worse
-##   than renaming).
-  if test -z "$gccflag"; then
-    gccflag=-MD,
-  fi
-  "$@" -Wp,"$gccflag$tmpdepfile"
-  stat=$?
-  if test $stat -ne 0; then
-    rm -f "$tmpdepfile"
-    exit $stat
-  fi
-  rm -f "$depfile"
-  echo "$object : \\" > "$depfile"
-  # The second -e expression handles DOS-style file names with drive
-  # letters.
-  sed -e 's/^[^:]*: / /' \
-      -e 's/^['$alpha']:\/[^:]*: / /' < "$tmpdepfile" >> "$depfile"
-## This next piece of magic avoids the "deleted header file" problem.
-## The problem is that when a header file which appears in a .P file
-## is deleted, the dependency causes make to die (because there is
-## typically no way to rebuild the header).  We avoid this by adding
-## dummy dependencies for each header file.  Too bad gcc doesn't do
-## this for us directly.
-## Some versions of gcc put a space before the ':'.  On the theory
-## that the space means something, we add a space to the output as
-## well.  hp depmode also adds that space, but also prefixes the VPATH
-## to the object.  Take care to not repeat it in the output.
-## Some versions of the HPUX 10.20 sed can't process this invocation
-## correctly.  Breaking it into two sed invocations is a workaround.
-  tr ' ' "$nl" < "$tmpdepfile" \
-    | sed -e 's/^\\$//' -e '/^$/d' -e "s|.*$object$||" -e '/:$/d' \
-    | sed -e 's/$/ :/' >> "$depfile"
-  rm -f "$tmpdepfile"
-  ;;
-
-hp)
-  # This case exists only to let depend.m4 do its work.  It works by
-  # looking at the text of this script.  This case will never be run,
-  # since it is checked for above.
-  exit 1
-  ;;
-
-sgi)
-  if test "$libtool" = yes; then
-    "$@" "-Wp,-MDupdate,$tmpdepfile"
-  else
-    "$@" -MDupdate "$tmpdepfile"
-  fi
-  stat=$?
-  if test $stat -ne 0; then
-    rm -f "$tmpdepfile"
-    exit $stat
-  fi
-  rm -f "$depfile"
-
-  if test -f "$tmpdepfile"; then  # yes, the sourcefile depend on other files
-    echo "$object : \\" > "$depfile"
-    # Clip off the initial element (the dependent).  Don't try to be
-    # clever and replace this with sed code, as IRIX sed won't handle
-    # lines with more than a fixed number of characters (4096 in
-    # IRIX 6.2 sed, 8192 in IRIX 6.5).  We also remove comment lines;
-    # the IRIX cc adds comments like '#:fec' to the end of the
-    # dependency line.
-    tr ' ' "$nl" < "$tmpdepfile" \
-      | sed -e 's/^.*\.o://' -e 's/#.*$//' -e '/^$/ d' \
-      | tr "$nl" ' ' >> "$depfile"
-    echo >> "$depfile"
-    # The second pass generates a dummy entry for each header file.
-    tr ' ' "$nl" < "$tmpdepfile" \
-      | sed -e 's/^.*\.o://' -e 's/#.*$//' -e '/^$/ d' -e 's/$/:/' \
-      >> "$depfile"
-  else
-    make_dummy_depfile
-  fi
-  rm -f "$tmpdepfile"
-  ;;
-
-xlc)
-  # This case exists only to let depend.m4 do its work.  It works by
-  # looking at the text of this script.  This case will never be run,
-  # since it is checked for above.
-  exit 1
-  ;;
-
-aix)
-  # The C for AIX Compiler uses -M and outputs the dependencies
-  # in a .u file.  In older versions, this file always lives in the
-  # current directory.  Also, the AIX compiler puts '$object:' at the
-  # start of each line; $object doesn't have directory information.
-  # Version 6 uses the directory in both cases.
-  set_dir_from "$object"
-  set_base_from "$object"
-  if test "$libtool" = yes; then
-    tmpdepfile1=$dir$base.u
-    tmpdepfile2=$base.u
-    tmpdepfile3=$dir.libs/$base.u
-    "$@" -Wc,-M
-  else
-    tmpdepfile1=$dir$base.u
-    tmpdepfile2=$dir$base.u
-    tmpdepfile3=$dir$base.u
-    "$@" -M
-  fi
-  stat=$?
-  if test $stat -ne 0; then
-    rm -f "$tmpdepfile1" "$tmpdepfile2" "$tmpdepfile3"
-    exit $stat
-  fi
-
-  for tmpdepfile in "$tmpdepfile1" "$tmpdepfile2" "$tmpdepfile3"
-  do
-    test -f "$tmpdepfile" && break
-  done
-  aix_post_process_depfile
-  ;;
-
-tcc)
-  # tcc (Tiny C Compiler) understand '-MD -MF file' since version 0.9.26
-  # FIXME: That version still under development at the moment of writing.
-  #        Make that this statement remains true also for stable, released
-  #        versions.
-  # It will wrap lines (doesn't matter whether long or short) with a
-  # trailing '\', as in:
-  #
-  #   foo.o : \
-  #    foo.c \
-  #    foo.h \
-  #
-  # It will put a trailing '\' even on the last line, and will use leading
-  # spaces rather than leading tabs (at least since its commit 0394caf7
-  # "Emit spaces for -MD").
-  "$@" -MD -MF "$tmpdepfile"
-  stat=$?
-  if test $stat -ne 0; then
-    rm -f "$tmpdepfile"
-    exit $stat
-  fi
-  rm -f "$depfile"
-  # Each non-empty line is of the form 'foo.o : \' or ' dep.h \'.
-  # We have to change lines of the first kind to '$object: \'.
-  sed -e "s|.*:|$object :|" < "$tmpdepfile" > "$depfile"
-  # And for each line of the second kind, we have to emit a 'dep.h:'
-  # dummy dependency, to avoid the deleted-header problem.
-  sed -n -e 's|^  *\(.*\) *\\$|\1:|p' < "$tmpdepfile" >> "$depfile"
-  rm -f "$tmpdepfile"
-  ;;
-
-## The order of this option in the case statement is important, since the
-## shell code in configure will try each of these formats in the order
-## listed in this file.  A plain '-MD' option would be understood by many
-## compilers, so we must ensure this comes after the gcc and icc options.
-pgcc)
-  # Portland's C compiler understands '-MD'.
-  # Will always output deps to 'file.d' where file is the root name of the
-  # source file under compilation, even if file resides in a subdirectory.
-  # The object file name does not affect the name of the '.d' file.
-  # pgcc 10.2 will output
-  #    foo.o: sub/foo.c sub/foo.h
-  # and will wrap long lines using '\' :
-  #    foo.o: sub/foo.c ... \
-  #     sub/foo.h ... \
-  #     ...
-  set_dir_from "$object"
-  # Use the source, not the object, to determine the base name, since
-  # that's sadly what pgcc will do too.
-  set_base_from "$source"
-  tmpdepfile=$base.d
-
-  # For projects that build the same source file twice into different object
-  # files, the pgcc approach of using the *source* file root name can cause
-  # problems in parallel builds.  Use a locking strategy to avoid stomping on
-  # the same $tmpdepfile.
-  lockdir=$base.d-lock
-  trap "
-    echo '$0: caught signal, cleaning up...' >&2
-    rmdir '$lockdir'
-    exit 1
-  " 1 2 13 15
-  numtries=100
-  i=$numtries
-  while test $i -gt 0; do
-    # mkdir is a portable test-and-set.
-    if mkdir "$lockdir" 2>/dev/null; then
-      # This process acquired the lock.
-      "$@" -MD
-      stat=$?
-      # Release the lock.
-      rmdir "$lockdir"
-      break
-    else
-      # If the lock is being held by a different process, wait
-      # until the winning process is done or we timeout.
-      while test -d "$lockdir" && test $i -gt 0; do
-        sleep 1
-        i=`expr $i - 1`
-      done
-    fi
-    i=`expr $i - 1`
-  done
-  trap - 1 2 13 15
-  if test $i -le 0; then
-    echo "$0: failed to acquire lock after $numtries attempts" >&2
-    echo "$0: check lockdir '$lockdir'" >&2
-    exit 1
-  fi
-
-  if test $stat -ne 0; then
-    rm -f "$tmpdepfile"
-    exit $stat
-  fi
-  rm -f "$depfile"
-  # Each line is of the form `foo.o: dependent.h',
-  # or `foo.o: dep1.h dep2.h \', or ` dep3.h dep4.h \'.
-  # Do two passes, one to just change these to
-  # `$object: dependent.h' and one to simply `dependent.h:'.
-  sed "s,^[^:]*:,$object :," < "$tmpdepfile" > "$depfile"
-  # Some versions of the HPUX 10.20 sed can't process this invocation
-  # correctly.  Breaking it into two sed invocations is a workaround.
-  sed 's,^[^:]*: \(.*\)$,\1,;s/^\\$//;/^$/d;/:$/d' < "$tmpdepfile" \
-    | sed -e 's/$/ :/' >> "$depfile"
-  rm -f "$tmpdepfile"
-  ;;
-
-hp2)
-  # The "hp" stanza above does not work with aCC (C++) and HP's ia64
-  # compilers, which have integrated preprocessors.  The correct option
-  # to use with these is +Maked; it writes dependencies to a file named
-  # 'foo.d', which lands next to the object file, wherever that
-  # happens to be.
-  # Much of this is similar to the tru64 case; see comments there.
-  set_dir_from  "$object"
-  set_base_from "$object"
-  if test "$libtool" = yes; then
-    tmpdepfile1=$dir$base.d
-    tmpdepfile2=$dir.libs/$base.d
-    "$@" -Wc,+Maked
-  else
-    tmpdepfile1=$dir$base.d
-    tmpdepfile2=$dir$base.d
-    "$@" +Maked
-  fi
-  stat=$?
-  if test $stat -ne 0; then
-     rm -f "$tmpdepfile1" "$tmpdepfile2"
-     exit $stat
-  fi
-
-  for tmpdepfile in "$tmpdepfile1" "$tmpdepfile2"
-  do
-    test -f "$tmpdepfile" && break
-  done
-  if test -f "$tmpdepfile"; then
-    sed -e "s,^.*\.[$lower]*:,$object:," "$tmpdepfile" > "$depfile"
-    # Add 'dependent.h:' lines.
-    sed -ne '2,${
-               s/^ *//
-               s/ \\*$//
-               s/$/:/
-               p
-             }' "$tmpdepfile" >> "$depfile"
-  else
-    make_dummy_depfile
-  fi
-  rm -f "$tmpdepfile" "$tmpdepfile2"
-  ;;
-
-tru64)
-  # The Tru64 compiler uses -MD to generate dependencies as a side
-  # effect.  'cc -MD -o foo.o ...' puts the dependencies into 'foo.o.d'.
-  # At least on Alpha/Redhat 6.1, Compaq CCC V6.2-504 seems to put
-  # dependencies in 'foo.d' instead, so we check for that too.
-  # Subdirectories are respected.
-  set_dir_from  "$object"
-  set_base_from "$object"
-
-  if test "$libtool" = yes; then
-    # Libtool generates 2 separate objects for the 2 libraries.  These
-    # two compilations output dependencies in $dir.libs/$base.o.d and
-    # in $dir$base.o.d.  We have to check for both files, because
-    # one of the two compilations can be disabled.  We should prefer
-    # $dir$base.o.d over $dir.libs/$base.o.d because the latter is
-    # automatically cleaned when .libs/ is deleted, while ignoring
-    # the former would cause a distcleancheck panic.
-    tmpdepfile1=$dir$base.o.d          # libtool 1.5
-    tmpdepfile2=$dir.libs/$base.o.d    # Likewise.
-    tmpdepfile3=$dir.libs/$base.d      # Compaq CCC V6.2-504
-    "$@" -Wc,-MD
-  else
-    tmpdepfile1=$dir$base.d
-    tmpdepfile2=$dir$base.d
-    tmpdepfile3=$dir$base.d
-    "$@" -MD
-  fi
-
-  stat=$?
-  if test $stat -ne 0; then
-    rm -f "$tmpdepfile1" "$tmpdepfile2" "$tmpdepfile3"
-    exit $stat
-  fi
-
-  for tmpdepfile in "$tmpdepfile1" "$tmpdepfile2" "$tmpdepfile3"
-  do
-    test -f "$tmpdepfile" && break
-  done
-  # Same post-processing that is required for AIX mode.
-  aix_post_process_depfile
-  ;;
-
-msvc7)
-  if test "$libtool" = yes; then
-    showIncludes=-Wc,-showIncludes
-  else
-    showIncludes=-showIncludes
-  fi
-  "$@" $showIncludes > "$tmpdepfile"
-  stat=$?
-  grep -v '^Note: including file: ' "$tmpdepfile"
-  if test $stat -ne 0; then
-    rm -f "$tmpdepfile"
-    exit $stat
-  fi
-  rm -f "$depfile"
-  echo "$object : \\" > "$depfile"
-  # The first sed program below extracts the file names and escapes
-  # backslashes for cygpath.  The second sed program outputs the file
-  # name when reading, but also accumulates all include files in the
-  # hold buffer in order to output them again at the end.  This only
-  # works with sed implementations that can handle large buffers.
-  sed < "$tmpdepfile" -n '
-/^Note: including file:  *\(.*\)/ {
-  s//\1/
-  s/\\/\\\\/g
-  p
-}' | $cygpath_u | sort -u | sed -n '
-s/ /\\ /g
-s/\(.*\)/'"$tab"'\1 \\/p
-s/.\(.*\) \\/\1:/
-H
-$ {
-  s/.*/'"$tab"'/
-  G
-  p
-}' >> "$depfile"
-  echo >> "$depfile" # make sure the fragment doesn't end with a backslash
-  rm -f "$tmpdepfile"
-  ;;
-
-msvc7msys)
-  # This case exists only to let depend.m4 do its work.  It works by
-  # looking at the text of this script.  This case will never be run,
-  # since it is checked for above.
-  exit 1
-  ;;
-
-#nosideeffect)
-  # This comment above is used by automake to tell side-effect
-  # dependency tracking mechanisms from slower ones.
-
-dashmstdout)
-  # Important note: in order to support this mode, a compiler *must*
-  # always write the preprocessed file to stdout, regardless of -o.
-  "$@" || exit $?
-
-  # Remove the call to Libtool.
-  if test "$libtool" = yes; then
-    while test "X$1" != 'X--mode=compile'; do
-      shift
-    done
-    shift
-  fi
-
-  # Remove '-o $object'.
-  IFS=" "
-  for arg
-  do
-    case $arg in
-    -o)
-      shift
-      ;;
-    $object)
-      shift
-      ;;
-    *)
-      set fnord "$@" "$arg"
-      shift # fnord
-      shift # $arg
-      ;;
-    esac
-  done
-
-  test -z "$dashmflag" && dashmflag=-M
-  # Require at least two characters before searching for ':'
-  # in the target name.  This is to cope with DOS-style filenames:
-  # a dependency such as 'c:/foo/bar' could be seen as target 'c' otherwise.
-  "$@" $dashmflag |
-    sed "s|^[$tab ]*[^:$tab ][^:][^:]*:[$tab ]*|$object: |" > "$tmpdepfile"
-  rm -f "$depfile"
-  cat < "$tmpdepfile" > "$depfile"
-  # Some versions of the HPUX 10.20 sed can't process this sed invocation
-  # correctly.  Breaking it into two sed invocations is a workaround.
-  tr ' ' "$nl" < "$tmpdepfile" \
-    | sed -e 's/^\\$//' -e '/^$/d' -e '/:$/d' \
-    | sed -e 's/$/ :/' >> "$depfile"
-  rm -f "$tmpdepfile"
-  ;;
-
-dashXmstdout)
-  # This case only exists to satisfy depend.m4.  It is never actually
-  # run, as this mode is specially recognized in the preamble.
-  exit 1
-  ;;
-
-makedepend)
-  "$@" || exit $?
-  # Remove any Libtool call
-  if test "$libtool" = yes; then
-    while test "X$1" != 'X--mode=compile'; do
-      shift
-    done
-    shift
-  fi
-  # X makedepend
-  shift
-  cleared=no eat=no
-  for arg
-  do
-    case $cleared in
-    no)
-      set ""; shift
-      cleared=yes ;;
-    esac
-    if test $eat = yes; then
-      eat=no
-      continue
-    fi
-    case "$arg" in
-    -D*|-I*)
-      set fnord "$@" "$arg"; shift ;;
-    # Strip any option that makedepend may not understand.  Remove
-    # the object too, otherwise makedepend will parse it as a source file.
-    -arch)
-      eat=yes ;;
-    -*|$object)
-      ;;
-    *)
-      set fnord "$@" "$arg"; shift ;;
-    esac
-  done
-  obj_suffix=`echo "$object" | sed 's/^.*\././'`
-  touch "$tmpdepfile"
-  ${MAKEDEPEND-makedepend} -o"$obj_suffix" -f"$tmpdepfile" "$@"
-  rm -f "$depfile"
-  # makedepend may prepend the VPATH from the source file name to the object.
-  # No need to regex-escape $object, excess matching of '.' is harmless.
-  sed "s|^.*\($object *:\)|\1|" "$tmpdepfile" > "$depfile"
-  # Some versions of the HPUX 10.20 sed can't process the last invocation
-  # correctly.  Breaking it into two sed invocations is a workaround.
-  sed '1,2d' "$tmpdepfile" \
-    | tr ' ' "$nl" \
-    | sed -e 's/^\\$//' -e '/^$/d' -e '/:$/d' \
-    | sed -e 's/$/ :/' >> "$depfile"
-  rm -f "$tmpdepfile" "$tmpdepfile".bak
-  ;;
-
-cpp)
-  # Important note: in order to support this mode, a compiler *must*
-  # always write the preprocessed file to stdout.
-  "$@" || exit $?
-
-  # Remove the call to Libtool.
-  if test "$libtool" = yes; then
-    while test "X$1" != 'X--mode=compile'; do
-      shift
-    done
-    shift
-  fi
-
-  # Remove '-o $object'.
-  IFS=" "
-  for arg
-  do
-    case $arg in
-    -o)
-      shift
-      ;;
-    $object)
-      shift
-      ;;
-    *)
-      set fnord "$@" "$arg"
-      shift # fnord
-      shift # $arg
-      ;;
-    esac
-  done
-
-  "$@" -E \
-    | sed -n -e '/^# [0-9][0-9]* "\([^"]*\)".*/ s:: \1 \\:p' \
-             -e '/^#line [0-9][0-9]* "\([^"]*\)".*/ s:: \1 \\:p' \
-    | sed '$ s: \\$::' > "$tmpdepfile"
-  rm -f "$depfile"
-  echo "$object : \\" > "$depfile"
-  cat < "$tmpdepfile" >> "$depfile"
-  sed < "$tmpdepfile" '/^$/d;s/^ //;s/ \\$//;s/$/ :/' >> "$depfile"
-  rm -f "$tmpdepfile"
-  ;;
-
-msvisualcpp)
-  # Important note: in order to support this mode, a compiler *must*
-  # always write the preprocessed file to stdout.
-  "$@" || exit $?
-
-  # Remove the call to Libtool.
-  if test "$libtool" = yes; then
-    while test "X$1" != 'X--mode=compile'; do
-      shift
-    done
-    shift
-  fi
-
-  IFS=" "
-  for arg
-  do
-    case "$arg" in
-    -o)
-      shift
-      ;;
-    $object)
-      shift
-      ;;
-    "-Gm"|"/Gm"|"-Gi"|"/Gi"|"-ZI"|"/ZI")
-        set fnord "$@"
-        shift
-        shift
-        ;;
-    *)
-        set fnord "$@" "$arg"
-        shift
-        shift
-        ;;
-    esac
-  done
-  "$@" -E 2>/dev/null |
-  sed -n '/^#line [0-9][0-9]* "\([^"]*\)"/ s::\1:p' | $cygpath_u | sort -u > "$tmpdepfile"
-  rm -f "$depfile"
-  echo "$object : \\" > "$depfile"
-  sed < "$tmpdepfile" -n -e 's% %\\ %g' -e '/^\(.*\)$/ s::'"$tab"'\1 \\:p' >> "$depfile"
-  echo "$tab" >> "$depfile"
-  sed < "$tmpdepfile" -n -e 's% %\\ %g' -e '/^\(.*\)$/ s::\1\::p' >> "$depfile"
-  rm -f "$tmpdepfile"
-  ;;
-
-msvcmsys)
-  # This case exists only to let depend.m4 do its work.  It works by
-  # looking at the text of this script.  This case will never be run,
-  # since it is checked for above.
-  exit 1
-  ;;
-
-none)
-  exec "$@"
-  ;;
-
-*)
-  echo "Unknown depmode $depmode" 1>&2
-  exit 1
-  ;;
-esac
-
-exit 0
-
-# Local Variables:
-# mode: shell-script
-# sh-indentation: 2
-# eval: (add-hook 'write-file-hooks 'time-stamp)
-# time-stamp-start: "scriptversion="
-# time-stamp-format: "%:y-%02m-%02d.%02H"
-# time-stamp-time-zone: "UTC"
-# time-stamp-end: "; # UTC"
-# End:
diff --git a/depcomp b/depcomp
new file mode 120000
index 00000000..b0ad20c0
--- /dev/null
+++ b/depcomp
@@ -0,0 +1 @@
+/usr/share/automake-1.11/depcomp
\ No newline at end of file
diff --git a/install-sh b/install-sh
deleted file mode 100755
index 0b0fdcbb..00000000
--- a/install-sh
+++ /dev/null
@@ -1,501 +0,0 @@
-#!/bin/sh
-# install - install a program, script, or datafile
-
-scriptversion=2013-12-25.23; # UTC
-
-# This originates from X11R5 (mit/util/scripts/install.sh), which was
-# later released in X11R6 (xc/config/util/install.sh) with the
-# following copyright and license.
-#
-# Copyright (C) 1994 X Consortium
-#
-# Permission is hereby granted, free of charge, to any person obtaining a copy
-# of this software and associated documentation files (the "Software"), to
-# deal in the Software without restriction, including without limitation the
-# rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
-# sell copies of the Software, and to permit persons to whom the Software is
-# furnished to do so, subject to the following conditions:
-#
-# The above copyright notice and this permission notice shall be included in
-# all copies or substantial portions of the Software.
-#
-# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL THE
-# X CONSORTIUM BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
-# AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNEC-
-# TION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-#
-# Except as contained in this notice, the name of the X Consortium shall not
-# be used in advertising or otherwise to promote the sale, use or other deal-
-# ings in this Software without prior written authorization from the X Consor-
-# tium.
-#
-#
-# FSF changes to this file are in the public domain.
-#
-# Calling this script install-sh is preferred over install.sh, to prevent
-# 'make' implicit rules from creating a file called install from it
-# when there is no Makefile.
-#
-# This script is compatible with the BSD install script, but was written
-# from scratch.
-
-tab='	'
-nl='
-'
-IFS=" $tab$nl"
-
-# Set DOITPROG to "echo" to test this script.
-
-doit=${DOITPROG-}
-doit_exec=${doit:-exec}
-
-# Put in absolute file names if you don't have them in your path;
-# or use environment vars.
-
-chgrpprog=${CHGRPPROG-chgrp}
-chmodprog=${CHMODPROG-chmod}
-chownprog=${CHOWNPROG-chown}
-cmpprog=${CMPPROG-cmp}
-cpprog=${CPPROG-cp}
-mkdirprog=${MKDIRPROG-mkdir}
-mvprog=${MVPROG-mv}
-rmprog=${RMPROG-rm}
-stripprog=${STRIPPROG-strip}
-
-posix_mkdir=
-
-# Desired mode of installed file.
-mode=0755
-
-chgrpcmd=
-chmodcmd=$chmodprog
-chowncmd=
-mvcmd=$mvprog
-rmcmd="$rmprog -f"
-stripcmd=
-
-src=
-dst=
-dir_arg=
-dst_arg=
-
-copy_on_change=false
-is_target_a_directory=possibly
-
-usage="\
-Usage: $0 [OPTION]... [-T] SRCFILE DSTFILE
-   or: $0 [OPTION]... SRCFILES... DIRECTORY
-   or: $0 [OPTION]... -t DIRECTORY SRCFILES...
-   or: $0 [OPTION]... -d DIRECTORIES...
-
-In the 1st form, copy SRCFILE to DSTFILE.
-In the 2nd and 3rd, copy all SRCFILES to DIRECTORY.
-In the 4th, create DIRECTORIES.
-
-Options:
-     --help     display this help and exit.
-     --version  display version info and exit.
-
-  -c            (ignored)
-  -C            install only if different (preserve the last data modification time)
-  -d            create directories instead of installing files.
-  -g GROUP      $chgrpprog installed files to GROUP.
-  -m MODE       $chmodprog installed files to MODE.
-  -o USER       $chownprog installed files to USER.
-  -s            $stripprog installed files.
-  -t DIRECTORY  install into DIRECTORY.
-  -T            report an error if DSTFILE is a directory.
-
-Environment variables override the default commands:
-  CHGRPPROG CHMODPROG CHOWNPROG CMPPROG CPPROG MKDIRPROG MVPROG
-  RMPROG STRIPPROG
-"
-
-while test $# -ne 0; do
-  case $1 in
-    -c) ;;
-
-    -C) copy_on_change=true;;
-
-    -d) dir_arg=true;;
-
-    -g) chgrpcmd="$chgrpprog $2"
-        shift;;
-
-    --help) echo "$usage"; exit $?;;
-
-    -m) mode=$2
-        case $mode in
-          *' '* | *"$tab"* | *"$nl"* | *'*'* | *'?'* | *'['*)
-            echo "$0: invalid mode: $mode" >&2
-            exit 1;;
-        esac
-        shift;;
-
-    -o) chowncmd="$chownprog $2"
-        shift;;
-
-    -s) stripcmd=$stripprog;;
-
-    -t)
-        is_target_a_directory=always
-        dst_arg=$2
-        # Protect names problematic for 'test' and other utilities.
-        case $dst_arg in
-          -* | [=\(\)!]) dst_arg=./$dst_arg;;
-        esac
-        shift;;
-
-    -T) is_target_a_directory=never;;
-
-    --version) echo "$0 $scriptversion"; exit $?;;
-
-    --) shift
-        break;;
-
-    -*) echo "$0: invalid option: $1" >&2
-        exit 1;;
-
-    *)  break;;
-  esac
-  shift
-done
-
-# We allow the use of options -d and -T together, by making -d
-# take the precedence; this is for compatibility with GNU install.
-
-if test -n "$dir_arg"; then
-  if test -n "$dst_arg"; then
-    echo "$0: target directory not allowed when installing a directory." >&2
-    exit 1
-  fi
-fi
-
-if test $# -ne 0 && test -z "$dir_arg$dst_arg"; then
-  # When -d is used, all remaining arguments are directories to create.
-  # When -t is used, the destination is already specified.
-  # Otherwise, the last argument is the destination.  Remove it from $@.
-  for arg
-  do
-    if test -n "$dst_arg"; then
-      # $@ is not empty: it contains at least $arg.
-      set fnord "$@" "$dst_arg"
-      shift # fnord
-    fi
-    shift # arg
-    dst_arg=$arg
-    # Protect names problematic for 'test' and other utilities.
-    case $dst_arg in
-      -* | [=\(\)!]) dst_arg=./$dst_arg;;
-    esac
-  done
-fi
-
-if test $# -eq 0; then
-  if test -z "$dir_arg"; then
-    echo "$0: no input file specified." >&2
-    exit 1
-  fi
-  # It's OK to call 'install-sh -d' without argument.
-  # This can happen when creating conditional directories.
-  exit 0
-fi
-
-if test -z "$dir_arg"; then
-  if test $# -gt 1 || test "$is_target_a_directory" = always; then
-    if test ! -d "$dst_arg"; then
-      echo "$0: $dst_arg: Is not a directory." >&2
-      exit 1
-    fi
-  fi
-fi
-
-if test -z "$dir_arg"; then
-  do_exit='(exit $ret); exit $ret'
-  trap "ret=129; $do_exit" 1
-  trap "ret=130; $do_exit" 2
-  trap "ret=141; $do_exit" 13
-  trap "ret=143; $do_exit" 15
-
-  # Set umask so as not to create temps with too-generous modes.
-  # However, 'strip' requires both read and write access to temps.
-  case $mode in
-    # Optimize common cases.
-    *644) cp_umask=133;;
-    *755) cp_umask=22;;
-
-    *[0-7])
-      if test -z "$stripcmd"; then
-        u_plus_rw=
-      else
-        u_plus_rw='% 200'
-      fi
-      cp_umask=`expr '(' 777 - $mode % 1000 ')' $u_plus_rw`;;
-    *)
-      if test -z "$stripcmd"; then
-        u_plus_rw=
-      else
-        u_plus_rw=,u+rw
-      fi
-      cp_umask=$mode$u_plus_rw;;
-  esac
-fi
-
-for src
-do
-  # Protect names problematic for 'test' and other utilities.
-  case $src in
-    -* | [=\(\)!]) src=./$src;;
-  esac
-
-  if test -n "$dir_arg"; then
-    dst=$src
-    dstdir=$dst
-    test -d "$dstdir"
-    dstdir_status=$?
-  else
-
-    # Waiting for this to be detected by the "$cpprog $src $dsttmp" command
-    # might cause directories to be created, which would be especially bad
-    # if $src (and thus $dsttmp) contains '*'.
-    if test ! -f "$src" && test ! -d "$src"; then
-      echo "$0: $src does not exist." >&2
-      exit 1
-    fi
-
-    if test -z "$dst_arg"; then
-      echo "$0: no destination specified." >&2
-      exit 1
-    fi
-    dst=$dst_arg
-
-    # If destination is a directory, append the input filename; won't work
-    # if double slashes aren't ignored.
-    if test -d "$dst"; then
-      if test "$is_target_a_directory" = never; then
-        echo "$0: $dst_arg: Is a directory" >&2
-        exit 1
-      fi
-      dstdir=$dst
-      dst=$dstdir/`basename "$src"`
-      dstdir_status=0
-    else
-      dstdir=`dirname "$dst"`
-      test -d "$dstdir"
-      dstdir_status=$?
-    fi
-  fi
-
-  obsolete_mkdir_used=false
-
-  if test $dstdir_status != 0; then
-    case $posix_mkdir in
-      '')
-        # Create intermediate dirs using mode 755 as modified by the umask.
-        # This is like FreeBSD 'install' as of 1997-10-28.
-        umask=`umask`
-        case $stripcmd.$umask in
-          # Optimize common cases.
-          *[2367][2367]) mkdir_umask=$umask;;
-          .*0[02][02] | .[02][02] | .[02]) mkdir_umask=22;;
-
-          *[0-7])
-            mkdir_umask=`expr $umask + 22 \
-              - $umask % 100 % 40 + $umask % 20 \
-              - $umask % 10 % 4 + $umask % 2
-            `;;
-          *) mkdir_umask=$umask,go-w;;
-        esac
-
-        # With -d, create the new directory with the user-specified mode.
-        # Otherwise, rely on $mkdir_umask.
-        if test -n "$dir_arg"; then
-          mkdir_mode=-m$mode
-        else
-          mkdir_mode=
-        fi
-
-        posix_mkdir=false
-        case $umask in
-          *[123567][0-7][0-7])
-            # POSIX mkdir -p sets u+wx bits regardless of umask, which
-            # is incompatible with FreeBSD 'install' when (umask & 300) != 0.
-            ;;
-          *)
-            tmpdir=${TMPDIR-/tmp}/ins$RANDOM-$$
-            trap 'ret=$?; rmdir "$tmpdir/d" "$tmpdir" 2>/dev/null; exit $ret' 0
-
-            if (umask $mkdir_umask &&
-                exec $mkdirprog $mkdir_mode -p -- "$tmpdir/d") >/dev/null 2>&1
-            then
-              if test -z "$dir_arg" || {
-                   # Check for POSIX incompatibilities with -m.
-                   # HP-UX 11.23 and IRIX 6.5 mkdir -m -p sets group- or
-                   # other-writable bit of parent directory when it shouldn't.
-                   # FreeBSD 6.1 mkdir -m -p sets mode of existing directory.
-                   ls_ld_tmpdir=`ls -ld "$tmpdir"`
-                   case $ls_ld_tmpdir in
-                     d????-?r-*) different_mode=700;;
-                     d????-?--*) different_mode=755;;
-                     *) false;;
-                   esac &&
-                   $mkdirprog -m$different_mode -p -- "$tmpdir" && {
-                     ls_ld_tmpdir_1=`ls -ld "$tmpdir"`
-                     test "$ls_ld_tmpdir" = "$ls_ld_tmpdir_1"
-                   }
-                 }
-              then posix_mkdir=:
-              fi
-              rmdir "$tmpdir/d" "$tmpdir"
-            else
-              # Remove any dirs left behind by ancient mkdir implementations.
-              rmdir ./$mkdir_mode ./-p ./-- 2>/dev/null
-            fi
-            trap '' 0;;
-        esac;;
-    esac
-
-    if
-      $posix_mkdir && (
-        umask $mkdir_umask &&
-        $doit_exec $mkdirprog $mkdir_mode -p -- "$dstdir"
-      )
-    then :
-    else
-
-      # The umask is ridiculous, or mkdir does not conform to POSIX,
-      # or it failed possibly due to a race condition.  Create the
-      # directory the slow way, step by step, checking for races as we go.
-
-      case $dstdir in
-        /*) prefix='/';;
-        [-=\(\)!]*) prefix='./';;
-        *)  prefix='';;
-      esac
-
-      oIFS=$IFS
-      IFS=/
-      set -f
-      set fnord $dstdir
-      shift
-      set +f
-      IFS=$oIFS
-
-      prefixes=
-
-      for d
-      do
-        test X"$d" = X && continue
-
-        prefix=$prefix$d
-        if test -d "$prefix"; then
-          prefixes=
-        else
-          if $posix_mkdir; then
-            (umask=$mkdir_umask &&
-             $doit_exec $mkdirprog $mkdir_mode -p -- "$dstdir") && break
-            # Don't fail if two instances are running concurrently.
-            test -d "$prefix" || exit 1
-          else
-            case $prefix in
-              *\'*) qprefix=`echo "$prefix" | sed "s/'/'\\\\\\\\''/g"`;;
-              *) qprefix=$prefix;;
-            esac
-            prefixes="$prefixes '$qprefix'"
-          fi
-        fi
-        prefix=$prefix/
-      done
-
-      if test -n "$prefixes"; then
-        # Don't fail if two instances are running concurrently.
-        (umask $mkdir_umask &&
-         eval "\$doit_exec \$mkdirprog $prefixes") ||
-          test -d "$dstdir" || exit 1
-        obsolete_mkdir_used=true
-      fi
-    fi
-  fi
-
-  if test -n "$dir_arg"; then
-    { test -z "$chowncmd" || $doit $chowncmd "$dst"; } &&
-    { test -z "$chgrpcmd" || $doit $chgrpcmd "$dst"; } &&
-    { test "$obsolete_mkdir_used$chowncmd$chgrpcmd" = false ||
-      test -z "$chmodcmd" || $doit $chmodcmd $mode "$dst"; } || exit 1
-  else
-
-    # Make a couple of temp file names in the proper directory.
-    dsttmp=$dstdir/_inst.$$_
-    rmtmp=$dstdir/_rm.$$_
-
-    # Trap to clean up those temp files at exit.
-    trap 'ret=$?; rm -f "$dsttmp" "$rmtmp" && exit $ret' 0
-
-    # Copy the file name to the temp name.
-    (umask $cp_umask && $doit_exec $cpprog "$src" "$dsttmp") &&
-
-    # and set any options; do chmod last to preserve setuid bits.
-    #
-    # If any of these fail, we abort the whole thing.  If we want to
-    # ignore errors from any of these, just make sure not to ignore
-    # errors from the above "$doit $cpprog $src $dsttmp" command.
-    #
-    { test -z "$chowncmd" || $doit $chowncmd "$dsttmp"; } &&
-    { test -z "$chgrpcmd" || $doit $chgrpcmd "$dsttmp"; } &&
-    { test -z "$stripcmd" || $doit $stripcmd "$dsttmp"; } &&
-    { test -z "$chmodcmd" || $doit $chmodcmd $mode "$dsttmp"; } &&
-
-    # If -C, don't bother to copy if it wouldn't change the file.
-    if $copy_on_change &&
-       old=`LC_ALL=C ls -dlL "$dst"     2>/dev/null` &&
-       new=`LC_ALL=C ls -dlL "$dsttmp"  2>/dev/null` &&
-       set -f &&
-       set X $old && old=:$2:$4:$5:$6 &&
-       set X $new && new=:$2:$4:$5:$6 &&
-       set +f &&
-       test "$old" = "$new" &&
-       $cmpprog "$dst" "$dsttmp" >/dev/null 2>&1
-    then
-      rm -f "$dsttmp"
-    else
-      # Rename the file to the real destination.
-      $doit $mvcmd -f "$dsttmp" "$dst" 2>/dev/null ||
-
-      # The rename failed, perhaps because mv can't rename something else
-      # to itself, or perhaps because mv is so ancient that it does not
-      # support -f.
-      {
-        # Now remove or move aside any old file at destination location.
-        # We try this two ways since rm can't unlink itself on some
-        # systems and the destination file might be busy for other
-        # reasons.  In this case, the final cleanup might fail but the new
-        # file should still install successfully.
-        {
-          test ! -f "$dst" ||
-          $doit $rmcmd -f "$dst" 2>/dev/null ||
-          { $doit $mvcmd -f "$dst" "$rmtmp" 2>/dev/null &&
-            { $doit $rmcmd -f "$rmtmp" 2>/dev/null; :; }
-          } ||
-          { echo "$0: cannot unlink or rename $dst" >&2
-            (exit 1); exit 1
-          }
-        } &&
-
-        # Now rename the file to the real destination.
-        $doit $mvcmd "$dsttmp" "$dst"
-      }
-    fi || exit 1
-
-    trap '' 0
-  fi
-done
-
-# Local variables:
-# eval: (add-hook 'write-file-hooks 'time-stamp)
-# time-stamp-start: "scriptversion="
-# time-stamp-format: "%:y-%02m-%02d.%02H"
-# time-stamp-time-zone: "UTC"
-# time-stamp-end: "; # UTC"
-# End:
diff --git a/install-sh b/install-sh
new file mode 120000
index 00000000..205f21c6
--- /dev/null
+++ b/install-sh
@@ -0,0 +1 @@
+/usr/share/automake-1.11/install-sh
\ No newline at end of file
diff --git a/lib/Grid_config.h b/lib/Grid_config.h
index 2397894f..be3a4b5e 100644
--- a/lib/Grid_config.h
+++ b/lib/Grid_config.h
@@ -2,7 +2,7 @@
 /* lib/Grid_config.h.in.  Generated from configure.ac by autoheader.  */
 
 /* AVX */
-/* #undef AVX1 */
+#define AVX1 1
 
 /* AVX2 */
 /* #undef AVX2 */
@@ -93,14 +93,11 @@
 /* Define to the one symbol short name of this package. */
 #define PACKAGE_TARNAME "grid"
 
-/* Define to the home page for this package. */
-#define PACKAGE_URL ""
-
 /* Define to the version of this package. */
 #define PACKAGE_VERSION "1.0"
 
 /* SSE4 */
-#define SSE4 1
+/* #undef SSE4 */
 
 /* Define to 1 if you have the ANSI C header files. */
 #define STDC_HEADERS 1
diff --git a/lib/Grid_config.h.in b/lib/Grid_config.h.in
index 6f05d6cb..ba21e1be 100644
--- a/lib/Grid_config.h.in
+++ b/lib/Grid_config.h.in
@@ -92,9 +92,6 @@
 /* Define to the one symbol short name of this package. */
 #undef PACKAGE_TARNAME
 
-/* Define to the home page for this package. */
-#undef PACKAGE_URL
-
 /* Define to the version of this package. */
 #undef PACKAGE_VERSION
 
diff --git a/lib/simd/Grid_avx.h b/lib/simd/Grid_avx.h
index 303f30c5..f9fa1f85 100644
--- a/lib/simd/Grid_avx.h
+++ b/lib/simd/Grid_avx.h
@@ -299,7 +299,7 @@ namespace Optimization {
   //////////////////////////////////////////////
   // Some Template specialization
   template < typename vtype > 
-    void permute(vtype a, vtype b, int perm) {
+    void permute(vtype &a, vtype &b, int perm) {
     union { 
       __m256 f;
       vtype v;
@@ -320,11 +320,16 @@ namespace Optimization {
   template<>
     inline Grid::ComplexF Reduce<Grid::ComplexF, __m256>::operator()(__m256 in){
     __m256 v1,v2;
+    union { 
+      __m256 v;
+      float f[8];
+    } conv;
     Optimization::permute(v1,in,0); // sse 128; paired complex single
     v1 = _mm256_add_ps(v1,in);
     Optimization::permute(v2,v1,1); // avx 256; quad complex single
     v1 = _mm256_add_ps(v1,v2);
-    return Grid::ComplexF(v1[0],v1[1]);
+    conv.v = v1;
+    return Grid::ComplexF(conv.f[0],conv.f[1]);
   }
   //Real float Reduce
   template<>
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index 41ce84ea..41706906 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -77,9 +77,7 @@ inline void Gpermute(vsimd &y,const vsimd &b,int perm){
       switch (perm){
 #if defined(AVX1)||defined(AVX2)
       // 8x32 bits=>3 permutes
-      case 2: 
-	conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); 
-	break;
+      case 2: conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); break;
       case 1: conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2)); break;
       case 0: conv.f = _mm256_permute2f128_ps(conv.f,conv.f,0x01); break;
 #endif
diff --git a/missing b/missing
deleted file mode 100755
index f62bbae3..00000000
--- a/missing
+++ /dev/null
@@ -1,215 +0,0 @@
-#! /bin/sh
-# Common wrapper for a few potentially missing GNU programs.
-
-scriptversion=2013-10-28.13; # UTC
-
-# Copyright (C) 1996-2014 Free Software Foundation, Inc.
-# Originally written by Fran,cois Pinard <pinard@iro.umontreal.ca>, 1996.
-
-# This program is free software; you can redistribute it and/or modify
-# it under the terms of the GNU General Public License as published by
-# the Free Software Foundation; either version 2, or (at your option)
-# any later version.
-
-# This program is distributed in the hope that it will be useful,
-# but WITHOUT ANY WARRANTY; without even the implied warranty of
-# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-# GNU General Public License for more details.
-
-# You should have received a copy of the GNU General Public License
-# along with this program.  If not, see <http://www.gnu.org/licenses/>.
-
-# As a special exception to the GNU General Public License, if you
-# distribute this file as part of a program that contains a
-# configuration script generated by Autoconf, you may include it under
-# the same distribution terms that you use for the rest of that program.
-
-if test $# -eq 0; then
-  echo 1>&2 "Try '$0 --help' for more information"
-  exit 1
-fi
-
-case $1 in
-
-  --is-lightweight)
-    # Used by our autoconf macros to check whether the available missing
-    # script is modern enough.
-    exit 0
-    ;;
-
-  --run)
-    # Back-compat with the calling convention used by older automake.
-    shift
-    ;;
-
-  -h|--h|--he|--hel|--help)
-    echo "\
-$0 [OPTION]... PROGRAM [ARGUMENT]...
-
-Run 'PROGRAM [ARGUMENT]...', returning a proper advice when this fails due
-to PROGRAM being missing or too old.
-
-Options:
-  -h, --help      display this help and exit
-  -v, --version   output version information and exit
-
-Supported PROGRAM values:
-  aclocal   autoconf  autoheader   autom4te  automake  makeinfo
-  bison     yacc      flex         lex       help2man
-
-Version suffixes to PROGRAM as well as the prefixes 'gnu-', 'gnu', and
-'g' are ignored when checking the name.
-
-Send bug reports to <bug-automake@gnu.org>."
-    exit $?
-    ;;
-
-  -v|--v|--ve|--ver|--vers|--versi|--versio|--version)
-    echo "missing $scriptversion (GNU Automake)"
-    exit $?
-    ;;
-
-  -*)
-    echo 1>&2 "$0: unknown '$1' option"
-    echo 1>&2 "Try '$0 --help' for more information"
-    exit 1
-    ;;
-
-esac
-
-# Run the given program, remember its exit status.
-"$@"; st=$?
-
-# If it succeeded, we are done.
-test $st -eq 0 && exit 0
-
-# Also exit now if we it failed (or wasn't found), and '--version' was
-# passed; such an option is passed most likely to detect whether the
-# program is present and works.
-case $2 in --version|--help) exit $st;; esac
-
-# Exit code 63 means version mismatch.  This often happens when the user
-# tries to use an ancient version of a tool on a file that requires a
-# minimum version.
-if test $st -eq 63; then
-  msg="probably too old"
-elif test $st -eq 127; then
-  # Program was missing.
-  msg="missing on your system"
-else
-  # Program was found and executed, but failed.  Give up.
-  exit $st
-fi
-
-perl_URL=http://www.perl.org/
-flex_URL=http://flex.sourceforge.net/
-gnu_software_URL=http://www.gnu.org/software
-
-program_details ()
-{
-  case $1 in
-    aclocal|automake)
-      echo "The '$1' program is part of the GNU Automake package:"
-      echo "<$gnu_software_URL/automake>"
-      echo "It also requires GNU Autoconf, GNU m4 and Perl in order to run:"
-      echo "<$gnu_software_URL/autoconf>"
-      echo "<$gnu_software_URL/m4/>"
-      echo "<$perl_URL>"
-      ;;
-    autoconf|autom4te|autoheader)
-      echo "The '$1' program is part of the GNU Autoconf package:"
-      echo "<$gnu_software_URL/autoconf/>"
-      echo "It also requires GNU m4 and Perl in order to run:"
-      echo "<$gnu_software_URL/m4/>"
-      echo "<$perl_URL>"
-      ;;
-  esac
-}
-
-give_advice ()
-{
-  # Normalize program name to check for.
-  normalized_program=`echo "$1" | sed '
-    s/^gnu-//; t
-    s/^gnu//; t
-    s/^g//; t'`
-
-  printf '%s\n' "'$1' is $msg."
-
-  configure_deps="'configure.ac' or m4 files included by 'configure.ac'"
-  case $normalized_program in
-    autoconf*)
-      echo "You should only need it if you modified 'configure.ac',"
-      echo "or m4 files included by it."
-      program_details 'autoconf'
-      ;;
-    autoheader*)
-      echo "You should only need it if you modified 'acconfig.h' or"
-      echo "$configure_deps."
-      program_details 'autoheader'
-      ;;
-    automake*)
-      echo "You should only need it if you modified 'Makefile.am' or"
-      echo "$configure_deps."
-      program_details 'automake'
-      ;;
-    aclocal*)
-      echo "You should only need it if you modified 'acinclude.m4' or"
-      echo "$configure_deps."
-      program_details 'aclocal'
-      ;;
-   autom4te*)
-      echo "You might have modified some maintainer files that require"
-      echo "the 'autom4te' program to be rebuilt."
-      program_details 'autom4te'
-      ;;
-    bison*|yacc*)
-      echo "You should only need it if you modified a '.y' file."
-      echo "You may want to install the GNU Bison package:"
-      echo "<$gnu_software_URL/bison/>"
-      ;;
-    lex*|flex*)
-      echo "You should only need it if you modified a '.l' file."
-      echo "You may want to install the Fast Lexical Analyzer package:"
-      echo "<$flex_URL>"
-      ;;
-    help2man*)
-      echo "You should only need it if you modified a dependency" \
-           "of a man page."
-      echo "You may want to install the GNU Help2man package:"
-      echo "<$gnu_software_URL/help2man/>"
-    ;;
-    makeinfo*)
-      echo "You should only need it if you modified a '.texi' file, or"
-      echo "any other file indirectly affecting the aspect of the manual."
-      echo "You might want to install the Texinfo package:"
-      echo "<$gnu_software_URL/texinfo/>"
-      echo "The spurious makeinfo call might also be the consequence of"
-      echo "using a buggy 'make' (AIX, DU, IRIX), in which case you might"
-      echo "want to install GNU make:"
-      echo "<$gnu_software_URL/make/>"
-      ;;
-    *)
-      echo "You might have modified some files without having the proper"
-      echo "tools for further handling them.  Check the 'README' file, it"
-      echo "often tells you about the needed prerequisites for installing"
-      echo "this package.  You may also peek at any GNU archive site, in"
-      echo "case some other package contains this missing '$1' program."
-      ;;
-  esac
-}
-
-give_advice "$1" | sed -e '1s/^/WARNING: /' \
-                       -e '2,$s/^/         /' >&2
-
-# Propagate the correct exit status (expected to be 127 for a program
-# not found, 63 for a program that failed due to version mismatch).
-exit $st
-
-# Local variables:
-# eval: (add-hook 'write-file-hooks 'time-stamp)
-# time-stamp-start: "scriptversion="
-# time-stamp-format: "%:y-%02m-%02d.%02H"
-# time-stamp-time-zone: "UTC"
-# time-stamp-end: "; # UTC"
-# End:
diff --git a/missing b/missing
new file mode 120000
index 00000000..20bc5b0e
--- /dev/null
+++ b/missing
@@ -0,0 +1 @@
+/usr/share/automake-1.11/missing
\ No newline at end of file
diff --git a/reconfigure_script b/reconfigure_script
index 6bdb081e..d8d7212d 100755
--- a/reconfigure_script
+++ b/reconfigure_script
@@ -1,4 +1,4 @@
 aclocal -I m4
 autoheader -f
-automake -f 
+automake -f --add-missing
 autoconf -f

From 4e3f4104ab4f31aaa28a7a5896e12341e48d12e1 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Wed, 27 May 2015 11:10:51 +0900
Subject: [PATCH 242/429] Added a .gitignore file to eliminate autoconf files
 from commits

---
 .gitignore           |    2 +
 INSTALL              |    2 +-
 Makefile.in          |  694 ------
 aclocal.m4           |  980 --------
 compile              |  347 ---
 config.guess         |    1 -
 config.sub           |    1 -
 configure            | 5168 ++++++++++++++++--------------------------
 depcomp              |    1 -
 install-sh           |    1 -
 lib/Grid_config.h    |    7 +-
 lib/Grid_config.h.in |    3 +
 missing              |    1 -
 13 files changed, 1976 insertions(+), 5232 deletions(-)
 delete mode 100644 Makefile.in
 delete mode 100644 aclocal.m4
 delete mode 100755 compile
 delete mode 120000 config.guess
 delete mode 120000 config.sub
 delete mode 120000 depcomp
 delete mode 120000 install-sh
 delete mode 120000 missing

diff --git a/.gitignore b/.gitignore
index ac77b76b..09504991 100644
--- a/.gitignore
+++ b/.gitignore
@@ -46,3 +46,5 @@ config.status
 /install-sh
 /missing
 /stamp-h1
+/config.sub
+/config.guess
diff --git a/INSTALL b/INSTALL
index cbd1c80d..f812f5a2 120000
--- a/INSTALL
+++ b/INSTALL
@@ -1 +1 @@
-/usr/share/automake-1.11/INSTALL
\ No newline at end of file
+/usr/share/automake-1.14/INSTALL
\ No newline at end of file
diff --git a/Makefile.in b/Makefile.in
deleted file mode 100644
index cfa68e94..00000000
--- a/Makefile.in
+++ /dev/null
@@ -1,694 +0,0 @@
-# Makefile.in generated by automake 1.11.1 from Makefile.am.
-# @configure_input@
-
-# Copyright (C) 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002,
-# 2003, 2004, 2005, 2006, 2007, 2008, 2009  Free Software Foundation,
-# Inc.
-# This Makefile.in is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# This program is distributed in the hope that it will be useful,
-# but WITHOUT ANY WARRANTY, to the extent permitted by law; without
-# even the implied warranty of MERCHANTABILITY or FITNESS FOR A
-# PARTICULAR PURPOSE.
-
-@SET_MAKE@
-VPATH = @srcdir@
-pkgdatadir = $(datadir)/@PACKAGE@
-pkgincludedir = $(includedir)/@PACKAGE@
-pkglibdir = $(libdir)/@PACKAGE@
-pkglibexecdir = $(libexecdir)/@PACKAGE@
-am__cd = CDPATH="$${ZSH_VERSION+.}$(PATH_SEPARATOR)" && cd
-install_sh_DATA = $(install_sh) -c -m 644
-install_sh_PROGRAM = $(install_sh) -c
-install_sh_SCRIPT = $(install_sh) -c
-INSTALL_HEADER = $(INSTALL_DATA)
-transform = $(program_transform_name)
-NORMAL_INSTALL = :
-PRE_INSTALL = :
-POST_INSTALL = :
-NORMAL_UNINSTALL = :
-PRE_UNINSTALL = :
-POST_UNINSTALL = :
-build_triplet = @build@
-host_triplet = @host@
-target_triplet = @target@
-subdir = .
-DIST_COMMON = README $(am__configure_deps) $(srcdir)/Makefile.am \
-	$(srcdir)/Makefile.in $(top_srcdir)/configure \
-	$(top_srcdir)/docs/doxy.cfg.in AUTHORS COPYING ChangeLog \
-	INSTALL NEWS TODO compile config.guess config.sub depcomp \
-	install-sh missing
-ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
-am__aclocal_m4_deps = $(top_srcdir)/m4/ac_prog_doxygen.m4 \
-	$(top_srcdir)/m4/ax_cxx_compile_stdcxx_11.m4 \
-	$(top_srcdir)/configure.ac
-am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
-	$(ACLOCAL_M4)
-am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \
- configure.lineno config.status.lineno
-mkinstalldirs = $(install_sh) -d
-CONFIG_HEADER = $(top_builddir)/lib/Grid_config.h
-CONFIG_CLEAN_FILES = docs/doxy.cfg
-CONFIG_CLEAN_VPATH_FILES =
-AM_V_GEN = $(am__v_GEN_$(V))
-am__v_GEN_ = $(am__v_GEN_$(AM_DEFAULT_VERBOSITY))
-am__v_GEN_0 = @echo "  GEN   " $@;
-AM_V_at = $(am__v_at_$(V))
-am__v_at_ = $(am__v_at_$(AM_DEFAULT_VERBOSITY))
-am__v_at_0 = @
-SOURCES =
-DIST_SOURCES =
-RECURSIVE_TARGETS = all-recursive check-recursive dvi-recursive \
-	html-recursive info-recursive install-data-recursive \
-	install-dvi-recursive install-exec-recursive \
-	install-html-recursive install-info-recursive \
-	install-pdf-recursive install-ps-recursive install-recursive \
-	installcheck-recursive installdirs-recursive pdf-recursive \
-	ps-recursive uninstall-recursive
-RECURSIVE_CLEAN_TARGETS = mostlyclean-recursive clean-recursive	\
-  distclean-recursive maintainer-clean-recursive
-AM_RECURSIVE_TARGETS = $(RECURSIVE_TARGETS:-recursive=) \
-	$(RECURSIVE_CLEAN_TARGETS:-recursive=) tags TAGS ctags CTAGS \
-	distdir dist dist-all distcheck
-ETAGS = etags
-CTAGS = ctags
-DIST_SUBDIRS = $(SUBDIRS)
-DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
-distdir = $(PACKAGE)-$(VERSION)
-top_distdir = $(distdir)
-am__remove_distdir = \
-  { test ! -d "$(distdir)" \
-    || { find "$(distdir)" -type d ! -perm -200 -exec chmod u+w {} ';' \
-         && rm -fr "$(distdir)"; }; }
-am__relativize = \
-  dir0=`pwd`; \
-  sed_first='s,^\([^/]*\)/.*$$,\1,'; \
-  sed_rest='s,^[^/]*/*,,'; \
-  sed_last='s,^.*/\([^/]*\)$$,\1,'; \
-  sed_butlast='s,/*[^/]*$$,,'; \
-  while test -n "$$dir1"; do \
-    first=`echo "$$dir1" | sed -e "$$sed_first"`; \
-    if test "$$first" != "."; then \
-      if test "$$first" = ".."; then \
-        dir2=`echo "$$dir0" | sed -e "$$sed_last"`/"$$dir2"; \
-        dir0=`echo "$$dir0" | sed -e "$$sed_butlast"`; \
-      else \
-        first2=`echo "$$dir2" | sed -e "$$sed_first"`; \
-        if test "$$first2" = "$$first"; then \
-          dir2=`echo "$$dir2" | sed -e "$$sed_rest"`; \
-        else \
-          dir2="../$$dir2"; \
-        fi; \
-        dir0="$$dir0"/"$$first"; \
-      fi; \
-    fi; \
-    dir1=`echo "$$dir1" | sed -e "$$sed_rest"`; \
-  done; \
-  reldir="$$dir2"
-DIST_ARCHIVES = $(distdir).tar.gz
-GZIP_ENV = --best
-distuninstallcheck_listfiles = find . -type f -print
-distcleancheck_listfiles = find . -type f -print
-ACLOCAL = @ACLOCAL@
-AMTAR = @AMTAR@
-AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
-AUTOCONF = @AUTOCONF@
-AUTOHEADER = @AUTOHEADER@
-AUTOMAKE = @AUTOMAKE@
-AWK = @AWK@
-CPPFLAGS = @CPPFLAGS@
-CXX = @CXX@
-CXXCPP = @CXXCPP@
-CXXDEPMODE = @CXXDEPMODE@
-CXXFLAGS = @CXXFLAGS@
-CYGPATH_W = @CYGPATH_W@
-DEFS = @DEFS@
-DEPDIR = @DEPDIR@
-DOT = @DOT@
-DOXYGEN = @DOXYGEN@
-ECHO_C = @ECHO_C@
-ECHO_N = @ECHO_N@
-ECHO_T = @ECHO_T@
-EGREP = @EGREP@
-EXEEXT = @EXEEXT@
-GREP = @GREP@
-HAVE_CXX11 = @HAVE_CXX11@
-INSTALL = @INSTALL@
-INSTALL_DATA = @INSTALL_DATA@
-INSTALL_PROGRAM = @INSTALL_PROGRAM@
-INSTALL_SCRIPT = @INSTALL_SCRIPT@
-INSTALL_STRIP_PROGRAM = @INSTALL_STRIP_PROGRAM@
-LDFLAGS = @LDFLAGS@
-LIBOBJS = @LIBOBJS@
-LIBS = @LIBS@
-LTLIBOBJS = @LTLIBOBJS@
-MAKEINFO = @MAKEINFO@
-MKDIR_P = @MKDIR_P@
-OBJEXT = @OBJEXT@
-OPENMP_CXXFLAGS = @OPENMP_CXXFLAGS@
-PACKAGE = @PACKAGE@
-PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
-PACKAGE_NAME = @PACKAGE_NAME@
-PACKAGE_STRING = @PACKAGE_STRING@
-PACKAGE_TARNAME = @PACKAGE_TARNAME@
-PACKAGE_VERSION = @PACKAGE_VERSION@
-PATH_SEPARATOR = @PATH_SEPARATOR@
-RANLIB = @RANLIB@
-SET_MAKE = @SET_MAKE@
-SHELL = @SHELL@
-STRIP = @STRIP@
-VERSION = @VERSION@
-abs_builddir = @abs_builddir@
-abs_srcdir = @abs_srcdir@
-abs_top_builddir = @abs_top_builddir@
-abs_top_srcdir = @abs_top_srcdir@
-ac_ct_CXX = @ac_ct_CXX@
-am__include = @am__include@
-am__leading_dot = @am__leading_dot@
-am__quote = @am__quote@
-am__tar = @am__tar@
-am__untar = @am__untar@
-bindir = @bindir@
-build = @build@
-build_alias = @build_alias@
-build_cpu = @build_cpu@
-build_os = @build_os@
-build_vendor = @build_vendor@
-builddir = @builddir@
-datadir = @datadir@
-datarootdir = @datarootdir@
-docdir = @docdir@
-dvidir = @dvidir@
-enable_dot = @enable_dot@
-enable_html_docs = @enable_html_docs@
-enable_latex_docs = @enable_latex_docs@
-exec_prefix = @exec_prefix@
-host = @host@
-host_alias = @host_alias@
-host_cpu = @host_cpu@
-host_os = @host_os@
-host_vendor = @host_vendor@
-htmldir = @htmldir@
-includedir = @includedir@
-infodir = @infodir@
-install_sh = @install_sh@
-libdir = @libdir@
-libexecdir = @libexecdir@
-localedir = @localedir@
-localstatedir = @localstatedir@
-mandir = @mandir@
-mkdir_p = @mkdir_p@
-oldincludedir = @oldincludedir@
-pdfdir = @pdfdir@
-prefix = @prefix@
-program_transform_name = @program_transform_name@
-psdir = @psdir@
-sbindir = @sbindir@
-sharedstatedir = @sharedstatedir@
-srcdir = @srcdir@
-sysconfdir = @sysconfdir@
-target = @target@
-target_alias = @target_alias@
-target_cpu = @target_cpu@
-target_os = @target_os@
-target_vendor = @target_vendor@
-top_build_prefix = @top_build_prefix@
-top_builddir = @top_builddir@
-top_srcdir = @top_srcdir@
-
-# additional include paths necessary to compile the C++ library
-AM_CXXFLAGS = -I$(top_srcdir)/
-SUBDIRS = lib tests benchmarks
-all: all-recursive
-
-.SUFFIXES:
-am--refresh:
-	@:
-$(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
-	@for dep in $?; do \
-	  case '$(am__configure_deps)' in \
-	    *$$dep*) \
-	      echo ' cd $(srcdir) && $(AUTOMAKE) --gnu'; \
-	      $(am__cd) $(srcdir) && $(AUTOMAKE) --gnu \
-		&& exit 0; \
-	      exit 1;; \
-	  esac; \
-	done; \
-	echo ' cd $(top_srcdir) && $(AUTOMAKE) --gnu Makefile'; \
-	$(am__cd) $(top_srcdir) && \
-	  $(AUTOMAKE) --gnu Makefile
-.PRECIOUS: Makefile
-Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
-	@case '$?' in \
-	  *config.status*) \
-	    echo ' $(SHELL) ./config.status'; \
-	    $(SHELL) ./config.status;; \
-	  *) \
-	    echo ' cd $(top_builddir) && $(SHELL) ./config.status $@ $(am__depfiles_maybe)'; \
-	    cd $(top_builddir) && $(SHELL) ./config.status $@ $(am__depfiles_maybe);; \
-	esac;
-
-$(top_builddir)/config.status: $(top_srcdir)/configure $(CONFIG_STATUS_DEPENDENCIES)
-	$(SHELL) ./config.status --recheck
-
-$(top_srcdir)/configure:  $(am__configure_deps)
-	$(am__cd) $(srcdir) && $(AUTOCONF)
-$(ACLOCAL_M4):  $(am__aclocal_m4_deps)
-	$(am__cd) $(srcdir) && $(ACLOCAL) $(ACLOCAL_AMFLAGS)
-$(am__aclocal_m4_deps):
-docs/doxy.cfg: $(top_builddir)/config.status $(top_srcdir)/docs/doxy.cfg.in
-	cd $(top_builddir) && $(SHELL) ./config.status $@
-
-# This directory's subdirectories are mostly independent; you can cd
-# into them and run `make' without going through this Makefile.
-# To change the values of `make' variables: instead of editing Makefiles,
-# (1) if the variable is set in `config.status', edit `config.status'
-#     (which will cause the Makefiles to be regenerated when you run `make');
-# (2) otherwise, pass the desired values on the `make' command line.
-$(RECURSIVE_TARGETS):
-	@fail= failcom='exit 1'; \
-	for f in x $$MAKEFLAGS; do \
-	  case $$f in \
-	    *=* | --[!k]*);; \
-	    *k*) failcom='fail=yes';; \
-	  esac; \
-	done; \
-	dot_seen=no; \
-	target=`echo $@ | sed s/-recursive//`; \
-	list='$(SUBDIRS)'; for subdir in $$list; do \
-	  echo "Making $$target in $$subdir"; \
-	  if test "$$subdir" = "."; then \
-	    dot_seen=yes; \
-	    local_target="$$target-am"; \
-	  else \
-	    local_target="$$target"; \
-	  fi; \
-	  ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) $$local_target) \
-	  || eval $$failcom; \
-	done; \
-	if test "$$dot_seen" = "no"; then \
-	  $(MAKE) $(AM_MAKEFLAGS) "$$target-am" || exit 1; \
-	fi; test -z "$$fail"
-
-$(RECURSIVE_CLEAN_TARGETS):
-	@fail= failcom='exit 1'; \
-	for f in x $$MAKEFLAGS; do \
-	  case $$f in \
-	    *=* | --[!k]*);; \
-	    *k*) failcom='fail=yes';; \
-	  esac; \
-	done; \
-	dot_seen=no; \
-	case "$@" in \
-	  distclean-* | maintainer-clean-*) list='$(DIST_SUBDIRS)' ;; \
-	  *) list='$(SUBDIRS)' ;; \
-	esac; \
-	rev=''; for subdir in $$list; do \
-	  if test "$$subdir" = "."; then :; else \
-	    rev="$$subdir $$rev"; \
-	  fi; \
-	done; \
-	rev="$$rev ."; \
-	target=`echo $@ | sed s/-recursive//`; \
-	for subdir in $$rev; do \
-	  echo "Making $$target in $$subdir"; \
-	  if test "$$subdir" = "."; then \
-	    local_target="$$target-am"; \
-	  else \
-	    local_target="$$target"; \
-	  fi; \
-	  ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) $$local_target) \
-	  || eval $$failcom; \
-	done && test -z "$$fail"
-tags-recursive:
-	list='$(SUBDIRS)'; for subdir in $$list; do \
-	  test "$$subdir" = . || ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) tags); \
-	done
-ctags-recursive:
-	list='$(SUBDIRS)'; for subdir in $$list; do \
-	  test "$$subdir" = . || ($(am__cd) $$subdir && $(MAKE) $(AM_MAKEFLAGS) ctags); \
-	done
-
-ID: $(HEADERS) $(SOURCES) $(LISP) $(TAGS_FILES)
-	list='$(SOURCES) $(HEADERS) $(LISP) $(TAGS_FILES)'; \
-	unique=`for i in $$list; do \
-	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
-	  done | \
-	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
-	      END { if (nonempty) { for (i in files) print i; }; }'`; \
-	mkid -fID $$unique
-tags: TAGS
-
-TAGS: tags-recursive $(HEADERS) $(SOURCES)  $(TAGS_DEPENDENCIES) \
-		$(TAGS_FILES) $(LISP)
-	set x; \
-	here=`pwd`; \
-	if ($(ETAGS) --etags-include --version) >/dev/null 2>&1; then \
-	  include_option=--etags-include; \
-	  empty_fix=.; \
-	else \
-	  include_option=--include; \
-	  empty_fix=; \
-	fi; \
-	list='$(SUBDIRS)'; for subdir in $$list; do \
-	  if test "$$subdir" = .; then :; else \
-	    test ! -f $$subdir/TAGS || \
-	      set "$$@" "$$include_option=$$here/$$subdir/TAGS"; \
-	  fi; \
-	done; \
-	list='$(SOURCES) $(HEADERS)  $(LISP) $(TAGS_FILES)'; \
-	unique=`for i in $$list; do \
-	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
-	  done | \
-	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
-	      END { if (nonempty) { for (i in files) print i; }; }'`; \
-	shift; \
-	if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
-	  test -n "$$unique" || unique=$$empty_fix; \
-	  if test $$# -gt 0; then \
-	    $(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
-	      "$$@" $$unique; \
-	  else \
-	    $(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
-	      $$unique; \
-	  fi; \
-	fi
-ctags: CTAGS
-CTAGS: ctags-recursive $(HEADERS) $(SOURCES)  $(TAGS_DEPENDENCIES) \
-		$(TAGS_FILES) $(LISP)
-	list='$(SOURCES) $(HEADERS)  $(LISP) $(TAGS_FILES)'; \
-	unique=`for i in $$list; do \
-	    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
-	  done | \
-	  $(AWK) '{ files[$$0] = 1; nonempty = 1; } \
-	      END { if (nonempty) { for (i in files) print i; }; }'`; \
-	test -z "$(CTAGS_ARGS)$$unique" \
-	  || $(CTAGS) $(CTAGSFLAGS) $(AM_CTAGSFLAGS) $(CTAGS_ARGS) \
-	     $$unique
-
-GTAGS:
-	here=`$(am__cd) $(top_builddir) && pwd` \
-	  && $(am__cd) $(top_srcdir) \
-	  && gtags -i $(GTAGS_ARGS) "$$here"
-
-distclean-tags:
-	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
-
-distdir: $(DISTFILES)
-	$(am__remove_distdir)
-	test -d "$(distdir)" || mkdir "$(distdir)"
-	@srcdirstrip=`echo "$(srcdir)" | sed 's/[].[^$$\\*]/\\\\&/g'`; \
-	topsrcdirstrip=`echo "$(top_srcdir)" | sed 's/[].[^$$\\*]/\\\\&/g'`; \
-	list='$(DISTFILES)'; \
-	  dist_files=`for file in $$list; do echo $$file; done | \
-	  sed -e "s|^$$srcdirstrip/||;t" \
-	      -e "s|^$$topsrcdirstrip/|$(top_builddir)/|;t"`; \
-	case $$dist_files in \
-	  */*) $(MKDIR_P) `echo "$$dist_files" | \
-			   sed '/\//!d;s|^|$(distdir)/|;s,/[^/]*$$,,' | \
-			   sort -u` ;; \
-	esac; \
-	for file in $$dist_files; do \
-	  if test -f $$file || test -d $$file; then d=.; else d=$(srcdir); fi; \
-	  if test -d $$d/$$file; then \
-	    dir=`echo "/$$file" | sed -e 's,/[^/]*$$,,'`; \
-	    if test -d "$(distdir)/$$file"; then \
-	      find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
-	    fi; \
-	    if test -d $(srcdir)/$$file && test $$d != $(srcdir); then \
-	      cp -fpR $(srcdir)/$$file "$(distdir)$$dir" || exit 1; \
-	      find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
-	    fi; \
-	    cp -fpR $$d/$$file "$(distdir)$$dir" || exit 1; \
-	  else \
-	    test -f "$(distdir)/$$file" \
-	    || cp -p $$d/$$file "$(distdir)/$$file" \
-	    || exit 1; \
-	  fi; \
-	done
-	@list='$(DIST_SUBDIRS)'; for subdir in $$list; do \
-	  if test "$$subdir" = .; then :; else \
-	    test -d "$(distdir)/$$subdir" \
-	    || $(MKDIR_P) "$(distdir)/$$subdir" \
-	    || exit 1; \
-	  fi; \
-	done
-	@list='$(DIST_SUBDIRS)'; for subdir in $$list; do \
-	  if test "$$subdir" = .; then :; else \
-	    dir1=$$subdir; dir2="$(distdir)/$$subdir"; \
-	    $(am__relativize); \
-	    new_distdir=$$reldir; \
-	    dir1=$$subdir; dir2="$(top_distdir)"; \
-	    $(am__relativize); \
-	    new_top_distdir=$$reldir; \
-	    echo " (cd $$subdir && $(MAKE) $(AM_MAKEFLAGS) top_distdir="$$new_top_distdir" distdir="$$new_distdir" \\"; \
-	    echo "     am__remove_distdir=: am__skip_length_check=: am__skip_mode_fix=: distdir)"; \
-	    ($(am__cd) $$subdir && \
-	      $(MAKE) $(AM_MAKEFLAGS) \
-	        top_distdir="$$new_top_distdir" \
-	        distdir="$$new_distdir" \
-		am__remove_distdir=: \
-		am__skip_length_check=: \
-		am__skip_mode_fix=: \
-	        distdir) \
-	      || exit 1; \
-	  fi; \
-	done
-	-test -n "$(am__skip_mode_fix)" \
-	|| find "$(distdir)" -type d ! -perm -755 \
-		-exec chmod u+rwx,go+rx {} \; -o \
-	  ! -type d ! -perm -444 -links 1 -exec chmod a+r {} \; -o \
-	  ! -type d ! -perm -400 -exec chmod a+r {} \; -o \
-	  ! -type d ! -perm -444 -exec $(install_sh) -c -m a+r {} {} \; \
-	|| chmod -R a+r "$(distdir)"
-dist-gzip: distdir
-	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
-	$(am__remove_distdir)
-
-dist-bzip2: distdir
-	tardir=$(distdir) && $(am__tar) | bzip2 -9 -c >$(distdir).tar.bz2
-	$(am__remove_distdir)
-
-dist-lzma: distdir
-	tardir=$(distdir) && $(am__tar) | lzma -9 -c >$(distdir).tar.lzma
-	$(am__remove_distdir)
-
-dist-xz: distdir
-	tardir=$(distdir) && $(am__tar) | xz -c >$(distdir).tar.xz
-	$(am__remove_distdir)
-
-dist-tarZ: distdir
-	tardir=$(distdir) && $(am__tar) | compress -c >$(distdir).tar.Z
-	$(am__remove_distdir)
-
-dist-shar: distdir
-	shar $(distdir) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).shar.gz
-	$(am__remove_distdir)
-
-dist-zip: distdir
-	-rm -f $(distdir).zip
-	zip -rq $(distdir).zip $(distdir)
-	$(am__remove_distdir)
-
-dist dist-all: distdir
-	tardir=$(distdir) && $(am__tar) | GZIP=$(GZIP_ENV) gzip -c >$(distdir).tar.gz
-	$(am__remove_distdir)
-
-# This target untars the dist file and tries a VPATH configuration.  Then
-# it guarantees that the distribution is self-contained by making another
-# tarfile.
-distcheck: dist
-	case '$(DIST_ARCHIVES)' in \
-	*.tar.gz*) \
-	  GZIP=$(GZIP_ENV) gzip -dc $(distdir).tar.gz | $(am__untar) ;;\
-	*.tar.bz2*) \
-	  bzip2 -dc $(distdir).tar.bz2 | $(am__untar) ;;\
-	*.tar.lzma*) \
-	  lzma -dc $(distdir).tar.lzma | $(am__untar) ;;\
-	*.tar.xz*) \
-	  xz -dc $(distdir).tar.xz | $(am__untar) ;;\
-	*.tar.Z*) \
-	  uncompress -c $(distdir).tar.Z | $(am__untar) ;;\
-	*.shar.gz*) \
-	  GZIP=$(GZIP_ENV) gzip -dc $(distdir).shar.gz | unshar ;;\
-	*.zip*) \
-	  unzip $(distdir).zip ;;\
-	esac
-	chmod -R a-w $(distdir); chmod u+w $(distdir)
-	mkdir $(distdir)/_build
-	mkdir $(distdir)/_inst
-	chmod a-w $(distdir)
-	test -d $(distdir)/_build || exit 0; \
-	dc_install_base=`$(am__cd) $(distdir)/_inst && pwd | sed -e 's,^[^:\\/]:[\\/],/,'` \
-	  && dc_destdir="$${TMPDIR-/tmp}/am-dc-$$$$/" \
-	  && am__cwd=`pwd` \
-	  && $(am__cd) $(distdir)/_build \
-	  && ../configure --srcdir=.. --prefix="$$dc_install_base" \
-	    $(DISTCHECK_CONFIGURE_FLAGS) \
-	  && $(MAKE) $(AM_MAKEFLAGS) \
-	  && $(MAKE) $(AM_MAKEFLAGS) dvi \
-	  && $(MAKE) $(AM_MAKEFLAGS) check \
-	  && $(MAKE) $(AM_MAKEFLAGS) install \
-	  && $(MAKE) $(AM_MAKEFLAGS) installcheck \
-	  && $(MAKE) $(AM_MAKEFLAGS) uninstall \
-	  && $(MAKE) $(AM_MAKEFLAGS) distuninstallcheck_dir="$$dc_install_base" \
-	        distuninstallcheck \
-	  && chmod -R a-w "$$dc_install_base" \
-	  && ({ \
-	       (cd ../.. && umask 077 && mkdir "$$dc_destdir") \
-	       && $(MAKE) $(AM_MAKEFLAGS) DESTDIR="$$dc_destdir" install \
-	       && $(MAKE) $(AM_MAKEFLAGS) DESTDIR="$$dc_destdir" uninstall \
-	       && $(MAKE) $(AM_MAKEFLAGS) DESTDIR="$$dc_destdir" \
-	            distuninstallcheck_dir="$$dc_destdir" distuninstallcheck; \
-	      } || { rm -rf "$$dc_destdir"; exit 1; }) \
-	  && rm -rf "$$dc_destdir" \
-	  && $(MAKE) $(AM_MAKEFLAGS) dist \
-	  && rm -rf $(DIST_ARCHIVES) \
-	  && $(MAKE) $(AM_MAKEFLAGS) distcleancheck \
-	  && cd "$$am__cwd" \
-	  || exit 1
-	$(am__remove_distdir)
-	@(echo "$(distdir) archives ready for distribution: "; \
-	  list='$(DIST_ARCHIVES)'; for i in $$list; do echo $$i; done) | \
-	  sed -e 1h -e 1s/./=/g -e 1p -e 1x -e '$$p' -e '$$x'
-distuninstallcheck:
-	@$(am__cd) '$(distuninstallcheck_dir)' \
-	&& test `$(distuninstallcheck_listfiles) | wc -l` -le 1 \
-	   || { echo "ERROR: files left after uninstall:" ; \
-	        if test -n "$(DESTDIR)"; then \
-	          echo "  (check DESTDIR support)"; \
-	        fi ; \
-	        $(distuninstallcheck_listfiles) ; \
-	        exit 1; } >&2
-distcleancheck: distclean
-	@if test '$(srcdir)' = . ; then \
-	  echo "ERROR: distcleancheck can only run from a VPATH build" ; \
-	  exit 1 ; \
-	fi
-	@test `$(distcleancheck_listfiles) | wc -l` -eq 0 \
-	  || { echo "ERROR: files left in build directory after distclean:" ; \
-	       $(distcleancheck_listfiles) ; \
-	       exit 1; } >&2
-check-am: all-am
-check: check-recursive
-all-am: Makefile
-installdirs: installdirs-recursive
-installdirs-am:
-install: install-recursive
-install-exec: install-exec-recursive
-install-data: install-data-recursive
-uninstall: uninstall-recursive
-
-install-am: all-am
-	@$(MAKE) $(AM_MAKEFLAGS) install-exec-am install-data-am
-
-installcheck: installcheck-recursive
-install-strip:
-	$(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
-	  install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
-	  `test -z '$(STRIP)' || \
-	    echo "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'"` install
-mostlyclean-generic:
-
-clean-generic:
-
-distclean-generic:
-	-test -z "$(CONFIG_CLEAN_FILES)" || rm -f $(CONFIG_CLEAN_FILES)
-	-test . = "$(srcdir)" || test -z "$(CONFIG_CLEAN_VPATH_FILES)" || rm -f $(CONFIG_CLEAN_VPATH_FILES)
-
-maintainer-clean-generic:
-	@echo "This command is intended for maintainers to use"
-	@echo "it deletes files that may require special tools to rebuild."
-clean: clean-recursive
-
-clean-am: clean-generic mostlyclean-am
-
-distclean: distclean-recursive
-	-rm -f $(am__CONFIG_DISTCLEAN_FILES)
-	-rm -f Makefile
-distclean-am: clean-am distclean-generic distclean-tags
-
-dvi: dvi-recursive
-
-dvi-am:
-
-html: html-recursive
-
-html-am:
-
-info: info-recursive
-
-info-am:
-
-install-data-am:
-
-install-dvi: install-dvi-recursive
-
-install-dvi-am:
-
-install-exec-am:
-
-install-html: install-html-recursive
-
-install-html-am:
-
-install-info: install-info-recursive
-
-install-info-am:
-
-install-man:
-
-install-pdf: install-pdf-recursive
-
-install-pdf-am:
-
-install-ps: install-ps-recursive
-
-install-ps-am:
-
-installcheck-am:
-
-maintainer-clean: maintainer-clean-recursive
-	-rm -f $(am__CONFIG_DISTCLEAN_FILES)
-	-rm -rf $(top_srcdir)/autom4te.cache
-	-rm -f Makefile
-maintainer-clean-am: distclean-am maintainer-clean-generic
-
-mostlyclean: mostlyclean-recursive
-
-mostlyclean-am: mostlyclean-generic
-
-pdf: pdf-recursive
-
-pdf-am:
-
-ps: ps-recursive
-
-ps-am:
-
-uninstall-am:
-
-.MAKE: $(RECURSIVE_CLEAN_TARGETS) $(RECURSIVE_TARGETS) ctags-recursive \
-	install-am install-strip tags-recursive
-
-.PHONY: $(RECURSIVE_CLEAN_TARGETS) $(RECURSIVE_TARGETS) CTAGS GTAGS \
-	all all-am am--refresh check check-am clean clean-generic \
-	ctags ctags-recursive dist dist-all dist-bzip2 dist-gzip \
-	dist-lzma dist-shar dist-tarZ dist-xz dist-zip distcheck \
-	distclean distclean-generic distclean-tags distcleancheck \
-	distdir distuninstallcheck dvi dvi-am html html-am info \
-	info-am install install-am install-data install-data-am \
-	install-dvi install-dvi-am install-exec install-exec-am \
-	install-html install-html-am install-info install-info-am \
-	install-man install-pdf install-pdf-am install-ps \
-	install-ps-am install-strip installcheck installcheck-am \
-	installdirs installdirs-am maintainer-clean \
-	maintainer-clean-generic mostlyclean mostlyclean-generic pdf \
-	pdf-am ps ps-am tags tags-recursive uninstall uninstall-am
-
-
-# Tell versions [3.59,3.63) of GNU make to not export all variables.
-# Otherwise a system limit (for SysV at least) may be exceeded.
-.NOEXPORT:
diff --git a/aclocal.m4 b/aclocal.m4
deleted file mode 100644
index e517c84c..00000000
--- a/aclocal.m4
+++ /dev/null
@@ -1,980 +0,0 @@
-# generated automatically by aclocal 1.11.1 -*- Autoconf -*-
-
-# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
-# 2005, 2006, 2007, 2008, 2009  Free Software Foundation, Inc.
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# This program is distributed in the hope that it will be useful,
-# but WITHOUT ANY WARRANTY, to the extent permitted by law; without
-# even the implied warranty of MERCHANTABILITY or FITNESS FOR A
-# PARTICULAR PURPOSE.
-
-m4_ifndef([AC_AUTOCONF_VERSION],
-  [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
-m4_if(m4_defn([AC_AUTOCONF_VERSION]), [2.63],,
-[m4_warning([this file was generated for autoconf 2.63.
-You have another version of autoconf.  It may work, but is not guaranteed to.
-If you have problems, you may need to regenerate the build system entirely.
-To do so, use the procedure documented by the package, typically `autoreconf'.])])
-
-# Copyright (C) 2002, 2003, 2005, 2006, 2007, 2008  Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# AM_AUTOMAKE_VERSION(VERSION)
-# ----------------------------
-# Automake X.Y traces this macro to ensure aclocal.m4 has been
-# generated from the m4 files accompanying Automake X.Y.
-# (This private macro should not be called outside this file.)
-AC_DEFUN([AM_AUTOMAKE_VERSION],
-[am__api_version='1.11'
-dnl Some users find AM_AUTOMAKE_VERSION and mistake it for a way to
-dnl require some minimum version.  Point them to the right macro.
-m4_if([$1], [1.11.1], [],
-      [AC_FATAL([Do not call $0, use AM_INIT_AUTOMAKE([$1]).])])dnl
-])
-
-# _AM_AUTOCONF_VERSION(VERSION)
-# -----------------------------
-# aclocal traces this macro to find the Autoconf version.
-# This is a private macro too.  Using m4_define simplifies
-# the logic in aclocal, which can simply ignore this definition.
-m4_define([_AM_AUTOCONF_VERSION], [])
-
-# AM_SET_CURRENT_AUTOMAKE_VERSION
-# -------------------------------
-# Call AM_AUTOMAKE_VERSION and AM_AUTOMAKE_VERSION so they can be traced.
-# This function is AC_REQUIREd by AM_INIT_AUTOMAKE.
-AC_DEFUN([AM_SET_CURRENT_AUTOMAKE_VERSION],
-[AM_AUTOMAKE_VERSION([1.11.1])dnl
-m4_ifndef([AC_AUTOCONF_VERSION],
-  [m4_copy([m4_PACKAGE_VERSION], [AC_AUTOCONF_VERSION])])dnl
-_AM_AUTOCONF_VERSION(m4_defn([AC_AUTOCONF_VERSION]))])
-
-# AM_AUX_DIR_EXPAND                                         -*- Autoconf -*-
-
-# Copyright (C) 2001, 2003, 2005  Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# For projects using AC_CONFIG_AUX_DIR([foo]), Autoconf sets
-# $ac_aux_dir to `$srcdir/foo'.  In other projects, it is set to
-# `$srcdir', `$srcdir/..', or `$srcdir/../..'.
-#
-# Of course, Automake must honor this variable whenever it calls a
-# tool from the auxiliary directory.  The problem is that $srcdir (and
-# therefore $ac_aux_dir as well) can be either absolute or relative,
-# depending on how configure is run.  This is pretty annoying, since
-# it makes $ac_aux_dir quite unusable in subdirectories: in the top
-# source directory, any form will work fine, but in subdirectories a
-# relative path needs to be adjusted first.
-#
-# $ac_aux_dir/missing
-#    fails when called from a subdirectory if $ac_aux_dir is relative
-# $top_srcdir/$ac_aux_dir/missing
-#    fails if $ac_aux_dir is absolute,
-#    fails when called from a subdirectory in a VPATH build with
-#          a relative $ac_aux_dir
-#
-# The reason of the latter failure is that $top_srcdir and $ac_aux_dir
-# are both prefixed by $srcdir.  In an in-source build this is usually
-# harmless because $srcdir is `.', but things will broke when you
-# start a VPATH build or use an absolute $srcdir.
-#
-# So we could use something similar to $top_srcdir/$ac_aux_dir/missing,
-# iff we strip the leading $srcdir from $ac_aux_dir.  That would be:
-#   am_aux_dir='\$(top_srcdir)/'`expr "$ac_aux_dir" : "$srcdir//*\(.*\)"`
-# and then we would define $MISSING as
-#   MISSING="\${SHELL} $am_aux_dir/missing"
-# This will work as long as MISSING is not called from configure, because
-# unfortunately $(top_srcdir) has no meaning in configure.
-# However there are other variables, like CC, which are often used in
-# configure, and could therefore not use this "fixed" $ac_aux_dir.
-#
-# Another solution, used here, is to always expand $ac_aux_dir to an
-# absolute PATH.  The drawback is that using absolute paths prevent a
-# configured tree to be moved without reconfiguration.
-
-AC_DEFUN([AM_AUX_DIR_EXPAND],
-[dnl Rely on autoconf to set up CDPATH properly.
-AC_PREREQ([2.50])dnl
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
-])
-
-# AM_CONDITIONAL                                            -*- Autoconf -*-
-
-# Copyright (C) 1997, 2000, 2001, 2003, 2004, 2005, 2006, 2008
-# Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# serial 9
-
-# AM_CONDITIONAL(NAME, SHELL-CONDITION)
-# -------------------------------------
-# Define a conditional.
-AC_DEFUN([AM_CONDITIONAL],
-[AC_PREREQ(2.52)dnl
- ifelse([$1], [TRUE],  [AC_FATAL([$0: invalid condition: $1])],
-	[$1], [FALSE], [AC_FATAL([$0: invalid condition: $1])])dnl
-AC_SUBST([$1_TRUE])dnl
-AC_SUBST([$1_FALSE])dnl
-_AM_SUBST_NOTMAKE([$1_TRUE])dnl
-_AM_SUBST_NOTMAKE([$1_FALSE])dnl
-m4_define([_AM_COND_VALUE_$1], [$2])dnl
-if $2; then
-  $1_TRUE=
-  $1_FALSE='#'
-else
-  $1_TRUE='#'
-  $1_FALSE=
-fi
-AC_CONFIG_COMMANDS_PRE(
-[if test -z "${$1_TRUE}" && test -z "${$1_FALSE}"; then
-  AC_MSG_ERROR([[conditional "$1" was never defined.
-Usually this means the macro was only invoked conditionally.]])
-fi])])
-
-# Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2009
-# Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# serial 10
-
-# There are a few dirty hacks below to avoid letting `AC_PROG_CC' be
-# written in clear, in which case automake, when reading aclocal.m4,
-# will think it sees a *use*, and therefore will trigger all it's
-# C support machinery.  Also note that it means that autoscan, seeing
-# CC etc. in the Makefile, will ask for an AC_PROG_CC use...
-
-
-# _AM_DEPENDENCIES(NAME)
-# ----------------------
-# See how the compiler implements dependency checking.
-# NAME is "CC", "CXX", "GCJ", or "OBJC".
-# We try a few techniques and use that to set a single cache variable.
-#
-# We don't AC_REQUIRE the corresponding AC_PROG_CC since the latter was
-# modified to invoke _AM_DEPENDENCIES(CC); we would have a circular
-# dependency, and given that the user is not expected to run this macro,
-# just rely on AC_PROG_CC.
-AC_DEFUN([_AM_DEPENDENCIES],
-[AC_REQUIRE([AM_SET_DEPDIR])dnl
-AC_REQUIRE([AM_OUTPUT_DEPENDENCY_COMMANDS])dnl
-AC_REQUIRE([AM_MAKE_INCLUDE])dnl
-AC_REQUIRE([AM_DEP_TRACK])dnl
-
-ifelse([$1], CC,   [depcc="$CC"   am_compiler_list=],
-       [$1], CXX,  [depcc="$CXX"  am_compiler_list=],
-       [$1], OBJC, [depcc="$OBJC" am_compiler_list='gcc3 gcc'],
-       [$1], UPC,  [depcc="$UPC"  am_compiler_list=],
-       [$1], GCJ,  [depcc="$GCJ"  am_compiler_list='gcc3 gcc'],
-                   [depcc="$$1"   am_compiler_list=])
-
-AC_CACHE_CHECK([dependency style of $depcc],
-               [am_cv_$1_dependencies_compiler_type],
-[if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
-  # We make a subdir and do the tests there.  Otherwise we can end up
-  # making bogus files that we don't know about and never remove.  For
-  # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named `D' -- because `-MD' means `put the output
-  # in D'.
-  mkdir conftest.dir
-  # Copy depcomp to subdir because otherwise we won't find it if we're
-  # using a relative directory.
-  cp "$am_depcomp" conftest.dir
-  cd conftest.dir
-  # We will build objects and dependencies in a subdirectory because
-  # it helps to detect inapplicable dependency modes.  For instance
-  # both Tru64's cc and ICC support -MD to output dependencies as a
-  # side effect of compilation, but ICC will put the dependencies in
-  # the current directory while Tru64 will put them in the object
-  # directory.
-  mkdir sub
-
-  am_cv_$1_dependencies_compiler_type=none
-  if test "$am_compiler_list" = ""; then
-     am_compiler_list=`sed -n ['s/^#*\([a-zA-Z0-9]*\))$/\1/p'] < ./depcomp`
-  fi
-  am__universal=false
-  m4_case([$1], [CC],
-    [case " $depcc " in #(
-     *\ -arch\ *\ -arch\ *) am__universal=true ;;
-     esac],
-    [CXX],
-    [case " $depcc " in #(
-     *\ -arch\ *\ -arch\ *) am__universal=true ;;
-     esac])
-
-  for depmode in $am_compiler_list; do
-    # Setup a source with many dependencies, because some compilers
-    # like to wrap large dependency lists on column 80 (with \), and
-    # we should not choose a depcomp mode which is confused by this.
-    #
-    # We need to recreate these files for each test, as the compiler may
-    # overwrite some of them when testing with obscure command lines.
-    # This happens at least with the AIX C compiler.
-    : > sub/conftest.c
-    for i in 1 2 3 4 5 6; do
-      echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
-      # Solaris 8's {/usr,}/bin/sh.
-      touch sub/conftst$i.h
-    done
-    echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
-
-    # We check with `-c' and `-o' for the sake of the "dashmstdout"
-    # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle `-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs
-    am__obj=sub/conftest.${OBJEXT-o}
-    am__minus_obj="-o $am__obj"
-    case $depmode in
-    gcc)
-      # This depmode causes a compiler race in universal mode.
-      test "$am__universal" = false || continue
-      ;;
-    nosideeffect)
-      # after this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested
-      if test "x$enable_dependency_tracking" = xyes; then
-	continue
-      else
-	break
-      fi
-      ;;
-    msvisualcpp | msvcmsys)
-      # This compiler won't grok `-c -o', but also, the minuso test has
-      # not run yet.  These depmodes are late enough in the game, and
-      # so weak that their functioning should not be impacted.
-      am__obj=conftest.${OBJEXT-o}
-      am__minus_obj=
-      ;;
-    none) break ;;
-    esac
-    if depmode=$depmode \
-       source=sub/conftest.c object=$am__obj \
-       depfile=sub/conftest.Po tmpdepfile=sub/conftest.TPo \
-       $SHELL ./depcomp $depcc -c $am__minus_obj sub/conftest.c \
-         >/dev/null 2>conftest.err &&
-       grep sub/conftst1.h sub/conftest.Po > /dev/null 2>&1 &&
-       grep sub/conftst6.h sub/conftest.Po > /dev/null 2>&1 &&
-       grep $am__obj sub/conftest.Po > /dev/null 2>&1 &&
-       ${MAKE-make} -s -f confmf > /dev/null 2>&1; then
-      # icc doesn't choke on unknown options, it will just issue warnings
-      # or remarks (even with -Werror).  So we grep stderr for any message
-      # that says an option was ignored or not supported.
-      # When given -MP, icc 7.0 and 7.1 complain thusly:
-      #   icc: Command line warning: ignoring option '-M'; no argument required
-      # The diagnosis changed in icc 8.0:
-      #   icc: Command line remark: option '-MP' not supported
-      if (grep 'ignoring option' conftest.err ||
-          grep 'not supported' conftest.err) >/dev/null 2>&1; then :; else
-        am_cv_$1_dependencies_compiler_type=$depmode
-        break
-      fi
-    fi
-  done
-
-  cd ..
-  rm -rf conftest.dir
-else
-  am_cv_$1_dependencies_compiler_type=none
-fi
-])
-AC_SUBST([$1DEPMODE], [depmode=$am_cv_$1_dependencies_compiler_type])
-AM_CONDITIONAL([am__fastdep$1], [
-  test "x$enable_dependency_tracking" != xno \
-  && test "$am_cv_$1_dependencies_compiler_type" = gcc3])
-])
-
-
-# AM_SET_DEPDIR
-# -------------
-# Choose a directory name for dependency files.
-# This macro is AC_REQUIREd in _AM_DEPENDENCIES
-AC_DEFUN([AM_SET_DEPDIR],
-[AC_REQUIRE([AM_SET_LEADING_DOT])dnl
-AC_SUBST([DEPDIR], ["${am__leading_dot}deps"])dnl
-])
-
-
-# AM_DEP_TRACK
-# ------------
-AC_DEFUN([AM_DEP_TRACK],
-[AC_ARG_ENABLE(dependency-tracking,
-[  --disable-dependency-tracking  speeds up one-time build
-  --enable-dependency-tracking   do not reject slow dependency extractors])
-if test "x$enable_dependency_tracking" != xno; then
-  am_depcomp="$ac_aux_dir/depcomp"
-  AMDEPBACKSLASH='\'
-fi
-AM_CONDITIONAL([AMDEP], [test "x$enable_dependency_tracking" != xno])
-AC_SUBST([AMDEPBACKSLASH])dnl
-_AM_SUBST_NOTMAKE([AMDEPBACKSLASH])dnl
-])
-
-# Generate code to set up dependency tracking.              -*- Autoconf -*-
-
-# Copyright (C) 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2008
-# Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-#serial 5
-
-# _AM_OUTPUT_DEPENDENCY_COMMANDS
-# ------------------------------
-AC_DEFUN([_AM_OUTPUT_DEPENDENCY_COMMANDS],
-[{
-  # Autoconf 2.62 quotes --file arguments for eval, but not when files
-  # are listed without --file.  Let's play safe and only enable the eval
-  # if we detect the quoting.
-  case $CONFIG_FILES in
-  *\'*) eval set x "$CONFIG_FILES" ;;
-  *)   set x $CONFIG_FILES ;;
-  esac
-  shift
-  for mf
-  do
-    # Strip MF so we end up with the name of the file.
-    mf=`echo "$mf" | sed -e 's/:.*$//'`
-    # Check whether this is an Automake generated Makefile or not.
-    # We used to match only the files named `Makefile.in', but
-    # some people rename them; so instead we look at the file content.
-    # Grep'ing the first line is not enough: some people post-process
-    # each Makefile.in and add a new line on top of each file to say so.
-    # Grep'ing the whole file is not good either: AIX grep has a line
-    # limit of 2048, but all sed's we know have understand at least 4000.
-    if sed -n 's,^#.*generated by automake.*,X,p' "$mf" | grep X >/dev/null 2>&1; then
-      dirpart=`AS_DIRNAME("$mf")`
-    else
-      continue
-    fi
-    # Extract the definition of DEPDIR, am__include, and am__quote
-    # from the Makefile without running `make'.
-    DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
-    test -z "$DEPDIR" && continue
-    am__include=`sed -n 's/^am__include = //p' < "$mf"`
-    test -z "am__include" && continue
-    am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
-    # When using ansi2knr, U may be empty or an underscore; expand it
-    U=`sed -n 's/^U = //p' < "$mf"`
-    # Find all dependency output files, they are included files with
-    # $(DEPDIR) in their names.  We invoke sed twice because it is the
-    # simplest approach to changing $(DEPDIR) to its actual value in the
-    # expansion.
-    for file in `sed -n "
-      s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
-	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g' -e 's/\$U/'"$U"'/g'`; do
-      # Make sure the directory exists.
-      test -f "$dirpart/$file" && continue
-      fdir=`AS_DIRNAME(["$file"])`
-      AS_MKDIR_P([$dirpart/$fdir])
-      # echo "creating $dirpart/$file"
-      echo '# dummy' > "$dirpart/$file"
-    done
-  done
-}
-])# _AM_OUTPUT_DEPENDENCY_COMMANDS
-
-
-# AM_OUTPUT_DEPENDENCY_COMMANDS
-# -----------------------------
-# This macro should only be invoked once -- use via AC_REQUIRE.
-#
-# This code is only required when automatic dependency tracking
-# is enabled.  FIXME.  This creates each `.P' file that we will
-# need in order to bootstrap the dependency handling code.
-AC_DEFUN([AM_OUTPUT_DEPENDENCY_COMMANDS],
-[AC_CONFIG_COMMANDS([depfiles],
-     [test x"$AMDEP_TRUE" != x"" || _AM_OUTPUT_DEPENDENCY_COMMANDS],
-     [AMDEP_TRUE="$AMDEP_TRUE" ac_aux_dir="$ac_aux_dir"])
-])
-
-# Do all the work for Automake.                             -*- Autoconf -*-
-
-# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
-# 2005, 2006, 2008, 2009 Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# serial 16
-
-# This macro actually does too much.  Some checks are only needed if
-# your package does certain things.  But this isn't really a big deal.
-
-# AM_INIT_AUTOMAKE(PACKAGE, VERSION, [NO-DEFINE])
-# AM_INIT_AUTOMAKE([OPTIONS])
-# -----------------------------------------------
-# The call with PACKAGE and VERSION arguments is the old style
-# call (pre autoconf-2.50), which is being phased out.  PACKAGE
-# and VERSION should now be passed to AC_INIT and removed from
-# the call to AM_INIT_AUTOMAKE.
-# We support both call styles for the transition.  After
-# the next Automake release, Autoconf can make the AC_INIT
-# arguments mandatory, and then we can depend on a new Autoconf
-# release and drop the old call support.
-AC_DEFUN([AM_INIT_AUTOMAKE],
-[AC_PREREQ([2.62])dnl
-dnl Autoconf wants to disallow AM_ names.  We explicitly allow
-dnl the ones we care about.
-m4_pattern_allow([^AM_[A-Z]+FLAGS$])dnl
-AC_REQUIRE([AM_SET_CURRENT_AUTOMAKE_VERSION])dnl
-AC_REQUIRE([AC_PROG_INSTALL])dnl
-if test "`cd $srcdir && pwd`" != "`pwd`"; then
-  # Use -I$(srcdir) only when $(srcdir) != ., so that make's output
-  # is not polluted with repeated "-I."
-  AC_SUBST([am__isrc], [' -I$(srcdir)'])_AM_SUBST_NOTMAKE([am__isrc])dnl
-  # test to see if srcdir already configured
-  if test -f $srcdir/config.status; then
-    AC_MSG_ERROR([source directory already configured; run "make distclean" there first])
-  fi
-fi
-
-# test whether we have cygpath
-if test -z "$CYGPATH_W"; then
-  if (cygpath --version) >/dev/null 2>/dev/null; then
-    CYGPATH_W='cygpath -w'
-  else
-    CYGPATH_W=echo
-  fi
-fi
-AC_SUBST([CYGPATH_W])
-
-# Define the identity of the package.
-dnl Distinguish between old-style and new-style calls.
-m4_ifval([$2],
-[m4_ifval([$3], [_AM_SET_OPTION([no-define])])dnl
- AC_SUBST([PACKAGE], [$1])dnl
- AC_SUBST([VERSION], [$2])],
-[_AM_SET_OPTIONS([$1])dnl
-dnl Diagnose old-style AC_INIT with new-style AM_AUTOMAKE_INIT.
-m4_if(m4_ifdef([AC_PACKAGE_NAME], 1)m4_ifdef([AC_PACKAGE_VERSION], 1), 11,,
-  [m4_fatal([AC_INIT should be called with package and version arguments])])dnl
- AC_SUBST([PACKAGE], ['AC_PACKAGE_TARNAME'])dnl
- AC_SUBST([VERSION], ['AC_PACKAGE_VERSION'])])dnl
-
-_AM_IF_OPTION([no-define],,
-[AC_DEFINE_UNQUOTED(PACKAGE, "$PACKAGE", [Name of package])
- AC_DEFINE_UNQUOTED(VERSION, "$VERSION", [Version number of package])])dnl
-
-# Some tools Automake needs.
-AC_REQUIRE([AM_SANITY_CHECK])dnl
-AC_REQUIRE([AC_ARG_PROGRAM])dnl
-AM_MISSING_PROG(ACLOCAL, aclocal-${am__api_version})
-AM_MISSING_PROG(AUTOCONF, autoconf)
-AM_MISSING_PROG(AUTOMAKE, automake-${am__api_version})
-AM_MISSING_PROG(AUTOHEADER, autoheader)
-AM_MISSING_PROG(MAKEINFO, makeinfo)
-AC_REQUIRE([AM_PROG_INSTALL_SH])dnl
-AC_REQUIRE([AM_PROG_INSTALL_STRIP])dnl
-AC_REQUIRE([AM_PROG_MKDIR_P])dnl
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
-AC_REQUIRE([AC_PROG_AWK])dnl
-AC_REQUIRE([AC_PROG_MAKE_SET])dnl
-AC_REQUIRE([AM_SET_LEADING_DOT])dnl
-_AM_IF_OPTION([tar-ustar], [_AM_PROG_TAR([ustar])],
-	      [_AM_IF_OPTION([tar-pax], [_AM_PROG_TAR([pax])],
-			     [_AM_PROG_TAR([v7])])])
-_AM_IF_OPTION([no-dependencies],,
-[AC_PROVIDE_IFELSE([AC_PROG_CC],
-		  [_AM_DEPENDENCIES(CC)],
-		  [define([AC_PROG_CC],
-			  defn([AC_PROG_CC])[_AM_DEPENDENCIES(CC)])])dnl
-AC_PROVIDE_IFELSE([AC_PROG_CXX],
-		  [_AM_DEPENDENCIES(CXX)],
-		  [define([AC_PROG_CXX],
-			  defn([AC_PROG_CXX])[_AM_DEPENDENCIES(CXX)])])dnl
-AC_PROVIDE_IFELSE([AC_PROG_OBJC],
-		  [_AM_DEPENDENCIES(OBJC)],
-		  [define([AC_PROG_OBJC],
-			  defn([AC_PROG_OBJC])[_AM_DEPENDENCIES(OBJC)])])dnl
-])
-_AM_IF_OPTION([silent-rules], [AC_REQUIRE([AM_SILENT_RULES])])dnl
-dnl The `parallel-tests' driver may need to know about EXEEXT, so add the
-dnl `am__EXEEXT' conditional if _AM_COMPILER_EXEEXT was seen.  This macro
-dnl is hooked onto _AC_COMPILER_EXEEXT early, see below.
-AC_CONFIG_COMMANDS_PRE(dnl
-[m4_provide_if([_AM_COMPILER_EXEEXT],
-  [AM_CONDITIONAL([am__EXEEXT], [test -n "$EXEEXT"])])])dnl
-])
-
-dnl Hook into `_AC_COMPILER_EXEEXT' early to learn its expansion.  Do not
-dnl add the conditional right here, as _AC_COMPILER_EXEEXT may be further
-dnl mangled by Autoconf and run in a shell conditional statement.
-m4_define([_AC_COMPILER_EXEEXT],
-m4_defn([_AC_COMPILER_EXEEXT])[m4_provide([_AM_COMPILER_EXEEXT])])
-
-
-# When config.status generates a header, we must update the stamp-h file.
-# This file resides in the same directory as the config header
-# that is generated.  The stamp files are numbered to have different names.
-
-# Autoconf calls _AC_AM_CONFIG_HEADER_HOOK (when defined) in the
-# loop where config.status creates the headers, so we can generate
-# our stamp files there.
-AC_DEFUN([_AC_AM_CONFIG_HEADER_HOOK],
-[# Compute $1's index in $config_headers.
-_am_arg=$1
-_am_stamp_count=1
-for _am_header in $config_headers :; do
-  case $_am_header in
-    $_am_arg | $_am_arg:* )
-      break ;;
-    * )
-      _am_stamp_count=`expr $_am_stamp_count + 1` ;;
-  esac
-done
-echo "timestamp for $_am_arg" >`AS_DIRNAME(["$_am_arg"])`/stamp-h[]$_am_stamp_count])
-
-# Copyright (C) 2001, 2003, 2005, 2008  Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# AM_PROG_INSTALL_SH
-# ------------------
-# Define $install_sh.
-AC_DEFUN([AM_PROG_INSTALL_SH],
-[AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
-if test x"${install_sh}" != xset; then
-  case $am_aux_dir in
-  *\ * | *\	*)
-    install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
-  *)
-    install_sh="\${SHELL} $am_aux_dir/install-sh"
-  esac
-fi
-AC_SUBST(install_sh)])
-
-# Copyright (C) 2003, 2005  Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# serial 2
-
-# Check whether the underlying file-system supports filenames
-# with a leading dot.  For instance MS-DOS doesn't.
-AC_DEFUN([AM_SET_LEADING_DOT],
-[rm -rf .tst 2>/dev/null
-mkdir .tst 2>/dev/null
-if test -d .tst; then
-  am__leading_dot=.
-else
-  am__leading_dot=_
-fi
-rmdir .tst 2>/dev/null
-AC_SUBST([am__leading_dot])])
-
-# Check to see how 'make' treats includes.	            -*- Autoconf -*-
-
-# Copyright (C) 2001, 2002, 2003, 2005, 2009  Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# serial 4
-
-# AM_MAKE_INCLUDE()
-# -----------------
-# Check to see how make treats includes.
-AC_DEFUN([AM_MAKE_INCLUDE],
-[am_make=${MAKE-make}
-cat > confinc << 'END'
-am__doit:
-	@echo this is the am__doit target
-.PHONY: am__doit
-END
-# If we don't find an include directive, just comment out the code.
-AC_MSG_CHECKING([for style of include used by $am_make])
-am__include="#"
-am__quote=
-_am_result=none
-# First try GNU make style include.
-echo "include confinc" > confmf
-# Ignore all kinds of additional output from `make'.
-case `$am_make -s -f confmf 2> /dev/null` in #(
-*the\ am__doit\ target*)
-  am__include=include
-  am__quote=
-  _am_result=GNU
-  ;;
-esac
-# Now try BSD make style include.
-if test "$am__include" = "#"; then
-   echo '.include "confinc"' > confmf
-   case `$am_make -s -f confmf 2> /dev/null` in #(
-   *the\ am__doit\ target*)
-     am__include=.include
-     am__quote="\""
-     _am_result=BSD
-     ;;
-   esac
-fi
-AC_SUBST([am__include])
-AC_SUBST([am__quote])
-AC_MSG_RESULT([$_am_result])
-rm -f confinc confmf
-])
-
-# Fake the existence of programs that GNU maintainers use.  -*- Autoconf -*-
-
-# Copyright (C) 1997, 1999, 2000, 2001, 2003, 2004, 2005, 2008
-# Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# serial 6
-
-# AM_MISSING_PROG(NAME, PROGRAM)
-# ------------------------------
-AC_DEFUN([AM_MISSING_PROG],
-[AC_REQUIRE([AM_MISSING_HAS_RUN])
-$1=${$1-"${am_missing_run}$2"}
-AC_SUBST($1)])
-
-
-# AM_MISSING_HAS_RUN
-# ------------------
-# Define MISSING if not defined so far and test if it supports --run.
-# If it does, set am_missing_run to use it, otherwise, to nothing.
-AC_DEFUN([AM_MISSING_HAS_RUN],
-[AC_REQUIRE([AM_AUX_DIR_EXPAND])dnl
-AC_REQUIRE_AUX_FILE([missing])dnl
-if test x"${MISSING+set}" != xset; then
-  case $am_aux_dir in
-  *\ * | *\	*)
-    MISSING="\${SHELL} \"$am_aux_dir/missing\"" ;;
-  *)
-    MISSING="\${SHELL} $am_aux_dir/missing" ;;
-  esac
-fi
-# Use eval to expand $SHELL
-if eval "$MISSING --run true"; then
-  am_missing_run="$MISSING --run "
-else
-  am_missing_run=
-  AC_MSG_WARN([`missing' script is too old or missing])
-fi
-])
-
-# Copyright (C) 2003, 2004, 2005, 2006  Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# AM_PROG_MKDIR_P
-# ---------------
-# Check for `mkdir -p'.
-AC_DEFUN([AM_PROG_MKDIR_P],
-[AC_PREREQ([2.60])dnl
-AC_REQUIRE([AC_PROG_MKDIR_P])dnl
-dnl Automake 1.8 to 1.9.6 used to define mkdir_p.  We now use MKDIR_P,
-dnl while keeping a definition of mkdir_p for backward compatibility.
-dnl @MKDIR_P@ is magic: AC_OUTPUT adjusts its value for each Makefile.
-dnl However we cannot define mkdir_p as $(MKDIR_P) for the sake of
-dnl Makefile.ins that do not define MKDIR_P, so we do our own
-dnl adjustment using top_builddir (which is defined more often than
-dnl MKDIR_P).
-AC_SUBST([mkdir_p], ["$MKDIR_P"])dnl
-case $mkdir_p in
-  [[\\/$]]* | ?:[[\\/]]*) ;;
-  */*) mkdir_p="\$(top_builddir)/$mkdir_p" ;;
-esac
-])
-
-# Helper functions for option handling.                     -*- Autoconf -*-
-
-# Copyright (C) 2001, 2002, 2003, 2005, 2008  Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# serial 4
-
-# _AM_MANGLE_OPTION(NAME)
-# -----------------------
-AC_DEFUN([_AM_MANGLE_OPTION],
-[[_AM_OPTION_]m4_bpatsubst($1, [[^a-zA-Z0-9_]], [_])])
-
-# _AM_SET_OPTION(NAME)
-# ------------------------------
-# Set option NAME.  Presently that only means defining a flag for this option.
-AC_DEFUN([_AM_SET_OPTION],
-[m4_define(_AM_MANGLE_OPTION([$1]), 1)])
-
-# _AM_SET_OPTIONS(OPTIONS)
-# ----------------------------------
-# OPTIONS is a space-separated list of Automake options.
-AC_DEFUN([_AM_SET_OPTIONS],
-[m4_foreach_w([_AM_Option], [$1], [_AM_SET_OPTION(_AM_Option)])])
-
-# _AM_IF_OPTION(OPTION, IF-SET, [IF-NOT-SET])
-# -------------------------------------------
-# Execute IF-SET if OPTION is set, IF-NOT-SET otherwise.
-AC_DEFUN([_AM_IF_OPTION],
-[m4_ifset(_AM_MANGLE_OPTION([$1]), [$2], [$3])])
-
-# Check to make sure that the build environment is sane.    -*- Autoconf -*-
-
-# Copyright (C) 1996, 1997, 2000, 2001, 2003, 2005, 2008
-# Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# serial 5
-
-# AM_SANITY_CHECK
-# ---------------
-AC_DEFUN([AM_SANITY_CHECK],
-[AC_MSG_CHECKING([whether build environment is sane])
-# Just in case
-sleep 1
-echo timestamp > conftest.file
-# Reject unsafe characters in $srcdir or the absolute working directory
-# name.  Accept space and tab only in the latter.
-am_lf='
-'
-case `pwd` in
-  *[[\\\"\#\$\&\'\`$am_lf]]*)
-    AC_MSG_ERROR([unsafe absolute working directory name]);;
-esac
-case $srcdir in
-  *[[\\\"\#\$\&\'\`$am_lf\ \	]]*)
-    AC_MSG_ERROR([unsafe srcdir value: `$srcdir']);;
-esac
-
-# Do `set' in a subshell so we don't clobber the current shell's
-# arguments.  Must try -L first in case configure is actually a
-# symlink; some systems play weird games with the mod time of symlinks
-# (eg FreeBSD returns the mod time of the symlink's containing
-# directory).
-if (
-   set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
-   if test "$[*]" = "X"; then
-      # -L didn't work.
-      set X `ls -t "$srcdir/configure" conftest.file`
-   fi
-   rm -f conftest.file
-   if test "$[*]" != "X $srcdir/configure conftest.file" \
-      && test "$[*]" != "X conftest.file $srcdir/configure"; then
-
-      # If neither matched, then we have a broken ls.  This can happen
-      # if, for instance, CONFIG_SHELL is bash and it inherits a
-      # broken ls alias from the environment.  This has actually
-      # happened.  Such a system could not be considered "sane".
-      AC_MSG_ERROR([ls -t appears to fail.  Make sure there is not a broken
-alias in your environment])
-   fi
-
-   test "$[2]" = conftest.file
-   )
-then
-   # Ok.
-   :
-else
-   AC_MSG_ERROR([newly created file is older than distributed files!
-Check your system clock])
-fi
-AC_MSG_RESULT(yes)])
-
-# Copyright (C) 2009  Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# serial 1
-
-# AM_SILENT_RULES([DEFAULT])
-# --------------------------
-# Enable less verbose build rules; with the default set to DEFAULT
-# (`yes' being less verbose, `no' or empty being verbose).
-AC_DEFUN([AM_SILENT_RULES],
-[AC_ARG_ENABLE([silent-rules],
-[  --enable-silent-rules          less verbose build output (undo: `make V=1')
-  --disable-silent-rules         verbose build output (undo: `make V=0')])
-case $enable_silent_rules in
-yes) AM_DEFAULT_VERBOSITY=0;;
-no)  AM_DEFAULT_VERBOSITY=1;;
-*)   AM_DEFAULT_VERBOSITY=m4_if([$1], [yes], [0], [1]);;
-esac
-AC_SUBST([AM_DEFAULT_VERBOSITY])dnl
-AM_BACKSLASH='\'
-AC_SUBST([AM_BACKSLASH])dnl
-_AM_SUBST_NOTMAKE([AM_BACKSLASH])dnl
-])
-
-# Copyright (C) 2001, 2003, 2005  Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# AM_PROG_INSTALL_STRIP
-# ---------------------
-# One issue with vendor `install' (even GNU) is that you can't
-# specify the program used to strip binaries.  This is especially
-# annoying in cross-compiling environments, where the build's strip
-# is unlikely to handle the host's binaries.
-# Fortunately install-sh will honor a STRIPPROG variable, so we
-# always use install-sh in `make install-strip', and initialize
-# STRIPPROG with the value of the STRIP variable (set by the user).
-AC_DEFUN([AM_PROG_INSTALL_STRIP],
-[AC_REQUIRE([AM_PROG_INSTALL_SH])dnl
-# Installed binaries are usually stripped using `strip' when the user
-# run `make install-strip'.  However `strip' might not be the right
-# tool to use in cross-compilation environments, therefore Automake
-# will honor the `STRIP' environment variable to overrule this program.
-dnl Don't test for $cross_compiling = yes, because it might be `maybe'.
-if test "$cross_compiling" != no; then
-  AC_CHECK_TOOL([STRIP], [strip], :)
-fi
-INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
-AC_SUBST([INSTALL_STRIP_PROGRAM])])
-
-# Copyright (C) 2006, 2008  Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# serial 2
-
-# _AM_SUBST_NOTMAKE(VARIABLE)
-# ---------------------------
-# Prevent Automake from outputting VARIABLE = @VARIABLE@ in Makefile.in.
-# This macro is traced by Automake.
-AC_DEFUN([_AM_SUBST_NOTMAKE])
-
-# AM_SUBST_NOTMAKE(VARIABLE)
-# ---------------------------
-# Public sister of _AM_SUBST_NOTMAKE.
-AC_DEFUN([AM_SUBST_NOTMAKE], [_AM_SUBST_NOTMAKE($@)])
-
-# Check how to create a tarball.                            -*- Autoconf -*-
-
-# Copyright (C) 2004, 2005  Free Software Foundation, Inc.
-#
-# This file is free software; the Free Software Foundation
-# gives unlimited permission to copy and/or distribute it,
-# with or without modifications, as long as this notice is preserved.
-
-# serial 2
-
-# _AM_PROG_TAR(FORMAT)
-# --------------------
-# Check how to create a tarball in format FORMAT.
-# FORMAT should be one of `v7', `ustar', or `pax'.
-#
-# Substitute a variable $(am__tar) that is a command
-# writing to stdout a FORMAT-tarball containing the directory
-# $tardir.
-#     tardir=directory && $(am__tar) > result.tar
-#
-# Substitute a variable $(am__untar) that extract such
-# a tarball read from stdin.
-#     $(am__untar) < result.tar
-AC_DEFUN([_AM_PROG_TAR],
-[# Always define AMTAR for backward compatibility.
-AM_MISSING_PROG([AMTAR], [tar])
-m4_if([$1], [v7],
-     [am__tar='${AMTAR} chof - "$$tardir"'; am__untar='${AMTAR} xf -'],
-     [m4_case([$1], [ustar],, [pax],,
-              [m4_fatal([Unknown tar format])])
-AC_MSG_CHECKING([how to create a $1 tar archive])
-# Loop over all known methods to create a tar archive until one works.
-_am_tools='gnutar m4_if([$1], [ustar], [plaintar]) pax cpio none'
-_am_tools=${am_cv_prog_tar_$1-$_am_tools}
-# Do not fold the above two line into one, because Tru64 sh and
-# Solaris sh will not grok spaces in the rhs of `-'.
-for _am_tool in $_am_tools
-do
-  case $_am_tool in
-  gnutar)
-    for _am_tar in tar gnutar gtar;
-    do
-      AM_RUN_LOG([$_am_tar --version]) && break
-    done
-    am__tar="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$$tardir"'
-    am__tar_="$_am_tar --format=m4_if([$1], [pax], [posix], [$1]) -chf - "'"$tardir"'
-    am__untar="$_am_tar -xf -"
-    ;;
-  plaintar)
-    # Must skip GNU tar: if it does not support --format= it doesn't create
-    # ustar tarball either.
-    (tar --version) >/dev/null 2>&1 && continue
-    am__tar='tar chf - "$$tardir"'
-    am__tar_='tar chf - "$tardir"'
-    am__untar='tar xf -'
-    ;;
-  pax)
-    am__tar='pax -L -x $1 -w "$$tardir"'
-    am__tar_='pax -L -x $1 -w "$tardir"'
-    am__untar='pax -r'
-    ;;
-  cpio)
-    am__tar='find "$$tardir" -print | cpio -o -H $1 -L'
-    am__tar_='find "$tardir" -print | cpio -o -H $1 -L'
-    am__untar='cpio -i -H $1 -d'
-    ;;
-  none)
-    am__tar=false
-    am__tar_=false
-    am__untar=false
-    ;;
-  esac
-
-  # If the value was cached, stop now.  We just wanted to have am__tar
-  # and am__untar set.
-  test -n "${am_cv_prog_tar_$1}" && break
-
-  # tar/untar a dummy directory, and stop if the command works
-  rm -rf conftest.dir
-  mkdir conftest.dir
-  echo GrepMe > conftest.dir/file
-  AM_RUN_LOG([tardir=conftest.dir && eval $am__tar_ >conftest.tar])
-  rm -rf conftest.dir
-  if test -s conftest.tar; then
-    AM_RUN_LOG([$am__untar <conftest.tar])
-    grep GrepMe conftest.dir/file >/dev/null 2>&1 && break
-  fi
-done
-rm -rf conftest.dir
-
-AC_CACHE_VAL([am_cv_prog_tar_$1], [am_cv_prog_tar_$1=$_am_tool])
-AC_MSG_RESULT([$am_cv_prog_tar_$1])])
-AC_SUBST([am__tar])
-AC_SUBST([am__untar])
-]) # _AM_PROG_TAR
-
-m4_include([m4/ac_prog_doxygen.m4])
-m4_include([m4/ax_cxx_compile_stdcxx_11.m4])
diff --git a/compile b/compile
deleted file mode 100755
index a85b723c..00000000
--- a/compile
+++ /dev/null
@@ -1,347 +0,0 @@
-#! /bin/sh
-# Wrapper for compilers which do not understand '-c -o'.
-
-scriptversion=2012-10-14.11; # UTC
-
-# Copyright (C) 1999-2014 Free Software Foundation, Inc.
-# Written by Tom Tromey <tromey@cygnus.com>.
-#
-# This program is free software; you can redistribute it and/or modify
-# it under the terms of the GNU General Public License as published by
-# the Free Software Foundation; either version 2, or (at your option)
-# any later version.
-#
-# This program is distributed in the hope that it will be useful,
-# but WITHOUT ANY WARRANTY; without even the implied warranty of
-# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-# GNU General Public License for more details.
-#
-# You should have received a copy of the GNU General Public License
-# along with this program.  If not, see <http://www.gnu.org/licenses/>.
-
-# As a special exception to the GNU General Public License, if you
-# distribute this file as part of a program that contains a
-# configuration script generated by Autoconf, you may include it under
-# the same distribution terms that you use for the rest of that program.
-
-# This file is maintained in Automake, please report
-# bugs to <bug-automake@gnu.org> or send patches to
-# <automake-patches@gnu.org>.
-
-nl='
-'
-
-# We need space, tab and new line, in precisely that order.  Quoting is
-# there to prevent tools from complaining about whitespace usage.
-IFS=" ""	$nl"
-
-file_conv=
-
-# func_file_conv build_file lazy
-# Convert a $build file to $host form and store it in $file
-# Currently only supports Windows hosts. If the determined conversion
-# type is listed in (the comma separated) LAZY, no conversion will
-# take place.
-func_file_conv ()
-{
-  file=$1
-  case $file in
-    / | /[!/]*) # absolute file, and not a UNC file
-      if test -z "$file_conv"; then
-	# lazily determine how to convert abs files
-	case `uname -s` in
-	  MINGW*)
-	    file_conv=mingw
-	    ;;
-	  CYGWIN*)
-	    file_conv=cygwin
-	    ;;
-	  *)
-	    file_conv=wine
-	    ;;
-	esac
-      fi
-      case $file_conv/,$2, in
-	*,$file_conv,*)
-	  ;;
-	mingw/*)
-	  file=`cmd //C echo "$file " | sed -e 's/"\(.*\) " *$/\1/'`
-	  ;;
-	cygwin/*)
-	  file=`cygpath -m "$file" || echo "$file"`
-	  ;;
-	wine/*)
-	  file=`winepath -w "$file" || echo "$file"`
-	  ;;
-      esac
-      ;;
-  esac
-}
-
-# func_cl_dashL linkdir
-# Make cl look for libraries in LINKDIR
-func_cl_dashL ()
-{
-  func_file_conv "$1"
-  if test -z "$lib_path"; then
-    lib_path=$file
-  else
-    lib_path="$lib_path;$file"
-  fi
-  linker_opts="$linker_opts -LIBPATH:$file"
-}
-
-# func_cl_dashl library
-# Do a library search-path lookup for cl
-func_cl_dashl ()
-{
-  lib=$1
-  found=no
-  save_IFS=$IFS
-  IFS=';'
-  for dir in $lib_path $LIB
-  do
-    IFS=$save_IFS
-    if $shared && test -f "$dir/$lib.dll.lib"; then
-      found=yes
-      lib=$dir/$lib.dll.lib
-      break
-    fi
-    if test -f "$dir/$lib.lib"; then
-      found=yes
-      lib=$dir/$lib.lib
-      break
-    fi
-    if test -f "$dir/lib$lib.a"; then
-      found=yes
-      lib=$dir/lib$lib.a
-      break
-    fi
-  done
-  IFS=$save_IFS
-
-  if test "$found" != yes; then
-    lib=$lib.lib
-  fi
-}
-
-# func_cl_wrapper cl arg...
-# Adjust compile command to suit cl
-func_cl_wrapper ()
-{
-  # Assume a capable shell
-  lib_path=
-  shared=:
-  linker_opts=
-  for arg
-  do
-    if test -n "$eat"; then
-      eat=
-    else
-      case $1 in
-	-o)
-	  # configure might choose to run compile as 'compile cc -o foo foo.c'.
-	  eat=1
-	  case $2 in
-	    *.o | *.[oO][bB][jJ])
-	      func_file_conv "$2"
-	      set x "$@" -Fo"$file"
-	      shift
-	      ;;
-	    *)
-	      func_file_conv "$2"
-	      set x "$@" -Fe"$file"
-	      shift
-	      ;;
-	  esac
-	  ;;
-	-I)
-	  eat=1
-	  func_file_conv "$2" mingw
-	  set x "$@" -I"$file"
-	  shift
-	  ;;
-	-I*)
-	  func_file_conv "${1#-I}" mingw
-	  set x "$@" -I"$file"
-	  shift
-	  ;;
-	-l)
-	  eat=1
-	  func_cl_dashl "$2"
-	  set x "$@" "$lib"
-	  shift
-	  ;;
-	-l*)
-	  func_cl_dashl "${1#-l}"
-	  set x "$@" "$lib"
-	  shift
-	  ;;
-	-L)
-	  eat=1
-	  func_cl_dashL "$2"
-	  ;;
-	-L*)
-	  func_cl_dashL "${1#-L}"
-	  ;;
-	-static)
-	  shared=false
-	  ;;
-	-Wl,*)
-	  arg=${1#-Wl,}
-	  save_ifs="$IFS"; IFS=','
-	  for flag in $arg; do
-	    IFS="$save_ifs"
-	    linker_opts="$linker_opts $flag"
-	  done
-	  IFS="$save_ifs"
-	  ;;
-	-Xlinker)
-	  eat=1
-	  linker_opts="$linker_opts $2"
-	  ;;
-	-*)
-	  set x "$@" "$1"
-	  shift
-	  ;;
-	*.cc | *.CC | *.cxx | *.CXX | *.[cC]++)
-	  func_file_conv "$1"
-	  set x "$@" -Tp"$file"
-	  shift
-	  ;;
-	*.c | *.cpp | *.CPP | *.lib | *.LIB | *.Lib | *.OBJ | *.obj | *.[oO])
-	  func_file_conv "$1" mingw
-	  set x "$@" "$file"
-	  shift
-	  ;;
-	*)
-	  set x "$@" "$1"
-	  shift
-	  ;;
-      esac
-    fi
-    shift
-  done
-  if test -n "$linker_opts"; then
-    linker_opts="-link$linker_opts"
-  fi
-  exec "$@" $linker_opts
-  exit 1
-}
-
-eat=
-
-case $1 in
-  '')
-     echo "$0: No command.  Try '$0 --help' for more information." 1>&2
-     exit 1;
-     ;;
-  -h | --h*)
-    cat <<\EOF
-Usage: compile [--help] [--version] PROGRAM [ARGS]
-
-Wrapper for compilers which do not understand '-c -o'.
-Remove '-o dest.o' from ARGS, run PROGRAM with the remaining
-arguments, and rename the output as expected.
-
-If you are trying to build a whole package this is not the
-right script to run: please start by reading the file 'INSTALL'.
-
-Report bugs to <bug-automake@gnu.org>.
-EOF
-    exit $?
-    ;;
-  -v | --v*)
-    echo "compile $scriptversion"
-    exit $?
-    ;;
-  cl | *[/\\]cl | cl.exe | *[/\\]cl.exe )
-    func_cl_wrapper "$@"      # Doesn't return...
-    ;;
-esac
-
-ofile=
-cfile=
-
-for arg
-do
-  if test -n "$eat"; then
-    eat=
-  else
-    case $1 in
-      -o)
-	# configure might choose to run compile as 'compile cc -o foo foo.c'.
-	# So we strip '-o arg' only if arg is an object.
-	eat=1
-	case $2 in
-	  *.o | *.obj)
-	    ofile=$2
-	    ;;
-	  *)
-	    set x "$@" -o "$2"
-	    shift
-	    ;;
-	esac
-	;;
-      *.c)
-	cfile=$1
-	set x "$@" "$1"
-	shift
-	;;
-      *)
-	set x "$@" "$1"
-	shift
-	;;
-    esac
-  fi
-  shift
-done
-
-if test -z "$ofile" || test -z "$cfile"; then
-  # If no '-o' option was seen then we might have been invoked from a
-  # pattern rule where we don't need one.  That is ok -- this is a
-  # normal compilation that the losing compiler can handle.  If no
-  # '.c' file was seen then we are probably linking.  That is also
-  # ok.
-  exec "$@"
-fi
-
-# Name of file we expect compiler to create.
-cofile=`echo "$cfile" | sed 's|^.*[\\/]||; s|^[a-zA-Z]:||; s/\.c$/.o/'`
-
-# Create the lock directory.
-# Note: use '[/\\:.-]' here to ensure that we don't use the same name
-# that we are using for the .o file.  Also, base the name on the expected
-# object file name, since that is what matters with a parallel build.
-lockdir=`echo "$cofile" | sed -e 's|[/\\:.-]|_|g'`.d
-while true; do
-  if mkdir "$lockdir" >/dev/null 2>&1; then
-    break
-  fi
-  sleep 1
-done
-# FIXME: race condition here if user kills between mkdir and trap.
-trap "rmdir '$lockdir'; exit 1" 1 2 15
-
-# Run the compile.
-"$@"
-ret=$?
-
-if test -f "$cofile"; then
-  test "$cofile" = "$ofile" || mv "$cofile" "$ofile"
-elif test -f "${cofile}bj"; then
-  test "${cofile}bj" = "$ofile" || mv "${cofile}bj" "$ofile"
-fi
-
-rmdir "$lockdir"
-exit $ret
-
-# Local Variables:
-# mode: shell-script
-# sh-indentation: 2
-# eval: (add-hook 'write-file-hooks 'time-stamp)
-# time-stamp-start: "scriptversion="
-# time-stamp-format: "%:y-%02m-%02d.%02H"
-# time-stamp-time-zone: "UTC"
-# time-stamp-end: "; # UTC"
-# End:
diff --git a/config.guess b/config.guess
deleted file mode 120000
index 405bc323..00000000
--- a/config.guess
+++ /dev/null
@@ -1 +0,0 @@
-/usr/share/automake-1.11/config.guess
\ No newline at end of file
diff --git a/config.sub b/config.sub
deleted file mode 120000
index 4d47fbcb..00000000
--- a/config.sub
+++ /dev/null
@@ -1 +0,0 @@
-/usr/share/automake-1.11/config.sub
\ No newline at end of file
diff --git a/configure b/configure
index e4dd388e..ccfea6b7 100755
--- a/configure
+++ b/configure
@@ -1,20 +1,22 @@
 #! /bin/sh
 # Guess values for system-dependent variables and create Makefiles.
-# Generated by GNU Autoconf 2.63 for Grid 1.0.
+# Generated by GNU Autoconf 2.69 for Grid 1.0.
 #
 # Report bugs to <paboyle@ph.ed.ac.uk>.
 #
-# Copyright (C) 1992, 1993, 1994, 1995, 1996, 1998, 1999, 2000, 2001,
-# 2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
+#
+# Copyright (C) 1992-1996, 1998-2012 Free Software Foundation, Inc.
+#
+#
 # This configure script is free software; the Free Software Foundation
 # gives unlimited permission to copy, distribute and modify it.
-## --------------------- ##
-## M4sh Initialization.  ##
-## --------------------- ##
+## -------------------- ##
+## M4sh Initialization. ##
+## -------------------- ##
 
 # Be more Bourne compatible
 DUALCASE=1; export DUALCASE # for MKS sh
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
   emulate sh
   NULLCMD=:
   # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
@@ -22,23 +24,15 @@ if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
   alias -g '${1+"$@"}'='"$@"'
   setopt NO_GLOB_SUBST
 else
-  case `(set -o) 2>/dev/null` in
-  *posix*) set -o posix ;;
+  case `(set -o) 2>/dev/null` in #(
+  *posix*) :
+    set -o posix ;; #(
+  *) :
+     ;;
 esac
-
 fi
 
 
-
-
-# PATH needs CR
-# Avoid depending upon Character Ranges.
-as_cr_letters='abcdefghijklmnopqrstuvwxyz'
-as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
-as_cr_Letters=$as_cr_letters$as_cr_LETTERS
-as_cr_digits='0123456789'
-as_cr_alnum=$as_cr_Letters$as_cr_digits
-
 as_nl='
 '
 export as_nl
@@ -46,7 +40,13 @@ export as_nl
 as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
-if (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+# Prefer a ksh shell builtin over an external printf program on Solaris,
+# but without wasting forks for bash or zsh.
+if test -z "$BASH_VERSION$ZSH_VERSION" \
+    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
+  as_echo='print -r --'
+  as_echo_n='print -rn --'
+elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
   as_echo='printf %s\n'
   as_echo_n='printf %s'
 else
@@ -57,7 +57,7 @@ else
     as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
     as_echo_n_body='eval
       arg=$1;
-      case $arg in
+      case $arg in #(
       *"$as_nl"*)
 	expr "X$arg" : "X\\(.*\\)$as_nl";
 	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
@@ -80,13 +80,6 @@ if test "${PATH_SEPARATOR+set}" != set; then
   }
 fi
 
-# Support unset when possible.
-if ( (MAIL=60; unset MAIL) || exit) >/dev/null 2>&1; then
-  as_unset=unset
-else
-  as_unset=false
-fi
-
 
 # IFS
 # We need space, tab and new line, in precisely that order.  Quoting is
@@ -96,15 +89,16 @@ fi
 IFS=" ""	$as_nl"
 
 # Find who we are.  Look in the path if we contain no directory separator.
-case $0 in
+as_myself=
+case $0 in #((
   *[\\/]* ) as_myself=$0 ;;
   *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
 for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
-done
+    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+  done
 IFS=$as_save_IFS
 
      ;;
@@ -116,12 +110,16 @@ if test "x$as_myself" = x; then
 fi
 if test ! -f "$as_myself"; then
   $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
-  { (exit 1); exit 1; }
+  exit 1
 fi
 
-# Work around bugs in pre-3.0 UWIN ksh.
-for as_var in ENV MAIL MAILPATH
-do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
+# Unset variables that we do not need and which cause bugs (e.g. in
+# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
+# suppresses any "Segmentation fault" message there.  '((' could
+# trigger a bug in pdksh 5.2.14.
+for as_var in BASH_ENV ENV MAIL MAILPATH
+do eval test x\${$as_var+set} = xset \
+  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
 done
 PS1='$ '
 PS2='> '
@@ -133,7 +131,294 @@ export LC_ALL
 LANGUAGE=C
 export LANGUAGE
 
-# Required to use basename.
+# CDPATH.
+(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
+
+# Use a proper internal environment variable to ensure we don't fall
+  # into an infinite loop, continuously re-executing ourselves.
+  if test x"${_as_can_reexec}" != xno && test "x$CONFIG_SHELL" != x; then
+    _as_can_reexec=no; export _as_can_reexec;
+    # We cannot yet assume a decent shell, so we have to provide a
+# neutralization value for shells without unset; and this also
+# works around shells that cannot unset nonexistent variables.
+# Preserve -v and -x to the replacement shell.
+BASH_ENV=/dev/null
+ENV=/dev/null
+(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
+case $- in # ((((
+  *v*x* | *x*v* ) as_opts=-vx ;;
+  *v* ) as_opts=-v ;;
+  *x* ) as_opts=-x ;;
+  * ) as_opts= ;;
+esac
+exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
+# Admittedly, this is quite paranoid, since all the known shells bail
+# out after a failed `exec'.
+$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
+as_fn_exit 255
+  fi
+  # We don't want this to propagate to other subprocesses.
+          { _as_can_reexec=; unset _as_can_reexec;}
+if test "x$CONFIG_SHELL" = x; then
+  as_bourne_compatible="if test -n \"\${ZSH_VERSION+set}\" && (emulate sh) >/dev/null 2>&1; then :
+  emulate sh
+  NULLCMD=:
+  # Pre-4.2 versions of Zsh do word splitting on \${1+\"\$@\"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '\${1+\"\$@\"}'='\"\$@\"'
+  setopt NO_GLOB_SUBST
+else
+  case \`(set -o) 2>/dev/null\` in #(
+  *posix*) :
+    set -o posix ;; #(
+  *) :
+     ;;
+esac
+fi
+"
+  as_required="as_fn_return () { (exit \$1); }
+as_fn_success () { as_fn_return 0; }
+as_fn_failure () { as_fn_return 1; }
+as_fn_ret_success () { return 0; }
+as_fn_ret_failure () { return 1; }
+
+exitcode=0
+as_fn_success || { exitcode=1; echo as_fn_success failed.; }
+as_fn_failure && { exitcode=1; echo as_fn_failure succeeded.; }
+as_fn_ret_success || { exitcode=1; echo as_fn_ret_success failed.; }
+as_fn_ret_failure && { exitcode=1; echo as_fn_ret_failure succeeded.; }
+if ( set x; as_fn_ret_success y && test x = \"\$1\" ); then :
+
+else
+  exitcode=1; echo positional parameters were not saved.
+fi
+test x\$exitcode = x0 || exit 1
+test -x / || exit 1"
+  as_suggested="  as_lineno_1=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_1a=\$LINENO
+  as_lineno_2=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_2a=\$LINENO
+  eval 'test \"x\$as_lineno_1'\$as_run'\" != \"x\$as_lineno_2'\$as_run'\" &&
+  test \"x\`expr \$as_lineno_1'\$as_run' + 1\`\" = \"x\$as_lineno_2'\$as_run'\"' || exit 1
+test \$(( 1 + 1 )) = 2 || exit 1"
+  if (eval "$as_required") 2>/dev/null; then :
+  as_have_required=yes
+else
+  as_have_required=no
+fi
+  if test x$as_have_required = xyes && (eval "$as_suggested") 2>/dev/null; then :
+
+else
+  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+as_found=false
+for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+  as_found=:
+  case $as_dir in #(
+	 /*)
+	   for as_base in sh bash ksh sh5; do
+	     # Try only shells that exist, to save several forks.
+	     as_shell=$as_dir/$as_base
+	     if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
+		    { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$as_shell"; } 2>/dev/null; then :
+  CONFIG_SHELL=$as_shell as_have_required=yes
+		   if { $as_echo "$as_bourne_compatible""$as_suggested" | as_run=a "$as_shell"; } 2>/dev/null; then :
+  break 2
+fi
+fi
+	   done;;
+       esac
+  as_found=false
+done
+$as_found || { if { test -f "$SHELL" || test -f "$SHELL.exe"; } &&
+	      { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$SHELL"; } 2>/dev/null; then :
+  CONFIG_SHELL=$SHELL as_have_required=yes
+fi; }
+IFS=$as_save_IFS
+
+
+      if test "x$CONFIG_SHELL" != x; then :
+  export CONFIG_SHELL
+             # We cannot yet assume a decent shell, so we have to provide a
+# neutralization value for shells without unset; and this also
+# works around shells that cannot unset nonexistent variables.
+# Preserve -v and -x to the replacement shell.
+BASH_ENV=/dev/null
+ENV=/dev/null
+(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
+case $- in # ((((
+  *v*x* | *x*v* ) as_opts=-vx ;;
+  *v* ) as_opts=-v ;;
+  *x* ) as_opts=-x ;;
+  * ) as_opts= ;;
+esac
+exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
+# Admittedly, this is quite paranoid, since all the known shells bail
+# out after a failed `exec'.
+$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
+exit 255
+fi
+
+    if test x$as_have_required = xno; then :
+  $as_echo "$0: This script requires a shell more modern than all"
+  $as_echo "$0: the shells that I found on your system."
+  if test x${ZSH_VERSION+set} = xset ; then
+    $as_echo "$0: In particular, zsh $ZSH_VERSION has bugs and should"
+    $as_echo "$0: be upgraded to zsh 4.3.4 or later."
+  else
+    $as_echo "$0: Please tell bug-autoconf@gnu.org and
+$0: paboyle@ph.ed.ac.uk about your system, including any
+$0: error possibly output before this message. Then install
+$0: a modern shell, or manually run the script under such a
+$0: shell if you do have one."
+  fi
+  exit 1
+fi
+fi
+fi
+SHELL=${CONFIG_SHELL-/bin/sh}
+export SHELL
+# Unset more variables known to interfere with behavior of common tools.
+CLICOLOR_FORCE= GREP_OPTIONS=
+unset CLICOLOR_FORCE GREP_OPTIONS
+
+## --------------------- ##
+## M4sh Shell Functions. ##
+## --------------------- ##
+# as_fn_unset VAR
+# ---------------
+# Portably unset VAR.
+as_fn_unset ()
+{
+  { eval $1=; unset $1;}
+}
+as_unset=as_fn_unset
+
+# as_fn_set_status STATUS
+# -----------------------
+# Set $? to STATUS, without forking.
+as_fn_set_status ()
+{
+  return $1
+} # as_fn_set_status
+
+# as_fn_exit STATUS
+# -----------------
+# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
+as_fn_exit ()
+{
+  set +e
+  as_fn_set_status $1
+  exit $1
+} # as_fn_exit
+
+# as_fn_mkdir_p
+# -------------
+# Create "$as_dir" as a directory, including parents if necessary.
+as_fn_mkdir_p ()
+{
+
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || eval $as_mkdir_p || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
+
+
+} # as_fn_mkdir_p
+
+# as_fn_executable_p FILE
+# -----------------------
+# Test if FILE is an executable regular file.
+as_fn_executable_p ()
+{
+  test -f "$1" && test -x "$1"
+} # as_fn_executable_p
+# as_fn_append VAR VALUE
+# ----------------------
+# Append the text in VALUE to the end of the definition contained in VAR. Take
+# advantage of any shell optimizations that allow amortized linear growth over
+# repeated appends, instead of the typical quadratic growth present in naive
+# implementations.
+if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
+  eval 'as_fn_append ()
+  {
+    eval $1+=\$2
+  }'
+else
+  as_fn_append ()
+  {
+    eval $1=\$$1\$2
+  }
+fi # as_fn_append
+
+# as_fn_arith ARG...
+# ------------------
+# Perform arithmetic evaluation on the ARGs, and store the result in the
+# global $as_val. Take advantage of shells that can avoid forks. The arguments
+# must be portable across $(()) and expr.
+if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
+  eval 'as_fn_arith ()
+  {
+    as_val=$(( $* ))
+  }'
+else
+  as_fn_arith ()
+  {
+    as_val=`expr "$@" || test $? -eq 1`
+  }
+fi # as_fn_arith
+
+
+# as_fn_error STATUS ERROR [LINENO LOG_FD]
+# ----------------------------------------
+# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
+# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
+# script with STATUS, using 1 if that was 0.
+as_fn_error ()
+{
+  as_status=$1; test $as_status -eq 0 && as_status=1
+  if test "$4"; then
+    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
+  fi
+  $as_echo "$as_me: error: $2" >&2
+  as_fn_exit $as_status
+} # as_fn_error
+
 if expr a : '\(a\)' >/dev/null 2>&1 &&
    test "X`expr 00001 : '.*\(...\)'`" = X001; then
   as_expr=expr
@@ -147,8 +432,12 @@ else
   as_basename=false
 fi
 
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
 
-# Name of the executable.
 as_me=`$as_basename -- "$0" ||
 $as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
 	 X"$0" : 'X\(//\)$' \| \
@@ -168,295 +457,19 @@ $as_echo X/"$0" |
 	  }
 	  s/.*/./; q'`
 
-# CDPATH.
-$as_unset CDPATH
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
 
 
-if test "x$CONFIG_SHELL" = x; then
-  if (eval ":") 2>/dev/null; then
-  as_have_required=yes
-else
-  as_have_required=no
-fi
-
-  if test $as_have_required = yes &&	 (eval ":
-(as_func_return () {
-  (exit \$1)
-}
-as_func_success () {
-  as_func_return 0
-}
-as_func_failure () {
-  as_func_return 1
-}
-as_func_ret_success () {
-  return 0
-}
-as_func_ret_failure () {
-  return 1
-}
-
-exitcode=0
-if as_func_success; then
-  :
-else
-  exitcode=1
-  echo as_func_success failed.
-fi
-
-if as_func_failure; then
-  exitcode=1
-  echo as_func_failure succeeded.
-fi
-
-if as_func_ret_success; then
-  :
-else
-  exitcode=1
-  echo as_func_ret_success failed.
-fi
-
-if as_func_ret_failure; then
-  exitcode=1
-  echo as_func_ret_failure succeeded.
-fi
-
-if ( set x; as_func_ret_success y && test x = \"\$1\" ); then
-  :
-else
-  exitcode=1
-  echo positional parameters were not saved.
-fi
-
-test \$exitcode = 0) || { (exit 1); exit 1; }
-
-(
-  as_lineno_1=\$LINENO
-  as_lineno_2=\$LINENO
-  test \"x\$as_lineno_1\" != \"x\$as_lineno_2\" &&
-  test \"x\`expr \$as_lineno_1 + 1\`\" = \"x\$as_lineno_2\") || { (exit 1); exit 1; }
-") 2> /dev/null; then
-  :
-else
-  as_candidate_shells=
-    as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-  case $as_dir in
-	 /*)
-	   for as_base in sh bash ksh sh5; do
-	     as_candidate_shells="$as_candidate_shells $as_dir/$as_base"
-	   done;;
-       esac
-done
-IFS=$as_save_IFS
-
-
-      for as_shell in $as_candidate_shells $SHELL; do
-	 # Try only shells that exist, to save several forks.
-	 if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
-		{ ("$as_shell") 2> /dev/null <<\_ASEOF
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
-  emulate sh
-  NULLCMD=:
-  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
-  # is contrary to our usage.  Disable this feature.
-  alias -g '${1+"$@"}'='"$@"'
-  setopt NO_GLOB_SUBST
-else
-  case `(set -o) 2>/dev/null` in
-  *posix*) set -o posix ;;
-esac
-
-fi
-
-
-:
-_ASEOF
-}; then
-  CONFIG_SHELL=$as_shell
-	       as_have_required=yes
-	       if { "$as_shell" 2> /dev/null <<\_ASEOF
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
-  emulate sh
-  NULLCMD=:
-  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
-  # is contrary to our usage.  Disable this feature.
-  alias -g '${1+"$@"}'='"$@"'
-  setopt NO_GLOB_SUBST
-else
-  case `(set -o) 2>/dev/null` in
-  *posix*) set -o posix ;;
-esac
-
-fi
-
-
-:
-(as_func_return () {
-  (exit $1)
-}
-as_func_success () {
-  as_func_return 0
-}
-as_func_failure () {
-  as_func_return 1
-}
-as_func_ret_success () {
-  return 0
-}
-as_func_ret_failure () {
-  return 1
-}
-
-exitcode=0
-if as_func_success; then
-  :
-else
-  exitcode=1
-  echo as_func_success failed.
-fi
-
-if as_func_failure; then
-  exitcode=1
-  echo as_func_failure succeeded.
-fi
-
-if as_func_ret_success; then
-  :
-else
-  exitcode=1
-  echo as_func_ret_success failed.
-fi
-
-if as_func_ret_failure; then
-  exitcode=1
-  echo as_func_ret_failure succeeded.
-fi
-
-if ( set x; as_func_ret_success y && test x = "$1" ); then
-  :
-else
-  exitcode=1
-  echo positional parameters were not saved.
-fi
-
-test $exitcode = 0) || { (exit 1); exit 1; }
-
-(
-  as_lineno_1=$LINENO
-  as_lineno_2=$LINENO
-  test "x$as_lineno_1" != "x$as_lineno_2" &&
-  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2") || { (exit 1); exit 1; }
-
-_ASEOF
-}; then
-  break
-fi
-
-fi
-
-      done
-
-      if test "x$CONFIG_SHELL" != x; then
-  for as_var in BASH_ENV ENV
-	do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
-	done
-	export CONFIG_SHELL
-	exec "$CONFIG_SHELL" "$as_myself" ${1+"$@"}
-fi
-
-
-    if test $as_have_required = no; then
-  echo This script requires a shell more modern than all the
-      echo shells that I found on your system.  Please install a
-      echo modern shell, or manually run the script under such a
-      echo shell if you do have one.
-      { (exit 1); exit 1; }
-fi
-
-
-fi
-
-fi
-
-
-
-(eval "as_func_return () {
-  (exit \$1)
-}
-as_func_success () {
-  as_func_return 0
-}
-as_func_failure () {
-  as_func_return 1
-}
-as_func_ret_success () {
-  return 0
-}
-as_func_ret_failure () {
-  return 1
-}
-
-exitcode=0
-if as_func_success; then
-  :
-else
-  exitcode=1
-  echo as_func_success failed.
-fi
-
-if as_func_failure; then
-  exitcode=1
-  echo as_func_failure succeeded.
-fi
-
-if as_func_ret_success; then
-  :
-else
-  exitcode=1
-  echo as_func_ret_success failed.
-fi
-
-if as_func_ret_failure; then
-  exitcode=1
-  echo as_func_ret_failure succeeded.
-fi
-
-if ( set x; as_func_ret_success y && test x = \"\$1\" ); then
-  :
-else
-  exitcode=1
-  echo positional parameters were not saved.
-fi
-
-test \$exitcode = 0") || {
-  echo No shell found that supports shell functions.
-  echo Please tell bug-autoconf@gnu.org about your system,
-  echo including any error possibly output before this message.
-  echo This can help us improve future autoconf versions.
-  echo Configuration will now proceed without shell functions.
-}
-
-
-
-  as_lineno_1=$LINENO
-  as_lineno_2=$LINENO
-  test "x$as_lineno_1" != "x$as_lineno_2" &&
-  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2" || {
-
-  # Create $as_me.lineno as a copy of $as_myself, but with $LINENO
-  # uniformly replaced by the line number.  The first 'sed' inserts a
-  # line-number line after each line using $LINENO; the second 'sed'
-  # does the real work.  The second script uses 'N' to pair each
-  # line-number line with the line containing $LINENO, and appends
-  # trailing '-' during substitution so that $LINENO is not a special
-  # case at line end.
-  # (Raja R Harinath suggested sed '=', and Paul Eggert wrote the
-  # scripts with optimization help from Paolo Bonzini.  Blame Lee
-  # E. McMahon (1931-1989) for sed's syntax.  :-)
+  as_lineno_1=$LINENO as_lineno_1a=$LINENO
+  as_lineno_2=$LINENO as_lineno_2a=$LINENO
+  eval 'test "x$as_lineno_1'$as_run'" != "x$as_lineno_2'$as_run'" &&
+  test "x`expr $as_lineno_1'$as_run' + 1`" = "x$as_lineno_2'$as_run'"' || {
+  # Blame Lee E. McMahon (1931-1989) for sed's syntax.  :-)
   sed -n '
     p
     /[$]LINENO/=
@@ -473,9 +486,12 @@ test \$exitcode = 0") || {
       s/-\n.*//
     ' >$as_me.lineno &&
   chmod +x "$as_me.lineno" ||
-    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2
-   { (exit 1); exit 1; }; }
+    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2; as_fn_exit 1; }
 
+  # If we had to re-execute with $CONFIG_SHELL, we're ensured to have
+  # already done that, so ensure we don't try to do so again and fall
+  # in an infinite loop.  This has already happened in practice.
+  _as_can_reexec=no; export _as_can_reexec
   # Don't try to exec as it changes $[0], causing all sort of problems
   # (the dirname of $[0] is not the place where we might find the
   # original and so on.  Autoconf is especially sensitive to this).
@@ -484,29 +500,18 @@ test \$exitcode = 0") || {
   exit
 }
 
-
-if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
-  as_dirname=dirname
-else
-  as_dirname=false
-fi
-
 ECHO_C= ECHO_N= ECHO_T=
-case `echo -n x` in
+case `echo -n x` in #(((((
 -n*)
-  case `echo 'x\c'` in
+  case `echo 'xy\c'` in
   *c*) ECHO_T='	';;	# ECHO_T is single tab character.
-  *)   ECHO_C='\c';;
+  xy)  ECHO_C='\c';;
+  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
+       ECHO_T='	';;
   esac;;
 *)
   ECHO_N='-n';;
 esac
-if expr a : '\(a\)' >/dev/null 2>&1 &&
-   test "X`expr 00001 : '.*\(...\)'`" = X001; then
-  as_expr=expr
-else
-  as_expr=false
-fi
 
 rm -f conf$$ conf$$.exe conf$$.file
 if test -d conf$$.dir; then
@@ -521,49 +526,29 @@ if (echo >conf$$.file) 2>/dev/null; then
     # ... but there are two gotchas:
     # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
     # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
-    # In both cases, we have to default to `cp -p'.
+    # In both cases, we have to default to `cp -pR'.
     ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
-      as_ln_s='cp -p'
+      as_ln_s='cp -pR'
   elif ln conf$$.file conf$$ 2>/dev/null; then
     as_ln_s=ln
   else
-    as_ln_s='cp -p'
+    as_ln_s='cp -pR'
   fi
 else
-  as_ln_s='cp -p'
+  as_ln_s='cp -pR'
 fi
 rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
 rmdir conf$$.dir 2>/dev/null
 
 if mkdir -p . 2>/dev/null; then
-  as_mkdir_p=:
+  as_mkdir_p='mkdir -p "$as_dir"'
 else
   test -d ./-p && rmdir ./-p
   as_mkdir_p=false
 fi
 
-if test -x / >/dev/null 2>&1; then
-  as_test_x='test -x'
-else
-  if ls -dL / >/dev/null 2>&1; then
-    as_ls_L_option=L
-  else
-    as_ls_L_option=
-  fi
-  as_test_x='
-    eval sh -c '\''
-      if test -d "$1"; then
-	test -d "$1/.";
-      else
-	case $1 in
-	-*)set "./$1";;
-	esac;
-	case `ls -ld'$as_ls_L_option' "$1" 2>/dev/null` in
-	???[sx]*):;;*)false;;esac;fi
-    '\'' sh
-  '
-fi
-as_executable_p=$as_test_x
+as_test_x='test -x'
+as_executable_p=as_fn_executable_p
 
 # Sed expression to map a string onto a valid CPP name.
 as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
@@ -572,11 +557,11 @@ as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
 as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
 
 
-
-exec 7<&0 </dev/null 6>&1
+test -n "$DJDIR" || exec 7<&0 </dev/null
+exec 6>&1
 
 # Name of the host.
-# hostname on some systems (SVR3.2, Linux) returns a bogus exit status,
+# hostname on some systems (SVR3.2, old GNU/Linux) returns a bogus exit status,
 # so uname gets run too.
 ac_hostname=`(hostname || uname -n) 2>/dev/null | sed 1q`
 
@@ -591,7 +576,6 @@ cross_compiling=no
 subdirs=
 MFLAGS=
 MAKEFLAGS=
-SHELL=${CONFIG_SHELL-/bin/sh}
 
 # Identity of this package.
 PACKAGE_NAME='Grid'
@@ -599,6 +583,7 @@ PACKAGE_TARNAME='grid'
 PACKAGE_VERSION='1.0'
 PACKAGE_STRING='Grid 1.0'
 PACKAGE_BUGREPORT='paboyle@ph.ed.ac.uk'
+PACKAGE_URL=''
 
 ac_unique_file="lib/Grid.h"
 # Factoring default headers for most tests.
@@ -661,6 +646,7 @@ OPENMP_CXXFLAGS
 am__fastdepCXX_FALSE
 am__fastdepCXX_TRUE
 CXXDEPMODE
+am__nodep
 AMDEPBACKSLASH
 AMDEP_FALSE
 AMDEP_TRUE
@@ -676,6 +662,8 @@ CXXFLAGS
 CXX
 AM_BACKSLASH
 AM_DEFAULT_VERBOSITY
+AM_DEFAULT_V
+AM_V
 am__untar
 am__tar
 AMTAR
@@ -741,6 +729,7 @@ bindir
 program_transform_name
 prefix
 exec_prefix
+PACKAGE_URL
 PACKAGE_BUGREPORT
 PACKAGE_STRING
 PACKAGE_VERSION
@@ -833,8 +822,9 @@ do
   fi
 
   case $ac_option in
-  *=*)	ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
-  *)	ac_optarg=yes ;;
+  *=?*) ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
+  *=)   ac_optarg= ;;
+  *)    ac_optarg=yes ;;
   esac
 
   # Accept the important Cygnus configure options, so we can diagnose typos.
@@ -879,8 +869,7 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*disable-\(.*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid feature name: $ac_useropt" >&2
-   { (exit 1); exit 1; }; }
+      as_fn_error $? "invalid feature name: $ac_useropt"
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -906,8 +895,7 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*enable-\([^=]*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid feature name: $ac_useropt" >&2
-   { (exit 1); exit 1; }; }
+      as_fn_error $? "invalid feature name: $ac_useropt"
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1111,8 +1099,7 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*with-\([^=]*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid package name: $ac_useropt" >&2
-   { (exit 1); exit 1; }; }
+      as_fn_error $? "invalid package name: $ac_useropt"
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1128,8 +1115,7 @@ do
     ac_useropt=`expr "x$ac_option" : 'x-*without-\(.*\)'`
     # Reject names that are not valid shell variable names.
     expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid package name: $ac_useropt" >&2
-   { (exit 1); exit 1; }; }
+      as_fn_error $? "invalid package name: $ac_useropt"
     ac_useropt_orig=$ac_useropt
     ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
     case $ac_user_opts in
@@ -1159,17 +1145,17 @@ do
   | --x-librar=* | --x-libra=* | --x-libr=* | --x-lib=* | --x-li=* | --x-l=*)
     x_libraries=$ac_optarg ;;
 
-  -*) { $as_echo "$as_me: error: unrecognized option: $ac_option
-Try \`$0 --help' for more information." >&2
-   { (exit 1); exit 1; }; }
+  -*) as_fn_error $? "unrecognized option: \`$ac_option'
+Try \`$0 --help' for more information"
     ;;
 
   *=*)
     ac_envvar=`expr "x$ac_option" : 'x\([^=]*\)='`
     # Reject names that are not valid shell variable names.
-    expr "x$ac_envvar" : ".*[^_$as_cr_alnum]" >/dev/null &&
-      { $as_echo "$as_me: error: invalid variable name: $ac_envvar" >&2
-   { (exit 1); exit 1; }; }
+    case $ac_envvar in #(
+      '' | [0-9]* | *[!_$as_cr_alnum]* )
+      as_fn_error $? "invalid variable name: \`$ac_envvar'" ;;
+    esac
     eval $ac_envvar=\$ac_optarg
     export $ac_envvar ;;
 
@@ -1178,7 +1164,7 @@ Try \`$0 --help' for more information." >&2
     $as_echo "$as_me: WARNING: you should use --build, --host, --target" >&2
     expr "x$ac_option" : ".*[^-._$as_cr_alnum]" >/dev/null &&
       $as_echo "$as_me: WARNING: invalid host type: $ac_option" >&2
-    : ${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}
+    : "${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}"
     ;;
 
   esac
@@ -1186,15 +1172,13 @@ done
 
 if test -n "$ac_prev"; then
   ac_option=--`echo $ac_prev | sed 's/_/-/g'`
-  { $as_echo "$as_me: error: missing argument to $ac_option" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "missing argument to $ac_option"
 fi
 
 if test -n "$ac_unrecognized_opts"; then
   case $enable_option_checking in
     no) ;;
-    fatal) { $as_echo "$as_me: error: unrecognized options: $ac_unrecognized_opts" >&2
-   { (exit 1); exit 1; }; } ;;
+    fatal) as_fn_error $? "unrecognized options: $ac_unrecognized_opts" ;;
     *)     $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2 ;;
   esac
 fi
@@ -1217,8 +1201,7 @@ do
     [\\/$]* | ?:[\\/]* )  continue;;
     NONE | '' ) case $ac_var in *prefix ) continue;; esac;;
   esac
-  { $as_echo "$as_me: error: expected an absolute directory name for --$ac_var: $ac_val" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "expected an absolute directory name for --$ac_var: $ac_val"
 done
 
 # There might be people who depend on the old broken behavior: `$host'
@@ -1232,8 +1215,6 @@ target=$target_alias
 if test "x$host_alias" != x; then
   if test "x$build_alias" = x; then
     cross_compiling=maybe
-    $as_echo "$as_me: WARNING: If you wanted to set the --build type, don't use --host.
-    If a cross compiler is detected then cross compile mode will be used." >&2
   elif test "x$build_alias" != "x$host_alias"; then
     cross_compiling=yes
   fi
@@ -1248,11 +1229,9 @@ test "$silent" = yes && exec 6>/dev/null
 ac_pwd=`pwd` && test -n "$ac_pwd" &&
 ac_ls_di=`ls -di .` &&
 ac_pwd_ls_di=`cd "$ac_pwd" && ls -di .` ||
-  { $as_echo "$as_me: error: working directory cannot be determined" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "working directory cannot be determined"
 test "X$ac_ls_di" = "X$ac_pwd_ls_di" ||
-  { $as_echo "$as_me: error: pwd does not report name of working directory" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "pwd does not report name of working directory"
 
 
 # Find the source files, if location was not specified.
@@ -1291,13 +1270,11 @@ else
 fi
 if test ! -r "$srcdir/$ac_unique_file"; then
   test "$ac_srcdir_defaulted" = yes && srcdir="$ac_confdir or .."
-  { $as_echo "$as_me: error: cannot find sources ($ac_unique_file) in $srcdir" >&2
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "cannot find sources ($ac_unique_file) in $srcdir"
 fi
 ac_msg="sources are in $srcdir, but \`cd $srcdir' does not work"
 ac_abs_confdir=`(
-	cd "$srcdir" && test -r "./$ac_unique_file" || { $as_echo "$as_me: error: $ac_msg" >&2
-   { (exit 1); exit 1; }; }
+	cd "$srcdir" && test -r "./$ac_unique_file" || as_fn_error $? "$ac_msg"
 	pwd)`
 # When building in place, set srcdir=.
 if test "$ac_abs_confdir" = "$ac_pwd"; then
@@ -1337,7 +1314,7 @@ Configuration:
       --help=short        display options specific to this package
       --help=recursive    display the short help of all the included packages
   -V, --version           display version information and exit
-  -q, --quiet, --silent   do not print \`checking...' messages
+  -q, --quiet, --silent   do not print \`checking ...' messages
       --cache-file=FILE   cache test results in FILE [disabled]
   -C, --config-cache      alias for \`--cache-file=config.cache'
   -n, --no-create         do not create output files
@@ -1402,10 +1379,12 @@ Optional Features:
   --disable-option-checking  ignore unrecognized --enable/--with options
   --disable-FEATURE       do not include FEATURE (same as --enable-FEATURE=no)
   --enable-FEATURE[=ARG]  include FEATURE [ARG=yes]
-  --enable-silent-rules          less verbose build output (undo: `make V=1')
-  --disable-silent-rules         verbose build output (undo: `make V=0')
-  --disable-dependency-tracking  speeds up one-time build
-  --enable-dependency-tracking   do not reject slow dependency extractors
+  --enable-silent-rules   less verbose build output (undo: "make V=1")
+  --disable-silent-rules  verbose build output (undo: "make V=0")
+  --enable-dependency-tracking
+                          do not reject slow dependency extractors
+  --disable-dependency-tracking
+                          speeds up one-time build
   --disable-openmp        do not use OpenMP
   --enable-simd=SSE|AVX|AVX2|AVX512|MIC
                           Select instructions to be SSE4.0, AVX 1.0, AVX
@@ -1423,7 +1402,7 @@ Some influential environment variables:
   LDFLAGS     linker flags, e.g. -L<lib dir> if you have libraries in a
               nonstandard directory <lib dir>
   LIBS        libraries to pass to the linker, e.g. -l<library>
-  CPPFLAGS    C/C++/Objective C preprocessor flags, e.g. -I<include dir> if
+  CPPFLAGS    (Objective) C/C++ preprocessor flags, e.g. -I<include dir> if
               you have headers in a nonstandard directory <include dir>
   CXXCPP      C++ preprocessor
 
@@ -1494,21 +1473,530 @@ test -n "$ac_init_help" && exit $ac_status
 if $ac_init_version; then
   cat <<\_ACEOF
 Grid configure 1.0
-generated by GNU Autoconf 2.63
+generated by GNU Autoconf 2.69
 
-Copyright (C) 1992, 1993, 1994, 1995, 1996, 1998, 1999, 2000, 2001,
-2002, 2003, 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
+Copyright (C) 2012 Free Software Foundation, Inc.
 This configure script is free software; the Free Software Foundation
 gives unlimited permission to copy, distribute and modify it.
 _ACEOF
   exit
 fi
+
+## ------------------------ ##
+## Autoconf initialization. ##
+## ------------------------ ##
+
+# ac_fn_cxx_try_compile LINENO
+# ----------------------------
+# Try to compile conftest.$ac_ext, and return whether this succeeded.
+ac_fn_cxx_try_compile ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  rm -f conftest.$ac_objext
+  if { { ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compile") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_retval=1
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_cxx_try_compile
+
+# ac_fn_cxx_try_link LINENO
+# -------------------------
+# Try to link conftest.$ac_ext, and return whether this succeeded.
+ac_fn_cxx_try_link ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  rm -f conftest.$ac_objext conftest$ac_exeext
+  if { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && {
+	 test -z "$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest$ac_exeext && {
+	 test "$cross_compiling" = yes ||
+	 test -x conftest$ac_exeext
+       }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_retval=1
+fi
+  # Delete the IPA/IPO (Inter Procedural Analysis/Optimization) information
+  # created by the PGI compiler (conftest_ipa8_conftest.oo), as it would
+  # interfere with the next link command; also delete a directory that is
+  # left behind by Apple's compiler.  We do this before executing the actions.
+  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_cxx_try_link
+
+# ac_fn_cxx_try_cpp LINENO
+# ------------------------
+# Try to preprocess conftest.$ac_ext, and return whether this succeeded.
+ac_fn_cxx_try_cpp ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if { { ac_try="$ac_cpp conftest.$ac_ext"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } > conftest.i && {
+	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
+	 test ! -s conftest.err
+       }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+    ac_retval=1
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_cxx_try_cpp
+
+# ac_fn_cxx_check_header_mongrel LINENO HEADER VAR INCLUDES
+# ---------------------------------------------------------
+# Tests whether HEADER exists, giving a warning if it cannot be compiled using
+# the include files in INCLUDES and setting the cache variable VAR
+# accordingly.
+ac_fn_cxx_check_header_mongrel ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if eval \${$3+:} false; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+else
+  # Is the header compilable?
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 usability" >&5
+$as_echo_n "checking $2 usability... " >&6; }
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+#include <$2>
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  ac_header_compiler=yes
+else
+  ac_header_compiler=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_compiler" >&5
+$as_echo "$ac_header_compiler" >&6; }
+
+# Is the header present?
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 presence" >&5
+$as_echo_n "checking $2 presence... " >&6; }
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <$2>
+_ACEOF
+if ac_fn_cxx_try_cpp "$LINENO"; then :
+  ac_header_preproc=yes
+else
+  ac_header_preproc=no
+fi
+rm -f conftest.err conftest.i conftest.$ac_ext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_preproc" >&5
+$as_echo "$ac_header_preproc" >&6; }
+
+# So?  What about this header?
+case $ac_header_compiler:$ac_header_preproc:$ac_cxx_preproc_warn_flag in #((
+  yes:no: )
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&5
+$as_echo "$as_me: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
+    ;;
+  no:yes:* )
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: present but cannot be compiled" >&5
+$as_echo "$as_me: WARNING: $2: present but cannot be compiled" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     check for missing prerequisite headers?" >&5
+$as_echo "$as_me: WARNING: $2:     check for missing prerequisite headers?" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: see the Autoconf documentation" >&5
+$as_echo "$as_me: WARNING: $2: see the Autoconf documentation" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&5
+$as_echo "$as_me: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&2;}
+    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
+$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
+( $as_echo "## ---------------------------------- ##
+## Report this to paboyle@ph.ed.ac.uk ##
+## ---------------------------------- ##"
+     ) | sed "s/^/$as_me: WARNING:     /" >&2
+    ;;
+esac
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  eval "$3=\$ac_header_compiler"
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_cxx_check_header_mongrel
+
+# ac_fn_cxx_try_run LINENO
+# ------------------------
+# Try to link conftest.$ac_ext, and return whether this succeeded. Assumes
+# that executables *can* be run.
+ac_fn_cxx_try_run ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && { ac_try='./conftest$ac_exeext'
+  { { case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: program exited with status $ac_status" >&5
+       $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+       ac_retval=$ac_status
+fi
+  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_cxx_try_run
+
+# ac_fn_cxx_check_header_compile LINENO HEADER VAR INCLUDES
+# ---------------------------------------------------------
+# Tests whether HEADER exists and can be compiled using the include files in
+# INCLUDES, setting the cache variable VAR accordingly.
+ac_fn_cxx_check_header_compile ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+#include <$2>
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  eval "$3=yes"
+else
+  eval "$3=no"
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_cxx_check_header_compile
+
+# ac_fn_cxx_check_decl LINENO SYMBOL VAR INCLUDES
+# -----------------------------------------------
+# Tests whether SYMBOL is declared in INCLUDES, setting cache variable VAR
+# accordingly.
+ac_fn_cxx_check_decl ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  as_decl_name=`echo $2|sed 's/ *(.*//'`
+  as_decl_use=`echo $2|sed -e 's/(/((/' -e 's/)/) 0&/' -e 's/,/) 0& (/g'`
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $as_decl_name is declared" >&5
+$as_echo_n "checking whether $as_decl_name is declared... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+#ifndef $as_decl_name
+#ifdef __cplusplus
+  (void) $as_decl_use;
+#else
+  (void) $as_decl_name;
+#endif
+#endif
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  eval "$3=yes"
+else
+  eval "$3=no"
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_cxx_check_decl
+
+# ac_fn_cxx_check_type LINENO TYPE VAR INCLUDES
+# ---------------------------------------------
+# Tests whether TYPE exists after having included INCLUDES, setting cache
+# variable VAR accordingly.
+ac_fn_cxx_check_type ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  eval "$3=no"
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+if (sizeof ($2))
+	 return 0;
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$4
+int
+main ()
+{
+if (sizeof (($2)))
+	    return 0;
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+
+else
+  eval "$3=yes"
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_cxx_check_type
+
+# ac_fn_c_find_uintX_t LINENO BITS VAR
+# ------------------------------------
+# Finds an unsigned integer type with width BITS, setting cache variable VAR
+# accordingly.
+ac_fn_c_find_uintX_t ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for uint$2_t" >&5
+$as_echo_n "checking for uint$2_t... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  eval "$3=no"
+     # Order is important - never check a type that is potentially smaller
+     # than half of the expected target width.
+     for ac_type in uint$2_t 'unsigned int' 'unsigned long int' \
+	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
+       cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+$ac_includes_default
+int
+main ()
+{
+static int test_array [1 - 2 * !((($ac_type) -1 >> ($2 / 2 - 1)) >> ($2 / 2 - 1) == 3)];
+test_array [0] = 0;
+return test_array [0];
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  case $ac_type in #(
+  uint$2_t) :
+    eval "$3=yes" ;; #(
+  *) :
+    eval "$3=\$ac_type" ;;
+esac
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+       if eval test \"x\$"$3"\" = x"no"; then :
+
+else
+  break
+fi
+     done
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_c_find_uintX_t
+
+# ac_fn_cxx_check_func LINENO FUNC VAR
+# ------------------------------------
+# Tests whether FUNC exists, setting the cache variable VAR accordingly
+ac_fn_cxx_check_func ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
+$as_echo_n "checking for $2... " >&6; }
+if eval \${$3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+/* Define $2 to an innocuous variant, in case <limits.h> declares $2.
+   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
+#define $2 innocuous_$2
+
+/* System header to define __stub macros and hopefully few prototypes,
+    which can conflict with char $2 (); below.
+    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
+    <limits.h> exists even on freestanding compilers.  */
+
+#ifdef __STDC__
+# include <limits.h>
+#else
+# include <assert.h>
+#endif
+
+#undef $2
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char $2 ();
+/* The GNU C library defines this for functions which it implements
+    to always fail with ENOSYS.  Some functions are actually named
+    something starting with __ and the normal name is an alias.  */
+#if defined __stub_$2 || defined __stub___$2
+choke me
+#endif
+
+int
+main ()
+{
+return $2 ();
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_link "$LINENO"; then :
+  eval "$3=yes"
+else
+  eval "$3=no"
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+fi
+eval ac_res=\$$3
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
+$as_echo "$ac_res" >&6; }
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+
+} # ac_fn_cxx_check_func
 cat >config.log <<_ACEOF
 This file contains any messages produced by compilers while
 running configure, to aid debugging if configure makes a mistake.
 
 It was created by Grid $as_me 1.0, which was
-generated by GNU Autoconf 2.63.  Invocation command line was
+generated by GNU Autoconf 2.69.  Invocation command line was
 
   $ $0 $@
 
@@ -1544,8 +2032,8 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  $as_echo "PATH: $as_dir"
-done
+    $as_echo "PATH: $as_dir"
+  done
 IFS=$as_save_IFS
 
 } >&5
@@ -1582,9 +2070,9 @@ do
       ac_arg=`$as_echo "$ac_arg" | sed "s/'/'\\\\\\\\''/g"` ;;
     esac
     case $ac_pass in
-    1) ac_configure_args0="$ac_configure_args0 '$ac_arg'" ;;
+    1) as_fn_append ac_configure_args0 " '$ac_arg'" ;;
     2)
-      ac_configure_args1="$ac_configure_args1 '$ac_arg'"
+      as_fn_append ac_configure_args1 " '$ac_arg'"
       if test $ac_must_keep_next = true; then
 	ac_must_keep_next=false # Got value, back to normal.
       else
@@ -1600,13 +2088,13 @@ do
 	  -* ) ac_must_keep_next=true ;;
 	esac
       fi
-      ac_configure_args="$ac_configure_args '$ac_arg'"
+      as_fn_append ac_configure_args " '$ac_arg'"
       ;;
     esac
   done
 done
-$as_unset ac_configure_args0 || test "${ac_configure_args0+set}" != set || { ac_configure_args0=; export ac_configure_args0; }
-$as_unset ac_configure_args1 || test "${ac_configure_args1+set}" != set || { ac_configure_args1=; export ac_configure_args1; }
+{ ac_configure_args0=; unset ac_configure_args0;}
+{ ac_configure_args1=; unset ac_configure_args1;}
 
 # When interrupted or exit'd, cleanup temporary files, and complete
 # config.log.  We remove comments because anyway the quotes in there
@@ -1618,11 +2106,9 @@ trap 'exit_status=$?
   {
     echo
 
-    cat <<\_ASBOX
-## ---------------- ##
+    $as_echo "## ---------------- ##
 ## Cache variables. ##
-## ---------------- ##
-_ASBOX
+## ---------------- ##"
     echo
     # The following way of writing the cache mishandles newlines in values,
 (
@@ -1631,13 +2117,13 @@ _ASBOX
     case $ac_val in #(
     *${as_nl}*)
       case $ac_var in #(
-      *_cv_*) { $as_echo "$as_me:$LINENO: WARNING: cache variable $ac_var contains a newline" >&5
+      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
 $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
       esac
       case $ac_var in #(
       _ | IFS | as_nl) ;; #(
       BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
-      *) $as_unset $ac_var ;;
+      *) { eval $ac_var=; unset $ac_var;} ;;
       esac ;;
     esac
   done
@@ -1656,11 +2142,9 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
 )
     echo
 
-    cat <<\_ASBOX
-## ----------------- ##
+    $as_echo "## ----------------- ##
 ## Output variables. ##
-## ----------------- ##
-_ASBOX
+## ----------------- ##"
     echo
     for ac_var in $ac_subst_vars
     do
@@ -1673,11 +2157,9 @@ _ASBOX
     echo
 
     if test -n "$ac_subst_files"; then
-      cat <<\_ASBOX
-## ------------------- ##
+      $as_echo "## ------------------- ##
 ## File substitutions. ##
-## ------------------- ##
-_ASBOX
+## ------------------- ##"
       echo
       for ac_var in $ac_subst_files
       do
@@ -1691,11 +2173,9 @@ _ASBOX
     fi
 
     if test -s confdefs.h; then
-      cat <<\_ASBOX
-## ----------- ##
+      $as_echo "## ----------- ##
 ## confdefs.h. ##
-## ----------- ##
-_ASBOX
+## ----------- ##"
       echo
       cat confdefs.h
       echo
@@ -1709,46 +2189,53 @@ _ASBOX
     exit $exit_status
 ' 0
 for ac_signal in 1 2 13 15; do
-  trap 'ac_signal='$ac_signal'; { (exit 1); exit 1; }' $ac_signal
+  trap 'ac_signal='$ac_signal'; as_fn_exit 1' $ac_signal
 done
 ac_signal=0
 
 # confdefs.h avoids OS command line length limits that DEFS can exceed.
 rm -f -r conftest* confdefs.h
 
+$as_echo "/* confdefs.h */" > confdefs.h
+
 # Predefined preprocessor variables.
 
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_NAME "$PACKAGE_NAME"
 _ACEOF
 
-
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_TARNAME "$PACKAGE_TARNAME"
 _ACEOF
 
-
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_VERSION "$PACKAGE_VERSION"
 _ACEOF
 
-
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_STRING "$PACKAGE_STRING"
 _ACEOF
 
-
 cat >>confdefs.h <<_ACEOF
 #define PACKAGE_BUGREPORT "$PACKAGE_BUGREPORT"
 _ACEOF
 
+cat >>confdefs.h <<_ACEOF
+#define PACKAGE_URL "$PACKAGE_URL"
+_ACEOF
+
 
 # Let the site file select an alternate cache file if it wants to.
 # Prefer an explicitly selected file to automatically selected ones.
 ac_site_file1=NONE
 ac_site_file2=NONE
 if test -n "$CONFIG_SITE"; then
-  ac_site_file1=$CONFIG_SITE
+  # We do not want a PATH search for config.site.
+  case $CONFIG_SITE in #((
+    -*)  ac_site_file1=./$CONFIG_SITE;;
+    */*) ac_site_file1=$CONFIG_SITE;;
+    *)   ac_site_file1=./$CONFIG_SITE;;
+  esac
 elif test "x$prefix" != xNONE; then
   ac_site_file1=$prefix/share/config.site
   ac_site_file2=$prefix/etc/config.site
@@ -1759,19 +2246,23 @@ fi
 for ac_site_file in "$ac_site_file1" "$ac_site_file2"
 do
   test "x$ac_site_file" = xNONE && continue
-  if test -r "$ac_site_file"; then
-    { $as_echo "$as_me:$LINENO: loading site script $ac_site_file" >&5
+  if test /dev/null != "$ac_site_file" && test -r "$ac_site_file"; then
+    { $as_echo "$as_me:${as_lineno-$LINENO}: loading site script $ac_site_file" >&5
 $as_echo "$as_me: loading site script $ac_site_file" >&6;}
     sed 's/^/| /' "$ac_site_file" >&5
-    . "$ac_site_file"
+    . "$ac_site_file" \
+      || { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "failed to load site script $ac_site_file
+See \`config.log' for more details" "$LINENO" 5; }
   fi
 done
 
 if test -r "$cache_file"; then
-  # Some versions of bash will fail to source /dev/null (special
-  # files actually), so we avoid doing that.
-  if test -f "$cache_file"; then
-    { $as_echo "$as_me:$LINENO: loading cache $cache_file" >&5
+  # Some versions of bash will fail to source /dev/null (special files
+  # actually), so we avoid doing that.  DJGPP emulates it as a regular file.
+  if test /dev/null != "$cache_file" && test -f "$cache_file"; then
+    { $as_echo "$as_me:${as_lineno-$LINENO}: loading cache $cache_file" >&5
 $as_echo "$as_me: loading cache $cache_file" >&6;}
     case $cache_file in
       [\\/]* | ?:[\\/]* ) . "$cache_file";;
@@ -1779,7 +2270,7 @@ $as_echo "$as_me: loading cache $cache_file" >&6;}
     esac
   fi
 else
-  { $as_echo "$as_me:$LINENO: creating cache $cache_file" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: creating cache $cache_file" >&5
 $as_echo "$as_me: creating cache $cache_file" >&6;}
   >$cache_file
 fi
@@ -1794,11 +2285,11 @@ for ac_var in $ac_precious_vars; do
   eval ac_new_val=\$ac_env_${ac_var}_value
   case $ac_old_set,$ac_new_set in
     set,)
-      { $as_echo "$as_me:$LINENO: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
+      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
 $as_echo "$as_me: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&2;}
       ac_cache_corrupted=: ;;
     ,set)
-      { $as_echo "$as_me:$LINENO: error: \`$ac_var' was not set in the previous run" >&5
+      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was not set in the previous run" >&5
 $as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
       ac_cache_corrupted=: ;;
     ,);;
@@ -1808,17 +2299,17 @@ $as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
 	ac_old_val_w=`echo x $ac_old_val`
 	ac_new_val_w=`echo x $ac_new_val`
 	if test "$ac_old_val_w" != "$ac_new_val_w"; then
-	  { $as_echo "$as_me:$LINENO: error: \`$ac_var' has changed since the previous run:" >&5
+	  { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' has changed since the previous run:" >&5
 $as_echo "$as_me: error: \`$ac_var' has changed since the previous run:" >&2;}
 	  ac_cache_corrupted=:
 	else
-	  { $as_echo "$as_me:$LINENO: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
+	  { $as_echo "$as_me:${as_lineno-$LINENO}: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
 $as_echo "$as_me: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&2;}
 	  eval $ac_var=\$ac_old_val
 	fi
-	{ $as_echo "$as_me:$LINENO:   former value:  \`$ac_old_val'" >&5
+	{ $as_echo "$as_me:${as_lineno-$LINENO}:   former value:  \`$ac_old_val'" >&5
 $as_echo "$as_me:   former value:  \`$ac_old_val'" >&2;}
-	{ $as_echo "$as_me:$LINENO:   current value: \`$ac_new_val'" >&5
+	{ $as_echo "$as_me:${as_lineno-$LINENO}:   current value: \`$ac_new_val'" >&5
 $as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
       fi;;
   esac
@@ -1830,43 +2321,20 @@ $as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
     esac
     case " $ac_configure_args " in
       *" '$ac_arg' "*) ;; # Avoid dups.  Use of quotes ensures accuracy.
-      *) ac_configure_args="$ac_configure_args '$ac_arg'" ;;
+      *) as_fn_append ac_configure_args " '$ac_arg'" ;;
     esac
   fi
 done
 if $ac_cache_corrupted; then
-  { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-  { $as_echo "$as_me:$LINENO: error: changes in the environment can compromise the build" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: error: changes in the environment can compromise the build" >&5
 $as_echo "$as_me: error: changes in the environment can compromise the build" >&2;}
-  { { $as_echo "$as_me:$LINENO: error: run \`make distclean' and/or \`rm $cache_file' and start over" >&5
-$as_echo "$as_me: error: run \`make distclean' and/or \`rm $cache_file' and start over" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "run \`make distclean' and/or \`rm $cache_file' and start over" "$LINENO" 5
 fi
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+## -------------------- ##
+## Main body of script. ##
+## -------------------- ##
 
 ac_ext=c
 ac_cpp='$CPP $CPPFLAGS'
@@ -1892,9 +2360,7 @@ for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
   fi
 done
 if test -z "$ac_aux_dir"; then
-  { { $as_echo "$as_me:$LINENO: error: cannot find install-sh or install.sh in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" >&5
-$as_echo "$as_me: error: cannot find install-sh or install.sh in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "cannot find install-sh, install.sh, or shtool in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" "$LINENO" 5
 fi
 
 # These three variables are undocumented and unsupported,
@@ -1908,35 +2374,27 @@ ac_configure="$SHELL $ac_aux_dir/configure"  # Please don't use this var.
 
 # Make sure we can run config.sub.
 $SHELL "$ac_aux_dir/config.sub" sun4 >/dev/null 2>&1 ||
-  { { $as_echo "$as_me:$LINENO: error: cannot run $SHELL $ac_aux_dir/config.sub" >&5
-$as_echo "$as_me: error: cannot run $SHELL $ac_aux_dir/config.sub" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "cannot run $SHELL $ac_aux_dir/config.sub" "$LINENO" 5
 
-{ $as_echo "$as_me:$LINENO: checking build system type" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking build system type" >&5
 $as_echo_n "checking build system type... " >&6; }
-if test "${ac_cv_build+set}" = set; then
+if ${ac_cv_build+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   ac_build_alias=$build_alias
 test "x$ac_build_alias" = x &&
   ac_build_alias=`$SHELL "$ac_aux_dir/config.guess"`
 test "x$ac_build_alias" = x &&
-  { { $as_echo "$as_me:$LINENO: error: cannot guess build type; you must specify one" >&5
-$as_echo "$as_me: error: cannot guess build type; you must specify one" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "cannot guess build type; you must specify one" "$LINENO" 5
 ac_cv_build=`$SHELL "$ac_aux_dir/config.sub" $ac_build_alias` ||
-  { { $as_echo "$as_me:$LINENO: error: $SHELL $ac_aux_dir/config.sub $ac_build_alias failed" >&5
-$as_echo "$as_me: error: $SHELL $ac_aux_dir/config.sub $ac_build_alias failed" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "$SHELL $ac_aux_dir/config.sub $ac_build_alias failed" "$LINENO" 5
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_build" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_build" >&5
 $as_echo "$ac_cv_build" >&6; }
 case $ac_cv_build in
 *-*-*) ;;
-*) { { $as_echo "$as_me:$LINENO: error: invalid value of canonical build" >&5
-$as_echo "$as_me: error: invalid value of canonical build" >&2;}
-   { (exit 1); exit 1; }; };;
+*) as_fn_error $? "invalid value of canonical build" "$LINENO" 5;;
 esac
 build=$ac_cv_build
 ac_save_IFS=$IFS; IFS='-'
@@ -1952,28 +2410,24 @@ IFS=$ac_save_IFS
 case $build_os in *\ *) build_os=`echo "$build_os" | sed 's/ /-/g'`;; esac
 
 
-{ $as_echo "$as_me:$LINENO: checking host system type" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking host system type" >&5
 $as_echo_n "checking host system type... " >&6; }
-if test "${ac_cv_host+set}" = set; then
+if ${ac_cv_host+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test "x$host_alias" = x; then
   ac_cv_host=$ac_cv_build
 else
   ac_cv_host=`$SHELL "$ac_aux_dir/config.sub" $host_alias` ||
-    { { $as_echo "$as_me:$LINENO: error: $SHELL $ac_aux_dir/config.sub $host_alias failed" >&5
-$as_echo "$as_me: error: $SHELL $ac_aux_dir/config.sub $host_alias failed" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "$SHELL $ac_aux_dir/config.sub $host_alias failed" "$LINENO" 5
 fi
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_host" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_host" >&5
 $as_echo "$ac_cv_host" >&6; }
 case $ac_cv_host in
 *-*-*) ;;
-*) { { $as_echo "$as_me:$LINENO: error: invalid value of canonical host" >&5
-$as_echo "$as_me: error: invalid value of canonical host" >&2;}
-   { (exit 1); exit 1; }; };;
+*) as_fn_error $? "invalid value of canonical host" "$LINENO" 5;;
 esac
 host=$ac_cv_host
 ac_save_IFS=$IFS; IFS='-'
@@ -1989,28 +2443,24 @@ IFS=$ac_save_IFS
 case $host_os in *\ *) host_os=`echo "$host_os" | sed 's/ /-/g'`;; esac
 
 
-{ $as_echo "$as_me:$LINENO: checking target system type" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking target system type" >&5
 $as_echo_n "checking target system type... " >&6; }
-if test "${ac_cv_target+set}" = set; then
+if ${ac_cv_target+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test "x$target_alias" = x; then
   ac_cv_target=$ac_cv_host
 else
   ac_cv_target=`$SHELL "$ac_aux_dir/config.sub" $target_alias` ||
-    { { $as_echo "$as_me:$LINENO: error: $SHELL $ac_aux_dir/config.sub $target_alias failed" >&5
-$as_echo "$as_me: error: $SHELL $ac_aux_dir/config.sub $target_alias failed" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "$SHELL $ac_aux_dir/config.sub $target_alias failed" "$LINENO" 5
 fi
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_target" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_target" >&5
 $as_echo "$ac_cv_target" >&6; }
 case $ac_cv_target in
 *-*-*) ;;
-*) { { $as_echo "$as_me:$LINENO: error: invalid value of canonical target" >&5
-$as_echo "$as_me: error: invalid value of canonical target" >&2;}
-   { (exit 1); exit 1; }; };;
+*) as_fn_error $? "invalid value of canonical target" "$LINENO" 5;;
 esac
 target=$ac_cv_target
 ac_save_IFS=$IFS; IFS='-'
@@ -2032,7 +2482,8 @@ test -n "$target_alias" &&
   test "$program_prefix$program_suffix$program_transform_name" = \
     NONENONEs,x,x, &&
   program_prefix=${target_alias}-
-am__api_version='1.11'
+
+am__api_version='1.14'
 
 # Find a good install program.  We prefer a C program (faster),
 # so one script is as good as another.  But avoid the broken or
@@ -2048,10 +2499,10 @@ am__api_version='1.11'
 # OS/2's system install, which has a completely different semantic
 # ./install, which can be erroneously created by make from ./install.sh.
 # Reject install programs that cannot install multiple files.
-{ $as_echo "$as_me:$LINENO: checking for a BSD-compatible install" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a BSD-compatible install" >&5
 $as_echo_n "checking for a BSD-compatible install... " >&6; }
 if test -z "$INSTALL"; then
-if test "${ac_cv_path_install+set}" = set; then
+if ${ac_cv_path_install+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
@@ -2059,11 +2510,11 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  # Account for people who put trailing slashes in PATH elements.
-case $as_dir/ in
-  ./ | .// | /cC/* | \
+    # Account for people who put trailing slashes in PATH elements.
+case $as_dir/ in #((
+  ./ | .// | /[cC]/* | \
   /etc/* | /usr/sbin/* | /usr/etc/* | /sbin/* | /usr/afsws/bin/* | \
-  ?:\\/os2\\/install\\/* | ?:\\/OS2\\/INSTALL\\/* | \
+  ?:[\\/]os2[\\/]install[\\/]* | ?:[\\/]OS2[\\/]INSTALL[\\/]* | \
   /usr/ucb/* ) ;;
   *)
     # OSF1 and SCO ODT 3.0 have their own names for install.
@@ -2071,7 +2522,7 @@ case $as_dir/ in
     # by default.
     for ac_prog in ginstall scoinst install; do
       for ac_exec_ext in '' $ac_executable_extensions; do
-	if { test -f "$as_dir/$ac_prog$ac_exec_ext" && $as_test_x "$as_dir/$ac_prog$ac_exec_ext"; }; then
+	if as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext"; then
 	  if test $ac_prog = install &&
 	    grep dspmsg "$as_dir/$ac_prog$ac_exec_ext" >/dev/null 2>&1; then
 	    # AIX install.  It has an incompatible calling convention.
@@ -2100,7 +2551,7 @@ case $as_dir/ in
     ;;
 esac
 
-done
+  done
 IFS=$as_save_IFS
 
 rm -rf conftest.one conftest.two conftest.dir
@@ -2116,7 +2567,7 @@ fi
     INSTALL=$ac_install_sh
   fi
 fi
-{ $as_echo "$as_me:$LINENO: result: $INSTALL" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $INSTALL" >&5
 $as_echo "$INSTALL" >&6; }
 
 # Use test -z because SunOS4 sh mishandles braces in ${var-val}.
@@ -2127,68 +2578,73 @@ test -z "$INSTALL_SCRIPT" && INSTALL_SCRIPT='${INSTALL}'
 
 test -z "$INSTALL_DATA" && INSTALL_DATA='${INSTALL} -m 644'
 
-{ $as_echo "$as_me:$LINENO: checking whether build environment is sane" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether build environment is sane" >&5
 $as_echo_n "checking whether build environment is sane... " >&6; }
-# Just in case
-sleep 1
-echo timestamp > conftest.file
 # Reject unsafe characters in $srcdir or the absolute working directory
 # name.  Accept space and tab only in the latter.
 am_lf='
 '
 case `pwd` in
   *[\\\"\#\$\&\'\`$am_lf]*)
-    { { $as_echo "$as_me:$LINENO: error: unsafe absolute working directory name" >&5
-$as_echo "$as_me: error: unsafe absolute working directory name" >&2;}
-   { (exit 1); exit 1; }; };;
+    as_fn_error $? "unsafe absolute working directory name" "$LINENO" 5;;
 esac
 case $srcdir in
   *[\\\"\#\$\&\'\`$am_lf\ \	]*)
-    { { $as_echo "$as_me:$LINENO: error: unsafe srcdir value: \`$srcdir'" >&5
-$as_echo "$as_me: error: unsafe srcdir value: \`$srcdir'" >&2;}
-   { (exit 1); exit 1; }; };;
+    as_fn_error $? "unsafe srcdir value: '$srcdir'" "$LINENO" 5;;
 esac
 
-# Do `set' in a subshell so we don't clobber the current shell's
+# Do 'set' in a subshell so we don't clobber the current shell's
 # arguments.  Must try -L first in case configure is actually a
 # symlink; some systems play weird games with the mod time of symlinks
 # (eg FreeBSD returns the mod time of the symlink's containing
 # directory).
 if (
-   set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
-   if test "$*" = "X"; then
-      # -L didn't work.
-      set X `ls -t "$srcdir/configure" conftest.file`
-   fi
-   rm -f conftest.file
-   if test "$*" != "X $srcdir/configure conftest.file" \
-      && test "$*" != "X conftest.file $srcdir/configure"; then
-
-      # If neither matched, then we have a broken ls.  This can happen
-      # if, for instance, CONFIG_SHELL is bash and it inherits a
-      # broken ls alias from the environment.  This has actually
-      # happened.  Such a system could not be considered "sane".
-      { { $as_echo "$as_me:$LINENO: error: ls -t appears to fail.  Make sure there is not a broken
-alias in your environment" >&5
-$as_echo "$as_me: error: ls -t appears to fail.  Make sure there is not a broken
-alias in your environment" >&2;}
-   { (exit 1); exit 1; }; }
-   fi
+   am_has_slept=no
+   for am_try in 1 2; do
+     echo "timestamp, slept: $am_has_slept" > conftest.file
+     set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
+     if test "$*" = "X"; then
+	# -L didn't work.
+	set X `ls -t "$srcdir/configure" conftest.file`
+     fi
+     if test "$*" != "X $srcdir/configure conftest.file" \
+	&& test "$*" != "X conftest.file $srcdir/configure"; then
 
+	# If neither matched, then we have a broken ls.  This can happen
+	# if, for instance, CONFIG_SHELL is bash and it inherits a
+	# broken ls alias from the environment.  This has actually
+	# happened.  Such a system could not be considered "sane".
+	as_fn_error $? "ls -t appears to fail.  Make sure there is not a broken
+  alias in your environment" "$LINENO" 5
+     fi
+     if test "$2" = conftest.file || test $am_try -eq 2; then
+       break
+     fi
+     # Just in case.
+     sleep 1
+     am_has_slept=yes
+   done
    test "$2" = conftest.file
    )
 then
    # Ok.
    :
 else
-   { { $as_echo "$as_me:$LINENO: error: newly created file is older than distributed files!
-Check your system clock" >&5
-$as_echo "$as_me: error: newly created file is older than distributed files!
-Check your system clock" >&2;}
-   { (exit 1); exit 1; }; }
+   as_fn_error $? "newly created file is older than distributed files!
+Check your system clock" "$LINENO" 5
 fi
-{ $as_echo "$as_me:$LINENO: result: yes" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
 $as_echo "yes" >&6; }
+# If we didn't sleep, we still need to ensure time stamps of config.status and
+# generated files are strictly newer.
+am_sleep_pid=
+if grep 'slept: no' conftest.file >/dev/null 2>&1; then
+  ( sleep 1 ) &
+  am_sleep_pid=$!
+fi
+
+rm -f conftest.file
+
 test "$program_prefix" != NONE &&
   program_transform_name="s&^&$program_prefix&;$program_transform_name"
 # Use a double $ so make ignores it.
@@ -2211,12 +2667,12 @@ if test x"${MISSING+set}" != xset; then
   esac
 fi
 # Use eval to expand $SHELL
-if eval "$MISSING --run true"; then
-  am_missing_run="$MISSING --run "
+if eval "$MISSING --is-lightweight"; then
+  am_missing_run="$MISSING "
 else
   am_missing_run=
-  { $as_echo "$as_me:$LINENO: WARNING: \`missing' script is too old or missing" >&5
-$as_echo "$as_me: WARNING: \`missing' script is too old or missing" >&2;}
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: 'missing' script is too old or missing" >&5
+$as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
 fi
 
 if test x"${install_sh}" != xset; then
@@ -2228,17 +2684,17 @@ if test x"${install_sh}" != xset; then
   esac
 fi
 
-# Installed binaries are usually stripped using `strip' when the user
-# run `make install-strip'.  However `strip' might not be the right
+# Installed binaries are usually stripped using 'strip' when the user
+# run "make install-strip".  However 'strip' might not be the right
 # tool to use in cross-compilation environments, therefore Automake
-# will honor the `STRIP' environment variable to overrule this program.
+# will honor the 'STRIP' environment variable to overrule this program.
 if test "$cross_compiling" != no; then
   if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}strip", so it can be a program name with args.
 set dummy ${ac_tool_prefix}strip; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_STRIP+set}" = set; then
+if ${ac_cv_prog_STRIP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$STRIP"; then
@@ -2249,24 +2705,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_STRIP="${ac_tool_prefix}strip"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 STRIP=$ac_cv_prog_STRIP
 if test -n "$STRIP"; then
-  { $as_echo "$as_me:$LINENO: result: $STRIP" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $STRIP" >&5
 $as_echo "$STRIP" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2276,9 +2732,9 @@ if test -z "$ac_cv_prog_STRIP"; then
   ac_ct_STRIP=$STRIP
   # Extract the first word of "strip", so it can be a program name with args.
 set dummy strip; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_ac_ct_STRIP+set}" = set; then
+if ${ac_cv_prog_ac_ct_STRIP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_STRIP"; then
@@ -2289,24 +2745,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_ac_ct_STRIP="strip"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_STRIP=$ac_cv_prog_ac_ct_STRIP
 if test -n "$ac_ct_STRIP"; then
-  { $as_echo "$as_me:$LINENO: result: $ac_ct_STRIP" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_STRIP" >&5
 $as_echo "$ac_ct_STRIP" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2315,7 +2771,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -2328,10 +2784,10 @@ fi
 fi
 INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
 
-{ $as_echo "$as_me:$LINENO: checking for a thread-safe mkdir -p" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a thread-safe mkdir -p" >&5
 $as_echo_n "checking for a thread-safe mkdir -p... " >&6; }
 if test -z "$MKDIR_P"; then
-  if test "${ac_cv_path_mkdir+set}" = set; then
+  if ${ac_cv_path_mkdir+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
@@ -2339,9 +2795,9 @@ for as_dir in $PATH$PATH_SEPARATOR/opt/sfw/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_prog in mkdir gmkdir; do
+    for ac_prog in mkdir gmkdir; do
 	 for ac_exec_ext in '' $ac_executable_extensions; do
-	   { test -f "$as_dir/$ac_prog$ac_exec_ext" && $as_test_x "$as_dir/$ac_prog$ac_exec_ext"; } || continue
+	   as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext" || continue
 	   case `"$as_dir/$ac_prog$ac_exec_ext" --version 2>&1` in #(
 	     'mkdir (GNU coreutils) '* | \
 	     'mkdir (coreutils) '* | \
@@ -2351,11 +2807,12 @@ do
 	   esac
 	 done
        done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 
+  test -d ./--version && rmdir ./--version
   if test "${ac_cv_path_mkdir+set}" = set; then
     MKDIR_P="$ac_cv_path_mkdir -p"
   else
@@ -2363,26 +2820,19 @@ fi
     # value for MKDIR_P within a source directory, because that will
     # break other packages using the cache if that directory is
     # removed, or if the value is a relative name.
-    test -d ./--version && rmdir ./--version
     MKDIR_P="$ac_install_sh -d"
   fi
 fi
-{ $as_echo "$as_me:$LINENO: result: $MKDIR_P" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $MKDIR_P" >&5
 $as_echo "$MKDIR_P" >&6; }
 
-mkdir_p="$MKDIR_P"
-case $mkdir_p in
-  [\\/$]* | ?:[\\/]*) ;;
-  */*) mkdir_p="\$(top_builddir)/$mkdir_p" ;;
-esac
-
 for ac_prog in gawk mawk nawk awk
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_AWK+set}" = set; then
+if ${ac_cv_prog_AWK+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$AWK"; then
@@ -2393,24 +2843,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_AWK="$ac_prog"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 AWK=$ac_cv_prog_AWK
 if test -n "$AWK"; then
-  { $as_echo "$as_me:$LINENO: result: $AWK" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $AWK" >&5
 $as_echo "$AWK" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2418,11 +2868,11 @@ fi
   test -n "$AWK" && break
 done
 
-{ $as_echo "$as_me:$LINENO: checking whether ${MAKE-make} sets \$(MAKE)" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether ${MAKE-make} sets \$(MAKE)" >&5
 $as_echo_n "checking whether ${MAKE-make} sets \$(MAKE)... " >&6; }
 set x ${MAKE-make}
 ac_make=`$as_echo "$2" | sed 's/+/p/g; s/[^a-zA-Z0-9_]/_/g'`
-if { as_var=ac_cv_prog_make_${ac_make}_set; eval "test \"\${$as_var+set}\" = set"; }; then
+if eval \${ac_cv_prog_make_${ac_make}_set+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   cat >conftest.make <<\_ACEOF
@@ -2430,7 +2880,7 @@ SHELL = /bin/sh
 all:
 	@echo '@@@%%%=$(MAKE)=@@@%%%'
 _ACEOF
-# GNU make sometimes prints "make[1]: Entering...", which would confuse us.
+# GNU make sometimes prints "make[1]: Entering ...", which would confuse us.
 case `${MAKE-make} -f conftest.make 2>/dev/null` in
   *@@@%%%=?*=@@@%%%*)
     eval ac_cv_prog_make_${ac_make}_set=yes;;
@@ -2440,11 +2890,11 @@ esac
 rm -f conftest.make
 fi
 if eval test \$ac_cv_prog_make_${ac_make}_set = yes; then
-  { $as_echo "$as_me:$LINENO: result: yes" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
 $as_echo "yes" >&6; }
   SET_MAKE=
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
   SET_MAKE="MAKE=${MAKE-make}"
 fi
@@ -2458,15 +2908,52 @@ else
 fi
 rmdir .tst 2>/dev/null
 
+# Check whether --enable-silent-rules was given.
+if test "${enable_silent_rules+set}" = set; then :
+  enableval=$enable_silent_rules;
+fi
+
+case $enable_silent_rules in # (((
+  yes) AM_DEFAULT_VERBOSITY=0;;
+   no) AM_DEFAULT_VERBOSITY=1;;
+    *) AM_DEFAULT_VERBOSITY=1;;
+esac
+am_make=${MAKE-make}
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $am_make supports nested variables" >&5
+$as_echo_n "checking whether $am_make supports nested variables... " >&6; }
+if ${am_cv_make_support_nested_variables+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if $as_echo 'TRUE=$(BAR$(V))
+BAR0=false
+BAR1=true
+V=1
+am__doit:
+	@$(TRUE)
+.PHONY: am__doit' | $am_make -f - >/dev/null 2>&1; then
+  am_cv_make_support_nested_variables=yes
+else
+  am_cv_make_support_nested_variables=no
+fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_make_support_nested_variables" >&5
+$as_echo "$am_cv_make_support_nested_variables" >&6; }
+if test $am_cv_make_support_nested_variables = yes; then
+    AM_V='$(V)'
+  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
+else
+  AM_V=$AM_DEFAULT_VERBOSITY
+  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
+fi
+AM_BACKSLASH='\'
+
 if test "`cd $srcdir && pwd`" != "`pwd`"; then
   # Use -I$(srcdir) only when $(srcdir) != ., so that make's output
   # is not polluted with repeated "-I."
   am__isrc=' -I$(srcdir)'
   # test to see if srcdir already configured
   if test -f $srcdir/config.status; then
-    { { $as_echo "$as_me:$LINENO: error: source directory already configured; run \"make distclean\" there first" >&5
-$as_echo "$as_me: error: source directory already configured; run \"make distclean\" there first" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "source directory already configured; run \"make distclean\" there first" "$LINENO" 5
   fi
 fi
 
@@ -2510,36 +2997,115 @@ AUTOHEADER=${AUTOHEADER-"${am_missing_run}autoheader"}
 
 MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 
+# For better backward compatibility.  To be removed once Automake 1.9.x
+# dies out for good.  For more background, see:
+# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
+# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
+mkdir_p='$(MKDIR_P)'
+
 # We need awk for the "check" target.  The system "awk" is bad on
 # some platforms.
-# Always define AMTAR for backward compatibility.
+# Always define AMTAR for backward compatibility.  Yes, it's still used
+# in the wild :-(  We should find a proper way to deprecate it ...
+AMTAR='$${TAR-tar}'
 
-AMTAR=${AMTAR-"${am_missing_run}tar"}
 
-am__tar='${AMTAR} chof - "$$tardir"'; am__untar='${AMTAR} xf -'
+# We'll loop over all known methods to create a tar archive until one works.
+_am_tools='gnutar  pax cpio none'
+
+am__tar='$${TAR-tar} chof - "$$tardir"' am__untar='$${TAR-tar} xf -'
 
 
 
 
 
 
+# POSIX will say in a future version that running "rm -f" with no argument
+# is OK; and we want to be able to make that assumption in our Makefile
+# recipes.  So use an aggressive probe to check that the usage we want is
+# actually supported "in the wild" to an acceptable degree.
+# See automake bug#10828.
+# To make any issue more visible, cause the running configure to be aborted
+# by default if the 'rm' program in use doesn't match our expectations; the
+# user can still override this though.
+if rm -f && rm -fr && rm -rf; then : OK; else
+  cat >&2 <<'END'
+Oops!
+
+Your 'rm' program seems unable to run without file operands specified
+on the command line, even when the '-f' option is present.  This is contrary
+to the behaviour of most rm programs out there, and not conforming with
+the upcoming POSIX standard: <http://austingroupbugs.net/view.php?id=542>
+
+Please tell bug-automake@gnu.org about your system, including the value
+of your $PATH and any error possibly output before this message.  This
+can help us improve future automake versions.
+
+END
+  if test x"$ACCEPT_INFERIOR_RM_PROGRAM" = x"yes"; then
+    echo 'Configuration will proceed anyway, since you have set the' >&2
+    echo 'ACCEPT_INFERIOR_RM_PROGRAM variable to "yes"' >&2
+    echo >&2
+  else
+    cat >&2 <<'END'
+Aborting the configuration process, to ensure you take notice of the issue.
+
+You can download and install GNU coreutils to get an 'rm' implementation
+that behaves properly: <http://www.gnu.org/software/coreutils/>.
+
+If you want to complete the configuration process using your problematic
+'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
+to "yes", and re-run configure.
+
+END
+    as_fn_error $? "Your 'rm' program is bad, sorry." "$LINENO" 5
+  fi
+fi
+
 
 ac_config_headers="$ac_config_headers lib/Grid_config.h"
 
 # Check whether --enable-silent-rules was given.
-if test "${enable_silent_rules+set}" = set; then
+if test "${enable_silent_rules+set}" = set; then :
   enableval=$enable_silent_rules;
 fi
 
-case $enable_silent_rules in
-yes) AM_DEFAULT_VERBOSITY=0;;
-no)  AM_DEFAULT_VERBOSITY=1;;
-*)   AM_DEFAULT_VERBOSITY=0;;
+case $enable_silent_rules in # (((
+  yes) AM_DEFAULT_VERBOSITY=0;;
+   no) AM_DEFAULT_VERBOSITY=1;;
+    *) AM_DEFAULT_VERBOSITY=0;;
 esac
+am_make=${MAKE-make}
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $am_make supports nested variables" >&5
+$as_echo_n "checking whether $am_make supports nested variables... " >&6; }
+if ${am_cv_make_support_nested_variables+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if $as_echo 'TRUE=$(BAR$(V))
+BAR0=false
+BAR1=true
+V=1
+am__doit:
+	@$(TRUE)
+.PHONY: am__doit' | $am_make -f - >/dev/null 2>&1; then
+  am_cv_make_support_nested_variables=yes
+else
+  am_cv_make_support_nested_variables=no
+fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_make_support_nested_variables" >&5
+$as_echo "$am_cv_make_support_nested_variables" >&6; }
+if test $am_cv_make_support_nested_variables = yes; then
+    AM_V='$(V)'
+  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
+else
+  AM_V=$AM_DEFAULT_VERBOSITY
+  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
+fi
 AM_BACKSLASH='\'
 
 
-{ $as_echo "$as_me:$LINENO:
+{ $as_echo "$as_me:${as_lineno-$LINENO}:
 
 :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
 Configuring $PACKAGE v$VERSION  for $host
@@ -2573,9 +3139,9 @@ if test -z "$CXX"; then
   do
     # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
 set dummy $ac_tool_prefix$ac_prog; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_CXX+set}" = set; then
+if ${ac_cv_prog_CXX+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$CXX"; then
@@ -2586,24 +3152,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_CXX="$ac_tool_prefix$ac_prog"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 CXX=$ac_cv_prog_CXX
 if test -n "$CXX"; then
-  { $as_echo "$as_me:$LINENO: result: $CXX" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CXX" >&5
 $as_echo "$CXX" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2617,9 +3183,9 @@ if test -z "$CXX"; then
 do
   # Extract the first word of "$ac_prog", so it can be a program name with args.
 set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_ac_ct_CXX+set}" = set; then
+if ${ac_cv_prog_ac_ct_CXX+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_CXX"; then
@@ -2630,24 +3196,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_ac_ct_CXX="$ac_prog"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_CXX=$ac_cv_prog_ac_ct_CXX
 if test -n "$ac_ct_CXX"; then
-  { $as_echo "$as_me:$LINENO: result: $ac_ct_CXX" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CXX" >&5
 $as_echo "$ac_ct_CXX" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -2660,7 +3226,7 @@ done
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -2671,48 +3237,31 @@ fi
   fi
 fi
 # Provide some information about the compiler.
-$as_echo "$as_me:$LINENO: checking for C++ compiler version" >&5
+$as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler version" >&5
 set X $ac_compile
 ac_compiler=$2
-{ (ac_try="$ac_compiler --version >&5"
+for ac_option in --version -v -V -qversion; do
+  { { ac_try="$ac_compiler $ac_option >&5"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler --version >&5") 2>&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
-{ (ac_try="$ac_compiler -v >&5"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler -v >&5") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
-{ (ac_try="$ac_compiler -V >&5"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compiler -V >&5") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }
+  if test -s conftest.err; then
+    sed '10a\
+... rest of stderr output deleted ...
+         10q' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+  fi
+  rm -f conftest.er1 conftest.err
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }
+done
 
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2728,8 +3277,8 @@ ac_clean_files="$ac_clean_files a.out a.out.dSYM a.exe b.out"
 # Try to create an executable without -o first, disregard a.out.
 # It will help us diagnose broken compilers, and finding out an intuition
 # of exeext.
-{ $as_echo "$as_me:$LINENO: checking for C++ compiler default output file name" >&5
-$as_echo_n "checking for C++ compiler default output file name... " >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether the C++ compiler works" >&5
+$as_echo_n "checking whether the C++ compiler works... " >&6; }
 ac_link_default=`$as_echo "$ac_link" | sed 's/ -o *conftest[^ ]*//'`
 
 # The possible output files:
@@ -2745,17 +3294,17 @@ do
 done
 rm -f $ac_rmfiles
 
-if { (ac_try="$ac_link_default"
+if { { ac_try="$ac_link_default"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
   (eval "$ac_link_default") 2>&5
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; then
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then :
   # Autoconf-2.13 could set the ac_cv_exeext variable to `no'.
 # So ignore a value of `no', otherwise this would lead to `EXEEXT = no'
 # in a Makefile.  We should not override ac_cv_exeext if it was cached,
@@ -2772,7 +3321,7 @@ do
 	# certainly right.
 	break;;
     *.* )
-        if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
+	if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
 	then :; else
 	   ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
 	fi
@@ -2791,84 +3340,41 @@ test "$ac_cv_exeext" = no && ac_cv_exeext=
 else
   ac_file=''
 fi
-
-{ $as_echo "$as_me:$LINENO: result: $ac_file" >&5
-$as_echo "$ac_file" >&6; }
-if test -z "$ac_file"; then
-  $as_echo "$as_me: failed program was:" >&5
+if test -z "$ac_file"; then :
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+$as_echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
 
-{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: C++ compiler cannot create executables
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: C++ compiler cannot create executables
-See \`config.log' for more details." >&2;}
-   { (exit 77); exit 77; }; }; }
-fi
-
-ac_exeext=$ac_cv_exeext
-
-# Check that the compiler produces executables we can run.  If not, either
-# the compiler is broken, or we cross compile.
-{ $as_echo "$as_me:$LINENO: checking whether the C++ compiler works" >&5
-$as_echo_n "checking whether the C++ compiler works... " >&6; }
-# FIXME: These cross compiler hacks should be removed for Autoconf 3.0
-# If not cross compiling, check that we can run a simple program.
-if test "$cross_compiling" != yes; then
-  if { ac_try='./$ac_file'
-  { (case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; }; then
-    cross_compiling=no
-  else
-    if test "$cross_compiling" = maybe; then
-	cross_compiling=yes
-    else
-	{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: cannot run C++ compiled programs.
-If you meant to cross compile, use \`--host'.
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: cannot run C++ compiled programs.
-If you meant to cross compile, use \`--host'.
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
-    fi
-  fi
-fi
-{ $as_echo "$as_me:$LINENO: result: yes" >&5
+as_fn_error 77 "C++ compiler cannot create executables
+See \`config.log' for more details" "$LINENO" 5; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
 $as_echo "yes" >&6; }
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler default output file name" >&5
+$as_echo_n "checking for C++ compiler default output file name... " >&6; }
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_file" >&5
+$as_echo "$ac_file" >&6; }
+ac_exeext=$ac_cv_exeext
 
 rm -f -r a.out a.out.dSYM a.exe conftest$ac_cv_exeext b.out
 ac_clean_files=$ac_clean_files_save
-# Check that the compiler produces executables we can run.  If not, either
-# the compiler is broken, or we cross compile.
-{ $as_echo "$as_me:$LINENO: checking whether we are cross compiling" >&5
-$as_echo_n "checking whether we are cross compiling... " >&6; }
-{ $as_echo "$as_me:$LINENO: result: $cross_compiling" >&5
-$as_echo "$cross_compiling" >&6; }
-
-{ $as_echo "$as_me:$LINENO: checking for suffix of executables" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of executables" >&5
 $as_echo_n "checking for suffix of executables... " >&6; }
-if { (ac_try="$ac_link"
+if { { ac_try="$ac_link"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
   (eval "$ac_link") 2>&5
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; then
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then :
   # If both `conftest.exe' and `conftest' are `present' (well, observable)
 # catch `conftest.exe'.  For instance with Cygwin, `ls conftest' will
 # work properly (i.e., refer to `conftest.exe'), while it won't with
@@ -2883,32 +3389,83 @@ for ac_file in conftest.exe conftest conftest.*; do
   esac
 done
 else
-  { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: cannot compute suffix of executables: cannot compile and link
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: cannot compute suffix of executables: cannot compile and link
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
+as_fn_error $? "cannot compute suffix of executables: cannot compile and link
+See \`config.log' for more details" "$LINENO" 5; }
 fi
-
-rm -f conftest$ac_cv_exeext
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_exeext" >&5
+rm -f conftest conftest$ac_cv_exeext
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_exeext" >&5
 $as_echo "$ac_cv_exeext" >&6; }
 
 rm -f conftest.$ac_ext
 EXEEXT=$ac_cv_exeext
 ac_exeext=$EXEEXT
-{ $as_echo "$as_me:$LINENO: checking for suffix of object files" >&5
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <stdio.h>
+int
+main ()
+{
+FILE *f = fopen ("conftest.out", "w");
+ return ferror (f) || fclose (f) != 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+ac_clean_files="$ac_clean_files conftest.out"
+# Check that the compiler produces executables we can run.  If not, either
+# the compiler is broken, or we cross compile.
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are cross compiling" >&5
+$as_echo_n "checking whether we are cross compiling... " >&6; }
+if test "$cross_compiling" != yes; then
+  { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }
+  if { ac_try='./conftest$ac_cv_exeext'
+  { { case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; }; then
+    cross_compiling=no
+  else
+    if test "$cross_compiling" = maybe; then
+	cross_compiling=yes
+    else
+	{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "cannot run C++ compiled programs.
+If you meant to cross compile, use \`--host'.
+See \`config.log' for more details" "$LINENO" 5; }
+    fi
+  fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $cross_compiling" >&5
+$as_echo "$cross_compiling" >&6; }
+
+rm -f conftest.$ac_ext conftest$ac_cv_exeext conftest.out
+ac_clean_files=$ac_clean_files_save
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of object files" >&5
 $as_echo_n "checking for suffix of object files... " >&6; }
-if test "${ac_cv_objext+set}" = set; then
+if ${ac_cv_objext+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2920,17 +3477,17 @@ main ()
 }
 _ACEOF
 rm -f conftest.o conftest.obj
-if { (ac_try="$ac_compile"
+if { { ac_try="$ac_compile"
 case "(($ac_try" in
   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
   *) ac_try_echo=$ac_try;;
 esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
   (eval "$ac_compile") 2>&5
   ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; then
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then :
   for ac_file in conftest.o conftest.obj conftest.*; do
   test -f "$ac_file" || continue;
   case $ac_file in
@@ -2943,31 +3500,23 @@ else
   $as_echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
 
-{ { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: cannot compute suffix of object files: cannot compile
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: cannot compute suffix of object files: cannot compile
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
+as_fn_error $? "cannot compute suffix of object files: cannot compile
+See \`config.log' for more details" "$LINENO" 5; }
 fi
-
 rm -f conftest.$ac_cv_objext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_objext" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_objext" >&5
 $as_echo "$ac_cv_objext" >&6; }
 OBJEXT=$ac_cv_objext
 ac_objext=$OBJEXT
-{ $as_echo "$as_me:$LINENO: checking whether we are using the GNU C++ compiler" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C++ compiler" >&5
 $as_echo_n "checking whether we are using the GNU C++ compiler... " >&6; }
-if test "${ac_cv_cxx_compiler_gnu+set}" = set; then
+if ${ac_cv_cxx_compiler_gnu+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -2981,37 +3530,16 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_cxx_try_compile "$LINENO"; then :
   ac_compiler_gnu=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_compiler_gnu=no
+  ac_compiler_gnu=no
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 ac_cv_cxx_compiler_gnu=$ac_compiler_gnu
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_cxx_compiler_gnu" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_cxx_compiler_gnu" >&5
 $as_echo "$ac_cv_cxx_compiler_gnu" >&6; }
 if test $ac_compiler_gnu = yes; then
   GXX=yes
@@ -3020,20 +3548,16 @@ else
 fi
 ac_test_CXXFLAGS=${CXXFLAGS+set}
 ac_save_CXXFLAGS=$CXXFLAGS
-{ $as_echo "$as_me:$LINENO: checking whether $CXX accepts -g" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX accepts -g" >&5
 $as_echo_n "checking whether $CXX accepts -g... " >&6; }
-if test "${ac_cv_prog_cxx_g+set}" = set; then
+if ${ac_cv_prog_cxx_g+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   ac_save_cxx_werror_flag=$ac_cxx_werror_flag
    ac_cxx_werror_flag=yes
    ac_cv_prog_cxx_g=no
    CXXFLAGS="-g"
-   cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -3044,35 +3568,11 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_cxx_try_compile "$LINENO"; then :
   ac_cv_prog_cxx_g=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	CXXFLAGS=""
-      cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  CXXFLAGS=""
+      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -3083,36 +3583,12 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+if ac_fn_cxx_try_compile "$LINENO"; then :
 
-	ac_cxx_werror_flag=$ac_save_cxx_werror_flag
+else
+  ac_cxx_werror_flag=$ac_save_cxx_werror_flag
 	 CXXFLAGS="-g"
-	 cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 int
@@ -3123,42 +3599,17 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_cxx_try_compile "$LINENO"; then :
   ac_cv_prog_cxx_g=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
    ac_cxx_werror_flag=$ac_save_cxx_werror_flag
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cxx_g" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cxx_g" >&5
 $as_echo "$ac_cv_prog_cxx_g" >&6; }
 if test "$ac_test_CXXFLAGS" = set; then
   CXXFLAGS=$ac_save_CXXFLAGS
@@ -3192,14 +3643,14 @@ am__doit:
 .PHONY: am__doit
 END
 # If we don't find an include directive, just comment out the code.
-{ $as_echo "$as_me:$LINENO: checking for style of include used by $am_make" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for style of include used by $am_make" >&5
 $as_echo_n "checking for style of include used by $am_make... " >&6; }
 am__include="#"
 am__quote=
 _am_result=none
 # First try GNU make style include.
 echo "include confinc" > confmf
-# Ignore all kinds of additional output from `make'.
+# Ignore all kinds of additional output from 'make'.
 case `$am_make -s -f confmf 2> /dev/null` in #(
 *the\ am__doit\ target*)
   am__include=include
@@ -3220,18 +3671,19 @@ if test "$am__include" = "#"; then
 fi
 
 
-{ $as_echo "$as_me:$LINENO: result: $_am_result" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $_am_result" >&5
 $as_echo "$_am_result" >&6; }
 rm -f confinc confmf
 
 # Check whether --enable-dependency-tracking was given.
-if test "${enable_dependency_tracking+set}" = set; then
+if test "${enable_dependency_tracking+set}" = set; then :
   enableval=$enable_dependency_tracking;
 fi
 
 if test "x$enable_dependency_tracking" != xno; then
   am_depcomp="$ac_aux_dir/depcomp"
   AMDEPBACKSLASH='\'
+  am__nodep='_no'
 fi
  if test "x$enable_dependency_tracking" != xno; then
   AMDEP_TRUE=
@@ -3245,17 +3697,18 @@ fi
 
 depcc="$CXX"  am_compiler_list=
 
-{ $as_echo "$as_me:$LINENO: checking dependency style of $depcc" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
 $as_echo_n "checking dependency style of $depcc... " >&6; }
-if test "${am_cv_CXX_dependencies_compiler_type+set}" = set; then
+if ${am_cv_CXX_dependencies_compiler_type+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
   # We make a subdir and do the tests there.  Otherwise we can end up
   # making bogus files that we don't know about and never remove.  For
   # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named `D' -- because `-MD' means `put the output
-  # in D'.
+  # making a dummy file named 'D' -- because '-MD' means "put the output
+  # in D".
+  rm -rf conftest.dir
   mkdir conftest.dir
   # Copy depcomp to subdir because otherwise we won't find it if we're
   # using a relative directory.
@@ -3289,16 +3742,16 @@ else
     : > sub/conftest.c
     for i in 1 2 3 4 5 6; do
       echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using `: > sub/conftst$i.h' creates only sub/conftst1.h with
-      # Solaris 8's {/usr,}/bin/sh.
-      touch sub/conftst$i.h
+      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
+      # Solaris 10 /bin/sh.
+      echo '/* dummy */' > sub/conftst$i.h
     done
     echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
 
-    # We check with `-c' and `-o' for the sake of the "dashmstdout"
+    # We check with '-c' and '-o' for the sake of the "dashmstdout"
     # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle `-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs
+    # handle '-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs.
     am__obj=sub/conftest.${OBJEXT-o}
     am__minus_obj="-o $am__obj"
     case $depmode in
@@ -3307,16 +3760,16 @@ else
       test "$am__universal" = false || continue
       ;;
     nosideeffect)
-      # after this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested
+      # After this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested.
       if test "x$enable_dependency_tracking" = xyes; then
 	continue
       else
 	break
       fi
       ;;
-    msvisualcpp | msvcmsys)
-      # This compiler won't grok `-c -o', but also, the minuso test has
+    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
+      # This compiler won't grok '-c -o', but also, the minuso test has
       # not run yet.  These depmodes are late enough in the game, and
       # so weak that their functioning should not be impacted.
       am__obj=conftest.${OBJEXT-o}
@@ -3355,7 +3808,7 @@ else
 fi
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $am_cv_CXX_dependencies_compiler_type" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CXX_dependencies_compiler_type" >&5
 $as_echo "$am_cv_CXX_dependencies_compiler_type" >&6; }
 CXXDEPMODE=depmode=$am_cv_CXX_dependencies_compiler_type
 
@@ -3374,17 +3827,18 @@ fi
 
   OPENMP_CXXFLAGS=
   # Check whether --enable-openmp was given.
-if test "${enable_openmp+set}" = set; then
+if test "${enable_openmp+set}" = set; then :
   enableval=$enable_openmp;
 fi
 
   if test "$enable_openmp" != no; then
-    { $as_echo "$as_me:$LINENO: checking for $CC option to support OpenMP" >&5
-$as_echo_n "checking for $CC option to support OpenMP... " >&6; }
-if test "${ac_cv_prog_cxx_openmp+set}" = set; then
+    { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CXX option to support OpenMP" >&5
+$as_echo_n "checking for $CXX option to support OpenMP... " >&6; }
+if ${ac_cv_prog_cxx_openmp+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
 
 #ifndef _OPENMP
  choke me
@@ -3393,37 +3847,16 @@ else
 int main () { return omp_get_num_threads (); }
 
 _ACEOF
-rm -f conftest.$ac_objext conftest$ac_exeext
-if { (ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_link") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 $as_test_x conftest$ac_exeext
-       }; then
+if ac_fn_cxx_try_link "$LINENO"; then :
   ac_cv_prog_cxx_openmp='none needed'
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_cv_prog_cxx_openmp='unsupported'
-	  	  	  	  	  	  	  	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp; do
+  ac_cv_prog_cxx_openmp='unsupported'
+	  	  	  	  	  	  	                                	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp -homp \
+                           -Popenmp --openmp; do
 	    ac_save_CXXFLAGS=$CXXFLAGS
 	    CXXFLAGS="$CXXFLAGS $ac_option"
-	    cat >conftest.$ac_ext <<_ACEOF
+	    cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
 
 #ifndef _OPENMP
  choke me
@@ -3432,56 +3865,27 @@ sed 's/^/| /' conftest.$ac_ext >&5
 int main () { return omp_get_num_threads (); }
 
 _ACEOF
-rm -f conftest.$ac_objext conftest$ac_exeext
-if { (ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_link") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 $as_test_x conftest$ac_exeext
-       }; then
+if ac_fn_cxx_try_link "$LINENO"; then :
   ac_cv_prog_cxx_openmp=$ac_option
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
 fi
-
-rm -rf conftest.dSYM
-rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
-      conftest$ac_exeext conftest.$ac_ext
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
 	    CXXFLAGS=$ac_save_CXXFLAGS
 	    if test "$ac_cv_prog_cxx_openmp" != unsupported; then
 	      break
 	    fi
 	  done
 fi
-
-rm -rf conftest.dSYM
-rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
-      conftest$ac_exeext conftest.$ac_ext
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_prog_cxx_openmp" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cxx_openmp" >&5
 $as_echo "$ac_cv_prog_cxx_openmp" >&6; }
     case $ac_cv_prog_cxx_openmp in #(
       "none needed" | unsupported)
-        ;; #(
+	;; #(
       *)
-        OPENMP_CXXFLAGS=$ac_cv_prog_cxx_openmp ;;
+	OPENMP_CXXFLAGS=$ac_cv_prog_cxx_openmp ;;
     esac
   fi
 
@@ -3489,9 +3893,9 @@ $as_echo "$ac_cv_prog_cxx_openmp" >&6; }
 if test -n "$ac_tool_prefix"; then
   # Extract the first word of "${ac_tool_prefix}ranlib", so it can be a program name with args.
 set dummy ${ac_tool_prefix}ranlib; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_RANLIB+set}" = set; then
+if ${ac_cv_prog_RANLIB+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$RANLIB"; then
@@ -3502,24 +3906,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_RANLIB="${ac_tool_prefix}ranlib"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 RANLIB=$ac_cv_prog_RANLIB
 if test -n "$RANLIB"; then
-  { $as_echo "$as_me:$LINENO: result: $RANLIB" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $RANLIB" >&5
 $as_echo "$RANLIB" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3529,9 +3933,9 @@ if test -z "$ac_cv_prog_RANLIB"; then
   ac_ct_RANLIB=$RANLIB
   # Extract the first word of "ranlib", so it can be a program name with args.
 set dummy ranlib; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_ac_ct_RANLIB+set}" = set; then
+if ${ac_cv_prog_ac_ct_RANLIB+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$ac_ct_RANLIB"; then
@@ -3542,24 +3946,24 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_ac_ct_RANLIB="ranlib"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 ac_ct_RANLIB=$ac_cv_prog_ac_ct_RANLIB
 if test -n "$ac_ct_RANLIB"; then
-  { $as_echo "$as_me:$LINENO: result: $ac_ct_RANLIB" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_RANLIB" >&5
 $as_echo "$ac_ct_RANLIB" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
@@ -3568,7 +3972,7 @@ fi
   else
     case $cross_compiling:$ac_tool_warned in
 yes:)
-{ $as_echo "$as_me:$LINENO: WARNING: using cross tools not prefixed with host triplet" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
 $as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
 ac_tool_warned=yes ;;
 esac
@@ -3585,16 +3989,12 @@ ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
 ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
 ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
   ac_success=no
-  { $as_echo "$as_me:$LINENO: checking whether $CXX supports C++11 features by default" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX supports C++11 features by default" >&5
 $as_echo_n "checking whether $CXX supports C++11 features by default... " >&6; }
-if test "${ax_cv_cxx_compile_cxx11+set}" = set; then
+if ${ax_cv_cxx_compile_cxx11+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
   template <typename T>
@@ -3646,35 +4046,14 @@ cat >>conftest.$ac_ext <<_ACEOF
     }
 
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_cxx_try_compile "$LINENO"; then :
   ax_cv_cxx_compile_cxx11=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ax_cv_cxx_compile_cxx11=no
+  ax_cv_cxx_compile_cxx11=no
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 fi
-{ $as_echo "$as_me:$LINENO: result: $ax_cv_cxx_compile_cxx11" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_cxx_compile_cxx11" >&5
 $as_echo "$ax_cv_cxx_compile_cxx11" >&6; }
   if test x$ax_cv_cxx_compile_cxx11 = xyes; then
     ac_success=yes
@@ -3685,18 +4064,14 @@ $as_echo "$ax_cv_cxx_compile_cxx11" >&6; }
     if test x$ac_success = xno; then
             for switch in -std=c++11 -std=c++0x +std=c++11; do
       cachevar=`$as_echo "ax_cv_cxx_compile_cxx11_$switch" | $as_tr_sh`
-      { $as_echo "$as_me:$LINENO: checking whether $CXX supports C++11 features with $switch" >&5
+      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX supports C++11 features with $switch" >&5
 $as_echo_n "checking whether $CXX supports C++11 features with $switch... " >&6; }
-if { as_var=$cachevar; eval "test \"\${$as_var+set}\" = set"; }; then
+if eval \${$cachevar+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   ac_save_CXXFLAGS="$CXXFLAGS"
          CXXFLAGS="$CXXFLAGS $switch"
-         cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+         cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
   template <typename T>
@@ -3748,38 +4123,16 @@ cat >>conftest.$ac_ext <<_ACEOF
     }
 
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_cxx_try_compile "$LINENO"; then :
   eval $cachevar=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	eval $cachevar=no
+  eval $cachevar=no
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
          CXXFLAGS="$ac_save_CXXFLAGS"
 fi
-ac_res=`eval 'as_val=${'$cachevar'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
+eval ac_res=\$$cachevar
+	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
 $as_echo "$ac_res" >&6; }
       if eval test x\$$cachevar = xyes; then
         CXXFLAGS="$CXXFLAGS $switch"
@@ -3796,21 +4149,17 @@ ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
 
   if test x$ax_cxx_compile_cxx11_required = xtrue; then
     if test x$ac_success = xno; then
-      { { $as_echo "$as_me:$LINENO: error: *** A compiler with support for C++11 language features is required." >&5
-$as_echo "$as_me: error: *** A compiler with support for C++11 language features is required." >&2;}
-   { (exit 1); exit 1; }; }
+      as_fn_error $? "*** A compiler with support for C++11 language features is required." "$LINENO" 5
     fi
   else
     if test x$ac_success = xno; then
       HAVE_CXX11=0
-      { $as_echo "$as_me:$LINENO: No compiler with C++11 support was found" >&5
+      { $as_echo "$as_me:${as_lineno-$LINENO}: No compiler with C++11 support was found" >&5
 $as_echo "$as_me: No compiler with C++11 support was found" >&6;}
     else
       HAVE_CXX11=1
 
-cat >>confdefs.h <<\_ACEOF
-#define HAVE_CXX11 1
-_ACEOF
+$as_echo "#define HAVE_CXX11 1" >>confdefs.h
 
     fi
 
@@ -3828,10 +4177,10 @@ ac_cpp='$CXXCPP $CPPFLAGS'
 ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
 ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
 ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
-{ $as_echo "$as_me:$LINENO: checking how to run the C++ preprocessor" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking how to run the C++ preprocessor" >&5
 $as_echo_n "checking how to run the C++ preprocessor... " >&6; }
 if test -z "$CXXCPP"; then
-  if test "${ac_cv_prog_CXXCPP+set}" = set; then
+  if ${ac_cv_prog_CXXCPP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
       # Double quotes because CXXCPP needs to be expanded
@@ -3846,11 +4195,7 @@ do
   # <limits.h> exists even on freestanding compilers.
   # On the NeXT, cc -E runs the code through the compiler's parser,
   # not just through cpp. "Syntax error" is here to catch this case.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #ifdef __STDC__
 # include <limits.h>
@@ -3859,78 +4204,34 @@ cat >>conftest.$ac_ext <<_ACEOF
 #endif
 		     Syntax error
 _ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+if ac_fn_cxx_try_cpp "$LINENO"; then :
 
+else
   # Broken: fails on valid input.
 continue
 fi
-
-rm -f conftest.err conftest.$ac_ext
+rm -f conftest.err conftest.i conftest.$ac_ext
 
   # OK, works on sane cases.  Now check whether nonexistent headers
   # can be detected and how.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <ac_nonexistent.h>
 _ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
+if ac_fn_cxx_try_cpp "$LINENO"; then :
   # Broken: success on invalid input.
 continue
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
   # Passes both tests.
 ac_preproc_ok=:
 break
 fi
-
-rm -f conftest.err conftest.$ac_ext
+rm -f conftest.err conftest.i conftest.$ac_ext
 
 done
 # Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
-rm -f conftest.err conftest.$ac_ext
-if $ac_preproc_ok; then
+rm -f conftest.i conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then :
   break
 fi
 
@@ -3942,7 +4243,7 @@ fi
 else
   ac_cv_prog_CXXCPP=$CXXCPP
 fi
-{ $as_echo "$as_me:$LINENO: result: $CXXCPP" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $CXXCPP" >&5
 $as_echo "$CXXCPP" >&6; }
 ac_preproc_ok=false
 for ac_cxx_preproc_warn_flag in '' yes
@@ -3953,11 +4254,7 @@ do
   # <limits.h> exists even on freestanding compilers.
   # On the NeXT, cc -E runs the code through the compiler's parser,
   # not just through cpp. "Syntax error" is here to catch this case.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #ifdef __STDC__
 # include <limits.h>
@@ -3966,87 +4263,40 @@ cat >>conftest.$ac_ext <<_ACEOF
 #endif
 		     Syntax error
 _ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+if ac_fn_cxx_try_cpp "$LINENO"; then :
 
+else
   # Broken: fails on valid input.
 continue
 fi
-
-rm -f conftest.err conftest.$ac_ext
+rm -f conftest.err conftest.i conftest.$ac_ext
 
   # OK, works on sane cases.  Now check whether nonexistent headers
   # can be detected and how.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <ac_nonexistent.h>
 _ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
+if ac_fn_cxx_try_cpp "$LINENO"; then :
   # Broken: success on invalid input.
 continue
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
   # Passes both tests.
 ac_preproc_ok=:
 break
 fi
-
-rm -f conftest.err conftest.$ac_ext
+rm -f conftest.err conftest.i conftest.$ac_ext
 
 done
 # Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
-rm -f conftest.err conftest.$ac_ext
-if $ac_preproc_ok; then
-  :
+rm -f conftest.i conftest.err conftest.$ac_ext
+if $ac_preproc_ok; then :
+
 else
-  { { $as_echo "$as_me:$LINENO: error: in \`$ac_pwd':" >&5
+  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
 $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-{ { $as_echo "$as_me:$LINENO: error: C++ preprocessor \"$CXXCPP\" fails sanity check
-See \`config.log' for more details." >&5
-$as_echo "$as_me: error: C++ preprocessor \"$CXXCPP\" fails sanity check
-See \`config.log' for more details." >&2;}
-   { (exit 1); exit 1; }; }; }
+as_fn_error $? "C++ preprocessor \"$CXXCPP\" fails sanity check
+See \`config.log' for more details" "$LINENO" 5; }
 fi
 
 ac_ext=cpp
@@ -4056,9 +4306,9 @@ ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ex
 ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
 
 
-{ $as_echo "$as_me:$LINENO: checking for grep that handles long lines and -e" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for grep that handles long lines and -e" >&5
 $as_echo_n "checking for grep that handles long lines and -e... " >&6; }
-if test "${ac_cv_path_GREP+set}" = set; then
+if ${ac_cv_path_GREP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -z "$GREP"; then
@@ -4069,10 +4319,10 @@ for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_prog in grep ggrep; do
+    for ac_prog in grep ggrep; do
     for ac_exec_ext in '' $ac_executable_extensions; do
       ac_path_GREP="$as_dir/$ac_prog$ac_exec_ext"
-      { test -f "$ac_path_GREP" && $as_test_x "$ac_path_GREP"; } || continue
+      as_fn_executable_p "$ac_path_GREP" || continue
 # Check for GNU ac_path_GREP and select it if it is found.
   # Check for GNU $ac_path_GREP
 case `"$ac_path_GREP" --version 2>&1` in
@@ -4089,7 +4339,7 @@ case `"$ac_path_GREP" --version 2>&1` in
     $as_echo 'GREP' >> "conftest.nl"
     "$ac_path_GREP" -e 'GREP$' -e '-(cannot match)-' < "conftest.nl" >"conftest.out" 2>/dev/null || break
     diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
-    ac_count=`expr $ac_count + 1`
+    as_fn_arith $ac_count + 1 && ac_count=$as_val
     if test $ac_count -gt ${ac_path_GREP_max-0}; then
       # Best one so far, save it but keep looking for a better one
       ac_cv_path_GREP="$ac_path_GREP"
@@ -4104,26 +4354,24 @@ esac
       $ac_path_GREP_found && break 3
     done
   done
-done
+  done
 IFS=$as_save_IFS
   if test -z "$ac_cv_path_GREP"; then
-    { { $as_echo "$as_me:$LINENO: error: no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&5
-$as_echo "$as_me: error: no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
   fi
 else
   ac_cv_path_GREP=$GREP
 fi
 
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_path_GREP" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_GREP" >&5
 $as_echo "$ac_cv_path_GREP" >&6; }
  GREP="$ac_cv_path_GREP"
 
 
-{ $as_echo "$as_me:$LINENO: checking for egrep" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for egrep" >&5
 $as_echo_n "checking for egrep... " >&6; }
-if test "${ac_cv_path_EGREP+set}" = set; then
+if ${ac_cv_path_EGREP+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if echo a | $GREP -E '(a|b)' >/dev/null 2>&1
@@ -4137,10 +4385,10 @@ for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_prog in egrep; do
+    for ac_prog in egrep; do
     for ac_exec_ext in '' $ac_executable_extensions; do
       ac_path_EGREP="$as_dir/$ac_prog$ac_exec_ext"
-      { test -f "$ac_path_EGREP" && $as_test_x "$ac_path_EGREP"; } || continue
+      as_fn_executable_p "$ac_path_EGREP" || continue
 # Check for GNU ac_path_EGREP and select it if it is found.
   # Check for GNU $ac_path_EGREP
 case `"$ac_path_EGREP" --version 2>&1` in
@@ -4157,7 +4405,7 @@ case `"$ac_path_EGREP" --version 2>&1` in
     $as_echo 'EGREP' >> "conftest.nl"
     "$ac_path_EGREP" 'EGREP$' < "conftest.nl" >"conftest.out" 2>/dev/null || break
     diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
-    ac_count=`expr $ac_count + 1`
+    as_fn_arith $ac_count + 1 && ac_count=$as_val
     if test $ac_count -gt ${ac_path_EGREP_max-0}; then
       # Best one so far, save it but keep looking for a better one
       ac_cv_path_EGREP="$ac_path_EGREP"
@@ -4172,12 +4420,10 @@ esac
       $ac_path_EGREP_found && break 3
     done
   done
-done
+  done
 IFS=$as_save_IFS
   if test -z "$ac_cv_path_EGREP"; then
-    { { $as_echo "$as_me:$LINENO: error: no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&5
-$as_echo "$as_me: error: no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
   fi
 else
   ac_cv_path_EGREP=$EGREP
@@ -4185,21 +4431,17 @@ fi
 
    fi
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_path_EGREP" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_EGREP" >&5
 $as_echo "$ac_cv_path_EGREP" >&6; }
  EGREP="$ac_cv_path_EGREP"
 
 
-{ $as_echo "$as_me:$LINENO: checking for ANSI C header files" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for ANSI C header files" >&5
 $as_echo_n "checking for ANSI C header files... " >&6; }
-if test "${ac_cv_header_stdc+set}" = set; then
+if ${ac_cv_header_stdc+:} false; then :
   $as_echo_n "(cached) " >&6
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <stdlib.h>
 #include <stdarg.h>
@@ -4214,48 +4456,23 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
+if ac_fn_cxx_try_compile "$LINENO"; then :
   ac_cv_header_stdc=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_cv_header_stdc=no
+  ac_cv_header_stdc=no
 fi
-
 rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
 
 if test $ac_cv_header_stdc = yes; then
   # SunOS 4.x string.h does not declare mem*, contrary to ANSI.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <string.h>
 
 _ACEOF
 if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
-  $EGREP "memchr" >/dev/null 2>&1; then
-  :
+  $EGREP "memchr" >/dev/null 2>&1; then :
+
 else
   ac_cv_header_stdc=no
 fi
@@ -4265,18 +4482,14 @@ fi
 
 if test $ac_cv_header_stdc = yes; then
   # ISC 2.0.2 stdlib.h does not declare free, contrary to ANSI.
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <stdlib.h>
 
 _ACEOF
 if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
-  $EGREP "free" >/dev/null 2>&1; then
-  :
+  $EGREP "free" >/dev/null 2>&1; then :
+
 else
   ac_cv_header_stdc=no
 fi
@@ -4286,14 +4499,10 @@ fi
 
 if test $ac_cv_header_stdc = yes; then
   # /bin/cc in Irix-4.0.5 gets non-ANSI ctype macros unless using -ansi.
-  if test "$cross_compiling" = yes; then
+  if test "$cross_compiling" = yes; then :
   :
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 #include <ctype.h>
 #include <stdlib.h>
@@ -4320,118 +4529,33 @@ main ()
   return 0;
 }
 _ACEOF
-rm -f conftest$ac_exeext
-if { (ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_link") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && { ac_try='./conftest$ac_exeext'
-  { (case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); }; }; then
-  :
+if ac_fn_cxx_try_run "$LINENO"; then :
+
 else
-  $as_echo "$as_me: program exited with status $ac_status" >&5
-$as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-( exit $ac_status )
-ac_cv_header_stdc=no
+  ac_cv_header_stdc=no
 fi
-rm -rf conftest.dSYM
-rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext conftest.$ac_objext conftest.$ac_ext
+rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
+  conftest.$ac_objext conftest.beam conftest.$ac_ext
 fi
 
-
 fi
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_header_stdc" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_header_stdc" >&5
 $as_echo "$ac_cv_header_stdc" >&6; }
 if test $ac_cv_header_stdc = yes; then
 
-cat >>confdefs.h <<\_ACEOF
-#define STDC_HEADERS 1
-_ACEOF
+$as_echo "#define STDC_HEADERS 1" >>confdefs.h
 
 fi
 
 # On IRIX 5.3, sys/types and inttypes.h are conflicting.
-
-
-
-
-
-
-
-
-
 for ac_header in sys/types.h sys/stat.h stdlib.h string.h memory.h strings.h \
 		  inttypes.h stdint.h unistd.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  eval "$as_ac_Header=yes"
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	eval "$as_ac_Header=no"
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
+ac_fn_cxx_check_header_compile "$LINENO" "$ac_header" "$as_ac_Header" "$ac_includes_default
+"
+if eval test \"x\$"$as_ac_Header"\" = x"yes"; then :
   cat >>confdefs.h <<_ACEOF
 #define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
@@ -4441,1150 +4565,107 @@ fi
 done
 
 
-
 for ac_header in stdint.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
-$as_echo_n "checking $ac_header usability... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_header_compiler=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_header_compiler=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
-$as_echo_n "checking $ac_header presence... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <$ac_header>
-_ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  ac_header_preproc=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-  ac_header_preproc=no
-fi
-
-rm -f conftest.err conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_cxx_preproc_warn_flag in
-  yes:no: )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
-    ac_header_preproc=yes
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
-$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
-    ( cat <<\_ASBOX
-## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##
-_ASBOX
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  eval "$as_ac_Header=\$ac_header_preproc"
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-
-fi
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_cxx_check_header_mongrel "$LINENO" "stdint.h" "ac_cv_header_stdint_h" "$ac_includes_default"
+if test "x$ac_cv_header_stdint_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+#define HAVE_STDINT_H 1
 _ACEOF
 
 fi
 
 done
 
-
 for ac_header in mm_malloc.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
-$as_echo_n "checking $ac_header usability... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_header_compiler=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_header_compiler=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
-$as_echo_n "checking $ac_header presence... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <$ac_header>
-_ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  ac_header_preproc=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-  ac_header_preproc=no
-fi
-
-rm -f conftest.err conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_cxx_preproc_warn_flag in
-  yes:no: )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
-    ac_header_preproc=yes
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
-$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
-    ( cat <<\_ASBOX
-## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##
-_ASBOX
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  eval "$as_ac_Header=\$ac_header_preproc"
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-
-fi
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_cxx_check_header_mongrel "$LINENO" "mm_malloc.h" "ac_cv_header_mm_malloc_h" "$ac_includes_default"
+if test "x$ac_cv_header_mm_malloc_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+#define HAVE_MM_MALLOC_H 1
 _ACEOF
 
 fi
 
 done
 
-
 for ac_header in malloc/malloc.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
-$as_echo_n "checking $ac_header usability... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_header_compiler=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_header_compiler=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
-$as_echo_n "checking $ac_header presence... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <$ac_header>
-_ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  ac_header_preproc=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-  ac_header_preproc=no
-fi
-
-rm -f conftest.err conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_cxx_preproc_warn_flag in
-  yes:no: )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
-    ac_header_preproc=yes
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
-$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
-    ( cat <<\_ASBOX
-## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##
-_ASBOX
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  eval "$as_ac_Header=\$ac_header_preproc"
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-
-fi
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_cxx_check_header_mongrel "$LINENO" "malloc/malloc.h" "ac_cv_header_malloc_malloc_h" "$ac_includes_default"
+if test "x$ac_cv_header_malloc_malloc_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+#define HAVE_MALLOC_MALLOC_H 1
 _ACEOF
 
 fi
 
 done
 
-
 for ac_header in malloc.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
-$as_echo_n "checking $ac_header usability... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_header_compiler=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_header_compiler=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
-$as_echo_n "checking $ac_header presence... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <$ac_header>
-_ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  ac_header_preproc=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-  ac_header_preproc=no
-fi
-
-rm -f conftest.err conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_cxx_preproc_warn_flag in
-  yes:no: )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
-    ac_header_preproc=yes
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
-$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
-    ( cat <<\_ASBOX
-## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##
-_ASBOX
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  eval "$as_ac_Header=\$ac_header_preproc"
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-
-fi
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_cxx_check_header_mongrel "$LINENO" "malloc.h" "ac_cv_header_malloc_h" "$ac_includes_default"
+if test "x$ac_cv_header_malloc_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+#define HAVE_MALLOC_H 1
 _ACEOF
 
 fi
 
 done
 
-
 for ac_header in endian.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
-$as_echo_n "checking $ac_header usability... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_header_compiler=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_header_compiler=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
-$as_echo_n "checking $ac_header presence... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <$ac_header>
-_ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  ac_header_preproc=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-  ac_header_preproc=no
-fi
-
-rm -f conftest.err conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_cxx_preproc_warn_flag in
-  yes:no: )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
-    ac_header_preproc=yes
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
-$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
-    ( cat <<\_ASBOX
-## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##
-_ASBOX
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  eval "$as_ac_Header=\$ac_header_preproc"
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-
-fi
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_cxx_check_header_mongrel "$LINENO" "endian.h" "ac_cv_header_endian_h" "$ac_includes_default"
+if test "x$ac_cv_header_endian_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+#define HAVE_ENDIAN_H 1
 _ACEOF
 
 fi
 
 done
 
-
 for ac_header in gmp.h
-do
-as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  { $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:$LINENO: checking $ac_header usability" >&5
-$as_echo_n "checking $ac_header usability... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-#include <$ac_header>
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_header_compiler=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_header_compiler=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:$LINENO: checking $ac_header presence" >&5
-$as_echo_n "checking $ac_header presence... " >&6; }
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <$ac_header>
-_ACEOF
-if { (ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } >/dev/null && {
-	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       }; then
-  ac_header_preproc=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-  ac_header_preproc=no
-fi
-
-rm -f conftest.err conftest.$ac_ext
-{ $as_echo "$as_me:$LINENO: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_cxx_preproc_warn_flag in
-  yes:no: )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $ac_header: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the compiler's result" >&2;}
-    ac_header_preproc=yes
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $ac_header: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $ac_header:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $ac_header: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $ac_header:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: proceeding with the preprocessor's result" >&5
-$as_echo "$as_me: WARNING: $ac_header: proceeding with the preprocessor's result" >&2;}
-    { $as_echo "$as_me:$LINENO: WARNING: $ac_header: in the future, the compiler will take precedence" >&5
-$as_echo "$as_me: WARNING: $ac_header: in the future, the compiler will take precedence" >&2;}
-    ( cat <<\_ASBOX
-## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##
-_ASBOX
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-{ $as_echo "$as_me:$LINENO: checking for $ac_header" >&5
-$as_echo_n "checking for $ac_header... " >&6; }
-if { as_var=$as_ac_Header; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  eval "$as_ac_Header=\$ac_header_preproc"
-fi
-ac_res=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-
-fi
-as_val=`eval 'as_val=${'$as_ac_Header'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_cxx_check_header_mongrel "$LINENO" "gmp.h" "ac_cv_header_gmp_h" "$ac_includes_default"
+if test "x$ac_cv_header_gmp_h" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
+#define HAVE_GMP_H 1
 _ACEOF
 
 fi
 
 done
 
-{ $as_echo "$as_me:$LINENO: checking whether ntohll is declared" >&5
-$as_echo_n "checking whether ntohll is declared... " >&6; }
-if test "${ac_cv_have_decl_ntohll+set}" = set; then
-  $as_echo_n "(cached) " >&6
+ac_fn_cxx_check_decl "$LINENO" "ntohll" "ac_cv_have_decl_ntohll" "#include <arpa/inet.h>
+"
+if test "x$ac_cv_have_decl_ntohll" = xyes; then :
+  ac_have_decl=1
 else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <arpa/inet.h>
-
-int
-main ()
-{
-#ifndef ntohll
-  (void) ntohll;
-#endif
-
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_cv_have_decl_ntohll=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_cv_have_decl_ntohll=no
+  ac_have_decl=0
 fi
 
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_have_decl_ntohll" >&5
-$as_echo "$ac_cv_have_decl_ntohll" >&6; }
-if test "x$ac_cv_have_decl_ntohll" = x""yes; then
-
 cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_NTOHLL 1
+#define HAVE_DECL_NTOHLL $ac_have_decl
 _ACEOF
 
-
+ac_fn_cxx_check_decl "$LINENO" "be64toh" "ac_cv_have_decl_be64toh" "#include <arpa/inet.h>
+"
+if test "x$ac_cv_have_decl_be64toh" = xyes; then :
+  ac_have_decl=1
 else
-  cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_NTOHLL 0
-_ACEOF
-
-
+  ac_have_decl=0
 fi
 
-
-{ $as_echo "$as_me:$LINENO: checking whether be64toh is declared" >&5
-$as_echo_n "checking whether be64toh is declared... " >&6; }
-if test "${ac_cv_have_decl_be64toh+set}" = set; then
-  $as_echo_n "(cached) " >&6
-else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-#include <arpa/inet.h>
-
-int
-main ()
-{
-#ifndef be64toh
-  (void) be64toh;
-#endif
-
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  ac_cv_have_decl_be64toh=yes
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_cv_have_decl_be64toh=no
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_have_decl_be64toh" >&5
-$as_echo "$ac_cv_have_decl_be64toh" >&6; }
-if test "x$ac_cv_have_decl_be64toh" = x""yes; then
-
 cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_BE64TOH 1
+#define HAVE_DECL_BE64TOH $ac_have_decl
 _ACEOF
 
 
-else
-  cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_BE64TOH 0
-_ACEOF
-
-
-fi
-
-
-
 # Checks for typedefs, structures, and compiler characteristics.
-{ $as_echo "$as_me:$LINENO: checking for size_t" >&5
-$as_echo_n "checking for size_t... " >&6; }
-if test "${ac_cv_type_size_t+set}" = set; then
-  $as_echo_n "(cached) " >&6
-else
-  ac_cv_type_size_t=no
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-if (sizeof (size_t))
-       return 0;
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-if (sizeof ((size_t)))
-	  return 0;
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  :
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
+ac_fn_cxx_check_type "$LINENO" "size_t" "ac_cv_type_size_t" "$ac_includes_default"
+if test "x$ac_cv_type_size_t" = xyes; then :
 
-	ac_cv_type_size_t=yes
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_type_size_t" >&5
-$as_echo "$ac_cv_type_size_t" >&6; }
-if test "x$ac_cv_type_size_t" = x""yes; then
-  :
 else
 
 cat >>confdefs.h <<_ACEOF
@@ -5593,75 +4674,12 @@ _ACEOF
 
 fi
 
-
-  { $as_echo "$as_me:$LINENO: checking for uint32_t" >&5
-$as_echo_n "checking for uint32_t... " >&6; }
-if test "${ac_cv_c_uint32_t+set}" = set; then
-  $as_echo_n "(cached) " >&6
-else
-  ac_cv_c_uint32_t=no
-     for ac_type in 'uint32_t' 'unsigned int' 'unsigned long int' \
-	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
-       cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-static int test_array [1 - 2 * !(($ac_type) -1 >> (32 - 1) == 1)];
-test_array [0] = 0
-
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  case $ac_type in
-  uint32_t) ac_cv_c_uint32_t=yes ;;
-  *) ac_cv_c_uint32_t=$ac_type ;;
-esac
-
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-       test "$ac_cv_c_uint32_t" != no && break
-     done
-fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_uint32_t" >&5
-$as_echo "$ac_cv_c_uint32_t" >&6; }
-  case $ac_cv_c_uint32_t in #(
+ac_fn_c_find_uintX_t "$LINENO" "32" "ac_cv_c_uint32_t"
+case $ac_cv_c_uint32_t in #(
   no|yes) ;; #(
   *)
 
-cat >>confdefs.h <<\_ACEOF
-#define _UINT32_T 1
-_ACEOF
+$as_echo "#define _UINT32_T 1" >>confdefs.h
 
 
 cat >>confdefs.h <<_ACEOF
@@ -5670,75 +4688,12 @@ _ACEOF
 ;;
   esac
 
-
-  { $as_echo "$as_me:$LINENO: checking for uint64_t" >&5
-$as_echo_n "checking for uint64_t... " >&6; }
-if test "${ac_cv_c_uint64_t+set}" = set; then
-  $as_echo_n "(cached) " >&6
-else
-  ac_cv_c_uint64_t=no
-     for ac_type in 'uint64_t' 'unsigned int' 'unsigned long int' \
-	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
-       cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-static int test_array [1 - 2 * !(($ac_type) -1 >> (64 - 1) == 1)];
-test_array [0] = 0
-
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext
-if { (ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_compile") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then
-  case $ac_type in
-  uint64_t) ac_cv_c_uint64_t=yes ;;
-  *) ac_cv_c_uint64_t=$ac_type ;;
-esac
-
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-
-fi
-
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-       test "$ac_cv_c_uint64_t" != no && break
-     done
-fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_c_uint64_t" >&5
-$as_echo "$ac_cv_c_uint64_t" >&6; }
-  case $ac_cv_c_uint64_t in #(
+ac_fn_c_find_uintX_t "$LINENO" "64" "ac_cv_c_uint64_t"
+case $ac_cv_c_uint64_t in #(
   no|yes) ;; #(
   *)
 
-cat >>confdefs.h <<\_ACEOF
-#define _UINT64_T 1
-_ACEOF
+$as_echo "#define _UINT64_T 1" >>confdefs.h
 
 
 cat >>confdefs.h <<_ACEOF
@@ -5754,121 +4709,26 @@ echo Checking libraries
 echo :::::::::::::::::::::::::::::::::::::::::::
 
 
-
 for ac_func in gettimeofday
-do
-as_ac_var=`$as_echo "ac_cv_func_$ac_func" | $as_tr_sh`
-{ $as_echo "$as_me:$LINENO: checking for $ac_func" >&5
-$as_echo_n "checking for $ac_func... " >&6; }
-if { as_var=$as_ac_var; eval "test \"\${$as_var+set}\" = set"; }; then
-  $as_echo_n "(cached) " >&6
-else
-  cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
-/* end confdefs.h.  */
-/* Define $ac_func to an innocuous variant, in case <limits.h> declares $ac_func.
-   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
-#define $ac_func innocuous_$ac_func
-
-/* System header to define __stub macros and hopefully few prototypes,
-    which can conflict with char $ac_func (); below.
-    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
-    <limits.h> exists even on freestanding compilers.  */
-
-#ifdef __STDC__
-# include <limits.h>
-#else
-# include <assert.h>
-#endif
-
-#undef $ac_func
-
-/* Override any GCC internal prototype to avoid an error.
-   Use char because int might match the return type of a GCC
-   builtin and then its argument prototype would still apply.  */
-#ifdef __cplusplus
-extern "C"
-#endif
-char $ac_func ();
-/* The GNU C library defines this for functions which it implements
-    to always fail with ENOSYS.  Some functions are actually named
-    something starting with __ and the normal name is an alias.  */
-#if defined __stub_$ac_func || defined __stub___$ac_func
-choke me
-#endif
-
-int
-main ()
-{
-return $ac_func ();
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.$ac_objext conftest$ac_exeext
-if { (ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_link") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 $as_test_x conftest$ac_exeext
-       }; then
-  eval "$as_ac_var=yes"
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	eval "$as_ac_var=no"
-fi
-
-rm -rf conftest.dSYM
-rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
-      conftest$ac_exeext conftest.$ac_ext
-fi
-ac_res=`eval 'as_val=${'$as_ac_var'}
-		 $as_echo "$as_val"'`
-	       { $as_echo "$as_me:$LINENO: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-as_val=`eval 'as_val=${'$as_ac_var'}
-		 $as_echo "$as_val"'`
-   if test "x$as_val" = x""yes; then
+do :
+  ac_fn_cxx_check_func "$LINENO" "gettimeofday" "ac_cv_func_gettimeofday"
+if test "x$ac_cv_func_gettimeofday" = xyes; then :
   cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_func" | $as_tr_cpp` 1
+#define HAVE_GETTIMEOFDAY 1
 _ACEOF
 
 fi
 done
 
 
-
-{ $as_echo "$as_me:$LINENO: checking for __gmpf_init in -lgmp" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for __gmpf_init in -lgmp" >&5
 $as_echo_n "checking for __gmpf_init in -lgmp... " >&6; }
-if test "${ac_cv_lib_gmp___gmpf_init+set}" = set; then
+if ${ac_cv_lib_gmp___gmpf_init+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   ac_check_lib_save_LIBS=$LIBS
 LIBS="-lgmp  $LIBS"
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 /* Override any GCC internal prototype to avoid an error.
@@ -5886,43 +4746,18 @@ return __gmpf_init ();
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext conftest$ac_exeext
-if { (ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_link") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 $as_test_x conftest$ac_exeext
-       }; then
+if ac_fn_cxx_try_link "$LINENO"; then :
   ac_cv_lib_gmp___gmpf_init=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_cv_lib_gmp___gmpf_init=no
+  ac_cv_lib_gmp___gmpf_init=no
 fi
-
-rm -rf conftest.dSYM
-rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
-      conftest$ac_exeext conftest.$ac_ext
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
 LIBS=$ac_check_lib_save_LIBS
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_lib_gmp___gmpf_init" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_gmp___gmpf_init" >&5
 $as_echo "$ac_cv_lib_gmp___gmpf_init" >&6; }
-if test "x$ac_cv_lib_gmp___gmpf_init" = x""yes; then
+if test "x$ac_cv_lib_gmp___gmpf_init" = xyes; then :
   cat >>confdefs.h <<_ACEOF
 #define HAVE_LIBGMP 1
 _ACEOF
@@ -5930,29 +4765,20 @@ _ACEOF
   LIBS="-lgmp $LIBS"
 
 else
-  { { $as_echo "$as_me:$LINENO: error: GNU Multiple Precision GMP library was not found in your system.
+  as_fn_error $? "GNU Multiple Precision GMP library was not found in your system.
 Please install or provide the correct path to your installation
-Info at: http://www.gmplib.org" >&5
-$as_echo "$as_me: error: GNU Multiple Precision GMP library was not found in your system.
-Please install or provide the correct path to your installation
-Info at: http://www.gmplib.org" >&2;}
-   { (exit 1); exit 1; }; }
+Info at: http://www.gmplib.org" "$LINENO" 5
 fi
 
 
-
-{ $as_echo "$as_me:$LINENO: checking for mpfr_init in -lmpfr" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for mpfr_init in -lmpfr" >&5
 $as_echo_n "checking for mpfr_init in -lmpfr... " >&6; }
-if test "${ac_cv_lib_mpfr_mpfr_init+set}" = set; then
+if ${ac_cv_lib_mpfr_mpfr_init+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   ac_check_lib_save_LIBS=$LIBS
 LIBS="-lmpfr  $LIBS"
-cat >conftest.$ac_ext <<_ACEOF
-/* confdefs.h.  */
-_ACEOF
-cat confdefs.h >>conftest.$ac_ext
-cat >>conftest.$ac_ext <<_ACEOF
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
 
 /* Override any GCC internal prototype to avoid an error.
@@ -5970,43 +4796,18 @@ return mpfr_init ();
   return 0;
 }
 _ACEOF
-rm -f conftest.$ac_objext conftest$ac_exeext
-if { (ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:$LINENO: $ac_try_echo\""
-$as_echo "$ac_try_echo") >&5
-  (eval "$ac_link") 2>conftest.er1
-  ac_status=$?
-  grep -v '^ *+' conftest.er1 >conftest.err
-  rm -f conftest.er1
-  cat conftest.err >&5
-  $as_echo "$as_me:$LINENO: \$? = $ac_status" >&5
-  (exit $ac_status); } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 $as_test_x conftest$ac_exeext
-       }; then
+if ac_fn_cxx_try_link "$LINENO"; then :
   ac_cv_lib_mpfr_mpfr_init=yes
 else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_cv_lib_mpfr_mpfr_init=no
+  ac_cv_lib_mpfr_mpfr_init=no
 fi
-
-rm -rf conftest.dSYM
-rm -f core conftest.err conftest.$ac_objext conftest_ipa8_conftest.oo \
-      conftest$ac_exeext conftest.$ac_ext
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
 LIBS=$ac_check_lib_save_LIBS
 fi
-{ $as_echo "$as_me:$LINENO: result: $ac_cv_lib_mpfr_mpfr_init" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_mpfr_mpfr_init" >&5
 $as_echo "$ac_cv_lib_mpfr_mpfr_init" >&6; }
-if test "x$ac_cv_lib_mpfr_mpfr_init" = x""yes; then
+if test "x$ac_cv_lib_mpfr_mpfr_init" = xyes; then :
   cat >>confdefs.h <<_ACEOF
 #define HAVE_LIBMPFR 1
 _ACEOF
@@ -6014,13 +4815,9 @@ _ACEOF
   LIBS="-lmpfr $LIBS"
 
 else
-  { { $as_echo "$as_me:$LINENO: error: GNU Multiple Precision MPFR library was not found in your system.
+  as_fn_error $? "GNU Multiple Precision MPFR library was not found in your system.
 Please install or provide the correct path to your installation
-Info at: http://www.mpfr.org/" >&5
-$as_echo "$as_me: error: GNU Multiple Precision MPFR library was not found in your system.
-Please install or provide the correct path to your installation
-Info at: http://www.mpfr.org/" >&2;}
-   { (exit 1); exit 1; }; }
+Info at: http://www.mpfr.org/" "$LINENO" 5
 fi
 
 
@@ -6029,7 +4826,7 @@ fi
 
 
 # Check whether --enable-simd was given.
-if test "${enable_simd+set}" = set; then
+if test "${enable_simd+set}" = set; then :
   enableval=$enable_simd; \
 	ac_SIMD=${enable_simd}
 else
@@ -6041,45 +4838,35 @@ case ${ac_SIMD} in
      SSE4)
        echo Configuring for SSE4
 
-cat >>confdefs.h <<\_ACEOF
-#define SSE4 1
-_ACEOF
+$as_echo "#define SSE4 1" >>confdefs.h
 
      ;;
      AVX)
        echo Configuring for AVX
 
-cat >>confdefs.h <<\_ACEOF
-#define AVX1 1
-_ACEOF
+$as_echo "#define AVX1 1" >>confdefs.h
 
      ;;
      AVX2)
        echo Configuring for AVX2
 
-cat >>confdefs.h <<\_ACEOF
-#define AVX2 1
-_ACEOF
+$as_echo "#define AVX2 1" >>confdefs.h
 
      ;;
      AVX512|MIC)
        echo Configuring for AVX512 and MIC
 
-cat >>confdefs.h <<\_ACEOF
-#define AVX512 1
-_ACEOF
+$as_echo "#define AVX512 1" >>confdefs.h
 
      ;;
      *)
-     { { $as_echo "$as_me:$LINENO: error: ${ac_SIMD} unsupported --enable-simd option" >&5
-$as_echo "$as_me: error: ${ac_SIMD} unsupported --enable-simd option" >&2;}
-   { (exit 1); exit 1; }; };
+     as_fn_error $? "${ac_SIMD} unsupported --enable-simd option" "$LINENO" 5;
      ;;
 esac
 
 
 # Check whether --enable-comms was given.
-if test "${enable_comms+set}" = set; then
+if test "${enable_comms+set}" = set; then :
   enableval=$enable_comms; ac_COMMS=${enable_comms}
 else
   ac_COMMS=none
@@ -6090,23 +4877,17 @@ case ${ac_COMMS} in
      none)
        echo Configuring for NO communications
 
-cat >>confdefs.h <<\_ACEOF
-#define GRID_COMMS_NONE 1
-_ACEOF
+$as_echo "#define GRID_COMMS_NONE 1" >>confdefs.h
 
      ;;
      mpi)
        echo Configuring for MPI communications
 
-cat >>confdefs.h <<\_ACEOF
-#define GRID_COMMS_MPI 1
-_ACEOF
+$as_echo "#define GRID_COMMS_MPI 1" >>confdefs.h
 
      ;;
      *)
-     { { $as_echo "$as_me:$LINENO: error: ${ac_COMMS} unsupported --enable-comms option" >&5
-$as_echo "$as_me: error: ${ac_COMMS} unsupported --enable-comms option" >&2;}
-   { (exit 1); exit 1; }; };
+     as_fn_error $? "${ac_COMMS} unsupported --enable-comms option" "$LINENO" 5;
      ;;
 esac
 
@@ -6136,25 +4917,25 @@ echo Checking doxygen support
 echo :::::::::::::::::::::::::::::::::::::::::::
 
 # Check whether --enable-doxygen was given.
-if test "${enable_doxygen+set}" = set; then
+if test "${enable_doxygen+set}" = set; then :
   enableval=$enable_doxygen;
 fi
 
 
 # Check whether --enable-dot was given.
-if test "${enable_dot+set}" = set; then
+if test "${enable_dot+set}" = set; then :
   enableval=$enable_dot;
 fi
 
 # Check whether --enable-html-docs was given.
-if test "${enable_html_docs+set}" = set; then
+if test "${enable_html_docs+set}" = set; then :
   enableval=$enable_html_docs;
 else
    enable_html_docs=yes
 fi
 
 # Check whether --enable-latex-docs was given.
-if test "${enable_latex_docs+set}" = set; then
+if test "${enable_latex_docs+set}" = set; then :
   enableval=$enable_latex_docs;
 else
   enable_latex_docs=no
@@ -6166,9 +4947,9 @@ if test "x$enable_doxygen" = xno; then
 else
  # Extract the first word of "doxygen", so it can be a program name with args.
 set dummy doxygen; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_DOXYGEN+set}" = set; then
+if ${ac_cv_prog_DOXYGEN+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$DOXYGEN"; then
@@ -6179,33 +4960,31 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_DOXYGEN="doxygen"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 DOXYGEN=$ac_cv_prog_DOXYGEN
 if test -n "$DOXYGEN"; then
-  { $as_echo "$as_me:$LINENO: result: $DOXYGEN" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $DOXYGEN" >&5
 $as_echo "$DOXYGEN" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
 
  if test x$DOXYGEN = x; then
   if test "x$enable_doxygen" = xyes; then
-   { { $as_echo "$as_me:$LINENO: error: could not find doxygen" >&5
-$as_echo "$as_me: error: could not find doxygen" >&2;}
-   { (exit 1); exit 1; }; }
+   as_fn_error $? "could not find doxygen" "$LINENO" 5
   fi
   enable_doc=no
  else
@@ -6216,9 +4995,9 @@ $as_echo "$as_me: error: could not find doxygen" >&2;}
    enable_doc=yes
    # Extract the first word of "dot", so it can be a program name with args.
 set dummy dot; ac_word=$2
-{ $as_echo "$as_me:$LINENO: checking for $ac_word" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
 $as_echo_n "checking for $ac_word... " >&6; }
-if test "${ac_cv_prog_DOT+set}" = set; then
+if ${ac_cv_prog_DOT+:} false; then :
   $as_echo_n "(cached) " >&6
 else
   if test -n "$DOT"; then
@@ -6229,30 +5008,30 @@ for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  for ac_exec_ext in '' $ac_executable_extensions; do
-  if { test -f "$as_dir/$ac_word$ac_exec_ext" && $as_test_x "$as_dir/$ac_word$ac_exec_ext"; }; then
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
     ac_cv_prog_DOT="dot"
-    $as_echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
     break 2
   fi
 done
-done
+  done
 IFS=$as_save_IFS
 
 fi
 fi
 DOT=$ac_cv_prog_DOT
 if test -n "$DOT"; then
-  { $as_echo "$as_me:$LINENO: result: $DOT" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $DOT" >&5
 $as_echo "$DOT" >&6; }
 else
-  { $as_echo "$as_me:$LINENO: result: no" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
 $as_echo "no" >&6; }
 fi
 
 
   else
-   { $as_echo "$as_me:$LINENO: WARNING: doxygen version $doxy_ver too old, doxygen will not be used." >&5
+   { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: doxygen version $doxy_ver too old, doxygen will not be used." >&5
 $as_echo "$as_me: WARNING: doxygen version $doxy_ver too old, doxygen will not be used." >&2;}
    enable_doc=no
   fi
@@ -6270,9 +5049,7 @@ fi
 
 if test x$DOT = x; then
  if test "x$enable_dot" = xyes; then
-  { { $as_echo "$as_me:$LINENO: error: could not find dot" >&5
-$as_echo "$as_me: error: could not find dot" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "could not find dot" "$LINENO" 5
  fi
  enable_dot=no
 else
@@ -6326,13 +5103,13 @@ _ACEOF
     case $ac_val in #(
     *${as_nl}*)
       case $ac_var in #(
-      *_cv_*) { $as_echo "$as_me:$LINENO: WARNING: cache variable $ac_var contains a newline" >&5
+      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
 $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
       esac
       case $ac_var in #(
       _ | IFS | as_nl) ;; #(
       BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
-      *) $as_unset $ac_var ;;
+      *) { eval $ac_var=; unset $ac_var;} ;;
       esac ;;
     esac
   done
@@ -6340,8 +5117,8 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
   (set) 2>&1 |
     case $as_nl`(ac_space=' '; set) 2>&1` in #(
     *${as_nl}ac_space=\ *)
-      # `set' does not quote correctly, so add quotes (double-quote
-      # substitution turns \\\\ into \\, and sed turns \\ into \).
+      # `set' does not quote correctly, so add quotes: double-quote
+      # substitution turns \\\\ into \\, and sed turns \\ into \.
       sed -n \
 	"s/'/'\\\\''/g;
 	  s/^\\([_$as_cr_alnum]*_cv_[_$as_cr_alnum]*\\)=\\(.*\\)/\\1='\\2'/p"
@@ -6363,12 +5140,23 @@ $as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
      :end' >>confcache
 if diff "$cache_file" confcache >/dev/null 2>&1; then :; else
   if test -w "$cache_file"; then
-    test "x$cache_file" != "x/dev/null" &&
-      { $as_echo "$as_me:$LINENO: updating cache $cache_file" >&5
+    if test "x$cache_file" != "x/dev/null"; then
+      { $as_echo "$as_me:${as_lineno-$LINENO}: updating cache $cache_file" >&5
 $as_echo "$as_me: updating cache $cache_file" >&6;}
-    cat confcache >$cache_file
+      if test ! -f "$cache_file" || test -h "$cache_file"; then
+	cat confcache >"$cache_file"
+      else
+        case $cache_file in #(
+        */* | ?:*)
+	  mv -f confcache "$cache_file"$$ &&
+	  mv -f "$cache_file"$$ "$cache_file" ;; #(
+        *)
+	  mv -f confcache "$cache_file" ;;
+	esac
+      fi
+    fi
   else
-    { $as_echo "$as_me:$LINENO: not updating unwritable cache $cache_file" >&5
+    { $as_echo "$as_me:${as_lineno-$LINENO}: not updating unwritable cache $cache_file" >&5
 $as_echo "$as_me: not updating unwritable cache $cache_file" >&6;}
   fi
 fi
@@ -6382,20 +5170,29 @@ DEFS=-DHAVE_CONFIG_H
 
 ac_libobjs=
 ac_ltlibobjs=
+U=
 for ac_i in : $LIBOBJS; do test "x$ac_i" = x: && continue
   # 1. Remove the extension, and $U if already installed.
   ac_script='s/\$U\././;s/\.o$//;s/\.obj$//'
   ac_i=`$as_echo "$ac_i" | sed "$ac_script"`
   # 2. Prepend LIBOBJDIR.  When used with automake>=1.10 LIBOBJDIR
   #    will be set to the directory where LIBOBJS objects are built.
-  ac_libobjs="$ac_libobjs \${LIBOBJDIR}$ac_i\$U.$ac_objext"
-  ac_ltlibobjs="$ac_ltlibobjs \${LIBOBJDIR}$ac_i"'$U.lo'
+  as_fn_append ac_libobjs " \${LIBOBJDIR}$ac_i\$U.$ac_objext"
+  as_fn_append ac_ltlibobjs " \${LIBOBJDIR}$ac_i"'$U.lo'
 done
 LIBOBJS=$ac_libobjs
 
 LTLIBOBJS=$ac_ltlibobjs
 
 
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking that generated files are newer than configure" >&5
+$as_echo_n "checking that generated files are newer than configure... " >&6; }
+   if test -n "$am_sleep_pid"; then
+     # Hide warnings about reused PIDs.
+     wait $am_sleep_pid 2>/dev/null
+   fi
+   { $as_echo "$as_me:${as_lineno-$LINENO}: result: done" >&5
+$as_echo "done" >&6; }
  if test -n "$EXEEXT"; then
   am__EXEEXT_TRUE=
   am__EXEEXT_FALSE='#'
@@ -6405,48 +5202,34 @@ else
 fi
 
 if test -z "${AMDEP_TRUE}" && test -z "${AMDEP_FALSE}"; then
-  { { $as_echo "$as_me:$LINENO: error: conditional \"AMDEP\" was never defined.
-Usually this means the macro was only invoked conditionally." >&5
-$as_echo "$as_me: error: conditional \"AMDEP\" was never defined.
-Usually this means the macro was only invoked conditionally." >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "conditional \"AMDEP\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 if test -z "${am__fastdepCXX_TRUE}" && test -z "${am__fastdepCXX_FALSE}"; then
-  { { $as_echo "$as_me:$LINENO: error: conditional \"am__fastdepCXX\" was never defined.
-Usually this means the macro was only invoked conditionally." >&5
-$as_echo "$as_me: error: conditional \"am__fastdepCXX\" was never defined.
-Usually this means the macro was only invoked conditionally." >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "conditional \"am__fastdepCXX\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 if test -z "${BUILD_COMMS_MPI_TRUE}" && test -z "${BUILD_COMMS_MPI_FALSE}"; then
-  { { $as_echo "$as_me:$LINENO: error: conditional \"BUILD_COMMS_MPI\" was never defined.
-Usually this means the macro was only invoked conditionally." >&5
-$as_echo "$as_me: error: conditional \"BUILD_COMMS_MPI\" was never defined.
-Usually this means the macro was only invoked conditionally." >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "conditional \"BUILD_COMMS_MPI\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 if test -z "${BUILD_COMMS_NONE_TRUE}" && test -z "${BUILD_COMMS_NONE_FALSE}"; then
-  { { $as_echo "$as_me:$LINENO: error: conditional \"BUILD_COMMS_NONE\" was never defined.
-Usually this means the macro was only invoked conditionally." >&5
-$as_echo "$as_me: error: conditional \"BUILD_COMMS_NONE\" was never defined.
-Usually this means the macro was only invoked conditionally." >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "conditional \"BUILD_COMMS_NONE\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 if test -z "${DOXYGEN_DOC_TRUE}" && test -z "${DOXYGEN_DOC_FALSE}"; then
-  { { $as_echo "$as_me:$LINENO: error: conditional \"DOXYGEN_DOC\" was never defined.
-Usually this means the macro was only invoked conditionally." >&5
-$as_echo "$as_me: error: conditional \"DOXYGEN_DOC\" was never defined.
-Usually this means the macro was only invoked conditionally." >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "conditional \"DOXYGEN_DOC\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 
-: ${CONFIG_STATUS=./config.status}
+: "${CONFIG_STATUS=./config.status}"
 ac_write_fail=0
 ac_clean_files_save=$ac_clean_files
 ac_clean_files="$ac_clean_files $CONFIG_STATUS"
-{ $as_echo "$as_me:$LINENO: creating $CONFIG_STATUS" >&5
+{ $as_echo "$as_me:${as_lineno-$LINENO}: creating $CONFIG_STATUS" >&5
 $as_echo "$as_me: creating $CONFIG_STATUS" >&6;}
-cat >$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+as_write_fail=0
+cat >$CONFIG_STATUS <<_ASEOF || as_write_fail=1
 #! $SHELL
 # Generated by $as_me.
 # Run this file to recreate the current configuration.
@@ -6456,17 +5239,18 @@ cat >$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 debug=false
 ac_cs_recheck=false
 ac_cs_silent=false
-SHELL=\${CONFIG_SHELL-$SHELL}
-_ACEOF
 
-cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-## --------------------- ##
-## M4sh Initialization.  ##
-## --------------------- ##
+SHELL=\${CONFIG_SHELL-$SHELL}
+export SHELL
+_ASEOF
+cat >>$CONFIG_STATUS <<\_ASEOF || as_write_fail=1
+## -------------------- ##
+## M4sh Initialization. ##
+## -------------------- ##
 
 # Be more Bourne compatible
 DUALCASE=1; export DUALCASE # for MKS sh
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
   emulate sh
   NULLCMD=:
   # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
@@ -6474,23 +5258,15 @@ if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
   alias -g '${1+"$@"}'='"$@"'
   setopt NO_GLOB_SUBST
 else
-  case `(set -o) 2>/dev/null` in
-  *posix*) set -o posix ;;
+  case `(set -o) 2>/dev/null` in #(
+  *posix*) :
+    set -o posix ;; #(
+  *) :
+     ;;
 esac
-
 fi
 
 
-
-
-# PATH needs CR
-# Avoid depending upon Character Ranges.
-as_cr_letters='abcdefghijklmnopqrstuvwxyz'
-as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
-as_cr_Letters=$as_cr_letters$as_cr_LETTERS
-as_cr_digits='0123456789'
-as_cr_alnum=$as_cr_Letters$as_cr_digits
-
 as_nl='
 '
 export as_nl
@@ -6498,7 +5274,13 @@ export as_nl
 as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
 as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
-if (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
+# Prefer a ksh shell builtin over an external printf program on Solaris,
+# but without wasting forks for bash or zsh.
+if test -z "$BASH_VERSION$ZSH_VERSION" \
+    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
+  as_echo='print -r --'
+  as_echo_n='print -rn --'
+elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
   as_echo='printf %s\n'
   as_echo_n='printf %s'
 else
@@ -6509,7 +5291,7 @@ else
     as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
     as_echo_n_body='eval
       arg=$1;
-      case $arg in
+      case $arg in #(
       *"$as_nl"*)
 	expr "X$arg" : "X\\(.*\\)$as_nl";
 	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
@@ -6532,13 +5314,6 @@ if test "${PATH_SEPARATOR+set}" != set; then
   }
 fi
 
-# Support unset when possible.
-if ( (MAIL=60; unset MAIL) || exit) >/dev/null 2>&1; then
-  as_unset=unset
-else
-  as_unset=false
-fi
-
 
 # IFS
 # We need space, tab and new line, in precisely that order.  Quoting is
@@ -6548,15 +5323,16 @@ fi
 IFS=" ""	$as_nl"
 
 # Find who we are.  Look in the path if we contain no directory separator.
-case $0 in
+as_myself=
+case $0 in #((
   *[\\/]* ) as_myself=$0 ;;
   *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
 for as_dir in $PATH
 do
   IFS=$as_save_IFS
   test -z "$as_dir" && as_dir=.
-  test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
-done
+    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
+  done
 IFS=$as_save_IFS
 
      ;;
@@ -6568,12 +5344,16 @@ if test "x$as_myself" = x; then
 fi
 if test ! -f "$as_myself"; then
   $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
-  { (exit 1); exit 1; }
+  exit 1
 fi
 
-# Work around bugs in pre-3.0 UWIN ksh.
-for as_var in ENV MAIL MAILPATH
-do ($as_unset $as_var) >/dev/null 2>&1 && $as_unset $as_var
+# Unset variables that we do not need and which cause bugs (e.g. in
+# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
+# suppresses any "Segmentation fault" message there.  '((' could
+# trigger a bug in pdksh 5.2.14.
+for as_var in BASH_ENV ENV MAIL MAILPATH
+do eval test x\${$as_var+set} = xset \
+  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
 done
 PS1='$ '
 PS2='> '
@@ -6585,7 +5365,89 @@ export LC_ALL
 LANGUAGE=C
 export LANGUAGE
 
-# Required to use basename.
+# CDPATH.
+(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
+
+
+# as_fn_error STATUS ERROR [LINENO LOG_FD]
+# ----------------------------------------
+# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
+# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
+# script with STATUS, using 1 if that was 0.
+as_fn_error ()
+{
+  as_status=$1; test $as_status -eq 0 && as_status=1
+  if test "$4"; then
+    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
+  fi
+  $as_echo "$as_me: error: $2" >&2
+  as_fn_exit $as_status
+} # as_fn_error
+
+
+# as_fn_set_status STATUS
+# -----------------------
+# Set $? to STATUS, without forking.
+as_fn_set_status ()
+{
+  return $1
+} # as_fn_set_status
+
+# as_fn_exit STATUS
+# -----------------
+# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
+as_fn_exit ()
+{
+  set +e
+  as_fn_set_status $1
+  exit $1
+} # as_fn_exit
+
+# as_fn_unset VAR
+# ---------------
+# Portably unset VAR.
+as_fn_unset ()
+{
+  { eval $1=; unset $1;}
+}
+as_unset=as_fn_unset
+# as_fn_append VAR VALUE
+# ----------------------
+# Append the text in VALUE to the end of the definition contained in VAR. Take
+# advantage of any shell optimizations that allow amortized linear growth over
+# repeated appends, instead of the typical quadratic growth present in naive
+# implementations.
+if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
+  eval 'as_fn_append ()
+  {
+    eval $1+=\$2
+  }'
+else
+  as_fn_append ()
+  {
+    eval $1=\$$1\$2
+  }
+fi # as_fn_append
+
+# as_fn_arith ARG...
+# ------------------
+# Perform arithmetic evaluation on the ARGs, and store the result in the
+# global $as_val. Take advantage of shells that can avoid forks. The arguments
+# must be portable across $(()) and expr.
+if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
+  eval 'as_fn_arith ()
+  {
+    as_val=$(( $* ))
+  }'
+else
+  as_fn_arith ()
+  {
+    as_val=`expr "$@" || test $? -eq 1`
+  }
+fi # as_fn_arith
+
+
 if expr a : '\(a\)' >/dev/null 2>&1 &&
    test "X`expr 00001 : '.*\(...\)'`" = X001; then
   as_expr=expr
@@ -6599,8 +5461,12 @@ else
   as_basename=false
 fi
 
+if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
+  as_dirname=dirname
+else
+  as_dirname=false
+fi
 
-# Name of the executable.
 as_me=`$as_basename -- "$0" ||
 $as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
 	 X"$0" : 'X\(//\)$' \| \
@@ -6620,76 +5486,25 @@ $as_echo X/"$0" |
 	  }
 	  s/.*/./; q'`
 
-# CDPATH.
-$as_unset CDPATH
-
-
-
-  as_lineno_1=$LINENO
-  as_lineno_2=$LINENO
-  test "x$as_lineno_1" != "x$as_lineno_2" &&
-  test "x`expr $as_lineno_1 + 1`" = "x$as_lineno_2" || {
-
-  # Create $as_me.lineno as a copy of $as_myself, but with $LINENO
-  # uniformly replaced by the line number.  The first 'sed' inserts a
-  # line-number line after each line using $LINENO; the second 'sed'
-  # does the real work.  The second script uses 'N' to pair each
-  # line-number line with the line containing $LINENO, and appends
-  # trailing '-' during substitution so that $LINENO is not a special
-  # case at line end.
-  # (Raja R Harinath suggested sed '=', and Paul Eggert wrote the
-  # scripts with optimization help from Paolo Bonzini.  Blame Lee
-  # E. McMahon (1931-1989) for sed's syntax.  :-)
-  sed -n '
-    p
-    /[$]LINENO/=
-  ' <$as_myself |
-    sed '
-      s/[$]LINENO.*/&-/
-      t lineno
-      b
-      :lineno
-      N
-      :loop
-      s/[$]LINENO\([^'$as_cr_alnum'_].*\n\)\(.*\)/\2\1\2/
-      t loop
-      s/-\n.*//
-    ' >$as_me.lineno &&
-  chmod +x "$as_me.lineno" ||
-    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2
-   { (exit 1); exit 1; }; }
-
-  # Don't try to exec as it changes $[0], causing all sort of problems
-  # (the dirname of $[0] is not the place where we might find the
-  # original and so on.  Autoconf is especially sensitive to this).
-  . "./$as_me.lineno"
-  # Exit status is that of the last command.
-  exit
-}
-
-
-if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
-  as_dirname=dirname
-else
-  as_dirname=false
-fi
+# Avoid depending upon Character Ranges.
+as_cr_letters='abcdefghijklmnopqrstuvwxyz'
+as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
+as_cr_Letters=$as_cr_letters$as_cr_LETTERS
+as_cr_digits='0123456789'
+as_cr_alnum=$as_cr_Letters$as_cr_digits
 
 ECHO_C= ECHO_N= ECHO_T=
-case `echo -n x` in
+case `echo -n x` in #(((((
 -n*)
-  case `echo 'x\c'` in
+  case `echo 'xy\c'` in
   *c*) ECHO_T='	';;	# ECHO_T is single tab character.
-  *)   ECHO_C='\c';;
+  xy)  ECHO_C='\c';;
+  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
+       ECHO_T='	';;
   esac;;
 *)
   ECHO_N='-n';;
 esac
-if expr a : '\(a\)' >/dev/null 2>&1 &&
-   test "X`expr 00001 : '.*\(...\)'`" = X001; then
-  as_expr=expr
-else
-  as_expr=false
-fi
 
 rm -f conf$$ conf$$.exe conf$$.file
 if test -d conf$$.dir; then
@@ -6704,49 +5519,85 @@ if (echo >conf$$.file) 2>/dev/null; then
     # ... but there are two gotchas:
     # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
     # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
-    # In both cases, we have to default to `cp -p'.
+    # In both cases, we have to default to `cp -pR'.
     ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
-      as_ln_s='cp -p'
+      as_ln_s='cp -pR'
   elif ln conf$$.file conf$$ 2>/dev/null; then
     as_ln_s=ln
   else
-    as_ln_s='cp -p'
+    as_ln_s='cp -pR'
   fi
 else
-  as_ln_s='cp -p'
+  as_ln_s='cp -pR'
 fi
 rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
 rmdir conf$$.dir 2>/dev/null
 
+
+# as_fn_mkdir_p
+# -------------
+# Create "$as_dir" as a directory, including parents if necessary.
+as_fn_mkdir_p ()
+{
+
+  case $as_dir in #(
+  -*) as_dir=./$as_dir;;
+  esac
+  test -d "$as_dir" || eval $as_mkdir_p || {
+    as_dirs=
+    while :; do
+      case $as_dir in #(
+      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
+      *) as_qdir=$as_dir;;
+      esac
+      as_dirs="'$as_qdir' $as_dirs"
+      as_dir=`$as_dirname -- "$as_dir" ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
+$as_echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)[^/].*/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\/\)$/{
+	    s//\1/
+	    q
+	  }
+	  /^X\(\/\).*/{
+	    s//\1/
+	    q
+	  }
+	  s/.*/./; q'`
+      test -d "$as_dir" && break
+    done
+    test -z "$as_dirs" || eval "mkdir $as_dirs"
+  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
+
+
+} # as_fn_mkdir_p
 if mkdir -p . 2>/dev/null; then
-  as_mkdir_p=:
+  as_mkdir_p='mkdir -p "$as_dir"'
 else
   test -d ./-p && rmdir ./-p
   as_mkdir_p=false
 fi
 
-if test -x / >/dev/null 2>&1; then
-  as_test_x='test -x'
-else
-  if ls -dL / >/dev/null 2>&1; then
-    as_ls_L_option=L
-  else
-    as_ls_L_option=
-  fi
-  as_test_x='
-    eval sh -c '\''
-      if test -d "$1"; then
-	test -d "$1/.";
-      else
-	case $1 in
-	-*)set "./$1";;
-	esac;
-	case `ls -ld'$as_ls_L_option' "$1" 2>/dev/null` in
-	???[sx]*):;;*)false;;esac;fi
-    '\'' sh
-  '
-fi
-as_executable_p=$as_test_x
+
+# as_fn_executable_p FILE
+# -----------------------
+# Test if FILE is an executable regular file.
+as_fn_executable_p ()
+{
+  test -f "$1" && test -x "$1"
+} # as_fn_executable_p
+as_test_x='test -x'
+as_executable_p=as_fn_executable_p
 
 # Sed expression to map a string onto a valid CPP name.
 as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
@@ -6756,13 +5607,19 @@ as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
 
 
 exec 6>&1
+## ----------------------------------- ##
+## Main body of $CONFIG_STATUS script. ##
+## ----------------------------------- ##
+_ASEOF
+test $as_write_fail = 0 && chmod +x $CONFIG_STATUS || ac_write_fail=1
 
-# Save the log message, to keep $[0] and so on meaningful, and to
+cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
+# Save the log message, to keep $0 and so on meaningful, and to
 # report actual input values of CONFIG_FILES etc. instead of their
 # values after options handling.
 ac_log="
 This file was extended by Grid $as_me 1.0, which was
-generated by GNU Autoconf 2.63.  Invocation command line was
+generated by GNU Autoconf 2.69.  Invocation command line was
 
   CONFIG_FILES    = $CONFIG_FILES
   CONFIG_HEADERS  = $CONFIG_HEADERS
@@ -6794,13 +5651,15 @@ _ACEOF
 
 cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 ac_cs_usage="\
-\`$as_me' instantiates files from templates according to the
-current configuration.
+\`$as_me' instantiates files and other configuration actions
+from templates according to the current configuration.  Unless the files
+and actions are specified as TAGs, all are instantiated by default.
 
-Usage: $0 [OPTION]... [FILE]...
+Usage: $0 [OPTION]... [TAG]...
 
   -h, --help       print this help, then exit
   -V, --version    print version number and configuration settings, then exit
+      --config     print configuration, then exit
   -q, --quiet, --silent
                    do not print progress messages
   -d, --debug      don't remove temporary files
@@ -6819,16 +5678,17 @@ $config_headers
 Configuration commands:
 $config_commands
 
-Report bugs to <bug-autoconf@gnu.org>."
+Report bugs to <paboyle@ph.ed.ac.uk>."
 
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
+ac_cs_config="`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`"
 ac_cs_version="\\
 Grid config.status 1.0
-configured by $0, generated by GNU Autoconf 2.63,
-  with options \\"`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`\\"
+configured by $0, generated by GNU Autoconf 2.69,
+  with options \\"\$ac_cs_config\\"
 
-Copyright (C) 2008 Free Software Foundation, Inc.
+Copyright (C) 2012 Free Software Foundation, Inc.
 This config.status script is free software; the Free Software Foundation
 gives unlimited permission to copy, distribute and modify it."
 
@@ -6846,11 +5706,16 @@ ac_need_defaults=:
 while test $# != 0
 do
   case $1 in
-  --*=*)
+  --*=?*)
     ac_option=`expr "X$1" : 'X\([^=]*\)='`
     ac_optarg=`expr "X$1" : 'X[^=]*=\(.*\)'`
     ac_shift=:
     ;;
+  --*=)
+    ac_option=`expr "X$1" : 'X\([^=]*\)='`
+    ac_optarg=
+    ac_shift=:
+    ;;
   *)
     ac_option=$1
     ac_optarg=$2
@@ -6864,27 +5729,29 @@ do
     ac_cs_recheck=: ;;
   --version | --versio | --versi | --vers | --ver | --ve | --v | -V )
     $as_echo "$ac_cs_version"; exit ;;
+  --config | --confi | --conf | --con | --co | --c )
+    $as_echo "$ac_cs_config"; exit ;;
   --debug | --debu | --deb | --de | --d | -d )
     debug=: ;;
   --file | --fil | --fi | --f )
     $ac_shift
     case $ac_optarg in
     *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
+    '') as_fn_error $? "missing file argument" ;;
     esac
-    CONFIG_FILES="$CONFIG_FILES '$ac_optarg'"
+    as_fn_append CONFIG_FILES " '$ac_optarg'"
     ac_need_defaults=false;;
   --header | --heade | --head | --hea )
     $ac_shift
     case $ac_optarg in
     *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
     esac
-    CONFIG_HEADERS="$CONFIG_HEADERS '$ac_optarg'"
+    as_fn_append CONFIG_HEADERS " '$ac_optarg'"
     ac_need_defaults=false;;
   --he | --h)
     # Conflict between --help and --header
-    { $as_echo "$as_me: error: ambiguous option: $1
-Try \`$0 --help' for more information." >&2
-   { (exit 1); exit 1; }; };;
+    as_fn_error $? "ambiguous option: \`$1'
+Try \`$0 --help' for more information.";;
   --help | --hel | -h )
     $as_echo "$ac_cs_usage"; exit ;;
   -q | -quiet | --quiet | --quie | --qui | --qu | --q \
@@ -6892,11 +5759,10 @@ Try \`$0 --help' for more information." >&2
     ac_cs_silent=: ;;
 
   # This is an error.
-  -*) { $as_echo "$as_me: error: unrecognized option: $1
-Try \`$0 --help' for more information." >&2
-   { (exit 1); exit 1; }; } ;;
+  -*) as_fn_error $? "unrecognized option: \`$1'
+Try \`$0 --help' for more information." ;;
 
-  *) ac_config_targets="$ac_config_targets $1"
+  *) as_fn_append ac_config_targets " $1"
      ac_need_defaults=false ;;
 
   esac
@@ -6913,7 +5779,7 @@ fi
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 if \$ac_cs_recheck; then
-  set X '$SHELL' '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
+  set X $SHELL '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
   shift
   \$as_echo "running CONFIG_SHELL=$SHELL \$*" >&6
   CONFIG_SHELL='$SHELL'
@@ -6955,9 +5821,7 @@ do
     "tests/Makefile") CONFIG_FILES="$CONFIG_FILES tests/Makefile" ;;
     "benchmarks/Makefile") CONFIG_FILES="$CONFIG_FILES benchmarks/Makefile" ;;
 
-  *) { { $as_echo "$as_me:$LINENO: error: invalid argument: $ac_config_target" >&5
-$as_echo "$as_me: error: invalid argument: $ac_config_target" >&2;}
-   { (exit 1); exit 1; }; };;
+  *) as_fn_error $? "invalid argument: \`$ac_config_target'" "$LINENO" 5;;
   esac
 done
 
@@ -6980,26 +5844,24 @@ fi
 # after its creation but before its name has been assigned to `$tmp'.
 $debug ||
 {
-  tmp=
+  tmp= ac_tmp=
   trap 'exit_status=$?
-  { test -z "$tmp" || test ! -d "$tmp" || rm -fr "$tmp"; } && exit $exit_status
+  : "${ac_tmp:=$tmp}"
+  { test ! -d "$ac_tmp" || rm -fr "$ac_tmp"; } && exit $exit_status
 ' 0
-  trap '{ (exit 1); exit 1; }' 1 2 13 15
+  trap 'as_fn_exit 1' 1 2 13 15
 }
 # Create a (secure) tmp directory for tmp files.
 
 {
   tmp=`(umask 077 && mktemp -d "./confXXXXXX") 2>/dev/null` &&
-  test -n "$tmp" && test -d "$tmp"
+  test -d "$tmp"
 }  ||
 {
   tmp=./conf$$-$RANDOM
   (umask 077 && mkdir "$tmp")
-} ||
-{
-   $as_echo "$as_me: cannot create a temporary directory in ." >&2
-   { (exit 1); exit 1; }
-}
+} || as_fn_error $? "cannot create a temporary directory in ." "$LINENO" 5
+ac_tmp=$tmp
 
 # Set up the scripts for CONFIG_FILES section.
 # No need to generate them if there are no CONFIG_FILES.
@@ -7007,7 +5869,13 @@ $debug ||
 if test -n "$CONFIG_FILES"; then
 
 
-ac_cr='
'
+ac_cr=`echo X | tr X '\015'`
+# On cygwin, bash can eat \r inside `` if the user requested igncr.
+# But we know of no other shell where ac_cr would be empty at this
+# point, so we can use a bashism as a fallback.
+if test "x$ac_cr" = x; then
+  eval ac_cr=\$\'\\r\'
+fi
 ac_cs_awk_cr=`$AWK 'BEGIN { print "a\rb" }' </dev/null 2>/dev/null`
 if test "$ac_cs_awk_cr" = "a${ac_cr}b"; then
   ac_cs_awk_cr='\\r'
@@ -7015,7 +5883,7 @@ else
   ac_cs_awk_cr=$ac_cr
 fi
 
-echo 'BEGIN {' >"$tmp/subs1.awk" &&
+echo 'BEGIN {' >"$ac_tmp/subs1.awk" &&
 _ACEOF
 
 
@@ -7024,24 +5892,18 @@ _ACEOF
   echo "$ac_subst_vars" | sed 's/.*/&!$&$ac_delim/' &&
   echo "_ACEOF"
 } >conf$$subs.sh ||
-  { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
-$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
-   { (exit 1); exit 1; }; }
-ac_delim_num=`echo "$ac_subst_vars" | grep -c '$'`
+  as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
+ac_delim_num=`echo "$ac_subst_vars" | grep -c '^'`
 ac_delim='%!_!# '
 for ac_last_try in false false false false false :; do
   . ./conf$$subs.sh ||
-    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
-$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
 
   ac_delim_n=`sed -n "s/.*$ac_delim\$/X/p" conf$$subs.awk | grep -c X`
   if test $ac_delim_n = $ac_delim_num; then
     break
   elif $ac_last_try; then
-    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_STATUS" >&5
-$as_echo "$as_me: error: could not make $CONFIG_STATUS" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
   else
     ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
   fi
@@ -7049,7 +5911,7 @@ done
 rm -f conf$$subs.sh
 
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-cat >>"\$tmp/subs1.awk" <<\\_ACAWK &&
+cat >>"\$ac_tmp/subs1.awk" <<\\_ACAWK &&
 _ACEOF
 sed -n '
 h
@@ -7063,7 +5925,7 @@ s/'"$ac_delim"'$//
 t delim
 :nl
 h
-s/\(.\{148\}\).*/\1/
+s/\(.\{148\}\)..*/\1/
 t more1
 s/["\\]/\\&/g; s/^/"/; s/$/\\n"\\/
 p
@@ -7077,7 +5939,7 @@ s/.\{148\}//
 t nl
 :delim
 h
-s/\(.\{148\}\).*/\1/
+s/\(.\{148\}\)..*/\1/
 t more2
 s/["\\]/\\&/g; s/^/"/; s/$/"/
 p
@@ -7097,7 +5959,7 @@ t delim
 rm -f conf$$subs.awk
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 _ACAWK
-cat >>"\$tmp/subs1.awk" <<_ACAWK &&
+cat >>"\$ac_tmp/subs1.awk" <<_ACAWK &&
   for (key in S) S_is_set[key] = 1
   FS = ""
 
@@ -7129,23 +5991,29 @@ if sed "s/$ac_cr//" < /dev/null > /dev/null 2>&1; then
   sed "s/$ac_cr\$//; s/$ac_cr/$ac_cs_awk_cr/g"
 else
   cat
-fi < "$tmp/subs1.awk" > "$tmp/subs.awk" \
-  || { { $as_echo "$as_me:$LINENO: error: could not setup config files machinery" >&5
-$as_echo "$as_me: error: could not setup config files machinery" >&2;}
-   { (exit 1); exit 1; }; }
+fi < "$ac_tmp/subs1.awk" > "$ac_tmp/subs.awk" \
+  || as_fn_error $? "could not setup config files machinery" "$LINENO" 5
 _ACEOF
 
-# VPATH may cause trouble with some makes, so we remove $(srcdir),
-# ${srcdir} and @srcdir@ from VPATH if srcdir is ".", strip leading and
+# VPATH may cause trouble with some makes, so we remove sole $(srcdir),
+# ${srcdir} and @srcdir@ entries from VPATH if srcdir is ".", strip leading and
 # trailing colons and then remove the whole line if VPATH becomes empty
 # (actually we leave an empty line to preserve line numbers).
 if test "x$srcdir" = x.; then
-  ac_vpsub='/^[	 ]*VPATH[	 ]*=/{
-s/:*\$(srcdir):*/:/
-s/:*\${srcdir}:*/:/
-s/:*@srcdir@:*/:/
-s/^\([^=]*=[	 ]*\):*/\1/
+  ac_vpsub='/^[	 ]*VPATH[	 ]*=[	 ]*/{
+h
+s///
+s/^/:/
+s/[	 ]*$/:/
+s/:\$(srcdir):/:/g
+s/:\${srcdir}:/:/g
+s/:@srcdir@:/:/g
+s/^:*//
 s/:*$//
+x
+s/\(=[	 ]*\).*/\1/
+G
+s/\n//
 s/^[^=]*=[	 ]*$//
 }'
 fi
@@ -7157,7 +6025,7 @@ fi # test -n "$CONFIG_FILES"
 # No need to generate them if there are no CONFIG_HEADERS.
 # This happens for instance with `./config.status Makefile'.
 if test -n "$CONFIG_HEADERS"; then
-cat >"$tmp/defines.awk" <<\_ACAWK ||
+cat >"$ac_tmp/defines.awk" <<\_ACAWK ||
 BEGIN {
 _ACEOF
 
@@ -7169,13 +6037,11 @@ _ACEOF
 # handling of long lines.
 ac_delim='%!_!# '
 for ac_last_try in false false :; do
-  ac_t=`sed -n "/$ac_delim/p" confdefs.h`
-  if test -z "$ac_t"; then
+  ac_tt=`sed -n "/$ac_delim/p" confdefs.h`
+  if test -z "$ac_tt"; then
     break
   elif $ac_last_try; then
-    { { $as_echo "$as_me:$LINENO: error: could not make $CONFIG_HEADERS" >&5
-$as_echo "$as_me: error: could not make $CONFIG_HEADERS" >&2;}
-   { (exit 1); exit 1; }; }
+    as_fn_error $? "could not make $CONFIG_HEADERS" "$LINENO" 5
   else
     ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
   fi
@@ -7260,9 +6126,7 @@ cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 _ACAWK
 _ACEOF
 cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-  { { $as_echo "$as_me:$LINENO: error: could not setup config headers machinery" >&5
-$as_echo "$as_me: error: could not setup config headers machinery" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "could not setup config headers machinery" "$LINENO" 5
 fi # test -n "$CONFIG_HEADERS"
 
 
@@ -7275,9 +6139,7 @@ do
   esac
   case $ac_mode$ac_tag in
   :[FHL]*:*);;
-  :L* | :C*:*) { { $as_echo "$as_me:$LINENO: error: invalid tag $ac_tag" >&5
-$as_echo "$as_me: error: invalid tag $ac_tag" >&2;}
-   { (exit 1); exit 1; }; };;
+  :L* | :C*:*) as_fn_error $? "invalid tag \`$ac_tag'" "$LINENO" 5;;
   :[FH]-) ac_tag=-:-;;
   :[FH]*) ac_tag=$ac_tag:$ac_tag.in;;
   esac
@@ -7296,7 +6158,7 @@ $as_echo "$as_me: error: invalid tag $ac_tag" >&2;}
     for ac_f
     do
       case $ac_f in
-      -) ac_f="$tmp/stdin";;
+      -) ac_f="$ac_tmp/stdin";;
       *) # Look for the file first in the build tree, then in the source tree
 	 # (if the path is not absolute).  The absolute path cannot be DOS-style,
 	 # because $ac_f cannot contain `:'.
@@ -7305,12 +6167,10 @@ $as_echo "$as_me: error: invalid tag $ac_tag" >&2;}
 	   [\\/$]*) false;;
 	   *) test -f "$srcdir/$ac_f" && ac_f="$srcdir/$ac_f";;
 	   esac ||
-	   { { $as_echo "$as_me:$LINENO: error: cannot find input file: $ac_f" >&5
-$as_echo "$as_me: error: cannot find input file: $ac_f" >&2;}
-   { (exit 1); exit 1; }; };;
+	   as_fn_error 1 "cannot find input file: \`$ac_f'" "$LINENO" 5;;
       esac
       case $ac_f in *\'*) ac_f=`$as_echo "$ac_f" | sed "s/'/'\\\\\\\\''/g"`;; esac
-      ac_file_inputs="$ac_file_inputs '$ac_f'"
+      as_fn_append ac_file_inputs " '$ac_f'"
     done
 
     # Let's still pretend it is `configure' which instantiates (i.e., don't
@@ -7321,7 +6181,7 @@ $as_echo "$as_me: error: cannot find input file: $ac_f" >&2;}
 	`' by configure.'
     if test x"$ac_file" != x-; then
       configure_input="$ac_file.  $configure_input"
-      { $as_echo "$as_me:$LINENO: creating $ac_file" >&5
+      { $as_echo "$as_me:${as_lineno-$LINENO}: creating $ac_file" >&5
 $as_echo "$as_me: creating $ac_file" >&6;}
     fi
     # Neutralize special characters interpreted by sed in replacement strings.
@@ -7333,10 +6193,8 @@ $as_echo "$as_me: creating $ac_file" >&6;}
     esac
 
     case $ac_tag in
-    *:-:* | *:-) cat >"$tmp/stdin" \
-      || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; } ;;
+    *:-:* | *:-) cat >"$ac_tmp/stdin" \
+      || as_fn_error $? "could not create $ac_file" "$LINENO" 5 ;;
     esac
     ;;
   esac
@@ -7364,47 +6222,7 @@ $as_echo X"$ac_file" |
 	    q
 	  }
 	  s/.*/./; q'`
-  { as_dir="$ac_dir"
-  case $as_dir in #(
-  -*) as_dir=./$as_dir;;
-  esac
-  test -d "$as_dir" || { $as_mkdir_p && mkdir -p "$as_dir"; } || {
-    as_dirs=
-    while :; do
-      case $as_dir in #(
-      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
-      *) as_qdir=$as_dir;;
-      esac
-      as_dirs="'$as_qdir' $as_dirs"
-      as_dir=`$as_dirname -- "$as_dir" ||
-$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_dir" : 'X\(//\)[^/]' \| \
-	 X"$as_dir" : 'X\(//\)$' \| \
-	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_dir" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      test -d "$as_dir" && break
-    done
-    test -z "$as_dirs" || eval "mkdir $as_dirs"
-  } || test -d "$as_dir" || { { $as_echo "$as_me:$LINENO: error: cannot create directory $as_dir" >&5
-$as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
-   { (exit 1); exit 1; }; }; }
+  as_dir="$ac_dir"; as_fn_mkdir_p
   ac_builddir=.
 
 case "$ac_dir" in
@@ -7461,7 +6279,6 @@ cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 # If the template does not know about datarootdir, expand it.
 # FIXME: This hack should be removed a few years after 2.60.
 ac_datarootdir_hack=; ac_datarootdir_seen=
-
 ac_sed_dataroot='
 /datarootdir/ {
   p
@@ -7471,12 +6288,11 @@ ac_sed_dataroot='
 /@docdir@/p
 /@infodir@/p
 /@localedir@/p
-/@mandir@/p
-'
+/@mandir@/p'
 case `eval "sed -n \"\$ac_sed_dataroot\" $ac_file_inputs"` in
 *datarootdir*) ac_datarootdir_seen=yes;;
 *@datadir@*|*@docdir@*|*@infodir@*|*@localedir@*|*@mandir@*)
-  { $as_echo "$as_me:$LINENO: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
 $as_echo "$as_me: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&2;}
 _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
@@ -7486,7 +6302,7 @@ cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
   s&@infodir@&$infodir&g
   s&@localedir@&$localedir&g
   s&@mandir@&$mandir&g
-    s&\\\${datarootdir}&$datarootdir&g' ;;
+  s&\\\${datarootdir}&$datarootdir&g' ;;
 esac
 _ACEOF
 
@@ -7514,27 +6330,24 @@ s&@INSTALL@&$ac_INSTALL&;t t
 s&@MKDIR_P@&$ac_MKDIR_P&;t t
 $ac_datarootdir_hack
 "
-eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$tmp/subs.awk" >$tmp/out \
-  || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; }
+eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$ac_tmp/subs.awk" \
+  >$ac_tmp/out || as_fn_error $? "could not create $ac_file" "$LINENO" 5
 
 test -z "$ac_datarootdir_hack$ac_datarootdir_seen" &&
-  { ac_out=`sed -n '/\${datarootdir}/p' "$tmp/out"`; test -n "$ac_out"; } &&
-  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' "$tmp/out"`; test -z "$ac_out"; } &&
-  { $as_echo "$as_me:$LINENO: WARNING: $ac_file contains a reference to the variable \`datarootdir'
-which seems to be undefined.  Please make sure it is defined." >&5
+  { ac_out=`sed -n '/\${datarootdir}/p' "$ac_tmp/out"`; test -n "$ac_out"; } &&
+  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' \
+      "$ac_tmp/out"`; test -z "$ac_out"; } &&
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file contains a reference to the variable \`datarootdir'
+which seems to be undefined.  Please make sure it is defined" >&5
 $as_echo "$as_me: WARNING: $ac_file contains a reference to the variable \`datarootdir'
-which seems to be undefined.  Please make sure it is defined." >&2;}
+which seems to be undefined.  Please make sure it is defined" >&2;}
 
-  rm -f "$tmp/stdin"
+  rm -f "$ac_tmp/stdin"
   case $ac_file in
-  -) cat "$tmp/out" && rm -f "$tmp/out";;
-  *) rm -f "$ac_file" && mv "$tmp/out" "$ac_file";;
+  -) cat "$ac_tmp/out" && rm -f "$ac_tmp/out";;
+  *) rm -f "$ac_file" && mv "$ac_tmp/out" "$ac_file";;
   esac \
-  || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; }
+  || as_fn_error $? "could not create $ac_file" "$LINENO" 5
  ;;
   :H)
   #
@@ -7543,27 +6356,21 @@ $as_echo "$as_me: error: could not create $ac_file" >&2;}
   if test x"$ac_file" != x-; then
     {
       $as_echo "/* $configure_input  */" \
-      && eval '$AWK -f "$tmp/defines.awk"' "$ac_file_inputs"
-    } >"$tmp/config.h" \
-      || { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; }
-    if diff "$ac_file" "$tmp/config.h" >/dev/null 2>&1; then
-      { $as_echo "$as_me:$LINENO: $ac_file is unchanged" >&5
+      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs"
+    } >"$ac_tmp/config.h" \
+      || as_fn_error $? "could not create $ac_file" "$LINENO" 5
+    if diff "$ac_file" "$ac_tmp/config.h" >/dev/null 2>&1; then
+      { $as_echo "$as_me:${as_lineno-$LINENO}: $ac_file is unchanged" >&5
 $as_echo "$as_me: $ac_file is unchanged" >&6;}
     else
       rm -f "$ac_file"
-      mv "$tmp/config.h" "$ac_file" \
-	|| { { $as_echo "$as_me:$LINENO: error: could not create $ac_file" >&5
-$as_echo "$as_me: error: could not create $ac_file" >&2;}
-   { (exit 1); exit 1; }; }
+      mv "$ac_tmp/config.h" "$ac_file" \
+	|| as_fn_error $? "could not create $ac_file" "$LINENO" 5
     fi
   else
     $as_echo "/* $configure_input  */" \
-      && eval '$AWK -f "$tmp/defines.awk"' "$ac_file_inputs" \
-      || { { $as_echo "$as_me:$LINENO: error: could not create -" >&5
-$as_echo "$as_me: error: could not create -" >&2;}
-   { (exit 1); exit 1; }; }
+      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs" \
+      || as_fn_error $? "could not create -" "$LINENO" 5
   fi
 # Compute "$ac_file"'s index in $config_headers.
 _am_arg="$ac_file"
@@ -7601,7 +6408,7 @@ $as_echo X"$_am_arg" |
 	  s/.*/./; q'`/stamp-h$_am_stamp_count
  ;;
 
-  :C)  { $as_echo "$as_me:$LINENO: executing $ac_file commands" >&5
+  :C)  { $as_echo "$as_me:${as_lineno-$LINENO}: executing $ac_file commands" >&5
 $as_echo "$as_me: executing $ac_file commands" >&6;}
  ;;
   esac
@@ -7609,7 +6416,7 @@ $as_echo "$as_me: executing $ac_file commands" >&6;}
 
   case $ac_file$ac_mode in
     "depfiles":C) test x"$AMDEP_TRUE" != x"" || {
-  # Autoconf 2.62 quotes --file arguments for eval, but not when files
+  # Older Autoconf quotes --file arguments for eval, but not when files
   # are listed without --file.  Let's play safe and only enable the eval
   # if we detect the quoting.
   case $CONFIG_FILES in
@@ -7622,7 +6429,7 @@ $as_echo "$as_me: executing $ac_file commands" >&6;}
     # Strip MF so we end up with the name of the file.
     mf=`echo "$mf" | sed -e 's/:.*$//'`
     # Check whether this is an Automake generated Makefile or not.
-    # We used to match only the files named `Makefile.in', but
+    # We used to match only the files named 'Makefile.in', but
     # some people rename them; so instead we look at the file content.
     # Grep'ing the first line is not enough: some people post-process
     # each Makefile.in and add a new line on top of each file to say so.
@@ -7656,21 +6463,19 @@ $as_echo X"$mf" |
       continue
     fi
     # Extract the definition of DEPDIR, am__include, and am__quote
-    # from the Makefile without running `make'.
+    # from the Makefile without running 'make'.
     DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
     test -z "$DEPDIR" && continue
     am__include=`sed -n 's/^am__include = //p' < "$mf"`
-    test -z "am__include" && continue
+    test -z "$am__include" && continue
     am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
-    # When using ansi2knr, U may be empty or an underscore; expand it
-    U=`sed -n 's/^U = //p' < "$mf"`
     # Find all dependency output files, they are included files with
     # $(DEPDIR) in their names.  We invoke sed twice because it is the
     # simplest approach to changing $(DEPDIR) to its actual value in the
     # expansion.
     for file in `sed -n "
       s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
-	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g' -e 's/\$U/'"$U"'/g'`; do
+	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
       # Make sure the directory exists.
       test -f "$dirpart/$file" && continue
       fdir=`$as_dirname -- "$file" ||
@@ -7696,47 +6501,7 @@ $as_echo X"$file" |
 	    q
 	  }
 	  s/.*/./; q'`
-      { as_dir=$dirpart/$fdir
-  case $as_dir in #(
-  -*) as_dir=./$as_dir;;
-  esac
-  test -d "$as_dir" || { $as_mkdir_p && mkdir -p "$as_dir"; } || {
-    as_dirs=
-    while :; do
-      case $as_dir in #(
-      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
-      *) as_qdir=$as_dir;;
-      esac
-      as_dirs="'$as_qdir' $as_dirs"
-      as_dir=`$as_dirname -- "$as_dir" ||
-$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_dir" : 'X\(//\)[^/]' \| \
-	 X"$as_dir" : 'X\(//\)$' \| \
-	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_dir" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      test -d "$as_dir" && break
-    done
-    test -z "$as_dirs" || eval "mkdir $as_dirs"
-  } || test -d "$as_dir" || { { $as_echo "$as_me:$LINENO: error: cannot create directory $as_dir" >&5
-$as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
-   { (exit 1); exit 1; }; }; }
+      as_dir=$dirpart/$fdir; as_fn_mkdir_p
       # echo "creating $dirpart/$file"
       echo '# dummy' > "$dirpart/$file"
     done
@@ -7748,15 +6513,12 @@ $as_echo "$as_me: error: cannot create directory $as_dir" >&2;}
 done # for ac_tag
 
 
-{ (exit 0); exit 0; }
+as_fn_exit 0
 _ACEOF
-chmod +x $CONFIG_STATUS
 ac_clean_files=$ac_clean_files_save
 
 test $ac_write_fail = 0 ||
-  { { $as_echo "$as_me:$LINENO: error: write failure creating $CONFIG_STATUS" >&5
-$as_echo "$as_me: error: write failure creating $CONFIG_STATUS" >&2;}
-   { (exit 1); exit 1; }; }
+  as_fn_error $? "write failure creating $CONFIG_STATUS" "$LINENO" 5
 
 
 # configure is writing to config.log, and then calls config.status.
@@ -7777,10 +6539,10 @@ if test "$no_create" != yes; then
   exec 5>>config.log
   # Use ||, not &&, to avoid exiting from the if with $? = 1, which
   # would make configure fail if this is the last instruction.
-  $ac_cs_success || { (exit 1); exit 1; }
+  $ac_cs_success || as_fn_exit 1
 fi
 if test -n "$ac_unrecognized_opts" && test "$enable_option_checking" != no; then
-  { $as_echo "$as_me:$LINENO: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
 $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2;}
 fi
 
diff --git a/depcomp b/depcomp
deleted file mode 120000
index b0ad20c0..00000000
--- a/depcomp
+++ /dev/null
@@ -1 +0,0 @@
-/usr/share/automake-1.11/depcomp
\ No newline at end of file
diff --git a/install-sh b/install-sh
deleted file mode 120000
index 205f21c6..00000000
--- a/install-sh
+++ /dev/null
@@ -1 +0,0 @@
-/usr/share/automake-1.11/install-sh
\ No newline at end of file
diff --git a/lib/Grid_config.h b/lib/Grid_config.h
index be3a4b5e..2397894f 100644
--- a/lib/Grid_config.h
+++ b/lib/Grid_config.h
@@ -2,7 +2,7 @@
 /* lib/Grid_config.h.in.  Generated from configure.ac by autoheader.  */
 
 /* AVX */
-#define AVX1 1
+/* #undef AVX1 */
 
 /* AVX2 */
 /* #undef AVX2 */
@@ -93,11 +93,14 @@
 /* Define to the one symbol short name of this package. */
 #define PACKAGE_TARNAME "grid"
 
+/* Define to the home page for this package. */
+#define PACKAGE_URL ""
+
 /* Define to the version of this package. */
 #define PACKAGE_VERSION "1.0"
 
 /* SSE4 */
-/* #undef SSE4 */
+#define SSE4 1
 
 /* Define to 1 if you have the ANSI C header files. */
 #define STDC_HEADERS 1
diff --git a/lib/Grid_config.h.in b/lib/Grid_config.h.in
index ba21e1be..6f05d6cb 100644
--- a/lib/Grid_config.h.in
+++ b/lib/Grid_config.h.in
@@ -92,6 +92,9 @@
 /* Define to the one symbol short name of this package. */
 #undef PACKAGE_TARNAME
 
+/* Define to the home page for this package. */
+#undef PACKAGE_URL
+
 /* Define to the version of this package. */
 #undef PACKAGE_VERSION
 
diff --git a/missing b/missing
deleted file mode 120000
index 20bc5b0e..00000000
--- a/missing
+++ /dev/null
@@ -1 +0,0 @@
-/usr/share/automake-1.11/missing
\ No newline at end of file

From 28ac219d818e0d164b738b6a04385d922e53fda9 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Wed, 27 May 2015 12:11:44 +0900
Subject: [PATCH 243/429] Included Gpermute in the new Grid_simd.h file style.
 Now tested for SSE4. OK

---
 lib/Grid_simd.h              | 14 ++++++++++
 lib/simd/Grid_avx.h          | 26 +++++++++++++++++-
 lib/simd/Grid_knc.h          | 27 +++++++++++++++++-
 lib/simd/Grid_qpx.h          |  3 +-
 lib/simd/Grid_sse4.h         | 32 ++++++++++++++++++++--
 lib/simd/Grid_vector_types.h | 53 ++----------------------------------
 6 files changed, 99 insertions(+), 56 deletions(-)

diff --git a/lib/Grid_simd.h b/lib/Grid_simd.h
index cccc82e0..26edb4c9 100644
--- a/lib/Grid_simd.h
+++ b/lib/Grid_simd.h
@@ -95,6 +95,20 @@ namespace Grid {
   template<>            inline void zeroit(RealF &arg){ arg=0; };
   template<>            inline void zeroit(RealD &arg){ arg=0; };
   
+
+  //////////////////////////////////////////////////////////
+  // Permute
+  // Permute 0 every ABCDEFGH -> BA DC FE HG
+  // Permute 1 every ABCDEFGH -> CD AB GH EF
+  // Permute 2 every ABCDEFGH -> EFGH ABCD
+  // Permute 3 possible on longer iVector lengths (512bit = 8 double = 16 single)
+  // Permute 4 possible on half precision @512bit vectors.
+  //
+  // Defined inside SIMD specialization files
+  //////////////////////////////////////////////////////////
+  template<class VectorSIMD>
+    inline void Gpermute(VectorSIMD &y,const VectorSIMD &b,int perm);
+
 };
 
 #include <simd/Grid_vector_types.h>
diff --git a/lib/simd/Grid_avx.h b/lib/simd/Grid_avx.h
index f9fa1f85..ec2df44b 100644
--- a/lib/simd/Grid_avx.h
+++ b/lib/simd/Grid_avx.h
@@ -4,7 +4,7 @@
 
   Using intrinsics
 */
-// Time-stamp: <2015-05-22 18:58:27 neo>
+// Time-stamp: <2015-05-27 12:07:15 neo>
 //----------------------------------------------------------------------
 
 #include <immintrin.h>
@@ -383,6 +383,30 @@ namespace Grid {
   typedef __m256d SIMD_Dtype; // Double precision type
   typedef __m256i SIMD_Itype; // Integer type
 
+  // prefecthing 
+  inline void v_prefetch0(int size, const char *ptr){
+    for(int i=0;i<size;i+=64){ //  Define L1 linesize above
+      _mm_prefetch(ptr+i+4096,_MM_HINT_T1);
+      _mm_prefetch(ptr+i+512,_MM_HINT_T0);
+    }
+  }
+  
+  template < typename VectorSIMD > 
+    inline void Gpermute(VectorSIMD &y,const VectorSIMD &b, int perm ) {
+    union { 
+      __m256 f;
+      decltype(VectorSIMD::v) v;
+    } conv;
+    conv.v = b.v;
+    switch(perm){
+    case 3: break; //empty for AVX1/2
+    case 2: conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); break;
+    case 1: conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2));  break; 
+    case 0: conv.f = _mm256_permute2f128_ps(conv.f,conv.f,0x01); break;
+    default: assert(0); break;
+    }
+    y.v=conv.v;
+  };
 
   // Function name aliases
   typedef Optimization::Vsplat   VsplatSIMD;
diff --git a/lib/simd/Grid_knc.h b/lib/simd/Grid_knc.h
index daec3973..bb914270 100644
--- a/lib/simd/Grid_knc.h
+++ b/lib/simd/Grid_knc.h
@@ -4,7 +4,7 @@
 
   Using intrinsics
 */
-// Time-stamp: <2015-05-22 17:12:44 neo>
+// Time-stamp: <2015-05-27 12:08:50 neo>
 //----------------------------------------------------------------------
 
 #include <immintrin.h>
@@ -302,7 +302,32 @@ namespace Grid {
   typedef __m512d SIMD_Dtype; // Double precision type
   typedef __m512i SIMD_Itype; // Integer type
 
+  // prefecth
+  inline void v_prefetch0(int size, const char *ptr){
+    for(int i=0;i<size;i+=64){ //  Define L1 linesize above
+      _mm_prefetch(ptr+i+4096,_MM_HINT_T1);
+      _mm_prefetch(ptr+i+512,_MM_HINT_T0);
+    }
+  }
 
+  // Gpermute utilities consider coalescing into 1 Gpermute
+  template < typename VectorSIMD > 
+    inline void Gpermute(VectorSIMD &y,const VectorSIMD &b, int perm ) {
+    union { 
+      __m512 f;
+      decltype(VectorSIMD::v) v;
+    } conv;
+    conv.v = b.v;
+    switch(perm){
+    case 3:  conv.f = _mm512_swizzle_ps(conv.f,_MM_SWIZ_REG_CDAB); break;
+    case 2:  conv.f = _mm512_swizzle_ps(conv.f,_MM_SWIZ_REG_BADC); break; 
+    case 1 : conv.f = _mm512_permute4f128_ps(conv.f,(_MM_PERM_ENUM)_MM_SHUFFLE(2,3,0,1)); break;
+    case 0 : conv.f = _mm512_permute4f128_ps(conv.f,(_MM_PERM_ENUM)_MM_SHUFFLE(1,0,3,2)); break;
+    default: assert(0); break;
+    }
+    y=conv.v;
+  };
+  
   // Function name aliases
   typedef Optimization::Vsplat   VsplatSIMD;
   typedef Optimization::Vstore   VstoreSIMD;
diff --git a/lib/simd/Grid_qpx.h b/lib/simd/Grid_qpx.h
index dc0251f8..2bd5a20e 100644
--- a/lib/simd/Grid_qpx.h
+++ b/lib/simd/Grid_qpx.h
@@ -4,7 +4,7 @@
 
   Using intrinsics
 */
-// Time-stamp: <2015-05-22 17:29:26 neo>
+// Time-stamp: <2015-05-27 11:30:21 neo>
 //----------------------------------------------------------------------
 
 // lot of undefined functions
@@ -251,6 +251,7 @@ namespace Grid {
   typedef vector4double SIMD_Dtype; // Double precision type
   typedef int SIMD_Itype;           // Integer type
 
+  inline void v_prefetch0(int size, const char *ptr){};
 
   // Function name aliases
   typedef Optimization::Vsplat   VsplatSIMD;
diff --git a/lib/simd/Grid_sse4.h b/lib/simd/Grid_sse4.h
index eb930003..ce2737d5 100644
--- a/lib/simd/Grid_sse4.h
+++ b/lib/simd/Grid_sse4.h
@@ -4,7 +4,7 @@
 
   Using intrinsics
 */
-// Time-stamp: <2015-05-21 18:06:30 neo>
+// Time-stamp: <2015-05-27 12:02:07 neo>
 //----------------------------------------------------------------------
 
 #include <pmmintrin.h>
@@ -221,7 +221,7 @@ namespace Optimization {
   };
 
 
-  
+
 
 
   //////////////////////////////////////////////
@@ -277,6 +277,10 @@ namespace Optimization {
     assert(0);
   }
   
+
+
+
+
   
 }
 
@@ -289,6 +293,28 @@ namespace Grid {
   typedef __m128i SIMD_Itype; // Integer type
 
 
+  inline void v_prefetch0(int size, const char *ptr){};  // prefetch utilities
+
+  // Gpermute function
+  template < typename VectorSIMD > 
+    inline void Gpermute(VectorSIMD &y,const VectorSIMD &b, int perm ) {
+    union { 
+      __m128 f;
+      decltype(VectorSIMD::v) v;
+    } conv;
+    conv.v = b.v;
+    switch(perm){
+    case 3: break; //empty for SSE4
+    case 2: break; //empty for SSE4
+    case 1: conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); break;
+    case 0: conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2)); break;
+    default: assert(0); break;
+    }
+    y.v=conv.v;
+  }; 
+  
+
+
   // Function name aliases
   typedef Optimization::Vsplat   VsplatSIMD;
   typedef Optimization::Vstore   VstoreSIMD;
@@ -296,6 +322,8 @@ namespace Grid {
   typedef Optimization::Vstream  VstreamSIMD;
   template <typename S, typename T> using ReduceSIMD = Optimization::Reduce<S,T>;
 
+ 
+
 
   // Arithmetic operations
   typedef Optimization::Sum         SumSIMD;
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index 41706906..0bedf32f 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -2,7 +2,7 @@
 /*! @file Grid_vector_types.h
   @brief Defines templated class Grid_simd to deal with inner vector types
 */
-// Time-stamp: <2015-05-26 14:08:13 neo>
+// Time-stamp: <2015-05-27 12:04:06 neo>
 //---------------------------------------------------------------------------
 #ifndef GRID_VECTOR_TYPES
 #define GRID_VECTOR_TYPES
@@ -58,56 +58,7 @@ namespace Grid {
   ///////////////////////////////////////////////
 
 
-  // Move to the simd files
-//////////////////////////////////////////////////////////
-// Permute
-// Permute 0 every ABCDEFGH -> BA DC FE HG
-// Permute 1 every ABCDEFGH -> CD AB GH EF
-// Permute 2 every ABCDEFGH -> EFGH ABCD
-// Permute 3 possible on longer iVector lengths (512bit = 8 double = 16 single)
-// Permute 4 possible on half precision @512bit vectors.
-//////////////////////////////////////////////////////////
-template<class vsimd>
-inline void Gpermute(vsimd &y,const vsimd &b,int perm){
-	union { 
-	  SIMD_Ftype f;
-	  decltype(vsimd::v) v;
-	} conv;
-	conv.v = b.v;
-      switch (perm){
-#if defined(AVX1)||defined(AVX2)
-      // 8x32 bits=>3 permutes
-      case 2: conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); break;
-      case 1: conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2)); break;
-      case 0: conv.f = _mm256_permute2f128_ps(conv.f,conv.f,0x01); break;
-#endif
-#ifdef SSE4
-      case 1: conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); break;
-      case 0: conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2));break;
-#endif
-#ifdef AVX512
-	// 16 floats=> permutes
-        // Permute 0 every abcd efgh ijkl mnop -> badc fehg jilk nmpo 
-        // Permute 1 every abcd efgh ijkl mnop -> cdab ghef jkij opmn 
-        // Permute 2 every abcd efgh ijkl mnop -> efgh abcd mnop ijkl
-        // Permute 3 every abcd efgh ijkl mnop -> ijkl mnop abcd efgh
-      case 3: conv.f = _mm512_swizzle_ps(conv.f,_MM_SWIZ_REG_CDAB); break;
-      case 2: conv.f = _mm512_swizzle_ps(conv.f,_MM_SWIZ_REG_BADC); break;
-      case 1: conv.f = _mm512_permute4f128_ps(conv.f,(_MM_PERM_ENUM)_MM_SHUFFLE(2,3,0,1)); break;
-      case 0: conv.f = _mm512_permute4f128_ps(conv.f,(_MM_PERM_ENUM)_MM_SHUFFLE(1,0,3,2)); break;
-#endif
-#ifdef QPX
-#error not implemented
-#endif
-      default: assert(0); break;
-      }
-      y.v=conv.v;
-
- };
-
-///////////////////////////////////////
-
-
+  
 
   /*
     @brief Grid_simd class for the SIMD vector type operations

From be66fdcfab9cd8559f51bb734ebbc20c328b7838 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Wed, 27 May 2015 18:30:11 +0900
Subject: [PATCH 244/429] Check at configure time if CPU supports the requested
 SIMD optimization

---
 .gitignore                                    |   31 +
 README.md                                     |    4 +-
 configure                                     | 1872 ++++++++++++++++-
 configure.ac                                  |   32 +-
 lib/Grid_config.h                             |   33 +
 lib/Grid_config.h.in                          |   33 +
 m4/ax_check_compile_flag.m4                   |   72 +
 m4/ax_ext.m4                                  |  288 +++
 m4/ax_gcc_x86_avx_xgetbv.m4                   |   79 +
 m4/ax_gcc_x86_cpuid.m4                        |   45 +
 .../reconfigure_script                        |    0
 11 files changed, 2475 insertions(+), 14 deletions(-)
 create mode 100644 m4/ax_check_compile_flag.m4
 create mode 100644 m4/ax_ext.m4
 create mode 100644 m4/ax_gcc_x86_avx_xgetbv.m4
 create mode 100644 m4/ax_gcc_x86_cpuid.m4
 rename reconfigure_script => scripts/reconfigure_script (100%)

diff --git a/.gitignore b/.gitignore
index 09504991..82e09bc0 100644
--- a/.gitignore
+++ b/.gitignore
@@ -5,6 +5,7 @@
 *.obj
 *~
 errs
+*#
 
 # Precompiled Headers
 *.gch
@@ -48,3 +49,33 @@ config.status
 /stamp-h1
 /config.sub
 /config.guess
+
+
+# Packages #
+############
+# it's better to unpack these files and commit the raw source
+# git has its own built in compression methods
+*.7z
+*.dmg
+*.gz
+*.iso
+*.jar
+*.rar
+*.tar
+*.zip
+ 
+# Logs and databases #
+######################
+*.log
+*.sql
+*.sqlite
+ 
+# OS generated files #
+######################
+.DS_Store
+.DS_Store?
+._*
+.Spotlight-V100
+.Trashes
+ehthumbs.db
+Thumbs.db
\ No newline at end of file
diff --git a/README.md b/README.md
index e18ca474..fe1453f7 100644
--- a/README.md
+++ b/README.md
@@ -20,7 +20,7 @@ optimally use MPI, OpenMP and SIMD parallelism under the hood. This is a signifi
 for most programmers.
 
 The layout transformations are parametrised by the SIMD vector length. This adapts according to the architecture.
-Presently SSE2 (128 bit) AVX, AVX2 (256 bit) and IMCI and AVX512 (512 bit) targets are supported.
+Presently SSE4 (128 bit) AVX, AVX2 (256 bit) and IMCI and AVX512 (512 bit) targets are supported.
 
 These are presented as 
 
@@ -46,3 +46,5 @@ are examples:
      ./configure CXX=icpc CXXFLAGS="-std=c++11 -O3 -mmic" --enable-simd=AVX512 --host=none
      
      
+For developers:
+Use reconfigure_script in the scripts/ directory to create the autotools environment 
diff --git a/configure b/configure
index ccfea6b7..ea0a71c2 100755
--- a/configure
+++ b/configure
@@ -640,6 +640,13 @@ BUILD_COMMS_MPI_TRUE
 EGREP
 GREP
 CXXCPP
+SIMD_FLAGS
+am__fastdepCC_FALSE
+am__fastdepCC_TRUE
+CCDEPMODE
+ac_ct_CC
+CFLAGS
+CC
 HAVE_CXX11
 RANLIB
 OPENMP_CXXFLAGS
@@ -759,6 +766,8 @@ LDFLAGS
 LIBS
 CPPFLAGS
 CCC
+CC
+CFLAGS
 CXXCPP'
 
 
@@ -1386,7 +1395,7 @@ Optional Features:
   --disable-dependency-tracking
                           speeds up one-time build
   --disable-openmp        do not use OpenMP
-  --enable-simd=SSE|AVX|AVX2|AVX512|MIC
+  --enable-simd=SSE4|AVX|AVX2|AVX512|MIC
                           Select instructions to be SSE4.0, AVX 1.0, AVX
                           2.0+FMA, AVX 512, MIC
   --enable-comms=none|mpi Select communications
@@ -1404,6 +1413,8 @@ Some influential environment variables:
   LIBS        libraries to pass to the linker, e.g. -l<library>
   CPPFLAGS    (Objective) C/C++ preprocessor flags, e.g. -I<include dir> if
               you have headers in a nonstandard directory <include dir>
+  CC          C compiler command
+  CFLAGS      C compiler flags
   CXXCPP      C++ preprocessor
 
 Use these variables to override the choices made by `configure' or to help
@@ -1570,6 +1581,86 @@ fi
 
 } # ac_fn_cxx_try_link
 
+# ac_fn_c_try_compile LINENO
+# --------------------------
+# Try to compile conftest.$ac_ext, and return whether this succeeded.
+ac_fn_c_try_compile ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  rm -f conftest.$ac_objext
+  if { { ac_try="$ac_compile"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compile") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    grep -v '^ *+' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+    mv -f conftest.er1 conftest.err
+  fi
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && {
+	 test -z "$ac_c_werror_flag" ||
+	 test ! -s conftest.err
+       } && test -s conftest.$ac_objext; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+	ac_retval=1
+fi
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_compile
+
+# ac_fn_c_try_run LINENO
+# ----------------------
+# Try to link conftest.$ac_ext, and return whether this succeeded. Assumes
+# that executables *can* be run.
+ac_fn_c_try_run ()
+{
+  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
+  if { { ac_try="$ac_link"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_link") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; } && { ac_try='./conftest$ac_exeext'
+  { { case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_try") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; }; then :
+  ac_retval=0
+else
+  $as_echo "$as_me: program exited with status $ac_status" >&5
+       $as_echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+       ac_retval=$ac_status
+fi
+  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
+  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
+  as_fn_set_status $ac_retval
+
+} # ac_fn_c_try_run
+
 # ac_fn_cxx_try_cpp LINENO
 # ------------------------
 # Try to preprocess conftest.$ac_ext, and return whether this succeeded.
@@ -4166,6 +4257,1749 @@ $as_echo "#define HAVE_CXX11 1" >>confdefs.h
 
   fi
 
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+if test -n "$ac_tool_prefix"; then
+  # Extract the first word of "${ac_tool_prefix}gcc", so it can be a program name with args.
+set dummy ${ac_tool_prefix}gcc; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_CC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$CC"; then
+  ac_cv_prog_CC="$CC" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_CC="${ac_tool_prefix}gcc"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+CC=$ac_cv_prog_CC
+if test -n "$CC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+$as_echo "$CC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+fi
+if test -z "$ac_cv_prog_CC"; then
+  ac_ct_CC=$CC
+  # Extract the first word of "gcc", so it can be a program name with args.
+set dummy gcc; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_ac_ct_CC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$ac_ct_CC"; then
+  ac_cv_prog_ac_ct_CC="$ac_ct_CC" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_ac_ct_CC="gcc"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+ac_ct_CC=$ac_cv_prog_ac_ct_CC
+if test -n "$ac_ct_CC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
+$as_echo "$ac_ct_CC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+  if test "x$ac_ct_CC" = x; then
+    CC=""
+  else
+    case $cross_compiling:$ac_tool_warned in
+yes:)
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+$as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
+ac_tool_warned=yes ;;
+esac
+    CC=$ac_ct_CC
+  fi
+else
+  CC="$ac_cv_prog_CC"
+fi
+
+if test -z "$CC"; then
+          if test -n "$ac_tool_prefix"; then
+    # Extract the first word of "${ac_tool_prefix}cc", so it can be a program name with args.
+set dummy ${ac_tool_prefix}cc; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_CC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$CC"; then
+  ac_cv_prog_CC="$CC" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_CC="${ac_tool_prefix}cc"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+CC=$ac_cv_prog_CC
+if test -n "$CC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+$as_echo "$CC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+  fi
+fi
+if test -z "$CC"; then
+  # Extract the first word of "cc", so it can be a program name with args.
+set dummy cc; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_CC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$CC"; then
+  ac_cv_prog_CC="$CC" # Let the user override the test.
+else
+  ac_prog_rejected=no
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    if test "$as_dir/$ac_word$ac_exec_ext" = "/usr/ucb/cc"; then
+       ac_prog_rejected=yes
+       continue
+     fi
+    ac_cv_prog_CC="cc"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+if test $ac_prog_rejected = yes; then
+  # We found a bogon in the path, so make sure we never use it.
+  set dummy $ac_cv_prog_CC
+  shift
+  if test $# != 0; then
+    # We chose a different compiler from the bogus one.
+    # However, it has the same basename, so the bogon will be chosen
+    # first if we set CC to just the basename; use the full file name.
+    shift
+    ac_cv_prog_CC="$as_dir/$ac_word${1+' '}$@"
+  fi
+fi
+fi
+fi
+CC=$ac_cv_prog_CC
+if test -n "$CC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+$as_echo "$CC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+fi
+if test -z "$CC"; then
+  if test -n "$ac_tool_prefix"; then
+  for ac_prog in cl.exe
+  do
+    # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
+set dummy $ac_tool_prefix$ac_prog; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_CC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$CC"; then
+  ac_cv_prog_CC="$CC" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_CC="$ac_tool_prefix$ac_prog"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+CC=$ac_cv_prog_CC
+if test -n "$CC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
+$as_echo "$CC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+    test -n "$CC" && break
+  done
+fi
+if test -z "$CC"; then
+  ac_ct_CC=$CC
+  for ac_prog in cl.exe
+do
+  # Extract the first word of "$ac_prog", so it can be a program name with args.
+set dummy $ac_prog; ac_word=$2
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
+$as_echo_n "checking for $ac_word... " >&6; }
+if ${ac_cv_prog_ac_ct_CC+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -n "$ac_ct_CC"; then
+  ac_cv_prog_ac_ct_CC="$ac_ct_CC" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+    for ac_exec_ext in '' $ac_executable_extensions; do
+  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_ac_ct_CC="$ac_prog"
+    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+  done
+IFS=$as_save_IFS
+
+fi
+fi
+ac_ct_CC=$ac_cv_prog_ac_ct_CC
+if test -n "$ac_ct_CC"; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
+$as_echo "$ac_ct_CC" >&6; }
+else
+  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+fi
+
+
+  test -n "$ac_ct_CC" && break
+done
+
+  if test "x$ac_ct_CC" = x; then
+    CC=""
+  else
+    case $cross_compiling:$ac_tool_warned in
+yes:)
+{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
+$as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
+ac_tool_warned=yes ;;
+esac
+    CC=$ac_ct_CC
+  fi
+fi
+
+fi
+
+
+test -z "$CC" && { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "no acceptable C compiler found in \$PATH
+See \`config.log' for more details" "$LINENO" 5; }
+
+# Provide some information about the compiler.
+$as_echo "$as_me:${as_lineno-$LINENO}: checking for C compiler version" >&5
+set X $ac_compile
+ac_compiler=$2
+for ac_option in --version -v -V -qversion; do
+  { { ac_try="$ac_compiler $ac_option >&5"
+case "(($ac_try" in
+  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
+  *) ac_try_echo=$ac_try;;
+esac
+eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
+$as_echo "$ac_try_echo"; } >&5
+  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
+  ac_status=$?
+  if test -s conftest.err; then
+    sed '10a\
+... rest of stderr output deleted ...
+         10q' conftest.err >conftest.er1
+    cat conftest.er1 >&5
+  fi
+  rm -f conftest.er1 conftest.err
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }
+done
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C compiler" >&5
+$as_echo_n "checking whether we are using the GNU C compiler... " >&6; }
+if ${ac_cv_c_compiler_gnu+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+#ifndef __GNUC__
+       choke me
+#endif
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_compiler_gnu=yes
+else
+  ac_compiler_gnu=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+ac_cv_c_compiler_gnu=$ac_compiler_gnu
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_c_compiler_gnu" >&5
+$as_echo "$ac_cv_c_compiler_gnu" >&6; }
+if test $ac_compiler_gnu = yes; then
+  GCC=yes
+else
+  GCC=
+fi
+ac_test_CFLAGS=${CFLAGS+set}
+ac_save_CFLAGS=$CFLAGS
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC accepts -g" >&5
+$as_echo_n "checking whether $CC accepts -g... " >&6; }
+if ${ac_cv_prog_cc_g+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_save_c_werror_flag=$ac_c_werror_flag
+   ac_c_werror_flag=yes
+   ac_cv_prog_cc_g=no
+   CFLAGS="-g"
+   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_cv_prog_cc_g=yes
+else
+  CFLAGS=""
+      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+
+else
+  ac_c_werror_flag=$ac_save_c_werror_flag
+	 CFLAGS="-g"
+	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_compile "$LINENO"; then :
+  ac_cv_prog_cc_g=yes
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+   ac_c_werror_flag=$ac_save_c_werror_flag
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_g" >&5
+$as_echo "$ac_cv_prog_cc_g" >&6; }
+if test "$ac_test_CFLAGS" = set; then
+  CFLAGS=$ac_save_CFLAGS
+elif test $ac_cv_prog_cc_g = yes; then
+  if test "$GCC" = yes; then
+    CFLAGS="-g -O2"
+  else
+    CFLAGS="-g"
+  fi
+else
+  if test "$GCC" = yes; then
+    CFLAGS="-O2"
+  else
+    CFLAGS=
+  fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to accept ISO C89" >&5
+$as_echo_n "checking for $CC option to accept ISO C89... " >&6; }
+if ${ac_cv_prog_cc_c89+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_cv_prog_cc_c89=no
+ac_save_CC=$CC
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <stdarg.h>
+#include <stdio.h>
+struct stat;
+/* Most of the following tests are stolen from RCS 5.7's src/conf.sh.  */
+struct buf { int x; };
+FILE * (*rcsopen) (struct buf *, struct stat *, int);
+static char *e (p, i)
+     char **p;
+     int i;
+{
+  return p[i];
+}
+static char *f (char * (*g) (char **, int), char **p, ...)
+{
+  char *s;
+  va_list v;
+  va_start (v,p);
+  s = g (p, va_arg (v,int));
+  va_end (v);
+  return s;
+}
+
+/* OSF 4.0 Compaq cc is some sort of almost-ANSI by default.  It has
+   function prototypes and stuff, but not '\xHH' hex character constants.
+   These don't provoke an error unfortunately, instead are silently treated
+   as 'x'.  The following induces an error, until -std is added to get
+   proper ANSI mode.  Curiously '\x00'!='x' always comes out true, for an
+   array size at least.  It's necessary to write '\x00'==0 to get something
+   that's true only with -std.  */
+int osf4_cc_array ['\x00' == 0 ? 1 : -1];
+
+/* IBM C 6 for AIX is almost-ANSI by default, but it replaces macro parameters
+   inside strings and character constants.  */
+#define FOO(x) 'x'
+int xlc6_cc_array[FOO(a) == 'x' ? 1 : -1];
+
+int test (int i, double x);
+struct s1 {int (*f) (int a);};
+struct s2 {int (*f) (double a);};
+int pairnames (int, char **, FILE *(*)(struct buf *, struct stat *, int), int, int);
+int argc;
+char **argv;
+int
+main ()
+{
+return f (e, argv, 0) != argv[0]  ||  f (e, argv, 1) != argv[1];
+  ;
+  return 0;
+}
+_ACEOF
+for ac_arg in '' -qlanglvl=extc89 -qlanglvl=ansi -std \
+	-Ae "-Aa -D_HPUX_SOURCE" "-Xc -D__EXTENSIONS__"
+do
+  CC="$ac_save_CC $ac_arg"
+  if ac_fn_c_try_compile "$LINENO"; then :
+  ac_cv_prog_cc_c89=$ac_arg
+fi
+rm -f core conftest.err conftest.$ac_objext
+  test "x$ac_cv_prog_cc_c89" != "xno" && break
+done
+rm -f conftest.$ac_ext
+CC=$ac_save_CC
+
+fi
+# AC_CACHE_VAL
+case "x$ac_cv_prog_cc_c89" in
+  x)
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: none needed" >&5
+$as_echo "none needed" >&6; } ;;
+  xno)
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: unsupported" >&5
+$as_echo "unsupported" >&6; } ;;
+  *)
+    CC="$CC $ac_cv_prog_cc_c89"
+    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_c89" >&5
+$as_echo "$ac_cv_prog_cc_c89" >&6; } ;;
+esac
+if test "x$ac_cv_prog_cc_c89" != xno; then :
+
+fi
+
+ac_ext=cpp
+ac_cpp='$CXXCPP $CPPFLAGS'
+ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
+
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC understands -c and -o together" >&5
+$as_echo_n "checking whether $CC understands -c and -o together... " >&6; }
+if ${am_cv_prog_cc_c_o+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+  # Make sure it works both with $CC and with simple cc.
+  # Following AC_PROG_CC_C_O, we do the test twice because some
+  # compilers refuse to overwrite an existing .o file with -o,
+  # though they will create one.
+  am_cv_prog_cc_c_o=yes
+  for am_i in 1 2; do
+    if { echo "$as_me:$LINENO: $CC -c conftest.$ac_ext -o conftest2.$ac_objext" >&5
+   ($CC -c conftest.$ac_ext -o conftest2.$ac_objext) >&5 2>&5
+   ac_status=$?
+   echo "$as_me:$LINENO: \$? = $ac_status" >&5
+   (exit $ac_status); } \
+         && test -f conftest2.$ac_objext; then
+      : OK
+    else
+      am_cv_prog_cc_c_o=no
+      break
+    fi
+  done
+  rm -f core conftest*
+  unset am_i
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_prog_cc_c_o" >&5
+$as_echo "$am_cv_prog_cc_c_o" >&6; }
+if test "$am_cv_prog_cc_c_o" != yes; then
+   # Losing compiler, so override with the script.
+   # FIXME: It is wrong to rewrite CC.
+   # But if we don't then we get into trouble of one sort or another.
+   # A longer-term fix would be to have automake use am__CC in this case,
+   # and then we could set am__CC="\$(top_srcdir)/compile \$(CC)"
+   CC="$am_aux_dir/compile $CC"
+fi
+ac_ext=cpp
+ac_cpp='$CXXCPP $CPPFLAGS'
+ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
+
+
+depcc="$CC"   am_compiler_list=
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
+$as_echo_n "checking dependency style of $depcc... " >&6; }
+if ${am_cv_CC_dependencies_compiler_type+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
+  # We make a subdir and do the tests there.  Otherwise we can end up
+  # making bogus files that we don't know about and never remove.  For
+  # instance it was reported that on HP-UX the gcc test will end up
+  # making a dummy file named 'D' -- because '-MD' means "put the output
+  # in D".
+  rm -rf conftest.dir
+  mkdir conftest.dir
+  # Copy depcomp to subdir because otherwise we won't find it if we're
+  # using a relative directory.
+  cp "$am_depcomp" conftest.dir
+  cd conftest.dir
+  # We will build objects and dependencies in a subdirectory because
+  # it helps to detect inapplicable dependency modes.  For instance
+  # both Tru64's cc and ICC support -MD to output dependencies as a
+  # side effect of compilation, but ICC will put the dependencies in
+  # the current directory while Tru64 will put them in the object
+  # directory.
+  mkdir sub
+
+  am_cv_CC_dependencies_compiler_type=none
+  if test "$am_compiler_list" = ""; then
+     am_compiler_list=`sed -n 's/^#*\([a-zA-Z0-9]*\))$/\1/p' < ./depcomp`
+  fi
+  am__universal=false
+  case " $depcc " in #(
+     *\ -arch\ *\ -arch\ *) am__universal=true ;;
+     esac
+
+  for depmode in $am_compiler_list; do
+    # Setup a source with many dependencies, because some compilers
+    # like to wrap large dependency lists on column 80 (with \), and
+    # we should not choose a depcomp mode which is confused by this.
+    #
+    # We need to recreate these files for each test, as the compiler may
+    # overwrite some of them when testing with obscure command lines.
+    # This happens at least with the AIX C compiler.
+    : > sub/conftest.c
+    for i in 1 2 3 4 5 6; do
+      echo '#include "conftst'$i'.h"' >> sub/conftest.c
+      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
+      # Solaris 10 /bin/sh.
+      echo '/* dummy */' > sub/conftst$i.h
+    done
+    echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
+
+    # We check with '-c' and '-o' for the sake of the "dashmstdout"
+    # mode.  It turns out that the SunPro C++ compiler does not properly
+    # handle '-M -o', and we need to detect this.  Also, some Intel
+    # versions had trouble with output in subdirs.
+    am__obj=sub/conftest.${OBJEXT-o}
+    am__minus_obj="-o $am__obj"
+    case $depmode in
+    gcc)
+      # This depmode causes a compiler race in universal mode.
+      test "$am__universal" = false || continue
+      ;;
+    nosideeffect)
+      # After this tag, mechanisms are not by side-effect, so they'll
+      # only be used when explicitly requested.
+      if test "x$enable_dependency_tracking" = xyes; then
+	continue
+      else
+	break
+      fi
+      ;;
+    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
+      # This compiler won't grok '-c -o', but also, the minuso test has
+      # not run yet.  These depmodes are late enough in the game, and
+      # so weak that their functioning should not be impacted.
+      am__obj=conftest.${OBJEXT-o}
+      am__minus_obj=
+      ;;
+    none) break ;;
+    esac
+    if depmode=$depmode \
+       source=sub/conftest.c object=$am__obj \
+       depfile=sub/conftest.Po tmpdepfile=sub/conftest.TPo \
+       $SHELL ./depcomp $depcc -c $am__minus_obj sub/conftest.c \
+         >/dev/null 2>conftest.err &&
+       grep sub/conftst1.h sub/conftest.Po > /dev/null 2>&1 &&
+       grep sub/conftst6.h sub/conftest.Po > /dev/null 2>&1 &&
+       grep $am__obj sub/conftest.Po > /dev/null 2>&1 &&
+       ${MAKE-make} -s -f confmf > /dev/null 2>&1; then
+      # icc doesn't choke on unknown options, it will just issue warnings
+      # or remarks (even with -Werror).  So we grep stderr for any message
+      # that says an option was ignored or not supported.
+      # When given -MP, icc 7.0 and 7.1 complain thusly:
+      #   icc: Command line warning: ignoring option '-M'; no argument required
+      # The diagnosis changed in icc 8.0:
+      #   icc: Command line remark: option '-MP' not supported
+      if (grep 'ignoring option' conftest.err ||
+          grep 'not supported' conftest.err) >/dev/null 2>&1; then :; else
+        am_cv_CC_dependencies_compiler_type=$depmode
+        break
+      fi
+    fi
+  done
+
+  cd ..
+  rm -rf conftest.dir
+else
+  am_cv_CC_dependencies_compiler_type=none
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CC_dependencies_compiler_type" >&5
+$as_echo "$am_cv_CC_dependencies_compiler_type" >&6; }
+CCDEPMODE=depmode=$am_cv_CC_dependencies_compiler_type
+
+ if
+  test "x$enable_dependency_tracking" != xno \
+  && test "$am_cv_CC_dependencies_compiler_type" = gcc3; then
+  am__fastdepCC_TRUE=
+  am__fastdepCC_FALSE='#'
+else
+  am__fastdepCC_TRUE='#'
+  am__fastdepCC_FALSE=
+fi
+
+
+
+
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for x86 cpuid  output" >&5
+$as_echo_n "checking for x86 cpuid  output... " >&6; }
+if ${ax_cv_gcc_x86_cpuid_+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test "$cross_compiling" = yes; then :
+  ax_cv_gcc_x86_cpuid_=unknown
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <stdio.h>
+int
+main ()
+{
+
+     int op = , eax, ebx, ecx, edx;
+     FILE *f;
+      __asm__("cpuid"
+        : "=a" (eax), "=b" (ebx), "=c" (ecx), "=d" (edx)
+        : "a" (op));
+     f = fopen("conftest_cpuid", "w"); if (!f) return 1;
+     fprintf(f, "%x:%x:%x:%x\n", eax, ebx, ecx, edx);
+     fclose(f);
+     return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_run "$LINENO"; then :
+  ax_cv_gcc_x86_cpuid_=`cat conftest_cpuid`; rm -f conftest_cpuid
+else
+  ax_cv_gcc_x86_cpuid_=unknown; rm -f conftest_cpuid
+fi
+rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
+  conftest.$ac_objext conftest.beam conftest.$ac_ext
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_gcc_x86_cpuid_" >&5
+$as_echo "$ax_cv_gcc_x86_cpuid_" >&6; }
+ac_ext=cpp
+ac_cpp='$CXXCPP $CPPFLAGS'
+ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
+
+
+
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for x86-AVX xgetbv  output" >&5
+$as_echo_n "checking for x86-AVX xgetbv  output... " >&6; }
+if ${ax_cv_gcc_x86_avx_xgetbv_+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test "$cross_compiling" = yes; then :
+  ax_cv_gcc_x86_avx_xgetbv_=unknown
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <stdio.h>
+int
+main ()
+{
+
+     int op = , eax, edx;
+     FILE *f;
+      /* Opcodes for xgetbv */
+      __asm__(".byte 0x0f, 0x01, 0xd0"
+        : "=a" (eax), "=d" (edx),
+        : "c" (op));
+     f = fopen("conftest_xgetbv", "w"); if (!f) return 1;
+     fprintf(f, "%x:%x\n", eax, edx);
+     fclose(f);
+     return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_run "$LINENO"; then :
+  ax_cv_gcc_x86_avx_xgetbv_=`cat conftest_xgetbv`; rm -f conftest_xgetbv
+else
+  ax_cv_gcc_x86_avx_xgetbv_=unknown; rm -f conftest_xgetbv
+fi
+rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
+  conftest.$ac_objext conftest.beam conftest.$ac_ext
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_gcc_x86_avx_xgetbv_" >&5
+$as_echo "$ax_cv_gcc_x86_avx_xgetbv_" >&6; }
+ac_ext=cpp
+ac_cpp='$CXXCPP $CPPFLAGS'
+ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
+
+
+
+
+
+  case $host_cpu in
+    powerpc*)
+      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether altivec is supported" >&5
+$as_echo_n "checking whether altivec is supported... " >&6; }
+if ${ax_cv_have_altivec_ext+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+            if test `/usr/sbin/sysctl -a 2>/dev/null| grep -c hw.optional.altivec` != 0; then
+                if test `/usr/sbin/sysctl -n hw.optional.altivec` = 1; then
+                  ax_cv_have_altivec_ext=yes
+                fi
+            fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_altivec_ext" >&5
+$as_echo "$ax_cv_have_altivec_ext" >&6; }
+
+          if test "$ax_cv_have_altivec_ext" = yes; then
+
+$as_echo "#define HAVE_ALTIVEC /**/" >>confdefs.h
+
+            { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -faltivec" >&5
+$as_echo_n "checking whether C++ compiler accepts -faltivec... " >&6; }
+if ${ax_cv_check_cxxflags___faltivec+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+  ax_check_save_flags=$CXXFLAGS
+  CXXFLAGS="$CXXFLAGS  -faltivec"
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  ax_cv_check_cxxflags___faltivec=yes
+else
+  ax_cv_check_cxxflags___faltivec=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+  CXXFLAGS=$ax_check_save_flags
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___faltivec" >&5
+$as_echo "$ax_cv_check_cxxflags___faltivec" >&6; }
+if test x"$ax_cv_check_cxxflags___faltivec" = xyes; then :
+  SIMD_FLAGS="$SIMD_FLAGS -faltivec"
+else
+  :
+fi
+
+          fi
+    ;;
+
+
+    i[3456]86*|x86_64*|amd64*)
+
+
+
+
+
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for x86 cpuid 0x00000001 output" >&5
+$as_echo_n "checking for x86 cpuid 0x00000001 output... " >&6; }
+if ${ax_cv_gcc_x86_cpuid_0x00000001+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test "$cross_compiling" = yes; then :
+  ax_cv_gcc_x86_cpuid_0x00000001=unknown
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <stdio.h>
+int
+main ()
+{
+
+     int op = 0x00000001, eax, ebx, ecx, edx;
+     FILE *f;
+      __asm__("cpuid"
+        : "=a" (eax), "=b" (ebx), "=c" (ecx), "=d" (edx)
+        : "a" (op));
+     f = fopen("conftest_cpuid", "w"); if (!f) return 1;
+     fprintf(f, "%x:%x:%x:%x\n", eax, ebx, ecx, edx);
+     fclose(f);
+     return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_run "$LINENO"; then :
+  ax_cv_gcc_x86_cpuid_0x00000001=`cat conftest_cpuid`; rm -f conftest_cpuid
+else
+  ax_cv_gcc_x86_cpuid_0x00000001=unknown; rm -f conftest_cpuid
+fi
+rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
+  conftest.$ac_objext conftest.beam conftest.$ac_ext
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_gcc_x86_cpuid_0x00000001" >&5
+$as_echo "$ax_cv_gcc_x86_cpuid_0x00000001" >&6; }
+ac_ext=cpp
+ac_cpp='$CXXCPP $CPPFLAGS'
+ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
+
+
+      ecx=0
+      edx=0
+      ebx=0
+      if test "$ax_cv_gcc_x86_cpuid_0x00000001" != "unknown";
+      then
+        ecx=`echo $ax_cv_gcc_x86_cpuid_0x00000001 | cut -d ":" -f 3`
+        edx=`echo $ax_cv_gcc_x86_cpuid_0x00000001 | cut -d ":" -f 4`
+      fi
+
+
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for x86 cpuid 0x00000007 output" >&5
+$as_echo_n "checking for x86 cpuid 0x00000007 output... " >&6; }
+if ${ax_cv_gcc_x86_cpuid_0x00000007+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test "$cross_compiling" = yes; then :
+  ax_cv_gcc_x86_cpuid_0x00000007=unknown
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <stdio.h>
+int
+main ()
+{
+
+     int op = 0x00000007, eax, ebx, ecx, edx;
+     FILE *f;
+      __asm__("cpuid"
+        : "=a" (eax), "=b" (ebx), "=c" (ecx), "=d" (edx)
+        : "a" (op));
+     f = fopen("conftest_cpuid", "w"); if (!f) return 1;
+     fprintf(f, "%x:%x:%x:%x\n", eax, ebx, ecx, edx);
+     fclose(f);
+     return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_run "$LINENO"; then :
+  ax_cv_gcc_x86_cpuid_0x00000007=`cat conftest_cpuid`; rm -f conftest_cpuid
+else
+  ax_cv_gcc_x86_cpuid_0x00000007=unknown; rm -f conftest_cpuid
+fi
+rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
+  conftest.$ac_objext conftest.beam conftest.$ac_ext
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_gcc_x86_cpuid_0x00000007" >&5
+$as_echo "$ax_cv_gcc_x86_cpuid_0x00000007" >&6; }
+ac_ext=cpp
+ac_cpp='$CXXCPP $CPPFLAGS'
+ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
+
+
+      if test "$ax_cv_gcc_x86_cpuid_0x00000007" != "unknown";
+      then
+        ebx=`echo $ax_cv_gcc_x86_cpuid_0x00000007 | cut -d ":" -f 2`
+      fi
+
+      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether mmx is supported" >&5
+$as_echo_n "checking whether mmx is supported... " >&6; }
+if ${ax_cv_have_mmx_ext+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+        ax_cv_have_mmx_ext=no
+        if test "$((0x$edx>>23&0x01))" = 1; then
+          ax_cv_have_mmx_ext=yes
+        fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_mmx_ext" >&5
+$as_echo "$ax_cv_have_mmx_ext" >&6; }
+
+      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether sse is supported" >&5
+$as_echo_n "checking whether sse is supported... " >&6; }
+if ${ax_cv_have_sse_ext+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+        ax_cv_have_sse_ext=no
+        if test "$((0x$edx>>25&0x01))" = 1; then
+          ax_cv_have_sse_ext=yes
+        fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_sse_ext" >&5
+$as_echo "$ax_cv_have_sse_ext" >&6; }
+
+      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether sse2 is supported" >&5
+$as_echo_n "checking whether sse2 is supported... " >&6; }
+if ${ax_cv_have_sse2_ext+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+        ax_cv_have_sse2_ext=no
+        if test "$((0x$edx>>26&0x01))" = 1; then
+          ax_cv_have_sse2_ext=yes
+        fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_sse2_ext" >&5
+$as_echo "$ax_cv_have_sse2_ext" >&6; }
+
+      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether sse3 is supported" >&5
+$as_echo_n "checking whether sse3 is supported... " >&6; }
+if ${ax_cv_have_sse3_ext+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+        ax_cv_have_sse3_ext=no
+        if test "$((0x$ecx&0x01))" = 1; then
+          ax_cv_have_sse3_ext=yes
+        fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_sse3_ext" >&5
+$as_echo "$ax_cv_have_sse3_ext" >&6; }
+
+      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether ssse3 is supported" >&5
+$as_echo_n "checking whether ssse3 is supported... " >&6; }
+if ${ax_cv_have_ssse3_ext+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+        ax_cv_have_ssse3_ext=no
+        if test "$((0x$ecx>>9&0x01))" = 1; then
+          ax_cv_have_ssse3_ext=yes
+        fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_ssse3_ext" >&5
+$as_echo "$ax_cv_have_ssse3_ext" >&6; }
+
+      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether sse4.1 is supported" >&5
+$as_echo_n "checking whether sse4.1 is supported... " >&6; }
+if ${ax_cv_have_sse41_ext+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+        ax_cv_have_sse41_ext=no
+        if test "$((0x$ecx>>19&0x01))" = 1; then
+          ax_cv_have_sse41_ext=yes
+        fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_sse41_ext" >&5
+$as_echo "$ax_cv_have_sse41_ext" >&6; }
+
+      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether sse4.2 is supported" >&5
+$as_echo_n "checking whether sse4.2 is supported... " >&6; }
+if ${ax_cv_have_sse42_ext+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+        ax_cv_have_sse42_ext=no
+        if test "$((0x$ecx>>20&0x01))" = 1; then
+          ax_cv_have_sse42_ext=yes
+        fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_sse42_ext" >&5
+$as_echo "$ax_cv_have_sse42_ext" >&6; }
+
+      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether avx is supported by processor" >&5
+$as_echo_n "checking whether avx is supported by processor... " >&6; }
+if ${ax_cv_have_avx_cpu_ext+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+        ax_cv_have_avx_cpu_ext=no
+        if test "$((0x$ecx>>28&0x01))" = 1; then
+          ax_cv_have_avx_cpu_ext=yes
+        fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_avx_cpu_ext" >&5
+$as_echo "$ax_cv_have_avx_cpu_ext" >&6; }
+
+      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether avx2 is supported by processor" >&5
+$as_echo_n "checking whether avx2 is supported by processor... " >&6; }
+if ${ax_cv_have_avx2_cpu_ext+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+        ax_cv_have_avx2_cpu_ext=no
+        if test "$((0x$ebx>>5&0x01))" = 1; then
+          ax_cv_have_avx2_cpu_ext=yes
+        fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_avx2_cpu_ext" >&5
+$as_echo "$ax_cv_have_avx2_cpu_ext" >&6; }
+
+
+      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether fma is supported by processor" >&5
+$as_echo_n "checking whether fma is supported by processor... " >&6; }
+if ${ax_cv_have_fma_cpu_ext+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+        ax_cv_have_fma_cpu_ext=no
+        if test "$((0x$ecx>>12&0x01))" = 1; then
+          ax_cv_have_fma_cpu_ext=yes
+        fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_fma_cpu_ext" >&5
+$as_echo "$ax_cv_have_fma_cpu_ext" >&6; }
+
+
+      if test x"$ax_cv_have_avx_cpu_ext" = x"yes"; then
+
+ac_ext=c
+ac_cpp='$CPP $CPPFLAGS'
+ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_c_compiler_gnu
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for x86-AVX xgetbv 0x00000000 output" >&5
+$as_echo_n "checking for x86-AVX xgetbv 0x00000000 output... " >&6; }
+if ${ax_cv_gcc_x86_avx_xgetbv_0x00000000+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  if test "$cross_compiling" = yes; then :
+  ax_cv_gcc_x86_avx_xgetbv_0x00000000=unknown
+else
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+#include <stdio.h>
+int
+main ()
+{
+
+     int op = 0x00000000, eax, edx;
+     FILE *f;
+      /* Opcodes for xgetbv */
+      __asm__(".byte 0x0f, 0x01, 0xd0"
+        : "=a" (eax), "=d" (edx),
+        : "c" (op));
+     f = fopen("conftest_xgetbv", "w"); if (!f) return 1;
+     fprintf(f, "%x:%x\n", eax, edx);
+     fclose(f);
+     return 0;
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_run "$LINENO"; then :
+  ax_cv_gcc_x86_avx_xgetbv_0x00000000=`cat conftest_xgetbv`; rm -f conftest_xgetbv
+else
+  ax_cv_gcc_x86_avx_xgetbv_0x00000000=unknown; rm -f conftest_xgetbv
+fi
+rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
+  conftest.$ac_objext conftest.beam conftest.$ac_ext
+fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_gcc_x86_avx_xgetbv_0x00000000" >&5
+$as_echo "$ax_cv_gcc_x86_avx_xgetbv_0x00000000" >&6; }
+ac_ext=cpp
+ac_cpp='$CXXCPP $CPPFLAGS'
+ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
+ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
+ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
+
+
+
+        xgetbv_eax="0"
+        if test x"$ax_cv_gcc_x86_avx_xgetbv_0x00000000" != x"unknown"; then
+          xgetbv_eax=`echo $ax_cv_gcc_x86_avx_xgetbv_0x00000000 | cut -d ":" -f 1`
+        fi
+
+        { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether avx is supported by operating system" >&5
+$as_echo_n "checking whether avx is supported by operating system... " >&6; }
+if ${ax_cv_have_avx_ext+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+          ax_cv_have_avx_ext=no
+
+          if test "$((0x$ecx>>27&0x01))" = 1; then
+            if test "$((0x$xgetbv_eax&0x6))" = 6; then
+              ax_cv_have_avx_ext=yes
+            fi
+          fi
+
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_avx_ext" >&5
+$as_echo "$ax_cv_have_avx_ext" >&6; }
+        if test x"$ax_cv_have_avx_ext" = x"no"; then
+          { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports AVX, but your operating system doesn't" >&5
+$as_echo "$as_me: WARNING: Your processor supports AVX, but your operating system doesn't" >&2;}
+        fi
+      fi
+
+      if test "$ax_cv_have_mmx_ext" = yes; then
+        { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -mmmx" >&5
+$as_echo_n "checking whether C++ compiler accepts -mmmx... " >&6; }
+if ${ax_cv_check_cxxflags___mmmx+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+  ax_check_save_flags=$CXXFLAGS
+  CXXFLAGS="$CXXFLAGS  -mmmx"
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  ax_cv_check_cxxflags___mmmx=yes
+else
+  ax_cv_check_cxxflags___mmmx=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+  CXXFLAGS=$ax_check_save_flags
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___mmmx" >&5
+$as_echo "$ax_cv_check_cxxflags___mmmx" >&6; }
+if test x"$ax_cv_check_cxxflags___mmmx" = xyes; then :
+  ax_cv_support_mmx_ext=yes
+else
+  :
+fi
+
+        if test x"$ax_cv_support_mmx_ext" = x"yes"; then
+          SIMD_FLAGS="$SIMD_FLAGS -mmmx"
+
+$as_echo "#define HAVE_MMX /**/" >>confdefs.h
+
+        else
+          { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports mmx instructions but not your compiler, can you try another compiler?" >&5
+$as_echo "$as_me: WARNING: Your processor supports mmx instructions but not your compiler, can you try another compiler?" >&2;}
+        fi
+      fi
+
+      if test "$ax_cv_have_sse_ext" = yes; then
+        { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -msse" >&5
+$as_echo_n "checking whether C++ compiler accepts -msse... " >&6; }
+if ${ax_cv_check_cxxflags___msse+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+  ax_check_save_flags=$CXXFLAGS
+  CXXFLAGS="$CXXFLAGS  -msse"
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  ax_cv_check_cxxflags___msse=yes
+else
+  ax_cv_check_cxxflags___msse=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+  CXXFLAGS=$ax_check_save_flags
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___msse" >&5
+$as_echo "$ax_cv_check_cxxflags___msse" >&6; }
+if test x"$ax_cv_check_cxxflags___msse" = xyes; then :
+  ax_cv_support_sse_ext=yes
+else
+  :
+fi
+
+        if test x"$ax_cv_support_sse_ext" = x"yes"; then
+          SIMD_FLAGS="$SIMD_FLAGS -msse"
+
+$as_echo "#define HAVE_SSE /**/" >>confdefs.h
+
+        else
+          { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports sse instructions but not your compiler, can you try another compiler?" >&5
+$as_echo "$as_me: WARNING: Your processor supports sse instructions but not your compiler, can you try another compiler?" >&2;}
+        fi
+      fi
+
+      if test "$ax_cv_have_sse2_ext" = yes; then
+        { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -msse2" >&5
+$as_echo_n "checking whether C++ compiler accepts -msse2... " >&6; }
+if ${ax_cv_check_cxxflags___msse2+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+  ax_check_save_flags=$CXXFLAGS
+  CXXFLAGS="$CXXFLAGS  -msse2"
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  ax_cv_check_cxxflags___msse2=yes
+else
+  ax_cv_check_cxxflags___msse2=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+  CXXFLAGS=$ax_check_save_flags
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___msse2" >&5
+$as_echo "$ax_cv_check_cxxflags___msse2" >&6; }
+if test x"$ax_cv_check_cxxflags___msse2" = xyes; then :
+  ax_cv_support_sse2_ext=yes
+else
+  :
+fi
+
+        if test x"$ax_cv_support_sse2_ext" = x"yes"; then
+          SIMD_FLAGS="$SIMD_FLAGS -msse2"
+
+$as_echo "#define HAVE_SSE2 /**/" >>confdefs.h
+
+        else
+          { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports sse2 instructions but not your compiler, can you try another compiler?" >&5
+$as_echo "$as_me: WARNING: Your processor supports sse2 instructions but not your compiler, can you try another compiler?" >&2;}
+        fi
+      fi
+
+      if test "$ax_cv_have_sse3_ext" = yes; then
+        { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -msse3" >&5
+$as_echo_n "checking whether C++ compiler accepts -msse3... " >&6; }
+if ${ax_cv_check_cxxflags___msse3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+  ax_check_save_flags=$CXXFLAGS
+  CXXFLAGS="$CXXFLAGS  -msse3"
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  ax_cv_check_cxxflags___msse3=yes
+else
+  ax_cv_check_cxxflags___msse3=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+  CXXFLAGS=$ax_check_save_flags
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___msse3" >&5
+$as_echo "$ax_cv_check_cxxflags___msse3" >&6; }
+if test x"$ax_cv_check_cxxflags___msse3" = xyes; then :
+  ax_cv_support_sse3_ext=yes
+else
+  :
+fi
+
+        if test x"$ax_cv_support_sse3_ext" = x"yes"; then
+          SIMD_FLAGS="$SIMD_FLAGS -msse3"
+
+$as_echo "#define HAVE_SSE3 /**/" >>confdefs.h
+
+        else
+          { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports sse3 instructions but not your compiler, can you try another compiler?" >&5
+$as_echo "$as_me: WARNING: Your processor supports sse3 instructions but not your compiler, can you try another compiler?" >&2;}
+        fi
+      fi
+
+      if test "$ax_cv_have_ssse3_ext" = yes; then
+        { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -mssse3" >&5
+$as_echo_n "checking whether C++ compiler accepts -mssse3... " >&6; }
+if ${ax_cv_check_cxxflags___mssse3+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+  ax_check_save_flags=$CXXFLAGS
+  CXXFLAGS="$CXXFLAGS  -mssse3"
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  ax_cv_check_cxxflags___mssse3=yes
+else
+  ax_cv_check_cxxflags___mssse3=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+  CXXFLAGS=$ax_check_save_flags
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___mssse3" >&5
+$as_echo "$ax_cv_check_cxxflags___mssse3" >&6; }
+if test x"$ax_cv_check_cxxflags___mssse3" = xyes; then :
+  ax_cv_support_ssse3_ext=yes
+else
+  :
+fi
+
+        if test x"$ax_cv_support_ssse3_ext" = x"yes"; then
+          SIMD_FLAGS="$SIMD_FLAGS -mssse3"
+
+$as_echo "#define HAVE_SSSE3 /**/" >>confdefs.h
+
+        else
+          { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports ssse3 instructions but not your compiler, can you try another compiler?" >&5
+$as_echo "$as_me: WARNING: Your processor supports ssse3 instructions but not your compiler, can you try another compiler?" >&2;}
+        fi
+      fi
+
+      if test "$ax_cv_have_sse41_ext" = yes; then
+        { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -msse4.1" >&5
+$as_echo_n "checking whether C++ compiler accepts -msse4.1... " >&6; }
+if ${ax_cv_check_cxxflags___msse4_1+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+  ax_check_save_flags=$CXXFLAGS
+  CXXFLAGS="$CXXFLAGS  -msse4.1"
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  ax_cv_check_cxxflags___msse4_1=yes
+else
+  ax_cv_check_cxxflags___msse4_1=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+  CXXFLAGS=$ax_check_save_flags
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___msse4_1" >&5
+$as_echo "$ax_cv_check_cxxflags___msse4_1" >&6; }
+if test x"$ax_cv_check_cxxflags___msse4_1" = xyes; then :
+  ax_cv_support_sse41_ext=yes
+else
+  :
+fi
+
+        if test x"$ax_cv_support_sse41_ext" = x"yes"; then
+          SIMD_FLAGS="$SIMD_FLAGS -msse4.1"
+
+$as_echo "#define HAVE_SSE4_1 /**/" >>confdefs.h
+
+        else
+          { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports sse4.1 instructions but not your compiler, can you try another compiler?" >&5
+$as_echo "$as_me: WARNING: Your processor supports sse4.1 instructions but not your compiler, can you try another compiler?" >&2;}
+        fi
+      fi
+
+      if test "$ax_cv_have_sse42_ext" = yes; then
+        { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -msse4.2" >&5
+$as_echo_n "checking whether C++ compiler accepts -msse4.2... " >&6; }
+if ${ax_cv_check_cxxflags___msse4_2+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+  ax_check_save_flags=$CXXFLAGS
+  CXXFLAGS="$CXXFLAGS  -msse4.2"
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  ax_cv_check_cxxflags___msse4_2=yes
+else
+  ax_cv_check_cxxflags___msse4_2=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+  CXXFLAGS=$ax_check_save_flags
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___msse4_2" >&5
+$as_echo "$ax_cv_check_cxxflags___msse4_2" >&6; }
+if test x"$ax_cv_check_cxxflags___msse4_2" = xyes; then :
+  ax_cv_support_sse42_ext=yes
+else
+  :
+fi
+
+        if test x"$ax_cv_support_sse42_ext" = x"yes"; then
+          SIMD_FLAGS="$SIMD_FLAGS -msse4.2"
+
+$as_echo "#define HAVE_SSE4_2 /**/" >>confdefs.h
+
+        else
+          { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports sse4.2 instructions but not your compiler, can you try another compiler?" >&5
+$as_echo "$as_me: WARNING: Your processor supports sse4.2 instructions but not your compiler, can you try another compiler?" >&2;}
+        fi
+      fi
+
+      if test "$ax_cv_have_avx_ext" = yes; then
+        { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -mavx" >&5
+$as_echo_n "checking whether C++ compiler accepts -mavx... " >&6; }
+if ${ax_cv_check_cxxflags___mavx+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+  ax_check_save_flags=$CXXFLAGS
+  CXXFLAGS="$CXXFLAGS  -mavx"
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  ax_cv_check_cxxflags___mavx=yes
+else
+  ax_cv_check_cxxflags___mavx=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+  CXXFLAGS=$ax_check_save_flags
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___mavx" >&5
+$as_echo "$ax_cv_check_cxxflags___mavx" >&6; }
+if test x"$ax_cv_check_cxxflags___mavx" = xyes; then :
+  ax_cv_support_avx_ext=yes
+else
+  :
+fi
+
+        if test x"$ax_cv_support_avx_ext" = x"yes"; then
+          SIMD_FLAGS="$SIMD_FLAGS -mavx"
+
+$as_echo "#define HAVE_AVX /**/" >>confdefs.h
+
+        else
+          { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports avx instructions but not your compiler, can you try another compiler?" >&5
+$as_echo "$as_me: WARNING: Your processor supports avx instructions but not your compiler, can you try another compiler?" >&2;}
+        fi
+      fi
+
+       if test "$ax_cv_have_avx2_ext" = yes; then
+         { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -mavx2" >&5
+$as_echo_n "checking whether C++ compiler accepts -mavx2... " >&6; }
+if ${ax_cv_check_cxxflags___mavx2+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+  ax_check_save_flags=$CXXFLAGS
+  CXXFLAGS="$CXXFLAGS  -mavx2"
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  ax_cv_check_cxxflags___mavx2=yes
+else
+  ax_cv_check_cxxflags___mavx2=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+  CXXFLAGS=$ax_check_save_flags
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___mavx2" >&5
+$as_echo "$ax_cv_check_cxxflags___mavx2" >&6; }
+if test x"$ax_cv_check_cxxflags___mavx2" = xyes; then :
+  ax_cv_support_avx2_ext=yes
+else
+  :
+fi
+
+         if test x"$ax_cv_support_avx2_ext" = x"yes"; then
+           SIMD_FLAGS="$SIMD_FLAGS -mavx2"
+
+$as_echo "#define HAVE_AVX2 /**/" >>confdefs.h
+
+         else
+           { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports avx2 instructions but not your compiler, can you try another compiler?" >&5
+$as_echo "$as_me: WARNING: Your processor supports avx2 instructions but not your compiler, can you try another compiler?" >&2;}
+         fi
+        fi
+
+       if test "$ax_cv_have_fma_ext" = yes; then
+         { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -mfma" >&5
+$as_echo_n "checking whether C++ compiler accepts -mfma... " >&6; }
+if ${ax_cv_check_cxxflags___mfma+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+
+  ax_check_save_flags=$CXXFLAGS
+  CXXFLAGS="$CXXFLAGS  -mfma"
+  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_cxx_try_compile "$LINENO"; then :
+  ax_cv_check_cxxflags___mfma=yes
+else
+  ax_cv_check_cxxflags___mfma=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
+  CXXFLAGS=$ax_check_save_flags
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___mfma" >&5
+$as_echo "$ax_cv_check_cxxflags___mfma" >&6; }
+if test x"$ax_cv_check_cxxflags___mfma" = xyes; then :
+  ax_cv_support_fma_ext=yes
+else
+  :
+fi
+
+         if test x"$ax_cv_support_fma_ext" = x"yes"; then
+           SIMD_FLAGS="$SIMD_FLAGS -mfma"
+
+$as_echo "#define HAVE_FMA /**/" >>confdefs.h
+
+         else
+           { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports fma instructions but not your compiler, can you try another compiler?" >&5
+$as_echo "$as_me: WARNING: Your processor supports fma instructions but not your compiler, can you try another compiler?" >&2;}
+         fi
+      fi
+
+  ;;
+  esac
+
+
 
 
 # Checks for libraries.
@@ -4834,30 +6668,48 @@ else
 fi
 
 
+supported=no
+
 case ${ac_SIMD} in
      SSE4)
        echo Configuring for SSE4
-
+       if test x"$ax_cv_support_ssse3_ext" = x"yes"; then
 $as_echo "#define SSE4 1" >>confdefs.h
 
+       supported=yes
+       else
+	{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor does not support SSE4 instructions" >&5
+$as_echo "$as_me: WARNING: Your processor does not support SSE4 instructions" >&2;}
+       fi
      ;;
      AVX)
        echo Configuring for AVX
-
+       if test x"$ax_cv_support_avx_ext" = x"yes"; then
 $as_echo "#define AVX1 1" >>confdefs.h
 
+       supported=yes
+       else
+       	{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor does not support AVX instructions" >&5
+$as_echo "$as_me: WARNING: Your processor does not support AVX instructions" >&2;}
+       fi
      ;;
      AVX2)
        echo Configuring for AVX2
-
+       if test x"$ax_cv_support_avx2_ext" = x"yes"; then
 $as_echo "#define AVX2 1" >>confdefs.h
 
+       supported=yes
+       else
+       { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor does not support AVX2 instructions" >&5
+$as_echo "$as_me: WARNING: Your processor does not support AVX2 instructions" >&2;}
+       fi
      ;;
      AVX512|MIC)
        echo Configuring for AVX512 and MIC
 
 $as_echo "#define AVX512 1" >>confdefs.h
 
+       supported=yes
      ;;
      *)
      as_fn_error $? "${ac_SIMD} unsupported --enable-simd option" "$LINENO" 5;
@@ -5067,7 +6919,9 @@ ac_config_files="$ac_config_files docs/doxy.cfg"
 
 fi
 
-
+echo
+echo Creating configuration files
+echo :::::::::::::::::::::::::::::::::::::::::::
 ac_config_files="$ac_config_files Makefile"
 
 ac_config_files="$ac_config_files lib/Makefile"
@@ -5209,6 +7063,10 @@ if test -z "${am__fastdepCXX_TRUE}" && test -z "${am__fastdepCXX_FALSE}"; then
   as_fn_error $? "conditional \"am__fastdepCXX\" was never defined.
 Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
+if test -z "${am__fastdepCC_TRUE}" && test -z "${am__fastdepCC_FALSE}"; then
+  as_fn_error $? "conditional \"am__fastdepCC\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
+fi
 if test -z "${BUILD_COMMS_MPI_TRUE}" && test -z "${BUILD_COMMS_MPI_FALSE}"; then
   as_fn_error $? "conditional \"BUILD_COMMS_MPI\" was never defined.
 Usually this means the macro was only invoked conditionally." "$LINENO" 5
@@ -6561,9 +8419,9 @@ The following features are enabled:
 - os (target)                   : $target_os
 - build DOXYGEN documentation   : `if test "x$enable_doc" = xyes; then echo yes; else echo no; fi`
 - graphs and diagrams           : `if test "x$enable_dot" = xyes; then echo yes; else echo no; fi`
-
+- Supported SIMD flags          : $SIMD_FLAGS
 ----------------------------------------------------------
-- enabled simd support          : ${ac_SIMD}
+- enabled simd support          : ${ac_SIMD}   (supported: $supported )
 - communications type           : ${ac_COMMS}
 
 
diff --git a/configure.ac b/configure.ac
index 5dcbea36..00622bb8 100644
--- a/configure.ac
+++ b/configure.ac
@@ -3,7 +3,7 @@
 #
 # Project Grid package  
 # 
-# Time-stamp: <2015-05-26 17:18:54 neo>
+# Time-stamp: <2015-05-27 18:29:04 neo>
 
 AC_PREREQ([2.63])
 AC_INIT([Grid], [1.0], [paboyle@ph.ed.ac.uk])
@@ -27,7 +27,7 @@ AC_PROG_CXX
 AC_OPENMP
 AC_PROG_RANLIB
 AX_CXX_COMPILE_STDCXX_11(noext, mandatory)
-
+AX_EXT
 
 # Checks for libraries.
 #AX_GCC_VAR_ATTRIBUTE(aligned)
@@ -69,26 +69,44 @@ Info at: http://www.mpfr.org/)])
 
 
 
-AC_ARG_ENABLE([simd],[AC_HELP_STRING([--enable-simd=SSE|AVX|AVX2|AVX512|MIC],\
+AC_ARG_ENABLE([simd],[AC_HELP_STRING([--enable-simd=SSE4|AVX|AVX2|AVX512|MIC],\
 	[Select instructions to be SSE4.0, AVX 1.0, AVX 2.0+FMA, AVX 512, MIC])],\
 	[ac_SIMD=${enable_simd}],[ac_SIMD=AVX2])
 
+supported=no
+
 case ${ac_SIMD} in
      SSE4)
        echo Configuring for SSE4
+       if test x"$ax_cv_support_ssse3_ext" = x"yes"; then  dnl minimal support for SSE4
        AC_DEFINE([SSE4],[1],[SSE4] )
+       supported=yes
+       else
+	AC_MSG_WARN([Your processor does not support SSE4 instructions])
+       fi
      ;;
      AVX)
        echo Configuring for AVX
+       if test x"$ax_cv_support_avx_ext" = x"yes"; then  dnl minimal support for AVX
        AC_DEFINE([AVX1],[1],[AVX] )
+       supported=yes			  
+       else
+       	AC_MSG_WARN([Your processor does not support AVX instructions])
+       fi
      ;;
      AVX2)
        echo Configuring for AVX2
+       if test x"$ax_cv_support_avx2_ext" = x"yes"; then  dnl minimal support for AVX2
        AC_DEFINE([AVX2],[1],[AVX2] )
+       supported=yes
+       else
+       AC_MSG_WARN([Your processor does not support AVX2 instructions])
+       fi
      ;;
      AVX512|MIC)
        echo Configuring for AVX512 and MIC
        AC_DEFINE([AVX512],[1],[AVX512] )
+       supported=yes
      ;;
      *)
      AC_MSG_ERROR([${ac_SIMD} unsupported --enable-simd option]); 
@@ -129,7 +147,9 @@ then
 AC_CONFIG_FILES([docs/doxy.cfg])
 fi
 
-
+echo
+echo Creating configuration files
+echo :::::::::::::::::::::::::::::::::::::::::::
 AC_CONFIG_FILES(Makefile)
 AC_CONFIG_FILES(lib/Makefile)
 AC_CONFIG_FILES(tests/Makefile)
@@ -150,9 +170,9 @@ The following features are enabled:
 - os (target)                   : $target_os
 - build DOXYGEN documentation   : `if test "x$enable_doc" = xyes; then echo yes; else echo no; fi`
 - graphs and diagrams           : `if test "x$enable_dot" = xyes; then echo yes; else echo no; fi`
-
+- Supported SIMD flags          : $SIMD_FLAGS
 ----------------------------------------------------------
-- enabled simd support          : ${ac_SIMD}
+- enabled simd support          : ${ac_SIMD}   (supported: $supported )
 - communications type           : ${ac_COMMS}
 
 
diff --git a/lib/Grid_config.h b/lib/Grid_config.h
index 2397894f..914bd06c 100644
--- a/lib/Grid_config.h
+++ b/lib/Grid_config.h
@@ -16,6 +16,15 @@
 /* GRID_COMMS_NONE */
 #define GRID_COMMS_NONE 1
 
+/* Support Altivec instructions */
+/* #undef HAVE_ALTIVEC */
+
+/* Support AVX (Advanced Vector Extensions) instructions */
+/* #undef HAVE_AVX */
+
+/* Support AVX2 (Advanced Vector Extensions 2) instructions */
+/* #undef HAVE_AVX2 */
+
 /* define if the compiler supports basic C++11 syntax */
 /* #undef HAVE_CXX11 */
 
@@ -30,6 +39,9 @@
 /* Define to 1 if you have the <endian.h> header file. */
 #define HAVE_ENDIAN_H 1
 
+/* Support FMA3 (Fused Multiply-Add) instructions */
+/* #undef HAVE_FMA */
+
 /* Define to 1 if you have the `gettimeofday' function. */
 #define HAVE_GETTIMEOFDAY 1
 
@@ -54,9 +66,30 @@
 /* Define to 1 if you have the <memory.h> header file. */
 #define HAVE_MEMORY_H 1
 
+/* Support mmx instructions */
+#define HAVE_MMX /**/
+
 /* Define to 1 if you have the <mm_malloc.h> header file. */
 #define HAVE_MM_MALLOC_H 1
 
+/* Support SSE (Streaming SIMD Extensions) instructions */
+#define HAVE_SSE /**/
+
+/* Support SSE2 (Streaming SIMD Extensions 2) instructions */
+#define HAVE_SSE2 /**/
+
+/* Support SSE3 (Streaming SIMD Extensions 3) instructions */
+#define HAVE_SSE3 /**/
+
+/* Support SSSE4.1 (Streaming SIMD Extensions 4.1) instructions */
+#define HAVE_SSE4_1 /**/
+
+/* Support SSSE4.2 (Streaming SIMD Extensions 4.2) instructions */
+#define HAVE_SSE4_2 /**/
+
+/* Support SSSE3 (Supplemental Streaming SIMD Extensions 3) instructions */
+#define HAVE_SSSE3 /**/
+
 /* Define to 1 if you have the <stdint.h> header file. */
 #define HAVE_STDINT_H 1
 
diff --git a/lib/Grid_config.h.in b/lib/Grid_config.h.in
index 6f05d6cb..8c958c82 100644
--- a/lib/Grid_config.h.in
+++ b/lib/Grid_config.h.in
@@ -15,6 +15,15 @@
 /* GRID_COMMS_NONE */
 #undef GRID_COMMS_NONE
 
+/* Support Altivec instructions */
+#undef HAVE_ALTIVEC
+
+/* Support AVX (Advanced Vector Extensions) instructions */
+#undef HAVE_AVX
+
+/* Support AVX2 (Advanced Vector Extensions 2) instructions */
+#undef HAVE_AVX2
+
 /* define if the compiler supports basic C++11 syntax */
 #undef HAVE_CXX11
 
@@ -29,6 +38,9 @@
 /* Define to 1 if you have the <endian.h> header file. */
 #undef HAVE_ENDIAN_H
 
+/* Support FMA3 (Fused Multiply-Add) instructions */
+#undef HAVE_FMA
+
 /* Define to 1 if you have the `gettimeofday' function. */
 #undef HAVE_GETTIMEOFDAY
 
@@ -53,9 +65,30 @@
 /* Define to 1 if you have the <memory.h> header file. */
 #undef HAVE_MEMORY_H
 
+/* Support mmx instructions */
+#undef HAVE_MMX
+
 /* Define to 1 if you have the <mm_malloc.h> header file. */
 #undef HAVE_MM_MALLOC_H
 
+/* Support SSE (Streaming SIMD Extensions) instructions */
+#undef HAVE_SSE
+
+/* Support SSE2 (Streaming SIMD Extensions 2) instructions */
+#undef HAVE_SSE2
+
+/* Support SSE3 (Streaming SIMD Extensions 3) instructions */
+#undef HAVE_SSE3
+
+/* Support SSSE4.1 (Streaming SIMD Extensions 4.1) instructions */
+#undef HAVE_SSE4_1
+
+/* Support SSSE4.2 (Streaming SIMD Extensions 4.2) instructions */
+#undef HAVE_SSE4_2
+
+/* Support SSSE3 (Supplemental Streaming SIMD Extensions 3) instructions */
+#undef HAVE_SSSE3
+
 /* Define to 1 if you have the <stdint.h> header file. */
 #undef HAVE_STDINT_H
 
diff --git a/m4/ax_check_compile_flag.m4 b/m4/ax_check_compile_flag.m4
new file mode 100644
index 00000000..c3a8d695
--- /dev/null
+++ b/m4/ax_check_compile_flag.m4
@@ -0,0 +1,72 @@
+# ===========================================================================
+#   http://www.gnu.org/software/autoconf-archive/ax_check_compile_flag.html
+# ===========================================================================
+#
+# SYNOPSIS
+#
+#   AX_CHECK_COMPILE_FLAG(FLAG, [ACTION-SUCCESS], [ACTION-FAILURE], [EXTRA-FLAGS])
+#
+# DESCRIPTION
+#
+#   Check whether the given FLAG works with the current language's compiler
+#   or gives an error.  (Warnings, however, are ignored)
+#
+#   ACTION-SUCCESS/ACTION-FAILURE are shell commands to execute on
+#   success/failure.
+#
+#   If EXTRA-FLAGS is defined, it is added to the current language's default
+#   flags (e.g. CFLAGS) when the check is done.  The check is thus made with
+#   the flags: "CFLAGS EXTRA-FLAGS FLAG".  This can for example be used to
+#   force the compiler to issue an error when a bad flag is given.
+#
+#   NOTE: Implementation based on AX_CFLAGS_GCC_OPTION. Please keep this
+#   macro in sync with AX_CHECK_{PREPROC,LINK}_FLAG.
+#
+# LICENSE
+#
+#   Copyright (c) 2008 Guido U. Draheim <guidod@gmx.de>
+#   Copyright (c) 2011 Maarten Bosmans <mkbosmans@gmail.com>
+#
+#   This program is free software: you can redistribute it and/or modify it
+#   under the terms of the GNU General Public License as published by the
+#   Free Software Foundation, either version 3 of the License, or (at your
+#   option) any later version.
+#
+#   This program is distributed in the hope that it will be useful, but
+#   WITHOUT ANY WARRANTY; without even the implied warranty of
+#   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General
+#   Public License for more details.
+#
+#   You should have received a copy of the GNU General Public License along
+#   with this program. If not, see <http://www.gnu.org/licenses/>.
+#
+#   As a special exception, the respective Autoconf Macro's copyright owner
+#   gives unlimited permission to copy, distribute and modify the configure
+#   scripts that are the output of Autoconf when processing the Macro. You
+#   need not follow the terms of the GNU General Public License when using
+#   or distributing such scripts, even though portions of the text of the
+#   Macro appear in them. The GNU General Public License (GPL) does govern
+#   all other use of the material that constitutes the Autoconf Macro.
+#
+#   This special exception to the GPL applies to versions of the Autoconf
+#   Macro released by the Autoconf Archive. When you make and distribute a
+#   modified version of the Autoconf Macro, you may extend this special
+#   exception to the GPL to apply to your modified version as well.
+
+#serial 2
+
+AC_DEFUN([AX_CHECK_COMPILE_FLAG],
+[AC_PREREQ(2.59)dnl for _AC_LANG_PREFIX
+AS_VAR_PUSHDEF([CACHEVAR],[ax_cv_check_[]_AC_LANG_ABBREV[]flags_$4_$1])dnl
+AC_CACHE_CHECK([whether _AC_LANG compiler accepts $1], CACHEVAR, [
+  ax_check_save_flags=$[]_AC_LANG_PREFIX[]FLAGS
+  _AC_LANG_PREFIX[]FLAGS="$[]_AC_LANG_PREFIX[]FLAGS $4 $1"
+  AC_COMPILE_IFELSE([AC_LANG_PROGRAM()],
+    [AS_VAR_SET(CACHEVAR,[yes])],
+    [AS_VAR_SET(CACHEVAR,[no])])
+  _AC_LANG_PREFIX[]FLAGS=$ax_check_save_flags])
+AS_IF([test x"AS_VAR_GET(CACHEVAR)" = xyes],
+  [m4_default([$2], :)],
+  [m4_default([$3], :)])
+AS_VAR_POPDEF([CACHEVAR])dnl
+])dnl AX_CHECK_COMPILE_FLAGS
diff --git a/m4/ax_ext.m4 b/m4/ax_ext.m4
new file mode 100644
index 00000000..97a0f3e6
--- /dev/null
+++ b/m4/ax_ext.m4
@@ -0,0 +1,288 @@
+# ===========================================================================
+#          http://www.gnu.org/software/autoconf-archive/ax_ext.html
+# ===========================================================================
+#
+# SYNOPSIS
+#
+#   AX_EXT
+#
+# DESCRIPTION
+#
+#   Find supported SIMD extensions by requesting cpuid. When an SIMD
+#   extension is found, the -m"simdextensionname" is added to SIMD_FLAGS if
+#   compiler supports it. For example, if "sse2" is available, then "-msse2"
+#   is added to SIMD_FLAGS.
+#
+#   This macro calls:
+#
+#     AC_SUBST(SIMD_FLAGS)
+#
+#   And defines:
+#
+#     HAVE_MMX / HAVE_SSE / HAVE_SSE2 / HAVE_SSE3 / HAVE_SSSE3 / HAVE_SSE4.1 / HAVE_SSE4.2 / HAVE_AVX
+#
+# LICENSE
+#
+#   Copyright (c) 2007 Christophe Tournayre <turn3r@users.sourceforge.net>
+#   Copyright (c) 2013 Michael Petch <mpetch@capp-sysware.com>
+#
+#   Copying and distribution of this file, with or without modification, are
+#   permitted in any medium without royalty provided the copyright notice
+#   and this notice are preserved. This file is offered as-is, without any
+#   warranty.
+
+#serial 13
+
+AC_DEFUN([AX_EXT],
+[
+  AC_REQUIRE([AC_CANONICAL_HOST])
+
+  case $host_cpu in
+    powerpc*)
+      AC_CACHE_CHECK([whether altivec is supported], [ax_cv_have_altivec_ext],
+          [
+            if test `/usr/sbin/sysctl -a 2>/dev/null| grep -c hw.optional.altivec` != 0; then
+                if test `/usr/sbin/sysctl -n hw.optional.altivec` = 1; then
+                  ax_cv_have_altivec_ext=yes
+                fi
+            fi
+          ])
+
+          if test "$ax_cv_have_altivec_ext" = yes; then
+            AC_DEFINE(HAVE_ALTIVEC,,[Support Altivec instructions])
+            AX_CHECK_COMPILE_FLAG(-faltivec, [SIMD_FLAGS="$SIMD_FLAGS -faltivec"], [])
+          fi
+    ;;
+
+
+    i[[3456]]86*|x86_64*|amd64*)
+
+      AC_REQUIRE([AX_GCC_X86_CPUID])
+      AC_REQUIRE([AX_GCC_X86_AVX_XGETBV])
+
+      AX_GCC_X86_CPUID(0x00000001)
+      ecx=0
+      edx=0
+      ebx=0
+      if test "$ax_cv_gcc_x86_cpuid_0x00000001" != "unknown";
+      then
+        ecx=`echo $ax_cv_gcc_x86_cpuid_0x00000001 | cut -d ":" -f 3`
+        edx=`echo $ax_cv_gcc_x86_cpuid_0x00000001 | cut -d ":" -f 4`
+      fi
+
+      AX_GCC_X86_CPUID(0x00000007)
+      if test "$ax_cv_gcc_x86_cpuid_0x00000007" != "unknown";
+      then
+        ebx=`echo $ax_cv_gcc_x86_cpuid_0x00000007 | cut -d ":" -f 2`
+      fi
+
+      AC_CACHE_CHECK([whether mmx is supported], [ax_cv_have_mmx_ext],
+      [
+        ax_cv_have_mmx_ext=no
+        if test "$((0x$edx>>23&0x01))" = 1; then
+          ax_cv_have_mmx_ext=yes
+        fi
+      ])
+
+      AC_CACHE_CHECK([whether sse is supported], [ax_cv_have_sse_ext],
+      [
+        ax_cv_have_sse_ext=no
+        if test "$((0x$edx>>25&0x01))" = 1; then
+          ax_cv_have_sse_ext=yes
+        fi
+      ])
+
+      AC_CACHE_CHECK([whether sse2 is supported], [ax_cv_have_sse2_ext],
+      [
+        ax_cv_have_sse2_ext=no
+        if test "$((0x$edx>>26&0x01))" = 1; then
+          ax_cv_have_sse2_ext=yes
+        fi
+      ])
+
+      AC_CACHE_CHECK([whether sse3 is supported], [ax_cv_have_sse3_ext],
+      [
+        ax_cv_have_sse3_ext=no
+        if test "$((0x$ecx&0x01))" = 1; then
+          ax_cv_have_sse3_ext=yes
+        fi
+      ])
+
+      AC_CACHE_CHECK([whether ssse3 is supported], [ax_cv_have_ssse3_ext],
+      [
+        ax_cv_have_ssse3_ext=no
+        if test "$((0x$ecx>>9&0x01))" = 1; then
+          ax_cv_have_ssse3_ext=yes
+        fi
+      ])
+
+      AC_CACHE_CHECK([whether sse4.1 is supported], [ax_cv_have_sse41_ext],
+      [
+        ax_cv_have_sse41_ext=no
+        if test "$((0x$ecx>>19&0x01))" = 1; then
+          ax_cv_have_sse41_ext=yes
+        fi
+      ])
+
+      AC_CACHE_CHECK([whether sse4.2 is supported], [ax_cv_have_sse42_ext],
+      [
+        ax_cv_have_sse42_ext=no
+        if test "$((0x$ecx>>20&0x01))" = 1; then
+          ax_cv_have_sse42_ext=yes
+        fi
+      ])
+
+      AC_CACHE_CHECK([whether avx is supported by processor], [ax_cv_have_avx_cpu_ext],
+      [
+        ax_cv_have_avx_cpu_ext=no
+        if test "$((0x$ecx>>28&0x01))" = 1; then
+          ax_cv_have_avx_cpu_ext=yes
+        fi
+      ])
+
+      AC_CACHE_CHECK([whether avx2 is supported by processor], [ax_cv_have_avx2_cpu_ext],
+      [
+        ax_cv_have_avx2_cpu_ext=no
+        if test "$((0x$ebx>>5&0x01))" = 1; then
+          ax_cv_have_avx2_cpu_ext=yes
+        fi
+      ])
+
+
+      AC_CACHE_CHECK([whether fma is supported by processor], [ax_cv_have_fma_cpu_ext],
+      [
+        ax_cv_have_fma_cpu_ext=no
+        if test "$((0x$ecx>>12&0x01))" = 1; then
+          ax_cv_have_fma_cpu_ext=yes
+        fi
+      ])
+
+
+      if test x"$ax_cv_have_avx_cpu_ext" = x"yes"; then
+        AX_GCC_X86_AVX_XGETBV(0x00000000)
+
+        xgetbv_eax="0"
+        if test x"$ax_cv_gcc_x86_avx_xgetbv_0x00000000" != x"unknown"; then
+          xgetbv_eax=`echo $ax_cv_gcc_x86_avx_xgetbv_0x00000000 | cut -d ":" -f 1`
+        fi
+
+        AC_CACHE_CHECK([whether avx is supported by operating system], [ax_cv_have_avx_ext],
+        [
+          ax_cv_have_avx_ext=no
+
+          if test "$((0x$ecx>>27&0x01))" = 1; then
+            if test "$((0x$xgetbv_eax&0x6))" = 6; then
+              ax_cv_have_avx_ext=yes
+            fi
+          fi
+        ])
+        if test x"$ax_cv_have_avx_ext" = x"no"; then
+          AC_MSG_WARN([Your processor supports AVX, but your operating system doesn't])
+        fi
+      fi
+
+      if test "$ax_cv_have_mmx_ext" = yes; then
+        AX_CHECK_COMPILE_FLAG(-mmmx, ax_cv_support_mmx_ext=yes, [])
+        if test x"$ax_cv_support_mmx_ext" = x"yes"; then
+          SIMD_FLAGS="$SIMD_FLAGS -mmmx"
+          AC_DEFINE(HAVE_MMX,,[Support mmx instructions])
+        else
+          AC_MSG_WARN([Your processor supports mmx instructions but not your compiler, can you try another compiler?])
+        fi
+      fi
+
+      if test "$ax_cv_have_sse_ext" = yes; then
+        AX_CHECK_COMPILE_FLAG(-msse, ax_cv_support_sse_ext=yes, [])
+        if test x"$ax_cv_support_sse_ext" = x"yes"; then
+          SIMD_FLAGS="$SIMD_FLAGS -msse"
+          AC_DEFINE(HAVE_SSE,,[Support SSE (Streaming SIMD Extensions) instructions])
+        else
+          AC_MSG_WARN([Your processor supports sse instructions but not your compiler, can you try another compiler?])
+        fi
+      fi
+
+      if test "$ax_cv_have_sse2_ext" = yes; then
+        AX_CHECK_COMPILE_FLAG(-msse2, ax_cv_support_sse2_ext=yes, [])
+        if test x"$ax_cv_support_sse2_ext" = x"yes"; then
+          SIMD_FLAGS="$SIMD_FLAGS -msse2"
+          AC_DEFINE(HAVE_SSE2,,[Support SSE2 (Streaming SIMD Extensions 2) instructions])
+        else
+          AC_MSG_WARN([Your processor supports sse2 instructions but not your compiler, can you try another compiler?])
+        fi
+      fi
+
+      if test "$ax_cv_have_sse3_ext" = yes; then
+        AX_CHECK_COMPILE_FLAG(-msse3, ax_cv_support_sse3_ext=yes, [])
+        if test x"$ax_cv_support_sse3_ext" = x"yes"; then
+          SIMD_FLAGS="$SIMD_FLAGS -msse3"
+          AC_DEFINE(HAVE_SSE3,,[Support SSE3 (Streaming SIMD Extensions 3) instructions])
+        else
+          AC_MSG_WARN([Your processor supports sse3 instructions but not your compiler, can you try another compiler?])
+        fi
+      fi
+
+      if test "$ax_cv_have_ssse3_ext" = yes; then
+        AX_CHECK_COMPILE_FLAG(-mssse3, ax_cv_support_ssse3_ext=yes, [])
+        if test x"$ax_cv_support_ssse3_ext" = x"yes"; then
+          SIMD_FLAGS="$SIMD_FLAGS -mssse3"
+          AC_DEFINE(HAVE_SSSE3,,[Support SSSE3 (Supplemental Streaming SIMD Extensions 3) instructions])
+        else
+          AC_MSG_WARN([Your processor supports ssse3 instructions but not your compiler, can you try another compiler?])
+        fi
+      fi
+
+      if test "$ax_cv_have_sse41_ext" = yes; then
+        AX_CHECK_COMPILE_FLAG(-msse4.1, ax_cv_support_sse41_ext=yes, [])
+        if test x"$ax_cv_support_sse41_ext" = x"yes"; then
+          SIMD_FLAGS="$SIMD_FLAGS -msse4.1"
+          AC_DEFINE(HAVE_SSE4_1,,[Support SSSE4.1 (Streaming SIMD Extensions 4.1) instructions])
+        else
+          AC_MSG_WARN([Your processor supports sse4.1 instructions but not your compiler, can you try another compiler?])
+        fi
+      fi
+
+      if test "$ax_cv_have_sse42_ext" = yes; then
+        AX_CHECK_COMPILE_FLAG(-msse4.2, ax_cv_support_sse42_ext=yes, [])
+        if test x"$ax_cv_support_sse42_ext" = x"yes"; then
+          SIMD_FLAGS="$SIMD_FLAGS -msse4.2"
+          AC_DEFINE(HAVE_SSE4_2,,[Support SSSE4.2 (Streaming SIMD Extensions 4.2) instructions])
+        else
+          AC_MSG_WARN([Your processor supports sse4.2 instructions but not your compiler, can you try another compiler?])
+        fi
+      fi
+
+      if test "$ax_cv_have_avx_ext" = yes; then
+        AX_CHECK_COMPILE_FLAG(-mavx, ax_cv_support_avx_ext=yes, [])
+        if test x"$ax_cv_support_avx_ext" = x"yes"; then
+          SIMD_FLAGS="$SIMD_FLAGS -mavx"
+          AC_DEFINE(HAVE_AVX,,[Support AVX (Advanced Vector Extensions) instructions])
+        else
+          AC_MSG_WARN([Your processor supports avx instructions but not your compiler, can you try another compiler?])
+        fi
+      fi
+
+       if test "$ax_cv_have_avx2_ext" = yes; then
+         AX_CHECK_COMPILE_FLAG(-mavx2, ax_cv_support_avx2_ext=yes, [])
+         if test x"$ax_cv_support_avx2_ext" = x"yes"; then
+           SIMD_FLAGS="$SIMD_FLAGS -mavx2"
+           AC_DEFINE(HAVE_AVX2,,[Support AVX2 (Advanced Vector Extensions 2) instructions])
+         else
+           AC_MSG_WARN([Your processor supports avx2 instructions but not your compiler, can you try another compiler?])
+         fi
+        fi
+
+       if test "$ax_cv_have_fma_ext" = yes; then
+         AX_CHECK_COMPILE_FLAG(-mfma, ax_cv_support_fma_ext=yes, [])
+         if test x"$ax_cv_support_fma_ext" = x"yes"; then
+           SIMD_FLAGS="$SIMD_FLAGS -mfma"
+           AC_DEFINE(HAVE_FMA,,[Support FMA3 (Fused Multiply-Add) instructions])
+         else
+           AC_MSG_WARN([Your processor supports fma instructions but not your compiler, can you try another compiler?])
+         fi
+      fi
+
+  ;;
+  esac
+
+  AC_SUBST(SIMD_FLAGS)
+])
diff --git a/m4/ax_gcc_x86_avx_xgetbv.m4 b/m4/ax_gcc_x86_avx_xgetbv.m4
new file mode 100644
index 00000000..2da9a6ab
--- /dev/null
+++ b/m4/ax_gcc_x86_avx_xgetbv.m4
@@ -0,0 +1,79 @@
+# ===========================================================================
+#   http://www.gnu.org/software/autoconf-archive/ax_gcc_x86_avx_xgetbv.html
+# ===========================================================================
+#
+# SYNOPSIS
+#
+#   AX_GCC_X86_AVX_XGETBV
+#
+# DESCRIPTION
+#
+#   On later x86 processors with AVX SIMD support, with gcc or a compiler
+#   that has a compatible syntax for inline assembly instructions, run a
+#   small program that executes the xgetbv instruction with input OP. This
+#   can be used to detect if the OS supports AVX instruction usage.
+#
+#   On output, the values of the eax and edx registers are stored as
+#   hexadecimal strings as "eax:edx" in the cache variable
+#   ax_cv_gcc_x86_avx_xgetbv.
+#
+#   If the xgetbv instruction fails (because you are running a
+#   cross-compiler, or because you are not using gcc, or because you are on
+#   a processor that doesn't have this instruction),
+#   ax_cv_gcc_x86_avx_xgetbv_OP is set to the string "unknown".
+#
+#   This macro mainly exists to be used in AX_EXT.
+#
+# LICENSE
+#
+#   Copyright (c) 2013 Michael Petch <mpetch@capp-sysware.com>
+#
+#   This program is free software: you can redistribute it and/or modify it
+#   under the terms of the GNU General Public License as published by the
+#   Free Software Foundation, either version 3 of the License, or (at your
+#   option) any later version.
+#
+#   This program is distributed in the hope that it will be useful, but
+#   WITHOUT ANY WARRANTY; without even the implied warranty of
+#   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General
+#   Public License for more details.
+#
+#   You should have received a copy of the GNU General Public License along
+#   with this program. If not, see <http://www.gnu.org/licenses/>.
+#
+#   As a special exception, the respective Autoconf Macro's copyright owner
+#   gives unlimited permission to copy, distribute and modify the configure
+#   scripts that are the output of Autoconf when processing the Macro. You
+#   need not follow the terms of the GNU General Public License when using
+#   or distributing such scripts, even though portions of the text of the
+#   Macro appear in them. The GNU General Public License (GPL) does govern
+#   all other use of the material that constitutes the Autoconf Macro.
+#
+#   This special exception to the GPL applies to versions of the Autoconf
+#   Macro released by the Autoconf Archive. When you make and distribute a
+#   modified version of the Autoconf Macro, you may extend this special
+#   exception to the GPL to apply to your modified version as well.
+
+#serial 1
+
+AC_DEFUN([AX_GCC_X86_AVX_XGETBV],
+[AC_REQUIRE([AC_PROG_CC])
+AC_LANG_PUSH([C])
+AC_CACHE_CHECK(for x86-AVX xgetbv $1 output, ax_cv_gcc_x86_avx_xgetbv_$1,
+ [AC_RUN_IFELSE([AC_LANG_PROGRAM([#include <stdio.h>], [
+     int op = $1, eax, edx;
+     FILE *f;
+      /* Opcodes for xgetbv */
+      __asm__(".byte 0x0f, 0x01, 0xd0"
+        : "=a" (eax), "=d" (edx), 
+        : "c" (op));
+     f = fopen("conftest_xgetbv", "w"); if (!f) return 1;
+     fprintf(f, "%x:%x\n", eax, edx);
+     fclose(f);
+     return 0;
+])],
+     [ax_cv_gcc_x86_avx_xgetbv_$1=`cat conftest_xgetbv`; rm -f conftest_xgetbv],
+     [ax_cv_gcc_x86_avx_xgetbv_$1=unknown; rm -f conftest_xgetbv],
+     [ax_cv_gcc_x86_avx_xgetbv_$1=unknown])])
+AC_LANG_POP([C])
+])
diff --git a/m4/ax_gcc_x86_cpuid.m4 b/m4/ax_gcc_x86_cpuid.m4
new file mode 100644
index 00000000..47dc8fda
--- /dev/null
+++ b/m4/ax_gcc_x86_cpuid.m4
@@ -0,0 +1,45 @@
+dnl @synopsis AX_GCC_X86_CPUID(OP)
+dnl
+dnl @summary run x86 cpuid instruction OP using gcc inline assembler
+dnl
+dnl On Pentium and later x86 processors, with gcc or a compiler that
+dnl has a compatible syntax for inline assembly instructions, run a
+dnl small program that executes the cpuid instruction with input OP.
+dnl This can be used to detect the CPU type.
+dnl
+dnl On output, the values of the eax, ebx, ecx, and edx registers are
+dnl stored as hexadecimal strings as "eax:ebx:ecx:edx" in the cache
+dnl variable ax_cv_gcc_x86_cpuid_OP.
+dnl
+dnl If the cpuid instruction fails (because you are running a
+dnl cross-compiler, or because you are not using gcc, or because you
+dnl are on a processor that doesn't have this instruction),
+dnl ax_cv_gcc_x86_cpuid_OP is set to the string "unknown".
+dnl
+dnl This macro mainly exists to be used in AX_GCC_ARCHFLAG.
+dnl
+dnl @category Misc
+dnl @author Steven G. Johnson <stevenj@alum.mit.edu> and Matteo Frigo.
+dnl @version 2005-05-30
+dnl @license GPLWithACException
+
+AC_DEFUN([AX_GCC_X86_CPUID],
+[AC_REQUIRE([AC_PROG_CC])
+AC_LANG_PUSH([C])
+AC_CACHE_CHECK(for x86 cpuid $1 output, ax_cv_gcc_x86_cpuid_$1,
+ [AC_RUN_IFELSE([AC_LANG_PROGRAM([#include <stdio.h>], [
+     int op = $1, eax, ebx, ecx, edx;
+     FILE *f;
+      __asm__("cpuid"
+        : "=a" (eax), "=b" (ebx), "=c" (ecx), "=d" (edx)
+        : "a" (op));
+     f = fopen("conftest_cpuid", "w"); if (!f) return 1;
+     fprintf(f, "%x:%x:%x:%x\n", eax, ebx, ecx, edx);
+     fclose(f);
+     return 0;
+])],
+     [ax_cv_gcc_x86_cpuid_$1=`cat conftest_cpuid`; rm -f conftest_cpuid],
+     [ax_cv_gcc_x86_cpuid_$1=unknown; rm -f conftest_cpuid],
+     [ax_cv_gcc_x86_cpuid_$1=unknown])])
+AC_LANG_POP([C])
+])
diff --git a/reconfigure_script b/scripts/reconfigure_script
similarity index 100%
rename from reconfigure_script
rename to scripts/reconfigure_script

From e8be96bfe73fafca9b3afd1f0bbb3ce703bec600 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 27 May 2015 16:07:17 +0100
Subject: [PATCH 245/429] Improving the reduction to go through our on permute.
 Must also do this for avx512

---
 INSTALL              |   2 +-
 configure            |  13 +++---
 lib/simd/Grid_avx.h  |  75 ++++++++++++-----------------
 lib/simd/Grid_sse4.h | 109 ++++++++++++++++++++-----------------------
 4 files changed, 89 insertions(+), 110 deletions(-)

diff --git a/INSTALL b/INSTALL
index f812f5a2..80a61507 120000
--- a/INSTALL
+++ b/INSTALL
@@ -1 +1 @@
-/usr/share/automake-1.14/INSTALL
\ No newline at end of file
+/opt/local/share/automake-1.15/INSTALL
\ No newline at end of file
diff --git a/configure b/configure
index ccfea6b7..615ba987 100755
--- a/configure
+++ b/configure
@@ -2483,7 +2483,7 @@ test -n "$target_alias" &&
     NONENONEs,x,x, &&
   program_prefix=${target_alias}-
 
-am__api_version='1.14'
+am__api_version='1.15'
 
 # Find a good install program.  We prefer a C program (faster),
 # so one script is as good as another.  But avoid the broken or
@@ -2655,8 +2655,8 @@ test "$program_suffix" != NONE &&
 ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
 program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
 
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 
 if test x"${MISSING+set}" != xset; then
   case $am_aux_dir in
@@ -2675,7 +2675,7 @@ else
 $as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
 fi
 
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -3003,8 +3003,8 @@ MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 mkdir_p='$(MKDIR_P)'
 
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
 # Always define AMTAR for backward compatibility.  Yes, it's still used
 # in the wild :-(  We should find a proper way to deprecate it ...
 AMTAR='$${TAR-tar}'
@@ -3063,6 +3063,7 @@ END
 fi
 
 
+
 ac_config_headers="$ac_config_headers lib/Grid_config.h"
 
 # Check whether --enable-silent-rules was given.
diff --git a/lib/simd/Grid_avx.h b/lib/simd/Grid_avx.h
index ec2df44b..3c8ebc61 100644
--- a/lib/simd/Grid_avx.h
+++ b/lib/simd/Grid_avx.h
@@ -14,7 +14,23 @@
 #endif
 
 namespace Optimization {
-  
+
+  template<class vtype>
+  union uconv {
+    __m256 f;
+    vtype v;
+  };
+
+  union u256f {
+    __m256 v;
+    float f[8];
+  };
+
+  union u256d {
+    __m256d v;
+    double f[4];
+  };
+
   struct Vsplat{
     //Complex float
     inline __m256 operator()(float a, float b){
@@ -54,7 +70,6 @@ namespace Optimization {
 
   };
 
-
   struct Vstream{
     //Float
     inline void operator()(float * a, __m256 b){
@@ -68,8 +83,6 @@ namespace Optimization {
 
   };
 
-
-
   struct Vset{
     // Complex float 
     inline __m256 operator()(Grid::ComplexF *a){
@@ -92,7 +105,6 @@ namespace Optimization {
       return _mm256_set_epi32(a[7],a[6],a[5],a[4],a[3],a[2],a[1],a[0]);
     }
 
-
   };
 
   template <typename Out_type, typename In_type>
@@ -106,9 +118,6 @@ namespace Optimization {
     }
   };
 
-
- 
-
   /////////////////////////////////////////////////////
   // Arithmetic operations
   /////////////////////////////////////////////////////
@@ -170,7 +179,6 @@ namespace Optimization {
     }
   };
 
-
   struct MultComplex{
     // Complex float
     inline __m256 operator()(__m256 a, __m256 b){
@@ -207,7 +215,6 @@ namespace Optimization {
 	IF IMM0[3] = 0
 	THEN DEST[255:192]=SRC2[191:128] ELSE DEST[255:192]=SRC2[255:192] FI; // Ox5 r<->i   ; 0xC unchanged
       */
-      
       __m256d ymm0,ymm1,ymm2;
       ymm0 = _mm256_shuffle_pd(a,a,0x0); // ymm0 <- ar ar, ar,ar b'00,00
       ymm0 = _mm256_mul_pd(ymm0,b);      // ymm0 <- ar bi, ar br
@@ -247,7 +254,6 @@ namespace Optimization {
     }
   };
 
-
   struct Conj{
     // Complex single
     inline __m256 operator()(__m256 in){
@@ -292,18 +298,13 @@ namespace Optimization {
     }
   };
 
-
-  
-
-
   //////////////////////////////////////////////
   // Some Template specialization
+  //////////////////////////////////////////////
+
   template < typename vtype > 
-    void permute(vtype &a, vtype &b, int perm) {
-    union { 
-      __m256 f;
-      vtype v;
-    } conv;
+    void permute(vtype &a,vtype b, int perm) {
+    uconv<vtype> conv;
     conv.v = b;
     switch (perm){
       // 8x32 bits=>3 permutes
@@ -313,24 +314,20 @@ namespace Optimization {
     default: assert(0); break;
     }
     a = conv.v;
-    
   }
-  
+
   //Complex float Reduce
   template<>
     inline Grid::ComplexF Reduce<Grid::ComplexF, __m256>::operator()(__m256 in){
     __m256 v1,v2;
-    union { 
-      __m256 v;
-      float f[8];
-    } conv;
     Optimization::permute(v1,in,0); // sse 128; paired complex single
     v1 = _mm256_add_ps(v1,in);
     Optimization::permute(v2,v1,1); // avx 256; quad complex single
     v1 = _mm256_add_ps(v1,v2);
-    conv.v = v1;
+    u256f conv; conv.v = v1;
     return Grid::ComplexF(conv.f[0],conv.f[1]);
   }
+
   //Real float Reduce
   template<>
   inline Grid::RealF Reduce<Grid::RealF, __m256>::operator()(__m256 in){
@@ -341,7 +338,8 @@ namespace Optimization {
     v1 = _mm256_add_ps(v1,v2);
     Optimization::permute(v2,v1,2); 
     v1 = _mm256_add_ps(v1,v2);
-    return v1[0];
+    u256f conv; conv.v=v1;
+    return conv.f[0];
   }
   
   
@@ -351,7 +349,8 @@ namespace Optimization {
     __m256d v1;
     Optimization::permute(v1,in,0); // sse 128; paired complex single
     v1 = _mm256_add_pd(v1,in);
-    return Grid::ComplexD(v1[0],v1[1]);
+    u256d conv; conv.v = v1;
+    return Grid::ComplexD(conv.f[0],conv.f[1]);
   }
   
   //Real double Reduce
@@ -362,7 +361,8 @@ namespace Optimization {
     v1 = _mm256_add_pd(v1,in);
     Optimization::permute(v2,v1,1); 
     v1 = _mm256_add_pd(v1,v2);
-    return v1[0];
+    u256d conv; conv.v = v1;
+    return conv.f[0];
   }
 
   //Integer Reduce
@@ -390,22 +390,9 @@ namespace Grid {
       _mm_prefetch(ptr+i+512,_MM_HINT_T0);
     }
   }
-  
   template < typename VectorSIMD > 
     inline void Gpermute(VectorSIMD &y,const VectorSIMD &b, int perm ) {
-    union { 
-      __m256 f;
-      decltype(VectorSIMD::v) v;
-    } conv;
-    conv.v = b.v;
-    switch(perm){
-    case 3: break; //empty for AVX1/2
-    case 2: conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); break;
-    case 1: conv.f = _mm256_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2));  break; 
-    case 0: conv.f = _mm256_permute2f128_ps(conv.f,conv.f,0x01); break;
-    default: assert(0); break;
-    }
-    y.v=conv.v;
+    Optimization::permute(y.v,b.v,perm);
   };
 
   // Function name aliases
diff --git a/lib/simd/Grid_sse4.h b/lib/simd/Grid_sse4.h
index ce2737d5..0da888f7 100644
--- a/lib/simd/Grid_sse4.h
+++ b/lib/simd/Grid_sse4.h
@@ -10,6 +10,21 @@
 #include <pmmintrin.h>
 
 namespace Optimization {
+
+  template<class vtype>
+  union uconv {
+    __m128 f;
+    vtype v;
+  };
+
+  union u128f {
+    __m128 v;
+    float f[4];
+  };
+  union u128d {
+    __m128d v;
+    double f[2];
+  };
   
   struct Vsplat{
     //Complex float
@@ -50,7 +65,6 @@ namespace Optimization {
 
   };
 
-
   struct Vstream{
     //Float
     inline void operator()(float * a, __m128 b){
@@ -64,8 +78,6 @@ namespace Optimization {
 
   };
 
-
-
   struct Vset{
     // Complex float 
     inline __m128 operator()(Grid::ComplexF *a){
@@ -102,9 +114,6 @@ namespace Optimization {
     }
   };
 
-
- 
-
   /////////////////////////////////////////////////////
   // Arithmetic operations
   /////////////////////////////////////////////////////
@@ -138,7 +147,6 @@ namespace Optimization {
     }
   };
 
-
   struct MultComplex{
     // Complex float
     inline __m128 operator()(__m128 a, __m128 b){
@@ -177,7 +185,6 @@ namespace Optimization {
     }
   };
 
-
   struct Conj{
     // Complex single
     inline __m128 operator()(__m128 in){
@@ -216,57 +223,61 @@ namespace Optimization {
       __m128d tmp = _mm_shuffle_pd(in,in,0x1);
       return _mm_addsub_pd(_mm_setzero_pd(),tmp); // r,-i
     }
-
-
   };
 
-
-
-
-
   //////////////////////////////////////////////
   // Some Template specialization
-  
+  template < typename vtype > 
+    void permute(vtype &a, vtype b, int perm) {
+    uconv<vtype> conv; 
+    conv.v = b;
+    switch(perm){
+    case 3: break; //empty for SSE4
+    case 2: break; //empty for SSE4
+    case 1: conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); break;
+    case 0: conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2)); break;
+    default: assert(0); break;
+    }
+    a=conv.v;
+  }; 
+
   //Complex float Reduce
   template<>
   inline Grid::ComplexF Reduce<Grid::ComplexF, __m128>::operator()(__m128 in){
-    union {
-      __m128 v1;  
-      float f[4]; 
-    } u128;
-    u128.v1 = _mm_add_ps(in, _mm_shuffle_ps(in,in, 0b01001110)); // FIXME Prefer to use _MM_SHUFFLE macros
-    return Grid::ComplexF(u128.f[0], u128.f[1]);   
+    __m128 v1; // two complex
+    Optimization::permute(v1,in,0); 
+    v1= _mm_add_ps(v1,in);
+    u128f conv;    conv.v=v1;
+    return Grid::ComplexF(conv.f[0],conv.f[1]);
   }
   //Real float Reduce
   template<>
   inline Grid::RealF Reduce<Grid::RealF, __m128>::operator()(__m128 in){
-    // FIXME Hack
-    const Grid::RealF * ptr = (const Grid::RealF *) &in;
-    Grid::RealF ret = 0; 
-    for(int i=0;i< 4 ;i++){ // 4 number of simd lanes for float
-      ret = ret+ptr[i];
-    }
-    return ret;
+    __m128 v1,v2; // quad single
+    Optimization::permute(v1,in,0); 
+    v1= _mm_add_ps(v1,in);
+    Optimization::permute(v2,v1,1); 
+    v1 = _mm_add_ps(v1,v2);
+    u128f conv; conv.v=v1;
+    return conv.f[0];
   }
   
   
   //Complex double Reduce
   template<>
   inline Grid::ComplexD Reduce<Grid::ComplexD, __m128d>::operator()(__m128d in){
-    printf("Reduce : Missing good complex double implementation -> FIX\n");
-    return Grid::ComplexD(in[0], in[1]); // inefficient
+    u128d conv; conv.v = in;
+    return Grid::ComplexD(conv.f[0],conv.f[1]);
   }
   
   //Real double Reduce
   template<>
   inline Grid::RealD Reduce<Grid::RealD, __m128d>::operator()(__m128d in){
-    // FIXME Hack
-    const Grid::RealD * ptr =(const Grid::RealD *)  &in;
-    Grid::RealD ret = 0; 
-    for(int i=0;i< 2 ;i++){// 2 number of simd lanes for float
-      ret = ret+ptr[i];
-    }
-    return ret;
+    __m128d v1;
+    Optimization::permute(v1,in,0); // avx 256; quad double
+    v1 = _mm_add_pd(v1,in);
+    u128d conv; conv.v = v1;
+    return conv.f[0];
   }
 
   //Integer Reduce
@@ -276,12 +287,6 @@ namespace Optimization {
    printf("Reduce : Missing integer implementation -> FIX\n");
     assert(0);
   }
-  
-
-
-
-
-  
 }
 
 //////////////////////////////////////////////////////////////////////////////////////
@@ -292,27 +297,13 @@ namespace Grid {
   typedef __m128d SIMD_Dtype; // Double precision type
   typedef __m128i SIMD_Itype; // Integer type
 
-
   inline void v_prefetch0(int size, const char *ptr){};  // prefetch utilities
 
   // Gpermute function
   template < typename VectorSIMD > 
     inline void Gpermute(VectorSIMD &y,const VectorSIMD &b, int perm ) {
-    union { 
-      __m128 f;
-      decltype(VectorSIMD::v) v;
-    } conv;
-    conv.v = b.v;
-    switch(perm){
-    case 3: break; //empty for SSE4
-    case 2: break; //empty for SSE4
-    case 1: conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(2,3,0,1)); break;
-    case 0: conv.f = _mm_shuffle_ps(conv.f,conv.f,_MM_SHUFFLE(1,0,3,2)); break;
-    default: assert(0); break;
-    }
-    y.v=conv.v;
-  }; 
-  
+    Optimization::permute(y.v,b.v,perm);
+  }
 
 
   // Function name aliases

From a5c3424cfbdb56890e22c9e9b741beacdd65c831 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 28 May 2015 11:35:43 +0100
Subject: [PATCH 246/429] Works now with Clang-avx, Clang-sse and ICPC-avx,
 ICPC-sse

---
 lib/Makefile.am                        |   2 +-
 lib/simd/{Grid_knc.h => Grid_avx512.h} |   0
 lib/simd/Grid_vector_types.h           | 391 ++++++++++++-------------
 3 files changed, 186 insertions(+), 207 deletions(-)
 rename lib/simd/{Grid_knc.h => Grid_avx512.h} (100%)

diff --git a/lib/Makefile.am b/lib/Makefile.am
index f761b59e..a5b89c0a 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -96,7 +96,7 @@ nobase_include_HEADERS = algorithms/approx/bigfloat.h		\
 	simd/Grid_vector_types.h				\
 	simd/Grid_sse4.h					\
 	simd/Grid_avx.h						\
-	simd/Grid_knc.h					
+	simd/Grid_avx512.h					
 
 
 
diff --git a/lib/simd/Grid_knc.h b/lib/simd/Grid_avx512.h
similarity index 100%
rename from lib/simd/Grid_knc.h
rename to lib/simd/Grid_avx512.h
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index 0bedf32f..e389f3c2 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -14,7 +14,7 @@
 #include "Grid_avx.h"
 #endif
 #if defined AVX512
-#include "Grid_knc.h"
+#include "Grid_avx512.h"
 #endif
 #if defined QPX
 #include "Grid_qpx.h"
@@ -32,13 +32,24 @@ namespace Grid {
 
   // type alias used to simplify the syntax of std::enable_if
   template <typename T> using Invoke                                  =  typename T::type;
-  template <typename Condition, typename ReturnType> using EnableIf   =    Invoke<std::enable_if<Condition::value, ReturnType>>;
-  template <typename Condition, typename ReturnType> using NotEnableIf=    Invoke<std::enable_if<!Condition::value, ReturnType>>;
-  
+  template <typename Condition, typename ReturnType> using EnableIf   =  Invoke<std::enable_if<Condition::value, ReturnType> >;
+  template <typename Condition, typename ReturnType> using NotEnableIf=  Invoke<std::enable_if<!Condition::value, ReturnType> >;
+
+
   ////////////////////////////////////////////////////////
   // Check for complexity with type traits
-  template <typename T>     struct is_complex : std::false_type {};
-  template < typename T >   struct is_complex< std::complex<T> >: std::true_type {};
+  template <typename T>   struct is_complex : public std::false_type {};
+  template <> struct is_complex<std::complex<double> >: public std::true_type {};
+  template <> struct is_complex<std::complex<float> > : public std::true_type {};
+
+  template <typename T> using IfReal    = Invoke<std::enable_if<std::is_floating_point<T>::value,int> > ;
+  template <typename T> using IfComplex = Invoke<std::enable_if<is_complex<T>::value,int> > ;
+  template <typename T> using IfInteger = Invoke<std::enable_if<std::is_integral<T>::value,int> > ;
+
+  template <typename T> using IfNotReal    = Invoke<std::enable_if<!std::is_floating_point<T>::value,int> > ;
+  template <typename T> using IfNotComplex = Invoke<std::enable_if<!is_complex<T>::value,int> > ;
+  template <typename T> using IfNotInteger = Invoke<std::enable_if<!std::is_integral<T>::value,int> > ;
+
   ////////////////////////////////////////////////////////
   // Define the operation templates functors
   // general forms to allow for vsplat syntax
@@ -54,11 +65,9 @@ namespace Grid {
     Out unary(Input src, Operation op){
     return op(src);
   } 
-
   ///////////////////////////////////////////////
 
 
-  
 
   /*
     @brief Grid_simd class for the SIMD vector type operations
@@ -73,27 +82,28 @@ namespace Grid {
    
     Vector_type v;
     
-    
     static inline int Nsimd(void) { return sizeof(Vector_type)/sizeof(Scalar_type);}
-    
-    // Constructors
-    Grid_simd & operator = ( Zero & z){
-      vzero(*this);
-      return (*this);
-    }
+
     
     Grid_simd& operator=(const Grid_simd&& rhs){v=rhs.v;return *this;};
     Grid_simd& operator=(const Grid_simd& rhs){v=rhs.v;return *this;}; //faster than not declaring it and leaving to the compiler
     Grid_simd()=default; 
-    Grid_simd(const Grid_simd& rhs):v(rhs.v){};    //compiles in movaps
+    Grid_simd(const Grid_simd& rhs) :v(rhs.v){};    //compiles in movaps
     Grid_simd(const Grid_simd&& rhs):v(rhs.v){};  
+
+    /////////////////////////////
+    // Constructors
+    /////////////////////////////
+    Grid_simd & operator = ( Zero & z){
+      vzero(*this);
+      return (*this);
+    }
   
     //Enable if complex type
-    template < class S = Scalar_type > 
+    template < typename S = Scalar_type > 
     Grid_simd(const typename std::enable_if< is_complex < S >::value, S>::type a){
       vsplat(*this,a);
     };
-    
 
     Grid_simd(const Real a){
       vsplat(*this,Scalar_type(a));
@@ -107,86 +117,16 @@ namespace Grid {
     friend inline void sub (Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) - (*r); }
     friend inline void add (Grid_simd * __restrict__ y,const Grid_simd * __restrict__ l,const Grid_simd *__restrict__ r){ *y = (*l) + (*r); }
 
-
     friend inline void mac (Grid_simd *__restrict__ y,const Scalar_type *__restrict__ a,const Grid_simd   *__restrict__ x){ *y = (*a)*(*x)+(*y); };
     friend inline void mult(Grid_simd *__restrict__ y,const Scalar_type *__restrict__ l,const Grid_simd   *__restrict__ r){ *y = (*l) * (*r); }
     friend inline void sub (Grid_simd *__restrict__ y,const Scalar_type *__restrict__ l,const Grid_simd   *__restrict__ r){ *y = (*l) - (*r); }
     friend inline void add (Grid_simd *__restrict__ y,const Scalar_type *__restrict__ l,const Grid_simd   *__restrict__ r){ *y = (*l) + (*r); }
+
     friend inline void mac (Grid_simd *__restrict__ y,const Grid_simd   *__restrict__ a,const Scalar_type *__restrict__ x){ *y = (*a)*(*x)+(*y); };
     friend inline void mult(Grid_simd *__restrict__ y,const Grid_simd   *__restrict__ l,const Scalar_type *__restrict__ r){ *y = (*l) * (*r); }
     friend inline void sub (Grid_simd *__restrict__ y,const Grid_simd   *__restrict__ l,const Scalar_type *__restrict__ r){ *y = (*l) - (*r); }
     friend inline void add (Grid_simd *__restrict__ y,const Grid_simd   *__restrict__ l,const Scalar_type *__restrict__ r){ *y = (*l) + (*r); }
 
-
-
-    //not for integer types... 
-    template <  class S = Scalar_type, NotEnableIf<std::is_integral < S >, int> = 0 > 
-    friend inline Grid_simd adj(const Grid_simd &in){ return conjugate(in); }
-        
-    ///////////////////////////////////////////////
-    // Initialise to 1,0,i for the correct types
-    ///////////////////////////////////////////////
-    // For complex types
-    template <  class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 
-      friend inline void vone(Grid_simd &ret)      { vsplat(ret,1.0,0.0); }
-    template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 
-      friend inline void vzero(Grid_simd &ret)     { vsplat(ret,0.0,0.0); }// use xor?
-    template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 
-      friend inline void vcomplex_i(Grid_simd &ret){ vsplat(ret,0.0,1.0);} 
-
-    // if not complex overload here 
-    template <  class S = Scalar_type, EnableIf<std::is_floating_point < S >,int> = 0 > 
-      friend inline void vone(Grid_simd &ret)      { vsplat(ret,1.0); }
-    template <  class S = Scalar_type, EnableIf<std::is_floating_point < S >,int> = 0 > 
-      friend inline void vzero(Grid_simd &ret)     { vsplat(ret,0.0); }
-    
-
-   
-    // For integral types
-    template <  class S = Scalar_type, EnableIf<std::is_integral < S >, int> = 0 > 
-      friend inline void vone(Grid_simd &ret)      { vsplat(ret,1); }
-    template <  class S = Scalar_type, EnableIf<std::is_integral < S >, int> = 0 > 
-      friend inline void vzero(Grid_simd &ret)      { vsplat(ret,0); }
-    template <  class S = Scalar_type, EnableIf<std::is_integral < S >, int> = 0 > 
-      friend inline void vtrue (Grid_simd &ret){vsplat(ret,0xFFFFFFFF);}
-    template <  class S = Scalar_type, EnableIf<std::is_integral < S >, int> = 0 > 
-      friend inline void vfalse(Grid_simd &ret){vsplat(ret,0);}
-   
-    ////////////////////////////////////
-    // Arithmetic operator overloads +,-,*
-    ////////////////////////////////////
-    friend inline Grid_simd operator + (Grid_simd a, Grid_simd b)
-    {
-      Grid_simd ret;
-      ret.v = binary<Vector_type>(a.v, b.v, SumSIMD());
-      return ret;
-    };
-        
-    friend inline Grid_simd operator - (Grid_simd a, Grid_simd b)
-    {
-      Grid_simd ret;
-      ret.v = binary<Vector_type>(a.v, b.v, SubSIMD());
-      return ret;
-    };
-        
-    // Distinguish between complex types and others
-    template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 >
-      friend inline Grid_simd operator * (Grid_simd a, Grid_simd b)
-      {
-	Grid_simd ret;
-	ret.v = binary<Vector_type>(a.v,b.v, MultComplexSIMD());
-	return ret;
-      };
-
-    // Real/Integer types
-    template <  class S = Scalar_type, NotEnableIf<is_complex < S >, int> = 0 > 
-    friend inline Grid_simd operator * (Grid_simd a, Grid_simd b)
-      {
-	Grid_simd ret;
-	ret.v = binary<Vector_type>(a.v,b.v, MultSIMD());
-	return ret;
-      };
-
     ////////////////////////////////////////////////////////////////////////
     // FIXME:  gonna remove these load/store, get, set, prefetch
     ////////////////////////////////////////////////////////////////////////
@@ -194,28 +134,6 @@ namespace Grid {
       ret.v = unary<Vector_type>(a, VsetSIMD());
     }
         
-    ///////////////////////
-    // Splat
-    ///////////////////////
-    // overload if complex
-    template < class S = Scalar_type > 
-    friend inline void vsplat(Grid_simd &ret, EnableIf<is_complex < S >, S> c){
-      Real a = real(c);
-      Real b = imag(c);
-      vsplat(ret,a,b);
-    }
-
-    // this is only for the complex version
-    template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 
-    friend inline void vsplat(Grid_simd &ret,Real a, Real b){
-      ret.v = binary<Vector_type>(a, b, VsplatSIMD());
-    }    
-
-    //if real fill with a, if complex fill with a in the real part (first function above)
-    friend inline void vsplat(Grid_simd &ret,Real a){
-      ret.v = unary<Vector_type>(a, VsplatSIMD());
-    }    
-
     ///////////////////////
     // Vstore
     ///////////////////////
@@ -223,19 +141,6 @@ namespace Grid {
       binary<void>(ret.v, (Real*)a, VstoreSIMD());
     }
 
-    ///////////////////////
-    // Vstream
-    ///////////////////////
-    template <  class S = Scalar_type, NotEnableIf<std::is_integral < S >, int> = 0 > 
-    friend inline void vstream(Grid_simd &out,const Grid_simd &in){
-      binary<void>((Real*)&out.v, in.v, VstreamSIMD());
-    }
-
-    template <  class S = Scalar_type, EnableIf<std::is_integral < S >, int> = 0 > 
-      friend inline void vstream(Grid_simd &out,const Grid_simd &in){
-      out=in;
-    }
-    
     ///////////////////////
     // Vprefetch
     ///////////////////////
@@ -244,7 +149,6 @@ namespace Grid {
       _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
     }
 
-
     ///////////////////////
     // Reduce
     ///////////////////////
@@ -265,63 +169,6 @@ namespace Grid {
       return a*b;
     }
 
-    ///////////////////////
-    // Conjugate
-    ///////////////////////
-    template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 							     
-    friend inline Grid_simd  conjugate(const Grid_simd  &in){
-      Grid_simd  ret ; 
-      ret.v = unary<Vector_type>(in.v, ConjSIMD());
-      return ret;
-    }
-    template < class S = Scalar_type, NotEnableIf<is_complex < S >, int> = 0 > 
-    friend inline Grid_simd  conjugate(const Grid_simd  &in){
-      return in; // for real objects
-    }
-
-
-    ///////////////////////
-    // timesMinusI
-    ///////////////////////
-    template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 
-    friend inline void timesMinusI( Grid_simd &ret,const Grid_simd &in){
-      ret.v = binary<Vector_type>(in.v, ret.v, TimesMinusISIMD());
-    }
-
-    template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 
-    friend inline Grid_simd timesMinusI(const Grid_simd &in){
-      Grid_simd ret; 
-      timesMinusI(ret,in);
-      return ret;
-    }
-
-    template < class S = Scalar_type, NotEnableIf<is_complex < S >, int> = 0 > 
-    friend inline Grid_simd timesMinusI(const Grid_simd &in){
-      return in;
-    }
-
-
-    ///////////////////////
-    // timesI
-    ///////////////////////
-    template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 
-    friend inline void timesI(Grid_simd &ret,const Grid_simd &in){
-      ret.v =   binary<Vector_type>(in.v, ret.v, TimesISIMD());     
-    }
-        
-    template < class S = Scalar_type, EnableIf<is_complex < S >, int> = 0 > 
-    friend inline Grid_simd timesI(const Grid_simd &in){
-      Grid_simd ret; 
-      timesI(ret,in);
-      return ret;
-    }
-
-    template < class S = Scalar_type, NotEnableIf<is_complex < S >, int> = 0 > 
-    friend inline Grid_simd timesI(const Grid_simd &in){
-      return in;
-    }
-
-
     ///////////////////////
     // Unary negation
     ///////////////////////
@@ -346,9 +193,6 @@ namespace Grid {
       return *this;
     }
 
-
-
-
     ////////////////////////////////////////////////////////////////////
     // General permute; assumes vector length is same across 
     // all subtypes; may not be a good assumption, but could
@@ -359,48 +203,183 @@ namespace Grid {
       Gpermute<Grid_simd>(y,b,perm);
     }
     
-    
   };// end of Grid_simd class definition 
 
 
+  ///////////////////////
+  // Splat
+  ///////////////////////
+  
+  // this is only for the complex version
+  template <class S, class V, IfComplex<S> =0, class ABtype> 
+    inline void vsplat(Grid_simd<S,V> &ret,ABtype a, ABtype b){
+    ret.v = binary<V>(a, b, VsplatSIMD());
+  }    
 
-  template<class scalar_type, class vector_type > 
-    inline Grid_simd< scalar_type, vector_type>  innerProduct(const Grid_simd< scalar_type, vector_type> & l, const Grid_simd< scalar_type, vector_type> & r) 
+  // overload if complex
+  template <class S,class V> inline void vsplat(Grid_simd<S,V> &ret, EnableIf<is_complex < S >, S> c) {
+    Real a = real(c);
+    Real b = imag(c);
+    vsplat(ret,a,b);
+  }
+
+  //if real fill with a, if complex fill with a in the real part (first function above)
+  template <class S,class V>
+    inline void vsplat(Grid_simd<S,V> &ret,NotEnableIf<is_complex< S>,S> a){
+    ret.v = unary<V>(a, VsplatSIMD());
+  }    
+  //////////////////////////
+
+  ///////////////////////////////////////////////
+  // Initialise to 1,0,i for the correct types
+  ///////////////////////////////////////////////
+  // For complex types
+  template <class S,class V, IfComplex<S> = 0 > inline void vone(Grid_simd<S,V>  &ret)     { vsplat(ret,S(1.0,0.0)); }
+  template <class S,class V, IfComplex<S> = 0 > inline void vzero(Grid_simd<S,V> &ret)     { vsplat(ret,S(0.0,0.0)); }// use xor?
+  template <class S,class V, IfComplex<S> = 0 > inline void vcomplex_i(Grid_simd<S,V> &ret){ vsplat(ret,S(0.0,1.0));} 
+
+  // if not complex overload here 
+  template <class S,class V, IfReal<S> = 0 > inline void vone (Grid_simd<S,V> &ret){ vsplat(ret,1.0); }
+  template <class S,class V, IfReal<S> = 0 > inline void vzero(Grid_simd<S,V> &ret)     { vsplat(ret,0.0); }
+   
+  // For integral types
+  template <class S,class V,IfInteger<S> = 0 > inline void vone(Grid_simd<S,V> &ret)  {vsplat(ret,1); }
+  template <class S,class V,IfInteger<S> = 0 > inline void vzero(Grid_simd<S,V> &ret) {vsplat(ret,0); }
+  template <class S,class V,IfInteger<S> = 0 > inline void vtrue (Grid_simd<S,V> &ret){vsplat(ret,0xFFFFFFFF);}
+  template <class S,class V,IfInteger<S> = 0 > inline void vfalse(Grid_simd<S,V> &ret){vsplat(ret,0);}
+
+  template<class S,class V> inline void zeroit(Grid_simd<S,V> &z){ vzero(z);}
+
+  ///////////////////////
+  // Vstream
+  ///////////////////////
+  template <class S,class V, IfNotInteger<S> = 0 > 
+    inline void vstream(Grid_simd<S,V> &out,const Grid_simd<S,V> &in){
+      binary<void>((Real*)&out.v, in.v, VstreamSIMD());
+    }
+
+  template <class S,class V, IfInteger<S> = 0 > 
+    inline void vstream(Grid_simd<S,V> &out,const Grid_simd<S,V> &in){
+    out=in;
+  }
+
+  ////////////////////////////////////
+  // Arithmetic operator overloads +,-,*
+  ////////////////////////////////////
+  template<class S,class V> inline Grid_simd<S,V> operator + (Grid_simd<S,V> a, Grid_simd<S,V> b) {
+    Grid_simd<S,V> ret;
+    ret.v = binary<V>(a.v, b.v, SumSIMD());
+    return ret;
+  };
+        
+  template<class S,class V> inline Grid_simd<S,V> operator - (Grid_simd<S,V> a, Grid_simd<S,V> b) {
+    Grid_simd<S,V> ret;
+    ret.v = binary<V>(a.v, b.v, SubSIMD());
+    return ret;
+  };
+        
+  // Distinguish between complex types and others
+  template<class S,class V, IfComplex<S> = 0 > inline Grid_simd<S,V> operator * (Grid_simd<S,V> a, Grid_simd<S,V> b) {
+    Grid_simd<S,V> ret;
+    ret.v = binary<V>(a.v,b.v, MultComplexSIMD());
+    return ret;
+  };
+
+    // Real/Integer types
+  template<class S,class V, IfNotComplex<S> = 0 > inline Grid_simd<S,V> operator * (Grid_simd<S,V> a, Grid_simd<S,V> b) {
+    Grid_simd<S,V> ret;
+    ret.v = binary<V>(a.v,b.v, MultSIMD());
+    return ret;
+  };
+  
+
+    ///////////////////////
+    // Conjugate
+    ///////////////////////
+  template <class S,class V, IfComplex<S> = 0 > 
+    inline Grid_simd<S,V> conjugate(const Grid_simd<S,V>  &in){
+    Grid_simd<S,V>  ret ; 
+    ret.v = unary<V>(in.v, ConjSIMD());
+    return ret;
+  }
+  template <class S,class V, IfNotComplex<S> = 0 > inline Grid_simd<S,V> conjugate(const Grid_simd<S,V>  &in){
+    return in; // for real objects
+  }
+
+  //Suppress adj for integer types... // odd; why conjugate above but not adj??
+  template < class S, class V, IfNotInteger<S> = 0 > 
+    inline Grid_simd<S,V> adj(const Grid_simd<S,V> &in){ return conjugate(in); }
+  
+  ///////////////////////
+  // timesMinusI
+  ///////////////////////
+  template<class S,class V,IfComplex<S> = 0 > 
+    inline void timesMinusI( Grid_simd<S,V> &ret,const Grid_simd<S,V> &in){
+    ret.v = binary<V>(in.v, ret.v, TimesMinusISIMD());
+  }
+
+  template<class S,class V,IfComplex<S> = 0 > 
+    inline Grid_simd<S,V> timesMinusI(const Grid_simd<S,V> &in){
+    Grid_simd<S,V> ret; 
+    timesMinusI(ret,in);
+    return ret;
+  }
+
+  template<class S,class V,IfNotComplex<S> = 0 > 
+    inline Grid_simd<S,V> timesMinusI(const Grid_simd<S,V> &in){
+    return in;
+  }
+
+    ///////////////////////
+    // timesI
+    ///////////////////////
+  template<class S,class V,IfComplex<S> = 0 > 
+    inline void timesI(Grid_simd<S,V> &ret,const Grid_simd<S,V> &in){
+    ret.v =   binary<V>(in.v, ret.v, TimesISIMD());     
+  }
+        
+  template<class S,class V,IfComplex<S> = 0 > 
+    inline Grid_simd<S,V> timesI(const Grid_simd<S,V> &in){
+    Grid_simd<S,V> ret; 
+    timesI(ret,in);
+    return ret;
+  }
+
+  template<class S,class V,IfNotComplex<S> = 0 > 
+    inline Grid_simd<S,V> timesI(const Grid_simd<S,V> &in){
+    return in;
+  }
+
+
+  /////////////////////
+  // Inner, outer
+  /////////////////////
+
+  template<class S, class V > 
+    inline Grid_simd< S, V>  innerProduct(const Grid_simd< S, V> & l, const Grid_simd< S, V> & r) 
   {
     return conjugate(l)*r; 
   }
 
-  template<class scalar_type, class vector_type >
-    inline void zeroit(Grid_simd< scalar_type, vector_type> &z){ vzero(z);}
-
-
-  template<class scalar_type, class vector_type >
-  inline Grid_simd< scalar_type, vector_type> outerProduct(const Grid_simd< scalar_type, vector_type> &l, const Grid_simd< scalar_type, vector_type>& r)
+  template<class S, class V >
+  inline Grid_simd< S, V> outerProduct(const Grid_simd< S, V> &l, const Grid_simd< S, V> & r)
   {
     return l*r;
   }
 
-
-  template<class scalar_type, class vector_type >
-  inline Grid_simd< scalar_type, vector_type> trace(const Grid_simd< scalar_type, vector_type> &arg){
+  template<class S, class V >
+  inline Grid_simd< S, V> trace(const Grid_simd< S, V> &arg){
     return arg;
   }
 
-
+  ///////////////////////////////
   // Define available types
-
+  ///////////////////////////////
   typedef Grid_simd< float                 , SIMD_Ftype > vRealF;
   typedef Grid_simd< double                , SIMD_Dtype > vRealD;
   typedef Grid_simd< std::complex< float > , SIMD_Ftype > vComplexF;
   typedef Grid_simd< std::complex< double >, SIMD_Dtype > vComplexD;
   typedef Grid_simd< Integer               , SIMD_Itype > vInteger;
-
-
-
-
-
-
-
 }
 
 #endif

From 445e38acf6721cc7d39760103e1677d41744f586 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 28 May 2015 13:47:01 +0100
Subject: [PATCH 247/429] Weak scale the benchmarks automatically.

---
 benchmarks/Grid_memory_bandwidth.cc | 31 +++++++++++----------
 benchmarks/Grid_su3.cc              | 31 +++++++++++----------
 lib/Grid_init.cc                    |  1 -
 lib/cshift/Grid_cshift_common.h     | 43 ++++++++++++++++++-----------
 4 files changed, 60 insertions(+), 46 deletions(-)

diff --git a/benchmarks/Grid_memory_bandwidth.cc b/benchmarks/Grid_memory_bandwidth.cc
index fc959c38..5abdc6a3 100644
--- a/benchmarks/Grid_memory_bandwidth.cc
+++ b/benchmarks/Grid_memory_bandwidth.cc
@@ -27,8 +27,8 @@ int main (int argc, char ** argv)
 
   for(int lat=4;lat<=32;lat+=4){
 
-      std::vector<int> latt_size  ({lat,lat,lat,lat});
-
+      std::vector<int> latt_size  ({lat*mpi_layout[0],lat*mpi_layout[1],lat*mpi_layout[2],lat*mpi_layout[3]});
+      int vol = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
       GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
 
       //GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
@@ -47,8 +47,8 @@ int main (int argc, char ** argv)
       double stop=usecond();
       double time = (stop-start)/Nloop*1000;
       
-      double flops=lat*lat*lat*lat*Nvec*2;// mul,add
-      double bytes=3*lat*lat*lat*lat*Nvec*sizeof(Real);
+      double flops=vol*Nvec*2;// mul,add
+      double bytes=3*vol*Nvec*sizeof(Real);
       std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"   \t\t"<<bytes/time<<"\t\t"<<flops/time<<std::endl;
 
     }
@@ -61,8 +61,8 @@ int main (int argc, char ** argv)
   
   for(int lat=4;lat<=32;lat+=4){
 
-      std::vector<int> latt_size  ({lat,lat,lat,lat});
-
+      std::vector<int> latt_size  ({lat*mpi_layout[0],lat*mpi_layout[1],lat*mpi_layout[2],lat*mpi_layout[3]});
+      int vol = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
       GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
 
       //GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
@@ -79,8 +79,8 @@ int main (int argc, char ** argv)
       double stop=usecond();
       double time = (stop-start)/Nloop*1000;
      
-      double flops=lat*lat*lat*lat*Nvec*2;// mul,add
-      double bytes=3*lat*lat*lat*lat*Nvec*sizeof(Real);
+      double flops=vol*Nvec*2;// mul,add
+      double bytes=3*vol*Nvec*sizeof(Real);
       std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"   \t\t"<<bytes/time<<"\t\t"<<flops/time<<std::endl;
 
     }
@@ -92,7 +92,8 @@ int main (int argc, char ** argv)
 
   for(int lat=4;lat<=32;lat+=4){
 
-      std::vector<int> latt_size  ({lat,lat,lat,lat});
+      std::vector<int> latt_size  ({lat*mpi_layout[0],lat*mpi_layout[1],lat*mpi_layout[2],lat*mpi_layout[3]});
+      int vol = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
 
       GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
 
@@ -111,8 +112,8 @@ int main (int argc, char ** argv)
       double stop=usecond();
       double time = (stop-start)/Nloop*1000;
       
-      double bytes=2*lat*lat*lat*lat*Nvec*sizeof(Real);
-      double flops=lat*lat*lat*lat*Nvec*1;// mul
+      double bytes=2*vol*Nvec*sizeof(Real);
+      double flops=vol*Nvec*1;// mul
       std::cout <<std::setprecision(3) << lat<<"\t\t"<<bytes<<"   \t\t"<<bytes/time<<"\t\t"<<flops/time<<std::endl;
 
   }
@@ -125,8 +126,8 @@ int main (int argc, char ** argv)
 
   for(int lat=4;lat<=32;lat+=4){
 
-      std::vector<int> latt_size  ({lat,lat,lat,lat});
-
+      std::vector<int> latt_size  ({lat*mpi_layout[0],lat*mpi_layout[1],lat*mpi_layout[2],lat*mpi_layout[3]});
+      int vol = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
       GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
 
       //GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
@@ -144,8 +145,8 @@ int main (int argc, char ** argv)
       double stop=usecond();
       double time = (stop-start)/Nloop*1000;
       
-      double bytes=lat*lat*lat*lat*Nvec*sizeof(Real);
-      double flops=lat*lat*lat*lat*Nvec*2;// mul,add
+      double bytes=vol*Nvec*sizeof(Real);
+      double flops=vol*Nvec*2;// mul,add
       std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"  \t\t"<<bytes/time<<"\t\t"<<flops/time<<std::endl;
 
   }    
diff --git a/benchmarks/Grid_su3.cc b/benchmarks/Grid_su3.cc
index d64e8706..903ea348 100644
--- a/benchmarks/Grid_su3.cc
+++ b/benchmarks/Grid_su3.cc
@@ -24,8 +24,8 @@ int main (int argc, char ** argv)
 
   for(int lat=2;lat<=24;lat+=2){
 
-      std::vector<int> latt_size  ({lat,lat,lat,lat});
-
+      std::vector<int> latt_size  ({lat*mpi_layout[0],lat*mpi_layout[1],lat*mpi_layout[2],lat*mpi_layout[3]});
+      int vol = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
       GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
       //      GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
 
@@ -40,9 +40,9 @@ int main (int argc, char ** argv)
       double stop=usecond();
       double time = (stop-start)/Nloop*1000.0;
       
-      double bytes=3.0*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
-      double footprint=2.0*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
-      double flops=Nc*Nc*(6.0+8.0+8.0)*lat*lat*lat*lat;
+      double bytes=3.0*vol*Nc*Nc*sizeof(Complex);
+      double footprint=2.0*vol*Nc*Nc*sizeof(Complex);
+      double flops=Nc*Nc*(6.0+8.0+8.0)*vol;
       std::cout<<std::setprecision(3) << lat<<"\t\t"<<footprint<<"    \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
 
     }
@@ -56,7 +56,8 @@ int main (int argc, char ** argv)
 
   for(int lat=2;lat<=24;lat+=2){
 
-      std::vector<int> latt_size  ({lat,lat,lat,lat});
+      std::vector<int> latt_size  ({lat*mpi_layout[0],lat*mpi_layout[1],lat*mpi_layout[2],lat*mpi_layout[3]});
+      int vol = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
 
       GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
       //      GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
@@ -72,8 +73,8 @@ int main (int argc, char ** argv)
       double stop=usecond();
       double time = (stop-start)/Nloop*1000.0;
       
-      double bytes=3*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
-      double flops=Nc*Nc*(6+8+8)*lat*lat*lat*lat;
+      double bytes=3*vol*Nc*Nc*sizeof(Complex);
+      double flops=Nc*Nc*(6+8+8)*vol;
       std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"    \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
 
     }
@@ -86,7 +87,8 @@ int main (int argc, char ** argv)
 
   for(int lat=2;lat<=24;lat+=2){
 
-      std::vector<int> latt_size  ({lat,lat,lat,lat});
+      std::vector<int> latt_size  ({lat*mpi_layout[0],lat*mpi_layout[1],lat*mpi_layout[2],lat*mpi_layout[3]});
+      int vol = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
 
       GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
       //      GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
@@ -102,8 +104,8 @@ int main (int argc, char ** argv)
       double stop=usecond();
       double time = (stop-start)/Nloop*1000.0;
       
-      double bytes=3*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
-      double flops=Nc*Nc*(6+8+8)*lat*lat*lat*lat;
+      double bytes=3*vol*Nc*Nc*sizeof(Complex);
+      double flops=Nc*Nc*(6+8+8)*vol;
       std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"    \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
 
     }
@@ -116,7 +118,8 @@ int main (int argc, char ** argv)
 
   for(int lat=2;lat<=24;lat+=2){
 
-      std::vector<int> latt_size  ({lat,lat,lat,lat});
+      std::vector<int> latt_size  ({lat*mpi_layout[0],lat*mpi_layout[1],lat*mpi_layout[2],lat*mpi_layout[3]});
+      int vol = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
 
       GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
       //      GridParallelRNG          pRNG(&Grid);      pRNG.SeedRandomDevice();
@@ -132,8 +135,8 @@ int main (int argc, char ** argv)
       double stop=usecond();
       double time = (stop-start)/Nloop*1000.0;
       
-      double bytes=3*lat*lat*lat*lat*Nc*Nc*sizeof(Complex);
-      double flops=Nc*Nc*(8+8+8)*lat*lat*lat*lat;
+      double bytes=3*vol*Nc*Nc*sizeof(Complex);
+      double flops=Nc*Nc*(8+8+8)*vol;
       std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"   \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
 
     }
diff --git a/lib/Grid_init.cc b/lib/Grid_init.cc
index 7fce4793..f72393cb 100644
--- a/lib/Grid_init.cc
+++ b/lib/Grid_init.cc
@@ -54,7 +54,6 @@ namespace Grid {
   const std::vector<int> &GridDefaultLatt(void)     {return Grid_default_latt;};
   const std::vector<int> &GridDefaultMpi(void)      {return Grid_default_mpi;};
 
-
   ////////////////////////////////////////////////////////////
   // Command line parsing assist for stock controls
   ////////////////////////////////////////////////////////////
diff --git a/lib/cshift/Grid_cshift_common.h b/lib/cshift/Grid_cshift_common.h
index 97e66b2c..c369fe1c 100644
--- a/lib/cshift/Grid_cshift_common.h
+++ b/lib/cshift/Grid_cshift_common.h
@@ -27,9 +27,11 @@ Gather_plane_simple (const Lattice<vobj> &rhs,std::vector<cobj,alignedAllocator<
 
   int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
   
+  int e1=rhs._grid->_slice_nblock[dimension];
+  int e2=rhs._grid->_slice_block[dimension];
 PARALLEL_NESTED_LOOP2
-  for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-    for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+  for(int n=0;n<e1;n++){
+    for(int b=0;b<e2;b++){
       int o  = n*rhs._grid->_slice_stride[dimension];
       int bo = n*rhs._grid->_slice_block[dimension];
       int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);// Could easily be a table lookup
@@ -54,10 +56,12 @@ Gather_plane_extract(const Lattice<vobj> &rhs,std::vector<typename cobj::scalar_
   }
 
   int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
-    
+
+  int e1=rhs._grid->_slice_nblock[dimension];
+  int e2=rhs._grid->_slice_block[dimension];
 PARALLEL_NESTED_LOOP2
-  for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-    for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+  for(int n=0;n<e1;n++){
+    for(int b=0;b<e2;b++){
 
       int o=n*rhs._grid->_slice_stride[dimension];
       int offset = b+n*rhs._grid->_slice_block[dimension];
@@ -103,9 +107,11 @@ template<class vobj> void Scatter_plane_simple (Lattice<vobj> &rhs,std::vector<v
 
   int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
     
+  int e1=rhs._grid->_slice_nblock[dimension];
+  int e2=rhs._grid->_slice_block[dimension];
 PARALLEL_NESTED_LOOP2
-  for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-    for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+  for(int n=0;n<e1;n++){
+    for(int b=0;b<e2;b++){
       int o   =n*rhs._grid->_slice_stride[dimension];
       int bo  =n*rhs._grid->_slice_block[dimension];
       int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);// Could easily be a table lookup
@@ -129,10 +135,11 @@ PARALLEL_NESTED_LOOP2
 
   int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
     
+  int e1=rhs._grid->_slice_nblock[dimension];
+  int e2=rhs._grid->_slice_block[dimension];
 PARALLEL_NESTED_LOOP2
-  for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-    for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
-
+  for(int n=0;n<e1;n++){
+    for(int b=0;b<e2;b++){
       int o      = n*rhs._grid->_slice_stride[dimension];
       int offset = b+n*rhs._grid->_slice_block[dimension];
       int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);
@@ -156,10 +163,12 @@ template<class vobj> void Copy_plane(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int
 
   int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
   int lo  = lplane*lhs._grid->_ostride[dimension]; // base offset for start of plane 
-  
+
+  int e1=rhs._grid->_slice_nblock[dimension]; // clearly loop invariant for icpc
+  int e2=rhs._grid->_slice_block[dimension];
 PARALLEL_NESTED_LOOP2
-  for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-    for(int b=0;b<rhs._grid->_slice_block[dimension];b++){
+  for(int n=0;n<e1;n++){
+    for(int b=0;b<e2;b++){
  
       int o =n*rhs._grid->_slice_stride[dimension]+b;
       int ocb=1<<lhs._grid->CheckerBoardFromOindex(o);
@@ -185,10 +194,12 @@ template<class vobj> void Copy_plane_permute(Lattice<vobj>& lhs,Lattice<vobj> &r
 
   int ro  = rplane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
   int lo  = lplane*lhs._grid->_ostride[dimension]; // base offset for start of plane 
-  
+
+  int e1=rhs._grid->_slice_nblock[dimension];
+  int e2=rhs._grid->_slice_block [dimension];
 PARALLEL_NESTED_LOOP2
-  for(int n=0;n<rhs._grid->_slice_nblock[dimension];n++){
-  for(int b=0;b<rhs._grid->_slice_block [dimension];b++){
+  for(int n=0;n<e1;n++){
+  for(int b=0;b<e2;b++){
 
       int o  =n*rhs._grid->_slice_stride[dimension];
       int ocb=1<<lhs._grid->CheckerBoardFromOindex(o+b);

From 4403e117a76529e73be8b881a3f3ad18c4215877 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Fri, 29 May 2015 11:41:02 +0900
Subject: [PATCH 248/429] Some modifications to the configure to check SIMD
 support

---
 configure    | 2 +-
 configure.ac | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/configure b/configure
index ea0a71c2..2d785792 100755
--- a/configure
+++ b/configure
@@ -6709,7 +6709,7 @@ $as_echo "$as_me: WARNING: Your processor does not support AVX2 instructions" >&
 
 $as_echo "#define AVX512 1" >>confdefs.h
 
-       supported=yes
+       supported="cross compilation"
      ;;
      *)
      as_fn_error $? "${ac_SIMD} unsupported --enable-simd option" "$LINENO" 5;
diff --git a/configure.ac b/configure.ac
index 00622bb8..1edb3e33 100644
--- a/configure.ac
+++ b/configure.ac
@@ -3,7 +3,7 @@
 #
 # Project Grid package  
 # 
-# Time-stamp: <2015-05-27 18:29:04 neo>
+# Time-stamp: <2015-05-27 18:51:47 neo>
 
 AC_PREREQ([2.63])
 AC_INIT([Grid], [1.0], [paboyle@ph.ed.ac.uk])
@@ -106,7 +106,7 @@ case ${ac_SIMD} in
      AVX512|MIC)
        echo Configuring for AVX512 and MIC
        AC_DEFINE([AVX512],[1],[AVX512] )
-       supported=yes
+       supported="cross compilation"
      ;;
      *)
      AC_MSG_ERROR([${ac_SIMD} unsupported --enable-simd option]); 

From 9cfc180334ab7aa16ce90405d7d11f51d164a94a Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 29 May 2015 14:11:34 +0100
Subject: [PATCH 249/429] Integer wrap problem fixed.

---
 benchmarks/Grid_comms.cc | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/benchmarks/Grid_comms.cc b/benchmarks/Grid_comms.cc
index 59d709a0..c218070b 100644
--- a/benchmarks/Grid_comms.cc
+++ b/benchmarks/Grid_comms.cc
@@ -77,7 +77,8 @@ int main (int argc, char ** argv)
       }
       double stop=usecond();
 
-      double xbytes    = Nloop*bytes*2*ncomm;
+      double dbytes    = bytes;
+      double xbytes    = Nloop*dbytes*2.0*ncomm;
       double rbytes    = xbytes;
       double bidibytes = xbytes+rbytes;
 
@@ -151,8 +152,9 @@ int main (int argc, char ** argv)
       }
 
       double stop=usecond();
-
-      double xbytes    = Nloop*bytes*2*ncomm;
+      
+      double dbytes    = bytes;
+      double xbytes    = Nloop*dbytes*2.0*ncomm;
       double rbytes    = xbytes;
       double bidibytes = xbytes+rbytes;
 

From a75b6f6e78fab299b6c08f983ca92b4c4ebfcaf8 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sun, 31 May 2015 15:09:02 +0100
Subject: [PATCH 250/429] Large scale change to support 5d fermion
 formulations. Have 5d replicated wilson with 4d gauge working and matrix
 regressing to Ls copies of wilson.

---
 benchmarks/Grid_comms.cc                      |   2 +-
 benchmarks/Grid_dwf.cc                        | 145 +++++++++++
 benchmarks/Grid_wilson.cc                     |   9 +-
 benchmarks/Grid_wilson_cg_prec.cc             |   6 +-
 benchmarks/Grid_wilson_cg_schur.cc            |   2 +-
 benchmarks/Grid_wilson_cg_unprec.cc           |   4 +-
 benchmarks/Grid_wilson_evenodd.cc             |   2 +-
 benchmarks/Makefile.am                        |   5 +-
 lib/Grid.h                                    |   2 +-
 lib/Grid_init.cc                              |   6 +-
 lib/Grid_stencil.h                            |  33 +--
 lib/Grid_threads.h                            |   2 +-
 lib/Makefile.am                               | 166 +++++++------
 lib/algorithms/SparseMatrix.h                 |   7 +-
 lib/algorithms/iterative/SchurRedBlack.h      |   4 +-
 lib/cartesian/Grid_cartesian_base.h           |  11 +-
 lib/cartesian/Grid_cartesian_full.h           |   7 +-
 lib/cartesian/Grid_cartesian_red_black.h      | 196 ++++++++++-----
 lib/cshift/Grid_cshift_common.h               |  14 +-
 lib/cshift/Grid_cshift_mpi.h                  |  16 +-
 lib/cshift/Grid_cshift_none.h                 |   2 +-
 lib/lattice/Grid_lattice_peekpoke.h           |   2 +-
 lib/lattice/Grid_lattice_transfer.h           |   3 +-
 lib/qcd/{Grid_qcd_dirac.cc => Dirac.cc}       |   0
 lib/qcd/{Grid_qcd_dirac.h => Dirac.h}         |   0
 lib/qcd/Grid_qcd_wilson_dop.cc                | 217 -----------------
 lib/qcd/Grid_qcd_wilson_dop.h                 | 105 --------
 lib/qcd/{Grid_qcd.h => QCD.h}                 |  17 +-
 lib/qcd/{Grid_qcd_2spinor.h => TwoSpinor.h}   |   0
 lib/qcd/action/Actions.h                      |  10 +
 lib/qcd/action/fermion/FermionAction.h        |  47 ++++
 .../action/fermion/FiveDimWilsonFermion.cc    | 228 ++++++++++++++++++
 lib/qcd/action/fermion/FiveDimWilsonFermion.h | 108 +++++++++
 lib/qcd/action/fermion/WilsonCompressor.h     |  61 +++++
 lib/qcd/action/fermion/WilsonFermion.cc       | 163 +++++++++++++
 lib/qcd/action/fermion/WilsonFermion.h        |  87 +++++++
 .../fermion/WilsonKernels.cc}                 |  46 ++--
 lib/qcd/action/fermion/WilsonKernels.h        |  42 ++++
 .../fermion/WilsonKernelsHand.cc}             |  57 ++---
 lib/stencil/Grid_lebesgue.cc                  | 103 ++++++++
 lib/stencil/Grid_lebesgue.h                   |  29 +++
 lib/stencil/Grid_stencil_common.cc            | 103 +-------
 scripts/linecount                             |   2 +-
 tests/Grid_cshift_red_black.cc                | 165 +++++++++++++
 tests/Makefile.am                             |   5 +-
 45 files changed, 1549 insertions(+), 692 deletions(-)
 create mode 100644 benchmarks/Grid_dwf.cc
 rename lib/qcd/{Grid_qcd_dirac.cc => Dirac.cc} (100%)
 rename lib/qcd/{Grid_qcd_dirac.h => Dirac.h} (100%)
 delete mode 100644 lib/qcd/Grid_qcd_wilson_dop.cc
 delete mode 100644 lib/qcd/Grid_qcd_wilson_dop.h
 rename lib/qcd/{Grid_qcd.h => QCD.h} (97%)
 rename lib/qcd/{Grid_qcd_2spinor.h => TwoSpinor.h} (100%)
 create mode 100644 lib/qcd/action/Actions.h
 create mode 100644 lib/qcd/action/fermion/FermionAction.h
 create mode 100644 lib/qcd/action/fermion/FiveDimWilsonFermion.cc
 create mode 100644 lib/qcd/action/fermion/FiveDimWilsonFermion.h
 create mode 100644 lib/qcd/action/fermion/WilsonCompressor.h
 create mode 100644 lib/qcd/action/fermion/WilsonFermion.cc
 create mode 100644 lib/qcd/action/fermion/WilsonFermion.h
 rename lib/qcd/{Grid_qcd_dhop.cc => action/fermion/WilsonKernels.cc} (87%)
 create mode 100644 lib/qcd/action/fermion/WilsonKernels.h
 rename lib/qcd/{Grid_qcd_dhop_hand.cc => action/fermion/WilsonKernelsHand.cc} (92%)
 create mode 100644 lib/stencil/Grid_lebesgue.cc
 create mode 100644 lib/stencil/Grid_lebesgue.h
 create mode 100644 tests/Grid_cshift_red_black.cc

diff --git a/benchmarks/Grid_comms.cc b/benchmarks/Grid_comms.cc
index c218070b..a24e7ed8 100644
--- a/benchmarks/Grid_comms.cc
+++ b/benchmarks/Grid_comms.cc
@@ -82,7 +82,7 @@ int main (int argc, char ** argv)
       double rbytes    = xbytes;
       double bidibytes = xbytes+rbytes;
 
-      double time = stop-start;
+      double time = stop-start; // microseconds
 
       std::cout << lat<<"\t\t"<<Ls<<"\t\t"<<bytes<<"\t\t"<<xbytes/time<<"\t\t"<<bidibytes/time<<std::endl;
     }
diff --git a/benchmarks/Grid_dwf.cc b/benchmarks/Grid_dwf.cc
new file mode 100644
index 00000000..eb1d9299
--- /dev/null
+++ b/benchmarks/Grid_dwf.cc
@@ -0,0 +1,145 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  int threads = GridThread::GetThreads();
+  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+
+  std::vector<int> latt4 = GridDefaultLatt();
+  std::vector<int> simd4 = GridDefaultSimd(Nd,vComplexF::Nsimd());
+  std::vector<int> mpi4  = GridDefaultMpi();
+
+  assert(latt4.size()==4 ); 
+  assert(simd4.size()==4 );
+  assert(mpi4.size() ==4 );
+
+  const int Ls=1;
+  std::vector<int> latt5({Ls,latt4[0],latt4[1],latt4[2],latt4[3]});
+  std::vector<int> simd5({1 ,simd4[0],simd4[1],simd4[2],simd4[3]}); 
+  std::vector<int>  mpi5({1 , mpi4[0], mpi4[1], mpi4[2], mpi4[3]}); 
+  std::vector<int>   cb5({0,1,1,1,1}); // Checkerboard 4d only
+  int                cbd=1;            // use dim-1 to reduce
+
+  // Four dim grid for gauge field U
+  GridCartesian               UGrid(latt4,simd4,mpi4); 
+  GridRedBlackCartesian     UrbGrid(&UGrid);
+
+  // Five dim grid for fermions F
+  GridCartesian               FGrid(latt5,simd5,mpi5); 
+  GridRedBlackCartesian     FrbGrid(latt5,simd5,mpi5,cb5,cbd); 
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+
+  GridParallelRNG          RNG5(&FGrid);  RNG5.SeedFixedIntegers(seeds5);
+  LatticeFermion src   (&FGrid); random(RNG5,src);
+  LatticeFermion result(&FGrid); result=zero;
+  LatticeFermion    ref(&FGrid);    ref=zero;
+  LatticeFermion    tmp(&FGrid);
+  LatticeFermion    err(&FGrid);
+
+  ColourMatrix cm = Complex(1.0,0.0);
+
+  GridParallelRNG          RNG4(&UGrid);  RNG4.SeedFixedIntegers(seeds4);
+  LatticeGaugeField Umu(&UGrid); random(RNG4,Umu);
+  LatticeGaugeField Umu5d(&FGrid); 
+
+  // replicate across fifth dimension
+  for(int ss=0;ss<Umu._grid->oSites();ss++){
+    for(int s=0;s<Ls;s++){
+      Umu5d._odata[Ls*ss+s] = Umu._odata[ss];
+    }
+  }
+
+  ////////////////////////////////////
+  // Naive wilson implementation
+  ////////////////////////////////////
+  std::vector<LatticeColourMatrix> U(4,&FGrid);
+  for(int mu=0;mu<Nd;mu++){
+    U[mu] = peekIndex<LorentzIndex>(Umu5d,mu);
+  }
+
+  if (1)
+  {
+    ref = zero;
+    for(int mu=0;mu<Nd;mu++){
+
+      tmp = U[mu]*Cshift(src,mu+1,1);
+      ref=ref + tmp + Gamma(Gmu[mu])*tmp;
+
+      tmp =adj(U[mu])*src;
+      tmp =Cshift(tmp,mu+1,-1);
+      ref=ref + tmp - Gamma(Gmu[mu])*tmp;
+    }
+    ref = -0.5*ref;
+  }
+
+  RealD mass=0.1;
+  FiveDimWilsonFermion Dw(Umu,FGrid,FrbGrid,UGrid,UrbGrid,mass);
+  
+  std::cout << "Calling Dw"<<std::endl;
+  int ncall=1000;
+  double t0=usecond();
+  for(int i=0;i<ncall;i++){
+    Dw.Dhop(src,result,0);
+  }
+  double t1=usecond();
+
+
+  double volume=Ls;  for(int mu=0;mu<Nd;mu++) volume=volume*latt4[mu];
+  double flops=1344*volume*ncall;
+  
+  std::cout << "Called Dw"<<std::endl;
+  std::cout << "norm result "<< norm2(result)<<std::endl;
+  std::cout << "norm ref    "<< norm2(ref)<<std::endl;
+  std::cout << "mflop/s =   "<< flops/(t1-t0)<<std::endl;
+  err = ref-result; 
+  std::cout << "norm diff   "<< norm2(err)<<std::endl;
+
+
+  if (1)
+  { // Naive wilson dag implementation
+    ref = zero;
+    for(int mu=0;mu<Nd;mu++){
+
+      //    ref =  src - Gamma(Gamma::GammaX)* src ; // 1+gamma_x
+      tmp = U[mu]*Cshift(src,mu+1,1);
+      for(int i=0;i<ref._odata.size();i++){
+	ref._odata[i]+= tmp._odata[i] - Gamma(Gmu[mu])*tmp._odata[i]; ;
+      }
+
+      tmp =adj(U[mu])*src;
+      tmp =Cshift(tmp,mu+1,-1);
+      for(int i=0;i<ref._odata.size();i++){
+	ref._odata[i]+= tmp._odata[i] + Gamma(Gmu[mu])*tmp._odata[i]; ;
+      }
+    }
+    ref = -0.5*ref;
+  }
+  Dw.Dhop(src,result,1);
+  std::cout << "Called DwDag"<<std::endl;
+  std::cout << "norm result "<< norm2(result)<<std::endl;
+  std::cout << "norm ref    "<< norm2(ref)<<std::endl;
+  err = ref-result; 
+  std::cout << "norm diff   "<< norm2(err)<<std::endl;
+
+  Grid_finalize();
+}
diff --git a/benchmarks/Grid_wilson.cc b/benchmarks/Grid_wilson.cc
index 3b0d04bc..c1050098 100644
--- a/benchmarks/Grid_wilson.cc
+++ b/benchmarks/Grid_wilson.cc
@@ -49,6 +49,7 @@ int main (int argc, char ** argv)
   }  
 
   // Only one non-zero (y)
+#if 0
   Umu=zero;
   Complex cone(1.0,0.0);
   for(int nn=0;nn<Nd;nn++){
@@ -59,6 +60,7 @@ int main (int argc, char ** argv)
     }
     pokeIndex<LorentzIndex>(Umu,U[nn],nn);
   }
+#endif
 
   for(int mu=0;mu<Nd;mu++){
     U[mu] = peekIndex<LorentzIndex>(Umu,mu);
@@ -80,9 +82,9 @@ int main (int argc, char ** argv)
       }
     }
   }
-
+  ref = -0.5*ref;
   RealD mass=0.1;
-  WilsonMatrix Dw(Umu,Grid,RBGrid,mass);
+  WilsonFermion Dw(Umu,Grid,RBGrid,mass);
   
   std::cout << "Calling Dw"<<std::endl;
   int ncall=10000;
@@ -91,7 +93,7 @@ int main (int argc, char ** argv)
     Dw.Dhop(src,result,0);
   }
   double t1=usecond();
-  double flops=1320*volume*ncall;
+  double flops=1344*volume*ncall;
   
   std::cout << "Called Dw"<<std::endl;
   std::cout << "norm result "<< norm2(result)<<std::endl;
@@ -129,6 +131,7 @@ int main (int argc, char ** argv)
       }
     }
   }
+  ref = -0.5*ref;
   Dw.Dhop(src,result,1);
   std::cout << "Called DwDag"<<std::endl;
   std::cout << "norm result "<< norm2(result)<<std::endl;
diff --git a/benchmarks/Grid_wilson_cg_prec.cc b/benchmarks/Grid_wilson_cg_prec.cc
index e86ec820..e376349c 100644
--- a/benchmarks/Grid_wilson_cg_prec.cc
+++ b/benchmarks/Grid_wilson_cg_prec.cc
@@ -42,9 +42,9 @@ int main (int argc, char ** argv)
   }
   
   RealD mass=0.5;
-  WilsonMatrix Dw(Umu,Grid,RBGrid,mass);
+  WilsonFermion Dw(Umu,Grid,RBGrid,mass);
 
-  //  HermitianOperator<WilsonMatrix,LatticeFermion> HermOp(Dw);
+  //  HermitianOperator<WilsonFermion,LatticeFermion> HermOp(Dw);
   //  ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   //  CG(HermOp,src,result);
   
@@ -53,7 +53,7 @@ int main (int argc, char ** argv)
   pickCheckerboard(Odd,src_o,src);
   result_o=zero;
 
-  HermitianCheckerBoardedOperator<WilsonMatrix,LatticeFermion> HermOpEO(Dw);
+  HermitianCheckerBoardedOperator<WilsonFermion,LatticeFermion> HermOpEO(Dw);
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   CG(HermOpEO,src_o,result_o);
 
diff --git a/benchmarks/Grid_wilson_cg_schur.cc b/benchmarks/Grid_wilson_cg_schur.cc
index af630ae1..28db1d4b 100644
--- a/benchmarks/Grid_wilson_cg_schur.cc
+++ b/benchmarks/Grid_wilson_cg_schur.cc
@@ -37,7 +37,7 @@ int main (int argc, char ** argv)
   LatticeFermion resid(&Grid); 
 
   RealD mass=0.5;
-  WilsonMatrix Dw(Umu,Grid,RBGrid,mass);
+  WilsonFermion Dw(Umu,Grid,RBGrid,mass);
 
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   SchurRedBlackSolve<LatticeFermion> SchurSolver(CG);
diff --git a/benchmarks/Grid_wilson_cg_unprec.cc b/benchmarks/Grid_wilson_cg_unprec.cc
index 15302aab..905dfde5 100644
--- a/benchmarks/Grid_wilson_cg_unprec.cc
+++ b/benchmarks/Grid_wilson_cg_unprec.cc
@@ -47,9 +47,9 @@ int main (int argc, char ** argv)
   }
   
   RealD mass=0.5;
-  WilsonMatrix Dw(Umu,Grid,RBGrid,mass);
+  WilsonFermion Dw(Umu,Grid,RBGrid,mass);
 
-  HermitianOperator<WilsonMatrix,LatticeFermion> HermOp(Dw);
+  HermitianOperator<WilsonFermion,LatticeFermion> HermOp(Dw);
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   CG(HermOp,src,result);
 
diff --git a/benchmarks/Grid_wilson_evenodd.cc b/benchmarks/Grid_wilson_evenodd.cc
index 4bc8c357..a073139d 100644
--- a/benchmarks/Grid_wilson_evenodd.cc
+++ b/benchmarks/Grid_wilson_evenodd.cc
@@ -60,7 +60,7 @@ int main (int argc, char ** argv)
 
   RealD mass=0.1;
 
-  WilsonMatrix Dw(Umu,Grid,RBGrid,mass);
+  WilsonFermion Dw(Umu,Grid,RBGrid,mass);
 
   LatticeFermion src_e   (&RBGrid);
   LatticeFermion src_o   (&RBGrid);
diff --git a/benchmarks/Makefile.am b/benchmarks/Makefile.am
index 95ae5eca..b2649669 100644
--- a/benchmarks/Makefile.am
+++ b/benchmarks/Makefile.am
@@ -5,11 +5,14 @@ AM_LDFLAGS = -L$(top_builddir)/lib
 #
 # Test code
 #
-bin_PROGRAMS = Grid_wilson Grid_comms Grid_memory_bandwidth Grid_su3 Grid_wilson_cg_unprec Grid_wilson_evenodd  Grid_wilson_cg_prec Grid_wilson_cg_schur
+bin_PROGRAMS = Grid_wilson Grid_comms Grid_memory_bandwidth Grid_su3 Grid_wilson_cg_unprec Grid_wilson_evenodd  Grid_wilson_cg_prec Grid_wilson_cg_schur Grid_dwf
 
 Grid_wilson_SOURCES = Grid_wilson.cc
 Grid_wilson_LDADD = -lGrid
 
+Grid_dwf_SOURCES = Grid_dwf.cc
+Grid_dwf_LDADD = -lGrid
+
 Grid_wilson_evenodd_SOURCES = Grid_wilson_evenodd.cc
 Grid_wilson_evenodd_LDADD = -lGrid
 
diff --git a/lib/Grid.h b/lib/Grid.h
index 1e513a46..7fa56892 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -60,7 +60,7 @@
 
 #include <Grid_algorithms.h>// subdir aggregate
 
-#include <qcd/Grid_qcd.h>
+#include <qcd/QCD.h>
 #include <parallelIO/GridNerscIO.h>
 
 namespace Grid {
diff --git a/lib/Grid_init.cc b/lib/Grid_init.cc
index f72393cb..580e602a 100644
--- a/lib/Grid_init.cc
+++ b/lib/Grid_init.cc
@@ -142,7 +142,11 @@ void Grid_init(int *argc,char ***argv)
     Grid_quiesce_nodes();
   }
   if( GridCmdOptionExists(*argv,*argv+*argc,"--dslash-opt") ){
-    WilsonMatrix::HandOptDslash=1;
+    WilsonFermion::HandOptDslash=1;
+    FiveDimWilsonFermion::HandOptDslash=1;
+  }
+  if( GridCmdOptionExists(*argv,*argv+*argc,"--lebesgue") ){
+    LebesgueOrder::UseLebesgueOrder=1;
   }
   GridParseLayout(*argv,*argc,
 		  Grid_default_latt,
diff --git a/lib/Grid_stencil.h b/lib/Grid_stencil.h
index fa22361b..50d22453 100644
--- a/lib/Grid_stencil.h
+++ b/lib/Grid_stencil.h
@@ -1,6 +1,8 @@
 #ifndef GRID_STENCIL_H
 #define GRID_STENCIL_H
 
+#include <stencil/Grid_lebesgue.h>   // subdir aggregate
+
 //////////////////////////////////////////////////////////////////////////////////////////
 // Must not lose sight that goal is to be able to construct really efficient
 // gather to a point stencil code. CSHIFT is not the best way, so need
@@ -38,29 +40,12 @@
 //////////////////////////////////////////////////////////////////////////////////////////
 
 namespace Grid {
-
-
+  
 
   class CartesianStencil { // Stencil runs along coordinate axes only; NO diagonal fill in.
   public:
 
       typedef uint32_t StencilInteger;
-      
-
-
-      StencilInteger alignup(StencilInteger n){
-	n--;           // 1000 0011 --> 1000 0010
-	n |= n >> 1;   // 1000 0010 | 0100 0001 = 1100 0011
-	n |= n >> 2;   // 1100 0011 | 0011 0000 = 1111 0011
-	n |= n >> 4;   // 1111 0011 | 0000 1111 = 1111 1111
-	n |= n >> 8;   // ... (At this point all bits are 1, so further bitwise-or
-	n |= n >> 16;  //      operations produce no effect.)
-	n++;           // 1111 1111 --> 1 0000 0000
-	return n;
-      };
-      void LebesgueOrder(void);
-
-      std::vector<StencilInteger> _LebesgueReorder;
 
       int                               _checkerboard;
       int                               _npoints; // Move to template param?
@@ -131,8 +116,8 @@ namespace Grid {
 	  // Gather phase
 	  int sshift [2];
 	  if ( comm_dim ) {
-	    sshift[0] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,0);
-	    sshift[1] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,1);
+	    sshift[0] = _grid->CheckerBoardShiftForCB(_checkerboard,dimension,shift,Even);
+	    sshift[1] = _grid->CheckerBoardShiftForCB(_checkerboard,dimension,shift,Odd);
 	    if ( sshift[0] == sshift[1] ) {
 	      if (splice_dim) {
 		GatherStartCommsSimd(source,dimension,shift,0x3,u_comm_buf,u_comm_offset,compress);
@@ -179,8 +164,8 @@ namespace Grid {
 	  std::vector<cobj,alignedAllocator<cobj> > send_buf(buffer_size); // hmm...
 	  std::vector<cobj,alignedAllocator<cobj> > recv_buf(buffer_size);
 	  
-	  int cb= (cbmask==0x2)? 1 : 0;
-	  int sshift= _grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,cb);
+	  int cb= (cbmask==0x2)? Odd : Even;
+	  int sshift= _grid->CheckerBoardShiftForCB(rhs.checkerboard,dimension,shift,cb);
 	  
 	  for(int x=0;x<rd;x++){       
 	    
@@ -266,8 +251,8 @@ namespace Grid {
 	  // Work out what to send where
 	  ///////////////////////////////////////////
 
-	  int cb    = (cbmask==0x2)? 1 : 0;
-	  int sshift= _grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,cb);
+	  int cb    = (cbmask==0x2)? Odd : Even;
+	  int sshift= _grid->CheckerBoardShiftForCB(rhs.checkerboard,dimension,shift,cb);
 	  
 	  // loop over outer coord planes orthog to dim
 	  for(int x=0;x<rd;x++){       
diff --git a/lib/Grid_threads.h b/lib/Grid_threads.h
index 24581855..f453b860 100644
--- a/lib/Grid_threads.h
+++ b/lib/Grid_threads.h
@@ -9,7 +9,7 @@
 
 #ifdef GRID_OMP
 #include <omp.h>
-#define PARALLEL_FOR_LOOP _Pragma("omp parallel for")
+#define PARALLEL_FOR_LOOP _Pragma("omp parallel for ")
 #define PARALLEL_NESTED_LOOP2 _Pragma("omp parallel for collapse(2)")
 #else
 #define PARALLEL_FOR_LOOP 
diff --git a/lib/Makefile.am b/lib/Makefile.am
index a5b89c0a..aa531df7 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -16,87 +16,103 @@ endif
 lib_LIBRARIES = libGrid.a
 libGrid_a_SOURCES =				\
 	Grid_init.cc				\
+	stencil/Grid_lebesgue.cc		\
 	stencil/Grid_stencil_common.cc		\
-	qcd/Grid_qcd_dirac.cc			\
-	qcd/Grid_qcd_dhop.cc			\
-	qcd/Grid_qcd_dhop_hand.cc		\
-	qcd/Grid_qcd_wilson_dop.cc		\
 	algorithms/approx/Zolotarev.cc		\
 	algorithms/approx/Remez.cc		\
+	qcd/action/fermion/FiveDimWilsonFermion.cc\
+	qcd/action/fermion/WilsonFermion.cc\
+	qcd/action/fermion/WilsonKernels.cc\
+	qcd/action/fermion/WilsonKernelsHand.cc\
+	qcd/Dirac.cc\
 	$(extra_sources)
 
 #
 # Include files
 #
-nobase_include_HEADERS = algorithms/approx/bigfloat.h		\
-	algorithms/approx/Chebyshev.h				\
-	algorithms/approx/Remez.h				\
-	algorithms/approx/Zolotarev.h				\
-	algorithms/iterative/ConjugateGradient.h		\
-	algorithms/iterative/NormalEquations.h			\
-	algorithms/iterative/SchurRedBlack.h			\
-	algorithms/LinearOperator.h				\
-	algorithms/SparseMatrix.h				\
-	cartesian/Grid_cartesian_base.h				\
-	cartesian/Grid_cartesian_full.h				\
-	cartesian/Grid_cartesian_red_black.h			\
-	communicator/Grid_communicator_base.h			\
-	cshift/Grid_cshift_common.h				\
-	cshift/Grid_cshift_mpi.h				\
-	cshift/Grid_cshift_none.h				\
-	Grid.h							\
-	Grid_algorithms.h					\
-	Grid_aligned_allocator.h				\
-	Grid_cartesian.h					\
-	Grid_communicator.h					\
-	Grid_comparison.h					\
-	Grid_cshift.h						\
-	Grid_extract.h						\
-	Grid_lattice.h						\
-	Grid_math.h						\
-	Grid_simd.h						\
-	Grid_stencil.h						\
-	Grid_threads.h						\
-	lattice/Grid_lattice_arith.h				\
-	lattice/Grid_lattice_base.h				\
-	lattice/Grid_lattice_comparison.h			\
-	lattice/Grid_lattice_conformable.h			\
-	lattice/Grid_lattice_coordinate.h			\
-	lattice/Grid_lattice_ET.h				\
-	lattice/Grid_lattice_local.h				\
-	lattice/Grid_lattice_overload.h				\
-	lattice/Grid_lattice_peekpoke.h				\
-	lattice/Grid_lattice_reality.h				\
-	lattice/Grid_lattice_reduction.h			\
-	lattice/Grid_lattice_rng.h				\
-	lattice/Grid_lattice_trace.h				\
-	lattice/Grid_lattice_transfer.h				\
-	lattice/Grid_lattice_transpose.h			\
-	lattice/Grid_lattice_where.h				\
-	math/Grid_math_arith.h					\
-	math/Grid_math_arith_add.h				\
-	math/Grid_math_arith_mac.h				\
-	math/Grid_math_arith_mul.h				\
-	math/Grid_math_arith_scalar.h				\
-	math/Grid_math_arith_sub.h				\
-	math/Grid_math_inner.h					\
-	math/Grid_math_outer.h					\
-	math/Grid_math_peek.h					\
-	math/Grid_math_poke.h					\
-	math/Grid_math_reality.h				\
-	math/Grid_math_tensors.h				\
-	math/Grid_math_trace.h					\
-	math/Grid_math_traits.h					\
-	math/Grid_math_transpose.h				\
-	parallelIO/GridNerscIO.h				\
-	qcd/Grid_qcd.h						\
-	qcd/Grid_qcd_2spinor.h					\
-	qcd/Grid_qcd_dirac.h					\
-	qcd/Grid_qcd_wilson_dop.h				\
-	simd/Grid_vector_types.h				\
-	simd/Grid_sse4.h					\
-	simd/Grid_avx.h						\
-	simd/Grid_avx512.h					
-
-
+nobase_include_HEADERS=\
+		./algorithms/approx/bigfloat.h\
+		./algorithms/approx/bigfloat_double.h\
+		./algorithms/approx/Chebyshev.h\
+		./algorithms/approx/Remez.h\
+		./algorithms/approx/Zolotarev.h\
+		./algorithms/iterative/ConjugateGradient.h\
+		./algorithms/iterative/NormalEquations.h\
+		./algorithms/iterative/SchurRedBlack.h\
+		./algorithms/LinearOperator.h\
+		./algorithms/SparseMatrix.h\
+		./cartesian/Grid_cartesian_base.h\
+		./cartesian/Grid_cartesian_full.h\
+		./cartesian/Grid_cartesian_red_black.h\
+		./communicator/Grid_communicator_base.h\
+		./cshift/Grid_cshift_common.h\
+		./cshift/Grid_cshift_mpi.h\
+		./cshift/Grid_cshift_none.h\
+		./Grid.h\
+		./Grid_algorithms.h\
+		./Grid_aligned_allocator.h\
+		./Grid_cartesian.h\
+		./Grid_communicator.h\
+		./Grid_comparison.h\
+		./Grid_config.h\
+		./Grid_cshift.h\
+		./Grid_extract.h\
+		./Grid_lattice.h\
+		./Grid_math.h\
+		./Grid_simd.h\
+		./Grid_stencil.h\
+		./Grid_threads.h\
+		./lattice/Grid_lattice_arith.h\
+		./lattice/Grid_lattice_base.h\
+		./lattice/Grid_lattice_comparison.h\
+		./lattice/Grid_lattice_conformable.h\
+		./lattice/Grid_lattice_coordinate.h\
+		./lattice/Grid_lattice_ET.h\
+		./lattice/Grid_lattice_local.h\
+		./lattice/Grid_lattice_overload.h\
+		./lattice/Grid_lattice_peekpoke.h\
+		./lattice/Grid_lattice_reality.h\
+		./lattice/Grid_lattice_reduction.h\
+		./lattice/Grid_lattice_rng.h\
+		./lattice/Grid_lattice_trace.h\
+		./lattice/Grid_lattice_transfer.h\
+		./lattice/Grid_lattice_transpose.h\
+		./lattice/Grid_lattice_where.h\
+		./math/Grid_math_arith.h\
+		./math/Grid_math_arith_add.h\
+		./math/Grid_math_arith_mac.h\
+		./math/Grid_math_arith_mul.h\
+		./math/Grid_math_arith_scalar.h\
+		./math/Grid_math_arith_sub.h\
+		./math/Grid_math_inner.h\
+		./math/Grid_math_outer.h\
+		./math/Grid_math_peek.h\
+		./math/Grid_math_poke.h\
+		./math/Grid_math_reality.h\
+		./math/Grid_math_tensors.h\
+		./math/Grid_math_trace.h\
+		./math/Grid_math_traits.h\
+		./math/Grid_math_transpose.h\
+		./parallelIO/GridNerscIO.h\
+		./qcd/action/Actions.h\
+		./qcd/action/fermion/FermionAction.h\
+		./qcd/action/fermion/FiveDimWilsonFermion.h\
+		./qcd/action/fermion/WilsonCompressor.h\
+		./qcd/action/fermion/WilsonFermion.h\
+		./qcd/action/fermion/WilsonKernels.h\
+		./qcd/Dirac.h\
+		./qcd/QCD.h\
+		./qcd/TwoSpinor.h\
+		./qcd/FermionAction.h\
+		./simd/Grid_avx.h\
+		./simd/Grid_avx512.h\
+		./simd/Grid_qpx.h\
+		./simd/Grid_sse4.h\
+		./simd/Grid_vector_types.h\
+		./simd/Old/Grid_vComplexD.h\
+		./simd/Old/Grid_vComplexF.h\
+		./simd/Old/Grid_vInteger.h\
+		./simd/Old/Grid_vRealD.h\
+		./simd/Old/Grid_vRealF.h\
+		./stencil/Grid_lebesgue.h
 
diff --git a/lib/algorithms/SparseMatrix.h b/lib/algorithms/SparseMatrix.h
index 7bfe959b..9c955e9a 100644
--- a/lib/algorithms/SparseMatrix.h
+++ b/lib/algorithms/SparseMatrix.h
@@ -10,7 +10,7 @@ namespace Grid {
   /////////////////////////////////////////////////////////////////////////////////////////////
     template<class Field> class SparseMatrixBase {
     public:
-      GridBase *_grid;
+      virtual GridBase *Grid(void) =0;
       // Full checkerboar operations
       virtual RealD M    (const Field &in, Field &out)=0;
       virtual RealD Mdag (const Field &in, Field &out)=0;
@@ -19,7 +19,6 @@ namespace Grid {
 	ni=M(in,tmp);
 	no=Mdag(tmp,out);
       }
-      SparseMatrixBase(GridBase *grid) : _grid(grid) {};
     };
 
   /////////////////////////////////////////////////////////////////////////////////////////////
@@ -27,7 +26,7 @@ namespace Grid {
   /////////////////////////////////////////////////////////////////////////////////////////////
     template<class Field> class CheckerBoardedSparseMatrixBase : public SparseMatrixBase<Field> {
     public:
-      GridBase *_cbgrid;
+      virtual GridBase *RedBlackGrid(void)=0;
       // half checkerboard operaions
       virtual  void Meooe    (const Field &in, Field &out)=0;
       virtual  void Mooee    (const Field &in, Field &out)=0;
@@ -62,9 +61,7 @@ namespace Grid {
 	Field tmp(in._grid);
 	ni=Mpc(in,tmp);
 	no=MpcDag(tmp,out);
-	//	std::cout<<"MpcDagMpc "<<ni<<" "<<no<<std::endl;
       }
-      CheckerBoardedSparseMatrixBase(GridBase *grid,GridBase *cbgrid) : SparseMatrixBase<Field>(grid), _cbgrid(cbgrid) {};
     };
 
 }
diff --git a/lib/algorithms/iterative/SchurRedBlack.h b/lib/algorithms/iterative/SchurRedBlack.h
index d3a76d7f..109f554b 100644
--- a/lib/algorithms/iterative/SchurRedBlack.h
+++ b/lib/algorithms/iterative/SchurRedBlack.h
@@ -60,8 +60,8 @@ namespace Grid {
 
       // FIXME CGdiagonalMee not implemented virtual function
       // FIXME use CBfactorise to control schur decomp
-      GridBase *grid = _Matrix._cbgrid;
-      GridBase *fgrid= _Matrix._grid;
+      GridBase *grid = _Matrix.RedBlackGrid();
+      GridBase *fgrid= _Matrix.Grid();
  
       Field src_e(grid);
       Field src_o(grid);
diff --git a/lib/cartesian/Grid_cartesian_base.h b/lib/cartesian/Grid_cartesian_base.h
index a74773f8..e93125c1 100644
--- a/lib/cartesian/Grid_cartesian_base.h
+++ b/lib/cartesian/Grid_cartesian_base.h
@@ -52,16 +52,13 @@ public:
     ////////////////////////////////////////////////////////////////
     virtual int CheckerBoarded(int dim)=0;
     virtual int CheckerBoard(std::vector<int> site)=0;
-    virtual int CheckerBoardDestination(int source_cb,int shift)=0;
+    virtual int CheckerBoardDestination(int source_cb,int shift,int dim)=0;
     virtual int CheckerBoardShift(int source_cb,int dim,int shift,int osite)=0;
-    inline int  CheckerBoardFromOindex (int Oindex){
+    virtual int CheckerBoardShiftForCB(int source_cb,int dim,int shift,int cb)=0;
+    int  CheckerBoardFromOindex (int Oindex){
       std::vector<int> ocoor;
       oCoorFromOindex(ocoor,Oindex); 
-      int ss=0;
-      for(int d=0;d<_ndimension;d++){
-	ss=ss+ocoor[d];
-      }      
-      return ss&0x1;
+      return CheckerBoard(ocoor);
     }
 
     //////////////////////////////////////////////////////////////////////////////////////////////
diff --git a/lib/cartesian/Grid_cartesian_full.h b/lib/cartesian/Grid_cartesian_full.h
index 73bd08b3..330bbfaf 100644
--- a/lib/cartesian/Grid_cartesian_full.h
+++ b/lib/cartesian/Grid_cartesian_full.h
@@ -18,11 +18,14 @@ public:
     virtual int CheckerBoard(std::vector<int> site){
         return 0;
     }
-    virtual int CheckerBoardDestination(int cb,int shift){
+    virtual int CheckerBoardDestination(int cb,int shift,int dim){
         return 0;
     }
+    virtual int CheckerBoardShiftForCB(int source_cb,int dim,int shift, int ocb){
+      return shift;
+    }
     virtual int CheckerBoardShift(int source_cb,int dim,int shift, int osite){
-        return shift;
+      return shift;
     }
     GridCartesian(std::vector<int> &dimensions,
 		  std::vector<int> &simd_layout,
diff --git a/lib/cartesian/Grid_cartesian_red_black.h b/lib/cartesian/Grid_cartesian_red_black.h
index ff2d3ba8..ace36edb 100644
--- a/lib/cartesian/Grid_cartesian_red_black.h
+++ b/lib/cartesian/Grid_cartesian_red_black.h
@@ -8,6 +8,10 @@ namespace Grid {
     static const int CbBlack=1;
     static const int Even   =CbRed;
     static const int Odd    =CbBlack;
+
+    // Perhaps these are misplaced and 
+    // should be in sparse matrix.
+    // Also should make these a named enum type
     static const int DaggerNo=0;
     static const int DaggerYes=1;
 
@@ -15,116 +19,174 @@ namespace Grid {
 class GridRedBlackCartesian : public GridBase
 {
 public:
+    std::vector<int> _checker_dim_mask;
+    int              _checker_dim;
+
     virtual int CheckerBoarded(int dim){
-      if( dim==0) return 1;
+      if( dim==_checker_dim) return 1;
       else return 0;
     }
     virtual int CheckerBoard(std::vector<int> site){
-      return (site[0]+site[1]+site[2]+site[3])&0x1;
+      int linear=0;
+      assert(site.size()==_ndimension);
+      for(int d=0;d<_ndimension;d++){ 
+	if(_checker_dim_mask[d])
+	  linear=linear+site[d];
+      }
+      return (linear&0x1);
     }
 
+
     // Depending on the cb of site, we toggle source cb.
     // for block #b, element #e = (b, e)
     // we need 
-    virtual int CheckerBoardShift(int source_cb,int dim,int shift,int osite){
+    virtual int CheckerBoardShiftForCB(int source_cb,int dim,int shift,int ocb){
+      if(dim != _checker_dim) return shift;
 
-      if(dim != 0) return shift;
-
-      int fulldim =_fdimensions[0];
+      int fulldim =_fdimensions[dim];
       shift = (shift+fulldim)%fulldim;
 
       // Probably faster with table lookup;
       // or by looping over x,y,z and multiply rather than computing checkerboard.
-      int ocb=CheckerBoardFromOindex(osite);
 	  
       if ( (source_cb+ocb)&1 ) {
+
 	return (shift)/2;
       } else {
 	return (shift+1)/2;
       }
     }
+    virtual int CheckerBoardShift(int source_cb,int dim,int shift,int osite){
 
-    virtual int CheckerBoardDestination(int source_cb,int shift){
-        if ((shift+_fdimensions[0])&0x1) {
+      if(dim != _checker_dim) return shift;
+
+      int ocb=CheckerBoardFromOindex(osite);
+      
+      return CheckerBoardShiftForCB(source_cb,dim,shift,ocb);
+    }
+    
+    virtual int CheckerBoardDestination(int source_cb,int shift,int dim){
+      if ( _checker_dim_mask[dim]  ) {
+	// If _fdimensions[checker_dim] is odd, then shifting by 1 in other dims
+	// does NOT cause a parity hop.
+	int add=(dim==_checker_dim) ? 0 : _fdimensions[_checker_dim];
+        if ( (shift+add) &0x1) {
             return 1-source_cb;
         } else {
             return source_cb;
         }
+      } else {
+	return source_cb;
+
+      }
     };
 
     GridRedBlackCartesian(GridBase *base) : GridRedBlackCartesian(base->_fdimensions,base->_simd_layout,base->_processors)  {};
 
     GridRedBlackCartesian(std::vector<int> &dimensions,
 			  std::vector<int> &simd_layout,
-			  std::vector<int> &processor_grid ) : GridBase(processor_grid)
+			  std::vector<int> &processor_grid,
+			  std::vector<int> &checker_dim_mask,
+			  int checker_dim
+			  ) :  GridBase(processor_grid) 
+    {
+      Init(dimensions,simd_layout,processor_grid,checker_dim_mask,checker_dim);
+    }
+    GridRedBlackCartesian(std::vector<int> &dimensions,
+			  std::vector<int> &simd_layout,
+			  std::vector<int> &processor_grid) : GridBase(processor_grid) 
+    {
+      std::vector<int> checker_dim_mask(dimensions.size(),1);
+      Init(dimensions,simd_layout,processor_grid,checker_dim_mask,0);
+    }
+    void Init(std::vector<int> &dimensions,
+	      std::vector<int> &simd_layout,
+	      std::vector<int> &processor_grid,
+	      std::vector<int> &checker_dim_mask,
+	      int checker_dim)
     {
     ///////////////////////
     // Grid information
     ///////////////////////
-        _ndimension = dimensions.size();
-        
-        _fdimensions.resize(_ndimension);
-        _gdimensions.resize(_ndimension);
-        _ldimensions.resize(_ndimension);
-        _rdimensions.resize(_ndimension);
-        _simd_layout.resize(_ndimension);
-        
-        _ostride.resize(_ndimension);
-        _istride.resize(_ndimension);
-        
-        _fsites = _gsites = _osites = _isites = 1;
+      _checker_dim = checker_dim;
+      assert(checker_dim_mask[checker_dim]==1);
+      _ndimension = dimensions.size();
+      assert(checker_dim_mask.size()==_ndimension);
+      assert(processor_grid.size()==_ndimension);
+      assert(simd_layout.size()==_ndimension);
+      
+      _fdimensions.resize(_ndimension);
+      _gdimensions.resize(_ndimension);
+      _ldimensions.resize(_ndimension);
+      _rdimensions.resize(_ndimension);
+      _simd_layout.resize(_ndimension);
+      
+      _ostride.resize(_ndimension);
+      _istride.resize(_ndimension);
+      
+      _fsites = _gsites = _osites = _isites = 1;
+	
+      _checker_dim_mask=checker_dim_mask;
 
-        for(int d=0;d<_ndimension;d++){
-            _fdimensions[d] = dimensions[d];
-            _gdimensions[d] = _fdimensions[d];
-	    _fsites = _fsites * _fdimensions[d];
-	    _gsites = _gsites * _gdimensions[d];
-                
-            if (d==0) _gdimensions[0] = _gdimensions[0]/2; // Remove a checkerboard
-            _ldimensions[d] = _gdimensions[d]/_processors[d];
-
-            // Use a reduced simd grid
-            _simd_layout[d] = simd_layout[d];
-            _rdimensions[d]= _ldimensions[d]/_simd_layout[d];
-
-            _osites *= _rdimensions[d];
-            _isites *= _simd_layout[d];
-                
-            // Addressing support
-            if ( d==0 ) {
-                _ostride[d] = 1;
-                _istride[d] = 1;
-            } else {
-                _ostride[d] = _ostride[d-1]*_rdimensions[d-1];
-                _istride[d] = _istride[d-1]*_simd_layout[d-1];
-            }
-        }
-            
-        ////////////////////////////////////////////////////////////////////////////////////////////
-        // subplane information
-        ////////////////////////////////////////////////////////////////////////////////////////////
-        _slice_block.resize(_ndimension);
-        _slice_stride.resize(_ndimension);
-        _slice_nblock.resize(_ndimension);
+      for(int d=0;d<_ndimension;d++){
+	_fdimensions[d] = dimensions[d];
+	_gdimensions[d] = _fdimensions[d];
+	_fsites = _fsites * _fdimensions[d];
+	_gsites = _gsites * _gdimensions[d];
         
-        int block =1;
-        int nblock=1;
-        for(int d=0;d<_ndimension;d++) nblock*=_rdimensions[d];
-            
-        for(int d=0;d<_ndimension;d++){
-            nblock/=_rdimensions[d];
-            _slice_block[d] =block;
-            _slice_stride[d]=_ostride[d]*_rdimensions[d];
-            _slice_nblock[d]=nblock;
-            block = block*_rdimensions[d];
-        }
+	if (d==_checker_dim) {
+	  _gdimensions[d] = _gdimensions[d]/2; // Remove a checkerboard
+	}
+	_ldimensions[d] = _gdimensions[d]/_processors[d];
+
+	// Use a reduced simd grid
+	_simd_layout[d] = simd_layout[d];
+	_rdimensions[d]= _ldimensions[d]/_simd_layout[d];
+	
+	_osites *= _rdimensions[d];
+	_isites *= _simd_layout[d];
+        
+	// Addressing support
+	if ( d==0 ) {
+	  _ostride[d] = 1;
+	  _istride[d] = 1;
+	} else {
+	  _ostride[d] = _ostride[d-1]*_rdimensions[d-1];
+	  _istride[d] = _istride[d-1]*_simd_layout[d-1];
+	}
+      }
             
+      ////////////////////////////////////////////////////////////////////////////////////////////
+      // subplane information
+      ////////////////////////////////////////////////////////////////////////////////////////////
+      _slice_block.resize(_ndimension);
+      _slice_stride.resize(_ndimension);
+      _slice_nblock.resize(_ndimension);
+        
+      int block =1;
+      int nblock=1;
+      for(int d=0;d<_ndimension;d++) nblock*=_rdimensions[d];
+      
+      for(int d=0;d<_ndimension;d++){
+	nblock/=_rdimensions[d];
+	_slice_block[d] =block;
+	_slice_stride[d]=_ostride[d]*_rdimensions[d];
+	_slice_nblock[d]=nblock;
+	block = block*_rdimensions[d];
+      }
+      
     };
 protected:
     virtual int oIndex(std::vector<int> &coor)
     {
-        int idx=_ostride[0]*((coor[0]/2)%_rdimensions[0]);
-        for(int d=1;d<_ndimension;d++) idx+=_ostride[d]*(coor[d]%_rdimensions[d]);
+      int idx=0;
+      for(int d=0;d<_ndimension;d++) {
+	if( d==_checker_dim ) {
+	  idx+=_ostride[d]*((coor[d]/2)%_rdimensions[d]);
+	} else {
+	  idx+=_ostride[d]*(coor[d]%_rdimensions[d]);
+	}
+      }
         return idx;
     };
         
diff --git a/lib/cshift/Grid_cshift_common.h b/lib/cshift/Grid_cshift_common.h
index c369fe1c..06e812d9 100644
--- a/lib/cshift/Grid_cshift_common.h
+++ b/lib/cshift/Grid_cshift_common.h
@@ -175,7 +175,6 @@ PARALLEL_NESTED_LOOP2
       if ( ocb&cbmask ) {
 	//lhs._odata[lo+o]=rhs._odata[ro+o];
 	vstream(lhs._odata[lo+o],rhs._odata[ro+o]);
-	
       }
 
     }
@@ -217,8 +216,8 @@ template<class vobj> void Cshift_local(Lattice<vobj>& ret,Lattice<vobj> &rhs,int
 {
   int sshift[2];
 
-  sshift[0] = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,0);
-  sshift[1] = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,1);
+  sshift[0] = rhs._grid->CheckerBoardShiftForCB(rhs.checkerboard,dimension,shift,Even);
+  sshift[1] = rhs._grid->CheckerBoardShiftForCB(rhs.checkerboard,dimension,shift,Odd);
 
   if ( sshift[0] == sshift[1] ) {
     Cshift_local(ret,rhs,dimension,shift,0x3);
@@ -239,8 +238,7 @@ template<class vobj> Lattice<vobj> Cshift_local(Lattice<vobj> &ret,Lattice<vobj>
   // Map to always positive shift modulo global full dimension.
   shift = (shift+fd)%fd;
 
-  ret.checkerboard = grid->CheckerBoardDestination(rhs.checkerboard,shift);
-        
+  ret.checkerboard = grid->CheckerBoardDestination(rhs.checkerboard,shift,dimension);
   // the permute type
   int permute_dim =grid->PermuteDim(dimension);
   int permute_type=grid->PermuteType(dimension);
@@ -250,11 +248,11 @@ template<class vobj> Lattice<vobj> Cshift_local(Lattice<vobj> &ret,Lattice<vobj>
     int o   = 0;
     int bo  = x * grid->_ostride[dimension];
     
-    int cb= (cbmask==0x2)? 1 : 0;
+    int cb= (cbmask==0x2)? Odd : Even;
 
-    int sshift = grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,cb);
+    int sshift = grid->CheckerBoardShiftForCB(rhs.checkerboard,dimension,shift,cb);
     int sx     = (x+sshift)%rd;
-	
+
     int permute_slice=0;
     if(permute_dim){
       int wrap = sshift/rd;
diff --git a/lib/cshift/Grid_cshift_mpi.h b/lib/cshift/Grid_cshift_mpi.h
index 569017d8..8c0badcd 100644
--- a/lib/cshift/Grid_cshift_mpi.h
+++ b/lib/cshift/Grid_cshift_mpi.h
@@ -39,8 +39,8 @@ template<class vobj> void Cshift_comms(Lattice<vobj>& ret,Lattice<vobj> &rhs,int
 {
   int sshift[2];
 
-  sshift[0] = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,0);
-  sshift[1] = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,1);
+  sshift[0] = rhs._grid->CheckerBoardShiftForCB(rhs.checkerboard,dimension,shift,Even);
+  sshift[1] = rhs._grid->CheckerBoardShiftForCB(rhs.checkerboard,dimension,shift,Odd);
 
   if ( sshift[0] == sshift[1] ) {
     Cshift_comms(ret,rhs,dimension,shift,0x3);
@@ -54,8 +54,8 @@ template<class vobj> void Cshift_comms_simd(Lattice<vobj>& ret,Lattice<vobj> &rh
 {
   int sshift[2];
 
-  sshift[0] = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,0);
-  sshift[1] = rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,1);
+  sshift[0] = rhs._grid->CheckerBoardShiftForCB(rhs.checkerboard,dimension,shift,Even);
+  sshift[1] = rhs._grid->CheckerBoardShiftForCB(rhs.checkerboard,dimension,shift,Odd);
 
   if ( sshift[0] == sshift[1] ) {
     Cshift_comms_simd(ret,rhs,dimension,shift,0x3);
@@ -87,8 +87,8 @@ template<class vobj> void Cshift_comms(Lattice<vobj> &ret,Lattice<vobj> &rhs,int
   std::vector<vobj,alignedAllocator<vobj> > send_buf(buffer_size);
   std::vector<vobj,alignedAllocator<vobj> > recv_buf(buffer_size);
 
-  int cb= (cbmask==0x2)? 1 : 0;
-  int sshift= rhs._grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,cb);
+  int cb= (cbmask==0x2)? Odd : Even;
+  int sshift= rhs._grid->CheckerBoardShiftForCB(rhs.checkerboard,dimension,shift,cb);
 
   for(int x=0;x<rd;x++){       
 
@@ -162,8 +162,8 @@ template<class vobj> void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &r
   ///////////////////////////////////////////
   // Work out what to send where
   ///////////////////////////////////////////
-  int cb    = (cbmask==0x2)? 1 : 0;
-  int sshift= grid->CheckerBoardShift(rhs.checkerboard,dimension,shift,cb);
+  int cb    = (cbmask==0x2)? Odd : Even;
+  int sshift= grid->CheckerBoardShiftForCB(rhs.checkerboard,dimension,shift,cb);
 
   // loop over outer coord planes orthog to dim
   for(int x=0;x<rd;x++){       
diff --git a/lib/cshift/Grid_cshift_none.h b/lib/cshift/Grid_cshift_none.h
index 3485a63e..be9467b1 100644
--- a/lib/cshift/Grid_cshift_none.h
+++ b/lib/cshift/Grid_cshift_none.h
@@ -4,7 +4,7 @@ namespace Grid {
 template<class vobj> Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
 {
   Lattice<vobj> ret(rhs._grid);
-  ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
+  ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift,dimension);
   Cshift_local(ret,rhs,dimension,shift);
   return ret;
 }
diff --git a/lib/lattice/Grid_lattice_peekpoke.h b/lib/lattice/Grid_lattice_peekpoke.h
index 7bffdbb8..9416226a 100644
--- a/lib/lattice/Grid_lattice_peekpoke.h
+++ b/lib/lattice/Grid_lattice_peekpoke.h
@@ -116,7 +116,7 @@ PARALLEL_FOR_LOOP
 
       int Nsimd = grid->Nsimd();
 
-      assert( l.checkerboard== l._grid->CheckerBoard(site));
+      assert( l.checkerboard == l._grid->CheckerBoard(site));
       assert( sizeof(sobj)*Nsimd == sizeof(vobj));
 
       int rank,odx,idx;
diff --git a/lib/lattice/Grid_lattice_transfer.h b/lib/lattice/Grid_lattice_transfer.h
index 27ae27e1..f0046b0e 100644
--- a/lib/lattice/Grid_lattice_transfer.h
+++ b/lib/lattice/Grid_lattice_transfer.h
@@ -32,7 +32,6 @@ PARALLEL_FOR_LOOP
       cbos=half._grid->CheckerBoard(coor);
       
       if (cbos==cb) {
-	
 	half._odata[ssh] = full._odata[ss];
 	ssh++;
       }
@@ -45,7 +44,7 @@ PARALLEL_FOR_LOOP
     for(int ss=0;ss<full._grid->oSites();ss++){
       std::vector<int> coor;
       int cbos;
-      
+
       full._grid->oCoorFromOindex(coor,ss);
       cbos=half._grid->CheckerBoard(coor);
       
diff --git a/lib/qcd/Grid_qcd_dirac.cc b/lib/qcd/Dirac.cc
similarity index 100%
rename from lib/qcd/Grid_qcd_dirac.cc
rename to lib/qcd/Dirac.cc
diff --git a/lib/qcd/Grid_qcd_dirac.h b/lib/qcd/Dirac.h
similarity index 100%
rename from lib/qcd/Grid_qcd_dirac.h
rename to lib/qcd/Dirac.h
diff --git a/lib/qcd/Grid_qcd_wilson_dop.cc b/lib/qcd/Grid_qcd_wilson_dop.cc
deleted file mode 100644
index 9a3f5f6a..00000000
--- a/lib/qcd/Grid_qcd_wilson_dop.cc
+++ /dev/null
@@ -1,217 +0,0 @@
-#include <Grid.h>
-
-namespace Grid {
-namespace QCD {
-
-const std::vector<int> WilsonMatrix::directions   ({0,1,2,3, 0, 1, 2, 3});
-const std::vector<int> WilsonMatrix::displacements({1,1,1,1,-1,-1,-1,-1});
-
-  int WilsonMatrix::HandOptDslash;
-
-  class WilsonCompressor {
-  public:
-    int mu;
-    int dag;
-
-    WilsonCompressor(int _dag){
-      mu=0;
-      dag=_dag;
-      assert((dag==0)||(dag==1));
-    }
-    void Point(int p) { 
-      mu=p;
-    };
-
-    vHalfSpinColourVector operator () (const vSpinColourVector &in)
-    {
-      vHalfSpinColourVector ret;
-      int mudag=mu;
-      if (dag) {
-	mudag=(mu+Nd)%(2*Nd);
-      }
-      switch(mudag) {
-      case Xp:
-	spProjXp(ret,in);
-	break;
-      case Yp:
-	spProjYp(ret,in);
-	break;
-      case Zp:
-	spProjZp(ret,in);
-	break;
-      case Tp:
-	spProjTp(ret,in);
-	break;
-      case Xm:
-	spProjXm(ret,in);
-	break;
-      case Ym:
-	spProjYm(ret,in);
-	break;
-      case Zm:
-	spProjZm(ret,in);
-	break;
-      case Tm:
-	spProjTm(ret,in);
-	break;
-      default: 
-	assert(0);
-	break;
-      }
-      return ret;
-    }
-  };
-
-  WilsonMatrix::WilsonMatrix(LatticeGaugeField &_Umu,GridCartesian &Fgrid,GridRedBlackCartesian &Hgrid, double _mass)  : 
-    CheckerBoardedSparseMatrixBase<LatticeFermion>(&Fgrid,&Hgrid),
-    Stencil    (  _grid,npoint,Even,directions,displacements),
-    StencilEven(_cbgrid,npoint,Even,directions,displacements), // source is Even
-    StencilOdd (_cbgrid,npoint,Odd ,directions,displacements), // source is Odd
-    mass(_mass),
-    Umu(_grid),
-    UmuEven(_cbgrid),
-    UmuOdd (_cbgrid)
-  {
-    // Allocate the required comms buffer
-    comm_buf.resize(Stencil._unified_buffer_size); // this is always big enough to contain EO
-    
-    DoubleStore(Umu,_Umu);
-    pickCheckerboard(Even,UmuEven,Umu);
-    pickCheckerboard(Odd ,UmuOdd,Umu);
-  }
-      
-void WilsonMatrix::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu)
-{
-  LatticeColourMatrix U(_grid);
-
-  for(int mu=0;mu<Nd;mu++){
-    U = peekIndex<LorentzIndex>(Umu,mu);
-    pokeIndex<LorentzIndex>(Uds,U,mu);
-    U = adj(Cshift(U,mu,-1));
-    pokeIndex<LorentzIndex>(Uds,U,mu+4);
-  }
-}
-
-RealD WilsonMatrix::M(const LatticeFermion &in, LatticeFermion &out)
-{
-  out.checkerboard=in.checkerboard;
-  Dhop(in,out,DaggerNo);
-  out = (4+mass)*in - 0.5*out  ; // FIXME : axpby_norm! fusion fun
-  return norm2(out);
-}
-RealD WilsonMatrix::Mdag(const LatticeFermion &in, LatticeFermion &out)
-{
-  out.checkerboard=in.checkerboard;
-  Dhop(in,out,DaggerYes);
-  out = (4+mass)*in - 0.5*out  ; // FIXME : axpby_norm! fusion fun
-  return norm2(out);
-}
-
-void WilsonMatrix::Meooe(const LatticeFermion &in, LatticeFermion &out)
-{
-  if ( in.checkerboard == Odd ) {
-    DhopEO(in,out,DaggerNo);
-  } else {
-    DhopOE(in,out,DaggerNo);
-  }
-  out = (-0.5)*out; // FIXME : scale factor in Dhop
-}
-void WilsonMatrix::MeooeDag(const LatticeFermion &in, LatticeFermion &out)
-{
-  if ( in.checkerboard == Odd ) {
-    DhopEO(in,out,DaggerYes);
-  } else {
-    DhopOE(in,out,DaggerYes);
-  }
-  out = (-0.5)*out; // FIXME : scale factor in Dhop
-}
-void WilsonMatrix::Mooee(const LatticeFermion &in, LatticeFermion &out)
-{
-  out.checkerboard = in.checkerboard;
-  out = (4.0+mass)*in;
-  return ;
-}
-void WilsonMatrix::MooeeDag(const LatticeFermion &in, LatticeFermion &out)
-{
-  out.checkerboard = in.checkerboard;
-  Mooee(in,out);
-}
-void WilsonMatrix::MooeeInv(const LatticeFermion &in, LatticeFermion &out)
-{
-  out.checkerboard = in.checkerboard;
-  out = (1.0/(4.0+mass))*in;
-  return ;
-}
-void WilsonMatrix::MooeeInvDag(const LatticeFermion &in, LatticeFermion &out)
-{
-  out.checkerboard = in.checkerboard;
-  MooeeInv(in,out);
-}
-
-void WilsonMatrix::DhopInternal(CartesianStencil & st,LatticeDoubledGaugeField & U,
-				const LatticeFermion &in, LatticeFermion &out,int dag)
-{
-  assert((dag==DaggerNo) ||(dag==DaggerYes));
-  WilsonCompressor compressor(dag);
-  st.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
-
-  if ( dag == DaggerYes ) {
-    if( HandOptDslash ) {
-PARALLEL_FOR_LOOP
-      for(int sss=0;sss<in._grid->oSites();sss++){
-        DiracOptHand::DhopSiteDag(st,U,comm_buf,sss,in,out);
-      }
-    } else { 
-PARALLEL_FOR_LOOP
-      for(int sss=0;sss<in._grid->oSites();sss++){
-        DiracOpt::DhopSiteDag(st,U,comm_buf,sss,in,out);
-      }
-    }
-  } else {
-    if( HandOptDslash ) {
-PARALLEL_FOR_LOOP
-      for(int sss=0;sss<in._grid->oSites();sss++){
-        DiracOptHand::DhopSite(st,U,comm_buf,sss,in,out);
-      }
-    } else { 
-PARALLEL_FOR_LOOP
-      for(int sss=0;sss<in._grid->oSites();sss++){
-        DiracOpt::DhopSite(st,U,comm_buf,sss,in,out);
-      }
-    }
-  }
-}
-void WilsonMatrix::DhopOE(const LatticeFermion &in, LatticeFermion &out,int dag)
-{
-  conformable(in._grid,_cbgrid);    // verifies half grid
-  conformable(in._grid,out._grid); // drops the cb check
-
-  assert(in.checkerboard==Even);
-  out.checkerboard = Odd;
-
-  DhopInternal(StencilEven,UmuOdd,in,out,dag);
-}
-void WilsonMatrix::DhopEO(const LatticeFermion &in, LatticeFermion &out,int dag)
-{
-  conformable(in._grid,_cbgrid);    // verifies half grid
-  conformable(in._grid,out._grid); // drops the cb check
-
-  assert(in.checkerboard==Odd);
-  out.checkerboard = Even;
-
-  DhopInternal(StencilOdd,UmuEven,in,out,dag);
-}
-void WilsonMatrix::Dhop(const LatticeFermion &in, LatticeFermion &out,int dag)
-{
-  conformable(in._grid,_grid); // verifies full grid
-  conformable(in._grid,out._grid);
-
-  out.checkerboard = in.checkerboard;
-
-  DhopInternal(Stencil,Umu,in,out,dag);
-}
-
-}}
-
-
-
diff --git a/lib/qcd/Grid_qcd_wilson_dop.h b/lib/qcd/Grid_qcd_wilson_dop.h
deleted file mode 100644
index 87418603..00000000
--- a/lib/qcd/Grid_qcd_wilson_dop.h
+++ /dev/null
@@ -1,105 +0,0 @@
-#ifndef  GRID_QCD_WILSON_DOP_H
-#define  GRID_QCD_WILSON_DOP_H
-
-
-namespace Grid {
-
-  namespace QCD {
-
-  // Should be in header?
-    const int Xp = 0;
-    const int Yp = 1;
-    const int Zp = 2;
-    const int Tp = 3;
-    const int Xm = 4;
-    const int Ym = 5;
-    const int Zm = 6;
-    const int Tm = 7;
-
-    class WilsonMatrix : public CheckerBoardedSparseMatrixBase<LatticeFermion>
-    {
-      //NB r=1;
-    public:
-      static int HandOptDslash;
-
-      double                        mass;
-      //      GridBase                     *    grid; // Inherited
-      //      GridBase                     *  cbgrid;
-
-      //Defines the stencils for even and odd
-      CartesianStencil Stencil; 
-      CartesianStencil StencilEven; 
-      CartesianStencil StencilOdd; 
-
-      // Copy of the gauge field , with even and odd subsets
-      LatticeDoubledGaugeField Umu;
-      LatticeDoubledGaugeField UmuEven;
-      LatticeDoubledGaugeField UmuOdd;
-
-      static const int npoint=8;
-      static const std::vector<int> directions   ;
-      static const std::vector<int> displacements;
-      static const int Xp,Xm,Yp,Ym,Zp,Zm,Tp,Tm;
-
-      // Comms buffer
-      std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  comm_buf;
-
-      // Constructor
-      WilsonMatrix(LatticeGaugeField &_Umu,GridCartesian &Fgrid,GridRedBlackCartesian &Hgrid,double _mass);
-
-      // DoubleStore
-      void DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu);
-
-      // override multiply
-      virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out);
-      virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out);
-
-      // half checkerboard operaions
-      virtual void   Meooe       (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
-
-      // non-hermitian hopping term; half cb or both
-      void Dhop  (const LatticeFermion &in, LatticeFermion &out,int dag);
-      void DhopOE(const LatticeFermion &in, LatticeFermion &out,int dag);
-      void DhopEO(const LatticeFermion &in, LatticeFermion &out,int dag);
-      void DhopInternal(CartesianStencil & st,LatticeDoubledGaugeField &U,
-			const LatticeFermion &in, LatticeFermion &out,int dag);
-
-      typedef iScalar<iMatrix<vComplex, Nc> > matrix;
-
-      
-    };
-
-
-    class DiracOpt {
-    public:
-      // These ones will need to be package intelligently. WilsonType base class
-      // for use by DWF etc..
-      static void DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
-		    std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-		    int ss,const LatticeFermion &in, LatticeFermion &out);
-      static void DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
-		       std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-		       int ss,const LatticeFermion &in, LatticeFermion &out);
-
-    };
-    class DiracOptHand {
-    public:
-      // These ones will need to be package intelligently. WilsonType base class
-      // for use by DWF etc..
-      static void DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
-		    std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-		    int ss,const LatticeFermion &in, LatticeFermion &out);
-      static void DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
-		       std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-		       int ss,const LatticeFermion &in, LatticeFermion &out);
-
-    };
-
-  }
-}
-#endif
diff --git a/lib/qcd/Grid_qcd.h b/lib/qcd/QCD.h
similarity index 97%
rename from lib/qcd/Grid_qcd.h
rename to lib/qcd/QCD.h
index 959f3529..7c45eb23 100644
--- a/lib/qcd/Grid_qcd.h
+++ b/lib/qcd/QCD.h
@@ -4,6 +4,16 @@ namespace Grid{
 
 namespace QCD {
 
+
+    static const int Xp = 0;
+    static const int Yp = 1;
+    static const int Zp = 2;
+    static const int Tp = 3;
+    static const int Xm = 4;
+    static const int Ym = 5;
+    static const int Zm = 6;
+    static const int Tm = 7;
+
     static const int Nc=3;
     static const int Ns=4;
     static const int Nd=4;
@@ -297,9 +307,8 @@ namespace QCD {
 }   //namespace QCD
 } // Grid
 
-#include <qcd/Grid_qcd_dirac.h>
-#include <qcd/Grid_qcd_2spinor.h>
-//#include <qcd/Grid_qcd_pauli.h>
-#include <qcd/Grid_qcd_wilson_dop.h>
+#include <qcd/Dirac.h>
+#include <qcd/TwoSpinor.h>
+#include <qcd/action/Actions.h>
 
 #endif
diff --git a/lib/qcd/Grid_qcd_2spinor.h b/lib/qcd/TwoSpinor.h
similarity index 100%
rename from lib/qcd/Grid_qcd_2spinor.h
rename to lib/qcd/TwoSpinor.h
diff --git a/lib/qcd/action/Actions.h b/lib/qcd/action/Actions.h
new file mode 100644
index 00000000..c4e8a2f0
--- /dev/null
+++ b/lib/qcd/action/Actions.h
@@ -0,0 +1,10 @@
+#ifndef GRID_QCD_ACTIONS_H
+#define GRID_QCD_ACTIONS_H
+
+#include <qcd/action/fermion/FermionAction.h>
+#include <qcd/action/fermion/WilsonCompressor.h>
+#include <qcd/action/fermion/WilsonKernels.h>
+#include <qcd/action/fermion/WilsonFermion.h>
+#include <qcd/action/fermion/FiveDimWilsonFermion.h>
+
+#endif
diff --git a/lib/qcd/action/fermion/FermionAction.h b/lib/qcd/action/fermion/FermionAction.h
new file mode 100644
index 00000000..1b05174b
--- /dev/null
+++ b/lib/qcd/action/fermion/FermionAction.h
@@ -0,0 +1,47 @@
+#ifndef  GRID_QCD_WILSON_DOP_H
+#define  GRID_QCD_WILSON_DOP_H
+
+namespace Grid {
+
+  namespace QCD {
+
+    //////////////////////////////////////////////////////////////////////////////
+    // Four component fermions
+    // Should type template the vector and gauge types
+    // Think about multiple representations
+    //////////////////////////////////////////////////////////////////////////////
+    template<class FermionField,class GaugeField>
+    class FermionAction : public CheckerBoardedSparseMatrixBase<FermionField>
+    {
+    public:
+
+      GridBase * Grid(void)   { return FermionGrid(); };   // this is all the linalg routines need to know
+      GridBase * RedBlackGrid(void) { return FermionRedBlackGrid(); };
+
+      virtual GridBase *FermionGrid(void)         =0;
+      virtual GridBase *FermionRedBlackGrid(void) =0;
+      virtual GridBase *GaugeGrid(void)           =0;
+      virtual GridBase *GaugeRedBlackGrid(void)   =0;
+
+      // override multiply
+      virtual RealD  M    (const FermionField &in, FermionField &out)=0;
+      virtual RealD  Mdag (const FermionField &in, FermionField &out)=0;
+
+      // half checkerboard operaions
+      virtual void   Meooe       (const FermionField &in, FermionField &out)=0;
+      virtual void   MeooeDag    (const FermionField &in, FermionField &out)=0;
+      virtual void   Mooee       (const FermionField &in, FermionField &out)=0;
+      virtual void   MooeeDag    (const FermionField &in, FermionField &out)=0;
+      virtual void   MooeeInv    (const FermionField &in, FermionField &out)=0;
+      virtual void   MooeeInvDag (const FermionField &in, FermionField &out)=0;
+
+      // non-hermitian hopping term; half cb or both
+      virtual void Dhop  (const FermionField &in, FermionField &out,int dag)=0;
+      virtual void DhopOE(const FermionField &in, FermionField &out,int dag)=0;
+      virtual void DhopEO(const FermionField &in, FermionField &out,int dag)=0;
+
+    };
+
+  }
+}
+#endif
diff --git a/lib/qcd/action/fermion/FiveDimWilsonFermion.cc b/lib/qcd/action/fermion/FiveDimWilsonFermion.cc
new file mode 100644
index 00000000..43645899
--- /dev/null
+++ b/lib/qcd/action/fermion/FiveDimWilsonFermion.cc
@@ -0,0 +1,228 @@
+#include <Grid.h>
+
+namespace Grid {
+namespace QCD {
+  
+  // S-direction is INNERMOST and takes no part in the parity.
+  const std::vector<int> FiveDimWilsonFermion::directions   ({1,2,3,4, 1, 2, 3, 4});
+  const std::vector<int> FiveDimWilsonFermion::displacements({1,1,1,1,-1,-1,-1,-1});
+
+  int FiveDimWilsonFermion::HandOptDslash;
+
+  // 5d lattice for DWF.
+  FiveDimWilsonFermion::FiveDimWilsonFermion(LatticeGaugeField &_Umu,
+					   GridCartesian         &FiveDimGrid,
+					   GridRedBlackCartesian &FiveDimRedBlackGrid,
+					   GridCartesian         &FourDimGrid,
+					   GridRedBlackCartesian &FourDimRedBlackGrid,
+					   double _mass) :
+  _FiveDimGrid(&FiveDimGrid),
+  _FiveDimRedBlackGrid(&FiveDimRedBlackGrid),
+  _FourDimGrid(&FourDimGrid),
+  _FourDimRedBlackGrid(&FourDimRedBlackGrid),
+  Stencil    (_FiveDimGrid,npoint,Even,directions,displacements),
+  StencilEven(_FiveDimRedBlackGrid,npoint,Even,directions,displacements), // source is Even
+  StencilOdd (_FiveDimRedBlackGrid,npoint,Odd ,directions,displacements), // source is Odd
+  mass(_mass),
+  Umu(_FourDimGrid),
+  UmuEven(_FourDimRedBlackGrid),
+  UmuOdd (_FourDimRedBlackGrid),
+  Lebesgue(_FourDimGrid),
+  LebesgueEvenOdd(_FourDimRedBlackGrid)
+{
+  // some assertions
+  assert(FiveDimGrid._ndimension==5);
+  assert(FourDimGrid._ndimension==4);
+  
+  assert(FiveDimRedBlackGrid._ndimension==5);
+  assert(FourDimRedBlackGrid._ndimension==4);
+
+  assert(FiveDimRedBlackGrid._checker_dim==1);
+
+  // Dimension zero of the five-d is the Ls direction
+  Ls=FiveDimGrid._fdimensions[0];
+  assert(FiveDimRedBlackGrid._fdimensions[0]==Ls);
+  assert(FiveDimRedBlackGrid._processors[0] ==1);
+  assert(FiveDimRedBlackGrid._simd_layout[0]==1);
+  assert(FiveDimGrid._processors[0]         ==1);
+  assert(FiveDimGrid._simd_layout[0]        ==1);
+
+  // Other dimensions must match the decomposition of the four-D fields 
+  for(int d=0;d<4;d++){
+    assert(FourDimRedBlackGrid._fdimensions[d]  ==FourDimGrid._fdimensions[d]);
+    assert(FiveDimRedBlackGrid._fdimensions[d+1]==FourDimGrid._fdimensions[d]);
+
+    assert(FourDimRedBlackGrid._processors[d]   ==FourDimGrid._processors[d]);
+    assert(FiveDimRedBlackGrid._processors[d+1] ==FourDimGrid._processors[d]);
+
+    assert(FourDimRedBlackGrid._simd_layout[d]  ==FourDimGrid._simd_layout[d]);
+    assert(FiveDimRedBlackGrid._simd_layout[d+1]==FourDimGrid._simd_layout[d]);
+
+    assert(FiveDimGrid._fdimensions[d+1]        ==FourDimGrid._fdimensions[d]);
+    assert(FiveDimGrid._processors[d+1]         ==FourDimGrid._processors[d]);
+    assert(FiveDimGrid._simd_layout[d+1]        ==FourDimGrid._simd_layout[d]);
+  }
+
+  // Allocate the required comms buffer
+  comm_buf.resize(Stencil._unified_buffer_size); // this is always big enough to contain EO
+  
+  DoubleStore(Umu,_Umu);
+  pickCheckerboard(Even,UmuEven,Umu);
+  pickCheckerboard(Odd ,UmuOdd,Umu);
+}
+void FiveDimWilsonFermion::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu)
+{
+  conformable(Uds._grid,GaugeGrid());
+  conformable(Umu._grid,GaugeGrid());
+  LatticeColourMatrix U(GaugeGrid());
+  for(int mu=0;mu<Nd;mu++){
+    U = peekIndex<LorentzIndex>(Umu,mu);
+    pokeIndex<LorentzIndex>(Uds,U,mu);
+    U = adj(Cshift(U,mu,-1));
+    pokeIndex<LorentzIndex>(Uds,U,mu+4);
+  }
+}
+
+RealD FiveDimWilsonFermion::M(const LatticeFermion &in, LatticeFermion &out)
+{
+  out.checkerboard=in.checkerboard;
+  Dhop(in,out,DaggerNo);
+  return axpy_norm(out,5.0-M5,in,out);
+}
+RealD FiveDimWilsonFermion::Mdag(const LatticeFermion &in, LatticeFermion &out)
+{
+  out.checkerboard=in.checkerboard;
+  Dhop(in,out,DaggerYes);
+  return axpy_norm(out,5.0-M5,in,out);
+}
+void FiveDimWilsonFermion::Meooe(const LatticeFermion &in, LatticeFermion &out)
+{
+  if ( in.checkerboard == Odd ) {
+    DhopEO(in,out,DaggerNo);
+  } else {
+    DhopOE(in,out,DaggerNo);
+  }
+}
+void FiveDimWilsonFermion::MeooeDag(const LatticeFermion &in, LatticeFermion &out)
+{
+  if ( in.checkerboard == Odd ) {
+    DhopEO(in,out,DaggerYes);
+  } else {
+    DhopOE(in,out,DaggerYes);
+  }
+}
+void FiveDimWilsonFermion::Mooee(const LatticeFermion &in, LatticeFermion &out)
+{
+  out.checkerboard = in.checkerboard;
+  out = (5.0-M5)*in;
+  return ;
+}
+void FiveDimWilsonFermion::MooeeDag(const LatticeFermion &in, LatticeFermion &out)
+{
+  out.checkerboard = in.checkerboard;
+  Mooee(in,out);
+}
+void FiveDimWilsonFermion::MooeeInv(const LatticeFermion &in, LatticeFermion &out)
+{
+  out.checkerboard = in.checkerboard;
+  out = (1.0/(5.0-M5))*in;
+  return ;
+}
+void FiveDimWilsonFermion::MooeeInvDag(const LatticeFermion &in, LatticeFermion &out)
+{
+  out.checkerboard = in.checkerboard;
+  MooeeInv(in,out);
+}
+void FiveDimWilsonFermion::DhopInternal(CartesianStencil & st, LebesgueOrder &lo,
+					LatticeDoubledGaugeField & U,
+					const LatticeFermion &in, LatticeFermion &out,int dag)
+{
+  assert((dag==DaggerNo) ||(dag==DaggerYes));
+
+  WilsonCompressor compressor(dag);
+
+  st.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
+  
+  // Dhop takes the 4d grid from U, and makes a 5d index for fermion
+  // Not loop ordering and data layout.
+  // Designed to create 
+  // - per thread reuse in L1 cache for U
+  // - 8 linear access unit stride streams per thread for Fermion for hw prefetchable.
+  if ( dag == DaggerYes ) {
+    if( HandOptDslash ) {
+      for(int ss=0;ss<U._grid->oSites();ss++){
+	int sU=lo.Reorder(ss);
+PARALLEL_FOR_LOOP
+	for(int s=0;s<Ls;s++){
+	  int sF = s+Ls*sU;
+	  DiracOptHand::DhopSiteDag(st,U,comm_buf,sF,sU,in,out);
+	}
+      }
+    } else { 
+      for(int ss=0;ss<U._grid->oSites();ss++){
+	int sU=lo.Reorder(ss);
+PARALLEL_FOR_LOOP
+	for(int s=0;s<Ls;s++){
+	  int sF = s+Ls*sU;
+	  DiracOpt::DhopSiteDag(st,U,comm_buf,sF,sU,in,out);
+	}
+      }
+    }
+  } else {
+    if( HandOptDslash ) {
+
+PARALLEL_FOR_LOOP
+      for(int ss=0;ss<U._grid->oSites();ss++){
+	int sU=lo.Reorder(ss);
+	for(int s=0;s<Ls;s++){
+	  int sF = s+Ls*sU;
+	  DiracOptHand::DhopSite(st,U,comm_buf,sF,sU,in,out);
+	}
+      }
+
+    } else { 
+      for(int ss=0;ss<U._grid->oSites();ss++){
+	int sU=lo.Reorder(ss);
+PARALLEL_FOR_LOOP
+	for(int s=0;s<Ls;s++){
+	  int sF = s+Ls*sU; 
+	  DiracOpt::DhopSite(st,U,comm_buf,sF,sU,in,out);
+	}
+      }
+    }
+  }
+}
+void FiveDimWilsonFermion::DhopOE(const LatticeFermion &in, LatticeFermion &out,int dag)
+{
+  conformable(in._grid,FermionRedBlackGrid());    // verifies half grid
+  conformable(in._grid,out._grid); // drops the cb check
+
+  assert(in.checkerboard==Even);
+  out.checkerboard = Odd;
+
+  DhopInternal(StencilEven,LebesgueEvenOdd,UmuOdd,in,out,dag);
+}
+void FiveDimWilsonFermion::DhopEO(const LatticeFermion &in, LatticeFermion &out,int dag)
+{
+  conformable(in._grid,FermionRedBlackGrid());    // verifies half grid
+  conformable(in._grid,out._grid); // drops the cb check
+
+  assert(in.checkerboard==Odd);
+  out.checkerboard = Even;
+
+  DhopInternal(StencilOdd,LebesgueEvenOdd,UmuEven,in,out,dag);
+}
+void FiveDimWilsonFermion::Dhop(const LatticeFermion &in, LatticeFermion &out,int dag)
+{
+  conformable(in._grid,FermionGrid()); // verifies full grid
+  conformable(in._grid,out._grid);
+
+  out.checkerboard = in.checkerboard;
+
+  DhopInternal(Stencil,Lebesgue,Umu,in,out,dag);
+}
+
+}}
+
+
+
diff --git a/lib/qcd/action/fermion/FiveDimWilsonFermion.h b/lib/qcd/action/fermion/FiveDimWilsonFermion.h
new file mode 100644
index 00000000..3aa85023
--- /dev/null
+++ b/lib/qcd/action/fermion/FiveDimWilsonFermion.h
@@ -0,0 +1,108 @@
+#ifndef  GRID_QCD_DWF_H
+#define  GRID_QCD_DWF_H
+
+namespace Grid {
+
+  namespace QCD {
+
+    ////////////////////////////////////////////////////////////////////////////////
+    // This is the 4d red black case appropriate to support
+    //
+    // parity = (x+y+z+t)|2;
+    // generalised five dim fermions like mobius, zolotarev etc..	
+    //
+    // i.e. even even contains fifth dim hopping term.
+    //
+    // [DIFFERS from original CPS red black implementation parity = (x+y+z+t+s)|2 ]
+    ////////////////////////////////////////////////////////////////////////////////
+    class FiveDimWilsonFermion : public FermionAction<LatticeFermion,LatticeGaugeField>
+    {
+    public:
+      ///////////////////////////////////////////////////////////////
+      // Implement the abstract base
+      ///////////////////////////////////////////////////////////////
+      GridBase *GaugeGrid(void)              { return _FourDimGrid ;}
+      GridBase *GaugeRedBlackGrid(void)      { return _FourDimRedBlackGrid ;}
+      GridBase *FermionGrid(void)            { return _FiveDimGrid;}
+      GridBase *FermionRedBlackGrid(void)    { return _FiveDimRedBlackGrid;}
+
+      // override multiply
+      virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out);
+      virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out);
+
+      // half checkerboard operaions
+      virtual void   Meooe       (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
+
+      // non-hermitian hopping term; half cb or both
+      void Dhop  (const LatticeFermion &in, LatticeFermion &out,int dag);
+      void DhopOE(const LatticeFermion &in, LatticeFermion &out,int dag);
+      void DhopEO(const LatticeFermion &in, LatticeFermion &out,int dag);
+
+      ///////////////////////////////////////////////////////////////
+      // New methods added 
+      ///////////////////////////////////////////////////////////////
+      void DhopInternal(CartesianStencil & st,
+			LebesgueOrder &lo,
+			LatticeDoubledGaugeField &U,
+			const LatticeFermion &in, 
+			LatticeFermion &out,
+			int dag);
+
+      // Constructors
+      FiveDimWilsonFermion(LatticeGaugeField &_Umu,
+			  GridCartesian         &FiveDimGrid,
+			  GridRedBlackCartesian &FiveDimRedBlackGrid,
+			  GridCartesian         &FourDimGrid,
+			  GridRedBlackCartesian &FourDimRedBlackGrid,
+			  double _mass);
+
+      // DoubleStore
+      void DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu);
+
+      ///////////////////////////////////////////////////////////////
+      // Data members require to support the functionality
+      ///////////////////////////////////////////////////////////////
+      static int HandOptDslash; // these are a temporary hack
+
+    protected:
+
+      // Add these to the support from Wilson
+      GridBase *_FourDimGrid;
+      GridBase *_FourDimRedBlackGrid;
+      GridBase *_FiveDimGrid;
+      GridBase *_FiveDimRedBlackGrid;
+
+      static const int npoint=8;
+      static const std::vector<int> directions   ;
+      static const std::vector<int> displacements;
+
+      double                        M5;
+      double                        mass;
+      int Ls;
+
+      //Defines the stencils for even and odd
+      CartesianStencil Stencil; 
+      CartesianStencil StencilEven; 
+      CartesianStencil StencilOdd; 
+
+      // Copy of the gauge field , with even and odd subsets
+      LatticeDoubledGaugeField Umu;
+      LatticeDoubledGaugeField UmuEven;
+      LatticeDoubledGaugeField UmuOdd;
+
+      LebesgueOrder Lebesgue;
+      LebesgueOrder LebesgueEvenOdd;
+
+      // Comms buffer
+      std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  comm_buf;
+      
+    };
+  }
+}
+
+#endif
diff --git a/lib/qcd/action/fermion/WilsonCompressor.h b/lib/qcd/action/fermion/WilsonCompressor.h
new file mode 100644
index 00000000..d9fe977a
--- /dev/null
+++ b/lib/qcd/action/fermion/WilsonCompressor.h
@@ -0,0 +1,61 @@
+#ifndef  GRID_QCD_WILSON_COMPRESSOR_H
+#define  GRID_QCD_WILSON_COMPRESSOR_H
+
+namespace Grid {
+namespace QCD {
+
+  class WilsonCompressor {
+  public:
+    int mu;
+    int dag;
+
+    WilsonCompressor(int _dag){
+      mu=0;
+      dag=_dag;
+      assert((dag==0)||(dag==1));
+    }
+    void Point(int p) { 
+      mu=p;
+    };
+
+    vHalfSpinColourVector operator () (const vSpinColourVector &in)
+    {
+      vHalfSpinColourVector ret;
+      int mudag=mu;
+      if (dag) {
+	mudag=(mu+Nd)%(2*Nd);
+      }
+      switch(mudag) {
+      case Xp:
+	spProjXp(ret,in);
+	break;
+      case Yp:
+	spProjYp(ret,in);
+	break;
+      case Zp:
+	spProjZp(ret,in);
+	break;
+      case Tp:
+	spProjTp(ret,in);
+	break;
+      case Xm:
+	spProjXm(ret,in);
+	break;
+      case Ym:
+	spProjYm(ret,in);
+	break;
+      case Zm:
+	spProjZm(ret,in);
+	break;
+      case Tm:
+	spProjTm(ret,in);
+	break;
+      default: 
+	assert(0);
+	break;
+      }
+      return ret;
+    }
+  };
+}} // namespace close
+#endif
diff --git a/lib/qcd/action/fermion/WilsonFermion.cc b/lib/qcd/action/fermion/WilsonFermion.cc
new file mode 100644
index 00000000..aa30a7fa
--- /dev/null
+++ b/lib/qcd/action/fermion/WilsonFermion.cc
@@ -0,0 +1,163 @@
+#include <Grid.h>
+
+namespace Grid {
+namespace QCD {
+
+const std::vector<int> WilsonFermion::directions   ({0,1,2,3, 0, 1, 2, 3});
+const std::vector<int> WilsonFermion::displacements({1,1,1,1,-1,-1,-1,-1});
+
+int WilsonFermion::HandOptDslash;
+
+WilsonFermion::WilsonFermion(LatticeGaugeField &_Umu,
+			   GridCartesian         &Fgrid,
+			   GridRedBlackCartesian &Hgrid, 
+			   double _mass) :
+  _grid(&Fgrid),
+  _cbgrid(&Hgrid),
+  Stencil    (&Fgrid,npoint,Even,directions,displacements),
+  StencilEven(&Hgrid,npoint,Even,directions,displacements), // source is Even
+  StencilOdd (&Hgrid,npoint,Odd ,directions,displacements), // source is Odd
+  mass(_mass),
+  Umu(&Fgrid),
+  UmuEven(&Hgrid),
+  UmuOdd (&Hgrid)
+{
+  // Allocate the required comms buffer
+  comm_buf.resize(Stencil._unified_buffer_size); // this is always big enough to contain EO
+  DoubleStore(Umu,_Umu);
+  pickCheckerboard(Even,UmuEven,Umu);
+  pickCheckerboard(Odd ,UmuOdd,Umu);
+}
+      
+void WilsonFermion::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu)
+{
+  conformable(Uds._grid,GaugeGrid());
+  conformable(Umu._grid,GaugeGrid());
+  LatticeColourMatrix U(GaugeGrid());
+  for(int mu=0;mu<Nd;mu++){
+    U = peekIndex<LorentzIndex>(Umu,mu);
+    pokeIndex<LorentzIndex>(Uds,U,mu);
+    U = adj(Cshift(U,mu,-1));
+    pokeIndex<LorentzIndex>(Uds,U,mu+4);
+  }
+}
+
+RealD WilsonFermion::M(const LatticeFermion &in, LatticeFermion &out)
+{
+  out.checkerboard=in.checkerboard;
+  Dhop(in,out,DaggerNo);
+  return axpy_norm(out,4+mass,in,out);
+}
+RealD WilsonFermion::Mdag(const LatticeFermion &in, LatticeFermion &out)
+{
+  out.checkerboard=in.checkerboard;
+  Dhop(in,out,DaggerYes);
+  return axpy_norm(out,4+mass,in,out);
+}
+
+void WilsonFermion::Meooe(const LatticeFermion &in, LatticeFermion &out)
+{
+  if ( in.checkerboard == Odd ) {
+    DhopEO(in,out,DaggerNo);
+  } else {
+    DhopOE(in,out,DaggerNo);
+  }
+}
+void WilsonFermion::MeooeDag(const LatticeFermion &in, LatticeFermion &out)
+{
+  if ( in.checkerboard == Odd ) {
+    DhopEO(in,out,DaggerYes);
+  } else {
+    DhopOE(in,out,DaggerYes);
+  }
+}
+void WilsonFermion::Mooee(const LatticeFermion &in, LatticeFermion &out)
+{
+  out.checkerboard = in.checkerboard;
+  out = (4.0+mass)*in;
+  return ;
+}
+void WilsonFermion::MooeeDag(const LatticeFermion &in, LatticeFermion &out)
+{
+  out.checkerboard = in.checkerboard;
+  Mooee(in,out);
+}
+void WilsonFermion::MooeeInv(const LatticeFermion &in, LatticeFermion &out)
+{
+  out.checkerboard = in.checkerboard;
+  out = (1.0/(4.0+mass))*in;
+  return ;
+}
+void WilsonFermion::MooeeInvDag(const LatticeFermion &in, LatticeFermion &out)
+{
+  out.checkerboard = in.checkerboard;
+  MooeeInv(in,out);
+}
+
+void WilsonFermion::DhopInternal(CartesianStencil & st,LatticeDoubledGaugeField & U,
+				const LatticeFermion &in, LatticeFermion &out,int dag)
+{
+  assert((dag==DaggerNo) ||(dag==DaggerYes));
+  WilsonCompressor compressor(dag);
+  st.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
+
+  if ( dag == DaggerYes ) {
+    if( HandOptDslash ) {
+PARALLEL_FOR_LOOP
+      for(int sss=0;sss<in._grid->oSites();sss++){
+        DiracOptHand::DhopSiteDag(st,U,comm_buf,sss,sss,in,out);
+      }
+    } else { 
+PARALLEL_FOR_LOOP
+      for(int sss=0;sss<in._grid->oSites();sss++){
+        DiracOpt::DhopSiteDag(st,U,comm_buf,sss,sss,in,out);
+      }
+    }
+  } else {
+    if( HandOptDslash ) {
+PARALLEL_FOR_LOOP
+      for(int sss=0;sss<in._grid->oSites();sss++){
+        DiracOptHand::DhopSite(st,U,comm_buf,sss,sss,in,out);
+      }
+    } else { 
+PARALLEL_FOR_LOOP
+      for(int sss=0;sss<in._grid->oSites();sss++){
+        DiracOpt::DhopSite(st,U,comm_buf,sss,sss,in,out);
+      }
+    }
+  }
+}
+void WilsonFermion::DhopOE(const LatticeFermion &in, LatticeFermion &out,int dag)
+{
+  conformable(in._grid,_cbgrid);    // verifies half grid
+  conformable(in._grid,out._grid); // drops the cb check
+
+  assert(in.checkerboard==Even);
+  out.checkerboard = Odd;
+
+  DhopInternal(StencilEven,UmuOdd,in,out,dag);
+}
+void WilsonFermion::DhopEO(const LatticeFermion &in, LatticeFermion &out,int dag)
+{
+  conformable(in._grid,_cbgrid);    // verifies half grid
+  conformable(in._grid,out._grid); // drops the cb check
+
+  assert(in.checkerboard==Odd);
+  out.checkerboard = Even;
+
+  DhopInternal(StencilOdd,UmuEven,in,out,dag);
+}
+void WilsonFermion::Dhop(const LatticeFermion &in, LatticeFermion &out,int dag)
+{
+  conformable(in._grid,_grid); // verifies full grid
+  conformable(in._grid,out._grid);
+
+  out.checkerboard = in.checkerboard;
+
+  DhopInternal(Stencil,Umu,in,out,dag);
+}
+
+}}
+
+
+
diff --git a/lib/qcd/action/fermion/WilsonFermion.h b/lib/qcd/action/fermion/WilsonFermion.h
new file mode 100644
index 00000000..5c208131
--- /dev/null
+++ b/lib/qcd/action/fermion/WilsonFermion.h
@@ -0,0 +1,87 @@
+#ifndef  GRID_QCD_WILSON_FERMION_H
+#define  GRID_QCD_WILSON_FERMION_H
+
+namespace Grid {
+
+  namespace QCD {
+
+    class WilsonFermion : public FermionAction<LatticeFermion,LatticeGaugeField>
+    {
+    public:
+
+      ///////////////////////////////////////////////////////////////
+      // Implement the abstract base
+      ///////////////////////////////////////////////////////////////
+      GridBase *GaugeGrid(void)              { return _grid ;}
+      GridBase *GaugeRedBlackGrid(void)      { return _cbgrid ;}
+      GridBase *FermionGrid(void)            { return _grid;}
+      GridBase *FermionRedBlackGrid(void)    { return _cbgrid;}
+
+      // override multiply
+      virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out);
+      virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out);
+
+      // half checkerboard operaions
+      void   Meooe       (const LatticeFermion &in, LatticeFermion &out);
+      void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out); // remain virtual so we 
+      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out); // can derive Clover
+      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out); // from Wilson bas
+      virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
+
+      // non-hermitian hopping term; half cb or both
+      void Dhop  (const LatticeFermion &in, LatticeFermion &out,int dag);
+      void DhopOE(const LatticeFermion &in, LatticeFermion &out,int dag);
+      void DhopEO(const LatticeFermion &in, LatticeFermion &out,int dag);
+
+      ///////////////////////////////////////////////////////////////
+      // Extra methods added by derived
+      ///////////////////////////////////////////////////////////////
+      void DhopInternal(CartesianStencil & st,
+			LatticeDoubledGaugeField &U,
+			const LatticeFermion &in, 
+			LatticeFermion &out,
+			int dag);
+
+      // Constructor
+      WilsonFermion(LatticeGaugeField &_Umu,GridCartesian &Fgrid,GridRedBlackCartesian &Hgrid,double _mass);
+
+      // DoubleStore
+      void DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu);
+
+      ///////////////////////////////////////////////////////////////
+      // Data members require to support the functionality
+      ///////////////////////////////////////////////////////////////
+      static int HandOptDslash; // these are a temporary hack
+      static int MortonOrder;
+
+    protected:
+
+      double                        mass;
+
+      GridBase                     *    _grid; 
+      GridBase                     *  _cbgrid;
+
+      static const int npoint=8;
+      static const std::vector<int> directions   ;
+      static const std::vector<int> displacements;
+
+      //Defines the stencils for even and odd
+      CartesianStencil Stencil; 
+      CartesianStencil StencilEven; 
+      CartesianStencil StencilOdd; 
+
+      // Copy of the gauge field , with even and odd subsets
+      LatticeDoubledGaugeField Umu;
+      LatticeDoubledGaugeField UmuEven;
+      LatticeDoubledGaugeField UmuOdd;
+
+      // Comms buffer
+      std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  comm_buf;
+
+      
+    };
+
+  }
+}
+#endif
diff --git a/lib/qcd/Grid_qcd_dhop.cc b/lib/qcd/action/fermion/WilsonKernels.cc
similarity index 87%
rename from lib/qcd/Grid_qcd_dhop.cc
rename to lib/qcd/action/fermion/WilsonKernels.cc
index 1e5dcd16..879bbb7a 100644
--- a/lib/qcd/Grid_qcd_dhop.cc
+++ b/lib/qcd/action/fermion/WilsonKernels.cc
@@ -4,8 +4,8 @@ namespace Grid {
 namespace QCD {
 
 void DiracOpt::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
-			    std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			    int ss,const LatticeFermion &in, LatticeFermion &out)
+			std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+			int sF,int sU,const LatticeFermion &in, LatticeFermion &out)
 {
     vHalfSpinColourVector  tmp;    
     vHalfSpinColourVector  chi;    
@@ -16,6 +16,7 @@ void DiracOpt::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     //#define VERBOSE( A)  if ( ss<10 ) { std::cout << "site " <<ss << " " #A " neigh " << offset << " perm "<< perm <<std::endl;}    
 
     // Xp
+    int ss = sF;
     offset = st._offsets [Xp][ss];
     local  = st._is_local[Xp][ss];
     perm   = st._permute[Xp][ss];
@@ -29,7 +30,7 @@ void DiracOpt::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[ss](Xp),&chi());
+    mult(&Uchi(),&U._odata[sU](Xp),&chi());
     spReconXp(result,Uchi);
 
     //    std::cout << "XP_RECON"<<std::endl;
@@ -51,7 +52,7 @@ void DiracOpt::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[ss](Yp),&chi());
+    mult(&Uchi(),&U._odata[sU](Yp),&chi());
     accumReconYp(result,Uchi);
 
     // Zp
@@ -67,7 +68,7 @@ void DiracOpt::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[ss](Zp),&chi());
+    mult(&Uchi(),&U._odata[sU](Zp),&chi());
     accumReconZp(result,Uchi);
 
     // Tp
@@ -83,7 +84,7 @@ void DiracOpt::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[ss](Tp),&chi());
+    mult(&Uchi(),&U._odata[sU](Tp),&chi());
     accumReconTp(result,Uchi);
 
     // Xm
@@ -101,7 +102,7 @@ void DiracOpt::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[ss](Xm),&chi());
+    mult(&Uchi(),&U._odata[sU](Xm),&chi());
     accumReconXm(result,Uchi);
     //  std::cout << "XM_RECON_ACCUM"<<std::endl;
     //    std::cout << result()(0)(0) <<" "<<result()(0)(1) <<" "<<result()(0)(2) <<std::endl;
@@ -124,7 +125,7 @@ void DiracOpt::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[ss](Ym),&chi());
+    mult(&Uchi(),&U._odata[sU](Ym),&chi());
     accumReconYm(result,Uchi);
 
     // Zm
@@ -140,7 +141,7 @@ void DiracOpt::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[ss](Zm),&chi());
+    mult(&Uchi(),&U._odata[sU](Zm),&chi());
     accumReconZm(result,Uchi);
 
     // Tm
@@ -156,15 +157,15 @@ void DiracOpt::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[ss](Tm),&chi());
+    mult(&Uchi(),&U._odata[sU](Tm),&chi());
     accumReconTm(result,Uchi);
 
-    vstream(out._odata[ss],result);
+    vstream(out._odata[ss],result*(-0.5));
 }
 
 void DiracOpt::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
-			       std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			       int ss,const LatticeFermion &in, LatticeFermion &out)
+			   std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+			   int sF,int sU,const LatticeFermion &in, LatticeFermion &out)
 {
     vHalfSpinColourVector  tmp;    
     vHalfSpinColourVector  chi;    
@@ -173,6 +174,7 @@ void DiracOpt::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
     int offset,local,perm, ptype;
 
     // Xp
+    int ss=sF;
     offset = st._offsets [Xm][ss];
     local  = st._is_local[Xm][ss];
     perm   = st._permute[Xm][ss];
@@ -186,7 +188,7 @@ void DiracOpt::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[ss](Xm),&chi());
+    mult(&Uchi(),&U._odata[sU](Xm),&chi());
     spReconXp(result,Uchi);
 
     // Yp
@@ -202,7 +204,7 @@ void DiracOpt::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[ss](Ym),&chi());
+    mult(&Uchi(),&U._odata[sU](Ym),&chi());
     accumReconYp(result,Uchi);
 
     // Zp
@@ -218,7 +220,7 @@ void DiracOpt::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[ss](Zm),&chi());
+    mult(&Uchi(),&U._odata[sU](Zm),&chi());
     accumReconZp(result,Uchi);
 
     // Tp
@@ -234,7 +236,7 @@ void DiracOpt::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[ss](Tm),&chi());
+    mult(&Uchi(),&U._odata[sU](Tm),&chi());
     accumReconTp(result,Uchi);
 
     // Xm
@@ -252,7 +254,7 @@ void DiracOpt::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[ss](Xp),&chi());
+    mult(&Uchi(),&U._odata[sU](Xp),&chi());
     accumReconXm(result,Uchi);
 
     // Ym
@@ -269,7 +271,7 @@ void DiracOpt::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[ss](Yp),&chi());
+    mult(&Uchi(),&U._odata[sU](Yp),&chi());
     accumReconYm(result,Uchi);
 
     // Zm
@@ -285,7 +287,7 @@ void DiracOpt::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[ss](Zp),&chi());
+    mult(&Uchi(),&U._odata[sU](Zp),&chi());
     accumReconZm(result,Uchi);
 
     // Tm
@@ -301,9 +303,9 @@ void DiracOpt::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[ss](Tp),&chi());
+    mult(&Uchi(),&U._odata[sU](Tp),&chi());
     accumReconTm(result,Uchi);
 
-    vstream(out._odata[ss],result);
+    vstream(out._odata[ss],result*(-0.5));
 }
 }}
diff --git a/lib/qcd/action/fermion/WilsonKernels.h b/lib/qcd/action/fermion/WilsonKernels.h
new file mode 100644
index 00000000..cc535022
--- /dev/null
+++ b/lib/qcd/action/fermion/WilsonKernels.h
@@ -0,0 +1,42 @@
+#ifndef  GRID_QCD_DHOP_H
+#define  GRID_QCD_DHOP_H
+
+namespace Grid {
+
+  namespace QCD {
+
+    ////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+    // Helper classes that implement Wilson stencil for a single site.
+    ////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+    
+    // Generic version works for any Nc and with extra flavour indices
+    class DiracOpt {
+    public:
+      // These ones will need to be package intelligently. WilsonType base class
+      // for use by DWF etc..
+      static void DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
+			   std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+			   int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
+      static void DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
+			      std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+			      int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
+
+    };
+
+    // Hand unrolled for Nc=3, one flavour
+    class DiracOptHand {
+    public:
+      // These ones will need to be package intelligently. WilsonType base class
+      // for use by DWF etc..
+      static void DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
+			   std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+			   int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
+      static void DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
+			      std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+			      int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
+
+    };
+
+  }
+}
+#endif
diff --git a/lib/qcd/Grid_qcd_dhop_hand.cc b/lib/qcd/action/fermion/WilsonKernelsHand.cc
similarity index 92%
rename from lib/qcd/Grid_qcd_dhop_hand.cc
rename to lib/qcd/action/fermion/WilsonKernelsHand.cc
index f8d464fb..019b668b 100644
--- a/lib/qcd/Grid_qcd_dhop_hand.cc
+++ b/lib/qcd/action/fermion/WilsonKernelsHand.cc
@@ -27,7 +27,7 @@
     Chi_12 = ref()(1)(2);
 
 #define MULT_2SPIN(A)\
-   auto & ref(U._odata[ss](A));	\
+   auto & ref(U._odata[sU](A));	\
     U_00 = ref()(0,0);\
     U_10 = ref()(1,0);\
     U_20 = ref()(2,0);\
@@ -282,7 +282,7 @@ namespace QCD {
 
 void DiracOptHand::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
 			    std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			    int ss,const LatticeFermion &in, LatticeFermion &out)
+			    int sF,int sU,const LatticeFermion &in, LatticeFermion &out)
 {
   REGISTER vComplex result_00; // 12 regs on knc
   REGISTER vComplex result_01;
@@ -338,7 +338,8 @@ void DiracOptHand::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
 
 
   int offset,local,perm, ptype;
-
+  int ss=sF;
+  
   // Xp
   offset = st._offsets [Xp][ss];
   local  = st._is_local[Xp][ss];
@@ -514,24 +515,24 @@ void DiracOptHand::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
 
   {
     vSpinColourVector & ref (out._odata[ss]);
-    vstream(ref()(0)(0),result_00);
-    vstream(ref()(0)(1),result_01);
-    vstream(ref()(0)(2),result_02);
-    vstream(ref()(1)(0),result_10);
-    vstream(ref()(1)(1),result_11);
-    vstream(ref()(1)(2),result_12);
-    vstream(ref()(2)(0),result_20);
-    vstream(ref()(2)(1),result_21);
-    vstream(ref()(2)(2),result_22);
-    vstream(ref()(3)(0),result_30);
-    vstream(ref()(3)(1),result_31);
-    vstream(ref()(3)(2),result_32);
+    vstream(ref()(0)(0),result_00*(-0.5));
+    vstream(ref()(0)(1),result_01*(-0.5));
+    vstream(ref()(0)(2),result_02*(-0.5));
+    vstream(ref()(1)(0),result_10*(-0.5));
+    vstream(ref()(1)(1),result_11*(-0.5));
+    vstream(ref()(1)(2),result_12*(-0.5));
+    vstream(ref()(2)(0),result_20*(-0.5));
+    vstream(ref()(2)(1),result_21*(-0.5));
+    vstream(ref()(2)(2),result_22*(-0.5));
+    vstream(ref()(3)(0),result_30*(-0.5));
+    vstream(ref()(3)(1),result_31*(-0.5));
+    vstream(ref()(3)(2),result_32*(-0.5));
   }
 }
 
 void DiracOptHand::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
 			       std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			       int ss,const LatticeFermion &in, LatticeFermion &out)
+			       int ss,int sU,const LatticeFermion &in, LatticeFermion &out)
 {
   REGISTER vComplex result_00; // 12 regs on knc
   REGISTER vComplex result_01;
@@ -752,18 +753,18 @@ void DiracOptHand::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
 
   {
     vSpinColourVector & ref (out._odata[ss]);
-    vstream(ref()(0)(0),result_00);
-    vstream(ref()(0)(1),result_01);
-    vstream(ref()(0)(2),result_02);
-    vstream(ref()(1)(0),result_10);
-    vstream(ref()(1)(1),result_11);
-    vstream(ref()(1)(2),result_12);
-    vstream(ref()(2)(0),result_20);
-    vstream(ref()(2)(1),result_21);
-    vstream(ref()(2)(2),result_22);
-    vstream(ref()(3)(0),result_30);
-    vstream(ref()(3)(1),result_31);
-    vstream(ref()(3)(2),result_32);
+    vstream(ref()(0)(0),result_00*(-0.5));
+    vstream(ref()(0)(1),result_01*(-0.5));
+    vstream(ref()(0)(2),result_02*(-0.5));
+    vstream(ref()(1)(0),result_10*(-0.5));
+    vstream(ref()(1)(1),result_11*(-0.5));
+    vstream(ref()(1)(2),result_12*(-0.5));
+    vstream(ref()(2)(0),result_20*(-0.5));
+    vstream(ref()(2)(1),result_21*(-0.5));
+    vstream(ref()(2)(2),result_22*(-0.5));
+    vstream(ref()(3)(0),result_30*(-0.5));
+    vstream(ref()(3)(1),result_31*(-0.5));
+    vstream(ref()(3)(2),result_32*(-0.5));
   }
 }
 }}
diff --git a/lib/stencil/Grid_lebesgue.cc b/lib/stencil/Grid_lebesgue.cc
new file mode 100644
index 00000000..977e3562
--- /dev/null
+++ b/lib/stencil/Grid_lebesgue.cc
@@ -0,0 +1,103 @@
+#include <Grid.h>
+
+namespace Grid {
+
+int LebesgueOrder::UseLebesgueOrder;
+
+LebesgueOrder::IndexInteger LebesgueOrder::alignup(IndexInteger n){
+  n--;           // 1000 0011 --> 1000 0010
+  n |= n >> 1;   // 1000 0010 | 0100 0001 = 1100 0011
+  n |= n >> 2;   // 1100 0011 | 0011 0000 = 1111 0011
+  n |= n >> 4;   // 1111 0011 | 0000 1111 = 1111 1111
+  n |= n >> 8;   // ... (At this point all bits are 1, so further bitwise-or
+  n |= n >> 16;  //      operations produce no effect.)
+  n++;           // 1111 1111 --> 1 0000 0000
+  return n;
+};
+
+LebesgueOrder::LebesgueOrder(GridBase *grid) 
+{
+  _LebesgueReorder.resize(0);
+  
+  // Align up dimensions to power of two.
+  const IndexInteger one=1;
+  IndexInteger ND = grid->_ndimension;
+  std::vector<IndexInteger> dims(ND);
+  std::vector<IndexInteger> adims(ND);
+  std::vector<std::vector<IndexInteger> > bitlist(ND);
+  
+  for(IndexInteger mu=0;mu<ND;mu++){
+    dims[mu] = grid->_rdimensions[mu];
+    assert ( dims[mu] != 0 );
+    adims[mu] = alignup(dims[mu]);
+  }
+  
+  // List which bits of padded volume coordinate contribute; this strategy 
+  // i) avoids recursion 
+  // ii) has loop lengths at most the width of a 32 bit word.
+  int sitebit=0;
+  int split=2;
+  for(int mu=0;mu<ND;mu++){   // mu 0 takes bit 0; mu 1 takes bit 1 etc...
+    for(int bit=0;bit<split;bit++){
+      IndexInteger mask = one<<bit;
+      if ( mask&(adims[mu]-1) ){
+	bitlist[mu].push_back(sitebit);
+	sitebit++;
+      }
+    }
+  }
+  for(int bit=split;bit<32;bit++){
+    IndexInteger mask = one<<bit;
+    for(int mu=0;mu<ND;mu++){   // mu 0 takes bit 0; mu 1 takes bit 1 etc...
+      if ( mask&(adims[mu]-1) ){
+	bitlist[mu].push_back(sitebit);
+	sitebit++;
+      }
+    }
+  }
+  
+  // Work out padded and unpadded volumes
+  IndexInteger avol = 1;
+  for(int mu=0;mu<ND;mu++) avol = avol * adims[mu];
+  
+  IndexInteger vol = 1;
+  for(int mu=0;mu<ND;mu++) vol = vol * dims[mu];
+  
+  // Loop over padded volume, following Lebesgue curve
+  // We interleave the bits from sequential "mu".
+  std::vector<IndexInteger> ax(ND);
+  
+  for(IndexInteger asite=0;asite<avol;asite++){
+    
+    // Start with zero and collect bits
+    for(int mu=0;mu<ND;mu++) ax[mu] = 0;
+    
+    int contained = 1;
+    for(int mu=0;mu<ND;mu++){
+      
+      // Build the coordinate on the aligned volume
+      for(int bit=0;bit<bitlist[mu].size();bit++){
+	int sbit=bitlist[mu][bit];
+	
+	if(asite&(one<<sbit)){
+	  ax[mu]|=one<<bit;
+	}
+      }
+      
+      // Is it contained in original box
+      if ( ax[mu]>dims[mu]-1 ) contained = 0;
+      
+    }
+    
+    if ( contained ) {
+      int site = ax[0]
+	+        dims[0]*ax[1]
+	+dims[0]*dims[1]*ax[2]
+	+dims[0]*dims[1]*dims[2]*ax[3];
+      
+      _LebesgueReorder.push_back(site);
+    }
+	}
+  assert( _LebesgueReorder.size() == vol );
+}
+}
diff --git a/lib/stencil/Grid_lebesgue.h b/lib/stencil/Grid_lebesgue.h
new file mode 100644
index 00000000..1d59b127
--- /dev/null
+++ b/lib/stencil/Grid_lebesgue.h
@@ -0,0 +1,29 @@
+#ifndef GRID_LEBESGUE_H
+#define GRID_LEBESGUE_H
+
+#include<vector>
+
+// Lebesgue, Morton, Z-graph ordering assistance
+namespace Grid {
+  
+  class LebesgueOrder { 
+  public:
+
+    static int UseLebesgueOrder;
+
+    typedef uint32_t IndexInteger;
+
+    inline IndexInteger Reorder(IndexInteger ss) { 
+      return UseLebesgueOrder ? _LebesgueReorder[ss] : ss; 
+    };
+
+    IndexInteger alignup(IndexInteger n);
+
+    LebesgueOrder(GridBase *grid);
+
+  private:
+    std::vector<IndexInteger> _LebesgueReorder;
+
+  };    
+}
+#endif
diff --git a/lib/stencil/Grid_stencil_common.cc b/lib/stencil/Grid_stencil_common.cc
index 6cbcc890..f0f8c581 100644
--- a/lib/stencil/Grid_stencil_common.cc
+++ b/lib/stencil/Grid_stencil_common.cc
@@ -3,95 +3,6 @@
 namespace Grid {
 
 
-
-void CartesianStencil::LebesgueOrder(void) 
-{
-  _LebesgueReorder.resize(0);
-  
-  // Align up dimensions to power of two.
-  const StencilInteger one=1;
-  StencilInteger ND = _grid->_ndimension;
-  std::vector<StencilInteger> dims(ND);
-  std::vector<StencilInteger> adims(ND);
-  std::vector<std::vector<StencilInteger> > bitlist(ND);
-
-
-  for(StencilInteger mu=0;mu<ND;mu++){
-    dims[mu] = _grid->_rdimensions[mu];
-    assert ( dims[mu] != 0 );
-    adims[mu] = alignup(dims[mu]);
-  }
-
-  // List which bits of padded volume coordinate contribute; this strategy 
-  // i) avoids recursion 
-  // ii) has loop lengths at most the width of a 32 bit word.
-  int sitebit=0;
-  int split=24;
-  for(int mu=0;mu<ND;mu++){   // mu 0 takes bit 0; mu 1 takes bit 1 etc...
-    for(int bit=0;bit<split;bit++){
-    StencilInteger mask = one<<bit;
-      if ( mask&(adims[mu]-1) ){
-	bitlist[mu].push_back(sitebit);
-	sitebit++;
-      }
-    }
-  }
-  for(int bit=split;bit<32;bit++){
-    StencilInteger mask = one<<bit;
-    for(int mu=0;mu<ND;mu++){   // mu 0 takes bit 0; mu 1 takes bit 1 etc...
-      if ( mask&(adims[mu]-1) ){
-	bitlist[mu].push_back(sitebit);
-	sitebit++;
-      }
-    }
-  }
-
-  // Work out padded and unpadded volumes
-  StencilInteger avol = 1;
-  for(int mu=0;mu<ND;mu++) avol = avol * adims[mu];
-
-  StencilInteger vol = 1;
-  for(int mu=0;mu<ND;mu++) vol = vol * dims[mu];
-  
-  // Loop over padded volume, following Lebesgue curve
-  // We interleave the bits from sequential "mu".
-  std::vector<StencilInteger> ax(ND);
-  
-  for(StencilInteger asite=0;asite<avol;asite++){
-
-    // Start with zero and collect bits
-    for(int mu=0;mu<ND;mu++) ax[mu] = 0;
-
-    int contained = 1;
-    for(int mu=0;mu<ND;mu++){
-
-      // Build the coordinate on the aligned volume
-      for(int bit=0;bit<bitlist[mu].size();bit++){
-	int sbit=bitlist[mu][bit];
-
-	if(asite&(one<<sbit)){
-	  ax[mu]|=one<<bit;
-	}
-      }
-
-      // Is it contained in original box
-      if ( ax[mu]>dims[mu]-1 ) contained = 0;
-
-    }
-
-    if ( contained ) {
-      int site = ax[0]
-	+        dims[0]*ax[1]
-        +dims[0]*dims[1]*ax[2]
-        +dims[0]*dims[1]*dims[2]*ax[3];
-
-      _LebesgueReorder.push_back(site);
-    }
-  }
-
-  assert( _LebesgueReorder.size() == vol );
-}
-
   CartesianStencil::CartesianStencil(GridBase *grid,
 				     int npoints,
 				     int checkerboard,
@@ -110,8 +21,6 @@ void CartesianStencil::LebesgueOrder(void)
       _unified_buffer_size=0;
       _request_count =0;
 
-      LebesgueOrder();
-
       int osites  = _grid->oSites();
 
       for(int i=0;i<npoints;i++){
@@ -143,8 +52,8 @@ void CartesianStencil::LebesgueOrder(void)
 	// up a table containing the npoint "neighbours" and whether they 
 	// live in lattice or a comms buffer.
 	if ( !comm_dim ) {
-	  sshift[0] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,0);
-	  sshift[1] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,1);
+	  sshift[0] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,Even);
+	  sshift[1] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,Odd);
 
 	  if ( sshift[0] == sshift[1] ) {
 	    Local(point,dimension,shift,0x3);
@@ -154,8 +63,8 @@ void CartesianStencil::LebesgueOrder(void)
 	  }
 	} else { // All permute extract done in comms phase prior to Stencil application
 	  //        So tables are the same whether comm_dim or splice_dim
-	  sshift[0] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,0);
-	  sshift[1] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,1);
+	  sshift[0] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,Even);
+	  sshift[1] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,Odd);
 	  if ( sshift[0] == sshift[1] ) {
 	    Comms(point,dimension,shift,0x3);
 	  } else {
@@ -185,7 +94,7 @@ void CartesianStencil::LebesgueOrder(void)
 	int o   = 0;
 	int bo  = x * _grid->_ostride[dimension];
 	
-	int cb= (cbmask==0x2)? 1 : 0;
+	int cb= (cbmask==0x2)? Odd : Even;
 	  
 	int sshift = _grid->CheckerBoardShift(_checkerboard,dimension,shift,cb);
 	int sx     = (x+sshift)%rd;
@@ -224,7 +133,7 @@ void CartesianStencil::LebesgueOrder(void)
       _comm_buf_size[point] = buffer_size; // Size of _one_ plane. Multiple planes may be gathered and
                                            // send to one or more remote nodes.
 
-      int cb= (cbmask==0x2)? 1 : 0;
+      int cb= (cbmask==0x2)? Odd : Even;
       int sshift= _grid->CheckerBoardShift(_checkerboard,dimension,shift,cb);
       
       for(int x=0;x<rd;x++){       
diff --git a/scripts/linecount b/scripts/linecount
index c0731b23..350abf1c 100755
--- a/scripts/linecount
+++ b/scripts/linecount
@@ -1,3 +1,3 @@
 #!/bin/sh 
 
-wc -l */*.h */*/*.h */*/*/*.h */*.cc */*/*.cc */*/*/*.cc
\ No newline at end of file
+wc -l lib/*.h lib/*/*.h lib/*/*/*.h lib/*.cc lib/*/*.cc lib/*/*/*.cc tests/*.cc benchmarks/*.cc
\ No newline at end of file
diff --git a/tests/Grid_cshift_red_black.cc b/tests/Grid_cshift_red_black.cc
new file mode 100644
index 00000000..212e9b6e
--- /dev/null
+++ b/tests/Grid_cshift_red_black.cc
@@ -0,0 +1,165 @@
+#include <Grid.h>
+#include <parallelIO/GridNerscIO.h>
+
+using namespace Grid;
+using namespace Grid::QCD;
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  int Nd = latt_size.size();
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+
+  std::vector<int> mask(Nd,1);
+  mask[0]=0;
+
+  GridCartesian         Fine  (latt_size,simd_layout,mpi_layout);
+  GridRedBlackCartesian RBFine(latt_size,simd_layout,mpi_layout,mask,1);
+
+  GridParallelRNG      FineRNG(&Fine);  FineRNG.SeedRandomDevice();
+
+  LatticeComplex U(&Fine);
+  LatticeComplex ShiftU(&Fine);
+  LatticeComplex rbShiftU(&Fine);
+  LatticeComplex Ue(&RBFine); 
+  LatticeComplex Uo(&RBFine);
+  LatticeComplex ShiftUe(&RBFine);
+  LatticeComplex ShiftUo(&RBFine);
+  LatticeComplex lex(&Fine);
+  lex=zero;
+  Integer stride =1;
+  {
+    double nrm;
+    LatticeComplex coor(&Fine);
+
+    for(int d=0;d<Nd;d++){
+      //      Integer i=10000;
+      Integer i=0;
+      LatticeCoordinate(coor,d);
+      lex = lex + coor*stride+i;
+      stride=stride*latt_size[d];
+    }
+    U=lex;
+  }
+
+  pickCheckerboard(Even,Ue,U);
+  pickCheckerboard(Odd,Uo,U);
+
+  //  std::cout << U<<std::endl;
+  std::cout << "Ue " <<norm2(Ue)<<std::endl;
+  std::cout << "Uo " <<norm2(Uo)<<std::endl;
+
+
+  TComplex cm;
+  for(int dir=0;dir<Nd;dir++){
+    if ( dir!=1 ) continue;
+    for(int shift=0;shift<latt_size[dir];shift++){
+
+	std::cout<<"Shifting by "<<shift<<" in direction"<<dir<<std::endl;
+
+	//	std::cout<<"Even grid"<<std::endl;
+	ShiftUe = Cshift(Ue,dir,shift);    // Shift everything cb by cb
+	//	std::cout << "\tShiftUe " <<norm2(ShiftUe)<<std::endl;
+
+	//	std::cout<<"Odd grid"<<std::endl;
+	ShiftUo = Cshift(Uo,dir,shift);    
+	//	std::cout << "\tShiftUo " <<norm2(ShiftUo)<<std::endl;
+
+	//	std::cout<<"Recombined Even/Odd grids"<<std::endl;
+	setCheckerboard(rbShiftU,ShiftUe);
+	setCheckerboard(rbShiftU,ShiftUo);
+	//	std::cout << "\trbShiftU " <<norm2(rbShiftU)<<std::endl;
+
+	//	std::cout<<"Full grid shift"<<std::endl;
+	ShiftU  = Cshift(U,dir,shift);    // Shift everything
+	//	std::cout << "\tShiftU " <<norm2(rbShiftU)<<std::endl;
+
+	std::vector<int> coor(4);
+
+	std::cout << "Checking the non-checkerboard shift"<<std::endl;
+	for(coor[3]=0;coor[3]<latt_size[3];coor[3]++){
+	for(coor[2]=0;coor[2]<latt_size[2];coor[2]++){
+	for(coor[1]=0;coor[1]<latt_size[1];coor[1]++){
+	for(coor[0]=0;coor[0]<latt_size[0];coor[0]++){
+	  
+	  peekSite(cm,ShiftU,coor);
+
+	  /////////	  double nrm=norm2(U);
+
+	  std::vector<int> scoor(coor);
+	  scoor[dir] = (scoor[dir]+shift)%latt_size[dir];
+	  
+	  Integer slex = scoor[0]
+	    + latt_size[0]*scoor[1]
+	    + latt_size[0]*latt_size[1]*scoor[2]
+	    + latt_size[0]*latt_size[1]*latt_size[2]*scoor[3];
+
+	  Complex scm(slex);
+	  
+	  double nrm = abs(scm-cm()()());
+	  std::vector<int> peer(4);
+	  int index=real(cm);
+	  Fine.CoorFromIndex(peer,index,latt_size);
+
+	  if (nrm > 0){
+	    std::cerr<<"FAIL shift "<< shift<<" in dir "<< dir
+		     <<" ["<<coor[0]<<","<<coor[1]<<","<<coor[2]<<","<<coor[3]<<"] = "
+		     << cm()()()<<" expect "<<scm<<"  "<<nrm<<std::endl;
+	    std::cerr<<"Got    "<<index<<" " << peer[0]<<","<<peer[1]<<","<<peer[2]<<","<<peer[3]<<std::endl;
+	    index=real(scm);
+	    Fine.CoorFromIndex(peer,index,latt_size);
+	    std::cerr<<"Expect "<<index<<" " << peer[0]<<","<<peer[1]<<","<<peer[2]<<","<<peer[3]<<std::endl;
+	    exit(-1);
+	  }
+	}}}}
+
+
+	std::cout << "Checking the checkerboard shift"<<std::endl;
+	for(coor[3]=0;coor[3]<latt_size[3];coor[3]++){
+	for(coor[2]=0;coor[2]<latt_size[2];coor[2]++){
+	for(coor[1]=0;coor[1]<latt_size[1];coor[1]++){
+	for(coor[0]=0;coor[0]<latt_size[0];coor[0]++){
+	  
+	  peekSite(cm,rbShiftU,coor);
+
+	  double nrm=norm2(U);
+
+	  std::vector<int> scoor(coor);
+	  scoor[dir] = (scoor[dir]+shift)%latt_size[dir];
+	  
+	  Integer slex = scoor[0]
+	    + latt_size[0]*scoor[1]
+	    + latt_size[0]*latt_size[1]*scoor[2]
+	    + latt_size[0]*latt_size[1]*latt_size[2]*scoor[3];
+
+	  Complex scm(slex);
+	  
+	  nrm = abs(scm-cm()()());
+	  std::vector<int> peer(4);
+	  int index=real(cm);
+	  Fine.CoorFromIndex(peer,index,latt_size);
+
+	  if (nrm > 0){
+	    std::cerr<<"FAIL shift "<< shift<<" in dir "<< dir
+		     <<" ["<<coor[0]<<","<<coor[1]<<","<<coor[2]<<","<<coor[3]<<"] = "
+		     << cm()()()<<" expect "<<scm<<"  "<<nrm<<std::endl;
+	    std::cerr<<"Got    "<<index<<" " << peer[0]<<","<<peer[1]<<","<<peer[2]<<","<<peer[3]<<std::endl;
+	    index=real(scm);
+	    Fine.CoorFromIndex(peer,index,latt_size);
+	    std::cerr<<"Expect "<<index<<" " << peer[0]<<","<<peer[1]<<","<<peer[2]<<","<<peer[3]<<std::endl;
+	    exit(-1);
+	  } else if (0) { 
+	    std::cout<<"PASS shift "<< shift<<" in dir "<< dir
+		     <<" ["<<coor[0]<<","<<coor[1]<<","<<coor[2]<<","<<coor[3]<<"] = "
+		     << cm()()()<<" expect "<<scm<<"  "<<nrm<<std::endl;
+	  }
+	}}}}
+
+    }
+  }
+
+  Grid_finalize();
+}
diff --git a/tests/Makefile.am b/tests/Makefile.am
index 3c79191c..7ef4eb1f 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -5,7 +5,7 @@ AM_LDFLAGS = -L$(top_builddir)/lib
 #
 # Test code
 #
-bin_PROGRAMS = Grid_main Grid_stencil Grid_nersc_io Grid_cshift Grid_gamma  Grid_simd Grid_rng Grid_remez Grid_rng_fixed 
+bin_PROGRAMS = Grid_main Grid_stencil Grid_nersc_io Grid_cshift Grid_gamma  Grid_simd Grid_rng Grid_remez Grid_rng_fixed Grid_cshift_red_black 
 
 Grid_main_SOURCES = Grid_main.cc
 Grid_main_LDADD = -lGrid
@@ -25,6 +25,9 @@ Grid_nersc_io_LDADD = -lGrid
 Grid_cshift_SOURCES = Grid_cshift.cc
 Grid_cshift_LDADD = -lGrid
 
+Grid_cshift_red_black_SOURCES = Grid_cshift_red_black.cc
+Grid_cshift_red_black_LDADD = -lGrid
+
 Grid_gamma_SOURCES = Grid_gamma.cc
 Grid_gamma_LDADD = -lGrid
 

From 8272e15bd65e439864b91877d69641c1fb5a1062 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sun, 31 May 2015 15:11:09 +0100
Subject: [PATCH 251/429] Updated line counter

---
 scripts/linecount | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/scripts/linecount b/scripts/linecount
index 350abf1c..7624c15f 100755
--- a/scripts/linecount
+++ b/scripts/linecount
@@ -1,3 +1,3 @@
 #!/bin/sh 
 
-wc -l lib/*.h lib/*/*.h lib/*/*/*.h lib/*.cc lib/*/*.cc lib/*/*/*.cc tests/*.cc benchmarks/*.cc
\ No newline at end of file
+wc -l lib/*.h lib/*/*.h lib/*/*/*.h lib/*.cc lib/*/*.cc lib/*/*/*.cc tests/*.cc benchmarks/*.cc lib/*/*/*/*.cc lib/*/*/*/*.h
\ No newline at end of file

From ce8c7a77b60ffa2d418097663125fc61163bf018 Mon Sep 17 00:00:00 2001
From: azusayamaguchi <ayamaguc@YAMAKAZE.local>
Date: Sun, 31 May 2015 18:50:08 +0100
Subject: [PATCH 252/429] Bug in Makefile.am fixed

---
 lib/Makefile.am | 1 -
 1 file changed, 1 deletion(-)

diff --git a/lib/Makefile.am b/lib/Makefile.am
index aa531df7..e983b432 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -103,7 +103,6 @@ nobase_include_HEADERS=\
 		./qcd/Dirac.h\
 		./qcd/QCD.h\
 		./qcd/TwoSpinor.h\
-		./qcd/FermionAction.h\
 		./simd/Grid_avx.h\
 		./simd/Grid_avx512.h\
 		./simd/Grid_qpx.h\

From 9ea64767b05bc4c468da089e2c49bbc6813c2fa5 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 31 May 2015 22:50:03 +0100
Subject: [PATCH 253/429] No compile fix on mpi target

---
 lib/cshift/Grid_cshift_mpi.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/cshift/Grid_cshift_mpi.h b/lib/cshift/Grid_cshift_mpi.h
index 8c0badcd..3b3fad29 100644
--- a/lib/cshift/Grid_cshift_mpi.h
+++ b/lib/cshift/Grid_cshift_mpi.h
@@ -17,7 +17,7 @@ template<class vobj> Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int s
   // Map to always positive shift modulo global full dimension.
   shift = (shift+fd)%fd;
 
-  ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift);
+  ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift,dimension);
         
   // the permute type
   int simd_layout     = rhs._grid->_simd_layout[dimension];

From 4c617c36430977fecec6366104e3d3b27d69caf0 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Mon, 1 Jun 2015 12:25:59 +0100
Subject: [PATCH 254/429] Const safety

---
 lib/cshift/Grid_cshift_common.h |  8 ++++----
 lib/cshift/Grid_cshift_mpi.h    | 10 +++++-----
 lib/cshift/Grid_cshift_none.h   |  2 +-
 3 files changed, 10 insertions(+), 10 deletions(-)

diff --git a/lib/cshift/Grid_cshift_common.h b/lib/cshift/Grid_cshift_common.h
index 06e812d9..ffa3f771 100644
--- a/lib/cshift/Grid_cshift_common.h
+++ b/lib/cshift/Grid_cshift_common.h
@@ -153,7 +153,7 @@ PARALLEL_NESTED_LOOP2
 //////////////////////////////////////////////////////
 // local to node block strided copies
 //////////////////////////////////////////////////////
-template<class vobj> void Copy_plane(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int dimension,int lplane,int rplane,int cbmask)
+template<class vobj> void Copy_plane(Lattice<vobj>& lhs,const Lattice<vobj> &rhs, int dimension,int lplane,int rplane,int cbmask)
 {
   int rd = rhs._grid->_rdimensions[dimension];
 
@@ -182,7 +182,7 @@ PARALLEL_NESTED_LOOP2
   
 }
 
-template<class vobj> void Copy_plane_permute(Lattice<vobj>& lhs,Lattice<vobj> &rhs, int dimension,int lplane,int rplane,int cbmask,int permute_type)
+template<class vobj> void Copy_plane_permute(Lattice<vobj>& lhs,const Lattice<vobj> &rhs, int dimension,int lplane,int rplane,int cbmask,int permute_type)
 {
  
   int rd = rhs._grid->_rdimensions[dimension];
@@ -212,7 +212,7 @@ PARALLEL_NESTED_LOOP2
 //////////////////////////////////////////////////////
 // Local to node Cshift
 //////////////////////////////////////////////////////
-template<class vobj> void Cshift_local(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimension,int shift)
+template<class vobj> void Cshift_local(Lattice<vobj>& ret,const Lattice<vobj> &rhs,int dimension,int shift)
 {
   int sshift[2];
 
@@ -227,7 +227,7 @@ template<class vobj> void Cshift_local(Lattice<vobj>& ret,Lattice<vobj> &rhs,int
   }
 }
 
-template<class vobj> Lattice<vobj> Cshift_local(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int shift,int cbmask)
+template<class vobj> Lattice<vobj> Cshift_local(Lattice<vobj> &ret,const Lattice<vobj> &rhs,int dimension,int shift,int cbmask)
 {
   GridBase *grid = rhs._grid;
   int fd = grid->_fdimensions[dimension];
diff --git a/lib/cshift/Grid_cshift_mpi.h b/lib/cshift/Grid_cshift_mpi.h
index 3b3fad29..9cdfd316 100644
--- a/lib/cshift/Grid_cshift_mpi.h
+++ b/lib/cshift/Grid_cshift_mpi.h
@@ -4,7 +4,7 @@
 
 namespace Grid { 
 
-template<class vobj> Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
+template<class vobj> Lattice<vobj> Cshift(const Lattice<vobj> &rhs,int dimension,int shift)
 {
   typedef typename vobj::vector_type vector_type;
   typedef typename vobj::scalar_type scalar_type;
@@ -35,7 +35,7 @@ template<class vobj> Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int s
   return ret;
 }
 
-template<class vobj> void Cshift_comms(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimension,int shift)
+template<class vobj> void Cshift_comms(Lattice<vobj>& ret,const Lattice<vobj> &rhs,int dimension,int shift)
 {
   int sshift[2];
 
@@ -50,7 +50,7 @@ template<class vobj> void Cshift_comms(Lattice<vobj>& ret,Lattice<vobj> &rhs,int
   }
 }
 
-template<class vobj> void Cshift_comms_simd(Lattice<vobj>& ret,Lattice<vobj> &rhs,int dimension,int shift)
+template<class vobj> void Cshift_comms_simd(Lattice<vobj>& ret,const Lattice<vobj> &rhs,int dimension,int shift)
 {
   int sshift[2];
 
@@ -65,7 +65,7 @@ template<class vobj> void Cshift_comms_simd(Lattice<vobj>& ret,Lattice<vobj> &rh
   }
 }
 
-template<class vobj> void Cshift_comms(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int shift,int cbmask)
+template<class vobj> void Cshift_comms(Lattice<vobj> &ret,const Lattice<vobj> &rhs,int dimension,int shift,int cbmask)
 {
   typedef typename vobj::vector_type vector_type;
   typedef typename vobj::scalar_type scalar_type;
@@ -124,7 +124,7 @@ template<class vobj> void Cshift_comms(Lattice<vobj> &ret,Lattice<vobj> &rhs,int
   }
 }
 
-template<class vobj> void  Cshift_comms_simd(Lattice<vobj> &ret,Lattice<vobj> &rhs,int dimension,int shift,int cbmask)
+template<class vobj> void  Cshift_comms_simd(Lattice<vobj> &ret,const Lattice<vobj> &rhs,int dimension,int shift,int cbmask)
 {
   GridBase *grid=rhs._grid;
   const int Nsimd = grid->Nsimd();
diff --git a/lib/cshift/Grid_cshift_none.h b/lib/cshift/Grid_cshift_none.h
index be9467b1..a8c6f6c0 100644
--- a/lib/cshift/Grid_cshift_none.h
+++ b/lib/cshift/Grid_cshift_none.h
@@ -1,7 +1,7 @@
 #ifndef _GRID_CSHIFT_NONE_H_
 #define _GRID_CSHIFT_NONE_H_
 namespace Grid {
-template<class vobj> Lattice<vobj> Cshift(Lattice<vobj> &rhs,int dimension,int shift)
+template<class vobj> Lattice<vobj> Cshift(const Lattice<vobj> &rhs,int dimension,int shift)
 {
   Lattice<vobj> ret(rhs._grid);
   ret.checkerboard = rhs._grid->CheckerBoardDestination(rhs.checkerboard,shift,dimension);

From 8bd9fb4427350dcf900825da19a630ba6e89ad46 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Mon, 1 Jun 2015 12:26:20 +0100
Subject: [PATCH 255/429] FIx miistake

---
 lib/lattice/Grid_lattice_base.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/lattice/Grid_lattice_base.h b/lib/lattice/Grid_lattice_base.h
index 4a6d3180..15ef0944 100644
--- a/lib/lattice/Grid_lattice_base.h
+++ b/lib/lattice/Grid_lattice_base.h
@@ -176,7 +176,7 @@ PARALLEL_FOR_LOOP
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
 #ifdef STREAMING_STORES
-      vobj tmp = eval(tmp,ss,expr);
+      vobj tmp = eval(ss,expr);
       vstream(_odata[ss] ,tmp);
 #else
       _odata[ss]=eval(ss,expr);

From 2583570e17e95591d0f9dac4bddf18d3c1f6260b Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 2 Jun 2015 16:57:12 +0100
Subject: [PATCH 256/429] Domain wall fermions now invert ; have the basis set
 up for Tanh/Zolo * (Cayley/PartFrac/ContFrac) * (Mobius/Shamir/Wilson) Approx
        Representation               Kernel.

All are done with space-time taking part in checkerboarding, Ls uncheckerboarded

Have only so far tested the Domain Wall limit of mobius, and at that only checked
that it
i)  Inverts
ii) 5dim DW == Ls copies of 4dim D2
iii) MeeInv Mee == 1
iv) Meo+Mee+Moe+Moo == M unprec.
v) MpcDagMpc is hermitan
vi) Mdag is the adjoint of M between stochastic vectors.

That said, the RB schur solve, RB MpcDagMpc solve, Unprec solve
all converge and the true residual becomes small; so pretty good tests.
---
 benchmarks/Grid_dwf.cc                        |  79 +++---
 benchmarks/Grid_dwf_cg_prec.cc                |  58 +++++
 benchmarks/Grid_dwf_cg_schur.cc               |  53 ++++
 benchmarks/Grid_dwf_cg_unprec.cc              |  53 ++++
 benchmarks/Grid_dwf_even_odd.cc               | 207 +++++++++++++++
 benchmarks/Grid_wilson_evenodd.cc             |   5 +-
 benchmarks/Makefile.am                        |  46 +++-
 lib/Grid_init.cc                              |   2 +-
 lib/Makefile.am                               |  16 +-
 lib/algorithms/LinearOperator.h               |  32 ---
 lib/algorithms/approx/Zolotarev.cc            |   9 +-
 lib/algorithms/approx/Zolotarev.h             |   9 +-
 lib/cartesian/Grid_cartesian_base.h           |   2 +-
 lib/cartesian/Grid_cartesian_full.h           |   6 +-
 lib/cartesian/Grid_cartesian_red_black.h      |  24 +-
 lib/communicator/Grid_communicator_base.h     |   2 +-
 lib/communicator/Grid_communicator_mpi.cc     |   2 +-
 lib/communicator/Grid_communicator_none.cc    |   2 +-
 lib/qcd/LinalgUtils.h                         | 113 +++++++++
 lib/qcd/QCD.h                                 |   2 +
 lib/qcd/SpaceTimeGrid.cc                      |  52 ++++
 lib/qcd/SpaceTimeGrid.h                       |  18 ++
 lib/qcd/action/Actions.h                      |  78 +++++-
 lib/qcd/action/fermion/CayleyFermion5D.cc     | 235 ++++++++++++++++++
 lib/qcd/action/fermion/CayleyFermion5D.h      |  61 +++++
 .../fermion/ContinuedFractionFermion5D.cc     | 119 +++++++++
 .../fermion/ContinuedFractionFermion5D.h      |  53 ++++
 lib/qcd/action/fermion/DomainWallFermion.h    | 118 +++++++++
 .../{FermionAction.h => FermionOperator.h}    |   7 +-
 .../fermion/PartialFractionFermion5D.cc       |  47 ++++
 .../action/fermion/PartialFractionFermion5D.h |  49 ++++
 lib/qcd/action/fermion/WilsonFermion.cc       |   6 +-
 lib/qcd/action/fermion/WilsonFermion.h        |   6 +-
 ...DimWilsonFermion.cc => WilsonFermion5D.cc} | 106 +++-----
 ...veDimWilsonFermion.h => WilsonFermion5D.h} |  31 +--
 lib/stencil/Grid_stencil_common.cc            |  12 +-
 36 files changed, 1500 insertions(+), 220 deletions(-)
 create mode 100644 benchmarks/Grid_dwf_cg_prec.cc
 create mode 100644 benchmarks/Grid_dwf_cg_schur.cc
 create mode 100644 benchmarks/Grid_dwf_cg_unprec.cc
 create mode 100644 benchmarks/Grid_dwf_even_odd.cc
 create mode 100644 lib/qcd/LinalgUtils.h
 create mode 100644 lib/qcd/SpaceTimeGrid.cc
 create mode 100644 lib/qcd/SpaceTimeGrid.h
 create mode 100644 lib/qcd/action/fermion/CayleyFermion5D.cc
 create mode 100644 lib/qcd/action/fermion/CayleyFermion5D.h
 create mode 100644 lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
 create mode 100644 lib/qcd/action/fermion/ContinuedFractionFermion5D.h
 create mode 100644 lib/qcd/action/fermion/DomainWallFermion.h
 rename lib/qcd/action/fermion/{FermionAction.h => FermionOperator.h} (92%)
 create mode 100644 lib/qcd/action/fermion/PartialFractionFermion5D.cc
 create mode 100644 lib/qcd/action/fermion/PartialFractionFermion5D.h
 rename lib/qcd/action/fermion/{FiveDimWilsonFermion.cc => WilsonFermion5D.cc} (68%)
 rename lib/qcd/action/fermion/{FiveDimWilsonFermion.h => WilsonFermion5D.h} (72%)

diff --git a/benchmarks/Grid_dwf.cc b/benchmarks/Grid_dwf.cc
index eb1d9299..62869b17 100644
--- a/benchmarks/Grid_dwf.cc
+++ b/benchmarks/Grid_dwf.cc
@@ -24,43 +24,28 @@ int main (int argc, char ** argv)
   std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
 
   std::vector<int> latt4 = GridDefaultLatt();
-  std::vector<int> simd4 = GridDefaultSimd(Nd,vComplexF::Nsimd());
-  std::vector<int> mpi4  = GridDefaultMpi();
-
-  assert(latt4.size()==4 ); 
-  assert(simd4.size()==4 );
-  assert(mpi4.size() ==4 );
-
-  const int Ls=1;
-  std::vector<int> latt5({Ls,latt4[0],latt4[1],latt4[2],latt4[3]});
-  std::vector<int> simd5({1 ,simd4[0],simd4[1],simd4[2],simd4[3]}); 
-  std::vector<int>  mpi5({1 , mpi4[0], mpi4[1], mpi4[2], mpi4[3]}); 
-  std::vector<int>   cb5({0,1,1,1,1}); // Checkerboard 4d only
-  int                cbd=1;            // use dim-1 to reduce
-
-  // Four dim grid for gauge field U
-  GridCartesian               UGrid(latt4,simd4,mpi4); 
-  GridRedBlackCartesian     UrbGrid(&UGrid);
-
-  // Five dim grid for fermions F
-  GridCartesian               FGrid(latt5,simd5,mpi5); 
-  GridRedBlackCartesian     FrbGrid(latt5,simd5,mpi5,cb5,cbd); 
+  const int Ls=8;
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
 
   std::vector<int> seeds4({1,2,3,4});
   std::vector<int> seeds5({5,6,7,8});
 
-  GridParallelRNG          RNG5(&FGrid);  RNG5.SeedFixedIntegers(seeds5);
-  LatticeFermion src   (&FGrid); random(RNG5,src);
-  LatticeFermion result(&FGrid); result=zero;
-  LatticeFermion    ref(&FGrid);    ref=zero;
-  LatticeFermion    tmp(&FGrid);
-  LatticeFermion    err(&FGrid);
+  GridParallelRNG          RNG4(UGrid);  RNG4.SeedFixedIntegers(seeds4);
+  GridParallelRNG          RNG5(FGrid);  RNG5.SeedFixedIntegers(seeds5);
+
+  LatticeFermion src   (FGrid); random(RNG5,src);
+  LatticeFermion result(FGrid); result=zero;
+  LatticeFermion    ref(FGrid);    ref=zero;
+  LatticeFermion    tmp(FGrid);
+  LatticeFermion    err(FGrid);
 
   ColourMatrix cm = Complex(1.0,0.0);
 
-  GridParallelRNG          RNG4(&UGrid);  RNG4.SeedFixedIntegers(seeds4);
-  LatticeGaugeField Umu(&UGrid); random(RNG4,Umu);
-  LatticeGaugeField Umu5d(&FGrid); 
+  LatticeGaugeField Umu(UGrid); random(RNG4,Umu);
+  LatticeGaugeField Umu5d(FGrid); 
 
   // replicate across fifth dimension
   for(int ss=0;ss<Umu._grid->oSites();ss++){
@@ -72,7 +57,7 @@ int main (int argc, char ** argv)
   ////////////////////////////////////
   // Naive wilson implementation
   ////////////////////////////////////
-  std::vector<LatticeColourMatrix> U(4,&FGrid);
+  std::vector<LatticeColourMatrix> U(4,FGrid);
   for(int mu=0;mu<Nd;mu++){
     U[mu] = peekIndex<LorentzIndex>(Umu5d,mu);
   }
@@ -93,17 +78,17 @@ int main (int argc, char ** argv)
   }
 
   RealD mass=0.1;
-  FiveDimWilsonFermion Dw(Umu,FGrid,FrbGrid,UGrid,UrbGrid,mass);
+  RealD M5  =1.8;
+  DomainWallFermion Dw(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
   
   std::cout << "Calling Dw"<<std::endl;
-  int ncall=1000;
+  int ncall=10;
   double t0=usecond();
   for(int i=0;i<ncall;i++){
     Dw.Dhop(src,result,0);
   }
   double t1=usecond();
 
-
   double volume=Ls;  for(int mu=0;mu<Nd;mu++) volume=volume*latt4[mu];
   double flops=1344*volume*ncall;
   
@@ -141,5 +126,31 @@ int main (int argc, char ** argv)
   err = ref-result; 
   std::cout << "norm diff   "<< norm2(err)<<std::endl;
 
+  LatticeFermion src_e (FrbGrid);
+  LatticeFermion src_o (FrbGrid);
+  LatticeFermion r_e   (FrbGrid);
+  LatticeFermion r_o   (FrbGrid);
+  LatticeFermion r_eo  (FGrid);
+
+
+  std::cout << "Calling Deo and Doe"<<std::endl;
+  pickCheckerboard(Even,src_e,src);
+  pickCheckerboard(Odd,src_o,src);
+
+  Dw.DhopEO(src_o,r_e,DaggerNo);
+  Dw.DhopOE(src_e,r_o,DaggerNo);
+  Dw.Dhop(src,result,DaggerNo);
+
+  setCheckerboard(r_eo,r_o);
+  setCheckerboard(r_eo,r_e);
+
+  err = r_eo-result; 
+  std::cout << "norm diff   "<< norm2(err)<<std::endl;
+
+  pickCheckerboard(Even,src_e,err);
+  pickCheckerboard(Odd,src_o,err);
+  std::cout << "norm diff even  "<< norm2(src_e)<<std::endl;
+  std::cout << "norm diff odd   "<< norm2(src_o)<<std::endl;
+
   Grid_finalize();
 }
diff --git a/benchmarks/Grid_dwf_cg_prec.cc b/benchmarks/Grid_dwf_cg_prec.cc
new file mode 100644
index 00000000..0cf86a19
--- /dev/null
+++ b/benchmarks/Grid_dwf_cg_prec.cc
@@ -0,0 +1,58 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  const int Ls=8;
+
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);  RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);  RNG4.SeedFixedIntegers(seeds4);
+
+  LatticeFermion    src(FGrid); random(RNG5,src);
+  LatticeFermion result(FGrid); result=zero;
+  LatticeGaugeField Umu(UGrid); random(RNG4,Umu);
+
+  std::vector<LatticeColourMatrix> U(4,UGrid);
+  for(int mu=0;mu<Nd;mu++){
+    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+  }
+  
+  RealD mass=0.1;
+  RealD M5=1.8;
+  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+
+  LatticeFermion    src_o(FrbGrid);
+  LatticeFermion result_o(FrbGrid);
+  pickCheckerboard(Odd,src_o,src);
+  result_o=zero;
+
+  HermitianCheckerBoardedOperator<DomainWallFermion,LatticeFermion> HermOpEO(Ddwf);
+  ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
+  CG(HermOpEO,src_o,result_o);
+
+  Grid_finalize();
+}
diff --git a/benchmarks/Grid_dwf_cg_schur.cc b/benchmarks/Grid_dwf_cg_schur.cc
new file mode 100644
index 00000000..aac4d3fd
--- /dev/null
+++ b/benchmarks/Grid_dwf_cg_schur.cc
@@ -0,0 +1,53 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  const int Ls=8;
+
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);  RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);  RNG4.SeedFixedIntegers(seeds4);
+
+  LatticeFermion    src(FGrid); random(RNG5,src);
+  LatticeFermion result(FGrid); result=zero;
+  LatticeGaugeField Umu(UGrid); random(RNG4,Umu);
+
+  std::vector<LatticeColourMatrix> U(4,UGrid);
+  for(int mu=0;mu<Nd;mu++){
+    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+  }
+  
+  RealD mass=0.1;
+  RealD M5=1.8;
+  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+
+  ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
+  SchurRedBlackSolve<LatticeFermion> SchurSolver(CG);
+  SchurSolver(Ddwf,src,result);
+
+  Grid_finalize();
+}
diff --git a/benchmarks/Grid_dwf_cg_unprec.cc b/benchmarks/Grid_dwf_cg_unprec.cc
new file mode 100644
index 00000000..5c9e7ad3
--- /dev/null
+++ b/benchmarks/Grid_dwf_cg_unprec.cc
@@ -0,0 +1,53 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  const int Ls=8;
+
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);  RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);  RNG4.SeedFixedIntegers(seeds4);
+
+  LatticeFermion    src(FGrid); random(RNG5,src);
+  LatticeFermion result(FGrid); result=zero;
+  LatticeGaugeField Umu(UGrid); random(RNG4,Umu);
+
+  std::vector<LatticeColourMatrix> U(4,UGrid);
+  for(int mu=0;mu<Nd;mu++){
+    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+  }
+  
+  RealD mass=0.1;
+  RealD M5=1.8;
+  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+
+  HermitianOperator<DomainWallFermion,LatticeFermion> HermOp(Ddwf);
+  ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
+  CG(HermOp,src,result);
+
+  Grid_finalize();
+}
diff --git a/benchmarks/Grid_dwf_even_odd.cc b/benchmarks/Grid_dwf_even_odd.cc
new file mode 100644
index 00000000..ac47bbf9
--- /dev/null
+++ b/benchmarks/Grid_dwf_even_odd.cc
@@ -0,0 +1,207 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  int threads = GridThread::GetThreads();
+  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+
+
+  const int Ls=8;
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+
+  GridParallelRNG          RNG4(UGrid);  RNG4.SeedFixedIntegers(seeds4);
+  GridParallelRNG          RNG5(FGrid);  RNG5.SeedFixedIntegers(seeds5);
+
+  LatticeFermion src   (FGrid); random(RNG5,src);
+  LatticeFermion phi   (FGrid); random(RNG5,phi);
+  LatticeFermion chi   (FGrid); random(RNG5,chi);
+  LatticeFermion result(FGrid); result=zero;
+  LatticeFermion    ref(FGrid);    ref=zero;
+  LatticeFermion    tmp(FGrid);    tmp=zero;
+  LatticeFermion    err(FGrid);    tmp=zero;
+  LatticeGaugeField Umu(UGrid); random(RNG4,Umu);
+  std::vector<LatticeColourMatrix> U(4,UGrid);
+
+  // Only one non-zero (y)
+  Umu=zero;
+  for(int nn=0;nn<Nd;nn++){
+    random(RNG4,U[nn]);
+    if ( nn>0 ) 
+      U[nn]=zero;
+    pokeIndex<LorentzIndex>(Umu,U[nn],nn);
+  }
+
+  RealD mass=0.1;
+  RealD M5  =1.8;
+  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+
+  LatticeFermion src_e (FrbGrid);
+  LatticeFermion src_o (FrbGrid);
+  LatticeFermion r_e   (FrbGrid);
+  LatticeFermion r_o   (FrbGrid);
+  LatticeFermion r_eo  (FGrid);
+  LatticeFermion r_eeoo(FGrid);
+
+  std::cout<<"=========================================================="<<std::endl;
+  std::cout<<"= Testing that Meo + Moe + Moo + Mee = Munprec "<<std::endl;
+  std::cout<<"=========================================================="<<std::endl;
+
+  pickCheckerboard(Even,src_e,src);
+  pickCheckerboard(Odd,src_o,src);
+
+  Ddwf.Meooe(src_e,r_o);  std::cout<<"Applied Meo"<<std::endl;
+  Ddwf.Meooe(src_o,r_e);  std::cout<<"Applied Moe"<<std::endl;
+  setCheckerboard(r_eo,r_o);
+  setCheckerboard(r_eo,r_e);
+
+  Ddwf.Mooee(src_e,r_e);  std::cout<<"Applied Mee"<<std::endl;
+  Ddwf.Mooee(src_o,r_o);  std::cout<<"Applied Moo"<<std::endl;
+  setCheckerboard(r_eeoo,r_e);
+  setCheckerboard(r_eeoo,r_o);
+
+  r_eo=r_eo+r_eeoo;
+  Ddwf.M(src,ref);  
+
+  //  std::cout << r_eo<<std::endl;
+  //  std::cout << ref <<std::endl;
+
+  err= ref - r_eo;
+  std::cout << "EO norm diff   "<< norm2(err)<< " "<<norm2(ref)<< " " << norm2(r_eo) <<std::endl;
+    
+  LatticeComplex cerr(FGrid);
+  cerr = localInnerProduct(err,err);
+  //  std::cout << cerr<<std::endl;
+
+  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<"= Test Ddagger is the dagger of D by requiring                "<<std::endl;
+  std::cout<<"=  < phi | Deo | chi > * = < chi | Deo^dag| phi>  "<<std::endl;
+  std::cout<<"=============================================================="<<std::endl;
+  
+  LatticeFermion chi_e   (FrbGrid);
+  LatticeFermion chi_o   (FrbGrid);
+
+  LatticeFermion dchi_e  (FrbGrid);
+  LatticeFermion dchi_o  (FrbGrid);
+
+  LatticeFermion phi_e   (FrbGrid);
+  LatticeFermion phi_o   (FrbGrid);
+
+  LatticeFermion dphi_e  (FrbGrid);
+  LatticeFermion dphi_o  (FrbGrid);
+
+
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+  pickCheckerboard(Even,phi_e,phi);
+  pickCheckerboard(Odd ,phi_o,phi);
+
+  Ddwf.Meooe(chi_e,dchi_o);
+  Ddwf.Meooe(chi_o,dchi_e);
+  Ddwf.MeooeDag(phi_e,dphi_o);
+  Ddwf.MeooeDag(phi_o,dphi_e);
+
+  ComplexD pDce = innerProduct(phi_e,dchi_e);
+  ComplexD pDco = innerProduct(phi_o,dchi_o);
+  ComplexD cDpe = innerProduct(chi_e,dphi_e);
+  ComplexD cDpo = innerProduct(chi_o,dphi_o);
+
+  std::cout <<"e "<<pDce<<" "<<cDpe <<std::endl;
+  std::cout <<"o "<<pDco<<" "<<cDpo <<std::endl;
+
+  std::cout <<"pDce - conj(cDpo) "<< pDce-conj(cDpo) <<std::endl;
+  std::cout <<"pDco - conj(cDpe) "<< pDco-conj(cDpe) <<std::endl;
+
+  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<"= Test MeeInv Mee = 1                                         "<<std::endl;
+  std::cout<<"=============================================================="<<std::endl;
+
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+
+  Ddwf.Mooee(chi_e,src_e);
+  Ddwf.MooeeInv(src_e,phi_e);
+
+  Ddwf.Mooee(chi_o,src_o);
+  Ddwf.MooeeInv(src_o,phi_o);
+  
+  setCheckerboard(phi,phi_e);
+  setCheckerboard(phi,phi_o);
+
+  err = phi-chi;
+  std::cout << "norm diff   "<< norm2(err)<< std::endl;
+
+  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<"= Test MeeInvDag MeeDag = 1                                   "<<std::endl;
+  std::cout<<"=============================================================="<<std::endl;
+
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+
+  Ddwf.MooeeDag(chi_e,src_e);
+  Ddwf.MooeeInvDag(src_e,phi_e);
+
+  Ddwf.MooeeDag(chi_o,src_o);
+  Ddwf.MooeeInvDag(src_o,phi_o);
+  
+  setCheckerboard(phi,phi_e);
+  setCheckerboard(phi,phi_o);
+
+  err = phi-chi;
+  std::cout << "norm diff   "<< norm2(err)<< std::endl;
+
+  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<"= Test MpcDagMpc is Hermitian              "<<std::endl;
+  std::cout<<"=============================================================="<<std::endl;
+  
+  random(RNG5,phi);
+  random(RNG5,chi);
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+  pickCheckerboard(Even,phi_e,phi);
+  pickCheckerboard(Odd ,phi_o,phi);
+  RealD t1,t2;
+
+  Ddwf.MpcDagMpc(chi_e,dchi_e,t1,t2);
+  Ddwf.MpcDagMpc(chi_o,dchi_o,t1,t2);
+
+  Ddwf.MpcDagMpc(phi_e,dphi_e,t1,t2);
+  Ddwf.MpcDagMpc(phi_o,dphi_o,t1,t2);
+
+  pDce = innerProduct(phi_e,dchi_e);
+  pDco = innerProduct(phi_o,dchi_o);
+  cDpe = innerProduct(chi_e,dphi_e);
+  cDpo = innerProduct(chi_o,dphi_o);
+
+  std::cout <<"e "<<pDce<<" "<<cDpe <<std::endl;
+  std::cout <<"o "<<pDco<<" "<<cDpo <<std::endl;
+
+  std::cout <<"pDce - conj(cDpo) "<< pDco-conj(cDpo) <<std::endl;
+  std::cout <<"pDco - conj(cDpe) "<< pDce-conj(cDpe) <<std::endl;
+  
+  Grid_finalize();
+}
diff --git a/benchmarks/Grid_wilson_evenodd.cc b/benchmarks/Grid_wilson_evenodd.cc
index a073139d..3ebc4709 100644
--- a/benchmarks/Grid_wilson_evenodd.cc
+++ b/benchmarks/Grid_wilson_evenodd.cc
@@ -68,8 +68,6 @@ int main (int argc, char ** argv)
   LatticeFermion r_o   (&RBGrid);
   LatticeFermion r_eo  (&Grid);
 
-  const int Even=0;
-  const int Odd=1;
   std::cout<<"=========================================================="<<std::endl;
   std::cout<<"= Testing that Deo + Doe = Dunprec "<<std::endl;
   std::cout<<"=========================================================="<<std::endl;
@@ -79,12 +77,11 @@ int main (int argc, char ** argv)
 
   Dw.Meooe(src_e,r_o);  std::cout<<"Applied Meo"<<std::endl;
   Dw.Meooe(src_o,r_e);  std::cout<<"Applied Moe"<<std::endl;
-  Dw.Dhop (src,ref,0);
+  Dw.Dhop (src,ref,DaggerNo);
 
   setCheckerboard(r_eo,r_o);
   setCheckerboard(r_eo,r_e);
 
-  ref = (-0.5)*ref;
   err= ref - r_eo;
   std::cout << "EO norm diff   "<< norm2(err)<< " "<<norm2(ref)<< " " << norm2(r_eo) <<std::endl;
 
diff --git a/benchmarks/Makefile.am b/benchmarks/Makefile.am
index b2649669..e92f2e84 100644
--- a/benchmarks/Makefile.am
+++ b/benchmarks/Makefile.am
@@ -5,14 +5,34 @@ AM_LDFLAGS = -L$(top_builddir)/lib
 #
 # Test code
 #
-bin_PROGRAMS = Grid_wilson Grid_comms Grid_memory_bandwidth Grid_su3 Grid_wilson_cg_unprec Grid_wilson_evenodd  Grid_wilson_cg_prec Grid_wilson_cg_schur Grid_dwf
+bin_PROGRAMS = \
+	Grid_comms \
+	Grid_memory_bandwidth \
+	Grid_su3 \
+	Grid_wilson \
+	Grid_wilson_evenodd \
+	Grid_wilson_cg_unprec \
+	Grid_wilson_cg_prec \
+	Grid_wilson_cg_schur \
+	Grid_dwf\
+	Grid_dwf_even_odd\
+	Grid_dwf_cg_unprec\
+	Grid_dwf_cg_prec\
+	Grid_dwf_cg_schur
+
+Grid_comms_SOURCES = Grid_comms.cc
+Grid_comms_LDADD = -lGrid
+
+Grid_su3_SOURCES = Grid_su3.cc Grid_su3_test.cc Grid_su3_expr.cc
+Grid_su3_LDADD = -lGrid
+
+Grid_memory_bandwidth_SOURCES = Grid_memory_bandwidth.cc
+Grid_memory_bandwidth_LDADD = -lGrid
+
 
 Grid_wilson_SOURCES = Grid_wilson.cc
 Grid_wilson_LDADD = -lGrid
 
-Grid_dwf_SOURCES = Grid_dwf.cc
-Grid_dwf_LDADD = -lGrid
-
 Grid_wilson_evenodd_SOURCES = Grid_wilson_evenodd.cc
 Grid_wilson_evenodd_LDADD = -lGrid
 
@@ -25,12 +45,18 @@ Grid_wilson_cg_prec_LDADD = -lGrid
 Grid_wilson_cg_schur_SOURCES = Grid_wilson_cg_schur.cc
 Grid_wilson_cg_schur_LDADD = -lGrid
 
-Grid_comms_SOURCES = Grid_comms.cc
-Grid_comms_LDADD = -lGrid
+Grid_dwf_SOURCES = Grid_dwf.cc
+Grid_dwf_LDADD = -lGrid
 
-Grid_su3_SOURCES = Grid_su3.cc Grid_su3_test.cc Grid_su3_expr.cc
-Grid_su3_LDADD = -lGrid
+Grid_dwf_even_odd_SOURCES = Grid_dwf_even_odd.cc
+Grid_dwf_even_odd_LDADD = -lGrid
 
-Grid_memory_bandwidth_SOURCES = Grid_memory_bandwidth.cc
-Grid_memory_bandwidth_LDADD = -lGrid
+Grid_dwf_cg_unprec_SOURCES = Grid_dwf_cg_unprec.cc
+Grid_dwf_cg_unprec_LDADD = -lGrid
+
+Grid_dwf_cg_prec_SOURCES = Grid_dwf_cg_prec.cc
+Grid_dwf_cg_prec_LDADD = -lGrid
+
+Grid_dwf_cg_schur_SOURCES = Grid_dwf_cg_schur.cc
+Grid_dwf_cg_schur_LDADD = -lGrid
 
diff --git a/lib/Grid_init.cc b/lib/Grid_init.cc
index 580e602a..20371a4b 100644
--- a/lib/Grid_init.cc
+++ b/lib/Grid_init.cc
@@ -143,7 +143,7 @@ void Grid_init(int *argc,char ***argv)
   }
   if( GridCmdOptionExists(*argv,*argv+*argc,"--dslash-opt") ){
     WilsonFermion::HandOptDslash=1;
-    FiveDimWilsonFermion::HandOptDslash=1;
+    WilsonFermion5D::HandOptDslash=1;
   }
   if( GridCmdOptionExists(*argv,*argv+*argc,"--lebesgue") ){
     LebesgueOrder::UseLebesgueOrder=1;
diff --git a/lib/Makefile.am b/lib/Makefile.am
index e983b432..b688d8b2 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -20,13 +20,17 @@ libGrid_a_SOURCES =				\
 	stencil/Grid_stencil_common.cc		\
 	algorithms/approx/Zolotarev.cc		\
 	algorithms/approx/Remez.cc		\
-	qcd/action/fermion/FiveDimWilsonFermion.cc\
-	qcd/action/fermion/WilsonFermion.cc\
+	qcd/SpaceTimeGrid.cc\
+	qcd/Dirac.cc\
 	qcd/action/fermion/WilsonKernels.cc\
 	qcd/action/fermion/WilsonKernelsHand.cc\
-	qcd/Dirac.cc\
+	qcd/action/fermion/WilsonFermion.cc\
+	qcd/action/fermion/WilsonFermion5D.cc\
+	qcd/action/fermion/CayleyFermion5D.cc \
+	qcd/action/fermion/ContinuedFractionFermion5D.cc	\
 	$(extra_sources)
 
+#	qcd/action/fermion/PartialFractionFermion5D.cc	\
 #
 # Include files
 #
@@ -95,11 +99,11 @@ nobase_include_HEADERS=\
 		./math/Grid_math_transpose.h\
 		./parallelIO/GridNerscIO.h\
 		./qcd/action/Actions.h\
-		./qcd/action/fermion/FermionAction.h\
-		./qcd/action/fermion/FiveDimWilsonFermion.h\
+		./qcd/action/fermion/FermionOperator.h\
 		./qcd/action/fermion/WilsonCompressor.h\
-		./qcd/action/fermion/WilsonFermion.h\
 		./qcd/action/fermion/WilsonKernels.h\
+		./qcd/action/fermion/WilsonFermion.h\
+		./qcd/action/fermion/WilsonFermion5D.h\
 		./qcd/Dirac.h\
 		./qcd/QCD.h\
 		./qcd/TwoSpinor.h\
diff --git a/lib/algorithms/LinearOperator.h b/lib/algorithms/LinearOperator.h
index cb6240ba..5839f2c1 100644
--- a/lib/algorithms/LinearOperator.h
+++ b/lib/algorithms/LinearOperator.h
@@ -125,39 +125,7 @@ namespace Grid {
      };
     */
 
-    // Chroma interface defining GaugeAction
-    /*
-      template<typename P, typename Q>   class GaugeAction
-  virtual const CreateGaugeState<P,Q>& getCreateState() const = 0;
-  virtual GaugeState<P,Q>* createState(const Q& q) const
-  virtual const GaugeBC<P,Q>& getGaugeBC() const
-  virtual const Set& getSet(void) const = 0;
-  virtual void deriv(P& result, const Handle< GaugeState<P,Q> >& state) const 
-  virtual Double S(const Handle< GaugeState<P,Q> >& state) const = 0;
 
-  class LinearGaugeAction : public GaugeAction< multi1d<LatticeColorMatrix>, multi1d<LatticeColorMatrix> >
-  typedef multi1d<LatticeColorMatrix>  P;
-  typedef multi1d<LatticeColorMatrix>  Q;
-  virtual void staple(LatticeColorMatrix& result,
-		      const Handle< GaugeState<P,Q> >& state,
-		      int mu, int cb) const = 0;
-    */
-
-    // Chroma interface defining FermionAction
-    /*
-     template<typename T, typename P, typename Q>  class FermAct4D : public FermionAction<T,P,Q>
-     virtual LinearOperator<T>* linOp(Handle< FermState<T,P,Q> > state) const = 0;
-     virtual LinearOperator<T>* lMdagM(Handle< FermState<T,P,Q> > state) const = 0;
-     virtual LinOpSystemSolver<T>* invLinOp(Handle< FermState<T,P,Q> > state,
-     virtual MdagMSystemSolver<T>* invMdagM(Handle< FermState<T,P,Q> > state,
-     virtual LinOpMultiSystemSolver<T>* mInvLinOp(Handle< FermState<T,P,Q> > state,
-     virtual MdagMMultiSystemSolver<T>* mInvMdagM(Handle< FermState<T,P,Q> > state,
-     virtual MdagMMultiSystemSolverAccumulate<T>* mInvMdagMAcc(Handle< FermState<T,P,Q> > state,
-     virtual SystemSolver<T>* qprop(Handle< FermState<T,P,Q> > state,
-     class DiffFermAct4D : public FermAct4D<T,P,Q>
-     virtual DiffLinearOperator<T,Q,P>* linOp(Handle< FermState<T,P,Q> > state) const = 0;
-     virtual DiffLinearOperator<T,Q,P>* lMdagM(Handle< FermState<T,P,Q> > state) const = 0;
-    */
 }
 
 #endif
diff --git a/lib/algorithms/approx/Zolotarev.cc b/lib/algorithms/approx/Zolotarev.cc
index 7629bbde..c73a3436 100644
--- a/lib/algorithms/approx/Zolotarev.cc
+++ b/lib/algorithms/approx/Zolotarev.cc
@@ -58,6 +58,8 @@
 
 /* Compute the partial fraction expansion coefficients (alpha) from the
  * factored form */
+namespace Grid {
+namespace Approx {
 
 static void construct_partfrac(izd *z) {
   int dn = z -> dn, dd = z -> dd, type = z -> type;
@@ -291,7 +293,7 @@ static void sncndnFK(INTERNAL_PRECISION u, INTERNAL_PRECISION k,
  * Set type = 0 for the Zolotarev approximation, which is zero at x = 0, and
  * type = 1 for the approximation which is infinite at x = 0. */
 
-zolotarev_data* bfm_zolotarev(PRECISION epsilon, int n, int type) {
+zolotarev_data* grid_zolotarev(PRECISION epsilon, int n, int type) {
   INTERNAL_PRECISION A, c, cp, kp, ksq, sn, cn, dn, Kp, Kj, z, z0, t, M, F,
     l, invlambda, xi, xisq, *tv, s, opl;
   int m, czero, ts;
@@ -412,7 +414,7 @@ zolotarev_data* bfm_zolotarev(PRECISION epsilon, int n, int type) {
   return zd;
 }
 
-zolotarev_data* bfm_higham(PRECISION epsilon, int n) {
+zolotarev_data* grid_higham(PRECISION epsilon, int n) {
   INTERNAL_PRECISION A, M, c, cp, z, z0, t, epssq;
   int m, czero;
   zolotarev_data *zd;
@@ -502,6 +504,7 @@ zolotarev_data* bfm_higham(PRECISION epsilon, int n) {
   free(d);
   return zd;
 }
+}}
 
 #ifdef TEST
 
@@ -707,4 +710,6 @@ int main(int argc, char** argv) {
 
   return EXIT_SUCCESS;
 }
+
+
 #endif /* TEST */
diff --git a/lib/algorithms/approx/Zolotarev.h b/lib/algorithms/approx/Zolotarev.h
index 3f0dc58e..869e5a89 100644
--- a/lib/algorithms/approx/Zolotarev.h
+++ b/lib/algorithms/approx/Zolotarev.h
@@ -1,7 +1,8 @@
 /* -*- Mode: C; comment-column: 22; fill-column: 79; -*- */
 
 #ifdef __cplusplus
-extern "C" {
+namespace Grid {
+namespace Approx {
 #endif
 
 #define HVERSION Header Time-stamp: <14-OCT-2004 09:26:51.00 adk@MISSCONTRARY>
@@ -76,10 +77,10 @@ typedef struct {
  * zolotarev_data structure. The arguments must satisfy the constraints that
  * epsilon > 0, n > 0, and type = 0 or 1. */
 
-ZOLOTAREV_DATA* bfm_higham(PRECISION epsilon, int n) ;
-ZOLOTAREV_DATA* bfm_zolotarev(PRECISION epsilon, int n, int type);
+ZOLOTAREV_DATA* grid_higham(PRECISION epsilon, int n) ;
+ZOLOTAREV_DATA* grid_zolotarev(PRECISION epsilon, int n, int type);
 #endif
 
 #ifdef __cplusplus
-}
+}}
 #endif
diff --git a/lib/cartesian/Grid_cartesian_base.h b/lib/cartesian/Grid_cartesian_base.h
index e93125c1..66339648 100644
--- a/lib/cartesian/Grid_cartesian_base.h
+++ b/lib/cartesian/Grid_cartesian_base.h
@@ -21,7 +21,7 @@ public:
     // Give Lattice access
     template<class object> friend class Lattice;
 
-    GridBase(std::vector<int> & processor_grid) : CartesianCommunicator(processor_grid) {};
+    GridBase(const std::vector<int> & processor_grid) : CartesianCommunicator(processor_grid) {};
 
 
     // Physics Grid information.
diff --git a/lib/cartesian/Grid_cartesian_full.h b/lib/cartesian/Grid_cartesian_full.h
index 330bbfaf..2a9e0be8 100644
--- a/lib/cartesian/Grid_cartesian_full.h
+++ b/lib/cartesian/Grid_cartesian_full.h
@@ -27,9 +27,9 @@ public:
     virtual int CheckerBoardShift(int source_cb,int dim,int shift, int osite){
       return shift;
     }
-    GridCartesian(std::vector<int> &dimensions,
-		  std::vector<int> &simd_layout,
-		  std::vector<int> &processor_grid
+    GridCartesian(const std::vector<int> &dimensions,
+		  const std::vector<int> &simd_layout,
+		  const std::vector<int> &processor_grid
 		  ) : GridBase(processor_grid)
     {
         ///////////////////////
diff --git a/lib/cartesian/Grid_cartesian_red_black.h b/lib/cartesian/Grid_cartesian_red_black.h
index ace36edb..3a84ed49 100644
--- a/lib/cartesian/Grid_cartesian_red_black.h
+++ b/lib/cartesian/Grid_cartesian_red_black.h
@@ -81,28 +81,28 @@ public:
       }
     };
 
-    GridRedBlackCartesian(GridBase *base) : GridRedBlackCartesian(base->_fdimensions,base->_simd_layout,base->_processors)  {};
+    GridRedBlackCartesian(const GridBase *base) : GridRedBlackCartesian(base->_fdimensions,base->_simd_layout,base->_processors)  {};
 
-    GridRedBlackCartesian(std::vector<int> &dimensions,
-			  std::vector<int> &simd_layout,
-			  std::vector<int> &processor_grid,
-			  std::vector<int> &checker_dim_mask,
+    GridRedBlackCartesian(const std::vector<int> &dimensions,
+			  const std::vector<int> &simd_layout,
+			  const std::vector<int> &processor_grid,
+			  const std::vector<int> &checker_dim_mask,
 			  int checker_dim
 			  ) :  GridBase(processor_grid) 
     {
       Init(dimensions,simd_layout,processor_grid,checker_dim_mask,checker_dim);
     }
-    GridRedBlackCartesian(std::vector<int> &dimensions,
-			  std::vector<int> &simd_layout,
-			  std::vector<int> &processor_grid) : GridBase(processor_grid) 
+    GridRedBlackCartesian(const std::vector<int> &dimensions,
+			  const std::vector<int> &simd_layout,
+			  const std::vector<int> &processor_grid) : GridBase(processor_grid) 
     {
       std::vector<int> checker_dim_mask(dimensions.size(),1);
       Init(dimensions,simd_layout,processor_grid,checker_dim_mask,0);
     }
-    void Init(std::vector<int> &dimensions,
-	      std::vector<int> &simd_layout,
-	      std::vector<int> &processor_grid,
-	      std::vector<int> &checker_dim_mask,
+    void Init(const std::vector<int> &dimensions,
+	      const std::vector<int> &simd_layout,
+	      const std::vector<int> &processor_grid,
+	      const std::vector<int> &checker_dim_mask,
 	      int checker_dim)
     {
     ///////////////////////
diff --git a/lib/communicator/Grid_communicator_base.h b/lib/communicator/Grid_communicator_base.h
index 47c1f525..61e19993 100644
--- a/lib/communicator/Grid_communicator_base.h
+++ b/lib/communicator/Grid_communicator_base.h
@@ -27,7 +27,7 @@ class CartesianCommunicator {
 #endif
 
     // Constructor
-    CartesianCommunicator(std::vector<int> &pdimensions_in);
+    CartesianCommunicator(const std::vector<int> &pdimensions_in);
 
     // Wraps MPI_Cart routines
     void ShiftedRanks(int dim,int shift,int & source, int & dest);
diff --git a/lib/communicator/Grid_communicator_mpi.cc b/lib/communicator/Grid_communicator_mpi.cc
index 6ef05c3d..5dd34705 100644
--- a/lib/communicator/Grid_communicator_mpi.cc
+++ b/lib/communicator/Grid_communicator_mpi.cc
@@ -5,7 +5,7 @@ namespace Grid {
 
   // Should error check all MPI calls.
 
-CartesianCommunicator::CartesianCommunicator(std::vector<int> &processors)
+CartesianCommunicator::CartesianCommunicator(const std::vector<int> &processors)
 {
   _ndimension = processors.size();
   std::vector<int> periodic(_ndimension,1);
diff --git a/lib/communicator/Grid_communicator_none.cc b/lib/communicator/Grid_communicator_none.cc
index 90eb26cc..d7eb9453 100644
--- a/lib/communicator/Grid_communicator_none.cc
+++ b/lib/communicator/Grid_communicator_none.cc
@@ -1,7 +1,7 @@
 #include "Grid.h"
 namespace Grid {
 
-CartesianCommunicator::CartesianCommunicator(std::vector<int> &processors)
+CartesianCommunicator::CartesianCommunicator(const std::vector<int> &processors)
 {
   _processors = processors;
   _ndimension = processors.size();
diff --git a/lib/qcd/LinalgUtils.h b/lib/qcd/LinalgUtils.h
new file mode 100644
index 00000000..2b83d115
--- /dev/null
+++ b/lib/qcd/LinalgUtils.h
@@ -0,0 +1,113 @@
+#ifndef GRID_QCD_LINALG_UTILS_H
+#define GRID_QCD_LINALG_UTILS_H
+
+namespace Grid{
+namespace QCD{
+////////////////////////////////////////////////////////////////////////
+//This file brings additional linear combination assist that is helpful
+//to QCD such as chiral projectors and spin matrices applied to one of the inputs.
+//These routines support five-D chiral fermions and contain s-subslice indexing 
+//on the 5d (rb4d) checkerboarded lattices
+////////////////////////////////////////////////////////////////////////
+template<class vobj> 
+void axpby_ssp(Lattice<vobj> &z, RealD a,const Lattice<vobj> &x,RealD b,const Lattice<vobj> &y,int s,int sp)
+{
+  z.checkerboard = x.checkerboard;
+  conformable(x,y);
+  conformable(x,z);
+  GridBase *grid=x._grid;
+  int Ls = grid->_rdimensions[0];
+PARALLEL_FOR_LOOP
+  for(int ss=0;ss<grid->oSites();ss+=Ls){ // adds Ls
+    vobj tmp = a*x._odata[ss+s]+b*y._odata[ss+sp];
+    vstream(z._odata[ss+s],tmp);
+  }
+}
+
+template<class vobj> 
+void ag5xpby_ssp(Lattice<vobj> &z,RealD a,const Lattice<vobj> &x,RealD b,const Lattice<vobj> &y,int s,int sp)
+{
+  z.checkerboard = x.checkerboard;
+  conformable(x,y);
+  conformable(x,z);
+  GridBase *grid=x._grid;
+  int Ls = grid->_rdimensions[0];
+PARALLEL_FOR_LOOP
+  for(int ss=0;ss<grid->oSites();ss+=Ls){ // adds Ls
+    vobj tmp;
+    multGamma5(tmp(),a*x._odata[ss+s]());
+    tmp = tmp + b*y._odata[ss+sp];
+    vstream(z._odata[ss+s],tmp);
+  }
+}
+
+template<class vobj> 
+void axpbg5y_ssp(Lattice<vobj> &z,RealD a,const Lattice<vobj> &x,RealD b,const Lattice<vobj> &y,int s,int sp)
+{
+  z.checkerboard = x.checkerboard;
+  conformable(x,y);
+  conformable(x,z);
+  GridBase *grid=x._grid;
+  int Ls = grid->_rdimensions[0];
+PARALLEL_FOR_LOOP
+  for(int ss=0;ss<grid->oSites();ss+=Ls){ // adds Ls
+    vobj tmp;
+    multGamma5(tmp(),b*y._odata[ss+sp]());
+    tmp = tmp + a*x._odata[ss+s];
+    vstream(z._odata[ss+s],tmp);
+  }
+}
+
+template<class vobj> 
+void ag5xpbg5y_ssp(Lattice<vobj> &z,RealD a,const Lattice<vobj> &x,RealD b,const Lattice<vobj> &y,int s,int sp)
+{
+  z.checkerboard = x.checkerboard;
+  conformable(x,y);
+  conformable(x,z);
+  GridBase *grid=x._grid;
+  int Ls = grid->_rdimensions[0];
+PARALLEL_FOR_LOOP
+  for(int ss=0;ss<grid->oSites();ss+=Ls){ // adds Ls
+    vobj tmp1;
+    vobj tmp2;
+    tmp1 = a*x._odata[ss+s]+b*y._odata[ss+sp];
+    multGamma5(tmp2(),tmp1());
+    vstream(z._odata[ss+s],tmp2);
+  }
+}
+
+template<class vobj> 
+void axpby_ssp_pminus(Lattice<vobj> &z,RealD a,const Lattice<vobj> &x,RealD b,const Lattice<vobj> &y,int s,int sp)
+{
+  z.checkerboard = x.checkerboard;
+  conformable(x,y);
+  conformable(x,z);
+  GridBase *grid=x._grid;
+  int Ls = grid->_rdimensions[0];
+PARALLEL_FOR_LOOP
+  for(int ss=0;ss<grid->oSites();ss+=Ls){ // adds Ls
+    vobj tmp;
+    spProj5m(tmp,y._odata[ss+sp]);
+    tmp = a*x._odata[ss+s]+b*tmp;
+    vstream(z._odata[ss+s],tmp);
+  }
+}
+
+template<class vobj> 
+void axpby_ssp_pplus(Lattice<vobj> &z,RealD a,const Lattice<vobj> &x,RealD b,const Lattice<vobj> &y,int s,int sp)
+{
+  z.checkerboard = x.checkerboard;
+  conformable(x,y);
+  conformable(x,z);
+  GridBase *grid=x._grid;
+  int Ls = grid->_rdimensions[0];
+PARALLEL_FOR_LOOP
+  for(int ss=0;ss<grid->oSites();ss+=Ls){ // adds Ls
+    vobj tmp;
+    spProj5p(tmp,y._odata[ss+sp]);
+    tmp = a*x._odata[ss+s]+b*tmp;
+    vstream(z._odata[ss+s],tmp);
+  }
+}
+}}
+#endif 
diff --git a/lib/qcd/QCD.h b/lib/qcd/QCD.h
index 7c45eb23..4b5edc5c 100644
--- a/lib/qcd/QCD.h
+++ b/lib/qcd/QCD.h
@@ -307,8 +307,10 @@ namespace QCD {
 }   //namespace QCD
 } // Grid
 
+#include <qcd/SpaceTimeGrid.h>
 #include <qcd/Dirac.h>
 #include <qcd/TwoSpinor.h>
+#include <qcd/LinalgUtils.h>
 #include <qcd/action/Actions.h>
 
 #endif
diff --git a/lib/qcd/SpaceTimeGrid.cc b/lib/qcd/SpaceTimeGrid.cc
new file mode 100644
index 00000000..284c5771
--- /dev/null
+++ b/lib/qcd/SpaceTimeGrid.cc
@@ -0,0 +1,52 @@
+#include <Grid.h>
+
+namespace Grid { 
+  namespace QCD {
+
+/////////////////////////////////////////////////////////////////
+// Public interface
+/////////////////////////////////////////////////////////////////
+GridCartesian *SpaceTimeGrid::makeFourDimGrid(const std::vector<int> & latt,const std::vector<int> &simd,const std::vector<int> &mpi)
+{
+  return new GridCartesian(latt,simd,mpi); 
+}
+GridRedBlackCartesian *SpaceTimeGrid::makeFourDimRedBlackGrid(const GridCartesian *FourDimGrid)
+{
+  return new GridRedBlackCartesian(FourDimGrid); 
+}
+
+GridCartesian         *SpaceTimeGrid::makeFiveDimGrid(int Ls,const GridCartesian *FourDimGrid)
+{
+  int N4=FourDimGrid->_ndimension;
+
+  std::vector<int> latt5(1,Ls);
+  std::vector<int> simd5(1,1);
+  std::vector<int>  mpi5(1,1);
+  
+  for(int d=0;d<N4;d++){
+    latt5.push_back(FourDimGrid->_fdimensions[d]);
+    simd5.push_back(FourDimGrid->_simd_layout[d]);
+     mpi5.push_back(FourDimGrid->_processors[d]);
+  }
+  return new GridCartesian(latt5,simd5,mpi5); 
+}
+
+GridRedBlackCartesian *SpaceTimeGrid::makeFiveDimRedBlackGrid(int Ls,const GridCartesian *FourDimGrid)
+{
+  int N4=FourDimGrid->_ndimension;
+  int cbd=1;
+  std::vector<int> latt5(1,Ls);
+  std::vector<int> simd5(1,1);
+  std::vector<int>  mpi5(1,1);
+  std::vector<int>   cb5(1,0);
+    
+  for(int d=0;d<N4;d++){
+    latt5.push_back(FourDimGrid->_fdimensions[d]);
+    simd5.push_back(FourDimGrid->_simd_layout[d]);
+     mpi5.push_back(FourDimGrid->_processors[d]);
+      cb5.push_back(  1);
+    }
+  return new GridRedBlackCartesian(latt5,simd5,mpi5,cb5,cbd); 
+}
+
+}}
diff --git a/lib/qcd/SpaceTimeGrid.h b/lib/qcd/SpaceTimeGrid.h
new file mode 100644
index 00000000..0b386a0e
--- /dev/null
+++ b/lib/qcd/SpaceTimeGrid.h
@@ -0,0 +1,18 @@
+#ifndef GRID_QCD_SPACE_TIME_GRID_H
+#define GRID_QCD_SPACE_TIME_GRID_H
+namespace Grid {
+namespace QCD {
+
+class SpaceTimeGrid {
+ public:
+
+  static GridCartesian         *makeFourDimGrid(const std::vector<int> & latt,const std::vector<int> &simd,const std::vector<int> &mpi);
+  static GridRedBlackCartesian *makeFourDimRedBlackGrid       (const GridCartesian *FourDimGrid);
+  static GridCartesian         *makeFiveDimGrid        (int Ls,const GridCartesian *FourDimGrid);
+  static GridRedBlackCartesian *makeFiveDimRedBlackGrid(int Ls,const GridCartesian *FourDimGrid);
+
+};
+
+}}
+
+#endif
diff --git a/lib/qcd/action/Actions.h b/lib/qcd/action/Actions.h
index c4e8a2f0..acbf027c 100644
--- a/lib/qcd/action/Actions.h
+++ b/lib/qcd/action/Actions.h
@@ -1,10 +1,80 @@
 #ifndef GRID_QCD_ACTIONS_H
 #define GRID_QCD_ACTIONS_H
 
-#include <qcd/action/fermion/FermionAction.h>
-#include <qcd/action/fermion/WilsonCompressor.h>
-#include <qcd/action/fermion/WilsonKernels.h>
+
+// Some reorganisation likely required as both Chroma and IroIro
+// are separating the concept of the operator from that of action.
+//
+// The FermAction contains methods to create 
+//
+// * Linear operators             (Hermitian and non-hermitian)  .. my LinearOperator
+// * System solvers               (Hermitian and non-hermitian)  .. my OperatorFunction
+// * MultiShift System solvers    (Hermitian and non-hermitian)  .. my OperatorFunction
+
+
+////////////////////////////////////////////
+// Abstract base interface
+////////////////////////////////////////////
+#include <qcd/action/fermion/FermionOperator.h>
+
+////////////////////////////////////////////
+// Utility functions
+////////////////////////////////////////////
+#include <qcd/action/fermion/WilsonCompressor.h>     //used by all wilson type fermions
+#include <qcd/action/fermion/WilsonKernels.h>        //used by all wilson type fermions
+
+////////////////////////////////////////////
+// 4D formulations
+////////////////////////////////////////////
 #include <qcd/action/fermion/WilsonFermion.h>
-#include <qcd/action/fermion/FiveDimWilsonFermion.h>
+//#include <qcd/action/fermion/CloverFermion.h>
+
+////////////////////////////////////////////
+// 5D formulations
+////////////////////////////////////////////
+#include <qcd/action/fermion/WilsonFermion5D.h> // used by all 5d overlap types
+#include <qcd/action/fermion/CayleyFermion5D.h>
+#include <qcd/action/fermion/ContinuedFractionFermion5D.h>
+//#include <qcd/action/fermion/PartialFraction.h>
+
+#include <qcd/action/fermion/DomainWallFermion.h>
+//#include <qcd/action/fermion/ScaledShamirCayleyTanh.h>
+
+
+    // Chroma interface defining FermionAction
+    /*
+     template<typename T, typename P, typename Q>  class FermAct4D : public FermionAction<T,P,Q>
+     virtual LinearOperator<T>* linOp(Handle< FermState<T,P,Q> > state) const = 0;
+     virtual LinearOperator<T>* lMdagM(Handle< FermState<T,P,Q> > state) const = 0;
+     virtual LinOpSystemSolver<T>* invLinOp(Handle< FermState<T,P,Q> > state,
+     virtual MdagMSystemSolver<T>* invMdagM(Handle< FermState<T,P,Q> > state,
+     virtual LinOpMultiSystemSolver<T>* mInvLinOp(Handle< FermState<T,P,Q> > state,
+     virtual MdagMMultiSystemSolver<T>* mInvMdagM(Handle< FermState<T,P,Q> > state,
+     virtual MdagMMultiSystemSolverAccumulate<T>* mInvMdagMAcc(Handle< FermState<T,P,Q> > state,
+     virtual SystemSolver<T>* qprop(Handle< FermState<T,P,Q> > state,
+     class DiffFermAct4D : public FermAct4D<T,P,Q>
+     virtual DiffLinearOperator<T,Q,P>* linOp(Handle< FermState<T,P,Q> > state) const = 0;
+     virtual DiffLinearOperator<T,Q,P>* lMdagM(Handle< FermState<T,P,Q> > state) const = 0;
+    */
+
+
+    // Chroma interface defining GaugeAction
+    /*
+      template<typename P, typename Q>   class GaugeAction
+  virtual const CreateGaugeState<P,Q>& getCreateState() const = 0;
+  virtual GaugeState<P,Q>* createState(const Q& q) const
+  virtual const GaugeBC<P,Q>& getGaugeBC() const
+  virtual const Set& getSet(void) const = 0;
+  virtual void deriv(P& result, const Handle< GaugeState<P,Q> >& state) const 
+  virtual Double S(const Handle< GaugeState<P,Q> >& state) const = 0;
+
+  class LinearGaugeAction : public GaugeAction< multi1d<LatticeColorMatrix>, multi1d<LatticeColorMatrix> >
+  typedef multi1d<LatticeColorMatrix>  P;
+  typedef multi1d<LatticeColorMatrix>  Q;
+  virtual void staple(LatticeColorMatrix& result,
+		      const Handle< GaugeState<P,Q> >& state,
+		      int mu, int cb) const = 0;
+    */
+
 
 #endif
diff --git a/lib/qcd/action/fermion/CayleyFermion5D.cc b/lib/qcd/action/fermion/CayleyFermion5D.cc
new file mode 100644
index 00000000..263cc28b
--- /dev/null
+++ b/lib/qcd/action/fermion/CayleyFermion5D.cc
@@ -0,0 +1,235 @@
+#include <Grid.h>
+namespace Grid {
+namespace QCD {
+
+ CayleyFermion5D::CayleyFermion5D(LatticeGaugeField &_Umu,
+				  GridCartesian         &FiveDimGrid,
+				  GridRedBlackCartesian &FiveDimRedBlackGrid,
+				  GridCartesian         &FourDimGrid,
+				  GridRedBlackCartesian &FourDimRedBlackGrid,
+				  RealD _mass,RealD _M5) :
+   WilsonFermion5D(_Umu,
+		   FiveDimGrid,
+		   FiveDimRedBlackGrid,
+		   FourDimGrid,
+		   FourDimRedBlackGrid,_M5),
+   mass(_mass)
+ {
+   std::cout << "Constructing a CayleyFermion5D"<<std::endl;
+ }
+
+  // override multiply
+  RealD CayleyFermion5D::M    (const LatticeFermion &psi, LatticeFermion &chi)
+  {
+    LatticeFermion Din(psi._grid);
+
+    // Assemble Din
+    for(int s=0;s<Ls;s++){
+      if ( s==0 ) {
+	//	Din = bs psi[s] + cs[s] psi[s+1}
+	axpby_ssp_pminus(Din,bs[s],psi,cs[s],psi,s,s+1);
+	//      Din+= -mass*cs[s] psi[s+1}
+	axpby_ssp_pplus (Din,1.0,Din,-mass*cs[s],psi,s,Ls-1);
+      } else if ( s==(Ls-1)) { 
+	axpby_ssp_pminus(Din,bs[s],psi,-mass*cs[s],psi,s,0);
+	axpby_ssp_pplus (Din,1.0,Din,cs[s],psi,s,s-1);
+      } else {
+	axpby_ssp_pminus(Din,bs[s],psi,cs[s],psi,s,s+1);
+	axpby_ssp_pplus(Din,1.0,Din,cs[s],psi,s,s-1);
+      }
+    }
+
+    DW(Din,chi,DaggerNo);
+    // ((b D_W + D_w hop terms +1) on s-diag
+    axpby(chi,1.0,1.0,chi,psi); 
+
+    for(int s=0;s<Ls;s++){
+      if ( s==0 ){
+	axpby_ssp_pminus(chi,1.0,chi,-1.0,psi,s,s+1);
+	axpby_ssp_pplus (chi,1.0,chi,mass,psi,s,Ls-1);
+      } else if ( s==(Ls-1)) {
+	axpby_ssp_pminus(chi,1.0,chi,mass,psi,s,0);
+	axpby_ssp_pplus (chi,1.0,chi,-1.0,psi,s,s-1);
+      } else {
+	axpby_ssp_pminus(chi,1.0,chi,-1.0,psi,s,s+1);
+	axpby_ssp_pplus (chi,1.0,chi,-1.0,psi,s,s-1);
+      }
+    }
+    return norm2(chi);
+  }
+
+  RealD CayleyFermion5D::Mdag (const LatticeFermion &psi, LatticeFermion &chi)
+  {
+    // Under adjoint
+    //D1+        D1- P-    ->   D1+^dag   P+ D2-^dag
+    //D2- P+     D2+            P-D1-^dag D2+dag
+
+    LatticeFermion Din(psi._grid);
+    // Apply Dw
+    DW(psi,Din,DaggerYes); 
+
+    for(int s=0;s<Ls;s++){
+      // Collect the terms in DW
+      //	Chi = bs Din[s] + cs[s] Din[s+1}
+      //    Chi+= -mass*cs[s] psi[s+1}
+      if ( s==0 ) {
+	axpby_ssp_pplus (chi,bs[s],Din,cs[s+1],Din,s,s+1);
+	axpby_ssp_pminus(chi,1.0,chi,-mass*cs[Ls-1],Din,s,Ls-1);
+      } else if ( s==(Ls-1)) { 
+	axpby_ssp_pplus (chi,bs[s],Din,-mass*cs[0],Din,s,0);
+	axpby_ssp_pminus(chi,1.0,chi,cs[s-1],Din,s,s-1);
+      } else {
+	axpby_ssp_pplus (chi,bs[s],Din,cs[s+1],Din,s,s+1);
+	axpby_ssp_pminus(chi,1.0,chi,cs[s-1],Din,s,s-1);
+      }
+      // Collect the terms indept of DW
+      if ( s==0 ){
+	axpby_ssp_pplus (chi,1.0,chi,-1.0,psi,s,s+1);
+	axpby_ssp_pminus(chi,1.0,chi,mass,psi,s,Ls-1);
+      } else if ( s==(Ls-1)) {
+	axpby_ssp_pplus (chi,1.0,chi,mass,psi,s,0);
+	axpby_ssp_pminus(chi,1.0,chi,-1.0,psi,s,s-1);
+      } else {
+	axpby_ssp_pplus(chi,1.0,chi,-1.0,psi,s,s+1);
+	axpby_ssp_pminus(chi,1.0,chi,-1.0,psi,s,s-1);
+      }
+    }
+    // ((b D_W + D_w hop terms +1) on s-diag
+    axpby (chi,1.0,1.0,chi,psi); 
+    return norm2(chi);
+  }
+
+  // half checkerboard operations
+  void CayleyFermion5D::Meooe       (const LatticeFermion &psi, LatticeFermion &chi)
+  {
+    LatticeFermion tmp(psi._grid);
+    // Assemble the 5d matrix
+    for(int s=0;s<Ls;s++){
+      if ( s==0 ) {
+	//	tmp = bs psi[s] + cs[s] psi[s+1}
+	//      tmp+= -mass*cs[s] psi[s+1}
+	axpby_ssp_pminus(tmp,beo[s],psi,-ceo[s],psi ,s, s+1);
+	axpby_ssp_pplus(tmp,1.0,tmp,mass*ceo[s],psi,s,Ls-1);
+      } else if ( s==(Ls-1)) { 
+	axpby_ssp_pminus(tmp,beo[s],psi,mass*ceo[s],psi,s,0);
+	axpby_ssp_pplus(tmp,1.0,tmp,-ceo[s],psi,s,s-1);
+      } else {
+	axpby_ssp_pminus(tmp,beo[s],psi,-ceo[s],psi,s,s+1);
+	axpby_ssp_pplus (tmp,1.0,tmp,-ceo[s],psi,s,s-1);
+      }
+    }
+    // Apply 4d dslash
+    if ( psi.checkerboard == Odd ) {
+      DhopEO(tmp,chi,DaggerNo);
+    } else {
+      DhopOE(tmp,chi,DaggerNo);
+    }
+  }
+
+  void CayleyFermion5D::MeooeDag    (const LatticeFermion &psi, LatticeFermion &chi)
+  {
+    LatticeFermion tmp(psi._grid);
+    // Apply 4d dslash
+    if ( psi.checkerboard == Odd ) {
+      DhopEO(psi,tmp,DaggerYes);
+    } else {
+      DhopOE(psi,tmp,DaggerYes);
+    }
+    // Assemble the 5d matrix
+    for(int s=0;s<Ls;s++){
+      if ( s==0 ) {
+	axpby_ssp_pplus(chi,beo[s],tmp,   -ceo[s+1]  ,tmp,s,s+1);
+	axpby_ssp_pminus(chi,   1.0,chi,mass*ceo[Ls-1],tmp,s,Ls-1);
+      } else if ( s==(Ls-1)) { 
+	axpby_ssp_pplus(chi,beo[s],tmp,mass*ceo[0],tmp,s,0);
+	axpby_ssp_pminus(chi,1.0,chi,-ceo[s-1],tmp,s,s-1);
+      } else {
+	axpby_ssp_pplus(chi,beo[s],tmp,-ceo[s+1],tmp,s,s+1);
+	axpby_ssp_pminus(chi,1.0   ,chi,-ceo[s-1],tmp,s,s-1);
+      }
+    }
+  }
+
+  void CayleyFermion5D::Mooee       (const LatticeFermion &psi, LatticeFermion &chi)
+  {
+    for (int s=0;s<Ls;s++){
+      if ( s==0 ) {
+	axpby_ssp_pminus(chi,bee[s],psi ,-cee[s],psi,s,s+1);
+	axpby_ssp_pplus (chi,1.0,chi,mass*cee[s],psi,s,Ls-1);
+      } else if ( s==(Ls-1)) { 
+	axpby_ssp_pminus(chi,bee[s],psi,mass*cee[s],psi,s,0);
+	axpby_ssp_pplus (chi,1.0,chi,-cee[s],psi,s,s-1);
+      } else {
+	axpby_ssp_pminus(chi,bee[s],psi,-cee[s],psi,s,s+1);
+	axpby_ssp_pplus (chi,1.0,chi,-cee[s],psi,s,s-1);
+      }
+    }
+  }
+
+  void CayleyFermion5D::MooeeDag    (const LatticeFermion &psi, LatticeFermion &chi)
+  {
+    for (int s=0;s<Ls;s++){
+      // Assemble the 5d matrix
+      if ( s==0 ) {
+	axpby_ssp_pplus(chi,bee[s],psi,-cee[s+1]  ,psi,s,s+1);
+	axpby_ssp_pminus(chi,1.0,chi,mass*cee[Ls-1],psi,s,Ls-1);
+      } else if ( s==(Ls-1)) { 
+	axpby_ssp_pplus(chi,bee[s],psi,mass*cee[0],psi,s,0);
+	axpby_ssp_pminus(chi,1.0,chi,-cee[s-1],psi,s,s-1);
+      } else {
+	axpby_ssp_pplus(chi,bee[s],psi,-cee[s+1],psi,s,s+1);
+	axpby_ssp_pminus(chi,1.0   ,chi,-cee[s-1],psi,s,s-1);
+      }
+    }
+  }
+
+  void CayleyFermion5D::MooeeInv    (const LatticeFermion &psi, LatticeFermion &chi)
+  {
+    // Apply (L^{\prime})^{-1}
+    axpby_ssp (chi,1.0,psi,     0.0,psi,0,0);      // chi[0]=psi[0]
+    for (int s=1;s<Ls;s++){
+      axpby_ssp_pplus(chi,1.0,psi,-lee[s-1],chi,s,s-1);// recursion Psi[s] -lee P_+ chi[s-1]
+    }
+    // L_m^{-1} 
+    for (int s=0;s<Ls-1;s++){ // Chi[ee] = 1 - sum[s<Ls-1] -leem[s]P_- chi
+      axpby_ssp_pminus(chi,1.0,chi,-leem[s],chi,Ls-1,s);
+    }
+    // U_m^{-1} D^{-1}
+    for (int s=0;s<Ls-1;s++){
+      // Chi[s] + 1/d chi[s] 
+      axpby_ssp_pplus(chi,1.0/dee[s],chi,-ueem[s]/dee[Ls-1],chi,s,Ls-1);
+    }	
+    axpby_ssp(chi,1.0/dee[Ls-1],chi,0.0,chi,Ls-1,Ls-1); // Modest avoidable 
+    
+    // Apply U^{-1}
+    for (int s=Ls-2;s>=0;s--){
+      axpby_ssp_pminus (chi,1.0,chi,-uee[s],chi,s,s+1);  // chi[Ls]
+    }
+  }
+
+  void CayleyFermion5D::MooeeInvDag (const LatticeFermion &psi, LatticeFermion &chi)
+  {
+    // Apply (U^{\prime})^{-dagger}
+    axpby_ssp (chi,1.0,psi,     0.0,psi,0,0);      // chi[0]=psi[0]
+    for (int s=1;s<Ls;s++){
+      axpby_ssp_pminus(chi,1.0,psi,-uee[s-1],chi,s,s-1);
+    }
+    // U_m^{-\dagger} 
+    for (int s=0;s<Ls-1;s++){
+      axpby_ssp_pplus(chi,1.0,chi,-ueem[s],chi,Ls-1,s);
+    }
+    // L_m^{-\dagger} D^{-dagger}
+    for (int s=0;s<Ls-1;s++){
+      axpby_ssp_pminus(chi,1.0/dee[s],chi,-leem[s]/dee[Ls-1],chi,s,Ls-1);
+    }	
+    axpby_ssp(chi,1.0/dee[Ls-1],chi,0.0,chi,Ls-1,Ls-1); // Modest avoidable 
+    
+    // Apply L^{-dagger}
+    for (int s=Ls-2;s>=0;s--){
+      axpby_ssp_pplus (chi,1.0,chi,-lee[s],chi,s,s+1);  // chi[Ls]
+    }
+  }
+
+}
+}
+
diff --git a/lib/qcd/action/fermion/CayleyFermion5D.h b/lib/qcd/action/fermion/CayleyFermion5D.h
new file mode 100644
index 00000000..857ac124
--- /dev/null
+++ b/lib/qcd/action/fermion/CayleyFermion5D.h
@@ -0,0 +1,61 @@
+#ifndef  GRID_QCD_CAYLEY_FERMION_H
+#define  GRID_QCD_CAYLEY_FERMION_H
+
+namespace Grid {
+
+  namespace QCD {
+
+    class CayleyFermion5D : public WilsonFermion5D
+    {
+    public:
+
+      // override multiply
+      virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out);
+      virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out);
+
+      // half checkerboard operations
+      virtual void   Meooe       (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
+
+      //    protected:
+
+      Approx::zolotarev_data *zdata;
+
+      RealD mass;
+      // Cayley form Moebius (tanh and zolotarev)
+      std::vector<RealD> omega; 
+      std::vector<RealD> bs;    // S dependent coeffs
+      std::vector<RealD> cs;    
+      std::vector<RealD> as;    
+      // For preconditioning Cayley form
+      std::vector<RealD> bee;    
+      std::vector<RealD> cee;    
+      std::vector<RealD> aee;    
+      std::vector<RealD> beo;    
+      std::vector<RealD> ceo;    
+      std::vector<RealD> aeo;    
+      // LDU factorisation of the eeoo matrix
+      std::vector<RealD> lee;    
+      std::vector<RealD> leem;    
+      std::vector<RealD> uee;    
+      std::vector<RealD> ueem;    
+      std::vector<RealD> dee;    
+
+      // Constructors
+      CayleyFermion5D(LatticeGaugeField &_Umu,
+		      GridCartesian         &FiveDimGrid,
+		      GridRedBlackCartesian &FiveDimRedBlackGrid,
+		      GridCartesian         &FourDimGrid,
+		      GridRedBlackCartesian &FourDimRedBlackGrid,
+		      RealD _mass,RealD _M5);
+
+    };
+
+  }
+}
+
+#endif
diff --git a/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc b/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
new file mode 100644
index 00000000..c281b486
--- /dev/null
+++ b/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
@@ -0,0 +1,119 @@
+#include <Grid.h>
+
+namespace Grid {
+
+  namespace QCD {
+
+    RealD  ContinuedFractionFermion5D::M           (const LatticeFermion &psi, LatticeFermion &chi)
+    {
+      LatticeFermion D(psi._grid);
+
+      DW(psi,D,DaggerNo); 
+
+      int sign=1;
+      for(int s=0;s<Ls;s++){
+	if ( s==0 ) {
+	  ag5xpby_ssp(chi,cc[0]*Beta[0]*sign*scale,D,sqrt_cc[0],psi,s,s+1); // Multiplies Dw by G5 so Hw
+	} else if ( s==(Ls-1) ){
+	  RealD R=(1.0+mass)/(1.0-mass);
+	  ag5xpby_ssp(chi,Beta[s]*scale,D,sqrt_cc[s-1],psi,s,s-1);
+	  ag5xpby_ssp(chi,R,psi,1.0,chi,s,s);
+	} else {
+	  ag5xpby_ssp(chi,cc[s]*Beta[s]*sign*scale,D,sqrt_cc[s],psi,s,s+1);
+  	  axpby_ssp(chi,1.0,chi,sqrt_cc[s-1],psi,s,s-1);
+	}
+	sign=-sign; 
+      }
+      return norm2(chi);
+    }
+    RealD  ContinuedFractionFermion5D::Mdag        (const LatticeFermion &psi, LatticeFermion &chi)
+    {
+      // This matrix is already hermitian. (g5 Dw) = Dw dag g5 = (g5 Dw)dag
+      // The rest of matrix is symmetric.
+      // Can ignore "dag"
+      return M(psi,chi);
+    }
+    void   ContinuedFractionFermion5D::Meooe       (const LatticeFermion &psi, LatticeFermion &chi)
+    {
+      Dhop(psi,chi,DaggerNo); // Dslash on diagonal. g5 Dslash is hermitian
+      
+      int sign=1;
+      for(int s=0;s<Ls;s++){
+	if ( s==(Ls-1) ){
+	  ag5xpby_ssp(chi,Beta[s]*scale,chi,0.0,chi,s,s);
+	} else {
+	  ag5xpby_ssp(chi,cc[s]*Beta[s]*sign*scale,chi,0.0,chi,s,s);
+	}
+	sign=-sign; 
+    }
+
+    }
+    void   ContinuedFractionFermion5D::MeooeDag    (const LatticeFermion &psi, LatticeFermion &chi)
+    {
+      Meooe(psi,chi);
+    }
+    void   ContinuedFractionFermion5D::Mooee       (const LatticeFermion &psi, LatticeFermion &chi)
+    {
+      double dw_diag = (4.0-this->M5)*scale;
+    
+      int sign=1;
+      for(int s=0;s<Ls;s++){
+	if ( s==0 ) {
+	  ag5xpby_ssp(chi,cc[0]*Beta[0]*sign*dw_diag,psi,sqrt_cc[0],psi,s,s+1); // Multiplies Dw by G5 so Hw
+	} else if ( s==(Ls-1) ){
+	  // Drop the CC here.
+	  double R=(1+this->mass)/(1-this->mass);
+	  ag5xpby_ssp(chi,Beta[s]*dw_diag,psi,sqrt_cc[s-1],psi,s,s-1);
+	  ag5xpby_ssp(chi,R,psi,1.0,chi,s,s);
+	} else {
+	  ag5xpby_ssp(chi,cc[s]*Beta[s]*sign*dw_diag,psi,sqrt_cc[s],psi,s,s+1);
+	  axpby_ssp(chi,1.0,chi,sqrt_cc[s-1],psi,s,s-1);
+	}
+	sign=-sign; 
+      }
+    }
+
+    void   ContinuedFractionFermion5D::MooeeDag    (const LatticeFermion &psi, LatticeFermion &chi)
+    {
+      Mooee(psi,chi);
+    }
+    void   ContinuedFractionFermion5D::MooeeInv    (const LatticeFermion &psi, LatticeFermion &chi)
+    {
+      // Apply Linv
+      axpby_ssp(chi,1.0/cc_d[0],psi,0.0,psi,0,0);
+      for(int s=1;s<Ls;s++){
+	axpbg5y_ssp(chi,1.0/cc_d[s],psi,-1.0/See[s-1],chi,s,s-1);
+      }
+      // Apply Dinv
+      for(int s=0;s<Ls;s++){
+	ag5xpby_ssp(chi,1.0/See[s],chi,0.0,chi,s,s); //only appearance of See[0]
+      }
+      // Apply Uinv = (Linv)^T
+      axpby_ssp(chi,1.0/cc_d[Ls-1],chi,0.0,chi,this->Ls-1,this->Ls-1);
+      for(int s=Ls-2;s>=0;s--){
+	axpbg5y_ssp(chi,1.0/cc_d[s],chi,-1.0*cc_d[s+1]/See[s]/cc_d[s],chi,s,s+1);
+      }
+    }
+    void   ContinuedFractionFermion5D::MooeeInvDag (const LatticeFermion &psi, LatticeFermion &chi)
+    {
+      MooeeInv(psi,chi);
+    }
+    
+    // Constructors
+    ContinuedFractionFermion5D::ContinuedFractionFermion5D(
+							   LatticeGaugeField &_Umu,
+							   GridCartesian         &FiveDimGrid,
+							   GridRedBlackCartesian &FiveDimRedBlackGrid,
+							   GridCartesian         &FourDimGrid,
+							   GridRedBlackCartesian &FourDimRedBlackGrid,
+							   RealD _mass,RealD M5) :
+      WilsonFermion5D(_Umu,
+		      FiveDimGrid, FiveDimRedBlackGrid,
+		      FourDimGrid, FourDimRedBlackGrid,M5),
+      mass(_mass)
+    {
+    }
+
+  }
+}
+
diff --git a/lib/qcd/action/fermion/ContinuedFractionFermion5D.h b/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
new file mode 100644
index 00000000..7f5c022a
--- /dev/null
+++ b/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
@@ -0,0 +1,53 @@
+#ifndef  GRID_QCD_CONTINUED_FRACTION_H
+#define  GRID_QCD_CONTINUED_FRACTION_H
+
+namespace Grid {
+
+  namespace QCD {
+
+    class ContinuedFractionFermion5D : public WilsonFermion5D
+    {
+    public:
+
+      // override multiply
+      virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out);
+      virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out);
+
+      // half checkerboard operaions
+      virtual void   Meooe       (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
+
+    private:
+
+      Approx::zolotarev_data *zdata;
+
+      // Cont frac
+      RealD mass;
+      RealD R;
+      RealD scale;
+      std::vector<double> Beta;
+      std::vector<double> cc;;
+      std::vector<double> cc_d;;
+      std::vector<double> sqrt_cc;
+      std::vector<double> See;
+      std::vector<double> Aee;
+
+      // Constructors
+      ContinuedFractionFermion5D(LatticeGaugeField &_Umu,
+				 GridCartesian         &FiveDimGrid,
+				 GridRedBlackCartesian &FiveDimRedBlackGrid,
+				 GridCartesian         &FourDimGrid,
+				 GridRedBlackCartesian &FourDimRedBlackGrid,
+				 RealD _mass,RealD M5);
+
+    };
+
+
+  }
+}
+
+#endif
diff --git a/lib/qcd/action/fermion/DomainWallFermion.h b/lib/qcd/action/fermion/DomainWallFermion.h
new file mode 100644
index 00000000..2abb6eb2
--- /dev/null
+++ b/lib/qcd/action/fermion/DomainWallFermion.h
@@ -0,0 +1,118 @@
+#ifndef  GRID_QCD_DOMAIN_WALL_FERMION_H
+#define  GRID_QCD_DOMAIN_WALL_FERMION_H
+
+#include <Grid.h>
+
+namespace Grid {
+
+  namespace QCD {
+
+    class DomainWallFermion : public CayleyFermion5D
+    {
+    public:
+
+      // Constructors
+      DomainWallFermion(LatticeGaugeField &_Umu,
+			GridCartesian         &FiveDimGrid,
+			GridRedBlackCartesian &FiveDimRedBlackGrid,
+			GridCartesian         &FourDimGrid,
+			GridRedBlackCartesian &FourDimRedBlackGrid,
+			RealD _mass,RealD _M5) : 
+
+      CayleyFermion5D(_Umu,
+		      FiveDimGrid,
+		      FiveDimRedBlackGrid,
+		      FourDimGrid,
+		      FourDimRedBlackGrid,_mass,_M5)
+
+      {
+	RealD eps = 1.0;
+
+	zdata = Approx::grid_higham(eps,this->Ls);// eps is ignored for higham
+	assert(zdata->n==this->Ls);
+ 
+	///////////////////////////////////////////////////////////
+	// The Cayley coeffs (unprec)
+	///////////////////////////////////////////////////////////
+	this->omega.resize(this->Ls);
+	this->bs.resize(this->Ls);
+	this->cs.resize(this->Ls);
+	this->as.resize(this->Ls);
+	
+	for(int i=0; i < this->Ls; i++){
+	  this->as[i] = 1.0;
+	  this->omega[i] = ((double)zdata -> gamma[i]);
+	  double bb=1.0;
+	  this->bs[i] = 0.5*(bb/(this->omega[i]) + 1.0);
+	  this->cs[i] = 0.5*(bb/(this->omega[i]) - 1.0);
+	}
+
+	////////////////////////////////////////////////////////
+	// Constants for the preconditioned matrix Cayley form
+	////////////////////////////////////////////////////////
+	this->bee.resize(this->Ls);
+	this->cee.resize(this->Ls);
+	this->beo.resize(this->Ls);
+	this->ceo.resize(this->Ls);
+
+	for(int i=0;i<this->Ls;i++){
+	  this->bee[i]=as[i]*(bs[i]*(4.0-M5) +1.0);
+	  this->cee[i]=as[i]*(1.0-cs[i]*(4.0-M5));
+	  this->beo[i]=as[i]*bs[i];
+	  this->ceo[i]=-as[i]*cs[i];
+	}
+
+	aee.resize(this->Ls);
+	aeo.resize(this->Ls);
+	for(int i=0;i<this->Ls;i++){
+	  aee[i]=cee[i];
+	  aeo[i]=ceo[i];
+	}
+
+	//////////////////////////////////////////
+	// LDU decomposition of eeoo
+	//////////////////////////////////////////
+	dee.resize(this->Ls);
+	lee.resize(this->Ls);
+	leem.resize(this->Ls);
+	uee.resize(this->Ls);
+	ueem.resize(this->Ls);
+
+	for(int i=0;i<this->Ls;i++){
+	  
+	  dee[i] = bee[i];
+	  
+	  if ( i < this->Ls-1 ) {
+
+	    lee[i] =-cee[i+1]/bee[i]; // sub-diag entry on the ith column
+	    
+	    leem[i]=this->mass*cee[this->Ls-1]/bee[0];
+	    for(int j=0;j<i;j++)  leem[i]*= aee[j]/bee[j+1];
+	    
+	    uee[i] =-aee[i]/bee[i];   // up-diag entry on the ith row
+	    
+	    ueem[i]=this->mass;
+	    for(int j=1;j<=i;j++) ueem[i]*= cee[j]/bee[j];
+	    ueem[i]*= aee[0]/bee[0];
+	    
+	  } else { 
+	    lee[i] =0.0;
+	    leem[i]=0.0;
+	    uee[i] =0.0;
+	    ueem[i]=0.0;
+	  }
+	}
+	
+	{ 
+	  double delta_d=mass*cee[this->Ls-1];
+	  for(int j=0;j<this->Ls-1;j++) delta_d *= cee[j]/bee[j];
+	  dee[this->Ls-1] += delta_d;
+	}
+      }
+
+    };
+
+  }
+}
+
+#endif
diff --git a/lib/qcd/action/fermion/FermionAction.h b/lib/qcd/action/fermion/FermionOperator.h
similarity index 92%
rename from lib/qcd/action/fermion/FermionAction.h
rename to lib/qcd/action/fermion/FermionOperator.h
index 1b05174b..47c47478 100644
--- a/lib/qcd/action/fermion/FermionAction.h
+++ b/lib/qcd/action/fermion/FermionOperator.h
@@ -1,5 +1,5 @@
-#ifndef  GRID_QCD_WILSON_DOP_H
-#define  GRID_QCD_WILSON_DOP_H
+#ifndef  GRID_QCD_FERMION_OPERATOR_H
+#define  GRID_QCD_FERMION_OPERATOR_H
 
 namespace Grid {
 
@@ -11,7 +11,7 @@ namespace Grid {
     // Think about multiple representations
     //////////////////////////////////////////////////////////////////////////////
     template<class FermionField,class GaugeField>
-    class FermionAction : public CheckerBoardedSparseMatrixBase<FermionField>
+    class FermionOperator : public CheckerBoardedSparseMatrixBase<FermionField>
     {
     public:
 
@@ -40,6 +40,7 @@ namespace Grid {
       virtual void DhopOE(const FermionField &in, FermionField &out,int dag)=0;
       virtual void DhopEO(const FermionField &in, FermionField &out,int dag)=0;
 
+
     };
 
   }
diff --git a/lib/qcd/action/fermion/PartialFractionFermion5D.cc b/lib/qcd/action/fermion/PartialFractionFermion5D.cc
new file mode 100644
index 00000000..21f62a75
--- /dev/null
+++ b/lib/qcd/action/fermion/PartialFractionFermion5D.cc
@@ -0,0 +1,47 @@
+#ifndef  GRID_QCD_PARTIAL_FRACTION_H
+#define  GRID_QCD_PARTIAL_FRACTION_H
+
+namespace Grid {
+
+  namespace QCD {
+
+    class PartialFractionFermion5D : public WilsonFermion5D
+    {
+    public:
+
+      // override multiply
+      virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out);
+      virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out);
+
+      // half checkerboard operaions
+      virtual void   Meooe       (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
+
+    private:
+
+      zolotarev_data *zdata;
+
+      // Part frac
+      double R=(1+this->mass)/(1-this->mass);
+      std::vector<double> p; 
+      std::vector<double> q;
+
+      // Constructors
+      PartialFractionFermion5D(LatticeGaugeField &_Umu,
+				    GridCartesian         &FiveDimGrid,
+				    GridRedBlackCartesian &FiveDimRedBlackGrid,
+				    GridCartesian         &FourDimGrid,
+				    GridRedBlackCartesian &FourDimRedBlackGrid,
+				    RealD _mass,RealD M5);
+
+    };
+
+
+  }
+}
+
+#endif
diff --git a/lib/qcd/action/fermion/PartialFractionFermion5D.h b/lib/qcd/action/fermion/PartialFractionFermion5D.h
new file mode 100644
index 00000000..c51aa563
--- /dev/null
+++ b/lib/qcd/action/fermion/PartialFractionFermion5D.h
@@ -0,0 +1,49 @@
+#ifndef  GRID_QCD_PARTIAL_FRACTION_H
+#define  GRID_QCD_PARTIAL_FRACTION_H
+
+namespace Grid {
+
+  namespace QCD {
+
+    class PartialFractionFermion5D : public WilsonFermion5D
+    {
+    public:
+
+      // override multiply
+      virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out);
+      virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out);
+
+      // half checkerboard operaions
+      virtual void   Meooe       (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
+
+    private:
+
+      virtual void PartialFractionCoefficients(void);
+
+      zolotarev_data *zdata;
+
+      // Part frac
+      double R=(1+this->mass)/(1-this->mass);
+      std::vector<double> p; 
+      std::vector<double> q;
+
+      // Constructors
+      PartialFractionFermion5D(LatticeGaugeField &_Umu,
+				    GridCartesian         &FiveDimGrid,
+				    GridRedBlackCartesian &FiveDimRedBlackGrid,
+				    GridCartesian         &FourDimGrid,
+				    GridRedBlackCartesian &FourDimRedBlackGrid,
+				    RealD _mass,RealD M5);
+
+    };
+
+
+  }
+}
+
+#endif
diff --git a/lib/qcd/action/fermion/WilsonFermion.cc b/lib/qcd/action/fermion/WilsonFermion.cc
index aa30a7fa..9f2da251 100644
--- a/lib/qcd/action/fermion/WilsonFermion.cc
+++ b/lib/qcd/action/fermion/WilsonFermion.cc
@@ -9,9 +9,9 @@ const std::vector<int> WilsonFermion::displacements({1,1,1,1,-1,-1,-1,-1});
 int WilsonFermion::HandOptDslash;
 
 WilsonFermion::WilsonFermion(LatticeGaugeField &_Umu,
-			   GridCartesian         &Fgrid,
-			   GridRedBlackCartesian &Hgrid, 
-			   double _mass) :
+			     GridCartesian         &Fgrid,
+			     GridRedBlackCartesian &Hgrid, 
+			     RealD _mass) :
   _grid(&Fgrid),
   _cbgrid(&Hgrid),
   Stencil    (&Fgrid,npoint,Even,directions,displacements),
diff --git a/lib/qcd/action/fermion/WilsonFermion.h b/lib/qcd/action/fermion/WilsonFermion.h
index 5c208131..6040d328 100644
--- a/lib/qcd/action/fermion/WilsonFermion.h
+++ b/lib/qcd/action/fermion/WilsonFermion.h
@@ -5,7 +5,7 @@ namespace Grid {
 
   namespace QCD {
 
-    class WilsonFermion : public FermionAction<LatticeFermion,LatticeGaugeField>
+    class WilsonFermion : public FermionOperator<LatticeFermion,LatticeGaugeField>
     {
     public:
 
@@ -44,7 +44,7 @@ namespace Grid {
 			int dag);
 
       // Constructor
-      WilsonFermion(LatticeGaugeField &_Umu,GridCartesian &Fgrid,GridRedBlackCartesian &Hgrid,double _mass);
+      WilsonFermion(LatticeGaugeField &_Umu,GridCartesian &Fgrid,GridRedBlackCartesian &Hgrid,RealD _mass);
 
       // DoubleStore
       void DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu);
@@ -57,7 +57,7 @@ namespace Grid {
 
     protected:
 
-      double                        mass;
+      RealD                        mass;
 
       GridBase                     *    _grid; 
       GridBase                     *  _cbgrid;
diff --git a/lib/qcd/action/fermion/FiveDimWilsonFermion.cc b/lib/qcd/action/fermion/WilsonFermion5D.cc
similarity index 68%
rename from lib/qcd/action/fermion/FiveDimWilsonFermion.cc
rename to lib/qcd/action/fermion/WilsonFermion5D.cc
index 43645899..d22701b0 100644
--- a/lib/qcd/action/fermion/FiveDimWilsonFermion.cc
+++ b/lib/qcd/action/fermion/WilsonFermion5D.cc
@@ -4,18 +4,18 @@ namespace Grid {
 namespace QCD {
   
   // S-direction is INNERMOST and takes no part in the parity.
-  const std::vector<int> FiveDimWilsonFermion::directions   ({1,2,3,4, 1, 2, 3, 4});
-  const std::vector<int> FiveDimWilsonFermion::displacements({1,1,1,1,-1,-1,-1,-1});
+  const std::vector<int> WilsonFermion5D::directions   ({1,2,3,4, 1, 2, 3, 4});
+  const std::vector<int> WilsonFermion5D::displacements({1,1,1,1,-1,-1,-1,-1});
 
-  int FiveDimWilsonFermion::HandOptDslash;
+  int WilsonFermion5D::HandOptDslash;
 
   // 5d lattice for DWF.
-  FiveDimWilsonFermion::FiveDimWilsonFermion(LatticeGaugeField &_Umu,
+  WilsonFermion5D::WilsonFermion5D(LatticeGaugeField &_Umu,
 					   GridCartesian         &FiveDimGrid,
 					   GridRedBlackCartesian &FiveDimRedBlackGrid,
 					   GridCartesian         &FourDimGrid,
 					   GridRedBlackCartesian &FourDimRedBlackGrid,
-					   double _mass) :
+					   RealD _M5) :
   _FiveDimGrid(&FiveDimGrid),
   _FiveDimRedBlackGrid(&FiveDimRedBlackGrid),
   _FourDimGrid(&FourDimGrid),
@@ -23,7 +23,7 @@ namespace QCD {
   Stencil    (_FiveDimGrid,npoint,Even,directions,displacements),
   StencilEven(_FiveDimRedBlackGrid,npoint,Even,directions,displacements), // source is Even
   StencilOdd (_FiveDimRedBlackGrid,npoint,Odd ,directions,displacements), // source is Odd
-  mass(_mass),
+  M5(_M5),
   Umu(_FourDimGrid),
   UmuEven(_FourDimRedBlackGrid),
   UmuOdd (_FourDimRedBlackGrid),
@@ -70,7 +70,7 @@ namespace QCD {
   pickCheckerboard(Even,UmuEven,Umu);
   pickCheckerboard(Odd ,UmuOdd,Umu);
 }
-void FiveDimWilsonFermion::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu)
+void WilsonFermion5D::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu)
 {
   conformable(Uds._grid,GaugeGrid());
   conformable(Umu._grid,GaugeGrid());
@@ -82,60 +82,9 @@ void FiveDimWilsonFermion::DoubleStore(LatticeDoubledGaugeField &Uds,const Latti
     pokeIndex<LorentzIndex>(Uds,U,mu+4);
   }
 }
-
-RealD FiveDimWilsonFermion::M(const LatticeFermion &in, LatticeFermion &out)
-{
-  out.checkerboard=in.checkerboard;
-  Dhop(in,out,DaggerNo);
-  return axpy_norm(out,5.0-M5,in,out);
-}
-RealD FiveDimWilsonFermion::Mdag(const LatticeFermion &in, LatticeFermion &out)
-{
-  out.checkerboard=in.checkerboard;
-  Dhop(in,out,DaggerYes);
-  return axpy_norm(out,5.0-M5,in,out);
-}
-void FiveDimWilsonFermion::Meooe(const LatticeFermion &in, LatticeFermion &out)
-{
-  if ( in.checkerboard == Odd ) {
-    DhopEO(in,out,DaggerNo);
-  } else {
-    DhopOE(in,out,DaggerNo);
-  }
-}
-void FiveDimWilsonFermion::MeooeDag(const LatticeFermion &in, LatticeFermion &out)
-{
-  if ( in.checkerboard == Odd ) {
-    DhopEO(in,out,DaggerYes);
-  } else {
-    DhopOE(in,out,DaggerYes);
-  }
-}
-void FiveDimWilsonFermion::Mooee(const LatticeFermion &in, LatticeFermion &out)
-{
-  out.checkerboard = in.checkerboard;
-  out = (5.0-M5)*in;
-  return ;
-}
-void FiveDimWilsonFermion::MooeeDag(const LatticeFermion &in, LatticeFermion &out)
-{
-  out.checkerboard = in.checkerboard;
-  Mooee(in,out);
-}
-void FiveDimWilsonFermion::MooeeInv(const LatticeFermion &in, LatticeFermion &out)
-{
-  out.checkerboard = in.checkerboard;
-  out = (1.0/(5.0-M5))*in;
-  return ;
-}
-void FiveDimWilsonFermion::MooeeInvDag(const LatticeFermion &in, LatticeFermion &out)
-{
-  out.checkerboard = in.checkerboard;
-  MooeeInv(in,out);
-}
-void FiveDimWilsonFermion::DhopInternal(CartesianStencil & st, LebesgueOrder &lo,
-					LatticeDoubledGaugeField & U,
-					const LatticeFermion &in, LatticeFermion &out,int dag)
+void WilsonFermion5D::DhopInternal(CartesianStencil & st, LebesgueOrder &lo,
+				   LatticeDoubledGaugeField & U,
+				   const LatticeFermion &in, LatticeFermion &out,int dag)
 {
   assert((dag==DaggerNo) ||(dag==DaggerYes));
 
@@ -150,19 +99,21 @@ void FiveDimWilsonFermion::DhopInternal(CartesianStencil & st, LebesgueOrder &lo
   // - 8 linear access unit stride streams per thread for Fermion for hw prefetchable.
   if ( dag == DaggerYes ) {
     if( HandOptDslash ) {
-      for(int ss=0;ss<U._grid->oSites();ss++){
-	int sU=lo.Reorder(ss);
 PARALLEL_FOR_LOOP
+      for(int ss=0;ss<U._grid->oSites();ss++){
 	for(int s=0;s<Ls;s++){
+	  //int sU=lo.Reorder(ss);
+	  int sU=ss;
 	  int sF = s+Ls*sU;
 	  DiracOptHand::DhopSiteDag(st,U,comm_buf,sF,sU,in,out);
 	}
       }
     } else { 
-      for(int ss=0;ss<U._grid->oSites();ss++){
-	int sU=lo.Reorder(ss);
 PARALLEL_FOR_LOOP
+      for(int ss=0;ss<U._grid->oSites();ss++){
 	for(int s=0;s<Ls;s++){
+	  //	  int sU=lo.Reorder(ss);
+	  int sU=ss;
 	  int sF = s+Ls*sU;
 	  DiracOpt::DhopSiteDag(st,U,comm_buf,sF,sU,in,out);
 	}
@@ -170,21 +121,22 @@ PARALLEL_FOR_LOOP
     }
   } else {
     if( HandOptDslash ) {
-
 PARALLEL_FOR_LOOP
       for(int ss=0;ss<U._grid->oSites();ss++){
-	int sU=lo.Reorder(ss);
 	for(int s=0;s<Ls;s++){
+	  //	  int sU=lo.Reorder(ss);
+	  int sU=ss;
 	  int sF = s+Ls*sU;
 	  DiracOptHand::DhopSite(st,U,comm_buf,sF,sU,in,out);
 	}
       }
 
     } else { 
-      for(int ss=0;ss<U._grid->oSites();ss++){
-	int sU=lo.Reorder(ss);
 PARALLEL_FOR_LOOP
+      for(int ss=0;ss<U._grid->oSites();ss++){
 	for(int s=0;s<Ls;s++){
+	  //	  int sU=lo.Reorder(ss);
+	  int sU=ss;
 	  int sF = s+Ls*sU; 
 	  DiracOpt::DhopSite(st,U,comm_buf,sF,sU,in,out);
 	}
@@ -192,7 +144,7 @@ PARALLEL_FOR_LOOP
     }
   }
 }
-void FiveDimWilsonFermion::DhopOE(const LatticeFermion &in, LatticeFermion &out,int dag)
+void WilsonFermion5D::DhopOE(const LatticeFermion &in, LatticeFermion &out,int dag)
 {
   conformable(in._grid,FermionRedBlackGrid());    // verifies half grid
   conformable(in._grid,out._grid); // drops the cb check
@@ -202,7 +154,7 @@ void FiveDimWilsonFermion::DhopOE(const LatticeFermion &in, LatticeFermion &out,
 
   DhopInternal(StencilEven,LebesgueEvenOdd,UmuOdd,in,out,dag);
 }
-void FiveDimWilsonFermion::DhopEO(const LatticeFermion &in, LatticeFermion &out,int dag)
+void WilsonFermion5D::DhopEO(const LatticeFermion &in, LatticeFermion &out,int dag)
 {
   conformable(in._grid,FermionRedBlackGrid());    // verifies half grid
   conformable(in._grid,out._grid); // drops the cb check
@@ -212,7 +164,7 @@ void FiveDimWilsonFermion::DhopEO(const LatticeFermion &in, LatticeFermion &out,
 
   DhopInternal(StencilOdd,LebesgueEvenOdd,UmuEven,in,out,dag);
 }
-void FiveDimWilsonFermion::Dhop(const LatticeFermion &in, LatticeFermion &out,int dag)
+void WilsonFermion5D::Dhop(const LatticeFermion &in, LatticeFermion &out,int dag)
 {
   conformable(in._grid,FermionGrid()); // verifies full grid
   conformable(in._grid,out._grid);
@@ -221,8 +173,14 @@ void FiveDimWilsonFermion::Dhop(const LatticeFermion &in, LatticeFermion &out,in
 
   DhopInternal(Stencil,Lebesgue,Umu,in,out,dag);
 }
-
-}}
+void WilsonFermion5D::DW(const LatticeFermion &in, LatticeFermion &out,int dag)
+{
+  out.checkerboard=in.checkerboard;
+  Dhop(in,out,dag); // -0.5 is included
+  axpy(out,4.0-M5,in,out);
+}
+}
+}
 
 
 
diff --git a/lib/qcd/action/fermion/FiveDimWilsonFermion.h b/lib/qcd/action/fermion/WilsonFermion5D.h
similarity index 72%
rename from lib/qcd/action/fermion/FiveDimWilsonFermion.h
rename to lib/qcd/action/fermion/WilsonFermion5D.h
index 3aa85023..d4777d01 100644
--- a/lib/qcd/action/fermion/FiveDimWilsonFermion.h
+++ b/lib/qcd/action/fermion/WilsonFermion5D.h
@@ -15,7 +15,7 @@ namespace Grid {
     //
     // [DIFFERS from original CPS red black implementation parity = (x+y+z+t+s)|2 ]
     ////////////////////////////////////////////////////////////////////////////////
-    class FiveDimWilsonFermion : public FermionAction<LatticeFermion,LatticeGaugeField>
+    class WilsonFermion5D : public FermionOperator<LatticeFermion,LatticeGaugeField>
     {
     public:
       ///////////////////////////////////////////////////////////////
@@ -26,19 +26,21 @@ namespace Grid {
       GridBase *FermionGrid(void)            { return _FiveDimGrid;}
       GridBase *FermionRedBlackGrid(void)    { return _FiveDimRedBlackGrid;}
 
-      // override multiply
-      virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out);
-      virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out);
+      // full checkerboard operations; leave unimplemented as abstract for now
+      //virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out)=0;
+      //virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out)=0;
 
-      // half checkerboard operaions
-      virtual void   Meooe       (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
+      // half checkerboard operations; leave unimplemented as abstract for now
+      //      virtual void   Meooe       (const LatticeFermion &in, LatticeFermion &out)=0;
+      //      virtual void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out)=0;
+      //      virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out)=0;
+      //      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out)=0;
+      //      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out)=0;
+      //      virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out)=0;
 
-      // non-hermitian hopping term; half cb or both
+      // Implement hopping term non-hermitian hopping term; half cb or both
+      // Implement s-diagonal DW
+      void DW    (const LatticeFermion &in, LatticeFermion &out,int dag);
       void Dhop  (const LatticeFermion &in, LatticeFermion &out,int dag);
       void DhopOE(const LatticeFermion &in, LatticeFermion &out,int dag);
       void DhopEO(const LatticeFermion &in, LatticeFermion &out,int dag);
@@ -54,12 +56,12 @@ namespace Grid {
 			int dag);
 
       // Constructors
-      FiveDimWilsonFermion(LatticeGaugeField &_Umu,
+      WilsonFermion5D(LatticeGaugeField &_Umu,
 			  GridCartesian         &FiveDimGrid,
 			  GridRedBlackCartesian &FiveDimRedBlackGrid,
 			  GridCartesian         &FourDimGrid,
 			  GridRedBlackCartesian &FourDimRedBlackGrid,
-			  double _mass);
+			  double _M5);
 
       // DoubleStore
       void DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu);
@@ -82,7 +84,6 @@ namespace Grid {
       static const std::vector<int> displacements;
 
       double                        M5;
-      double                        mass;
       int Ls;
 
       //Defines the stencils for even and odd
diff --git a/lib/stencil/Grid_stencil_common.cc b/lib/stencil/Grid_stencil_common.cc
index f0f8c581..7f894faf 100644
--- a/lib/stencil/Grid_stencil_common.cc
+++ b/lib/stencil/Grid_stencil_common.cc
@@ -52,8 +52,8 @@ namespace Grid {
 	// up a table containing the npoint "neighbours" and whether they 
 	// live in lattice or a comms buffer.
 	if ( !comm_dim ) {
-	  sshift[0] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,Even);
-	  sshift[1] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,Odd);
+	  sshift[0] = _grid->CheckerBoardShiftForCB(_checkerboard,dimension,shift,Even);
+	  sshift[1] = _grid->CheckerBoardShiftForCB(_checkerboard,dimension,shift,Odd);
 
 	  if ( sshift[0] == sshift[1] ) {
 	    Local(point,dimension,shift,0x3);
@@ -63,8 +63,8 @@ namespace Grid {
 	  }
 	} else { // All permute extract done in comms phase prior to Stencil application
 	  //        So tables are the same whether comm_dim or splice_dim
-	  sshift[0] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,Even);
-	  sshift[1] = _grid->CheckerBoardShift(_checkerboard,dimension,shift,Odd);
+	  sshift[0] = _grid->CheckerBoardShiftForCB(_checkerboard,dimension,shift,Even);
+	  sshift[1] = _grid->CheckerBoardShiftForCB(_checkerboard,dimension,shift,Odd);
 	  if ( sshift[0] == sshift[1] ) {
 	    Comms(point,dimension,shift,0x3);
 	  } else {
@@ -96,7 +96,7 @@ namespace Grid {
 	
 	int cb= (cbmask==0x2)? Odd : Even;
 	  
-	int sshift = _grid->CheckerBoardShift(_checkerboard,dimension,shift,cb);
+	int sshift = _grid->CheckerBoardShiftForCB(_checkerboard,dimension,shift,cb);
 	int sx     = (x+sshift)%rd;
 	  
 	int permute_slice=0;
@@ -134,7 +134,7 @@ namespace Grid {
                                            // send to one or more remote nodes.
 
       int cb= (cbmask==0x2)? Odd : Even;
-      int sshift= _grid->CheckerBoardShift(_checkerboard,dimension,shift,cb);
+      int sshift= _grid->CheckerBoardShiftForCB(_checkerboard,dimension,shift,cb);
       
       for(int x=0;x<rd;x++){       
 	

From 35fdba81dd5bf8546b9d93dcba869a7ba1ed915a Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 2 Jun 2015 17:25:26 +0100
Subject: [PATCH 257/429] Reorg; moving prec/unprec/schur CG for Wilson and DWF
 into tests as these are really tests and not benchmarks (no performance
 reports, only convergence test).

---
 benchmarks/Makefile.am                        | 39 ++-------------
 {benchmarks => tests}/Grid_dwf_cg_prec.cc     |  0
 {benchmarks => tests}/Grid_dwf_cg_schur.cc    |  0
 {benchmarks => tests}/Grid_dwf_cg_unprec.cc   |  0
 {benchmarks => tests}/Grid_dwf_even_odd.cc    |  0
 {benchmarks => tests}/Grid_wilson_cg_prec.cc  |  0
 {benchmarks => tests}/Grid_wilson_cg_schur.cc |  0
 .../Grid_wilson_cg_unprec.cc                  |  0
 {benchmarks => tests}/Grid_wilson_evenodd.cc  |  0
 tests/Makefile.am                             | 50 ++++++++++++++++++-
 10 files changed, 52 insertions(+), 37 deletions(-)
 rename {benchmarks => tests}/Grid_dwf_cg_prec.cc (100%)
 rename {benchmarks => tests}/Grid_dwf_cg_schur.cc (100%)
 rename {benchmarks => tests}/Grid_dwf_cg_unprec.cc (100%)
 rename {benchmarks => tests}/Grid_dwf_even_odd.cc (100%)
 rename {benchmarks => tests}/Grid_wilson_cg_prec.cc (100%)
 rename {benchmarks => tests}/Grid_wilson_cg_schur.cc (100%)
 rename {benchmarks => tests}/Grid_wilson_cg_unprec.cc (100%)
 rename {benchmarks => tests}/Grid_wilson_evenodd.cc (100%)

diff --git a/benchmarks/Makefile.am b/benchmarks/Makefile.am
index e92f2e84..e1e092f4 100644
--- a/benchmarks/Makefile.am
+++ b/benchmarks/Makefile.am
@@ -10,53 +10,20 @@ bin_PROGRAMS = \
 	Grid_memory_bandwidth \
 	Grid_su3 \
 	Grid_wilson \
-	Grid_wilson_evenodd \
-	Grid_wilson_cg_unprec \
-	Grid_wilson_cg_prec \
-	Grid_wilson_cg_schur \
-	Grid_dwf\
-	Grid_dwf_even_odd\
-	Grid_dwf_cg_unprec\
-	Grid_dwf_cg_prec\
-	Grid_dwf_cg_schur
+	Grid_dwf
 
 Grid_comms_SOURCES = Grid_comms.cc
 Grid_comms_LDADD = -lGrid
 
-Grid_su3_SOURCES = Grid_su3.cc Grid_su3_test.cc Grid_su3_expr.cc
-Grid_su3_LDADD = -lGrid
-
 Grid_memory_bandwidth_SOURCES = Grid_memory_bandwidth.cc
 Grid_memory_bandwidth_LDADD = -lGrid
 
+Grid_su3_SOURCES = Grid_su3.cc Grid_su3_test.cc Grid_su3_expr.cc
+Grid_su3_LDADD = -lGrid
 
 Grid_wilson_SOURCES = Grid_wilson.cc
 Grid_wilson_LDADD = -lGrid
 
-Grid_wilson_evenodd_SOURCES = Grid_wilson_evenodd.cc
-Grid_wilson_evenodd_LDADD = -lGrid
-
-Grid_wilson_cg_unprec_SOURCES = Grid_wilson_cg_unprec.cc
-Grid_wilson_cg_unprec_LDADD = -lGrid
-
-Grid_wilson_cg_prec_SOURCES = Grid_wilson_cg_prec.cc
-Grid_wilson_cg_prec_LDADD = -lGrid
-
-Grid_wilson_cg_schur_SOURCES = Grid_wilson_cg_schur.cc
-Grid_wilson_cg_schur_LDADD = -lGrid
-
 Grid_dwf_SOURCES = Grid_dwf.cc
 Grid_dwf_LDADD = -lGrid
 
-Grid_dwf_even_odd_SOURCES = Grid_dwf_even_odd.cc
-Grid_dwf_even_odd_LDADD = -lGrid
-
-Grid_dwf_cg_unprec_SOURCES = Grid_dwf_cg_unprec.cc
-Grid_dwf_cg_unprec_LDADD = -lGrid
-
-Grid_dwf_cg_prec_SOURCES = Grid_dwf_cg_prec.cc
-Grid_dwf_cg_prec_LDADD = -lGrid
-
-Grid_dwf_cg_schur_SOURCES = Grid_dwf_cg_schur.cc
-Grid_dwf_cg_schur_LDADD = -lGrid
-
diff --git a/benchmarks/Grid_dwf_cg_prec.cc b/tests/Grid_dwf_cg_prec.cc
similarity index 100%
rename from benchmarks/Grid_dwf_cg_prec.cc
rename to tests/Grid_dwf_cg_prec.cc
diff --git a/benchmarks/Grid_dwf_cg_schur.cc b/tests/Grid_dwf_cg_schur.cc
similarity index 100%
rename from benchmarks/Grid_dwf_cg_schur.cc
rename to tests/Grid_dwf_cg_schur.cc
diff --git a/benchmarks/Grid_dwf_cg_unprec.cc b/tests/Grid_dwf_cg_unprec.cc
similarity index 100%
rename from benchmarks/Grid_dwf_cg_unprec.cc
rename to tests/Grid_dwf_cg_unprec.cc
diff --git a/benchmarks/Grid_dwf_even_odd.cc b/tests/Grid_dwf_even_odd.cc
similarity index 100%
rename from benchmarks/Grid_dwf_even_odd.cc
rename to tests/Grid_dwf_even_odd.cc
diff --git a/benchmarks/Grid_wilson_cg_prec.cc b/tests/Grid_wilson_cg_prec.cc
similarity index 100%
rename from benchmarks/Grid_wilson_cg_prec.cc
rename to tests/Grid_wilson_cg_prec.cc
diff --git a/benchmarks/Grid_wilson_cg_schur.cc b/tests/Grid_wilson_cg_schur.cc
similarity index 100%
rename from benchmarks/Grid_wilson_cg_schur.cc
rename to tests/Grid_wilson_cg_schur.cc
diff --git a/benchmarks/Grid_wilson_cg_unprec.cc b/tests/Grid_wilson_cg_unprec.cc
similarity index 100%
rename from benchmarks/Grid_wilson_cg_unprec.cc
rename to tests/Grid_wilson_cg_unprec.cc
diff --git a/benchmarks/Grid_wilson_evenodd.cc b/tests/Grid_wilson_evenodd.cc
similarity index 100%
rename from benchmarks/Grid_wilson_evenodd.cc
rename to tests/Grid_wilson_evenodd.cc
diff --git a/tests/Makefile.am b/tests/Makefile.am
index 7ef4eb1f..6f92a94d 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -5,7 +5,30 @@ AM_LDFLAGS = -L$(top_builddir)/lib
 #
 # Test code
 #
-bin_PROGRAMS = Grid_main Grid_stencil Grid_nersc_io Grid_cshift Grid_gamma  Grid_simd Grid_rng Grid_remez Grid_rng_fixed Grid_cshift_red_black 
+bin_PROGRAMS = Grid_main \
+	Grid_simd \
+	Grid_gamma  \
+	Grid_cshift \
+	Grid_cshift_red_black \
+	Grid_stencil \
+	Grid_nersc_io \
+	Grid_rng \
+	Grid_remez \
+	Grid_rng_fixed \
+	Grid_wilson_evenodd \
+	Grid_wilson_cg_unprec \
+	Grid_wilson_cg_prec \
+	Grid_wilson_cg_schur \
+	Grid_dwf_even_odd\
+	Grid_dwf_cg_unprec\
+	Grid_dwf_cg_prec\
+	Grid_dwf_cg_schur
+
+test: 
+	for f in $bin_PROGRAMS
+	do
+	./$f > $f.log
+	done
 
 Grid_main_SOURCES = Grid_main.cc
 Grid_main_LDADD = -lGrid
@@ -39,3 +62,28 @@ Grid_simd_LDADD = -lGrid
 
 #Grid_simd_new_SOURCES = Grid_simd_new.cc
 #Grid_simd_new_LDADD = -lGrid
+
+Grid_wilson_evenodd_SOURCES = Grid_wilson_evenodd.cc
+Grid_wilson_evenodd_LDADD = -lGrid
+
+Grid_wilson_cg_unprec_SOURCES = Grid_wilson_cg_unprec.cc
+Grid_wilson_cg_unprec_LDADD = -lGrid
+
+Grid_wilson_cg_prec_SOURCES = Grid_wilson_cg_prec.cc
+Grid_wilson_cg_prec_LDADD = -lGrid
+
+Grid_wilson_cg_schur_SOURCES = Grid_wilson_cg_schur.cc
+Grid_wilson_cg_schur_LDADD = -lGrid
+
+Grid_dwf_even_odd_SOURCES = Grid_dwf_even_odd.cc
+Grid_dwf_even_odd_LDADD = -lGrid
+
+Grid_dwf_cg_unprec_SOURCES = Grid_dwf_cg_unprec.cc
+Grid_dwf_cg_unprec_LDADD = -lGrid
+
+Grid_dwf_cg_prec_SOURCES = Grid_dwf_cg_prec.cc
+Grid_dwf_cg_prec_LDADD = -lGrid
+
+Grid_dwf_cg_schur_SOURCES = Grid_dwf_cg_schur.cc
+Grid_dwf_cg_schur_LDADD = -lGrid
+

From 59163862420a81a5e9d41bd11eae2c5c329e96b9 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 3 Jun 2015 09:36:26 +0100
Subject: [PATCH 258/429] Mobius Caley form, Mobius Zolotarev operators. Pass
 Even Odd vs unprec test and hermiticity checks in tests/Grid_any_evenodd.cc;
 will work on inversion tests shortly.

---
 lib/qcd/action/Actions.h                      |   8 +-
 lib/qcd/action/fermion/CayleyFermion5D.cc     | 110 ++++++++-
 lib/qcd/action/fermion/CayleyFermion5D.h      |   6 +-
 lib/qcd/action/fermion/DomainWallFermion.h    |  83 +------
 lib/qcd/action/fermion/MobiusFermion.h        |  46 ++++
 .../action/fermion/MobiusZolotarevFermion.h   |  48 ++++
 lib/qcd/action/fermion/ScaledShamir.h         |  51 ++++
 lib/qcd/action/fermion/WilsonFermion5D.h      |  12 +
 tests/Grid_any_evenodd.cc                     | 226 ++++++++++++++++++
 tests/Makefile.am                             |  12 +-
 10 files changed, 507 insertions(+), 95 deletions(-)
 create mode 100644 lib/qcd/action/fermion/MobiusFermion.h
 create mode 100644 lib/qcd/action/fermion/MobiusZolotarevFermion.h
 create mode 100644 lib/qcd/action/fermion/ScaledShamir.h
 create mode 100644 tests/Grid_any_evenodd.cc

diff --git a/lib/qcd/action/Actions.h b/lib/qcd/action/Actions.h
index acbf027c..d37d1cd4 100644
--- a/lib/qcd/action/Actions.h
+++ b/lib/qcd/action/Actions.h
@@ -38,6 +38,9 @@
 //#include <qcd/action/fermion/PartialFraction.h>
 
 #include <qcd/action/fermion/DomainWallFermion.h>
+#include <qcd/action/fermion/DomainWallFermion.h>
+#include <qcd/action/fermion/MobiusFermion.h>
+#include <qcd/action/fermion/MobiusZolotarevFermion.h>
 //#include <qcd/action/fermion/ScaledShamirCayleyTanh.h>
 
 
@@ -70,11 +73,6 @@
 
   class LinearGaugeAction : public GaugeAction< multi1d<LatticeColorMatrix>, multi1d<LatticeColorMatrix> >
   typedef multi1d<LatticeColorMatrix>  P;
-  typedef multi1d<LatticeColorMatrix>  Q;
-  virtual void staple(LatticeColorMatrix& result,
-		      const Handle< GaugeState<P,Q> >& state,
-		      int mu, int cb) const = 0;
     */
 
-
 #endif
diff --git a/lib/qcd/action/fermion/CayleyFermion5D.cc b/lib/qcd/action/fermion/CayleyFermion5D.cc
index 263cc28b..be528e79 100644
--- a/lib/qcd/action/fermion/CayleyFermion5D.cc
+++ b/lib/qcd/action/fermion/CayleyFermion5D.cc
@@ -15,7 +15,6 @@ namespace QCD {
 		   FourDimRedBlackGrid,_M5),
    mass(_mass)
  {
-   std::cout << "Constructing a CayleyFermion5D"<<std::endl;
  }
 
   // override multiply
@@ -229,7 +228,112 @@ namespace QCD {
       axpby_ssp_pplus (chi,1.0,chi,-lee[s],chi,s,s+1);  // chi[Ls]
     }
   }
+  
+  void CayleyFermion5D::SetCoefficients(RealD scale,Approx::zolotarev_data *zdata,RealD b,RealD c)
+  {
+
+    ///////////////////////////////////////////////////////////
+    // The Cayley coeffs (unprec)
+    ///////////////////////////////////////////////////////////
+    omega.resize(Ls);
+    bs.resize(Ls);
+    cs.resize(Ls);
+    as.resize(Ls);
+    
+    // 
+    // Ts = (    [bs+cs]Dw        )^-1 (    (bs+cs) Dw         )
+    //     -(g5  -------       -1 )    ( g5 ---------     + 1  )
+    //      (   {2+(bs-cs)Dw}     )    (    2+(bs-cs) Dw       )
+    //
+    //  bs = 1/2( (1/omega_s + 1)*b + (1/omega - 1)*c ) = 1/2(  1/omega(b+c) + (b-c) )
+    //  cs = 1/2( (1/omega_s - 1)*b + (1/omega + 1)*c ) = 1/2(  1/omega(b+c) - (b-c) )
+    //
+    // bs+cs = 0.5*( 1/omega(b+c) + (b-c) + 1/omega(b+c) - (b-c) ) = 1/omega(b+c)
+    // bs-cs = 0.5*( 1/omega(b+c) + (b-c) - 1/omega(b+c) + (b-c) ) = b-c
+    //
+    // So 
+    //
+    // Ts = (    [b+c]Dw/omega_s    )^-1 (    (b+c) Dw /omega_s        )
+    //     -(g5  -------         -1 )    ( g5 ---------           + 1  )
+    //      (   {2+(b-c)Dw}         )    (    2+(b-c) Dw               )
+    //
+    // Ts = (    [b+c]Dw            )^-1 (    (b+c) Dw                 )
+    //     -(g5  -------    -omega_s)    ( g5 ---------      + omega_s )
+    //      (   {2+(b-c)Dw}         )    (    2+(b-c) Dw               )
+    // 
+    
+    double bpc = b+c;
+    double bmc = b-c;
+    for(int i=0; i < Ls; i++){
+      as[i] = 1.0;
+      omega[i] = ((double)zdata->gamma[i]); //NB reciprocal relative to Chroma NEF code
+      bs[i] = 0.5*(bpc/omega[i] + bmc);
+      cs[i] = 0.5*(bpc/omega[i] - bmc);
+    }
+
+    ////////////////////////////////////////////////////////
+    // Constants for the preconditioned matrix Cayley form
+    ////////////////////////////////////////////////////////
+    bee.resize(Ls);
+    cee.resize(Ls);
+    beo.resize(Ls);
+    ceo.resize(Ls);
+    
+    for(int i=0;i<Ls;i++){
+      bee[i]=as[i]*(bs[i]*(4.0-M5) +1.0);
+      cee[i]=as[i]*(1.0-cs[i]*(4.0-M5));
+      beo[i]=as[i]*bs[i];
+      ceo[i]=-as[i]*cs[i];
+    }
+
+    aee.resize(Ls);
+    aeo.resize(Ls);
+    for(int i=0;i<Ls;i++){
+      aee[i]=cee[i];
+      aeo[i]=ceo[i];
+    }
+
+    //////////////////////////////////////////
+    // LDU decomposition of eeoo
+    //////////////////////////////////////////
+    dee.resize(Ls);
+    lee.resize(Ls);
+    leem.resize(Ls);
+    uee.resize(Ls);
+    ueem.resize(Ls);
+    
+    for(int i=0;i<Ls;i++){
+      
+      dee[i] = bee[i];
+      
+      if ( i < Ls-1 ) {
+	
+	lee[i] =-cee[i+1]/bee[i]; // sub-diag entry on the ith column
+	    
+	leem[i]=mass*cee[Ls-1]/bee[0];
+	for(int j=0;j<i;j++)  leem[i]*= aee[j]/bee[j+1];
+	
+	uee[i] =-aee[i]/bee[i];   // up-diag entry on the ith row
+	
+	ueem[i]=mass;
+	for(int j=1;j<=i;j++) ueem[i]*= cee[j]/bee[j];
+	ueem[i]*= aee[0]/bee[0];
+	    
+      } else { 
+	lee[i] =0.0;
+	leem[i]=0.0;
+	uee[i] =0.0;
+	ueem[i]=0.0;
+      }
+    }
+	
+    { 
+      double delta_d=mass*cee[Ls-1];
+      for(int j=0;j<Ls-1;j++) delta_d *= cee[j]/bee[j];
+      dee[Ls-1] += delta_d;
+    }
+  }
+
+}}
 
-}
-}
 
diff --git a/lib/qcd/action/fermion/CayleyFermion5D.h b/lib/qcd/action/fermion/CayleyFermion5D.h
index 857ac124..57c71992 100644
--- a/lib/qcd/action/fermion/CayleyFermion5D.h
+++ b/lib/qcd/action/fermion/CayleyFermion5D.h
@@ -22,10 +22,8 @@ namespace Grid {
       virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
 
       //    protected:
-
-      Approx::zolotarev_data *zdata;
-
       RealD mass;
+
       // Cayley form Moebius (tanh and zolotarev)
       std::vector<RealD> omega; 
       std::vector<RealD> bs;    // S dependent coeffs
@@ -53,6 +51,8 @@ namespace Grid {
 		      GridRedBlackCartesian &FourDimRedBlackGrid,
 		      RealD _mass,RealD _M5);
 
+    protected:
+      void SetCoefficients(RealD scale,Approx::zolotarev_data *zdata,RealD b,RealD c);
     };
 
   }
diff --git a/lib/qcd/action/fermion/DomainWallFermion.h b/lib/qcd/action/fermion/DomainWallFermion.h
index 2abb6eb2..3e6a9739 100644
--- a/lib/qcd/action/fermion/DomainWallFermion.h
+++ b/lib/qcd/action/fermion/DomainWallFermion.h
@@ -28,86 +28,13 @@ namespace Grid {
       {
 	RealD eps = 1.0;
 
-	zdata = Approx::grid_higham(eps,this->Ls);// eps is ignored for higham
+	Approx::zolotarev_data *zdata = Approx::grid_higham(eps,this->Ls);// eps is ignored for higham
 	assert(zdata->n==this->Ls);
+	
+	std::cout << "DomainWallFermion with Ls="<<Ls<<std::endl;
+	// Call base setter
+	this->CayleyFermion5D::SetCoefficients(1.0,zdata,1.0,0.0);
  
-	///////////////////////////////////////////////////////////
-	// The Cayley coeffs (unprec)
-	///////////////////////////////////////////////////////////
-	this->omega.resize(this->Ls);
-	this->bs.resize(this->Ls);
-	this->cs.resize(this->Ls);
-	this->as.resize(this->Ls);
-	
-	for(int i=0; i < this->Ls; i++){
-	  this->as[i] = 1.0;
-	  this->omega[i] = ((double)zdata -> gamma[i]);
-	  double bb=1.0;
-	  this->bs[i] = 0.5*(bb/(this->omega[i]) + 1.0);
-	  this->cs[i] = 0.5*(bb/(this->omega[i]) - 1.0);
-	}
-
-	////////////////////////////////////////////////////////
-	// Constants for the preconditioned matrix Cayley form
-	////////////////////////////////////////////////////////
-	this->bee.resize(this->Ls);
-	this->cee.resize(this->Ls);
-	this->beo.resize(this->Ls);
-	this->ceo.resize(this->Ls);
-
-	for(int i=0;i<this->Ls;i++){
-	  this->bee[i]=as[i]*(bs[i]*(4.0-M5) +1.0);
-	  this->cee[i]=as[i]*(1.0-cs[i]*(4.0-M5));
-	  this->beo[i]=as[i]*bs[i];
-	  this->ceo[i]=-as[i]*cs[i];
-	}
-
-	aee.resize(this->Ls);
-	aeo.resize(this->Ls);
-	for(int i=0;i<this->Ls;i++){
-	  aee[i]=cee[i];
-	  aeo[i]=ceo[i];
-	}
-
-	//////////////////////////////////////////
-	// LDU decomposition of eeoo
-	//////////////////////////////////////////
-	dee.resize(this->Ls);
-	lee.resize(this->Ls);
-	leem.resize(this->Ls);
-	uee.resize(this->Ls);
-	ueem.resize(this->Ls);
-
-	for(int i=0;i<this->Ls;i++){
-	  
-	  dee[i] = bee[i];
-	  
-	  if ( i < this->Ls-1 ) {
-
-	    lee[i] =-cee[i+1]/bee[i]; // sub-diag entry on the ith column
-	    
-	    leem[i]=this->mass*cee[this->Ls-1]/bee[0];
-	    for(int j=0;j<i;j++)  leem[i]*= aee[j]/bee[j+1];
-	    
-	    uee[i] =-aee[i]/bee[i];   // up-diag entry on the ith row
-	    
-	    ueem[i]=this->mass;
-	    for(int j=1;j<=i;j++) ueem[i]*= cee[j]/bee[j];
-	    ueem[i]*= aee[0]/bee[0];
-	    
-	  } else { 
-	    lee[i] =0.0;
-	    leem[i]=0.0;
-	    uee[i] =0.0;
-	    ueem[i]=0.0;
-	  }
-	}
-	
-	{ 
-	  double delta_d=mass*cee[this->Ls-1];
-	  for(int j=0;j<this->Ls-1;j++) delta_d *= cee[j]/bee[j];
-	  dee[this->Ls-1] += delta_d;
-	}
       }
 
     };
diff --git a/lib/qcd/action/fermion/MobiusFermion.h b/lib/qcd/action/fermion/MobiusFermion.h
new file mode 100644
index 00000000..4c291fad
--- /dev/null
+++ b/lib/qcd/action/fermion/MobiusFermion.h
@@ -0,0 +1,46 @@
+#ifndef  GRID_QCD_MOBIUS_FERMION_H
+#define  GRID_QCD_MOBIUS_FERMION_H
+
+#include <Grid.h>
+
+namespace Grid {
+
+  namespace QCD {
+
+    class MobiusFermion : public CayleyFermion5D
+    {
+    public:
+
+      // Constructors
+      MobiusFermion(LatticeGaugeField &_Umu,
+		    GridCartesian         &FiveDimGrid,
+		    GridRedBlackCartesian &FiveDimRedBlackGrid,
+		    GridCartesian         &FourDimGrid,
+		    GridRedBlackCartesian &FourDimRedBlackGrid,
+		    RealD _mass,RealD _M5,
+		    RealD b, RealD c) : 
+      
+      CayleyFermion5D(_Umu,
+		      FiveDimGrid,
+		      FiveDimRedBlackGrid,
+		      FourDimGrid,
+		      FourDimRedBlackGrid,_mass,_M5)
+
+      {
+	RealD eps = 1.0;
+
+	std::cout << "MobiusFermion (b="<<b<<",c="<<c<<") with Ls= "<<Ls<<" Tanh approx"<<std::endl;
+	Approx::zolotarev_data *zdata = Approx::grid_higham(eps,this->Ls);// eps is ignored for higham
+	assert(zdata->n==this->Ls);
+	
+	// Call base setter
+	this->CayleyFermion5D::SetCoefficients(1.0,zdata,b,c);
+ 
+      }
+
+    };
+
+  }
+}
+
+#endif
diff --git a/lib/qcd/action/fermion/MobiusZolotarevFermion.h b/lib/qcd/action/fermion/MobiusZolotarevFermion.h
new file mode 100644
index 00000000..866d0c39
--- /dev/null
+++ b/lib/qcd/action/fermion/MobiusZolotarevFermion.h
@@ -0,0 +1,48 @@
+#ifndef  GRID_QCD_MOBIUS_ZOLOTAREV_FERMION_H
+#define  GRID_QCD_MOBIUS_ZOLOTAREV_FERMION_H
+
+#include <Grid.h>
+
+namespace Grid {
+
+  namespace QCD {
+
+    class MobiusZolotarevFermion : public CayleyFermion5D
+    {
+    public:
+
+      // Constructors
+       MobiusZolotarevFermion(LatticeGaugeField &_Umu,
+			      GridCartesian         &FiveDimGrid,
+			      GridRedBlackCartesian &FiveDimRedBlackGrid,
+			      GridCartesian         &FourDimGrid,
+			      GridRedBlackCartesian &FourDimRedBlackGrid,
+			      RealD _mass,RealD _M5,
+			      RealD b, RealD c,
+			      RealD lo, RealD hi) : 
+      
+      CayleyFermion5D(_Umu,
+		      FiveDimGrid,
+		      FiveDimRedBlackGrid,
+		      FourDimGrid,
+		      FourDimRedBlackGrid,_mass,_M5)
+
+      {
+	RealD eps = lo/hi;
+
+	Approx::zolotarev_data *zdata = Approx::grid_zolotarev(eps,this->Ls,0);// eps is ignored for higham
+	assert(zdata->n==this->Ls);
+
+	std::cout << "MobiusZolotarevFermion (b="<<b<<",c="<<c<<") with Ls= "<<Ls<<" Zolotarev range ["<<lo<<","<<hi<<"]"<<std::endl;
+	
+	// Call base setter
+	this->CayleyFermion5D::SetCoefficients(1.0,zdata,b,c);
+ 
+      }
+
+    };
+
+  }
+}
+
+#endif
diff --git a/lib/qcd/action/fermion/ScaledShamir.h b/lib/qcd/action/fermion/ScaledShamir.h
new file mode 100644
index 00000000..a1fd33d0
--- /dev/null
+++ b/lib/qcd/action/fermion/ScaledShamir.h
@@ -0,0 +1,51 @@
+#ifndef  GRID_QCD_DOMAIN_WALL_FERMION_H
+#define  GRID_QCD_DOMAIN_WALL_FERMION_H
+
+#include <Grid.h>
+
+namespace Grid {
+
+  namespace QCD {
+
+    class ScaledShamirFermion : public CayleyFermion5D
+    {
+    public:
+
+      // Constructors
+      ScaledShamirFermion(LatticeGaugeField &_Umu,
+			  GridCartesian         &FiveDimGrid,
+			  GridRedBlackCartesian &FiveDimRedBlackGrid,
+			  GridCartesian         &FourDimGrid,
+			  GridRedBlackCartesian &FourDimRedBlackGrid,
+			  RealD _mass,RealD _M5, RealD scale) : 
+      
+      CayleyFermion5D(_Umu,
+		      FiveDimGrid,
+		      FiveDimRedBlackGrid,
+		      FourDimGrid,
+		      FourDimRedBlackGrid,_mass,_M5,
+		      RealD b, 
+		      RealD c)
+
+      {
+	RealD eps = 1.0;
+
+	Approx::zolotarev_data *zdata = Approx::grid_higham(eps,this->Ls);// eps is ignored for higham
+	assert(zdata->n==this->Ls);
+	
+	//b+c = scale;
+	//b-c = 1
+	//b   = 0.5(scale+1);
+	//c   = 0.5(scale-1);
+	
+	// Call base setter
+	this->CayleyFermion5D::SetCoefficients(1.0,zdata,0.5*(scale+1.0),0.5*(scale-1.0));
+
+       }
+
+    };
+
+  }
+}
+
+#endif
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.h b/lib/qcd/action/fermion/WilsonFermion5D.h
index d4777d01..062c4d82 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.h
+++ b/lib/qcd/action/fermion/WilsonFermion5D.h
@@ -14,6 +14,18 @@ namespace Grid {
     // i.e. even even contains fifth dim hopping term.
     //
     // [DIFFERS from original CPS red black implementation parity = (x+y+z+t+s)|2 ]
+    ////////////////////////////
+    //ContFrac:
+    //  Ls always odd. Rational poly deg is either Ls or Ls-1
+    //PartFrac 
+    //  Ls always odd. Rational poly deg is either Ls or Ls-1
+    //
+    //Cayley: Ls always even, Rational poly deg is Ls
+    // 
+    // Just set nrational as Ls. Forget about Ls-1 cases.
+    //
+    // Require odd Ls for cont and part frac
+    ////////////////////////////
     ////////////////////////////////////////////////////////////////////////////////
     class WilsonFermion5D : public FermionOperator<LatticeFermion,LatticeGaugeField>
     {
diff --git a/tests/Grid_any_evenodd.cc b/tests/Grid_any_evenodd.cc
new file mode 100644
index 00000000..8d8580c1
--- /dev/null
+++ b/tests/Grid_any_evenodd.cc
@@ -0,0 +1,226 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+
+template<class What> 
+void  TestWhat(What & Ddwf,
+	       GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+	       GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+	       RealD mass, RealD M5,
+	       GridParallelRNG *RNG4,   GridParallelRNG *RNG5);
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  int threads = GridThread::GetThreads();
+  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+
+  const int Ls=8;
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);  RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);  RNG4.SeedFixedIntegers(seeds4);
+
+  LatticeGaugeField Umu(UGrid); random(RNG4,Umu);
+  std::vector<LatticeColourMatrix> U(4,UGrid);
+
+  RealD mass=0.1;
+  RealD M5  =1.8;
+  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  TestWhat<DomainWallFermion>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  RealD b=1.5;// Scale factor b+c=2, b-c=1
+  RealD c=0.5;
+  MobiusFermion Dmob(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c);
+  TestWhat<MobiusFermion>(Dmob,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  MobiusZolotarevFermion Dzolo(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c,0.1,4.0);
+  TestWhat<MobiusZolotarevFermion>(Dzolo,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+
+  Grid_finalize();
+}
+
+template<class What> 
+void  TestWhat(What & Ddwf, 
+	       GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+	       GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+	       RealD mass, RealD M5,
+	       GridParallelRNG *RNG4,
+	       GridParallelRNG *RNG5)
+{
+
+  LatticeFermion src   (FGrid); random(*RNG5,src);
+  LatticeFermion phi   (FGrid); random(*RNG5,phi);
+  LatticeFermion chi   (FGrid); random(*RNG5,chi);
+  LatticeFermion result(FGrid); result=zero;
+  LatticeFermion    ref(FGrid);    ref=zero;
+  LatticeFermion    tmp(FGrid);    tmp=zero;
+  LatticeFermion    err(FGrid);    tmp=zero;
+
+  LatticeFermion src_e (FrbGrid);
+  LatticeFermion src_o (FrbGrid);
+  LatticeFermion r_e   (FrbGrid);
+  LatticeFermion r_o   (FrbGrid);
+  LatticeFermion r_eo  (FGrid);
+  LatticeFermion r_eeoo(FGrid);
+
+  std::cout<<"=========================================================="<<std::endl;
+  std::cout<<"= Testing that Meo + Moe + Moo + Mee = Munprec "<<std::endl;
+  std::cout<<"=========================================================="<<std::endl;
+
+  pickCheckerboard(Even,src_e,src);
+  pickCheckerboard(Odd,src_o,src);
+
+  Ddwf.Meooe(src_e,r_o);  std::cout<<"Applied Meo"<<std::endl;
+  Ddwf.Meooe(src_o,r_e);  std::cout<<"Applied Moe"<<std::endl;
+  setCheckerboard(r_eo,r_o);
+  setCheckerboard(r_eo,r_e);
+
+  Ddwf.Mooee(src_e,r_e);  std::cout<<"Applied Mee"<<std::endl;
+  Ddwf.Mooee(src_o,r_o);  std::cout<<"Applied Moo"<<std::endl;
+  setCheckerboard(r_eeoo,r_e);
+  setCheckerboard(r_eeoo,r_o);
+
+  r_eo=r_eo+r_eeoo;
+  Ddwf.M(src,ref);  
+
+  //  std::cout << r_eo<<std::endl;
+  //  std::cout << ref <<std::endl;
+
+  err= ref - r_eo;
+  std::cout << "EO norm diff   "<< norm2(err)<< " "<<norm2(ref)<< " " << norm2(r_eo) <<std::endl;
+    
+  LatticeComplex cerr(FGrid);
+  cerr = localInnerProduct(err,err);
+  //  std::cout << cerr<<std::endl;
+
+  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<"= Test Ddagger is the dagger of D by requiring                "<<std::endl;
+  std::cout<<"=  < phi | Deo | chi > * = < chi | Deo^dag| phi>  "<<std::endl;
+  std::cout<<"=============================================================="<<std::endl;
+  
+  LatticeFermion chi_e   (FrbGrid);
+  LatticeFermion chi_o   (FrbGrid);
+
+  LatticeFermion dchi_e  (FrbGrid);
+  LatticeFermion dchi_o  (FrbGrid);
+
+  LatticeFermion phi_e   (FrbGrid);
+  LatticeFermion phi_o   (FrbGrid);
+
+  LatticeFermion dphi_e  (FrbGrid);
+  LatticeFermion dphi_o  (FrbGrid);
+
+
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+  pickCheckerboard(Even,phi_e,phi);
+  pickCheckerboard(Odd ,phi_o,phi);
+
+  Ddwf.Meooe(chi_e,dchi_o);
+  Ddwf.Meooe(chi_o,dchi_e);
+  Ddwf.MeooeDag(phi_e,dphi_o);
+  Ddwf.MeooeDag(phi_o,dphi_e);
+
+  ComplexD pDce = innerProduct(phi_e,dchi_e);
+  ComplexD pDco = innerProduct(phi_o,dchi_o);
+  ComplexD cDpe = innerProduct(chi_e,dphi_e);
+  ComplexD cDpo = innerProduct(chi_o,dphi_o);
+
+  std::cout <<"e "<<pDce<<" "<<cDpe <<std::endl;
+  std::cout <<"o "<<pDco<<" "<<cDpo <<std::endl;
+
+  std::cout <<"pDce - conj(cDpo) "<< pDce-conj(cDpo) <<std::endl;
+  std::cout <<"pDco - conj(cDpe) "<< pDco-conj(cDpe) <<std::endl;
+
+  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<"= Test MeeInv Mee = 1                                         "<<std::endl;
+  std::cout<<"=============================================================="<<std::endl;
+
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+
+  Ddwf.Mooee(chi_e,src_e);
+  Ddwf.MooeeInv(src_e,phi_e);
+
+  Ddwf.Mooee(chi_o,src_o);
+  Ddwf.MooeeInv(src_o,phi_o);
+  
+  setCheckerboard(phi,phi_e);
+  setCheckerboard(phi,phi_o);
+
+  err = phi-chi;
+  std::cout << "norm diff   "<< norm2(err)<< std::endl;
+
+  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<"= Test MeeInvDag MeeDag = 1                                   "<<std::endl;
+  std::cout<<"=============================================================="<<std::endl;
+
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+
+  Ddwf.MooeeDag(chi_e,src_e);
+  Ddwf.MooeeInvDag(src_e,phi_e);
+
+  Ddwf.MooeeDag(chi_o,src_o);
+  Ddwf.MooeeInvDag(src_o,phi_o);
+  
+  setCheckerboard(phi,phi_e);
+  setCheckerboard(phi,phi_o);
+
+  err = phi-chi;
+  std::cout << "norm diff   "<< norm2(err)<< std::endl;
+
+  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<"= Test MpcDagMpc is Hermitian              "<<std::endl;
+  std::cout<<"=============================================================="<<std::endl;
+  
+  random(*RNG5,phi);
+  random(*RNG5,chi);
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+  pickCheckerboard(Even,phi_e,phi);
+  pickCheckerboard(Odd ,phi_o,phi);
+  RealD t1,t2;
+
+  Ddwf.MpcDagMpc(chi_e,dchi_e,t1,t2);
+  Ddwf.MpcDagMpc(chi_o,dchi_o,t1,t2);
+
+  Ddwf.MpcDagMpc(phi_e,dphi_e,t1,t2);
+  Ddwf.MpcDagMpc(phi_o,dphi_o,t1,t2);
+
+  pDce = innerProduct(phi_e,dchi_e);
+  pDco = innerProduct(phi_o,dchi_o);
+  cDpe = innerProduct(chi_e,dphi_e);
+  cDpo = innerProduct(chi_o,dphi_o);
+
+  std::cout <<"e "<<pDce<<" "<<cDpe <<std::endl;
+  std::cout <<"o "<<pDco<<" "<<cDpo <<std::endl;
+
+  std::cout <<"pDce - conj(cDpo) "<< pDco-conj(cDpo) <<std::endl;
+  std::cout <<"pDco - conj(cDpe) "<< pDce-conj(cDpe) <<std::endl;
+  
+}
diff --git a/tests/Makefile.am b/tests/Makefile.am
index 6f92a94d..f8f5df29 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -22,13 +22,10 @@ bin_PROGRAMS = Grid_main \
 	Grid_dwf_even_odd\
 	Grid_dwf_cg_unprec\
 	Grid_dwf_cg_prec\
-	Grid_dwf_cg_schur
+	Grid_dwf_cg_schur\
+	Grid_any_evenodd
+
 
-test: 
-	for f in $bin_PROGRAMS
-	do
-	./$f > $f.log
-	done
 
 Grid_main_SOURCES = Grid_main.cc
 Grid_main_LDADD = -lGrid
@@ -66,6 +63,9 @@ Grid_simd_LDADD = -lGrid
 Grid_wilson_evenodd_SOURCES = Grid_wilson_evenodd.cc
 Grid_wilson_evenodd_LDADD = -lGrid
 
+Grid_any_evenodd_SOURCES = Grid_any_evenodd.cc
+Grid_any_evenodd_LDADD = -lGrid
+
 Grid_wilson_cg_unprec_SOURCES = Grid_wilson_cg_unprec.cc
 Grid_wilson_cg_unprec_LDADD = -lGrid
 

From 343d039b378647f494a9a15f66f14b8e0b1ab4db Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 3 Jun 2015 09:51:06 +0100
Subject: [PATCH 259/429] Scaled Shamir and Scaled Shamir Zolotarev aliases for
 special cases of Mobius.

---
 lib/qcd/action/Actions.h                      |  3 +-
 lib/qcd/action/fermion/ScaledShamir.h         | 51 -------------------
 lib/qcd/action/fermion/ScaledShamirFermion.h  | 37 ++++++++++++++
 .../fermion/ScaledShamirZolotarevFermion.h    | 39 ++++++++++++++
 tests/Grid_any_evenodd.cc                     |  8 ++-
 5 files changed, 85 insertions(+), 53 deletions(-)
 delete mode 100644 lib/qcd/action/fermion/ScaledShamir.h
 create mode 100644 lib/qcd/action/fermion/ScaledShamirFermion.h
 create mode 100644 lib/qcd/action/fermion/ScaledShamirZolotarevFermion.h

diff --git a/lib/qcd/action/Actions.h b/lib/qcd/action/Actions.h
index d37d1cd4..893564e0 100644
--- a/lib/qcd/action/Actions.h
+++ b/lib/qcd/action/Actions.h
@@ -41,7 +41,8 @@
 #include <qcd/action/fermion/DomainWallFermion.h>
 #include <qcd/action/fermion/MobiusFermion.h>
 #include <qcd/action/fermion/MobiusZolotarevFermion.h>
-//#include <qcd/action/fermion/ScaledShamirCayleyTanh.h>
+#include <qcd/action/fermion/ScaledShamirFermion.h>
+#include <qcd/action/fermion/ScaledShamirZolotarevFermion.h>
 
 
     // Chroma interface defining FermionAction
diff --git a/lib/qcd/action/fermion/ScaledShamir.h b/lib/qcd/action/fermion/ScaledShamir.h
deleted file mode 100644
index a1fd33d0..00000000
--- a/lib/qcd/action/fermion/ScaledShamir.h
+++ /dev/null
@@ -1,51 +0,0 @@
-#ifndef  GRID_QCD_DOMAIN_WALL_FERMION_H
-#define  GRID_QCD_DOMAIN_WALL_FERMION_H
-
-#include <Grid.h>
-
-namespace Grid {
-
-  namespace QCD {
-
-    class ScaledShamirFermion : public CayleyFermion5D
-    {
-    public:
-
-      // Constructors
-      ScaledShamirFermion(LatticeGaugeField &_Umu,
-			  GridCartesian         &FiveDimGrid,
-			  GridRedBlackCartesian &FiveDimRedBlackGrid,
-			  GridCartesian         &FourDimGrid,
-			  GridRedBlackCartesian &FourDimRedBlackGrid,
-			  RealD _mass,RealD _M5, RealD scale) : 
-      
-      CayleyFermion5D(_Umu,
-		      FiveDimGrid,
-		      FiveDimRedBlackGrid,
-		      FourDimGrid,
-		      FourDimRedBlackGrid,_mass,_M5,
-		      RealD b, 
-		      RealD c)
-
-      {
-	RealD eps = 1.0;
-
-	Approx::zolotarev_data *zdata = Approx::grid_higham(eps,this->Ls);// eps is ignored for higham
-	assert(zdata->n==this->Ls);
-	
-	//b+c = scale;
-	//b-c = 1
-	//b   = 0.5(scale+1);
-	//c   = 0.5(scale-1);
-	
-	// Call base setter
-	this->CayleyFermion5D::SetCoefficients(1.0,zdata,0.5*(scale+1.0),0.5*(scale-1.0));
-
-       }
-
-    };
-
-  }
-}
-
-#endif
diff --git a/lib/qcd/action/fermion/ScaledShamirFermion.h b/lib/qcd/action/fermion/ScaledShamirFermion.h
new file mode 100644
index 00000000..59fb16a8
--- /dev/null
+++ b/lib/qcd/action/fermion/ScaledShamirFermion.h
@@ -0,0 +1,37 @@
+#ifndef  GRID_QCD_SCALED_SHAMIR_FERMION_H
+#define  GRID_QCD_SCALED_SHAMIR_FERMION_H
+
+#include <Grid.h>
+
+namespace Grid {
+
+  namespace QCD {
+
+    class ScaledShamirFermion : public MobiusFermion
+    {
+    public:
+
+      // Constructors
+    ScaledShamirFermion(LatticeGaugeField &_Umu,
+			GridCartesian         &FiveDimGrid,
+			GridRedBlackCartesian &FiveDimRedBlackGrid,
+			GridCartesian         &FourDimGrid,
+			GridRedBlackCartesian &FourDimRedBlackGrid,
+			RealD _mass,RealD _M5,
+			RealD scale) :
+      
+      // b+c=scale, b-c = 1 <=> 2b = scale+1; 2c = scale-1
+      MobiusFermion(_Umu,
+		    FiveDimGrid,
+		    FiveDimRedBlackGrid,
+		    FourDimGrid,
+		    FourDimRedBlackGrid,_mass,_M5,0.5*(scale+1.0),0.5*(scale-1.0))
+      {
+      }
+
+    };
+
+  }
+}
+
+#endif
diff --git a/lib/qcd/action/fermion/ScaledShamirZolotarevFermion.h b/lib/qcd/action/fermion/ScaledShamirZolotarevFermion.h
new file mode 100644
index 00000000..a4c88d5d
--- /dev/null
+++ b/lib/qcd/action/fermion/ScaledShamirZolotarevFermion.h
@@ -0,0 +1,39 @@
+#ifndef  GRID_QCD_SCALED_SHAMIR_ZOLOTAREV_FERMION_H
+#define  GRID_QCD_SCALED_SHAMIR_ZOLOTAREV_FERMION_H
+
+#include <Grid.h>
+
+namespace Grid {
+
+  namespace QCD {
+
+    class ScaledShamirZolotarevFermion : public MobiusZolotarevFermion
+    {
+    public:
+
+      // Constructors
+
+
+    ScaledShamirZolotarevFermion(LatticeGaugeField &_Umu,
+				 GridCartesian         &FiveDimGrid,
+				 GridRedBlackCartesian &FiveDimRedBlackGrid,
+				 GridCartesian         &FourDimGrid,
+				 GridRedBlackCartesian &FourDimRedBlackGrid,
+				 RealD _mass,RealD _M5,
+				 RealD scale,
+				 RealD lo, RealD hi) : 
+      
+      MobiusZolotarevFermion(_Umu,
+			       FiveDimGrid,
+			       FiveDimRedBlackGrid,
+			       FourDimGrid,
+			       FourDimRedBlackGrid,_mass,_M5,0.5*(scale+1.0),0.5*(scale-1.0),lo,hi)
+
+      {}
+
+    };
+
+  }
+}
+
+#endif
diff --git a/tests/Grid_any_evenodd.cc b/tests/Grid_any_evenodd.cc
index 8d8580c1..de1e3c78 100644
--- a/tests/Grid_any_evenodd.cc
+++ b/tests/Grid_any_evenodd.cc
@@ -56,9 +56,15 @@ int main (int argc, char ** argv)
   MobiusFermion Dmob(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c);
   TestWhat<MobiusFermion>(Dmob,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
-  MobiusZolotarevFermion Dzolo(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c,0.1,4.0);
+  MobiusZolotarevFermion Dzolo(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c,0.1,2.0);
   TestWhat<MobiusZolotarevFermion>(Dzolo,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
+  ScaledShamirFermion Dsham(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,2.0);
+  TestWhat<ScaledShamirFermion>(Dsham,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  ScaledShamirZolotarevFermion Dshamz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,2.0,0.1,2.0);
+  TestWhat<ScaledShamirZolotarevFermion>(Dshamz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
 
   Grid_finalize();
 }

From 26e9b04fab8c9e6db1b902ef0a9564785dea172b Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 3 Jun 2015 10:54:03 +0100
Subject: [PATCH 260/429] CG test written and passes i.e. converges with small
 true residual in RedBlack MpcDagMpc, Unprec MdagM and Schur red black solver
 for each of.

DomainWallFermion
MobiusFermion
MobiusZolotarevFermion
ScaledShamirFermion
ScaledShamirZolotarevFermion
---
 .../{Grid_comms.cc => Benchmark_comms.cc}     |   0
 benchmarks/{Grid_dwf.cc => Benchmark_dwf.cc}  |   0
 ...width.cc => Benchmark_memory_bandwidth.cc} |   0
 benchmarks/{Grid_su3.cc => Benchmark_su3.cc}  |   0
 ...Grid_su3_expr.cc => Benchmark_su3_expr.cc} |   0
 ...Grid_su3_test.cc => Benchmark_su3_test.cc} |   0
 .../{Grid_wilson.cc => Benchmark_wilson.cc}   |   0
 benchmarks/Makefile.am                        |  30 ++--
 tests/Makefile.am                             | 121 ++++++-------
 tests/{Grid_cshift.cc => Test_cshift.cc}      |   0
 ..._red_black.cc => Test_cshift_red_black.cc} |   0
 ...rid_dwf_cg_prec.cc => Test_dwf_cg_prec.cc} |   0
 ...d_dwf_cg_schur.cc => Test_dwf_cg_schur.cc} |   0
 ...dwf_cg_unprec.cc => Test_dwf_cg_unprec.cc} |   0
 ...d_dwf_even_odd.cc => Test_dwf_even_odd.cc} |   0
 tests/{Grid_gamma.cc => Test_gamma.cc}        |   0
 tests/{Grid_main.cc => Test_main.cc}          |   0
 tests/Test_many_cg.cc                         | 164 ++++++++++++++++++
 ...id_any_evenodd.cc => Test_many_evenodd.cc} |   0
 tests/{Grid_nersc_io.cc => Test_nersc_io.cc}  |   0
 tests/{Grid_remez.cc => Test_remez.cc}        |   0
 tests/{Grid_rng.cc => Test_rng.cc}            |   0
 .../{Grid_rng_fixed.cc => Test_rng_fixed.cc}  |   0
 tests/{Grid_simd.cc => Test_simd.cc}          |   0
 tests/{Grid_simd_new.cc => Test_simd_new.cc}  |   0
 tests/{Grid_stencil.cc => Test_stencil.cc}    |   0
 ...lson_cg_prec.cc => Test_wilson_cg_prec.cc} |   0
 ...on_cg_schur.cc => Test_wilson_cg_schur.cc} |   0
 ..._cg_unprec.cc => Test_wilson_cg_unprec.cc} |   0
 ...lson_evenodd.cc => Test_wilson_evenodd.cc} |   0
 30 files changed, 241 insertions(+), 74 deletions(-)
 rename benchmarks/{Grid_comms.cc => Benchmark_comms.cc} (100%)
 rename benchmarks/{Grid_dwf.cc => Benchmark_dwf.cc} (100%)
 rename benchmarks/{Grid_memory_bandwidth.cc => Benchmark_memory_bandwidth.cc} (100%)
 rename benchmarks/{Grid_su3.cc => Benchmark_su3.cc} (100%)
 rename benchmarks/{Grid_su3_expr.cc => Benchmark_su3_expr.cc} (100%)
 rename benchmarks/{Grid_su3_test.cc => Benchmark_su3_test.cc} (100%)
 rename benchmarks/{Grid_wilson.cc => Benchmark_wilson.cc} (100%)
 rename tests/{Grid_cshift.cc => Test_cshift.cc} (100%)
 rename tests/{Grid_cshift_red_black.cc => Test_cshift_red_black.cc} (100%)
 rename tests/{Grid_dwf_cg_prec.cc => Test_dwf_cg_prec.cc} (100%)
 rename tests/{Grid_dwf_cg_schur.cc => Test_dwf_cg_schur.cc} (100%)
 rename tests/{Grid_dwf_cg_unprec.cc => Test_dwf_cg_unprec.cc} (100%)
 rename tests/{Grid_dwf_even_odd.cc => Test_dwf_even_odd.cc} (100%)
 rename tests/{Grid_gamma.cc => Test_gamma.cc} (100%)
 rename tests/{Grid_main.cc => Test_main.cc} (100%)
 create mode 100644 tests/Test_many_cg.cc
 rename tests/{Grid_any_evenodd.cc => Test_many_evenodd.cc} (100%)
 rename tests/{Grid_nersc_io.cc => Test_nersc_io.cc} (100%)
 rename tests/{Grid_remez.cc => Test_remez.cc} (100%)
 rename tests/{Grid_rng.cc => Test_rng.cc} (100%)
 rename tests/{Grid_rng_fixed.cc => Test_rng_fixed.cc} (100%)
 rename tests/{Grid_simd.cc => Test_simd.cc} (100%)
 rename tests/{Grid_simd_new.cc => Test_simd_new.cc} (100%)
 rename tests/{Grid_stencil.cc => Test_stencil.cc} (100%)
 rename tests/{Grid_wilson_cg_prec.cc => Test_wilson_cg_prec.cc} (100%)
 rename tests/{Grid_wilson_cg_schur.cc => Test_wilson_cg_schur.cc} (100%)
 rename tests/{Grid_wilson_cg_unprec.cc => Test_wilson_cg_unprec.cc} (100%)
 rename tests/{Grid_wilson_evenodd.cc => Test_wilson_evenodd.cc} (100%)

diff --git a/benchmarks/Grid_comms.cc b/benchmarks/Benchmark_comms.cc
similarity index 100%
rename from benchmarks/Grid_comms.cc
rename to benchmarks/Benchmark_comms.cc
diff --git a/benchmarks/Grid_dwf.cc b/benchmarks/Benchmark_dwf.cc
similarity index 100%
rename from benchmarks/Grid_dwf.cc
rename to benchmarks/Benchmark_dwf.cc
diff --git a/benchmarks/Grid_memory_bandwidth.cc b/benchmarks/Benchmark_memory_bandwidth.cc
similarity index 100%
rename from benchmarks/Grid_memory_bandwidth.cc
rename to benchmarks/Benchmark_memory_bandwidth.cc
diff --git a/benchmarks/Grid_su3.cc b/benchmarks/Benchmark_su3.cc
similarity index 100%
rename from benchmarks/Grid_su3.cc
rename to benchmarks/Benchmark_su3.cc
diff --git a/benchmarks/Grid_su3_expr.cc b/benchmarks/Benchmark_su3_expr.cc
similarity index 100%
rename from benchmarks/Grid_su3_expr.cc
rename to benchmarks/Benchmark_su3_expr.cc
diff --git a/benchmarks/Grid_su3_test.cc b/benchmarks/Benchmark_su3_test.cc
similarity index 100%
rename from benchmarks/Grid_su3_test.cc
rename to benchmarks/Benchmark_su3_test.cc
diff --git a/benchmarks/Grid_wilson.cc b/benchmarks/Benchmark_wilson.cc
similarity index 100%
rename from benchmarks/Grid_wilson.cc
rename to benchmarks/Benchmark_wilson.cc
diff --git a/benchmarks/Makefile.am b/benchmarks/Makefile.am
index e1e092f4..74ff03c6 100644
--- a/benchmarks/Makefile.am
+++ b/benchmarks/Makefile.am
@@ -6,24 +6,24 @@ AM_LDFLAGS = -L$(top_builddir)/lib
 # Test code
 #
 bin_PROGRAMS = \
-	Grid_comms \
-	Grid_memory_bandwidth \
-	Grid_su3 \
-	Grid_wilson \
-	Grid_dwf
+	Benchmark_comms \
+	Benchmark_memory_bandwidth \
+	Benchmark_su3 \
+	Benchmark_wilson \
+	Benchmark_dwf
 
-Grid_comms_SOURCES = Grid_comms.cc
-Grid_comms_LDADD = -lGrid
+Benchmark_comms_SOURCES = Benchmark_comms.cc
+Benchmark_comms_LDADD = -lGrid
 
-Grid_memory_bandwidth_SOURCES = Grid_memory_bandwidth.cc
-Grid_memory_bandwidth_LDADD = -lGrid
+Benchmark_memory_bandwidth_SOURCES = Benchmark_memory_bandwidth.cc
+Benchmark_memory_bandwidth_LDADD = -lGrid
 
-Grid_su3_SOURCES = Grid_su3.cc Grid_su3_test.cc Grid_su3_expr.cc
-Grid_su3_LDADD = -lGrid
+Benchmark_su3_SOURCES = Benchmark_su3.cc Benchmark_su3_test.cc Benchmark_su3_expr.cc
+Benchmark_su3_LDADD = -lGrid
 
-Grid_wilson_SOURCES = Grid_wilson.cc
-Grid_wilson_LDADD = -lGrid
+Benchmark_wilson_SOURCES = Benchmark_wilson.cc
+Benchmark_wilson_LDADD = -lGrid
 
-Grid_dwf_SOURCES = Grid_dwf.cc
-Grid_dwf_LDADD = -lGrid
+Benchmark_dwf_SOURCES = Benchmark_dwf.cc
+Benchmark_dwf_LDADD = -lGrid
 
diff --git a/tests/Makefile.am b/tests/Makefile.am
index f8f5df29..bb3448f1 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -5,85 +5,88 @@ AM_LDFLAGS = -L$(top_builddir)/lib
 #
 # Test code
 #
-bin_PROGRAMS = Grid_main \
-	Grid_simd \
-	Grid_gamma  \
-	Grid_cshift \
-	Grid_cshift_red_black \
-	Grid_stencil \
-	Grid_nersc_io \
-	Grid_rng \
-	Grid_remez \
-	Grid_rng_fixed \
-	Grid_wilson_evenodd \
-	Grid_wilson_cg_unprec \
-	Grid_wilson_cg_prec \
-	Grid_wilson_cg_schur \
-	Grid_dwf_even_odd\
-	Grid_dwf_cg_unprec\
-	Grid_dwf_cg_prec\
-	Grid_dwf_cg_schur\
-	Grid_any_evenodd
+bin_PROGRAMS = Test_main \
+	Test_simd \
+	Test_gamma  \
+	Test_cshift \
+	Test_cshift_red_black \
+	Test_stencil \
+	Test_nersc_io \
+	Test_rng \
+	Test_remez \
+	Test_rng_fixed \
+	Test_wilson_evenodd \
+	Test_wilson_cg_unprec \
+	Test_wilson_cg_prec \
+	Test_wilson_cg_schur \
+	Test_dwf_even_odd\
+	Test_dwf_cg_unprec\
+	Test_dwf_cg_prec\
+	Test_dwf_cg_schur\
+	Test_many_evenodd\
+	Test_many_cg
 
 
 
-Grid_main_SOURCES = Grid_main.cc
-Grid_main_LDADD = -lGrid
+Test_main_SOURCES = Test_main.cc
+Test_main_LDADD = -lGrid
 
-Grid_rng_SOURCES = Grid_rng.cc
-Grid_rng_LDADD = -lGrid
+Test_rng_SOURCES = Test_rng.cc
+Test_rng_LDADD = -lGrid
 
-Grid_rng_fixed_SOURCES = Grid_rng_fixed.cc
-Grid_rng_fixed_LDADD = -lGrid
+Test_rng_fixed_SOURCES = Test_rng_fixed.cc
+Test_rng_fixed_LDADD = -lGrid
 
-Grid_remez_SOURCES = Grid_remez.cc
-Grid_remez_LDADD = -lGrid
+Test_remez_SOURCES = Test_remez.cc
+Test_remez_LDADD = -lGrid
 
-Grid_nersc_io_SOURCES = Grid_nersc_io.cc
-Grid_nersc_io_LDADD = -lGrid
+Test_nersc_io_SOURCES = Test_nersc_io.cc
+Test_nersc_io_LDADD = -lGrid
 
-Grid_cshift_SOURCES = Grid_cshift.cc
-Grid_cshift_LDADD = -lGrid
+Test_cshift_SOURCES = Test_cshift.cc
+Test_cshift_LDADD = -lGrid
 
-Grid_cshift_red_black_SOURCES = Grid_cshift_red_black.cc
-Grid_cshift_red_black_LDADD = -lGrid
+Test_cshift_red_black_SOURCES = Test_cshift_red_black.cc
+Test_cshift_red_black_LDADD = -lGrid
 
-Grid_gamma_SOURCES = Grid_gamma.cc
-Grid_gamma_LDADD = -lGrid
+Test_gamma_SOURCES = Test_gamma.cc
+Test_gamma_LDADD = -lGrid
 
-Grid_stencil_SOURCES = Grid_stencil.cc
-Grid_stencil_LDADD = -lGrid
+Test_stencil_SOURCES = Test_stencil.cc
+Test_stencil_LDADD = -lGrid
 
-Grid_simd_SOURCES = Grid_simd.cc
-Grid_simd_LDADD = -lGrid
+Test_simd_SOURCES = Test_simd.cc
+Test_simd_LDADD = -lGrid
 
-#Grid_simd_new_SOURCES = Grid_simd_new.cc
-#Grid_simd_new_LDADD = -lGrid
+#Test_simd_new_SOURCES = Test_simd_new.cc
+#Test_simd_new_LDADD = -lGrid
 
-Grid_wilson_evenodd_SOURCES = Grid_wilson_evenodd.cc
-Grid_wilson_evenodd_LDADD = -lGrid
+Test_wilson_evenodd_SOURCES = Test_wilson_evenodd.cc
+Test_wilson_evenodd_LDADD = -lGrid
 
-Grid_any_evenodd_SOURCES = Grid_any_evenodd.cc
-Grid_any_evenodd_LDADD = -lGrid
+Test_wilson_cg_unprec_SOURCES = Test_wilson_cg_unprec.cc
+Test_wilson_cg_unprec_LDADD = -lGrid
 
-Grid_wilson_cg_unprec_SOURCES = Grid_wilson_cg_unprec.cc
-Grid_wilson_cg_unprec_LDADD = -lGrid
+Test_wilson_cg_prec_SOURCES = Test_wilson_cg_prec.cc
+Test_wilson_cg_prec_LDADD = -lGrid
 
-Grid_wilson_cg_prec_SOURCES = Grid_wilson_cg_prec.cc
-Grid_wilson_cg_prec_LDADD = -lGrid
+Test_wilson_cg_schur_SOURCES = Test_wilson_cg_schur.cc
+Test_wilson_cg_schur_LDADD = -lGrid
 
-Grid_wilson_cg_schur_SOURCES = Grid_wilson_cg_schur.cc
-Grid_wilson_cg_schur_LDADD = -lGrid
+Test_dwf_even_odd_SOURCES = Test_dwf_even_odd.cc
+Test_dwf_even_odd_LDADD = -lGrid
 
-Grid_dwf_even_odd_SOURCES = Grid_dwf_even_odd.cc
-Grid_dwf_even_odd_LDADD = -lGrid
+Test_dwf_cg_unprec_SOURCES = Test_dwf_cg_unprec.cc
+Test_dwf_cg_unprec_LDADD = -lGrid
 
-Grid_dwf_cg_unprec_SOURCES = Grid_dwf_cg_unprec.cc
-Grid_dwf_cg_unprec_LDADD = -lGrid
+Test_dwf_cg_prec_SOURCES = Test_dwf_cg_prec.cc
+Test_dwf_cg_prec_LDADD = -lGrid
 
-Grid_dwf_cg_prec_SOURCES = Grid_dwf_cg_prec.cc
-Grid_dwf_cg_prec_LDADD = -lGrid
+Test_dwf_cg_schur_SOURCES = Test_dwf_cg_schur.cc
+Test_dwf_cg_schur_LDADD = -lGrid
 
-Grid_dwf_cg_schur_SOURCES = Grid_dwf_cg_schur.cc
-Grid_dwf_cg_schur_LDADD = -lGrid
+Test_many_evenodd_SOURCES = Test_many_evenodd.cc
+Test_many_evenodd_LDADD = -lGrid
 
+Test_many_cg_SOURCES = Test_many_cg.cc
+Test_many_cg_LDADD = -lGrid
diff --git a/tests/Grid_cshift.cc b/tests/Test_cshift.cc
similarity index 100%
rename from tests/Grid_cshift.cc
rename to tests/Test_cshift.cc
diff --git a/tests/Grid_cshift_red_black.cc b/tests/Test_cshift_red_black.cc
similarity index 100%
rename from tests/Grid_cshift_red_black.cc
rename to tests/Test_cshift_red_black.cc
diff --git a/tests/Grid_dwf_cg_prec.cc b/tests/Test_dwf_cg_prec.cc
similarity index 100%
rename from tests/Grid_dwf_cg_prec.cc
rename to tests/Test_dwf_cg_prec.cc
diff --git a/tests/Grid_dwf_cg_schur.cc b/tests/Test_dwf_cg_schur.cc
similarity index 100%
rename from tests/Grid_dwf_cg_schur.cc
rename to tests/Test_dwf_cg_schur.cc
diff --git a/tests/Grid_dwf_cg_unprec.cc b/tests/Test_dwf_cg_unprec.cc
similarity index 100%
rename from tests/Grid_dwf_cg_unprec.cc
rename to tests/Test_dwf_cg_unprec.cc
diff --git a/tests/Grid_dwf_even_odd.cc b/tests/Test_dwf_even_odd.cc
similarity index 100%
rename from tests/Grid_dwf_even_odd.cc
rename to tests/Test_dwf_even_odd.cc
diff --git a/tests/Grid_gamma.cc b/tests/Test_gamma.cc
similarity index 100%
rename from tests/Grid_gamma.cc
rename to tests/Test_gamma.cc
diff --git a/tests/Grid_main.cc b/tests/Test_main.cc
similarity index 100%
rename from tests/Grid_main.cc
rename to tests/Test_main.cc
diff --git a/tests/Test_many_cg.cc b/tests/Test_many_cg.cc
new file mode 100644
index 00000000..3241a025
--- /dev/null
+++ b/tests/Test_many_cg.cc
@@ -0,0 +1,164 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+template<class What> 
+void  TestCGinversions(What & Ddwf, 
+		       GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+		       GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+		       RealD mass, RealD M5,
+		       GridParallelRNG *RNG4,
+		       GridParallelRNG *RNG5);
+template<class What> 
+void  TestCGschur(What & Ddwf, 
+		  GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+		  GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+		  RealD mass, RealD M5,
+		  GridParallelRNG *RNG4,
+		  GridParallelRNG *RNG5);
+
+template<class What> 
+void  TestCGunprec(What & Ddwf, 
+		   GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+		   GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+		   RealD mass, RealD M5,
+		   GridParallelRNG *RNG4,
+		   GridParallelRNG *RNG5);
+
+template<class What> 
+void  TestCGprec(What & Ddwf, 
+		 GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+		 GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+		 RealD mass, RealD M5,
+		 GridParallelRNG *RNG4,
+		 GridParallelRNG *RNG5);
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  int threads = GridThread::GetThreads();
+  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+
+  const int Ls=8;
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);  RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);  RNG4.SeedFixedIntegers(seeds4);
+
+  LatticeGaugeField Umu(UGrid); random(RNG4,Umu);
+  std::vector<LatticeColourMatrix> U(4,UGrid);
+
+  RealD mass=0.1;
+  RealD M5  =1.8;
+  std::cout <<"DomainWallFermion test"<<std::endl;
+  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  TestCGinversions<DomainWallFermion>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  RealD b=1.5;// Scale factor b+c=2, b-c=1
+  RealD c=0.5;
+  std::cout <<"MobiusFermion test"<<std::endl;
+  MobiusFermion Dmob(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c);
+  TestCGinversions<MobiusFermion>(Dmob,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  std::cout <<"MobiusZolotarevFermion test"<<std::endl;
+  MobiusZolotarevFermion Dzolo(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c,0.1,2.0);
+  TestCGinversions<MobiusZolotarevFermion>(Dzolo,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  std::cout <<"ScaledShamirFermion test"<<std::endl;
+  ScaledShamirFermion Dsham(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,2.0);
+  TestCGinversions<ScaledShamirFermion>(Dsham,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  std::cout <<"ScaledShamirZolotarevFermion test"<<std::endl;
+  ScaledShamirZolotarevFermion Dshamz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,2.0,0.1,2.0);
+  TestCGinversions<ScaledShamirZolotarevFermion>(Dshamz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  Grid_finalize();
+}
+template<class What> 
+void  TestCGinversions(What & Ddwf, 
+		       GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+		       GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+		       RealD mass, RealD M5,
+		       GridParallelRNG *RNG4,
+		       GridParallelRNG *RNG5)
+{
+  std::cout << "Testing unpreconditioned inverter"<<std::endl;
+  TestCGunprec<What>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,RNG4,RNG5);
+  std::cout << "Testing red black preconditioned inverter"<<std::endl;
+  TestCGprec<What>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,RNG4,RNG5);
+  std::cout << "Testing red black Schur inverter"<<std::endl;
+  TestCGschur<What>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,RNG4,RNG5);
+}
+
+template<class What> 
+void  TestCGunprec(What & Ddwf, 
+		   GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+		   GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+		   RealD mass, RealD M5,
+		   GridParallelRNG *RNG4,
+		   GridParallelRNG *RNG5)
+{
+  LatticeFermion src   (FGrid); random(*RNG5,src);
+  LatticeFermion result(FGrid); result=zero;
+
+  HermitianOperator<What,LatticeFermion> HermOp(Ddwf);
+  ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
+  CG(HermOp,src,result);
+
+}
+template<class What> 
+void  TestCGprec(What & Ddwf, 
+		 GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+		 GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+		 RealD mass, RealD M5,
+		 GridParallelRNG *RNG4,
+		 GridParallelRNG *RNG5)
+{
+  LatticeFermion src   (FGrid); random(*RNG5,src);
+  LatticeFermion    src_o(FrbGrid);
+  LatticeFermion result_o(FrbGrid);
+  pickCheckerboard(Odd,src_o,src);
+  result_o=zero;
+
+  HermitianCheckerBoardedOperator<What,LatticeFermion> HermOpEO(Ddwf);
+  ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
+  CG(HermOpEO,src_o,result_o);
+}
+
+
+template<class What> 
+void  TestCGschur(What & Ddwf, 
+		   GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+		   GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+		   RealD mass, RealD M5,
+		   GridParallelRNG *RNG4,
+		   GridParallelRNG *RNG5)
+{
+  LatticeFermion src   (FGrid); random(*RNG5,src);
+  LatticeFermion result(FGrid); result=zero;
+
+  ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
+  SchurRedBlackSolve<LatticeFermion> SchurSolver(CG);
+  SchurSolver(Ddwf,src,result);
+}
diff --git a/tests/Grid_any_evenodd.cc b/tests/Test_many_evenodd.cc
similarity index 100%
rename from tests/Grid_any_evenodd.cc
rename to tests/Test_many_evenodd.cc
diff --git a/tests/Grid_nersc_io.cc b/tests/Test_nersc_io.cc
similarity index 100%
rename from tests/Grid_nersc_io.cc
rename to tests/Test_nersc_io.cc
diff --git a/tests/Grid_remez.cc b/tests/Test_remez.cc
similarity index 100%
rename from tests/Grid_remez.cc
rename to tests/Test_remez.cc
diff --git a/tests/Grid_rng.cc b/tests/Test_rng.cc
similarity index 100%
rename from tests/Grid_rng.cc
rename to tests/Test_rng.cc
diff --git a/tests/Grid_rng_fixed.cc b/tests/Test_rng_fixed.cc
similarity index 100%
rename from tests/Grid_rng_fixed.cc
rename to tests/Test_rng_fixed.cc
diff --git a/tests/Grid_simd.cc b/tests/Test_simd.cc
similarity index 100%
rename from tests/Grid_simd.cc
rename to tests/Test_simd.cc
diff --git a/tests/Grid_simd_new.cc b/tests/Test_simd_new.cc
similarity index 100%
rename from tests/Grid_simd_new.cc
rename to tests/Test_simd_new.cc
diff --git a/tests/Grid_stencil.cc b/tests/Test_stencil.cc
similarity index 100%
rename from tests/Grid_stencil.cc
rename to tests/Test_stencil.cc
diff --git a/tests/Grid_wilson_cg_prec.cc b/tests/Test_wilson_cg_prec.cc
similarity index 100%
rename from tests/Grid_wilson_cg_prec.cc
rename to tests/Test_wilson_cg_prec.cc
diff --git a/tests/Grid_wilson_cg_schur.cc b/tests/Test_wilson_cg_schur.cc
similarity index 100%
rename from tests/Grid_wilson_cg_schur.cc
rename to tests/Test_wilson_cg_schur.cc
diff --git a/tests/Grid_wilson_cg_unprec.cc b/tests/Test_wilson_cg_unprec.cc
similarity index 100%
rename from tests/Grid_wilson_cg_unprec.cc
rename to tests/Test_wilson_cg_unprec.cc
diff --git a/tests/Grid_wilson_evenodd.cc b/tests/Test_wilson_evenodd.cc
similarity index 100%
rename from tests/Grid_wilson_evenodd.cc
rename to tests/Test_wilson_evenodd.cc

From 8fe3d4f97126a300f76da35f3cbe9d943f6541c7 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 3 Jun 2015 11:26:54 +0100
Subject: [PATCH 261/429] Overlap Wilson Cayley tanh & zolo

---
 lib/qcd/action/Actions.h                      | 27 ++++++++++---
 .../action/fermion/MobiusZolotarevFermion.h   |  1 +
 .../fermion/OverlapWilsonCayleyTanhFermion.h  | 34 ++++++++++++++++
 .../OverlapWilsonCayleyZolotarevFermion.h     | 37 ++++++++++++++++++
 .../action/fermion/PartialFractionFermion5D.h |  4 +-
 .../fermion/ScaledShamirZolotarevFermion.h    | 39 -------------------
 .../action/fermion/ShamirZolotarevFermion.h   | 39 +++++++++++++++++++
 tests/Test_many_cg.cc                         | 14 +++++--
 tests/Test_many_evenodd.cc                    | 17 +++++++-
 9 files changed, 161 insertions(+), 51 deletions(-)
 create mode 100644 lib/qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h
 create mode 100644 lib/qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h
 delete mode 100644 lib/qcd/action/fermion/ScaledShamirZolotarevFermion.h
 create mode 100644 lib/qcd/action/fermion/ShamirZolotarevFermion.h

diff --git a/lib/qcd/action/Actions.h b/lib/qcd/action/Actions.h
index 893564e0..8a8c4642 100644
--- a/lib/qcd/action/Actions.h
+++ b/lib/qcd/action/Actions.h
@@ -30,19 +30,36 @@
 //#include <qcd/action/fermion/CloverFermion.h>
 
 ////////////////////////////////////////////
-// 5D formulations
+// 5D formulations...
 ////////////////////////////////////////////
+
 #include <qcd/action/fermion/WilsonFermion5D.h> // used by all 5d overlap types
+
+//////////
+// Cayley
+//////////
 #include <qcd/action/fermion/CayleyFermion5D.h>
-#include <qcd/action/fermion/ContinuedFractionFermion5D.h>
-//#include <qcd/action/fermion/PartialFraction.h>
 
 #include <qcd/action/fermion/DomainWallFermion.h>
 #include <qcd/action/fermion/DomainWallFermion.h>
+
 #include <qcd/action/fermion/MobiusFermion.h>
-#include <qcd/action/fermion/MobiusZolotarevFermion.h>
 #include <qcd/action/fermion/ScaledShamirFermion.h>
-#include <qcd/action/fermion/ScaledShamirZolotarevFermion.h>
+#include <qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h>
+
+#include <qcd/action/fermion/MobiusZolotarevFermion.h>
+#include <qcd/action/fermion/ShamirZolotarevFermion.h>
+#include <qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h>
+
+//////////////////////
+// Continued fraction
+//////////////////////
+#include <qcd/action/fermion/ContinuedFractionFermion5D.h>
+
+//////////////////////
+// Partial fraction
+//////////////////////
+#include <qcd/action/fermion/PartialFractionFermion5D.h>
 
 
     // Chroma interface defining FermionAction
diff --git a/lib/qcd/action/fermion/MobiusZolotarevFermion.h b/lib/qcd/action/fermion/MobiusZolotarevFermion.h
index 866d0c39..9ac795d9 100644
--- a/lib/qcd/action/fermion/MobiusZolotarevFermion.h
+++ b/lib/qcd/action/fermion/MobiusZolotarevFermion.h
@@ -34,6 +34,7 @@ namespace Grid {
 	assert(zdata->n==this->Ls);
 
 	std::cout << "MobiusZolotarevFermion (b="<<b<<",c="<<c<<") with Ls= "<<Ls<<" Zolotarev range ["<<lo<<","<<hi<<"]"<<std::endl;
+	std::cout << "MobiusZolotarevFermion : note there is a degeneracy between (b+c) and Zolo param hi"<<std::endl;
 	
 	// Call base setter
 	this->CayleyFermion5D::SetCoefficients(1.0,zdata,b,c);
diff --git a/lib/qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h b/lib/qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h
new file mode 100644
index 00000000..e764c8ae
--- /dev/null
+++ b/lib/qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h
@@ -0,0 +1,34 @@
+#ifndef OVERLAP_WILSON_CAYLEY_TANH_FERMION_H
+#define OVERLAP_WILSON_CAYLEY_TANH_FERMION_H
+
+#include <Grid.h>
+
+namespace Grid {
+
+  namespace QCD {
+
+    class OverlapWilsonCayleyTanhFermion : public MobiusFermion
+    {
+    public:
+
+      // Constructors
+    OverlapWilsonCayleyTanhFermion(LatticeGaugeField &_Umu,
+				   GridCartesian         &FiveDimGrid,
+				   GridRedBlackCartesian &FiveDimRedBlackGrid,
+				   GridCartesian         &FourDimGrid,
+				   GridRedBlackCartesian &FourDimRedBlackGrid,
+				   RealD _mass,RealD _M5,
+				   RealD scale) :
+      
+      // b+c=scale, b-c = 0 <=> b =c = scale/2
+      MobiusFermion(_Umu,
+		    FiveDimGrid,
+		    FiveDimRedBlackGrid,
+		    FourDimGrid,
+		    FourDimRedBlackGrid,_mass,_M5,0.5*scale,0.5*scale)
+	{
+	}
+    };
+  }
+}
+#endif
diff --git a/lib/qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h b/lib/qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h
new file mode 100644
index 00000000..82c43fb7
--- /dev/null
+++ b/lib/qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h
@@ -0,0 +1,37 @@
+#ifndef  OVERLAP_WILSON_CAYLEY_ZOLOTAREV_FERMION_H
+#define  OVERLAP_WILSON_CAYLEY_ZOLOTAREV_FERMION_H
+
+#include <Grid.h>
+
+namespace Grid {
+
+  namespace QCD {
+
+    class OverlapWilsonCayleyZolotarevFermion : public MobiusZolotarevFermion
+    {
+    public:
+
+      // Constructors
+
+    OverlapWilsonCayleyZolotarevFermion(LatticeGaugeField &_Umu,
+					GridCartesian         &FiveDimGrid,
+					GridRedBlackCartesian &FiveDimRedBlackGrid,
+					GridCartesian         &FourDimGrid,
+					GridRedBlackCartesian &FourDimRedBlackGrid,
+					RealD _mass,RealD _M5,
+					RealD lo, RealD hi) : 
+      // b+c=1.0, b-c = 0 <=> b =c = 1/2
+      MobiusZolotarevFermion(_Umu,
+			     FiveDimGrid,
+			     FiveDimRedBlackGrid,
+			     FourDimGrid,
+			     FourDimRedBlackGrid,_mass,_M5,0.5,0.5,lo,hi)
+
+      {}
+
+    };
+
+  }
+}
+
+#endif
diff --git a/lib/qcd/action/fermion/PartialFractionFermion5D.h b/lib/qcd/action/fermion/PartialFractionFermion5D.h
index c51aa563..95f8c0f9 100644
--- a/lib/qcd/action/fermion/PartialFractionFermion5D.h
+++ b/lib/qcd/action/fermion/PartialFractionFermion5D.h
@@ -25,10 +25,10 @@ namespace Grid {
 
       virtual void PartialFractionCoefficients(void);
 
-      zolotarev_data *zdata;
+      Approx::zolotarev_data *zdata;
 
       // Part frac
-      double R=(1+this->mass)/(1-this->mass);
+      double R;
       std::vector<double> p; 
       std::vector<double> q;
 
diff --git a/lib/qcd/action/fermion/ScaledShamirZolotarevFermion.h b/lib/qcd/action/fermion/ScaledShamirZolotarevFermion.h
deleted file mode 100644
index a4c88d5d..00000000
--- a/lib/qcd/action/fermion/ScaledShamirZolotarevFermion.h
+++ /dev/null
@@ -1,39 +0,0 @@
-#ifndef  GRID_QCD_SCALED_SHAMIR_ZOLOTAREV_FERMION_H
-#define  GRID_QCD_SCALED_SHAMIR_ZOLOTAREV_FERMION_H
-
-#include <Grid.h>
-
-namespace Grid {
-
-  namespace QCD {
-
-    class ScaledShamirZolotarevFermion : public MobiusZolotarevFermion
-    {
-    public:
-
-      // Constructors
-
-
-    ScaledShamirZolotarevFermion(LatticeGaugeField &_Umu,
-				 GridCartesian         &FiveDimGrid,
-				 GridRedBlackCartesian &FiveDimRedBlackGrid,
-				 GridCartesian         &FourDimGrid,
-				 GridRedBlackCartesian &FourDimRedBlackGrid,
-				 RealD _mass,RealD _M5,
-				 RealD scale,
-				 RealD lo, RealD hi) : 
-      
-      MobiusZolotarevFermion(_Umu,
-			       FiveDimGrid,
-			       FiveDimRedBlackGrid,
-			       FourDimGrid,
-			       FourDimRedBlackGrid,_mass,_M5,0.5*(scale+1.0),0.5*(scale-1.0),lo,hi)
-
-      {}
-
-    };
-
-  }
-}
-
-#endif
diff --git a/lib/qcd/action/fermion/ShamirZolotarevFermion.h b/lib/qcd/action/fermion/ShamirZolotarevFermion.h
new file mode 100644
index 00000000..6a7df439
--- /dev/null
+++ b/lib/qcd/action/fermion/ShamirZolotarevFermion.h
@@ -0,0 +1,39 @@
+#ifndef  GRID_QCD_SHAMIR_ZOLOTAREV_FERMION_H
+#define  GRID_QCD_SHAMIR_ZOLOTAREV_FERMION_H
+
+#include <Grid.h>
+
+namespace Grid {
+
+  namespace QCD {
+
+    class ShamirZolotarevFermion : public MobiusZolotarevFermion
+    {
+    public:
+
+      // Constructors
+
+
+    ShamirZolotarevFermion(LatticeGaugeField &_Umu,
+			   GridCartesian         &FiveDimGrid,
+			   GridRedBlackCartesian &FiveDimRedBlackGrid,
+			   GridCartesian         &FourDimGrid,
+			   GridRedBlackCartesian &FourDimRedBlackGrid,
+			   RealD _mass,RealD _M5,
+			   RealD lo, RealD hi) : 
+      
+      // b+c = 1; b-c = 1 => b=1, c=0
+      MobiusZolotarevFermion(_Umu,
+			     FiveDimGrid,
+			     FiveDimRedBlackGrid,
+			     FourDimGrid,
+			     FourDimRedBlackGrid,_mass,_M5,1.0,0.0,lo,hi)
+      
+      {}
+
+    };
+
+  }
+}
+
+#endif
diff --git a/tests/Test_many_cg.cc b/tests/Test_many_cg.cc
index 3241a025..4510d4b5 100644
--- a/tests/Test_many_cg.cc
+++ b/tests/Test_many_cg.cc
@@ -89,9 +89,17 @@ int main (int argc, char ** argv)
   ScaledShamirFermion Dsham(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,2.0);
   TestCGinversions<ScaledShamirFermion>(Dsham,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
-  std::cout <<"ScaledShamirZolotarevFermion test"<<std::endl;
-  ScaledShamirZolotarevFermion Dshamz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,2.0,0.1,2.0);
-  TestCGinversions<ScaledShamirZolotarevFermion>(Dshamz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  std::cout <<"ShamirZolotarevFermion test"<<std::endl;
+  ShamirZolotarevFermion Dshamz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,2.0);
+  TestCGinversions<ShamirZolotarevFermion>(Dshamz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  std::cout <<"OverlapWilsonCayleyTanhFermion test"<<std::endl;
+  OverlapWilsonCayleyTanhFermion Dov(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+  TestCGinversions<OverlapWilsonCayleyTanhFermion>(Dov,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  std::cout <<"OverlapWilsonCayleyZolotarevFermion test"<<std::endl;
+  OverlapWilsonCayleyZolotarevFermion Dovz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,2.0);
+  TestCGinversions<OverlapWilsonCayleyZolotarevFermion>(Dovz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   Grid_finalize();
 }
diff --git a/tests/Test_many_evenodd.cc b/tests/Test_many_evenodd.cc
index de1e3c78..df28981b 100644
--- a/tests/Test_many_evenodd.cc
+++ b/tests/Test_many_evenodd.cc
@@ -48,23 +48,36 @@ int main (int argc, char ** argv)
 
   RealD mass=0.1;
   RealD M5  =1.8;
+  std::cout <<"DomainWallFermion test"<<std::endl;
   DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
   TestWhat<DomainWallFermion>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   RealD b=1.5;// Scale factor b+c=2, b-c=1
   RealD c=0.5;
+  std::cout <<"MobiusFermion test"<<std::endl;
   MobiusFermion Dmob(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c);
   TestWhat<MobiusFermion>(Dmob,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
+  std::cout <<"MobiusZolotarevFermion test"<<std::endl;
   MobiusZolotarevFermion Dzolo(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c,0.1,2.0);
   TestWhat<MobiusZolotarevFermion>(Dzolo,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
+  std::cout <<"ScaledShamirFermion test"<<std::endl;
   ScaledShamirFermion Dsham(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,2.0);
   TestWhat<ScaledShamirFermion>(Dsham,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
-  ScaledShamirZolotarevFermion Dshamz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,2.0,0.1,2.0);
-  TestWhat<ScaledShamirZolotarevFermion>(Dshamz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
+  std::cout <<"ShamirZolotarevFermion test"<<std::endl;
+  ShamirZolotarevFermion Dshamz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,2.0);
+  TestWhat<ShamirZolotarevFermion>(Dshamz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  std::cout <<"OverlapWilsonCayleyTanhFermion test"<<std::endl;
+  OverlapWilsonCayleyTanhFermion Dov(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+  TestWhat<OverlapWilsonCayleyTanhFermion>(Dov,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  std::cout <<"OverlapWilsonCayleyZolotarevFermion test"<<std::endl;
+  OverlapWilsonCayleyZolotarevFermion Dovz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,2.0);
+  TestWhat<OverlapWilsonCayleyZolotarevFermion>(Dovz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   Grid_finalize();
 }

From 4bcc319e11c2ac4babbbed22fb65a51f1c3787b1 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 3 Jun 2015 12:47:05 +0100
Subject: [PATCH 262/429] Reorganise of file naming

---
 Makefile.am                                   |   2 +
 configure                                     |   4 +-
 configure.ac                                  |   2 +-
 lib/{Grid_algorithms.h => Algorithms.h}       |   0
 ...aligned_allocator.h => AlignedAllocator.h} |   0
 lib/Cartesian.h                               |   8 ++
 lib/{Grid_communicator.h => Communicator.h}   |   2 +-
 lib/{Grid_comparison.h => Comparison.h}       |   4 +-
 lib/{Grid_cshift.h => Cshift.h}               |   6 +-
 lib/Grid.h                                    |  26 ++--
 lib/{Grid_config.h => GridConfig.h}           |   0
 lib/{Grid_config.h.in => GridConfig.h.in}     |   2 +-
 lib/{Grid_init.cc => GridInit.cc}             |   0
 lib/Grid_cartesian.h                          |   8 --
 lib/Grid_math.h                               |  16 ---
 lib/{Grid_lattice.h => Lattice.h}             |   2 +-
 lib/Makefile.am                               | 113 ++----------------
 lib/{Grid_simd.h => Simd.h}                   |   0
 lib/{Grid_stencil.h => Stencil.h}             |   2 +-
 lib/Tensors.h                                 |  16 +++
 lib/{Grid_threads.h => Threads.h}             |   0
 ...Grid_cartesian_base.h => Cartesian_base.h} |   1 -
 ...Grid_cartesian_full.h => Cartesian_full.h} |   0
 ...sian_red_black.h => Cartesian_red_black.h} |   0
 ...ommunicator_base.h => Communicator_base.h} |   0
 ...ommunicator_mpi.cc => Communicator_mpi.cc} |   0
 ...municator_none.cc => Communicator_none.cc} |   0
 .../{Grid_cshift_common.h => Cshift_common.h} |   0
 .../{Grid_cshift_mpi.h => Cshift_mpi.h}       |   0
 .../{Grid_cshift_none.h => Cshift_none.h}     |   0
 .../{Grid_lattice_ET.h => Lattice_ET.h}       |   0
 .../{Grid_lattice_arith.h => Lattice_arith.h} |   0
 .../{Grid_lattice_base.h => Lattice_base.h}   |  27 ++---
 ...tice_comparison.h => Lattice_comparison.h} |   0
 ...ce_conformable.h => Lattice_conformable.h} |   0
 ...tice_coordinate.h => Lattice_coordinate.h} |   0
 .../{Grid_lattice_local.h => Lattice_local.h} |   0
 ..._lattice_overload.h => Lattice_overload.h} |   0
 ..._lattice_peekpoke.h => Lattice_peekpoke.h} |   0
 ...id_lattice_reality.h => Lattice_reality.h} |   0
 ...attice_reduction.h => Lattice_reduction.h} |   0
 .../{Grid_lattice_rng.h => Lattice_rng.h}     |   0
 .../{Grid_lattice_trace.h => Lattice_trace.h} |   0
 ..._lattice_transfer.h => Lattice_transfer.h} |   0
 ...attice_transpose.h => Lattice_transpose.h} |   0
 .../{Grid_lattice_where.h => Lattice_where.h} |   0
 lib/math/Grid_math_arith.h                    |  11 --
 lib/parallelIO/{GridNerscIO.h => NerscIO.h}   |   0
 .../fermion/PartialFractionFermion5D.cc       |  46 -------
 lib/stencil/{Grid_lebesgue.cc => Lebesgue.cc} |   0
 lib/stencil/{Grid_lebesgue.h => Lebesgue.h}   |   0
 ...id_stencil_common.cc => Stencil_common.cc} |   0
 lib/tensors/Tensor_arith.h                    |  11 ++
 .../Tensor_arith_add.h}                       |   0
 .../Tensor_arith_mac.h}                       |   0
 .../Tensor_arith_mul.h}                       |   0
 .../Tensor_arith_scalar.h}                    |   0
 .../Tensor_arith_sub.h}                       |   0
 .../Tensor_class.h}                           |   0
 .../Tensor_extract_merge.h}                   |   0
 .../Tensor_inner.h}                           |   0
 .../Tensor_outer.h}                           |   0
 .../Tensor_peek.h}                            |   0
 .../Tensor_poke.h}                            |   0
 .../Tensor_reality.h}                         |   0
 .../Tensor_trace.h}                           |   0
 .../Tensor_traits.h}                          |   0
 .../Tensor_transpose.h}                       |   0
 tests/Test_cshift.cc                          |   1 -
 tests/Test_cshift_red_black.cc                |   1 -
 tests/Test_gamma.cc                           |   1 -
 tests/Test_nersc_io.cc                        |   1 -
 tests/Test_rng.cc                             |   1 -
 tests/Test_rng_fixed.cc                       |   1 -
 tests/Test_simd.cc                            |   1 -
 75 files changed, 85 insertions(+), 231 deletions(-)
 rename lib/{Grid_algorithms.h => Algorithms.h} (100%)
 rename lib/{Grid_aligned_allocator.h => AlignedAllocator.h} (100%)
 create mode 100644 lib/Cartesian.h
 rename lib/{Grid_communicator.h => Communicator.h} (57%)
 rename lib/{Grid_comparison.h => Comparison.h} (98%)
 rename lib/{Grid_cshift.h => Cshift.h} (51%)
 rename lib/{Grid_config.h => GridConfig.h} (100%)
 rename lib/{Grid_config.h.in => GridConfig.h.in} (97%)
 rename lib/{Grid_init.cc => GridInit.cc} (100%)
 delete mode 100644 lib/Grid_cartesian.h
 delete mode 100644 lib/Grid_math.h
 rename lib/{Grid_lattice.h => Lattice.h} (58%)
 rename lib/{Grid_simd.h => Simd.h} (100%)
 rename lib/{Grid_stencil.h => Stencil.h} (99%)
 create mode 100644 lib/Tensors.h
 rename lib/{Grid_threads.h => Threads.h} (100%)
 rename lib/cartesian/{Grid_cartesian_base.h => Cartesian_base.h} (99%)
 rename lib/cartesian/{Grid_cartesian_full.h => Cartesian_full.h} (100%)
 rename lib/cartesian/{Grid_cartesian_red_black.h => Cartesian_red_black.h} (100%)
 rename lib/communicator/{Grid_communicator_base.h => Communicator_base.h} (100%)
 rename lib/communicator/{Grid_communicator_mpi.cc => Communicator_mpi.cc} (100%)
 rename lib/communicator/{Grid_communicator_none.cc => Communicator_none.cc} (100%)
 rename lib/cshift/{Grid_cshift_common.h => Cshift_common.h} (100%)
 rename lib/cshift/{Grid_cshift_mpi.h => Cshift_mpi.h} (100%)
 rename lib/cshift/{Grid_cshift_none.h => Cshift_none.h} (100%)
 rename lib/lattice/{Grid_lattice_ET.h => Lattice_ET.h} (100%)
 rename lib/lattice/{Grid_lattice_arith.h => Lattice_arith.h} (100%)
 rename lib/lattice/{Grid_lattice_base.h => Lattice_base.h} (93%)
 rename lib/lattice/{Grid_lattice_comparison.h => Lattice_comparison.h} (100%)
 rename lib/lattice/{Grid_lattice_conformable.h => Lattice_conformable.h} (100%)
 rename lib/lattice/{Grid_lattice_coordinate.h => Lattice_coordinate.h} (100%)
 rename lib/lattice/{Grid_lattice_local.h => Lattice_local.h} (100%)
 rename lib/lattice/{Grid_lattice_overload.h => Lattice_overload.h} (100%)
 rename lib/lattice/{Grid_lattice_peekpoke.h => Lattice_peekpoke.h} (100%)
 rename lib/lattice/{Grid_lattice_reality.h => Lattice_reality.h} (100%)
 rename lib/lattice/{Grid_lattice_reduction.h => Lattice_reduction.h} (100%)
 rename lib/lattice/{Grid_lattice_rng.h => Lattice_rng.h} (100%)
 rename lib/lattice/{Grid_lattice_trace.h => Lattice_trace.h} (100%)
 rename lib/lattice/{Grid_lattice_transfer.h => Lattice_transfer.h} (100%)
 rename lib/lattice/{Grid_lattice_transpose.h => Lattice_transpose.h} (100%)
 rename lib/lattice/{Grid_lattice_where.h => Lattice_where.h} (100%)
 delete mode 100644 lib/math/Grid_math_arith.h
 rename lib/parallelIO/{GridNerscIO.h => NerscIO.h} (100%)
 rename lib/stencil/{Grid_lebesgue.cc => Lebesgue.cc} (100%)
 rename lib/stencil/{Grid_lebesgue.h => Lebesgue.h} (100%)
 rename lib/stencil/{Grid_stencil_common.cc => Stencil_common.cc} (100%)
 create mode 100644 lib/tensors/Tensor_arith.h
 rename lib/{math/Grid_math_arith_add.h => tensors/Tensor_arith_add.h} (100%)
 rename lib/{math/Grid_math_arith_mac.h => tensors/Tensor_arith_mac.h} (100%)
 rename lib/{math/Grid_math_arith_mul.h => tensors/Tensor_arith_mul.h} (100%)
 rename lib/{math/Grid_math_arith_scalar.h => tensors/Tensor_arith_scalar.h} (100%)
 rename lib/{math/Grid_math_arith_sub.h => tensors/Tensor_arith_sub.h} (100%)
 rename lib/{math/Grid_math_tensors.h => tensors/Tensor_class.h} (100%)
 rename lib/{Grid_extract.h => tensors/Tensor_extract_merge.h} (100%)
 rename lib/{math/Grid_math_inner.h => tensors/Tensor_inner.h} (100%)
 rename lib/{math/Grid_math_outer.h => tensors/Tensor_outer.h} (100%)
 rename lib/{math/Grid_math_peek.h => tensors/Tensor_peek.h} (100%)
 rename lib/{math/Grid_math_poke.h => tensors/Tensor_poke.h} (100%)
 rename lib/{math/Grid_math_reality.h => tensors/Tensor_reality.h} (100%)
 rename lib/{math/Grid_math_trace.h => tensors/Tensor_trace.h} (100%)
 rename lib/{math/Grid_math_traits.h => tensors/Tensor_traits.h} (100%)
 rename lib/{math/Grid_math_transpose.h => tensors/Tensor_transpose.h} (100%)

diff --git a/Makefile.am b/Makefile.am
index fc3f6a0a..3b1d5690 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -1,3 +1,5 @@
 # additional include paths necessary to compile the C++ library
 AM_CXXFLAGS = -I$(top_srcdir)/
 SUBDIRS = lib tests benchmarks
+
+filelist: $(SUBDIRS)
\ No newline at end of file
diff --git a/configure b/configure
index 615ba987..20328052 100755
--- a/configure
+++ b/configure
@@ -3064,7 +3064,7 @@ fi
 
 
 
-ac_config_headers="$ac_config_headers lib/Grid_config.h"
+ac_config_headers="$ac_config_headers lib/GridConfig.h"
 
 # Check whether --enable-silent-rules was given.
 if test "${enable_silent_rules+set}" = set; then :
@@ -5814,7 +5814,7 @@ cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 for ac_config_target in $ac_config_targets
 do
   case $ac_config_target in
-    "lib/Grid_config.h") CONFIG_HEADERS="$CONFIG_HEADERS lib/Grid_config.h" ;;
+    "lib/GridConfig.h") CONFIG_HEADERS="$CONFIG_HEADERS lib/GridConfig.h" ;;
     "depfiles") CONFIG_COMMANDS="$CONFIG_COMMANDS depfiles" ;;
     "docs/doxy.cfg") CONFIG_FILES="$CONFIG_FILES docs/doxy.cfg" ;;
     "Makefile") CONFIG_FILES="$CONFIG_FILES Makefile" ;;
diff --git a/configure.ac b/configure.ac
index 5dcbea36..03fd7900 100644
--- a/configure.ac
+++ b/configure.ac
@@ -11,7 +11,7 @@ AC_CANONICAL_SYSTEM
 AM_INIT_AUTOMAKE(subdir-objects)
 AC_CONFIG_MACRO_DIR([m4])
 AC_CONFIG_SRCDIR([lib/Grid.h])
-AC_CONFIG_HEADERS([lib/Grid_config.h])
+AC_CONFIG_HEADERS([lib/GridConfig.h])
 m4_ifdef([AM_SILENT_RULES], [AM_SILENT_RULES([yes])])
 
 AC_MSG_NOTICE([
diff --git a/lib/Grid_algorithms.h b/lib/Algorithms.h
similarity index 100%
rename from lib/Grid_algorithms.h
rename to lib/Algorithms.h
diff --git a/lib/Grid_aligned_allocator.h b/lib/AlignedAllocator.h
similarity index 100%
rename from lib/Grid_aligned_allocator.h
rename to lib/AlignedAllocator.h
diff --git a/lib/Cartesian.h b/lib/Cartesian.h
new file mode 100644
index 00000000..db384b53
--- /dev/null
+++ b/lib/Cartesian.h
@@ -0,0 +1,8 @@
+#ifndef GRID_CARTESIAN_H
+#define GRID_CARTESIAN_H
+
+#include <cartesian/Cartesian_base.h>
+#include <cartesian/Cartesian_full.h>
+#include <cartesian/Cartesian_red_black.h> 
+
+#endif
diff --git a/lib/Grid_communicator.h b/lib/Communicator.h
similarity index 57%
rename from lib/Grid_communicator.h
rename to lib/Communicator.h
index cfa6e0a7..6880adda 100644
--- a/lib/Grid_communicator.h
+++ b/lib/Communicator.h
@@ -1,6 +1,6 @@
 #ifndef GRID_COMMUNICATOR_H
 #define GRID_COMMUNICATOR_H
 
-#include <communicator/Grid_communicator_base.h>
+#include <communicator/Communicator_base.h>
 
 #endif
diff --git a/lib/Grid_comparison.h b/lib/Comparison.h
similarity index 98%
rename from lib/Grid_comparison.h
rename to lib/Comparison.h
index 3f9c206d..ecd6ece0 100644
--- a/lib/Grid_comparison.h
+++ b/lib/Comparison.h
@@ -141,7 +141,7 @@ namespace Grid {
     }
 }
 
-#include <lattice/Grid_lattice_comparison.h>
-#include <lattice/Grid_lattice_where.h>
+#include <lattice/Lattice_comparison.h>
+#include <lattice/Lattice_where.h>
 
 #endif
diff --git a/lib/Grid_cshift.h b/lib/Cshift.h
similarity index 51%
rename from lib/Grid_cshift.h
rename to lib/Cshift.h
index 10c7a3c4..3caccbf9 100644
--- a/lib/Grid_cshift.h
+++ b/lib/Cshift.h
@@ -1,13 +1,13 @@
 #ifndef _GRID_CSHIFT_H_
 #define _GRID_CSHIFT_H_
 
-#include <cshift/Grid_cshift_common.h>
+#include <cshift/Cshift_common.h>
 
 #ifdef GRID_COMMS_NONE
-#include <cshift/Grid_cshift_none.h>
+#include <cshift/Cshift_none.h>
 #endif
 
 #ifdef GRID_COMMS_MPI
-#include <cshift/Grid_cshift_mpi.h>
+#include <cshift/Cshift_mpi.h>
 #endif 
 #endif
diff --git a/lib/Grid.h b/lib/Grid.h
index 7fa56892..16530434 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -33,7 +33,7 @@
 
 #define strong_inline __attribute__((always_inline)) inline
 
-#include <Grid_config.h>
+#include <GridConfig.h>
 
 ////////////////////////////////////////////////////////////
 // Tunable header includes
@@ -46,22 +46,22 @@
 #include <malloc.h>
 #endif
 
-#include <Grid_aligned_allocator.h>
+#include <AlignedAllocator.h>
 
-#include <Grid_simd.h>
-#include <Grid_threads.h>
+#include <Simd.h>
+#include <Threads.h>
 
-#include <Grid_cartesian.h> // subdir aggregate
-#include <Grid_math.h>      // subdir aggregate
-#include <Grid_lattice.h>   // subdir aggregate
-#include <Grid_comparison.h>
-#include <Grid_cshift.h>    // subdir aggregate
-#include <Grid_stencil.h>   // subdir aggregate
-
-#include <Grid_algorithms.h>// subdir aggregate
+#include <Communicator.h> // subdir aggregate
+#include <Cartesian.h> // subdir aggregate
+#include <Tensors.h>   // subdir aggregate
+#include <Lattice.h>   // subdir aggregate
+#include <Comparison.h>
+#include <Cshift.h>    // subdir aggregate
+#include <Stencil.h>   // subdir aggregate
+#include <Algorithms.h>// subdir aggregate
 
 #include <qcd/QCD.h>
-#include <parallelIO/GridNerscIO.h>
+#include <parallelIO/NerscIO.h>
 
 namespace Grid {
 
diff --git a/lib/Grid_config.h b/lib/GridConfig.h
similarity index 100%
rename from lib/Grid_config.h
rename to lib/GridConfig.h
diff --git a/lib/Grid_config.h.in b/lib/GridConfig.h.in
similarity index 97%
rename from lib/Grid_config.h.in
rename to lib/GridConfig.h.in
index 6f05d6cb..03f91ff4 100644
--- a/lib/Grid_config.h.in
+++ b/lib/GridConfig.h.in
@@ -1,4 +1,4 @@
-/* lib/Grid_config.h.in.  Generated from configure.ac by autoheader.  */
+/* lib/GridConfig.h.in.  Generated from configure.ac by autoheader.  */
 
 /* AVX */
 #undef AVX1
diff --git a/lib/Grid_init.cc b/lib/GridInit.cc
similarity index 100%
rename from lib/Grid_init.cc
rename to lib/GridInit.cc
diff --git a/lib/Grid_cartesian.h b/lib/Grid_cartesian.h
deleted file mode 100644
index c01be20a..00000000
--- a/lib/Grid_cartesian.h
+++ /dev/null
@@ -1,8 +0,0 @@
-#ifndef GRID_CARTESIAN_H
-#define GRID_CARTESIAN_H
-
-#include <cartesian/Grid_cartesian_base.h>
-#include <cartesian/Grid_cartesian_full.h>
-#include <cartesian/Grid_cartesian_red_black.h> 
-
-#endif
diff --git a/lib/Grid_math.h b/lib/Grid_math.h
deleted file mode 100644
index 17bc09a5..00000000
--- a/lib/Grid_math.h
+++ /dev/null
@@ -1,16 +0,0 @@
-#ifndef GRID_MATH_H
-#define GRID_MATH_H
-
-#include <math/Grid_math_traits.h>
-#include <math/Grid_math_tensors.h>
-#include <math/Grid_math_arith.h>
-#include <math/Grid_math_inner.h>
-#include <math/Grid_math_outer.h>
-#include <math/Grid_math_transpose.h>
-#include <math/Grid_math_trace.h>
-#include <math/Grid_math_peek.h>
-#include <math/Grid_math_poke.h>
-#include <math/Grid_math_reality.h>
-
-    
-#endif
diff --git a/lib/Grid_lattice.h b/lib/Lattice.h
similarity index 58%
rename from lib/Grid_lattice.h
rename to lib/Lattice.h
index 35664aee..1f29a908 100644
--- a/lib/Grid_lattice.h
+++ b/lib/Lattice.h
@@ -1,6 +1,6 @@
 #ifndef GRID_LATTICE_H
 #define GRID_LATTICE_H
 
-#include <lattice/Grid_lattice_base.h>
+#include <lattice/Lattice_base.h>
 
 #endif
diff --git a/lib/Makefile.am b/lib/Makefile.am
index b688d8b2..7ee52e0e 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -3,119 +3,26 @@ AM_CXXFLAGS = -I$(top_srcdir)/
 
 extra_sources=
 if BUILD_COMMS_MPI
-  extra_sources+=communicator/Grid_communicator_mpi.cc
+  extra_sources+=communicator/Communicator_mpi.cc
 endif
 
 if BUILD_COMMS_NONE
-  extra_sources+=communicator/Grid_communicator_none.cc
+  extra_sources+=communicator/Communicator_none.cc
 endif
 
 #
 # Libraries
 #
-lib_LIBRARIES = libGrid.a
-libGrid_a_SOURCES =				\
-	Grid_init.cc				\
-	stencil/Grid_lebesgue.cc		\
-	stencil/Grid_stencil_common.cc		\
-	algorithms/approx/Zolotarev.cc		\
-	algorithms/approx/Remez.cc		\
-	qcd/SpaceTimeGrid.cc\
-	qcd/Dirac.cc\
-	qcd/action/fermion/WilsonKernels.cc\
-	qcd/action/fermion/WilsonKernelsHand.cc\
-	qcd/action/fermion/WilsonFermion.cc\
-	qcd/action/fermion/WilsonFermion5D.cc\
-	qcd/action/fermion/CayleyFermion5D.cc \
-	qcd/action/fermion/ContinuedFractionFermion5D.cc	\
-	$(extra_sources)
 
-#	qcd/action/fermion/PartialFractionFermion5D.cc	\
+include Make.inc
+
+lib_LIBRARIES = libGrid.a
+libGrid_a_SOURCES = $(CCFILES) $(extra_sources)
+
+
+#	qcd/action/fermion/PartialFractionFermion5D.cc\	\
 #
 # Include files
 #
-nobase_include_HEADERS=\
-		./algorithms/approx/bigfloat.h\
-		./algorithms/approx/bigfloat_double.h\
-		./algorithms/approx/Chebyshev.h\
-		./algorithms/approx/Remez.h\
-		./algorithms/approx/Zolotarev.h\
-		./algorithms/iterative/ConjugateGradient.h\
-		./algorithms/iterative/NormalEquations.h\
-		./algorithms/iterative/SchurRedBlack.h\
-		./algorithms/LinearOperator.h\
-		./algorithms/SparseMatrix.h\
-		./cartesian/Grid_cartesian_base.h\
-		./cartesian/Grid_cartesian_full.h\
-		./cartesian/Grid_cartesian_red_black.h\
-		./communicator/Grid_communicator_base.h\
-		./cshift/Grid_cshift_common.h\
-		./cshift/Grid_cshift_mpi.h\
-		./cshift/Grid_cshift_none.h\
-		./Grid.h\
-		./Grid_algorithms.h\
-		./Grid_aligned_allocator.h\
-		./Grid_cartesian.h\
-		./Grid_communicator.h\
-		./Grid_comparison.h\
-		./Grid_config.h\
-		./Grid_cshift.h\
-		./Grid_extract.h\
-		./Grid_lattice.h\
-		./Grid_math.h\
-		./Grid_simd.h\
-		./Grid_stencil.h\
-		./Grid_threads.h\
-		./lattice/Grid_lattice_arith.h\
-		./lattice/Grid_lattice_base.h\
-		./lattice/Grid_lattice_comparison.h\
-		./lattice/Grid_lattice_conformable.h\
-		./lattice/Grid_lattice_coordinate.h\
-		./lattice/Grid_lattice_ET.h\
-		./lattice/Grid_lattice_local.h\
-		./lattice/Grid_lattice_overload.h\
-		./lattice/Grid_lattice_peekpoke.h\
-		./lattice/Grid_lattice_reality.h\
-		./lattice/Grid_lattice_reduction.h\
-		./lattice/Grid_lattice_rng.h\
-		./lattice/Grid_lattice_trace.h\
-		./lattice/Grid_lattice_transfer.h\
-		./lattice/Grid_lattice_transpose.h\
-		./lattice/Grid_lattice_where.h\
-		./math/Grid_math_arith.h\
-		./math/Grid_math_arith_add.h\
-		./math/Grid_math_arith_mac.h\
-		./math/Grid_math_arith_mul.h\
-		./math/Grid_math_arith_scalar.h\
-		./math/Grid_math_arith_sub.h\
-		./math/Grid_math_inner.h\
-		./math/Grid_math_outer.h\
-		./math/Grid_math_peek.h\
-		./math/Grid_math_poke.h\
-		./math/Grid_math_reality.h\
-		./math/Grid_math_tensors.h\
-		./math/Grid_math_trace.h\
-		./math/Grid_math_traits.h\
-		./math/Grid_math_transpose.h\
-		./parallelIO/GridNerscIO.h\
-		./qcd/action/Actions.h\
-		./qcd/action/fermion/FermionOperator.h\
-		./qcd/action/fermion/WilsonCompressor.h\
-		./qcd/action/fermion/WilsonKernels.h\
-		./qcd/action/fermion/WilsonFermion.h\
-		./qcd/action/fermion/WilsonFermion5D.h\
-		./qcd/Dirac.h\
-		./qcd/QCD.h\
-		./qcd/TwoSpinor.h\
-		./simd/Grid_avx.h\
-		./simd/Grid_avx512.h\
-		./simd/Grid_qpx.h\
-		./simd/Grid_sse4.h\
-		./simd/Grid_vector_types.h\
-		./simd/Old/Grid_vComplexD.h\
-		./simd/Old/Grid_vComplexF.h\
-		./simd/Old/Grid_vInteger.h\
-		./simd/Old/Grid_vRealD.h\
-		./simd/Old/Grid_vRealF.h\
-		./stencil/Grid_lebesgue.h
+nobase_include_HEADERS=$(HFILES)
 
diff --git a/lib/Grid_simd.h b/lib/Simd.h
similarity index 100%
rename from lib/Grid_simd.h
rename to lib/Simd.h
diff --git a/lib/Grid_stencil.h b/lib/Stencil.h
similarity index 99%
rename from lib/Grid_stencil.h
rename to lib/Stencil.h
index 50d22453..8529e73a 100644
--- a/lib/Grid_stencil.h
+++ b/lib/Stencil.h
@@ -1,7 +1,7 @@
 #ifndef GRID_STENCIL_H
 #define GRID_STENCIL_H
 
-#include <stencil/Grid_lebesgue.h>   // subdir aggregate
+#include <stencil/Lebesgue.h>   // subdir aggregate
 
 //////////////////////////////////////////////////////////////////////////////////////////
 // Must not lose sight that goal is to be able to construct really efficient
diff --git a/lib/Tensors.h b/lib/Tensors.h
new file mode 100644
index 00000000..e812149f
--- /dev/null
+++ b/lib/Tensors.h
@@ -0,0 +1,16 @@
+#ifndef GRID_MATH_H
+#define GRID_MATH_H
+
+#include <tensors/Tensor_traits.h>
+#include <tensors/Tensor_class.h>
+#include <tensors/Tensor_arith.h>
+#include <tensors/Tensor_inner.h>
+#include <tensors/Tensor_outer.h>
+#include <tensors/Tensor_transpose.h>
+#include <tensors/Tensor_trace.h>
+#include <tensors/Tensor_peek.h>
+#include <tensors/Tensor_poke.h>
+#include <tensors/Tensor_reality.h>
+#include <tensors/Tensor_extract_merge.h>
+    
+#endif
diff --git a/lib/Grid_threads.h b/lib/Threads.h
similarity index 100%
rename from lib/Grid_threads.h
rename to lib/Threads.h
diff --git a/lib/cartesian/Grid_cartesian_base.h b/lib/cartesian/Cartesian_base.h
similarity index 99%
rename from lib/cartesian/Grid_cartesian_base.h
rename to lib/cartesian/Cartesian_base.h
index 66339648..6303e38e 100644
--- a/lib/cartesian/Grid_cartesian_base.h
+++ b/lib/cartesian/Cartesian_base.h
@@ -2,7 +2,6 @@
 #define GRID_CARTESIAN_BASE_H
 
 #include <Grid.h>
-#include <Grid_communicator.h>
 
 namespace Grid{
 
diff --git a/lib/cartesian/Grid_cartesian_full.h b/lib/cartesian/Cartesian_full.h
similarity index 100%
rename from lib/cartesian/Grid_cartesian_full.h
rename to lib/cartesian/Cartesian_full.h
diff --git a/lib/cartesian/Grid_cartesian_red_black.h b/lib/cartesian/Cartesian_red_black.h
similarity index 100%
rename from lib/cartesian/Grid_cartesian_red_black.h
rename to lib/cartesian/Cartesian_red_black.h
diff --git a/lib/communicator/Grid_communicator_base.h b/lib/communicator/Communicator_base.h
similarity index 100%
rename from lib/communicator/Grid_communicator_base.h
rename to lib/communicator/Communicator_base.h
diff --git a/lib/communicator/Grid_communicator_mpi.cc b/lib/communicator/Communicator_mpi.cc
similarity index 100%
rename from lib/communicator/Grid_communicator_mpi.cc
rename to lib/communicator/Communicator_mpi.cc
diff --git a/lib/communicator/Grid_communicator_none.cc b/lib/communicator/Communicator_none.cc
similarity index 100%
rename from lib/communicator/Grid_communicator_none.cc
rename to lib/communicator/Communicator_none.cc
diff --git a/lib/cshift/Grid_cshift_common.h b/lib/cshift/Cshift_common.h
similarity index 100%
rename from lib/cshift/Grid_cshift_common.h
rename to lib/cshift/Cshift_common.h
diff --git a/lib/cshift/Grid_cshift_mpi.h b/lib/cshift/Cshift_mpi.h
similarity index 100%
rename from lib/cshift/Grid_cshift_mpi.h
rename to lib/cshift/Cshift_mpi.h
diff --git a/lib/cshift/Grid_cshift_none.h b/lib/cshift/Cshift_none.h
similarity index 100%
rename from lib/cshift/Grid_cshift_none.h
rename to lib/cshift/Cshift_none.h
diff --git a/lib/lattice/Grid_lattice_ET.h b/lib/lattice/Lattice_ET.h
similarity index 100%
rename from lib/lattice/Grid_lattice_ET.h
rename to lib/lattice/Lattice_ET.h
diff --git a/lib/lattice/Grid_lattice_arith.h b/lib/lattice/Lattice_arith.h
similarity index 100%
rename from lib/lattice/Grid_lattice_arith.h
rename to lib/lattice/Lattice_arith.h
diff --git a/lib/lattice/Grid_lattice_base.h b/lib/lattice/Lattice_base.h
similarity index 93%
rename from lib/lattice/Grid_lattice_base.h
rename to lib/lattice/Lattice_base.h
index 15ef0944..6b5fe261 100644
--- a/lib/lattice/Grid_lattice_base.h
+++ b/lib/lattice/Lattice_base.h
@@ -283,24 +283,23 @@ PARALLEL_FOR_LOOP
 
 
 
-#include <lattice/Grid_lattice_conformable.h>
+#include <lattice/Lattice_conformable.h>
 #define GRID_LATTICE_EXPRESSION_TEMPLATES
 #ifdef  GRID_LATTICE_EXPRESSION_TEMPLATES
-#include <lattice/Grid_lattice_ET.h>
+#include <lattice/Lattice_ET.h>
 #else 
-#include <lattice/Grid_lattice_overload.h>
+#include <lattice/Lattice_overload.h>
 #endif
-#include <lattice/Grid_lattice_arith.h>
-#include <lattice/Grid_lattice_trace.h>
-#include <lattice/Grid_lattice_transpose.h>
-#include <lattice/Grid_lattice_local.h>
-#include <lattice/Grid_lattice_reduction.h>
-#include <lattice/Grid_lattice_peekpoke.h>
-#include <lattice/Grid_lattice_reality.h>
-#include <Grid_extract.h>
-#include <lattice/Grid_lattice_coordinate.h>
-#include <lattice/Grid_lattice_rng.h>
-#include <lattice/Grid_lattice_transfer.h>
+#include <lattice/Lattice_arith.h>
+#include <lattice/Lattice_trace.h>
+#include <lattice/Lattice_transpose.h>
+#include <lattice/Lattice_local.h>
+#include <lattice/Lattice_reduction.h>
+#include <lattice/Lattice_peekpoke.h>
+#include <lattice/Lattice_reality.h>
+#include <lattice/Lattice_coordinate.h>
+#include <lattice/Lattice_rng.h>
+#include <lattice/Lattice_transfer.h>
 
 
 
diff --git a/lib/lattice/Grid_lattice_comparison.h b/lib/lattice/Lattice_comparison.h
similarity index 100%
rename from lib/lattice/Grid_lattice_comparison.h
rename to lib/lattice/Lattice_comparison.h
diff --git a/lib/lattice/Grid_lattice_conformable.h b/lib/lattice/Lattice_conformable.h
similarity index 100%
rename from lib/lattice/Grid_lattice_conformable.h
rename to lib/lattice/Lattice_conformable.h
diff --git a/lib/lattice/Grid_lattice_coordinate.h b/lib/lattice/Lattice_coordinate.h
similarity index 100%
rename from lib/lattice/Grid_lattice_coordinate.h
rename to lib/lattice/Lattice_coordinate.h
diff --git a/lib/lattice/Grid_lattice_local.h b/lib/lattice/Lattice_local.h
similarity index 100%
rename from lib/lattice/Grid_lattice_local.h
rename to lib/lattice/Lattice_local.h
diff --git a/lib/lattice/Grid_lattice_overload.h b/lib/lattice/Lattice_overload.h
similarity index 100%
rename from lib/lattice/Grid_lattice_overload.h
rename to lib/lattice/Lattice_overload.h
diff --git a/lib/lattice/Grid_lattice_peekpoke.h b/lib/lattice/Lattice_peekpoke.h
similarity index 100%
rename from lib/lattice/Grid_lattice_peekpoke.h
rename to lib/lattice/Lattice_peekpoke.h
diff --git a/lib/lattice/Grid_lattice_reality.h b/lib/lattice/Lattice_reality.h
similarity index 100%
rename from lib/lattice/Grid_lattice_reality.h
rename to lib/lattice/Lattice_reality.h
diff --git a/lib/lattice/Grid_lattice_reduction.h b/lib/lattice/Lattice_reduction.h
similarity index 100%
rename from lib/lattice/Grid_lattice_reduction.h
rename to lib/lattice/Lattice_reduction.h
diff --git a/lib/lattice/Grid_lattice_rng.h b/lib/lattice/Lattice_rng.h
similarity index 100%
rename from lib/lattice/Grid_lattice_rng.h
rename to lib/lattice/Lattice_rng.h
diff --git a/lib/lattice/Grid_lattice_trace.h b/lib/lattice/Lattice_trace.h
similarity index 100%
rename from lib/lattice/Grid_lattice_trace.h
rename to lib/lattice/Lattice_trace.h
diff --git a/lib/lattice/Grid_lattice_transfer.h b/lib/lattice/Lattice_transfer.h
similarity index 100%
rename from lib/lattice/Grid_lattice_transfer.h
rename to lib/lattice/Lattice_transfer.h
diff --git a/lib/lattice/Grid_lattice_transpose.h b/lib/lattice/Lattice_transpose.h
similarity index 100%
rename from lib/lattice/Grid_lattice_transpose.h
rename to lib/lattice/Lattice_transpose.h
diff --git a/lib/lattice/Grid_lattice_where.h b/lib/lattice/Lattice_where.h
similarity index 100%
rename from lib/lattice/Grid_lattice_where.h
rename to lib/lattice/Lattice_where.h
diff --git a/lib/math/Grid_math_arith.h b/lib/math/Grid_math_arith.h
deleted file mode 100644
index ca90ba88..00000000
--- a/lib/math/Grid_math_arith.h
+++ /dev/null
@@ -1,11 +0,0 @@
-#ifndef GRID_MATH_ARITH_H
-#define GRID_MATH_ARITH_H
-
-#include <math/Grid_math_arith_add.h>
-#include <math/Grid_math_arith_sub.h>
-#include <math/Grid_math_arith_mac.h>
-#include <math/Grid_math_arith_mul.h>
-#include <math/Grid_math_arith_scalar.h>
-
-#endif
-
diff --git a/lib/parallelIO/GridNerscIO.h b/lib/parallelIO/NerscIO.h
similarity index 100%
rename from lib/parallelIO/GridNerscIO.h
rename to lib/parallelIO/NerscIO.h
diff --git a/lib/qcd/action/fermion/PartialFractionFermion5D.cc b/lib/qcd/action/fermion/PartialFractionFermion5D.cc
index 21f62a75..8b137891 100644
--- a/lib/qcd/action/fermion/PartialFractionFermion5D.cc
+++ b/lib/qcd/action/fermion/PartialFractionFermion5D.cc
@@ -1,47 +1 @@
-#ifndef  GRID_QCD_PARTIAL_FRACTION_H
-#define  GRID_QCD_PARTIAL_FRACTION_H
 
-namespace Grid {
-
-  namespace QCD {
-
-    class PartialFractionFermion5D : public WilsonFermion5D
-    {
-    public:
-
-      // override multiply
-      virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out);
-      virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out);
-
-      // half checkerboard operaions
-      virtual void   Meooe       (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
-
-    private:
-
-      zolotarev_data *zdata;
-
-      // Part frac
-      double R=(1+this->mass)/(1-this->mass);
-      std::vector<double> p; 
-      std::vector<double> q;
-
-      // Constructors
-      PartialFractionFermion5D(LatticeGaugeField &_Umu,
-				    GridCartesian         &FiveDimGrid,
-				    GridRedBlackCartesian &FiveDimRedBlackGrid,
-				    GridCartesian         &FourDimGrid,
-				    GridRedBlackCartesian &FourDimRedBlackGrid,
-				    RealD _mass,RealD M5);
-
-    };
-
-
-  }
-}
-
-#endif
diff --git a/lib/stencil/Grid_lebesgue.cc b/lib/stencil/Lebesgue.cc
similarity index 100%
rename from lib/stencil/Grid_lebesgue.cc
rename to lib/stencil/Lebesgue.cc
diff --git a/lib/stencil/Grid_lebesgue.h b/lib/stencil/Lebesgue.h
similarity index 100%
rename from lib/stencil/Grid_lebesgue.h
rename to lib/stencil/Lebesgue.h
diff --git a/lib/stencil/Grid_stencil_common.cc b/lib/stencil/Stencil_common.cc
similarity index 100%
rename from lib/stencil/Grid_stencil_common.cc
rename to lib/stencil/Stencil_common.cc
diff --git a/lib/tensors/Tensor_arith.h b/lib/tensors/Tensor_arith.h
new file mode 100644
index 00000000..853a19a5
--- /dev/null
+++ b/lib/tensors/Tensor_arith.h
@@ -0,0 +1,11 @@
+#ifndef GRID_MATH_ARITH_H
+#define GRID_MATH_ARITH_H
+
+#include <tensors/Tensor_arith_add.h>
+#include <tensors/Tensor_arith_sub.h>
+#include <tensors/Tensor_arith_mac.h>
+#include <tensors/Tensor_arith_mul.h>
+#include <tensors/Tensor_arith_scalar.h>
+
+#endif
+
diff --git a/lib/math/Grid_math_arith_add.h b/lib/tensors/Tensor_arith_add.h
similarity index 100%
rename from lib/math/Grid_math_arith_add.h
rename to lib/tensors/Tensor_arith_add.h
diff --git a/lib/math/Grid_math_arith_mac.h b/lib/tensors/Tensor_arith_mac.h
similarity index 100%
rename from lib/math/Grid_math_arith_mac.h
rename to lib/tensors/Tensor_arith_mac.h
diff --git a/lib/math/Grid_math_arith_mul.h b/lib/tensors/Tensor_arith_mul.h
similarity index 100%
rename from lib/math/Grid_math_arith_mul.h
rename to lib/tensors/Tensor_arith_mul.h
diff --git a/lib/math/Grid_math_arith_scalar.h b/lib/tensors/Tensor_arith_scalar.h
similarity index 100%
rename from lib/math/Grid_math_arith_scalar.h
rename to lib/tensors/Tensor_arith_scalar.h
diff --git a/lib/math/Grid_math_arith_sub.h b/lib/tensors/Tensor_arith_sub.h
similarity index 100%
rename from lib/math/Grid_math_arith_sub.h
rename to lib/tensors/Tensor_arith_sub.h
diff --git a/lib/math/Grid_math_tensors.h b/lib/tensors/Tensor_class.h
similarity index 100%
rename from lib/math/Grid_math_tensors.h
rename to lib/tensors/Tensor_class.h
diff --git a/lib/Grid_extract.h b/lib/tensors/Tensor_extract_merge.h
similarity index 100%
rename from lib/Grid_extract.h
rename to lib/tensors/Tensor_extract_merge.h
diff --git a/lib/math/Grid_math_inner.h b/lib/tensors/Tensor_inner.h
similarity index 100%
rename from lib/math/Grid_math_inner.h
rename to lib/tensors/Tensor_inner.h
diff --git a/lib/math/Grid_math_outer.h b/lib/tensors/Tensor_outer.h
similarity index 100%
rename from lib/math/Grid_math_outer.h
rename to lib/tensors/Tensor_outer.h
diff --git a/lib/math/Grid_math_peek.h b/lib/tensors/Tensor_peek.h
similarity index 100%
rename from lib/math/Grid_math_peek.h
rename to lib/tensors/Tensor_peek.h
diff --git a/lib/math/Grid_math_poke.h b/lib/tensors/Tensor_poke.h
similarity index 100%
rename from lib/math/Grid_math_poke.h
rename to lib/tensors/Tensor_poke.h
diff --git a/lib/math/Grid_math_reality.h b/lib/tensors/Tensor_reality.h
similarity index 100%
rename from lib/math/Grid_math_reality.h
rename to lib/tensors/Tensor_reality.h
diff --git a/lib/math/Grid_math_trace.h b/lib/tensors/Tensor_trace.h
similarity index 100%
rename from lib/math/Grid_math_trace.h
rename to lib/tensors/Tensor_trace.h
diff --git a/lib/math/Grid_math_traits.h b/lib/tensors/Tensor_traits.h
similarity index 100%
rename from lib/math/Grid_math_traits.h
rename to lib/tensors/Tensor_traits.h
diff --git a/lib/math/Grid_math_transpose.h b/lib/tensors/Tensor_transpose.h
similarity index 100%
rename from lib/math/Grid_math_transpose.h
rename to lib/tensors/Tensor_transpose.h
diff --git a/tests/Test_cshift.cc b/tests/Test_cshift.cc
index 76125977..cc1da333 100644
--- a/tests/Test_cshift.cc
+++ b/tests/Test_cshift.cc
@@ -1,5 +1,4 @@
 #include <Grid.h>
-#include <parallelIO/GridNerscIO.h>
 
 using namespace Grid;
 using namespace Grid::QCD;
diff --git a/tests/Test_cshift_red_black.cc b/tests/Test_cshift_red_black.cc
index 212e9b6e..9ffa66b1 100644
--- a/tests/Test_cshift_red_black.cc
+++ b/tests/Test_cshift_red_black.cc
@@ -1,5 +1,4 @@
 #include <Grid.h>
-#include <parallelIO/GridNerscIO.h>
 
 using namespace Grid;
 using namespace Grid::QCD;
diff --git a/tests/Test_gamma.cc b/tests/Test_gamma.cc
index e803029b..bae77ecc 100644
--- a/tests/Test_gamma.cc
+++ b/tests/Test_gamma.cc
@@ -1,5 +1,4 @@
 #include <Grid.h>
-#include <parallelIO/GridNerscIO.h>
 
 using namespace std;
 using namespace Grid;
diff --git a/tests/Test_nersc_io.cc b/tests/Test_nersc_io.cc
index 6fe587a6..80d78291 100644
--- a/tests/Test_nersc_io.cc
+++ b/tests/Test_nersc_io.cc
@@ -1,5 +1,4 @@
 #include <Grid.h>
-#include <parallelIO/GridNerscIO.h>
 
 using namespace std;
 using namespace Grid;
diff --git a/tests/Test_rng.cc b/tests/Test_rng.cc
index 97f6c6b7..1731b740 100644
--- a/tests/Test_rng.cc
+++ b/tests/Test_rng.cc
@@ -1,5 +1,4 @@
 #include <Grid.h>
-#include <parallelIO/GridNerscIO.h>
 
 using namespace std;
 using namespace Grid;
diff --git a/tests/Test_rng_fixed.cc b/tests/Test_rng_fixed.cc
index 01e3315b..c836c93f 100644
--- a/tests/Test_rng_fixed.cc
+++ b/tests/Test_rng_fixed.cc
@@ -1,5 +1,4 @@
 #include <Grid.h>
-#include <parallelIO/GridNerscIO.h>
 
 using namespace std;
 using namespace Grid;
diff --git a/tests/Test_simd.cc b/tests/Test_simd.cc
index db600fe4..f4858d06 100644
--- a/tests/Test_simd.cc
+++ b/tests/Test_simd.cc
@@ -1,5 +1,4 @@
 #include <Grid.h>
-#include <parallelIO/GridNerscIO.h>
 
 using namespace std;
 using namespace Grid;

From 98dcb6831bc110e85106168453d3c2400d9ff2c7 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 3 Jun 2015 12:47:46 +0100
Subject: [PATCH 263/429] Convenience script to build the list of headers and
 .cc files in the library

---
 scripts/filelist | 9 +++++++++
 1 file changed, 9 insertions(+)
 create mode 100755 scripts/filelist

diff --git a/scripts/filelist b/scripts/filelist
new file mode 100755
index 00000000..73c6db51
--- /dev/null
+++ b/scripts/filelist
@@ -0,0 +1,9 @@
+#!/bin/bash
+
+cd lib
+HFILES=`find . -type f -name '*.h'`
+CCFILES=`find . -type f -name '*.cc' -not  -name '*ommunicator*.cc'`
+echo HFILES=$HFILES > Make.inc
+echo >> Make.inc
+echo CCFILES=$CCFILES >> Make.inc
+cd ..

From 4ef11d96e958627deda5c06dabce9d1a5d2b5bfc Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 3 Jun 2015 12:49:36 +0100
Subject: [PATCH 264/429] Make.inc needed in repo

---
 lib/Make.inc | 3 +++
 1 file changed, 3 insertions(+)
 create mode 100644 lib/Make.inc

diff --git a/lib/Make.inc b/lib/Make.inc
new file mode 100644
index 00000000..31683256
--- /dev/null
+++ b/lib/Make.inc
@@ -0,0 +1,3 @@
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Comparison.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/Dirac.h ./qcd/LinalgUtils.h ./qcd/QCD.h ./qcd/SpaceTimeGrid.h ./qcd/TwoSpinor.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Old/Grid_vComplexD.h ./simd/Old/Grid_vComplexF.h ./simd/Old/Grid_vInteger.h ./simd/Old/Grid_vRealD.h ./simd/Old/Grid_vRealF.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./Tensors.h ./Threads.h
+
+CCFILES=./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/Dirac.cc ./qcd/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc

From eaa3e6aaf6aff695eb8bc60e384133864092a0d0 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 3 Jun 2015 13:07:00 +0100
Subject: [PATCH 265/429] Assist for generating file lists contained in
 Make.inc files for convenience when things are added

---
 benchmarks/Make.inc                           |  23 +++
 benchmarks/Makefile.am                        |  23 +--
 ...nchmark_su3_expr.cc => simple_su3_expr.cc} |   0
 ...nchmark_su3_test.cc => simple_su3_test.cc} |   0
 lib/Make.inc                                  |   1 +
 scripts/filelist                              |  49 +++++-
 tests/Make.inc                                |  83 +++++++++
 tests/Makefile.am                             |  89 +---------
 tests/Test_simd_new.cc                        | 165 ------------------
 9 files changed, 157 insertions(+), 276 deletions(-)
 create mode 100644 benchmarks/Make.inc
 rename benchmarks/{Benchmark_su3_expr.cc => simple_su3_expr.cc} (100%)
 rename benchmarks/{Benchmark_su3_test.cc => simple_su3_test.cc} (100%)
 create mode 100644 tests/Make.inc
 delete mode 100644 tests/Test_simd_new.cc

diff --git a/benchmarks/Make.inc b/benchmarks/Make.inc
new file mode 100644
index 00000000..9871149c
--- /dev/null
+++ b/benchmarks/Make.inc
@@ -0,0 +1,23 @@
+
+bin_PROGRAMS = Benchmark_comms Benchmark_dwf Benchmark_memory_bandwidth Benchmark_su3 Benchmark_wilson
+
+
+Benchmark_comms_SOURCES=Benchmark_comms.cc
+Benchmark_comms_LDADD=-lGrid
+
+
+Benchmark_dwf_SOURCES=Benchmark_dwf.cc
+Benchmark_dwf_LDADD=-lGrid
+
+
+Benchmark_memory_bandwidth_SOURCES=Benchmark_memory_bandwidth.cc
+Benchmark_memory_bandwidth_LDADD=-lGrid
+
+
+Benchmark_su3_SOURCES=Benchmark_su3.cc
+Benchmark_su3_LDADD=-lGrid
+
+
+Benchmark_wilson_SOURCES=Benchmark_wilson.cc
+Benchmark_wilson_LDADD=-lGrid
+
diff --git a/benchmarks/Makefile.am b/benchmarks/Makefile.am
index 74ff03c6..8081844f 100644
--- a/benchmarks/Makefile.am
+++ b/benchmarks/Makefile.am
@@ -5,25 +5,4 @@ AM_LDFLAGS = -L$(top_builddir)/lib
 #
 # Test code
 #
-bin_PROGRAMS = \
-	Benchmark_comms \
-	Benchmark_memory_bandwidth \
-	Benchmark_su3 \
-	Benchmark_wilson \
-	Benchmark_dwf
-
-Benchmark_comms_SOURCES = Benchmark_comms.cc
-Benchmark_comms_LDADD = -lGrid
-
-Benchmark_memory_bandwidth_SOURCES = Benchmark_memory_bandwidth.cc
-Benchmark_memory_bandwidth_LDADD = -lGrid
-
-Benchmark_su3_SOURCES = Benchmark_su3.cc Benchmark_su3_test.cc Benchmark_su3_expr.cc
-Benchmark_su3_LDADD = -lGrid
-
-Benchmark_wilson_SOURCES = Benchmark_wilson.cc
-Benchmark_wilson_LDADD = -lGrid
-
-Benchmark_dwf_SOURCES = Benchmark_dwf.cc
-Benchmark_dwf_LDADD = -lGrid
-
+include Make.inc
diff --git a/benchmarks/Benchmark_su3_expr.cc b/benchmarks/simple_su3_expr.cc
similarity index 100%
rename from benchmarks/Benchmark_su3_expr.cc
rename to benchmarks/simple_su3_expr.cc
diff --git a/benchmarks/Benchmark_su3_test.cc b/benchmarks/simple_su3_test.cc
similarity index 100%
rename from benchmarks/Benchmark_su3_test.cc
rename to benchmarks/simple_su3_test.cc
diff --git a/lib/Make.inc b/lib/Make.inc
index 31683256..dd5269ac 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,3 +1,4 @@
+
 HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Comparison.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/Dirac.h ./qcd/LinalgUtils.h ./qcd/QCD.h ./qcd/SpaceTimeGrid.h ./qcd/TwoSpinor.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Old/Grid_vComplexD.h ./simd/Old/Grid_vComplexF.h ./simd/Old/Grid_vInteger.h ./simd/Old/Grid_vRealD.h ./simd/Old/Grid_vRealF.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./Tensors.h ./Threads.h
 
 CCFILES=./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/Dirac.cc ./qcd/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/scripts/filelist b/scripts/filelist
index 73c6db51..1b575611 100755
--- a/scripts/filelist
+++ b/scripts/filelist
@@ -1,9 +1,56 @@
 #!/bin/bash
 
 cd lib
+
 HFILES=`find . -type f -name '*.h'`
 CCFILES=`find . -type f -name '*.cc' -not  -name '*ommunicator*.cc'`
-echo HFILES=$HFILES > Make.inc
+echo> Make.inc
+echo HFILES=$HFILES >> Make.inc
 echo >> Make.inc
 echo CCFILES=$CCFILES >> Make.inc
+
 cd ..
+
+
+
+cd tests
+
+echo> Make.inc
+TESTS=`ls T*.cc`
+TESTLIST=`echo ${TESTS} | sed s/.cc//g `
+
+echo > Make.inc
+echo bin_PROGRAMS = ${TESTLIST} >> Make.inc
+echo >> Make.inc
+
+for f in $TESTS
+do
+BNAME=`basename $f .cc`
+echo >> Make.inc
+echo ${BNAME}_SOURCES=$f  >> Make.inc
+echo ${BNAME}_LDADD=-lGrid>> Make.inc
+echo >> Make.inc
+done
+
+cd ..
+
+
+cd benchmarks
+
+
+echo> Make.inc
+TESTS=`ls B*.cc`
+TESTLIST=`echo ${TESTS} | sed s/.cc//g `
+
+echo > Make.inc
+echo bin_PROGRAMS = ${TESTLIST} >> Make.inc
+echo >> Make.inc
+
+for f in $TESTS
+do
+BNAME=`basename $f .cc`
+echo >> Make.inc
+echo ${BNAME}_SOURCES=$f  >> Make.inc
+echo ${BNAME}_LDADD=-lGrid>> Make.inc
+echo >> Make.inc
+done
diff --git a/tests/Make.inc b/tests/Make.inc
new file mode 100644
index 00000000..d592f218
--- /dev/null
+++ b/tests/Make.inc
@@ -0,0 +1,83 @@
+
+bin_PROGRAMS = Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_even_odd Test_gamma Test_main Test_many_cg Test_many_evenodd Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_evenodd
+
+
+Test_cshift_SOURCES=Test_cshift.cc
+Test_cshift_LDADD=-lGrid
+
+
+Test_cshift_red_black_SOURCES=Test_cshift_red_black.cc
+Test_cshift_red_black_LDADD=-lGrid
+
+
+Test_dwf_cg_prec_SOURCES=Test_dwf_cg_prec.cc
+Test_dwf_cg_prec_LDADD=-lGrid
+
+
+Test_dwf_cg_schur_SOURCES=Test_dwf_cg_schur.cc
+Test_dwf_cg_schur_LDADD=-lGrid
+
+
+Test_dwf_cg_unprec_SOURCES=Test_dwf_cg_unprec.cc
+Test_dwf_cg_unprec_LDADD=-lGrid
+
+
+Test_dwf_even_odd_SOURCES=Test_dwf_even_odd.cc
+Test_dwf_even_odd_LDADD=-lGrid
+
+
+Test_gamma_SOURCES=Test_gamma.cc
+Test_gamma_LDADD=-lGrid
+
+
+Test_main_SOURCES=Test_main.cc
+Test_main_LDADD=-lGrid
+
+
+Test_many_cg_SOURCES=Test_many_cg.cc
+Test_many_cg_LDADD=-lGrid
+
+
+Test_many_evenodd_SOURCES=Test_many_evenodd.cc
+Test_many_evenodd_LDADD=-lGrid
+
+
+Test_nersc_io_SOURCES=Test_nersc_io.cc
+Test_nersc_io_LDADD=-lGrid
+
+
+Test_remez_SOURCES=Test_remez.cc
+Test_remez_LDADD=-lGrid
+
+
+Test_rng_SOURCES=Test_rng.cc
+Test_rng_LDADD=-lGrid
+
+
+Test_rng_fixed_SOURCES=Test_rng_fixed.cc
+Test_rng_fixed_LDADD=-lGrid
+
+
+Test_simd_SOURCES=Test_simd.cc
+Test_simd_LDADD=-lGrid
+
+
+Test_stencil_SOURCES=Test_stencil.cc
+Test_stencil_LDADD=-lGrid
+
+
+Test_wilson_cg_prec_SOURCES=Test_wilson_cg_prec.cc
+Test_wilson_cg_prec_LDADD=-lGrid
+
+
+Test_wilson_cg_schur_SOURCES=Test_wilson_cg_schur.cc
+Test_wilson_cg_schur_LDADD=-lGrid
+
+
+Test_wilson_cg_unprec_SOURCES=Test_wilson_cg_unprec.cc
+Test_wilson_cg_unprec_LDADD=-lGrid
+
+
+Test_wilson_evenodd_SOURCES=Test_wilson_evenodd.cc
+Test_wilson_evenodd_LDADD=-lGrid
+
diff --git a/tests/Makefile.am b/tests/Makefile.am
index bb3448f1..83385001 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -2,91 +2,4 @@
 AM_CXXFLAGS = -I$(top_srcdir)/lib
 AM_LDFLAGS = -L$(top_builddir)/lib
 
-#
-# Test code
-#
-bin_PROGRAMS = Test_main \
-	Test_simd \
-	Test_gamma  \
-	Test_cshift \
-	Test_cshift_red_black \
-	Test_stencil \
-	Test_nersc_io \
-	Test_rng \
-	Test_remez \
-	Test_rng_fixed \
-	Test_wilson_evenodd \
-	Test_wilson_cg_unprec \
-	Test_wilson_cg_prec \
-	Test_wilson_cg_schur \
-	Test_dwf_even_odd\
-	Test_dwf_cg_unprec\
-	Test_dwf_cg_prec\
-	Test_dwf_cg_schur\
-	Test_many_evenodd\
-	Test_many_cg
-
-
-
-Test_main_SOURCES = Test_main.cc
-Test_main_LDADD = -lGrid
-
-Test_rng_SOURCES = Test_rng.cc
-Test_rng_LDADD = -lGrid
-
-Test_rng_fixed_SOURCES = Test_rng_fixed.cc
-Test_rng_fixed_LDADD = -lGrid
-
-Test_remez_SOURCES = Test_remez.cc
-Test_remez_LDADD = -lGrid
-
-Test_nersc_io_SOURCES = Test_nersc_io.cc
-Test_nersc_io_LDADD = -lGrid
-
-Test_cshift_SOURCES = Test_cshift.cc
-Test_cshift_LDADD = -lGrid
-
-Test_cshift_red_black_SOURCES = Test_cshift_red_black.cc
-Test_cshift_red_black_LDADD = -lGrid
-
-Test_gamma_SOURCES = Test_gamma.cc
-Test_gamma_LDADD = -lGrid
-
-Test_stencil_SOURCES = Test_stencil.cc
-Test_stencil_LDADD = -lGrid
-
-Test_simd_SOURCES = Test_simd.cc
-Test_simd_LDADD = -lGrid
-
-#Test_simd_new_SOURCES = Test_simd_new.cc
-#Test_simd_new_LDADD = -lGrid
-
-Test_wilson_evenodd_SOURCES = Test_wilson_evenodd.cc
-Test_wilson_evenodd_LDADD = -lGrid
-
-Test_wilson_cg_unprec_SOURCES = Test_wilson_cg_unprec.cc
-Test_wilson_cg_unprec_LDADD = -lGrid
-
-Test_wilson_cg_prec_SOURCES = Test_wilson_cg_prec.cc
-Test_wilson_cg_prec_LDADD = -lGrid
-
-Test_wilson_cg_schur_SOURCES = Test_wilson_cg_schur.cc
-Test_wilson_cg_schur_LDADD = -lGrid
-
-Test_dwf_even_odd_SOURCES = Test_dwf_even_odd.cc
-Test_dwf_even_odd_LDADD = -lGrid
-
-Test_dwf_cg_unprec_SOURCES = Test_dwf_cg_unprec.cc
-Test_dwf_cg_unprec_LDADD = -lGrid
-
-Test_dwf_cg_prec_SOURCES = Test_dwf_cg_prec.cc
-Test_dwf_cg_prec_LDADD = -lGrid
-
-Test_dwf_cg_schur_SOURCES = Test_dwf_cg_schur.cc
-Test_dwf_cg_schur_LDADD = -lGrid
-
-Test_many_evenodd_SOURCES = Test_many_evenodd.cc
-Test_many_evenodd_LDADD = -lGrid
-
-Test_many_cg_SOURCES = Test_many_cg.cc
-Test_many_cg_LDADD = -lGrid
+include Make.inc
diff --git a/tests/Test_simd_new.cc b/tests/Test_simd_new.cc
deleted file mode 100644
index 41781304..00000000
--- a/tests/Test_simd_new.cc
+++ /dev/null
@@ -1,165 +0,0 @@
-#include <Grid.h>
-#include "simd/Grid_vector_types.h"
-#include <parallelIO/GridNerscIO.h>
-
-using namespace std;
-using namespace Grid;
-using namespace Grid::QCD;
-
-class funcPlus {
-public:
-  funcPlus() {};
-  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = i1+i2;}
-  std::string name(void) const { return std::string("Plus"); }
-};
-class funcMinus {
-public:
-  funcMinus() {};
-  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = i1-i2;}
-  std::string name(void) const { return std::string("Minus"); }
-};
-class funcTimes {
-public:
-  funcTimes() {};
-  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = i1*i2;}
-  std::string name(void) const { return std::string("Times"); }
-};
-class funcConj {
-public:
-  funcConj() {};
-  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = conjugate(i1);}
-  std::string name(void) const { return std::string("Conj"); }
-};
-class funcAdj {
-public:
-  funcAdj() {};
-  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = adj(i1);}
-  std::string name(void) const { return std::string("Adj"); }
-};
-
-class funcTimesI {
-public:
-  funcTimesI() {};
-  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = timesI(i1);}
-  std::string name(void) const { return std::string("timesI"); }
-};
-
-class funcTimesMinusI {
-public:
-  funcTimesMinusI() {};
-  template<class vec> void operator()(vec &rr,vec &i1,vec &i2) const { rr = timesMinusI(i1);}
-  std::string name(void) const { return std::string("timesMinusI"); }
-};
-
-template<class scal, class vec,class functor > 
-void Tester(const functor &func)
-{
-  GridSerialRNG          sRNG;
-  sRNG.SeedRandomDevice();
-  
-  int Nsimd = vec::Nsimd();
-
-  std::vector<scal> input1(Nsimd);
-  std::vector<scal> input2(Nsimd);
-  std::vector<scal> result(Nsimd);
-  std::vector<scal> reference(Nsimd);
-
-  std::vector<vec,alignedAllocator<vec> > buf(3);
-  vec & v_input1 = buf[0];
-  vec & v_input2 = buf[1];
-  vec & v_result = buf[2];
-
-
-  for(int i=0;i<Nsimd;i++){
-    random(sRNG,input1[i]);
-    random(sRNG,input2[i]);
-    random(sRNG,result[i]);
-  }
-
-  merge<vec,scal>(v_input1,input1);
-  merge<vec,scal>(v_input2,input2);
-  merge<vec,scal>(v_result,result);
-
-  func(v_result,v_input1,v_input2);
-
-  for(int i=0;i<Nsimd;i++) {
-    func(reference[i],input1[i],input2[i]);
-  }
-
-  extract<vec,scal>(v_result,result);
-  std::cout << " " << func.name()<<std::endl;
-
-  int ok=0;
-  for(int i=0;i<Nsimd;i++){
-    if ( abs(reference[i]-result[i])>0){
-      std::cout<< "*****" << std::endl;
-      std::cout<< "["<<i<<"] "<< abs(reference[i]-result[i]) << " " <<reference[i]<< " " << result[i]<<std::endl;
-      ok++;
-    }
-  }
-  if ( ok==0 ) std::cout << " OK!" <<std::endl;
-
-}
-
-
-
-int main (int argc, char ** argv)
-{
-  Grid_init(&argc,&argv);
-
-  std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd(4,MyComplexF::Nsimd());
-  std::vector<int> mpi_layout  = GridDefaultMpi();
-    
-  GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
-  std::vector<int> seeds({1,2,3,4});
-
-  // Insist that operations on random scalars gives
-  // identical results to on vectors.
-
-  std::cout << "==================================="<<  std::endl;
-  std::cout << "Testing MyComplexF "<<std::endl;
-  std::cout << "==================================="<<  std::endl;
-
-  Tester<ComplexF,MyComplexF>(funcTimesI());
-  Tester<ComplexF,MyComplexF>(funcTimesMinusI());
-  Tester<ComplexF,MyComplexF>(funcPlus());
-  Tester<ComplexF,MyComplexF>(funcMinus());
-  Tester<ComplexF,MyComplexF>(funcTimes());
-  Tester<ComplexF,MyComplexF>(funcConj());
-  Tester<ComplexF,MyComplexF>(funcAdj());
-
-  std::cout << "==================================="<<  std::endl;
-  std::cout << "Testing MyComplexD "<<std::endl;
-  std::cout << "==================================="<<  std::endl;
-
-
-  Tester<ComplexD,MyComplexD>(funcTimesI());
-  Tester<ComplexD,MyComplexD>(funcTimesMinusI());
-  Tester<ComplexD,MyComplexD>(funcPlus());
-  Tester<ComplexD,MyComplexD>(funcMinus());
-  Tester<ComplexD,MyComplexD>(funcTimes());
-  Tester<ComplexD,MyComplexD>(funcConj());
-  Tester<ComplexD,MyComplexD>(funcAdj());
-
-  std::cout << "==================================="<<  std::endl;
-  std::cout << "Testing MyRealF "<<std::endl;
-  std::cout << "==================================="<<  std::endl;
-
-
-  Tester<RealF,MyRealF>(funcPlus());
-  Tester<RealF,MyRealF>(funcMinus());
-  Tester<RealF,MyRealF>(funcTimes());
-  Tester<RealF,MyRealF>(funcAdj());
-
-  std::cout << "==================================="<<  std::endl;
-  std::cout << "Testing MyRealD "<<std::endl;
-  std::cout << "==================================="<<  std::endl;
-
-  Tester<RealD,MyRealD>(funcPlus());
-  Tester<RealD,MyRealD>(funcMinus());
-  Tester<RealD,MyRealD>(funcTimes());
-  Tester<RealD,MyRealD>(funcAdj());
-
-  Grid_finalize();
-}

From 50bd2935278d283b53fe24091aa7bb0abb05ef97 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 4 Jun 2015 00:00:45 +0100
Subject: [PATCH 266/429] First pass at continued fraction; solver and even odd
 decomposition tests pass. Have to make ContFrac class virtual and derive end
 non-abstract actions for the particular cases.

---
 lib/qcd/action/fermion/CayleyFermion5D.cc     |  11 +-
 lib/qcd/action/fermion/CayleyFermion5D.h      |   5 +-
 .../fermion/ContinuedFractionFermion5D.cc     |  81 +++++--
 .../fermion/ContinuedFractionFermion5D.h      |  34 +--
 lib/qcd/action/fermion/DomainWallFermion.h    |   3 +-
 lib/qcd/action/fermion/MobiusFermion.h        |   3 +-
 .../action/fermion/MobiusZolotarevFermion.h   |   4 +-
 tests/InvSqrt.gnu                             |   0
 tests/Make.inc                                |  30 ++-
 tests/Sqrt.gnu                                |   2 -
 tests/{Test_many_cg.cc => Test_cayley_cg.cc}  |   0
 ...any_evenodd.cc => Test_cayley_even_odd.cc} |   0
 tests/Test_contfrac_cg.cc                     | 147 ++++++++++++
 tests/Test_contfrac_even_odd.cc               | 218 ++++++++++++++++++
 ...son_evenodd.cc => Test_wilson_even_odd.cc} |   0
 15 files changed, 490 insertions(+), 48 deletions(-)
 delete mode 100644 tests/InvSqrt.gnu
 delete mode 100644 tests/Sqrt.gnu
 rename tests/{Test_many_cg.cc => Test_cayley_cg.cc} (100%)
 rename tests/{Test_many_evenodd.cc => Test_cayley_even_odd.cc} (100%)
 create mode 100644 tests/Test_contfrac_cg.cc
 create mode 100644 tests/Test_contfrac_even_odd.cc
 rename tests/{Test_wilson_evenodd.cc => Test_wilson_even_odd.cc} (100%)

diff --git a/lib/qcd/action/fermion/CayleyFermion5D.cc b/lib/qcd/action/fermion/CayleyFermion5D.cc
index be528e79..e47ff331 100644
--- a/lib/qcd/action/fermion/CayleyFermion5D.cc
+++ b/lib/qcd/action/fermion/CayleyFermion5D.cc
@@ -229,7 +229,14 @@ namespace QCD {
     }
   }
   
-  void CayleyFermion5D::SetCoefficients(RealD scale,Approx::zolotarev_data *zdata,RealD b,RealD c)
+  // Tanh
+  void CayleyFermion5D::SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD b,RealD c)
+  {
+    SetCoefficientsZolotarev(1.0,zdata,b,c);
+
+  }
+  //Zolo
+  void CayleyFermion5D::SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata,RealD b,RealD c)
   {
 
     ///////////////////////////////////////////////////////////
@@ -266,7 +273,7 @@ namespace QCD {
     double bmc = b-c;
     for(int i=0; i < Ls; i++){
       as[i] = 1.0;
-      omega[i] = ((double)zdata->gamma[i]); //NB reciprocal relative to Chroma NEF code
+      omega[i] = ((double)zdata->gamma[i])*zolo_hi; //NB reciprocal relative to Chroma NEF code
       bs[i] = 0.5*(bpc/omega[i] + bmc);
       cs[i] = 0.5*(bpc/omega[i] - bmc);
     }
diff --git a/lib/qcd/action/fermion/CayleyFermion5D.h b/lib/qcd/action/fermion/CayleyFermion5D.h
index 57c71992..e2175d77 100644
--- a/lib/qcd/action/fermion/CayleyFermion5D.h
+++ b/lib/qcd/action/fermion/CayleyFermion5D.h
@@ -20,7 +20,7 @@ namespace Grid {
       virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out);
       virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out);
       virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
-
+      virtual void   Instantiatable(void)=0;
       //    protected:
       RealD mass;
 
@@ -52,7 +52,8 @@ namespace Grid {
 		      RealD _mass,RealD _M5);
 
     protected:
-      void SetCoefficients(RealD scale,Approx::zolotarev_data *zdata,RealD b,RealD c);
+      void SetCoefficientsZolotarev(RealD zolohi,Approx::zolotarev_data *zdata,RealD b,RealD c);
+      void SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD b,RealD c);
     };
 
   }
diff --git a/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc b/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
index c281b486..250e365f 100644
--- a/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
+++ b/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
@@ -1,9 +1,56 @@
 #include <Grid.h>
 
 namespace Grid {
-
   namespace QCD {
 
+    void ContinuedFractionFermion5D::SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD b,RealD c)
+    {
+      SetCoefficientsZolotarev(1.0,zdata,b,c);
+    }
+    void ContinuedFractionFermion5D::SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata,RealD b,RealD c)
+    {
+      R=(1+this->mass)/(1-this->mass);
+
+      Beta.resize(Ls);
+      cc.resize(Ls);
+      cc_d.resize(Ls);
+      sqrt_cc.resize(Ls);
+      for(int i=0; i < Ls ; i++){
+	Beta[i] = zdata -> beta[i];
+	cc[i] = 1.0/Beta[i];
+	cc_d[i]=sqrt(cc[i]);
+      }
+    
+      cc_d[Ls-1]=1.0;
+      for(int i=0; i < Ls-1 ; i++){
+	sqrt_cc[i]= sqrt(cc[i]*cc[i+1]);
+      }    
+      sqrt_cc[Ls-2]=sqrt(cc[Ls-2]);
+
+
+      ZoloHiInv =1.0/zolo_hi;
+      double dw_diag = (4.0-M5)*ZoloHiInv;
+    
+      See.resize(Ls);
+      Aee.resize(Ls);
+      int sign=1;
+      for(int s=0;s<Ls;s++){
+	Aee[s] = sign * Beta[s] * dw_diag;
+	sign   = - sign;
+      }
+      Aee[Ls-1] += R;
+    
+      See[0] = Aee[0];
+      for(int s=1;s<Ls;s++){
+	See[s] = Aee[s] - 1.0/See[s-1];
+      }
+      for(int s=0;s<Ls;s++){
+	std::cout <<"s = "<<s<<" Beta "<<Beta[s]<<" Aee "<<Aee[s] <<" See "<<See[s] <<std::endl;
+      }
+    }
+
+
+
     RealD  ContinuedFractionFermion5D::M           (const LatticeFermion &psi, LatticeFermion &chi)
     {
       LatticeFermion D(psi._grid);
@@ -13,13 +60,13 @@ namespace Grid {
       int sign=1;
       for(int s=0;s<Ls;s++){
 	if ( s==0 ) {
-	  ag5xpby_ssp(chi,cc[0]*Beta[0]*sign*scale,D,sqrt_cc[0],psi,s,s+1); // Multiplies Dw by G5 so Hw
+	  ag5xpby_ssp(chi,cc[0]*Beta[0]*sign*ZoloHiInv,D,sqrt_cc[0],psi,s,s+1); // Multiplies Dw by G5 so Hw
 	} else if ( s==(Ls-1) ){
 	  RealD R=(1.0+mass)/(1.0-mass);
-	  ag5xpby_ssp(chi,Beta[s]*scale,D,sqrt_cc[s-1],psi,s,s-1);
+	  ag5xpby_ssp(chi,Beta[s]*ZoloHiInv,D,sqrt_cc[s-1],psi,s,s-1);
 	  ag5xpby_ssp(chi,R,psi,1.0,chi,s,s);
 	} else {
-	  ag5xpby_ssp(chi,cc[s]*Beta[s]*sign*scale,D,sqrt_cc[s],psi,s,s+1);
+	  ag5xpby_ssp(chi,cc[s]*Beta[s]*sign*ZoloHiInv,D,sqrt_cc[s],psi,s,s+1);
   	  axpby_ssp(chi,1.0,chi,sqrt_cc[s-1],psi,s,s-1);
 	}
 	sign=-sign; 
@@ -35,18 +82,22 @@ namespace Grid {
     }
     void   ContinuedFractionFermion5D::Meooe       (const LatticeFermion &psi, LatticeFermion &chi)
     {
-      Dhop(psi,chi,DaggerNo); // Dslash on diagonal. g5 Dslash is hermitian
+      // Apply 4d dslash
+      if ( psi.checkerboard == Odd ) {
+	DhopEO(psi,chi,DaggerNo); // Dslash on diagonal. g5 Dslash is hermitian
+      } else {
+	DhopOE(psi,chi,DaggerNo); // Dslash on diagonal. g5 Dslash is hermitian
+      }
       
       int sign=1;
       for(int s=0;s<Ls;s++){
 	if ( s==(Ls-1) ){
-	  ag5xpby_ssp(chi,Beta[s]*scale,chi,0.0,chi,s,s);
+	  ag5xpby_ssp(chi,Beta[s]*ZoloHiInv,chi,0.0,chi,s,s);
 	} else {
-	  ag5xpby_ssp(chi,cc[s]*Beta[s]*sign*scale,chi,0.0,chi,s,s);
+	  ag5xpby_ssp(chi,cc[s]*Beta[s]*sign*ZoloHiInv,chi,0.0,chi,s,s);
 	}
 	sign=-sign; 
-    }
-
+      }
     }
     void   ContinuedFractionFermion5D::MeooeDag    (const LatticeFermion &psi, LatticeFermion &chi)
     {
@@ -54,7 +105,7 @@ namespace Grid {
     }
     void   ContinuedFractionFermion5D::Mooee       (const LatticeFermion &psi, LatticeFermion &chi)
     {
-      double dw_diag = (4.0-this->M5)*scale;
+      double dw_diag = (4.0-M5)*ZoloHiInv;
     
       int sign=1;
       for(int s=0;s<Ls;s++){
@@ -62,7 +113,7 @@ namespace Grid {
 	  ag5xpby_ssp(chi,cc[0]*Beta[0]*sign*dw_diag,psi,sqrt_cc[0],psi,s,s+1); // Multiplies Dw by G5 so Hw
 	} else if ( s==(Ls-1) ){
 	  // Drop the CC here.
-	  double R=(1+this->mass)/(1-this->mass);
+	  double R=(1+mass)/(1-mass);
 	  ag5xpby_ssp(chi,Beta[s]*dw_diag,psi,sqrt_cc[s-1],psi,s,s-1);
 	  ag5xpby_ssp(chi,R,psi,1.0,chi,s,s);
 	} else {
@@ -80,7 +131,7 @@ namespace Grid {
     void   ContinuedFractionFermion5D::MooeeInv    (const LatticeFermion &psi, LatticeFermion &chi)
     {
       // Apply Linv
-      axpby_ssp(chi,1.0/cc_d[0],psi,0.0,psi,0,0);
+      axpby_ssp(chi,1.0/cc_d[0],psi,0.0,psi,0,0); 
       for(int s=1;s<Ls;s++){
 	axpbg5y_ssp(chi,1.0/cc_d[s],psi,-1.0/See[s-1],chi,s,s-1);
       }
@@ -89,7 +140,7 @@ namespace Grid {
 	ag5xpby_ssp(chi,1.0/See[s],chi,0.0,chi,s,s); //only appearance of See[0]
       }
       // Apply Uinv = (Linv)^T
-      axpby_ssp(chi,1.0/cc_d[Ls-1],chi,0.0,chi,this->Ls-1,this->Ls-1);
+      axpby_ssp(chi,1.0/cc_d[Ls-1],chi,0.0,chi,Ls-1,Ls-1);
       for(int s=Ls-2;s>=0;s--){
 	axpbg5y_ssp(chi,1.0/cc_d[s],chi,-1.0*cc_d[s+1]/See[s]/cc_d[s],chi,s,s+1);
       }
@@ -112,6 +163,10 @@ namespace Grid {
 		      FourDimGrid, FourDimRedBlackGrid,M5),
       mass(_mass)
     {
+      assert((Ls&0x1)==1); // Odd Ls required
+      int nrational=Ls-1;// Even rational order
+      zdata = Approx::grid_higham(1.0,nrational);// eps is ignored for higham
+      SetCoefficientsTanh(zdata,1.0,0.0);
     }
 
   }
diff --git a/lib/qcd/action/fermion/ContinuedFractionFermion5D.h b/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
index 7f5c022a..99365009 100644
--- a/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
+++ b/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
@@ -21,20 +21,8 @@ namespace Grid {
       virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out);
       virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
 
-    private:
-
-      Approx::zolotarev_data *zdata;
-
-      // Cont frac
-      RealD mass;
-      RealD R;
-      RealD scale;
-      std::vector<double> Beta;
-      std::vector<double> cc;;
-      std::vector<double> cc_d;;
-      std::vector<double> sqrt_cc;
-      std::vector<double> See;
-      std::vector<double> Aee;
+      //      virtual void   Instantiatable(void)=0;
+      virtual void   Instantiatable(void) {};
 
       // Constructors
       ContinuedFractionFermion5D(LatticeGaugeField &_Umu,
@@ -44,6 +32,24 @@ namespace Grid {
 				 GridRedBlackCartesian &FourDimRedBlackGrid,
 				 RealD _mass,RealD M5);
 
+    protected:
+
+      void SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD b,RealD c);
+      void SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata,RealD b,RealD c);
+
+      Approx::zolotarev_data *zdata;
+
+      // Cont frac
+      RealD mass;
+      RealD R;
+      RealD ZoloHiInv;
+      std::vector<double> Beta;
+      std::vector<double> cc;;
+      std::vector<double> cc_d;;
+      std::vector<double> sqrt_cc;
+      std::vector<double> See;
+      std::vector<double> Aee;
+
     };
 
 
diff --git a/lib/qcd/action/fermion/DomainWallFermion.h b/lib/qcd/action/fermion/DomainWallFermion.h
index 3e6a9739..a25c0c3c 100644
--- a/lib/qcd/action/fermion/DomainWallFermion.h
+++ b/lib/qcd/action/fermion/DomainWallFermion.h
@@ -11,6 +11,7 @@ namespace Grid {
     {
     public:
 
+      virtual void   Instantiatable(void) {};
       // Constructors
       DomainWallFermion(LatticeGaugeField &_Umu,
 			GridCartesian         &FiveDimGrid,
@@ -33,7 +34,7 @@ namespace Grid {
 	
 	std::cout << "DomainWallFermion with Ls="<<Ls<<std::endl;
 	// Call base setter
-	this->CayleyFermion5D::SetCoefficients(1.0,zdata,1.0,0.0);
+	this->CayleyFermion5D::SetCoefficientsTanh(zdata,1.0,0.0);
  
       }
 
diff --git a/lib/qcd/action/fermion/MobiusFermion.h b/lib/qcd/action/fermion/MobiusFermion.h
index 4c291fad..33f94089 100644
--- a/lib/qcd/action/fermion/MobiusFermion.h
+++ b/lib/qcd/action/fermion/MobiusFermion.h
@@ -11,6 +11,7 @@ namespace Grid {
     {
     public:
 
+      virtual void   Instantiatable(void) {};
       // Constructors
       MobiusFermion(LatticeGaugeField &_Umu,
 		    GridCartesian         &FiveDimGrid,
@@ -34,7 +35,7 @@ namespace Grid {
 	assert(zdata->n==this->Ls);
 	
 	// Call base setter
-	this->CayleyFermion5D::SetCoefficients(1.0,zdata,b,c);
+	this->CayleyFermion5D::SetCoefficientsTanh(zdata,b,c);
  
       }
 
diff --git a/lib/qcd/action/fermion/MobiusZolotarevFermion.h b/lib/qcd/action/fermion/MobiusZolotarevFermion.h
index 9ac795d9..1be61601 100644
--- a/lib/qcd/action/fermion/MobiusZolotarevFermion.h
+++ b/lib/qcd/action/fermion/MobiusZolotarevFermion.h
@@ -11,6 +11,7 @@ namespace Grid {
     {
     public:
 
+      virtual void   Instantiatable(void) {};
       // Constructors
        MobiusZolotarevFermion(LatticeGaugeField &_Umu,
 			      GridCartesian         &FiveDimGrid,
@@ -34,10 +35,9 @@ namespace Grid {
 	assert(zdata->n==this->Ls);
 
 	std::cout << "MobiusZolotarevFermion (b="<<b<<",c="<<c<<") with Ls= "<<Ls<<" Zolotarev range ["<<lo<<","<<hi<<"]"<<std::endl;
-	std::cout << "MobiusZolotarevFermion : note there is a degeneracy between (b+c) and Zolo param hi"<<std::endl;
 	
 	// Call base setter
-	this->CayleyFermion5D::SetCoefficients(1.0,zdata,b,c);
+	this->CayleyFermion5D::SetCoefficientsZolotarev(hi,zdata,b,c);
  
       }
 
diff --git a/tests/InvSqrt.gnu b/tests/InvSqrt.gnu
deleted file mode 100644
index e69de29b..00000000
diff --git a/tests/Make.inc b/tests/Make.inc
index d592f218..b525874d 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,21 @@
 
-bin_PROGRAMS = Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_even_odd Test_gamma Test_main Test_many_cg Test_many_evenodd Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_evenodd
+bin_PROGRAMS = Test_cayley_cg Test_cayley_even_odd Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_even_odd Test_gamma Test_main Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_even_odd
+
+
+Test_cayley_cg_SOURCES=Test_cayley_cg.cc
+Test_cayley_cg_LDADD=-lGrid
+
+
+Test_cayley_even_odd_SOURCES=Test_cayley_even_odd.cc
+Test_cayley_even_odd_LDADD=-lGrid
+
+
+Test_contfrac_cg_SOURCES=Test_contfrac_cg.cc
+Test_contfrac_cg_LDADD=-lGrid
+
+
+Test_contfrac_even_odd_SOURCES=Test_contfrac_even_odd.cc
+Test_contfrac_even_odd_LDADD=-lGrid
 
 
 Test_cshift_SOURCES=Test_cshift.cc
@@ -34,14 +50,6 @@ Test_main_SOURCES=Test_main.cc
 Test_main_LDADD=-lGrid
 
 
-Test_many_cg_SOURCES=Test_many_cg.cc
-Test_many_cg_LDADD=-lGrid
-
-
-Test_many_evenodd_SOURCES=Test_many_evenodd.cc
-Test_many_evenodd_LDADD=-lGrid
-
-
 Test_nersc_io_SOURCES=Test_nersc_io.cc
 Test_nersc_io_LDADD=-lGrid
 
@@ -78,6 +86,6 @@ Test_wilson_cg_unprec_SOURCES=Test_wilson_cg_unprec.cc
 Test_wilson_cg_unprec_LDADD=-lGrid
 
 
-Test_wilson_evenodd_SOURCES=Test_wilson_evenodd.cc
-Test_wilson_evenodd_LDADD=-lGrid
+Test_wilson_even_odd_SOURCES=Test_wilson_even_odd.cc
+Test_wilson_even_odd_LDADD=-lGrid
 
diff --git a/tests/Sqrt.gnu b/tests/Sqrt.gnu
deleted file mode 100644
index ae56ab97..00000000
--- a/tests/Sqrt.gnu
+++ /dev/null
@@ -1,2 +0,0 @@
-f(x) = 6.81384+(-2.34645e-06/(x+0.000228091))+(-1.51593e-05/(x+0.00112084))+(-6.89254e-05/(x+0.003496))+(-0.000288983/(x+0.00954309))+(-0.00119277/(x+0.024928))+(-0.0050183/(x+0.0646627))+(-0.0226449/(x+0.171576))+(-0.123767/(x+0.491792))+(-1.1705/(x+1.78667))+(-102.992/(x+18.4866));
-f(x) = 0.14676+(0.00952992/(x+5.40933e-05))+(0.0115952/(x+0.000559699))+(0.0161824/(x+0.00203338))+(0.0243252/(x+0.00582831))+(0.0379533/(x+0.0154649))+(0.060699/(x+0.0401156))+(0.100345/(x+0.104788))+(0.178335/(x+0.286042))+(0.381586/(x+0.892189))+(1.42625/(x+4.38422));
diff --git a/tests/Test_many_cg.cc b/tests/Test_cayley_cg.cc
similarity index 100%
rename from tests/Test_many_cg.cc
rename to tests/Test_cayley_cg.cc
diff --git a/tests/Test_many_evenodd.cc b/tests/Test_cayley_even_odd.cc
similarity index 100%
rename from tests/Test_many_evenodd.cc
rename to tests/Test_cayley_even_odd.cc
diff --git a/tests/Test_contfrac_cg.cc b/tests/Test_contfrac_cg.cc
new file mode 100644
index 00000000..7fa0d6fc
--- /dev/null
+++ b/tests/Test_contfrac_cg.cc
@@ -0,0 +1,147 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+
+template<class What> 
+void  TestCGinversions(What & Ddwf, 
+		       GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+		       GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+		       RealD mass, RealD M5,
+		       GridParallelRNG *RNG4,
+		       GridParallelRNG *RNG5);
+template<class What> 
+void  TestCGschur(What & Ddwf, 
+		  GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+		  GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+		  RealD mass, RealD M5,
+		  GridParallelRNG *RNG4,
+		  GridParallelRNG *RNG5);
+
+template<class What> 
+void  TestCGunprec(What & Ddwf, 
+		   GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+		   GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+		   RealD mass, RealD M5,
+		   GridParallelRNG *RNG4,
+		   GridParallelRNG *RNG5);
+
+template<class What> 
+void  TestCGprec(What & Ddwf, 
+		 GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+		 GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+		 RealD mass, RealD M5,
+		 GridParallelRNG *RNG4,
+		 GridParallelRNG *RNG5);
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  int threads = GridThread::GetThreads();
+  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+
+  const int Ls=9;
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);  RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);  RNG4.SeedFixedIntegers(seeds4);
+
+  LatticeGaugeField Umu(UGrid); random(RNG4,Umu);
+  std::vector<LatticeColourMatrix> U(4,UGrid);
+
+  RealD mass=0.1;
+  RealD M5  =1.8;
+  std::cout <<"ContinuedFractionFermion test"<<std::endl;
+  ContinuedFractionFermion5D Dcf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  TestCGinversions<ContinuedFractionFermion5D>(Dcf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  Grid_finalize();
+}
+template<class What> 
+void  TestCGinversions(What & Ddwf, 
+		       GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+		       GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+		       RealD mass, RealD M5,
+		       GridParallelRNG *RNG4,
+		       GridParallelRNG *RNG5)
+{
+  std::cout << "Testing unpreconditioned inverter"<<std::endl;
+  TestCGunprec<What>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,RNG4,RNG5);
+  std::cout << "Testing red black preconditioned inverter"<<std::endl;
+  TestCGprec<What>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,RNG4,RNG5);
+  std::cout << "Testing red black Schur inverter"<<std::endl;
+  TestCGschur<What>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,RNG4,RNG5);
+}
+
+template<class What> 
+void  TestCGunprec(What & Ddwf, 
+		   GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+		   GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+		   RealD mass, RealD M5,
+		   GridParallelRNG *RNG4,
+		   GridParallelRNG *RNG5)
+{
+  LatticeFermion src   (FGrid); random(*RNG5,src);
+  LatticeFermion result(FGrid); result=zero;
+
+  HermitianOperator<What,LatticeFermion> HermOp(Ddwf);
+  ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
+  CG(HermOp,src,result);
+
+}
+template<class What> 
+void  TestCGprec(What & Ddwf, 
+		 GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+		 GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+		 RealD mass, RealD M5,
+		 GridParallelRNG *RNG4,
+		 GridParallelRNG *RNG5)
+{
+  LatticeFermion src   (FGrid); random(*RNG5,src);
+  LatticeFermion    src_o(FrbGrid);
+  LatticeFermion result_o(FrbGrid);
+  pickCheckerboard(Odd,src_o,src);
+  result_o=zero;
+
+  HermitianCheckerBoardedOperator<What,LatticeFermion> HermOpEO(Ddwf);
+  ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
+  CG(HermOpEO,src_o,result_o);
+}
+
+
+template<class What> 
+void  TestCGschur(What & Ddwf, 
+		   GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+		   GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+		   RealD mass, RealD M5,
+		   GridParallelRNG *RNG4,
+		   GridParallelRNG *RNG5)
+{
+  LatticeFermion src   (FGrid); random(*RNG5,src);
+  LatticeFermion result(FGrid); result=zero;
+
+  ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
+  SchurRedBlackSolve<LatticeFermion> SchurSolver(CG);
+  SchurSolver(Ddwf,src,result);
+}
diff --git a/tests/Test_contfrac_even_odd.cc b/tests/Test_contfrac_even_odd.cc
new file mode 100644
index 00000000..801bd955
--- /dev/null
+++ b/tests/Test_contfrac_even_odd.cc
@@ -0,0 +1,218 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+
+template<class What> 
+void  TestWhat(What & Ddwf,
+	       GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+	       GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+	       RealD mass, RealD M5,
+	       GridParallelRNG *RNG4,   GridParallelRNG *RNG5);
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  int threads = GridThread::GetThreads();
+  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+
+  const int Ls=9;
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);  RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);  RNG4.SeedFixedIntegers(seeds4);
+
+  LatticeGaugeField Umu(UGrid); random(RNG4,Umu);
+  std::vector<LatticeColourMatrix> U(4,UGrid);
+
+  RealD mass=0.1;
+  RealD M5  =1.8;
+  std::cout <<"ContinuedFractionFermion test"<<std::endl;
+  ContinuedFractionFermion5D Dcf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  TestWhat<ContinuedFractionFermion5D>(Dcf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  Grid_finalize();
+}
+
+template<class What> 
+void  TestWhat(What & Ddwf, 
+	       GridCartesian         * FGrid,	       GridRedBlackCartesian * FrbGrid,
+	       GridCartesian         * UGrid,	       GridRedBlackCartesian * UrbGrid,
+	       RealD mass, RealD M5,
+	       GridParallelRNG *RNG4,
+	       GridParallelRNG *RNG5)
+{
+
+  LatticeFermion src   (FGrid); random(*RNG5,src);
+  LatticeFermion phi   (FGrid); random(*RNG5,phi);
+  LatticeFermion chi   (FGrid); random(*RNG5,chi);
+  LatticeFermion result(FGrid); result=zero;
+  LatticeFermion    ref(FGrid);    ref=zero;
+  LatticeFermion    tmp(FGrid);    tmp=zero;
+  LatticeFermion    err(FGrid);    tmp=zero;
+
+  LatticeFermion src_e (FrbGrid);
+  LatticeFermion src_o (FrbGrid);
+  LatticeFermion r_e   (FrbGrid);
+  LatticeFermion r_o   (FrbGrid);
+  LatticeFermion r_eo  (FGrid);
+  LatticeFermion r_eeoo(FGrid);
+
+  std::cout<<"=========================================================="<<std::endl;
+  std::cout<<"= Testing that Meo + Moe + Moo + Mee = Munprec "<<std::endl;
+  std::cout<<"=========================================================="<<std::endl;
+
+  pickCheckerboard(Even,src_e,src);
+  pickCheckerboard( Odd,src_o,src);
+
+  Ddwf.Meooe(src_e,r_o);  std::cout<<"Applied Meo "<<norm2(r_o)<<std::endl;
+  Ddwf.Meooe(src_o,r_e);  std::cout<<"Applied Moe "<<norm2(r_e)<<std::endl;
+  setCheckerboard(r_eo,r_o);
+  setCheckerboard(r_eo,r_e);
+
+  Ddwf.Mooee(src_e,r_e);  std::cout<<"Applied Mee"<<norm2(r_e)<<std::endl;
+  Ddwf.Mooee(src_o,r_o);  std::cout<<"Applied Moo"<<norm2(r_o)<<std::endl;
+  setCheckerboard(r_eeoo,r_e);
+  setCheckerboard(r_eeoo,r_o);
+
+  r_eo=r_eo+r_eeoo;
+  Ddwf.M(src,ref);  
+
+  //  std::cout << r_eo<<std::endl;
+  //  std::cout << ref <<std::endl;
+
+  err= ref - r_eo;
+  std::cout << "EO norm diff   "<< norm2(err)<< " "<<norm2(ref)<< " " << norm2(r_eo) <<std::endl;
+    
+  LatticeComplex cerr(FGrid);
+  cerr = localInnerProduct(err,err);
+  //  std::cout << cerr<<std::endl;
+
+  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<"= Test Ddagger is the dagger of D by requiring                "<<std::endl;
+  std::cout<<"=  < phi | Deo | chi > * = < chi | Deo^dag| phi>  "<<std::endl;
+  std::cout<<"=============================================================="<<std::endl;
+  
+  LatticeFermion chi_e   (FrbGrid);
+  LatticeFermion chi_o   (FrbGrid);
+
+  LatticeFermion dchi_e  (FrbGrid);
+  LatticeFermion dchi_o  (FrbGrid);
+
+  LatticeFermion phi_e   (FrbGrid);
+  LatticeFermion phi_o   (FrbGrid);
+
+  LatticeFermion dphi_e  (FrbGrid);
+  LatticeFermion dphi_o  (FrbGrid);
+
+
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+  pickCheckerboard(Even,phi_e,phi);
+  pickCheckerboard(Odd ,phi_o,phi);
+
+  Ddwf.Meooe(chi_e,dchi_o);
+  Ddwf.Meooe(chi_o,dchi_e);
+  Ddwf.MeooeDag(phi_e,dphi_o);
+  Ddwf.MeooeDag(phi_o,dphi_e);
+
+  ComplexD pDce = innerProduct(phi_e,dchi_e);
+  ComplexD pDco = innerProduct(phi_o,dchi_o);
+  ComplexD cDpe = innerProduct(chi_e,dphi_e);
+  ComplexD cDpo = innerProduct(chi_o,dphi_o);
+
+  std::cout <<"e "<<pDce<<" "<<cDpe <<std::endl;
+  std::cout <<"o "<<pDco<<" "<<cDpo <<std::endl;
+
+  std::cout <<"pDce - conj(cDpo) "<< pDce-conj(cDpo) <<std::endl;
+  std::cout <<"pDco - conj(cDpe) "<< pDco-conj(cDpe) <<std::endl;
+
+  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<"= Test MeeInv Mee = 1                                         "<<std::endl;
+  std::cout<<"=============================================================="<<std::endl;
+
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+
+  Ddwf.Mooee(chi_e,src_e);
+  Ddwf.MooeeInv(src_e,phi_e);
+
+  Ddwf.Mooee(chi_o,src_o);
+  Ddwf.MooeeInv(src_o,phi_o);
+  
+  setCheckerboard(phi,phi_e);
+  setCheckerboard(phi,phi_o);
+
+  err = phi-chi;
+  std::cout << "norm diff   "<< norm2(err)<< std::endl;
+
+  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<"= Test MeeInvDag MeeDag = 1                                   "<<std::endl;
+  std::cout<<"=============================================================="<<std::endl;
+
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+
+  Ddwf.MooeeDag(chi_e,src_e);
+  Ddwf.MooeeInvDag(src_e,phi_e);
+
+  Ddwf.MooeeDag(chi_o,src_o);
+  Ddwf.MooeeInvDag(src_o,phi_o);
+  
+  setCheckerboard(phi,phi_e);
+  setCheckerboard(phi,phi_o);
+
+  err = phi-chi;
+  std::cout << "norm diff   "<< norm2(err)<< std::endl;
+
+  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<"= Test MpcDagMpc is Hermitian              "<<std::endl;
+  std::cout<<"=============================================================="<<std::endl;
+  
+  random(*RNG5,phi);
+  random(*RNG5,chi);
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+  pickCheckerboard(Even,phi_e,phi);
+  pickCheckerboard(Odd ,phi_o,phi);
+  RealD t1,t2;
+
+  Ddwf.MpcDagMpc(chi_e,dchi_e,t1,t2);
+  Ddwf.MpcDagMpc(chi_o,dchi_o,t1,t2);
+
+  Ddwf.MpcDagMpc(phi_e,dphi_e,t1,t2);
+  Ddwf.MpcDagMpc(phi_o,dphi_o,t1,t2);
+
+  pDce = innerProduct(phi_e,dchi_e);
+  pDco = innerProduct(phi_o,dchi_o);
+  cDpe = innerProduct(chi_e,dphi_e);
+  cDpo = innerProduct(chi_o,dphi_o);
+
+  std::cout <<"e "<<pDce<<" "<<cDpe <<std::endl;
+  std::cout <<"o "<<pDco<<" "<<cDpo <<std::endl;
+
+  std::cout <<"pDce - conj(cDpo) "<< pDco-conj(cDpo) <<std::endl;
+  std::cout <<"pDco - conj(cDpe) "<< pDce-conj(cDpe) <<std::endl;
+  
+}
diff --git a/tests/Test_wilson_evenodd.cc b/tests/Test_wilson_even_odd.cc
similarity index 100%
rename from tests/Test_wilson_evenodd.cc
rename to tests/Test_wilson_even_odd.cc

From c327019574dfa1e5086594a3660ab235016109cc Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 4 Jun 2015 00:23:16 +0100
Subject: [PATCH 267/429] Implementing the Hw kernel continued fraction 5d
 overlap cases

---
 lib/qcd/action/Actions.h                      |  2 +
 .../fermion/ContinuedFractionFermion5D.cc     |  9 ++--
 .../fermion/ContinuedFractionFermion5D.h      |  6 +--
 .../OverlapWilsonContfracTanhFermion.h        | 39 ++++++++++++++++
 .../OverlapWilsonContfracZolotarevFermion.h   | 44 +++++++++++++++++++
 5 files changed, 91 insertions(+), 9 deletions(-)
 create mode 100644 lib/qcd/action/fermion/OverlapWilsonContfracTanhFermion.h
 create mode 100644 lib/qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h

diff --git a/lib/qcd/action/Actions.h b/lib/qcd/action/Actions.h
index 8a8c4642..b31e9136 100644
--- a/lib/qcd/action/Actions.h
+++ b/lib/qcd/action/Actions.h
@@ -55,6 +55,8 @@
 // Continued fraction
 //////////////////////
 #include <qcd/action/fermion/ContinuedFractionFermion5D.h>
+#include <qcd/action/fermion/OverlapWilsonContfracTanhFermion.h>
+#include <qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h>
 
 //////////////////////
 // Partial fraction
diff --git a/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc b/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
index 250e365f..92f6473e 100644
--- a/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
+++ b/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
@@ -3,11 +3,11 @@
 namespace Grid {
   namespace QCD {
 
-    void ContinuedFractionFermion5D::SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD b,RealD c)
+    void ContinuedFractionFermion5D::SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD scale)
     {
-      SetCoefficientsZolotarev(1.0,zdata,b,c);
+      SetCoefficientsZolotarev(1.0/scale,zdata);
     }
-    void ContinuedFractionFermion5D::SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata,RealD b,RealD c)
+    void ContinuedFractionFermion5D::SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata)
     {
       R=(1+this->mass)/(1-this->mass);
 
@@ -164,9 +164,6 @@ namespace Grid {
       mass(_mass)
     {
       assert((Ls&0x1)==1); // Odd Ls required
-      int nrational=Ls-1;// Even rational order
-      zdata = Approx::grid_higham(1.0,nrational);// eps is ignored for higham
-      SetCoefficientsTanh(zdata,1.0,0.0);
     }
 
   }
diff --git a/lib/qcd/action/fermion/ContinuedFractionFermion5D.h b/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
index 99365009..f363878f 100644
--- a/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
+++ b/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
@@ -22,7 +22,7 @@ namespace Grid {
       virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
 
       //      virtual void   Instantiatable(void)=0;
-      virtual void   Instantiatable(void) {};
+      virtual void   Instantiatable(void) =0;
 
       // Constructors
       ContinuedFractionFermion5D(LatticeGaugeField &_Umu,
@@ -34,8 +34,8 @@ namespace Grid {
 
     protected:
 
-      void SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD b,RealD c);
-      void SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata,RealD b,RealD c);
+      void SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD scale);
+      void SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata);;
 
       Approx::zolotarev_data *zdata;
 
diff --git a/lib/qcd/action/fermion/OverlapWilsonContfracTanhFermion.h b/lib/qcd/action/fermion/OverlapWilsonContfracTanhFermion.h
new file mode 100644
index 00000000..4865f169
--- /dev/null
+++ b/lib/qcd/action/fermion/OverlapWilsonContfracTanhFermion.h
@@ -0,0 +1,39 @@
+#ifndef OVERLAP_WILSON_CAYLEY_TANH_FERMION_H
+#define OVERLAP_WILSON_CAYLEY_TANH_FERMION_H
+
+#include <Grid.h>
+
+namespace Grid {
+
+  namespace QCD {
+
+    class OverlapWilsonContFracTanhFermion : public ContinuedFractionFermion5D
+    {
+    public:
+
+      virtual void   Instantiatable(void){};
+      // Constructors
+    OverlapWilsonContFracTanhFermion(LatticeGaugeField &_Umu,
+				     GridCartesian         &FiveDimGrid,
+				     GridRedBlackCartesian &FiveDimRedBlackGrid,
+				     GridCartesian         &FourDimGrid,
+				     GridRedBlackCartesian &FourDimRedBlackGrid,
+				     RealD _mass,RealD _M5,
+				     RealD scale) :
+      
+      // b+c=scale, b-c = 0 <=> b =c = scale/2
+      ContinuedFractionFermion5D(_Umu,
+		    FiveDimGrid,
+		    FiveDimRedBlackGrid,
+		    FourDimGrid,
+		    FourDimRedBlackGrid,_mass)
+	{
+	  assert((Ls&0x1)==1); // Odd Ls required
+	  int nrational=Ls-1;// Even rational order
+	  zdata = Approx::grid_higham(1.0,nrational);// eps is ignored for higham
+	  SetCoefficientsTanh(zdata,scale);
+	}
+    };
+  }
+}
+#endif
diff --git a/lib/qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h b/lib/qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h
new file mode 100644
index 00000000..7478c062
--- /dev/null
+++ b/lib/qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h
@@ -0,0 +1,44 @@
+#ifndef OVERLAP_WILSON_CAYLEY_TANH_FERMION_H
+#define OVERLAP_WILSON_CAYLEY_TANH_FERMION_H
+
+#include <Grid.h>
+
+namespace Grid {
+
+  namespace QCD {
+
+    class OverlapWilsonContFracZolotarevFermion : public ContinuedFractionFermion5D
+    {
+    public:
+
+      virtual void   Instantiatable(void){};
+      // Constructors
+    OverlapWilsonContFracZolotarevFermion(LatticeGaugeField &_Umu,
+					  GridCartesian         &FiveDimGrid,
+					  GridRedBlackCartesian &FiveDimRedBlackGrid,
+					  GridCartesian         &FourDimGrid,
+					  GridRedBlackCartesian &FourDimRedBlackGrid,
+					  RealD _mass,RealD _M5,
+					  RealD lo,RealD hi):
+      
+      // b+c=scale, b-c = 0 <=> b =c = scale/2
+      ContinuedFractionFermion5D(_Umu,
+				 FiveDimGrid,
+				 FiveDimRedBlackGrid,
+				 FourDimGrid,
+				 FourDimRedBlackGrid,_mass)
+	{
+	  assert((Ls&0x1)==1); // Odd Ls required
+
+	  int nrational=Ls-1;// Even rational order
+	  RealD eps = lo/hi;
+
+	  Approx::zolotarev_data *zdata = Approx::grid_zolotarev(eps,nrational,0);
+
+	  SetCoefficientsZolotarev(hi,zdata);
+
+	}
+    };
+  }
+}
+#endif

From 37aa74dfd2c6634a017925e5dbccfd3ecfbefc6f Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 4 Jun 2015 06:02:00 +0100
Subject: [PATCH 268/429] CG Tests work for wilson kernel cont frac zolo and
 tanh

---
 .../fermion/OverlapWilsonContfracTanhFermion.h       | 12 ++++++------
 .../fermion/OverlapWilsonContfracZolotarevFermion.h  |  6 +++---
 tests/Test_contfrac_cg.cc                            | 12 +++++++++---
 tests/Test_contfrac_even_odd.cc                      | 11 ++++++++---
 4 files changed, 26 insertions(+), 15 deletions(-)

diff --git a/lib/qcd/action/fermion/OverlapWilsonContfracTanhFermion.h b/lib/qcd/action/fermion/OverlapWilsonContfracTanhFermion.h
index 4865f169..ed0c24dc 100644
--- a/lib/qcd/action/fermion/OverlapWilsonContfracTanhFermion.h
+++ b/lib/qcd/action/fermion/OverlapWilsonContfracTanhFermion.h
@@ -1,5 +1,5 @@
-#ifndef OVERLAP_WILSON_CAYLEY_TANH_FERMION_H
-#define OVERLAP_WILSON_CAYLEY_TANH_FERMION_H
+#ifndef OVERLAP_WILSON_CONTFRAC_TANH_FERMION_H
+#define OVERLAP_WILSON_CONTFRAC_TANH_FERMION_H
 
 #include <Grid.h>
 
@@ -23,10 +23,10 @@ namespace Grid {
       
       // b+c=scale, b-c = 0 <=> b =c = scale/2
       ContinuedFractionFermion5D(_Umu,
-		    FiveDimGrid,
-		    FiveDimRedBlackGrid,
-		    FourDimGrid,
-		    FourDimRedBlackGrid,_mass)
+				 FiveDimGrid,
+				 FiveDimRedBlackGrid,
+				 FourDimGrid,
+				 FourDimRedBlackGrid,_mass,_M5)
 	{
 	  assert((Ls&0x1)==1); // Odd Ls required
 	  int nrational=Ls-1;// Even rational order
diff --git a/lib/qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h b/lib/qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h
index 7478c062..caf01133 100644
--- a/lib/qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h
+++ b/lib/qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h
@@ -1,5 +1,5 @@
-#ifndef OVERLAP_WILSON_CAYLEY_TANH_FERMION_H
-#define OVERLAP_WILSON_CAYLEY_TANH_FERMION_H
+#ifndef OVERLAP_WILSON_CONTFRAC_ZOLOTAREV_FERMION_H
+#define OVERLAP_WILSON_CONTFRAC_ZOLOTAREV_FERMION_H
 
 #include <Grid.h>
 
@@ -26,7 +26,7 @@ namespace Grid {
 				 FiveDimGrid,
 				 FiveDimRedBlackGrid,
 				 FourDimGrid,
-				 FourDimRedBlackGrid,_mass)
+				 FourDimRedBlackGrid,_mass,_M5)
 	{
 	  assert((Ls&0x1)==1); // Odd Ls required
 
diff --git a/tests/Test_contfrac_cg.cc b/tests/Test_contfrac_cg.cc
index 7fa0d6fc..83475254 100644
--- a/tests/Test_contfrac_cg.cc
+++ b/tests/Test_contfrac_cg.cc
@@ -72,9 +72,15 @@ int main (int argc, char ** argv)
 
   RealD mass=0.1;
   RealD M5  =1.8;
-  std::cout <<"ContinuedFractionFermion test"<<std::endl;
-  ContinuedFractionFermion5D Dcf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
-  TestCGinversions<ContinuedFractionFermion5D>(Dcf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+
+  std::cout <<"OverlapWilsonContFracTanhFermion  test"<<std::endl;
+  OverlapWilsonContFracTanhFermion Dcf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+  TestCGinversions<OverlapWilsonContFracTanhFermion>(Dcf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  std::cout <<"OverlapWilsonContFracZolotarevFermion  test"<<std::endl;
+  OverlapWilsonContFracZolotarevFermion Dcfz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,6.0);
+  TestCGinversions<OverlapWilsonContFracZolotarevFermion>(Dcfz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   Grid_finalize();
 }
diff --git a/tests/Test_contfrac_even_odd.cc b/tests/Test_contfrac_even_odd.cc
index 801bd955..e13c1189 100644
--- a/tests/Test_contfrac_even_odd.cc
+++ b/tests/Test_contfrac_even_odd.cc
@@ -48,9 +48,14 @@ int main (int argc, char ** argv)
 
   RealD mass=0.1;
   RealD M5  =1.8;
-  std::cout <<"ContinuedFractionFermion test"<<std::endl;
-  ContinuedFractionFermion5D Dcf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
-  TestWhat<ContinuedFractionFermion5D>(Dcf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  std::cout <<"OverlapWilsonContFracTanhFermion  test"<<std::endl;
+  OverlapWilsonContFracTanhFermion Dcf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+  TestWhat<OverlapWilsonContFracTanhFermion>(Dcf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  std::cout <<"OverlapWilsonContFracZolotarevFermion  test"<<std::endl;
+  OverlapWilsonContFracZolotarevFermion Dcfz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,6.0);
+  TestWhat<OverlapWilsonContFracZolotarevFermion>(Dcfz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   Grid_finalize();
 }

From ff9340d4d534595274e164b00be7a7f215619d06 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Thu, 4 Jun 2015 14:17:58 +0900
Subject: [PATCH 269/429] Small modification to the configure files

---
 INSTALL                      |    2 +-
 TODO                         |    3 +
 configure                    |   13 +-
 docs/doxy.cfg.test           | 2305 ----------------------------------
 lib/simd/Grid_avx.h          |   10 +-
 lib/simd/Grid_vector_types.h |    3 +-
 6 files changed, 13 insertions(+), 2323 deletions(-)
 delete mode 100644 docs/doxy.cfg.test

diff --git a/INSTALL b/INSTALL
index 80a61507..f812f5a2 120000
--- a/INSTALL
+++ b/INSTALL
@@ -1 +1 @@
-/opt/local/share/automake-1.15/INSTALL
\ No newline at end of file
+/usr/share/automake-1.14/INSTALL
\ No newline at end of file
diff --git a/TODO b/TODO
index 948fb99f..9a428f98 100644
--- a/TODO
+++ b/TODO
@@ -66,6 +66,9 @@ Insert/Extract
 
 * Support for ILDG
 
+* Support different boundary conditions (finite temp, chem. potential ... )
+
+* Support different fermion representations? 
 
 Actions -- coherent framework for implementing actions and their forces.
 
diff --git a/configure b/configure
index 9713ae55..2d785792 100755
--- a/configure
+++ b/configure
@@ -2574,7 +2574,7 @@ test -n "$target_alias" &&
     NONENONEs,x,x, &&
   program_prefix=${target_alias}-
 
-am__api_version='1.15'
+am__api_version='1.14'
 
 # Find a good install program.  We prefer a C program (faster),
 # so one script is as good as another.  But avoid the broken or
@@ -2746,8 +2746,8 @@ test "$program_suffix" != NONE &&
 ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
 program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
 
-# Expand $ac_aux_dir to an absolute path.
-am_aux_dir=`cd "$ac_aux_dir" && pwd`
+# expand $ac_aux_dir to an absolute path
+am_aux_dir=`cd $ac_aux_dir && pwd`
 
 if test x"${MISSING+set}" != xset; then
   case $am_aux_dir in
@@ -2766,7 +2766,7 @@ else
 $as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
 fi
 
-if test x"${install_sh+set}" != xset; then
+if test x"${install_sh}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -3094,8 +3094,8 @@ MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 mkdir_p='$(MKDIR_P)'
 
-# We need awk for the "check" target (and possibly the TAP driver).  The
-# system "awk" is bad on some platforms.
+# We need awk for the "check" target.  The system "awk" is bad on
+# some platforms.
 # Always define AMTAR for backward compatibility.  Yes, it's still used
 # in the wild :-(  We should find a proper way to deprecate it ...
 AMTAR='$${TAR-tar}'
@@ -3154,7 +3154,6 @@ END
 fi
 
 
-
 ac_config_headers="$ac_config_headers lib/Grid_config.h"
 
 # Check whether --enable-silent-rules was given.
diff --git a/docs/doxy.cfg.test b/docs/doxy.cfg.test
deleted file mode 100644
index c4b82094..00000000
--- a/docs/doxy.cfg.test
+++ /dev/null
@@ -1,2305 +0,0 @@
-# Doxyfile 1.8.6
-
-# This file describes the settings to be used by the documentation system
-# doxygen (www.doxygen.org) for a project.
-#
-# All text after a double hash (##) is considered a comment and is placed in
-# front of the TAG it is preceding.
-#
-# All text after a single hash (#) is considered a comment and will be ignored.
-# The format is:
-# TAG = value [value, ...]
-# For lists, items can also be appended using:
-# TAG += value [value, ...]
-# Values that contain spaces should be placed between quotes (\" \").
-
-#---------------------------------------------------------------------------
-# Project related configuration options
-#---------------------------------------------------------------------------
-
-# This tag specifies the encoding used for all characters in the config file
-# that follow. The default is UTF-8 which is also the encoding used for all text
-# before the first occurrence of this tag. Doxygen uses libiconv (or the iconv
-# built into libc) for the transcoding. See http://www.gnu.org/software/libiconv
-# for the list of possible encodings.
-# The default value is: UTF-8.
-
-DOXYFILE_ENCODING      = UTF-8
-
-# The PROJECT_NAME tag is a single word (or a sequence of words surrounded by
-# double-quotes, unless you are using Doxywizard) that should identify the
-# project for which the documentation is generated. This name is used in the
-# title of most generated pages and in a few other places.
-# The default value is: My Project.
-
-PROJECT_NAME           = "Grid"
-
-# The PROJECT_NUMBER tag can be used to enter a project or revision number. This
-# could be handy for archiving the generated documentation or if some version
-# control system is used.
-
-PROJECT_NUMBER         = 1.0
-
-# Using the PROJECT_BRIEF tag one can provide an optional one line description
-# for a project that appears at the top of each page and should give viewer a
-# quick idea about the purpose of the project. Keep the description short.
-
-PROJECT_BRIEF          =
-
-# With the PROJECT_LOGO tag one can specify an logo or icon that is included in
-# the documentation. The maximum height of the logo should not exceed 55 pixels
-# and the maximum width should not exceed 200 pixels. Doxygen will copy the logo
-# to the output directory.
-
-PROJECT_LOGO           =
-
-# The OUTPUT_DIRECTORY tag is used to specify the (relative or absolute) path
-# into which the generated documentation will be written. If a relative path is
-# entered, it will be relative to the location where doxygen was started. If
-# left blank the current directory will be used.
-
-OUTPUT_DIRECTORY       = ./doxy-en/
-
-# If the CREATE_SUBDIRS tag is set to YES, then doxygen will create 4096 sub-
-# directories (in 2 levels) under the output directory of each output format and
-# will distribute the generated files over these directories. Enabling this
-# option can be useful when feeding doxygen a huge amount of source files, where
-# putting all generated files in the same directory would otherwise causes
-# performance problems for the file system.
-# The default value is: NO.
-
-CREATE_SUBDIRS         = NO
-
-# The OUTPUT_LANGUAGE tag is used to specify the language in which all
-# documentation generated by doxygen is written. Doxygen will use this
-# information to generate all constant output in the proper language.
-# Possible values are: Afrikaans, Arabic, Armenian, Brazilian, Catalan, Chinese,
-# Chinese-Traditional, Croatian, Czech, Danish, Dutch, English (United States),
-# Esperanto, Farsi (Persian), Finnish, French, German, Greek, Hungarian,
-# Indonesian, Italian, Japanese, Japanese-en (Japanese with English messages),
-# Korean, Korean-en (Korean with English messages), Latvian, Lithuanian,
-# Macedonian, Norwegian, Persian (Farsi), Polish, Portuguese, Romanian, Russian,
-# Serbian, Serbian-Cyrillic, Slovak, Slovene, Spanish, Swedish, Turkish,
-# Ukrainian and Vietnamese.
-# The default value is: English.
-
-OUTPUT_LANGUAGE        = English
-
-# If the BRIEF_MEMBER_DESC tag is set to YES doxygen will include brief member
-# descriptions after the members that are listed in the file and class
-# documentation (similar to Javadoc). Set to NO to disable this.
-# The default value is: YES.
-
-BRIEF_MEMBER_DESC      = YES
-
-# If the REPEAT_BRIEF tag is set to YES doxygen will prepend the brief
-# description of a member or function before the detailed description
-#
-# Note: If both HIDE_UNDOC_MEMBERS and BRIEF_MEMBER_DESC are set to NO, the
-# brief descriptions will be completely suppressed.
-# The default value is: YES.
-
-REPEAT_BRIEF           = YES
-
-# This tag implements a quasi-intelligent brief description abbreviator that is
-# used to form the text in various listings. Each string in this list, if found
-# as the leading text of the brief description, will be stripped from the text
-# and the result, after processing the whole list, is used as the annotated
-# text. Otherwise, the brief description is used as-is. If left blank, the
-# following values are used ($name is automatically replaced with the name of
-# the entity):The $name class, The $name widget, The $name file, is, provides,
-# specifies, contains, represents, a, an and the.
-
-ABBREVIATE_BRIEF       =
-
-# If the ALWAYS_DETAILED_SEC and REPEAT_BRIEF tags are both set to YES then
-# doxygen will generate a detailed section even if there is only a brief
-# description.
-# The default value is: NO.
-
-ALWAYS_DETAILED_SEC    = NO
-
-# If the INLINE_INHERITED_MEMB tag is set to YES, doxygen will show all
-# inherited members of a class in the documentation of that class as if those
-# members were ordinary class members. Constructors, destructors and assignment
-# operators of the base classes will not be shown.
-# The default value is: NO.
-
-INLINE_INHERITED_MEMB  = NO
-
-# If the FULL_PATH_NAMES tag is set to YES doxygen will prepend the full path
-# before files name in the file list and in the header files. If set to NO the
-# shortest path that makes the file name unique will be used
-# The default value is: YES.
-
-FULL_PATH_NAMES        = YES
-
-# The STRIP_FROM_PATH tag can be used to strip a user-defined part of the path.
-# Stripping is only done if one of the specified strings matches the left-hand
-# part of the path. The tag can be used to show relative paths in the file list.
-# If left blank the directory from which doxygen is run is used as the path to
-# strip.
-#
-# Note that you can specify absolute paths here, but also relative paths, which
-# will be relative from the directory where doxygen is started.
-# This tag requires that the tag FULL_PATH_NAMES is set to YES.
-
-STRIP_FROM_PATH        =
-
-# The STRIP_FROM_INC_PATH tag can be used to strip a user-defined part of the
-# path mentioned in the documentation of a class, which tells the reader which
-# header file to include in order to use a class. If left blank only the name of
-# the header file containing the class definition is used. Otherwise one should
-# specify the list of include paths that are normally passed to the compiler
-# using the -I flag.
-
-STRIP_FROM_INC_PATH    =
-
-# If the SHORT_NAMES tag is set to YES, doxygen will generate much shorter (but
-# less readable) file names. This can be useful is your file systems doesn't
-# support long names like on DOS, Mac, or CD-ROM.
-# The default value is: NO.
-
-SHORT_NAMES            = NO
-
-# If the JAVADOC_AUTOBRIEF tag is set to YES then doxygen will interpret the
-# first line (until the first dot) of a Javadoc-style comment as the brief
-# description. If set to NO, the Javadoc-style will behave just like regular Qt-
-# style comments (thus requiring an explicit @brief command for a brief
-# description.)
-# The default value is: NO.
-
-JAVADOC_AUTOBRIEF      = NO
-
-# If the QT_AUTOBRIEF tag is set to YES then doxygen will interpret the first
-# line (until the first dot) of a Qt-style comment as the brief description. If
-# set to NO, the Qt-style will behave just like regular Qt-style comments (thus
-# requiring an explicit \brief command for a brief description.)
-# The default value is: NO.
-
-QT_AUTOBRIEF           = NO
-
-# The MULTILINE_CPP_IS_BRIEF tag can be set to YES to make doxygen treat a
-# multi-line C++ special comment block (i.e. a block of //! or /// comments) as
-# a brief description. This used to be the default behavior. The new default is
-# to treat a multi-line C++ comment block as a detailed description. Set this
-# tag to YES if you prefer the old behavior instead.
-#
-# Note that setting this tag to YES also means that rational rose comments are
-# not recognized any more.
-# The default value is: NO.
-
-MULTILINE_CPP_IS_BRIEF = NO
-
-# If the INHERIT_DOCS tag is set to YES then an undocumented member inherits the
-# documentation from any documented member that it re-implements.
-# The default value is: YES.
-
-INHERIT_DOCS           = YES
-
-# If the SEPARATE_MEMBER_PAGES tag is set to YES, then doxygen will produce a
-# new page for each member. If set to NO, the documentation of a member will be
-# part of the file/class/namespace that contains it.
-# The default value is: NO.
-
-SEPARATE_MEMBER_PAGES  = NO
-
-# The TAB_SIZE tag can be used to set the number of spaces in a tab. Doxygen
-# uses this value to replace tabs by spaces in code fragments.
-# Minimum value: 1, maximum value: 16, default value: 4.
-
-TAB_SIZE               = 4
-
-# This tag can be used to specify a number of aliases that act as commands in
-# the documentation. An alias has the form:
-# name=value
-# For example adding
-# "sideeffect=@par Side Effects:\n"
-# will allow you to put the command \sideeffect (or @sideeffect) in the
-# documentation, which will result in a user-defined paragraph with heading
-# "Side Effects:". You can put \n's in the value part of an alias to insert
-# newlines.
-
-ALIASES                =
-
-# This tag can be used to specify a number of word-keyword mappings (TCL only).
-# A mapping has the form "name=value". For example adding "class=itcl::class"
-# will allow you to use the command class in the itcl::class meaning.
-
-TCL_SUBST              =
-
-# Set the OPTIMIZE_OUTPUT_FOR_C tag to YES if your project consists of C sources
-# only. Doxygen will then generate output that is more tailored for C. For
-# instance, some of the names that are used will be different. The list of all
-# members will be omitted, etc.
-# The default value is: NO.
-
-OPTIMIZE_OUTPUT_FOR_C  = NO
-
-# Set the OPTIMIZE_OUTPUT_JAVA tag to YES if your project consists of Java or
-# Python sources only. Doxygen will then generate output that is more tailored
-# for that language. For instance, namespaces will be presented as packages,
-# qualified scopes will look different, etc.
-# The default value is: NO.
-
-OPTIMIZE_OUTPUT_JAVA   = NO
-
-# Set the OPTIMIZE_FOR_FORTRAN tag to YES if your project consists of Fortran
-# sources. Doxygen will then generate output that is tailored for Fortran.
-# The default value is: NO.
-
-OPTIMIZE_FOR_FORTRAN   = NO
-
-# Set the OPTIMIZE_OUTPUT_VHDL tag to YES if your project consists of VHDL
-# sources. Doxygen will then generate output that is tailored for VHDL.
-# The default value is: NO.
-
-OPTIMIZE_OUTPUT_VHDL   = NO
-
-# Doxygen selects the parser to use depending on the extension of the files it
-# parses. With this tag you can assign which parser to use for a given
-# extension. Doxygen has a built-in mapping, but you can override or extend it
-# using this tag. The format is ext=language, where ext is a file extension, and
-# language is one of the parsers supported by doxygen: IDL, Java, Javascript,
-# C#, C, C++, D, PHP, Objective-C, Python, Fortran, VHDL. For instance to make
-# doxygen treat .inc files as Fortran files (default is PHP), and .f files as C
-# (default is Fortran), use: inc=Fortran f=C.
-#
-# Note For files without extension you can use no_extension as a placeholder.
-#
-# Note that for custom extensions you also need to set FILE_PATTERNS otherwise
-# the files are not read by doxygen.
-
-EXTENSION_MAPPING      =
-
-# If the MARKDOWN_SUPPORT tag is enabled then doxygen pre-processes all comments
-# according to the Markdown format, which allows for more readable
-# documentation. See http://daringfireball.net/projects/markdown/ for details.
-# The output of markdown processing is further processed by doxygen, so you can
-# mix doxygen, HTML, and XML commands with Markdown formatting. Disable only in
-# case of backward compatibilities issues.
-# The default value is: YES.
-
-MARKDOWN_SUPPORT       = YES
-
-# When enabled doxygen tries to link words that correspond to documented
-# classes, or namespaces to their corresponding documentation. Such a link can
-# be prevented in individual cases by by putting a % sign in front of the word
-# or globally by setting AUTOLINK_SUPPORT to NO.
-# The default value is: YES.
-
-AUTOLINK_SUPPORT       = YES
-
-# If you use STL classes (i.e. std::string, std::vector, etc.) but do not want
-# to include (a tag file for) the STL sources as input, then you should set this
-# tag to YES in order to let doxygen match functions declarations and
-# definitions whose arguments contain STL classes (e.g. func(std::string);
-# versus func(std::string) {}). This also make the inheritance and collaboration
-# diagrams that involve STL classes more complete and accurate.
-# The default value is: NO.
-
-BUILTIN_STL_SUPPORT    = YES
-
-# If you use Microsoft's C++/CLI language, you should set this option to YES to
-# enable parsing support.
-# The default value is: NO.
-
-CPP_CLI_SUPPORT        = NO
-
-# Set the SIP_SUPPORT tag to YES if your project consists of sip (see:
-# http://www.riverbankcomputing.co.uk/software/sip/intro) sources only. Doxygen
-# will parse them like normal C++ but will assume all classes use public instead
-# of private inheritance when no explicit protection keyword is present.
-# The default value is: NO.
-
-SIP_SUPPORT            = NO
-
-# For Microsoft's IDL there are propget and propput attributes to indicate
-# getter and setter methods for a property. Setting this option to YES will make
-# doxygen to replace the get and set methods by a property in the documentation.
-# This will only work if the methods are indeed getting or setting a simple
-# type. If this is not the case, or you want to show the methods anyway, you
-# should set this option to NO.
-# The default value is: YES.
-
-IDL_PROPERTY_SUPPORT   = YES
-
-# If member grouping is used in the documentation and the DISTRIBUTE_GROUP_DOC
-# tag is set to YES, then doxygen will reuse the documentation of the first
-# member in the group (if any) for the other members of the group. By default
-# all members of a group must be documented explicitly.
-# The default value is: NO.
-
-DISTRIBUTE_GROUP_DOC   = NO
-
-# Set the SUBGROUPING tag to YES to allow class member groups of the same type
-# (for instance a group of public functions) to be put as a subgroup of that
-# type (e.g. under the Public Functions section). Set it to NO to prevent
-# subgrouping. Alternatively, this can be done per class using the
-# \nosubgrouping command.
-# The default value is: YES.
-
-SUBGROUPING            = YES
-
-# When the INLINE_GROUPED_CLASSES tag is set to YES, classes, structs and unions
-# are shown inside the group in which they are included (e.g. using \ingroup)
-# instead of on a separate page (for HTML and Man pages) or section (for LaTeX
-# and RTF).
-#
-# Note that this feature does not work in combination with
-# SEPARATE_MEMBER_PAGES.
-# The default value is: NO.
-
-INLINE_GROUPED_CLASSES = NO
-
-# When the INLINE_SIMPLE_STRUCTS tag is set to YES, structs, classes, and unions
-# with only public data fields or simple typedef fields will be shown inline in
-# the documentation of the scope in which they are defined (i.e. file,
-# namespace, or group documentation), provided this scope is documented. If set
-# to NO, structs, classes, and unions are shown on a separate page (for HTML and
-# Man pages) or section (for LaTeX and RTF).
-# The default value is: NO.
-
-INLINE_SIMPLE_STRUCTS  = NO
-
-# When TYPEDEF_HIDES_STRUCT tag is enabled, a typedef of a struct, union, or
-# enum is documented as struct, union, or enum with the name of the typedef. So
-# typedef struct TypeS {} TypeT, will appear in the documentation as a struct
-# with name TypeT. When disabled the typedef will appear as a member of a file,
-# namespace, or class. And the struct will be named TypeS. This can typically be
-# useful for C code in case the coding convention dictates that all compound
-# types are typedef'ed and only the typedef is referenced, never the tag name.
-# The default value is: NO.
-
-TYPEDEF_HIDES_STRUCT   = NO
-
-# The size of the symbol lookup cache can be set using LOOKUP_CACHE_SIZE. This
-# cache is used to resolve symbols given their name and scope. Since this can be
-# an expensive process and often the same symbol appears multiple times in the
-# code, doxygen keeps a cache of pre-resolved symbols. If the cache is too small
-# doxygen will become slower. If the cache is too large, memory is wasted. The
-# cache size is given by this formula: 2^(16+LOOKUP_CACHE_SIZE). The valid range
-# is 0..9, the default is 0, corresponding to a cache size of 2^16=65536
-# symbols. At the end of a run doxygen will report the cache usage and suggest
-# the optimal cache size from a speed point of view.
-# Minimum value: 0, maximum value: 9, default value: 0.
-
-LOOKUP_CACHE_SIZE      = 0
-
-#---------------------------------------------------------------------------
-# Build related configuration options
-#---------------------------------------------------------------------------
-
-# If the EXTRACT_ALL tag is set to YES doxygen will assume all entities in
-# documentation are documented, even if no documentation was available. Private
-# class members and static file members will be hidden unless the
-# EXTRACT_PRIVATE respectively EXTRACT_STATIC tags are set to YES.
-# Note: This will also disable the warnings about undocumented members that are
-# normally produced when WARNINGS is set to YES.
-# The default value is: NO.
-
-EXTRACT_ALL            = YES
-
-# If the EXTRACT_PRIVATE tag is set to YES all private members of a class will
-# be included in the documentation.
-# The default value is: NO.
-
-EXTRACT_PRIVATE        = YES
-
-# If the EXTRACT_PACKAGE tag is set to YES all members with package or internal
-# scope will be included in the documentation.
-# The default value is: NO.
-
-EXTRACT_PACKAGE        = YES
-
-# If the EXTRACT_STATIC tag is set to YES all static members of a file will be
-# included in the documentation.
-# The default value is: NO.
-
-EXTRACT_STATIC         = YES
-
-# If the EXTRACT_LOCAL_CLASSES tag is set to YES classes (and structs) defined
-# locally in source files will be included in the documentation. If set to NO
-# only classes defined in header files are included. Does not have any effect
-# for Java sources.
-# The default value is: YES.
-
-EXTRACT_LOCAL_CLASSES  = YES
-
-# This flag is only useful for Objective-C code. When set to YES local methods,
-# which are defined in the implementation section but not in the interface are
-# included in the documentation. If set to NO only methods in the interface are
-# included.
-# The default value is: NO.
-
-EXTRACT_LOCAL_METHODS  = YES
-
-# If this flag is set to YES, the members of anonymous namespaces will be
-# extracted and appear in the documentation as a namespace called
-# 'anonymous_namespace{file}', where file will be replaced with the base name of
-# the file that contains the anonymous namespace. By default anonymous namespace
-# are hidden.
-# The default value is: NO.
-
-EXTRACT_ANON_NSPACES   = YES
-
-# If the HIDE_UNDOC_MEMBERS tag is set to YES, doxygen will hide all
-# undocumented members inside documented classes or files. If set to NO these
-# members will be included in the various overviews, but no documentation
-# section is generated. This option has no effect if EXTRACT_ALL is enabled.
-# The default value is: NO.
-
-HIDE_UNDOC_MEMBERS     = NO
-
-# If the HIDE_UNDOC_CLASSES tag is set to YES, doxygen will hide all
-# undocumented classes that are normally visible in the class hierarchy. If set
-# to NO these classes will be included in the various overviews. This option has
-# no effect if EXTRACT_ALL is enabled.
-# The default value is: NO.
-
-HIDE_UNDOC_CLASSES     = NO
-
-# If the HIDE_FRIEND_COMPOUNDS tag is set to YES, doxygen will hide all friend
-# (class|struct|union) declarations. If set to NO these declarations will be
-# included in the documentation.
-# The default value is: NO.
-
-HIDE_FRIEND_COMPOUNDS  = NO
-
-# If the HIDE_IN_BODY_DOCS tag is set to YES, doxygen will hide any
-# documentation blocks found inside the body of a function. If set to NO these
-# blocks will be appended to the function's detailed documentation block.
-# The default value is: NO.
-
-HIDE_IN_BODY_DOCS      = NO
-
-# The INTERNAL_DOCS tag determines if documentation that is typed after a
-# \internal command is included. If the tag is set to NO then the documentation
-# will be excluded. Set it to YES to include the internal documentation.
-# The default value is: NO.
-
-INTERNAL_DOCS          = NO
-
-# If the CASE_SENSE_NAMES tag is set to NO then doxygen will only generate file
-# names in lower-case letters. If set to YES upper-case letters are also
-# allowed. This is useful if you have classes or files whose names only differ
-# in case and if your file system supports case sensitive file names. Windows
-# and Mac users are advised to set this option to NO.
-# The default value is: system dependent.
-
-CASE_SENSE_NAMES       = YES
-
-# If the HIDE_SCOPE_NAMES tag is set to NO then doxygen will show members with
-# their full class and namespace scopes in the documentation. If set to YES the
-# scope will be hidden.
-# The default value is: NO.
-
-HIDE_SCOPE_NAMES       = NO
-
-# If the SHOW_INCLUDE_FILES tag is set to YES then doxygen will put a list of
-# the files that are included by a file in the documentation of that file.
-# The default value is: YES.
-
-SHOW_INCLUDE_FILES     = YES
-
-# If the SHOW_GROUPED_MEMB_INC tag is set to YES then Doxygen will add for each
-# grouped member an include statement to the documentation, telling the reader
-# which file to include in order to use the member.
-# The default value is: NO.
-
-SHOW_GROUPED_MEMB_INC  = NO
-
-# If the FORCE_LOCAL_INCLUDES tag is set to YES then doxygen will list include
-# files with double quotes in the documentation rather than with sharp brackets.
-# The default value is: NO.
-
-FORCE_LOCAL_INCLUDES   = NO
-
-# If the INLINE_INFO tag is set to YES then a tag [inline] is inserted in the
-# documentation for inline members.
-# The default value is: YES.
-
-INLINE_INFO            = YES
-
-# If the SORT_MEMBER_DOCS tag is set to YES then doxygen will sort the
-# (detailed) documentation of file and class members alphabetically by member
-# name. If set to NO the members will appear in declaration order.
-# The default value is: YES.
-
-SORT_MEMBER_DOCS       = YES
-
-# If the SORT_BRIEF_DOCS tag is set to YES then doxygen will sort the brief
-# descriptions of file, namespace and class members alphabetically by member
-# name. If set to NO the members will appear in declaration order. Note that
-# this will also influence the order of the classes in the class list.
-# The default value is: NO.
-
-SORT_BRIEF_DOCS        = NO
-
-# If the SORT_MEMBERS_CTORS_1ST tag is set to YES then doxygen will sort the
-# (brief and detailed) documentation of class members so that constructors and
-# destructors are listed first. If set to NO the constructors will appear in the
-# respective orders defined by SORT_BRIEF_DOCS and SORT_MEMBER_DOCS.
-# Note: If SORT_BRIEF_DOCS is set to NO this option is ignored for sorting brief
-# member documentation.
-# Note: If SORT_MEMBER_DOCS is set to NO this option is ignored for sorting
-# detailed member documentation.
-# The default value is: NO.
-
-SORT_MEMBERS_CTORS_1ST = YES
-
-# If the SORT_GROUP_NAMES tag is set to YES then doxygen will sort the hierarchy
-# of group names into alphabetical order. If set to NO the group names will
-# appear in their defined order.
-# The default value is: NO.
-
-SORT_GROUP_NAMES       = NO
-
-# If the SORT_BY_SCOPE_NAME tag is set to YES, the class list will be sorted by
-# fully-qualified names, including namespaces. If set to NO, the class list will
-# be sorted only by class name, not including the namespace part.
-# Note: This option is not very useful if HIDE_SCOPE_NAMES is set to YES.
-# Note: This option applies only to the class list, not to the alphabetical
-# list.
-# The default value is: NO.
-
-SORT_BY_SCOPE_NAME     = NO
-
-# If the STRICT_PROTO_MATCHING option is enabled and doxygen fails to do proper
-# type resolution of all parameters of a function it will reject a match between
-# the prototype and the implementation of a member function even if there is
-# only one candidate or it is obvious which candidate to choose by doing a
-# simple string match. By disabling STRICT_PROTO_MATCHING doxygen will still
-# accept a match between prototype and implementation in such cases.
-# The default value is: NO.
-
-STRICT_PROTO_MATCHING  = NO
-
-# The GENERATE_TODOLIST tag can be used to enable ( YES) or disable ( NO) the
-# todo list. This list is created by putting \todo commands in the
-# documentation.
-# The default value is: YES.
-
-GENERATE_TODOLIST      = YES
-
-# The GENERATE_TESTLIST tag can be used to enable ( YES) or disable ( NO) the
-# test list. This list is created by putting \test commands in the
-# documentation.
-# The default value is: YES.
-
-GENERATE_TESTLIST      = YES
-
-# The GENERATE_BUGLIST tag can be used to enable ( YES) or disable ( NO) the bug
-# list. This list is created by putting \bug commands in the documentation.
-# The default value is: YES.
-
-GENERATE_BUGLIST       = YES
-
-# The GENERATE_DEPRECATEDLIST tag can be used to enable ( YES) or disable ( NO)
-# the deprecated list. This list is created by putting \deprecated commands in
-# the documentation.
-# The default value is: YES.
-
-GENERATE_DEPRECATEDLIST= YES
-
-# The ENABLED_SECTIONS tag can be used to enable conditional documentation
-# sections, marked by \if <section_label> ... \endif and \cond <section_label>
-# ... \endcond blocks.
-
-ENABLED_SECTIONS       =
-
-# The MAX_INITIALIZER_LINES tag determines the maximum number of lines that the
-# initial value of a variable or macro / define can have for it to appear in the
-# documentation. If the initializer consists of more lines than specified here
-# it will be hidden. Use a value of 0 to hide initializers completely. The
-# appearance of the value of individual variables and macros / defines can be
-# controlled using \showinitializer or \hideinitializer command in the
-# documentation regardless of this setting.
-# Minimum value: 0, maximum value: 10000, default value: 30.
-
-MAX_INITIALIZER_LINES  = 30
-
-# Set the SHOW_USED_FILES tag to NO to disable the list of files generated at
-# the bottom of the documentation of classes and structs. If set to YES the list
-# will mention the files that were used to generate the documentation.
-# The default value is: YES.
-
-SHOW_USED_FILES        = YES
-
-# Set the SHOW_FILES tag to NO to disable the generation of the Files page. This
-# will remove the Files entry from the Quick Index and from the Folder Tree View
-# (if specified).
-# The default value is: YES.
-
-SHOW_FILES             = YES
-
-# Set the SHOW_NAMESPACES tag to NO to disable the generation of the Namespaces
-# page. This will remove the Namespaces entry from the Quick Index and from the
-# Folder Tree View (if specified).
-# The default value is: YES.
-
-SHOW_NAMESPACES        = YES
-
-# The FILE_VERSION_FILTER tag can be used to specify a program or script that
-# doxygen should invoke to get the current version for each file (typically from
-# the version control system). Doxygen will invoke the program by executing (via
-# popen()) the command command input-file, where command is the value of the
-# FILE_VERSION_FILTER tag, and input-file is the name of an input file provided
-# by doxygen. Whatever the program writes to standard output is used as the file
-# version. For an example see the documentation.
-
-FILE_VERSION_FILTER    =
-
-# The LAYOUT_FILE tag can be used to specify a layout file which will be parsed
-# by doxygen. The layout file controls the global structure of the generated
-# output files in an output format independent way. To create the layout file
-# that represents doxygen's defaults, run doxygen with the -l option. You can
-# optionally specify a file name after the option, if omitted DoxygenLayout.xml
-# will be used as the name of the layout file.
-#
-# Note that if you run doxygen from a directory containing a file called
-# DoxygenLayout.xml, doxygen will parse it automatically even if the LAYOUT_FILE
-# tag is left empty.
-
-LAYOUT_FILE            =
-
-# The CITE_BIB_FILES tag can be used to specify one or more bib files containing
-# the reference definitions. This must be a list of .bib files. The .bib
-# extension is automatically appended if omitted. This requires the bibtex tool
-# to be installed. See also http://en.wikipedia.org/wiki/BibTeX for more info.
-# For LaTeX the style of the bibliography can be controlled using
-# LATEX_BIB_STYLE. To use this feature you need bibtex and perl available in the
-# search path. Do not use file names with spaces, bibtex cannot handle them. See
-# also \cite for info how to create references.
-
-CITE_BIB_FILES         =
-
-#---------------------------------------------------------------------------
-# Configuration options related to warning and progress messages
-#---------------------------------------------------------------------------
-
-# The QUIET tag can be used to turn on/off the messages that are generated to
-# standard output by doxygen. If QUIET is set to YES this implies that the
-# messages are off.
-# The default value is: NO.
-
-QUIET                  = NO
-
-# The WARNINGS tag can be used to turn on/off the warning messages that are
-# generated to standard error ( stderr) by doxygen. If WARNINGS is set to YES
-# this implies that the warnings are on.
-#
-# Tip: Turn warnings on while writing the documentation.
-# The default value is: YES.
-
-WARNINGS               = YES
-
-# If the WARN_IF_UNDOCUMENTED tag is set to YES, then doxygen will generate
-# warnings for undocumented members. If EXTRACT_ALL is set to YES then this flag
-# will automatically be disabled.
-# The default value is: YES.
-
-WARN_IF_UNDOCUMENTED   = YES
-
-# If the WARN_IF_DOC_ERROR tag is set to YES, doxygen will generate warnings for
-# potential errors in the documentation, such as not documenting some parameters
-# in a documented function, or documenting parameters that don't exist or using
-# markup commands wrongly.
-# The default value is: YES.
-
-WARN_IF_DOC_ERROR      = YES
-
-# This WARN_NO_PARAMDOC option can be enabled to get warnings for functions that
-# are documented, but have no documentation for their parameters or return
-# value. If set to NO doxygen will only warn about wrong or incomplete parameter
-# documentation, but not about the absence of documentation.
-# The default value is: NO.
-
-WARN_NO_PARAMDOC       = NO
-
-# The WARN_FORMAT tag determines the format of the warning messages that doxygen
-# can produce. The string should contain the $file, $line, and $text tags, which
-# will be replaced by the file and line number from which the warning originated
-# and the warning text. Optionally the format may contain $version, which will
-# be replaced by the version of the file (if it could be obtained via
-# FILE_VERSION_FILTER)
-# The default value is: $file:$line: $text.
-
-WARN_FORMAT            = "$file:$line: $text"
-
-# The WARN_LOGFILE tag can be used to specify a file to which warning and error
-# messages should be written. If left blank the output is written to standard
-# error (stderr).
-
-WARN_LOGFILE           =
-
-#---------------------------------------------------------------------------
-# Configuration options related to the input files
-#---------------------------------------------------------------------------
-
-# The INPUT tag is used to specify the files and/or directories that contain
-# documented source files. You may enter file names like myfile.cpp or
-# directories like /usr/src/myproject. Separate the files or directories with
-# spaces.
-# Note: If this tag is empty the current directory is searched.
-
-INPUT                  = ../lib \
-		         ../tests \
-			 ../benchmarks
-
-# This tag can be used to specify the character encoding of the source files
-# that doxygen parses. Internally doxygen uses the UTF-8 encoding. Doxygen uses
-# libiconv (or the iconv built into libc) for the transcoding. See the libiconv
-# documentation (see: http://www.gnu.org/software/libiconv) for the list of
-# possible encodings.
-# The default value is: UTF-8.
-
-INPUT_ENCODING         = UTF-8
-
-# If the value of the INPUT tag contains directories, you can use the
-# FILE_PATTERNS tag to specify one or more wildcard patterns (like *.cpp and
-# *.h) to filter out the source-files in the directories. If left blank the
-# following patterns are tested:*.c, *.cc, *.cxx, *.cpp, *.c++, *.java, *.ii,
-# *.ixx, *.ipp, *.i++, *.inl, *.idl, *.ddl, *.odl, *.h, *.hh, *.hxx, *.hpp,
-# *.h++, *.cs, *.d, *.php, *.php4, *.php5, *.phtml, *.inc, *.m, *.markdown,
-# *.md, *.mm, *.dox, *.py, *.f90, *.f, *.for, *.tcl, *.vhd, *.vhdl, *.ucf,
-# *.qsf, *.as and *.js.
-
-FILE_PATTERNS          =
-
-# The RECURSIVE tag can be used to specify whether or not subdirectories should
-# be searched for input files as well.
-# The default value is: NO.
-
-RECURSIVE              = YES
-
-# The EXCLUDE tag can be used to specify files and/or directories that should be
-# excluded from the INPUT source files. This way you can easily exclude a
-# subdirectory from a directory tree whose root is specified with the INPUT tag.
-#
-# Note that relative paths are relative to the directory from which doxygen is
-# run.
-
-EXCLUDE                =
-
-# The EXCLUDE_SYMLINKS tag can be used to select whether or not files or
-# directories that are symbolic links (a Unix file system feature) are excluded
-# from the input.
-# The default value is: NO.
-
-EXCLUDE_SYMLINKS       = NO
-
-# If the value of the INPUT tag contains directories, you can use the
-# EXCLUDE_PATTERNS tag to specify one or more wildcard patterns to exclude
-# certain files from those directories.
-#
-# Note that the wildcards are matched against the file with absolute path, so to
-# exclude all test directories for example use the pattern */test/*
-
-EXCLUDE_PATTERNS       =
-
-# The EXCLUDE_SYMBOLS tag can be used to specify one or more symbol names
-# (namespaces, classes, functions, etc.) that should be excluded from the
-# output. The symbol name can be a fully qualified name, a word, or if the
-# wildcard * is used, a substring. Examples: ANamespace, AClass,
-# AClass::ANamespace, ANamespace::*Test
-#
-# Note that the wildcards are matched against the file with absolute path, so to
-# exclude all test directories use the pattern */test/*
-
-EXCLUDE_SYMBOLS        =
-
-# The EXAMPLE_PATH tag can be used to specify one or more files or directories
-# that contain example code fragments that are included (see the \include
-# command).
-
-EXAMPLE_PATH           =
-
-# If the value of the EXAMPLE_PATH tag contains directories, you can use the
-# EXAMPLE_PATTERNS tag to specify one or more wildcard pattern (like *.cpp and
-# *.h) to filter out the source-files in the directories. If left blank all
-# files are included.
-
-EXAMPLE_PATTERNS       =
-
-# If the EXAMPLE_RECURSIVE tag is set to YES then subdirectories will be
-# searched for input files to be used with the \include or \dontinclude commands
-# irrespective of the value of the RECURSIVE tag.
-# The default value is: NO.
-
-EXAMPLE_RECURSIVE      = NO
-
-# The IMAGE_PATH tag can be used to specify one or more files or directories
-# that contain images that are to be included in the documentation (see the
-# \image command).
-
-IMAGE_PATH             =
-
-# The INPUT_FILTER tag can be used to specify a program that doxygen should
-# invoke to filter for each input file. Doxygen will invoke the filter program
-# by executing (via popen()) the command:
-#
-# <filter> <input-file>
-#
-# where <filter> is the value of the INPUT_FILTER tag, and <input-file> is the
-# name of an input file. Doxygen will then use the output that the filter
-# program writes to standard output. If FILTER_PATTERNS is specified, this tag
-# will be ignored.
-#
-# Note that the filter must not add or remove lines; it is applied before the
-# code is scanned, but not when the output code is generated. If lines are added
-# or removed, the anchors will not be placed correctly.
-
-INPUT_FILTER           =
-
-# The FILTER_PATTERNS tag can be used to specify filters on a per file pattern
-# basis. Doxygen will compare the file name with each pattern and apply the
-# filter if there is a match. The filters are a list of the form: pattern=filter
-# (like *.cpp=my_cpp_filter). See INPUT_FILTER for further information on how
-# filters are used. If the FILTER_PATTERNS tag is empty or if none of the
-# patterns match the file name, INPUT_FILTER is applied.
-
-FILTER_PATTERNS        =
-
-# If the FILTER_SOURCE_FILES tag is set to YES, the input filter (if set using
-# INPUT_FILTER ) will also be used to filter the input files that are used for
-# producing the source files to browse (i.e. when SOURCE_BROWSER is set to YES).
-# The default value is: NO.
-
-FILTER_SOURCE_FILES    = NO
-
-# The FILTER_SOURCE_PATTERNS tag can be used to specify source filters per file
-# pattern. A pattern will override the setting for FILTER_PATTERN (if any) and
-# it is also possible to disable source filtering for a specific pattern using
-# *.ext= (so without naming a filter).
-# This tag requires that the tag FILTER_SOURCE_FILES is set to YES.
-
-FILTER_SOURCE_PATTERNS =
-
-# If the USE_MDFILE_AS_MAINPAGE tag refers to the name of a markdown file that
-# is part of the input, its contents will be placed on the main page
-# (index.html). This can be useful if you have a project on for instance GitHub
-# and want to reuse the introduction page also for the doxygen output.
-
-USE_MDFILE_AS_MAINPAGE =
-
-#---------------------------------------------------------------------------
-# Configuration options related to source browsing
-#---------------------------------------------------------------------------
-
-# If the SOURCE_BROWSER tag is set to YES then a list of source files will be
-# generated. Documented entities will be cross-referenced with these sources.
-#
-# Note: To get rid of all source code in the generated output, make sure that
-# also VERBATIM_HEADERS is set to NO.
-# The default value is: NO.
-
-SOURCE_BROWSER         = YES
-
-# Setting the INLINE_SOURCES tag to YES will include the body of functions,
-# classes and enums directly into the documentation.
-# The default value is: NO.
-
-INLINE_SOURCES         = NO
-
-# Setting the STRIP_CODE_COMMENTS tag to YES will instruct doxygen to hide any
-# special comment blocks from generated source code fragments. Normal C, C++ and
-# Fortran comments will always remain visible.
-# The default value is: YES.
-
-STRIP_CODE_COMMENTS    = YES
-
-# If the REFERENCED_BY_RELATION tag is set to YES then for each documented
-# function all documented functions referencing it will be listed.
-# The default value is: NO.
-
-REFERENCED_BY_RELATION = NO
-
-# If the REFERENCES_RELATION tag is set to YES then for each documented function
-# all documented entities called/used by that function will be listed.
-# The default value is: NO.
-
-REFERENCES_RELATION    = NO
-
-# If the REFERENCES_LINK_SOURCE tag is set to YES and SOURCE_BROWSER tag is set
-# to YES, then the hyperlinks from functions in REFERENCES_RELATION and
-# REFERENCED_BY_RELATION lists will link to the source code. Otherwise they will
-# link to the documentation.
-# The default value is: YES.
-
-REFERENCES_LINK_SOURCE = YES
-
-# If SOURCE_TOOLTIPS is enabled (the default) then hovering a hyperlink in the
-# source code will show a tooltip with additional information such as prototype,
-# brief description and links to the definition and documentation. Since this
-# will make the HTML file larger and loading of large files a bit slower, you
-# can opt to disable this feature.
-# The default value is: YES.
-# This tag requires that the tag SOURCE_BROWSER is set to YES.
-
-SOURCE_TOOLTIPS        = YES
-
-# If the USE_HTAGS tag is set to YES then the references to source code will
-# point to the HTML generated by the htags(1) tool instead of doxygen built-in
-# source browser. The htags tool is part of GNU's global source tagging system
-# (see http://www.gnu.org/software/global/global.html). You will need version
-# 4.8.6 or higher.
-#
-# To use it do the following:
-# - Install the latest version of global
-# - Enable SOURCE_BROWSER and USE_HTAGS in the config file
-# - Make sure the INPUT points to the root of the source tree
-# - Run doxygen as normal
-#
-# Doxygen will invoke htags (and that will in turn invoke gtags), so these
-# tools must be available from the command line (i.e. in the search path).
-#
-# The result: instead of the source browser generated by doxygen, the links to
-# source code will now point to the output of htags.
-# The default value is: NO.
-# This tag requires that the tag SOURCE_BROWSER is set to YES.
-
-USE_HTAGS              = NO
-
-# If the VERBATIM_HEADERS tag is set the YES then doxygen will generate a
-# verbatim copy of the header file for each class for which an include is
-# specified. Set to NO to disable this.
-# See also: Section \class.
-# The default value is: YES.
-
-VERBATIM_HEADERS       = YES
-
-#---------------------------------------------------------------------------
-# Configuration options related to the alphabetical class index
-#---------------------------------------------------------------------------
-
-# If the ALPHABETICAL_INDEX tag is set to YES, an alphabetical index of all
-# compounds will be generated. Enable this if the project contains a lot of
-# classes, structs, unions or interfaces.
-# The default value is: YES.
-
-ALPHABETICAL_INDEX     = YES
-
-# The COLS_IN_ALPHA_INDEX tag can be used to specify the number of columns in
-# which the alphabetical index list will be split.
-# Minimum value: 1, maximum value: 20, default value: 5.
-# This tag requires that the tag ALPHABETICAL_INDEX is set to YES.
-
-COLS_IN_ALPHA_INDEX    = 5
-
-# In case all classes in a project start with a common prefix, all classes will
-# be put under the same header in the alphabetical index. The IGNORE_PREFIX tag
-# can be used to specify a prefix (or a list of prefixes) that should be ignored
-# while generating the index headers.
-# This tag requires that the tag ALPHABETICAL_INDEX is set to YES.
-
-IGNORE_PREFIX          =
-
-#---------------------------------------------------------------------------
-# Configuration options related to the HTML output
-#---------------------------------------------------------------------------
-
-# If the GENERATE_HTML tag is set to YES doxygen will generate HTML output
-# The default value is: YES.
-
-GENERATE_HTML          = YES
-
-# The HTML_OUTPUT tag is used to specify where the HTML docs will be put. If a
-# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
-# it.
-# The default directory is: html.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-HTML_OUTPUT            = html
-
-# The HTML_FILE_EXTENSION tag can be used to specify the file extension for each
-# generated HTML page (for example: .htm, .php, .asp).
-# The default value is: .html.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-HTML_FILE_EXTENSION    = .html
-
-# The HTML_HEADER tag can be used to specify a user-defined HTML header file for
-# each generated HTML page. If the tag is left blank doxygen will generate a
-# standard header.
-#
-# To get valid HTML the header file that includes any scripts and style sheets
-# that doxygen needs, which is dependent on the configuration options used (e.g.
-# the setting GENERATE_TREEVIEW). It is highly recommended to start with a
-# default header using
-# doxygen -w html new_header.html new_footer.html new_stylesheet.css
-# YourConfigFile
-# and then modify the file new_header.html. See also section "Doxygen usage"
-# for information on how to generate the default header that doxygen normally
-# uses.
-# Note: The header is subject to change so you typically have to regenerate the
-# default header when upgrading to a newer version of doxygen. For a description
-# of the possible markers and block names see the documentation.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-HTML_HEADER            =
-
-# The HTML_FOOTER tag can be used to specify a user-defined HTML footer for each
-# generated HTML page. If the tag is left blank doxygen will generate a standard
-# footer. See HTML_HEADER for more information on how to generate a default
-# footer and what special commands can be used inside the footer. See also
-# section "Doxygen usage" for information on how to generate the default footer
-# that doxygen normally uses.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-HTML_FOOTER            =
-
-# The HTML_STYLESHEET tag can be used to specify a user-defined cascading style
-# sheet that is used by each HTML page. It can be used to fine-tune the look of
-# the HTML output. If left blank doxygen will generate a default style sheet.
-# See also section "Doxygen usage" for information on how to generate the style
-# sheet that doxygen normally uses.
-# Note: It is recommended to use HTML_EXTRA_STYLESHEET instead of this tag, as
-# it is more robust and this tag (HTML_STYLESHEET) will in the future become
-# obsolete.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-HTML_STYLESHEET        =
-
-# The HTML_EXTRA_STYLESHEET tag can be used to specify an additional user-
-# defined cascading style sheet that is included after the standard style sheets
-# created by doxygen. Using this option one can overrule certain style aspects.
-# This is preferred over using HTML_STYLESHEET since it does not replace the
-# standard style sheet and is therefor more robust against future updates.
-# Doxygen will copy the style sheet file to the output directory. For an example
-# see the documentation.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-HTML_EXTRA_STYLESHEET  =
-
-# The HTML_EXTRA_FILES tag can be used to specify one or more extra images or
-# other source files which should be copied to the HTML output directory. Note
-# that these files will be copied to the base HTML output directory. Use the
-# $relpath^ marker in the HTML_HEADER and/or HTML_FOOTER files to load these
-# files. In the HTML_STYLESHEET file, use the file name only. Also note that the
-# files will be copied as-is; there are no commands or markers available.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-HTML_EXTRA_FILES       =
-
-# The HTML_COLORSTYLE_HUE tag controls the color of the HTML output. Doxygen
-# will adjust the colors in the stylesheet and background images according to
-# this color. Hue is specified as an angle on a colorwheel, see
-# http://en.wikipedia.org/wiki/Hue for more information. For instance the value
-# 0 represents red, 60 is yellow, 120 is green, 180 is cyan, 240 is blue, 300
-# purple, and 360 is red again.
-# Minimum value: 0, maximum value: 359, default value: 220.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-HTML_COLORSTYLE_HUE    = 220
-
-# The HTML_COLORSTYLE_SAT tag controls the purity (or saturation) of the colors
-# in the HTML output. For a value of 0 the output will use grayscales only. A
-# value of 255 will produce the most vivid colors.
-# Minimum value: 0, maximum value: 255, default value: 100.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-HTML_COLORSTYLE_SAT    = 100
-
-# The HTML_COLORSTYLE_GAMMA tag controls the gamma correction applied to the
-# luminance component of the colors in the HTML output. Values below 100
-# gradually make the output lighter, whereas values above 100 make the output
-# darker. The value divided by 100 is the actual gamma applied, so 80 represents
-# a gamma of 0.8, The value 220 represents a gamma of 2.2, and 100 does not
-# change the gamma.
-# Minimum value: 40, maximum value: 240, default value: 80.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-HTML_COLORSTYLE_GAMMA  = 80
-
-# If the HTML_TIMESTAMP tag is set to YES then the footer of each generated HTML
-# page will contain the date and time when the page was generated. Setting this
-# to NO can help when comparing the output of multiple runs.
-# The default value is: YES.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-HTML_TIMESTAMP         = YES
-
-# If the HTML_DYNAMIC_SECTIONS tag is set to YES then the generated HTML
-# documentation will contain sections that can be hidden and shown after the
-# page has loaded.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-HTML_DYNAMIC_SECTIONS  = NO
-
-# With HTML_INDEX_NUM_ENTRIES one can control the preferred number of entries
-# shown in the various tree structured indices initially; the user can expand
-# and collapse entries dynamically later on. Doxygen will expand the tree to
-# such a level that at most the specified number of entries are visible (unless
-# a fully collapsed tree already exceeds this amount). So setting the number of
-# entries 1 will produce a full collapsed tree by default. 0 is a special value
-# representing an infinite number of entries and will result in a full expanded
-# tree by default.
-# Minimum value: 0, maximum value: 9999, default value: 100.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-HTML_INDEX_NUM_ENTRIES = 100
-
-# If the GENERATE_DOCSET tag is set to YES, additional index files will be
-# generated that can be used as input for Apple's Xcode 3 integrated development
-# environment (see: http://developer.apple.com/tools/xcode/), introduced with
-# OSX 10.5 (Leopard). To create a documentation set, doxygen will generate a
-# Makefile in the HTML output directory. Running make will produce the docset in
-# that directory and running make install will install the docset in
-# ~/Library/Developer/Shared/Documentation/DocSets so that Xcode will find it at
-# startup. See http://developer.apple.com/tools/creatingdocsetswithdoxygen.html
-# for more information.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-GENERATE_DOCSET        = NO
-
-# This tag determines the name of the docset feed. A documentation feed provides
-# an umbrella under which multiple documentation sets from a single provider
-# (such as a company or product suite) can be grouped.
-# The default value is: Doxygen generated docs.
-# This tag requires that the tag GENERATE_DOCSET is set to YES.
-
-DOCSET_FEEDNAME        = "Doxygen generated docs"
-
-# This tag specifies a string that should uniquely identify the documentation
-# set bundle. This should be a reverse domain-name style string, e.g.
-# com.mycompany.MyDocSet. Doxygen will append .docset to the name.
-# The default value is: org.doxygen.Project.
-# This tag requires that the tag GENERATE_DOCSET is set to YES.
-
-DOCSET_BUNDLE_ID       = org.doxygen.Project
-
-# The DOCSET_PUBLISHER_ID tag specifies a string that should uniquely identify
-# the documentation publisher. This should be a reverse domain-name style
-# string, e.g. com.mycompany.MyDocSet.documentation.
-# The default value is: org.doxygen.Publisher.
-# This tag requires that the tag GENERATE_DOCSET is set to YES.
-
-DOCSET_PUBLISHER_ID    = org.doxygen.Publisher
-
-# The DOCSET_PUBLISHER_NAME tag identifies the documentation publisher.
-# The default value is: Publisher.
-# This tag requires that the tag GENERATE_DOCSET is set to YES.
-
-DOCSET_PUBLISHER_NAME  = Publisher
-
-# If the GENERATE_HTMLHELP tag is set to YES then doxygen generates three
-# additional HTML index files: index.hhp, index.hhc, and index.hhk. The
-# index.hhp is a project file that can be read by Microsoft's HTML Help Workshop
-# (see: http://www.microsoft.com/en-us/download/details.aspx?id=21138) on
-# Windows.
-#
-# The HTML Help Workshop contains a compiler that can convert all HTML output
-# generated by doxygen into a single compiled HTML file (.chm). Compiled HTML
-# files are now used as the Windows 98 help format, and will replace the old
-# Windows help format (.hlp) on all Windows platforms in the future. Compressed
-# HTML files also contain an index, a table of contents, and you can search for
-# words in the documentation. The HTML workshop also contains a viewer for
-# compressed HTML files.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-GENERATE_HTMLHELP      = NO
-
-# The CHM_FILE tag can be used to specify the file name of the resulting .chm
-# file. You can add a path in front of the file if the result should not be
-# written to the html output directory.
-# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
-
-CHM_FILE               =
-
-# The HHC_LOCATION tag can be used to specify the location (absolute path
-# including file name) of the HTML help compiler ( hhc.exe). If non-empty
-# doxygen will try to run the HTML help compiler on the generated index.hhp.
-# The file has to be specified with full path.
-# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
-
-HHC_LOCATION           =
-
-# The GENERATE_CHI flag controls if a separate .chi index file is generated (
-# YES) or that it should be included in the master .chm file ( NO).
-# The default value is: NO.
-# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
-
-GENERATE_CHI           = NO
-
-# The CHM_INDEX_ENCODING is used to encode HtmlHelp index ( hhk), content ( hhc)
-# and project file content.
-# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
-
-CHM_INDEX_ENCODING     =
-
-# The BINARY_TOC flag controls whether a binary table of contents is generated (
-# YES) or a normal table of contents ( NO) in the .chm file.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
-
-BINARY_TOC             = NO
-
-# The TOC_EXPAND flag can be set to YES to add extra items for group members to
-# the table of contents of the HTML help documentation and to the tree view.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_HTMLHELP is set to YES.
-
-TOC_EXPAND             = NO
-
-# If the GENERATE_QHP tag is set to YES and both QHP_NAMESPACE and
-# QHP_VIRTUAL_FOLDER are set, an additional index file will be generated that
-# can be used as input for Qt's qhelpgenerator to generate a Qt Compressed Help
-# (.qch) of the generated HTML documentation.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-GENERATE_QHP           = NO
-
-# If the QHG_LOCATION tag is specified, the QCH_FILE tag can be used to specify
-# the file name of the resulting .qch file. The path specified is relative to
-# the HTML output folder.
-# This tag requires that the tag GENERATE_QHP is set to YES.
-
-QCH_FILE               =
-
-# The QHP_NAMESPACE tag specifies the namespace to use when generating Qt Help
-# Project output. For more information please see Qt Help Project / Namespace
-# (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#namespace).
-# The default value is: org.doxygen.Project.
-# This tag requires that the tag GENERATE_QHP is set to YES.
-
-QHP_NAMESPACE          = org.doxygen.Project
-
-# The QHP_VIRTUAL_FOLDER tag specifies the namespace to use when generating Qt
-# Help Project output. For more information please see Qt Help Project / Virtual
-# Folders (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#virtual-
-# folders).
-# The default value is: doc.
-# This tag requires that the tag GENERATE_QHP is set to YES.
-
-QHP_VIRTUAL_FOLDER     = doc
-
-# If the QHP_CUST_FILTER_NAME tag is set, it specifies the name of a custom
-# filter to add. For more information please see Qt Help Project / Custom
-# Filters (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#custom-
-# filters).
-# This tag requires that the tag GENERATE_QHP is set to YES.
-
-QHP_CUST_FILTER_NAME   =
-
-# The QHP_CUST_FILTER_ATTRS tag specifies the list of the attributes of the
-# custom filter to add. For more information please see Qt Help Project / Custom
-# Filters (see: http://qt-project.org/doc/qt-4.8/qthelpproject.html#custom-
-# filters).
-# This tag requires that the tag GENERATE_QHP is set to YES.
-
-QHP_CUST_FILTER_ATTRS  =
-
-# The QHP_SECT_FILTER_ATTRS tag specifies the list of the attributes this
-# project's filter section matches. Qt Help Project / Filter Attributes (see:
-# http://qt-project.org/doc/qt-4.8/qthelpproject.html#filter-attributes).
-# This tag requires that the tag GENERATE_QHP is set to YES.
-
-QHP_SECT_FILTER_ATTRS  =
-
-# The QHG_LOCATION tag can be used to specify the location of Qt's
-# qhelpgenerator. If non-empty doxygen will try to run qhelpgenerator on the
-# generated .qhp file.
-# This tag requires that the tag GENERATE_QHP is set to YES.
-
-QHG_LOCATION           =
-
-# If the GENERATE_ECLIPSEHELP tag is set to YES, additional index files will be
-# generated, together with the HTML files, they form an Eclipse help plugin. To
-# install this plugin and make it available under the help contents menu in
-# Eclipse, the contents of the directory containing the HTML and XML files needs
-# to be copied into the plugins directory of eclipse. The name of the directory
-# within the plugins directory should be the same as the ECLIPSE_DOC_ID value.
-# After copying Eclipse needs to be restarted before the help appears.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-GENERATE_ECLIPSEHELP   = NO
-
-# A unique identifier for the Eclipse help plugin. When installing the plugin
-# the directory name containing the HTML and XML files should also have this
-# name. Each documentation set should have its own identifier.
-# The default value is: org.doxygen.Project.
-# This tag requires that the tag GENERATE_ECLIPSEHELP is set to YES.
-
-ECLIPSE_DOC_ID         = org.doxygen.Project
-
-# If you want full control over the layout of the generated HTML pages it might
-# be necessary to disable the index and replace it with your own. The
-# DISABLE_INDEX tag can be used to turn on/off the condensed index (tabs) at top
-# of each HTML page. A value of NO enables the index and the value YES disables
-# it. Since the tabs in the index contain the same information as the navigation
-# tree, you can set this option to YES if you also set GENERATE_TREEVIEW to YES.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-DISABLE_INDEX          = NO
-
-# The GENERATE_TREEVIEW tag is used to specify whether a tree-like index
-# structure should be generated to display hierarchical information. If the tag
-# value is set to YES, a side panel will be generated containing a tree-like
-# index structure (just like the one that is generated for HTML Help). For this
-# to work a browser that supports JavaScript, DHTML, CSS and frames is required
-# (i.e. any modern browser). Windows users are probably better off using the
-# HTML help feature. Via custom stylesheets (see HTML_EXTRA_STYLESHEET) one can
-# further fine-tune the look of the index. As an example, the default style
-# sheet generated by doxygen has an example that shows how to put an image at
-# the root of the tree instead of the PROJECT_NAME. Since the tree basically has
-# the same information as the tab index, you could consider setting
-# DISABLE_INDEX to YES when enabling this option.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-GENERATE_TREEVIEW      = YES
-
-# The ENUM_VALUES_PER_LINE tag can be used to set the number of enum values that
-# doxygen will group on one line in the generated HTML documentation.
-#
-# Note that a value of 0 will completely suppress the enum values from appearing
-# in the overview section.
-# Minimum value: 0, maximum value: 20, default value: 4.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-ENUM_VALUES_PER_LINE   = 4
-
-# If the treeview is enabled (see GENERATE_TREEVIEW) then this tag can be used
-# to set the initial width (in pixels) of the frame in which the tree is shown.
-# Minimum value: 0, maximum value: 1500, default value: 250.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-TREEVIEW_WIDTH         = 250
-
-# When the EXT_LINKS_IN_WINDOW option is set to YES doxygen will open links to
-# external symbols imported via tag files in a separate window.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-EXT_LINKS_IN_WINDOW    = NO
-
-# Use this tag to change the font size of LaTeX formulas included as images in
-# the HTML documentation. When you change the font size after a successful
-# doxygen run you need to manually remove any form_*.png images from the HTML
-# output directory to force them to be regenerated.
-# Minimum value: 8, maximum value: 50, default value: 10.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-FORMULA_FONTSIZE       = 10
-
-# Use the FORMULA_TRANPARENT tag to determine whether or not the images
-# generated for formulas are transparent PNGs. Transparent PNGs are not
-# supported properly for IE 6.0, but are supported on all modern browsers.
-#
-# Note that when changing this option you need to delete any form_*.png files in
-# the HTML output directory before the changes have effect.
-# The default value is: YES.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-FORMULA_TRANSPARENT    = YES
-
-# Enable the USE_MATHJAX option to render LaTeX formulas using MathJax (see
-# http://www.mathjax.org) which uses client side Javascript for the rendering
-# instead of using prerendered bitmaps. Use this if you do not have LaTeX
-# installed or if you want to formulas look prettier in the HTML output. When
-# enabled you may also need to install MathJax separately and configure the path
-# to it using the MATHJAX_RELPATH option.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-USE_MATHJAX            = YES
-
-# When MathJax is enabled you can set the default output format to be used for
-# the MathJax output. See the MathJax site (see:
-# http://docs.mathjax.org/en/latest/output.html) for more details.
-# Possible values are: HTML-CSS (which is slower, but has the best
-# compatibility), NativeMML (i.e. MathML) and SVG.
-# The default value is: HTML-CSS.
-# This tag requires that the tag USE_MATHJAX is set to YES.
-
-MATHJAX_FORMAT         = HTML-CSS
-
-# When MathJax is enabled you need to specify the location relative to the HTML
-# output directory using the MATHJAX_RELPATH option. The destination directory
-# should contain the MathJax.js script. For instance, if the mathjax directory
-# is located at the same level as the HTML output directory, then
-# MATHJAX_RELPATH should be ../mathjax. The default value points to the MathJax
-# Content Delivery Network so you can quickly see the result without installing
-# MathJax. However, it is strongly recommended to install a local copy of
-# MathJax from http://www.mathjax.org before deployment.
-# The default value is: http://cdn.mathjax.org/mathjax/latest.
-# This tag requires that the tag USE_MATHJAX is set to YES.
-
-MATHJAX_RELPATH        = http://cdn.mathjax.org/mathjax/latest
-
-# The MATHJAX_EXTENSIONS tag can be used to specify one or more MathJax
-# extension names that should be enabled during MathJax rendering. For example
-# MATHJAX_EXTENSIONS = TeX/AMSmath TeX/AMSsymbols
-# This tag requires that the tag USE_MATHJAX is set to YES.
-
-MATHJAX_EXTENSIONS     =
-
-# The MATHJAX_CODEFILE tag can be used to specify a file with javascript pieces
-# of code that will be used on startup of the MathJax code. See the MathJax site
-# (see: http://docs.mathjax.org/en/latest/output.html) for more details. For an
-# example see the documentation.
-# This tag requires that the tag USE_MATHJAX is set to YES.
-
-MATHJAX_CODEFILE       =
-
-# When the SEARCHENGINE tag is enabled doxygen will generate a search box for
-# the HTML output. The underlying search engine uses javascript and DHTML and
-# should work on any modern browser. Note that when using HTML help
-# (GENERATE_HTMLHELP), Qt help (GENERATE_QHP), or docsets (GENERATE_DOCSET)
-# there is already a search function so this one should typically be disabled.
-# For large projects the javascript based search engine can be slow, then
-# enabling SERVER_BASED_SEARCH may provide a better solution. It is possible to
-# search using the keyboard; to jump to the search box use <access key> + S
-# (what the <access key> is depends on the OS and browser, but it is typically
-# <CTRL>, <ALT>/<option>, or both). Inside the search box use the <cursor down
-# key> to jump into the search results window, the results can be navigated
-# using the <cursor keys>. Press <Enter> to select an item or <escape> to cancel
-# the search. The filter options can be selected when the cursor is inside the
-# search box by pressing <Shift>+<cursor down>. Also here use the <cursor keys>
-# to select a filter and <Enter> or <escape> to activate or cancel the filter
-# option.
-# The default value is: YES.
-# This tag requires that the tag GENERATE_HTML is set to YES.
-
-SEARCHENGINE           = YES
-
-# When the SERVER_BASED_SEARCH tag is enabled the search engine will be
-# implemented using a web server instead of a web client using Javascript. There
-# are two flavours of web server based searching depending on the
-# EXTERNAL_SEARCH setting. When disabled, doxygen will generate a PHP script for
-# searching and an index file used by the script. When EXTERNAL_SEARCH is
-# enabled the indexing and searching needs to be provided by external tools. See
-# the section "External Indexing and Searching" for details.
-# The default value is: NO.
-# This tag requires that the tag SEARCHENGINE is set to YES.
-
-SERVER_BASED_SEARCH    = NO
-
-# When EXTERNAL_SEARCH tag is enabled doxygen will no longer generate the PHP
-# script for searching. Instead the search results are written to an XML file
-# which needs to be processed by an external indexer. Doxygen will invoke an
-# external search engine pointed to by the SEARCHENGINE_URL option to obtain the
-# search results.
-#
-# Doxygen ships with an example indexer ( doxyindexer) and search engine
-# (doxysearch.cgi) which are based on the open source search engine library
-# Xapian (see: http://xapian.org/).
-#
-# See the section "External Indexing and Searching" for details.
-# The default value is: NO.
-# This tag requires that the tag SEARCHENGINE is set to YES.
-
-EXTERNAL_SEARCH        = NO
-
-# The SEARCHENGINE_URL should point to a search engine hosted by a web server
-# which will return the search results when EXTERNAL_SEARCH is enabled.
-#
-# Doxygen ships with an example indexer ( doxyindexer) and search engine
-# (doxysearch.cgi) which are based on the open source search engine library
-# Xapian (see: http://xapian.org/). See the section "External Indexing and
-# Searching" for details.
-# This tag requires that the tag SEARCHENGINE is set to YES.
-
-SEARCHENGINE_URL       =
-
-# When SERVER_BASED_SEARCH and EXTERNAL_SEARCH are both enabled the unindexed
-# search data is written to a file for indexing by an external tool. With the
-# SEARCHDATA_FILE tag the name of this file can be specified.
-# The default file is: searchdata.xml.
-# This tag requires that the tag SEARCHENGINE is set to YES.
-
-SEARCHDATA_FILE        = searchdata.xml
-
-# When SERVER_BASED_SEARCH and EXTERNAL_SEARCH are both enabled the
-# EXTERNAL_SEARCH_ID tag can be used as an identifier for the project. This is
-# useful in combination with EXTRA_SEARCH_MAPPINGS to search through multiple
-# projects and redirect the results back to the right project.
-# This tag requires that the tag SEARCHENGINE is set to YES.
-
-EXTERNAL_SEARCH_ID     =
-
-# The EXTRA_SEARCH_MAPPINGS tag can be used to enable searching through doxygen
-# projects other than the one defined by this configuration file, but that are
-# all added to the same external search index. Each project needs to have a
-# unique id set via EXTERNAL_SEARCH_ID. The search mapping then maps the id of
-# to a relative location where the documentation can be found. The format is:
-# EXTRA_SEARCH_MAPPINGS = tagname1=loc1 tagname2=loc2 ...
-# This tag requires that the tag SEARCHENGINE is set to YES.
-
-EXTRA_SEARCH_MAPPINGS  =
-
-#---------------------------------------------------------------------------
-# Configuration options related to the LaTeX output
-#---------------------------------------------------------------------------
-
-# If the GENERATE_LATEX tag is set to YES doxygen will generate LaTeX output.
-# The default value is: YES.
-
-GENERATE_LATEX         = YES
-
-# The LATEX_OUTPUT tag is used to specify where the LaTeX docs will be put. If a
-# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
-# it.
-# The default directory is: latex.
-# This tag requires that the tag GENERATE_LATEX is set to YES.
-
-LATEX_OUTPUT           = latex
-
-# The LATEX_CMD_NAME tag can be used to specify the LaTeX command name to be
-# invoked.
-#
-# Note that when enabling USE_PDFLATEX this option is only used for generating
-# bitmaps for formulas in the HTML output, but not in the Makefile that is
-# written to the output directory.
-# The default file is: latex.
-# This tag requires that the tag GENERATE_LATEX is set to YES.
-
-LATEX_CMD_NAME         = latex
-
-# The MAKEINDEX_CMD_NAME tag can be used to specify the command name to generate
-# index for LaTeX.
-# The default file is: makeindex.
-# This tag requires that the tag GENERATE_LATEX is set to YES.
-
-MAKEINDEX_CMD_NAME     = makeindex
-
-# If the COMPACT_LATEX tag is set to YES doxygen generates more compact LaTeX
-# documents. This may be useful for small projects and may help to save some
-# trees in general.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_LATEX is set to YES.
-
-COMPACT_LATEX          = NO
-
-# The PAPER_TYPE tag can be used to set the paper type that is used by the
-# printer.
-# Possible values are: a4 (210 x 297 mm), letter (8.5 x 11 inches), legal (8.5 x
-# 14 inches) and executive (7.25 x 10.5 inches).
-# The default value is: a4.
-# This tag requires that the tag GENERATE_LATEX is set to YES.
-
-PAPER_TYPE             = a4
-
-# The EXTRA_PACKAGES tag can be used to specify one or more LaTeX package names
-# that should be included in the LaTeX output. To get the times font for
-# instance you can specify
-# EXTRA_PACKAGES=times
-# If left blank no extra packages will be included.
-# This tag requires that the tag GENERATE_LATEX is set to YES.
-
-EXTRA_PACKAGES         =
-
-# The LATEX_HEADER tag can be used to specify a personal LaTeX header for the
-# generated LaTeX document. The header should contain everything until the first
-# chapter. If it is left blank doxygen will generate a standard header. See
-# section "Doxygen usage" for information on how to let doxygen write the
-# default header to a separate file.
-#
-# Note: Only use a user-defined header if you know what you are doing! The
-# following commands have a special meaning inside the header: $title,
-# $datetime, $date, $doxygenversion, $projectname, $projectnumber. Doxygen will
-# replace them by respectively the title of the page, the current date and time,
-# only the current date, the version number of doxygen, the project name (see
-# PROJECT_NAME), or the project number (see PROJECT_NUMBER).
-# This tag requires that the tag GENERATE_LATEX is set to YES.
-
-LATEX_HEADER           =
-
-# The LATEX_FOOTER tag can be used to specify a personal LaTeX footer for the
-# generated LaTeX document. The footer should contain everything after the last
-# chapter. If it is left blank doxygen will generate a standard footer.
-#
-# Note: Only use a user-defined footer if you know what you are doing!
-# This tag requires that the tag GENERATE_LATEX is set to YES.
-
-LATEX_FOOTER           =
-
-# The LATEX_EXTRA_FILES tag can be used to specify one or more extra images or
-# other source files which should be copied to the LATEX_OUTPUT output
-# directory. Note that the files will be copied as-is; there are no commands or
-# markers available.
-# This tag requires that the tag GENERATE_LATEX is set to YES.
-
-LATEX_EXTRA_FILES      =
-
-# If the PDF_HYPERLINKS tag is set to YES, the LaTeX that is generated is
-# prepared for conversion to PDF (using ps2pdf or pdflatex). The PDF file will
-# contain links (just like the HTML output) instead of page references. This
-# makes the output suitable for online browsing using a PDF viewer.
-# The default value is: YES.
-# This tag requires that the tag GENERATE_LATEX is set to YES.
-
-PDF_HYPERLINKS         = YES
-
-# If the LATEX_PDFLATEX tag is set to YES, doxygen will use pdflatex to generate
-# the PDF file directly from the LaTeX files. Set this option to YES to get a
-# higher quality PDF documentation.
-# The default value is: YES.
-# This tag requires that the tag GENERATE_LATEX is set to YES.
-
-USE_PDFLATEX           = YES
-
-# If the LATEX_BATCHMODE tag is set to YES, doxygen will add the \batchmode
-# command to the generated LaTeX files. This will instruct LaTeX to keep running
-# if errors occur, instead of asking the user for help. This option is also used
-# when generating formulas in HTML.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_LATEX is set to YES.
-
-LATEX_BATCHMODE        = NO
-
-# If the LATEX_HIDE_INDICES tag is set to YES then doxygen will not include the
-# index chapters (such as File Index, Compound Index, etc.) in the output.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_LATEX is set to YES.
-
-LATEX_HIDE_INDICES     = NO
-
-# If the LATEX_SOURCE_CODE tag is set to YES then doxygen will include source
-# code with syntax highlighting in the LaTeX output.
-#
-# Note that which sources are shown also depends on other settings such as
-# SOURCE_BROWSER.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_LATEX is set to YES.
-
-LATEX_SOURCE_CODE      = NO
-
-# The LATEX_BIB_STYLE tag can be used to specify the style to use for the
-# bibliography, e.g. plainnat, or ieeetr. See
-# http://en.wikipedia.org/wiki/BibTeX and \cite for more info.
-# The default value is: plain.
-# This tag requires that the tag GENERATE_LATEX is set to YES.
-
-LATEX_BIB_STYLE        = plain
-
-#---------------------------------------------------------------------------
-# Configuration options related to the RTF output
-#---------------------------------------------------------------------------
-
-# If the GENERATE_RTF tag is set to YES doxygen will generate RTF output. The
-# RTF output is optimized for Word 97 and may not look too pretty with other RTF
-# readers/editors.
-# The default value is: NO.
-
-GENERATE_RTF           = NO
-
-# The RTF_OUTPUT tag is used to specify where the RTF docs will be put. If a
-# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
-# it.
-# The default directory is: rtf.
-# This tag requires that the tag GENERATE_RTF is set to YES.
-
-RTF_OUTPUT             = rtf
-
-# If the COMPACT_RTF tag is set to YES doxygen generates more compact RTF
-# documents. This may be useful for small projects and may help to save some
-# trees in general.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_RTF is set to YES.
-
-COMPACT_RTF            = NO
-
-# If the RTF_HYPERLINKS tag is set to YES, the RTF that is generated will
-# contain hyperlink fields. The RTF file will contain links (just like the HTML
-# output) instead of page references. This makes the output suitable for online
-# browsing using Word or some other Word compatible readers that support those
-# fields.
-#
-# Note: WordPad (write) and others do not support links.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_RTF is set to YES.
-
-RTF_HYPERLINKS         = NO
-
-# Load stylesheet definitions from file. Syntax is similar to doxygen's config
-# file, i.e. a series of assignments. You only have to provide replacements,
-# missing definitions are set to their default value.
-#
-# See also section "Doxygen usage" for information on how to generate the
-# default style sheet that doxygen normally uses.
-# This tag requires that the tag GENERATE_RTF is set to YES.
-
-RTF_STYLESHEET_FILE    =
-
-# Set optional variables used in the generation of an RTF document. Syntax is
-# similar to doxygen's config file. A template extensions file can be generated
-# using doxygen -e rtf extensionFile.
-# This tag requires that the tag GENERATE_RTF is set to YES.
-
-RTF_EXTENSIONS_FILE    =
-
-#---------------------------------------------------------------------------
-# Configuration options related to the man page output
-#---------------------------------------------------------------------------
-
-# If the GENERATE_MAN tag is set to YES doxygen will generate man pages for
-# classes and files.
-# The default value is: NO.
-
-GENERATE_MAN           = NO
-
-# The MAN_OUTPUT tag is used to specify where the man pages will be put. If a
-# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
-# it. A directory man3 will be created inside the directory specified by
-# MAN_OUTPUT.
-# The default directory is: man.
-# This tag requires that the tag GENERATE_MAN is set to YES.
-
-MAN_OUTPUT             = man
-
-# The MAN_EXTENSION tag determines the extension that is added to the generated
-# man pages. In case the manual section does not start with a number, the number
-# 3 is prepended. The dot (.) at the beginning of the MAN_EXTENSION tag is
-# optional.
-# The default value is: .3.
-# This tag requires that the tag GENERATE_MAN is set to YES.
-
-MAN_EXTENSION          = .3
-
-# If the MAN_LINKS tag is set to YES and doxygen generates man output, then it
-# will generate one additional man file for each entity documented in the real
-# man page(s). These additional files only source the real man page, but without
-# them the man command would be unable to find the correct page.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_MAN is set to YES.
-
-MAN_LINKS              = NO
-
-#---------------------------------------------------------------------------
-# Configuration options related to the XML output
-#---------------------------------------------------------------------------
-
-# If the GENERATE_XML tag is set to YES doxygen will generate an XML file that
-# captures the structure of the code including all documentation.
-# The default value is: NO.
-
-GENERATE_XML           = NO
-
-# The XML_OUTPUT tag is used to specify where the XML pages will be put. If a
-# relative path is entered the value of OUTPUT_DIRECTORY will be put in front of
-# it.
-# The default directory is: xml.
-# This tag requires that the tag GENERATE_XML is set to YES.
-
-XML_OUTPUT             = xml
-
-# The XML_SCHEMA tag can be used to specify a XML schema, which can be used by a
-# validating XML parser to check the syntax of the XML files.
-# This tag requires that the tag GENERATE_XML is set to YES.
-
-XML_SCHEMA             =
-
-# The XML_DTD tag can be used to specify a XML DTD, which can be used by a
-# validating XML parser to check the syntax of the XML files.
-# This tag requires that the tag GENERATE_XML is set to YES.
-
-XML_DTD                =
-
-# If the XML_PROGRAMLISTING tag is set to YES doxygen will dump the program
-# listings (including syntax highlighting and cross-referencing information) to
-# the XML output. Note that enabling this will significantly increase the size
-# of the XML output.
-# The default value is: YES.
-# This tag requires that the tag GENERATE_XML is set to YES.
-
-XML_PROGRAMLISTING     = YES
-
-#---------------------------------------------------------------------------
-# Configuration options related to the DOCBOOK output
-#---------------------------------------------------------------------------
-
-# If the GENERATE_DOCBOOK tag is set to YES doxygen will generate Docbook files
-# that can be used to generate PDF.
-# The default value is: NO.
-
-GENERATE_DOCBOOK       = NO
-
-# The DOCBOOK_OUTPUT tag is used to specify where the Docbook pages will be put.
-# If a relative path is entered the value of OUTPUT_DIRECTORY will be put in
-# front of it.
-# The default directory is: docbook.
-# This tag requires that the tag GENERATE_DOCBOOK is set to YES.
-
-DOCBOOK_OUTPUT         = docbook
-
-#---------------------------------------------------------------------------
-# Configuration options for the AutoGen Definitions output
-#---------------------------------------------------------------------------
-
-# If the GENERATE_AUTOGEN_DEF tag is set to YES doxygen will generate an AutoGen
-# Definitions (see http://autogen.sf.net) file that captures the structure of
-# the code including all documentation. Note that this feature is still
-# experimental and incomplete at the moment.
-# The default value is: NO.
-
-GENERATE_AUTOGEN_DEF   = NO
-
-#---------------------------------------------------------------------------
-# Configuration options related to the Perl module output
-#---------------------------------------------------------------------------
-
-# If the GENERATE_PERLMOD tag is set to YES doxygen will generate a Perl module
-# file that captures the structure of the code including all documentation.
-#
-# Note that this feature is still experimental and incomplete at the moment.
-# The default value is: NO.
-
-GENERATE_PERLMOD       = NO
-
-# If the PERLMOD_LATEX tag is set to YES doxygen will generate the necessary
-# Makefile rules, Perl scripts and LaTeX code to be able to generate PDF and DVI
-# output from the Perl module output.
-# The default value is: NO.
-# This tag requires that the tag GENERATE_PERLMOD is set to YES.
-
-PERLMOD_LATEX          = NO
-
-# If the PERLMOD_PRETTY tag is set to YES the Perl module output will be nicely
-# formatted so it can be parsed by a human reader. This is useful if you want to
-# understand what is going on. On the other hand, if this tag is set to NO the
-# size of the Perl module output will be much smaller and Perl will parse it
-# just the same.
-# The default value is: YES.
-# This tag requires that the tag GENERATE_PERLMOD is set to YES.
-
-PERLMOD_PRETTY         = YES
-
-# The names of the make variables in the generated doxyrules.make file are
-# prefixed with the string contained in PERLMOD_MAKEVAR_PREFIX. This is useful
-# so different doxyrules.make files included by the same Makefile don't
-# overwrite each other's variables.
-# This tag requires that the tag GENERATE_PERLMOD is set to YES.
-
-PERLMOD_MAKEVAR_PREFIX =
-
-#---------------------------------------------------------------------------
-# Configuration options related to the preprocessor
-#---------------------------------------------------------------------------
-
-# If the ENABLE_PREPROCESSING tag is set to YES doxygen will evaluate all
-# C-preprocessor directives found in the sources and include files.
-# The default value is: YES.
-
-ENABLE_PREPROCESSING   = YES
-
-# If the MACRO_EXPANSION tag is set to YES doxygen will expand all macro names
-# in the source code. If set to NO only conditional compilation will be
-# performed. Macro expansion can be done in a controlled way by setting
-# EXPAND_ONLY_PREDEF to YES.
-# The default value is: NO.
-# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
-
-MACRO_EXPANSION        = NO
-
-# If the EXPAND_ONLY_PREDEF and MACRO_EXPANSION tags are both set to YES then
-# the macro expansion is limited to the macros specified with the PREDEFINED and
-# EXPAND_AS_DEFINED tags.
-# The default value is: NO.
-# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
-
-EXPAND_ONLY_PREDEF     = NO
-
-# If the SEARCH_INCLUDES tag is set to YES the includes files in the
-# INCLUDE_PATH will be searched if a #include is found.
-# The default value is: YES.
-# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
-
-SEARCH_INCLUDES        = YES
-
-# The INCLUDE_PATH tag can be used to specify one or more directories that
-# contain include files that are not input files but should be processed by the
-# preprocessor.
-# This tag requires that the tag SEARCH_INCLUDES is set to YES.
-
-INCLUDE_PATH           =
-
-# You can use the INCLUDE_FILE_PATTERNS tag to specify one or more wildcard
-# patterns (like *.h and *.hpp) to filter out the header-files in the
-# directories. If left blank, the patterns specified with FILE_PATTERNS will be
-# used.
-# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
-
-INCLUDE_FILE_PATTERNS  =
-
-# The PREDEFINED tag can be used to specify one or more macro names that are
-# defined before the preprocessor is started (similar to the -D option of e.g.
-# gcc). The argument of the tag is a list of macros of the form: name or
-# name=definition (no spaces). If the definition and the "=" are omitted, "=1"
-# is assumed. To prevent a macro definition from being undefined via #undef or
-# recursively expanded use the := operator instead of the = operator.
-# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
-
-PREDEFINED             =
-
-# If the MACRO_EXPANSION and EXPAND_ONLY_PREDEF tags are set to YES then this
-# tag can be used to specify a list of macro names that should be expanded. The
-# macro definition that is found in the sources will be used. Use the PREDEFINED
-# tag if you want to use a different macro definition that overrules the
-# definition found in the source code.
-# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
-
-EXPAND_AS_DEFINED      =
-
-# If the SKIP_FUNCTION_MACROS tag is set to YES then doxygen's preprocessor will
-# remove all refrences to function-like macros that are alone on a line, have an
-# all uppercase name, and do not end with a semicolon. Such function macros are
-# typically used for boiler-plate code, and will confuse the parser if not
-# removed.
-# The default value is: YES.
-# This tag requires that the tag ENABLE_PREPROCESSING is set to YES.
-
-SKIP_FUNCTION_MACROS   = YES
-
-#---------------------------------------------------------------------------
-# Configuration options related to external references
-#---------------------------------------------------------------------------
-
-# The TAGFILES tag can be used to specify one or more tag files. For each tag
-# file the location of the external documentation should be added. The format of
-# a tag file without this location is as follows:
-# TAGFILES = file1 file2 ...
-# Adding location for the tag files is done as follows:
-# TAGFILES = file1=loc1 "file2 = loc2" ...
-# where loc1 and loc2 can be relative or absolute paths or URLs. See the
-# section "Linking to external documentation" for more information about the use
-# of tag files.
-# Note: Each tag file must have an unique name (where the name does NOT include
-# the path). If a tag file is not located in the directory in which doxygen is
-# run, you must also specify the path to the tagfile here.
-
-TAGFILES               =
-
-# When a file name is specified after GENERATE_TAGFILE, doxygen will create a
-# tag file that is based on the input files it reads. See section "Linking to
-# external documentation" for more information about the usage of tag files.
-
-GENERATE_TAGFILE       =
-
-# If the ALLEXTERNALS tag is set to YES all external class will be listed in the
-# class index. If set to NO only the inherited external classes will be listed.
-# The default value is: NO.
-
-ALLEXTERNALS           = NO
-
-# If the EXTERNAL_GROUPS tag is set to YES all external groups will be listed in
-# the modules index. If set to NO, only the current project's groups will be
-# listed.
-# The default value is: YES.
-
-EXTERNAL_GROUPS        = YES
-
-# If the EXTERNAL_PAGES tag is set to YES all external pages will be listed in
-# the related pages index. If set to NO, only the current project's pages will
-# be listed.
-# The default value is: YES.
-
-EXTERNAL_PAGES         = YES
-
-# The PERL_PATH should be the absolute path and name of the perl script
-# interpreter (i.e. the result of 'which perl').
-# The default file (with absolute path) is: /usr/bin/perl.
-
-PERL_PATH              = /usr/bin/perl
-
-#---------------------------------------------------------------------------
-# Configuration options related to the dot tool
-#---------------------------------------------------------------------------
-
-# If the CLASS_DIAGRAMS tag is set to YES doxygen will generate a class diagram
-# (in HTML and LaTeX) for classes with base or super classes. Setting the tag to
-# NO turns the diagrams off. Note that this option also works with HAVE_DOT
-# disabled, but it is recommended to install and use dot, since it yields more
-# powerful graphs.
-# The default value is: YES.
-
-CLASS_DIAGRAMS         = YES
-
-# You can define message sequence charts within doxygen comments using the \msc
-# command. Doxygen will then run the mscgen tool (see:
-# http://www.mcternan.me.uk/mscgen/)) to produce the chart and insert it in the
-# documentation. The MSCGEN_PATH tag allows you to specify the directory where
-# the mscgen tool resides. If left empty the tool is assumed to be found in the
-# default search path.
-
-MSCGEN_PATH            =
-
-# You can include diagrams made with dia in doxygen documentation. Doxygen will
-# then run dia to produce the diagram and insert it in the documentation. The
-# DIA_PATH tag allows you to specify the directory where the dia binary resides.
-# If left empty dia is assumed to be found in the default search path.
-
-DIA_PATH               =
-
-# If set to YES, the inheritance and collaboration graphs will hide inheritance
-# and usage relations if the target is undocumented or is not a class.
-# The default value is: YES.
-
-HIDE_UNDOC_RELATIONS   = YES
-
-# If you set the HAVE_DOT tag to YES then doxygen will assume the dot tool is
-# available from the path. This tool is part of Graphviz (see:
-# http://www.graphviz.org/), a graph visualization toolkit from AT&T and Lucent
-# Bell Labs. The other options in this section have no effect if this option is
-# set to NO
-# The default value is: NO.
-
-HAVE_DOT               = YES
-
-# The DOT_NUM_THREADS specifies the number of dot invocations doxygen is allowed
-# to run in parallel. When set to 0 doxygen will base this on the number of
-# processors available in the system. You can set it explicitly to a value
-# larger than 0 to get control over the balance between CPU load and processing
-# speed.
-# Minimum value: 0, maximum value: 32, default value: 0.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-DOT_NUM_THREADS        = 2
-
-# When you want a differently looking font n the dot files that doxygen
-# generates you can specify the font name using DOT_FONTNAME. You need to make
-# sure dot is able to find the font, which can be done by putting it in a
-# standard location or by setting the DOTFONTPATH environment variable or by
-# setting DOT_FONTPATH to the directory containing the font.
-# The default value is: Helvetica.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-DOT_FONTNAME           = Helvetica
-
-# The DOT_FONTSIZE tag can be used to set the size (in points) of the font of
-# dot graphs.
-# Minimum value: 4, maximum value: 24, default value: 10.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-DOT_FONTSIZE           = 10
-
-# By default doxygen will tell dot to use the default font as specified with
-# DOT_FONTNAME. If you specify a different font using DOT_FONTNAME you can set
-# the path where dot can find it using this tag.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-DOT_FONTPATH           =
-
-# If the CLASS_GRAPH tag is set to YES then doxygen will generate a graph for
-# each documented class showing the direct and indirect inheritance relations.
-# Setting this tag to YES will force the CLASS_DIAGRAMS tag to NO.
-# The default value is: YES.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-CLASS_GRAPH            = YES
-
-# If the COLLABORATION_GRAPH tag is set to YES then doxygen will generate a
-# graph for each documented class showing the direct and indirect implementation
-# dependencies (inheritance, containment, and class references variables) of the
-# class with other documented classes.
-# The default value is: YES.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-COLLABORATION_GRAPH    = YES
-
-# If the GROUP_GRAPHS tag is set to YES then doxygen will generate a graph for
-# groups, showing the direct groups dependencies.
-# The default value is: YES.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-GROUP_GRAPHS           = YES
-
-# If the UML_LOOK tag is set to YES doxygen will generate inheritance and
-# collaboration diagrams in a style similar to the OMG's Unified Modeling
-# Language.
-# The default value is: NO.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-UML_LOOK               = NO
-
-# If the UML_LOOK tag is enabled, the fields and methods are shown inside the
-# class node. If there are many fields or methods and many nodes the graph may
-# become too big to be useful. The UML_LIMIT_NUM_FIELDS threshold limits the
-# number of items for each type to make the size more manageable. Set this to 0
-# for no limit. Note that the threshold may be exceeded by 50% before the limit
-# is enforced. So when you set the threshold to 10, up to 15 fields may appear,
-# but if the number exceeds 15, the total amount of fields shown is limited to
-# 10.
-# Minimum value: 0, maximum value: 100, default value: 10.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-UML_LIMIT_NUM_FIELDS   = 10
-
-# If the TEMPLATE_RELATIONS tag is set to YES then the inheritance and
-# collaboration graphs will show the relations between templates and their
-# instances.
-# The default value is: NO.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-TEMPLATE_RELATIONS     = NO
-
-# If the INCLUDE_GRAPH, ENABLE_PREPROCESSING and SEARCH_INCLUDES tags are set to
-# YES then doxygen will generate a graph for each documented file showing the
-# direct and indirect include dependencies of the file with other documented
-# files.
-# The default value is: YES.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-INCLUDE_GRAPH          = YES
-
-# If the INCLUDED_BY_GRAPH, ENABLE_PREPROCESSING and SEARCH_INCLUDES tags are
-# set to YES then doxygen will generate a graph for each documented file showing
-# the direct and indirect include dependencies of the file with other documented
-# files.
-# The default value is: YES.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-INCLUDED_BY_GRAPH      = YES
-
-# If the CALL_GRAPH tag is set to YES then doxygen will generate a call
-# dependency graph for every global function or class method.
-#
-# Note that enabling this option will significantly increase the time of a run.
-# So in most cases it will be better to enable call graphs for selected
-# functions only using the \callgraph command.
-# The default value is: NO.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-CALL_GRAPH             = NO
-
-# If the CALLER_GRAPH tag is set to YES then doxygen will generate a caller
-# dependency graph for every global function or class method.
-#
-# Note that enabling this option will significantly increase the time of a run.
-# So in most cases it will be better to enable caller graphs for selected
-# functions only using the \callergraph command.
-# The default value is: NO.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-CALLER_GRAPH           = NO
-
-# If the GRAPHICAL_HIERARCHY tag is set to YES then doxygen will graphical
-# hierarchy of all classes instead of a textual one.
-# The default value is: YES.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-GRAPHICAL_HIERARCHY    = YES
-
-# If the DIRECTORY_GRAPH tag is set to YES then doxygen will show the
-# dependencies a directory has on other directories in a graphical way. The
-# dependency relations are determined by the #include relations between the
-# files in the directories.
-# The default value is: YES.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-DIRECTORY_GRAPH        = YES
-
-# The DOT_IMAGE_FORMAT tag can be used to set the image format of the images
-# generated by dot.
-# Note: If you choose svg you need to set HTML_FILE_EXTENSION to xhtml in order
-# to make the SVG files visible in IE 9+ (other browsers do not have this
-# requirement).
-# Possible values are: png, jpg, gif and svg.
-# The default value is: png.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-DOT_IMAGE_FORMAT       = png
-
-# If DOT_IMAGE_FORMAT is set to svg, then this option can be set to YES to
-# enable generation of interactive SVG images that allow zooming and panning.
-#
-# Note that this requires a modern browser other than Internet Explorer. Tested
-# and working are Firefox, Chrome, Safari, and Opera.
-# Note: For IE 9+ you need to set HTML_FILE_EXTENSION to xhtml in order to make
-# the SVG files visible. Older versions of IE do not have SVG support.
-# The default value is: NO.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-INTERACTIVE_SVG        = NO
-
-# The DOT_PATH tag can be used to specify the path where the dot tool can be
-# found. If left blank, it is assumed the dot tool can be found in the path.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-DOT_PATH               =
-
-# The DOTFILE_DIRS tag can be used to specify one or more directories that
-# contain dot files that are included in the documentation (see the \dotfile
-# command).
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-DOTFILE_DIRS           =
-
-# The MSCFILE_DIRS tag can be used to specify one or more directories that
-# contain msc files that are included in the documentation (see the \mscfile
-# command).
-
-MSCFILE_DIRS           =
-
-# The DIAFILE_DIRS tag can be used to specify one or more directories that
-# contain dia files that are included in the documentation (see the \diafile
-# command).
-
-DIAFILE_DIRS           =
-
-# The DOT_GRAPH_MAX_NODES tag can be used to set the maximum number of nodes
-# that will be shown in the graph. If the number of nodes in a graph becomes
-# larger than this value, doxygen will truncate the graph, which is visualized
-# by representing a node as a red box. Note that doxygen if the number of direct
-# children of the root node in a graph is already larger than
-# DOT_GRAPH_MAX_NODES then the graph will not be shown at all. Also note that
-# the size of a graph can be further restricted by MAX_DOT_GRAPH_DEPTH.
-# Minimum value: 0, maximum value: 10000, default value: 50.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-DOT_GRAPH_MAX_NODES    = 50
-
-# The MAX_DOT_GRAPH_DEPTH tag can be used to set the maximum depth of the graphs
-# generated by dot. A depth value of 3 means that only nodes reachable from the
-# root by following a path via at most 3 edges will be shown. Nodes that lay
-# further from the root node will be omitted. Note that setting this option to 1
-# or 2 may greatly reduce the computation time needed for large code bases. Also
-# note that the size of a graph can be further restricted by
-# DOT_GRAPH_MAX_NODES. Using a depth of 0 means no depth restriction.
-# Minimum value: 0, maximum value: 1000, default value: 0.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-MAX_DOT_GRAPH_DEPTH    = 0
-
-# Set the DOT_TRANSPARENT tag to YES to generate images with a transparent
-# background. This is disabled by default, because dot on Windows does not seem
-# to support this out of the box.
-#
-# Warning: Depending on the platform used, enabling this option may lead to
-# badly anti-aliased labels on the edges of a graph (i.e. they become hard to
-# read).
-# The default value is: NO.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-DOT_TRANSPARENT        = YES
-
-# Set the DOT_MULTI_TARGETS tag to YES allow dot to generate multiple output
-# files in one run (i.e. multiple -o and -T options on the command line). This
-# makes dot run faster, but since only newer versions of dot (>1.8.10) support
-# this, this feature is disabled by default.
-# The default value is: NO.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-DOT_MULTI_TARGETS      = YES
-
-# If the GENERATE_LEGEND tag is set to YES doxygen will generate a legend page
-# explaining the meaning of the various boxes and arrows in the dot generated
-# graphs.
-# The default value is: YES.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-GENERATE_LEGEND        = YES
-
-# If the DOT_CLEANUP tag is set to YES doxygen will remove the intermediate dot
-# files that are used to generate the various graphs.
-# The default value is: YES.
-# This tag requires that the tag HAVE_DOT is set to YES.
-
-DOT_CLEANUP            = YES
diff --git a/lib/simd/Grid_avx.h b/lib/simd/Grid_avx.h
index 3c8ebc61..ee4d03d5 100644
--- a/lib/simd/Grid_avx.h
+++ b/lib/simd/Grid_avx.h
@@ -4,7 +4,7 @@
 
   Using intrinsics
 */
-// Time-stamp: <2015-05-27 12:07:15 neo>
+// Time-stamp: <2015-05-29 14:13:30 neo>
 //----------------------------------------------------------------------
 
 #include <immintrin.h>
@@ -261,13 +261,7 @@ namespace Optimization {
     }
     // Complex double
     inline __m256d operator()(__m256d in){
-      return _mm256_xor_pd(_mm256_addsub_pd(_mm256_setzero_pd(),in), _mm256_set1_pd(-0.f));//untested
-      /*
-	// original 
-	//      addsubps 0, inv=>0+in.v[3] 0-in.v[2], 0+in.v[1], 0-in.v[0], ...
-	__m256d tmp = _mm256_addsub_pd(_mm256_setzero_pd(),_mm256_shuffle_pd(in,in,0x5));
-	return _mm256_shuffle_pd(tmp,tmp,0x5);
-      */
+      return _mm256_xor_pd(_mm256_addsub_pd(_mm256_setzero_pd(),in), _mm256_set1_pd(-0.f));
     }
     // do not define for integer input
   };
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index e389f3c2..aa1abb47 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -2,7 +2,7 @@
 /*! @file Grid_vector_types.h
   @brief Defines templated class Grid_simd to deal with inner vector types
 */
-// Time-stamp: <2015-05-27 12:04:06 neo>
+// Time-stamp: <2015-05-29 14:19:48 neo>
 //---------------------------------------------------------------------------
 #ifndef GRID_VECTOR_TYPES
 #define GRID_VECTOR_TYPES
@@ -55,7 +55,6 @@ namespace Grid {
   // general forms to allow for vsplat syntax
   // need explicit declaration of types when used since
   // clang cannot automatically determine the output type sometimes
-  // use decltype?
   template < class Out, class Input1, class Input2, class Operation > 
     Out binary(Input1 src_1, Input2 src_2, Operation op){
     return op(src_1, src_2);

From 201d6d097dbd9102a0cdf2b68f72dca60a7f4c16 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 4 Jun 2015 09:41:16 +0100
Subject: [PATCH 270/429] Allow traceIndex on a different index to distribute
 replicated across a vector index

---
 lib/tensors/Tensor_trace.h | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/lib/tensors/Tensor_trace.h b/lib/tensors/Tensor_trace.h
index f6778496..9c85080e 100644
--- a/lib/tensors/Tensor_trace.h
+++ b/lib/tensors/Tensor_trace.h
@@ -72,5 +72,18 @@ auto traceIndex(const iMatrix<vtype,N> &arg) ->  iMatrix<decltype(traceIndex<Lev
   return ret;
 }
 
+// Allow to recurse if vector, but never terminate on a vector
+// trace of a different index can distribute across the vector index in a replicated way
+// but we do not trace a vector index.
+template<int Level,class vtype,int N,typename std::enable_if< iVector<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+auto traceIndex(const iVector<vtype,N> &arg) ->  iVector<decltype(traceIndex<Level>(arg._internal[0])),N> 
+{
+  iVector<decltype(traceIndex<Level>(arg._internal[0])),N> ret;
+  for(int i=0;i<N;i++){
+    ret._internal[i] = traceIndex<Level>(arg._internal[i]);
+  }
+  return ret;
+}
+
 }
 #endif

From 7fc54fc90403ed696edaaf22a2a4fe34d22414a9 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Thu, 4 Jun 2015 18:29:55 +0900
Subject: [PATCH 271/429] Added support for Ta to Lattice types

---
 lib/lattice/Lattice_reality.h | 11 +++++++++++
 lib/tensors/Tensor_Ta.h       | 17 +++++++++++------
 tests/Test_main.cc            | 13 +++++++------
 3 files changed, 29 insertions(+), 12 deletions(-)

diff --git a/lib/lattice/Lattice_reality.h b/lib/lattice/Lattice_reality.h
index 2f52ed22..4ad6fb7b 100644
--- a/lib/lattice/Lattice_reality.h
+++ b/lib/lattice/Lattice_reality.h
@@ -48,5 +48,16 @@ PARALLEL_FOR_LOOP
     }
 
 
+    template<class vobj> inline auto Ta(const Lattice<vobj> &z) -> Lattice<decltype(Ta(z._odata[0]))>
+    {
+      Lattice<decltype(Ta(z._odata[0]))> ret(z._grid);
+PARALLEL_FOR_LOOP
+        for(int ss=0;ss<z._grid->oSites();ss++){
+            ret._odata[ss] = Ta(z._odata[ss]);
+        }
+      return ret;
+    }
+
+
 }
 #endif
diff --git a/lib/tensors/Tensor_Ta.h b/lib/tensors/Tensor_Ta.h
index ef4b86e1..3a903364 100644
--- a/lib/tensors/Tensor_Ta.h
+++ b/lib/tensors/Tensor_Ta.h
@@ -5,10 +5,10 @@ namespace Grid {
   /////////////////////////////////////////////// 
   // Ta function for scalar, vector, matrix
   /////////////////////////////////////////////// 
-  /* inline ComplexF Ta( const ComplexF &arg){    return arg;} */
-  /* inline ComplexD Ta( const ComplexD &arg){    return arg;} */
-  /* inline RealF Ta( const RealF &arg){    return arg;} */
-  /* inline RealD Ta( const RealD &arg){    return arg;} */
+  inline ComplexF Ta( const ComplexF &arg){    return arg;}
+  inline ComplexD Ta( const ComplexD &arg){    return arg;}
+  inline RealF Ta( const RealF &arg){    return arg;}
+  inline RealD Ta( const RealD &arg){    return arg;}
 
 
   template<class vtype> inline iScalar<vtype> Ta(const iScalar<vtype>&r)
@@ -28,8 +28,13 @@ namespace Grid {
   template<class vtype,int N> inline iMatrix<vtype,N> Ta(const iMatrix<vtype,N> &arg)
     {
       iMatrix<vtype,N> ret(arg);
-      vtype factor = (1/(double)N);
-      ret = (ret - adj(arg))*0.5;
+      double factor = (1/(double)N);
+      for(int c1=0;c1<N;c1++){
+	for(int c2=0;c2<N;c2++){
+	  ret._internal[c1][c2]= (ret._internal[c1][c2] - adj(arg._internal[c2][c1]));
+	  ret._internal[c1][c2] *= 0.5;
+	}}
+      //ret = (ret - adj(arg))*0.5;
       ret -= trace(ret)*factor;
       return ret;
     }
diff --git a/tests/Test_main.cc b/tests/Test_main.cc
index 724e1563..cef2d848 100644
--- a/tests/Test_main.cc
+++ b/tests/Test_main.cc
@@ -123,7 +123,9 @@ int main (int argc, char ** argv)
     cMat = outerProduct(cVec,cVec);
     scalar = localInnerProduct(cVec,cVec);
 
- 
+    cMat = Ta(cMat);  //traceless antihermitian
+
+
     scalar += scalar;
     scalar -= scalar;
     scalar *= scalar;
@@ -214,13 +216,12 @@ int main (int argc, char ** argv)
     scm=transposeIndex<1>(scm);
     
    
-    random(SerialRNG, cm);
-    std::cout << cm << std::endl;
+    //random(SerialRNG, cm);
+    //std::cout << cm << std::endl;
 
     cm = Ta(cm);
-    //cm = adj(cm);
-    TComplex tracecm= trace(cm);      
-    std::cout << cm << "  "<< tracecm << std::endl;
+    //TComplex tracecm= trace(cm);      
+    //std::cout << cm << "  "<< tracecm << std::endl;
 
 
 //    Foo = Foo+scalar; // LatticeColourMatrix+Scalar

From 769b27b7f2123ae8ef8809697fb82c1bae235a40 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Thu, 4 Jun 2015 19:01:43 +0900
Subject: [PATCH 272/429] Corrected small compilation bug in traceIndex for
 iVectors

---
 lib/tensors/Tensor_trace.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/tensors/Tensor_trace.h b/lib/tensors/Tensor_trace.h
index 9c85080e..749b292e 100644
--- a/lib/tensors/Tensor_trace.h
+++ b/lib/tensors/Tensor_trace.h
@@ -75,7 +75,7 @@ auto traceIndex(const iMatrix<vtype,N> &arg) ->  iMatrix<decltype(traceIndex<Lev
 // Allow to recurse if vector, but never terminate on a vector
 // trace of a different index can distribute across the vector index in a replicated way
 // but we do not trace a vector index.
-template<int Level,class vtype,int N,typename std::enable_if< iVector<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+ template<int Level,class vtype,int N,typename std::enable_if< iVector<vtype, N>::TensorLevel != Level >::type * =nullptr> inline 
 auto traceIndex(const iVector<vtype,N> &arg) ->  iVector<decltype(traceIndex<Level>(arg._internal[0])),N> 
 {
   iVector<decltype(traceIndex<Level>(arg._internal[0])),N> ret;

From 51ec6b722e2d7ad5fadfbb23b5a792fe06856c82 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 4 Jun 2015 12:01:51 +0100
Subject: [PATCH 273/429] Mistaken commit that prevented compile ; fixing

---
 lib/tensors/Tensor_trace.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/tensors/Tensor_trace.h b/lib/tensors/Tensor_trace.h
index 9c85080e..961a1339 100644
--- a/lib/tensors/Tensor_trace.h
+++ b/lib/tensors/Tensor_trace.h
@@ -75,7 +75,7 @@ auto traceIndex(const iMatrix<vtype,N> &arg) ->  iMatrix<decltype(traceIndex<Lev
 // Allow to recurse if vector, but never terminate on a vector
 // trace of a different index can distribute across the vector index in a replicated way
 // but we do not trace a vector index.
-template<int Level,class vtype,int N,typename std::enable_if< iVector<vtype>::TensorLevel != Level >::type * =nullptr> inline 
+template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
 auto traceIndex(const iVector<vtype,N> &arg) ->  iVector<decltype(traceIndex<Level>(arg._internal[0])),N> 
 {
   iVector<decltype(traceIndex<Level>(arg._internal[0])),N> ret;

From b9e9777912f104493c5e78c83ed0ff945fd17eb5 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 4 Jun 2015 13:28:37 +0100
Subject: [PATCH 274/429] PartialFraction Hw with Zolo and Tanh approx
 converged under CG and passed EO breakdown and hermiticity tests.

---
 INSTALL                                       |   2 +-
 configure                                     |  30 +-
 configure.ac                                  |  13 +-
 lib/Make.inc                                  |   4 +-
 lib/algorithms/approx/Zolotarev.cc            |  16 +-
 lib/algorithms/approx/Zolotarev.h             |   5 +-
 lib/qcd/action/Actions.h                      |   2 +
 .../fermion/ContinuedFractionFermion5D.cc     |  14 +-
 .../fermion/ContinuedFractionFermion5D.h      |   3 +-
 lib/qcd/action/fermion/DomainWallFermion.h    |   5 +-
 lib/qcd/action/fermion/MobiusFermion.h        |   4 +-
 .../action/fermion/MobiusZolotarevFermion.h   |   3 +-
 .../OverlapWilsonContfracTanhFermion.h        |   3 +-
 .../OverlapWilsonContfracZolotarevFermion.h   |   6 +-
 .../OverlapWilsonPartialFractionTanhFermion.h |  40 +++
 ...lapWilsonPartialFractionZolotarevFermion.h |  44 +++
 .../fermion/PartialFractionFermion5D.cc       | 309 ++++++++++++++++++
 .../action/fermion/PartialFractionFermion5D.h |  32 +-
 lib/qcd/action/fermion/WilsonFermion5D.h      |   4 +-
 tests/Test_contfrac_cg.cc                     |  10 +
 tests/Test_contfrac_even_odd.cc               |   8 +
 21 files changed, 501 insertions(+), 56 deletions(-)
 create mode 100644 lib/qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h
 create mode 100644 lib/qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h

diff --git a/INSTALL b/INSTALL
index f812f5a2..80a61507 120000
--- a/INSTALL
+++ b/INSTALL
@@ -1 +1 @@
-/usr/share/automake-1.14/INSTALL
\ No newline at end of file
+/opt/local/share/automake-1.15/INSTALL
\ No newline at end of file
diff --git a/configure b/configure
index e8c4a9ca..36f85edb 100755
--- a/configure
+++ b/configure
@@ -2574,7 +2574,7 @@ test -n "$target_alias" &&
     NONENONEs,x,x, &&
   program_prefix=${target_alias}-
 
-am__api_version='1.14'
+am__api_version='1.15'
 
 # Find a good install program.  We prefer a C program (faster),
 # so one script is as good as another.  But avoid the broken or
@@ -2746,8 +2746,8 @@ test "$program_suffix" != NONE &&
 ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
 program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
 
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 
 if test x"${MISSING+set}" != xset; then
   case $am_aux_dir in
@@ -2766,7 +2766,7 @@ else
 $as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
 fi
 
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -3094,8 +3094,8 @@ MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 mkdir_p='$(MKDIR_P)'
 
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
 # Always define AMTAR for backward compatibility.  Yes, it's still used
 # in the wild :-(  We should find a proper way to deprecate it ...
 AMTAR='$${TAR-tar}'
@@ -3154,6 +3154,7 @@ END
 fi
 
 
+
 ac_config_headers="$ac_config_headers lib/GridConfig.h"
 
 # Check whether --enable-silent-rules was given.
@@ -6656,9 +6657,6 @@ fi
 
 
 
-
-
-
 # Check whether --enable-simd was given.
 if test "${enable_simd+set}" = set; then :
   enableval=$enable_simd; \
@@ -6673,21 +6671,21 @@ supported=no
 case ${ac_SIMD} in
      SSE4)
        echo Configuring for SSE4
-       if test x"$ax_cv_support_ssse3_ext" = x"yes"; then
+
 $as_echo "#define SSE4 1" >>confdefs.h
 
-       supported=yes
+       if test x"$ax_cv_support_ssse3_ext" = x"yes"; then           supported=yes
        else
-	{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor does not support SSE4 instructions" >&5
+  	{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor does not support SSE4 instructions" >&5
 $as_echo "$as_me: WARNING: Your processor does not support SSE4 instructions" >&2;}
        fi
      ;;
      AVX)
        echo Configuring for AVX
-       if test x"$ax_cv_support_avx_ext" = x"yes"; then
+
 $as_echo "#define AVX1 1" >>confdefs.h
 
-       supported=yes
+       if test x"$ax_cv_support_avx_ext" = x"yes"; then         supported=yes
        else
        	{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor does not support AVX instructions" >&5
 $as_echo "$as_me: WARNING: Your processor does not support AVX instructions" >&2;}
@@ -6695,10 +6693,10 @@ $as_echo "$as_me: WARNING: Your processor does not support AVX instructions" >&2
      ;;
      AVX2)
        echo Configuring for AVX2
-       if test x"$ax_cv_support_avx2_ext" = x"yes"; then
+
 $as_echo "#define AVX2 1" >>confdefs.h
 
-       supported=yes
+       if test x"$ax_cv_support_avx2_ext" = x"yes"; then         supported=yes
        else
        { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor does not support AVX2 instructions" >&5
 $as_echo "$as_me: WARNING: Your processor does not support AVX2 instructions" >&2;}
diff --git a/configure.ac b/configure.ac
index 74a0eebc..59792fb0 100644
--- a/configure.ac
+++ b/configure.ac
@@ -66,9 +66,6 @@ Please install or provide the correct path to your installation
 Info at: http://www.mpfr.org/)])
 
 
-
-
-
 AC_ARG_ENABLE([simd],[AC_HELP_STRING([--enable-simd=SSE4|AVX|AVX2|AVX512|MIC],\
 	[Select instructions to be SSE4.0, AVX 1.0, AVX 2.0+FMA, AVX 512, MIC])],\
 	[ac_SIMD=${enable_simd}],[ac_SIMD=AVX2])
@@ -78,17 +75,17 @@ supported=no
 case ${ac_SIMD} in
      SSE4)
        echo Configuring for SSE4
-       if test x"$ax_cv_support_ssse3_ext" = x"yes"; then  dnl minimal support for SSE4
        AC_DEFINE([SSE4],[1],[SSE4] )
-       supported=yes
+       if test x"$ax_cv_support_ssse3_ext" = x"yes"; then  dnl minimal support for SSE4
+         supported=yes
        else
-	AC_MSG_WARN([Your processor does not support SSE4 instructions])
+  	AC_MSG_WARN([Your processor does not support SSE4 instructions])
        fi
      ;;
      AVX)
        echo Configuring for AVX
-       if test x"$ax_cv_support_avx_ext" = x"yes"; then  dnl minimal support for AVX
        AC_DEFINE([AVX1],[1],[AVX] )
+       if test x"$ax_cv_support_avx_ext" = x"yes"; then  dnl minimal support for AVX
        supported=yes			  
        else
        	AC_MSG_WARN([Your processor does not support AVX instructions])
@@ -96,8 +93,8 @@ case ${ac_SIMD} in
      ;;
      AVX2)
        echo Configuring for AVX2
-       if test x"$ax_cv_support_avx2_ext" = x"yes"; then  dnl minimal support for AVX2
        AC_DEFINE([AVX2],[1],[AVX2] )
+       if test x"$ax_cv_support_avx2_ext" = x"yes"; then  dnl minimal support for AVX2
        supported=yes
        else
        AC_MSG_WARN([Your processor does not support AVX2 instructions])
diff --git a/lib/Make.inc b/lib/Make.inc
index 787539b7..be0a939d 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./Cshift.h ./simd/Grid_avx.h ./simd/Grid_vector_types.h ./simd/Grid_sse4.h ./simd/Grid_avx512.h ./simd/Old/Grid_vRealD.h ./simd/Old/Grid_vComplexD.h ./simd/Old/Grid_vInteger.h ./simd/Old/Grid_vComplexF.h ./simd/Old/Grid_vRealF.h ./simd/Grid_qpx.h ./Tensors.h ./Algorithms.h ./communicator/Communicator_base.h ./lattice/Lattice_rng.h ./lattice/Lattice_reduction.h ./lattice/Lattice_transfer.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_comparison.h ./lattice/Lattice_overload.h ./lattice/Lattice_reality.h ./lattice/Lattice_local.h ./lattice/Lattice_conformable.h ./lattice/Lattice_where.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_ET.h ./lattice/Lattice_transpose.h ./lattice/Lattice_trace.h ./Stencil.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_poke.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_class.h ./tensors/Tensor_transpose.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_reality.h ./tensors/Tensor_trace.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_outer.h ./tensors/Tensor_inner.h ./tensors/Tensor_traits.h ./tensors/Tensor_Ta.h ./tensors/Tensor_peek.h ./tensors/Tensor_arith.h ./tensors/Tensor_extract_merge.h ./Communicator.h ./Cartesian.h ./parallelIO/NerscIO.h ./qcd/QCD.h ./qcd/SpaceTimeGrid.h ./qcd/LinalgUtils.h ./qcd/TwoSpinor.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/Dirac.h ./cshift/Cshift_common.h ./cshift/Cshift_none.h ./cshift/Cshift_mpi.h ./Simd.h ./GridConfig.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_red_black.h ./cartesian/Cartesian_full.h ./AlignedAllocator.h ./Lattice.h ./Threads.h ./Comparison.h ./Grid.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Zolotarev.h ./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Remez.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./stencil/Lebesgue.h
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Comparison.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/Dirac.h ./qcd/LinalgUtils.h ./qcd/QCD.h ./qcd/SpaceTimeGrid.h ./qcd/TwoSpinor.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Old/Grid_vComplexD.h ./simd/Old/Grid_vComplexF.h ./simd/Old/Grid_vInteger.h ./simd/Old/Grid_vRealD.h ./simd/Old/Grid_vRealF.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./Tensors.h ./Threads.h
 
-CCFILES=./qcd/SpaceTimeGrid.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/Dirac.cc ./GridInit.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
+CCFILES=./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/Dirac.cc ./qcd/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/algorithms/approx/Zolotarev.cc b/lib/algorithms/approx/Zolotarev.cc
index c73a3436..1aba7aa8 100644
--- a/lib/algorithms/approx/Zolotarev.cc
+++ b/lib/algorithms/approx/Zolotarev.cc
@@ -293,7 +293,7 @@ static void sncndnFK(INTERNAL_PRECISION u, INTERNAL_PRECISION k,
  * Set type = 0 for the Zolotarev approximation, which is zero at x = 0, and
  * type = 1 for the approximation which is infinite at x = 0. */
 
-zolotarev_data* grid_zolotarev(PRECISION epsilon, int n, int type) {
+zolotarev_data* zolotarev(PRECISION epsilon, int n, int type) {
   INTERNAL_PRECISION A, c, cp, kp, ksq, sn, cn, dn, Kp, Kj, z, z0, t, M, F,
     l, invlambda, xi, xisq, *tv, s, opl;
   int m, czero, ts;
@@ -414,7 +414,19 @@ zolotarev_data* grid_zolotarev(PRECISION epsilon, int n, int type) {
   return zd;
 }
 
-zolotarev_data* grid_higham(PRECISION epsilon, int n) {
+
+void zolotarev_free(zolotarev_data *zdata)
+{
+    free(zdata -> a);
+    free(zdata -> ap);
+    free(zdata -> alpha);
+    free(zdata -> beta);
+    free(zdata -> gamma);
+    free(zdata);
+}
+
+
+zolotarev_data* higham(PRECISION epsilon, int n) {
   INTERNAL_PRECISION A, M, c, cp, z, z0, t, epssq;
   int m, czero;
   zolotarev_data *zd;
diff --git a/lib/algorithms/approx/Zolotarev.h b/lib/algorithms/approx/Zolotarev.h
index 869e5a89..f6ee56b1 100644
--- a/lib/algorithms/approx/Zolotarev.h
+++ b/lib/algorithms/approx/Zolotarev.h
@@ -77,8 +77,9 @@ typedef struct {
  * zolotarev_data structure. The arguments must satisfy the constraints that
  * epsilon > 0, n > 0, and type = 0 or 1. */
 
-ZOLOTAREV_DATA* grid_higham(PRECISION epsilon, int n) ;
-ZOLOTAREV_DATA* grid_zolotarev(PRECISION epsilon, int n, int type);
+ZOLOTAREV_DATA* higham(PRECISION epsilon, int n) ;
+ZOLOTAREV_DATA* zolotarev(PRECISION epsilon, int n, int type);
+void zolotarev_free(zolotarev_data *zdata);
 #endif
 
 #ifdef __cplusplus
diff --git a/lib/qcd/action/Actions.h b/lib/qcd/action/Actions.h
index b31e9136..edef1b53 100644
--- a/lib/qcd/action/Actions.h
+++ b/lib/qcd/action/Actions.h
@@ -62,6 +62,8 @@
 // Partial fraction
 //////////////////////
 #include <qcd/action/fermion/PartialFractionFermion5D.h>
+#include <qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h>
+#include <qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h>
 
 
     // Chroma interface defining FermionAction
diff --git a/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc b/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
index 92f6473e..6f99aa7f 100644
--- a/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
+++ b/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
@@ -9,6 +9,16 @@ namespace Grid {
     }
     void ContinuedFractionFermion5D::SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata)
     {
+      // How to check Ls matches??
+      //      std::cout << Ls << " Ls"<<std::endl;
+      //      std::cout << zdata->n  << " - n"<<std::endl;
+      //      std::cout << zdata->da << " -da "<<std::endl;
+      //      std::cout << zdata->db << " -db"<<std::endl;
+      //      std::cout << zdata->dn << " -dn"<<std::endl;
+      //      std::cout << zdata->dd << " -dd"<<std::endl;
+
+      assert(zdata->db==Ls);// Beta has Ls coeffs
+
       R=(1+this->mass)/(1-this->mass);
 
       Beta.resize(Ls);
@@ -29,7 +39,7 @@ namespace Grid {
 
 
       ZoloHiInv =1.0/zolo_hi;
-      double dw_diag = (4.0-M5)*ZoloHiInv;
+      dw_diag = (4.0-M5)*ZoloHiInv;
     
       See.resize(Ls);
       Aee.resize(Ls);
@@ -105,8 +115,6 @@ namespace Grid {
     }
     void   ContinuedFractionFermion5D::Mooee       (const LatticeFermion &psi, LatticeFermion &chi)
     {
-      double dw_diag = (4.0-M5)*ZoloHiInv;
-    
       int sign=1;
       for(int s=0;s<Ls;s++){
 	if ( s==0 ) {
diff --git a/lib/qcd/action/fermion/ContinuedFractionFermion5D.h b/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
index f363878f..b7e321c9 100644
--- a/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
+++ b/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
@@ -37,9 +37,8 @@ namespace Grid {
       void SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD scale);
       void SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata);;
 
-      Approx::zolotarev_data *zdata;
-
       // Cont frac
+      RealD dw_diag;
       RealD mass;
       RealD R;
       RealD ZoloHiInv;
diff --git a/lib/qcd/action/fermion/DomainWallFermion.h b/lib/qcd/action/fermion/DomainWallFermion.h
index a25c0c3c..34f01be5 100644
--- a/lib/qcd/action/fermion/DomainWallFermion.h
+++ b/lib/qcd/action/fermion/DomainWallFermion.h
@@ -29,13 +29,14 @@ namespace Grid {
       {
 	RealD eps = 1.0;
 
-	Approx::zolotarev_data *zdata = Approx::grid_higham(eps,this->Ls);// eps is ignored for higham
+	Approx::zolotarev_data *zdata = Approx::higham(eps,this->Ls);// eps is ignored for higham
 	assert(zdata->n==this->Ls);
 	
 	std::cout << "DomainWallFermion with Ls="<<Ls<<std::endl;
 	// Call base setter
 	this->CayleyFermion5D::SetCoefficientsTanh(zdata,1.0,0.0);
- 
+
+	Approx::zolotarev_free(zdata);
       }
 
     };
diff --git a/lib/qcd/action/fermion/MobiusFermion.h b/lib/qcd/action/fermion/MobiusFermion.h
index 33f94089..6603e3f2 100644
--- a/lib/qcd/action/fermion/MobiusFermion.h
+++ b/lib/qcd/action/fermion/MobiusFermion.h
@@ -31,11 +31,13 @@ namespace Grid {
 	RealD eps = 1.0;
 
 	std::cout << "MobiusFermion (b="<<b<<",c="<<c<<") with Ls= "<<Ls<<" Tanh approx"<<std::endl;
-	Approx::zolotarev_data *zdata = Approx::grid_higham(eps,this->Ls);// eps is ignored for higham
+	Approx::zolotarev_data *zdata = Approx::higham(eps,this->Ls);// eps is ignored for higham
 	assert(zdata->n==this->Ls);
 	
 	// Call base setter
 	this->CayleyFermion5D::SetCoefficientsTanh(zdata,b,c);
+
+	Approx::zolotarev_free(zdata);
  
       }
 
diff --git a/lib/qcd/action/fermion/MobiusZolotarevFermion.h b/lib/qcd/action/fermion/MobiusZolotarevFermion.h
index 1be61601..d3c0b3a0 100644
--- a/lib/qcd/action/fermion/MobiusZolotarevFermion.h
+++ b/lib/qcd/action/fermion/MobiusZolotarevFermion.h
@@ -31,7 +31,7 @@ namespace Grid {
       {
 	RealD eps = lo/hi;
 
-	Approx::zolotarev_data *zdata = Approx::grid_zolotarev(eps,this->Ls,0);// eps is ignored for higham
+	Approx::zolotarev_data *zdata = Approx::zolotarev(eps,this->Ls,0);
 	assert(zdata->n==this->Ls);
 
 	std::cout << "MobiusZolotarevFermion (b="<<b<<",c="<<c<<") with Ls= "<<Ls<<" Zolotarev range ["<<lo<<","<<hi<<"]"<<std::endl;
@@ -39,6 +39,7 @@ namespace Grid {
 	// Call base setter
 	this->CayleyFermion5D::SetCoefficientsZolotarev(hi,zdata,b,c);
  
+	Approx::zolotarev_free(zdata);
       }
 
     };
diff --git a/lib/qcd/action/fermion/OverlapWilsonContfracTanhFermion.h b/lib/qcd/action/fermion/OverlapWilsonContfracTanhFermion.h
index ed0c24dc..6ecd7822 100644
--- a/lib/qcd/action/fermion/OverlapWilsonContfracTanhFermion.h
+++ b/lib/qcd/action/fermion/OverlapWilsonContfracTanhFermion.h
@@ -30,8 +30,9 @@ namespace Grid {
 	{
 	  assert((Ls&0x1)==1); // Odd Ls required
 	  int nrational=Ls-1;// Even rational order
-	  zdata = Approx::grid_higham(1.0,nrational);// eps is ignored for higham
+	  Approx::zolotarev_data *zdata = Approx::higham(1.0,nrational);// eps is ignored for higham
 	  SetCoefficientsTanh(zdata,scale);
+	  Approx::zolotarev_free(zdata);
 	}
     };
   }
diff --git a/lib/qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h b/lib/qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h
index caf01133..595daef3 100644
--- a/lib/qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h
+++ b/lib/qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h
@@ -30,12 +30,12 @@ namespace Grid {
 	{
 	  assert((Ls&0x1)==1); // Odd Ls required
 
-	  int nrational=Ls-1;// Even rational order
+	  int nrational=Ls;// Odd rational order
 	  RealD eps = lo/hi;
 
-	  Approx::zolotarev_data *zdata = Approx::grid_zolotarev(eps,nrational,0);
-
+	  Approx::zolotarev_data *zdata = Approx::zolotarev(eps,nrational,0);
 	  SetCoefficientsZolotarev(hi,zdata);
+	  Approx::zolotarev_free(zdata);
 
 	}
     };
diff --git a/lib/qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h b/lib/qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h
new file mode 100644
index 00000000..1fa6c099
--- /dev/null
+++ b/lib/qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h
@@ -0,0 +1,40 @@
+#ifndef OVERLAP_WILSON_PARTFRAC_TANH_FERMION_H
+#define OVERLAP_WILSON_PARTFRAC_TANH_FERMION_H
+
+#include <Grid.h>
+
+namespace Grid {
+
+  namespace QCD {
+
+    class OverlapWilsonPartialFractionTanhFermion : public PartialFractionFermion5D
+    {
+    public:
+
+      virtual void   Instantiatable(void){};
+      // Constructors
+    OverlapWilsonPartialFractionTanhFermion(LatticeGaugeField &_Umu,
+					    GridCartesian         &FiveDimGrid,
+					    GridRedBlackCartesian &FiveDimRedBlackGrid,
+					    GridCartesian         &FourDimGrid,
+					    GridRedBlackCartesian &FourDimRedBlackGrid,
+					    RealD _mass,RealD _M5,
+					    RealD scale) :
+      
+      // b+c=scale, b-c = 0 <=> b =c = scale/2
+      PartialFractionFermion5D(_Umu,
+			       FiveDimGrid,
+			       FiveDimRedBlackGrid,
+			       FourDimGrid,
+			       FourDimRedBlackGrid,_mass,_M5)
+	{
+	  assert((Ls&0x1)==1); // Odd Ls required
+	  int nrational=Ls-1;// Even rational order
+	  Approx::zolotarev_data *zdata = Approx::higham(1.0,nrational);// eps is ignored for higham
+	  SetCoefficientsTanh(zdata,scale);
+	  Approx::zolotarev_free(zdata);
+	}
+    };
+  }
+}
+#endif
diff --git a/lib/qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h b/lib/qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h
new file mode 100644
index 00000000..3997c17f
--- /dev/null
+++ b/lib/qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h
@@ -0,0 +1,44 @@
+#ifndef OVERLAP_WILSON_PARTFRAC_ZOLOTAREV_FERMION_H
+#define OVERLAP_WILSON_PARTFRAC_ZOLOTAREV_FERMION_H
+
+#include <Grid.h>
+
+namespace Grid {
+
+  namespace QCD {
+
+    class OverlapWilsonPartialFractionZolotarevFermion : public PartialFractionFermion5D
+    {
+    public:
+
+      virtual void   Instantiatable(void){};
+      // Constructors
+    OverlapWilsonPartialFractionZolotarevFermion(LatticeGaugeField &_Umu,
+					  GridCartesian         &FiveDimGrid,
+					  GridRedBlackCartesian &FiveDimRedBlackGrid,
+					  GridCartesian         &FourDimGrid,
+					  GridRedBlackCartesian &FourDimRedBlackGrid,
+					  RealD _mass,RealD _M5,
+					  RealD lo,RealD hi):
+      
+      // b+c=scale, b-c = 0 <=> b =c = scale/2
+      PartialFractionFermion5D(_Umu,
+			       FiveDimGrid,
+			       FiveDimRedBlackGrid,
+			       FourDimGrid,
+			       FourDimRedBlackGrid,_mass,_M5)
+	{
+	  assert((Ls&0x1)==1); // Odd Ls required
+
+	  int nrational=Ls;// Odd rational order
+	  RealD eps = lo/hi;
+
+	  Approx::zolotarev_data *zdata = Approx::zolotarev(eps,nrational,0);
+	  SetCoefficientsZolotarev(hi,zdata);
+	  Approx::zolotarev_free(zdata);
+
+	}
+    };
+  }
+}
+#endif
diff --git a/lib/qcd/action/fermion/PartialFractionFermion5D.cc b/lib/qcd/action/fermion/PartialFractionFermion5D.cc
index 8b137891..285144c6 100644
--- a/lib/qcd/action/fermion/PartialFractionFermion5D.cc
+++ b/lib/qcd/action/fermion/PartialFractionFermion5D.cc
@@ -1 +1,310 @@
+#include <Grid.h>
+namespace Grid {
+  namespace QCD {
+
+    void   PartialFractionFermion5D::Meooe_internal(const LatticeFermion &psi, LatticeFermion &chi,int dag)
+    {
+      // this does both dag and undag but is trivial; make a common helper routing
+      int sign = dag ? (-1) : 1;
+
+      if ( psi.checkerboard == Odd ) {
+	DhopEO(psi,chi,DaggerNo);
+      } else {
+	DhopOE(psi,chi,DaggerNo);
+      }
+
+      int nblock=(Ls-1)/2;
+      for(int b=0;b<nblock;b++){
+	int s = 2*b;
+	ag5xpby_ssp(chi,-scale,chi,0.0,chi,s,s); 
+	ag5xpby_ssp(chi, scale,chi,0.0,chi,s+1,s+1); 
+      }
+      ag5xpby_ssp(chi,p[nblock]*scale/amax,chi,0.0,chi,Ls-1,Ls-1);
+    }
+
+    void   PartialFractionFermion5D::Mooee_internal(const LatticeFermion &psi, LatticeFermion &chi,int dag)
+    {
+      // again dag and undag are trivially related
+      int sign = dag ? (-1) : 1;
+      
+      int nblock=(Ls-1)/2;
+      for(int b=0;b<nblock;b++){
+	
+	int s = 2*b;
+	RealD pp = p[nblock-1-b];
+	RealD qq = q[nblock-1-b];
+	
+	// Do each 2x2 block aligned at s and multiplies Dw site diagonal by G5 so Hw
+	ag5xpby_ssp(chi,-dw_diag*scale,psi,amax*sqrt(qq)*scale,psi, s  ,s+1); 
+	ag5xpby_ssp(chi, dw_diag*scale,psi,amax*sqrt(qq)*scale,psi, s+1,s);
+	axpby_ssp  (chi, 1.0, chi,sqrt(amax*pp)*scale*sign,psi,s+1,Ls-1);
+      }
+      
+      {
+	RealD R=(1+mass)/(1-mass);
+	//R g5 psi[Ls-1] + p[0] H
+	ag5xpbg5y_ssp(chi,R*scale,psi,p[nblock]*scale*dw_diag/amax,psi,Ls-1,Ls-1);
+	
+	for(int b=0;b<nblock;b++){
+	  int s = 2*b+1;
+	  RealD pp = p[nblock-1-b];
+	  axpby_ssp(chi,1.0,chi,-sqrt(amax*pp)*scale*sign,psi,Ls-1,s);
+	}
+      }
+    }
+
+    void   PartialFractionFermion5D::MooeeInv_internal(const LatticeFermion &psi, LatticeFermion &chi,int dag)
+    {
+      int sign = dag ? (-1) : 1;
+
+      LatticeFermion tmp(psi._grid);
+      
+      ///////////////////////////////////////////////////////////////////////////////////////
+      //Linv
+      ///////////////////////////////////////////////////////////////////////////////////////
+      int nblock=(Ls-1)/2;
+
+      axpy(chi,0.0,psi,psi); // Identity piece
+      
+      for(int b=0;b<nblock;b++){
+	int s = 2*b;
+	RealD pp = p[nblock-1-b];
+	RealD qq = q[nblock-1-b];
+	RealD coeff1=sign*sqrt(amax*amax*amax*pp*qq) / ( dw_diag*dw_diag + amax*amax* qq);
+	RealD coeff2=sign*sqrt(amax*pp)*dw_diag / ( dw_diag*dw_diag + amax*amax* qq); // Implicit g5 here
+	axpby_ssp  (chi,1.0,chi,coeff1,psi,Ls-1,s);
+	axpbg5y_ssp(chi,1.0,chi,coeff2,psi,Ls-1,s+1);
+      }
+      
+      ///////////////////////////////////////////////////////////////////////////////////////
+      //Dinv (note D isn't really diagonal -- just diagonal enough that we can still invert)
+      // Compute Seeinv (coeff of gamma5)
+      ///////////////////////////////////////////////////////////////////////////////////////
+      RealD R=(1+mass)/(1-mass);
+      RealD Seeinv = R + p[nblock]*dw_diag/amax;
+      for(int b=0;b<nblock;b++){
+	Seeinv += p[nblock-1-b]*dw_diag/amax / ( dw_diag*dw_diag/amax/amax + q[nblock-1-b]);
+      }    
+      Seeinv = 1.0/Seeinv;
+      
+      for(int b=0;b<nblock;b++){
+	int s = 2*b;
+	RealD pp = p[nblock-1-b];
+	RealD qq = q[nblock-1-b];
+	RealD coeff1=dw_diag / ( dw_diag*dw_diag + amax*amax* qq); // Implicit g5 here
+	RealD coeff2=amax*sqrt(qq) / ( dw_diag*dw_diag + amax*amax* qq);
+	ag5xpby_ssp  (tmp,-coeff1,chi,coeff2,chi,s,s+1);
+	ag5xpby_ssp  (tmp, coeff1,chi,coeff2,chi,s+1,s);
+      }
+      ag5xpby_ssp  (tmp, Seeinv,chi,0.0,chi,Ls-1,Ls-1);
+      
+      ///////////////////////////////////////////////////////////////////////////////////////
+      // Uinv
+      ///////////////////////////////////////////////////////////////////////////////////////
+      for(int b=0;b<nblock;b++){
+	int s = 2*b;
+	RealD pp = p[nblock-1-b];
+	RealD qq = q[nblock-1-b];
+	RealD coeff1=-sign*sqrt(amax*amax*amax*pp*qq) / ( dw_diag*dw_diag + amax*amax* qq);
+	RealD coeff2=-sign*sqrt(amax*pp)*dw_diag / ( dw_diag*dw_diag + amax*amax* qq); // Implicit g5 here
+	axpby_ssp  (chi,1.0/scale,tmp,coeff1/scale,tmp,s,Ls-1);
+	axpbg5y_ssp(chi,1.0/scale,tmp,coeff2/scale,tmp,s+1,Ls-1);
+      }
+      axpby_ssp  (chi, 1.0/scale,tmp,0.0,tmp,Ls-1,Ls-1);
+    }
+
+    void   PartialFractionFermion5D::M_internal(const LatticeFermion &psi, LatticeFermion &chi,int dag)
+    {
+      LatticeFermion D(psi._grid);
+  
+      int sign = dag ? (-1) : 1;
+
+      // For partial frac Hw case (b5=c5=1) chroma quirkily computes
+      //
+      // Conventions for partfrac appear to be a mess.
+      // Tony's Nara lectures have
+      //
+      // BlockDiag(  H/p_i  1             | 1       )    
+      //          (  1      p_i H / q_i^2 | 0       )  
+      //           ---------------------------------
+      //           ( -1      0                | R  +p0 H  )
+      //
+      //Chroma     ( -2H    2sqrt(q_i)    |   0         )
+      //           (2 sqrt(q_i)   2H      |  2 sqrt(p_i) )
+      //           ---------------------------------
+      //           ( 0     -2 sqrt(p_i)   |  2 R gamma_5 + p0 2H
+      //
+      // Edwards/Joo/Kennedy/Wenger
+      //
+      // Here, the "beta's" selected by chroma to scale the unphysical bulk constraint fields
+      // incorporate the approx scale factor. This is obtained by propagating the
+      // scale on "H" out to the off diagonal elements as follows:
+      //
+      // BlockDiag(  H/p_i  1             | 1       ) 
+      //          (  1      p_i H / q_i^2 | 0       )  
+      //           ---------------------------------
+      //          ( -1      0                | R  + p_0 H  )
+      //
+      // becomes:
+      // BlockDiag(  H/ sp_i  1               | 1             ) 
+      //          (  1      sp_i H / s^2q_i^2 | 0             )  
+      //           ---------------------------------
+      //           ( -1      0                | R + p_0/s H   )
+      //
+      //
+      // This is implemented in Chroma by
+      //           p0' = p0/approxMax
+      //           p_i' = p_i*approxMax
+      //           q_i' = q_i*approxMax*approxMax
+      //
+      // After the equivalence transform is applied the matrix becomes
+      // 
+      //Chroma     ( -2H    sqrt(q'_i)    |   0         )
+      //           (sqrt(q'_i)   2H       |   sqrt(p'_i) )
+      //           ---------------------------------
+      //           ( 0     -sqrt(p'_i)    |  2 R gamma_5 + p'0 2H
+      //
+      //     =     ( -2H    sqrt(q_i)amax    |   0              )
+      //           (sqrt(q_i)amax   2H       |   sqrt(p_i*amax) )
+      //           ---------------------------------
+      //           ( 0     -sqrt(p_i)*amax   |  2 R gamma_5 + p0/amax 2H
+      //
+
+      DW(psi,D,DaggerNo); 
+
+      int nblock=(Ls-1)/2;
+      for(int b=0;b<nblock;b++){
+	
+	int s = 2*b;
+	double pp = p[nblock-1-b];
+	double qq = q[nblock-1-b];
+	
+	// Do each 2x2 block aligned at s and
+	ag5xpby_ssp(chi,-1.0*scale,D,amax*sqrt(qq)*scale,psi, s  ,s+1); // Multiplies Dw by G5 so Hw
+	ag5xpby_ssp(chi, 1.0*scale,D,amax*sqrt(qq)*scale,psi, s+1,s);
+	
+	// Pick up last column
+	axpby_ssp  (chi, 1.0, chi,sqrt(amax*pp)*scale*sign,psi,s+1,Ls-1);
+      }
+	
+      {
+	double R=(1+this->mass)/(1-this->mass);
+	//R g5 psi[Ls] + p[0] H
+	ag5xpbg5y_ssp(chi,R*scale,psi,p[nblock]*scale/amax,D,Ls-1,Ls-1);
+	
+	for(int b=0;b<nblock;b++){
+	  int s = 2*b+1;
+	  double pp = p[nblock-1-b];
+	  axpby_ssp(chi,1.0,chi,-sqrt(amax*pp)*scale*sign,psi,Ls-1,s);
+	}
+      }
+
+    }
+
+    RealD  PartialFractionFermion5D::M    (const LatticeFermion &in, LatticeFermion &out)
+    {
+      M_internal(in,out,DaggerNo);
+      return norm2(out);
+    }
+    RealD  PartialFractionFermion5D::Mdag (const LatticeFermion &in, LatticeFermion &out)
+    {
+      M_internal(in,out,DaggerYes);
+      return norm2(out);
+    }
+
+    void PartialFractionFermion5D::Meooe       (const LatticeFermion &in, LatticeFermion &out)
+    {
+      Meooe_internal(in,out,DaggerNo);
+    }
+    void PartialFractionFermion5D::MeooeDag    (const LatticeFermion &in, LatticeFermion &out)
+    {
+      Meooe_internal(in,out,DaggerYes);
+    }
+    void PartialFractionFermion5D::Mooee       (const LatticeFermion &in, LatticeFermion &out)
+    {
+      Mooee_internal(in,out,DaggerNo);
+    }
+    void PartialFractionFermion5D::MooeeDag    (const LatticeFermion &in, LatticeFermion &out)
+    {
+      Mooee_internal(in,out,DaggerYes);
+    }
+
+    void PartialFractionFermion5D::MooeeInv    (const LatticeFermion &in, LatticeFermion &out)
+    {
+      MooeeInv_internal(in,out,DaggerNo);
+    }
+    void PartialFractionFermion5D::MooeeInvDag (const LatticeFermion &in, LatticeFermion &out)
+    {
+      MooeeInv_internal(in,out,DaggerYes);
+    }
+
+    void  PartialFractionFermion5D::SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD scale){
+      SetCoefficientsZolotarev(1.0/scale,zdata);
+    }
+    void  PartialFractionFermion5D::SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata){
+
+      // check on degree matching
+      //      std::cout << Ls << " Ls"<<std::endl;
+      //      std::cout << zdata->n  << " - n"<<std::endl;
+      //      std::cout << zdata->da << " -da "<<std::endl;
+      //      std::cout << zdata->db << " -db"<<std::endl;
+      //      std::cout << zdata->dn << " -dn"<<std::endl;
+      //      std::cout << zdata->dd << " -dd"<<std::endl;
+      assert(Ls == (2*zdata->da -1) );
+
+      // Part frac
+      //      RealD R;
+      R=(1+mass)/(1-mass);
+      dw_diag = (4.0-M5);
+
+      //      std::vector<RealD> p; 
+      //      std::vector<RealD> q;
+      p.resize(zdata->da);
+      q.resize(zdata->dd);
+	
+      for(int n=0;n<zdata->da;n++){
+	p[n] = zdata -> alpha[n];
+      }
+      for(int n=0;n<zdata->dd;n++){
+	q[n] = -zdata -> ap[n];
+      }
+      
+      scale= part_frac_chroma_convention ? 2.0 : 1.0; // Chroma conventions annoy me
+
+      amax=zolo_hi;
+    }
+
+      // Constructors
+    PartialFractionFermion5D::PartialFractionFermion5D(LatticeGaugeField &_Umu,
+						       GridCartesian         &FiveDimGrid,
+						       GridRedBlackCartesian &FiveDimRedBlackGrid,
+						       GridCartesian         &FourDimGrid,
+						       GridRedBlackCartesian &FourDimRedBlackGrid,
+						       RealD _mass,RealD M5) :
+      WilsonFermion5D(_Umu,
+		      FiveDimGrid, FiveDimRedBlackGrid,
+		      FourDimGrid, FourDimRedBlackGrid,M5),
+      mass(_mass)
+
+    {
+      assert((Ls&0x1)==1); // Odd Ls required
+      int nrational=Ls-1;
+
+
+      Approx::zolotarev_data *zdata = Approx::higham(1.0,nrational);
+
+      // NB: chroma uses a cast to "float" for the zolotarev range(!?).
+      // this creates a real difference in the operator which I do not like but we can replicate here
+      // to demonstrate compatibility
+      //      RealD eps = (zolo_lo / zolo_hi);
+      //      zdata = bfm_zolotarev(eps,nrational,0);
+      
+      SetCoefficientsTanh(zdata,1.0);
+
+      Approx::zolotarev_free(zdata);
+
+    }
+ 
+ }
+}
 
diff --git a/lib/qcd/action/fermion/PartialFractionFermion5D.h b/lib/qcd/action/fermion/PartialFractionFermion5D.h
index 95f8c0f9..301b07c0 100644
--- a/lib/qcd/action/fermion/PartialFractionFermion5D.h
+++ b/lib/qcd/action/fermion/PartialFractionFermion5D.h
@@ -9,6 +9,13 @@ namespace Grid {
     {
     public:
 
+      const int part_frac_chroma_convention=1;
+
+      void   Meooe_internal(const LatticeFermion &in, LatticeFermion &out,int dag);
+      void   Mooee_internal(const LatticeFermion &in, LatticeFermion &out,int dag);
+      void   MooeeInv_internal(const LatticeFermion &in, LatticeFermion &out,int dag);
+      void   M_internal(const LatticeFermion &in, LatticeFermion &out,int dag);
+
       // override multiply
       virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out);
       virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out);
@@ -21,16 +28,7 @@ namespace Grid {
       virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out);
       virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
 
-    private:
-
-      virtual void PartialFractionCoefficients(void);
-
-      Approx::zolotarev_data *zdata;
-
-      // Part frac
-      double R;
-      std::vector<double> p; 
-      std::vector<double> q;
+      virtual void   Instantiatable(void) =0; // ensure no make-eee
 
       // Constructors
       PartialFractionFermion5D(LatticeGaugeField &_Umu,
@@ -40,6 +38,20 @@ namespace Grid {
 				    GridRedBlackCartesian &FourDimRedBlackGrid,
 				    RealD _mass,RealD M5);
 
+    protected:
+
+      virtual void SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD scale);
+      virtual void SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata);
+
+      // Part frac
+      RealD mass;
+      RealD dw_diag;
+      RealD R;
+      RealD amax;
+      RealD scale;
+      std::vector<double> p; 
+      std::vector<double> q;
+
     };
 
 
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.h b/lib/qcd/action/fermion/WilsonFermion5D.h
index 062c4d82..4d4a7aed 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.h
+++ b/lib/qcd/action/fermion/WilsonFermion5D.h
@@ -1,5 +1,5 @@
-#ifndef  GRID_QCD_DWF_H
-#define  GRID_QCD_DWF_H
+#ifndef  GRID_QCD_WILSON_FERMION_5D_H
+#define  GRID_QCD_WILSON_FERMION_5D_H
 
 namespace Grid {
 
diff --git a/tests/Test_contfrac_cg.cc b/tests/Test_contfrac_cg.cc
index 83475254..b91faae2 100644
--- a/tests/Test_contfrac_cg.cc
+++ b/tests/Test_contfrac_cg.cc
@@ -82,6 +82,16 @@ int main (int argc, char ** argv)
   OverlapWilsonContFracZolotarevFermion Dcfz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,6.0);
   TestCGinversions<OverlapWilsonContFracZolotarevFermion>(Dcfz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
+
+  std::cout <<"OverlapWilsonPartialFractionTanhFermion  test"<<std::endl;
+  OverlapWilsonPartialFractionTanhFermion Dpf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+  TestCGinversions<OverlapWilsonPartialFractionTanhFermion>(Dpf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  std::cout <<"OverlapWilsonPartialFractionZolotarevFermion  test"<<std::endl;
+  OverlapWilsonPartialFractionZolotarevFermion Dpfz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,6.0);
+  TestCGinversions<OverlapWilsonPartialFractionZolotarevFermion>(Dpfz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+
   Grid_finalize();
 }
 template<class What> 
diff --git a/tests/Test_contfrac_even_odd.cc b/tests/Test_contfrac_even_odd.cc
index e13c1189..d3f5fb1b 100644
--- a/tests/Test_contfrac_even_odd.cc
+++ b/tests/Test_contfrac_even_odd.cc
@@ -57,6 +57,14 @@ int main (int argc, char ** argv)
   OverlapWilsonContFracZolotarevFermion Dcfz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,6.0);
   TestWhat<OverlapWilsonContFracZolotarevFermion>(Dcfz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
+  std::cout <<"OverlapWilsonPartialFractionTanhFermion  test"<<std::endl;
+  OverlapWilsonPartialFractionTanhFermion Dpf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+  TestWhat<OverlapWilsonPartialFractionTanhFermion>(Dpf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
+  std::cout <<"OverlapWilsonPartialFractionZolotarevFermion  test"<<std::endl;
+  OverlapWilsonPartialFractionZolotarevFermion Dpfz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,6.0);
+  TestWhat<OverlapWilsonPartialFractionZolotarevFermion>(Dpfz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+
   Grid_finalize();
 }
 

From cadd4310f69edf9b4078dd4be97d8f7c5f9b0181 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 5 Jun 2015 05:31:27 +0100
Subject: [PATCH 275/429] comment improvement

---
 lib/algorithms/iterative/NormalEquations.h | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/lib/algorithms/iterative/NormalEquations.h b/lib/algorithms/iterative/NormalEquations.h
index 3f40f8a5..1d5e0504 100644
--- a/lib/algorithms/iterative/NormalEquations.h
+++ b/lib/algorithms/iterative/NormalEquations.h
@@ -4,8 +4,7 @@
 namespace Grid {
 
   ///////////////////////////////////////////////////////////////////////////////////////////////////////
-  // Take a matrix and form a Red Black solver calling a Herm solver
-  // Use of RB info prevents making SchurRedBlackSolve conform to standard interface
+  // Take a matrix and form an NE solver calling a Herm solver
   ///////////////////////////////////////////////////////////////////////////////////////////////////////
   template<class Field> class NormalEquations : public OperatorFunction<Field>{
   private:
@@ -15,7 +14,7 @@ namespace Grid {
   public:
 
     /////////////////////////////////////////////////////
-    // Wrap the usual normal equations Schur trick
+    // Wrap the usual normal equations trick
     /////////////////////////////////////////////////////
   NormalEquations(SparseMatrixBase<Field> &Matrix, OperatorFunction<Field> &HermitianSolver) 
     :  _Matrix(Matrix), _HermitianSolver(HermitianSolver) {}; 

From 58cdcbb5e444fd2a12f669739aa3d72b3fdaade4 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Fri, 5 Jun 2015 10:02:36 +0100
Subject: [PATCH 276/429] Adding some wilson loop support

---
 lib/qcd/action/gauge/GaugeActionBase.h   |  20 ++++
 lib/qcd/action/gauge/WilsonGaugeAction.h |  24 +++++
 lib/qcd/utils/CovariantCshift.h          |  22 ++++
 lib/qcd/utils/WilsonLoops.h              | 130 +++++++++++++++++++++++
 4 files changed, 196 insertions(+)
 create mode 100644 lib/qcd/action/gauge/GaugeActionBase.h
 create mode 100644 lib/qcd/action/gauge/WilsonGaugeAction.h
 create mode 100644 lib/qcd/utils/CovariantCshift.h
 create mode 100644 lib/qcd/utils/WilsonLoops.h

diff --git a/lib/qcd/action/gauge/GaugeActionBase.h b/lib/qcd/action/gauge/GaugeActionBase.h
new file mode 100644
index 00000000..55141965
--- /dev/null
+++ b/lib/qcd/action/gauge/GaugeActionBase.h
@@ -0,0 +1,20 @@
+#ifndef QCD_GAUGE_ACTION_BASE
+#define QCD_GAUGE_ACTION_BASE
+namespace Grid {
+namespace QCD{
+
+template<class GaugeField>
+class GaugeActionBase { // derive this from TermInAction?
+
+ public:
+  virtual RealD S(const GaugeField &U)                        = 0;  // evaluate the action
+  virtual void  deriv(const GaugeField &U,GaugeField & dSdU ) = 0;  // evaluate the action derivative
+  virtual void  staple(const GaugeField &U,GaugeField & dSdU ) = 0;  // evaluate the action derivative
+  virtual void refresh(const GaugeField & ) {}; 
+  // Boundary conditions?
+  // Heatbath?
+  virtual ~GaugeActionBase() {};
+};
+
+}}
+#endif
diff --git a/lib/qcd/action/gauge/WilsonGaugeAction.h b/lib/qcd/action/gauge/WilsonGaugeAction.h
new file mode 100644
index 00000000..8ecb1f81
--- /dev/null
+++ b/lib/qcd/action/gauge/WilsonGaugeAction.h
@@ -0,0 +1,24 @@
+#ifndef QCD_WILSON_GAUGE_ACTION_H
+#define QCD_WILSON_GAUGE_ACTION_H
+
+////////////////////////////////////////////////////////////////////////
+// Wilson Gauge Action .. should I template the Nc etc..
+////////////////////////////////////////////////////////////////////////
+template<class GaugeField,class MatrixField>
+class WilsonGaugeAction : public GaugeActionBase<GaugeField> {
+ public:
+  virtual RealD S(const GaugeField &U) {
+    return WilsonLoops<MatrixField,GaugeField>::sumPlaquette(U);
+  };
+  virtual RealD deriv(GaugeField &U,GaugeField & dSdU ) {
+    WilsonLoops<MatrixField,GaugeField>::Staple(dSdU,U,mu);
+  };
+  virtual void  staple(const MatrixField &stap,GaugeField & U,int mu ) {
+    WilsonLoops<MatrixField,GaugeField>::Staple(stap,U,mu);
+  };
+};
+
+
+#endif
+
+#endif
diff --git a/lib/qcd/utils/CovariantCshift.h b/lib/qcd/utils/CovariantCshift.h
new file mode 100644
index 00000000..0c621069
--- /dev/null
+++ b/lib/qcd/utils/CovariantCshift.h
@@ -0,0 +1,22 @@
+#ifndef QCD_UTILS_COVARIANT_CSHIFT_H
+#define QCD_UTILS_COVARIANT_CSHIFT_H
+namespace Grid {
+namespace QCD {
+////////////////////////////////////////////////////////////////////////
+// Low performance implementation of CovariantCshift API
+////////////////////////////////////////////////////////////////////////
+template<class covariant,class gauge> Lattice<covariant> CovShiftForward(const Lattice<gauge> &Link, 
+									    int mu,
+									    const Lattice<covariant> &field)
+{
+  return Link*Cshift(field,mu,1);// moves towards negative mu
+}
+template<class covariant,class gauge> Lattice<covariant> CovShiftBackward(const Lattice<gauge> &Link, 
+									     int mu,
+									     const Lattice<covariant> &field)
+{
+  Lattice<covariant> tmp(field._grid);
+  tmp = adj(Link)*field;
+  return Cshift(tmp,mu,-1);// moves towards positive mu
+}
+}}
diff --git a/lib/qcd/utils/WilsonLoops.h b/lib/qcd/utils/WilsonLoops.h
new file mode 100644
index 00000000..f2d6d4de
--- /dev/null
+++ b/lib/qcd/utils/WilsonLoops.h
@@ -0,0 +1,130 @@
+#ifndef QCD_UTILS_WILSON_LOOPS_H
+#define QCD_UTILS_WILSON_LOOPS_H
+namespace Grid {
+namespace QCD {
+
+// Common wilson loop observables
+template<class GaugeMat,class GaugeLorentz>
+class WilsonLoops {
+public:
+
+  //////////////////////////////////////////////////
+  // directed plaquette oriented in mu,nu plane
+  //////////////////////////////////////////////////
+  static void dirPlaquette(GaugeMat &plaq,const std::vector<GaugeMat> &U, const int mu, const int nu)
+  {
+    plaq=CovShiftForward(U[mu],mu,U[nu])*adj(CovShiftForward(U[nu],nu,U[mu]));
+  }
+  //////////////////////////////////////////////////
+  // trace of directed plaquette oriented in mu,nu plane
+  //////////////////////////////////////////////////
+  static void traceDirPlaquette(LatticeComplex &plaq, const std::vector<GaugeMat> &U, const int mu, const int nu)
+  {
+    GaugeMat sp(U[0]._grid);
+    dirPlaquette(sp,U,mu,nu);
+    plaq=trace(sp);
+  }
+  //////////////////////////////////////////////////
+  // sum over all planes of plaquette
+  //////////////////////////////////////////////////
+  static void sitePlaquette(LatticeComplex &Plaq,const std::vector<GaugeMat> &U)
+  {
+    LatticeComplex sitePlaq(U[0]._grid);
+    for(int mu=1;mu<Nd;mu++){
+      for(int nu=0;nu<mu;nu++){
+	traceDirPlaquette(sitePlaq,U,mu,nu);
+	Plaq = Plaq + sitePlaq;
+      }
+    }
+  }
+  //////////////////////////////////////////////////
+  // sum over all x,y,z,t and over all planes of plaquette
+  //////////////////////////////////////////////////
+  static RealD sumPlaquette(const GaugeLorentz &Umu){
+    std::vector<GaugeMat> U(4,Umu._grid);
+    for(int mu=0;mu<Nd;mu++){
+      U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+    }
+
+    LatticeComplex Plaq(Umu._grid);
+    
+    sitePlaquette(Plaq,U);
+    
+    TComplex Tp = sum(Plaq);
+    Complex p  = TensorRemove(Tp);
+    return p.real();
+  }
+  //////////////////////////////////////////////////
+  // average over all x,y,z,t and over all planes of plaquette
+  //////////////////////////////////////////////////
+  static RealD avgPlaquette(const GaugeLorentz &Umu){
+
+    RealD sumplaq = sumPlaquette(Umu);
+    
+    double vol = Umu._grid->gSites();
+    
+    double faces = (1.0*Nd*(Nd-1))/2.0;
+    
+    return sumplaq/vol/faces/Nc; // Nd , Nc dependent... FIXME
+  }
+  //////////////////////////////////////////////////
+  // the sum over all staples on each site
+  //////////////////////////////////////////////////
+  static void Staple(GaugeMat &staple,GaugeLorentz &Umu,int mu){
+
+    std::vector<GaugeMat> U(4,Umu._grid);
+    for(int d=0;d<Nd;d++){
+      U[d] = peekIndex<LorentzIndex>(Umu,d);
+    }
+
+    staple = zero;
+    GaugeMat tmp(U[0]._grid);
+
+    for(int nu=0;nu<Nd;nu++){
+
+      if(nu != mu) {
+
+      // mu
+      // ^
+      // |__  nu
+
+      //    __                                         __
+      //      |         |                          
+      //    __|  =    __|                         *
+      //
+
+	staple   += CovShiftForward(U[nu],nu,U[mu])*Cshift(adj(U[nu]),mu,+1);
+
+      //
+      //  __        __                         
+      // |         |                          
+      // |__     = |                              *       __
+      //
+      //
+	tmp    = CovShiftForward (U[mu],mu,U[nu]);
+	staple+= CovShiftBackward(U[nu],nu,tmp);
+      
+      }
+    }
+  }
+
+  //////////////////////////////////////////////////////
+  // Similar to above for rectangle is required
+  //////////////////////////////////////////////////////
+  /*
+void siteRectangle(GaugeMat &plaq,const std::vector<GaugeMat> &U, const int mu, const int nu){
+RealD avgRectangle(const std::vector<GaugeMat> &U){}
+RealD avgRectangle(const std::vector<GaugeMat> &U, const int mu, const int nu){}
+void traceRectangle(LatticeComplex &plaq,const std::vector<GaugeMat> &U, const int mu, const int nu){}
+void siteRectangle(GaugeMat &plaq,const std::vector<GaugeMat> &U, const int mu, const int nu){}
+  */
+
+};
+}}
+
+
+
+
+
+
+#endif

From a8b86e747befe4ab6aab868c8a956998d0304d55 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Fri, 5 Jun 2015 10:15:31 +0100
Subject: [PATCH 277/429] merge to the head

---
 lib/Make.inc                    |   4 +-
 lib/algorithms/approx/.dirstamp |   0
 lib/communicator/.dirstamp      |   0
 lib/qcd/.dirstamp               |   0
 lib/qcd/QCD.h                   |   2 +
 lib/stencil/.dirstamp           |   0
 tests/Make.inc                  |   6 +-
 tests/Test_GaugeAction.cc       | 128 ++++++++++++++++++++++++++++++++
 8 files changed, 137 insertions(+), 3 deletions(-)
 delete mode 100644 lib/algorithms/approx/.dirstamp
 delete mode 100644 lib/communicator/.dirstamp
 delete mode 100644 lib/qcd/.dirstamp
 delete mode 100644 lib/stencil/.dirstamp
 create mode 100644 tests/Test_GaugeAction.cc

diff --git a/lib/Make.inc b/lib/Make.inc
index be0a939d..002f0f30 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Comparison.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/Dirac.h ./qcd/LinalgUtils.h ./qcd/QCD.h ./qcd/SpaceTimeGrid.h ./qcd/TwoSpinor.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Old/Grid_vComplexD.h ./simd/Old/Grid_vComplexF.h ./simd/Old/Grid_vInteger.h ./simd/Old/Grid_vRealD.h ./simd/Old/Grid_vRealF.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./Tensors.h ./Threads.h
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Comparison.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/Dirac.h ./qcd/LinalgUtils.h ./qcd/QCD.h ./qcd/SpaceTimeGrid.h ./qcd/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Old/Grid_vComplexD.h ./simd/Old/Grid_vComplexF.h ./simd/Old/Grid_vInteger.h ./simd/Old/Grid_vRealD.h ./simd/Old/Grid_vRealF.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./Tensors.h ./Threads.h
 
-CCFILES=./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/Dirac.cc ./qcd/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
+CCFILES=./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/action/gauge/WilsonGaugeAction.cc ./qcd/Dirac.cc ./qcd/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/algorithms/approx/.dirstamp b/lib/algorithms/approx/.dirstamp
deleted file mode 100644
index e69de29b..00000000
diff --git a/lib/communicator/.dirstamp b/lib/communicator/.dirstamp
deleted file mode 100644
index e69de29b..00000000
diff --git a/lib/qcd/.dirstamp b/lib/qcd/.dirstamp
deleted file mode 100644
index e69de29b..00000000
diff --git a/lib/qcd/QCD.h b/lib/qcd/QCD.h
index 4b5edc5c..c3f70c43 100644
--- a/lib/qcd/QCD.h
+++ b/lib/qcd/QCD.h
@@ -311,6 +311,8 @@ namespace QCD {
 #include <qcd/Dirac.h>
 #include <qcd/TwoSpinor.h>
 #include <qcd/LinalgUtils.h>
+#include <qcd/utils/CovariantCshift.h>
+#include <qcd/utils/WilsonLoops.h>
 #include <qcd/action/Actions.h>
 
 #endif
diff --git a/lib/stencil/.dirstamp b/lib/stencil/.dirstamp
deleted file mode 100644
index e69de29b..00000000
diff --git a/tests/Make.inc b/tests/Make.inc
index b525874d..533cbf17 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,9 @@
 
-bin_PROGRAMS = Test_cayley_cg Test_cayley_even_odd Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_even_odd Test_gamma Test_main Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_even_odd
+bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_even_odd Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_even_odd Test_gamma Test_main Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_even_odd
+
+
+Test_GaugeAction_SOURCES=Test_GaugeAction.cc
+Test_GaugeAction_LDADD=-lGrid
 
 
 Test_cayley_cg_SOURCES=Test_cayley_cg.cc
diff --git a/tests/Test_GaugeAction.cc b/tests/Test_GaugeAction.cc
new file mode 100644
index 00000000..07e7a4eb
--- /dev/null
+++ b/tests/Test_GaugeAction.cc
@@ -0,0 +1,128 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+
+
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+  std::vector<int> latt_size  ({16,16,16,32});
+  std::vector<int> clatt_size  ({4,4,4,8});
+  int orthodir=3;
+  int orthosz =latt_size[orthodir];
+    
+  GridCartesian     Fine(latt_size,simd_layout,mpi_layout);
+  GridCartesian     Coarse(clatt_size,simd_layout,mpi_layout);
+
+  LatticeGaugeField Umu(&Fine);
+
+  std::vector<LatticeColourMatrix> U(4,&Fine);
+  
+  NerscField header;
+  
+  std::string file("./ckpoint_lat.4000");
+  readNerscConfiguration(Umu,header,file);
+
+  for(int mu=0;mu<Nd;mu++){
+    U[mu] = peekIndex<3>(Umu,mu);
+  }
+
+  // Painful ; fix syntactical niceness : to check reader
+  LatticeComplex LinkTrace(&Fine);
+  LinkTrace=zero;
+  for(int mu=0;mu<Nd;mu++){
+    LinkTrace = LinkTrace + trace(U[mu]);
+  }
+
+  // (1+2+3)=6 = N(N-1)/2 terms // equals to Double S Chorma
+  // in LatticeGaugeField out Plaq
+  // class WilsonLoop {
+  // RealD  plaquette(LatticeGaugeField &Umu);
+  // void   staple(LatticeSomethingorOther,LatticeGaugeField &Umu);
+  // RealD  rectangle(LatticeGaugeField &Umu);
+  // LatticeComplex sitePlaquette()
+  // } 
+  // covariantCshift ???
+  // GaugeActionBase
+  // GaugeActionPlaquette
+  // GaugeActionPlaquettePlusRectangle
+  // GaugeActionIwasaki
+  // GaugeActionSymanzik
+  // GaugeActionWilson
+  // Heatbath and quenched update.
+  //  
+  LatticeColourMatrix tmpU(&Fine);
+
+  LatticeComplex Plaq(&Fine);
+  LatticeComplex cPlaq(&Coarse);
+  Plaq = zero;
+  for(int mu=1;mu<Nd;mu++){
+    for(int nu=0;nu<mu;nu++){
+      Plaq = Plaq + trace(CovShiftForward(U[mu],mu,U[nu])*adj(CovShiftForward(U[nu],nu,U[mu])));
+    }
+  }
+  
+  double vol = Fine.gSites();
+  Complex PlaqScale(1.0/vol/6.0/3.0);
+  RealD   StapScale(1.0/vol/6.0/3.0);
+  std::cout <<"PlaqScale" << PlaqScale<<std::endl;
+  std::vector<TComplex> Plaq_T(orthosz);
+  sliceSum(Plaq,Plaq_T,Nd-1);
+  int Nt = Plaq_T.size();
+
+
+  TComplex Plaq_T_sum; 
+  Plaq_T_sum=zero;
+  for(int t=0;t<Nt;t++){
+    Plaq_T_sum = Plaq_T_sum+Plaq_T[t];
+    Complex Pt=TensorRemove(Plaq_T[t]);
+    std::cout << "sliced ["<<t<<"]" <<Pt*PlaqScale*Real(Nt) << std::endl;
+  }
+
+  {
+    Complex Pt = TensorRemove(Plaq_T_sum);
+    std::cout << "total " <<Pt*PlaqScale<<std::endl;
+  }  
+
+  TComplex Tp = sum(Plaq);
+  Complex p  = TensorRemove(Tp);
+  std::cout << "calculated plaquettes " <<p*PlaqScale<<std::endl;
+
+  RealD avg_plaq =  WilsonLoops::avgPlaquette(Umu);
+  std::cout << "NEW : calculated real plaquettes " <<avg_plaq<<std::endl;
+
+  RealD stap_plaq=0.0;
+  LatticeColourMatrix stap(&Fine);
+  LatticeComplex stap_tr(&Fine);
+  for(int mu=0;mu<Nd;mu++){
+    WilsonLoops::Staple(stap,Umu,mu);
+    stap_tr = trace(stap*adj(U[mu]));
+    TComplex Ts = sum(stap_tr);
+    Complex s  = TensorRemove(Ts);
+    stap_plaq+=real(s);
+  }
+  std::cout << "NEW : plaquette via staples"<< stap_plaq*StapScale*0.25<< std::endl;
+  Complex LinkTraceScale(1.0/vol/4.0/3.0);
+  TComplex Tl = sum(LinkTrace);
+  Complex l  = TensorRemove(Tl);
+  std::cout << "calculated link trace " <<l*LinkTraceScale<<std::endl;
+
+  sumBlocks(cPlaq,Plaq);
+  TComplex TcP = sum(cPlaq);
+  Complex ll= TensorRemove(TcP);
+  std::cout << "coarsened plaquettes sum to " <<ll*PlaqScale<<std::endl;
+
+
+  Grid_finalize();
+  }
+
+
+

From f3e60a9feb27087c88c16d011aa607ba10dbcac3 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 5 Jun 2015 10:17:10 +0100
Subject: [PATCH 278/429] Rework the linop support to get different forms of
 red black schur solver Moo on diag, or MooInv Moe MeeInv Meo

---
 lib/algorithms/LinearOperator.h              | 154 +++++++++++++------
 lib/algorithms/SparseMatrix.h                |  26 ----
 lib/algorithms/iterative/ConjugateGradient.h |  21 +--
 lib/algorithms/iterative/SchurRedBlack.h     |  15 +-
 tests/Test_cayley_cg.cc                      |   6 +-
 tests/Test_cayley_even_odd.cc                |  11 +-
 tests/Test_contfrac_cg.cc                    |   6 +-
 tests/Test_contfrac_even_odd.cc              |  11 +-
 tests/Test_dwf_cg_prec.cc                    |   2 +-
 tests/Test_dwf_cg_schur.cc                   |   2 +-
 tests/Test_dwf_cg_unprec.cc                  |   2 +-
 tests/Test_dwf_even_odd.cc                   |  10 +-
 tests/Test_wilson_cg_prec.cc                 |   2 +-
 tests/Test_wilson_cg_schur.cc                |   2 +-
 tests/Test_wilson_cg_unprec.cc               |   2 +-
 tests/Test_wilson_even_odd.cc                |   9 +-
 16 files changed, 155 insertions(+), 126 deletions(-)

diff --git a/lib/algorithms/LinearOperator.h b/lib/algorithms/LinearOperator.h
index 5839f2c1..710fcde7 100644
--- a/lib/algorithms/LinearOperator.h
+++ b/lib/algorithms/LinearOperator.h
@@ -18,84 +18,144 @@ namespace Grid {
     public:
       virtual void Op     (const Field &in, Field &out) = 0; // Abstract base
       virtual void AdjOp  (const Field &in, Field &out) = 0; // Abstract base
+      virtual void HermOpAndNorm(const Field &in, Field &out,double &n1,double &n2)=0;
     };
 
-  /////////////////////////////////////////////////////////////////////////////////////////////
-  // Hermitian operators are self adjoint and only require Op to be defined, so refine the base
-  /////////////////////////////////////////////////////////////////////////////////////////////
-    template<class Field> class HermitianOperatorBase : public LinearOperatorBase<Field> {
-    public:
-      virtual void OpAndNorm(const Field &in, Field &out,double &n1,double &n2)=0;
-      void AdjOp(const Field &in, Field &out) {
-	Op(in,out);
-      };
-      void Op(const Field &in, Field &out) {
-	double n1,n2;
-	OpAndNorm(in,out,n1,n2);
-      };
-    };
 
   /////////////////////////////////////////////////////////////////////////////////////////////
-  // Whereas non hermitian takes a generic sparse matrix (e.g. lattice action)
-  // conforming to sparse matrix interface and builds the full checkerboard non-herm operator
-  // Op and AdjOp distinct.
   // By sharing the class for Sparse Matrix across multiple operator wrappers, we can share code
   // between RB and non-RB variants. Sparse matrix is like the fermion action def, and then
   // the wrappers implement the specialisation of "Op" and "AdjOp" to the cases minimising
   // replication of code.
+  //
+  // I'm not entirely happy with implementation; to share the Schur code between herm and non-herm
+  // while still having a "OpAndNorm" in the abstract base I had to implement it in both cases
+  // with an assert trap in the non-herm. This isn't right; there must be a better C++ way to
+  // do it, but I fear it required multiple inheritance and mixed in abstract base classes
   /////////////////////////////////////////////////////////////////////////////////////////////
+
+    ////////////////////////////////////////////////////////////////////
+    // Construct herm op from non-herm matrix
+    ////////////////////////////////////////////////////////////////////
     template<class Matrix,class Field>
-    class NonHermitianOperator : public LinearOperatorBase<Field> {
+    class MdagMLinearOperator : public LinearOperatorBase<Field> {
       Matrix &_Mat;
     public:
-      NonHermitianOperator(Matrix &Mat): _Mat(Mat){};
+    MdagMLinearOperator(Matrix &Mat): _Mat(Mat){};
       void Op     (const Field &in, Field &out){
 	_Mat.M(in,out);
       }
       void AdjOp     (const Field &in, Field &out){
 	_Mat.Mdag(in,out);
       }
+      void HermOpAndNorm(const Field &in, Field &out,double &n1,double &n2){
+	_Mat.MdagM(in,out,n1,n2);
+      }
     };
-    
-    ////////////////////////////////////////////////////////////////////////////////////
-    // Redblack Non hermitian wrapper
-    ////////////////////////////////////////////////////////////////////////////////////
+
+    ////////////////////////////////////////////////////////////////////
+    // Wrap an already herm matrix
+    ////////////////////////////////////////////////////////////////////
     template<class Matrix,class Field>
-    class NonHermitianCheckerBoardedOperator : public LinearOperatorBase<Field> {
+    class HermitianLinearOperator : public LinearOperatorBase<Field> {
       Matrix &_Mat;
     public:
-      NonHermitianCheckerBoardedOperator(Matrix &Mat): _Mat(Mat){};
+    HermitianLinearOperator(Matrix &Mat): _Mat(Mat){};
       void Op     (const Field &in, Field &out){
-	_Mat.Mpc(in,out);
+	_Mat.M(in,out);
       }
-      void AdjOp     (const Field &in, Field &out){ //
-	_Mat.MpcDag(in,out);
+      void AdjOp     (const Field &in, Field &out){
+	_Mat.M(in,out);
+      }
+      void HermOpAndNorm(const Field &in, Field &out,double &n1,double &n2){
+	ComplexD dot;
+
+	_Mat.M(in,out);
+	
+	dot= innerProduct(in,out);
+	n1=real(dot);
+
+	dot = innerProduct(out,out);
+	n2=real(dot);
       }
     };
 
-    ////////////////////////////////////////////////////////////////////////////////////
-    // Hermitian wrapper
-    ////////////////////////////////////////////////////////////////////////////////////
-    template<class Matrix,class Field>
-    class HermitianOperator : public HermitianOperatorBase<Field> {
-      Matrix &_Mat;
+    //////////////////////////////////////////////////////////
+    // Even Odd Schur decomp operators; there are several
+    // ways to introduce the even odd checkerboarding
+    //////////////////////////////////////////////////////////
+
+    template<class Field>
+      class SchurOperatorBase :  public LinearOperatorBase<Field> {
     public:
-      HermitianOperator(Matrix &Mat): _Mat(Mat) {};
-      void OpAndNorm(const Field &in, Field &out,double &n1,double &n2){
-	return _Mat.MdagM(in,out,n1,n2);
+      virtual  RealD Mpc      (const Field &in, Field &out) =0;
+      virtual  RealD MpcDag   (const Field &in, Field &out) =0;
+      virtual void MpcDagMpc(const Field &in, Field &out,RealD &ni,RealD &no) {
+	Field tmp(in._grid);
+	ni=Mpc(in,tmp);
+	no=MpcDag(tmp,out);
+      }
+      void HermOpAndNorm(const Field &in, Field &out,RealD &n1,RealD &n2){
+	MpcDagMpc(in,out,n1,n2);
+      }
+      void Op     (const Field &in, Field &out){
+	Mpc(in,out);
+      }
+      void AdjOp     (const Field &in, Field &out){ 
+	MpcDag(in,out);
       }
     };
-
-    ////////////////////////////////////////////////////////////////////////////////////
-    // Hermitian CheckerBoarded wrapper
-    ////////////////////////////////////////////////////////////////////////////////////
     template<class Matrix,class Field>
-    class HermitianCheckerBoardedOperator : public HermitianOperatorBase<Field> {
+      class SchurDiagMooeeOperator :  public SchurOperatorBase<Field> {
       Matrix &_Mat;
     public:
-      HermitianCheckerBoardedOperator(Matrix &Mat): _Mat(Mat) {};
-      void OpAndNorm(const Field &in, Field &out,RealD &n1,RealD &n2){
-	_Mat.MpcDagMpc(in,out,n1,n2);
+      SchurDiagMooeeOperator (Matrix &Mat): _Mat(Mat){};
+      virtual  RealD Mpc      (const Field &in, Field &out) {
+	Field tmp(in._grid);
+
+	_Mat.Meooe(in,tmp);
+	_Mat.MooeeInv(tmp,out);
+	_Mat.Meooe(out,tmp);
+
+	_Mat.Mooee(in,out);
+	return axpy_norm(out,-1.0,tmp,out);
+      }
+      virtual  RealD MpcDag   (const Field &in, Field &out){
+	Field tmp(in._grid);
+
+	_Mat.MeooeDag(in,tmp);
+	_Mat.MooeeInvDag(tmp,out);
+	_Mat.MeooeDag(out,tmp);
+
+	_Mat.MooeeDag(in,out);
+	return axpy_norm(out,-1.0,tmp,out);
+      }
+    };
+    template<class Matrix,class Field>
+      class SchurDiagOneOperator :  public SchurOperatorBase<Field> {
+      Matrix &_Mat;
+    public:
+      SchurDiagOneOperator (Matrix &Mat): _Mat(Mat){};
+
+      virtual  RealD Mpc      (const Field &in, Field &out) {
+	Field tmp(in._grid);
+
+	_Mat.Meooe(in,tmp);
+	_Mat.MooeeInv(tmp,out);
+	_Mat.Meooe(out,tmp);
+	_Mat.MooeeInv(tmp,out);
+
+	return axpy_norm(out,-1.0,tmp,in);
+      }
+      virtual  RealD MpcDag   (const Field &in, Field &out){
+	Field tmp(in._grid);
+
+	_Mat.MooeeInvDag(in,out);
+	_Mat.MeooeDag(out,tmp);
+	_Mat.MooeeInvDag(tmp,out);
+	_Mat.MeooeDag(out,tmp);
+
+	return axpy_norm(out,-1.0,tmp,in);
       }
     };
 
@@ -106,10 +166,6 @@ namespace Grid {
     public:
       virtual void operator() (LinearOperatorBase<Field> &Linop, const Field &in, Field &out) = 0;
     };
-    template<class Field> class HermitianOperatorFunction {
-    public:
-      virtual void operator() (HermitianOperatorBase<Field> &Linop, const Field &in, Field &out) = 0;
-    };
 
     // FIXME : To think about
 
diff --git a/lib/algorithms/SparseMatrix.h b/lib/algorithms/SparseMatrix.h
index 9c955e9a..9146648f 100644
--- a/lib/algorithms/SparseMatrix.h
+++ b/lib/algorithms/SparseMatrix.h
@@ -36,32 +36,6 @@ namespace Grid {
       virtual  void MooeeDag    (const Field &in, Field &out)=0;
       virtual  void MooeeInvDag (const Field &in, Field &out)=0;
 
-      // Schur decomp operators
-      virtual  RealD Mpc      (const Field &in, Field &out) {
-	Field tmp(in._grid);
-
-	Meooe(in,tmp);
-	MooeeInv(tmp,out);
-	Meooe(out,tmp);
-
-	Mooee(in,out);
-	return axpy_norm(out,-1.0,tmp,out);
-      }
-      virtual  RealD MpcDag   (const Field &in, Field &out){
-	Field tmp(in._grid);
-
-	MeooeDag(in,tmp);
-	MooeeInvDag(tmp,out);
-	MeooeDag(out,tmp);
-
-	MooeeDag(in,out);
-	return axpy_norm(out,-1.0,tmp,out);
-      }
-      virtual void MpcDagMpc(const Field &in, Field &out,RealD &ni,RealD &no) {
-	Field tmp(in._grid);
-	ni=Mpc(in,tmp);
-	no=MpcDag(tmp,out);
-      }
     };
 
 }
diff --git a/lib/algorithms/iterative/ConjugateGradient.h b/lib/algorithms/iterative/ConjugateGradient.h
index d4f89662..e6915d83 100644
--- a/lib/algorithms/iterative/ConjugateGradient.h
+++ b/lib/algorithms/iterative/ConjugateGradient.h
@@ -9,17 +9,17 @@ namespace Grid {
     /////////////////////////////////////////////////////////////
 
   template<class Field> 
-    class ConjugateGradient : public HermitianOperatorFunction<Field> {
+    class ConjugateGradient : public OperatorFunction<Field> {
 public:                                                
     RealD   Tolerance;
     Integer MaxIterations;
     int verbose;
     ConjugateGradient(RealD tol,Integer maxit) : Tolerance(tol), MaxIterations(maxit) { 
-      verbose=0;
+      verbose=1;
     };
 
 
-    void operator() (HermitianOperatorBase<Field> &Linop,const Field &src, Field &psi){
+    void operator() (LinearOperatorBase<Field> &Linop,const Field &src, Field &psi){
 
       psi.checkerboard = src.checkerboard;
       conformable(psi,src);
@@ -33,7 +33,7 @@ public:
       //Initial residual computation & set up
       RealD guess = norm2(psi);
       
-      Linop.OpAndNorm(psi,mmp,d,b);
+      Linop.HermOpAndNorm(psi,mmp,d,b);
       
       r= src-mmp;
       p= r;
@@ -65,7 +65,7 @@ public:
 	
 	c=cp;
 	
-	Linop.OpAndNorm(p,mmp,d,qq);
+	Linop.HermOpAndNorm(p,mmp,d,qq);
 
 	RealD    qqck = norm2(mmp);
 	ComplexD dck  = innerProduct(p,mmp);
@@ -86,19 +86,10 @@ public:
 	  
 	if (verbose) std::cout<<"ConjugateGradient: Iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
 
-	// Hack
-	if (0) {
-	  Field   tt(src);
-      	  Linop.Op(psi,mmp);
-	  tt=mmp-src;
-	  RealD resnorm = norm2(tt);
-	  std::cout<<"ConjugateGradient: Iteration " <<k<<" true residual "<<resnorm << " computed " << cp <<std::endl;
-	}
-
 	// Stopping condition
 	if ( cp <= rsq ) { 
 	  
-	  Linop.Op(psi,mmp);
+	  Linop.HermOpAndNorm(psi,mmp,d,qq);
 	  p=mmp-src;
 	  
 	  RealD mmpnorm = sqrt(norm2(mmp));
diff --git a/lib/algorithms/iterative/SchurRedBlack.h b/lib/algorithms/iterative/SchurRedBlack.h
index 109f554b..01f2e3d5 100644
--- a/lib/algorithms/iterative/SchurRedBlack.h
+++ b/lib/algorithms/iterative/SchurRedBlack.h
@@ -40,16 +40,16 @@ namespace Grid {
   // Take a matrix and form a Red Black solver calling a Herm solver
   // Use of RB info prevents making SchurRedBlackSolve conform to standard interface
   ///////////////////////////////////////////////////////////////////////////////////////////////////////
-  template<class Field> class SchurRedBlackSolve {
+  template<class Field> class SchurRedBlackDiagMooeeSolve {
   private:
-    HermitianOperatorFunction<Field> & _HermitianRBSolver;
+    OperatorFunction<Field> & _HermitianRBSolver;
     int CBfactorise;
   public:
 
     /////////////////////////////////////////////////////
     // Wrap the usual normal equations Schur trick
     /////////////////////////////////////////////////////
-  SchurRedBlackSolve(HermitianOperatorFunction<Field> &HermitianRBSolver)  :
+  SchurRedBlackDiagMooeeSolve(OperatorFunction<Field> &HermitianRBSolver)  :
      _HermitianRBSolver(HermitianRBSolver) 
     { 
       CBfactorise=0;
@@ -62,6 +62,8 @@ namespace Grid {
       // FIXME use CBfactorise to control schur decomp
       GridBase *grid = _Matrix.RedBlackGrid();
       GridBase *fgrid= _Matrix.Grid();
+
+      SchurDiagMooeeOperator<Matrix,Field> _HermOpEO(_Matrix);
  
       Field src_e(grid);
       Field src_o(grid);
@@ -80,12 +82,13 @@ namespace Grid {
       _Matrix.MooeeInv(src_e,tmp);     assert(  tmp.checkerboard ==Even);
       _Matrix.Meooe   (tmp,Mtmp);      assert( Mtmp.checkerboard ==Odd);     
       tmp=src_o-Mtmp;                  assert(  tmp.checkerboard ==Odd);     
-      _Matrix.MpcDag(tmp,src_o);       assert(src_o.checkerboard ==Odd);       
+
+      // get the right MpcDag
+      _HermOpEO.MpcDag(tmp,src_o);     assert(src_o.checkerboard ==Odd);       
 
       //////////////////////////////////////////////////////////////
       // Call the red-black solver
       //////////////////////////////////////////////////////////////
-      HermitianCheckerBoardedOperator<Matrix,Field> _HermOpEO(_Matrix);
       std::cout << "SchurRedBlack solver calling the MpcDagMp solver" <<std::endl;
       _HermitianRBSolver(_HermOpEO,src_o,sol_o);  assert(sol_o.checkerboard==Odd);
 
@@ -105,7 +108,7 @@ namespace Grid {
       RealD ns = norm2(in);
       RealD nr = norm2(resid);
 
-      std::cout << "SchurRedBlack solver true unprec resid "<< sqrt(nr/ns) <<" nr "<< nr <<" ns "<<ns << std::endl;
+      std::cout << "SchurRedBlackDiagMooee solver true unprec resid "<< sqrt(nr/ns) <<" nr "<< nr <<" ns "<<ns << std::endl;
     }     
   };
 
diff --git a/tests/Test_cayley_cg.cc b/tests/Test_cayley_cg.cc
index 4510d4b5..32174035 100644
--- a/tests/Test_cayley_cg.cc
+++ b/tests/Test_cayley_cg.cc
@@ -130,7 +130,7 @@ void  TestCGunprec(What & Ddwf,
   LatticeFermion src   (FGrid); random(*RNG5,src);
   LatticeFermion result(FGrid); result=zero;
 
-  HermitianOperator<What,LatticeFermion> HermOp(Ddwf);
+  MdagMLinearOperator<What,LatticeFermion> HermOp(Ddwf);
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   CG(HermOp,src,result);
 
@@ -149,7 +149,7 @@ void  TestCGprec(What & Ddwf,
   pickCheckerboard(Odd,src_o,src);
   result_o=zero;
 
-  HermitianCheckerBoardedOperator<What,LatticeFermion> HermOpEO(Ddwf);
+  SchurDiagMooeeOperator<What,LatticeFermion> HermOpEO(Ddwf);
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   CG(HermOpEO,src_o,result_o);
 }
@@ -167,6 +167,6 @@ void  TestCGschur(What & Ddwf,
   LatticeFermion result(FGrid); result=zero;
 
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
-  SchurRedBlackSolve<LatticeFermion> SchurSolver(CG);
+  SchurRedBlackDiagMooeeSolve<LatticeFermion> SchurSolver(CG);
   SchurSolver(Ddwf,src,result);
 }
diff --git a/tests/Test_cayley_even_odd.cc b/tests/Test_cayley_even_odd.cc
index df28981b..4b5630f7 100644
--- a/tests/Test_cayley_even_odd.cc
+++ b/tests/Test_cayley_even_odd.cc
@@ -214,7 +214,7 @@ void  TestWhat(What & Ddwf,
   std::cout << "norm diff   "<< norm2(err)<< std::endl;
 
   std::cout<<"=============================================================="<<std::endl;
-  std::cout<<"= Test MpcDagMpc is Hermitian              "<<std::endl;
+  std::cout<<"= Test DiagMoo MpcDagMpc is Hermitian              "<<std::endl;
   std::cout<<"=============================================================="<<std::endl;
   
   random(*RNG5,phi);
@@ -225,11 +225,12 @@ void  TestWhat(What & Ddwf,
   pickCheckerboard(Odd ,phi_o,phi);
   RealD t1,t2;
 
-  Ddwf.MpcDagMpc(chi_e,dchi_e,t1,t2);
-  Ddwf.MpcDagMpc(chi_o,dchi_o,t1,t2);
+  SchurDiagMooeeOperator<What,LatticeFermion> HermOpEO(Ddwf);
+  HermOpEO.MpcDagMpc(chi_e,dchi_e,t1,t2);
+  HermOpEO.MpcDagMpc(chi_o,dchi_o,t1,t2);
 
-  Ddwf.MpcDagMpc(phi_e,dphi_e,t1,t2);
-  Ddwf.MpcDagMpc(phi_o,dphi_o,t1,t2);
+  HermOpEO.MpcDagMpc(phi_e,dphi_e,t1,t2);
+  HermOpEO.MpcDagMpc(phi_o,dphi_o,t1,t2);
 
   pDce = innerProduct(phi_e,dchi_e);
   pDco = innerProduct(phi_o,dchi_o);
diff --git a/tests/Test_contfrac_cg.cc b/tests/Test_contfrac_cg.cc
index b91faae2..5851f5e6 100644
--- a/tests/Test_contfrac_cg.cc
+++ b/tests/Test_contfrac_cg.cc
@@ -121,7 +121,7 @@ void  TestCGunprec(What & Ddwf,
   LatticeFermion src   (FGrid); random(*RNG5,src);
   LatticeFermion result(FGrid); result=zero;
 
-  HermitianOperator<What,LatticeFermion> HermOp(Ddwf);
+  MdagMLinearOperator<What,LatticeFermion> HermOp(Ddwf);
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   CG(HermOp,src,result);
 
@@ -140,7 +140,7 @@ void  TestCGprec(What & Ddwf,
   pickCheckerboard(Odd,src_o,src);
   result_o=zero;
 
-  HermitianCheckerBoardedOperator<What,LatticeFermion> HermOpEO(Ddwf);
+  SchurDiagMooeeOperator<What,LatticeFermion> HermOpEO(Ddwf);
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   CG(HermOpEO,src_o,result_o);
 }
@@ -158,6 +158,6 @@ void  TestCGschur(What & Ddwf,
   LatticeFermion result(FGrid); result=zero;
 
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
-  SchurRedBlackSolve<LatticeFermion> SchurSolver(CG);
+  SchurRedBlackDiagMooeeSolve<LatticeFermion> SchurSolver(CG);
   SchurSolver(Ddwf,src,result);
 }
diff --git a/tests/Test_contfrac_even_odd.cc b/tests/Test_contfrac_even_odd.cc
index d3f5fb1b..a246b904 100644
--- a/tests/Test_contfrac_even_odd.cc
+++ b/tests/Test_contfrac_even_odd.cc
@@ -211,11 +211,12 @@ void  TestWhat(What & Ddwf,
   pickCheckerboard(Odd ,phi_o,phi);
   RealD t1,t2;
 
-  Ddwf.MpcDagMpc(chi_e,dchi_e,t1,t2);
-  Ddwf.MpcDagMpc(chi_o,dchi_o,t1,t2);
-
-  Ddwf.MpcDagMpc(phi_e,dphi_e,t1,t2);
-  Ddwf.MpcDagMpc(phi_o,dphi_o,t1,t2);
+  SchurDiagMooeeOperator<What,LatticeFermion> HermOpEO(Ddwf);
+  HermOpEO.MpcDagMpc(chi_e,dchi_e,t1,t2);
+  HermOpEO.MpcDagMpc(chi_o,dchi_o,t1,t2);
+  
+  HermOpEO.MpcDagMpc(phi_e,dphi_e,t1,t2);
+  HermOpEO.MpcDagMpc(phi_o,dphi_o,t1,t2);
 
   pDce = innerProduct(phi_e,dchi_e);
   pDco = innerProduct(phi_o,dchi_o);
diff --git a/tests/Test_dwf_cg_prec.cc b/tests/Test_dwf_cg_prec.cc
index 0cf86a19..32a0c3ae 100644
--- a/tests/Test_dwf_cg_prec.cc
+++ b/tests/Test_dwf_cg_prec.cc
@@ -50,7 +50,7 @@ int main (int argc, char ** argv)
   pickCheckerboard(Odd,src_o,src);
   result_o=zero;
 
-  HermitianCheckerBoardedOperator<DomainWallFermion,LatticeFermion> HermOpEO(Ddwf);
+  SchurDiagMooeeOperator<DomainWallFermion,LatticeFermion> HermOpEO(Ddwf);
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   CG(HermOpEO,src_o,result_o);
 
diff --git a/tests/Test_dwf_cg_schur.cc b/tests/Test_dwf_cg_schur.cc
index aac4d3fd..d080045b 100644
--- a/tests/Test_dwf_cg_schur.cc
+++ b/tests/Test_dwf_cg_schur.cc
@@ -46,7 +46,7 @@ int main (int argc, char ** argv)
   DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
 
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
-  SchurRedBlackSolve<LatticeFermion> SchurSolver(CG);
+  SchurRedBlackDiagMooeeSolve<LatticeFermion> SchurSolver(CG);
   SchurSolver(Ddwf,src,result);
 
   Grid_finalize();
diff --git a/tests/Test_dwf_cg_unprec.cc b/tests/Test_dwf_cg_unprec.cc
index 5c9e7ad3..7a13be43 100644
--- a/tests/Test_dwf_cg_unprec.cc
+++ b/tests/Test_dwf_cg_unprec.cc
@@ -45,7 +45,7 @@ int main (int argc, char ** argv)
   RealD M5=1.8;
   DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
 
-  HermitianOperator<DomainWallFermion,LatticeFermion> HermOp(Ddwf);
+  MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermOp(Ddwf);
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   CG(HermOp,src,result);
 
diff --git a/tests/Test_dwf_even_odd.cc b/tests/Test_dwf_even_odd.cc
index ac47bbf9..82b39f68 100644
--- a/tests/Test_dwf_even_odd.cc
+++ b/tests/Test_dwf_even_odd.cc
@@ -186,11 +186,13 @@ int main (int argc, char ** argv)
   pickCheckerboard(Odd ,phi_o,phi);
   RealD t1,t2;
 
-  Ddwf.MpcDagMpc(chi_e,dchi_e,t1,t2);
-  Ddwf.MpcDagMpc(chi_o,dchi_o,t1,t2);
 
-  Ddwf.MpcDagMpc(phi_e,dphi_e,t1,t2);
-  Ddwf.MpcDagMpc(phi_o,dphi_o,t1,t2);
+  SchurDiagMooeeOperator<DomainWallFermion,LatticeFermion> HermOpEO(Ddwf);
+  HermOpEO.MpcDagMpc(chi_e,dchi_e,t1,t2);
+  HermOpEO.MpcDagMpc(chi_o,dchi_o,t1,t2);
+
+  HermOpEO.MpcDagMpc(phi_e,dphi_e,t1,t2);
+  HermOpEO.MpcDagMpc(phi_o,dphi_o,t1,t2);
 
   pDce = innerProduct(phi_e,dchi_e);
   pDco = innerProduct(phi_o,dchi_o);
diff --git a/tests/Test_wilson_cg_prec.cc b/tests/Test_wilson_cg_prec.cc
index e376349c..dd8f2821 100644
--- a/tests/Test_wilson_cg_prec.cc
+++ b/tests/Test_wilson_cg_prec.cc
@@ -53,7 +53,7 @@ int main (int argc, char ** argv)
   pickCheckerboard(Odd,src_o,src);
   result_o=zero;
 
-  HermitianCheckerBoardedOperator<WilsonFermion,LatticeFermion> HermOpEO(Dw);
+  SchurDiagMooeeOperator<WilsonFermion,LatticeFermion> HermOpEO(Dw);
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   CG(HermOpEO,src_o,result_o);
 
diff --git a/tests/Test_wilson_cg_schur.cc b/tests/Test_wilson_cg_schur.cc
index 28db1d4b..56abcd4a 100644
--- a/tests/Test_wilson_cg_schur.cc
+++ b/tests/Test_wilson_cg_schur.cc
@@ -40,7 +40,7 @@ int main (int argc, char ** argv)
   WilsonFermion Dw(Umu,Grid,RBGrid,mass);
 
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
-  SchurRedBlackSolve<LatticeFermion> SchurSolver(CG);
+  SchurRedBlackDiagMooeeSolve<LatticeFermion> SchurSolver(CG);
 
   SchurSolver(Dw,src,result);
   
diff --git a/tests/Test_wilson_cg_unprec.cc b/tests/Test_wilson_cg_unprec.cc
index 905dfde5..c9f2856b 100644
--- a/tests/Test_wilson_cg_unprec.cc
+++ b/tests/Test_wilson_cg_unprec.cc
@@ -49,7 +49,7 @@ int main (int argc, char ** argv)
   RealD mass=0.5;
   WilsonFermion Dw(Umu,Grid,RBGrid,mass);
 
-  HermitianOperator<WilsonFermion,LatticeFermion> HermOp(Dw);
+  MdagMLinearOperator<WilsonFermion,LatticeFermion> HermOp(Dw);
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   CG(HermOp,src,result);
 
diff --git a/tests/Test_wilson_even_odd.cc b/tests/Test_wilson_even_odd.cc
index 3ebc4709..16019c9c 100644
--- a/tests/Test_wilson_even_odd.cc
+++ b/tests/Test_wilson_even_odd.cc
@@ -177,11 +177,12 @@ int main (int argc, char ** argv)
   pickCheckerboard(Odd ,phi_o,phi);
   RealD t1,t2;
 
-  Dw.MpcDagMpc(chi_e,dchi_e,t1,t2);
-  Dw.MpcDagMpc(chi_o,dchi_o,t1,t2);
+  SchurDiagMooeeOperator<WilsonFermion,LatticeFermion> HermOpEO(Dw);
+  HermOpEO.MpcDagMpc(chi_e,dchi_e,t1,t2);
+  HermOpEO.MpcDagMpc(chi_o,dchi_o,t1,t2);
 
-  Dw.MpcDagMpc(phi_e,dphi_e,t1,t2);
-  Dw.MpcDagMpc(phi_o,dphi_o,t1,t2);
+  HermOpEO.MpcDagMpc(phi_e,dphi_e,t1,t2);
+  HermOpEO.MpcDagMpc(phi_o,dphi_o,t1,t2);
 
   pDce = innerProduct(phi_e,dchi_e);
   pDco = innerProduct(phi_o,dchi_o);

From 1d7f9567ee6e23c91ca40c03b42695bf61be0f15 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Fri, 5 Jun 2015 10:19:42 +0100
Subject: [PATCH 279/429] Endif terminated

---
 lib/qcd/utils/CovariantCshift.h | 1 +
 1 file changed, 1 insertion(+)

diff --git a/lib/qcd/utils/CovariantCshift.h b/lib/qcd/utils/CovariantCshift.h
index 0c621069..e090f497 100644
--- a/lib/qcd/utils/CovariantCshift.h
+++ b/lib/qcd/utils/CovariantCshift.h
@@ -20,3 +20,4 @@ template<class covariant,class gauge> Lattice<covariant> CovShiftBackward(const
   return Cshift(tmp,mu,-1);// moves towards positive mu
 }
 }}
+#endif

From d86c248d05433d0718e067412fa6d60d752d8c72 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Fri, 5 Jun 2015 10:21:28 +0100
Subject: [PATCH 280/429] Fix

---
 lib/Make.inc | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/Make.inc b/lib/Make.inc
index 002f0f30..94a566fb 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
 HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Comparison.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/Dirac.h ./qcd/LinalgUtils.h ./qcd/QCD.h ./qcd/SpaceTimeGrid.h ./qcd/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Old/Grid_vComplexD.h ./simd/Old/Grid_vComplexF.h ./simd/Old/Grid_vInteger.h ./simd/Old/Grid_vRealD.h ./simd/Old/Grid_vRealF.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./Tensors.h ./Threads.h
 
-CCFILES=./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/action/gauge/WilsonGaugeAction.cc ./qcd/Dirac.cc ./qcd/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
+CCFILES=./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/Dirac.cc ./qcd/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc

From ad18df92d00254113016d3939f89ddce41aef33f Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Fri, 5 Jun 2015 10:29:42 +0100
Subject: [PATCH 281/429] Compile fix

---
 lib/qcd/utils/WilsonLoops.h |  4 ++++
 tests/Test_GaugeAction.cc   | 10 +++++-----
 2 files changed, 9 insertions(+), 5 deletions(-)

diff --git a/lib/qcd/utils/WilsonLoops.h b/lib/qcd/utils/WilsonLoops.h
index f2d6d4de..82230aa8 100644
--- a/lib/qcd/utils/WilsonLoops.h
+++ b/lib/qcd/utils/WilsonLoops.h
@@ -120,6 +120,10 @@ void siteRectangle(GaugeMat &plaq,const std::vector<GaugeMat> &U, const int mu,
   */
 
 };
+
+
+ typedef WilsonLoops<LatticeColourMatrix,LatticeGaugeField> ColourWilsonLoops;
+
 }}
 
 
diff --git a/tests/Test_GaugeAction.cc b/tests/Test_GaugeAction.cc
index 07e7a4eb..0dde6d8f 100644
--- a/tests/Test_GaugeAction.cc
+++ b/tests/Test_GaugeAction.cc
@@ -1,12 +1,12 @@
 #include <Grid.h>
 
+#include <qcd/utils/CovariantCshift.h>
+#include <qcd/utils/WilsonLoops.h>
+
 using namespace std;
 using namespace Grid;
 using namespace Grid::QCD;
 
-
-
-
 int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
@@ -96,14 +96,14 @@ int main (int argc, char ** argv)
   Complex p  = TensorRemove(Tp);
   std::cout << "calculated plaquettes " <<p*PlaqScale<<std::endl;
 
-  RealD avg_plaq =  WilsonLoops::avgPlaquette(Umu);
+  RealD avg_plaq = ColourWilsonLoops::avgPlaquette(Umu);
   std::cout << "NEW : calculated real plaquettes " <<avg_plaq<<std::endl;
 
   RealD stap_plaq=0.0;
   LatticeColourMatrix stap(&Fine);
   LatticeComplex stap_tr(&Fine);
   for(int mu=0;mu<Nd;mu++){
-    WilsonLoops::Staple(stap,Umu,mu);
+    ColourWilsonLoops::Staple(stap,Umu,mu);
     stap_tr = trace(stap*adj(U[mu]));
     TComplex Ts = sum(stap_tr);
     Complex s  = TensorRemove(Ts);

From 50e8b2160e4e2ddc81c38d5dd90b86a17be197ee Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 5 Jun 2015 18:16:25 +0100
Subject: [PATCH 282/429] Conjugate residual added

---
 lib/Algorithms.h                             |  1 +
 lib/Make.inc                                 |  4 +-
 lib/algorithms/LinearOperator.h              | 18 ++++-
 lib/algorithms/SparseMatrix.h                |  1 +
 lib/algorithms/iterative/ConjugateResidual.h | 85 ++++++++++++++++++++
 lib/qcd/QCD.h                                |  8 +-
 lib/qcd/action/Actions.h                     | 35 +-------
 lib/qcd/action/fermion/g5HermitianLinop.h    | 38 +++++++++
 lib/qcd/{ => spin}/Dirac.cc                  |  0
 lib/qcd/{ => spin}/Dirac.h                   |  0
 lib/qcd/{ => spin}/TwoSpinor.h               |  0
 lib/qcd/{ => utils}/LinalgUtils.h            | 19 +++++
 lib/qcd/{ => utils}/SpaceTimeGrid.cc         |  0
 lib/qcd/{ => utils}/SpaceTimeGrid.h          |  0
 tests/Make.inc                               | 12 +++
 15 files changed, 181 insertions(+), 40 deletions(-)
 create mode 100644 lib/algorithms/iterative/ConjugateResidual.h
 create mode 100644 lib/qcd/action/fermion/g5HermitianLinop.h
 rename lib/qcd/{ => spin}/Dirac.cc (100%)
 rename lib/qcd/{ => spin}/Dirac.h (100%)
 rename lib/qcd/{ => spin}/TwoSpinor.h (100%)
 rename lib/qcd/{ => utils}/LinalgUtils.h (88%)
 rename lib/qcd/{ => utils}/SpaceTimeGrid.cc (100%)
 rename lib/qcd/{ => utils}/SpaceTimeGrid.h (100%)

diff --git a/lib/Algorithms.h b/lib/Algorithms.h
index b5a2814a..c2491cf4 100644
--- a/lib/Algorithms.h
+++ b/lib/Algorithms.h
@@ -5,6 +5,7 @@
 #include <algorithms/LinearOperator.h>
 
 #include <algorithms/iterative/ConjugateGradient.h>
+#include <algorithms/iterative/ConjugateResidual.h>
 #include <algorithms/iterative/NormalEquations.h>
 #include <algorithms/iterative/SchurRedBlack.h>
 
diff --git a/lib/Make.inc b/lib/Make.inc
index 94a566fb..ba35e25f 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Comparison.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/Dirac.h ./qcd/LinalgUtils.h ./qcd/QCD.h ./qcd/SpaceTimeGrid.h ./qcd/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Old/Grid_vComplexD.h ./simd/Old/Grid_vComplexF.h ./simd/Old/Grid_vInteger.h ./simd/Old/Grid_vRealD.h ./simd/Old/Grid_vRealF.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./Tensors.h ./Threads.h
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Comparison.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Old/Grid_vComplexD.h ./simd/Old/Grid_vComplexF.h ./simd/Old/Grid_vInteger.h ./simd/Old/Grid_vRealD.h ./simd/Old/Grid_vRealF.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./Tensors.h ./Threads.h
 
-CCFILES=./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/Dirac.cc ./qcd/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
+CCFILES=./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/algorithms/LinearOperator.h b/lib/algorithms/LinearOperator.h
index 710fcde7..492b34db 100644
--- a/lib/algorithms/LinearOperator.h
+++ b/lib/algorithms/LinearOperator.h
@@ -18,7 +18,8 @@ namespace Grid {
     public:
       virtual void Op     (const Field &in, Field &out) = 0; // Abstract base
       virtual void AdjOp  (const Field &in, Field &out) = 0; // Abstract base
-      virtual void HermOpAndNorm(const Field &in, Field &out,double &n1,double &n2)=0;
+      virtual void HermOpAndNorm(const Field &in, Field &out,RealD &n1,RealD &n2)=0;
+      virtual void HermOp(const Field &in, Field &out)=0;
     };
 
 
@@ -48,9 +49,13 @@ namespace Grid {
       void AdjOp     (const Field &in, Field &out){
 	_Mat.Mdag(in,out);
       }
-      void HermOpAndNorm(const Field &in, Field &out,double &n1,double &n2){
+      void HermOpAndNorm(const Field &in, Field &out,RealD &n1,RealD &n2){
 	_Mat.MdagM(in,out,n1,n2);
       }
+      void HermOp(const Field &in, Field &out){
+	RealD n1,n2;
+	HermOpAndNorm(in,out,n1,n2);
+      }
     };
 
     ////////////////////////////////////////////////////////////////////
@@ -67,7 +72,7 @@ namespace Grid {
       void AdjOp     (const Field &in, Field &out){
 	_Mat.M(in,out);
       }
-      void HermOpAndNorm(const Field &in, Field &out,double &n1,double &n2){
+      void HermOpAndNorm(const Field &in, Field &out,RealD &n1,RealD &n2){
 	ComplexD dot;
 
 	_Mat.M(in,out);
@@ -78,6 +83,9 @@ namespace Grid {
 	dot = innerProduct(out,out);
 	n2=real(dot);
       }
+      void HermOp(const Field &in, Field &out){
+	_Mat.M(in,out);
+      }
     };
 
     //////////////////////////////////////////////////////////
@@ -98,6 +106,10 @@ namespace Grid {
       void HermOpAndNorm(const Field &in, Field &out,RealD &n1,RealD &n2){
 	MpcDagMpc(in,out,n1,n2);
       }
+      void HermOp(const Field &in, Field &out){
+	RealD n1,n2;
+	HermOpAndNorm(in,out,n1,n2);
+      }
       void Op     (const Field &in, Field &out){
 	Mpc(in,out);
       }
diff --git a/lib/algorithms/SparseMatrix.h b/lib/algorithms/SparseMatrix.h
index 9146648f..a51657dc 100644
--- a/lib/algorithms/SparseMatrix.h
+++ b/lib/algorithms/SparseMatrix.h
@@ -18,6 +18,7 @@ namespace Grid {
 	Field tmp (in._grid);
 	ni=M(in,tmp);
 	no=Mdag(tmp,out);
+	std::cout << "MdagM "<< ni<<" "<<no<<std::endl;
       }
     };
 
diff --git a/lib/algorithms/iterative/ConjugateResidual.h b/lib/algorithms/iterative/ConjugateResidual.h
new file mode 100644
index 00000000..ae1f4378
--- /dev/null
+++ b/lib/algorithms/iterative/ConjugateResidual.h
@@ -0,0 +1,85 @@
+#ifndef GRID_CONJUGATE_RESIDUAL_H
+#define GRID_CONJUGATE_RESIDUAL_H
+
+namespace Grid {
+
+    /////////////////////////////////////////////////////////////
+    // Base classes for iterative processes based on operators
+    // single input vec, single output vec.
+    /////////////////////////////////////////////////////////////
+
+  template<class Field> 
+    class ConjugateResidual : public OperatorFunction<Field> {
+  public:                                                
+    RealD   Tolerance;
+    Integer MaxIterations;
+    int verbose;
+
+    ConjugateResidual(RealD tol,Integer maxit) : Tolerance(tol), MaxIterations(maxit) { 
+      verbose=1;
+    };
+
+    void operator() (LinearOperatorBase<Field> &Linop,const Field &src, Field &psi){
+
+      RealD a, b, c, d;
+      RealD cp, ssq,rsq;
+      
+      RealD rAr, rAAr, rArp;
+      RealD pAp, pAAp;
+
+      GridBase *grid = src._grid;
+      psi=zero;
+      Field r(grid),  p(grid), Ap(grid), Ar(grid);
+      
+      r=src;
+      p=src;
+
+      Linop.HermOpAndNorm(p,Ap,pAp,pAAp);
+      Linop.HermOpAndNorm(r,Ar,rAr,rAAr);
+
+      std::cout << "pAp, pAAp"<< pAp<<" "<<pAAp<<std::endl;
+      std::cout << "rAr, rAAr"<< rAr<<" "<<rAAr<<std::endl;
+
+      cp =norm2(r);
+      ssq=norm2(src);
+      rsq=Tolerance*Tolerance*ssq;
+
+      std::cout<<"ConjugateResidual: iteration " <<0<<" residual "<<cp<< " target"<< rsq<<std::endl;
+
+      for(int k=1;k<MaxIterations;k++){
+
+	a = rAr/pAAp;
+
+	axpy(psi,a,p,psi);
+
+	cp = axpy_norm(r,-a,Ap,r);
+
+	rArp=rAr;
+
+	Linop.HermOpAndNorm(r,Ar,rAr,rAAr);
+
+	b   =rAr/rArp;
+ 
+	axpy(p,b,p,r);
+	pAAp=axpy_norm(Ap,b,Ap,Ar);
+
+	std::cout<<"ConjugateResidual: iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
+
+	if(cp<rsq) {
+	  Linop.HermOp(psi,Ap);
+	  axpy(r,-1.0,src,Ap);
+	  RealD true_resid = norm2(r);
+	  std::cout<<"ConjugateResidual: Converged on iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
+	  std::cout<<"ConjugateResidual: true   residual  is "<<true_resid<<std::endl;
+	  std::cout<<"ConjugateResidual: target residual was "<<Tolerance <<std::endl;
+	  return;
+	}
+
+      }
+
+      std::cout<<"ConjugateResidual did NOT converge"<<std::endl;
+      assert(0);
+    }
+  };
+}
+#endif
diff --git a/lib/qcd/QCD.h b/lib/qcd/QCD.h
index c3f70c43..f17c557b 100644
--- a/lib/qcd/QCD.h
+++ b/lib/qcd/QCD.h
@@ -307,10 +307,10 @@ namespace QCD {
 }   //namespace QCD
 } // Grid
 
-#include <qcd/SpaceTimeGrid.h>
-#include <qcd/Dirac.h>
-#include <qcd/TwoSpinor.h>
-#include <qcd/LinalgUtils.h>
+#include <qcd/utils/SpaceTimeGrid.h>
+#include <qcd/spin/Dirac.h>
+#include <qcd/spin/TwoSpinor.h>
+#include <qcd/utils/LinalgUtils.h>
 #include <qcd/utils/CovariantCshift.h>
 #include <qcd/utils/WilsonLoops.h>
 #include <qcd/action/Actions.h>
diff --git a/lib/qcd/action/Actions.h b/lib/qcd/action/Actions.h
index edef1b53..56b453c6 100644
--- a/lib/qcd/action/Actions.h
+++ b/lib/qcd/action/Actions.h
@@ -65,36 +65,9 @@
 #include <qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h>
 #include <qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h>
 
-
-    // Chroma interface defining FermionAction
-    /*
-     template<typename T, typename P, typename Q>  class FermAct4D : public FermionAction<T,P,Q>
-     virtual LinearOperator<T>* linOp(Handle< FermState<T,P,Q> > state) const = 0;
-     virtual LinearOperator<T>* lMdagM(Handle< FermState<T,P,Q> > state) const = 0;
-     virtual LinOpSystemSolver<T>* invLinOp(Handle< FermState<T,P,Q> > state,
-     virtual MdagMSystemSolver<T>* invMdagM(Handle< FermState<T,P,Q> > state,
-     virtual LinOpMultiSystemSolver<T>* mInvLinOp(Handle< FermState<T,P,Q> > state,
-     virtual MdagMMultiSystemSolver<T>* mInvMdagM(Handle< FermState<T,P,Q> > state,
-     virtual MdagMMultiSystemSolverAccumulate<T>* mInvMdagMAcc(Handle< FermState<T,P,Q> > state,
-     virtual SystemSolver<T>* qprop(Handle< FermState<T,P,Q> > state,
-     class DiffFermAct4D : public FermAct4D<T,P,Q>
-     virtual DiffLinearOperator<T,Q,P>* linOp(Handle< FermState<T,P,Q> > state) const = 0;
-     virtual DiffLinearOperator<T,Q,P>* lMdagM(Handle< FermState<T,P,Q> > state) const = 0;
-    */
-
-
-    // Chroma interface defining GaugeAction
-    /*
-      template<typename P, typename Q>   class GaugeAction
-  virtual const CreateGaugeState<P,Q>& getCreateState() const = 0;
-  virtual GaugeState<P,Q>* createState(const Q& q) const
-  virtual const GaugeBC<P,Q>& getGaugeBC() const
-  virtual const Set& getSet(void) const = 0;
-  virtual void deriv(P& result, const Handle< GaugeState<P,Q> >& state) const 
-  virtual Double S(const Handle< GaugeState<P,Q> >& state) const = 0;
-
-  class LinearGaugeAction : public GaugeAction< multi1d<LatticeColorMatrix>, multi1d<LatticeColorMatrix> >
-  typedef multi1d<LatticeColorMatrix>  P;
-    */
+///////////////////////////////////////////////////////////////////////////////
+// G5 herm -- this has to live in QCD since dirac matrix is not in the broader sector of code
+///////////////////////////////////////////////////////////////////////////////
+#include <qcd/action/fermion/g5HermitianLinop.h>
 
 #endif
diff --git a/lib/qcd/action/fermion/g5HermitianLinop.h b/lib/qcd/action/fermion/g5HermitianLinop.h
new file mode 100644
index 00000000..9a8cc2eb
--- /dev/null
+++ b/lib/qcd/action/fermion/g5HermitianLinop.h
@@ -0,0 +1,38 @@
+#ifndef G5_HERMITIAN_LINOP
+#define G5_HERMITIAN_LINOP
+namespace Grid {
+  namespace QCD {
+////////////////////////////////////////////////////////////////////
+// Wrap an already herm matrix
+////////////////////////////////////////////////////////////////////
+template<class Matrix,class Field>
+class Gamma5HermitianLinearOperator : public LinearOperatorBase<Field> {
+  Matrix &_Mat;
+public:
+  Gamma5HermitianLinearOperator(Matrix &Mat): _Mat(Mat){};
+  void Op     (const Field &in, Field &out){
+    _Mat.M(in,out);
+  }
+  void AdjOp     (const Field &in, Field &out){
+    _Mat.M(in,out);
+  }
+  void HermOpAndNorm(const Field &in, Field &out,RealD &n1,RealD &n2){
+
+    HermOp(in,out);
+    
+    ComplexD dot;
+    dot= innerProduct(in,out);
+    n1=real(dot);
+    
+    dot = innerProduct(out,out);
+    n2=real(dot);
+  }
+  void HermOp(const Field &in, Field &out){
+    Field tmp(in._grid);
+    _Mat.M(in,tmp);
+    G5R5(out,tmp);
+  }
+};
+
+}}
+#endif
diff --git a/lib/qcd/Dirac.cc b/lib/qcd/spin/Dirac.cc
similarity index 100%
rename from lib/qcd/Dirac.cc
rename to lib/qcd/spin/Dirac.cc
diff --git a/lib/qcd/Dirac.h b/lib/qcd/spin/Dirac.h
similarity index 100%
rename from lib/qcd/Dirac.h
rename to lib/qcd/spin/Dirac.h
diff --git a/lib/qcd/TwoSpinor.h b/lib/qcd/spin/TwoSpinor.h
similarity index 100%
rename from lib/qcd/TwoSpinor.h
rename to lib/qcd/spin/TwoSpinor.h
diff --git a/lib/qcd/LinalgUtils.h b/lib/qcd/utils/LinalgUtils.h
similarity index 88%
rename from lib/qcd/LinalgUtils.h
rename to lib/qcd/utils/LinalgUtils.h
index 2b83d115..ef85e728 100644
--- a/lib/qcd/LinalgUtils.h
+++ b/lib/qcd/utils/LinalgUtils.h
@@ -109,5 +109,24 @@ PARALLEL_FOR_LOOP
     vstream(z._odata[ss+s],tmp);
   }
 }
+
+template<class vobj> 
+void G5R5(Lattice<vobj> &z,const Lattice<vobj> &x)
+{
+  GridBase *grid=x._grid;
+  z.checkerboard = x.checkerboard;
+  conformable(x,z);
+  int Ls = grid->_rdimensions[0];
+PARALLEL_FOR_LOOP
+  for(int ss=0;ss<grid->oSites();ss+=Ls){ // adds Ls
+    vobj tmp;
+    for(int s=0;s<Ls;s++){
+      int sp = Ls-1-s;
+      multGamma5(tmp(),x._odata[ss+s]());
+      vstream(z._odata[ss+sp],tmp);
+    }
+  }
+}
+
 }}
 #endif 
diff --git a/lib/qcd/SpaceTimeGrid.cc b/lib/qcd/utils/SpaceTimeGrid.cc
similarity index 100%
rename from lib/qcd/SpaceTimeGrid.cc
rename to lib/qcd/utils/SpaceTimeGrid.cc
diff --git a/lib/qcd/SpaceTimeGrid.h b/lib/qcd/utils/SpaceTimeGrid.h
similarity index 100%
rename from lib/qcd/SpaceTimeGrid.h
rename to lib/qcd/utils/SpaceTimeGrid.h
diff --git a/tests/Make.inc b/tests/Make.inc
index 533cbf17..de9d0709 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -14,6 +14,10 @@ Test_cayley_even_odd_SOURCES=Test_cayley_even_odd.cc
 Test_cayley_even_odd_LDADD=-lGrid
 
 
+Test_cf_cr_unprec_SOURCES=Test_cf_cr_unprec.cc
+Test_cf_cr_unprec_LDADD=-lGrid
+
+
 Test_contfrac_cg_SOURCES=Test_contfrac_cg.cc
 Test_contfrac_cg_LDADD=-lGrid
 
@@ -42,6 +46,10 @@ Test_dwf_cg_unprec_SOURCES=Test_dwf_cg_unprec.cc
 Test_dwf_cg_unprec_LDADD=-lGrid
 
 
+Test_dwf_cr_unprec_SOURCES=Test_dwf_cr_unprec.cc
+Test_dwf_cr_unprec_LDADD=-lGrid
+
+
 Test_dwf_even_odd_SOURCES=Test_dwf_even_odd.cc
 Test_dwf_even_odd_LDADD=-lGrid
 
@@ -90,6 +98,10 @@ Test_wilson_cg_unprec_SOURCES=Test_wilson_cg_unprec.cc
 Test_wilson_cg_unprec_LDADD=-lGrid
 
 
+Test_wilson_cr_unprec_SOURCES=Test_wilson_cr_unprec.cc
+Test_wilson_cr_unprec_LDADD=-lGrid
+
+
 Test_wilson_even_odd_SOURCES=Test_wilson_even_odd.cc
 Test_wilson_even_odd_LDADD=-lGrid
 

From 7537a2751d778534cfb1f4a2f851e8a1c870b71d Mon Sep 17 00:00:00 2001
From: Matt Spraggs <matthew.spraggs@gmail.com>
Date: Sun, 7 Jun 2015 17:06:25 +0100
Subject: [PATCH 283/429] Removed std::string calls from NerscIO map indexing

---
 lib/parallelIO/NerscIO.h | 44 ++++++++++++++++++++--------------------
 1 file changed, 22 insertions(+), 22 deletions(-)

diff --git a/lib/parallelIO/NerscIO.h b/lib/parallelIO/NerscIO.h
index 8a1775f1..4db182e0 100644
--- a/lib/parallelIO/NerscIO.h
+++ b/lib/parallelIO/NerscIO.h
@@ -164,37 +164,37 @@ inline int readNerscHeader(std::string file,GridBase *grid,  NerscField &field)
   // chomp the values
   //////////////////////////////////////////////////
 
-  field.hdr_version    = header[std::string("HDR_VERSION")];
-  field.data_type      = header[std::string("DATATYPE")];
-  field.storage_format = header[std::string("STORAGE_FORMAT")];
+  field.hdr_version    = header["HDR_VERSION"];
+  field.data_type      = header["DATATYPE"];
+  field.storage_format = header["STORAGE_FORMAT"];
   
-  field.dimension[0] = std::stol(header[std::string("DIMENSION_1")]);
-  field.dimension[1] = std::stol(header[std::string("DIMENSION_2")]);
-  field.dimension[2] = std::stol(header[std::string("DIMENSION_3")]);
-  field.dimension[3] = std::stol(header[std::string("DIMENSION_4")]);
+  field.dimension[0] = std::stol(header["DIMENSION_1"]);
+  field.dimension[1] = std::stol(header["DIMENSION_2"]);
+  field.dimension[2] = std::stol(header["DIMENSION_3"]);
+  field.dimension[3] = std::stol(header["DIMENSION_4"]);
 
   assert(grid->_ndimension == 4);
   for(int d=0;d<4;d++){
     assert(grid->_fdimensions[d]==field.dimension[d]);
   }
 
-  field.link_trace = std::stod(header[std::string("LINK_TRACE")]);
-  field.plaquette  = std::stod(header[std::string("PLAQUETTE")]);
+  field.link_trace = std::stod(header["LINK_TRACE"]);
+  field.plaquette  = std::stod(header["PLAQUETTE"]);
 
-  field.boundary[0] = header[std::string("BOUNDARY_1")];
-  field.boundary[1] = header[std::string("BOUNDARY_2")];
-  field.boundary[2] = header[std::string("BOUNDARY_3")];
-  field.boundary[3] = header[std::string("BOUNDARY_4")];
+  field.boundary[0] = header["BOUNDARY_1"];
+  field.boundary[1] = header["BOUNDARY_2"];
+  field.boundary[2] = header["BOUNDARY_3"];
+  field.boundary[3] = header["BOUNDARY_4"];
 
-  field.checksum = std::stoul(header[std::string("CHECKSUM")],0,16);
-  field.ensemble_id      = header[std::string("ENSEMBLE_ID")];
-  field.ensemble_label   = header[std::string("ENSEMBLE_LABEL")];
-  field.sequence_number  = std::stol(header[std::string("SEQUENCE_NUMBER")]);
-  field.creator          = header[std::string("CREATOR")];
-  field.creator_hardware = header[std::string("CREATOR_HARDWARE")];
-  field.creation_date    = header[std::string("CREATION_DATE")];
-  field.archive_date     = header[std::string("ARCHIVE_DATE")];
-  field.floating_point   = header[std::string("FLOATING_POINT")];
+  field.checksum = std::stoul(header["CHECKSUM"],0,16);
+  field.ensemble_id      = header["ENSEMBLE_ID"];
+  field.ensemble_label   = header["ENSEMBLE_LABEL"];
+  field.sequence_number  = std::stol(header["SEQUENCE_NUMBER"]);
+  field.creator          = header["CREATOR"];
+  field.creator_hardware = header["CREATOR_HARDWARE"];
+  field.creation_date    = header["CREATION_DATE"];
+  field.archive_date     = header["ARCHIVE_DATE"];
+  field.floating_point   = header["FLOATING_POINT"];
 
   return field.data_start;
 }

From 54aec05989076a0c6df9f35cacbf344ebb932519 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Mon, 8 Jun 2015 11:52:44 +0100
Subject: [PATCH 284/429] multishift conjugate gradient added and a strong
 test: take a diagonal but non-identity matrix l1 0  0  0 .... 0  l2 0  0 ....
 0  0  l3 0 ... .  .   . .  .   . .  .   .

And apply the multishift CG to it. Sum the poles and residues.
Insist that this be the same as the exactly taken square root
where l1,l2,l3 >= 0.
---
 lib/Algorithms.h                              |   9 +-
 lib/Make.inc                                  |   4 +-
 lib/algorithms/LinearOperator.h               |   8 +
 lib/algorithms/approx/MultiShiftFunction.cc   |  29 ++
 lib/algorithms/approx/MultiShiftFunction.h    |  28 ++
 lib/algorithms/approx/Remez.h                 |   9 +
 .../iterative/ConjugateGradientMultiShift.h   | 250 ++++++++++++++++++
 tests/Make.inc                                |   6 +-
 tests/Test_multishift_sqrt.cc                 | 110 ++++++++
 tests/Test_remez.cc                           |  53 +---
 tests/Test_wilson_cg_unprec.cc                |   1 -
 11 files changed, 453 insertions(+), 54 deletions(-)
 create mode 100644 lib/algorithms/approx/MultiShiftFunction.cc
 create mode 100644 lib/algorithms/approx/MultiShiftFunction.h
 create mode 100644 lib/algorithms/iterative/ConjugateGradientMultiShift.h
 create mode 100644 tests/Test_multishift_sqrt.cc

diff --git a/lib/Algorithms.h b/lib/Algorithms.h
index b5a2814a..f588bf41 100644
--- a/lib/Algorithms.h
+++ b/lib/Algorithms.h
@@ -4,13 +4,16 @@
 #include <algorithms/SparseMatrix.h>
 #include <algorithms/LinearOperator.h>
 
+#include <algorithms/approx/Zolotarev.h>
+#include <algorithms/approx/Chebyshev.h>
+#include <algorithms/approx/Remez.h>
+#include <algorithms/approx/MultiShiftFunction.h>
+
 #include <algorithms/iterative/ConjugateGradient.h>
 #include <algorithms/iterative/NormalEquations.h>
 #include <algorithms/iterative/SchurRedBlack.h>
 
-#include <algorithms/approx/Zolotarev.h>
-#include <algorithms/approx/Chebyshev.h>
-#include <algorithms/approx/Remez.h>
+#include <algorithms/iterative/ConjugateGradientMultiShift.h>
 
 // Eigen/lanczos
 // EigCg
diff --git a/lib/Make.inc b/lib/Make.inc
index 94a566fb..cfe14de7 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Comparison.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/Dirac.h ./qcd/LinalgUtils.h ./qcd/QCD.h ./qcd/SpaceTimeGrid.h ./qcd/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Old/Grid_vComplexD.h ./simd/Old/Grid_vComplexF.h ./simd/Old/Grid_vInteger.h ./simd/Old/Grid_vRealD.h ./simd/Old/Grid_vRealF.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./Tensors.h ./Threads.h
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Comparison.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/Dirac.h ./qcd/LinalgUtils.h ./qcd/QCD.h ./qcd/SpaceTimeGrid.h ./qcd/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Old/Grid_vComplexD.h ./simd/Old/Grid_vComplexF.h ./simd/Old/Grid_vInteger.h ./simd/Old/Grid_vRealD.h ./simd/Old/Grid_vRealF.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./Tensors.h ./Threads.h
 
-CCFILES=./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/Dirac.cc ./qcd/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
+CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/Dirac.cc ./qcd/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/algorithms/LinearOperator.h b/lib/algorithms/LinearOperator.h
index 710fcde7..9f4459c8 100644
--- a/lib/algorithms/LinearOperator.h
+++ b/lib/algorithms/LinearOperator.h
@@ -167,6 +167,14 @@ namespace Grid {
       virtual void operator() (LinearOperatorBase<Field> &Linop, const Field &in, Field &out) = 0;
     };
 
+    /////////////////////////////////////////////////////////////
+    // Base classes for Multishift solvers for operators
+    /////////////////////////////////////////////////////////////
+    template<class Field> class OperatorMultiFunction {
+    public:
+      virtual void operator() (LinearOperatorBase<Field> &Linop, const Field &in, std::vector<Field> &out) = 0;
+    };
+
     // FIXME : To think about
 
     // Chroma functionality list defining LinearOperator
diff --git a/lib/algorithms/approx/MultiShiftFunction.cc b/lib/algorithms/approx/MultiShiftFunction.cc
new file mode 100644
index 00000000..8f62dffd
--- /dev/null
+++ b/lib/algorithms/approx/MultiShiftFunction.cc
@@ -0,0 +1,29 @@
+#include <Grid.h>
+
+namespace Grid {
+double MultiShiftFunction::approx(double x)
+{
+  double a = norm;
+  for(int n=0;n<poles.size();n++){
+    a = a + residues[n]/(x+poles[n]);
+  }
+  return a;
+}
+void MultiShiftFunction::gnuplot(std::ostream &out)
+{
+  out<<"f(x) = "<<norm<<"";
+  for(int n=0;n<poles.size();n++){
+    out<<"+("<<residues[n]<<"/(x+"<<poles[n]<<"))";
+  }
+  out<<";"<<std::endl;
+}
+void MultiShiftFunction::csv(std::ostream &out)
+{
+  for (double x=lo; x<hi; x*=1.05) {
+    double f = approx(x);
+    double r = sqrt(x);
+    out<< x<<","<<r<<","<<f<<","<<r-f<<std::endl;
+  }
+  return;
+}
+}
diff --git a/lib/algorithms/approx/MultiShiftFunction.h b/lib/algorithms/approx/MultiShiftFunction.h
new file mode 100644
index 00000000..4136736e
--- /dev/null
+++ b/lib/algorithms/approx/MultiShiftFunction.h
@@ -0,0 +1,28 @@
+#ifndef MULTI_SHIFT_FUNCTION
+#define MULTI_SHIFT_FUNCTION
+namespace Grid {
+class MultiShiftFunction {
+public:
+  int order;
+  std::vector<RealD> poles;
+  std::vector<RealD> residues;
+  std::vector<RealD> tolerances;
+  RealD norm;
+  RealD lo,hi;
+  MultiShiftFunction(int n,RealD _lo,RealD _hi): poles(n), residues(n), lo(_lo), hi(_hi) {;};
+  RealD approx(RealD x);
+  void csv(std::ostream &out);
+  void gnuplot(std::ostream &out);
+  MultiShiftFunction(AlgRemez & remez,double tol,bool inverse) :
+      order(remez.getDegree()),
+      tolerances(remez.getDegree(),tol),
+      poles(remez.getDegree()),
+      residues(remez.getDegree())
+  {
+    remez.getBounds(lo,hi);
+    if ( inverse ) remez.getIPFE (&residues[0],&poles[0],&norm);
+    else remez.getPFE (&residues[0],&poles[0],&norm);
+  }
+};
+}
+#endif
diff --git a/lib/algorithms/approx/Remez.h b/lib/algorithms/approx/Remez.h
index 906811be..336382d2 100644
--- a/lib/algorithms/approx/Remez.h
+++ b/lib/algorithms/approx/Remez.h
@@ -125,8 +125,17 @@ class AlgRemez
   // Destructor
   virtual ~AlgRemez();
 
+  int getDegree(void){ 
+    assert(n==d);
+    return n;
+  }
   // Reset the bounds of the approximation
   void setBounds(double lower, double upper);
+  // Reset the bounds of the approximation
+  void getBounds(double &lower, double &upper) { 
+    lower=(double)apstrt;
+    upper=(double)apend;
+  }
 
   // Generate the rational approximation x^(pnum/pden)
   double generateApprox(int num_degree, int den_degree, 
diff --git a/lib/algorithms/iterative/ConjugateGradientMultiShift.h b/lib/algorithms/iterative/ConjugateGradientMultiShift.h
new file mode 100644
index 00000000..580ab838
--- /dev/null
+++ b/lib/algorithms/iterative/ConjugateGradientMultiShift.h
@@ -0,0 +1,250 @@
+#ifndef GRID_CONJUGATE_MULTI_SHIFT_GRADIENT_H
+#define GRID_CONJUGATE_MULTI_SHIFT_GRADIENT_H
+
+namespace Grid {
+
+    /////////////////////////////////////////////////////////////
+    // Base classes for iterative processes based on operators
+    // single input vec, single output vec.
+    /////////////////////////////////////////////////////////////
+
+  template<class Field> 
+    class ConjugateGradientMultiShift : public OperatorMultiFunction<Field>,
+                                        public OperatorFunction<Field>
+    {
+public:                                                
+    RealD   Tolerance;
+    Integer MaxIterations;
+    int verbose;
+    MultiShiftFunction shifts;
+
+    ConjugateGradientMultiShift(Integer maxit,MultiShiftFunction &_shifts) : 
+	MaxIterations(maxit),
+	shifts(_shifts)
+    { 
+      verbose=1;
+    }
+
+void operator() (LinearOperatorBase<Field> &Linop, const Field &src, Field &psi)
+{
+
+  GridBase *grid = src._grid;
+  int nshift = shifts.order;
+  std::vector<Field> results(nshift,grid);
+
+  (*this)(Linop,src,results);
+  
+  psi = shifts.norm*src;
+  for(int i=0;i<nshift;i++){
+    psi = psi + shifts.residues[i]*results[i];
+  }
+
+  return;
+}
+
+void operator() (LinearOperatorBase<Field> &Linop, const Field &src, std::vector<Field> &psi)
+{
+  
+  GridBase *grid = src._grid;
+  
+  ////////////////////////////////////////////////////////////////////////
+  // Convenience references to the info stored in "MultiShiftFunction"
+  ////////////////////////////////////////////////////////////////////////
+  int nshift = shifts.order;
+
+  std::vector<RealD> &mass(shifts.poles); // Make references to array in "shifts"
+  std::vector<RealD> &mresidual(shifts.tolerances);
+  std::vector<RealD> alpha(nshift,1.0);
+  std::vector<Field>   ps(nshift,grid);// Search directions
+
+  assert(psi.size()==nshift);
+  assert(mass.size()==nshift);
+  assert(mresidual.size()==nshift);
+  
+  // dynamic sized arrays on stack; 2d is a pain with vector
+  RealD  bs[nshift];
+  RealD  rsq[nshift];
+  RealD  z[nshift][2];
+  int     converged[nshift];
+  
+  const int       primary =0;
+  
+  //Primary shift fields CG iteration
+  RealD a,b,c,d;
+  RealD cp,bp,qq; //prev
+  
+  // Matrix mult fields
+  Field r(grid);
+  Field p(grid);
+  Field tmp(grid);
+  Field mmp(grid);
+  
+  // Check lightest mass
+  for(int s=0;s<nshift;s++){
+    assert( mass[s]>= mass[primary] );
+    converged[s]=0;
+  }
+  
+  // Wire guess to zero
+  // Residuals "r" are src
+  // First search direction "p" is also src
+  cp = norm2(src);
+  for(int s=0;s<nshift;s++){
+    rsq[s] = cp * mresidual[s] * mresidual[s];
+    std::cout<<"ConjugateGradientMultiShift: shift "<<s
+	     <<" target resid "<<rsq[s]<<std::endl;
+    ps[s] = src;
+  }
+  // r and p for primary
+  r=src;
+  p=src;
+  
+  //MdagM+m[0]
+  Linop.HermOpAndNorm(p,mmp,d,qq);
+  axpy(mmp,mass[0],p,mmp);
+  RealD rn = norm2(p);
+  d += rn*mass[0];
+  
+  // have verified that inner product of 
+  // p and mmp is equal to d after this since
+  // the d computation is tricky
+  //  qq = real(innerProduct(p,mmp));
+  //  std::cout << "debug equal ?  qq "<<qq<<" d "<< d<<std::endl;
+  
+  b = -cp /d;
+  
+  // Set up the various shift variables
+  int       iz=0;
+  z[0][1-iz] = 1.0;
+  z[0][iz]   = 1.0;
+  bs[0]      = b;
+  for(int s=1;s<nshift;s++){
+    z[s][1-iz] = 1.0;
+    z[s][iz]   = 1.0/( 1.0 - b*(mass[s]-mass[0]));
+    bs[s]      = b*z[s][iz]; 
+  }
+  
+  // r += b[0] A.p[0]
+  // c= norm(r)
+  c=axpy_norm(r,b,mmp,r);
+  
+  for(int s=0;s<nshift;s++) {
+    axpby(psi[s],0.,-bs[s]*alpha[s],src,src);
+  }
+  
+  
+  // Iteration loop
+  int k;
+  
+  for (k=1;k<=MaxIterations;k++){
+    
+    a = c /cp;
+    axpy(p,a,p,r);
+    
+    // Note to self - direction ps is iterated seperately
+    // for each shift. Does not appear to have any scope
+    // for avoiding linear algebra in "single" case.
+    // 
+    // However SAME r is used. Could load "r" and update
+    // ALL ps[s]. 2/3 Bandwidth saving
+    // New Kernel: Load r, vector of coeffs, vector of pointers ps
+    for(int s=0;s<nshift;s++){
+      if ( ! converged[s] ) { 
+	if (s==0){
+	  axpy(ps[s],a,ps[s],r);
+	} else{
+	  RealD as =a *z[s][iz]*bs[s] /(z[s][1-iz]*b);
+	  axpby(ps[s],z[s][iz],as,r,ps[s]);
+	}
+      }
+    }
+    
+    cp=c;
+    
+    Linop.HermOpAndNorm(p,mmp,d,qq);
+    axpy(mmp,mass[0],p,mmp);
+    RealD rn = norm2(p);
+    d += rn*mass[0];
+    
+    bp=b;
+    b=-cp/d;
+    
+    c=axpy_norm(r,b,mmp,r);
+
+    // Toggle the recurrence history
+    bs[0] = b;
+    iz = 1-iz;
+    for(int s=1;s<nshift;s++){
+      if((!converged[s])){
+	RealD z0 = z[s][1-iz];
+	RealD z1 = z[s][iz];
+	z[s][iz] = z0*z1*bp
+	  / (b*a*(z1-z0) + z1*bp*(1- (mass[s]-mass[0])*b)); 
+	bs[s] = b*z[s][iz]/z0; // NB sign  rel to Mike
+      }
+    }
+    
+    for(int s=0;s<nshift;s++){
+      int ss = s;
+      // Scope for optimisation here in case of "single".
+      // Could load psi[0] and pull all ps[s] in.
+      //      if ( single ) ss=primary;
+      // Bandwith saving in single case is Ls * 3 -> 2+Ls, so ~ 3x saving
+      // Pipelined CG gain:
+      //
+      // New Kernel: Load r, vector of coeffs, vector of pointers ps
+      // New Kernel: Load psi[0], vector of coeffs, vector of pointers ps
+      // If can predict the coefficient bs then we can fuse these and avoid write reread cyce
+      //  on ps[s].
+      // Before:  3 x npole  + 3 x npole
+      // After :  2 x npole (ps[s])        => 3x speed up of multishift CG.
+      
+      if( (!converged[s]) ) { 
+	axpy(psi[ss],-bs[s]*alpha[s],ps[s],psi[ss]);
+      }
+    }
+    
+    // Convergence checks
+    int all_converged = 1;
+    for(int s=0;s<nshift;s++){
+      
+      if ( (!converged[s]) ){
+	
+	RealD css  = c * z[s][iz]* z[s][iz];
+	
+	if(css<rsq[s]){
+	  if ( ! converged[s] )
+	    std::cout<<"ConjugateGradientMultiShift k="<<k<<" Shift "<<s<<" has converged"<<std::endl;
+	      converged[s]=1;
+	} else {
+	  all_converged=0;
+	}
+
+      }
+    }
+    
+    if ( all_converged ){
+
+      std::cout<< "CGMultiShift: All shifts have converged iteration "<<k<<std::endl;
+      std::cout<< "CGMultiShift: Checking solutions"<<std::endl;
+      
+      // Check answers 
+      for(int s=0; s < nshift; s++) { 
+	Linop.HermOpAndNorm(psi[s],mmp,d,qq);
+	axpy(tmp,mass[s],psi[s],mmp);
+	axpy(r,-alpha[s],src,tmp);
+	RealD rn = norm2(r);
+	RealD cn = norm2(src);
+	std::cout<<"CGMultiShift: shift["<<s<<"] true residual "<<std::sqrt(rn/cn)<<std::endl;
+      }
+      return;
+    }
+  }
+  // ugly hack
+  std::cout<<"CG multi shift did not converge"<<std::endl;
+  assert(0);
+}
+
+  };
+}
+#endif
diff --git a/tests/Make.inc b/tests/Make.inc
index 533cbf17..681ca95f 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,5 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_even_odd Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_even_odd Test_gamma Test_main Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_even_odd
+bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_even_odd Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_even_odd Test_gamma Test_main Test_multishift_sqrt Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_even_odd
 
 
 Test_GaugeAction_SOURCES=Test_GaugeAction.cc
@@ -54,6 +54,10 @@ Test_main_SOURCES=Test_main.cc
 Test_main_LDADD=-lGrid
 
 
+Test_multishift_sqrt_SOURCES=Test_multishift_sqrt.cc
+Test_multishift_sqrt_LDADD=-lGrid
+
+
 Test_nersc_io_SOURCES=Test_nersc_io.cc
 Test_nersc_io_LDADD=-lGrid
 
diff --git a/tests/Test_multishift_sqrt.cc b/tests/Test_multishift_sqrt.cc
new file mode 100644
index 00000000..8e6b08d4
--- /dev/null
+++ b/tests/Test_multishift_sqrt.cc
@@ -0,0 +1,110 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class Field> class DumbOperator  : public LinearOperatorBase<Field> {
+public:
+  LatticeComplex scale;
+  LatticeComplex sqrtscale;
+  DumbOperator(GridBase *grid)
+    : scale(grid),
+      sqrtscale(grid)
+  {
+    GridParallelRNG  pRNG(grid);  
+    std::vector<int> seeds({5,6,7,8});
+    pRNG.SeedFixedIntegers(seeds);
+
+    random(pRNG,sqrtscale);
+    sqrtscale = sqrtscale * adj(sqrtscale);// force real pos def
+    scale = sqrtscale * sqrtscale; 
+  }
+  void Op     (const Field &in, Field &out){
+    out = scale * in;
+  }
+  void AdjOp  (const Field &in, Field &out){
+    out = scale * in;
+  }
+  void HermOpAndNorm(const Field &in, Field &out,double &n1,double &n2){
+    ComplexD dot;
+
+    out = scale * in;
+
+    dot= innerProduct(in,out);
+    n1=real(dot);
+
+    dot = innerProduct(out,out);
+    n2=real(dot);
+  }
+  void ApplySqrt(const Field &in, Field &out){
+    out = sqrtscale * in;
+  }
+};
+
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  GridCartesian *grid = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), 
+						       GridDefaultSimd(Nd,vComplexF::Nsimd()),
+						       GridDefaultMpi());
+
+  double     lo=0.001;
+  double     hi=1.0;
+  int precision=64;
+  int    degree=10;
+  AlgRemez remez(lo,hi,precision);
+
+  ////////////////////////////////////////
+  // sqrt and inverse sqrt
+  ////////////////////////////////////////
+  std::cout << "Generating degree "<<degree<<" for x^(1/2)"<<std::endl;
+  remez.generateApprox(degree,1,2);
+
+  MultiShiftFunction Sqrt(remez,1.0e-6,false);
+
+  GridParallelRNG  pRNG(grid);  
+  std::vector<int> seeds({1,2,3,4});
+  pRNG.SeedFixedIntegers(seeds);
+
+  LatticeFermion    src(grid); random(pRNG,src);
+  LatticeFermion    combined(grid);
+  LatticeFermion    reference(grid);
+  LatticeFermion    error(grid);
+  std::vector<LatticeFermion> result(degree,grid);
+  LatticeFermion    summed(grid);
+
+  ConjugateGradientMultiShift<LatticeFermion> MSCG(10000,Sqrt);
+
+  DumbOperator<LatticeFermion> Diagonal(grid);
+
+  MSCG(Diagonal,src,result);
+
+
+  //  double a = norm;
+  //  for(int n=0;n<poles.size();n++){
+  //    a = a + residues[n]/(x+poles[n]);
+  //  }
+  assert(Sqrt.order==degree);
+
+  combined = Sqrt.norm*src;
+  for(int i=0;i<degree;i++){
+    combined = combined + Sqrt.residues[i]*result[i];
+  }
+  
+  Diagonal.ApplySqrt(src,reference);
+
+  error = reference - combined;
+
+  std::cout << " Reference "<<norm2(reference)<<std::endl;
+  std::cout << " combined  "<<norm2(combined) <<std::endl;
+  std::cout << " error     "<<norm2(error)    <<std::endl;
+
+  MSCG(Diagonal,src,summed);
+  error = summed - combined;
+  std::cout << " summed-combined "<<norm2(error)    <<std::endl;
+
+  Grid_finalize();
+}
diff --git a/tests/Test_remez.cc b/tests/Test_remez.cc
index b5993917..3a51b985 100644
--- a/tests/Test_remez.cc
+++ b/tests/Test_remez.cc
@@ -4,43 +4,6 @@ using namespace std;
 using namespace Grid;
 using namespace Grid::QCD;
 
-class MultiShiftFunction {
-public:
-  std::vector<double> poles;
-  std::vector<double> residues;
-  double norm;
-  double lo,hi;
-  MultiShiftFunction(int n,double _lo,double _hi): poles(n), residues(n), lo(_lo), hi(_hi) {;};
-  double approx(double x);
-  void csv(std::ostream &out);
-  void gnuplot(std::ostream &out);
-};
-double MultiShiftFunction::approx(double x)
-{
-  double a = norm;
-  for(int n=0;n<poles.size();n++){
-    a = a + residues[n]/(x+poles[n]);
-  }
-  return a;
-}
-void MultiShiftFunction::gnuplot(std::ostream &out)
-{
-  out<<"f(x) = "<<norm<<"";
-  for(int n=0;n<poles.size();n++){
-    out<<"+("<<residues[n]<<"/(x+"<<poles[n]<<"))";
-  }
-  out<<";"<<std::endl;
-}
-void MultiShiftFunction::csv(std::ostream &out)
-{
-  for (double x=lo; x<hi; x*=1.05) {
-    double f = approx(x);
-    double r = sqrt(x);
-    out<< x<<","<<r<<","<<f<<","<<r-f<<std::endl;
-  }
-  return;
-}
-
 int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
@@ -56,22 +19,18 @@ int main (int argc, char ** argv)
   ////////////////////////////////////////
   // sqrt and inverse sqrt
   ////////////////////////////////////////
-  MultiShiftFunction Sqrt(degree,lo,hi);
-  MultiShiftFunction InvSqrt(degree,lo,hi);
-
-  MultiShiftFunction SqrtSqrt(degree,lo,hi);
-  MultiShiftFunction InvSqrtSqrt(degree,lo,hi);
-
 
   std::cout << "Generating degree "<<degree<<" for x^(1/2)"<<std::endl;
   remez.generateApprox(degree,1,2);
-  remez.getPFE (&   Sqrt.residues[0],&   Sqrt.poles[0],&   Sqrt.norm);
-  remez.getIPFE(&InvSqrt.residues[0],&InvSqrt.poles[0],&InvSqrt.norm);
+  MultiShiftFunction Sqrt(remez,1.0,false);
+  MultiShiftFunction InvSqrt(remez,1.0,true);
+
 
   std::cout << "Generating degree "<<degree<<" for x^(1/4)"<<std::endl;
   remez.generateApprox(degree,1,4);
-  remez.getPFE (&SqrtSqrt.residues[0],&SqrtSqrt.poles[0],&SqrtSqrt.norm);
-  remez.getIPFE(&InvSqrtSqrt.residues[0],&InvSqrtSqrt.poles[0],&InvSqrtSqrt.norm);
+  MultiShiftFunction SqrtSqrt(remez,1.0,false);
+  MultiShiftFunction InvSqrtSqrt(remez,1.0,true);
+
   
   ofstream gnuplot(std::string("Sqrt.gnu"),std::ios::out|std::ios::trunc);
   Sqrt.gnuplot(gnuplot);
diff --git a/tests/Test_wilson_cg_unprec.cc b/tests/Test_wilson_cg_unprec.cc
index c9f2856b..13996017 100644
--- a/tests/Test_wilson_cg_unprec.cc
+++ b/tests/Test_wilson_cg_unprec.cc
@@ -20,7 +20,6 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-
   std::vector<int> latt_size   = GridDefaultLatt();
   std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();

From ea583e2e53c8b7f9c2084999b2dd23d9daf1a08b Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 8 Jun 2015 12:02:26 +0100
Subject: [PATCH 285/429] Prep for mgrid

---
 lib/Algorithms.h | 1 +
 1 file changed, 1 insertion(+)

diff --git a/lib/Algorithms.h b/lib/Algorithms.h
index c2491cf4..0252606f 100644
--- a/lib/Algorithms.h
+++ b/lib/Algorithms.h
@@ -3,6 +3,7 @@
 
 #include <algorithms/SparseMatrix.h>
 #include <algorithms/LinearOperator.h>
+#include <algorithms/CoarsenedMatrix.h>
 
 #include <algorithms/iterative/ConjugateGradient.h>
 #include <algorithms/iterative/ConjugateResidual.h>

From 5bdf89e3f00f6df888488b7015b5f5710fa561b2 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 8 Jun 2015 12:02:53 +0100
Subject: [PATCH 286/429] Prep for multigrid

---
 lib/algorithms/LinearOperator.h | 28 ++++++++++++++++++++++++++++
 1 file changed, 28 insertions(+)

diff --git a/lib/algorithms/LinearOperator.h b/lib/algorithms/LinearOperator.h
index 492b34db..ee0b2ec1 100644
--- a/lib/algorithms/LinearOperator.h
+++ b/lib/algorithms/LinearOperator.h
@@ -16,6 +16,11 @@ namespace Grid {
   /////////////////////////////////////////////////////////////////////////////////////////////
     template<class Field> class LinearOperatorBase {
     public:
+
+      // Support for coarsening to a multigrid
+      virtual void OpDiag (const Field &in, Field &out) = 0; // Abstract base
+      virtual void OpDir  (const Field &in, Field &out,int dir,int disp) = 0; // Abstract base
+
       virtual void Op     (const Field &in, Field &out) = 0; // Abstract base
       virtual void AdjOp  (const Field &in, Field &out) = 0; // Abstract base
       virtual void HermOpAndNorm(const Field &in, Field &out,RealD &n1,RealD &n2)=0;
@@ -43,6 +48,14 @@ namespace Grid {
       Matrix &_Mat;
     public:
     MdagMLinearOperator(Matrix &Mat): _Mat(Mat){};
+
+      // Support for coarsening to a multigrid
+      void OpDiag (const Field &in, Field &out) {
+	_Mat.Mdiag(in,out);
+      }
+      void OpDir  (const Field &in, Field &out,int dir,int disp) {
+	_Mat.Mdir(in,out,dir,disp);
+      }
       void Op     (const Field &in, Field &out){
 	_Mat.M(in,out);
       }
@@ -66,6 +79,13 @@ namespace Grid {
       Matrix &_Mat;
     public:
     HermitianLinearOperator(Matrix &Mat): _Mat(Mat){};
+      // Support for coarsening to a multigrid
+      void OpDiag (const Field &in, Field &out) {
+	_Mat.Mdiag(in,out);
+      }
+      void OpDir  (const Field &in, Field &out,int dir,int disp) {
+	_Mat.Mdir(in,out,dir,disp);
+      }
       void Op     (const Field &in, Field &out){
 	_Mat.M(in,out);
       }
@@ -116,6 +136,14 @@ namespace Grid {
       void AdjOp     (const Field &in, Field &out){ 
 	MpcDag(in,out);
       }
+      // Support for coarsening to a multigrid
+      void OpDiag (const Field &in, Field &out) {
+	assert(0); // must coarsen the unpreconditioned system
+      }
+      void OpDir  (const Field &in, Field &out,int dir,int disp) {
+	assert(0);
+      }
+
     };
     template<class Matrix,class Field>
       class SchurDiagMooeeOperator :  public SchurOperatorBase<Field> {

From 690397e7c67a6c3df62b5db60bed31254fec6a11 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 8 Jun 2015 12:03:36 +0100
Subject: [PATCH 287/429] sqrt

---
 tests/Test_remez.cc | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/tests/Test_remez.cc b/tests/Test_remez.cc
index b5993917..0d65ac7b 100644
--- a/tests/Test_remez.cc
+++ b/tests/Test_remez.cc
@@ -35,7 +35,7 @@ void MultiShiftFunction::csv(std::ostream &out)
 {
   for (double x=lo; x<hi; x*=1.05) {
     double f = approx(x);
-    double r = sqrt(x);
+    double r = std::sqrt(x);
     out<< x<<","<<r<<","<<f<<","<<r-f<<std::endl;
   }
   return;
@@ -80,8 +80,8 @@ int main (int argc, char ** argv)
   InvSqrt.gnuplot(gnuplot);
 
   double x=0.6789;
-  double sx=sqrt(x);
-  double ssx=sqrt(sx);
+  double sx=std::sqrt(x);
+  double ssx=std::sqrt(sx);
   double isx=1.0/sx;
   double issx=1.0/ssx;
 

From 9e7035f5dc622359a9f954f85a50febad7ac050a Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 8 Jun 2015 12:04:59 +0100
Subject: [PATCH 288/429] Conjugate residual algorithm; some more unary
 functions

---
 lib/Make.inc                                  |   2 +-
 lib/Simd.h                                    |   3 +
 lib/Tensors.h                                 |   1 +
 lib/algorithms/CoarsenedMatrix.h              | 146 ++++++++++++++
 lib/algorithms/SparseMatrix.h                 |   3 +-
 lib/algorithms/iterative/SchurRedBlack.h      |   2 +-
 lib/lattice/Lattice_base.h                    |   1 +
 lib/lattice/Lattice_local.h                   |   3 +-
 lib/lattice/Lattice_reality.h                 |   8 +-
 lib/lattice/Lattice_transfer.h                | 186 +++++++++++++-----
 lib/lattice/Lattice_unary.h                   |  32 +++
 lib/qcd/action/fermion/CayleyFermion5D.cc     |  21 ++
 lib/qcd/action/fermion/CayleyFermion5D.h      |   4 +
 .../fermion/ContinuedFractionFermion5D.cc     |  12 ++
 .../fermion/ContinuedFractionFermion5D.h      |   3 +
 lib/qcd/action/fermion/FermionOperator.h      |   3 +
 .../fermion/PartialFractionFermion5D.cc       |  16 ++
 .../action/fermion/PartialFractionFermion5D.h |   3 +
 lib/qcd/action/fermion/WilsonFermion.cc       |  20 ++
 lib/qcd/action/fermion/WilsonFermion.h        |   6 +-
 lib/qcd/action/fermion/WilsonFermion5D.cc     |  22 +++
 lib/qcd/action/fermion/WilsonFermion5D.h      |   4 +
 lib/qcd/action/fermion/WilsonKernels.cc       | 148 ++++++++++++--
 lib/qcd/action/fermion/WilsonKernels.h        |   3 +
 lib/qcd/action/fermion/g5HermitianLinop.h     |  15 +-
 lib/simd/Grid_avx.h                           |   5 +-
 lib/simd/Grid_vector_types.h                  |  30 ++-
 lib/simd/Grid_vector_unops.h                  |  60 ++++++
 lib/tensors/Tensor_class.h                    |   7 +
 lib/tensors/Tensor_unary.h                    |  37 ++++
 tests/Make.inc                                |  10 +-
 tests/Test_GaugeAction.cc                     |   2 +-
 tests/Test_cayley_coarsen_support.cc          | 108 ++++++++++
 tests/Test_cf_coarsen_support.cc              |  87 ++++++++
 tests/Test_cf_cr_unprec.cc                    |  58 ++++++
 tests/Test_dwf_cr_unprec.cc                   |  63 ++++++
 tests/Test_nersc_io.cc                        |   2 +-
 tests/Test_wilson_cr_unprec.cc                |  59 ++++++
 38 files changed, 1116 insertions(+), 79 deletions(-)
 create mode 100644 lib/algorithms/CoarsenedMatrix.h
 create mode 100644 lib/lattice/Lattice_unary.h
 create mode 100644 lib/simd/Grid_vector_unops.h
 create mode 100644 lib/tensors/Tensor_unary.h
 create mode 100644 tests/Test_cayley_coarsen_support.cc
 create mode 100644 tests/Test_cf_coarsen_support.cc
 create mode 100644 tests/Test_cf_cr_unprec.cc
 create mode 100644 tests/Test_dwf_cr_unprec.cc
 create mode 100644 tests/Test_wilson_cr_unprec.cc

diff --git a/lib/Make.inc b/lib/Make.inc
index ba35e25f..069e818d 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Comparison.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Old/Grid_vComplexD.h ./simd/Old/Grid_vComplexF.h ./simd/Old/Grid_vInteger.h ./simd/Old/Grid_vRealD.h ./simd/Old/Grid_vRealF.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./Tensors.h ./Threads.h
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Comparison.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Old/Grid_vComplexD.h ./simd/Old/Grid_vComplexF.h ./simd/Old/Grid_vInteger.h ./simd/Old/Grid_vRealD.h ./simd/Old/Grid_vRealF.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./Tensors.h ./Threads.h
 
 CCFILES=./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/Simd.h b/lib/Simd.h
index 26edb4c9..f93c3070 100644
--- a/lib/Simd.h
+++ b/lib/Simd.h
@@ -35,6 +35,8 @@ namespace Grid {
   inline RealD conjugate(const RealD  & r){ return r; }
   inline RealD real(const RealD  & r){ return r; }
 
+  inline RealD sqrt(const RealD  & r){ return std::sqrt(r); }
+
   inline ComplexD conjugate(const ComplexD& r){ return(conj(r)); }
   inline ComplexD adj(const ComplexD& r){ return(conjugate(r)); }
   inline ComplexF conjugate(const ComplexF& r ){ return(conj(r)); }
@@ -112,6 +114,7 @@ namespace Grid {
 };
 
 #include <simd/Grid_vector_types.h>
+#include <simd/Grid_vector_unops.h>
 
 namespace Grid {
   // Default precision
diff --git a/lib/Tensors.h b/lib/Tensors.h
index fa0ca7a9..e585b116 100644
--- a/lib/Tensors.h
+++ b/lib/Tensors.h
@@ -12,6 +12,7 @@
 #include <tensors/Tensor_peek.h>
 #include <tensors/Tensor_poke.h>
 #include <tensors/Tensor_reality.h>
+#include <tensors/Tensor_unary.h>
 #include <tensors/Tensor_extract_merge.h>
     
 #endif
diff --git a/lib/algorithms/CoarsenedMatrix.h b/lib/algorithms/CoarsenedMatrix.h
new file mode 100644
index 00000000..54ea279e
--- /dev/null
+++ b/lib/algorithms/CoarsenedMatrix.h
@@ -0,0 +1,146 @@
+#ifndef  GRID_ALGORITHM_COARSENED_MATRIX_H
+#define  GRID_ALGORITHM_COARSENED_MATRIX_H
+
+#include <Grid.h>
+
+namespace Grid {
+
+  class Geometry {
+  public:
+    int npoint;
+    int dimension;
+    std::vector<int> directions   ;
+    std::vector<int> displacements;
+
+    Geometry(int _d) : dimension(_d), npoint(2*_d+1), directions(npoint), displacements(npoint) {
+      for(int d=0;d<dimension;d++){
+	directions[2*d  ] = d;
+	directions[2*d+1] = d;
+	displacements[2*d  ] = +1;
+	displacements[2*d+1] = -1;
+      }
+      directions   [2*dimension]=0;
+      displacements[2*dimension]=0;
+    }
+    
+    std::vector<int> GetDelta(int point) {
+      std::vector<int> delta(dimension,0);
+      delta[directions[point]] = displacements[point];
+      return delta;
+    };
+
+  };
+  
+  // Fine Object == (per site) type of fine field
+  // nbasis      == number of deflation vectors
+  template<class Fobj,class CComplex,int nbasis>
+  class CoarsenedMatrix : public SparseMatrixBase<Lattice<iVector<vComplex,nbasis > > >  {
+  public:
+    
+    typedef iVector<vComplex,nbasis > siteVector;
+    typedef Lattice<iVector<vComplex,nbasis > > CoarseVector;
+    typedef Lattice<iMatrix<vComplex,nbasis > > CoarseMatrix;
+
+    typedef Lattice< CComplex >   CoarseScalar; // used for inner products on fine field
+    typedef Lattice<Fobj >        FineField;
+
+    ////////////////////
+    // Data members
+    ////////////////////
+    Geometry         geom;
+    GridBase *       _grid; 
+    CartesianStencil Stencil; 
+    std::vector<CoarseMatrix> A;
+    std::vector<siteVector,alignedAllocator<siteVector> >   comm_buf;
+      
+    ///////////////////////
+    // Interface
+    ///////////////////////
+    GridBase * Grid(void)         { return _grid; };   // this is all the linalg routines need to know
+
+    RealD M    (const CoarseVector &in, CoarseVector &out){
+      
+      SimpleCompressor<siteVector> compressor;
+      Stencil.HaloExchange(in,comm_buf,compressor);
+
+      //PARALLEL_FOR_LOOP
+      for(int ss=0;ss<Grid()->oSites();ss++){
+        siteVector res = zero;
+	siteVector tmp;
+	siteVector nbr;
+
+	int offset,local,perm;
+	for(int point=0;point<geom.npoint;point++){
+	  offset = Stencil._offsets [point][ss];
+	  local  = Stencil._is_local[point][ss];
+	  perm   = Stencil._permute[point][ss];
+	  
+	  if(local&&perm) { 
+	    permute(nbr,in._odata[offset],perm);
+	  } else if(local) { 
+	    nbr = in._odata[offset];
+	  } else {
+	    nbr = comm_buf[offset];
+	  }
+	  res = res + A[point]._odata[ss]*nbr;
+	}
+	vstream(out._odata[ss],res);
+      }
+      return norm2(out);
+    };
+
+    RealD Mdag (const CoarseVector &in, CoarseVector &out){ 
+      return M(in,out);
+    };
+    // Defer support for further coarsening for now
+    void Mdiag    (const CoarseVector &in,  CoarseVector &out){};
+    void Mdir     (const CoarseVector &in,  CoarseVector &out,int dir, int disp){};
+
+    CoarsenedMatrix(GridCartesian &CoarseGrid) 	: 
+
+      _grid(&CoarseGrid),
+      geom(CoarseGrid._ndimension),
+      Stencil(&CoarseGrid,geom.npoint,Even,geom.directions,geom.displacements),
+      A(geom.npoint,&CoarseGrid)
+    {
+      comm_buf.resize(Stencil._unified_buffer_size);
+    };
+
+    void CoarsenOperator(GridBase *FineGrid,LinearOperatorBase<Lattice<Fobj> > &linop,std::vector<Lattice<Fobj> > & subspace){
+
+      FineField  phi(FineGrid);
+      FineField Mphi(FineGrid);
+      CoarseVector Proj(Grid()); 
+      CoarseScalar InnerProd(Grid()); 
+
+      // Orthogonalise the subblocks over the basis
+      blockOrthogonalise(InnerProd,subspace);
+
+      // Compute the matrix elements of linop between this orthonormal
+      // set of vectors.
+      for(int i=0;i<nbasis;i++){
+	phi=subspace[i];
+	for(int p=0;p<geom.npoint;p++){ 
+	  int dir = geom.directions[p];
+	  int disp= geom.displacements[p];
+
+	  if ( disp==0 )linop.OpDiag(phi,Mphi);
+	  else linop.OpDir(phi,Mphi,dir,disp); 
+
+	  blockProject(Proj,Mphi,subspace);
+
+	  for(int ss=0;ss<Grid()->oSites();ss++){
+	    for(int j=0;j<nbasis;j++){
+	      A[p]._odata[ss](j,i) = Proj._odata[ss](j);
+	    }
+	  }
+
+	}	
+      }
+      std::cout<<"Computed Coarse Operator"<<std::endl;
+    }
+    
+  };
+
+}
+#endif
diff --git a/lib/algorithms/SparseMatrix.h b/lib/algorithms/SparseMatrix.h
index a51657dc..eb807555 100644
--- a/lib/algorithms/SparseMatrix.h
+++ b/lib/algorithms/SparseMatrix.h
@@ -18,8 +18,9 @@ namespace Grid {
 	Field tmp (in._grid);
 	ni=M(in,tmp);
 	no=Mdag(tmp,out);
-	std::cout << "MdagM "<< ni<<" "<<no<<std::endl;
       }
+      virtual  void Mdiag    (const Field &in, Field &out)=0;
+      virtual  void Mdir     (const Field &in, Field &out,int dir, int disp)=0;
     };
 
   /////////////////////////////////////////////////////////////////////////////////////////////
diff --git a/lib/algorithms/iterative/SchurRedBlack.h b/lib/algorithms/iterative/SchurRedBlack.h
index 01f2e3d5..3ebc5f1d 100644
--- a/lib/algorithms/iterative/SchurRedBlack.h
+++ b/lib/algorithms/iterative/SchurRedBlack.h
@@ -108,7 +108,7 @@ namespace Grid {
       RealD ns = norm2(in);
       RealD nr = norm2(resid);
 
-      std::cout << "SchurRedBlackDiagMooee solver true unprec resid "<< sqrt(nr/ns) <<" nr "<< nr <<" ns "<<ns << std::endl;
+      std::cout << "SchurRedBlackDiagMooee solver true unprec resid "<< std::sqrt(nr/ns) <<" nr "<< nr <<" ns "<<ns << std::endl;
     }     
   };
 
diff --git a/lib/lattice/Lattice_base.h b/lib/lattice/Lattice_base.h
index 6b5fe261..ce2990a5 100644
--- a/lib/lattice/Lattice_base.h
+++ b/lib/lattice/Lattice_base.h
@@ -299,6 +299,7 @@ PARALLEL_FOR_LOOP
 #include <lattice/Lattice_reality.h>
 #include <lattice/Lattice_coordinate.h>
 #include <lattice/Lattice_rng.h>
+#include <lattice/Lattice_unary.h>
 #include <lattice/Lattice_transfer.h>
 
 
diff --git a/lib/lattice/Lattice_local.h b/lib/lattice/Lattice_local.h
index 7b8bdf9b..ad7a1ff5 100644
--- a/lib/lattice/Lattice_local.h
+++ b/lib/lattice/Lattice_local.h
@@ -25,8 +25,7 @@ PARALLEL_FOR_LOOP
     
     // localInnerProduct
     template<class vobj>
-    inline auto localInnerProduct (const Lattice<vobj> &lhs,const Lattice<vobj> &rhs)
-      -> Lattice<typename vobj::tensor_reduced>
+    inline auto localInnerProduct (const Lattice<vobj> &lhs,const Lattice<vobj> &rhs) -> Lattice<typename vobj::tensor_reduced>
     {
       Lattice<typename vobj::tensor_reduced> ret(rhs._grid);
 PARALLEL_FOR_LOOP
diff --git a/lib/lattice/Lattice_reality.h b/lib/lattice/Lattice_reality.h
index 4ad6fb7b..8c8bc17a 100644
--- a/lib/lattice/Lattice_reality.h
+++ b/lib/lattice/Lattice_reality.h
@@ -27,9 +27,9 @@ PARALLEL_FOR_LOOP
         return ret;
     };
 
-    template<class vobj> inline auto real(const Lattice<vobj> &z) -> Lattice<decltype(real(z._odata[0]))>
+    template<class vobj> inline auto real(const Lattice<vobj> &z) -> Lattice<vobj>
     {
-      Lattice<decltype(real(z._odata[0]))> ret(z._grid);
+      Lattice<vobj> ret(z._grid);
 PARALLEL_FOR_LOOP
         for(int ss=0;ss<z._grid->oSites();ss++){
             ret._odata[ss] = real(z._odata[ss]);
@@ -37,9 +37,9 @@ PARALLEL_FOR_LOOP
       return ret;
     }
 
-    template<class vobj> inline auto imag(const Lattice<vobj> &z) -> Lattice<decltype(imag(z._odata[0]))>
+    template<class vobj> inline auto imag(const Lattice<vobj> &z) -> Lattice<vobj>
     {
-      Lattice<decltype(imag(z._odata[0]))> ret(z._grid);
+      Lattice<vobj> ret(z._grid);
 PARALLEL_FOR_LOOP
         for(int ss=0;ss<z._grid->oSites();ss++){
             ret._odata[ss] = imag(z._odata[ss]);
diff --git a/lib/lattice/Lattice_transfer.h b/lib/lattice/Lattice_transfer.h
index f0046b0e..871ed893 100644
--- a/lib/lattice/Lattice_transfer.h
+++ b/lib/lattice/Lattice_transfer.h
@@ -56,10 +56,10 @@ PARALLEL_FOR_LOOP
   }
   
 
-template<class vobj,int nbasis>
-inline void projectBlockBasis(Lattice<iVector<vComplex,nbasis > > &coarseData,
-			      const             Lattice<vobj>   &fineData,
-			      const std::vector<Lattice<vobj> > &Basis)
+template<class vobj,class CComplex,int nbasis>
+inline void blockProject(Lattice<iVector<CComplex,nbasis > > &coarseData,
+			 const             Lattice<vobj>   &fineData,
+			 const std::vector<Lattice<vobj> > &Basis)
 {
   GridBase * fine  = fineData._grid;
   GridBase * coarse= coarseData._grid;
@@ -69,13 +69,14 @@ inline void projectBlockBasis(Lattice<iVector<vComplex,nbasis > > &coarseData,
   assert( nbasis == Basis.size() );
   subdivides(coarse,fine); 
   for(int i=0;i<nbasis;i++){
-    conformable(Basis,fineData);
+    conformable(Basis[i],fineData);
   }
 
   std::vector<int>  block_r      (_ndimension);
   
   for(int d=0 ; d<_ndimension;d++){
     block_r[d] = fine->_rdimensions[d] / coarse->_rdimensions[d];
+    assert(block_r[d]*coarse->_rdimensions[d] == fine->_rdimensions[d]);
   }
 
   coarseData=zero;
@@ -92,20 +93,149 @@ inline void projectBlockBasis(Lattice<iVector<vComplex,nbasis > > &coarseData,
 
     for(int i=0;i<nbasis;i++) {
       
-      coarseData._odata[sc][i]=coarseData._odata[sc][i]
-	+ innerProduct(Basis[i]._odata[sf],fineData._odata[sf]);
+      coarseData._odata[sc](i)=coarseData._odata[sc](i)
+	+ TensorRemove( innerProduct(Basis[i]._odata[sf],fineData._odata[sf]));
 
     }
   }
   return;
+}
+
+template<class vobj,class CComplex>
+inline void blockZAXPY(Lattice<vobj> &fineZ,
+		       const Lattice<CComplex> &coarseA,
+		       const Lattice<vobj> &fineX,
+		       const Lattice<vobj> &fineY)
+{
+  GridBase * fine  = fineZ._grid;
+  GridBase * coarse= coarseA._grid;
+
+  fineZ.checkerboard=fineX.checkerboard;
+  subdivides(coarse,fine); // require they map
+  conformable(fineX,fineY);
+  conformable(fineX,fineZ);
+
+  int _ndimension = coarse->_ndimension;
   
+  std::vector<int>  block_r      (_ndimension);
+
+  // FIXME merge with subdivide checking routine as this is redundant
+  for(int d=0 ; d<_ndimension;d++){
+    block_r[d] = fine->_rdimensions[d] / coarse->_rdimensions[d];
+    assert(block_r[d]*coarse->_rdimensions[d]==fine->_rdimensions[d]);
+  }
+
+PARALLEL_FOR_LOOP
+  for(int sf=0;sf<fine->oSites();sf++){
+    
+    int sc;
+    std::vector<int> coor_c(_ndimension);
+    std::vector<int> coor_f(_ndimension);
+
+    GridBase::CoorFromIndex(coor_f,sf,fine->_rdimensions);
+    for(int d=0;d<_ndimension;d++) coor_c[d]=coor_f[d]/block_r[d];
+    GridBase::IndexFromCoor(coor_c,sc,coarse->_rdimensions);
+
+    // z = A x + y
+    fineZ._odata[sf]=coarseA._odata[sc]*fineX._odata[sf]+fineY._odata[sf];
+
+  }
+
+  return;
+}
+template<class vobj,class CComplex>
+  inline void blockInnerProduct(Lattice<CComplex> &CoarseInner,
+				const Lattice<vobj> &fineX,
+				const Lattice<vobj> &fineY)
+{
+  typedef decltype(innerProduct(fineX._odata[0],fineY._odata[0])) dotp;
+
+  GridBase *coarse(CoarseInner._grid);
+  GridBase *fine  (fineX._grid);
+
+  Lattice<dotp> fine_inner(fine);
+  Lattice<dotp> coarse_inner(coarse);
+
+  fine_inner = localInnerProduct(fineX,fineY);
+  blockSum(coarse_inner,fine_inner);
+  for(int ss=0;ss<coarse->oSites();ss++){
+    CoarseInner._odata[ss] = coarse_inner._odata[ss];
+  }
+}
+template<class vobj,class CComplex>
+inline void blockNormalise(Lattice<CComplex> &ip,Lattice<vobj> &fineX)
+{
+  GridBase *coarse = ip._grid;
+  blockInnerProduct(ip,fineX,fineX);
+  ip = rsqrt(ip);
+  blockZAXPY(fineX,ip,fineX,fineX);
+}
+// useful in multigrid project;
+// Generic name : Coarsen?
+template<class vobj>
+inline void blockSum(Lattice<vobj> &coarseData,const Lattice<vobj> &fineData)
+{
+  GridBase * fine  = fineData._grid;
+  GridBase * coarse= coarseData._grid;
+
+  subdivides(coarse,fine); // require they map
+
+  int _ndimension = coarse->_ndimension;
+  
+  std::vector<int>  block_r      (_ndimension);
+  
+  for(int d=0 ; d<_ndimension;d++){
+    block_r[d] = fine->_rdimensions[d] / coarse->_rdimensions[d];
+  }
+
+  coarseData=zero;
+  for(int sf=0;sf<fine->oSites();sf++){
+    
+    int sc;
+    std::vector<int> coor_c(_ndimension);
+    std::vector<int> coor_f(_ndimension);
+
+    GridBase::CoorFromIndex(coor_f,sf,fine->_rdimensions);
+    for(int d=0;d<_ndimension;d++) coor_c[d]=coor_f[d]/block_r[d];
+    GridBase::IndexFromCoor(coor_c,sc,coarse->_rdimensions);
+
+    coarseData._odata[sc]=coarseData._odata[sc]+fineData._odata[sf];
+
+  }
+  return;
 }
 
 
-template<class vobj,int nbasis>
-inline void promoteBlockBasis(const Lattice<iVector<vComplex,nbasis > > &coarseData,
-			                        Lattice<vobj>   &fineData,
-			      const std::vector<Lattice<vobj> > &Basis)
+template<class vobj,class CComplex>
+inline void blockOrthogonalise(Lattice<CComplex> &ip,std::vector<Lattice<vobj> > &Basis)
+{
+  GridBase *coarse = ip._grid;
+  GridBase *fine   = Basis[0]._grid;
+
+  int       nbasis = Basis.size() ;
+  int  _ndimension = coarse->_ndimension;
+
+  // checks
+  subdivides(coarse,fine); 
+  for(int i=0;i<nbasis;i++){
+    conformable(Basis[i]._grid,fine);
+  }
+
+  for(int v=0;v<nbasis;v++) {
+    for(int u=0;u<v;u++) {
+      //Inner product & remove component 
+      blockInnerProduct(ip,Basis[u],Basis[v]);
+      ip = -ip;
+      blockZAXPY<vobj,CComplex> (Basis[v],ip,Basis[u],Basis[v]);
+    }
+    blockNormalise(ip,Basis[v]);
+  }
+}
+
+template<class vobj,class CComplex,int nbasis>
+inline void blockPromote(const Lattice<iVector<CComplex,nbasis > > &coarseData,
+			 Lattice<vobj>   &fineData,
+			 const std::vector<Lattice<vobj> > &Basis)
 {
   GridBase * fine  = fineData._grid;
   GridBase * coarse= coarseData._grid;
@@ -146,40 +276,6 @@ inline void promoteBlockBasis(const Lattice<iVector<vComplex,nbasis > > &coarseD
   
 }
 
-// useful in multigrid project;
-// Generic name : Coarsen?
-template<class vobj>
-inline void sumBlocks(Lattice<vobj> &coarseData,const Lattice<vobj> &fineData)
-{
-  GridBase * fine  = fineData._grid;
-  GridBase * coarse= coarseData._grid;
-
-  subdivides(coarse,fine); // require they map
-
-  int _ndimension = coarse->_ndimension;
-  
-  std::vector<int>  block_r      (_ndimension);
-  
-  for(int d=0 ; d<_ndimension;d++){
-    block_r[d] = fine->_rdimensions[d] / coarse->_rdimensions[d];
-  }
-
-  coarseData=zero;
-  for(int sf=0;sf<fine->oSites();sf++){
-    
-    int sc;
-    std::vector<int> coor_c(_ndimension);
-    std::vector<int> coor_f(_ndimension);
-
-    GridBase::CoorFromIndex(coor_f,sf,fine->_rdimensions);
-    for(int d=0;d<_ndimension;d++) coor_c[d]=coor_f[d]/block_r[d];
-    GridBase::IndexFromCoor(coor_c,sc,coarse->_rdimensions);
-
-    coarseData._odata[sc]=coarseData._odata[sc]+fineData._odata[sf];
-
-  }
-  return;
-}
 
 }
 #endif
diff --git a/lib/lattice/Lattice_unary.h b/lib/lattice/Lattice_unary.h
new file mode 100644
index 00000000..131d7d68
--- /dev/null
+++ b/lib/lattice/Lattice_unary.h
@@ -0,0 +1,32 @@
+#ifndef GRID_LATTICE_UNARY_H
+#define GRID_LATTICE_UNARY_H
+
+namespace Grid {
+
+  //////////////////////////////////////////////////////////////////////////////////////////////////////
+  //  avoid copy back routines for mult, mac, sub, add
+  //////////////////////////////////////////////////////////////////////////////////////////////////////
+  template<class obj> Lattice<obj> sqrt(const Lattice<obj> &rhs){
+    Lattice<obj> ret(rhs._grid);
+    ret.checkerboard = rhs.checkerboard;
+    conformable(ret,rhs);
+PARALLEL_FOR_LOOP
+    for(int ss=0;ss<rhs._grid->oSites();ss++){
+      ret._odata[ss]=sqrt(rhs._odata[ss]);
+    }
+    return ret;
+  }
+  template<class obj> Lattice<obj> rsqrt(const Lattice<obj> &rhs){
+    Lattice<obj> ret(rhs._grid);
+    ret.checkerboard = rhs.checkerboard;
+    conformable(ret,rhs);
+PARALLEL_FOR_LOOP
+    for(int ss=0;ss<rhs._grid->oSites();ss++){
+      ret._odata[ss]=rsqrt(rhs._odata[ss]);
+    }
+    return ret;
+  }
+
+
+}
+#endif
diff --git a/lib/qcd/action/fermion/CayleyFermion5D.cc b/lib/qcd/action/fermion/CayleyFermion5D.cc
index e47ff331..7db67dd7 100644
--- a/lib/qcd/action/fermion/CayleyFermion5D.cc
+++ b/lib/qcd/action/fermion/CayleyFermion5D.cc
@@ -165,6 +165,27 @@ namespace QCD {
     }
   }
 
+  void  CayleyFermion5D::Mdir (const LatticeFermion &psi, LatticeFermion &chi,int dir,int disp){
+    LatticeFermion tmp(psi._grid);
+    // Assemble the 5d matrix
+    for(int s=0;s<Ls;s++){
+      if ( s==0 ) {
+	//	tmp = bs psi[s] + cs[s] psi[s+1}
+	//      tmp+= -mass*cs[s] psi[s+1}
+	axpby_ssp_pminus(tmp,beo[s],psi,-ceo[s],psi ,s, s+1);
+	axpby_ssp_pplus(tmp,1.0,tmp,mass*ceo[s],psi,s,Ls-1);
+      } else if ( s==(Ls-1)) { 
+	axpby_ssp_pminus(tmp,beo[s],psi,mass*ceo[s],psi,s,0);
+	axpby_ssp_pplus(tmp,1.0,tmp,-ceo[s],psi,s,s-1);
+      } else {
+	axpby_ssp_pminus(tmp,beo[s],psi,-ceo[s],psi,s,s+1);
+	axpby_ssp_pplus (tmp,1.0,tmp,-ceo[s],psi,s,s-1);
+      }
+    }
+    // Apply 4d dslash fragment
+    DhopDir(tmp,chi,dir,disp);
+  }
+
   void CayleyFermion5D::MooeeDag    (const LatticeFermion &psi, LatticeFermion &chi)
   {
     for (int s=0;s<Ls;s++){
diff --git a/lib/qcd/action/fermion/CayleyFermion5D.h b/lib/qcd/action/fermion/CayleyFermion5D.h
index e2175d77..208ab0bb 100644
--- a/lib/qcd/action/fermion/CayleyFermion5D.h
+++ b/lib/qcd/action/fermion/CayleyFermion5D.h
@@ -21,6 +21,10 @@ namespace Grid {
       virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out);
       virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
       virtual void   Instantiatable(void)=0;
+
+      // Efficient support for multigrid coarsening
+      virtual void  Mdir (const LatticeFermion &in, LatticeFermion &out,int dir,int disp);
+
       //    protected:
       RealD mass;
 
diff --git a/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc b/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
index 6f99aa7f..0d54f272 100644
--- a/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
+++ b/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
@@ -90,6 +90,18 @@ namespace Grid {
       // Can ignore "dag"
       return M(psi,chi);
     }
+    void  ContinuedFractionFermion5D::Mdir (const LatticeFermion &psi, LatticeFermion &chi,int dir,int disp){
+      DhopDir(psi,chi,dir,disp); // Dslash on diagonal. g5 Dslash is hermitian
+      int sign=1;
+      for(int s=0;s<Ls;s++){
+	if ( s==(Ls-1) ){
+	  ag5xpby_ssp(chi,Beta[s]*ZoloHiInv,chi,0.0,chi,s,s);
+	} else {
+	  ag5xpby_ssp(chi,cc[s]*Beta[s]*sign*ZoloHiInv,chi,0.0,chi,s,s);
+	}
+	sign=-sign; 
+      }
+    }
     void   ContinuedFractionFermion5D::Meooe       (const LatticeFermion &psi, LatticeFermion &chi)
     {
       // Apply 4d dslash
diff --git a/lib/qcd/action/fermion/ContinuedFractionFermion5D.h b/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
index b7e321c9..43b0f756 100644
--- a/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
+++ b/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
@@ -24,6 +24,9 @@ namespace Grid {
       //      virtual void   Instantiatable(void)=0;
       virtual void   Instantiatable(void) =0;
 
+      // Efficient support for multigrid coarsening
+      virtual void  Mdir (const LatticeFermion &in, LatticeFermion &out,int dir,int disp);
+
       // Constructors
       ContinuedFractionFermion5D(LatticeGaugeField &_Umu,
 				 GridCartesian         &FiveDimGrid,
diff --git a/lib/qcd/action/fermion/FermionOperator.h b/lib/qcd/action/fermion/FermionOperator.h
index 47c47478..1aa2f8f0 100644
--- a/lib/qcd/action/fermion/FermionOperator.h
+++ b/lib/qcd/action/fermion/FermionOperator.h
@@ -40,6 +40,9 @@ namespace Grid {
       virtual void DhopOE(const FermionField &in, FermionField &out,int dag)=0;
       virtual void DhopEO(const FermionField &in, FermionField &out,int dag)=0;
 
+      virtual void  Mdiag(const FermionField &in, FermionField &out) { Mooee(in,out);};   // Same as Mooee applied to both CB's
+      virtual void  Mdir (const FermionField &in, FermionField &out,int dir,int disp)=0;   // case by case Wilson, Clover, Cayley, ContFrac, PartFrac
+      virtual void  DhopDir(const FermionField &in, FermionField &out,int dir,int disp)=0; // implemented by WilsonFermion and WilsonFermion5D
 
     };
 
diff --git a/lib/qcd/action/fermion/PartialFractionFermion5D.cc b/lib/qcd/action/fermion/PartialFractionFermion5D.cc
index 285144c6..a148fe40 100644
--- a/lib/qcd/action/fermion/PartialFractionFermion5D.cc
+++ b/lib/qcd/action/fermion/PartialFractionFermion5D.cc
@@ -2,6 +2,22 @@
 namespace Grid {
   namespace QCD {
 
+    void  PartialFractionFermion5D::Mdir (const LatticeFermion &psi, LatticeFermion &chi,int dir,int disp){
+      // this does both dag and undag but is trivial; make a common helper routing
+
+      int sign = 1;
+
+      DhopDir(psi,chi,dir,disp);
+
+      int nblock=(Ls-1)/2;
+      for(int b=0;b<nblock;b++){
+	int s = 2*b;
+	ag5xpby_ssp(chi,-scale,chi,0.0,chi,s,s); 
+	ag5xpby_ssp(chi, scale,chi,0.0,chi,s+1,s+1); 
+      }
+      ag5xpby_ssp(chi,p[nblock]*scale/amax,chi,0.0,chi,Ls-1,Ls-1);
+
+    }
     void   PartialFractionFermion5D::Meooe_internal(const LatticeFermion &psi, LatticeFermion &chi,int dag)
     {
       // this does both dag and undag but is trivial; make a common helper routing
diff --git a/lib/qcd/action/fermion/PartialFractionFermion5D.h b/lib/qcd/action/fermion/PartialFractionFermion5D.h
index 301b07c0..e734caa0 100644
--- a/lib/qcd/action/fermion/PartialFractionFermion5D.h
+++ b/lib/qcd/action/fermion/PartialFractionFermion5D.h
@@ -30,6 +30,9 @@ namespace Grid {
 
       virtual void   Instantiatable(void) =0; // ensure no make-eee
 
+      // Efficient support for multigrid coarsening
+      virtual void  Mdir (const LatticeFermion &in, LatticeFermion &out,int dir,int disp);
+
       // Constructors
       PartialFractionFermion5D(LatticeGaugeField &_Umu,
 				    GridCartesian         &FiveDimGrid,
diff --git a/lib/qcd/action/fermion/WilsonFermion.cc b/lib/qcd/action/fermion/WilsonFermion.cc
index 9f2da251..13f16c5a 100644
--- a/lib/qcd/action/fermion/WilsonFermion.cc
+++ b/lib/qcd/action/fermion/WilsonFermion.cc
@@ -93,6 +93,26 @@ void WilsonFermion::MooeeInvDag(const LatticeFermion &in, LatticeFermion &out)
   out.checkerboard = in.checkerboard;
   MooeeInv(in,out);
 }
+void WilsonFermion::Mdir (const LatticeFermion &in, LatticeFermion &out,int dir,int disp)
+{
+  DhopDir(in,out,dir,disp);
+}
+void WilsonFermion::DhopDir(const LatticeFermion &in, LatticeFermion &out,int dir,int disp){
+  WilsonCompressor compressor(DaggerNo);
+  Stencil.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
+  
+  assert( (disp==1)||(disp==-1) );
+
+  int skip = (disp==1) ? 0 : 1;
+
+  int dirdisp = dir+skip*4;
+
+PARALLEL_FOR_LOOP
+  for(int sss=0;sss<in._grid->oSites();sss++){
+    DiracOpt::DhopDir(Stencil,Umu,comm_buf,sss,sss,in,out,dirdisp);
+  }
+
+};
 
 void WilsonFermion::DhopInternal(CartesianStencil & st,LatticeDoubledGaugeField & U,
 				const LatticeFermion &in, LatticeFermion &out,int dag)
diff --git a/lib/qcd/action/fermion/WilsonFermion.h b/lib/qcd/action/fermion/WilsonFermion.h
index 6040d328..60726fdd 100644
--- a/lib/qcd/action/fermion/WilsonFermion.h
+++ b/lib/qcd/action/fermion/WilsonFermion.h
@@ -26,7 +26,7 @@ namespace Grid {
       void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out);
       virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out); // remain virtual so we 
       virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out); // can derive Clover
-      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out); // from Wilson bas
+      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out); // from Wilson base
       virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
 
       // non-hermitian hopping term; half cb or both
@@ -34,6 +34,10 @@ namespace Grid {
       void DhopOE(const LatticeFermion &in, LatticeFermion &out,int dag);
       void DhopEO(const LatticeFermion &in, LatticeFermion &out,int dag);
 
+      // Multigrid assistance
+      void   Mdir (const LatticeFermion &in, LatticeFermion &out,int dir,int disp);
+      void DhopDir(const LatticeFermion &in, LatticeFermion &out,int dir,int disp);
+
       ///////////////////////////////////////////////////////////////
       // Extra methods added by derived
       ///////////////////////////////////////////////////////////////
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.cc b/lib/qcd/action/fermion/WilsonFermion5D.cc
index d22701b0..4ddd14ab 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.cc
+++ b/lib/qcd/action/fermion/WilsonFermion5D.cc
@@ -82,6 +82,28 @@ void WilsonFermion5D::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGau
     pokeIndex<LorentzIndex>(Uds,U,mu+4);
   }
 }
+void WilsonFermion5D::DhopDir(const LatticeFermion &in, LatticeFermion &out,int dir,int disp)
+{
+  assert( (disp==1)||(disp==-1) );
+
+  WilsonCompressor compressor(DaggerNo);
+  Stencil.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
+  
+  int skip = (disp==1) ? 0 : 1;
+
+  int dirdisp = dir+skip*4;
+
+PARALLEL_FOR_LOOP
+  for(int ss=0;ss<Umu._grid->oSites();ss++){
+    for(int s=0;s<Ls;s++){
+      int sU=ss;
+      int sF = s+Ls*sU; 
+      DiracOpt::DhopDir(Stencil,Umu,comm_buf,sF,sU,in,out,dirdisp);
+    }
+  }
+
+};
+
 void WilsonFermion5D::DhopInternal(CartesianStencil & st, LebesgueOrder &lo,
 				   LatticeDoubledGaugeField & U,
 				   const LatticeFermion &in, LatticeFermion &out,int dag)
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.h b/lib/qcd/action/fermion/WilsonFermion5D.h
index 4d4a7aed..10138991 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.h
+++ b/lib/qcd/action/fermion/WilsonFermion5D.h
@@ -57,6 +57,10 @@ namespace Grid {
       void DhopOE(const LatticeFermion &in, LatticeFermion &out,int dag);
       void DhopEO(const LatticeFermion &in, LatticeFermion &out,int dag);
 
+      // add a DhopComm
+      // -- suboptimal interface will presently trigger multiple comms.
+      void DhopDir(const LatticeFermion &in, LatticeFermion &out,int dir,int disp);
+
       ///////////////////////////////////////////////////////////////
       // New methods added 
       ///////////////////////////////////////////////////////////////
diff --git a/lib/qcd/action/fermion/WilsonKernels.cc b/lib/qcd/action/fermion/WilsonKernels.cc
index 879bbb7a..518b8004 100644
--- a/lib/qcd/action/fermion/WilsonKernels.cc
+++ b/lib/qcd/action/fermion/WilsonKernels.cc
@@ -13,7 +13,6 @@ void DiracOpt::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     vHalfSpinColourVector Uchi;
     int offset,local,perm, ptype;
 
-    //#define VERBOSE( A)  if ( ss<10 ) { std::cout << "site " <<ss << " " #A " neigh " << offset << " perm "<< perm <<std::endl;}    
 
     // Xp
     int ss = sF;
@@ -33,12 +32,6 @@ void DiracOpt::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     mult(&Uchi(),&U._odata[sU](Xp),&chi());
     spReconXp(result,Uchi);
 
-    //    std::cout << "XP_RECON"<<std::endl;
-    //    std::cout << result()(0)(0) <<" "<<result()(0)(1) <<" "<<result()(0)(2) <<std::endl;
-    //    std::cout << result()(1)(0) <<" "<<result()(1)(1) <<" "<<result()(1)(2) <<std::endl;
-    //    std::cout << result()(2)(0) <<" "<<result()(2)(1) <<" "<<result()(2)(2) <<std::endl;
-    //    std::cout << result()(3)(0) <<" "<<result()(3)(1) <<" "<<result()(3)(2) <<std::endl;
-
     // Yp
     offset = st._offsets [Yp][ss];
     local  = st._is_local[Yp][ss];
@@ -93,8 +86,7 @@ void DiracOpt::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     perm   = st._permute[Xm][ss];
     ptype  = st._permute_type[Xm];
 
-    if ( local && perm ) 
-    {
+    if ( local && perm ) {
       spProjXm(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
     } else if ( local ) {
@@ -104,12 +96,6 @@ void DiracOpt::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     }
     mult(&Uchi(),&U._odata[sU](Xm),&chi());
     accumReconXm(result,Uchi);
-    //  std::cout << "XM_RECON_ACCUM"<<std::endl;
-    //    std::cout << result()(0)(0) <<" "<<result()(0)(1) <<" "<<result()(0)(2) <<std::endl;
-    //    std::cout << result()(1)(0) <<" "<<result()(1)(1) <<" "<<result()(1)(2) <<std::endl;
-    //    std::cout << result()(2)(0) <<" "<<result()(2)(1) <<" "<<result()(2)(2) <<std::endl;
-    //    std::cout << result()(3)(0) <<" "<<result()(3)(1) <<" "<<result()(3)(2) <<std::endl;
-
 
     // Ym
     offset = st._offsets [Ym][ss];
@@ -308,4 +294,136 @@ void DiracOpt::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
 
     vstream(out._odata[ss],result*(-0.5));
 }
+
+void DiracOpt::DhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
+			std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+		       int sF,int sU,const LatticeFermion &in, LatticeFermion &out,int dirdisp)
+{
+    vHalfSpinColourVector  tmp;    
+    vHalfSpinColourVector  chi;    
+    vSpinColourVector result;
+    vHalfSpinColourVector Uchi;
+    int offset,local,perm, ptype;
+    int ss=sF;
+    
+    offset = st._offsets [dirdisp][ss];
+    local  = st._is_local[dirdisp][ss];
+    perm   = st._permute[dirdisp][ss];
+    ptype  = st._permute_type[dirdisp];
+
+    // Xp
+    if(dirdisp==Xp){
+      if ( local && perm ) {
+	spProjXp(tmp,in._odata[offset]);
+	permute(chi,tmp,ptype);
+      } else if ( local ) {
+	spProjXp(chi,in._odata[offset]);
+      } else { 
+	chi=buf[offset];
+      }
+      mult(&Uchi(),&U._odata[sU](Xp),&chi());
+      spReconXp(result,Uchi);
+    }
+
+    // Yp
+    if ( dirdisp==Yp ){
+      if ( local && perm ) {
+	spProjYp(tmp,in._odata[offset]);
+	permute(chi,tmp,ptype);
+      } else if ( local ) {
+	spProjYp(chi,in._odata[offset]);
+      } else { 
+	chi=buf[offset];
+      }
+      mult(&Uchi(),&U._odata[sU](Yp),&chi());
+      spReconYp(result,Uchi);
+    }
+
+    // Zp
+    if ( dirdisp ==Zp ){
+      if ( local && perm ) {
+	spProjZp(tmp,in._odata[offset]);
+	permute(chi,tmp,ptype);
+      } else if ( local ) {
+	spProjZp(chi,in._odata[offset]);
+      } else { 
+	chi=buf[offset];
+      }
+      mult(&Uchi(),&U._odata[sU](Zp),&chi());
+      spReconZp(result,Uchi);
+    }
+
+    // Tp
+    if ( dirdisp ==Tp ){
+      if ( local && perm ) {
+	spProjTp(tmp,in._odata[offset]);
+	permute(chi,tmp,ptype);
+      } else if ( local ) {
+	spProjTp(chi,in._odata[offset]);
+      } else { 
+	chi=buf[offset];
+      }
+      mult(&Uchi(),&U._odata[sU](Tp),&chi());
+      spReconTp(result,Uchi);
+    }
+
+    // Xm
+    if ( dirdisp==Xm ){
+      if ( local && perm ) {
+	spProjXm(tmp,in._odata[offset]);
+	permute(chi,tmp,ptype);
+      } else if ( local ) {
+	spProjXm(chi,in._odata[offset]);
+      } else { 
+	chi=buf[offset];
+      }
+      mult(&Uchi(),&U._odata[sU](Xm),&chi());
+      spReconXm(result,Uchi);
+    }
+
+    // Ym
+    if ( dirdisp == Ym ){
+      if ( local && perm ) {
+	spProjYm(tmp,in._odata[offset]);
+	permute(chi,tmp,ptype);
+      } else if ( local ) {
+	spProjYm(chi,in._odata[offset]);
+      } else { 
+	chi=buf[offset];
+      }
+      mult(&Uchi(),&U._odata[sU](Ym),&chi());
+      spReconYm(result,Uchi);
+    }
+
+    // Zm
+    if ( dirdisp == Zm ){
+      if ( local && perm ) {
+	spProjZm(tmp,in._odata[offset]);
+	permute(chi,tmp,ptype);
+      } else if ( local ) {
+	spProjZm(chi,in._odata[offset]);
+      } else { 
+	chi=buf[offset];
+      }
+      mult(&Uchi(),&U._odata[sU](Zm),&chi());
+      spReconZm(result,Uchi);
+    }
+
+    // Tm
+    if ( dirdisp==Tm ) {
+      if ( local && perm ) {
+	spProjTm(tmp,in._odata[offset]);
+	permute(chi,tmp,ptype);
+      } else if ( local ) {
+	spProjTm(chi,in._odata[offset]);
+      } else { 
+	chi=buf[offset];
+      }
+      mult(&Uchi(),&U._odata[sU](Tm),&chi());
+      spReconTm(result,Uchi);
+    }
+
+    vstream(out._odata[ss],result*(-0.5));
+}
+
 }}
diff --git a/lib/qcd/action/fermion/WilsonKernels.h b/lib/qcd/action/fermion/WilsonKernels.h
index cc535022..db384ddf 100644
--- a/lib/qcd/action/fermion/WilsonKernels.h
+++ b/lib/qcd/action/fermion/WilsonKernels.h
@@ -20,6 +20,9 @@ namespace Grid {
       static void DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
 			      std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
 			      int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
+      static void DhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
+			   std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+			  int sF,int sU,const LatticeFermion &in, LatticeFermion &out,int dirdisp);
 
     };
 
diff --git a/lib/qcd/action/fermion/g5HermitianLinop.h b/lib/qcd/action/fermion/g5HermitianLinop.h
index 9a8cc2eb..5b3411ec 100644
--- a/lib/qcd/action/fermion/g5HermitianLinop.h
+++ b/lib/qcd/action/fermion/g5HermitianLinop.h
@@ -11,11 +11,22 @@ class Gamma5HermitianLinearOperator : public LinearOperatorBase<Field> {
 public:
   Gamma5HermitianLinearOperator(Matrix &Mat): _Mat(Mat){};
   void Op     (const Field &in, Field &out){
-    _Mat.M(in,out);
+    HermOp(in,out);
   }
   void AdjOp     (const Field &in, Field &out){
-    _Mat.M(in,out);
+    HermOp(in,out);
   }
+  void OpDiag (const Field &in, Field &out) {
+    Field tmp(in._grid);
+    _Mat.Mdiag(in,tmp);
+    G5R5(out,tmp);
+  }
+  void OpDir  (const Field &in, Field &out,int dir,int disp) {
+    Field tmp(in._grid);
+    _Mat.Mdir(in,tmp,dir,disp);
+    G5R5(out,tmp);
+  }
+
   void HermOpAndNorm(const Field &in, Field &out,RealD &n1,RealD &n2){
 
     HermOp(in,out);
diff --git a/lib/simd/Grid_avx.h b/lib/simd/Grid_avx.h
index ee4d03d5..ba023146 100644
--- a/lib/simd/Grid_avx.h
+++ b/lib/simd/Grid_avx.h
@@ -314,9 +314,9 @@ namespace Optimization {
   template<>
     inline Grid::ComplexF Reduce<Grid::ComplexF, __m256>::operator()(__m256 in){
     __m256 v1,v2;
-    Optimization::permute(v1,in,0); // sse 128; paired complex single
+    Optimization::permute(v1,in,0); // avx 256; quad complex single
     v1 = _mm256_add_ps(v1,in);
-    Optimization::permute(v2,v1,1); // avx 256; quad complex single
+    Optimization::permute(v2,v1,1); 
     v1 = _mm256_add_ps(v1,v2);
     u256f conv; conv.v = v1;
     return Grid::ComplexF(conv.f[0],conv.f[1]);
@@ -367,7 +367,6 @@ namespace Optimization {
     assert(0);
   }
   
-  
 }
 
 //////////////////////////////////////////////////////////////////////////////////////
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index aa1abb47..9525e45e 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -78,11 +78,18 @@ namespace Grid {
     typedef typename RealPart < Scalar_type >::type Real; 
     typedef Vector_type     vector_type;
     typedef Scalar_type     scalar_type;
+
+
+    typedef union conv_t_union {
+	Vector_type v;
+	Scalar_type s[sizeof(Vector_type)/sizeof(Scalar_type)];
+      conv_t_union(){};
+      } conv_t;
+    
    
     Vector_type v;
     
     static inline int Nsimd(void) { return sizeof(Vector_type)/sizeof(Scalar_type);}
-
     
     Grid_simd& operator=(const Grid_simd&& rhs){v=rhs.v;return *this;};
     Grid_simd& operator=(const Grid_simd& rhs){v=rhs.v;return *this;}; //faster than not declaring it and leaving to the compiler
@@ -192,6 +199,27 @@ namespace Grid {
       return *this;
     }
 
+
+    ///////////////////////////////////////
+    // Not all functions are supported
+    // through SIMD and must breakout to 
+    // scalar type and back again. This
+    // provides support
+    ///////////////////////////////////////
+
+    template<class functor> friend inline Grid_simd SimdApply (const functor &func,const Grid_simd &v) {
+
+      Grid_simd ret;
+      Grid_simd::conv_t conv;
+
+      conv.v = v.v;
+      for(int i=0;i<Nsimd();i++){
+	conv.s[i]=func(conv.s[i]);
+      }
+      ret.v = conv.v;
+      return ret;
+    }
+
     ////////////////////////////////////////////////////////////////////
     // General permute; assumes vector length is same across 
     // all subtypes; may not be a good assumption, but could
diff --git a/lib/simd/Grid_vector_unops.h b/lib/simd/Grid_vector_unops.h
new file mode 100644
index 00000000..9aeec350
--- /dev/null
+++ b/lib/simd/Grid_vector_unops.h
@@ -0,0 +1,60 @@
+#ifndef GRID_VECTOR_UNOPS
+#define GRID_VECTOR_UNOPS
+
+namespace Grid { 
+
+  template<class scalar> struct SqrtRealFunctor {
+    scalar operator()(const scalar &a) const {
+      return sqrt(real(a));
+    }
+  };
+
+  template<class scalar> struct RSqrtRealFunctor {
+    scalar operator()(const scalar &a)  const {
+      return scalar(1.0/sqrt(real(a)));
+    }
+  };
+
+  template<class scalar> struct CosRealFunctor {
+    scalar operator()(const scalar &a)  const {
+      return cos(real(a));
+    }
+  };
+
+  template<class scalar> struct SinRealFunctor {
+    scalar operator()(const scalar &a)  const {
+      return sin(real(a));
+    }
+  };
+
+  template<class scalar> struct PowRealFunctor {
+    double y;
+  PowRealFunctor(double _y) : y(_y) {};
+    scalar operator()(const scalar &a)  const {
+      return pow(real(a),y);
+    }
+  };
+
+  template < class S, class V > 
+  inline Grid_simd<S,V> sqrt(const Grid_simd<S,V> &r) {
+    return SimdApply(SqrtRealFunctor<S>(),r);
+  }
+  template < class S, class V > 
+  inline Grid_simd<S,V> rsqrt(const Grid_simd<S,V> &r) {
+    return SimdApply(RSqrtRealFunctor<S>(),r);
+  }
+  template < class S, class V > 
+  inline Grid_simd<S,V> cos(const Grid_simd<S,V> &r) {
+    return SimdApply(CosRealFunctor<S>(),r);
+  }
+  template < class S, class V > 
+  inline Grid_simd<S,V> sin(const Grid_simd<S,V> &r) {
+    return SimdApply(CosRealFunctor<S>(),r);
+  }
+  template < class S, class V > 
+  inline Grid_simd<S,V> pow(const Grid_simd<S,V> &r,double y) {
+    return SimdApply(PowRealFunctor<S>(y),r);
+  }
+
+}
+#endif
diff --git a/lib/tensors/Tensor_class.h b/lib/tensors/Tensor_class.h
index bd1902b8..3ed4f865 100644
--- a/lib/tensors/Tensor_class.h
+++ b/lib/tensors/Tensor_class.h
@@ -123,6 +123,13 @@ public:
   typedef iScalar<tensor_reduced_v> tensor_reduced;
   typedef iVector<recurse_scalar_object,N> scalar_object;
 
+  template<class T,typename std::enable_if<!isGridTensor<T>::value, T>::type* = nullptr > strong_inline auto operator = (T arg) -> iVector<vtype,N>
+    { 
+      zeroit(*this);
+      for(int i=0;i<N;i++)
+	_internal[i] = arg;
+      return *this;
+    }
 
   enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
   iVector(const Zero &z){ *this = zero; };
diff --git a/lib/tensors/Tensor_unary.h b/lib/tensors/Tensor_unary.h
new file mode 100644
index 00000000..1fbec985
--- /dev/null
+++ b/lib/tensors/Tensor_unary.h
@@ -0,0 +1,37 @@
+#ifndef GRID_TENSOR_UNARY_H
+#define GRID_TENSOR_UNARY_H
+namespace Grid {
+
+#define UNARY_REAL(func)\
+template<class obj> inline auto func(const iScalar<obj> &z) -> iScalar<obj>\
+{\
+    iScalar<obj> ret;\
+    ret._internal = func( (z._internal));\
+    return ret;\
+}\
+template<class obj,int N> inline auto func(const iVector<obj,N> &z) -> iVector<obj,N>\
+{\
+    iVector<obj,N> ret;\
+    for(int c1=0;c1<N;c1++){\
+      ret._internal[c1] = func( (z._internal[c1]));\
+    }\
+    return ret;\
+}\
+template<class obj,int N> inline auto func(const iMatrix<obj,N> &z) -> iMatrix<obj,N>\
+{\
+    iMatrix<obj,N> ret;\
+    for(int c1=0;c1<N;c1++){\
+    for(int c2=0;c2<N;c2++){\
+      ret._internal[c1][c2] = func( (z._internal[c1][c2]));\
+    }}\
+    return ret;\
+}
+
+UNARY_REAL(sqrt);
+UNARY_REAL(rsqrt);
+UNARY_REAL(sin);
+UNARY_REAL(cos);
+
+
+}
+#endif
diff --git a/tests/Make.inc b/tests/Make.inc
index de9d0709..d417ab56 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,5 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_even_odd Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_even_odd Test_gamma Test_main Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_even_odd
+bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_gamma Test_main Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
 
 
 Test_GaugeAction_SOURCES=Test_GaugeAction.cc
@@ -10,10 +10,18 @@ Test_cayley_cg_SOURCES=Test_cayley_cg.cc
 Test_cayley_cg_LDADD=-lGrid
 
 
+Test_cayley_coarsen_support_SOURCES=Test_cayley_coarsen_support.cc
+Test_cayley_coarsen_support_LDADD=-lGrid
+
+
 Test_cayley_even_odd_SOURCES=Test_cayley_even_odd.cc
 Test_cayley_even_odd_LDADD=-lGrid
 
 
+Test_cf_coarsen_support_SOURCES=Test_cf_coarsen_support.cc
+Test_cf_coarsen_support_LDADD=-lGrid
+
+
 Test_cf_cr_unprec_SOURCES=Test_cf_cr_unprec.cc
 Test_cf_cr_unprec_LDADD=-lGrid
 
diff --git a/tests/Test_GaugeAction.cc b/tests/Test_GaugeAction.cc
index 0dde6d8f..bbfd8717 100644
--- a/tests/Test_GaugeAction.cc
+++ b/tests/Test_GaugeAction.cc
@@ -115,7 +115,7 @@ int main (int argc, char ** argv)
   Complex l  = TensorRemove(Tl);
   std::cout << "calculated link trace " <<l*LinkTraceScale<<std::endl;
 
-  sumBlocks(cPlaq,Plaq);
+  blockSum(cPlaq,Plaq);
   TComplex TcP = sum(cPlaq);
   Complex ll= TensorRemove(TcP);
   std::cout << "coarsened plaquettes sum to " <<ll*PlaqScale<<std::endl;
diff --git a/tests/Test_cayley_coarsen_support.cc b/tests/Test_cayley_coarsen_support.cc
new file mode 100644
index 00000000..acd96da3
--- /dev/null
+++ b/tests/Test_cayley_coarsen_support.cc
@@ -0,0 +1,108 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+
+
+  const int Ls=8;
+
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+  // Construct a coarsened grid
+  std::vector<int> clatt = GridDefaultLatt();
+  for(int d=0;d<clatt.size();d++){
+    clatt[d] = clatt[d]/2;
+  }
+  GridCartesian *Coarse4d =  SpaceTimeGrid::makeFourDimGrid(clatt, GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());;
+  GridCartesian *Coarse5d =  SpaceTimeGrid::makeFiveDimGrid(1,Coarse4d);
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);  RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);  RNG4.SeedFixedIntegers(seeds4);
+
+  LatticeFermion    src(FGrid); random(RNG5,src);
+  LatticeFermion result(FGrid); result=zero;
+  LatticeFermion    ref(FGrid); ref=zero;
+  LatticeFermion    tmp(FGrid);
+  LatticeFermion    err(FGrid);
+  LatticeGaugeField Umu(UGrid); random(RNG4,Umu);
+
+  std::vector<LatticeColourMatrix> U(4,UGrid);
+
+  for(int mu=0;mu<Nd;mu++){
+    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+  }
+  
+  RealD mass=0.5;
+  RealD M5=1.8;
+
+  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  Gamma5HermitianLinearOperator<DomainWallFermion,LatticeFermion> HermIndefOp(Ddwf);
+
+  HermIndefOp.Op(src,ref);
+  HermIndefOp.OpDiag(src,result);
+  
+  for(int d=0;d<4;d++){
+    HermIndefOp.OpDir(src,tmp,d,+1); result=result+tmp; 
+    std::cout<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
+    HermIndefOp.OpDir(src,tmp,d,-1); result=result+tmp;
+    std::cout<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
+  }
+  err = result-ref;
+  std::cout<<"Error "<<norm2(err)<<std::endl;
+
+  const int nbasis = 8;
+  std::vector<LatticeFermion> subspace(nbasis,FGrid);
+  
+  for(int b=0;b<nbasis;b++){
+    random(RNG5,subspace[b]);
+  }
+  std::cout << "Computed randoms"<< std::endl;
+
+  CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis> LittleDiracOp(*Coarse5d);
+
+  LittleDiracOp.CoarsenOperator(FGrid,HermIndefOp,subspace);
+
+  typedef   Lattice<iVector<vComplex,nbasis > > coarse_vec;
+  
+  coarse_vec c_src (Coarse5d);  c_src= zero;
+  coarse_vec c_res (Coarse5d);
+  
+  Complex one(1.0);
+  c_src = one;  // 1 in every element for vector 1.
+
+  //  TODO
+  // -- promote from subspace, check we get the vector we wanted
+  // -- apply ldop; check we get the same as inner product of M times big vec
+  // -- pick blocks one by one. Evaluate matrix elements.
+
+  std::cout << "Multiplying by LittleDiracOp "<< std::endl;
+  LittleDiracOp.M(c_src,c_res);
+
+  std::cout << "Done "<< std::endl;
+  Grid_finalize();
+}
diff --git a/tests/Test_cf_coarsen_support.cc b/tests/Test_cf_coarsen_support.cc
new file mode 100644
index 00000000..9d81a798
--- /dev/null
+++ b/tests/Test_cf_coarsen_support.cc
@@ -0,0 +1,87 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  const int Ls=9;
+
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);  RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);  RNG4.SeedFixedIntegers(seeds4);
+
+  LatticeFermion    src(FGrid); random(RNG5,src);
+  LatticeFermion result(FGrid); result=zero;
+  LatticeFermion    ref(FGrid); ref=zero;
+  LatticeFermion    tmp(FGrid);
+  LatticeFermion    err(FGrid);
+  LatticeGaugeField Umu(UGrid); random(RNG4,Umu);
+
+  std::vector<LatticeColourMatrix> U(4,UGrid);
+  for(int mu=0;mu<Nd;mu++){
+    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+  }
+  
+  RealD mass=0.1;
+  RealD M5=1.8;
+
+  {
+    OverlapWilsonContFracTanhFermion Dcf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+    HermitianLinearOperator<OverlapWilsonContFracTanhFermion,LatticeFermion> HermIndefOp(Dcf);
+
+    HermIndefOp.Op(src,ref);
+    HermIndefOp.OpDiag(src,result);
+    
+    for(int d=0;d<4;d++){
+      HermIndefOp.OpDir(src,tmp,d,+1); result=result+tmp; 
+      std::cout<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
+      HermIndefOp.OpDir(src,tmp,d,-1); result=result+tmp;
+      std::cout<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
+    }
+    err = result-ref;
+    std::cout<<"Error "<<norm2(err)<<std::endl;
+  }
+
+  {
+    OverlapWilsonPartialFractionTanhFermion Dpf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+    HermitianLinearOperator<OverlapWilsonPartialFractionTanhFermion,LatticeFermion> HermIndefOp(Dpf);
+    
+    HermIndefOp.Op(src,ref);
+    HermIndefOp.OpDiag(src,result);
+    
+    for(int d=0;d<4;d++){
+      HermIndefOp.OpDir(src,tmp,d,+1); result=result+tmp; 
+      std::cout<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
+      HermIndefOp.OpDir(src,tmp,d,-1); result=result+tmp;
+      std::cout<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
+    }
+
+    err = result-ref;
+    std::cout<<"Error "<<norm2(err)<<std::endl;
+  }
+
+
+  Grid_finalize();
+}
diff --git a/tests/Test_cf_cr_unprec.cc b/tests/Test_cf_cr_unprec.cc
new file mode 100644
index 00000000..9c1aea60
--- /dev/null
+++ b/tests/Test_cf_cr_unprec.cc
@@ -0,0 +1,58 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  const int Ls=9;
+
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);  RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);  RNG4.SeedFixedIntegers(seeds4);
+
+  LatticeFermion    src(FGrid); random(RNG5,src);
+  LatticeFermion result(FGrid); result=zero;
+  LatticeGaugeField Umu(UGrid); random(RNG4,Umu);
+
+  std::vector<LatticeColourMatrix> U(4,UGrid);
+  for(int mu=0;mu<Nd;mu++){
+    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+  }
+  
+  RealD mass=0.1;
+  RealD M5=1.8;
+
+  OverlapWilsonContFracTanhFermion Dcf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+
+  ConjugateResidual<LatticeFermion> MCR(1.0e-8,10000);
+
+  MdagMLinearOperator<OverlapWilsonContFracTanhFermion,LatticeFermion> HermPosDefOp(Dcf);
+  MCR(HermPosDefOp,src,result);
+
+  HermitianLinearOperator<OverlapWilsonContFracTanhFermion,LatticeFermion> HermIndefOp(Dcf);
+  MCR(HermIndefOp,src,result);
+
+  Grid_finalize();
+}
diff --git a/tests/Test_dwf_cr_unprec.cc b/tests/Test_dwf_cr_unprec.cc
new file mode 100644
index 00000000..cffa726c
--- /dev/null
+++ b/tests/Test_dwf_cr_unprec.cc
@@ -0,0 +1,63 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+
+
+  const int Ls=8;
+
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);  RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);  RNG4.SeedFixedIntegers(seeds4);
+
+  LatticeFermion    src(FGrid); random(RNG5,src);
+  LatticeFermion result(FGrid); result=zero;
+  LatticeGaugeField Umu(UGrid); random(RNG4,Umu);
+
+  std::vector<LatticeColourMatrix> U(4,UGrid);
+
+  for(int mu=0;mu<Nd;mu++){
+    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+  }
+
+  ConjugateResidual<LatticeFermion> MCR(1.0e-8,10000);
+  
+  RealD mass=0.5;
+  RealD M5=1.8;
+  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+
+  MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermOp(Ddwf);
+  MCR(HermOp,src,result);
+
+  Gamma5HermitianLinearOperator<DomainWallFermion,LatticeFermion> g5HermOp(Ddwf);
+  MCR(g5HermOp,src,result);
+
+
+  Grid_finalize();
+}
diff --git a/tests/Test_nersc_io.cc b/tests/Test_nersc_io.cc
index 80d78291..403d2cab 100644
--- a/tests/Test_nersc_io.cc
+++ b/tests/Test_nersc_io.cc
@@ -83,7 +83,7 @@ int main (int argc, char ** argv)
   Complex l  = TensorRemove(Tl);
   std::cout << "calculated link trace " <<l*LinkTraceScale<<std::endl;
 
-  sumBlocks(cPlaq,Plaq);
+  blockSum(cPlaq,Plaq);
   TComplex TcP = sum(cPlaq);
   Complex ll= TensorRemove(TcP);
   std::cout << "coarsened plaquettes sum to " <<ll*PlaqScale<<std::endl;
diff --git a/tests/Test_wilson_cr_unprec.cc b/tests/Test_wilson_cr_unprec.cc
new file mode 100644
index 00000000..7fbf8c8f
--- /dev/null
+++ b/tests/Test_wilson_cr_unprec.cc
@@ -0,0 +1,59 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+  GridCartesian               Grid(latt_size,simd_layout,mpi_layout);
+  GridRedBlackCartesian     RBGrid(latt_size,simd_layout,mpi_layout);
+
+  std::vector<int> seeds({1,2,3,4});
+  GridParallelRNG          pRNG(&Grid);  pRNG.SeedFixedIntegers(seeds);
+
+  LatticeFermion src(&Grid); random(pRNG,src);
+  RealD nrm = norm2(src);
+  LatticeFermion result(&Grid); result=zero;
+  LatticeGaugeField Umu(&Grid); random(pRNG,Umu);
+
+  std::vector<LatticeColourMatrix> U(4,&Grid);
+
+  double volume=1;
+  for(int mu=0;mu<Nd;mu++){
+    volume=volume*latt_size[mu];
+  }  
+
+  for(int mu=0;mu<Nd;mu++){
+    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+  }
+  
+  RealD mass=0.5;
+  WilsonFermion Dw(Umu,Grid,RBGrid,mass);
+
+  MdagMLinearOperator<WilsonFermion,LatticeFermion> HermOp(Dw);
+
+  ConjugateResidual<LatticeFermion> MCR(1.0e-8,10000);
+
+  MCR(HermOp,src,result);
+
+  Grid_finalize();
+}

From 78607950ac8a4922d35af95c3c7e46c2e98c7aa6 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 8 Jun 2015 12:12:13 +0100
Subject: [PATCH 289/429] No compile fix after merge

---
 tests/Test_multishift_sqrt.cc | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/tests/Test_multishift_sqrt.cc b/tests/Test_multishift_sqrt.cc
index 8e6b08d4..1ecc1fde 100644
--- a/tests/Test_multishift_sqrt.cc
+++ b/tests/Test_multishift_sqrt.cc
@@ -20,12 +20,20 @@ public:
     sqrtscale = sqrtscale * adj(sqrtscale);// force real pos def
     scale = sqrtscale * sqrtscale; 
   }
+  // Support for coarsening to a multigrid
+  void OpDiag (const Field &in, Field &out) {};
+  void OpDir  (const Field &in, Field &out,int dir,int disp){};
+
   void Op     (const Field &in, Field &out){
     out = scale * in;
   }
   void AdjOp  (const Field &in, Field &out){
     out = scale * in;
   }
+  void HermOp(const Field &in, Field &out){
+    double n1, n2;
+    HermOpAndNorm(in,out,n1,n2);
+  }
   void HermOpAndNorm(const Field &in, Field &out,double &n1,double &n2){
     ComplexD dot;
 

From 6b8fe0405415eb798d3f509bf39e71db27f57b85 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Tue, 9 Jun 2015 15:46:21 +0900
Subject: [PATCH 290/429] Experimental support for ARM

---
 .gitignore                         |   2 +-
 INSTALL                            |   2 +-
 configure                          |  29 ++-
 configure.ac                       |  22 ++-
 lib/GridConfig.h                   |  14 +-
 lib/GridConfig.h.in                |  14 +-
 lib/Make.inc                       |   4 +-
 lib/algorithms/approx/.dirstamp    |   0
 lib/communicator/.dirstamp         |   0
 lib/qcd/spin/.dirstamp             |   0
 lib/qcd/utils/.dirstamp            |   0
 lib/simd/Grid_avx.h                |   8 +-
 lib/simd/Grid_avx512.h             |   8 +-
 lib/simd/Grid_empty.h              | 289 +++++++++++++++++++++++++++
 lib/simd/Grid_neon.h               | 308 +++++++++++++++++++++++++++++
 lib/simd/Grid_sse4.h               |   9 +-
 lib/simd/Grid_vector_types.h       |  10 +-
 lib/stencil/.dirstamp              |   0
 scripts/arm_configure.experimental |   1 +
 tests/Make.inc                     |  10 +-
 20 files changed, 693 insertions(+), 37 deletions(-)
 create mode 100644 lib/algorithms/approx/.dirstamp
 create mode 100644 lib/communicator/.dirstamp
 create mode 100644 lib/qcd/spin/.dirstamp
 create mode 100644 lib/qcd/utils/.dirstamp
 create mode 100644 lib/simd/Grid_empty.h
 create mode 100644 lib/simd/Grid_neon.h
 create mode 100644 lib/stencil/.dirstamp
 create mode 100644 scripts/arm_configure.experimental

diff --git a/.gitignore b/.gitignore
index 82e09bc0..be00eb30 100644
--- a/.gitignore
+++ b/.gitignore
@@ -49,7 +49,7 @@ config.status
 /stamp-h1
 /config.sub
 /config.guess
-
+/INSTALL
 
 # Packages #
 ############
diff --git a/INSTALL b/INSTALL
index 80a61507..f812f5a2 120000
--- a/INSTALL
+++ b/INSTALL
@@ -1 +1 @@
-/opt/local/share/automake-1.15/INSTALL
\ No newline at end of file
+/usr/share/automake-1.14/INSTALL
\ No newline at end of file
diff --git a/configure b/configure
index 36f85edb..702a52db 100755
--- a/configure
+++ b/configure
@@ -2574,7 +2574,7 @@ test -n "$target_alias" &&
     NONENONEs,x,x, &&
   program_prefix=${target_alias}-
 
-am__api_version='1.15'
+am__api_version='1.14'
 
 # Find a good install program.  We prefer a C program (faster),
 # so one script is as good as another.  But avoid the broken or
@@ -2746,8 +2746,8 @@ test "$program_suffix" != NONE &&
 ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
 program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
 
-# Expand $ac_aux_dir to an absolute path.
-am_aux_dir=`cd "$ac_aux_dir" && pwd`
+# expand $ac_aux_dir to an absolute path
+am_aux_dir=`cd $ac_aux_dir && pwd`
 
 if test x"${MISSING+set}" != xset; then
   case $am_aux_dir in
@@ -2766,7 +2766,7 @@ else
 $as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
 fi
 
-if test x"${install_sh+set}" != xset; then
+if test x"${install_sh}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -3094,8 +3094,8 @@ MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 mkdir_p='$(MKDIR_P)'
 
-# We need awk for the "check" target (and possibly the TAP driver).  The
-# system "awk" is bad on some platforms.
+# We need awk for the "check" target.  The system "awk" is bad on
+# some platforms.
 # Always define AMTAR for backward compatibility.  Yes, it's still used
 # in the wild :-(  We should find a proper way to deprecate it ...
 AMTAR='$${TAR-tar}'
@@ -3154,7 +3154,6 @@ END
 fi
 
 
-
 ac_config_headers="$ac_config_headers lib/GridConfig.h"
 
 # Check whether --enable-silent-rules was given.
@@ -6003,6 +6002,7 @@ $as_echo "$as_me: WARNING: Your processor supports fma instructions but not your
 
 
 
+
 # Checks for libraries.
 #AX_GCC_VAR_ATTRIBUTE(aligned)
 
@@ -6709,8 +6709,21 @@ $as_echo "#define AVX512 1" >>confdefs.h
 
        supported="cross compilation"
      ;;
+     NEONv7)
+       echo Configuring for experimental ARMv7 support
+
+$as_echo "#define NEONv7 1" >>confdefs.h
+
+       supported="cross compilation"
+     ;;
+     DEBUG)
+       echo Configuring without SIMD support - only for compiler DEBUGGING!
+
+$as_echo "#define EMPTY_SIMD 1" >>confdefs.h
+
+      ;;
      *)
-     as_fn_error $? "${ac_SIMD} unsupported --enable-simd option" "$LINENO" 5;
+     as_fn_error $? "${ac_SIMD} flag unsupported as --enable-simd option\nRun ./configure --help for the list of options" "$LINENO" 5;
      ;;
 esac
 
diff --git a/configure.ac b/configure.ac
index 59792fb0..cf0da9eb 100644
--- a/configure.ac
+++ b/configure.ac
@@ -3,7 +3,7 @@
 #
 # Project Grid package  
 # 
-# Time-stamp: <2015-05-27 18:51:47 neo>
+# Time-stamp: <2015-06-09 15:26:39 neo>
 
 AC_PREREQ([2.63])
 AC_INIT([Grid], [1.0], [paboyle@ph.ed.ac.uk])
@@ -29,6 +29,7 @@ AC_PROG_RANLIB
 AX_CXX_COMPILE_STDCXX_11(noext, mandatory)
 AX_EXT
 
+
 # Checks for libraries.
 #AX_GCC_VAR_ATTRIBUTE(aligned)
 
@@ -75,7 +76,7 @@ supported=no
 case ${ac_SIMD} in
      SSE4)
        echo Configuring for SSE4
-       AC_DEFINE([SSE4],[1],[SSE4] )
+       AC_DEFINE([SSE4],[1],[SSE4 Intrinsics] )
        if test x"$ax_cv_support_ssse3_ext" = x"yes"; then  dnl minimal support for SSE4
          supported=yes
        else
@@ -84,7 +85,7 @@ case ${ac_SIMD} in
      ;;
      AVX)
        echo Configuring for AVX
-       AC_DEFINE([AVX1],[1],[AVX] )
+       AC_DEFINE([AVX1],[1],[AVX Intrinsics] )
        if test x"$ax_cv_support_avx_ext" = x"yes"; then  dnl minimal support for AVX
        supported=yes			  
        else
@@ -93,7 +94,7 @@ case ${ac_SIMD} in
      ;;
      AVX2)
        echo Configuring for AVX2
-       AC_DEFINE([AVX2],[1],[AVX2] )
+       AC_DEFINE([AVX2],[1],[AVX2 Intrinsics] )
        if test x"$ax_cv_support_avx2_ext" = x"yes"; then  dnl minimal support for AVX2
        supported=yes
        else
@@ -102,11 +103,20 @@ case ${ac_SIMD} in
      ;;
      AVX512|MIC)
        echo Configuring for AVX512 and MIC
-       AC_DEFINE([AVX512],[1],[AVX512] )
+       AC_DEFINE([AVX512],[1],[AVX512 Intrinsics for Knights Corner] )
        supported="cross compilation"
      ;;
+     NEONv7)
+       echo Configuring for experimental ARMv7 support 
+       AC_DEFINE([NEONv7],[1],[NEON ARMv7 Experimental support ] )
+       supported="cross compilation"
+     ;;
+     DEBUG)
+       echo Configuring without SIMD support - only for compiler DEBUGGING!
+       AC_DEFINE([EMPTY_SIMD],[1],[EMPTY_SIMD only for DEBUGGING] )
+      ;;     
      *)
-     AC_MSG_ERROR([${ac_SIMD} unsupported --enable-simd option]); 
+     AC_MSG_ERROR([${ac_SIMD} flag unsupported as --enable-simd option\nRun ./configure --help for the list of options]); 
      ;;
 esac
 
diff --git a/lib/GridConfig.h b/lib/GridConfig.h
index e5b75382..c9dea1cb 100644
--- a/lib/GridConfig.h
+++ b/lib/GridConfig.h
@@ -1,15 +1,18 @@
 /* lib/GridConfig.h.  Generated from GridConfig.h.in by configure.  */
 /* lib/GridConfig.h.in.  Generated from configure.ac by autoheader.  */
 
-/* AVX */
+/* AVX Intrinsics */
 /* #undef AVX1 */
 
-/* AVX2 */
+/* AVX2 Intrinsics */
 /* #undef AVX2 */
 
-/* AVX512 */
+/* AVX512 Intrinsics for Knights Corner */
 /* #undef AVX512 */
 
+/* EMPTY_SIMD only for DEBUGGING */
+/* #undef EMPTY_SIMD */
+
 /* GRID_COMMS_MPI */
 /* #undef GRID_COMMS_MPI */
 
@@ -111,6 +114,9 @@
 /* Define to 1 if you have the <unistd.h> header file. */
 #define HAVE_UNISTD_H 1
 
+/* NEON ARMv7 Experimental support */
+/* #undef NEONv7 */
+
 /* Name of package */
 #define PACKAGE "grid"
 
@@ -132,7 +138,7 @@
 /* Define to the version of this package. */
 #define PACKAGE_VERSION "1.0"
 
-/* SSE4 */
+/* SSE4 Intrinsics */
 #define SSE4 1
 
 /* Define to 1 if you have the ANSI C header files. */
diff --git a/lib/GridConfig.h.in b/lib/GridConfig.h.in
index afa1d1e4..8ff77f3d 100644
--- a/lib/GridConfig.h.in
+++ b/lib/GridConfig.h.in
@@ -1,14 +1,17 @@
 /* lib/GridConfig.h.in.  Generated from configure.ac by autoheader.  */
 
-/* AVX */
+/* AVX Intrinsics */
 #undef AVX1
 
-/* AVX2 */
+/* AVX2 Intrinsics */
 #undef AVX2
 
-/* AVX512 */
+/* AVX512 Intrinsics for Knights Corner */
 #undef AVX512
 
+/* EMPTY_SIMD only for DEBUGGING */
+#undef EMPTY_SIMD
+
 /* GRID_COMMS_MPI */
 #undef GRID_COMMS_MPI
 
@@ -110,6 +113,9 @@
 /* Define to 1 if you have the <unistd.h> header file. */
 #undef HAVE_UNISTD_H
 
+/* NEON ARMv7 Experimental support */
+#undef NEONv7
+
 /* Name of package */
 #undef PACKAGE
 
@@ -131,7 +137,7 @@
 /* Define to the version of this package. */
 #undef PACKAGE_VERSION
 
-/* SSE4 */
+/* SSE4 Intrinsics */
 #undef SSE4
 
 /* Define to 1 if you have the ANSI C header files. */
diff --git a/lib/Make.inc b/lib/Make.inc
index c38606e2..edbcff7f 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Comparison.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./simd/Old/Grid_vComplexD.h ./simd/Old/Grid_vComplexF.h ./simd/Old/Grid_vInteger.h ./simd/Old/Grid_vRealD.h ./simd/Old/Grid_vRealF.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
+HFILES=./Cshift.h ./simd/Grid_avx.h ./simd/Grid_vector_types.h ./simd/Grid_sse4.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_vector_unops.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./Tensors.h ./Algorithms.h ./communicator/Communicator_base.h ./lattice/Lattice_rng.h ./lattice/Lattice_reduction.h ./lattice/Lattice_transfer.h ./lattice/Lattice_unary.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_comparison.h ./lattice/Lattice_overload.h ./lattice/Lattice_reality.h ./lattice/Lattice_local.h ./lattice/Lattice_conformable.h ./lattice/Lattice_where.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_ET.h ./lattice/Lattice_transpose.h ./lattice/Lattice_trace.h ./Stencil.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_poke.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_class.h ./tensors/Tensor_transpose.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_reality.h ./tensors/Tensor_trace.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_outer.h ./tensors/Tensor_inner.h ./tensors/Tensor_traits.h ./tensors/Tensor_Ta.h ./tensors/Tensor_unary.h ./tensors/Tensor_peek.h ./tensors/Tensor_arith.h ./tensors/Tensor_extract_merge.h ./Communicator.h ./Cartesian.h ./parallelIO/NerscIO.h ./qcd/QCD.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/LinalgUtils.h ./qcd/utils/CovariantCshift.h ./qcd/utils/WilsonLoops.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/spin/TwoSpinor.h ./qcd/spin/Dirac.h ./cshift/Cshift_common.h ./cshift/Cshift_none.h ./cshift/Cshift_mpi.h ./Simd.h ./GridConfig.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_red_black.h ./cartesian/Cartesian_full.h ./AlignedAllocator.h ./Lattice.h ./Threads.h ./Comparison.h ./Grid.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Zolotarev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Remez.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./algorithms/CoarsenedMatrix.h ./stencil/Lebesgue.h
 
-CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
+CCFILES=./qcd/utils/SpaceTimeGrid.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/spin/Dirac.cc ./GridInit.cc ./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/algorithms/approx/.dirstamp b/lib/algorithms/approx/.dirstamp
new file mode 100644
index 00000000..e69de29b
diff --git a/lib/communicator/.dirstamp b/lib/communicator/.dirstamp
new file mode 100644
index 00000000..e69de29b
diff --git a/lib/qcd/spin/.dirstamp b/lib/qcd/spin/.dirstamp
new file mode 100644
index 00000000..e69de29b
diff --git a/lib/qcd/utils/.dirstamp b/lib/qcd/utils/.dirstamp
new file mode 100644
index 00000000..e69de29b
diff --git a/lib/simd/Grid_avx.h b/lib/simd/Grid_avx.h
index ba023146..2dc38cca 100644
--- a/lib/simd/Grid_avx.h
+++ b/lib/simd/Grid_avx.h
@@ -4,7 +4,7 @@
 
   Using intrinsics
 */
-// Time-stamp: <2015-05-29 14:13:30 neo>
+// Time-stamp: <2015-06-09 14:26:59 neo>
 //----------------------------------------------------------------------
 
 #include <immintrin.h>
@@ -383,6 +383,12 @@ namespace Grid {
       _mm_prefetch(ptr+i+512,_MM_HINT_T0);
     }
   }
+  inline void prefetch_HINT_T0(const char *ptr){
+    _mm_prefetch(ptr,_MM_HINT_T0);
+  }
+
+
+
   template < typename VectorSIMD > 
     inline void Gpermute(VectorSIMD &y,const VectorSIMD &b, int perm ) {
     Optimization::permute(y.v,b.v,perm);
diff --git a/lib/simd/Grid_avx512.h b/lib/simd/Grid_avx512.h
index bb914270..10741174 100644
--- a/lib/simd/Grid_avx512.h
+++ b/lib/simd/Grid_avx512.h
@@ -4,7 +4,7 @@
 
   Using intrinsics
 */
-// Time-stamp: <2015-05-27 12:08:50 neo>
+// Time-stamp: <2015-06-09 14:27:28 neo>
 //----------------------------------------------------------------------
 
 #include <immintrin.h>
@@ -309,6 +309,12 @@ namespace Grid {
       _mm_prefetch(ptr+i+512,_MM_HINT_T0);
     }
   }
+  inline void prefetch_HINT_T0(const char *ptr){
+    _mm_prefetch(ptr,_MM_HINT_T0);
+  }
+
+
+
 
   // Gpermute utilities consider coalescing into 1 Gpermute
   template < typename VectorSIMD > 
diff --git a/lib/simd/Grid_empty.h b/lib/simd/Grid_empty.h
new file mode 100644
index 00000000..1c088d87
--- /dev/null
+++ b/lib/simd/Grid_empty.h
@@ -0,0 +1,289 @@
+//----------------------------------------------------------------------
+/*! @file Grid_sse4.h
+  @brief Empty Optimization libraries for debugging
+
+  Using intrinsics
+*/
+// Time-stamp: <2015-06-09 14:28:02 neo>
+//----------------------------------------------------------------------
+
+namespace Optimization {
+
+  template<class vtype>
+  union uconv {
+    float f;
+    vtype v;
+  };
+
+  union u128f {
+    float v;
+    float f[4];
+  };
+  union u128d {
+    double v;
+    double f[2];
+  };
+  
+  struct Vsplat{
+    //Complex float
+    inline float operator()(float a, float b){
+      return 0;
+    }
+    // Real float
+    inline float operator()(float a){
+      return 0;
+    }
+    //Complex double
+    inline double operator()(double a, double b){
+      return 0;
+    }
+    //Real double
+    inline double operator()(double a){
+      return 0;
+    }
+    //Integer
+    inline int operator()(Integer a){
+      return 0;
+    }
+  };
+
+  struct Vstore{
+    //Float 
+    inline void operator()(float a, float* F){
+      
+    }
+    //Double
+    inline void operator()(double a, double* D){
+     
+    }
+    //Integer
+    inline void operator()(int a, Integer* I){
+      
+    }
+
+  };
+
+  struct Vstream{
+    //Float
+    inline void operator()(float * a, float b){
+     
+    }
+    //Double
+    inline void operator()(double * a, double b){
+     
+    }
+
+
+  };
+
+  struct Vset{
+    // Complex float 
+    inline float operator()(Grid::ComplexF *a){
+      return 0;
+    }
+    // Complex double 
+    inline double operator()(Grid::ComplexD *a){
+      return 0;
+    }
+    // Real float 
+    inline float operator()(float *a){
+      return  0;
+    }
+    // Real double
+    inline double operator()(double *a){
+      return 0;
+    }
+    // Integer
+    inline int operator()(Integer *a){
+      return 0;
+    }
+
+
+  };
+
+  template <typename Out_type, typename In_type>
+  struct Reduce{
+    //Need templated class to overload output type
+    //General form must generate error if compiled
+    inline Out_type operator()(In_type in){
+      printf("Error, using wrong Reduce function\n");
+      exit(1);
+      return 0;
+    }
+  };
+
+  /////////////////////////////////////////////////////
+  // Arithmetic operations
+  /////////////////////////////////////////////////////
+  struct Sum{
+    //Complex/Real float
+    inline float operator()(float a, float b){
+      return 0;
+    }
+    //Complex/Real double
+    inline double operator()(double a, double b){
+      return 0;
+    }
+    //Integer
+    inline int operator()(int a, int b){
+      return 0;
+    }
+  };
+
+  struct Sub{
+    //Complex/Real float
+    inline float operator()(float a, float b){
+      return 0;
+    }
+    //Complex/Real double
+    inline double operator()(double a, double b){
+      return 0;
+    }
+    //Integer
+    inline int operator()(int a, int b){
+      return 0;
+    }
+  };
+
+  struct MultComplex{
+    // Complex float
+    inline float operator()(float a, float b){
+      return 0;
+    }
+    // Complex double
+    inline double operator()(double a, double b){
+      return 0;
+    }
+  };
+
+  struct Mult{
+    // Real float
+    inline float operator()(float a, float b){
+      return 0;
+    }
+    // Real double
+    inline double operator()(double a, double b){
+      return 0;
+    }
+    // Integer
+    inline int operator()(int a, int b){
+      return 0;
+    }
+  };
+
+  struct Conj{
+    // Complex single
+    inline float operator()(float in){
+      return 0;
+    }
+    // Complex double
+    inline double operator()(double in){
+      return 0;
+    }
+    // do not define for integer input
+  };
+
+  struct TimesMinusI{
+    //Complex single
+    inline float operator()(float in, float ret){
+      return 0;
+    }
+    //Complex double
+    inline double operator()(double in, double ret){
+      return 0;
+    }
+
+
+  };
+
+  struct TimesI{
+    //Complex single
+    inline float operator()(float in, float ret){
+      return 0;
+    }
+    //Complex double
+    inline double operator()(double in, double ret){
+      return 0;
+    }
+  };
+
+  //////////////////////////////////////////////
+  // Some Template specialization
+  template < typename vtype > 
+    void permute(vtype &a, vtype b, int perm) {
+   }; 
+
+  //Complex float Reduce
+  template<>
+  inline Grid::ComplexF Reduce<Grid::ComplexF, float>::operator()(float in){
+    return 0;
+  }
+  //Real float Reduce
+  template<>
+  inline Grid::RealF Reduce<Grid::RealF, float>::operator()(float in){
+    return 0;
+  }
+  
+  
+  //Complex double Reduce
+  template<>
+  inline Grid::ComplexD Reduce<Grid::ComplexD, double>::operator()(double in){
+    return 0;
+  }
+  
+  //Real double Reduce
+  template<>
+  inline Grid::RealD Reduce<Grid::RealD, double>::operator()(double in){
+    return 0;
+  }
+
+  //Integer Reduce
+  template<>
+  inline Integer Reduce<Integer, int>::operator()(int in){
+    // FIXME unimplemented
+   printf("Reduce : Missing integer implementation -> FIX\n");
+    assert(0);
+  }
+}
+
+//////////////////////////////////////////////////////////////////////////////////////
+// Here assign types 
+namespace Grid {
+
+  typedef float SIMD_Ftype;  // Single precision type
+  typedef double SIMD_Dtype; // Double precision type
+  typedef int SIMD_Itype; // Integer type
+
+  // prefetch utilities
+  inline void v_prefetch0(int size, const char *ptr){};
+  inline void prefetch_HINT_T0(const char *ptr){};
+
+
+
+  // Gpermute function
+  template < typename VectorSIMD > 
+    inline void Gpermute(VectorSIMD &y,const VectorSIMD &b, int perm ) {
+    Optimization::permute(y.v,b.v,perm);
+  }
+
+
+  // Function name aliases
+  typedef Optimization::Vsplat   VsplatSIMD;
+  typedef Optimization::Vstore   VstoreSIMD;
+  typedef Optimization::Vset     VsetSIMD;
+  typedef Optimization::Vstream  VstreamSIMD;
+  template <typename S, typename T> using ReduceSIMD = Optimization::Reduce<S,T>;
+
+ 
+
+
+  // Arithmetic operations
+  typedef Optimization::Sum         SumSIMD;
+  typedef Optimization::Sub         SubSIMD;
+  typedef Optimization::Mult        MultSIMD;
+  typedef Optimization::MultComplex MultComplexSIMD;
+  typedef Optimization::Conj        ConjSIMD;
+  typedef Optimization::TimesMinusI TimesMinusISIMD;
+  typedef Optimization::TimesI      TimesISIMD;
+
+}
diff --git a/lib/simd/Grid_neon.h b/lib/simd/Grid_neon.h
new file mode 100644
index 00000000..73660b40
--- /dev/null
+++ b/lib/simd/Grid_neon.h
@@ -0,0 +1,308 @@
+//----------------------------------------------------------------------
+/*! @file Grid_sse4.h
+  @brief Optimization libraries for NEON (ARM) instructions set ARMv7
+
+  Experimental - Using intrinsics - DEVELOPING! 
+*/
+// Time-stamp: <2015-06-09 15:25:40 neo>
+//----------------------------------------------------------------------
+
+#include <arm_neon.h>
+
+namespace Optimization {
+
+  template<class vtype>
+  union uconv {
+    float32x4_t f;
+    vtype v;
+  };
+
+  union u128f {
+    float32x4_t v;
+    float f[4];
+  };
+  union u128d {
+    float32x4_t v;
+    float f[4];
+  };
+  
+  struct Vsplat{
+    //Complex float
+    inline float32x4_t operator()(float a, float b){
+      float32x4_t foo;
+      return foo;
+    }
+    // Real float
+    inline float32x4_t operator()(float a){
+      float32x4_t foo;
+      return foo;
+    }
+    //Complex double
+    inline float32x4_t operator()(double a, double b){
+      float32x4_t foo;
+      return foo;
+    }
+    //Real double
+    inline float32x4_t operator()(double a){
+      float32x4_t foo;
+      return foo;
+    }
+    //Integer
+    inline uint32x4_t operator()(Integer a){
+      uint32x4_t foo;
+      return foo;
+    }
+  };
+
+  struct Vstore{
+    //Float 
+    inline void operator()(float32x4_t a, float* F){
+      
+    }
+    //Double
+    inline void operator()(float32x4_t a, double* D){
+      
+    }
+    //Integer
+    inline void operator()(uint32x4_t a, Integer* I){
+     
+    }
+
+  };
+
+  struct Vstream{
+    //Float
+    inline void operator()(float * a, float32x4_t b){
+    
+    }
+    //Double
+    inline void operator()(double * a, float32x4_t b){
+  
+    }
+
+
+  };
+
+  struct Vset{
+    // Complex float 
+    inline float32x4_t operator()(Grid::ComplexF *a){
+      float32x4_t foo;
+      return foo;
+    }
+    // Complex double 
+    inline float32x4_t operator()(Grid::ComplexD *a){
+      float32x4_t foo;
+      return foo;
+    }
+    // Real float 
+    inline float32x4_t operator()(float *a){
+      float32x4_t foo;
+      return foo;
+    }
+    // Real double
+    inline float32x4_t operator()(double *a){
+      float32x4_t foo;
+      return foo;
+    }
+    // Integer
+    inline uint32x4_t operator()(Integer *a){
+      uint32x4_t foo;
+      return foo;
+    }
+
+
+  };
+
+  template <typename Out_type, typename In_type>
+  struct Reduce{
+    //Need templated class to overload output type
+    //General form must generate error if compiled
+    inline Out_type operator()(In_type in){
+      printf("Error, using wrong Reduce function\n");
+      exit(1);
+      return 0;
+    }
+  };
+
+  /////////////////////////////////////////////////////
+  // Arithmetic operations
+  /////////////////////////////////////////////////////
+  struct Sum{
+    //Complex/Real float
+    inline float32x4_t operator()(float32x4_t a, float32x4_t b){
+      float32x4_t foo;
+      return foo;
+    }
+    //Complex/Real double
+    //inline float32x4_t operator()(float32x4_t a, float32x4_t b){
+    //  float32x4_t foo;
+    //  return foo;
+    //}
+    //Integer
+    inline uint32x4_t operator()(uint32x4_t a, uint32x4_t b){
+      uint32x4_t foo;
+      return foo;
+    }
+  };
+
+  struct Sub{
+    //Complex/Real float
+    inline float32x4_t operator()(float32x4_t a, float32x4_t b){
+      float32x4_t foo;
+      return foo;
+    }
+    //Complex/Real double
+    //inline float32x4_t operator()(float32x4_t a, float32x4_t b){
+    //  float32x4_t foo;
+    //  return foo;
+    //}
+    //Integer
+    inline uint32x4_t operator()(uint32x4_t a, uint32x4_t b){
+      uint32x4_t foo;
+      return foo;
+    }
+  };
+
+  struct MultComplex{
+    // Complex float
+    inline float32x4_t operator()(float32x4_t a, float32x4_t b){
+      float32x4_t foo;
+      return foo;
+    }
+    // Complex double
+    //inline float32x4_t operator()(float32x4_t a, float32x4_t b){
+    //  float32x4_t foo;
+    //  return foo;
+    //}
+  };
+
+  struct Mult{
+    // Real float
+    inline float32x4_t operator()(float32x4_t a, float32x4_t b){
+      return a;
+    }
+    // Real double
+    //inline float32x4_t operator()(float32x4_t a, float32x4_t b){
+    //  return 0;
+    //}
+    // Integer
+    inline uint32x4_t operator()(uint32x4_t a, uint32x4_t b){
+      return a;
+    }
+  };
+
+  struct Conj{
+    // Complex single
+    inline float32x4_t operator()(float32x4_t in){
+      return in;
+    }
+    // Complex double
+    //inline float32x4_t operator()(float32x4_t in){
+    // return 0;
+    //}
+    // do not define for integer input
+  };
+
+  struct TimesMinusI{
+    //Complex single
+    inline float32x4_t operator()(float32x4_t in, float32x4_t ret){
+      return in;
+    }
+    //Complex double
+    //inline float32x4_t operator()(float32x4_t in, float32x4_t ret){
+    //  return in;
+    //}
+
+
+  };
+
+  struct TimesI{
+    //Complex single
+    inline float32x4_t operator()(float32x4_t in, float32x4_t ret){
+      return in;
+    }
+    //Complex double
+    //inline float32x4_t operator()(float32x4_t in, float32x4_t ret){
+    //  return 0;
+    //}
+  };
+
+  //////////////////////////////////////////////
+  // Some Template specialization
+  template < typename vtype > 
+    void permute(vtype &a, vtype b, int perm) {
+
+  }; 
+
+  //Complex float Reduce
+  template<>
+  inline Grid::ComplexF Reduce<Grid::ComplexF, float32x4_t>::operator()(float32x4_t in){
+    return 0;
+  }
+  //Real float Reduce
+  template<>
+  inline Grid::RealF Reduce<Grid::RealF, float32x4_t>::operator()(float32x4_t in){
+    return 0;
+  }
+  
+  
+  //Complex double Reduce
+  template<>
+  inline Grid::ComplexD Reduce<Grid::ComplexD, float32x4_t>::operator()(float32x4_t in){
+    return 0;
+  }
+  
+  //Real double Reduce
+  template<>
+  inline Grid::RealD Reduce<Grid::RealD, float32x4_t>::operator()(float32x4_t in){
+    return 0;
+  }
+
+  //Integer Reduce
+  template<>
+  inline Integer Reduce<Integer, uint32x4_t>::operator()(uint32x4_t in){
+    // FIXME unimplemented
+   printf("Reduce : Missing integer implementation -> FIX\n");
+    assert(0);
+  }
+}
+
+//////////////////////////////////////////////////////////////////////////////////////
+// Here assign types 
+namespace Grid {
+
+  typedef float32x4_t  SIMD_Ftype; // Single precision type
+  typedef float32x4_t  SIMD_Dtype; // Double precision type - no double on ARMv7
+  typedef uint32x4_t   SIMD_Itype; // Integer type
+
+  inline void v_prefetch0(int size, const char *ptr){};  // prefetch utilities
+  inline void prefetch_HINT_T0(const char *ptr){};
+
+
+  // Gpermute function
+  template < typename VectorSIMD > 
+    inline void Gpermute(VectorSIMD &y,const VectorSIMD &b, int perm ) {
+    Optimization::permute(y.v,b.v,perm);
+  }
+
+
+  // Function name aliases
+  typedef Optimization::Vsplat   VsplatSIMD;
+  typedef Optimization::Vstore   VstoreSIMD;
+  typedef Optimization::Vset     VsetSIMD;
+  typedef Optimization::Vstream  VstreamSIMD;
+  template <typename S, typename T> using ReduceSIMD = Optimization::Reduce<S,T>;
+
+ 
+
+
+  // Arithmetic operations
+  typedef Optimization::Sum         SumSIMD;
+  typedef Optimization::Sub         SubSIMD;
+  typedef Optimization::Mult        MultSIMD;
+  typedef Optimization::MultComplex MultComplexSIMD;
+  typedef Optimization::Conj        ConjSIMD;
+  typedef Optimization::TimesMinusI TimesMinusISIMD;
+  typedef Optimization::TimesI      TimesISIMD;
+
+}
diff --git a/lib/simd/Grid_sse4.h b/lib/simd/Grid_sse4.h
index 0da888f7..3a54c0b3 100644
--- a/lib/simd/Grid_sse4.h
+++ b/lib/simd/Grid_sse4.h
@@ -4,7 +4,7 @@
 
   Using intrinsics
 */
-// Time-stamp: <2015-05-27 12:02:07 neo>
+// Time-stamp: <2015-06-09 14:24:01 neo>
 //----------------------------------------------------------------------
 
 #include <pmmintrin.h>
@@ -297,7 +297,12 @@ namespace Grid {
   typedef __m128d SIMD_Dtype; // Double precision type
   typedef __m128i SIMD_Itype; // Integer type
 
-  inline void v_prefetch0(int size, const char *ptr){};  // prefetch utilities
+  // prefetch utilities
+  inline void v_prefetch0(int size, const char *ptr){};
+  inline void prefetch_HINT_T0(const char *ptr){
+    _mm_prefetch(ptr,_MM_HINT_T0);
+  }
+  
 
   // Gpermute function
   template < typename VectorSIMD > 
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index 9525e45e..c670f531 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -2,11 +2,14 @@
 /*! @file Grid_vector_types.h
   @brief Defines templated class Grid_simd to deal with inner vector types
 */
-// Time-stamp: <2015-05-29 14:19:48 neo>
+// Time-stamp: <2015-06-09 15:00:47 neo>
 //---------------------------------------------------------------------------
 #ifndef GRID_VECTOR_TYPES
 #define GRID_VECTOR_TYPES
 
+#ifdef EMPTY_SIMD
+#include "Grid_empty.h"
+#endif
 #ifdef SSE4
 #include "Grid_sse4.h"
 #endif
@@ -19,6 +22,9 @@
 #if defined QPX
 #include "Grid_qpx.h"
 #endif
+#ifdef NEONv7
+#include "Grid_neon.h"
+#endif
 
 namespace Grid {
 
@@ -152,7 +158,7 @@ namespace Grid {
     ///////////////////////
     friend inline void vprefetch(const Grid_simd &v)
     {
-      _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
+      prefetch_HINT_T0((const char*)&v.v);
     }
 
     ///////////////////////
diff --git a/lib/stencil/.dirstamp b/lib/stencil/.dirstamp
new file mode 100644
index 00000000..e69de29b
diff --git a/scripts/arm_configure.experimental b/scripts/arm_configure.experimental
new file mode 100644
index 00000000..fbad84a5
--- /dev/null
+++ b/scripts/arm_configure.experimental
@@ -0,0 +1 @@
+./configure --host=arm-linux-gnueabihf  CXX=clang++-3.5 CXXFLAGS='-std=c++11 -O3 -target arm-linux-gnueabihf -I/usr/arm-linux-gnueabihf/include/ -I/home/neo/Codes/gmp6.0/gmp-arm/include/ -I/usr/arm-linux-gnueabihf/include/c++/4.8.2/arm-linux-gnueabihf/ -L/home/neo/Codes/gmp6.0/gmp-arm/lib/ -I/home/neo/Codes/mpfr3.1.2/mpfr-arm/include/ -L/home/neo/Codes/mpfr3.1.2/mpfr-arm/lib/ -static -mcpu=cortex-a7' --enable-simd=NEONv7
diff --git a/tests/Make.inc b/tests/Make.inc
index 2b132fd0..1f5573ca 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,9 +1,5 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_gamma Test_main Test_multishift_sqrt Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
-
-
-Test_GaugeAction_SOURCES=Test_GaugeAction.cc
-Test_GaugeAction_LDADD=-lGrid
+bin_PROGRAMS = Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_gamma Test_GaugeAction Test_main Test_multishift_sqrt Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
 
 
 Test_cayley_cg_SOURCES=Test_cayley_cg.cc
@@ -66,6 +62,10 @@ Test_gamma_SOURCES=Test_gamma.cc
 Test_gamma_LDADD=-lGrid
 
 
+Test_GaugeAction_SOURCES=Test_GaugeAction.cc
+Test_GaugeAction_LDADD=-lGrid
+
+
 Test_main_SOURCES=Test_main.cc
 Test_main_LDADD=-lGrid
 

From 9269126fbafd37625bcc0ec3590e8abeeb6da406 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 9 Jun 2015 10:25:29 +0100
Subject: [PATCH 291/429] Unary ops and coarse grid support

---
 lib/tensors/Tensor_traits.h | 1 -
 1 file changed, 1 deletion(-)

diff --git a/lib/tensors/Tensor_traits.h b/lib/tensors/Tensor_traits.h
index c47dca10..b5715c93 100644
--- a/lib/tensors/Tensor_traits.h
+++ b/lib/tensors/Tensor_traits.h
@@ -119,7 +119,6 @@ namespace Grid {
     static const bool value = true;
     static const bool notvalue = false;
   };
-
   template<> struct isGridTensor<int > {
     static const bool value = false;
     static const bool notvalue = true;

From 506dfd1517caf9fc753c299b15567e4120fd17ca Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 9 Jun 2015 10:26:19 +0100
Subject: [PATCH 292/429] Some unary ops and coarse grid support

---
 lib/Grid.h                                    |   1 -
 lib/Make.inc                                  |   2 +-
 lib/algorithms/CoarsenedMatrix.h              | 134 +++++++++++++++---
 lib/lattice/Lattice_base.h                    |   4 +-
 .../Lattice_comparison_utils.h}               |   2 -
 lib/lattice/Lattice_coordinate.h              |  14 +-
 lib/lattice/Lattice_transfer.h                |  23 ++-
 lib/lattice/Lattice_unary.h                   |  41 ++++++
 lib/qcd/action/fermion/WilsonFermion5D.cc     |   4 +
 lib/simd/Grid_vector_unops.h                  |  12 ++
 lib/tensors/Tensor_class.h                    |   9 +-
 lib/tensors/Tensor_unary.h                    |  29 ++++
 scripts/linecount                             |  11 +-
 tests/Test_cayley_coarsen_support.cc          |  19 ++-
 tests/Test_cshift.cc                          |   1 -
 15 files changed, 258 insertions(+), 48 deletions(-)
 rename lib/{Comparison.h => lattice/Lattice_comparison_utils.h} (98%)

diff --git a/lib/Grid.h b/lib/Grid.h
index 16530434..aec0e9ac 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -55,7 +55,6 @@
 #include <Cartesian.h> // subdir aggregate
 #include <Tensors.h>   // subdir aggregate
 #include <Lattice.h>   // subdir aggregate
-#include <Comparison.h>
 #include <Cshift.h>    // subdir aggregate
 #include <Stencil.h>   // subdir aggregate
 #include <Algorithms.h>// subdir aggregate
diff --git a/lib/Make.inc b/lib/Make.inc
index c38606e2..7c7e02c8 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Comparison.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./simd/Old/Grid_vComplexD.h ./simd/Old/Grid_vComplexF.h ./simd/Old/Grid_vInteger.h ./simd/Old/Grid_vRealD.h ./simd/Old/Grid_vRealF.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./simd/Old/Grid_vComplexD.h ./simd/Old/Grid_vComplexF.h ./simd/Old/Grid_vInteger.h ./simd/Old/Grid_vRealD.h ./simd/Old/Grid_vRealF.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
 
 CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/algorithms/CoarsenedMatrix.h b/lib/algorithms/CoarsenedMatrix.h
index 54ea279e..006da07f 100644
--- a/lib/algorithms/CoarsenedMatrix.h
+++ b/lib/algorithms/CoarsenedMatrix.h
@@ -6,12 +6,42 @@
 namespace Grid {
 
   class Geometry {
+    //    int dimension;
   public:
     int npoint;
-    int dimension;
     std::vector<int> directions   ;
     std::vector<int> displacements;
+    std::vector<int> opdirs;
 
+    // FIXME -- don't like xposing the operator directions
+    // as different to the geometrical dirs
+    // Also don't like special casing five dim.. should pass an object in template
+  Geometry(int _d)  {
+  
+      int base = (_d==5) ? 1:0;
+
+      // make coarse grid stencil for 4d , not 5d
+      if ( _d==5 ) _d=4;
+
+      npoint = 2*_d+1;
+      directions.resize(npoint);
+      displacements.resize(npoint);
+      opdirs.resize(npoint);
+      for(int d=0;d<_d;d++){
+	directions[2*d  ] = d+base;
+	directions[2*d+1] = d+base;
+	opdirs[2*d  ] = d;
+	opdirs[2*d+1] = d;
+	displacements[2*d  ] = +1;
+	displacements[2*d+1] = -1;
+      }
+      directions   [2*_d]=0;
+      displacements[2*_d]=0;
+      opdirs       [2*_d]=0;
+    }
+  
+    /*
+      // Original cleaner code
     Geometry(int _d) : dimension(_d), npoint(2*_d+1), directions(npoint), displacements(npoint) {
       for(int d=0;d<dimension;d++){
 	directions[2*d  ] = d;
@@ -22,12 +52,12 @@ namespace Grid {
       directions   [2*dimension]=0;
       displacements[2*dimension]=0;
     }
-    
     std::vector<int> GetDelta(int point) {
       std::vector<int> delta(dimension,0);
       delta[directions[point]] = displacements[point];
       return delta;
     };
+    */    
 
   };
   
@@ -50,7 +80,10 @@ namespace Grid {
     Geometry         geom;
     GridBase *       _grid; 
     CartesianStencil Stencil; 
+
     std::vector<CoarseMatrix> A;
+    std::vector<CoarseMatrix> Aslow;
+
     std::vector<siteVector,alignedAllocator<siteVector> >   comm_buf;
       
     ///////////////////////
@@ -63,7 +96,7 @@ namespace Grid {
       SimpleCompressor<siteVector> compressor;
       Stencil.HaloExchange(in,comm_buf,compressor);
 
-      //PARALLEL_FOR_LOOP
+PARALLEL_FOR_LOOP
       for(int ss=0;ss<Grid()->oSites();ss++){
         siteVector res = zero;
 	siteVector tmp;
@@ -101,43 +134,112 @@ namespace Grid {
       _grid(&CoarseGrid),
       geom(CoarseGrid._ndimension),
       Stencil(&CoarseGrid,geom.npoint,Even,geom.directions,geom.displacements),
-      A(geom.npoint,&CoarseGrid)
+	A(geom.npoint,&CoarseGrid),
+	Aslow(geom.npoint,&CoarseGrid)
     {
       comm_buf.resize(Stencil._unified_buffer_size);
     };
 
     void CoarsenOperator(GridBase *FineGrid,LinearOperatorBase<Lattice<Fobj> > &linop,std::vector<Lattice<Fobj> > & subspace){
 
-      FineField  phi(FineGrid);
-      FineField Mphi(FineGrid);
-      CoarseVector Proj(Grid()); 
+      FineField iblock(FineGrid); // contributions from within this block
+      FineField oblock(FineGrid); // contributions from outwith this block
+
+      FineField     phi(FineGrid);
+      FineField     tmp(FineGrid);
+      FineField     zz(FineGrid); zz=zero;
+      FineField    Mphi(FineGrid);
+
+      Lattice<iScalar<vInteger> > coor(FineGrid);
+
+      CoarseVector iProj(Grid()); 
+      CoarseVector oProj(Grid()); 
       CoarseScalar InnerProd(Grid()); 
 
       // Orthogonalise the subblocks over the basis
       blockOrthogonalise(InnerProd,subspace);
+      blockProject(iProj,subspace[0],subspace);
 
       // Compute the matrix elements of linop between this orthonormal
       // set of vectors.
+      int self_stencil=-1;
+      for(int p=0;p<geom.npoint;p++){ 
+	A[p]=zero;
+	if( geom.displacements[p]==0){
+	  self_stencil=p;
+	}
+      }
+      assert(self_stencil!=-1);
+
       for(int i=0;i<nbasis;i++){
 	phi=subspace[i];
 	for(int p=0;p<geom.npoint;p++){ 
-	  int dir = geom.directions[p];
+
+	  int dir   = geom.directions[p];
+	  int opdir = geom.opdirs[p];
 	  int disp= geom.displacements[p];
 
-	  if ( disp==0 )linop.OpDiag(phi,Mphi);
-	  else linop.OpDir(phi,Mphi,dir,disp); 
+	  int block=(FineGrid->_rdimensions[dir])/(Grid()->_rdimensions[dir]);
 
-	  blockProject(Proj,Mphi,subspace);
+	  LatticeCoordinate(coor,dir);
 
-	  for(int ss=0;ss<Grid()->oSites();ss++){
-	    for(int j=0;j<nbasis;j++){
-	      A[p]._odata[ss](j,i) = Proj._odata[ss](j);
-	    }
+	  if ( disp==0 ){
+	    linop.OpDiag(phi,Mphi);
+	  }
+	  else  {
+	    linop.OpDir(phi,Mphi,opdir,disp); 
 	  }
 
-	}	
+	  ////////////////////////////////////////////////////////////////////////
+	  // Pick out contributions coming from this cell and neighbour cell
+	  ////////////////////////////////////////////////////////////////////////
+	  if ( disp==0 ) {
+	    iblock = Mphi;
+	    oblock = zero;
+	  } else if ( disp==1 ) {
+	    oblock = where(mod(coor,block)==(block-1),Mphi,zz);
+	    iblock = where(mod(coor,block)!=(block-1),Mphi,zz);
+	  } else if ( disp==-1 ) {
+	    oblock = where(mod(coor,block)==0,Mphi,zz);
+	    iblock = where(mod(coor,block)!=0,Mphi,zz);
+	  } else {
+	    assert(0);
+	  }
+
+	  blockProject(iProj,iblock,subspace);
+	  blockProject(oProj,oblock,subspace);
+	  for(int ss=0;ss<Grid()->oSites();ss++){
+	    for(int j=0;j<nbasis;j++){
+	      if( disp!= 0 ) {
+		A[p]._odata[ss](j,i) = oProj._odata[ss](j);
+	      }
+	      A[self_stencil]._odata[ss](j,i) =	A[self_stencil]._odata[ss](j,i) + iProj._odata[ss](j);
+	    }
+	  }
+	}
       }
+
+#if 0
+      ///////////////////////////
+      // test code worth preserving in if block
+      ///////////////////////////
+      std::cout<< " Computed matrix elements "<< self_stencil <<std::endl;
+      for(int p=0;p<geom.npoint;p++){
+	std::cout<< "A["<<p<<"]" << std::endl;
+	std::cout<< A[p] << std::endl;
+      }
+      std::cout<< " picking by block0 "<< self_stencil <<std::endl;
+
+      phi=subspace[0];
+      std::vector<int> bc(FineGrid->_ndimension,0);
+
+      blockPick(Grid(),phi,tmp,bc);      // Pick out a block
+      linop.Op(tmp,Mphi);                // Apply big dop
+      blockProject(iProj,Mphi,subspace); // project it and print it
+      std::cout<< " Computed matrix elements from block zero only "<<std::endl;
+      std::cout<< iProj <<std::endl;
       std::cout<<"Computed Coarse Operator"<<std::endl;
+#endif
     }
     
   };
diff --git a/lib/lattice/Lattice_base.h b/lib/lattice/Lattice_base.h
index ce2990a5..5a17826b 100644
--- a/lib/lattice/Lattice_base.h
+++ b/lib/lattice/Lattice_base.h
@@ -297,11 +297,13 @@ PARALLEL_FOR_LOOP
 #include <lattice/Lattice_reduction.h>
 #include <lattice/Lattice_peekpoke.h>
 #include <lattice/Lattice_reality.h>
+#include <lattice/Lattice_comparison_utils.h>
+#include <lattice/Lattice_comparison.h>
 #include <lattice/Lattice_coordinate.h>
+#include <lattice/Lattice_where.h>
 #include <lattice/Lattice_rng.h>
 #include <lattice/Lattice_unary.h>
 #include <lattice/Lattice_transfer.h>
 
 
-
 #endif
diff --git a/lib/Comparison.h b/lib/lattice/Lattice_comparison_utils.h
similarity index 98%
rename from lib/Comparison.h
rename to lib/lattice/Lattice_comparison_utils.h
index ecd6ece0..d5e3b3e5 100644
--- a/lib/Comparison.h
+++ b/lib/lattice/Lattice_comparison_utils.h
@@ -141,7 +141,5 @@ namespace Grid {
     }
 }
 
-#include <lattice/Lattice_comparison.h>
-#include <lattice/Lattice_where.h>
 
 #endif
diff --git a/lib/lattice/Lattice_coordinate.h b/lib/lattice/Lattice_coordinate.h
index c49aed89..a83dbe49 100644
--- a/lib/lattice/Lattice_coordinate.h
+++ b/lib/lattice/Lattice_coordinate.h
@@ -3,20 +3,8 @@
 
 namespace Grid {
 
-  /*
-  depbase=`echo Grid_main.o | sed 's|[^/]*$|.deps/&|;s|\.o$||'`;\
-        icpc -DHAVE_CONFIG_H -I. -I../lib    -I../lib -mmic -O3 -std=c++11 -fopenmp -MT Grid_main.o -MD -MP -MF $depbase.Tpo -c -o Grid_main.o Grid_main.cc &&\
-        mv -f $depbase.Tpo $depbase.Po
-	  ../lib/lattice/Grid_lattice_coordinate.h(25): error: no suitable user-defined conversion from "vector_type" to "const Grid::iScalar<Grid::iScalar<Grid::iScalar<Grid::vInteger>>>" exists
-    l._odata[o]=vI;
-                    ^
-          detected during instantiation of "void Grid::LatticeCoordinate(Grid::Lattice<iobj> &, int) [with iobj=Grid::QCD::vTInteger]" at line 283 of "Grid_main.cc"
-
-	    compilation aborted for Grid_main.cc (code 2)
-*/
     template<class iobj> inline void LatticeCoordinate(Lattice<iobj> &l,int mu)
     {
-      typedef typename iobj::scalar_object scalar_object;
       typedef typename iobj::scalar_type scalar_type;
       typedef typename iobj::vector_type vector_type;
 
@@ -33,7 +21,7 @@ namespace Grid {
 	  mergebuf[i]=(Integer)gcoor[mu];
 	}
 	merge<vector_type,scalar_type>(vI,mergebuf);
-	l._odata[o]._internal._internal._internal=vI;
+	l._odata[o]=vI;
       }
     };
 
diff --git a/lib/lattice/Lattice_transfer.h b/lib/lattice/Lattice_transfer.h
index 871ed893..12993680 100644
--- a/lib/lattice/Lattice_transfer.h
+++ b/lib/lattice/Lattice_transfer.h
@@ -166,9 +166,10 @@ template<class vobj,class CComplex>
 inline void blockNormalise(Lattice<CComplex> &ip,Lattice<vobj> &fineX)
 {
   GridBase *coarse = ip._grid;
+  Lattice<vobj> zz(fineX._grid); zz=zero;
   blockInnerProduct(ip,fineX,fineX);
   ip = rsqrt(ip);
-  blockZAXPY(fineX,ip,fineX,fineX);
+  blockZAXPY(fineX,ip,fineX,zz);
 }
 // useful in multigrid project;
 // Generic name : Coarsen?
@@ -205,6 +206,26 @@ inline void blockSum(Lattice<vobj> &coarseData,const Lattice<vobj> &fineData)
   return;
 }
 
+template<class vobj>
+inline void blockPick(GridBase *coarse,const Lattice<vobj> &unpicked,Lattice<vobj> &picked,std::vector<int> coor)
+{
+  GridBase * fine = unpicked._grid;
+
+  Lattice<vobj> zz(fine);
+  Lattice<iScalar<vInteger> > fcoor(fine);
+
+  zz = zero;
+
+  picked = unpicked;
+  for(int d=0;d<fine->_ndimension;d++){
+    LatticeCoordinate(fcoor,d);
+    int block= fine->_rdimensions[d] / coarse->_rdimensions[d];
+    int lo   = (coor[d])*block;
+    int hi   = (coor[d]+1)*block;
+    picked = where( (fcoor<hi) , picked, zz);
+    picked = where( (fcoor>=lo), picked, zz);
+  }
+}
 
 template<class vobj,class CComplex>
 inline void blockOrthogonalise(Lattice<CComplex> &ip,std::vector<Lattice<vobj> > &Basis)
diff --git a/lib/lattice/Lattice_unary.h b/lib/lattice/Lattice_unary.h
index 131d7d68..fcdcc484 100644
--- a/lib/lattice/Lattice_unary.h
+++ b/lib/lattice/Lattice_unary.h
@@ -26,7 +26,48 @@ PARALLEL_FOR_LOOP
     }
     return ret;
   }
+  template<class obj> Lattice<obj> sin(const Lattice<obj> &rhs){
+    Lattice<obj> ret(rhs._grid);
+    ret.checkerboard = rhs.checkerboard;
+    conformable(ret,rhs);
+PARALLEL_FOR_LOOP
+    for(int ss=0;ss<rhs._grid->oSites();ss++){
+      ret._odata[ss]=sin(rhs._odata[ss]);
+    }
+    return ret;
+  }
+  template<class obj> Lattice<obj> cos(const Lattice<obj> &rhs){
+    Lattice<obj> ret(rhs._grid);
+    ret.checkerboard = rhs.checkerboard;
+    conformable(ret,rhs);
+PARALLEL_FOR_LOOP
+    for(int ss=0;ss<rhs._grid->oSites();ss++){
+      ret._odata[ss]=cos(rhs._odata[ss]);
+    }
+    return ret;
+  }
 
 
+  template<class obj> Lattice<obj> pow(const Lattice<obj> &rhs,RealD y){
+    Lattice<obj> ret(rhs._grid);
+    ret.checkerboard = rhs.checkerboard;
+    conformable(ret,rhs);
+PARALLEL_FOR_LOOP
+    for(int ss=0;ss<rhs._grid->oSites();ss++){
+      ret._odata[ss]=pow(rhs._odata[ss],y);
+    }
+    return ret;
+  }
+  template<class obj> Lattice<obj> mod(const Lattice<obj> &rhs,Integer y){
+    Lattice<obj> ret(rhs._grid);
+    ret.checkerboard = rhs.checkerboard;
+    conformable(ret,rhs);
+PARALLEL_FOR_LOOP
+    for(int ss=0;ss<rhs._grid->oSites();ss++){
+      ret._odata[ss]=mod(rhs._odata[ss],y);
+    }
+    return ret;
+  }
+
 }
 #endif
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.cc b/lib/qcd/action/fermion/WilsonFermion5D.cc
index 4ddd14ab..7d9ffe22 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.cc
+++ b/lib/qcd/action/fermion/WilsonFermion5D.cc
@@ -85,6 +85,7 @@ void WilsonFermion5D::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGau
 void WilsonFermion5D::DhopDir(const LatticeFermion &in, LatticeFermion &out,int dir,int disp)
 {
   assert( (disp==1)||(disp==-1) );
+  assert( (dir>=0)&&(dir<4) ); //must do x,y,z or t;
 
   WilsonCompressor compressor(DaggerNo);
   Stencil.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
@@ -93,6 +94,9 @@ void WilsonFermion5D::DhopDir(const LatticeFermion &in, LatticeFermion &out,int
 
   int dirdisp = dir+skip*4;
 
+  assert(dirdisp<=7);
+  assert(dirdisp>=0);
+
 PARALLEL_FOR_LOOP
   for(int ss=0;ss<Umu._grid->oSites();ss++){
     for(int s=0;s<Ls;s++){
diff --git a/lib/simd/Grid_vector_unops.h b/lib/simd/Grid_vector_unops.h
index 9aeec350..8e2b6956 100644
--- a/lib/simd/Grid_vector_unops.h
+++ b/lib/simd/Grid_vector_unops.h
@@ -35,6 +35,14 @@ namespace Grid {
     }
   };
 
+  template<class scalar> struct ModIntFunctor {
+    Integer y;
+  ModIntFunctor(Integer _y) : y(_y) {};
+    scalar operator()(const scalar &a)  const {
+      return Integer(a)%y;
+    }
+  };
+
   template < class S, class V > 
   inline Grid_simd<S,V> sqrt(const Grid_simd<S,V> &r) {
     return SimdApply(SqrtRealFunctor<S>(),r);
@@ -55,6 +63,10 @@ namespace Grid {
   inline Grid_simd<S,V> pow(const Grid_simd<S,V> &r,double y) {
     return SimdApply(PowRealFunctor<S>(y),r);
   }
+  template < class S, class V > 
+  inline Grid_simd<S,V> mod(const Grid_simd<S,V> &r,Integer y) {
+    return SimdApply(ModIntFunctor<S>(y),r);
+  }
 
 }
 #endif
diff --git a/lib/tensors/Tensor_class.h b/lib/tensors/Tensor_class.h
index 3ed4f865..019e6fb1 100644
--- a/lib/tensors/Tensor_class.h
+++ b/lib/tensors/Tensor_class.h
@@ -90,10 +90,10 @@ public:
   operator ComplexD () const { return(TensorRemove(_internal)); };
   operator RealD () const { return(real(TensorRemove(_internal))); }
   
-  // convert from a something to a scalar
-  template<class T,typename std::enable_if<!isGridTensor<T>::value, T>::type* = nullptr > strong_inline auto operator = (T arg) -> iScalar<vtype>
+  // convert from a something to a scalar via constructor of something arg
+  template<class T,typename std::enable_if<!isGridTensor<T>::value, T>::type* = nullptr > strong_inline iScalar<vtype> operator = (T arg)
     { 
-      _internal = vtype(arg);
+      _internal = arg;
       return *this;
     }
 
@@ -316,7 +316,8 @@ public:
 	stream<<o._internal[i][j];
 	if (i<N-1)	stream<<",";
       }
-      stream<<"}\n\t\t";
+      stream<<"}";
+      if(i!=N-1) stream<<"\n\t\t";
     }
     stream<<"}";
     return stream;
diff --git a/lib/tensors/Tensor_unary.h b/lib/tensors/Tensor_unary.h
index 1fbec985..2a546701 100644
--- a/lib/tensors/Tensor_unary.h
+++ b/lib/tensors/Tensor_unary.h
@@ -27,11 +27,40 @@ template<class obj,int N> inline auto func(const iMatrix<obj,N> &z) -> iMatrix<o
     return ret;\
 }
 
+
+#define BINARY_RSCALAR(func,scal)					\
+template<class obj> inline iScalar<obj> func(const iScalar<obj> &z,scal y)	\
+{\
+    iScalar<obj> ret;\
+    ret._internal = func(z._internal,y);	\
+    return ret;\
+}\
+ template<class obj,int N> inline iVector<obj,N> func(const iVector<obj,N> &z,scal y) \
+{\
+    iVector<obj,N> ret;\
+    for(int c1=0;c1<N;c1++){\
+      ret._internal[c1] = func(z._internal[c1],y);	\
+    }\
+    return ret;\
+}\
+ template<class obj,int N> inline  iMatrix<obj,N> func(const iMatrix<obj,N> &z, scal y) \
+{\
+    iMatrix<obj,N> ret;\
+    for(int c1=0;c1<N;c1++){\
+    for(int c2=0;c2<N;c2++){\
+      ret._internal[c1][c2] = func(z._internal[c1][c2],y);	\
+    }}\
+    return ret;\
+}
+
 UNARY_REAL(sqrt);
 UNARY_REAL(rsqrt);
 UNARY_REAL(sin);
 UNARY_REAL(cos);
 
+BINARY_RSCALAR(mod,Integer);
+BINARY_RSCALAR(pow,RealD);
+
 
 }
 #endif
diff --git a/scripts/linecount b/scripts/linecount
index 7624c15f..0be67142 100755
--- a/scripts/linecount
+++ b/scripts/linecount
@@ -1,3 +1,12 @@
 #!/bin/sh 
 
-wc -l lib/*.h lib/*/*.h lib/*/*/*.h lib/*.cc lib/*/*.cc lib/*/*/*.cc tests/*.cc benchmarks/*.cc lib/*/*/*/*.cc lib/*/*/*/*.h
\ No newline at end of file
+wc -l lib/*.h lib/*/*.h lib/*/*/*.h lib/*.cc lib/*/*.cc lib/*/*/*.cc tests/*.cc benchmarks/*.cc lib/*/*/*/*.cc lib/*/*/*/*.h
+
+echo "lib"
+wc -l lib/*.h lib/*/*.h lib/*/*/*.h lib/*.cc lib/*/*.cc lib/*/*/*.cc lib/*/*/*/*.cc lib/*/*/*/*.h | grep total
+
+echo "tests"
+wc -l tests/*.cc | grep total
+
+echo "benchmarks"
+wc -l benchmarks/*.cc | grep total
diff --git a/tests/Test_cayley_coarsen_support.cc b/tests/Test_cayley_coarsen_support.cc
index acd96da3..064f5f13 100644
--- a/tests/Test_cayley_coarsen_support.cc
+++ b/tests/Test_cayley_coarsen_support.cc
@@ -23,7 +23,7 @@ int main (int argc, char ** argv)
 
 
 
-  const int Ls=8;
+  const int Ls=4;
 
   GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
   GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
@@ -50,13 +50,18 @@ int main (int argc, char ** argv)
   LatticeFermion    tmp(FGrid);
   LatticeFermion    err(FGrid);
   LatticeGaugeField Umu(UGrid); random(RNG4,Umu);
-
+#if 0
   std::vector<LatticeColourMatrix> U(4,UGrid);
-
-  for(int mu=0;mu<Nd;mu++){
-    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+  Umu=zero;
+  Complex cone(1.0,0.0);
+  for(int nn=0;nn<Nd;nn++){
+    if(1) {
+      if (nn!=0) { U[nn]=zero; std::cout << "zeroing gauge field in dir "<<nn<<std::endl; }
+      else       { U[nn] = cone;std::cout << "unit gauge field in dir "<<nn<<std::endl; }
+    }
+    pokeIndex<LorentzIndex>(Umu,U[nn],nn);
   }
-  
+#endif  
   RealD mass=0.5;
   RealD M5=1.8;
 
@@ -75,7 +80,7 @@ int main (int argc, char ** argv)
   err = result-ref;
   std::cout<<"Error "<<norm2(err)<<std::endl;
 
-  const int nbasis = 8;
+  const int nbasis = 2;
   std::vector<LatticeFermion> subspace(nbasis,FGrid);
   
   for(int b=0;b<nbasis;b++){
diff --git a/tests/Test_cshift.cc b/tests/Test_cshift.cc
index cc1da333..85f8529c 100644
--- a/tests/Test_cshift.cc
+++ b/tests/Test_cshift.cc
@@ -34,7 +34,6 @@ int main (int argc, char ** argv)
   }
 
 
-
   TComplex cm;
   
   for(int dir=0;dir<4;dir++){

From d60cfe31a78a7df7056dc2eb1fb49c4d55ed47d6 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Tue, 9 Jun 2015 18:59:45 +0900
Subject: [PATCH 293/429] Adding support for iMatrix exponentiation

---
 lib/simd/Old/Grid_vComplexD.h | 430 -------------------------------
 lib/simd/Old/Grid_vComplexF.h | 463 ----------------------------------
 lib/simd/Old/Grid_vInteger.h  | 259 -------------------
 lib/simd/Old/Grid_vRealD.h    | 276 --------------------
 lib/simd/Old/Grid_vRealF.h    | 313 -----------------------
 lib/tensors/Tensor_Ta.h       |  84 +++++-
 tests/InvSqrt.gnu             |   0
 tests/Sqrt.gnu                |   2 +
 tests/Test_main.cc            |   3 +
 9 files changed, 83 insertions(+), 1747 deletions(-)
 delete mode 100644 lib/simd/Old/Grid_vComplexD.h
 delete mode 100644 lib/simd/Old/Grid_vComplexF.h
 delete mode 100644 lib/simd/Old/Grid_vInteger.h
 delete mode 100644 lib/simd/Old/Grid_vRealD.h
 delete mode 100644 lib/simd/Old/Grid_vRealF.h
 create mode 100644 tests/InvSqrt.gnu
 create mode 100644 tests/Sqrt.gnu

diff --git a/lib/simd/Old/Grid_vComplexD.h b/lib/simd/Old/Grid_vComplexD.h
deleted file mode 100644
index 116c2866..00000000
--- a/lib/simd/Old/Grid_vComplexD.h
+++ /dev/null
@@ -1,430 +0,0 @@
-#ifndef GRID_VCOMPLEXD_H
-#define  GRID_VCOMPLEXD_H
-
-namespace Grid {
-    class vComplexD {
-    public:
-        zvec v;
-    public:
-	typedef zvec     vector_type;
-	typedef ComplexD scalar_type;
-
-        vComplexD & operator = ( Zero & z){
-            vzero(*this);
-            return (*this);
-        }
-        vComplexD( Zero & z){
-            vzero(*this);
-        }
-        vComplexD()=default;
-        vComplexD(ComplexD a){
-	  vsplat(*this,a);
-	};
-        vComplexD(double a){
-	  vsplat(*this,ComplexD(a));
-	};
- 
-        ///////////////////////////////////////////////
-        // mac, mult, sub, add, adj
-        // Should do an AVX2 version with mac.
-       ///////////////////////////////////////////////
-        friend inline void mac (vComplexD * __restrict__ y,const vComplexD * __restrict__ a,const vComplexD *__restrict__ x) {*y = (*a)*(*x)+(*y);};
-        friend inline void mult(vComplexD * __restrict__ y,const vComplexD * __restrict__ l,const vComplexD *__restrict__ r) {*y = (*l) * (*r);}
-        friend inline void sub (vComplexD * __restrict__ y,const vComplexD * __restrict__ l,const vComplexD *__restrict__ r) {*y = (*l) - (*r);}
-        friend inline void add (vComplexD * __restrict__ y,const vComplexD * __restrict__ l,const vComplexD *__restrict__ r) {*y = (*l) + (*r);}
-        friend inline vComplexD adj(const vComplexD &in){ return conjugate(in); }
-
-        //////////////////////////////////
-        // Initialise to 1,0,i
-        //////////////////////////////////
-        friend inline void vone      (vComplexD &ret){ vsplat(ret,1.0,0.0);}
-        friend inline void vzero     (vComplexD &ret){ vsplat(ret,0.0,0.0);}
-        friend inline void vcomplex_i(vComplexD &ret){ vsplat(ret,0.0,1.0);}
-        
-        ////////////////////////////////////
-        // Arithmetic operator overloads +,-,*
-        ////////////////////////////////////
-        friend inline vComplexD operator + (vComplexD a, vComplexD b)
-        {
-            vComplexD ret;
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_add_pd(a.v,b.v);
-#endif
-#ifdef SSE4
-            ret.v = _mm_add_pd(a.v,b.v);
-#endif
-#ifdef AVX512
-            ret.v = _mm512_add_pd(a.v,b.v);
-#endif
-#ifdef QPX
-            ret.v = vec_add(a.v,b.v);
-#endif
-            return ret;
-        };
-        
-        friend inline vComplexD operator - (vComplexD a, vComplexD b)
-        {
-            vComplexD ret;
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_sub_pd(a.v,b.v);
-#endif
-#ifdef SSE4
-            ret.v = _mm_sub_pd(a.v,b.v);
-#endif
-#ifdef AVX512
-            ret.v = _mm512_sub_pd(a.v,b.v);
-#endif
-#ifdef QPX
-            ret.v = vec_sub(a.v,b.v);
-#endif
-            return ret;
-        };
-        
-        friend inline vComplexD operator * (vComplexD a, vComplexD b)
-        {
-            vComplexD ret;
-            
-            //Multiplicationof (ak+ibk)*(ck+idk)
-            // a + i b can be stored as a data structure
-            //From intel optimisation reference guide
-            /*
-             movsldup xmm0, Src1; load real parts into the destination,
-             ; a1, a1, a0, a0
-             movaps xmm1, src2; load the 2nd pair of complex values, ; i.e. d1, c1, d0, c0
-             mulps xmm0, xmm1; temporary results, a1d1, a1c1, a0d0, ; a0c0
-             shufps xmm1, xmm1, b1; reorder the real and imaginary ; parts, c1, d1, c0, d0
-             movshdup xmm2, Src1; load the imaginary parts into the ; destination, b1, b1, b0, b0
-             mulps xmm2, xmm1; temporary results, b1c1, b1d1, b0c0, ; b0d0
-             addsubps xmm0, xmm2; b1c1+a1d1, a1c1 -b1d1, b0c0+a0d
-             VSHUFPD (VEX.256 encoded version)
-             IF IMM0[0] = 0
-             THEN DEST[63:0]=SRC1[63:0] ELSE DEST[63:0]=SRC1[127:64] FI;
-             IF IMM0[1] = 0
-             THEN DEST[127:64]=SRC2[63:0] ELSE DEST[127:64]=SRC2[127:64] FI;
-             IF IMM0[2] = 0
-             THEN DEST[191:128]=SRC1[191:128] ELSE DEST[191:128]=SRC1[255:192] FI;
-             IF IMM0[3] = 0
-             THEN DEST[255:192]=SRC2[191:128] ELSE DEST[255:192]=SRC2[255:192] FI; // Ox5 r<->i   ; 0xC unchanged
-             */
-#if defined (AVX1)|| defined (AVX2)
-            zvec ymm0,ymm1,ymm2;
-            ymm0 = _mm256_shuffle_pd(a.v,a.v,0x0); // ymm0 <- ar ar, ar,ar b'00,00
-            ymm0 = _mm256_mul_pd(ymm0,b.v);        // ymm0 <- ar bi, ar br
-            ymm1 = _mm256_shuffle_pd(b.v,b.v,0x5); // ymm1 <- br,bi  b'01,01 
-            ymm2 = _mm256_shuffle_pd(a.v,a.v,0xF); // ymm2 <- ai,ai  b'11,11
-            ymm1 = _mm256_mul_pd(ymm1,ymm2);       // ymm1 <- br ai, ai bi
-            ret.v= _mm256_addsub_pd(ymm0,ymm1);
-#endif
-#ifdef SSE4
-            zvec ymm0,ymm1,ymm2;
-            ymm0 = _mm_shuffle_pd(a.v,a.v,0x0); // ymm0 <- ar ar,
-            ymm0 = _mm_mul_pd(ymm0,b.v);        // ymm0 <- ar bi, ar br
-            ymm1 = _mm_shuffle_pd(b.v,b.v,0x1); // ymm1 <- br,bi   b01
-            ymm2 = _mm_shuffle_pd(a.v,a.v,0x3); // ymm2 <- ai,ai   b11
-            ymm1 = _mm_mul_pd(ymm1,ymm2);       // ymm1 <- br ai, ai bi
-            ret.v= _mm_addsub_pd(ymm0,ymm1);
-#endif
-#ifdef AVX512
-            /* This is from
-             * Automatic SIMD Vectorization of Fast Fourier Transforms for the Larrabee and AVX Instruction Sets 
-             * @inproceedings{McFarlin:2011:ASV:1995896.1995938,
-             * author = {McFarlin, Daniel S. and Arbatov, Volodymyr and Franchetti, Franz and P\"{u}schel, Markus},
-             * title = {Automatic SIMD Vectorization of Fast Fourier Transforms for the Larrabee and AVX Instruction Sets},
-             * booktitle = {Proceedings of the International Conference on Supercomputing},
-             * series = {ICS '11},
-             * year = {2011},
-             * isbn = {978-1-4503-0102-2},
-             * location = {Tucson, Arizona, USA},
-             * pages = {265--274},
-             * numpages = {10},
-             * url = {http://doi.acm.org/10.1145/1995896.1995938},
-             * doi = {10.1145/1995896.1995938},
-             * acmid = {1995938},
-             * publisher = {ACM},
-             * address = {New York, NY, USA},
-             * keywords = {autovectorization, fourier transform, program generation, simd, super-optimization},
-	     *                } 
-             */
-            zvec vzero,ymm0,ymm1,real,imag;
-            vzero =(zvec)_mm512_setzero();
-            ymm0 =  _mm512_swizzle_pd(a.v, _MM_SWIZ_REG_CDAB); // 
-            real =(zvec)_mm512_mask_or_epi64((__m512i)a.v, 0xAA,(__m512i)vzero,(__m512i) ymm0);
-            imag =  _mm512_mask_sub_pd(a.v, 0x55,vzero, ymm0);
-            ymm1 =  _mm512_mul_pd(real, b.v);
-            ymm0 =  _mm512_swizzle_pd(b.v, _MM_SWIZ_REG_CDAB); // OK
-            ret.v=  _mm512_fmadd_pd(ymm0,imag,ymm1);
-             /* Imag OK */
-#endif
-#ifdef QPX
-            ret.v = vec_mul(a.v,b.v);
-#endif
-            return ret;
-        };
-
-	////////////////////////////////////////////////////////////////////
-	// General permute; assumes vector length is same across 
-	// all subtypes; may not be a good assumption, but could
-	// add the vector width as a template param for BG/Q for example
-	////////////////////////////////////////////////////////////////////
-	friend inline void permute(vComplexD &y,vComplexD b,int perm)
-	{
-	  Gpermute<vComplexD>(y,b,perm);
-	}
-	/*
-	friend inline void merge(vComplexD &y,std::vector<ComplexD *> &extracted)
-	{
-	  Gmerge<vComplexD,ComplexD >(y,extracted);
-	}
-	friend inline void extract(const vComplexD &y,std::vector<ComplexD *> &extracted)
-	{
-	  Gextract<vComplexD,ComplexD>(y,extracted);
-	}
-	friend inline void merge(vComplexD &y,std::vector<ComplexD > &extracted)
-	{
-	  Gmerge<vComplexD,ComplexD >(y,extracted);
-	}
-	friend inline void extract(const vComplexD &y,std::vector<ComplexD > &extracted)
-	{
-	  Gextract<vComplexD,ComplexD>(y,extracted);
-	}
-	*/
-
-        ///////////////////////
-        // Splat
-        ///////////////////////
-        friend inline void vsplat(vComplexD &ret,ComplexD c){
-            float a= real(c);
-            float b= imag(c);
-            vsplat(ret,a,b);
-        }
-
-
-        friend inline void vsplat(vComplexD &ret,double rl,double ig){
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_set_pd(ig,rl,ig,rl);
-#endif
-#ifdef SSE4
-            ret.v = _mm_set_pd(ig,rl);
-#endif
-#ifdef AVX512
-            ret.v = _mm512_set_pd(ig,rl,ig,rl,ig,rl,ig,rl);
-#endif
-#ifdef QPX
-            ret.v = {ig,rl,ig,rl};
-#endif
-        }
-
-        friend inline void vset(vComplexD &ret,ComplexD *a){
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_set_pd(a[1].imag(),a[1].real(),a[0].imag(),a[0].real());
-#endif
-#ifdef SSE4
-            ret.v = _mm_set_pd(a[0].imag(),a[0].real());
-#endif
-#ifdef AVX512
-            ret.v = _mm512_set_pd(a[3].imag(),a[3].real(),a[2].imag(),a[2].real(),a[1].imag(),a[1].real(),a[0].imag(),a[0].real());
-            // Note v has a0 a1 a2 a3 
-#endif
-#ifdef QPX
-            ret.v = {a[0].real(),a[0].imag(),a[1].real(),a[3].imag()};
-#endif
-        }
-
-friend inline void vstore(const vComplexD &ret, ComplexD *a){
-#if defined (AVX1)|| defined (AVX2)
-       _mm256_store_pd((double *)a,ret.v);
-#endif
-#ifdef SSE4
-       _mm_store_pd((double *)a,ret.v);
-#endif
-#ifdef AVX512
-       _mm512_store_pd((double *)a,ret.v);
-   //Note v has a3 a2 a1 a0
-#endif
-#ifdef QPX
-	assert(0);
-#endif
-        }
-        friend inline void vstream(vComplexD &out,const vComplexD &in){
-#if defined (AVX1)|| defined (AVX2)
-	  _mm256_stream_pd((double *)&out.v,in.v);
-#endif
-#ifdef SSE4
-	  _mm_stream_pd((double *)&out.v,in.v);
-#endif
-#ifdef AVX512
-	  _mm512_storenrngo_pd((double *)&out.v,in.v);
-	  //	  _mm512_stream_pd((double *)&out.v,in.v);
-	  //Note v has a3 a2 a1 a0
-#endif
-#ifdef QPX
-	  assert(0);
-#endif
-	}
-      friend inline void prefetch(const vComplexD &v)
-        {
-            _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
-        }
-
-        ////////////////////////
-        // Conjugate
-        ////////////////////////
-        friend inline vComplexD conjugate(const vComplexD &in){
-            vComplexD ret ; vzero(ret);
-#if defined (AVX1)|| defined (AVX2)
-	    //	    addsubps 0, inv=>0+in.v[3] 0-in.v[2], 0+in.v[1], 0-in.v[0], ...
-	    zvec tmp = _mm256_addsub_pd(ret.v,_mm256_shuffle_pd(in.v,in.v,0x5));
-	    ret.v    =_mm256_shuffle_pd(tmp,tmp,0x5);
-#endif
-#ifdef SSE4
-	    zvec tmp = _mm_addsub_pd(ret.v,_mm_shuffle_pd(in.v,in.v,0x1));
-	    ret.v    = _mm_shuffle_pd(tmp,tmp,0x1);
-#endif
-#ifdef AVX512
-	    ret.v = _mm512_mask_sub_pd(in.v, 0xaa,ret.v, in.v);             
-#endif
-#ifdef QPX
-	     assert(0);
-#endif
-            return ret;
-        }
-
-        friend inline void timesMinusI(vComplexD &ret,const vComplexD &in){
-	  vzero(ret);
-	  vComplexD tmp;
-#if defined (AVX1)|| defined (AVX2)
-	  tmp.v    =_mm256_addsub_pd(ret.v,in.v); // r,-i
-          ret.v    =_mm256_shuffle_pd(tmp.v,tmp.v,0x5);
-#endif
-#ifdef SSE4
-	  tmp.v    =_mm_addsub_pd(ret.v,in.v); // r,-i
-          ret.v    =_mm_shuffle_pd(tmp.v,tmp.v,0x1);
-#endif
-#ifdef AVX512
-          ret.v = _mm512_mask_sub_pd(in.v,0xaa,ret.v,in.v); // real -imag 
-	  ret.v = _mm512_swizzle_pd(ret.v, _MM_SWIZ_REG_CDAB);// OK
-#endif
-#ifdef QPX
-            assert(0);
-#endif
-	}
-
-        friend inline void timesI(vComplexD &ret, const vComplexD &in){
-	  vzero(ret);
-	  vComplexD tmp;
-#if defined (AVX1)|| defined (AVX2)
-	  tmp.v    =_mm256_shuffle_pd(in.v,in.v,0x5);
-          ret.v    =_mm256_addsub_pd(ret.v,tmp.v); // i,-r
-#endif
-#ifdef SSE4
-	  tmp.v    =_mm_shuffle_pd(in.v,in.v,0x1);
-          ret.v    =_mm_addsub_pd(ret.v,tmp.v); // r,-i
-#endif
-#ifdef AVX512
-          tmp.v    = _mm512_swizzle_pd(in.v, _MM_SWIZ_REG_CDAB);// OK
-	  ret.v    = _mm512_mask_sub_pd(tmp.v,0xaa,ret.v,tmp.v); // real -imag 
-#endif
-#ifdef QPX
-            assert(0);
-#endif
-	}
-
-        friend inline vComplexD timesMinusI(const vComplexD &in){
-	  vComplexD ret; 
-	  timesMinusI(ret,in);
-	  return ret;
-	}
-
-        friend inline vComplexD timesI(const vComplexD &in){
-	  vComplexD ret; 
-	  timesI(ret,in);
-	  return ret;
-	}
-
-
-// REDUCE FIXME must be a cleaner implementation
-       friend inline ComplexD Reduce(const vComplexD & in)
-       { 
-	 vComplexD v1,v2;
-	 union { 
-	   zvec v;
-	   double f[sizeof(zvec)/sizeof(double)];
-	 } conv;
-	   
-#ifdef SSE4
-	 v1=in;
-#endif
-#if defined(AVX1) || defined (AVX2)
-	 permute(v1,in,0); // sse 128; paired complex single
-	 v1=v1+in;
-#endif
-#ifdef AVX512
-	 permute(v1,in,0); // sse 128; paired complex single
-	 v1=v1+in;
-	 permute(v2,v1,1); // avx 256; quad complex single
-	 v1=v1+v2;
-#endif 
-#ifdef QPX
-#error
-#endif
-	 conv.v = v1.v;
-	 return ComplexD(conv.f[0],conv.f[1]);
-        }
-        
-        // Unary negation
-        friend inline vComplexD operator -(const vComplexD &r) {
-            vComplexD ret;
-            vzero(ret);
-            ret = ret - r;
-            return ret;
-        }
-        // *=,+=,-= operators
-        inline vComplexD &operator *=(const vComplexD &r) {
-            *this = (*this)*r;
-            return *this;
-        }
-        inline vComplexD &operator +=(const vComplexD &r) {
-            *this = *this+r;
-            return *this;
-        }
-        inline vComplexD &operator -=(const vComplexD &r) {
-            *this = *this-r;
-            return *this;
-        }
-
-    public:
-        static int Nsimd(void) { return sizeof(zvec)/sizeof(double)/2;}
-    };
-
-
-    inline vComplexD innerProduct(const vComplexD & l, const vComplexD & r) { return conjugate(l)*r; }
-
-
-    typedef  vComplexD vDComplex;
-    inline void zeroit(vComplexD &z){ vzero(z);}
-
-    inline vComplexD outerProduct(const vComplexD &l, const vComplexD& r)
-    {
-        return l*r;
-    }
-    inline vComplexD trace(const vComplexD &arg){
-        return arg;
-    }
-/////////////////////////////////////////////////////////////////////////
-//// Generic routine to promote object<complex> -> object<vcomplex>
-//// Supports the array reordering transformation that gives me SIMD utilisation
-///////////////////////////////////////////////////////////////////////////
-/*
-template<template<class> class object>
-inline object<vComplex> splat(object<Complex >s){
-      object<vComplex> ret;
-      vComplex * v_ptr = (vComplex *)& ret;
-      Complex * s_ptr = (Complex *) &s;
-      for(int i=0;i<sizeof(ret);i+=sizeof(vComplex)){
-          vsplat(*(v_ptr++),*(s_ptr++));
-      }
-      return ret;
-    }
-*/
-}
-#endif
diff --git a/lib/simd/Old/Grid_vComplexF.h b/lib/simd/Old/Grid_vComplexF.h
deleted file mode 100644
index 828145d6..00000000
--- a/lib/simd/Old/Grid_vComplexF.h
+++ /dev/null
@@ -1,463 +0,0 @@
-#ifndef GRID_VCOMPLEXF
-#define GRID_VCOMPLEXF
-
-namespace Grid {
-
-  /*
-  inline void Print(const char *A,cvec c) { 
-	  float *fp=(float *)&c; 
-	  printf(A); 
-	  printf(" %le %le %le %le %le %le %le %le\n",
-		 fp[0],fp[1],fp[2],fp[3],fp[4],fp[5],fp[6],fp[7]);
-	}
-  */
-
-    class vComplexF {
-      //    protected:
-
-    public:
-        cvec v;
-        
-    public:
-        static inline int Nsimd(void) { return sizeof(cvec)/sizeof(float)/2;}
-    public:
-	typedef cvec     vector_type;
-	typedef ComplexF scalar_type;
-
-        vComplexF & operator = ( Zero & z){
-            vzero(*this);
-            return (*this);
-        }
-	//        vComplexF( Zero & z){
-	//            vzero(*this);
-	//        }
-        vComplexF()=default;
-        vComplexF(ComplexF a){
-	  vsplat(*this,a);
-	};
-        vComplexF(double a){
-	  vsplat(*this,ComplexF(a));
-	};
-       
-        ///////////////////////////////////////////////
-        // mac, mult, sub, add, adj
-        // Should do an AVX2 version with mac.
-        ///////////////////////////////////////////////
-        friend inline void mac (vComplexF * __restrict__ y,const vComplexF * __restrict__ a,const vComplexF *__restrict__ x){ *y = (*a)*(*x)+(*y); };
-        friend inline void mult(vComplexF * __restrict__ y,const vComplexF * __restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) * (*r); }
-        friend inline void sub (vComplexF * __restrict__ y,const vComplexF * __restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) - (*r); }
-        friend inline void add (vComplexF * __restrict__ y,const vComplexF * __restrict__ l,const vComplexF *__restrict__ r){ *y = (*l) + (*r); }
-        friend inline vComplexF adj(const vComplexF &in){ return conjugate(in); }
-        
-        //////////////////////////////////
-        // Initialise to 1,0,i
-        //////////////////////////////////
-        friend inline void vone(vComplexF &ret)      { vsplat(ret,1.0,0.0); }
-        friend inline void vzero(vComplexF &ret)     { vsplat(ret,0.0,0.0); }
-        friend inline void vcomplex_i(vComplexF &ret){ vsplat(ret,0.0,1.0); }
-          
-        ////////////////////////////////////
-        // Arithmetic operator overloads +,-,*
-        ////////////////////////////////////
-        friend inline vComplexF operator + (vComplexF a, vComplexF b)
-        {
-            vComplexF ret;
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_add_ps(a.v,b.v);
-#endif
-#ifdef SSE4
-            ret.v = _mm_add_ps(a.v,b.v);
-#endif
-#ifdef AVX512
-            ret.v = _mm512_add_ps(a.v,b.v);
-#endif
-#ifdef QPX
-#error
-#endif
-            return ret;
-        };
-        
-        friend inline vComplexF operator - (vComplexF a, vComplexF b)
-        {
-            vComplexF ret;
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_sub_ps(a.v,b.v);
-#endif
-#ifdef SSE4
-            ret.v = _mm_sub_ps(a.v,b.v);
-#endif
-#ifdef AVX512
-            ret.v = _mm512_sub_ps(a.v,b.v);
-#endif
-#ifdef QPX
-#error
-#endif
-            return ret;
-        };
-        
-        friend inline vComplexF operator * (vComplexF a, vComplexF b)
-        {
-            vComplexF ret;
-            
-            //Multiplicationof (ak+ibk)*(ck+idk)
-            // a + i b can be stored as a data structure
-            //From intel optimisation reference
-            /*
-             movsldup xmm0, Src1; load real parts into the destination,
-             ; a1, a1, a0, a0
-             movaps xmm1, src2; load the 2nd pair of complex values, ; i.e. d1, c1, d0, c0
-             mulps xmm0, xmm1; temporary results, a1d1, a1c1, a0d0, ; a0c0
-             shufps xmm1, xmm1, b1; reorder the real and imaginary ; parts, c1, d1, c0, d0
-             movshdup xmm2, Src1; load the imaginary parts into the ; destination, b1, b1, b0, b0
-             mulps xmm2, xmm1; temporary results, b1c1, b1d1, b0c0, ; b0d0
-             addsubps xmm0, xmm2; b1c1+a1d1, a1c1 -b1d1, b0c0+a0d
-             */
-#if defined (AVX1)|| defined (AVX2)
-            cvec ymm0,ymm1,ymm2;
-            ymm0 = _mm256_shuffle_ps(a.v,a.v,_MM_SHUFFLE(2,2,0,0)); // ymm0 <- ar ar,
-            ymm0 = _mm256_mul_ps(ymm0,b.v);        // ymm0 <- ar bi, ar br
-            // FIXME AVX2 could MAC
-            ymm1 = _mm256_shuffle_ps(b.v,b.v,_MM_SHUFFLE(2,3,0,1)); // ymm1 <- br,bi
-            ymm2 = _mm256_shuffle_ps(a.v,a.v,_MM_SHUFFLE(3,3,1,1)); // ymm2 <- ai,ai
-            ymm1 = _mm256_mul_ps(ymm1,ymm2);       // ymm1 <- br ai, ai bi
-            ret.v= _mm256_addsub_ps(ymm0,ymm1);    
-#endif
-#ifdef SSE4
-            cvec ymm0,ymm1,ymm2;
-            ymm0 = _mm_shuffle_ps(a.v,a.v,_MM_SHUFFLE(2,2,0,0)); // ymm0 <- ar ar,
-            ymm0 = _mm_mul_ps(ymm0,b.v);        // ymm0 <- ar bi, ar br
-            ymm1 = _mm_shuffle_ps(b.v,b.v,_MM_SHUFFLE(2,3,0,1)); // ymm1 <- br,bi
-            ymm2 = _mm_shuffle_ps(a.v,a.v,_MM_SHUFFLE(3,3,1,1)); // ymm2 <- ai,ai
-            ymm1 = _mm_mul_ps(ymm1,ymm2);       // ymm1 <- br ai, ai bi
-            ret.v= _mm_addsub_ps(ymm0,ymm1);
-#endif
-#ifdef AVX512
-//
-            cvec vzero,ymm0,ymm1,real, imag;
-            vzero = _mm512_setzero();
-            ymm0  = _mm512_swizzle_ps(a.v, _MM_SWIZ_REG_CDAB); // 
-            real  = (__m512)_mm512_mask_or_epi32((__m512i)a.v, 0xAAAA,(__m512i)vzero,(__m512i)ymm0);
-            imag  = _mm512_mask_sub_ps(a.v, 0x5555,vzero, ymm0);
-            ymm1  = _mm512_mul_ps(real, b.v);
-            ymm0  = _mm512_swizzle_ps(b.v, _MM_SWIZ_REG_CDAB); // OK
-            ret.v = _mm512_fmadd_ps(ymm0,imag,ymm1);
-
-
-#endif
-#ifdef QPX
-            ret.v = vec_mul(a.v,b.v);
-#endif
-            return ret;
-        };
-      
-
-	////////////////////////////////////////////////////////////////////////
-	// FIXME:  gonna remove these load/store, get, set, prefetch
-	////////////////////////////////////////////////////////////////////////
-        friend inline void vset(vComplexF &ret, ComplexF *a){
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_set_ps(a[3].imag(),a[3].real(),a[2].imag(),a[2].real(),a[1].imag(),a[1].real(),a[0].imag(),a[0].real());
-#endif
-#ifdef SSE4
-            ret.v = _mm_set_ps(a[1].imag(), a[1].real(),a[0].imag(),a[0].real());
-#endif
-#ifdef AVX512
-            ret.v = _mm512_set_ps(a[7].imag(),a[7].real(),a[6].imag(),a[6].real(),a[5].imag(),a[5].real(),a[4].imag(),a[4].real(),a[3].imag(),a[3].real(),a[2].imag(),a[2].real(),a[1].imag(),a[1].real(),a[0].imag(),a[0].real());
-            // Note v has a0 a1 a2 a3 a4 a5 a6 a7
-#endif
-#ifdef QPX
-	    ret.v = {a[0].real(),a[0].imag(),a[1].real(),a[1].imag(),a[2].real(),a[2].imag(),a[3].real(),a[3].imag()};
-#endif
-        }
-        
-        ///////////////////////
-        // Splat
-        ///////////////////////
-        friend inline void vsplat(vComplexF &ret,ComplexF c){
-            float a= real(c);
-            float b= imag(c);
-            vsplat(ret,a,b);
-        }
-
-	friend inline void vstore(const vComplexF &ret, ComplexF *a){
-#if defined (AVX1)|| defined (AVX2)
-	  _mm256_store_ps((float *)a,ret.v);
-#endif
-#ifdef SSE4
-	  _mm_store_ps((float *)a,ret.v);
-#endif
-#ifdef AVX512
-	  _mm512_store_ps((float *)a,ret.v);
-	  //Note v has a3 a2 a1 a0
-#endif
-#ifdef QPX
-	  assert(0);
-#endif
-	}
-        friend inline void vstream(vComplexF &out,const vComplexF &in){
-#if defined (AVX1)|| defined (AVX2)
-	  _mm256_stream_ps((float *)&out.v,in.v);
-#endif
-#ifdef SSE4
-	  _mm_stream_ps((float *)&out.v,in.v);
-#endif
-#ifdef AVX512
-	  _mm512_storenrngo_ps((float *)&out.v,in.v);
-	  //Note v has a3 a2 a1 a0
-#endif
-#ifdef QPX
-	  assert(0);
-#endif
-	}
-      friend inline void prefetch(const vComplexF &v)
-        {
-            _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
-        }
-
-        friend inline void vsplat(vComplexF &ret,float a,float b){
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_set_ps(b,a,b,a,b,a,b,a);
-#endif
-#ifdef SSE4
-            ret.v = _mm_set_ps(b,a,b,a);
-#endif
-#ifdef AVX512
-            ret.v = _mm512_set_ps(b,a,b,a,b,a,b,a,b,a,b,a,b,a,b,a);
-#endif
-#ifdef QPX
-            ret.v = {a,b,a,b};
-#endif
-        }
-	friend inline void permute(vComplexF &y,vComplexF b,int perm)
-	{
-	  Gpermute<vComplexF>(y,b,perm);
-	}
-	friend inline ComplexF Reduce(const vComplexF & in)
-	{
-	 vComplexF v1,v2;
-	 union { 
-	   cvec v;
-	   float f[sizeof(cvec)/sizeof(float)];
-	 } conv;
-#ifdef SSE4
-	 permute(v1,in,0); // sse 128; paired complex single
-	 v1=v1+in;
-#endif
-#if defined(AVX1) || defined (AVX2)
-	 permute(v1,in,0); // sse 128; paired complex single
-	 v1=v1+in;
-	 permute(v2,v1,1); // avx 256; quad complex single
-	 v1=v1+v2;
-#endif
-#ifdef AVX512
-	 permute(v1,in,0); // avx512 octo-complex single
-	 v1=v1+in;
-	 permute(v2,v1,1); 
-	 v1=v1+v2;
-	 permute(v2,v1,2); 
-	 v1=v1+v2;
-#endif
-#ifdef QPX
-#error
-#endif
-	 conv.v = v1.v;
-	 return ComplexF(conv.f[0],conv.f[1]);
-        }
-
-        friend inline vComplexF operator * (const ComplexF &a, vComplexF b){
-            vComplexF va;
-            vsplat(va,a);
-            return va*b;
-        }
-        friend inline vComplexF operator * (vComplexF b,const ComplexF &a){
-	  return a*b;
-        }
-
-       /*
-	template<class real>
-        friend inline vComplexF operator * (vComplexF b,const real &a){
-            vComplexF va;
-	    Complex ca(a,0);
-            vsplat(va,ca);
-            return va*b;
-        }
-	template<class real>
-	friend inline vComplexF operator * (const real &a,vComplexF b){
-	  return a*b;
-	}
-
-        friend inline vComplexF operator + (const Complex &a, vComplexF b){
-            vComplexF va;
-            vsplat(va,a);
-            return va+b;
-        }
-        friend inline vComplexF operator + (vComplexF b,const Complex &a){
-            return a+b;
-        }
-	template<class real>
-        friend inline vComplexF operator + (vComplexF b,const real &a){
-            vComplexF va;
-	    Complex ca(a,0);
-            vsplat(va,ca);
-            return va+b;
-        }
-	template<class real>
-	friend inline vComplexF operator + (const real &a,vComplexF b){
-	  return a+b;
-	}
-        friend inline vComplexF operator - (const Complex &a, vComplexF b){
-            vComplexF va;
-            vsplat(va,a);
-            return va-b;
-        }
-        friend inline vComplexF operator - (vComplexF b,const Complex &a){
-            vComplexF va;
-            vsplat(va,a);
-            return b-va;
-        }
-	template<class real>
-        friend inline vComplexF operator - (vComplexF b,const real &a){
-            vComplexF va;
-	    Complex ca(a,0);
-            vsplat(va,ca);
-            return b-va;
-        }
-	template<class real>
-	friend inline vComplexF operator - (const real &a,vComplexF b){
-            vComplexF va;
-	    Complex ca(a,0);
-            vsplat(va,ca);
-            return va-b;
-	}
-       */
-       
-
-        ///////////////////////
-        // Conjugate
-        ///////////////////////
-								     
-        friend inline vComplexF conjugate(const vComplexF &in){
-            vComplexF ret ; vzero(ret);
-#if defined (AVX1)|| defined (AVX2)
-	    ret.v = _mm256_xor_ps(_mm256_addsub_ps(ret.v,in.v), _mm256_set1_ps(-0.f));
-#endif
-#ifdef SSE4
-	    ret.v = _mm_xor_ps(_mm_addsub_ps(ret.v,in.v), _mm_set1_ps(-0.f));
-#endif
-#ifdef AVX512
-            ret.v = _mm512_mask_sub_ps(in.v,0xaaaa,ret.v,in.v); // Zero out 0+real 0-imag 
-#endif
-#ifdef QPX
-            assert(0);
-#endif
-            return ret;
-        }
-        friend inline void timesMinusI( vComplexF &ret,const vComplexF &in){
-	  vzero(ret);
-#if defined (AVX1)|| defined (AVX2)
-	  cvec tmp =_mm256_addsub_ps(ret.v,in.v); // r,-i
-          ret.v = _mm256_shuffle_ps(tmp,tmp,_MM_SHUFFLE(2,3,0,1)); //-i,r
-#endif
-#ifdef SSE4
-	  cvec tmp =_mm_addsub_ps(ret.v,in.v); // r,-i
-          ret.v = _mm_shuffle_ps(tmp,tmp,_MM_SHUFFLE(2,3,0,1));
-#endif
-#ifdef AVX512
-          ret.v = _mm512_mask_sub_ps(in.v,0xaaaa,ret.v,in.v); // real -imag 
-	  ret.v = _mm512_swizzle_ps(ret.v, _MM_SWIZ_REG_CDAB);// OK
-#endif
-#ifdef QPX
-            assert(0);
-#endif
-	}
-        friend inline void timesI(vComplexF &ret,const vComplexF &in){
-	  vzero(ret);
-#if defined (AVX1)|| defined (AVX2)
-	  cvec tmp =_mm256_shuffle_ps(in.v,in.v,_MM_SHUFFLE(2,3,0,1));//i,r
-          ret.v    =_mm256_addsub_ps(ret.v,tmp);     //i,-r
-#endif
-#ifdef SSE4
-	  cvec tmp =_mm_shuffle_ps(in.v,in.v,_MM_SHUFFLE(2,3,0,1));
-          ret.v = _mm_addsub_ps(ret.v,tmp); // r,-i
-#endif
-#ifdef AVX512
-          cvec tmp = _mm512_swizzle_ps(in.v, _MM_SWIZ_REG_CDAB);// OK
-	  ret.v    = _mm512_mask_sub_ps(tmp,0xaaaa,ret.v,tmp); // real -imag 
-#endif
-#ifdef QPX
-            assert(0);
-#endif
-	}
-        friend inline vComplexF timesMinusI(const vComplexF &in){
-	  vComplexF ret; 
-	  timesMinusI(ret,in);
-	  return ret;
-	}
-        friend inline vComplexF timesI(const vComplexF &in){
-	  vComplexF ret; 
-	  timesI(ret,in);
-	  return ret;
-	}
-
-        
-        // Unary negation
-        friend inline vComplexF operator -(const vComplexF &r) {
-            vComplexF ret;
-            vzero(ret);
-            ret = ret - r;
-            return ret;
-        }
-        // *=,+=,-= operators
-        inline vComplexF &operator *=(const vComplexF &r) {
-            *this = (*this)*r;
-            return *this;
-        }
-        inline vComplexF &operator +=(const vComplexF &r) {
-            *this = *this+r;
-            return *this;
-        }
-        inline vComplexF &operator -=(const vComplexF &r) {
-            *this = *this-r;
-            return *this;
-        }
-
-      /*
-      friend inline void merge(vComplexF &y,std::vector<ComplexF *> &extracted)
-      {
-	Gmerge<vComplexF,ComplexF >(y,extracted);
-      }
-      friend inline void extract(const vComplexF &y,std::vector<ComplexF *> &extracted)
-      {
-	Gextract<vComplexF,ComplexF>(y,extracted);
-      }
-      friend inline void merge(vComplexF &y,std::vector<ComplexF > &extracted)
-      {
-	Gmerge<vComplexF,ComplexF >(y,extracted);
-      }
-      friend inline void extract(const vComplexF &y,std::vector<ComplexF > &extracted)
-      {
-	Gextract<vComplexF,ComplexF>(y,extracted);
-      }
-      */
-
-    };
-
-    inline vComplexF innerProduct(const vComplexF & l, const vComplexF & r) 
-    {
-      return conjugate(l)*r; 
-    }
-
-    inline void zeroit(vComplexF &z){ vzero(z);}
-
-    inline vComplexF outerProduct(const vComplexF &l, const vComplexF& r)
-    {
-        return l*r;
-    }
-    inline vComplexF trace(const vComplexF &arg){
-        return arg;
-    }
-
-
-}
-
-#endif
diff --git a/lib/simd/Old/Grid_vInteger.h b/lib/simd/Old/Grid_vInteger.h
deleted file mode 100644
index ee4a3633..00000000
--- a/lib/simd/Old/Grid_vInteger.h
+++ /dev/null
@@ -1,259 +0,0 @@
-#ifndef GRID_VINTEGER_H
-#define GRID_VINTEGER_H
-
-namespace Grid {
-
-
-  class vInteger {
-    protected:
-
-    public:
-
-      ivec v;
-
-	typedef ivec     vector_type;
-	typedef Integer scalar_type;
-
-        vInteger(){};
-        vInteger & operator = (const Zero & z){
-            vzero(*this);
-            return (*this);
-        }
-        vInteger(Integer a){
-	  vsplat(*this,a);
-	};
-        ////////////////////////////////////
-        // Arithmetic operator overloads +,-,*
-        ////////////////////////////////////
-        friend inline vInteger operator + ( vInteger a,  vInteger b)
-        {
-	  vInteger ret;
-#if defined (AVX1) 
-	  __m128i a0,a1;
-	  __m128i b0,b1;
-	  a0 = _mm256_extractf128_si256(a.v,0);
-	  b0 = _mm256_extractf128_si256(b.v,0);
-	  a1 = _mm256_extractf128_si256(a.v,1);
-	  b1 = _mm256_extractf128_si256(b.v,1);
-	  a0 = _mm_add_epi32(a0,b0);
-	  a1 = _mm_add_epi32(a1,b1);
-	  ret.v = _mm256_set_m128i(a1,a0);
-#endif
-#if defined (AVX2)
-            ret.v = _mm256_add_epi32(a.v,b.v);
-#endif
-#ifdef SSE4
-            ret.v = _mm_add_epi32(a.v,b.v);
-#endif
-#ifdef AVX512
-            ret.v = _mm512_add_epi32(a.v,b.v);
-#endif
-#ifdef QPX
-            // Implement as array of ints is only option
-#error
-#endif
-            return ret;
-        };
-        
-        friend inline vInteger operator - ( vInteger a, vInteger b)
-        {
-            vInteger ret;
-#if defined (AVX1) 
-	  __m128i a0,a1;
-	  __m128i b0,b1;
-	  a0 = _mm256_extractf128_si256(a.v,0);
-	  b0 = _mm256_extractf128_si256(b.v,0);
-	  a1 = _mm256_extractf128_si256(a.v,1);
-	  b1 = _mm256_extractf128_si256(b.v,1);
-	  a0 = _mm_sub_epi32(a0,b0);
-	  a1 = _mm_sub_epi32(a1,b1);
-	  ret.v = _mm256_set_m128i(a1,a0);
-#endif
-#if defined (AVX2)
-            ret.v = _mm256_sub_epi32(a.v,b.v);
-#endif
-#ifdef SSE4
-            ret.v = _mm_sub_epi32(a.v,b.v);
-#endif
-#ifdef AVX512
-            ret.v = _mm512_sub_epi32(a.v,b.v);
-#endif
-#ifdef QPX
-            // Implement as array of ints is only option
-#error
-#endif
-            return ret;
-        };
-
-        friend inline vInteger operator * ( vInteger a, vInteger b)
-        {
-            vInteger ret;
-#if defined (AVX1) 
-	  __m128i a0,a1;
-	  __m128i b0,b1;
-	  a0 = _mm256_extractf128_si256(a.v,0);
-	  b0 = _mm256_extractf128_si256(b.v,0);
-	  a1 = _mm256_extractf128_si256(a.v,1);
-	  b1 = _mm256_extractf128_si256(b.v,1);
-	  a0 = _mm_mul_epi32(a0,b0);
-	  a1 = _mm_mul_epi32(a1,b1);
-	  ret.v = _mm256_set_m128i(a1,a0);
-#endif
-#if defined (AVX2)
-            ret.v = _mm256_mul_epi32(a.v,b.v);
-#endif
-#ifdef SSE4
-            ret.v = _mm_mul_epi32(a.v,b.v);
-#endif
-#ifdef AVX512
-	    //            ret.v = _mm512_mul_epi32(a.v,b.v);
-	    ret.v = _mm512_mullo_epi32(a.v,b.v);
-#endif
-#ifdef QPX
-            // Implement as array of ints is only option
-#error
-#endif
-            return ret;
-        };
-        
-        ///////////////////////////////////////////////
-        // mult, sub, add, adj,conjugate, mac functions
-        ///////////////////////////////////////////////
-        friend inline void mult(vInteger * __restrict__ y,const vInteger * __restrict__ l,const vInteger *__restrict__ r) {*y = (*l) * (*r);}
-        friend inline void sub (vInteger * __restrict__ y,const vInteger * __restrict__ l,const vInteger *__restrict__ r) {*y = (*l) - (*r);}
-        friend inline void add (vInteger * __restrict__ y,const vInteger * __restrict__ l,const vInteger *__restrict__ r) {*y = (*l) + (*r);}
-        friend inline void mac (vInteger &y,const vInteger a,const vInteger x){
-            y = a*x+y;
-	}
-        
-        //////////////////////////////////
-        // Initialise to 1,0,i
-        //////////////////////////////////
-        friend inline void vone (vInteger &ret){vsplat(ret,1);}
-        friend inline void vzero(vInteger &ret){vsplat(ret,0);}
-        friend inline void vtrue (vInteger &ret){vsplat(ret,0xFFFFFFFF);}
-        friend inline void vfalse(vInteger &ret){vsplat(ret,0);}
-
-        
-        /////////////////////////////////////////////////////
-        // Broadcast a value across Nsimd copies.
-        /////////////////////////////////////////////////////
-        friend inline void vsplat(vInteger &ret,scalar_type a){
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_set1_epi32(a);
-#endif
-#ifdef SSE4
-            ret.v = _mm_set1_epi32(a);
-#endif
-#ifdef AVX512
-            ret.v = _mm512_set1_epi32(a);
-#endif
-#ifdef QPX
-#error
-#endif
-        }
-        friend inline void vset(vInteger &ret,scalar_type *a){
-#if defined (AVX1)|| defined (AVX2)
-	  ret.v = _mm256_set_epi32(a[7],a[6],a[5],a[4],a[3],a[2],a[1],a[0]);
-#endif
-#ifdef SSE4
-	  ret.v = _mm_set_epi32(a[0],a[1],a[2],a[3]);
-#endif
-#ifdef AVX512
-	  ret.v = _mm512_set_epi32( a[15],a[14],a[13],a[12],a[11],a[10],a[9],a[8],
-                                   a[7],a[6],a[5],a[4],a[3],a[2],a[1],a[0]);
-#endif
-#ifdef QPX
-#error
-#endif
-	}
-
-	friend inline void vstore(const vInteger &ret, Integer *a){
-#if defined (AVX1)|| defined (AVX2)
-	  _mm256_store_si256((__m256i*)a,ret.v);
-#endif
-#ifdef SSE4
-	  _mm_store_si128((__m128i *)a,ret.v);
-#endif
-#ifdef AVX512
-	  _mm512_store_si512(a,ret.v);
-#endif
-#ifdef QPX
-	  assert(0);
-#endif
-        }
-
-	friend inline void vstream(vInteger & out,const vInteger &in){
-	  out=in;
-        }
-
-        friend inline void prefetch(const vInteger &v)
-        {
-            _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
-        }
-
-        // Unary negation
-        friend inline vInteger operator -(const vInteger &r) {
-            vInteger ret;
-            vzero(ret);
-            ret = ret - r;
-            return ret;
-        }
-       friend inline Integer Reduce(const vInteger & in)
-       {
-	 // unimplemented
-	 assert(0);
-       }
-        // *=,+=,-= operators
-        inline vInteger &operator *=(const vInteger &r) {
-            *this = (*this)*r;
-            return *this;
-        }
-        inline vInteger &operator +=(const vInteger &r) {
-            *this = *this+r;
-            return *this;
-        }
-        inline vInteger &operator -=(const vInteger &r) {
-            *this = *this-r;
-            return *this;
-        }
-
-      friend inline void permute(vInteger &y,const vInteger b,int perm)
-      {
-	Gpermute<vInteger>(y,b,perm);
-      }
-      /*
-      friend inline void merge(vInteger &y,std::vector<Integer *> &extracted)
-      {
-	Gmerge<vInteger,Integer>(y,extracted);
-      }
-      friend inline void extract(const vInteger &y,std::vector<Integer *> &extracted)
-      {
-	Gextract<vInteger,Integer>(y,extracted);
-      }
-      friend inline void merge(vInteger &y,std::vector<Integer> &extracted)
-      {
-	Gmerge<vInteger,Integer>(y,extracted);
-      }
-      friend inline void extract(const vInteger &y,std::vector<Integer> &extracted)
-      {
-	Gextract<vInteger,Integer>(y,extracted);
-      }
-      */
-
-    public:
-        static inline int Nsimd(void) { return sizeof(ivec)/sizeof(Integer);}
-  };
-
-    inline vInteger localInnerProduct(const vInteger & l, const vInteger & r) { return l*r; }
-
-    inline void  zeroit(vInteger &z){ vzero(z);}
-
-    inline vInteger outerProduct(const vInteger &l, const vInteger& r)
-    {
-        return l*r;
-    }
-    
-}
-
-#endif
diff --git a/lib/simd/Old/Grid_vRealD.h b/lib/simd/Old/Grid_vRealD.h
deleted file mode 100644
index a32e579e..00000000
--- a/lib/simd/Old/Grid_vRealD.h
+++ /dev/null
@@ -1,276 +0,0 @@
-#ifndef GRID_VREALD_H
-#define GRID_VREALD_H
-
-namespace Grid {
-    class vRealD  {
-    public:
-        dvec v; // dvec is double precision vector
-
-    public:
-	typedef dvec  vector_type;
-	typedef RealD scalar_type;
-
-        vRealD()=default;
-        vRealD(RealD a){
-	  vsplat(*this,a);
-	};
-        vRealD(Zero &zero){
-	  zeroit(*this);
-	}
-        vRealD & operator = ( Zero & z){
-	  vzero(*this);
-	  return (*this);
-        }
-
-        friend inline void mult(vRealD * __restrict__ y,const vRealD * __restrict__ l,const vRealD *__restrict__ r) {*y = (*l) * (*r);}
-        friend inline void sub (vRealD * __restrict__ y,const vRealD * __restrict__ l,const vRealD *__restrict__ r) {*y = (*l) - (*r);}
-        friend inline void add (vRealD * __restrict__ y,const vRealD * __restrict__ l,const vRealD *__restrict__ r) {*y = (*l) + (*r);}
-        friend inline vRealD adj(const vRealD &in) { return in; }
-        friend inline vRealD conjugate(const vRealD &in){ return in; }
-        
-        friend inline void mac (vRealD &y,const vRealD a,const vRealD x){
-#if defined (AVX1) || defined (SSE4)
-            y = a*x+y;
-#endif
-#ifdef AVX2     // AVX 2 introduced FMA support. FMA4 eliminates a copy, but AVX only has FMA3
-            // accelerates multiply accumulate, but not general multiply add
-            y.v = _mm256_fmadd_pd(a.v,x.v,y.v);
-#endif
-#ifdef AVX512
-            // here precision of vector are still single
-            y.v = _mm512_fmadd_pd(a.v,x.v,y.v);
-#endif
-#ifdef QPX
-            y.v = vec_madd(a.v,x.v,y.v);
-#endif
-        }
-        //////////////////////////////////
-        // Initialise to 1,0
-        //////////////////////////////////
-        friend inline void vone (vRealD &ret){ vsplat(ret,1.0);}
-        friend inline void vzero(vRealD &ret){ vsplat(ret,0.0);}
-        
-        
-        ////////////////////////////////////
-        // Arithmetic operator overloads +,-,*
-        ////////////////////////////////////
-        friend inline vRealD operator + (vRealD a, vRealD b)
-        {
-            vRealD ret;
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_add_pd(a.v,b.v);
-#endif
-#ifdef SSE4
-            ret.v = _mm_add_pd(a.v,b.v);
-#endif
-#ifdef AVX512
-            ret.v = _mm512_add_pd(a.v,b.v);
-#endif
-#ifdef QPX
-            ret.v = vec_add(a.v,b.v);
-#endif
-            return ret;
-        };
-        friend inline vRealD operator - (vRealD a, vRealD b)
-        {
-            vRealD ret;
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_sub_pd(a.v,b.v);
-#endif
-#ifdef SSE4
-            ret.v = _mm_sub_pd(a.v,b.v);
-#endif
-#ifdef AVX512
-            ret.v = _mm512_sub_pd(a.v,b.v);
-#endif
-#ifdef QPX
-            ret.v = vec_sub(a.v,b.v);
-#endif
-            return ret;
-        };
-        
-        friend inline vRealD operator * (vRealD a, vRealD b)
-        {
-            vRealD ret;
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_mul_pd(a.v,b.v);
-#endif
-#ifdef SSE4
-            ret.v = _mm_mul_pd(a.v,b.v);
-#endif
-#ifdef AVX512
-            ret.v = _mm512_mul_pd(a.v,b.v);
-#endif
-#ifdef QPX
-            ret.v = vec_mul(a.v,b.v);
-#endif
-            return ret;
-        };
-
-	////////////////////////////////////////////////////////////////////
-	// General permute; assumes vector length is same across 
-	// all subtypes; may not be a good assumption, but could
-	// add the vector width as a template param for BG/Q for example
-	////////////////////////////////////////////////////////////////////
-
-	friend inline void permute(vRealD &y,vRealD b,int perm)
-	{
-	  Gpermute<vRealD>(y,b,perm);
-	}
-	/*
-	friend inline void merge(vRealD &y,std::vector<RealD *> &extracted)
-	{
-	  Gmerge<vRealD,RealD >(y,extracted);
-	}
-	friend inline void extract(const vRealD &y,std::vector<RealD *> &extracted)
-	{
-	  Gextract<vRealD,RealD>(y,extracted);
-	}
-	friend inline void merge(vRealD &y,std::vector<RealD > &extracted)
-	{
-	  Gmerge<vRealD,RealD >(y,extracted);
-	}
-	friend inline void extract(const vRealD &y,std::vector<RealD > &extracted)
-	{
-	  Gextract<vRealD,RealD>(y,extracted);
-	}
-	*/
-        
-        friend inline void vsplat(vRealD &ret,double a){
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_set_pd(a,a,a,a);
-#endif
-#ifdef SSE4
-            ret.v = _mm_set_pd(a,a);
-#endif
-#ifdef AVX512
-            ret.v = _mm512_set1_pd(a);
-#endif
-#ifdef QPX
-            ret.v = {a,a,a,a};
-#endif
-        }
-	friend inline void vset(vRealD &ret, double *a){
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_set_pd(a[3],a[2],a[1],a[0]);
-#endif
-#ifdef SSE4
-            ret.v = _mm_set_pd(a[1],a[0]);
-#endif
-#ifdef AVX512
-            ret.v = _mm512_set_pd(a[7],a[6],a[5],a[4],a[3],a[2],a[1],a[0]);
-            // Note v has a0 a1 a2 a3 a4 a5 a6 a7
-#endif
-#ifdef QPX
-            ret.v = {a[0],a[1],a[2],a[3]};
-#endif
-	}
-
-	friend inline void vstore(const vRealD &ret, double *a){
-#if defined (AVX1)|| defined (AVX2)
-            _mm256_store_pd(a,ret.v);
-#endif
-#ifdef SSE4
-            _mm_store_pd(a,ret.v);
-#endif
-#ifdef AVX512
-            _mm512_store_pd(a,ret.v);
-            // Note v has a7 a6 a5ba4 a3 a2 a1 a0
-#endif
-#ifdef QPX
-	    assert(0);
-#endif
-	}
-        friend inline void vstream(vRealD &out,const vRealD &in){
-#if defined (AVX1)|| defined (AVX2)
-	  _mm256_stream_pd((double *)&out.v,in.v);
-#endif
-#ifdef SSE4
-	  _mm_stream_pd((double *)&out.v,in.v);
-#endif
-#ifdef AVX512
-	  _mm512_storenrngo_pd((double *)&out.v,in.v);
-	  //Note v has a3 a2 a1 a0
-#endif
-#ifdef QPX
-	  assert(0);
-#endif
-	}
-        friend inline void prefetch(const vRealD &v)
-        {
-            _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
-        }
-        // Unary negation
-        friend inline vRealD operator -(const vRealD &r) {
-            vRealD ret;
-            vzero(ret);
-            ret = ret - r;
-            return ret;
-        }
-
-       friend inline RealD Reduce(const vRealD & in)
-       {
-	 vRealD v1,v2;
-	 union { 
-	   dvec v;
-	   double f[sizeof(dvec)/sizeof(double)];
-	 } conv;
-#ifdef SSE4
-	 permute(v1,in,0); // sse 128; paired real double
-	 v1=v1+in;
-#endif
-#if defined(AVX1) || defined (AVX2)
-	 permute(v1,in,0); // avx 256; quad double
-	 v1=v1+in;
-	 permute(v2,v1,1); 
-	 v1=v1+v2;
-#endif
-#ifdef AVX512
-	 permute(v1,in,0); // avx 512; octo-double
-	 v1=v1+in;
-	 permute(v2,v1,1); 
-	 v1=v1+v2;
-	 permute(v2,v1,2); 
-	 v1=v1+v2;
-#endif
-#ifdef QPX
-#endif
-	 conv.v=v1.v;
-	 return conv.f[0];
-       }
-
-        // *=,+=,-= operators
-        inline vRealD &operator *=(const vRealD &r) {
-            *this = (*this)*r;
-            return *this;
-        }
-        inline vRealD &operator +=(const vRealD &r) {
-            *this = *this+r;
-            return *this;
-        }
-        inline vRealD &operator -=(const vRealD &r) {
-            *this = *this-r;
-            return *this;
-        }
-
-    public:
-        static int Nsimd(void) { return sizeof(dvec)/sizeof(double);}
-    };
-
-    inline vRealD innerProduct(const vRealD & l, const vRealD & r) { return conjugate(l)*r; }
-    inline void zeroit(vRealD &z){ vzero(z);}
-
-    inline vRealD outerProduct(const vRealD &l, const vRealD& r)
-    {
-        return l*r;
-    }
-    inline vRealD trace(const vRealD &arg){
-        return arg;
-    }
-    inline vRealD real(const vRealD &arg){
-        return arg;
-    }
-
-
-}
-#endif
diff --git a/lib/simd/Old/Grid_vRealF.h b/lib/simd/Old/Grid_vRealF.h
deleted file mode 100644
index 05bfa2f5..00000000
--- a/lib/simd/Old/Grid_vRealF.h
+++ /dev/null
@@ -1,313 +0,0 @@
-#ifndef GRID_VREALF_H
-#define GRID_VREALF_H
-
-
-namespace Grid {
-    class vRealF  {
-    public:
-        fvec v;
-
-    public:
-	typedef fvec  vector_type;
-	typedef RealF scalar_type;
-
-        vRealF()=default;
-        vRealF(RealF a){
-	  vsplat(*this,a);
-	};
-        vRealF(Zero &zero){
-	  zeroit(*this);
-	}
-        vRealF & operator = ( Zero & z){
-	  vzero(*this);
-	  return (*this);
-        }
-        ////////////////////////////////////
-        // Arithmetic operator overloads +,-,*
-        ////////////////////////////////////
-        friend inline vRealF operator + ( vRealF a,  vRealF b)
-        {
-            vRealF ret;
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_add_ps(a.v,b.v);
-#endif
-#ifdef SSE4
-            ret.v = _mm_add_ps(a.v,b.v);
-#endif
-#ifdef AVX512
-            ret.v = _mm512_add_ps(a.v,b.v);
-#endif
-#ifdef QPX
-            vector4double aa,bb,cc;
-            aa = vec_lda(0,(float *)&a);
-            bb = vec_lda(0,(float *)&b);
-            cc = vec_add(aa,bb);
-            vec_sta(cc,0,(float *)&ret.v);
-#endif
-            return ret;
-        };
-        
-        friend inline vRealF operator - ( vRealF a, vRealF b)
-        {
-            vRealF ret;
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_sub_ps(a.v,b.v);
-#endif
-#ifdef SSE4
-            ret.v = _mm_sub_ps(a.v,b.v);
-#endif
-#ifdef AVX512
-            ret.v = _mm512_sub_ps(a.v,b.v);
-#endif
-#ifdef QPX
-            vector4double aa,bb,cc;
-            aa = vec_lda(0,(float *)&a);
-            bb = vec_lda(0,(float *)&b);
-            cc = vec_sub(aa,bb);
-            vec_sta(cc,0,(float *)&ret.v);
-#endif
-            return ret;
-        };
-
-        friend inline vRealF operator * ( vRealF a, vRealF b)
-        {
-            vRealF ret;
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_mul_ps(a.v,b.v);
-#endif
-#ifdef SSE4
-            ret.v = _mm_mul_ps(a.v,b.v);
-#endif
-#ifdef AVX512
-            ret.v = _mm512_mul_ps(a.v,b.v);
-#endif
-#ifdef QPX
-            vector4double aa,bb,cc; // QPX single we are forced to load as this promotes single mem->double regs.
-            aa = vec_lda(0,(float *)&a);
-            bb = vec_lda(0,(float *)&b);
-            cc = vec_mul(aa,bb);
-            vec_sta(cc,0,(float *)&ret.v);
-#endif
-            return ret;
-        };
-        
-        ///////////////////////////////////////////////
-        // mult, sub, add, adj,conjugate, mac functions
-        ///////////////////////////////////////////////
-        friend inline void mult(vRealF * __restrict__ y,const vRealF * __restrict__ l,const vRealF *__restrict__ r) {*y = (*l) * (*r);}
-        friend inline void sub (vRealF * __restrict__ y,const vRealF * __restrict__ l,const vRealF *__restrict__ r) {*y = (*l) - (*r);}
-        friend inline void add (vRealF * __restrict__ y,const vRealF * __restrict__ l,const vRealF *__restrict__ r) {*y = (*l) + (*r);}
-        friend inline vRealF adj(const vRealF &in) { return in; }
-        friend inline vRealF conjugate(const vRealF &in){ return in; }
-
-        friend inline void mac (vRealF &y,const vRealF a,const vRealF x){
-#if defined (AVX1) || defined (SSE4)
-            y = a*x+y;
-#endif
-#ifdef AVX2     // AVX 2 introduced FMA support. FMA4 eliminates a copy, but AVX only has FMA3
-            // accelerates multiply accumulate, but not general multiply add
-            y.v = _mm256_fmadd_ps(a.v,x.v,y.v);
-#endif
-#ifdef AVX512
-            y.v = _mm512_fmadd_ps(a.v,x.v,y.v);
-#endif
-#ifdef QPX
-            vector4double aa,xx,yy; // QPX single we are forced to load as this promotes single mem->double regs.
-            aa = vec_lda(0,(float *)&a.v);
-            xx = vec_lda(0,(float *)&x.v);
-            yy = vec_lda(0,(float *)&y.v);
-            yy = vec_madd(aa,xx,yy);
-            vec_sta(yy,0,(float *)&y.v);
-#endif
-        }
-        
-        //////////////////////////////////
-        // Initialise to 1,0,i
-        //////////////////////////////////
-        friend inline void vone (vRealF &ret){vsplat(ret,1.0);}
-        friend inline void vzero(vRealF &ret){vsplat(ret,0.0);}
-
-
-	////////////////////////////////////////////////////////////////////
-	// General permute; assumes vector length is same across 
-	// all subtypes; may not be a good assumption, but could
-	// add the vector width as a template param for BG/Q for example
-	////////////////////////////////////////////////////////////////////
-
-	friend inline void permute(vRealF &y,vRealF b,int perm)
-	{
-	  Gpermute<vRealF>(y,b,perm);
-	}
-	/*
-	friend inline void merge(vRealF &y,std::vector<RealF *> &extracted)
-	{
-	  Gmerge<vRealF,RealF >(y,extracted);
-	}
-	friend inline void extract(const vRealF &y,std::vector<RealF *> &extracted)
-	{
-	  Gextract<vRealF,RealF>(y,extracted);
-	}
-	friend inline void merge(vRealF &y,std::vector<RealF> &extracted)
-	{
-	  Gmerge<vRealF,RealF >(y,extracted);
-	}
-	friend inline void extract(const vRealF &y,std::vector<RealF> &extracted)
-	{
-	  Gextract<vRealF,RealF>(y,extracted);
-	}
-	*/
-        
-        /////////////////////////////////////////////////////
-        // Broadcast a value across Nsimd copies.
-        /////////////////////////////////////////////////////
-        friend inline void vsplat(vRealF &ret,float a){
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_set_ps(a,a,a,a,a,a,a,a);
-#endif
-#ifdef SSE4
-            ret.v = _mm_set_ps(a,a,a,a);
-#endif
-#ifdef AVX512
-            //ret.v = _mm512_set_ps(a,a,a,a,a,a,a,a,a,a,a,a,a,a,a,a);
-            ret.v = _mm512_set1_ps(a);
-#endif
-#ifdef QPX
-            ret.v = {a,a,a,a};
-#endif
-        }
-
-
-        friend inline void vset(vRealF &ret, float *a){
-#if defined (AVX1)|| defined (AVX2)
-            ret.v = _mm256_set_ps(a[7],a[6],a[5],a[4],a[3],a[2],a[1],a[0]);
-#endif
-#ifdef SSE4
-            ret.v = _mm_set_ps(a[3],a[2],a[1],a[0]);
-#endif
-#ifdef AVX512
-            ret.v = _mm512_set_ps( a[15],a[14],a[13],a[12],a[11],a[10],a[9],a[8],
-                                   a[7],a[6],a[5],a[4],a[3],a[2],a[1],a[0]);
-            // Note v has a0 a1 a2 a3 a4 a5 a6 a7
-#endif
-#ifdef QPX
-            ret.v = {a[0],a[1],a[2],a[3],a[4],a[5],a[6],a[7]};
-#endif
-	}
-
-	////////////////////////////////////////////////////////////////////////
-	// FIXME:  gonna remove these load/store, get, set, prefetch
-	////////////////////////////////////////////////////////////////////////
-friend inline void vstore(const vRealF &ret, float *a){
-#if defined (AVX1)|| defined (AVX2)
-	_mm256_store_ps(a,ret.v);
-#endif
-#ifdef SSE4
-	_mm_store_ps(a,ret.v);
-#endif
-#ifdef AVX512
-	_mm512_store_ps(a,ret.v);
-	// Note v has a7 a6 a5ba4 a3 a2 a1 a0
-#endif
-#ifdef QPX
-	assert(0);
-#endif
-        }
-        friend inline void vstream(vRealF &out,const vRealF &in){
-#if defined (AVX1)|| defined (AVX2)
-	  _mm256_stream_ps((float *)&out.v,in.v);
-#endif
-#ifdef SSE4
-	  _mm_stream_ps((float *)&out.v,in.v);
-#endif
-#ifdef AVX512
-	  _mm512_storenrngo_ps((float *)&out.v,in.v);
-	  //	  _mm512_stream_ps((float *)&out.v,in.v);
-	  //Note v has a3 a2 a1 a0
-#endif
-#ifdef QPX
-	  assert(0);
-#endif
-	}
-
-
-        friend inline void prefetch(const vRealF &v)
-        {
-            _mm_prefetch((const char*)&v.v,_MM_HINT_T0);
-        }
-        // Unary negation
-        friend inline vRealF operator -(const vRealF &r) {
-            vRealF ret;
-            vzero(ret);
-            ret = ret - r;
-            return ret;
-        }
-       friend inline RealF Reduce(const vRealF & in)
-       {
-	 vRealF v1,v2;
-	 union { 
-	   fvec v;
-	   float f[sizeof(fvec)/sizeof(double)];
-	 } conv;
-#ifdef SSE4
-	 permute(v1,in,0); // sse 128; quad single
-	 v1=v1+in;
-	 permute(v2,v1,1); 
-	 v1=v1+v2;
-#endif
-#if defined(AVX1) || defined (AVX2)
-	 permute(v1,in,0); // avx 256; octo-double
-	 v1=v1+in;
-	 permute(v2,v1,1); 
-	 v1=v1+v2;
-	 permute(v2,v1,2); 
-	 v1=v1+v2;
-#endif
-#ifdef AVX512
-	 permute(v1,in,0); // avx 256; octo-double
-	 v1=v1+in;
-	 permute(v2,v1,1); 
-	 v1=v1+v2;
-	 permute(v2,v1,2); 
-	 v1=v1+v2;
-	 permute(v2,v1,3); 
-	 v1=v1+v2;
-#endif
-#ifdef QPX
-#endif
-	 conv.v=v1.v;
-	 return conv.f[0];
-        }
-
-        // *=,+=,-= operators
-        inline vRealF &operator *=(const vRealF &r) {
-            *this = (*this)*r;
-            return *this;
-        }
-        inline vRealF &operator +=(const vRealF &r) {
-            *this = *this+r;
-            return *this;
-        }
-        inline vRealF &operator -=(const vRealF &r) {
-            *this = *this-r;
-            return *this;
-        }
-    public:
-        static inline int Nsimd(void) { return sizeof(fvec)/sizeof(float);}
-    };
-    inline vRealF innerProduct(const vRealF & l, const vRealF & r) { return conjugate(l)*r; }
-    inline void  zeroit(vRealF &z){ vzero(z);}
-
-    inline vRealF outerProduct(const vRealF &l, const vRealF& r)
-    {
-        return l*r;
-    }
-    inline vRealF trace(const vRealF &arg){
-        return arg;
-    }
-    inline vRealF real(const vRealF &arg){
-        return arg;
-    }
-
-    
-}
-#endif
diff --git a/lib/tensors/Tensor_Ta.h b/lib/tensors/Tensor_Ta.h
index 3a903364..f06dbf04 100644
--- a/lib/tensors/Tensor_Ta.h
+++ b/lib/tensors/Tensor_Ta.h
@@ -29,15 +29,87 @@ namespace Grid {
     {
       iMatrix<vtype,N> ret(arg);
       double factor = (1/(double)N);
-      for(int c1=0;c1<N;c1++){
-	for(int c2=0;c2<N;c2++){
-	  ret._internal[c1][c2]= (ret._internal[c1][c2] - adj(arg._internal[c2][c1]));
-	  ret._internal[c1][c2] *= 0.5;
-	}}
-      //ret = (ret - adj(arg))*0.5;
+      ret = (ret - adj(arg))*0.5;
       ret -= trace(ret)*factor;
       return ret;
     }
 
+
+  /////////////////////////////////////////////// 
+  // ProjectOnGroup function for scalar, vector, matrix
+  /////////////////////////////////////////////// 
+
+
+  template<class vtype> inline iScalar<vtype> ProjectOnGroup(const iScalar<vtype>&r)
+    {
+      iScalar<vtype> ret;
+      ret._internal = ProjectOnGroup(r._internal);
+      return ret;
+    }
+  template<class vtype,int N> inline iVector<vtype,N> ProjectOnGroup(const iVector<vtype,N>&r)
+    {
+      iVector<vtype,N> ret;
+      for(int i=0;i<N;i++){
+        ret._internal[i] = ProjectOnGroup(r._internal[i]);
+      }
+      return ret;
+    }
+  template<class vtype,int N, typename std::enable_if< GridTypeMapper<vtype>::TensorLevel == 0 >::type * =nullptr> 
+    inline iMatrix<vtype,N> ProjectOnGroup(const iMatrix<vtype,N> &arg)
+    {
+      // need a check for the group type?
+      iMatrix<vtype,N> ret(arg);
+      double nrm;
+      for(int c1=0;c1<N;c1++){
+	nrm = 0.0; 
+	for(int c2=0;c2<N;c2++)
+	  nrm = real(innerProduct(ret._internal[c1][c2],ret._internal[c1][c2]));
+	nrm = 1.0/sqrt(nrm);
+	for(int c2=0;c2<N;c2++)
+	  ret._internal[c1][c2]*= nrm;
+      
+	for (int b=c1+1; b<N; ++b){
+	  decltype(ret._internal[b][b]*ret._internal[b][b]) pr = 0.0;
+	  for(int c=0; c<N; ++c)
+	    pr += ret._internal[c1][c]*ret._internal[b][c];
+	  
+	  for(int c=0; c<N; ++c){
+	    ret._internal[b][c] -= pr * ret._internal[c1][c];
+	  }
+	}
+	  
+      }
+      return ret;
+    }
+
+
+  /////////////////////////////////////////////// 
+  // Exponentiate function for scalar, vector, matrix
+  /////////////////////////////////////////////// 
+
+
+  template<class vtype> inline iScalar<vtype> Exponentiate(const iScalar<vtype>&r, double alpha,  int Nexp)
+    {
+      iScalar<vtype> ret;
+      ret._internal = Exponentiate(r._internal, alpha, Nexp);
+      return ret;
+    }
+
+
+  template<class vtype,int N, typename std::enable_if< GridTypeMapper<vtype>::TensorLevel == 0 >::type * =nullptr> 
+    inline iMatrix<vtype,N> Exponentiate(const iMatrix<vtype,N> &arg, double alpha, int Nexp)
+    {
+      iMatrix<vtype,N> unit(1.0);
+      iMatrix<vtype,N> temp(unit);
+      
+      for(int i=Nexp; i>=1;--i){
+	temp *= alpha/double(i);
+	temp = unit + temp*arg;
+      }
+      return ProjectOnGroup(temp);
+    }
+
+
+
 }
 #endif
diff --git a/tests/InvSqrt.gnu b/tests/InvSqrt.gnu
new file mode 100644
index 00000000..e69de29b
diff --git a/tests/Sqrt.gnu b/tests/Sqrt.gnu
new file mode 100644
index 00000000..ae56ab97
--- /dev/null
+++ b/tests/Sqrt.gnu
@@ -0,0 +1,2 @@
+f(x) = 6.81384+(-2.34645e-06/(x+0.000228091))+(-1.51593e-05/(x+0.00112084))+(-6.89254e-05/(x+0.003496))+(-0.000288983/(x+0.00954309))+(-0.00119277/(x+0.024928))+(-0.0050183/(x+0.0646627))+(-0.0226449/(x+0.171576))+(-0.123767/(x+0.491792))+(-1.1705/(x+1.78667))+(-102.992/(x+18.4866));
+f(x) = 0.14676+(0.00952992/(x+5.40933e-05))+(0.0115952/(x+0.000559699))+(0.0161824/(x+0.00203338))+(0.0243252/(x+0.00582831))+(0.0379533/(x+0.0154649))+(0.060699/(x+0.0401156))+(0.100345/(x+0.104788))+(0.178335/(x+0.286042))+(0.381586/(x+0.892189))+(1.42625/(x+4.38422));
diff --git a/tests/Test_main.cc b/tests/Test_main.cc
index cef2d848..88d69192 100644
--- a/tests/Test_main.cc
+++ b/tests/Test_main.cc
@@ -223,6 +223,9 @@ int main (int argc, char ** argv)
     //TComplex tracecm= trace(cm);      
     //std::cout << cm << "  "<< tracecm << std::endl;
 
+    cm = ProjectOnGroup(cm);
+
+    cm = Exponentiate(cm, 1.0, 10);
 
 //    Foo = Foo+scalar; // LatticeColourMatrix+Scalar
 //    Foo = Foo*scalar; // LatticeColourMatrix*Scalar

From f0530e9f5243bb50c62a521c4ac64fd97ec51635 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 9 Jun 2015 22:36:48 +0100
Subject: [PATCH 294/429] silly change

---
 tests/Test_nersc_io.cc | 1 -
 1 file changed, 1 deletion(-)

diff --git a/tests/Test_nersc_io.cc b/tests/Test_nersc_io.cc
index 403d2cab..f473781e 100644
--- a/tests/Test_nersc_io.cc
+++ b/tests/Test_nersc_io.cc
@@ -25,7 +25,6 @@ int main (int argc, char ** argv)
   std::vector<LatticeColourMatrix> U(4,&Fine);
   
   NerscField header;
-  
   std::string file("./ckpoint_lat.4000");
   readNerscConfiguration(Umu,header,file);
 

From a5b75a095cb69a22fc684a2eee63b4442598a3ed Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 9 Jun 2015 22:37:21 +0100
Subject: [PATCH 295/429] Debugged finally. A silly mistake in permute cost me
 a day of debug.

---
 lib/algorithms/CoarsenedMatrix.h | 104 ++++++++++++++++++++++++-------
 1 file changed, 82 insertions(+), 22 deletions(-)

diff --git a/lib/algorithms/CoarsenedMatrix.h b/lib/algorithms/CoarsenedMatrix.h
index 006da07f..ba417be3 100644
--- a/lib/algorithms/CoarsenedMatrix.h
+++ b/lib/algorithms/CoarsenedMatrix.h
@@ -11,7 +11,6 @@ namespace Grid {
     int npoint;
     std::vector<int> directions   ;
     std::vector<int> displacements;
-    std::vector<int> opdirs;
 
     // FIXME -- don't like xposing the operator directions
     // as different to the geometrical dirs
@@ -26,18 +25,22 @@ namespace Grid {
       npoint = 2*_d+1;
       directions.resize(npoint);
       displacements.resize(npoint);
-      opdirs.resize(npoint);
       for(int d=0;d<_d;d++){
 	directions[2*d  ] = d+base;
 	directions[2*d+1] = d+base;
-	opdirs[2*d  ] = d;
-	opdirs[2*d+1] = d;
 	displacements[2*d  ] = +1;
 	displacements[2*d+1] = -1;
       }
       directions   [2*_d]=0;
       displacements[2*_d]=0;
-      opdirs       [2*_d]=0;
+      
+      //// report back
+      std::cout<<"directions    :";
+      for(int d=0;d<npoint;d++) std::cout<< directions[d]<< " ";
+      std::cout <<std::endl;
+      std::cout<<"displacements :";
+      for(int d=0;d<npoint;d++) std::cout<< displacements[d]<< " ";
+      std::cout <<std::endl;
     }
   
     /*
@@ -64,12 +67,12 @@ namespace Grid {
   // Fine Object == (per site) type of fine field
   // nbasis      == number of deflation vectors
   template<class Fobj,class CComplex,int nbasis>
-  class CoarsenedMatrix : public SparseMatrixBase<Lattice<iVector<vComplex,nbasis > > >  {
+  class CoarsenedMatrix : public SparseMatrixBase<Lattice<iVector<CComplex,nbasis > > >  {
   public:
     
-    typedef iVector<vComplex,nbasis > siteVector;
-    typedef Lattice<iVector<vComplex,nbasis > > CoarseVector;
-    typedef Lattice<iMatrix<vComplex,nbasis > > CoarseMatrix;
+    typedef iVector<CComplex,nbasis >             siteVector;
+    typedef Lattice<siteVector>                 CoarseVector;
+    typedef Lattice<iMatrix<CComplex,nbasis > > CoarseMatrix;
 
     typedef Lattice< CComplex >   CoarseScalar; // used for inner products on fine field
     typedef Lattice<Fobj >        FineField;
@@ -82,7 +85,6 @@ namespace Grid {
     CartesianStencil Stencil; 
 
     std::vector<CoarseMatrix> A;
-    std::vector<CoarseMatrix> Aslow;
 
     std::vector<siteVector,alignedAllocator<siteVector> >   comm_buf;
       
@@ -91,25 +93,28 @@ namespace Grid {
     ///////////////////////
     GridBase * Grid(void)         { return _grid; };   // this is all the linalg routines need to know
 
-    RealD M    (const CoarseVector &in, CoarseVector &out){
-      
+    RealD M (const CoarseVector &in, CoarseVector &out){
+
+      conformable(_grid,in._grid);
+      conformable(in._grid,out._grid);
+
       SimpleCompressor<siteVector> compressor;
       Stencil.HaloExchange(in,comm_buf,compressor);
 
-PARALLEL_FOR_LOOP
+      //PARALLEL_FOR_LOOP
       for(int ss=0;ss<Grid()->oSites();ss++){
         siteVector res = zero;
-	siteVector tmp;
 	siteVector nbr;
+	int offset,local,perm,ptype;
 
-	int offset,local,perm;
 	for(int point=0;point<geom.npoint;point++){
 	  offset = Stencil._offsets [point][ss];
 	  local  = Stencil._is_local[point][ss];
-	  perm   = Stencil._permute[point][ss];
+	  perm   = Stencil._permute [point][ss];
+	  ptype  = Stencil._permute_type[point];
 	  
 	  if(local&&perm) { 
-	    permute(nbr,in._odata[offset],perm);
+	    permute(nbr,in._odata[offset],ptype);
 	  } else if(local) { 
 	    nbr = in._odata[offset];
 	  } else {
@@ -125,6 +130,7 @@ PARALLEL_FOR_LOOP
     RealD Mdag (const CoarseVector &in, CoarseVector &out){ 
       return M(in,out);
     };
+
     // Defer support for further coarsening for now
     void Mdiag    (const CoarseVector &in,  CoarseVector &out){};
     void Mdir     (const CoarseVector &in,  CoarseVector &out,int dir, int disp){};
@@ -134,8 +140,7 @@ PARALLEL_FOR_LOOP
       _grid(&CoarseGrid),
       geom(CoarseGrid._ndimension),
       Stencil(&CoarseGrid,geom.npoint,Even,geom.directions,geom.displacements),
-	A(geom.npoint,&CoarseGrid),
-	Aslow(geom.npoint,&CoarseGrid)
+      A(geom.npoint,&CoarseGrid)
     {
       comm_buf.resize(Stencil._unified_buffer_size);
     };
@@ -176,8 +181,7 @@ PARALLEL_FOR_LOOP
 	for(int p=0;p<geom.npoint;p++){ 
 
 	  int dir   = geom.directions[p];
-	  int opdir = geom.opdirs[p];
-	  int disp= geom.displacements[p];
+	  int disp  = geom.displacements[p];
 
 	  int block=(FineGrid->_rdimensions[dir])/(Grid()->_rdimensions[dir]);
 
@@ -187,7 +191,7 @@ PARALLEL_FOR_LOOP
 	    linop.OpDiag(phi,Mphi);
 	  }
 	  else  {
-	    linop.OpDir(phi,Mphi,opdir,disp); 
+	    linop.OpDir(phi,Mphi,dir,disp); 
 	  }
 
 	  ////////////////////////////////////////////////////////////////////////
@@ -240,6 +244,62 @@ PARALLEL_FOR_LOOP
       std::cout<< iProj <<std::endl;
       std::cout<<"Computed Coarse Operator"<<std::endl;
 #endif
+      AssertHermitian();
+      //      ForceHermitian();
+      //      ForceDiagonal();
+    }
+    void ForceDiagonal(void) {
+
+
+      std::cout<<"**************************************************"<<std::endl;
+      std::cout<<"****   Forcing coarse operator to be diagonal ****"<<std::endl;
+      std::cout<<"**************************************************"<<std::endl;
+      for(int p=0;p<8;p++){
+	A[p]=zero;
+      }
+
+      GridParallelRNG  RNG(Grid()); RNG.SeedRandomDevice();
+      Lattice<iScalar<CComplex> > val(Grid()); random(RNG,val);
+
+      Complex one(1.0);
+
+      iMatrix<Complex,nbasis> ident;  ident=one;
+
+      val = val*adj(val);
+      val = val + 1.0;
+
+      A[8] = val*ident;
+
+      //      for(int s=0;s<Grid()->oSites();s++) {
+      //	A[8]._odata[s]=val._odata[s];
+      //      }
+    }
+    void ForceHermitian(void) {
+      for(int d=0;d<4;d++){
+	int dd=d+1;
+	A[2*d] = adj(Cshift(A[2*d+1],dd,1));
+      }
+      A[8] = 0.5*(A[8] + adj(A[8]));
+    }
+    void AssertHermitian(void) {
+      CoarseMatrix AA    (Grid());
+      CoarseMatrix AAc   (Grid());
+      CoarseMatrix Diff  (Grid());
+      for(int d=0;d<4;d++){
+	
+	int dd=d+1;
+	AAc = Cshift(A[2*d+1],dd,1);
+	AA  = A[2*d];
+	
+	Diff = AA - adj(AAc);
+
+	std::cout<<"Norm diff dim "<<d<<" "<< norm2(Diff)<<std::endl;
+	std::cout<<"Norm dim "<<d<<" "<< norm2(AA)<<std::endl;
+	  
+      }
+      Diff = A[8] - adj(A[8]);
+      std::cout<<"Norm diff local "<< norm2(Diff)<<std::endl;
+      std::cout<<"Norm local "<< norm2(A[8])<<std::endl;
     }
     
   };

From e19391ef7ac8bc674316a9a0c963aa3e72cf448c Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 9 Jun 2015 22:38:13 +0100
Subject: [PATCH 296/429] Starting to use

---
 lib/algorithms/approx/Chebyshev.h | 16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)

diff --git a/lib/algorithms/approx/Chebyshev.h b/lib/algorithms/approx/Chebyshev.h
index 5c7741e4..0dad7179 100644
--- a/lib/algorithms/approx/Chebyshev.h
+++ b/lib/algorithms/approx/Chebyshev.h
@@ -42,6 +42,14 @@ namespace Grid {
     double lo;
 
   public:
+    void csv(std::ostream &out){
+      for (double x=lo; x<hi; x+=(hi-lo)/1000) {
+	double f = approx(x);
+	out<< x<<" "<<f<<std::endl;
+      }
+      return;
+    }
+
     Chebyshev(double _lo,double _hi,int _order, double (* func)(double) ){
       lo=_lo;
       hi=_hi;
@@ -52,16 +60,16 @@ namespace Grid {
       for(int j=0;j<order;j++){
 	double s=0;
 	for(int k=0;k<order;k++){
-	  double y=cos(M_PI*(k+0.5)/order);
+	  double y=std::cos(M_PI*(k+0.5)/order);
 	  double x=0.5*(y*(hi-lo)+(hi+lo));
 	  double f=func(x);
-	  s=s+f*cos( j*M_PI*(k+0.5)/order );
+	  s=s+f*std::cos( j*M_PI*(k+0.5)/order );
 	}
 	Coeffs[j] = s * 2.0/order;
       }
     };
 
-    double Evaluate(double x) // Convenience for plotting the approximation
+    double approx(double x) // Convenience for plotting the approximation
     {
       double Tn;
       double Tnm;
@@ -91,7 +99,7 @@ namespace Grid {
     void   PlotApprox(std::ostream &out) {
       out<<"Polynomial approx ["<<lo<<","<<hi<<"]"<<std::endl;
       for(double x=lo;x<hi;x+=(hi-lo)/50.0){
-	out <<x<<"\t"<<Evaluate(x)<<std::endl;
+	out <<x<<"\t"<<approx(x)<<std::endl;
       }
     };
 

From 4a105403652ec806dfc67d7300c2ad06084a0397 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 9 Jun 2015 22:39:13 +0100
Subject: [PATCH 297/429] Got this sorted with the promote working in a test

---
 lib/lattice/Lattice_transfer.h | 9 ++++-----
 1 file changed, 4 insertions(+), 5 deletions(-)

diff --git a/lib/lattice/Lattice_transfer.h b/lib/lattice/Lattice_transfer.h
index 12993680..fb6b39f1 100644
--- a/lib/lattice/Lattice_transfer.h
+++ b/lib/lattice/Lattice_transfer.h
@@ -94,7 +94,7 @@ inline void blockProject(Lattice<iVector<CComplex,nbasis > > &coarseData,
     for(int i=0;i<nbasis;i++) {
       
       coarseData._odata[sc](i)=coarseData._odata[sc](i)
-	+ TensorRemove( innerProduct(Basis[i]._odata[sf],fineData._odata[sf]));
+	+ innerProduct(Basis[i]._odata[sf],fineData._odata[sf]);
 
     }
   }
@@ -266,7 +266,7 @@ inline void blockPromote(const Lattice<iVector<CComplex,nbasis > > &coarseData,
   assert( nbasis == Basis.size() );
   subdivides(coarse,fine); 
   for(int i=0;i<nbasis;i++){
-    conformable(Basis,fineData);
+    conformable(Basis[i]._grid,fine);
   }
 
   std::vector<int>  block_r      (_ndimension);
@@ -287,9 +287,8 @@ inline void blockPromote(const Lattice<iVector<CComplex,nbasis > > &coarseData,
     GridBase::IndexFromCoor(coor_c,sc,coarse->_rdimensions);
 
     for(int i=0;i<nbasis;i++) {
-
-      if(i==0) fineData._odata[sf]=                    coarseData._odata[sc][i]*Basis[i]._odata[sf];
-      else     fineData._odata[sf]=fineData._odata[sf]+coarseData._odata[sc][i]*Basis[i]._odata[sf];
+      if(i==0) fineData._odata[sf]=coarseData._odata[sc](i) * Basis[i]._odata[sf];
+      else     fineData._odata[sf]=fineData._odata[sf]+coarseData._odata[sc](i)*Basis[i]._odata[sf];
 
     }
   }

From 2e43d8fe92496a77da3e267895621d1e0e9f8af9 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 9 Jun 2015 22:39:37 +0100
Subject: [PATCH 298/429] Prettier reporting

---
 scripts/linecount | 32 +++++++++++++++++++++++++-------
 1 file changed, 25 insertions(+), 7 deletions(-)

diff --git a/scripts/linecount b/scripts/linecount
index 0be67142..4463ce84 100755
--- a/scripts/linecount
+++ b/scripts/linecount
@@ -1,12 +1,30 @@
 #!/bin/sh 
+if [ $# -eq 1 ]
+then
+  wc -l lib/*.h lib/*/*.h lib/*/*/*.h lib/*.cc lib/*/*.cc lib/*/*/*.cc tests/*.cc benchmarks/*.cc lib/*/*/*/*.cc lib/*/*/*/*.h >& tmp.fil
+  count=`grep total tmp.fil`
+  echo $count " in Grid library"
+else 
+  wc -l lib/*.h lib/*/*.h lib/*/*/*.h lib/*.cc lib/*/*.cc lib/*/*/*.cc tests/*.cc benchmarks/*.cc lib/*/*/*/*.cc lib/*/*/*/*.h 
+fi
 
-wc -l lib/*.h lib/*/*.h lib/*/*/*.h lib/*.cc lib/*/*.cc lib/*/*/*.cc tests/*.cc benchmarks/*.cc lib/*/*/*/*.cc lib/*/*/*/*.h
+wc -l lib/*.h lib/*/*.h lib/*/*/*.h lib/*.cc lib/*/*.cc lib/*/*/*.cc lib/*/*/*/*.cc lib/*/*/*/*.h >& tmp.fil
+count=`grep total tmp.fil`
+echo  $count " in lib"
 
-echo "lib"
-wc -l lib/*.h lib/*/*.h lib/*/*/*.h lib/*.cc lib/*/*.cc lib/*/*/*.cc lib/*/*/*/*.cc lib/*/*/*/*.h | grep total
+for sdir in `ls -F lib/| grep /`
+do
+wc -l lib/${sdir}/*.h lib/${sdir}/*/*.h lib/${sdir}/*.cc lib/${sdir}/*/*.cc lib/${sdir}/*/*/*.cc lib/${sdir}/*/*/*.h >& tmp.fil
+count=`grep total tmp.fil`
+echo $count " in lib/${sdir}" 
+done
 
-echo "tests"
-wc -l tests/*.cc | grep total
+wc -l tests/*.cc | grep total >& tmp.fil
+count=`grep total tmp.fil`
+echo  $count " in tests"
 
-echo "benchmarks"
-wc -l benchmarks/*.cc | grep total
+wc -l benchmarks/*.cc | grep total >& tmp.fil
+count=`grep total tmp.fil`
+echo  $count " in benchmarks"
+
+rm tmp.fil

From 930463a226f2c1f551dec812cf02a0fbd65ed379 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 9 Jun 2015 22:40:12 +0100
Subject: [PATCH 299/429] Files update

---
 tests/Make.inc | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/tests/Make.inc b/tests/Make.inc
index 2b132fd0..64ba4768 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,5 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_gamma Test_main Test_multishift_sqrt Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
+bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cg Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_gamma Test_main Test_multishift_sqrt Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
 
 
 Test_GaugeAction_SOURCES=Test_GaugeAction.cc
@@ -18,6 +18,14 @@ Test_cayley_even_odd_SOURCES=Test_cayley_even_odd.cc
 Test_cayley_even_odd_LDADD=-lGrid
 
 
+Test_cayley_ldop_cg_SOURCES=Test_cayley_ldop_cg.cc
+Test_cayley_ldop_cg_LDADD=-lGrid
+
+
+Test_cayley_ldop_cr_SOURCES=Test_cayley_ldop_cr.cc
+Test_cayley_ldop_cr_LDADD=-lGrid
+
+
 Test_cf_coarsen_support_SOURCES=Test_cf_coarsen_support.cc
 Test_cf_coarsen_support_LDADD=-lGrid
 

From c7152c520aef4b0b472146635483cc08418deced Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 9 Jun 2015 22:40:58 +0100
Subject: [PATCH 300/429] g5 and g5R5 hermitian are now differentiated

---
 lib/qcd/action/fermion/g5HermitianLinop.h | 47 ++++++++++++++++++++++-
 1 file changed, 45 insertions(+), 2 deletions(-)

diff --git a/lib/qcd/action/fermion/g5HermitianLinop.h b/lib/qcd/action/fermion/g5HermitianLinop.h
index 5b3411ec..e5b691d8 100644
--- a/lib/qcd/action/fermion/g5HermitianLinop.h
+++ b/lib/qcd/action/fermion/g5HermitianLinop.h
@@ -6,10 +6,10 @@ namespace Grid {
 // Wrap an already herm matrix
 ////////////////////////////////////////////////////////////////////
 template<class Matrix,class Field>
-class Gamma5HermitianLinearOperator : public LinearOperatorBase<Field> {
+class Gamma5R5HermitianLinearOperator : public LinearOperatorBase<Field> {
   Matrix &_Mat;
 public:
-  Gamma5HermitianLinearOperator(Matrix &Mat): _Mat(Mat){};
+  Gamma5R5HermitianLinearOperator(Matrix &Mat): _Mat(Mat){};
   void Op     (const Field &in, Field &out){
     HermOp(in,out);
   }
@@ -45,5 +45,48 @@ public:
   }
 };
 
+
+template<class Matrix,class Field>
+class Gamma5HermitianLinearOperator : public LinearOperatorBase<Field> {
+  Matrix &_Mat;
+  Gamma g5;
+public:
+    Gamma5HermitianLinearOperator(Matrix &Mat): _Mat(Mat), g5(Gamma::Gamma5) {};
+  void Op     (const Field &in, Field &out){
+    HermOp(in,out);
+  }
+  void AdjOp     (const Field &in, Field &out){
+    HermOp(in,out);
+  }
+  void OpDiag (const Field &in, Field &out) {
+    Field tmp(in._grid);
+    _Mat.Mdiag(in,tmp);
+    out=g5*tmp;
+  }
+  void OpDir  (const Field &in, Field &out,int dir,int disp) {
+    Field tmp(in._grid);
+    _Mat.Mdir(in,tmp,dir,disp);
+    out=g5*tmp;
+  }
+
+  void HermOpAndNorm(const Field &in, Field &out,RealD &n1,RealD &n2){
+
+    HermOp(in,out);
+    
+    ComplexD dot;
+    dot= innerProduct(in,out);
+    n1=real(dot);
+    
+    dot = innerProduct(out,out);
+    n2=real(dot);
+  }
+  void HermOp(const Field &in, Field &out){
+    Field tmp(in._grid);
+    _Mat.M(in,tmp);
+    out=g5*tmp;
+  }
+};
+
+
 }}
 #endif

From 971d22498341b34da97fec53404e31f3785c7146 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 9 Jun 2015 22:41:27 +0100
Subject: [PATCH 301/429] G5R5 update

---
 tests/Test_dwf_cr_unprec.cc | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tests/Test_dwf_cr_unprec.cc b/tests/Test_dwf_cr_unprec.cc
index cffa726c..6cd2b32a 100644
--- a/tests/Test_dwf_cr_unprec.cc
+++ b/tests/Test_dwf_cr_unprec.cc
@@ -55,7 +55,7 @@ int main (int argc, char ** argv)
   MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermOp(Ddwf);
   MCR(HermOp,src,result);
 
-  Gamma5HermitianLinearOperator<DomainWallFermion,LatticeFermion> g5HermOp(Ddwf);
+  Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> g5HermOp(Ddwf);
   MCR(g5HermOp,src,result);
 
 

From 6abbd35d819268a15b3b5a1544b07b5d783d0b6d Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 9 Jun 2015 22:41:59 +0100
Subject: [PATCH 302/429] 5d OpDir direction interface refers to the 5d dims,
 not 4d to present a sensible and consistent external interface.

---
 lib/qcd/action/fermion/WilsonFermion5D.cc | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/lib/qcd/action/fermion/WilsonFermion5D.cc b/lib/qcd/action/fermion/WilsonFermion5D.cc
index 7d9ffe22..7bbeeef9 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.cc
+++ b/lib/qcd/action/fermion/WilsonFermion5D.cc
@@ -72,6 +72,7 @@ namespace QCD {
 }
 void WilsonFermion5D::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu)
 {
+  assert(GaugeGrid()->_ndimension==4);
   conformable(Uds._grid,GaugeGrid());
   conformable(Umu._grid,GaugeGrid());
   LatticeColourMatrix U(GaugeGrid());
@@ -82,8 +83,10 @@ void WilsonFermion5D::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGau
     pokeIndex<LorentzIndex>(Uds,U,mu+4);
   }
 }
-void WilsonFermion5D::DhopDir(const LatticeFermion &in, LatticeFermion &out,int dir,int disp)
+void WilsonFermion5D::DhopDir(const LatticeFermion &in, LatticeFermion &out,int dir5,int disp)
 {
+  int dir = dir5-1; // Maps to the ordering above in "directions" that is passed to stencil
+                    // we drop off the innermost fifth dimension
   assert( (disp==1)||(disp==-1) );
   assert( (dir>=0)&&(dir<4) ); //must do x,y,z or t;
 

From 850054d162999e72f1c46d856552a9745e76569a Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 9 Jun 2015 22:43:10 +0100
Subject: [PATCH 303/429] Some extra tests

---
 tests/Test_cayley_coarsen_support.cc | 99 ++++++++++++++++++++++------
 1 file changed, 80 insertions(+), 19 deletions(-)

diff --git a/tests/Test_cayley_coarsen_support.cc b/tests/Test_cayley_coarsen_support.cc
index 064f5f13..b68a5cc3 100644
--- a/tests/Test_cayley_coarsen_support.cc
+++ b/tests/Test_cayley_coarsen_support.cc
@@ -21,9 +21,7 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-
-
-  const int Ls=4;
+  const int Ls=8;
 
   GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
   GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
@@ -41,8 +39,10 @@ int main (int argc, char ** argv)
 
   std::vector<int> seeds4({1,2,3,4});
   std::vector<int> seeds5({5,6,7,8});
-  GridParallelRNG          RNG5(FGrid);  RNG5.SeedFixedIntegers(seeds5);
-  GridParallelRNG          RNG4(UGrid);  RNG4.SeedFixedIntegers(seeds4);
+  std::vector<int> cseeds({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);   RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);   RNG4.SeedFixedIntegers(seeds4);
+  GridParallelRNG          CRNG(Coarse5d);CRNG.SeedFixedIntegers(cseeds);
 
   LatticeFermion    src(FGrid); random(RNG5,src);
   LatticeFermion result(FGrid); result=zero;
@@ -50,31 +50,33 @@ int main (int argc, char ** argv)
   LatticeFermion    tmp(FGrid);
   LatticeFermion    err(FGrid);
   LatticeGaugeField Umu(UGrid); random(RNG4,Umu);
+
 #if 0
   std::vector<LatticeColourMatrix> U(4,UGrid);
   Umu=zero;
   Complex cone(1.0,0.0);
   for(int nn=0;nn<Nd;nn++){
     if(1) {
-      if (nn!=0) { U[nn]=zero; std::cout << "zeroing gauge field in dir "<<nn<<std::endl; }
-      else       { U[nn] = cone;std::cout << "unit gauge field in dir "<<nn<<std::endl; }
+      if (nn>2) { U[nn]=zero; std::cout << "zeroing gauge field in dir "<<nn<<std::endl; }
+      else      { U[nn]=cone; std::cout << "unit gauge field in dir "<<nn<<std::endl; }
     }
     pokeIndex<LorentzIndex>(Umu,U[nn],nn);
   }
 #endif  
+
   RealD mass=0.5;
   RealD M5=1.8;
 
   DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
-  Gamma5HermitianLinearOperator<DomainWallFermion,LatticeFermion> HermIndefOp(Ddwf);
+  Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> HermIndefOp(Ddwf);
 
   HermIndefOp.Op(src,ref);
   HermIndefOp.OpDiag(src,result);
   
   for(int d=0;d<4;d++){
-    HermIndefOp.OpDir(src,tmp,d,+1); result=result+tmp; 
+    HermIndefOp.OpDir(src,tmp,d+1,+1); result=result+tmp; 
     std::cout<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
-    HermIndefOp.OpDir(src,tmp,d,-1); result=result+tmp;
+    HermIndefOp.OpDir(src,tmp,d+1,-1); result=result+tmp;
     std::cout<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
   }
   err = result-ref;
@@ -82,32 +84,91 @@ int main (int argc, char ** argv)
 
   const int nbasis = 2;
   std::vector<LatticeFermion> subspace(nbasis,FGrid);
-  
+  LatticeFermion prom(FGrid);
+
   for(int b=0;b<nbasis;b++){
     random(RNG5,subspace[b]);
   }
   std::cout << "Computed randoms"<< std::endl;
 
-  CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis> LittleDiracOp(*Coarse5d);
+  typedef CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis> LittleDiracOperator;
+  typedef LittleDiracOperator::CoarseVector CoarseVector;
+
+  LittleDiracOperator LittleDiracOp(*Coarse5d);
 
   LittleDiracOp.CoarsenOperator(FGrid,HermIndefOp,subspace);
-
-  typedef   Lattice<iVector<vComplex,nbasis > > coarse_vec;
   
-  coarse_vec c_src (Coarse5d);  c_src= zero;
-  coarse_vec c_res (Coarse5d);
+  CoarseVector c_src (Coarse5d);
+  CoarseVector c_res (Coarse5d);
+  CoarseVector c_proj(Coarse5d);
   
-  Complex one(1.0);
-  c_src = one;  // 1 in every element for vector 1.
-
   //  TODO
   // -- promote from subspace, check we get the vector we wanted
   // -- apply ldop; check we get the same as inner product of M times big vec
   // -- pick blocks one by one. Evaluate matrix elements.
+  Complex one(1.0);
+  c_src = one;  // 1 in every element for vector 1.
+  
+  blockPromote(c_src,err,subspace);
+
+
+  prom=zero;
+  for(int b=0;b<nbasis;b++){
+    prom=prom+subspace[b];
+  }
+  err=err-prom; 
+  std::cout<<"Promoted back from subspace err "<<norm2(err)<<std::endl;
+
+  HermIndefOp.HermOp(prom,tmp);
+  blockProject(c_proj,tmp,subspace);
+
+  LittleDiracOp.M(c_src,c_res);
+
+  c_proj = c_proj - c_res;
+  std::cout<<"Representation of ldop within subspace "<<norm2(c_proj)<<std::endl;
 
   std::cout << "Multiplying by LittleDiracOp "<< std::endl;
   LittleDiracOp.M(c_src,c_res);
 
+  std::cout<<"Testing hermiticity explicitly by inspecting matrix elements"<<std::endl;
+  LittleDiracOp.AssertHermitian();
+
+  std::cout << "Testing Hermiticity stochastically "<< std::endl;
+  CoarseVector phi(Coarse5d);
+  CoarseVector chi(Coarse5d);
+  CoarseVector Aphi(Coarse5d);
+  CoarseVector Achi(Coarse5d);
+
+  random(CRNG,phi);
+  random(CRNG,chi);
+
+
+  std::cout<<"Made randoms"<<std::endl;
+
+  LittleDiracOp.M(phi,Aphi);
+  LittleDiracOp.Mdag(chi,Achi);
+
+  ComplexD pAc = innerProduct(chi,Aphi);
+  ComplexD cAp = innerProduct(phi,Achi);
+  ComplexD cAc = innerProduct(chi,Achi);
+  ComplexD pAp = innerProduct(phi,Aphi);
+
+  std::cout<< "pAc "<<pAc<<" cAp "<< cAp<< " diff "<<pAc-adj(cAp)<<std::endl;
+
+  std::cout<< "pAp "<<pAp<<" cAc "<< cAc<<"Should be real"<< std::endl;
+
+  std::cout<<"Testing linearity"<<std::endl;
+  CoarseVector PhiPlusChi(Coarse5d);
+  CoarseVector APhiPlusChi(Coarse5d);
+  CoarseVector linerr(Coarse5d);
+  PhiPlusChi = phi+chi;
+  LittleDiracOp.M(PhiPlusChi,APhiPlusChi);
+
+  linerr= APhiPlusChi-Aphi;
+  linerr= linerr-Achi;
+  std::cout<<"**Diff "<<norm2(linerr)<<std::endl;
+
+
   std::cout << "Done "<< std::endl;
   Grid_finalize();
 }

From 60a96e3d0ddf584e933db294a8d99177b43d2768 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 9 Jun 2015 22:43:41 +0100
Subject: [PATCH 304/429] Remove extra layers of checks now it works

---
 lib/algorithms/CoarsenedMatrix.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/algorithms/CoarsenedMatrix.h b/lib/algorithms/CoarsenedMatrix.h
index ba417be3..96a40fe7 100644
--- a/lib/algorithms/CoarsenedMatrix.h
+++ b/lib/algorithms/CoarsenedMatrix.h
@@ -244,7 +244,7 @@ namespace Grid {
       std::cout<< iProj <<std::endl;
       std::cout<<"Computed Coarse Operator"<<std::endl;
 #endif
-      AssertHermitian();
+      //      AssertHermitian();
       //      ForceHermitian();
       //      ForceDiagonal();
     }

From d50cf43e5eb40f5896f1ede0923ec5ba0443293d Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 9 Jun 2015 22:50:45 +0100
Subject: [PATCH 305/429] Solver converges

---
 tests/Test_cayley_ldop_cg.cc | 136 +++++++++++++++++++++++++++++++++++
 1 file changed, 136 insertions(+)
 create mode 100644 tests/Test_cayley_ldop_cg.cc

diff --git a/tests/Test_cayley_ldop_cg.cc b/tests/Test_cayley_ldop_cg.cc
new file mode 100644
index 00000000..b146002d
--- /dev/null
+++ b/tests/Test_cayley_ldop_cg.cc
@@ -0,0 +1,136 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+double lowpass(double x)
+{
+  return pow(x*x+1.0,-2);
+}
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  Chebyshev<LatticeFermion> filter(-150.0, 150.0,16, lowpass);
+  ofstream csv(std::string("filter.dat"),std::ios::out|std::ios::trunc);
+  filter.csv(csv);
+  csv.close();
+
+  const int Ls=8;
+
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+  // Construct a coarsened grid
+  std::vector<int> clatt = GridDefaultLatt();
+  for(int d=0;d<clatt.size();d++){
+    clatt[d] = clatt[d]/2;
+  }
+  GridCartesian *Coarse4d =  SpaceTimeGrid::makeFourDimGrid(clatt, GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());;
+  GridCartesian *Coarse5d =  SpaceTimeGrid::makeFiveDimGrid(1,Coarse4d);
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  std::vector<int> cseeds({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);   RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);   RNG4.SeedFixedIntegers(seeds4);
+  GridParallelRNG          CRNG(Coarse5d);CRNG.SeedFixedIntegers(cseeds);
+
+  LatticeFermion    src(FGrid); gaussian(RNG5,src);
+  LatticeFermion result(FGrid); result=zero;
+  LatticeFermion    ref(FGrid); ref=zero;
+  LatticeFermion    tmp(FGrid);
+  LatticeFermion    err(FGrid);
+  LatticeGaugeField Umu(UGrid); gaussian(RNG4,Umu);
+
+
+  //random(RNG4,Umu);
+  NerscField header;
+  std::string file("./ckpoint_lat.400");
+  readNerscConfiguration(Umu,header,file);
+
+#if 0
+  LatticeColourMatrix U(UGrid);
+  U=zero;
+  for(int nn=0;nn<Nd;nn++){
+    if (nn>2) {
+      pokeIndex<LorentzIndex>(Umu,U,nn);
+    }
+  }
+#endif  
+  RealD mass=0.1;
+  RealD M5=1.0;
+
+  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> HermIndefOp(Ddwf);
+
+  const int nbasis = 8;
+  std::vector<LatticeFermion> subspace(nbasis,FGrid);
+  LatticeFermion noise(FGrid);
+  LatticeFermion ms(FGrid);
+  for(int b=0;b<nbasis;b++){
+    Gamma g5(Gamma::Gamma5);
+    gaussian(RNG5,noise);
+    RealD scale = pow(norm2(noise),-0.5); 
+    noise=noise*scale;
+
+    HermIndefOp.HermOp(noise,ms); std::cout << "Noise    "<<b<<" Ms "<<norm2(ms)<< " "<< norm2(noise)<<std::endl;
+
+
+    //    filter(HermIndefOp,noise,subspace[b]);
+    // inverse iteration
+    MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
+    ConjugateGradient<LatticeFermion> CG(1.0e-5,10000);
+
+    for(int i=0;i<4;i++){
+
+      CG(HermDefOp,noise,subspace[b]);
+      noise = subspace[b];
+
+      scale = pow(norm2(noise),-0.5); 
+      noise=noise*scale;
+      HermDefOp.HermOp(noise,ms); std::cout << "filt    "<<b<<" <u|H|u> "<<norm2(ms)<< " "<< norm2(noise)<<std::endl;
+    }
+
+    subspace[b] = noise;
+    HermIndefOp.HermOp(subspace[b],ms); std::cout << "Filtered "<<b<<" Ms "<<norm2(ms)<< " "<<norm2(subspace[b]) <<std::endl;
+  }
+  std::cout << "Computed randoms"<< std::endl;
+
+  typedef CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis> LittleDiracOperator;
+  typedef LittleDiracOperator::CoarseVector CoarseVector;
+
+  LittleDiracOperator LittleDiracOp(*Coarse5d);
+  LittleDiracOp.CoarsenOperator(FGrid,HermIndefOp,subspace);
+  
+  CoarseVector c_src (Coarse5d);
+  CoarseVector c_res (Coarse5d);
+  gaussian(CRNG,c_src);
+  c_res=zero;
+
+  std::cout << "Solving CG on coarse space "<< std::endl;
+  MdagMLinearOperator<LittleDiracOperator,CoarseVector> PosdefLdop(LittleDiracOp);
+  ConjugateGradient<CoarseVector> CG(1.0e-8,10000);
+  CG(PosdefLdop,c_src,c_res);
+
+  std::cout << "Done "<< std::endl;
+  Grid_finalize();
+}

From 62cb914488ad868996b35b6251421ed6886b00c7 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 9 Jun 2015 22:51:02 +0100
Subject: [PATCH 306/429] Solver converges

---
 tests/Test_cayley_ldop_cr.cc | 138 +++++++++++++++++++++++++++++++++++
 1 file changed, 138 insertions(+)
 create mode 100644 tests/Test_cayley_ldop_cr.cc

diff --git a/tests/Test_cayley_ldop_cr.cc b/tests/Test_cayley_ldop_cr.cc
new file mode 100644
index 00000000..afa773b2
--- /dev/null
+++ b/tests/Test_cayley_ldop_cr.cc
@@ -0,0 +1,138 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+double lowpass(double x)
+{
+  return pow(x*x+1.0,-2);
+}
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  Chebyshev<LatticeFermion> filter(-150.0, 150.0,16, lowpass);
+  ofstream csv(std::string("filter.dat"),std::ios::out|std::ios::trunc);
+  filter.csv(csv);
+  csv.close();
+
+  const int Ls=8;
+
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+  // Construct a coarsened grid
+  std::vector<int> clatt = GridDefaultLatt();
+  for(int d=0;d<clatt.size();d++){
+    clatt[d] = clatt[d]/2;
+  }
+  GridCartesian *Coarse4d =  SpaceTimeGrid::makeFourDimGrid(clatt, GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());;
+  GridCartesian *Coarse5d =  SpaceTimeGrid::makeFiveDimGrid(1,Coarse4d);
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  std::vector<int> cseeds({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);   RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);   RNG4.SeedFixedIntegers(seeds4);
+  GridParallelRNG          CRNG(Coarse5d);CRNG.SeedFixedIntegers(cseeds);
+
+  LatticeFermion    src(FGrid); gaussian(RNG5,src);
+  LatticeFermion result(FGrid); result=zero;
+  LatticeFermion    ref(FGrid); ref=zero;
+  LatticeFermion    tmp(FGrid);
+  LatticeFermion    err(FGrid);
+  LatticeGaugeField Umu(UGrid); gaussian(RNG4,Umu);
+
+
+  //random(RNG4,Umu);
+  NerscField header;
+  std::string file("./ckpoint_lat.400");
+  readNerscConfiguration(Umu,header,file);
+
+#if 0
+  LatticeColourMatrix U(UGrid);
+  Complex cone(1.0,0.0);
+  for(int nn=0;nn<Nd;nn++){
+    if (nn==3) { 
+      U=zero; std::cout << "zeroing gauge field in dir "<<nn<<std::endl;
+    //    else       { U[nn]= cone;std::cout << "unit gauge field in dir "<<nn<<std::endl; }
+      pokeIndex<LorentzIndex>(Umu,U,nn);
+    }
+  }
+#endif  
+  RealD mass=0.5;
+  RealD M5=1.8;
+
+  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> HermIndefOp(Ddwf);
+
+  const int nbasis = 8;
+  std::vector<LatticeFermion> subspace(nbasis,FGrid);
+  LatticeFermion noise(FGrid);
+  LatticeFermion ms(FGrid);
+  for(int b=0;b<nbasis;b++){
+    Gamma g5(Gamma::Gamma5);
+    gaussian(RNG5,noise);
+    RealD scale = pow(norm2(noise),-0.5); 
+    noise=noise*scale;
+
+    HermIndefOp.HermOp(noise,ms); std::cout << "Noise    "<<b<<" Ms "<<norm2(ms)<< " "<< norm2(noise)<<std::endl;
+
+
+    //    filter(HermIndefOp,noise,subspace[b]);
+    // inverse iteration
+    MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
+    ConjugateGradient<LatticeFermion> CG(1.0e-5,10000);
+
+    for(int i=0;i<4;i++){
+
+      CG(HermDefOp,noise,subspace[b]);
+      noise = subspace[b];
+
+      scale = pow(norm2(noise),-0.5); 
+      noise=noise*scale;
+      HermDefOp.HermOp(noise,ms); std::cout << "filt    "<<b<<" <u|H|u> "<<norm2(ms)<< " "<< norm2(noise)<<std::endl;
+    }
+
+    subspace[b] = noise;
+    HermIndefOp.HermOp(subspace[b],ms); std::cout << "Filtered "<<b<<" Ms "<<norm2(ms)<< " "<<norm2(subspace[b]) <<std::endl;
+  }
+  std::cout << "Computed randoms"<< std::endl;
+
+  typedef CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis> LittleDiracOperator;
+  typedef LittleDiracOperator::CoarseVector CoarseVector;
+
+  LittleDiracOperator LittleDiracOp(*Coarse5d);
+  LittleDiracOp.CoarsenOperator(FGrid,HermIndefOp,subspace);
+  
+  CoarseVector c_src (Coarse5d);
+  CoarseVector c_res (Coarse5d);
+  gaussian(CRNG,c_src);
+  c_res=zero;
+
+  std::cout << "Solving MCR on coarse space "<< std::endl;
+  HermitianLinearOperator<LittleDiracOperator,CoarseVector> HermIndefLdop(LittleDiracOp);
+  ConjugateResidual<CoarseVector> MCR(1.0e-8,10000);
+  MCR(HermIndefLdop,c_src,c_res);
+
+  std::cout << "Done "<< std::endl;
+  Grid_finalize();
+}

From 965a92ce404fb189d9824174d4fd4ece1b047312 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Wed, 10 Jun 2015 14:16:33 +0900
Subject: [PATCH 307/429] Adding several iMatrix utilities

---
 lib/Make.inc            |  4 ++--
 lib/tensors/Tensor_Ta.h | 48 +++++++++++++++++++++++++++++++++++++++--
 tests/Make.inc          | 10 ++++-----
 tests/Test_main.cc      | 16 +++++++++-----
 4 files changed, 64 insertions(+), 14 deletions(-)

diff --git a/lib/Make.inc b/lib/Make.inc
index c3070f51..7be4eeaa 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./simd/Old/Grid_vComplexD.h ./simd/Old/Grid_vComplexF.h ./simd/Old/Grid_vInteger.h ./simd/Old/Grid_vRealD.h ./simd/Old/Grid_vRealF.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
+HFILES=./Cshift.h ./simd/Grid_avx.h ./simd/Grid_vector_types.h ./simd/Grid_sse4.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_vector_unops.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./Tensors.h ./Algorithms.h ./communicator/Communicator_base.h ./lattice/Lattice_rng.h ./lattice/Lattice_reduction.h ./lattice/Lattice_transfer.h ./lattice/Lattice_unary.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_comparison.h ./lattice/Lattice_overload.h ./lattice/Lattice_reality.h ./lattice/Lattice_local.h ./lattice/Lattice_conformable.h ./lattice/Lattice_where.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_ET.h ./lattice/Lattice_transpose.h ./lattice/Lattice_trace.h ./Stencil.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_poke.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_class.h ./tensors/Tensor_transpose.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_reality.h ./tensors/Tensor_trace.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_outer.h ./tensors/Tensor_inner.h ./tensors/Tensor_traits.h ./tensors/Tensor_Ta.h ./tensors/Tensor_unary.h ./tensors/Tensor_peek.h ./tensors/Tensor_arith.h ./tensors/Tensor_extract_merge.h ./Communicator.h ./Cartesian.h ./parallelIO/NerscIO.h ./qcd/QCD.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/LinalgUtils.h ./qcd/utils/CovariantCshift.h ./qcd/utils/WilsonLoops.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/spin/TwoSpinor.h ./qcd/spin/Dirac.h ./cshift/Cshift_common.h ./cshift/Cshift_none.h ./cshift/Cshift_mpi.h ./Simd.h ./GridConfig.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_red_black.h ./cartesian/Cartesian_full.h ./AlignedAllocator.h ./Lattice.h ./Threads.h ./Grid.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Zolotarev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Remez.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./algorithms/CoarsenedMatrix.h ./stencil/Lebesgue.h
 
-CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
+CCFILES=./qcd/utils/SpaceTimeGrid.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/spin/Dirac.cc ./GridInit.cc ./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/tensors/Tensor_Ta.h b/lib/tensors/Tensor_Ta.h
index f06dbf04..202c6c71 100644
--- a/lib/tensors/Tensor_Ta.h
+++ b/lib/tensors/Tensor_Ta.h
@@ -63,26 +63,70 @@ namespace Grid {
       for(int c1=0;c1<N;c1++){
 	nrm = 0.0; 
 	for(int c2=0;c2<N;c2++)
-	  nrm = real(innerProduct(ret._internal[c1][c2],ret._internal[c1][c2]));
+	  nrm += real(innerProduct(ret._internal[c1][c2],ret._internal[c1][c2]));
 	nrm = 1.0/sqrt(nrm);
+	std::cout << "norm : "<< nrm << "\n";
 	for(int c2=0;c2<N;c2++)
 	  ret._internal[c1][c2]*= nrm;
       
 	for (int b=c1+1; b<N; ++b){
 	  decltype(ret._internal[b][b]*ret._internal[b][b]) pr = 0.0;
 	  for(int c=0; c<N; ++c)
-	    pr += ret._internal[c1][c]*ret._internal[b][c];
+	    pr += conjugate(ret._internal[c1][c])*ret._internal[b][c];
 	  
+	  std::cout << "pr : "<< pr << "\n";
 	  for(int c=0; c<N; ++c){
 	    ret._internal[b][c] -= pr * ret._internal[c1][c];
 	  }
 	}
 	  
       }
+      // assuming the determinant is ok
       return ret;
     }
 
 
+  /////////////////////////////////////////////// 
+  // Determinant function for scalar, vector, matrix
+  /////////////////////////////////////////////// 
+  inline ComplexF Determinant( const ComplexF &arg){    return arg;}
+  inline ComplexD Determinant( const ComplexD &arg){    return arg;}
+  inline RealF Determinant( const RealF &arg){    return arg;}
+  inline RealD Determinant( const RealD &arg){    return arg;}
+
+  template<class vtype> inline auto Determinant(const iScalar<vtype>&r) -> iScalar<decltype(Determinant(r._internal))>
+    {
+      iScalar<decltype(Determinant(r._internal))> ret;
+      ret._internal = Determinant(r._internal);
+      return ret;
+    }
+
+  template<class vtype,int N, typename std::enable_if< GridTypeMapper<vtype>::TensorLevel == 0 >::type * =nullptr> 
+    inline auto Determinant(const iMatrix<vtype,N> &arg)-> iScalar<decltype(Determinant(arg._internal[0][0]))>
+    {
+      iMatrix<vtype,N> ret(arg);
+      iScalar<decltype(Determinant(arg._internal[0][0]))> det = 1.0;
+      /* Conversion of matrix to upper triangular */
+      for(int i = 0; i < N; i++){
+        for(int j = 0; j < N; j++){
+	  if(j>i){
+	    vtype ratio = ret._internal[j][i]/ret._internal[i][i];
+	    for(int k = 0; k < N; k++){
+	      ret._internal[j][k] -= ratio * ret._internal[i][k];
+	    }
+	  }
+        }
+      }      
+
+      for(int i = 0; i < N; i++)
+	det *= ret._internal[i][i];   
+
+      return det;
+    }
+
+
+
+
   /////////////////////////////////////////////// 
   // Exponentiate function for scalar, vector, matrix
   /////////////////////////////////////////////// 
diff --git a/tests/Make.inc b/tests/Make.inc
index 64ba4768..6102bc0c 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,9 +1,5 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cg Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_gamma Test_main Test_multishift_sqrt Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
-
-
-Test_GaugeAction_SOURCES=Test_GaugeAction.cc
-Test_GaugeAction_LDADD=-lGrid
+bin_PROGRAMS = Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cg Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_gamma Test_GaugeAction Test_main Test_multishift_sqrt Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
 
 
 Test_cayley_cg_SOURCES=Test_cayley_cg.cc
@@ -74,6 +70,10 @@ Test_gamma_SOURCES=Test_gamma.cc
 Test_gamma_LDADD=-lGrid
 
 
+Test_GaugeAction_SOURCES=Test_GaugeAction.cc
+Test_GaugeAction_LDADD=-lGrid
+
+
 Test_main_SOURCES=Test_main.cc
 Test_main_LDADD=-lGrid
 
diff --git a/tests/Test_main.cc b/tests/Test_main.cc
index 88d69192..a42c3dd6 100644
--- a/tests/Test_main.cc
+++ b/tests/Test_main.cc
@@ -216,16 +216,22 @@ int main (int argc, char ** argv)
     scm=transposeIndex<1>(scm);
     
    
-    //random(SerialRNG, cm);
+    random(SerialRNG, cm);
+    std::cout << cm << std::endl;
+
+    //cm = Ta(cm);
+    //TComplex tracecm= trace(cm);      
     //std::cout << cm << std::endl;
 
-    cm = Ta(cm);
-    //TComplex tracecm= trace(cm);      
-    //std::cout << cm << "  "<< tracecm << std::endl;
-
     cm = ProjectOnGroup(cm);
+    std::cout << cm << "  " << std::endl;
 
+    TComplex det = Determinant(cm);
+    
+    std::cout << "determinant: " << det <<  std::endl;
     cm = Exponentiate(cm, 1.0, 10);
+    std::cout << cm << "  " << std::endl;
+
 
 //    Foo = Foo+scalar; // LatticeColourMatrix+Scalar
 //    Foo = Foo*scalar; // LatticeColourMatrix*Scalar

From 2f4a0c5de4fc64ca5006e62f809e8efc1dbf4121 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Wed, 10 Jun 2015 11:24:14 +0100
Subject: [PATCH 308/429] Successful generation of general SU N generators This
 class will move into a utils class and gain exponentiation and ta projection
 to support HMC and heatbath

---
 tests/Test_lie_generators.cc | 229 +++++++++++++++++++++++++++++++++++
 1 file changed, 229 insertions(+)
 create mode 100644 tests/Test_lie_generators.cc

diff --git a/tests/Test_lie_generators.cc b/tests/Test_lie_generators.cc
new file mode 100644
index 00000000..b29682a0
--- /dev/null
+++ b/tests/Test_lie_generators.cc
@@ -0,0 +1,229 @@
+#include <Grid.h>
+
+#include <qcd/utils/CovariantCshift.h>
+#include <qcd/utils/WilsonLoops.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+class suN {
+public:
+
+  static int generators(int ncolour)   { return ncolour*ncolour-1; }
+  static int su2subgroups(int ncolour) { return (ncolour*(ncolour-1))/2; }
+  
+  template<typename CComplex,int N> using suNmatrix = iScalar<iScalar<iMatrix<CComplex,N> > > ;
+
+  ////////////////////////////////////////////////////////////////////////
+  // There are N^2-1 generators for SU(N).
+  //
+  // We take a traceless hermitian generator basis as follows
+  //
+  // * Normalisation: trace ta tb = 1/2 delta_ab
+  // 
+  // * Off diagonal
+  //    - pairs of rows i1,i2 behaving like pauli matrices signma_x, sigma_y
+  //     
+  //    - there are (Nc-1-i1) slots for i2 on each row [ x  0  x ]                                   
+  //      direct count off each row                                                                    
+  //
+  //    - Sum of all pairs is Nc(Nc-1)/2: proof arithmetic series
+  //
+  //      (Nc-1) + (Nc-2)+...  1      ==> Nc*(Nc-1)/2 
+  //      1+ 2+          +   + Nc-1                        
+  // 
+  //    - There are 2 x Nc (Nc-1)/ 2 of these = Nc^2 - Nc
+  //
+  //    - We enumerate the row-col pairs.
+  //    - for each row col pair there is a (sigma_x) and a (sigma_y) like generator
+  //
+  //
+  //   t^a_ij = { in 0.. Nc(Nc-1)/2 -1} =>  delta_{i,i1} delta_{j,i2} +  delta_{i,i1} delta_{j,i2}  
+  //   t^a_ij = { in Nc(Nc-1)/2 ... Nc^(Nc-1) -1} =>  i delta_{i,i1} delta_{j,i2} - i delta_{i,i1} delta_{j,i2}  
+  //   
+  // * Diagonal; must be traceless and normalised
+  //   - Sequence is 
+  //   N  (1,-1,0,0...)
+  //   N  (1, 1,-2,0...)
+  //   N  (1, 1, 1,-3,0...)
+  //   N  (1, 1, 1, 1,-4,0...)
+  //
+  //   where 1/2 = N^2 (1+.. m^2)etc.... for the m-th diagonal generator
+  //   NB this gives the famous SU3 result for su2 index 8
+  //
+  //   N= sqrt(1/2 . 1/6 ) = 1/2 . 1/sqrt(3) 
+  //
+  //   ( 1      )
+  //   (    1   ) / sqrt(3) /2  = 1/2 lambda_8
+  //   (      -2)
+  ////////////////////////////////////////////////////////////////////////
+  template<class CComplex,int Ncolour> 
+  static void suNgenerator(int lieIndex,suNmatrix<CComplex,Ncolour> &ta){
+    // map lie index to which type of generator
+    int diagIndex;
+    int su2Index;
+    int sigxy;
+    int NNm1 =  Ncolour*(Ncolour-1);
+    if ( lieIndex>= NNm1 ) {
+      diagIndex = lieIndex -NNm1;
+      suNgeneratorDiagonal(diagIndex,ta);
+      return;
+    }
+    sigxy   = lieIndex&0x1;
+    su2Index= lieIndex>>1;
+    if ( sigxy ) suNgeneratorSigmaY(su2Index,ta);
+    else         suNgeneratorSigmaX(su2Index,ta);
+  }
+  template<class CComplex,int Ncolour> 
+  static void suNgeneratorSigmaX(int su2Index,suNmatrix<CComplex,Ncolour> &ta){
+    ta=zero;
+    int i1,i2;
+    su2SubGroupIndex<Ncolour>(i1,i2,su2Index);
+    ta()()(i1,i2)=1.0;
+    ta()()(i2,i1)=1.0;
+    ta= ta *0.5;
+  }
+  template<class CComplex,int Ncolour> 
+  static void suNgeneratorSigmaY(int su2Index,suNmatrix<CComplex,Ncolour> &ta){
+    ta=zero;
+    Complex i(0.0,1.0);
+    int i1,i2;
+    su2SubGroupIndex<Ncolour>(i1,i2,su2Index);
+    ta()()(i1,i2)=-i;
+    ta()()(i2,i1)= i;
+    ta= ta *0.5;
+  }
+  template<class CComplex,int Ncolour> 
+  static void suNgeneratorDiagonal(int diagIndex,suNmatrix<CComplex,Ncolour> &ta){
+    ta=zero;
+    int trsq=0;
+    int last=diagIndex+1;
+    for(int i=0;i<=diagIndex;i++){
+      ta()()(i,i) = 1.0;
+      trsq++;
+    }
+    ta()()(last,last) = -last;
+    trsq+=last*last;
+    RealD nrm = 1.0/std::sqrt(2.0*trsq);
+    ta = ta *nrm;
+  }
+  ////////////////////////////////////////////////////////////////////////
+  // Map a 
+  //
+  ////////////////////////////////////////////////////////////////////////
+  template<int Ncolour>
+  static void su2SubGroupIndex(int &i1,int &i2,int su2_index){
+
+    assert( (su2_index>=0) && (su2_index< (Ncolour*(Ncolour-1))/2) );
+
+    int spare=su2_index;
+    for(i1=0;spare >= (Ncolour-1-i1);i1++ ){
+      spare = spare - (Ncolour-1-i1);  // remove the Nc-1-i1 terms                                 
+    }
+    i2=i1+1+spare;
+  }
+  template<class CComplex,int Ncolour>
+  static void su2Extract(std::vector<LatticeComplex> &r,const Lattice<suNmatrix<CComplex,Ncolour> > &source, int su2_index)
+  {
+    assert(r.size() == 4); // store in 4 real parts
+    
+    for(int i=0;i<4;i++){
+      conformable(r[i],source);
+    }
+    
+    int i1,i2;    
+    su2SubGroupIndex<Ncolour>(i1,i2,su2_index);
+    
+    /* Compute the b(k) of A_SU(2) = b0 + i sum_k bk sigma_k */ 
+    r[0] = real(source()()(i1,i1) + source()()(i2,i2));
+    r[1] = imag(source()()(i1,i2) + source()()(i2,i1));
+    r[2] = real(source()()(i1,i2) - source()()(i2,i1));
+    r[3] = imag(source()()(i1,i1) - source()()(i2,i2));
+  }
+
+
+  template<int Ncolour> static void printGenerators(void)
+  {
+    for(int gen=0;gen<suN::generators(Ncolour);gen++){
+      suN::suNmatrix<Complex,Ncolour> ta;
+      suN::suNgenerator(gen,ta);
+      std::cout<< "Nc = "<<Ncolour<<" t_"<<gen<<std::endl;
+      std::cout<<ta<<std::endl;
+    }
+  }
+
+  template<int Ncolour> static void testGenerators(void){
+    suNmatrix<Complex,Ncolour> ta;
+    suNmatrix<Complex,Ncolour> tb;
+    std::cout<<"Checking trace ta tb is 0.5 delta_ab"<<std::endl;
+    for(int a=0;a<generators(Ncolour);a++){
+      for(int b=0;b<generators(Ncolour);b++){
+	suNgenerator(a,ta);
+	suNgenerator(b,tb);
+	Complex tr =TensorRemove(trace(ta*tb)); 
+	std::cout<<tr<<" ";
+	if(a==b) assert(abs(tr-Complex(0.5))<1.0e-6);
+	if(a!=b) assert(abs(tr)<1.0e-6);
+      }
+      std::cout<<std::endl;
+    }
+    std::cout<<"Checking hermitian"<<std::endl;
+    for(int a=0;a<generators(Ncolour);a++){
+      suNgenerator(a,ta);
+      std::cout<<a<<" ";
+      assert(norm2(ta-adj(ta))<1.0e-6);
+    }    
+    std::cout<<std::endl;
+
+    std::cout<<"Checking traceless"<<std::endl;
+    for(int a=0;a<generators(Ncolour);a++){
+      suNgenerator(a,ta);
+      Complex tr =TensorRemove(trace(ta)); 
+      std::cout<<a<<" ";
+      assert(abs(tr)<1.0e-6);
+    }    
+    std::cout<<std::endl;
+  }
+
+};
+
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+  std::vector<int> latt_size  ({4,4,4,4});
+
+  GridCartesian     Fine(latt_size,simd_layout,mpi_layout);
+
+  LatticeGaugeField Umu(&Fine);
+
+  std::cout<<"*********************************************"<<std::endl;
+  std::cout<<"* Generators for SU(2)"<<std::endl;
+  std::cout<<"*********************************************"<<std::endl;
+  suN::printGenerators<2>();
+  suN::testGenerators<2>();
+  std::cout<<"*********************************************"<<std::endl;
+  std::cout<<"* Generators for SU(3)"<<std::endl;
+  std::cout<<"*********************************************"<<std::endl;
+  suN::printGenerators<3>();
+  suN::testGenerators<3>();
+  std::cout<<"*********************************************"<<std::endl;
+  std::cout<<"* Generators for SU(4)"<<std::endl;
+  std::cout<<"*********************************************"<<std::endl;
+  suN::printGenerators<4>();
+  suN::testGenerators<4>();
+  std::cout<<"*********************************************"<<std::endl;
+  std::cout<<"* Generators for SU(5)"<<std::endl;
+  std::cout<<"*********************************************"<<std::endl;
+  suN::printGenerators<5>();
+  suN::testGenerators<5>();
+
+  Grid_finalize();
+}
+
+

From 89f961faef68812e35b27a7c3ca3f0df54fafd82 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Wed, 10 Jun 2015 11:26:46 +0100
Subject: [PATCH 309/429] commit file list

---
 lib/Make.inc   | 2 +-
 tests/Make.inc | 6 +++++-
 2 files changed, 6 insertions(+), 2 deletions(-)

diff --git a/lib/Make.inc b/lib/Make.inc
index c3070f51..417c70ba 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./simd/Old/Grid_vComplexD.h ./simd/Old/Grid_vComplexF.h ./simd/Old/Grid_vInteger.h ./simd/Old/Grid_vRealD.h ./simd/Old/Grid_vRealF.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
 
 CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/tests/Make.inc b/tests/Make.inc
index 64ba4768..714878f6 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,5 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cg Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_gamma Test_main Test_multishift_sqrt Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
+bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cg Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_gamma Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
 
 
 Test_GaugeAction_SOURCES=Test_GaugeAction.cc
@@ -74,6 +74,10 @@ Test_gamma_SOURCES=Test_gamma.cc
 Test_gamma_LDADD=-lGrid
 
 
+Test_lie_generators_SOURCES=Test_lie_generators.cc
+Test_lie_generators_LDADD=-lGrid
+
+
 Test_main_SOURCES=Test_main.cc
 Test_main_LDADD=-lGrid
 

From e5d30fe5e29b7bd5a076c39bedd3bff0b2f9c766 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Wed, 10 Jun 2015 11:30:27 +0100
Subject: [PATCH 310/429] Fix compile

---
 tests/Test_GaugeAction.cc | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)

diff --git a/tests/Test_GaugeAction.cc b/tests/Test_GaugeAction.cc
index bbfd8717..fc49cf2e 100644
--- a/tests/Test_GaugeAction.cc
+++ b/tests/Test_GaugeAction.cc
@@ -7,6 +7,26 @@ using namespace std;
 using namespace Grid;
 using namespace Grid::QCD;
 
+
+/* For Metropolis */
+class Metropolis {
+public:
+  GridSerialRNG & sRNG;
+  Metropolis(GridSerialRNG & _sRNG) : sRNG(_sRNG) {};
+  bool AcceptReject(const RealD Delta)
+  {
+    RealD rand;
+
+    if(Delta <=0.0) return true;
+
+    random(sRNG,rand);
+    if(rand <= exp(-Delta))
+      return true;
+    else 
+      return false;
+  }
+};
+
 int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);

From 50c4f416b6ae174d110faf5386e46a70ab3482d2 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 10 Jun 2015 12:25:15 +0100
Subject: [PATCH 311/429] file list

---
 lib/Make.inc   | 2 +-
 tests/Make.inc | 6 +++++-
 2 files changed, 6 insertions(+), 2 deletions(-)

diff --git a/lib/Make.inc b/lib/Make.inc
index 417c70ba..11362e27 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
 
 CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/tests/Make.inc b/tests/Make.inc
index 714878f6..d0d7d001 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,5 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cg Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_gamma Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
+bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cg Test_cayley_ldop_cr Test_cayley_ldop_cr_chebyshev Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_gamma Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
 
 
 Test_GaugeAction_SOURCES=Test_GaugeAction.cc
@@ -26,6 +26,10 @@ Test_cayley_ldop_cr_SOURCES=Test_cayley_ldop_cr.cc
 Test_cayley_ldop_cr_LDADD=-lGrid
 
 
+Test_cayley_ldop_cr_chebyshev_SOURCES=Test_cayley_ldop_cr_chebyshev.cc
+Test_cayley_ldop_cr_chebyshev_LDADD=-lGrid
+
+
 Test_cf_coarsen_support_SOURCES=Test_cf_coarsen_support.cc
 Test_cf_coarsen_support_LDADD=-lGrid
 

From 1cc25837b2aa622494bf468c02e791aabb30cca7 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 10 Jun 2015 12:26:25 +0100
Subject: [PATCH 312/429] Updates to ldop tests

---
 .gitignore                             |   3 +-
 INSTALL                                |   2 +-
 configure                              |  13 +--
 tests/Test_cayley_ldop_cr.cc           |  17 +---
 tests/Test_cayley_ldop_cr_chebyshev.cc | 124 +++++++++++++++++++++++++
 5 files changed, 137 insertions(+), 22 deletions(-)
 create mode 100644 tests/Test_cayley_ldop_cr_chebyshev.cc

diff --git a/.gitignore b/.gitignore
index be00eb30..a68020a3 100644
--- a/.gitignore
+++ b/.gitignore
@@ -6,6 +6,7 @@
 *~
 errs
 *#
+INSTALL
 
 # Precompiled Headers
 *.gch
@@ -78,4 +79,4 @@ config.status
 .Spotlight-V100
 .Trashes
 ehthumbs.db
-Thumbs.db
\ No newline at end of file
+Thumbs.db
diff --git a/INSTALL b/INSTALL
index f812f5a2..80a61507 120000
--- a/INSTALL
+++ b/INSTALL
@@ -1 +1 @@
-/usr/share/automake-1.14/INSTALL
\ No newline at end of file
+/opt/local/share/automake-1.15/INSTALL
\ No newline at end of file
diff --git a/configure b/configure
index 702a52db..07b99339 100755
--- a/configure
+++ b/configure
@@ -2574,7 +2574,7 @@ test -n "$target_alias" &&
     NONENONEs,x,x, &&
   program_prefix=${target_alias}-
 
-am__api_version='1.14'
+am__api_version='1.15'
 
 # Find a good install program.  We prefer a C program (faster),
 # so one script is as good as another.  But avoid the broken or
@@ -2746,8 +2746,8 @@ test "$program_suffix" != NONE &&
 ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
 program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
 
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 
 if test x"${MISSING+set}" != xset; then
   case $am_aux_dir in
@@ -2766,7 +2766,7 @@ else
 $as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
 fi
 
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -3094,8 +3094,8 @@ MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 mkdir_p='$(MKDIR_P)'
 
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
 # Always define AMTAR for backward compatibility.  Yes, it's still used
 # in the wild :-(  We should find a proper way to deprecate it ...
 AMTAR='$${TAR-tar}'
@@ -3154,6 +3154,7 @@ END
 fi
 
 
+
 ac_config_headers="$ac_config_headers lib/GridConfig.h"
 
 # Check whether --enable-silent-rules was given.
diff --git a/tests/Test_cayley_ldop_cr.cc b/tests/Test_cayley_ldop_cr.cc
index afa773b2..698bbb28 100644
--- a/tests/Test_cayley_ldop_cr.cc
+++ b/tests/Test_cayley_ldop_cr.cc
@@ -64,21 +64,10 @@ int main (int argc, char ** argv)
 
   //random(RNG4,Umu);
   NerscField header;
-  std::string file("./ckpoint_lat.400");
+  std::string file("./ckpoint_lat.4000");
   readNerscConfiguration(Umu,header,file);
 
-#if 0
-  LatticeColourMatrix U(UGrid);
-  Complex cone(1.0,0.0);
-  for(int nn=0;nn<Nd;nn++){
-    if (nn==3) { 
-      U=zero; std::cout << "zeroing gauge field in dir "<<nn<<std::endl;
-    //    else       { U[nn]= cone;std::cout << "unit gauge field in dir "<<nn<<std::endl; }
-      pokeIndex<LorentzIndex>(Umu,U,nn);
-    }
-  }
-#endif  
-  RealD mass=0.5;
+  RealD mass=0.01;
   RealD M5=1.8;
 
   DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
@@ -100,7 +89,7 @@ int main (int argc, char ** argv)
     //    filter(HermIndefOp,noise,subspace[b]);
     // inverse iteration
     MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
-    ConjugateGradient<LatticeFermion> CG(1.0e-5,10000);
+    ConjugateGradient<LatticeFermion> CG(1.0e-4,10000);
 
     for(int i=0;i<4;i++){
 
diff --git a/tests/Test_cayley_ldop_cr_chebyshev.cc b/tests/Test_cayley_ldop_cr_chebyshev.cc
new file mode 100644
index 00000000..091b8eff
--- /dev/null
+++ b/tests/Test_cayley_ldop_cr_chebyshev.cc
@@ -0,0 +1,124 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+};
+
+double lowpass(double x)
+{
+  return pow(x*x,-2);
+}
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  Chebyshev<LatticeFermion> filter(-120.0, 120.0,256, lowpass);
+  ofstream csv(std::string("filter.dat"),std::ios::out|std::ios::trunc);
+  filter.csv(csv);
+  csv.close();
+
+  const int Ls=8;
+
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+  // Construct a coarsened grid
+  std::vector<int> clatt = GridDefaultLatt();
+  for(int d=0;d<clatt.size();d++){
+    clatt[d] = clatt[d]/2;
+  }
+  GridCartesian *Coarse4d =  SpaceTimeGrid::makeFourDimGrid(clatt, GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());;
+  GridCartesian *Coarse5d =  SpaceTimeGrid::makeFiveDimGrid(1,Coarse4d);
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  std::vector<int> cseeds({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);   RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);   RNG4.SeedFixedIntegers(seeds4);
+  GridParallelRNG          CRNG(Coarse5d);CRNG.SeedFixedIntegers(cseeds);
+
+  LatticeFermion    src(FGrid); gaussian(RNG5,src);
+  LatticeFermion result(FGrid); result=zero;
+  LatticeFermion    ref(FGrid); ref=zero;
+  LatticeFermion    tmp(FGrid);
+  LatticeFermion    err(FGrid);
+  LatticeGaugeField Umu(UGrid); gaussian(RNG4,Umu);
+
+
+  //random(RNG4,Umu);
+  NerscField header;
+  std::string file("./ckpoint_lat.400");
+  readNerscConfiguration(Umu,header,file);
+
+#if 0
+  LatticeColourMatrix U(UGrid);
+  Complex cone(1.0,0.0);
+  for(int nn=0;nn<Nd;nn++){
+    if (nn==3) { 
+      U=zero; std::cout << "zeroing gauge field in dir "<<nn<<std::endl;
+    //    else       { U[nn]= cone;std::cout << "unit gauge field in dir "<<nn<<std::endl; }
+      pokeIndex<LorentzIndex>(Umu,U,nn);
+    }
+  }
+#endif  
+  RealD mass=0.5;
+  RealD M5=1.8;
+
+  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> HermIndefOp(Ddwf);
+
+  const int nbasis = 8;
+  std::vector<LatticeFermion> subspace(nbasis,FGrid);
+  LatticeFermion noise(FGrid);
+  LatticeFermion ms(FGrid);
+  for(int b=0;b<nbasis;b++){
+    Gamma g5(Gamma::Gamma5);
+    gaussian(RNG5,noise);
+    RealD scale = pow(norm2(noise),-0.5); 
+    noise=noise*scale;
+
+    HermIndefOp.HermOp(noise,ms); std::cout << "Noise    "<<b<<" Ms "<<norm2(ms)<< " "<< norm2(noise)<<std::endl;
+
+    filter(HermIndefOp,noise,subspace[b]);
+    scale = pow(norm2(subspace[b]),-0.5); 
+    subspace[b]=subspace[b]*scale;
+    HermIndefOp.HermOp(subspace[b],ms); std::cout << "Filtered "<<b<<" Ms "<<norm2(ms)<< " "<<norm2(subspace[b]) <<std::endl;
+  }
+  std::cout << "Computed randoms"<< std::endl;
+
+  typedef CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis> LittleDiracOperator;
+  typedef LittleDiracOperator::CoarseVector CoarseVector;
+
+  LittleDiracOperator LittleDiracOp(*Coarse5d);
+  LittleDiracOp.CoarsenOperator(FGrid,HermIndefOp,subspace);
+  
+  CoarseVector c_src (Coarse5d);
+  CoarseVector c_res (Coarse5d);
+  gaussian(CRNG,c_src);
+  c_res=zero;
+
+  std::cout << "Solving MCR on coarse space "<< std::endl;
+  HermitianLinearOperator<LittleDiracOperator,CoarseVector> HermIndefLdop(LittleDiracOp);
+  ConjugateResidual<CoarseVector> MCR(1.0e-8,10000);
+  MCR(HermIndefLdop,c_src,c_res);
+
+  std::cout << "Done "<< std::endl;
+  Grid_finalize();
+}

From 46eafa520d518497a0cd229db212381c38469867 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:48:41 +0100
Subject: [PATCH 313/429] fix no compile

---
 lib/AlignedAllocator.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/AlignedAllocator.h b/lib/AlignedAllocator.h
index 0e74e842..fd7bbb78 100644
--- a/lib/AlignedAllocator.h
+++ b/lib/AlignedAllocator.h
@@ -33,7 +33,7 @@ public:
   ~alignedAllocator() throw() { }
 
   pointer       address(reference __x)       const { return &__x; }
-  const_pointer address(const_reference __x) const { return &__x; }
+  //  const_pointer address(const_reference __x) const { return &__x; }
 
   size_type  max_size() const throw() { return size_t(-1) / sizeof(_Tp); }
 

From 9184cc68da79677672216fd44ff4fd7adc5efd59 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:49:26 +0100
Subject: [PATCH 314/429] TensorRemove not needed now

---
 tests/Test_gamma.cc | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/tests/Test_gamma.cc b/tests/Test_gamma.cc
index bae77ecc..118362e4 100644
--- a/tests/Test_gamma.cc
+++ b/tests/Test_gamma.cc
@@ -95,7 +95,7 @@ int main (int argc, char ** argv)
   for(int mu=0;mu<6;mu++){
     result =  Gamma(g[mu])* ident * Gamma(g[mu]);
     result = result - ident;
-    double mag = TensorRemove(norm2(result));
+    RealD mag = norm2(result);
     std::cout << list[mu]<<" " << mag<<std::endl;
   }
 
@@ -103,7 +103,7 @@ int main (int argc, char ** argv)
   for(int mu=0;mu<6;mu++){
     result =          rnd * Gamma(g[mu]);
     result = result + rnd * Gamma(g[mu+6]);
-    double mag = TensorRemove(norm2(result));
+    RealD mag = norm2(result);
     std::cout << list[mu]<<" " << mag<<std::endl;
   }
 
@@ -111,7 +111,7 @@ int main (int argc, char ** argv)
   for(int mu=0;mu<6;mu++){
     result =           Gamma(g[mu])  *rnd;
     result = result +  Gamma(g[mu+6])*rnd;
-    double mag = TensorRemove(norm2(result));
+    RealD mag = norm2(result);
     std::cout << list[mu]<<" " << mag<<std::endl;
   }
 

From e4e91d304216b6dc90d0c93dc41aef08f25489da Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:49:57 +0100
Subject: [PATCH 315/429] be more precise on typing

---
 tests/Test_main.cc | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/tests/Test_main.cc b/tests/Test_main.cc
index 88d69192..daa6c3aa 100644
--- a/tests/Test_main.cc
+++ b/tests/Test_main.cc
@@ -242,7 +242,7 @@ int main (int argc, char ** argv)
     
 
     { // Peek-ology and Poke-ology, with a little app-ology
-      TComplex      c;
+      Complex      c;
       ColourMatrix c_m;   
       SpinMatrix   s_m;   
       SpinColourMatrix sc_m; 
@@ -299,7 +299,7 @@ int main (int argc, char ** argv)
     }
 
     Bar = zero;
-    Bar = where(lex<10,Foo,Bar);
+    Bar = where(lex<Integer(10),Foo,Bar);
     {
       std::vector<int> coor(4);
       for(coor[3]=0;coor[3]<latt_size[3]/mpi_layout[3];coor[3]++){
@@ -467,7 +467,8 @@ int main (int argc, char ** argv)
         mdiff = shifted1-shifted2;
         amdiff=adj(mdiff);
         ColourMatrix prod = amdiff*mdiff;
-        Real Ttr=real(trace(prod));
+	Complex trprod = trace(prod);
+        Real Ttr=real(trprod);
         double nn=Ttr;
         if ( nn > 0 )
             cout<<"Shift real trace fail "<<coor[0]<<coor[1]<<coor[2]<<coor[3] <<endl;

From 0c8c44b3e3d40fd48d4fe5b60009929c87d1038a Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:50:26 +0100
Subject: [PATCH 316/429] Minor change

---
 tests/Test_cshift.cc | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/tests/Test_cshift.cc b/tests/Test_cshift.cc
index 85f8529c..e571a338 100644
--- a/tests/Test_cshift.cc
+++ b/tests/Test_cshift.cc
@@ -66,7 +66,8 @@ int main (int argc, char ** argv)
 	  
 	  nrm = abs(scm-cm()()());
 	  std::vector<int> peer(4);
-	  int index=real(cm);
+	  Complex tmp  =cm;
+	  Integer index=real(tmp);
 	  Fine.CoorFromIndex(peer,index,latt_size);
 
 	  if (nrm > 0){

From f3aebd4b33f126275630bed192844ae5a6ecd404 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:50:59 +0100
Subject: [PATCH 317/429] Apply a heatbath sweep

---
 tests/Test_lie_generators.cc | 271 +++++++++--------------------------
 1 file changed, 68 insertions(+), 203 deletions(-)

diff --git a/tests/Test_lie_generators.cc b/tests/Test_lie_generators.cc
index b29682a0..91f2d33f 100644
--- a/tests/Test_lie_generators.cc
+++ b/tests/Test_lie_generators.cc
@@ -2,226 +2,91 @@
 
 #include <qcd/utils/CovariantCshift.h>
 #include <qcd/utils/WilsonLoops.h>
+#include <qcd/utils/SUn.h>
 
 using namespace std;
 using namespace Grid;
 using namespace Grid::QCD;
 
-class suN {
-public:
-
-  static int generators(int ncolour)   { return ncolour*ncolour-1; }
-  static int su2subgroups(int ncolour) { return (ncolour*(ncolour-1))/2; }
-  
-  template<typename CComplex,int N> using suNmatrix = iScalar<iScalar<iMatrix<CComplex,N> > > ;
-
-  ////////////////////////////////////////////////////////////////////////
-  // There are N^2-1 generators for SU(N).
-  //
-  // We take a traceless hermitian generator basis as follows
-  //
-  // * Normalisation: trace ta tb = 1/2 delta_ab
-  // 
-  // * Off diagonal
-  //    - pairs of rows i1,i2 behaving like pauli matrices signma_x, sigma_y
-  //     
-  //    - there are (Nc-1-i1) slots for i2 on each row [ x  0  x ]                                   
-  //      direct count off each row                                                                    
-  //
-  //    - Sum of all pairs is Nc(Nc-1)/2: proof arithmetic series
-  //
-  //      (Nc-1) + (Nc-2)+...  1      ==> Nc*(Nc-1)/2 
-  //      1+ 2+          +   + Nc-1                        
-  // 
-  //    - There are 2 x Nc (Nc-1)/ 2 of these = Nc^2 - Nc
-  //
-  //    - We enumerate the row-col pairs.
-  //    - for each row col pair there is a (sigma_x) and a (sigma_y) like generator
-  //
-  //
-  //   t^a_ij = { in 0.. Nc(Nc-1)/2 -1} =>  delta_{i,i1} delta_{j,i2} +  delta_{i,i1} delta_{j,i2}  
-  //   t^a_ij = { in Nc(Nc-1)/2 ... Nc^(Nc-1) -1} =>  i delta_{i,i1} delta_{j,i2} - i delta_{i,i1} delta_{j,i2}  
-  //   
-  // * Diagonal; must be traceless and normalised
-  //   - Sequence is 
-  //   N  (1,-1,0,0...)
-  //   N  (1, 1,-2,0...)
-  //   N  (1, 1, 1,-3,0...)
-  //   N  (1, 1, 1, 1,-4,0...)
-  //
-  //   where 1/2 = N^2 (1+.. m^2)etc.... for the m-th diagonal generator
-  //   NB this gives the famous SU3 result for su2 index 8
-  //
-  //   N= sqrt(1/2 . 1/6 ) = 1/2 . 1/sqrt(3) 
-  //
-  //   ( 1      )
-  //   (    1   ) / sqrt(3) /2  = 1/2 lambda_8
-  //   (      -2)
-  ////////////////////////////////////////////////////////////////////////
-  template<class CComplex,int Ncolour> 
-  static void suNgenerator(int lieIndex,suNmatrix<CComplex,Ncolour> &ta){
-    // map lie index to which type of generator
-    int diagIndex;
-    int su2Index;
-    int sigxy;
-    int NNm1 =  Ncolour*(Ncolour-1);
-    if ( lieIndex>= NNm1 ) {
-      diagIndex = lieIndex -NNm1;
-      suNgeneratorDiagonal(diagIndex,ta);
-      return;
-    }
-    sigxy   = lieIndex&0x1;
-    su2Index= lieIndex>>1;
-    if ( sigxy ) suNgeneratorSigmaY(su2Index,ta);
-    else         suNgeneratorSigmaX(su2Index,ta);
-  }
-  template<class CComplex,int Ncolour> 
-  static void suNgeneratorSigmaX(int su2Index,suNmatrix<CComplex,Ncolour> &ta){
-    ta=zero;
-    int i1,i2;
-    su2SubGroupIndex<Ncolour>(i1,i2,su2Index);
-    ta()()(i1,i2)=1.0;
-    ta()()(i2,i1)=1.0;
-    ta= ta *0.5;
-  }
-  template<class CComplex,int Ncolour> 
-  static void suNgeneratorSigmaY(int su2Index,suNmatrix<CComplex,Ncolour> &ta){
-    ta=zero;
-    Complex i(0.0,1.0);
-    int i1,i2;
-    su2SubGroupIndex<Ncolour>(i1,i2,su2Index);
-    ta()()(i1,i2)=-i;
-    ta()()(i2,i1)= i;
-    ta= ta *0.5;
-  }
-  template<class CComplex,int Ncolour> 
-  static void suNgeneratorDiagonal(int diagIndex,suNmatrix<CComplex,Ncolour> &ta){
-    ta=zero;
-    int trsq=0;
-    int last=diagIndex+1;
-    for(int i=0;i<=diagIndex;i++){
-      ta()()(i,i) = 1.0;
-      trsq++;
-    }
-    ta()()(last,last) = -last;
-    trsq+=last*last;
-    RealD nrm = 1.0/std::sqrt(2.0*trsq);
-    ta = ta *nrm;
-  }
-  ////////////////////////////////////////////////////////////////////////
-  // Map a 
-  //
-  ////////////////////////////////////////////////////////////////////////
-  template<int Ncolour>
-  static void su2SubGroupIndex(int &i1,int &i2,int su2_index){
-
-    assert( (su2_index>=0) && (su2_index< (Ncolour*(Ncolour-1))/2) );
-
-    int spare=su2_index;
-    for(i1=0;spare >= (Ncolour-1-i1);i1++ ){
-      spare = spare - (Ncolour-1-i1);  // remove the Nc-1-i1 terms                                 
-    }
-    i2=i1+1+spare;
-  }
-  template<class CComplex,int Ncolour>
-  static void su2Extract(std::vector<LatticeComplex> &r,const Lattice<suNmatrix<CComplex,Ncolour> > &source, int su2_index)
-  {
-    assert(r.size() == 4); // store in 4 real parts
-    
-    for(int i=0;i<4;i++){
-      conformable(r[i],source);
-    }
-    
-    int i1,i2;    
-    su2SubGroupIndex<Ncolour>(i1,i2,su2_index);
-    
-    /* Compute the b(k) of A_SU(2) = b0 + i sum_k bk sigma_k */ 
-    r[0] = real(source()()(i1,i1) + source()()(i2,i2));
-    r[1] = imag(source()()(i1,i2) + source()()(i2,i1));
-    r[2] = real(source()()(i1,i2) - source()()(i2,i1));
-    r[3] = imag(source()()(i1,i1) - source()()(i2,i2));
-  }
-
-
-  template<int Ncolour> static void printGenerators(void)
-  {
-    for(int gen=0;gen<suN::generators(Ncolour);gen++){
-      suN::suNmatrix<Complex,Ncolour> ta;
-      suN::suNgenerator(gen,ta);
-      std::cout<< "Nc = "<<Ncolour<<" t_"<<gen<<std::endl;
-      std::cout<<ta<<std::endl;
-    }
-  }
-
-  template<int Ncolour> static void testGenerators(void){
-    suNmatrix<Complex,Ncolour> ta;
-    suNmatrix<Complex,Ncolour> tb;
-    std::cout<<"Checking trace ta tb is 0.5 delta_ab"<<std::endl;
-    for(int a=0;a<generators(Ncolour);a++){
-      for(int b=0;b<generators(Ncolour);b++){
-	suNgenerator(a,ta);
-	suNgenerator(b,tb);
-	Complex tr =TensorRemove(trace(ta*tb)); 
-	std::cout<<tr<<" ";
-	if(a==b) assert(abs(tr-Complex(0.5))<1.0e-6);
-	if(a!=b) assert(abs(tr)<1.0e-6);
-      }
-      std::cout<<std::endl;
-    }
-    std::cout<<"Checking hermitian"<<std::endl;
-    for(int a=0;a<generators(Ncolour);a++){
-      suNgenerator(a,ta);
-      std::cout<<a<<" ";
-      assert(norm2(ta-adj(ta))<1.0e-6);
-    }    
-    std::cout<<std::endl;
-
-    std::cout<<"Checking traceless"<<std::endl;
-    for(int a=0;a<generators(Ncolour);a++){
-      suNgenerator(a,ta);
-      Complex tr =TensorRemove(trace(ta)); 
-      std::cout<<a<<" ";
-      assert(abs(tr)<1.0e-6);
-    }    
-    std::cout<<std::endl;
-  }
-
-};
-
 
 int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-
-  std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
-  std::vector<int> mpi_layout  = GridDefaultMpi();
-  std::vector<int> latt_size  ({4,4,4,4});
-
-  GridCartesian     Fine(latt_size,simd_layout,mpi_layout);
-
-  LatticeGaugeField Umu(&Fine);
+  std::vector<int> latt({4,4,4,8});
+  GridCartesian * grid = SpaceTimeGrid::makeFourDimGrid(latt, 
+							GridDefaultSimd(Nd,vComplexF::Nsimd()),
+							GridDefaultMpi());
+  
+  GridRedBlackCartesian * rbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(grid);
 
   std::cout<<"*********************************************"<<std::endl;
   std::cout<<"* Generators for SU(2)"<<std::endl;
   std::cout<<"*********************************************"<<std::endl;
-  suN::printGenerators<2>();
-  suN::testGenerators<2>();
+  SU2::printGenerators();
+  SU2::testGenerators();
+
   std::cout<<"*********************************************"<<std::endl;
   std::cout<<"* Generators for SU(3)"<<std::endl;
   std::cout<<"*********************************************"<<std::endl;
-  suN::printGenerators<3>();
-  suN::testGenerators<3>();
-  std::cout<<"*********************************************"<<std::endl;
-  std::cout<<"* Generators for SU(4)"<<std::endl;
-  std::cout<<"*********************************************"<<std::endl;
-  suN::printGenerators<4>();
-  suN::testGenerators<4>();
-  std::cout<<"*********************************************"<<std::endl;
-  std::cout<<"* Generators for SU(5)"<<std::endl;
-  std::cout<<"*********************************************"<<std::endl;
-  suN::printGenerators<5>();
-  suN::testGenerators<5>();
+  SU3::printGenerators();
+  SU3::testGenerators();
+
+  //  std::cout<<"*********************************************"<<std::endl;
+  //  std::cout<<"* Generators for SU(4)"<<std::endl;
+  //  std::cout<<"*********************************************"<<std::endl;
+  //  SU4::printGenerators();
+  //  SU4::testGenerators();
+
+  //  std::cout<<"*********************************************"<<std::endl;
+  //  std::cout<<"* Generators for SU(5)"<<std::endl;
+  //  std::cout<<"*********************************************"<<std::endl;
+  //  SU5::printGenerators();
+  //  SU5::testGenerators();
+
+  ///////////////////////////////
+  // Configuration of known size
+  ///////////////////////////////
+  NerscField header;
+  std::string file("./ckpoint_lat.400");
+  LatticeGaugeField Umu(grid);
+  //  readNerscConfiguration(Umu,header,file);
+  Umu=1.0; // Cold start
+
+  // RNG set up for test
+  std::vector<int> pseeds({1,2,3,4,5}); // once I caught a fish alive
+  std::vector<int> sseeds({6,7,8,9,10});// then i let it go again
+  GridParallelRNG  pRNG(grid); pRNG.SeedFixedIntegers(pseeds);
+  GridSerialRNG    sRNG;       sRNG.SeedFixedIntegers(sseeds);
+
+  // SU3 colour operatoions
+  LatticeColourMatrix link(grid);
+  LatticeColourMatrix staple(grid);
+  int mu=0;
+
+  // Get Staple
+  ColourWilsonLoops::Staple(staple,Umu,mu);
+  // Get Link
+  link = peekIndex<LorentzIndex>(Umu,mu);
+
+  // Apply heatbath to the link
+  RealD beta=6.0;
+  int subgroup=0;
+  int nhb=1;
+  int trials=0;
+  int fails=0;
+
+  LatticeInteger one(rbGrid);  one = 1; // fill with ones
+  LatticeInteger mask(grid);   mask= zero;
+  one.checkerboard=Even;
+  setCheckerboard(mask,one);
+
+  // update Even checkerboard
+
+  SU3::SubGroupHeatBath(sRNG,pRNG,beta,link,staple,subgroup,
+			nhb,trials,fails,mask);
+
+
 
   Grid_finalize();
 }

From ff3db9ceee3faf5bb499188961a233c08678d15d Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:51:37 +0100
Subject: [PATCH 318/429] Typing

---
 tests/Test_cshift_red_black.cc | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/tests/Test_cshift_red_black.cc b/tests/Test_cshift_red_black.cc
index 9ffa66b1..5cafd04f 100644
--- a/tests/Test_cshift_red_black.cc
+++ b/tests/Test_cshift_red_black.cc
@@ -100,7 +100,8 @@ int main (int argc, char ** argv)
 	  
 	  double nrm = abs(scm-cm()()());
 	  std::vector<int> peer(4);
-	  int index=real(cm);
+	  Complex ctmp = cm;
+	  Integer index=real(ctmp);
 	  Fine.CoorFromIndex(peer,index,latt_size);
 
 	  if (nrm > 0){
@@ -138,7 +139,8 @@ int main (int argc, char ** argv)
 	  
 	  nrm = abs(scm-cm()()());
 	  std::vector<int> peer(4);
-	  int index=real(cm);
+	  Complex ctmp=cm;
+	  Integer index=real(ctmp);
 	  Fine.CoorFromIndex(peer,index,latt_size);
 
 	  if (nrm > 0){

From c79d85f76364ac8e73b0ae1e7911dfc61fffcf3a Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:51:56 +0100
Subject: [PATCH 319/429] more accurate comment

---
 tests/Test_cayley_coarsen_support.cc | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/tests/Test_cayley_coarsen_support.cc b/tests/Test_cayley_coarsen_support.cc
index b68a5cc3..8cdc9565 100644
--- a/tests/Test_cayley_coarsen_support.cc
+++ b/tests/Test_cayley_coarsen_support.cc
@@ -102,16 +102,11 @@ int main (int argc, char ** argv)
   CoarseVector c_res (Coarse5d);
   CoarseVector c_proj(Coarse5d);
   
-  //  TODO
-  // -- promote from subspace, check we get the vector we wanted
-  // -- apply ldop; check we get the same as inner product of M times big vec
-  // -- pick blocks one by one. Evaluate matrix elements.
   Complex one(1.0);
   c_src = one;  // 1 in every element for vector 1.
   
   blockPromote(c_src,err,subspace);
 
-
   prom=zero;
   for(int b=0;b<nbasis;b++){
     prom=prom+subspace[b];

From 3e261b3d9e002ad1266ae6458e7ebd570347e89c Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:52:26 +0100
Subject: [PATCH 320/429] Minor

---
 tests/Test_GaugeAction.cc | 1 +
 1 file changed, 1 insertion(+)

diff --git a/tests/Test_GaugeAction.cc b/tests/Test_GaugeAction.cc
index fc49cf2e..56634341 100644
--- a/tests/Test_GaugeAction.cc
+++ b/tests/Test_GaugeAction.cc
@@ -27,6 +27,7 @@ public:
   }
 };
 
+
 int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);

From 5c66b5c71286b706c5e2f731c360cc4ca9f236f9 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:52:39 +0100
Subject: [PATCH 321/429] File list

---
 tests/Make.inc | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/tests/Make.inc b/tests/Make.inc
index d0d7d001..d04bf0ac 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,11 +1,15 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cg Test_cayley_ldop_cr Test_cayley_ldop_cr_chebyshev Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_gamma Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
+bin_PROGRAMS = Test_GaugeAction Test_Metropolis Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cg Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_gamma Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
 
 
 Test_GaugeAction_SOURCES=Test_GaugeAction.cc
 Test_GaugeAction_LDADD=-lGrid
 
 
+Test_Metropolis_SOURCES=Test_Metropolis.cc
+Test_Metropolis_LDADD=-lGrid
+
+
 Test_cayley_cg_SOURCES=Test_cayley_cg.cc
 Test_cayley_cg_LDADD=-lGrid
 
@@ -26,10 +30,6 @@ Test_cayley_ldop_cr_SOURCES=Test_cayley_ldop_cr.cc
 Test_cayley_ldop_cr_LDADD=-lGrid
 
 
-Test_cayley_ldop_cr_chebyshev_SOURCES=Test_cayley_ldop_cr_chebyshev.cc
-Test_cayley_ldop_cr_chebyshev_LDADD=-lGrid
-
-
 Test_cf_coarsen_support_SOURCES=Test_cf_coarsen_support.cc
 Test_cf_coarsen_support_LDADD=-lGrid
 

From d66cab3f01838e851acb3943df4b0887223223e1 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:53:18 +0100
Subject: [PATCH 322/429] Unary funcs update

---
 lib/tensors/Tensor_unary.h | 67 +++++++++++++++++++++++++++++++++++---
 1 file changed, 62 insertions(+), 5 deletions(-)

diff --git a/lib/tensors/Tensor_unary.h b/lib/tensors/Tensor_unary.h
index 2a546701..045097a3 100644
--- a/lib/tensors/Tensor_unary.h
+++ b/lib/tensors/Tensor_unary.h
@@ -2,7 +2,7 @@
 #define GRID_TENSOR_UNARY_H
 namespace Grid {
 
-#define UNARY_REAL(func)\
+#define UNARY(func)\
 template<class obj> inline auto func(const iScalar<obj> &z) -> iScalar<obj>\
 {\
     iScalar<obj> ret;\
@@ -53,14 +53,71 @@ template<class obj> inline iScalar<obj> func(const iScalar<obj> &z,scal y)	\
     return ret;\
 }
 
-UNARY_REAL(sqrt);
-UNARY_REAL(rsqrt);
-UNARY_REAL(sin);
-UNARY_REAL(cos);
+UNARY(sqrt);
+UNARY(rsqrt);
+UNARY(sin);
+UNARY(cos);
+UNARY(log);
+UNARY(exp);
+UNARY(abs);
+UNARY(Not);
+
+
+template<class obj> inline auto toReal(const iScalar<obj> &z) -> typename iScalar<obj>::Realified
+{
+  typename iScalar<obj>::Realified ret;
+  ret._internal = toReal(z._internal);
+  return ret;
+}
+ template<class obj,int N> inline auto toReal(const iVector<obj,N> &z) -> typename iVector<obj,N>::Realified
+{
+  typename iVector<obj,N>::Realified ret;
+  for(int c1=0;c1<N;c1++){  
+    ret._internal[c1] = toReal(z._internal[c1]); 
+  }
+  return ret;
+}
+template<class obj,int N> inline auto toReal(const iMatrix<obj,N> &z) -> typename iMatrix<obj,N>::Realified
+{
+  typename iMatrix<obj,N>::Realified ret;
+  for(int c1=0;c1<N;c1++){
+  for(int c2=0;c2<N;c2++){
+    ret._internal[c1][c2] = toReal(z._internal[c1][c2]);
+  }}
+  return ret;
+}
+
+template<class obj> inline auto toComplex(const iScalar<obj> &z) -> typename iScalar<obj>::Complexified
+{
+  typename iScalar<obj>::Complexified ret;
+  ret._internal = toComplex(z._internal);
+  return ret;
+}
+ template<class obj,int N> inline auto toComplex(const iVector<obj,N> &z) -> typename iVector<obj,N>::Complexified
+{
+  typename iVector<obj,N>::Complexified ret;
+  for(int c1=0;c1<N;c1++){  
+    ret._internal[c1] = toComplex(z._internal[c1]); 
+  }
+  return ret;
+}
+template<class obj,int N> inline auto toComplex(const iMatrix<obj,N> &z) -> typename iMatrix<obj,N>::Complexified
+{
+  typename iMatrix<obj,N>::Complexified ret;
+  for(int c1=0;c1<N;c1++){
+  for(int c2=0;c2<N;c2++){
+    ret._internal[c1][c2] = toComplex(z._internal[c1][c2]);
+  }}
+  return ret;
+}
+
 
 BINARY_RSCALAR(mod,Integer);
 BINARY_RSCALAR(pow,RealD);
 
+#undef UNARY
+#undef BINARY_RSCALAR
+
 
 }
 #endif

From d74aa0e1e6f046f0c95711dddbdb795b1e40e2fc Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:53:52 +0100
Subject: [PATCH 323/429] Real handling improvement

---
 lib/tensors/Tensor_traits.h | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)

diff --git a/lib/tensors/Tensor_traits.h b/lib/tensors/Tensor_traits.h
index b5715c93..703a2ff5 100644
--- a/lib/tensors/Tensor_traits.h
+++ b/lib/tensors/Tensor_traits.h
@@ -26,6 +26,8 @@ namespace Grid {
     typedef typename T::vector_type vector_type;
     typedef typename T::tensor_reduced tensor_reduced;
     typedef typename T::scalar_object scalar_object;
+    typedef typename T::Complexified Complexified;
+    typedef typename T::Realified Realified;
     enum { TensorLevel = T::TensorLevel };
   };
 
@@ -38,6 +40,8 @@ namespace Grid {
     typedef RealF vector_type;
     typedef RealF tensor_reduced ;
     typedef RealF scalar_object;
+    typedef ComplexF Complexified;
+    typedef RealF Realified;
     enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<RealD> {
@@ -46,6 +50,8 @@ namespace Grid {
     typedef RealD vector_type;
     typedef RealD tensor_reduced;
     typedef RealD scalar_object;
+    typedef ComplexD Complexified;
+    typedef RealD Realified;
     enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<ComplexF> {
@@ -54,6 +60,8 @@ namespace Grid {
     typedef ComplexF vector_type;
     typedef ComplexF tensor_reduced;
     typedef ComplexF scalar_object;
+    typedef ComplexF Complexified;
+    typedef RealF Realified;
     enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<ComplexD> {
@@ -62,6 +70,8 @@ namespace Grid {
     typedef ComplexD vector_type;
     typedef ComplexD tensor_reduced;
     typedef ComplexD scalar_object;
+    typedef ComplexD Complexified;
+    typedef RealD Realified;
     enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<Integer> {
@@ -70,6 +80,8 @@ namespace Grid {
     typedef Integer vector_type;
     typedef Integer tensor_reduced;
     typedef Integer scalar_object;
+    typedef void Complexified;
+    typedef void Realified;
     enum { TensorLevel = 0 };
   };
 
@@ -79,6 +91,8 @@ namespace Grid {
     typedef vRealF vector_type;
     typedef vRealF tensor_reduced;
     typedef RealF  scalar_object;
+    typedef vComplexF Complexified;
+    typedef vRealF Realified;
     enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<vRealD> {
@@ -87,6 +101,8 @@ namespace Grid {
     typedef vRealD vector_type;
     typedef vRealD tensor_reduced;
     typedef RealD  scalar_object;
+    typedef vComplexD Complexified;
+    typedef vRealD Realified;
     enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<vComplexF> {
@@ -95,6 +111,8 @@ namespace Grid {
     typedef vComplexF vector_type;
     typedef vComplexF tensor_reduced;
     typedef ComplexF  scalar_object;
+    typedef vComplexF Complexified;
+    typedef vRealF Realified;
     enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<vComplexD> {
@@ -103,6 +121,8 @@ namespace Grid {
     typedef vComplexD vector_type;
     typedef vComplexD tensor_reduced;
     typedef ComplexD  scalar_object;
+    typedef vComplexD Complexified;
+    typedef vRealD Realified;
     enum { TensorLevel = 0 };
   };
   template<> class GridTypeMapper<vInteger> {
@@ -111,6 +131,8 @@ namespace Grid {
     typedef vInteger vector_type;
     typedef vInteger tensor_reduced;
     typedef  Integer scalar_object;
+    typedef void Complexified;
+    typedef void Realified;
     enum { TensorLevel = 0 };
   };
 

From 17de4bce207d4528f483d9e9d9ca5d92cf528aae Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:54:18 +0100
Subject: [PATCH 324/429] Real/Complex improvements

---
 lib/tensors/Tensor_inner.h | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/lib/tensors/Tensor_inner.h b/lib/tensors/Tensor_inner.h
index bbb35605..c77adff1 100644
--- a/lib/tensors/Tensor_inner.h
+++ b/lib/tensors/Tensor_inner.h
@@ -10,7 +10,8 @@ namespace Grid {
       typedef typename sobj::scalar_type scalar;
       decltype(innerProduct(arg,arg)) nrm;
       nrm = innerProduct(arg,arg);
-      return real(nrm);
+      RealD ret = real(nrm);
+      return ret;
     }
 
     template<class l,class r,int N> inline

From 51ebc4f40208c455ef717d399c7b6498d14bc35f Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:55:21 +0100
Subject: [PATCH 325/429] Handle case of simd_layout not filling whole vector.
 Useful if real complex live on same grid

---
 lib/tensors/Tensor_extract_merge.h | 67 ++++++++++++++----------------
 1 file changed, 31 insertions(+), 36 deletions(-)

diff --git a/lib/tensors/Tensor_extract_merge.h b/lib/tensors/Tensor_extract_merge.h
index 75868463..cea9bb64 100644
--- a/lib/tensors/Tensor_extract_merge.h
+++ b/lib/tensors/Tensor_extract_merge.h
@@ -31,18 +31,17 @@ inline void merge(typename std::enable_if<!isGridTensor<vsimd>::value, vsimd >::
 		  std::vector<scalar *> &extracted,int offset){
   int Nextr=extracted.size();
   int Nsimd=vsimd::Nsimd();
-  int s=Nsimd/Nextr;
-
+  int s=Nsimd/Nextr; // can have sparse occupation of simd vector if simd_layout does not fill it
+                     // replicate n-fold. Use to allow Integer masks to 
+                     // predicate floating point of various width assignments and maintain conformable.
   scalar *buf =(scalar *) y;
   for(int i=0;i<Nextr;i++){
     for(int ii=0;ii<s;ii++){
       buf[i*s+ii]=extracted[i][offset];
     }
   }
-
 };
 
-
 ////////////////////////////////////////////////////////////////////////////////////////////////
 // Extract a fundamental vector type to scalar array 
 ////////////////////////////////////////////////////////////////////////////////////////////////
@@ -55,8 +54,17 @@ inline void extract(typename std::enable_if<!isGridTensor<vsimd>::value, const v
 
   scalar *buf = (scalar *)&y;
   for(int i=0;i<Nextr;i++){
-    for(int ii=0;ii<s;ii++){
-      extracted[i]=buf[i*s+ii];
+    extracted[i]=buf[i*s];
+    for(int ii=1;ii<s;ii++){
+      if ( buf[i*s]!=buf[i*s+ii] ){
+	std::cout << " SIMD extract failure splat="<<s<<" ii "<<ii<<" " <<Nextr<<" "<< Nsimd<<" "<<std::endl;
+	for(int vv=0;vv<Nsimd;vv++) {
+	  std::cout<< buf[vv]<<" ";
+	}
+	std::cout<<std::endl;
+	assert(0);
+      }
+      assert(buf[i*s]==buf[i*s+ii]);
     }
   }
 
@@ -74,21 +82,7 @@ inline void merge(typename std::enable_if<!isGridTensor<vsimd>::value, vsimd >::
 
   for(int i=0;i<Nextr;i++){
     for(int ii=0;ii<s;ii++){
-      buf[i*s+ii]=extracted[i];
-    }
-  }
-
-};
-template<class vsimd,class scalar>
-inline void AmergeA(typename std::enable_if<!isGridTensor<vsimd>::value, vsimd >::type  &y,std::vector<scalar> &extracted){
-  int Nextr=extracted.size();
-  int Nsimd=vsimd::Nsimd();
-  int s=Nsimd/Nextr;
-
-  scalar *buf = (scalar *)&y;
-  for(int i=0;i<Nextr;i++){
-    for(int ii=0;ii<s;ii++){
-      buf[i*s+ii]=extracted[i];
+      buf[i*s+ii]=extracted[i]; // replicates value
     }
   }
 };
@@ -102,12 +96,12 @@ template<class vobj> inline void extract(const vobj &vec,std::vector<typename vo
   typedef typename vobj::vector_type vector_type ;
 
   const int Nsimd=vobj::vector_type::Nsimd();
+  int Nextr=extracted.size();
   const int words=sizeof(vobj)/sizeof(vector_type);
+  int s=Nsimd/Nextr;
 
-  extracted.resize(Nsimd);
-
-  std::vector<scalar_type *> pointers(Nsimd);
-  for(int i=0;i<Nsimd;i++) 
+  std::vector<scalar_type *> pointers(Nextr);
+  for(int i=0;i<Nextr;i++) 
     pointers[i] =(scalar_type *)& extracted[i];
 
   vector_type *vp = (vector_type *)&vec;
@@ -127,11 +121,11 @@ void extract(const vobj &vec,std::vector<typename vobj::scalar_object *> &extrac
 
   const int words=sizeof(vobj)/sizeof(vector_type);
   const int Nsimd=vobj::vector_type::Nsimd();
-
-  assert(extracted.size()==Nsimd);
+  int Nextr=extracted.size();
+  int s = Nsimd/Nextr;
 
   std::vector<scalar_type *> pointers(Nsimd);
-  for(int i=0;i<Nsimd;i++) {
+  for(int i=0;i<Nextr;i++) {
     pointers[i] =(scalar_type *)& extracted[i][offset];
   }
 
@@ -153,10 +147,11 @@ void merge(vobj &vec,std::vector<typename vobj::scalar_object> &extracted)
   const int Nsimd=vobj::vector_type::Nsimd();
   const int words=sizeof(vobj)/sizeof(vector_type);
 
-  assert(extracted.size()==Nsimd);
+  int Nextr = extracted.size();
+  int splat=Nsimd/Nextr;
 
-  std::vector<scalar_type *> pointers(Nsimd);
-  for(int i=0;i<Nsimd;i++) 
+  std::vector<scalar_type *> pointers(Nextr);
+  for(int i=0;i<Nextr;i++) 
     pointers[i] =(scalar_type *)& extracted[i];
   
   vector_type *vp = (vector_type *)&vec;
@@ -177,14 +172,14 @@ void merge(vobj &vec,std::vector<typename vobj::scalar_object *> &extracted,int
   const int Nsimd=vobj::vector_type::Nsimd();
   const int words=sizeof(vobj)/sizeof(vector_type);
 
-  assert(extracted.size()==Nsimd);
+  int Nextr=extracted.size();
 
-  std::vector<scalar_type *> pointers(Nsimd);
-  for(int i=0;i<Nsimd;i++) 
+  std::vector<scalar_type *> pointers(Nextr);
+  for(int i=0;i<Nextr;i++) 
     pointers[i] =(scalar_type *)& extracted[i][offset];
-  
+
   vector_type *vp = (vector_type *)&vec;
-  assert((void *)vp!=NULL);
+
   for(int w=0;w<words;w++){
     merge<vector_type,scalar_type>(&vp[w],pointers,w);
   }

From b521155218d29592012a8110c2dfa6638dcb8e9a Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:56:08 +0100
Subject: [PATCH 326/429] Real complex improved

---
 lib/tensors/Tensor_class.h | 32 ++++++++++++++++++++++++++++----
 1 file changed, 28 insertions(+), 4 deletions(-)

diff --git a/lib/tensors/Tensor_class.h b/lib/tensors/Tensor_class.h
index 019e6fb1..7962517c 100644
--- a/lib/tensors/Tensor_class.h
+++ b/lib/tensors/Tensor_class.h
@@ -23,13 +23,17 @@ template<class vtype> class iScalar
 public:
   vtype _internal;
 
-  typedef typename GridTypeMapper<vtype>::scalar_type   scalar_type;
+  typedef typename GridTypeMapper<vtype>::scalar_type scalar_type;
   typedef typename GridTypeMapper<vtype>::vector_type vector_type;
   typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
   typedef iScalar<tensor_reduced_v> tensor_reduced;
   typedef typename GridTypeMapper<vtype>::scalar_object recurse_scalar_object;
   typedef iScalar<recurse_scalar_object> scalar_object;
 
+  // substitutes a real or complex version with same tensor structure
+  typedef iScalar<typename GridTypeMapper<vtype>::Complexified > Complexified;
+  typedef iScalar<typename GridTypeMapper<vtype>::Realified >    Realified;
+
   enum { TensorLevel = GridTypeMapper<vtype>::TensorLevel + 1};
 
   // Scalar no action
@@ -86,9 +90,19 @@ public:
   strong_inline const vtype & operator ()(void) const {
     return _internal;
   }
-  
-  operator ComplexD () const { return(TensorRemove(_internal)); };
-  operator RealD () const { return(real(TensorRemove(_internal))); }
+
+  // Type casts meta programmed
+  template<class U=vtype,class V=scalar_type,IfComplex<V> = 0,IfNotSimd<U> = 0>  
+    operator ComplexF () const { return(TensorRemove(_internal)); };
+  template<class U=vtype,class V=scalar_type,IfComplex<V> = 0,IfNotSimd<U> = 0>  
+    operator ComplexD () const { return(TensorRemove(_internal)); };
+  template<class U=vtype,class V=scalar_type,IfComplex<V> = 0,IfNotSimd<U> = 0>  
+    operator RealD () const { return(real(TensorRemove(_internal))); }
+  template<class U=vtype,class V=scalar_type,IfReal<V>    = 0,IfNotSimd<U> = 0>  
+    operator RealD    () const { return TensorRemove(_internal); }
+  template<class U=vtype,class V=scalar_type,IfInteger<V> = 0,IfNotSimd<U> = 0>  
+    operator Integer  () const { return Integer(TensorRemove(_internal)); }
+
   
   // convert from a something to a scalar via constructor of something arg
   template<class T,typename std::enable_if<!isGridTensor<T>::value, T>::type* = nullptr > strong_inline iScalar<vtype> operator = (T arg)
@@ -123,6 +137,10 @@ public:
   typedef iScalar<tensor_reduced_v> tensor_reduced;
   typedef iVector<recurse_scalar_object,N> scalar_object;
 
+  // substitutes a real or complex version with same tensor structure
+  typedef iVector<typename GridTypeMapper<vtype>::Complexified,N > Complexified;
+  typedef iVector<typename GridTypeMapper<vtype>::Realified,N >    Realified;
+
   template<class T,typename std::enable_if<!isGridTensor<T>::value, T>::type* = nullptr > strong_inline auto operator = (T arg) -> iVector<vtype,N>
     { 
       zeroit(*this);
@@ -211,6 +229,12 @@ public:
   typedef typename GridTypeMapper<vtype>::vector_type vector_type;
   typedef typename GridTypeMapper<vtype>::tensor_reduced tensor_reduced_v;
   typedef typename GridTypeMapper<vtype>::scalar_object recurse_scalar_object;
+
+  // substitutes a real or complex version with same tensor structure
+  typedef iMatrix<typename GridTypeMapper<vtype>::Complexified,N > Complexified;
+  typedef iMatrix<typename GridTypeMapper<vtype>::Realified,N >    Realified;
+
+  // Tensure removal
   typedef iScalar<tensor_reduced_v> tensor_reduced;
   typedef iMatrix<recurse_scalar_object,N> scalar_object;
 

From d5810eaef18cc30982a809a7605f6f33e9f3026d Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:56:40 +0100
Subject: [PATCH 327/429] Logical ops element by element

---
 lib/tensors/Tensor_logical.h | 32 ++++++++++++++++++++++++++++++++
 1 file changed, 32 insertions(+)
 create mode 100644 lib/tensors/Tensor_logical.h

diff --git a/lib/tensors/Tensor_logical.h b/lib/tensors/Tensor_logical.h
new file mode 100644
index 00000000..5ded9f9c
--- /dev/null
+++ b/lib/tensors/Tensor_logical.h
@@ -0,0 +1,32 @@
+#ifndef GRID_TENSOR_LOGICAL_H
+#define GRID_TENSOR_LOGICAL_H
+
+namespace Grid {
+
+#define LOGICAL_BINOP(Op)\
+template<class v> strong_inline iScalar<v> operator Op (const iScalar<v>& lhs,const iScalar<v>& rhs) \
+{\
+  iScalar<v> ret;\
+  ret._internal = lhs._internal Op rhs._internal ;\
+  return ret;\
+}\
+template<class l> strong_inline iScalar<l> operator Op (const iScalar<l>& lhs,Integer rhs) \
+{\
+  typename iScalar<l>::scalar_type t; t=rhs;\
+  typename iScalar<l>::tensor_reduced srhs; srhs=t;\
+  return lhs Op srhs;\
+}\
+template<class l> strong_inline iScalar<l> operator Op (Integer lhs,const iScalar<l>& rhs) \
+{\
+  typename iScalar<l>::scalar_type t;t=lhs;\
+  typename iScalar<l>::tensor_reduced slhs;slhs=t;\
+  return slhs Op rhs;\
+}
+
+LOGICAL_BINOP(|);
+LOGICAL_BINOP(&);
+LOGICAL_BINOP(||);
+LOGICAL_BINOP(&&);
+
+}
+#endif

From 9006f7a34a91ab3371664e83f5d141930879a3db Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:57:23 +0100
Subject: [PATCH 328/429] typo fix -- remove extra template arg

---
 lib/tensors/Tensor_arith_scalar.h | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/lib/tensors/Tensor_arith_scalar.h b/lib/tensors/Tensor_arith_scalar.h
index c6418638..5999f935 100644
--- a/lib/tensors/Tensor_arith_scalar.h
+++ b/lib/tensors/Tensor_arith_scalar.h
@@ -9,12 +9,12 @@ namespace Grid {
 //////////////////////////////////////////////////////////////////////////////////////////
 
 // multiplication by fundamental scalar type
-template<class l,int N> strong_inline iScalar<l> operator * (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
+template<class l> strong_inline iScalar<l> operator * (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
 {
   typename iScalar<l>::tensor_reduced srhs; srhs=rhs;
   return lhs*srhs;
 }
-template<class l,int N> strong_inline iScalar<l> operator * (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
+template<class l> strong_inline iScalar<l> operator * (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs*lhs; }
 
 template<class l,int N> strong_inline iVector<l,N> operator * (const iVector<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
 {
@@ -118,12 +118,12 @@ template<class l,int N> strong_inline iMatrix<l,N> operator * (Integer lhs,const
 ///////////////////////////////////////////////////////////////////////////////////////////////
 // addition by fundamental scalar type applies to matrix(down diag) and scalar
 ///////////////////////////////////////////////////////////////////////////////////////////////
-template<class l,int N> strong_inline iScalar<l> operator + (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
+template<class l> strong_inline iScalar<l> operator + (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
 {
   typename iScalar<l>::tensor_reduced srhs; srhs=rhs;
   return lhs+srhs;
 }
-template<class l,int N> strong_inline iScalar<l> operator + (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
+template<class l> strong_inline iScalar<l> operator + (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) {  return rhs+lhs; }
 
 template<class l,int N> strong_inline iMatrix<l,N> operator + (const iMatrix<l,N>& lhs,const typename iScalar<l>::scalar_type rhs) 
 {
@@ -176,12 +176,12 @@ template<class l,int N> strong_inline iMatrix<l,N> operator + (Integer lhs,const
 ///////////////////////////////////////////////////////////////////////////////////////////////
 // subtraction of fundamental scalar type applies to matrix(down diag) and scalar
 ///////////////////////////////////////////////////////////////////////////////////////////////
-template<class l,int N> strong_inline iScalar<l> operator - (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
+template<class l> strong_inline iScalar<l> operator - (const iScalar<l>& lhs,const typename iScalar<l>::scalar_type rhs) 
 {
   typename iScalar<l>::tensor_reduced srhs; srhs=rhs;
   return lhs-srhs;
 }
-template<class l,int N> strong_inline iScalar<l> operator - (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) 
+template<class l> strong_inline iScalar<l> operator - (const typename iScalar<l>::scalar_type lhs,const iScalar<l>& rhs) 
 {
   typename iScalar<l>::tensor_reduced slhs;slhs=lhs;
   return slhs-rhs;

From 8245585deef256d7baeccff3c10bfbc6f8e8a457 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:58:14 +0100
Subject: [PATCH 329/429] More functions broken out into element by element

---
 lib/simd/Grid_vector_unops.h | 108 +++++++++++++++++++++++++++++++++++
 1 file changed, 108 insertions(+)

diff --git a/lib/simd/Grid_vector_unops.h b/lib/simd/Grid_vector_unops.h
index 8e2b6956..3ba77d82 100644
--- a/lib/simd/Grid_vector_unops.h
+++ b/lib/simd/Grid_vector_unops.h
@@ -1,6 +1,8 @@
 #ifndef GRID_VECTOR_UNOPS
 #define GRID_VECTOR_UNOPS
 
+#include <cmath>
+
 namespace Grid { 
 
   template<class scalar> struct SqrtRealFunctor {
@@ -27,6 +29,28 @@ namespace Grid {
     }
   };
 
+  template<class scalar> struct LogRealFunctor {
+    scalar operator()(const scalar &a)  const {
+      return log(real(a));
+    }
+  };
+
+  template<class scalar> struct ExpRealFunctor {
+    scalar operator()(const scalar &a)  const {
+      return exp(real(a));
+    }
+  };
+  template<class scalar> struct NotFunctor {
+    scalar operator()(const scalar &a)  const {
+      return (!a);
+    }
+  };
+  template<class scalar> struct AbsRealFunctor {
+    scalar operator()(const scalar &a)  const {
+      return std::abs(real(a));
+    }
+  };
+
   template<class scalar> struct PowRealFunctor {
     double y;
   PowRealFunctor(double _y) : y(_y) {};
@@ -43,6 +67,25 @@ namespace Grid {
     }
   };
 
+  template<class scalar> struct RealFunctor {
+    scalar operator()(const scalar &a)  const {
+      return real(a);
+    }
+  };
+  template<class scalar> struct ImagFunctor {
+    scalar operator()(const scalar &a)  const {
+      return imag(a);
+    }
+  };
+  template < class S, class V > 
+  inline Grid_simd<S,V> real(const Grid_simd<S,V> &r) {
+    return SimdApply(RealFunctor<S>(),r);
+  }
+  template < class S, class V > 
+  inline Grid_simd<S,V> imag(const Grid_simd<S,V> &r) {
+    return SimdApply(ImagFunctor<S>(),r);
+  }
+
   template < class S, class V > 
   inline Grid_simd<S,V> sqrt(const Grid_simd<S,V> &r) {
     return SimdApply(SqrtRealFunctor<S>(),r);
@@ -60,6 +103,22 @@ namespace Grid {
     return SimdApply(CosRealFunctor<S>(),r);
   }
   template < class S, class V > 
+  inline Grid_simd<S,V> log(const Grid_simd<S,V> &r) {
+    return SimdApply(LogRealFunctor<S>(),r);
+  }
+  template < class S, class V > 
+  inline Grid_simd<S,V> abs(const Grid_simd<S,V> &r) {
+    return SimdApply(AbsRealFunctor<S>(),r);
+  }
+  template < class S, class V > 
+  inline Grid_simd<S,V> exp(const Grid_simd<S,V> &r) {
+    return SimdApply(ExpRealFunctor<S>(),r);
+  }
+  template < class S, class V > 
+  inline Grid_simd<S,V> Not(const Grid_simd<S,V> &r) {
+    return SimdApply(NotFunctor<S>(),r);
+  }
+  template < class S, class V > 
   inline Grid_simd<S,V> pow(const Grid_simd<S,V> &r,double y) {
     return SimdApply(PowRealFunctor<S>(y),r);
   }
@@ -67,6 +126,55 @@ namespace Grid {
   inline Grid_simd<S,V> mod(const Grid_simd<S,V> &r,Integer y) {
     return SimdApply(ModIntFunctor<S>(y),r);
   }
+  ////////////////////////////////////////////////////////////////////////////
+  // Allows us to assign into **conformable** real vectors from complex
+  ////////////////////////////////////////////////////////////////////////////
+  //  template < class S, class V > 
+  //  inline auto ComplexRemove(const Grid_simd<S,V> &c) -> Grid_simd<Grid_simd<S,V>::Real,V> {
+  //    Grid_simd<Grid_simd<S,V>::Real,V> ret;
+  //    ret.v = c.v;
+  //    return ret;
+  //  }
+  template<class scalar> struct AndFunctor {
+    scalar operator()(const scalar &x, const scalar &y)  const {
+      return x & y;
+    }
+  };
+  template<class scalar> struct OrFunctor {
+    scalar operator()(const scalar &x, const scalar &y)  const {
+      return x | y;
+    }
+  };
+  template<class scalar> struct AndAndFunctor {
+    scalar operator()(const scalar &x, const scalar &y)  const {
+      return x && y;
+    }
+  };
+  template<class scalar> struct OrOrFunctor {
+    scalar operator()(const scalar &x, const scalar &y)  const {
+      return x || y;
+    }
+  };
+
+  ////////////////////////////////
+  // Calls to simd binop functors
+  ////////////////////////////////
+  template < class S, class V > 
+  inline Grid_simd<S,V> operator &(const Grid_simd<S,V> &x,const Grid_simd<S,V> &y) {
+    return SimdApplyBinop(AndFunctor<S>(),x,y);
+  }
+  template < class S, class V > 
+  inline Grid_simd<S,V> operator &&(const Grid_simd<S,V> &x,const Grid_simd<S,V> &y) {
+    return SimdApplyBinop(AndAndFunctor<S>(),x,y);
+  }
+  template < class S, class V > 
+  inline Grid_simd<S,V> operator |(const Grid_simd<S,V> &x,const Grid_simd<S,V> &y) {
+    return SimdApplyBinop(OrFunctor<S>(),x,y);
+  }
+  template < class S, class V > 
+  inline Grid_simd<S,V> operator ||(const Grid_simd<S,V> &x,const Grid_simd<S,V> &y) {
+    return SimdApplyBinop(OrOrFunctor<S>(),x,y);
+  }
 
 }
 #endif

From 19e8d2809aeca8a3b5515ac2c9cec675725d35a7 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:59:07 +0100
Subject: [PATCH 330/429] Binop assist and real/complex improvements

---
 lib/simd/Grid_vector_types.h | 79 ++++++++++++++++++++++++++++++++++--
 1 file changed, 75 insertions(+), 4 deletions(-)

diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index c670f531..034d5314 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -28,13 +28,19 @@
 
 namespace Grid {
 
+  //////////////////////////////////////
   // To take the floating point type of real/complex type
+  //////////////////////////////////////
   template <typename T> struct RealPart {
     typedef T type;
   };
   template <typename T> struct RealPart< std::complex<T> >{
     typedef T type;
   };
+  
+  //////////////////////////////////////
+  // demote a vector to real type
+  //////////////////////////////////////
 
   // type alias used to simplify the syntax of std::enable_if
   template <typename T> using Invoke                                  =  typename T::type;
@@ -90,7 +96,7 @@ namespace Grid {
 	Vector_type v;
 	Scalar_type s[sizeof(Vector_type)/sizeof(Scalar_type)];
       conv_t_union(){};
-      } conv_t;
+    } conv_t;
     
    
     Vector_type v;
@@ -205,7 +211,6 @@ namespace Grid {
       return *this;
     }
 
-
     ///////////////////////////////////////
     // Not all functions are supported
     // through SIMD and must breakout to 
@@ -214,7 +219,6 @@ namespace Grid {
     ///////////////////////////////////////
 
     template<class functor> friend inline Grid_simd SimdApply (const functor &func,const Grid_simd &v) {
-
       Grid_simd ret;
       Grid_simd::conv_t conv;
 
@@ -225,6 +229,19 @@ namespace Grid {
       ret.v = conv.v;
       return ret;
     }
+    template<class functor> friend inline Grid_simd SimdApplyBinop (const functor &func,const Grid_simd &x,const Grid_simd &y) {
+      Grid_simd ret;
+      Grid_simd::conv_t cx;
+      Grid_simd::conv_t cy;
+
+      cx.v = x.v;
+      cy.v = y.v;
+      for(int i=0;i<Nsimd();i++){
+	cx.s[i]=func(cx.s[i],cy.s[i]);
+      }
+      ret.v = cx.v;
+      return ret;
+    }
 
     ////////////////////////////////////////////////////////////////////
     // General permute; assumes vector length is same across 
@@ -235,6 +252,7 @@ namespace Grid {
     {
       Gpermute<Grid_simd>(y,b,perm);
     }
+
     
   };// end of Grid_simd class definition 
 
@@ -383,7 +401,6 @@ namespace Grid {
     return in;
   }
 
-
   /////////////////////
   // Inner, outer
   /////////////////////
@@ -405,6 +422,46 @@ namespace Grid {
     return arg;
   }
 
+
+  ////////////////////////////////////////////////////////////
+  // copy/splat complex real parts into real;
+  // insert real into complex and zero imag;
+  ////////////////////////////////////////////////////////////
+
+  //real = toReal( complex )
+  template<class S,class V,IfReal<S>  = 0>	
+  inline Grid_simd<S,V> toReal(const Grid_simd<std::complex<S>,V> &in)
+  {
+    typedef Grid_simd<S,V> simd;
+    simd ret;
+    typename simd::conv_t conv;
+    conv.v = in.v;
+    for(int i=0;i<simd::Nsimd();i+=2){
+      conv.s[i+1]=conv.s[i];    // duplicate (r,r);(r,r);(r,r); etc...
+    }
+    ret.v = conv.v;
+    return ret;
+  }
+  
+  //complex = toComplex( real )
+  template<class R,class V,IfReal<R> = 0 >	// must be a real arg
+  inline Grid_simd<std::complex<R>,V> toComplex (const Grid_simd<R,V> &in)
+  {
+    typedef Grid_simd<R,V> Rsimd;
+    typedef Grid_simd<std::complex<R>,V> Csimd;
+    typename Rsimd::conv_t conv;// address as real
+    
+    conv.v = in.v;
+    for(int i=0;i<Rsimd::Nsimd();i+=2){
+      assert(conv.s[i+1]==conv.s[i]); // trap any cases where real was not duplicated 
+      // indicating the SIMD grids of real and imag assignment did not correctly match
+      conv.s[i+1]=0.0;                // zero imaginary parts
+    }
+    Csimd ret;
+    ret.v = conv.v;
+    return ret;
+  }
+
   ///////////////////////////////
   // Define available types
   ///////////////////////////////
@@ -413,6 +470,20 @@ namespace Grid {
   typedef Grid_simd< std::complex< float > , SIMD_Ftype > vComplexF;
   typedef Grid_simd< std::complex< double >, SIMD_Dtype > vComplexD;
   typedef Grid_simd< Integer               , SIMD_Itype > vInteger;
+
+  /////////////////////////////////////////
+  // Some traits to recognise the types
+  /////////////////////////////////////////
+  template <typename T> struct is_simd : public std::false_type{};
+  template <> struct is_simd<vRealF>   : public std::true_type {};
+  template <> struct is_simd<vRealD>   : public std::true_type {};
+  template <> struct is_simd<vComplexF>: public std::true_type {};
+  template <> struct is_simd<vComplexD>: public std::true_type {};
+  template <> struct is_simd<vInteger> : public std::true_type {};
+
+  template <typename T> using IfSimd     = Invoke<std::enable_if< is_simd<T>::value,int> > ;
+  template <typename T> using IfNotSimd  = Invoke<std::enable_if<!is_simd<T>::value,unsigned> > ;
+
 }
 
 #endif

From 66e57186104232ea8a8f10e9b51ffb372177dd9b Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 00:59:50 +0100
Subject: [PATCH 331/429] const safety

---
 lib/qcd/utils/WilsonLoops.h | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/lib/qcd/utils/WilsonLoops.h b/lib/qcd/utils/WilsonLoops.h
index 82230aa8..69b86859 100644
--- a/lib/qcd/utils/WilsonLoops.h
+++ b/lib/qcd/utils/WilsonLoops.h
@@ -70,7 +70,7 @@ public:
   //////////////////////////////////////////////////
   // the sum over all staples on each site
   //////////////////////////////////////////////////
-  static void Staple(GaugeMat &staple,GaugeLorentz &Umu,int mu){
+  static void Staple(GaugeMat &staple,const GaugeLorentz &Umu,int mu){
 
     std::vector<GaugeMat> U(4,Umu._grid);
     for(int d=0;d<Nd;d++){
@@ -123,12 +123,10 @@ void siteRectangle(GaugeMat &plaq,const std::vector<GaugeMat> &U, const int mu,
 
 
  typedef WilsonLoops<LatticeColourMatrix,LatticeGaugeField> ColourWilsonLoops;
+ typedef WilsonLoops<LatticeColourMatrix,LatticeGaugeField> U1WilsonLoops;
+ typedef WilsonLoops<LatticeColourMatrix,LatticeGaugeField> SU2WilsonLoops;
+ typedef WilsonLoops<LatticeColourMatrix,LatticeGaugeField> SU3WilsonLoops;
 
 }}
 
-
-
-
-
-
 #endif

From 87cc0d4ca3588063558b0513ddac4d58e74aabe7 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 01:00:11 +0100
Subject: [PATCH 332/429] Peek poke colour/spin/complex and trace transpose
 support

---
 lib/qcd/QCD.h | 112 ++++++++++++++++++++++++++++++++++++++++++++++----
 1 file changed, 105 insertions(+), 7 deletions(-)

diff --git a/lib/qcd/QCD.h b/lib/qcd/QCD.h
index f17c557b..ab180ff6 100644
--- a/lib/qcd/QCD.h
+++ b/lib/qcd/QCD.h
@@ -250,6 +250,7 @@ namespace QCD {
     //////////////////////////////////////////////////////////////////////////////
     // Peek and Poke named after physics attributes
     //////////////////////////////////////////////////////////////////////////////
+
     //spin
     template<class vobj> auto peekSpin(const vobj &rhs,int i) -> decltype(peekIndex<SpinIndex>(rhs,0))
     {
@@ -289,20 +290,117 @@ namespace QCD {
     {
       return peekIndex<LorentzIndex>(rhs,i);
     }
-    template<class vobj> auto peekLorentz(const vobj &rhs,int i,int j) -> decltype(peekIndex<LorentzIndex>(rhs,0,0))
-    {
-      return peekIndex<LorentzIndex>(rhs,i,j);
-    }
     template<class vobj> auto peekLorentz(const Lattice<vobj> &rhs,int i) -> decltype(peekIndex<LorentzIndex>(rhs,0))
     {
       return peekIndex<LorentzIndex>(rhs,i);
     }
-    template<class vobj> auto peekLorentz(const Lattice<vobj> &rhs,int i,int j) -> decltype(peekIndex<LorentzIndex>(rhs,0,0))
+
+    //////////////////////////////////////////////
+    // Poke lattice
+    //////////////////////////////////////////////
+    template<class vobj> 
+      void pokeColour(Lattice<vobj> &lhs,
+		      const Lattice<decltype(peekIndex<ColourIndex>(lhs._odata[0],0))> & rhs,
+		      int i)
     {
-      return peekIndex<LorentzIndex>(rhs,i,j);
+      pokeIndex<ColourIndex>(lhs,rhs,i);
+    }
+    template<class vobj> 
+      void pokeColour(Lattice<vobj> &lhs,
+		      const Lattice<decltype(peekIndex<ColourIndex>(lhs._odata[0],0,0))> & rhs,
+		      int i,int j)
+    {
+      pokeIndex<ColourIndex>(lhs,rhs,i,j);
+    }
+    template<class vobj> 
+      void pokeSpin(Lattice<vobj> &lhs,
+		      const Lattice<decltype(peekIndex<SpinIndex>(lhs._odata[0],0))> & rhs,
+		      int i)
+    {
+      pokeIndex<SpinIndex>(lhs,rhs,i);
+    }
+    template<class vobj> 
+      void pokeSpin(Lattice<vobj> &lhs,
+		      const Lattice<decltype(peekIndex<SpinIndex>(lhs._odata[0],0,0))> & rhs,
+		      int i,int j)
+    {
+      pokeIndex<SpinIndex>(lhs,rhs,i,j);
+    }
+    template<class vobj> 
+      void pokeLorentz(Lattice<vobj> &lhs,
+		      const Lattice<decltype(peekIndex<LorentzIndex>(lhs._odata[0],0))> & rhs,
+		      int i)
+    {
+      pokeIndex<LorentzIndex>(lhs,rhs,i);
     }
 
-    // FIXME transpose Colour, transpose Spin, traceColour traceSpin
+    //////////////////////////////////////////////
+    // Poke scalars
+    //////////////////////////////////////////////
+
+    template<class vobj> void pokeSpin(vobj &lhs,const decltype(peekIndex<SpinIndex>(lhs,0)) & rhs,int i)
+    {
+      pokeIndex<SpinIndex>(lhs,rhs,i);
+    }
+    template<class vobj> void pokeSpin(vobj &lhs,const decltype(peekIndex<SpinIndex>(lhs,0,0)) & rhs,int i,int j)
+    {
+      pokeIndex<SpinIndex>(lhs,rhs,i,j);
+    }
+
+    template<class vobj> void pokeColour(vobj &lhs,const decltype(peekIndex<ColourIndex>(lhs,0)) & rhs,int i)
+    {
+      pokeIndex<ColourIndex>(lhs,rhs,i);
+    }
+    template<class vobj> void pokeColour(vobj &lhs,const decltype(peekIndex<ColourIndex>(lhs,0,0)) & rhs,int i,int j)
+    {
+      pokeIndex<ColourIndex>(lhs,rhs,i,j);
+    }
+
+    template<class vobj> void pokeLorentz(vobj &lhs,const decltype(peekIndex<LorentzIndex>(lhs,0)) & rhs,int i)
+    {
+      pokeIndex<LorentzIndex>(lhs,rhs,i);
+    }
+
+
+    //////////////////////////////////////////////
+    // transpose array and scalar
+    //////////////////////////////////////////////
+    template<int Index,class vobj> inline Lattice<vobj> transposeSpin(const Lattice<vobj> &lhs){
+      return transposeIndex<SpinIndex>(lhs);
+    }
+    template<int Index,class vobj> inline Lattice<vobj> transposeColour(const Lattice<vobj> &lhs){
+      return transposeIndex<ColourIndex>(lhs);
+    }
+    template<int Index,class vobj> inline vobj transposeSpin(const vobj &lhs){
+      return transposeIndex<SpinIndex>(lhs);
+    }
+    template<int Index,class vobj> inline vobj transposeColour(const vobj &lhs){
+      return transposeIndex<ColourIndex>(lhs);
+    }
+
+    //////////////////////////////////////////
+    // Trace lattice and non-lattice
+    //////////////////////////////////////////
+    template<int Index,class vobj>
+    inline auto traceSpin(const Lattice<vobj> &lhs) -> Lattice<decltype(traceIndex<SpinIndex>(lhs._odata[0]))>
+    {
+      return traceIndex<SpinIndex>(lhs);
+    }
+    template<int Index,class vobj>
+    inline auto traceColour(const Lattice<vobj> &lhs) -> Lattice<decltype(traceIndex<ColourIndex>(lhs._odata[0]))>
+    {
+      return traceIndex<ColourIndex>(lhs);
+    }
+    template<int Index,class vobj>
+    inline auto traceSpin(const vobj &lhs) -> Lattice<decltype(traceIndex<SpinIndex>(lhs))>
+    {
+      return traceIndex<SpinIndex>(lhs);
+    }
+    template<int Index,class vobj>
+    inline auto traceColour(const vobj &lhs) -> Lattice<decltype(traceIndex<ColourIndex>(lhs))>
+    {
+      return traceIndex<ColourIndex>(lhs);
+    }
 
 }   //namespace QCD
 } // Grid

From 8b9a9aea81e59cbcde007e57e68ecef0c7111448 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 01:00:46 +0100
Subject: [PATCH 333/429] More TODO as ever

---
 TODO | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/TODO b/TODO
index 9a428f98..5c7cf35d 100644
--- a/TODO
+++ b/TODO
@@ -1,11 +1,15 @@
 ================================================================
 *** Hacks and bug fixes to clean up and Audits
 ================================================================
-* Base class to share common code between vRealF, VComplexF etc... done
-  - Performance check on Guido's reimplementation strategy  - (GUIDO) tested and no difference was found, merged
+
+*  Extract/merge/set cleanup ; too many variants; rationalise and call simpler ones
+*  Used #define repetitive sequences to minimise code.
+*  Rewrite core tensor arithmetic support to be more systematic
+*  Ensure we ET as much as possible; move unop functions into ET framework.
+   - tests with expression args to all functions
+
 
 * FIXME audit
-
 * const audit
 
 Insert/Extract

From 0f7cf40867df7b6c1efc552b209943b1fb49d63c Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 01:01:21 +0100
Subject: [PATCH 334/429] Moving where in to the expression template system;
 deprecate

---
 lib/lattice/Lattice_where.h | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/lib/lattice/Lattice_where.h b/lib/lattice/Lattice_where.h
index 84d7471c..c83d80c1 100644
--- a/lib/lattice/Lattice_where.h
+++ b/lib/lattice/Lattice_where.h
@@ -8,13 +8,14 @@ namespace Grid {
 //                              and blow away the tensor structures.
 //
 template<class vobj,class iobj>
-inline void where(Lattice<vobj> &ret,const Lattice<iobj> &predicate,Lattice<vobj> &iftrue,Lattice<vobj> &iffalse)
+inline void whereWolf(Lattice<vobj> &ret,const Lattice<iobj> &predicate,Lattice<vobj> &iftrue,Lattice<vobj> &iffalse)
 {
   conformable(iftrue,iffalse);
   conformable(iftrue,predicate);
   conformable(iftrue,ret);
 
   GridBase *grid=iftrue._grid;
+
   typedef typename vobj::scalar_object scalar_object;
   typedef typename vobj::scalar_type scalar_type;
   typedef typename vobj::vector_type vector_type;
@@ -43,7 +44,7 @@ PARALLEL_FOR_LOOP
 }
 
 template<class vobj,class iobj>
-inline Lattice<vobj> where(const Lattice<iobj> &predicate,Lattice<vobj> &iftrue,Lattice<vobj> &iffalse)
+inline Lattice<vobj> whereWolf(const Lattice<iobj> &predicate,Lattice<vobj> &iftrue,Lattice<vobj> &iffalse)
 {
   conformable(iftrue,iffalse);
   conformable(iftrue,predicate);

From 4d81d402b977c2964eadc22cc22ece70eeb49f24 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 01:01:55 +0100
Subject: [PATCH 335/429] Extra lattice unaries

---
 lib/lattice/Lattice_unary.h | 45 -------------------------------------
 1 file changed, 45 deletions(-)

diff --git a/lib/lattice/Lattice_unary.h b/lib/lattice/Lattice_unary.h
index fcdcc484..91b39007 100644
--- a/lib/lattice/Lattice_unary.h
+++ b/lib/lattice/Lattice_unary.h
@@ -3,51 +3,6 @@
 
 namespace Grid {
 
-  //////////////////////////////////////////////////////////////////////////////////////////////////////
-  //  avoid copy back routines for mult, mac, sub, add
-  //////////////////////////////////////////////////////////////////////////////////////////////////////
-  template<class obj> Lattice<obj> sqrt(const Lattice<obj> &rhs){
-    Lattice<obj> ret(rhs._grid);
-    ret.checkerboard = rhs.checkerboard;
-    conformable(ret,rhs);
-PARALLEL_FOR_LOOP
-    for(int ss=0;ss<rhs._grid->oSites();ss++){
-      ret._odata[ss]=sqrt(rhs._odata[ss]);
-    }
-    return ret;
-  }
-  template<class obj> Lattice<obj> rsqrt(const Lattice<obj> &rhs){
-    Lattice<obj> ret(rhs._grid);
-    ret.checkerboard = rhs.checkerboard;
-    conformable(ret,rhs);
-PARALLEL_FOR_LOOP
-    for(int ss=0;ss<rhs._grid->oSites();ss++){
-      ret._odata[ss]=rsqrt(rhs._odata[ss]);
-    }
-    return ret;
-  }
-  template<class obj> Lattice<obj> sin(const Lattice<obj> &rhs){
-    Lattice<obj> ret(rhs._grid);
-    ret.checkerboard = rhs.checkerboard;
-    conformable(ret,rhs);
-PARALLEL_FOR_LOOP
-    for(int ss=0;ss<rhs._grid->oSites();ss++){
-      ret._odata[ss]=sin(rhs._odata[ss]);
-    }
-    return ret;
-  }
-  template<class obj> Lattice<obj> cos(const Lattice<obj> &rhs){
-    Lattice<obj> ret(rhs._grid);
-    ret.checkerboard = rhs.checkerboard;
-    conformable(ret,rhs);
-PARALLEL_FOR_LOOP
-    for(int ss=0;ss<rhs._grid->oSites();ss++){
-      ret._odata[ss]=cos(rhs._odata[ss]);
-    }
-    return ret;
-  }
-
-
   template<class obj> Lattice<obj> pow(const Lattice<obj> &rhs,RealD y){
     Lattice<obj> ret(rhs._grid);
     ret.checkerboard = rhs.checkerboard;

From 6a697be06081c1a2be28c13bc510616a9369579f Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 01:02:31 +0100
Subject: [PATCH 336/429] Transpose always returns self image

---
 lib/lattice/Lattice_transpose.h | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/lib/lattice/Lattice_transpose.h b/lib/lattice/Lattice_transpose.h
index c8e7433f..73ef1747 100644
--- a/lib/lattice/Lattice_transpose.h
+++ b/lib/lattice/Lattice_transpose.h
@@ -24,8 +24,7 @@ PARALLEL_FOR_LOOP
     // Index level dependent transpose
     ////////////////////////////////////////////////////////////////////////////////////////////////////
     template<int Index,class vobj>
-    inline auto transposeIndex(const Lattice<vobj> &lhs)
-      -> Lattice<decltype(transposeIndex<Index>(lhs._odata[0]))>
+    inline auto transposeIndex(const Lattice<vobj> &lhs) -> Lattice<decltype(transposeIndex<Index>(lhs._odata[0]))>
     {
       Lattice<decltype(transposeIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
 PARALLEL_FOR_LOOP

From 02885ee58398f3fe774212b727c3236beac7a7cd Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 01:03:28 +0100
Subject: [PATCH 337/429] trying to find a way to remove functions from the ET
 system using explicit expression closure statements. Not sure if this works
 yet

---
 lib/lattice/Lattice_reduction.h | 25 +++++++++++++++++++++++++
 1 file changed, 25 insertions(+)

diff --git a/lib/lattice/Lattice_reduction.h b/lib/lattice/Lattice_reduction.h
index a6d3e0dc..922821ed 100644
--- a/lib/lattice/Lattice_reduction.h
+++ b/lib/lattice/Lattice_reduction.h
@@ -51,6 +51,31 @@ PARALLEL_FOR_LOOP
       return nrm;
     }
 
+    template<class Op,class T1>
+      inline auto sum(const LatticeUnaryExpression<Op,T1> & expr)
+      ->typename decltype(expr.first.func(eval(0,std::get<0>(expr.second))))::scalar_object
+    {
+      return sum(closure(expr));
+    }
+
+    template<class Op,class T1,class T2>
+      inline auto sum(const LatticeBinaryExpression<Op,T1,T2> & expr)
+      ->typename decltype(expr.first.func(eval(0,std::get<0>(expr.second)),eval(0,std::get<1>(expr.second))))::scalar_object
+    {
+      return sum(closure(expr));
+    }
+
+
+    template<class Op,class T1,class T2,class T3>
+      inline auto sum(const LatticeTrinaryExpression<Op,T1,T2,T3> & expr)
+      ->typename decltype(expr.first.func(eval(0,std::get<0>(expr.second)),
+				 eval(0,std::get<1>(expr.second)),
+				 eval(0,std::get<2>(expr.second))
+				 ))::scalar_object
+    {
+      return sum(closure(expr));
+    }
+
     template<class vobj>
     inline typename vobj::scalar_object sum(const Lattice<vobj> &arg){
 

From 38e8a61035ca60b2a548e189cca1e6d55dba3268 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 01:04:32 +0100
Subject: [PATCH 338/429] Moving more into the ET system

---
 lib/lattice/Lattice_reality.h | 31 -------------------------------
 1 file changed, 31 deletions(-)

diff --git a/lib/lattice/Lattice_reality.h b/lib/lattice/Lattice_reality.h
index 8c8bc17a..29a7976a 100644
--- a/lib/lattice/Lattice_reality.h
+++ b/lib/lattice/Lattice_reality.h
@@ -27,37 +27,6 @@ PARALLEL_FOR_LOOP
         return ret;
     };
 
-    template<class vobj> inline auto real(const Lattice<vobj> &z) -> Lattice<vobj>
-    {
-      Lattice<vobj> ret(z._grid);
-PARALLEL_FOR_LOOP
-        for(int ss=0;ss<z._grid->oSites();ss++){
-            ret._odata[ss] = real(z._odata[ss]);
-        }
-      return ret;
-    }
-
-    template<class vobj> inline auto imag(const Lattice<vobj> &z) -> Lattice<vobj>
-    {
-      Lattice<vobj> ret(z._grid);
-PARALLEL_FOR_LOOP
-        for(int ss=0;ss<z._grid->oSites();ss++){
-            ret._odata[ss] = imag(z._odata[ss]);
-        }
-      return ret;
-    }
-
-
-    template<class vobj> inline auto Ta(const Lattice<vobj> &z) -> Lattice<decltype(Ta(z._odata[0]))>
-    {
-      Lattice<decltype(Ta(z._odata[0]))> ret(z._grid);
-PARALLEL_FOR_LOOP
-        for(int ss=0;ss<z._grid->oSites();ss++){
-            ret._odata[ss] = Ta(z._odata[ss]);
-        }
-      return ret;
-    }
-
 
 }
 #endif

From 9bdb8ffb3f6f8ba7618a32a7d1b79a1321647e5f Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 01:05:06 +0100
Subject: [PATCH 339/429] Allow sparse occupation of vectors in some cases

---
 lib/lattice/Lattice_peekpoke.h | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/lib/lattice/Lattice_peekpoke.h b/lib/lattice/Lattice_peekpoke.h
index 9416226a..5455fdd6 100644
--- a/lib/lattice/Lattice_peekpoke.h
+++ b/lib/lattice/Lattice_peekpoke.h
@@ -117,7 +117,9 @@ PARALLEL_FOR_LOOP
       int Nsimd = grid->Nsimd();
 
       assert( l.checkerboard == l._grid->CheckerBoard(site));
-      assert( sizeof(sobj)*Nsimd == sizeof(vobj));
+
+      // FIXME
+      //      assert( sizeof(sobj)*Nsimd == sizeof(vobj));
 
       int rank,odx,idx;
       grid->GlobalCoorToRankIndex(rank,odx,idx,site);
@@ -132,6 +134,7 @@ PARALLEL_FOR_LOOP
       return;
     };
 
+
     //////////////////////////////////////////////////////////
     // Peek a scalar object from the SIMD array
     //////////////////////////////////////////////////////////

From 2303a2231b0beeea023bdfc71273b1baa0dd96f3 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 01:05:39 +0100
Subject: [PATCH 340/429] Allow real comparisons and expressions in comparisons

---
 lib/lattice/Lattice_comparison_utils.h | 299 +++++++++++++++----------
 1 file changed, 178 insertions(+), 121 deletions(-)

diff --git a/lib/lattice/Lattice_comparison_utils.h b/lib/lattice/Lattice_comparison_utils.h
index d5e3b3e5..d25c49f2 100644
--- a/lib/lattice/Lattice_comparison_utils.h
+++ b/lib/lattice/Lattice_comparison_utils.h
@@ -5,140 +5,197 @@ namespace Grid {
 
   /////////////////////////////////////////
   // This implementation is a bit poor.
-  // Only support logical operations (== etc)
-  // on scalar objects. Strip any tensor structures.
+  //
+  // Only support relational logical operations (<, >  etc)
+  // on scalar objects. Therefore can strip any tensor structures.
+  //
   // Should guard this with isGridTensor<> enable if?
   /////////////////////////////////////////
-    // Generic list of functors
-    template<class lobj,class robj> class veq {
+  //
+  // Generic list of functors
+  //
+  template<class lobj,class robj> class veq {
+  public:
+    vInteger operator()(const lobj &lhs, const robj &rhs)
+    { 
+      return (lhs) == (rhs);
+    }
+  };
+  template<class lobj,class robj> class vne {
+  public:
+    vInteger operator()(const lobj &lhs, const robj &rhs)
+    { 
+      return (lhs) != (rhs);
+    }
+  };
+  template<class lobj,class robj> class vlt {
+  public:
+    vInteger operator()(const lobj &lhs, const robj &rhs)
+    { 
+      return (lhs) < (rhs);
+    }
+  };
+  template<class lobj,class robj> class vle {
+  public:
+    vInteger operator()(const lobj &lhs, const robj &rhs)
+    { 
+      return (lhs) <= (rhs);
+    }
+  };
+  template<class lobj,class robj> class vgt {
+  public:
+    vInteger operator()(const lobj &lhs, const robj &rhs)
+    { 
+      return (lhs) > (rhs);
+    }
+  };
+  template<class lobj,class robj> class vge {
     public:
-      vInteger operator()(const lobj &lhs, const robj &rhs)
-	{ 
-	  return TensorRemove(lhs) == TensorRemove(rhs);
-	}
-    };
-    template<class lobj,class robj> class vne {
-    public:
-      vInteger operator()(const lobj &lhs, const robj &rhs)
-	{ 
-	  return TensorRemove(lhs) != TensorRemove(rhs);
-	}
-    };
-    template<class lobj,class robj> class vlt {
-    public:
-      vInteger operator()(const lobj &lhs, const robj &rhs)
-	{ 
-	  return TensorRemove(lhs) < TensorRemove(rhs);
-	}
-    };
-    template<class lobj,class robj> class vle {
-    public:
-      vInteger operator()(const lobj &lhs, const robj &rhs)
-	{ 
-	  return TensorRemove(lhs) <= TensorRemove(rhs);
-	}
-    };
-    template<class lobj,class robj> class vgt {
-    public:
-      vInteger operator()(const lobj &lhs, const robj &rhs)
-	{ 
-	  return TensorRemove(lhs) > TensorRemove(rhs);
-	}
-    };
-    template<class lobj,class robj> class vge {
-    public:
-      vInteger operator()(const lobj &lhs, const robj &rhs)
-	{ 
-	  return TensorRemove(lhs) >= TensorRemove(rhs);
-	}
-    };
+    vInteger operator()(const lobj &lhs, const robj &rhs)
+    { 
+      return (lhs) >= (rhs);
+    }
+  };
+  
+  // Generic list of functors
+  template<class lobj,class robj> class seq {
+  public:
+    Integer operator()(const lobj &lhs, const robj &rhs)
+    { 
+      return (lhs) == (rhs);
+    }
+  };
+  template<class lobj,class robj> class sne {
+  public:
+    Integer operator()(const lobj &lhs, const robj &rhs)
+    { 
+      return (lhs) != (rhs);
+    }
+  };
+  template<class lobj,class robj> class slt {
+  public:
+    Integer operator()(const lobj &lhs, const robj &rhs)
+    { 
+      return (lhs) < (rhs);
+    }
+  };
+  template<class lobj,class robj> class sle {
+  public:
+    Integer operator()(const lobj &lhs, const robj &rhs)
+    { 
+      return (lhs) <= (rhs);
+    }
+  };
+  template<class lobj,class robj> class sgt {
+  public:
+    Integer operator()(const lobj &lhs, const robj &rhs)
+    { 
+      return (lhs) > (rhs);
+    }
+  };
+  template<class lobj,class robj> class sge {
+  public:
+    Integer operator()(const lobj &lhs, const robj &rhs)
+    { 
+      return (lhs) >= (rhs);
+    }
+  };
 
-    // Generic list of functors
-    template<class lobj,class robj> class seq {
-    public:
-      Integer operator()(const lobj &lhs, const robj &rhs)
-	{ 
-	  return TensorRemove(lhs) == TensorRemove(rhs);
-	}
-    };
-    template<class lobj,class robj> class sne {
-    public:
-      Integer operator()(const lobj &lhs, const robj &rhs)
-	{ 
-	  return TensorRemove(lhs) != TensorRemove(rhs);
-	}
-    };
-    template<class lobj,class robj> class slt {
-    public:
-      Integer operator()(const lobj &lhs, const robj &rhs)
-	{ 
-	  return TensorRemove(lhs) < TensorRemove(rhs);
-	}
-    };
-    template<class lobj,class robj> class sle {
-    public:
-      Integer operator()(const lobj &lhs, const robj &rhs)
-	{ 
-	  return TensorRemove(lhs) <= TensorRemove(rhs);
-	}
-    };
-    template<class lobj,class robj> class sgt {
-    public:
-      Integer operator()(const lobj &lhs, const robj &rhs)
-	{ 
-	  return TensorRemove(lhs) > TensorRemove(rhs);
-	}
-    };
-    template<class lobj,class robj> class sge {
-    public:
-      Integer operator()(const lobj &lhs, const robj &rhs)
-	{ 
-	  return TensorRemove(lhs) >= TensorRemove(rhs);
-	}
-    };
-
-
-    //////////////////////////////////////////////////////////////////////////////////////////////////////
-    // Integer gets extra relational functions. Could also implement these for RealF, RealD etc..
-    //////////////////////////////////////////////////////////////////////////////////////////////////////
-    template<class sfunctor> 
-    inline vInteger Comparison(sfunctor sop,const vInteger & lhs, const vInteger & rhs)
+  //////////////////////////////////////////////////////////////////////////////////////////////////////
+  // Integer and real get extra relational functions.
+  //////////////////////////////////////////////////////////////////////////////////////////////////////
+  template<class sfunctor, class vsimd,IfNotComplex<vsimd> = 0> 
+    inline vInteger Comparison(sfunctor sop,const vsimd & lhs, const vsimd & rhs)
     {
-      std::vector<Integer> vlhs(vInteger::Nsimd());   // Use functors to reduce this to single implementation
-      std::vector<Integer> vrhs(vInteger::Nsimd());
+      typedef typename vsimd::scalar_type scalar;
+      std::vector<scalar> vlhs(vsimd::Nsimd());   // Use functors to reduce this to single implementation
+      std::vector<scalar> vrhs(vsimd::Nsimd());
+      std::vector<Integer> vpred(vsimd::Nsimd());
       vInteger ret;
-      extract<vInteger,Integer>(lhs,vlhs);
-      extract<vInteger,Integer>(rhs,vrhs);
-      for(int s=0;s<vInteger::Nsimd();s++){
-	vlhs[s] = sop(vlhs[s],vrhs[s]);
+      extract<vsimd,scalar>(lhs,vlhs);
+      extract<vsimd,scalar>(rhs,vrhs);
+      for(int s=0;s<vsimd::Nsimd();s++){
+	vpred[s] = sop(vlhs[s],vrhs[s]);
       }
-      merge<vInteger,Integer>(ret,vlhs);
+      merge<vInteger,Integer>(ret,vpred);
       return ret;
     }
-    inline vInteger operator < (const vInteger & lhs, const vInteger & rhs)
+
+  template<class sfunctor, class vsimd,IfNotComplex<vsimd> = 0> 
+    inline vInteger Comparison(sfunctor sop,const vsimd & lhs, const typename vsimd::scalar_type & rhs)
     {
-      return Comparison(slt<Integer,Integer>(),lhs,rhs);
+      typedef typename vsimd::scalar_type scalar;
+      std::vector<scalar> vlhs(vsimd::Nsimd());   // Use functors to reduce this to single implementation
+      std::vector<Integer> vpred(vsimd::Nsimd());
+      vInteger ret;
+      extract<vsimd,scalar>(lhs,vlhs);
+      for(int s=0;s<vsimd::Nsimd();s++){
+	vpred[s] = sop(vlhs[s],rhs);
+      }
+      merge<vInteger,Integer>(ret,vpred);
+      return ret;
     }
-    inline vInteger operator <= (const vInteger & lhs, const vInteger & rhs)
+
+  template<class sfunctor, class vsimd,IfNotComplex<vsimd> = 0> 
+    inline vInteger Comparison(sfunctor sop,const typename vsimd::scalar_type & lhs, const vsimd & rhs)
     {
-      return Comparison(sle<Integer,Integer>(),lhs,rhs);
-    }
-    inline vInteger operator > (const vInteger & lhs, const vInteger & rhs)
-    {
-      return Comparison(sgt<Integer,Integer>(),lhs,rhs);
-    }
-    inline vInteger operator >= (const vInteger & lhs, const vInteger & rhs)
-    {
-      return Comparison(sge<Integer,Integer>(),lhs,rhs);
-    }
-    inline vInteger operator == (const vInteger & lhs, const vInteger & rhs)
-    {
-      return Comparison(seq<Integer,Integer>(),lhs,rhs);
-    }
-    inline vInteger operator != (const vInteger & lhs, const vInteger & rhs)
-    {
-      return Comparison(sne<Integer,Integer>(),lhs,rhs);
+      typedef typename vsimd::scalar_type scalar;
+      std::vector<scalar> vrhs(vsimd::Nsimd());   // Use functors to reduce this to single implementation
+      std::vector<Integer> vpred(vsimd::Nsimd());
+      vInteger ret;
+      extract<vsimd,scalar>(rhs,vrhs);
+      for(int s=0;s<vsimd::Nsimd();s++){
+	vpred[s] = sop(lhs,vrhs[s]);
+      }
+      merge<vInteger,Integer>(ret,vpred);
+      return ret;
     }
+
+#define DECLARE_RELATIONAL(op,functor) \
+  template<class vsimd,IfSimd<vsimd> = 0>\
+    inline vInteger operator op (const vsimd & lhs, const vsimd & rhs)\
+    {\
+      typedef typename vsimd::scalar_type scalar;\
+      return Comparison(functor<scalar,scalar>(),lhs,rhs);\
+    }\
+  template<class vsimd,IfSimd<vsimd> = 0>\
+    inline vInteger operator op (const vsimd & lhs, const typename vsimd::scalar_type & rhs) \
+    {\
+      typedef typename vsimd::scalar_type scalar;\
+      return Comparison(functor<scalar,scalar>(),lhs,rhs);\
+    }\
+  template<class vsimd,IfSimd<vsimd> = 0>\
+    inline vInteger operator op (const typename vsimd::scalar_type & lhs, const vsimd & rhs) \
+    {\
+      typedef typename vsimd::scalar_type scalar;\
+      return Comparison(functor<scalar,scalar>(),lhs,rhs);\
+    }\
+  template<class vsimd>\
+    inline vInteger operator op(const iScalar<vsimd> &lhs,const iScalar<vsimd> &rhs)\
+    {									\
+      return lhs._internal op rhs._internal;				\
+    }									\
+  template<class vsimd>\
+    inline vInteger operator op(const iScalar<vsimd> &lhs,const typename vsimd::scalar_type &rhs) \
+    {									\
+      return lhs._internal op rhs;					\
+    }									\
+  template<class vsimd>\
+    inline vInteger operator op(const typename vsimd::scalar_type &lhs,const iScalar<vsimd> &rhs) \
+    {									\
+      return lhs op rhs._internal;					\
+    }									
+
+
+DECLARE_RELATIONAL(<,slt);
+DECLARE_RELATIONAL(<=,sle);
+DECLARE_RELATIONAL(>,sgt);
+DECLARE_RELATIONAL(>=,sge);
+DECLARE_RELATIONAL(==,seq);
+DECLARE_RELATIONAL(!=,sne);
+
+#undef DECLARE_RELATIONAL
+
 }
 
 

From bc238177b70b013649125a11a51e7ad404464ae6 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 01:05:57 +0100
Subject: [PATCH 341/429] real comps and expression comps

---
 lib/lattice/Lattice_comparison.h | 16 ++++++++++++++--
 1 file changed, 14 insertions(+), 2 deletions(-)

diff --git a/lib/lattice/Lattice_comparison.h b/lib/lattice/Lattice_comparison.h
index 431cf911..604ea4b9 100644
--- a/lib/lattice/Lattice_comparison.h
+++ b/lib/lattice/Lattice_comparison.h
@@ -11,7 +11,11 @@ namespace Grid {
     //Query supporting bitwise &, |, ^, !
     //Query supporting logical &&, ||, 
     //////////////////////////////////////////////////////////////////////////
-    template<class vfunctor,class lobj,class robj> 
+
+  //////////////////////////////////////////////////////////////////////////
+  // compare lattice to lattice
+  //////////////////////////////////////////////////////////////////////////
+  template<class vfunctor,class lobj,class robj>  
     inline Lattice<vInteger> LLComparison(vfunctor op,const Lattice<lobj> &lhs,const Lattice<robj> &rhs)
     {
       Lattice<vInteger> ret(rhs._grid);
@@ -21,6 +25,9 @@ PARALLEL_FOR_LOOP
         }
         return ret;
     }
+  //////////////////////////////////////////////////////////////////////////
+  // compare lattice to scalar
+  //////////////////////////////////////////////////////////////////////////
     template<class vfunctor,class lobj,class robj> 
     inline Lattice<vInteger> LSComparison(vfunctor op,const Lattice<lobj> &lhs,const robj &rhs)
     {
@@ -31,6 +38,9 @@ PARALLEL_FOR_LOOP
         }
         return ret;
     }
+  //////////////////////////////////////////////////////////////////////////
+  // compare scalar to lattice
+  //////////////////////////////////////////////////////////////////////////
     template<class vfunctor,class lobj,class robj> 
     inline Lattice<vInteger> SLComparison(vfunctor op,const lobj &lhs,const Lattice<robj> &rhs)
     {
@@ -42,6 +52,9 @@ PARALLEL_FOR_LOOP
         return ret;
     }
 
+  //////////////////////////////////////////////////////////////////////////
+  // Map to functors
+  //////////////////////////////////////////////////////////////////////////
     // Less than
    template<class lobj,class robj>
    inline Lattice<vInteger> operator < (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {
@@ -99,7 +112,6 @@ PARALLEL_FOR_LOOP
      return SLComparison(vge<lobj,robj>(),lhs,rhs);
    }
 
-
    // equal
    template<class lobj,class robj>
    inline Lattice<vInteger> operator == (const Lattice<lobj> & lhs, const Lattice<robj> & rhs) {

From 667abe888aa3a7133f930628e5fb23da06564d63 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 01:06:56 +0100
Subject: [PATCH 342/429] Cosmetic

---
 lib/lattice/Lattice_base.h | 13 ++++---------
 1 file changed, 4 insertions(+), 9 deletions(-)

diff --git a/lib/lattice/Lattice_base.h b/lib/lattice/Lattice_base.h
index 5a17826b..cd2fa855 100644
--- a/lib/lattice/Lattice_base.h
+++ b/lib/lattice/Lattice_base.h
@@ -93,7 +93,7 @@ PARALLEL_FOR_LOOP
     }
     return *this;
   }
-  template <typename Op, typename T1,typename T2>             strong_inline Lattice<vobj> & operator=(const LatticeBinaryExpression<Op,T1,T2> &expr)
+  template <typename Op, typename T1,typename T2> strong_inline Lattice<vobj> & operator=(const LatticeBinaryExpression<Op,T1,T2> &expr)
   {
     GridBase *egrid(nullptr);
     GridFromExpression(egrid,expr);
@@ -131,8 +131,8 @@ PARALLEL_FOR_LOOP
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
 #ifdef STREAMING_STORES
-      vobj tmp = eval(ss,expr);
-      vstream(_odata[ss] ,tmp);
+      //vobj tmp = eval(ss,expr);
+      vstream(_odata[ss] ,eval(ss,expr));
 #else
       _odata[ss] = eval(ss,expr);
 #endif
@@ -196,12 +196,7 @@ PARALLEL_FOR_LOOP
     _odata.resize(_grid->oSites());
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
-#ifdef STREAMING_STORES
-      vobj tmp = eval(ss,expr);
-      vstream(_odata[ss] ,tmp);
-#else
       _odata[ss]=eval(ss,expr);
-#endif
     }
   };
 
@@ -254,7 +249,7 @@ PARALLEL_FOR_LOOP
         Lattice<vobj> ret(lhs._grid);
 PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){
-            ret._odata[ss] = lhs._odata[ss]/rhs._odata[ss];
+	  ret._odata[ss] = lhs._odata[ss]*pow(rhs._odata[ss],-1.0);
         }
         return ret;
     };

From 84a207f2e60d2d2cb59bcc090f841739c2239fdd Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 01:07:25 +0100
Subject: [PATCH 343/429] Where and many other functions (sin cos abs log exp)
 into ET system

---
 lib/lattice/Lattice_ET.h | 118 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 118 insertions(+)

diff --git a/lib/lattice/Lattice_ET.h b/lib/lattice/Lattice_ET.h
index 2a7172ea..caa45e69 100644
--- a/lib/lattice/Lattice_ET.h
+++ b/lib/lattice/Lattice_ET.h
@@ -9,6 +9,37 @@
 
 namespace Grid {
 
+  ////////////////////////////////////////////////////
+  // Predicated where support
+  ////////////////////////////////////////////////////
+  template<class iobj,class vobj,class robj>
+    inline vobj predicatedWhere(const iobj &predicate,const vobj &iftrue,const robj &iffalse) {
+
+    typename std::remove_const<vobj>::type ret;
+
+    typedef typename vobj::scalar_object scalar_object;
+    typedef typename vobj::scalar_type scalar_type;
+    typedef typename vobj::vector_type vector_type;
+
+    const int Nsimd = vobj::vector_type::Nsimd();
+    const int words = sizeof(vobj)/sizeof(vector_type);
+
+    std::vector<Integer> mask(Nsimd);
+    std::vector<scalar_object> truevals (Nsimd);
+    std::vector<scalar_object> falsevals(Nsimd);
+
+    extract(iftrue   ,truevals);
+    extract(iffalse  ,falsevals);
+    extract<vInteger,Integer>(TensorRemove(predicate),mask);
+
+    for(int s=0;s<Nsimd;s++){
+      if (mask[s]) falsevals[s]=truevals[s];
+    }
+
+    merge(ret,falsevals);
+    return ret;
+  }
+
 ////////////////////////////////////////////
 // recursive evaluation of expressions; Could
 // switch to generic approach with variadics, a la
@@ -142,10 +173,23 @@ template <class arg> struct name\
 };
 
 GridUnopClass(UnarySub,-a);
+GridUnopClass(UnaryNot,Not(a));
 GridUnopClass(UnaryAdj,adj(a));
 GridUnopClass(UnaryConj,conjugate(a));
 GridUnopClass(UnaryTrace,trace(a));
 GridUnopClass(UnaryTranspose,transpose(a));
+GridUnopClass(UnaryTa,Ta(a));
+GridUnopClass(UnaryReal,real(a));
+GridUnopClass(UnaryImag,imag(a));
+GridUnopClass(UnaryToReal,toReal(a));
+GridUnopClass(UnaryToComplex,toComplex(a));
+GridUnopClass(UnaryAbs,abs(a));
+GridUnopClass(UnarySqrt,sqrt(a));
+GridUnopClass(UnaryRsqrt,rsqrt(a));
+GridUnopClass(UnarySin,sin(a));
+GridUnopClass(UnaryCos,cos(a));
+GridUnopClass(UnaryLog,log(a));
+GridUnopClass(UnaryExp,exp(a));
 
 ////////////////////////////////////////////
 // Binary operators
@@ -163,6 +207,28 @@ GridBinOpClass(BinaryAdd,lhs+rhs);
 GridBinOpClass(BinarySub,lhs-rhs);
 GridBinOpClass(BinaryMul,lhs*rhs);
 
+GridBinOpClass(BinaryAnd   ,lhs&rhs);
+GridBinOpClass(BinaryOr    ,lhs|rhs);
+GridBinOpClass(BinaryAndAnd,lhs&&rhs);
+GridBinOpClass(BinaryOrOr  ,lhs||rhs);
+
+////////////////////////////////////////////////////
+// Trinary conditional op
+////////////////////////////////////////////////////
+#define GridTrinOpClass(name,combination)\
+template <class predicate,class left, class right>	\
+struct name\
+{\
+  static auto inline func(const predicate &pred,const left &lhs,const right &rhs)-> decltype(combination) const \
+    {\
+      return combination;\
+    }\
+}
+
+GridTrinOpClass(TrinaryWhere,(predicatedWhere<predicate, \
+			       typename std::remove_reference<left>::type, \
+			       typename std::remove_reference<right>::type> (pred,lhs,rhs)));
+
 ////////////////////////////////////////////
 // Operator syntactical glue
 ////////////////////////////////////////////
@@ -218,15 +284,67 @@ template <typename T1,typename T2,typename T3> inline auto op(const T1 &pred,con
 ////////////////////////
 
 GRID_DEF_UNOP(operator -,UnarySub);
+GRID_DEF_UNOP(Not,UnaryNot);
+GRID_DEF_UNOP(operator !,UnaryNot);
 GRID_DEF_UNOP(adj,UnaryAdj);
 GRID_DEF_UNOP(conjugate,UnaryConj);
 GRID_DEF_UNOP(trace,UnaryTrace);
 GRID_DEF_UNOP(transpose,UnaryTranspose);
+GRID_DEF_UNOP(Ta,UnaryTa);
+GRID_DEF_UNOP(real,UnaryReal);
+GRID_DEF_UNOP(imag,UnaryImag);
+GRID_DEF_UNOP(toReal,UnaryToReal);
+GRID_DEF_UNOP(toComplex,UnaryToComplex);
+GRID_DEF_UNOP(abs  ,UnaryAbs); //abs overloaded in cmath C++98; DON'T do the abs-fabs-dabs-labs thing
+GRID_DEF_UNOP(sqrt ,UnarySqrt);
+GRID_DEF_UNOP(rsqrt,UnarySqrt);
+GRID_DEF_UNOP(sin  ,UnarySin);
+GRID_DEF_UNOP(cos  ,UnaryCos);
+GRID_DEF_UNOP(log  ,UnaryLog);
+GRID_DEF_UNOP(exp  ,UnaryExp);
 
 GRID_DEF_BINOP(operator+,BinaryAdd);
 GRID_DEF_BINOP(operator-,BinarySub);
 GRID_DEF_BINOP(operator*,BinaryMul);
 
+GRID_DEF_BINOP(operator&,BinaryAnd);
+GRID_DEF_BINOP(operator|,BinaryOr);
+GRID_DEF_BINOP(operator&&,BinaryAndAnd);
+GRID_DEF_BINOP(operator||,BinaryOrOr);
+
+GRID_DEF_TRINOP(where,TrinaryWhere);
+
+/////////////////////////////////////////////////////////////
+// Closure convenience to force expression to evaluate
+/////////////////////////////////////////////////////////////
+template<class Op,class T1>
+  auto closure(const LatticeUnaryExpression<Op,T1> & expr)
+  -> Lattice<decltype(expr.first.func(eval(0,std::get<0>(expr.second))))>
+{
+  Lattice<decltype(expr.first.func(eval(0,std::get<0>(expr.second))))> ret(expr);
+  return ret;
+}
+template<class Op,class T1, class T2>
+  auto closure(const LatticeBinaryExpression<Op,T1,T2> & expr)
+  -> Lattice<decltype(expr.first.func(eval(0,std::get<0>(expr.second)),
+				      eval(0,std::get<1>(expr.second))))>
+{
+  Lattice<decltype(expr.first.func(eval(0,std::get<0>(expr.second)),
+				   eval(0,std::get<1>(expr.second))))> ret(expr);
+  return ret;
+}
+template<class Op,class T1, class T2, class T3>
+  auto closure(const LatticeTrinaryExpression<Op,T1,T2,T3> & expr)
+  -> Lattice<decltype(expr.first.func(eval(0,std::get<0>(expr.second)),
+				      eval(0,std::get<1>(expr.second)),
+				      eval(0,std::get<2>(expr.second))))>
+{
+  Lattice<decltype(expr.first.func(eval(0,std::get<0>(expr.second)),
+				   eval(0,std::get<1>(expr.second)),
+				   eval(0,std::get<2>(expr.second))))> ret(expr);
+  return ret;
+}
+
 #undef GRID_UNOP
 #undef GRID_BINOP
 #undef GRID_TRINOP

From a5bb9472b6d025b46ee70cf32f03a3f473746635 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 01:07:58 +0100
Subject: [PATCH 344/429] WHere should always have precisely matching types in
 ET system

---
 lib/algorithms/CoarsenedMatrix.h | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/lib/algorithms/CoarsenedMatrix.h b/lib/algorithms/CoarsenedMatrix.h
index 96a40fe7..fe801106 100644
--- a/lib/algorithms/CoarsenedMatrix.h
+++ b/lib/algorithms/CoarsenedMatrix.h
@@ -183,7 +183,7 @@ namespace Grid {
 	  int dir   = geom.directions[p];
 	  int disp  = geom.displacements[p];
 
-	  int block=(FineGrid->_rdimensions[dir])/(Grid()->_rdimensions[dir]);
+	  Integer block=(FineGrid->_rdimensions[dir])/(Grid()->_rdimensions[dir]);
 
 	  LatticeCoordinate(coor,dir);
 
@@ -204,8 +204,8 @@ namespace Grid {
 	    oblock = where(mod(coor,block)==(block-1),Mphi,zz);
 	    iblock = where(mod(coor,block)!=(block-1),Mphi,zz);
 	  } else if ( disp==-1 ) {
-	    oblock = where(mod(coor,block)==0,Mphi,zz);
-	    iblock = where(mod(coor,block)!=0,Mphi,zz);
+	    oblock = where(mod(coor,block)==(Integer)0,Mphi,zz);
+	    iblock = where(mod(coor,block)!=(Integer)0,Mphi,zz);
 	  } else {
 	    assert(0);
 	  }

From 353c9598c113311f18c9d1d44bb9d87cc8a806f0 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 01:08:29 +0100
Subject: [PATCH 345/429] Logical elemenet by element on tensors

---
 lib/Tensors.h | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/lib/Tensors.h b/lib/Tensors.h
index e585b116..a2246f4a 100644
--- a/lib/Tensors.h
+++ b/lib/Tensors.h
@@ -14,5 +14,6 @@
 #include <tensors/Tensor_reality.h>
 #include <tensors/Tensor_unary.h>
 #include <tensors/Tensor_extract_merge.h>
-    
+#include <tensors/Tensor_logical.h>
+
 #endif

From ae0873bc77222ba48f3f6edc71621a56a43e8551 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 14 Jun 2015 01:28:54 +0100
Subject: [PATCH 346/429] First cut at SUN support for quenched updates

---
 lib/qcd/utils/SUn.h | 414 ++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 414 insertions(+)
 create mode 100644 lib/qcd/utils/SUn.h

diff --git a/lib/qcd/utils/SUn.h b/lib/qcd/utils/SUn.h
new file mode 100644
index 00000000..3a8b5f4d
--- /dev/null
+++ b/lib/qcd/utils/SUn.h
@@ -0,0 +1,414 @@
+#ifndef QCD_UTIL_SUN_H
+#define QCD_UTIL_SUN_H
+
+
+namespace Grid {
+  namespace QCD {
+
+template<int ncolour>
+class SU {
+public:
+
+  static int generators(void)   { return ncolour*ncolour-1; }
+  static int su2subgroups(void) { return (ncolour*(ncolour-1))/2; }
+
+  template<typename vtype> using iSUnMatrix              = iScalar<iScalar<iMatrix<vtype, ncolour> > > ;
+
+  //////////////////////////////////////////////////////////////////////////////////////////////////
+  // Types can be accessed as SU<2>::Matrix , SU<2>::vSUnMatrix, SU<2>::LatticeMatrix etc...
+  //////////////////////////////////////////////////////////////////////////////////////////////////
+  typedef iSUnMatrix<Complex>         Matrix;
+  typedef iSUnMatrix<ComplexF>        MatrixF;
+  typedef iSUnMatrix<ComplexD>        MatrixD;
+
+  typedef iSUnMatrix<vComplex>       vMatrix;
+  typedef iSUnMatrix<vComplexF>      vMatrixF;
+  typedef iSUnMatrix<vComplexD>      vMatrixD;
+
+  typedef Lattice<vMatrix>     LatticeMatrix;
+  typedef Lattice<vMatrixF>    LatticeMatrixF;
+  typedef Lattice<vMatrixD>    LatticeMatrixD;
+
+  ////////////////////////////////////////////////////////////////////////
+  // There are N^2-1 generators for SU(N).
+  //
+  // We take a traceless hermitian generator basis as follows
+  //
+  // * Normalisation: trace ta tb = 1/2 delta_ab
+  // 
+  // * Off diagonal
+  //    - pairs of rows i1,i2 behaving like pauli matrices signma_x, sigma_y
+  //     
+  //    - there are (Nc-1-i1) slots for i2 on each row [ x  0  x ]                                   
+  //      direct count off each row                                                                    
+  //
+  //    - Sum of all pairs is Nc(Nc-1)/2: proof arithmetic series
+  //
+  //      (Nc-1) + (Nc-2)+...  1      ==> Nc*(Nc-1)/2 
+  //      1+ 2+          +   + Nc-1                        
+  // 
+  //    - There are 2 x Nc (Nc-1)/ 2 of these = Nc^2 - Nc
+  //
+  //    - We enumerate the row-col pairs.
+  //    - for each row col pair there is a (sigma_x) and a (sigma_y) like generator
+  //
+  //
+  //   t^a_ij = { in 0.. Nc(Nc-1)/2 -1} =>  delta_{i,i1} delta_{j,i2} +  delta_{i,i1} delta_{j,i2}  
+  //   t^a_ij = { in Nc(Nc-1)/2 ... Nc^(Nc-1) -1} =>  i delta_{i,i1} delta_{j,i2} - i delta_{i,i1} delta_{j,i2}  
+  //   
+  // * Diagonal; must be traceless and normalised
+  //   - Sequence is 
+  //   N  (1,-1,0,0...)
+  //   N  (1, 1,-2,0...)
+  //   N  (1, 1, 1,-3,0...)
+  //   N  (1, 1, 1, 1,-4,0...)
+  //
+  //   where 1/2 = N^2 (1+.. m^2)etc.... for the m-th diagonal generator
+  //   NB this gives the famous SU3 result for su2 index 8
+  //
+  //   N= sqrt(1/2 . 1/6 ) = 1/2 . 1/sqrt(3) 
+  //
+  //   ( 1      )
+  //   (    1   ) / sqrt(3) /2  = 1/2 lambda_8
+  //   (      -2)
+  ////////////////////////////////////////////////////////////////////////
+  template<class cplx>
+  static void generator(int lieIndex,iSUnMatrix<cplx> &ta){
+    // map lie index to which type of generator
+    int diagIndex;
+    int su2Index;
+    int sigxy;
+    int NNm1 =  ncolour*(ncolour-1);
+    if ( lieIndex>= NNm1 ) {
+      diagIndex = lieIndex -NNm1;
+      generatorDiagonal(diagIndex,ta);
+      return;
+    }
+    sigxy   = lieIndex&0x1;
+    su2Index= lieIndex>>1;
+    if ( sigxy ) generatorSigmaY(su2Index,ta);
+    else         generatorSigmaX(su2Index,ta);
+  }
+  template<class cplx>
+  static void generatorSigmaX(int su2Index,iSUnMatrix<cplx> &ta){
+    ta=zero;
+    int i1,i2;
+    su2SubGroupIndex(i1,i2,su2Index);
+    ta()()(i1,i2)=1.0;
+    ta()()(i2,i1)=1.0;
+    ta= ta *0.5;
+  }
+  template<class cplx>
+  static void generatorSigmaY(int su2Index,iSUnMatrix<cplx> &ta){
+    ta=zero;
+    cplx i(0.0,1.0);
+    int i1,i2;
+    su2SubGroupIndex(i1,i2,su2Index);
+    ta()()(i1,i2)=-i;
+    ta()()(i2,i1)= i;
+    ta= ta *0.5;
+  }
+  template<class cplx>
+  static void generatorDiagonal(int diagIndex,iSUnMatrix<cplx> &ta){
+    ta=zero;
+    int trsq=0;
+    int last=diagIndex+1;
+    for(int i=0;i<=diagIndex;i++){
+      ta()()(i,i) = 1.0;
+      trsq++;
+    }
+    ta()()(last,last) = -last;
+    trsq+=last*last;
+    RealD nrm = 1.0/std::sqrt(2.0*trsq);
+    ta = ta *nrm;
+  }
+  ////////////////////////////////////////////////////////////////////////
+  // Map a su2 subgroup number to the pair of rows that are non zero
+  ////////////////////////////////////////////////////////////////////////
+  static void su2SubGroupIndex(int &i1,int &i2,int su2_index){
+
+    assert( (su2_index>=0) && (su2_index< (ncolour*(ncolour-1))/2) );
+
+    int spare=su2_index;
+    for(i1=0;spare >= (ncolour-1-i1);i1++ ){
+      spare = spare - (ncolour-1-i1);  // remove the Nc-1-i1 terms  
+    }
+    i2=i1+1+spare;
+  }
+  template<class vreal,class vcplx>
+  static void su2Extract(std::vector<Lattice<iSinglet  <vreal> > > &r,
+			       const Lattice<iSUnMatrix<vcplx> > &source, 
+			 int su2_index)
+  {
+    GridBase *grid(source._grid);
+
+    assert(r.size() == 4); // store in 4 real parts
+    
+    for(int i=0;i<4;i++){
+      conformable(r[i],source);
+    }
+    
+    int i1,i2;    
+    su2SubGroupIndex(i1,i2,su2_index);
+
+    /* Compute the b(k) of A_SU(2) = b0 + i sum_k bk sigma_k */ 
+    
+    //    r[0] = toReal(real(peekColour(source,i1,i1)) + real(peekColour(source,i2,i2)));
+    //    r[1] = toReal(imag(peekColour(source,i1,i2)) + imag(peekColour(source,i2,i1)));
+    //    r[2] = toReal(real(peekColour(source,i1,i2)) - real(peekColour(source,i2,i1)));
+    //    r[3] = toReal(imag(peekColour(source,i1,i1)) - imag(peekColour(source,i2,i2)));
+    r[0] = toReal(real(peekColour(source,i1,i1)) + real(peekColour(source,i2,i2)));
+    r[1] = toReal(imag(peekColour(source,i1,i2)) + imag(peekColour(source,i2,i1)));
+    r[2] = toReal(real(peekColour(source,i1,i2)) - real(peekColour(source,i2,i1)));
+    r[3] = toReal(imag(peekColour(source,i1,i1)) - imag(peekColour(source,i2,i2)));
+    
+  }
+
+  template<class vreal,class vcplx>
+  static void su2Insert(const std::vector<Lattice<iSinglet<vreal> > > &r,
+			Lattice<iSUnMatrix<vcplx> > &dest,
+			int su2_index)
+  {
+    typedef typename Lattice<iSUnMatrix<vcplx> >::scalar_type cplx;
+    typedef Lattice<iSinglet<vcplx> > Lcomplex;
+    GridBase * grid = dest._grid;
+
+    assert(r.size() == 4); // store in 4 real parts
+
+    Lcomplex tmp(grid);
+
+    std::vector<Lcomplex > cr(4,grid);
+    for(int i=0;i<r.size();i++){
+      conformable(r[i],dest);
+      cr[i]=toComplex(r[i]);
+    }
+
+    int i1,i2;
+    su2SubGroupIndex(i1,i2,su2_index);
+
+    cplx one   (1.0,0.0);
+    cplx cplx_i(0.0,1.0);
+
+    tmp =   cr[0]*one+ cr[3]*cplx_i;   pokeColour(dest,tmp,i1,i2); 
+    tmp =   cr[2]*one+ cr[1]*cplx_i;   pokeColour(dest,tmp,i1,i2);
+    tmp =  -cr[2]*one+ cr[1]*cplx_i;   pokeColour(dest,tmp,i2,i1);
+    tmp =   cr[0]*one- cr[3]*cplx_i;   pokeColour(dest,tmp,i2,i2);
+  }
+
+  static void SubGroupHeatBath(      GridSerialRNG       &sRNG,
+				     GridParallelRNG     &pRNG,
+			             RealD beta,
+				     LatticeMatrix &link, 
+			       const LatticeMatrix &staple, 
+				     int su2_subgroup,
+				     int nheatbath,
+				     int& ntrials,
+				     int& nfails,
+				     LatticeInteger &wheremask)
+  {
+    GridBase *grid = link._grid;
+
+    LatticeMatrix V(grid);
+    V = link*staple;
+
+    std::vector<LatticeReal> r(4,grid);
+    std::vector<LatticeReal> a(4,grid);
+    su2Extract(r,V,su2_subgroup); // HERE
+
+    LatticeReal r_l(grid);
+    r_l  = r[0]*r[0]+r[1]*r[1]+r[2]*r[2]+r[3]*r[3];
+    r_l = sqrt(r_l);
+
+    LatticeReal ftmp(grid);
+    LatticeReal ftmp1(grid);
+    LatticeReal ftmp2(grid);
+    LatticeReal one (grid);    one = 1.0;
+    LatticeReal zz  (grid);    zz  = zero;
+    LatticeReal recip(grid);   recip = one/r_l;
+    
+    Real machine_epsilon= 1.0e-14;
+    
+    ftmp = where(r_l>machine_epsilon,recip,one);
+    a[0] = where(r_l>machine_epsilon,   r[0] * ftmp , one);
+    a[1] = where(r_l>machine_epsilon, -(r[1] * ftmp), zz);
+    a[2] = where(r_l>machine_epsilon, -(r[2] * ftmp), zz);
+    a[3] = where(r_l>machine_epsilon, -(r[3] * ftmp), zz);
+
+    r_l *=  beta / ncolour;
+    
+    ftmp1 = where(wheremask,one,zz);
+    Real num_sites = TensorRemove(sum(ftmp1));
+
+    Integer itrials = (Integer)num_sites;
+    ntrials = 0;
+    nfails = 0;
+
+    LatticeInteger lupdate(grid);
+    LatticeInteger lbtmp(grid);
+    LatticeInteger lbtmp2(grid); lbtmp2=zero;
+
+    int n_done = 0;
+    int nhb = 0;
+
+    r[0] = a[0];
+    lupdate = 1;
+
+    LatticeReal ones (grid); ones = 1.0;
+    LatticeReal zeros(grid); zeros=zero;
+
+    const RealD twopi=2.0*M_PI;
+    while ( nhb < nheatbath && n_done < num_sites ) {
+
+       ntrials += itrials;
+
+       random(pRNG,r[1]);
+       std::cout<<"RANDOM SPARSE FLOAT r[1]"<<std::endl;
+       std::cout<<r[1]<<std::endl;
+
+       random(pRNG,r[2]);
+       random(pRNG,ftmp);
+
+       r[1] = log(r[1]);
+       r[2] = log(r[2]);
+
+       ftmp = ftmp*twopi;
+       r[3] = cos(ftmp);
+       r[3] = r[3]*r[3];
+
+       r[1] += r[2] * r[3];
+       r[2]  = r[1] / r_l;
+
+       random(pRNG,ftmp);
+       r[1] = ftmp*ftmp;
+
+       {
+	 LatticeInteger mask_true (grid); mask_true = 1;
+	 LatticeInteger mask_false(grid); mask_false= 0;
+	 LatticeReal    thresh(grid); thresh = (1.0 + 0.5*r[2]);
+	 lbtmp = where(r[1] <= thresh,mask_true,mask_false);
+       }
+       lbtmp2= lbtmp && lupdate;       
+       r[0]  = where(lbtmp2, 1.0+r[2], r[0]);
+
+       ftmp1 = where(lbtmp2,ones,zeros);
+       RealD sitesum = sum(ftmp1);
+       Integer itmp = sitesum;
+
+       n_done += itmp;
+       itrials -= itmp;
+       nfails += itrials;
+       lbtmp   = !lbtmp;
+       lupdate = lupdate & lbtmp;
+       ++nhb;
+    }
+    // Now create r[1], r[2] and r[3] according to the spherical measure 
+    // Take absolute value to guard against round-off 
+    random(pRNG,ftmp1);
+    r[2] = 1.0 - 2.0*ftmp1;
+ 
+    ftmp1 = abs(1.0 - r[0]*r[0]);
+    r[3]   = -(sqrt(ftmp1) * r[2]);
+
+    // Take absolute value to guard against round-off 
+    r_l    = sqrt(abs(ftmp1 - r[3]*r[3]));
+ 
+    random(pRNG,ftmp1);
+    ftmp1 *= twopi;
+    r[1]    = r_l * cos(ftmp1);
+    r[2]    = r_l * sin(ftmp1);
+
+    // Update matrix is B = R * A, with B, R and A given by b_i, r_i and a_i 
+    std::vector<LatticeReal> b(4,grid);
+    b[0] = r[0]*a[0] - r[1]*a[1] - r[2]*a[2] - r[3]*a[3];
+    b[1] = r[0]*a[1] + r[1]*a[0] - r[2]*a[3] + r[3]*a[2];
+    b[2] = r[0]*a[2] + r[2]*a[0] - r[3]*a[1] + r[1]*a[3];
+    b[3] = r[0]*a[3] + r[3]*a[0] - r[1]*a[2] + r[2]*a[1];
+
+     //
+     // Now fill an SU(3) matrix V with the SU(2) submatrix su2_index
+     // parametrized by a_k in the sigma matrix basis.
+     //
+    su2Insert(b,V,su2_subgroup);
+
+    // U = V*U
+    LatticeMatrix tmp(grid);
+    tmp = V * link;
+
+    //mask the assignment back
+    link = where(wheremask,tmp,link);
+
+  }
+
+  static void printGenerators(void)
+  {
+    for(int gen=0;gen<generators();gen++){
+      Matrix ta;
+      generator(gen,ta);
+      std::cout<< "Nc = "<<ncolour<<" t_"<<gen<<std::endl;
+      std::cout<<ta<<std::endl;
+    }
+  }
+
+  static void testGenerators(void){
+    Matrix ta;
+    Matrix tb;
+    std::cout<<"Checking trace ta tb is 0.5 delta_ab"<<std::endl;
+    for(int a=0;a<generators();a++){
+      for(int b=0;b<generators();b++){
+	generator(a,ta);
+	generator(b,tb);
+	Complex tr =TensorRemove(trace(ta*tb)); 
+	std::cout<<tr<<" ";
+	if(a==b) assert(abs(tr-Complex(0.5))<1.0e-6);
+	if(a!=b) assert(abs(tr)<1.0e-6);
+      }
+      std::cout<<std::endl;
+    }
+    std::cout<<"Checking hermitian"<<std::endl;
+    for(int a=0;a<generators();a++){
+      generator(a,ta);
+      std::cout<<a<<" ";
+      assert(norm2(ta-adj(ta))<1.0e-6);
+    }    
+    std::cout<<std::endl;
+
+    std::cout<<"Checking traceless"<<std::endl;
+    for(int a=0;a<generators();a++){
+      generator(a,ta);
+      Complex tr =TensorRemove(trace(ta)); 
+      std::cout<<a<<" ";
+      assert(abs(tr)<1.0e-6);
+    }    
+    std::cout<<std::endl;
+  }
+
+  // reunitarise??
+
+  static void taProj( const LatticeMatrix &in,  LatticeMatrix &out){
+    out = Ta(in);
+  }
+  static void taExp( const LatticeMatrix &x,  LatticeMatrix &ex){ 
+    LatticeMatrix xn   = x;
+
+    RealD nfac = 1.0;
+    ex = 1+x; // 1+x
+
+    // Do a 12th order exponentiation
+    for(int i= 2; i <= 12; ++i)
+    {
+      nfac = nfac/i; 
+      xn   = xn * x; // x2, x3,x4....
+      ex  += xn*nfac;// x2/2!, x3/3!....
+    }
+  }
+
+};
+
+ typedef SU<2> SU2;
+ typedef SU<3> SU3;
+ typedef SU<4> SU4;
+ typedef SU<5> SU5;
+
+  }
+}
+#endif

From 625bb1d7ee99d1e9544e23b1246579a0370f9e81 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Tue, 16 Jun 2015 14:04:33 +0100
Subject: [PATCH 347/429] pointer cast

---
 lib/AlignedAllocator.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/lib/AlignedAllocator.h b/lib/AlignedAllocator.h
index fd7bbb78..3e90371a 100644
--- a/lib/AlignedAllocator.h
+++ b/lib/AlignedAllocator.h
@@ -49,9 +49,9 @@ public:
 
   void deallocate(pointer __p, size_type) { 
 #ifdef HAVE_MM_MALLOC_H
-    _mm_free(__p); 
+    _mm_free((void *)__p); 
 #else
-    free(__p);
+    free((void *)__p);
 #endif
   }
   void construct(pointer __p, const _Tp& __val) { };

From cd7eac062ae52ae0dabce4086aba54fd37ce76e3 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Tue, 16 Jun 2015 14:04:56 +0100
Subject: [PATCH 348/429] Extra check that failed in quenched test

---
 lib/cartesian/Cartesian_red_black.h | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/lib/cartesian/Cartesian_red_black.h b/lib/cartesian/Cartesian_red_black.h
index 3a84ed49..2d885d36 100644
--- a/lib/cartesian/Cartesian_red_black.h
+++ b/lib/cartesian/Cartesian_red_black.h
@@ -142,7 +142,10 @@ public:
 	// Use a reduced simd grid
 	_simd_layout[d] = simd_layout[d];
 	_rdimensions[d]= _ldimensions[d]/_simd_layout[d];
-	
+
+	// all elements of a simd vector must have same checkerboard.
+	if ( simd_layout[d]>1 ) assert((_rdimensions[d]&0x1)==0); 
+
 	_osites *= _rdimensions[d];
 	_isites *= _simd_layout[d];
         

From 633ee06faf90a2ecbdb8deb687f031617c1cd119 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Tue, 16 Jun 2015 14:06:31 +0100
Subject: [PATCH 349/429] Typo fix

---
 lib/lattice/Lattice_base.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/lib/lattice/Lattice_base.h b/lib/lattice/Lattice_base.h
index cd2fa855..0503732c 100644
--- a/lib/lattice/Lattice_base.h
+++ b/lib/lattice/Lattice_base.h
@@ -158,7 +158,7 @@ PARALLEL_FOR_LOOP
       vobj tmp = eval(ss,expr);
       vstream(_odata[ss] ,tmp);
 #else
-      _odata[ss]=_eval(ss,expr);
+      _odata[ss]=eval(ss,expr);
 #endif
     }
   };
@@ -196,7 +196,7 @@ PARALLEL_FOR_LOOP
     _odata.resize(_grid->oSites());
 PARALLEL_FOR_LOOP
     for(int ss=0;ss<_grid->oSites();ss++){
-      _odata[ss]=eval(ss,expr);
+      vstream(_odata[ss] ,eval(ss,expr));
     }
   };
 

From 173b31ce0520c74931295074db387dbaa8c19523 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Tue, 16 Jun 2015 14:07:05 +0100
Subject: [PATCH 350/429] uninitialised bug fix

---
 lib/qcd/utils/WilsonLoops.h | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/lib/qcd/utils/WilsonLoops.h b/lib/qcd/utils/WilsonLoops.h
index 69b86859..e3885432 100644
--- a/lib/qcd/utils/WilsonLoops.h
+++ b/lib/qcd/utils/WilsonLoops.h
@@ -30,6 +30,7 @@ public:
   static void sitePlaquette(LatticeComplex &Plaq,const std::vector<GaugeMat> &U)
   {
     LatticeComplex sitePlaq(U[0]._grid);
+    Plaq=zero;
     for(int mu=1;mu<Nd;mu++){
       for(int nu=0;nu<mu;nu++){
 	traceDirPlaquette(sitePlaq,U,mu,nu);
@@ -42,6 +43,7 @@ public:
   //////////////////////////////////////////////////
   static RealD sumPlaquette(const GaugeLorentz &Umu){
     std::vector<GaugeMat> U(4,Umu._grid);
+
     for(int mu=0;mu<Nd;mu++){
       U[mu] = peekIndex<LorentzIndex>(Umu,mu);
     }
@@ -72,13 +74,15 @@ public:
   //////////////////////////////////////////////////
   static void Staple(GaugeMat &staple,const GaugeLorentz &Umu,int mu){
 
-    std::vector<GaugeMat> U(4,Umu._grid);
+    GridBase *grid = Umu._grid;
+
+    std::vector<GaugeMat> U(4,grid);
     for(int d=0;d<Nd;d++){
       U[d] = peekIndex<LorentzIndex>(Umu,d);
     }
 
     staple = zero;
-    GaugeMat tmp(U[0]._grid);
+    GaugeMat tmp(grid);
 
     for(int nu=0;nu<Nd;nu++){
 

From 79a9f8b9c99d5c12e1b68b3bbde0062d08379e0b Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Tue, 16 Jun 2015 14:17:11 +0100
Subject: [PATCH 351/429] Quenched works for wilson gauge

---
 lib/qcd/utils/SUn.h | 465 ++++++++++++++++++++++++++++----------------
 1 file changed, 302 insertions(+), 163 deletions(-)

diff --git a/lib/qcd/utils/SUn.h b/lib/qcd/utils/SUn.h
index 3a8b5f4d..0ede7965 100644
--- a/lib/qcd/utils/SUn.h
+++ b/lib/qcd/utils/SUn.h
@@ -1,7 +1,6 @@
 #ifndef QCD_UTIL_SUN_H
 #define QCD_UTIL_SUN_H
 
-
 namespace Grid {
   namespace QCD {
 
@@ -13,6 +12,7 @@ public:
   static int su2subgroups(void) { return (ncolour*(ncolour-1))/2; }
 
   template<typename vtype> using iSUnMatrix              = iScalar<iScalar<iMatrix<vtype, ncolour> > > ;
+  template<typename vtype> using iSU2Matrix              = iScalar<iScalar<iMatrix<vtype, 2> > > ;
 
   //////////////////////////////////////////////////////////////////////////////////////////////////
   // Types can be accessed as SU<2>::Matrix , SU<2>::vSUnMatrix, SU<2>::LatticeMatrix etc...
@@ -29,6 +29,19 @@ public:
   typedef Lattice<vMatrixF>    LatticeMatrixF;
   typedef Lattice<vMatrixD>    LatticeMatrixD;
 
+  typedef iSU2Matrix<Complex>         SU2Matrix;
+  typedef iSU2Matrix<ComplexF>        SU2MatrixF;
+  typedef iSU2Matrix<ComplexD>        SU2MatrixD;
+
+  typedef iSU2Matrix<vComplex>       vSU2Matrix;
+  typedef iSU2Matrix<vComplexF>      vSU2MatrixF;
+  typedef iSU2Matrix<vComplexD>      vSU2MatrixD;
+
+  typedef Lattice<vSU2Matrix>     LatticeSU2Matrix;
+  typedef Lattice<vSU2MatrixF>    LatticeSU2MatrixF;
+  typedef Lattice<vSU2MatrixD>    LatticeSU2MatrixD;
+
+
   ////////////////////////////////////////////////////////////////////////
   // There are N^2-1 generators for SU(N).
   //
@@ -122,6 +135,7 @@ public:
     RealD nrm = 1.0/std::sqrt(2.0*trsq);
     ta = ta *nrm;
   }
+
   ////////////////////////////////////////////////////////////////////////
   // Map a su2 subgroup number to the pair of rows that are non zero
   ////////////////////////////////////////////////////////////////////////
@@ -135,208 +149,333 @@ public:
     }
     i2=i1+1+spare;
   }
-  template<class vreal,class vcplx>
-  static void su2Extract(std::vector<Lattice<iSinglet  <vreal> > > &r,
-			       const Lattice<iSUnMatrix<vcplx> > &source, 
-			 int su2_index)
+
+  //////////////////////////////////////////////////////////////////////////////////////////
+  // Pull out a subgroup and project on to real coeffs x pauli basis
+  //////////////////////////////////////////////////////////////////////////////////////////
+  template<class vcplx>
+  static void su2Extract(    Lattice<iSinglet<vcplx> > &Determinant,
+			     Lattice<iSU2Matrix<vcplx> > &subgroup, 
+			     const Lattice<iSUnMatrix<vcplx> > &source, 
+			     int su2_index)
   {
     GridBase *grid(source._grid);
+    conformable(subgroup,source);
+    conformable(subgroup,Determinant);
+    int i0,i1;    
+    su2SubGroupIndex(i0,i1,su2_index);
+
+PARALLEL_FOR_LOOP
+    for(int ss=0;ss!=grid->oSites();ss++){
+      subgroup._odata[ss]()()(0,0) = source._odata[ss]()()(i0,i0);
+      subgroup._odata[ss]()()(0,1) = source._odata[ss]()()(i0,i1);
+      subgroup._odata[ss]()()(1,0) = source._odata[ss]()()(i1,i0);
+      subgroup._odata[ss]()()(1,1) = source._odata[ss]()()(i1,i1);
+
+      iSU2Matrix<vcplx> Sigma = subgroup._odata[ss];
+
+      Sigma = Sigma-adj(Sigma)+trace(adj(Sigma));
+
+      subgroup._odata[ss] = Sigma;
+
+      // this should be purely real
+      Determinant._odata[ss] = Sigma()()(0,0)*Sigma()()(1,1)
+   	                     - Sigma()()(0,1)*Sigma()()(1,0);
 
-    assert(r.size() == 4); // store in 4 real parts
-    
-    for(int i=0;i<4;i++){
-      conformable(r[i],source);
     }
-    
-    int i1,i2;    
-    su2SubGroupIndex(i1,i2,su2_index);
 
-    /* Compute the b(k) of A_SU(2) = b0 + i sum_k bk sigma_k */ 
-    
-    //    r[0] = toReal(real(peekColour(source,i1,i1)) + real(peekColour(source,i2,i2)));
-    //    r[1] = toReal(imag(peekColour(source,i1,i2)) + imag(peekColour(source,i2,i1)));
-    //    r[2] = toReal(real(peekColour(source,i1,i2)) - real(peekColour(source,i2,i1)));
-    //    r[3] = toReal(imag(peekColour(source,i1,i1)) - imag(peekColour(source,i2,i2)));
-    r[0] = toReal(real(peekColour(source,i1,i1)) + real(peekColour(source,i2,i2)));
-    r[1] = toReal(imag(peekColour(source,i1,i2)) + imag(peekColour(source,i2,i1)));
-    r[2] = toReal(real(peekColour(source,i1,i2)) - real(peekColour(source,i2,i1)));
-    r[3] = toReal(imag(peekColour(source,i1,i1)) - imag(peekColour(source,i2,i2)));
-    
   }
 
-  template<class vreal,class vcplx>
-  static void su2Insert(const std::vector<Lattice<iSinglet<vreal> > > &r,
-			Lattice<iSUnMatrix<vcplx> > &dest,
+  //////////////////////////////////////////////////////////////////////////////////////////
+  // Set matrix to one and insert a pauli subgroup
+  //////////////////////////////////////////////////////////////////////////////////////////
+  template<class vcplx>
+  static void su2Insert( const Lattice<iSU2Matrix<vcplx> > &subgroup, 
+			       Lattice<iSUnMatrix<vcplx> > &dest,
 			int su2_index)
   {
-    typedef typename Lattice<iSUnMatrix<vcplx> >::scalar_type cplx;
-    typedef Lattice<iSinglet<vcplx> > Lcomplex;
-    GridBase * grid = dest._grid;
+    GridBase *grid(dest._grid);
+    conformable(subgroup,dest);
+    int i0,i1;    
+    su2SubGroupIndex(i0,i1,su2_index);
 
-    assert(r.size() == 4); // store in 4 real parts
-
-    Lcomplex tmp(grid);
-
-    std::vector<Lcomplex > cr(4,grid);
-    for(int i=0;i<r.size();i++){
-      conformable(r[i],dest);
-      cr[i]=toComplex(r[i]);
+    dest = 1.0; // start out with identity
+PARALLEL_FOR_LOOP
+    for(int ss=0;ss!=grid->oSites();ss++){
+      dest._odata[ss]()()(i0,i0) = subgroup._odata[ss]()()(0,0);
+      dest._odata[ss]()()(i0,i1) = subgroup._odata[ss]()()(0,1);
+      dest._odata[ss]()()(i1,i0) = subgroup._odata[ss]()()(1,0);
+      dest._odata[ss]()()(i1,i1) = subgroup._odata[ss]()()(1,1);
     }
-
-    int i1,i2;
-    su2SubGroupIndex(i1,i2,su2_index);
-
-    cplx one   (1.0,0.0);
-    cplx cplx_i(0.0,1.0);
-
-    tmp =   cr[0]*one+ cr[3]*cplx_i;   pokeColour(dest,tmp,i1,i2); 
-    tmp =   cr[2]*one+ cr[1]*cplx_i;   pokeColour(dest,tmp,i1,i2);
-    tmp =  -cr[2]*one+ cr[1]*cplx_i;   pokeColour(dest,tmp,i2,i1);
-    tmp =   cr[0]*one- cr[3]*cplx_i;   pokeColour(dest,tmp,i2,i2);
   }
 
+
+  ///////////////////////////////////////////////
+  // Generate e^{ Re Tr Staple Link} dlink 
+  //
+  // *** Note Staple should be appropriate linear compbination between all staples.
+  // *** If already by beta pass coefficient 1.0.
+  // *** This routine applies the additional 1/Nc factor that comes after trace in action.
+  //
+  ///////////////////////////////////////////////
   static void SubGroupHeatBath(      GridSerialRNG       &sRNG,
 				     GridParallelRNG     &pRNG,
-			             RealD beta,
+				     RealD beta,//coeff multiplying staple in action (with no 1/Nc)
 				     LatticeMatrix &link, 
-			       const LatticeMatrix &staple, 
+				     const LatticeMatrix &barestaple, // multiplied by action coeffs so th
 				     int su2_subgroup,
 				     int nheatbath,
-				     int& ntrials,
-				     int& nfails,
 				     LatticeInteger &wheremask)
   {
     GridBase *grid = link._grid;
 
-    LatticeMatrix V(grid);
-    V = link*staple;
-
-    std::vector<LatticeReal> r(4,grid);
-    std::vector<LatticeReal> a(4,grid);
-    su2Extract(r,V,su2_subgroup); // HERE
-
-    LatticeReal r_l(grid);
-    r_l  = r[0]*r[0]+r[1]*r[1]+r[2]*r[2]+r[3]*r[3];
-    r_l = sqrt(r_l);
-
-    LatticeReal ftmp(grid);
-    LatticeReal ftmp1(grid);
-    LatticeReal ftmp2(grid);
-    LatticeReal one (grid);    one = 1.0;
-    LatticeReal zz  (grid);    zz  = zero;
-    LatticeReal recip(grid);   recip = one/r_l;
-    
-    Real machine_epsilon= 1.0e-14;
-    
-    ftmp = where(r_l>machine_epsilon,recip,one);
-    a[0] = where(r_l>machine_epsilon,   r[0] * ftmp , one);
-    a[1] = where(r_l>machine_epsilon, -(r[1] * ftmp), zz);
-    a[2] = where(r_l>machine_epsilon, -(r[2] * ftmp), zz);
-    a[3] = where(r_l>machine_epsilon, -(r[3] * ftmp), zz);
-
-    r_l *=  beta / ncolour;
-    
-    ftmp1 = where(wheremask,one,zz);
-    Real num_sites = TensorRemove(sum(ftmp1));
-
-    Integer itrials = (Integer)num_sites;
-    ntrials = 0;
-    nfails = 0;
-
-    LatticeInteger lupdate(grid);
-    LatticeInteger lbtmp(grid);
-    LatticeInteger lbtmp2(grid); lbtmp2=zero;
-
-    int n_done = 0;
-    int nhb = 0;
-
-    r[0] = a[0];
-    lupdate = 1;
-
-    LatticeReal ones (grid); ones = 1.0;
-    LatticeReal zeros(grid); zeros=zero;
-
+    int ntrials=0;
+    int nfails=0;
     const RealD twopi=2.0*M_PI;
-    while ( nhb < nheatbath && n_done < num_sites ) {
 
-       ntrials += itrials;
+    LatticeMatrix staple(grid); 
 
-       random(pRNG,r[1]);
-       std::cout<<"RANDOM SPARSE FLOAT r[1]"<<std::endl;
-       std::cout<<r[1]<<std::endl;
+    staple = barestaple * (beta/ncolour);
 
-       random(pRNG,r[2]);
-       random(pRNG,ftmp);
+    LatticeMatrix    V(grid);    V = link*staple;
 
-       r[1] = log(r[1]);
-       r[2] = log(r[2]);
+    // Subgroup manipulation in the lie algebra space
+    LatticeSU2Matrix    u(grid);   // Kennedy pendleton "u" real projected normalised Sigma
+    LatticeSU2Matrix uinv(grid);
+    LatticeSU2Matrix   ua(grid);   // a in pauli form
+    LatticeSU2Matrix    b(grid);   // rotated matrix after hb
 
-       ftmp = ftmp*twopi;
-       r[3] = cos(ftmp);
-       r[3] = r[3]*r[3];
+    // Some handy constant fields
+    LatticeComplex ones (grid); ones = 1.0;
+    LatticeComplex zeros(grid); zeros=zero;
+    LatticeReal rones (grid); rones = 1.0;
+    LatticeReal rzeros(grid); rzeros=zero;
+    LatticeComplex udet(grid); // determinant of real(staple)
+    LatticeInteger mask_true (grid); mask_true = 1;
+    LatticeInteger mask_false(grid); mask_false= 0;
 
-       r[1] += r[2] * r[3];
-       r[2]  = r[1] / r_l;
+  /*
+PLB 156 P393 (1985) (Kennedy and Pendleton)
 
-       random(pRNG,ftmp);
-       r[1] = ftmp*ftmp;
+Note: absorb "beta" into the def of sigma compared to KP paper; staple
+passed to this routine has "beta" already multiplied in
 
-       {
-	 LatticeInteger mask_true (grid); mask_true = 1;
-	 LatticeInteger mask_false(grid); mask_false= 0;
-	 LatticeReal    thresh(grid); thresh = (1.0 + 0.5*r[2]);
-	 lbtmp = where(r[1] <= thresh,mask_true,mask_false);
-       }
-       lbtmp2= lbtmp && lupdate;       
-       r[0]  = where(lbtmp2, 1.0+r[2], r[0]);
+Action linear in links h and of form:
 
-       ftmp1 = where(lbtmp2,ones,zeros);
-       RealD sitesum = sum(ftmp1);
-       Integer itmp = sitesum;
+    beta S = beta  Sum_p (1 - 1/Nc Re Tr Plaq )
 
-       n_done += itmp;
-       itrials -= itmp;
-       nfails += itrials;
-       lbtmp   = !lbtmp;
-       lupdate = lupdate & lbtmp;
-       ++nhb;
-    }
-    // Now create r[1], r[2] and r[3] according to the spherical measure 
-    // Take absolute value to guard against round-off 
-    random(pRNG,ftmp1);
-    r[2] = 1.0 - 2.0*ftmp1;
- 
-    ftmp1 = abs(1.0 - r[0]*r[0]);
-    r[3]   = -(sqrt(ftmp1) * r[2]);
+Writing Sigma = 1/Nc (beta Sigma') where sum over staples is "Sigma' "
 
-    // Take absolute value to guard against round-off 
-    r_l    = sqrt(abs(ftmp1 - r[3]*r[3]));
- 
-    random(pRNG,ftmp1);
-    ftmp1 *= twopi;
-    r[1]    = r_l * cos(ftmp1);
-    r[2]    = r_l * sin(ftmp1);
+     beta S = const - beta/Nc Re Tr h Sigma'
+            = const - Re Tr h Sigma
+      
+Decompose h and Sigma into (1, sigma_j) ; h_i real, h^2=1, Sigma_i complex arbitrary.
 
-    // Update matrix is B = R * A, with B, R and A given by b_i, r_i and a_i 
-    std::vector<LatticeReal> b(4,grid);
-    b[0] = r[0]*a[0] - r[1]*a[1] - r[2]*a[2] - r[3]*a[3];
-    b[1] = r[0]*a[1] + r[1]*a[0] - r[2]*a[3] + r[3]*a[2];
-    b[2] = r[0]*a[2] + r[2]*a[0] - r[3]*a[1] + r[1]*a[3];
-    b[3] = r[0]*a[3] + r[3]*a[0] - r[1]*a[2] + r[2]*a[1];
+    Tr h Sigma = h_i Sigma_j Tr (sigma_i sigma_j)  = h_i Sigma_j 2 delta_ij
+ Re Tr h Sigma = 2 h_j Re Sigma_j
 
-     //
-     // Now fill an SU(3) matrix V with the SU(2) submatrix su2_index
-     // parametrized by a_k in the sigma matrix basis.
-     //
+Normalised re Sigma_j = xi u_j 
+
+With u_j a unit vector and U can be in SU(2);
+
+Re Tr h Sigma = 2 h_j Re Sigma_j = 2 xi (h.u)
+
+4xi^2 = Det [ Sig - Sig^dag  + 1 Tr Sigdag]
+ u   = 1/2xi [ Sig - Sig^dag  + 1 Tr Sigdag]
+
+ xi = sqrt(Det)/2;
+
+Write a= u h in SU(2); a has pauli decomp a_j;
+
+Note: Product b' xi is unvariant because scaling Sigma leaves
+      normalised vector "u" fixed; Can rescale Sigma so b' = 1.
+  */
+
+    ////////////////////////////////////////////////////////
+    // Real part of Pauli decomposition
+    // Note a subgroup can project to zero in cold start
+    ////////////////////////////////////////////////////////
+    su2Extract(udet,u,V,su2_subgroup); 
+
+    //////////////////////////////////////////////////////
+    // Normalising this vector if possible; else identity
+    //////////////////////////////////////////////////////
+    LatticeComplex xi(grid);  
+
+    LatticeSU2Matrix lident(grid);
+    
+    SU2Matrix ident = Complex(1.0); 
+    SU2Matrix pauli1; SU<2>::generator(0,pauli1);
+    SU2Matrix pauli2; SU<2>::generator(1,pauli2);
+    SU2Matrix pauli3; SU<2>::generator(2,pauli3);
+    pauli1 = timesI(pauli1)*2.0;
+    pauli2 = timesI(pauli2)*2.0;
+    pauli3 = timesI(pauli3)*2.0;
+
+    LatticeComplex cone(grid);
+    LatticeReal adet(grid);
+    adet = abs(toReal(udet));
+    lident=Complex(1.0);
+    cone  =Complex(1.0);
+    Real machine_epsilon=1.0e-7;
+    u   = where(adet>machine_epsilon,u,lident);
+    udet= where(adet>machine_epsilon,udet,cone);
+
+    xi  = 0.5*sqrt(udet);     //4xi^2 = Det [ Sig - Sig^dag  + 1 Tr Sigdag]
+    u   = 0.5*u*pow(xi,-1.0); //  u   = 1/2xi [ Sig - Sig^dag  + 1 Tr Sigdag]
+
+    // Debug test for sanity
+    uinv=adj(u);
+    b=u*uinv-1.0;
+    assert(norm2(b)<1.0e-4);
+
+  /*
+Measure: Haar measure dh has d^4a delta(1-|a^2|)
+In polars:
+  da = da0 r^2 sin theta dr dtheta dphi delta( 1 - r^2 -a0^2) 
+     = da0 r^2 sin theta dr dtheta dphi delta( (sqrt(1-a0^) - r)(sqrt(1-a0^) + r) )
+     = da0 r/2 sin theta dr dtheta dphi delta( (sqrt(1-a0^) - r) )
+
+Action factor Q(h) dh  = e^-S[h]  dh =  e^{  xi Tr uh} dh    // beta enters through xi
+                                     =  e^{2 xi (h.u)} dh
+                                     =  e^{2 xi h0u0}.e^{2 xi h1u1}.e^{2 xi h2u2}.e^{2 xi h3u3} dh
+
+Therefore for each site, take xi for that site
+i) generate  |a0|<1 with dist 
+   (1-a0^2)^0.5 e^{2 xi a0 } da0
+
+Take alpha = 2 xi  = 2 xi [ recall 2 beta/Nc unmod staple norm]; hence 2.0/Nc factor in Chroma ]
+A. Generate two uniformly distributed pseudo-random numbers R and R', R'', R''' in the unit interval; 
+B. Set X = -(ln R)/alpha, X' =-(ln R')/alpha;
+C. Set C = cos^2(2pi R"), with R" another uniform random number in [0,1] ;
+D. Set A = XC;
+E. Let d  = X'+A;
+F. If R'''^2 :> 1 - 0.5 d,  go back to A;
+G. Set a0 = 1 - d;
+
+Note that in step D setting B ~ X - A and using B in place of A in step E will generate a second independent a 0 value.
+  */
+
+    /////////////////////////////////////////////////////////
+    // count the number of sites by picking "1"'s out of hat
+    /////////////////////////////////////////////////////////
+    Integer hit=0;
+    LatticeReal rtmp(grid);
+    rtmp=where(wheremask,rones,rzeros);
+    RealD numSites = sum(rtmp);
+    RealD numAccepted;
+    LatticeInteger      Accepted(grid); Accepted = zero;
+    LatticeInteger newlyAccepted(grid);
+   
+    std::vector<LatticeReal> xr(4,grid);
+    std::vector<LatticeReal>  a(4,grid);
+    LatticeReal d(grid); d=zero;
+    LatticeReal alpha(grid);
+
+    //    std::cout<<"xi "<<xi <<std::endl;
+    alpha = toReal(2.0*xi);
+
+    do { 
+
+      // A. Generate two uniformly distributed pseudo-random numbers R and R', R'', R''' in the unit interval; 
+      random(pRNG,xr[0]); 
+      random(pRNG,xr[1]); 
+      random(pRNG,xr[2]); 
+      random(pRNG,xr[3]); 
+
+      // B. Set X = - ln R/alpha, X' = -ln R'/alpha
+      xr[1] = -log(xr[1])/alpha;  
+      xr[2] = -log(xr[2])/alpha;
+
+      // C. Set C = cos^2(2piR'')
+      xr[3] = cos(xr[3]*twopi);
+      xr[3] = xr[3]*xr[3];
+
+      LatticeReal xrsq(grid);
+
+      //D. Set A = XC;
+      //E. Let d  = X'+A;
+      xrsq = xr[2]+xr[1]*xr[3];
+
+      d = where(Accepted,d,xr[2]+xr[1]*xr[3]);
+	      
+      //F. If R'''^2 :> 1 - 0.5 d,  go back to A;
+      LatticeReal thresh(grid); thresh = 1.0-d*0.5;
+      xrsq = xr[0]*xr[0];
+      LatticeInteger ione(grid); ione = 1;
+      LatticeInteger izero(grid); izero=zero;
+
+      newlyAccepted = where ( xrsq < thresh,ione,izero);
+      Accepted = where ( newlyAccepted, newlyAccepted,Accepted);
+      Accepted = where ( wheremask, Accepted,izero);
+
+      // FIXME need an iSum for integer to avoid overload on return type??
+      rtmp=where(Accepted,rones,rzeros);
+      numAccepted = sum(rtmp);
+
+      hit++;
+
+    } while ( (numAccepted < numSites) && ( hit < nheatbath) );
+
+    // G. Set a0 = 1 - d;
+    a[0]=zero;
+    a[0]=where(wheremask,1.0-d,a[0]);
+
+    //////////////////////////////////////////
+    //    ii) generate a_i uniform on two sphere radius (1-a0^2)^0.5
+    //////////////////////////////////////////
+
+    LatticeReal a123mag(grid);
+    a123mag = sqrt(abs(1.0-a[0]*a[0]));
+
+    LatticeReal cos_theta (grid);
+    LatticeReal sin_theta (grid);
+    LatticeReal       phi (grid);
+
+    random(pRNG,phi);            phi = phi * twopi;       // uniform in [0,2pi]
+    random(pRNG,cos_theta); cos_theta=(cos_theta*2.0)-1.0;  // uniform in [-1,1]
+    sin_theta = sqrt(abs(1.0-cos_theta*cos_theta));
+
+    a[1]    = a123mag * sin_theta * cos(phi);
+    a[2]    = a123mag * sin_theta * sin(phi);
+    a[3]    = a123mag * cos_theta;
+    
+    ua = toComplex(a[0])*ident 
+       + toComplex(a[1])*pauli1 
+       + toComplex(a[2])*pauli2 
+       + toComplex(a[3])*pauli3;
+
+    b = 1.0;
+    b = where(wheremask,uinv*ua,b);
     su2Insert(b,V,su2_subgroup);
 
-    // U = V*U
-    LatticeMatrix tmp(grid);
-    tmp = V * link;
+    //mask the assignment back based on Accptance
+    link = where(Accepted,V * link,link);
 
-    //mask the assignment back
-    link = where(wheremask,tmp,link);
+    //////////////////////////////
+    // Debug Checks
+    // SU2 check
+    LatticeSU2Matrix    check(grid);   // rotated matrix after hb
+    u = zero;
+    check = ua * adj(ua) - 1.0;
+    check = where(Accepted,check,u);
+    assert(norm2(check)<1.0e-4);
 
+    check = b*adj(b)-1.0;
+    check = where(Accepted,check,u);
+    assert(norm2(check)<1.0e-4);
+    
+    LatticeMatrix Vcheck(grid);
+    Vcheck = zero;
+    Vcheck = where(Accepted,V*adj(V) - 1.0,Vcheck);
+    //    std::cout << "SU3 check " <<norm2(Vcheck)<<std::endl;
+    assert(norm2(Vcheck)<1.0e-4);
+    
+    // Verify the link stays in SU(3)
+    //    std::cout <<"Checking the modified link"<<std::endl;
+    Vcheck = link*adj(link) - 1.0;
+    assert(norm2(Vcheck)<1.0e-4);
+    /////////////////////////////////
   }
 
   static void printGenerators(void)

From 9c16bccbf401d4153b08f72ad766a1be6c1bea2c Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Tue, 16 Jun 2015 14:17:45 +0100
Subject: [PATCH 352/429] Critical bug fix of sin/cos typo

---
 lib/simd/Grid_vector_unops.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/simd/Grid_vector_unops.h b/lib/simd/Grid_vector_unops.h
index 3ba77d82..37d1f6de 100644
--- a/lib/simd/Grid_vector_unops.h
+++ b/lib/simd/Grid_vector_unops.h
@@ -100,7 +100,7 @@ namespace Grid {
   }
   template < class S, class V > 
   inline Grid_simd<S,V> sin(const Grid_simd<S,V> &r) {
-    return SimdApply(CosRealFunctor<S>(),r);
+    return SimdApply(SinRealFunctor<S>(),r);
   }
   template < class S, class V > 
   inline Grid_simd<S,V> log(const Grid_simd<S,V> &r) {

From ddbfb026d5a14d11b74776994b80752186e0255b Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Tue, 16 Jun 2015 14:18:48 +0100
Subject: [PATCH 353/429] Heatbath and config related removed

---
 tests/Test_lie_generators.cc | 43 ------------------------------------
 1 file changed, 43 deletions(-)

diff --git a/tests/Test_lie_generators.cc b/tests/Test_lie_generators.cc
index 91f2d33f..d36e463e 100644
--- a/tests/Test_lie_generators.cc
+++ b/tests/Test_lie_generators.cc
@@ -44,49 +44,6 @@ int main (int argc, char ** argv)
   //  SU5::printGenerators();
   //  SU5::testGenerators();
 
-  ///////////////////////////////
-  // Configuration of known size
-  ///////////////////////////////
-  NerscField header;
-  std::string file("./ckpoint_lat.400");
-  LatticeGaugeField Umu(grid);
-  //  readNerscConfiguration(Umu,header,file);
-  Umu=1.0; // Cold start
-
-  // RNG set up for test
-  std::vector<int> pseeds({1,2,3,4,5}); // once I caught a fish alive
-  std::vector<int> sseeds({6,7,8,9,10});// then i let it go again
-  GridParallelRNG  pRNG(grid); pRNG.SeedFixedIntegers(pseeds);
-  GridSerialRNG    sRNG;       sRNG.SeedFixedIntegers(sseeds);
-
-  // SU3 colour operatoions
-  LatticeColourMatrix link(grid);
-  LatticeColourMatrix staple(grid);
-  int mu=0;
-
-  // Get Staple
-  ColourWilsonLoops::Staple(staple,Umu,mu);
-  // Get Link
-  link = peekIndex<LorentzIndex>(Umu,mu);
-
-  // Apply heatbath to the link
-  RealD beta=6.0;
-  int subgroup=0;
-  int nhb=1;
-  int trials=0;
-  int fails=0;
-
-  LatticeInteger one(rbGrid);  one = 1; // fill with ones
-  LatticeInteger mask(grid);   mask= zero;
-  one.checkerboard=Even;
-  setCheckerboard(mask,one);
-
-  // update Even checkerboard
-
-  SU3::SubGroupHeatBath(sRNG,pRNG,beta,link,staple,subgroup,
-			nhb,trials,fails,mask);
-
-
 
   Grid_finalize();
 }

From 318e24474864bf1951845402bd92e0f18b742e80 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Wed, 17 Jun 2015 02:02:06 +0900
Subject: [PATCH 354/429] Check for SUN projection and Exponential

---
 tests/Test_main.cc | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/tests/Test_main.cc b/tests/Test_main.cc
index f0618c04..9db38059 100644
--- a/tests/Test_main.cc
+++ b/tests/Test_main.cc
@@ -1,9 +1,5 @@
 #include "Grid.h"
 
-//DEBUG
-#ifdef SSE4
-#include "simd/Grid_vector_types.h"
-#endif
 
 using namespace std;
 using namespace Grid;
@@ -225,12 +221,17 @@ int main (int argc, char ** argv)
 
     cm = ProjectOnGroup(cm);
     std::cout << cm << "  " << std::endl;
+    cm = ProjectOnGroup(cm);
+    std::cout << cm << "  " << std::endl;
+
 
     TComplex det = Determinant(cm);
     
     std::cout << "determinant: " << det <<  std::endl;
     cm = Exponentiate(cm, 1.0, 10);
     std::cout << cm << "  " << std::endl;
+    det = Determinant(cm);
+    std::cout << "determinant: " << det <<  std::endl;
 
 
 //    Foo = Foo+scalar; // LatticeColourMatrix+Scalar

From 06047b83c75cd64e43061792b29b9872f494997e Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Tue, 16 Jun 2015 20:23:27 +0100
Subject: [PATCH 355/429] add bug-fixed Test_nersc)_io.

---
 lib/Make.inc           | 2 +-
 tests/Test_nersc_io.cc | 5 ++++-
 2 files changed, 5 insertions(+), 2 deletions(-)

diff --git a/lib/Make.inc b/lib/Make.inc
index 11362e27..3020a157 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
 
 CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/tests/Test_nersc_io.cc b/tests/Test_nersc_io.cc
index f473781e..b1b9d7ae 100644
--- a/tests/Test_nersc_io.cc
+++ b/tests/Test_nersc_io.cc
@@ -12,7 +12,7 @@ int main (int argc, char ** argv)
 
   std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
-  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> latt_size  ({16,16,16,32});
   std::vector<int> clatt_size  ({4,4,4,8});
   int orthodir=3;
   int orthosz =latt_size[orthodir];
@@ -21,6 +21,8 @@ int main (int argc, char ** argv)
   GridCartesian     Coarse(clatt_size,simd_layout,mpi_layout);
 
   LatticeGaugeField Umu(&Fine);
+  LatticeGaugeField Umu_diff(&Fine);
+  LatticeGaugeField Umu_saved(&Fine);
 
   std::vector<LatticeColourMatrix> U(4,&Fine);
   
@@ -72,6 +74,7 @@ int main (int argc, char ** argv)
     std::cout << "total " <<Pt*PlaqScale<<std::endl;
   }  
 
+
   TComplex Tp = sum(Plaq);
   Complex p  = TensorRemove(Tp);
   std::cout << "calculated plaquettes " <<p*PlaqScale<<std::endl;

From 09757cbf0cbffe96588fb74045dfd3e60ae0cc4c Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Wed, 17 Jun 2015 20:41:07 +0900
Subject: [PATCH 356/429] Lattice matrix exponential ok

---
 lib/Make.inc                     |  2 +-
 lib/Simd.h                       | 12 +++++
 lib/Tensors.h                    |  2 +
 lib/lattice/Lattice_ET.h         |  1 -
 lib/lattice/Lattice_unary.h      | 13 +++++
 lib/tensors/Tensor_Ta.h          | 88 +++++---------------------------
 lib/tensors/Tensor_determinant.h | 45 ++++++++++++++++
 lib/tensors/Tensor_exp.h         | 37 ++++++++++++++
 tests/Test_main.cc               | 25 +++++----
 9 files changed, 136 insertions(+), 89 deletions(-)
 create mode 100644 lib/tensors/Tensor_determinant.h
 create mode 100644 lib/tensors/Tensor_exp.h

diff --git a/lib/Make.inc b/lib/Make.inc
index a660a775..dab78b37 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./Cshift.h ./simd/Grid_avx.h ./simd/Grid_vector_types.h ./simd/Grid_sse4.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_vector_unops.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./Tensors.h ./Algorithms.h ./communicator/Communicator_base.h ./lattice/Lattice_rng.h ./lattice/Lattice_reduction.h ./lattice/Lattice_transfer.h ./lattice/Lattice_unary.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_comparison.h ./lattice/Lattice_overload.h ./lattice/Lattice_reality.h ./lattice/Lattice_local.h ./lattice/Lattice_conformable.h ./lattice/Lattice_where.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_ET.h ./lattice/Lattice_transpose.h ./lattice/Lattice_trace.h ./Stencil.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_poke.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_class.h ./tensors/Tensor_logical.h ./tensors/Tensor_transpose.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_reality.h ./tensors/Tensor_trace.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_outer.h ./tensors/Tensor_inner.h ./tensors/Tensor_traits.h ./tensors/Tensor_Ta.h ./tensors/Tensor_unary.h ./tensors/Tensor_peek.h ./tensors/Tensor_arith.h ./tensors/Tensor_extract_merge.h ./Communicator.h ./Cartesian.h ./parallelIO/NerscIO.h ./qcd/QCD.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/LinalgUtils.h ./qcd/utils/CovariantCshift.h ./qcd/utils/WilsonLoops.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/spin/TwoSpinor.h ./qcd/spin/Dirac.h ./cshift/Cshift_common.h ./cshift/Cshift_none.h ./cshift/Cshift_mpi.h ./Simd.h ./GridConfig.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_red_black.h ./cartesian/Cartesian_full.h ./AlignedAllocator.h ./Lattice.h ./Threads.h ./Grid.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Zolotarev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Remez.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./algorithms/CoarsenedMatrix.h ./stencil/Lebesgue.h
+HFILES=./Cshift.h ./simd/Grid_avx.h ./simd/Grid_vector_types.h ./simd/Grid_sse4.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_vector_unops.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./Tensors.h ./Algorithms.h ./communicator/Communicator_base.h ./lattice/Lattice_rng.h ./lattice/Lattice_reduction.h ./lattice/Lattice_transfer.h ./lattice/Lattice_unary.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_comparison.h ./lattice/Lattice_overload.h ./lattice/Lattice_reality.h ./lattice/Lattice_local.h ./lattice/Lattice_conformable.h ./lattice/Lattice_where.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_ET.h ./lattice/Lattice_transpose.h ./lattice/Lattice_trace.h ./Stencil.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_poke.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_class.h ./tensors/Tensor_logical.h ./tensors/Tensor_transpose.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_reality.h ./tensors/Tensor_trace.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_outer.h ./tensors/Tensor_inner.h ./tensors/Tensor_traits.h ./tensors/Tensor_Ta.h ./tensors/Tensor_unary.h ./tensors/Tensor_determinant.h ./tensors/Tensor_peek.h ./tensors/Tensor_arith.h ./tensors/Tensor_extract_merge.h ./Communicator.h ./Cartesian.h ./parallelIO/NerscIO.h ./qcd/QCD.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/LinalgUtils.h ./qcd/utils/CovariantCshift.h ./qcd/utils/WilsonLoops.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/spin/TwoSpinor.h ./qcd/spin/Dirac.h ./cshift/Cshift_common.h ./cshift/Cshift_none.h ./cshift/Cshift_mpi.h ./Simd.h ./GridConfig.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_red_black.h ./cartesian/Cartesian_full.h ./AlignedAllocator.h ./Lattice.h ./Threads.h ./Grid.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Zolotarev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Remez.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./algorithms/CoarsenedMatrix.h ./stencil/Lebesgue.h
 
 CCFILES=./qcd/utils/SpaceTimeGrid.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/spin/Dirac.cc ./GridInit.cc ./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/Simd.h b/lib/Simd.h
index f93c3070..7da0d7ad 100644
--- a/lib/Simd.h
+++ b/lib/Simd.h
@@ -46,6 +46,18 @@ namespace Grid {
   inline ComplexF innerProduct(const ComplexF & l, const ComplexF & r) { return conjugate(l)*r; }
   inline RealD innerProduct(const RealD & l, const RealD & r) { return l*r; }
   inline RealF innerProduct(const RealF & l, const RealF & r) { return l*r; }
+
+  inline ComplexD Reduce(const ComplexD& r){ return r; }
+  inline ComplexF Reduce(const ComplexF& r){ return r; }
+  inline RealD Reduce(const RealD& r){ return r; }
+  inline RealF Reduce(const RealF& r){ return r; }
+
+  inline RealD toReal(const ComplexD& r){ return real(r); }
+  inline RealF toReal(const ComplexF& r){ return real(r); }
+  inline RealD toReal(const RealD& r){ return r; }
+  inline RealF toReal(const RealF& r){ return r; }
+
+  
   
   ////////////////////////////////////////////////////////////////////////////////
   //Provide support functions for basic real and complex data types required by Grid
diff --git a/lib/Tensors.h b/lib/Tensors.h
index a2246f4a..90c9d7a8 100644
--- a/lib/Tensors.h
+++ b/lib/Tensors.h
@@ -9,6 +9,8 @@
 #include <tensors/Tensor_transpose.h>
 #include <tensors/Tensor_trace.h>
 #include <tensors/Tensor_Ta.h>
+#include <tensors/Tensor_determinant.h>
+#include <tensors/Tensor_exp.h>
 #include <tensors/Tensor_peek.h>
 #include <tensors/Tensor_poke.h>
 #include <tensors/Tensor_reality.h>
diff --git a/lib/lattice/Lattice_ET.h b/lib/lattice/Lattice_ET.h
index caa45e69..82d11218 100644
--- a/lib/lattice/Lattice_ET.h
+++ b/lib/lattice/Lattice_ET.h
@@ -1,4 +1,3 @@
-
 #ifndef GRID_LATTICE_ET_H
 #define GRID_LATTICE_ET_H
 
diff --git a/lib/lattice/Lattice_unary.h b/lib/lattice/Lattice_unary.h
index 91b39007..6d72ef31 100644
--- a/lib/lattice/Lattice_unary.h
+++ b/lib/lattice/Lattice_unary.h
@@ -24,5 +24,18 @@ PARALLEL_FOR_LOOP
     return ret;
   }
 
+  template<class obj> Lattice<obj> expMat(const Lattice<obj> &rhs, ComplexD alpha, Integer Nexp = DEFAULT_MAT_EXP){
+    Lattice<obj> ret(rhs._grid);
+    ret.checkerboard = rhs.checkerboard;
+    conformable(ret,rhs);
+PARALLEL_FOR_LOOP
+    for(int ss=0;ss<rhs._grid->oSites();ss++){
+      ret._odata[ss]=Exponentiate(rhs._odata[ss],alpha, Nexp);
+    }
+    return ret;
+  }
+
+
+
 }
 #endif
diff --git a/lib/tensors/Tensor_Ta.h b/lib/tensors/Tensor_Ta.h
index 0eea1e6d..e49ede35 100644
--- a/lib/tensors/Tensor_Ta.h
+++ b/lib/tensors/Tensor_Ta.h
@@ -1,5 +1,7 @@
 #ifndef GRID_MATH_TA_H
 #define GRID_MATH_TA_H
+
+
 namespace Grid {
 
   /////////////////////////////////////////////// 
@@ -36,7 +38,8 @@ namespace Grid {
 
 
   /////////////////////////////////////////////// 
-  // ProjectOnGroup function for scalar, vector, matrix
+  // ProjectOnGroup function for scalar, vector, matrix 
+  // Projects on orthogonal, unitary group
   /////////////////////////////////////////////// 
 
 
@@ -59,22 +62,23 @@ namespace Grid {
     {
       // need a check for the group type?
       iMatrix<vtype,N> ret(arg);
-      double nrm;
+      RealD nrm;
+      vtype inner;
       for(int c1=0;c1<N;c1++){
-	nrm = 0.0; 
+	zeroit(inner);	
 	for(int c2=0;c2<N;c2++)
-	  nrm += real(innerProduct(ret._internal[c1][c2],ret._internal[c1][c2]));
-	nrm = 1.0/sqrt(nrm);
-	std::cout << "norm : "<< nrm << "\n";
+	  inner += innerProduct(ret._internal[c1][c2],ret._internal[c1][c2]);
+
+	nrm = 1.0/sqrt(Reduce(toReal(inner)));
 	for(int c2=0;c2<N;c2++)
 	  ret._internal[c1][c2]*= nrm;
       
 	for (int b=c1+1; b<N; ++b){
-	  decltype(ret._internal[b][b]*ret._internal[b][b]) pr = 0.0;
+	  decltype(ret._internal[b][b]*ret._internal[b][b]) pr;
+	  zeroit(pr);
 	  for(int c=0; c<N; ++c)
 	    pr += conjugate(ret._internal[c1][c])*ret._internal[b][c];
 	  
-	  std::cout << "pr : "<< pr << "\n";
 	  for(int c=0; c<N; ++c){
 	    ret._internal[b][c] -= pr * ret._internal[c1][c];
 	  }
@@ -86,74 +90,6 @@ namespace Grid {
     }
 
 
-  /////////////////////////////////////////////// 
-  // Determinant function for scalar, vector, matrix
-  /////////////////////////////////////////////// 
-  inline ComplexF Determinant( const ComplexF &arg){    return arg;}
-  inline ComplexD Determinant( const ComplexD &arg){    return arg;}
-  inline RealF Determinant( const RealF &arg){    return arg;}
-  inline RealD Determinant( const RealD &arg){    return arg;}
-
-  template<class vtype> inline auto Determinant(const iScalar<vtype>&r) -> iScalar<decltype(Determinant(r._internal))>
-    {
-      iScalar<decltype(Determinant(r._internal))> ret;
-      ret._internal = Determinant(r._internal);
-      return ret;
-    }
-
-  template<class vtype,int N, typename std::enable_if< GridTypeMapper<vtype>::TensorLevel == 0 >::type * =nullptr> 
-    inline iScalar<vtype> Determinant(const iMatrix<vtype,N> &arg)
-    {
-      iMatrix<vtype,N> ret(arg);
-      iScalar<vtype> det = vtype(1.0);
-      /* Conversion of matrix to upper triangular */
-      for(int i = 0; i < N; i++){
-        for(int j = 0; j < N; j++){
-	  if(j>i){
-	    vtype ratio = ret._internal[j][i]/ret._internal[i][i];
-	    for(int k = 0; k < N; k++){
-	      ret._internal[j][k] -= ratio * ret._internal[i][k];
-	    }
-	  }
-        }
-      }      
-
-      for(int i = 0; i < N; i++)
-	det *= ret._internal[i][i];   
-
-      return det;
-    }
-
-
-
-
-  /////////////////////////////////////////////// 
-  // Exponentiate function for scalar, vector, matrix
-  /////////////////////////////////////////////// 
-
-
-  template<class vtype> inline iScalar<vtype> Exponentiate(const iScalar<vtype>&r, double alpha,  int Nexp)
-    {
-      iScalar<vtype> ret;
-      ret._internal = Exponentiate(r._internal, alpha, Nexp);
-      return ret;
-    }
-
-
-  template<class vtype,int N, typename std::enable_if< GridTypeMapper<vtype>::TensorLevel == 0 >::type * =nullptr> 
-    inline iMatrix<vtype,N> Exponentiate(const iMatrix<vtype,N> &arg, double alpha, int Nexp)
-    {
-      iMatrix<vtype,N> unit(1.0);
-      iMatrix<vtype,N> temp(unit);
-      
-      for(int i=Nexp; i>=1;--i){
-	temp *= alpha/double(i);
-	temp = unit + temp*arg;
-      }
-      return ProjectOnGroup(temp);
-    }
-
-
 
 }
 #endif
diff --git a/lib/tensors/Tensor_determinant.h b/lib/tensors/Tensor_determinant.h
new file mode 100644
index 00000000..7c8af9ed
--- /dev/null
+++ b/lib/tensors/Tensor_determinant.h
@@ -0,0 +1,45 @@
+#ifndef GRID_MATH_DET_H
+#define GRID_MATH_DET_H
+namespace Grid {
+  /////////////////////////////////////////////// 
+  // Determinant function for scalar, vector, matrix
+  /////////////////////////////////////////////// 
+  inline ComplexF Determinant( const ComplexF &arg){    return arg;}
+  inline ComplexD Determinant( const ComplexD &arg){    return arg;}
+  inline RealF Determinant( const RealF &arg){    return arg;}
+  inline RealD Determinant( const RealD &arg){    return arg;}
+
+  template<class vtype> inline auto Determinant(const iScalar<vtype>&r) -> iScalar<decltype(Determinant(r._internal))>
+    {
+      iScalar<decltype(Determinant(r._internal))> ret;
+      ret._internal = Determinant(r._internal);
+      return ret;
+    }
+
+  template<class vtype,int N, typename std::enable_if< GridTypeMapper<vtype>::TensorLevel == 0 >::type * =nullptr> 
+    inline iScalar<vtype> Determinant(const iMatrix<vtype,N> &arg)
+    {
+      iMatrix<vtype,N> ret(arg);
+      iScalar<vtype> det = vtype(1.0);
+      /* Conversion of matrix to upper triangular */
+      for(int i = 0; i < N; i++){
+        for(int j = 0; j < N; j++){
+	  if(j>i){
+	    vtype ratio = ret._internal[j][i]/ret._internal[i][i];
+	    for(int k = 0; k < N; k++){
+	      ret._internal[j][k] -= ratio * ret._internal[i][k];
+	    }
+	  }
+        }
+      }      
+
+      for(int i = 0; i < N; i++)
+	det *= ret._internal[i][i];   
+
+      return det;
+    }
+
+
+
+}
+#endif
diff --git a/lib/tensors/Tensor_exp.h b/lib/tensors/Tensor_exp.h
new file mode 100644
index 00000000..f8437116
--- /dev/null
+++ b/lib/tensors/Tensor_exp.h
@@ -0,0 +1,37 @@
+#ifndef GRID_MATH_EXP_H
+#define GRID_MATH_EXP_H
+
+#define DEFAULT_MAT_EXP 12
+
+namespace Grid {
+
+  /////////////////////////////////////////////// 
+  // Exponentiate function for scalar, vector, matrix
+  /////////////////////////////////////////////// 
+
+
+  template<class vtype> inline iScalar<vtype> Exponentiate(const iScalar<vtype>&r, ComplexD alpha ,  Integer Nexp = DEFAULT_MAT_EXP)
+    {
+      iScalar<vtype> ret;
+      ret._internal = Exponentiate(r._internal, alpha, Nexp);
+      return ret;
+    }
+
+
+  template<class vtype,int N, typename std::enable_if< GridTypeMapper<vtype>::TensorLevel == 0 >::type * =nullptr> 
+    inline iMatrix<vtype,N> Exponentiate(const iMatrix<vtype,N> &arg, ComplexD alpha  , Integer Nexp = DEFAULT_MAT_EXP )
+    {
+      iMatrix<vtype,N> unit(1.0);
+      iMatrix<vtype,N> temp(unit);
+      
+      for(int i=Nexp; i>=1;--i){
+	temp *= alpha/ComplexD(i);
+	temp = unit + temp*arg;
+      }
+      return ProjectOnGroup(temp);//maybe not strictly necessary
+    }
+
+
+
+}
+#endif
diff --git a/tests/Test_main.cc b/tests/Test_main.cc
index 9db38059..8eecf0c3 100644
--- a/tests/Test_main.cc
+++ b/tests/Test_main.cc
@@ -215,21 +215,21 @@ int main (int argc, char ** argv)
     random(SerialRNG, cm);
     std::cout << cm << std::endl;
 
-    //cm = Ta(cm);
-    //TComplex tracecm= trace(cm);      
-    //std::cout << cm << std::endl;
+    cm = Ta(cm);
+    TComplex tracecm= trace(cm);      
+    std::cout << cm << std::endl;
 
-    cm = ProjectOnGroup(cm);
+
+    cm = Exponentiate(cm, 1.0, 12);
     std::cout << cm << "  " << std::endl;
-    cm = ProjectOnGroup(cm);
-    std::cout << cm << "  " << std::endl;
-
-
-    TComplex det = Determinant(cm);
-    
+    Complex det = Determinant(cm);
     std::cout << "determinant: " << det <<  std::endl;
-    cm = Exponentiate(cm, 1.0, 10);
+
+    cm = ProjectOnGroup(cm);
     std::cout << cm << "  " << std::endl;
+    cm = ProjectOnGroup(cm);
+    std::cout << cm << "  " << std::endl;
+
     det = Determinant(cm);
     std::cout << "determinant: " << det <<  std::endl;
 
@@ -244,6 +244,9 @@ int main (int argc, char ** argv)
     LatticeComplex trscMat(&Fine);
     trscMat = trace(scMat); // Trace
 
+    // Exponentiate test
+    cMat = expMat(cMat, ComplexD(1.0, 0.0));
+
     // LatticeComplex trlcMat(&Fine);
     // trlcMat = trace(lcMat); // Trace involving iVector - now generates error
     

From fd208ca11c23d6281335a43a804b5864c5a1d80a Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sat, 20 Jun 2015 14:09:26 +0100
Subject: [PATCH 357/429] Add the test_quenched files

---
 tests/Test_quenched_update.cc | 89 +++++++++++++++++++++++++++++++++++
 1 file changed, 89 insertions(+)
 create mode 100644 tests/Test_quenched_update.cc

diff --git a/tests/Test_quenched_update.cc b/tests/Test_quenched_update.cc
new file mode 100644
index 00000000..2d5ea2ee
--- /dev/null
+++ b/tests/Test_quenched_update.cc
@@ -0,0 +1,89 @@
+#include <Grid.h>
+
+#include <qcd/utils/CovariantCshift.h>
+#include <qcd/utils/WilsonLoops.h>
+#include <qcd/utils/SUn.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> latt({8,8,8,8});
+  GridCartesian * grid = SpaceTimeGrid::makeFourDimGrid(latt, 
+							GridDefaultSimd(Nd,vComplexF::Nsimd()),
+							GridDefaultMpi());
+  
+  GridRedBlackCartesian * rbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(grid);
+
+  ///////////////////////////////
+  // Configuration of known size
+  ///////////////////////////////
+  LatticeGaugeField Umu(grid);
+  Umu=1.0; // Cold start
+
+  // RNG set up for test
+  std::vector<int> pseeds({1,2,3,4,5}); // once I caught a fish alive
+  std::vector<int> sseeds({6,7,8,9,10});// then i let it go again
+  GridParallelRNG  pRNG(grid); pRNG.SeedFixedIntegers(pseeds);
+  GridSerialRNG    sRNG;       sRNG.SeedFixedIntegers(sseeds);
+
+  // SU3 colour operatoions
+  LatticeColourMatrix link(grid);
+  LatticeColourMatrix staple(grid);
+  int mu=0;
+
+  // Apply heatbath to the link
+  RealD beta=6.0;
+
+  int subsets[2] = { Even, Odd};
+  LatticeInteger one(rbGrid);  one = 1; // fill with ones
+  LatticeInteger mask(grid); 
+
+  for(int sweep=0;sweep<1000;sweep++){
+
+    RealD plaq = ColourWilsonLoops::avgPlaquette(Umu);
+
+    std::cout<<"sweep "<<sweep<<" PLAQUETTE "<<plaq<<std::endl;
+
+    for( int cb=0;cb<2;cb++ ) {
+
+      one.checkerboard=subsets[cb];
+      mask= zero;
+      setCheckerboard(mask,one);
+
+      //      std::cout<<mask<<std::endl;
+      for(int mu=0;mu<Nd;mu++){
+	
+	// Get Link and Staple term in action; must contain Beta and 
+	// any other coeffs
+	ColourWilsonLoops::Staple(staple,Umu,mu);
+	staple = adj(staple);
+
+	link = peekIndex<LorentzIndex>(Umu,mu);
+
+	for( int subgroup=0;subgroup<SU3::su2subgroups();subgroup++ ) {
+
+	  // update Even checkerboard
+	  SU3::SubGroupHeatBath(sRNG,pRNG,beta,link,staple,subgroup,20,mask);
+
+	}
+
+	pokeIndex<LorentzIndex>(Umu,link,mu);
+	//reunitarise link;
+	
+      }
+
+    }
+    
+  }
+
+  Grid_finalize();
+}
+
+
+
+

From 6ad96f7383bd1adc0ae30cd7e6ea0f95b9ebb862 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 20 Jun 2015 22:22:56 +0100
Subject: [PATCH 358/429] Patches for beginnings of an overlap multigrid

---
 lib/Algorithms.h                             |   4 +-
 lib/algorithms/CoarsenedMatrix.h             | 126 ++++++++++++++++---
 lib/algorithms/iterative/ConjugateGradient.h |  25 ++--
 lib/algorithms/iterative/ConjugateResidual.h |  15 +--
 lib/lattice/Lattice_ET.h                     |   2 +-
 lib/lattice/Lattice_transfer.h               |   2 +-
 lib/qcd/utils/SUn.h                          |  54 +++++++-
 tests/Test_cayley_coarsen_support.cc         |  19 ++-
 tests/Test_cayley_ldop_cg.cc                 |  49 +++++---
 tests/Test_cayley_ldop_cr.cc                 |  63 +++++++---
 tests/Test_cayley_ldop_cr_chebyshev.cc       | 124 ------------------
 11 files changed, 285 insertions(+), 198 deletions(-)
 delete mode 100644 tests/Test_cayley_ldop_cr_chebyshev.cc

diff --git a/lib/Algorithms.h b/lib/Algorithms.h
index 2007cf60..751cf6b2 100644
--- a/lib/Algorithms.h
+++ b/lib/Algorithms.h
@@ -3,7 +3,6 @@
 
 #include <algorithms/SparseMatrix.h>
 #include <algorithms/LinearOperator.h>
-#include <algorithms/CoarsenedMatrix.h>
 
 #include <algorithms/approx/Zolotarev.h>
 #include <algorithms/approx/Chebyshev.h>
@@ -17,6 +16,9 @@
 
 #include <algorithms/iterative/ConjugateGradientMultiShift.h>
 
+
+#include <algorithms/CoarsenedMatrix.h>
+
 // Eigen/lanczos
 // EigCg
 // MCR
diff --git a/lib/algorithms/CoarsenedMatrix.h b/lib/algorithms/CoarsenedMatrix.h
index fe801106..2576a696 100644
--- a/lib/algorithms/CoarsenedMatrix.h
+++ b/lib/algorithms/CoarsenedMatrix.h
@@ -12,9 +12,6 @@ namespace Grid {
     std::vector<int> directions   ;
     std::vector<int> displacements;
 
-    // FIXME -- don't like xposing the operator directions
-    // as different to the geometrical dirs
-    // Also don't like special casing five dim.. should pass an object in template
   Geometry(int _d)  {
   
       int base = (_d==5) ? 1:0;
@@ -64,6 +61,99 @@ namespace Grid {
 
   };
   
+  template<class Fobj,class CComplex,int nbasis>
+  class Aggregation   {
+  public:
+    typedef iVector<CComplex,nbasis >             siteVector;
+    typedef Lattice<siteVector>                 CoarseVector;
+    typedef Lattice<iMatrix<CComplex,nbasis > > CoarseMatrix;
+
+    typedef Lattice< CComplex >   CoarseScalar; // used for inner products on fine field
+    typedef Lattice<Fobj >        FineField;
+
+    GridBase *CoarseGrid;
+    GridBase *FineGrid;
+    std::vector<Lattice<Fobj> > subspace;
+
+    Aggregation(GridBase *_CoarseGrid,GridBase *_FineGrid) : 
+      CoarseGrid(_CoarseGrid),
+      FineGrid(_FineGrid),
+      subspace(nbasis,_FineGrid)
+	{
+	};
+  
+    void Orthogonalise(void){
+      CoarseScalar InnerProd(CoarseGrid); 
+      blockOrthogonalise(InnerProd,subspace);
+#if 1
+      //      CheckOrthogonal();
+#endif
+
+    } 
+    void CheckOrthogonal(void){
+      CoarseVector iProj(CoarseGrid); 
+      CoarseVector eProj(CoarseGrid); 
+      Lattice<CComplex> pokey(CoarseGrid);
+
+      
+      for(int i=0;i<nbasis;i++){
+	blockProject(iProj,subspace[i],subspace);
+
+	eProj=zero; 
+	for(int ss=0;ss<CoarseGrid->oSites();ss++){
+	  eProj._odata[ss](i)=CComplex(1.0);
+	}
+	eProj=eProj - iProj;
+	std::cout<<"Orthog check error "<<i<<" " << norm2(eProj)<<std::endl;
+      }
+      std::cout <<"CheckOrthog done"<<std::endl;
+    }
+    void ProjectToSubspace(CoarseVector &CoarseVec,const FineField &FineVec){
+      blockProject(CoarseVec,FineVec,subspace);
+    }
+    void PromoteFromSubspace(const CoarseVector &CoarseVec,FineField &FineVec){
+      blockPromote(CoarseVec,FineVec,subspace);
+    }
+    void CreateSubspaceRandom(GridParallelRNG &RNG){
+      for(int i=0;i<nbasis;i++){
+	random(RNG,subspace[i]);
+	std::cout<<" norm subspace["<<i<<"] "<<norm2(subspace[i])<<std::endl;
+      }
+      Orthogonalise();
+    }
+    virtual void CreateSubspace(GridParallelRNG  &RNG,LinearOperatorBase<FineField> &hermop) {
+
+      RealD scale;
+
+      ConjugateGradient<FineField> CG(1.0e-4,10000);
+      FineField noise(FineGrid);
+      FineField Mn(FineGrid);
+
+      for(int b=0;b<nbasis;b++){
+	
+	gaussian(RNG,noise);
+	scale = std::pow(norm2(noise),-0.5); 
+	noise=noise*scale;
+
+	hermop.Op(noise,Mn); std::cout << "Noise    "<<b<<" nMdagMn "<<norm2(Mn)<< " "<< norm2(noise)<<std::endl;
+	for(int i=0;i<2;i++){
+
+	  CG(hermop,noise,subspace[b]);
+
+	  noise = subspace[b];
+	  scale = std::pow(norm2(noise),-0.5); 
+	  noise=noise*scale;
+
+	}
+
+	hermop.Op(noise,Mn); std::cout << "filtered    "<<b<<" <s|MdagM|s> "<<norm2(Mn)<< " "<< norm2(noise)<<std::endl;
+	subspace[b] = noise;
+      }
+
+      Orthogonalise();
+
+    }
+  };
   // Fine Object == (per site) type of fine field
   // nbasis      == number of deflation vectors
   template<class Fobj,class CComplex,int nbasis>
@@ -145,7 +235,8 @@ namespace Grid {
       comm_buf.resize(Stencil._unified_buffer_size);
     };
 
-    void CoarsenOperator(GridBase *FineGrid,LinearOperatorBase<Lattice<Fobj> > &linop,std::vector<Lattice<Fobj> > & subspace){
+    void CoarsenOperator(GridBase *FineGrid,LinearOperatorBase<Lattice<Fobj> > &linop,
+			 Aggregation<Fobj,CComplex,nbasis> & Subspace){
 
       FineField iblock(FineGrid); // contributions from within this block
       FineField oblock(FineGrid); // contributions from outwith this block
@@ -162,8 +253,11 @@ namespace Grid {
       CoarseScalar InnerProd(Grid()); 
 
       // Orthogonalise the subblocks over the basis
-      blockOrthogonalise(InnerProd,subspace);
-      blockProject(iProj,subspace[0],subspace);
+      blockOrthogonalise(InnerProd,Subspace.subspace);
+      //Subspace.Orthogonalise();
+      //      Subspace.CheckOrthogonal();
+      //Subspace.Orthogonalise();
+      //      Subspace.CheckOrthogonal();
 
       // Compute the matrix elements of linop between this orthonormal
       // set of vectors.
@@ -177,7 +271,7 @@ namespace Grid {
       assert(self_stencil!=-1);
 
       for(int i=0;i<nbasis;i++){
-	phi=subspace[i];
+	phi=Subspace.subspace[i];
 	for(int p=0;p<geom.npoint;p++){ 
 
 	  int dir   = geom.directions[p];
@@ -210,8 +304,10 @@ namespace Grid {
 	    assert(0);
 	  }
 
-	  blockProject(iProj,iblock,subspace);
-	  blockProject(oProj,oblock,subspace);
+	  Subspace.ProjectToSubspace(iProj,iblock);
+	  Subspace.ProjectToSubspace(oProj,oblock);
+	  //	  blockProject(iProj,iblock,Subspace.subspace);
+	  //	  blockProject(oProj,oblock,Subspace.subspace);
 	  for(int ss=0;ss<Grid()->oSites();ss++){
 	    for(int j=0;j<nbasis;j++){
 	      if( disp!= 0 ) {
@@ -234,19 +330,19 @@ namespace Grid {
       }
       std::cout<< " picking by block0 "<< self_stencil <<std::endl;
 
-      phi=subspace[0];
+      phi=Subspace.subspace[0];
       std::vector<int> bc(FineGrid->_ndimension,0);
 
       blockPick(Grid(),phi,tmp,bc);      // Pick out a block
       linop.Op(tmp,Mphi);                // Apply big dop
-      blockProject(iProj,Mphi,subspace); // project it and print it
+      blockProject(iProj,Mphi,Subspace.subspace); // project it and print it
       std::cout<< " Computed matrix elements from block zero only "<<std::endl;
       std::cout<< iProj <<std::endl;
       std::cout<<"Computed Coarse Operator"<<std::endl;
 #endif
-      //      AssertHermitian();
       //      ForceHermitian();
-      //      ForceDiagonal();
+      AssertHermitian();
+      // ForceDiagonal();
     }
     void ForceDiagonal(void) {
 
@@ -263,7 +359,7 @@ namespace Grid {
 
       Complex one(1.0);
 
-      iMatrix<Complex,nbasis> ident;  ident=one;
+      iMatrix<CComplex,nbasis> ident;  ident=one;
 
       val = val*adj(val);
       val = val + 1.0;
@@ -279,7 +375,7 @@ namespace Grid {
 	int dd=d+1;
 	A[2*d] = adj(Cshift(A[2*d+1],dd,1));
       }
-      A[8] = 0.5*(A[8] + adj(A[8]));
+      //      A[8] = 0.5*(A[8] + adj(A[8]));
     }
     void AssertHermitian(void) {
       CoarseMatrix AA    (Grid());
diff --git a/lib/algorithms/iterative/ConjugateGradient.h b/lib/algorithms/iterative/ConjugateGradient.h
index e6915d83..0f366ebf 100644
--- a/lib/algorithms/iterative/ConjugateGradient.h
+++ b/lib/algorithms/iterative/ConjugateGradient.h
@@ -15,7 +15,7 @@ public:
     Integer MaxIterations;
     int verbose;
     ConjugateGradient(RealD tol,Integer maxit) : Tolerance(tol), MaxIterations(maxit) { 
-      verbose=1;
+      verbose=0;
     };
 
 
@@ -58,7 +58,7 @@ public:
 	return;
       }
       
-      std::cout << std::setprecision(4)<< "ConjugateGradient: k=0 residual "<<cp<<" rsq"<<rsq<<std::endl;
+      if(verbose) std::cout << std::setprecision(4)<< "ConjugateGradient: k=0 residual "<<cp<<" rsq"<<rsq<<std::endl;
       
       int k;
       for (k=1;k<=MaxIterations;k++){
@@ -69,23 +69,29 @@ public:
 
 	RealD    qqck = norm2(mmp);
 	ComplexD dck  = innerProduct(p,mmp);
-	//	if (verbose) std::cout <<std::setprecision(4)<< "ConjugateGradient:  d,qq "<<d<< " "<<qq <<" qqcheck "<< qqck<< " dck "<< dck<<std::endl;
+	if (verbose) std::cout <<std::setprecision(4)<< "ConjugateGradient:  d,qq "<<d<< " "<<qq <<" qqcheck "<< qqck<< " dck "<< dck<<std::endl;
       
 	a      = c/d;
 	b_pred = a*(a*qq-d)/c;
 
 
-	//	if (verbose) std::cout <<std::setprecision(4)<< "ConjugateGradient:  a,bp "<<a<< " "<<b_pred <<std::endl;
+	if (verbose) std::cout <<std::setprecision(4)<< "ConjugateGradient:  a,bp "<<a<< " "<<b_pred <<std::endl;
 	cp = axpy_norm(r,-a,mmp,r);
 	b = cp/c;
-	//	std::cout <<std::setprecision(4)<< "ConjugateGradient:  cp,b "<<cp<< " "<<b <<std::endl;
+	if (verbose) std::cout <<std::setprecision(4)<< "ConjugateGradient:  cp,b "<<cp<< " "<<b <<std::endl;
 	
 	// Fuse these loops ; should be really easy
 	psi= a*p+psi;
 	p  = p*b+r;
 	  
 	if (verbose) std::cout<<"ConjugateGradient: Iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
-
+	
+	if (0) { 
+	  Field tmp(src._grid);
+	  Linop.HermOpAndNorm(psi,tmp,qqck,qqck);
+	  tmp=tmp-src;
+	  std::cout<<"ConjugateGradient: true   residual  is "<< norm2(tmp);
+	}
 	// Stopping condition
 	if ( cp <= rsq ) { 
 	  
@@ -98,9 +104,10 @@ public:
 	  RealD resnorm = sqrt(norm2(p));
 	  RealD true_residual = resnorm/srcnorm;
 
-	  std::cout<<"ConjugateGradient: Converged on iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
-	  std::cout<<"ConjugateGradient: true   residual  is "<<true_residual<<" sol "<<psinorm<<" src "<<srcnorm<<std::endl;
-	  std::cout<<"ConjugateGradient: target residual was "<<Tolerance<<std::endl;
+	  std::cout<<"ConjugateGradient: Converged on iteration " <<k
+		   <<" computed residual "<<sqrt(cp/ssq)
+		   <<" true residual     "<<true_residual
+		   <<" target "<<Tolerance<<std::endl;
 	  return;
 	}
       }
diff --git a/lib/algorithms/iterative/ConjugateResidual.h b/lib/algorithms/iterative/ConjugateResidual.h
index ae1f4378..2a42557e 100644
--- a/lib/algorithms/iterative/ConjugateResidual.h
+++ b/lib/algorithms/iterative/ConjugateResidual.h
@@ -16,7 +16,7 @@ namespace Grid {
     int verbose;
 
     ConjugateResidual(RealD tol,Integer maxit) : Tolerance(tol), MaxIterations(maxit) { 
-      verbose=1;
+      verbose=0;
     };
 
     void operator() (LinearOperatorBase<Field> &Linop,const Field &src, Field &psi){
@@ -37,8 +37,8 @@ namespace Grid {
       Linop.HermOpAndNorm(p,Ap,pAp,pAAp);
       Linop.HermOpAndNorm(r,Ar,rAr,rAAr);
 
-      std::cout << "pAp, pAAp"<< pAp<<" "<<pAAp<<std::endl;
-      std::cout << "rAr, rAAr"<< rAr<<" "<<rAAr<<std::endl;
+      if(verbose) std::cout << "pAp, pAAp"<< pAp<<" "<<pAAp<<std::endl;
+      if(verbose) std::cout << "rAr, rAAr"<< rAr<<" "<<rAAr<<std::endl;
 
       cp =norm2(r);
       ssq=norm2(src);
@@ -63,15 +63,16 @@ namespace Grid {
 	axpy(p,b,p,r);
 	pAAp=axpy_norm(Ap,b,Ap,Ar);
 
-	std::cout<<"ConjugateResidual: iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
+	if(verbose) std::cout<<"ConjugateResidual: iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
 
 	if(cp<rsq) {
 	  Linop.HermOp(psi,Ap);
 	  axpy(r,-1.0,src,Ap);
 	  RealD true_resid = norm2(r);
-	  std::cout<<"ConjugateResidual: Converged on iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
-	  std::cout<<"ConjugateResidual: true   residual  is "<<true_resid<<std::endl;
-	  std::cout<<"ConjugateResidual: target residual was "<<Tolerance <<std::endl;
+	  std::cout<<"ConjugateResidual: Converged on iteration " <<k
+		   << " computed residual "<<sqrt(cp/ssq)
+	           << " true residual "<<true_resid
+	           << " target "       <<Tolerance <<std::endl;
 	  return;
 	}
 
diff --git a/lib/lattice/Lattice_ET.h b/lib/lattice/Lattice_ET.h
index 82d11218..cbcca2ce 100644
--- a/lib/lattice/Lattice_ET.h
+++ b/lib/lattice/Lattice_ET.h
@@ -296,7 +296,7 @@ GRID_DEF_UNOP(toReal,UnaryToReal);
 GRID_DEF_UNOP(toComplex,UnaryToComplex);
 GRID_DEF_UNOP(abs  ,UnaryAbs); //abs overloaded in cmath C++98; DON'T do the abs-fabs-dabs-labs thing
 GRID_DEF_UNOP(sqrt ,UnarySqrt);
-GRID_DEF_UNOP(rsqrt,UnarySqrt);
+GRID_DEF_UNOP(rsqrt,UnaryRsqrt);
 GRID_DEF_UNOP(sin  ,UnarySin);
 GRID_DEF_UNOP(cos  ,UnaryCos);
 GRID_DEF_UNOP(log  ,UnaryLog);
diff --git a/lib/lattice/Lattice_transfer.h b/lib/lattice/Lattice_transfer.h
index fb6b39f1..ac628faa 100644
--- a/lib/lattice/Lattice_transfer.h
+++ b/lib/lattice/Lattice_transfer.h
@@ -168,7 +168,7 @@ inline void blockNormalise(Lattice<CComplex> &ip,Lattice<vobj> &fineX)
   GridBase *coarse = ip._grid;
   Lattice<vobj> zz(fineX._grid); zz=zero;
   blockInnerProduct(ip,fineX,fineX);
-  ip = rsqrt(ip);
+  ip = pow(ip,-0.5);
   blockZAXPY(fineX,ip,fineX,zz);
 }
 // useful in multigrid project;
diff --git a/lib/qcd/utils/SUn.h b/lib/qcd/utils/SUn.h
index 0ede7965..647699c8 100644
--- a/lib/qcd/utils/SUn.h
+++ b/lib/qcd/utils/SUn.h
@@ -522,22 +522,66 @@ Note that in step D setting B ~ X - A and using B in place of A in step E will g
   }
 
   // reunitarise??
+  static void LieRandomize(GridParallelRNG     &pRNG,LatticeMatrix &out,double scale=1.0){
+    GridBase *grid = out._grid;
+    LatticeComplex ca (grid);
+    LatticeMatrix  lie(grid);
+    LatticeMatrix  la (grid);
+    Complex ci(0.0,scale);
+    Matrix ta;
+
+    lie=zero;
+    for(int a=0;a<generators();a++){
+
+      random(pRNG,ca); ca=real(ca)-0.5;
+      generator(a,ta);
+
+      la=ci*ca*ta;
+
+      lie = lie+la; // e^{i la ta}
+    }
+    taExp(lie,out);
+  }
+
+  static void HotConfiguration(GridParallelRNG &pRNG,LatticeGaugeField &out){
+    LatticeMatrix Umu(out._grid);
+    for(int mu=0;mu<Nd;mu++){
+      LieRandomize(pRNG,Umu,1.0);
+      pokeLorentz(out,Umu,mu);
+    }
+  }
+  static void TepidConfiguration(GridParallelRNG &pRNG,LatticeGaugeField &out){
+    LatticeMatrix Umu(out._grid);
+    for(int mu=0;mu<Nd;mu++){
+      LieRandomize(pRNG,Umu,0.01);
+      pokeLorentz(out,Umu,mu);
+    }
+  }
+  static void ColdConfiguration(GridParallelRNG &pRNG,LatticeGaugeField &out){
+    LatticeMatrix Umu(out._grid);
+    Umu=1.0;
+    for(int mu=0;mu<Nd;mu++){
+      pokeLorentz(out,Umu,mu);
+    }
+  }
 
   static void taProj( const LatticeMatrix &in,  LatticeMatrix &out){
     out = Ta(in);
   }
   static void taExp( const LatticeMatrix &x,  LatticeMatrix &ex){ 
-    LatticeMatrix xn   = x;
 
+    LatticeMatrix xn(x._grid);
     RealD nfac = 1.0;
-    ex = 1+x; // 1+x
+
+    xn = x;
+    ex =xn+Complex(1.0); // 1+x
 
     // Do a 12th order exponentiation
-    for(int i= 2; i <= 12; ++i)
+    for(int i=2; i <= 12; ++i)
     {
-      nfac = nfac/i; 
+      nfac = nfac/RealD(i); //1/2, 1/2.3 ...
       xn   = xn * x; // x2, x3,x4....
-      ex  += xn*nfac;// x2/2!, x3/3!....
+      ex   = ex+ xn*nfac;// x2/2!, x3/3!....
     }
   }
 
diff --git a/tests/Test_cayley_coarsen_support.cc b/tests/Test_cayley_coarsen_support.cc
index 8cdc9565..72536942 100644
--- a/tests/Test_cayley_coarsen_support.cc
+++ b/tests/Test_cayley_coarsen_support.cc
@@ -83,20 +83,27 @@ int main (int argc, char ** argv)
   std::cout<<"Error "<<norm2(err)<<std::endl;
 
   const int nbasis = 2;
-  std::vector<LatticeFermion> subspace(nbasis,FGrid);
   LatticeFermion prom(FGrid);
 
-  for(int b=0;b<nbasis;b++){
-    random(RNG5,subspace[b]);
-  }
-  std::cout << "Computed randoms"<< std::endl;
+  std::vector<LatticeFermion> subspace(nbasis,FGrid);
+
+  std::cout<<"Calling Aggregation class" <<std::endl;
+
+  MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
+  typedef Aggregation<vSpinColourVector,vTComplex,nbasis> Subspace;
+  Subspace Aggregates(Coarse5d,FGrid);
+  Aggregates.CreateSubspaceRandom(RNG5);
+
+  subspace=Aggregates.subspace;
+
+  std::cout << "Called aggregation class"<< std::endl;
 
   typedef CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis> LittleDiracOperator;
   typedef LittleDiracOperator::CoarseVector CoarseVector;
 
   LittleDiracOperator LittleDiracOp(*Coarse5d);
 
-  LittleDiracOp.CoarsenOperator(FGrid,HermIndefOp,subspace);
+  LittleDiracOp.CoarsenOperator(FGrid,HermIndefOp,Aggregates);
   
   CoarseVector c_src (Coarse5d);
   CoarseVector c_res (Coarse5d);
diff --git a/tests/Test_cayley_ldop_cg.cc b/tests/Test_cayley_ldop_cg.cc
index b146002d..f960ab7b 100644
--- a/tests/Test_cayley_ldop_cg.cc
+++ b/tests/Test_cayley_ldop_cg.cc
@@ -1,4 +1,6 @@
 #include <Grid.h>
+#include <qcd/utils/WilsonLoops.h>
+#include <qcd/utils/SUn.h>
 
 using namespace std;
 using namespace Grid;
@@ -59,67 +61,84 @@ int main (int argc, char ** argv)
   LatticeFermion    ref(FGrid); ref=zero;
   LatticeFermion    tmp(FGrid);
   LatticeFermion    err(FGrid);
-  LatticeGaugeField Umu(UGrid); gaussian(RNG4,Umu);
+  LatticeGaugeField Umu(UGrid); 
 
 
+
+  //gaussian(RNG4,Umu);
   //random(RNG4,Umu);
+
   NerscField header;
   std::string file("./ckpoint_lat.400");
   readNerscConfiguration(Umu,header,file);
+  //  SU3::ColdConfiguration(RNG4,Umu);
+  //  SU3::TepidConfiguration(RNG4,Umu);
+  //  SU3::HotConfiguration(RNG4,Umu);
+  //  Umu=zero;
 
 #if 0
   LatticeColourMatrix U(UGrid);
-  U=zero;
   for(int nn=0;nn<Nd;nn++){
-    if (nn>2) {
-      pokeIndex<LorentzIndex>(Umu,U,nn);
-    }
+    U=peekIndex<LorentzIndex>(Umu,nn);
+    U=U*adj(U)-1.0;
+    std::cout<<"SU3 test "<<norm2(U)<<std::endl;
   }
 #endif  
   RealD mass=0.1;
-  RealD M5=1.0;
+  RealD M5=1.5;
 
   DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
   Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> HermIndefOp(Ddwf);
 
   const int nbasis = 8;
+
+#if 0
   std::vector<LatticeFermion> subspace(nbasis,FGrid);
   LatticeFermion noise(FGrid);
   LatticeFermion ms(FGrid);
   for(int b=0;b<nbasis;b++){
-    Gamma g5(Gamma::Gamma5);
+
     gaussian(RNG5,noise);
     RealD scale = pow(norm2(noise),-0.5); 
     noise=noise*scale;
 
-    HermIndefOp.HermOp(noise,ms); std::cout << "Noise    "<<b<<" Ms "<<norm2(ms)<< " "<< norm2(noise)<<std::endl;
-
+    HermIndefOp.Op(noise,ms); std::cout << "Noise    "<<b<<" Ms "<<norm2(ms)<< " "<< norm2(noise)<<std::endl;
 
     //    filter(HermIndefOp,noise,subspace[b]);
     // inverse iteration
     MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
-    ConjugateGradient<LatticeFermion> CG(1.0e-5,10000);
+    ConjugateGradient<LatticeFermion> CG(1.0e-4,10000);
 
-    for(int i=0;i<4;i++){
+    for(int i=0;i<1;i++){
 
       CG(HermDefOp,noise,subspace[b]);
       noise = subspace[b];
 
       scale = pow(norm2(noise),-0.5); 
       noise=noise*scale;
-      HermDefOp.HermOp(noise,ms); std::cout << "filt    "<<b<<" <u|H|u> "<<norm2(ms)<< " "<< norm2(noise)<<std::endl;
+      HermDefOp.Op(noise,ms); std::cout << "filt    "<<b<<" <u|H|u> "<<norm2(ms)<< " "<< norm2(noise)<<std::endl;
     }
 
     subspace[b] = noise;
-    HermIndefOp.HermOp(subspace[b],ms); std::cout << "Filtered "<<b<<" Ms "<<norm2(ms)<< " "<<norm2(subspace[b]) <<std::endl;
+    HermIndefOp.Op(subspace[b],ms); std::cout << "Filtered "<<b<<" Ms "<<norm2(ms)<< " "<<norm2(subspace[b]) <<std::endl;
+
   }
   std::cout << "Computed randoms"<< std::endl;
+#else
+  std::cout<<"Calling Aggregation class" <<std::endl;
+  MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
+  typedef Aggregation<vSpinColourVector,vTComplex,nbasis> Subspace;
+  Subspace Aggregates(Coarse5d,FGrid);
+  Aggregates.CreateSubspace(RNG5,HermDefOp);
+  std::cout << "Called aggregation class"<< std::endl;
+#endif
+
 
   typedef CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis> LittleDiracOperator;
   typedef LittleDiracOperator::CoarseVector CoarseVector;
 
   LittleDiracOperator LittleDiracOp(*Coarse5d);
-  LittleDiracOp.CoarsenOperator(FGrid,HermIndefOp,subspace);
+  LittleDiracOp.CoarsenOperator(FGrid,HermIndefOp,Aggregates);
   
   CoarseVector c_src (Coarse5d);
   CoarseVector c_res (Coarse5d);
@@ -128,7 +147,7 @@ int main (int argc, char ** argv)
 
   std::cout << "Solving CG on coarse space "<< std::endl;
   MdagMLinearOperator<LittleDiracOperator,CoarseVector> PosdefLdop(LittleDiracOp);
-  ConjugateGradient<CoarseVector> CG(1.0e-8,10000);
+  ConjugateGradient<CoarseVector> CG(1.0e-6,10000);
   CG(PosdefLdop,c_src,c_res);
 
   std::cout << "Done "<< std::endl;
diff --git a/tests/Test_cayley_ldop_cr.cc b/tests/Test_cayley_ldop_cr.cc
index 698bbb28..c4001b55 100644
--- a/tests/Test_cayley_ldop_cr.cc
+++ b/tests/Test_cayley_ldop_cr.cc
@@ -1,4 +1,6 @@
 #include <Grid.h>
+#include <qcd/utils/WilsonLoops.h>
+#include <qcd/utils/SUn.h>
 
 using namespace std;
 using namespace Grid;
@@ -59,67 +61,100 @@ int main (int argc, char ** argv)
   LatticeFermion    ref(FGrid); ref=zero;
   LatticeFermion    tmp(FGrid);
   LatticeFermion    err(FGrid);
-  LatticeGaugeField Umu(UGrid); gaussian(RNG4,Umu);
+  LatticeGaugeField Umu(UGrid); 
 
 
+
+  //gaussian(RNG4,Umu);
   //random(RNG4,Umu);
-  NerscField header;
-  std::string file("./ckpoint_lat.4000");
-  readNerscConfiguration(Umu,header,file);
 
-  RealD mass=0.01;
-  RealD M5=1.8;
+  NerscField header;
+  std::string file("./ckpoint_lat.400");
+  readNerscConfiguration(Umu,header,file);
+  //  SU3::ColdConfiguration(RNG4,Umu);
+  //  SU3::TepidConfiguration(RNG4,Umu);
+  //  SU3::HotConfiguration(RNG4,Umu);
+  //  Umu=zero;
+
+#if 0
+  LatticeColourMatrix U(UGrid);
+  for(int nn=0;nn<Nd;nn++){
+    U=peekIndex<LorentzIndex>(Umu,nn);
+    U=U*adj(U)-1.0;
+    std::cout<<"SU3 test "<<norm2(U)<<std::endl;
+  }
+#endif  
+  RealD mass=0.1;
+  RealD M5=1.5;
 
   DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
   Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> HermIndefOp(Ddwf);
 
   const int nbasis = 8;
+
+#if 0
   std::vector<LatticeFermion> subspace(nbasis,FGrid);
   LatticeFermion noise(FGrid);
   LatticeFermion ms(FGrid);
   for(int b=0;b<nbasis;b++){
-    Gamma g5(Gamma::Gamma5);
+
     gaussian(RNG5,noise);
     RealD scale = pow(norm2(noise),-0.5); 
     noise=noise*scale;
 
-    HermIndefOp.HermOp(noise,ms); std::cout << "Noise    "<<b<<" Ms "<<norm2(ms)<< " "<< norm2(noise)<<std::endl;
-
+    HermIndefOp.Op(noise,ms); std::cout << "Noise    "<<b<<" Ms "<<norm2(ms)<< " "<< norm2(noise)<<std::endl;
 
     //    filter(HermIndefOp,noise,subspace[b]);
     // inverse iteration
     MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
     ConjugateGradient<LatticeFermion> CG(1.0e-4,10000);
 
-    for(int i=0;i<4;i++){
+    for(int i=0;i<1;i++){
 
       CG(HermDefOp,noise,subspace[b]);
       noise = subspace[b];
 
       scale = pow(norm2(noise),-0.5); 
       noise=noise*scale;
-      HermDefOp.HermOp(noise,ms); std::cout << "filt    "<<b<<" <u|H|u> "<<norm2(ms)<< " "<< norm2(noise)<<std::endl;
+      HermDefOp.Op(noise,ms); std::cout << "filt    "<<b<<" <u|H|u> "<<norm2(ms)<< " "<< norm2(noise)<<std::endl;
     }
 
     subspace[b] = noise;
-    HermIndefOp.HermOp(subspace[b],ms); std::cout << "Filtered "<<b<<" Ms "<<norm2(ms)<< " "<<norm2(subspace[b]) <<std::endl;
+    HermIndefOp.Op(subspace[b],ms); std::cout << "Filtered "<<b<<" Ms "<<norm2(ms)<< " "<<norm2(subspace[b]) <<std::endl;
+
   }
   std::cout << "Computed randoms"<< std::endl;
+#else
+  std::cout<<"Calling Aggregation class" <<std::endl;
+  MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
+  typedef Aggregation<vSpinColourVector,vTComplex,nbasis> Subspace;
+  Subspace Aggregates(Coarse5d,FGrid);
+  Aggregates.CreateSubspace(RNG5,HermDefOp);
+  std::cout << "Called aggregation class"<< std::endl;
+#endif
+
 
   typedef CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis> LittleDiracOperator;
   typedef LittleDiracOperator::CoarseVector CoarseVector;
 
   LittleDiracOperator LittleDiracOp(*Coarse5d);
-  LittleDiracOp.CoarsenOperator(FGrid,HermIndefOp,subspace);
+  LittleDiracOp.CoarsenOperator(FGrid,HermIndefOp,Aggregates);
   
   CoarseVector c_src (Coarse5d);
   CoarseVector c_res (Coarse5d);
   gaussian(CRNG,c_src);
   c_res=zero;
 
+
+  std::cout << "Solving CG on coarse space "<< std::endl;
+  MdagMLinearOperator<LittleDiracOperator,CoarseVector> PosdefLdop(LittleDiracOp);
+  ConjugateGradient<CoarseVector> CG(1.0e-6,10000);
+  CG(PosdefLdop,c_src,c_res);
+
+
   std::cout << "Solving MCR on coarse space "<< std::endl;
   HermitianLinearOperator<LittleDiracOperator,CoarseVector> HermIndefLdop(LittleDiracOp);
-  ConjugateResidual<CoarseVector> MCR(1.0e-8,10000);
+  ConjugateResidual<CoarseVector> MCR(1.0e-6,10000);
   MCR(HermIndefLdop,c_src,c_res);
 
   std::cout << "Done "<< std::endl;
diff --git a/tests/Test_cayley_ldop_cr_chebyshev.cc b/tests/Test_cayley_ldop_cr_chebyshev.cc
deleted file mode 100644
index 091b8eff..00000000
--- a/tests/Test_cayley_ldop_cr_chebyshev.cc
+++ /dev/null
@@ -1,124 +0,0 @@
-#include <Grid.h>
-
-using namespace std;
-using namespace Grid;
-using namespace Grid::QCD;
-
-
-template<class d>
-struct scal {
-  d internal;
-};
-
-Gamma::GammaMatrix Gmu [] = {
-    Gamma::GammaX,
-    Gamma::GammaY,
-    Gamma::GammaZ,
-    Gamma::GammaT
-};
-
-double lowpass(double x)
-{
-  return pow(x*x,-2);
-}
-
-int main (int argc, char ** argv)
-{
-  Grid_init(&argc,&argv);
-
-  Chebyshev<LatticeFermion> filter(-120.0, 120.0,256, lowpass);
-  ofstream csv(std::string("filter.dat"),std::ios::out|std::ios::trunc);
-  filter.csv(csv);
-  csv.close();
-
-  const int Ls=8;
-
-  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
-  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
-
-  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
-  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
-
-  // Construct a coarsened grid
-  std::vector<int> clatt = GridDefaultLatt();
-  for(int d=0;d<clatt.size();d++){
-    clatt[d] = clatt[d]/2;
-  }
-  GridCartesian *Coarse4d =  SpaceTimeGrid::makeFourDimGrid(clatt, GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());;
-  GridCartesian *Coarse5d =  SpaceTimeGrid::makeFiveDimGrid(1,Coarse4d);
-
-  std::vector<int> seeds4({1,2,3,4});
-  std::vector<int> seeds5({5,6,7,8});
-  std::vector<int> cseeds({5,6,7,8});
-  GridParallelRNG          RNG5(FGrid);   RNG5.SeedFixedIntegers(seeds5);
-  GridParallelRNG          RNG4(UGrid);   RNG4.SeedFixedIntegers(seeds4);
-  GridParallelRNG          CRNG(Coarse5d);CRNG.SeedFixedIntegers(cseeds);
-
-  LatticeFermion    src(FGrid); gaussian(RNG5,src);
-  LatticeFermion result(FGrid); result=zero;
-  LatticeFermion    ref(FGrid); ref=zero;
-  LatticeFermion    tmp(FGrid);
-  LatticeFermion    err(FGrid);
-  LatticeGaugeField Umu(UGrid); gaussian(RNG4,Umu);
-
-
-  //random(RNG4,Umu);
-  NerscField header;
-  std::string file("./ckpoint_lat.400");
-  readNerscConfiguration(Umu,header,file);
-
-#if 0
-  LatticeColourMatrix U(UGrid);
-  Complex cone(1.0,0.0);
-  for(int nn=0;nn<Nd;nn++){
-    if (nn==3) { 
-      U=zero; std::cout << "zeroing gauge field in dir "<<nn<<std::endl;
-    //    else       { U[nn]= cone;std::cout << "unit gauge field in dir "<<nn<<std::endl; }
-      pokeIndex<LorentzIndex>(Umu,U,nn);
-    }
-  }
-#endif  
-  RealD mass=0.5;
-  RealD M5=1.8;
-
-  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
-  Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> HermIndefOp(Ddwf);
-
-  const int nbasis = 8;
-  std::vector<LatticeFermion> subspace(nbasis,FGrid);
-  LatticeFermion noise(FGrid);
-  LatticeFermion ms(FGrid);
-  for(int b=0;b<nbasis;b++){
-    Gamma g5(Gamma::Gamma5);
-    gaussian(RNG5,noise);
-    RealD scale = pow(norm2(noise),-0.5); 
-    noise=noise*scale;
-
-    HermIndefOp.HermOp(noise,ms); std::cout << "Noise    "<<b<<" Ms "<<norm2(ms)<< " "<< norm2(noise)<<std::endl;
-
-    filter(HermIndefOp,noise,subspace[b]);
-    scale = pow(norm2(subspace[b]),-0.5); 
-    subspace[b]=subspace[b]*scale;
-    HermIndefOp.HermOp(subspace[b],ms); std::cout << "Filtered "<<b<<" Ms "<<norm2(ms)<< " "<<norm2(subspace[b]) <<std::endl;
-  }
-  std::cout << "Computed randoms"<< std::endl;
-
-  typedef CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis> LittleDiracOperator;
-  typedef LittleDiracOperator::CoarseVector CoarseVector;
-
-  LittleDiracOperator LittleDiracOp(*Coarse5d);
-  LittleDiracOp.CoarsenOperator(FGrid,HermIndefOp,subspace);
-  
-  CoarseVector c_src (Coarse5d);
-  CoarseVector c_res (Coarse5d);
-  gaussian(CRNG,c_src);
-  c_res=zero;
-
-  std::cout << "Solving MCR on coarse space "<< std::endl;
-  HermitianLinearOperator<LittleDiracOperator,CoarseVector> HermIndefLdop(LittleDiracOp);
-  ConjugateResidual<CoarseVector> MCR(1.0e-8,10000);
-  MCR(HermIndefLdop,c_src,c_res);
-
-  std::cout << "Done "<< std::endl;
-  Grid_finalize();
-}

From bce59d9911530eb5588498f59a94b6473a53f4f6 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 20 Jun 2015 22:23:57 +0100
Subject: [PATCH 359/429] HDCG but this is not complete and placeholder for
 later completion

---
 lib/algorithms/iterative/AdefGeneric.h | 370 +++++++++++++++++++++++++
 1 file changed, 370 insertions(+)
 create mode 100644 lib/algorithms/iterative/AdefGeneric.h

diff --git a/lib/algorithms/iterative/AdefGeneric.h b/lib/algorithms/iterative/AdefGeneric.h
new file mode 100644
index 00000000..0d1c3d91
--- /dev/null
+++ b/lib/algorithms/iterative/AdefGeneric.h
@@ -0,0 +1,370 @@
+#ifndef GRID_ALGORITHMS_ITERATIVE_GENERIC_PCG
+#define GRID_ALGORITHMS_ITERATIVE_GENERIC_PCG
+
+  /*
+   * Compared to Tang-2009:  P=Pleft. P^T = PRight Q=MssInv. 
+   * Script A = SolverMatrix 
+   * Script P = Preconditioner
+   *
+   * Deflation methods considered
+   *      -- Solve P A x = P b        [ like Luscher ]
+   * DEF-1        M P A x = M P b     [i.e. left precon]
+   * DEF-2        P^T M A x = P^T M b
+   * ADEF-1       Preconditioner = M P + Q      [ Q + M + M A Q]
+   * ADEF-2       Preconditioner = P^T M + Q
+   * BNN          Preconditioner = P^T M P + Q
+   * BNN2         Preconditioner = M P + P^TM +Q - M P A M 
+   * 
+   * Implement ADEF-2
+   *
+   * Vstart = P^Tx + Qb
+   * M1 = P^TM + Q
+   * M2=M3=1
+   * Vout = x
+   */
+
+// abstract base
+template<class Field, class CoarseField>
+class TwoLevelFlexiblePcg : public LinearFunction<Field>
+{
+ public:
+  int verbose;
+  RealD   Tolerance;
+  Integer MaxIterations;
+  const int mmax = 5;
+  GridBase *grid;
+  GridBase *coarsegrid;
+
+  LinearOperatorBase<Field>   *_Linop
+  OperatorFunction<Field>     *_Smoother,
+  LinearFunction<CoarseField> *_CoarseSolver;
+
+  // Need somthing that knows how to get from Coarse to fine and back again
+  
+  // more most opertor functions
+  TwoLevelFlexiblePcg(RealD tol,
+		     Integer maxit,
+		     LinearOperatorBase<Field> *Linop,
+		     LinearOperatorBase<Field> *SmootherLinop,
+		     OperatorFunction<Field>   *Smoother,
+		     OperatorFunction<CoarseField>  CoarseLinop
+		     ) : 
+      Tolerance(tol), 
+      MaxIterations(maxit),
+      _Linop(Linop),
+      _PreconditionerLinop(PrecLinop),
+      _Preconditioner(Preconditioner)
+  { 
+    verbose=0;
+  };
+
+  // The Pcg routine is common to all, but the various matrices differ from derived 
+  // implementation to derived implmentation
+  void operator() (const Field &src, Field &psi){
+  void operator() (const Field &src, Field &psi){
+
+    psi.checkerboard = src.checkerboard;
+    grid             = src._grid;
+
+    RealD f;
+    RealD rtzp,rtz,a,d,b;
+    RealD rptzp;
+    RealD tn;
+    RealD guess = norm2(psi);
+    RealD ssq   = norm2(src);
+    RealD rsq   = ssq*Tolerance*Tolerance;
+    
+    /////////////////////////////
+    // Set up history vectors
+    /////////////////////////////
+    std::vector<Field> p  (mmax,grid);
+    std::vector<Field> mmp(mmax,grid);
+    std::vector<RealD> pAp(mmax);
+
+    Field x  (grid); x = psi;
+    Field z  (grid);
+    Field tmp(grid);
+    Field r  (grid);
+    Field mu (grid);
+  
+    //////////////////////////
+    // x0 = Vstart -- possibly modify guess
+    //////////////////////////
+    x=src;
+    Vstart(x,src);
+
+    // r0 = b -A x0
+    HermOp(x,mmp); // Shouldn't this be something else?
+    axpy (r, -1.0,mmp[0], src);    // Recomputes r=src-Ax0
+
+    //////////////////////////////////
+    // Compute z = M1 x
+    //////////////////////////////////
+    M1(r,z,tmp,mp,SmootherMirs);
+    rtzp =real(innerProduct(r,z));
+
+    ///////////////////////////////////////
+    // Solve for Mss mu = P A z and set p = z-mu
+    // Def2: p = 1 - Q Az = Pright z 
+    // Other algos M2 is trivial
+    ///////////////////////////////////////
+    M2(z,p[0]);
+
+    for (int k=0;k<=MaxIterations;k++){
+    
+      int peri_k  = k % mmax;
+      int peri_kp = (k+1) % mmax;
+
+      rtz=rtzp;
+      d= M3(p[peri_k],mp,mmp[peri_k],tmp);
+      a = rtz/d;
+    
+      // Memorise this
+      pAp[peri_k] = d;
+
+      axpy(x,a,p[peri_k],x);
+      RealD rn = axpy_norm(r,-a,mmp[peri_k],r);
+
+      // Compute z = M x
+      M1(r,z,tmp,mp);
+
+      rtzp =real(innerProduct(r,z));
+
+      M2(z,mu); // ADEF-2 this is identity. Axpy possible to eliminate
+
+      p[peri_kp]=p[peri_k];
+
+      // Standard search direction  p -> z + b p    ; b = 
+      b = (rtzp)/rtz;
+
+      int northog;
+      //    northog     = (peri_kp==0)?1:peri_kp; // This is the fCG(mmax) algorithm
+      northog     = (k>mmax-1)?(mmax-1):k;        // This is the fCG-Tr(mmax-1) algorithm
+    
+      for(int back=0; back < northog; back++){
+	int peri_back = (k-back)%mmax;
+	RealD pbApk= real(innerProduct(mmp[peri_back],p[peri_kp]));
+	RealD beta = -pbApk/pAp[peri_back];
+	axpy(p[peri_kp],beta,p[peri_back],p[peri_kp]);
+      }
+
+      RealD rrn=sqrt(rn/ssq);
+      std::cout<<"TwoLevelfPcg: k= "<<k<<" residual = "<<rrn<<std::endl;
+
+      // Stopping condition
+      if ( rn <= rsq ) { 
+
+	HermOp(x,mmp); // Shouldn't this be something else?
+	axpy(tmp,-1.0,src,mmp[0]);
+	
+	RealD psinorm = sqrt(norm2(x));
+	RealD srcnorm = sqrt(norm2(src));
+	RealD tmpnorm = sqrt(norm2(tmp));
+	RealD true_residual = tmpnorm/srcnorm;
+	std::cout<<"TwoLevelfPcg:   true residual is "<<true_residual<<std::endl;
+	std::cout<<"TwoLevelfPcg: target residual was"<<Tolerance<<std::endl;
+	return k;
+      }
+    }
+    // Non-convergence
+    assert(0);
+  }
+
+ public:
+
+  virtual void M(Field & in,Field & out,Field & tmp) {
+
+  }
+
+  virtual void M1(Field & in, Field & out) {// the smoother
+
+    // [PTM+Q] in = [1 - Q A] M in + Q in = Min + Q [ in -A Min]
+    Field tmp(grid);
+    Field Min(grid);
+
+    PcgM(in,Min); // Smoother call
+
+    HermOp(Min,out);
+    axpy(tmp,-1.0,out,in);          // tmp  = in - A Min
+
+    ProjectToSubspace(tmp,PleftProj);     
+    ApplyInverse(PleftProj,PleftMss_proj); // Ass^{-1} [in - A Min]_s
+    PromoteFromSubspace(PleftMss_proj,tmp);// tmp = Q[in - A Min]  
+    axpy(out,1.0,Min,tmp); // Min+tmp
+  }
+
+  virtual void M2(const Field & in, Field & out) {
+    out=in;
+    // Must override for Def2 only
+    //  case PcgDef2:
+    //    Pright(in,out);
+    //    break;
+  }
+
+  virtual RealD M3(const Field & p, Field & mmp){
+    double d,dd;
+    HermOpAndNorm(p,mmp,d,dd);
+    return dd;
+    // Must override for Def1 only
+    //  case PcgDef1:
+    //    d=linop_d->Mprec(p,mmp,tmp,0,1);// Dag no
+    //      linop_d->Mprec(mmp,mp,tmp,1);// Dag yes
+    //    Pleft(mp,mmp);
+    //    d=real(linop_d->inner(p,mmp));
+  }
+
+  virtual void VstartDef2(Field & xconst Field & src){
+    //case PcgDef2:
+    //case PcgAdef2: 
+    //case PcgAdef2f:
+    //case PcgV11f:
+    ///////////////////////////////////
+    // Choose x_0 such that 
+    // x_0 = guess +  (A_ss^inv) r_s = guess + Ass_inv [src -Aguess]
+    //                               = [1 - Ass_inv A] Guess + Assinv src
+    //                               = P^T guess + Assinv src 
+    //                               = Vstart  [Tang notation]
+    // This gives:
+    // W^T (src - A x_0) = src_s - A guess_s - r_s
+    //                   = src_s - (A guess)_s - src_s  + (A guess)_s 
+    //                   = 0 
+    ///////////////////////////////////
+    Field r(grid);
+    Field mmp(grid);
+    
+    HermOp(x,mmp);
+    axpy (r, -1.0, mmp, src);        // r_{-1} = src - A x
+    ProjectToSubspace(r,PleftProj);     
+    ApplyInverseCG(PleftProj,PleftMss_proj); // Ass^{-1} r_s
+    PromoteFromSubspace(PleftMss_proj,mmp);  
+    x=x+mmp;
+
+  }
+
+  virtual void Vstart(Field & x,const Field & src){
+    return;
+  }
+
+  /////////////////////////////////////////////////////////////////////
+  // Only Def1 has non-trivial Vout. Override in Def1
+  /////////////////////////////////////////////////////////////////////
+  virtual void   Vout  (Field & in, Field & out,Field & src){
+    out = in;
+    //case PcgDef1:
+    //    //Qb + PT x
+    //    ProjectToSubspace(src,PleftProj);     
+    //    ApplyInverse(PleftProj,PleftMss_proj); // Ass^{-1} r_s
+    //    PromoteFromSubspace(PleftMss_proj,tmp);  
+    //    
+    //    Pright(in,out);
+    //    
+    //    linop_d->axpy(out,tmp,out,1.0);
+    //    break;
+  }
+
+  ////////////////////////////////////////////////////////////////////////////////////////////////
+  // Pright and Pleft are common to all implementations
+  ////////////////////////////////////////////////////////////////////////////////////////////////
+  virtual void Pright(Field & in,Field & out){
+    // P_R  = [ 1              0 ] 
+    //        [ -Mss^-1 Msb    0 ] 
+    Field in_sbar(grid);
+
+    ProjectToSubspace(in,PleftProj);     
+    PromoteFromSubspace(PleftProj,out);  
+    axpy(in_sbar,-1.0,out,in);       // in_sbar = in - in_s 
+
+    HermOp(in_sbar,out);
+    ProjectToSubspace(out,PleftProj);           // Mssbar in_sbar  (project)
+
+    ApplyInverse     (PleftProj,PleftMss_proj); // Mss^{-1} Mssbar 
+    PromoteFromSubspace(PleftMss_proj,out);     // 
+
+    axpy(out,-1.0,out,in_sbar);     // in_sbar - Mss^{-1} Mssbar in_sbar
+  }
+  virtual void Pleft (Field & in,Field & out){
+    // P_L  = [ 1  -Mbs Mss^-1] 
+    //        [ 0   0         ] 
+    Field in_sbar(grid);
+    Field    tmp2(grid);
+    Field    Mtmp(grid);
+
+    ProjectToSubspace(in,PleftProj);     
+    PromoteFromSubspace(PleftProj,out);  
+    axpy(in_sbar,-1.0,out,in);      // in_sbar = in - in_s
+
+    ApplyInverse(PleftProj,PleftMss_proj); // Mss^{-1} in_s
+    PromoteFromSubspace(PleftMss_proj,out);
+
+    HermOp(out,Mtmp);
+
+    ProjectToSubspace(Mtmp,PleftProj);      // Msbar s Mss^{-1}
+    PromoteFromSubspace(PleftProj,tmp2);
+
+    axpy(out,-1.0,tmp2,Mtmp);
+    axpy(out,-1.0,out,in_sbar);     // in_sbar - Msbars Mss^{-1} in_s
+  }
+}
+
+template<class Field>
+class TwoLevelFlexiblePcgADef2 : public TwoLevelFlexiblePcg<Field> {
+ public:
+  virtual void M(Field & in,Field & out,Field & tmp){
+
+  } 
+  virtual void M1(Field & in, Field & out,Field & tmp,Field & mp){
+
+  }
+  virtual void M2(Field & in, Field & out){
+
+  }
+  virtual RealD M3(Field & p, Field & mp,Field & mmp, Field & tmp){
+
+  }
+  virtual void Vstart(Field & in, Field & src, Field & r, Field & mp, Field & mmp, Field & tmp){
+
+  }
+}
+/*
+template<class Field>
+class TwoLevelFlexiblePcgAD : public TwoLevelFlexiblePcg<Field> {
+ public:
+  virtual void M(Field & in,Field & out,Field & tmp); 
+  virtual void M1(Field & in, Field & out,Field & tmp,Field & mp);
+  virtual void M2(Field & in, Field & out);
+  virtual RealD M3(Field & p, Field & mp,Field & mmp, Field & tmp);
+  virtual void Vstart(Field & in, Field & src, Field & r, Field & mp, Field & mmp, Field & tmp);
+}
+
+template<class Field>
+class TwoLevelFlexiblePcgDef1 : public TwoLevelFlexiblePcg<Field> {
+ public:
+  virtual void M(Field & in,Field & out,Field & tmp); 
+  virtual void M1(Field & in, Field & out,Field & tmp,Field & mp);
+  virtual void M2(Field & in, Field & out);
+  virtual RealD M3(Field & p, Field & mp,Field & mmp, Field & tmp);
+  virtual void Vstart(Field & in, Field & src, Field & r, Field & mp, Field & mmp, Field & tmp);
+  virtual void   Vout  (Field & in, Field & out,Field & src,Field & tmp);
+}
+
+template<class Field>
+class TwoLevelFlexiblePcgDef2 : public TwoLevelFlexiblePcg<Field> {
+ public:
+  virtual void M(Field & in,Field & out,Field & tmp); 
+  virtual void M1(Field & in, Field & out,Field & tmp,Field & mp);
+  virtual void M2(Field & in, Field & out);
+  virtual RealD M3(Field & p, Field & mp,Field & mmp, Field & tmp);
+  virtual void Vstart(Field & in, Field & src, Field & r, Field & mp, Field & mmp, Field & tmp);
+}
+
+template<class Field>
+class TwoLevelFlexiblePcgV11: public TwoLevelFlexiblePcg<Field> {
+ public:
+  virtual void M(Field & in,Field & out,Field & tmp); 
+  virtual void M1(Field & in, Field & out,Field & tmp,Field & mp);
+  virtual void M2(Field & in, Field & out);
+  virtual RealD M3(Field & p, Field & mp,Field & mmp, Field & tmp);
+  virtual void Vstart(Field & in, Field & src, Field & r, Field & mp, Field & mmp, Field & tmp);
+}
+*/
+#endif

From eb5ad2884c2bff4d77ae3f16862e2fc762033587 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 20 Jun 2015 22:24:21 +0100
Subject: [PATCH 360/429] Will start this as a two level algorithm

---
 .../PrecGeneralisedConjugateResidual.h        | 86 +++++++++++++++++++
 1 file changed, 86 insertions(+)
 create mode 100644 lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h

diff --git a/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h b/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h
new file mode 100644
index 00000000..a269ec8a
--- /dev/null
+++ b/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h
@@ -0,0 +1,86 @@
+#ifndef GRID_PGCR_H
+#define GRID_PGCR_H
+
+namespace Grid {
+
+    /////////////////////////////////////////////////////////////
+    // Base classes for iterative processes based on operators
+    // single input vec, single output vec.
+    /////////////////////////////////////////////////////////////
+
+  template<class Field> 
+    class ConjugateResidual : public OperatorFunction<Field> {
+  public:                                                
+    RealD   Tolerance;
+    Integer MaxIterations;
+    int verbose;
+
+    ConjugateResidual(RealD tol,Integer maxit) : Tolerance(tol), MaxIterations(maxit) { 
+      verbose=0;
+    };
+
+    void operator() (LinearOperatorBase<Field> &Linop,const Field &src, Field &psi){
+
+      RealD a, b, c, d;
+      RealD cp, ssq,rsq;
+      
+      RealD rAr, rAAr, rArp;
+      RealD pAp, pAAp;
+
+      GridBase *grid = src._grid;
+      psi=zero;
+      Field r(grid),  p(grid), Ap(grid), Ar(grid);
+      
+      r=src;
+      p=src;
+
+      Linop.HermOpAndNorm(p,Ap,pAp,pAAp);
+      Linop.HermOpAndNorm(r,Ar,rAr,rAAr);
+
+      if(verbose) std::cout << "pAp, pAAp"<< pAp<<" "<<pAAp<<std::endl;
+      if(verbose) std::cout << "rAr, rAAr"<< rAr<<" "<<rAAr<<std::endl;
+
+      cp =norm2(r);
+      ssq=norm2(src);
+      rsq=Tolerance*Tolerance*ssq;
+
+      std::cout<<"ConjugateResidual: iteration " <<0<<" residual "<<cp<< " target"<< rsq<<std::endl;
+
+      for(int k=1;k<MaxIterations;k++){
+
+	a = rAr/pAAp;
+
+	axpy(psi,a,p,psi);
+
+	cp = axpy_norm(r,-a,Ap,r);
+
+	rArp=rAr;
+
+	Linop.HermOpAndNorm(r,Ar,rAr,rAAr);
+
+	b   =rAr/rArp;
+ 
+	axpy(p,b,p,r);
+	pAAp=axpy_norm(Ap,b,Ap,Ar);
+
+	if(verbose) std::cout<<"ConjugateResidual: iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
+
+	if(cp<rsq) {
+	  Linop.HermOp(psi,Ap);
+	  axpy(r,-1.0,src,Ap);
+	  RealD true_resid = norm2(r);
+	  std::cout<<"ConjugateResidual: Converged on iteration " <<k
+		   << " computed residual "<<sqrt(cp/ssq)
+	           << " true residual "<<true_resid
+	           << " target "       <<Tolerance <<std::endl;
+	  return;
+	}
+
+      }
+
+      std::cout<<"ConjugateResidual did NOT converge"<<std::endl;
+      assert(0);
+    }
+  };
+}
+#endif

From 945bb93e488114bd2885c2b965e6ce0945c99098 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 21 Jun 2015 10:58:46 +0100
Subject: [PATCH 361/429] Variable preconditioned GCR with restarting.
 Orthogonalisation depth and restart frequency is controllable via constructor

---
 lib/Algorithms.h                              |   1 +
 lib/Make.inc                                  |   2 +-
 lib/algorithms/LinearOperator.h               |   5 +
 lib/algorithms/iterative/ConjugateGradient.h  |  10 --
 lib/algorithms/iterative/ConjugateResidual.h  |  11 +-
 .../PrecGeneralisedConjugateResidual.h        | 168 ++++++++++++------
 lib/qcd/QCD.h                                 |   1 +
 tests/Make.inc                                |   6 +-
 tests/Test_dwf_fpgcr.cc                       |  90 ++++++++++
 9 files changed, 222 insertions(+), 72 deletions(-)
 create mode 100644 tests/Test_dwf_fpgcr.cc

diff --git a/lib/Algorithms.h b/lib/Algorithms.h
index 751cf6b2..311d12db 100644
--- a/lib/Algorithms.h
+++ b/lib/Algorithms.h
@@ -3,6 +3,7 @@
 
 #include <algorithms/SparseMatrix.h>
 #include <algorithms/LinearOperator.h>
+#include <algorithms/Preconditioner.h>
 
 #include <algorithms/approx/Zolotarev.h>
 #include <algorithms/approx/Chebyshev.h>
diff --git a/lib/Make.inc b/lib/Make.inc
index e1167bfa..6b5ae64f 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
 
 CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/algorithms/LinearOperator.h b/lib/algorithms/LinearOperator.h
index 1cf78501..a804bceb 100644
--- a/lib/algorithms/LinearOperator.h
+++ b/lib/algorithms/LinearOperator.h
@@ -207,6 +207,11 @@ namespace Grid {
       virtual void operator() (LinearOperatorBase<Field> &Linop, const Field &in, Field &out) = 0;
     };
 
+    template<class Field> class LinearFunction {
+    public:
+      virtual void operator() (const Field &in, Field &out) = 0;
+    };
+
     /////////////////////////////////////////////////////////////
     // Base classes for Multishift solvers for operators
     /////////////////////////////////////////////////////////////
diff --git a/lib/algorithms/iterative/ConjugateGradient.h b/lib/algorithms/iterative/ConjugateGradient.h
index 0f366ebf..e281d765 100644
--- a/lib/algorithms/iterative/ConjugateGradient.h
+++ b/lib/algorithms/iterative/ConjugateGradient.h
@@ -69,16 +69,12 @@ public:
 
 	RealD    qqck = norm2(mmp);
 	ComplexD dck  = innerProduct(p,mmp);
-	if (verbose) std::cout <<std::setprecision(4)<< "ConjugateGradient:  d,qq "<<d<< " "<<qq <<" qqcheck "<< qqck<< " dck "<< dck<<std::endl;
       
 	a      = c/d;
 	b_pred = a*(a*qq-d)/c;
 
-
-	if (verbose) std::cout <<std::setprecision(4)<< "ConjugateGradient:  a,bp "<<a<< " "<<b_pred <<std::endl;
 	cp = axpy_norm(r,-a,mmp,r);
 	b = cp/c;
-	if (verbose) std::cout <<std::setprecision(4)<< "ConjugateGradient:  cp,b "<<cp<< " "<<b <<std::endl;
 	
 	// Fuse these loops ; should be really easy
 	psi= a*p+psi;
@@ -86,12 +82,6 @@ public:
 	  
 	if (verbose) std::cout<<"ConjugateGradient: Iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
 	
-	if (0) { 
-	  Field tmp(src._grid);
-	  Linop.HermOpAndNorm(psi,tmp,qqck,qqck);
-	  tmp=tmp-src;
-	  std::cout<<"ConjugateGradient: true   residual  is "<< norm2(tmp);
-	}
 	// Stopping condition
 	if ( cp <= rsq ) { 
 	  
diff --git a/lib/algorithms/iterative/ConjugateResidual.h b/lib/algorithms/iterative/ConjugateResidual.h
index 2a42557e..197a0b15 100644
--- a/lib/algorithms/iterative/ConjugateResidual.h
+++ b/lib/algorithms/iterative/ConjugateResidual.h
@@ -37,14 +37,11 @@ namespace Grid {
       Linop.HermOpAndNorm(p,Ap,pAp,pAAp);
       Linop.HermOpAndNorm(r,Ar,rAr,rAAr);
 
-      if(verbose) std::cout << "pAp, pAAp"<< pAp<<" "<<pAAp<<std::endl;
-      if(verbose) std::cout << "rAr, rAAr"<< rAr<<" "<<rAAr<<std::endl;
-
       cp =norm2(r);
       ssq=norm2(src);
       rsq=Tolerance*Tolerance*ssq;
 
-      std::cout<<"ConjugateResidual: iteration " <<0<<" residual "<<cp<< " target"<< rsq<<std::endl;
+      if (verbose) std::cout<<"ConjugateResidual: iteration " <<0<<" residual "<<cp<< " target"<< rsq<<std::endl;
 
       for(int k=1;k<MaxIterations;k++){
 
@@ -62,16 +59,16 @@ namespace Grid {
  
 	axpy(p,b,p,r);
 	pAAp=axpy_norm(Ap,b,Ap,Ar);
-
+	
 	if(verbose) std::cout<<"ConjugateResidual: iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
 
 	if(cp<rsq) {
 	  Linop.HermOp(psi,Ap);
 	  axpy(r,-1.0,src,Ap);
-	  RealD true_resid = norm2(r);
+	  RealD true_resid = norm2(r)/ssq;
 	  std::cout<<"ConjugateResidual: Converged on iteration " <<k
 		   << " computed residual "<<sqrt(cp/ssq)
-	           << " true residual "<<true_resid
+	           << " true residual "<<sqrt(true_resid)
 	           << " target "       <<Tolerance <<std::endl;
 	  return;
 	}
diff --git a/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h b/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h
index a269ec8a..bd5bfbfe 100644
--- a/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h
+++ b/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h
@@ -1,75 +1,55 @@
-#ifndef GRID_PGCR_H
-#define GRID_PGCR_H
+#ifndef GRID_PREC_GCR_H
+#define GRID_PREC_GCR_H
 
+///////////////////////////////////////////////////////////////////////////////////////////////////////
+//VPGCR Abe and Zhang, 2005.
+//INTERNATIONAL JOURNAL OF NUMERICAL ANALYSIS AND MODELING
+//Computing and Information Volume 2, Number 2, Pages 147-161
+///////////////////////////////////////////////////////////////////////////////////////////////////////
 namespace Grid {
 
-    /////////////////////////////////////////////////////////////
-    // Base classes for iterative processes based on operators
-    // single input vec, single output vec.
-    /////////////////////////////////////////////////////////////
-
-  template<class Field> 
-    class ConjugateResidual : public OperatorFunction<Field> {
+  template<class Field>
+    class PrecGeneralisedConjugateResidual : public OperatorFunction<Field> {
   public:                                                
     RealD   Tolerance;
     Integer MaxIterations;
     int verbose;
+    int mmax;
+    int nstep;
+    int steps;
+    LinearFunction<Field> &Preconditioner;
 
-    ConjugateResidual(RealD tol,Integer maxit) : Tolerance(tol), MaxIterations(maxit) { 
-      verbose=0;
+  PrecGeneralisedConjugateResidual(RealD tol,Integer maxit,LinearFunction<Field> &Prec,int _mmax,int _nstep) : 
+      Tolerance(tol), 
+      MaxIterations(maxit),
+      Preconditioner(Prec),
+      mmax(_mmax),
+      nstep(_nstep)
+    { 
+      verbose=1;
     };
 
     void operator() (LinearOperatorBase<Field> &Linop,const Field &src, Field &psi){
 
-      RealD a, b, c, d;
-      RealD cp, ssq,rsq;
-      
-      RealD rAr, rAAr, rArp;
-      RealD pAp, pAAp;
-
-      GridBase *grid = src._grid;
       psi=zero;
-      Field r(grid),  p(grid), Ap(grid), Ar(grid);
-      
-      r=src;
-      p=src;
-
-      Linop.HermOpAndNorm(p,Ap,pAp,pAAp);
-      Linop.HermOpAndNorm(r,Ar,rAr,rAAr);
-
-      if(verbose) std::cout << "pAp, pAAp"<< pAp<<" "<<pAAp<<std::endl;
-      if(verbose) std::cout << "rAr, rAAr"<< rAr<<" "<<rAAr<<std::endl;
-
-      cp =norm2(r);
+      RealD cp, ssq,rsq;
       ssq=norm2(src);
       rsq=Tolerance*Tolerance*ssq;
+      
+      Field r(src._grid);
 
-      std::cout<<"ConjugateResidual: iteration " <<0<<" residual "<<cp<< " target"<< rsq<<std::endl;
+      steps=0;
+      for(int k=0;k<MaxIterations;k++){
 
-      for(int k=1;k<MaxIterations;k++){
+	cp=GCRnStep(Linop,src,psi,rsq);
 
-	a = rAr/pAAp;
-
-	axpy(psi,a,p,psi);
-
-	cp = axpy_norm(r,-a,Ap,r);
-
-	rArp=rAr;
-
-	Linop.HermOpAndNorm(r,Ar,rAr,rAAr);
-
-	b   =rAr/rArp;
- 
-	axpy(p,b,p,r);
-	pAAp=axpy_norm(Ap,b,Ap,Ar);
-
-	if(verbose) std::cout<<"ConjugateResidual: iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
+	if ( verbose ) std::cout<<"VPGCR("<<mmax<<","<<nstep<<") "<< steps <<" steps cp = "<<cp<<std::endl;
 
 	if(cp<rsq) {
-	  Linop.HermOp(psi,Ap);
-	  axpy(r,-1.0,src,Ap);
+	  Linop.HermOp(psi,r);
+	  axpy(r,-1.0,src,r);
 	  RealD true_resid = norm2(r);
-	  std::cout<<"ConjugateResidual: Converged on iteration " <<k
+	  std::cout<<"PrecGeneralisedConjugateResidual: Converged on iteration " <<steps
 		   << " computed residual "<<sqrt(cp/ssq)
 	           << " true residual "<<true_resid
 	           << " target "       <<Tolerance <<std::endl;
@@ -77,10 +57,92 @@ namespace Grid {
 	}
 
       }
-
-      std::cout<<"ConjugateResidual did NOT converge"<<std::endl;
+      std::cout<<"Variable Preconditioned GCR did not converge"<<std::endl;
       assert(0);
     }
+    RealD GCRnStep(LinearOperatorBase<Field> &Linop,const Field &src, Field &psi,RealD rsq){
+
+      RealD cp;
+      RealD a, b, c, d;
+      RealD zAz, zAAz;
+      RealD rAq, rq;
+
+      GridBase *grid = src._grid;
+
+      Field r(grid);
+      Field z(grid);
+      Field Az(grid);
+
+      ////////////////////////////////
+      // history for flexible orthog
+      ////////////////////////////////
+      std::vector<Field> q(mmax,grid);
+      std::vector<Field> p(mmax,grid);
+      std::vector<RealD> qq(mmax);
+      
+      //////////////////////////////////
+      // initial guess x0 is taken as nonzero.
+      // r0=src-A x0 = src
+      //////////////////////////////////
+      Linop.HermOpAndNorm(psi,Az,zAz,zAAz); 
+      r=src-Az;
+      
+      /////////////////////
+      // p = Prec(r)
+      /////////////////////
+      Preconditioner(r,z);
+      Linop.HermOpAndNorm(z,Az,zAz,zAAz); 
+
+      //p[0],q[0],qq[0] 
+      p[0]= z;
+      q[0]= Az;
+      qq[0]= zAAz;
+
+      cp =norm2(r);
+
+      for(int k=0;k<nstep;k++){
+
+	steps++;
+
+	int kp     = k+1;
+	int peri_k = k %mmax;
+	int peri_kp= kp%mmax;
+
+	rq= real(innerProduct(r,q[peri_k])); // what if rAr not real?
+	a = rq/qq[peri_k];
+
+	axpy(psi,a,p[peri_k],psi);         
+
+	cp = axpy_norm(r,-a,q[peri_k],r);  
+
+	if((k==nstep-1)||(cp<rsq)){
+	  return cp;
+	}
+
+	Preconditioner(r,z);// solve Az = r
+
+	Linop.HermOpAndNorm(z,Az,zAz,zAAz);
+
+	q[peri_kp]=Az;
+	p[peri_kp]=z;
+
+	int northog = ((kp)>(mmax-1))?(mmax-1):(kp);  // if more than mmax done, we orthog all mmax history.
+	for(int back=0;back<northog;back++){
+
+	  int peri_back=(k-back)%mmax;   	  assert((k-back)>=0);
+
+	  b=-real(innerProduct(q[peri_back],Az))/qq[peri_back];
+	  p[peri_kp]=p[peri_kp]+b*p[peri_back];
+	  q[peri_kp]=q[peri_kp]+b*q[peri_back];
+
+	}
+	qq[peri_kp]=norm2(q[peri_kp]); // could use axpy_norm
+
+
+      }
+      assert(0); // never reached
+      return cp;
+    }
   };
 }
 #endif
diff --git a/lib/qcd/QCD.h b/lib/qcd/QCD.h
index ab180ff6..b3c6d138 100644
--- a/lib/qcd/QCD.h
+++ b/lib/qcd/QCD.h
@@ -411,6 +411,7 @@ namespace QCD {
 #include <qcd/utils/LinalgUtils.h>
 #include <qcd/utils/CovariantCshift.h>
 #include <qcd/utils/WilsonLoops.h>
+#include <qcd/utils/SUn.h>
 #include <qcd/action/Actions.h>
 
 #endif
diff --git a/tests/Make.inc b/tests/Make.inc
index 41d4ab8f..4d636cc5 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,5 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_Metropolis Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cg Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_gamma Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
+bin_PROGRAMS = Test_GaugeAction Test_Metropolis Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cg Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_gamma Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
 
 
 Test_GaugeAction_SOURCES=Test_GaugeAction.cc
@@ -74,6 +74,10 @@ Test_dwf_even_odd_SOURCES=Test_dwf_even_odd.cc
 Test_dwf_even_odd_LDADD=-lGrid
 
 
+Test_dwf_fpgcr_SOURCES=Test_dwf_fpgcr.cc
+Test_dwf_fpgcr_LDADD=-lGrid
+
+
 Test_gamma_SOURCES=Test_gamma.cc
 Test_gamma_LDADD=-lGrid
 
diff --git a/tests/Test_dwf_fpgcr.cc b/tests/Test_dwf_fpgcr.cc
new file mode 100644
index 00000000..a0d6ee96
--- /dev/null
+++ b/tests/Test_dwf_fpgcr.cc
@@ -0,0 +1,90 @@
+#include <Grid.h>
+#include <algorithms/iterative/PrecGeneralisedConjugateResidual.h>
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+
+
+  const int Ls=8;
+
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);  RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);  RNG4.SeedFixedIntegers(seeds4);
+
+  LatticeFermion    src(FGrid); random(RNG5,src);
+  LatticeFermion result(FGrid); result=zero;
+  LatticeGaugeField Umu(UGrid); 
+
+  SU3::HotConfiguration(RNG4,Umu);
+
+  TrivialPrecon<LatticeFermion> simple;
+
+  PrecGeneralisedConjugateResidual<LatticeFermion> PGCR(1.0e-6,10000,simple,4,160);
+
+  ConjugateResidual<LatticeFermion> CR(1.0e-6,10000);
+
+  ConjugateGradient<LatticeFermion> CG(1.0e-6,10000);
+  
+  RealD mass=0.5;
+  RealD M5=1.8;
+  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+
+  std::cout<<"*********************************************************"<<std::endl;
+  std::cout<<"* Solving with MdagM VPGCR "<<std::endl;
+  std::cout<<"*********************************************************"<<std::endl;
+  MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermOp(Ddwf);
+  result=zero;
+  PGCR(HermOp,src,result);
+
+  std::cout<<"*********************************************************"<<std::endl;
+  std::cout<<"* Solving with g5-VPGCR "<<std::endl;
+  std::cout<<"*********************************************************"<<std::endl;
+  Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> g5HermOp(Ddwf);
+  result=zero;
+  PGCR(g5HermOp,src,result);
+
+  std::cout<<"*********************************************************"<<std::endl;
+  std::cout<<"* Solving with MdagM-CR "<<std::endl;
+  std::cout<<"*********************************************************"<<std::endl;
+  result=zero;
+  CR(HermOp,src,result);
+
+  std::cout<<"*********************************************************"<<std::endl;
+  std::cout<<"* Solving with g5-CR "<<std::endl;
+  std::cout<<"*********************************************************"<<std::endl;
+  result=zero;
+  CR(g5HermOp,src,result);
+
+  std::cout<<"*********************************************************"<<std::endl;
+  std::cout<<"* Solving with MdagM-CG "<<std::endl;
+  std::cout<<"*********************************************************"<<std::endl;
+  result=zero;
+  CG(HermOp,src,result);
+
+  Grid_finalize();
+}

From e415587e8feccef8dd6bf559d197c1434355d6b5 Mon Sep 17 00:00:00 2001
From: Azusa Yamaguchi <ayamaguc@staffmail.ed.ac.uk>
Date: Sun, 21 Jun 2015 11:03:55 +0100
Subject: [PATCH 362/429] Abstract preconditioner

---
 lib/algorithms/Preconditioner.h | 19 +++++++++++++++++++
 1 file changed, 19 insertions(+)
 create mode 100644 lib/algorithms/Preconditioner.h

diff --git a/lib/algorithms/Preconditioner.h b/lib/algorithms/Preconditioner.h
new file mode 100644
index 00000000..e62c3ab7
--- /dev/null
+++ b/lib/algorithms/Preconditioner.h
@@ -0,0 +1,19 @@
+#ifndef GRID_PRECONDITIONER_H
+#define GRID_PRECONDITIONER_H
+
+namespace Grid {
+
+  template<class Field> class Preconditioner :  public LinearFunction<Field> { 
+    virtual void operator()(const Field &src, Field & psi)=0;
+  };
+
+  template<class Field> class TrivialPrecon :  public Preconditioner<Field> { 
+  public:
+    void operator()(const Field &src, Field & psi){
+      psi = src;
+    }
+    TrivialPrecon(void){};
+  };
+
+}
+#endif

From dec68e5c0e6b524a30ad5594478a33e7f45f1293 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 22 Jun 2015 12:49:44 +0100
Subject: [PATCH 363/429] Some small steps towards a multigrid

---
 lib/Make.inc                                  |   2 +-
 lib/algorithms/CoarsenedMatrix.h              |   6 +-
 lib/algorithms/iterative/ConjugateGradient.h  |   2 +-
 .../PrecGeneralisedConjugateResidual.h        |   6 +-
 lib/qcd/action/fermion/WilsonFermion5D.cc     |  39 ++--
 lib/qcd/utils/SUn.h                           |   4 +-
 tests/InvSqrt.gnu                             |   0
 tests/Make.inc                                |  14 +-
 tests/Sqrt.gnu                                |   2 -
 tests/Test_cayley_ldop_cg.cc                  | 155 ---------------
 tests/Test_cayley_ldop_cr.cc                  | 103 ++--------
 tests/Test_dwf_hdcr.cc                        | 188 ++++++++++++++++++
 12 files changed, 243 insertions(+), 278 deletions(-)
 delete mode 100644 tests/InvSqrt.gnu
 delete mode 100644 tests/Sqrt.gnu
 delete mode 100644 tests/Test_cayley_ldop_cg.cc
 create mode 100644 tests/Test_dwf_hdcr.cc

diff --git a/lib/Make.inc b/lib/Make.inc
index 6b5ae64f..91a06fcb 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/BfmHDCG.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
 
 CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/algorithms/CoarsenedMatrix.h b/lib/algorithms/CoarsenedMatrix.h
index 2576a696..9ea77b84 100644
--- a/lib/algorithms/CoarsenedMatrix.h
+++ b/lib/algorithms/CoarsenedMatrix.h
@@ -85,10 +85,6 @@ namespace Grid {
     void Orthogonalise(void){
       CoarseScalar InnerProd(CoarseGrid); 
       blockOrthogonalise(InnerProd,subspace);
-#if 1
-      //      CheckOrthogonal();
-#endif
-
     } 
     void CheckOrthogonal(void){
       CoarseVector iProj(CoarseGrid); 
@@ -125,7 +121,7 @@ namespace Grid {
 
       RealD scale;
 
-      ConjugateGradient<FineField> CG(1.0e-4,10000);
+      ConjugateGradient<FineField> CG(1.0e-3,10000);
       FineField noise(FineGrid);
       FineField Mn(FineGrid);
 
diff --git a/lib/algorithms/iterative/ConjugateGradient.h b/lib/algorithms/iterative/ConjugateGradient.h
index e281d765..3829dad8 100644
--- a/lib/algorithms/iterative/ConjugateGradient.h
+++ b/lib/algorithms/iterative/ConjugateGradient.h
@@ -15,7 +15,7 @@ public:
     Integer MaxIterations;
     int verbose;
     ConjugateGradient(RealD tol,Integer maxit) : Tolerance(tol), MaxIterations(maxit) { 
-      verbose=0;
+      verbose=1;
     };
 
 
diff --git a/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h b/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h
index bd5bfbfe..77b7df49 100644
--- a/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h
+++ b/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h
@@ -48,11 +48,11 @@ namespace Grid {
 	if(cp<rsq) {
 	  Linop.HermOp(psi,r);
 	  axpy(r,-1.0,src,r);
-	  RealD true_resid = norm2(r);
+	  RealD tr = norm2(r);
 	  std::cout<<"PrecGeneralisedConjugateResidual: Converged on iteration " <<steps
 		   << " computed residual "<<sqrt(cp/ssq)
-	           << " true residual "<<true_resid
-	           << " target "       <<Tolerance <<std::endl;
+	           << " true residual "    <<sqrt(tr/ssq)
+	           << " target "           <<Tolerance <<std::endl;
 	  return;
 	}
 
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.cc b/lib/qcd/action/fermion/WilsonFermion5D.cc
index 358b0c9a..ed57d37d 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.cc
+++ b/lib/qcd/action/fermion/WilsonFermion5D.cc
@@ -87,8 +87,8 @@ void WilsonFermion5D::DhopDir(const LatticeFermion &in, LatticeFermion &out,int
 {
   int dir = dir5-1; // Maps to the ordering above in "directions" that is passed to stencil
                     // we drop off the innermost fifth dimension
-  assert( (disp==1)||(disp==-1) );
-  assert( (dir>=0)&&(dir<4) ); //must do x,y,z or t;
+  //  assert( (disp==1)||(disp==-1) );
+  //  assert( (dir>=0)&&(dir<4) ); //must do x,y,z or t;
 
   WilsonCompressor compressor(DaggerNo);
   Stencil.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
@@ -100,7 +100,7 @@ void WilsonFermion5D::DhopDir(const LatticeFermion &in, LatticeFermion &out,int
   assert(dirdisp<=7);
   assert(dirdisp>=0);
 
-PARALLEL_FOR_LOOP
+//PARALLEL_FOR_LOOP
   for(int ss=0;ss<Umu._grid->oSites();ss++){
     for(int s=0;s<Ls;s++){
       int sU=ss;
@@ -114,7 +114,7 @@ void WilsonFermion5D::DhopInternal(CartesianStencil & st, LebesgueOrder &lo,
 				   LatticeDoubledGaugeField & U,
 			   const LatticeFermion &in, LatticeFermion &out,int dag)
 {
-  assert((dag==DaggerNo) ||(dag==DaggerYes));
+  //  assert((dag==DaggerNo) ||(dag==DaggerYes));
 
   WilsonCompressor compressor(dag);
 
@@ -127,29 +127,32 @@ void WilsonFermion5D::DhopInternal(CartesianStencil & st, LebesgueOrder &lo,
   // - 8 linear access unit stride streams per thread for Fermion for hw prefetchable.
   if ( dag == DaggerYes ) {
     if( HandOptDslash ) {
-PARALLEL_FOR_LOOP
+#pragma parallel for
       for(int ss=0;ss<U._grid->oSites();ss++){
-	for(int s=0;s<Ls;s++){
-	  //int sU=lo.Reorder(ss);
-	  int sU=ss;
-	  int sF = s+Ls*sU;
-	  DiracOptHand::DhopSiteDag(st,U,comm_buf,sF,sU,in,out);
+	{
+	  for(int s=0;s<Ls;s++){
+	    int sU=ss;
+	    int sF = s+Ls*sU;
+	    DiracOptHand::DhopSiteDag(st,U,comm_buf,sF,sU,in,out);
+	  }
 	}
       }
     } else { 
-PARALLEL_FOR_LOOP
+#pragma parallel for
       for(int ss=0;ss<U._grid->oSites();ss++){
-	for(int s=0;s<Ls;s++){
-	  //	  int sU=lo.Reorder(ss);
-	  int sU=ss;
-	  int sF = s+Ls*sU;
-	  DiracOpt::DhopSiteDag(st,U,comm_buf,sF,sU,in,out);
+	{
+	  int sd;
+	  for(sd=0;sd<Ls;sd++){
+	    int sU=ss;
+	    int sF = sd+Ls*sU;
+	    DiracOpt::DhopSiteDag(st,U,comm_buf,sF,sU,in,out);
+	  }
 	}
       }
     }
   } else {
     if( HandOptDslash ) {
-PARALLEL_FOR_LOOP
+#pragma parallel for
       for(int ss=0;ss<U._grid->oSites();ss++){
 	for(int s=0;s<Ls;s++){
 	  //	  int sU=lo.Reorder(ss);
@@ -160,7 +163,7 @@ PARALLEL_FOR_LOOP
       }
 
     } else { 
-PARALLEL_FOR_LOOP
+#pragma parallel for
       for(int ss=0;ss<U._grid->oSites();ss++){
 	for(int s=0;s<Ls;s++){
 	  //	  int sU=lo.Reorder(ss);
diff --git a/lib/qcd/utils/SUn.h b/lib/qcd/utils/SUn.h
index 647699c8..5e551274 100644
--- a/lib/qcd/utils/SUn.h
+++ b/lib/qcd/utils/SUn.h
@@ -166,7 +166,7 @@ public:
     su2SubGroupIndex(i0,i1,su2_index);
 
 PARALLEL_FOR_LOOP
-    for(int ss=0;ss!=grid->oSites();ss++){
+    for(int ss=0;ss<grid->oSites();ss++){
       subgroup._odata[ss]()()(0,0) = source._odata[ss]()()(i0,i0);
       subgroup._odata[ss]()()(0,1) = source._odata[ss]()()(i0,i1);
       subgroup._odata[ss]()()(1,0) = source._odata[ss]()()(i1,i0);
@@ -201,7 +201,7 @@ PARALLEL_FOR_LOOP
 
     dest = 1.0; // start out with identity
 PARALLEL_FOR_LOOP
-    for(int ss=0;ss!=grid->oSites();ss++){
+    for(int ss=0;ss<grid->oSites();ss++){
       dest._odata[ss]()()(i0,i0) = subgroup._odata[ss]()()(0,0);
       dest._odata[ss]()()(i0,i1) = subgroup._odata[ss]()()(0,1);
       dest._odata[ss]()()(i1,i0) = subgroup._odata[ss]()()(1,0);
diff --git a/tests/InvSqrt.gnu b/tests/InvSqrt.gnu
deleted file mode 100644
index e69de29b..00000000
diff --git a/tests/Make.inc b/tests/Make.inc
index 4d636cc5..4ad511bd 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,15 +1,11 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_Metropolis Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cg Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_gamma Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
+bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
 
 
 Test_GaugeAction_SOURCES=Test_GaugeAction.cc
 Test_GaugeAction_LDADD=-lGrid
 
 
-Test_Metropolis_SOURCES=Test_Metropolis.cc
-Test_Metropolis_LDADD=-lGrid
-
-
 Test_cayley_cg_SOURCES=Test_cayley_cg.cc
 Test_cayley_cg_LDADD=-lGrid
 
@@ -22,10 +18,6 @@ Test_cayley_even_odd_SOURCES=Test_cayley_even_odd.cc
 Test_cayley_even_odd_LDADD=-lGrid
 
 
-Test_cayley_ldop_cg_SOURCES=Test_cayley_ldop_cg.cc
-Test_cayley_ldop_cg_LDADD=-lGrid
-
-
 Test_cayley_ldop_cr_SOURCES=Test_cayley_ldop_cr.cc
 Test_cayley_ldop_cr_LDADD=-lGrid
 
@@ -78,6 +70,10 @@ Test_dwf_fpgcr_SOURCES=Test_dwf_fpgcr.cc
 Test_dwf_fpgcr_LDADD=-lGrid
 
 
+Test_dwf_hdcr_SOURCES=Test_dwf_hdcr.cc
+Test_dwf_hdcr_LDADD=-lGrid
+
+
 Test_gamma_SOURCES=Test_gamma.cc
 Test_gamma_LDADD=-lGrid
 
diff --git a/tests/Sqrt.gnu b/tests/Sqrt.gnu
deleted file mode 100644
index ae56ab97..00000000
--- a/tests/Sqrt.gnu
+++ /dev/null
@@ -1,2 +0,0 @@
-f(x) = 6.81384+(-2.34645e-06/(x+0.000228091))+(-1.51593e-05/(x+0.00112084))+(-6.89254e-05/(x+0.003496))+(-0.000288983/(x+0.00954309))+(-0.00119277/(x+0.024928))+(-0.0050183/(x+0.0646627))+(-0.0226449/(x+0.171576))+(-0.123767/(x+0.491792))+(-1.1705/(x+1.78667))+(-102.992/(x+18.4866));
-f(x) = 0.14676+(0.00952992/(x+5.40933e-05))+(0.0115952/(x+0.000559699))+(0.0161824/(x+0.00203338))+(0.0243252/(x+0.00582831))+(0.0379533/(x+0.0154649))+(0.060699/(x+0.0401156))+(0.100345/(x+0.104788))+(0.178335/(x+0.286042))+(0.381586/(x+0.892189))+(1.42625/(x+4.38422));
diff --git a/tests/Test_cayley_ldop_cg.cc b/tests/Test_cayley_ldop_cg.cc
deleted file mode 100644
index f960ab7b..00000000
--- a/tests/Test_cayley_ldop_cg.cc
+++ /dev/null
@@ -1,155 +0,0 @@
-#include <Grid.h>
-#include <qcd/utils/WilsonLoops.h>
-#include <qcd/utils/SUn.h>
-
-using namespace std;
-using namespace Grid;
-using namespace Grid::QCD;
-
-
-template<class d>
-struct scal {
-  d internal;
-};
-
-  Gamma::GammaMatrix Gmu [] = {
-    Gamma::GammaX,
-    Gamma::GammaY,
-    Gamma::GammaZ,
-    Gamma::GammaT
-  };
-
-double lowpass(double x)
-{
-  return pow(x*x+1.0,-2);
-}
-
-int main (int argc, char ** argv)
-{
-  Grid_init(&argc,&argv);
-
-  Chebyshev<LatticeFermion> filter(-150.0, 150.0,16, lowpass);
-  ofstream csv(std::string("filter.dat"),std::ios::out|std::ios::trunc);
-  filter.csv(csv);
-  csv.close();
-
-  const int Ls=8;
-
-  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
-  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
-
-  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
-  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
-
-  // Construct a coarsened grid
-  std::vector<int> clatt = GridDefaultLatt();
-  for(int d=0;d<clatt.size();d++){
-    clatt[d] = clatt[d]/2;
-  }
-  GridCartesian *Coarse4d =  SpaceTimeGrid::makeFourDimGrid(clatt, GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());;
-  GridCartesian *Coarse5d =  SpaceTimeGrid::makeFiveDimGrid(1,Coarse4d);
-
-  std::vector<int> seeds4({1,2,3,4});
-  std::vector<int> seeds5({5,6,7,8});
-  std::vector<int> cseeds({5,6,7,8});
-  GridParallelRNG          RNG5(FGrid);   RNG5.SeedFixedIntegers(seeds5);
-  GridParallelRNG          RNG4(UGrid);   RNG4.SeedFixedIntegers(seeds4);
-  GridParallelRNG          CRNG(Coarse5d);CRNG.SeedFixedIntegers(cseeds);
-
-  LatticeFermion    src(FGrid); gaussian(RNG5,src);
-  LatticeFermion result(FGrid); result=zero;
-  LatticeFermion    ref(FGrid); ref=zero;
-  LatticeFermion    tmp(FGrid);
-  LatticeFermion    err(FGrid);
-  LatticeGaugeField Umu(UGrid); 
-
-
-
-  //gaussian(RNG4,Umu);
-  //random(RNG4,Umu);
-
-  NerscField header;
-  std::string file("./ckpoint_lat.400");
-  readNerscConfiguration(Umu,header,file);
-  //  SU3::ColdConfiguration(RNG4,Umu);
-  //  SU3::TepidConfiguration(RNG4,Umu);
-  //  SU3::HotConfiguration(RNG4,Umu);
-  //  Umu=zero;
-
-#if 0
-  LatticeColourMatrix U(UGrid);
-  for(int nn=0;nn<Nd;nn++){
-    U=peekIndex<LorentzIndex>(Umu,nn);
-    U=U*adj(U)-1.0;
-    std::cout<<"SU3 test "<<norm2(U)<<std::endl;
-  }
-#endif  
-  RealD mass=0.1;
-  RealD M5=1.5;
-
-  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
-  Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> HermIndefOp(Ddwf);
-
-  const int nbasis = 8;
-
-#if 0
-  std::vector<LatticeFermion> subspace(nbasis,FGrid);
-  LatticeFermion noise(FGrid);
-  LatticeFermion ms(FGrid);
-  for(int b=0;b<nbasis;b++){
-
-    gaussian(RNG5,noise);
-    RealD scale = pow(norm2(noise),-0.5); 
-    noise=noise*scale;
-
-    HermIndefOp.Op(noise,ms); std::cout << "Noise    "<<b<<" Ms "<<norm2(ms)<< " "<< norm2(noise)<<std::endl;
-
-    //    filter(HermIndefOp,noise,subspace[b]);
-    // inverse iteration
-    MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
-    ConjugateGradient<LatticeFermion> CG(1.0e-4,10000);
-
-    for(int i=0;i<1;i++){
-
-      CG(HermDefOp,noise,subspace[b]);
-      noise = subspace[b];
-
-      scale = pow(norm2(noise),-0.5); 
-      noise=noise*scale;
-      HermDefOp.Op(noise,ms); std::cout << "filt    "<<b<<" <u|H|u> "<<norm2(ms)<< " "<< norm2(noise)<<std::endl;
-    }
-
-    subspace[b] = noise;
-    HermIndefOp.Op(subspace[b],ms); std::cout << "Filtered "<<b<<" Ms "<<norm2(ms)<< " "<<norm2(subspace[b]) <<std::endl;
-
-  }
-  std::cout << "Computed randoms"<< std::endl;
-#else
-  std::cout<<"Calling Aggregation class" <<std::endl;
-  MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
-  typedef Aggregation<vSpinColourVector,vTComplex,nbasis> Subspace;
-  Subspace Aggregates(Coarse5d,FGrid);
-  Aggregates.CreateSubspace(RNG5,HermDefOp);
-  std::cout << "Called aggregation class"<< std::endl;
-#endif
-
-
-  typedef CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis> LittleDiracOperator;
-  typedef LittleDiracOperator::CoarseVector CoarseVector;
-
-  LittleDiracOperator LittleDiracOp(*Coarse5d);
-  LittleDiracOp.CoarsenOperator(FGrid,HermIndefOp,Aggregates);
-  
-  CoarseVector c_src (Coarse5d);
-  CoarseVector c_res (Coarse5d);
-  gaussian(CRNG,c_src);
-  c_res=zero;
-
-  std::cout << "Solving CG on coarse space "<< std::endl;
-  MdagMLinearOperator<LittleDiracOperator,CoarseVector> PosdefLdop(LittleDiracOp);
-  ConjugateGradient<CoarseVector> CG(1.0e-6,10000);
-  CG(PosdefLdop,c_src,c_res);
-
-  std::cout << "Done "<< std::endl;
-  Grid_finalize();
-}
diff --git a/tests/Test_cayley_ldop_cr.cc b/tests/Test_cayley_ldop_cr.cc
index c4001b55..8926c19c 100644
--- a/tests/Test_cayley_ldop_cr.cc
+++ b/tests/Test_cayley_ldop_cr.cc
@@ -1,38 +1,13 @@
 #include <Grid.h>
-#include <qcd/utils/WilsonLoops.h>
-#include <qcd/utils/SUn.h>
 
 using namespace std;
 using namespace Grid;
 using namespace Grid::QCD;
 
-
-template<class d>
-struct scal {
-  d internal;
-};
-
-  Gamma::GammaMatrix Gmu [] = {
-    Gamma::GammaX,
-    Gamma::GammaY,
-    Gamma::GammaZ,
-    Gamma::GammaT
-  };
-
-double lowpass(double x)
-{
-  return pow(x*x+1.0,-2);
-}
-
 int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  Chebyshev<LatticeFermion> filter(-150.0, 150.0,16, lowpass);
-  ofstream csv(std::string("filter.dat"),std::ios::out|std::ios::trunc);
-  filter.csv(csv);
-  csv.close();
-
   const int Ls=8;
 
   GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
@@ -41,7 +16,9 @@ int main (int argc, char ** argv)
   GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
   GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
 
-  // Construct a coarsened grid
+  ///////////////////////////////////////////////////
+  // Construct a coarsened grid; utility for this?
+  ///////////////////////////////////////////////////
   std::vector<int> clatt = GridDefaultLatt();
   for(int d=0;d<clatt.size();d++){
     clatt[d] = clatt[d]/2;
@@ -63,80 +40,38 @@ int main (int argc, char ** argv)
   LatticeFermion    err(FGrid);
   LatticeGaugeField Umu(UGrid); 
 
-
-
-  //gaussian(RNG4,Umu);
-  //random(RNG4,Umu);
-
   NerscField header;
   std::string file("./ckpoint_lat.400");
   readNerscConfiguration(Umu,header,file);
+
   //  SU3::ColdConfiguration(RNG4,Umu);
   //  SU3::TepidConfiguration(RNG4,Umu);
   //  SU3::HotConfiguration(RNG4,Umu);
   //  Umu=zero;
 
-#if 0
-  LatticeColourMatrix U(UGrid);
-  for(int nn=0;nn<Nd;nn++){
-    U=peekIndex<LorentzIndex>(Umu,nn);
-    U=U*adj(U)-1.0;
-    std::cout<<"SU3 test "<<norm2(U)<<std::endl;
-  }
-#endif  
   RealD mass=0.1;
   RealD M5=1.5;
 
+  std::cout << "**************************************************"<< std::endl;
+  std::cout << "Building g5R5 hermitian DWF operator" <<std::endl;
+  std::cout << "**************************************************"<< std::endl;
   DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
   Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> HermIndefOp(Ddwf);
 
   const int nbasis = 8;
 
-#if 0
-  std::vector<LatticeFermion> subspace(nbasis,FGrid);
-  LatticeFermion noise(FGrid);
-  LatticeFermion ms(FGrid);
-  for(int b=0;b<nbasis;b++){
+  typedef Aggregation<vSpinColourVector,vTComplex,nbasis>     Subspace;
+  typedef CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis> LittleDiracOperator;
+  typedef LittleDiracOperator::CoarseVector                   CoarseVector;
 
-    gaussian(RNG5,noise);
-    RealD scale = pow(norm2(noise),-0.5); 
-    noise=noise*scale;
-
-    HermIndefOp.Op(noise,ms); std::cout << "Noise    "<<b<<" Ms "<<norm2(ms)<< " "<< norm2(noise)<<std::endl;
-
-    //    filter(HermIndefOp,noise,subspace[b]);
-    // inverse iteration
-    MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
-    ConjugateGradient<LatticeFermion> CG(1.0e-4,10000);
-
-    for(int i=0;i<1;i++){
-
-      CG(HermDefOp,noise,subspace[b]);
-      noise = subspace[b];
-
-      scale = pow(norm2(noise),-0.5); 
-      noise=noise*scale;
-      HermDefOp.Op(noise,ms); std::cout << "filt    "<<b<<" <u|H|u> "<<norm2(ms)<< " "<< norm2(noise)<<std::endl;
-    }
-
-    subspace[b] = noise;
-    HermIndefOp.Op(subspace[b],ms); std::cout << "Filtered "<<b<<" Ms "<<norm2(ms)<< " "<<norm2(subspace[b]) <<std::endl;
-
-  }
-  std::cout << "Computed randoms"<< std::endl;
-#else
-  std::cout<<"Calling Aggregation class" <<std::endl;
+  std::cout << "**************************************************"<< std::endl;
+  std::cout << "Calling Aggregation class to build subspace" <<std::endl;
+  std::cout << "**************************************************"<< std::endl;
   MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
-  typedef Aggregation<vSpinColourVector,vTComplex,nbasis> Subspace;
   Subspace Aggregates(Coarse5d,FGrid);
   Aggregates.CreateSubspace(RNG5,HermDefOp);
-  std::cout << "Called aggregation class"<< std::endl;
-#endif
 
 
-  typedef CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis> LittleDiracOperator;
-  typedef LittleDiracOperator::CoarseVector CoarseVector;
-
   LittleDiracOperator LittleDiracOp(*Coarse5d);
   LittleDiracOp.CoarsenOperator(FGrid,HermIndefOp,Aggregates);
   
@@ -145,18 +80,22 @@ int main (int argc, char ** argv)
   gaussian(CRNG,c_src);
   c_res=zero;
 
-
-  std::cout << "Solving CG on coarse space "<< std::endl;
+  std::cout << "**************************************************"<< std::endl;
+  std::cout << "Solving mdagm-CG on coarse space "<< std::endl;
+  std::cout << "**************************************************"<< std::endl;
   MdagMLinearOperator<LittleDiracOperator,CoarseVector> PosdefLdop(LittleDiracOp);
   ConjugateGradient<CoarseVector> CG(1.0e-6,10000);
   CG(PosdefLdop,c_src,c_res);
 
-
-  std::cout << "Solving MCR on coarse space "<< std::endl;
+  std::cout << "**************************************************"<< std::endl;
+  std::cout << "Solving indef-MCR on coarse space "<< std::endl;
+  std::cout << "**************************************************"<< std::endl;
   HermitianLinearOperator<LittleDiracOperator,CoarseVector> HermIndefLdop(LittleDiracOp);
   ConjugateResidual<CoarseVector> MCR(1.0e-6,10000);
   MCR(HermIndefLdop,c_src,c_res);
 
+  std::cout << "**************************************************"<< std::endl;
   std::cout << "Done "<< std::endl;
+  std::cout << "**************************************************"<< std::endl;
   Grid_finalize();
 }
diff --git a/tests/Test_dwf_hdcr.cc b/tests/Test_dwf_hdcr.cc
new file mode 100644
index 00000000..44b05714
--- /dev/null
+++ b/tests/Test_dwf_hdcr.cc
@@ -0,0 +1,188 @@
+#include <Grid.h>
+#include <algorithms/iterative/PrecGeneralisedConjugateResidual.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class Fobj,class CComplex,int nbasis>
+class MultiGridPreconditioner : public LinearFunction< Lattice<Fobj> > {
+public:
+
+  typedef Aggregation<Fobj,CComplex,nbasis> Aggregates;
+  typedef CoarsenedMatrix<Fobj,CComplex,nbasis> CoarseOperator;
+
+  typedef typename Aggregation<Fobj,CComplex,nbasis>::siteVector     siteVector;
+  typedef typename Aggregation<Fobj,CComplex,nbasis>::CoarseScalar CoarseScalar;
+  typedef typename Aggregation<Fobj,CComplex,nbasis>::CoarseVector CoarseVector;
+  typedef typename Aggregation<Fobj,CComplex,nbasis>::CoarseMatrix CoarseMatrix;
+  typedef typename Aggregation<Fobj,CComplex,nbasis>::FineField    FineField;
+  typedef LinearOperatorBase<FineField>                            FineOperator;
+
+  Aggregates     & _Aggregates;
+  CoarseOperator & _CoarseOperator;
+  FineOperator   & _FineOperator;
+
+  // Constructor
+  MultiGridPreconditioner(Aggregates &Agg, CoarseOperator &Coarse, FineOperator &Fine) 
+    : _Aggregates(Agg),
+      _CoarseOperator(Coarse),
+      _FineOperator(Fine)
+  {
+  }
+
+  void operator()(const FineField &in, FineField & out) {
+
+    FineField Min(in._grid);
+
+    CoarseVector Csrc(_CoarseOperator.Grid());
+    CoarseVector Ctmp(_CoarseOperator.Grid());
+    CoarseVector Csol(_CoarseOperator.Grid());
+
+    // Monitor completeness of low mode space
+    _Aggregates.ProjectToSubspace  (Csrc,in);
+    _Aggregates.PromoteFromSubspace(Csrc,out);
+    std::cout<<"Completeness: "<<std::sqrt(norm2(out)/norm2(in))<<std::endl;
+
+    ConjugateResidual<FineField>    MCR(1.0e-2,1000);
+    ConjugateGradient<CoarseVector>  CG(1.0e-2,10000);
+
+    // [PTM+Q] in = [1 - Q A] M in + Q in = Min + Q [ in -A Min]
+    // Smoothing step, followed by coarse grid correction
+
+    MCR(_FineOperator,in,Min);
+    _FineOperator.Op(Min,out);
+    out = in -out; // out = in - A Min
+
+    MdagMLinearOperator<CoarseOperator,CoarseVector>     MdagMOp(_CoarseOperator);
+    HermitianLinearOperator<CoarseOperator,CoarseVector> HermOp(_CoarseOperator);
+    Csol=zero;
+    _Aggregates.ProjectToSubspace  (Csrc,out);
+    HermOp.AdjOp(Csrc,Ctmp);// Normal equations
+    CG(MdagMOp  ,Ctmp,Csol);
+    _Aggregates.PromoteFromSubspace(Csol,out);
+
+    out = Min + out;;
+  }
+
+};
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  const int Ls=8;
+
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+  ///////////////////////////////////////////////////
+  // Construct a coarsened grid; utility for this?
+  ///////////////////////////////////////////////////
+  std::vector<int> clatt = GridDefaultLatt();
+  for(int d=0;d<clatt.size();d++){
+    clatt[d] = clatt[d]/4;
+  }
+  GridCartesian *Coarse4d =  SpaceTimeGrid::makeFourDimGrid(clatt, GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());;
+  GridCartesian *Coarse5d =  SpaceTimeGrid::makeFiveDimGrid(1,Coarse4d);
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  std::vector<int> cseeds({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);   RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);   RNG4.SeedFixedIntegers(seeds4);
+  GridParallelRNG          CRNG(Coarse5d);CRNG.SeedFixedIntegers(cseeds);
+
+  LatticeFermion    src(FGrid); gaussian(RNG5,src);
+  LatticeFermion result(FGrid); result=zero;
+  LatticeFermion    ref(FGrid); ref=zero;
+  LatticeFermion    tmp(FGrid);
+  LatticeFermion    err(FGrid);
+  LatticeGaugeField Umu(UGrid); 
+
+  NerscField header;
+  std::string file("./ckpoint_lat.4000");
+  readNerscConfiguration(Umu,header,file);
+
+  //  SU3::ColdConfiguration(RNG4,Umu);
+  //  SU3::TepidConfiguration(RNG4,Umu);
+  //  SU3::HotConfiguration(RNG4,Umu);
+  //  Umu=zero;
+
+  RealD mass=0.04;
+  RealD M5=1.8;
+
+  std::cout << "**************************************************"<< std::endl;
+  std::cout << "Building g5R5 hermitian DWF operator" <<std::endl;
+  std::cout << "**************************************************"<< std::endl;
+  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+
+  const int nbasis = 4;
+
+  typedef Aggregation<vSpinColourVector,vTComplex,nbasis>              Subspace;
+  typedef CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis>          CoarseOperator;
+  typedef CoarseOperator::CoarseVector                                 CoarseVector;
+
+  std::cout << "**************************************************"<< std::endl;
+  std::cout << "Calling Aggregation class to build subspace" <<std::endl;
+  std::cout << "**************************************************"<< std::endl;
+  MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
+  Subspace Aggregates(Coarse5d,FGrid);
+  Aggregates.CreateSubspace(RNG5,HermDefOp);
+
+  std::cout << "**************************************************"<< std::endl;
+  std::cout << "Building coarse representation of Indef operator" <<std::endl;
+  std::cout << "**************************************************"<< std::endl;
+  Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> HermIndefOp(Ddwf);
+  CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis> LDOp(*Coarse5d);
+  LDOp.CoarsenOperator(FGrid,HermIndefOp,Aggregates);
+
+  std::cout << "**************************************************"<< std::endl;
+  std::cout << "Testing some coarse space solvers  " <<std::endl;
+  std::cout << "**************************************************"<< std::endl;
+  CoarseVector c_src (Coarse5d);
+  CoarseVector c_res (Coarse5d);
+  gaussian(CRNG,c_src);
+  c_res=zero;
+
+  std::cout << "**************************************************"<< std::endl;
+  std::cout << "Solving posdef-CG on coarse space "<< std::endl;
+  std::cout << "**************************************************"<< std::endl;
+  MdagMLinearOperator<CoarseOperator,CoarseVector> PosdefLdop(LDOp);
+  ConjugateGradient<CoarseVector> CG(1.0e-6,10000);
+  CG(PosdefLdop,c_src,c_res);
+
+  std::cout << "**************************************************"<< std::endl;
+  std::cout << "Solving indef-MCR on coarse space "<< std::endl;
+  std::cout << "**************************************************"<< std::endl;
+  HermitianLinearOperator<CoarseOperator,CoarseVector> HermIndefLdop(LDOp);
+  ConjugateResidual<CoarseVector> MCR(1.0e-6,10000);
+  //MCR(HermIndefLdop,c_src,c_res);
+
+  std::cout << "**************************************************"<< std::endl;
+  std::cout << "Building deflation preconditioner "<< std::endl;
+  std::cout << "**************************************************"<< std::endl;
+
+  MultiGridPreconditioner <vSpinColourVector,vTComplex,nbasis> Precon(Aggregates, LDOp,HermIndefOp);
+
+  std::cout << "**************************************************"<< std::endl;
+  std::cout << "Building a one level PGCR "<< std::endl;
+  std::cout << "**************************************************"<< std::endl;
+  TrivialPrecon<LatticeFermion> simple;
+  PrecGeneralisedConjugateResidual<LatticeFermion> GCR(1.0e-6,10000,simple,8,64);
+  GCR(HermIndefOp,src,result);
+
+  std::cout << "**************************************************"<< std::endl;
+  std::cout << "Building a two level PGCR "<< std::endl;
+  std::cout << "**************************************************"<< std::endl;
+  PrecGeneralisedConjugateResidual<LatticeFermion> PGCR(1.0e-6,10000,Precon,8,64);
+  PGCR(HermIndefOp,src,result);
+
+  std::cout << "**************************************************"<< std::endl;
+  std::cout << "Done "<< std::endl;
+  std::cout << "**************************************************"<< std::endl;
+  Grid_finalize();
+}

From a67b44ffa4f8659444c1874c30bae75b089f30dc Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 25 Jun 2015 10:59:53 +0100
Subject: [PATCH 364/429] Update Benchmark_comms.cc

---
 benchmarks/Benchmark_comms.cc | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/benchmarks/Benchmark_comms.cc b/benchmarks/Benchmark_comms.cc
index a24e7ed8..5a85269e 100644
--- a/benchmarks/Benchmark_comms.cc
+++ b/benchmarks/Benchmark_comms.cc
@@ -26,7 +26,10 @@ int main (int argc, char ** argv)
   for(int lat=4;lat<=32;lat+=2){
     for(int Ls=1;Ls<=16;Ls*=2){
 
-      std::vector<int> latt_size  ({lat,lat,lat,lat});
+      std::vector<int> latt_size  ({lat*mpi_layout[0],
+      				    lat*mpi_layout[1],
+      				    lat*mpi_layout[2],
+      				    lat*mpi_layout[3]});
 
       GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
 

From 7de5ccb879345c77082b8b79f129182d1e39a653 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 30 Jun 2015 15:00:19 +0100
Subject: [PATCH 365/429] big commit fixing nocompiles in defective C++11
 compilers (gcc, icpc). stared getting to near the bleeding edge I guess

---
 lib/qcd/utils/SUn.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/qcd/utils/SUn.h b/lib/qcd/utils/SUn.h
index 5e551274..4088a23b 100644
--- a/lib/qcd/utils/SUn.h
+++ b/lib/qcd/utils/SUn.h
@@ -547,7 +547,7 @@ Note that in step D setting B ~ X - A and using B in place of A in step E will g
     LatticeMatrix Umu(out._grid);
     for(int mu=0;mu<Nd;mu++){
       LieRandomize(pRNG,Umu,1.0);
-      pokeLorentz(out,Umu,mu);
+      pokeIndex<LorentzIndex>(out,Umu,mu);
     }
   }
   static void TepidConfiguration(GridParallelRNG &pRNG,LatticeGaugeField &out){

From 0971522f43bac7d7c71e361e4922c208917fa0ef Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 30 Jun 2015 15:01:26 +0100
Subject: [PATCH 366/429] big commit fixing nocompiles in defective C++11
 compilers (gcc, icpc). stared getting to near the bleeding edge I guess

---
 tests/Test_GaugeAction.cc        |   2 +-
 tests/Test_cf_coarsen_support.cc |   2 +-
 tests/Test_cf_cr_unprec.cc       |   2 +-
 tests/Test_dwf_cg_prec.cc        |   2 +-
 tests/Test_dwf_cg_schur.cc       |   2 +-
 tests/Test_dwf_cg_unprec.cc      |   2 +-
 tests/Test_dwf_cr_unprec.cc      |   2 +-
 tests/Test_dwf_hdcr.cc           | 218 ++++++++++++++++++++++++++-----
 tests/Test_gamma.cc              |   8 +-
 tests/Test_main.cc               |  30 ++---
 tests/Test_nersc_io.cc           |   4 +-
 tests/Test_quenched_update.cc    |   2 +-
 tests/Test_wilson_cg_prec.cc     |   2 +-
 tests/Test_wilson_cg_unprec.cc   |   6 -
 tests/Test_wilson_cr_unprec.cc   |   4 -
 15 files changed, 218 insertions(+), 70 deletions(-)

diff --git a/tests/Test_GaugeAction.cc b/tests/Test_GaugeAction.cc
index 56634341..1ce8306e 100644
--- a/tests/Test_GaugeAction.cc
+++ b/tests/Test_GaugeAction.cc
@@ -53,7 +53,7 @@ int main (int argc, char ** argv)
   readNerscConfiguration(Umu,header,file);
 
   for(int mu=0;mu<Nd;mu++){
-    U[mu] = peekIndex<3>(Umu,mu);
+    U[mu] = PeekIndex<LorentzIndex>(Umu,mu);
   }
 
   // Painful ; fix syntactical niceness : to check reader
diff --git a/tests/Test_cf_coarsen_support.cc b/tests/Test_cf_coarsen_support.cc
index 9d81a798..4a4cf3e9 100644
--- a/tests/Test_cf_coarsen_support.cc
+++ b/tests/Test_cf_coarsen_support.cc
@@ -41,7 +41,7 @@ int main (int argc, char ** argv)
 
   std::vector<LatticeColourMatrix> U(4,UGrid);
   for(int mu=0;mu<Nd;mu++){
-    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+    U[mu] = PeekIndex<LorentzIndex>(Umu,mu);
   }
   
   RealD mass=0.1;
diff --git a/tests/Test_cf_cr_unprec.cc b/tests/Test_cf_cr_unprec.cc
index 9c1aea60..e74fc85e 100644
--- a/tests/Test_cf_cr_unprec.cc
+++ b/tests/Test_cf_cr_unprec.cc
@@ -38,7 +38,7 @@ int main (int argc, char ** argv)
 
   std::vector<LatticeColourMatrix> U(4,UGrid);
   for(int mu=0;mu<Nd;mu++){
-    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+    U[mu] = PeekIndex<LorentzIndex>(Umu,mu);
   }
   
   RealD mass=0.1;
diff --git a/tests/Test_dwf_cg_prec.cc b/tests/Test_dwf_cg_prec.cc
index 32a0c3ae..28848ff9 100644
--- a/tests/Test_dwf_cg_prec.cc
+++ b/tests/Test_dwf_cg_prec.cc
@@ -38,7 +38,7 @@ int main (int argc, char ** argv)
 
   std::vector<LatticeColourMatrix> U(4,UGrid);
   for(int mu=0;mu<Nd;mu++){
-    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+    U[mu] = PeekIndex<LorentzIndex>(Umu,mu);
   }
   
   RealD mass=0.1;
diff --git a/tests/Test_dwf_cg_schur.cc b/tests/Test_dwf_cg_schur.cc
index d080045b..8415c489 100644
--- a/tests/Test_dwf_cg_schur.cc
+++ b/tests/Test_dwf_cg_schur.cc
@@ -38,7 +38,7 @@ int main (int argc, char ** argv)
 
   std::vector<LatticeColourMatrix> U(4,UGrid);
   for(int mu=0;mu<Nd;mu++){
-    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+    U[mu] = PeekIndex<LorentzIndex>(Umu,mu);
   }
   
   RealD mass=0.1;
diff --git a/tests/Test_dwf_cg_unprec.cc b/tests/Test_dwf_cg_unprec.cc
index 7a13be43..f4f63c50 100644
--- a/tests/Test_dwf_cg_unprec.cc
+++ b/tests/Test_dwf_cg_unprec.cc
@@ -38,7 +38,7 @@ int main (int argc, char ** argv)
 
   std::vector<LatticeColourMatrix> U(4,UGrid);
   for(int mu=0;mu<Nd;mu++){
-    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+    U[mu] = PeekIndex<LorentzIndex>(Umu,mu);
   }
   
   RealD mass=0.1;
diff --git a/tests/Test_dwf_cr_unprec.cc b/tests/Test_dwf_cr_unprec.cc
index 6cd2b32a..2cf7065c 100644
--- a/tests/Test_dwf_cr_unprec.cc
+++ b/tests/Test_dwf_cr_unprec.cc
@@ -43,7 +43,7 @@ int main (int argc, char ** argv)
   std::vector<LatticeColourMatrix> U(4,UGrid);
 
   for(int mu=0;mu<Nd;mu++){
-    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+    U[mu] = PeekIndex<LorentzIndex>(Umu,mu);
   }
 
   ConjugateResidual<LatticeFermion> MCR(1.0e-8,10000);
diff --git a/tests/Test_dwf_hdcr.cc b/tests/Test_dwf_hdcr.cc
index 44b05714..8d1a5fce 100644
--- a/tests/Test_dwf_hdcr.cc
+++ b/tests/Test_dwf_hdcr.cc
@@ -1,11 +1,12 @@
 #include <Grid.h>
 #include <algorithms/iterative/PrecGeneralisedConjugateResidual.h>
+#include <algorithms/iterative/PrecConjugateResidual.h>
 
 using namespace std;
 using namespace Grid;
 using namespace Grid::QCD;
 
-template<class Fobj,class CComplex,int nbasis>
+template<class Fobj,class CComplex,int nbasis, class Matrix>
 class MultiGridPreconditioner : public LinearFunction< Lattice<Fobj> > {
 public:
 
@@ -21,19 +22,22 @@ public:
 
   Aggregates     & _Aggregates;
   CoarseOperator & _CoarseOperator;
+  Matrix         & _Matrix;
   FineOperator   & _FineOperator;
 
   // Constructor
-  MultiGridPreconditioner(Aggregates &Agg, CoarseOperator &Coarse, FineOperator &Fine) 
+  MultiGridPreconditioner(Aggregates &Agg, CoarseOperator &Coarse, FineOperator &Fine,Matrix &FineMatrix) 
     : _Aggregates(Agg),
       _CoarseOperator(Coarse),
-      _FineOperator(Fine)
+      _FineOperator(Fine),
+      _Matrix(FineMatrix)
   {
   }
-
+#if 0
   void operator()(const FineField &in, FineField & out) {
 
     FineField Min(in._grid);
+    FineField tmp(in._grid);
 
     CoarseVector Csrc(_CoarseOperator.Grid());
     CoarseVector Ctmp(_CoarseOperator.Grid());
@@ -44,17 +48,40 @@ public:
     _Aggregates.PromoteFromSubspace(Csrc,out);
     std::cout<<"Completeness: "<<std::sqrt(norm2(out)/norm2(in))<<std::endl;
 
-    ConjugateResidual<FineField>    MCR(1.0e-2,1000);
-    ConjugateGradient<CoarseVector>  CG(1.0e-2,10000);
+    // Build some solvers
+    ConjugateGradient<FineField>    fCG(1.0e-1,1000);
+    ConjugateGradient<CoarseVector>  CG(1.0e-8,100000);
+
+    ////////////////////////////////////////////////////////////////////////
+    // ADEF2: [PTM+Q] in = [1 - Q A] M in + Q in = Min + Q [ in -A Min]
+    ////////////////////////////////////////////////////////////////////////
 
-    // [PTM+Q] in = [1 - Q A] M in + Q in = Min + Q [ in -A Min]
     // Smoothing step, followed by coarse grid correction
+    MdagMLinearOperator<Matrix,FineField> MdagMOp(_Matrix);
 
-    MCR(_FineOperator,in,Min);
-    _FineOperator.Op(Min,out);
-    out = in -out; // out = in - A Min
+    Min=in;
+    std::cout<< " Preconditioner in  " << norm2(in)<<std::endl; 
+    _FineOperator.AdjOp(Min,tmp);
+    std::cout<< " Preconditioner tmp  " << norm2(in)<<std::endl; 
 
-    MdagMLinearOperator<CoarseOperator,CoarseVector>     MdagMOp(_CoarseOperator);
+    fCG(MdagMOp,tmp,out);
+
+    _FineOperator.Op(out,tmp);
+
+    std::cout<< " Preconditioner in  " << norm2(in)<<std::endl; 
+    std::cout<< " Preconditioner out " << norm2(out)<<std::endl; 
+    std::cout<< " Preconditioner Aout" << norm2(tmp)<<std::endl; 
+
+    tmp = tmp - in;
+    
+    std::cout<<"preconditioner thinks residual is "<<std::sqrt(norm2(tmp)/norm2(in))<<std::endl;
+
+    /*
+    //    _FineOperator.Op(Min,out);
+    //    out = in -out; // out = in - A Min
+    out = in;
+
+        MdagMLinearOperator<CoarseOperator,CoarseVector> MdagMOp(_CoarseOperator);
     HermitianLinearOperator<CoarseOperator,CoarseVector> HermOp(_CoarseOperator);
     Csol=zero;
     _Aggregates.ProjectToSubspace  (Csrc,out);
@@ -63,6 +90,121 @@ public:
     _Aggregates.PromoteFromSubspace(Csol,out);
 
     out = Min + out;;
+    */
+
+  }
+#endif
+
+    ////////////////////////////////////////////////////////////////////////
+    // ADEF2: [PTM+Q] in = [1 - Q A] M in + Q in = Min + Q [ in -A Min]
+    // ADEF1: [MP+Q ] in =M [1 - A Q] in + Q in  
+    ////////////////////////////////////////////////////////////////////////
+#if 0
+  void operator()(const FineField &in, FineField & out) {
+
+    CoarseVector Csrc(_CoarseOperator.Grid());
+    CoarseVector Ctmp(_CoarseOperator.Grid());
+    CoarseVector Csol(_CoarseOperator.Grid());
+
+    ConjugateGradient<CoarseVector>  CG(1.0e-10,100000);
+    ConjugateGradient<FineField>    fCG(3.0e-2,1000);
+
+    HermitianLinearOperator<CoarseOperator,CoarseVector>  HermOp(_CoarseOperator);
+    MdagMLinearOperator<CoarseOperator,CoarseVector>     MdagMOp(_CoarseOperator);
+    MdagMLinearOperator<Matrix,FineField>               fMdagMOp(_Matrix);
+
+    FineField tmp(in._grid);
+    FineField res(in._grid);
+    FineField Min(in._grid);
+
+    // Monitor completeness of low mode space
+    _Aggregates.ProjectToSubspace  (Csrc,in);
+    _Aggregates.PromoteFromSubspace(Csrc,out);
+    std::cout<<"Coarse Grid Preconditioner\nCompleteness in: "<<std::sqrt(norm2(out)/norm2(in))<<std::endl;
+
+    // [PTM+Q] in = [1 - Q A] M in + Q in = Min + Q [ in -A Min]
+    _FineOperator.Op(in,tmp);// this is the G5 herm bit
+    fCG(fMdagMOp,tmp,Min);    // solves  MdagM = g5 M g5M
+
+    // Monitor completeness of low mode space
+    _Aggregates.ProjectToSubspace  (Csrc,Min);
+    _Aggregates.PromoteFromSubspace(Csrc,out);
+    std::cout<<"Completeness Min: "<<std::sqrt(norm2(out)/norm2(Min))<<std::endl;
+
+    _FineOperator.Op(Min,tmp);
+    tmp = in - tmp;   // in - A Min
+
+    Csol=zero;
+    _Aggregates.ProjectToSubspace  (Csrc,tmp);
+    HermOp.AdjOp(Csrc,Ctmp);// Normal equations
+    CG(MdagMOp,Ctmp,Csol);
+
+    HermOp.Op(Csol,Ctmp);
+    Ctmp=Ctmp-Csrc;
+    std::cout<<"coarse space true residual "<<std::sqrt(norm2(Ctmp)/norm2(Csrc))<<std::endl;
+    _Aggregates.PromoteFromSubspace(Csol,out);
+
+    _FineOperator.Op(out,res);
+    res=res-tmp;
+    std::cout<<"promoted sol residual "<<std::sqrt(norm2(res)/norm2(tmp))<<std::endl;
+    _Aggregates.ProjectToSubspace  (Csrc,res);
+    std::cout<<"coarse space proj of residual "<<norm2(Csrc)<<std::endl;
+
+    
+    out = out+Min; // additive coarse space correction
+    //    out = Min; // no additive coarse space correction
+
+    _FineOperator.Op(out,tmp);
+    tmp=tmp-in;         // tmp is new residual
+
+    std::cout<< " Preconditioner in  " << norm2(in)<<std::endl; 
+    std::cout<< " Preconditioner out " << norm2(out)<<std::endl; 
+    std::cout<<"preconditioner thinks residual is "<<std::sqrt(norm2(tmp)/norm2(in))<<std::endl;
+
+  }
+#endif
+  // ADEF1: [MP+Q ] in =M [1 - A Q] in + Q in  
+  void operator()(const FineField &in, FineField & out) {
+
+    CoarseVector Csrc(_CoarseOperator.Grid());
+    CoarseVector Ctmp(_CoarseOperator.Grid());
+    CoarseVector Csol(_CoarseOperator.Grid()); Csol=zero;
+
+    ConjugateGradient<CoarseVector>  CG(1.0e-10,100000);
+    ConjugateGradient<FineField>    fCG(1.0e-3,1000);
+
+    HermitianLinearOperator<CoarseOperator,CoarseVector>  HermOp(_CoarseOperator);
+    MdagMLinearOperator<CoarseOperator,CoarseVector>     MdagMOp(_CoarseOperator);
+    MdagMLinearOperator<Matrix,FineField>               fMdagMOp(_Matrix);
+
+    FineField tmp(in._grid);
+    FineField res(in._grid);
+    FineField Qin(in._grid);
+
+    // Monitor completeness of low mode space
+    //    _Aggregates.ProjectToSubspace  (Csrc,in);
+    //    _Aggregates.PromoteFromSubspace(Csrc,out);
+    //    std::cout<<"Coarse Grid Preconditioner\nCompleteness in: "<<std::sqrt(norm2(out)/norm2(in))<<std::endl;
+    
+    _Aggregates.ProjectToSubspace  (Csrc,in);
+    HermOp.AdjOp(Csrc,Ctmp);// Normal equations
+    CG(MdagMOp,Ctmp,Csol);
+    _Aggregates.PromoteFromSubspace(Csol,Qin);
+
+
+    _FineOperator.Op(Qin,tmp);// A Q in
+    tmp = in - tmp;            // in - A Q in
+
+    _FineOperator.Op(tmp,res);// this is the G5 herm bit
+    fCG(fMdagMOp,res,out);    // solves  MdagM = g5 M g5M
+
+    out = out + Qin;
+
+    _FineOperator.Op(out,tmp);
+    tmp=tmp-in;         // tmp is new residual
+
+    std::cout<<"preconditioner thinks residual is "<<std::sqrt(norm2(tmp)/norm2(in))<<std::endl;
+
   }
 
 };
@@ -73,7 +215,7 @@ int main (int argc, char ** argv)
 
   const int Ls=8;
 
-  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
   GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
 
   GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
@@ -82,11 +224,12 @@ int main (int argc, char ** argv)
   ///////////////////////////////////////////////////
   // Construct a coarsened grid; utility for this?
   ///////////////////////////////////////////////////
+  const int block=4;
   std::vector<int> clatt = GridDefaultLatt();
   for(int d=0;d<clatt.size();d++){
-    clatt[d] = clatt[d]/4;
+    clatt[d] = clatt[d]/block;
   }
-  GridCartesian *Coarse4d =  SpaceTimeGrid::makeFourDimGrid(clatt, GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());;
+  GridCartesian *Coarse4d =  SpaceTimeGrid::makeFourDimGrid(clatt, GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());;
   GridCartesian *Coarse5d =  SpaceTimeGrid::makeFiveDimGrid(1,Coarse4d);
 
   std::vector<int> seeds4({1,2,3,4});
@@ -96,7 +239,9 @@ int main (int argc, char ** argv)
   GridParallelRNG          RNG4(UGrid);   RNG4.SeedFixedIntegers(seeds4);
   GridParallelRNG          CRNG(Coarse5d);CRNG.SeedFixedIntegers(cseeds);
 
-  LatticeFermion    src(FGrid); gaussian(RNG5,src);
+  Gamma g5(Gamma::Gamma5);
+
+  LatticeFermion    src(FGrid); gaussian(RNG5,src);// src=src+g5*src;
   LatticeFermion result(FGrid); result=zero;
   LatticeFermion    ref(FGrid); ref=zero;
   LatticeFermion    tmp(FGrid);
@@ -112,7 +257,7 @@ int main (int argc, char ** argv)
   //  SU3::HotConfiguration(RNG4,Umu);
   //  Umu=zero;
 
-  RealD mass=0.04;
+  RealD mass=0.01;
   RealD M5=1.8;
 
   std::cout << "**************************************************"<< std::endl;
@@ -120,7 +265,7 @@ int main (int argc, char ** argv)
   std::cout << "**************************************************"<< std::endl;
   DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
 
-  const int nbasis = 4;
+  const int nbasis = 6;
 
   typedef Aggregation<vSpinColourVector,vTComplex,nbasis>              Subspace;
   typedef CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis>          CoarseOperator;
@@ -132,7 +277,12 @@ int main (int argc, char ** argv)
   MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
   Subspace Aggregates(Coarse5d,FGrid);
   Aggregates.CreateSubspace(RNG5,HermDefOp);
-
+//  for(int i=0;i<nbasis;i++){
+//    result =     Aggregates.subspace[i];
+//    Aggregates.subspace[i]=result+g5*result;
+//  }
+  result=zero;
+  
   std::cout << "**************************************************"<< std::endl;
   std::cout << "Building coarse representation of Indef operator" <<std::endl;
   std::cout << "**************************************************"<< std::endl;
@@ -152,33 +302,41 @@ int main (int argc, char ** argv)
   std::cout << "Solving posdef-CG on coarse space "<< std::endl;
   std::cout << "**************************************************"<< std::endl;
   MdagMLinearOperator<CoarseOperator,CoarseVector> PosdefLdop(LDOp);
-  ConjugateGradient<CoarseVector> CG(1.0e-6,10000);
+  ConjugateGradient<CoarseVector> CG(1.0e-6,100000);
   CG(PosdefLdop,c_src,c_res);
 
-  std::cout << "**************************************************"<< std::endl;
-  std::cout << "Solving indef-MCR on coarse space "<< std::endl;
-  std::cout << "**************************************************"<< std::endl;
-  HermitianLinearOperator<CoarseOperator,CoarseVector> HermIndefLdop(LDOp);
-  ConjugateResidual<CoarseVector> MCR(1.0e-6,10000);
+  //  std::cout << "**************************************************"<< std::endl;
+  //  std::cout << "Solving indef-MCR on coarse space "<< std::endl;
+  //  std::cout << "**************************************************"<< std::endl;
+  //  HermitianLinearOperator<CoarseOperator,CoarseVector> HermIndefLdop(LDOp);
+  //  ConjugateResidual<CoarseVector> MCR(1.0e-6,100000);
   //MCR(HermIndefLdop,c_src,c_res);
 
   std::cout << "**************************************************"<< std::endl;
   std::cout << "Building deflation preconditioner "<< std::endl;
   std::cout << "**************************************************"<< std::endl;
 
-  MultiGridPreconditioner <vSpinColourVector,vTComplex,nbasis> Precon(Aggregates, LDOp,HermIndefOp);
+  MultiGridPreconditioner <vSpinColourVector,vTComplex,nbasis,DomainWallFermion> Precon(Aggregates, LDOp,HermIndefOp,Ddwf);
+  TrivialPrecon<LatticeFermion> simple;
 
   std::cout << "**************************************************"<< std::endl;
-  std::cout << "Building a one level PGCR "<< std::endl;
+  std::cout << "Unprec CG "<< std::endl;
   std::cout << "**************************************************"<< std::endl;
-  TrivialPrecon<LatticeFermion> simple;
-  PrecGeneralisedConjugateResidual<LatticeFermion> GCR(1.0e-6,10000,simple,8,64);
-  GCR(HermIndefOp,src,result);
+  //  TrivialPrecon<LatticeFermion> simple;
+  ConjugateGradient<LatticeFermion> fCG(1.0e-8,100000);
+  fCG(HermDefOp,src,result);
+
+  std::cout << "**************************************************"<< std::endl;
+  std::cout << "Testing GCR on indef matrix "<< std::endl;
+  std::cout << "**************************************************"<< std::endl;
+  //  PrecGeneralisedConjugateResidual<LatticeFermion> UPGCR(1.0e-8,100000,simple,8,128);
+  //  UPGCR(HermIndefOp,src,result);
 
   std::cout << "**************************************************"<< std::endl;
   std::cout << "Building a two level PGCR "<< std::endl;
   std::cout << "**************************************************"<< std::endl;
-  PrecGeneralisedConjugateResidual<LatticeFermion> PGCR(1.0e-6,10000,Precon,8,64);
+  PrecGeneralisedConjugateResidual<LatticeFermion> PGCR(1.0e-8,100000,Precon,8,128);
+  std::cout<<"checking norm src "<<norm2(src)<<std::endl;
   PGCR(HermIndefOp,src,result);
 
   std::cout << "**************************************************"<< std::endl;
diff --git a/tests/Test_gamma.cc b/tests/Test_gamma.cc
index 118362e4..f4f507a7 100644
--- a/tests/Test_gamma.cc
+++ b/tests/Test_gamma.cc
@@ -73,13 +73,13 @@ int main (int argc, char ** argv)
       for(int j=0;j<Ns;j++){
 	if ( abs(result()(i,j)())==0 ) {
 	  std::cout<< " 0";
-	} else if ( abs(result()(i,j)() - ComplexF(0,1))==0){
+	} else if ( abs(result()(i,j)() - Complex(0,1))==0){
 	  std::cout<< " i";
-	} else if ( abs(result()(i,j)() + ComplexF(0,1))==0){
+	} else if ( abs(result()(i,j)() + Complex(0,1))==0){
 	  std::cout<< "-i";
-	} else if ( abs(result()(i,j)() - ComplexF(1,0))==0){
+	} else if ( abs(result()(i,j)() - Complex(1,0))==0){
 	  std::cout<< " 1";
-	} else if ( abs(result()(i,j)() + ComplexF(1,0))==0){
+	} else if ( abs(result()(i,j)() + Complex(1,0))==0){
 	  std::cout<< "-1";
 	}
 	std::cout<< ((j==Ns-1) ? ")" : "," );
diff --git a/tests/Test_main.cc b/tests/Test_main.cc
index 8eecf0c3..6bef8629 100644
--- a/tests/Test_main.cc
+++ b/tests/Test_main.cc
@@ -141,13 +141,13 @@ int main (int argc, char ** argv)
     //    rscalar=real(scalar);
     //    iscalar=imag(scalar);
     //    scalar =cmplx(rscalar,iscalar);
-    pokeIndex<1>(cVec,scalar,1);
+    pokeIndex<2>(cVec,scalar,1);
 
 
     scalar=transpose(scalar);
-    scalar=transposeIndex<1>(scalar);
-    scalar=traceIndex<1>(scalar);
-    scalar=peekIndex<1>(cVec,0);
+    scalar=TransposeIndex<ColourIndex>(scalar);
+    scalar=TraceIndex<SpinIndex>(scalar);
+    scalar=PeekIndex<ColourIndex>(cVec,0);
 
     scalar=trace(scalar);
     scalar=localInnerProduct(cVec,cVec);
@@ -230,8 +230,8 @@ int main (int argc, char ** argv)
     cm = ProjectOnGroup(cm);
     std::cout << cm << "  " << std::endl;
 
-    det = Determinant(cm);
-    std::cout << "determinant: " << det <<  std::endl;
+    //    det = Determinant(cm);
+    //    std::cout << "determinant: " << det <<  std::endl;
 
 
 //    Foo = Foo+scalar; // LatticeColourMatrix+Scalar
@@ -257,16 +257,16 @@ int main (int argc, char ** argv)
       SpinMatrix   s_m;   
       SpinColourMatrix sc_m; 
 
-      s_m = traceIndex<1>(sc_m); // Map to traceColour
-      c_m = traceIndex<2>(sc_m); // map to traceSpin
+      s_m = TensorIndexRecursion<ColourIndex>::traceIndex(sc_m); // Map to traceColour
+      c_m = TensorIndexRecursion<SpinIndex>::traceIndex(sc_m); // map to traceSpin
 
-      c   = traceIndex<2>(s_m); 
-      c   = traceIndex<1>(c_m);
+      c   = TensorIndexRecursion<SpinIndex>::traceIndex(s_m); 
+      c   = TensorIndexRecursion<ColourIndex>::traceIndex(c_m);
       
-      s_m = peekIndex<1>(scm,0,0);
-      c_m = peekIndex<2>(scm,1,2);
+      s_m = TensorIndexRecursion<ColourIndex>::peekIndex(scm,0,0);
+      c_m = TensorIndexRecursion<SpinIndex>::peekIndex(scm,1,2);
       //      c_m = peekSpin<SpinColourMatrix>(scm,1,2);
-      c_m = peekIdiot<SpinColourMatrix>(scm,1,2);
+      //      c_m = peekIdiot<SpinColourMatrix>(scm,1,2);
 
       printf("c. Level %d\n",c_m.TensorLevel);
       printf("c. Level %d\n",c_m().TensorLevel);
@@ -277,7 +277,7 @@ int main (int argc, char ** argv)
       c          = scm()(1,1)(1,2);
       scm()(1,1)(2,1) = c;
 
-      pokeIndex<1> (c_m,c,0,0);
+      pokeIndex<ColourIndex> (c_m,c,0,0);
     }
 
     FooBar = Bar;
@@ -346,7 +346,7 @@ int main (int argc, char ** argv)
 
     LatticeGaugeField U(&Fine);
     //    LatticeColourMatrix Uy = peekLorentz(U,1);
-    LatticeColourMatrix Uy = peekDumKopf(U,1);
+    //    LatticeColourMatrix Uy = peekDumKopf(U,1);
 
     flops = ncall*1.0*volume*(8*Nc*Nc*Nc);
     bytes = ncall*1.0*volume*Nc*Nc    *2*3*sizeof(Grid::Real);
diff --git a/tests/Test_nersc_io.cc b/tests/Test_nersc_io.cc
index b1b9d7ae..f4afb6d0 100644
--- a/tests/Test_nersc_io.cc
+++ b/tests/Test_nersc_io.cc
@@ -10,7 +10,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
 
-  std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplex::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
   std::vector<int> latt_size  ({16,16,16,32});
   std::vector<int> clatt_size  ({4,4,4,8});
@@ -31,7 +31,7 @@ int main (int argc, char ** argv)
   readNerscConfiguration(Umu,header,file);
 
   for(int mu=0;mu<Nd;mu++){
-    U[mu] = peekIndex<3>(Umu,mu);
+    U[mu] = PeekIndex<LorentzIndex>(Umu,mu);
   }
 
   // Painful ; fix syntactical niceness
diff --git a/tests/Test_quenched_update.cc b/tests/Test_quenched_update.cc
index 2d5ea2ee..887a27a8 100644
--- a/tests/Test_quenched_update.cc
+++ b/tests/Test_quenched_update.cc
@@ -63,7 +63,7 @@ int main (int argc, char ** argv)
 	ColourWilsonLoops::Staple(staple,Umu,mu);
 	staple = adj(staple);
 
-	link = peekIndex<LorentzIndex>(Umu,mu);
+	link = PeekIndex<LorentzIndex>(Umu,mu);
 
 	for( int subgroup=0;subgroup<SU3::su2subgroups();subgroup++ ) {
 
diff --git a/tests/Test_wilson_cg_prec.cc b/tests/Test_wilson_cg_prec.cc
index dd8f2821..6365ee32 100644
--- a/tests/Test_wilson_cg_prec.cc
+++ b/tests/Test_wilson_cg_prec.cc
@@ -38,7 +38,7 @@ int main (int argc, char ** argv)
   std::vector<LatticeColourMatrix> U(4,&Grid);
 
   for(int mu=0;mu<Nd;mu++){
-    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+    U[mu] = PeekIndex<LorentzIndex>(Umu,mu);
   }
   
   RealD mass=0.5;
diff --git a/tests/Test_wilson_cg_unprec.cc b/tests/Test_wilson_cg_unprec.cc
index 13996017..4e42af42 100644
--- a/tests/Test_wilson_cg_unprec.cc
+++ b/tests/Test_wilson_cg_unprec.cc
@@ -34,16 +34,10 @@ int main (int argc, char ** argv)
   LatticeFermion result(&Grid); result=zero;
   LatticeGaugeField Umu(&Grid); random(pRNG,Umu);
 
-  std::vector<LatticeColourMatrix> U(4,&Grid);
-
   double volume=1;
   for(int mu=0;mu<Nd;mu++){
     volume=volume*latt_size[mu];
   }  
-
-  for(int mu=0;mu<Nd;mu++){
-    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
-  }
   
   RealD mass=0.5;
   WilsonFermion Dw(Umu,Grid,RBGrid,mass);
diff --git a/tests/Test_wilson_cr_unprec.cc b/tests/Test_wilson_cr_unprec.cc
index 7fbf8c8f..1aa04632 100644
--- a/tests/Test_wilson_cr_unprec.cc
+++ b/tests/Test_wilson_cr_unprec.cc
@@ -42,10 +42,6 @@ int main (int argc, char ** argv)
     volume=volume*latt_size[mu];
   }  
 
-  for(int mu=0;mu<Nd;mu++){
-    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
-  }
-  
   RealD mass=0.5;
   WilsonFermion Dw(Umu,Grid,RBGrid,mass);
 

From 490042f8e15eb0a761f4260d8850f92dd7b595fe Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 30 Jun 2015 15:01:44 +0100
Subject: [PATCH 367/429] big commit fixing nocompiles in defective C++11
 compilers (gcc, icpc). stared getting to near the bleeding edge I guess

---
 benchmarks/Benchmark_dwf.cc    | 2 +-
 benchmarks/Benchmark_wilson.cc | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/benchmarks/Benchmark_dwf.cc b/benchmarks/Benchmark_dwf.cc
index 62869b17..290bf1df 100644
--- a/benchmarks/Benchmark_dwf.cc
+++ b/benchmarks/Benchmark_dwf.cc
@@ -59,7 +59,7 @@ int main (int argc, char ** argv)
   ////////////////////////////////////
   std::vector<LatticeColourMatrix> U(4,FGrid);
   for(int mu=0;mu<Nd;mu++){
-    U[mu] = peekIndex<LorentzIndex>(Umu5d,mu);
+    U[mu] = PeekIndex<LorentzIndex>(Umu5d,mu);
   }
 
   if (1)
diff --git a/benchmarks/Benchmark_wilson.cc b/benchmarks/Benchmark_wilson.cc
index c1050098..7d1bd8df 100644
--- a/benchmarks/Benchmark_wilson.cc
+++ b/benchmarks/Benchmark_wilson.cc
@@ -63,7 +63,7 @@ int main (int argc, char ** argv)
 #endif
 
   for(int mu=0;mu<Nd;mu++){
-    U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+    U[mu] = PeekIndex<LorentzIndex>(Umu,mu);
   }
   
   { // Naive wilson implementation

From 7cfe432ee2dac71797aeaa1590e64690576ee694 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 30 Jun 2015 15:02:27 +0100
Subject: [PATCH 368/429] VPGCR updates

---
 .../PrecGeneralisedConjugateResidual.h        | 31 +++++++++++++++++--
 1 file changed, 29 insertions(+), 2 deletions(-)

diff --git a/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h b/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h
index 77b7df49..148359a8 100644
--- a/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h
+++ b/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h
@@ -5,6 +5,8 @@
 //VPGCR Abe and Zhang, 2005.
 //INTERNATIONAL JOURNAL OF NUMERICAL ANALYSIS AND MODELING
 //Computing and Information Volume 2, Number 2, Pages 147-161
+//NB. Likely not original reference since they are focussing on a preconditioner variant.
+//    but VPGCR was nicely written up in their paper
 ///////////////////////////////////////////////////////////////////////////////////////////////////////
 namespace Grid {
 
@@ -19,7 +21,7 @@ namespace Grid {
     int steps;
     LinearFunction<Field> &Preconditioner;
 
-  PrecGeneralisedConjugateResidual(RealD tol,Integer maxit,LinearFunction<Field> &Prec,int _mmax,int _nstep) : 
+   PrecGeneralisedConjugateResidual(RealD tol,Integer maxit,LinearFunction<Field> &Prec,int _mmax,int _nstep) : 
       Tolerance(tol), 
       MaxIterations(maxit),
       Preconditioner(Prec),
@@ -71,6 +73,8 @@ namespace Grid {
 
       Field r(grid);
       Field z(grid);
+      Field tmp(grid);
+      Field ttmp(grid);
       Field Az(grid);
 
       ////////////////////////////////
@@ -91,6 +95,24 @@ namespace Grid {
       // p = Prec(r)
       /////////////////////
       Preconditioner(r,z);
+
+      std::cout<< " Preconditioner in " << norm2(r)<<std::endl; 
+      std::cout<< " Preconditioner out " << norm2(z)<<std::endl; 
+      
+      Linop.HermOp(z,tmp); 
+
+      std::cout<< " Preconditioner Aout " << norm2(tmp)<<std::endl; 
+      ttmp=tmp;
+      tmp=tmp-r;
+
+      std::cout<< " Preconditioner resid " << std::sqrt(norm2(tmp)/norm2(r))<<std::endl; 
+      /*
+      std::cout<<r<<std::endl;
+      std::cout<<z<<std::endl;
+      std::cout<<ttmp<<std::endl;
+      std::cout<<tmp<<std::endl;
+      */
+
       Linop.HermOpAndNorm(z,Az,zAz,zAAz); 
 
       //p[0],q[0],qq[0] 
@@ -115,14 +137,19 @@ namespace Grid {
 
 	cp = axpy_norm(r,-a,q[peri_k],r);  
 
+	std::cout<< " VPCG_step resid" <<sqrt(cp/rsq)<<std::endl; 
 	if((k==nstep-1)||(cp<rsq)){
 	  return cp;
 	}
 
 	Preconditioner(r,z);// solve Az = r
-
 	Linop.HermOpAndNorm(z,Az,zAz,zAAz);
 
+
+	Linop.HermOp(z,tmp);
+        tmp=tmp-r;
+	std::cout<< " Preconditioner resid" <<sqrt(norm2(tmp)/norm2(r))<<std::endl; 
+
 	q[peri_kp]=Az;
 	p[peri_kp]=z;
 

From 74e397b29cdcc4fb54baaa48dc825ddf2c575764 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 30 Jun 2015 15:03:11 +0100
Subject: [PATCH 369/429] big commit fixing nocompiles in defective C++11
 compilers (gcc, icpc). stared getting to near the bleeding edge I guess

---
 lib/qcd/QCD.h                               |   6 +-
 lib/qcd/action/fermion/WilsonFermion.cc     |  12 +-
 lib/qcd/action/fermion/WilsonFermion5D.cc   |  28 +--
 lib/qcd/action/fermion/WilsonFermion5D.h    |  16 +-
 lib/qcd/action/fermion/WilsonKernels.cc     |   6 +-
 lib/qcd/action/fermion/WilsonKernels.h      |  49 ++--
 lib/qcd/action/fermion/WilsonKernelsHand.cc |   4 +-
 lib/qcd/spin/Dirac.h                        |  20 +-
 lib/qcd/spin/TwoSpinor.h                    | 257 ++++++++++----------
 lib/qcd/utils/WilsonLoops.h                 |   4 +-
 10 files changed, 204 insertions(+), 198 deletions(-)

diff --git a/lib/qcd/QCD.h b/lib/qcd/QCD.h
index b3c6d138..0e841d8e 100644
--- a/lib/qcd/QCD.h
+++ b/lib/qcd/QCD.h
@@ -24,9 +24,9 @@ namespace QCD {
     // QCD iMatrix types
     // Index conventions:                            Lorentz x Spin x Colour
     //////////////////////////////////////////////////////////////////////////////
-    static const int ColourIndex = 1;
-    static const int SpinIndex   = 2;
-    static const int LorentzIndex= 3;
+    static const int ColourIndex = 2;
+    static const int SpinIndex   = 1;
+    static const int LorentzIndex= 0;
     
 
     // ChrisK very keen to add extra space for Gparity doubling.
diff --git a/lib/qcd/action/fermion/WilsonFermion.cc b/lib/qcd/action/fermion/WilsonFermion.cc
index 13f16c5a..067d5113 100644
--- a/lib/qcd/action/fermion/WilsonFermion.cc
+++ b/lib/qcd/action/fermion/WilsonFermion.cc
@@ -35,7 +35,7 @@ void WilsonFermion::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGauge
   conformable(Umu._grid,GaugeGrid());
   LatticeColourMatrix U(GaugeGrid());
   for(int mu=0;mu<Nd;mu++){
-    U = peekIndex<LorentzIndex>(Umu,mu);
+    U = PeekIndex<LorentzIndex>(Umu,mu);
     pokeIndex<LorentzIndex>(Uds,U,mu);
     U = adj(Cshift(U,mu,-1));
     pokeIndex<LorentzIndex>(Uds,U,mu+4);
@@ -109,7 +109,7 @@ void WilsonFermion::DhopDir(const LatticeFermion &in, LatticeFermion &out,int di
 
 PARALLEL_FOR_LOOP
   for(int sss=0;sss<in._grid->oSites();sss++){
-    DiracOpt::DhopDir(Stencil,Umu,comm_buf,sss,sss,in,out,dirdisp);
+    DiracOptDhopDir(Stencil,Umu,comm_buf,sss,sss,in,out,dirdisp);
   }
 
 };
@@ -125,24 +125,24 @@ void WilsonFermion::DhopInternal(CartesianStencil & st,LatticeDoubledGaugeField
     if( HandOptDslash ) {
 PARALLEL_FOR_LOOP
       for(int sss=0;sss<in._grid->oSites();sss++){
-        DiracOptHand::DhopSiteDag(st,U,comm_buf,sss,sss,in,out);
+        DiracOptHandDhopSiteDag(st,U,comm_buf,sss,sss,in,out);
       }
     } else { 
 PARALLEL_FOR_LOOP
       for(int sss=0;sss<in._grid->oSites();sss++){
-        DiracOpt::DhopSiteDag(st,U,comm_buf,sss,sss,in,out);
+        DiracOptDhopSiteDag(st,U,comm_buf,sss,sss,in,out);
       }
     }
   } else {
     if( HandOptDslash ) {
 PARALLEL_FOR_LOOP
       for(int sss=0;sss<in._grid->oSites();sss++){
-        DiracOptHand::DhopSite(st,U,comm_buf,sss,sss,in,out);
+        DiracOptHandDhopSite(st,U,comm_buf,sss,sss,in,out);
       }
     } else { 
 PARALLEL_FOR_LOOP
       for(int sss=0;sss<in._grid->oSites();sss++){
-        DiracOpt::DhopSite(st,U,comm_buf,sss,sss,in,out);
+        DiracOptDhopSite(st,U,comm_buf,sss,sss,in,out);
       }
     }
   }
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.cc b/lib/qcd/action/fermion/WilsonFermion5D.cc
index ed57d37d..6d6097cf 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.cc
+++ b/lib/qcd/action/fermion/WilsonFermion5D.cc
@@ -77,7 +77,7 @@ void WilsonFermion5D::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGau
   conformable(Umu._grid,GaugeGrid());
   LatticeColourMatrix U(GaugeGrid());
   for(int mu=0;mu<Nd;mu++){
-    U = peekIndex<LorentzIndex>(Umu,mu);
+    U = PeekIndex<LorentzIndex>(Umu,mu);
     pokeIndex<LorentzIndex>(Uds,U,mu);
     U = adj(Cshift(U,mu,-1));
     pokeIndex<LorentzIndex>(Uds,U,mu+4);
@@ -105,7 +105,7 @@ void WilsonFermion5D::DhopDir(const LatticeFermion &in, LatticeFermion &out,int
     for(int s=0;s<Ls;s++){
       int sU=ss;
       int sF = s+Ls*sU; 
-      DiracOpt::DhopDir(Stencil,Umu,comm_buf,sF,sU,in,out,dirdisp);
+      DiracOptDhopDir(Stencil,Umu,comm_buf,sF,sU,in,out,dirdisp);
     }
   }
 };
@@ -127,49 +127,47 @@ void WilsonFermion5D::DhopInternal(CartesianStencil & st, LebesgueOrder &lo,
   // - 8 linear access unit stride streams per thread for Fermion for hw prefetchable.
   if ( dag == DaggerYes ) {
     if( HandOptDslash ) {
-#pragma parallel for
+PARALLEL_FOR_LOOP
       for(int ss=0;ss<U._grid->oSites();ss++){
-	{
-	  for(int s=0;s<Ls;s++){
-	    int sU=ss;
-	    int sF = s+Ls*sU;
-	    DiracOptHand::DhopSiteDag(st,U,comm_buf,sF,sU,in,out);
+	for(int s=0;s<Ls;s++){
+	  int sU=ss;
+	  int sF = s+Ls*sU;
+	  DiracOptHandDhopSiteDag(st,U,comm_buf,sF,sU,in,out);
 	  }
-	}
       }
     } else { 
-#pragma parallel for
+PARALLEL_FOR_LOOP
       for(int ss=0;ss<U._grid->oSites();ss++){
 	{
 	  int sd;
 	  for(sd=0;sd<Ls;sd++){
 	    int sU=ss;
 	    int sF = sd+Ls*sU;
-	    DiracOpt::DhopSiteDag(st,U,comm_buf,sF,sU,in,out);
+	    DiracOptDhopSiteDag(st,U,comm_buf,sF,sU,in,out);
 	  }
 	}
       }
     }
   } else {
     if( HandOptDslash ) {
-#pragma parallel for
+PARALLEL_FOR_LOOP
       for(int ss=0;ss<U._grid->oSites();ss++){
 	for(int s=0;s<Ls;s++){
 	  //	  int sU=lo.Reorder(ss);
 	  int sU=ss;
 	  int sF = s+Ls*sU;
-	  DiracOptHand::DhopSite(st,U,comm_buf,sF,sU,in,out);
+	  DiracOptHandDhopSite(st,U,comm_buf,sF,sU,in,out);
 	}
       }
 
     } else { 
-#pragma parallel for
+PARALLEL_FOR_LOOP
       for(int ss=0;ss<U._grid->oSites();ss++){
 	for(int s=0;s<Ls;s++){
 	  //	  int sU=lo.Reorder(ss);
 	  int sU=ss;
 	  int sF = s+Ls*sU; 
-	  DiracOpt::DhopSite(st,U,comm_buf,sF,sU,in,out);
+	  DiracOptDhopSite(st,U,comm_buf,sF,sU,in,out);
 	}
       }
     }
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.h b/lib/qcd/action/fermion/WilsonFermion5D.h
index 10138991..e345d2a4 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.h
+++ b/lib/qcd/action/fermion/WilsonFermion5D.h
@@ -39,16 +39,16 @@ namespace Grid {
       GridBase *FermionRedBlackGrid(void)    { return _FiveDimRedBlackGrid;}
 
       // full checkerboard operations; leave unimplemented as abstract for now
-      //virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out)=0;
-      //virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out)=0;
+      virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out){assert(0); return 0.0;};
+      virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out){assert(0); return 0.0;};
 
       // half checkerboard operations; leave unimplemented as abstract for now
-      //      virtual void   Meooe       (const LatticeFermion &in, LatticeFermion &out)=0;
-      //      virtual void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out)=0;
-      //      virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out)=0;
-      //      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out)=0;
-      //      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out)=0;
-      //      virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out)=0;
+      virtual void   Meooe       (const LatticeFermion &in, LatticeFermion &out){assert(0);};
+      virtual void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out){assert(0);};
+      virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out){assert(0);};
+      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out){assert(0);};
+      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out){assert(0);};
+      virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out){assert(0);};
 
       // Implement hopping term non-hermitian hopping term; half cb or both
       // Implement s-diagonal DW
diff --git a/lib/qcd/action/fermion/WilsonKernels.cc b/lib/qcd/action/fermion/WilsonKernels.cc
index 518b8004..4054280a 100644
--- a/lib/qcd/action/fermion/WilsonKernels.cc
+++ b/lib/qcd/action/fermion/WilsonKernels.cc
@@ -3,7 +3,7 @@
 namespace Grid {
 namespace QCD {
 
-void DiracOpt::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
+void DiracOptDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
 			std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
 			int sF,int sU,const LatticeFermion &in, LatticeFermion &out)
 {
@@ -149,7 +149,7 @@ void DiracOpt::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     vstream(out._odata[ss],result*(-0.5));
 }
 
-void DiracOpt::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
+void DiracOptDhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
 			   std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
 			   int sF,int sU,const LatticeFermion &in, LatticeFermion &out)
 {
@@ -295,7 +295,7 @@ void DiracOpt::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
     vstream(out._odata[ss],result*(-0.5));
 }
 
-void DiracOpt::DhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
+void DiracOptDhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
 			std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
 		       int sF,int sU,const LatticeFermion &in, LatticeFermion &out,int dirdisp)
 {
diff --git a/lib/qcd/action/fermion/WilsonKernels.h b/lib/qcd/action/fermion/WilsonKernels.h
index db384ddf..ec807af5 100644
--- a/lib/qcd/action/fermion/WilsonKernels.h
+++ b/lib/qcd/action/fermion/WilsonKernels.h
@@ -10,35 +10,40 @@ namespace Grid {
     ////////////////////////////////////////////////////////////////////////////////////////////////////////////////
     
     // Generic version works for any Nc and with extra flavour indices
-    class DiracOpt {
-    public:
+    //    namespace DiracOpt {
+
       // These ones will need to be package intelligently. WilsonType base class
       // for use by DWF etc..
-      static void DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
+      void DiracOptDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
+			    std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+			    int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
+      void DiracOptDhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
+			       std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+			       int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
+      void DiracOptDhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
 			   std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			   int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
-      static void DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
-			      std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			      int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
-      static void DhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
-			   std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			  int sF,int sU,const LatticeFermion &in, LatticeFermion &out,int dirdisp);
+			   int sF,int sU,const LatticeFermion &in, LatticeFermion &out,int dirdisp);
+      
+      //  };
 
-    };
-
-    // Hand unrolled for Nc=3, one flavour
-    class DiracOptHand {
-    public:
+      // Hand unrolled for Nc=3, one flavour
+      //    namespace DiracOptHand {
       // These ones will need to be package intelligently. WilsonType base class
       // for use by DWF etc..
-      static void DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
-			   std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			   int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
-      static void DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
-			      std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			      int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
 
-    };
+      void DiracOptHandDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
+				std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+				int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
+      void DiracOptHandDhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
+				   std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+				   int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
+
+      //    };
+  
+
+    void DiracOptHandDhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
+				 std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+				 int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
 
   }
 }
diff --git a/lib/qcd/action/fermion/WilsonKernelsHand.cc b/lib/qcd/action/fermion/WilsonKernelsHand.cc
index 019b668b..8c31d65b 100644
--- a/lib/qcd/action/fermion/WilsonKernelsHand.cc
+++ b/lib/qcd/action/fermion/WilsonKernelsHand.cc
@@ -280,7 +280,7 @@
 namespace Grid {
 namespace QCD {
 
-void DiracOptHand::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
+void DiracOptHandDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
 			    std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
 			    int sF,int sU,const LatticeFermion &in, LatticeFermion &out)
 {
@@ -530,7 +530,7 @@ void DiracOptHand::DhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
   }
 }
 
-void DiracOptHand::DhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
+void DiracOptHandDhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
 			       std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
 			       int ss,int sU,const LatticeFermion &in, LatticeFermion &out)
 {
diff --git a/lib/qcd/spin/Dirac.h b/lib/qcd/spin/Dirac.h
index 2623e683..3208d0b5 100644
--- a/lib/qcd/spin/Dirac.h
+++ b/lib/qcd/spin/Dirac.h
@@ -4,6 +4,8 @@ namespace Grid{
 
 namespace QCD {
 
+  const int SpinorIndex = 2;
+
   class Gamma {
 
   public:
@@ -331,7 +333,7 @@ namespace QCD {
 
     //left multiply
     template<class vtype> inline auto operator * ( const Gamma &G,const iScalar<vtype> &arg) ->
-      typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type 
+      typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type 
 
 	{
 	  iScalar<vtype> ret;
@@ -339,14 +341,14 @@ namespace QCD {
 	  return ret;
 	}
     template<class vtype,int N> inline auto operator * ( const Gamma &G,const iVector<vtype,N> &arg) ->
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type 
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type 
 	{
 	  iVector<vtype,N> ret;
 	  ret._internal=G*arg._internal;
 	  return ret;
 	}
     template<class vtype,int N> inline auto operator * ( const Gamma &G,const iMatrix<vtype,N> &arg) ->
-      typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type 
+      typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type 
 	{
 	  iMatrix<vtype,N> ret;
 	  ret._internal=G*arg._internal;
@@ -356,7 +358,7 @@ namespace QCD {
 
     //right multiply
     template<class vtype> inline auto operator * (const iScalar<vtype> &arg, const Gamma &G) ->
-      typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type 
+      typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type 
 
 	{
 	  iScalar<vtype> ret;
@@ -364,14 +366,14 @@ namespace QCD {
 	  return ret;
 	}
     template<class vtype,int N> inline auto operator * (const iVector<vtype,N> &arg, const Gamma &G) ->
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type 
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type 
 	{
 	  iVector<vtype,N> ret;
 	  ret._internal=arg._internal*G;
 	  return ret;
 	}
     template<class vtype,int N> inline auto operator * (const iMatrix<vtype,N> &arg, const Gamma &G) ->
-      typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type 
+      typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type 
 	{
 	  iMatrix<vtype,N> ret;
 	  ret._internal=arg._internal*G;
@@ -382,7 +384,7 @@ namespace QCD {
     // When we hit the spin index this matches and we stop
     ////////////////////////////////////////////////////////
     template<class vtype> inline auto operator * ( const Gamma &G,const iMatrix<vtype,Ns> &arg) ->
-      typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,Ns>,SpinIndex>::value,iMatrix<vtype,Ns> >::type 
+      typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,Ns>,SpinorIndex>::value,iMatrix<vtype,Ns> >::type 
       {
 	iMatrix<vtype,Ns> ret;
 	switch (G._g) {
@@ -430,7 +432,7 @@ namespace QCD {
       }
     // Could have used type trait for Matrix/vector and then an enable if to share code
     template<class vtype> inline auto operator * ( const Gamma &G,const iVector<vtype,Ns> &arg) ->
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type 
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type 
       {
 	iVector<vtype,Ns> ret;
 	switch (G._g) {
@@ -478,7 +480,7 @@ namespace QCD {
       }
 
     template<class vtype> inline auto operator * (const iMatrix<vtype,Ns> &arg, const Gamma &G) ->
-      typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,Ns>,SpinIndex>::value,iMatrix<vtype,Ns> >::type 
+      typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,Ns>,SpinorIndex>::value,iMatrix<vtype,Ns> >::type 
       {
 	iMatrix<vtype,Ns> ret;
 	switch (G._g) {
diff --git a/lib/qcd/spin/TwoSpinor.h b/lib/qcd/spin/TwoSpinor.h
index 97de13cf..919f0f76 100644
--- a/lib/qcd/spin/TwoSpinor.h
+++ b/lib/qcd/spin/TwoSpinor.h
@@ -42,17 +42,18 @@ namespace QCD {
    * -i 0  0  0
    */
 
+
   template<class vtype> strong_inline void 
     spProjXp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
       // To fail is not to err (Cryptic clue: suggest to Google SFINAE ;) )
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)+timesI(fspin(3));
       hspin(1)=fspin(1)+timesI(fspin(2));
     }
   template<class vtype> strong_inline void spProjXm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)-timesI(fspin(3));
       hspin(1)=fspin(1)-timesI(fspin(2));
     }
@@ -63,13 +64,13 @@ namespace QCD {
       // -1 0  0  0
   template<class vtype> strong_inline void spProjYp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)-fspin(3);
       hspin(1)=fspin(1)+fspin(2);
     }
   template<class vtype> strong_inline void spProjYm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)+fspin(3);
       hspin(1)=fspin(1)-fspin(2);
     }
@@ -81,13 +82,13 @@ namespace QCD {
 	     */
   template<class vtype> strong_inline void spProjZp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)+timesI(fspin(2));
       hspin(1)=fspin(1)-timesI(fspin(3));
     }
   template<class vtype> strong_inline void spProjZm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)-timesI(fspin(2));
       hspin(1)=fspin(1)+timesI(fspin(3));
     }
@@ -99,13 +100,13 @@ namespace QCD {
 	     */
   template<class vtype> strong_inline void spProjTp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)+fspin(2);
       hspin(1)=fspin(1)+fspin(3);
     }
   template<class vtype> strong_inline void spProjTm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)-fspin(2);
       hspin(1)=fspin(1)-fspin(3);
     }
@@ -118,13 +119,13 @@ namespace QCD {
 
   template<class vtype> strong_inline void spProj5p (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0);
       hspin(1)=fspin(1);
     }
   template<class vtype> strong_inline void spProj5m (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(2);
       hspin(1)=fspin(3);
     }
@@ -132,7 +133,7 @@ namespace QCD {
   //  template<class vtype> strong_inline void fspProj5p (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
   template<class vtype> strong_inline void spProj5p (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       rfspin(0)=fspin(0);
       rfspin(1)=fspin(1);
       rfspin(2)=zero;
@@ -141,7 +142,7 @@ namespace QCD {
   //  template<class vtype> strong_inline void fspProj5m (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
   template<class vtype> strong_inline void spProj5m (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       rfspin(0)=zero;
       rfspin(1)=zero;
       rfspin(2)=fspin(2);
@@ -159,7 +160,7 @@ namespace QCD {
    */
   template<class vtype> strong_inline void spReconXp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
       fspin(1)=hspin(1);
       fspin(2)=timesMinusI(hspin(1));
@@ -167,7 +168,7 @@ namespace QCD {
     }
   template<class vtype> strong_inline void spReconXm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
       fspin(1)=hspin(1);
       fspin(2)=timesI(hspin(1));
@@ -175,7 +176,7 @@ namespace QCD {
     }
   template<class vtype> strong_inline void accumReconXp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
       fspin(1)+=hspin(1);
       fspin(2)-=timesI(hspin(1));
@@ -183,7 +184,7 @@ namespace QCD {
     }
   template<class vtype> strong_inline void accumReconXm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
       fspin(1)+=hspin(1);
       fspin(2)+=timesI(hspin(1));
@@ -197,7 +198,7 @@ namespace QCD {
 
   template<class vtype> strong_inline void spReconYp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
       fspin(1)=hspin(1);
       fspin(2)= hspin(1);
@@ -205,7 +206,7 @@ namespace QCD {
     }
   template<class vtype> strong_inline void spReconYm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
       fspin(1)=hspin(1);
       fspin(2)=-hspin(1);
@@ -213,7 +214,7 @@ namespace QCD {
     }
   template<class vtype> strong_inline void accumReconYp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
       fspin(1)+=hspin(1);
       fspin(2)+=hspin(1);
@@ -221,7 +222,7 @@ namespace QCD {
     }
   template<class vtype> strong_inline void accumReconYm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
       fspin(1)+=hspin(1);
       fspin(2)-=hspin(1);
@@ -236,7 +237,7 @@ namespace QCD {
 	     */
   template<class vtype> strong_inline void spReconZp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
       fspin(1)=hspin(1);
       fspin(2)=timesMinusI(hspin(0));
@@ -244,7 +245,7 @@ namespace QCD {
     }
   template<class vtype> strong_inline void spReconZm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
       fspin(1)=hspin(1);
       fspin(2)=     timesI(hspin(0));
@@ -252,7 +253,7 @@ namespace QCD {
     }
   template<class vtype> strong_inline void accumReconZp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
       fspin(1)+=hspin(1);
       fspin(2)-=timesI(hspin(0));
@@ -260,7 +261,7 @@ namespace QCD {
     }
   template<class vtype> strong_inline void accumReconZm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
       fspin(1)+=hspin(1);
       fspin(2)+=timesI(hspin(0));
@@ -274,7 +275,7 @@ namespace QCD {
 	     */
   template<class vtype> strong_inline void spReconTp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
       fspin(1)=hspin(1);
       fspin(2)=hspin(0);
@@ -282,7 +283,7 @@ namespace QCD {
     }
   template<class vtype> strong_inline void spReconTm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
       fspin(1)=hspin(1);
       fspin(2)=-hspin(0);
@@ -290,7 +291,7 @@ namespace QCD {
     }
   template<class vtype> strong_inline void accumReconTp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
       fspin(1)+=hspin(1);
       fspin(2)+=hspin(0);
@@ -298,7 +299,7 @@ namespace QCD {
     }
   template<class vtype> strong_inline void accumReconTm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
       fspin(1)+=hspin(1);
       fspin(2)-=hspin(0);
@@ -312,7 +313,7 @@ namespace QCD {
 	     */
   template<class vtype> strong_inline void spRecon5p (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0)+hspin(0); // add is lower latency than mul
       fspin(1)=hspin(1)+hspin(1); // probably no measurable diffence though
       fspin(2)=zero;
@@ -320,7 +321,7 @@ namespace QCD {
     }
   template<class vtype> strong_inline void spRecon5m (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=zero;
       fspin(1)=zero;
       fspin(2)=hspin(0)+hspin(0);
@@ -328,13 +329,13 @@ namespace QCD {
     }
   template<class vtype> strong_inline void accumRecon5p (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0)+hspin(0);
       fspin(1)+=hspin(1)+hspin(1);
     }
   template<class vtype> strong_inline void accumRecon5m (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(2)+=hspin(0)+hspin(0);
       fspin(3)+=hspin(1)+hspin(1);
     }
@@ -348,19 +349,19 @@ namespace QCD {
   //////////
   template<class rtype,class vtype> strong_inline void spProjXp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjXp(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spProjXp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjXp(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spProjXp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProjXp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -370,19 +371,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spReconXp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconXp(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spReconXp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconXp(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spReconXp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spReconXp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -391,19 +392,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void accumReconXp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconXp(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void accumReconXp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconXp(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void accumReconXp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumReconXp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -417,19 +418,19 @@ namespace QCD {
   ////////
   template<class rtype,class vtype> strong_inline void spProjXm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjXm(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spProjXm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjXm(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spProjXm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProjXm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -439,19 +440,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spReconXm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconXm(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spReconXm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconXm(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spReconXm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spReconXm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -460,19 +461,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void accumReconXm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconXm(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void accumReconXm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconXm(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void accumReconXm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumReconXm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -486,19 +487,19 @@ namespace QCD {
   ////////
   template<class rtype,class vtype> strong_inline void spProjYp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjYp(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spProjYp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjYp(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spProjYp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProjYp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -508,19 +509,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spReconYp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconYp(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spReconYp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconYp(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spReconYp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spReconYp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -529,19 +530,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void accumReconYp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconYp(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void accumReconYp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconYp(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void accumReconYp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumReconYp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -554,19 +555,19 @@ namespace QCD {
   ////////
   template<class rtype,class vtype> strong_inline void spProjYm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjYm(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spProjYm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjYm(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spProjYm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProjYm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -576,19 +577,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spReconYm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconYm(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spReconYm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconYm(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spReconYm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spReconYm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -597,19 +598,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void accumReconYm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconYm(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void accumReconYm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconYm(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void accumReconYm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumReconYm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -622,19 +623,19 @@ namespace QCD {
   ////////
   template<class rtype,class vtype> strong_inline void spProjZp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjZp(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spProjZp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjZp(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spProjZp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProjZp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -644,19 +645,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spReconZp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconZp(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spReconZp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconZp(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spReconZp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spReconZp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -665,19 +666,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void accumReconZp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconZp(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void accumReconZp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconZp(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void accumReconZp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumReconZp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -690,19 +691,19 @@ namespace QCD {
   ////////
   template<class rtype,class vtype> strong_inline void spProjZm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjZm(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spProjZm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjZm(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spProjZm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProjZm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -712,19 +713,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spReconZm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconZm(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spReconZm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconZm(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spReconZm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spReconZm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -733,19 +734,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void accumReconZm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconZm(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void accumReconZm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconZm(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void accumReconZm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumReconZm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -758,19 +759,19 @@ namespace QCD {
   ////////
   template<class rtype,class vtype> strong_inline void spProjTp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjTp(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spProjTp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjTp(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spProjTp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProjTp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -780,19 +781,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spReconTp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconTp(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spReconTp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconTp(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spReconTp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spReconTp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -801,19 +802,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void accumReconTp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconTp(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void accumReconTp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconTp(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void accumReconTp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumReconTp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -826,19 +827,19 @@ namespace QCD {
   ////////
   template<class rtype,class vtype> strong_inline void spProjTm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjTm(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spProjTm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjTm(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spProjTm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProjTm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -848,19 +849,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spReconTm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconTm(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spReconTm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconTm(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spReconTm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spReconTm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -869,19 +870,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void accumReconTm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconTm(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void accumReconTm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconTm(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void accumReconTm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumReconTm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -894,19 +895,19 @@ namespace QCD {
   ////////
   template<class rtype,class vtype> strong_inline void spProj5p (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProj5p(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spProj5p (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProj5p(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spProj5p (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProj5p(hspin._internal[i][j],fspin._internal[i][j]);
@@ -915,19 +916,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spRecon5p (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spRecon5p(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spRecon5p (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spRecon5p(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spRecon5p (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spRecon5p(hspin._internal[i][j],fspin._internal[i][j]);
@@ -936,19 +937,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void accumRecon5p (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumRecon5p(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void accumRecon5p (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumRecon5p(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void accumRecon5p (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumRecon5p(hspin._internal[i][j],fspin._internal[i][j]);
@@ -959,13 +960,13 @@ namespace QCD {
   //  template<class vtype> strong_inline void fspProj5p (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
   template<class vtype> strong_inline void spProj5p (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProj5p(hspin._internal,fspin._internal);
   }
   //  template<class vtype,int N> strong_inline void fspProj5p (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
   template<class vtype,int N> strong_inline void spProj5p (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProj5p(hspin._internal[i],fspin._internal[i]);
     }
@@ -973,7 +974,7 @@ namespace QCD {
   //  template<class vtype,int N> strong_inline void fspProj5p (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
   template<class vtype,int N> strong_inline void spProj5p (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProj5p(hspin._internal[i][j],fspin._internal[i][j]);
@@ -986,19 +987,19 @@ namespace QCD {
   ////////
   template<class rtype,class vtype> strong_inline void spProj5m (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProj5m(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spProj5m (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProj5m(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spProj5m (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProj5m(hspin._internal[i][j],fspin._internal[i][j]);
@@ -1007,19 +1008,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spRecon5m (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spRecon5m(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void spRecon5m (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spRecon5m(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void spRecon5m (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spRecon5m(hspin._internal[i][j],fspin._internal[i][j]);
@@ -1028,19 +1029,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void accumRecon5m (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumRecon5m(hspin._internal,fspin._internal);
   }
   template<class rtype,class vtype,int N> strong_inline void accumRecon5m (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumRecon5m(hspin._internal[i],fspin._internal[i]);
     }
   }
   template<class rtype,class vtype,int N> strong_inline void accumRecon5m (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumRecon5m(hspin._internal[i][j],fspin._internal[i][j]);
@@ -1052,13 +1053,13 @@ namespace QCD {
   //  template<class vtype> strong_inline void fspProj5m (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
   template<class vtype> strong_inline void spProj5m (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinIndex>::notvalue,iScalar<vtype> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProj5m(hspin._internal,fspin._internal);
   }
   //  template<class vtype,int N> strong_inline void fspProj5m (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
   template<class vtype,int N> strong_inline void spProj5m (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProj5m(hspin._internal[i],fspin._internal[i]);
     }
@@ -1066,7 +1067,7 @@ namespace QCD {
   //  template<class vtype,int N> strong_inline void fspProj5m (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
   template<class vtype,int N> strong_inline void spProj5m (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProj5m(hspin._internal[i][j],fspin._internal[i][j]);
diff --git a/lib/qcd/utils/WilsonLoops.h b/lib/qcd/utils/WilsonLoops.h
index e3885432..9b8e8724 100644
--- a/lib/qcd/utils/WilsonLoops.h
+++ b/lib/qcd/utils/WilsonLoops.h
@@ -45,7 +45,7 @@ public:
     std::vector<GaugeMat> U(4,Umu._grid);
 
     for(int mu=0;mu<Nd;mu++){
-      U[mu] = peekIndex<LorentzIndex>(Umu,mu);
+      U[mu] = PeekIndex<LorentzIndex>(Umu,mu);
     }
 
     LatticeComplex Plaq(Umu._grid);
@@ -78,7 +78,7 @@ public:
 
     std::vector<GaugeMat> U(4,grid);
     for(int d=0;d<Nd;d++){
-      U[d] = peekIndex<LorentzIndex>(Umu,d);
+      U[d] = PeekIndex<LorentzIndex>(Umu,d);
     }
 
     staple = zero;

From f41c7dffef6ddeb643679afe9c383125b987923f Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 30 Jun 2015 15:17:27 +0100
Subject: [PATCH 370/429] Big commit fixing nocompiles in defective C++11
 compilers (gcc, icpc). stared getting to near the bleeding edge I guess

---
 lib/Simd.h                                |   1 +
 lib/algorithms/CoarsenedMatrix.h          |   7 +-
 lib/lattice/Lattice_peekpoke.h            |  26 +-
 lib/lattice/Lattice_trace.h               |   2 +-
 lib/lattice/Lattice_transpose.h           |   2 +-
 lib/qcd/action/fermion/WilsonFermion.cc   |   4 +-
 lib/qcd/action/fermion/WilsonFermion5D.cc |   4 +-
 lib/qcd/utils/SUn.h                       |   6 +-
 lib/tensors/Tensor_class.h                |  18 +-
 lib/tensors/Tensor_peek.h                 |   4 +-
 lib/tensors/Tensor_poke.h                 |   7 +-
 lib/tensors/Tensor_trace.h                | 361 +++++++++++++++++++---
 tests/Test_dwf_even_odd.cc                |   2 +-
 tests/Test_main.cc                        |   2 +-
 tests/Test_quenched_update.cc             |   2 +-
 tests/Test_wilson_even_odd.cc             |   2 +-
 16 files changed, 349 insertions(+), 101 deletions(-)

diff --git a/lib/Simd.h b/lib/Simd.h
index 7da0d7ad..474e5dd3 100644
--- a/lib/Simd.h
+++ b/lib/Simd.h
@@ -17,6 +17,7 @@ namespace Grid {
 
   typedef  float  RealF;
   typedef  double RealD;
+#define GRID_DEFAULT_PRECISION_DOUBLE
 #ifdef GRID_DEFAULT_PRECISION_DOUBLE
   typedef RealD   Real;
 #else
diff --git a/lib/algorithms/CoarsenedMatrix.h b/lib/algorithms/CoarsenedMatrix.h
index 9ea77b84..e257294c 100644
--- a/lib/algorithms/CoarsenedMatrix.h
+++ b/lib/algorithms/CoarsenedMatrix.h
@@ -121,7 +121,7 @@ namespace Grid {
 
       RealD scale;
 
-      ConjugateGradient<FineField> CG(1.0e-3,10000);
+      ConjugateGradient<FineField> CG(1.0e-4,10000);
       FineField noise(FineGrid);
       FineField Mn(FineGrid);
 
@@ -131,7 +131,8 @@ namespace Grid {
 	scale = std::pow(norm2(noise),-0.5); 
 	noise=noise*scale;
 
-	hermop.Op(noise,Mn); std::cout << "Noise    "<<b<<" nMdagMn "<<norm2(Mn)<< " "<< norm2(noise)<<std::endl;
+	hermop.Op(noise,Mn); std::cout << "noise   ["<<b<<"] <n|MdagM|n> "<<norm2(Mn)<<std::endl;
+
 	for(int i=0;i<2;i++){
 
 	  CG(hermop,noise,subspace[b]);
@@ -142,7 +143,7 @@ namespace Grid {
 
 	}
 
-	hermop.Op(noise,Mn); std::cout << "filtered    "<<b<<" <s|MdagM|s> "<<norm2(Mn)<< " "<< norm2(noise)<<std::endl;
+	hermop.Op(noise,Mn); std::cout << "filtered["<<b<<"] <f|MdagM|f> "<<norm2(Mn)<<std::endl;
 	subspace[b] = noise;
       }
 
diff --git a/lib/lattice/Lattice_peekpoke.h b/lib/lattice/Lattice_peekpoke.h
index 5455fdd6..e60618c5 100644
--- a/lib/lattice/Lattice_peekpoke.h
+++ b/lib/lattice/Lattice_peekpoke.h
@@ -10,18 +10,8 @@ namespace Grid {
     ////////////////////////////////////////////////////////////////////////////////////////////////////
     // Peek internal indices of a Lattice object
     ////////////////////////////////////////////////////////////////////////////////////////////////////
-  template<int Index,class vobj>
-     auto peekIndex(const Lattice<vobj> &lhs) -> Lattice<decltype(peekIndex<Index>(lhs._odata[0]))>
-    {
-      Lattice<decltype(peekIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
-PARALLEL_FOR_LOOP
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-            ret._odata[ss] = peekIndex<Index>(lhs._odata[ss]);
-        }
-        return ret;
-    };
     template<int Index,class vobj>
-       auto peekIndex(const Lattice<vobj> &lhs,int i) -> Lattice<decltype(peekIndex<Index>(lhs._odata[0],i))>
+       auto PeekIndex(const Lattice<vobj> &lhs,int i) -> Lattice<decltype(peekIndex<Index>(lhs._odata[0],i))>
     {
       Lattice<decltype(peekIndex<Index>(lhs._odata[0],i))> ret(lhs._grid);
 PARALLEL_FOR_LOOP
@@ -31,7 +21,7 @@ PARALLEL_FOR_LOOP
         return ret;
     };
     template<int Index,class vobj>
-       auto peekIndex(const Lattice<vobj> &lhs,int i,int j) -> Lattice<decltype(peekIndex<Index>(lhs._odata[0],i,j))>
+       auto PeekIndex(const Lattice<vobj> &lhs,int i,int j) -> Lattice<decltype(peekIndex<Index>(lhs._odata[0],i,j))>
     {
       Lattice<decltype(peekIndex<Index>(lhs._odata[0],i,j))> ret(lhs._grid);
 PARALLEL_FOR_LOOP
@@ -44,16 +34,8 @@ PARALLEL_FOR_LOOP
     ////////////////////////////////////////////////////////////////////////////////////////////////////
     // Poke internal indices of a Lattice object
     ////////////////////////////////////////////////////////////////////////////////////////////////////
-    template<int Index,class vobj>
-    void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0]))> & rhs)
-    {
-PARALLEL_FOR_LOOP
-        for(int ss=0;ss<lhs._grid->oSites();ss++){
-	  pokeIndex<Index>(lhs._odata[ss],rhs._odata[ss]);
-	}      
-    }
     template<int Index,class vobj> 
-    void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0],0))> & rhs,int i)
+    void PokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0],0))> & rhs,int i)
     {
 PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){
@@ -61,7 +43,7 @@ PARALLEL_FOR_LOOP
 	}      
     }
     template<int Index,class vobj>
-      void pokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0],0,0))> & rhs,int i,int j)
+      void PokeIndex(Lattice<vobj> &lhs,const Lattice<decltype(peekIndex<Index>(lhs._odata[0],0,0))> & rhs,int i,int j)
     {
 PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){
diff --git a/lib/lattice/Lattice_trace.h b/lib/lattice/Lattice_trace.h
index 75cc5b87..1b35b412 100644
--- a/lib/lattice/Lattice_trace.h
+++ b/lib/lattice/Lattice_trace.h
@@ -26,7 +26,7 @@ PARALLEL_FOR_LOOP
     // Trace Index level dependent operation
     ////////////////////////////////////////////////////////////////////////////////////////////////////
     template<int Index,class vobj>
-    inline auto traceIndex(const Lattice<vobj> &lhs) -> Lattice<decltype(traceIndex<Index>(lhs._odata[0]))>
+    inline auto TraceIndex(const Lattice<vobj> &lhs) -> Lattice<decltype(traceIndex<Index>(lhs._odata[0]))>
     {
       Lattice<decltype(traceIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
 PARALLEL_FOR_LOOP
diff --git a/lib/lattice/Lattice_transpose.h b/lib/lattice/Lattice_transpose.h
index 73ef1747..a23ce48c 100644
--- a/lib/lattice/Lattice_transpose.h
+++ b/lib/lattice/Lattice_transpose.h
@@ -24,7 +24,7 @@ PARALLEL_FOR_LOOP
     // Index level dependent transpose
     ////////////////////////////////////////////////////////////////////////////////////////////////////
     template<int Index,class vobj>
-    inline auto transposeIndex(const Lattice<vobj> &lhs) -> Lattice<decltype(transposeIndex<Index>(lhs._odata[0]))>
+    inline auto TransposeIndex(const Lattice<vobj> &lhs) -> Lattice<decltype(transposeIndex<Index>(lhs._odata[0]))>
     {
       Lattice<decltype(transposeIndex<Index>(lhs._odata[0]))> ret(lhs._grid);
 PARALLEL_FOR_LOOP
diff --git a/lib/qcd/action/fermion/WilsonFermion.cc b/lib/qcd/action/fermion/WilsonFermion.cc
index 067d5113..f5ff7cb0 100644
--- a/lib/qcd/action/fermion/WilsonFermion.cc
+++ b/lib/qcd/action/fermion/WilsonFermion.cc
@@ -36,9 +36,9 @@ void WilsonFermion::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGauge
   LatticeColourMatrix U(GaugeGrid());
   for(int mu=0;mu<Nd;mu++){
     U = PeekIndex<LorentzIndex>(Umu,mu);
-    pokeIndex<LorentzIndex>(Uds,U,mu);
+    PokeIndex<LorentzIndex>(Uds,U,mu);
     U = adj(Cshift(U,mu,-1));
-    pokeIndex<LorentzIndex>(Uds,U,mu+4);
+    PokeIndex<LorentzIndex>(Uds,U,mu+4);
   }
 }
 
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.cc b/lib/qcd/action/fermion/WilsonFermion5D.cc
index 6d6097cf..316b0676 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.cc
+++ b/lib/qcd/action/fermion/WilsonFermion5D.cc
@@ -78,9 +78,9 @@ void WilsonFermion5D::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGau
   LatticeColourMatrix U(GaugeGrid());
   for(int mu=0;mu<Nd;mu++){
     U = PeekIndex<LorentzIndex>(Umu,mu);
-    pokeIndex<LorentzIndex>(Uds,U,mu);
+    PokeIndex<LorentzIndex>(Uds,U,mu);
     U = adj(Cshift(U,mu,-1));
-    pokeIndex<LorentzIndex>(Uds,U,mu+4);
+    PokeIndex<LorentzIndex>(Uds,U,mu+4);
   }
 }
 void WilsonFermion5D::DhopDir(const LatticeFermion &in, LatticeFermion &out,int dir5,int disp)
diff --git a/lib/qcd/utils/SUn.h b/lib/qcd/utils/SUn.h
index 4088a23b..69de3f9f 100644
--- a/lib/qcd/utils/SUn.h
+++ b/lib/qcd/utils/SUn.h
@@ -547,21 +547,21 @@ Note that in step D setting B ~ X - A and using B in place of A in step E will g
     LatticeMatrix Umu(out._grid);
     for(int mu=0;mu<Nd;mu++){
       LieRandomize(pRNG,Umu,1.0);
-      pokeIndex<LorentzIndex>(out,Umu,mu);
+      PokeIndex<LorentzIndex>(out,Umu,mu);
     }
   }
   static void TepidConfiguration(GridParallelRNG &pRNG,LatticeGaugeField &out){
     LatticeMatrix Umu(out._grid);
     for(int mu=0;mu<Nd;mu++){
       LieRandomize(pRNG,Umu,0.01);
-      pokeLorentz(out,Umu,mu);
+      PokeLorentz(out,Umu,mu);
     }
   }
   static void ColdConfiguration(GridParallelRNG &pRNG,LatticeGaugeField &out){
     LatticeMatrix Umu(out._grid);
     Umu=1.0;
     for(int mu=0;mu<Nd;mu++){
-      pokeLorentz(out,Umu,mu);
+      PokeLorentz(out,Umu,mu);
     }
   }
 
diff --git a/lib/tensors/Tensor_class.h b/lib/tensors/Tensor_class.h
index 7962517c..80247164 100644
--- a/lib/tensors/Tensor_class.h
+++ b/lib/tensors/Tensor_class.h
@@ -91,18 +91,12 @@ public:
     return _internal;
   }
 
-  // Type casts meta programmed
-  template<class U=vtype,class V=scalar_type,IfComplex<V> = 0,IfNotSimd<U> = 0>  
-    operator ComplexF () const { return(TensorRemove(_internal)); };
-  template<class U=vtype,class V=scalar_type,IfComplex<V> = 0,IfNotSimd<U> = 0>  
-    operator ComplexD () const { return(TensorRemove(_internal)); };
-  template<class U=vtype,class V=scalar_type,IfComplex<V> = 0,IfNotSimd<U> = 0>  
-    operator RealD () const { return(real(TensorRemove(_internal))); }
-  template<class U=vtype,class V=scalar_type,IfReal<V>    = 0,IfNotSimd<U> = 0>  
-    operator RealD    () const { return TensorRemove(_internal); }
-  template<class U=vtype,class V=scalar_type,IfInteger<V> = 0,IfNotSimd<U> = 0>  
-    operator Integer  () const { return Integer(TensorRemove(_internal)); }
-
+  // Type casts meta programmed, must be pure scalar to match TensorRemove
+  template<class U=vtype,class V=scalar_type,IfComplex<V> = 0,IfNotSimd<U> = 0> operator ComplexF () const { return(TensorRemove(_internal)); };
+  template<class U=vtype,class V=scalar_type,IfComplex<V> = 0,IfNotSimd<U> = 0> operator ComplexD () const { return(TensorRemove(_internal)); };
+  //  template<class U=vtype,class V=scalar_type,IfComplex<V> = 0,IfNotSimd<U> = 0> operator RealD    () const { return(real(TensorRemove(_internal))); }
+  template<class U=vtype,class V=scalar_type,IfReal<V>    = 0,IfNotSimd<U> = 0> operator RealD    () const { return TensorRemove(_internal); }
+  template<class U=vtype,class V=scalar_type,IfInteger<V> = 0,IfNotSimd<U> = 0> operator Integer  () const { return Integer(TensorRemove(_internal)); }
   
   // convert from a something to a scalar via constructor of something arg
   template<class T,typename std::enable_if<!isGridTensor<T>::value, T>::type* = nullptr > strong_inline iScalar<vtype> operator = (T arg)
diff --git a/lib/tensors/Tensor_peek.h b/lib/tensors/Tensor_peek.h
index 5c8d4a93..9c037581 100644
--- a/lib/tensors/Tensor_peek.h
+++ b/lib/tensors/Tensor_peek.h
@@ -11,7 +11,7 @@ namespace Grid {
 //template<int Level> inline ComplexD peekIndex(const ComplexD arg) { return arg;}
 //template<int Level> inline RealF peekIndex(const RealF arg) { return arg;}
 //template<int Level> inline RealD peekIndex(const RealD arg) { return arg;}
-
+#if 0
 // Scalar peek, no indices
 template<int Level,class vtype,typename std::enable_if< iScalar<vtype>::TensorLevel == Level >::type * =nullptr> inline 
   auto peekIndex(const iScalar<vtype> &arg) ->  iScalar<vtype> 
@@ -88,6 +88,7 @@ template<int Level,class vtype,int N, typename std::enable_if< iScalar<vtype>::T
   }
   return ret;
 }
+
 // matrix
 template<int Level,class vtype,int N, typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
 auto peekIndex(const iMatrix<vtype,N> &arg) ->   iMatrix<decltype(peekIndex<Level>(arg._internal[0][0])),N> 
@@ -119,6 +120,7 @@ template<int Level,class vtype,int N, typename std::enable_if< iScalar<vtype>::T
   }}
   return ret;
 }
+#endif
 
 
 }
diff --git a/lib/tensors/Tensor_poke.h b/lib/tensors/Tensor_poke.h
index a015478e..8a3fd69d 100644
--- a/lib/tensors/Tensor_poke.h
+++ b/lib/tensors/Tensor_poke.h
@@ -5,7 +5,7 @@ namespace Grid {
 //////////////////////////////////////////////////////////////////////////////
 // Poke a specific index; 
 //////////////////////////////////////////////////////////////////////////////
-
+#if 0
 // Scalar poke
 template<int Level,class vtype,typename std::enable_if< iScalar<vtype>::TensorLevel == Level >::type * =nullptr> inline 
   void pokeIndex(iScalar<vtype> &ret, const iScalar<vtype> &arg)
@@ -18,7 +18,7 @@ template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::Te
 {
   ret._internal[i] = arg._internal;
 }
-// Vector poke, two indices
+//Matrix poke, two indices
 template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::TensorLevel == Level >::type * =nullptr> inline 
   void pokeIndex(iMatrix<vtype,N> &ret, const iScalar<vtype> &arg,int i,int j)
 {
@@ -31,7 +31,6 @@ template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::Te
 // scalar
 template<int Level,class vtype,typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
 void pokeIndex(iScalar<vtype> &ret, const iScalar<decltype(peekIndex<Level>(ret._internal))>  &arg)
-		 
 {
   pokeIndex<Level>(ret._internal,arg._internal);
 }
@@ -95,7 +94,7 @@ template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::Te
     pokeIndex<Level>(ret._internal[ii][jj],arg._internal[ii][jj],i,j);
   }}
 }
-
+#endif
 
 }
 #endif
diff --git a/lib/tensors/Tensor_trace.h b/lib/tensors/Tensor_trace.h
index 749b292e..bcde670c 100644
--- a/lib/tensors/Tensor_trace.h
+++ b/lib/tensors/Tensor_trace.h
@@ -32,58 +32,327 @@ inline auto trace(const iScalar<vtype> &arg) -> iScalar<decltype(trace(arg._inte
     ret._internal=trace(arg._internal);
     return ret;
 }
-////////////////////////////////////////////////////////////////////////////////////////////////////////
-// Trace Specific indices.
-////////////////////////////////////////////////////////////////////////////////////////////////////////
-template<int Level,class vtype,typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline auto 
-traceIndex (const iScalar<vtype> &arg) -> iScalar<decltype(traceIndex<Level>(arg._internal))>
-{
-  iScalar<decltype(traceIndex<Level>(arg._internal))> ret;
-  ret._internal=traceIndex<Level>(arg._internal);
-  return ret;
-}
-template<int Level,class vtype,typename std::enable_if< iScalar<vtype>::TensorLevel == Level >::type * =nullptr> inline auto
-traceIndex (const iScalar<vtype> &arg) -> iScalar<vtype>
-{
-  return arg;
-}
 
-// If we hit the right index, return scalar and trace it with no further recursion
-template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::TensorLevel == Level >::type * =nullptr> inline 
-auto traceIndex(const iMatrix<vtype,N> &arg) ->  iScalar<vtype> 
-{
-  iScalar<vtype> ret;
-  zeroit(ret._internal);
-  for(int i=0;i<N;i++){
-    ret._internal = ret._internal + arg._internal[i][i];
-  }
-  return ret;
-}
-
-// not this level, so recurse
-template<int Level,class vtype,int N,typename std::enable_if< iScalar<vtype>::TensorLevel != Level >::type * =nullptr> inline 
-auto traceIndex(const iMatrix<vtype,N> &arg) ->  iMatrix<decltype(traceIndex<Level>(arg._internal[0][0])),N> 
-{
-  iMatrix<decltype(traceIndex<Level>(arg._internal[0][0])),N> ret;
-  for(int i=0;i<N;i++){
-  for(int j=0;j<N;j++){
-    ret._internal[i][j] = traceIndex<Level>(arg._internal[i][j]);
-  }}
-  return ret;
-}
+/////////////////////////////////////
+// Alternate implementation .. count 
+// indices from left. Annoying but
+// I'm working around compiler bugs in gcc and earlier icpc
+/////////////////////////////////////
 
 // Allow to recurse if vector, but never terminate on a vector
-// trace of a different index can distribute across the vector index in a replicated way
-// but we do not trace a vector index.
- template<int Level,class vtype,int N,typename std::enable_if< iVector<vtype, N>::TensorLevel != Level >::type * =nullptr> inline 
-auto traceIndex(const iVector<vtype,N> &arg) ->  iVector<decltype(traceIndex<Level>(arg._internal[0])),N> 
-{
-  iVector<decltype(traceIndex<Level>(arg._internal[0])),N> ret;
-  for(int i=0;i<N;i++){
-    ret._internal[i] = traceIndex<Level>(arg._internal[i]);
+// ltrace of a different index can distribute across the vector index in a replicated way
+// but we do not ltrace a vector index.
+template<int Level> 
+class TensorIndexRecursion {
+
+ public:
+  ////////////////////////////////////////////
+  // Recursion for tracing a specific index
+  ////////////////////////////////////////////
+  template<class vtype>
+  static auto traceIndex(const iScalar<vtype> arg) ->  iScalar<decltype(TensorIndexRecursion<Level-1>::traceIndex(arg._internal))> 
+  {
+    iScalar<decltype(TensorIndexRecursion<Level-1>::traceIndex(arg._internal))> ret;
+    ret._internal = TensorIndexRecursion<Level-1>::traceIndex(arg._internal);
+    return ret;
   }
+  template<class vtype,int N>
+  static auto traceIndex(const iVector<vtype,N> arg) ->  iVector<decltype(TensorIndexRecursion<Level-1>::traceIndex(arg._internal[0])),N> 
+  {
+    iVector<decltype(TensorIndexRecursion<Level-1>::traceIndex(arg._internal[0])),N> ret;
+    for(int i=0;i<N;i++){
+      ret._internal[i] = TensorIndexRecursion<Level-1>::traceIndex(arg._internal[i]);
+    }
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto traceIndex(const iMatrix<vtype,N> arg) ->  iMatrix<decltype(TensorIndexRecursion<Level-1>::traceIndex(arg._internal[0][0])),N> 
+  {
+    iMatrix<decltype(TensorIndexRecursion<Level-1>::traceIndex(arg._internal[0][0])),N> ret;
+    for(int i=0;i<N;i++){
+    for(int j=0;j<N;j++){
+      ret._internal[i][j] = TensorIndexRecursion<Level-1>::traceIndex(arg._internal[i][j]);
+    }}
+    return ret;
+  }
+  ////////////////////////////////////////////
+  // Recursion for peeking a specific index
+  ////////////////////////////////////////////
+  template<class vtype>
+  static auto peekIndex(const iScalar<vtype> arg,int i) ->  iScalar<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal,0))> 
+  {
+    iScalar<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal,0))> ret;
+    ret._internal = TensorIndexRecursion<Level-1>::peekIndex(arg._internal,i);
+    return ret;
+  }
+  template<class vtype>
+  static auto peekIndex(const iScalar<vtype> arg,int i,int j) ->  iScalar<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal,0,0))> 
+  {
+    iScalar<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal,0,0))> ret;
+    ret._internal = TensorIndexRecursion<Level-1>::peekIndex(arg._internal,i,j);
+    return ret;
+  }
+
+  template<class vtype,int N>
+  static auto peekIndex(const iVector<vtype,N> arg,int ii) ->  iVector<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0],0)),N> 
+  {
+    iVector<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0],0)),N> ret;
+    for(int i=0;i<N;i++){
+      ret._internal[i] = TensorIndexRecursion<Level-1>::peekIndex(arg._internal[i],ii);
+    }
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto peekIndex(const iVector<vtype,N> arg,int ii,int jj) ->  iVector<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0],0,0)),N> 
+  {
+    iVector<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0],0,0)),N> ret;
+    for(int i=0;i<N;i++){
+      ret._internal[i] = TensorIndexRecursion<Level-1>::peekIndex(arg._internal[i],ii,jj);
+    }
+    return ret;
+  }
+
+  template<class vtype,int N>
+  static auto peekIndex(const iMatrix<vtype,N> arg,int ii) ->  iMatrix<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0][0],0)),N> 
+  {
+    iMatrix<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0][0],0)),N> ret;
+    for(int i=0;i<N;i++){
+    for(int j=0;j<N;j++){
+      ret._internal[i][j] = TensorIndexRecursion<Level-1>::peekIndex(arg._internal[i][j],ii);
+    }}
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto peekIndex(const iMatrix<vtype,N> arg,int ii,int jj) ->  iMatrix<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0][0],0,0)),N> 
+  {
+    iMatrix<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0][0],0,0)),N> ret;
+    for(int i=0;i<N;i++){
+    for(int j=0;j<N;j++){
+      ret._internal[i][j] = TensorIndexRecursion<Level-1>::peekIndex(arg._internal[i][j],ii,jj);
+    }}
+    return ret;
+  }
+  ////////////////////////////////////////////
+  // Recursion for poking a specific index
+  ////////////////////////////////////////////
+
+  template<class vtype> inline static 
+    void pokeIndex(iScalar<vtype> &ret, const iScalar<decltype(TensorIndexRecursion<Level-1>::peekIndex(ret._internal,0))> &arg, int i)
+    {
+      TensorIndexRecursion<Level-1>::pokeIndex(ret._internal,arg._internal,i);
+    }
+  template<class vtype> inline static 
+    void pokeIndex(iScalar<vtype> &ret, const iScalar<decltype(TensorIndexRecursion<Level-1>::peekIndex(ret._internal,0,0))> &arg, int i,int j)
+    {
+      TensorIndexRecursion<Level-1>::pokeIndex(ret._internal,arg._internal,i,j);
+    }
+
+  template<class vtype,int N> inline static 
+    void pokeIndex(iVector<vtype,N> &ret, const iVector<decltype(TensorIndexRecursion<Level-1>::peekIndex(ret._internal,0)),N> &arg, int i)
+    {
+      for(int ii=0;ii<N;ii++){
+	TensorIndexRecursion<Level-1>::pokeIndex(ret._internal[ii],arg._internal[ii],i);
+      }
+    }
+  template<class vtype,int N> inline static 
+    void pokeIndex(iVector<vtype,N> &ret, const iVector<decltype(TensorIndexRecursion<Level-1>::peekIndex(ret._internal,0)),N> &arg, int i,int j)
+    {
+      for(int ii=0;ii<N;ii++){
+	TensorIndexRecursion<Level-1>::pokeIndex(ret._internal[ii],arg._internal[ii],i,j);
+      }
+    }
+
+  template<class vtype,int N> inline static 
+    void pokeIndex(iMatrix<vtype,N> &ret, const iMatrix<decltype(TensorIndexRecursion<Level-1>::peekIndex(ret._internal,0)),N> &arg, int i)
+    {
+      for(int ii=0;ii<N;ii++){
+      for(int jj=0;jj<N;jj++){
+	TensorIndexRecursion<Level-1>::pokeIndex(ret._internal[ii][jj],arg._internal[ii][jj],i);
+      }}
+    }
+  template<class vtype,int N> inline static 
+    void pokeIndex(iMatrix<vtype,N> &ret, const iMatrix<decltype(TensorIndexRecursion<Level-1>::peekIndex(ret._internal,0)),N> &arg, int i,int j)
+    {
+      for(int ii=0;ii<N;ii++){
+      for(int jj=0;jj<N;jj++){
+	TensorIndexRecursion<Level-1>::pokeIndex(ret._internal[ii][jj],arg._internal[ii][jj],i,j);
+      }}
+    }
+
+  ////////////////////////////////////////////
+  // Recursion for transposing a specific index
+  ////////////////////////////////////////////
+  template<class vtype>
+  static auto transposeIndex(const iScalar<vtype> arg) ->  iScalar<vtype> 
+  {
+    iScalar<vtype> ret;
+    ret._internal = TensorIndexRecursion<Level-1>::transposeIndex(arg._internal);
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto transposeIndex(const iVector<vtype,N> arg) ->  iVector<vtype,N> 
+  {
+    iVector<vtype,N> ret;
+    for(int i=0;i<N;i++){
+      ret._internal[i] = TensorIndexRecursion<Level-1>::transposeIndex(arg._internal[i]);
+    }
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto transposeIndex(const iMatrix<vtype,N> arg) ->  iMatrix<vtype,N> 
+  {
+    iMatrix<vtype,N> ret;
+    for(int i=0;i<N;i++){
+    for(int j=0;j<N;j++){
+      ret._internal[i][j] = TensorIndexRecursion<Level-1>::transposeIndex(arg._internal[i][j]);
+    }}
+    return ret;
+  }
+};
+
+////////////////////////////
+// strip const & ref quali's
+////////////////////////////
+#define RemoveCRV(a) typename std::remove_const<typename std::remove_reference<decltype(a)>::type>::type
+template<>
+class TensorIndexRecursion<0> {
+ public:
+
+  /////////////////////////////////////////
+  // Ends recursion for trace (scalar/vector/matrix)
+  /////////////////////////////////////////
+  template<class vtype>
+  static auto traceIndex(const iScalar<vtype> arg) ->  iScalar<RemoveCRV(arg._internal)>
+  {
+    iScalar<RemoveCRV(arg._internal)> ret;
+    ret._internal = arg._internal;
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto traceIndex(const iVector<vtype,N> arg) ->  iScalar<RemoveCRV(arg._internal[0])>
+  {
+    iScalar<RemoveCRV(arg._internal[0])> ret;
+    ret._internal=zero;
+    for(int i=0;i<N;i++){
+      ret._internal = ret._internal+ arg._internal[i];
+    }
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto traceIndex(const iMatrix<vtype,N> arg) ->  iScalar<RemoveCRV(arg._internal[0][0])> 
+  {
+    iScalar<RemoveCRV(arg._internal[0][0])> ret;
+    ret=zero;
+    for(int i=0;i<N;i++){
+      ret._internal = ret._internal+arg._internal[i][i];
+    }
+    return ret;
+  }
+  /////////////////////////////////////////
+  // Ends recursion for transpose scalar/vector/matrix
+  /////////////////////////////////////////
+  template<class vtype>
+  static auto transposeIndex(const iScalar<vtype> arg) ->  iScalar<vtype>
+  {
+    iScalar<vtype> ret;
+    ret._internal = arg._internal;
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto transposeIndex(const iVector<vtype,N> arg) ->  iVector<vtype,N>
+  {
+    iScalar<vtype> ret;
+    ret._internal=zero;
+    for(int i=0;i<N;i++){
+      ret._internal = ret._internal+ arg._internal[i];
+    }
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto transposeIndex(const iMatrix<vtype,N> arg)  ->  iMatrix<vtype,N> 
+  {
+    iScalar<vtype> ret;
+    ret=zero;
+    for(int i=0;i<N;i++){
+      ret._internal = ret._internal+arg._internal[i][i];
+    }
+    return ret;
+  }
+  ////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+  // End recursion for peeking a specific index; single index on vector, double index on matrix
+  ////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+  template<class vtype,int N>
+  static auto peekIndex(const iVector<vtype,N> arg,int ii) ->  iScalar<vtype> 
+  {
+    iScalar<vtype> ret;
+    ret._internal = arg._internal[ii];
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto peekIndex(const iMatrix<vtype,N> arg,int ii,int jj) ->  iScalar<vtype>
+  {
+    iScalar<vtype> ret;
+    ret._internal = arg._internal[ii][jj];
+    return ret;
+  }
+  // Vector poke, one index
+  template<class vtype,int N> inline static 
+    void pokeIndex(iVector<vtype,N> &ret, const iScalar<vtype> &arg,int i)
+    {
+      ret._internal[i] = arg._internal;
+    }
+  // Matrix poke two indices
+  template<class vtype,int N> inline static 
+    void pokeIndex(iMatrix<vtype,N> &ret, const iScalar<vtype> &arg,int i,int j)
+    {
+      ret._internal[i][j] = arg._internal;
+    }
+
+};
+
+////////////////////////////////////////////////////////////////////////////////////////////////////////
+// External wrappers
+////////////////////////////////////////////////////////////////////////////////////////////////////////
+
+template<int Level,class vtype> inline auto traceIndex (const vtype &arg) -> RemoveCRV(TensorIndexRecursion<Level>::traceIndex(arg))
+{
+  RemoveCRV(TensorIndexRecursion<Level>::traceIndex(arg)) ret;
+  ret=TensorIndexRecursion<Level>::traceIndex(arg);
   return ret;
 }
+template<int Level,class vtype> inline auto transposeIndex (const vtype &arg) -> RemoveCRV(TensorIndexRecursion<Level>::transposeIndex(arg))
+{
+  RemoveCRV(TensorIndexRecursion<Level>::transposeIndex(arg)) ret;
+  ret=TensorIndexRecursion<Level>::transposeIndex(arg);
+  return ret;
+}
+template<int Level,class vtype> inline auto peekIndex (const vtype &arg,int i) -> RemoveCRV(TensorIndexRecursion<Level>::peekIndex(arg,0))
+{
+  RemoveCRV(TensorIndexRecursion<Level>::peekIndex(arg,0)) ret;
+  ret=TensorIndexRecursion<Level>::peekIndex(arg,i);
+  return ret;
+}
+template<int Level,class vtype> inline auto peekIndex (const vtype &arg,int i,int j) -> RemoveCRV(TensorIndexRecursion<Level>::peekIndex(arg,0,0))
+{
+  RemoveCRV(TensorIndexRecursion<Level>::peekIndex(arg,0,0)) ret;
+  ret=TensorIndexRecursion<Level>::peekIndex(arg,i,j);
+  return ret;
+}
+
+template<int Level,class vtype> inline 
+void pokeIndex (vtype &ret,const decltype(TensorIndexRecursion<Level>::peekIndex(ret,0)) &arg,int i) 
+{
+  TensorIndexRecursion<Level>::pokeIndex(ret,arg,i);
+}
+
+template<int Level,class vtype> inline 
+void pokeIndex (vtype &ret,const decltype(TensorIndexRecursion<Level>::peekIndex(ret,0,0)) &arg,int i,int j) 
+{
+  TensorIndexRecursion<Level>::pokeIndex(ret,arg,i,j);
+}
+
+
+#undef RemoveCRV
 
 }
 #endif
diff --git a/tests/Test_dwf_even_odd.cc b/tests/Test_dwf_even_odd.cc
index 82b39f68..5ec73212 100644
--- a/tests/Test_dwf_even_odd.cc
+++ b/tests/Test_dwf_even_odd.cc
@@ -53,7 +53,7 @@ int main (int argc, char ** argv)
     random(RNG4,U[nn]);
     if ( nn>0 ) 
       U[nn]=zero;
-    pokeIndex<LorentzIndex>(Umu,U[nn],nn);
+    PokeIndex<LorentzIndex>(Umu,U[nn],nn);
   }
 
   RealD mass=0.1;
diff --git a/tests/Test_main.cc b/tests/Test_main.cc
index 6bef8629..85450047 100644
--- a/tests/Test_main.cc
+++ b/tests/Test_main.cc
@@ -141,7 +141,7 @@ int main (int argc, char ** argv)
     //    rscalar=real(scalar);
     //    iscalar=imag(scalar);
     //    scalar =cmplx(rscalar,iscalar);
-    pokeIndex<2>(cVec,scalar,1);
+    PokeIndex<ColourIndex>(cVec,scalar,1);
 
 
     scalar=transpose(scalar);
diff --git a/tests/Test_quenched_update.cc b/tests/Test_quenched_update.cc
index 887a27a8..cb58144b 100644
--- a/tests/Test_quenched_update.cc
+++ b/tests/Test_quenched_update.cc
@@ -72,7 +72,7 @@ int main (int argc, char ** argv)
 
 	}
 
-	pokeIndex<LorentzIndex>(Umu,link,mu);
+	PokeIndex<LorentzIndex>(Umu,link,mu);
 	//reunitarise link;
 	
       }
diff --git a/tests/Test_wilson_even_odd.cc b/tests/Test_wilson_even_odd.cc
index 16019c9c..91fbb90f 100644
--- a/tests/Test_wilson_even_odd.cc
+++ b/tests/Test_wilson_even_odd.cc
@@ -55,7 +55,7 @@ int main (int argc, char ** argv)
   Umu=zero;
   for(int nn=0;nn<Nd;nn++){
     random(pRNG,U[nn]);
-    pokeIndex<LorentzIndex>(Umu,U[nn],nn);
+    PokeIndex<LorentzIndex>(Umu,U[nn],nn);
   }
 
   RealD mass=0.1;

From b1399414236020a5fa5bbfd468dd66ac1dd2c333 Mon Sep 17 00:00:00 2001
From: paboyle <paboyle@ph.ed.ac.uk>
Date: Wed, 1 Jul 2015 13:04:02 +0100
Subject: [PATCH 371/429] Remove dependency on wrong file

---
 tests/Test_dwf_hdcr.cc | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tests/Test_dwf_hdcr.cc b/tests/Test_dwf_hdcr.cc
index 8d1a5fce..bafdb639 100644
--- a/tests/Test_dwf_hdcr.cc
+++ b/tests/Test_dwf_hdcr.cc
@@ -1,6 +1,6 @@
 #include <Grid.h>
 #include <algorithms/iterative/PrecGeneralisedConjugateResidual.h>
-#include <algorithms/iterative/PrecConjugateResidual.h>
+//#include <algorithms/iterative/PrecConjugateResidual.h>
 
 using namespace std;
 using namespace Grid;

From dc66161f477aef5470db0ba4acdcbfd46b9bb731 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 1 Jul 2015 22:45:15 +0100
Subject: [PATCH 372/429] Change the SIMD command correctly with precision =
 double vs. single and connect the "Real" default precisoin to a configure
 flag. Have RealF, RealD and Real types, where Real is compile target
 dependent single/double, RealF is single and RealD is double etc..

---
 benchmarks/Benchmark_dwf.cc          |  2 +-
 benchmarks/Benchmark_wilson.cc       |  5 ++++-
 configure                            | 24 ++++++++++++++++++++++++
 configure.ac                         | 11 +++++++++++
 lib/GridConfig.h.in                  |  6 ++++++
 lib/Simd.h                           |  1 -
 tests/Test_GaugeAction.cc            |  2 +-
 tests/Test_cayley_cg.cc              |  2 +-
 tests/Test_cayley_coarsen_support.cc |  4 ++--
 tests/Test_cayley_even_odd.cc        |  2 +-
 tests/Test_cayley_ldop_cr.cc         |  4 ++--
 tests/Test_cf_coarsen_support.cc     |  2 +-
 tests/Test_cf_cr_unprec.cc           |  2 +-
 tests/Test_contfrac_cg.cc            |  2 +-
 tests/Test_contfrac_even_odd.cc      |  2 +-
 tests/Test_cshift.cc                 |  2 +-
 tests/Test_cshift_red_black.cc       |  2 +-
 tests/Test_dwf_cg_prec.cc            |  2 +-
 tests/Test_dwf_cg_schur.cc           |  2 +-
 tests/Test_dwf_cg_unprec.cc          |  2 +-
 tests/Test_dwf_cr_unprec.cc          |  2 +-
 tests/Test_dwf_even_odd.cc           |  2 +-
 tests/Test_dwf_fpgcr.cc              |  2 +-
 tests/Test_gamma.cc                  |  2 +-
 tests/Test_lie_generators.cc         |  2 +-
 tests/Test_main.cc                   |  2 +-
 tests/Test_multishift_sqrt.cc        |  2 +-
 tests/Test_quenched_update.cc        |  2 +-
 tests/Test_rng.cc                    |  2 +-
 tests/Test_rng_fixed.cc              |  2 +-
 tests/Test_simd.cc                   |  2 +-
 tests/Test_stencil.cc                |  2 +-
 tests/Test_wilson_cg_prec.cc         |  2 +-
 tests/Test_wilson_cg_schur.cc        |  2 +-
 tests/Test_wilson_cg_unprec.cc       |  2 +-
 tests/Test_wilson_cr_unprec.cc       |  2 +-
 tests/Test_wilson_even_odd.cc        |  4 +++-
 37 files changed, 81 insertions(+), 36 deletions(-)

diff --git a/benchmarks/Benchmark_dwf.cc b/benchmarks/Benchmark_dwf.cc
index 290bf1df..9ce8f0db 100644
--- a/benchmarks/Benchmark_dwf.cc
+++ b/benchmarks/Benchmark_dwf.cc
@@ -25,7 +25,7 @@ int main (int argc, char ** argv)
 
   std::vector<int> latt4 = GridDefaultLatt();
   const int Ls=8;
-  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
   GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
   GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
   GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
diff --git a/benchmarks/Benchmark_wilson.cc b/benchmarks/Benchmark_wilson.cc
index 7d1bd8df..aa34ccf4 100644
--- a/benchmarks/Benchmark_wilson.cc
+++ b/benchmarks/Benchmark_wilson.cc
@@ -22,13 +22,16 @@ int main (int argc, char ** argv)
 
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
   GridCartesian               Grid(latt_size,simd_layout,mpi_layout);
   GridRedBlackCartesian     RBGrid(latt_size,simd_layout,mpi_layout);
 
   int threads = GridThread::GetThreads();
   std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+  std::cout << "Grid floating point word size is REALF"<< sizeof(RealF)<<std::endl;
+  std::cout << "Grid floating point word size is REALD"<< sizeof(RealD)<<std::endl;
+  std::cout << "Grid floating point word size is REAL"<< sizeof(Real)<<std::endl;
 
   std::vector<int> seeds({1,2,3,4});
   GridParallelRNG          pRNG(&Grid);
diff --git a/configure b/configure
index 07b99339..06ca42d5 100755
--- a/configure
+++ b/configure
@@ -751,6 +751,7 @@ enable_silent_rules
 enable_dependency_tracking
 enable_openmp
 enable_simd
+enable_precision
 enable_comms
 enable_doxygen
 enable_dot
@@ -1398,6 +1399,8 @@ Optional Features:
   --enable-simd=SSE4|AVX|AVX2|AVX512|MIC
                           Select instructions to be SSE4.0, AVX 1.0, AVX
                           2.0+FMA, AVX 512, MIC
+  --enable-precision=single|double
+                          Select default word size of Real
   --enable-comms=none|mpi Select communications
   --enable-doxygen        enable documentation generation with doxygen (auto)
   --enable-dot            use 'dot' to generate graphs in doxygen (auto)
@@ -6728,6 +6731,27 @@ $as_echo "#define EMPTY_SIMD 1" >>confdefs.h
      ;;
 esac
 
+# Check whether --enable-precision was given.
+if test "${enable_precision+set}" = set; then :
+  enableval=$enable_precision; ac_PRECISION=${enable_precision}
+else
+  ac_PRECISION=single
+fi
+
+case ${ac_PRECISION} in
+     single)
+       echo default precision is single
+
+$as_echo "#define GRID_DEFAULT_PRECISION_SINGLE 1" >>confdefs.h
+
+     ;;
+     double)
+       echo default precision is double
+
+$as_echo "#define GRID_DEFAULT_PRECISION_DOUBLE 1" >>confdefs.h
+
+     ;;
+esac
 
 # Check whether --enable-comms was given.
 if test "${enable_comms+set}" = set; then :
diff --git a/configure.ac b/configure.ac
index cf0da9eb..5ef6fa8d 100644
--- a/configure.ac
+++ b/configure.ac
@@ -120,6 +120,17 @@ case ${ac_SIMD} in
      ;;
 esac
 
+AC_ARG_ENABLE([precision],[AC_HELP_STRING([--enable-precision=single|double],[Select default word size of Real])],[ac_PRECISION=${enable_precision}],[ac_PRECISION=single])
+case ${ac_PRECISION} in
+     single)
+       echo default precision is single
+       AC_DEFINE([GRID_DEFAULT_PRECISION_SINGLE],[1],[GRID_DEFAULT_PRECISION is SINGLE] )
+     ;;
+     double)
+       echo default precision is double
+       AC_DEFINE([GRID_DEFAULT_PRECISION_DOUBLE],[1],[GRID_DEFAULT_PRECISION is DOUBLE] )
+     ;;
+esac
 
 AC_ARG_ENABLE([comms],[AC_HELP_STRING([--enable-comms=none|mpi],[Select communications])],[ac_COMMS=${enable_comms}],[ac_COMMS=none])
 
diff --git a/lib/GridConfig.h.in b/lib/GridConfig.h.in
index 8ff77f3d..1ec0ecf5 100644
--- a/lib/GridConfig.h.in
+++ b/lib/GridConfig.h.in
@@ -18,6 +18,12 @@
 /* GRID_COMMS_NONE */
 #undef GRID_COMMS_NONE
 
+/* GRID_DEFAULT_PRECISION is DOUBLE */
+#undef GRID_DEFAULT_PRECISION_DOUBLE
+
+/* GRID_DEFAULT_PRECISION is SINGLE */
+#undef GRID_DEFAULT_PRECISION_SINGLE
+
 /* Support Altivec instructions */
 #undef HAVE_ALTIVEC
 
diff --git a/lib/Simd.h b/lib/Simd.h
index 474e5dd3..7da0d7ad 100644
--- a/lib/Simd.h
+++ b/lib/Simd.h
@@ -17,7 +17,6 @@ namespace Grid {
 
   typedef  float  RealF;
   typedef  double RealD;
-#define GRID_DEFAULT_PRECISION_DOUBLE
 #ifdef GRID_DEFAULT_PRECISION_DOUBLE
   typedef RealD   Real;
 #else
diff --git a/tests/Test_GaugeAction.cc b/tests/Test_GaugeAction.cc
index 1ce8306e..612eb693 100644
--- a/tests/Test_GaugeAction.cc
+++ b/tests/Test_GaugeAction.cc
@@ -33,7 +33,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
 
-  std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplex::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
   std::vector<int> latt_size  ({16,16,16,32});
   std::vector<int> clatt_size  ({4,4,4,8});
diff --git a/tests/Test_cayley_cg.cc b/tests/Test_cayley_cg.cc
index 32174035..23004be9 100644
--- a/tests/Test_cayley_cg.cc
+++ b/tests/Test_cayley_cg.cc
@@ -55,7 +55,7 @@ int main (int argc, char ** argv)
   std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
 
   const int Ls=8;
-  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
   GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
   GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
   GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
diff --git a/tests/Test_cayley_coarsen_support.cc b/tests/Test_cayley_coarsen_support.cc
index 72536942..473fa10a 100644
--- a/tests/Test_cayley_coarsen_support.cc
+++ b/tests/Test_cayley_coarsen_support.cc
@@ -23,7 +23,7 @@ int main (int argc, char ** argv)
 
   const int Ls=8;
 
-  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
   GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
 
   GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
@@ -34,7 +34,7 @@ int main (int argc, char ** argv)
   for(int d=0;d<clatt.size();d++){
     clatt[d] = clatt[d]/2;
   }
-  GridCartesian *Coarse4d =  SpaceTimeGrid::makeFourDimGrid(clatt, GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());;
+  GridCartesian *Coarse4d =  SpaceTimeGrid::makeFourDimGrid(clatt, GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());;
   GridCartesian *Coarse5d =  SpaceTimeGrid::makeFiveDimGrid(1,Coarse4d);
 
   std::vector<int> seeds4({1,2,3,4});
diff --git a/tests/Test_cayley_even_odd.cc b/tests/Test_cayley_even_odd.cc
index 4b5630f7..21c828db 100644
--- a/tests/Test_cayley_even_odd.cc
+++ b/tests/Test_cayley_even_odd.cc
@@ -32,7 +32,7 @@ int main (int argc, char ** argv)
   std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
 
   const int Ls=8;
-  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
   GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
   GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
   GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
diff --git a/tests/Test_cayley_ldop_cr.cc b/tests/Test_cayley_ldop_cr.cc
index 8926c19c..2e67df26 100644
--- a/tests/Test_cayley_ldop_cr.cc
+++ b/tests/Test_cayley_ldop_cr.cc
@@ -10,7 +10,7 @@ int main (int argc, char ** argv)
 
   const int Ls=8;
 
-  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
   GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
 
   GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
@@ -23,7 +23,7 @@ int main (int argc, char ** argv)
   for(int d=0;d<clatt.size();d++){
     clatt[d] = clatt[d]/2;
   }
-  GridCartesian *Coarse4d =  SpaceTimeGrid::makeFourDimGrid(clatt, GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());;
+  GridCartesian *Coarse4d =  SpaceTimeGrid::makeFourDimGrid(clatt, GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());;
   GridCartesian *Coarse5d =  SpaceTimeGrid::makeFiveDimGrid(1,Coarse4d);
 
   std::vector<int> seeds4({1,2,3,4});
diff --git a/tests/Test_cf_coarsen_support.cc b/tests/Test_cf_coarsen_support.cc
index 4a4cf3e9..959f5a9b 100644
--- a/tests/Test_cf_coarsen_support.cc
+++ b/tests/Test_cf_coarsen_support.cc
@@ -22,7 +22,7 @@ int main (int argc, char ** argv)
 
   const int Ls=9;
 
-  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
   GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
   GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
   GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
diff --git a/tests/Test_cf_cr_unprec.cc b/tests/Test_cf_cr_unprec.cc
index e74fc85e..90acb503 100644
--- a/tests/Test_cf_cr_unprec.cc
+++ b/tests/Test_cf_cr_unprec.cc
@@ -22,7 +22,7 @@ int main (int argc, char ** argv)
 
   const int Ls=9;
 
-  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
   GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
   GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
   GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
diff --git a/tests/Test_contfrac_cg.cc b/tests/Test_contfrac_cg.cc
index 5851f5e6..1b937cef 100644
--- a/tests/Test_contfrac_cg.cc
+++ b/tests/Test_contfrac_cg.cc
@@ -56,7 +56,7 @@ int main (int argc, char ** argv)
   std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
 
   const int Ls=9;
-  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
   GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
   GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
   GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
diff --git a/tests/Test_contfrac_even_odd.cc b/tests/Test_contfrac_even_odd.cc
index a246b904..5d39b7df 100644
--- a/tests/Test_contfrac_even_odd.cc
+++ b/tests/Test_contfrac_even_odd.cc
@@ -32,7 +32,7 @@ int main (int argc, char ** argv)
   std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
 
   const int Ls=9;
-  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
   GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
   GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
   GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
diff --git a/tests/Test_cshift.cc b/tests/Test_cshift.cc
index e571a338..cd85439c 100644
--- a/tests/Test_cshift.cc
+++ b/tests/Test_cshift.cc
@@ -8,7 +8,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplex::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
 
   GridCartesian        Fine(latt_size,simd_layout,mpi_layout);
diff --git a/tests/Test_cshift_red_black.cc b/tests/Test_cshift_red_black.cc
index 5cafd04f..79ebcfc5 100644
--- a/tests/Test_cshift_red_black.cc
+++ b/tests/Test_cshift_red_black.cc
@@ -9,7 +9,7 @@ int main (int argc, char ** argv)
 
   std::vector<int> latt_size   = GridDefaultLatt();
   int Nd = latt_size.size();
-  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
 
   std::vector<int> mask(Nd,1);
diff --git a/tests/Test_dwf_cg_prec.cc b/tests/Test_dwf_cg_prec.cc
index 28848ff9..d429ee92 100644
--- a/tests/Test_dwf_cg_prec.cc
+++ b/tests/Test_dwf_cg_prec.cc
@@ -22,7 +22,7 @@ int main (int argc, char ** argv)
 
   const int Ls=8;
 
-  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
   GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
   GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
   GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
diff --git a/tests/Test_dwf_cg_schur.cc b/tests/Test_dwf_cg_schur.cc
index 8415c489..ac3989d9 100644
--- a/tests/Test_dwf_cg_schur.cc
+++ b/tests/Test_dwf_cg_schur.cc
@@ -22,7 +22,7 @@ int main (int argc, char ** argv)
 
   const int Ls=8;
 
-  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
   GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
   GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
   GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
diff --git a/tests/Test_dwf_cg_unprec.cc b/tests/Test_dwf_cg_unprec.cc
index f4f63c50..055e1bef 100644
--- a/tests/Test_dwf_cg_unprec.cc
+++ b/tests/Test_dwf_cg_unprec.cc
@@ -22,7 +22,7 @@ int main (int argc, char ** argv)
 
   const int Ls=8;
 
-  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
   GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
   GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
   GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
diff --git a/tests/Test_dwf_cr_unprec.cc b/tests/Test_dwf_cr_unprec.cc
index 2cf7065c..c2b75853 100644
--- a/tests/Test_dwf_cr_unprec.cc
+++ b/tests/Test_dwf_cr_unprec.cc
@@ -25,7 +25,7 @@ int main (int argc, char ** argv)
 
   const int Ls=8;
 
-  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
   GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
   GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
   GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
diff --git a/tests/Test_dwf_even_odd.cc b/tests/Test_dwf_even_odd.cc
index 5ec73212..3ad01076 100644
--- a/tests/Test_dwf_even_odd.cc
+++ b/tests/Test_dwf_even_odd.cc
@@ -26,7 +26,7 @@ int main (int argc, char ** argv)
 
 
   const int Ls=8;
-  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
   GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
   GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
   GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
diff --git a/tests/Test_dwf_fpgcr.cc b/tests/Test_dwf_fpgcr.cc
index a0d6ee96..969377a0 100644
--- a/tests/Test_dwf_fpgcr.cc
+++ b/tests/Test_dwf_fpgcr.cc
@@ -25,7 +25,7 @@ int main (int argc, char ** argv)
 
   const int Ls=8;
 
-  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplexF::Nsimd()),GridDefaultMpi());
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
   GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
   GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
   GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
diff --git a/tests/Test_gamma.cc b/tests/Test_gamma.cc
index f4f507a7..1f9cb491 100644
--- a/tests/Test_gamma.cc
+++ b/tests/Test_gamma.cc
@@ -14,7 +14,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplex::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
      
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
diff --git a/tests/Test_lie_generators.cc b/tests/Test_lie_generators.cc
index d36e463e..588baeda 100644
--- a/tests/Test_lie_generators.cc
+++ b/tests/Test_lie_generators.cc
@@ -15,7 +15,7 @@ int main (int argc, char ** argv)
 
   std::vector<int> latt({4,4,4,8});
   GridCartesian * grid = SpaceTimeGrid::makeFourDimGrid(latt, 
-							GridDefaultSimd(Nd,vComplexF::Nsimd()),
+							GridDefaultSimd(Nd,vComplex::Nsimd()),
 							GridDefaultMpi());
   
   GridRedBlackCartesian * rbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(grid);
diff --git a/tests/Test_main.cc b/tests/Test_main.cc
index 85450047..295b1981 100644
--- a/tests/Test_main.cc
+++ b/tests/Test_main.cc
@@ -27,7 +27,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplex::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
 
   latt_size.resize(4);
diff --git a/tests/Test_multishift_sqrt.cc b/tests/Test_multishift_sqrt.cc
index 1ecc1fde..3d69fdf1 100644
--- a/tests/Test_multishift_sqrt.cc
+++ b/tests/Test_multishift_sqrt.cc
@@ -56,7 +56,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   GridCartesian *grid = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), 
-						       GridDefaultSimd(Nd,vComplexF::Nsimd()),
+						       GridDefaultSimd(Nd,vComplex::Nsimd()),
 						       GridDefaultMpi());
 
   double     lo=0.001;
diff --git a/tests/Test_quenched_update.cc b/tests/Test_quenched_update.cc
index cb58144b..e419d002 100644
--- a/tests/Test_quenched_update.cc
+++ b/tests/Test_quenched_update.cc
@@ -14,7 +14,7 @@ int main (int argc, char ** argv)
 
   std::vector<int> latt({8,8,8,8});
   GridCartesian * grid = SpaceTimeGrid::makeFourDimGrid(latt, 
-							GridDefaultSimd(Nd,vComplexF::Nsimd()),
+							GridDefaultSimd(Nd,vComplex::Nsimd()),
 							GridDefaultMpi());
   
   GridRedBlackCartesian * rbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(grid);
diff --git a/tests/Test_rng.cc b/tests/Test_rng.cc
index 1731b740..9cd05a22 100644
--- a/tests/Test_rng.cc
+++ b/tests/Test_rng.cc
@@ -9,7 +9,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplex::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
      
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
diff --git a/tests/Test_rng_fixed.cc b/tests/Test_rng_fixed.cc
index c836c93f..a4171401 100644
--- a/tests/Test_rng_fixed.cc
+++ b/tests/Test_rng_fixed.cc
@@ -9,7 +9,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplex::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
      
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
diff --git a/tests/Test_simd.cc b/tests/Test_simd.cc
index f4858d06..800001be 100644
--- a/tests/Test_simd.cc
+++ b/tests/Test_simd.cc
@@ -185,7 +185,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd(4,vComplexF::Nsimd());
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplex::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
     
   GridCartesian     Grid(latt_size,simd_layout,mpi_layout);
diff --git a/tests/Test_stencil.cc b/tests/Test_stencil.cc
index 9770ed7c..3f6bb4fe 100644
--- a/tests/Test_stencil.cc
+++ b/tests/Test_stencil.cc
@@ -10,7 +10,7 @@ int main (int argc, char ** argv)
 
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
 
   double volume = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
diff --git a/tests/Test_wilson_cg_prec.cc b/tests/Test_wilson_cg_prec.cc
index 6365ee32..c59699e4 100644
--- a/tests/Test_wilson_cg_prec.cc
+++ b/tests/Test_wilson_cg_prec.cc
@@ -22,7 +22,7 @@ int main (int argc, char ** argv)
 
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
   GridCartesian               Grid(latt_size,simd_layout,mpi_layout);
   GridRedBlackCartesian     RBGrid(latt_size,simd_layout,mpi_layout);
diff --git a/tests/Test_wilson_cg_schur.cc b/tests/Test_wilson_cg_schur.cc
index 56abcd4a..1484a6a8 100644
--- a/tests/Test_wilson_cg_schur.cc
+++ b/tests/Test_wilson_cg_schur.cc
@@ -22,7 +22,7 @@ int main (int argc, char ** argv)
 
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
   GridCartesian               Grid(latt_size,simd_layout,mpi_layout);
   GridRedBlackCartesian     RBGrid(latt_size,simd_layout,mpi_layout);
diff --git a/tests/Test_wilson_cg_unprec.cc b/tests/Test_wilson_cg_unprec.cc
index 4e42af42..5984cac9 100644
--- a/tests/Test_wilson_cg_unprec.cc
+++ b/tests/Test_wilson_cg_unprec.cc
@@ -21,7 +21,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
   GridCartesian               Grid(latt_size,simd_layout,mpi_layout);
   GridRedBlackCartesian     RBGrid(latt_size,simd_layout,mpi_layout);
diff --git a/tests/Test_wilson_cr_unprec.cc b/tests/Test_wilson_cr_unprec.cc
index 1aa04632..468f0d5a 100644
--- a/tests/Test_wilson_cr_unprec.cc
+++ b/tests/Test_wilson_cr_unprec.cc
@@ -22,7 +22,7 @@ int main (int argc, char ** argv)
 
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
   GridCartesian               Grid(latt_size,simd_layout,mpi_layout);
   GridRedBlackCartesian     RBGrid(latt_size,simd_layout,mpi_layout);
diff --git a/tests/Test_wilson_even_odd.cc b/tests/Test_wilson_even_odd.cc
index 91fbb90f..90d15e99 100644
--- a/tests/Test_wilson_even_odd.cc
+++ b/tests/Test_wilson_even_odd.cc
@@ -21,7 +21,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   std::vector<int> latt_size   = GridDefaultLatt();
-  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexF::Nsimd());
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
   GridCartesian               Grid(latt_size,simd_layout,mpi_layout);
   GridRedBlackCartesian     RBGrid(latt_size,simd_layout,mpi_layout);
@@ -55,7 +55,9 @@ int main (int argc, char ** argv)
   Umu=zero;
   for(int nn=0;nn<Nd;nn++){
     random(pRNG,U[nn]);
+    std::cout<<"U[nn]"<<norm2(U[nn])<<std::endl;
     PokeIndex<LorentzIndex>(Umu,U[nn],nn);
+    std::cout<<"Umu"<<norm2(Umu)<<std::endl;
   }
 
   RealD mass=0.1;

From 0aec35bfc0cb3e401ae482e5973a735ce8adc55a Mon Sep 17 00:00:00 2001
From: paboyle <paboyle@ph.ed.ac.uk>
Date: Wed, 1 Jul 2015 22:46:53 +0100
Subject: [PATCH 373/429] Modified memory bw test to display word size

---
 benchmarks/Benchmark_memory_bandwidth.cc | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/benchmarks/Benchmark_memory_bandwidth.cc b/benchmarks/Benchmark_memory_bandwidth.cc
index 5abdc6a3..5b436001 100644
--- a/benchmarks/Benchmark_memory_bandwidth.cc
+++ b/benchmarks/Benchmark_memory_bandwidth.cc
@@ -20,7 +20,7 @@ int main (int argc, char ** argv)
   std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
   
   std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "= Benchmarking fused AXPY bandwidth"<<std::endl;
+  std::cout << "= Benchmarking fused AXPY bandwidth ; sizeof(Real) "<<sizeof(Real)<<std::endl;
   std::cout << "===================================================================================================="<<std::endl;
   std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s"<<"\t\t"<<"Gflop/s"<<std::endl;
   std::cout << "----------------------------------------------------------"<<std::endl;

From c0ea404a7cc3b0f1e7341c06630fa0091a5c6635 Mon Sep 17 00:00:00 2001
From: paboyle <paboyle@ph.ed.ac.uk>
Date: Wed, 1 Jul 2015 22:47:33 +0100
Subject: [PATCH 374/429] Temporarily disable gmp dependency simply because
 Cray XC30's I'm benchmarking have a downlevel gmp version that chokes on
 ::max_align_t where gmp had a bug as far as I recall.

---
 lib/algorithms/approx/Remez.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/algorithms/approx/Remez.h b/lib/algorithms/approx/Remez.h
index 336382d2..e6a973df 100644
--- a/lib/algorithms/approx/Remez.h
+++ b/lib/algorithms/approx/Remez.h
@@ -15,7 +15,7 @@
 #ifndef INCLUDED_ALG_REMEZ_H
 #define INCLUDED_ALG_REMEZ_H
 
-#include <algorithms/approx/bigfloat.h>
+#include <algorithms/approx/bigfloat_double.h>
 
 #define JMAX 10000 //Maximum number of iterations of Newton's approximation
 #define SUM_MAX 10 // Maximum number of terms in exponential

From d4c4ce49fc41959cb4128c79a54d70b8fb2b94e4 Mon Sep 17 00:00:00 2001
From: paboyle <paboyle@ph.ed.ac.uk>
Date: Wed, 1 Jul 2015 22:48:58 +0100
Subject: [PATCH 375/429] More xc30 config commansd

---
 scripts/configure-commands | 16 ++++++++++++++--
 1 file changed, 14 insertions(+), 2 deletions(-)

diff --git a/scripts/configure-commands b/scripts/configure-commands
index 538f3661..4cb4b44f 100755
--- a/scripts/configure-commands
+++ b/scripts/configure-commands
@@ -16,10 +16,16 @@ echo -e $RED ==============================
 echo -e $YELLOW
 
 case $WD in
-g++-sse4)
+g++-avx)
+  CXX=g++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11" LIBS="-lgmp -lmpfr" --enable-comms=none
+  ;;
+g++-avx-openmp)
+  CXX=g++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -std=c++11" LIBS="-fopenmp -lgmp -lmpfr" --enable-comms=none
+  ;;
+g++5-sse4)
   CXX=g++-5 ../../configure --enable-simd=SSE4 CXXFLAGS="-msse4 -O3 -std=c++11" LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;
-g++-avx)
+g++5-avx)
   CXX=g++-5 ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11" LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;
 icpc-avx)
@@ -52,6 +58,12 @@ CXX=clang++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -std=c
 clang-avx-openmp)
 CXX=clang-omp++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -std=c++11" LDFLAGS="-fopenmp"  LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;
+clang-xc30)
+CXX=$HOME/Clang/install/bin/clang++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -std=c++11 -I/opt/gcc/4.9.2/snos/include/g++/x86_64-suse-linux/ -I/opt/gcc/4.9.2/snos/include/g++/ " LDFLAGS=""  LIBS="-lgmp -lmpfr" --enable-comms=none
+  ;;
+clang-xc30-openmp)
+CXX=$HOME/Clang/install/bin/clang++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -fopenmp -O3 -std=c++11 -I/opt/gcc/4.9.2/snos/include/g++/x86_64-suse-linux/ -I/opt/gcc/4.9.2/snos/include/g++/ " LDFLAGS="-fopenmp"  LIBS="-lgmp -lmpfr" --enable-comms=none
+  ;;
 clang-avx2-openmp)
 CXX=clang-omp++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -fopenmp -O3 -std=c++11" LDFLAGS="-fopenmp"  LIBS="-lgmp -lmpfr" --enable-comms=none
   ;;

From 4cbcc7fd2346cdad7d44ebf395ea2607a0104d31 Mon Sep 17 00:00:00 2001
From: paboyle <paboyle@ph.ed.ac.uk>
Date: Wed, 1 Jul 2015 22:50:13 +0100
Subject: [PATCH 376/429] Some useful XC30 commands

---
 scripts/configure-cray | 10 ++++++++++
 scripts/cray-modules   |  2 ++
 2 files changed, 12 insertions(+)
 create mode 100755 scripts/configure-cray
 create mode 100644 scripts/cray-modules

diff --git a/scripts/configure-cray b/scripts/configure-cray
new file mode 100755
index 00000000..db581493
--- /dev/null
+++ b/scripts/configure-cray
@@ -0,0 +1,10 @@
+#!/bin/bash
+DIRS="g++-avx-openmp g++-avx clang-xc30 clang-xc30-openmp"
+
+for D in $DIRS
+do
+  mkdir -p builds/$D
+  cd builds/$D
+  ../../scripts/configure-commands $D
+  cd ../..
+done
diff --git a/scripts/cray-modules b/scripts/cray-modules
new file mode 100644
index 00000000..95de2b0b
--- /dev/null
+++ b/scripts/cray-modules
@@ -0,0 +1,2 @@
+module swap PrgEnv-cray PrgEnv-intel
+module swap intel/14.0.4.211 intel/15.0.2.164

From 84ec7c40cd7786dbe914559f1440ce473cf22cc6 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 1 Jul 2015 23:43:57 +0100
Subject: [PATCH 377/429] Cleaning up the recursion for traceIndex<n> after the
 changes the enable G++ to compile it again.

---
 lib/Make.inc                       |   2 +-
 lib/{tensors => Old}/Tensor_peek.h |   0
 lib/{tensors => Old}/Tensor_poke.h |   0
 lib/Tensors.h                      |   5 +-
 lib/tensors/Tensor_index.h         | 321 ++++++++++++++++++++++++++++
 lib/tensors/Tensor_trace.h         | 330 +----------------------------
 lib/tensors/Tensor_transpose.h     |   2 +
 7 files changed, 331 insertions(+), 329 deletions(-)
 rename lib/{tensors => Old}/Tensor_peek.h (100%)
 rename lib/{tensors => Old}/Tensor_poke.h (100%)
 create mode 100644 lib/tensors/Tensor_index.h

diff --git a/lib/Make.inc b/lib/Make.inc
index 91a06fcb..a1fe777e 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/BfmHDCG.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_peek.h ./tensors/Tensor_poke.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/BfmHDCG.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./Old/Tensor_peek.h ./Old/Tensor_poke.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_index.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
 
 CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/tensors/Tensor_peek.h b/lib/Old/Tensor_peek.h
similarity index 100%
rename from lib/tensors/Tensor_peek.h
rename to lib/Old/Tensor_peek.h
diff --git a/lib/tensors/Tensor_poke.h b/lib/Old/Tensor_poke.h
similarity index 100%
rename from lib/tensors/Tensor_poke.h
rename to lib/Old/Tensor_poke.h
diff --git a/lib/Tensors.h b/lib/Tensors.h
index 90c9d7a8..402fb9e3 100644
--- a/lib/Tensors.h
+++ b/lib/Tensors.h
@@ -8,11 +8,12 @@
 #include <tensors/Tensor_outer.h>
 #include <tensors/Tensor_transpose.h>
 #include <tensors/Tensor_trace.h>
+#include <tensors/Tensor_index.h>
 #include <tensors/Tensor_Ta.h>
 #include <tensors/Tensor_determinant.h>
 #include <tensors/Tensor_exp.h>
-#include <tensors/Tensor_peek.h>
-#include <tensors/Tensor_poke.h>
+//#include <tensors/Tensor_peek.h>
+//#include <tensors/Tensor_poke.h>
 #include <tensors/Tensor_reality.h>
 #include <tensors/Tensor_unary.h>
 #include <tensors/Tensor_extract_merge.h>
diff --git a/lib/tensors/Tensor_index.h b/lib/tensors/Tensor_index.h
new file mode 100644
index 00000000..842f3e1d
--- /dev/null
+++ b/lib/tensors/Tensor_index.h
@@ -0,0 +1,321 @@
+#ifndef GRID_TENSOR_INDEX_H
+#define GRID_TENSOR_INDEX_H
+
+////////////////////////////////////////////////////////////////////////////////////////
+// Recursion for trace, transpose, peek, poke a specific index
+////////////////////////////////////////////////////////////////////////////////////////
+// Allow trace to recurse if vector, but never terminate on a vector
+// trace of a different index can distribute across the vector index in a replicated way
+// but we do not trace a vector index.
+
+namespace Grid {
+
+  /* Needed?
+template<int Level> inline ComplexF traceIndex(const ComplexF arg) { return arg;}
+template<int Level> inline ComplexD traceIndex(const ComplexD arg) { return arg;}
+template<int Level> inline RealF traceIndex(const RealF arg) { return arg;}
+template<int Level> inline RealD traceIndex(const RealD arg) { return arg;}
+  */
+template<int Level> 
+class TensorIndexRecursion {
+
+ public:
+  template<class vtype>
+  static auto traceIndex(const iScalar<vtype> arg) ->  iScalar<decltype(TensorIndexRecursion<Level-1>::traceIndex(arg._internal))> 
+  {
+    iScalar<decltype(TensorIndexRecursion<Level-1>::traceIndex(arg._internal))> ret;
+    ret._internal = TensorIndexRecursion<Level-1>::traceIndex(arg._internal);
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto traceIndex(const iVector<vtype,N> arg) ->  iVector<decltype(TensorIndexRecursion<Level-1>::traceIndex(arg._internal[0])),N> 
+  {
+    iVector<decltype(TensorIndexRecursion<Level-1>::traceIndex(arg._internal[0])),N> ret;
+    for(int i=0;i<N;i++){
+      ret._internal[i] = TensorIndexRecursion<Level-1>::traceIndex(arg._internal[i]);
+    }
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto traceIndex(const iMatrix<vtype,N> arg) ->  iMatrix<decltype(TensorIndexRecursion<Level-1>::traceIndex(arg._internal[0][0])),N> 
+  {
+    iMatrix<decltype(TensorIndexRecursion<Level-1>::traceIndex(arg._internal[0][0])),N> ret;
+    for(int i=0;i<N;i++){
+    for(int j=0;j<N;j++){
+      ret._internal[i][j] = TensorIndexRecursion<Level-1>::traceIndex(arg._internal[i][j]);
+    }}
+    return ret;
+  }
+  ////////////////////////////////////////////
+  // Recursion for peeking a specific index
+  ////////////////////////////////////////////
+  template<class vtype>
+  static auto peekIndex(const iScalar<vtype> arg,int i) ->  iScalar<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal,0))> 
+  {
+    iScalar<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal,0))> ret;
+    ret._internal = TensorIndexRecursion<Level-1>::peekIndex(arg._internal,i);
+    return ret;
+  }
+  template<class vtype>
+  static auto peekIndex(const iScalar<vtype> arg,int i,int j) ->  iScalar<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal,0,0))> 
+  {
+    iScalar<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal,0,0))> ret;
+    ret._internal = TensorIndexRecursion<Level-1>::peekIndex(arg._internal,i,j);
+    return ret;
+  }
+
+  template<class vtype,int N>
+  static auto peekIndex(const iVector<vtype,N> arg,int ii) ->  iVector<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0],0)),N> 
+  {
+    iVector<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0],0)),N> ret;
+    for(int i=0;i<N;i++){
+      ret._internal[i] = TensorIndexRecursion<Level-1>::peekIndex(arg._internal[i],ii);
+    }
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto peekIndex(const iVector<vtype,N> arg,int ii,int jj) ->  iVector<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0],0,0)),N> 
+  {
+    iVector<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0],0,0)),N> ret;
+    for(int i=0;i<N;i++){
+      ret._internal[i] = TensorIndexRecursion<Level-1>::peekIndex(arg._internal[i],ii,jj);
+    }
+    return ret;
+  }
+
+  template<class vtype,int N>
+  static auto peekIndex(const iMatrix<vtype,N> arg,int ii) ->  iMatrix<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0][0],0)),N> 
+  {
+    iMatrix<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0][0],0)),N> ret;
+    for(int i=0;i<N;i++){
+    for(int j=0;j<N;j++){
+      ret._internal[i][j] = TensorIndexRecursion<Level-1>::peekIndex(arg._internal[i][j],ii);
+    }}
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto peekIndex(const iMatrix<vtype,N> arg,int ii,int jj) ->  iMatrix<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0][0],0,0)),N> 
+  {
+    iMatrix<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0][0],0,0)),N> ret;
+    for(int i=0;i<N;i++){
+    for(int j=0;j<N;j++){
+      ret._internal[i][j] = TensorIndexRecursion<Level-1>::peekIndex(arg._internal[i][j],ii,jj);
+    }}
+    return ret;
+  }
+  ////////////////////////////////////////////
+  // Recursion for poking a specific index
+  ////////////////////////////////////////////
+
+  template<class vtype> inline static 
+    void pokeIndex(iScalar<vtype> &ret, const iScalar<decltype(TensorIndexRecursion<Level-1>::peekIndex(ret._internal,0))> &arg, int i)
+    {
+      TensorIndexRecursion<Level-1>::pokeIndex(ret._internal,arg._internal,i);
+    }
+  template<class vtype> inline static 
+    void pokeIndex(iScalar<vtype> &ret, const iScalar<decltype(TensorIndexRecursion<Level-1>::peekIndex(ret._internal,0,0))> &arg, int i,int j)
+    {
+      TensorIndexRecursion<Level-1>::pokeIndex(ret._internal,arg._internal,i,j);
+    }
+
+  template<class vtype,int N> inline static 
+    void pokeIndex(iVector<vtype,N> &ret, const iVector<decltype(TensorIndexRecursion<Level-1>::peekIndex(ret._internal,0)),N> &arg, int i)
+    {
+      for(int ii=0;ii<N;ii++){
+	TensorIndexRecursion<Level-1>::pokeIndex(ret._internal[ii],arg._internal[ii],i);
+      }
+    }
+  template<class vtype,int N> inline static 
+    void pokeIndex(iVector<vtype,N> &ret, const iVector<decltype(TensorIndexRecursion<Level-1>::peekIndex(ret._internal,0)),N> &arg, int i,int j)
+    {
+      for(int ii=0;ii<N;ii++){
+	TensorIndexRecursion<Level-1>::pokeIndex(ret._internal[ii],arg._internal[ii],i,j);
+      }
+    }
+
+  template<class vtype,int N> inline static 
+    void pokeIndex(iMatrix<vtype,N> &ret, const iMatrix<decltype(TensorIndexRecursion<Level-1>::peekIndex(ret._internal,0)),N> &arg, int i)
+    {
+      for(int ii=0;ii<N;ii++){
+      for(int jj=0;jj<N;jj++){
+	TensorIndexRecursion<Level-1>::pokeIndex(ret._internal[ii][jj],arg._internal[ii][jj],i);
+      }}
+    }
+  template<class vtype,int N> inline static 
+    void pokeIndex(iMatrix<vtype,N> &ret, const iMatrix<decltype(TensorIndexRecursion<Level-1>::peekIndex(ret._internal,0)),N> &arg, int i,int j)
+    {
+      for(int ii=0;ii<N;ii++){
+      for(int jj=0;jj<N;jj++){
+	TensorIndexRecursion<Level-1>::pokeIndex(ret._internal[ii][jj],arg._internal[ii][jj],i,j);
+      }}
+    }
+
+  ////////////////////////////////////////////
+  // Recursion for transposing a specific index
+  ////////////////////////////////////////////
+  template<class vtype>
+  static auto transposeIndex(const iScalar<vtype> arg) ->  iScalar<vtype> 
+  {
+    iScalar<vtype> ret;
+    ret._internal = TensorIndexRecursion<Level-1>::transposeIndex(arg._internal);
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto transposeIndex(const iVector<vtype,N> arg) ->  iVector<vtype,N> 
+  {
+    iVector<vtype,N> ret;
+    for(int i=0;i<N;i++){
+      ret._internal[i] = TensorIndexRecursion<Level-1>::transposeIndex(arg._internal[i]);
+    }
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto transposeIndex(const iMatrix<vtype,N> arg) ->  iMatrix<vtype,N> 
+  {
+    iMatrix<vtype,N> ret;
+    for(int i=0;i<N;i++){
+    for(int j=0;j<N;j++){
+      ret._internal[i][j] = TensorIndexRecursion<Level-1>::transposeIndex(arg._internal[i][j]);
+    }}
+    return ret;
+  }
+};
+
+////////////////////////////
+// strip const & ref quali's
+////////////////////////////
+#define RemoveCRV(a) typename std::remove_const<typename std::remove_reference<decltype(a)>::type>::type
+template<>
+class TensorIndexRecursion<0> {
+ public:
+
+  /////////////////////////////////////////
+  // Ends recursion for trace (scalar/vector/matrix)
+  /////////////////////////////////////////
+  template<class vtype>
+  static auto traceIndex(const iScalar<vtype> arg) ->  iScalar<RemoveCRV(arg._internal)>
+  {
+    iScalar<RemoveCRV(arg._internal)> ret;
+    ret._internal = arg._internal;
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto traceIndex(const iVector<vtype,N> arg) ->  iScalar<RemoveCRV(arg._internal[0])>
+  {
+    iScalar<RemoveCRV(arg._internal[0])> ret;
+    ret._internal=zero;
+    for(int i=0;i<N;i++){
+      ret._internal = ret._internal+ arg._internal[i];
+    }
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto traceIndex(const iMatrix<vtype,N> arg) ->  iScalar<RemoveCRV(arg._internal[0][0])> 
+  {
+    iScalar<RemoveCRV(arg._internal[0][0])> ret;
+    ret=zero;
+    for(int i=0;i<N;i++){
+      ret._internal = ret._internal+arg._internal[i][i];
+    }
+    return ret;
+  }
+  /////////////////////////////////////////
+  // Ends recursion for transpose scalar/matrix ; no way to terminate on vector
+  /////////////////////////////////////////
+  template<class vtype>
+  static auto transposeIndex(const iScalar<vtype> arg) ->  iScalar<vtype>
+  {
+    iScalar<vtype> ret;
+    ret._internal = arg._internal;
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto transposeIndex(const iMatrix<vtype,N> arg)  ->  iMatrix<vtype,N> 
+  {
+    iMatrix<vtype,N> ret;
+    ret=zero;
+    for(int i=0;i<N;i++){
+    for(int j=0;j<N;j++){
+      ret._internal[i][j] = ret._internal[i][j]+arg._internal[i][j];
+    }}
+    return ret;
+  }
+  ////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+  // End recursion for peeking a specific index; single index on vector, double index on matrix
+  ////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+  template<class vtype,int N>
+  static auto peekIndex(const iVector<vtype,N> arg,int ii) ->  iScalar<vtype> 
+  {
+    iScalar<vtype> ret;
+    ret._internal = arg._internal[ii];
+    return ret;
+  }
+  template<class vtype,int N>
+  static auto peekIndex(const iMatrix<vtype,N> arg,int ii,int jj) ->  iScalar<vtype>
+  {
+    iScalar<vtype> ret;
+    ret._internal = arg._internal[ii][jj];
+    return ret;
+  }
+  // Vector poke, one index
+  template<class vtype,int N> inline static 
+    void pokeIndex(iVector<vtype,N> &ret, const iScalar<vtype> &arg,int i)
+    {
+      ret._internal[i] = arg._internal;
+    }
+  // Matrix poke two indices
+  template<class vtype,int N> inline static 
+    void pokeIndex(iMatrix<vtype,N> &ret, const iScalar<vtype> &arg,int i,int j)
+    {
+      ret._internal[i][j] = arg._internal;
+    }
+
+};
+
+////////////////////////////////////////////////////////////////////////////////////////////////////////
+// External wrappers
+////////////////////////////////////////////////////////////////////////////////////////////////////////
+
+template<int Level,class vtype> inline auto traceIndex (const vtype &arg) -> RemoveCRV(TensorIndexRecursion<Level>::traceIndex(arg))
+{
+  RemoveCRV(TensorIndexRecursion<Level>::traceIndex(arg)) ret;
+  ret=TensorIndexRecursion<Level>::traceIndex(arg);
+  return ret;
+}
+template<int Level,class vtype> inline auto transposeIndex (const vtype &arg) -> RemoveCRV(TensorIndexRecursion<Level>::transposeIndex(arg))
+{
+  RemoveCRV(TensorIndexRecursion<Level>::transposeIndex(arg)) ret;
+  ret=TensorIndexRecursion<Level>::transposeIndex(arg);
+  return ret;
+}
+
+template<int Level,class vtype> inline auto peekIndex (const vtype &arg,int i) -> RemoveCRV(TensorIndexRecursion<Level>::peekIndex(arg,0))
+{
+  RemoveCRV(TensorIndexRecursion<Level>::peekIndex(arg,0)) ret;
+  ret=TensorIndexRecursion<Level>::peekIndex(arg,i);
+  return ret;
+}
+template<int Level,class vtype> inline auto peekIndex (const vtype &arg,int i,int j) -> RemoveCRV(TensorIndexRecursion<Level>::peekIndex(arg,0,0))
+{
+  RemoveCRV(TensorIndexRecursion<Level>::peekIndex(arg,0,0)) ret;
+  ret=TensorIndexRecursion<Level>::peekIndex(arg,i,j);
+  return ret;
+}
+
+template<int Level,class vtype> inline 
+void pokeIndex (vtype &ret,const decltype(TensorIndexRecursion<Level>::peekIndex(ret,0)) &arg,int i) 
+{
+  TensorIndexRecursion<Level>::pokeIndex(ret,arg,i);
+}
+
+template<int Level,class vtype> inline 
+void pokeIndex (vtype &ret,const decltype(TensorIndexRecursion<Level>::peekIndex(ret,0,0)) &arg,int i,int j) 
+{
+  TensorIndexRecursion<Level>::pokeIndex(ret,arg,i,j);
+}
+
+
+#undef RemoveCRV
+
+}
+#endif
diff --git a/lib/tensors/Tensor_trace.h b/lib/tensors/Tensor_trace.h
index bcde670c..4cf5edc2 100644
--- a/lib/tensors/Tensor_trace.h
+++ b/lib/tensors/Tensor_trace.h
@@ -1,8 +1,10 @@
 #ifndef GRID_MATH_TRACE_H
 #define GRID_MATH_TRACE_H
 namespace Grid {
+
 //////////////////////////////////////////////////////////////////
-// Traces: both all indices and a specific index 
+// Traces: both all indices and a specific index. Indices must be
+// either scalar or matrix
 /////////////////////////////////////////////////////////////////
 
 inline ComplexF trace( const ComplexF &arg){    return arg;}
@@ -10,11 +12,6 @@ inline ComplexD trace( const ComplexD &arg){    return arg;}
 inline RealF trace( const RealF &arg){    return arg;}
 inline RealD trace( const RealD &arg){    return arg;}
 
-template<int Level> inline ComplexF traceIndex(const ComplexF arg) { return arg;}
-template<int Level> inline ComplexD traceIndex(const ComplexD arg) { return arg;}
-template<int Level> inline RealF traceIndex(const RealF arg) { return arg;}
-template<int Level> inline RealD traceIndex(const RealD arg) { return arg;}
-
 template<class vtype,int N>
 inline auto trace(const iMatrix<vtype,N> &arg) -> iScalar<decltype(trace(arg._internal[0][0]))>
 {
@@ -25,6 +22,7 @@ inline auto trace(const iMatrix<vtype,N> &arg) -> iScalar<decltype(trace(arg._in
     }
     return ret;
 }
+
 template<class vtype>
 inline auto trace(const iScalar<vtype> &arg) -> iScalar<decltype(trace(arg._internal))>
 {
@@ -33,326 +31,6 @@ inline auto trace(const iScalar<vtype> &arg) -> iScalar<decltype(trace(arg._inte
     return ret;
 }
 
-/////////////////////////////////////
-// Alternate implementation .. count 
-// indices from left. Annoying but
-// I'm working around compiler bugs in gcc and earlier icpc
-/////////////////////////////////////
-
-// Allow to recurse if vector, but never terminate on a vector
-// ltrace of a different index can distribute across the vector index in a replicated way
-// but we do not ltrace a vector index.
-template<int Level> 
-class TensorIndexRecursion {
-
- public:
-  ////////////////////////////////////////////
-  // Recursion for tracing a specific index
-  ////////////////////////////////////////////
-  template<class vtype>
-  static auto traceIndex(const iScalar<vtype> arg) ->  iScalar<decltype(TensorIndexRecursion<Level-1>::traceIndex(arg._internal))> 
-  {
-    iScalar<decltype(TensorIndexRecursion<Level-1>::traceIndex(arg._internal))> ret;
-    ret._internal = TensorIndexRecursion<Level-1>::traceIndex(arg._internal);
-    return ret;
-  }
-  template<class vtype,int N>
-  static auto traceIndex(const iVector<vtype,N> arg) ->  iVector<decltype(TensorIndexRecursion<Level-1>::traceIndex(arg._internal[0])),N> 
-  {
-    iVector<decltype(TensorIndexRecursion<Level-1>::traceIndex(arg._internal[0])),N> ret;
-    for(int i=0;i<N;i++){
-      ret._internal[i] = TensorIndexRecursion<Level-1>::traceIndex(arg._internal[i]);
-    }
-    return ret;
-  }
-  template<class vtype,int N>
-  static auto traceIndex(const iMatrix<vtype,N> arg) ->  iMatrix<decltype(TensorIndexRecursion<Level-1>::traceIndex(arg._internal[0][0])),N> 
-  {
-    iMatrix<decltype(TensorIndexRecursion<Level-1>::traceIndex(arg._internal[0][0])),N> ret;
-    for(int i=0;i<N;i++){
-    for(int j=0;j<N;j++){
-      ret._internal[i][j] = TensorIndexRecursion<Level-1>::traceIndex(arg._internal[i][j]);
-    }}
-    return ret;
-  }
-  ////////////////////////////////////////////
-  // Recursion for peeking a specific index
-  ////////////////////////////////////////////
-  template<class vtype>
-  static auto peekIndex(const iScalar<vtype> arg,int i) ->  iScalar<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal,0))> 
-  {
-    iScalar<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal,0))> ret;
-    ret._internal = TensorIndexRecursion<Level-1>::peekIndex(arg._internal,i);
-    return ret;
-  }
-  template<class vtype>
-  static auto peekIndex(const iScalar<vtype> arg,int i,int j) ->  iScalar<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal,0,0))> 
-  {
-    iScalar<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal,0,0))> ret;
-    ret._internal = TensorIndexRecursion<Level-1>::peekIndex(arg._internal,i,j);
-    return ret;
-  }
-
-  template<class vtype,int N>
-  static auto peekIndex(const iVector<vtype,N> arg,int ii) ->  iVector<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0],0)),N> 
-  {
-    iVector<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0],0)),N> ret;
-    for(int i=0;i<N;i++){
-      ret._internal[i] = TensorIndexRecursion<Level-1>::peekIndex(arg._internal[i],ii);
-    }
-    return ret;
-  }
-  template<class vtype,int N>
-  static auto peekIndex(const iVector<vtype,N> arg,int ii,int jj) ->  iVector<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0],0,0)),N> 
-  {
-    iVector<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0],0,0)),N> ret;
-    for(int i=0;i<N;i++){
-      ret._internal[i] = TensorIndexRecursion<Level-1>::peekIndex(arg._internal[i],ii,jj);
-    }
-    return ret;
-  }
-
-  template<class vtype,int N>
-  static auto peekIndex(const iMatrix<vtype,N> arg,int ii) ->  iMatrix<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0][0],0)),N> 
-  {
-    iMatrix<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0][0],0)),N> ret;
-    for(int i=0;i<N;i++){
-    for(int j=0;j<N;j++){
-      ret._internal[i][j] = TensorIndexRecursion<Level-1>::peekIndex(arg._internal[i][j],ii);
-    }}
-    return ret;
-  }
-  template<class vtype,int N>
-  static auto peekIndex(const iMatrix<vtype,N> arg,int ii,int jj) ->  iMatrix<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0][0],0,0)),N> 
-  {
-    iMatrix<decltype(TensorIndexRecursion<Level-1>::peekIndex(arg._internal[0][0],0,0)),N> ret;
-    for(int i=0;i<N;i++){
-    for(int j=0;j<N;j++){
-      ret._internal[i][j] = TensorIndexRecursion<Level-1>::peekIndex(arg._internal[i][j],ii,jj);
-    }}
-    return ret;
-  }
-  ////////////////////////////////////////////
-  // Recursion for poking a specific index
-  ////////////////////////////////////////////
-
-  template<class vtype> inline static 
-    void pokeIndex(iScalar<vtype> &ret, const iScalar<decltype(TensorIndexRecursion<Level-1>::peekIndex(ret._internal,0))> &arg, int i)
-    {
-      TensorIndexRecursion<Level-1>::pokeIndex(ret._internal,arg._internal,i);
-    }
-  template<class vtype> inline static 
-    void pokeIndex(iScalar<vtype> &ret, const iScalar<decltype(TensorIndexRecursion<Level-1>::peekIndex(ret._internal,0,0))> &arg, int i,int j)
-    {
-      TensorIndexRecursion<Level-1>::pokeIndex(ret._internal,arg._internal,i,j);
-    }
-
-  template<class vtype,int N> inline static 
-    void pokeIndex(iVector<vtype,N> &ret, const iVector<decltype(TensorIndexRecursion<Level-1>::peekIndex(ret._internal,0)),N> &arg, int i)
-    {
-      for(int ii=0;ii<N;ii++){
-	TensorIndexRecursion<Level-1>::pokeIndex(ret._internal[ii],arg._internal[ii],i);
-      }
-    }
-  template<class vtype,int N> inline static 
-    void pokeIndex(iVector<vtype,N> &ret, const iVector<decltype(TensorIndexRecursion<Level-1>::peekIndex(ret._internal,0)),N> &arg, int i,int j)
-    {
-      for(int ii=0;ii<N;ii++){
-	TensorIndexRecursion<Level-1>::pokeIndex(ret._internal[ii],arg._internal[ii],i,j);
-      }
-    }
-
-  template<class vtype,int N> inline static 
-    void pokeIndex(iMatrix<vtype,N> &ret, const iMatrix<decltype(TensorIndexRecursion<Level-1>::peekIndex(ret._internal,0)),N> &arg, int i)
-    {
-      for(int ii=0;ii<N;ii++){
-      for(int jj=0;jj<N;jj++){
-	TensorIndexRecursion<Level-1>::pokeIndex(ret._internal[ii][jj],arg._internal[ii][jj],i);
-      }}
-    }
-  template<class vtype,int N> inline static 
-    void pokeIndex(iMatrix<vtype,N> &ret, const iMatrix<decltype(TensorIndexRecursion<Level-1>::peekIndex(ret._internal,0)),N> &arg, int i,int j)
-    {
-      for(int ii=0;ii<N;ii++){
-      for(int jj=0;jj<N;jj++){
-	TensorIndexRecursion<Level-1>::pokeIndex(ret._internal[ii][jj],arg._internal[ii][jj],i,j);
-      }}
-    }
-
-  ////////////////////////////////////////////
-  // Recursion for transposing a specific index
-  ////////////////////////////////////////////
-  template<class vtype>
-  static auto transposeIndex(const iScalar<vtype> arg) ->  iScalar<vtype> 
-  {
-    iScalar<vtype> ret;
-    ret._internal = TensorIndexRecursion<Level-1>::transposeIndex(arg._internal);
-    return ret;
-  }
-  template<class vtype,int N>
-  static auto transposeIndex(const iVector<vtype,N> arg) ->  iVector<vtype,N> 
-  {
-    iVector<vtype,N> ret;
-    for(int i=0;i<N;i++){
-      ret._internal[i] = TensorIndexRecursion<Level-1>::transposeIndex(arg._internal[i]);
-    }
-    return ret;
-  }
-  template<class vtype,int N>
-  static auto transposeIndex(const iMatrix<vtype,N> arg) ->  iMatrix<vtype,N> 
-  {
-    iMatrix<vtype,N> ret;
-    for(int i=0;i<N;i++){
-    for(int j=0;j<N;j++){
-      ret._internal[i][j] = TensorIndexRecursion<Level-1>::transposeIndex(arg._internal[i][j]);
-    }}
-    return ret;
-  }
-};
-
-////////////////////////////
-// strip const & ref quali's
-////////////////////////////
-#define RemoveCRV(a) typename std::remove_const<typename std::remove_reference<decltype(a)>::type>::type
-template<>
-class TensorIndexRecursion<0> {
- public:
-
-  /////////////////////////////////////////
-  // Ends recursion for trace (scalar/vector/matrix)
-  /////////////////////////////////////////
-  template<class vtype>
-  static auto traceIndex(const iScalar<vtype> arg) ->  iScalar<RemoveCRV(arg._internal)>
-  {
-    iScalar<RemoveCRV(arg._internal)> ret;
-    ret._internal = arg._internal;
-    return ret;
-  }
-  template<class vtype,int N>
-  static auto traceIndex(const iVector<vtype,N> arg) ->  iScalar<RemoveCRV(arg._internal[0])>
-  {
-    iScalar<RemoveCRV(arg._internal[0])> ret;
-    ret._internal=zero;
-    for(int i=0;i<N;i++){
-      ret._internal = ret._internal+ arg._internal[i];
-    }
-    return ret;
-  }
-  template<class vtype,int N>
-  static auto traceIndex(const iMatrix<vtype,N> arg) ->  iScalar<RemoveCRV(arg._internal[0][0])> 
-  {
-    iScalar<RemoveCRV(arg._internal[0][0])> ret;
-    ret=zero;
-    for(int i=0;i<N;i++){
-      ret._internal = ret._internal+arg._internal[i][i];
-    }
-    return ret;
-  }
-  /////////////////////////////////////////
-  // Ends recursion for transpose scalar/vector/matrix
-  /////////////////////////////////////////
-  template<class vtype>
-  static auto transposeIndex(const iScalar<vtype> arg) ->  iScalar<vtype>
-  {
-    iScalar<vtype> ret;
-    ret._internal = arg._internal;
-    return ret;
-  }
-  template<class vtype,int N>
-  static auto transposeIndex(const iVector<vtype,N> arg) ->  iVector<vtype,N>
-  {
-    iScalar<vtype> ret;
-    ret._internal=zero;
-    for(int i=0;i<N;i++){
-      ret._internal = ret._internal+ arg._internal[i];
-    }
-    return ret;
-  }
-  template<class vtype,int N>
-  static auto transposeIndex(const iMatrix<vtype,N> arg)  ->  iMatrix<vtype,N> 
-  {
-    iScalar<vtype> ret;
-    ret=zero;
-    for(int i=0;i<N;i++){
-      ret._internal = ret._internal+arg._internal[i][i];
-    }
-    return ret;
-  }
-  ////////////////////////////////////////////////////////////////////////////////////////////////////////////////
-  // End recursion for peeking a specific index; single index on vector, double index on matrix
-  ////////////////////////////////////////////////////////////////////////////////////////////////////////////////
-  template<class vtype,int N>
-  static auto peekIndex(const iVector<vtype,N> arg,int ii) ->  iScalar<vtype> 
-  {
-    iScalar<vtype> ret;
-    ret._internal = arg._internal[ii];
-    return ret;
-  }
-  template<class vtype,int N>
-  static auto peekIndex(const iMatrix<vtype,N> arg,int ii,int jj) ->  iScalar<vtype>
-  {
-    iScalar<vtype> ret;
-    ret._internal = arg._internal[ii][jj];
-    return ret;
-  }
-  // Vector poke, one index
-  template<class vtype,int N> inline static 
-    void pokeIndex(iVector<vtype,N> &ret, const iScalar<vtype> &arg,int i)
-    {
-      ret._internal[i] = arg._internal;
-    }
-  // Matrix poke two indices
-  template<class vtype,int N> inline static 
-    void pokeIndex(iMatrix<vtype,N> &ret, const iScalar<vtype> &arg,int i,int j)
-    {
-      ret._internal[i][j] = arg._internal;
-    }
-
-};
-
-////////////////////////////////////////////////////////////////////////////////////////////////////////
-// External wrappers
-////////////////////////////////////////////////////////////////////////////////////////////////////////
-
-template<int Level,class vtype> inline auto traceIndex (const vtype &arg) -> RemoveCRV(TensorIndexRecursion<Level>::traceIndex(arg))
-{
-  RemoveCRV(TensorIndexRecursion<Level>::traceIndex(arg)) ret;
-  ret=TensorIndexRecursion<Level>::traceIndex(arg);
-  return ret;
-}
-template<int Level,class vtype> inline auto transposeIndex (const vtype &arg) -> RemoveCRV(TensorIndexRecursion<Level>::transposeIndex(arg))
-{
-  RemoveCRV(TensorIndexRecursion<Level>::transposeIndex(arg)) ret;
-  ret=TensorIndexRecursion<Level>::transposeIndex(arg);
-  return ret;
-}
-template<int Level,class vtype> inline auto peekIndex (const vtype &arg,int i) -> RemoveCRV(TensorIndexRecursion<Level>::peekIndex(arg,0))
-{
-  RemoveCRV(TensorIndexRecursion<Level>::peekIndex(arg,0)) ret;
-  ret=TensorIndexRecursion<Level>::peekIndex(arg,i);
-  return ret;
-}
-template<int Level,class vtype> inline auto peekIndex (const vtype &arg,int i,int j) -> RemoveCRV(TensorIndexRecursion<Level>::peekIndex(arg,0,0))
-{
-  RemoveCRV(TensorIndexRecursion<Level>::peekIndex(arg,0,0)) ret;
-  ret=TensorIndexRecursion<Level>::peekIndex(arg,i,j);
-  return ret;
-}
-
-template<int Level,class vtype> inline 
-void pokeIndex (vtype &ret,const decltype(TensorIndexRecursion<Level>::peekIndex(ret,0)) &arg,int i) 
-{
-  TensorIndexRecursion<Level>::pokeIndex(ret,arg,i);
-}
-
-template<int Level,class vtype> inline 
-void pokeIndex (vtype &ret,const decltype(TensorIndexRecursion<Level>::peekIndex(ret,0,0)) &arg,int i,int j) 
-{
-  TensorIndexRecursion<Level>::pokeIndex(ret,arg,i,j);
-}
-
-
-#undef RemoveCRV
 
 }
 #endif
diff --git a/lib/tensors/Tensor_transpose.h b/lib/tensors/Tensor_transpose.h
index 76f80b1b..3798f4bf 100644
--- a/lib/tensors/Tensor_transpose.h
+++ b/lib/tensors/Tensor_transpose.h
@@ -59,6 +59,7 @@ template<class vtype>
 // Transpose a specific index; instructive to compare this style of recursion termination
 // to that of adj; which is easiers?
 ////////////////////////////////////////////////////////////////////////////////////////////
+#if 0
 template<int Level,class vtype,int N> inline 
   typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,Level>::value, iMatrix<vtype,N> >::type 
 transposeIndex (const iMatrix<vtype,N> &arg)
@@ -96,6 +97,7 @@ transposeIndex (const iScalar<vtype> &arg)
 {
   return arg;
 }
+#endif
 
 }
 #endif

From a666e66e36ec01d2f9a1e4e580541b1f98c5364c Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 2 Jul 2015 02:03:09 +0100
Subject: [PATCH 378/429] No compile fix

---
 lib/simd/Grid_avx512.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/simd/Grid_avx512.h b/lib/simd/Grid_avx512.h
index 10741174..b874f15c 100644
--- a/lib/simd/Grid_avx512.h
+++ b/lib/simd/Grid_avx512.h
@@ -331,7 +331,7 @@ namespace Grid {
     case 0 : conv.f = _mm512_permute4f128_ps(conv.f,(_MM_PERM_ENUM)_MM_SHUFFLE(1,0,3,2)); break;
     default: assert(0); break;
     }
-    y=conv.v;
+    y.v=conv.v;
   };
   
   // Function name aliases

From 9655d43017b3cbabda565ae1b8e02f1ca51f090a Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Fri, 3 Jul 2015 16:51:41 +0900
Subject: [PATCH 379/429] Skeleton of HMC/Integrators

---
 INSTALL                                    |  2 +-
 configure                                  | 13 ++-
 lib/GridConfig.h                           |  6 ++
 lib/Make.inc                               |  4 +-
 lib/qcd/QCD.h                              |  3 +
 lib/qcd/hmc/.dirstamp                      |  0
 lib/qcd/hmc/HMC.cc                         | 98 ++++++++++++++++++++++
 lib/qcd/hmc/HMC.h                          | 50 +++++++++++
 lib/qcd/hmc/integrators/.dirstamp          |  0
 lib/qcd/hmc/integrators/Integrator_base.cc | 22 +++++
 lib/qcd/hmc/integrators/Integrator_base.h  | 50 +++++++++++
 lib/qcd/utils/SUn.h                        | 20 +++++
 tests/Make.inc                             | 10 +--
 13 files changed, 263 insertions(+), 15 deletions(-)
 create mode 100644 lib/qcd/hmc/.dirstamp
 create mode 100644 lib/qcd/hmc/HMC.cc
 create mode 100644 lib/qcd/hmc/HMC.h
 create mode 100644 lib/qcd/hmc/integrators/.dirstamp
 create mode 100644 lib/qcd/hmc/integrators/Integrator_base.cc
 create mode 100644 lib/qcd/hmc/integrators/Integrator_base.h

diff --git a/INSTALL b/INSTALL
index 80a61507..f812f5a2 120000
--- a/INSTALL
+++ b/INSTALL
@@ -1 +1 @@
-/opt/local/share/automake-1.15/INSTALL
\ No newline at end of file
+/usr/share/automake-1.14/INSTALL
\ No newline at end of file
diff --git a/configure b/configure
index 06ca42d5..935f1018 100755
--- a/configure
+++ b/configure
@@ -2577,7 +2577,7 @@ test -n "$target_alias" &&
     NONENONEs,x,x, &&
   program_prefix=${target_alias}-
 
-am__api_version='1.15'
+am__api_version='1.14'
 
 # Find a good install program.  We prefer a C program (faster),
 # so one script is as good as another.  But avoid the broken or
@@ -2749,8 +2749,8 @@ test "$program_suffix" != NONE &&
 ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
 program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
 
-# Expand $ac_aux_dir to an absolute path.
-am_aux_dir=`cd "$ac_aux_dir" && pwd`
+# expand $ac_aux_dir to an absolute path
+am_aux_dir=`cd $ac_aux_dir && pwd`
 
 if test x"${MISSING+set}" != xset; then
   case $am_aux_dir in
@@ -2769,7 +2769,7 @@ else
 $as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
 fi
 
-if test x"${install_sh+set}" != xset; then
+if test x"${install_sh}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -3097,8 +3097,8 @@ MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 mkdir_p='$(MKDIR_P)'
 
-# We need awk for the "check" target (and possibly the TAP driver).  The
-# system "awk" is bad on some platforms.
+# We need awk for the "check" target.  The system "awk" is bad on
+# some platforms.
 # Always define AMTAR for backward compatibility.  Yes, it's still used
 # in the wild :-(  We should find a proper way to deprecate it ...
 AMTAR='$${TAR-tar}'
@@ -3157,7 +3157,6 @@ END
 fi
 
 
-
 ac_config_headers="$ac_config_headers lib/GridConfig.h"
 
 # Check whether --enable-silent-rules was given.
diff --git a/lib/GridConfig.h b/lib/GridConfig.h
index c9dea1cb..f3a1be6a 100644
--- a/lib/GridConfig.h
+++ b/lib/GridConfig.h
@@ -19,6 +19,12 @@
 /* GRID_COMMS_NONE */
 #define GRID_COMMS_NONE 1
 
+/* GRID_DEFAULT_PRECISION is DOUBLE */
+/* #undef GRID_DEFAULT_PRECISION_DOUBLE */
+
+/* GRID_DEFAULT_PRECISION is SINGLE */
+#define GRID_DEFAULT_PRECISION_SINGLE 1
+
 /* Support Altivec instructions */
 /* #undef HAVE_ALTIVEC */
 
diff --git a/lib/Make.inc b/lib/Make.inc
index a1fe777e..18a29228 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/BfmHDCG.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./Old/Tensor_peek.h ./Old/Tensor_poke.h ./parallelIO/NerscIO.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_index.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
+HFILES=./Cshift.h ./simd/Grid_avx.h ./simd/Grid_vector_types.h ./simd/Grid_sse4.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_vector_unops.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./Tensors.h ./Algorithms.h ./communicator/Communicator_base.h ./lattice/Lattice_rng.h ./lattice/Lattice_reduction.h ./lattice/Lattice_transfer.h ./lattice/Lattice_unary.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_comparison.h ./lattice/Lattice_overload.h ./lattice/Lattice_reality.h ./lattice/Lattice_local.h ./lattice/Lattice_conformable.h ./lattice/Lattice_where.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_ET.h ./lattice/Lattice_transpose.h ./lattice/Lattice_trace.h ./Stencil.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_exp.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_class.h ./tensors/Tensor_logical.h ./tensors/Tensor_transpose.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_reality.h ./tensors/Tensor_trace.h ./tensors/Tensor_index.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_outer.h ./tensors/Tensor_inner.h ./tensors/Tensor_traits.h ./tensors/Tensor_Ta.h ./tensors/Tensor_unary.h ./tensors/Tensor_determinant.h ./tensors/Tensor_arith.h ./tensors/Tensor_extract_merge.h ./Communicator.h ./Cartesian.h ./parallelIO/NerscIO.h ./qcd/QCD.h ./qcd/hmc/integrators/Integrator_base.h ./qcd/hmc/HMC.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/LinalgUtils.h ./qcd/utils/CovariantCshift.h ./qcd/utils/WilsonLoops.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/spin/TwoSpinor.h ./qcd/spin/Dirac.h ./cshift/Cshift_common.h ./cshift/Cshift_none.h ./cshift/Cshift_mpi.h ./Simd.h ./GridConfig.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_red_black.h ./cartesian/Cartesian_full.h ./AlignedAllocator.h ./Lattice.h ./Old/Tensor_poke.h ./Old/Tensor_peek.h ./Threads.h ./Grid.h ./algorithms/Preconditioner.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/AdefGeneric.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Zolotarev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Remez.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./algorithms/CoarsenedMatrix.h ./stencil/Lebesgue.h
 
-CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
+CCFILES=./qcd/hmc/integrators/Integrator_base.cc ./qcd/hmc/HMC.cc ./qcd/utils/SpaceTimeGrid.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/spin/Dirac.cc ./GridInit.cc ./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/qcd/QCD.h b/lib/qcd/QCD.h
index 0e841d8e..cf8ff6d8 100644
--- a/lib/qcd/QCD.h
+++ b/lib/qcd/QCD.h
@@ -413,5 +413,8 @@ namespace QCD {
 #include <qcd/utils/WilsonLoops.h>
 #include <qcd/utils/SUn.h>
 #include <qcd/action/Actions.h>
+#include <qcd/hmc/integrators/Integrator_base.h>
+#include <qcd/hmc/HMC.h>
+
 
 #endif
diff --git a/lib/qcd/hmc/.dirstamp b/lib/qcd/hmc/.dirstamp
new file mode 100644
index 00000000..e69de29b
diff --git a/lib/qcd/hmc/HMC.cc b/lib/qcd/hmc/HMC.cc
new file mode 100644
index 00000000..32ef8ca0
--- /dev/null
+++ b/lib/qcd/hmc/HMC.cc
@@ -0,0 +1,98 @@
+#include <Grid.h>
+
+namespace Grid{
+  namespace QCD{
+
+    HMCparameters::HMCparameters(){
+	// FIXME fill this constructor  now just default values
+	  
+	////////////////////////////// Default values
+	Nsweeps             = 10;
+	TotalSweeps         = 10;
+	ThermalizationSteps = 0;
+	StartingConfig      = 0;
+	SaveInterval        = 1;
+	Filename_prefix     = "Conf_";
+	/////////////////////////////////
+	  
+      }
+
+
+    /////////////////////////////////////////////////////////////////
+
+    HybridMonteCarlo::HybridMonteCarlo(GridSerialRNG& R):RNG(R){
+	//FIXME
+      }
+
+
+    void HybridMonteCarlo::evolve(LatticeColourMatrix& Uin)const{
+      Real DeltaH;
+      Real timer;
+      
+      // Thermalizations
+      for(int iter=1; iter <= Params.ThermalizationSteps; ++iter){
+	std::cout << "-- # Thermalization step = "<< iter <<  "\n";
+	
+	DeltaH = evolve_step(Uin);
+	
+	std::cout<< "[Timing] Trajectory time (s) : "<< timer/1000.0 << "\n";
+	std::cout<< "dH = "<< DeltaH << "\n";
+	
+	// Update matrix
+	//Uin = md_->get_U();  //accept every time
+      }
+
+      // Actual updates
+      for(int iter=Params.StartingConfig; 
+	  iter < Params.Nsweeps+Params.StartingConfig; ++iter){
+	std::cout << "-- # Sweep = "<< iter <<  "\n";
+		
+	DeltaH = evolve_step(Uin);
+		
+	if(metropolis_test(DeltaH))  {};//Uin = md_->get_U();
+	// need sync?	
+      }
+
+      
+    }
+
+
+    RealD HybridMonteCarlo::evolve_step(LatticeColourMatrix& Uin)const{
+      /*
+      md_->init(Uin,RNG);     // set U and initialize P and phi's 
+      RealD H0 = md_->S();     // current state            
+      std::cout<<"Total H_before = "<< H0 << "\n";
+      
+      md_->integrator();
+      
+      RealD H1 = md_->calc_H();     // updated state            
+      std::cout<<"Total H_after = "<< H1 << "\n";
+      
+      return (H1-H0);
+      */
+      return 0;
+    }
+
+
+
+
+    bool HybridMonteCarlo::metropolis_test(const RealD DeltaH)const{
+      RealD rn_test;
+      RealD prob = std::exp(-DeltaH);
+      random(RNG,rn_test);
+      
+      std::cout<< "--------------------------------------------\n";
+      std::cout<< "dH = "<<DeltaH << "  Random = "<< rn_test 
+	       << "\nAcc. Probability = " << ((prob<1.0)? prob: 1.0)<< "   ";
+      
+      if((prob >1.0) || (rn_test <= prob)){       // accepted
+	std::cout <<"-- ACCEPTED\n";
+	return true;
+      } else {                               // rejected
+	std::cout <<"-- REJECTED\n";
+	return false;
+      }
+    }
+
+  }
+}
diff --git a/lib/qcd/hmc/HMC.h b/lib/qcd/hmc/HMC.h
new file mode 100644
index 00000000..e25bb3ed
--- /dev/null
+++ b/lib/qcd/hmc/HMC.h
@@ -0,0 +1,50 @@
+//--------------------------------------------------------------------
+/*! @file HMC.h
+ * @brief Declaration of classes for HybridMonteCarlo update
+ *
+ * @author Guido Cossu
+ */
+//--------------------------------------------------------------------
+#ifndef HMC_INCLUDED
+#define HMC_INCLUDED
+
+#include <string>
+#include <memory>
+
+namespace Grid{
+  namespace QCD{
+    
+    struct HMCparameters{
+      Integer Nsweeps; /* @brief Number of sweeps in this run */
+      Integer TotalSweeps; /* @brief If provided, the total number of sweeps */
+      Integer ThermalizationSteps;
+      Integer StartingConfig;
+      Integer SaveInterval; //Setting to 0 does not save configurations
+      std::string Filename_prefix; // To save configurations
+      
+      HMCparameters();
+    };
+    
+ 
+    class HybridMonteCarlo{
+      const HMCparameters Params;
+      GridSerialRNG& RNG; 
+      // FIXME need the integrator
+      
+      bool metropolis_test(const RealD DeltaH)const;
+      RealD evolve_step(LatticeColourMatrix&)const;
+      
+      
+    public:
+      HybridMonteCarlo(GridSerialRNG&);
+      ~HybridMonteCarlo(){};
+
+      void evolve(LatticeColourMatrix& Uin)const;
+
+    };
+    
+  }// QCD
+}// Grid
+
+
+#endif 
diff --git a/lib/qcd/hmc/integrators/.dirstamp b/lib/qcd/hmc/integrators/.dirstamp
new file mode 100644
index 00000000..e69de29b
diff --git a/lib/qcd/hmc/integrators/Integrator_base.cc b/lib/qcd/hmc/integrators/Integrator_base.cc
new file mode 100644
index 00000000..5fc89482
--- /dev/null
+++ b/lib/qcd/hmc/integrators/Integrator_base.cc
@@ -0,0 +1,22 @@
+/*! 
+  @file Integrator_base.cc
+  @brief utilities for MD including funcs to generate initial HMC momentum
+ */
+
+
+#include <Grid.h>
+
+static const double sq3i = 1.0/sqrt(3.0);
+
+namespace Grid{
+  namespace QCD{
+
+    
+
+    void MDutils::generate_momenta_su3(LatticeColourMatrix& P,GridParallelRNG& pRNG){
+      SU3::GaussianLieAlgebraMatrix(pRNG, P);
+    }
+
+
+  }
+}
diff --git a/lib/qcd/hmc/integrators/Integrator_base.h b/lib/qcd/hmc/integrators/Integrator_base.h
new file mode 100644
index 00000000..f1d6f711
--- /dev/null
+++ b/lib/qcd/hmc/integrators/Integrator_base.h
@@ -0,0 +1,50 @@
+//--------------------------------------------------------------------
+/*! @file Integrator_base.h
+ * @brief Declaration of classes for the abstract Molecular Dynamics integrator
+ *
+ * @author Guido Cossu
+ */
+//--------------------------------------------------------------------
+
+#ifndef INTEGRATOR_INCLUDED
+#define INTEGRATOR_INCLUDED
+
+class Action;
+class RandNum;
+class Observer;
+
+typedef std::vector<Action*> ActionLevel;
+typedef std::vector<ActionLevel> ActionSet;
+typedef std::vector<Observer*> ObserverList;
+
+/*! @brief Abstract base class for Molecular Dynamics management */
+
+namespace Grid{
+  namespace QCD{
+    
+    class Integrator{
+    private:
+      virtual void update_P(int lv,double ep) = 0;
+      virtual void update_U(double ep) = 0;
+      
+      virtual void register_observers() = 0;
+      virtual void notify_observers() = 0;
+      
+    public:
+      virtual ~Integrator(){}
+      virtual void init(const LatticeColourMatrix&,
+			const GridParallelRNG& RNG)=0;
+      virtual double S()const =0;
+      virtual void integrate(int level) =0;
+      virtual const LatticeColourMatrix get_U() const =0;
+      
+      void generate_momenta(LatticeColourMatrix& P,const RandNum& rand);
+    };
+    
+    namespace MDutils{
+      void generate_momenta_su3(LatticeColourMatrix& P,GridParallelRNG& RNG);
+    }
+    
+  }
+}
+#endif//INTEGRATOR_INCLUDED
diff --git a/lib/qcd/utils/SUn.h b/lib/qcd/utils/SUn.h
index 69de3f9f..9a34f3e3 100644
--- a/lib/qcd/utils/SUn.h
+++ b/lib/qcd/utils/SUn.h
@@ -543,6 +543,26 @@ Note that in step D setting B ~ X - A and using B in place of A in step E will g
     taExp(lie,out);
   }
 
+  static void GaussianLieAlgebraMatrix(GridParallelRNG     &pRNG,LatticeMatrix &out,double scale=1.0){
+    GridBase *grid = out._grid;
+    LatticeComplex ca (grid);
+    LatticeMatrix  lie(grid);
+    LatticeMatrix  la (grid);
+    Complex ci(0.0,scale);
+    Matrix ta;
+
+    out=zero;
+    for(int a=0;a<generators();a++){
+
+      gaussian(pRNG,ca); 
+      generator(a,ta);
+
+      la=ci*ca*ta;
+      out = lie+la; // e^{i la ta}
+    }
+  }
+
+
   static void HotConfiguration(GridParallelRNG &pRNG,LatticeGaugeField &out){
     LatticeMatrix Umu(out._grid);
     for(int mu=0;mu<Nd;mu++){
diff --git a/tests/Make.inc b/tests/Make.inc
index 4ad511bd..4592cd36 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,9 +1,5 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
-
-
-Test_GaugeAction_SOURCES=Test_GaugeAction.cc
-Test_GaugeAction_LDADD=-lGrid
+bin_PROGRAMS = Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_GaugeAction Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
 
 
 Test_cayley_cg_SOURCES=Test_cayley_cg.cc
@@ -78,6 +74,10 @@ Test_gamma_SOURCES=Test_gamma.cc
 Test_gamma_LDADD=-lGrid
 
 
+Test_GaugeAction_SOURCES=Test_GaugeAction.cc
+Test_GaugeAction_LDADD=-lGrid
+
+
 Test_lie_generators_SOURCES=Test_lie_generators.cc
 Test_lie_generators_LDADD=-lGrid
 

From 30c9dc473dcfa751364e6563055c2d42220b36c3 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Sat, 4 Jul 2015 02:43:14 +0900
Subject: [PATCH 380/429] More progress in the HMC construction

---
 lib/Make.inc                                  |  2 +-
 lib/qcd/QCD.h                                 | 10 +--
 .../{gauge/GaugeActionBase.h => ActionBase.h} | 11 +--
 lib/qcd/action/Actions.h                      |  9 ++
 lib/qcd/action/gauge/WilsonGaugeAction.h      | 52 +++++++-----
 lib/qcd/hmc/HMC.cc                            | 76 +----------------
 lib/qcd/hmc/HMC.h                             | 84 +++++++++++++++++--
 lib/qcd/hmc/integrators/Integrator_base.cc    | 16 ++--
 lib/qcd/hmc/integrators/Integrator_base.h     | 67 +++++++++++----
 lib/qcd/utils/SUn.h                           |  4 +-
 tests/Make.inc                                |  6 +-
 tests/Test_hmc.cc                             | 36 ++++++++
 12 files changed, 234 insertions(+), 139 deletions(-)
 rename lib/qcd/action/{gauge/GaugeActionBase.h => ActionBase.h} (65%)
 create mode 100644 tests/Test_hmc.cc

diff --git a/lib/Make.inc b/lib/Make.inc
index 18a29228..85ed84c2 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./Cshift.h ./simd/Grid_avx.h ./simd/Grid_vector_types.h ./simd/Grid_sse4.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_vector_unops.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./Tensors.h ./Algorithms.h ./communicator/Communicator_base.h ./lattice/Lattice_rng.h ./lattice/Lattice_reduction.h ./lattice/Lattice_transfer.h ./lattice/Lattice_unary.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_comparison.h ./lattice/Lattice_overload.h ./lattice/Lattice_reality.h ./lattice/Lattice_local.h ./lattice/Lattice_conformable.h ./lattice/Lattice_where.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_ET.h ./lattice/Lattice_transpose.h ./lattice/Lattice_trace.h ./Stencil.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_exp.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_class.h ./tensors/Tensor_logical.h ./tensors/Tensor_transpose.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_reality.h ./tensors/Tensor_trace.h ./tensors/Tensor_index.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_outer.h ./tensors/Tensor_inner.h ./tensors/Tensor_traits.h ./tensors/Tensor_Ta.h ./tensors/Tensor_unary.h ./tensors/Tensor_determinant.h ./tensors/Tensor_arith.h ./tensors/Tensor_extract_merge.h ./Communicator.h ./Cartesian.h ./parallelIO/NerscIO.h ./qcd/QCD.h ./qcd/hmc/integrators/Integrator_base.h ./qcd/hmc/HMC.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/LinalgUtils.h ./qcd/utils/CovariantCshift.h ./qcd/utils/WilsonLoops.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/gauge/GaugeActionBase.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/spin/TwoSpinor.h ./qcd/spin/Dirac.h ./cshift/Cshift_common.h ./cshift/Cshift_none.h ./cshift/Cshift_mpi.h ./Simd.h ./GridConfig.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_red_black.h ./cartesian/Cartesian_full.h ./AlignedAllocator.h ./Lattice.h ./Old/Tensor_poke.h ./Old/Tensor_peek.h ./Threads.h ./Grid.h ./algorithms/Preconditioner.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/AdefGeneric.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Zolotarev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Remez.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./algorithms/CoarsenedMatrix.h ./stencil/Lebesgue.h
+HFILES=./Cshift.h ./simd/Grid_avx.h ./simd/Grid_vector_types.h ./simd/Grid_sse4.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_vector_unops.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./Tensors.h ./Algorithms.h ./communicator/Communicator_base.h ./lattice/Lattice_rng.h ./lattice/Lattice_reduction.h ./lattice/Lattice_transfer.h ./lattice/Lattice_unary.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_comparison.h ./lattice/Lattice_overload.h ./lattice/Lattice_reality.h ./lattice/Lattice_local.h ./lattice/Lattice_conformable.h ./lattice/Lattice_where.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_ET.h ./lattice/Lattice_transpose.h ./lattice/Lattice_trace.h ./Stencil.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_exp.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_class.h ./tensors/Tensor_logical.h ./tensors/Tensor_transpose.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_reality.h ./tensors/Tensor_trace.h ./tensors/Tensor_index.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_outer.h ./tensors/Tensor_inner.h ./tensors/Tensor_traits.h ./tensors/Tensor_Ta.h ./tensors/Tensor_unary.h ./tensors/Tensor_determinant.h ./tensors/Tensor_arith.h ./tensors/Tensor_extract_merge.h ./Communicator.h ./Cartesian.h ./parallelIO/NerscIO.h ./qcd/QCD.h ./qcd/hmc/integrators/Integrator_base.h ./qcd/hmc/HMC.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/LinalgUtils.h ./qcd/utils/CovariantCshift.h ./qcd/utils/WilsonLoops.h ./qcd/action/ActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/spin/TwoSpinor.h ./qcd/spin/Dirac.h ./cshift/Cshift_common.h ./cshift/Cshift_none.h ./cshift/Cshift_mpi.h ./Simd.h ./GridConfig.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_red_black.h ./cartesian/Cartesian_full.h ./AlignedAllocator.h ./Lattice.h ./Old/Tensor_poke.h ./Old/Tensor_peek.h ./Threads.h ./Grid.h ./algorithms/Preconditioner.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/AdefGeneric.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Zolotarev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Remez.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./algorithms/CoarsenedMatrix.h ./stencil/Lebesgue.h
 
 CCFILES=./qcd/hmc/integrators/Integrator_base.cc ./qcd/hmc/HMC.cc ./qcd/utils/SpaceTimeGrid.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/spin/Dirac.cc ./GridInit.cc ./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/qcd/QCD.h b/lib/qcd/QCD.h
index cf8ff6d8..a318141d 100644
--- a/lib/qcd/QCD.h
+++ b/lib/qcd/QCD.h
@@ -303,35 +303,35 @@ namespace QCD {
 		      const Lattice<decltype(peekIndex<ColourIndex>(lhs._odata[0],0))> & rhs,
 		      int i)
     {
-      pokeIndex<ColourIndex>(lhs,rhs,i);
+      PokeIndex<ColourIndex>(lhs,rhs,i);
     }
     template<class vobj> 
       void pokeColour(Lattice<vobj> &lhs,
 		      const Lattice<decltype(peekIndex<ColourIndex>(lhs._odata[0],0,0))> & rhs,
 		      int i,int j)
     {
-      pokeIndex<ColourIndex>(lhs,rhs,i,j);
+      PokeIndex<ColourIndex>(lhs,rhs,i,j);
     }
     template<class vobj> 
       void pokeSpin(Lattice<vobj> &lhs,
 		      const Lattice<decltype(peekIndex<SpinIndex>(lhs._odata[0],0))> & rhs,
 		      int i)
     {
-      pokeIndex<SpinIndex>(lhs,rhs,i);
+      PokeIndex<SpinIndex>(lhs,rhs,i);
     }
     template<class vobj> 
       void pokeSpin(Lattice<vobj> &lhs,
 		      const Lattice<decltype(peekIndex<SpinIndex>(lhs._odata[0],0,0))> & rhs,
 		      int i,int j)
     {
-      pokeIndex<SpinIndex>(lhs,rhs,i,j);
+      PokeIndex<SpinIndex>(lhs,rhs,i,j);
     }
     template<class vobj> 
       void pokeLorentz(Lattice<vobj> &lhs,
 		      const Lattice<decltype(peekIndex<LorentzIndex>(lhs._odata[0],0))> & rhs,
 		      int i)
     {
-      pokeIndex<LorentzIndex>(lhs,rhs,i);
+      PokeIndex<LorentzIndex>(lhs,rhs,i);
     }
 
     //////////////////////////////////////////////
diff --git a/lib/qcd/action/gauge/GaugeActionBase.h b/lib/qcd/action/ActionBase.h
similarity index 65%
rename from lib/qcd/action/gauge/GaugeActionBase.h
rename to lib/qcd/action/ActionBase.h
index 55141965..e3b13f0e 100644
--- a/lib/qcd/action/gauge/GaugeActionBase.h
+++ b/lib/qcd/action/ActionBase.h
@@ -1,19 +1,20 @@
-#ifndef QCD_GAUGE_ACTION_BASE
-#define QCD_GAUGE_ACTION_BASE
+#ifndef QCD_ACTION_BASE
+#define QCD_ACTION_BASE
 namespace Grid {
 namespace QCD{
 
 template<class GaugeField>
-class GaugeActionBase { // derive this from TermInAction?
+class Action { 
 
  public:
+  virtual void  init(const GaugeField &U)                      = 0;
   virtual RealD S(const GaugeField &U)                        = 0;  // evaluate the action
   virtual void  deriv(const GaugeField &U,GaugeField & dSdU ) = 0;  // evaluate the action derivative
   virtual void  staple(const GaugeField &U,GaugeField & dSdU ) = 0;  // evaluate the action derivative
-  virtual void refresh(const GaugeField & ) {}; 
+  //virtual void  refresh(const GaugeField & ) {}                ; 
   // Boundary conditions?
   // Heatbath?
-  virtual ~GaugeActionBase() {};
+  virtual ~Action() {};
 };
 
 }}
diff --git a/lib/qcd/action/Actions.h b/lib/qcd/action/Actions.h
index 56b453c6..94809723 100644
--- a/lib/qcd/action/Actions.h
+++ b/lib/qcd/action/Actions.h
@@ -15,8 +15,17 @@
 ////////////////////////////////////////////
 // Abstract base interface
 ////////////////////////////////////////////
+#include <qcd/action/ActionBase.h>
+
 #include <qcd/action/fermion/FermionOperator.h>
 
+
+////////////////////////////////////////////
+// Gauge Actions
+////////////////////////////////////////////
+#include <qcd/action/gauge/WilsonGaugeAction.h>
+
+
 ////////////////////////////////////////////
 // Utility functions
 ////////////////////////////////////////////
diff --git a/lib/qcd/action/gauge/WilsonGaugeAction.h b/lib/qcd/action/gauge/WilsonGaugeAction.h
index 8ecb1f81..7726db80 100644
--- a/lib/qcd/action/gauge/WilsonGaugeAction.h
+++ b/lib/qcd/action/gauge/WilsonGaugeAction.h
@@ -1,24 +1,38 @@
 #ifndef QCD_WILSON_GAUGE_ACTION_H
 #define QCD_WILSON_GAUGE_ACTION_H
 
-////////////////////////////////////////////////////////////////////////
-// Wilson Gauge Action .. should I template the Nc etc..
-////////////////////////////////////////////////////////////////////////
-template<class GaugeField,class MatrixField>
-class WilsonGaugeAction : public GaugeActionBase<GaugeField> {
- public:
-  virtual RealD S(const GaugeField &U) {
-    return WilsonLoops<MatrixField,GaugeField>::sumPlaquette(U);
-  };
-  virtual RealD deriv(GaugeField &U,GaugeField & dSdU ) {
-    WilsonLoops<MatrixField,GaugeField>::Staple(dSdU,U,mu);
-  };
-  virtual void  staple(const MatrixField &stap,GaugeField & U,int mu ) {
-    WilsonLoops<MatrixField,GaugeField>::Staple(stap,U,mu);
-  };
-};
-
-
-#endif
+namespace Grid{
+  namespace QCD{
+    
+    ////////////////////////////////////////////////////////////////////////
+    // Wilson Gauge Action .. should I template the Nc etc..
+    ////////////////////////////////////////////////////////////////////////
+    template<class GaugeField,class MatrixField>
+      class WilsonGaugeAction : public Action<GaugeField> {
+    private:
+      RealD beta;
+    public:
+    WilsonGaugeAction(RealD b):beta(b){};
+      
+      virtual void init(const GaugeField &U) {//FIXME
+      };
+      
+      virtual RealD S(const GaugeField &U) {
+	return WilsonLoops<MatrixField,GaugeField>::sumPlaquette(U);
+      };
+      virtual void deriv(const GaugeField &U,GaugeField & dSdU) {
+	//FIXME loop on directions
+	MatrixField dSdU_mu(U._grid);
+	WilsonLoops<MatrixField,GaugeField>::Staple(dSdU_mu,U,0);
+      };
+      virtual void  staple(const GaugeField &stap,GaugeField & U) {
+	//FIXME loop on directions
+	MatrixField stap_mu(U._grid);
+	WilsonLoops<MatrixField,GaugeField>::Staple(stap_mu,U,0);
+      };
+    };
+    
+  }
+}
 
 #endif
diff --git a/lib/qcd/hmc/HMC.cc b/lib/qcd/hmc/HMC.cc
index 32ef8ca0..859d6646 100644
--- a/lib/qcd/hmc/HMC.cc
+++ b/lib/qcd/hmc/HMC.cc
@@ -18,81 +18,7 @@ namespace Grid{
       }
 
 
-    /////////////////////////////////////////////////////////////////
-
-    HybridMonteCarlo::HybridMonteCarlo(GridSerialRNG& R):RNG(R){
-	//FIXME
-      }
-
-
-    void HybridMonteCarlo::evolve(LatticeColourMatrix& Uin)const{
-      Real DeltaH;
-      Real timer;
-      
-      // Thermalizations
-      for(int iter=1; iter <= Params.ThermalizationSteps; ++iter){
-	std::cout << "-- # Thermalization step = "<< iter <<  "\n";
-	
-	DeltaH = evolve_step(Uin);
-	
-	std::cout<< "[Timing] Trajectory time (s) : "<< timer/1000.0 << "\n";
-	std::cout<< "dH = "<< DeltaH << "\n";
-	
-	// Update matrix
-	//Uin = md_->get_U();  //accept every time
-      }
-
-      // Actual updates
-      for(int iter=Params.StartingConfig; 
-	  iter < Params.Nsweeps+Params.StartingConfig; ++iter){
-	std::cout << "-- # Sweep = "<< iter <<  "\n";
-		
-	DeltaH = evolve_step(Uin);
-		
-	if(metropolis_test(DeltaH))  {};//Uin = md_->get_U();
-	// need sync?	
-      }
-
-      
-    }
-
-
-    RealD HybridMonteCarlo::evolve_step(LatticeColourMatrix& Uin)const{
-      /*
-      md_->init(Uin,RNG);     // set U and initialize P and phi's 
-      RealD H0 = md_->S();     // current state            
-      std::cout<<"Total H_before = "<< H0 << "\n";
-      
-      md_->integrator();
-      
-      RealD H1 = md_->calc_H();     // updated state            
-      std::cout<<"Total H_after = "<< H1 << "\n";
-      
-      return (H1-H0);
-      */
-      return 0;
-    }
-
-
-
-
-    bool HybridMonteCarlo::metropolis_test(const RealD DeltaH)const{
-      RealD rn_test;
-      RealD prob = std::exp(-DeltaH);
-      random(RNG,rn_test);
-      
-      std::cout<< "--------------------------------------------\n";
-      std::cout<< "dH = "<<DeltaH << "  Random = "<< rn_test 
-	       << "\nAcc. Probability = " << ((prob<1.0)? prob: 1.0)<< "   ";
-      
-      if((prob >1.0) || (rn_test <= prob)){       // accepted
-	std::cout <<"-- ACCEPTED\n";
-	return true;
-      } else {                               // rejected
-	std::cout <<"-- REJECTED\n";
-	return false;
-      }
-    }
+   
 
   }
 }
diff --git a/lib/qcd/hmc/HMC.h b/lib/qcd/hmc/HMC.h
index e25bb3ed..e4d8740c 100644
--- a/lib/qcd/hmc/HMC.h
+++ b/lib/qcd/hmc/HMC.h
@@ -25,21 +25,91 @@ namespace Grid{
       HMCparameters();
     };
     
- 
+    template <class IntegType> 
     class HybridMonteCarlo{
       const HMCparameters Params;
-      GridSerialRNG& RNG; 
-      // FIXME need the integrator
+      GridSerialRNG sRNG;
+      GridParallelRNG pRNG;
+      std::unique_ptr< Integrator<IntegType> > MD;
       
-      bool metropolis_test(const RealD DeltaH)const;
-      RealD evolve_step(LatticeColourMatrix&)const;
+
+      bool metropolis_test(const RealD DeltaH){
+	RealD rn_test;
+	RealD prob = std::exp(-DeltaH);
+	random(sRNG,rn_test);
+      
+	std::cout<< "--------------------------------------------\n";
+	std::cout<< "dH = "<<DeltaH << "  Random = "<< rn_test 
+		 << "\nAcc. Probability = " << ((prob<1.0)? prob: 1.0)<< "   ";
+      
+	if((prob >1.0) || (rn_test <= prob)){       // accepted
+	  std::cout <<"-- ACCEPTED\n";
+	  return true;
+	} else {                               // rejected
+	  std::cout <<"-- REJECTED\n";
+	  return false;
+	}
+      }
+
+      RealD evolve_step(LatticeColourMatrix& Uin){
+     
+	MD->init(Uin,pRNG);     // set U and initialize P and phi's 
+	RealD H0 = MD->S();     // current state            
+	std::cout<<"Total H_before = "<< H0 << "\n";
+      
+	MD->integrate(0);
+      
+	RealD H1 = MD->S();     // updated state            
+	std::cout<<"Total H_after = "<< H1 << "\n";
+      
+	return (H1-H0);
+      }
+
       
       
     public:
-      HybridMonteCarlo(GridSerialRNG&);
+      HybridMonteCarlo(Integrator<IntegType>& MolDyn, GridBase* grid):MD(&MolDyn),pRNG(grid){
+	//FIXME
+
+	// initialize RNGs
+	sRNG.SeedRandomDevice();
+	pRNG.SeedRandomDevice();
+      }
       ~HybridMonteCarlo(){};
 
-      void evolve(LatticeColourMatrix& Uin)const;
+
+
+      void evolve(LatticeColourMatrix& Uin){
+	Real DeltaH;
+	Real timer;
+      
+	// Thermalizations
+	for(int iter=1; iter <= Params.ThermalizationSteps; ++iter){
+	  std::cout << "-- # Thermalization step = "<< iter <<  "\n";
+	
+	  DeltaH = evolve_step(Uin);
+	
+	  std::cout<< "[Timing] Trajectory time (s) : "<< timer/1000.0 << "\n";
+	  std::cout<< "dH = "<< DeltaH << "\n";
+	
+	  // Update matrix
+	  Uin = MD->get_U();  //accept every time
+	}
+
+	// Actual updates
+	for(int iter=Params.StartingConfig; 
+	    iter < Params.Nsweeps+Params.StartingConfig; ++iter){
+	  std::cout << "-- # Sweep = "<< iter <<  "\n";
+		
+	  DeltaH = evolve_step(Uin);
+		
+	  if(metropolis_test(DeltaH))  Uin = MD->get_U();
+	  // need sync?	
+	}
+
+      
+      }
+
 
     };
     
diff --git a/lib/qcd/hmc/integrators/Integrator_base.cc b/lib/qcd/hmc/integrators/Integrator_base.cc
index 5fc89482..0a256444 100644
--- a/lib/qcd/hmc/integrators/Integrator_base.cc
+++ b/lib/qcd/hmc/integrators/Integrator_base.cc
@@ -3,18 +3,22 @@
   @brief utilities for MD including funcs to generate initial HMC momentum
  */
 
-
 #include <Grid.h>
 
-static const double sq3i = 1.0/sqrt(3.0);
-
 namespace Grid{
   namespace QCD{
 
-    
+    void MDutils::generate_momenta(LatticeLorentzColourMatrix& P,GridParallelRNG& pRNG){
+      // for future support of different groups
+      MDutils::generate_momenta_su3(P, pRNG);
+    }
 
-    void MDutils::generate_momenta_su3(LatticeColourMatrix& P,GridParallelRNG& pRNG){
-      SU3::GaussianLieAlgebraMatrix(pRNG, P);
+    void MDutils::generate_momenta_su3(LatticeLorentzColourMatrix& P,GridParallelRNG& pRNG){
+      LatticeColourMatrix Pmu(P._grid);
+      for(int mu=0;mu<Nd;mu++){
+	SU3::GaussianLieAlgebraMatrix(pRNG, Pmu);
+	pokeLorentz(P, Pmu, mu);
+      }
     }
 
 
diff --git a/lib/qcd/hmc/integrators/Integrator_base.h b/lib/qcd/hmc/integrators/Integrator_base.h
index f1d6f711..61abf118 100644
--- a/lib/qcd/hmc/integrators/Integrator_base.h
+++ b/lib/qcd/hmc/integrators/Integrator_base.h
@@ -9,40 +9,71 @@
 #ifndef INTEGRATOR_INCLUDED
 #define INTEGRATOR_INCLUDED
 
-class Action;
-class RandNum;
 class Observer;
 
-typedef std::vector<Action*> ActionLevel;
-typedef std::vector<ActionLevel> ActionSet;
-typedef std::vector<Observer*> ObserverList;
 
 /*! @brief Abstract base class for Molecular Dynamics management */
 
 namespace Grid{
   namespace QCD{
+
+    typedef Action<LatticeLorentzColourMatrix>*  ActPtr; // now force the same size as the rest of the code
+    typedef std::vector<ActPtr> ActionLevel;
+    typedef std::vector<ActionLevel> ActionSet;
+    typedef std::vector<Observer*> ObserverList;
     
+    
+    class Integrator2MN{
+      const double lambda = 0.1931833275037836;
+      void step (LatticeColourMatrix&, LatticeColourMatrix&, 
+		 int, std::vector<int>&);
+      
+    };
+    
+    class IntegratorLeapFrog{
+      void step (LatticeColourMatrix&, LatticeColourMatrix&,
+		 int, std::vector<int>&);
+    };
+    
+
+
+    template< class IntegratorPolicy >
     class Integrator{
     private:
-      virtual void update_P(int lv,double ep) = 0;
-      virtual void update_U(double ep) = 0;
+      int Nexp;
+      int MDsteps;  // number of outer steps
+      RealD trajL;  // trajectory length 
+      RealD stepsize;
+
+      const std::vector<int> Nrel; // relative steps per level
+      const ActionSet as;
+      ObserverList observers;
+      //      LatticeColourMatrix* const U;  // is shared among all actions - or use a singleton...
+      LatticeColourMatrix P;
       
-      virtual void register_observers() = 0;
-      virtual void notify_observers() = 0;
+      IntegratorPolicy TheIntegrator;// contains parameters too
+      void update_P(int lv,double ep);
+      void update_U(double ep);
+      
+      void register_observers();
+      void notify_observers();
+      void integrator_step(int level ,std::vector<Integer>& clock);
+
       
     public:
-      virtual ~Integrator(){}
-      virtual void init(const LatticeColourMatrix&,
-			const GridParallelRNG& RNG)=0;
-      virtual double S()const =0;
-      virtual void integrate(int level) =0;
-      virtual const LatticeColourMatrix get_U() const =0;
-      
-      void generate_momenta(LatticeColourMatrix& P,const RandNum& rand);
+      Integrator(int Nexp_, int MDsteps_, RealD trajL_,
+		 ActionSet& Aset, ObserverList obs):as(Aset), observers(obs){};
+      ~Integrator(){}
+      void init(LatticeLorentzColourMatrix&,
+		GridParallelRNG&);
+      double S();
+      void integrate(int level);
+      LatticeColourMatrix get_U();
     };
     
     namespace MDutils{
-      void generate_momenta_su3(LatticeColourMatrix& P,GridParallelRNG& RNG);
+      void generate_momenta(LatticeLorentzColourMatrix&,GridParallelRNG&);
+      void generate_momenta_su3(LatticeLorentzColourMatrix&,GridParallelRNG&);
     }
     
   }
diff --git a/lib/qcd/utils/SUn.h b/lib/qcd/utils/SUn.h
index 9a34f3e3..70d0dcb2 100644
--- a/lib/qcd/utils/SUn.h
+++ b/lib/qcd/utils/SUn.h
@@ -574,14 +574,14 @@ Note that in step D setting B ~ X - A and using B in place of A in step E will g
     LatticeMatrix Umu(out._grid);
     for(int mu=0;mu<Nd;mu++){
       LieRandomize(pRNG,Umu,0.01);
-      PokeLorentz(out,Umu,mu);
+      pokeLorentz(out,Umu,mu);
     }
   }
   static void ColdConfiguration(GridParallelRNG &pRNG,LatticeGaugeField &out){
     LatticeMatrix Umu(out._grid);
     Umu=1.0;
     for(int mu=0;mu<Nd;mu++){
-      PokeLorentz(out,Umu,mu);
+      pokeLorentz(out,Umu,mu);
     }
   }
 
diff --git a/tests/Make.inc b/tests/Make.inc
index 4592cd36..9e69dffe 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,5 @@
 
-bin_PROGRAMS = Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_GaugeAction Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
+bin_PROGRAMS = Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_GaugeAction Test_hmc Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
 
 
 Test_cayley_cg_SOURCES=Test_cayley_cg.cc
@@ -78,6 +78,10 @@ Test_GaugeAction_SOURCES=Test_GaugeAction.cc
 Test_GaugeAction_LDADD=-lGrid
 
 
+Test_hmc_SOURCES=Test_hmc.cc
+Test_hmc_LDADD=-lGrid
+
+
 Test_lie_generators_SOURCES=Test_lie_generators.cc
 Test_lie_generators_LDADD=-lGrid
 
diff --git a/tests/Test_hmc.cc b/tests/Test_hmc.cc
new file mode 100644
index 00000000..2a3d5d74
--- /dev/null
+++ b/tests/Test_hmc.cc
@@ -0,0 +1,36 @@
+#include "Grid.h"
+
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplex::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+
+  latt_size.resize(4);
+
+  latt_size[0] = 8;
+  latt_size[1] = 8;
+  latt_size[2] = 8;
+  latt_size[3] = 8;
+  double volume = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
+  
+  GridCartesian           Fine(latt_size,simd_layout,mpi_layout);
+  GridParallelRNG       FineRNG(&Fine);
+  GridSerialRNG       SerialRNG;
+  FineRNG.SeedRandomDevice();
+  
+  WilsonGaugeAction<LatticeLorentzColourMatrix, LatticeColourMatrix> Waction(6.0);
+  //Collect actions
+  ActionLevel Level;
+  Level.push_back(&Waction);
+  
+
+  //Integrator<IntegratorLeapFrog(12,10,1.0);
+}

From b1f94fa292a5fed0015a5a18ac645d109f21f408 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Sat, 4 Jul 2015 17:47:50 +0900
Subject: [PATCH 381/429] HMC ready but untested

---
 lib/qcd/QCD.h                             |  40 ++---
 lib/qcd/action/ActionBase.h               |   8 +-
 lib/qcd/action/gauge/WilsonGaugeAction.h  |   3 +-
 lib/qcd/hmc/HMC.h                         |  32 ++--
 lib/qcd/hmc/integrators/Integrator_base.h | 201 ++++++++++++++++++----
 tests/Test_hmc.cc                         |  24 ++-
 6 files changed, 225 insertions(+), 83 deletions(-)

diff --git a/lib/qcd/QCD.h b/lib/qcd/QCD.h
index a318141d..a61102a3 100644
--- a/lib/qcd/QCD.h
+++ b/lib/qcd/QCD.h
@@ -252,47 +252,47 @@ namespace QCD {
     //////////////////////////////////////////////////////////////////////////////
 
     //spin
-    template<class vobj> auto peekSpin(const vobj &rhs,int i) -> decltype(peekIndex<SpinIndex>(rhs,0))
+    template<class vobj> auto peekSpin(const vobj &rhs,int i) -> decltype(PeekIndex<SpinIndex>(rhs,0))
     {
-      return peekIndex<SpinIndex>(rhs,i);
+      return PeekIndex<SpinIndex>(rhs,i);
     }
-    template<class vobj> auto peekSpin(const vobj &rhs,int i,int j) -> decltype(peekIndex<SpinIndex>(rhs,0,0))
+    template<class vobj> auto peekSpin(const vobj &rhs,int i,int j) -> decltype(PeekIndex<SpinIndex>(rhs,0,0))
     {
-      return peekIndex<SpinIndex>(rhs,i,j);
+      return PeekIndex<SpinIndex>(rhs,i,j);
     }
-    template<class vobj> auto peekSpin(const Lattice<vobj> &rhs,int i) -> decltype(peekIndex<SpinIndex>(rhs,0))
+    template<class vobj> auto peekSpin(const Lattice<vobj> &rhs,int i) -> decltype(PeekIndex<SpinIndex>(rhs,0))
     {
-      return peekIndex<SpinIndex>(rhs,i);
+      return PeekIndex<SpinIndex>(rhs,i);
     }
-    template<class vobj> auto peekSpin(const Lattice<vobj> &rhs,int i,int j) -> decltype(peekIndex<SpinIndex>(rhs,0,0))
+    template<class vobj> auto peekSpin(const Lattice<vobj> &rhs,int i,int j) -> decltype(PeekIndex<SpinIndex>(rhs,0,0))
     {
-      return peekIndex<SpinIndex>(rhs,i,j);
+      return PeekIndex<SpinIndex>(rhs,i,j);
     }
     //colour
-    template<class vobj> auto peekColour(const vobj &rhs,int i) -> decltype(peekIndex<ColourIndex>(rhs,0))
+    template<class vobj> auto peekColour(const vobj &rhs,int i) -> decltype(PeekIndex<ColourIndex>(rhs,0))
     {
-      return peekIndex<ColourIndex>(rhs,i);
+      return PeekIndex<ColourIndex>(rhs,i);
     }
-    template<class vobj> auto peekColour(const vobj &rhs,int i,int j) -> decltype(peekIndex<ColourIndex>(rhs,0,0))
+    template<class vobj> auto peekColour(const vobj &rhs,int i,int j) -> decltype(PeekIndex<ColourIndex>(rhs,0,0))
     {
-      return peekIndex<ColourIndex>(rhs,i,j);
+      return PeekIndex<ColourIndex>(rhs,i,j);
     }
-    template<class vobj> auto peekColour(const Lattice<vobj> &rhs,int i) -> decltype(peekIndex<ColourIndex>(rhs,0))
+    template<class vobj> auto peekColour(const Lattice<vobj> &rhs,int i) -> decltype(PeekIndex<ColourIndex>(rhs,0))
     {
-      return peekIndex<ColourIndex>(rhs,i);
+      return PeekIndex<ColourIndex>(rhs,i);
     }
-    template<class vobj> auto peekColour(const Lattice<vobj> &rhs,int i,int j) -> decltype(peekIndex<ColourIndex>(rhs,0,0))
+    template<class vobj> auto peekColour(const Lattice<vobj> &rhs,int i,int j) -> decltype(PeekIndex<ColourIndex>(rhs,0,0))
     {
-      return peekIndex<ColourIndex>(rhs,i,j);
+      return PeekIndex<ColourIndex>(rhs,i,j);
     }
     //lorentz
-    template<class vobj> auto peekLorentz(const vobj &rhs,int i) -> decltype(peekIndex<LorentzIndex>(rhs,0))
+    template<class vobj> auto peekLorentz(const vobj &rhs,int i) -> decltype(PeekIndex<LorentzIndex>(rhs,0))
     {
-      return peekIndex<LorentzIndex>(rhs,i);
+      return PeekIndex<LorentzIndex>(rhs,i);
     }
-    template<class vobj> auto peekLorentz(const Lattice<vobj> &rhs,int i) -> decltype(peekIndex<LorentzIndex>(rhs,0))
+    template<class vobj> auto peekLorentz(const Lattice<vobj> &rhs,int i) -> decltype(PeekIndex<LorentzIndex>(rhs,0))
     {
-      return peekIndex<LorentzIndex>(rhs,i);
+      return PeekIndex<LorentzIndex>(rhs,i);
     }
 
     //////////////////////////////////////////////
diff --git a/lib/qcd/action/ActionBase.h b/lib/qcd/action/ActionBase.h
index e3b13f0e..51531750 100644
--- a/lib/qcd/action/ActionBase.h
+++ b/lib/qcd/action/ActionBase.h
@@ -7,10 +7,10 @@ template<class GaugeField>
 class Action { 
 
  public:
-  virtual void  init(const GaugeField &U)                      = 0;
-  virtual RealD S(const GaugeField &U)                        = 0;  // evaluate the action
-  virtual void  deriv(const GaugeField &U,GaugeField & dSdU ) = 0;  // evaluate the action derivative
-  virtual void  staple(const GaugeField &U,GaugeField & dSdU ) = 0;  // evaluate the action derivative
+  virtual void  init(const GaugeField &U, GridParallelRNG& pRNG) = 0;
+  virtual RealD S(const GaugeField &U)                           = 0;  // evaluate the action
+  virtual void  deriv(const GaugeField &U,GaugeField & dSdU )    = 0;  // evaluate the action derivative
+  virtual void  staple(const GaugeField &U,GaugeField & dSdU )   = 0;  // evaluate the action derivative
   //virtual void  refresh(const GaugeField & ) {}                ; 
   // Boundary conditions?
   // Heatbath?
diff --git a/lib/qcd/action/gauge/WilsonGaugeAction.h b/lib/qcd/action/gauge/WilsonGaugeAction.h
index 7726db80..c6ddd4a3 100644
--- a/lib/qcd/action/gauge/WilsonGaugeAction.h
+++ b/lib/qcd/action/gauge/WilsonGaugeAction.h
@@ -14,8 +14,7 @@ namespace Grid{
     public:
     WilsonGaugeAction(RealD b):beta(b){};
       
-      virtual void init(const GaugeField &U) {//FIXME
-      };
+      virtual void init(const GaugeField &U, GridParallelRNG& pRNG) {};
       
       virtual RealD S(const GaugeField &U) {
 	return WilsonLoops<MatrixField,GaugeField>::sumPlaquette(U);
diff --git a/lib/qcd/hmc/HMC.h b/lib/qcd/hmc/HMC.h
index e4d8740c..e783bee0 100644
--- a/lib/qcd/hmc/HMC.h
+++ b/lib/qcd/hmc/HMC.h
@@ -1,6 +1,6 @@
 //--------------------------------------------------------------------
 /*! @file HMC.h
- * @brief Declaration of classes for HybridMonteCarlo update
+ * @brief Declaration of classes for Hybrid Monte Carlo update
  *
  * @author Guido Cossu
  */
@@ -30,7 +30,7 @@ namespace Grid{
       const HMCparameters Params;
       GridSerialRNG sRNG;
       GridParallelRNG pRNG;
-      std::unique_ptr< Integrator<IntegType> > MD;
+      Integrator<IntegType>& MD;
       
 
       bool metropolis_test(const RealD DeltaH){
@@ -51,13 +51,13 @@ namespace Grid{
 	}
       }
 
-      RealD evolve_step(LatticeColourMatrix& Uin){
-     
-	MD->init(Uin,pRNG);     // set U and initialize P and phi's 
+      RealD evolve_step(LatticeLorentzColourMatrix& U){
+
+	MD->init(U,pRNG);     // set U and initialize P and phi's 
 	RealD H0 = MD->S();     // current state            
 	std::cout<<"Total H_before = "<< H0 << "\n";
       
-	MD->integrate(0);
+	MD->integrate(U,0);
       
 	RealD H1 = MD->S();     // updated state            
 	std::cout<<"Total H_after = "<< H1 << "\n";
@@ -68,7 +68,10 @@ namespace Grid{
       
       
     public:
-      HybridMonteCarlo(Integrator<IntegType>& MolDyn, GridBase* grid):MD(&MolDyn),pRNG(grid){
+    HybridMonteCarlo(HMCparameters Pms, 
+		     Integrator<IntegType>& MolDyn, 
+		     GridBase* grid):
+      Params(Pms),MD(MolDyn),pRNG(grid){
 	//FIXME
 
 	// initialize RNGs
@@ -79,7 +82,7 @@ namespace Grid{
 
 
 
-      void evolve(LatticeColourMatrix& Uin){
+      void evolve(LatticeLorentzColourMatrix& Uin){
 	Real DeltaH;
 	Real timer;
       
@@ -90,20 +93,21 @@ namespace Grid{
 	  DeltaH = evolve_step(Uin);
 	
 	  std::cout<< "[Timing] Trajectory time (s) : "<< timer/1000.0 << "\n";
-	  std::cout<< "dH = "<< DeltaH << "\n";
+	  std::cout<< " dH = "<< DeltaH << "\n";
 	
 	  // Update matrix
-	  Uin = MD->get_U();  //accept every time
+	  //Uin = MD->get_U();  //accept every time
 	}
 
-	// Actual updates
+	// Actual updates (evolve a copy Ucopy then copy back eventually)
+	LatticeLorentzColourMatrix Ucopy(Uin._grid);
 	for(int iter=Params.StartingConfig; 
 	    iter < Params.Nsweeps+Params.StartingConfig; ++iter){
 	  std::cout << "-- # Sweep = "<< iter <<  "\n";
+	  Ucopy = Uin;
+	  DeltaH = evolve_step(Ucopy);
 		
-	  DeltaH = evolve_step(Uin);
-		
-	  if(metropolis_test(DeltaH))  Uin = MD->get_U();
+	  if(metropolis_test(DeltaH))  Uin = Ucopy;
 	  // need sync?	
 	}
 
diff --git a/lib/qcd/hmc/integrators/Integrator_base.h b/lib/qcd/hmc/integrators/Integrator_base.h
index 61abf118..3f610958 100644
--- a/lib/qcd/hmc/integrators/Integrator_base.h
+++ b/lib/qcd/hmc/integrators/Integrator_base.h
@@ -9,6 +9,8 @@
 #ifndef INTEGRATOR_INCLUDED
 #define INTEGRATOR_INCLUDED
 
+#include <memory>
+
 class Observer;
 
 
@@ -22,59 +24,188 @@ namespace Grid{
     typedef std::vector<ActionLevel> ActionSet;
     typedef std::vector<Observer*> ObserverList;
     
-    
-    class Integrator2MN{
-      const double lambda = 0.1931833275037836;
-      void step (LatticeColourMatrix&, LatticeColourMatrix&, 
-		 int, std::vector<int>&);
-      
+    class LeapFrog;
+
+
+    struct IntegratorParameters{
+      int Nexp;
+      int MDsteps;  // number of outer steps
+      RealD trajL;  // trajectory length 
+      RealD stepsize;
+
+      IntegratorParameters(int Nexp_,
+			   int MDsteps_, 
+			   RealD trajL_):
+	Nexp(Nexp_),MDsteps(MDsteps_),trajL(trajL_),stepsize(trajL/MDsteps){};
     };
+
+
+    namespace MDutils{
+      void generate_momenta(LatticeLorentzColourMatrix&,GridParallelRNG&);
+      void generate_momenta_su3(LatticeLorentzColourMatrix&,GridParallelRNG&);
+    }
+
     
-    class IntegratorLeapFrog{
-      void step (LatticeColourMatrix&, LatticeColourMatrix&,
-		 int, std::vector<int>&);
-    };
     
 
 
     template< class IntegratorPolicy >
     class Integrator{
     private:
-      int Nexp;
-      int MDsteps;  // number of outer steps
-      RealD trajL;  // trajectory length 
-      RealD stepsize;
-
-      const std::vector<int> Nrel; // relative steps per level
+      IntegratorParameters Params;
       const ActionSet as;
-      ObserverList observers;
-      //      LatticeColourMatrix* const U;  // is shared among all actions - or use a singleton...
-      LatticeColourMatrix P;
+      const std::vector<int> Nrel; //relative step size per level
+      //ObserverList observers; // not yet
+      std::unique_ptr<LatticeLorentzColourMatrix> P;
       
       IntegratorPolicy TheIntegrator;// contains parameters too
-      void update_P(int lv,double ep);
-      void update_U(double ep);
+
+
+      void update_P(LatticeLorentzColourMatrix&U, int level,double ep){
+	for(int a=0; a<as[level].size(); ++a){
+	  LatticeLorentzColourMatrix force(U._grid);
+	  as[level].at(a)->deriv(U,force);
+	  *P -= force*ep;
+	}
+      }
+
+
+      void update_U(LatticeLorentzColourMatrix&U, double ep){
+	//rewrite exponential to deal with the lorentz index?
+	LatticeColourMatrix Umu(U._grid);
+	LatticeColourMatrix Pmu(U._grid);
+	for (int mu = 0; mu < Nd; mu++){
+	  Umu=peekLorentz(U, mu);
+	  Pmu=peekLorentz(*P, mu);
+	  Umu = expMat(Pmu, Complex(ep, 0.0))*Umu;
+	}
+
+      }
       
       void register_observers();
       void notify_observers();
-      void integrator_step(int level ,std::vector<Integer>& clock);
+      
+      friend void IntegratorPolicy::step (LatticeLorentzColourMatrix& U, 
+				     int level, std::vector<int>& clock,
+				     Integrator<LeapFrog>* Integ);
+    public:
+      Integrator(IntegratorParameters Par,
+		 ActionSet& Aset, std::vector<int> Nrel_):
+	Params(Par),as(Aset),Nrel(Nrel_){
+	assert(as.size() == Nrel.size());
+      };
+      
+      ~Integrator(){}
+
+
+      //Initialization of momenta and actions
+      void init(LatticeLorentzColourMatrix& U,
+		GridParallelRNG& pRNG){
+	std::cout<< "Integrator init\n";
+	if (!P)
+	  P = new LatticeLorentzColourMatrix(U._grid);
+	MDutils::generate_momenta(*P,pRNG);
+	
+	for(int level=0; level< as.size(); ++level){
+	  for(int actionID=0; actionID<as.at(level).size(); ++actionID){
+	    as[level].at(actionID)->init(U, pRNG);
+	  }
+	}
+      }
 
       
-    public:
-      Integrator(int Nexp_, int MDsteps_, RealD trajL_,
-		 ActionSet& Aset, ObserverList obs):as(Aset), observers(obs){};
-      ~Integrator(){}
-      void init(LatticeLorentzColourMatrix&,
-		GridParallelRNG&);
-      double S();
-      void integrate(int level);
-      LatticeColourMatrix get_U();
+
+      RealD S(LatticeLorentzColourMatrix& U){
+	// Momenta
+	LatticeComplex Hloc = - trace((*P)*adj(*P));
+	Complex Hsum = sum(Hloc);
+	
+	RealD H = Hsum.real();
+
+	// Actions
+	for(int level=0; level<as.size(); ++level)
+	  for(int actionID=0; actionID<as.at(level).size(); ++actionID)
+	    H += as[level].at(actionID)->S(U);
+	
+	return H;
+      }
+      
+      
+
+      void integrate(LatticeLorentzColourMatrix& U, int level){
+	std::vector<int> clock;
+	clock.resize(as.size(),0);
+	for(int step=0; step< Params.MDsteps; ++step)   // MD step
+	  TheIntegrator.step(U,0,clock, *(this));
+      }
     };
     
-    namespace MDutils{
-      void generate_momenta(LatticeLorentzColourMatrix&,GridParallelRNG&);
-      void generate_momenta_su3(LatticeLorentzColourMatrix&,GridParallelRNG&);
-    }
+
+    class MinimumNorm2{
+      const double lambda = 0.1931833275037836;
+    public:
+      void step (LatticeLorentzColourMatrix& U, int level, std::vector<int>& clock);
+      
+    };
+    
+    class LeapFrog{
+    public:
+      void step (LatticeLorentzColourMatrix& U, 
+		 int level, std::vector<int>& clock,
+		 Integrator<LeapFrog>* Integ){
+	// cl  : current level
+	// fl  : final level
+	// eps : current step size
+	
+	int fl = Integ->as.size() -1;
+	double eps = Integ->Params.stepsize;
+	
+	// Get current level step size
+	for(int l=0; l<=level; ++l) eps/= Integ->Nrel[l];
+	
+	int fin = 1;
+	for(int l=0; l<=level; ++l) fin*= Integ->Nrel[l];
+	fin = 2*Integ->Params.MDsteps*fin - 1;
+	
+	for(int e=0; e<Integ->Nrel[level]; ++e){
+	  
+	  if(clock[level] == 0){    // initial half step
+	    Integ->update_P(U, level,eps/2);
+	    ++clock[level];
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
+	  }
+	  if(level == fl){          // lowest level
+	    Integ->update_U(U, eps);
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"U "<< 0.5*(clock[level]+1) <<std::endl;
+	  }else{                 // recursive function call
+	    step(U, level+1,clock, Integ);
+	  }
+	  if(clock[level] == fin){  // final half step
+	    Integ->update_P(U, level,eps/2);
+	    
+	    ++clock[level];
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
+	  }else{                  // bulk step
+	    Integ->update_P(U, level,eps);
+	    
+	    clock[level]+=2;
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
+	  }
+	}
+
+
+
+
+      }
+    };
+
+
+
+
     
   }
 }
diff --git a/tests/Test_hmc.cc b/tests/Test_hmc.cc
index 2a3d5d74..cbae27ab 100644
--- a/tests/Test_hmc.cc
+++ b/tests/Test_hmc.cc
@@ -22,15 +22,23 @@ int main (int argc, char ** argv)
   double volume = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
   
   GridCartesian           Fine(latt_size,simd_layout,mpi_layout);
-  GridParallelRNG       FineRNG(&Fine);
-  GridSerialRNG       SerialRNG;
-  FineRNG.SeedRandomDevice();
   
+  // simplify template?
   WilsonGaugeAction<LatticeLorentzColourMatrix, LatticeColourMatrix> Waction(6.0);
-  //Collect actions
-  ActionLevel Level;
-  Level.push_back(&Waction);
-  
 
-  //Integrator<IntegratorLeapFrog(12,10,1.0);
+  //Collect actions
+  ActionLevel Level1;
+  Level1.push_back(&Waction);
+  ActionSet FullSet;
+  FullSet.push_back(Level1);
+
+  // Create integrator
+  IntegratorParameters MDpar(12,10,1.0);
+  std::vector<int> rel ={1};
+  Integrator<LeapFrog> MDleapfrog(MDpar, FullSet,rel);
+
+  // Create HMC
+  HMCparameters HMCpar;
+  HybridMonteCarlo<LeapFrog>  HMCrun(HMCpar, MDleapfrog, &Fine);
+
 }

From 62d8952c0a35d8dc226b4ffdf14e5fecf797d849 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Mon, 6 Jul 2015 02:24:58 +0900
Subject: [PATCH 382/429] Debugged vector version of ProjectOnGroup

---
 lib/qcd/action/ActionBase.h               |  1 -
 lib/qcd/action/gauge/WilsonGaugeAction.h  | 27 ++++++++++++++++-------
 lib/qcd/hmc/HMC.cc                        |  2 +-
 lib/qcd/hmc/HMC.h                         | 11 ++++-----
 lib/qcd/hmc/integrators/Integrator_base.h | 24 +++++++++++++++-----
 lib/simd/Grid_vector_unops.h              |  5 +++++
 lib/tensors/Tensor_Ta.h                   |  6 +++--
 lib/tensors/Tensor_exp.h                  |  2 ++
 tests/Test_hmc.cc                         | 14 +++++++++---
 tests/Test_main.cc                        | 23 ++++++++++++++++++-
 10 files changed, 89 insertions(+), 26 deletions(-)

diff --git a/lib/qcd/action/ActionBase.h b/lib/qcd/action/ActionBase.h
index 51531750..05edce86 100644
--- a/lib/qcd/action/ActionBase.h
+++ b/lib/qcd/action/ActionBase.h
@@ -10,7 +10,6 @@ class Action {
   virtual void  init(const GaugeField &U, GridParallelRNG& pRNG) = 0;
   virtual RealD S(const GaugeField &U)                           = 0;  // evaluate the action
   virtual void  deriv(const GaugeField &U,GaugeField & dSdU )    = 0;  // evaluate the action derivative
-  virtual void  staple(const GaugeField &U,GaugeField & dSdU )   = 0;  // evaluate the action derivative
   //virtual void  refresh(const GaugeField & ) {}                ; 
   // Boundary conditions?
   // Heatbath?
diff --git a/lib/qcd/action/gauge/WilsonGaugeAction.h b/lib/qcd/action/gauge/WilsonGaugeAction.h
index c6ddd4a3..d5eb56e8 100644
--- a/lib/qcd/action/gauge/WilsonGaugeAction.h
+++ b/lib/qcd/action/gauge/WilsonGaugeAction.h
@@ -17,17 +17,28 @@ namespace Grid{
       virtual void init(const GaugeField &U, GridParallelRNG& pRNG) {};
       
       virtual RealD S(const GaugeField &U) {
-	return WilsonLoops<MatrixField,GaugeField>::sumPlaquette(U);
+	RealD plaq = WilsonLoops<MatrixField,GaugeField>::avgPlaquette(U);
+	std::cout << "Plaq : "<<plaq << "\n";
+	double vol = U._grid->gSites();
+	return beta*(1.0 -plaq)*(Nd*(Nd-1.0))*vol*0.5;
       };
       virtual void deriv(const GaugeField &U,GaugeField & dSdU) {
-	//FIXME loop on directions
+
+	//not optimal implementation FIXME
+	//extend Ta to include Lorentz indexes
+	RealD factor = 0.5*beta/RealD(Nc);
+	std::cout << "Debug force Wilson  "<< factor << "\n";
 	MatrixField dSdU_mu(U._grid);
-	WilsonLoops<MatrixField,GaugeField>::Staple(dSdU_mu,U,0);
-      };
-      virtual void  staple(const GaugeField &stap,GaugeField & U) {
-	//FIXME loop on directions
-	MatrixField stap_mu(U._grid);
-	WilsonLoops<MatrixField,GaugeField>::Staple(stap_mu,U,0);
+	MatrixField Umu(U._grid);
+	for (int mu=0; mu < Nd; mu++){
+	  Umu = PeekIndex<LorentzIndex>(U,mu);
+	  // Staple in direction mu
+	  WilsonLoops<MatrixField,GaugeField>::Staple(dSdU_mu,U,mu);
+	  std::cout << "Staple norm ["<<mu<<"] = "<< norm2(dSdU_mu) <<"\n";
+	  dSdU_mu = Ta(Umu*adj(dSdU_mu))*factor;
+	  std::cout << "Force norm ["<<mu<<"] = "<< norm2(dSdU_mu) << "  : Umu "<< norm2(Umu)<<"\n";
+	  pokeLorentz(dSdU, dSdU_mu, mu);
+	}
       };
     };
     
diff --git a/lib/qcd/hmc/HMC.cc b/lib/qcd/hmc/HMC.cc
index 859d6646..343fb32c 100644
--- a/lib/qcd/hmc/HMC.cc
+++ b/lib/qcd/hmc/HMC.cc
@@ -9,7 +9,7 @@ namespace Grid{
 	////////////////////////////// Default values
 	Nsweeps             = 10;
 	TotalSweeps         = 10;
-	ThermalizationSteps = 0;
+	ThermalizationSteps = 1;
 	StartingConfig      = 0;
 	SaveInterval        = 1;
 	Filename_prefix     = "Conf_";
diff --git a/lib/qcd/hmc/HMC.h b/lib/qcd/hmc/HMC.h
index e783bee0..83a66a48 100644
--- a/lib/qcd/hmc/HMC.h
+++ b/lib/qcd/hmc/HMC.h
@@ -53,13 +53,13 @@ namespace Grid{
 
       RealD evolve_step(LatticeLorentzColourMatrix& U){
 
-	MD->init(U,pRNG);     // set U and initialize P and phi's 
-	RealD H0 = MD->S();     // current state            
+	MD.init(U,pRNG);     // set U and initialize P and phi's 
+	RealD H0 = MD.S(U);     // current state            
 	std::cout<<"Total H_before = "<< H0 << "\n";
       
-	MD->integrate(U,0);
+	MD.integrate(U,0);
       
-	RealD H1 = MD->S();     // updated state            
+	RealD H1 = MD.S(U);     // updated state            
 	std::cout<<"Total H_after = "<< H1 << "\n";
       
 	return (H1-H0);
@@ -104,10 +104,11 @@ namespace Grid{
 	for(int iter=Params.StartingConfig; 
 	    iter < Params.Nsweeps+Params.StartingConfig; ++iter){
 	  std::cout << "-- # Sweep = "<< iter <<  "\n";
+	  
 	  Ucopy = Uin;
 	  DeltaH = evolve_step(Ucopy);
 		
-	  if(metropolis_test(DeltaH))  Uin = Ucopy;
+	  if(metropolis_test(DeltaH)) Uin = Ucopy;
 	  // need sync?	
 	}
 
diff --git a/lib/qcd/hmc/integrators/Integrator_base.h b/lib/qcd/hmc/integrators/Integrator_base.h
index 3f610958..acb420f7 100644
--- a/lib/qcd/hmc/integrators/Integrator_base.h
+++ b/lib/qcd/hmc/integrators/Integrator_base.h
@@ -25,7 +25,7 @@ namespace Grid{
     typedef std::vector<Observer*> ObserverList;
     
     class LeapFrog;
-
+   
 
     struct IntegratorParameters{
       int Nexp;
@@ -77,7 +77,9 @@ namespace Grid{
 	for (int mu = 0; mu < Nd; mu++){
 	  Umu=peekLorentz(U, mu);
 	  Pmu=peekLorentz(*P, mu);
+	  std::cout << "U norm ["<<mu<<"] = "<< norm2(Umu) << "  : Pmu "<< norm2(Pmu)<<"\n";
 	  Umu = expMat(Pmu, Complex(ep, 0.0))*Umu;
+	  pokeLorentz(U, Umu, mu);
 	}
 
       }
@@ -102,10 +104,15 @@ namespace Grid{
       void init(LatticeLorentzColourMatrix& U,
 		GridParallelRNG& pRNG){
 	std::cout<< "Integrator init\n";
-	if (!P)
-	  P = new LatticeLorentzColourMatrix(U._grid);
+	if (!P){
+	  std::unique_ptr<LatticeLorentzColourMatrix> Pnew(new LatticeLorentzColourMatrix(U._grid));
+	  P = std::move(Pnew);
+
+	}
 	MDutils::generate_momenta(*P,pRNG);
 	
+       
+	
 	for(int level=0; level< as.size(); ++level){
 	  for(int actionID=0; actionID<as.at(level).size(); ++actionID){
 	    as[level].at(actionID)->init(U, pRNG);
@@ -116,12 +123,19 @@ namespace Grid{
       
 
       RealD S(LatticeLorentzColourMatrix& U){
+	LatticeComplex Hloc(U._grid);
+	Hloc = zero;
 	// Momenta
-	LatticeComplex Hloc = - trace((*P)*adj(*P));
+	for (int mu=0; mu <Nd; mu++){
+	  LatticeColourMatrix Pmu = peekLorentz(*P, mu);
+	  Hloc -= trace(Pmu*adj(Pmu));
+	}
 	Complex Hsum = sum(Hloc);
 	
 	RealD H = Hsum.real();
 
+	std::cout << "H_p = "<< H << "\n";
+
 	// Actions
 	for(int level=0; level<as.size(); ++level)
 	  for(int actionID=0; actionID<as.at(level).size(); ++actionID)
@@ -136,7 +150,7 @@ namespace Grid{
 	std::vector<int> clock;
 	clock.resize(as.size(),0);
 	for(int step=0; step< Params.MDsteps; ++step)   // MD step
-	  TheIntegrator.step(U,0,clock, *(this));
+	  TheIntegrator.step(U,0,clock, (this));
       }
     };
     
diff --git a/lib/simd/Grid_vector_unops.h b/lib/simd/Grid_vector_unops.h
index 37d1f6de..e4942937 100644
--- a/lib/simd/Grid_vector_unops.h
+++ b/lib/simd/Grid_vector_unops.h
@@ -94,6 +94,11 @@ namespace Grid {
   inline Grid_simd<S,V> rsqrt(const Grid_simd<S,V> &r) {
     return SimdApply(RSqrtRealFunctor<S>(),r);
   }
+  template < class Scalar > 
+  inline Scalar rsqrt(const Scalar &r) {
+    return (RSqrtRealFunctor<Scalar>(),r);
+  }
+
   template < class S, class V > 
   inline Grid_simd<S,V> cos(const Grid_simd<S,V> &r) {
     return SimdApply(CosRealFunctor<S>(),r);
diff --git a/lib/tensors/Tensor_Ta.h b/lib/tensors/Tensor_Ta.h
index e49ede35..cba07362 100644
--- a/lib/tensors/Tensor_Ta.h
+++ b/lib/tensors/Tensor_Ta.h
@@ -62,14 +62,16 @@ namespace Grid {
     {
       // need a check for the group type?
       iMatrix<vtype,N> ret(arg);
-      RealD nrm;
+      vtype nrm;
       vtype inner;
       for(int c1=0;c1<N;c1++){
 	zeroit(inner);	
 	for(int c2=0;c2<N;c2++)
 	  inner += innerProduct(ret._internal[c1][c2],ret._internal[c1][c2]);
 
-	nrm = 1.0/sqrt(Reduce(toReal(inner)));
+	//nrm = 1.0/sqrt(Reduce(toReal(inner)));
+	nrm = rsqrt(inner);
+
 	for(int c2=0;c2<N;c2++)
 	  ret._internal[c1][c2]*= nrm;
       
diff --git a/lib/tensors/Tensor_exp.h b/lib/tensors/Tensor_exp.h
index f8437116..6a514330 100644
--- a/lib/tensors/Tensor_exp.h
+++ b/lib/tensors/Tensor_exp.h
@@ -28,7 +28,9 @@ namespace Grid {
 	temp *= alpha/ComplexD(i);
 	temp = unit + temp*arg;
       }
+      
       return ProjectOnGroup(temp);//maybe not strictly necessary
+      
     }
 
 
diff --git a/tests/Test_hmc.cc b/tests/Test_hmc.cc
index cbae27ab..b292c57d 100644
--- a/tests/Test_hmc.cc
+++ b/tests/Test_hmc.cc
@@ -22,7 +22,13 @@ int main (int argc, char ** argv)
   double volume = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
   
   GridCartesian           Fine(latt_size,simd_layout,mpi_layout);
-  
+  GridParallelRNG  pRNG(&Fine);
+  pRNG.SeedRandomDevice();
+  LatticeLorentzColourMatrix     U(&Fine);
+
+  SU3::ColdConfiguration(pRNG, U);
+ 
+
   // simplify template?
   WilsonGaugeAction<LatticeLorentzColourMatrix, LatticeColourMatrix> Waction(6.0);
 
@@ -33,12 +39,14 @@ int main (int argc, char ** argv)
   FullSet.push_back(Level1);
 
   // Create integrator
-  IntegratorParameters MDpar(12,10,1.0);
+  IntegratorParameters MDpar(12,50,1.0);
   std::vector<int> rel ={1};
   Integrator<LeapFrog> MDleapfrog(MDpar, FullSet,rel);
 
   // Create HMC
   HMCparameters HMCpar;
-  HybridMonteCarlo<LeapFrog>  HMCrun(HMCpar, MDleapfrog, &Fine);
+  HybridMonteCarlo<LeapFrog>  HMC(HMCpar, MDleapfrog, &Fine);
+
+  HMC.evolve(U);
 
 }
diff --git a/tests/Test_main.cc b/tests/Test_main.cc
index 295b1981..a98fbeb7 100644
--- a/tests/Test_main.cc
+++ b/tests/Test_main.cc
@@ -220,15 +220,19 @@ int main (int argc, char ** argv)
     std::cout << cm << std::endl;
 
 
-    cm = Exponentiate(cm, 1.0, 12);
+    cm = Exponentiate(cm, 2.0, 12);
     std::cout << cm << "  " << std::endl;
     Complex det = Determinant(cm);
     std::cout << "determinant: " << det <<  std::endl;
+    std::cout << "norm: " << norm2(cm) <<  std::endl;
 
     cm = ProjectOnGroup(cm);
     std::cout << cm << "  " << std::endl;
+    std::cout << "norm: " << norm2(cm) <<  std::endl;
     cm = ProjectOnGroup(cm);
     std::cout << cm << "  " << std::endl;
+    std::cout << "norm: " << norm2(cm) <<  std::endl;
+
 
     //    det = Determinant(cm);
     //    std::cout << "determinant: " << det <<  std::endl;
@@ -245,7 +249,24 @@ int main (int argc, char ** argv)
     trscMat = trace(scMat); // Trace
 
     // Exponentiate test
+    std::vector<int> mysite {0,0,0,0};
+    random(FineRNG,cMat);
+    cMat = Ta(cMat);
+    peekSite(cm, cMat, mysite);
+    std::cout << cm << "  " << std::endl;
+    cm = Exponentiate(cm, 1.0, 12);
+    std::cout << cm << "  " << std::endl;
+    std::cout << "norm: " << norm2(cm) <<  std::endl;
+
+
+    std::cout << "norm cMmat : " << norm2(cMat) <<  std::endl;
     cMat = expMat(cMat, ComplexD(1.0, 0.0));
+    std::cout << "norm expMat: " << norm2(cMat) <<  std::endl;
+    peekSite(cm, cMat, mysite);
+    std::cout << cm << "  " << std::endl;
+    std::cout << "determinant: " << Determinant(cm) <<  std::endl;
+    std::cout << "norm: " << norm2(cm) <<  std::endl;
+
 
     // LatticeComplex trlcMat(&Fine);
     // trlcMat = trace(lcMat); // Trace involving iVector - now generates error

From 2718038977de0267017d94f005a16a934e8989e4 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Mon, 6 Jul 2015 12:58:49 +0900
Subject: [PATCH 383/429] HMC for Wilson Gauge action works

Fixed bug in momenta generation
---
 lib/qcd/action/gauge/WilsonGaugeAction.h       |  3 ---
 lib/qcd/hmc/HMC.cc                             |  6 +++---
 lib/qcd/hmc/HMC.h                              | 16 ++++------------
 lib/qcd/hmc/integrators/Integrator_base.cc     |  2 ++
 lib/qcd/hmc/integrators/Integrator_base.h      | 13 ++++---------
 lib/qcd/utils/SUn.h                            | 12 ++++++------
 lib/tensors/Tensor_Ta.h                        |  2 --
 tests/Make.inc                                 |  6 +++---
 tests/{Test_hmc.cc => Test_hmc_WilsonGauge.cc} |  4 ++--
 9 files changed, 24 insertions(+), 40 deletions(-)
 rename tests/{Test_hmc.cc => Test_hmc_WilsonGauge.cc} (93%)

diff --git a/lib/qcd/action/gauge/WilsonGaugeAction.h b/lib/qcd/action/gauge/WilsonGaugeAction.h
index d5eb56e8..aaedf2f9 100644
--- a/lib/qcd/action/gauge/WilsonGaugeAction.h
+++ b/lib/qcd/action/gauge/WilsonGaugeAction.h
@@ -27,16 +27,13 @@ namespace Grid{
 	//not optimal implementation FIXME
 	//extend Ta to include Lorentz indexes
 	RealD factor = 0.5*beta/RealD(Nc);
-	std::cout << "Debug force Wilson  "<< factor << "\n";
 	MatrixField dSdU_mu(U._grid);
 	MatrixField Umu(U._grid);
 	for (int mu=0; mu < Nd; mu++){
 	  Umu = PeekIndex<LorentzIndex>(U,mu);
 	  // Staple in direction mu
 	  WilsonLoops<MatrixField,GaugeField>::Staple(dSdU_mu,U,mu);
-	  std::cout << "Staple norm ["<<mu<<"] = "<< norm2(dSdU_mu) <<"\n";
 	  dSdU_mu = Ta(Umu*adj(dSdU_mu))*factor;
-	  std::cout << "Force norm ["<<mu<<"] = "<< norm2(dSdU_mu) << "  : Umu "<< norm2(Umu)<<"\n";
 	  pokeLorentz(dSdU, dSdU_mu, mu);
 	}
       };
diff --git a/lib/qcd/hmc/HMC.cc b/lib/qcd/hmc/HMC.cc
index 343fb32c..9db2c604 100644
--- a/lib/qcd/hmc/HMC.cc
+++ b/lib/qcd/hmc/HMC.cc
@@ -7,9 +7,9 @@ namespace Grid{
 	// FIXME fill this constructor  now just default values
 	  
 	////////////////////////////// Default values
-	Nsweeps             = 10;
-	TotalSweeps         = 10;
-	ThermalizationSteps = 1;
+	Nsweeps             = 100;
+	TotalSweeps         = 20;
+	ThermalizationSteps = 20;
 	StartingConfig      = 0;
 	SaveInterval        = 1;
 	Filename_prefix     = "Conf_";
diff --git a/lib/qcd/hmc/HMC.h b/lib/qcd/hmc/HMC.h
index 83a66a48..01fb2f9c 100644
--- a/lib/qcd/hmc/HMC.h
+++ b/lib/qcd/hmc/HMC.h
@@ -10,6 +10,7 @@
 
 #include <string>
 #include <memory>
+#include <unistd.h>
 
 namespace Grid{
   namespace QCD{
@@ -55,12 +56,12 @@ namespace Grid{
 
 	MD.init(U,pRNG);     // set U and initialize P and phi's 
 	RealD H0 = MD.S(U);     // current state            
-	std::cout<<"Total H_before = "<< H0 << "\n";
+	std::cout<<"Total H before = "<< H0 << "\n";
       
 	MD.integrate(U,0);
       
 	RealD H1 = MD.S(U);     // updated state            
-	std::cout<<"Total H_after = "<< H1 << "\n";
+	std::cout<<"Total H after = "<< H1 << "\n";
       
 	return (H1-H0);
       }
@@ -84,19 +85,14 @@ namespace Grid{
 
       void evolve(LatticeLorentzColourMatrix& Uin){
 	Real DeltaH;
-	Real timer;
       
 	// Thermalizations
 	for(int iter=1; iter <= Params.ThermalizationSteps; ++iter){
 	  std::cout << "-- # Thermalization step = "<< iter <<  "\n";
 	
 	  DeltaH = evolve_step(Uin);
-	
-	  std::cout<< "[Timing] Trajectory time (s) : "<< timer/1000.0 << "\n";
 	  std::cout<< " dH = "<< DeltaH << "\n";
 	
-	  // Update matrix
-	  //Uin = MD->get_U();  //accept every time
 	}
 
 	// Actual updates (evolve a copy Ucopy then copy back eventually)
@@ -109,13 +105,9 @@ namespace Grid{
 	  DeltaH = evolve_step(Ucopy);
 		
 	  if(metropolis_test(DeltaH)) Uin = Ucopy;
-	  // need sync?	
+	  //need sync?
 	}
-
-      
       }
-
-
     };
     
   }// QCD
diff --git a/lib/qcd/hmc/integrators/Integrator_base.cc b/lib/qcd/hmc/integrators/Integrator_base.cc
index 0a256444..8947405e 100644
--- a/lib/qcd/hmc/integrators/Integrator_base.cc
+++ b/lib/qcd/hmc/integrators/Integrator_base.cc
@@ -15,10 +15,12 @@ namespace Grid{
 
     void MDutils::generate_momenta_su3(LatticeLorentzColourMatrix& P,GridParallelRNG& pRNG){
       LatticeColourMatrix Pmu(P._grid);
+      Pmu = zero;
       for(int mu=0;mu<Nd;mu++){
 	SU3::GaussianLieAlgebraMatrix(pRNG, Pmu);
 	pokeLorentz(P, Pmu, mu);
       }
+      
     }
 
 
diff --git a/lib/qcd/hmc/integrators/Integrator_base.h b/lib/qcd/hmc/integrators/Integrator_base.h
index acb420f7..4891eb5e 100644
--- a/lib/qcd/hmc/integrators/Integrator_base.h
+++ b/lib/qcd/hmc/integrators/Integrator_base.h
@@ -71,14 +71,13 @@ namespace Grid{
 
 
       void update_U(LatticeLorentzColourMatrix&U, double ep){
-	//rewrite exponential to deal with the lorentz index?
+	//rewrite exponential to deal automatically  with the lorentz index?
 	LatticeColourMatrix Umu(U._grid);
 	LatticeColourMatrix Pmu(U._grid);
 	for (int mu = 0; mu < Nd; mu++){
 	  Umu=peekLorentz(U, mu);
 	  Pmu=peekLorentz(*P, mu);
-	  std::cout << "U norm ["<<mu<<"] = "<< norm2(Umu) << "  : Pmu "<< norm2(Pmu)<<"\n";
-	  Umu = expMat(Pmu, Complex(ep, 0.0))*Umu;
+	  Umu = expMat(Pmu, ep)*Umu;
 	  pokeLorentz(U, Umu, mu);
 	}
 
@@ -105,14 +104,10 @@ namespace Grid{
 		GridParallelRNG& pRNG){
 	std::cout<< "Integrator init\n";
 	if (!P){
-	  std::unique_ptr<LatticeLorentzColourMatrix> Pnew(new LatticeLorentzColourMatrix(U._grid));
+	  std::unique_ptr<LatticeLorentzColourMatrix> Pnew(new LatticeLorentzColourMatrix(U._grid));	  
 	  P = std::move(Pnew);
-
 	}
 	MDutils::generate_momenta(*P,pRNG);
-	
-       
-	
 	for(int level=0; level< as.size(); ++level){
 	  for(int actionID=0; actionID<as.at(level).size(); ++actionID){
 	    as[level].at(actionID)->init(U, pRNG);
@@ -128,7 +123,7 @@ namespace Grid{
 	// Momenta
 	for (int mu=0; mu <Nd; mu++){
 	  LatticeColourMatrix Pmu = peekLorentz(*P, mu);
-	  Hloc -= trace(Pmu*adj(Pmu));
+	  Hloc -= trace(Pmu*Pmu);
 	}
 	Complex Hsum = sum(Hloc);
 	
diff --git a/lib/qcd/utils/SUn.h b/lib/qcd/utils/SUn.h
index 70d0dcb2..10313f5c 100644
--- a/lib/qcd/utils/SUn.h
+++ b/lib/qcd/utils/SUn.h
@@ -545,21 +545,21 @@ Note that in step D setting B ~ X - A and using B in place of A in step E will g
 
   static void GaussianLieAlgebraMatrix(GridParallelRNG     &pRNG,LatticeMatrix &out,double scale=1.0){
     GridBase *grid = out._grid;
-    LatticeComplex ca (grid);
-    LatticeMatrix  lie(grid);
+    LatticeReal ca (grid);
     LatticeMatrix  la (grid);
     Complex ci(0.0,scale);
     Matrix ta;
 
     out=zero;
     for(int a=0;a<generators();a++){
-
       gaussian(pRNG,ca); 
       generator(a,ta);
-
-      la=ci*ca*ta;
-      out = lie+la; // e^{i la ta}
+      
+      la=toComplex(ca)*ci*ta;
+   
+      out += la; 
     }
+
   }
 
 
diff --git a/lib/tensors/Tensor_Ta.h b/lib/tensors/Tensor_Ta.h
index cba07362..724cbd6e 100644
--- a/lib/tensors/Tensor_Ta.h
+++ b/lib/tensors/Tensor_Ta.h
@@ -69,9 +69,7 @@ namespace Grid {
 	for(int c2=0;c2<N;c2++)
 	  inner += innerProduct(ret._internal[c1][c2],ret._internal[c1][c2]);
 
-	//nrm = 1.0/sqrt(Reduce(toReal(inner)));
 	nrm = rsqrt(inner);
-
 	for(int c2=0;c2<N;c2++)
 	  ret._internal[c1][c2]*= nrm;
       
diff --git a/tests/Make.inc b/tests/Make.inc
index 9e69dffe..31c39cd4 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,5 @@
 
-bin_PROGRAMS = Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_GaugeAction Test_hmc Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
+bin_PROGRAMS = Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_GaugeAction Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
 
 
 Test_cayley_cg_SOURCES=Test_cayley_cg.cc
@@ -78,8 +78,8 @@ Test_GaugeAction_SOURCES=Test_GaugeAction.cc
 Test_GaugeAction_LDADD=-lGrid
 
 
-Test_hmc_SOURCES=Test_hmc.cc
-Test_hmc_LDADD=-lGrid
+Test_hmc_WilsonGauge_SOURCES=Test_hmc_WilsonGauge.cc
+Test_hmc_WilsonGauge_LDADD=-lGrid
 
 
 Test_lie_generators_SOURCES=Test_lie_generators.cc
diff --git a/tests/Test_hmc.cc b/tests/Test_hmc_WilsonGauge.cc
similarity index 93%
rename from tests/Test_hmc.cc
rename to tests/Test_hmc_WilsonGauge.cc
index b292c57d..4fc468c9 100644
--- a/tests/Test_hmc.cc
+++ b/tests/Test_hmc_WilsonGauge.cc
@@ -26,7 +26,7 @@ int main (int argc, char ** argv)
   pRNG.SeedRandomDevice();
   LatticeLorentzColourMatrix     U(&Fine);
 
-  SU3::ColdConfiguration(pRNG, U);
+  SU3::HotConfiguration(pRNG, U);
  
 
   // simplify template?
@@ -39,7 +39,7 @@ int main (int argc, char ** argv)
   FullSet.push_back(Level1);
 
   // Create integrator
-  IntegratorParameters MDpar(12,50,1.0);
+  IntegratorParameters MDpar(12,30,1.0);
   std::vector<int> rel ={1};
   Integrator<LeapFrog> MDleapfrog(MDpar, FullSet,rel);
 

From 199185202563e7bea9576c0d9c5310aa32f5a96e Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Mon, 6 Jul 2015 16:17:32 +0900
Subject: [PATCH 384/429] Added minimum norm integrator

Little rearrangement of HMC and integrator classes
---
 lib/qcd/hmc/HMC.h                         |  15 +--
 lib/qcd/hmc/integrators/Integrator_base.h | 111 +++++++++++++++++-----
 tests/Test_hmc_WilsonGauge.cc             |   7 +-
 3 files changed, 94 insertions(+), 39 deletions(-)

diff --git a/lib/qcd/hmc/HMC.h b/lib/qcd/hmc/HMC.h
index 01fb2f9c..f2bb4bcb 100644
--- a/lib/qcd/hmc/HMC.h
+++ b/lib/qcd/hmc/HMC.h
@@ -10,7 +10,6 @@
 
 #include <string>
 #include <memory>
-#include <unistd.h>
 
 namespace Grid{
   namespace QCD{
@@ -30,10 +29,8 @@ namespace Grid{
     class HybridMonteCarlo{
       const HMCparameters Params;
       GridSerialRNG sRNG;
-      GridParallelRNG pRNG;
       Integrator<IntegType>& MD;
       
-
       bool metropolis_test(const RealD DeltaH){
 	RealD rn_test;
 	RealD prob = std::exp(-DeltaH);
@@ -54,7 +51,7 @@ namespace Grid{
 
       RealD evolve_step(LatticeLorentzColourMatrix& U){
 
-	MD.init(U,pRNG);     // set U and initialize P and phi's 
+	MD.init(U);     // set U and initialize P and phi's 
 	RealD H0 = MD.S(U);     // current state            
 	std::cout<<"Total H before = "<< H0 << "\n";
       
@@ -70,14 +67,12 @@ namespace Grid{
       
     public:
     HybridMonteCarlo(HMCparameters Pms, 
-		     Integrator<IntegType>& MolDyn, 
-		     GridBase* grid):
-      Params(Pms),MD(MolDyn),pRNG(grid){
+		     Integrator<IntegType>& MolDyn):
+      Params(Pms),MD(MolDyn){
 	//FIXME
 
-	// initialize RNGs
+	// initialize RNGs also with seed
 	sRNG.SeedRandomDevice();
-	pRNG.SeedRandomDevice();
       }
       ~HybridMonteCarlo(){};
 
@@ -85,7 +80,7 @@ namespace Grid{
 
       void evolve(LatticeLorentzColourMatrix& Uin){
 	Real DeltaH;
-      
+	
 	// Thermalizations
 	for(int iter=1; iter <= Params.ThermalizationSteps; ++iter){
 	  std::cout << "-- # Thermalization step = "<< iter <<  "\n";
diff --git a/lib/qcd/hmc/integrators/Integrator_base.h b/lib/qcd/hmc/integrators/Integrator_base.h
index 4891eb5e..9b47a19d 100644
--- a/lib/qcd/hmc/integrators/Integrator_base.h
+++ b/lib/qcd/hmc/integrators/Integrator_base.h
@@ -24,9 +24,6 @@ namespace Grid{
     typedef std::vector<ActionLevel> ActionSet;
     typedef std::vector<Observer*> ObserverList;
     
-    class LeapFrog;
-   
-
     struct IntegratorParameters{
       int Nexp;
       int MDsteps;  // number of outer steps
@@ -46,20 +43,20 @@ namespace Grid{
     }
 
     
-    
-
-
     template< class IntegratorPolicy >
     class Integrator{
     private:
       IntegratorParameters Params;
       const ActionSet as;
       const std::vector<int> Nrel; //relative step size per level
-      //ObserverList observers; // not yet
       std::unique_ptr<LatticeLorentzColourMatrix> P;
-      
-      IntegratorPolicy TheIntegrator;// contains parameters too
+      GridParallelRNG pRNG;
+      //ObserverList observers; // not yet
+     
+      IntegratorPolicy TheIntegrator;
 
+      void register_observers();
+      void notify_observers();
 
       void update_P(LatticeLorentzColourMatrix&U, int level,double ep){
 	for(int a=0; a<as[level].size(); ++a){
@@ -77,36 +74,32 @@ namespace Grid{
 	for (int mu = 0; mu < Nd; mu++){
 	  Umu=peekLorentz(U, mu);
 	  Pmu=peekLorentz(*P, mu);
-	  Umu = expMat(Pmu, ep)*Umu;
+	  Umu = expMat(Pmu, ep, Params.Nexp)*Umu;
 	  pokeLorentz(U, Umu, mu);
 	}
 
       }
       
-      void register_observers();
-      void notify_observers();
+
       
       friend void IntegratorPolicy::step (LatticeLorentzColourMatrix& U, 
 				     int level, std::vector<int>& clock,
-				     Integrator<LeapFrog>* Integ);
+				     Integrator<IntegratorPolicy>* Integ);
     public:
-      Integrator(IntegratorParameters Par,
+    Integrator(GridBase* grid, IntegratorParameters Par,
 		 ActionSet& Aset, std::vector<int> Nrel_):
-	Params(Par),as(Aset),Nrel(Nrel_){
+      Params(Par),as(Aset),Nrel(Nrel_),P(new LatticeLorentzColourMatrix(grid)),pRNG(grid){
 	assert(as.size() == Nrel.size());
+	pRNG.SeedRandomDevice();
       };
       
       ~Integrator(){}
 
 
       //Initialization of momenta and actions
-      void init(LatticeLorentzColourMatrix& U,
-		GridParallelRNG& pRNG){
+      void init(LatticeLorentzColourMatrix& U){
 	std::cout<< "Integrator init\n";
-	if (!P){
-	  std::unique_ptr<LatticeLorentzColourMatrix> Pnew(new LatticeLorentzColourMatrix(U._grid));	  
-	  P = std::move(Pnew);
-	}
+
 	MDutils::generate_momenta(*P,pRNG);
 	for(int level=0; level< as.size(); ++level){
 	  for(int actionID=0; actionID<as.at(level).size(); ++actionID){
@@ -116,7 +109,7 @@ namespace Grid{
       }
 
       
-
+      // Calculate action
       RealD S(LatticeLorentzColourMatrix& U){
 	LatticeComplex Hloc(U._grid);
 	Hloc = zero;
@@ -153,7 +146,74 @@ namespace Grid{
     class MinimumNorm2{
       const double lambda = 0.1931833275037836;
     public:
-      void step (LatticeLorentzColourMatrix& U, int level, std::vector<int>& clock);
+      void step (LatticeLorentzColourMatrix& U, 
+		 int level, std::vector<int>& clock,
+		 Integrator<MinimumNorm2>* Integ){
+	// level  : current level
+	// fl  : final level
+	// eps : current step size
+
+	int fl = Integ->as.size() -1;
+	double eps = Integ->Params.stepsize;
+	
+	for(int l=0; l<=level; ++l) eps/= 2.0*Integ->Nrel[l];
+	
+	int fin = Integ->Nrel[0];
+	for(int l=1; l<=level; ++l) fin*= 2.0*Integ->Nrel[l];
+	fin = 3*Integ->Params.MDsteps*fin -1;
+	
+	
+	for(int e=0; e<Integ->Nrel[level]; ++e){
+	  
+	  if(clock[level] == 0){    // initial half step 
+	    Integ->update_P(U,level,lambda*eps);
+	    ++clock[level];
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"P "<< clock[level] <<std::endl;
+	  }
+	  
+	  if(level == fl){          // lowest level 
+	    Integ->update_U(U,0.5*eps);
+	    
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"U "<< (clock[level]+1) <<std::endl;
+	  }else{                 // recursive function call 
+	    step(U,level+1,clock, Integ);
+	  }
+	  
+	  Integ->update_P(U,level,(1.0-2.0*lambda)*eps);
+	  ++clock[level];
+	  for(int l=0; l<level;++l) std::cout<<"   ";
+	  std::cout<<"P "<< (clock[level]) <<std::endl;
+	  
+	  if(level == fl){          // lowest level 
+	    Integ->update_U(U,0.5*eps);
+	    
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"U "<< (clock[level]+1) <<std::endl;
+	  }else{                 // recursive function call 
+	    step(U,level+1,clock, Integ);
+	  }    
+	  
+	  
+	  if(clock[level] == fin){  // final half step
+	    Integ->update_P(U,level,lambda*eps);
+	    
+	    ++clock[level];
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"P "<< clock[level] <<std::endl;
+	  }else{                  // bulk step
+	    Integ->update_P(U,level,lambda*2.0*eps);
+	    
+	    clock[level]+=2;
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"P "<< clock[level] <<std::endl;
+	  }
+	}
+	
+	
+	
+      }
       
     };
     
@@ -162,7 +222,6 @@ namespace Grid{
       void step (LatticeLorentzColourMatrix& U, 
 		 int level, std::vector<int>& clock,
 		 Integrator<LeapFrog>* Integ){
-	// cl  : current level
 	// fl  : final level
 	// eps : current step size
 	
@@ -179,7 +238,7 @@ namespace Grid{
 	for(int e=0; e<Integ->Nrel[level]; ++e){
 	  
 	  if(clock[level] == 0){    // initial half step
-	    Integ->update_P(U, level,eps/2);
+	    Integ->update_P(U, level,eps/2.0);
 	    ++clock[level];
 	    for(int l=0; l<level;++l) std::cout<<"   ";
 	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
@@ -192,7 +251,7 @@ namespace Grid{
 	    step(U, level+1,clock, Integ);
 	  }
 	  if(clock[level] == fin){  // final half step
-	    Integ->update_P(U, level,eps/2);
+	    Integ->update_P(U, level,eps/2.0);
 	    
 	    ++clock[level];
 	    for(int l=0; l<level;++l) std::cout<<"   ";
diff --git a/tests/Test_hmc_WilsonGauge.cc b/tests/Test_hmc_WilsonGauge.cc
index 4fc468c9..98f75b3f 100644
--- a/tests/Test_hmc_WilsonGauge.cc
+++ b/tests/Test_hmc_WilsonGauge.cc
@@ -29,7 +29,7 @@ int main (int argc, char ** argv)
   SU3::HotConfiguration(pRNG, U);
  
 
-  // simplify template?
+  // simplify template declaration? Strip the lorentz from the second template
   WilsonGaugeAction<LatticeLorentzColourMatrix, LatticeColourMatrix> Waction(6.0);
 
   //Collect actions
@@ -39,13 +39,14 @@ int main (int argc, char ** argv)
   FullSet.push_back(Level1);
 
   // Create integrator
+  typedef MinimumNorm2  IntegratorAlgorithm;// change here to change the algorithm
   IntegratorParameters MDpar(12,30,1.0);
   std::vector<int> rel ={1};
-  Integrator<LeapFrog> MDleapfrog(MDpar, FullSet,rel);
+  Integrator<IntegratorAlgorithm> MDynamics(&Fine,MDpar, FullSet,rel);
 
   // Create HMC
   HMCparameters HMCpar;
-  HybridMonteCarlo<LeapFrog>  HMC(HMCpar, MDleapfrog, &Fine);
+  HybridMonteCarlo<IntegratorAlgorithm>  HMC(HMCpar, MDynamics);
 
   HMC.evolve(U);
 

From 32e6887d5f8348a20abcf9fcd65523b40698cfd0 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Mon, 6 Jul 2015 16:46:43 +0900
Subject: [PATCH 385/429] Rearranging files in hmc

---
 lib/Make.inc                                  |   4 +-
 lib/qcd/QCD.h                                 |   3 +-
 lib/qcd/hmc/HMC.h                             |  17 +-
 .../{Integrator_base.cc => Integrator.cc}     |   0
 lib/qcd/hmc/integrators/Integrator.h          | 148 +++++++++
 .../hmc/integrators/Integrator_algorithm.h    | 152 ++++++++++
 lib/qcd/hmc/integrators/Integrator_base.h     | 280 ------------------
 tests/Test_hmc_WilsonGauge.cc                 |   2 +-
 8 files changed, 312 insertions(+), 294 deletions(-)
 rename lib/qcd/hmc/integrators/{Integrator_base.cc => Integrator.cc} (100%)
 create mode 100644 lib/qcd/hmc/integrators/Integrator.h
 create mode 100644 lib/qcd/hmc/integrators/Integrator_algorithm.h
 delete mode 100644 lib/qcd/hmc/integrators/Integrator_base.h

diff --git a/lib/Make.inc b/lib/Make.inc
index 85ed84c2..dfa8401b 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./Cshift.h ./simd/Grid_avx.h ./simd/Grid_vector_types.h ./simd/Grid_sse4.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_vector_unops.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./Tensors.h ./Algorithms.h ./communicator/Communicator_base.h ./lattice/Lattice_rng.h ./lattice/Lattice_reduction.h ./lattice/Lattice_transfer.h ./lattice/Lattice_unary.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_comparison.h ./lattice/Lattice_overload.h ./lattice/Lattice_reality.h ./lattice/Lattice_local.h ./lattice/Lattice_conformable.h ./lattice/Lattice_where.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_ET.h ./lattice/Lattice_transpose.h ./lattice/Lattice_trace.h ./Stencil.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_exp.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_class.h ./tensors/Tensor_logical.h ./tensors/Tensor_transpose.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_reality.h ./tensors/Tensor_trace.h ./tensors/Tensor_index.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_outer.h ./tensors/Tensor_inner.h ./tensors/Tensor_traits.h ./tensors/Tensor_Ta.h ./tensors/Tensor_unary.h ./tensors/Tensor_determinant.h ./tensors/Tensor_arith.h ./tensors/Tensor_extract_merge.h ./Communicator.h ./Cartesian.h ./parallelIO/NerscIO.h ./qcd/QCD.h ./qcd/hmc/integrators/Integrator_base.h ./qcd/hmc/HMC.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/LinalgUtils.h ./qcd/utils/CovariantCshift.h ./qcd/utils/WilsonLoops.h ./qcd/action/ActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/spin/TwoSpinor.h ./qcd/spin/Dirac.h ./cshift/Cshift_common.h ./cshift/Cshift_none.h ./cshift/Cshift_mpi.h ./Simd.h ./GridConfig.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_red_black.h ./cartesian/Cartesian_full.h ./AlignedAllocator.h ./Lattice.h ./Old/Tensor_poke.h ./Old/Tensor_peek.h ./Threads.h ./Grid.h ./algorithms/Preconditioner.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/AdefGeneric.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Zolotarev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Remez.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./algorithms/CoarsenedMatrix.h ./stencil/Lebesgue.h
+HFILES=./Cshift.h ./simd/Grid_avx.h ./simd/Grid_vector_types.h ./simd/Grid_sse4.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_vector_unops.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./Tensors.h ./Algorithms.h ./communicator/Communicator_base.h ./lattice/Lattice_rng.h ./lattice/Lattice_reduction.h ./lattice/Lattice_transfer.h ./lattice/Lattice_unary.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_comparison.h ./lattice/Lattice_overload.h ./lattice/Lattice_reality.h ./lattice/Lattice_local.h ./lattice/Lattice_conformable.h ./lattice/Lattice_where.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_ET.h ./lattice/Lattice_transpose.h ./lattice/Lattice_trace.h ./Stencil.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_exp.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_class.h ./tensors/Tensor_logical.h ./tensors/Tensor_transpose.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_reality.h ./tensors/Tensor_trace.h ./tensors/Tensor_index.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_outer.h ./tensors/Tensor_inner.h ./tensors/Tensor_traits.h ./tensors/Tensor_Ta.h ./tensors/Tensor_unary.h ./tensors/Tensor_determinant.h ./tensors/Tensor_arith.h ./tensors/Tensor_extract_merge.h ./Communicator.h ./Cartesian.h ./parallelIO/NerscIO.h ./qcd/QCD.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/hmc/HMC.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/LinalgUtils.h ./qcd/utils/CovariantCshift.h ./qcd/utils/WilsonLoops.h ./qcd/action/ActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/spin/TwoSpinor.h ./qcd/spin/Dirac.h ./cshift/Cshift_common.h ./cshift/Cshift_none.h ./cshift/Cshift_mpi.h ./Simd.h ./GridConfig.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_red_black.h ./cartesian/Cartesian_full.h ./AlignedAllocator.h ./Lattice.h ./Old/Tensor_poke.h ./Old/Tensor_peek.h ./Threads.h ./Grid.h ./algorithms/Preconditioner.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/AdefGeneric.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Zolotarev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Remez.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./algorithms/CoarsenedMatrix.h ./stencil/Lebesgue.h
 
-CCFILES=./qcd/hmc/integrators/Integrator_base.cc ./qcd/hmc/HMC.cc ./qcd/utils/SpaceTimeGrid.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/spin/Dirac.cc ./GridInit.cc ./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
+CCFILES=./qcd/hmc/integrators/Integrator.cc ./qcd/hmc/HMC.cc ./qcd/utils/SpaceTimeGrid.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/spin/Dirac.cc ./GridInit.cc ./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/qcd/QCD.h b/lib/qcd/QCD.h
index a61102a3..1227ad74 100644
--- a/lib/qcd/QCD.h
+++ b/lib/qcd/QCD.h
@@ -413,7 +413,8 @@ namespace QCD {
 #include <qcd/utils/WilsonLoops.h>
 #include <qcd/utils/SUn.h>
 #include <qcd/action/Actions.h>
-#include <qcd/hmc/integrators/Integrator_base.h>
+#include <qcd/hmc/integrators/Integrator.h>
+#include <qcd/hmc/integrators/Integrator_algorithm.h>
 #include <qcd/hmc/HMC.h>
 
 
diff --git a/lib/qcd/hmc/HMC.h b/lib/qcd/hmc/HMC.h
index f2bb4bcb..c0f06b64 100644
--- a/lib/qcd/hmc/HMC.h
+++ b/lib/qcd/hmc/HMC.h
@@ -25,11 +25,11 @@ namespace Grid{
       HMCparameters();
     };
     
-    template <class IntegType> 
+    template <class Algorithm> 
     class HybridMonteCarlo{
       const HMCparameters Params;
       GridSerialRNG sRNG;
-      Integrator<IntegType>& MD;
+      Integrator<Algorithm>& MD;
       
       bool metropolis_test(const RealD DeltaH){
 	RealD rn_test;
@@ -51,23 +51,21 @@ namespace Grid{
 
       RealD evolve_step(LatticeLorentzColourMatrix& U){
 
-	MD.init(U);     // set U and initialize P and phi's 
-	RealD H0 = MD.S(U);     // current state            
+	MD.init(U); // set U and initialize P and phi's 
+	RealD H0 = MD.S(U); // initial state action  
 	std::cout<<"Total H before = "<< H0 << "\n";
       
-	MD.integrate(U,0);
+	MD.integrate(U);
       
-	RealD H1 = MD.S(U);     // updated state            
+	RealD H1 = MD.S(U); // updated state action            
 	std::cout<<"Total H after = "<< H1 << "\n";
       
 	return (H1-H0);
       }
-
-      
       
     public:
     HybridMonteCarlo(HMCparameters Pms, 
-		     Integrator<IntegType>& MolDyn):
+		     Integrator<Algorithm>& MolDyn):
       Params(Pms),MD(MolDyn){
 	//FIXME
 
@@ -87,7 +85,6 @@ namespace Grid{
 	
 	  DeltaH = evolve_step(Uin);
 	  std::cout<< " dH = "<< DeltaH << "\n";
-	
 	}
 
 	// Actual updates (evolve a copy Ucopy then copy back eventually)
diff --git a/lib/qcd/hmc/integrators/Integrator_base.cc b/lib/qcd/hmc/integrators/Integrator.cc
similarity index 100%
rename from lib/qcd/hmc/integrators/Integrator_base.cc
rename to lib/qcd/hmc/integrators/Integrator.cc
diff --git a/lib/qcd/hmc/integrators/Integrator.h b/lib/qcd/hmc/integrators/Integrator.h
new file mode 100644
index 00000000..5eb319b2
--- /dev/null
+++ b/lib/qcd/hmc/integrators/Integrator.h
@@ -0,0 +1,148 @@
+//--------------------------------------------------------------------
+/*! @file Integrator.h
+ * @brief Declaration of classes for the Molecular Dynamics integrator
+ *
+ * @author Guido Cossu
+ */
+//--------------------------------------------------------------------
+
+#ifndef INTEGRATOR_INCLUDED
+#define INTEGRATOR_INCLUDED
+
+class Observer;
+
+#include <memory>
+
+namespace Grid{
+  namespace QCD{
+
+    typedef Action<LatticeLorentzColourMatrix>*  ActPtr; // now force the same colours as the rest of the code
+    typedef std::vector<ActPtr> ActionLevel;
+    typedef std::vector<ActionLevel> ActionSet;
+    typedef std::vector<Observer*> ObserverList;
+    
+    struct IntegratorParameters{
+      int Nexp;
+      int MDsteps;  // number of outer steps
+      RealD trajL;  // trajectory length 
+      RealD stepsize;
+
+      IntegratorParameters(int Nexp_,
+			   int MDsteps_, 
+			   RealD trajL_):
+      Nexp(Nexp_),MDsteps(MDsteps_),trajL(trajL_),stepsize(trajL/MDsteps){};
+    };
+
+
+    namespace MDutils{
+      void generate_momenta(LatticeLorentzColourMatrix&,GridParallelRNG&);
+      void generate_momenta_su3(LatticeLorentzColourMatrix&,GridParallelRNG&);
+    }
+
+    /*! @brief Class for Molecular Dynamics management */   
+    template< class IntegratorAlgorithm >
+    class Integrator{
+    private:
+      IntegratorParameters Params;
+      const ActionSet as;
+      const std::vector<int> Nrel; //relative step size per level
+      std::unique_ptr<LatticeLorentzColourMatrix> P;
+      GridParallelRNG pRNG;
+      //ObserverList observers; // not yet
+     
+      IntegratorAlgorithm TheIntegrator;
+
+      void register_observers();
+      void notify_observers();
+
+      void update_P(LatticeLorentzColourMatrix&U, int level,double ep){
+	for(int a=0; a<as[level].size(); ++a){
+	  LatticeLorentzColourMatrix force(U._grid);
+	  as[level].at(a)->deriv(U,force);
+	  *P -= force*ep;
+	}
+      }
+
+
+      void update_U(LatticeLorentzColourMatrix&U, double ep){
+	//rewrite exponential to deal automatically  with the lorentz index?
+	LatticeColourMatrix Umu(U._grid);
+	LatticeColourMatrix Pmu(U._grid);
+	for (int mu = 0; mu < Nd; mu++){
+	  Umu=peekLorentz(U, mu);
+	  Pmu=peekLorentz(*P, mu);
+	  Umu = expMat(Pmu, ep, Params.Nexp)*Umu;
+	  pokeLorentz(U, Umu, mu);
+	}
+
+      }
+      
+
+      
+      friend void IntegratorAlgorithm::step (LatticeLorentzColourMatrix& U, 
+					     int level, std::vector<int>& clock,
+					     Integrator<IntegratorAlgorithm>* Integ);
+    public:
+    Integrator(GridBase* grid, IntegratorParameters Par,
+		 ActionSet& Aset, std::vector<int> Nrel_):
+      Params(Par),as(Aset),Nrel(Nrel_),P(new LatticeLorentzColourMatrix(grid)),pRNG(grid){
+	assert(as.size() == Nrel.size());
+	pRNG.SeedRandomDevice();
+      };
+      
+      ~Integrator(){}
+
+
+      //Initialization of momenta and actions
+      void init(LatticeLorentzColourMatrix& U){
+	std::cout<< "Integrator init\n";
+
+	MDutils::generate_momenta(*P,pRNG);
+	for(int level=0; level< as.size(); ++level){
+	  for(int actionID=0; actionID<as.at(level).size(); ++actionID){
+	    as[level].at(actionID)->init(U, pRNG);
+	  }
+	}
+      }
+
+      
+      // Calculate action
+      RealD S(LatticeLorentzColourMatrix& U){
+	LatticeComplex Hloc(U._grid);
+	Hloc = zero;
+	// Momenta
+	for (int mu=0; mu <Nd; mu++){
+	  LatticeColourMatrix Pmu = peekLorentz(*P, mu);
+	  Hloc -= trace(Pmu*Pmu);
+	}
+	Complex Hsum = sum(Hloc);
+	
+	RealD H = Hsum.real();
+
+	std::cout << "H_p = "<< H << "\n";
+
+	// Actions
+	for(int level=0; level<as.size(); ++level)
+	  for(int actionID=0; actionID<as.at(level).size(); ++actionID)
+	    H += as[level].at(actionID)->S(U);
+	
+	return H;
+      }
+
+      void integrate(LatticeLorentzColourMatrix& U){
+	std::vector<int> clock;
+	clock.resize(as.size(),0);
+	for(int step=0; step< Params.MDsteps; ++step)   // MD step
+	  TheIntegrator.step(U,0,clock, (this));
+      }
+    };
+    
+
+
+
+
+
+    
+  }
+}
+#endif//INTEGRATOR_INCLUDED
diff --git a/lib/qcd/hmc/integrators/Integrator_algorithm.h b/lib/qcd/hmc/integrators/Integrator_algorithm.h
new file mode 100644
index 00000000..0dce2978
--- /dev/null
+++ b/lib/qcd/hmc/integrators/Integrator_algorithm.h
@@ -0,0 +1,152 @@
+//--------------------------------------------------------------------
+/*! @file Integrator_algorithm.h
+ * @brief Declaration of classes for the Molecular Dynamics algorithms
+ *
+ * @author Guido Cossu
+ */
+//--------------------------------------------------------------------
+
+#ifndef INTEGRATOR_ALG_INCLUDED
+#define INTEGRATOR_ALG_INCLUDED
+
+
+namespace Grid{
+  namespace QCD{
+
+
+    class MinimumNorm2{
+      const double lambda = 0.1931833275037836;
+    public:
+      void step (LatticeLorentzColourMatrix& U, 
+		 int level, std::vector<int>& clock,
+		 Integrator<MinimumNorm2>* Integ){
+	// level  : current level
+	// fl     : final level
+	// eps    : current step size
+
+	int fl = Integ->as.size() -1;
+	double eps = Integ->Params.stepsize;
+	
+	for(int l=0; l<=level; ++l) eps/= 2.0*Integ->Nrel[l];
+	
+	int fin = Integ->Nrel[0];
+	for(int l=1; l<=level; ++l) fin*= 2.0*Integ->Nrel[l];
+	fin = 3*Integ->Params.MDsteps*fin -1;
+	
+	
+	for(int e=0; e<Integ->Nrel[level]; ++e){
+	  
+	  if(clock[level] == 0){    // initial half step 
+	    Integ->update_P(U,level,lambda*eps);
+	    ++clock[level];
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"P "<< clock[level] <<std::endl;
+	  }
+	  
+	  if(level == fl){          // lowest level 
+	    Integ->update_U(U,0.5*eps);
+	    
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"U "<< (clock[level]+1) <<std::endl;
+	  }else{                 // recursive function call 
+	    step(U,level+1,clock, Integ);
+	  }
+	  
+	  Integ->update_P(U,level,(1.0-2.0*lambda)*eps);
+	  ++clock[level];
+	  for(int l=0; l<level;++l) std::cout<<"   ";
+	  std::cout<<"P "<< (clock[level]) <<std::endl;
+	  
+	  if(level == fl){          // lowest level 
+	    Integ->update_U(U,0.5*eps);
+	    
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"U "<< (clock[level]+1) <<std::endl;
+	  }else{                 // recursive function call 
+	    step(U,level+1,clock, Integ);
+	  }    
+	  
+	  
+	  if(clock[level] == fin){  // final half step
+	    Integ->update_P(U,level,lambda*eps);
+	    
+	    ++clock[level];
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"P "<< clock[level] <<std::endl;
+	  }else{                  // bulk step
+	    Integ->update_P(U,level,lambda*2.0*eps);
+	    
+	    clock[level]+=2;
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"P "<< clock[level] <<std::endl;
+	  }
+	}
+	
+	
+	
+      }
+      
+    };
+    
+    class LeapFrog{
+    public:
+      void step (LatticeLorentzColourMatrix& U, 
+		 int level, std::vector<int>& clock,
+		 Integrator<LeapFrog>* Integ){
+	// level  : current level
+	// fl     : final level
+	// eps    : current step size
+	
+	int fl = Integ->as.size() -1;
+	double eps = Integ->Params.stepsize;
+	
+	// Get current level step size
+	for(int l=0; l<=level; ++l) eps/= Integ->Nrel[l];
+	
+	int fin = 1;
+	for(int l=0; l<=level; ++l) fin*= Integ->Nrel[l];
+	fin = 2*Integ->Params.MDsteps*fin - 1;
+	
+	for(int e=0; e<Integ->Nrel[level]; ++e){
+	  
+	  if(clock[level] == 0){    // initial half step
+	    Integ->update_P(U, level,eps/2.0);
+	    ++clock[level];
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
+	  }
+	  if(level == fl){          // lowest level
+	    Integ->update_U(U, eps);
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"U "<< 0.5*(clock[level]+1) <<std::endl;
+	  }else{                 // recursive function call
+	    step(U, level+1,clock, Integ);
+	  }
+	  if(clock[level] == fin){  // final half step
+	    Integ->update_P(U, level,eps/2.0);
+	    
+	    ++clock[level];
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
+	  }else{                  // bulk step
+	    Integ->update_P(U, level,eps);
+	    
+	    clock[level]+=2;
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
+	  }
+	}
+
+
+
+
+      }
+    };
+
+
+  }
+}
+
+
+
+#endif//INTEGRATOR_INCLUDED
diff --git a/lib/qcd/hmc/integrators/Integrator_base.h b/lib/qcd/hmc/integrators/Integrator_base.h
deleted file mode 100644
index 9b47a19d..00000000
--- a/lib/qcd/hmc/integrators/Integrator_base.h
+++ /dev/null
@@ -1,280 +0,0 @@
-//--------------------------------------------------------------------
-/*! @file Integrator_base.h
- * @brief Declaration of classes for the abstract Molecular Dynamics integrator
- *
- * @author Guido Cossu
- */
-//--------------------------------------------------------------------
-
-#ifndef INTEGRATOR_INCLUDED
-#define INTEGRATOR_INCLUDED
-
-#include <memory>
-
-class Observer;
-
-
-/*! @brief Abstract base class for Molecular Dynamics management */
-
-namespace Grid{
-  namespace QCD{
-
-    typedef Action<LatticeLorentzColourMatrix>*  ActPtr; // now force the same size as the rest of the code
-    typedef std::vector<ActPtr> ActionLevel;
-    typedef std::vector<ActionLevel> ActionSet;
-    typedef std::vector<Observer*> ObserverList;
-    
-    struct IntegratorParameters{
-      int Nexp;
-      int MDsteps;  // number of outer steps
-      RealD trajL;  // trajectory length 
-      RealD stepsize;
-
-      IntegratorParameters(int Nexp_,
-			   int MDsteps_, 
-			   RealD trajL_):
-	Nexp(Nexp_),MDsteps(MDsteps_),trajL(trajL_),stepsize(trajL/MDsteps){};
-    };
-
-
-    namespace MDutils{
-      void generate_momenta(LatticeLorentzColourMatrix&,GridParallelRNG&);
-      void generate_momenta_su3(LatticeLorentzColourMatrix&,GridParallelRNG&);
-    }
-
-    
-    template< class IntegratorPolicy >
-    class Integrator{
-    private:
-      IntegratorParameters Params;
-      const ActionSet as;
-      const std::vector<int> Nrel; //relative step size per level
-      std::unique_ptr<LatticeLorentzColourMatrix> P;
-      GridParallelRNG pRNG;
-      //ObserverList observers; // not yet
-     
-      IntegratorPolicy TheIntegrator;
-
-      void register_observers();
-      void notify_observers();
-
-      void update_P(LatticeLorentzColourMatrix&U, int level,double ep){
-	for(int a=0; a<as[level].size(); ++a){
-	  LatticeLorentzColourMatrix force(U._grid);
-	  as[level].at(a)->deriv(U,force);
-	  *P -= force*ep;
-	}
-      }
-
-
-      void update_U(LatticeLorentzColourMatrix&U, double ep){
-	//rewrite exponential to deal automatically  with the lorentz index?
-	LatticeColourMatrix Umu(U._grid);
-	LatticeColourMatrix Pmu(U._grid);
-	for (int mu = 0; mu < Nd; mu++){
-	  Umu=peekLorentz(U, mu);
-	  Pmu=peekLorentz(*P, mu);
-	  Umu = expMat(Pmu, ep, Params.Nexp)*Umu;
-	  pokeLorentz(U, Umu, mu);
-	}
-
-      }
-      
-
-      
-      friend void IntegratorPolicy::step (LatticeLorentzColourMatrix& U, 
-				     int level, std::vector<int>& clock,
-				     Integrator<IntegratorPolicy>* Integ);
-    public:
-    Integrator(GridBase* grid, IntegratorParameters Par,
-		 ActionSet& Aset, std::vector<int> Nrel_):
-      Params(Par),as(Aset),Nrel(Nrel_),P(new LatticeLorentzColourMatrix(grid)),pRNG(grid){
-	assert(as.size() == Nrel.size());
-	pRNG.SeedRandomDevice();
-      };
-      
-      ~Integrator(){}
-
-
-      //Initialization of momenta and actions
-      void init(LatticeLorentzColourMatrix& U){
-	std::cout<< "Integrator init\n";
-
-	MDutils::generate_momenta(*P,pRNG);
-	for(int level=0; level< as.size(); ++level){
-	  for(int actionID=0; actionID<as.at(level).size(); ++actionID){
-	    as[level].at(actionID)->init(U, pRNG);
-	  }
-	}
-      }
-
-      
-      // Calculate action
-      RealD S(LatticeLorentzColourMatrix& U){
-	LatticeComplex Hloc(U._grid);
-	Hloc = zero;
-	// Momenta
-	for (int mu=0; mu <Nd; mu++){
-	  LatticeColourMatrix Pmu = peekLorentz(*P, mu);
-	  Hloc -= trace(Pmu*Pmu);
-	}
-	Complex Hsum = sum(Hloc);
-	
-	RealD H = Hsum.real();
-
-	std::cout << "H_p = "<< H << "\n";
-
-	// Actions
-	for(int level=0; level<as.size(); ++level)
-	  for(int actionID=0; actionID<as.at(level).size(); ++actionID)
-	    H += as[level].at(actionID)->S(U);
-	
-	return H;
-      }
-      
-      
-
-      void integrate(LatticeLorentzColourMatrix& U, int level){
-	std::vector<int> clock;
-	clock.resize(as.size(),0);
-	for(int step=0; step< Params.MDsteps; ++step)   // MD step
-	  TheIntegrator.step(U,0,clock, (this));
-      }
-    };
-    
-
-    class MinimumNorm2{
-      const double lambda = 0.1931833275037836;
-    public:
-      void step (LatticeLorentzColourMatrix& U, 
-		 int level, std::vector<int>& clock,
-		 Integrator<MinimumNorm2>* Integ){
-	// level  : current level
-	// fl  : final level
-	// eps : current step size
-
-	int fl = Integ->as.size() -1;
-	double eps = Integ->Params.stepsize;
-	
-	for(int l=0; l<=level; ++l) eps/= 2.0*Integ->Nrel[l];
-	
-	int fin = Integ->Nrel[0];
-	for(int l=1; l<=level; ++l) fin*= 2.0*Integ->Nrel[l];
-	fin = 3*Integ->Params.MDsteps*fin -1;
-	
-	
-	for(int e=0; e<Integ->Nrel[level]; ++e){
-	  
-	  if(clock[level] == 0){    // initial half step 
-	    Integ->update_P(U,level,lambda*eps);
-	    ++clock[level];
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"P "<< clock[level] <<std::endl;
-	  }
-	  
-	  if(level == fl){          // lowest level 
-	    Integ->update_U(U,0.5*eps);
-	    
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"U "<< (clock[level]+1) <<std::endl;
-	  }else{                 // recursive function call 
-	    step(U,level+1,clock, Integ);
-	  }
-	  
-	  Integ->update_P(U,level,(1.0-2.0*lambda)*eps);
-	  ++clock[level];
-	  for(int l=0; l<level;++l) std::cout<<"   ";
-	  std::cout<<"P "<< (clock[level]) <<std::endl;
-	  
-	  if(level == fl){          // lowest level 
-	    Integ->update_U(U,0.5*eps);
-	    
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"U "<< (clock[level]+1) <<std::endl;
-	  }else{                 // recursive function call 
-	    step(U,level+1,clock, Integ);
-	  }    
-	  
-	  
-	  if(clock[level] == fin){  // final half step
-	    Integ->update_P(U,level,lambda*eps);
-	    
-	    ++clock[level];
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"P "<< clock[level] <<std::endl;
-	  }else{                  // bulk step
-	    Integ->update_P(U,level,lambda*2.0*eps);
-	    
-	    clock[level]+=2;
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"P "<< clock[level] <<std::endl;
-	  }
-	}
-	
-	
-	
-      }
-      
-    };
-    
-    class LeapFrog{
-    public:
-      void step (LatticeLorentzColourMatrix& U, 
-		 int level, std::vector<int>& clock,
-		 Integrator<LeapFrog>* Integ){
-	// fl  : final level
-	// eps : current step size
-	
-	int fl = Integ->as.size() -1;
-	double eps = Integ->Params.stepsize;
-	
-	// Get current level step size
-	for(int l=0; l<=level; ++l) eps/= Integ->Nrel[l];
-	
-	int fin = 1;
-	for(int l=0; l<=level; ++l) fin*= Integ->Nrel[l];
-	fin = 2*Integ->Params.MDsteps*fin - 1;
-	
-	for(int e=0; e<Integ->Nrel[level]; ++e){
-	  
-	  if(clock[level] == 0){    // initial half step
-	    Integ->update_P(U, level,eps/2.0);
-	    ++clock[level];
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
-	  }
-	  if(level == fl){          // lowest level
-	    Integ->update_U(U, eps);
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"U "<< 0.5*(clock[level]+1) <<std::endl;
-	  }else{                 // recursive function call
-	    step(U, level+1,clock, Integ);
-	  }
-	  if(clock[level] == fin){  // final half step
-	    Integ->update_P(U, level,eps/2.0);
-	    
-	    ++clock[level];
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
-	  }else{                  // bulk step
-	    Integ->update_P(U, level,eps);
-	    
-	    clock[level]+=2;
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
-	  }
-	}
-
-
-
-
-      }
-    };
-
-
-
-
-    
-  }
-}
-#endif//INTEGRATOR_INCLUDED
diff --git a/tests/Test_hmc_WilsonGauge.cc b/tests/Test_hmc_WilsonGauge.cc
index 98f75b3f..9b9bd388 100644
--- a/tests/Test_hmc_WilsonGauge.cc
+++ b/tests/Test_hmc_WilsonGauge.cc
@@ -40,7 +40,7 @@ int main (int argc, char ** argv)
 
   // Create integrator
   typedef MinimumNorm2  IntegratorAlgorithm;// change here to change the algorithm
-  IntegratorParameters MDpar(12,30,1.0);
+  IntegratorParameters MDpar(12,5,1.0);
   std::vector<int> rel ={1};
   Integrator<IntegratorAlgorithm> MDynamics(&Fine,MDpar, FullSet,rel);
 

From 19a1ffedcc11877e98e60c5a0ad36b3c0f46b450 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Mon, 6 Jul 2015 18:32:20 +0900
Subject: [PATCH 386/429] Simplifying HMC syntax for the final user

---
 lib/GridConfig.h                              |  4 +--
 lib/qcd/hmc/integrators/Integrator.h          | 28 +++++++++++--------
 .../hmc/integrators/Integrator_algorithm.h    | 14 +++++-----
 tests/Test_hmc_WilsonGauge.cc                 |  5 ++--
 4 files changed, 28 insertions(+), 23 deletions(-)

diff --git a/lib/GridConfig.h b/lib/GridConfig.h
index f3a1be6a..45d07e55 100644
--- a/lib/GridConfig.h
+++ b/lib/GridConfig.h
@@ -20,10 +20,10 @@
 #define GRID_COMMS_NONE 1
 
 /* GRID_DEFAULT_PRECISION is DOUBLE */
-/* #undef GRID_DEFAULT_PRECISION_DOUBLE */
+#define GRID_DEFAULT_PRECISION_DOUBLE 1
 
 /* GRID_DEFAULT_PRECISION is SINGLE */
-#define GRID_DEFAULT_PRECISION_SINGLE 1
+/* #undef GRID_DEFAULT_PRECISION_SINGLE */
 
 /* Support Altivec instructions */
 /* #undef HAVE_ALTIVEC */
diff --git a/lib/qcd/hmc/integrators/Integrator.h b/lib/qcd/hmc/integrators/Integrator.h
index 5eb319b2..522991ac 100644
--- a/lib/qcd/hmc/integrators/Integrator.h
+++ b/lib/qcd/hmc/integrators/Integrator.h
@@ -17,7 +17,15 @@ namespace Grid{
   namespace QCD{
 
     typedef Action<LatticeLorentzColourMatrix>*  ActPtr; // now force the same colours as the rest of the code
-    typedef std::vector<ActPtr> ActionLevel;
+    struct ActionLevel{
+      int multiplier;
+    public:
+      std::vector<ActPtr> actions;
+      explicit ActionLevel(int mul = 1):multiplier(mul){assert (mul > 0);};
+      void push_back(ActPtr ptr){
+	actions.push_back(ptr);
+      }
+    };
     typedef std::vector<ActionLevel> ActionSet;
     typedef std::vector<Observer*> ObserverList;
     
@@ -45,7 +53,6 @@ namespace Grid{
     private:
       IntegratorParameters Params;
       const ActionSet as;
-      const std::vector<int> Nrel; //relative step size per level
       std::unique_ptr<LatticeLorentzColourMatrix> P;
       GridParallelRNG pRNG;
       //ObserverList observers; // not yet
@@ -56,9 +63,9 @@ namespace Grid{
       void notify_observers();
 
       void update_P(LatticeLorentzColourMatrix&U, int level,double ep){
-	for(int a=0; a<as[level].size(); ++a){
+	for(int a=0; a<as[level].actions.size(); ++a){
 	  LatticeLorentzColourMatrix force(U._grid);
-	  as[level].at(a)->deriv(U,force);
+	  as[level].actions.at(a)->deriv(U,force);
 	  *P -= force*ep;
 	}
       }
@@ -84,9 +91,8 @@ namespace Grid{
 					     Integrator<IntegratorAlgorithm>* Integ);
     public:
     Integrator(GridBase* grid, IntegratorParameters Par,
-		 ActionSet& Aset, std::vector<int> Nrel_):
-      Params(Par),as(Aset),Nrel(Nrel_),P(new LatticeLorentzColourMatrix(grid)),pRNG(grid){
-	assert(as.size() == Nrel.size());
+		 ActionSet& Aset):
+      Params(Par),as(Aset),P(new LatticeLorentzColourMatrix(grid)),pRNG(grid){
 	pRNG.SeedRandomDevice();
       };
       
@@ -99,8 +105,8 @@ namespace Grid{
 
 	MDutils::generate_momenta(*P,pRNG);
 	for(int level=0; level< as.size(); ++level){
-	  for(int actionID=0; actionID<as.at(level).size(); ++actionID){
-	    as[level].at(actionID)->init(U, pRNG);
+	  for(int actionID=0; actionID<as[level].actions.size(); ++actionID){
+	    as[level].actions.at(actionID)->init(U, pRNG);
 	  }
 	}
       }
@@ -123,8 +129,8 @@ namespace Grid{
 
 	// Actions
 	for(int level=0; level<as.size(); ++level)
-	  for(int actionID=0; actionID<as.at(level).size(); ++actionID)
-	    H += as[level].at(actionID)->S(U);
+	  for(int actionID=0; actionID<as[level].actions.size(); ++actionID)
+	    H += as[level].actions.at(actionID)->S(U);
 	
 	return H;
       }
diff --git a/lib/qcd/hmc/integrators/Integrator_algorithm.h b/lib/qcd/hmc/integrators/Integrator_algorithm.h
index 0dce2978..c192b404 100644
--- a/lib/qcd/hmc/integrators/Integrator_algorithm.h
+++ b/lib/qcd/hmc/integrators/Integrator_algorithm.h
@@ -27,14 +27,14 @@ namespace Grid{
 	int fl = Integ->as.size() -1;
 	double eps = Integ->Params.stepsize;
 	
-	for(int l=0; l<=level; ++l) eps/= 2.0*Integ->Nrel[l];
+	for(int l=0; l<=level; ++l) eps/= 2.0*Integ->as[l].multiplier;
 	
-	int fin = Integ->Nrel[0];
-	for(int l=1; l<=level; ++l) fin*= 2.0*Integ->Nrel[l];
+	int fin = Integ->as[0].multiplier;
+	for(int l=1; l<=level; ++l) fin*= 2.0*Integ->as[l].multiplier;
 	fin = 3*Integ->Params.MDsteps*fin -1;
 	
 	
-	for(int e=0; e<Integ->Nrel[level]; ++e){
+	for(int e=0; e<Integ->as[level].multiplier; ++e){
 	  
 	  if(clock[level] == 0){    // initial half step 
 	    Integ->update_P(U,level,lambda*eps);
@@ -101,13 +101,13 @@ namespace Grid{
 	double eps = Integ->Params.stepsize;
 	
 	// Get current level step size
-	for(int l=0; l<=level; ++l) eps/= Integ->Nrel[l];
+	for(int l=0; l<=level; ++l) eps/= Integ->as[l].multiplier;
 	
 	int fin = 1;
-	for(int l=0; l<=level; ++l) fin*= Integ->Nrel[l];
+	for(int l=0; l<=level; ++l) fin*= Integ->as[l].multiplier;
 	fin = 2*Integ->Params.MDsteps*fin - 1;
 	
-	for(int e=0; e<Integ->Nrel[level]; ++e){
+	for(int e=0; e<Integ->as[level].multiplier; ++e){
 	  
 	  if(clock[level] == 0){    // initial half step
 	    Integ->update_P(U, level,eps/2.0);
diff --git a/tests/Test_hmc_WilsonGauge.cc b/tests/Test_hmc_WilsonGauge.cc
index 9b9bd388..d39ec9b9 100644
--- a/tests/Test_hmc_WilsonGauge.cc
+++ b/tests/Test_hmc_WilsonGauge.cc
@@ -39,10 +39,9 @@ int main (int argc, char ** argv)
   FullSet.push_back(Level1);
 
   // Create integrator
-  typedef MinimumNorm2  IntegratorAlgorithm;// change here to change the algorithm
+  typedef MinimumNorm2  IntegratorAlgorithm;// change here to modify the algorithm
   IntegratorParameters MDpar(12,5,1.0);
-  std::vector<int> rel ={1};
-  Integrator<IntegratorAlgorithm> MDynamics(&Fine,MDpar, FullSet,rel);
+  Integrator<IntegratorAlgorithm> MDynamics(&Fine,MDpar, FullSet);
 
   // Create HMC
   HMCparameters HMCpar;

From 48ae886c3202710bc079a07ea64bd5031209127f Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Tue, 7 Jul 2015 14:59:37 +0900
Subject: [PATCH 387/429] Cleaning up files for HMC

---
 lib/qcd/hmc/HMC.h                    | 19 +++++++++----------
 lib/qcd/hmc/integrators/Integrator.h | 27 ++++++++++++++-------------
 2 files changed, 23 insertions(+), 23 deletions(-)

diff --git a/lib/qcd/hmc/HMC.h b/lib/qcd/hmc/HMC.h
index c0f06b64..2da8dd0f 100644
--- a/lib/qcd/hmc/HMC.h
+++ b/lib/qcd/hmc/HMC.h
@@ -1,15 +1,16 @@
 //--------------------------------------------------------------------
 /*! @file HMC.h
- * @brief Declaration of classes for Hybrid Monte Carlo update
+ * @brief Classes for Hybrid Monte Carlo update
  *
  * @author Guido Cossu
+ * Time-stamp: <2015-07-07 14:58:13 neo>
  */
 //--------------------------------------------------------------------
 #ifndef HMC_INCLUDED
 #define HMC_INCLUDED
 
 #include <string>
-#include <memory>
+
 
 namespace Grid{
   namespace QCD{
@@ -20,7 +21,7 @@ namespace Grid{
       Integer ThermalizationSteps;
       Integer StartingConfig;
       Integer SaveInterval; //Setting to 0 does not save configurations
-      std::string Filename_prefix; // To save configurations
+      std::string Filename_prefix; // To save configurations and rng seed
       
       HMCparameters();
     };
@@ -49,8 +50,7 @@ namespace Grid{
 	}
       }
 
-      RealD evolve_step(LatticeLorentzColourMatrix& U){
-
+      RealD evolve_step(LatticeGaugeField& U){
 	MD.init(U); // set U and initialize P and phi's 
 	RealD H0 = MD.S(U); // initial state action  
 	std::cout<<"Total H before = "<< H0 << "\n";
@@ -74,9 +74,7 @@ namespace Grid{
       }
       ~HybridMonteCarlo(){};
 
-
-
-      void evolve(LatticeLorentzColourMatrix& Uin){
+      void evolve(LatticeGaugeField& Uin){
 	Real DeltaH;
 	
 	// Thermalizations
@@ -88,7 +86,7 @@ namespace Grid{
 	}
 
 	// Actual updates (evolve a copy Ucopy then copy back eventually)
-	LatticeLorentzColourMatrix Ucopy(Uin._grid);
+	LatticeGaugeField Ucopy(Uin._grid);
 	for(int iter=Params.StartingConfig; 
 	    iter < Params.Nsweeps+Params.StartingConfig; ++iter){
 	  std::cout << "-- # Sweep = "<< iter <<  "\n";
@@ -97,7 +95,8 @@ namespace Grid{
 	  DeltaH = evolve_step(Ucopy);
 		
 	  if(metropolis_test(DeltaH)) Uin = Ucopy;
-	  //need sync?
+	  
+	  // here save config and RNG seed
 	}
       }
     };
diff --git a/lib/qcd/hmc/integrators/Integrator.h b/lib/qcd/hmc/integrators/Integrator.h
index 522991ac..f752baab 100644
--- a/lib/qcd/hmc/integrators/Integrator.h
+++ b/lib/qcd/hmc/integrators/Integrator.h
@@ -1,8 +1,9 @@
 //--------------------------------------------------------------------
 /*! @file Integrator.h
- * @brief Declaration of classes for the Molecular Dynamics integrator
+ * @brief Classes for the Molecular Dynamics integrator
  *
  * @author Guido Cossu
+ * Time-stamp: <2015-07-07 14:58:40 neo>
  */
 //--------------------------------------------------------------------
 
@@ -16,7 +17,7 @@ class Observer;
 namespace Grid{
   namespace QCD{
 
-    typedef Action<LatticeLorentzColourMatrix>*  ActPtr; // now force the same colours as the rest of the code
+    typedef Action<LatticeGaugeField>*  ActPtr; // now force the same colours as the rest of the code
     struct ActionLevel{
       int multiplier;
     public:
@@ -43,8 +44,8 @@ namespace Grid{
 
 
     namespace MDutils{
-      void generate_momenta(LatticeLorentzColourMatrix&,GridParallelRNG&);
-      void generate_momenta_su3(LatticeLorentzColourMatrix&,GridParallelRNG&);
+      void generate_momenta(LatticeGaugeField&,GridParallelRNG&);
+      void generate_momenta_su3(LatticeGaugeField&,GridParallelRNG&);
     }
 
     /*! @brief Class for Molecular Dynamics management */   
@@ -53,7 +54,7 @@ namespace Grid{
     private:
       IntegratorParameters Params;
       const ActionSet as;
-      std::unique_ptr<LatticeLorentzColourMatrix> P;
+      std::unique_ptr<LatticeGaugeField> P;
       GridParallelRNG pRNG;
       //ObserverList observers; // not yet
      
@@ -62,16 +63,16 @@ namespace Grid{
       void register_observers();
       void notify_observers();
 
-      void update_P(LatticeLorentzColourMatrix&U, int level,double ep){
+      void update_P(LatticeGaugeField&U, int level,double ep){
 	for(int a=0; a<as[level].actions.size(); ++a){
-	  LatticeLorentzColourMatrix force(U._grid);
+	  LatticeGaugeField force(U._grid);
 	  as[level].actions.at(a)->deriv(U,force);
 	  *P -= force*ep;
 	}
       }
 
 
-      void update_U(LatticeLorentzColourMatrix&U, double ep){
+      void update_U(LatticeGaugeField&U, double ep){
 	//rewrite exponential to deal automatically  with the lorentz index?
 	LatticeColourMatrix Umu(U._grid);
 	LatticeColourMatrix Pmu(U._grid);
@@ -86,13 +87,13 @@ namespace Grid{
       
 
       
-      friend void IntegratorAlgorithm::step (LatticeLorentzColourMatrix& U, 
+      friend void IntegratorAlgorithm::step (LatticeGaugeField& U, 
 					     int level, std::vector<int>& clock,
 					     Integrator<IntegratorAlgorithm>* Integ);
     public:
     Integrator(GridBase* grid, IntegratorParameters Par,
 		 ActionSet& Aset):
-      Params(Par),as(Aset),P(new LatticeLorentzColourMatrix(grid)),pRNG(grid){
+      Params(Par),as(Aset),P(new LatticeGaugeField(grid)),pRNG(grid){
 	pRNG.SeedRandomDevice();
       };
       
@@ -100,7 +101,7 @@ namespace Grid{
 
 
       //Initialization of momenta and actions
-      void init(LatticeLorentzColourMatrix& U){
+      void init(LatticeGaugeField& U){
 	std::cout<< "Integrator init\n";
 
 	MDutils::generate_momenta(*P,pRNG);
@@ -113,7 +114,7 @@ namespace Grid{
 
       
       // Calculate action
-      RealD S(LatticeLorentzColourMatrix& U){
+      RealD S(LatticeGaugeField& U){
 	LatticeComplex Hloc(U._grid);
 	Hloc = zero;
 	// Momenta
@@ -135,7 +136,7 @@ namespace Grid{
 	return H;
       }
 
-      void integrate(LatticeLorentzColourMatrix& U){
+      void integrate(LatticeGaugeField& U){
 	std::vector<int> clock;
 	clock.resize(as.size(),0);
 	for(int step=0; step< Params.MDsteps; ++step)   // MD step

From ab509d3f8e351e5463472e994dcae0776ed3d803 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 11 Jul 2015 23:06:31 +0900
Subject: [PATCH 388/429] This file drives me crazy

---
 INSTALL | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/INSTALL b/INSTALL
index f812f5a2..80a61507 120000
--- a/INSTALL
+++ b/INSTALL
@@ -1 +1 @@
-/usr/share/automake-1.14/INSTALL
\ No newline at end of file
+/opt/local/share/automake-1.15/INSTALL
\ No newline at end of file

From ce7c24989d6c7c0acbfb7cb6e5c1e7dce37890ef Mon Sep 17 00:00:00 2001
From: Antonin Portelli <antonin.portelli@me.com>
Date: Fri, 17 Jul 2015 11:15:17 +0900
Subject: [PATCH 389/429] gitignore update

---
 .gitignore |   66 +-
 INSTALL    |    1 -
 configure  | 8463 ----------------------------------------------------
 3 files changed, 42 insertions(+), 8488 deletions(-)
 delete mode 120000 INSTALL
 delete mode 100755 configure

diff --git a/.gitignore b/.gitignore
index a68020a3..fb988e6b 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,56 +1,65 @@
-# Compiled Object files
+# Compiled Object files #
+#########################
 *.slo
 *.lo
 *.o
 *.obj
-*~
-errs
-*#
-INSTALL
 
-# Precompiled Headers
+
+# Editor files #
+################
+*~
+*#
+
+# Precompiled Headers #
+#######################
 *.gch
 *.pch
 
-# Compiled Dynamic libraries
+# Compiled Dynamic libraries #
+##############################
 *.so
 *.dylib
 *.dll
 
-# Fortran module files
+# Fortran module files #
+########################
 *.mod
 
-# Compiled Static libraries
+# Compiled Static libraries #
+#############################
 *.lai
 *.la
 *.a
 *.lib
 
-# Executables
+# Executables #
+###############
 *.exe
 *.out
 *.app
-# http://www.gnu.org/software/automake
 
+# http://www.gnu.org/software/automake #
+########################################
 Makefile.in
 Makefile
 config.log
 config.status
 .deps
 
-# http://www.gnu.org/software/autoconf
-
-/autom4te.cache
-/aclocal.m4
-/compile
-/configure
-/depcomp
-/install-sh
-/missing
-/stamp-h1
-/config.sub
-/config.guess
-/INSTALL
+# http://www.gnu.org/software/autoconf #
+########################################
+autom4te.cache
+aclocal.m4
+compile
+configure
+depcomp
+install-sh
+missing
+stamp-h1
+config.sub
+config.guess
+INSTALL
 
 # Packages #
 ############
@@ -80,3 +89,12 @@ config.status
 .Trashes
 ehthumbs.db
 Thumbs.db
+
+# build directory #
+###################
+build/*
+
+# IDE related files #
+#####################
+*.xcodeproj/*
+build.sh
diff --git a/INSTALL b/INSTALL
deleted file mode 120000
index f812f5a2..00000000
--- a/INSTALL
+++ /dev/null
@@ -1 +0,0 @@
-/usr/share/automake-1.14/INSTALL
\ No newline at end of file
diff --git a/configure b/configure
deleted file mode 100755
index 935f1018..00000000
--- a/configure
+++ /dev/null
@@ -1,8463 +0,0 @@
-#! /bin/sh
-# Guess values for system-dependent variables and create Makefiles.
-# Generated by GNU Autoconf 2.69 for Grid 1.0.
-#
-# Report bugs to <paboyle@ph.ed.ac.uk>.
-#
-#
-# Copyright (C) 1992-1996, 1998-2012 Free Software Foundation, Inc.
-#
-#
-# This configure script is free software; the Free Software Foundation
-# gives unlimited permission to copy, distribute and modify it.
-## -------------------- ##
-## M4sh Initialization. ##
-## -------------------- ##
-
-# Be more Bourne compatible
-DUALCASE=1; export DUALCASE # for MKS sh
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
-  emulate sh
-  NULLCMD=:
-  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
-  # is contrary to our usage.  Disable this feature.
-  alias -g '${1+"$@"}'='"$@"'
-  setopt NO_GLOB_SUBST
-else
-  case `(set -o) 2>/dev/null` in #(
-  *posix*) :
-    set -o posix ;; #(
-  *) :
-     ;;
-esac
-fi
-
-
-as_nl='
-'
-export as_nl
-# Printing a long string crashes Solaris 7 /usr/bin/printf.
-as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
-as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
-as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
-# Prefer a ksh shell builtin over an external printf program on Solaris,
-# but without wasting forks for bash or zsh.
-if test -z "$BASH_VERSION$ZSH_VERSION" \
-    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
-  as_echo='print -r --'
-  as_echo_n='print -rn --'
-elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
-  as_echo='printf %s\n'
-  as_echo_n='printf %s'
-else
-  if test "X`(/usr/ucb/echo -n -n $as_echo) 2>/dev/null`" = "X-n $as_echo"; then
-    as_echo_body='eval /usr/ucb/echo -n "$1$as_nl"'
-    as_echo_n='/usr/ucb/echo -n'
-  else
-    as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
-    as_echo_n_body='eval
-      arg=$1;
-      case $arg in #(
-      *"$as_nl"*)
-	expr "X$arg" : "X\\(.*\\)$as_nl";
-	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
-      esac;
-      expr "X$arg" : "X\\(.*\\)" | tr -d "$as_nl"
-    '
-    export as_echo_n_body
-    as_echo_n='sh -c $as_echo_n_body as_echo'
-  fi
-  export as_echo_body
-  as_echo='sh -c $as_echo_body as_echo'
-fi
-
-# The user is always right.
-if test "${PATH_SEPARATOR+set}" != set; then
-  PATH_SEPARATOR=:
-  (PATH='/bin;/bin'; FPATH=$PATH; sh -c :) >/dev/null 2>&1 && {
-    (PATH='/bin:/bin'; FPATH=$PATH; sh -c :) >/dev/null 2>&1 ||
-      PATH_SEPARATOR=';'
-  }
-fi
-
-
-# IFS
-# We need space, tab and new line, in precisely that order.  Quoting is
-# there to prevent editors from complaining about space-tab.
-# (If _AS_PATH_WALK were called with IFS unset, it would disable word
-# splitting by setting IFS to empty value.)
-IFS=" ""	$as_nl"
-
-# Find who we are.  Look in the path if we contain no directory separator.
-as_myself=
-case $0 in #((
-  *[\\/]* ) as_myself=$0 ;;
-  *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
-  done
-IFS=$as_save_IFS
-
-     ;;
-esac
-# We did not find ourselves, most probably we were run as `sh COMMAND'
-# in which case we are not to be found in the path.
-if test "x$as_myself" = x; then
-  as_myself=$0
-fi
-if test ! -f "$as_myself"; then
-  $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
-  exit 1
-fi
-
-# Unset variables that we do not need and which cause bugs (e.g. in
-# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
-# suppresses any "Segmentation fault" message there.  '((' could
-# trigger a bug in pdksh 5.2.14.
-for as_var in BASH_ENV ENV MAIL MAILPATH
-do eval test x\${$as_var+set} = xset \
-  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
-done
-PS1='$ '
-PS2='> '
-PS4='+ '
-
-# NLS nuisances.
-LC_ALL=C
-export LC_ALL
-LANGUAGE=C
-export LANGUAGE
-
-# CDPATH.
-(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
-
-# Use a proper internal environment variable to ensure we don't fall
-  # into an infinite loop, continuously re-executing ourselves.
-  if test x"${_as_can_reexec}" != xno && test "x$CONFIG_SHELL" != x; then
-    _as_can_reexec=no; export _as_can_reexec;
-    # We cannot yet assume a decent shell, so we have to provide a
-# neutralization value for shells without unset; and this also
-# works around shells that cannot unset nonexistent variables.
-# Preserve -v and -x to the replacement shell.
-BASH_ENV=/dev/null
-ENV=/dev/null
-(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
-case $- in # ((((
-  *v*x* | *x*v* ) as_opts=-vx ;;
-  *v* ) as_opts=-v ;;
-  *x* ) as_opts=-x ;;
-  * ) as_opts= ;;
-esac
-exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
-# Admittedly, this is quite paranoid, since all the known shells bail
-# out after a failed `exec'.
-$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
-as_fn_exit 255
-  fi
-  # We don't want this to propagate to other subprocesses.
-          { _as_can_reexec=; unset _as_can_reexec;}
-if test "x$CONFIG_SHELL" = x; then
-  as_bourne_compatible="if test -n \"\${ZSH_VERSION+set}\" && (emulate sh) >/dev/null 2>&1; then :
-  emulate sh
-  NULLCMD=:
-  # Pre-4.2 versions of Zsh do word splitting on \${1+\"\$@\"}, which
-  # is contrary to our usage.  Disable this feature.
-  alias -g '\${1+\"\$@\"}'='\"\$@\"'
-  setopt NO_GLOB_SUBST
-else
-  case \`(set -o) 2>/dev/null\` in #(
-  *posix*) :
-    set -o posix ;; #(
-  *) :
-     ;;
-esac
-fi
-"
-  as_required="as_fn_return () { (exit \$1); }
-as_fn_success () { as_fn_return 0; }
-as_fn_failure () { as_fn_return 1; }
-as_fn_ret_success () { return 0; }
-as_fn_ret_failure () { return 1; }
-
-exitcode=0
-as_fn_success || { exitcode=1; echo as_fn_success failed.; }
-as_fn_failure && { exitcode=1; echo as_fn_failure succeeded.; }
-as_fn_ret_success || { exitcode=1; echo as_fn_ret_success failed.; }
-as_fn_ret_failure && { exitcode=1; echo as_fn_ret_failure succeeded.; }
-if ( set x; as_fn_ret_success y && test x = \"\$1\" ); then :
-
-else
-  exitcode=1; echo positional parameters were not saved.
-fi
-test x\$exitcode = x0 || exit 1
-test -x / || exit 1"
-  as_suggested="  as_lineno_1=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_1a=\$LINENO
-  as_lineno_2=";as_suggested=$as_suggested$LINENO;as_suggested=$as_suggested" as_lineno_2a=\$LINENO
-  eval 'test \"x\$as_lineno_1'\$as_run'\" != \"x\$as_lineno_2'\$as_run'\" &&
-  test \"x\`expr \$as_lineno_1'\$as_run' + 1\`\" = \"x\$as_lineno_2'\$as_run'\"' || exit 1
-test \$(( 1 + 1 )) = 2 || exit 1"
-  if (eval "$as_required") 2>/dev/null; then :
-  as_have_required=yes
-else
-  as_have_required=no
-fi
-  if test x$as_have_required = xyes && (eval "$as_suggested") 2>/dev/null; then :
-
-else
-  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-as_found=false
-for as_dir in /bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-  as_found=:
-  case $as_dir in #(
-	 /*)
-	   for as_base in sh bash ksh sh5; do
-	     # Try only shells that exist, to save several forks.
-	     as_shell=$as_dir/$as_base
-	     if { test -f "$as_shell" || test -f "$as_shell.exe"; } &&
-		    { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$as_shell"; } 2>/dev/null; then :
-  CONFIG_SHELL=$as_shell as_have_required=yes
-		   if { $as_echo "$as_bourne_compatible""$as_suggested" | as_run=a "$as_shell"; } 2>/dev/null; then :
-  break 2
-fi
-fi
-	   done;;
-       esac
-  as_found=false
-done
-$as_found || { if { test -f "$SHELL" || test -f "$SHELL.exe"; } &&
-	      { $as_echo "$as_bourne_compatible""$as_required" | as_run=a "$SHELL"; } 2>/dev/null; then :
-  CONFIG_SHELL=$SHELL as_have_required=yes
-fi; }
-IFS=$as_save_IFS
-
-
-      if test "x$CONFIG_SHELL" != x; then :
-  export CONFIG_SHELL
-             # We cannot yet assume a decent shell, so we have to provide a
-# neutralization value for shells without unset; and this also
-# works around shells that cannot unset nonexistent variables.
-# Preserve -v and -x to the replacement shell.
-BASH_ENV=/dev/null
-ENV=/dev/null
-(unset BASH_ENV) >/dev/null 2>&1 && unset BASH_ENV ENV
-case $- in # ((((
-  *v*x* | *x*v* ) as_opts=-vx ;;
-  *v* ) as_opts=-v ;;
-  *x* ) as_opts=-x ;;
-  * ) as_opts= ;;
-esac
-exec $CONFIG_SHELL $as_opts "$as_myself" ${1+"$@"}
-# Admittedly, this is quite paranoid, since all the known shells bail
-# out after a failed `exec'.
-$as_echo "$0: could not re-execute with $CONFIG_SHELL" >&2
-exit 255
-fi
-
-    if test x$as_have_required = xno; then :
-  $as_echo "$0: This script requires a shell more modern than all"
-  $as_echo "$0: the shells that I found on your system."
-  if test x${ZSH_VERSION+set} = xset ; then
-    $as_echo "$0: In particular, zsh $ZSH_VERSION has bugs and should"
-    $as_echo "$0: be upgraded to zsh 4.3.4 or later."
-  else
-    $as_echo "$0: Please tell bug-autoconf@gnu.org and
-$0: paboyle@ph.ed.ac.uk about your system, including any
-$0: error possibly output before this message. Then install
-$0: a modern shell, or manually run the script under such a
-$0: shell if you do have one."
-  fi
-  exit 1
-fi
-fi
-fi
-SHELL=${CONFIG_SHELL-/bin/sh}
-export SHELL
-# Unset more variables known to interfere with behavior of common tools.
-CLICOLOR_FORCE= GREP_OPTIONS=
-unset CLICOLOR_FORCE GREP_OPTIONS
-
-## --------------------- ##
-## M4sh Shell Functions. ##
-## --------------------- ##
-# as_fn_unset VAR
-# ---------------
-# Portably unset VAR.
-as_fn_unset ()
-{
-  { eval $1=; unset $1;}
-}
-as_unset=as_fn_unset
-
-# as_fn_set_status STATUS
-# -----------------------
-# Set $? to STATUS, without forking.
-as_fn_set_status ()
-{
-  return $1
-} # as_fn_set_status
-
-# as_fn_exit STATUS
-# -----------------
-# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
-as_fn_exit ()
-{
-  set +e
-  as_fn_set_status $1
-  exit $1
-} # as_fn_exit
-
-# as_fn_mkdir_p
-# -------------
-# Create "$as_dir" as a directory, including parents if necessary.
-as_fn_mkdir_p ()
-{
-
-  case $as_dir in #(
-  -*) as_dir=./$as_dir;;
-  esac
-  test -d "$as_dir" || eval $as_mkdir_p || {
-    as_dirs=
-    while :; do
-      case $as_dir in #(
-      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
-      *) as_qdir=$as_dir;;
-      esac
-      as_dirs="'$as_qdir' $as_dirs"
-      as_dir=`$as_dirname -- "$as_dir" ||
-$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_dir" : 'X\(//\)[^/]' \| \
-	 X"$as_dir" : 'X\(//\)$' \| \
-	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_dir" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      test -d "$as_dir" && break
-    done
-    test -z "$as_dirs" || eval "mkdir $as_dirs"
-  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
-
-
-} # as_fn_mkdir_p
-
-# as_fn_executable_p FILE
-# -----------------------
-# Test if FILE is an executable regular file.
-as_fn_executable_p ()
-{
-  test -f "$1" && test -x "$1"
-} # as_fn_executable_p
-# as_fn_append VAR VALUE
-# ----------------------
-# Append the text in VALUE to the end of the definition contained in VAR. Take
-# advantage of any shell optimizations that allow amortized linear growth over
-# repeated appends, instead of the typical quadratic growth present in naive
-# implementations.
-if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
-  eval 'as_fn_append ()
-  {
-    eval $1+=\$2
-  }'
-else
-  as_fn_append ()
-  {
-    eval $1=\$$1\$2
-  }
-fi # as_fn_append
-
-# as_fn_arith ARG...
-# ------------------
-# Perform arithmetic evaluation on the ARGs, and store the result in the
-# global $as_val. Take advantage of shells that can avoid forks. The arguments
-# must be portable across $(()) and expr.
-if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
-  eval 'as_fn_arith ()
-  {
-    as_val=$(( $* ))
-  }'
-else
-  as_fn_arith ()
-  {
-    as_val=`expr "$@" || test $? -eq 1`
-  }
-fi # as_fn_arith
-
-
-# as_fn_error STATUS ERROR [LINENO LOG_FD]
-# ----------------------------------------
-# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
-# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
-# script with STATUS, using 1 if that was 0.
-as_fn_error ()
-{
-  as_status=$1; test $as_status -eq 0 && as_status=1
-  if test "$4"; then
-    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
-  fi
-  $as_echo "$as_me: error: $2" >&2
-  as_fn_exit $as_status
-} # as_fn_error
-
-if expr a : '\(a\)' >/dev/null 2>&1 &&
-   test "X`expr 00001 : '.*\(...\)'`" = X001; then
-  as_expr=expr
-else
-  as_expr=false
-fi
-
-if (basename -- /) >/dev/null 2>&1 && test "X`basename -- / 2>&1`" = "X/"; then
-  as_basename=basename
-else
-  as_basename=false
-fi
-
-if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
-  as_dirname=dirname
-else
-  as_dirname=false
-fi
-
-as_me=`$as_basename -- "$0" ||
-$as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
-	 X"$0" : 'X\(//\)$' \| \
-	 X"$0" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X/"$0" |
-    sed '/^.*\/\([^/][^/]*\)\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\/\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\/\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-
-# Avoid depending upon Character Ranges.
-as_cr_letters='abcdefghijklmnopqrstuvwxyz'
-as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
-as_cr_Letters=$as_cr_letters$as_cr_LETTERS
-as_cr_digits='0123456789'
-as_cr_alnum=$as_cr_Letters$as_cr_digits
-
-
-  as_lineno_1=$LINENO as_lineno_1a=$LINENO
-  as_lineno_2=$LINENO as_lineno_2a=$LINENO
-  eval 'test "x$as_lineno_1'$as_run'" != "x$as_lineno_2'$as_run'" &&
-  test "x`expr $as_lineno_1'$as_run' + 1`" = "x$as_lineno_2'$as_run'"' || {
-  # Blame Lee E. McMahon (1931-1989) for sed's syntax.  :-)
-  sed -n '
-    p
-    /[$]LINENO/=
-  ' <$as_myself |
-    sed '
-      s/[$]LINENO.*/&-/
-      t lineno
-      b
-      :lineno
-      N
-      :loop
-      s/[$]LINENO\([^'$as_cr_alnum'_].*\n\)\(.*\)/\2\1\2/
-      t loop
-      s/-\n.*//
-    ' >$as_me.lineno &&
-  chmod +x "$as_me.lineno" ||
-    { $as_echo "$as_me: error: cannot create $as_me.lineno; rerun with a POSIX shell" >&2; as_fn_exit 1; }
-
-  # If we had to re-execute with $CONFIG_SHELL, we're ensured to have
-  # already done that, so ensure we don't try to do so again and fall
-  # in an infinite loop.  This has already happened in practice.
-  _as_can_reexec=no; export _as_can_reexec
-  # Don't try to exec as it changes $[0], causing all sort of problems
-  # (the dirname of $[0] is not the place where we might find the
-  # original and so on.  Autoconf is especially sensitive to this).
-  . "./$as_me.lineno"
-  # Exit status is that of the last command.
-  exit
-}
-
-ECHO_C= ECHO_N= ECHO_T=
-case `echo -n x` in #(((((
--n*)
-  case `echo 'xy\c'` in
-  *c*) ECHO_T='	';;	# ECHO_T is single tab character.
-  xy)  ECHO_C='\c';;
-  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
-       ECHO_T='	';;
-  esac;;
-*)
-  ECHO_N='-n';;
-esac
-
-rm -f conf$$ conf$$.exe conf$$.file
-if test -d conf$$.dir; then
-  rm -f conf$$.dir/conf$$.file
-else
-  rm -f conf$$.dir
-  mkdir conf$$.dir 2>/dev/null
-fi
-if (echo >conf$$.file) 2>/dev/null; then
-  if ln -s conf$$.file conf$$ 2>/dev/null; then
-    as_ln_s='ln -s'
-    # ... but there are two gotchas:
-    # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
-    # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
-    # In both cases, we have to default to `cp -pR'.
-    ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
-      as_ln_s='cp -pR'
-  elif ln conf$$.file conf$$ 2>/dev/null; then
-    as_ln_s=ln
-  else
-    as_ln_s='cp -pR'
-  fi
-else
-  as_ln_s='cp -pR'
-fi
-rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
-rmdir conf$$.dir 2>/dev/null
-
-if mkdir -p . 2>/dev/null; then
-  as_mkdir_p='mkdir -p "$as_dir"'
-else
-  test -d ./-p && rmdir ./-p
-  as_mkdir_p=false
-fi
-
-as_test_x='test -x'
-as_executable_p=as_fn_executable_p
-
-# Sed expression to map a string onto a valid CPP name.
-as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
-
-# Sed expression to map a string onto a valid variable name.
-as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
-
-
-test -n "$DJDIR" || exec 7<&0 </dev/null
-exec 6>&1
-
-# Name of the host.
-# hostname on some systems (SVR3.2, old GNU/Linux) returns a bogus exit status,
-# so uname gets run too.
-ac_hostname=`(hostname || uname -n) 2>/dev/null | sed 1q`
-
-#
-# Initializations.
-#
-ac_default_prefix=/usr/local
-ac_clean_files=
-ac_config_libobj_dir=.
-LIBOBJS=
-cross_compiling=no
-subdirs=
-MFLAGS=
-MAKEFLAGS=
-
-# Identity of this package.
-PACKAGE_NAME='Grid'
-PACKAGE_TARNAME='grid'
-PACKAGE_VERSION='1.0'
-PACKAGE_STRING='Grid 1.0'
-PACKAGE_BUGREPORT='paboyle@ph.ed.ac.uk'
-PACKAGE_URL=''
-
-ac_unique_file="lib/Grid.h"
-# Factoring default headers for most tests.
-ac_includes_default="\
-#include <stdio.h>
-#ifdef HAVE_SYS_TYPES_H
-# include <sys/types.h>
-#endif
-#ifdef HAVE_SYS_STAT_H
-# include <sys/stat.h>
-#endif
-#ifdef STDC_HEADERS
-# include <stdlib.h>
-# include <stddef.h>
-#else
-# ifdef HAVE_STDLIB_H
-#  include <stdlib.h>
-# endif
-#endif
-#ifdef HAVE_STRING_H
-# if !defined STDC_HEADERS && defined HAVE_MEMORY_H
-#  include <memory.h>
-# endif
-# include <string.h>
-#endif
-#ifdef HAVE_STRINGS_H
-# include <strings.h>
-#endif
-#ifdef HAVE_INTTYPES_H
-# include <inttypes.h>
-#endif
-#ifdef HAVE_STDINT_H
-# include <stdint.h>
-#endif
-#ifdef HAVE_UNISTD_H
-# include <unistd.h>
-#endif"
-
-ac_subst_vars='am__EXEEXT_FALSE
-am__EXEEXT_TRUE
-LTLIBOBJS
-LIBOBJS
-enable_latex_docs
-enable_html_docs
-enable_dot
-DOXYGEN_DOC_FALSE
-DOXYGEN_DOC_TRUE
-DOT
-DOXYGEN
-BUILD_COMMS_NONE_FALSE
-BUILD_COMMS_NONE_TRUE
-BUILD_COMMS_MPI_FALSE
-BUILD_COMMS_MPI_TRUE
-EGREP
-GREP
-CXXCPP
-SIMD_FLAGS
-am__fastdepCC_FALSE
-am__fastdepCC_TRUE
-CCDEPMODE
-ac_ct_CC
-CFLAGS
-CC
-HAVE_CXX11
-RANLIB
-OPENMP_CXXFLAGS
-am__fastdepCXX_FALSE
-am__fastdepCXX_TRUE
-CXXDEPMODE
-am__nodep
-AMDEPBACKSLASH
-AMDEP_FALSE
-AMDEP_TRUE
-am__quote
-am__include
-DEPDIR
-OBJEXT
-EXEEXT
-ac_ct_CXX
-CPPFLAGS
-LDFLAGS
-CXXFLAGS
-CXX
-AM_BACKSLASH
-AM_DEFAULT_VERBOSITY
-AM_DEFAULT_V
-AM_V
-am__untar
-am__tar
-AMTAR
-am__leading_dot
-SET_MAKE
-AWK
-mkdir_p
-MKDIR_P
-INSTALL_STRIP_PROGRAM
-STRIP
-install_sh
-MAKEINFO
-AUTOHEADER
-AUTOMAKE
-AUTOCONF
-ACLOCAL
-VERSION
-PACKAGE
-CYGPATH_W
-am__isrc
-INSTALL_DATA
-INSTALL_SCRIPT
-INSTALL_PROGRAM
-target_os
-target_vendor
-target_cpu
-target
-host_os
-host_vendor
-host_cpu
-host
-build_os
-build_vendor
-build_cpu
-build
-target_alias
-host_alias
-build_alias
-LIBS
-ECHO_T
-ECHO_N
-ECHO_C
-DEFS
-mandir
-localedir
-libdir
-psdir
-pdfdir
-dvidir
-htmldir
-infodir
-docdir
-oldincludedir
-includedir
-localstatedir
-sharedstatedir
-sysconfdir
-datadir
-datarootdir
-libexecdir
-sbindir
-bindir
-program_transform_name
-prefix
-exec_prefix
-PACKAGE_URL
-PACKAGE_BUGREPORT
-PACKAGE_STRING
-PACKAGE_VERSION
-PACKAGE_TARNAME
-PACKAGE_NAME
-PATH_SEPARATOR
-SHELL'
-ac_subst_files=''
-ac_user_opts='
-enable_option_checking
-enable_silent_rules
-enable_dependency_tracking
-enable_openmp
-enable_simd
-enable_precision
-enable_comms
-enable_doxygen
-enable_dot
-enable_html_docs
-enable_latex_docs
-'
-      ac_precious_vars='build_alias
-host_alias
-target_alias
-CXX
-CXXFLAGS
-LDFLAGS
-LIBS
-CPPFLAGS
-CCC
-CC
-CFLAGS
-CXXCPP'
-
-
-# Initialize some variables set by options.
-ac_init_help=
-ac_init_version=false
-ac_unrecognized_opts=
-ac_unrecognized_sep=
-# The variables have the same names as the options, with
-# dashes changed to underlines.
-cache_file=/dev/null
-exec_prefix=NONE
-no_create=
-no_recursion=
-prefix=NONE
-program_prefix=NONE
-program_suffix=NONE
-program_transform_name=s,x,x,
-silent=
-site=
-srcdir=
-verbose=
-x_includes=NONE
-x_libraries=NONE
-
-# Installation directory options.
-# These are left unexpanded so users can "make install exec_prefix=/foo"
-# and all the variables that are supposed to be based on exec_prefix
-# by default will actually change.
-# Use braces instead of parens because sh, perl, etc. also accept them.
-# (The list follows the same order as the GNU Coding Standards.)
-bindir='${exec_prefix}/bin'
-sbindir='${exec_prefix}/sbin'
-libexecdir='${exec_prefix}/libexec'
-datarootdir='${prefix}/share'
-datadir='${datarootdir}'
-sysconfdir='${prefix}/etc'
-sharedstatedir='${prefix}/com'
-localstatedir='${prefix}/var'
-includedir='${prefix}/include'
-oldincludedir='/usr/include'
-docdir='${datarootdir}/doc/${PACKAGE_TARNAME}'
-infodir='${datarootdir}/info'
-htmldir='${docdir}'
-dvidir='${docdir}'
-pdfdir='${docdir}'
-psdir='${docdir}'
-libdir='${exec_prefix}/lib'
-localedir='${datarootdir}/locale'
-mandir='${datarootdir}/man'
-
-ac_prev=
-ac_dashdash=
-for ac_option
-do
-  # If the previous option needs an argument, assign it.
-  if test -n "$ac_prev"; then
-    eval $ac_prev=\$ac_option
-    ac_prev=
-    continue
-  fi
-
-  case $ac_option in
-  *=?*) ac_optarg=`expr "X$ac_option" : '[^=]*=\(.*\)'` ;;
-  *=)   ac_optarg= ;;
-  *)    ac_optarg=yes ;;
-  esac
-
-  # Accept the important Cygnus configure options, so we can diagnose typos.
-
-  case $ac_dashdash$ac_option in
-  --)
-    ac_dashdash=yes ;;
-
-  -bindir | --bindir | --bindi | --bind | --bin | --bi)
-    ac_prev=bindir ;;
-  -bindir=* | --bindir=* | --bindi=* | --bind=* | --bin=* | --bi=*)
-    bindir=$ac_optarg ;;
-
-  -build | --build | --buil | --bui | --bu)
-    ac_prev=build_alias ;;
-  -build=* | --build=* | --buil=* | --bui=* | --bu=*)
-    build_alias=$ac_optarg ;;
-
-  -cache-file | --cache-file | --cache-fil | --cache-fi \
-  | --cache-f | --cache- | --cache | --cach | --cac | --ca | --c)
-    ac_prev=cache_file ;;
-  -cache-file=* | --cache-file=* | --cache-fil=* | --cache-fi=* \
-  | --cache-f=* | --cache-=* | --cache=* | --cach=* | --cac=* | --ca=* | --c=*)
-    cache_file=$ac_optarg ;;
-
-  --config-cache | -C)
-    cache_file=config.cache ;;
-
-  -datadir | --datadir | --datadi | --datad)
-    ac_prev=datadir ;;
-  -datadir=* | --datadir=* | --datadi=* | --datad=*)
-    datadir=$ac_optarg ;;
-
-  -datarootdir | --datarootdir | --datarootdi | --datarootd | --dataroot \
-  | --dataroo | --dataro | --datar)
-    ac_prev=datarootdir ;;
-  -datarootdir=* | --datarootdir=* | --datarootdi=* | --datarootd=* \
-  | --dataroot=* | --dataroo=* | --dataro=* | --datar=*)
-    datarootdir=$ac_optarg ;;
-
-  -disable-* | --disable-*)
-    ac_useropt=`expr "x$ac_option" : 'x-*disable-\(.*\)'`
-    # Reject names that are not valid shell variable names.
-    expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid feature name: $ac_useropt"
-    ac_useropt_orig=$ac_useropt
-    ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
-    case $ac_user_opts in
-      *"
-"enable_$ac_useropt"
-"*) ;;
-      *) ac_unrecognized_opts="$ac_unrecognized_opts$ac_unrecognized_sep--disable-$ac_useropt_orig"
-	 ac_unrecognized_sep=', ';;
-    esac
-    eval enable_$ac_useropt=no ;;
-
-  -docdir | --docdir | --docdi | --doc | --do)
-    ac_prev=docdir ;;
-  -docdir=* | --docdir=* | --docdi=* | --doc=* | --do=*)
-    docdir=$ac_optarg ;;
-
-  -dvidir | --dvidir | --dvidi | --dvid | --dvi | --dv)
-    ac_prev=dvidir ;;
-  -dvidir=* | --dvidir=* | --dvidi=* | --dvid=* | --dvi=* | --dv=*)
-    dvidir=$ac_optarg ;;
-
-  -enable-* | --enable-*)
-    ac_useropt=`expr "x$ac_option" : 'x-*enable-\([^=]*\)'`
-    # Reject names that are not valid shell variable names.
-    expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid feature name: $ac_useropt"
-    ac_useropt_orig=$ac_useropt
-    ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
-    case $ac_user_opts in
-      *"
-"enable_$ac_useropt"
-"*) ;;
-      *) ac_unrecognized_opts="$ac_unrecognized_opts$ac_unrecognized_sep--enable-$ac_useropt_orig"
-	 ac_unrecognized_sep=', ';;
-    esac
-    eval enable_$ac_useropt=\$ac_optarg ;;
-
-  -exec-prefix | --exec_prefix | --exec-prefix | --exec-prefi \
-  | --exec-pref | --exec-pre | --exec-pr | --exec-p | --exec- \
-  | --exec | --exe | --ex)
-    ac_prev=exec_prefix ;;
-  -exec-prefix=* | --exec_prefix=* | --exec-prefix=* | --exec-prefi=* \
-  | --exec-pref=* | --exec-pre=* | --exec-pr=* | --exec-p=* | --exec-=* \
-  | --exec=* | --exe=* | --ex=*)
-    exec_prefix=$ac_optarg ;;
-
-  -gas | --gas | --ga | --g)
-    # Obsolete; use --with-gas.
-    with_gas=yes ;;
-
-  -help | --help | --hel | --he | -h)
-    ac_init_help=long ;;
-  -help=r* | --help=r* | --hel=r* | --he=r* | -hr*)
-    ac_init_help=recursive ;;
-  -help=s* | --help=s* | --hel=s* | --he=s* | -hs*)
-    ac_init_help=short ;;
-
-  -host | --host | --hos | --ho)
-    ac_prev=host_alias ;;
-  -host=* | --host=* | --hos=* | --ho=*)
-    host_alias=$ac_optarg ;;
-
-  -htmldir | --htmldir | --htmldi | --htmld | --html | --htm | --ht)
-    ac_prev=htmldir ;;
-  -htmldir=* | --htmldir=* | --htmldi=* | --htmld=* | --html=* | --htm=* \
-  | --ht=*)
-    htmldir=$ac_optarg ;;
-
-  -includedir | --includedir | --includedi | --included | --include \
-  | --includ | --inclu | --incl | --inc)
-    ac_prev=includedir ;;
-  -includedir=* | --includedir=* | --includedi=* | --included=* | --include=* \
-  | --includ=* | --inclu=* | --incl=* | --inc=*)
-    includedir=$ac_optarg ;;
-
-  -infodir | --infodir | --infodi | --infod | --info | --inf)
-    ac_prev=infodir ;;
-  -infodir=* | --infodir=* | --infodi=* | --infod=* | --info=* | --inf=*)
-    infodir=$ac_optarg ;;
-
-  -libdir | --libdir | --libdi | --libd)
-    ac_prev=libdir ;;
-  -libdir=* | --libdir=* | --libdi=* | --libd=*)
-    libdir=$ac_optarg ;;
-
-  -libexecdir | --libexecdir | --libexecdi | --libexecd | --libexec \
-  | --libexe | --libex | --libe)
-    ac_prev=libexecdir ;;
-  -libexecdir=* | --libexecdir=* | --libexecdi=* | --libexecd=* | --libexec=* \
-  | --libexe=* | --libex=* | --libe=*)
-    libexecdir=$ac_optarg ;;
-
-  -localedir | --localedir | --localedi | --localed | --locale)
-    ac_prev=localedir ;;
-  -localedir=* | --localedir=* | --localedi=* | --localed=* | --locale=*)
-    localedir=$ac_optarg ;;
-
-  -localstatedir | --localstatedir | --localstatedi | --localstated \
-  | --localstate | --localstat | --localsta | --localst | --locals)
-    ac_prev=localstatedir ;;
-  -localstatedir=* | --localstatedir=* | --localstatedi=* | --localstated=* \
-  | --localstate=* | --localstat=* | --localsta=* | --localst=* | --locals=*)
-    localstatedir=$ac_optarg ;;
-
-  -mandir | --mandir | --mandi | --mand | --man | --ma | --m)
-    ac_prev=mandir ;;
-  -mandir=* | --mandir=* | --mandi=* | --mand=* | --man=* | --ma=* | --m=*)
-    mandir=$ac_optarg ;;
-
-  -nfp | --nfp | --nf)
-    # Obsolete; use --without-fp.
-    with_fp=no ;;
-
-  -no-create | --no-create | --no-creat | --no-crea | --no-cre \
-  | --no-cr | --no-c | -n)
-    no_create=yes ;;
-
-  -no-recursion | --no-recursion | --no-recursio | --no-recursi \
-  | --no-recurs | --no-recur | --no-recu | --no-rec | --no-re | --no-r)
-    no_recursion=yes ;;
-
-  -oldincludedir | --oldincludedir | --oldincludedi | --oldincluded \
-  | --oldinclude | --oldinclud | --oldinclu | --oldincl | --oldinc \
-  | --oldin | --oldi | --old | --ol | --o)
-    ac_prev=oldincludedir ;;
-  -oldincludedir=* | --oldincludedir=* | --oldincludedi=* | --oldincluded=* \
-  | --oldinclude=* | --oldinclud=* | --oldinclu=* | --oldincl=* | --oldinc=* \
-  | --oldin=* | --oldi=* | --old=* | --ol=* | --o=*)
-    oldincludedir=$ac_optarg ;;
-
-  -prefix | --prefix | --prefi | --pref | --pre | --pr | --p)
-    ac_prev=prefix ;;
-  -prefix=* | --prefix=* | --prefi=* | --pref=* | --pre=* | --pr=* | --p=*)
-    prefix=$ac_optarg ;;
-
-  -program-prefix | --program-prefix | --program-prefi | --program-pref \
-  | --program-pre | --program-pr | --program-p)
-    ac_prev=program_prefix ;;
-  -program-prefix=* | --program-prefix=* | --program-prefi=* \
-  | --program-pref=* | --program-pre=* | --program-pr=* | --program-p=*)
-    program_prefix=$ac_optarg ;;
-
-  -program-suffix | --program-suffix | --program-suffi | --program-suff \
-  | --program-suf | --program-su | --program-s)
-    ac_prev=program_suffix ;;
-  -program-suffix=* | --program-suffix=* | --program-suffi=* \
-  | --program-suff=* | --program-suf=* | --program-su=* | --program-s=*)
-    program_suffix=$ac_optarg ;;
-
-  -program-transform-name | --program-transform-name \
-  | --program-transform-nam | --program-transform-na \
-  | --program-transform-n | --program-transform- \
-  | --program-transform | --program-transfor \
-  | --program-transfo | --program-transf \
-  | --program-trans | --program-tran \
-  | --progr-tra | --program-tr | --program-t)
-    ac_prev=program_transform_name ;;
-  -program-transform-name=* | --program-transform-name=* \
-  | --program-transform-nam=* | --program-transform-na=* \
-  | --program-transform-n=* | --program-transform-=* \
-  | --program-transform=* | --program-transfor=* \
-  | --program-transfo=* | --program-transf=* \
-  | --program-trans=* | --program-tran=* \
-  | --progr-tra=* | --program-tr=* | --program-t=*)
-    program_transform_name=$ac_optarg ;;
-
-  -pdfdir | --pdfdir | --pdfdi | --pdfd | --pdf | --pd)
-    ac_prev=pdfdir ;;
-  -pdfdir=* | --pdfdir=* | --pdfdi=* | --pdfd=* | --pdf=* | --pd=*)
-    pdfdir=$ac_optarg ;;
-
-  -psdir | --psdir | --psdi | --psd | --ps)
-    ac_prev=psdir ;;
-  -psdir=* | --psdir=* | --psdi=* | --psd=* | --ps=*)
-    psdir=$ac_optarg ;;
-
-  -q | -quiet | --quiet | --quie | --qui | --qu | --q \
-  | -silent | --silent | --silen | --sile | --sil)
-    silent=yes ;;
-
-  -sbindir | --sbindir | --sbindi | --sbind | --sbin | --sbi | --sb)
-    ac_prev=sbindir ;;
-  -sbindir=* | --sbindir=* | --sbindi=* | --sbind=* | --sbin=* \
-  | --sbi=* | --sb=*)
-    sbindir=$ac_optarg ;;
-
-  -sharedstatedir | --sharedstatedir | --sharedstatedi \
-  | --sharedstated | --sharedstate | --sharedstat | --sharedsta \
-  | --sharedst | --shareds | --shared | --share | --shar \
-  | --sha | --sh)
-    ac_prev=sharedstatedir ;;
-  -sharedstatedir=* | --sharedstatedir=* | --sharedstatedi=* \
-  | --sharedstated=* | --sharedstate=* | --sharedstat=* | --sharedsta=* \
-  | --sharedst=* | --shareds=* | --shared=* | --share=* | --shar=* \
-  | --sha=* | --sh=*)
-    sharedstatedir=$ac_optarg ;;
-
-  -site | --site | --sit)
-    ac_prev=site ;;
-  -site=* | --site=* | --sit=*)
-    site=$ac_optarg ;;
-
-  -srcdir | --srcdir | --srcdi | --srcd | --src | --sr)
-    ac_prev=srcdir ;;
-  -srcdir=* | --srcdir=* | --srcdi=* | --srcd=* | --src=* | --sr=*)
-    srcdir=$ac_optarg ;;
-
-  -sysconfdir | --sysconfdir | --sysconfdi | --sysconfd | --sysconf \
-  | --syscon | --sysco | --sysc | --sys | --sy)
-    ac_prev=sysconfdir ;;
-  -sysconfdir=* | --sysconfdir=* | --sysconfdi=* | --sysconfd=* | --sysconf=* \
-  | --syscon=* | --sysco=* | --sysc=* | --sys=* | --sy=*)
-    sysconfdir=$ac_optarg ;;
-
-  -target | --target | --targe | --targ | --tar | --ta | --t)
-    ac_prev=target_alias ;;
-  -target=* | --target=* | --targe=* | --targ=* | --tar=* | --ta=* | --t=*)
-    target_alias=$ac_optarg ;;
-
-  -v | -verbose | --verbose | --verbos | --verbo | --verb)
-    verbose=yes ;;
-
-  -version | --version | --versio | --versi | --vers | -V)
-    ac_init_version=: ;;
-
-  -with-* | --with-*)
-    ac_useropt=`expr "x$ac_option" : 'x-*with-\([^=]*\)'`
-    # Reject names that are not valid shell variable names.
-    expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid package name: $ac_useropt"
-    ac_useropt_orig=$ac_useropt
-    ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
-    case $ac_user_opts in
-      *"
-"with_$ac_useropt"
-"*) ;;
-      *) ac_unrecognized_opts="$ac_unrecognized_opts$ac_unrecognized_sep--with-$ac_useropt_orig"
-	 ac_unrecognized_sep=', ';;
-    esac
-    eval with_$ac_useropt=\$ac_optarg ;;
-
-  -without-* | --without-*)
-    ac_useropt=`expr "x$ac_option" : 'x-*without-\(.*\)'`
-    # Reject names that are not valid shell variable names.
-    expr "x$ac_useropt" : ".*[^-+._$as_cr_alnum]" >/dev/null &&
-      as_fn_error $? "invalid package name: $ac_useropt"
-    ac_useropt_orig=$ac_useropt
-    ac_useropt=`$as_echo "$ac_useropt" | sed 's/[-+.]/_/g'`
-    case $ac_user_opts in
-      *"
-"with_$ac_useropt"
-"*) ;;
-      *) ac_unrecognized_opts="$ac_unrecognized_opts$ac_unrecognized_sep--without-$ac_useropt_orig"
-	 ac_unrecognized_sep=', ';;
-    esac
-    eval with_$ac_useropt=no ;;
-
-  --x)
-    # Obsolete; use --with-x.
-    with_x=yes ;;
-
-  -x-includes | --x-includes | --x-include | --x-includ | --x-inclu \
-  | --x-incl | --x-inc | --x-in | --x-i)
-    ac_prev=x_includes ;;
-  -x-includes=* | --x-includes=* | --x-include=* | --x-includ=* | --x-inclu=* \
-  | --x-incl=* | --x-inc=* | --x-in=* | --x-i=*)
-    x_includes=$ac_optarg ;;
-
-  -x-libraries | --x-libraries | --x-librarie | --x-librari \
-  | --x-librar | --x-libra | --x-libr | --x-lib | --x-li | --x-l)
-    ac_prev=x_libraries ;;
-  -x-libraries=* | --x-libraries=* | --x-librarie=* | --x-librari=* \
-  | --x-librar=* | --x-libra=* | --x-libr=* | --x-lib=* | --x-li=* | --x-l=*)
-    x_libraries=$ac_optarg ;;
-
-  -*) as_fn_error $? "unrecognized option: \`$ac_option'
-Try \`$0 --help' for more information"
-    ;;
-
-  *=*)
-    ac_envvar=`expr "x$ac_option" : 'x\([^=]*\)='`
-    # Reject names that are not valid shell variable names.
-    case $ac_envvar in #(
-      '' | [0-9]* | *[!_$as_cr_alnum]* )
-      as_fn_error $? "invalid variable name: \`$ac_envvar'" ;;
-    esac
-    eval $ac_envvar=\$ac_optarg
-    export $ac_envvar ;;
-
-  *)
-    # FIXME: should be removed in autoconf 3.0.
-    $as_echo "$as_me: WARNING: you should use --build, --host, --target" >&2
-    expr "x$ac_option" : ".*[^-._$as_cr_alnum]" >/dev/null &&
-      $as_echo "$as_me: WARNING: invalid host type: $ac_option" >&2
-    : "${build_alias=$ac_option} ${host_alias=$ac_option} ${target_alias=$ac_option}"
-    ;;
-
-  esac
-done
-
-if test -n "$ac_prev"; then
-  ac_option=--`echo $ac_prev | sed 's/_/-/g'`
-  as_fn_error $? "missing argument to $ac_option"
-fi
-
-if test -n "$ac_unrecognized_opts"; then
-  case $enable_option_checking in
-    no) ;;
-    fatal) as_fn_error $? "unrecognized options: $ac_unrecognized_opts" ;;
-    *)     $as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2 ;;
-  esac
-fi
-
-# Check all directory arguments for consistency.
-for ac_var in	exec_prefix prefix bindir sbindir libexecdir datarootdir \
-		datadir sysconfdir sharedstatedir localstatedir includedir \
-		oldincludedir docdir infodir htmldir dvidir pdfdir psdir \
-		libdir localedir mandir
-do
-  eval ac_val=\$$ac_var
-  # Remove trailing slashes.
-  case $ac_val in
-    */ )
-      ac_val=`expr "X$ac_val" : 'X\(.*[^/]\)' \| "X$ac_val" : 'X\(.*\)'`
-      eval $ac_var=\$ac_val;;
-  esac
-  # Be sure to have absolute directory names.
-  case $ac_val in
-    [\\/$]* | ?:[\\/]* )  continue;;
-    NONE | '' ) case $ac_var in *prefix ) continue;; esac;;
-  esac
-  as_fn_error $? "expected an absolute directory name for --$ac_var: $ac_val"
-done
-
-# There might be people who depend on the old broken behavior: `$host'
-# used to hold the argument of --host etc.
-# FIXME: To remove some day.
-build=$build_alias
-host=$host_alias
-target=$target_alias
-
-# FIXME: To remove some day.
-if test "x$host_alias" != x; then
-  if test "x$build_alias" = x; then
-    cross_compiling=maybe
-  elif test "x$build_alias" != "x$host_alias"; then
-    cross_compiling=yes
-  fi
-fi
-
-ac_tool_prefix=
-test -n "$host_alias" && ac_tool_prefix=$host_alias-
-
-test "$silent" = yes && exec 6>/dev/null
-
-
-ac_pwd=`pwd` && test -n "$ac_pwd" &&
-ac_ls_di=`ls -di .` &&
-ac_pwd_ls_di=`cd "$ac_pwd" && ls -di .` ||
-  as_fn_error $? "working directory cannot be determined"
-test "X$ac_ls_di" = "X$ac_pwd_ls_di" ||
-  as_fn_error $? "pwd does not report name of working directory"
-
-
-# Find the source files, if location was not specified.
-if test -z "$srcdir"; then
-  ac_srcdir_defaulted=yes
-  # Try the directory containing this script, then the parent directory.
-  ac_confdir=`$as_dirname -- "$as_myself" ||
-$as_expr X"$as_myself" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_myself" : 'X\(//\)[^/]' \| \
-	 X"$as_myself" : 'X\(//\)$' \| \
-	 X"$as_myself" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_myself" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-  srcdir=$ac_confdir
-  if test ! -r "$srcdir/$ac_unique_file"; then
-    srcdir=..
-  fi
-else
-  ac_srcdir_defaulted=no
-fi
-if test ! -r "$srcdir/$ac_unique_file"; then
-  test "$ac_srcdir_defaulted" = yes && srcdir="$ac_confdir or .."
-  as_fn_error $? "cannot find sources ($ac_unique_file) in $srcdir"
-fi
-ac_msg="sources are in $srcdir, but \`cd $srcdir' does not work"
-ac_abs_confdir=`(
-	cd "$srcdir" && test -r "./$ac_unique_file" || as_fn_error $? "$ac_msg"
-	pwd)`
-# When building in place, set srcdir=.
-if test "$ac_abs_confdir" = "$ac_pwd"; then
-  srcdir=.
-fi
-# Remove unnecessary trailing slashes from srcdir.
-# Double slashes in file names in object file debugging info
-# mess up M-x gdb in Emacs.
-case $srcdir in
-*/) srcdir=`expr "X$srcdir" : 'X\(.*[^/]\)' \| "X$srcdir" : 'X\(.*\)'`;;
-esac
-for ac_var in $ac_precious_vars; do
-  eval ac_env_${ac_var}_set=\${${ac_var}+set}
-  eval ac_env_${ac_var}_value=\$${ac_var}
-  eval ac_cv_env_${ac_var}_set=\${${ac_var}+set}
-  eval ac_cv_env_${ac_var}_value=\$${ac_var}
-done
-
-#
-# Report the --help message.
-#
-if test "$ac_init_help" = "long"; then
-  # Omit some internal or obsolete options to make the list less imposing.
-  # This message is too long to be a string in the A/UX 3.1 sh.
-  cat <<_ACEOF
-\`configure' configures Grid 1.0 to adapt to many kinds of systems.
-
-Usage: $0 [OPTION]... [VAR=VALUE]...
-
-To assign environment variables (e.g., CC, CFLAGS...), specify them as
-VAR=VALUE.  See below for descriptions of some of the useful variables.
-
-Defaults for the options are specified in brackets.
-
-Configuration:
-  -h, --help              display this help and exit
-      --help=short        display options specific to this package
-      --help=recursive    display the short help of all the included packages
-  -V, --version           display version information and exit
-  -q, --quiet, --silent   do not print \`checking ...' messages
-      --cache-file=FILE   cache test results in FILE [disabled]
-  -C, --config-cache      alias for \`--cache-file=config.cache'
-  -n, --no-create         do not create output files
-      --srcdir=DIR        find the sources in DIR [configure dir or \`..']
-
-Installation directories:
-  --prefix=PREFIX         install architecture-independent files in PREFIX
-                          [$ac_default_prefix]
-  --exec-prefix=EPREFIX   install architecture-dependent files in EPREFIX
-                          [PREFIX]
-
-By default, \`make install' will install all the files in
-\`$ac_default_prefix/bin', \`$ac_default_prefix/lib' etc.  You can specify
-an installation prefix other than \`$ac_default_prefix' using \`--prefix',
-for instance \`--prefix=\$HOME'.
-
-For better control, use the options below.
-
-Fine tuning of the installation directories:
-  --bindir=DIR            user executables [EPREFIX/bin]
-  --sbindir=DIR           system admin executables [EPREFIX/sbin]
-  --libexecdir=DIR        program executables [EPREFIX/libexec]
-  --sysconfdir=DIR        read-only single-machine data [PREFIX/etc]
-  --sharedstatedir=DIR    modifiable architecture-independent data [PREFIX/com]
-  --localstatedir=DIR     modifiable single-machine data [PREFIX/var]
-  --libdir=DIR            object code libraries [EPREFIX/lib]
-  --includedir=DIR        C header files [PREFIX/include]
-  --oldincludedir=DIR     C header files for non-gcc [/usr/include]
-  --datarootdir=DIR       read-only arch.-independent data root [PREFIX/share]
-  --datadir=DIR           read-only architecture-independent data [DATAROOTDIR]
-  --infodir=DIR           info documentation [DATAROOTDIR/info]
-  --localedir=DIR         locale-dependent data [DATAROOTDIR/locale]
-  --mandir=DIR            man documentation [DATAROOTDIR/man]
-  --docdir=DIR            documentation root [DATAROOTDIR/doc/grid]
-  --htmldir=DIR           html documentation [DOCDIR]
-  --dvidir=DIR            dvi documentation [DOCDIR]
-  --pdfdir=DIR            pdf documentation [DOCDIR]
-  --psdir=DIR             ps documentation [DOCDIR]
-_ACEOF
-
-  cat <<\_ACEOF
-
-Program names:
-  --program-prefix=PREFIX            prepend PREFIX to installed program names
-  --program-suffix=SUFFIX            append SUFFIX to installed program names
-  --program-transform-name=PROGRAM   run sed PROGRAM on installed program names
-
-System types:
-  --build=BUILD     configure for building on BUILD [guessed]
-  --host=HOST       cross-compile to build programs to run on HOST [BUILD]
-  --target=TARGET   configure for building compilers for TARGET [HOST]
-_ACEOF
-fi
-
-if test -n "$ac_init_help"; then
-  case $ac_init_help in
-     short | recursive ) echo "Configuration of Grid 1.0:";;
-   esac
-  cat <<\_ACEOF
-
-Optional Features:
-  --disable-option-checking  ignore unrecognized --enable/--with options
-  --disable-FEATURE       do not include FEATURE (same as --enable-FEATURE=no)
-  --enable-FEATURE[=ARG]  include FEATURE [ARG=yes]
-  --enable-silent-rules   less verbose build output (undo: "make V=1")
-  --disable-silent-rules  verbose build output (undo: "make V=0")
-  --enable-dependency-tracking
-                          do not reject slow dependency extractors
-  --disable-dependency-tracking
-                          speeds up one-time build
-  --disable-openmp        do not use OpenMP
-  --enable-simd=SSE4|AVX|AVX2|AVX512|MIC
-                          Select instructions to be SSE4.0, AVX 1.0, AVX
-                          2.0+FMA, AVX 512, MIC
-  --enable-precision=single|double
-                          Select default word size of Real
-  --enable-comms=none|mpi Select communications
-  --enable-doxygen        enable documentation generation with doxygen (auto)
-  --enable-dot            use 'dot' to generate graphs in doxygen (auto)
-  --enable-html-docs      enable HTML generation with doxygen (yes)
-  --enable-latex-docs     enable LaTeX documentation generation with doxygen
-                          (no)
-
-Some influential environment variables:
-  CXX         C++ compiler command
-  CXXFLAGS    C++ compiler flags
-  LDFLAGS     linker flags, e.g. -L<lib dir> if you have libraries in a
-              nonstandard directory <lib dir>
-  LIBS        libraries to pass to the linker, e.g. -l<library>
-  CPPFLAGS    (Objective) C/C++ preprocessor flags, e.g. -I<include dir> if
-              you have headers in a nonstandard directory <include dir>
-  CC          C compiler command
-  CFLAGS      C compiler flags
-  CXXCPP      C++ preprocessor
-
-Use these variables to override the choices made by `configure' or to help
-it to find libraries and programs with nonstandard names/locations.
-
-Report bugs to <paboyle@ph.ed.ac.uk>.
-_ACEOF
-ac_status=$?
-fi
-
-if test "$ac_init_help" = "recursive"; then
-  # If there are subdirs, report their specific --help.
-  for ac_dir in : $ac_subdirs_all; do test "x$ac_dir" = x: && continue
-    test -d "$ac_dir" ||
-      { cd "$srcdir" && ac_pwd=`pwd` && srcdir=. && test -d "$ac_dir"; } ||
-      continue
-    ac_builddir=.
-
-case "$ac_dir" in
-.) ac_dir_suffix= ac_top_builddir_sub=. ac_top_build_prefix= ;;
-*)
-  ac_dir_suffix=/`$as_echo "$ac_dir" | sed 's|^\.[\\/]||'`
-  # A ".." for each directory in $ac_dir_suffix.
-  ac_top_builddir_sub=`$as_echo "$ac_dir_suffix" | sed 's|/[^\\/]*|/..|g;s|/||'`
-  case $ac_top_builddir_sub in
-  "") ac_top_builddir_sub=. ac_top_build_prefix= ;;
-  *)  ac_top_build_prefix=$ac_top_builddir_sub/ ;;
-  esac ;;
-esac
-ac_abs_top_builddir=$ac_pwd
-ac_abs_builddir=$ac_pwd$ac_dir_suffix
-# for backward compatibility:
-ac_top_builddir=$ac_top_build_prefix
-
-case $srcdir in
-  .)  # We are building in place.
-    ac_srcdir=.
-    ac_top_srcdir=$ac_top_builddir_sub
-    ac_abs_top_srcdir=$ac_pwd ;;
-  [\\/]* | ?:[\\/]* )  # Absolute name.
-    ac_srcdir=$srcdir$ac_dir_suffix;
-    ac_top_srcdir=$srcdir
-    ac_abs_top_srcdir=$srcdir ;;
-  *) # Relative name.
-    ac_srcdir=$ac_top_build_prefix$srcdir$ac_dir_suffix
-    ac_top_srcdir=$ac_top_build_prefix$srcdir
-    ac_abs_top_srcdir=$ac_pwd/$srcdir ;;
-esac
-ac_abs_srcdir=$ac_abs_top_srcdir$ac_dir_suffix
-
-    cd "$ac_dir" || { ac_status=$?; continue; }
-    # Check for guested configure.
-    if test -f "$ac_srcdir/configure.gnu"; then
-      echo &&
-      $SHELL "$ac_srcdir/configure.gnu" --help=recursive
-    elif test -f "$ac_srcdir/configure"; then
-      echo &&
-      $SHELL "$ac_srcdir/configure" --help=recursive
-    else
-      $as_echo "$as_me: WARNING: no configuration information is in $ac_dir" >&2
-    fi || ac_status=$?
-    cd "$ac_pwd" || { ac_status=$?; break; }
-  done
-fi
-
-test -n "$ac_init_help" && exit $ac_status
-if $ac_init_version; then
-  cat <<\_ACEOF
-Grid configure 1.0
-generated by GNU Autoconf 2.69
-
-Copyright (C) 2012 Free Software Foundation, Inc.
-This configure script is free software; the Free Software Foundation
-gives unlimited permission to copy, distribute and modify it.
-_ACEOF
-  exit
-fi
-
-## ------------------------ ##
-## Autoconf initialization. ##
-## ------------------------ ##
-
-# ac_fn_cxx_try_compile LINENO
-# ----------------------------
-# Try to compile conftest.$ac_ext, and return whether this succeeded.
-ac_fn_cxx_try_compile ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  rm -f conftest.$ac_objext
-  if { { ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compile") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_retval=1
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_cxx_try_compile
-
-# ac_fn_cxx_try_link LINENO
-# -------------------------
-# Try to link conftest.$ac_ext, and return whether this succeeded.
-ac_fn_cxx_try_link ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  rm -f conftest.$ac_objext conftest$ac_exeext
-  if { { ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_link") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && {
-	 test -z "$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest$ac_exeext && {
-	 test "$cross_compiling" = yes ||
-	 test -x conftest$ac_exeext
-       }; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_retval=1
-fi
-  # Delete the IPA/IPO (Inter Procedural Analysis/Optimization) information
-  # created by the PGI compiler (conftest_ipa8_conftest.oo), as it would
-  # interfere with the next link command; also delete a directory that is
-  # left behind by Apple's compiler.  We do this before executing the actions.
-  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_cxx_try_link
-
-# ac_fn_c_try_compile LINENO
-# --------------------------
-# Try to compile conftest.$ac_ext, and return whether this succeeded.
-ac_fn_c_try_compile ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  rm -f conftest.$ac_objext
-  if { { ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compile") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && {
-	 test -z "$ac_c_werror_flag" ||
-	 test ! -s conftest.err
-       } && test -s conftest.$ac_objext; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-	ac_retval=1
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_c_try_compile
-
-# ac_fn_c_try_run LINENO
-# ----------------------
-# Try to link conftest.$ac_ext, and return whether this succeeded. Assumes
-# that executables *can* be run.
-ac_fn_c_try_run ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  if { { ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_link") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && { ac_try='./conftest$ac_exeext'
-  { { case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; }; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: program exited with status $ac_status" >&5
-       $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-       ac_retval=$ac_status
-fi
-  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_c_try_run
-
-# ac_fn_cxx_try_cpp LINENO
-# ------------------------
-# Try to preprocess conftest.$ac_ext, and return whether this succeeded.
-ac_fn_cxx_try_cpp ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  if { { ac_try="$ac_cpp conftest.$ac_ext"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_cpp conftest.$ac_ext") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    grep -v '^ *+' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-    mv -f conftest.er1 conftest.err
-  fi
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } > conftest.i && {
-	 test -z "$ac_cxx_preproc_warn_flag$ac_cxx_werror_flag" ||
-	 test ! -s conftest.err
-       }; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-    ac_retval=1
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_cxx_try_cpp
-
-# ac_fn_cxx_check_header_mongrel LINENO HEADER VAR INCLUDES
-# ---------------------------------------------------------
-# Tests whether HEADER exists, giving a warning if it cannot be compiled using
-# the include files in INCLUDES and setting the cache variable VAR
-# accordingly.
-ac_fn_cxx_check_header_mongrel ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  if eval \${$3+:} false; then :
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-else
-  # Is the header compilable?
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 usability" >&5
-$as_echo_n "checking $2 usability... " >&6; }
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-#include <$2>
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  ac_header_compiler=yes
-else
-  ac_header_compiler=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_compiler" >&5
-$as_echo "$ac_header_compiler" >&6; }
-
-# Is the header present?
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking $2 presence" >&5
-$as_echo_n "checking $2 presence... " >&6; }
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <$2>
-_ACEOF
-if ac_fn_cxx_try_cpp "$LINENO"; then :
-  ac_header_preproc=yes
-else
-  ac_header_preproc=no
-fi
-rm -f conftest.err conftest.i conftest.$ac_ext
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_header_preproc" >&5
-$as_echo "$ac_header_preproc" >&6; }
-
-# So?  What about this header?
-case $ac_header_compiler:$ac_header_preproc:$ac_cxx_preproc_warn_flag in #((
-  yes:no: )
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&5
-$as_echo "$as_me: WARNING: $2: accepted by the compiler, rejected by the preprocessor!" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
-    ;;
-  no:yes:* )
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: present but cannot be compiled" >&5
-$as_echo "$as_me: WARNING: $2: present but cannot be compiled" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     check for missing prerequisite headers?" >&5
-$as_echo "$as_me: WARNING: $2:     check for missing prerequisite headers?" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: see the Autoconf documentation" >&5
-$as_echo "$as_me: WARNING: $2: see the Autoconf documentation" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&5
-$as_echo "$as_me: WARNING: $2:     section \"Present But Cannot Be Compiled\"" >&2;}
-    { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $2: proceeding with the compiler's result" >&5
-$as_echo "$as_me: WARNING: $2: proceeding with the compiler's result" >&2;}
-( $as_echo "## ---------------------------------- ##
-## Report this to paboyle@ph.ed.ac.uk ##
-## ---------------------------------- ##"
-     ) | sed "s/^/$as_me: WARNING:     /" >&2
-    ;;
-esac
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  eval "$3=\$ac_header_compiler"
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-fi
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_cxx_check_header_mongrel
-
-# ac_fn_cxx_try_run LINENO
-# ------------------------
-# Try to link conftest.$ac_ext, and return whether this succeeded. Assumes
-# that executables *can* be run.
-ac_fn_cxx_try_run ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  if { { ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_link") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; } && { ac_try='./conftest$ac_exeext'
-  { { case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; }; then :
-  ac_retval=0
-else
-  $as_echo "$as_me: program exited with status $ac_status" >&5
-       $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-       ac_retval=$ac_status
-fi
-  rm -rf conftest.dSYM conftest_ipa8_conftest.oo
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-  as_fn_set_status $ac_retval
-
-} # ac_fn_cxx_try_run
-
-# ac_fn_cxx_check_header_compile LINENO HEADER VAR INCLUDES
-# ---------------------------------------------------------
-# Tests whether HEADER exists and can be compiled using the include files in
-# INCLUDES, setting the cache variable VAR accordingly.
-ac_fn_cxx_check_header_compile ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-#include <$2>
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  eval "$3=yes"
-else
-  eval "$3=no"
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_cxx_check_header_compile
-
-# ac_fn_cxx_check_decl LINENO SYMBOL VAR INCLUDES
-# -----------------------------------------------
-# Tests whether SYMBOL is declared in INCLUDES, setting cache variable VAR
-# accordingly.
-ac_fn_cxx_check_decl ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  as_decl_name=`echo $2|sed 's/ *(.*//'`
-  as_decl_use=`echo $2|sed -e 's/(/((/' -e 's/)/) 0&/' -e 's/,/) 0& (/g'`
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $as_decl_name is declared" >&5
-$as_echo_n "checking whether $as_decl_name is declared... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-int
-main ()
-{
-#ifndef $as_decl_name
-#ifdef __cplusplus
-  (void) $as_decl_use;
-#else
-  (void) $as_decl_name;
-#endif
-#endif
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  eval "$3=yes"
-else
-  eval "$3=no"
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_cxx_check_decl
-
-# ac_fn_cxx_check_type LINENO TYPE VAR INCLUDES
-# ---------------------------------------------
-# Tests whether TYPE exists after having included INCLUDES, setting cache
-# variable VAR accordingly.
-ac_fn_cxx_check_type ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  eval "$3=no"
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-int
-main ()
-{
-if (sizeof ($2))
-	 return 0;
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$4
-int
-main ()
-{
-if (sizeof (($2)))
-	    return 0;
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-
-else
-  eval "$3=yes"
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_cxx_check_type
-
-# ac_fn_c_find_uintX_t LINENO BITS VAR
-# ------------------------------------
-# Finds an unsigned integer type with width BITS, setting cache variable VAR
-# accordingly.
-ac_fn_c_find_uintX_t ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for uint$2_t" >&5
-$as_echo_n "checking for uint$2_t... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  eval "$3=no"
-     # Order is important - never check a type that is potentially smaller
-     # than half of the expected target width.
-     for ac_type in uint$2_t 'unsigned int' 'unsigned long int' \
-	 'unsigned long long int' 'unsigned short int' 'unsigned char'; do
-       cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-$ac_includes_default
-int
-main ()
-{
-static int test_array [1 - 2 * !((($ac_type) -1 >> ($2 / 2 - 1)) >> ($2 / 2 - 1) == 3)];
-test_array [0] = 0;
-return test_array [0];
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  case $ac_type in #(
-  uint$2_t) :
-    eval "$3=yes" ;; #(
-  *) :
-    eval "$3=\$ac_type" ;;
-esac
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-       if eval test \"x\$"$3"\" = x"no"; then :
-
-else
-  break
-fi
-     done
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_c_find_uintX_t
-
-# ac_fn_cxx_check_func LINENO FUNC VAR
-# ------------------------------------
-# Tests whether FUNC exists, setting the cache variable VAR accordingly
-ac_fn_cxx_check_func ()
-{
-  as_lineno=${as_lineno-"$1"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $2" >&5
-$as_echo_n "checking for $2... " >&6; }
-if eval \${$3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-/* Define $2 to an innocuous variant, in case <limits.h> declares $2.
-   For example, HP-UX 11i <limits.h> declares gettimeofday.  */
-#define $2 innocuous_$2
-
-/* System header to define __stub macros and hopefully few prototypes,
-    which can conflict with char $2 (); below.
-    Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
-    <limits.h> exists even on freestanding compilers.  */
-
-#ifdef __STDC__
-# include <limits.h>
-#else
-# include <assert.h>
-#endif
-
-#undef $2
-
-/* Override any GCC internal prototype to avoid an error.
-   Use char because int might match the return type of a GCC
-   builtin and then its argument prototype would still apply.  */
-#ifdef __cplusplus
-extern "C"
-#endif
-char $2 ();
-/* The GNU C library defines this for functions which it implements
-    to always fail with ENOSYS.  Some functions are actually named
-    something starting with __ and the normal name is an alias.  */
-#if defined __stub_$2 || defined __stub___$2
-choke me
-#endif
-
-int
-main ()
-{
-return $2 ();
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_link "$LINENO"; then :
-  eval "$3=yes"
-else
-  eval "$3=no"
-fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
-fi
-eval ac_res=\$$3
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-  eval $as_lineno_stack; ${as_lineno_stack:+:} unset as_lineno
-
-} # ac_fn_cxx_check_func
-cat >config.log <<_ACEOF
-This file contains any messages produced by compilers while
-running configure, to aid debugging if configure makes a mistake.
-
-It was created by Grid $as_me 1.0, which was
-generated by GNU Autoconf 2.69.  Invocation command line was
-
-  $ $0 $@
-
-_ACEOF
-exec 5>>config.log
-{
-cat <<_ASUNAME
-## --------- ##
-## Platform. ##
-## --------- ##
-
-hostname = `(hostname || uname -n) 2>/dev/null | sed 1q`
-uname -m = `(uname -m) 2>/dev/null || echo unknown`
-uname -r = `(uname -r) 2>/dev/null || echo unknown`
-uname -s = `(uname -s) 2>/dev/null || echo unknown`
-uname -v = `(uname -v) 2>/dev/null || echo unknown`
-
-/usr/bin/uname -p = `(/usr/bin/uname -p) 2>/dev/null || echo unknown`
-/bin/uname -X     = `(/bin/uname -X) 2>/dev/null     || echo unknown`
-
-/bin/arch              = `(/bin/arch) 2>/dev/null              || echo unknown`
-/usr/bin/arch -k       = `(/usr/bin/arch -k) 2>/dev/null       || echo unknown`
-/usr/convex/getsysinfo = `(/usr/convex/getsysinfo) 2>/dev/null || echo unknown`
-/usr/bin/hostinfo      = `(/usr/bin/hostinfo) 2>/dev/null      || echo unknown`
-/bin/machine           = `(/bin/machine) 2>/dev/null           || echo unknown`
-/usr/bin/oslevel       = `(/usr/bin/oslevel) 2>/dev/null       || echo unknown`
-/bin/universe          = `(/bin/universe) 2>/dev/null          || echo unknown`
-
-_ASUNAME
-
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    $as_echo "PATH: $as_dir"
-  done
-IFS=$as_save_IFS
-
-} >&5
-
-cat >&5 <<_ACEOF
-
-
-## ----------- ##
-## Core tests. ##
-## ----------- ##
-
-_ACEOF
-
-
-# Keep a trace of the command line.
-# Strip out --no-create and --no-recursion so they do not pile up.
-# Strip out --silent because we don't want to record it for future runs.
-# Also quote any args containing shell meta-characters.
-# Make two passes to allow for proper duplicate-argument suppression.
-ac_configure_args=
-ac_configure_args0=
-ac_configure_args1=
-ac_must_keep_next=false
-for ac_pass in 1 2
-do
-  for ac_arg
-  do
-    case $ac_arg in
-    -no-create | --no-c* | -n | -no-recursion | --no-r*) continue ;;
-    -q | -quiet | --quiet | --quie | --qui | --qu | --q \
-    | -silent | --silent | --silen | --sile | --sil)
-      continue ;;
-    *\'*)
-      ac_arg=`$as_echo "$ac_arg" | sed "s/'/'\\\\\\\\''/g"` ;;
-    esac
-    case $ac_pass in
-    1) as_fn_append ac_configure_args0 " '$ac_arg'" ;;
-    2)
-      as_fn_append ac_configure_args1 " '$ac_arg'"
-      if test $ac_must_keep_next = true; then
-	ac_must_keep_next=false # Got value, back to normal.
-      else
-	case $ac_arg in
-	  *=* | --config-cache | -C | -disable-* | --disable-* \
-	  | -enable-* | --enable-* | -gas | --g* | -nfp | --nf* \
-	  | -q | -quiet | --q* | -silent | --sil* | -v | -verb* \
-	  | -with-* | --with-* | -without-* | --without-* | --x)
-	    case "$ac_configure_args0 " in
-	      "$ac_configure_args1"*" '$ac_arg' "* ) continue ;;
-	    esac
-	    ;;
-	  -* ) ac_must_keep_next=true ;;
-	esac
-      fi
-      as_fn_append ac_configure_args " '$ac_arg'"
-      ;;
-    esac
-  done
-done
-{ ac_configure_args0=; unset ac_configure_args0;}
-{ ac_configure_args1=; unset ac_configure_args1;}
-
-# When interrupted or exit'd, cleanup temporary files, and complete
-# config.log.  We remove comments because anyway the quotes in there
-# would cause problems or look ugly.
-# WARNING: Use '\'' to represent an apostrophe within the trap.
-# WARNING: Do not start the trap code with a newline, due to a FreeBSD 4.0 bug.
-trap 'exit_status=$?
-  # Save into config.log some information that might help in debugging.
-  {
-    echo
-
-    $as_echo "## ---------------- ##
-## Cache variables. ##
-## ---------------- ##"
-    echo
-    # The following way of writing the cache mishandles newlines in values,
-(
-  for ac_var in `(set) 2>&1 | sed -n '\''s/^\([a-zA-Z_][a-zA-Z0-9_]*\)=.*/\1/p'\''`; do
-    eval ac_val=\$$ac_var
-    case $ac_val in #(
-    *${as_nl}*)
-      case $ac_var in #(
-      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
-$as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
-      esac
-      case $ac_var in #(
-      _ | IFS | as_nl) ;; #(
-      BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
-      *) { eval $ac_var=; unset $ac_var;} ;;
-      esac ;;
-    esac
-  done
-  (set) 2>&1 |
-    case $as_nl`(ac_space='\'' '\''; set) 2>&1` in #(
-    *${as_nl}ac_space=\ *)
-      sed -n \
-	"s/'\''/'\''\\\\'\'''\''/g;
-	  s/^\\([_$as_cr_alnum]*_cv_[_$as_cr_alnum]*\\)=\\(.*\\)/\\1='\''\\2'\''/p"
-      ;; #(
-    *)
-      sed -n "/^[_$as_cr_alnum]*_cv_[_$as_cr_alnum]*=/p"
-      ;;
-    esac |
-    sort
-)
-    echo
-
-    $as_echo "## ----------------- ##
-## Output variables. ##
-## ----------------- ##"
-    echo
-    for ac_var in $ac_subst_vars
-    do
-      eval ac_val=\$$ac_var
-      case $ac_val in
-      *\'\''*) ac_val=`$as_echo "$ac_val" | sed "s/'\''/'\''\\\\\\\\'\'''\''/g"`;;
-      esac
-      $as_echo "$ac_var='\''$ac_val'\''"
-    done | sort
-    echo
-
-    if test -n "$ac_subst_files"; then
-      $as_echo "## ------------------- ##
-## File substitutions. ##
-## ------------------- ##"
-      echo
-      for ac_var in $ac_subst_files
-      do
-	eval ac_val=\$$ac_var
-	case $ac_val in
-	*\'\''*) ac_val=`$as_echo "$ac_val" | sed "s/'\''/'\''\\\\\\\\'\'''\''/g"`;;
-	esac
-	$as_echo "$ac_var='\''$ac_val'\''"
-      done | sort
-      echo
-    fi
-
-    if test -s confdefs.h; then
-      $as_echo "## ----------- ##
-## confdefs.h. ##
-## ----------- ##"
-      echo
-      cat confdefs.h
-      echo
-    fi
-    test "$ac_signal" != 0 &&
-      $as_echo "$as_me: caught signal $ac_signal"
-    $as_echo "$as_me: exit $exit_status"
-  } >&5
-  rm -f core *.core core.conftest.* &&
-    rm -f -r conftest* confdefs* conf$$* $ac_clean_files &&
-    exit $exit_status
-' 0
-for ac_signal in 1 2 13 15; do
-  trap 'ac_signal='$ac_signal'; as_fn_exit 1' $ac_signal
-done
-ac_signal=0
-
-# confdefs.h avoids OS command line length limits that DEFS can exceed.
-rm -f -r conftest* confdefs.h
-
-$as_echo "/* confdefs.h */" > confdefs.h
-
-# Predefined preprocessor variables.
-
-cat >>confdefs.h <<_ACEOF
-#define PACKAGE_NAME "$PACKAGE_NAME"
-_ACEOF
-
-cat >>confdefs.h <<_ACEOF
-#define PACKAGE_TARNAME "$PACKAGE_TARNAME"
-_ACEOF
-
-cat >>confdefs.h <<_ACEOF
-#define PACKAGE_VERSION "$PACKAGE_VERSION"
-_ACEOF
-
-cat >>confdefs.h <<_ACEOF
-#define PACKAGE_STRING "$PACKAGE_STRING"
-_ACEOF
-
-cat >>confdefs.h <<_ACEOF
-#define PACKAGE_BUGREPORT "$PACKAGE_BUGREPORT"
-_ACEOF
-
-cat >>confdefs.h <<_ACEOF
-#define PACKAGE_URL "$PACKAGE_URL"
-_ACEOF
-
-
-# Let the site file select an alternate cache file if it wants to.
-# Prefer an explicitly selected file to automatically selected ones.
-ac_site_file1=NONE
-ac_site_file2=NONE
-if test -n "$CONFIG_SITE"; then
-  # We do not want a PATH search for config.site.
-  case $CONFIG_SITE in #((
-    -*)  ac_site_file1=./$CONFIG_SITE;;
-    */*) ac_site_file1=$CONFIG_SITE;;
-    *)   ac_site_file1=./$CONFIG_SITE;;
-  esac
-elif test "x$prefix" != xNONE; then
-  ac_site_file1=$prefix/share/config.site
-  ac_site_file2=$prefix/etc/config.site
-else
-  ac_site_file1=$ac_default_prefix/share/config.site
-  ac_site_file2=$ac_default_prefix/etc/config.site
-fi
-for ac_site_file in "$ac_site_file1" "$ac_site_file2"
-do
-  test "x$ac_site_file" = xNONE && continue
-  if test /dev/null != "$ac_site_file" && test -r "$ac_site_file"; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: loading site script $ac_site_file" >&5
-$as_echo "$as_me: loading site script $ac_site_file" >&6;}
-    sed 's/^/| /' "$ac_site_file" >&5
-    . "$ac_site_file" \
-      || { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "failed to load site script $ac_site_file
-See \`config.log' for more details" "$LINENO" 5; }
-  fi
-done
-
-if test -r "$cache_file"; then
-  # Some versions of bash will fail to source /dev/null (special files
-  # actually), so we avoid doing that.  DJGPP emulates it as a regular file.
-  if test /dev/null != "$cache_file" && test -f "$cache_file"; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: loading cache $cache_file" >&5
-$as_echo "$as_me: loading cache $cache_file" >&6;}
-    case $cache_file in
-      [\\/]* | ?:[\\/]* ) . "$cache_file";;
-      *)                      . "./$cache_file";;
-    esac
-  fi
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: creating cache $cache_file" >&5
-$as_echo "$as_me: creating cache $cache_file" >&6;}
-  >$cache_file
-fi
-
-# Check that the precious variables saved in the cache have kept the same
-# value.
-ac_cache_corrupted=false
-for ac_var in $ac_precious_vars; do
-  eval ac_old_set=\$ac_cv_env_${ac_var}_set
-  eval ac_new_set=\$ac_env_${ac_var}_set
-  eval ac_old_val=\$ac_cv_env_${ac_var}_value
-  eval ac_new_val=\$ac_env_${ac_var}_value
-  case $ac_old_set,$ac_new_set in
-    set,)
-      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&5
-$as_echo "$as_me: error: \`$ac_var' was set to \`$ac_old_val' in the previous run" >&2;}
-      ac_cache_corrupted=: ;;
-    ,set)
-      { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' was not set in the previous run" >&5
-$as_echo "$as_me: error: \`$ac_var' was not set in the previous run" >&2;}
-      ac_cache_corrupted=: ;;
-    ,);;
-    *)
-      if test "x$ac_old_val" != "x$ac_new_val"; then
-	# differences in whitespace do not lead to failure.
-	ac_old_val_w=`echo x $ac_old_val`
-	ac_new_val_w=`echo x $ac_new_val`
-	if test "$ac_old_val_w" != "$ac_new_val_w"; then
-	  { $as_echo "$as_me:${as_lineno-$LINENO}: error: \`$ac_var' has changed since the previous run:" >&5
-$as_echo "$as_me: error: \`$ac_var' has changed since the previous run:" >&2;}
-	  ac_cache_corrupted=:
-	else
-	  { $as_echo "$as_me:${as_lineno-$LINENO}: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&5
-$as_echo "$as_me: warning: ignoring whitespace changes in \`$ac_var' since the previous run:" >&2;}
-	  eval $ac_var=\$ac_old_val
-	fi
-	{ $as_echo "$as_me:${as_lineno-$LINENO}:   former value:  \`$ac_old_val'" >&5
-$as_echo "$as_me:   former value:  \`$ac_old_val'" >&2;}
-	{ $as_echo "$as_me:${as_lineno-$LINENO}:   current value: \`$ac_new_val'" >&5
-$as_echo "$as_me:   current value: \`$ac_new_val'" >&2;}
-      fi;;
-  esac
-  # Pass precious variables to config.status.
-  if test "$ac_new_set" = set; then
-    case $ac_new_val in
-    *\'*) ac_arg=$ac_var=`$as_echo "$ac_new_val" | sed "s/'/'\\\\\\\\''/g"` ;;
-    *) ac_arg=$ac_var=$ac_new_val ;;
-    esac
-    case " $ac_configure_args " in
-      *" '$ac_arg' "*) ;; # Avoid dups.  Use of quotes ensures accuracy.
-      *) as_fn_append ac_configure_args " '$ac_arg'" ;;
-    esac
-  fi
-done
-if $ac_cache_corrupted; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-  { $as_echo "$as_me:${as_lineno-$LINENO}: error: changes in the environment can compromise the build" >&5
-$as_echo "$as_me: error: changes in the environment can compromise the build" >&2;}
-  as_fn_error $? "run \`make distclean' and/or \`rm $cache_file' and start over" "$LINENO" 5
-fi
-## -------------------- ##
-## Main body of script. ##
-## -------------------- ##
-
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-
-
-ac_aux_dir=
-for ac_dir in "$srcdir" "$srcdir/.." "$srcdir/../.."; do
-  if test -f "$ac_dir/install-sh"; then
-    ac_aux_dir=$ac_dir
-    ac_install_sh="$ac_aux_dir/install-sh -c"
-    break
-  elif test -f "$ac_dir/install.sh"; then
-    ac_aux_dir=$ac_dir
-    ac_install_sh="$ac_aux_dir/install.sh -c"
-    break
-  elif test -f "$ac_dir/shtool"; then
-    ac_aux_dir=$ac_dir
-    ac_install_sh="$ac_aux_dir/shtool install -c"
-    break
-  fi
-done
-if test -z "$ac_aux_dir"; then
-  as_fn_error $? "cannot find install-sh, install.sh, or shtool in \"$srcdir\" \"$srcdir/..\" \"$srcdir/../..\"" "$LINENO" 5
-fi
-
-# These three variables are undocumented and unsupported,
-# and are intended to be withdrawn in a future Autoconf release.
-# They can cause serious problems if a builder's source tree is in a directory
-# whose full name contains unusual characters.
-ac_config_guess="$SHELL $ac_aux_dir/config.guess"  # Please don't use this var.
-ac_config_sub="$SHELL $ac_aux_dir/config.sub"  # Please don't use this var.
-ac_configure="$SHELL $ac_aux_dir/configure"  # Please don't use this var.
-
-
-# Make sure we can run config.sub.
-$SHELL "$ac_aux_dir/config.sub" sun4 >/dev/null 2>&1 ||
-  as_fn_error $? "cannot run $SHELL $ac_aux_dir/config.sub" "$LINENO" 5
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking build system type" >&5
-$as_echo_n "checking build system type... " >&6; }
-if ${ac_cv_build+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  ac_build_alias=$build_alias
-test "x$ac_build_alias" = x &&
-  ac_build_alias=`$SHELL "$ac_aux_dir/config.guess"`
-test "x$ac_build_alias" = x &&
-  as_fn_error $? "cannot guess build type; you must specify one" "$LINENO" 5
-ac_cv_build=`$SHELL "$ac_aux_dir/config.sub" $ac_build_alias` ||
-  as_fn_error $? "$SHELL $ac_aux_dir/config.sub $ac_build_alias failed" "$LINENO" 5
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_build" >&5
-$as_echo "$ac_cv_build" >&6; }
-case $ac_cv_build in
-*-*-*) ;;
-*) as_fn_error $? "invalid value of canonical build" "$LINENO" 5;;
-esac
-build=$ac_cv_build
-ac_save_IFS=$IFS; IFS='-'
-set x $ac_cv_build
-shift
-build_cpu=$1
-build_vendor=$2
-shift; shift
-# Remember, the first character of IFS is used to create $*,
-# except with old shells:
-build_os=$*
-IFS=$ac_save_IFS
-case $build_os in *\ *) build_os=`echo "$build_os" | sed 's/ /-/g'`;; esac
-
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking host system type" >&5
-$as_echo_n "checking host system type... " >&6; }
-if ${ac_cv_host+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test "x$host_alias" = x; then
-  ac_cv_host=$ac_cv_build
-else
-  ac_cv_host=`$SHELL "$ac_aux_dir/config.sub" $host_alias` ||
-    as_fn_error $? "$SHELL $ac_aux_dir/config.sub $host_alias failed" "$LINENO" 5
-fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_host" >&5
-$as_echo "$ac_cv_host" >&6; }
-case $ac_cv_host in
-*-*-*) ;;
-*) as_fn_error $? "invalid value of canonical host" "$LINENO" 5;;
-esac
-host=$ac_cv_host
-ac_save_IFS=$IFS; IFS='-'
-set x $ac_cv_host
-shift
-host_cpu=$1
-host_vendor=$2
-shift; shift
-# Remember, the first character of IFS is used to create $*,
-# except with old shells:
-host_os=$*
-IFS=$ac_save_IFS
-case $host_os in *\ *) host_os=`echo "$host_os" | sed 's/ /-/g'`;; esac
-
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking target system type" >&5
-$as_echo_n "checking target system type... " >&6; }
-if ${ac_cv_target+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test "x$target_alias" = x; then
-  ac_cv_target=$ac_cv_host
-else
-  ac_cv_target=`$SHELL "$ac_aux_dir/config.sub" $target_alias` ||
-    as_fn_error $? "$SHELL $ac_aux_dir/config.sub $target_alias failed" "$LINENO" 5
-fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_target" >&5
-$as_echo "$ac_cv_target" >&6; }
-case $ac_cv_target in
-*-*-*) ;;
-*) as_fn_error $? "invalid value of canonical target" "$LINENO" 5;;
-esac
-target=$ac_cv_target
-ac_save_IFS=$IFS; IFS='-'
-set x $ac_cv_target
-shift
-target_cpu=$1
-target_vendor=$2
-shift; shift
-# Remember, the first character of IFS is used to create $*,
-# except with old shells:
-target_os=$*
-IFS=$ac_save_IFS
-case $target_os in *\ *) target_os=`echo "$target_os" | sed 's/ /-/g'`;; esac
-
-
-# The aliases save the names the user supplied, while $host etc.
-# will get canonicalized.
-test -n "$target_alias" &&
-  test "$program_prefix$program_suffix$program_transform_name" = \
-    NONENONEs,x,x, &&
-  program_prefix=${target_alias}-
-
-am__api_version='1.14'
-
-# Find a good install program.  We prefer a C program (faster),
-# so one script is as good as another.  But avoid the broken or
-# incompatible versions:
-# SysV /etc/install, /usr/sbin/install
-# SunOS /usr/etc/install
-# IRIX /sbin/install
-# AIX /bin/install
-# AmigaOS /C/install, which installs bootblocks on floppy discs
-# AIX 4 /usr/bin/installbsd, which doesn't work without a -g flag
-# AFS /usr/afsws/bin/install, which mishandles nonexistent args
-# SVR4 /usr/ucb/install, which tries to use the nonexistent group "staff"
-# OS/2's system install, which has a completely different semantic
-# ./install, which can be erroneously created by make from ./install.sh.
-# Reject install programs that cannot install multiple files.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a BSD-compatible install" >&5
-$as_echo_n "checking for a BSD-compatible install... " >&6; }
-if test -z "$INSTALL"; then
-if ${ac_cv_path_install+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    # Account for people who put trailing slashes in PATH elements.
-case $as_dir/ in #((
-  ./ | .// | /[cC]/* | \
-  /etc/* | /usr/sbin/* | /usr/etc/* | /sbin/* | /usr/afsws/bin/* | \
-  ?:[\\/]os2[\\/]install[\\/]* | ?:[\\/]OS2[\\/]INSTALL[\\/]* | \
-  /usr/ucb/* ) ;;
-  *)
-    # OSF1 and SCO ODT 3.0 have their own names for install.
-    # Don't use installbsd from OSF since it installs stuff as root
-    # by default.
-    for ac_prog in ginstall scoinst install; do
-      for ac_exec_ext in '' $ac_executable_extensions; do
-	if as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext"; then
-	  if test $ac_prog = install &&
-	    grep dspmsg "$as_dir/$ac_prog$ac_exec_ext" >/dev/null 2>&1; then
-	    # AIX install.  It has an incompatible calling convention.
-	    :
-	  elif test $ac_prog = install &&
-	    grep pwplus "$as_dir/$ac_prog$ac_exec_ext" >/dev/null 2>&1; then
-	    # program-specific install script used by HP pwplus--don't use.
-	    :
-	  else
-	    rm -rf conftest.one conftest.two conftest.dir
-	    echo one > conftest.one
-	    echo two > conftest.two
-	    mkdir conftest.dir
-	    if "$as_dir/$ac_prog$ac_exec_ext" -c conftest.one conftest.two "`pwd`/conftest.dir" &&
-	      test -s conftest.one && test -s conftest.two &&
-	      test -s conftest.dir/conftest.one &&
-	      test -s conftest.dir/conftest.two
-	    then
-	      ac_cv_path_install="$as_dir/$ac_prog$ac_exec_ext -c"
-	      break 3
-	    fi
-	  fi
-	fi
-      done
-    done
-    ;;
-esac
-
-  done
-IFS=$as_save_IFS
-
-rm -rf conftest.one conftest.two conftest.dir
-
-fi
-  if test "${ac_cv_path_install+set}" = set; then
-    INSTALL=$ac_cv_path_install
-  else
-    # As a last resort, use the slow shell script.  Don't cache a
-    # value for INSTALL within a source directory, because that will
-    # break other packages using the cache if that directory is
-    # removed, or if the value is a relative name.
-    INSTALL=$ac_install_sh
-  fi
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $INSTALL" >&5
-$as_echo "$INSTALL" >&6; }
-
-# Use test -z because SunOS4 sh mishandles braces in ${var-val}.
-# It thinks the first close brace ends the variable substitution.
-test -z "$INSTALL_PROGRAM" && INSTALL_PROGRAM='${INSTALL}'
-
-test -z "$INSTALL_SCRIPT" && INSTALL_SCRIPT='${INSTALL}'
-
-test -z "$INSTALL_DATA" && INSTALL_DATA='${INSTALL} -m 644'
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether build environment is sane" >&5
-$as_echo_n "checking whether build environment is sane... " >&6; }
-# Reject unsafe characters in $srcdir or the absolute working directory
-# name.  Accept space and tab only in the latter.
-am_lf='
-'
-case `pwd` in
-  *[\\\"\#\$\&\'\`$am_lf]*)
-    as_fn_error $? "unsafe absolute working directory name" "$LINENO" 5;;
-esac
-case $srcdir in
-  *[\\\"\#\$\&\'\`$am_lf\ \	]*)
-    as_fn_error $? "unsafe srcdir value: '$srcdir'" "$LINENO" 5;;
-esac
-
-# Do 'set' in a subshell so we don't clobber the current shell's
-# arguments.  Must try -L first in case configure is actually a
-# symlink; some systems play weird games with the mod time of symlinks
-# (eg FreeBSD returns the mod time of the symlink's containing
-# directory).
-if (
-   am_has_slept=no
-   for am_try in 1 2; do
-     echo "timestamp, slept: $am_has_slept" > conftest.file
-     set X `ls -Lt "$srcdir/configure" conftest.file 2> /dev/null`
-     if test "$*" = "X"; then
-	# -L didn't work.
-	set X `ls -t "$srcdir/configure" conftest.file`
-     fi
-     if test "$*" != "X $srcdir/configure conftest.file" \
-	&& test "$*" != "X conftest.file $srcdir/configure"; then
-
-	# If neither matched, then we have a broken ls.  This can happen
-	# if, for instance, CONFIG_SHELL is bash and it inherits a
-	# broken ls alias from the environment.  This has actually
-	# happened.  Such a system could not be considered "sane".
-	as_fn_error $? "ls -t appears to fail.  Make sure there is not a broken
-  alias in your environment" "$LINENO" 5
-     fi
-     if test "$2" = conftest.file || test $am_try -eq 2; then
-       break
-     fi
-     # Just in case.
-     sleep 1
-     am_has_slept=yes
-   done
-   test "$2" = conftest.file
-   )
-then
-   # Ok.
-   :
-else
-   as_fn_error $? "newly created file is older than distributed files!
-Check your system clock" "$LINENO" 5
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
-$as_echo "yes" >&6; }
-# If we didn't sleep, we still need to ensure time stamps of config.status and
-# generated files are strictly newer.
-am_sleep_pid=
-if grep 'slept: no' conftest.file >/dev/null 2>&1; then
-  ( sleep 1 ) &
-  am_sleep_pid=$!
-fi
-
-rm -f conftest.file
-
-test "$program_prefix" != NONE &&
-  program_transform_name="s&^&$program_prefix&;$program_transform_name"
-# Use a double $ so make ignores it.
-test "$program_suffix" != NONE &&
-  program_transform_name="s&\$&$program_suffix&;$program_transform_name"
-# Double any \ or $.
-# By default was `s,x,x', remove it if useless.
-ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
-program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
-
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
-
-if test x"${MISSING+set}" != xset; then
-  case $am_aux_dir in
-  *\ * | *\	*)
-    MISSING="\${SHELL} \"$am_aux_dir/missing\"" ;;
-  *)
-    MISSING="\${SHELL} $am_aux_dir/missing" ;;
-  esac
-fi
-# Use eval to expand $SHELL
-if eval "$MISSING --is-lightweight"; then
-  am_missing_run="$MISSING "
-else
-  am_missing_run=
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: 'missing' script is too old or missing" >&5
-$as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
-fi
-
-if test x"${install_sh}" != xset; then
-  case $am_aux_dir in
-  *\ * | *\	*)
-    install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
-  *)
-    install_sh="\${SHELL} $am_aux_dir/install-sh"
-  esac
-fi
-
-# Installed binaries are usually stripped using 'strip' when the user
-# run "make install-strip".  However 'strip' might not be the right
-# tool to use in cross-compilation environments, therefore Automake
-# will honor the 'STRIP' environment variable to overrule this program.
-if test "$cross_compiling" != no; then
-  if test -n "$ac_tool_prefix"; then
-  # Extract the first word of "${ac_tool_prefix}strip", so it can be a program name with args.
-set dummy ${ac_tool_prefix}strip; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_STRIP+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$STRIP"; then
-  ac_cv_prog_STRIP="$STRIP" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_STRIP="${ac_tool_prefix}strip"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-STRIP=$ac_cv_prog_STRIP
-if test -n "$STRIP"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $STRIP" >&5
-$as_echo "$STRIP" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-
-fi
-if test -z "$ac_cv_prog_STRIP"; then
-  ac_ct_STRIP=$STRIP
-  # Extract the first word of "strip", so it can be a program name with args.
-set dummy strip; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_STRIP+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$ac_ct_STRIP"; then
-  ac_cv_prog_ac_ct_STRIP="$ac_ct_STRIP" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_ac_ct_STRIP="strip"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-ac_ct_STRIP=$ac_cv_prog_ac_ct_STRIP
-if test -n "$ac_ct_STRIP"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_STRIP" >&5
-$as_echo "$ac_ct_STRIP" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-  if test "x$ac_ct_STRIP" = x; then
-    STRIP=":"
-  else
-    case $cross_compiling:$ac_tool_warned in
-yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
-$as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
-ac_tool_warned=yes ;;
-esac
-    STRIP=$ac_ct_STRIP
-  fi
-else
-  STRIP="$ac_cv_prog_STRIP"
-fi
-
-fi
-INSTALL_STRIP_PROGRAM="\$(install_sh) -c -s"
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for a thread-safe mkdir -p" >&5
-$as_echo_n "checking for a thread-safe mkdir -p... " >&6; }
-if test -z "$MKDIR_P"; then
-  if ${ac_cv_path_mkdir+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH$PATH_SEPARATOR/opt/sfw/bin
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_prog in mkdir gmkdir; do
-	 for ac_exec_ext in '' $ac_executable_extensions; do
-	   as_fn_executable_p "$as_dir/$ac_prog$ac_exec_ext" || continue
-	   case `"$as_dir/$ac_prog$ac_exec_ext" --version 2>&1` in #(
-	     'mkdir (GNU coreutils) '* | \
-	     'mkdir (coreutils) '* | \
-	     'mkdir (fileutils) '4.1*)
-	       ac_cv_path_mkdir=$as_dir/$ac_prog$ac_exec_ext
-	       break 3;;
-	   esac
-	 done
-       done
-  done
-IFS=$as_save_IFS
-
-fi
-
-  test -d ./--version && rmdir ./--version
-  if test "${ac_cv_path_mkdir+set}" = set; then
-    MKDIR_P="$ac_cv_path_mkdir -p"
-  else
-    # As a last resort, use the slow shell script.  Don't cache a
-    # value for MKDIR_P within a source directory, because that will
-    # break other packages using the cache if that directory is
-    # removed, or if the value is a relative name.
-    MKDIR_P="$ac_install_sh -d"
-  fi
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $MKDIR_P" >&5
-$as_echo "$MKDIR_P" >&6; }
-
-for ac_prog in gawk mawk nawk awk
-do
-  # Extract the first word of "$ac_prog", so it can be a program name with args.
-set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_AWK+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$AWK"; then
-  ac_cv_prog_AWK="$AWK" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_AWK="$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-AWK=$ac_cv_prog_AWK
-if test -n "$AWK"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $AWK" >&5
-$as_echo "$AWK" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-
-  test -n "$AWK" && break
-done
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether ${MAKE-make} sets \$(MAKE)" >&5
-$as_echo_n "checking whether ${MAKE-make} sets \$(MAKE)... " >&6; }
-set x ${MAKE-make}
-ac_make=`$as_echo "$2" | sed 's/+/p/g; s/[^a-zA-Z0-9_]/_/g'`
-if eval \${ac_cv_prog_make_${ac_make}_set+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat >conftest.make <<\_ACEOF
-SHELL = /bin/sh
-all:
-	@echo '@@@%%%=$(MAKE)=@@@%%%'
-_ACEOF
-# GNU make sometimes prints "make[1]: Entering ...", which would confuse us.
-case `${MAKE-make} -f conftest.make 2>/dev/null` in
-  *@@@%%%=?*=@@@%%%*)
-    eval ac_cv_prog_make_${ac_make}_set=yes;;
-  *)
-    eval ac_cv_prog_make_${ac_make}_set=no;;
-esac
-rm -f conftest.make
-fi
-if eval test \$ac_cv_prog_make_${ac_make}_set = yes; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
-$as_echo "yes" >&6; }
-  SET_MAKE=
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-  SET_MAKE="MAKE=${MAKE-make}"
-fi
-
-rm -rf .tst 2>/dev/null
-mkdir .tst 2>/dev/null
-if test -d .tst; then
-  am__leading_dot=.
-else
-  am__leading_dot=_
-fi
-rmdir .tst 2>/dev/null
-
-# Check whether --enable-silent-rules was given.
-if test "${enable_silent_rules+set}" = set; then :
-  enableval=$enable_silent_rules;
-fi
-
-case $enable_silent_rules in # (((
-  yes) AM_DEFAULT_VERBOSITY=0;;
-   no) AM_DEFAULT_VERBOSITY=1;;
-    *) AM_DEFAULT_VERBOSITY=1;;
-esac
-am_make=${MAKE-make}
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $am_make supports nested variables" >&5
-$as_echo_n "checking whether $am_make supports nested variables... " >&6; }
-if ${am_cv_make_support_nested_variables+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if $as_echo 'TRUE=$(BAR$(V))
-BAR0=false
-BAR1=true
-V=1
-am__doit:
-	@$(TRUE)
-.PHONY: am__doit' | $am_make -f - >/dev/null 2>&1; then
-  am_cv_make_support_nested_variables=yes
-else
-  am_cv_make_support_nested_variables=no
-fi
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_make_support_nested_variables" >&5
-$as_echo "$am_cv_make_support_nested_variables" >&6; }
-if test $am_cv_make_support_nested_variables = yes; then
-    AM_V='$(V)'
-  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
-else
-  AM_V=$AM_DEFAULT_VERBOSITY
-  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
-fi
-AM_BACKSLASH='\'
-
-if test "`cd $srcdir && pwd`" != "`pwd`"; then
-  # Use -I$(srcdir) only when $(srcdir) != ., so that make's output
-  # is not polluted with repeated "-I."
-  am__isrc=' -I$(srcdir)'
-  # test to see if srcdir already configured
-  if test -f $srcdir/config.status; then
-    as_fn_error $? "source directory already configured; run \"make distclean\" there first" "$LINENO" 5
-  fi
-fi
-
-# test whether we have cygpath
-if test -z "$CYGPATH_W"; then
-  if (cygpath --version) >/dev/null 2>/dev/null; then
-    CYGPATH_W='cygpath -w'
-  else
-    CYGPATH_W=echo
-  fi
-fi
-
-
-# Define the identity of the package.
- PACKAGE='grid'
- VERSION='1.0'
-
-
-cat >>confdefs.h <<_ACEOF
-#define PACKAGE "$PACKAGE"
-_ACEOF
-
-
-cat >>confdefs.h <<_ACEOF
-#define VERSION "$VERSION"
-_ACEOF
-
-# Some tools Automake needs.
-
-ACLOCAL=${ACLOCAL-"${am_missing_run}aclocal-${am__api_version}"}
-
-
-AUTOCONF=${AUTOCONF-"${am_missing_run}autoconf"}
-
-
-AUTOMAKE=${AUTOMAKE-"${am_missing_run}automake-${am__api_version}"}
-
-
-AUTOHEADER=${AUTOHEADER-"${am_missing_run}autoheader"}
-
-
-MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
-
-# For better backward compatibility.  To be removed once Automake 1.9.x
-# dies out for good.  For more background, see:
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00001.html>
-# <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
-mkdir_p='$(MKDIR_P)'
-
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
-# Always define AMTAR for backward compatibility.  Yes, it's still used
-# in the wild :-(  We should find a proper way to deprecate it ...
-AMTAR='$${TAR-tar}'
-
-
-# We'll loop over all known methods to create a tar archive until one works.
-_am_tools='gnutar  pax cpio none'
-
-am__tar='$${TAR-tar} chof - "$$tardir"' am__untar='$${TAR-tar} xf -'
-
-
-
-
-
-
-# POSIX will say in a future version that running "rm -f" with no argument
-# is OK; and we want to be able to make that assumption in our Makefile
-# recipes.  So use an aggressive probe to check that the usage we want is
-# actually supported "in the wild" to an acceptable degree.
-# See automake bug#10828.
-# To make any issue more visible, cause the running configure to be aborted
-# by default if the 'rm' program in use doesn't match our expectations; the
-# user can still override this though.
-if rm -f && rm -fr && rm -rf; then : OK; else
-  cat >&2 <<'END'
-Oops!
-
-Your 'rm' program seems unable to run without file operands specified
-on the command line, even when the '-f' option is present.  This is contrary
-to the behaviour of most rm programs out there, and not conforming with
-the upcoming POSIX standard: <http://austingroupbugs.net/view.php?id=542>
-
-Please tell bug-automake@gnu.org about your system, including the value
-of your $PATH and any error possibly output before this message.  This
-can help us improve future automake versions.
-
-END
-  if test x"$ACCEPT_INFERIOR_RM_PROGRAM" = x"yes"; then
-    echo 'Configuration will proceed anyway, since you have set the' >&2
-    echo 'ACCEPT_INFERIOR_RM_PROGRAM variable to "yes"' >&2
-    echo >&2
-  else
-    cat >&2 <<'END'
-Aborting the configuration process, to ensure you take notice of the issue.
-
-You can download and install GNU coreutils to get an 'rm' implementation
-that behaves properly: <http://www.gnu.org/software/coreutils/>.
-
-If you want to complete the configuration process using your problematic
-'rm' anyway, export the environment variable ACCEPT_INFERIOR_RM_PROGRAM
-to "yes", and re-run configure.
-
-END
-    as_fn_error $? "Your 'rm' program is bad, sorry." "$LINENO" 5
-  fi
-fi
-
-
-ac_config_headers="$ac_config_headers lib/GridConfig.h"
-
-# Check whether --enable-silent-rules was given.
-if test "${enable_silent_rules+set}" = set; then :
-  enableval=$enable_silent_rules;
-fi
-
-case $enable_silent_rules in # (((
-  yes) AM_DEFAULT_VERBOSITY=0;;
-   no) AM_DEFAULT_VERBOSITY=1;;
-    *) AM_DEFAULT_VERBOSITY=0;;
-esac
-am_make=${MAKE-make}
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $am_make supports nested variables" >&5
-$as_echo_n "checking whether $am_make supports nested variables... " >&6; }
-if ${am_cv_make_support_nested_variables+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if $as_echo 'TRUE=$(BAR$(V))
-BAR0=false
-BAR1=true
-V=1
-am__doit:
-	@$(TRUE)
-.PHONY: am__doit' | $am_make -f - >/dev/null 2>&1; then
-  am_cv_make_support_nested_variables=yes
-else
-  am_cv_make_support_nested_variables=no
-fi
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_make_support_nested_variables" >&5
-$as_echo "$am_cv_make_support_nested_variables" >&6; }
-if test $am_cv_make_support_nested_variables = yes; then
-    AM_V='$(V)'
-  AM_DEFAULT_V='$(AM_DEFAULT_VERBOSITY)'
-else
-  AM_V=$AM_DEFAULT_VERBOSITY
-  AM_DEFAULT_V=$AM_DEFAULT_VERBOSITY
-fi
-AM_BACKSLASH='\'
-
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}:
-
-:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
-Configuring $PACKAGE v$VERSION  for $host
-:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
-" >&5
-$as_echo "$as_me:
-
-:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
-Configuring $PACKAGE v$VERSION  for $host
-:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
-" >&6;}
-
-# Checks for programs.
-ac_ext=cpp
-ac_cpp='$CXXCPP $CPPFLAGS'
-ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
-
-ac_ext=cpp
-ac_cpp='$CXXCPP $CPPFLAGS'
-ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
-if test -z "$CXX"; then
-  if test -n "$CCC"; then
-    CXX=$CCC
-  else
-    if test -n "$ac_tool_prefix"; then
-  for ac_prog in g++ c++ gpp aCC CC cxx cc++ cl.exe FCC KCC RCC xlC_r xlC
-  do
-    # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
-set dummy $ac_tool_prefix$ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CXX+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$CXX"; then
-  ac_cv_prog_CXX="$CXX" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_CXX="$ac_tool_prefix$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-CXX=$ac_cv_prog_CXX
-if test -n "$CXX"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CXX" >&5
-$as_echo "$CXX" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-
-    test -n "$CXX" && break
-  done
-fi
-if test -z "$CXX"; then
-  ac_ct_CXX=$CXX
-  for ac_prog in g++ c++ gpp aCC CC cxx cc++ cl.exe FCC KCC RCC xlC_r xlC
-do
-  # Extract the first word of "$ac_prog", so it can be a program name with args.
-set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_CXX+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$ac_ct_CXX"; then
-  ac_cv_prog_ac_ct_CXX="$ac_ct_CXX" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_ac_ct_CXX="$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-ac_ct_CXX=$ac_cv_prog_ac_ct_CXX
-if test -n "$ac_ct_CXX"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CXX" >&5
-$as_echo "$ac_ct_CXX" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-
-  test -n "$ac_ct_CXX" && break
-done
-
-  if test "x$ac_ct_CXX" = x; then
-    CXX="g++"
-  else
-    case $cross_compiling:$ac_tool_warned in
-yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
-$as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
-ac_tool_warned=yes ;;
-esac
-    CXX=$ac_ct_CXX
-  fi
-fi
-
-  fi
-fi
-# Provide some information about the compiler.
-$as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler version" >&5
-set X $ac_compile
-ac_compiler=$2
-for ac_option in --version -v -V -qversion; do
-  { { ac_try="$ac_compiler $ac_option >&5"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    sed '10a\
-... rest of stderr output deleted ...
-         10q' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-  fi
-  rm -f conftest.er1 conftest.err
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }
-done
-
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-ac_clean_files_save=$ac_clean_files
-ac_clean_files="$ac_clean_files a.out a.out.dSYM a.exe b.out"
-# Try to create an executable without -o first, disregard a.out.
-# It will help us diagnose broken compilers, and finding out an intuition
-# of exeext.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether the C++ compiler works" >&5
-$as_echo_n "checking whether the C++ compiler works... " >&6; }
-ac_link_default=`$as_echo "$ac_link" | sed 's/ -o *conftest[^ ]*//'`
-
-# The possible output files:
-ac_files="a.out conftest.exe conftest a.exe a_out.exe b.out conftest.*"
-
-ac_rmfiles=
-for ac_file in $ac_files
-do
-  case $ac_file in
-    *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | *.map | *.inf | *.dSYM | *.o | *.obj ) ;;
-    * ) ac_rmfiles="$ac_rmfiles $ac_file";;
-  esac
-done
-rm -f $ac_rmfiles
-
-if { { ac_try="$ac_link_default"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_link_default") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; then :
-  # Autoconf-2.13 could set the ac_cv_exeext variable to `no'.
-# So ignore a value of `no', otherwise this would lead to `EXEEXT = no'
-# in a Makefile.  We should not override ac_cv_exeext if it was cached,
-# so that the user can short-circuit this test for compilers unknown to
-# Autoconf.
-for ac_file in $ac_files ''
-do
-  test -f "$ac_file" || continue
-  case $ac_file in
-    *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | *.map | *.inf | *.dSYM | *.o | *.obj )
-	;;
-    [ab].out )
-	# We found the default executable, but exeext='' is most
-	# certainly right.
-	break;;
-    *.* )
-	if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
-	then :; else
-	   ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
-	fi
-	# We set ac_cv_exeext here because the later test for it is not
-	# safe: cross compilers may not add the suffix if given an `-o'
-	# argument, so we may need to know it at that point already.
-	# Even if this section looks crufty: it has the advantage of
-	# actually working.
-	break;;
-    * )
-	break;;
-  esac
-done
-test "$ac_cv_exeext" = no && ac_cv_exeext=
-
-else
-  ac_file=''
-fi
-if test -z "$ac_file"; then :
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-$as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error 77 "C++ compiler cannot create executables
-See \`config.log' for more details" "$LINENO" 5; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
-$as_echo "yes" >&6; }
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for C++ compiler default output file name" >&5
-$as_echo_n "checking for C++ compiler default output file name... " >&6; }
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_file" >&5
-$as_echo "$ac_file" >&6; }
-ac_exeext=$ac_cv_exeext
-
-rm -f -r a.out a.out.dSYM a.exe conftest$ac_cv_exeext b.out
-ac_clean_files=$ac_clean_files_save
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of executables" >&5
-$as_echo_n "checking for suffix of executables... " >&6; }
-if { { ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_link") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; then :
-  # If both `conftest.exe' and `conftest' are `present' (well, observable)
-# catch `conftest.exe'.  For instance with Cygwin, `ls conftest' will
-# work properly (i.e., refer to `conftest.exe'), while it won't with
-# `rm'.
-for ac_file in conftest.exe conftest conftest.*; do
-  test -f "$ac_file" || continue
-  case $ac_file in
-    *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | *.map | *.inf | *.dSYM | *.o | *.obj ) ;;
-    *.* ) ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
-	  break;;
-    * ) break;;
-  esac
-done
-else
-  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "cannot compute suffix of executables: cannot compile and link
-See \`config.log' for more details" "$LINENO" 5; }
-fi
-rm -f conftest conftest$ac_cv_exeext
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_exeext" >&5
-$as_echo "$ac_cv_exeext" >&6; }
-
-rm -f conftest.$ac_ext
-EXEEXT=$ac_cv_exeext
-ac_exeext=$EXEEXT
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <stdio.h>
-int
-main ()
-{
-FILE *f = fopen ("conftest.out", "w");
- return ferror (f) || fclose (f) != 0;
-
-  ;
-  return 0;
-}
-_ACEOF
-ac_clean_files="$ac_clean_files conftest.out"
-# Check that the compiler produces executables we can run.  If not, either
-# the compiler is broken, or we cross compile.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are cross compiling" >&5
-$as_echo_n "checking whether we are cross compiling... " >&6; }
-if test "$cross_compiling" != yes; then
-  { { ac_try="$ac_link"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_link") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }
-  if { ac_try='./conftest$ac_cv_exeext'
-  { { case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_try") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; }; then
-    cross_compiling=no
-  else
-    if test "$cross_compiling" = maybe; then
-	cross_compiling=yes
-    else
-	{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "cannot run C++ compiled programs.
-If you meant to cross compile, use \`--host'.
-See \`config.log' for more details" "$LINENO" 5; }
-    fi
-  fi
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $cross_compiling" >&5
-$as_echo "$cross_compiling" >&6; }
-
-rm -f conftest.$ac_ext conftest$ac_cv_exeext conftest.out
-ac_clean_files=$ac_clean_files_save
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for suffix of object files" >&5
-$as_echo_n "checking for suffix of object files... " >&6; }
-if ${ac_cv_objext+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-rm -f conftest.o conftest.obj
-if { { ac_try="$ac_compile"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compile") 2>&5
-  ac_status=$?
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }; then :
-  for ac_file in conftest.o conftest.obj conftest.*; do
-  test -f "$ac_file" || continue;
-  case $ac_file in
-    *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | *.map | *.inf | *.dSYM ) ;;
-    *) ac_cv_objext=`expr "$ac_file" : '.*\.\(.*\)'`
-       break;;
-  esac
-done
-else
-  $as_echo "$as_me: failed program was:" >&5
-sed 's/^/| /' conftest.$ac_ext >&5
-
-{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "cannot compute suffix of object files: cannot compile
-See \`config.log' for more details" "$LINENO" 5; }
-fi
-rm -f conftest.$ac_cv_objext conftest.$ac_ext
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_objext" >&5
-$as_echo "$ac_cv_objext" >&6; }
-OBJEXT=$ac_cv_objext
-ac_objext=$OBJEXT
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C++ compiler" >&5
-$as_echo_n "checking whether we are using the GNU C++ compiler... " >&6; }
-if ${ac_cv_cxx_compiler_gnu+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-#ifndef __GNUC__
-       choke me
-#endif
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  ac_compiler_gnu=yes
-else
-  ac_compiler_gnu=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-ac_cv_cxx_compiler_gnu=$ac_compiler_gnu
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_cxx_compiler_gnu" >&5
-$as_echo "$ac_cv_cxx_compiler_gnu" >&6; }
-if test $ac_compiler_gnu = yes; then
-  GXX=yes
-else
-  GXX=
-fi
-ac_test_CXXFLAGS=${CXXFLAGS+set}
-ac_save_CXXFLAGS=$CXXFLAGS
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX accepts -g" >&5
-$as_echo_n "checking whether $CXX accepts -g... " >&6; }
-if ${ac_cv_prog_cxx_g+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  ac_save_cxx_werror_flag=$ac_cxx_werror_flag
-   ac_cxx_werror_flag=yes
-   ac_cv_prog_cxx_g=no
-   CXXFLAGS="-g"
-   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  ac_cv_prog_cxx_g=yes
-else
-  CXXFLAGS=""
-      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-
-else
-  ac_cxx_werror_flag=$ac_save_cxx_werror_flag
-	 CXXFLAGS="-g"
-	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  ac_cv_prog_cxx_g=yes
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-   ac_cxx_werror_flag=$ac_save_cxx_werror_flag
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cxx_g" >&5
-$as_echo "$ac_cv_prog_cxx_g" >&6; }
-if test "$ac_test_CXXFLAGS" = set; then
-  CXXFLAGS=$ac_save_CXXFLAGS
-elif test $ac_cv_prog_cxx_g = yes; then
-  if test "$GXX" = yes; then
-    CXXFLAGS="-g -O2"
-  else
-    CXXFLAGS="-g"
-  fi
-else
-  if test "$GXX" = yes; then
-    CXXFLAGS="-O2"
-  else
-    CXXFLAGS=
-  fi
-fi
-ac_ext=cpp
-ac_cpp='$CXXCPP $CPPFLAGS'
-ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
-DEPDIR="${am__leading_dot}deps"
-
-ac_config_commands="$ac_config_commands depfiles"
-
-
-am_make=${MAKE-make}
-cat > confinc << 'END'
-am__doit:
-	@echo this is the am__doit target
-.PHONY: am__doit
-END
-# If we don't find an include directive, just comment out the code.
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for style of include used by $am_make" >&5
-$as_echo_n "checking for style of include used by $am_make... " >&6; }
-am__include="#"
-am__quote=
-_am_result=none
-# First try GNU make style include.
-echo "include confinc" > confmf
-# Ignore all kinds of additional output from 'make'.
-case `$am_make -s -f confmf 2> /dev/null` in #(
-*the\ am__doit\ target*)
-  am__include=include
-  am__quote=
-  _am_result=GNU
-  ;;
-esac
-# Now try BSD make style include.
-if test "$am__include" = "#"; then
-   echo '.include "confinc"' > confmf
-   case `$am_make -s -f confmf 2> /dev/null` in #(
-   *the\ am__doit\ target*)
-     am__include=.include
-     am__quote="\""
-     _am_result=BSD
-     ;;
-   esac
-fi
-
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $_am_result" >&5
-$as_echo "$_am_result" >&6; }
-rm -f confinc confmf
-
-# Check whether --enable-dependency-tracking was given.
-if test "${enable_dependency_tracking+set}" = set; then :
-  enableval=$enable_dependency_tracking;
-fi
-
-if test "x$enable_dependency_tracking" != xno; then
-  am_depcomp="$ac_aux_dir/depcomp"
-  AMDEPBACKSLASH='\'
-  am__nodep='_no'
-fi
- if test "x$enable_dependency_tracking" != xno; then
-  AMDEP_TRUE=
-  AMDEP_FALSE='#'
-else
-  AMDEP_TRUE='#'
-  AMDEP_FALSE=
-fi
-
-
-
-depcc="$CXX"  am_compiler_list=
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
-$as_echo_n "checking dependency style of $depcc... " >&6; }
-if ${am_cv_CXX_dependencies_compiler_type+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
-  # We make a subdir and do the tests there.  Otherwise we can end up
-  # making bogus files that we don't know about and never remove.  For
-  # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named 'D' -- because '-MD' means "put the output
-  # in D".
-  rm -rf conftest.dir
-  mkdir conftest.dir
-  # Copy depcomp to subdir because otherwise we won't find it if we're
-  # using a relative directory.
-  cp "$am_depcomp" conftest.dir
-  cd conftest.dir
-  # We will build objects and dependencies in a subdirectory because
-  # it helps to detect inapplicable dependency modes.  For instance
-  # both Tru64's cc and ICC support -MD to output dependencies as a
-  # side effect of compilation, but ICC will put the dependencies in
-  # the current directory while Tru64 will put them in the object
-  # directory.
-  mkdir sub
-
-  am_cv_CXX_dependencies_compiler_type=none
-  if test "$am_compiler_list" = ""; then
-     am_compiler_list=`sed -n 's/^#*\([a-zA-Z0-9]*\))$/\1/p' < ./depcomp`
-  fi
-  am__universal=false
-  case " $depcc " in #(
-     *\ -arch\ *\ -arch\ *) am__universal=true ;;
-     esac
-
-  for depmode in $am_compiler_list; do
-    # Setup a source with many dependencies, because some compilers
-    # like to wrap large dependency lists on column 80 (with \), and
-    # we should not choose a depcomp mode which is confused by this.
-    #
-    # We need to recreate these files for each test, as the compiler may
-    # overwrite some of them when testing with obscure command lines.
-    # This happens at least with the AIX C compiler.
-    : > sub/conftest.c
-    for i in 1 2 3 4 5 6; do
-      echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
-      # Solaris 10 /bin/sh.
-      echo '/* dummy */' > sub/conftst$i.h
-    done
-    echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
-
-    # We check with '-c' and '-o' for the sake of the "dashmstdout"
-    # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle '-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs.
-    am__obj=sub/conftest.${OBJEXT-o}
-    am__minus_obj="-o $am__obj"
-    case $depmode in
-    gcc)
-      # This depmode causes a compiler race in universal mode.
-      test "$am__universal" = false || continue
-      ;;
-    nosideeffect)
-      # After this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested.
-      if test "x$enable_dependency_tracking" = xyes; then
-	continue
-      else
-	break
-      fi
-      ;;
-    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
-      # This compiler won't grok '-c -o', but also, the minuso test has
-      # not run yet.  These depmodes are late enough in the game, and
-      # so weak that their functioning should not be impacted.
-      am__obj=conftest.${OBJEXT-o}
-      am__minus_obj=
-      ;;
-    none) break ;;
-    esac
-    if depmode=$depmode \
-       source=sub/conftest.c object=$am__obj \
-       depfile=sub/conftest.Po tmpdepfile=sub/conftest.TPo \
-       $SHELL ./depcomp $depcc -c $am__minus_obj sub/conftest.c \
-         >/dev/null 2>conftest.err &&
-       grep sub/conftst1.h sub/conftest.Po > /dev/null 2>&1 &&
-       grep sub/conftst6.h sub/conftest.Po > /dev/null 2>&1 &&
-       grep $am__obj sub/conftest.Po > /dev/null 2>&1 &&
-       ${MAKE-make} -s -f confmf > /dev/null 2>&1; then
-      # icc doesn't choke on unknown options, it will just issue warnings
-      # or remarks (even with -Werror).  So we grep stderr for any message
-      # that says an option was ignored or not supported.
-      # When given -MP, icc 7.0 and 7.1 complain thusly:
-      #   icc: Command line warning: ignoring option '-M'; no argument required
-      # The diagnosis changed in icc 8.0:
-      #   icc: Command line remark: option '-MP' not supported
-      if (grep 'ignoring option' conftest.err ||
-          grep 'not supported' conftest.err) >/dev/null 2>&1; then :; else
-        am_cv_CXX_dependencies_compiler_type=$depmode
-        break
-      fi
-    fi
-  done
-
-  cd ..
-  rm -rf conftest.dir
-else
-  am_cv_CXX_dependencies_compiler_type=none
-fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CXX_dependencies_compiler_type" >&5
-$as_echo "$am_cv_CXX_dependencies_compiler_type" >&6; }
-CXXDEPMODE=depmode=$am_cv_CXX_dependencies_compiler_type
-
- if
-  test "x$enable_dependency_tracking" != xno \
-  && test "$am_cv_CXX_dependencies_compiler_type" = gcc3; then
-  am__fastdepCXX_TRUE=
-  am__fastdepCXX_FALSE='#'
-else
-  am__fastdepCXX_TRUE='#'
-  am__fastdepCXX_FALSE=
-fi
-
-
-
-
-  OPENMP_CXXFLAGS=
-  # Check whether --enable-openmp was given.
-if test "${enable_openmp+set}" = set; then :
-  enableval=$enable_openmp;
-fi
-
-  if test "$enable_openmp" != no; then
-    { $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CXX option to support OpenMP" >&5
-$as_echo_n "checking for $CXX option to support OpenMP... " >&6; }
-if ${ac_cv_prog_cxx_openmp+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-#ifndef _OPENMP
- choke me
-#endif
-#include <omp.h>
-int main () { return omp_get_num_threads (); }
-
-_ACEOF
-if ac_fn_cxx_try_link "$LINENO"; then :
-  ac_cv_prog_cxx_openmp='none needed'
-else
-  ac_cv_prog_cxx_openmp='unsupported'
-	  	  	  	  	  	  	                                	  	  	  	  	  	  for ac_option in -fopenmp -xopenmp -openmp -mp -omp -qsmp=omp -homp \
-                           -Popenmp --openmp; do
-	    ac_save_CXXFLAGS=$CXXFLAGS
-	    CXXFLAGS="$CXXFLAGS $ac_option"
-	    cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-#ifndef _OPENMP
- choke me
-#endif
-#include <omp.h>
-int main () { return omp_get_num_threads (); }
-
-_ACEOF
-if ac_fn_cxx_try_link "$LINENO"; then :
-  ac_cv_prog_cxx_openmp=$ac_option
-fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
-	    CXXFLAGS=$ac_save_CXXFLAGS
-	    if test "$ac_cv_prog_cxx_openmp" != unsupported; then
-	      break
-	    fi
-	  done
-fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cxx_openmp" >&5
-$as_echo "$ac_cv_prog_cxx_openmp" >&6; }
-    case $ac_cv_prog_cxx_openmp in #(
-      "none needed" | unsupported)
-	;; #(
-      *)
-	OPENMP_CXXFLAGS=$ac_cv_prog_cxx_openmp ;;
-    esac
-  fi
-
-
-if test -n "$ac_tool_prefix"; then
-  # Extract the first word of "${ac_tool_prefix}ranlib", so it can be a program name with args.
-set dummy ${ac_tool_prefix}ranlib; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_RANLIB+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$RANLIB"; then
-  ac_cv_prog_RANLIB="$RANLIB" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_RANLIB="${ac_tool_prefix}ranlib"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-RANLIB=$ac_cv_prog_RANLIB
-if test -n "$RANLIB"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $RANLIB" >&5
-$as_echo "$RANLIB" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-
-fi
-if test -z "$ac_cv_prog_RANLIB"; then
-  ac_ct_RANLIB=$RANLIB
-  # Extract the first word of "ranlib", so it can be a program name with args.
-set dummy ranlib; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_RANLIB+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$ac_ct_RANLIB"; then
-  ac_cv_prog_ac_ct_RANLIB="$ac_ct_RANLIB" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_ac_ct_RANLIB="ranlib"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-ac_ct_RANLIB=$ac_cv_prog_ac_ct_RANLIB
-if test -n "$ac_ct_RANLIB"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_RANLIB" >&5
-$as_echo "$ac_ct_RANLIB" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-  if test "x$ac_ct_RANLIB" = x; then
-    RANLIB=":"
-  else
-    case $cross_compiling:$ac_tool_warned in
-yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
-$as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
-ac_tool_warned=yes ;;
-esac
-    RANLIB=$ac_ct_RANLIB
-  fi
-else
-  RANLIB="$ac_cv_prog_RANLIB"
-fi
-
-    ax_cxx_compile_cxx11_required=true
-  ac_ext=cpp
-ac_cpp='$CXXCPP $CPPFLAGS'
-ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
-  ac_success=no
-  { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX supports C++11 features by default" >&5
-$as_echo_n "checking whether $CXX supports C++11 features by default... " >&6; }
-if ${ax_cv_cxx_compile_cxx11+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-  template <typename T>
-    struct check
-    {
-      static_assert(sizeof(int) <= sizeof(T), "not big enough");
-    };
-
-    struct Base {
-    virtual void f() {}
-    };
-    struct Child : public Base {
-    virtual void f() override {}
-    };
-
-    typedef check<check<bool>> right_angle_brackets;
-
-    int a;
-    decltype(a) b;
-
-    typedef check<int> check_type;
-    check_type c;
-    check_type&& cr = static_cast<check_type&&>(c);
-
-    auto d = a;
-    auto l = [](){};
-    // Prevent Clang error: unused variable 'l' [-Werror,-Wunused-variable]
-    struct use_l { use_l() { l(); } };
-
-    // http://stackoverflow.com/questions/13728184/template-aliases-and-sfinae
-    // Clang 3.1 fails with headers of libstd++ 4.8.3 when using std::function because of this
-    namespace test_template_alias_sfinae {
-        struct foo {};
-
-        template<typename T>
-        using member = typename T::member_type;
-
-        template<typename T>
-        void func(...) {}
-
-        template<typename T>
-        void func(member<T>*) {}
-
-        void test();
-
-        void test() {
-            func<foo>(0);
-        }
-    }
-
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  ax_cv_cxx_compile_cxx11=yes
-else
-  ax_cv_cxx_compile_cxx11=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_cxx_compile_cxx11" >&5
-$as_echo "$ax_cv_cxx_compile_cxx11" >&6; }
-  if test x$ax_cv_cxx_compile_cxx11 = xyes; then
-    ac_success=yes
-  fi
-
-
-
-    if test x$ac_success = xno; then
-            for switch in -std=c++11 -std=c++0x +std=c++11; do
-      cachevar=`$as_echo "ax_cv_cxx_compile_cxx11_$switch" | $as_tr_sh`
-      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CXX supports C++11 features with $switch" >&5
-$as_echo_n "checking whether $CXX supports C++11 features with $switch... " >&6; }
-if eval \${$cachevar+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  ac_save_CXXFLAGS="$CXXFLAGS"
-         CXXFLAGS="$CXXFLAGS $switch"
-         cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-  template <typename T>
-    struct check
-    {
-      static_assert(sizeof(int) <= sizeof(T), "not big enough");
-    };
-
-    struct Base {
-    virtual void f() {}
-    };
-    struct Child : public Base {
-    virtual void f() override {}
-    };
-
-    typedef check<check<bool>> right_angle_brackets;
-
-    int a;
-    decltype(a) b;
-
-    typedef check<int> check_type;
-    check_type c;
-    check_type&& cr = static_cast<check_type&&>(c);
-
-    auto d = a;
-    auto l = [](){};
-    // Prevent Clang error: unused variable 'l' [-Werror,-Wunused-variable]
-    struct use_l { use_l() { l(); } };
-
-    // http://stackoverflow.com/questions/13728184/template-aliases-and-sfinae
-    // Clang 3.1 fails with headers of libstd++ 4.8.3 when using std::function because of this
-    namespace test_template_alias_sfinae {
-        struct foo {};
-
-        template<typename T>
-        using member = typename T::member_type;
-
-        template<typename T>
-        void func(...) {}
-
-        template<typename T>
-        void func(member<T>*) {}
-
-        void test();
-
-        void test() {
-            func<foo>(0);
-        }
-    }
-
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  eval $cachevar=yes
-else
-  eval $cachevar=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-         CXXFLAGS="$ac_save_CXXFLAGS"
-fi
-eval ac_res=\$$cachevar
-	       { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_res" >&5
-$as_echo "$ac_res" >&6; }
-      if eval test x\$$cachevar = xyes; then
-        CXXFLAGS="$CXXFLAGS $switch"
-        ac_success=yes
-        break
-      fi
-    done
-  fi
-  ac_ext=cpp
-ac_cpp='$CXXCPP $CPPFLAGS'
-ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
-
-  if test x$ax_cxx_compile_cxx11_required = xtrue; then
-    if test x$ac_success = xno; then
-      as_fn_error $? "*** A compiler with support for C++11 language features is required." "$LINENO" 5
-    fi
-  else
-    if test x$ac_success = xno; then
-      HAVE_CXX11=0
-      { $as_echo "$as_me:${as_lineno-$LINENO}: No compiler with C++11 support was found" >&5
-$as_echo "$as_me: No compiler with C++11 support was found" >&6;}
-    else
-      HAVE_CXX11=1
-
-$as_echo "#define HAVE_CXX11 1" >>confdefs.h
-
-    fi
-
-
-  fi
-
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-if test -n "$ac_tool_prefix"; then
-  # Extract the first word of "${ac_tool_prefix}gcc", so it can be a program name with args.
-set dummy ${ac_tool_prefix}gcc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$CC"; then
-  ac_cv_prog_CC="$CC" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_CC="${ac_tool_prefix}gcc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-CC=$ac_cv_prog_CC
-if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
-$as_echo "$CC" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-
-fi
-if test -z "$ac_cv_prog_CC"; then
-  ac_ct_CC=$CC
-  # Extract the first word of "gcc", so it can be a program name with args.
-set dummy gcc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_CC+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$ac_ct_CC"; then
-  ac_cv_prog_ac_ct_CC="$ac_ct_CC" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_ac_ct_CC="gcc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-ac_ct_CC=$ac_cv_prog_ac_ct_CC
-if test -n "$ac_ct_CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
-$as_echo "$ac_ct_CC" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-  if test "x$ac_ct_CC" = x; then
-    CC=""
-  else
-    case $cross_compiling:$ac_tool_warned in
-yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
-$as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
-ac_tool_warned=yes ;;
-esac
-    CC=$ac_ct_CC
-  fi
-else
-  CC="$ac_cv_prog_CC"
-fi
-
-if test -z "$CC"; then
-          if test -n "$ac_tool_prefix"; then
-    # Extract the first word of "${ac_tool_prefix}cc", so it can be a program name with args.
-set dummy ${ac_tool_prefix}cc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$CC"; then
-  ac_cv_prog_CC="$CC" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_CC="${ac_tool_prefix}cc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-CC=$ac_cv_prog_CC
-if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
-$as_echo "$CC" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-
-  fi
-fi
-if test -z "$CC"; then
-  # Extract the first word of "cc", so it can be a program name with args.
-set dummy cc; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$CC"; then
-  ac_cv_prog_CC="$CC" # Let the user override the test.
-else
-  ac_prog_rejected=no
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    if test "$as_dir/$ac_word$ac_exec_ext" = "/usr/ucb/cc"; then
-       ac_prog_rejected=yes
-       continue
-     fi
-    ac_cv_prog_CC="cc"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-if test $ac_prog_rejected = yes; then
-  # We found a bogon in the path, so make sure we never use it.
-  set dummy $ac_cv_prog_CC
-  shift
-  if test $# != 0; then
-    # We chose a different compiler from the bogus one.
-    # However, it has the same basename, so the bogon will be chosen
-    # first if we set CC to just the basename; use the full file name.
-    shift
-    ac_cv_prog_CC="$as_dir/$ac_word${1+' '}$@"
-  fi
-fi
-fi
-fi
-CC=$ac_cv_prog_CC
-if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
-$as_echo "$CC" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-
-fi
-if test -z "$CC"; then
-  if test -n "$ac_tool_prefix"; then
-  for ac_prog in cl.exe
-  do
-    # Extract the first word of "$ac_tool_prefix$ac_prog", so it can be a program name with args.
-set dummy $ac_tool_prefix$ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_CC+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$CC"; then
-  ac_cv_prog_CC="$CC" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_CC="$ac_tool_prefix$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-CC=$ac_cv_prog_CC
-if test -n "$CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $CC" >&5
-$as_echo "$CC" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-
-    test -n "$CC" && break
-  done
-fi
-if test -z "$CC"; then
-  ac_ct_CC=$CC
-  for ac_prog in cl.exe
-do
-  # Extract the first word of "$ac_prog", so it can be a program name with args.
-set dummy $ac_prog; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_ac_ct_CC+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$ac_ct_CC"; then
-  ac_cv_prog_ac_ct_CC="$ac_ct_CC" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_ac_ct_CC="$ac_prog"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-ac_ct_CC=$ac_cv_prog_ac_ct_CC
-if test -n "$ac_ct_CC"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_ct_CC" >&5
-$as_echo "$ac_ct_CC" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-
-  test -n "$ac_ct_CC" && break
-done
-
-  if test "x$ac_ct_CC" = x; then
-    CC=""
-  else
-    case $cross_compiling:$ac_tool_warned in
-yes:)
-{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: using cross tools not prefixed with host triplet" >&5
-$as_echo "$as_me: WARNING: using cross tools not prefixed with host triplet" >&2;}
-ac_tool_warned=yes ;;
-esac
-    CC=$ac_ct_CC
-  fi
-fi
-
-fi
-
-
-test -z "$CC" && { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "no acceptable C compiler found in \$PATH
-See \`config.log' for more details" "$LINENO" 5; }
-
-# Provide some information about the compiler.
-$as_echo "$as_me:${as_lineno-$LINENO}: checking for C compiler version" >&5
-set X $ac_compile
-ac_compiler=$2
-for ac_option in --version -v -V -qversion; do
-  { { ac_try="$ac_compiler $ac_option >&5"
-case "(($ac_try" in
-  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
-  *) ac_try_echo=$ac_try;;
-esac
-eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
-$as_echo "$ac_try_echo"; } >&5
-  (eval "$ac_compiler $ac_option >&5") 2>conftest.err
-  ac_status=$?
-  if test -s conftest.err; then
-    sed '10a\
-... rest of stderr output deleted ...
-         10q' conftest.err >conftest.er1
-    cat conftest.er1 >&5
-  fi
-  rm -f conftest.er1 conftest.err
-  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
-  test $ac_status = 0; }
-done
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether we are using the GNU C compiler" >&5
-$as_echo_n "checking whether we are using the GNU C compiler... " >&6; }
-if ${ac_cv_c_compiler_gnu+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-#ifndef __GNUC__
-       choke me
-#endif
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  ac_compiler_gnu=yes
-else
-  ac_compiler_gnu=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-ac_cv_c_compiler_gnu=$ac_compiler_gnu
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_c_compiler_gnu" >&5
-$as_echo "$ac_cv_c_compiler_gnu" >&6; }
-if test $ac_compiler_gnu = yes; then
-  GCC=yes
-else
-  GCC=
-fi
-ac_test_CFLAGS=${CFLAGS+set}
-ac_save_CFLAGS=$CFLAGS
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC accepts -g" >&5
-$as_echo_n "checking whether $CC accepts -g... " >&6; }
-if ${ac_cv_prog_cc_g+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  ac_save_c_werror_flag=$ac_c_werror_flag
-   ac_c_werror_flag=yes
-   ac_cv_prog_cc_g=no
-   CFLAGS="-g"
-   cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  ac_cv_prog_cc_g=yes
-else
-  CFLAGS=""
-      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-
-else
-  ac_c_werror_flag=$ac_save_c_werror_flag
-	 CFLAGS="-g"
-	 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"; then :
-  ac_cv_prog_cc_g=yes
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-   ac_c_werror_flag=$ac_save_c_werror_flag
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_g" >&5
-$as_echo "$ac_cv_prog_cc_g" >&6; }
-if test "$ac_test_CFLAGS" = set; then
-  CFLAGS=$ac_save_CFLAGS
-elif test $ac_cv_prog_cc_g = yes; then
-  if test "$GCC" = yes; then
-    CFLAGS="-g -O2"
-  else
-    CFLAGS="-g"
-  fi
-else
-  if test "$GCC" = yes; then
-    CFLAGS="-O2"
-  else
-    CFLAGS=
-  fi
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $CC option to accept ISO C89" >&5
-$as_echo_n "checking for $CC option to accept ISO C89... " >&6; }
-if ${ac_cv_prog_cc_c89+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  ac_cv_prog_cc_c89=no
-ac_save_CC=$CC
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <stdarg.h>
-#include <stdio.h>
-struct stat;
-/* Most of the following tests are stolen from RCS 5.7's src/conf.sh.  */
-struct buf { int x; };
-FILE * (*rcsopen) (struct buf *, struct stat *, int);
-static char *e (p, i)
-     char **p;
-     int i;
-{
-  return p[i];
-}
-static char *f (char * (*g) (char **, int), char **p, ...)
-{
-  char *s;
-  va_list v;
-  va_start (v,p);
-  s = g (p, va_arg (v,int));
-  va_end (v);
-  return s;
-}
-
-/* OSF 4.0 Compaq cc is some sort of almost-ANSI by default.  It has
-   function prototypes and stuff, but not '\xHH' hex character constants.
-   These don't provoke an error unfortunately, instead are silently treated
-   as 'x'.  The following induces an error, until -std is added to get
-   proper ANSI mode.  Curiously '\x00'!='x' always comes out true, for an
-   array size at least.  It's necessary to write '\x00'==0 to get something
-   that's true only with -std.  */
-int osf4_cc_array ['\x00' == 0 ? 1 : -1];
-
-/* IBM C 6 for AIX is almost-ANSI by default, but it replaces macro parameters
-   inside strings and character constants.  */
-#define FOO(x) 'x'
-int xlc6_cc_array[FOO(a) == 'x' ? 1 : -1];
-
-int test (int i, double x);
-struct s1 {int (*f) (int a);};
-struct s2 {int (*f) (double a);};
-int pairnames (int, char **, FILE *(*)(struct buf *, struct stat *, int), int, int);
-int argc;
-char **argv;
-int
-main ()
-{
-return f (e, argv, 0) != argv[0]  ||  f (e, argv, 1) != argv[1];
-  ;
-  return 0;
-}
-_ACEOF
-for ac_arg in '' -qlanglvl=extc89 -qlanglvl=ansi -std \
-	-Ae "-Aa -D_HPUX_SOURCE" "-Xc -D__EXTENSIONS__"
-do
-  CC="$ac_save_CC $ac_arg"
-  if ac_fn_c_try_compile "$LINENO"; then :
-  ac_cv_prog_cc_c89=$ac_arg
-fi
-rm -f core conftest.err conftest.$ac_objext
-  test "x$ac_cv_prog_cc_c89" != "xno" && break
-done
-rm -f conftest.$ac_ext
-CC=$ac_save_CC
-
-fi
-# AC_CACHE_VAL
-case "x$ac_cv_prog_cc_c89" in
-  x)
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: none needed" >&5
-$as_echo "none needed" >&6; } ;;
-  xno)
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: unsupported" >&5
-$as_echo "unsupported" >&6; } ;;
-  *)
-    CC="$CC $ac_cv_prog_cc_c89"
-    { $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_prog_cc_c89" >&5
-$as_echo "$ac_cv_prog_cc_c89" >&6; } ;;
-esac
-if test "x$ac_cv_prog_cc_c89" != xno; then :
-
-fi
-
-ac_ext=cpp
-ac_cpp='$CXXCPP $CPPFLAGS'
-ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
-
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether $CC understands -c and -o together" >&5
-$as_echo_n "checking whether $CC understands -c and -o together... " >&6; }
-if ${am_cv_prog_cc_c_o+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-  # Make sure it works both with $CC and with simple cc.
-  # Following AC_PROG_CC_C_O, we do the test twice because some
-  # compilers refuse to overwrite an existing .o file with -o,
-  # though they will create one.
-  am_cv_prog_cc_c_o=yes
-  for am_i in 1 2; do
-    if { echo "$as_me:$LINENO: $CC -c conftest.$ac_ext -o conftest2.$ac_objext" >&5
-   ($CC -c conftest.$ac_ext -o conftest2.$ac_objext) >&5 2>&5
-   ac_status=$?
-   echo "$as_me:$LINENO: \$? = $ac_status" >&5
-   (exit $ac_status); } \
-         && test -f conftest2.$ac_objext; then
-      : OK
-    else
-      am_cv_prog_cc_c_o=no
-      break
-    fi
-  done
-  rm -f core conftest*
-  unset am_i
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_prog_cc_c_o" >&5
-$as_echo "$am_cv_prog_cc_c_o" >&6; }
-if test "$am_cv_prog_cc_c_o" != yes; then
-   # Losing compiler, so override with the script.
-   # FIXME: It is wrong to rewrite CC.
-   # But if we don't then we get into trouble of one sort or another.
-   # A longer-term fix would be to have automake use am__CC in this case,
-   # and then we could set am__CC="\$(top_srcdir)/compile \$(CC)"
-   CC="$am_aux_dir/compile $CC"
-fi
-ac_ext=cpp
-ac_cpp='$CXXCPP $CPPFLAGS'
-ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
-
-
-depcc="$CC"   am_compiler_list=
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking dependency style of $depcc" >&5
-$as_echo_n "checking dependency style of $depcc... " >&6; }
-if ${am_cv_CC_dependencies_compiler_type+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -z "$AMDEP_TRUE" && test -f "$am_depcomp"; then
-  # We make a subdir and do the tests there.  Otherwise we can end up
-  # making bogus files that we don't know about and never remove.  For
-  # instance it was reported that on HP-UX the gcc test will end up
-  # making a dummy file named 'D' -- because '-MD' means "put the output
-  # in D".
-  rm -rf conftest.dir
-  mkdir conftest.dir
-  # Copy depcomp to subdir because otherwise we won't find it if we're
-  # using a relative directory.
-  cp "$am_depcomp" conftest.dir
-  cd conftest.dir
-  # We will build objects and dependencies in a subdirectory because
-  # it helps to detect inapplicable dependency modes.  For instance
-  # both Tru64's cc and ICC support -MD to output dependencies as a
-  # side effect of compilation, but ICC will put the dependencies in
-  # the current directory while Tru64 will put them in the object
-  # directory.
-  mkdir sub
-
-  am_cv_CC_dependencies_compiler_type=none
-  if test "$am_compiler_list" = ""; then
-     am_compiler_list=`sed -n 's/^#*\([a-zA-Z0-9]*\))$/\1/p' < ./depcomp`
-  fi
-  am__universal=false
-  case " $depcc " in #(
-     *\ -arch\ *\ -arch\ *) am__universal=true ;;
-     esac
-
-  for depmode in $am_compiler_list; do
-    # Setup a source with many dependencies, because some compilers
-    # like to wrap large dependency lists on column 80 (with \), and
-    # we should not choose a depcomp mode which is confused by this.
-    #
-    # We need to recreate these files for each test, as the compiler may
-    # overwrite some of them when testing with obscure command lines.
-    # This happens at least with the AIX C compiler.
-    : > sub/conftest.c
-    for i in 1 2 3 4 5 6; do
-      echo '#include "conftst'$i'.h"' >> sub/conftest.c
-      # Using ": > sub/conftst$i.h" creates only sub/conftst1.h with
-      # Solaris 10 /bin/sh.
-      echo '/* dummy */' > sub/conftst$i.h
-    done
-    echo "${am__include} ${am__quote}sub/conftest.Po${am__quote}" > confmf
-
-    # We check with '-c' and '-o' for the sake of the "dashmstdout"
-    # mode.  It turns out that the SunPro C++ compiler does not properly
-    # handle '-M -o', and we need to detect this.  Also, some Intel
-    # versions had trouble with output in subdirs.
-    am__obj=sub/conftest.${OBJEXT-o}
-    am__minus_obj="-o $am__obj"
-    case $depmode in
-    gcc)
-      # This depmode causes a compiler race in universal mode.
-      test "$am__universal" = false || continue
-      ;;
-    nosideeffect)
-      # After this tag, mechanisms are not by side-effect, so they'll
-      # only be used when explicitly requested.
-      if test "x$enable_dependency_tracking" = xyes; then
-	continue
-      else
-	break
-      fi
-      ;;
-    msvc7 | msvc7msys | msvisualcpp | msvcmsys)
-      # This compiler won't grok '-c -o', but also, the minuso test has
-      # not run yet.  These depmodes are late enough in the game, and
-      # so weak that their functioning should not be impacted.
-      am__obj=conftest.${OBJEXT-o}
-      am__minus_obj=
-      ;;
-    none) break ;;
-    esac
-    if depmode=$depmode \
-       source=sub/conftest.c object=$am__obj \
-       depfile=sub/conftest.Po tmpdepfile=sub/conftest.TPo \
-       $SHELL ./depcomp $depcc -c $am__minus_obj sub/conftest.c \
-         >/dev/null 2>conftest.err &&
-       grep sub/conftst1.h sub/conftest.Po > /dev/null 2>&1 &&
-       grep sub/conftst6.h sub/conftest.Po > /dev/null 2>&1 &&
-       grep $am__obj sub/conftest.Po > /dev/null 2>&1 &&
-       ${MAKE-make} -s -f confmf > /dev/null 2>&1; then
-      # icc doesn't choke on unknown options, it will just issue warnings
-      # or remarks (even with -Werror).  So we grep stderr for any message
-      # that says an option was ignored or not supported.
-      # When given -MP, icc 7.0 and 7.1 complain thusly:
-      #   icc: Command line warning: ignoring option '-M'; no argument required
-      # The diagnosis changed in icc 8.0:
-      #   icc: Command line remark: option '-MP' not supported
-      if (grep 'ignoring option' conftest.err ||
-          grep 'not supported' conftest.err) >/dev/null 2>&1; then :; else
-        am_cv_CC_dependencies_compiler_type=$depmode
-        break
-      fi
-    fi
-  done
-
-  cd ..
-  rm -rf conftest.dir
-else
-  am_cv_CC_dependencies_compiler_type=none
-fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $am_cv_CC_dependencies_compiler_type" >&5
-$as_echo "$am_cv_CC_dependencies_compiler_type" >&6; }
-CCDEPMODE=depmode=$am_cv_CC_dependencies_compiler_type
-
- if
-  test "x$enable_dependency_tracking" != xno \
-  && test "$am_cv_CC_dependencies_compiler_type" = gcc3; then
-  am__fastdepCC_TRUE=
-  am__fastdepCC_FALSE='#'
-else
-  am__fastdepCC_TRUE='#'
-  am__fastdepCC_FALSE=
-fi
-
-
-
-
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for x86 cpuid  output" >&5
-$as_echo_n "checking for x86 cpuid  output... " >&6; }
-if ${ax_cv_gcc_x86_cpuid_+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test "$cross_compiling" = yes; then :
-  ax_cv_gcc_x86_cpuid_=unknown
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <stdio.h>
-int
-main ()
-{
-
-     int op = , eax, ebx, ecx, edx;
-     FILE *f;
-      __asm__("cpuid"
-        : "=a" (eax), "=b" (ebx), "=c" (ecx), "=d" (edx)
-        : "a" (op));
-     f = fopen("conftest_cpuid", "w"); if (!f) return 1;
-     fprintf(f, "%x:%x:%x:%x\n", eax, ebx, ecx, edx);
-     fclose(f);
-     return 0;
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_run "$LINENO"; then :
-  ax_cv_gcc_x86_cpuid_=`cat conftest_cpuid`; rm -f conftest_cpuid
-else
-  ax_cv_gcc_x86_cpuid_=unknown; rm -f conftest_cpuid
-fi
-rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
-  conftest.$ac_objext conftest.beam conftest.$ac_ext
-fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_gcc_x86_cpuid_" >&5
-$as_echo "$ax_cv_gcc_x86_cpuid_" >&6; }
-ac_ext=cpp
-ac_cpp='$CXXCPP $CPPFLAGS'
-ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
-
-
-
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for x86-AVX xgetbv  output" >&5
-$as_echo_n "checking for x86-AVX xgetbv  output... " >&6; }
-if ${ax_cv_gcc_x86_avx_xgetbv_+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test "$cross_compiling" = yes; then :
-  ax_cv_gcc_x86_avx_xgetbv_=unknown
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <stdio.h>
-int
-main ()
-{
-
-     int op = , eax, edx;
-     FILE *f;
-      /* Opcodes for xgetbv */
-      __asm__(".byte 0x0f, 0x01, 0xd0"
-        : "=a" (eax), "=d" (edx),
-        : "c" (op));
-     f = fopen("conftest_xgetbv", "w"); if (!f) return 1;
-     fprintf(f, "%x:%x\n", eax, edx);
-     fclose(f);
-     return 0;
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_run "$LINENO"; then :
-  ax_cv_gcc_x86_avx_xgetbv_=`cat conftest_xgetbv`; rm -f conftest_xgetbv
-else
-  ax_cv_gcc_x86_avx_xgetbv_=unknown; rm -f conftest_xgetbv
-fi
-rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
-  conftest.$ac_objext conftest.beam conftest.$ac_ext
-fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_gcc_x86_avx_xgetbv_" >&5
-$as_echo "$ax_cv_gcc_x86_avx_xgetbv_" >&6; }
-ac_ext=cpp
-ac_cpp='$CXXCPP $CPPFLAGS'
-ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
-
-
-
-
-
-  case $host_cpu in
-    powerpc*)
-      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether altivec is supported" >&5
-$as_echo_n "checking whether altivec is supported... " >&6; }
-if ${ax_cv_have_altivec_ext+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-            if test `/usr/sbin/sysctl -a 2>/dev/null| grep -c hw.optional.altivec` != 0; then
-                if test `/usr/sbin/sysctl -n hw.optional.altivec` = 1; then
-                  ax_cv_have_altivec_ext=yes
-                fi
-            fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_altivec_ext" >&5
-$as_echo "$ax_cv_have_altivec_ext" >&6; }
-
-          if test "$ax_cv_have_altivec_ext" = yes; then
-
-$as_echo "#define HAVE_ALTIVEC /**/" >>confdefs.h
-
-            { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -faltivec" >&5
-$as_echo_n "checking whether C++ compiler accepts -faltivec... " >&6; }
-if ${ax_cv_check_cxxflags___faltivec+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-  ax_check_save_flags=$CXXFLAGS
-  CXXFLAGS="$CXXFLAGS  -faltivec"
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  ax_cv_check_cxxflags___faltivec=yes
-else
-  ax_cv_check_cxxflags___faltivec=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-  CXXFLAGS=$ax_check_save_flags
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___faltivec" >&5
-$as_echo "$ax_cv_check_cxxflags___faltivec" >&6; }
-if test x"$ax_cv_check_cxxflags___faltivec" = xyes; then :
-  SIMD_FLAGS="$SIMD_FLAGS -faltivec"
-else
-  :
-fi
-
-          fi
-    ;;
-
-
-    i[3456]86*|x86_64*|amd64*)
-
-
-
-
-
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for x86 cpuid 0x00000001 output" >&5
-$as_echo_n "checking for x86 cpuid 0x00000001 output... " >&6; }
-if ${ax_cv_gcc_x86_cpuid_0x00000001+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test "$cross_compiling" = yes; then :
-  ax_cv_gcc_x86_cpuid_0x00000001=unknown
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <stdio.h>
-int
-main ()
-{
-
-     int op = 0x00000001, eax, ebx, ecx, edx;
-     FILE *f;
-      __asm__("cpuid"
-        : "=a" (eax), "=b" (ebx), "=c" (ecx), "=d" (edx)
-        : "a" (op));
-     f = fopen("conftest_cpuid", "w"); if (!f) return 1;
-     fprintf(f, "%x:%x:%x:%x\n", eax, ebx, ecx, edx);
-     fclose(f);
-     return 0;
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_run "$LINENO"; then :
-  ax_cv_gcc_x86_cpuid_0x00000001=`cat conftest_cpuid`; rm -f conftest_cpuid
-else
-  ax_cv_gcc_x86_cpuid_0x00000001=unknown; rm -f conftest_cpuid
-fi
-rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
-  conftest.$ac_objext conftest.beam conftest.$ac_ext
-fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_gcc_x86_cpuid_0x00000001" >&5
-$as_echo "$ax_cv_gcc_x86_cpuid_0x00000001" >&6; }
-ac_ext=cpp
-ac_cpp='$CXXCPP $CPPFLAGS'
-ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
-
-
-      ecx=0
-      edx=0
-      ebx=0
-      if test "$ax_cv_gcc_x86_cpuid_0x00000001" != "unknown";
-      then
-        ecx=`echo $ax_cv_gcc_x86_cpuid_0x00000001 | cut -d ":" -f 3`
-        edx=`echo $ax_cv_gcc_x86_cpuid_0x00000001 | cut -d ":" -f 4`
-      fi
-
-
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for x86 cpuid 0x00000007 output" >&5
-$as_echo_n "checking for x86 cpuid 0x00000007 output... " >&6; }
-if ${ax_cv_gcc_x86_cpuid_0x00000007+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test "$cross_compiling" = yes; then :
-  ax_cv_gcc_x86_cpuid_0x00000007=unknown
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <stdio.h>
-int
-main ()
-{
-
-     int op = 0x00000007, eax, ebx, ecx, edx;
-     FILE *f;
-      __asm__("cpuid"
-        : "=a" (eax), "=b" (ebx), "=c" (ecx), "=d" (edx)
-        : "a" (op));
-     f = fopen("conftest_cpuid", "w"); if (!f) return 1;
-     fprintf(f, "%x:%x:%x:%x\n", eax, ebx, ecx, edx);
-     fclose(f);
-     return 0;
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_run "$LINENO"; then :
-  ax_cv_gcc_x86_cpuid_0x00000007=`cat conftest_cpuid`; rm -f conftest_cpuid
-else
-  ax_cv_gcc_x86_cpuid_0x00000007=unknown; rm -f conftest_cpuid
-fi
-rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
-  conftest.$ac_objext conftest.beam conftest.$ac_ext
-fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_gcc_x86_cpuid_0x00000007" >&5
-$as_echo "$ax_cv_gcc_x86_cpuid_0x00000007" >&6; }
-ac_ext=cpp
-ac_cpp='$CXXCPP $CPPFLAGS'
-ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
-
-
-      if test "$ax_cv_gcc_x86_cpuid_0x00000007" != "unknown";
-      then
-        ebx=`echo $ax_cv_gcc_x86_cpuid_0x00000007 | cut -d ":" -f 2`
-      fi
-
-      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether mmx is supported" >&5
-$as_echo_n "checking whether mmx is supported... " >&6; }
-if ${ax_cv_have_mmx_ext+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-        ax_cv_have_mmx_ext=no
-        if test "$((0x$edx>>23&0x01))" = 1; then
-          ax_cv_have_mmx_ext=yes
-        fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_mmx_ext" >&5
-$as_echo "$ax_cv_have_mmx_ext" >&6; }
-
-      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether sse is supported" >&5
-$as_echo_n "checking whether sse is supported... " >&6; }
-if ${ax_cv_have_sse_ext+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-        ax_cv_have_sse_ext=no
-        if test "$((0x$edx>>25&0x01))" = 1; then
-          ax_cv_have_sse_ext=yes
-        fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_sse_ext" >&5
-$as_echo "$ax_cv_have_sse_ext" >&6; }
-
-      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether sse2 is supported" >&5
-$as_echo_n "checking whether sse2 is supported... " >&6; }
-if ${ax_cv_have_sse2_ext+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-        ax_cv_have_sse2_ext=no
-        if test "$((0x$edx>>26&0x01))" = 1; then
-          ax_cv_have_sse2_ext=yes
-        fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_sse2_ext" >&5
-$as_echo "$ax_cv_have_sse2_ext" >&6; }
-
-      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether sse3 is supported" >&5
-$as_echo_n "checking whether sse3 is supported... " >&6; }
-if ${ax_cv_have_sse3_ext+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-        ax_cv_have_sse3_ext=no
-        if test "$((0x$ecx&0x01))" = 1; then
-          ax_cv_have_sse3_ext=yes
-        fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_sse3_ext" >&5
-$as_echo "$ax_cv_have_sse3_ext" >&6; }
-
-      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether ssse3 is supported" >&5
-$as_echo_n "checking whether ssse3 is supported... " >&6; }
-if ${ax_cv_have_ssse3_ext+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-        ax_cv_have_ssse3_ext=no
-        if test "$((0x$ecx>>9&0x01))" = 1; then
-          ax_cv_have_ssse3_ext=yes
-        fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_ssse3_ext" >&5
-$as_echo "$ax_cv_have_ssse3_ext" >&6; }
-
-      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether sse4.1 is supported" >&5
-$as_echo_n "checking whether sse4.1 is supported... " >&6; }
-if ${ax_cv_have_sse41_ext+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-        ax_cv_have_sse41_ext=no
-        if test "$((0x$ecx>>19&0x01))" = 1; then
-          ax_cv_have_sse41_ext=yes
-        fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_sse41_ext" >&5
-$as_echo "$ax_cv_have_sse41_ext" >&6; }
-
-      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether sse4.2 is supported" >&5
-$as_echo_n "checking whether sse4.2 is supported... " >&6; }
-if ${ax_cv_have_sse42_ext+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-        ax_cv_have_sse42_ext=no
-        if test "$((0x$ecx>>20&0x01))" = 1; then
-          ax_cv_have_sse42_ext=yes
-        fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_sse42_ext" >&5
-$as_echo "$ax_cv_have_sse42_ext" >&6; }
-
-      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether avx is supported by processor" >&5
-$as_echo_n "checking whether avx is supported by processor... " >&6; }
-if ${ax_cv_have_avx_cpu_ext+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-        ax_cv_have_avx_cpu_ext=no
-        if test "$((0x$ecx>>28&0x01))" = 1; then
-          ax_cv_have_avx_cpu_ext=yes
-        fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_avx_cpu_ext" >&5
-$as_echo "$ax_cv_have_avx_cpu_ext" >&6; }
-
-      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether avx2 is supported by processor" >&5
-$as_echo_n "checking whether avx2 is supported by processor... " >&6; }
-if ${ax_cv_have_avx2_cpu_ext+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-        ax_cv_have_avx2_cpu_ext=no
-        if test "$((0x$ebx>>5&0x01))" = 1; then
-          ax_cv_have_avx2_cpu_ext=yes
-        fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_avx2_cpu_ext" >&5
-$as_echo "$ax_cv_have_avx2_cpu_ext" >&6; }
-
-
-      { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether fma is supported by processor" >&5
-$as_echo_n "checking whether fma is supported by processor... " >&6; }
-if ${ax_cv_have_fma_cpu_ext+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-        ax_cv_have_fma_cpu_ext=no
-        if test "$((0x$ecx>>12&0x01))" = 1; then
-          ax_cv_have_fma_cpu_ext=yes
-        fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_fma_cpu_ext" >&5
-$as_echo "$ax_cv_have_fma_cpu_ext" >&6; }
-
-
-      if test x"$ax_cv_have_avx_cpu_ext" = x"yes"; then
-
-ac_ext=c
-ac_cpp='$CPP $CPPFLAGS'
-ac_compile='$CC -c $CFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CC -o conftest$ac_exeext $CFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_c_compiler_gnu
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for x86-AVX xgetbv 0x00000000 output" >&5
-$as_echo_n "checking for x86-AVX xgetbv 0x00000000 output... " >&6; }
-if ${ax_cv_gcc_x86_avx_xgetbv_0x00000000+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test "$cross_compiling" = yes; then :
-  ax_cv_gcc_x86_avx_xgetbv_0x00000000=unknown
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <stdio.h>
-int
-main ()
-{
-
-     int op = 0x00000000, eax, edx;
-     FILE *f;
-      /* Opcodes for xgetbv */
-      __asm__(".byte 0x0f, 0x01, 0xd0"
-        : "=a" (eax), "=d" (edx),
-        : "c" (op));
-     f = fopen("conftest_xgetbv", "w"); if (!f) return 1;
-     fprintf(f, "%x:%x\n", eax, edx);
-     fclose(f);
-     return 0;
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_run "$LINENO"; then :
-  ax_cv_gcc_x86_avx_xgetbv_0x00000000=`cat conftest_xgetbv`; rm -f conftest_xgetbv
-else
-  ax_cv_gcc_x86_avx_xgetbv_0x00000000=unknown; rm -f conftest_xgetbv
-fi
-rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
-  conftest.$ac_objext conftest.beam conftest.$ac_ext
-fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_gcc_x86_avx_xgetbv_0x00000000" >&5
-$as_echo "$ax_cv_gcc_x86_avx_xgetbv_0x00000000" >&6; }
-ac_ext=cpp
-ac_cpp='$CXXCPP $CPPFLAGS'
-ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
-
-
-
-        xgetbv_eax="0"
-        if test x"$ax_cv_gcc_x86_avx_xgetbv_0x00000000" != x"unknown"; then
-          xgetbv_eax=`echo $ax_cv_gcc_x86_avx_xgetbv_0x00000000 | cut -d ":" -f 1`
-        fi
-
-        { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether avx is supported by operating system" >&5
-$as_echo_n "checking whether avx is supported by operating system... " >&6; }
-if ${ax_cv_have_avx_ext+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-          ax_cv_have_avx_ext=no
-
-          if test "$((0x$ecx>>27&0x01))" = 1; then
-            if test "$((0x$xgetbv_eax&0x6))" = 6; then
-              ax_cv_have_avx_ext=yes
-            fi
-          fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_have_avx_ext" >&5
-$as_echo "$ax_cv_have_avx_ext" >&6; }
-        if test x"$ax_cv_have_avx_ext" = x"no"; then
-          { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports AVX, but your operating system doesn't" >&5
-$as_echo "$as_me: WARNING: Your processor supports AVX, but your operating system doesn't" >&2;}
-        fi
-      fi
-
-      if test "$ax_cv_have_mmx_ext" = yes; then
-        { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -mmmx" >&5
-$as_echo_n "checking whether C++ compiler accepts -mmmx... " >&6; }
-if ${ax_cv_check_cxxflags___mmmx+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-  ax_check_save_flags=$CXXFLAGS
-  CXXFLAGS="$CXXFLAGS  -mmmx"
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  ax_cv_check_cxxflags___mmmx=yes
-else
-  ax_cv_check_cxxflags___mmmx=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-  CXXFLAGS=$ax_check_save_flags
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___mmmx" >&5
-$as_echo "$ax_cv_check_cxxflags___mmmx" >&6; }
-if test x"$ax_cv_check_cxxflags___mmmx" = xyes; then :
-  ax_cv_support_mmx_ext=yes
-else
-  :
-fi
-
-        if test x"$ax_cv_support_mmx_ext" = x"yes"; then
-          SIMD_FLAGS="$SIMD_FLAGS -mmmx"
-
-$as_echo "#define HAVE_MMX /**/" >>confdefs.h
-
-        else
-          { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports mmx instructions but not your compiler, can you try another compiler?" >&5
-$as_echo "$as_me: WARNING: Your processor supports mmx instructions but not your compiler, can you try another compiler?" >&2;}
-        fi
-      fi
-
-      if test "$ax_cv_have_sse_ext" = yes; then
-        { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -msse" >&5
-$as_echo_n "checking whether C++ compiler accepts -msse... " >&6; }
-if ${ax_cv_check_cxxflags___msse+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-  ax_check_save_flags=$CXXFLAGS
-  CXXFLAGS="$CXXFLAGS  -msse"
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  ax_cv_check_cxxflags___msse=yes
-else
-  ax_cv_check_cxxflags___msse=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-  CXXFLAGS=$ax_check_save_flags
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___msse" >&5
-$as_echo "$ax_cv_check_cxxflags___msse" >&6; }
-if test x"$ax_cv_check_cxxflags___msse" = xyes; then :
-  ax_cv_support_sse_ext=yes
-else
-  :
-fi
-
-        if test x"$ax_cv_support_sse_ext" = x"yes"; then
-          SIMD_FLAGS="$SIMD_FLAGS -msse"
-
-$as_echo "#define HAVE_SSE /**/" >>confdefs.h
-
-        else
-          { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports sse instructions but not your compiler, can you try another compiler?" >&5
-$as_echo "$as_me: WARNING: Your processor supports sse instructions but not your compiler, can you try another compiler?" >&2;}
-        fi
-      fi
-
-      if test "$ax_cv_have_sse2_ext" = yes; then
-        { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -msse2" >&5
-$as_echo_n "checking whether C++ compiler accepts -msse2... " >&6; }
-if ${ax_cv_check_cxxflags___msse2+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-  ax_check_save_flags=$CXXFLAGS
-  CXXFLAGS="$CXXFLAGS  -msse2"
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  ax_cv_check_cxxflags___msse2=yes
-else
-  ax_cv_check_cxxflags___msse2=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-  CXXFLAGS=$ax_check_save_flags
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___msse2" >&5
-$as_echo "$ax_cv_check_cxxflags___msse2" >&6; }
-if test x"$ax_cv_check_cxxflags___msse2" = xyes; then :
-  ax_cv_support_sse2_ext=yes
-else
-  :
-fi
-
-        if test x"$ax_cv_support_sse2_ext" = x"yes"; then
-          SIMD_FLAGS="$SIMD_FLAGS -msse2"
-
-$as_echo "#define HAVE_SSE2 /**/" >>confdefs.h
-
-        else
-          { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports sse2 instructions but not your compiler, can you try another compiler?" >&5
-$as_echo "$as_me: WARNING: Your processor supports sse2 instructions but not your compiler, can you try another compiler?" >&2;}
-        fi
-      fi
-
-      if test "$ax_cv_have_sse3_ext" = yes; then
-        { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -msse3" >&5
-$as_echo_n "checking whether C++ compiler accepts -msse3... " >&6; }
-if ${ax_cv_check_cxxflags___msse3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-  ax_check_save_flags=$CXXFLAGS
-  CXXFLAGS="$CXXFLAGS  -msse3"
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  ax_cv_check_cxxflags___msse3=yes
-else
-  ax_cv_check_cxxflags___msse3=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-  CXXFLAGS=$ax_check_save_flags
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___msse3" >&5
-$as_echo "$ax_cv_check_cxxflags___msse3" >&6; }
-if test x"$ax_cv_check_cxxflags___msse3" = xyes; then :
-  ax_cv_support_sse3_ext=yes
-else
-  :
-fi
-
-        if test x"$ax_cv_support_sse3_ext" = x"yes"; then
-          SIMD_FLAGS="$SIMD_FLAGS -msse3"
-
-$as_echo "#define HAVE_SSE3 /**/" >>confdefs.h
-
-        else
-          { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports sse3 instructions but not your compiler, can you try another compiler?" >&5
-$as_echo "$as_me: WARNING: Your processor supports sse3 instructions but not your compiler, can you try another compiler?" >&2;}
-        fi
-      fi
-
-      if test "$ax_cv_have_ssse3_ext" = yes; then
-        { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -mssse3" >&5
-$as_echo_n "checking whether C++ compiler accepts -mssse3... " >&6; }
-if ${ax_cv_check_cxxflags___mssse3+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-  ax_check_save_flags=$CXXFLAGS
-  CXXFLAGS="$CXXFLAGS  -mssse3"
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  ax_cv_check_cxxflags___mssse3=yes
-else
-  ax_cv_check_cxxflags___mssse3=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-  CXXFLAGS=$ax_check_save_flags
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___mssse3" >&5
-$as_echo "$ax_cv_check_cxxflags___mssse3" >&6; }
-if test x"$ax_cv_check_cxxflags___mssse3" = xyes; then :
-  ax_cv_support_ssse3_ext=yes
-else
-  :
-fi
-
-        if test x"$ax_cv_support_ssse3_ext" = x"yes"; then
-          SIMD_FLAGS="$SIMD_FLAGS -mssse3"
-
-$as_echo "#define HAVE_SSSE3 /**/" >>confdefs.h
-
-        else
-          { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports ssse3 instructions but not your compiler, can you try another compiler?" >&5
-$as_echo "$as_me: WARNING: Your processor supports ssse3 instructions but not your compiler, can you try another compiler?" >&2;}
-        fi
-      fi
-
-      if test "$ax_cv_have_sse41_ext" = yes; then
-        { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -msse4.1" >&5
-$as_echo_n "checking whether C++ compiler accepts -msse4.1... " >&6; }
-if ${ax_cv_check_cxxflags___msse4_1+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-  ax_check_save_flags=$CXXFLAGS
-  CXXFLAGS="$CXXFLAGS  -msse4.1"
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  ax_cv_check_cxxflags___msse4_1=yes
-else
-  ax_cv_check_cxxflags___msse4_1=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-  CXXFLAGS=$ax_check_save_flags
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___msse4_1" >&5
-$as_echo "$ax_cv_check_cxxflags___msse4_1" >&6; }
-if test x"$ax_cv_check_cxxflags___msse4_1" = xyes; then :
-  ax_cv_support_sse41_ext=yes
-else
-  :
-fi
-
-        if test x"$ax_cv_support_sse41_ext" = x"yes"; then
-          SIMD_FLAGS="$SIMD_FLAGS -msse4.1"
-
-$as_echo "#define HAVE_SSE4_1 /**/" >>confdefs.h
-
-        else
-          { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports sse4.1 instructions but not your compiler, can you try another compiler?" >&5
-$as_echo "$as_me: WARNING: Your processor supports sse4.1 instructions but not your compiler, can you try another compiler?" >&2;}
-        fi
-      fi
-
-      if test "$ax_cv_have_sse42_ext" = yes; then
-        { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -msse4.2" >&5
-$as_echo_n "checking whether C++ compiler accepts -msse4.2... " >&6; }
-if ${ax_cv_check_cxxflags___msse4_2+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-  ax_check_save_flags=$CXXFLAGS
-  CXXFLAGS="$CXXFLAGS  -msse4.2"
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  ax_cv_check_cxxflags___msse4_2=yes
-else
-  ax_cv_check_cxxflags___msse4_2=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-  CXXFLAGS=$ax_check_save_flags
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___msse4_2" >&5
-$as_echo "$ax_cv_check_cxxflags___msse4_2" >&6; }
-if test x"$ax_cv_check_cxxflags___msse4_2" = xyes; then :
-  ax_cv_support_sse42_ext=yes
-else
-  :
-fi
-
-        if test x"$ax_cv_support_sse42_ext" = x"yes"; then
-          SIMD_FLAGS="$SIMD_FLAGS -msse4.2"
-
-$as_echo "#define HAVE_SSE4_2 /**/" >>confdefs.h
-
-        else
-          { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports sse4.2 instructions but not your compiler, can you try another compiler?" >&5
-$as_echo "$as_me: WARNING: Your processor supports sse4.2 instructions but not your compiler, can you try another compiler?" >&2;}
-        fi
-      fi
-
-      if test "$ax_cv_have_avx_ext" = yes; then
-        { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -mavx" >&5
-$as_echo_n "checking whether C++ compiler accepts -mavx... " >&6; }
-if ${ax_cv_check_cxxflags___mavx+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-  ax_check_save_flags=$CXXFLAGS
-  CXXFLAGS="$CXXFLAGS  -mavx"
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  ax_cv_check_cxxflags___mavx=yes
-else
-  ax_cv_check_cxxflags___mavx=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-  CXXFLAGS=$ax_check_save_flags
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___mavx" >&5
-$as_echo "$ax_cv_check_cxxflags___mavx" >&6; }
-if test x"$ax_cv_check_cxxflags___mavx" = xyes; then :
-  ax_cv_support_avx_ext=yes
-else
-  :
-fi
-
-        if test x"$ax_cv_support_avx_ext" = x"yes"; then
-          SIMD_FLAGS="$SIMD_FLAGS -mavx"
-
-$as_echo "#define HAVE_AVX /**/" >>confdefs.h
-
-        else
-          { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports avx instructions but not your compiler, can you try another compiler?" >&5
-$as_echo "$as_me: WARNING: Your processor supports avx instructions but not your compiler, can you try another compiler?" >&2;}
-        fi
-      fi
-
-       if test "$ax_cv_have_avx2_ext" = yes; then
-         { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -mavx2" >&5
-$as_echo_n "checking whether C++ compiler accepts -mavx2... " >&6; }
-if ${ax_cv_check_cxxflags___mavx2+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-  ax_check_save_flags=$CXXFLAGS
-  CXXFLAGS="$CXXFLAGS  -mavx2"
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  ax_cv_check_cxxflags___mavx2=yes
-else
-  ax_cv_check_cxxflags___mavx2=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-  CXXFLAGS=$ax_check_save_flags
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___mavx2" >&5
-$as_echo "$ax_cv_check_cxxflags___mavx2" >&6; }
-if test x"$ax_cv_check_cxxflags___mavx2" = xyes; then :
-  ax_cv_support_avx2_ext=yes
-else
-  :
-fi
-
-         if test x"$ax_cv_support_avx2_ext" = x"yes"; then
-           SIMD_FLAGS="$SIMD_FLAGS -mavx2"
-
-$as_echo "#define HAVE_AVX2 /**/" >>confdefs.h
-
-         else
-           { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports avx2 instructions but not your compiler, can you try another compiler?" >&5
-$as_echo "$as_me: WARNING: Your processor supports avx2 instructions but not your compiler, can you try another compiler?" >&2;}
-         fi
-        fi
-
-       if test "$ax_cv_have_fma_ext" = yes; then
-         { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether C++ compiler accepts -mfma" >&5
-$as_echo_n "checking whether C++ compiler accepts -mfma... " >&6; }
-if ${ax_cv_check_cxxflags___mfma+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-
-  ax_check_save_flags=$CXXFLAGS
-  CXXFLAGS="$CXXFLAGS  -mfma"
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  ax_cv_check_cxxflags___mfma=yes
-else
-  ax_cv_check_cxxflags___mfma=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-  CXXFLAGS=$ax_check_save_flags
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ax_cv_check_cxxflags___mfma" >&5
-$as_echo "$ax_cv_check_cxxflags___mfma" >&6; }
-if test x"$ax_cv_check_cxxflags___mfma" = xyes; then :
-  ax_cv_support_fma_ext=yes
-else
-  :
-fi
-
-         if test x"$ax_cv_support_fma_ext" = x"yes"; then
-           SIMD_FLAGS="$SIMD_FLAGS -mfma"
-
-$as_echo "#define HAVE_FMA /**/" >>confdefs.h
-
-         else
-           { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor supports fma instructions but not your compiler, can you try another compiler?" >&5
-$as_echo "$as_me: WARNING: Your processor supports fma instructions but not your compiler, can you try another compiler?" >&2;}
-         fi
-      fi
-
-  ;;
-  esac
-
-
-
-
-
-# Checks for libraries.
-#AX_GCC_VAR_ATTRIBUTE(aligned)
-
-# Checks for header files.
-ac_ext=cpp
-ac_cpp='$CXXCPP $CPPFLAGS'
-ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking how to run the C++ preprocessor" >&5
-$as_echo_n "checking how to run the C++ preprocessor... " >&6; }
-if test -z "$CXXCPP"; then
-  if ${ac_cv_prog_CXXCPP+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-      # Double quotes because CXXCPP needs to be expanded
-    for CXXCPP in "$CXX -E" "/lib/cpp"
-    do
-      ac_preproc_ok=false
-for ac_cxx_preproc_warn_flag in '' yes
-do
-  # Use a header file that comes with gcc, so configuring glibc
-  # with a fresh cross-compiler works.
-  # Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
-  # <limits.h> exists even on freestanding compilers.
-  # On the NeXT, cc -E runs the code through the compiler's parser,
-  # not just through cpp. "Syntax error" is here to catch this case.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#ifdef __STDC__
-# include <limits.h>
-#else
-# include <assert.h>
-#endif
-		     Syntax error
-_ACEOF
-if ac_fn_cxx_try_cpp "$LINENO"; then :
-
-else
-  # Broken: fails on valid input.
-continue
-fi
-rm -f conftest.err conftest.i conftest.$ac_ext
-
-  # OK, works on sane cases.  Now check whether nonexistent headers
-  # can be detected and how.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <ac_nonexistent.h>
-_ACEOF
-if ac_fn_cxx_try_cpp "$LINENO"; then :
-  # Broken: success on invalid input.
-continue
-else
-  # Passes both tests.
-ac_preproc_ok=:
-break
-fi
-rm -f conftest.err conftest.i conftest.$ac_ext
-
-done
-# Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
-rm -f conftest.i conftest.err conftest.$ac_ext
-if $ac_preproc_ok; then :
-  break
-fi
-
-    done
-    ac_cv_prog_CXXCPP=$CXXCPP
-
-fi
-  CXXCPP=$ac_cv_prog_CXXCPP
-else
-  ac_cv_prog_CXXCPP=$CXXCPP
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $CXXCPP" >&5
-$as_echo "$CXXCPP" >&6; }
-ac_preproc_ok=false
-for ac_cxx_preproc_warn_flag in '' yes
-do
-  # Use a header file that comes with gcc, so configuring glibc
-  # with a fresh cross-compiler works.
-  # Prefer <limits.h> to <assert.h> if __STDC__ is defined, since
-  # <limits.h> exists even on freestanding compilers.
-  # On the NeXT, cc -E runs the code through the compiler's parser,
-  # not just through cpp. "Syntax error" is here to catch this case.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#ifdef __STDC__
-# include <limits.h>
-#else
-# include <assert.h>
-#endif
-		     Syntax error
-_ACEOF
-if ac_fn_cxx_try_cpp "$LINENO"; then :
-
-else
-  # Broken: fails on valid input.
-continue
-fi
-rm -f conftest.err conftest.i conftest.$ac_ext
-
-  # OK, works on sane cases.  Now check whether nonexistent headers
-  # can be detected and how.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <ac_nonexistent.h>
-_ACEOF
-if ac_fn_cxx_try_cpp "$LINENO"; then :
-  # Broken: success on invalid input.
-continue
-else
-  # Passes both tests.
-ac_preproc_ok=:
-break
-fi
-rm -f conftest.err conftest.i conftest.$ac_ext
-
-done
-# Because of `break', _AC_PREPROC_IFELSE's cleaning code was skipped.
-rm -f conftest.i conftest.err conftest.$ac_ext
-if $ac_preproc_ok; then :
-
-else
-  { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
-$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
-as_fn_error $? "C++ preprocessor \"$CXXCPP\" fails sanity check
-See \`config.log' for more details" "$LINENO" 5; }
-fi
-
-ac_ext=cpp
-ac_cpp='$CXXCPP $CPPFLAGS'
-ac_compile='$CXX -c $CXXFLAGS $CPPFLAGS conftest.$ac_ext >&5'
-ac_link='$CXX -o conftest$ac_exeext $CXXFLAGS $CPPFLAGS $LDFLAGS conftest.$ac_ext $LIBS >&5'
-ac_compiler_gnu=$ac_cv_cxx_compiler_gnu
-
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for grep that handles long lines and -e" >&5
-$as_echo_n "checking for grep that handles long lines and -e... " >&6; }
-if ${ac_cv_path_GREP+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -z "$GREP"; then
-  ac_path_GREP_found=false
-  # Loop through the user's path and test for each of PROGNAME-LIST
-  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_prog in grep ggrep; do
-    for ac_exec_ext in '' $ac_executable_extensions; do
-      ac_path_GREP="$as_dir/$ac_prog$ac_exec_ext"
-      as_fn_executable_p "$ac_path_GREP" || continue
-# Check for GNU ac_path_GREP and select it if it is found.
-  # Check for GNU $ac_path_GREP
-case `"$ac_path_GREP" --version 2>&1` in
-*GNU*)
-  ac_cv_path_GREP="$ac_path_GREP" ac_path_GREP_found=:;;
-*)
-  ac_count=0
-  $as_echo_n 0123456789 >"conftest.in"
-  while :
-  do
-    cat "conftest.in" "conftest.in" >"conftest.tmp"
-    mv "conftest.tmp" "conftest.in"
-    cp "conftest.in" "conftest.nl"
-    $as_echo 'GREP' >> "conftest.nl"
-    "$ac_path_GREP" -e 'GREP$' -e '-(cannot match)-' < "conftest.nl" >"conftest.out" 2>/dev/null || break
-    diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
-    as_fn_arith $ac_count + 1 && ac_count=$as_val
-    if test $ac_count -gt ${ac_path_GREP_max-0}; then
-      # Best one so far, save it but keep looking for a better one
-      ac_cv_path_GREP="$ac_path_GREP"
-      ac_path_GREP_max=$ac_count
-    fi
-    # 10*(2^10) chars as input seems more than enough
-    test $ac_count -gt 10 && break
-  done
-  rm -f conftest.in conftest.tmp conftest.nl conftest.out;;
-esac
-
-      $ac_path_GREP_found && break 3
-    done
-  done
-  done
-IFS=$as_save_IFS
-  if test -z "$ac_cv_path_GREP"; then
-    as_fn_error $? "no acceptable grep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
-  fi
-else
-  ac_cv_path_GREP=$GREP
-fi
-
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_GREP" >&5
-$as_echo "$ac_cv_path_GREP" >&6; }
- GREP="$ac_cv_path_GREP"
-
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for egrep" >&5
-$as_echo_n "checking for egrep... " >&6; }
-if ${ac_cv_path_EGREP+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if echo a | $GREP -E '(a|b)' >/dev/null 2>&1
-   then ac_cv_path_EGREP="$GREP -E"
-   else
-     if test -z "$EGREP"; then
-  ac_path_EGREP_found=false
-  # Loop through the user's path and test for each of PROGNAME-LIST
-  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH$PATH_SEPARATOR/usr/xpg4/bin
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_prog in egrep; do
-    for ac_exec_ext in '' $ac_executable_extensions; do
-      ac_path_EGREP="$as_dir/$ac_prog$ac_exec_ext"
-      as_fn_executable_p "$ac_path_EGREP" || continue
-# Check for GNU ac_path_EGREP and select it if it is found.
-  # Check for GNU $ac_path_EGREP
-case `"$ac_path_EGREP" --version 2>&1` in
-*GNU*)
-  ac_cv_path_EGREP="$ac_path_EGREP" ac_path_EGREP_found=:;;
-*)
-  ac_count=0
-  $as_echo_n 0123456789 >"conftest.in"
-  while :
-  do
-    cat "conftest.in" "conftest.in" >"conftest.tmp"
-    mv "conftest.tmp" "conftest.in"
-    cp "conftest.in" "conftest.nl"
-    $as_echo 'EGREP' >> "conftest.nl"
-    "$ac_path_EGREP" 'EGREP$' < "conftest.nl" >"conftest.out" 2>/dev/null || break
-    diff "conftest.out" "conftest.nl" >/dev/null 2>&1 || break
-    as_fn_arith $ac_count + 1 && ac_count=$as_val
-    if test $ac_count -gt ${ac_path_EGREP_max-0}; then
-      # Best one so far, save it but keep looking for a better one
-      ac_cv_path_EGREP="$ac_path_EGREP"
-      ac_path_EGREP_max=$ac_count
-    fi
-    # 10*(2^10) chars as input seems more than enough
-    test $ac_count -gt 10 && break
-  done
-  rm -f conftest.in conftest.tmp conftest.nl conftest.out;;
-esac
-
-      $ac_path_EGREP_found && break 3
-    done
-  done
-  done
-IFS=$as_save_IFS
-  if test -z "$ac_cv_path_EGREP"; then
-    as_fn_error $? "no acceptable egrep could be found in $PATH$PATH_SEPARATOR/usr/xpg4/bin" "$LINENO" 5
-  fi
-else
-  ac_cv_path_EGREP=$EGREP
-fi
-
-   fi
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_path_EGREP" >&5
-$as_echo "$ac_cv_path_EGREP" >&6; }
- EGREP="$ac_cv_path_EGREP"
-
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for ANSI C header files" >&5
-$as_echo_n "checking for ANSI C header files... " >&6; }
-if ${ac_cv_header_stdc+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <stdlib.h>
-#include <stdarg.h>
-#include <string.h>
-#include <float.h>
-
-int
-main ()
-{
-
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_compile "$LINENO"; then :
-  ac_cv_header_stdc=yes
-else
-  ac_cv_header_stdc=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.$ac_ext
-
-if test $ac_cv_header_stdc = yes; then
-  # SunOS 4.x string.h does not declare mem*, contrary to ANSI.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <string.h>
-
-_ACEOF
-if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
-  $EGREP "memchr" >/dev/null 2>&1; then :
-
-else
-  ac_cv_header_stdc=no
-fi
-rm -f conftest*
-
-fi
-
-if test $ac_cv_header_stdc = yes; then
-  # ISC 2.0.2 stdlib.h does not declare free, contrary to ANSI.
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <stdlib.h>
-
-_ACEOF
-if (eval "$ac_cpp conftest.$ac_ext") 2>&5 |
-  $EGREP "free" >/dev/null 2>&1; then :
-
-else
-  ac_cv_header_stdc=no
-fi
-rm -f conftest*
-
-fi
-
-if test $ac_cv_header_stdc = yes; then
-  # /bin/cc in Irix-4.0.5 gets non-ANSI ctype macros unless using -ansi.
-  if test "$cross_compiling" = yes; then :
-  :
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <ctype.h>
-#include <stdlib.h>
-#if ((' ' & 0x0FF) == 0x020)
-# define ISLOWER(c) ('a' <= (c) && (c) <= 'z')
-# define TOUPPER(c) (ISLOWER(c) ? 'A' + ((c) - 'a') : (c))
-#else
-# define ISLOWER(c) \
-		   (('a' <= (c) && (c) <= 'i') \
-		     || ('j' <= (c) && (c) <= 'r') \
-		     || ('s' <= (c) && (c) <= 'z'))
-# define TOUPPER(c) (ISLOWER(c) ? ((c) | 0x40) : (c))
-#endif
-
-#define XOR(e, f) (((e) && !(f)) || (!(e) && (f)))
-int
-main ()
-{
-  int i;
-  for (i = 0; i < 256; i++)
-    if (XOR (islower (i), ISLOWER (i))
-	|| toupper (i) != TOUPPER (i))
-      return 2;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_run "$LINENO"; then :
-
-else
-  ac_cv_header_stdc=no
-fi
-rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
-  conftest.$ac_objext conftest.beam conftest.$ac_ext
-fi
-
-fi
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_header_stdc" >&5
-$as_echo "$ac_cv_header_stdc" >&6; }
-if test $ac_cv_header_stdc = yes; then
-
-$as_echo "#define STDC_HEADERS 1" >>confdefs.h
-
-fi
-
-# On IRIX 5.3, sys/types and inttypes.h are conflicting.
-for ac_header in sys/types.h sys/stat.h stdlib.h string.h memory.h strings.h \
-		  inttypes.h stdint.h unistd.h
-do :
-  as_ac_Header=`$as_echo "ac_cv_header_$ac_header" | $as_tr_sh`
-ac_fn_cxx_check_header_compile "$LINENO" "$ac_header" "$as_ac_Header" "$ac_includes_default
-"
-if eval test \"x\$"$as_ac_Header"\" = x"yes"; then :
-  cat >>confdefs.h <<_ACEOF
-#define `$as_echo "HAVE_$ac_header" | $as_tr_cpp` 1
-_ACEOF
-
-fi
-
-done
-
-
-for ac_header in stdint.h
-do :
-  ac_fn_cxx_check_header_mongrel "$LINENO" "stdint.h" "ac_cv_header_stdint_h" "$ac_includes_default"
-if test "x$ac_cv_header_stdint_h" = xyes; then :
-  cat >>confdefs.h <<_ACEOF
-#define HAVE_STDINT_H 1
-_ACEOF
-
-fi
-
-done
-
-for ac_header in mm_malloc.h
-do :
-  ac_fn_cxx_check_header_mongrel "$LINENO" "mm_malloc.h" "ac_cv_header_mm_malloc_h" "$ac_includes_default"
-if test "x$ac_cv_header_mm_malloc_h" = xyes; then :
-  cat >>confdefs.h <<_ACEOF
-#define HAVE_MM_MALLOC_H 1
-_ACEOF
-
-fi
-
-done
-
-for ac_header in malloc/malloc.h
-do :
-  ac_fn_cxx_check_header_mongrel "$LINENO" "malloc/malloc.h" "ac_cv_header_malloc_malloc_h" "$ac_includes_default"
-if test "x$ac_cv_header_malloc_malloc_h" = xyes; then :
-  cat >>confdefs.h <<_ACEOF
-#define HAVE_MALLOC_MALLOC_H 1
-_ACEOF
-
-fi
-
-done
-
-for ac_header in malloc.h
-do :
-  ac_fn_cxx_check_header_mongrel "$LINENO" "malloc.h" "ac_cv_header_malloc_h" "$ac_includes_default"
-if test "x$ac_cv_header_malloc_h" = xyes; then :
-  cat >>confdefs.h <<_ACEOF
-#define HAVE_MALLOC_H 1
-_ACEOF
-
-fi
-
-done
-
-for ac_header in endian.h
-do :
-  ac_fn_cxx_check_header_mongrel "$LINENO" "endian.h" "ac_cv_header_endian_h" "$ac_includes_default"
-if test "x$ac_cv_header_endian_h" = xyes; then :
-  cat >>confdefs.h <<_ACEOF
-#define HAVE_ENDIAN_H 1
-_ACEOF
-
-fi
-
-done
-
-for ac_header in gmp.h
-do :
-  ac_fn_cxx_check_header_mongrel "$LINENO" "gmp.h" "ac_cv_header_gmp_h" "$ac_includes_default"
-if test "x$ac_cv_header_gmp_h" = xyes; then :
-  cat >>confdefs.h <<_ACEOF
-#define HAVE_GMP_H 1
-_ACEOF
-
-fi
-
-done
-
-ac_fn_cxx_check_decl "$LINENO" "ntohll" "ac_cv_have_decl_ntohll" "#include <arpa/inet.h>
-"
-if test "x$ac_cv_have_decl_ntohll" = xyes; then :
-  ac_have_decl=1
-else
-  ac_have_decl=0
-fi
-
-cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_NTOHLL $ac_have_decl
-_ACEOF
-
-ac_fn_cxx_check_decl "$LINENO" "be64toh" "ac_cv_have_decl_be64toh" "#include <arpa/inet.h>
-"
-if test "x$ac_cv_have_decl_be64toh" = xyes; then :
-  ac_have_decl=1
-else
-  ac_have_decl=0
-fi
-
-cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_BE64TOH $ac_have_decl
-_ACEOF
-
-
-# Checks for typedefs, structures, and compiler characteristics.
-ac_fn_cxx_check_type "$LINENO" "size_t" "ac_cv_type_size_t" "$ac_includes_default"
-if test "x$ac_cv_type_size_t" = xyes; then :
-
-else
-
-cat >>confdefs.h <<_ACEOF
-#define size_t unsigned int
-_ACEOF
-
-fi
-
-ac_fn_c_find_uintX_t "$LINENO" "32" "ac_cv_c_uint32_t"
-case $ac_cv_c_uint32_t in #(
-  no|yes) ;; #(
-  *)
-
-$as_echo "#define _UINT32_T 1" >>confdefs.h
-
-
-cat >>confdefs.h <<_ACEOF
-#define uint32_t $ac_cv_c_uint32_t
-_ACEOF
-;;
-  esac
-
-ac_fn_c_find_uintX_t "$LINENO" "64" "ac_cv_c_uint64_t"
-case $ac_cv_c_uint64_t in #(
-  no|yes) ;; #(
-  *)
-
-$as_echo "#define _UINT64_T 1" >>confdefs.h
-
-
-cat >>confdefs.h <<_ACEOF
-#define uint64_t $ac_cv_c_uint64_t
-_ACEOF
-;;
-  esac
-
-
-# Checks for library functions.
-echo
-echo Checking libraries
-echo :::::::::::::::::::::::::::::::::::::::::::
-
-
-for ac_func in gettimeofday
-do :
-  ac_fn_cxx_check_func "$LINENO" "gettimeofday" "ac_cv_func_gettimeofday"
-if test "x$ac_cv_func_gettimeofday" = xyes; then :
-  cat >>confdefs.h <<_ACEOF
-#define HAVE_GETTIMEOFDAY 1
-_ACEOF
-
-fi
-done
-
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for __gmpf_init in -lgmp" >&5
-$as_echo_n "checking for __gmpf_init in -lgmp... " >&6; }
-if ${ac_cv_lib_gmp___gmpf_init+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  ac_check_lib_save_LIBS=$LIBS
-LIBS="-lgmp  $LIBS"
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-/* Override any GCC internal prototype to avoid an error.
-   Use char because int might match the return type of a GCC
-   builtin and then its argument prototype would still apply.  */
-#ifdef __cplusplus
-extern "C"
-#endif
-char __gmpf_init ();
-int
-main ()
-{
-return __gmpf_init ();
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_link "$LINENO"; then :
-  ac_cv_lib_gmp___gmpf_init=yes
-else
-  ac_cv_lib_gmp___gmpf_init=no
-fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
-LIBS=$ac_check_lib_save_LIBS
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_gmp___gmpf_init" >&5
-$as_echo "$ac_cv_lib_gmp___gmpf_init" >&6; }
-if test "x$ac_cv_lib_gmp___gmpf_init" = xyes; then :
-  cat >>confdefs.h <<_ACEOF
-#define HAVE_LIBGMP 1
-_ACEOF
-
-  LIBS="-lgmp $LIBS"
-
-else
-  as_fn_error $? "GNU Multiple Precision GMP library was not found in your system.
-Please install or provide the correct path to your installation
-Info at: http://www.gmplib.org" "$LINENO" 5
-fi
-
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for mpfr_init in -lmpfr" >&5
-$as_echo_n "checking for mpfr_init in -lmpfr... " >&6; }
-if ${ac_cv_lib_mpfr_mpfr_init+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  ac_check_lib_save_LIBS=$LIBS
-LIBS="-lmpfr  $LIBS"
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-/* Override any GCC internal prototype to avoid an error.
-   Use char because int might match the return type of a GCC
-   builtin and then its argument prototype would still apply.  */
-#ifdef __cplusplus
-extern "C"
-#endif
-char mpfr_init ();
-int
-main ()
-{
-return mpfr_init ();
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_cxx_try_link "$LINENO"; then :
-  ac_cv_lib_mpfr_mpfr_init=yes
-else
-  ac_cv_lib_mpfr_mpfr_init=no
-fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
-LIBS=$ac_check_lib_save_LIBS
-fi
-{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_mpfr_mpfr_init" >&5
-$as_echo "$ac_cv_lib_mpfr_mpfr_init" >&6; }
-if test "x$ac_cv_lib_mpfr_mpfr_init" = xyes; then :
-  cat >>confdefs.h <<_ACEOF
-#define HAVE_LIBMPFR 1
-_ACEOF
-
-  LIBS="-lmpfr $LIBS"
-
-else
-  as_fn_error $? "GNU Multiple Precision MPFR library was not found in your system.
-Please install or provide the correct path to your installation
-Info at: http://www.mpfr.org/" "$LINENO" 5
-fi
-
-
-
-# Check whether --enable-simd was given.
-if test "${enable_simd+set}" = set; then :
-  enableval=$enable_simd; \
-	ac_SIMD=${enable_simd}
-else
-  ac_SIMD=AVX2
-fi
-
-
-supported=no
-
-case ${ac_SIMD} in
-     SSE4)
-       echo Configuring for SSE4
-
-$as_echo "#define SSE4 1" >>confdefs.h
-
-       if test x"$ax_cv_support_ssse3_ext" = x"yes"; then           supported=yes
-       else
-  	{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor does not support SSE4 instructions" >&5
-$as_echo "$as_me: WARNING: Your processor does not support SSE4 instructions" >&2;}
-       fi
-     ;;
-     AVX)
-       echo Configuring for AVX
-
-$as_echo "#define AVX1 1" >>confdefs.h
-
-       if test x"$ax_cv_support_avx_ext" = x"yes"; then         supported=yes
-       else
-       	{ $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor does not support AVX instructions" >&5
-$as_echo "$as_me: WARNING: Your processor does not support AVX instructions" >&2;}
-       fi
-     ;;
-     AVX2)
-       echo Configuring for AVX2
-
-$as_echo "#define AVX2 1" >>confdefs.h
-
-       if test x"$ax_cv_support_avx2_ext" = x"yes"; then         supported=yes
-       else
-       { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: Your processor does not support AVX2 instructions" >&5
-$as_echo "$as_me: WARNING: Your processor does not support AVX2 instructions" >&2;}
-       fi
-     ;;
-     AVX512|MIC)
-       echo Configuring for AVX512 and MIC
-
-$as_echo "#define AVX512 1" >>confdefs.h
-
-       supported="cross compilation"
-     ;;
-     NEONv7)
-       echo Configuring for experimental ARMv7 support
-
-$as_echo "#define NEONv7 1" >>confdefs.h
-
-       supported="cross compilation"
-     ;;
-     DEBUG)
-       echo Configuring without SIMD support - only for compiler DEBUGGING!
-
-$as_echo "#define EMPTY_SIMD 1" >>confdefs.h
-
-      ;;
-     *)
-     as_fn_error $? "${ac_SIMD} flag unsupported as --enable-simd option\nRun ./configure --help for the list of options" "$LINENO" 5;
-     ;;
-esac
-
-# Check whether --enable-precision was given.
-if test "${enable_precision+set}" = set; then :
-  enableval=$enable_precision; ac_PRECISION=${enable_precision}
-else
-  ac_PRECISION=single
-fi
-
-case ${ac_PRECISION} in
-     single)
-       echo default precision is single
-
-$as_echo "#define GRID_DEFAULT_PRECISION_SINGLE 1" >>confdefs.h
-
-     ;;
-     double)
-       echo default precision is double
-
-$as_echo "#define GRID_DEFAULT_PRECISION_DOUBLE 1" >>confdefs.h
-
-     ;;
-esac
-
-# Check whether --enable-comms was given.
-if test "${enable_comms+set}" = set; then :
-  enableval=$enable_comms; ac_COMMS=${enable_comms}
-else
-  ac_COMMS=none
-fi
-
-
-case ${ac_COMMS} in
-     none)
-       echo Configuring for NO communications
-
-$as_echo "#define GRID_COMMS_NONE 1" >>confdefs.h
-
-     ;;
-     mpi)
-       echo Configuring for MPI communications
-
-$as_echo "#define GRID_COMMS_MPI 1" >>confdefs.h
-
-     ;;
-     *)
-     as_fn_error $? "${ac_COMMS} unsupported --enable-comms option" "$LINENO" 5;
-     ;;
-esac
-
- if  test "X${ac_COMMS}X" == "XmpiX" ; then
-  BUILD_COMMS_MPI_TRUE=
-  BUILD_COMMS_MPI_FALSE='#'
-else
-  BUILD_COMMS_MPI_TRUE='#'
-  BUILD_COMMS_MPI_FALSE=
-fi
-
- if  test "X${ac_COMMS}X" == "XnoneX" ; then
-  BUILD_COMMS_NONE_TRUE=
-  BUILD_COMMS_NONE_FALSE='#'
-else
-  BUILD_COMMS_NONE_TRUE='#'
-  BUILD_COMMS_NONE_FALSE=
-fi
-
-
-
-###################################################################
-# Checks for doxygen support
-# if present enables the "make doxyfile" command
-echo
-echo Checking doxygen support
-echo :::::::::::::::::::::::::::::::::::::::::::
-
-# Check whether --enable-doxygen was given.
-if test "${enable_doxygen+set}" = set; then :
-  enableval=$enable_doxygen;
-fi
-
-
-# Check whether --enable-dot was given.
-if test "${enable_dot+set}" = set; then :
-  enableval=$enable_dot;
-fi
-
-# Check whether --enable-html-docs was given.
-if test "${enable_html_docs+set}" = set; then :
-  enableval=$enable_html_docs;
-else
-   enable_html_docs=yes
-fi
-
-# Check whether --enable-latex-docs was given.
-if test "${enable_latex_docs+set}" = set; then :
-  enableval=$enable_latex_docs;
-else
-  enable_latex_docs=no
-fi
-
-
-if test "x$enable_doxygen" = xno; then
- enable_doc=no
-else
- # Extract the first word of "doxygen", so it can be a program name with args.
-set dummy doxygen; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_DOXYGEN+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$DOXYGEN"; then
-  ac_cv_prog_DOXYGEN="$DOXYGEN" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_DOXYGEN="doxygen"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-DOXYGEN=$ac_cv_prog_DOXYGEN
-if test -n "$DOXYGEN"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $DOXYGEN" >&5
-$as_echo "$DOXYGEN" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-
- if test x$DOXYGEN = x; then
-  if test "x$enable_doxygen" = xyes; then
-   as_fn_error $? "could not find doxygen" "$LINENO" 5
-  fi
-  enable_doc=no
- else
-  doxy_ver=`doxygen --version`
-  doxy_major=`expr "$doxy_ver" : '\([0-9]\)\..*'`
-  doxy_minor=`expr "$doxy_ver" : '[0-9]\.\([0-9]\).*'`
-  if test $doxy_major -eq "1" -a $doxy_minor -ge "3" ; then
-   enable_doc=yes
-   # Extract the first word of "dot", so it can be a program name with args.
-set dummy dot; ac_word=$2
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for $ac_word" >&5
-$as_echo_n "checking for $ac_word... " >&6; }
-if ${ac_cv_prog_DOT+:} false; then :
-  $as_echo_n "(cached) " >&6
-else
-  if test -n "$DOT"; then
-  ac_cv_prog_DOT="$DOT" # Let the user override the test.
-else
-as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    for ac_exec_ext in '' $ac_executable_extensions; do
-  if as_fn_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
-    ac_cv_prog_DOT="dot"
-    $as_echo "$as_me:${as_lineno-$LINENO}: found $as_dir/$ac_word$ac_exec_ext" >&5
-    break 2
-  fi
-done
-  done
-IFS=$as_save_IFS
-
-fi
-fi
-DOT=$ac_cv_prog_DOT
-if test -n "$DOT"; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: $DOT" >&5
-$as_echo "$DOT" >&6; }
-else
-  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
-$as_echo "no" >&6; }
-fi
-
-
-  else
-   { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: doxygen version $doxy_ver too old, doxygen will not be used." >&5
-$as_echo "$as_me: WARNING: doxygen version $doxy_ver too old, doxygen will not be used." >&2;}
-   enable_doc=no
-  fi
- fi
-fi
-
- if test x$enable_doc = xyes; then
-  DOXYGEN_DOC_TRUE=
-  DOXYGEN_DOC_FALSE='#'
-else
-  DOXYGEN_DOC_TRUE='#'
-  DOXYGEN_DOC_FALSE=
-fi
-
-
-if test x$DOT = x; then
- if test "x$enable_dot" = xyes; then
-  as_fn_error $? "could not find dot" "$LINENO" 5
- fi
- enable_dot=no
-else
- enable_dot=yes
-fi
-
-
-
-
-
-
-if test -n "$DOXYGEN"
-then
-ac_config_files="$ac_config_files docs/doxy.cfg"
-
-fi
-
-echo
-echo Creating configuration files
-echo :::::::::::::::::::::::::::::::::::::::::::
-ac_config_files="$ac_config_files Makefile"
-
-ac_config_files="$ac_config_files lib/Makefile"
-
-ac_config_files="$ac_config_files tests/Makefile"
-
-ac_config_files="$ac_config_files benchmarks/Makefile"
-
-cat >confcache <<\_ACEOF
-# This file is a shell script that caches the results of configure
-# tests run on this system so they can be shared between configure
-# scripts and configure runs, see configure's option --config-cache.
-# It is not useful on other systems.  If it contains results you don't
-# want to keep, you may remove or edit it.
-#
-# config.status only pays attention to the cache file if you give it
-# the --recheck option to rerun configure.
-#
-# `ac_cv_env_foo' variables (set or unset) will be overridden when
-# loading this file, other *unset* `ac_cv_foo' will be assigned the
-# following values.
-
-_ACEOF
-
-# The following way of writing the cache mishandles newlines in values,
-# but we know of no workaround that is simple, portable, and efficient.
-# So, we kill variables containing newlines.
-# Ultrix sh set writes to stderr and can't be redirected directly,
-# and sets the high bit in the cache file unless we assign to the vars.
-(
-  for ac_var in `(set) 2>&1 | sed -n 's/^\([a-zA-Z_][a-zA-Z0-9_]*\)=.*/\1/p'`; do
-    eval ac_val=\$$ac_var
-    case $ac_val in #(
-    *${as_nl}*)
-      case $ac_var in #(
-      *_cv_*) { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: cache variable $ac_var contains a newline" >&5
-$as_echo "$as_me: WARNING: cache variable $ac_var contains a newline" >&2;} ;;
-      esac
-      case $ac_var in #(
-      _ | IFS | as_nl) ;; #(
-      BASH_ARGV | BASH_SOURCE) eval $ac_var= ;; #(
-      *) { eval $ac_var=; unset $ac_var;} ;;
-      esac ;;
-    esac
-  done
-
-  (set) 2>&1 |
-    case $as_nl`(ac_space=' '; set) 2>&1` in #(
-    *${as_nl}ac_space=\ *)
-      # `set' does not quote correctly, so add quotes: double-quote
-      # substitution turns \\\\ into \\, and sed turns \\ into \.
-      sed -n \
-	"s/'/'\\\\''/g;
-	  s/^\\([_$as_cr_alnum]*_cv_[_$as_cr_alnum]*\\)=\\(.*\\)/\\1='\\2'/p"
-      ;; #(
-    *)
-      # `set' quotes correctly as required by POSIX, so do not add quotes.
-      sed -n "/^[_$as_cr_alnum]*_cv_[_$as_cr_alnum]*=/p"
-      ;;
-    esac |
-    sort
-) |
-  sed '
-     /^ac_cv_env_/b end
-     t clear
-     :clear
-     s/^\([^=]*\)=\(.*[{}].*\)$/test "${\1+set}" = set || &/
-     t end
-     s/^\([^=]*\)=\(.*\)$/\1=${\1=\2}/
-     :end' >>confcache
-if diff "$cache_file" confcache >/dev/null 2>&1; then :; else
-  if test -w "$cache_file"; then
-    if test "x$cache_file" != "x/dev/null"; then
-      { $as_echo "$as_me:${as_lineno-$LINENO}: updating cache $cache_file" >&5
-$as_echo "$as_me: updating cache $cache_file" >&6;}
-      if test ! -f "$cache_file" || test -h "$cache_file"; then
-	cat confcache >"$cache_file"
-      else
-        case $cache_file in #(
-        */* | ?:*)
-	  mv -f confcache "$cache_file"$$ &&
-	  mv -f "$cache_file"$$ "$cache_file" ;; #(
-        *)
-	  mv -f confcache "$cache_file" ;;
-	esac
-      fi
-    fi
-  else
-    { $as_echo "$as_me:${as_lineno-$LINENO}: not updating unwritable cache $cache_file" >&5
-$as_echo "$as_me: not updating unwritable cache $cache_file" >&6;}
-  fi
-fi
-rm -f confcache
-
-test "x$prefix" = xNONE && prefix=$ac_default_prefix
-# Let make expand exec_prefix.
-test "x$exec_prefix" = xNONE && exec_prefix='${prefix}'
-
-DEFS=-DHAVE_CONFIG_H
-
-ac_libobjs=
-ac_ltlibobjs=
-U=
-for ac_i in : $LIBOBJS; do test "x$ac_i" = x: && continue
-  # 1. Remove the extension, and $U if already installed.
-  ac_script='s/\$U\././;s/\.o$//;s/\.obj$//'
-  ac_i=`$as_echo "$ac_i" | sed "$ac_script"`
-  # 2. Prepend LIBOBJDIR.  When used with automake>=1.10 LIBOBJDIR
-  #    will be set to the directory where LIBOBJS objects are built.
-  as_fn_append ac_libobjs " \${LIBOBJDIR}$ac_i\$U.$ac_objext"
-  as_fn_append ac_ltlibobjs " \${LIBOBJDIR}$ac_i"'$U.lo'
-done
-LIBOBJS=$ac_libobjs
-
-LTLIBOBJS=$ac_ltlibobjs
-
-
-{ $as_echo "$as_me:${as_lineno-$LINENO}: checking that generated files are newer than configure" >&5
-$as_echo_n "checking that generated files are newer than configure... " >&6; }
-   if test -n "$am_sleep_pid"; then
-     # Hide warnings about reused PIDs.
-     wait $am_sleep_pid 2>/dev/null
-   fi
-   { $as_echo "$as_me:${as_lineno-$LINENO}: result: done" >&5
-$as_echo "done" >&6; }
- if test -n "$EXEEXT"; then
-  am__EXEEXT_TRUE=
-  am__EXEEXT_FALSE='#'
-else
-  am__EXEEXT_TRUE='#'
-  am__EXEEXT_FALSE=
-fi
-
-if test -z "${AMDEP_TRUE}" && test -z "${AMDEP_FALSE}"; then
-  as_fn_error $? "conditional \"AMDEP\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
-fi
-if test -z "${am__fastdepCXX_TRUE}" && test -z "${am__fastdepCXX_FALSE}"; then
-  as_fn_error $? "conditional \"am__fastdepCXX\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
-fi
-if test -z "${am__fastdepCC_TRUE}" && test -z "${am__fastdepCC_FALSE}"; then
-  as_fn_error $? "conditional \"am__fastdepCC\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
-fi
-if test -z "${BUILD_COMMS_MPI_TRUE}" && test -z "${BUILD_COMMS_MPI_FALSE}"; then
-  as_fn_error $? "conditional \"BUILD_COMMS_MPI\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
-fi
-if test -z "${BUILD_COMMS_NONE_TRUE}" && test -z "${BUILD_COMMS_NONE_FALSE}"; then
-  as_fn_error $? "conditional \"BUILD_COMMS_NONE\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
-fi
-if test -z "${DOXYGEN_DOC_TRUE}" && test -z "${DOXYGEN_DOC_FALSE}"; then
-  as_fn_error $? "conditional \"DOXYGEN_DOC\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
-fi
-
-: "${CONFIG_STATUS=./config.status}"
-ac_write_fail=0
-ac_clean_files_save=$ac_clean_files
-ac_clean_files="$ac_clean_files $CONFIG_STATUS"
-{ $as_echo "$as_me:${as_lineno-$LINENO}: creating $CONFIG_STATUS" >&5
-$as_echo "$as_me: creating $CONFIG_STATUS" >&6;}
-as_write_fail=0
-cat >$CONFIG_STATUS <<_ASEOF || as_write_fail=1
-#! $SHELL
-# Generated by $as_me.
-# Run this file to recreate the current configuration.
-# Compiler output produced by configure, useful for debugging
-# configure, is in config.log if it exists.
-
-debug=false
-ac_cs_recheck=false
-ac_cs_silent=false
-
-SHELL=\${CONFIG_SHELL-$SHELL}
-export SHELL
-_ASEOF
-cat >>$CONFIG_STATUS <<\_ASEOF || as_write_fail=1
-## -------------------- ##
-## M4sh Initialization. ##
-## -------------------- ##
-
-# Be more Bourne compatible
-DUALCASE=1; export DUALCASE # for MKS sh
-if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then :
-  emulate sh
-  NULLCMD=:
-  # Pre-4.2 versions of Zsh do word splitting on ${1+"$@"}, which
-  # is contrary to our usage.  Disable this feature.
-  alias -g '${1+"$@"}'='"$@"'
-  setopt NO_GLOB_SUBST
-else
-  case `(set -o) 2>/dev/null` in #(
-  *posix*) :
-    set -o posix ;; #(
-  *) :
-     ;;
-esac
-fi
-
-
-as_nl='
-'
-export as_nl
-# Printing a long string crashes Solaris 7 /usr/bin/printf.
-as_echo='\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\'
-as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo
-as_echo=$as_echo$as_echo$as_echo$as_echo$as_echo$as_echo
-# Prefer a ksh shell builtin over an external printf program on Solaris,
-# but without wasting forks for bash or zsh.
-if test -z "$BASH_VERSION$ZSH_VERSION" \
-    && (test "X`print -r -- $as_echo`" = "X$as_echo") 2>/dev/null; then
-  as_echo='print -r --'
-  as_echo_n='print -rn --'
-elif (test "X`printf %s $as_echo`" = "X$as_echo") 2>/dev/null; then
-  as_echo='printf %s\n'
-  as_echo_n='printf %s'
-else
-  if test "X`(/usr/ucb/echo -n -n $as_echo) 2>/dev/null`" = "X-n $as_echo"; then
-    as_echo_body='eval /usr/ucb/echo -n "$1$as_nl"'
-    as_echo_n='/usr/ucb/echo -n'
-  else
-    as_echo_body='eval expr "X$1" : "X\\(.*\\)"'
-    as_echo_n_body='eval
-      arg=$1;
-      case $arg in #(
-      *"$as_nl"*)
-	expr "X$arg" : "X\\(.*\\)$as_nl";
-	arg=`expr "X$arg" : ".*$as_nl\\(.*\\)"`;;
-      esac;
-      expr "X$arg" : "X\\(.*\\)" | tr -d "$as_nl"
-    '
-    export as_echo_n_body
-    as_echo_n='sh -c $as_echo_n_body as_echo'
-  fi
-  export as_echo_body
-  as_echo='sh -c $as_echo_body as_echo'
-fi
-
-# The user is always right.
-if test "${PATH_SEPARATOR+set}" != set; then
-  PATH_SEPARATOR=:
-  (PATH='/bin;/bin'; FPATH=$PATH; sh -c :) >/dev/null 2>&1 && {
-    (PATH='/bin:/bin'; FPATH=$PATH; sh -c :) >/dev/null 2>&1 ||
-      PATH_SEPARATOR=';'
-  }
-fi
-
-
-# IFS
-# We need space, tab and new line, in precisely that order.  Quoting is
-# there to prevent editors from complaining about space-tab.
-# (If _AS_PATH_WALK were called with IFS unset, it would disable word
-# splitting by setting IFS to empty value.)
-IFS=" ""	$as_nl"
-
-# Find who we are.  Look in the path if we contain no directory separator.
-as_myself=
-case $0 in #((
-  *[\\/]* ) as_myself=$0 ;;
-  *) as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
-for as_dir in $PATH
-do
-  IFS=$as_save_IFS
-  test -z "$as_dir" && as_dir=.
-    test -r "$as_dir/$0" && as_myself=$as_dir/$0 && break
-  done
-IFS=$as_save_IFS
-
-     ;;
-esac
-# We did not find ourselves, most probably we were run as `sh COMMAND'
-# in which case we are not to be found in the path.
-if test "x$as_myself" = x; then
-  as_myself=$0
-fi
-if test ! -f "$as_myself"; then
-  $as_echo "$as_myself: error: cannot find myself; rerun with an absolute file name" >&2
-  exit 1
-fi
-
-# Unset variables that we do not need and which cause bugs (e.g. in
-# pre-3.0 UWIN ksh).  But do not cause bugs in bash 2.01; the "|| exit 1"
-# suppresses any "Segmentation fault" message there.  '((' could
-# trigger a bug in pdksh 5.2.14.
-for as_var in BASH_ENV ENV MAIL MAILPATH
-do eval test x\${$as_var+set} = xset \
-  && ( (unset $as_var) || exit 1) >/dev/null 2>&1 && unset $as_var || :
-done
-PS1='$ '
-PS2='> '
-PS4='+ '
-
-# NLS nuisances.
-LC_ALL=C
-export LC_ALL
-LANGUAGE=C
-export LANGUAGE
-
-# CDPATH.
-(unset CDPATH) >/dev/null 2>&1 && unset CDPATH
-
-
-# as_fn_error STATUS ERROR [LINENO LOG_FD]
-# ----------------------------------------
-# Output "`basename $0`: error: ERROR" to stderr. If LINENO and LOG_FD are
-# provided, also output the error to LOG_FD, referencing LINENO. Then exit the
-# script with STATUS, using 1 if that was 0.
-as_fn_error ()
-{
-  as_status=$1; test $as_status -eq 0 && as_status=1
-  if test "$4"; then
-    as_lineno=${as_lineno-"$3"} as_lineno_stack=as_lineno_stack=$as_lineno_stack
-    $as_echo "$as_me:${as_lineno-$LINENO}: error: $2" >&$4
-  fi
-  $as_echo "$as_me: error: $2" >&2
-  as_fn_exit $as_status
-} # as_fn_error
-
-
-# as_fn_set_status STATUS
-# -----------------------
-# Set $? to STATUS, without forking.
-as_fn_set_status ()
-{
-  return $1
-} # as_fn_set_status
-
-# as_fn_exit STATUS
-# -----------------
-# Exit the shell with STATUS, even in a "trap 0" or "set -e" context.
-as_fn_exit ()
-{
-  set +e
-  as_fn_set_status $1
-  exit $1
-} # as_fn_exit
-
-# as_fn_unset VAR
-# ---------------
-# Portably unset VAR.
-as_fn_unset ()
-{
-  { eval $1=; unset $1;}
-}
-as_unset=as_fn_unset
-# as_fn_append VAR VALUE
-# ----------------------
-# Append the text in VALUE to the end of the definition contained in VAR. Take
-# advantage of any shell optimizations that allow amortized linear growth over
-# repeated appends, instead of the typical quadratic growth present in naive
-# implementations.
-if (eval "as_var=1; as_var+=2; test x\$as_var = x12") 2>/dev/null; then :
-  eval 'as_fn_append ()
-  {
-    eval $1+=\$2
-  }'
-else
-  as_fn_append ()
-  {
-    eval $1=\$$1\$2
-  }
-fi # as_fn_append
-
-# as_fn_arith ARG...
-# ------------------
-# Perform arithmetic evaluation on the ARGs, and store the result in the
-# global $as_val. Take advantage of shells that can avoid forks. The arguments
-# must be portable across $(()) and expr.
-if (eval "test \$(( 1 + 1 )) = 2") 2>/dev/null; then :
-  eval 'as_fn_arith ()
-  {
-    as_val=$(( $* ))
-  }'
-else
-  as_fn_arith ()
-  {
-    as_val=`expr "$@" || test $? -eq 1`
-  }
-fi # as_fn_arith
-
-
-if expr a : '\(a\)' >/dev/null 2>&1 &&
-   test "X`expr 00001 : '.*\(...\)'`" = X001; then
-  as_expr=expr
-else
-  as_expr=false
-fi
-
-if (basename -- /) >/dev/null 2>&1 && test "X`basename -- / 2>&1`" = "X/"; then
-  as_basename=basename
-else
-  as_basename=false
-fi
-
-if (as_dir=`dirname -- /` && test "X$as_dir" = X/) >/dev/null 2>&1; then
-  as_dirname=dirname
-else
-  as_dirname=false
-fi
-
-as_me=`$as_basename -- "$0" ||
-$as_expr X/"$0" : '.*/\([^/][^/]*\)/*$' \| \
-	 X"$0" : 'X\(//\)$' \| \
-	 X"$0" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X/"$0" |
-    sed '/^.*\/\([^/][^/]*\)\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\/\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\/\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-
-# Avoid depending upon Character Ranges.
-as_cr_letters='abcdefghijklmnopqrstuvwxyz'
-as_cr_LETTERS='ABCDEFGHIJKLMNOPQRSTUVWXYZ'
-as_cr_Letters=$as_cr_letters$as_cr_LETTERS
-as_cr_digits='0123456789'
-as_cr_alnum=$as_cr_Letters$as_cr_digits
-
-ECHO_C= ECHO_N= ECHO_T=
-case `echo -n x` in #(((((
--n*)
-  case `echo 'xy\c'` in
-  *c*) ECHO_T='	';;	# ECHO_T is single tab character.
-  xy)  ECHO_C='\c';;
-  *)   echo `echo ksh88 bug on AIX 6.1` > /dev/null
-       ECHO_T='	';;
-  esac;;
-*)
-  ECHO_N='-n';;
-esac
-
-rm -f conf$$ conf$$.exe conf$$.file
-if test -d conf$$.dir; then
-  rm -f conf$$.dir/conf$$.file
-else
-  rm -f conf$$.dir
-  mkdir conf$$.dir 2>/dev/null
-fi
-if (echo >conf$$.file) 2>/dev/null; then
-  if ln -s conf$$.file conf$$ 2>/dev/null; then
-    as_ln_s='ln -s'
-    # ... but there are two gotchas:
-    # 1) On MSYS, both `ln -s file dir' and `ln file dir' fail.
-    # 2) DJGPP < 2.04 has no symlinks; `ln -s' creates a wrapper executable.
-    # In both cases, we have to default to `cp -pR'.
-    ln -s conf$$.file conf$$.dir 2>/dev/null && test ! -f conf$$.exe ||
-      as_ln_s='cp -pR'
-  elif ln conf$$.file conf$$ 2>/dev/null; then
-    as_ln_s=ln
-  else
-    as_ln_s='cp -pR'
-  fi
-else
-  as_ln_s='cp -pR'
-fi
-rm -f conf$$ conf$$.exe conf$$.dir/conf$$.file conf$$.file
-rmdir conf$$.dir 2>/dev/null
-
-
-# as_fn_mkdir_p
-# -------------
-# Create "$as_dir" as a directory, including parents if necessary.
-as_fn_mkdir_p ()
-{
-
-  case $as_dir in #(
-  -*) as_dir=./$as_dir;;
-  esac
-  test -d "$as_dir" || eval $as_mkdir_p || {
-    as_dirs=
-    while :; do
-      case $as_dir in #(
-      *\'*) as_qdir=`$as_echo "$as_dir" | sed "s/'/'\\\\\\\\''/g"`;; #'(
-      *) as_qdir=$as_dir;;
-      esac
-      as_dirs="'$as_qdir' $as_dirs"
-      as_dir=`$as_dirname -- "$as_dir" ||
-$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$as_dir" : 'X\(//\)[^/]' \| \
-	 X"$as_dir" : 'X\(//\)$' \| \
-	 X"$as_dir" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$as_dir" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      test -d "$as_dir" && break
-    done
-    test -z "$as_dirs" || eval "mkdir $as_dirs"
-  } || test -d "$as_dir" || as_fn_error $? "cannot create directory $as_dir"
-
-
-} # as_fn_mkdir_p
-if mkdir -p . 2>/dev/null; then
-  as_mkdir_p='mkdir -p "$as_dir"'
-else
-  test -d ./-p && rmdir ./-p
-  as_mkdir_p=false
-fi
-
-
-# as_fn_executable_p FILE
-# -----------------------
-# Test if FILE is an executable regular file.
-as_fn_executable_p ()
-{
-  test -f "$1" && test -x "$1"
-} # as_fn_executable_p
-as_test_x='test -x'
-as_executable_p=as_fn_executable_p
-
-# Sed expression to map a string onto a valid CPP name.
-as_tr_cpp="eval sed 'y%*$as_cr_letters%P$as_cr_LETTERS%;s%[^_$as_cr_alnum]%_%g'"
-
-# Sed expression to map a string onto a valid variable name.
-as_tr_sh="eval sed 'y%*+%pp%;s%[^_$as_cr_alnum]%_%g'"
-
-
-exec 6>&1
-## ----------------------------------- ##
-## Main body of $CONFIG_STATUS script. ##
-## ----------------------------------- ##
-_ASEOF
-test $as_write_fail = 0 && chmod +x $CONFIG_STATUS || ac_write_fail=1
-
-cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-# Save the log message, to keep $0 and so on meaningful, and to
-# report actual input values of CONFIG_FILES etc. instead of their
-# values after options handling.
-ac_log="
-This file was extended by Grid $as_me 1.0, which was
-generated by GNU Autoconf 2.69.  Invocation command line was
-
-  CONFIG_FILES    = $CONFIG_FILES
-  CONFIG_HEADERS  = $CONFIG_HEADERS
-  CONFIG_LINKS    = $CONFIG_LINKS
-  CONFIG_COMMANDS = $CONFIG_COMMANDS
-  $ $0 $@
-
-on `(hostname || uname -n) 2>/dev/null | sed 1q`
-"
-
-_ACEOF
-
-case $ac_config_files in *"
-"*) set x $ac_config_files; shift; ac_config_files=$*;;
-esac
-
-case $ac_config_headers in *"
-"*) set x $ac_config_headers; shift; ac_config_headers=$*;;
-esac
-
-
-cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-# Files that config.status was made for.
-config_files="$ac_config_files"
-config_headers="$ac_config_headers"
-config_commands="$ac_config_commands"
-
-_ACEOF
-
-cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-ac_cs_usage="\
-\`$as_me' instantiates files and other configuration actions
-from templates according to the current configuration.  Unless the files
-and actions are specified as TAGs, all are instantiated by default.
-
-Usage: $0 [OPTION]... [TAG]...
-
-  -h, --help       print this help, then exit
-  -V, --version    print version number and configuration settings, then exit
-      --config     print configuration, then exit
-  -q, --quiet, --silent
-                   do not print progress messages
-  -d, --debug      don't remove temporary files
-      --recheck    update $as_me by reconfiguring in the same conditions
-      --file=FILE[:TEMPLATE]
-                   instantiate the configuration file FILE
-      --header=FILE[:TEMPLATE]
-                   instantiate the configuration header FILE
-
-Configuration files:
-$config_files
-
-Configuration headers:
-$config_headers
-
-Configuration commands:
-$config_commands
-
-Report bugs to <paboyle@ph.ed.ac.uk>."
-
-_ACEOF
-cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-ac_cs_config="`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`"
-ac_cs_version="\\
-Grid config.status 1.0
-configured by $0, generated by GNU Autoconf 2.69,
-  with options \\"\$ac_cs_config\\"
-
-Copyright (C) 2012 Free Software Foundation, Inc.
-This config.status script is free software; the Free Software Foundation
-gives unlimited permission to copy, distribute and modify it."
-
-ac_pwd='$ac_pwd'
-srcdir='$srcdir'
-INSTALL='$INSTALL'
-MKDIR_P='$MKDIR_P'
-AWK='$AWK'
-test -n "\$AWK" || AWK=awk
-_ACEOF
-
-cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-# The default lists apply if the user does not specify any file.
-ac_need_defaults=:
-while test $# != 0
-do
-  case $1 in
-  --*=?*)
-    ac_option=`expr "X$1" : 'X\([^=]*\)='`
-    ac_optarg=`expr "X$1" : 'X[^=]*=\(.*\)'`
-    ac_shift=:
-    ;;
-  --*=)
-    ac_option=`expr "X$1" : 'X\([^=]*\)='`
-    ac_optarg=
-    ac_shift=:
-    ;;
-  *)
-    ac_option=$1
-    ac_optarg=$2
-    ac_shift=shift
-    ;;
-  esac
-
-  case $ac_option in
-  # Handling of the options.
-  -recheck | --recheck | --rechec | --reche | --rech | --rec | --re | --r)
-    ac_cs_recheck=: ;;
-  --version | --versio | --versi | --vers | --ver | --ve | --v | -V )
-    $as_echo "$ac_cs_version"; exit ;;
-  --config | --confi | --conf | --con | --co | --c )
-    $as_echo "$ac_cs_config"; exit ;;
-  --debug | --debu | --deb | --de | --d | -d )
-    debug=: ;;
-  --file | --fil | --fi | --f )
-    $ac_shift
-    case $ac_optarg in
-    *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
-    '') as_fn_error $? "missing file argument" ;;
-    esac
-    as_fn_append CONFIG_FILES " '$ac_optarg'"
-    ac_need_defaults=false;;
-  --header | --heade | --head | --hea )
-    $ac_shift
-    case $ac_optarg in
-    *\'*) ac_optarg=`$as_echo "$ac_optarg" | sed "s/'/'\\\\\\\\''/g"` ;;
-    esac
-    as_fn_append CONFIG_HEADERS " '$ac_optarg'"
-    ac_need_defaults=false;;
-  --he | --h)
-    # Conflict between --help and --header
-    as_fn_error $? "ambiguous option: \`$1'
-Try \`$0 --help' for more information.";;
-  --help | --hel | -h )
-    $as_echo "$ac_cs_usage"; exit ;;
-  -q | -quiet | --quiet | --quie | --qui | --qu | --q \
-  | -silent | --silent | --silen | --sile | --sil | --si | --s)
-    ac_cs_silent=: ;;
-
-  # This is an error.
-  -*) as_fn_error $? "unrecognized option: \`$1'
-Try \`$0 --help' for more information." ;;
-
-  *) as_fn_append ac_config_targets " $1"
-     ac_need_defaults=false ;;
-
-  esac
-  shift
-done
-
-ac_configure_extra_args=
-
-if $ac_cs_silent; then
-  exec 6>/dev/null
-  ac_configure_extra_args="$ac_configure_extra_args --silent"
-fi
-
-_ACEOF
-cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-if \$ac_cs_recheck; then
-  set X $SHELL '$0' $ac_configure_args \$ac_configure_extra_args --no-create --no-recursion
-  shift
-  \$as_echo "running CONFIG_SHELL=$SHELL \$*" >&6
-  CONFIG_SHELL='$SHELL'
-  export CONFIG_SHELL
-  exec "\$@"
-fi
-
-_ACEOF
-cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-exec 5>>config.log
-{
-  echo
-  sed 'h;s/./-/g;s/^.../## /;s/...$/ ##/;p;x;p;x' <<_ASBOX
-## Running $as_me. ##
-_ASBOX
-  $as_echo "$ac_log"
-} >&5
-
-_ACEOF
-cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-#
-# INIT-COMMANDS
-#
-AMDEP_TRUE="$AMDEP_TRUE" ac_aux_dir="$ac_aux_dir"
-
-_ACEOF
-
-cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-
-# Handling of arguments.
-for ac_config_target in $ac_config_targets
-do
-  case $ac_config_target in
-    "lib/GridConfig.h") CONFIG_HEADERS="$CONFIG_HEADERS lib/GridConfig.h" ;;
-    "depfiles") CONFIG_COMMANDS="$CONFIG_COMMANDS depfiles" ;;
-    "docs/doxy.cfg") CONFIG_FILES="$CONFIG_FILES docs/doxy.cfg" ;;
-    "Makefile") CONFIG_FILES="$CONFIG_FILES Makefile" ;;
-    "lib/Makefile") CONFIG_FILES="$CONFIG_FILES lib/Makefile" ;;
-    "tests/Makefile") CONFIG_FILES="$CONFIG_FILES tests/Makefile" ;;
-    "benchmarks/Makefile") CONFIG_FILES="$CONFIG_FILES benchmarks/Makefile" ;;
-
-  *) as_fn_error $? "invalid argument: \`$ac_config_target'" "$LINENO" 5;;
-  esac
-done
-
-
-# If the user did not use the arguments to specify the items to instantiate,
-# then the envvar interface is used.  Set only those that are not.
-# We use the long form for the default assignment because of an extremely
-# bizarre bug on SunOS 4.1.3.
-if $ac_need_defaults; then
-  test "${CONFIG_FILES+set}" = set || CONFIG_FILES=$config_files
-  test "${CONFIG_HEADERS+set}" = set || CONFIG_HEADERS=$config_headers
-  test "${CONFIG_COMMANDS+set}" = set || CONFIG_COMMANDS=$config_commands
-fi
-
-# Have a temporary directory for convenience.  Make it in the build tree
-# simply because there is no reason against having it here, and in addition,
-# creating and moving files from /tmp can sometimes cause problems.
-# Hook for its removal unless debugging.
-# Note that there is a small window in which the directory will not be cleaned:
-# after its creation but before its name has been assigned to `$tmp'.
-$debug ||
-{
-  tmp= ac_tmp=
-  trap 'exit_status=$?
-  : "${ac_tmp:=$tmp}"
-  { test ! -d "$ac_tmp" || rm -fr "$ac_tmp"; } && exit $exit_status
-' 0
-  trap 'as_fn_exit 1' 1 2 13 15
-}
-# Create a (secure) tmp directory for tmp files.
-
-{
-  tmp=`(umask 077 && mktemp -d "./confXXXXXX") 2>/dev/null` &&
-  test -d "$tmp"
-}  ||
-{
-  tmp=./conf$$-$RANDOM
-  (umask 077 && mkdir "$tmp")
-} || as_fn_error $? "cannot create a temporary directory in ." "$LINENO" 5
-ac_tmp=$tmp
-
-# Set up the scripts for CONFIG_FILES section.
-# No need to generate them if there are no CONFIG_FILES.
-# This happens for instance with `./config.status config.h'.
-if test -n "$CONFIG_FILES"; then
-
-
-ac_cr=`echo X | tr X '\015'`
-# On cygwin, bash can eat \r inside `` if the user requested igncr.
-# But we know of no other shell where ac_cr would be empty at this
-# point, so we can use a bashism as a fallback.
-if test "x$ac_cr" = x; then
-  eval ac_cr=\$\'\\r\'
-fi
-ac_cs_awk_cr=`$AWK 'BEGIN { print "a\rb" }' </dev/null 2>/dev/null`
-if test "$ac_cs_awk_cr" = "a${ac_cr}b"; then
-  ac_cs_awk_cr='\\r'
-else
-  ac_cs_awk_cr=$ac_cr
-fi
-
-echo 'BEGIN {' >"$ac_tmp/subs1.awk" &&
-_ACEOF
-
-
-{
-  echo "cat >conf$$subs.awk <<_ACEOF" &&
-  echo "$ac_subst_vars" | sed 's/.*/&!$&$ac_delim/' &&
-  echo "_ACEOF"
-} >conf$$subs.sh ||
-  as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
-ac_delim_num=`echo "$ac_subst_vars" | grep -c '^'`
-ac_delim='%!_!# '
-for ac_last_try in false false false false false :; do
-  . ./conf$$subs.sh ||
-    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
-
-  ac_delim_n=`sed -n "s/.*$ac_delim\$/X/p" conf$$subs.awk | grep -c X`
-  if test $ac_delim_n = $ac_delim_num; then
-    break
-  elif $ac_last_try; then
-    as_fn_error $? "could not make $CONFIG_STATUS" "$LINENO" 5
-  else
-    ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
-  fi
-done
-rm -f conf$$subs.sh
-
-cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-cat >>"\$ac_tmp/subs1.awk" <<\\_ACAWK &&
-_ACEOF
-sed -n '
-h
-s/^/S["/; s/!.*/"]=/
-p
-g
-s/^[^!]*!//
-:repl
-t repl
-s/'"$ac_delim"'$//
-t delim
-:nl
-h
-s/\(.\{148\}\)..*/\1/
-t more1
-s/["\\]/\\&/g; s/^/"/; s/$/\\n"\\/
-p
-n
-b repl
-:more1
-s/["\\]/\\&/g; s/^/"/; s/$/"\\/
-p
-g
-s/.\{148\}//
-t nl
-:delim
-h
-s/\(.\{148\}\)..*/\1/
-t more2
-s/["\\]/\\&/g; s/^/"/; s/$/"/
-p
-b
-:more2
-s/["\\]/\\&/g; s/^/"/; s/$/"\\/
-p
-g
-s/.\{148\}//
-t delim
-' <conf$$subs.awk | sed '
-/^[^""]/{
-  N
-  s/\n//
-}
-' >>$CONFIG_STATUS || ac_write_fail=1
-rm -f conf$$subs.awk
-cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-_ACAWK
-cat >>"\$ac_tmp/subs1.awk" <<_ACAWK &&
-  for (key in S) S_is_set[key] = 1
-  FS = ""
-
-}
-{
-  line = $ 0
-  nfields = split(line, field, "@")
-  substed = 0
-  len = length(field[1])
-  for (i = 2; i < nfields; i++) {
-    key = field[i]
-    keylen = length(key)
-    if (S_is_set[key]) {
-      value = S[key]
-      line = substr(line, 1, len) "" value "" substr(line, len + keylen + 3)
-      len += length(value) + length(field[++i])
-      substed = 1
-    } else
-      len += 1 + keylen
-  }
-
-  print line
-}
-
-_ACAWK
-_ACEOF
-cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-if sed "s/$ac_cr//" < /dev/null > /dev/null 2>&1; then
-  sed "s/$ac_cr\$//; s/$ac_cr/$ac_cs_awk_cr/g"
-else
-  cat
-fi < "$ac_tmp/subs1.awk" > "$ac_tmp/subs.awk" \
-  || as_fn_error $? "could not setup config files machinery" "$LINENO" 5
-_ACEOF
-
-# VPATH may cause trouble with some makes, so we remove sole $(srcdir),
-# ${srcdir} and @srcdir@ entries from VPATH if srcdir is ".", strip leading and
-# trailing colons and then remove the whole line if VPATH becomes empty
-# (actually we leave an empty line to preserve line numbers).
-if test "x$srcdir" = x.; then
-  ac_vpsub='/^[	 ]*VPATH[	 ]*=[	 ]*/{
-h
-s///
-s/^/:/
-s/[	 ]*$/:/
-s/:\$(srcdir):/:/g
-s/:\${srcdir}:/:/g
-s/:@srcdir@:/:/g
-s/^:*//
-s/:*$//
-x
-s/\(=[	 ]*\).*/\1/
-G
-s/\n//
-s/^[^=]*=[	 ]*$//
-}'
-fi
-
-cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-fi # test -n "$CONFIG_FILES"
-
-# Set up the scripts for CONFIG_HEADERS section.
-# No need to generate them if there are no CONFIG_HEADERS.
-# This happens for instance with `./config.status Makefile'.
-if test -n "$CONFIG_HEADERS"; then
-cat >"$ac_tmp/defines.awk" <<\_ACAWK ||
-BEGIN {
-_ACEOF
-
-# Transform confdefs.h into an awk script `defines.awk', embedded as
-# here-document in config.status, that substitutes the proper values into
-# config.h.in to produce config.h.
-
-# Create a delimiter string that does not exist in confdefs.h, to ease
-# handling of long lines.
-ac_delim='%!_!# '
-for ac_last_try in false false :; do
-  ac_tt=`sed -n "/$ac_delim/p" confdefs.h`
-  if test -z "$ac_tt"; then
-    break
-  elif $ac_last_try; then
-    as_fn_error $? "could not make $CONFIG_HEADERS" "$LINENO" 5
-  else
-    ac_delim="$ac_delim!$ac_delim _$ac_delim!! "
-  fi
-done
-
-# For the awk script, D is an array of macro values keyed by name,
-# likewise P contains macro parameters if any.  Preserve backslash
-# newline sequences.
-
-ac_word_re=[_$as_cr_Letters][_$as_cr_alnum]*
-sed -n '
-s/.\{148\}/&'"$ac_delim"'/g
-t rset
-:rset
-s/^[	 ]*#[	 ]*define[	 ][	 ]*/ /
-t def
-d
-:def
-s/\\$//
-t bsnl
-s/["\\]/\\&/g
-s/^ \('"$ac_word_re"'\)\(([^()]*)\)[	 ]*\(.*\)/P["\1"]="\2"\
-D["\1"]=" \3"/p
-s/^ \('"$ac_word_re"'\)[	 ]*\(.*\)/D["\1"]=" \2"/p
-d
-:bsnl
-s/["\\]/\\&/g
-s/^ \('"$ac_word_re"'\)\(([^()]*)\)[	 ]*\(.*\)/P["\1"]="\2"\
-D["\1"]=" \3\\\\\\n"\\/p
-t cont
-s/^ \('"$ac_word_re"'\)[	 ]*\(.*\)/D["\1"]=" \2\\\\\\n"\\/p
-t cont
-d
-:cont
-n
-s/.\{148\}/&'"$ac_delim"'/g
-t clear
-:clear
-s/\\$//
-t bsnlc
-s/["\\]/\\&/g; s/^/"/; s/$/"/p
-d
-:bsnlc
-s/["\\]/\\&/g; s/^/"/; s/$/\\\\\\n"\\/p
-b cont
-' <confdefs.h | sed '
-s/'"$ac_delim"'/"\\\
-"/g' >>$CONFIG_STATUS || ac_write_fail=1
-
-cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-  for (key in D) D_is_set[key] = 1
-  FS = ""
-}
-/^[\t ]*#[\t ]*(define|undef)[\t ]+$ac_word_re([\t (]|\$)/ {
-  line = \$ 0
-  split(line, arg, " ")
-  if (arg[1] == "#") {
-    defundef = arg[2]
-    mac1 = arg[3]
-  } else {
-    defundef = substr(arg[1], 2)
-    mac1 = arg[2]
-  }
-  split(mac1, mac2, "(") #)
-  macro = mac2[1]
-  prefix = substr(line, 1, index(line, defundef) - 1)
-  if (D_is_set[macro]) {
-    # Preserve the white space surrounding the "#".
-    print prefix "define", macro P[macro] D[macro]
-    next
-  } else {
-    # Replace #undef with comments.  This is necessary, for example,
-    # in the case of _POSIX_SOURCE, which is predefined and required
-    # on some systems where configure will not decide to define it.
-    if (defundef == "undef") {
-      print "/*", prefix defundef, macro, "*/"
-      next
-    }
-  }
-}
-{ print }
-_ACAWK
-_ACEOF
-cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-  as_fn_error $? "could not setup config headers machinery" "$LINENO" 5
-fi # test -n "$CONFIG_HEADERS"
-
-
-eval set X "  :F $CONFIG_FILES  :H $CONFIG_HEADERS    :C $CONFIG_COMMANDS"
-shift
-for ac_tag
-do
-  case $ac_tag in
-  :[FHLC]) ac_mode=$ac_tag; continue;;
-  esac
-  case $ac_mode$ac_tag in
-  :[FHL]*:*);;
-  :L* | :C*:*) as_fn_error $? "invalid tag \`$ac_tag'" "$LINENO" 5;;
-  :[FH]-) ac_tag=-:-;;
-  :[FH]*) ac_tag=$ac_tag:$ac_tag.in;;
-  esac
-  ac_save_IFS=$IFS
-  IFS=:
-  set x $ac_tag
-  IFS=$ac_save_IFS
-  shift
-  ac_file=$1
-  shift
-
-  case $ac_mode in
-  :L) ac_source=$1;;
-  :[FH])
-    ac_file_inputs=
-    for ac_f
-    do
-      case $ac_f in
-      -) ac_f="$ac_tmp/stdin";;
-      *) # Look for the file first in the build tree, then in the source tree
-	 # (if the path is not absolute).  The absolute path cannot be DOS-style,
-	 # because $ac_f cannot contain `:'.
-	 test -f "$ac_f" ||
-	   case $ac_f in
-	   [\\/$]*) false;;
-	   *) test -f "$srcdir/$ac_f" && ac_f="$srcdir/$ac_f";;
-	   esac ||
-	   as_fn_error 1 "cannot find input file: \`$ac_f'" "$LINENO" 5;;
-      esac
-      case $ac_f in *\'*) ac_f=`$as_echo "$ac_f" | sed "s/'/'\\\\\\\\''/g"`;; esac
-      as_fn_append ac_file_inputs " '$ac_f'"
-    done
-
-    # Let's still pretend it is `configure' which instantiates (i.e., don't
-    # use $as_me), people would be surprised to read:
-    #    /* config.h.  Generated by config.status.  */
-    configure_input='Generated from '`
-	  $as_echo "$*" | sed 's|^[^:]*/||;s|:[^:]*/|, |g'
-	`' by configure.'
-    if test x"$ac_file" != x-; then
-      configure_input="$ac_file.  $configure_input"
-      { $as_echo "$as_me:${as_lineno-$LINENO}: creating $ac_file" >&5
-$as_echo "$as_me: creating $ac_file" >&6;}
-    fi
-    # Neutralize special characters interpreted by sed in replacement strings.
-    case $configure_input in #(
-    *\&* | *\|* | *\\* )
-       ac_sed_conf_input=`$as_echo "$configure_input" |
-       sed 's/[\\\\&|]/\\\\&/g'`;; #(
-    *) ac_sed_conf_input=$configure_input;;
-    esac
-
-    case $ac_tag in
-    *:-:* | *:-) cat >"$ac_tmp/stdin" \
-      || as_fn_error $? "could not create $ac_file" "$LINENO" 5 ;;
-    esac
-    ;;
-  esac
-
-  ac_dir=`$as_dirname -- "$ac_file" ||
-$as_expr X"$ac_file" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$ac_file" : 'X\(//\)[^/]' \| \
-	 X"$ac_file" : 'X\(//\)$' \| \
-	 X"$ac_file" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$ac_file" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-  as_dir="$ac_dir"; as_fn_mkdir_p
-  ac_builddir=.
-
-case "$ac_dir" in
-.) ac_dir_suffix= ac_top_builddir_sub=. ac_top_build_prefix= ;;
-*)
-  ac_dir_suffix=/`$as_echo "$ac_dir" | sed 's|^\.[\\/]||'`
-  # A ".." for each directory in $ac_dir_suffix.
-  ac_top_builddir_sub=`$as_echo "$ac_dir_suffix" | sed 's|/[^\\/]*|/..|g;s|/||'`
-  case $ac_top_builddir_sub in
-  "") ac_top_builddir_sub=. ac_top_build_prefix= ;;
-  *)  ac_top_build_prefix=$ac_top_builddir_sub/ ;;
-  esac ;;
-esac
-ac_abs_top_builddir=$ac_pwd
-ac_abs_builddir=$ac_pwd$ac_dir_suffix
-# for backward compatibility:
-ac_top_builddir=$ac_top_build_prefix
-
-case $srcdir in
-  .)  # We are building in place.
-    ac_srcdir=.
-    ac_top_srcdir=$ac_top_builddir_sub
-    ac_abs_top_srcdir=$ac_pwd ;;
-  [\\/]* | ?:[\\/]* )  # Absolute name.
-    ac_srcdir=$srcdir$ac_dir_suffix;
-    ac_top_srcdir=$srcdir
-    ac_abs_top_srcdir=$srcdir ;;
-  *) # Relative name.
-    ac_srcdir=$ac_top_build_prefix$srcdir$ac_dir_suffix
-    ac_top_srcdir=$ac_top_build_prefix$srcdir
-    ac_abs_top_srcdir=$ac_pwd/$srcdir ;;
-esac
-ac_abs_srcdir=$ac_abs_top_srcdir$ac_dir_suffix
-
-
-  case $ac_mode in
-  :F)
-  #
-  # CONFIG_FILE
-  #
-
-  case $INSTALL in
-  [\\/$]* | ?:[\\/]* ) ac_INSTALL=$INSTALL ;;
-  *) ac_INSTALL=$ac_top_build_prefix$INSTALL ;;
-  esac
-  ac_MKDIR_P=$MKDIR_P
-  case $MKDIR_P in
-  [\\/$]* | ?:[\\/]* ) ;;
-  */*) ac_MKDIR_P=$ac_top_build_prefix$MKDIR_P ;;
-  esac
-_ACEOF
-
-cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-# If the template does not know about datarootdir, expand it.
-# FIXME: This hack should be removed a few years after 2.60.
-ac_datarootdir_hack=; ac_datarootdir_seen=
-ac_sed_dataroot='
-/datarootdir/ {
-  p
-  q
-}
-/@datadir@/p
-/@docdir@/p
-/@infodir@/p
-/@localedir@/p
-/@mandir@/p'
-case `eval "sed -n \"\$ac_sed_dataroot\" $ac_file_inputs"` in
-*datarootdir*) ac_datarootdir_seen=yes;;
-*@datadir@*|*@docdir@*|*@infodir@*|*@localedir@*|*@mandir@*)
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&5
-$as_echo "$as_me: WARNING: $ac_file_inputs seems to ignore the --datarootdir setting" >&2;}
-_ACEOF
-cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-  ac_datarootdir_hack='
-  s&@datadir@&$datadir&g
-  s&@docdir@&$docdir&g
-  s&@infodir@&$infodir&g
-  s&@localedir@&$localedir&g
-  s&@mandir@&$mandir&g
-  s&\\\${datarootdir}&$datarootdir&g' ;;
-esac
-_ACEOF
-
-# Neutralize VPATH when `$srcdir' = `.'.
-# Shell code in configure.ac might set extrasub.
-# FIXME: do we really want to maintain this feature?
-cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
-ac_sed_extra="$ac_vpsub
-$extrasub
-_ACEOF
-cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
-:t
-/@[a-zA-Z_][a-zA-Z_0-9]*@/!b
-s|@configure_input@|$ac_sed_conf_input|;t t
-s&@top_builddir@&$ac_top_builddir_sub&;t t
-s&@top_build_prefix@&$ac_top_build_prefix&;t t
-s&@srcdir@&$ac_srcdir&;t t
-s&@abs_srcdir@&$ac_abs_srcdir&;t t
-s&@top_srcdir@&$ac_top_srcdir&;t t
-s&@abs_top_srcdir@&$ac_abs_top_srcdir&;t t
-s&@builddir@&$ac_builddir&;t t
-s&@abs_builddir@&$ac_abs_builddir&;t t
-s&@abs_top_builddir@&$ac_abs_top_builddir&;t t
-s&@INSTALL@&$ac_INSTALL&;t t
-s&@MKDIR_P@&$ac_MKDIR_P&;t t
-$ac_datarootdir_hack
-"
-eval sed \"\$ac_sed_extra\" "$ac_file_inputs" | $AWK -f "$ac_tmp/subs.awk" \
-  >$ac_tmp/out || as_fn_error $? "could not create $ac_file" "$LINENO" 5
-
-test -z "$ac_datarootdir_hack$ac_datarootdir_seen" &&
-  { ac_out=`sed -n '/\${datarootdir}/p' "$ac_tmp/out"`; test -n "$ac_out"; } &&
-  { ac_out=`sed -n '/^[	 ]*datarootdir[	 ]*:*=/p' \
-      "$ac_tmp/out"`; test -z "$ac_out"; } &&
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: $ac_file contains a reference to the variable \`datarootdir'
-which seems to be undefined.  Please make sure it is defined" >&5
-$as_echo "$as_me: WARNING: $ac_file contains a reference to the variable \`datarootdir'
-which seems to be undefined.  Please make sure it is defined" >&2;}
-
-  rm -f "$ac_tmp/stdin"
-  case $ac_file in
-  -) cat "$ac_tmp/out" && rm -f "$ac_tmp/out";;
-  *) rm -f "$ac_file" && mv "$ac_tmp/out" "$ac_file";;
-  esac \
-  || as_fn_error $? "could not create $ac_file" "$LINENO" 5
- ;;
-  :H)
-  #
-  # CONFIG_HEADER
-  #
-  if test x"$ac_file" != x-; then
-    {
-      $as_echo "/* $configure_input  */" \
-      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs"
-    } >"$ac_tmp/config.h" \
-      || as_fn_error $? "could not create $ac_file" "$LINENO" 5
-    if diff "$ac_file" "$ac_tmp/config.h" >/dev/null 2>&1; then
-      { $as_echo "$as_me:${as_lineno-$LINENO}: $ac_file is unchanged" >&5
-$as_echo "$as_me: $ac_file is unchanged" >&6;}
-    else
-      rm -f "$ac_file"
-      mv "$ac_tmp/config.h" "$ac_file" \
-	|| as_fn_error $? "could not create $ac_file" "$LINENO" 5
-    fi
-  else
-    $as_echo "/* $configure_input  */" \
-      && eval '$AWK -f "$ac_tmp/defines.awk"' "$ac_file_inputs" \
-      || as_fn_error $? "could not create -" "$LINENO" 5
-  fi
-# Compute "$ac_file"'s index in $config_headers.
-_am_arg="$ac_file"
-_am_stamp_count=1
-for _am_header in $config_headers :; do
-  case $_am_header in
-    $_am_arg | $_am_arg:* )
-      break ;;
-    * )
-      _am_stamp_count=`expr $_am_stamp_count + 1` ;;
-  esac
-done
-echo "timestamp for $_am_arg" >`$as_dirname -- "$_am_arg" ||
-$as_expr X"$_am_arg" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$_am_arg" : 'X\(//\)[^/]' \| \
-	 X"$_am_arg" : 'X\(//\)$' \| \
-	 X"$_am_arg" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$_am_arg" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`/stamp-h$_am_stamp_count
- ;;
-
-  :C)  { $as_echo "$as_me:${as_lineno-$LINENO}: executing $ac_file commands" >&5
-$as_echo "$as_me: executing $ac_file commands" >&6;}
- ;;
-  esac
-
-
-  case $ac_file$ac_mode in
-    "depfiles":C) test x"$AMDEP_TRUE" != x"" || {
-  # Older Autoconf quotes --file arguments for eval, but not when files
-  # are listed without --file.  Let's play safe and only enable the eval
-  # if we detect the quoting.
-  case $CONFIG_FILES in
-  *\'*) eval set x "$CONFIG_FILES" ;;
-  *)   set x $CONFIG_FILES ;;
-  esac
-  shift
-  for mf
-  do
-    # Strip MF so we end up with the name of the file.
-    mf=`echo "$mf" | sed -e 's/:.*$//'`
-    # Check whether this is an Automake generated Makefile or not.
-    # We used to match only the files named 'Makefile.in', but
-    # some people rename them; so instead we look at the file content.
-    # Grep'ing the first line is not enough: some people post-process
-    # each Makefile.in and add a new line on top of each file to say so.
-    # Grep'ing the whole file is not good either: AIX grep has a line
-    # limit of 2048, but all sed's we know have understand at least 4000.
-    if sed -n 's,^#.*generated by automake.*,X,p' "$mf" | grep X >/dev/null 2>&1; then
-      dirpart=`$as_dirname -- "$mf" ||
-$as_expr X"$mf" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$mf" : 'X\(//\)[^/]' \| \
-	 X"$mf" : 'X\(//\)$' \| \
-	 X"$mf" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$mf" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-    else
-      continue
-    fi
-    # Extract the definition of DEPDIR, am__include, and am__quote
-    # from the Makefile without running 'make'.
-    DEPDIR=`sed -n 's/^DEPDIR = //p' < "$mf"`
-    test -z "$DEPDIR" && continue
-    am__include=`sed -n 's/^am__include = //p' < "$mf"`
-    test -z "$am__include" && continue
-    am__quote=`sed -n 's/^am__quote = //p' < "$mf"`
-    # Find all dependency output files, they are included files with
-    # $(DEPDIR) in their names.  We invoke sed twice because it is the
-    # simplest approach to changing $(DEPDIR) to its actual value in the
-    # expansion.
-    for file in `sed -n "
-      s/^$am__include $am__quote\(.*(DEPDIR).*\)$am__quote"'$/\1/p' <"$mf" | \
-	 sed -e 's/\$(DEPDIR)/'"$DEPDIR"'/g'`; do
-      # Make sure the directory exists.
-      test -f "$dirpart/$file" && continue
-      fdir=`$as_dirname -- "$file" ||
-$as_expr X"$file" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
-	 X"$file" : 'X\(//\)[^/]' \| \
-	 X"$file" : 'X\(//\)$' \| \
-	 X"$file" : 'X\(/\)' \| . 2>/dev/null ||
-$as_echo X"$file" |
-    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)[^/].*/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\/\)$/{
-	    s//\1/
-	    q
-	  }
-	  /^X\(\/\).*/{
-	    s//\1/
-	    q
-	  }
-	  s/.*/./; q'`
-      as_dir=$dirpart/$fdir; as_fn_mkdir_p
-      # echo "creating $dirpart/$file"
-      echo '# dummy' > "$dirpart/$file"
-    done
-  done
-}
- ;;
-
-  esac
-done # for ac_tag
-
-
-as_fn_exit 0
-_ACEOF
-ac_clean_files=$ac_clean_files_save
-
-test $ac_write_fail = 0 ||
-  as_fn_error $? "write failure creating $CONFIG_STATUS" "$LINENO" 5
-
-
-# configure is writing to config.log, and then calls config.status.
-# config.status does its own redirection, appending to config.log.
-# Unfortunately, on DOS this fails, as config.log is still kept open
-# by configure, so config.status won't be able to write to it; its
-# output is simply discarded.  So we exec the FD to /dev/null,
-# effectively closing config.log, so it can be properly (re)opened and
-# appended to by config.status.  When coming back to configure, we
-# need to make the FD available again.
-if test "$no_create" != yes; then
-  ac_cs_success=:
-  ac_config_status_args=
-  test "$silent" = yes &&
-    ac_config_status_args="$ac_config_status_args --quiet"
-  exec 5>/dev/null
-  $SHELL $CONFIG_STATUS $ac_config_status_args || ac_cs_success=false
-  exec 5>>config.log
-  # Use ||, not &&, to avoid exiting from the if with $? = 1, which
-  # would make configure fail if this is the last instruction.
-  $ac_cs_success || as_fn_exit 1
-fi
-if test -n "$ac_unrecognized_opts" && test "$enable_option_checking" != no; then
-  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: unrecognized options: $ac_unrecognized_opts" >&5
-$as_echo "$as_me: WARNING: unrecognized options: $ac_unrecognized_opts" >&2;}
-fi
-
-
-
-echo "
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-Summary of configuration for $PACKAGE v$VERSION
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-
-The following features are enabled:
-
-- architecture (build)          : $build_cpu
-- os (build)                    : $build_os
-- architecture (target)         : $target_cpu
-- os (target)                   : $target_os
-- build DOXYGEN documentation   : `if test "x$enable_doc" = xyes; then echo yes; else echo no; fi`
-- graphs and diagrams           : `if test "x$enable_dot" = xyes; then echo yes; else echo no; fi`
-- Supported SIMD flags          : $SIMD_FLAGS
-----------------------------------------------------------
-- enabled simd support          : ${ac_SIMD}   (supported: $supported )
-- communications type           : ${ac_COMMS}
-
-
-"

From 73c4a1dac9acfadb72a1326e9ba2661cba5162c1 Mon Sep 17 00:00:00 2001
From: Antonin Portelli <antonin.portelli@me.com>
Date: Fri, 17 Jul 2015 11:15:57 +0900
Subject: [PATCH 390/429] fix of AX_GCC_X86_AVX_XGETBV macro

---
 m4/ax_gcc_x86_avx_xgetbv.m4 | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/m4/ax_gcc_x86_avx_xgetbv.m4 b/m4/ax_gcc_x86_avx_xgetbv.m4
index 2da9a6ab..c8c1c504 100644
--- a/m4/ax_gcc_x86_avx_xgetbv.m4
+++ b/m4/ax_gcc_x86_avx_xgetbv.m4
@@ -65,7 +65,7 @@ AC_CACHE_CHECK(for x86-AVX xgetbv $1 output, ax_cv_gcc_x86_avx_xgetbv_$1,
      FILE *f;
       /* Opcodes for xgetbv */
       __asm__(".byte 0x0f, 0x01, 0xd0"
-        : "=a" (eax), "=d" (edx), 
+        : "=a" (eax), "=d" (edx) 
         : "c" (op));
      f = fopen("conftest_xgetbv", "w"); if (!f) return 1;
      fprintf(f, "%x:%x\n", eax, edx);

From 7343a95772b9cd96fd7eced914e0066828bcd15b Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Tue, 21 Jul 2015 11:52:15 +0900
Subject: [PATCH 391/429] More NEON functionalities

---
 configure                                   |  6 +-
 configure.ac                                |  8 +-
 lib/GridConfig.h.in                         |  4 +-
 lib/qcd/action/gauge/WilsonGaugeAction.h    |  3 +-
 lib/qcd/utils/WilsonLoops.h                 |  1 -
 lib/simd/Grid_neon.h                        | 93 ++++++++++-----------
 lib/simd/Grid_vector_types.h                |  4 +-
 scripts/arm_configure.experimental_cortex57 |  3 +
 tests/Test_hmc_WilsonGauge.cc               |  4 +-
 tests/Test_simd.cc                          | 49 +++++------
 10 files changed, 88 insertions(+), 87 deletions(-)
 create mode 100644 scripts/arm_configure.experimental_cortex57

diff --git a/configure b/configure
index 935f1018..cc11a210 100755
--- a/configure
+++ b/configure
@@ -6712,10 +6712,10 @@ $as_echo "#define AVX512 1" >>confdefs.h
 
        supported="cross compilation"
      ;;
-     NEONv7)
-       echo Configuring for experimental ARMv7 support
+     NEONv8)
+       echo Configuring for experimental ARMv8a support
 
-$as_echo "#define NEONv7 1" >>confdefs.h
+$as_echo "#define NEONv8 1" >>confdefs.h
 
        supported="cross compilation"
      ;;
diff --git a/configure.ac b/configure.ac
index 5ef6fa8d..be01228c 100644
--- a/configure.ac
+++ b/configure.ac
@@ -3,7 +3,7 @@
 #
 # Project Grid package  
 # 
-# Time-stamp: <2015-06-09 15:26:39 neo>
+# Time-stamp: <2015-07-10 17:46:21 neo>
 
 AC_PREREQ([2.63])
 AC_INIT([Grid], [1.0], [paboyle@ph.ed.ac.uk])
@@ -106,9 +106,9 @@ case ${ac_SIMD} in
        AC_DEFINE([AVX512],[1],[AVX512 Intrinsics for Knights Corner] )
        supported="cross compilation"
      ;;
-     NEONv7)
-       echo Configuring for experimental ARMv7 support 
-       AC_DEFINE([NEONv7],[1],[NEON ARMv7 Experimental support ] )
+     NEONv8)
+       echo Configuring for experimental ARMv8a support 
+       AC_DEFINE([NEONv8],[1],[NEON ARMv8 Experimental support ] )
        supported="cross compilation"
      ;;
      DEBUG)
diff --git a/lib/GridConfig.h.in b/lib/GridConfig.h.in
index 1ec0ecf5..136d7bfb 100644
--- a/lib/GridConfig.h.in
+++ b/lib/GridConfig.h.in
@@ -119,8 +119,8 @@
 /* Define to 1 if you have the <unistd.h> header file. */
 #undef HAVE_UNISTD_H
 
-/* NEON ARMv7 Experimental support */
-#undef NEONv7
+/* NEON ARMv8 Experimental support */
+#undef NEONv8
 
 /* Name of package */
 #undef PACKAGE
diff --git a/lib/qcd/action/gauge/WilsonGaugeAction.h b/lib/qcd/action/gauge/WilsonGaugeAction.h
index aaedf2f9..ad1aecab 100644
--- a/lib/qcd/action/gauge/WilsonGaugeAction.h
+++ b/lib/qcd/action/gauge/WilsonGaugeAction.h
@@ -7,7 +7,7 @@ namespace Grid{
     ////////////////////////////////////////////////////////////////////////
     // Wilson Gauge Action .. should I template the Nc etc..
     ////////////////////////////////////////////////////////////////////////
-    template<class GaugeField,class MatrixField>
+    template<class GaugeField, class MatrixField>
       class WilsonGaugeAction : public Action<GaugeField> {
     private:
       RealD beta;
@@ -23,7 +23,6 @@ namespace Grid{
 	return beta*(1.0 -plaq)*(Nd*(Nd-1.0))*vol*0.5;
       };
       virtual void deriv(const GaugeField &U,GaugeField & dSdU) {
-
 	//not optimal implementation FIXME
 	//extend Ta to include Lorentz indexes
 	RealD factor = 0.5*beta/RealD(Nc);
diff --git a/lib/qcd/utils/WilsonLoops.h b/lib/qcd/utils/WilsonLoops.h
index 9b8e8724..4565c8e8 100644
--- a/lib/qcd/utils/WilsonLoops.h
+++ b/lib/qcd/utils/WilsonLoops.h
@@ -7,7 +7,6 @@ namespace QCD {
 template<class GaugeMat,class GaugeLorentz>
 class WilsonLoops {
 public:
-
   //////////////////////////////////////////////////
   // directed plaquette oriented in mu,nu plane
   //////////////////////////////////////////////////
diff --git a/lib/simd/Grid_neon.h b/lib/simd/Grid_neon.h
index 73660b40..3d36ea95 100644
--- a/lib/simd/Grid_neon.h
+++ b/lib/simd/Grid_neon.h
@@ -1,14 +1,16 @@
 //----------------------------------------------------------------------
 /*! @file Grid_sse4.h
-  @brief Optimization libraries for NEON (ARM) instructions set ARMv7
+  @brief Optimization libraries for NEON (ARM) instructions set ARMv8
 
   Experimental - Using intrinsics - DEVELOPING! 
 */
-// Time-stamp: <2015-06-09 15:25:40 neo>
+// Time-stamp: <2015-07-10 17:45:09 neo>
 //----------------------------------------------------------------------
 
 #include <arm_neon.h>
 
+// ARMv8 supports double precision
+
 namespace Optimization {
 
   template<class vtype>
@@ -22,50 +24,47 @@ namespace Optimization {
     float f[4];
   };
   union u128d {
-    float32x4_t v;
-    float f[4];
+    float64x2_t v;
+    double f[4];
   };
   
   struct Vsplat{
     //Complex float
     inline float32x4_t operator()(float a, float b){
-      float32x4_t foo;
-      return foo;
+      float tmp[4]={a,b,a,b};
+      return vld1q_f32(tmp);
     }
     // Real float
     inline float32x4_t operator()(float a){
-      float32x4_t foo;
-      return foo;
+      return vld1q_dup_f32(&a);
     }
     //Complex double
     inline float32x4_t operator()(double a, double b){
-      float32x4_t foo;
-      return foo;
+      float tmp[4]={(float)a,(float)b,(float)a,(float)b};
+      return vld1q_f32(tmp);
     }
     //Real double
     inline float32x4_t operator()(double a){
-      float32x4_t foo;
-      return foo;
+      return vld1q_dup_f32(&a);
     }
     //Integer
     inline uint32x4_t operator()(Integer a){
-      uint32x4_t foo;
-      return foo;
+      return vld1q_dup_u32(&a);
     }
   };
 
   struct Vstore{
     //Float 
     inline void operator()(float32x4_t a, float* F){
-      
+      vst1q_f32(F, a);
     }
     //Double
     inline void operator()(float32x4_t a, double* D){
-      
+      vst1q_f32((float*)D, a);
     }
     //Integer
     inline void operator()(uint32x4_t a, Integer* I){
-     
+      vst1q_u32(I, a);
     }
 
   };
@@ -130,36 +129,30 @@ namespace Optimization {
   struct Sum{
     //Complex/Real float
     inline float32x4_t operator()(float32x4_t a, float32x4_t b){
-      float32x4_t foo;
-      return foo;
+      return vaddq_f32(a,b);
     }
     //Complex/Real double
-    //inline float32x4_t operator()(float32x4_t a, float32x4_t b){
-    //  float32x4_t foo;
-    //  return foo;
-    //}
+    inline float64x2_t operator()(float64x2_t a, float64x2_t b){
+      return vaddq_f64(a,b);
+    }
     //Integer
     inline uint32x4_t operator()(uint32x4_t a, uint32x4_t b){
-      uint32x4_t foo;
-      return foo;
+      return vaddq_u32(a,b);
     }
   };
 
   struct Sub{
     //Complex/Real float
     inline float32x4_t operator()(float32x4_t a, float32x4_t b){
-      float32x4_t foo;
-      return foo;
+      return vsubq_f32(a,b);
     }
     //Complex/Real double
-    //inline float32x4_t operator()(float32x4_t a, float32x4_t b){
-    //  float32x4_t foo;
-    //  return foo;
-    //}
+    inline float64x2_t operator()(float64x2_t a, float64x2_t b){
+      return vsubq_f64(a,b);
+    }
     //Integer
     inline uint32x4_t operator()(uint32x4_t a, uint32x4_t b){
-      uint32x4_t foo;
-      return foo;
+      return vsubq_u32(a,b);
     }
   };
 
@@ -170,24 +163,24 @@ namespace Optimization {
       return foo;
     }
     // Complex double
-    //inline float32x4_t operator()(float32x4_t a, float32x4_t b){
-    //  float32x4_t foo;
-    //  return foo;
-    //}
+    inline float64x2_t operator()(float64x2_t a, float64x2_t b){
+      float32x4_t foo;
+      return foo;
+    }
   };
 
   struct Mult{
     // Real float
     inline float32x4_t operator()(float32x4_t a, float32x4_t b){
-      return a;
+      return vmulq_f32(a,b);
     }
     // Real double
-    //inline float32x4_t operator()(float32x4_t a, float32x4_t b){
-    //  return 0;
-    //}
+    inline float64x2_t operator()(float64x2_t a, float64x2_t b){
+      return vmulq_f64(a,b);
+    }
     // Integer
     inline uint32x4_t operator()(uint32x4_t a, uint32x4_t b){
-      return a;
+      return vmulq_u32(a,b);
     }
   };
 
@@ -219,6 +212,7 @@ namespace Optimization {
   struct TimesI{
     //Complex single
     inline float32x4_t operator()(float32x4_t in, float32x4_t ret){
+      //need shuffle
       return in;
     }
     //Complex double
@@ -242,20 +236,25 @@ namespace Optimization {
   //Real float Reduce
   template<>
   inline Grid::RealF Reduce<Grid::RealF, float32x4_t>::operator()(float32x4_t in){
-    return 0;
+    float32x2_t high = vget_high_f32(in);
+    float32x2_t low = vget_low_f32(in);
+    float32x2_t tmp = vadd_f32(low, high);
+    float32x2_t sum = vpadd_f32(tmp, tmp);
+    return vget_lane_f32(sum,0);
   }
   
   
   //Complex double Reduce
   template<>
-  inline Grid::ComplexD Reduce<Grid::ComplexD, float32x4_t>::operator()(float32x4_t in){
+  inline Grid::ComplexD Reduce<Grid::ComplexD, float64x2_t>::operator()(float64x2_t in){
     return 0;
   }
   
   //Real double Reduce
   template<>
-  inline Grid::RealD Reduce<Grid::RealD, float32x4_t>::operator()(float32x4_t in){
-    return 0;
+  inline Grid::RealD Reduce<Grid::RealD, float64x2_t>::operator()(float64x2_t in){
+    float64x2_t sum = vpaddq_f64(in, in);
+    return vgetq_lane_f64(sum,0);
   }
 
   //Integer Reduce
@@ -272,7 +271,7 @@ namespace Optimization {
 namespace Grid {
 
   typedef float32x4_t  SIMD_Ftype; // Single precision type
-  typedef float32x4_t  SIMD_Dtype; // Double precision type - no double on ARMv7
+  typedef float64x2_t  SIMD_Dtype; // Double precision type
   typedef uint32x4_t   SIMD_Itype; // Integer type
 
   inline void v_prefetch0(int size, const char *ptr){};  // prefetch utilities
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index 034d5314..4a0ea33b 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -2,7 +2,7 @@
 /*! @file Grid_vector_types.h
   @brief Defines templated class Grid_simd to deal with inner vector types
 */
-// Time-stamp: <2015-06-09 15:00:47 neo>
+// Time-stamp: <2015-07-10 17:45:33 neo>
 //---------------------------------------------------------------------------
 #ifndef GRID_VECTOR_TYPES
 #define GRID_VECTOR_TYPES
@@ -22,7 +22,7 @@
 #if defined QPX
 #include "Grid_qpx.h"
 #endif
-#ifdef NEONv7
+#ifdef NEONv8
 #include "Grid_neon.h"
 #endif
 
diff --git a/scripts/arm_configure.experimental_cortex57 b/scripts/arm_configure.experimental_cortex57
new file mode 100644
index 00000000..d229763e
--- /dev/null
+++ b/scripts/arm_configure.experimental_cortex57
@@ -0,0 +1,3 @@
+#./configure --host=arm-linux-gnueabihf  CXX=clang++-3.5 CXXFLAGS='-std=c++11 -O3 -target arm-linux-gnueabihf -I/usr/arm-linux-gnueabihf/include/ -I/home/neo/Codes/gmp6.0/gmp-arm/include/ -I/usr/lib/llvm-3.5/lib/clang/3.5.0/include/ -L/home/neo/Codes/gmp6.0/gmp-arm/lib/ -I/home/neo/Codes/mpfr3.1.2/mpfr-arm/include/ -L/home/neo/Codes/mpfr3.1.2/mpfr-arm/lib/ -static -mcpu=cortex-a57' --enable-simd=NEONv7
+
+./configure --host=aarch64-linux-gnu  CXX=clang++-3.5 CXXFLAGS='-std=c++11 -O3 -target aarch64-linux-gnu -static -I/home/neo/Codes/gmp6.0/gmp-armv8/include/ -L/home/neo/Codes/gmp6.0/gmp-armv8/lib/ -I/home/neo/Codes/mpfr3.1.2/mpfr-armv8/include/ -L/home/neo/Codes/mpfr3.1.2/mpfr-armv8/lib/ -I/usr/aarch64-linux-gnu/include/ -I/usr/aarch64-linux-gnu/include/c++/4.8.2/aarch64-linux-gnu/' --enable-simd=NEONv7
diff --git a/tests/Test_hmc_WilsonGauge.cc b/tests/Test_hmc_WilsonGauge.cc
index d39ec9b9..7682581d 100644
--- a/tests/Test_hmc_WilsonGauge.cc
+++ b/tests/Test_hmc_WilsonGauge.cc
@@ -24,13 +24,13 @@ int main (int argc, char ** argv)
   GridCartesian           Fine(latt_size,simd_layout,mpi_layout);
   GridParallelRNG  pRNG(&Fine);
   pRNG.SeedRandomDevice();
-  LatticeLorentzColourMatrix     U(&Fine);
+  LatticeGaugeField U(&Fine);
 
   SU3::HotConfiguration(pRNG, U);
  
 
   // simplify template declaration? Strip the lorentz from the second template
-  WilsonGaugeAction<LatticeLorentzColourMatrix, LatticeColourMatrix> Waction(6.0);
+  WilsonGaugeAction<LatticeGaugeField, LatticeColourMatrix> Waction(6.0);
 
   //Collect actions
   ActionLevel Level1;
diff --git a/tests/Test_simd.cc b/tests/Test_simd.cc
index 800001be..ee50a312 100644
--- a/tests/Test_simd.cc
+++ b/tests/Test_simd.cc
@@ -194,6 +194,31 @@ int main (int argc, char ** argv)
   // Insist that operations on random scalars gives
   // identical results to on vectors.
 
+  std::cout << "==================================="<<  std::endl;
+  std::cout << "Testing vRealF "<<std::endl;
+  std::cout << "==================================="<<  std::endl;
+
+
+  Tester<RealF,vRealF>(funcPlus());
+  Tester<RealF,vRealF>(funcMinus());
+  Tester<RealF,vRealF>(funcTimes());
+  Tester<RealF,vRealF>(funcAdj());
+  Tester<RealF,vRealF>(funcConj());
+  Tester<RealF,vRealF>(funcInnerProduct());
+  ReductionTester<RealF,RealF,vRealF>(funcReduce());
+
+  std::cout << "==================================="<<  std::endl;
+  std::cout << "Testing vRealD "<<std::endl;
+  std::cout << "==================================="<<  std::endl;
+
+  Tester<RealD,vRealD>(funcPlus());
+  Tester<RealD,vRealD>(funcMinus());
+  Tester<RealD,vRealD>(funcTimes());
+  Tester<RealD,vRealD>(funcAdj());
+  Tester<RealD,vRealD>(funcConj());
+  Tester<RealD,vRealD>(funcInnerProduct());
+  ReductionTester<RealD,RealD,vRealD>(funcReduce());
+
   std::cout << "==================================="<<  std::endl;
   std::cout << "Testing vComplexF "<<std::endl;
   std::cout << "==================================="<<  std::endl;
@@ -223,30 +248,6 @@ int main (int argc, char ** argv)
   Tester<ComplexD,vComplexD>(funcInnerProduct());
   ReductionTester<ComplexD,ComplexD,vComplexD>(funcReduce());
 
-  std::cout << "==================================="<<  std::endl;
-  std::cout << "Testing vRealF "<<std::endl;
-  std::cout << "==================================="<<  std::endl;
-
-
-  Tester<RealF,vRealF>(funcPlus());
-  Tester<RealF,vRealF>(funcMinus());
-  Tester<RealF,vRealF>(funcTimes());
-  Tester<RealF,vRealF>(funcAdj());
-  Tester<RealF,vRealF>(funcConj());
-  Tester<RealF,vRealF>(funcInnerProduct());
-  ReductionTester<RealF,RealF,vRealF>(funcReduce());
-
-  std::cout << "==================================="<<  std::endl;
-  std::cout << "Testing vRealD "<<std::endl;
-  std::cout << "==================================="<<  std::endl;
-
-  Tester<RealD,vRealD>(funcPlus());
-  Tester<RealD,vRealD>(funcMinus());
-  Tester<RealD,vRealD>(funcTimes());
-  Tester<RealD,vRealD>(funcAdj());
-  Tester<RealD,vRealD>(funcConj());
-  Tester<RealD,vRealD>(funcInnerProduct());
-  ReductionTester<RealD,RealD,vRealD>(funcReduce());
 
   Grid_finalize();
 }

From 8a4f9d236724437d50330cdd87608106002b4b89 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 21 Jul 2015 12:12:29 +0900
Subject: [PATCH 392/429] verbose

---
 lib/algorithms/iterative/ConjugateGradient.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/algorithms/iterative/ConjugateGradient.h b/lib/algorithms/iterative/ConjugateGradient.h
index 3829dad8..e281d765 100644
--- a/lib/algorithms/iterative/ConjugateGradient.h
+++ b/lib/algorithms/iterative/ConjugateGradient.h
@@ -15,7 +15,7 @@ public:
     Integer MaxIterations;
     int verbose;
     ConjugateGradient(RealD tol,Integer maxit) : Tolerance(tol), MaxIterations(maxit) { 
-      verbose=1;
+      verbose=0;
     };
 
 

From 59baa15d9fc2b85b06c96423247b8ac29f00661a Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 21 Jul 2015 12:13:03 +0900
Subject: [PATCH 393/429] Tweaks to subspace set up to put in g5 r5 hermiticity

---
 lib/algorithms/CoarsenedMatrix.h | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/lib/algorithms/CoarsenedMatrix.h b/lib/algorithms/CoarsenedMatrix.h
index e257294c..998292b2 100644
--- a/lib/algorithms/CoarsenedMatrix.h
+++ b/lib/algorithms/CoarsenedMatrix.h
@@ -117,15 +117,15 @@ namespace Grid {
       }
       Orthogonalise();
     }
-    virtual void CreateSubspace(GridParallelRNG  &RNG,LinearOperatorBase<FineField> &hermop) {
+    virtual void CreateSubspace(GridParallelRNG  &RNG,LinearOperatorBase<FineField> &hermop,int nn=nbasis) {
 
       RealD scale;
 
-      ConjugateGradient<FineField> CG(1.0e-4,10000);
+      ConjugateGradient<FineField> CG(2.0e-3,10000);
       FineField noise(FineGrid);
       FineField Mn(FineGrid);
 
-      for(int b=0;b<nbasis;b++){
+      for(int b=0;b<nn;b++){
 	
 	gaussian(RNG,noise);
 	scale = std::pow(norm2(noise),-0.5); 
@@ -133,7 +133,7 @@ namespace Grid {
 
 	hermop.Op(noise,Mn); std::cout << "noise   ["<<b<<"] <n|MdagM|n> "<<norm2(Mn)<<std::endl;
 
-	for(int i=0;i<2;i++){
+	for(int i=0;i<1;i++){
 
 	  CG(hermop,noise,subspace[b]);
 
@@ -144,7 +144,8 @@ namespace Grid {
 	}
 
 	hermop.Op(noise,Mn); std::cout << "filtered["<<b<<"] <f|MdagM|f> "<<norm2(Mn)<<std::endl;
-	subspace[b] = noise;
+	subspace[b]   = noise;
+
       }
 
       Orthogonalise();

From 01f9b1f6a4058082c488378d2e321b6227ca5f31 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 21 Jul 2015 13:48:57 +0900
Subject: [PATCH 394/429] More info

---
 lib/algorithms/approx/Chebyshev.h | 46 ++++++++++++++++---------------
 1 file changed, 24 insertions(+), 22 deletions(-)

diff --git a/lib/algorithms/approx/Chebyshev.h b/lib/algorithms/approx/Chebyshev.h
index 0dad7179..f629ab08 100644
--- a/lib/algorithms/approx/Chebyshev.h
+++ b/lib/algorithms/approx/Chebyshev.h
@@ -50,6 +50,15 @@ namespace Grid {
       return;
     }
 
+    // Convenience for plotting the approximation
+    void   PlotApprox(std::ostream &out) {
+      out<<"Polynomial approx ["<<lo<<","<<hi<<"]"<<std::endl;
+      for(double x=lo;x<hi;x+=(hi-lo)/50.0){
+	out <<x<<"\t"<<approx(x)<<std::endl;
+      }
+    };
+
+
     Chebyshev(double _lo,double _hi,int _order, double (* func)(double) ){
       lo=_lo;
       hi=_hi;
@@ -95,46 +104,39 @@ namespace Grid {
       return sum;
     };
 
-    // Convenience for plotting the approximation
-    void   PlotApprox(std::ostream &out) {
-      out<<"Polynomial approx ["<<lo<<","<<hi<<"]"<<std::endl;
-      for(double x=lo;x<hi;x+=(hi-lo)/50.0){
-	out <<x<<"\t"<<approx(x)<<std::endl;
-      }
-    };
-
-    // Implement the required interface; could require Lattice base class
+    // Implement the required interface
     void operator() (LinearOperatorBase<Field> &Linop, const Field &in, Field &out) {
 
-      Field T0 = in;
-      Field T1 = T0; // Field T1(T0._grid); more efficient but hardwires Lattice class
-      Field T2 = T1;
+      GridBase *grid=in._grid;
+
+      int vol=grid->gSites();
+
+      Field T0(grid); T0 = in;  
+      Field T1(grid); 
+      Field T2(grid);
+      Field y(grid);
       
-      // use a pointer trick to eliminate copies
       Field *Tnm = &T0;
       Field *Tn  = &T1;
       Field *Tnp = &T2;
-      Field y   = in;
-  
+
+      std::cout << "Chebyshev ["<<lo<<","<<hi<<"]"<< " order "<<order <<std::endl;
+      // Tn=T1 = (xscale M + mscale)in
       double xscale = 2.0/(hi-lo);
       double mscale = -(hi+lo)/(hi-lo);
-
-      // Tn=T1 = (xscale M + mscale)in
-      Linop.Op(T0,y);
-
+      Linop.HermOp(T0,y);
       T1=y*xscale+in*mscale;
 
       // sum = .5 c[0] T0 + c[1] T1
       out = (0.5*Coeffs[0])*T0 + Coeffs[1]*T1;
-
       for(int n=2;n<order;n++){
 	
-	Linop.Op(*Tn,y);
+	Linop.HermOp(*Tn,y);
 
 	y=xscale*y+mscale*(*Tn);
 
 	*Tnp=2.0*y-(*Tnm);
-	
+
 	out=out+Coeffs[n]* (*Tnp);
 
 	// Cycle pointers to avoid copies

From 821ac7b6f4ea87f5d89de207295c73d95cd26264 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 21 Jul 2015 13:51:56 +0900
Subject: [PATCH 395/429] Printing change

---
 lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h b/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h
index 148359a8..fee41fa8 100644
--- a/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h
+++ b/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h
@@ -137,7 +137,7 @@ namespace Grid {
 
 	cp = axpy_norm(r,-a,q[peri_k],r);  
 
-	std::cout<< " VPCG_step resid" <<sqrt(cp/rsq)<<std::endl; 
+	std::cout<< " VPGCR_step resid" <<sqrt(cp/rsq)<<std::endl; 
 	if((k==nstep-1)||(cp<rsq)){
 	  return cp;
 	}

From 81987b64a6424abba3369a22c08262f494d05750 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 21 Jul 2015 13:52:23 +0900
Subject: [PATCH 396/429] This file is being developed and will remain hacky
 until the new algorithm is complete

---
 tests/Test_dwf_hdcr.cc | 176 +++++++++++++++++++++++++++++++++++++++--
 1 file changed, 169 insertions(+), 7 deletions(-)

diff --git a/tests/Test_dwf_hdcr.cc b/tests/Test_dwf_hdcr.cc
index bafdb639..488f7277 100644
--- a/tests/Test_dwf_hdcr.cc
+++ b/tests/Test_dwf_hdcr.cc
@@ -6,6 +6,10 @@ using namespace std;
 using namespace Grid;
 using namespace Grid::QCD;
 
+RealD InverseApproximation(RealD x){
+  return 1.0/x;
+}
+
 template<class Fobj,class CComplex,int nbasis, class Matrix>
 class MultiGridPreconditioner : public LinearFunction< Lattice<Fobj> > {
 public:
@@ -33,6 +37,27 @@ public:
       _Matrix(FineMatrix)
   {
   }
+
+  void PowerMethod(const FineField &in) {
+
+    FineField p1(in._grid);
+    FineField p2(in._grid);
+
+    MdagMLinearOperator<Matrix,FineField>   fMdagMOp(_Matrix);
+
+    p1=in;
+    RealD absp2;
+    for(int i=0;i<20;i++){
+      RealD absp1=std::sqrt(norm2(p1));
+      fMdagMOp.HermOp(p1,p2);// this is the G5 herm bit      
+      //      _FineOperator.Op(p1,p2);// this is the G5 herm bit      
+      RealD absp2=std::sqrt(norm2(p2));
+      if(i%10==9)
+	std::cout << "Power method on mdagm "<<i<<" " << absp2/absp1<<std::endl;
+      p1=p2*(1.0/std::sqrt(absp2));
+    }
+  }
+
 #if 0
   void operator()(const FineField &in, FineField & out) {
 
@@ -49,7 +74,7 @@ public:
     std::cout<<"Completeness: "<<std::sqrt(norm2(out)/norm2(in))<<std::endl;
 
     // Build some solvers
-    ConjugateGradient<FineField>    fCG(1.0e-1,1000);
+    ConjugateGradient<FineField>    fCG(1.0e-3,1000);
     ConjugateGradient<CoarseVector>  CG(1.0e-8,100000);
 
     ////////////////////////////////////////////////////////////////////////
@@ -164,6 +189,7 @@ public:
   }
 #endif
   // ADEF1: [MP+Q ] in =M [1 - A Q] in + Q in  
+#if 0
   void operator()(const FineField &in, FineField & out) {
 
     CoarseVector Csrc(_CoarseOperator.Grid());
@@ -171,11 +197,11 @@ public:
     CoarseVector Csol(_CoarseOperator.Grid()); Csol=zero;
 
     ConjugateGradient<CoarseVector>  CG(1.0e-10,100000);
-    ConjugateGradient<FineField>    fCG(1.0e-3,1000);
+    ConjugateGradient<FineField>    fCG(3.0e-2,1000);
 
     HermitianLinearOperator<CoarseOperator,CoarseVector>  HermOp(_CoarseOperator);
     MdagMLinearOperator<CoarseOperator,CoarseVector>     MdagMOp(_CoarseOperator);
-    MdagMLinearOperator<Matrix,FineField>               fMdagMOp(_Matrix);
+    ShiftedMdagMLinearOperator<Matrix,FineField>        fMdagMOp(_Matrix,0.1);
 
     FineField tmp(in._grid);
     FineField res(in._grid);
@@ -191,7 +217,7 @@ public:
     CG(MdagMOp,Ctmp,Csol);
     _Aggregates.PromoteFromSubspace(Csol,Qin);
 
-
+    //    Qin=0;
     _FineOperator.Op(Qin,tmp);// A Q in
     tmp = in - tmp;            // in - A Q in
 
@@ -206,6 +232,116 @@ public:
     std::cout<<"preconditioner thinks residual is "<<std::sqrt(norm2(tmp)/norm2(in))<<std::endl;
 
   }
+#endif
+
+  void SmootherTest (const FineField & in){
+    
+    FineField vec1(in._grid);
+    FineField vec2(in._grid);
+    RealD lo[3] = { 0.5, 1.0, 2.0};
+
+    //    MdagMLinearOperator<Matrix,FineField>        fMdagMOp(_Matrix);
+    ShiftedMdagMLinearOperator<Matrix,FineField> fMdagMOp(_Matrix,0.5);
+
+    RealD Ni,r;
+
+    Ni = norm2(in);
+
+    for(int ilo=0;ilo<4;ilo++){
+      for(int ord=10;ord<60;ord+=10){
+
+	_FineOperator.AdjOp(in,vec1);
+
+	Chebyshev<FineField> Cheby  (lo[ilo],70.0,ord,InverseApproximation);
+	Cheby(fMdagMOp,vec1,vec2);    // solves  MdagM = g5 M g5M
+
+	_FineOperator.Op(vec2,vec1);// this is the G5 herm bit
+	vec1  = in - vec1;   // tmp  = in - A Min
+	r=norm2(vec1);
+	std::cout << "Smoother resid "<<std::sqrt(r/Ni)<<std::endl;
+
+      }
+    }
+  }
+
+  void operator()(const FineField &in, FineField & out) {
+
+    CoarseVector Csrc(_CoarseOperator.Grid());
+    CoarseVector Ctmp(_CoarseOperator.Grid());
+    CoarseVector Csol(_CoarseOperator.Grid()); Csol=zero;
+
+    ConjugateGradient<CoarseVector>  CG(1.0e-5,100000);
+    ConjugateGradient<FineField>    fCG(3.0e-2,1000);
+
+    HermitianLinearOperator<CoarseOperator,CoarseVector>  HermOp(_CoarseOperator);
+    MdagMLinearOperator<CoarseOperator,CoarseVector>     MdagMOp(_CoarseOperator);
+    //    MdagMLinearOperator<Matrix,FineField>        fMdagMOp(_Matrix);
+    ShiftedMdagMLinearOperator<Matrix,FineField> fMdagMOp(_Matrix,0.5);
+
+    FineField vec1(in._grid);
+    FineField vec2(in._grid);
+
+    //    Chebyshev<FineField> Cheby    (0.5,70.0,30,InverseApproximation);
+    //    Chebyshev<FineField> ChebyAccu(0.5,70.0,30,InverseApproximation);
+    Chebyshev<FineField> Cheby    (1.0,70.0,20,InverseApproximation);
+    Chebyshev<FineField> ChebyAccu(1.0,70.0,20,InverseApproximation);
+
+    _Aggregates.ProjectToSubspace  (Csrc,in);
+    _Aggregates.PromoteFromSubspace(Csrc,out);
+    std::cout<<"Completeness: "<<std::sqrt(norm2(out)/norm2(in))<<std::endl;
+    
+    //    ofstream fout("smoother");
+    //    Cheby.csv(fout);
+
+    // V11 multigrid.
+    // Use a fixed chebyshev and hope hermiticity helps.
+
+    // To make a working smoother for indefinite operator
+    // must multiply by "Mdag" (ouch loses all low mode content)
+    // and apply to poly approx of (mdagm)^-1.
+    // so that we end up with an odd polynomial.
+
+    RealD Ni = norm2(in);
+
+    _FineOperator.AdjOp(in,vec1);// this is the G5 herm bit
+    ChebyAccu(fMdagMOp,vec1,out);    // solves  MdagM = g5 M g5M
+
+    std::cout << "Smoother norm "<<norm2(out)<<std::endl;
+
+    // Update with residual for out
+    _FineOperator.Op(out,vec1);// this is the G5 herm bit
+    vec1  = in - vec1;   // tmp  = in - A Min
+
+    RealD r = norm2(vec1);
+
+    std::cout << "Smoother resid "<<std::sqrt(r/Ni)<< " " << r << " " << Ni <<std::endl;
+    
+    _Aggregates.ProjectToSubspace  (Csrc,vec1);
+    HermOp.AdjOp(Csrc,Ctmp);// Normal equations
+    CG(MdagMOp,Ctmp,Csol);
+    _Aggregates.PromoteFromSubspace(Csol,vec1); // Ass^{-1} [in - A Min]_s
+                                             // Q = Q[in - A Min]  
+    out = out+vec1;
+
+    // Three preconditioner smoothing -- hermitian if C3 = C1
+    // Recompute error
+    _FineOperator.Op(out,vec1);// this is the G5 herm bit
+    vec1  = in - vec1;   // tmp  = in - A Min
+    r=norm2(vec1);
+
+    std::cout << "Coarse resid "<<std::sqrt(r/Ni)<<std::endl;
+
+    // Reapply smoother
+    _FineOperator.Op(vec1,vec2);  // this is the G5 herm bit
+    ChebyAccu(fMdagMOp,vec2,vec1);    // solves  MdagM = g5 M g5M
+
+    out =out+vec1;
+    _FineOperator.Op(out,vec1);// this is the G5 herm bit
+    vec1  = in - vec1;   // tmp  = in - A Min
+    r=norm2(vec1);
+    std::cout << "Smoother resid "<<std::sqrt(r/Ni)<<std::endl;
+
+  }
 
 };
 
@@ -224,7 +360,7 @@ int main (int argc, char ** argv)
   ///////////////////////////////////////////////////
   // Construct a coarsened grid; utility for this?
   ///////////////////////////////////////////////////
-  const int block=4;
+  const int block=2;
   std::vector<int> clatt = GridDefaultLatt();
   for(int d=0;d<clatt.size();d++){
     clatt[d] = clatt[d]/block;
@@ -265,7 +401,8 @@ int main (int argc, char ** argv)
   std::cout << "**************************************************"<< std::endl;
   DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
 
-  const int nbasis = 6;
+  const int nbasis = 32;
+  //  const int nbasis = 4;
 
   typedef Aggregation<vSpinColourVector,vTComplex,nbasis>              Subspace;
   typedef CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis>          CoarseOperator;
@@ -276,7 +413,19 @@ int main (int argc, char ** argv)
   std::cout << "**************************************************"<< std::endl;
   MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
   Subspace Aggregates(Coarse5d,FGrid);
-  Aggregates.CreateSubspace(RNG5,HermDefOp);
+  //  Aggregates.CreateSubspace(RNG5,HermDefOp,nbasis);
+  assert ( (nbasis & 0x1)==0);
+  int nb=nbasis/2;
+  std::cout << " nbasis/2 = "<<nb<<std::endl;
+  Aggregates.CreateSubspace(RNG5,HermDefOp,nb);
+  for(int n=0;n<nb;n++){
+    G5R5(Aggregates.subspace[n+nb],Aggregates.subspace[n]);
+    std::cout<<n<<" subspace "<<norm2(Aggregates.subspace[n+nb])<<" "<<norm2(Aggregates.subspace[n]) <<std::endl;
+  }
+  for(int n=0;n<nbasis;n++){
+    std::cout << "vec["<<n<<"] = "<<norm2(Aggregates.subspace[n])  <<std::endl;
+  }
+
 //  for(int i=0;i<nbasis;i++){
 //    result =     Aggregates.subspace[i];
 //    Aggregates.subspace[i]=result+g5*result;
@@ -319,12 +468,18 @@ int main (int argc, char ** argv)
   MultiGridPreconditioner <vSpinColourVector,vTComplex,nbasis,DomainWallFermion> Precon(Aggregates, LDOp,HermIndefOp,Ddwf);
   TrivialPrecon<LatticeFermion> simple;
 
+  std::cout << "**************************************************"<< std::endl;
+  std::cout << "Testing smoother efficacy"<< std::endl;
+  std::cout << "**************************************************"<< std::endl;
+  Precon.SmootherTest(src);
+
   std::cout << "**************************************************"<< std::endl;
   std::cout << "Unprec CG "<< std::endl;
   std::cout << "**************************************************"<< std::endl;
   //  TrivialPrecon<LatticeFermion> simple;
   ConjugateGradient<LatticeFermion> fCG(1.0e-8,100000);
   fCG(HermDefOp,src,result);
+  //  exit(0);
 
   std::cout << "**************************************************"<< std::endl;
   std::cout << "Testing GCR on indef matrix "<< std::endl;
@@ -332,6 +487,13 @@ int main (int argc, char ** argv)
   //  PrecGeneralisedConjugateResidual<LatticeFermion> UPGCR(1.0e-8,100000,simple,8,128);
   //  UPGCR(HermIndefOp,src,result);
 
+  
+  /// Get themax eval
+  std::cout << "**************************************************"<< std::endl;
+  std::cout <<" Applying power method to find spectral range      "<<std::endl;
+  std::cout << "**************************************************"<< std::endl;
+  Precon.PowerMethod(src);
+
   std::cout << "**************************************************"<< std::endl;
   std::cout << "Building a two level PGCR "<< std::endl;
   std::cout << "**************************************************"<< std::endl;

From 9d18773fbc6272bc1c6cb6629438e394a88dca61 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 21 Jul 2015 13:52:59 +0900
Subject: [PATCH 397/429] This was needed to compile on gcc

---
 lib/qcd/hmc/integrators/Integrator.h | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/lib/qcd/hmc/integrators/Integrator.h b/lib/qcd/hmc/integrators/Integrator.h
index 5eb319b2..90707d51 100644
--- a/lib/qcd/hmc/integrators/Integrator.h
+++ b/lib/qcd/hmc/integrators/Integrator.h
@@ -69,10 +69,10 @@ namespace Grid{
 	LatticeColourMatrix Umu(U._grid);
 	LatticeColourMatrix Pmu(U._grid);
 	for (int mu = 0; mu < Nd; mu++){
-	  Umu=peekLorentz(U, mu);
-	  Pmu=peekLorentz(*P, mu);
+	  Umu=PeekIndex<LorentzIndex>(U, mu);
+	  Pmu=PeekIndex<LorentzIndex>(*P, mu);
 	  Umu = expMat(Pmu, ep, Params.Nexp)*Umu;
-	  pokeLorentz(U, Umu, mu);
+	  PokeIndex<LorentzIndex>(U, Umu, mu);
 	}
 
       }

From c7925e5c9b8c0b9bacecbef206b91a944d9a652d Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 21 Jul 2015 13:53:23 +0900
Subject: [PATCH 398/429] Small pretty layout change

---
 lib/qcd/action/fermion/g5HermitianLinop.h | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/lib/qcd/action/fermion/g5HermitianLinop.h b/lib/qcd/action/fermion/g5HermitianLinop.h
index e5b691d8..864af86d 100644
--- a/lib/qcd/action/fermion/g5HermitianLinop.h
+++ b/lib/qcd/action/fermion/g5HermitianLinop.h
@@ -1,7 +1,9 @@
 #ifndef G5_HERMITIAN_LINOP
 #define G5_HERMITIAN_LINOP
+
 namespace Grid {
   namespace QCD {
+
 ////////////////////////////////////////////////////////////////////
 // Wrap an already herm matrix
 ////////////////////////////////////////////////////////////////////

From e34f8adbf4c43ca149234278a3b90078c0c107e9 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 21 Jul 2015 13:54:09 +0900
Subject: [PATCH 399/429] No changes shown on git diff

---
 lib/algorithms/LinearOperator.h | 41 +++++++++++++++++++++++++++++++++
 1 file changed, 41 insertions(+)

diff --git a/lib/algorithms/LinearOperator.h b/lib/algorithms/LinearOperator.h
index a804bceb..31e086ba 100644
--- a/lib/algorithms/LinearOperator.h
+++ b/lib/algorithms/LinearOperator.h
@@ -71,6 +71,47 @@ namespace Grid {
       }
     };
 
+    ////////////////////////////////////////////////////////////////////
+    // Construct herm op and shift it for mgrid smoother
+    ////////////////////////////////////////////////////////////////////
+    template<class Matrix,class Field>
+    class ShiftedMdagMLinearOperator : public LinearOperatorBase<Field> {
+      Matrix &_Mat;
+      RealD _shift;
+    public:
+    ShiftedMdagMLinearOperator(Matrix &Mat,RealD shift): _Mat(Mat), _shift(shift){};
+      // Support for coarsening to a multigrid
+      void OpDiag (const Field &in, Field &out) {
+	_Mat.Mdiag(in,out);
+	assert(0);
+      }
+      void OpDir  (const Field &in, Field &out,int dir,int disp) {
+	_Mat.Mdir(in,out,dir,disp);
+	assert(0);
+      }
+      void Op     (const Field &in, Field &out){
+	_Mat.M(in,out);
+	assert(0);
+      }
+      void AdjOp     (const Field &in, Field &out){
+	_Mat.Mdag(in,out);
+	assert(0);
+      }
+      void HermOpAndNorm(const Field &in, Field &out,RealD &n1,RealD &n2){
+	_Mat.MdagM(in,out,n1,n2);
+	out = out + _shift*in;
+
+	ComplexD dot;	
+	dot= innerProduct(in,out);
+	n1=real(dot);
+	n2=norm2(out);
+      }
+      void HermOp(const Field &in, Field &out){
+	RealD n1,n2;
+	HermOpAndNorm(in,out,n1,n2);
+      }
+    };
+
     ////////////////////////////////////////////////////////////////////
     // Wrap an already herm matrix
     ////////////////////////////////////////////////////////////////////

From 89258456848b53f4556e12495969ded1d9d28661 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 21 Jul 2015 13:56:22 +0900
Subject: [PATCH 400/429] Merge

---
 lib/algorithms/approx/Remez.h             |   2 +-
 lib/qcd/action/fermion/WilsonKernels.cc   |   2 -
 lib/qcd/hmc/integrators/Integrator.cc     |   2 +-
 lib/qcd/hmc/integrators/Integrator_base.h | 212 ++++++++++++++++++++++
 tests/Test_multishift_sqrt.cc             |  35 +++-
 5 files changed, 247 insertions(+), 6 deletions(-)
 create mode 100644 lib/qcd/hmc/integrators/Integrator_base.h

diff --git a/lib/algorithms/approx/Remez.h b/lib/algorithms/approx/Remez.h
index e6a973df..336382d2 100644
--- a/lib/algorithms/approx/Remez.h
+++ b/lib/algorithms/approx/Remez.h
@@ -15,7 +15,7 @@
 #ifndef INCLUDED_ALG_REMEZ_H
 #define INCLUDED_ALG_REMEZ_H
 
-#include <algorithms/approx/bigfloat_double.h>
+#include <algorithms/approx/bigfloat.h>
 
 #define JMAX 10000 //Maximum number of iterations of Newton's approximation
 #define SUM_MAX 10 // Maximum number of terms in exponential
diff --git a/lib/qcd/action/fermion/WilsonKernels.cc b/lib/qcd/action/fermion/WilsonKernels.cc
index 4054280a..bdc18eee 100644
--- a/lib/qcd/action/fermion/WilsonKernels.cc
+++ b/lib/qcd/action/fermion/WilsonKernels.cc
@@ -1,5 +1,4 @@
 #include <Grid.h>
-
 namespace Grid {
 namespace QCD {
 
@@ -13,7 +12,6 @@ void DiracOptDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     vHalfSpinColourVector Uchi;
     int offset,local,perm, ptype;
 
-
     // Xp
     int ss = sF;
     offset = st._offsets [Xp][ss];
diff --git a/lib/qcd/hmc/integrators/Integrator.cc b/lib/qcd/hmc/integrators/Integrator.cc
index 8947405e..16a6513e 100644
--- a/lib/qcd/hmc/integrators/Integrator.cc
+++ b/lib/qcd/hmc/integrators/Integrator.cc
@@ -18,7 +18,7 @@ namespace Grid{
       Pmu = zero;
       for(int mu=0;mu<Nd;mu++){
 	SU3::GaussianLieAlgebraMatrix(pRNG, Pmu);
-	pokeLorentz(P, Pmu, mu);
+	PokeIndex<LorentzIndex>(P, Pmu, mu);
       }
       
     }
diff --git a/lib/qcd/hmc/integrators/Integrator_base.h b/lib/qcd/hmc/integrators/Integrator_base.h
new file mode 100644
index 00000000..38798d67
--- /dev/null
+++ b/lib/qcd/hmc/integrators/Integrator_base.h
@@ -0,0 +1,212 @@
+//--------------------------------------------------------------------
+/*! @file Integrator_base.h
+ * @brief Declaration of classes for the abstract Molecular Dynamics integrator
+ *
+ * @author Guido Cossu
+ */
+//--------------------------------------------------------------------
+
+#ifndef INTEGRATOR_INCLUDED
+#define INTEGRATOR_INCLUDED
+
+#include <memory>
+
+class Observer;
+
+
+/*! @brief Abstract base class for Molecular Dynamics management */
+
+namespace Grid{
+  namespace QCD{
+
+    typedef Action<LatticeLorentzColourMatrix>*  ActPtr; // now force the same size as the rest of the code
+    typedef std::vector<ActPtr> ActionLevel;
+    typedef std::vector<ActionLevel> ActionSet;
+    typedef std::vector<Observer*> ObserverList;
+    
+    class LeapFrog;
+
+
+    struct IntegratorParameters{
+      int Nexp;
+      int MDsteps;  // number of outer steps
+      RealD trajL;  // trajectory length 
+      RealD stepsize;
+
+      IntegratorParameters(int Nexp_,
+			   int MDsteps_, 
+			   RealD trajL_):
+	Nexp(Nexp_),MDsteps(MDsteps_),trajL(trajL_),stepsize(trajL/MDsteps){};
+    };
+
+
+    namespace MDutils{
+      void generate_momenta(LatticeLorentzColourMatrix&,GridParallelRNG&);
+      void generate_momenta_su3(LatticeLorentzColourMatrix&,GridParallelRNG&);
+    }
+
+    
+    
+
+
+    template< class IntegratorPolicy >
+    class Integrator{
+    private:
+      IntegratorParameters Params;
+      const ActionSet as;
+      const std::vector<int> Nrel; //relative step size per level
+      //ObserverList observers; // not yet
+      std::unique_ptr<LatticeLorentzColourMatrix> P;
+      
+      IntegratorPolicy TheIntegrator;// contains parameters too
+
+
+      void update_P(LatticeLorentzColourMatrix&U, int level,double ep){
+	for(int a=0; a<as[level].size(); ++a){
+	  LatticeLorentzColourMatrix force(U._grid);
+	  as[level].at(a)->deriv(U,force);
+	  *P -= force*ep;
+	}
+      }
+
+
+      void update_U(LatticeLorentzColourMatrix&U, double ep){
+	//rewrite exponential to deal with the lorentz index?
+	LatticeColourMatrix Umu(U._grid);
+	LatticeColourMatrix Pmu(U._grid);
+	for (int mu = 0; mu < Nd; mu++){
+	  Umu=PeekIndex<LorentzIndex>(U, mu);
+	  Pmu=PeekIndex<LorentzIndex>(*P, mu);
+	  Umu = expMat(Pmu, Complex(ep, 0.0))*Umu;
+	}
+
+      }
+      
+      void register_observers();
+      void notify_observers();
+      
+      friend void IntegratorPolicy::step (LatticeLorentzColourMatrix& U, 
+				     int level, std::vector<int>& clock,
+				     Integrator<LeapFrog>* Integ);
+    public:
+      Integrator(IntegratorParameters Par,
+		 ActionSet& Aset, std::vector<int> Nrel_):
+	Params(Par),as(Aset),Nrel(Nrel_){
+	assert(as.size() == Nrel.size());
+      };
+      
+      ~Integrator(){}
+
+
+      //Initialization of momenta and actions
+      void init(LatticeLorentzColourMatrix& U,
+		GridParallelRNG& pRNG){
+	std::cout<< "Integrator init\n";
+	if (!P)
+	  P = new LatticeLorentzColourMatrix(U._grid);
+	MDutils::generate_momenta(*P,pRNG);
+	
+	for(int level=0; level< as.size(); ++level){
+	  for(int actionID=0; actionID<as.at(level).size(); ++actionID){
+	    as[level].at(actionID)->init(U, pRNG);
+	  }
+	}
+      }
+
+      
+
+      RealD S(LatticeLorentzColourMatrix& U){
+	// Momenta
+	LatticeComplex Hloc = - trace((*P)*adj(*P));
+	Complex Hsum = sum(Hloc);
+	
+	RealD H = Hsum.real();
+
+	// Actions
+	for(int level=0; level<as.size(); ++level)
+	  for(int actionID=0; actionID<as.at(level).size(); ++actionID)
+	    H += as[level].at(actionID)->S(U);
+	
+	return H;
+      }
+      
+      
+
+      void integrate(LatticeLorentzColourMatrix& U, int level){
+	std::vector<int> clock;
+	clock.resize(as.size(),0);
+	for(int step=0; step< Params.MDsteps; ++step)   // MD step
+	  TheIntegrator.step(U,0,clock, *(this));
+      }
+    };
+    
+
+    class MinimumNorm2{
+      const double lambda = 0.1931833275037836;
+    public:
+      void step (LatticeLorentzColourMatrix& U, int level, std::vector<int>& clock);
+      
+    };
+    
+    class LeapFrog{
+    public:
+      void step (LatticeLorentzColourMatrix& U, 
+		 int level, std::vector<int>& clock,
+		 Integrator<LeapFrog>* Integ){
+	// cl  : current level
+	// fl  : final level
+	// eps : current step size
+	
+	int fl = Integ->as.size() -1;
+	double eps = Integ->Params.stepsize;
+	
+	// Get current level step size
+	for(int l=0; l<=level; ++l) eps/= Integ->Nrel[l];
+	
+	int fin = 1;
+	for(int l=0; l<=level; ++l) fin*= Integ->Nrel[l];
+	fin = 2*Integ->Params.MDsteps*fin - 1;
+	
+	for(int e=0; e<Integ->Nrel[level]; ++e){
+	  
+	  if(clock[level] == 0){    // initial half step
+	    Integ->update_P(U, level,eps/2);
+	    ++clock[level];
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
+	  }
+	  if(level == fl){          // lowest level
+	    Integ->update_U(U, eps);
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"U "<< 0.5*(clock[level]+1) <<std::endl;
+	  }else{                 // recursive function call
+	    step(U, level+1,clock, Integ);
+	  }
+	  if(clock[level] == fin){  // final half step
+	    Integ->update_P(U, level,eps/2);
+	    
+	    ++clock[level];
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
+	  }else{                  // bulk step
+	    Integ->update_P(U, level,eps);
+	    
+	    clock[level]+=2;
+	    for(int l=0; l<level;++l) std::cout<<"   ";
+	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
+	  }
+	}
+
+
+
+
+      }
+    };
+
+
+
+
+    
+  }
+}
+#endif//INTEGRATOR_INCLUDED
diff --git a/tests/Test_multishift_sqrt.cc b/tests/Test_multishift_sqrt.cc
index 3d69fdf1..4fff263f 100644
--- a/tests/Test_multishift_sqrt.cc
+++ b/tests/Test_multishift_sqrt.cc
@@ -17,8 +17,12 @@ public:
     pRNG.SeedFixedIntegers(seeds);
 
     random(pRNG,sqrtscale);
-    sqrtscale = sqrtscale * adj(sqrtscale);// force real pos def
-    scale = sqrtscale * sqrtscale; 
+    sqrtscale = real(sqrtscale)*3.0+0.5;// force real pos def
+    scale = sqrtscale *sqrtscale; //scale should be bounded by 12.25
+
+    // 
+    //    sqrtscale = 2.0;
+    //    scale = 4.0;
   }
   // Support for coarsening to a multigrid
   void OpDiag (const Field &in, Field &out) {};
@@ -48,8 +52,18 @@ public:
   void ApplySqrt(const Field &in, Field &out){
     out = sqrtscale * in;
   }
+  void ApplyInverse(const Field &in, Field &out){
+    out = pow(scale,-1.0) * in;
+  }
 };
 
+RealD InverseApproximation(RealD x){
+  return 1.0/x;
+}
+RealD SqrtApproximation(RealD x){
+  return std::sqrt(x);
+}
+
 
 int main (int argc, char ** argv)
 {
@@ -114,5 +128,22 @@ int main (int argc, char ** argv)
   error = summed - combined;
   std::cout << " summed-combined "<<norm2(error)    <<std::endl;
 
+
+  src=1.0;
+  Chebyshev<LatticeFermion> Cheby(0.1,40.0,200,InverseApproximation);
+
+  std::cout<<"Chebuy approx vector "<<std::endl;
+  Cheby(Diagonal,src,combined);
+  std::ofstream of("cheby");
+  Cheby.csv(of);
+
+  Diagonal.ApplyInverse(src,reference);
+  error = reference - combined;
+
+  std::cout << "Chebyshev inverse test "<<std::endl;
+  std::cout << " Reference "<<norm2(reference)<<std::endl;
+  std::cout << " combined  "<<norm2(combined) <<std::endl;
+  std::cout << " error     "<<norm2(error)    <<std::endl;
+
   Grid_finalize();
 }

From d6a2d734d3019e17dcee2c1a6c1a6a9f46f0070e Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 22 Jul 2015 00:30:05 +0900
Subject: [PATCH 401/429] 5x speed up now

---
 tests/Test_dwf_hdcr.cc | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/tests/Test_dwf_hdcr.cc b/tests/Test_dwf_hdcr.cc
index 488f7277..ebdf68c8 100644
--- a/tests/Test_dwf_hdcr.cc
+++ b/tests/Test_dwf_hdcr.cc
@@ -276,15 +276,15 @@ public:
     HermitianLinearOperator<CoarseOperator,CoarseVector>  HermOp(_CoarseOperator);
     MdagMLinearOperator<CoarseOperator,CoarseVector>     MdagMOp(_CoarseOperator);
     //    MdagMLinearOperator<Matrix,FineField>        fMdagMOp(_Matrix);
-    ShiftedMdagMLinearOperator<Matrix,FineField> fMdagMOp(_Matrix,0.5);
+    ShiftedMdagMLinearOperator<Matrix,FineField> fMdagMOp(_Matrix,1.0);
 
     FineField vec1(in._grid);
     FineField vec2(in._grid);
 
     //    Chebyshev<FineField> Cheby    (0.5,70.0,30,InverseApproximation);
     //    Chebyshev<FineField> ChebyAccu(0.5,70.0,30,InverseApproximation);
-    Chebyshev<FineField> Cheby    (1.0,70.0,20,InverseApproximation);
-    Chebyshev<FineField> ChebyAccu(1.0,70.0,20,InverseApproximation);
+    Chebyshev<FineField> Cheby    (2.0,70.0,10,InverseApproximation);
+    Chebyshev<FineField> ChebyAccu(2.0,70.0,10,InverseApproximation);
 
     _Aggregates.ProjectToSubspace  (Csrc,in);
     _Aggregates.PromoteFromSubspace(Csrc,out);
@@ -477,8 +477,8 @@ int main (int argc, char ** argv)
   std::cout << "Unprec CG "<< std::endl;
   std::cout << "**************************************************"<< std::endl;
   //  TrivialPrecon<LatticeFermion> simple;
-  ConjugateGradient<LatticeFermion> fCG(1.0e-8,100000);
-  fCG(HermDefOp,src,result);
+  //  ConjugateGradient<LatticeFermion> fCG(1.0e-8,100000);
+  //  fCG(HermDefOp,src,result);
   //  exit(0);
 
   std::cout << "**************************************************"<< std::endl;

From 91629a24fe43cbb46f705d84133e11e3f6f91511 Mon Sep 17 00:00:00 2001
From: paboyle <paboyle@ph.ed.ac.uk>
Date: Tue, 21 Jul 2015 22:41:01 -0700
Subject: [PATCH 402/429] Removed troublesome macros

---
 configure.ac | 20 +++++++++-----------
 1 file changed, 9 insertions(+), 11 deletions(-)

diff --git a/configure.ac b/configure.ac
index 5ef6fa8d..f9283a32 100644
--- a/configure.ac
+++ b/configure.ac
@@ -26,10 +26,9 @@ AC_LANG(C++)
 AC_PROG_CXX
 AC_OPENMP
 AC_PROG_RANLIB
-AX_CXX_COMPILE_STDCXX_11(noext, mandatory)
+#AX_CXX_COMPILE_STDCXX_11(noext, mandatory)
 AX_EXT
 
-
 # Checks for libraries.
 #AX_GCC_VAR_ATTRIBUTE(aligned)
 
@@ -66,7 +65,6 @@ AC_CHECK_LIB([mpfr],[mpfr_init],,
 Please install or provide the correct path to your installation
 Info at: http://www.mpfr.org/)])
 
-
 AC_ARG_ENABLE([simd],[AC_HELP_STRING([--enable-simd=SSE4|AVX|AVX2|AVX512|MIC],\
 	[Select instructions to be SSE4.0, AVX 1.0, AVX 2.0+FMA, AVX 512, MIC])],\
 	[ac_SIMD=${enable_simd}],[ac_SIMD=AVX2])
@@ -155,15 +153,15 @@ AM_CONDITIONAL(BUILD_COMMS_NONE,[ test "X${ac_COMMS}X" == "XnoneX" ])
 ###################################################################
 # Checks for doxygen support
 # if present enables the "make doxyfile" command
-echo
-echo Checking doxygen support 
-echo :::::::::::::::::::::::::::::::::::::::::::
-AC_PROG_DOXYGEN
+#echo
+#echo Checking doxygen support 
+#echo :::::::::::::::::::::::::::::::::::::::::::
+#AC_PROG_DOXYGEN
 
-if test -n "$DOXYGEN"
-then
-AC_CONFIG_FILES([docs/doxy.cfg])
-fi
+#if test -n "$DOXYGEN"
+#then
+#AC_CONFIG_FILES([docs/doxy.cfg])
+#fi
 
 echo
 echo Creating configuration files

From c67bc303dbd3e7514bec62aa12dfcd1fcbf1ce97 Mon Sep 17 00:00:00 2001
From: paboyle <paboyle@ph.ed.ac.uk>
Date: Tue, 21 Jul 2015 22:49:36 -0700
Subject: [PATCH 403/429] Bug work around

---
 lib/algorithms/approx/Remez.h | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/lib/algorithms/approx/Remez.h b/lib/algorithms/approx/Remez.h
index 336382d2..518fa939 100644
--- a/lib/algorithms/approx/Remez.h
+++ b/lib/algorithms/approx/Remez.h
@@ -15,6 +15,8 @@
 #ifndef INCLUDED_ALG_REMEZ_H
 #define INCLUDED_ALG_REMEZ_H
 
+#include <stddef.h>
+
 #include <algorithms/approx/bigfloat.h>
 
 #define JMAX 10000 //Maximum number of iterations of Newton's approximation

From 28bdc909088613ed410dea981f0f5256e79000f2 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 24 Jul 2015 01:31:13 +0900
Subject: [PATCH 404/429] Sizable improvement in multigrid for unsquared.

6000 matmuls CG unprec
2000 matmuls CG prec (4000 eo muls)
1050 matmuls PGCR on 16^3 x 32 x 8 m=.01

Substantial effort on timing and logging infrastructure
---
 benchmarks/Benchmark_comms.cc                 |  22 +--
 benchmarks/Benchmark_dwf.cc                   |  40 +++--
 benchmarks/Benchmark_memory_bandwidth.cc      |  48 +++---
 benchmarks/Benchmark_su3.cc                   |  50 +++---
 benchmarks/Benchmark_wilson.cc                |  34 ++--
 gcc-bug-report/broken.cc                      |   4 +-
 lib/Grid.h                                    |   3 +
 lib/GridInit.cc                               | 123 ++++++++------
 lib/GridLog.cc                                |  62 +++++++
 lib/GridLog.h                                 |  46 +++++
 lib/GridTime.h                                |  44 +++++
 lib/Make.inc                                  |   4 +-
 lib/algorithms/CoarsenedMatrix.h              |  53 +++---
 lib/algorithms/approx/Chebyshev.h             |   2 +-
 lib/algorithms/approx/Remez.cc                |   1 +
 lib/algorithms/approx/Remez.h                 |   1 +
 lib/algorithms/iterative/AdefGeneric.h        |   6 +-
 lib/algorithms/iterative/ConjugateGradient.h  |  20 +--
 .../iterative/ConjugateGradientMultiShift.h   |  14 +-
 lib/algorithms/iterative/ConjugateResidual.h  |   8 +-
 .../PrecGeneralisedConjugateResidual.h        |  26 +--
 lib/algorithms/iterative/SchurRedBlack.h      |   4 +-
 lib/lattice/Lattice_ET.h                      |  12 +-
 lib/lattice/Lattice_base.h                    |   2 +-
 lib/lattice/Lattice_reduction.h               |   2 +-
 lib/lattice/Lattice_rng.h                     |   3 +-
 lib/lattice/Lattice_transfer.h                |   5 +-
 .../fermion/ContinuedFractionFermion5D.cc     |  14 +-
 lib/qcd/action/fermion/DomainWallFermion.h    |   2 +-
 lib/qcd/action/fermion/MobiusFermion.h        |   2 +-
 .../action/fermion/MobiusZolotarevFermion.h   |   2 +-
 .../fermion/PartialFractionFermion5D.cc       |  12 +-
 lib/qcd/action/fermion/WilsonFermion5D.cc     |   2 +-
 lib/qcd/action/fermion/WilsonKernelsHand.cc   |  20 +--
 lib/qcd/action/gauge/WilsonGaugeAction.h      |   2 +-
 lib/qcd/hmc/HMC.h                             |  18 +-
 lib/qcd/hmc/integrators/Integrator.h          |   4 +-
 .../hmc/integrators/Integrator_algorithm.h    |  40 ++---
 lib/qcd/hmc/integrators/Integrator_base.h     |  18 +-
 lib/qcd/utils/SUn.h                           |  28 +--
 lib/tensors/Tensor_extract_merge.h            |   6 +-
 tests/Make.inc                                |  10 +-
 tests/Test_GaugeAction.cc                     |  16 +-
 tests/Test_cayley_cg.cc                       |  22 +--
 tests/Test_cayley_coarsen_support.cc          |  36 ++--
 tests/Test_cayley_even_odd.cc                 |  84 ++++-----
 tests/Test_cayley_ldop_cr.cc                  |  30 ++--
 tests/Test_cf_coarsen_support.cc              |  12 +-
 tests/Test_contfrac_cg.cc                     |  16 +-
 tests/Test_contfrac_even_odd.cc               |  78 ++++-----
 tests/Test_cshift.cc                          |   2 +-
 tests/Test_cshift_red_black.cc                |  30 ++--
 tests/Test_dwf_even_odd.cc                    |  70 ++++----
 tests/Test_dwf_fpgcr.cc                       |  30 ++--
 tests/Test_dwf_hdcr.cc                        | 159 ++++++++++--------
 tests/Test_gamma.cc                           |  32 ++--
 tests/Test_lie_generators.cc                  |  24 +--
 tests/Test_main.cc                            |  42 ++---
 tests/Test_multishift_sqrt.cc                 |  20 +--
 tests/Test_nersc_io.cc                        |  12 +-
 tests/Test_quenched_update.cc                 |   4 +-
 tests/Test_remez.cc                           |  14 +-
 tests/Test_rng.cc                             |  14 +-
 tests/Test_rng_fixed.cc                       |  22 +--
 tests/Test_simd.cc                            |  40 ++---
 tests/Test_stencil.cc                         |   6 +-
 tests/Test_wilson_even_odd.cc                 |  64 +++----
 67 files changed, 945 insertions(+), 753 deletions(-)
 create mode 100644 lib/GridLog.cc
 create mode 100644 lib/GridLog.h
 create mode 100644 lib/GridTime.h

diff --git a/benchmarks/Benchmark_comms.cc b/benchmarks/Benchmark_comms.cc
index 5a85269e..216172f9 100644
--- a/benchmarks/Benchmark_comms.cc
+++ b/benchmarks/Benchmark_comms.cc
@@ -11,15 +11,15 @@ int main (int argc, char ** argv)
   std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplexD::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi();
   int threads = GridThread::GetThreads();
-  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+  std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
 
   int Nloop=10;
   int nmu=0;
   for(int mu=0;mu<4;mu++) if (mpi_layout[mu]>1) nmu++;
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "= Benchmarking concurrent halo exchange in "<<nmu<<" dimensions"<<std::endl;
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<" Ls  "<<"\t\t"<<"bytes"<<"\t\t"<<"MB/s uni"<<"\t\t"<<"MB/s bidi"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "= Benchmarking concurrent halo exchange in "<<nmu<<" dimensions"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "  L  "<<"\t\t"<<" Ls  "<<"\t\t"<<"bytes"<<"\t\t"<<"MB/s uni"<<"\t\t"<<"MB/s bidi"<<std::endl;
 
 
 
@@ -87,15 +87,15 @@ int main (int argc, char ** argv)
 
       double time = stop-start; // microseconds
 
-      std::cout << lat<<"\t\t"<<Ls<<"\t\t"<<bytes<<"\t\t"<<xbytes/time<<"\t\t"<<bidibytes/time<<std::endl;
+      std::cout<<GridLogMessage << lat<<"\t\t"<<Ls<<"\t\t"<<bytes<<"\t\t"<<xbytes/time<<"\t\t"<<bidibytes/time<<std::endl;
     }
   }    
 
 
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "= Benchmarking sequential halo exchange in "<<nmu<<" dimensions"<<std::endl;
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<" Ls  "<<"\t\t"<<"bytes"<<"\t\t"<<"MB/s uni"<<"\t\t"<<"MB/s bidi"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "= Benchmarking sequential halo exchange in "<<nmu<<" dimensions"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "  L  "<<"\t\t"<<" Ls  "<<"\t\t"<<"bytes"<<"\t\t"<<"MB/s uni"<<"\t\t"<<"MB/s bidi"<<std::endl;
 
 
   for(int lat=4;lat<=32;lat+=2){
@@ -163,7 +163,7 @@ int main (int argc, char ** argv)
 
       double time = stop-start;
 
-      std::cout << lat<<"\t\t"<<Ls<<"\t\t"<<bytes<<"\t\t"<<xbytes/time<<"\t\t"<<bidibytes/time<<std::endl;
+      std::cout<<GridLogMessage << lat<<"\t\t"<<Ls<<"\t\t"<<bytes<<"\t\t"<<xbytes/time<<"\t\t"<<bidibytes/time<<std::endl;
     }
   }  
 
diff --git a/benchmarks/Benchmark_dwf.cc b/benchmarks/Benchmark_dwf.cc
index 9ce8f0db..9d9303ae 100644
--- a/benchmarks/Benchmark_dwf.cc
+++ b/benchmarks/Benchmark_dwf.cc
@@ -21,7 +21,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   int threads = GridThread::GetThreads();
-  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+  std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
 
   std::vector<int> latt4 = GridDefaultLatt();
   const int Ls=8;
@@ -81,7 +81,7 @@ int main (int argc, char ** argv)
   RealD M5  =1.8;
   DomainWallFermion Dw(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
   
-  std::cout << "Calling Dw"<<std::endl;
+  std::cout<<GridLogMessage << "Calling Dw"<<std::endl;
   int ncall=10;
   double t0=usecond();
   for(int i=0;i<ncall;i++){
@@ -92,12 +92,12 @@ int main (int argc, char ** argv)
   double volume=Ls;  for(int mu=0;mu<Nd;mu++) volume=volume*latt4[mu];
   double flops=1344*volume*ncall;
   
-  std::cout << "Called Dw"<<std::endl;
-  std::cout << "norm result "<< norm2(result)<<std::endl;
-  std::cout << "norm ref    "<< norm2(ref)<<std::endl;
-  std::cout << "mflop/s =   "<< flops/(t1-t0)<<std::endl;
+  std::cout<<GridLogMessage << "Called Dw"<<std::endl;
+  std::cout<<GridLogMessage << "norm result "<< norm2(result)<<std::endl;
+  std::cout<<GridLogMessage << "norm ref    "<< norm2(ref)<<std::endl;
+  std::cout<<GridLogMessage << "mflop/s =   "<< flops/(t1-t0)<<std::endl;
   err = ref-result; 
-  std::cout << "norm diff   "<< norm2(err)<<std::endl;
+  std::cout<<GridLogMessage << "norm diff   "<< norm2(err)<<std::endl;
 
 
   if (1)
@@ -120,11 +120,11 @@ int main (int argc, char ** argv)
     ref = -0.5*ref;
   }
   Dw.Dhop(src,result,1);
-  std::cout << "Called DwDag"<<std::endl;
-  std::cout << "norm result "<< norm2(result)<<std::endl;
-  std::cout << "norm ref    "<< norm2(ref)<<std::endl;
+  std::cout<<GridLogMessage << "Called DwDag"<<std::endl;
+  std::cout<<GridLogMessage << "norm result "<< norm2(result)<<std::endl;
+  std::cout<<GridLogMessage << "norm ref    "<< norm2(ref)<<std::endl;
   err = ref-result; 
-  std::cout << "norm diff   "<< norm2(err)<<std::endl;
+  std::cout<<GridLogMessage << "norm diff   "<< norm2(err)<<std::endl;
 
   LatticeFermion src_e (FrbGrid);
   LatticeFermion src_o (FrbGrid);
@@ -133,24 +133,32 @@ int main (int argc, char ** argv)
   LatticeFermion r_eo  (FGrid);
 
 
-  std::cout << "Calling Deo and Doe"<<std::endl;
+  std::cout<<GridLogMessage << "Calling Deo and Doe"<<std::endl;
   pickCheckerboard(Even,src_e,src);
   pickCheckerboard(Odd,src_o,src);
 
+  std::cout<<GridLogMessage << "src_e"<<norm2(src_e)<<std::endl;
+  std::cout<<GridLogMessage << "src_o"<<norm2(src_o)<<std::endl;
+
+
   Dw.DhopEO(src_o,r_e,DaggerNo);
   Dw.DhopOE(src_e,r_o,DaggerNo);
-  Dw.Dhop(src,result,DaggerNo);
+  Dw.Dhop  (src  ,result,DaggerNo);
+
+  std::cout<<GridLogMessage << "r_e"<<norm2(r_e)<<std::endl;
+  std::cout<<GridLogMessage << "r_o"<<norm2(r_o)<<std::endl;
+  std::cout<<GridLogMessage << "res"<<norm2(result)<<std::endl;
 
   setCheckerboard(r_eo,r_o);
   setCheckerboard(r_eo,r_e);
 
   err = r_eo-result; 
-  std::cout << "norm diff   "<< norm2(err)<<std::endl;
+  std::cout<<GridLogMessage << "norm diff   "<< norm2(err)<<std::endl;
 
   pickCheckerboard(Even,src_e,err);
   pickCheckerboard(Odd,src_o,err);
-  std::cout << "norm diff even  "<< norm2(src_e)<<std::endl;
-  std::cout << "norm diff odd   "<< norm2(src_o)<<std::endl;
+  std::cout<<GridLogMessage << "norm diff even  "<< norm2(src_e)<<std::endl;
+  std::cout<<GridLogMessage << "norm diff odd   "<< norm2(src_o)<<std::endl;
 
   Grid_finalize();
 }
diff --git a/benchmarks/Benchmark_memory_bandwidth.cc b/benchmarks/Benchmark_memory_bandwidth.cc
index 5b436001..16dad525 100644
--- a/benchmarks/Benchmark_memory_bandwidth.cc
+++ b/benchmarks/Benchmark_memory_bandwidth.cc
@@ -17,13 +17,13 @@ int main (int argc, char ** argv)
   std::vector<int> mpi_layout  = GridDefaultMpi();
 
   int threads = GridThread::GetThreads();
-  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+  std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
   
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "= Benchmarking fused AXPY bandwidth ; sizeof(Real) "<<sizeof(Real)<<std::endl;
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s"<<"\t\t"<<"Gflop/s"<<std::endl;
-  std::cout << "----------------------------------------------------------"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "= Benchmarking fused AXPY bandwidth ; sizeof(Real) "<<sizeof(Real)<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s"<<"\t\t"<<"Gflop/s"<<std::endl;
+  std::cout<<GridLogMessage << "----------------------------------------------------------"<<std::endl;
 
   for(int lat=4;lat<=32;lat+=4){
 
@@ -49,15 +49,15 @@ int main (int argc, char ** argv)
       
       double flops=vol*Nvec*2;// mul,add
       double bytes=3*vol*Nvec*sizeof(Real);
-      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"   \t\t"<<bytes/time<<"\t\t"<<flops/time<<std::endl;
+      std::cout<<GridLogMessage<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"   \t\t"<<bytes/time<<"\t\t"<<flops/time<<std::endl;
 
     }
 
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "= Benchmarking a*x + y bandwidth"<<std::endl;
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s"<<"\t\t"<<"Gflop/s"<<std::endl;
-  std::cout << "----------------------------------------------------------"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "= Benchmarking a*x + y bandwidth"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s"<<"\t\t"<<"Gflop/s"<<std::endl;
+  std::cout<<GridLogMessage << "----------------------------------------------------------"<<std::endl;
   
   for(int lat=4;lat<=32;lat+=4){
 
@@ -81,14 +81,14 @@ int main (int argc, char ** argv)
      
       double flops=vol*Nvec*2;// mul,add
       double bytes=3*vol*Nvec*sizeof(Real);
-      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"   \t\t"<<bytes/time<<"\t\t"<<flops/time<<std::endl;
+      std::cout<<GridLogMessage<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"   \t\t"<<bytes/time<<"\t\t"<<flops/time<<std::endl;
 
     }
 
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "= Benchmarking SCALE bandwidth"<<std::endl;
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s"<<"\t\t"<<"Gflop/s"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "= Benchmarking SCALE bandwidth"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s"<<"\t\t"<<"Gflop/s"<<std::endl;
 
   for(int lat=4;lat<=32;lat+=4){
 
@@ -114,15 +114,15 @@ int main (int argc, char ** argv)
       
       double bytes=2*vol*Nvec*sizeof(Real);
       double flops=vol*Nvec*1;// mul
-      std::cout <<std::setprecision(3) << lat<<"\t\t"<<bytes<<"   \t\t"<<bytes/time<<"\t\t"<<flops/time<<std::endl;
+      std::cout<<GridLogMessage <<std::setprecision(3) << lat<<"\t\t"<<bytes<<"   \t\t"<<bytes/time<<"\t\t"<<flops/time<<std::endl;
 
   }
 
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "= Benchmarking READ bandwidth"<<std::endl;
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s"<<"\t\t"<<"Gflop/s"<<std::endl;
-  std::cout << "----------------------------------------------------------"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "= Benchmarking READ bandwidth"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s"<<"\t\t"<<"Gflop/s"<<std::endl;
+  std::cout<<GridLogMessage << "----------------------------------------------------------"<<std::endl;
 
   for(int lat=4;lat<=32;lat+=4){
 
@@ -147,7 +147,7 @@ int main (int argc, char ** argv)
       
       double bytes=vol*Nvec*sizeof(Real);
       double flops=vol*Nvec*2;// mul,add
-      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"  \t\t"<<bytes/time<<"\t\t"<<flops/time<<std::endl;
+      std::cout<<GridLogMessage<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"  \t\t"<<bytes/time<<"\t\t"<<flops/time<<std::endl;
 
   }    
 
diff --git a/benchmarks/Benchmark_su3.cc b/benchmarks/Benchmark_su3.cc
index 903ea348..3312e3c8 100644
--- a/benchmarks/Benchmark_su3.cc
+++ b/benchmarks/Benchmark_su3.cc
@@ -14,13 +14,13 @@ int main (int argc, char ** argv)
   std::vector<int> mpi_layout  = GridDefaultMpi();
 
   int threads = GridThread::GetThreads();
-  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+  std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
 
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "= Benchmarking SU3xSU3  x= x*y"<<std::endl;
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
-  std::cout << "----------------------------------------------------------"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "= Benchmarking SU3xSU3  x= x*y"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
+  std::cout<<GridLogMessage << "----------------------------------------------------------"<<std::endl;
 
   for(int lat=2;lat<=24;lat+=2){
 
@@ -43,16 +43,16 @@ int main (int argc, char ** argv)
       double bytes=3.0*vol*Nc*Nc*sizeof(Complex);
       double footprint=2.0*vol*Nc*Nc*sizeof(Complex);
       double flops=Nc*Nc*(6.0+8.0+8.0)*vol;
-      std::cout<<std::setprecision(3) << lat<<"\t\t"<<footprint<<"    \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
+      std::cout<<GridLogMessage<<std::setprecision(3) << lat<<"\t\t"<<footprint<<"    \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
 
     }
 
 
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "= Benchmarking SU3xSU3  z= x*y"<<std::endl;
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
-  std::cout << "----------------------------------------------------------"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "= Benchmarking SU3xSU3  z= x*y"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
+  std::cout<<GridLogMessage << "----------------------------------------------------------"<<std::endl;
 
   for(int lat=2;lat<=24;lat+=2){
 
@@ -75,15 +75,15 @@ int main (int argc, char ** argv)
       
       double bytes=3*vol*Nc*Nc*sizeof(Complex);
       double flops=Nc*Nc*(6+8+8)*vol;
-      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"    \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
+      std::cout<<GridLogMessage<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"    \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
 
     }
 
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "= Benchmarking SU3xSU3  mult(z,x,y)"<<std::endl;
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
-  std::cout << "----------------------------------------------------------"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "= Benchmarking SU3xSU3  mult(z,x,y)"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
+  std::cout<<GridLogMessage << "----------------------------------------------------------"<<std::endl;
 
   for(int lat=2;lat<=24;lat+=2){
 
@@ -106,15 +106,15 @@ int main (int argc, char ** argv)
       
       double bytes=3*vol*Nc*Nc*sizeof(Complex);
       double flops=Nc*Nc*(6+8+8)*vol;
-      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"    \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
+      std::cout<<GridLogMessage<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"    \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
 
     }
 
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "= Benchmarking SU3xSU3  mac(z,x,y)"<<std::endl;
-  std::cout << "===================================================================================================="<<std::endl;
-  std::cout << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
-  std::cout << "----------------------------------------------------------"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "= Benchmarking SU3xSU3  mac(z,x,y)"<<std::endl;
+  std::cout<<GridLogMessage << "===================================================================================================="<<std::endl;
+  std::cout<<GridLogMessage << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
+  std::cout<<GridLogMessage << "----------------------------------------------------------"<<std::endl;
 
   for(int lat=2;lat<=24;lat+=2){
 
@@ -137,7 +137,7 @@ int main (int argc, char ** argv)
       
       double bytes=3*vol*Nc*Nc*sizeof(Complex);
       double flops=Nc*Nc*(8+8+8)*vol;
-      std::cout<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"   \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
+      std::cout<<GridLogMessage<<std::setprecision(3) << lat<<"\t\t"<<bytes<<"   \t\t"<<bytes/time<<"\t\t" << flops/time<<std::endl;
 
     }
 
diff --git a/benchmarks/Benchmark_wilson.cc b/benchmarks/Benchmark_wilson.cc
index aa34ccf4..7d23dec3 100644
--- a/benchmarks/Benchmark_wilson.cc
+++ b/benchmarks/Benchmark_wilson.cc
@@ -28,10 +28,10 @@ int main (int argc, char ** argv)
   GridRedBlackCartesian     RBGrid(latt_size,simd_layout,mpi_layout);
 
   int threads = GridThread::GetThreads();
-  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
-  std::cout << "Grid floating point word size is REALF"<< sizeof(RealF)<<std::endl;
-  std::cout << "Grid floating point word size is REALD"<< sizeof(RealD)<<std::endl;
-  std::cout << "Grid floating point word size is REAL"<< sizeof(Real)<<std::endl;
+  std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+  std::cout<<GridLogMessage << "Grid floating point word size is REALF"<< sizeof(RealF)<<std::endl;
+  std::cout<<GridLogMessage << "Grid floating point word size is REALD"<< sizeof(RealD)<<std::endl;
+  std::cout<<GridLogMessage << "Grid floating point word size is REAL"<< sizeof(Real)<<std::endl;
 
   std::vector<int> seeds({1,2,3,4});
   GridParallelRNG          pRNG(&Grid);
@@ -58,8 +58,8 @@ int main (int argc, char ** argv)
   for(int nn=0;nn<Nd;nn++){
     random(pRNG,U[nn]);
     if(0) {
-      if (nn==-1) { U[nn]=zero; std::cout << "zeroing gauge field in dir "<<nn<<std::endl; }
-      else       { U[nn] = cone;std::cout << "unit gauge field in dir "<<nn<<std::endl; }
+      if (nn==-1) { U[nn]=zero; std::cout<<GridLogMessage << "zeroing gauge field in dir "<<nn<<std::endl; }
+      else       { U[nn] = cone;std::cout<<GridLogMessage << "unit gauge field in dir "<<nn<<std::endl; }
     }
     pokeIndex<LorentzIndex>(Umu,U[nn],nn);
   }
@@ -89,7 +89,7 @@ int main (int argc, char ** argv)
   RealD mass=0.1;
   WilsonFermion Dw(Umu,Grid,RBGrid,mass);
   
-  std::cout << "Calling Dw"<<std::endl;
+  std::cout<<GridLogMessage << "Calling Dw"<<std::endl;
   int ncall=10000;
   double t0=usecond();
   for(int i=0;i<ncall;i++){
@@ -98,12 +98,12 @@ int main (int argc, char ** argv)
   double t1=usecond();
   double flops=1344*volume*ncall;
   
-  std::cout << "Called Dw"<<std::endl;
-  std::cout << "norm result "<< norm2(result)<<std::endl;
-  std::cout << "norm ref    "<< norm2(ref)<<std::endl;
-  std::cout << "mflop/s =   "<< flops/(t1-t0)<<std::endl;
+  std::cout<<GridLogMessage << "Called Dw"<<std::endl;
+  std::cout<<GridLogMessage << "norm result "<< norm2(result)<<std::endl;
+  std::cout<<GridLogMessage << "norm ref    "<< norm2(ref)<<std::endl;
+  std::cout<<GridLogMessage << "mflop/s =   "<< flops/(t1-t0)<<std::endl;
   err = ref-result; 
-  std::cout << "norm diff   "<< norm2(err)<<std::endl;
+  std::cout<<GridLogMessage << "norm diff   "<< norm2(err)<<std::endl;
 
 
   //  for(int ss=0;ss<10;ss++ ){
@@ -112,7 +112,7 @@ int main (int argc, char ** argv)
       for(int j=0;j<Nc;j++){
 	ComplexF * ref_p = (ComplexF *)&ref._odata[ss]()(i)(j);
 	ComplexF * res_p = (ComplexF *)&result._odata[ss]()(i)(j);
-	std::cout << ss<< " "<<i<<" "<<j<<" "<< (*ref_p)<<" " <<(*res_p)<<std::endl;
+	std::cout<<GridLogMessage << ss<< " "<<i<<" "<<j<<" "<< (*ref_p)<<" " <<(*res_p)<<std::endl;
       }
     }
   }
@@ -136,11 +136,11 @@ int main (int argc, char ** argv)
   }
   ref = -0.5*ref;
   Dw.Dhop(src,result,1);
-  std::cout << "Called DwDag"<<std::endl;
-  std::cout << "norm result "<< norm2(result)<<std::endl;
-  std::cout << "norm ref    "<< norm2(ref)<<std::endl;
+  std::cout<<GridLogMessage << "Called DwDag"<<std::endl;
+  std::cout<<GridLogMessage << "norm result "<< norm2(result)<<std::endl;
+  std::cout<<GridLogMessage << "norm ref    "<< norm2(ref)<<std::endl;
   err = ref-result; 
-  std::cout << "norm diff   "<< norm2(err)<<std::endl;
+  std::cout<<GridLogMessage << "norm diff   "<< norm2(err)<<std::endl;
 
   Grid_finalize();
 }
diff --git a/gcc-bug-report/broken.cc b/gcc-bug-report/broken.cc
index d23572d3..edf878c5 100644
--- a/gcc-bug-report/broken.cc
+++ b/gcc-bug-report/broken.cc
@@ -29,12 +29,12 @@ public:
 
 template<int N,class obj,typename std::enable_if<N==obj::NestLevel >::type * = nullptr > auto function(const obj &arg)-> obj
 {
-  std::cout<<"Leaf "<<obj::NestLevel<<std::endl;
+  std::cout<<GridLogMessage<<"Leaf "<<obj::NestLevel<<std::endl;
   return arg;
 }
 template<int N,class obj,typename std::enable_if<N!=obj::NestLevel >::type * = nullptr > auto function(const obj &arg)-> obj
 {
-  std::cout<<"Node "<<obj::NestLevel<<std::endl;
+  std::cout<<GridLogMessage<<"Node "<<obj::NestLevel<<std::endl;
   obj ret;
   ret.internal=function<N>(arg.internal);
   return ret;
diff --git a/lib/Grid.h b/lib/Grid.h
index aec0e9ac..751b609d 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -35,6 +35,9 @@
 
 #include <GridConfig.h>
 
+#include <GridTime.h>
+#include <GridLog.h>
+
 ////////////////////////////////////////////////////////////
 // Tunable header includes
 ////////////////////////////////////////////////////////////
diff --git a/lib/GridInit.cc b/lib/GridInit.cc
index 20371a4b..194fd589 100644
--- a/lib/GridInit.cc
+++ b/lib/GridInit.cc
@@ -25,17 +25,19 @@
 
 namespace Grid {
 
-  //////////////////////////////////////////////////////
-  // Convenience functions to access stadard command line arg
-  // driven parallelism controls
-  //////////////////////////////////////////////////////
-  static std::vector<int> Grid_default_latt;
-  static std::vector<int> Grid_default_mpi;
+//////////////////////////////////////////////////////
+// Convenience functions to access stadard command line arg
+// driven parallelism controls
+//////////////////////////////////////////////////////
+static std::vector<int> Grid_default_latt;
+static std::vector<int> Grid_default_mpi;
+int GridThread::_threads;
 
-  int GridThread::_threads;
 
-  const std::vector<int> GridDefaultSimd(int dims,int nsimd)
-  {
+const std::vector<int> &GridDefaultLatt(void)     {return Grid_default_latt;};
+const std::vector<int> &GridDefaultMpi(void)      {return Grid_default_mpi;};
+const std::vector<int> GridDefaultSimd(int dims,int nsimd)
+{
     std::vector<int> layout(dims);
     int nn=nsimd;
     for(int d=dims-1;d>=0;d--){
@@ -48,15 +50,11 @@ namespace Grid {
     }
     assert(nn==1);
     return layout;
-  }
-
+}
   
-  const std::vector<int> &GridDefaultLatt(void)     {return Grid_default_latt;};
-  const std::vector<int> &GridDefaultMpi(void)      {return Grid_default_mpi;};
-
-  ////////////////////////////////////////////////////////////
-  // Command line parsing assist for stock controls
-  ////////////////////////////////////////////////////////////
+////////////////////////////////////////////////////////////
+// Command line parsing assist for stock controls
+////////////////////////////////////////////////////////////
 std::string GridCmdOptionPayload(char ** begin, char ** end, const std::string & option)
 {
   char ** itr = std::find(begin, end, option);
@@ -70,6 +68,23 @@ bool GridCmdOptionExists(char** begin, char** end, const std::string& option)
 {
   return std::find(begin, end, option) != end;
 }
+  // Comma separated list
+void GridCmdOptionCSL(std::string str,std::vector<std::string> & vec)
+{
+  size_t pos = 0;
+  std::string token;
+  std::string delimiter(",");
+
+  vec.resize(0);
+  while ((pos = str.find(delimiter)) != std::string::npos) {
+    token = str.substr(0, pos);
+    vec.push_back(token);
+    str.erase(0, pos + delimiter.length());
+  }
+  token = str;
+  vec.push_back(token);
+  return;
+}
 
 void GridCmdOptionIntVector(std::string &str,std::vector<int> & vec)
 {
@@ -84,6 +99,7 @@ void GridCmdOptionIntVector(std::string &str,std::vector<int> & vec)
   return;
 }
 
+
 void GridParseLayout(char **argv,int argc,
 		     std::vector<int> &latt,
 		     std::vector<int> &mpi)
@@ -117,8 +133,9 @@ std::string GridCmdVectorIntToString(const std::vector<int> & vec){
   std::copy(vec.begin(), vec.end(),std::ostream_iterator<int>(oss, " "));
   return oss.str();
 }
-  /////////////////////////////////////////////////////////
-  /////////////////////////////////////////////////////////
+/////////////////////////////////////////////////////////
+//
+/////////////////////////////////////////////////////////
 void Grid_init(int *argc,char ***argv)
 {
 #ifdef GRID_COMMS_MPI
@@ -126,15 +143,33 @@ void Grid_init(int *argc,char ***argv)
 #endif
   // Parse command line args.
 
+  GridLogger::StopWatch.Start();
+
+  std::string arg;
+  std::vector<std::string> logstreams;
+  std::string defaultLog("Error,Warning,Message,Performance");
+
+  GridCmdOptionCSL(defaultLog,logstreams);
+  GridLogConfigure(logstreams);
+
   if( GridCmdOptionExists(*argv,*argv+*argc,"--help") ){
-    std::cout<<"--help : this message"<<std::endl;
-    std::cout<<"--debug-signals : catch sigsegv and print a blame report"<<std::endl;
-    std::cout<<"--debug-stdout  : print stdout from EVERY node"<<std::endl;    
-    std::cout<<"--decomposition : report on default omp,mpi and simd decomposition"<<std::endl;    
-    std::cout<<"--mpi n.n.n.n   : default MPI decomposition"<<std::endl;    
-    std::cout<<"--omp n         : default number of OMP threads"<<std::endl;    
-    std::cout<<"--grid n.n.n.n  : default Grid size"<<std::endl;    
+    std::cout<<GridLogMessage<<"--help : this message"<<std::endl;
+    std::cout<<GridLogMessage<<"--debug-signals : catch sigsegv and print a blame report"<<std::endl;
+    std::cout<<GridLogMessage<<"--debug-stdout  : print stdout from EVERY node"<<std::endl;    
+    std::cout<<GridLogMessage<<"--decomposition : report on default omp,mpi and simd decomposition"<<std::endl;    
+    std::cout<<GridLogMessage<<"--mpi n.n.n.n   : default MPI decomposition"<<std::endl;    
+    std::cout<<GridLogMessage<<"--omp n         : default number of OMP threads"<<std::endl;    
+    std::cout<<GridLogMessage<<"--grid n.n.n.n  : default Grid size"<<std::endl;    
+    std::cout<<GridLogMessage<<"--log list      : comma separted list of streams from Error,Warning,Message,Performance,Iterative,Debug"<<std::endl;    
   }
+
+  if( GridCmdOptionExists(*argv,*argv+*argc,"--log") ){
+    arg = GridCmdOptionPayload(*argv,*argv+*argc,"--log");
+    GridCmdOptionCSL(arg,logstreams);
+    GridLogConfigure(logstreams);
+  }
+
+
   if( GridCmdOptionExists(*argv,*argv+*argc,"--debug-signals") ){
     Grid_debug_handler_init();
   }
@@ -152,38 +187,18 @@ void Grid_init(int *argc,char ***argv)
 		  Grid_default_latt,
 		  Grid_default_mpi);
   if( GridCmdOptionExists(*argv,*argv+*argc,"--decomposition") ){
-    std::cout<<"Grid Decomposition\n";
-    std::cout<<"\tOpenMP threads : "<<GridThread::GetThreads()<<std::endl;
-    std::cout<<"\tMPI tasks      : "<<GridCmdVectorIntToString(GridDefaultMpi())<<std::endl;
-    std::cout<<"\tvRealF         : "<<sizeof(vRealF)*8    <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vRealF::Nsimd()))<<std::endl;
-    std::cout<<"\tvRealD         : "<<sizeof(vRealD)*8    <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vRealD::Nsimd()))<<std::endl;
-    std::cout<<"\tvComplexF      : "<<sizeof(vComplexF)*8 <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vComplexF::Nsimd()))<<std::endl;
-    std::cout<<"\tvComplexD      : "<<sizeof(vComplexD)*8 <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vComplexD::Nsimd()))<<std::endl;
+    std::cout<<GridLogMessage<<"Grid Decomposition\n";
+    std::cout<<GridLogMessage<<"\tOpenMP threads : "<<GridThread::GetThreads()<<std::endl;
+    std::cout<<GridLogMessage<<"\tMPI tasks      : "<<GridCmdVectorIntToString(GridDefaultMpi())<<std::endl;
+    std::cout<<GridLogMessage<<"\tvRealF         : "<<sizeof(vRealF)*8    <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vRealF::Nsimd()))<<std::endl;
+    std::cout<<GridLogMessage<<"\tvRealD         : "<<sizeof(vRealD)*8    <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vRealD::Nsimd()))<<std::endl;
+    std::cout<<GridLogMessage<<"\tvComplexF      : "<<sizeof(vComplexF)*8 <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vComplexF::Nsimd()))<<std::endl;
+    std::cout<<GridLogMessage<<"\tvComplexD      : "<<sizeof(vComplexD)*8 <<"bits ; " <<GridCmdVectorIntToString(GridDefaultSimd(4,vComplexD::Nsimd()))<<std::endl;
   }
 
+
 }
 
-
-  ////////////////////////////////////////////////////////////
-  // Verbose limiter on MPI tasks
-  ////////////////////////////////////////////////////////////
-  void Grid_quiesce_nodes(void)
-  {
-#ifdef GRID_COMMS_MPI
-    int me;
-    MPI_Comm_rank(MPI_COMM_WORLD,&me);
-    if ( me ) { 
-      std::cout.setstate(std::ios::badbit);
-    }
-#endif
-  }
-  void Grid_unquiesce_nodes(void)
-  {
-#ifdef GRID_COMMS_MPI
-    std::cout.clear();
-#endif
-  }
-
   
 void Grid_finalize(void)
 {
diff --git a/lib/GridLog.cc b/lib/GridLog.cc
new file mode 100644
index 00000000..b171a828
--- /dev/null
+++ b/lib/GridLog.cc
@@ -0,0 +1,62 @@
+#include <Grid.h>
+
+namespace Grid {
+
+GridStopWatch GridLogger::StopWatch;
+std::ostream  GridLogger::devnull(0);
+
+GridLogger GridLogError      (1,"Error");
+GridLogger GridLogWarning    (1,"Warning");
+GridLogger GridLogMessage    (1,"Message");
+GridLogger GridLogDebug      (1,"Debug");
+GridLogger GridLogPerformance(1,"Performance");
+GridLogger GridLogIterative  (1,"Iterative");
+
+void GridLogConfigure(std::vector<std::string> &logstreams)
+{
+  GridLogError.Active(0);
+  GridLogWarning.Active(0);
+  GridLogMessage.Active(0);
+  GridLogIterative.Active(0);
+  GridLogDebug.Active(0);
+  GridLogPerformance.Active(0);
+
+  for(int i=0;i<logstreams.size();i++){
+    if ( logstreams[i]== std::string("Error")       ) GridLogError.Active(1);
+    if ( logstreams[i]== std::string("Warning")     ) GridLogWarning.Active(1);
+    if ( logstreams[i]== std::string("Message")     ) GridLogMessage.Active(1);
+    if ( logstreams[i]== std::string("Iterative")   ) GridLogIterative.Active(1);
+    if ( logstreams[i]== std::string("Debug")       ) GridLogDebug.Active(1);
+    if ( logstreams[i]== std::string("Performance") ) GridLogPerformance.Active(1);
+  }
+}
+
+////////////////////////////////////////////////////////////
+// Verbose limiter on MPI tasks
+////////////////////////////////////////////////////////////
+void Grid_quiesce_nodes(void)
+{
+#ifdef GRID_COMMS_MPI
+  int me;
+  MPI_Comm_rank(MPI_COMM_WORLD,&me);
+  if ( me ) { 
+    std::cout<<GridLogMessage.setstate(std::ios::badbit);
+  }
+#endif
+}
+
+void Grid_unquiesce_nodes(void)
+{
+#ifdef GRID_COMMS_MPI
+    std::cout<<GridLogMessage.clear();
+#endif
+}
+
+std::ostream& operator<< (std::ostream& stream, const GridTime& time)
+{
+  stream << time.count()<<" ms";
+  return stream;
+}
+
+}
+
diff --git a/lib/GridLog.h b/lib/GridLog.h
new file mode 100644
index 00000000..4cf094bc
--- /dev/null
+++ b/lib/GridLog.h
@@ -0,0 +1,46 @@
+#ifndef GRID_LOG_H
+#define GRID_LOG_H
+namespace Grid {
+
+// Dress the output; use std::chrono for time stamping via the StopWatch class
+
+std::ostream& operator<< (std::ostream& stream, const GridTime& time);
+
+class GridLogger { 
+  int active;
+  std::string name;
+public:
+
+  static GridStopWatch StopWatch;
+  static std::ostream devnull;
+  
+  GridLogger(int on, std::string nm): active(on), name(nm) { 
+  };
+  
+  void Active(int on) {active = on;};
+
+  friend std::ostream& operator<< (std::ostream& stream, const GridLogger& log){
+    if ( log.active ) {
+      StopWatch.Stop();
+      GridTime now = StopWatch.Elapsed();
+      StopWatch.Start();
+      stream << "Grid : "<<log.name << " : " << now << " : ";
+      return stream;
+    } else { 
+      return devnull;
+    }
+  }
+
+};
+
+void GridLogConfigure(std::vector<std::string> &logstreams);
+
+extern GridLogger GridLogError;
+extern GridLogger GridLogWarning;
+extern GridLogger GridLogMessage;
+extern GridLogger GridLogDebug  ;
+extern GridLogger GridLogPerformance;
+extern GridLogger GridLogIterative  ;
+
+}
+#endif
diff --git a/lib/GridTime.h b/lib/GridTime.h
new file mode 100644
index 00000000..841ace11
--- /dev/null
+++ b/lib/GridTime.h
@@ -0,0 +1,44 @@
+#ifndef GRID_TIME_H
+#define GRID_TIME_H
+namespace Grid {
+  // Dress the output; use std::chrono
+#include <chrono>
+#include <ctime>
+
+typedef  std::chrono::system_clock          GridClock;
+typedef  std::chrono::time_point<GridClock> GridTimePoint;
+typedef  std::chrono::milliseconds          GridTime;
+
+ 
+class GridStopWatch {
+private:
+  bool running;
+  GridTimePoint start;
+  GridTime accumulator;
+public:
+  GridStopWatch () { 
+    Reset();
+  }
+  void     Start(void) { 
+    assert(running == false);
+    start = GridClock::now(); 
+    running = true;
+  }
+  void     Stop(void)  { 
+    assert(running == true);
+    accumulator+= std::chrono::duration_cast<GridTime>(GridClock::now()-start); 
+    running = false; 
+  };
+  void     Reset(void){
+    running = false;
+    start = GridClock::now();
+    accumulator = std::chrono::duration_cast<GridTime>(start-start); 
+  }
+  GridTime Elapsed(void) {
+    assert(running == false);
+    return accumulator;
+  }
+};
+
+}
+#endif
diff --git a/lib/Make.inc b/lib/Make.inc
index dfa8401b..621ef7e7 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./Cshift.h ./simd/Grid_avx.h ./simd/Grid_vector_types.h ./simd/Grid_sse4.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_vector_unops.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./Tensors.h ./Algorithms.h ./communicator/Communicator_base.h ./lattice/Lattice_rng.h ./lattice/Lattice_reduction.h ./lattice/Lattice_transfer.h ./lattice/Lattice_unary.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_comparison.h ./lattice/Lattice_overload.h ./lattice/Lattice_reality.h ./lattice/Lattice_local.h ./lattice/Lattice_conformable.h ./lattice/Lattice_where.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_ET.h ./lattice/Lattice_transpose.h ./lattice/Lattice_trace.h ./Stencil.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_exp.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_class.h ./tensors/Tensor_logical.h ./tensors/Tensor_transpose.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_reality.h ./tensors/Tensor_trace.h ./tensors/Tensor_index.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_outer.h ./tensors/Tensor_inner.h ./tensors/Tensor_traits.h ./tensors/Tensor_Ta.h ./tensors/Tensor_unary.h ./tensors/Tensor_determinant.h ./tensors/Tensor_arith.h ./tensors/Tensor_extract_merge.h ./Communicator.h ./Cartesian.h ./parallelIO/NerscIO.h ./qcd/QCD.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/hmc/HMC.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/LinalgUtils.h ./qcd/utils/CovariantCshift.h ./qcd/utils/WilsonLoops.h ./qcd/action/ActionBase.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/spin/TwoSpinor.h ./qcd/spin/Dirac.h ./cshift/Cshift_common.h ./cshift/Cshift_none.h ./cshift/Cshift_mpi.h ./Simd.h ./GridConfig.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_red_black.h ./cartesian/Cartesian_full.h ./AlignedAllocator.h ./Lattice.h ./Old/Tensor_poke.h ./Old/Tensor_peek.h ./Threads.h ./Grid.h ./algorithms/Preconditioner.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/AdefGeneric.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/Zolotarev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Remez.h ./algorithms/LinearOperator.h ./algorithms/SparseMatrix.h ./algorithms/CoarsenedMatrix.h ./stencil/Lebesgue.h
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/BfmHDCG.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/GeneralisedMinimumResidual.h ./algorithms/iterative/gmres.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./GridLog.h ./GridTime.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./Old/Tensor_peek.h ./Old/Tensor_poke.h ./parallelIO/NerscIO.h ./qcd/action/ActionBase.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/hmc/HMC.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/hmc/integrators/Integrator_base.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_index.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
 
-CCFILES=./qcd/hmc/integrators/Integrator.cc ./qcd/hmc/HMC.cc ./qcd/utils/SpaceTimeGrid.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/spin/Dirac.cc ./GridInit.cc ./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
+CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./GridLog.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/hmc/HMC.cc ./qcd/hmc/integrators/Integrator.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/algorithms/CoarsenedMatrix.h b/lib/algorithms/CoarsenedMatrix.h
index 998292b2..c4b039f1 100644
--- a/lib/algorithms/CoarsenedMatrix.h
+++ b/lib/algorithms/CoarsenedMatrix.h
@@ -32,12 +32,12 @@ namespace Grid {
       displacements[2*_d]=0;
       
       //// report back
-      std::cout<<"directions    :";
+      std::cout<<GridLogMessage<<"directions    :";
       for(int d=0;d<npoint;d++) std::cout<< directions[d]<< " ";
       std::cout <<std::endl;
-      std::cout<<"displacements :";
+      std::cout<<GridLogMessage<<"displacements :";
       for(int d=0;d<npoint;d++) std::cout<< displacements[d]<< " ";
-      std::cout <<std::endl;
+      std::cout<<std::endl;
     }
   
     /*
@@ -100,9 +100,9 @@ namespace Grid {
 	  eProj._odata[ss](i)=CComplex(1.0);
 	}
 	eProj=eProj - iProj;
-	std::cout<<"Orthog check error "<<i<<" " << norm2(eProj)<<std::endl;
+	std::cout<<GridLogMessage<<"Orthog check error "<<i<<" " << norm2(eProj)<<std::endl;
       }
-      std::cout <<"CheckOrthog done"<<std::endl;
+      std::cout<<GridLogMessage <<"CheckOrthog done"<<std::endl;
     }
     void ProjectToSubspace(CoarseVector &CoarseVec,const FineField &FineVec){
       blockProject(CoarseVec,FineVec,subspace);
@@ -113,7 +113,7 @@ namespace Grid {
     void CreateSubspaceRandom(GridParallelRNG &RNG){
       for(int i=0;i<nbasis;i++){
 	random(RNG,subspace[i]);
-	std::cout<<" norm subspace["<<i<<"] "<<norm2(subspace[i])<<std::endl;
+	std::cout<<GridLogMessage<<" norm subspace["<<i<<"] "<<norm2(subspace[i])<<std::endl;
       }
       Orthogonalise();
     }
@@ -121,7 +121,7 @@ namespace Grid {
 
       RealD scale;
 
-      ConjugateGradient<FineField> CG(2.0e-3,10000);
+      ConjugateGradient<FineField> CG(1.0e-2,10000);
       FineField noise(FineGrid);
       FineField Mn(FineGrid);
 
@@ -131,7 +131,7 @@ namespace Grid {
 	scale = std::pow(norm2(noise),-0.5); 
 	noise=noise*scale;
 
-	hermop.Op(noise,Mn); std::cout << "noise   ["<<b<<"] <n|MdagM|n> "<<norm2(Mn)<<std::endl;
+	hermop.Op(noise,Mn); std::cout<<GridLogMessage << "noise   ["<<b<<"] <n|MdagM|n> "<<norm2(Mn)<<std::endl;
 
 	for(int i=0;i<1;i++){
 
@@ -143,7 +143,7 @@ namespace Grid {
 
 	}
 
-	hermop.Op(noise,Mn); std::cout << "filtered["<<b<<"] <f|MdagM|f> "<<norm2(Mn)<<std::endl;
+	hermop.Op(noise,Mn); std::cout<<GridLogMessage << "filtered["<<b<<"] <f|MdagM|f> "<<norm2(Mn)<<std::endl;
 	subspace[b]   = noise;
 
       }
@@ -189,7 +189,7 @@ namespace Grid {
       SimpleCompressor<siteVector> compressor;
       Stencil.HaloExchange(in,comm_buf,compressor);
 
-      //PARALLEL_FOR_LOOP
+PARALLEL_FOR_LOOP
       for(int ss=0;ss<Grid()->oSites();ss++){
         siteVector res = zero;
 	siteVector nbr;
@@ -252,10 +252,6 @@ namespace Grid {
 
       // Orthogonalise the subblocks over the basis
       blockOrthogonalise(InnerProd,Subspace.subspace);
-      //Subspace.Orthogonalise();
-      //      Subspace.CheckOrthogonal();
-      //Subspace.Orthogonalise();
-      //      Subspace.CheckOrthogonal();
 
       // Compute the matrix elements of linop between this orthonormal
       // set of vectors.
@@ -306,6 +302,7 @@ namespace Grid {
 	  Subspace.ProjectToSubspace(oProj,oblock);
 	  //	  blockProject(iProj,iblock,Subspace.subspace);
 	  //	  blockProject(oProj,oblock,Subspace.subspace);
+PARALLEL_FOR_LOOP
 	  for(int ss=0;ss<Grid()->oSites();ss++){
 	    for(int j=0;j<nbasis;j++){
 	      if( disp!= 0 ) {
@@ -321,12 +318,12 @@ namespace Grid {
       ///////////////////////////
       // test code worth preserving in if block
       ///////////////////////////
-      std::cout<< " Computed matrix elements "<< self_stencil <<std::endl;
+      std::cout<<GridLogMessage<< " Computed matrix elements "<< self_stencil <<std::endl;
       for(int p=0;p<geom.npoint;p++){
-	std::cout<< "A["<<p<<"]" << std::endl;
-	std::cout<< A[p] << std::endl;
+	std::cout<<GridLogMessage<< "A["<<p<<"]" << std::endl;
+	std::cout<<GridLogMessage<< A[p] << std::endl;
       }
-      std::cout<< " picking by block0 "<< self_stencil <<std::endl;
+      std::cout<<GridLogMessage<< " picking by block0 "<< self_stencil <<std::endl;
 
       phi=Subspace.subspace[0];
       std::vector<int> bc(FineGrid->_ndimension,0);
@@ -334,9 +331,9 @@ namespace Grid {
       blockPick(Grid(),phi,tmp,bc);      // Pick out a block
       linop.Op(tmp,Mphi);                // Apply big dop
       blockProject(iProj,Mphi,Subspace.subspace); // project it and print it
-      std::cout<< " Computed matrix elements from block zero only "<<std::endl;
-      std::cout<< iProj <<std::endl;
-      std::cout<<"Computed Coarse Operator"<<std::endl;
+      std::cout<<GridLogMessage<< " Computed matrix elements from block zero only "<<std::endl;
+      std::cout<<GridLogMessage<< iProj <<std::endl;
+      std::cout<<GridLogMessage<<"Computed Coarse Operator"<<std::endl;
 #endif
       //      ForceHermitian();
       AssertHermitian();
@@ -345,9 +342,9 @@ namespace Grid {
     void ForceDiagonal(void) {
 
 
-      std::cout<<"**************************************************"<<std::endl;
-      std::cout<<"****   Forcing coarse operator to be diagonal ****"<<std::endl;
-      std::cout<<"**************************************************"<<std::endl;
+      std::cout<<GridLogMessage<<"**************************************************"<<std::endl;
+      std::cout<<GridLogMessage<<"****   Forcing coarse operator to be diagonal ****"<<std::endl;
+      std::cout<<GridLogMessage<<"**************************************************"<<std::endl;
       for(int p=0;p<8;p++){
 	A[p]=zero;
       }
@@ -387,13 +384,13 @@ namespace Grid {
 	
 	Diff = AA - adj(AAc);
 
-	std::cout<<"Norm diff dim "<<d<<" "<< norm2(Diff)<<std::endl;
-	std::cout<<"Norm dim "<<d<<" "<< norm2(AA)<<std::endl;
+	std::cout<<GridLogMessage<<"Norm diff dim "<<d<<" "<< norm2(Diff)<<std::endl;
+	std::cout<<GridLogMessage<<"Norm dim "<<d<<" "<< norm2(AA)<<std::endl;
 	  
       }
       Diff = A[8] - adj(A[8]);
-      std::cout<<"Norm diff local "<< norm2(Diff)<<std::endl;
-      std::cout<<"Norm local "<< norm2(A[8])<<std::endl;
+      std::cout<<GridLogMessage<<"Norm diff local "<< norm2(Diff)<<std::endl;
+      std::cout<<GridLogMessage<<"Norm local "<< norm2(A[8])<<std::endl;
     }
     
   };
diff --git a/lib/algorithms/approx/Chebyshev.h b/lib/algorithms/approx/Chebyshev.h
index f629ab08..e4fa0b61 100644
--- a/lib/algorithms/approx/Chebyshev.h
+++ b/lib/algorithms/approx/Chebyshev.h
@@ -120,7 +120,7 @@ namespace Grid {
       Field *Tn  = &T1;
       Field *Tnp = &T2;
 
-      std::cout << "Chebyshev ["<<lo<<","<<hi<<"]"<< " order "<<order <<std::endl;
+      std::cout<<GridLogMessage << "Chebyshev ["<<lo<<","<<hi<<"]"<< " order "<<order <<std::endl;
       // Tn=T1 = (xscale M + mscale)in
       double xscale = 2.0/(hi-lo);
       double mscale = -(hi+lo)/(hi-lo);
diff --git a/lib/algorithms/approx/Remez.cc b/lib/algorithms/approx/Remez.cc
index 4ec2270d..38d60088 100644
--- a/lib/algorithms/approx/Remez.cc
+++ b/lib/algorithms/approx/Remez.cc
@@ -757,3 +757,4 @@ void AlgRemez::csv(std::ostream & os)
   }
   return;
 }
+
diff --git a/lib/algorithms/approx/Remez.h b/lib/algorithms/approx/Remez.h
index 336382d2..4e8ae3c9 100644
--- a/lib/algorithms/approx/Remez.h
+++ b/lib/algorithms/approx/Remez.h
@@ -28,6 +28,7 @@
   remez.getIPFE(res,pole,&norm);
   remez.csv(ostream &os);
  */
+
 class AlgRemez
 {
  private:
diff --git a/lib/algorithms/iterative/AdefGeneric.h b/lib/algorithms/iterative/AdefGeneric.h
index 0d1c3d91..6b839d7a 100644
--- a/lib/algorithms/iterative/AdefGeneric.h
+++ b/lib/algorithms/iterative/AdefGeneric.h
@@ -149,7 +149,7 @@ class TwoLevelFlexiblePcg : public LinearFunction<Field>
       }
 
       RealD rrn=sqrt(rn/ssq);
-      std::cout<<"TwoLevelfPcg: k= "<<k<<" residual = "<<rrn<<std::endl;
+      std::cout<<GridLogMessage<<"TwoLevelfPcg: k= "<<k<<" residual = "<<rrn<<std::endl;
 
       // Stopping condition
       if ( rn <= rsq ) { 
@@ -161,8 +161,8 @@ class TwoLevelFlexiblePcg : public LinearFunction<Field>
 	RealD srcnorm = sqrt(norm2(src));
 	RealD tmpnorm = sqrt(norm2(tmp));
 	RealD true_residual = tmpnorm/srcnorm;
-	std::cout<<"TwoLevelfPcg:   true residual is "<<true_residual<<std::endl;
-	std::cout<<"TwoLevelfPcg: target residual was"<<Tolerance<<std::endl;
+	std::cout<<GridLogMessage<<"TwoLevelfPcg:   true residual is "<<true_residual<<std::endl;
+	std::cout<<GridLogMessage<<"TwoLevelfPcg: target residual was"<<Tolerance<<std::endl;
 	return k;
       }
     }
diff --git a/lib/algorithms/iterative/ConjugateGradient.h b/lib/algorithms/iterative/ConjugateGradient.h
index e281d765..70b89f90 100644
--- a/lib/algorithms/iterative/ConjugateGradient.h
+++ b/lib/algorithms/iterative/ConjugateGradient.h
@@ -43,12 +43,12 @@ public:
       ssq=norm2(src);
 
       if ( verbose ) {
-	std::cout <<std::setprecision(4)<< "ConjugateGradient: guess "<<guess<<std::endl;
-	std::cout <<std::setprecision(4)<< "ConjugateGradient:   src "<<ssq  <<std::endl;
-	std::cout <<std::setprecision(4)<< "ConjugateGradient:    mp "<<d    <<std::endl;
-	std::cout <<std::setprecision(4)<< "ConjugateGradient:   mmp "<<b    <<std::endl;
-	std::cout <<std::setprecision(4)<< "ConjugateGradient:  cp,r "<<cp   <<std::endl;
-	std::cout <<std::setprecision(4)<< "ConjugateGradient:     p "<<a    <<std::endl;
+	std::cout<<GridLogMessage <<std::setprecision(4)<< "ConjugateGradient: guess "<<guess<<std::endl;
+	std::cout<<GridLogMessage <<std::setprecision(4)<< "ConjugateGradient:   src "<<ssq  <<std::endl;
+	std::cout<<GridLogMessage <<std::setprecision(4)<< "ConjugateGradient:    mp "<<d    <<std::endl;
+	std::cout<<GridLogMessage <<std::setprecision(4)<< "ConjugateGradient:   mmp "<<b    <<std::endl;
+	std::cout<<GridLogMessage <<std::setprecision(4)<< "ConjugateGradient:  cp,r "<<cp   <<std::endl;
+	std::cout<<GridLogMessage <<std::setprecision(4)<< "ConjugateGradient:     p "<<a    <<std::endl;
       }
 
       RealD rsq =  Tolerance* Tolerance*ssq;
@@ -58,7 +58,7 @@ public:
 	return;
       }
       
-      if(verbose) std::cout << std::setprecision(4)<< "ConjugateGradient: k=0 residual "<<cp<<" rsq"<<rsq<<std::endl;
+      if(verbose) std::cout<<GridLogMessage << std::setprecision(4)<< "ConjugateGradient: k=0 residual "<<cp<<" rsq"<<rsq<<std::endl;
       
       int k;
       for (k=1;k<=MaxIterations;k++){
@@ -80,7 +80,7 @@ public:
 	psi= a*p+psi;
 	p  = p*b+r;
 	  
-	if (verbose) std::cout<<"ConjugateGradient: Iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
+	if (verbose) std::cout<<GridLogMessage<<"ConjugateGradient: Iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
 	
 	// Stopping condition
 	if ( cp <= rsq ) { 
@@ -94,14 +94,14 @@ public:
 	  RealD resnorm = sqrt(norm2(p));
 	  RealD true_residual = resnorm/srcnorm;
 
-	  std::cout<<"ConjugateGradient: Converged on iteration " <<k
+	  std::cout<<GridLogMessage<<"ConjugateGradient: Converged on iteration " <<k
 		   <<" computed residual "<<sqrt(cp/ssq)
 		   <<" true residual     "<<true_residual
 		   <<" target "<<Tolerance<<std::endl;
 	  return;
 	}
       }
-      std::cout<<"ConjugateGradient did NOT converge"<<std::endl;
+      std::cout<<GridLogMessage<<"ConjugateGradient did NOT converge"<<std::endl;
       assert(0);
     }
   };
diff --git a/lib/algorithms/iterative/ConjugateGradientMultiShift.h b/lib/algorithms/iterative/ConjugateGradientMultiShift.h
index 580ab838..c6e7650d 100644
--- a/lib/algorithms/iterative/ConjugateGradientMultiShift.h
+++ b/lib/algorithms/iterative/ConjugateGradientMultiShift.h
@@ -91,7 +91,7 @@ void operator() (LinearOperatorBase<Field> &Linop, const Field &src, std::vector
   cp = norm2(src);
   for(int s=0;s<nshift;s++){
     rsq[s] = cp * mresidual[s] * mresidual[s];
-    std::cout<<"ConjugateGradientMultiShift: shift "<<s
+    std::cout<<GridLogMessage<<"ConjugateGradientMultiShift: shift "<<s
 	     <<" target resid "<<rsq[s]<<std::endl;
     ps[s] = src;
   }
@@ -109,7 +109,7 @@ void operator() (LinearOperatorBase<Field> &Linop, const Field &src, std::vector
   // p and mmp is equal to d after this since
   // the d computation is tricky
   //  qq = real(innerProduct(p,mmp));
-  //  std::cout << "debug equal ?  qq "<<qq<<" d "<< d<<std::endl;
+  //  std::cout<<GridLogMessage << "debug equal ?  qq "<<qq<<" d "<< d<<std::endl;
   
   b = -cp /d;
   
@@ -214,7 +214,7 @@ void operator() (LinearOperatorBase<Field> &Linop, const Field &src, std::vector
 	
 	if(css<rsq[s]){
 	  if ( ! converged[s] )
-	    std::cout<<"ConjugateGradientMultiShift k="<<k<<" Shift "<<s<<" has converged"<<std::endl;
+	    std::cout<<GridLogMessage<<"ConjugateGradientMultiShift k="<<k<<" Shift "<<s<<" has converged"<<std::endl;
 	      converged[s]=1;
 	} else {
 	  all_converged=0;
@@ -225,8 +225,8 @@ void operator() (LinearOperatorBase<Field> &Linop, const Field &src, std::vector
     
     if ( all_converged ){
 
-      std::cout<< "CGMultiShift: All shifts have converged iteration "<<k<<std::endl;
-      std::cout<< "CGMultiShift: Checking solutions"<<std::endl;
+      std::cout<<GridLogMessage<< "CGMultiShift: All shifts have converged iteration "<<k<<std::endl;
+      std::cout<<GridLogMessage<< "CGMultiShift: Checking solutions"<<std::endl;
       
       // Check answers 
       for(int s=0; s < nshift; s++) { 
@@ -235,13 +235,13 @@ void operator() (LinearOperatorBase<Field> &Linop, const Field &src, std::vector
 	axpy(r,-alpha[s],src,tmp);
 	RealD rn = norm2(r);
 	RealD cn = norm2(src);
-	std::cout<<"CGMultiShift: shift["<<s<<"] true residual "<<std::sqrt(rn/cn)<<std::endl;
+	std::cout<<GridLogMessage<<"CGMultiShift: shift["<<s<<"] true residual "<<std::sqrt(rn/cn)<<std::endl;
       }
       return;
     }
   }
   // ugly hack
-  std::cout<<"CG multi shift did not converge"<<std::endl;
+  std::cout<<GridLogMessage<<"CG multi shift did not converge"<<std::endl;
   assert(0);
 }
 
diff --git a/lib/algorithms/iterative/ConjugateResidual.h b/lib/algorithms/iterative/ConjugateResidual.h
index 197a0b15..1c3c6a91 100644
--- a/lib/algorithms/iterative/ConjugateResidual.h
+++ b/lib/algorithms/iterative/ConjugateResidual.h
@@ -41,7 +41,7 @@ namespace Grid {
       ssq=norm2(src);
       rsq=Tolerance*Tolerance*ssq;
 
-      if (verbose) std::cout<<"ConjugateResidual: iteration " <<0<<" residual "<<cp<< " target"<< rsq<<std::endl;
+      if (verbose) std::cout<<GridLogMessage<<"ConjugateResidual: iteration " <<0<<" residual "<<cp<< " target"<< rsq<<std::endl;
 
       for(int k=1;k<MaxIterations;k++){
 
@@ -60,13 +60,13 @@ namespace Grid {
 	axpy(p,b,p,r);
 	pAAp=axpy_norm(Ap,b,Ap,Ar);
 	
-	if(verbose) std::cout<<"ConjugateResidual: iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
+	if(verbose) std::cout<<GridLogMessage<<"ConjugateResidual: iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
 
 	if(cp<rsq) {
 	  Linop.HermOp(psi,Ap);
 	  axpy(r,-1.0,src,Ap);
 	  RealD true_resid = norm2(r)/ssq;
-	  std::cout<<"ConjugateResidual: Converged on iteration " <<k
+	  std::cout<<GridLogMessage<<"ConjugateResidual: Converged on iteration " <<k
 		   << " computed residual "<<sqrt(cp/ssq)
 	           << " true residual "<<sqrt(true_resid)
 	           << " target "       <<Tolerance <<std::endl;
@@ -75,7 +75,7 @@ namespace Grid {
 
       }
 
-      std::cout<<"ConjugateResidual did NOT converge"<<std::endl;
+      std::cout<<GridLogMessage<<"ConjugateResidual did NOT converge"<<std::endl;
       assert(0);
     }
   };
diff --git a/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h b/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h
index fee41fa8..3a9bb731 100644
--- a/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h
+++ b/lib/algorithms/iterative/PrecGeneralisedConjugateResidual.h
@@ -45,13 +45,13 @@ namespace Grid {
 
 	cp=GCRnStep(Linop,src,psi,rsq);
 
-	if ( verbose ) std::cout<<"VPGCR("<<mmax<<","<<nstep<<") "<< steps <<" steps cp = "<<cp<<std::endl;
+	if ( verbose ) std::cout<<GridLogMessage<<"VPGCR("<<mmax<<","<<nstep<<") "<< steps <<" steps cp = "<<cp<<std::endl;
 
 	if(cp<rsq) {
 	  Linop.HermOp(psi,r);
 	  axpy(r,-1.0,src,r);
 	  RealD tr = norm2(r);
-	  std::cout<<"PrecGeneralisedConjugateResidual: Converged on iteration " <<steps
+	  std::cout<<GridLogMessage<<"PrecGeneralisedConjugateResidual: Converged on iteration " <<steps
 		   << " computed residual "<<sqrt(cp/ssq)
 	           << " true residual "    <<sqrt(tr/ssq)
 	           << " target "           <<Tolerance <<std::endl;
@@ -59,7 +59,7 @@ namespace Grid {
 	}
 
       }
-      std::cout<<"Variable Preconditioned GCR did not converge"<<std::endl;
+      std::cout<<GridLogMessage<<"Variable Preconditioned GCR did not converge"<<std::endl;
       assert(0);
     }
     RealD GCRnStep(LinearOperatorBase<Field> &Linop,const Field &src, Field &psi,RealD rsq){
@@ -96,21 +96,21 @@ namespace Grid {
       /////////////////////
       Preconditioner(r,z);
 
-      std::cout<< " Preconditioner in " << norm2(r)<<std::endl; 
-      std::cout<< " Preconditioner out " << norm2(z)<<std::endl; 
+      std::cout<<GridLogMessage<< " Preconditioner in " << norm2(r)<<std::endl; 
+      std::cout<<GridLogMessage<< " Preconditioner out " << norm2(z)<<std::endl; 
       
       Linop.HermOp(z,tmp); 
 
-      std::cout<< " Preconditioner Aout " << norm2(tmp)<<std::endl; 
+      std::cout<<GridLogMessage<< " Preconditioner Aout " << norm2(tmp)<<std::endl; 
       ttmp=tmp;
       tmp=tmp-r;
 
-      std::cout<< " Preconditioner resid " << std::sqrt(norm2(tmp)/norm2(r))<<std::endl; 
+      std::cout<<GridLogMessage<< " Preconditioner resid " << std::sqrt(norm2(tmp)/norm2(r))<<std::endl; 
       /*
-      std::cout<<r<<std::endl;
-      std::cout<<z<<std::endl;
-      std::cout<<ttmp<<std::endl;
-      std::cout<<tmp<<std::endl;
+      std::cout<<GridLogMessage<<r<<std::endl;
+      std::cout<<GridLogMessage<<z<<std::endl;
+      std::cout<<GridLogMessage<<ttmp<<std::endl;
+      std::cout<<GridLogMessage<<tmp<<std::endl;
       */
 
       Linop.HermOpAndNorm(z,Az,zAz,zAAz); 
@@ -137,7 +137,7 @@ namespace Grid {
 
 	cp = axpy_norm(r,-a,q[peri_k],r);  
 
-	std::cout<< " VPGCR_step resid" <<sqrt(cp/rsq)<<std::endl; 
+	std::cout<<GridLogMessage<< " VPGCR_step resid" <<sqrt(cp/rsq)<<std::endl; 
 	if((k==nstep-1)||(cp<rsq)){
 	  return cp;
 	}
@@ -148,7 +148,7 @@ namespace Grid {
 
 	Linop.HermOp(z,tmp);
         tmp=tmp-r;
-	std::cout<< " Preconditioner resid" <<sqrt(norm2(tmp)/norm2(r))<<std::endl; 
+	std::cout<<GridLogMessage<< " Preconditioner resid" <<sqrt(norm2(tmp)/norm2(r))<<std::endl; 
 
 	q[peri_kp]=Az;
 	p[peri_kp]=z;
diff --git a/lib/algorithms/iterative/SchurRedBlack.h b/lib/algorithms/iterative/SchurRedBlack.h
index 3ebc5f1d..b023fb7b 100644
--- a/lib/algorithms/iterative/SchurRedBlack.h
+++ b/lib/algorithms/iterative/SchurRedBlack.h
@@ -89,7 +89,7 @@ namespace Grid {
       //////////////////////////////////////////////////////////////
       // Call the red-black solver
       //////////////////////////////////////////////////////////////
-      std::cout << "SchurRedBlack solver calling the MpcDagMp solver" <<std::endl;
+      std::cout<<GridLogMessage << "SchurRedBlack solver calling the MpcDagMp solver" <<std::endl;
       _HermitianRBSolver(_HermOpEO,src_o,sol_o);  assert(sol_o.checkerboard==Odd);
 
       ///////////////////////////////////////////////////
@@ -108,7 +108,7 @@ namespace Grid {
       RealD ns = norm2(in);
       RealD nr = norm2(resid);
 
-      std::cout << "SchurRedBlackDiagMooee solver true unprec resid "<< std::sqrt(nr/ns) <<" nr "<< nr <<" ns "<<ns << std::endl;
+      std::cout<<GridLogMessage << "SchurRedBlackDiagMooee solver true unprec resid "<< std::sqrt(nr/ns) <<" nr "<< nr <<" ns "<<ns << std::endl;
     }     
   };
 
diff --git a/lib/lattice/Lattice_ET.h b/lib/lattice/Lattice_ET.h
index cbcca2ce..a7910747 100644
--- a/lib/lattice/Lattice_ET.h
+++ b/lib/lattice/Lattice_ET.h
@@ -132,18 +132,18 @@ inline void CBFromExpression(int &cb,const T1& lat)   // Lattice leaf
     assert(cb==lat.checkerboard);
   } 
   cb=lat.checkerboard;
-  //  std::cout<<"Lattice leaf cb "<<cb<<std::endl;
+  //  std::cout<<GridLogMessage<<"Lattice leaf cb "<<cb<<std::endl;
 }
 template<class T1,typename std::enable_if<!is_lattice<T1>::value, T1>::type * = nullptr >
 inline void CBFromExpression(int &cb,const T1& notlat)   // non-lattice leaf
 {
-  //  std::cout<<"Non lattice leaf cb"<<cb<<std::endl;
+  //  std::cout<<GridLogMessage<<"Non lattice leaf cb"<<cb<<std::endl;
 }
 template <typename Op, typename T1>
 inline void CBFromExpression(int &cb,const LatticeUnaryExpression<Op,T1 > &expr)
 {
   CBFromExpression(cb,std::get<0>(expr.second));// recurse 
-  //  std::cout<<"Unary node cb "<<cb<<std::endl;
+  //  std::cout<<GridLogMessage<<"Unary node cb "<<cb<<std::endl;
 }
 
 template <typename Op, typename T1, typename T2>
@@ -151,7 +151,7 @@ inline void CBFromExpression(int &cb,const LatticeBinaryExpression<Op,T1,T2> &ex
 {
   CBFromExpression(cb,std::get<0>(expr.second));// recurse
   CBFromExpression(cb,std::get<1>(expr.second));
-  //  std::cout<<"Binary node cb "<<cb<<std::endl;
+  //  std::cout<<GridLogMessage<<"Binary node cb "<<cb<<std::endl;
 }
 template <typename Op, typename T1, typename T2, typename T3>
 inline void CBFromExpression( int &cb,const LatticeTrinaryExpression<Op,T1,T2,T3 > &expr) 
@@ -159,7 +159,7 @@ inline void CBFromExpression( int &cb,const LatticeTrinaryExpression<Op,T1,T2,T3
   CBFromExpression(cb,std::get<0>(expr.second));// recurse
   CBFromExpression(cb,std::get<1>(expr.second));
   CBFromExpression(cb,std::get<2>(expr.second));
-  //  std::cout<<"Trinary node cb "<<cb<<std::endl;
+  //  std::cout<<GridLogMessage<<"Trinary node cb "<<cb<<std::endl;
 }
 
 ////////////////////////////////////////////
@@ -370,7 +370,7 @@ using namespace Grid;
    tmp.func(eval(0,v1),eval(0,v2));
 
    auto var = v1+v2;
-   std::cout<<typeid(var).name()<<std::endl;
+   std::cout<<GridLogMessage<<typeid(var).name()<<std::endl;
 
    v3=v1+v2;
    v3=v1+v2+v1*v2;
diff --git a/lib/lattice/Lattice_base.h b/lib/lattice/Lattice_base.h
index 0503732c..2a365cdd 100644
--- a/lib/lattice/Lattice_base.h
+++ b/lib/lattice/Lattice_base.h
@@ -221,7 +221,7 @@ PARALLEL_FOR_LOOP
     template<class robj> strong_inline Lattice<vobj> & operator = (const Lattice<robj> & r){
       this->checkerboard = r.checkerboard;
       conformable(*this,r);
-      std::cout<<"Lattice operator ="<<std::endl;
+      std::cout<<GridLogMessage<<"Lattice operator ="<<std::endl;
 PARALLEL_FOR_LOOP
         for(int ss=0;ss<_grid->oSites();ss++){
             this->_odata[ss]=r._odata[ss];
diff --git a/lib/lattice/Lattice_reduction.h b/lib/lattice/Lattice_reduction.h
index 922821ed..e6287b41 100644
--- a/lib/lattice/Lattice_reduction.h
+++ b/lib/lattice/Lattice_reduction.h
@@ -125,7 +125,7 @@ template<class vobj> inline void sliceSum(const Lattice<vobj> &Data,std::vector<
   assert(grid!=NULL);
 
   // FIXME
-  std::cout<<"WARNING ! SliceSum is unthreaded "<<grid->SumArraySize()<<" threads "<<std::endl;
+  std::cout<<GridLogMessage<<"WARNING ! SliceSum is unthreaded "<<grid->SumArraySize()<<" threads "<<std::endl;
 
   const int    Nd = grid->_ndimension;
   const int Nsimd = grid->Nsimd();
diff --git a/lib/lattice/Lattice_rng.h b/lib/lattice/Lattice_rng.h
index f3cc0ddc..55246016 100644
--- a/lib/lattice/Lattice_rng.h
+++ b/lib/lattice/Lattice_rng.h
@@ -233,7 +233,8 @@ namespace Grid {
       int words=sizeof(scalar_object)/sizeof(scalar_type);
 
       std::vector<scalar_object> buf(Nsimd);
-      
+
+PARALLEL_FOR_LOOP
       for(int ss=0;ss<osites;ss++){
 	for(int si=0;si<Nsimd;si++){
 
diff --git a/lib/lattice/Lattice_transfer.h b/lib/lattice/Lattice_transfer.h
index ac628faa..b5c43e43 100644
--- a/lib/lattice/Lattice_transfer.h
+++ b/lib/lattice/Lattice_transfer.h
@@ -23,7 +23,7 @@ inline void subdivides(GridBase *coarse,GridBase *fine)
   template<class vobj> inline void pickCheckerboard(int cb,Lattice<vobj> &half,const Lattice<vobj> &full){
     half.checkerboard = cb;
     int ssh=0;
-PARALLEL_FOR_LOOP
+    //PARALLEL_FOR_LOOP
     for(int ss=0;ss<full._grid->oSites();ss++){
       std::vector<int> coor;
       int cbos;
@@ -40,7 +40,7 @@ PARALLEL_FOR_LOOP
   template<class vobj> inline void setCheckerboard(Lattice<vobj> &full,const Lattice<vobj> &half){
     int cb = half.checkerboard;
     int ssh=0;
-PARALLEL_FOR_LOOP
+    //PARALLEL_FOR_LOOP
     for(int ss=0;ss<full._grid->oSites();ss++){
       std::vector<int> coor;
       int cbos;
@@ -158,6 +158,7 @@ template<class vobj,class CComplex>
 
   fine_inner = localInnerProduct(fineX,fineY);
   blockSum(coarse_inner,fine_inner);
+PARALLEL_FOR_LOOP
   for(int ss=0;ss<coarse->oSites();ss++){
     CoarseInner._odata[ss] = coarse_inner._odata[ss];
   }
diff --git a/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc b/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
index 0d54f272..2206bbc8 100644
--- a/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
+++ b/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
@@ -10,12 +10,12 @@ namespace Grid {
     void ContinuedFractionFermion5D::SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata)
     {
       // How to check Ls matches??
-      //      std::cout << Ls << " Ls"<<std::endl;
-      //      std::cout << zdata->n  << " - n"<<std::endl;
-      //      std::cout << zdata->da << " -da "<<std::endl;
-      //      std::cout << zdata->db << " -db"<<std::endl;
-      //      std::cout << zdata->dn << " -dn"<<std::endl;
-      //      std::cout << zdata->dd << " -dd"<<std::endl;
+      //      std::cout<<GridLogMessage << Ls << " Ls"<<std::endl;
+      //      std::cout<<GridLogMessage << zdata->n  << " - n"<<std::endl;
+      //      std::cout<<GridLogMessage << zdata->da << " -da "<<std::endl;
+      //      std::cout<<GridLogMessage << zdata->db << " -db"<<std::endl;
+      //      std::cout<<GridLogMessage << zdata->dn << " -dn"<<std::endl;
+      //      std::cout<<GridLogMessage << zdata->dd << " -dd"<<std::endl;
 
       assert(zdata->db==Ls);// Beta has Ls coeffs
 
@@ -55,7 +55,7 @@ namespace Grid {
 	See[s] = Aee[s] - 1.0/See[s-1];
       }
       for(int s=0;s<Ls;s++){
-	std::cout <<"s = "<<s<<" Beta "<<Beta[s]<<" Aee "<<Aee[s] <<" See "<<See[s] <<std::endl;
+	std::cout<<GridLogMessage <<"s = "<<s<<" Beta "<<Beta[s]<<" Aee "<<Aee[s] <<" See "<<See[s] <<std::endl;
       }
     }
 
diff --git a/lib/qcd/action/fermion/DomainWallFermion.h b/lib/qcd/action/fermion/DomainWallFermion.h
index 34f01be5..498d8b34 100644
--- a/lib/qcd/action/fermion/DomainWallFermion.h
+++ b/lib/qcd/action/fermion/DomainWallFermion.h
@@ -32,7 +32,7 @@ namespace Grid {
 	Approx::zolotarev_data *zdata = Approx::higham(eps,this->Ls);// eps is ignored for higham
 	assert(zdata->n==this->Ls);
 	
-	std::cout << "DomainWallFermion with Ls="<<Ls<<std::endl;
+	std::cout<<GridLogMessage << "DomainWallFermion with Ls="<<Ls<<std::endl;
 	// Call base setter
 	this->CayleyFermion5D::SetCoefficientsTanh(zdata,1.0,0.0);
 
diff --git a/lib/qcd/action/fermion/MobiusFermion.h b/lib/qcd/action/fermion/MobiusFermion.h
index 6603e3f2..5aaaa188 100644
--- a/lib/qcd/action/fermion/MobiusFermion.h
+++ b/lib/qcd/action/fermion/MobiusFermion.h
@@ -30,7 +30,7 @@ namespace Grid {
       {
 	RealD eps = 1.0;
 
-	std::cout << "MobiusFermion (b="<<b<<",c="<<c<<") with Ls= "<<Ls<<" Tanh approx"<<std::endl;
+	std::cout<<GridLogMessage << "MobiusFermion (b="<<b<<",c="<<c<<") with Ls= "<<Ls<<" Tanh approx"<<std::endl;
 	Approx::zolotarev_data *zdata = Approx::higham(eps,this->Ls);// eps is ignored for higham
 	assert(zdata->n==this->Ls);
 	
diff --git a/lib/qcd/action/fermion/MobiusZolotarevFermion.h b/lib/qcd/action/fermion/MobiusZolotarevFermion.h
index d3c0b3a0..5f74702c 100644
--- a/lib/qcd/action/fermion/MobiusZolotarevFermion.h
+++ b/lib/qcd/action/fermion/MobiusZolotarevFermion.h
@@ -34,7 +34,7 @@ namespace Grid {
 	Approx::zolotarev_data *zdata = Approx::zolotarev(eps,this->Ls,0);
 	assert(zdata->n==this->Ls);
 
-	std::cout << "MobiusZolotarevFermion (b="<<b<<",c="<<c<<") with Ls= "<<Ls<<" Zolotarev range ["<<lo<<","<<hi<<"]"<<std::endl;
+	std::cout<<GridLogMessage << "MobiusZolotarevFermion (b="<<b<<",c="<<c<<") with Ls= "<<Ls<<" Zolotarev range ["<<lo<<","<<hi<<"]"<<std::endl;
 	
 	// Call base setter
 	this->CayleyFermion5D::SetCoefficientsZolotarev(hi,zdata,b,c);
diff --git a/lib/qcd/action/fermion/PartialFractionFermion5D.cc b/lib/qcd/action/fermion/PartialFractionFermion5D.cc
index a148fe40..958cea30 100644
--- a/lib/qcd/action/fermion/PartialFractionFermion5D.cc
+++ b/lib/qcd/action/fermion/PartialFractionFermion5D.cc
@@ -260,12 +260,12 @@ namespace Grid {
     void  PartialFractionFermion5D::SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata){
 
       // check on degree matching
-      //      std::cout << Ls << " Ls"<<std::endl;
-      //      std::cout << zdata->n  << " - n"<<std::endl;
-      //      std::cout << zdata->da << " -da "<<std::endl;
-      //      std::cout << zdata->db << " -db"<<std::endl;
-      //      std::cout << zdata->dn << " -dn"<<std::endl;
-      //      std::cout << zdata->dd << " -dd"<<std::endl;
+      //      std::cout<<GridLogMessage << Ls << " Ls"<<std::endl;
+      //      std::cout<<GridLogMessage << zdata->n  << " - n"<<std::endl;
+      //      std::cout<<GridLogMessage << zdata->da << " -da "<<std::endl;
+      //      std::cout<<GridLogMessage << zdata->db << " -db"<<std::endl;
+      //      std::cout<<GridLogMessage << zdata->dn << " -dn"<<std::endl;
+      //      std::cout<<GridLogMessage << zdata->dd << " -dd"<<std::endl;
       assert(Ls == (2*zdata->da -1) );
 
       // Part frac
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.cc b/lib/qcd/action/fermion/WilsonFermion5D.cc
index 316b0676..43829ea7 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.cc
+++ b/lib/qcd/action/fermion/WilsonFermion5D.cc
@@ -100,7 +100,7 @@ void WilsonFermion5D::DhopDir(const LatticeFermion &in, LatticeFermion &out,int
   assert(dirdisp<=7);
   assert(dirdisp>=0);
 
-//PARALLEL_FOR_LOOP
+PARALLEL_FOR_LOOP
   for(int ss=0;ss<Umu._grid->oSites();ss++){
     for(int s=0;s<Ls;s++){
       int sU=ss;
diff --git a/lib/qcd/action/fermion/WilsonKernelsHand.cc b/lib/qcd/action/fermion/WilsonKernelsHand.cc
index 8c31d65b..cff700fb 100644
--- a/lib/qcd/action/fermion/WilsonKernelsHand.cc
+++ b/lib/qcd/action/fermion/WilsonKernelsHand.cc
@@ -360,11 +360,11 @@ void DiracOptHandDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     MULT_2SPIN(Xp);
   }
   XP_RECON;
-  //  std::cout << "XP_RECON"<<std::endl;
-  //  std::cout << result_00 <<" "<<result_01 <<" "<<result_02 <<std::endl;
-  //  std::cout << result_10 <<" "<<result_11 <<" "<<result_12 <<std::endl;
-  //  std::cout << result_20 <<" "<<result_21 <<" "<<result_22 <<std::endl;
-  //  std::cout << result_30 <<" "<<result_31 <<" "<<result_32 <<std::endl;
+  //  std::cout<<GridLogMessage << "XP_RECON"<<std::endl;
+  //  std::cout<<GridLogMessage << result_00 <<" "<<result_01 <<" "<<result_02 <<std::endl;
+  //  std::cout<<GridLogMessage << result_10 <<" "<<result_11 <<" "<<result_12 <<std::endl;
+  //  std::cout<<GridLogMessage << result_20 <<" "<<result_21 <<" "<<result_22 <<std::endl;
+  //  std::cout<<GridLogMessage << result_30 <<" "<<result_31 <<" "<<result_32 <<std::endl;
 
   // Yp
   offset = st._offsets [Yp][ss];
@@ -446,11 +446,11 @@ void DiracOptHandDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     MULT_2SPIN(Xm);
   }
   XM_RECON_ACCUM;
-  //  std::cout << "XM_RECON_ACCUM"<<std::endl;
-  //  std::cout << result_00 <<" "<<result_01 <<" "<<result_02 <<std::endl;
-  //  std::cout << result_10 <<" "<<result_11 <<" "<<result_12 <<std::endl;
-  //  std::cout << result_20 <<" "<<result_21 <<" "<<result_22 <<std::endl;
-  //  std::cout << result_30 <<" "<<result_31 <<" "<<result_32 <<std::endl;
+  //  std::cout<<GridLogMessage << "XM_RECON_ACCUM"<<std::endl;
+  //  std::cout<<GridLogMessage << result_00 <<" "<<result_01 <<" "<<result_02 <<std::endl;
+  //  std::cout<<GridLogMessage << result_10 <<" "<<result_11 <<" "<<result_12 <<std::endl;
+  //  std::cout<<GridLogMessage << result_20 <<" "<<result_21 <<" "<<result_22 <<std::endl;
+  //  std::cout<<GridLogMessage << result_30 <<" "<<result_31 <<" "<<result_32 <<std::endl;
   
   
   // Ym
diff --git a/lib/qcd/action/gauge/WilsonGaugeAction.h b/lib/qcd/action/gauge/WilsonGaugeAction.h
index aaedf2f9..fc31a5c3 100644
--- a/lib/qcd/action/gauge/WilsonGaugeAction.h
+++ b/lib/qcd/action/gauge/WilsonGaugeAction.h
@@ -18,7 +18,7 @@ namespace Grid{
       
       virtual RealD S(const GaugeField &U) {
 	RealD plaq = WilsonLoops<MatrixField,GaugeField>::avgPlaquette(U);
-	std::cout << "Plaq : "<<plaq << "\n";
+	std::cout<<GridLogMessage << "Plaq : "<<plaq << "\n";
 	double vol = U._grid->gSites();
 	return beta*(1.0 -plaq)*(Nd*(Nd-1.0))*vol*0.5;
       };
diff --git a/lib/qcd/hmc/HMC.h b/lib/qcd/hmc/HMC.h
index c0f06b64..e3fe57ed 100644
--- a/lib/qcd/hmc/HMC.h
+++ b/lib/qcd/hmc/HMC.h
@@ -36,15 +36,15 @@ namespace Grid{
 	RealD prob = std::exp(-DeltaH);
 	random(sRNG,rn_test);
       
-	std::cout<< "--------------------------------------------\n";
-	std::cout<< "dH = "<<DeltaH << "  Random = "<< rn_test 
+	std::cout<<GridLogMessage<< "--------------------------------------------\n";
+	std::cout<<GridLogMessage<< "dH = "<<DeltaH << "  Random = "<< rn_test 
 		 << "\nAcc. Probability = " << ((prob<1.0)? prob: 1.0)<< "   ";
       
 	if((prob >1.0) || (rn_test <= prob)){       // accepted
-	  std::cout <<"-- ACCEPTED\n";
+	  std::cout<<GridLogMessage <<"-- ACCEPTED\n";
 	  return true;
 	} else {                               // rejected
-	  std::cout <<"-- REJECTED\n";
+	  std::cout<<GridLogMessage <<"-- REJECTED\n";
 	  return false;
 	}
       }
@@ -53,12 +53,12 @@ namespace Grid{
 
 	MD.init(U); // set U and initialize P and phi's 
 	RealD H0 = MD.S(U); // initial state action  
-	std::cout<<"Total H before = "<< H0 << "\n";
+	std::cout<<GridLogMessage<<"Total H before = "<< H0 << "\n";
       
 	MD.integrate(U);
       
 	RealD H1 = MD.S(U); // updated state action            
-	std::cout<<"Total H after = "<< H1 << "\n";
+	std::cout<<GridLogMessage<<"Total H after = "<< H1 << "\n";
       
 	return (H1-H0);
       }
@@ -81,17 +81,17 @@ namespace Grid{
 	
 	// Thermalizations
 	for(int iter=1; iter <= Params.ThermalizationSteps; ++iter){
-	  std::cout << "-- # Thermalization step = "<< iter <<  "\n";
+	  std::cout<<GridLogMessage << "-- # Thermalization step = "<< iter <<  "\n";
 	
 	  DeltaH = evolve_step(Uin);
-	  std::cout<< " dH = "<< DeltaH << "\n";
+	  std::cout<<GridLogMessage<< " dH = "<< DeltaH << "\n";
 	}
 
 	// Actual updates (evolve a copy Ucopy then copy back eventually)
 	LatticeLorentzColourMatrix Ucopy(Uin._grid);
 	for(int iter=Params.StartingConfig; 
 	    iter < Params.Nsweeps+Params.StartingConfig; ++iter){
-	  std::cout << "-- # Sweep = "<< iter <<  "\n";
+	  std::cout<<GridLogMessage << "-- # Sweep = "<< iter <<  "\n";
 	  
 	  Ucopy = Uin;
 	  DeltaH = evolve_step(Ucopy);
diff --git a/lib/qcd/hmc/integrators/Integrator.h b/lib/qcd/hmc/integrators/Integrator.h
index 90707d51..c0ed74ff 100644
--- a/lib/qcd/hmc/integrators/Integrator.h
+++ b/lib/qcd/hmc/integrators/Integrator.h
@@ -95,7 +95,7 @@ namespace Grid{
 
       //Initialization of momenta and actions
       void init(LatticeLorentzColourMatrix& U){
-	std::cout<< "Integrator init\n";
+	std::cout<<GridLogMessage<< "Integrator init\n";
 
 	MDutils::generate_momenta(*P,pRNG);
 	for(int level=0; level< as.size(); ++level){
@@ -119,7 +119,7 @@ namespace Grid{
 	
 	RealD H = Hsum.real();
 
-	std::cout << "H_p = "<< H << "\n";
+	std::cout<<GridLogMessage << "H_p = "<< H << "\n";
 
 	// Actions
 	for(int level=0; level<as.size(); ++level)
diff --git a/lib/qcd/hmc/integrators/Integrator_algorithm.h b/lib/qcd/hmc/integrators/Integrator_algorithm.h
index 0dce2978..0369021f 100644
--- a/lib/qcd/hmc/integrators/Integrator_algorithm.h
+++ b/lib/qcd/hmc/integrators/Integrator_algorithm.h
@@ -39,29 +39,29 @@ namespace Grid{
 	  if(clock[level] == 0){    // initial half step 
 	    Integ->update_P(U,level,lambda*eps);
 	    ++clock[level];
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"P "<< clock[level] <<std::endl;
+	    for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
+	    std::cout<<GridLogMessage<<"P "<< clock[level] <<std::endl;
 	  }
 	  
 	  if(level == fl){          // lowest level 
 	    Integ->update_U(U,0.5*eps);
 	    
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"U "<< (clock[level]+1) <<std::endl;
+	    for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
+	    std::cout<<GridLogMessage<<"U "<< (clock[level]+1) <<std::endl;
 	  }else{                 // recursive function call 
 	    step(U,level+1,clock, Integ);
 	  }
 	  
 	  Integ->update_P(U,level,(1.0-2.0*lambda)*eps);
 	  ++clock[level];
-	  for(int l=0; l<level;++l) std::cout<<"   ";
-	  std::cout<<"P "<< (clock[level]) <<std::endl;
+	  for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
+	  std::cout<<GridLogMessage<<"P "<< (clock[level]) <<std::endl;
 	  
 	  if(level == fl){          // lowest level 
 	    Integ->update_U(U,0.5*eps);
 	    
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"U "<< (clock[level]+1) <<std::endl;
+	    for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
+	    std::cout<<GridLogMessage<<"U "<< (clock[level]+1) <<std::endl;
 	  }else{                 // recursive function call 
 	    step(U,level+1,clock, Integ);
 	  }    
@@ -71,14 +71,14 @@ namespace Grid{
 	    Integ->update_P(U,level,lambda*eps);
 	    
 	    ++clock[level];
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"P "<< clock[level] <<std::endl;
+	    for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
+	    std::cout<<GridLogMessage<<"P "<< clock[level] <<std::endl;
 	  }else{                  // bulk step
 	    Integ->update_P(U,level,lambda*2.0*eps);
 	    
 	    clock[level]+=2;
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"P "<< clock[level] <<std::endl;
+	    for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
+	    std::cout<<GridLogMessage<<"P "<< clock[level] <<std::endl;
 	  }
 	}
 	
@@ -112,13 +112,13 @@ namespace Grid{
 	  if(clock[level] == 0){    // initial half step
 	    Integ->update_P(U, level,eps/2.0);
 	    ++clock[level];
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
+	    for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
+	    std::cout<<GridLogMessage<<"P "<< 0.5*clock[level] <<std::endl;
 	  }
 	  if(level == fl){          // lowest level
 	    Integ->update_U(U, eps);
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"U "<< 0.5*(clock[level]+1) <<std::endl;
+	    for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
+	    std::cout<<GridLogMessage<<"U "<< 0.5*(clock[level]+1) <<std::endl;
 	  }else{                 // recursive function call
 	    step(U, level+1,clock, Integ);
 	  }
@@ -126,14 +126,14 @@ namespace Grid{
 	    Integ->update_P(U, level,eps/2.0);
 	    
 	    ++clock[level];
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
+	    for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
+	    std::cout<<GridLogMessage<<"P "<< 0.5*clock[level] <<std::endl;
 	  }else{                  // bulk step
 	    Integ->update_P(U, level,eps);
 	    
 	    clock[level]+=2;
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
+	    for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
+	    std::cout<<GridLogMessage<<"P "<< 0.5*clock[level] <<std::endl;
 	  }
 	}
 
diff --git a/lib/qcd/hmc/integrators/Integrator_base.h b/lib/qcd/hmc/integrators/Integrator_base.h
index 38798d67..ef65f694 100644
--- a/lib/qcd/hmc/integrators/Integrator_base.h
+++ b/lib/qcd/hmc/integrators/Integrator_base.h
@@ -101,7 +101,7 @@ namespace Grid{
       //Initialization of momenta and actions
       void init(LatticeLorentzColourMatrix& U,
 		GridParallelRNG& pRNG){
-	std::cout<< "Integrator init\n";
+	std::cout<<GridLogMessage<< "Integrator init\n";
 	if (!P)
 	  P = new LatticeLorentzColourMatrix(U._grid);
 	MDutils::generate_momenta(*P,pRNG);
@@ -172,13 +172,13 @@ namespace Grid{
 	  if(clock[level] == 0){    // initial half step
 	    Integ->update_P(U, level,eps/2);
 	    ++clock[level];
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
+	    for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
+	    std::cout<<GridLogMessage<<"P "<< 0.5*clock[level] <<std::endl;
 	  }
 	  if(level == fl){          // lowest level
 	    Integ->update_U(U, eps);
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"U "<< 0.5*(clock[level]+1) <<std::endl;
+	    for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
+	    std::cout<<GridLogMessage<<"U "<< 0.5*(clock[level]+1) <<std::endl;
 	  }else{                 // recursive function call
 	    step(U, level+1,clock, Integ);
 	  }
@@ -186,14 +186,14 @@ namespace Grid{
 	    Integ->update_P(U, level,eps/2);
 	    
 	    ++clock[level];
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
+	    for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
+	    std::cout<<GridLogMessage<<"P "<< 0.5*clock[level] <<std::endl;
 	  }else{                  // bulk step
 	    Integ->update_P(U, level,eps);
 	    
 	    clock[level]+=2;
-	    for(int l=0; l<level;++l) std::cout<<"   ";
-	    std::cout<<"P "<< 0.5*clock[level] <<std::endl;
+	    for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
+	    std::cout<<GridLogMessage<<"P "<< 0.5*clock[level] <<std::endl;
 	  }
 	}
 
diff --git a/lib/qcd/utils/SUn.h b/lib/qcd/utils/SUn.h
index 10313f5c..4ae69a0e 100644
--- a/lib/qcd/utils/SUn.h
+++ b/lib/qcd/utils/SUn.h
@@ -372,7 +372,7 @@ Note that in step D setting B ~ X - A and using B in place of A in step E will g
     LatticeReal d(grid); d=zero;
     LatticeReal alpha(grid);
 
-    //    std::cout<<"xi "<<xi <<std::endl;
+    //    std::cout<<GridLogMessage<<"xi "<<xi <<std::endl;
     alpha = toReal(2.0*xi);
 
     do { 
@@ -468,11 +468,11 @@ Note that in step D setting B ~ X - A and using B in place of A in step E will g
     LatticeMatrix Vcheck(grid);
     Vcheck = zero;
     Vcheck = where(Accepted,V*adj(V) - 1.0,Vcheck);
-    //    std::cout << "SU3 check " <<norm2(Vcheck)<<std::endl;
+    //    std::cout<<GridLogMessage << "SU3 check " <<norm2(Vcheck)<<std::endl;
     assert(norm2(Vcheck)<1.0e-4);
     
     // Verify the link stays in SU(3)
-    //    std::cout <<"Checking the modified link"<<std::endl;
+    //    std::cout<<GridLogMessage <<"Checking the modified link"<<std::endl;
     Vcheck = link*adj(link) - 1.0;
     assert(norm2(Vcheck)<1.0e-4);
     /////////////////////////////////
@@ -483,42 +483,42 @@ Note that in step D setting B ~ X - A and using B in place of A in step E will g
     for(int gen=0;gen<generators();gen++){
       Matrix ta;
       generator(gen,ta);
-      std::cout<< "Nc = "<<ncolour<<" t_"<<gen<<std::endl;
-      std::cout<<ta<<std::endl;
+      std::cout<<GridLogMessage<< "Nc = "<<ncolour<<" t_"<<gen<<std::endl;
+      std::cout<<GridLogMessage<<ta<<std::endl;
     }
   }
 
   static void testGenerators(void){
     Matrix ta;
     Matrix tb;
-    std::cout<<"Checking trace ta tb is 0.5 delta_ab"<<std::endl;
+    std::cout<<GridLogMessage<<"Checking trace ta tb is 0.5 delta_ab"<<std::endl;
     for(int a=0;a<generators();a++){
       for(int b=0;b<generators();b++){
 	generator(a,ta);
 	generator(b,tb);
 	Complex tr =TensorRemove(trace(ta*tb)); 
-	std::cout<<tr<<" ";
+	std::cout<<GridLogMessage<<tr<<" ";
 	if(a==b) assert(abs(tr-Complex(0.5))<1.0e-6);
 	if(a!=b) assert(abs(tr)<1.0e-6);
       }
-      std::cout<<std::endl;
+      std::cout<<GridLogMessage<<std::endl;
     }
-    std::cout<<"Checking hermitian"<<std::endl;
+    std::cout<<GridLogMessage<<"Checking hermitian"<<std::endl;
     for(int a=0;a<generators();a++){
       generator(a,ta);
-      std::cout<<a<<" ";
+      std::cout<<GridLogMessage<<a<<" ";
       assert(norm2(ta-adj(ta))<1.0e-6);
     }    
-    std::cout<<std::endl;
+    std::cout<<GridLogMessage<<std::endl;
 
-    std::cout<<"Checking traceless"<<std::endl;
+    std::cout<<GridLogMessage<<"Checking traceless"<<std::endl;
     for(int a=0;a<generators();a++){
       generator(a,ta);
       Complex tr =TensorRemove(trace(ta)); 
-      std::cout<<a<<" ";
+      std::cout<<GridLogMessage<<a<<" ";
       assert(abs(tr)<1.0e-6);
     }    
-    std::cout<<std::endl;
+    std::cout<<GridLogMessage<<std::endl;
   }
 
   // reunitarise??
diff --git a/lib/tensors/Tensor_extract_merge.h b/lib/tensors/Tensor_extract_merge.h
index 21ffd778..6209209f 100644
--- a/lib/tensors/Tensor_extract_merge.h
+++ b/lib/tensors/Tensor_extract_merge.h
@@ -57,11 +57,11 @@ inline void extract(typename std::enable_if<!isGridTensor<vsimd>::value, const v
     extracted[i]=buf[i*s];
     for(int ii=1;ii<s;ii++){
       if ( buf[i*s]!=buf[i*s+ii] ){
-	std::cout << " SIMD extract failure splat = "<<s<<" ii "<<ii<<" " <<Nextr<<" "<< Nsimd<<" "<<std::endl;
+	std::cout<<GridLogMessage << " SIMD extract failure splat = "<<s<<" ii "<<ii<<" " <<Nextr<<" "<< Nsimd<<" "<<std::endl;
 	for(int vv=0;vv<Nsimd;vv++) {
-	  std::cout<< buf[vv]<<" ";
+	  std::cout<<GridLogMessage<< buf[vv]<<" ";
 	}
-	std::cout<<std::endl;
+	std::cout<<GridLogMessage<<std::endl;
 	assert(0);
       }
       assert(buf[i*s]==buf[i*s+ii]);
diff --git a/tests/Make.inc b/tests/Make.inc
index 31c39cd4..f04da4eb 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,9 @@
 
-bin_PROGRAMS = Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_GaugeAction Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
+bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
+
+
+Test_GaugeAction_SOURCES=Test_GaugeAction.cc
+Test_GaugeAction_LDADD=-lGrid
 
 
 Test_cayley_cg_SOURCES=Test_cayley_cg.cc
@@ -74,10 +78,6 @@ Test_gamma_SOURCES=Test_gamma.cc
 Test_gamma_LDADD=-lGrid
 
 
-Test_GaugeAction_SOURCES=Test_GaugeAction.cc
-Test_GaugeAction_LDADD=-lGrid
-
-
 Test_hmc_WilsonGauge_SOURCES=Test_hmc_WilsonGauge.cc
 Test_hmc_WilsonGauge_LDADD=-lGrid
 
diff --git a/tests/Test_GaugeAction.cc b/tests/Test_GaugeAction.cc
index 612eb693..d779cd20 100644
--- a/tests/Test_GaugeAction.cc
+++ b/tests/Test_GaugeAction.cc
@@ -94,7 +94,7 @@ int main (int argc, char ** argv)
   double vol = Fine.gSites();
   Complex PlaqScale(1.0/vol/6.0/3.0);
   RealD   StapScale(1.0/vol/6.0/3.0);
-  std::cout <<"PlaqScale" << PlaqScale<<std::endl;
+  std::cout<<GridLogMessage <<"PlaqScale" << PlaqScale<<std::endl;
   std::vector<TComplex> Plaq_T(orthosz);
   sliceSum(Plaq,Plaq_T,Nd-1);
   int Nt = Plaq_T.size();
@@ -105,20 +105,20 @@ int main (int argc, char ** argv)
   for(int t=0;t<Nt;t++){
     Plaq_T_sum = Plaq_T_sum+Plaq_T[t];
     Complex Pt=TensorRemove(Plaq_T[t]);
-    std::cout << "sliced ["<<t<<"]" <<Pt*PlaqScale*Real(Nt) << std::endl;
+    std::cout<<GridLogMessage << "sliced ["<<t<<"]" <<Pt*PlaqScale*Real(Nt) << std::endl;
   }
 
   {
     Complex Pt = TensorRemove(Plaq_T_sum);
-    std::cout << "total " <<Pt*PlaqScale<<std::endl;
+    std::cout<<GridLogMessage << "total " <<Pt*PlaqScale<<std::endl;
   }  
 
   TComplex Tp = sum(Plaq);
   Complex p  = TensorRemove(Tp);
-  std::cout << "calculated plaquettes " <<p*PlaqScale<<std::endl;
+  std::cout<<GridLogMessage << "calculated plaquettes " <<p*PlaqScale<<std::endl;
 
   RealD avg_plaq = ColourWilsonLoops::avgPlaquette(Umu);
-  std::cout << "NEW : calculated real plaquettes " <<avg_plaq<<std::endl;
+  std::cout<<GridLogMessage << "NEW : calculated real plaquettes " <<avg_plaq<<std::endl;
 
   RealD stap_plaq=0.0;
   LatticeColourMatrix stap(&Fine);
@@ -130,16 +130,16 @@ int main (int argc, char ** argv)
     Complex s  = TensorRemove(Ts);
     stap_plaq+=real(s);
   }
-  std::cout << "NEW : plaquette via staples"<< stap_plaq*StapScale*0.25<< std::endl;
+  std::cout<<GridLogMessage << "NEW : plaquette via staples"<< stap_plaq*StapScale*0.25<< std::endl;
   Complex LinkTraceScale(1.0/vol/4.0/3.0);
   TComplex Tl = sum(LinkTrace);
   Complex l  = TensorRemove(Tl);
-  std::cout << "calculated link trace " <<l*LinkTraceScale<<std::endl;
+  std::cout<<GridLogMessage << "calculated link trace " <<l*LinkTraceScale<<std::endl;
 
   blockSum(cPlaq,Plaq);
   TComplex TcP = sum(cPlaq);
   Complex ll= TensorRemove(TcP);
-  std::cout << "coarsened plaquettes sum to " <<ll*PlaqScale<<std::endl;
+  std::cout<<GridLogMessage << "coarsened plaquettes sum to " <<ll*PlaqScale<<std::endl;
 
 
   Grid_finalize();
diff --git a/tests/Test_cayley_cg.cc b/tests/Test_cayley_cg.cc
index 23004be9..f8d1187c 100644
--- a/tests/Test_cayley_cg.cc
+++ b/tests/Test_cayley_cg.cc
@@ -52,7 +52,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   int threads = GridThread::GetThreads();
-  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+  std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
 
   const int Ls=8;
   GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
@@ -71,33 +71,33 @@ int main (int argc, char ** argv)
 
   RealD mass=0.1;
   RealD M5  =1.8;
-  std::cout <<"DomainWallFermion test"<<std::endl;
+  std::cout<<GridLogMessage <<"DomainWallFermion test"<<std::endl;
   DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
   TestCGinversions<DomainWallFermion>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   RealD b=1.5;// Scale factor b+c=2, b-c=1
   RealD c=0.5;
-  std::cout <<"MobiusFermion test"<<std::endl;
+  std::cout<<GridLogMessage <<"MobiusFermion test"<<std::endl;
   MobiusFermion Dmob(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c);
   TestCGinversions<MobiusFermion>(Dmob,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
-  std::cout <<"MobiusZolotarevFermion test"<<std::endl;
+  std::cout<<GridLogMessage <<"MobiusZolotarevFermion test"<<std::endl;
   MobiusZolotarevFermion Dzolo(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c,0.1,2.0);
   TestCGinversions<MobiusZolotarevFermion>(Dzolo,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
-  std::cout <<"ScaledShamirFermion test"<<std::endl;
+  std::cout<<GridLogMessage <<"ScaledShamirFermion test"<<std::endl;
   ScaledShamirFermion Dsham(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,2.0);
   TestCGinversions<ScaledShamirFermion>(Dsham,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
-  std::cout <<"ShamirZolotarevFermion test"<<std::endl;
+  std::cout<<GridLogMessage <<"ShamirZolotarevFermion test"<<std::endl;
   ShamirZolotarevFermion Dshamz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,2.0);
   TestCGinversions<ShamirZolotarevFermion>(Dshamz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
-  std::cout <<"OverlapWilsonCayleyTanhFermion test"<<std::endl;
+  std::cout<<GridLogMessage <<"OverlapWilsonCayleyTanhFermion test"<<std::endl;
   OverlapWilsonCayleyTanhFermion Dov(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
   TestCGinversions<OverlapWilsonCayleyTanhFermion>(Dov,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
-  std::cout <<"OverlapWilsonCayleyZolotarevFermion test"<<std::endl;
+  std::cout<<GridLogMessage <<"OverlapWilsonCayleyZolotarevFermion test"<<std::endl;
   OverlapWilsonCayleyZolotarevFermion Dovz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,2.0);
   TestCGinversions<OverlapWilsonCayleyZolotarevFermion>(Dovz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
@@ -111,11 +111,11 @@ void  TestCGinversions(What & Ddwf,
 		       GridParallelRNG *RNG4,
 		       GridParallelRNG *RNG5)
 {
-  std::cout << "Testing unpreconditioned inverter"<<std::endl;
+  std::cout<<GridLogMessage << "Testing unpreconditioned inverter"<<std::endl;
   TestCGunprec<What>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,RNG4,RNG5);
-  std::cout << "Testing red black preconditioned inverter"<<std::endl;
+  std::cout<<GridLogMessage << "Testing red black preconditioned inverter"<<std::endl;
   TestCGprec<What>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,RNG4,RNG5);
-  std::cout << "Testing red black Schur inverter"<<std::endl;
+  std::cout<<GridLogMessage << "Testing red black Schur inverter"<<std::endl;
   TestCGschur<What>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,RNG4,RNG5);
 }
 
diff --git a/tests/Test_cayley_coarsen_support.cc b/tests/Test_cayley_coarsen_support.cc
index 473fa10a..3ddf6205 100644
--- a/tests/Test_cayley_coarsen_support.cc
+++ b/tests/Test_cayley_coarsen_support.cc
@@ -57,8 +57,8 @@ int main (int argc, char ** argv)
   Complex cone(1.0,0.0);
   for(int nn=0;nn<Nd;nn++){
     if(1) {
-      if (nn>2) { U[nn]=zero; std::cout << "zeroing gauge field in dir "<<nn<<std::endl; }
-      else      { U[nn]=cone; std::cout << "unit gauge field in dir "<<nn<<std::endl; }
+      if (nn>2) { U[nn]=zero; std::cout<<GridLogMessage << "zeroing gauge field in dir "<<nn<<std::endl; }
+      else      { U[nn]=cone; std::cout<<GridLogMessage << "unit gauge field in dir "<<nn<<std::endl; }
     }
     pokeIndex<LorentzIndex>(Umu,U[nn],nn);
   }
@@ -75,19 +75,19 @@ int main (int argc, char ** argv)
   
   for(int d=0;d<4;d++){
     HermIndefOp.OpDir(src,tmp,d+1,+1); result=result+tmp; 
-    std::cout<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
+    std::cout<<GridLogMessage<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
     HermIndefOp.OpDir(src,tmp,d+1,-1); result=result+tmp;
-    std::cout<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
+    std::cout<<GridLogMessage<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
   }
   err = result-ref;
-  std::cout<<"Error "<<norm2(err)<<std::endl;
+  std::cout<<GridLogMessage<<"Error "<<norm2(err)<<std::endl;
 
   const int nbasis = 2;
   LatticeFermion prom(FGrid);
 
   std::vector<LatticeFermion> subspace(nbasis,FGrid);
 
-  std::cout<<"Calling Aggregation class" <<std::endl;
+  std::cout<<GridLogMessage<<"Calling Aggregation class" <<std::endl;
 
   MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
   typedef Aggregation<vSpinColourVector,vTComplex,nbasis> Subspace;
@@ -96,7 +96,7 @@ int main (int argc, char ** argv)
 
   subspace=Aggregates.subspace;
 
-  std::cout << "Called aggregation class"<< std::endl;
+  std::cout<<GridLogMessage << "Called aggregation class"<< std::endl;
 
   typedef CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis> LittleDiracOperator;
   typedef LittleDiracOperator::CoarseVector CoarseVector;
@@ -119,7 +119,7 @@ int main (int argc, char ** argv)
     prom=prom+subspace[b];
   }
   err=err-prom; 
-  std::cout<<"Promoted back from subspace err "<<norm2(err)<<std::endl;
+  std::cout<<GridLogMessage<<"Promoted back from subspace err "<<norm2(err)<<std::endl;
 
   HermIndefOp.HermOp(prom,tmp);
   blockProject(c_proj,tmp,subspace);
@@ -127,15 +127,15 @@ int main (int argc, char ** argv)
   LittleDiracOp.M(c_src,c_res);
 
   c_proj = c_proj - c_res;
-  std::cout<<"Representation of ldop within subspace "<<norm2(c_proj)<<std::endl;
+  std::cout<<GridLogMessage<<"Representation of ldop within subspace "<<norm2(c_proj)<<std::endl;
 
-  std::cout << "Multiplying by LittleDiracOp "<< std::endl;
+  std::cout<<GridLogMessage << "Multiplying by LittleDiracOp "<< std::endl;
   LittleDiracOp.M(c_src,c_res);
 
-  std::cout<<"Testing hermiticity explicitly by inspecting matrix elements"<<std::endl;
+  std::cout<<GridLogMessage<<"Testing hermiticity explicitly by inspecting matrix elements"<<std::endl;
   LittleDiracOp.AssertHermitian();
 
-  std::cout << "Testing Hermiticity stochastically "<< std::endl;
+  std::cout<<GridLogMessage << "Testing Hermiticity stochastically "<< std::endl;
   CoarseVector phi(Coarse5d);
   CoarseVector chi(Coarse5d);
   CoarseVector Aphi(Coarse5d);
@@ -145,7 +145,7 @@ int main (int argc, char ** argv)
   random(CRNG,chi);
 
 
-  std::cout<<"Made randoms"<<std::endl;
+  std::cout<<GridLogMessage<<"Made randoms"<<std::endl;
 
   LittleDiracOp.M(phi,Aphi);
   LittleDiracOp.Mdag(chi,Achi);
@@ -155,11 +155,11 @@ int main (int argc, char ** argv)
   ComplexD cAc = innerProduct(chi,Achi);
   ComplexD pAp = innerProduct(phi,Aphi);
 
-  std::cout<< "pAc "<<pAc<<" cAp "<< cAp<< " diff "<<pAc-adj(cAp)<<std::endl;
+  std::cout<<GridLogMessage<< "pAc "<<pAc<<" cAp "<< cAp<< " diff "<<pAc-adj(cAp)<<std::endl;
 
-  std::cout<< "pAp "<<pAp<<" cAc "<< cAc<<"Should be real"<< std::endl;
+  std::cout<<GridLogMessage<< "pAp "<<pAp<<" cAc "<< cAc<<"Should be real"<< std::endl;
 
-  std::cout<<"Testing linearity"<<std::endl;
+  std::cout<<GridLogMessage<<"Testing linearity"<<std::endl;
   CoarseVector PhiPlusChi(Coarse5d);
   CoarseVector APhiPlusChi(Coarse5d);
   CoarseVector linerr(Coarse5d);
@@ -168,9 +168,9 @@ int main (int argc, char ** argv)
 
   linerr= APhiPlusChi-Aphi;
   linerr= linerr-Achi;
-  std::cout<<"**Diff "<<norm2(linerr)<<std::endl;
+  std::cout<<GridLogMessage<<"**Diff "<<norm2(linerr)<<std::endl;
 
 
-  std::cout << "Done "<< std::endl;
+  std::cout<<GridLogMessage << "Done "<< std::endl;
   Grid_finalize();
 }
diff --git a/tests/Test_cayley_even_odd.cc b/tests/Test_cayley_even_odd.cc
index 21c828db..d0c7b74a 100644
--- a/tests/Test_cayley_even_odd.cc
+++ b/tests/Test_cayley_even_odd.cc
@@ -29,7 +29,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   int threads = GridThread::GetThreads();
-  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+  std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
 
   const int Ls=8;
   GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
@@ -48,34 +48,34 @@ int main (int argc, char ** argv)
 
   RealD mass=0.1;
   RealD M5  =1.8;
-  std::cout <<"DomainWallFermion test"<<std::endl;
+  std::cout<<GridLogMessage <<"DomainWallFermion test"<<std::endl;
   DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
   TestWhat<DomainWallFermion>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   RealD b=1.5;// Scale factor b+c=2, b-c=1
   RealD c=0.5;
-  std::cout <<"MobiusFermion test"<<std::endl;
+  std::cout<<GridLogMessage <<"MobiusFermion test"<<std::endl;
   MobiusFermion Dmob(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c);
   TestWhat<MobiusFermion>(Dmob,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
-  std::cout <<"MobiusZolotarevFermion test"<<std::endl;
+  std::cout<<GridLogMessage <<"MobiusZolotarevFermion test"<<std::endl;
   MobiusZolotarevFermion Dzolo(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c,0.1,2.0);
   TestWhat<MobiusZolotarevFermion>(Dzolo,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
-  std::cout <<"ScaledShamirFermion test"<<std::endl;
+  std::cout<<GridLogMessage <<"ScaledShamirFermion test"<<std::endl;
   ScaledShamirFermion Dsham(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,2.0);
   TestWhat<ScaledShamirFermion>(Dsham,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
 
-  std::cout <<"ShamirZolotarevFermion test"<<std::endl;
+  std::cout<<GridLogMessage <<"ShamirZolotarevFermion test"<<std::endl;
   ShamirZolotarevFermion Dshamz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,2.0);
   TestWhat<ShamirZolotarevFermion>(Dshamz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
-  std::cout <<"OverlapWilsonCayleyTanhFermion test"<<std::endl;
+  std::cout<<GridLogMessage <<"OverlapWilsonCayleyTanhFermion test"<<std::endl;
   OverlapWilsonCayleyTanhFermion Dov(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
   TestWhat<OverlapWilsonCayleyTanhFermion>(Dov,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
-  std::cout <<"OverlapWilsonCayleyZolotarevFermion test"<<std::endl;
+  std::cout<<GridLogMessage <<"OverlapWilsonCayleyZolotarevFermion test"<<std::endl;
   OverlapWilsonCayleyZolotarevFermion Dovz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,2.0);
   TestWhat<OverlapWilsonCayleyZolotarevFermion>(Dovz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
@@ -106,40 +106,40 @@ void  TestWhat(What & Ddwf,
   LatticeFermion r_eo  (FGrid);
   LatticeFermion r_eeoo(FGrid);
 
-  std::cout<<"=========================================================="<<std::endl;
-  std::cout<<"= Testing that Meo + Moe + Moo + Mee = Munprec "<<std::endl;
-  std::cout<<"=========================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=========================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Testing that Meo + Moe + Moo + Mee = Munprec "<<std::endl;
+  std::cout<<GridLogMessage<<"=========================================================="<<std::endl;
 
   pickCheckerboard(Even,src_e,src);
   pickCheckerboard(Odd,src_o,src);
 
-  Ddwf.Meooe(src_e,r_o);  std::cout<<"Applied Meo"<<std::endl;
-  Ddwf.Meooe(src_o,r_e);  std::cout<<"Applied Moe"<<std::endl;
+  Ddwf.Meooe(src_e,r_o);  std::cout<<GridLogMessage<<"Applied Meo"<<std::endl;
+  Ddwf.Meooe(src_o,r_e);  std::cout<<GridLogMessage<<"Applied Moe"<<std::endl;
   setCheckerboard(r_eo,r_o);
   setCheckerboard(r_eo,r_e);
 
-  Ddwf.Mooee(src_e,r_e);  std::cout<<"Applied Mee"<<std::endl;
-  Ddwf.Mooee(src_o,r_o);  std::cout<<"Applied Moo"<<std::endl;
+  Ddwf.Mooee(src_e,r_e);  std::cout<<GridLogMessage<<"Applied Mee"<<std::endl;
+  Ddwf.Mooee(src_o,r_o);  std::cout<<GridLogMessage<<"Applied Moo"<<std::endl;
   setCheckerboard(r_eeoo,r_e);
   setCheckerboard(r_eeoo,r_o);
 
   r_eo=r_eo+r_eeoo;
   Ddwf.M(src,ref);  
 
-  //  std::cout << r_eo<<std::endl;
-  //  std::cout << ref <<std::endl;
+  //  std::cout<<GridLogMessage << r_eo<<std::endl;
+  //  std::cout<<GridLogMessage << ref <<std::endl;
 
   err= ref - r_eo;
-  std::cout << "EO norm diff   "<< norm2(err)<< " "<<norm2(ref)<< " " << norm2(r_eo) <<std::endl;
+  std::cout<<GridLogMessage << "EO norm diff   "<< norm2(err)<< " "<<norm2(ref)<< " " << norm2(r_eo) <<std::endl;
     
   LatticeComplex cerr(FGrid);
   cerr = localInnerProduct(err,err);
-  //  std::cout << cerr<<std::endl;
+  //  std::cout<<GridLogMessage << cerr<<std::endl;
 
-  std::cout<<"=============================================================="<<std::endl;
-  std::cout<<"= Test Ddagger is the dagger of D by requiring                "<<std::endl;
-  std::cout<<"=  < phi | Deo | chi > * = < chi | Deo^dag| phi>  "<<std::endl;
-  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test Ddagger is the dagger of D by requiring                "<<std::endl;
+  std::cout<<GridLogMessage<<"=  < phi | Deo | chi > * = < chi | Deo^dag| phi>  "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
   
   LatticeFermion chi_e   (FrbGrid);
   LatticeFermion chi_o   (FrbGrid);
@@ -169,15 +169,15 @@ void  TestWhat(What & Ddwf,
   ComplexD cDpe = innerProduct(chi_e,dphi_e);
   ComplexD cDpo = innerProduct(chi_o,dphi_o);
 
-  std::cout <<"e "<<pDce<<" "<<cDpe <<std::endl;
-  std::cout <<"o "<<pDco<<" "<<cDpo <<std::endl;
+  std::cout<<GridLogMessage <<"e "<<pDce<<" "<<cDpe <<std::endl;
+  std::cout<<GridLogMessage <<"o "<<pDco<<" "<<cDpo <<std::endl;
 
-  std::cout <<"pDce - conj(cDpo) "<< pDce-conj(cDpo) <<std::endl;
-  std::cout <<"pDco - conj(cDpe) "<< pDco-conj(cDpe) <<std::endl;
+  std::cout<<GridLogMessage <<"pDce - conj(cDpo) "<< pDce-conj(cDpo) <<std::endl;
+  std::cout<<GridLogMessage <<"pDco - conj(cDpe) "<< pDco-conj(cDpe) <<std::endl;
 
-  std::cout<<"=============================================================="<<std::endl;
-  std::cout<<"= Test MeeInv Mee = 1                                         "<<std::endl;
-  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test MeeInv Mee = 1                                         "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
 
   pickCheckerboard(Even,chi_e,chi);
   pickCheckerboard(Odd ,chi_o,chi);
@@ -192,11 +192,11 @@ void  TestWhat(What & Ddwf,
   setCheckerboard(phi,phi_o);
 
   err = phi-chi;
-  std::cout << "norm diff   "<< norm2(err)<< std::endl;
+  std::cout<<GridLogMessage << "norm diff   "<< norm2(err)<< std::endl;
 
-  std::cout<<"=============================================================="<<std::endl;
-  std::cout<<"= Test MeeInvDag MeeDag = 1                                   "<<std::endl;
-  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test MeeInvDag MeeDag = 1                                   "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
 
   pickCheckerboard(Even,chi_e,chi);
   pickCheckerboard(Odd ,chi_o,chi);
@@ -211,11 +211,11 @@ void  TestWhat(What & Ddwf,
   setCheckerboard(phi,phi_o);
 
   err = phi-chi;
-  std::cout << "norm diff   "<< norm2(err)<< std::endl;
+  std::cout<<GridLogMessage << "norm diff   "<< norm2(err)<< std::endl;
 
-  std::cout<<"=============================================================="<<std::endl;
-  std::cout<<"= Test DiagMoo MpcDagMpc is Hermitian              "<<std::endl;
-  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test DiagMoo MpcDagMpc is Hermitian              "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
   
   random(*RNG5,phi);
   random(*RNG5,chi);
@@ -237,10 +237,10 @@ void  TestWhat(What & Ddwf,
   cDpe = innerProduct(chi_e,dphi_e);
   cDpo = innerProduct(chi_o,dphi_o);
 
-  std::cout <<"e "<<pDce<<" "<<cDpe <<std::endl;
-  std::cout <<"o "<<pDco<<" "<<cDpo <<std::endl;
+  std::cout<<GridLogMessage <<"e "<<pDce<<" "<<cDpe <<std::endl;
+  std::cout<<GridLogMessage <<"o "<<pDco<<" "<<cDpo <<std::endl;
 
-  std::cout <<"pDce - conj(cDpo) "<< pDco-conj(cDpo) <<std::endl;
-  std::cout <<"pDco - conj(cDpe) "<< pDce-conj(cDpe) <<std::endl;
+  std::cout<<GridLogMessage <<"pDce - conj(cDpo) "<< pDco-conj(cDpo) <<std::endl;
+  std::cout<<GridLogMessage <<"pDco - conj(cDpe) "<< pDce-conj(cDpe) <<std::endl;
   
 }
diff --git a/tests/Test_cayley_ldop_cr.cc b/tests/Test_cayley_ldop_cr.cc
index 2e67df26..48d7c9dc 100644
--- a/tests/Test_cayley_ldop_cr.cc
+++ b/tests/Test_cayley_ldop_cr.cc
@@ -52,9 +52,9 @@ int main (int argc, char ** argv)
   RealD mass=0.1;
   RealD M5=1.5;
 
-  std::cout << "**************************************************"<< std::endl;
-  std::cout << "Building g5R5 hermitian DWF operator" <<std::endl;
-  std::cout << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "Building g5R5 hermitian DWF operator" <<std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
   Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> HermIndefOp(Ddwf);
 
@@ -64,9 +64,9 @@ int main (int argc, char ** argv)
   typedef CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis> LittleDiracOperator;
   typedef LittleDiracOperator::CoarseVector                   CoarseVector;
 
-  std::cout << "**************************************************"<< std::endl;
-  std::cout << "Calling Aggregation class to build subspace" <<std::endl;
-  std::cout << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "Calling Aggregation class to build subspace" <<std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
   Subspace Aggregates(Coarse5d,FGrid);
   Aggregates.CreateSubspace(RNG5,HermDefOp);
@@ -80,22 +80,22 @@ int main (int argc, char ** argv)
   gaussian(CRNG,c_src);
   c_res=zero;
 
-  std::cout << "**************************************************"<< std::endl;
-  std::cout << "Solving mdagm-CG on coarse space "<< std::endl;
-  std::cout << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "Solving mdagm-CG on coarse space "<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   MdagMLinearOperator<LittleDiracOperator,CoarseVector> PosdefLdop(LittleDiracOp);
   ConjugateGradient<CoarseVector> CG(1.0e-6,10000);
   CG(PosdefLdop,c_src,c_res);
 
-  std::cout << "**************************************************"<< std::endl;
-  std::cout << "Solving indef-MCR on coarse space "<< std::endl;
-  std::cout << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "Solving indef-MCR on coarse space "<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   HermitianLinearOperator<LittleDiracOperator,CoarseVector> HermIndefLdop(LittleDiracOp);
   ConjugateResidual<CoarseVector> MCR(1.0e-6,10000);
   MCR(HermIndefLdop,c_src,c_res);
 
-  std::cout << "**************************************************"<< std::endl;
-  std::cout << "Done "<< std::endl;
-  std::cout << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "Done "<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   Grid_finalize();
 }
diff --git a/tests/Test_cf_coarsen_support.cc b/tests/Test_cf_coarsen_support.cc
index 959f5a9b..caa862be 100644
--- a/tests/Test_cf_coarsen_support.cc
+++ b/tests/Test_cf_coarsen_support.cc
@@ -56,12 +56,12 @@ int main (int argc, char ** argv)
     
     for(int d=0;d<4;d++){
       HermIndefOp.OpDir(src,tmp,d,+1); result=result+tmp; 
-      std::cout<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
+      std::cout<<GridLogMessage<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
       HermIndefOp.OpDir(src,tmp,d,-1); result=result+tmp;
-      std::cout<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
+      std::cout<<GridLogMessage<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
     }
     err = result-ref;
-    std::cout<<"Error "<<norm2(err)<<std::endl;
+    std::cout<<GridLogMessage<<"Error "<<norm2(err)<<std::endl;
   }
 
   {
@@ -73,13 +73,13 @@ int main (int argc, char ** argv)
     
     for(int d=0;d<4;d++){
       HermIndefOp.OpDir(src,tmp,d,+1); result=result+tmp; 
-      std::cout<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
+      std::cout<<GridLogMessage<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
       HermIndefOp.OpDir(src,tmp,d,-1); result=result+tmp;
-      std::cout<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
+      std::cout<<GridLogMessage<<"dir "<<d<<" tmp "<<norm2(tmp)<<std::endl;
     }
 
     err = result-ref;
-    std::cout<<"Error "<<norm2(err)<<std::endl;
+    std::cout<<GridLogMessage<<"Error "<<norm2(err)<<std::endl;
   }
 
 
diff --git a/tests/Test_contfrac_cg.cc b/tests/Test_contfrac_cg.cc
index 1b937cef..da6a81b9 100644
--- a/tests/Test_contfrac_cg.cc
+++ b/tests/Test_contfrac_cg.cc
@@ -53,7 +53,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   int threads = GridThread::GetThreads();
-  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+  std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
 
   const int Ls=9;
   GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
@@ -74,20 +74,20 @@ int main (int argc, char ** argv)
   RealD M5  =1.8;
 
 
-  std::cout <<"OverlapWilsonContFracTanhFermion  test"<<std::endl;
+  std::cout<<GridLogMessage <<"OverlapWilsonContFracTanhFermion  test"<<std::endl;
   OverlapWilsonContFracTanhFermion Dcf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
   TestCGinversions<OverlapWilsonContFracTanhFermion>(Dcf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
-  std::cout <<"OverlapWilsonContFracZolotarevFermion  test"<<std::endl;
+  std::cout<<GridLogMessage <<"OverlapWilsonContFracZolotarevFermion  test"<<std::endl;
   OverlapWilsonContFracZolotarevFermion Dcfz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,6.0);
   TestCGinversions<OverlapWilsonContFracZolotarevFermion>(Dcfz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
 
-  std::cout <<"OverlapWilsonPartialFractionTanhFermion  test"<<std::endl;
+  std::cout<<GridLogMessage <<"OverlapWilsonPartialFractionTanhFermion  test"<<std::endl;
   OverlapWilsonPartialFractionTanhFermion Dpf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
   TestCGinversions<OverlapWilsonPartialFractionTanhFermion>(Dpf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
-  std::cout <<"OverlapWilsonPartialFractionZolotarevFermion  test"<<std::endl;
+  std::cout<<GridLogMessage <<"OverlapWilsonPartialFractionZolotarevFermion  test"<<std::endl;
   OverlapWilsonPartialFractionZolotarevFermion Dpfz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,6.0);
   TestCGinversions<OverlapWilsonPartialFractionZolotarevFermion>(Dpfz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
@@ -102,11 +102,11 @@ void  TestCGinversions(What & Ddwf,
 		       GridParallelRNG *RNG4,
 		       GridParallelRNG *RNG5)
 {
-  std::cout << "Testing unpreconditioned inverter"<<std::endl;
+  std::cout<<GridLogMessage << "Testing unpreconditioned inverter"<<std::endl;
   TestCGunprec<What>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,RNG4,RNG5);
-  std::cout << "Testing red black preconditioned inverter"<<std::endl;
+  std::cout<<GridLogMessage << "Testing red black preconditioned inverter"<<std::endl;
   TestCGprec<What>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,RNG4,RNG5);
-  std::cout << "Testing red black Schur inverter"<<std::endl;
+  std::cout<<GridLogMessage << "Testing red black Schur inverter"<<std::endl;
   TestCGschur<What>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,RNG4,RNG5);
 }
 
diff --git a/tests/Test_contfrac_even_odd.cc b/tests/Test_contfrac_even_odd.cc
index 5d39b7df..c8e5b093 100644
--- a/tests/Test_contfrac_even_odd.cc
+++ b/tests/Test_contfrac_even_odd.cc
@@ -29,7 +29,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   int threads = GridThread::GetThreads();
-  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+  std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
 
   const int Ls=9;
   GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
@@ -49,19 +49,19 @@ int main (int argc, char ** argv)
   RealD mass=0.1;
   RealD M5  =1.8;
 
-  std::cout <<"OverlapWilsonContFracTanhFermion  test"<<std::endl;
+  std::cout<<GridLogMessage <<"OverlapWilsonContFracTanhFermion  test"<<std::endl;
   OverlapWilsonContFracTanhFermion Dcf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
   TestWhat<OverlapWilsonContFracTanhFermion>(Dcf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
-  std::cout <<"OverlapWilsonContFracZolotarevFermion  test"<<std::endl;
+  std::cout<<GridLogMessage <<"OverlapWilsonContFracZolotarevFermion  test"<<std::endl;
   OverlapWilsonContFracZolotarevFermion Dcfz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,6.0);
   TestWhat<OverlapWilsonContFracZolotarevFermion>(Dcfz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
-  std::cout <<"OverlapWilsonPartialFractionTanhFermion  test"<<std::endl;
+  std::cout<<GridLogMessage <<"OverlapWilsonPartialFractionTanhFermion  test"<<std::endl;
   OverlapWilsonPartialFractionTanhFermion Dpf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
   TestWhat<OverlapWilsonPartialFractionTanhFermion>(Dpf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
-  std::cout <<"OverlapWilsonPartialFractionZolotarevFermion  test"<<std::endl;
+  std::cout<<GridLogMessage <<"OverlapWilsonPartialFractionZolotarevFermion  test"<<std::endl;
   OverlapWilsonPartialFractionZolotarevFermion Dpfz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,6.0);
   TestWhat<OverlapWilsonPartialFractionZolotarevFermion>(Dpfz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
@@ -92,40 +92,40 @@ void  TestWhat(What & Ddwf,
   LatticeFermion r_eo  (FGrid);
   LatticeFermion r_eeoo(FGrid);
 
-  std::cout<<"=========================================================="<<std::endl;
-  std::cout<<"= Testing that Meo + Moe + Moo + Mee = Munprec "<<std::endl;
-  std::cout<<"=========================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=========================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Testing that Meo + Moe + Moo + Mee = Munprec "<<std::endl;
+  std::cout<<GridLogMessage<<"=========================================================="<<std::endl;
 
   pickCheckerboard(Even,src_e,src);
   pickCheckerboard( Odd,src_o,src);
 
-  Ddwf.Meooe(src_e,r_o);  std::cout<<"Applied Meo "<<norm2(r_o)<<std::endl;
-  Ddwf.Meooe(src_o,r_e);  std::cout<<"Applied Moe "<<norm2(r_e)<<std::endl;
+  Ddwf.Meooe(src_e,r_o);  std::cout<<GridLogMessage<<"Applied Meo "<<norm2(r_o)<<std::endl;
+  Ddwf.Meooe(src_o,r_e);  std::cout<<GridLogMessage<<"Applied Moe "<<norm2(r_e)<<std::endl;
   setCheckerboard(r_eo,r_o);
   setCheckerboard(r_eo,r_e);
 
-  Ddwf.Mooee(src_e,r_e);  std::cout<<"Applied Mee"<<norm2(r_e)<<std::endl;
-  Ddwf.Mooee(src_o,r_o);  std::cout<<"Applied Moo"<<norm2(r_o)<<std::endl;
+  Ddwf.Mooee(src_e,r_e);  std::cout<<GridLogMessage<<"Applied Mee"<<norm2(r_e)<<std::endl;
+  Ddwf.Mooee(src_o,r_o);  std::cout<<GridLogMessage<<"Applied Moo"<<norm2(r_o)<<std::endl;
   setCheckerboard(r_eeoo,r_e);
   setCheckerboard(r_eeoo,r_o);
 
   r_eo=r_eo+r_eeoo;
   Ddwf.M(src,ref);  
 
-  //  std::cout << r_eo<<std::endl;
-  //  std::cout << ref <<std::endl;
+  //  std::cout<<GridLogMessage << r_eo<<std::endl;
+  //  std::cout<<GridLogMessage << ref <<std::endl;
 
   err= ref - r_eo;
-  std::cout << "EO norm diff   "<< norm2(err)<< " "<<norm2(ref)<< " " << norm2(r_eo) <<std::endl;
+  std::cout<<GridLogMessage << "EO norm diff   "<< norm2(err)<< " "<<norm2(ref)<< " " << norm2(r_eo) <<std::endl;
     
   LatticeComplex cerr(FGrid);
   cerr = localInnerProduct(err,err);
-  //  std::cout << cerr<<std::endl;
+  //  std::cout<<GridLogMessage << cerr<<std::endl;
 
-  std::cout<<"=============================================================="<<std::endl;
-  std::cout<<"= Test Ddagger is the dagger of D by requiring                "<<std::endl;
-  std::cout<<"=  < phi | Deo | chi > * = < chi | Deo^dag| phi>  "<<std::endl;
-  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test Ddagger is the dagger of D by requiring                "<<std::endl;
+  std::cout<<GridLogMessage<<"=  < phi | Deo | chi > * = < chi | Deo^dag| phi>  "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
   
   LatticeFermion chi_e   (FrbGrid);
   LatticeFermion chi_o   (FrbGrid);
@@ -155,15 +155,15 @@ void  TestWhat(What & Ddwf,
   ComplexD cDpe = innerProduct(chi_e,dphi_e);
   ComplexD cDpo = innerProduct(chi_o,dphi_o);
 
-  std::cout <<"e "<<pDce<<" "<<cDpe <<std::endl;
-  std::cout <<"o "<<pDco<<" "<<cDpo <<std::endl;
+  std::cout<<GridLogMessage <<"e "<<pDce<<" "<<cDpe <<std::endl;
+  std::cout<<GridLogMessage <<"o "<<pDco<<" "<<cDpo <<std::endl;
 
-  std::cout <<"pDce - conj(cDpo) "<< pDce-conj(cDpo) <<std::endl;
-  std::cout <<"pDco - conj(cDpe) "<< pDco-conj(cDpe) <<std::endl;
+  std::cout<<GridLogMessage <<"pDce - conj(cDpo) "<< pDce-conj(cDpo) <<std::endl;
+  std::cout<<GridLogMessage <<"pDco - conj(cDpe) "<< pDco-conj(cDpe) <<std::endl;
 
-  std::cout<<"=============================================================="<<std::endl;
-  std::cout<<"= Test MeeInv Mee = 1                                         "<<std::endl;
-  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test MeeInv Mee = 1                                         "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
 
   pickCheckerboard(Even,chi_e,chi);
   pickCheckerboard(Odd ,chi_o,chi);
@@ -178,11 +178,11 @@ void  TestWhat(What & Ddwf,
   setCheckerboard(phi,phi_o);
 
   err = phi-chi;
-  std::cout << "norm diff   "<< norm2(err)<< std::endl;
+  std::cout<<GridLogMessage << "norm diff   "<< norm2(err)<< std::endl;
 
-  std::cout<<"=============================================================="<<std::endl;
-  std::cout<<"= Test MeeInvDag MeeDag = 1                                   "<<std::endl;
-  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test MeeInvDag MeeDag = 1                                   "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
 
   pickCheckerboard(Even,chi_e,chi);
   pickCheckerboard(Odd ,chi_o,chi);
@@ -197,11 +197,11 @@ void  TestWhat(What & Ddwf,
   setCheckerboard(phi,phi_o);
 
   err = phi-chi;
-  std::cout << "norm diff   "<< norm2(err)<< std::endl;
+  std::cout<<GridLogMessage << "norm diff   "<< norm2(err)<< std::endl;
 
-  std::cout<<"=============================================================="<<std::endl;
-  std::cout<<"= Test MpcDagMpc is Hermitian              "<<std::endl;
-  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test MpcDagMpc is Hermitian              "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
   
   random(*RNG5,phi);
   random(*RNG5,chi);
@@ -223,10 +223,10 @@ void  TestWhat(What & Ddwf,
   cDpe = innerProduct(chi_e,dphi_e);
   cDpo = innerProduct(chi_o,dphi_o);
 
-  std::cout <<"e "<<pDce<<" "<<cDpe <<std::endl;
-  std::cout <<"o "<<pDco<<" "<<cDpo <<std::endl;
+  std::cout<<GridLogMessage <<"e "<<pDce<<" "<<cDpe <<std::endl;
+  std::cout<<GridLogMessage <<"o "<<pDco<<" "<<cDpo <<std::endl;
 
-  std::cout <<"pDce - conj(cDpo) "<< pDco-conj(cDpo) <<std::endl;
-  std::cout <<"pDco - conj(cDpe) "<< pDce-conj(cDpe) <<std::endl;
+  std::cout<<GridLogMessage <<"pDce - conj(cDpo) "<< pDco-conj(cDpo) <<std::endl;
+  std::cout<<GridLogMessage <<"pDco - conj(cDpe) "<< pDce-conj(cDpe) <<std::endl;
   
 }
diff --git a/tests/Test_cshift.cc b/tests/Test_cshift.cc
index cd85439c..febf1837 100644
--- a/tests/Test_cshift.cc
+++ b/tests/Test_cshift.cc
@@ -39,7 +39,7 @@ int main (int argc, char ** argv)
   for(int dir=0;dir<4;dir++){
     for(int shift=0;shift<latt_size[dir];shift++){
       if ( Fine.IsBoss() ) 
-	std::cout<<"Shifting by "<<shift<<" in direction"<<dir<<std::endl;
+	std::cout<<GridLogMessage<<"Shifting by "<<shift<<" in direction"<<dir<<std::endl;
 
 	ShiftU  = Cshift(U,dir,shift);    // Shift everything
 
diff --git a/tests/Test_cshift_red_black.cc b/tests/Test_cshift_red_black.cc
index 79ebcfc5..65af9857 100644
--- a/tests/Test_cshift_red_black.cc
+++ b/tests/Test_cshift_red_black.cc
@@ -47,9 +47,9 @@ int main (int argc, char ** argv)
   pickCheckerboard(Even,Ue,U);
   pickCheckerboard(Odd,Uo,U);
 
-  //  std::cout << U<<std::endl;
-  std::cout << "Ue " <<norm2(Ue)<<std::endl;
-  std::cout << "Uo " <<norm2(Uo)<<std::endl;
+  //  std::cout<<GridLogMessage << U<<std::endl;
+  std::cout<<GridLogMessage << "Ue " <<norm2(Ue)<<std::endl;
+  std::cout<<GridLogMessage << "Uo " <<norm2(Uo)<<std::endl;
 
 
   TComplex cm;
@@ -57,28 +57,28 @@ int main (int argc, char ** argv)
     if ( dir!=1 ) continue;
     for(int shift=0;shift<latt_size[dir];shift++){
 
-	std::cout<<"Shifting by "<<shift<<" in direction"<<dir<<std::endl;
+	std::cout<<GridLogMessage<<"Shifting by "<<shift<<" in direction"<<dir<<std::endl;
 
-	//	std::cout<<"Even grid"<<std::endl;
+	//	std::cout<<GridLogMessage<<"Even grid"<<std::endl;
 	ShiftUe = Cshift(Ue,dir,shift);    // Shift everything cb by cb
-	//	std::cout << "\tShiftUe " <<norm2(ShiftUe)<<std::endl;
+	//	std::cout<<GridLogMessage << "\tShiftUe " <<norm2(ShiftUe)<<std::endl;
 
-	//	std::cout<<"Odd grid"<<std::endl;
+	//	std::cout<<GridLogMessage<<"Odd grid"<<std::endl;
 	ShiftUo = Cshift(Uo,dir,shift);    
-	//	std::cout << "\tShiftUo " <<norm2(ShiftUo)<<std::endl;
+	//	std::cout<<GridLogMessage << "\tShiftUo " <<norm2(ShiftUo)<<std::endl;
 
-	//	std::cout<<"Recombined Even/Odd grids"<<std::endl;
+	//	std::cout<<GridLogMessage<<"Recombined Even/Odd grids"<<std::endl;
 	setCheckerboard(rbShiftU,ShiftUe);
 	setCheckerboard(rbShiftU,ShiftUo);
-	//	std::cout << "\trbShiftU " <<norm2(rbShiftU)<<std::endl;
+	//	std::cout<<GridLogMessage << "\trbShiftU " <<norm2(rbShiftU)<<std::endl;
 
-	//	std::cout<<"Full grid shift"<<std::endl;
+	//	std::cout<<GridLogMessage<<"Full grid shift"<<std::endl;
 	ShiftU  = Cshift(U,dir,shift);    // Shift everything
-	//	std::cout << "\tShiftU " <<norm2(rbShiftU)<<std::endl;
+	//	std::cout<<GridLogMessage << "\tShiftU " <<norm2(rbShiftU)<<std::endl;
 
 	std::vector<int> coor(4);
 
-	std::cout << "Checking the non-checkerboard shift"<<std::endl;
+	std::cout<<GridLogMessage << "Checking the non-checkerboard shift"<<std::endl;
 	for(coor[3]=0;coor[3]<latt_size[3];coor[3]++){
 	for(coor[2]=0;coor[2]<latt_size[2];coor[2]++){
 	for(coor[1]=0;coor[1]<latt_size[1];coor[1]++){
@@ -117,7 +117,7 @@ int main (int argc, char ** argv)
 	}}}}
 
 
-	std::cout << "Checking the checkerboard shift"<<std::endl;
+	std::cout<<GridLogMessage << "Checking the checkerboard shift"<<std::endl;
 	for(coor[3]=0;coor[3]<latt_size[3];coor[3]++){
 	for(coor[2]=0;coor[2]<latt_size[2];coor[2]++){
 	for(coor[1]=0;coor[1]<latt_size[1];coor[1]++){
@@ -153,7 +153,7 @@ int main (int argc, char ** argv)
 	    std::cerr<<"Expect "<<index<<" " << peer[0]<<","<<peer[1]<<","<<peer[2]<<","<<peer[3]<<std::endl;
 	    exit(-1);
 	  } else if (0) { 
-	    std::cout<<"PASS shift "<< shift<<" in dir "<< dir
+	    std::cout<<GridLogMessage<<"PASS shift "<< shift<<" in dir "<< dir
 		     <<" ["<<coor[0]<<","<<coor[1]<<","<<coor[2]<<","<<coor[3]<<"] = "
 		     << cm()()()<<" expect "<<scm<<"  "<<nrm<<std::endl;
 	  }
diff --git a/tests/Test_dwf_even_odd.cc b/tests/Test_dwf_even_odd.cc
index 3ad01076..4e3d9799 100644
--- a/tests/Test_dwf_even_odd.cc
+++ b/tests/Test_dwf_even_odd.cc
@@ -22,7 +22,7 @@ int main (int argc, char ** argv)
   Grid_init(&argc,&argv);
 
   int threads = GridThread::GetThreads();
-  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+  std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
 
 
   const int Ls=8;
@@ -67,40 +67,40 @@ int main (int argc, char ** argv)
   LatticeFermion r_eo  (FGrid);
   LatticeFermion r_eeoo(FGrid);
 
-  std::cout<<"=========================================================="<<std::endl;
-  std::cout<<"= Testing that Meo + Moe + Moo + Mee = Munprec "<<std::endl;
-  std::cout<<"=========================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=========================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Testing that Meo + Moe + Moo + Mee = Munprec "<<std::endl;
+  std::cout<<GridLogMessage<<"=========================================================="<<std::endl;
 
   pickCheckerboard(Even,src_e,src);
   pickCheckerboard(Odd,src_o,src);
 
-  Ddwf.Meooe(src_e,r_o);  std::cout<<"Applied Meo"<<std::endl;
-  Ddwf.Meooe(src_o,r_e);  std::cout<<"Applied Moe"<<std::endl;
+  Ddwf.Meooe(src_e,r_o);  std::cout<<GridLogMessage<<"Applied Meo"<<std::endl;
+  Ddwf.Meooe(src_o,r_e);  std::cout<<GridLogMessage<<"Applied Moe"<<std::endl;
   setCheckerboard(r_eo,r_o);
   setCheckerboard(r_eo,r_e);
 
-  Ddwf.Mooee(src_e,r_e);  std::cout<<"Applied Mee"<<std::endl;
-  Ddwf.Mooee(src_o,r_o);  std::cout<<"Applied Moo"<<std::endl;
+  Ddwf.Mooee(src_e,r_e);  std::cout<<GridLogMessage<<"Applied Mee"<<std::endl;
+  Ddwf.Mooee(src_o,r_o);  std::cout<<GridLogMessage<<"Applied Moo"<<std::endl;
   setCheckerboard(r_eeoo,r_e);
   setCheckerboard(r_eeoo,r_o);
 
   r_eo=r_eo+r_eeoo;
   Ddwf.M(src,ref);  
 
-  //  std::cout << r_eo<<std::endl;
-  //  std::cout << ref <<std::endl;
+  //  std::cout<<GridLogMessage << r_eo<<std::endl;
+  //  std::cout<<GridLogMessage << ref <<std::endl;
 
   err= ref - r_eo;
-  std::cout << "EO norm diff   "<< norm2(err)<< " "<<norm2(ref)<< " " << norm2(r_eo) <<std::endl;
+  std::cout<<GridLogMessage << "EO norm diff   "<< norm2(err)<< " "<<norm2(ref)<< " " << norm2(r_eo) <<std::endl;
     
   LatticeComplex cerr(FGrid);
   cerr = localInnerProduct(err,err);
-  //  std::cout << cerr<<std::endl;
+  //  std::cout<<GridLogMessage << cerr<<std::endl;
 
-  std::cout<<"=============================================================="<<std::endl;
-  std::cout<<"= Test Ddagger is the dagger of D by requiring                "<<std::endl;
-  std::cout<<"=  < phi | Deo | chi > * = < chi | Deo^dag| phi>  "<<std::endl;
-  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test Ddagger is the dagger of D by requiring                "<<std::endl;
+  std::cout<<GridLogMessage<<"=  < phi | Deo | chi > * = < chi | Deo^dag| phi>  "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
   
   LatticeFermion chi_e   (FrbGrid);
   LatticeFermion chi_o   (FrbGrid);
@@ -130,15 +130,15 @@ int main (int argc, char ** argv)
   ComplexD cDpe = innerProduct(chi_e,dphi_e);
   ComplexD cDpo = innerProduct(chi_o,dphi_o);
 
-  std::cout <<"e "<<pDce<<" "<<cDpe <<std::endl;
-  std::cout <<"o "<<pDco<<" "<<cDpo <<std::endl;
+  std::cout<<GridLogMessage <<"e "<<pDce<<" "<<cDpe <<std::endl;
+  std::cout<<GridLogMessage <<"o "<<pDco<<" "<<cDpo <<std::endl;
 
-  std::cout <<"pDce - conj(cDpo) "<< pDce-conj(cDpo) <<std::endl;
-  std::cout <<"pDco - conj(cDpe) "<< pDco-conj(cDpe) <<std::endl;
+  std::cout<<GridLogMessage <<"pDce - conj(cDpo) "<< pDce-conj(cDpo) <<std::endl;
+  std::cout<<GridLogMessage <<"pDco - conj(cDpe) "<< pDco-conj(cDpe) <<std::endl;
 
-  std::cout<<"=============================================================="<<std::endl;
-  std::cout<<"= Test MeeInv Mee = 1                                         "<<std::endl;
-  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test MeeInv Mee = 1                                         "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
 
   pickCheckerboard(Even,chi_e,chi);
   pickCheckerboard(Odd ,chi_o,chi);
@@ -153,11 +153,11 @@ int main (int argc, char ** argv)
   setCheckerboard(phi,phi_o);
 
   err = phi-chi;
-  std::cout << "norm diff   "<< norm2(err)<< std::endl;
+  std::cout<<GridLogMessage << "norm diff   "<< norm2(err)<< std::endl;
 
-  std::cout<<"=============================================================="<<std::endl;
-  std::cout<<"= Test MeeInvDag MeeDag = 1                                   "<<std::endl;
-  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test MeeInvDag MeeDag = 1                                   "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
 
   pickCheckerboard(Even,chi_e,chi);
   pickCheckerboard(Odd ,chi_o,chi);
@@ -172,11 +172,11 @@ int main (int argc, char ** argv)
   setCheckerboard(phi,phi_o);
 
   err = phi-chi;
-  std::cout << "norm diff   "<< norm2(err)<< std::endl;
+  std::cout<<GridLogMessage << "norm diff   "<< norm2(err)<< std::endl;
 
-  std::cout<<"=============================================================="<<std::endl;
-  std::cout<<"= Test MpcDagMpc is Hermitian              "<<std::endl;
-  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test MpcDagMpc is Hermitian              "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
   
   random(RNG5,phi);
   random(RNG5,chi);
@@ -199,11 +199,11 @@ int main (int argc, char ** argv)
   cDpe = innerProduct(chi_e,dphi_e);
   cDpo = innerProduct(chi_o,dphi_o);
 
-  std::cout <<"e "<<pDce<<" "<<cDpe <<std::endl;
-  std::cout <<"o "<<pDco<<" "<<cDpo <<std::endl;
+  std::cout<<GridLogMessage <<"e "<<pDce<<" "<<cDpe <<std::endl;
+  std::cout<<GridLogMessage <<"o "<<pDco<<" "<<cDpo <<std::endl;
 
-  std::cout <<"pDce - conj(cDpo) "<< pDco-conj(cDpo) <<std::endl;
-  std::cout <<"pDco - conj(cDpe) "<< pDce-conj(cDpe) <<std::endl;
+  std::cout<<GridLogMessage <<"pDce - conj(cDpo) "<< pDco-conj(cDpo) <<std::endl;
+  std::cout<<GridLogMessage <<"pDco - conj(cDpe) "<< pDce-conj(cDpe) <<std::endl;
   
   Grid_finalize();
 }
diff --git a/tests/Test_dwf_fpgcr.cc b/tests/Test_dwf_fpgcr.cc
index 969377a0..6c589c6c 100644
--- a/tests/Test_dwf_fpgcr.cc
+++ b/tests/Test_dwf_fpgcr.cc
@@ -54,35 +54,35 @@ int main (int argc, char ** argv)
   RealD M5=1.8;
   DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
 
-  std::cout<<"*********************************************************"<<std::endl;
-  std::cout<<"* Solving with MdagM VPGCR "<<std::endl;
-  std::cout<<"*********************************************************"<<std::endl;
+  std::cout<<GridLogMessage<<"*********************************************************"<<std::endl;
+  std::cout<<GridLogMessage<<"* Solving with MdagM VPGCR "<<std::endl;
+  std::cout<<GridLogMessage<<"*********************************************************"<<std::endl;
   MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermOp(Ddwf);
   result=zero;
   PGCR(HermOp,src,result);
 
-  std::cout<<"*********************************************************"<<std::endl;
-  std::cout<<"* Solving with g5-VPGCR "<<std::endl;
-  std::cout<<"*********************************************************"<<std::endl;
+  std::cout<<GridLogMessage<<"*********************************************************"<<std::endl;
+  std::cout<<GridLogMessage<<"* Solving with g5-VPGCR "<<std::endl;
+  std::cout<<GridLogMessage<<"*********************************************************"<<std::endl;
   Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> g5HermOp(Ddwf);
   result=zero;
   PGCR(g5HermOp,src,result);
 
-  std::cout<<"*********************************************************"<<std::endl;
-  std::cout<<"* Solving with MdagM-CR "<<std::endl;
-  std::cout<<"*********************************************************"<<std::endl;
+  std::cout<<GridLogMessage<<"*********************************************************"<<std::endl;
+  std::cout<<GridLogMessage<<"* Solving with MdagM-CR "<<std::endl;
+  std::cout<<GridLogMessage<<"*********************************************************"<<std::endl;
   result=zero;
   CR(HermOp,src,result);
 
-  std::cout<<"*********************************************************"<<std::endl;
-  std::cout<<"* Solving with g5-CR "<<std::endl;
-  std::cout<<"*********************************************************"<<std::endl;
+  std::cout<<GridLogMessage<<"*********************************************************"<<std::endl;
+  std::cout<<GridLogMessage<<"* Solving with g5-CR "<<std::endl;
+  std::cout<<GridLogMessage<<"*********************************************************"<<std::endl;
   result=zero;
   CR(g5HermOp,src,result);
 
-  std::cout<<"*********************************************************"<<std::endl;
-  std::cout<<"* Solving with MdagM-CG "<<std::endl;
-  std::cout<<"*********************************************************"<<std::endl;
+  std::cout<<GridLogMessage<<"*********************************************************"<<std::endl;
+  std::cout<<GridLogMessage<<"* Solving with MdagM-CG "<<std::endl;
+  std::cout<<GridLogMessage<<"*********************************************************"<<std::endl;
   result=zero;
   CG(HermOp,src,result);
 
diff --git a/tests/Test_dwf_hdcr.cc b/tests/Test_dwf_hdcr.cc
index ebdf68c8..29f6e998 100644
--- a/tests/Test_dwf_hdcr.cc
+++ b/tests/Test_dwf_hdcr.cc
@@ -53,7 +53,7 @@ public:
       //      _FineOperator.Op(p1,p2);// this is the G5 herm bit      
       RealD absp2=std::sqrt(norm2(p2));
       if(i%10==9)
-	std::cout << "Power method on mdagm "<<i<<" " << absp2/absp1<<std::endl;
+	std::cout<<GridLogMessage << "Power method on mdagm "<<i<<" " << absp2/absp1<<std::endl;
       p1=p2*(1.0/std::sqrt(absp2));
     }
   }
@@ -71,7 +71,7 @@ public:
     // Monitor completeness of low mode space
     _Aggregates.ProjectToSubspace  (Csrc,in);
     _Aggregates.PromoteFromSubspace(Csrc,out);
-    std::cout<<"Completeness: "<<std::sqrt(norm2(out)/norm2(in))<<std::endl;
+    std::cout<<GridLogMessage<<"Completeness: "<<std::sqrt(norm2(out)/norm2(in))<<std::endl;
 
     // Build some solvers
     ConjugateGradient<FineField>    fCG(1.0e-3,1000);
@@ -85,21 +85,21 @@ public:
     MdagMLinearOperator<Matrix,FineField> MdagMOp(_Matrix);
 
     Min=in;
-    std::cout<< " Preconditioner in  " << norm2(in)<<std::endl; 
+    std::cout<<GridLogMessage<< " Preconditioner in  " << norm2(in)<<std::endl; 
     _FineOperator.AdjOp(Min,tmp);
-    std::cout<< " Preconditioner tmp  " << norm2(in)<<std::endl; 
+    std::cout<<GridLogMessage<< " Preconditioner tmp  " << norm2(in)<<std::endl; 
 
     fCG(MdagMOp,tmp,out);
 
     _FineOperator.Op(out,tmp);
 
-    std::cout<< " Preconditioner in  " << norm2(in)<<std::endl; 
-    std::cout<< " Preconditioner out " << norm2(out)<<std::endl; 
-    std::cout<< " Preconditioner Aout" << norm2(tmp)<<std::endl; 
+    std::cout<<GridLogMessage<< " Preconditioner in  " << norm2(in)<<std::endl; 
+    std::cout<<GridLogMessage<< " Preconditioner out " << norm2(out)<<std::endl; 
+    std::cout<<GridLogMessage<< " Preconditioner Aout" << norm2(tmp)<<std::endl; 
 
     tmp = tmp - in;
     
-    std::cout<<"preconditioner thinks residual is "<<std::sqrt(norm2(tmp)/norm2(in))<<std::endl;
+    std::cout<<GridLogMessage<<"preconditioner thinks residual is "<<std::sqrt(norm2(tmp)/norm2(in))<<std::endl;
 
     /*
     //    _FineOperator.Op(Min,out);
@@ -145,7 +145,7 @@ public:
     // Monitor completeness of low mode space
     _Aggregates.ProjectToSubspace  (Csrc,in);
     _Aggregates.PromoteFromSubspace(Csrc,out);
-    std::cout<<"Coarse Grid Preconditioner\nCompleteness in: "<<std::sqrt(norm2(out)/norm2(in))<<std::endl;
+    std::cout<<GridLogMessage<<"Coarse Grid Preconditioner\nCompleteness in: "<<std::sqrt(norm2(out)/norm2(in))<<std::endl;
 
     // [PTM+Q] in = [1 - Q A] M in + Q in = Min + Q [ in -A Min]
     _FineOperator.Op(in,tmp);// this is the G5 herm bit
@@ -154,7 +154,7 @@ public:
     // Monitor completeness of low mode space
     _Aggregates.ProjectToSubspace  (Csrc,Min);
     _Aggregates.PromoteFromSubspace(Csrc,out);
-    std::cout<<"Completeness Min: "<<std::sqrt(norm2(out)/norm2(Min))<<std::endl;
+    std::cout<<GridLogMessage<<"Completeness Min: "<<std::sqrt(norm2(out)/norm2(Min))<<std::endl;
 
     _FineOperator.Op(Min,tmp);
     tmp = in - tmp;   // in - A Min
@@ -166,14 +166,14 @@ public:
 
     HermOp.Op(Csol,Ctmp);
     Ctmp=Ctmp-Csrc;
-    std::cout<<"coarse space true residual "<<std::sqrt(norm2(Ctmp)/norm2(Csrc))<<std::endl;
+    std::cout<<GridLogMessage<<"coarse space true residual "<<std::sqrt(norm2(Ctmp)/norm2(Csrc))<<std::endl;
     _Aggregates.PromoteFromSubspace(Csol,out);
 
     _FineOperator.Op(out,res);
     res=res-tmp;
-    std::cout<<"promoted sol residual "<<std::sqrt(norm2(res)/norm2(tmp))<<std::endl;
+    std::cout<<GridLogMessage<<"promoted sol residual "<<std::sqrt(norm2(res)/norm2(tmp))<<std::endl;
     _Aggregates.ProjectToSubspace  (Csrc,res);
-    std::cout<<"coarse space proj of residual "<<norm2(Csrc)<<std::endl;
+    std::cout<<GridLogMessage<<"coarse space proj of residual "<<norm2(Csrc)<<std::endl;
 
     
     out = out+Min; // additive coarse space correction
@@ -182,9 +182,9 @@ public:
     _FineOperator.Op(out,tmp);
     tmp=tmp-in;         // tmp is new residual
 
-    std::cout<< " Preconditioner in  " << norm2(in)<<std::endl; 
-    std::cout<< " Preconditioner out " << norm2(out)<<std::endl; 
-    std::cout<<"preconditioner thinks residual is "<<std::sqrt(norm2(tmp)/norm2(in))<<std::endl;
+    std::cout<<GridLogMessage<< " Preconditioner in  " << norm2(in)<<std::endl; 
+    std::cout<<GridLogMessage<< " Preconditioner out " << norm2(out)<<std::endl; 
+    std::cout<<GridLogMessage<<"preconditioner thinks residual is "<<std::sqrt(norm2(tmp)/norm2(in))<<std::endl;
 
   }
 #endif
@@ -210,7 +210,7 @@ public:
     // Monitor completeness of low mode space
     //    _Aggregates.ProjectToSubspace  (Csrc,in);
     //    _Aggregates.PromoteFromSubspace(Csrc,out);
-    //    std::cout<<"Coarse Grid Preconditioner\nCompleteness in: "<<std::sqrt(norm2(out)/norm2(in))<<std::endl;
+    //    std::cout<<GridLogMessage<<"Coarse Grid Preconditioner\nCompleteness in: "<<std::sqrt(norm2(out)/norm2(in))<<std::endl;
     
     _Aggregates.ProjectToSubspace  (Csrc,in);
     HermOp.AdjOp(Csrc,Ctmp);// Normal equations
@@ -229,7 +229,7 @@ public:
     _FineOperator.Op(out,tmp);
     tmp=tmp-in;         // tmp is new residual
 
-    std::cout<<"preconditioner thinks residual is "<<std::sqrt(norm2(tmp)/norm2(in))<<std::endl;
+    std::cout<<GridLogMessage<<"preconditioner thinks residual is "<<std::sqrt(norm2(tmp)/norm2(in))<<std::endl;
 
   }
 #endif
@@ -247,8 +247,8 @@ public:
 
     Ni = norm2(in);
 
-    for(int ilo=0;ilo<4;ilo++){
-      for(int ord=10;ord<60;ord+=10){
+    for(int ilo=0;ilo<3;ilo++){
+      for(int ord=5;ord<50;ord*=2){
 
 	_FineOperator.AdjOp(in,vec1);
 
@@ -258,7 +258,7 @@ public:
 	_FineOperator.Op(vec2,vec1);// this is the G5 herm bit
 	vec1  = in - vec1;   // tmp  = in - A Min
 	r=norm2(vec1);
-	std::cout << "Smoother resid "<<std::sqrt(r/Ni)<<std::endl;
+	std::cout<<GridLogMessage << "Smoother resid "<<std::sqrt(r/Ni)<<std::endl;
 
       }
     }
@@ -270,13 +270,13 @@ public:
     CoarseVector Ctmp(_CoarseOperator.Grid());
     CoarseVector Csol(_CoarseOperator.Grid()); Csol=zero;
 
-    ConjugateGradient<CoarseVector>  CG(1.0e-5,100000);
-    ConjugateGradient<FineField>    fCG(3.0e-2,1000);
+    ConjugateGradient<CoarseVector>  CG(1.0e-3,100000);
+    //    ConjugateGradient<FineField>    fCG(3.0e-2,1000);
 
     HermitianLinearOperator<CoarseOperator,CoarseVector>  HermOp(_CoarseOperator);
     MdagMLinearOperator<CoarseOperator,CoarseVector>     MdagMOp(_CoarseOperator);
     //    MdagMLinearOperator<Matrix,FineField>        fMdagMOp(_Matrix);
-    ShiftedMdagMLinearOperator<Matrix,FineField> fMdagMOp(_Matrix,1.0);
+    ShiftedMdagMLinearOperator<Matrix,FineField> fMdagMOp(_Matrix,0.0);
 
     FineField vec1(in._grid);
     FineField vec2(in._grid);
@@ -288,7 +288,7 @@ public:
 
     _Aggregates.ProjectToSubspace  (Csrc,in);
     _Aggregates.PromoteFromSubspace(Csrc,out);
-    std::cout<<"Completeness: "<<std::sqrt(norm2(out)/norm2(in))<<std::endl;
+    std::cout<<GridLogMessage<<"Completeness: "<<std::sqrt(norm2(out)/norm2(in))<<std::endl;
     
     //    ofstream fout("smoother");
     //    Cheby.csv(fout);
@@ -306,7 +306,7 @@ public:
     _FineOperator.AdjOp(in,vec1);// this is the G5 herm bit
     ChebyAccu(fMdagMOp,vec1,out);    // solves  MdagM = g5 M g5M
 
-    std::cout << "Smoother norm "<<norm2(out)<<std::endl;
+    std::cout<<GridLogMessage << "Smoother norm "<<norm2(out)<<std::endl;
 
     // Update with residual for out
     _FineOperator.Op(out,vec1);// this is the G5 herm bit
@@ -314,7 +314,7 @@ public:
 
     RealD r = norm2(vec1);
 
-    std::cout << "Smoother resid "<<std::sqrt(r/Ni)<< " " << r << " " << Ni <<std::endl;
+    std::cout<<GridLogMessage << "Smoother resid "<<std::sqrt(r/Ni)<< " " << r << " " << Ni <<std::endl;
     
     _Aggregates.ProjectToSubspace  (Csrc,vec1);
     HermOp.AdjOp(Csrc,Ctmp);// Normal equations
@@ -329,7 +329,7 @@ public:
     vec1  = in - vec1;   // tmp  = in - A Min
     r=norm2(vec1);
 
-    std::cout << "Coarse resid "<<std::sqrt(r/Ni)<<std::endl;
+    std::cout<<GridLogMessage << "Coarse resid "<<std::sqrt(r/Ni)<<std::endl;
 
     // Reapply smoother
     _FineOperator.Op(vec1,vec2);  // this is the G5 herm bit
@@ -339,7 +339,7 @@ public:
     _FineOperator.Op(out,vec1);// this is the G5 herm bit
     vec1  = in - vec1;   // tmp  = in - A Min
     r=norm2(vec1);
-    std::cout << "Smoother resid "<<std::sqrt(r/Ni)<<std::endl;
+    std::cout<<GridLogMessage << "Smoother resid "<<std::sqrt(r/Ni)<<std::endl;
 
   }
 
@@ -385,7 +385,7 @@ int main (int argc, char ** argv)
   LatticeGaugeField Umu(UGrid); 
 
   NerscField header;
-  std::string file("./ckpoint_lat.4000");
+  std::string file("./ckpoint_lat.400");
   readNerscConfiguration(Umu,header,file);
 
   //  SU3::ColdConfiguration(RNG4,Umu);
@@ -396,9 +396,9 @@ int main (int argc, char ** argv)
   RealD mass=0.01;
   RealD M5=1.8;
 
-  std::cout << "**************************************************"<< std::endl;
-  std::cout << "Building g5R5 hermitian DWF operator" <<std::endl;
-  std::cout << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "Building g5R5 hermitian DWF operator" <<std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
 
   const int nbasis = 32;
@@ -408,22 +408,22 @@ int main (int argc, char ** argv)
   typedef CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis>          CoarseOperator;
   typedef CoarseOperator::CoarseVector                                 CoarseVector;
 
-  std::cout << "**************************************************"<< std::endl;
-  std::cout << "Calling Aggregation class to build subspace" <<std::endl;
-  std::cout << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "Calling Aggregation class to build subspace" <<std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
   Subspace Aggregates(Coarse5d,FGrid);
   //  Aggregates.CreateSubspace(RNG5,HermDefOp,nbasis);
   assert ( (nbasis & 0x1)==0);
   int nb=nbasis/2;
-  std::cout << " nbasis/2 = "<<nb<<std::endl;
+  std::cout<<GridLogMessage << " nbasis/2 = "<<nb<<std::endl;
   Aggregates.CreateSubspace(RNG5,HermDefOp,nb);
   for(int n=0;n<nb;n++){
     G5R5(Aggregates.subspace[n+nb],Aggregates.subspace[n]);
-    std::cout<<n<<" subspace "<<norm2(Aggregates.subspace[n+nb])<<" "<<norm2(Aggregates.subspace[n]) <<std::endl;
+    std::cout<<GridLogMessage<<n<<" subspace "<<norm2(Aggregates.subspace[n+nb])<<" "<<norm2(Aggregates.subspace[n]) <<std::endl;
   }
   for(int n=0;n<nbasis;n++){
-    std::cout << "vec["<<n<<"] = "<<norm2(Aggregates.subspace[n])  <<std::endl;
+    std::cout<<GridLogMessage << "vec["<<n<<"] = "<<norm2(Aggregates.subspace[n])  <<std::endl;
   }
 
 //  for(int i=0;i<nbasis;i++){
@@ -432,77 +432,90 @@ int main (int argc, char ** argv)
 //  }
   result=zero;
   
-  std::cout << "**************************************************"<< std::endl;
-  std::cout << "Building coarse representation of Indef operator" <<std::endl;
-  std::cout << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "Building coarse representation of Indef operator" <<std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> HermIndefOp(Ddwf);
   CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis> LDOp(*Coarse5d);
   LDOp.CoarsenOperator(FGrid,HermIndefOp,Aggregates);
 
-  std::cout << "**************************************************"<< std::endl;
-  std::cout << "Testing some coarse space solvers  " <<std::endl;
-  std::cout << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "Testing some coarse space solvers  " <<std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   CoarseVector c_src (Coarse5d);
   CoarseVector c_res (Coarse5d);
   gaussian(CRNG,c_src);
   c_res=zero;
 
-  std::cout << "**************************************************"<< std::endl;
-  std::cout << "Solving posdef-CG on coarse space "<< std::endl;
-  std::cout << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "Solving posdef-CG on coarse space "<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   MdagMLinearOperator<CoarseOperator,CoarseVector> PosdefLdop(LDOp);
   ConjugateGradient<CoarseVector> CG(1.0e-6,100000);
   CG(PosdefLdop,c_src,c_res);
 
-  //  std::cout << "**************************************************"<< std::endl;
-  //  std::cout << "Solving indef-MCR on coarse space "<< std::endl;
-  //  std::cout << "**************************************************"<< std::endl;
+  //  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  //  std::cout<<GridLogMessage << "Solving indef-MCR on coarse space "<< std::endl;
+  //  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   //  HermitianLinearOperator<CoarseOperator,CoarseVector> HermIndefLdop(LDOp);
   //  ConjugateResidual<CoarseVector> MCR(1.0e-6,100000);
   //MCR(HermIndefLdop,c_src,c_res);
 
-  std::cout << "**************************************************"<< std::endl;
-  std::cout << "Building deflation preconditioner "<< std::endl;
-  std::cout << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "Building deflation preconditioner "<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
 
   MultiGridPreconditioner <vSpinColourVector,vTComplex,nbasis,DomainWallFermion> Precon(Aggregates, LDOp,HermIndefOp,Ddwf);
   TrivialPrecon<LatticeFermion> simple;
 
-  std::cout << "**************************************************"<< std::endl;
-  std::cout << "Testing smoother efficacy"<< std::endl;
-  std::cout << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "Testing smoother efficacy"<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   Precon.SmootherTest(src);
 
-  std::cout << "**************************************************"<< std::endl;
-  std::cout << "Unprec CG "<< std::endl;
-  std::cout << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "Unprec CG "<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   //  TrivialPrecon<LatticeFermion> simple;
   //  ConjugateGradient<LatticeFermion> fCG(1.0e-8,100000);
   //  fCG(HermDefOp,src,result);
   //  exit(0);
 
-  std::cout << "**************************************************"<< std::endl;
-  std::cout << "Testing GCR on indef matrix "<< std::endl;
-  std::cout << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "Testing GCR on indef matrix "<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   //  PrecGeneralisedConjugateResidual<LatticeFermion> UPGCR(1.0e-8,100000,simple,8,128);
   //  UPGCR(HermIndefOp,src,result);
 
   
   /// Get themax eval
-  std::cout << "**************************************************"<< std::endl;
-  std::cout <<" Applying power method to find spectral range      "<<std::endl;
-  std::cout << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage <<" Applying power method to find spectral range      "<<std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   Precon.PowerMethod(src);
 
-  std::cout << "**************************************************"<< std::endl;
-  std::cout << "Building a two level PGCR "<< std::endl;
-  std::cout << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "Building a two level PGCR "<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   PrecGeneralisedConjugateResidual<LatticeFermion> PGCR(1.0e-8,100000,Precon,8,128);
-  std::cout<<"checking norm src "<<norm2(src)<<std::endl;
+  std::cout<<GridLogMessage<<"checking norm src "<<norm2(src)<<std::endl;
   PGCR(HermIndefOp,src,result);
 
-  std::cout << "**************************************************"<< std::endl;
-  std::cout << "Done "<< std::endl;
-  std::cout << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "Red Black Prec CG "<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  SchurDiagMooeeOperator<DomainWallFermion,LatticeFermion> HermOpEO(Ddwf);
+  ConjugateGradient<LatticeFermion> pCG(1.0e-8,10000);
+
+  LatticeFermion    src_o(FrbGrid);
+  LatticeFermion result_o(FrbGrid);
+  pickCheckerboard(Odd,src_o,src);
+  result_o=zero;
+
+  pCG(HermOpEO,src_o,result_o);
+
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
+  std::cout<<GridLogMessage << "Done "<< std::endl;
+  std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   Grid_finalize();
 }
diff --git a/tests/Test_gamma.cc b/tests/Test_gamma.cc
index 1f9cb491..748409b1 100644
--- a/tests/Test_gamma.cc
+++ b/tests/Test_gamma.cc
@@ -35,19 +35,19 @@ int main (int argc, char ** argv)
   SpinVector lv; random(sRNG,lv);
   SpinVector rv; random(sRNG,rv);
 
-  //  std::cout << " Is pod " << std::is_pod<SpinVector>::value  << std::endl;
-  //  std::cout << " Is pod double   " << std::is_pod<double>::value  << std::endl;
-  //  std::cout << " Is pod ComplexF " << std::is_pod<ComplexF>::value  << std::endl;
-  //  std::cout << " Is triv double " << std::has_trivial_default_constructor<double>::value  << std::endl;
-  //  std::cout << " Is triv ComplexF " << std::has_trivial_default_constructor<ComplexF>::value  << std::endl;
-  //  std::cout << " Is pod Scalar<double> " << std::is_pod<iScalar<double> >::value  << std::endl;
-  //  std::cout << " Is pod Scalar<ComplexF> " << std::is_pod<iScalar<ComplexF> >::value  << std::endl;
-  //  std::cout << " Is pod Scalar<vComplexF> " << std::is_pod<iScalar<vComplexF> >::value  << std::endl;
-  //  std::cout << " Is pod Scalar<vComplexD> " << std::is_pod<iScalar<vComplexD> >::value  << std::endl;
-  //  std::cout << " Is pod Scalar<vRealF> " << std::is_pod<iScalar<vRealF> >::value  << std::endl;
-  //  std::cout << " Is pod Scalar<vRealD> " << std::is_pod<iScalar<vRealD> >::value  << std::endl;
-  //  std::cout << " Is triv Scalar<double> " <<std::has_trivial_default_constructor<iScalar<double> >::value << std::endl;
-  //  std::cout << " Is triv Scalar<vComplexD> "<<std::has_trivial_default_constructor<iScalar<vComplexD> >::value  << std::endl;
+  //  std::cout<<GridLogMessage << " Is pod " << std::is_pod<SpinVector>::value  << std::endl;
+  //  std::cout<<GridLogMessage << " Is pod double   " << std::is_pod<double>::value  << std::endl;
+  //  std::cout<<GridLogMessage << " Is pod ComplexF " << std::is_pod<ComplexF>::value  << std::endl;
+  //  std::cout<<GridLogMessage << " Is triv double " << std::has_trivial_default_constructor<double>::value  << std::endl;
+  //  std::cout<<GridLogMessage << " Is triv ComplexF " << std::has_trivial_default_constructor<ComplexF>::value  << std::endl;
+  //  std::cout<<GridLogMessage << " Is pod Scalar<double> " << std::is_pod<iScalar<double> >::value  << std::endl;
+  //  std::cout<<GridLogMessage << " Is pod Scalar<ComplexF> " << std::is_pod<iScalar<ComplexF> >::value  << std::endl;
+  //  std::cout<<GridLogMessage << " Is pod Scalar<vComplexF> " << std::is_pod<iScalar<vComplexF> >::value  << std::endl;
+  //  std::cout<<GridLogMessage << " Is pod Scalar<vComplexD> " << std::is_pod<iScalar<vComplexD> >::value  << std::endl;
+  //  std::cout<<GridLogMessage << " Is pod Scalar<vRealF> " << std::is_pod<iScalar<vRealF> >::value  << std::endl;
+  //  std::cout<<GridLogMessage << " Is pod Scalar<vRealD> " << std::is_pod<iScalar<vRealD> >::value  << std::endl;
+  //  std::cout<<GridLogMessage << " Is triv Scalar<double> " <<std::has_trivial_default_constructor<iScalar<double> >::value << std::endl;
+  //  std::cout<<GridLogMessage << " Is triv Scalar<vComplexD> "<<std::has_trivial_default_constructor<iScalar<vComplexD> >::value  << std::endl;
 
   for(int a=0;a<Ns;a++){
     ident()(a,a) = ComplexF(1.0);
@@ -66,8 +66,8 @@ int main (int argc, char ** argv)
 
     for(int i=0;i<Ns;i++){
 
-      if(i==0) std::cout << list[mu];
-      else     std::cout << list[12];
+      if(i==0) std::cout<<GridLogMessage << list[mu];
+      else     std::cout<<GridLogMessage << list[12];
 
       std::cout<<"(";
       for(int j=0;j<Ns;j++){
@@ -82,7 +82,7 @@ int main (int argc, char ** argv)
 	} else if ( abs(result()(i,j)() + Complex(1,0))==0){
 	  std::cout<< "-1";
 	}
-	std::cout<< ((j==Ns-1) ? ")" : "," );
+	std::cout<<((j==Ns-1) ? ")" : "," );
       }
       std::cout << std::endl;
     }
diff --git a/tests/Test_lie_generators.cc b/tests/Test_lie_generators.cc
index 588baeda..25d11759 100644
--- a/tests/Test_lie_generators.cc
+++ b/tests/Test_lie_generators.cc
@@ -20,27 +20,27 @@ int main (int argc, char ** argv)
   
   GridRedBlackCartesian * rbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(grid);
 
-  std::cout<<"*********************************************"<<std::endl;
-  std::cout<<"* Generators for SU(2)"<<std::endl;
-  std::cout<<"*********************************************"<<std::endl;
+  std::cout<<GridLogMessage<<"*********************************************"<<std::endl;
+  std::cout<<GridLogMessage<<"* Generators for SU(2)"<<std::endl;
+  std::cout<<GridLogMessage<<"*********************************************"<<std::endl;
   SU2::printGenerators();
   SU2::testGenerators();
 
-  std::cout<<"*********************************************"<<std::endl;
-  std::cout<<"* Generators for SU(3)"<<std::endl;
-  std::cout<<"*********************************************"<<std::endl;
+  std::cout<<GridLogMessage<<"*********************************************"<<std::endl;
+  std::cout<<GridLogMessage<<"* Generators for SU(3)"<<std::endl;
+  std::cout<<GridLogMessage<<"*********************************************"<<std::endl;
   SU3::printGenerators();
   SU3::testGenerators();
 
-  //  std::cout<<"*********************************************"<<std::endl;
-  //  std::cout<<"* Generators for SU(4)"<<std::endl;
-  //  std::cout<<"*********************************************"<<std::endl;
+  //  std::cout<<GridLogMessage<<"*********************************************"<<std::endl;
+  //  std::cout<<GridLogMessage<<"* Generators for SU(4)"<<std::endl;
+  //  std::cout<<GridLogMessage<<"*********************************************"<<std::endl;
   //  SU4::printGenerators();
   //  SU4::testGenerators();
 
-  //  std::cout<<"*********************************************"<<std::endl;
-  //  std::cout<<"* Generators for SU(5)"<<std::endl;
-  //  std::cout<<"*********************************************"<<std::endl;
+  //  std::cout<<GridLogMessage<<"*********************************************"<<std::endl;
+  //  std::cout<<GridLogMessage<<"* Generators for SU(5)"<<std::endl;
+  //  std::cout<<GridLogMessage<<"*********************************************"<<std::endl;
   //  SU5::printGenerators();
   //  SU5::testGenerators();
 
diff --git a/tests/Test_main.cc b/tests/Test_main.cc
index a98fbeb7..b2b2ab31 100644
--- a/tests/Test_main.cc
+++ b/tests/Test_main.cc
@@ -213,29 +213,29 @@ int main (int argc, char ** argv)
     
    
     random(SerialRNG, cm);
-    std::cout << cm << std::endl;
+    std::cout<<GridLogMessage << cm << std::endl;
 
     cm = Ta(cm);
     TComplex tracecm= trace(cm);      
-    std::cout << cm << std::endl;
+    std::cout<<GridLogMessage << cm << std::endl;
 
 
     cm = Exponentiate(cm, 2.0, 12);
-    std::cout << cm << "  " << std::endl;
+    std::cout<<GridLogMessage << cm << "  " << std::endl;
     Complex det = Determinant(cm);
-    std::cout << "determinant: " << det <<  std::endl;
-    std::cout << "norm: " << norm2(cm) <<  std::endl;
+    std::cout<<GridLogMessage << "determinant: " << det <<  std::endl;
+    std::cout<<GridLogMessage << "norm: " << norm2(cm) <<  std::endl;
 
     cm = ProjectOnGroup(cm);
-    std::cout << cm << "  " << std::endl;
-    std::cout << "norm: " << norm2(cm) <<  std::endl;
+    std::cout<<GridLogMessage << cm << "  " << std::endl;
+    std::cout<<GridLogMessage << "norm: " << norm2(cm) <<  std::endl;
     cm = ProjectOnGroup(cm);
-    std::cout << cm << "  " << std::endl;
-    std::cout << "norm: " << norm2(cm) <<  std::endl;
+    std::cout<<GridLogMessage << cm << "  " << std::endl;
+    std::cout<<GridLogMessage << "norm: " << norm2(cm) <<  std::endl;
 
 
     //    det = Determinant(cm);
-    //    std::cout << "determinant: " << det <<  std::endl;
+    //    std::cout<<GridLogMessage << "determinant: " << det <<  std::endl;
 
 
 //    Foo = Foo+scalar; // LatticeColourMatrix+Scalar
@@ -253,19 +253,19 @@ int main (int argc, char ** argv)
     random(FineRNG,cMat);
     cMat = Ta(cMat);
     peekSite(cm, cMat, mysite);
-    std::cout << cm << "  " << std::endl;
+    std::cout<<GridLogMessage << cm << "  " << std::endl;
     cm = Exponentiate(cm, 1.0, 12);
-    std::cout << cm << "  " << std::endl;
-    std::cout << "norm: " << norm2(cm) <<  std::endl;
+    std::cout<<GridLogMessage << cm << "  " << std::endl;
+    std::cout<<GridLogMessage << "norm: " << norm2(cm) <<  std::endl;
 
 
-    std::cout << "norm cMmat : " << norm2(cMat) <<  std::endl;
+    std::cout<<GridLogMessage << "norm cMmat : " << norm2(cMat) <<  std::endl;
     cMat = expMat(cMat, ComplexD(1.0, 0.0));
-    std::cout << "norm expMat: " << norm2(cMat) <<  std::endl;
+    std::cout<<GridLogMessage << "norm expMat: " << norm2(cMat) <<  std::endl;
     peekSite(cm, cMat, mysite);
-    std::cout << cm << "  " << std::endl;
-    std::cout << "determinant: " << Determinant(cm) <<  std::endl;
-    std::cout << "norm: " << norm2(cm) <<  std::endl;
+    std::cout<<GridLogMessage << cm << "  " << std::endl;
+    std::cout<<GridLogMessage << "determinant: " << Determinant(cm) <<  std::endl;
+    std::cout<<GridLogMessage << "norm: " << norm2(cm) <<  std::endl;
 
 
     // LatticeComplex trlcMat(&Fine);
@@ -434,7 +434,7 @@ int main (int argc, char ** argv)
 	pickCheckerboard(1,bFoo,Foo);
     
 	if ( Fine.IsBoss() ) {
-	  std::cout << "Shifting both parities by "<< shift <<" direction "<< dir <<std::endl;
+	  std::cout<<GridLogMessage << "Shifting both parities by "<< shift <<" direction "<< dir <<std::endl;
 	}
 	Shifted  = Cshift(Foo,dir,shift);    // Shift everything
 
@@ -545,7 +545,7 @@ int main (int argc, char ** argv)
         }}
     }}}}
 	if( Fine.IsBoss() ){
-	  std::cout << "LatticeColorMatrix * LatticeColorMatrix nrm diff = "<<nrm<<std::endl;
+	  std::cout<<GridLogMessage << "LatticeColorMatrix * LatticeColorMatrix nrm diff = "<<nrm<<std::endl;
 	}
       }}
 
@@ -553,7 +553,7 @@ int main (int argc, char ** argv)
  } // loop for omp
 
 
- std::cout << sizeof(vComplexF) << std::endl;
+ std::cout<<GridLogMessage << sizeof(vComplexF) << std::endl;
  
  Grid_finalize();
 }
diff --git a/tests/Test_multishift_sqrt.cc b/tests/Test_multishift_sqrt.cc
index 4fff263f..111a521f 100644
--- a/tests/Test_multishift_sqrt.cc
+++ b/tests/Test_multishift_sqrt.cc
@@ -82,7 +82,7 @@ int main (int argc, char ** argv)
   ////////////////////////////////////////
   // sqrt and inverse sqrt
   ////////////////////////////////////////
-  std::cout << "Generating degree "<<degree<<" for x^(1/2)"<<std::endl;
+  std::cout<<GridLogMessage << "Generating degree "<<degree<<" for x^(1/2)"<<std::endl;
   remez.generateApprox(degree,1,2);
 
   MultiShiftFunction Sqrt(remez,1.0e-6,false);
@@ -120,19 +120,19 @@ int main (int argc, char ** argv)
 
   error = reference - combined;
 
-  std::cout << " Reference "<<norm2(reference)<<std::endl;
-  std::cout << " combined  "<<norm2(combined) <<std::endl;
-  std::cout << " error     "<<norm2(error)    <<std::endl;
+  std::cout<<GridLogMessage << " Reference "<<norm2(reference)<<std::endl;
+  std::cout<<GridLogMessage << " combined  "<<norm2(combined) <<std::endl;
+  std::cout<<GridLogMessage << " error     "<<norm2(error)    <<std::endl;
 
   MSCG(Diagonal,src,summed);
   error = summed - combined;
-  std::cout << " summed-combined "<<norm2(error)    <<std::endl;
+  std::cout<<GridLogMessage << " summed-combined "<<norm2(error)    <<std::endl;
 
 
   src=1.0;
   Chebyshev<LatticeFermion> Cheby(0.1,40.0,200,InverseApproximation);
 
-  std::cout<<"Chebuy approx vector "<<std::endl;
+  std::cout<<GridLogMessage<<"Chebuy approx vector "<<std::endl;
   Cheby(Diagonal,src,combined);
   std::ofstream of("cheby");
   Cheby.csv(of);
@@ -140,10 +140,10 @@ int main (int argc, char ** argv)
   Diagonal.ApplyInverse(src,reference);
   error = reference - combined;
 
-  std::cout << "Chebyshev inverse test "<<std::endl;
-  std::cout << " Reference "<<norm2(reference)<<std::endl;
-  std::cout << " combined  "<<norm2(combined) <<std::endl;
-  std::cout << " error     "<<norm2(error)    <<std::endl;
+  std::cout<<GridLogMessage << "Chebyshev inverse test "<<std::endl;
+  std::cout<<GridLogMessage << " Reference "<<norm2(reference)<<std::endl;
+  std::cout<<GridLogMessage << " combined  "<<norm2(combined) <<std::endl;
+  std::cout<<GridLogMessage << " error     "<<norm2(error)    <<std::endl;
 
   Grid_finalize();
 }
diff --git a/tests/Test_nersc_io.cc b/tests/Test_nersc_io.cc
index f4afb6d0..34821e97 100644
--- a/tests/Test_nersc_io.cc
+++ b/tests/Test_nersc_io.cc
@@ -55,7 +55,7 @@ int main (int argc, char ** argv)
 #endif
   double vol = Fine.gSites();
   Complex PlaqScale(1.0/vol/6.0/3.0);
-  std::cout <<"PlaqScale" << PlaqScale<<std::endl;
+  std::cout<<GridLogMessage <<"PlaqScale" << PlaqScale<<std::endl;
 
   std::vector<TComplex> Plaq_T(orthosz);
   sliceSum(Plaq,Plaq_T,Nd-1);
@@ -66,29 +66,29 @@ int main (int argc, char ** argv)
   for(int t=0;t<Nt;t++){
     Plaq_T_sum = Plaq_T_sum+Plaq_T[t];
     Complex Pt=TensorRemove(Plaq_T[t]);
-    std::cout << "sliced ["<<t<<"]" <<Pt*PlaqScale*Real(Nt)<<std::endl;
+    std::cout<<GridLogMessage << "sliced ["<<t<<"]" <<Pt*PlaqScale*Real(Nt)<<std::endl;
   }
 
   {
     Complex Pt = TensorRemove(Plaq_T_sum);
-    std::cout << "total " <<Pt*PlaqScale<<std::endl;
+    std::cout<<GridLogMessage << "total " <<Pt*PlaqScale<<std::endl;
   }  
 
 
   TComplex Tp = sum(Plaq);
   Complex p  = TensorRemove(Tp);
-  std::cout << "calculated plaquettes " <<p*PlaqScale<<std::endl;
+  std::cout<<GridLogMessage << "calculated plaquettes " <<p*PlaqScale<<std::endl;
 
 
   Complex LinkTraceScale(1.0/vol/4.0/3.0);
   TComplex Tl = sum(LinkTrace);
   Complex l  = TensorRemove(Tl);
-  std::cout << "calculated link trace " <<l*LinkTraceScale<<std::endl;
+  std::cout<<GridLogMessage << "calculated link trace " <<l*LinkTraceScale<<std::endl;
 
   blockSum(cPlaq,Plaq);
   TComplex TcP = sum(cPlaq);
   Complex ll= TensorRemove(TcP);
-  std::cout << "coarsened plaquettes sum to " <<ll*PlaqScale<<std::endl;
+  std::cout<<GridLogMessage << "coarsened plaquettes sum to " <<ll*PlaqScale<<std::endl;
   
   Grid_finalize();
 }
diff --git a/tests/Test_quenched_update.cc b/tests/Test_quenched_update.cc
index e419d002..b084c9c8 100644
--- a/tests/Test_quenched_update.cc
+++ b/tests/Test_quenched_update.cc
@@ -47,7 +47,7 @@ int main (int argc, char ** argv)
 
     RealD plaq = ColourWilsonLoops::avgPlaquette(Umu);
 
-    std::cout<<"sweep "<<sweep<<" PLAQUETTE "<<plaq<<std::endl;
+    std::cout<<GridLogMessage<<"sweep "<<sweep<<" PLAQUETTE "<<plaq<<std::endl;
 
     for( int cb=0;cb<2;cb++ ) {
 
@@ -55,7 +55,7 @@ int main (int argc, char ** argv)
       mask= zero;
       setCheckerboard(mask,one);
 
-      //      std::cout<<mask<<std::endl;
+      //      std::cout<<GridLogMessage<<mask<<std::endl;
       for(int mu=0;mu<Nd;mu++){
 	
 	// Get Link and Staple term in action; must contain Beta and 
diff --git a/tests/Test_remez.cc b/tests/Test_remez.cc
index 5a6300e7..2401d254 100644
--- a/tests/Test_remez.cc
+++ b/tests/Test_remez.cc
@@ -8,7 +8,7 @@ int main (int argc, char ** argv)
 {
   Grid_init(&argc,&argv);
 
-  std::cout << "Testing Remez"<<std::endl;
+  std::cout<<GridLogMessage << "Testing Remez"<<std::endl;
 
   double     lo=0.01;
   double     hi=1.0;
@@ -20,13 +20,13 @@ int main (int argc, char ** argv)
   // sqrt and inverse sqrt
   ////////////////////////////////////////
 
-  std::cout << "Generating degree "<<degree<<" for x^(1/2)"<<std::endl;
+  std::cout<<GridLogMessage << "Generating degree "<<degree<<" for x^(1/2)"<<std::endl;
   remez.generateApprox(degree,1,2);
   MultiShiftFunction Sqrt(remez,1.0,false);
   MultiShiftFunction InvSqrt(remez,1.0,true);
 
 
-  std::cout << "Generating degree "<<degree<<" for x^(1/4)"<<std::endl;
+  std::cout<<GridLogMessage << "Generating degree "<<degree<<" for x^(1/4)"<<std::endl;
   remez.generateApprox(degree,1,4);
   MultiShiftFunction SqrtSqrt(remez,1.0,false);
   MultiShiftFunction InvSqrtSqrt(remez,1.0,true);
@@ -49,10 +49,10 @@ int main (int argc, char ** argv)
   double aisx =InvSqrt.approx(x);
   double aissx=InvSqrtSqrt.approx(x);
 
-  std::cout << "x^(1/2) : "<<sx<<" "<<asx<<std::endl;
-  std::cout << "x^(1/4) : "<<ssx<<" "<<assx<<std::endl;
-  std::cout << "x^(-1/2): "<<isx<<" "<<aisx<<std::endl;
-  std::cout << "x^(-1/4): "<<issx<<" "<<aissx<<std::endl;
+  std::cout<<GridLogMessage << "x^(1/2) : "<<sx<<" "<<asx<<std::endl;
+  std::cout<<GridLogMessage << "x^(1/4) : "<<ssx<<" "<<assx<<std::endl;
+  std::cout<<GridLogMessage << "x^(-1/2): "<<isx<<" "<<aisx<<std::endl;
+  std::cout<<GridLogMessage << "x^(-1/4): "<<issx<<" "<<aissx<<std::endl;
   assert(fabs(sx-asx)<1.0e-6);
   assert(fabs(ssx-assx)<1.0e-6);
   assert(fabs(isx-aisx)<1.0e-6);
diff --git a/tests/Test_rng.cc b/tests/Test_rng.cc
index 9cd05a22..da8da734 100644
--- a/tests/Test_rng.cc
+++ b/tests/Test_rng.cc
@@ -24,27 +24,27 @@ int main (int argc, char ** argv)
 
   SpinMatrix rnd  ; 
   random(sRNG,rnd);
-  std::cout<<"Random Spin Matrix (random_device)\n"<< rnd<<std::endl;
+  std::cout<<GridLogMessage<<"Random Spin Matrix (random_device)\n"<< rnd<<std::endl;
 
   random(fsRNG,rnd);
-  std::cout<<"Random Spin Matrix (fixed seed)\n"<< rnd<<std::endl;
+  std::cout<<GridLogMessage<<"Random Spin Matrix (fixed seed)\n"<< rnd<<std::endl;
 
   SpinVector rv; 
   random(sRNG,rv);
-  std::cout<<"Random Spin Vector (random device)\n"<< rv<<std::endl;
+  std::cout<<GridLogMessage<<"Random Spin Vector (random device)\n"<< rv<<std::endl;
 
   random(fsRNG,rv);
-  std::cout<<"Random Spin Vector (fixed seed)\n"<< rv<<std::endl;
+  std::cout<<GridLogMessage<<"Random Spin Vector (fixed seed)\n"<< rv<<std::endl;
 
   gaussian(fsRNG,rv);
-  std::cout<<"Gaussian Spin Vector (fixed seed)\n"<< rv<<std::endl;
+  std::cout<<GridLogMessage<<"Gaussian Spin Vector (fixed seed)\n"<< rv<<std::endl;
 
   LatticeColourVector lcv(&Grid);
   random(pRNG,lcv);
-  std::cout<<"Random Lattice Colour Vector (random device)\n"<< lcv<<std::endl;
+  std::cout<<GridLogMessage<<"Random Lattice Colour Vector (random device)\n"<< lcv<<std::endl;
 
   random(fpRNG,lcv);
-  std::cout<<"Random Lattice Colour Vector (fixed seed)\n"<< lcv<<std::endl;
+  std::cout<<GridLogMessage<<"Random Lattice Colour Vector (fixed seed)\n"<< lcv<<std::endl;
 
   Grid_finalize();
 }
diff --git a/tests/Test_rng_fixed.cc b/tests/Test_rng_fixed.cc
index a4171401..f2271ca1 100644
--- a/tests/Test_rng_fixed.cc
+++ b/tests/Test_rng_fixed.cc
@@ -20,31 +20,31 @@ int main (int argc, char ** argv)
   GridParallelRNG          fpRNG(&Grid);  fpRNG.SeedFixedIntegers(seeds);
 
   vComplexF tmp; random(fsRNG,tmp);
-  std::cout<<"Random vComplexF (fixed seed)\n"<< tmp<<std::endl;
-  std::cout<<"conjugate(tmp)\n"<< conjugate(tmp)<<std::endl;
-  std::cout<<"conjugate(tmp)*tmp\n"<< conjugate(tmp)*tmp<<std::endl;
-  std::cout<<"innerProduct"<< innerProduct(tmp,tmp)<<std::endl;
-  std::cout<<"Reduce(innerProduct)"<< Reduce(innerProduct(tmp,tmp))<<std::endl;
+  std::cout<<GridLogMessage<<"Random vComplexF (fixed seed)\n"<< tmp<<std::endl;
+  std::cout<<GridLogMessage<<"conjugate(tmp)\n"<< conjugate(tmp)<<std::endl;
+  std::cout<<GridLogMessage<<"conjugate(tmp)*tmp\n"<< conjugate(tmp)*tmp<<std::endl;
+  std::cout<<GridLogMessage<<"innerProduct"<< innerProduct(tmp,tmp)<<std::endl;
+  std::cout<<GridLogMessage<<"Reduce(innerProduct)"<< Reduce(innerProduct(tmp,tmp))<<std::endl;
 
   SpinMatrix rnd  ; 
   random(fsRNG,rnd);
-  std::cout<<"Random Spin Matrix (fixed seed)\n"<< rnd<<std::endl;
+  std::cout<<GridLogMessage<<"Random Spin Matrix (fixed seed)\n"<< rnd<<std::endl;
 
   SpinVector rv; 
   random(fsRNG,rv);
-  std::cout<<"Random Spin Vector (fixed seed)\n"<< rv<<std::endl;
+  std::cout<<GridLogMessage<<"Random Spin Vector (fixed seed)\n"<< rv<<std::endl;
 
   gaussian(fsRNG,rv);
-  std::cout<<"Gaussian Spin Vector (fixed seed)\n"<< rv<<std::endl;
+  std::cout<<GridLogMessage<<"Gaussian Spin Vector (fixed seed)\n"<< rv<<std::endl;
 
   LatticeColourVector lcv(&Grid);
 
   LatticeFermion src(&Grid); random(fpRNG,src);
-  std::cout << "src norm : " << norm2(src)<<std::endl;
-  std::cout << "src " << src<<std::endl;
+  std::cout<<GridLogMessage << "src norm : " << norm2(src)<<std::endl;
+  std::cout<<GridLogMessage << "src " << src<<std::endl;
 
   random(fpRNG,lcv);
-  std::cout<<"Random Lattice Colour Vector (fixed seed)\n"<< lcv<<std::endl;
+  std::cout<<GridLogMessage<<"Random Lattice Colour Vector (fixed seed)\n"<< lcv<<std::endl;
 
 
   Grid_finalize();
diff --git a/tests/Test_simd.cc b/tests/Test_simd.cc
index 800001be..d4256872 100644
--- a/tests/Test_simd.cc
+++ b/tests/Test_simd.cc
@@ -113,18 +113,18 @@ void Tester(const functor &func)
   }
 
   extract<vec,scal>(v_result,result);
-  std::cout << " " << func.name()<<std::endl;
+  std::cout<<GridLogMessage << " " << func.name()<<std::endl;
 
   int ok=0;
   for(int i=0;i<Nsimd;i++){
     if ( abs(reference[i]-result[i])>0){
-      std::cout<< "*****" << std::endl;
-      std::cout<< "["<<i<<"] "<< abs(reference[i]-result[i]) << " " <<reference[i]<< " " << result[i]<<std::endl;
+      std::cout<<GridLogMessage<< "*****" << std::endl;
+      std::cout<<GridLogMessage<< "["<<i<<"] "<< abs(reference[i]-result[i]) << " " <<reference[i]<< " " << result[i]<<std::endl;
       ok++;
     }
   }
   if ( ok==0 ) {
-    std::cout << " OK!" <<std::endl;
+    std::cout<<GridLogMessage << " OK!" <<std::endl;
   }
   assert(ok==0);
 }
@@ -164,16 +164,16 @@ void ReductionTester(const functor &func)
     reference+=tmp;
   }
 
-  std::cout << " " << func.name()<<std::endl;
+  std::cout<<GridLogMessage << " " << func.name()<<std::endl;
 
   int ok=0;
   if ( abs(reference-result)/abs(reference) > 1.0e-6 ){ // rounding is possible for reduce order
-    std::cout<< "*****" << std::endl;
-    std::cout<< abs(reference-result) << " " <<reference<< " " << result<<std::endl;
+    std::cout<<GridLogMessage<< "*****" << std::endl;
+    std::cout<<GridLogMessage<< abs(reference-result) << " " <<reference<< " " << result<<std::endl;
     ok++;
   }
   if ( ok==0 ) {
-    std::cout << " OK!" <<std::endl;
+    std::cout<<GridLogMessage << " OK!" <<std::endl;
   }
   assert(ok==0);
 }
@@ -194,9 +194,9 @@ int main (int argc, char ** argv)
   // Insist that operations on random scalars gives
   // identical results to on vectors.
 
-  std::cout << "==================================="<<  std::endl;
-  std::cout << "Testing vComplexF "<<std::endl;
-  std::cout << "==================================="<<  std::endl;
+  std::cout<<GridLogMessage << "==================================="<<  std::endl;
+  std::cout<<GridLogMessage << "Testing vComplexF "<<std::endl;
+  std::cout<<GridLogMessage << "==================================="<<  std::endl;
 
   Tester<ComplexF,vComplexF>(funcTimesI());
   Tester<ComplexF,vComplexF>(funcTimesMinusI());
@@ -208,9 +208,9 @@ int main (int argc, char ** argv)
   Tester<ComplexF,vComplexF>(funcInnerProduct());
   ReductionTester<ComplexF,ComplexF,vComplexF>(funcReduce());
 
-  std::cout << "==================================="<<  std::endl;
-  std::cout << "Testing vComplexD "<<std::endl;
-  std::cout << "==================================="<<  std::endl;
+  std::cout<<GridLogMessage << "==================================="<<  std::endl;
+  std::cout<<GridLogMessage << "Testing vComplexD "<<std::endl;
+  std::cout<<GridLogMessage << "==================================="<<  std::endl;
 
 
   Tester<ComplexD,vComplexD>(funcTimesI());
@@ -223,9 +223,9 @@ int main (int argc, char ** argv)
   Tester<ComplexD,vComplexD>(funcInnerProduct());
   ReductionTester<ComplexD,ComplexD,vComplexD>(funcReduce());
 
-  std::cout << "==================================="<<  std::endl;
-  std::cout << "Testing vRealF "<<std::endl;
-  std::cout << "==================================="<<  std::endl;
+  std::cout<<GridLogMessage << "==================================="<<  std::endl;
+  std::cout<<GridLogMessage << "Testing vRealF "<<std::endl;
+  std::cout<<GridLogMessage << "==================================="<<  std::endl;
 
 
   Tester<RealF,vRealF>(funcPlus());
@@ -236,9 +236,9 @@ int main (int argc, char ** argv)
   Tester<RealF,vRealF>(funcInnerProduct());
   ReductionTester<RealF,RealF,vRealF>(funcReduce());
 
-  std::cout << "==================================="<<  std::endl;
-  std::cout << "Testing vRealD "<<std::endl;
-  std::cout << "==================================="<<  std::endl;
+  std::cout<<GridLogMessage << "==================================="<<  std::endl;
+  std::cout<<GridLogMessage << "Testing vRealD "<<std::endl;
+  std::cout<<GridLogMessage << "==================================="<<  std::endl;
 
   Tester<RealD,vRealD>(funcPlus());
   Tester<RealD,vRealD>(funcMinus());
diff --git a/tests/Test_stencil.cc b/tests/Test_stencil.cc
index 3f6bb4fe..d455c1c1 100644
--- a/tests/Test_stencil.cc
+++ b/tests/Test_stencil.cc
@@ -34,7 +34,7 @@ int main (int argc, char ** argv)
     for(int dir=0;dir<4;dir++){
       for(int disp=0;disp<Fine._fdimensions[dir];disp++){
 
-	std::cout << "Using stencil to shift dim "<<dir<< " by "<<disp<<std::endl;
+	std::cout<<GridLogMessage << "Using stencil to shift dim "<<dir<< " by "<<disp<<std::endl;
 	// start to test the Cartesian npoint stencil infrastructure
 	int npoint=1;
 	std::vector<int> directions(npoint,dir);
@@ -73,7 +73,7 @@ int main (int argc, char ** argv)
 	Real nrmB = norm2(Bar);
 	Diff = Check-Bar;
 	Real nrm  = norm2(Diff);
-	std::cout<<"N2diff ="<<nrm<<" "<<nrmC<<" " <<nrmB<<std::endl;
+	std::cout<<GridLogMessage<<"N2diff ="<<nrm<<" "<<nrmC<<" " <<nrmB<<std::endl;
 
 	Real snrmC =0;
 	Real snrmB =0;
@@ -110,7 +110,7 @@ int main (int argc, char ** argv)
 	 
 	}}}}
 
-	std::cout<<"scalar N2diff = "<<snrm<<" " <<snrmC<<" "<<snrmB<<std::endl;
+	std::cout<<GridLogMessage<<"scalar N2diff = "<<snrm<<" " <<snrmC<<" "<<snrmB<<std::endl;
 
 
       }
diff --git a/tests/Test_wilson_even_odd.cc b/tests/Test_wilson_even_odd.cc
index 90d15e99..66ac89dc 100644
--- a/tests/Test_wilson_even_odd.cc
+++ b/tests/Test_wilson_even_odd.cc
@@ -27,7 +27,7 @@ int main (int argc, char ** argv)
   GridRedBlackCartesian     RBGrid(latt_size,simd_layout,mpi_layout);
 
   int threads = GridThread::GetThreads();
-  std::cout << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+  std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
 
   std::vector<int> seeds({1,2,3,4});
 
@@ -55,9 +55,9 @@ int main (int argc, char ** argv)
   Umu=zero;
   for(int nn=0;nn<Nd;nn++){
     random(pRNG,U[nn]);
-    std::cout<<"U[nn]"<<norm2(U[nn])<<std::endl;
+    std::cout<<GridLogMessage<<"U[nn]"<<norm2(U[nn])<<std::endl;
     PokeIndex<LorentzIndex>(Umu,U[nn],nn);
-    std::cout<<"Umu"<<norm2(Umu)<<std::endl;
+    std::cout<<GridLogMessage<<"Umu"<<norm2(Umu)<<std::endl;
   }
 
   RealD mass=0.1;
@@ -70,30 +70,30 @@ int main (int argc, char ** argv)
   LatticeFermion r_o   (&RBGrid);
   LatticeFermion r_eo  (&Grid);
 
-  std::cout<<"=========================================================="<<std::endl;
-  std::cout<<"= Testing that Deo + Doe = Dunprec "<<std::endl;
-  std::cout<<"=========================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=========================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Testing that Deo + Doe = Dunprec "<<std::endl;
+  std::cout<<GridLogMessage<<"=========================================================="<<std::endl;
 
   pickCheckerboard(Even,src_e,src);
   pickCheckerboard(Odd,src_o,src);
 
-  Dw.Meooe(src_e,r_o);  std::cout<<"Applied Meo"<<std::endl;
-  Dw.Meooe(src_o,r_e);  std::cout<<"Applied Moe"<<std::endl;
+  Dw.Meooe(src_e,r_o);  std::cout<<GridLogMessage<<"Applied Meo"<<std::endl;
+  Dw.Meooe(src_o,r_e);  std::cout<<GridLogMessage<<"Applied Moe"<<std::endl;
   Dw.Dhop (src,ref,DaggerNo);
 
   setCheckerboard(r_eo,r_o);
   setCheckerboard(r_eo,r_e);
 
   err= ref - r_eo;
-  std::cout << "EO norm diff   "<< norm2(err)<< " "<<norm2(ref)<< " " << norm2(r_eo) <<std::endl;
+  std::cout<<GridLogMessage << "EO norm diff   "<< norm2(err)<< " "<<norm2(ref)<< " " << norm2(r_eo) <<std::endl;
 
   LatticeComplex cerr(&Grid);
   cerr = localInnerProduct(err,err);
 
-  std::cout<<"=============================================================="<<std::endl;
-  std::cout<<"= Test Ddagger is the dagger of D by requiring                "<<std::endl;
-  std::cout<<"=  < phi | Deo | chi > * = < chi | Deo^dag| phi>  "<<std::endl;
-  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test Ddagger is the dagger of D by requiring                "<<std::endl;
+  std::cout<<GridLogMessage<<"=  < phi | Deo | chi > * = < chi | Deo^dag| phi>  "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
   
   LatticeFermion chi_e   (&RBGrid);
   LatticeFermion chi_o   (&RBGrid);
@@ -123,15 +123,15 @@ int main (int argc, char ** argv)
   ComplexD cDpe = innerProduct(chi_e,dphi_e);
   ComplexD cDpo = innerProduct(chi_o,dphi_o);
 
-  std::cout <<"e "<<pDce<<" "<<cDpe <<std::endl;
-  std::cout <<"o "<<pDco<<" "<<cDpo <<std::endl;
+  std::cout<<GridLogMessage <<"e "<<pDce<<" "<<cDpe <<std::endl;
+  std::cout<<GridLogMessage <<"o "<<pDco<<" "<<cDpo <<std::endl;
 
-  std::cout <<"pDce - conj(cDpo) "<< pDce-conj(cDpo) <<std::endl;
-  std::cout <<"pDco - conj(cDpe) "<< pDco-conj(cDpe) <<std::endl;
+  std::cout<<GridLogMessage <<"pDce - conj(cDpo) "<< pDce-conj(cDpo) <<std::endl;
+  std::cout<<GridLogMessage <<"pDco - conj(cDpe) "<< pDco-conj(cDpe) <<std::endl;
 
-  std::cout<<"=============================================================="<<std::endl;
-  std::cout<<"= Test MeeInv Mee = 1                                         "<<std::endl;
-  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test MeeInv Mee = 1                                         "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
 
   pickCheckerboard(Even,chi_e,chi);
   pickCheckerboard(Odd ,chi_o,chi);
@@ -146,11 +146,11 @@ int main (int argc, char ** argv)
   setCheckerboard(phi,phi_o);
 
   err = phi-chi;
-  std::cout << "norm diff   "<< norm2(err)<< std::endl;
+  std::cout<<GridLogMessage << "norm diff   "<< norm2(err)<< std::endl;
 
-  std::cout<<"=============================================================="<<std::endl;
-  std::cout<<"= Test MeeInvDag MeeDag = 1                                   "<<std::endl;
-  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test MeeInvDag MeeDag = 1                                   "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
 
   pickCheckerboard(Even,chi_e,chi);
   pickCheckerboard(Odd ,chi_o,chi);
@@ -165,11 +165,11 @@ int main (int argc, char ** argv)
   setCheckerboard(phi,phi_o);
 
   err = phi-chi;
-  std::cout << "norm diff   "<< norm2(err)<< std::endl;
+  std::cout<<GridLogMessage << "norm diff   "<< norm2(err)<< std::endl;
 
-  std::cout<<"=============================================================="<<std::endl;
-  std::cout<<"= Test MpcDagMpc is Hermitian              "<<std::endl;
-  std::cout<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test MpcDagMpc is Hermitian              "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
   
   random(pRNG,phi);
   random(pRNG,chi);
@@ -191,11 +191,11 @@ int main (int argc, char ** argv)
   cDpe = innerProduct(chi_e,dphi_e);
   cDpo = innerProduct(chi_o,dphi_o);
 
-  std::cout <<"e "<<pDce<<" "<<cDpe <<std::endl;
-  std::cout <<"o "<<pDco<<" "<<cDpo <<std::endl;
+  std::cout<<GridLogMessage <<"e "<<pDce<<" "<<cDpe <<std::endl;
+  std::cout<<GridLogMessage <<"o "<<pDco<<" "<<cDpo <<std::endl;
 
-  std::cout <<"pDce - conj(cDpo) "<< pDco-conj(cDpo) <<std::endl;
-  std::cout <<"pDco - conj(cDpe) "<< pDce-conj(cDpe) <<std::endl;
+  std::cout<<GridLogMessage <<"pDce - conj(cDpo) "<< pDco-conj(cDpo) <<std::endl;
+  std::cout<<GridLogMessage <<"pDco - conj(cDpe) "<< pDce-conj(cDpe) <<std::endl;
   
   Grid_finalize();
 }

From d7e6b65a76a2cf4de6c985a63003c4847b9d06b7 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sun, 26 Jul 2015 10:54:38 +0900
Subject: [PATCH 405/429] Elemental force term for Wilson dslash added and
 tests thereof passing. Now need to construct pseudofermion two flavour,
 ratio, one flavour, ratio action fragments.

---
 lib/GridConfig.h.in                       |   3 -
 lib/lattice/Lattice_base.h                |   5 +
 lib/qcd/action/fermion/FermionOperator.h  |  19 ++-
 lib/qcd/action/fermion/WilsonFermion.cc   |  88 +++++++++++-
 lib/qcd/action/fermion/WilsonFermion.h    |  88 +++++++++++-
 lib/qcd/action/fermion/WilsonFermion5D.cc |  95 ++++++++++++-
 lib/qcd/action/fermion/WilsonFermion5D.h  |  18 ++-
 lib/qcd/action/fermion/WilsonKernels.cc   |  44 +++---
 lib/qcd/action/fermion/WilsonKernels.h    |   2 +-
 lib/qcd/hmc/HMC.h                         |   2 -
 lib/simd/Grid_vector_types.h              |   2 +-
 lib/tensors/Tensor_outer.h                |   6 +-
 lib/tensors/Tensor_trace.h                |  10 ++
 tests/Make.inc                            |  10 +-
 tests/Test_dwf_cg_prec.cc                 |   4 +-
 tests/Test_wilson_force.cc                | 102 ++++++++++++++
 tests/Test_wilson_force_phiMdagMphi.cc    | 140 +++++++++++++++++++
 tests/Test_wilson_force_phiMphi.cc        | 162 ++++++++++++++++++++++
 18 files changed, 758 insertions(+), 42 deletions(-)
 create mode 100644 tests/Test_wilson_force.cc
 create mode 100644 tests/Test_wilson_force_phiMdagMphi.cc
 create mode 100644 tests/Test_wilson_force_phiMphi.cc

diff --git a/lib/GridConfig.h.in b/lib/GridConfig.h.in
index 1ec0ecf5..bcb8d811 100644
--- a/lib/GridConfig.h.in
+++ b/lib/GridConfig.h.in
@@ -33,9 +33,6 @@
 /* Support AVX2 (Advanced Vector Extensions 2) instructions */
 #undef HAVE_AVX2
 
-/* define if the compiler supports basic C++11 syntax */
-#undef HAVE_CXX11
-
 /* Define to 1 if you have the declaration of `be64toh', and to 0 if you
    don't. */
 #undef HAVE_DECL_BE64TOH
diff --git a/lib/lattice/Lattice_base.h b/lib/lattice/Lattice_base.h
index 2a365cdd..9c9ef044 100644
--- a/lib/lattice/Lattice_base.h
+++ b/lib/lattice/Lattice_base.h
@@ -60,6 +60,11 @@ public:
     GridBase *_grid;
     int checkerboard;
     std::vector<vobj,alignedAllocator<vobj> > _odata;
+    
+    // to pthread need a computable loop where loop induction is not required
+    int begin(void) { return 0;};
+    int end(void)   { return _odata.size(); }
+    vobj & operator[](int i) { return _odata[i]; };
 
 public:
     typedef typename vobj::scalar_type scalar_type;
diff --git a/lib/qcd/action/fermion/FermionOperator.h b/lib/qcd/action/fermion/FermionOperator.h
index 1aa2f8f0..a8f10380 100644
--- a/lib/qcd/action/fermion/FermionOperator.h
+++ b/lib/qcd/action/fermion/FermionOperator.h
@@ -39,10 +39,23 @@ namespace Grid {
       virtual void Dhop  (const FermionField &in, FermionField &out,int dag)=0;
       virtual void DhopOE(const FermionField &in, FermionField &out,int dag)=0;
       virtual void DhopEO(const FermionField &in, FermionField &out,int dag)=0;
+      virtual void DhopDir(const FermionField &in, FermionField &out,int dir,int disp)=0; // implemented by WilsonFermion and WilsonFermion5D
+
+      // force terms; five routines; default to Dhop on diagonal
+      virtual void MDeriv  (LatticeGaugeField &mat,const FermionField &U,const FermionField &V,int dag){DhopDeriv(mat,U,V,dag);};
+      virtual void MoeDeriv(LatticeGaugeField &mat,const FermionField &U,const FermionField &V,int dag){DhopDerivOE(mat,U,V,dag);};
+      virtual void MeoDeriv(LatticeGaugeField &mat,const FermionField &U,const FermionField &V,int dag){DhopDerivEO(mat,U,V,dag);};
+      virtual void MooDeriv(LatticeGaugeField &mat,const FermionField &U,const FermionField &V,int dag){mat=zero;};
+      virtual void MeeDeriv(LatticeGaugeField &mat,const FermionField &U,const FermionField &V,int dag){mat=zero;};
+
+      virtual void DhopDeriv  (LatticeGaugeField &mat,const FermionField &U,const FermionField &V,int dag)=0;
+      virtual void DhopDerivEO(LatticeGaugeField &mat,const FermionField &U,const FermionField &V,int dag)=0;
+      virtual void DhopDerivOE(LatticeGaugeField &mat,const FermionField &U,const FermionField &V,int dag)=0;
+
+
+      virtual void  Mdiag  (const FermionField &in, FermionField &out) { Mooee(in,out);};   // Same as Mooee applied to both CB's
+      virtual void  Mdir   (const FermionField &in, FermionField &out,int dir,int disp)=0;   // case by case Wilson, Clover, Cayley, ContFrac, PartFrac
 
-      virtual void  Mdiag(const FermionField &in, FermionField &out) { Mooee(in,out);};   // Same as Mooee applied to both CB's
-      virtual void  Mdir (const FermionField &in, FermionField &out,int dir,int disp)=0;   // case by case Wilson, Clover, Cayley, ContFrac, PartFrac
-      virtual void  DhopDir(const FermionField &in, FermionField &out,int dir,int disp)=0; // implemented by WilsonFermion and WilsonFermion5D
 
     };
 
diff --git a/lib/qcd/action/fermion/WilsonFermion.cc b/lib/qcd/action/fermion/WilsonFermion.cc
index f5ff7cb0..afd7edd7 100644
--- a/lib/qcd/action/fermion/WilsonFermion.cc
+++ b/lib/qcd/action/fermion/WilsonFermion.cc
@@ -98,7 +98,9 @@ void WilsonFermion::Mdir (const LatticeFermion &in, LatticeFermion &out,int dir,
   DhopDir(in,out,dir,disp);
 }
 void WilsonFermion::DhopDir(const LatticeFermion &in, LatticeFermion &out,int dir,int disp){
+
   WilsonCompressor compressor(DaggerNo);
+
   Stencil.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
   
   assert( (disp==1)||(disp==-1) );
@@ -109,9 +111,22 @@ void WilsonFermion::DhopDir(const LatticeFermion &in, LatticeFermion &out,int di
 
 PARALLEL_FOR_LOOP
   for(int sss=0;sss<in._grid->oSites();sss++){
-    DiracOptDhopDir(Stencil,Umu,comm_buf,sss,sss,in,out,dirdisp);
+    DiracOptDhopDir(Stencil,Umu,comm_buf,sss,sss,in,out,dirdisp,dirdisp);
   }
+ 
+};
+void WilsonFermion::DhopDirDisp(const LatticeFermion &in, LatticeFermion &out,int dirdisp,int gamma,int dag)
+{
 
+  WilsonCompressor compressor(dag);
+
+  Stencil.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
+
+PARALLEL_FOR_LOOP
+  for(int sss=0;sss<in._grid->oSites();sss++){
+    DiracOptDhopDir(Stencil,Umu,comm_buf,sss,sss,in,out,dirdisp,gamma);
+  }
+ 
 };
 
 void WilsonFermion::DhopInternal(CartesianStencil & st,LatticeDoubledGaugeField & U,
@@ -177,6 +192,77 @@ void WilsonFermion::Dhop(const LatticeFermion &in, LatticeFermion &out,int dag)
   DhopInternal(Stencil,Umu,in,out,dag);
 }
 
+void WilsonFermion::DerivInternal(CartesianStencil & st,LatticeDoubledGaugeField & U,
+				  LatticeGaugeField &mat,const LatticeFermion &A,const LatticeFermion &B,int dag)
+{
+  assert((dag==DaggerNo) ||(dag==DaggerYes));
+
+  WilsonCompressor compressor(dag);
+
+  LatticeColourMatrix tmp(B._grid);
+  LatticeFermion Btilde(B._grid);
+
+  st.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(B,comm_buf,compressor);
+  
+  for(int mu=0;mu<Nd;mu++){
+      
+    ////////////////////////////////////////////////////////////////////////
+    // Flip gamma (1+g)<->(1-g) if dag
+    ////////////////////////////////////////////////////////////////////////
+    int gamma = mu;
+    if ( dag ) gamma+= Nd;
+
+    ////////////////////////
+    // Call the single hop
+    ////////////////////////
+PARALLEL_FOR_LOOP
+    for(int sss=0;sss<B._grid->oSites();sss++){
+      DiracOptDhopDir(st,U,comm_buf,sss,sss,B,Btilde,mu,gamma);
+    }
+
+    //////////////////////////////////////////////////
+    // spin trace outer product
+    //////////////////////////////////////////////////
+    tmp = TraceIndex<SpinIndex>(outerProduct(Btilde,A)); 
+    PokeIndex<LorentzIndex>(mat,tmp,mu);
+
+  }
+}
+void WilsonFermion::DhopDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+{
+  conformable(U._grid,_grid);  
+  conformable(U._grid,V._grid);
+  conformable(U._grid,mat._grid);
+
+  mat.checkerboard = U.checkerboard;
+
+  DerivInternal(Stencil,Umu,mat,U,V,dag);
+}
+void WilsonFermion::DhopDerivOE(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+{
+  conformable(U._grid,_cbgrid);  
+  conformable(U._grid,V._grid);
+  conformable(U._grid,mat._grid);
+
+  assert(V.checkerboard==Even);
+  assert(U.checkerboard==Odd);
+  mat.checkerboard = Odd;
+
+  DerivInternal(StencilEven,UmuOdd,mat,U,V,dag);
+}
+void WilsonFermion::DhopDerivEO(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+{
+  conformable(U._grid,_cbgrid);  
+  conformable(U._grid,V._grid);
+  conformable(U._grid,mat._grid);
+
+  assert(V.checkerboard==Odd);
+  assert(U.checkerboard==Even);
+  mat.checkerboard = Even;
+
+  DerivInternal(StencilOdd,UmuEven,mat,U,V,dag);
+}
+
 }}
 
 
diff --git a/lib/qcd/action/fermion/WilsonFermion.h b/lib/qcd/action/fermion/WilsonFermion.h
index 60726fdd..5422ca52 100644
--- a/lib/qcd/action/fermion/WilsonFermion.h
+++ b/lib/qcd/action/fermion/WilsonFermion.h
@@ -24,11 +24,95 @@ namespace Grid {
       // half checkerboard operaions
       void   Meooe       (const LatticeFermion &in, LatticeFermion &out);
       void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out);
+
       virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out); // remain virtual so we 
       virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out); // can derive Clover
       virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out); // from Wilson base
       virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
 
+      ////////////////////////
+      // 
+      // Force term: d/dtau S = 0
+      //
+      // It is simplest to consider the two flavour force term 
+      // 
+      //       S[U,phi] = phidag (MdagM)^-1 phi
+      //
+      // But simplify even this to
+      //
+      //       S[U,phi] = phidag MdagM phi
+      //
+      // (other options exist depending on nature of action fragment.)
+      // 
+      // Require momentum be traceless anti-hermitian to move within group manifold [ P = i P^a T^a ]
+      //
+      // Define the HMC hamiltonian
+      //
+      //       H = 1/2 Tr P^2 + S(U,phi)
+      // 
+      // .
+      // U =  P U    (lorentz & color indices multiplied)
+      //
+      // Hence
+      //
+      // .c    c  c       c
+      // U =  U  P   = - U  P        (c == dagger)
+      //
+      // So, taking some liberty with implicit indices
+      //                  .      .          .c      c
+      // dH/dt = 0 = Tr P P +Tr[ U  dS/dU + U  dS/dU  ]
+      //
+      //                .                          c      c
+      //           = Tr P P + i Tr[  P U dS/dU  - U   P dS/dU  ]
+      //
+      //                   .                        c  c
+      //           = Tr P (P + i ( U dS/dU - P dS/dU  U ]
+      //
+      //              .                        c  c
+      //           => P  = -i [ U dS/dU - dS/dU  U ]      generates HMC EoM
+      //
+      // Simple case work this out using S = phi^dag MdagM phi for wilson:
+      //                               c       c
+      // dSdt     = dU_xdt  dSdUx  + dUxdt dSdUx
+      //          
+      //         = Tr i P U_x [ (\phi^\dag)_x (1+g) (M \phi)_x+\mu   +(\phi^\dag M^\dag)_x (1-g) \phi_{x+\mu} ]
+      //                 c
+      //            - i U_x P [ (\phi^\dag)_x+mu (1-g) (M \phi)_x    +(\phi^\dag M^\dag)_(x+\mu) (1+g) \phi_{x} ]
+      //
+      //         = i [(\phi^\dag)_x      ]_j P_jk  [U_x(1+g) (M \phi)_x+\mu]_k        (1)
+      //         + i [(\phi^\dagM^\dag)_x]_j P_jk  [U_x(1-g) (\phi)_x+\mu]_k          (2)
+      //         - i [(\phi^\dag)_x+mu (1-g) U^dag_x]_j P_jk  [(M \phi)_xk            (3)
+      //         - i [(\phi^\dagM^\dag)_x+mu (1+g) U^dag_x]_j P_jk  [ \phi]_xk        (4)
+      //
+      // Observe that (1)* = (4)
+      //              (2)* = (3)
+      //
+      // Write as    .
+      //             P_{kj}  = - i (  [U_x(1+g) (M \phi)_x+\mu] (x) [(\phi^\dag)_x] + [U_x(1-g) (\phi)_x+\mu] (x) [(\phi^\dagM^\dag)_x] - h.c )
+      //
+      // where (x) denotes outer product in colour and spins are traced.
+      //
+      // Need only evaluate (1) and (2)  [Chroma] or (2) and (4) [IroIro] and take the 
+      // traceless anti hermitian part (of term in brackets w/o the "i")
+      // 
+      // Generalisation to S=phi^dag (MdagM)^{-1} phi is simple:
+      //
+      // For more complicated DWF etc... apply product rule in differentiation
+      //
+      ////////////////////////
+      void DhopDeriv  (LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+      void DhopDerivEO(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+      void DhopDerivOE(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+
+      // Extra support internal
+      void DerivInternal(CartesianStencil & st,
+			 LatticeDoubledGaugeField & U,
+			 LatticeGaugeField &mat,
+			 const LatticeFermion &A,
+			 const LatticeFermion &B,
+			 int dag);
+
+
       // non-hermitian hopping term; half cb or both
       void Dhop  (const LatticeFermion &in, LatticeFermion &out,int dag);
       void DhopOE(const LatticeFermion &in, LatticeFermion &out,int dag);
@@ -37,6 +121,7 @@ namespace Grid {
       // Multigrid assistance
       void   Mdir (const LatticeFermion &in, LatticeFermion &out,int dir,int disp);
       void DhopDir(const LatticeFermion &in, LatticeFermion &out,int dir,int disp);
+      void DhopDirDisp(const LatticeFermion &in, LatticeFermion &out,int dirdisp,int gamma,int dag);
 
       ///////////////////////////////////////////////////////////////
       // Extra methods added by derived
@@ -59,7 +144,8 @@ namespace Grid {
       static int HandOptDslash; // these are a temporary hack
       static int MortonOrder;
 
-    protected:
+      //    protected:
+    public:
 
       RealD                        mass;
 
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.cc b/lib/qcd/action/fermion/WilsonFermion5D.cc
index 43829ea7..523934c4 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.cc
+++ b/lib/qcd/action/fermion/WilsonFermion5D.cc
@@ -105,14 +105,105 @@ PARALLEL_FOR_LOOP
     for(int s=0;s<Ls;s++){
       int sU=ss;
       int sF = s+Ls*sU; 
-      DiracOptDhopDir(Stencil,Umu,comm_buf,sF,sU,in,out,dirdisp);
+      DiracOptDhopDir(Stencil,Umu,comm_buf,sF,sU,in,out,dirdisp,dirdisp);
     }
   }
 };
 
+void WilsonFermion5D::DerivInternal(CartesianStencil & st,
+				    LatticeDoubledGaugeField & U,
+				    LatticeGaugeField &mat,
+				    const LatticeFermion &A,
+				    const LatticeFermion &B,
+				    int dag)
+{
+  assert((dag==DaggerNo) ||(dag==DaggerYes));
+
+  WilsonCompressor compressor(dag);
+  
+  LatticeColourMatrix tmp(B._grid);
+  LatticeFermion Btilde(B._grid);
+
+  st.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(B,comm_buf,compressor);
+  
+  for(int mu=0;mu<Nd;mu++){
+      
+    ////////////////////////////////////////////////////////////////////////
+    // Flip gamma if dag
+    ////////////////////////////////////////////////////////////////////////
+    int gamma = mu;
+    if ( dag ) gamma+= Nd;
+
+    ////////////////////////
+    // Call the single hop
+    ////////////////////////
+PARALLEL_FOR_LOOP
+    for(int sss=0;sss<B._grid->oSites();sss++){
+      for(int s=0;s<Ls;s++){
+	int sU=sss;
+	int sF = s+Ls*sU;
+	DiracOptDhopDir(st,U,comm_buf,sF,sU,B,Btilde,mu,gamma);
+      }
+    }
+
+    ////////////////////////////
+    // spin trace outer product
+    ////////////////////////////
+    tmp = TraceIndex<SpinIndex>(outerProduct(Btilde,A)); // ordering here
+    PokeIndex<LorentzIndex>(mat,tmp,mu);
+
+  }
+}
+
+void WilsonFermion5D::DhopDeriv(      LatticeGaugeField &mat,
+				const LatticeFermion &A,
+				const LatticeFermion &B,
+				int dag)
+{
+  conformable(A._grid,FermionGrid());  
+  conformable(A._grid,B._grid);
+  conformable(GaugeGrid(),mat._grid);
+
+  mat.checkerboard = A.checkerboard;
+
+  DerivInternal(Stencil,Umu,mat,A,B,dag);
+}
+
+void WilsonFermion5D::DhopDerivEO(LatticeGaugeField &mat,
+				  const LatticeFermion &A,
+				  const LatticeFermion &B,
+				  int dag)
+{
+  conformable(A._grid,FermionRedBlackGrid());
+  conformable(GaugeRedBlackGrid(),mat._grid);
+  conformable(A._grid,B._grid);
+
+  assert(B.checkerboard==Odd);
+  assert(A.checkerboard==Even);
+  mat.checkerboard = Even;
+
+  DerivInternal(StencilOdd,UmuEven,mat,A,B,dag);
+}
+
+void WilsonFermion5D::DhopDerivOE(LatticeGaugeField &mat,
+				  const LatticeFermion &A,
+				  const LatticeFermion &B,
+				  int dag)
+{
+  conformable(A._grid,FermionRedBlackGrid());
+  conformable(GaugeRedBlackGrid(),mat._grid);
+  conformable(A._grid,B._grid);
+
+  assert(B.checkerboard==Even);
+  assert(A.checkerboard==Odd);
+  mat.checkerboard = Odd;
+
+  DerivInternal(StencilEven,UmuOdd,mat,A,B,dag);
+}
+
 void WilsonFermion5D::DhopInternal(CartesianStencil & st, LebesgueOrder &lo,
 				   LatticeDoubledGaugeField & U,
-			   const LatticeFermion &in, LatticeFermion &out,int dag)
+				   const LatticeFermion &in, LatticeFermion &out,int dag)
 {
   //  assert((dag==DaggerNo) ||(dag==DaggerYes));
 
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.h b/lib/qcd/action/fermion/WilsonFermion5D.h
index e345d2a4..7ff5d1c3 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.h
+++ b/lib/qcd/action/fermion/WilsonFermion5D.h
@@ -44,12 +44,18 @@ namespace Grid {
 
       // half checkerboard operations; leave unimplemented as abstract for now
       virtual void   Meooe       (const LatticeFermion &in, LatticeFermion &out){assert(0);};
-      virtual void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out){assert(0);};
       virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out){assert(0);};
-      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out){assert(0);};
       virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out){assert(0);};
+
+      virtual void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out){assert(0);};
+      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out){assert(0);};
       virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out){assert(0);};
 
+      // These can be overridden by fancy 5d chiral actions
+      virtual void DhopDeriv  (LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+      virtual void DhopDerivEO(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+      virtual void DhopDerivOE(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+
       // Implement hopping term non-hermitian hopping term; half cb or both
       // Implement s-diagonal DW
       void DW    (const LatticeFermion &in, LatticeFermion &out,int dag);
@@ -64,6 +70,14 @@ namespace Grid {
       ///////////////////////////////////////////////////////////////
       // New methods added 
       ///////////////////////////////////////////////////////////////
+
+      void DerivInternal(CartesianStencil & st,
+			 LatticeDoubledGaugeField & U,
+			 LatticeGaugeField &mat,
+			 const LatticeFermion &A,
+			 const LatticeFermion &B,
+			 int dag);
+
       void DhopInternal(CartesianStencil & st,
 			LebesgueOrder &lo,
 			LatticeDoubledGaugeField &U,
diff --git a/lib/qcd/action/fermion/WilsonKernels.cc b/lib/qcd/action/fermion/WilsonKernels.cc
index bdc18eee..43ccff63 100644
--- a/lib/qcd/action/fermion/WilsonKernels.cc
+++ b/lib/qcd/action/fermion/WilsonKernels.cc
@@ -294,8 +294,8 @@ void DiracOptDhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
 }
 
 void DiracOptDhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
-			std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-		       int sF,int sU,const LatticeFermion &in, LatticeFermion &out,int dirdisp)
+		     std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
+		     int sF,int sU,const LatticeFermion &in, LatticeFermion &out,int dir,int gamma)
 {
     vHalfSpinColourVector  tmp;    
     vHalfSpinColourVector  chi;    
@@ -304,13 +304,13 @@ void DiracOptDhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
     int offset,local,perm, ptype;
     int ss=sF;
     
-    offset = st._offsets [dirdisp][ss];
-    local  = st._is_local[dirdisp][ss];
-    perm   = st._permute[dirdisp][ss];
-    ptype  = st._permute_type[dirdisp];
+    offset = st._offsets [dir][ss];
+    local  = st._is_local[dir][ss];
+    perm   = st._permute[dir][ss];
+    ptype  = st._permute_type[dir];
 
     // Xp
-    if(dirdisp==Xp){
+    if(gamma==Xp){
       if ( local && perm ) {
 	spProjXp(tmp,in._odata[offset]);
 	permute(chi,tmp,ptype);
@@ -319,12 +319,12 @@ void DiracOptDhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
       } else { 
 	chi=buf[offset];
       }
-      mult(&Uchi(),&U._odata[sU](Xp),&chi());
+      mult(&Uchi(),&U._odata[sU](dir),&chi());
       spReconXp(result,Uchi);
     }
 
     // Yp
-    if ( dirdisp==Yp ){
+    if ( gamma==Yp ){
       if ( local && perm ) {
 	spProjYp(tmp,in._odata[offset]);
 	permute(chi,tmp,ptype);
@@ -333,12 +333,12 @@ void DiracOptDhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
       } else { 
 	chi=buf[offset];
       }
-      mult(&Uchi(),&U._odata[sU](Yp),&chi());
+      mult(&Uchi(),&U._odata[sU](dir),&chi());
       spReconYp(result,Uchi);
     }
 
     // Zp
-    if ( dirdisp ==Zp ){
+    if ( gamma ==Zp ){
       if ( local && perm ) {
 	spProjZp(tmp,in._odata[offset]);
 	permute(chi,tmp,ptype);
@@ -347,12 +347,12 @@ void DiracOptDhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
       } else { 
 	chi=buf[offset];
       }
-      mult(&Uchi(),&U._odata[sU](Zp),&chi());
+      mult(&Uchi(),&U._odata[sU](dir),&chi());
       spReconZp(result,Uchi);
     }
 
     // Tp
-    if ( dirdisp ==Tp ){
+    if ( gamma ==Tp ){
       if ( local && perm ) {
 	spProjTp(tmp,in._odata[offset]);
 	permute(chi,tmp,ptype);
@@ -361,12 +361,12 @@ void DiracOptDhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
       } else { 
 	chi=buf[offset];
       }
-      mult(&Uchi(),&U._odata[sU](Tp),&chi());
+      mult(&Uchi(),&U._odata[sU](dir),&chi());
       spReconTp(result,Uchi);
     }
 
     // Xm
-    if ( dirdisp==Xm ){
+    if ( gamma==Xm ){
       if ( local && perm ) {
 	spProjXm(tmp,in._odata[offset]);
 	permute(chi,tmp,ptype);
@@ -375,12 +375,12 @@ void DiracOptDhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
       } else { 
 	chi=buf[offset];
       }
-      mult(&Uchi(),&U._odata[sU](Xm),&chi());
+      mult(&Uchi(),&U._odata[sU](dir),&chi());
       spReconXm(result,Uchi);
     }
 
     // Ym
-    if ( dirdisp == Ym ){
+    if ( gamma == Ym ){
       if ( local && perm ) {
 	spProjYm(tmp,in._odata[offset]);
 	permute(chi,tmp,ptype);
@@ -389,12 +389,12 @@ void DiracOptDhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
       } else { 
 	chi=buf[offset];
       }
-      mult(&Uchi(),&U._odata[sU](Ym),&chi());
+      mult(&Uchi(),&U._odata[sU](dir),&chi());
       spReconYm(result,Uchi);
     }
 
     // Zm
-    if ( dirdisp == Zm ){
+    if ( gamma == Zm ){
       if ( local && perm ) {
 	spProjZm(tmp,in._odata[offset]);
 	permute(chi,tmp,ptype);
@@ -403,12 +403,12 @@ void DiracOptDhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
       } else { 
 	chi=buf[offset];
       }
-      mult(&Uchi(),&U._odata[sU](Zm),&chi());
+      mult(&Uchi(),&U._odata[sU](dir),&chi());
       spReconZm(result,Uchi);
     }
 
     // Tm
-    if ( dirdisp==Tm ) {
+    if ( gamma==Tm ) {
       if ( local && perm ) {
 	spProjTm(tmp,in._odata[offset]);
 	permute(chi,tmp,ptype);
@@ -417,7 +417,7 @@ void DiracOptDhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
       } else { 
 	chi=buf[offset];
       }
-      mult(&Uchi(),&U._odata[sU](Tm),&chi());
+      mult(&Uchi(),&U._odata[sU](dir),&chi());
       spReconTm(result,Uchi);
     }
 
diff --git a/lib/qcd/action/fermion/WilsonKernels.h b/lib/qcd/action/fermion/WilsonKernels.h
index ec807af5..cb0e07bb 100644
--- a/lib/qcd/action/fermion/WilsonKernels.h
+++ b/lib/qcd/action/fermion/WilsonKernels.h
@@ -22,7 +22,7 @@ namespace Grid {
 			       int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
       void DiracOptDhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
 			   std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			   int sF,int sU,const LatticeFermion &in, LatticeFermion &out,int dirdisp);
+			   int sF,int sU,const LatticeFermion &in, LatticeFermion &out,int dirdisp,int gamma);
       
       //  };
 
diff --git a/lib/qcd/hmc/HMC.h b/lib/qcd/hmc/HMC.h
index e3fe57ed..21c6bb0d 100644
--- a/lib/qcd/hmc/HMC.h
+++ b/lib/qcd/hmc/HMC.h
@@ -68,14 +68,12 @@ namespace Grid{
 		     Integrator<Algorithm>& MolDyn):
       Params(Pms),MD(MolDyn){
 	//FIXME
-
 	// initialize RNGs also with seed
 	sRNG.SeedRandomDevice();
       }
       ~HybridMonteCarlo(){};
 
 
-
       void evolve(LatticeLorentzColourMatrix& Uin){
 	Real DeltaH;
 	
diff --git a/lib/simd/Grid_vector_types.h b/lib/simd/Grid_vector_types.h
index 034d5314..71472407 100644
--- a/lib/simd/Grid_vector_types.h
+++ b/lib/simd/Grid_vector_types.h
@@ -414,7 +414,7 @@ namespace Grid {
   template<class S, class V >
   inline Grid_simd< S, V> outerProduct(const Grid_simd< S, V> &l, const Grid_simd< S, V> & r)
   {
-    return l*r;
+    return l*conjugate(r);
   }
 
   template<class S, class V >
diff --git a/lib/tensors/Tensor_outer.h b/lib/tensors/Tensor_outer.h
index 816f6ad8..d0150709 100644
--- a/lib/tensors/Tensor_outer.h
+++ b/lib/tensors/Tensor_outer.h
@@ -28,11 +28,13 @@ auto outerProduct (const iScalar<l>& lhs,const iScalar<r>& rhs) -> iScalar<declt
 
 inline ComplexF outerProduct(const ComplexF &l, const ComplexF& r)
 {
-  return l*r;
+  std::cout << "outer product taking conj "<<r<<" "<<conj(r)<<std::endl;
+  return l*conj(r);
 }
 inline ComplexD outerProduct(const ComplexD &l, const ComplexD& r)
 {
-  return l*r;
+  std::cout << "outer product taking conj "<<r<<" "<<conj(r)<<std::endl;
+  return l*conj(r);
 }
 inline RealF outerProduct(const RealF &l, const RealF& r)
 {
diff --git a/lib/tensors/Tensor_trace.h b/lib/tensors/Tensor_trace.h
index 4cf5edc2..bd75bb91 100644
--- a/lib/tensors/Tensor_trace.h
+++ b/lib/tensors/Tensor_trace.h
@@ -31,6 +31,16 @@ inline auto trace(const iScalar<vtype> &arg) -> iScalar<decltype(trace(arg._inte
     return ret;
 }
 
+template<class vtype,int N>
+  inline auto trace(const iVector<vtype,N> &arg) -> iVector<decltype(trace(arg._internal[0])),N>
+{
+  iVector<decltype(trace(arg._internal[0])),N> ret;
+  for(int i=0;i<N;i++){
+    ret._internal[i]=trace(arg._internal[i]);
+  }
+  return ret;
+}
+
 
 }
 #endif
diff --git a/tests/Make.inc b/tests/Make.inc
index f04da4eb..b78b4d16 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,5 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd
+bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd Test_wilson_force Test_wilson_force_phiMphi
 
 
 Test_GaugeAction_SOURCES=Test_GaugeAction.cc
@@ -141,3 +141,11 @@ Test_wilson_cr_unprec_LDADD=-lGrid
 Test_wilson_even_odd_SOURCES=Test_wilson_even_odd.cc
 Test_wilson_even_odd_LDADD=-lGrid
 
+
+Test_wilson_force_SOURCES=Test_wilson_force.cc
+Test_wilson_force_LDADD=-lGrid
+
+
+Test_wilson_force_phiMphi_SOURCES=Test_wilson_force_phiMphi.cc
+Test_wilson_force_phiMphi_LDADD=-lGrid
+
diff --git a/tests/Test_dwf_cg_prec.cc b/tests/Test_dwf_cg_prec.cc
index d429ee92..633d280f 100644
--- a/tests/Test_dwf_cg_prec.cc
+++ b/tests/Test_dwf_cg_prec.cc
@@ -34,7 +34,9 @@ int main (int argc, char ** argv)
 
   LatticeFermion    src(FGrid); random(RNG5,src);
   LatticeFermion result(FGrid); result=zero;
-  LatticeGaugeField Umu(UGrid); random(RNG4,Umu);
+  LatticeGaugeField Umu(UGrid); 
+
+  SU3::HotConfiguration(RNG4,Umu);
 
   std::vector<LatticeColourMatrix> U(4,UGrid);
   for(int mu=0;mu<Nd;mu++){
diff --git a/tests/Test_wilson_force.cc b/tests/Test_wilson_force.cc
new file mode 100644
index 00000000..6a8d7611
--- /dev/null
+++ b/tests/Test_wilson_force.cc
@@ -0,0 +1,102 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+#define parallel_for PARALLEL_FOR_LOOP for
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+
+  GridCartesian               Grid(latt_size,simd_layout,mpi_layout);
+  GridRedBlackCartesian     RBGrid(latt_size,simd_layout,mpi_layout);
+
+  int threads = GridThread::GetThreads();
+  std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+
+  std::vector<int> seeds({1,2,3,4});
+
+  GridParallelRNG          pRNG(&Grid);
+  pRNG.SeedRandomDevice();
+
+  LatticeFermion phi        (&Grid); gaussian(pRNG,phi);
+  LatticeFermion Mphi       (&Grid); 
+  LatticeFermion MphiPrime  (&Grid); 
+
+  LatticeGaugeField U(&Grid);
+
+  SU3::HotConfiguration(pRNG,U);
+  //  SU3::ColdConfiguration(pRNG,U);
+  
+  ////////////////////////////////////
+  // Unmodified matrix element
+  ////////////////////////////////////
+  RealD mass=-4.0; //kills the diagonal term
+  WilsonFermion Dw     (U,     Grid,RBGrid,mass);
+  Dw.M   (phi,Mphi);
+
+  ComplexD S    = innerProduct(Mphi,Mphi); // pdag MdagM p
+
+  // get the deriv of phidag MdagM phi with respect to "U"
+  LatticeGaugeField UdSdU(&Grid);
+  LatticeGaugeField tmp(&Grid);
+
+  Dw.MDeriv(tmp , Mphi,  phi,DaggerNo );  UdSdU=tmp;
+  Dw.MDeriv(tmp , phi,  Mphi,DaggerYes ); UdSdU=UdSdU+tmp;
+
+
+  LatticeFermion Ftmp      (&Grid);
+
+  ////////////////////////////////////
+  // Modify the gauge field a little 
+  ////////////////////////////////////
+  RealD dt = 1.0e-6;
+
+  LatticeColourMatrix mommu(&Grid); 
+  LatticeGaugeField mom(&Grid); 
+  LatticeGaugeField Uprime(&Grid); 
+
+  for(int mu=0;mu<Nd;mu++){
+
+    SU3::GaussianLieAlgebraMatrix(pRNG, mommu); // Traceless antihermitian momentum; gaussian in lie alg
+
+    PokeIndex<LorentzIndex>(mom,mommu,mu);
+    parallel_for(auto i=mom.begin();i<mom.end();i++){
+      Uprime[i](mu) =U[i](mu)+ mom[i](mu)*U[i](mu)*dt;
+    }
+
+  }
+
+  Dw.DoubleStore(Dw.Umu,Uprime);
+  Dw.M          (phi,MphiPrime);
+
+  ComplexD Sprime    = innerProduct(MphiPrime   ,MphiPrime);
+
+  //////////////////////////////////////////////
+  // Use derivative to estimate dS
+  //////////////////////////////////////////////
+  LatticeComplex dS(&Grid); dS = zero;
+
+  parallel_for(auto i=mom.begin();i<mom.end();i++){
+    for(int mu=0;mu<Nd;mu++){
+      //      dS[i]() = dS[i]()+trace(mom[i](mu) * UdSdU[i](mu) - mom[i](mu)* adj( UdSdU[i](mu)) )*dt;
+      dS[i]() =    dS[i]()+trace(mom[i](mu) * (UdSdU[i](mu)))*dt;
+      dS[i]() =    dS[i]()-trace(mom[i](mu) * adj(UdSdU[i](mu)))*dt;
+    }
+  }
+  Complex dSpred    = sum(dS);
+
+  std::cout << GridLogMessage << " S      "<<S<<std::endl;
+  std::cout << GridLogMessage << " Sprime "<<Sprime<<std::endl;
+  std::cout << GridLogMessage << "dS      "<<Sprime-S<<std::endl;
+  std::cout << GridLogMessage << "predict dS    "<< dSpred <<std::endl;
+
+  std::cout<< GridLogMessage << "Done" <<std::endl;
+  Grid_finalize();
+}
diff --git a/tests/Test_wilson_force_phiMdagMphi.cc b/tests/Test_wilson_force_phiMdagMphi.cc
new file mode 100644
index 00000000..33c5f6e1
--- /dev/null
+++ b/tests/Test_wilson_force_phiMdagMphi.cc
@@ -0,0 +1,140 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+#define parallel_for PARALLEL_FOR_LOOP for
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+
+  GridCartesian               Grid(latt_size,simd_layout,mpi_layout);
+  GridRedBlackCartesian     RBGrid(latt_size,simd_layout,mpi_layout);
+
+  int threads = GridThread::GetThreads();
+  std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+
+  std::vector<int> seeds({1,2,3,4});
+
+  GridParallelRNG          pRNG(&Grid);
+  pRNG.SeedRandomDevice();
+
+  LatticeFermion phi        (&Grid); gaussian(pRNG,phi);
+  LatticeFermion Mphi       (&Grid); 
+  LatticeFermion Mdagphi       (&Grid); 
+  LatticeFermion MphiPrime  (&Grid); 
+  LatticeFermion MdagphiPrime  (&Grid); 
+  LatticeFermion dMphi      (&Grid); 
+
+  LatticeGaugeField U(&Grid);
+
+
+  SU3::HotConfiguration(pRNG,U);
+  //  SU3::ColdConfiguration(pRNG,U);
+  
+  ////////////////////////////////////
+  // Unmodified matrix element
+  ////////////////////////////////////
+  RealD mass=-4.0; //kills the diagonal term
+  WilsonFermion Dw     (U,     Grid,RBGrid,mass);
+  Dw.M   (phi,Mphi);
+  Dw.Mdag(phi,Mdagphi);
+
+  ComplexD S    = innerProduct(Mphi,Mphi); // pdag MdagM p
+  ComplexD Sdag = innerProduct(Mdagphi,Mdagphi); // pdag MMdag p
+
+  // get the deriv of phidag MdagM phi with respect to "U"
+  LatticeGaugeField UdSdU(&Grid);
+  LatticeGaugeField UdSdUdag(&Grid);
+  LatticeGaugeField tmp(&Grid);
+
+  Dw.MDeriv(tmp , Mphi,  phi,DaggerNo );  UdSdU=tmp;
+
+  Dw.MDeriv(tmp , Mdagphi,  phi,DaggerYes );  UdSdUdag=tmp;
+
+
+  LatticeFermion dMdagphi      (&Grid);  dMdagphi=zero;
+  LatticeFermion Ftmp      (&Grid);
+
+
+  //  Dw.MDeriv(UdSdU,Mdagphi,  phi,DaggerYes );// UdSdU =UdSdU +tmp;
+
+  ////////////////////////////////////
+  // Modify the gauge field a little in one dir
+  ////////////////////////////////////
+  RealD dt = 1.0e-3;
+
+  LatticeColourMatrix mommu(&Grid); 
+  LatticeGaugeField mom(&Grid); 
+  LatticeGaugeField Uprime(&Grid); 
+
+  for(int mu=0;mu<Nd;mu++){
+
+    SU3::GaussianLieAlgebraMatrix(pRNG, mommu); // Traceless antihermitian momentum; gaussian in lie alg
+
+    //    Dw.DoubleStore(Dw.Umu,Uprime); // update U _and_ Udag
+    Dw.DhopDirDisp(phi,Ftmp,mu,mu+4,DaggerYes); 
+    dMdagphi=dMdagphi+mommu*Ftmp*dt;
+    
+    PokeIndex<LorentzIndex>(mom,mommu,mu);
+
+    parallel_for(auto i=mom.begin();i<mom.end();i++){
+      Uprime[i](mu) =U[i](mu)+ mom[i](mu)*U[i](mu)*dt;
+      Dw.Umu[i](mu) =Uprime[i](mu); // update U but _not_ Udag
+    }
+
+  }
+
+  Dw.Mdag(phi,MdagphiPrime);
+  Dw.M   (phi,MphiPrime);
+
+  std::cout << GridLogMessage << "deltaMdag phi    "<< norm2(dMdagphi) <<std::endl;
+  Ftmp=MdagphiPrime - Mdagphi;
+  std::cout << GridLogMessage << "diff Mdag phi    "<< norm2(Ftmp) <<std::endl;
+  Ftmp = Ftmp - dMdagphi;
+  std::cout << GridLogMessage << "err  Mdag phi    "<< norm2(Ftmp) <<std::endl;
+  std::cout << dMdagphi<<std::endl;
+  Ftmp=MdagphiPrime - Mdagphi;
+  std::cout << Ftmp<<std::endl;
+
+
+  ComplexD Sprime    = innerProduct(Mphi   ,MphiPrime);
+  ComplexD Sprimedag = innerProduct(Mdagphi,MdagphiPrime);
+
+  ComplexD deltaSdag = innerProduct(Mdagphi,dMdagphi);
+  std::cout << GridLogMessage << "deltaSdag from inner prod of mom* M[u]     "<<deltaSdag<<std::endl;
+
+  //////////////////////////////////////////////
+  // Use derivative to estimate dS
+  //////////////////////////////////////////////
+  LatticeComplex dS(&Grid); dS = zero;
+  LatticeComplex dSdag(&Grid); dSdag = zero;
+  parallel_for(auto i=mom.begin();i<mom.end();i++){
+    for(int mu=0;mu<Nd;mu++){
+      //      dS[i]() = dS[i]()+trace(mom[i](mu) * UdSdU[i](mu) - mom[i](mu)* adj( UdSdU[i](mu)) )*dt;
+      dS[i]()    =    dS[i]()+trace(mom[i](mu) * UdSdU[i](mu) )*dt;
+      dSdag[i]() = dSdag[i]()+trace(mom[i](mu) * UdSdUdag[i](mu) )*dt;
+    }
+  }
+  Complex dSpred    = sum(dS);
+  Complex dSdagpred = sum(dSdag);
+
+  std::cout << GridLogMessage << " S      "<<S<<std::endl;
+  std::cout << GridLogMessage << " Sprime "<<Sprime<<std::endl;
+  std::cout << GridLogMessage << "dS      "<<Sprime-S<<std::endl;
+  std::cout << GridLogMessage << "predict dS    "<< dSpred <<std::endl;
+  std::cout << "\n\n"<<std::endl;
+  std::cout << GridLogMessage << " Sdag      "<<Sdag<<std::endl;
+  std::cout << GridLogMessage << " Sprimedag "<<Sprimedag<<std::endl;
+  std::cout << GridLogMessage << "dSdag      "<<Sprimedag-Sdag<<std::endl;
+  std::cout << GridLogMessage << "predict dSdag    "<< dSdagpred <<std::endl;
+
+  std::cout<< GridLogMessage << "Done" <<std::endl;
+  Grid_finalize();
+}
diff --git a/tests/Test_wilson_force_phiMphi.cc b/tests/Test_wilson_force_phiMphi.cc
new file mode 100644
index 00000000..f1e42195
--- /dev/null
+++ b/tests/Test_wilson_force_phiMphi.cc
@@ -0,0 +1,162 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+#define parallel_for PARALLEL_FOR_LOOP for
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+
+  GridCartesian               Grid(latt_size,simd_layout,mpi_layout);
+  GridRedBlackCartesian     RBGrid(latt_size,simd_layout,mpi_layout);
+
+  int threads = GridThread::GetThreads();
+  std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+
+  std::vector<int> seeds({1,2,3,4});
+
+  GridParallelRNG          pRNG(&Grid);
+  pRNG.SeedRandomDevice();
+
+  LatticeFermion phi        (&Grid); gaussian(pRNG,phi);
+  LatticeFermion Mphi       (&Grid); 
+  LatticeFermion MphiPrime  (&Grid); 
+  LatticeFermion dMphi      (&Grid); 
+
+  LatticeGaugeField U(&Grid);
+
+
+  SU3::HotConfiguration(pRNG,U);
+
+  ////////////////////////////////////
+  // Unmodified matrix element
+  ////////////////////////////////////
+  RealD mass=-4.0; //kills the diagonal term
+  WilsonFermion Dw     (U,     Grid,RBGrid,mass);
+  Dw.M(phi,Mphi);
+
+  ComplexD S = innerProduct(phi,Mphi);
+
+  // get the deriv
+  LatticeGaugeField UdSdU(&Grid);
+  Dw.MDeriv(UdSdU,phi, phi,DaggerNo ); 
+
+
+  ////////////////////////////////////
+  // Modify the gauge field a little in one dir
+  ////////////////////////////////////
+  RealD dt = 1.0e-3;
+  Complex Complex_i(0,1);
+
+  LatticeColourMatrix Umu(&Grid);
+  LatticeColourMatrix Umu_save(&Grid);
+  LatticeColourMatrix dU (&Grid);
+  LatticeColourMatrix mom(&Grid); 
+  SU3::GaussianLieAlgebraMatrix(pRNG, mom); // Traceless antihermitian momentum; gaussian in lie alg
+
+
+  // check mom is as i expect
+  LatticeColourMatrix tmpmom(&Grid); 
+  tmpmom = mom+adj(mom);
+  std::cout << GridLogMessage << "mom anti-herm check "<< norm2(tmpmom)<<std::endl;
+  std::cout << GridLogMessage << "mom tr check "<< norm2(trace(mom))<<std::endl;
+  
+  const int mu=0;
+  Umu = PeekIndex<LorentzIndex>(U,mu);
+  Umu_save=Umu;
+  dU = mom * Umu * dt;
+  Umu= Umu+dU;
+  PokeIndex<LorentzIndex>(Dw.Umu,Umu,mu);
+
+  Dw.M(phi,MphiPrime);
+
+  ComplexD Sprime = innerProduct(phi,MphiPrime);
+
+  std::cout << GridLogMessage << " S      "<<S<<std::endl;
+  std::cout << GridLogMessage << " Sprime "<<Sprime<<std::endl;
+  std::cout << GridLogMessage << "dS      "<<Sprime-S<<std::endl;
+
+  Dw.Umu=zero;
+  PokeIndex<LorentzIndex>(Dw.Umu,dU,mu);
+  Dw.M(phi,dMphi);
+
+
+  ComplexD deltaS = innerProduct(phi,dMphi);
+  std::cout << GridLogMessage << "deltaS      "<<deltaS<<std::endl;
+
+  Dw.Umu=zero;
+  PokeIndex<LorentzIndex>(Dw.Umu,Umu_save,mu);
+  Dw.Mdir(phi,dMphi,mu,1);
+  dMphi = dt*mom*dMphi;
+
+  deltaS = innerProduct(phi,dMphi);
+  std::cout << GridLogMessage << "deltaS from inner prod of mom* M[u]     "<<deltaS<<std::endl;
+
+  deltaS = sum(trace(outerProduct(dMphi,phi)));
+  std::cout << GridLogMessage << "deltaS from trace outer prod of deltaM      "<<deltaS<<std::endl;
+
+/*
+  LatticeComplex lip(&Grid);
+  lip  = localInnerProduct(phi,dMphi);
+  
+  LatticeComplex trop(&Grid);
+  trop = trace(outerProduct(dMphi,phi));
+
+  LatticeSpinColourMatrix op(&Grid);
+  op = outerProduct(dMphi,phi);
+
+  LatticeSpinColourMatrix hop(&Grid);
+  LatticeComplex op_cpt(&Grid);
+  for(int s1=0;s1<Ns;s1++){
+  for(int s2=0;s2<Ns;s2++){
+  for(int c1=0;c1<Nc;c1++){
+  for(int c2=0;c2<Nc;c2++){
+
+    op_cpt = peekColour(peekSpin(dMphi,s1),c1) * adj(peekColour(peekSpin(phi,s2),c2));
+
+    parallel_for(auto i=hop.begin();i<hop.end();i++){
+      hop[i]()(s1,s2)(c1,c2) = op_cpt[i]()()();
+    }
+    
+  }}}}
+
+  LatticeSpinColourMatrix diffop(&Grid);
+
+  diffop = hop - op;
+  std::cout << GridLogMessage << "hand outer prod diff   "<<norm2(diffop)<<std::endl;
+
+  deltaS = sum(trace(hop));
+  std::cout << GridLogMessage << "deltaS hop   "<<deltaS<<std::endl;
+
+  std::cout << GridLogMessage<< "  phi[0] : "<<  phi._odata[0]<<std::endl;
+  std::cout << GridLogMessage<< "dMphi[0] : "<<dMphi._odata[0]<<std::endl;
+  std::cout << GridLogMessage<< "hop[0]   : "<<  hop._odata[0]<<std::endl;
+  std::cout << GridLogMessage<< " op[0]   : "<<   op._odata[0]<<std::endl;
+
+
+  std::cout << GridLogMessage << "lip      "<<lip<<std::endl;
+  std::cout << GridLogMessage << "trop     "<<trop<<std::endl;
+
+*/  
+  
+  //  std::cout << GridLogMessage << " UdSdU " << UdSdU << std::endl;
+
+  LatticeComplex dS(&Grid); dS = zero;
+  parallel_for(auto i=mom.begin();i<mom.end();i++){
+    dS[i]() = trace(mom[i]() * UdSdU[i](mu) )*dt;
+  }
+  Complex dSpred = sum(dS);
+
+  std::cout << GridLogMessage << "predict dS    "<< dSpred <<std::endl;
+
+
+  cout<< GridLogMessage << "Done" <<std::endl;
+  Grid_finalize();
+}

From 9de40578d37b6525c9c907eb5bcd3f9765a50762 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sun, 26 Jul 2015 12:28:03 +0900
Subject: [PATCH 406/429] Two flavour pseudofermion action

---
 lib/qcd/action/pseudofermion/TwoFlavour.h | 178 ++++++++++++++++++++++
 1 file changed, 178 insertions(+)
 create mode 100644 lib/qcd/action/pseudofermion/TwoFlavour.h

diff --git a/lib/qcd/action/pseudofermion/TwoFlavour.h b/lib/qcd/action/pseudofermion/TwoFlavour.h
new file mode 100644
index 00000000..a65f4ee3
--- /dev/null
+++ b/lib/qcd/action/pseudofermion/TwoFlavour.h
@@ -0,0 +1,178 @@
+#ifndef QCD_PSEUDOFERMION_TWO_FLAVOUR_H
+#define QCD_PSEUDOFERMION_TWO_FLAVOUR_H
+
+namespace Grid{
+  namespace QCD{
+
+    // Placeholder comments:
+
+    ///////////////////////////////////////
+    // Two flavour ratio
+    ///////////////////////////////////////
+    // S = phi^dag V (Mdag M)^-1 V^dag  phi
+    // dS/du = phi^dag dV (Mdag M)^-1 V^dag  phi
+    //       - phi^dag V (Mdag M)^-1 [ Mdag dM + dMdag M ]  (Mdag M)^-1 V^dag  phi
+    //       + phi^dag V (Mdag M)^-1 dV^dag  phi
+
+    ///////////////////////////////////////
+    // One flavour rational
+    ///////////////////////////////////////
+
+    // S_f = chi^dag *  N(M^dag*M)/D(M^dag*M) * chi
+    //
+    // Here, M is some operator 
+    // N and D makeup the rat. poly 
+    //
+    // Need
+    // dS_f/dU = chi^dag   P/Q d[N/D]  P/Q  chi
+    //
+    // Here N/D \sim R_{-1/2} ~ (M^dagM)^{-1/2}
+    //
+    // N/D is expressed as partial fraction expansion:
+    //
+    //           a0 + \sum_k ak/(M^dagM + bk)
+    //
+    // d[N/D] is then
+    //
+    //          \sum_k -ak [M^dagM+bk]^{-1}  [ dM^dag M + M^dag dM ] [M^dag M + bk]^{-1}
+    //
+    // Need
+    //
+    //       Mf Phi_k = [MdagM+bk]^{-1} Phi
+    //       Mf Phi   = \sum_k ak [MdagM+bk]^{-1} Phi
+    //
+    // With these building blocks
+    //
+    //       dS/dU =  \sum_k -ak Mf Phi_k^dag      [ dM^dag M + M^dag dM ] Mf Phi_k
+    //        S    = innerprodReal(Phi,Mf Phi);
+    
+    ///////////////////////////////////////
+    // One flavour rational ratio
+    ///////////////////////////////////////
+
+    // S_f = chi^dag* P(V^dag*V)/Q(V^dag*V)* N(M^dag*M)/D(M^dag*M)* P(V^dag*V)/Q(V^dag*V)* chi
+    //
+    // Here, M is some 5D operator and V is the Pauli-Villars field
+    // N and D makeup the rat. poly of the M term and P and & makeup the rat.poly of the denom term
+    //
+    // Need
+    // dS_f/dU =  chi^dag d[P/Q]  N/D   P/Q  chi
+    //         +  chi^dag   P/Q d[N/D]  P/Q  chi
+    //         +  chi^dag   P/Q   N/D d[P/Q] chi
+    //
+    // Here P/Q \sim R_{1/4}  ~ (V^dagV)^{1/4}
+    // Here N/D \sim R_{-1/2} ~ (M^dagM)^{-1/2}
+    //
+    // P/Q is expressed as partial fraction expansion:
+    //
+    //           a0 + \sum_k ak/(V^dagV + bk)
+    //
+    // d[P/Q] is then
+    //
+    //          \sum_k -ak [V^dagV+bk]^{-1}  [ dV^dag V + V^dag dV ] [V^dag V + bk]^{-1}
+    //
+    // and similar for N/D.
+    // 
+    // Need
+    //       MpvPhi_k   = [Vdag V + bk]^{-1} chi
+    //
+    //       MpvPhi     = {a0 +  \sum_k ak [Vdag V + bk]^{-1} }chi
+    //
+    //       MfMpvPhi_k = [MdagM+bk]^{-1} MpvPhi
+    //      
+    //       MfMpvPhi   = {a0 +  \sum_k ak [Mdag M + bk]^{-1} } MpvPhi
+    //
+    //       MpvMfMpvPhi_k = [Vdag V + bk]^{-1} MfMpvchi
+    //
+    // With these building blocks
+    //
+    //       dS/dU =  
+    //                 \sum_k -ak MpvPhi_k^dag        [ dV^dag V + V^dag dV ] MpvMfMpvPhi_k           <- deriv on P left
+    //             +   \sum_k -ak MpvMfMpvPhi_k^\dag  [ dV^dag V + V^dag dV ] MpvPhi_k
+    //             +   \sum_k -ak MfMpvPhi_k^dag      [ dM^dag M + M^dag dM ] MfMpvPhi_k
+
+    
+    ////////////////////////////////////////////////////////////////////////
+    // Two flavour pseudofermion action for any dop
+    ////////////////////////////////////////////////////////////////////////
+    template<class GaugeField,class MatrixField,class FermionField,class FermionOperator>
+      class TwoFlavourPseudoFermionAction : public Action<GaugeField> {
+
+    private:
+
+      FermionOperator<FermionField,GaugeField> & FermOp;// the basic operator
+
+      OperatorFunction<FermionField> &DerivativeSolver;
+
+      OperatorFunction<FermionField> &ActionSolver;
+
+      GridBase *Grid;
+
+      FermionField Phi; // the pseudo fermion field for this trajectory
+
+    public:
+      /////////////////////////////////////////////////
+      // Pass in required objects.
+      /////////////////////////////////////////////////
+      TwoFlavourPseudoFermionAction(FermionOperator &Op, 
+				    OperatorFunction<FermionField> & DS,
+				    OperatorFunction<FermionField> & AS
+				    ) : FermOp(Op), DerivativeSolver(DS), ActionSolver(AS) {
+      };
+      
+      //////////////////////////////////////////////////////////////////////////////////////
+      // Push the gauge field in to the dops. Assume any BC's and smearing already applied
+      //////////////////////////////////////////////////////////////////////////////////////
+      virtual void init(const GaugeField &U, GridParallelRNG& pRNG) {
+
+	// width? Must check
+	gaussian(Phi,pRNG);
+
+      };
+
+      //////////////////////////////////////////////////////
+      // S = phi^dag (Mdag M)^-1 phi
+      //////////////////////////////////////////////////////
+      virtual RealD S(const GaugeField &U) {
+
+	FermionField X(Grid);
+	FermionField Y(Grid);
+	
+	MdagMLinearOperator<FermionOperator<FermionField,GaugeField>,FermionField> MdagMOp(FermOp);
+
+	ActionSolver(MdagMop,Phi,X);
+	MdagMOp.Op(X,Y);
+
+	RealD action = norm2(Y);
+	
+	return action;
+      };
+
+      //////////////////////////////////////////////////////
+      // dS/du = - phi^dag  (Mdag M)^-1 [ Mdag dM + dMdag M ]  (Mdag M)^-1 phi
+      //////////////////////////////////////////////////////
+      virtual void deriv(const GaugeField &U,GaugeField & dSdU) {
+
+	FermionField X(Grid);
+	FermionField Y(Grid);
+	GaugeField   tmp(Grid);
+
+	MdagMLinearOperator<FermionOperator<FermionField,GaugeField>,FermionField> MdagMOp(FermOp);
+
+	DerivativeSolver(MdagMop,Phi,X);
+	MdagMOp.Op(X,Y);
+
+	// Our conventions really make this UdSdU; We do not differentiate wrt Udag here.
+	// So must take dSdU - adj(dSdU) and left multiply by mom to get dS/dt.
+
+	FermOp.MDeriv(tmp , Y, X,DaggerNo );  dSdU=tmp;
+	FermOp.MDeriv(tmp , X, Y,DaggerYes);  dSdU=-UdSdU-tmp;
+
+      };
+
+    };
+    
+  }
+}
+
+#endif

From 4c3f36b80c2117a27abd18a10fc442f861a0d8a9 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 27 Jul 2015 18:30:19 +0900
Subject: [PATCH 407/429] Files renamed

---
 configure.ac | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/configure.ac b/configure.ac
index f9283a32..fcb21fd0 100644
--- a/configure.ac
+++ b/configure.ac
@@ -11,7 +11,7 @@ AC_CANONICAL_SYSTEM
 AM_INIT_AUTOMAKE(subdir-objects)
 AC_CONFIG_MACRO_DIR([m4])
 AC_CONFIG_SRCDIR([lib/Grid.h])
-AC_CONFIG_HEADERS([lib/GridConfig.h])
+AC_CONFIG_HEADERS([lib/Config.h])
 m4_ifdef([AM_SILENT_RULES], [AM_SILENT_RULES([yes])])
 
 AC_MSG_NOTICE([

From bc09d7c3bd4146e914eb5114d8cb404a9ca84260 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 27 Jul 2015 18:32:28 +0900
Subject: [PATCH 408/429] Committing incomplete work for parameter file I/O.
 MacroMagic.h is central. Guido and I plan to move over to generating virtual
 (XML, JSON, YAML, text, binary) encoding from macro based system.

---
 lib/AlignedAllocator.h                     |     7 +
 lib/{GridConfig.h => Config.h}             |     0
 lib/{GridConfig.h.in => Config.h.in}       |     2 +-
 lib/Grid.h                                 |    89 +-
 lib/{GridInit.cc => Init.cc}               |     0
 lib/Init.h                                 |    32 +
 lib/{GridLog.cc => Log.cc}                 |     4 +-
 lib/{GridLog.h => Log.h}                   |     0
 lib/MacroMagic.h                           |    78 +
 lib/Make.inc                               |     4 +-
 lib/{GridTime.h => Timer.h}                |    14 +-
 lib/algorithms/approx/MultiShiftFunction.h |     2 +
 lib/pugixml/README.md                      |    44 +
 lib/pugixml/pugiconfig.hpp                 |    71 +
 lib/pugixml/pugixml.cpp                    | 12485 +++++++++++++++++++
 lib/pugixml/pugixml.hpp                    |  1400 +++
 lib/pugixml/readme.txt                     |    52 +
 lib/qcd/action/ActionBase.h                |    24 +-
 lib/qcd/action/fermion/WilsonFermion5D.cc  |     1 +
 lib/qcd/hmc/HMC.cc                         |     3 -
 lib/qcd/hmc/HMC.h                          |    37 +-
 tests/Make.inc                             |     6 +-
 22 files changed, 14262 insertions(+), 93 deletions(-)
 rename lib/{GridConfig.h => Config.h} (100%)
 rename lib/{GridConfig.h.in => Config.h.in} (98%)
 rename lib/{GridInit.cc => Init.cc} (100%)
 create mode 100644 lib/Init.h
 rename lib/{GridLog.cc => Log.cc} (94%)
 rename lib/{GridLog.h => Log.h} (100%)
 create mode 100644 lib/MacroMagic.h
 rename lib/{GridTime.h => Timer.h} (92%)
 create mode 100644 lib/pugixml/README.md
 create mode 100644 lib/pugixml/pugiconfig.hpp
 create mode 100644 lib/pugixml/pugixml.cpp
 create mode 100644 lib/pugixml/pugixml.hpp
 create mode 100644 lib/pugixml/readme.txt

diff --git a/lib/AlignedAllocator.h b/lib/AlignedAllocator.h
index 3e90371a..ea25f6a8 100644
--- a/lib/AlignedAllocator.h
+++ b/lib/AlignedAllocator.h
@@ -1,6 +1,13 @@
 #ifndef GRID_ALIGNED_ALLOCATOR_H
 #define GRID_ALIGNED_ALLOCATOR_H
 
+#ifdef HAVE_MALLOC_MALLOC_H
+#include <malloc/malloc.h>
+#endif
+#ifdef HAVE_MALLOC_H
+#include <malloc.h>
+#endif
+
 #include <immintrin.h>
 #ifdef HAVE_MM_MALLOC_H
 #include <mm_malloc.h>
diff --git a/lib/GridConfig.h b/lib/Config.h
similarity index 100%
rename from lib/GridConfig.h
rename to lib/Config.h
diff --git a/lib/GridConfig.h.in b/lib/Config.h.in
similarity index 98%
rename from lib/GridConfig.h.in
rename to lib/Config.h.in
index bcb8d811..d5e53e9b 100644
--- a/lib/GridConfig.h.in
+++ b/lib/Config.h.in
@@ -1,4 +1,4 @@
-/* lib/GridConfig.h.in.  Generated from configure.ac by autoheader.  */
+/* lib/Config.h.in.  Generated from configure.ac by autoheader.  */
 
 /* AVX Intrinsics */
 #undef AVX1
diff --git a/lib/Grid.h b/lib/Grid.h
index 751b609d..624bc5a7 100644
--- a/lib/Grid.h
+++ b/lib/Grid.h
@@ -6,95 +6,48 @@
 //  Copyright (c) 2014 University of Edinburgh. All rights reserved.
 //
 
-
 #ifndef GRID_H
 #define GRID_H
 
+///////////////////
+// Std C++ dependencies
+///////////////////
 #include <cassert>
-
 #include <complex>
 #include <vector>
-
 #include <iostream>
 #include <iomanip>
 #include <random>
 #include <functional>
-
 #include <stdio.h>
 #include <stdlib.h>
-#include <sys/time.h>
 #include <stdio.h>
 #include <signal.h>
+#include <ctime>
+#include <sys/time.h>
+#include <chrono>
 
-#ifndef MAX
-#define MAX(x,y) ((x)>(y)?(x):(y))
-#define MIN(x,y) ((x)>(y)?(y):(x))
-#endif
-
-#define strong_inline __attribute__((always_inline)) inline
-
-#include <GridConfig.h>
-
-#include <GridTime.h>
-#include <GridLog.h>
-
-////////////////////////////////////////////////////////////
-// Tunable header includes
-////////////////////////////////////////////////////////////
-
-#ifdef HAVE_MALLOC_MALLOC_H
-#include <malloc/malloc.h>
-#endif
-#ifdef HAVE_MALLOC_H
-#include <malloc.h>
-#endif
-
+///////////////////
+// Grid headers
+///////////////////
+#include <MacroMagic.h>
+#include <Config.h>
+#include <Timer.h>
+#include <Log.h>
 #include <AlignedAllocator.h>
-
 #include <Simd.h>
 #include <Threads.h>
-
-#include <Communicator.h> // subdir aggregate
-#include <Cartesian.h> // subdir aggregate
-#include <Tensors.h>   // subdir aggregate
-#include <Lattice.h>   // subdir aggregate
-#include <Cshift.h>    // subdir aggregate
-#include <Stencil.h>   // subdir aggregate
-#include <Algorithms.h>// subdir aggregate
-
+#include <Communicator.h> 
+#include <Cartesian.h>    
+#include <Tensors.h>      
+#include <Lattice.h>      
+#include <Cshift.h>       
+#include <Stencil.h>      
+#include <Algorithms.h>   
 #include <qcd/QCD.h>
 #include <parallelIO/NerscIO.h>
 
-namespace Grid {
+#include <Init.h>
 
-  void Grid_init(int *argc,char ***argv);
-  void Grid_finalize(void);
-  // internal, controled with --handle
-  void Grid_sa_signal_handler(int sig,siginfo_t *si,void * ptr);
-  void Grid_debug_handler_init(void);
-  void Grid_quiesce_nodes(void);
-  void Grid_unquiesce_nodes(void);
-
-  // C++11 time facilities better?
-  double usecond(void);
-
-  const std::vector<int> GridDefaultSimd(int dims,int nsimd);
-  const std::vector<int> &GridDefaultLatt(void);
-  const std::vector<int> &GridDefaultMpi(void);
-  const int              &GridThreads(void)  ;
-  void                 GridSetThreads(int t) ;
-
-  // Common parsing chores
-  std::string GridCmdOptionPayload(char ** begin, char ** end, const std::string & option);
-  bool        GridCmdOptionExists(char** begin, char** end, const std::string& option);
-  std::string GridCmdVectorIntToString(const std::vector<int> & vec);
-
-  void GridParseLayout(char **argv,int argc,
-		       std::vector<int> &latt,
-		       std::vector<int> &simd,
-		       std::vector<int> &mpi);
-
-
-};
 
 #endif
diff --git a/lib/GridInit.cc b/lib/Init.cc
similarity index 100%
rename from lib/GridInit.cc
rename to lib/Init.cc
diff --git a/lib/Init.h b/lib/Init.h
new file mode 100644
index 00000000..aa0b74cc
--- /dev/null
+++ b/lib/Init.h
@@ -0,0 +1,32 @@
+#ifndef GRID_INIT_H
+#define GRID_INIT_H
+
+namespace Grid {
+
+  void Grid_init(int *argc,char ***argv);
+  void Grid_finalize(void);
+  // internal, controled with --handle
+  void Grid_sa_signal_handler(int sig,siginfo_t *si,void * ptr);
+  void Grid_debug_handler_init(void);
+  void Grid_quiesce_nodes(void);
+  void Grid_unquiesce_nodes(void);
+
+  const std::vector<int> GridDefaultSimd(int dims,int nsimd);
+  const std::vector<int> &GridDefaultLatt(void);
+  const std::vector<int> &GridDefaultMpi(void);
+  const int              &GridThreads(void)  ;
+  void                    GridSetThreads(int t) ;
+
+  // Common parsing chores
+  std::string GridCmdOptionPayload(char ** begin, char ** end, const std::string & option);
+  bool        GridCmdOptionExists(char** begin, char** end, const std::string& option);
+  std::string GridCmdVectorIntToString(const std::vector<int> & vec);
+
+  void GridParseLayout(char **argv,int argc,
+		       std::vector<int> &latt,
+		       std::vector<int> &simd,
+		       std::vector<int> &mpi);
+
+
+};
+#endif
diff --git a/lib/GridLog.cc b/lib/Log.cc
similarity index 94%
rename from lib/GridLog.cc
rename to lib/Log.cc
index b171a828..468d9116 100644
--- a/lib/GridLog.cc
+++ b/lib/Log.cc
@@ -40,7 +40,7 @@ void Grid_quiesce_nodes(void)
   int me;
   MPI_Comm_rank(MPI_COMM_WORLD,&me);
   if ( me ) { 
-    std::cout<<GridLogMessage.setstate(std::ios::badbit);
+    std::cout.setstate(std::ios::badbit);
   }
 #endif
 }
@@ -48,7 +48,7 @@ void Grid_quiesce_nodes(void)
 void Grid_unquiesce_nodes(void)
 {
 #ifdef GRID_COMMS_MPI
-    std::cout<<GridLogMessage.clear();
+    std::cout.clear();
 #endif
 }
 
diff --git a/lib/GridLog.h b/lib/Log.h
similarity index 100%
rename from lib/GridLog.h
rename to lib/Log.h
diff --git a/lib/MacroMagic.h b/lib/MacroMagic.h
new file mode 100644
index 00000000..00f60089
--- /dev/null
+++ b/lib/MacroMagic.h
@@ -0,0 +1,78 @@
+#ifndef GRID_MACRO_MAGIC_H
+#define GRID_MACRO_MAGIC_H
+
+#define strong_inline __attribute__((always_inline)) inline
+
+#ifndef MAX
+#define MAX(x,y) ((x)>(y)?(x):(y))
+#define MIN(x,y) ((x)>(y)?(y):(x))
+#endif
+
+#define GRID_MACRO_FIRST(a, ...) a
+#define GRID_MACRO_SECOND(a, b, ...) b
+
+#define GRID_MACRO_EMPTY()
+
+#define GRID_MACRO_EVAL(...)     GRID_MACRO_EVAL1024(__VA_ARGS__)
+#define GRID_MACRO_EVAL1024(...) GRID_MACRO_EVAL512(GRID_MACRO_EVAL512(__VA_ARGS__))
+#define GRID_MACRO_EVAL512(...)  GRID_MACRO_EVAL256(GRID_MACRO_EVAL256(__VA_ARGS__))
+#define GRID_MACRO_EVAL256(...)  GRID_MACRO_EVAL128(GRID_MACRO_EVAL128(__VA_ARGS__))
+#define GRID_MACRO_EVAL128(...)  GRID_MACRO_EVAL64(GRID_MACRO_EVAL64(__VA_ARGS__))
+#define GRID_MACRO_EVAL64(...)   GRID_MACRO_EVAL32(GRID_MACRO_EVAL32(__VA_ARGS__))
+#define GRID_MACRO_EVAL32(...)   GRID_MACRO_EVAL16(GRID_MACRO_EVAL16(__VA_ARGS__))
+#define GRID_MACRO_EVAL16(...)   GRID_MACRO_EVAL8(GRID_MACRO_EVAL8(__VA_ARGS__))
+#define GRID_MACRO_EVAL8(...)    GRID_MACRO_EVAL4(GRID_MACRO_EVAL4(__VA_ARGS__))
+#define GRID_MACRO_EVAL4(...)    GRID_MACRO_EVAL2(GRID_MACRO_EVAL2(__VA_ARGS__))
+#define GRID_MACRO_EVAL2(...)    GRID_MACRO_EVAL1(GRID_MACRO_EVAL1(__VA_ARGS__))
+#define GRID_MACRO_EVAL1(...) __VA_ARGS__
+
+#define GRID_MACRO_DEFER1(m) m GRID_MACRO_EMPTY()
+#define GRID_MACRO_DEFER2(m) m GRID_MACRO_EMPTY GRID_MACRO_EMPTY()()
+#define GRID_MACRO_DEFER3(m) m GRID_MACRO_EMPTY GRID_MACRO_EMPTY GRID_MACRO_EMPTY()()()
+#define GRID_MACRO_DEFER4(m) m GRID_MACRO_EMPTY GRID_MACRO_EMPTY GRID_MACRO_EMPTY GRID_MACRO_EMPTY()()()()
+
+#define GRID_MACRO_IS_PROBE(...) GRID_MACRO_SECOND(__VA_ARGS__, 0)
+#define GRID_MACRO_PROBE() ~, 1
+
+#define GRID_MACRO_CAT(a,b) a ## b
+
+#define GRID_MACRO_NOT(x) GRID_MACRO_IS_PROBE(GRID_MACRO_CAT(_GRID_MACRO_NOT_, x))
+#define _GRID_MACRO_NOT_0 GRID_MACRO_PROBE()
+
+#define GRID_MACRO_BOOL(x) GRID_MACRO_NOT(GRID_MACRO_NOT(x))
+
+#define GRID_MACRO_IF_ELSE(condition) _GRID_MACRO_IF_ELSE(GRID_MACRO_BOOL(condition))
+#define _GRID_MACRO_IF_ELSE(condition) GRID_MACRO_CAT(_GRID_MACRO_IF_, condition)
+
+#define _GRID_MACRO_IF_1(...) __VA_ARGS__ _GRID_MACRO_IF_1_ELSE
+#define _GRID_MACRO_IF_0(...)             _GRID_MACRO_IF_0_ELSE
+
+#define _GRID_MACRO_IF_1_ELSE(...)
+#define _GRID_MACRO_IF_0_ELSE(...) __VA_ARGS__
+
+#define GRID_MACRO_HAS_ARGS(...) GRID_MACRO_BOOL(GRID_MACRO_FIRST(_GRID_MACRO_END_OF_ARGUMENTS_ __VA_ARGS__)())
+#define _GRID_MACRO_END_OF_ARGUMENTS_() 0
+
+#define GRID_MACRO_MAP(m, first, second, ...)   \
+  m(first,second)                           \
+  GRID_MACRO_IF_ELSE(GRID_MACRO_HAS_ARGS(__VA_ARGS__))(				       \
+				 GRID_MACRO_DEFER4(_GRID_MACRO_MAP)()(m, __VA_ARGS__)   \
+				     )(                                 \
+				       /* Do nothing, just terminate */ \
+									)
+#define _GRID_MACRO_MAP() GRID_MACRO_MAP
+
+#define GRID_MACRO_MEMBER(A,B)        A B;
+
+#define GRID_MACRO_OS_WRITE_MEMBER(A,B) os<< #A <<" "#B <<" = "<< obj. B <<" ; " <<std::endl;
+
+#define GRID_DECL_CLASS_MEMBERS(cname,...)		\
+  GRID_MACRO_EVAL(GRID_MACRO_MAP(GRID_MACRO_MEMBER,__VA_ARGS__))		\
+  friend std::ostream & operator << (std::ostream &os, const cname &obj ) {	\
+    os<<"class "<<#cname<<" {"<<std::endl;\
+    GRID_MACRO_EVAL(GRID_MACRO_MAP(GRID_MACRO_OS_WRITE_MEMBER,__VA_ARGS__))	\
+      os<<"}";								\
+    return os;\
+  };  
+
+#endif
diff --git a/lib/Make.inc b/lib/Make.inc
index 621ef7e7..92699dbe 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/BfmHDCG.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/GeneralisedMinimumResidual.h ./algorithms/iterative/gmres.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./GridConfig.h ./GridLog.h ./GridTime.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./Old/Tensor_peek.h ./Old/Tensor_poke.h ./parallelIO/NerscIO.h ./qcd/action/ActionBase.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/hmc/HMC.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/hmc/integrators/Integrator_base.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_index.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/BfmHDCG.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/GeneralisedMinimumResidual.h ./algorithms/iterative/gmres.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Config.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./Init.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./Log.h ./MacroMagic.h ./Old/Tensor_peek.h ./Old/Tensor_poke.h ./parallelIO/NerscIO.h ./qcd/action/ActionBase.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/pseudofermion/TwoFlavour.h ./qcd/hmc/HMC.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/hmc/integrators/Integrator_base.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_index.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h ./Timer.h
 
-CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./GridInit.cc ./GridLog.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/hmc/HMC.cc ./qcd/hmc/integrators/Integrator.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
+CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./Init.cc ./Log.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/hmc/HMC.cc ./qcd/hmc/integrators/Integrator.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/GridTime.h b/lib/Timer.h
similarity index 92%
rename from lib/GridTime.h
rename to lib/Timer.h
index 841ace11..43ace890 100644
--- a/lib/GridTime.h
+++ b/lib/Timer.h
@@ -1,9 +1,17 @@
 #ifndef GRID_TIME_H
 #define GRID_TIME_H
-namespace Grid {
-  // Dress the output; use std::chrono
-#include <chrono>
+
+#include <sys/time.h>
 #include <ctime>
+#include <chrono>
+
+namespace Grid {
+
+
+  // Dress the output; use std::chrono
+
+// C++11 time facilities better?
+double usecond(void);
 
 typedef  std::chrono::system_clock          GridClock;
 typedef  std::chrono::time_point<GridClock> GridTimePoint;
diff --git a/lib/algorithms/approx/MultiShiftFunction.h b/lib/algorithms/approx/MultiShiftFunction.h
index 4136736e..fceed2eb 100644
--- a/lib/algorithms/approx/MultiShiftFunction.h
+++ b/lib/algorithms/approx/MultiShiftFunction.h
@@ -1,6 +1,8 @@
 #ifndef MULTI_SHIFT_FUNCTION
 #define MULTI_SHIFT_FUNCTION
+
 namespace Grid {
+
 class MultiShiftFunction {
 public:
   int order;
diff --git a/lib/pugixml/README.md b/lib/pugixml/README.md
new file mode 100644
index 00000000..9d8a935f
--- /dev/null
+++ b/lib/pugixml/README.md
@@ -0,0 +1,44 @@
+pugixml [![Build Status](https://travis-ci.org/zeux/pugixml.svg?branch=master)](https://travis-ci.org/zeux/pugixml) [![Build status](https://ci.appveyor.com/api/projects/status/9hdks1doqvq8pwe7/branch/master?svg=true)](https://ci.appveyor.com/project/zeux/pugixml)
+=======
+
+pugixml is a C++ XML processing library, which consists of a DOM-like interface with rich traversal/modification
+capabilities, an extremely fast XML parser which constructs the DOM tree from an XML file/buffer, and an XPath 1.0
+implementation for complex data-driven tree queries. Full Unicode support is also available, with Unicode interface
+variants and conversions between different Unicode encodings (which happen automatically during parsing/saving).
+
+pugixml is used by a lot of projects, both open-source and proprietary, for performance and easy-to-use interface.
+
+## Documentation
+
+Documentation for the current release of pugixml is available on-line as two separate documents:
+
+* [Quick-start guide](http://pugixml.org/docs/quickstart.html), that aims to provide enough information to start using the library;
+* [Complete reference manual](http://pugixml.org/docs/manual.html), that describes all features of the library in detail.
+
+You’re advised to start with the quick-start guide; however, many important library features are either not described in it at all or only mentioned briefly; if you require more information you should read the complete manual.
+
+## License
+This library is available to anybody free of charge, under the terms of MIT License:
+
+Copyright (c) 2006-2015 Arseny Kapoulkine
+
+Permission is hereby granted, free of charge, to any person
+obtaining a copy of this software and associated documentation
+files (the "Software"), to deal in the Software without
+restriction, including without limitation the rights to use,
+copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the
+Software is furnished to do so, subject to the following
+conditions:
+
+The above copyright notice and this permission notice shall be
+included in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
+OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
+HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+OTHER DEALINGS IN THE SOFTWARE.
diff --git a/lib/pugixml/pugiconfig.hpp b/lib/pugixml/pugiconfig.hpp
new file mode 100644
index 00000000..5ee5131f
--- /dev/null
+++ b/lib/pugixml/pugiconfig.hpp
@@ -0,0 +1,71 @@
+/**
+ * pugixml parser - version 1.6
+ * --------------------------------------------------------
+ * Copyright (C) 2006-2015, by Arseny Kapoulkine (arseny.kapoulkine@gmail.com)
+ * Report bugs and download new versions at http://pugixml.org/
+ *
+ * This library is distributed under the MIT License. See notice at the end
+ * of this file.
+ *
+ * This work is based on the pugxml parser, which is:
+ * Copyright (C) 2003, by Kristen Wegner (kristen@tima.net)
+ */
+
+#ifndef HEADER_PUGICONFIG_HPP
+#define HEADER_PUGICONFIG_HPP
+
+// Uncomment this to enable wchar_t mode
+// #define PUGIXML_WCHAR_MODE
+
+// Uncomment this to disable XPath
+// #define PUGIXML_NO_XPATH
+
+// Uncomment this to disable STL
+// #define PUGIXML_NO_STL
+
+// Uncomment this to disable exceptions
+// #define PUGIXML_NO_EXCEPTIONS
+
+// Set this to control attributes for public classes/functions, i.e.:
+// #define PUGIXML_API __declspec(dllexport) // to export all public symbols from DLL
+// #define PUGIXML_CLASS __declspec(dllimport) // to import all classes from DLL
+// #define PUGIXML_FUNCTION __fastcall // to set calling conventions to all public functions to fastcall
+// In absence of PUGIXML_CLASS/PUGIXML_FUNCTION definitions PUGIXML_API is used instead
+
+// Tune these constants to adjust memory-related behavior
+// #define PUGIXML_MEMORY_PAGE_SIZE 32768
+// #define PUGIXML_MEMORY_OUTPUT_STACK 10240
+// #define PUGIXML_MEMORY_XPATH_PAGE_SIZE 4096
+
+// Uncomment this to switch to header-only version
+// #define PUGIXML_HEADER_ONLY
+
+// Uncomment this to enable long long support
+// #define PUGIXML_HAS_LONG_LONG
+
+#endif
+
+/**
+ * Copyright (c) 2006-2015 Arseny Kapoulkine
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use,
+ * copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following
+ * conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
+ * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
+ * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
diff --git a/lib/pugixml/pugixml.cpp b/lib/pugixml/pugixml.cpp
new file mode 100644
index 00000000..07f3a336
--- /dev/null
+++ b/lib/pugixml/pugixml.cpp
@@ -0,0 +1,12485 @@
+/**
+ * pugixml parser - version 1.6
+ * --------------------------------------------------------
+ * Copyright (C) 2006-2015, by Arseny Kapoulkine (arseny.kapoulkine@gmail.com)
+ * Report bugs and download new versions at http://pugixml.org/
+ *
+ * This library is distributed under the MIT License. See notice at the end
+ * of this file.
+ *
+ * This work is based on the pugxml parser, which is:
+ * Copyright (C) 2003, by Kristen Wegner (kristen@tima.net)
+ */
+
+#ifndef SOURCE_PUGIXML_CPP
+#define SOURCE_PUGIXML_CPP
+
+#include "pugixml.hpp"
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <assert.h>
+
+#ifdef PUGIXML_WCHAR_MODE
+#	include <wchar.h>
+#endif
+
+#ifndef PUGIXML_NO_XPATH
+#	include <math.h>
+#	include <float.h>
+#	ifdef PUGIXML_NO_EXCEPTIONS
+#		include <setjmp.h>
+#	endif
+#endif
+
+#ifndef PUGIXML_NO_STL
+#	include <istream>
+#	include <ostream>
+#	include <string>
+#endif
+
+// For placement new
+#include <new>
+
+#ifdef _MSC_VER
+#	pragma warning(push)
+#	pragma warning(disable: 4127) // conditional expression is constant
+#	pragma warning(disable: 4324) // structure was padded due to __declspec(align())
+#	pragma warning(disable: 4611) // interaction between '_setjmp' and C++ object destruction is non-portable
+#	pragma warning(disable: 4702) // unreachable code
+#	pragma warning(disable: 4996) // this function or variable may be unsafe
+#	pragma warning(disable: 4793) // function compiled as native: presence of '_setjmp' makes a function unmanaged
+#endif
+
+#ifdef __INTEL_COMPILER
+#	pragma warning(disable: 177) // function was declared but never referenced 
+#	pragma warning(disable: 279) // controlling expression is constant
+#	pragma warning(disable: 1478 1786) // function was declared "deprecated"
+#	pragma warning(disable: 1684) // conversion from pointer to same-sized integral type
+#endif
+
+#if defined(__BORLANDC__) && defined(PUGIXML_HEADER_ONLY)
+#	pragma warn -8080 // symbol is declared but never used; disabling this inside push/pop bracket does not make the warning go away
+#endif
+
+#ifdef __BORLANDC__
+#	pragma option push
+#	pragma warn -8008 // condition is always false
+#	pragma warn -8066 // unreachable code
+#endif
+
+#ifdef __SNC__
+// Using diag_push/diag_pop does not disable the warnings inside templates due to a compiler bug
+#	pragma diag_suppress=178 // function was declared but never referenced
+#	pragma diag_suppress=237 // controlling expression is constant
+#endif
+
+// Inlining controls
+#if defined(_MSC_VER) && _MSC_VER >= 1300
+#	define PUGI__NO_INLINE __declspec(noinline)
+#elif defined(__GNUC__)
+#	define PUGI__NO_INLINE __attribute__((noinline))
+#else
+#	define PUGI__NO_INLINE 
+#endif
+
+// Branch weight controls
+#if defined(__GNUC__)
+#	define PUGI__UNLIKELY(cond) __builtin_expect(cond, 0)
+#else
+#	define PUGI__UNLIKELY(cond) (cond)
+#endif
+
+// Simple static assertion
+#define PUGI__STATIC_ASSERT(cond) { static const char condition_failed[(cond) ? 1 : -1] = {0}; (void)condition_failed[0]; }
+
+// Digital Mars C++ bug workaround for passing char loaded from memory via stack
+#ifdef __DMC__
+#	define PUGI__DMC_VOLATILE volatile
+#else
+#	define PUGI__DMC_VOLATILE
+#endif
+
+// Borland C++ bug workaround for not defining ::memcpy depending on header include order (can't always use std::memcpy because some compilers don't have it at all)
+#if defined(__BORLANDC__) && !defined(__MEM_H_USING_LIST)
+using std::memcpy;
+using std::memmove;
+#endif
+
+// In some environments MSVC is a compiler but the CRT lacks certain MSVC-specific features
+#if defined(_MSC_VER) && !defined(__S3E__)
+#	define PUGI__MSVC_CRT_VERSION _MSC_VER
+#endif
+
+#ifdef PUGIXML_HEADER_ONLY
+#	define PUGI__NS_BEGIN namespace pugi { namespace impl {
+#	define PUGI__NS_END } }
+#	define PUGI__FN inline
+#	define PUGI__FN_NO_INLINE inline
+#else
+#	if defined(_MSC_VER) && _MSC_VER < 1300 // MSVC6 seems to have an amusing bug with anonymous namespaces inside namespaces
+#		define PUGI__NS_BEGIN namespace pugi { namespace impl {
+#		define PUGI__NS_END } }
+#	else
+#		define PUGI__NS_BEGIN namespace pugi { namespace impl { namespace {
+#		define PUGI__NS_END } } }
+#	endif
+#	define PUGI__FN
+#	define PUGI__FN_NO_INLINE PUGI__NO_INLINE
+#endif
+
+// uintptr_t
+#if !defined(_MSC_VER) || _MSC_VER >= 1600
+#	include <stdint.h>
+#else
+namespace pugi
+{
+#	ifndef _UINTPTR_T_DEFINED
+	typedef size_t uintptr_t;
+#	endif
+
+	typedef unsigned __int8 uint8_t;
+	typedef unsigned __int16 uint16_t;
+	typedef unsigned __int32 uint32_t;
+}
+#endif
+
+// Memory allocation
+PUGI__NS_BEGIN
+	PUGI__FN void* default_allocate(size_t size)
+	{
+		return malloc(size);
+	}
+
+	PUGI__FN void default_deallocate(void* ptr)
+	{
+		free(ptr);
+	}
+
+	template <typename T>
+	struct xml_memory_management_function_storage
+	{
+		static allocation_function allocate;
+		static deallocation_function deallocate;
+	};
+
+	// Global allocation functions are stored in class statics so that in header mode linker deduplicates them
+	// Without a template<> we'll get multiple definitions of the same static
+	template <typename T> allocation_function xml_memory_management_function_storage<T>::allocate = default_allocate;
+	template <typename T> deallocation_function xml_memory_management_function_storage<T>::deallocate = default_deallocate;
+
+	typedef xml_memory_management_function_storage<int> xml_memory;
+PUGI__NS_END
+
+// String utilities
+PUGI__NS_BEGIN
+	// Get string length
+	PUGI__FN size_t strlength(const char_t* s)
+	{
+		assert(s);
+
+	#ifdef PUGIXML_WCHAR_MODE
+		return wcslen(s);
+	#else
+		return strlen(s);
+	#endif
+	}
+
+	// Compare two strings
+	PUGI__FN bool strequal(const char_t* src, const char_t* dst)
+	{
+		assert(src && dst);
+
+	#ifdef PUGIXML_WCHAR_MODE
+		return wcscmp(src, dst) == 0;
+	#else
+		return strcmp(src, dst) == 0;
+	#endif
+	}
+
+	// Compare lhs with [rhs_begin, rhs_end)
+	PUGI__FN bool strequalrange(const char_t* lhs, const char_t* rhs, size_t count)
+	{
+		for (size_t i = 0; i < count; ++i)
+			if (lhs[i] != rhs[i])
+				return false;
+	
+		return lhs[count] == 0;
+	}
+
+	// Get length of wide string, even if CRT lacks wide character support
+	PUGI__FN size_t strlength_wide(const wchar_t* s)
+	{
+		assert(s);
+
+	#ifdef PUGIXML_WCHAR_MODE
+		return wcslen(s);
+	#else
+		const wchar_t* end = s;
+		while (*end) end++;
+		return static_cast<size_t>(end - s);
+	#endif
+	}
+
+#ifdef PUGIXML_WCHAR_MODE
+	// Convert string to wide string, assuming all symbols are ASCII
+	PUGI__FN void widen_ascii(wchar_t* dest, const char* source)
+	{
+		for (const char* i = source; *i; ++i) *dest++ = *i;
+		*dest = 0;
+	}
+#endif
+PUGI__NS_END
+
+// auto_ptr-like object for exception recovery
+PUGI__NS_BEGIN
+	template <typename T, typename D = void(*)(T*)> struct auto_deleter
+	{
+		T* data;
+		D deleter;
+
+		auto_deleter(T* data_, D deleter_): data(data_), deleter(deleter_)
+		{
+		}
+
+		~auto_deleter()
+		{
+			if (data) deleter(data);
+		}
+
+		T* release()
+		{
+			T* result = data;
+			data = 0;
+			return result;
+		}
+	};
+PUGI__NS_END
+
+#ifdef PUGIXML_COMPACT
+PUGI__NS_BEGIN
+	class compact_hash_table
+	{
+	public:
+		compact_hash_table(): _items(0), _capacity(0), _count(0)
+		{
+		}
+
+		void clear()
+		{
+			if (_items)
+			{
+				xml_memory::deallocate(_items);
+				_items = 0;
+				_capacity = 0;
+				_count = 0;
+			}
+		}
+
+		void** find(const void* key)
+		{
+			assert(key);
+
+			if (_capacity == 0) return 0;
+
+			size_t hashmod = _capacity - 1;
+			size_t bucket = hash(key) & hashmod;
+
+			for (size_t probe = 0; probe <= hashmod; ++probe)
+			{
+				item_t& probe_item = _items[bucket];
+
+				if (probe_item.key == key)
+					return &probe_item.value;
+
+				if (probe_item.key == 0)
+					return 0;
+
+				// hash collision, quadratic probing
+				bucket = (bucket + probe + 1) & hashmod;
+			}
+
+			assert(!"Hash table is full");
+			return 0;
+		}
+
+		void** insert(const void* key)
+		{
+			assert(key);
+			assert(_count < _capacity * 3 / 4);
+
+			size_t hashmod = _capacity - 1;
+			size_t bucket = hash(key) & hashmod;
+
+			for (size_t probe = 0; probe <= hashmod; ++probe)
+			{
+				item_t& probe_item = _items[bucket];
+
+				if (probe_item.key == 0)
+				{
+					probe_item.key = key;
+					_count++;
+					return &probe_item.value;
+				}
+
+				if (probe_item.key == key)
+					return &probe_item.value;
+
+				// hash collision, quadratic probing
+				bucket = (bucket + probe + 1) & hashmod;
+			}
+
+			assert(!"Hash table is full");
+			return 0;
+		}
+
+		bool reserve()
+		{
+			if (_count + 16 >= _capacity - _capacity / 4)
+				return rehash();
+
+			return true;
+		}
+
+	private:
+		struct item_t
+		{
+			const void* key;
+			void* value;
+		};
+
+		item_t* _items;
+		size_t _capacity;
+
+		size_t _count;
+
+		bool rehash();
+
+		static unsigned int hash(const void* key)
+		{
+			unsigned int h = static_cast<unsigned int>(reinterpret_cast<uintptr_t>(key));
+
+			// MurmurHash3 32-bit finalizer
+			h ^= h >> 16;
+			h *= 0x85ebca6bu;
+			h ^= h >> 13;
+			h *= 0xc2b2ae35u;
+			h ^= h >> 16;
+
+			return h;
+		}
+	};
+
+	PUGI__FN_NO_INLINE bool compact_hash_table::rehash()
+	{
+		compact_hash_table rt;
+		rt._capacity = (_capacity == 0) ? 32 : _capacity * 2;
+		rt._items = static_cast<item_t*>(xml_memory::allocate(sizeof(item_t) * rt._capacity));
+
+		if (!rt._items)
+			return false;
+
+		memset(rt._items, 0, sizeof(item_t) * rt._capacity);
+
+		for (size_t i = 0; i < _capacity; ++i)
+			if (_items[i].key)
+				*rt.insert(_items[i].key) = _items[i].value;
+
+		if (_items)
+			xml_memory::deallocate(_items);
+
+		_capacity = rt._capacity;
+		_items = rt._items;
+
+		return true;
+	}
+
+PUGI__NS_END
+#endif
+
+PUGI__NS_BEGIN
+	static const size_t xml_memory_page_size =
+	#ifdef PUGIXML_MEMORY_PAGE_SIZE
+		PUGIXML_MEMORY_PAGE_SIZE
+	#else
+		32768
+	#endif
+		;
+
+#ifdef PUGIXML_COMPACT
+	static const uintptr_t xml_memory_block_alignment = 4;
+
+	static const uintptr_t xml_memory_page_alignment = sizeof(void*);
+#else
+	static const uintptr_t xml_memory_block_alignment = sizeof(void*);
+
+	static const uintptr_t xml_memory_page_alignment = 64;
+	static const uintptr_t xml_memory_page_pointer_mask = ~(xml_memory_page_alignment - 1);
+#endif
+
+	// extra metadata bits
+	static const uintptr_t xml_memory_page_contents_shared_mask = 32;
+	static const uintptr_t xml_memory_page_name_allocated_mask = 16;
+	static const uintptr_t xml_memory_page_value_allocated_mask = 8;
+	static const uintptr_t xml_memory_page_type_mask = 7;
+
+	// combined masks for string uniqueness
+	static const uintptr_t xml_memory_page_name_allocated_or_shared_mask = xml_memory_page_name_allocated_mask | xml_memory_page_contents_shared_mask;
+	static const uintptr_t xml_memory_page_value_allocated_or_shared_mask = xml_memory_page_value_allocated_mask | xml_memory_page_contents_shared_mask;
+
+#ifdef PUGIXML_COMPACT
+	#define PUGI__GETPAGE_IMPL(header) (header).get_page()
+#else
+	#define PUGI__GETPAGE_IMPL(header) reinterpret_cast<impl::xml_memory_page*>((header) & impl::xml_memory_page_pointer_mask)
+#endif
+
+	#define PUGI__GETPAGE(n) PUGI__GETPAGE_IMPL((n)->header)
+	#define PUGI__NODETYPE(n) static_cast<xml_node_type>(((n)->header & impl::xml_memory_page_type_mask) + 1)
+
+	struct xml_allocator;
+
+	struct xml_memory_page
+	{
+		static xml_memory_page* construct(void* memory)
+		{
+			xml_memory_page* result = static_cast<xml_memory_page*>(memory);
+
+			result->allocator = 0;
+			result->prev = 0;
+			result->next = 0;
+			result->busy_size = 0;
+			result->freed_size = 0;
+
+		#ifdef PUGIXML_COMPACT
+			result->compact_string_base = 0;
+			result->compact_shared_parent = 0;
+			result->compact_page_marker = 0;
+		#endif
+
+			return result;
+		}
+
+		xml_allocator* allocator;
+
+		xml_memory_page* prev;
+		xml_memory_page* next;
+
+		size_t busy_size;
+		size_t freed_size;
+
+	#ifdef PUGIXML_COMPACT
+		char_t* compact_string_base;
+		void* compact_shared_parent;
+		uint32_t* compact_page_marker;
+	#endif
+	};
+
+	struct xml_memory_string_header
+	{
+		uint16_t page_offset; // offset from page->data
+		uint16_t full_size; // 0 if string occupies whole page
+	};
+
+	struct xml_allocator
+	{
+		xml_allocator(xml_memory_page* root): _root(root), _busy_size(root->busy_size)
+		{
+		#ifdef PUGIXML_COMPACT
+			_hash = 0;
+		#endif
+		}
+
+		xml_memory_page* allocate_page(size_t data_size)
+		{
+			size_t size = sizeof(xml_memory_page) + data_size;
+
+			// allocate block with some alignment, leaving memory for worst-case padding
+			void* memory = xml_memory::allocate(size + xml_memory_page_alignment);
+			if (!memory) return 0;
+
+			// align to next page boundary (note: this guarantees at least 1 usable byte before the page)
+			char* page_memory = reinterpret_cast<char*>((reinterpret_cast<uintptr_t>(memory) + xml_memory_page_alignment) & ~(xml_memory_page_alignment - 1));
+
+			// prepare page structure
+			xml_memory_page* page = xml_memory_page::construct(page_memory);
+			assert(page);
+
+			page->allocator = _root->allocator;
+
+			// record the offset for freeing the memory block
+			assert(page_memory > memory && page_memory - static_cast<char*>(memory) <= 127);
+			page_memory[-1] = static_cast<char>(page_memory - static_cast<char*>(memory));
+
+			return page;
+		}
+
+		static void deallocate_page(xml_memory_page* page)
+		{
+			char* page_memory = reinterpret_cast<char*>(page);
+
+			xml_memory::deallocate(page_memory - page_memory[-1]);
+		}
+
+		void* allocate_memory_oob(size_t size, xml_memory_page*& out_page);
+
+		void* allocate_memory(size_t size, xml_memory_page*& out_page)
+		{
+			if (PUGI__UNLIKELY(_busy_size + size > xml_memory_page_size))
+				return allocate_memory_oob(size, out_page);
+
+			void* buf = reinterpret_cast<char*>(_root) + sizeof(xml_memory_page) + _busy_size;
+
+			_busy_size += size;
+
+			out_page = _root;
+
+			return buf;
+		}
+
+	#ifdef PUGIXML_COMPACT
+		void* allocate_object(size_t size, xml_memory_page*& out_page)
+		{
+			void* result = allocate_memory(size + sizeof(uint32_t), out_page);
+			if (!result) return 0;
+
+			// adjust for marker
+			ptrdiff_t offset = static_cast<char*>(result) - reinterpret_cast<char*>(out_page->compact_page_marker);
+
+			if (PUGI__UNLIKELY(static_cast<uintptr_t>(offset) >= 256 * xml_memory_block_alignment))
+			{
+				// insert new marker
+				uint32_t* marker = static_cast<uint32_t*>(result);
+
+				*marker = static_cast<uint32_t>(reinterpret_cast<char*>(marker) - reinterpret_cast<char*>(out_page));
+				out_page->compact_page_marker = marker;
+
+				// since we don't reuse the page space until we reallocate it, we can just pretend that we freed the marker block
+				// this will make sure deallocate_memory correctly tracks the size
+				out_page->freed_size += sizeof(uint32_t);
+
+				return marker + 1;
+			}
+			else
+			{
+				// roll back uint32_t part
+				_busy_size -= sizeof(uint32_t);
+
+				return result;
+			}
+		}
+	#else
+		void* allocate_object(size_t size, xml_memory_page*& out_page)
+		{
+			return allocate_memory(size, out_page);
+		}
+	#endif
+
+		void deallocate_memory(void* ptr, size_t size, xml_memory_page* page)
+		{
+			if (page == _root) page->busy_size = _busy_size;
+
+			assert(ptr >= reinterpret_cast<char*>(page) + sizeof(xml_memory_page) && ptr < reinterpret_cast<char*>(page) + sizeof(xml_memory_page) + page->busy_size);
+			(void)!ptr;
+
+			page->freed_size += size;
+			assert(page->freed_size <= page->busy_size);
+
+			if (page->freed_size == page->busy_size)
+			{
+				if (page->next == 0)
+				{
+					assert(_root == page);
+
+					// top page freed, just reset sizes
+					page->busy_size = 0;
+					page->freed_size = 0;
+
+				#ifdef PUGIXML_COMPACT
+					// reset compact state to maximize efficiency
+					page->compact_string_base = 0;
+					page->compact_shared_parent = 0;
+					page->compact_page_marker = 0;
+				#endif
+
+					_busy_size = 0;
+				}
+				else
+				{
+					assert(_root != page);
+					assert(page->prev);
+
+					// remove from the list
+					page->prev->next = page->next;
+					page->next->prev = page->prev;
+
+					// deallocate
+					deallocate_page(page);
+				}
+			}
+		}
+
+		char_t* allocate_string(size_t length)
+		{
+			static const size_t max_encoded_offset = (1 << 16) * xml_memory_block_alignment;
+
+			PUGI__STATIC_ASSERT(xml_memory_page_size <= max_encoded_offset);
+
+			// allocate memory for string and header block
+			size_t size = sizeof(xml_memory_string_header) + length * sizeof(char_t);
+			
+			// round size up to block alignment boundary
+			size_t full_size = (size + (xml_memory_block_alignment - 1)) & ~(xml_memory_block_alignment - 1);
+
+			xml_memory_page* page;
+			xml_memory_string_header* header = static_cast<xml_memory_string_header*>(allocate_memory(full_size, page));
+
+			if (!header) return 0;
+
+			// setup header
+			ptrdiff_t page_offset = reinterpret_cast<char*>(header) - reinterpret_cast<char*>(page) - sizeof(xml_memory_page);
+
+			assert(page_offset % xml_memory_block_alignment == 0);
+			assert(page_offset >= 0 && static_cast<size_t>(page_offset) < max_encoded_offset);
+			header->page_offset = static_cast<uint16_t>(static_cast<size_t>(page_offset) / xml_memory_block_alignment);
+
+			// full_size == 0 for large strings that occupy the whole page
+			assert(full_size % xml_memory_block_alignment == 0);
+			assert(full_size < max_encoded_offset || (page->busy_size == full_size && page_offset == 0));
+			header->full_size = static_cast<uint16_t>(full_size < max_encoded_offset ? full_size / xml_memory_block_alignment : 0);
+
+			// round-trip through void* to avoid 'cast increases required alignment of target type' warning
+			// header is guaranteed a pointer-sized alignment, which should be enough for char_t
+			return static_cast<char_t*>(static_cast<void*>(header + 1));
+		}
+
+		void deallocate_string(char_t* string)
+		{
+			// this function casts pointers through void* to avoid 'cast increases required alignment of target type' warnings
+			// we're guaranteed the proper (pointer-sized) alignment on the input string if it was allocated via allocate_string
+
+			// get header
+			xml_memory_string_header* header = static_cast<xml_memory_string_header*>(static_cast<void*>(string)) - 1;
+			assert(header);
+
+			// deallocate
+			size_t page_offset = sizeof(xml_memory_page) + header->page_offset * xml_memory_block_alignment;
+			xml_memory_page* page = reinterpret_cast<xml_memory_page*>(static_cast<void*>(reinterpret_cast<char*>(header) - page_offset));
+
+			// if full_size == 0 then this string occupies the whole page
+			size_t full_size = header->full_size == 0 ? page->busy_size : header->full_size * xml_memory_block_alignment;
+
+			deallocate_memory(header, full_size, page);
+		}
+
+		bool reserve()
+		{
+		#ifdef PUGIXML_COMPACT
+			return _hash->reserve();
+		#else
+			return true;
+		#endif
+		}
+
+		xml_memory_page* _root;
+		size_t _busy_size;
+
+	#ifdef PUGIXML_COMPACT
+		compact_hash_table* _hash;
+	#endif
+	};
+
+	PUGI__FN_NO_INLINE void* xml_allocator::allocate_memory_oob(size_t size, xml_memory_page*& out_page)
+	{
+		const size_t large_allocation_threshold = xml_memory_page_size / 4;
+
+		xml_memory_page* page = allocate_page(size <= large_allocation_threshold ? xml_memory_page_size : size);
+		out_page = page;
+
+		if (!page) return 0;
+
+		if (size <= large_allocation_threshold)
+		{
+			_root->busy_size = _busy_size;
+
+			// insert page at the end of linked list
+			page->prev = _root;
+			_root->next = page;
+			_root = page;
+
+			_busy_size = size;
+		}
+		else
+		{
+			// insert page before the end of linked list, so that it is deleted as soon as possible
+			// the last page is not deleted even if it's empty (see deallocate_memory)
+			assert(_root->prev);
+
+			page->prev = _root->prev;
+			page->next = _root;
+
+			_root->prev->next = page;
+			_root->prev = page;
+
+			page->busy_size = size;
+		}
+
+		return reinterpret_cast<char*>(page) + sizeof(xml_memory_page);
+	}
+PUGI__NS_END
+
+#ifdef PUGIXML_COMPACT
+PUGI__NS_BEGIN
+	static const uintptr_t compact_alignment_log2 = 2;
+	static const uintptr_t compact_alignment = 1 << compact_alignment_log2;
+
+	class compact_header
+	{
+	public:
+		compact_header(xml_memory_page* page, unsigned int flags)
+		{
+			PUGI__STATIC_ASSERT(xml_memory_block_alignment == compact_alignment);
+
+			ptrdiff_t offset = (reinterpret_cast<char*>(this) - reinterpret_cast<char*>(page->compact_page_marker));
+			assert(offset % compact_alignment == 0 && static_cast<uintptr_t>(offset) < 256 * compact_alignment);
+
+			_page = static_cast<unsigned char>(offset >> compact_alignment_log2);
+			_flags = static_cast<unsigned char>(flags);
+		}
+
+		void operator&=(uintptr_t mod)
+		{
+			_flags &= mod;
+		}
+
+		void operator|=(uintptr_t mod)
+		{
+			_flags |= mod;
+		}
+
+		uintptr_t operator&(uintptr_t mod) const
+		{
+			return _flags & mod;
+		}
+
+		xml_memory_page* get_page() const
+		{
+			const char* page_marker = reinterpret_cast<const char*>(this) - (_page << compact_alignment_log2);
+			const char* page = page_marker - *reinterpret_cast<const uint32_t*>(page_marker);
+
+			return const_cast<xml_memory_page*>(reinterpret_cast<const xml_memory_page*>(page));
+		}
+
+	private:
+		unsigned char _page;
+		unsigned char _flags;
+	};
+
+	PUGI__FN xml_memory_page* compact_get_page(const void* object, int header_offset)
+	{
+		const compact_header* header = reinterpret_cast<const compact_header*>(static_cast<const char*>(object) - header_offset);
+
+		return header->get_page();
+	}
+
+	template <int header_offset, typename T> PUGI__FN_NO_INLINE T* compact_get_value(const void* object)
+	{
+		return static_cast<T*>(*compact_get_page(object, header_offset)->allocator->_hash->find(object));
+	}
+
+	template <int header_offset, typename T> PUGI__FN_NO_INLINE void compact_set_value(const void* object, T* value)
+	{
+		*compact_get_page(object, header_offset)->allocator->_hash->insert(object) = value;
+	}
+
+	template <typename T, int header_offset, int start = -126> class compact_pointer
+	{
+	public:
+		compact_pointer(): _data(0)
+		{
+		}
+
+		void operator=(const compact_pointer& rhs)
+		{
+			*this = rhs + 0;
+		}
+
+		void operator=(T* value)
+		{
+			if (value)
+			{
+				// value is guaranteed to be compact-aligned; 'this' is not
+				// our decoding is based on 'this' aligned to compact alignment downwards (see operator T*)
+				// so for negative offsets (e.g. -3) we need to adjust the diff by compact_alignment - 1 to
+				// compensate for arithmetic shift rounding for negative values
+				ptrdiff_t diff = reinterpret_cast<char*>(value) - reinterpret_cast<char*>(this);
+				ptrdiff_t offset = ((diff + int(compact_alignment - 1)) >> compact_alignment_log2) - start;
+
+				if (static_cast<uintptr_t>(offset) <= 253)
+					_data = static_cast<unsigned char>(offset + 1);
+				else
+				{
+					compact_set_value<header_offset>(this, value);
+
+					_data = 255;
+				}
+			}
+			else
+				_data = 0;
+		}
+
+		operator T*() const
+		{
+			if (_data)
+			{
+				if (_data < 255)
+				{
+					uintptr_t base = reinterpret_cast<uintptr_t>(this) & ~(compact_alignment - 1);
+
+					return reinterpret_cast<T*>(base + ((_data - 1 + start) << compact_alignment_log2));
+				}
+				else
+					return compact_get_value<header_offset, T>(this);
+			}
+			else
+				return 0;
+		}
+
+		T* operator->() const
+		{
+			return operator T*();
+		}
+
+	private:
+		unsigned char _data;
+	};
+
+	template <typename T, int header_offset> class compact_pointer_parent
+	{
+	public:
+		compact_pointer_parent(): _data(0)
+		{
+		}
+
+		void operator=(const compact_pointer_parent& rhs)
+		{
+			*this = rhs + 0;
+		}
+
+		void operator=(T* value)
+		{
+			if (value)
+			{
+				// value is guaranteed to be compact-aligned; 'this' is not
+				// our decoding is based on 'this' aligned to compact alignment downwards (see operator T*)
+				// so for negative offsets (e.g. -3) we need to adjust the diff by compact_alignment - 1 to
+				// compensate for arithmetic shift behavior for negative values
+				ptrdiff_t diff = reinterpret_cast<char*>(value) - reinterpret_cast<char*>(this);
+				ptrdiff_t offset = ((diff + int(compact_alignment - 1)) >> compact_alignment_log2) + 65533;
+
+				if (static_cast<uintptr_t>(offset) <= 65533)
+				{
+					_data = static_cast<unsigned short>(offset + 1);
+				}
+				else
+				{
+					xml_memory_page* page = compact_get_page(this, header_offset);
+
+					if (PUGI__UNLIKELY(page->compact_shared_parent == 0))
+						page->compact_shared_parent = value;
+
+					if (page->compact_shared_parent == value)
+					{
+						_data = 65534;
+					}
+					else
+					{
+						compact_set_value<header_offset>(this, value);
+
+						_data = 65535;
+					}
+				}
+			}
+			else
+			{
+				_data = 0;
+			}
+		}
+
+		operator T*() const
+		{
+			if (_data)
+			{
+				if (_data < 65534)
+				{
+					uintptr_t base = reinterpret_cast<uintptr_t>(this) & ~(compact_alignment - 1);
+
+					return reinterpret_cast<T*>(base + ((_data - 1 - 65533) << compact_alignment_log2));
+				}
+				else if (_data == 65534)
+					return static_cast<T*>(compact_get_page(this, header_offset)->compact_shared_parent);
+				else
+					return compact_get_value<header_offset, T>(this);
+			}
+			else
+				return 0;
+		}
+
+		T* operator->() const
+		{
+			return operator T*();
+		}
+
+	private:
+		uint16_t _data;
+	};
+
+	template <int header_offset, int base_offset> class compact_string
+	{
+	public:
+		compact_string(): _data(0)
+		{
+		}
+
+		void operator=(const compact_string& rhs)
+		{
+			*this = rhs + 0;
+		}
+
+		void operator=(char_t* value)
+		{
+			if (value)
+			{
+				xml_memory_page* page = compact_get_page(this, header_offset);
+
+				if (PUGI__UNLIKELY(page->compact_string_base == 0))
+					page->compact_string_base = value;
+
+				ptrdiff_t offset = value - page->compact_string_base;
+
+				if (static_cast<uintptr_t>(offset) < (65535 << 7))
+				{
+					uint16_t* base = reinterpret_cast<uint16_t*>(reinterpret_cast<char*>(this) - base_offset);
+
+					if (*base == 0)
+					{
+						*base = static_cast<uint16_t>((offset >> 7) + 1);
+						_data = static_cast<unsigned char>((offset & 127) + 1);
+					}
+					else
+					{
+						ptrdiff_t remainder = offset - ((*base - 1) << 7);
+
+						if (static_cast<uintptr_t>(remainder) <= 253)
+						{
+							_data = static_cast<unsigned char>(remainder + 1);
+						}
+						else
+						{
+							compact_set_value<header_offset>(this, value);
+
+							_data = 255;
+						}
+					}
+				}
+				else
+				{
+					compact_set_value<header_offset>(this, value);
+
+					_data = 255;
+				}
+			}
+			else
+			{
+				_data = 0;
+			}
+		}
+
+		operator char_t*() const
+		{
+			if (_data)
+			{
+				if (_data < 255)
+				{
+					xml_memory_page* page = compact_get_page(this, header_offset);
+
+					const uint16_t* base = reinterpret_cast<const uint16_t*>(reinterpret_cast<const char*>(this) - base_offset);
+					assert(*base);
+
+					ptrdiff_t offset = ((*base - 1) << 7) + (_data - 1);
+
+					return page->compact_string_base + offset;
+				}
+				else
+				{
+					return compact_get_value<header_offset, char_t>(this);
+				}
+			}
+			else
+				return 0;
+		}
+
+	private:
+		unsigned char _data;
+	};
+PUGI__NS_END
+#endif
+
+#ifdef PUGIXML_COMPACT
+namespace pugi
+{
+	struct xml_attribute_struct
+	{
+		xml_attribute_struct(impl::xml_memory_page* page): header(page, 0), namevalue_base(0)
+		{
+			PUGI__STATIC_ASSERT(sizeof(xml_attribute_struct) == 8);
+		}
+
+		impl::compact_header header;
+
+		uint16_t namevalue_base;
+
+		impl::compact_string<4, 2> name;
+		impl::compact_string<5, 3> value;
+
+		impl::compact_pointer<xml_attribute_struct, 6> prev_attribute_c;
+		impl::compact_pointer<xml_attribute_struct, 7, 0> next_attribute;
+	};
+
+	struct xml_node_struct
+	{
+		xml_node_struct(impl::xml_memory_page* page, xml_node_type type): header(page, type - 1), namevalue_base(0)
+		{
+			PUGI__STATIC_ASSERT(sizeof(xml_node_struct) == 12);
+		}
+
+		impl::compact_header header;
+
+		uint16_t namevalue_base;
+
+		impl::compact_string<4, 2> name;
+		impl::compact_string<5, 3> value;
+
+		impl::compact_pointer_parent<xml_node_struct, 6> parent;
+
+		impl::compact_pointer<xml_node_struct, 8, 0> first_child;
+
+		impl::compact_pointer<xml_node_struct,  9>    prev_sibling_c;
+		impl::compact_pointer<xml_node_struct, 10, 0> next_sibling;
+
+		impl::compact_pointer<xml_attribute_struct, 11, 0> first_attribute;
+	};
+}
+#else
+namespace pugi
+{
+	struct xml_attribute_struct
+	{
+		xml_attribute_struct(impl::xml_memory_page* page): header(reinterpret_cast<uintptr_t>(page)), name(0), value(0), prev_attribute_c(0), next_attribute(0)
+		{
+		}
+
+		uintptr_t header;
+
+		char_t*	name;
+		char_t*	value;
+
+		xml_attribute_struct* prev_attribute_c;
+		xml_attribute_struct* next_attribute;
+	};
+
+	struct xml_node_struct
+	{
+		xml_node_struct(impl::xml_memory_page* page, xml_node_type type): header(reinterpret_cast<uintptr_t>(page) | (type - 1)), name(0), value(0), parent(0), first_child(0), prev_sibling_c(0), next_sibling(0), first_attribute(0)
+		{
+		}
+
+		uintptr_t header;
+
+		char_t* name;
+		char_t* value;
+
+		xml_node_struct* parent;
+
+		xml_node_struct* first_child;
+
+		xml_node_struct* prev_sibling_c;
+		xml_node_struct* next_sibling;
+
+		xml_attribute_struct* first_attribute;
+	};
+}
+#endif
+
+PUGI__NS_BEGIN
+	struct xml_extra_buffer
+	{
+		char_t* buffer;
+		xml_extra_buffer* next;
+	};
+
+	struct xml_document_struct: public xml_node_struct, public xml_allocator
+	{
+		xml_document_struct(xml_memory_page* page): xml_node_struct(page, node_document), xml_allocator(page), buffer(0), extra_buffers(0)
+		{
+		#ifdef PUGIXML_COMPACT
+			_hash = &hash;
+		#endif
+		}
+
+		const char_t* buffer;
+
+		xml_extra_buffer* extra_buffers;
+
+	#ifdef PUGIXML_COMPACT
+		compact_hash_table hash;
+	#endif
+	};
+
+	template <typename Object> inline xml_allocator& get_allocator(const Object* object)
+	{
+		assert(object);
+
+		return *PUGI__GETPAGE(object)->allocator;
+	}
+
+	template <typename Object> inline xml_document_struct& get_document(const Object* object)
+	{
+		assert(object);
+
+		return *static_cast<xml_document_struct*>(PUGI__GETPAGE(object)->allocator);
+	}
+PUGI__NS_END
+
+// Low-level DOM operations
+PUGI__NS_BEGIN
+	inline xml_attribute_struct* allocate_attribute(xml_allocator& alloc)
+	{
+		xml_memory_page* page;
+		void* memory = alloc.allocate_object(sizeof(xml_attribute_struct), page);
+		if (!memory) return 0;
+
+		return new (memory) xml_attribute_struct(page);
+	}
+
+	inline xml_node_struct* allocate_node(xml_allocator& alloc, xml_node_type type)
+	{
+		xml_memory_page* page;
+		void* memory = alloc.allocate_object(sizeof(xml_node_struct), page);
+		if (!memory) return 0;
+
+		return new (memory) xml_node_struct(page, type);
+	}
+
+	inline void destroy_attribute(xml_attribute_struct* a, xml_allocator& alloc)
+	{
+		if (a->header & impl::xml_memory_page_name_allocated_mask)
+			alloc.deallocate_string(a->name);
+
+		if (a->header & impl::xml_memory_page_value_allocated_mask)
+			alloc.deallocate_string(a->value);
+
+		alloc.deallocate_memory(a, sizeof(xml_attribute_struct), PUGI__GETPAGE(a));
+	}
+
+	inline void destroy_node(xml_node_struct* n, xml_allocator& alloc)
+	{
+		if (n->header & impl::xml_memory_page_name_allocated_mask)
+			alloc.deallocate_string(n->name);
+
+		if (n->header & impl::xml_memory_page_value_allocated_mask)
+			alloc.deallocate_string(n->value);
+
+		for (xml_attribute_struct* attr = n->first_attribute; attr; )
+		{
+			xml_attribute_struct* next = attr->next_attribute;
+
+			destroy_attribute(attr, alloc);
+
+			attr = next;
+		}
+
+		for (xml_node_struct* child = n->first_child; child; )
+		{
+			xml_node_struct* next = child->next_sibling;
+
+			destroy_node(child, alloc);
+
+			child = next;
+		}
+
+		alloc.deallocate_memory(n, sizeof(xml_node_struct), PUGI__GETPAGE(n));
+	}
+
+	inline void append_node(xml_node_struct* child, xml_node_struct* node)
+	{
+		child->parent = node;
+
+		xml_node_struct* head = node->first_child;
+
+		if (head)
+		{
+			xml_node_struct* tail = head->prev_sibling_c;
+
+			tail->next_sibling = child;
+			child->prev_sibling_c = tail;
+			head->prev_sibling_c = child;
+		}
+		else
+		{
+			node->first_child = child;
+			child->prev_sibling_c = child;
+		}
+	}
+
+	inline void prepend_node(xml_node_struct* child, xml_node_struct* node)
+	{
+		child->parent = node;
+
+		xml_node_struct* head = node->first_child;
+
+		if (head)
+		{
+			child->prev_sibling_c = head->prev_sibling_c;
+			head->prev_sibling_c = child;
+		}
+		else
+			child->prev_sibling_c = child;
+
+		child->next_sibling = head;
+		node->first_child = child;
+	}
+
+	inline void insert_node_after(xml_node_struct* child, xml_node_struct* node)
+	{
+		xml_node_struct* parent = node->parent;
+
+		child->parent = parent;
+
+		if (node->next_sibling)
+			node->next_sibling->prev_sibling_c = child;
+		else
+			parent->first_child->prev_sibling_c = child;
+
+		child->next_sibling = node->next_sibling;
+		child->prev_sibling_c = node;
+
+		node->next_sibling = child;
+	}
+
+	inline void insert_node_before(xml_node_struct* child, xml_node_struct* node)
+	{
+		xml_node_struct* parent = node->parent;
+
+		child->parent = parent;
+
+		if (node->prev_sibling_c->next_sibling)
+			node->prev_sibling_c->next_sibling = child;
+		else
+			parent->first_child = child;
+
+		child->prev_sibling_c = node->prev_sibling_c;
+		child->next_sibling = node;
+
+		node->prev_sibling_c = child;
+	}
+
+	inline void remove_node(xml_node_struct* node)
+	{
+		xml_node_struct* parent = node->parent;
+
+		if (node->next_sibling)
+			node->next_sibling->prev_sibling_c = node->prev_sibling_c;
+		else
+			parent->first_child->prev_sibling_c = node->prev_sibling_c;
+
+		if (node->prev_sibling_c->next_sibling)
+			node->prev_sibling_c->next_sibling = node->next_sibling;
+		else
+			parent->first_child = node->next_sibling;
+
+		node->parent = 0;
+		node->prev_sibling_c = 0;
+		node->next_sibling = 0;
+	}
+
+	inline void append_attribute(xml_attribute_struct* attr, xml_node_struct* node)
+	{
+		xml_attribute_struct* head = node->first_attribute;
+
+		if (head)
+		{
+			xml_attribute_struct* tail = head->prev_attribute_c;
+
+			tail->next_attribute = attr;
+			attr->prev_attribute_c = tail;
+			head->prev_attribute_c = attr;
+		}
+		else
+		{
+			node->first_attribute = attr;
+			attr->prev_attribute_c = attr;
+		}
+	}
+
+	inline void prepend_attribute(xml_attribute_struct* attr, xml_node_struct* node)
+	{
+		xml_attribute_struct* head = node->first_attribute;
+
+		if (head)
+		{
+			attr->prev_attribute_c = head->prev_attribute_c;
+			head->prev_attribute_c = attr;
+		}
+		else
+			attr->prev_attribute_c = attr;
+
+		attr->next_attribute = head;
+		node->first_attribute = attr;
+	}
+
+	inline void insert_attribute_after(xml_attribute_struct* attr, xml_attribute_struct* place, xml_node_struct* node)
+	{
+		if (place->next_attribute)
+			place->next_attribute->prev_attribute_c = attr;
+		else
+			node->first_attribute->prev_attribute_c = attr;
+
+		attr->next_attribute = place->next_attribute;
+		attr->prev_attribute_c = place;
+		place->next_attribute = attr;
+	}
+
+	inline void insert_attribute_before(xml_attribute_struct* attr, xml_attribute_struct* place, xml_node_struct* node)
+	{
+		if (place->prev_attribute_c->next_attribute)
+			place->prev_attribute_c->next_attribute = attr;
+		else
+			node->first_attribute = attr;
+
+		attr->prev_attribute_c = place->prev_attribute_c;
+		attr->next_attribute = place;
+		place->prev_attribute_c = attr;
+	}
+
+	inline void remove_attribute(xml_attribute_struct* attr, xml_node_struct* node)
+	{
+		if (attr->next_attribute)
+			attr->next_attribute->prev_attribute_c = attr->prev_attribute_c;
+		else
+			node->first_attribute->prev_attribute_c = attr->prev_attribute_c;
+
+		if (attr->prev_attribute_c->next_attribute)
+			attr->prev_attribute_c->next_attribute = attr->next_attribute;
+		else
+			node->first_attribute = attr->next_attribute;
+
+		attr->prev_attribute_c = 0;
+		attr->next_attribute = 0;
+	}
+
+	PUGI__FN_NO_INLINE xml_node_struct* append_new_node(xml_node_struct* node, xml_allocator& alloc, xml_node_type type = node_element)
+	{
+		if (!alloc.reserve()) return 0;
+
+		xml_node_struct* child = allocate_node(alloc, type);
+		if (!child) return 0;
+
+		append_node(child, node);
+
+		return child;
+	}
+
+	PUGI__FN_NO_INLINE xml_attribute_struct* append_new_attribute(xml_node_struct* node, xml_allocator& alloc)
+	{
+		if (!alloc.reserve()) return 0;
+
+		xml_attribute_struct* attr = allocate_attribute(alloc);
+		if (!attr) return 0;
+
+		append_attribute(attr, node);
+
+		return attr;
+	}
+PUGI__NS_END
+
+// Helper classes for code generation
+PUGI__NS_BEGIN
+	struct opt_false
+	{
+		enum { value = 0 };
+	};
+
+	struct opt_true
+	{
+		enum { value = 1 };
+	};
+PUGI__NS_END
+
+// Unicode utilities
+PUGI__NS_BEGIN
+	inline uint16_t endian_swap(uint16_t value)
+	{
+		return static_cast<uint16_t>(((value & 0xff) << 8) | (value >> 8));
+	}
+
+	inline uint32_t endian_swap(uint32_t value)
+	{
+		return ((value & 0xff) << 24) | ((value & 0xff00) << 8) | ((value & 0xff0000) >> 8) | (value >> 24);
+	}
+
+	struct utf8_counter
+	{
+		typedef size_t value_type;
+
+		static value_type low(value_type result, uint32_t ch)
+		{
+			// U+0000..U+007F
+			if (ch < 0x80) return result + 1;
+			// U+0080..U+07FF
+			else if (ch < 0x800) return result + 2;
+			// U+0800..U+FFFF
+			else return result + 3;
+		}
+
+		static value_type high(value_type result, uint32_t)
+		{
+			// U+10000..U+10FFFF
+			return result + 4;
+		}
+	};
+
+	struct utf8_writer
+	{
+		typedef uint8_t* value_type;
+
+		static value_type low(value_type result, uint32_t ch)
+		{
+			// U+0000..U+007F
+			if (ch < 0x80)
+			{
+				*result = static_cast<uint8_t>(ch);
+				return result + 1;
+			}
+			// U+0080..U+07FF
+			else if (ch < 0x800)
+			{
+				result[0] = static_cast<uint8_t>(0xC0 | (ch >> 6));
+				result[1] = static_cast<uint8_t>(0x80 | (ch & 0x3F));
+				return result + 2;
+			}
+			// U+0800..U+FFFF
+			else
+			{
+				result[0] = static_cast<uint8_t>(0xE0 | (ch >> 12));
+				result[1] = static_cast<uint8_t>(0x80 | ((ch >> 6) & 0x3F));
+				result[2] = static_cast<uint8_t>(0x80 | (ch & 0x3F));
+				return result + 3;
+			}
+		}
+
+		static value_type high(value_type result, uint32_t ch)
+		{
+			// U+10000..U+10FFFF
+			result[0] = static_cast<uint8_t>(0xF0 | (ch >> 18));
+			result[1] = static_cast<uint8_t>(0x80 | ((ch >> 12) & 0x3F));
+			result[2] = static_cast<uint8_t>(0x80 | ((ch >> 6) & 0x3F));
+			result[3] = static_cast<uint8_t>(0x80 | (ch & 0x3F));
+			return result + 4;
+		}
+
+		static value_type any(value_type result, uint32_t ch)
+		{
+			return (ch < 0x10000) ? low(result, ch) : high(result, ch);
+		}
+	};
+
+	struct utf16_counter
+	{
+		typedef size_t value_type;
+
+		static value_type low(value_type result, uint32_t)
+		{
+			return result + 1;
+		}
+
+		static value_type high(value_type result, uint32_t)
+		{
+			return result + 2;
+		}
+	};
+
+	struct utf16_writer
+	{
+		typedef uint16_t* value_type;
+
+		static value_type low(value_type result, uint32_t ch)
+		{
+			*result = static_cast<uint16_t>(ch);
+
+			return result + 1;
+		}
+
+		static value_type high(value_type result, uint32_t ch)
+		{
+			uint32_t msh = static_cast<uint32_t>(ch - 0x10000) >> 10;
+			uint32_t lsh = static_cast<uint32_t>(ch - 0x10000) & 0x3ff;
+
+			result[0] = static_cast<uint16_t>(0xD800 + msh);
+			result[1] = static_cast<uint16_t>(0xDC00 + lsh);
+
+			return result + 2;
+		}
+
+		static value_type any(value_type result, uint32_t ch)
+		{
+			return (ch < 0x10000) ? low(result, ch) : high(result, ch);
+		}
+	};
+
+	struct utf32_counter
+	{
+		typedef size_t value_type;
+
+		static value_type low(value_type result, uint32_t)
+		{
+			return result + 1;
+		}
+
+		static value_type high(value_type result, uint32_t)
+		{
+			return result + 1;
+		}
+	};
+
+	struct utf32_writer
+	{
+		typedef uint32_t* value_type;
+
+		static value_type low(value_type result, uint32_t ch)
+		{
+			*result = ch;
+
+			return result + 1;
+		}
+
+		static value_type high(value_type result, uint32_t ch)
+		{
+			*result = ch;
+
+			return result + 1;
+		}
+
+		static value_type any(value_type result, uint32_t ch)
+		{
+			*result = ch;
+
+			return result + 1;
+		}
+	};
+
+	struct latin1_writer
+	{
+		typedef uint8_t* value_type;
+
+		static value_type low(value_type result, uint32_t ch)
+		{
+			*result = static_cast<uint8_t>(ch > 255 ? '?' : ch);
+
+			return result + 1;
+		}
+
+		static value_type high(value_type result, uint32_t ch)
+		{
+			(void)ch;
+
+			*result = '?';
+
+			return result + 1;
+		}
+	};
+
+	template <size_t size> struct wchar_selector;
+
+	template <> struct wchar_selector<2>
+	{
+		typedef uint16_t type;
+		typedef utf16_counter counter;
+		typedef utf16_writer writer;
+	};
+
+	template <> struct wchar_selector<4>
+	{
+		typedef uint32_t type;
+		typedef utf32_counter counter;
+		typedef utf32_writer writer;
+	};
+
+	typedef wchar_selector<sizeof(wchar_t)>::counter wchar_counter;
+	typedef wchar_selector<sizeof(wchar_t)>::writer wchar_writer;
+
+	template <typename Traits, typename opt_swap = opt_false> struct utf_decoder
+	{
+		static inline typename Traits::value_type decode_utf8_block(const uint8_t* data, size_t size, typename Traits::value_type result)
+		{
+			const uint8_t utf8_byte_mask = 0x3f;
+
+			while (size)
+			{
+				uint8_t lead = *data;
+
+				// 0xxxxxxx -> U+0000..U+007F
+				if (lead < 0x80)
+				{
+					result = Traits::low(result, lead);
+					data += 1;
+					size -= 1;
+
+					// process aligned single-byte (ascii) blocks
+					if ((reinterpret_cast<uintptr_t>(data) & 3) == 0)
+					{
+						// round-trip through void* to silence 'cast increases required alignment of target type' warnings
+						while (size >= 4 && (*static_cast<const uint32_t*>(static_cast<const void*>(data)) & 0x80808080) == 0)
+						{
+							result = Traits::low(result, data[0]);
+							result = Traits::low(result, data[1]);
+							result = Traits::low(result, data[2]);
+							result = Traits::low(result, data[3]);
+							data += 4;
+							size -= 4;
+						}
+					}
+				}
+				// 110xxxxx -> U+0080..U+07FF
+				else if (static_cast<unsigned int>(lead - 0xC0) < 0x20 && size >= 2 && (data[1] & 0xc0) == 0x80)
+				{
+					result = Traits::low(result, ((lead & ~0xC0) << 6) | (data[1] & utf8_byte_mask));
+					data += 2;
+					size -= 2;
+				}
+				// 1110xxxx -> U+0800-U+FFFF
+				else if (static_cast<unsigned int>(lead - 0xE0) < 0x10 && size >= 3 && (data[1] & 0xc0) == 0x80 && (data[2] & 0xc0) == 0x80)
+				{
+					result = Traits::low(result, ((lead & ~0xE0) << 12) | ((data[1] & utf8_byte_mask) << 6) | (data[2] & utf8_byte_mask));
+					data += 3;
+					size -= 3;
+				}
+				// 11110xxx -> U+10000..U+10FFFF
+				else if (static_cast<unsigned int>(lead - 0xF0) < 0x08 && size >= 4 && (data[1] & 0xc0) == 0x80 && (data[2] & 0xc0) == 0x80 && (data[3] & 0xc0) == 0x80)
+				{
+					result = Traits::high(result, ((lead & ~0xF0) << 18) | ((data[1] & utf8_byte_mask) << 12) | ((data[2] & utf8_byte_mask) << 6) | (data[3] & utf8_byte_mask));
+					data += 4;
+					size -= 4;
+				}
+				// 10xxxxxx or 11111xxx -> invalid
+				else
+				{
+					data += 1;
+					size -= 1;
+				}
+			}
+
+			return result;
+		}
+
+		static inline typename Traits::value_type decode_utf16_block(const uint16_t* data, size_t size, typename Traits::value_type result)
+		{
+			const uint16_t* end = data + size;
+
+			while (data < end)
+			{
+				unsigned int lead = opt_swap::value ? endian_swap(*data) : *data;
+
+				// U+0000..U+D7FF
+				if (lead < 0xD800)
+				{
+					result = Traits::low(result, lead);
+					data += 1;
+				}
+				// U+E000..U+FFFF
+				else if (static_cast<unsigned int>(lead - 0xE000) < 0x2000)
+				{
+					result = Traits::low(result, lead);
+					data += 1;
+				}
+				// surrogate pair lead
+				else if (static_cast<unsigned int>(lead - 0xD800) < 0x400 && data + 1 < end)
+				{
+					uint16_t next = opt_swap::value ? endian_swap(data[1]) : data[1];
+
+					if (static_cast<unsigned int>(next - 0xDC00) < 0x400)
+					{
+						result = Traits::high(result, 0x10000 + ((lead & 0x3ff) << 10) + (next & 0x3ff));
+						data += 2;
+					}
+					else
+					{
+						data += 1;
+					}
+				}
+				else
+				{
+					data += 1;
+				}
+			}
+
+			return result;
+		}
+
+		static inline typename Traits::value_type decode_utf32_block(const uint32_t* data, size_t size, typename Traits::value_type result)
+		{
+			const uint32_t* end = data + size;
+
+			while (data < end)
+			{
+				uint32_t lead = opt_swap::value ? endian_swap(*data) : *data;
+
+				// U+0000..U+FFFF
+				if (lead < 0x10000)
+				{
+					result = Traits::low(result, lead);
+					data += 1;
+				}
+				// U+10000..U+10FFFF
+				else
+				{
+					result = Traits::high(result, lead);
+					data += 1;
+				}
+			}
+
+			return result;
+		}
+
+		static inline typename Traits::value_type decode_latin1_block(const uint8_t* data, size_t size, typename Traits::value_type result)
+		{
+			for (size_t i = 0; i < size; ++i)
+			{
+				result = Traits::low(result, data[i]);
+			}
+
+			return result;
+		}
+
+		static inline typename Traits::value_type decode_wchar_block_impl(const uint16_t* data, size_t size, typename Traits::value_type result)
+		{
+			return decode_utf16_block(data, size, result);
+		}
+
+		static inline typename Traits::value_type decode_wchar_block_impl(const uint32_t* data, size_t size, typename Traits::value_type result)
+		{
+			return decode_utf32_block(data, size, result);
+		}
+
+		static inline typename Traits::value_type decode_wchar_block(const wchar_t* data, size_t size, typename Traits::value_type result)
+		{
+			return decode_wchar_block_impl(reinterpret_cast<const wchar_selector<sizeof(wchar_t)>::type*>(data), size, result);
+		}
+	};
+
+	template <typename T> PUGI__FN void convert_utf_endian_swap(T* result, const T* data, size_t length)
+	{
+		for (size_t i = 0; i < length; ++i) result[i] = endian_swap(data[i]);
+	}
+
+#ifdef PUGIXML_WCHAR_MODE
+	PUGI__FN void convert_wchar_endian_swap(wchar_t* result, const wchar_t* data, size_t length)
+	{
+		for (size_t i = 0; i < length; ++i) result[i] = static_cast<wchar_t>(endian_swap(static_cast<wchar_selector<sizeof(wchar_t)>::type>(data[i])));
+	}
+#endif
+PUGI__NS_END
+
+PUGI__NS_BEGIN
+	enum chartype_t
+	{
+		ct_parse_pcdata = 1,	// \0, &, \r, <
+		ct_parse_attr = 2,		// \0, &, \r, ', "
+		ct_parse_attr_ws = 4,	// \0, &, \r, ', ", \n, tab
+		ct_space = 8,			// \r, \n, space, tab
+		ct_parse_cdata = 16,	// \0, ], >, \r
+		ct_parse_comment = 32,	// \0, -, >, \r
+		ct_symbol = 64,			// Any symbol > 127, a-z, A-Z, 0-9, _, :, -, .
+		ct_start_symbol = 128	// Any symbol > 127, a-z, A-Z, _, :
+	};
+
+	static const unsigned char chartype_table[256] =
+	{
+		55,  0,   0,   0,   0,   0,   0,   0,      0,   12,  12,  0,   0,   63,  0,   0,   // 0-15
+		0,   0,   0,   0,   0,   0,   0,   0,      0,   0,   0,   0,   0,   0,   0,   0,   // 16-31
+		8,   0,   6,   0,   0,   0,   7,   6,      0,   0,   0,   0,   0,   96,  64,  0,   // 32-47
+		64,  64,  64,  64,  64,  64,  64,  64,     64,  64,  192, 0,   1,   0,   48,  0,   // 48-63
+		0,   192, 192, 192, 192, 192, 192, 192,    192, 192, 192, 192, 192, 192, 192, 192, // 64-79
+		192, 192, 192, 192, 192, 192, 192, 192,    192, 192, 192, 0,   0,   16,  0,   192, // 80-95
+		0,   192, 192, 192, 192, 192, 192, 192,    192, 192, 192, 192, 192, 192, 192, 192, // 96-111
+		192, 192, 192, 192, 192, 192, 192, 192,    192, 192, 192, 0, 0, 0, 0, 0,           // 112-127
+
+		192, 192, 192, 192, 192, 192, 192, 192,    192, 192, 192, 192, 192, 192, 192, 192, // 128+
+		192, 192, 192, 192, 192, 192, 192, 192,    192, 192, 192, 192, 192, 192, 192, 192,
+		192, 192, 192, 192, 192, 192, 192, 192,    192, 192, 192, 192, 192, 192, 192, 192,
+		192, 192, 192, 192, 192, 192, 192, 192,    192, 192, 192, 192, 192, 192, 192, 192,
+		192, 192, 192, 192, 192, 192, 192, 192,    192, 192, 192, 192, 192, 192, 192, 192,
+		192, 192, 192, 192, 192, 192, 192, 192,    192, 192, 192, 192, 192, 192, 192, 192,
+		192, 192, 192, 192, 192, 192, 192, 192,    192, 192, 192, 192, 192, 192, 192, 192,
+		192, 192, 192, 192, 192, 192, 192, 192,    192, 192, 192, 192, 192, 192, 192, 192
+	};
+
+	enum chartypex_t
+	{
+		ctx_special_pcdata = 1,   // Any symbol >= 0 and < 32 (except \t, \r, \n), &, <, >
+		ctx_special_attr = 2,     // Any symbol >= 0 and < 32 (except \t), &, <, >, "
+		ctx_start_symbol = 4,	  // Any symbol > 127, a-z, A-Z, _
+		ctx_digit = 8,			  // 0-9
+		ctx_symbol = 16			  // Any symbol > 127, a-z, A-Z, 0-9, _, -, .
+	};
+	
+	static const unsigned char chartypex_table[256] =
+	{
+		3,  3,  3,  3,  3,  3,  3,  3,     3,  0,  2,  3,  3,  2,  3,  3,     // 0-15
+		3,  3,  3,  3,  3,  3,  3,  3,     3,  3,  3,  3,  3,  3,  3,  3,     // 16-31
+		0,  0,  2,  0,  0,  0,  3,  0,     0,  0,  0,  0,  0, 16, 16,  0,     // 32-47
+		24, 24, 24, 24, 24, 24, 24, 24,    24, 24, 0,  0,  3,  0,  3,  0,     // 48-63
+
+		0,  20, 20, 20, 20, 20, 20, 20,    20, 20, 20, 20, 20, 20, 20, 20,    // 64-79
+		20, 20, 20, 20, 20, 20, 20, 20,    20, 20, 20, 0,  0,  0,  0,  20,    // 80-95
+		0,  20, 20, 20, 20, 20, 20, 20,    20, 20, 20, 20, 20, 20, 20, 20,    // 96-111
+		20, 20, 20, 20, 20, 20, 20, 20,    20, 20, 20, 0,  0,  0,  0,  0,     // 112-127
+
+		20, 20, 20, 20, 20, 20, 20, 20,    20, 20, 20, 20, 20, 20, 20, 20,    // 128+
+		20, 20, 20, 20, 20, 20, 20, 20,    20, 20, 20, 20, 20, 20, 20, 20,
+		20, 20, 20, 20, 20, 20, 20, 20,    20, 20, 20, 20, 20, 20, 20, 20,
+		20, 20, 20, 20, 20, 20, 20, 20,    20, 20, 20, 20, 20, 20, 20, 20,
+		20, 20, 20, 20, 20, 20, 20, 20,    20, 20, 20, 20, 20, 20, 20, 20,
+		20, 20, 20, 20, 20, 20, 20, 20,    20, 20, 20, 20, 20, 20, 20, 20,
+		20, 20, 20, 20, 20, 20, 20, 20,    20, 20, 20, 20, 20, 20, 20, 20,
+		20, 20, 20, 20, 20, 20, 20, 20,    20, 20, 20, 20, 20, 20, 20, 20
+	};
+	
+#ifdef PUGIXML_WCHAR_MODE
+	#define PUGI__IS_CHARTYPE_IMPL(c, ct, table) ((static_cast<unsigned int>(c) < 128 ? table[static_cast<unsigned int>(c)] : table[128]) & (ct))
+#else
+	#define PUGI__IS_CHARTYPE_IMPL(c, ct, table) (table[static_cast<unsigned char>(c)] & (ct))
+#endif
+
+	#define PUGI__IS_CHARTYPE(c, ct) PUGI__IS_CHARTYPE_IMPL(c, ct, chartype_table)
+	#define PUGI__IS_CHARTYPEX(c, ct) PUGI__IS_CHARTYPE_IMPL(c, ct, chartypex_table)
+
+	PUGI__FN bool is_little_endian()
+	{
+		unsigned int ui = 1;
+
+		return *reinterpret_cast<unsigned char*>(&ui) == 1;
+	}
+
+	PUGI__FN xml_encoding get_wchar_encoding()
+	{
+		PUGI__STATIC_ASSERT(sizeof(wchar_t) == 2 || sizeof(wchar_t) == 4);
+
+		if (sizeof(wchar_t) == 2)
+			return is_little_endian() ? encoding_utf16_le : encoding_utf16_be;
+		else 
+			return is_little_endian() ? encoding_utf32_le : encoding_utf32_be;
+	}
+
+	PUGI__FN xml_encoding guess_buffer_encoding(uint8_t d0, uint8_t d1, uint8_t d2, uint8_t d3)
+	{
+		// look for BOM in first few bytes
+		if (d0 == 0 && d1 == 0 && d2 == 0xfe && d3 == 0xff) return encoding_utf32_be;
+		if (d0 == 0xff && d1 == 0xfe && d2 == 0 && d3 == 0) return encoding_utf32_le;
+		if (d0 == 0xfe && d1 == 0xff) return encoding_utf16_be;
+		if (d0 == 0xff && d1 == 0xfe) return encoding_utf16_le;
+		if (d0 == 0xef && d1 == 0xbb && d2 == 0xbf) return encoding_utf8;
+
+		// look for <, <? or <?xm in various encodings
+		if (d0 == 0 && d1 == 0 && d2 == 0 && d3 == 0x3c) return encoding_utf32_be;
+		if (d0 == 0x3c && d1 == 0 && d2 == 0 && d3 == 0) return encoding_utf32_le;
+		if (d0 == 0 && d1 == 0x3c && d2 == 0 && d3 == 0x3f) return encoding_utf16_be;
+		if (d0 == 0x3c && d1 == 0 && d2 == 0x3f && d3 == 0) return encoding_utf16_le;
+		if (d0 == 0x3c && d1 == 0x3f && d2 == 0x78 && d3 == 0x6d) return encoding_utf8;
+
+		// look for utf16 < followed by node name (this may fail, but is better than utf8 since it's zero terminated so early)
+		if (d0 == 0 && d1 == 0x3c) return encoding_utf16_be;
+		if (d0 == 0x3c && d1 == 0) return encoding_utf16_le;
+
+		// no known BOM detected, assume utf8
+		return encoding_utf8;
+	}
+
+	PUGI__FN xml_encoding get_buffer_encoding(xml_encoding encoding, const void* contents, size_t size)
+	{
+		// replace wchar encoding with utf implementation
+		if (encoding == encoding_wchar) return get_wchar_encoding();
+
+		// replace utf16 encoding with utf16 with specific endianness
+		if (encoding == encoding_utf16) return is_little_endian() ? encoding_utf16_le : encoding_utf16_be;
+
+		// replace utf32 encoding with utf32 with specific endianness
+		if (encoding == encoding_utf32) return is_little_endian() ? encoding_utf32_le : encoding_utf32_be;
+
+		// only do autodetection if no explicit encoding is requested
+		if (encoding != encoding_auto) return encoding;
+
+		// skip encoding autodetection if input buffer is too small
+		if (size < 4) return encoding_utf8;
+
+		// try to guess encoding (based on XML specification, Appendix F.1)
+		const uint8_t* data = static_cast<const uint8_t*>(contents);
+
+		PUGI__DMC_VOLATILE uint8_t d0 = data[0], d1 = data[1], d2 = data[2], d3 = data[3];
+
+		return guess_buffer_encoding(d0, d1, d2, d3);
+	}
+
+	PUGI__FN bool get_mutable_buffer(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, bool is_mutable)
+	{
+		size_t length = size / sizeof(char_t);
+
+		if (is_mutable)
+		{
+			out_buffer = static_cast<char_t*>(const_cast<void*>(contents));
+			out_length = length;
+		}
+		else
+		{
+			char_t* buffer = static_cast<char_t*>(xml_memory::allocate((length + 1) * sizeof(char_t)));
+			if (!buffer) return false;
+
+			if (contents)
+				memcpy(buffer, contents, length * sizeof(char_t));
+			else
+				assert(length == 0);
+
+			buffer[length] = 0;
+
+			out_buffer = buffer;
+			out_length = length + 1;
+		}
+
+		return true;
+	}
+
+#ifdef PUGIXML_WCHAR_MODE
+	PUGI__FN bool need_endian_swap_utf(xml_encoding le, xml_encoding re)
+	{
+		return (le == encoding_utf16_be && re == encoding_utf16_le) || (le == encoding_utf16_le && re == encoding_utf16_be) ||
+			   (le == encoding_utf32_be && re == encoding_utf32_le) || (le == encoding_utf32_le && re == encoding_utf32_be);
+	}
+
+	PUGI__FN bool convert_buffer_endian_swap(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, bool is_mutable)
+	{
+		const char_t* data = static_cast<const char_t*>(contents);
+		size_t length = size / sizeof(char_t);
+
+		if (is_mutable)
+		{
+			char_t* buffer = const_cast<char_t*>(data);
+
+			convert_wchar_endian_swap(buffer, data, length);
+
+			out_buffer = buffer;
+			out_length = length;
+		}
+		else
+		{
+			char_t* buffer = static_cast<char_t*>(xml_memory::allocate((length + 1) * sizeof(char_t)));
+			if (!buffer) return false;
+
+			convert_wchar_endian_swap(buffer, data, length);
+			buffer[length] = 0;
+
+			out_buffer = buffer;
+			out_length = length + 1;
+		}
+
+		return true;
+	}
+
+	PUGI__FN bool convert_buffer_utf8(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size)
+	{
+		const uint8_t* data = static_cast<const uint8_t*>(contents);
+		size_t data_length = size;
+
+		// first pass: get length in wchar_t units
+		size_t length = utf_decoder<wchar_counter>::decode_utf8_block(data, data_length, 0);
+
+		// allocate buffer of suitable length
+		char_t* buffer = static_cast<char_t*>(xml_memory::allocate((length + 1) * sizeof(char_t)));
+		if (!buffer) return false;
+
+		// second pass: convert utf8 input to wchar_t
+		wchar_writer::value_type obegin = reinterpret_cast<wchar_writer::value_type>(buffer);
+		wchar_writer::value_type oend = utf_decoder<wchar_writer>::decode_utf8_block(data, data_length, obegin);
+
+		assert(oend == obegin + length);
+		*oend = 0;
+
+		out_buffer = buffer;
+		out_length = length + 1;
+
+		return true;
+	}
+
+	template <typename opt_swap> PUGI__FN bool convert_buffer_utf16(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, opt_swap)
+	{
+		const uint16_t* data = static_cast<const uint16_t*>(contents);
+		size_t data_length = size / sizeof(uint16_t);
+
+		// first pass: get length in wchar_t units
+		size_t length = utf_decoder<wchar_counter, opt_swap>::decode_utf16_block(data, data_length, 0);
+
+		// allocate buffer of suitable length
+		char_t* buffer = static_cast<char_t*>(xml_memory::allocate((length + 1) * sizeof(char_t)));
+		if (!buffer) return false;
+
+		// second pass: convert utf16 input to wchar_t
+		wchar_writer::value_type obegin = reinterpret_cast<wchar_writer::value_type>(buffer);
+		wchar_writer::value_type oend = utf_decoder<wchar_writer, opt_swap>::decode_utf16_block(data, data_length, obegin);
+
+		assert(oend == obegin + length);
+		*oend = 0;
+
+		out_buffer = buffer;
+		out_length = length + 1;
+
+		return true;
+	}
+
+	template <typename opt_swap> PUGI__FN bool convert_buffer_utf32(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, opt_swap)
+	{
+		const uint32_t* data = static_cast<const uint32_t*>(contents);
+		size_t data_length = size / sizeof(uint32_t);
+
+		// first pass: get length in wchar_t units
+		size_t length = utf_decoder<wchar_counter, opt_swap>::decode_utf32_block(data, data_length, 0);
+
+		// allocate buffer of suitable length
+		char_t* buffer = static_cast<char_t*>(xml_memory::allocate((length + 1) * sizeof(char_t)));
+		if (!buffer) return false;
+
+		// second pass: convert utf32 input to wchar_t
+		wchar_writer::value_type obegin = reinterpret_cast<wchar_writer::value_type>(buffer);
+		wchar_writer::value_type oend = utf_decoder<wchar_writer, opt_swap>::decode_utf32_block(data, data_length, obegin);
+
+		assert(oend == obegin + length);
+		*oend = 0;
+
+		out_buffer = buffer;
+		out_length = length + 1;
+
+		return true;
+	}
+
+	PUGI__FN bool convert_buffer_latin1(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size)
+	{
+		const uint8_t* data = static_cast<const uint8_t*>(contents);
+		size_t data_length = size;
+
+		// get length in wchar_t units
+		size_t length = data_length;
+
+		// allocate buffer of suitable length
+		char_t* buffer = static_cast<char_t*>(xml_memory::allocate((length + 1) * sizeof(char_t)));
+		if (!buffer) return false;
+
+		// convert latin1 input to wchar_t
+		wchar_writer::value_type obegin = reinterpret_cast<wchar_writer::value_type>(buffer);
+		wchar_writer::value_type oend = utf_decoder<wchar_writer>::decode_latin1_block(data, data_length, obegin);
+
+		assert(oend == obegin + length);
+		*oend = 0;
+
+		out_buffer = buffer;
+		out_length = length + 1;
+
+		return true;
+	}
+
+	PUGI__FN bool convert_buffer(char_t*& out_buffer, size_t& out_length, xml_encoding encoding, const void* contents, size_t size, bool is_mutable)
+	{
+		// get native encoding
+		xml_encoding wchar_encoding = get_wchar_encoding();
+
+		// fast path: no conversion required
+		if (encoding == wchar_encoding) return get_mutable_buffer(out_buffer, out_length, contents, size, is_mutable);
+
+		// only endian-swapping is required
+		if (need_endian_swap_utf(encoding, wchar_encoding)) return convert_buffer_endian_swap(out_buffer, out_length, contents, size, is_mutable);
+
+		// source encoding is utf8
+		if (encoding == encoding_utf8) return convert_buffer_utf8(out_buffer, out_length, contents, size);
+
+		// source encoding is utf16
+		if (encoding == encoding_utf16_be || encoding == encoding_utf16_le)
+		{
+			xml_encoding native_encoding = is_little_endian() ? encoding_utf16_le : encoding_utf16_be;
+
+			return (native_encoding == encoding) ?
+				convert_buffer_utf16(out_buffer, out_length, contents, size, opt_false()) :
+				convert_buffer_utf16(out_buffer, out_length, contents, size, opt_true());
+		}
+
+		// source encoding is utf32
+		if (encoding == encoding_utf32_be || encoding == encoding_utf32_le)
+		{
+			xml_encoding native_encoding = is_little_endian() ? encoding_utf32_le : encoding_utf32_be;
+
+			return (native_encoding == encoding) ?
+				convert_buffer_utf32(out_buffer, out_length, contents, size, opt_false()) :
+				convert_buffer_utf32(out_buffer, out_length, contents, size, opt_true());
+		}
+
+		// source encoding is latin1
+		if (encoding == encoding_latin1) return convert_buffer_latin1(out_buffer, out_length, contents, size);
+
+		assert(!"Invalid encoding");
+		return false;
+	}
+#else
+	template <typename opt_swap> PUGI__FN bool convert_buffer_utf16(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, opt_swap)
+	{
+		const uint16_t* data = static_cast<const uint16_t*>(contents);
+		size_t data_length = size / sizeof(uint16_t);
+
+		// first pass: get length in utf8 units
+		size_t length = utf_decoder<utf8_counter, opt_swap>::decode_utf16_block(data, data_length, 0);
+
+		// allocate buffer of suitable length
+		char_t* buffer = static_cast<char_t*>(xml_memory::allocate((length + 1) * sizeof(char_t)));
+		if (!buffer) return false;
+
+		// second pass: convert utf16 input to utf8
+		uint8_t* obegin = reinterpret_cast<uint8_t*>(buffer);
+		uint8_t* oend = utf_decoder<utf8_writer, opt_swap>::decode_utf16_block(data, data_length, obegin);
+
+		assert(oend == obegin + length);
+		*oend = 0;
+
+		out_buffer = buffer;
+		out_length = length + 1;
+
+		return true;
+	}
+
+	template <typename opt_swap> PUGI__FN bool convert_buffer_utf32(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, opt_swap)
+	{
+		const uint32_t* data = static_cast<const uint32_t*>(contents);
+		size_t data_length = size / sizeof(uint32_t);
+
+		// first pass: get length in utf8 units
+		size_t length = utf_decoder<utf8_counter, opt_swap>::decode_utf32_block(data, data_length, 0);
+
+		// allocate buffer of suitable length
+		char_t* buffer = static_cast<char_t*>(xml_memory::allocate((length + 1) * sizeof(char_t)));
+		if (!buffer) return false;
+
+		// second pass: convert utf32 input to utf8
+		uint8_t* obegin = reinterpret_cast<uint8_t*>(buffer);
+		uint8_t* oend = utf_decoder<utf8_writer, opt_swap>::decode_utf32_block(data, data_length, obegin);
+
+		assert(oend == obegin + length);
+		*oend = 0;
+
+		out_buffer = buffer;
+		out_length = length + 1;
+
+		return true;
+	}
+
+	PUGI__FN size_t get_latin1_7bit_prefix_length(const uint8_t* data, size_t size)
+	{
+		for (size_t i = 0; i < size; ++i)
+			if (data[i] > 127)
+				return i;
+
+		return size;
+	}
+
+	PUGI__FN bool convert_buffer_latin1(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, bool is_mutable)
+	{
+		const uint8_t* data = static_cast<const uint8_t*>(contents);
+		size_t data_length = size;
+
+		// get size of prefix that does not need utf8 conversion
+		size_t prefix_length = get_latin1_7bit_prefix_length(data, data_length);
+		assert(prefix_length <= data_length);
+
+		const uint8_t* postfix = data + prefix_length;
+		size_t postfix_length = data_length - prefix_length;
+
+		// if no conversion is needed, just return the original buffer
+		if (postfix_length == 0) return get_mutable_buffer(out_buffer, out_length, contents, size, is_mutable);
+
+		// first pass: get length in utf8 units
+		size_t length = prefix_length + utf_decoder<utf8_counter>::decode_latin1_block(postfix, postfix_length, 0);
+
+		// allocate buffer of suitable length
+		char_t* buffer = static_cast<char_t*>(xml_memory::allocate((length + 1) * sizeof(char_t)));
+		if (!buffer) return false;
+
+		// second pass: convert latin1 input to utf8
+		memcpy(buffer, data, prefix_length);
+
+		uint8_t* obegin = reinterpret_cast<uint8_t*>(buffer);
+		uint8_t* oend = utf_decoder<utf8_writer>::decode_latin1_block(postfix, postfix_length, obegin + prefix_length);
+
+		assert(oend == obegin + length);
+		*oend = 0;
+
+		out_buffer = buffer;
+		out_length = length + 1;
+
+		return true;
+	}
+
+	PUGI__FN bool convert_buffer(char_t*& out_buffer, size_t& out_length, xml_encoding encoding, const void* contents, size_t size, bool is_mutable)
+	{
+		// fast path: no conversion required
+		if (encoding == encoding_utf8) return get_mutable_buffer(out_buffer, out_length, contents, size, is_mutable);
+
+		// source encoding is utf16
+		if (encoding == encoding_utf16_be || encoding == encoding_utf16_le)
+		{
+			xml_encoding native_encoding = is_little_endian() ? encoding_utf16_le : encoding_utf16_be;
+
+			return (native_encoding == encoding) ?
+				convert_buffer_utf16(out_buffer, out_length, contents, size, opt_false()) :
+				convert_buffer_utf16(out_buffer, out_length, contents, size, opt_true());
+		}
+
+		// source encoding is utf32
+		if (encoding == encoding_utf32_be || encoding == encoding_utf32_le)
+		{
+			xml_encoding native_encoding = is_little_endian() ? encoding_utf32_le : encoding_utf32_be;
+
+			return (native_encoding == encoding) ?
+				convert_buffer_utf32(out_buffer, out_length, contents, size, opt_false()) :
+				convert_buffer_utf32(out_buffer, out_length, contents, size, opt_true());
+		}
+
+		// source encoding is latin1
+		if (encoding == encoding_latin1) return convert_buffer_latin1(out_buffer, out_length, contents, size, is_mutable);
+
+		assert(!"Invalid encoding");
+		return false;
+	}
+#endif
+
+	PUGI__FN size_t as_utf8_begin(const wchar_t* str, size_t length)
+	{
+		// get length in utf8 characters
+		return utf_decoder<utf8_counter>::decode_wchar_block(str, length, 0);
+	}
+
+	PUGI__FN void as_utf8_end(char* buffer, size_t size, const wchar_t* str, size_t length)
+	{
+		// convert to utf8
+		uint8_t* begin = reinterpret_cast<uint8_t*>(buffer);
+		uint8_t* end = utf_decoder<utf8_writer>::decode_wchar_block(str, length, begin);
+	
+		assert(begin + size == end);
+		(void)!end;
+		(void)!size;
+	}
+	
+#ifndef PUGIXML_NO_STL
+	PUGI__FN std::string as_utf8_impl(const wchar_t* str, size_t length)
+	{
+		// first pass: get length in utf8 characters
+		size_t size = as_utf8_begin(str, length);
+
+		// allocate resulting string
+		std::string result;
+		result.resize(size);
+
+		// second pass: convert to utf8
+		if (size > 0) as_utf8_end(&result[0], size, str, length);
+
+		return result;
+	}
+
+	PUGI__FN std::basic_string<wchar_t> as_wide_impl(const char* str, size_t size)
+	{
+		const uint8_t* data = reinterpret_cast<const uint8_t*>(str);
+
+		// first pass: get length in wchar_t units
+		size_t length = utf_decoder<wchar_counter>::decode_utf8_block(data, size, 0);
+
+		// allocate resulting string
+		std::basic_string<wchar_t> result;
+		result.resize(length);
+
+		// second pass: convert to wchar_t
+		if (length > 0)
+		{
+			wchar_writer::value_type begin = reinterpret_cast<wchar_writer::value_type>(&result[0]);
+			wchar_writer::value_type end = utf_decoder<wchar_writer>::decode_utf8_block(data, size, begin);
+
+			assert(begin + length == end);
+			(void)!end;
+		}
+
+		return result;
+	}
+#endif
+
+	template <typename Header>
+	inline bool strcpy_insitu_allow(size_t length, const Header& header, uintptr_t header_mask, char_t* target)
+	{
+		// never reuse shared memory
+		if (header & xml_memory_page_contents_shared_mask) return false;
+
+		size_t target_length = strlength(target);
+
+		// always reuse document buffer memory if possible
+		if ((header & header_mask) == 0) return target_length >= length;
+
+		// reuse heap memory if waste is not too great
+		const size_t reuse_threshold = 32;
+
+		return target_length >= length && (target_length < reuse_threshold || target_length - length < target_length / 2);
+	}
+
+	template <typename String, typename Header>
+	PUGI__FN bool strcpy_insitu(String& dest, Header& header, uintptr_t header_mask, const char_t* source)
+	{
+		size_t source_length = strlength(source);
+
+		if (source_length == 0)
+		{
+			// empty string and null pointer are equivalent, so just deallocate old memory
+			xml_allocator* alloc = PUGI__GETPAGE_IMPL(header)->allocator;
+
+			if (header & header_mask) alloc->deallocate_string(dest);
+			
+			// mark the string as not allocated
+			dest = 0;
+			header &= ~header_mask;
+
+			return true;
+		}
+		else if (dest && strcpy_insitu_allow(source_length, header, header_mask, dest))
+		{
+			// we can reuse old buffer, so just copy the new data (including zero terminator)
+			memcpy(dest, source, (source_length + 1) * sizeof(char_t));
+			
+			return true;
+		}
+		else
+		{
+			xml_allocator* alloc = PUGI__GETPAGE_IMPL(header)->allocator;
+
+			if (!alloc->reserve()) return false;
+
+			// allocate new buffer
+			char_t* buf = alloc->allocate_string(source_length + 1);
+			if (!buf) return false;
+
+			// copy the string (including zero terminator)
+			memcpy(buf, source, (source_length + 1) * sizeof(char_t));
+
+			// deallocate old buffer (*after* the above to protect against overlapping memory and/or allocation failures)
+			if (header & header_mask) alloc->deallocate_string(dest);
+			
+			// the string is now allocated, so set the flag
+			dest = buf;
+			header |= header_mask;
+
+			return true;
+		}
+	}
+
+	struct gap
+	{
+		char_t* end;
+		size_t size;
+			
+		gap(): end(0), size(0)
+		{
+		}
+			
+		// Push new gap, move s count bytes further (skipping the gap).
+		// Collapse previous gap.
+		void push(char_t*& s, size_t count)
+		{
+			if (end) // there was a gap already; collapse it
+			{
+				// Move [old_gap_end, new_gap_start) to [old_gap_start, ...)
+				assert(s >= end);
+				memmove(end - size, end, reinterpret_cast<char*>(s) - reinterpret_cast<char*>(end));
+			}
+				
+			s += count; // end of current gap
+				
+			// "merge" two gaps
+			end = s;
+			size += count;
+		}
+			
+		// Collapse all gaps, return past-the-end pointer
+		char_t* flush(char_t* s)
+		{
+			if (end)
+			{
+				// Move [old_gap_end, current_pos) to [old_gap_start, ...)
+				assert(s >= end);
+				memmove(end - size, end, reinterpret_cast<char*>(s) - reinterpret_cast<char*>(end));
+
+				return s - size;
+			}
+			else return s;
+		}
+	};
+	
+	PUGI__FN char_t* strconv_escape(char_t* s, gap& g)
+	{
+		char_t* stre = s + 1;
+
+		switch (*stre)
+		{
+			case '#':	// &#...
+			{
+				unsigned int ucsc = 0;
+
+				if (stre[1] == 'x') // &#x... (hex code)
+				{
+					stre += 2;
+
+					char_t ch = *stre;
+
+					if (ch == ';') return stre;
+
+					for (;;)
+					{
+						if (static_cast<unsigned int>(ch - '0') <= 9)
+							ucsc = 16 * ucsc + (ch - '0');
+						else if (static_cast<unsigned int>((ch | ' ') - 'a') <= 5)
+							ucsc = 16 * ucsc + ((ch | ' ') - 'a' + 10);
+						else if (ch == ';')
+							break;
+						else // cancel
+							return stre;
+
+						ch = *++stre;
+					}
+					
+					++stre;
+				}
+				else	// &#... (dec code)
+				{
+					char_t ch = *++stre;
+
+					if (ch == ';') return stre;
+
+					for (;;)
+					{
+						if (static_cast<unsigned int>(static_cast<unsigned int>(ch) - '0') <= 9)
+							ucsc = 10 * ucsc + (ch - '0');
+						else if (ch == ';')
+							break;
+						else // cancel
+							return stre;
+
+						ch = *++stre;
+					}
+					
+					++stre;
+				}
+
+			#ifdef PUGIXML_WCHAR_MODE
+				s = reinterpret_cast<char_t*>(wchar_writer::any(reinterpret_cast<wchar_writer::value_type>(s), ucsc));
+			#else
+				s = reinterpret_cast<char_t*>(utf8_writer::any(reinterpret_cast<uint8_t*>(s), ucsc));
+			#endif
+					
+				g.push(s, stre - s);
+				return stre;
+			}
+
+			case 'a':	// &a
+			{
+				++stre;
+
+				if (*stre == 'm') // &am
+				{
+					if (*++stre == 'p' && *++stre == ';') // &amp;
+					{
+						*s++ = '&';
+						++stre;
+							
+						g.push(s, stre - s);
+						return stre;
+					}
+				}
+				else if (*stre == 'p') // &ap
+				{
+					if (*++stre == 'o' && *++stre == 's' && *++stre == ';') // &apos;
+					{
+						*s++ = '\'';
+						++stre;
+
+						g.push(s, stre - s);
+						return stre;
+					}
+				}
+				break;
+			}
+
+			case 'g': // &g
+			{
+				if (*++stre == 't' && *++stre == ';') // &gt;
+				{
+					*s++ = '>';
+					++stre;
+					
+					g.push(s, stre - s);
+					return stre;
+				}
+				break;
+			}
+
+			case 'l': // &l
+			{
+				if (*++stre == 't' && *++stre == ';') // &lt;
+				{
+					*s++ = '<';
+					++stre;
+						
+					g.push(s, stre - s);
+					return stre;
+				}
+				break;
+			}
+
+			case 'q': // &q
+			{
+				if (*++stre == 'u' && *++stre == 'o' && *++stre == 't' && *++stre == ';') // &quot;
+				{
+					*s++ = '"';
+					++stre;
+					
+					g.push(s, stre - s);
+					return stre;
+				}
+				break;
+			}
+
+			default:
+				break;
+		}
+		
+		return stre;
+	}
+
+	// Parser utilities
+	#define PUGI__ENDSWITH(c, e)        ((c) == (e) || ((c) == 0 && endch == (e)))
+	#define PUGI__SKIPWS()              { while (PUGI__IS_CHARTYPE(*s, ct_space)) ++s; }
+	#define PUGI__OPTSET(OPT)           ( optmsk & (OPT) )
+	#define PUGI__PUSHNODE(TYPE)        { cursor = append_new_node(cursor, alloc, TYPE); if (!cursor) PUGI__THROW_ERROR(status_out_of_memory, s); }
+	#define PUGI__POPNODE()             { cursor = cursor->parent; }
+	#define PUGI__SCANFOR(X)            { while (*s != 0 && !(X)) ++s; }
+	#define PUGI__SCANWHILE(X)          { while (X) ++s; }
+	#define PUGI__SCANWHILE_UNROLL(X)   { for (;;) { char_t ss = s[0]; if (PUGI__UNLIKELY(!(X))) { break; } ss = s[1]; if (PUGI__UNLIKELY(!(X))) { s += 1; break; } ss = s[2]; if (PUGI__UNLIKELY(!(X))) { s += 2; break; } ss = s[3]; if (PUGI__UNLIKELY(!(X))) { s += 3; break; } s += 4; } }
+	#define PUGI__ENDSEG()              { ch = *s; *s = 0; ++s; }
+	#define PUGI__THROW_ERROR(err, m)   return error_offset = m, error_status = err, static_cast<char_t*>(0)
+	#define PUGI__CHECK_ERROR(err, m)   { if (*s == 0) PUGI__THROW_ERROR(err, m); }
+
+	PUGI__FN char_t* strconv_comment(char_t* s, char_t endch)
+	{
+		gap g;
+		
+		while (true)
+		{
+			PUGI__SCANWHILE_UNROLL(!PUGI__IS_CHARTYPE(ss, ct_parse_comment));
+		
+			if (*s == '\r') // Either a single 0x0d or 0x0d 0x0a pair
+			{
+				*s++ = '\n'; // replace first one with 0x0a
+				
+				if (*s == '\n') g.push(s, 1);
+			}
+			else if (s[0] == '-' && s[1] == '-' && PUGI__ENDSWITH(s[2], '>')) // comment ends here
+			{
+				*g.flush(s) = 0;
+				
+				return s + (s[2] == '>' ? 3 : 2);
+			}
+			else if (*s == 0)
+			{
+				return 0;
+			}
+			else ++s;
+		}
+	}
+
+	PUGI__FN char_t* strconv_cdata(char_t* s, char_t endch)
+	{
+		gap g;
+			
+		while (true)
+		{
+			PUGI__SCANWHILE_UNROLL(!PUGI__IS_CHARTYPE(ss, ct_parse_cdata));
+			
+			if (*s == '\r') // Either a single 0x0d or 0x0d 0x0a pair
+			{
+				*s++ = '\n'; // replace first one with 0x0a
+				
+				if (*s == '\n') g.push(s, 1);
+			}
+			else if (s[0] == ']' && s[1] == ']' && PUGI__ENDSWITH(s[2], '>')) // CDATA ends here
+			{
+				*g.flush(s) = 0;
+				
+				return s + 1;
+			}
+			else if (*s == 0)
+			{
+				return 0;
+			}
+			else ++s;
+		}
+	}
+	
+	typedef char_t* (*strconv_pcdata_t)(char_t*);
+		
+	template <typename opt_trim, typename opt_eol, typename opt_escape> struct strconv_pcdata_impl
+	{
+		static char_t* parse(char_t* s)
+		{
+			gap g;
+
+			char_t* begin = s;
+
+			while (true)
+			{
+				PUGI__SCANWHILE_UNROLL(!PUGI__IS_CHARTYPE(ss, ct_parse_pcdata));
+
+				if (*s == '<') // PCDATA ends here
+				{
+					char_t* end = g.flush(s);
+
+					if (opt_trim::value)
+						while (end > begin && PUGI__IS_CHARTYPE(end[-1], ct_space))
+							--end;
+
+					*end = 0;
+					
+					return s + 1;
+				}
+				else if (opt_eol::value && *s == '\r') // Either a single 0x0d or 0x0d 0x0a pair
+				{
+					*s++ = '\n'; // replace first one with 0x0a
+					
+					if (*s == '\n') g.push(s, 1);
+				}
+				else if (opt_escape::value && *s == '&')
+				{
+					s = strconv_escape(s, g);
+				}
+				else if (*s == 0)
+				{
+					char_t* end = g.flush(s);
+
+					if (opt_trim::value)
+						while (end > begin && PUGI__IS_CHARTYPE(end[-1], ct_space))
+							--end;
+
+					*end = 0;
+
+					return s;
+				}
+				else ++s;
+			}
+		}
+	};
+	
+	PUGI__FN strconv_pcdata_t get_strconv_pcdata(unsigned int optmask)
+	{
+		PUGI__STATIC_ASSERT(parse_escapes == 0x10 && parse_eol == 0x20 && parse_trim_pcdata == 0x0800);
+
+		switch (((optmask >> 4) & 3) | ((optmask >> 9) & 4)) // get bitmask for flags (eol escapes trim)
+		{
+		case 0: return strconv_pcdata_impl<opt_false, opt_false, opt_false>::parse;
+		case 1: return strconv_pcdata_impl<opt_false, opt_false, opt_true>::parse;
+		case 2: return strconv_pcdata_impl<opt_false, opt_true, opt_false>::parse;
+		case 3: return strconv_pcdata_impl<opt_false, opt_true, opt_true>::parse;
+		case 4: return strconv_pcdata_impl<opt_true, opt_false, opt_false>::parse;
+		case 5: return strconv_pcdata_impl<opt_true, opt_false, opt_true>::parse;
+		case 6: return strconv_pcdata_impl<opt_true, opt_true, opt_false>::parse;
+		case 7: return strconv_pcdata_impl<opt_true, opt_true, opt_true>::parse;
+		default: assert(false); return 0; // should not get here
+		}
+	}
+
+	typedef char_t* (*strconv_attribute_t)(char_t*, char_t);
+	
+	template <typename opt_escape> struct strconv_attribute_impl
+	{
+		static char_t* parse_wnorm(char_t* s, char_t end_quote)
+		{
+			gap g;
+
+			// trim leading whitespaces
+			if (PUGI__IS_CHARTYPE(*s, ct_space))
+			{
+				char_t* str = s;
+				
+				do ++str;
+				while (PUGI__IS_CHARTYPE(*str, ct_space));
+				
+				g.push(s, str - s);
+			}
+
+			while (true)
+			{
+				PUGI__SCANWHILE_UNROLL(!PUGI__IS_CHARTYPE(ss, ct_parse_attr_ws | ct_space));
+				
+				if (*s == end_quote)
+				{
+					char_t* str = g.flush(s);
+					
+					do *str-- = 0;
+					while (PUGI__IS_CHARTYPE(*str, ct_space));
+				
+					return s + 1;
+				}
+				else if (PUGI__IS_CHARTYPE(*s, ct_space))
+				{
+					*s++ = ' ';
+		
+					if (PUGI__IS_CHARTYPE(*s, ct_space))
+					{
+						char_t* str = s + 1;
+						while (PUGI__IS_CHARTYPE(*str, ct_space)) ++str;
+						
+						g.push(s, str - s);
+					}
+				}
+				else if (opt_escape::value && *s == '&')
+				{
+					s = strconv_escape(s, g);
+				}
+				else if (!*s)
+				{
+					return 0;
+				}
+				else ++s;
+			}
+		}
+
+		static char_t* parse_wconv(char_t* s, char_t end_quote)
+		{
+			gap g;
+
+			while (true)
+			{
+				PUGI__SCANWHILE_UNROLL(!PUGI__IS_CHARTYPE(ss, ct_parse_attr_ws));
+				
+				if (*s == end_quote)
+				{
+					*g.flush(s) = 0;
+				
+					return s + 1;
+				}
+				else if (PUGI__IS_CHARTYPE(*s, ct_space))
+				{
+					if (*s == '\r')
+					{
+						*s++ = ' ';
+				
+						if (*s == '\n') g.push(s, 1);
+					}
+					else *s++ = ' ';
+				}
+				else if (opt_escape::value && *s == '&')
+				{
+					s = strconv_escape(s, g);
+				}
+				else if (!*s)
+				{
+					return 0;
+				}
+				else ++s;
+			}
+		}
+
+		static char_t* parse_eol(char_t* s, char_t end_quote)
+		{
+			gap g;
+
+			while (true)
+			{
+				PUGI__SCANWHILE_UNROLL(!PUGI__IS_CHARTYPE(ss, ct_parse_attr));
+				
+				if (*s == end_quote)
+				{
+					*g.flush(s) = 0;
+				
+					return s + 1;
+				}
+				else if (*s == '\r')
+				{
+					*s++ = '\n';
+					
+					if (*s == '\n') g.push(s, 1);
+				}
+				else if (opt_escape::value && *s == '&')
+				{
+					s = strconv_escape(s, g);
+				}
+				else if (!*s)
+				{
+					return 0;
+				}
+				else ++s;
+			}
+		}
+
+		static char_t* parse_simple(char_t* s, char_t end_quote)
+		{
+			gap g;
+
+			while (true)
+			{
+				PUGI__SCANWHILE_UNROLL(!PUGI__IS_CHARTYPE(ss, ct_parse_attr));
+				
+				if (*s == end_quote)
+				{
+					*g.flush(s) = 0;
+				
+					return s + 1;
+				}
+				else if (opt_escape::value && *s == '&')
+				{
+					s = strconv_escape(s, g);
+				}
+				else if (!*s)
+				{
+					return 0;
+				}
+				else ++s;
+			}
+		}
+	};
+
+	PUGI__FN strconv_attribute_t get_strconv_attribute(unsigned int optmask)
+	{
+		PUGI__STATIC_ASSERT(parse_escapes == 0x10 && parse_eol == 0x20 && parse_wconv_attribute == 0x40 && parse_wnorm_attribute == 0x80);
+		
+		switch ((optmask >> 4) & 15) // get bitmask for flags (wconv wnorm eol escapes)
+		{
+		case 0:  return strconv_attribute_impl<opt_false>::parse_simple;
+		case 1:  return strconv_attribute_impl<opt_true>::parse_simple;
+		case 2:  return strconv_attribute_impl<opt_false>::parse_eol;
+		case 3:  return strconv_attribute_impl<opt_true>::parse_eol;
+		case 4:  return strconv_attribute_impl<opt_false>::parse_wconv;
+		case 5:  return strconv_attribute_impl<opt_true>::parse_wconv;
+		case 6:  return strconv_attribute_impl<opt_false>::parse_wconv;
+		case 7:  return strconv_attribute_impl<opt_true>::parse_wconv;
+		case 8:  return strconv_attribute_impl<opt_false>::parse_wnorm;
+		case 9:  return strconv_attribute_impl<opt_true>::parse_wnorm;
+		case 10: return strconv_attribute_impl<opt_false>::parse_wnorm;
+		case 11: return strconv_attribute_impl<opt_true>::parse_wnorm;
+		case 12: return strconv_attribute_impl<opt_false>::parse_wnorm;
+		case 13: return strconv_attribute_impl<opt_true>::parse_wnorm;
+		case 14: return strconv_attribute_impl<opt_false>::parse_wnorm;
+		case 15: return strconv_attribute_impl<opt_true>::parse_wnorm;
+		default: assert(false); return 0; // should not get here
+		}
+	}
+
+	inline xml_parse_result make_parse_result(xml_parse_status status, ptrdiff_t offset = 0)
+	{
+		xml_parse_result result;
+		result.status = status;
+		result.offset = offset;
+
+		return result;
+	}
+
+	struct xml_parser
+	{
+		xml_allocator alloc;
+		xml_allocator* alloc_state;
+		char_t* error_offset;
+		xml_parse_status error_status;
+		
+		xml_parser(xml_allocator* alloc_): alloc(*alloc_), alloc_state(alloc_), error_offset(0), error_status(status_ok)
+		{
+		}
+
+		~xml_parser()
+		{
+			*alloc_state = alloc;
+		}
+
+		// DOCTYPE consists of nested sections of the following possible types:
+		// <!-- ... -->, <? ... ?>, "...", '...'
+		// <![...]]>
+		// <!...>
+		// First group can not contain nested groups
+		// Second group can contain nested groups of the same type
+		// Third group can contain all other groups
+		char_t* parse_doctype_primitive(char_t* s)
+		{
+			if (*s == '"' || *s == '\'')
+			{
+				// quoted string
+				char_t ch = *s++;
+				PUGI__SCANFOR(*s == ch);
+				if (!*s) PUGI__THROW_ERROR(status_bad_doctype, s);
+
+				s++;
+			}
+			else if (s[0] == '<' && s[1] == '?')
+			{
+				// <? ... ?>
+				s += 2;
+				PUGI__SCANFOR(s[0] == '?' && s[1] == '>'); // no need for ENDSWITH because ?> can't terminate proper doctype
+				if (!*s) PUGI__THROW_ERROR(status_bad_doctype, s);
+
+				s += 2;
+			}
+			else if (s[0] == '<' && s[1] == '!' && s[2] == '-' && s[3] == '-')
+			{
+				s += 4;
+				PUGI__SCANFOR(s[0] == '-' && s[1] == '-' && s[2] == '>'); // no need for ENDSWITH because --> can't terminate proper doctype
+				if (!*s) PUGI__THROW_ERROR(status_bad_doctype, s);
+
+				s += 3;
+			}
+			else PUGI__THROW_ERROR(status_bad_doctype, s);
+
+			return s;
+		}
+
+		char_t* parse_doctype_ignore(char_t* s)
+		{
+			size_t depth = 0;
+
+			assert(s[0] == '<' && s[1] == '!' && s[2] == '[');
+			s += 3;
+
+			while (*s)
+			{
+				if (s[0] == '<' && s[1] == '!' && s[2] == '[')
+				{
+					// nested ignore section
+					s += 3;
+					depth++;
+				}
+				else if (s[0] == ']' && s[1] == ']' && s[2] == '>')
+				{
+					// ignore section end
+					s += 3;
+
+					if (depth == 0)
+						return s;
+
+					depth--;
+				}
+				else s++;
+			}
+
+			PUGI__THROW_ERROR(status_bad_doctype, s);
+		}
+
+		char_t* parse_doctype_group(char_t* s, char_t endch)
+		{
+			size_t depth = 0;
+
+			assert((s[0] == '<' || s[0] == 0) && s[1] == '!');
+			s += 2;
+
+			while (*s)
+			{
+				if (s[0] == '<' && s[1] == '!' && s[2] != '-')
+				{
+					if (s[2] == '[')
+					{
+						// ignore
+						s = parse_doctype_ignore(s);
+						if (!s) return s;
+					}
+					else
+					{
+						// some control group
+						s += 2;
+						depth++;
+					}
+				}
+				else if (s[0] == '<' || s[0] == '"' || s[0] == '\'')
+				{
+					// unknown tag (forbidden), or some primitive group
+					s = parse_doctype_primitive(s);
+					if (!s) return s;
+				}
+				else if (*s == '>')
+				{
+					if (depth == 0)
+						return s;
+
+					depth--;
+					s++;
+				}
+				else s++;
+			}
+
+			if (depth != 0 || endch != '>') PUGI__THROW_ERROR(status_bad_doctype, s);
+
+			return s;
+		}
+
+		char_t* parse_exclamation(char_t* s, xml_node_struct* cursor, unsigned int optmsk, char_t endch)
+		{
+			// parse node contents, starting with exclamation mark
+			++s;
+
+			if (*s == '-') // '<!-...'
+			{
+				++s;
+
+				if (*s == '-') // '<!--...'
+				{
+					++s;
+
+					if (PUGI__OPTSET(parse_comments))
+					{
+						PUGI__PUSHNODE(node_comment); // Append a new node on the tree.
+						cursor->value = s; // Save the offset.
+					}
+
+					if (PUGI__OPTSET(parse_eol) && PUGI__OPTSET(parse_comments))
+					{
+						s = strconv_comment(s, endch);
+
+						if (!s) PUGI__THROW_ERROR(status_bad_comment, cursor->value);
+					}
+					else
+					{
+						// Scan for terminating '-->'.
+						PUGI__SCANFOR(s[0] == '-' && s[1] == '-' && PUGI__ENDSWITH(s[2], '>'));
+						PUGI__CHECK_ERROR(status_bad_comment, s);
+
+						if (PUGI__OPTSET(parse_comments))
+							*s = 0; // Zero-terminate this segment at the first terminating '-'.
+
+						s += (s[2] == '>' ? 3 : 2); // Step over the '\0->'.
+					}
+				}
+				else PUGI__THROW_ERROR(status_bad_comment, s);
+			}
+			else if (*s == '[')
+			{
+				// '<![CDATA[...'
+				if (*++s=='C' && *++s=='D' && *++s=='A' && *++s=='T' && *++s=='A' && *++s == '[')
+				{
+					++s;
+
+					if (PUGI__OPTSET(parse_cdata))
+					{
+						PUGI__PUSHNODE(node_cdata); // Append a new node on the tree.
+						cursor->value = s; // Save the offset.
+
+						if (PUGI__OPTSET(parse_eol))
+						{
+							s = strconv_cdata(s, endch);
+
+							if (!s) PUGI__THROW_ERROR(status_bad_cdata, cursor->value);
+						}
+						else
+						{
+							// Scan for terminating ']]>'.
+							PUGI__SCANFOR(s[0] == ']' && s[1] == ']' && PUGI__ENDSWITH(s[2], '>'));
+							PUGI__CHECK_ERROR(status_bad_cdata, s);
+
+							*s++ = 0; // Zero-terminate this segment.
+						}
+					}
+					else // Flagged for discard, but we still have to scan for the terminator.
+					{
+						// Scan for terminating ']]>'.
+						PUGI__SCANFOR(s[0] == ']' && s[1] == ']' && PUGI__ENDSWITH(s[2], '>'));
+						PUGI__CHECK_ERROR(status_bad_cdata, s);
+
+						++s;
+					}
+
+					s += (s[1] == '>' ? 2 : 1); // Step over the last ']>'.
+				}
+				else PUGI__THROW_ERROR(status_bad_cdata, s);
+			}
+			else if (s[0] == 'D' && s[1] == 'O' && s[2] == 'C' && s[3] == 'T' && s[4] == 'Y' && s[5] == 'P' && PUGI__ENDSWITH(s[6], 'E'))
+			{
+				s -= 2;
+
+				if (cursor->parent) PUGI__THROW_ERROR(status_bad_doctype, s);
+
+				char_t* mark = s + 9;
+
+				s = parse_doctype_group(s, endch);
+				if (!s) return s;
+
+				assert((*s == 0 && endch == '>') || *s == '>');
+				if (*s) *s++ = 0;
+
+				if (PUGI__OPTSET(parse_doctype))
+				{
+					while (PUGI__IS_CHARTYPE(*mark, ct_space)) ++mark;
+
+					PUGI__PUSHNODE(node_doctype);
+
+					cursor->value = mark;
+				}
+			}
+			else if (*s == 0 && endch == '-') PUGI__THROW_ERROR(status_bad_comment, s);
+			else if (*s == 0 && endch == '[') PUGI__THROW_ERROR(status_bad_cdata, s);
+			else PUGI__THROW_ERROR(status_unrecognized_tag, s);
+
+			return s;
+		}
+
+		char_t* parse_question(char_t* s, xml_node_struct*& ref_cursor, unsigned int optmsk, char_t endch)
+		{
+			// load into registers
+			xml_node_struct* cursor = ref_cursor;
+			char_t ch = 0;
+
+			// parse node contents, starting with question mark
+			++s;
+
+			// read PI target
+			char_t* target = s;
+
+			if (!PUGI__IS_CHARTYPE(*s, ct_start_symbol)) PUGI__THROW_ERROR(status_bad_pi, s);
+
+			PUGI__SCANWHILE(PUGI__IS_CHARTYPE(*s, ct_symbol));
+			PUGI__CHECK_ERROR(status_bad_pi, s);
+
+			// determine node type; stricmp / strcasecmp is not portable
+			bool declaration = (target[0] | ' ') == 'x' && (target[1] | ' ') == 'm' && (target[2] | ' ') == 'l' && target + 3 == s;
+
+			if (declaration ? PUGI__OPTSET(parse_declaration) : PUGI__OPTSET(parse_pi))
+			{
+				if (declaration)
+				{
+					// disallow non top-level declarations
+					if (cursor->parent) PUGI__THROW_ERROR(status_bad_pi, s);
+
+					PUGI__PUSHNODE(node_declaration);
+				}
+				else
+				{
+					PUGI__PUSHNODE(node_pi);
+				}
+
+				cursor->name = target;
+
+				PUGI__ENDSEG();
+
+				// parse value/attributes
+				if (ch == '?')
+				{
+					// empty node
+					if (!PUGI__ENDSWITH(*s, '>')) PUGI__THROW_ERROR(status_bad_pi, s);
+					s += (*s == '>');
+
+					PUGI__POPNODE();
+				}
+				else if (PUGI__IS_CHARTYPE(ch, ct_space))
+				{
+					PUGI__SKIPWS();
+
+					// scan for tag end
+					char_t* value = s;
+
+					PUGI__SCANFOR(s[0] == '?' && PUGI__ENDSWITH(s[1], '>'));
+					PUGI__CHECK_ERROR(status_bad_pi, s);
+
+					if (declaration)
+					{
+						// replace ending ? with / so that 'element' terminates properly
+						*s = '/';
+
+						// we exit from this function with cursor at node_declaration, which is a signal to parse() to go to LOC_ATTRIBUTES
+						s = value;
+					}
+					else
+					{
+						// store value and step over >
+						cursor->value = value;
+
+						PUGI__POPNODE();
+
+						PUGI__ENDSEG();
+
+						s += (*s == '>');
+					}
+				}
+				else PUGI__THROW_ERROR(status_bad_pi, s);
+			}
+			else
+			{
+				// scan for tag end
+				PUGI__SCANFOR(s[0] == '?' && PUGI__ENDSWITH(s[1], '>'));
+				PUGI__CHECK_ERROR(status_bad_pi, s);
+
+				s += (s[1] == '>' ? 2 : 1);
+			}
+
+			// store from registers
+			ref_cursor = cursor;
+
+			return s;
+		}
+
+		char_t* parse_tree(char_t* s, xml_node_struct* root, unsigned int optmsk, char_t endch)
+		{
+			strconv_attribute_t strconv_attribute = get_strconv_attribute(optmsk);
+			strconv_pcdata_t strconv_pcdata = get_strconv_pcdata(optmsk);
+			
+			char_t ch = 0;
+			xml_node_struct* cursor = root;
+			char_t* mark = s;
+
+			while (*s != 0)
+			{
+				if (*s == '<')
+				{
+					++s;
+
+				LOC_TAG:
+					if (PUGI__IS_CHARTYPE(*s, ct_start_symbol)) // '<#...'
+					{
+						PUGI__PUSHNODE(node_element); // Append a new node to the tree.
+
+						cursor->name = s;
+
+						PUGI__SCANWHILE_UNROLL(PUGI__IS_CHARTYPE(ss, ct_symbol)); // Scan for a terminator.
+						PUGI__ENDSEG(); // Save char in 'ch', terminate & step over.
+
+						if (ch == '>')
+						{
+							// end of tag
+						}
+						else if (PUGI__IS_CHARTYPE(ch, ct_space))
+						{
+						LOC_ATTRIBUTES:
+							while (true)
+							{
+								PUGI__SKIPWS(); // Eat any whitespace.
+						
+								if (PUGI__IS_CHARTYPE(*s, ct_start_symbol)) // <... #...
+								{
+									xml_attribute_struct* a = append_new_attribute(cursor, alloc); // Make space for this attribute.
+									if (!a) PUGI__THROW_ERROR(status_out_of_memory, s);
+
+									a->name = s; // Save the offset.
+
+									PUGI__SCANWHILE_UNROLL(PUGI__IS_CHARTYPE(ss, ct_symbol)); // Scan for a terminator.
+									PUGI__ENDSEG(); // Save char in 'ch', terminate & step over.
+
+									if (PUGI__IS_CHARTYPE(ch, ct_space))
+									{
+										PUGI__SKIPWS(); // Eat any whitespace.
+
+										ch = *s;
+										++s;
+									}
+									
+									if (ch == '=') // '<... #=...'
+									{
+										PUGI__SKIPWS(); // Eat any whitespace.
+
+										if (*s == '"' || *s == '\'') // '<... #="...'
+										{
+											ch = *s; // Save quote char to avoid breaking on "''" -or- '""'.
+											++s; // Step over the quote.
+											a->value = s; // Save the offset.
+
+											s = strconv_attribute(s, ch);
+										
+											if (!s) PUGI__THROW_ERROR(status_bad_attribute, a->value);
+
+											// After this line the loop continues from the start;
+											// Whitespaces, / and > are ok, symbols and EOF are wrong,
+											// everything else will be detected
+											if (PUGI__IS_CHARTYPE(*s, ct_start_symbol)) PUGI__THROW_ERROR(status_bad_attribute, s);
+										}
+										else PUGI__THROW_ERROR(status_bad_attribute, s);
+									}
+									else PUGI__THROW_ERROR(status_bad_attribute, s);
+								}
+								else if (*s == '/')
+								{
+									++s;
+									
+									if (*s == '>')
+									{
+										PUGI__POPNODE();
+										s++;
+										break;
+									}
+									else if (*s == 0 && endch == '>')
+									{
+										PUGI__POPNODE();
+										break;
+									}
+									else PUGI__THROW_ERROR(status_bad_start_element, s);
+								}
+								else if (*s == '>')
+								{
+									++s;
+
+									break;
+								}
+								else if (*s == 0 && endch == '>')
+								{
+									break;
+								}
+								else PUGI__THROW_ERROR(status_bad_start_element, s);
+							}
+
+							// !!!
+						}
+						else if (ch == '/') // '<#.../'
+						{
+							if (!PUGI__ENDSWITH(*s, '>')) PUGI__THROW_ERROR(status_bad_start_element, s);
+
+							PUGI__POPNODE(); // Pop.
+
+							s += (*s == '>');
+						}
+						else if (ch == 0)
+						{
+							// we stepped over null terminator, backtrack & handle closing tag
+							--s;
+							
+							if (endch != '>') PUGI__THROW_ERROR(status_bad_start_element, s);
+						}
+						else PUGI__THROW_ERROR(status_bad_start_element, s);
+					}
+					else if (*s == '/')
+					{
+						++s;
+
+						char_t* name = cursor->name;
+						if (!name) PUGI__THROW_ERROR(status_end_element_mismatch, s);
+						
+						while (PUGI__IS_CHARTYPE(*s, ct_symbol))
+						{
+							if (*s++ != *name++) PUGI__THROW_ERROR(status_end_element_mismatch, s);
+						}
+
+						if (*name)
+						{
+							if (*s == 0 && name[0] == endch && name[1] == 0) PUGI__THROW_ERROR(status_bad_end_element, s);
+							else PUGI__THROW_ERROR(status_end_element_mismatch, s);
+						}
+							
+						PUGI__POPNODE(); // Pop.
+
+						PUGI__SKIPWS();
+
+						if (*s == 0)
+						{
+							if (endch != '>') PUGI__THROW_ERROR(status_bad_end_element, s);
+						}
+						else
+						{
+							if (*s != '>') PUGI__THROW_ERROR(status_bad_end_element, s);
+							++s;
+						}
+					}
+					else if (*s == '?') // '<?...'
+					{
+						s = parse_question(s, cursor, optmsk, endch);
+						if (!s) return s;
+
+						assert(cursor);
+						if (PUGI__NODETYPE(cursor) == node_declaration) goto LOC_ATTRIBUTES;
+					}
+					else if (*s == '!') // '<!...'
+					{
+						s = parse_exclamation(s, cursor, optmsk, endch);
+						if (!s) return s;
+					}
+					else if (*s == 0 && endch == '?') PUGI__THROW_ERROR(status_bad_pi, s);
+					else PUGI__THROW_ERROR(status_unrecognized_tag, s);
+				}
+				else
+				{
+					mark = s; // Save this offset while searching for a terminator.
+
+					PUGI__SKIPWS(); // Eat whitespace if no genuine PCDATA here.
+
+					if (*s == '<' || !*s)
+					{
+						// We skipped some whitespace characters because otherwise we would take the tag branch instead of PCDATA one
+						assert(mark != s);
+
+						if (!PUGI__OPTSET(parse_ws_pcdata | parse_ws_pcdata_single) || PUGI__OPTSET(parse_trim_pcdata))
+						{
+							continue;
+						}
+						else if (PUGI__OPTSET(parse_ws_pcdata_single))
+						{
+							if (s[0] != '<' || s[1] != '/' || cursor->first_child) continue;
+						}
+					}
+
+					if (!PUGI__OPTSET(parse_trim_pcdata))
+						s = mark;
+							
+					if (cursor->parent || PUGI__OPTSET(parse_fragment))
+					{
+						PUGI__PUSHNODE(node_pcdata); // Append a new node on the tree.
+						cursor->value = s; // Save the offset.
+
+						s = strconv_pcdata(s);
+								
+						PUGI__POPNODE(); // Pop since this is a standalone.
+						
+						if (!*s) break;
+					}
+					else
+					{
+						PUGI__SCANFOR(*s == '<'); // '...<'
+						if (!*s) break;
+						
+						++s;
+					}
+
+					// We're after '<'
+					goto LOC_TAG;
+				}
+			}
+
+			// check that last tag is closed
+			if (cursor != root) PUGI__THROW_ERROR(status_end_element_mismatch, s);
+
+			return s;
+		}
+
+	#ifdef PUGIXML_WCHAR_MODE
+		static char_t* parse_skip_bom(char_t* s)
+		{
+			unsigned int bom = 0xfeff;
+			return (s[0] == static_cast<wchar_t>(bom)) ? s + 1 : s;
+		}
+	#else
+		static char_t* parse_skip_bom(char_t* s)
+		{
+			return (s[0] == '\xef' && s[1] == '\xbb' && s[2] == '\xbf') ? s + 3 : s;
+		}
+	#endif
+
+		static bool has_element_node_siblings(xml_node_struct* node)
+		{
+			while (node)
+			{
+				if (PUGI__NODETYPE(node) == node_element) return true;
+
+				node = node->next_sibling;
+			}
+
+			return false;
+		}
+
+		static xml_parse_result parse(char_t* buffer, size_t length, xml_document_struct* xmldoc, xml_node_struct* root, unsigned int optmsk)
+		{
+			// early-out for empty documents
+			if (length == 0)
+				return make_parse_result(PUGI__OPTSET(parse_fragment) ? status_ok : status_no_document_element);
+
+			// get last child of the root before parsing
+			xml_node_struct* last_root_child = root->first_child ? root->first_child->prev_sibling_c + 0 : 0;
+	
+			// create parser on stack
+			xml_parser parser(static_cast<xml_allocator*>(xmldoc));
+
+			// save last character and make buffer zero-terminated (speeds up parsing)
+			char_t endch = buffer[length - 1];
+			buffer[length - 1] = 0;
+			
+			// skip BOM to make sure it does not end up as part of parse output
+			char_t* buffer_data = parse_skip_bom(buffer);
+
+			// perform actual parsing
+			parser.parse_tree(buffer_data, root, optmsk, endch);
+
+			xml_parse_result result = make_parse_result(parser.error_status, parser.error_offset ? parser.error_offset - buffer : 0);
+			assert(result.offset >= 0 && static_cast<size_t>(result.offset) <= length);
+
+			if (result)
+			{
+				// since we removed last character, we have to handle the only possible false positive (stray <)
+				if (endch == '<')
+					return make_parse_result(status_unrecognized_tag, length - 1);
+
+				// check if there are any element nodes parsed
+				xml_node_struct* first_root_child_parsed = last_root_child ? last_root_child->next_sibling + 0 : root->first_child;
+
+				if (!PUGI__OPTSET(parse_fragment) && !has_element_node_siblings(first_root_child_parsed))
+					return make_parse_result(status_no_document_element, length - 1);
+			}
+			else
+			{
+				// roll back offset if it occurs on a null terminator in the source buffer
+				if (result.offset > 0 && static_cast<size_t>(result.offset) == length - 1 && endch == 0)
+					result.offset--;
+			}
+
+			return result;
+		}
+	};
+
+	// Output facilities
+	PUGI__FN xml_encoding get_write_native_encoding()
+	{
+	#ifdef PUGIXML_WCHAR_MODE
+		return get_wchar_encoding();
+	#else
+		return encoding_utf8;
+	#endif
+	}
+
+	PUGI__FN xml_encoding get_write_encoding(xml_encoding encoding)
+	{
+		// replace wchar encoding with utf implementation
+		if (encoding == encoding_wchar) return get_wchar_encoding();
+
+		// replace utf16 encoding with utf16 with specific endianness
+		if (encoding == encoding_utf16) return is_little_endian() ? encoding_utf16_le : encoding_utf16_be;
+
+		// replace utf32 encoding with utf32 with specific endianness
+		if (encoding == encoding_utf32) return is_little_endian() ? encoding_utf32_le : encoding_utf32_be;
+
+		// only do autodetection if no explicit encoding is requested
+		if (encoding != encoding_auto) return encoding;
+
+		// assume utf8 encoding
+		return encoding_utf8;
+	}
+
+#ifdef PUGIXML_WCHAR_MODE
+	PUGI__FN size_t get_valid_length(const char_t* data, size_t length)
+	{
+		if (length < 1) return 0;
+
+		// discard last character if it's the lead of a surrogate pair 
+		return (sizeof(wchar_t) == 2 && static_cast<unsigned int>(static_cast<uint16_t>(data[length - 1]) - 0xD800) < 0x400) ? length - 1 : length;
+	}
+
+	PUGI__FN size_t convert_buffer_output(char_t* r_char, uint8_t* r_u8, uint16_t* r_u16, uint32_t* r_u32, const char_t* data, size_t length, xml_encoding encoding)
+	{
+		// only endian-swapping is required
+		if (need_endian_swap_utf(encoding, get_wchar_encoding()))
+		{
+			convert_wchar_endian_swap(r_char, data, length);
+
+			return length * sizeof(char_t);
+		}
+	
+		// convert to utf8
+		if (encoding == encoding_utf8)
+		{
+			uint8_t* dest = r_u8;
+			uint8_t* end = utf_decoder<utf8_writer>::decode_wchar_block(data, length, dest);
+
+			return static_cast<size_t>(end - dest);
+		}
+
+		// convert to utf16
+		if (encoding == encoding_utf16_be || encoding == encoding_utf16_le)
+		{
+			uint16_t* dest = r_u16;
+
+			// convert to native utf16
+			uint16_t* end = utf_decoder<utf16_writer>::decode_wchar_block(data, length, dest);
+
+			// swap if necessary
+			xml_encoding native_encoding = is_little_endian() ? encoding_utf16_le : encoding_utf16_be;
+
+			if (native_encoding != encoding) convert_utf_endian_swap(dest, dest, static_cast<size_t>(end - dest));
+
+			return static_cast<size_t>(end - dest) * sizeof(uint16_t);
+		}
+
+		// convert to utf32
+		if (encoding == encoding_utf32_be || encoding == encoding_utf32_le)
+		{
+			uint32_t* dest = r_u32;
+
+			// convert to native utf32
+			uint32_t* end = utf_decoder<utf32_writer>::decode_wchar_block(data, length, dest);
+
+			// swap if necessary
+			xml_encoding native_encoding = is_little_endian() ? encoding_utf32_le : encoding_utf32_be;
+
+			if (native_encoding != encoding) convert_utf_endian_swap(dest, dest, static_cast<size_t>(end - dest));
+
+			return static_cast<size_t>(end - dest) * sizeof(uint32_t);
+		}
+
+		// convert to latin1
+		if (encoding == encoding_latin1)
+		{
+			uint8_t* dest = r_u8;
+			uint8_t* end = utf_decoder<latin1_writer>::decode_wchar_block(data, length, dest);
+
+			return static_cast<size_t>(end - dest);
+		}
+
+		assert(!"Invalid encoding");
+		return 0;
+	}
+#else
+	PUGI__FN size_t get_valid_length(const char_t* data, size_t length)
+	{
+		if (length < 5) return 0;
+
+		for (size_t i = 1; i <= 4; ++i)
+		{
+			uint8_t ch = static_cast<uint8_t>(data[length - i]);
+
+			// either a standalone character or a leading one
+			if ((ch & 0xc0) != 0x80) return length - i;
+		}
+
+		// there are four non-leading characters at the end, sequence tail is broken so might as well process the whole chunk
+		return length;
+	}
+
+	PUGI__FN size_t convert_buffer_output(char_t* /* r_char */, uint8_t* r_u8, uint16_t* r_u16, uint32_t* r_u32, const char_t* data, size_t length, xml_encoding encoding)
+	{
+		if (encoding == encoding_utf16_be || encoding == encoding_utf16_le)
+		{
+			uint16_t* dest = r_u16;
+
+			// convert to native utf16
+			uint16_t* end = utf_decoder<utf16_writer>::decode_utf8_block(reinterpret_cast<const uint8_t*>(data), length, dest);
+
+			// swap if necessary
+			xml_encoding native_encoding = is_little_endian() ? encoding_utf16_le : encoding_utf16_be;
+
+			if (native_encoding != encoding) convert_utf_endian_swap(dest, dest, static_cast<size_t>(end - dest));
+
+			return static_cast<size_t>(end - dest) * sizeof(uint16_t);
+		}
+
+		if (encoding == encoding_utf32_be || encoding == encoding_utf32_le)
+		{
+			uint32_t* dest = r_u32;
+
+			// convert to native utf32
+			uint32_t* end = utf_decoder<utf32_writer>::decode_utf8_block(reinterpret_cast<const uint8_t*>(data), length, dest);
+
+			// swap if necessary
+			xml_encoding native_encoding = is_little_endian() ? encoding_utf32_le : encoding_utf32_be;
+
+			if (native_encoding != encoding) convert_utf_endian_swap(dest, dest, static_cast<size_t>(end - dest));
+
+			return static_cast<size_t>(end - dest) * sizeof(uint32_t);
+		}
+
+		if (encoding == encoding_latin1)
+		{
+			uint8_t* dest = r_u8;
+			uint8_t* end = utf_decoder<latin1_writer>::decode_utf8_block(reinterpret_cast<const uint8_t*>(data), length, dest);
+
+			return static_cast<size_t>(end - dest);
+		}
+
+		assert(!"Invalid encoding");
+		return 0;
+	}
+#endif
+
+	class xml_buffered_writer
+	{
+		xml_buffered_writer(const xml_buffered_writer&);
+		xml_buffered_writer& operator=(const xml_buffered_writer&);
+
+	public:
+		xml_buffered_writer(xml_writer& writer_, xml_encoding user_encoding): writer(writer_), bufsize(0), encoding(get_write_encoding(user_encoding))
+		{
+			PUGI__STATIC_ASSERT(bufcapacity >= 8);
+		}
+
+		size_t flush()
+		{
+			flush(buffer, bufsize);
+			bufsize = 0;
+			return 0;
+		}
+
+		void flush(const char_t* data, size_t size)
+		{
+			if (size == 0) return;
+
+			// fast path, just write data
+			if (encoding == get_write_native_encoding())
+				writer.write(data, size * sizeof(char_t));
+			else
+			{
+				// convert chunk
+				size_t result = convert_buffer_output(scratch.data_char, scratch.data_u8, scratch.data_u16, scratch.data_u32, data, size, encoding);
+				assert(result <= sizeof(scratch));
+
+				// write data
+				writer.write(scratch.data_u8, result);
+			}
+		}
+
+		void write_direct(const char_t* data, size_t length)
+		{
+			// flush the remaining buffer contents
+			flush();
+
+			// handle large chunks
+			if (length > bufcapacity)
+			{
+				if (encoding == get_write_native_encoding())
+				{
+					// fast path, can just write data chunk
+					writer.write(data, length * sizeof(char_t));
+					return;
+				}
+
+				// need to convert in suitable chunks
+				while (length > bufcapacity)
+				{
+					// get chunk size by selecting such number of characters that are guaranteed to fit into scratch buffer
+					// and form a complete codepoint sequence (i.e. discard start of last codepoint if necessary)
+					size_t chunk_size = get_valid_length(data, bufcapacity);
+					assert(chunk_size);
+
+					// convert chunk and write
+					flush(data, chunk_size);
+
+					// iterate
+					data += chunk_size;
+					length -= chunk_size;
+				}
+
+				// small tail is copied below
+				bufsize = 0;
+			}
+
+			memcpy(buffer + bufsize, data, length * sizeof(char_t));
+			bufsize += length;
+		}
+
+		void write_buffer(const char_t* data, size_t length)
+		{
+			size_t offset = bufsize;
+
+			if (offset + length <= bufcapacity)
+			{
+				memcpy(buffer + offset, data, length * sizeof(char_t));
+				bufsize = offset + length;
+			}
+			else
+			{
+				write_direct(data, length);
+			}
+		}
+
+		void write_string(const char_t* data)
+		{
+			// write the part of the string that fits in the buffer
+			size_t offset = bufsize;
+
+			while (*data && offset < bufcapacity)
+				buffer[offset++] = *data++;
+
+			// write the rest
+			if (offset < bufcapacity)
+			{
+				bufsize = offset;
+			}
+			else
+			{
+				// backtrack a bit if we have split the codepoint
+				size_t length = offset - bufsize;
+				size_t extra = length - get_valid_length(data - length, length);
+
+				bufsize = offset - extra;
+
+				write_direct(data - extra, strlength(data) + extra);
+			}
+		}
+
+		void write(char_t d0)
+		{
+			size_t offset = bufsize;
+			if (offset > bufcapacity - 1) offset = flush();
+
+			buffer[offset + 0] = d0;
+			bufsize = offset + 1;
+		}
+
+		void write(char_t d0, char_t d1)
+		{
+			size_t offset = bufsize;
+			if (offset > bufcapacity - 2) offset = flush();
+
+			buffer[offset + 0] = d0;
+			buffer[offset + 1] = d1;
+			bufsize = offset + 2;
+		}
+
+		void write(char_t d0, char_t d1, char_t d2)
+		{
+			size_t offset = bufsize;
+			if (offset > bufcapacity - 3) offset = flush();
+
+			buffer[offset + 0] = d0;
+			buffer[offset + 1] = d1;
+			buffer[offset + 2] = d2;
+			bufsize = offset + 3;
+		}
+
+		void write(char_t d0, char_t d1, char_t d2, char_t d3)
+		{
+			size_t offset = bufsize;
+			if (offset > bufcapacity - 4) offset = flush();
+
+			buffer[offset + 0] = d0;
+			buffer[offset + 1] = d1;
+			buffer[offset + 2] = d2;
+			buffer[offset + 3] = d3;
+			bufsize = offset + 4;
+		}
+
+		void write(char_t d0, char_t d1, char_t d2, char_t d3, char_t d4)
+		{
+			size_t offset = bufsize;
+			if (offset > bufcapacity - 5) offset = flush();
+
+			buffer[offset + 0] = d0;
+			buffer[offset + 1] = d1;
+			buffer[offset + 2] = d2;
+			buffer[offset + 3] = d3;
+			buffer[offset + 4] = d4;
+			bufsize = offset + 5;
+		}
+
+		void write(char_t d0, char_t d1, char_t d2, char_t d3, char_t d4, char_t d5)
+		{
+			size_t offset = bufsize;
+			if (offset > bufcapacity - 6) offset = flush();
+
+			buffer[offset + 0] = d0;
+			buffer[offset + 1] = d1;
+			buffer[offset + 2] = d2;
+			buffer[offset + 3] = d3;
+			buffer[offset + 4] = d4;
+			buffer[offset + 5] = d5;
+			bufsize = offset + 6;
+		}
+
+		// utf8 maximum expansion: x4 (-> utf32)
+		// utf16 maximum expansion: x2 (-> utf32)
+		// utf32 maximum expansion: x1
+		enum
+		{
+			bufcapacitybytes =
+			#ifdef PUGIXML_MEMORY_OUTPUT_STACK
+				PUGIXML_MEMORY_OUTPUT_STACK
+			#else
+				10240
+			#endif
+			,
+			bufcapacity = bufcapacitybytes / (sizeof(char_t) + 4)
+		};
+
+		char_t buffer[bufcapacity];
+
+		union
+		{
+			uint8_t data_u8[4 * bufcapacity];
+			uint16_t data_u16[2 * bufcapacity];
+			uint32_t data_u32[bufcapacity];
+			char_t data_char[bufcapacity];
+		} scratch;
+
+		xml_writer& writer;
+		size_t bufsize;
+		xml_encoding encoding;
+	};
+
+	PUGI__FN void text_output_escaped(xml_buffered_writer& writer, const char_t* s, chartypex_t type)
+	{
+		while (*s)
+		{
+			const char_t* prev = s;
+			
+			// While *s is a usual symbol
+			PUGI__SCANWHILE_UNROLL(!PUGI__IS_CHARTYPEX(ss, type));
+		
+			writer.write_buffer(prev, static_cast<size_t>(s - prev));
+
+			switch (*s)
+			{
+				case 0: break;
+				case '&':
+					writer.write('&', 'a', 'm', 'p', ';');
+					++s;
+					break;
+				case '<':
+					writer.write('&', 'l', 't', ';');
+					++s;
+					break;
+				case '>':
+					writer.write('&', 'g', 't', ';');
+					++s;
+					break;
+				case '"':
+					writer.write('&', 'q', 'u', 'o', 't', ';');
+					++s;
+					break;
+				default: // s is not a usual symbol
+				{
+					unsigned int ch = static_cast<unsigned int>(*s++);
+					assert(ch < 32);
+
+					writer.write('&', '#', static_cast<char_t>((ch / 10) + '0'), static_cast<char_t>((ch % 10) + '0'), ';');
+				}
+			}
+		}
+	}
+
+	PUGI__FN void text_output(xml_buffered_writer& writer, const char_t* s, chartypex_t type, unsigned int flags)
+	{
+		if (flags & format_no_escapes)
+			writer.write_string(s);
+		else
+			text_output_escaped(writer, s, type);
+	}
+
+	PUGI__FN void text_output_cdata(xml_buffered_writer& writer, const char_t* s)
+	{
+		do
+		{
+			writer.write('<', '!', '[', 'C', 'D');
+			writer.write('A', 'T', 'A', '[');
+
+			const char_t* prev = s;
+
+			// look for ]]> sequence - we can't output it as is since it terminates CDATA
+			while (*s && !(s[0] == ']' && s[1] == ']' && s[2] == '>')) ++s;
+
+			// skip ]] if we stopped at ]]>, > will go to the next CDATA section
+			if (*s) s += 2;
+
+			writer.write_buffer(prev, static_cast<size_t>(s - prev));
+
+			writer.write(']', ']', '>');
+		}
+		while (*s);
+	}
+
+	PUGI__FN void text_output_indent(xml_buffered_writer& writer, const char_t* indent, size_t indent_length, unsigned int depth)
+	{
+		switch (indent_length)
+		{
+		case 1:
+		{
+			for (unsigned int i = 0; i < depth; ++i)
+				writer.write(indent[0]);
+			break;
+		}
+
+		case 2:
+		{
+			for (unsigned int i = 0; i < depth; ++i)
+				writer.write(indent[0], indent[1]);
+			break;
+		}
+
+		case 3:
+		{
+			for (unsigned int i = 0; i < depth; ++i)
+				writer.write(indent[0], indent[1], indent[2]);
+			break;
+		}
+
+		case 4:
+		{
+			for (unsigned int i = 0; i < depth; ++i)
+				writer.write(indent[0], indent[1], indent[2], indent[3]);
+			break;
+		}
+
+		default:
+		{
+			for (unsigned int i = 0; i < depth; ++i)
+				writer.write_buffer(indent, indent_length);
+		}
+		}
+	}
+
+	PUGI__FN void node_output_comment(xml_buffered_writer& writer, const char_t* s)
+	{
+		writer.write('<', '!', '-', '-');
+
+		while (*s)
+		{
+			const char_t* prev = s;
+
+			// look for -\0 or -- sequence - we can't output it since -- is illegal in comment body
+			while (*s && !(s[0] == '-' && (s[1] == '-' || s[1] == 0))) ++s;
+
+			writer.write_buffer(prev, static_cast<size_t>(s - prev));
+
+			if (*s)
+			{
+				assert(*s == '-');
+
+				writer.write('-', ' ');
+				++s;
+			}
+		}
+
+		writer.write('-', '-', '>');
+	}
+
+	PUGI__FN void node_output_pi_value(xml_buffered_writer& writer, const char_t* s)
+	{
+		while (*s)
+		{
+			const char_t* prev = s;
+
+			// look for ?> sequence - we can't output it since ?> terminates PI
+			while (*s && !(s[0] == '?' && s[1] == '>')) ++s;
+
+			writer.write_buffer(prev, static_cast<size_t>(s - prev));
+
+			if (*s)
+			{
+				assert(s[0] == '?' && s[1] == '>');
+
+				writer.write('?', ' ', '>');
+				s += 2;
+			}
+		}
+	}
+
+	PUGI__FN void node_output_attributes(xml_buffered_writer& writer, xml_node_struct* node, const char_t* indent, size_t indent_length, unsigned int flags, unsigned int depth)
+	{
+		const char_t* default_name = PUGIXML_TEXT(":anonymous");
+
+		for (xml_attribute_struct* a = node->first_attribute; a; a = a->next_attribute)
+		{
+			if ((flags & (format_indent_attributes | format_raw)) == format_indent_attributes)
+			{
+				writer.write('\n');
+
+				text_output_indent(writer, indent, indent_length, depth + 1);
+			}
+			else
+			{
+				writer.write(' ');
+			}
+
+			writer.write_string(a->name ? a->name : default_name);
+			writer.write('=', '"');
+
+			if (a->value)
+				text_output(writer, a->value, ctx_special_attr, flags);
+
+			writer.write('"');
+		}
+	}
+
+	PUGI__FN bool node_output_start(xml_buffered_writer& writer, xml_node_struct* node, const char_t* indent, size_t indent_length, unsigned int flags, unsigned int depth)
+	{
+		const char_t* default_name = PUGIXML_TEXT(":anonymous");
+		const char_t* name = node->name ? node->name : default_name;
+
+		writer.write('<');
+		writer.write_string(name);
+
+		if (node->first_attribute)
+			node_output_attributes(writer, node, indent, indent_length, flags, depth);
+
+		if (!node->first_child)
+		{
+			writer.write(' ', '/', '>');
+
+			return false;
+		}
+		else
+		{
+			writer.write('>');
+
+			return true;
+		}
+	}
+
+	PUGI__FN void node_output_end(xml_buffered_writer& writer, xml_node_struct* node)
+	{
+		const char_t* default_name = PUGIXML_TEXT(":anonymous");
+		const char_t* name = node->name ? node->name : default_name;
+
+		writer.write('<', '/');
+		writer.write_string(name);
+		writer.write('>');
+	}
+
+	PUGI__FN void node_output_simple(xml_buffered_writer& writer, xml_node_struct* node, unsigned int flags)
+	{
+		const char_t* default_name = PUGIXML_TEXT(":anonymous");
+
+		switch (PUGI__NODETYPE(node))
+		{
+			case node_pcdata:
+				text_output(writer, node->value ? node->value + 0 : PUGIXML_TEXT(""), ctx_special_pcdata, flags);
+				break;
+
+			case node_cdata:
+				text_output_cdata(writer, node->value ? node->value + 0 : PUGIXML_TEXT(""));
+				break;
+
+			case node_comment:
+				node_output_comment(writer, node->value ? node->value + 0 : PUGIXML_TEXT(""));
+				break;
+
+			case node_pi:
+				writer.write('<', '?');
+				writer.write_string(node->name ? node->name : default_name);
+
+				if (node->value)
+				{
+					writer.write(' ');
+					node_output_pi_value(writer, node->value);
+				}
+
+				writer.write('?', '>');
+				break;
+
+			case node_declaration:
+				writer.write('<', '?');
+				writer.write_string(node->name ? node->name : default_name);
+				node_output_attributes(writer, node, PUGIXML_TEXT(""), 0, flags | format_raw, 0);
+				writer.write('?', '>');
+				break;
+
+			case node_doctype:
+				writer.write('<', '!', 'D', 'O', 'C');
+				writer.write('T', 'Y', 'P', 'E');
+
+				if (node->value)
+				{
+					writer.write(' ');
+					writer.write_string(node->value);
+				}
+
+				writer.write('>');
+				break;
+
+			default:
+				assert(!"Invalid node type");
+		}
+	}
+
+	enum indent_flags_t
+	{
+		indent_newline = 1,
+		indent_indent = 2
+	};
+
+	PUGI__FN void node_output(xml_buffered_writer& writer, xml_node_struct* root, const char_t* indent, unsigned int flags, unsigned int depth)
+	{
+		size_t indent_length = ((flags & (format_indent | format_indent_attributes)) && (flags & format_raw) == 0) ? strlength(indent) : 0;
+		unsigned int indent_flags = indent_indent;
+
+		xml_node_struct* node = root;
+
+		do
+		{
+			assert(node);
+
+			// begin writing current node
+			if (PUGI__NODETYPE(node) == node_pcdata || PUGI__NODETYPE(node) == node_cdata)
+			{
+				node_output_simple(writer, node, flags);
+
+				indent_flags = 0;
+			}
+			else
+			{
+				if ((indent_flags & indent_newline) && (flags & format_raw) == 0)
+					writer.write('\n');
+
+				if ((indent_flags & indent_indent) && indent_length)
+					text_output_indent(writer, indent, indent_length, depth);
+
+				if (PUGI__NODETYPE(node) == node_element)
+				{
+					indent_flags = indent_newline | indent_indent;
+
+					if (node_output_start(writer, node, indent, indent_length, flags, depth))
+					{
+						node = node->first_child;
+						depth++;
+						continue;
+					}
+				}
+				else if (PUGI__NODETYPE(node) == node_document)
+				{
+					indent_flags = indent_indent;
+
+					if (node->first_child)
+					{
+						node = node->first_child;
+						continue;
+					}
+				}
+				else
+				{
+					node_output_simple(writer, node, flags);
+
+					indent_flags = indent_newline | indent_indent;
+				}
+			}
+
+			// continue to the next node
+			while (node != root)
+			{
+				if (node->next_sibling)
+				{
+					node = node->next_sibling;
+					break;
+				}
+
+				node = node->parent;
+
+				// write closing node
+				if (PUGI__NODETYPE(node) == node_element)
+				{
+					depth--;
+
+					if ((indent_flags & indent_newline) && (flags & format_raw) == 0)
+						writer.write('\n');
+
+					if ((indent_flags & indent_indent) && indent_length)
+						text_output_indent(writer, indent, indent_length, depth);
+
+					node_output_end(writer, node);
+
+					indent_flags = indent_newline | indent_indent;
+				}
+			}
+		}
+		while (node != root);
+
+		if ((indent_flags & indent_newline) && (flags & format_raw) == 0)
+			writer.write('\n');
+	}
+
+	PUGI__FN bool has_declaration(xml_node_struct* node)
+	{
+		for (xml_node_struct* child = node->first_child; child; child = child->next_sibling)
+		{
+			xml_node_type type = PUGI__NODETYPE(child);
+
+			if (type == node_declaration) return true;
+			if (type == node_element) return false;
+		}
+
+		return false;
+	}
+
+	PUGI__FN bool is_attribute_of(xml_attribute_struct* attr, xml_node_struct* node)
+	{
+		for (xml_attribute_struct* a = node->first_attribute; a; a = a->next_attribute)
+			if (a == attr)
+				return true;
+
+		return false;
+	}
+
+	PUGI__FN bool allow_insert_attribute(xml_node_type parent)
+	{
+		return parent == node_element || parent == node_declaration;
+	}
+
+	PUGI__FN bool allow_insert_child(xml_node_type parent, xml_node_type child)
+	{
+		if (parent != node_document && parent != node_element) return false;
+		if (child == node_document || child == node_null) return false;
+		if (parent != node_document && (child == node_declaration || child == node_doctype)) return false;
+
+		return true;
+	}
+
+	PUGI__FN bool allow_move(xml_node parent, xml_node child)
+	{
+		// check that child can be a child of parent
+		if (!allow_insert_child(parent.type(), child.type()))
+			return false;
+
+		// check that node is not moved between documents
+		if (parent.root() != child.root())
+			return false;
+
+		// check that new parent is not in the child subtree
+		xml_node cur = parent;
+
+		while (cur)
+		{
+			if (cur == child)
+				return false;
+
+			cur = cur.parent();
+		}
+
+		return true;
+	}
+
+	template <typename String, typename Header>
+	PUGI__FN void node_copy_string(String& dest, Header& header, uintptr_t header_mask, char_t* source, Header& source_header, xml_allocator* alloc)
+	{
+		assert(!dest && (header & header_mask) == 0);
+
+		if (source)
+		{
+			if (alloc && (source_header & header_mask) == 0)
+			{
+				dest = source;
+
+				// since strcpy_insitu can reuse document buffer memory we need to mark both source and dest as shared
+				header |= xml_memory_page_contents_shared_mask;
+				source_header |= xml_memory_page_contents_shared_mask;
+			}
+			else
+				strcpy_insitu(dest, header, header_mask, source);
+		}
+	}
+
+	PUGI__FN void node_copy_contents(xml_node_struct* dn, xml_node_struct* sn, xml_allocator* shared_alloc)
+	{
+		node_copy_string(dn->name, dn->header, xml_memory_page_name_allocated_mask, sn->name, sn->header, shared_alloc);
+		node_copy_string(dn->value, dn->header, xml_memory_page_value_allocated_mask, sn->value, sn->header, shared_alloc);
+
+		for (xml_attribute_struct* sa = sn->first_attribute; sa; sa = sa->next_attribute)
+		{
+			xml_attribute_struct* da = append_new_attribute(dn, get_allocator(dn));
+
+			if (da)
+			{
+				node_copy_string(da->name, da->header, xml_memory_page_name_allocated_mask, sa->name, sa->header, shared_alloc);
+				node_copy_string(da->value, da->header, xml_memory_page_value_allocated_mask, sa->value, sa->header, shared_alloc);
+			}
+		}
+	}
+
+	PUGI__FN void node_copy_tree(xml_node_struct* dn, xml_node_struct* sn)
+	{
+		xml_allocator& alloc = get_allocator(dn);
+		xml_allocator* shared_alloc = (&alloc == &get_allocator(sn)) ? &alloc : 0;
+
+		node_copy_contents(dn, sn, shared_alloc);
+
+		xml_node_struct* dit = dn;
+		xml_node_struct* sit = sn->first_child;
+
+		while (sit && sit != sn)
+		{
+			if (sit != dn)
+			{
+				xml_node_struct* copy = append_new_node(dit, alloc, PUGI__NODETYPE(sit));
+
+				if (copy)
+				{
+					node_copy_contents(copy, sit, shared_alloc);
+
+					if (sit->first_child)
+					{
+						dit = copy;
+						sit = sit->first_child;
+						continue;
+					}
+				}
+			}
+
+			// continue to the next node
+			do
+			{
+				if (sit->next_sibling)
+				{
+					sit = sit->next_sibling;
+					break;
+				}
+
+				sit = sit->parent;
+				dit = dit->parent;
+			}
+			while (sit != sn);
+		}
+	}
+
+	PUGI__FN void node_copy_attribute(xml_attribute_struct* da, xml_attribute_struct* sa)
+	{
+		xml_allocator& alloc = get_allocator(da);
+		xml_allocator* shared_alloc = (&alloc == &get_allocator(sa)) ? &alloc : 0;
+
+		node_copy_string(da->name, da->header, xml_memory_page_name_allocated_mask, sa->name, sa->header, shared_alloc);
+		node_copy_string(da->value, da->header, xml_memory_page_value_allocated_mask, sa->value, sa->header, shared_alloc);
+	}
+
+	inline bool is_text_node(xml_node_struct* node)
+	{
+		xml_node_type type = PUGI__NODETYPE(node);
+
+		return type == node_pcdata || type == node_cdata;
+	}
+
+	// get value with conversion functions
+	PUGI__FN int get_integer_base(const char_t* value)
+	{
+		const char_t* s = value;
+
+		while (PUGI__IS_CHARTYPE(*s, ct_space))
+			s++;
+
+		if (*s == '-')
+			s++;
+
+		return (s[0] == '0' && (s[1] == 'x' || s[1] == 'X')) ? 16 : 10;
+	}
+
+	PUGI__FN int get_value_int(const char_t* value)
+	{
+		int base = get_integer_base(value);
+
+	#ifdef PUGIXML_WCHAR_MODE
+		return static_cast<int>(wcstol(value, 0, base));
+	#else
+		return static_cast<int>(strtol(value, 0, base));
+	#endif
+	}
+
+	PUGI__FN unsigned int get_value_uint(const char_t* value)
+	{
+		int base = get_integer_base(value);
+
+	#ifdef PUGIXML_WCHAR_MODE
+		return static_cast<unsigned int>(wcstoul(value, 0, base));
+	#else
+		return static_cast<unsigned int>(strtoul(value, 0, base));
+	#endif
+	}
+
+	PUGI__FN double get_value_double(const char_t* value)
+	{
+	#ifdef PUGIXML_WCHAR_MODE
+		return wcstod(value, 0);
+	#else
+		return strtod(value, 0);
+	#endif
+	}
+
+	PUGI__FN float get_value_float(const char_t* value)
+	{
+	#ifdef PUGIXML_WCHAR_MODE
+		return static_cast<float>(wcstod(value, 0));
+	#else
+		return static_cast<float>(strtod(value, 0));
+	#endif
+	}
+
+	PUGI__FN bool get_value_bool(const char_t* value)
+	{
+		// only look at first char
+		char_t first = *value;
+
+		// 1*, t* (true), T* (True), y* (yes), Y* (YES)
+		return (first == '1' || first == 't' || first == 'T' || first == 'y' || first == 'Y');
+	}
+
+#ifdef PUGIXML_HAS_LONG_LONG
+	PUGI__FN long long get_value_llong(const char_t* value)
+	{
+		int base = get_integer_base(value);
+
+	#ifdef PUGIXML_WCHAR_MODE
+		#ifdef PUGI__MSVC_CRT_VERSION
+			return _wcstoi64(value, 0, base);
+		#else
+			return wcstoll(value, 0, base);
+		#endif
+	#else
+		#ifdef PUGI__MSVC_CRT_VERSION
+			return _strtoi64(value, 0, base);
+		#else
+			return strtoll(value, 0, base);
+		#endif
+	#endif
+	}
+
+	PUGI__FN unsigned long long get_value_ullong(const char_t* value)
+	{
+		int base = get_integer_base(value);
+
+	#ifdef PUGIXML_WCHAR_MODE
+		#ifdef PUGI__MSVC_CRT_VERSION
+			return _wcstoui64(value, 0, base);
+		#else
+			return wcstoull(value, 0, base);
+		#endif
+	#else
+		#ifdef PUGI__MSVC_CRT_VERSION
+			return _strtoui64(value, 0, base);
+		#else
+			return strtoull(value, 0, base);
+		#endif
+	#endif
+	}
+#endif
+
+	// set value with conversion functions
+	template <typename String, typename Header>
+	PUGI__FN bool set_value_buffer(String& dest, Header& header, uintptr_t header_mask, char (&buf)[128])
+	{
+	#ifdef PUGIXML_WCHAR_MODE
+		char_t wbuf[128];
+		impl::widen_ascii(wbuf, buf);
+
+		return strcpy_insitu(dest, header, header_mask, wbuf);
+	#else
+		return strcpy_insitu(dest, header, header_mask, buf);
+	#endif
+	}
+
+	template <typename String, typename Header>
+	PUGI__FN bool set_value_convert(String& dest, Header& header, uintptr_t header_mask, int value)
+	{
+		char buf[128];
+		sprintf(buf, "%d", value);
+	
+		return set_value_buffer(dest, header, header_mask, buf);
+	}
+
+	template <typename String, typename Header>
+	PUGI__FN bool set_value_convert(String& dest, Header& header, uintptr_t header_mask, unsigned int value)
+	{
+		char buf[128];
+		sprintf(buf, "%u", value);
+
+		return set_value_buffer(dest, header, header_mask, buf);
+	}
+
+	template <typename String, typename Header>
+	PUGI__FN bool set_value_convert(String& dest, Header& header, uintptr_t header_mask, float value)
+	{
+		char buf[128];
+		sprintf(buf, "%.9g", value);
+
+		return set_value_buffer(dest, header, header_mask, buf);
+	}
+
+	template <typename String, typename Header>
+	PUGI__FN bool set_value_convert(String& dest, Header& header, uintptr_t header_mask, double value)
+	{
+		char buf[128];
+		sprintf(buf, "%.17g", value);
+
+		return set_value_buffer(dest, header, header_mask, buf);
+	}
+	
+	template <typename String, typename Header>
+	PUGI__FN bool set_value_convert(String& dest, Header& header, uintptr_t header_mask, bool value)
+	{
+		return strcpy_insitu(dest, header, header_mask, value ? PUGIXML_TEXT("true") : PUGIXML_TEXT("false"));
+	}
+
+#ifdef PUGIXML_HAS_LONG_LONG
+	template <typename String, typename Header>
+	PUGI__FN bool set_value_convert(String& dest, Header& header, uintptr_t header_mask, long long value)
+	{
+		char buf[128];
+		sprintf(buf, "%lld", value);
+	
+		return set_value_buffer(dest, header, header_mask, buf);
+	}
+
+	template <typename String, typename Header>
+	PUGI__FN bool set_value_convert(String& dest, Header& header, uintptr_t header_mask, unsigned long long value)
+	{
+		char buf[128];
+		sprintf(buf, "%llu", value);
+	
+		return set_value_buffer(dest, header, header_mask, buf);
+	}
+#endif
+
+	// we need to get length of entire file to load it in memory; the only (relatively) sane way to do it is via seek/tell trick
+	PUGI__FN xml_parse_status get_file_size(FILE* file, size_t& out_result)
+	{
+	#if defined(PUGI__MSVC_CRT_VERSION) && PUGI__MSVC_CRT_VERSION >= 1400 && !defined(_WIN32_WCE)
+		// there are 64-bit versions of fseek/ftell, let's use them
+		typedef __int64 length_type;
+
+		_fseeki64(file, 0, SEEK_END);
+		length_type length = _ftelli64(file);
+		_fseeki64(file, 0, SEEK_SET);
+	#elif defined(__MINGW32__) && !defined(__NO_MINGW_LFS) && (!defined(__STRICT_ANSI__) || defined(__MINGW64_VERSION_MAJOR))
+		// there are 64-bit versions of fseek/ftell, let's use them
+		typedef off64_t length_type;
+
+		fseeko64(file, 0, SEEK_END);
+		length_type length = ftello64(file);
+		fseeko64(file, 0, SEEK_SET);
+	#else
+		// if this is a 32-bit OS, long is enough; if this is a unix system, long is 64-bit, which is enough; otherwise we can't do anything anyway.
+		typedef long length_type;
+
+		fseek(file, 0, SEEK_END);
+		length_type length = ftell(file);
+		fseek(file, 0, SEEK_SET);
+	#endif
+
+		// check for I/O errors
+		if (length < 0) return status_io_error;
+		
+		// check for overflow
+		size_t result = static_cast<size_t>(length);
+
+		if (static_cast<length_type>(result) != length) return status_out_of_memory;
+
+		// finalize
+		out_result = result;
+
+		return status_ok;
+	}
+
+	// This function assumes that buffer has extra sizeof(char_t) writable bytes after size
+	PUGI__FN size_t zero_terminate_buffer(void* buffer, size_t size, xml_encoding encoding) 
+	{
+		// We only need to zero-terminate if encoding conversion does not do it for us
+	#ifdef PUGIXML_WCHAR_MODE
+		xml_encoding wchar_encoding = get_wchar_encoding();
+
+		if (encoding == wchar_encoding || need_endian_swap_utf(encoding, wchar_encoding))
+		{
+			size_t length = size / sizeof(char_t);
+
+			static_cast<char_t*>(buffer)[length] = 0;
+			return (length + 1) * sizeof(char_t);
+		}
+	#else
+		if (encoding == encoding_utf8)
+		{
+			static_cast<char*>(buffer)[size] = 0;
+			return size + 1;
+		}
+	#endif
+
+		return size;
+	}
+
+	PUGI__FN xml_parse_result load_file_impl(xml_document& doc, FILE* file, unsigned int options, xml_encoding encoding)
+	{
+		if (!file) return make_parse_result(status_file_not_found);
+
+		// get file size (can result in I/O errors)
+		size_t size = 0;
+		xml_parse_status size_status = get_file_size(file, size);
+		if (size_status != status_ok) return make_parse_result(size_status);
+		
+		size_t max_suffix_size = sizeof(char_t);
+
+		// allocate buffer for the whole file
+		char* contents = static_cast<char*>(xml_memory::allocate(size + max_suffix_size));
+		if (!contents) return make_parse_result(status_out_of_memory);
+
+		// read file in memory
+		size_t read_size = fread(contents, 1, size, file);
+
+		if (read_size != size)
+		{
+			xml_memory::deallocate(contents);
+			return make_parse_result(status_io_error);
+		}
+
+		xml_encoding real_encoding = get_buffer_encoding(encoding, contents, size);
+		
+		return doc.load_buffer_inplace_own(contents, zero_terminate_buffer(contents, size, real_encoding), options, real_encoding);
+	}
+
+#ifndef PUGIXML_NO_STL
+	template <typename T> struct xml_stream_chunk
+	{
+		static xml_stream_chunk* create()
+		{
+			void* memory = xml_memory::allocate(sizeof(xml_stream_chunk));
+			if (!memory) return 0;
+			
+			return new (memory) xml_stream_chunk();
+		}
+
+		static void destroy(xml_stream_chunk* chunk)
+		{
+			// free chunk chain
+			while (chunk)
+			{
+				xml_stream_chunk* next_ = chunk->next;
+
+				xml_memory::deallocate(chunk);
+
+				chunk = next_;
+			}
+		}
+
+		xml_stream_chunk(): next(0), size(0)
+		{
+		}
+
+		xml_stream_chunk* next;
+		size_t size;
+
+		T data[xml_memory_page_size / sizeof(T)];
+	};
+
+	template <typename T> PUGI__FN xml_parse_status load_stream_data_noseek(std::basic_istream<T>& stream, void** out_buffer, size_t* out_size)
+	{
+		auto_deleter<xml_stream_chunk<T> > chunks(0, xml_stream_chunk<T>::destroy);
+
+		// read file to a chunk list
+		size_t total = 0;
+		xml_stream_chunk<T>* last = 0;
+
+		while (!stream.eof())
+		{
+			// allocate new chunk
+			xml_stream_chunk<T>* chunk = xml_stream_chunk<T>::create();
+			if (!chunk) return status_out_of_memory;
+
+			// append chunk to list
+			if (last) last = last->next = chunk;
+			else chunks.data = last = chunk;
+
+			// read data to chunk
+			stream.read(chunk->data, static_cast<std::streamsize>(sizeof(chunk->data) / sizeof(T)));
+			chunk->size = static_cast<size_t>(stream.gcount()) * sizeof(T);
+
+			// read may set failbit | eofbit in case gcount() is less than read length, so check for other I/O errors
+			if (stream.bad() || (!stream.eof() && stream.fail())) return status_io_error;
+
+			// guard against huge files (chunk size is small enough to make this overflow check work)
+			if (total + chunk->size < total) return status_out_of_memory;
+			total += chunk->size;
+		}
+
+		size_t max_suffix_size = sizeof(char_t);
+
+		// copy chunk list to a contiguous buffer
+		char* buffer = static_cast<char*>(xml_memory::allocate(total + max_suffix_size));
+		if (!buffer) return status_out_of_memory;
+
+		char* write = buffer;
+
+		for (xml_stream_chunk<T>* chunk = chunks.data; chunk; chunk = chunk->next)
+		{
+			assert(write + chunk->size <= buffer + total);
+			memcpy(write, chunk->data, chunk->size);
+			write += chunk->size;
+		}
+
+		assert(write == buffer + total);
+
+		// return buffer
+		*out_buffer = buffer;
+		*out_size = total;
+
+		return status_ok;
+	}
+
+	template <typename T> PUGI__FN xml_parse_status load_stream_data_seek(std::basic_istream<T>& stream, void** out_buffer, size_t* out_size)
+	{
+		// get length of remaining data in stream
+		typename std::basic_istream<T>::pos_type pos = stream.tellg();
+		stream.seekg(0, std::ios::end);
+		std::streamoff length = stream.tellg() - pos;
+		stream.seekg(pos);
+
+		if (stream.fail() || pos < 0) return status_io_error;
+
+		// guard against huge files
+		size_t read_length = static_cast<size_t>(length);
+
+		if (static_cast<std::streamsize>(read_length) != length || length < 0) return status_out_of_memory;
+
+		size_t max_suffix_size = sizeof(char_t);
+
+		// read stream data into memory (guard against stream exceptions with buffer holder)
+		auto_deleter<void> buffer(xml_memory::allocate(read_length * sizeof(T) + max_suffix_size), xml_memory::deallocate);
+		if (!buffer.data) return status_out_of_memory;
+
+		stream.read(static_cast<T*>(buffer.data), static_cast<std::streamsize>(read_length));
+
+		// read may set failbit | eofbit in case gcount() is less than read_length (i.e. line ending conversion), so check for other I/O errors
+		if (stream.bad() || (!stream.eof() && stream.fail())) return status_io_error;
+
+		// return buffer
+		size_t actual_length = static_cast<size_t>(stream.gcount());
+		assert(actual_length <= read_length);
+		
+		*out_buffer = buffer.release();
+		*out_size = actual_length * sizeof(T);
+
+		return status_ok;
+	}
+
+	template <typename T> PUGI__FN xml_parse_result load_stream_impl(xml_document& doc, std::basic_istream<T>& stream, unsigned int options, xml_encoding encoding)
+	{
+		void* buffer = 0;
+		size_t size = 0;
+		xml_parse_status status = status_ok;
+
+		// if stream has an error bit set, bail out (otherwise tellg() can fail and we'll clear error bits)
+		if (stream.fail()) return make_parse_result(status_io_error);
+
+		// load stream to memory (using seek-based implementation if possible, since it's faster and takes less memory)
+		if (stream.tellg() < 0)
+		{
+			stream.clear(); // clear error flags that could be set by a failing tellg
+			status = load_stream_data_noseek(stream, &buffer, &size);
+		}
+		else
+			status = load_stream_data_seek(stream, &buffer, &size);
+
+		if (status != status_ok) return make_parse_result(status);
+
+		xml_encoding real_encoding = get_buffer_encoding(encoding, buffer, size);
+		
+		return doc.load_buffer_inplace_own(buffer, zero_terminate_buffer(buffer, size, real_encoding), options, real_encoding);
+	}
+#endif
+
+#if defined(PUGI__MSVC_CRT_VERSION) || defined(__BORLANDC__) || (defined(__MINGW32__) && (!defined(__STRICT_ANSI__) || defined(__MINGW64_VERSION_MAJOR)))
+	PUGI__FN FILE* open_file_wide(const wchar_t* path, const wchar_t* mode)
+	{
+		return _wfopen(path, mode);
+	}
+#else
+	PUGI__FN char* convert_path_heap(const wchar_t* str)
+	{
+		assert(str);
+
+		// first pass: get length in utf8 characters
+		size_t length = strlength_wide(str);
+		size_t size = as_utf8_begin(str, length);
+
+		// allocate resulting string
+		char* result = static_cast<char*>(xml_memory::allocate(size + 1));
+		if (!result) return 0;
+
+		// second pass: convert to utf8
+		as_utf8_end(result, size, str, length);
+
+		// zero-terminate
+		result[size] = 0;
+
+		return result;
+	}
+
+	PUGI__FN FILE* open_file_wide(const wchar_t* path, const wchar_t* mode)
+	{
+		// there is no standard function to open wide paths, so our best bet is to try utf8 path
+		char* path_utf8 = convert_path_heap(path);
+		if (!path_utf8) return 0;
+
+		// convert mode to ASCII (we mirror _wfopen interface)
+		char mode_ascii[4] = {0};
+		for (size_t i = 0; mode[i]; ++i) mode_ascii[i] = static_cast<char>(mode[i]);
+
+		// try to open the utf8 path
+		FILE* result = fopen(path_utf8, mode_ascii);
+
+		// free dummy buffer
+		xml_memory::deallocate(path_utf8);
+
+		return result;
+	}
+#endif
+
+	PUGI__FN bool save_file_impl(const xml_document& doc, FILE* file, const char_t* indent, unsigned int flags, xml_encoding encoding)
+	{
+		if (!file) return false;
+
+		xml_writer_file writer(file);
+		doc.save(writer, indent, flags, encoding);
+
+		return ferror(file) == 0;
+	}
+
+	PUGI__FN xml_parse_result load_buffer_impl(xml_document_struct* doc, xml_node_struct* root, void* contents, size_t size, unsigned int options, xml_encoding encoding, bool is_mutable, bool own, char_t** out_buffer)
+	{
+		// check input buffer
+		if (!contents && size) return make_parse_result(status_io_error);
+
+		// get actual encoding
+		xml_encoding buffer_encoding = impl::get_buffer_encoding(encoding, contents, size);
+
+		// get private buffer
+		char_t* buffer = 0;
+		size_t length = 0;
+
+		if (!impl::convert_buffer(buffer, length, buffer_encoding, contents, size, is_mutable)) return impl::make_parse_result(status_out_of_memory);
+		
+		// delete original buffer if we performed a conversion
+		if (own && buffer != contents && contents) impl::xml_memory::deallocate(contents);
+
+		// grab onto buffer if it's our buffer, user is responsible for deallocating contents himself
+		if (own || buffer != contents) *out_buffer = buffer;
+
+		// store buffer for offset_debug
+		doc->buffer = buffer;
+
+		// parse
+		xml_parse_result res = impl::xml_parser::parse(buffer, length, doc, root, options);
+
+		// remember encoding
+		res.encoding = buffer_encoding;
+
+		return res;
+	}
+PUGI__NS_END
+
+namespace pugi
+{
+	PUGI__FN xml_writer_file::xml_writer_file(void* file_): file(file_)
+	{
+	}
+
+	PUGI__FN void xml_writer_file::write(const void* data, size_t size)
+	{
+		size_t result = fwrite(data, 1, size, static_cast<FILE*>(file));
+		(void)!result; // unfortunately we can't do proper error handling here
+	}
+
+#ifndef PUGIXML_NO_STL
+	PUGI__FN xml_writer_stream::xml_writer_stream(std::basic_ostream<char, std::char_traits<char> >& stream): narrow_stream(&stream), wide_stream(0)
+	{
+	}
+
+	PUGI__FN xml_writer_stream::xml_writer_stream(std::basic_ostream<wchar_t, std::char_traits<wchar_t> >& stream): narrow_stream(0), wide_stream(&stream)
+	{
+	}
+
+	PUGI__FN void xml_writer_stream::write(const void* data, size_t size)
+	{
+		if (narrow_stream)
+		{
+			assert(!wide_stream);
+			narrow_stream->write(reinterpret_cast<const char*>(data), static_cast<std::streamsize>(size));
+		}
+		else
+		{
+			assert(wide_stream);
+			assert(size % sizeof(wchar_t) == 0);
+
+			wide_stream->write(reinterpret_cast<const wchar_t*>(data), static_cast<std::streamsize>(size / sizeof(wchar_t)));
+		}
+	}
+#endif
+
+	PUGI__FN xml_tree_walker::xml_tree_walker(): _depth(0)
+	{
+	}
+	
+	PUGI__FN xml_tree_walker::~xml_tree_walker()
+	{
+	}
+
+	PUGI__FN int xml_tree_walker::depth() const
+	{
+		return _depth;
+	}
+
+	PUGI__FN bool xml_tree_walker::begin(xml_node&)
+	{
+		return true;
+	}
+
+	PUGI__FN bool xml_tree_walker::end(xml_node&)
+	{
+		return true;
+	}
+
+	PUGI__FN xml_attribute::xml_attribute(): _attr(0)
+	{
+	}
+
+	PUGI__FN xml_attribute::xml_attribute(xml_attribute_struct* attr): _attr(attr)
+	{
+	}
+
+	PUGI__FN static void unspecified_bool_xml_attribute(xml_attribute***)
+	{
+	}
+
+	PUGI__FN xml_attribute::operator xml_attribute::unspecified_bool_type() const
+	{
+		return _attr ? unspecified_bool_xml_attribute : 0;
+	}
+
+	PUGI__FN bool xml_attribute::operator!() const
+	{
+		return !_attr;
+	}
+
+	PUGI__FN bool xml_attribute::operator==(const xml_attribute& r) const
+	{
+		return (_attr == r._attr);
+	}
+	
+	PUGI__FN bool xml_attribute::operator!=(const xml_attribute& r) const
+	{
+		return (_attr != r._attr);
+	}
+
+	PUGI__FN bool xml_attribute::operator<(const xml_attribute& r) const
+	{
+		return (_attr < r._attr);
+	}
+	
+	PUGI__FN bool xml_attribute::operator>(const xml_attribute& r) const
+	{
+		return (_attr > r._attr);
+	}
+	
+	PUGI__FN bool xml_attribute::operator<=(const xml_attribute& r) const
+	{
+		return (_attr <= r._attr);
+	}
+	
+	PUGI__FN bool xml_attribute::operator>=(const xml_attribute& r) const
+	{
+		return (_attr >= r._attr);
+	}
+
+	PUGI__FN xml_attribute xml_attribute::next_attribute() const
+	{
+		return _attr ? xml_attribute(_attr->next_attribute) : xml_attribute();
+	}
+
+	PUGI__FN xml_attribute xml_attribute::previous_attribute() const
+	{
+		return _attr && _attr->prev_attribute_c->next_attribute ? xml_attribute(_attr->prev_attribute_c) : xml_attribute();
+	}
+
+	PUGI__FN const char_t* xml_attribute::as_string(const char_t* def) const
+	{
+		return (_attr && _attr->value) ? _attr->value : def;
+	}
+
+	PUGI__FN int xml_attribute::as_int(int def) const
+	{
+		return (_attr && _attr->value) ? impl::get_value_int(_attr->value) : def;
+	}
+
+	PUGI__FN unsigned int xml_attribute::as_uint(unsigned int def) const
+	{
+		return (_attr && _attr->value) ? impl::get_value_uint(_attr->value) : def;
+	}
+
+	PUGI__FN double xml_attribute::as_double(double def) const
+	{
+		return (_attr && _attr->value) ? impl::get_value_double(_attr->value) : def;
+	}
+
+	PUGI__FN float xml_attribute::as_float(float def) const
+	{
+		return (_attr && _attr->value) ? impl::get_value_float(_attr->value) : def;
+	}
+
+	PUGI__FN bool xml_attribute::as_bool(bool def) const
+	{
+		return (_attr && _attr->value) ? impl::get_value_bool(_attr->value) : def;
+	}
+
+#ifdef PUGIXML_HAS_LONG_LONG
+	PUGI__FN long long xml_attribute::as_llong(long long def) const
+	{
+		return (_attr && _attr->value) ? impl::get_value_llong(_attr->value) : def;
+	}
+
+	PUGI__FN unsigned long long xml_attribute::as_ullong(unsigned long long def) const
+	{
+		return (_attr && _attr->value) ? impl::get_value_ullong(_attr->value) : def;
+	}
+#endif
+
+	PUGI__FN bool xml_attribute::empty() const
+	{
+		return !_attr;
+	}
+
+	PUGI__FN const char_t* xml_attribute::name() const
+	{
+		return (_attr && _attr->name) ? _attr->name + 0 : PUGIXML_TEXT("");
+	}
+
+	PUGI__FN const char_t* xml_attribute::value() const
+	{
+		return (_attr && _attr->value) ? _attr->value + 0 : PUGIXML_TEXT("");
+	}
+
+	PUGI__FN size_t xml_attribute::hash_value() const
+	{
+		return static_cast<size_t>(reinterpret_cast<uintptr_t>(_attr) / sizeof(xml_attribute_struct));
+	}
+
+	PUGI__FN xml_attribute_struct* xml_attribute::internal_object() const
+	{
+		return _attr;
+	}
+
+	PUGI__FN xml_attribute& xml_attribute::operator=(const char_t* rhs)
+	{
+		set_value(rhs);
+		return *this;
+	}
+	
+	PUGI__FN xml_attribute& xml_attribute::operator=(int rhs)
+	{
+		set_value(rhs);
+		return *this;
+	}
+
+	PUGI__FN xml_attribute& xml_attribute::operator=(unsigned int rhs)
+	{
+		set_value(rhs);
+		return *this;
+	}
+
+	PUGI__FN xml_attribute& xml_attribute::operator=(double rhs)
+	{
+		set_value(rhs);
+		return *this;
+	}
+	
+	PUGI__FN xml_attribute& xml_attribute::operator=(float rhs)
+	{
+		set_value(rhs);
+		return *this;
+	}
+
+	PUGI__FN xml_attribute& xml_attribute::operator=(bool rhs)
+	{
+		set_value(rhs);
+		return *this;
+	}
+
+#ifdef PUGIXML_HAS_LONG_LONG
+	PUGI__FN xml_attribute& xml_attribute::operator=(long long rhs)
+	{
+		set_value(rhs);
+		return *this;
+	}
+
+	PUGI__FN xml_attribute& xml_attribute::operator=(unsigned long long rhs)
+	{
+		set_value(rhs);
+		return *this;
+	}
+#endif
+
+	PUGI__FN bool xml_attribute::set_name(const char_t* rhs)
+	{
+		if (!_attr) return false;
+		
+		return impl::strcpy_insitu(_attr->name, _attr->header, impl::xml_memory_page_name_allocated_mask, rhs);
+	}
+		
+	PUGI__FN bool xml_attribute::set_value(const char_t* rhs)
+	{
+		if (!_attr) return false;
+
+		return impl::strcpy_insitu(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs);
+	}
+
+	PUGI__FN bool xml_attribute::set_value(int rhs)
+	{
+		if (!_attr) return false;
+
+		return impl::set_value_convert(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs);
+	}
+
+	PUGI__FN bool xml_attribute::set_value(unsigned int rhs)
+	{
+		if (!_attr) return false;
+
+		return impl::set_value_convert(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs);
+	}
+
+	PUGI__FN bool xml_attribute::set_value(double rhs)
+	{
+		if (!_attr) return false;
+
+		return impl::set_value_convert(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs);
+	}
+	
+	PUGI__FN bool xml_attribute::set_value(float rhs)
+	{
+		if (!_attr) return false;
+
+		return impl::set_value_convert(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs);
+	}
+
+	PUGI__FN bool xml_attribute::set_value(bool rhs)
+	{
+		if (!_attr) return false;
+
+		return impl::set_value_convert(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs);
+	}
+
+#ifdef PUGIXML_HAS_LONG_LONG
+	PUGI__FN bool xml_attribute::set_value(long long rhs)
+	{
+		if (!_attr) return false;
+
+		return impl::set_value_convert(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs);
+	}
+
+	PUGI__FN bool xml_attribute::set_value(unsigned long long rhs)
+	{
+		if (!_attr) return false;
+
+		return impl::set_value_convert(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs);
+	}
+#endif
+
+#ifdef __BORLANDC__
+	PUGI__FN bool operator&&(const xml_attribute& lhs, bool rhs)
+	{
+		return (bool)lhs && rhs;
+	}
+
+	PUGI__FN bool operator||(const xml_attribute& lhs, bool rhs)
+	{
+		return (bool)lhs || rhs;
+	}
+#endif
+
+	PUGI__FN xml_node::xml_node(): _root(0)
+	{
+	}
+
+	PUGI__FN xml_node::xml_node(xml_node_struct* p): _root(p)
+	{
+	}
+	
+	PUGI__FN static void unspecified_bool_xml_node(xml_node***)
+	{
+	}
+
+	PUGI__FN xml_node::operator xml_node::unspecified_bool_type() const
+	{
+		return _root ? unspecified_bool_xml_node : 0;
+	}
+
+	PUGI__FN bool xml_node::operator!() const
+	{
+		return !_root;
+	}
+
+	PUGI__FN xml_node::iterator xml_node::begin() const
+	{
+		return iterator(_root ? _root->first_child + 0 : 0, _root);
+	}
+
+	PUGI__FN xml_node::iterator xml_node::end() const
+	{
+		return iterator(0, _root);
+	}
+	
+	PUGI__FN xml_node::attribute_iterator xml_node::attributes_begin() const
+	{
+		return attribute_iterator(_root ? _root->first_attribute + 0 : 0, _root);
+	}
+
+	PUGI__FN xml_node::attribute_iterator xml_node::attributes_end() const
+	{
+		return attribute_iterator(0, _root);
+	}
+	
+	PUGI__FN xml_object_range<xml_node_iterator> xml_node::children() const
+	{
+		return xml_object_range<xml_node_iterator>(begin(), end());
+	}
+
+	PUGI__FN xml_object_range<xml_named_node_iterator> xml_node::children(const char_t* name_) const
+	{
+		return xml_object_range<xml_named_node_iterator>(xml_named_node_iterator(child(name_)._root, _root, name_), xml_named_node_iterator(0, _root, name_));
+	}
+
+	PUGI__FN xml_object_range<xml_attribute_iterator> xml_node::attributes() const
+	{
+		return xml_object_range<xml_attribute_iterator>(attributes_begin(), attributes_end());
+	}
+
+	PUGI__FN bool xml_node::operator==(const xml_node& r) const
+	{
+		return (_root == r._root);
+	}
+
+	PUGI__FN bool xml_node::operator!=(const xml_node& r) const
+	{
+		return (_root != r._root);
+	}
+
+	PUGI__FN bool xml_node::operator<(const xml_node& r) const
+	{
+		return (_root < r._root);
+	}
+	
+	PUGI__FN bool xml_node::operator>(const xml_node& r) const
+	{
+		return (_root > r._root);
+	}
+	
+	PUGI__FN bool xml_node::operator<=(const xml_node& r) const
+	{
+		return (_root <= r._root);
+	}
+	
+	PUGI__FN bool xml_node::operator>=(const xml_node& r) const
+	{
+		return (_root >= r._root);
+	}
+
+	PUGI__FN bool xml_node::empty() const
+	{
+		return !_root;
+	}
+	
+	PUGI__FN const char_t* xml_node::name() const
+	{
+		return (_root && _root->name) ? _root->name + 0 : PUGIXML_TEXT("");
+	}
+
+	PUGI__FN xml_node_type xml_node::type() const
+	{
+		return _root ? PUGI__NODETYPE(_root) : node_null;
+	}
+	
+	PUGI__FN const char_t* xml_node::value() const
+	{
+		return (_root && _root->value) ? _root->value + 0 : PUGIXML_TEXT("");
+	}
+	
+	PUGI__FN xml_node xml_node::child(const char_t* name_) const
+	{
+		if (!_root) return xml_node();
+
+		for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling)
+			if (i->name && impl::strequal(name_, i->name)) return xml_node(i);
+
+		return xml_node();
+	}
+
+	PUGI__FN xml_attribute xml_node::attribute(const char_t* name_) const
+	{
+		if (!_root) return xml_attribute();
+
+		for (xml_attribute_struct* i = _root->first_attribute; i; i = i->next_attribute)
+			if (i->name && impl::strequal(name_, i->name))
+				return xml_attribute(i);
+		
+		return xml_attribute();
+	}
+	
+	PUGI__FN xml_node xml_node::next_sibling(const char_t* name_) const
+	{
+		if (!_root) return xml_node();
+		
+		for (xml_node_struct* i = _root->next_sibling; i; i = i->next_sibling)
+			if (i->name && impl::strequal(name_, i->name)) return xml_node(i);
+
+		return xml_node();
+	}
+
+	PUGI__FN xml_node xml_node::next_sibling() const
+	{
+		return _root ? xml_node(_root->next_sibling) : xml_node();
+	}
+
+	PUGI__FN xml_node xml_node::previous_sibling(const char_t* name_) const
+	{
+		if (!_root) return xml_node();
+		
+		for (xml_node_struct* i = _root->prev_sibling_c; i->next_sibling; i = i->prev_sibling_c)
+			if (i->name && impl::strequal(name_, i->name)) return xml_node(i);
+
+		return xml_node();
+	}
+
+	PUGI__FN xml_attribute xml_node::attribute(const char_t* name_, xml_attribute& hint_) const
+	{
+		xml_attribute_struct* hint = hint_._attr;
+
+		// if hint is not an attribute of node, behavior is not defined
+		assert(!hint || (_root && impl::is_attribute_of(hint, _root)));
+
+		if (!_root) return xml_attribute();
+
+		// optimistically search from hint up until the end
+		for (xml_attribute_struct* i = hint; i; i = i->next_attribute)
+			if (i->name && impl::strequal(name_, i->name))
+			{
+				// update hint to maximize efficiency of searching for consecutive attributes
+				hint_._attr = i->next_attribute;
+
+				return xml_attribute(i);
+			}
+
+		// wrap around and search from the first attribute until the hint
+		// 'j' null pointer check is technically redundant, but it prevents a crash in case the assertion above fails
+		for (xml_attribute_struct* j = _root->first_attribute; j && j != hint; j = j->next_attribute)
+			if (j->name && impl::strequal(name_, j->name))
+			{
+				// update hint to maximize efficiency of searching for consecutive attributes
+				hint_._attr = j->next_attribute;
+
+				return xml_attribute(j);
+			}
+
+		return xml_attribute();
+	}
+
+	PUGI__FN xml_node xml_node::previous_sibling() const
+	{
+		if (!_root) return xml_node();
+		
+		if (_root->prev_sibling_c->next_sibling) return xml_node(_root->prev_sibling_c);
+		else return xml_node();
+	}
+
+	PUGI__FN xml_node xml_node::parent() const
+	{
+		return _root ? xml_node(_root->parent) : xml_node();
+	}
+
+	PUGI__FN xml_node xml_node::root() const
+	{
+		return _root ? xml_node(&impl::get_document(_root)) : xml_node();
+	}
+
+	PUGI__FN xml_text xml_node::text() const
+	{
+		return xml_text(_root);
+	}
+
+	PUGI__FN const char_t* xml_node::child_value() const
+	{
+		if (!_root) return PUGIXML_TEXT("");
+		
+		for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling)
+			if (impl::is_text_node(i) && i->value)
+				return i->value;
+
+		return PUGIXML_TEXT("");
+	}
+
+	PUGI__FN const char_t* xml_node::child_value(const char_t* name_) const
+	{
+		return child(name_).child_value();
+	}
+
+	PUGI__FN xml_attribute xml_node::first_attribute() const
+	{
+		return _root ? xml_attribute(_root->first_attribute) : xml_attribute();
+	}
+
+	PUGI__FN xml_attribute xml_node::last_attribute() const
+	{
+		return _root && _root->first_attribute ? xml_attribute(_root->first_attribute->prev_attribute_c) : xml_attribute();
+	}
+
+	PUGI__FN xml_node xml_node::first_child() const
+	{
+		return _root ? xml_node(_root->first_child) : xml_node();
+	}
+
+	PUGI__FN xml_node xml_node::last_child() const
+	{
+		return _root && _root->first_child ? xml_node(_root->first_child->prev_sibling_c) : xml_node();
+	}
+
+	PUGI__FN bool xml_node::set_name(const char_t* rhs)
+	{
+		static const bool has_name[] = { false, false, true, false, false, false, true, true, false };
+
+		if (!_root || !has_name[PUGI__NODETYPE(_root)])
+			return false;
+
+		return impl::strcpy_insitu(_root->name, _root->header, impl::xml_memory_page_name_allocated_mask, rhs);
+	}
+		
+	PUGI__FN bool xml_node::set_value(const char_t* rhs)
+	{
+		static const bool has_value[] = { false, false, false, true, true, true, true, false, true };
+
+		if (!_root || !has_value[PUGI__NODETYPE(_root)])
+			return false;
+
+		return impl::strcpy_insitu(_root->value, _root->header, impl::xml_memory_page_value_allocated_mask, rhs);
+	}
+
+	PUGI__FN xml_attribute xml_node::append_attribute(const char_t* name_)
+	{
+		if (!impl::allow_insert_attribute(type())) return xml_attribute();
+		
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_attribute();
+
+		xml_attribute a(impl::allocate_attribute(alloc));
+		if (!a) return xml_attribute();
+
+		impl::append_attribute(a._attr, _root);
+
+		a.set_name(name_);
+		
+		return a;
+	}
+
+	PUGI__FN xml_attribute xml_node::prepend_attribute(const char_t* name_)
+	{
+		if (!impl::allow_insert_attribute(type())) return xml_attribute();
+		
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_attribute();
+
+		xml_attribute a(impl::allocate_attribute(alloc));
+		if (!a) return xml_attribute();
+
+		impl::prepend_attribute(a._attr, _root);
+
+		a.set_name(name_);
+
+		return a;
+	}
+
+	PUGI__FN xml_attribute xml_node::insert_attribute_after(const char_t* name_, const xml_attribute& attr)
+	{
+		if (!impl::allow_insert_attribute(type())) return xml_attribute();
+		if (!attr || !impl::is_attribute_of(attr._attr, _root)) return xml_attribute();
+		
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_attribute();
+
+		xml_attribute a(impl::allocate_attribute(alloc));
+		if (!a) return xml_attribute();
+
+		impl::insert_attribute_after(a._attr, attr._attr, _root);
+
+		a.set_name(name_);
+
+		return a;
+	}
+
+	PUGI__FN xml_attribute xml_node::insert_attribute_before(const char_t* name_, const xml_attribute& attr)
+	{
+		if (!impl::allow_insert_attribute(type())) return xml_attribute();
+		if (!attr || !impl::is_attribute_of(attr._attr, _root)) return xml_attribute();
+		
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_attribute();
+
+		xml_attribute a(impl::allocate_attribute(alloc));
+		if (!a) return xml_attribute();
+
+		impl::insert_attribute_before(a._attr, attr._attr, _root);
+
+		a.set_name(name_);
+
+		return a;
+	}
+
+	PUGI__FN xml_attribute xml_node::append_copy(const xml_attribute& proto)
+	{
+		if (!proto) return xml_attribute();
+		if (!impl::allow_insert_attribute(type())) return xml_attribute();
+
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_attribute();
+
+		xml_attribute a(impl::allocate_attribute(alloc));
+		if (!a) return xml_attribute();
+
+		impl::append_attribute(a._attr, _root);
+		impl::node_copy_attribute(a._attr, proto._attr);
+
+		return a;
+	}
+
+	PUGI__FN xml_attribute xml_node::prepend_copy(const xml_attribute& proto)
+	{
+		if (!proto) return xml_attribute();
+		if (!impl::allow_insert_attribute(type())) return xml_attribute();
+
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_attribute();
+
+		xml_attribute a(impl::allocate_attribute(alloc));
+		if (!a) return xml_attribute();
+
+		impl::prepend_attribute(a._attr, _root);
+		impl::node_copy_attribute(a._attr, proto._attr);
+
+		return a;
+	}
+
+	PUGI__FN xml_attribute xml_node::insert_copy_after(const xml_attribute& proto, const xml_attribute& attr)
+	{
+		if (!proto) return xml_attribute();
+		if (!impl::allow_insert_attribute(type())) return xml_attribute();
+		if (!attr || !impl::is_attribute_of(attr._attr, _root)) return xml_attribute();
+
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_attribute();
+
+		xml_attribute a(impl::allocate_attribute(alloc));
+		if (!a) return xml_attribute();
+
+		impl::insert_attribute_after(a._attr, attr._attr, _root);
+		impl::node_copy_attribute(a._attr, proto._attr);
+
+		return a;
+	}
+
+	PUGI__FN xml_attribute xml_node::insert_copy_before(const xml_attribute& proto, const xml_attribute& attr)
+	{
+		if (!proto) return xml_attribute();
+		if (!impl::allow_insert_attribute(type())) return xml_attribute();
+		if (!attr || !impl::is_attribute_of(attr._attr, _root)) return xml_attribute();
+
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_attribute();
+
+		xml_attribute a(impl::allocate_attribute(alloc));
+		if (!a) return xml_attribute();
+
+		impl::insert_attribute_before(a._attr, attr._attr, _root);
+		impl::node_copy_attribute(a._attr, proto._attr);
+
+		return a;
+	}
+
+	PUGI__FN xml_node xml_node::append_child(xml_node_type type_)
+	{
+		if (!impl::allow_insert_child(type(), type_)) return xml_node();
+		
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_node();
+
+		xml_node n(impl::allocate_node(alloc, type_));
+		if (!n) return xml_node();
+
+		impl::append_node(n._root, _root);
+
+		if (type_ == node_declaration) n.set_name(PUGIXML_TEXT("xml"));
+
+		return n;
+	}
+
+	PUGI__FN xml_node xml_node::prepend_child(xml_node_type type_)
+	{
+		if (!impl::allow_insert_child(type(), type_)) return xml_node();
+
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_node();
+		
+		xml_node n(impl::allocate_node(alloc, type_));
+		if (!n) return xml_node();
+
+		impl::prepend_node(n._root, _root);
+				
+		if (type_ == node_declaration) n.set_name(PUGIXML_TEXT("xml"));
+
+		return n;
+	}
+
+	PUGI__FN xml_node xml_node::insert_child_before(xml_node_type type_, const xml_node& node)
+	{
+		if (!impl::allow_insert_child(type(), type_)) return xml_node();
+		if (!node._root || node._root->parent != _root) return xml_node();
+
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_node();
+	
+		xml_node n(impl::allocate_node(alloc, type_));
+		if (!n) return xml_node();
+
+		impl::insert_node_before(n._root, node._root);
+
+		if (type_ == node_declaration) n.set_name(PUGIXML_TEXT("xml"));
+
+		return n;
+	}
+
+	PUGI__FN xml_node xml_node::insert_child_after(xml_node_type type_, const xml_node& node)
+	{
+		if (!impl::allow_insert_child(type(), type_)) return xml_node();
+		if (!node._root || node._root->parent != _root) return xml_node();
+
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_node();
+	
+		xml_node n(impl::allocate_node(alloc, type_));
+		if (!n) return xml_node();
+
+		impl::insert_node_after(n._root, node._root);
+
+		if (type_ == node_declaration) n.set_name(PUGIXML_TEXT("xml"));
+
+		return n;
+	}
+
+	PUGI__FN xml_node xml_node::append_child(const char_t* name_)
+	{
+		xml_node result = append_child(node_element);
+
+		result.set_name(name_);
+
+		return result;
+	}
+
+	PUGI__FN xml_node xml_node::prepend_child(const char_t* name_)
+	{
+		xml_node result = prepend_child(node_element);
+
+		result.set_name(name_);
+
+		return result;
+	}
+
+	PUGI__FN xml_node xml_node::insert_child_after(const char_t* name_, const xml_node& node)
+	{
+		xml_node result = insert_child_after(node_element, node);
+
+		result.set_name(name_);
+
+		return result;
+	}
+
+	PUGI__FN xml_node xml_node::insert_child_before(const char_t* name_, const xml_node& node)
+	{
+		xml_node result = insert_child_before(node_element, node);
+
+		result.set_name(name_);
+
+		return result;
+	}
+
+	PUGI__FN xml_node xml_node::append_copy(const xml_node& proto)
+	{
+		xml_node_type type_ = proto.type();
+		if (!impl::allow_insert_child(type(), type_)) return xml_node();
+
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_node();
+
+		xml_node n(impl::allocate_node(alloc, type_));
+		if (!n) return xml_node();
+
+		impl::append_node(n._root, _root);
+		impl::node_copy_tree(n._root, proto._root);
+
+		return n;
+	}
+
+	PUGI__FN xml_node xml_node::prepend_copy(const xml_node& proto)
+	{
+		xml_node_type type_ = proto.type();
+		if (!impl::allow_insert_child(type(), type_)) return xml_node();
+
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_node();
+
+		xml_node n(impl::allocate_node(alloc, type_));
+		if (!n) return xml_node();
+
+		impl::prepend_node(n._root, _root);
+		impl::node_copy_tree(n._root, proto._root);
+
+		return n;
+	}
+
+	PUGI__FN xml_node xml_node::insert_copy_after(const xml_node& proto, const xml_node& node)
+	{
+		xml_node_type type_ = proto.type();
+		if (!impl::allow_insert_child(type(), type_)) return xml_node();
+		if (!node._root || node._root->parent != _root) return xml_node();
+
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_node();
+
+		xml_node n(impl::allocate_node(alloc, type_));
+		if (!n) return xml_node();
+
+		impl::insert_node_after(n._root, node._root);
+		impl::node_copy_tree(n._root, proto._root);
+
+		return n;
+	}
+
+	PUGI__FN xml_node xml_node::insert_copy_before(const xml_node& proto, const xml_node& node)
+	{
+		xml_node_type type_ = proto.type();
+		if (!impl::allow_insert_child(type(), type_)) return xml_node();
+		if (!node._root || node._root->parent != _root) return xml_node();
+
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_node();
+
+		xml_node n(impl::allocate_node(alloc, type_));
+		if (!n) return xml_node();
+
+		impl::insert_node_before(n._root, node._root);
+		impl::node_copy_tree(n._root, proto._root);
+
+		return n;
+	}
+
+	PUGI__FN xml_node xml_node::append_move(const xml_node& moved)
+	{
+		if (!impl::allow_move(*this, moved)) return xml_node();
+
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_node();
+
+		// disable document_buffer_order optimization since moving nodes around changes document order without changing buffer pointers
+		impl::get_document(_root).header |= impl::xml_memory_page_contents_shared_mask;
+
+		impl::remove_node(moved._root);
+		impl::append_node(moved._root, _root);
+
+		return moved;
+	}
+
+	PUGI__FN xml_node xml_node::prepend_move(const xml_node& moved)
+	{
+		if (!impl::allow_move(*this, moved)) return xml_node();
+
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_node();
+
+		// disable document_buffer_order optimization since moving nodes around changes document order without changing buffer pointers
+		impl::get_document(_root).header |= impl::xml_memory_page_contents_shared_mask;
+
+		impl::remove_node(moved._root);
+		impl::prepend_node(moved._root, _root);
+
+		return moved;
+	}
+
+	PUGI__FN xml_node xml_node::insert_move_after(const xml_node& moved, const xml_node& node)
+	{
+		if (!impl::allow_move(*this, moved)) return xml_node();
+		if (!node._root || node._root->parent != _root) return xml_node();
+		if (moved._root == node._root) return xml_node();
+
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_node();
+
+		// disable document_buffer_order optimization since moving nodes around changes document order without changing buffer pointers
+		impl::get_document(_root).header |= impl::xml_memory_page_contents_shared_mask;
+
+		impl::remove_node(moved._root);
+		impl::insert_node_after(moved._root, node._root);
+
+		return moved;
+	}
+
+	PUGI__FN xml_node xml_node::insert_move_before(const xml_node& moved, const xml_node& node)
+	{
+		if (!impl::allow_move(*this, moved)) return xml_node();
+		if (!node._root || node._root->parent != _root) return xml_node();
+		if (moved._root == node._root) return xml_node();
+
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return xml_node();
+
+		// disable document_buffer_order optimization since moving nodes around changes document order without changing buffer pointers
+		impl::get_document(_root).header |= impl::xml_memory_page_contents_shared_mask;
+
+		impl::remove_node(moved._root);
+		impl::insert_node_before(moved._root, node._root);
+
+		return moved;
+	}
+
+	PUGI__FN bool xml_node::remove_attribute(const char_t* name_)
+	{
+		return remove_attribute(attribute(name_));
+	}
+
+	PUGI__FN bool xml_node::remove_attribute(const xml_attribute& a)
+	{
+		if (!_root || !a._attr) return false;
+		if (!impl::is_attribute_of(a._attr, _root)) return false;
+
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return false;
+
+		impl::remove_attribute(a._attr, _root);
+		impl::destroy_attribute(a._attr, alloc);
+
+		return true;
+	}
+
+	PUGI__FN bool xml_node::remove_child(const char_t* name_)
+	{
+		return remove_child(child(name_));
+	}
+
+	PUGI__FN bool xml_node::remove_child(const xml_node& n)
+	{
+		if (!_root || !n._root || n._root->parent != _root) return false;
+
+		impl::xml_allocator& alloc = impl::get_allocator(_root);
+		if (!alloc.reserve()) return false;
+
+		impl::remove_node(n._root);
+		impl::destroy_node(n._root, alloc);
+
+		return true;
+	}
+
+	PUGI__FN xml_parse_result xml_node::append_buffer(const void* contents, size_t size, unsigned int options, xml_encoding encoding)
+	{
+		// append_buffer is only valid for elements/documents
+		if (!impl::allow_insert_child(type(), node_element)) return impl::make_parse_result(status_append_invalid_root);
+
+		// get document node
+		impl::xml_document_struct* doc = &impl::get_document(_root);
+
+		// disable document_buffer_order optimization since in a document with multiple buffers comparing buffer pointers does not make sense
+		doc->header |= impl::xml_memory_page_contents_shared_mask;
+		
+		// get extra buffer element (we'll store the document fragment buffer there so that we can deallocate it later)
+		impl::xml_memory_page* page = 0;
+		impl::xml_extra_buffer* extra = static_cast<impl::xml_extra_buffer*>(doc->allocate_memory(sizeof(impl::xml_extra_buffer), page));
+		(void)page;
+
+		if (!extra) return impl::make_parse_result(status_out_of_memory);
+
+		// add extra buffer to the list
+		extra->buffer = 0;
+		extra->next = doc->extra_buffers;
+		doc->extra_buffers = extra;
+
+		// name of the root has to be NULL before parsing - otherwise closing node mismatches will not be detected at the top level
+		struct name_sentry
+		{
+			xml_node_struct* node;
+			char_t* name;
+
+			~name_sentry() { node->name = name; }
+		};
+
+		name_sentry sentry = { _root, _root->name };
+
+		sentry.node->name = 0;
+
+		return impl::load_buffer_impl(doc, _root, const_cast<void*>(contents), size, options, encoding, false, false, &extra->buffer);
+	}
+
+	PUGI__FN xml_node xml_node::find_child_by_attribute(const char_t* name_, const char_t* attr_name, const char_t* attr_value) const
+	{
+		if (!_root) return xml_node();
+		
+		for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling)
+			if (i->name && impl::strequal(name_, i->name))
+			{
+				for (xml_attribute_struct* a = i->first_attribute; a; a = a->next_attribute)
+					if (a->name && impl::strequal(attr_name, a->name) && impl::strequal(attr_value, a->value ? a->value + 0 : PUGIXML_TEXT("")))
+						return xml_node(i);
+			}
+
+		return xml_node();
+	}
+
+	PUGI__FN xml_node xml_node::find_child_by_attribute(const char_t* attr_name, const char_t* attr_value) const
+	{
+		if (!_root) return xml_node();
+		
+		for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling)
+			for (xml_attribute_struct* a = i->first_attribute; a; a = a->next_attribute)
+				if (a->name && impl::strequal(attr_name, a->name) && impl::strequal(attr_value, a->value ? a->value + 0 : PUGIXML_TEXT("")))
+					return xml_node(i);
+
+		return xml_node();
+	}
+
+#ifndef PUGIXML_NO_STL
+	PUGI__FN string_t xml_node::path(char_t delimiter) const
+	{
+		if (!_root) return string_t();
+
+		size_t offset = 0;
+
+		for (xml_node_struct* i = _root; i; i = i->parent)
+		{
+			offset += (i != _root);
+			offset += i->name ? impl::strlength(i->name) : 0;
+		}
+
+		string_t result;
+		result.resize(offset);
+
+		for (xml_node_struct* j = _root; j; j = j->parent)
+		{
+			if (j != _root)
+				result[--offset] = delimiter;
+
+			if (j->name && *j->name)
+			{
+				size_t length = impl::strlength(j->name);
+
+				offset -= length;
+				memcpy(&result[offset], j->name, length * sizeof(char_t));
+			}
+		}
+
+		assert(offset == 0);
+
+		return result;
+	}
+#endif
+
+	PUGI__FN xml_node xml_node::first_element_by_path(const char_t* path_, char_t delimiter) const
+	{
+		xml_node found = *this; // Current search context.
+
+		if (!_root || !path_ || !path_[0]) return found;
+
+		if (path_[0] == delimiter)
+		{
+			// Absolute path; e.g. '/foo/bar'
+			found = found.root();
+			++path_;
+		}
+
+		const char_t* path_segment = path_;
+
+		while (*path_segment == delimiter) ++path_segment;
+
+		const char_t* path_segment_end = path_segment;
+
+		while (*path_segment_end && *path_segment_end != delimiter) ++path_segment_end;
+
+		if (path_segment == path_segment_end) return found;
+
+		const char_t* next_segment = path_segment_end;
+
+		while (*next_segment == delimiter) ++next_segment;
+
+		if (*path_segment == '.' && path_segment + 1 == path_segment_end)
+			return found.first_element_by_path(next_segment, delimiter);
+		else if (*path_segment == '.' && *(path_segment+1) == '.' && path_segment + 2 == path_segment_end)
+			return found.parent().first_element_by_path(next_segment, delimiter);
+		else
+		{
+			for (xml_node_struct* j = found._root->first_child; j; j = j->next_sibling)
+			{
+				if (j->name && impl::strequalrange(j->name, path_segment, static_cast<size_t>(path_segment_end - path_segment)))
+				{
+					xml_node subsearch = xml_node(j).first_element_by_path(next_segment, delimiter);
+
+					if (subsearch) return subsearch;
+				}
+			}
+
+			return xml_node();
+		}
+	}
+
+	PUGI__FN bool xml_node::traverse(xml_tree_walker& walker)
+	{
+		walker._depth = -1;
+		
+		xml_node arg_begin = *this;
+		if (!walker.begin(arg_begin)) return false;
+
+		xml_node cur = first_child();
+				
+		if (cur)
+		{
+			++walker._depth;
+
+			do 
+			{
+				xml_node arg_for_each = cur;
+				if (!walker.for_each(arg_for_each))
+					return false;
+						
+				if (cur.first_child())
+				{
+					++walker._depth;
+					cur = cur.first_child();
+				}
+				else if (cur.next_sibling())
+					cur = cur.next_sibling();
+				else
+				{
+					// Borland C++ workaround
+					while (!cur.next_sibling() && cur != *this && !cur.parent().empty())
+					{
+						--walker._depth;
+						cur = cur.parent();
+					}
+						
+					if (cur != *this)
+						cur = cur.next_sibling();
+				}
+			}
+			while (cur && cur != *this);
+		}
+
+		assert(walker._depth == -1);
+
+		xml_node arg_end = *this;
+		return walker.end(arg_end);
+	}
+
+	PUGI__FN size_t xml_node::hash_value() const
+	{
+		return static_cast<size_t>(reinterpret_cast<uintptr_t>(_root) / sizeof(xml_node_struct));
+	}
+
+	PUGI__FN xml_node_struct* xml_node::internal_object() const
+	{
+		return _root;
+	}
+
+	PUGI__FN void xml_node::print(xml_writer& writer, const char_t* indent, unsigned int flags, xml_encoding encoding, unsigned int depth) const
+	{
+		if (!_root) return;
+
+		impl::xml_buffered_writer buffered_writer(writer, encoding);
+
+		impl::node_output(buffered_writer, _root, indent, flags, depth);
+
+		buffered_writer.flush();
+	}
+
+#ifndef PUGIXML_NO_STL
+	PUGI__FN void xml_node::print(std::basic_ostream<char, std::char_traits<char> >& stream, const char_t* indent, unsigned int flags, xml_encoding encoding, unsigned int depth) const
+	{
+		xml_writer_stream writer(stream);
+
+		print(writer, indent, flags, encoding, depth);
+	}
+
+	PUGI__FN void xml_node::print(std::basic_ostream<wchar_t, std::char_traits<wchar_t> >& stream, const char_t* indent, unsigned int flags, unsigned int depth) const
+	{
+		xml_writer_stream writer(stream);
+
+		print(writer, indent, flags, encoding_wchar, depth);
+	}
+#endif
+
+	PUGI__FN ptrdiff_t xml_node::offset_debug() const
+	{
+		if (!_root) return -1;
+
+		impl::xml_document_struct& doc = impl::get_document(_root);
+
+		// we can determine the offset reliably only if there is exactly once parse buffer
+		if (!doc.buffer || doc.extra_buffers) return -1;
+
+		switch (type())
+		{
+		case node_document:
+			return 0;
+
+		case node_element:
+		case node_declaration:
+		case node_pi:
+			return _root->name && (_root->header & impl::xml_memory_page_name_allocated_or_shared_mask) == 0 ? _root->name - doc.buffer : -1;
+
+		case node_pcdata:
+		case node_cdata:
+		case node_comment:
+		case node_doctype:
+			return _root->value && (_root->header & impl::xml_memory_page_value_allocated_or_shared_mask) == 0 ? _root->value - doc.buffer : -1;
+
+		default:
+			return -1;
+		}
+	}
+
+#ifdef __BORLANDC__
+	PUGI__FN bool operator&&(const xml_node& lhs, bool rhs)
+	{
+		return (bool)lhs && rhs;
+	}
+
+	PUGI__FN bool operator||(const xml_node& lhs, bool rhs)
+	{
+		return (bool)lhs || rhs;
+	}
+#endif
+
+	PUGI__FN xml_text::xml_text(xml_node_struct* root): _root(root)
+	{
+	}
+
+	PUGI__FN xml_node_struct* xml_text::_data() const
+	{
+		if (!_root || impl::is_text_node(_root)) return _root;
+
+		for (xml_node_struct* node = _root->first_child; node; node = node->next_sibling)
+			if (impl::is_text_node(node))
+				return node;
+
+		return 0;
+	}
+
+	PUGI__FN xml_node_struct* xml_text::_data_new()
+	{
+		xml_node_struct* d = _data();
+		if (d) return d;
+
+		return xml_node(_root).append_child(node_pcdata).internal_object();
+	}
+
+	PUGI__FN xml_text::xml_text(): _root(0)
+	{
+	}
+
+	PUGI__FN static void unspecified_bool_xml_text(xml_text***)
+	{
+	}
+
+	PUGI__FN xml_text::operator xml_text::unspecified_bool_type() const
+	{
+		return _data() ? unspecified_bool_xml_text : 0;
+	}
+
+	PUGI__FN bool xml_text::operator!() const
+	{
+		return !_data();
+	}
+
+	PUGI__FN bool xml_text::empty() const
+	{
+		return _data() == 0;
+	}
+
+	PUGI__FN const char_t* xml_text::get() const
+	{
+		xml_node_struct* d = _data();
+
+		return (d && d->value) ? d->value + 0 : PUGIXML_TEXT("");
+	}
+
+	PUGI__FN const char_t* xml_text::as_string(const char_t* def) const
+	{
+		xml_node_struct* d = _data();
+
+		return (d && d->value) ? d->value : def;
+	}
+
+	PUGI__FN int xml_text::as_int(int def) const
+	{
+		xml_node_struct* d = _data();
+
+		return (d && d->value) ? impl::get_value_int(d->value) : def;
+	}
+
+	PUGI__FN unsigned int xml_text::as_uint(unsigned int def) const
+	{
+		xml_node_struct* d = _data();
+
+		return (d && d->value) ? impl::get_value_uint(d->value) : def;
+	}
+
+	PUGI__FN double xml_text::as_double(double def) const
+	{
+		xml_node_struct* d = _data();
+
+		return (d && d->value) ? impl::get_value_double(d->value) : def;
+	}
+
+	PUGI__FN float xml_text::as_float(float def) const
+	{
+		xml_node_struct* d = _data();
+
+		return (d && d->value) ? impl::get_value_float(d->value) : def;
+	}
+
+	PUGI__FN bool xml_text::as_bool(bool def) const
+	{
+		xml_node_struct* d = _data();
+
+		return (d && d->value) ? impl::get_value_bool(d->value) : def;
+	}
+
+#ifdef PUGIXML_HAS_LONG_LONG
+	PUGI__FN long long xml_text::as_llong(long long def) const
+	{
+		xml_node_struct* d = _data();
+
+		return (d && d->value) ? impl::get_value_llong(d->value) : def;
+	}
+
+	PUGI__FN unsigned long long xml_text::as_ullong(unsigned long long def) const
+	{
+		xml_node_struct* d = _data();
+
+		return (d && d->value) ? impl::get_value_ullong(d->value) : def;
+	}
+#endif
+
+	PUGI__FN bool xml_text::set(const char_t* rhs)
+	{
+		xml_node_struct* dn = _data_new();
+
+		return dn ? impl::strcpy_insitu(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs) : false;
+	}
+
+	PUGI__FN bool xml_text::set(int rhs)
+	{
+		xml_node_struct* dn = _data_new();
+
+		return dn ? impl::set_value_convert(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs) : false;
+	}
+
+	PUGI__FN bool xml_text::set(unsigned int rhs)
+	{
+		xml_node_struct* dn = _data_new();
+
+		return dn ? impl::set_value_convert(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs) : false;
+	}
+
+	PUGI__FN bool xml_text::set(float rhs)
+	{
+		xml_node_struct* dn = _data_new();
+
+		return dn ? impl::set_value_convert(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs) : false;
+	}
+
+	PUGI__FN bool xml_text::set(double rhs)
+	{
+		xml_node_struct* dn = _data_new();
+
+		return dn ? impl::set_value_convert(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs) : false;
+	}
+
+	PUGI__FN bool xml_text::set(bool rhs)
+	{
+		xml_node_struct* dn = _data_new();
+
+		return dn ? impl::set_value_convert(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs) : false;
+	}
+
+#ifdef PUGIXML_HAS_LONG_LONG
+	PUGI__FN bool xml_text::set(long long rhs)
+	{
+		xml_node_struct* dn = _data_new();
+
+		return dn ? impl::set_value_convert(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs) : false;
+	}
+
+	PUGI__FN bool xml_text::set(unsigned long long rhs)
+	{
+		xml_node_struct* dn = _data_new();
+
+		return dn ? impl::set_value_convert(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs) : false;
+	}
+#endif
+
+	PUGI__FN xml_text& xml_text::operator=(const char_t* rhs)
+	{
+		set(rhs);
+		return *this;
+	}
+
+	PUGI__FN xml_text& xml_text::operator=(int rhs)
+	{
+		set(rhs);
+		return *this;
+	}
+
+	PUGI__FN xml_text& xml_text::operator=(unsigned int rhs)
+	{
+		set(rhs);
+		return *this;
+	}
+
+	PUGI__FN xml_text& xml_text::operator=(double rhs)
+	{
+		set(rhs);
+		return *this;
+	}
+
+	PUGI__FN xml_text& xml_text::operator=(float rhs)
+	{
+		set(rhs);
+		return *this;
+	}
+
+	PUGI__FN xml_text& xml_text::operator=(bool rhs)
+	{
+		set(rhs);
+		return *this;
+	}
+
+#ifdef PUGIXML_HAS_LONG_LONG
+	PUGI__FN xml_text& xml_text::operator=(long long rhs)
+	{
+		set(rhs);
+		return *this;
+	}
+
+	PUGI__FN xml_text& xml_text::operator=(unsigned long long rhs)
+	{
+		set(rhs);
+		return *this;
+	}
+#endif
+
+	PUGI__FN xml_node xml_text::data() const
+	{
+		return xml_node(_data());
+	}
+
+#ifdef __BORLANDC__
+	PUGI__FN bool operator&&(const xml_text& lhs, bool rhs)
+	{
+		return (bool)lhs && rhs;
+	}
+
+	PUGI__FN bool operator||(const xml_text& lhs, bool rhs)
+	{
+		return (bool)lhs || rhs;
+	}
+#endif
+
+	PUGI__FN xml_node_iterator::xml_node_iterator()
+	{
+	}
+
+	PUGI__FN xml_node_iterator::xml_node_iterator(const xml_node& node): _wrap(node), _parent(node.parent())
+	{
+	}
+
+	PUGI__FN xml_node_iterator::xml_node_iterator(xml_node_struct* ref, xml_node_struct* parent): _wrap(ref), _parent(parent)
+	{
+	}
+
+	PUGI__FN bool xml_node_iterator::operator==(const xml_node_iterator& rhs) const
+	{
+		return _wrap._root == rhs._wrap._root && _parent._root == rhs._parent._root;
+	}
+	
+	PUGI__FN bool xml_node_iterator::operator!=(const xml_node_iterator& rhs) const
+	{
+		return _wrap._root != rhs._wrap._root || _parent._root != rhs._parent._root;
+	}
+
+	PUGI__FN xml_node& xml_node_iterator::operator*() const
+	{
+		assert(_wrap._root);
+		return _wrap;
+	}
+
+	PUGI__FN xml_node* xml_node_iterator::operator->() const
+	{
+		assert(_wrap._root);
+		return const_cast<xml_node*>(&_wrap); // BCC32 workaround
+	}
+
+	PUGI__FN const xml_node_iterator& xml_node_iterator::operator++()
+	{
+		assert(_wrap._root);
+		_wrap._root = _wrap._root->next_sibling;
+		return *this;
+	}
+
+	PUGI__FN xml_node_iterator xml_node_iterator::operator++(int)
+	{
+		xml_node_iterator temp = *this;
+		++*this;
+		return temp;
+	}
+
+	PUGI__FN const xml_node_iterator& xml_node_iterator::operator--()
+	{
+		_wrap = _wrap._root ? _wrap.previous_sibling() : _parent.last_child();
+		return *this;
+	}
+
+	PUGI__FN xml_node_iterator xml_node_iterator::operator--(int)
+	{
+		xml_node_iterator temp = *this;
+		--*this;
+		return temp;
+	}
+
+	PUGI__FN xml_attribute_iterator::xml_attribute_iterator()
+	{
+	}
+
+	PUGI__FN xml_attribute_iterator::xml_attribute_iterator(const xml_attribute& attr, const xml_node& parent): _wrap(attr), _parent(parent)
+	{
+	}
+
+	PUGI__FN xml_attribute_iterator::xml_attribute_iterator(xml_attribute_struct* ref, xml_node_struct* parent): _wrap(ref), _parent(parent)
+	{
+	}
+
+	PUGI__FN bool xml_attribute_iterator::operator==(const xml_attribute_iterator& rhs) const
+	{
+		return _wrap._attr == rhs._wrap._attr && _parent._root == rhs._parent._root;
+	}
+	
+	PUGI__FN bool xml_attribute_iterator::operator!=(const xml_attribute_iterator& rhs) const
+	{
+		return _wrap._attr != rhs._wrap._attr || _parent._root != rhs._parent._root;
+	}
+
+	PUGI__FN xml_attribute& xml_attribute_iterator::operator*() const
+	{
+		assert(_wrap._attr);
+		return _wrap;
+	}
+
+	PUGI__FN xml_attribute* xml_attribute_iterator::operator->() const
+	{
+		assert(_wrap._attr);
+		return const_cast<xml_attribute*>(&_wrap); // BCC32 workaround
+	}
+
+	PUGI__FN const xml_attribute_iterator& xml_attribute_iterator::operator++()
+	{
+		assert(_wrap._attr);
+		_wrap._attr = _wrap._attr->next_attribute;
+		return *this;
+	}
+
+	PUGI__FN xml_attribute_iterator xml_attribute_iterator::operator++(int)
+	{
+		xml_attribute_iterator temp = *this;
+		++*this;
+		return temp;
+	}
+
+	PUGI__FN const xml_attribute_iterator& xml_attribute_iterator::operator--()
+	{
+		_wrap = _wrap._attr ? _wrap.previous_attribute() : _parent.last_attribute();
+		return *this;
+	}
+
+	PUGI__FN xml_attribute_iterator xml_attribute_iterator::operator--(int)
+	{
+		xml_attribute_iterator temp = *this;
+		--*this;
+		return temp;
+	}
+
+	PUGI__FN xml_named_node_iterator::xml_named_node_iterator(): _name(0)
+	{
+	}
+
+	PUGI__FN xml_named_node_iterator::xml_named_node_iterator(const xml_node& node, const char_t* name): _wrap(node), _parent(node.parent()), _name(name)
+	{
+	}
+
+	PUGI__FN xml_named_node_iterator::xml_named_node_iterator(xml_node_struct* ref, xml_node_struct* parent, const char_t* name): _wrap(ref), _parent(parent), _name(name)
+	{
+	}
+
+	PUGI__FN bool xml_named_node_iterator::operator==(const xml_named_node_iterator& rhs) const
+	{
+		return _wrap._root == rhs._wrap._root && _parent._root == rhs._parent._root;
+	}
+
+	PUGI__FN bool xml_named_node_iterator::operator!=(const xml_named_node_iterator& rhs) const
+	{
+		return _wrap._root != rhs._wrap._root || _parent._root != rhs._parent._root;
+	}
+
+	PUGI__FN xml_node& xml_named_node_iterator::operator*() const
+	{
+		assert(_wrap._root);
+		return _wrap;
+	}
+
+	PUGI__FN xml_node* xml_named_node_iterator::operator->() const
+	{
+		assert(_wrap._root);
+		return const_cast<xml_node*>(&_wrap); // BCC32 workaround
+	}
+
+	PUGI__FN const xml_named_node_iterator& xml_named_node_iterator::operator++()
+	{
+		assert(_wrap._root);
+		_wrap = _wrap.next_sibling(_name);
+		return *this;
+	}
+
+	PUGI__FN xml_named_node_iterator xml_named_node_iterator::operator++(int)
+	{
+		xml_named_node_iterator temp = *this;
+		++*this;
+		return temp;
+	}
+
+	PUGI__FN const xml_named_node_iterator& xml_named_node_iterator::operator--()
+	{
+		if (_wrap._root)
+			_wrap = _wrap.previous_sibling(_name);
+		else
+		{
+			_wrap = _parent.last_child();
+
+			if (!impl::strequal(_wrap.name(), _name))
+				_wrap = _wrap.previous_sibling(_name);
+		}
+
+		return *this;
+	}
+
+	PUGI__FN xml_named_node_iterator xml_named_node_iterator::operator--(int)
+	{
+		xml_named_node_iterator temp = *this;
+		--*this;
+		return temp;
+	}
+
+	PUGI__FN xml_parse_result::xml_parse_result(): status(status_internal_error), offset(0), encoding(encoding_auto)
+	{
+	}
+
+	PUGI__FN xml_parse_result::operator bool() const
+	{
+		return status == status_ok;
+	}
+
+	PUGI__FN const char* xml_parse_result::description() const
+	{
+		switch (status)
+		{
+		case status_ok: return "No error";
+
+		case status_file_not_found: return "File was not found";
+		case status_io_error: return "Error reading from file/stream";
+		case status_out_of_memory: return "Could not allocate memory";
+		case status_internal_error: return "Internal error occurred";
+
+		case status_unrecognized_tag: return "Could not determine tag type";
+
+		case status_bad_pi: return "Error parsing document declaration/processing instruction";
+		case status_bad_comment: return "Error parsing comment";
+		case status_bad_cdata: return "Error parsing CDATA section";
+		case status_bad_doctype: return "Error parsing document type declaration";
+		case status_bad_pcdata: return "Error parsing PCDATA section";
+		case status_bad_start_element: return "Error parsing start element tag";
+		case status_bad_attribute: return "Error parsing element attribute";
+		case status_bad_end_element: return "Error parsing end element tag";
+		case status_end_element_mismatch: return "Start-end tags mismatch";
+
+		case status_append_invalid_root: return "Unable to append nodes: root is not an element or document";
+
+		case status_no_document_element: return "No document element found";
+
+		default: return "Unknown error";
+		}
+	}
+
+	PUGI__FN xml_document::xml_document(): _buffer(0)
+	{
+		create();
+	}
+
+	PUGI__FN xml_document::~xml_document()
+	{
+		destroy();
+	}
+
+	PUGI__FN void xml_document::reset()
+	{
+		destroy();
+		create();
+	}
+
+	PUGI__FN void xml_document::reset(const xml_document& proto)
+	{
+		reset();
+
+		for (xml_node cur = proto.first_child(); cur; cur = cur.next_sibling())
+			append_copy(cur);
+	}
+
+	PUGI__FN void xml_document::create()
+	{
+		assert(!_root);
+
+	#ifdef PUGIXML_COMPACT
+		const size_t page_offset = sizeof(uint32_t);
+	#else
+		const size_t page_offset = 0;
+	#endif
+
+		// initialize sentinel page
+		PUGI__STATIC_ASSERT(sizeof(impl::xml_memory_page) + sizeof(impl::xml_document_struct) + impl::xml_memory_page_alignment - sizeof(void*) + page_offset <= sizeof(_memory));
+
+		// align upwards to page boundary
+		void* page_memory = reinterpret_cast<void*>((reinterpret_cast<uintptr_t>(_memory) + (impl::xml_memory_page_alignment - 1)) & ~(impl::xml_memory_page_alignment - 1));
+
+		// prepare page structure
+		impl::xml_memory_page* page = impl::xml_memory_page::construct(page_memory);
+		assert(page);
+
+		page->busy_size = impl::xml_memory_page_size;
+
+		// setup first page marker
+	#ifdef PUGIXML_COMPACT
+		page->compact_page_marker = reinterpret_cast<uint32_t*>(reinterpret_cast<char*>(page) + sizeof(impl::xml_memory_page));
+		*page->compact_page_marker = sizeof(impl::xml_memory_page);
+	#endif
+
+		// allocate new root
+		_root = new (reinterpret_cast<char*>(page) + sizeof(impl::xml_memory_page) + page_offset) impl::xml_document_struct(page);
+		_root->prev_sibling_c = _root;
+
+		// setup sentinel page
+		page->allocator = static_cast<impl::xml_document_struct*>(_root);
+
+		// verify the document allocation
+		assert(reinterpret_cast<char*>(_root) + sizeof(impl::xml_document_struct) <= _memory + sizeof(_memory));
+	}
+
+	PUGI__FN void xml_document::destroy()
+	{
+		assert(_root);
+
+		// destroy static storage
+		if (_buffer)
+		{
+			impl::xml_memory::deallocate(_buffer);
+			_buffer = 0;
+		}
+
+		// destroy extra buffers (note: no need to destroy linked list nodes, they're allocated using document allocator)
+		for (impl::xml_extra_buffer* extra = static_cast<impl::xml_document_struct*>(_root)->extra_buffers; extra; extra = extra->next)
+		{
+			if (extra->buffer) impl::xml_memory::deallocate(extra->buffer);
+		}
+
+		// destroy dynamic storage, leave sentinel page (it's in static memory)
+		impl::xml_memory_page* root_page = PUGI__GETPAGE(_root);
+		assert(root_page && !root_page->prev);
+		assert(reinterpret_cast<char*>(root_page) >= _memory && reinterpret_cast<char*>(root_page) < _memory + sizeof(_memory));
+
+		for (impl::xml_memory_page* page = root_page->next; page; )
+		{
+			impl::xml_memory_page* next = page->next;
+
+			impl::xml_allocator::deallocate_page(page);
+
+			page = next;
+		}
+
+	#ifdef PUGIXML_COMPACT
+		// destroy hash table
+		static_cast<impl::xml_document_struct*>(_root)->hash.clear();
+	#endif
+
+		_root = 0;
+	}
+
+#ifndef PUGIXML_NO_STL
+	PUGI__FN xml_parse_result xml_document::load(std::basic_istream<char, std::char_traits<char> >& stream, unsigned int options, xml_encoding encoding)
+	{
+		reset();
+
+		return impl::load_stream_impl(*this, stream, options, encoding);
+	}
+
+	PUGI__FN xml_parse_result xml_document::load(std::basic_istream<wchar_t, std::char_traits<wchar_t> >& stream, unsigned int options)
+	{
+		reset();
+
+		return impl::load_stream_impl(*this, stream, options, encoding_wchar);
+	}
+#endif
+
+	PUGI__FN xml_parse_result xml_document::load_string(const char_t* contents, unsigned int options)
+	{
+		// Force native encoding (skip autodetection)
+	#ifdef PUGIXML_WCHAR_MODE
+		xml_encoding encoding = encoding_wchar;
+	#else
+		xml_encoding encoding = encoding_utf8;
+	#endif
+
+		return load_buffer(contents, impl::strlength(contents) * sizeof(char_t), options, encoding);
+	}
+
+	PUGI__FN xml_parse_result xml_document::load(const char_t* contents, unsigned int options)
+	{
+		return load_string(contents, options);
+	}
+
+	PUGI__FN xml_parse_result xml_document::load_file(const char* path_, unsigned int options, xml_encoding encoding)
+	{
+		reset();
+
+		using impl::auto_deleter; // MSVC7 workaround
+		auto_deleter<FILE, int(*)(FILE*)> file(fopen(path_, "rb"), fclose);
+
+		return impl::load_file_impl(*this, file.data, options, encoding);
+	}
+
+	PUGI__FN xml_parse_result xml_document::load_file(const wchar_t* path_, unsigned int options, xml_encoding encoding)
+	{
+		reset();
+
+		using impl::auto_deleter; // MSVC7 workaround
+		auto_deleter<FILE, int(*)(FILE*)> file(impl::open_file_wide(path_, L"rb"), fclose);
+
+		return impl::load_file_impl(*this, file.data, options, encoding);
+	}
+
+	PUGI__FN xml_parse_result xml_document::load_buffer(const void* contents, size_t size, unsigned int options, xml_encoding encoding)
+	{
+		reset();
+
+		return impl::load_buffer_impl(static_cast<impl::xml_document_struct*>(_root), _root, const_cast<void*>(contents), size, options, encoding, false, false, &_buffer);
+	}
+
+	PUGI__FN xml_parse_result xml_document::load_buffer_inplace(void* contents, size_t size, unsigned int options, xml_encoding encoding)
+	{
+		reset();
+
+		return impl::load_buffer_impl(static_cast<impl::xml_document_struct*>(_root), _root, contents, size, options, encoding, true, false, &_buffer);
+	}
+
+	PUGI__FN xml_parse_result xml_document::load_buffer_inplace_own(void* contents, size_t size, unsigned int options, xml_encoding encoding)
+	{
+		reset();
+
+		return impl::load_buffer_impl(static_cast<impl::xml_document_struct*>(_root), _root, contents, size, options, encoding, true, true, &_buffer);
+	}
+
+	PUGI__FN void xml_document::save(xml_writer& writer, const char_t* indent, unsigned int flags, xml_encoding encoding) const
+	{
+		impl::xml_buffered_writer buffered_writer(writer, encoding);
+
+		if ((flags & format_write_bom) && encoding != encoding_latin1)
+		{
+			// BOM always represents the codepoint U+FEFF, so just write it in native encoding
+		#ifdef PUGIXML_WCHAR_MODE
+			unsigned int bom = 0xfeff;
+			buffered_writer.write(static_cast<wchar_t>(bom));
+		#else
+			buffered_writer.write('\xef', '\xbb', '\xbf');
+		#endif
+		}
+
+		if (!(flags & format_no_declaration) && !impl::has_declaration(_root))
+		{
+			buffered_writer.write_string(PUGIXML_TEXT("<?xml version=\"1.0\""));
+			if (encoding == encoding_latin1) buffered_writer.write_string(PUGIXML_TEXT(" encoding=\"ISO-8859-1\""));
+			buffered_writer.write('?', '>');
+			if (!(flags & format_raw)) buffered_writer.write('\n');
+		}
+
+		impl::node_output(buffered_writer, _root, indent, flags, 0);
+
+		buffered_writer.flush();
+	}
+
+#ifndef PUGIXML_NO_STL
+	PUGI__FN void xml_document::save(std::basic_ostream<char, std::char_traits<char> >& stream, const char_t* indent, unsigned int flags, xml_encoding encoding) const
+	{
+		xml_writer_stream writer(stream);
+
+		save(writer, indent, flags, encoding);
+	}
+
+	PUGI__FN void xml_document::save(std::basic_ostream<wchar_t, std::char_traits<wchar_t> >& stream, const char_t* indent, unsigned int flags) const
+	{
+		xml_writer_stream writer(stream);
+
+		save(writer, indent, flags, encoding_wchar);
+	}
+#endif
+
+	PUGI__FN bool xml_document::save_file(const char* path_, const char_t* indent, unsigned int flags, xml_encoding encoding) const
+	{
+		using impl::auto_deleter; // MSVC7 workaround
+		auto_deleter<FILE, int(*)(FILE*)> file(fopen(path_, (flags & format_save_file_text) ? "w" : "wb"), fclose);
+
+		return impl::save_file_impl(*this, file.data, indent, flags, encoding);
+	}
+
+	PUGI__FN bool xml_document::save_file(const wchar_t* path_, const char_t* indent, unsigned int flags, xml_encoding encoding) const
+	{
+		using impl::auto_deleter; // MSVC7 workaround
+		auto_deleter<FILE, int(*)(FILE*)> file(impl::open_file_wide(path_, (flags & format_save_file_text) ? L"w" : L"wb"), fclose);
+
+		return impl::save_file_impl(*this, file.data, indent, flags, encoding);
+	}
+
+	PUGI__FN xml_node xml_document::document_element() const
+	{
+		assert(_root);
+
+		for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling)
+			if (PUGI__NODETYPE(i) == node_element)
+				return xml_node(i);
+
+		return xml_node();
+	}
+
+#ifndef PUGIXML_NO_STL
+	PUGI__FN std::string PUGIXML_FUNCTION as_utf8(const wchar_t* str)
+	{
+		assert(str);
+
+		return impl::as_utf8_impl(str, impl::strlength_wide(str));
+	}
+
+	PUGI__FN std::string PUGIXML_FUNCTION as_utf8(const std::basic_string<wchar_t>& str)
+	{
+		return impl::as_utf8_impl(str.c_str(), str.size());
+	}
+	
+	PUGI__FN std::basic_string<wchar_t> PUGIXML_FUNCTION as_wide(const char* str)
+	{
+		assert(str);
+
+		return impl::as_wide_impl(str, strlen(str));
+	}
+	
+	PUGI__FN std::basic_string<wchar_t> PUGIXML_FUNCTION as_wide(const std::string& str)
+	{
+		return impl::as_wide_impl(str.c_str(), str.size());
+	}
+#endif
+
+	PUGI__FN void PUGIXML_FUNCTION set_memory_management_functions(allocation_function allocate, deallocation_function deallocate)
+	{
+		impl::xml_memory::allocate = allocate;
+		impl::xml_memory::deallocate = deallocate;
+	}
+
+	PUGI__FN allocation_function PUGIXML_FUNCTION get_memory_allocation_function()
+	{
+		return impl::xml_memory::allocate;
+	}
+
+	PUGI__FN deallocation_function PUGIXML_FUNCTION get_memory_deallocation_function()
+	{
+		return impl::xml_memory::deallocate;
+	}
+}
+
+#if !defined(PUGIXML_NO_STL) && (defined(_MSC_VER) || defined(__ICC))
+namespace std
+{
+	// Workarounds for (non-standard) iterator category detection for older versions (MSVC7/IC8 and earlier)
+	PUGI__FN std::bidirectional_iterator_tag _Iter_cat(const pugi::xml_node_iterator&)
+	{
+		return std::bidirectional_iterator_tag();
+	}
+
+	PUGI__FN std::bidirectional_iterator_tag _Iter_cat(const pugi::xml_attribute_iterator&)
+	{
+		return std::bidirectional_iterator_tag();
+	}
+
+	PUGI__FN std::bidirectional_iterator_tag _Iter_cat(const pugi::xml_named_node_iterator&)
+	{
+		return std::bidirectional_iterator_tag();
+	}
+}
+#endif
+
+#if !defined(PUGIXML_NO_STL) && defined(__SUNPRO_CC)
+namespace std
+{
+	// Workarounds for (non-standard) iterator category detection
+	PUGI__FN std::bidirectional_iterator_tag __iterator_category(const pugi::xml_node_iterator&)
+	{
+		return std::bidirectional_iterator_tag();
+	}
+
+	PUGI__FN std::bidirectional_iterator_tag __iterator_category(const pugi::xml_attribute_iterator&)
+	{
+		return std::bidirectional_iterator_tag();
+	}
+
+	PUGI__FN std::bidirectional_iterator_tag __iterator_category(const pugi::xml_named_node_iterator&)
+	{
+		return std::bidirectional_iterator_tag();
+	}
+}
+#endif
+
+#ifndef PUGIXML_NO_XPATH
+// STL replacements
+PUGI__NS_BEGIN
+	struct equal_to
+	{
+		template <typename T> bool operator()(const T& lhs, const T& rhs) const
+		{
+			return lhs == rhs;
+		}
+	};
+
+	struct not_equal_to
+	{
+		template <typename T> bool operator()(const T& lhs, const T& rhs) const
+		{
+			return lhs != rhs;
+		}
+	};
+
+	struct less
+	{
+		template <typename T> bool operator()(const T& lhs, const T& rhs) const
+		{
+			return lhs < rhs;
+		}
+	};
+
+	struct less_equal
+	{
+		template <typename T> bool operator()(const T& lhs, const T& rhs) const
+		{
+			return lhs <= rhs;
+		}
+	};
+
+	template <typename T> void swap(T& lhs, T& rhs)
+	{
+		T temp = lhs;
+		lhs = rhs;
+		rhs = temp;
+	}
+
+	template <typename I, typename Pred> I min_element(I begin, I end, const Pred& pred)
+	{
+		I result = begin;
+
+		for (I it = begin + 1; it != end; ++it)
+			if (pred(*it, *result))
+				result = it;
+
+		return result;
+	}
+
+	template <typename I> void reverse(I begin, I end)
+	{
+		while (end - begin > 1) swap(*begin++, *--end);
+	}
+
+	template <typename I> I unique(I begin, I end)
+	{
+		// fast skip head
+		while (end - begin > 1 && *begin != *(begin + 1)) begin++;
+
+		if (begin == end) return begin;
+
+		// last written element
+		I write = begin++; 
+
+		// merge unique elements
+		while (begin != end)
+		{
+			if (*begin != *write)
+				*++write = *begin++;
+			else
+				begin++;
+		}
+
+		// past-the-end (write points to live element)
+		return write + 1;
+	}
+
+	template <typename I> void copy_backwards(I begin, I end, I target)
+	{
+		while (begin != end) *--target = *--end;
+	}
+
+	template <typename I, typename Pred, typename T> void insertion_sort(I begin, I end, const Pred& pred, T*)
+	{
+		assert(begin != end);
+
+		for (I it = begin + 1; it != end; ++it)
+		{
+			T val = *it;
+
+			if (pred(val, *begin))
+			{
+				// move to front
+				copy_backwards(begin, it, it + 1);
+				*begin = val;
+			}
+			else
+			{
+				I hole = it;
+
+				// move hole backwards
+				while (pred(val, *(hole - 1)))
+				{
+					*hole = *(hole - 1);
+					hole--;
+				}
+
+				// fill hole with element
+				*hole = val;
+			}
+		}
+	}
+
+	// std variant for elements with ==
+	template <typename I, typename Pred> void partition(I begin, I middle, I end, const Pred& pred, I* out_eqbeg, I* out_eqend)
+	{
+		I eqbeg = middle, eqend = middle + 1;
+
+		// expand equal range
+		while (eqbeg != begin && *(eqbeg - 1) == *eqbeg) --eqbeg;
+		while (eqend != end && *eqend == *eqbeg) ++eqend;
+
+		// process outer elements
+		I ltend = eqbeg, gtbeg = eqend;
+
+		for (;;)
+		{
+			// find the element from the right side that belongs to the left one
+			for (; gtbeg != end; ++gtbeg)
+				if (!pred(*eqbeg, *gtbeg))
+				{
+					if (*gtbeg == *eqbeg) swap(*gtbeg, *eqend++);
+					else break;
+				}
+
+			// find the element from the left side that belongs to the right one
+			for (; ltend != begin; --ltend)
+				if (!pred(*(ltend - 1), *eqbeg))
+				{
+					if (*eqbeg == *(ltend - 1)) swap(*(ltend - 1), *--eqbeg);
+					else break;
+				}
+
+			// scanned all elements
+			if (gtbeg == end && ltend == begin)
+			{
+				*out_eqbeg = eqbeg;
+				*out_eqend = eqend;
+				return;
+			}
+
+			// make room for elements by moving equal area
+			if (gtbeg == end)
+			{
+				if (--ltend != --eqbeg) swap(*ltend, *eqbeg);
+				swap(*eqbeg, *--eqend);
+			}
+			else if (ltend == begin)
+			{
+				if (eqend != gtbeg) swap(*eqbeg, *eqend);
+				++eqend;
+				swap(*gtbeg++, *eqbeg++);
+			}
+			else swap(*gtbeg++, *--ltend);
+		}
+	}
+
+	template <typename I, typename Pred> void median3(I first, I middle, I last, const Pred& pred)
+	{
+		if (pred(*middle, *first)) swap(*middle, *first);
+		if (pred(*last, *middle)) swap(*last, *middle);
+		if (pred(*middle, *first)) swap(*middle, *first);
+	}
+
+	template <typename I, typename Pred> void median(I first, I middle, I last, const Pred& pred)
+	{
+		if (last - first <= 40)
+		{
+			// median of three for small chunks
+			median3(first, middle, last, pred);
+		}
+		else
+		{
+			// median of nine
+			size_t step = (last - first + 1) / 8;
+
+			median3(first, first + step, first + 2 * step, pred);
+			median3(middle - step, middle, middle + step, pred);
+			median3(last - 2 * step, last - step, last, pred);
+			median3(first + step, middle, last - step, pred);
+		}
+	}
+
+	template <typename I, typename Pred> void sort(I begin, I end, const Pred& pred)
+	{
+		// sort large chunks
+		while (end - begin > 32)
+		{
+			// find median element
+			I middle = begin + (end - begin) / 2;
+			median(begin, middle, end - 1, pred);
+
+			// partition in three chunks (< = >)
+			I eqbeg, eqend;
+			partition(begin, middle, end, pred, &eqbeg, &eqend);
+
+			// loop on larger half
+			if (eqbeg - begin > end - eqend)
+			{
+				sort(eqend, end, pred);
+				end = eqbeg;
+			}
+			else
+			{
+				sort(begin, eqbeg, pred);
+				begin = eqend;
+			}
+		}
+
+		// insertion sort small chunk
+		if (begin != end) insertion_sort(begin, end, pred, &*begin);
+	}
+PUGI__NS_END
+
+// Allocator used for AST and evaluation stacks
+PUGI__NS_BEGIN
+	static const size_t xpath_memory_page_size =
+	#ifdef PUGIXML_MEMORY_XPATH_PAGE_SIZE
+		PUGIXML_MEMORY_XPATH_PAGE_SIZE
+	#else
+		4096
+	#endif
+		;
+
+	static const uintptr_t xpath_memory_block_alignment = sizeof(double) > sizeof(void*) ? sizeof(double) : sizeof(void*);
+
+	struct xpath_memory_block
+	{	
+		xpath_memory_block* next;
+		size_t capacity;
+
+		union
+		{
+			char data[xpath_memory_page_size];
+			double alignment;
+		};
+	};
+		
+	class xpath_allocator
+	{
+		xpath_memory_block* _root;
+		size_t _root_size;
+
+	public:
+	#ifdef PUGIXML_NO_EXCEPTIONS
+		jmp_buf* error_handler;
+	#endif
+
+		xpath_allocator(xpath_memory_block* root, size_t root_size = 0): _root(root), _root_size(root_size)
+		{
+		#ifdef PUGIXML_NO_EXCEPTIONS
+			error_handler = 0;
+		#endif
+		}
+		
+		void* allocate_nothrow(size_t size)
+		{
+			// round size up to block alignment boundary
+			size = (size + xpath_memory_block_alignment - 1) & ~(xpath_memory_block_alignment - 1);
+
+			if (_root_size + size <= _root->capacity)
+			{
+				void* buf = &_root->data[0] + _root_size;
+				_root_size += size;
+				return buf;
+			}
+			else
+			{
+				// make sure we have at least 1/4th of the page free after allocation to satisfy subsequent allocation requests
+				size_t block_capacity_base = sizeof(_root->data);
+				size_t block_capacity_req = size + block_capacity_base / 4;
+				size_t block_capacity = (block_capacity_base > block_capacity_req) ? block_capacity_base : block_capacity_req;
+
+				size_t block_size = block_capacity + offsetof(xpath_memory_block, data);
+
+				xpath_memory_block* block = static_cast<xpath_memory_block*>(xml_memory::allocate(block_size));
+				if (!block) return 0;
+				
+				block->next = _root;
+				block->capacity = block_capacity;
+				
+				_root = block;
+				_root_size = size;
+				
+				return block->data;
+			}
+		}
+
+		void* allocate(size_t size)
+		{
+			void* result = allocate_nothrow(size);
+
+			if (!result)
+			{
+			#ifdef PUGIXML_NO_EXCEPTIONS
+				assert(error_handler);
+				longjmp(*error_handler, 1);
+			#else
+				throw std::bad_alloc();
+			#endif
+			}
+
+			return result;
+		}
+
+		void* reallocate(void* ptr, size_t old_size, size_t new_size)
+		{
+			// round size up to block alignment boundary
+			old_size = (old_size + xpath_memory_block_alignment - 1) & ~(xpath_memory_block_alignment - 1);
+			new_size = (new_size + xpath_memory_block_alignment - 1) & ~(xpath_memory_block_alignment - 1);
+
+			// we can only reallocate the last object
+			assert(ptr == 0 || static_cast<char*>(ptr) + old_size == &_root->data[0] + _root_size);
+
+			// adjust root size so that we have not allocated the object at all
+			bool only_object = (_root_size == old_size);
+
+			if (ptr) _root_size -= old_size;
+
+			// allocate a new version (this will obviously reuse the memory if possible)
+			void* result = allocate(new_size);
+			assert(result);
+
+			// we have a new block
+			if (result != ptr && ptr)
+			{
+				// copy old data
+				assert(new_size >= old_size);
+				memcpy(result, ptr, old_size);
+
+				// free the previous page if it had no other objects
+				if (only_object)
+				{
+					assert(_root->data == result);
+					assert(_root->next);
+
+					xpath_memory_block* next = _root->next->next;
+
+					if (next)
+					{
+						// deallocate the whole page, unless it was the first one
+						xml_memory::deallocate(_root->next);
+						_root->next = next;
+					}
+				}
+			}
+
+			return result;
+		}
+
+		void revert(const xpath_allocator& state)
+		{
+			// free all new pages
+			xpath_memory_block* cur = _root;
+
+			while (cur != state._root)
+			{
+				xpath_memory_block* next = cur->next;
+
+				xml_memory::deallocate(cur);
+
+				cur = next;
+			}
+
+			// restore state
+			_root = state._root;
+			_root_size = state._root_size;
+		}
+
+		void release()
+		{
+			xpath_memory_block* cur = _root;
+			assert(cur);
+
+			while (cur->next)
+			{
+				xpath_memory_block* next = cur->next;
+
+				xml_memory::deallocate(cur);
+
+				cur = next;
+			}
+		}
+	};
+
+	struct xpath_allocator_capture
+	{
+		xpath_allocator_capture(xpath_allocator* alloc): _target(alloc), _state(*alloc)
+		{
+		}
+
+		~xpath_allocator_capture()
+		{
+			_target->revert(_state);
+		}
+
+		xpath_allocator* _target;
+		xpath_allocator _state;
+	};
+
+	struct xpath_stack
+	{
+		xpath_allocator* result;
+		xpath_allocator* temp;
+	};
+
+	struct xpath_stack_data
+	{
+		xpath_memory_block blocks[2];
+		xpath_allocator result;
+		xpath_allocator temp;
+		xpath_stack stack;
+
+	#ifdef PUGIXML_NO_EXCEPTIONS
+		jmp_buf error_handler;
+	#endif
+
+		xpath_stack_data(): result(blocks + 0), temp(blocks + 1)
+		{
+			blocks[0].next = blocks[1].next = 0;
+			blocks[0].capacity = blocks[1].capacity = sizeof(blocks[0].data);
+
+			stack.result = &result;
+			stack.temp = &temp;
+
+		#ifdef PUGIXML_NO_EXCEPTIONS
+			result.error_handler = temp.error_handler = &error_handler;
+		#endif
+		}
+
+		~xpath_stack_data()
+		{
+			result.release();
+			temp.release();
+		}
+	};
+PUGI__NS_END
+
+// String class
+PUGI__NS_BEGIN
+	class xpath_string
+	{
+		const char_t* _buffer;
+		bool _uses_heap;
+		size_t _length_heap;
+
+		static char_t* duplicate_string(const char_t* string, size_t length, xpath_allocator* alloc)
+		{
+			char_t* result = static_cast<char_t*>(alloc->allocate((length + 1) * sizeof(char_t)));
+			assert(result);
+
+			memcpy(result, string, length * sizeof(char_t));
+			result[length] = 0;
+
+			return result;
+		}
+
+		xpath_string(const char_t* buffer, bool uses_heap_, size_t length_heap): _buffer(buffer), _uses_heap(uses_heap_), _length_heap(length_heap)
+		{
+		}
+
+	public:
+		static xpath_string from_const(const char_t* str)
+		{
+			return xpath_string(str, false, 0);
+		}
+
+		static xpath_string from_heap_preallocated(const char_t* begin, const char_t* end)
+		{
+			assert(begin <= end && *end == 0);
+
+			return xpath_string(begin, true, static_cast<size_t>(end - begin));
+		}
+
+		static xpath_string from_heap(const char_t* begin, const char_t* end, xpath_allocator* alloc)
+		{
+			assert(begin <= end);
+
+			size_t length = static_cast<size_t>(end - begin);
+
+			return length == 0 ? xpath_string() : xpath_string(duplicate_string(begin, length, alloc), true, length);
+		}
+
+		xpath_string(): _buffer(PUGIXML_TEXT("")), _uses_heap(false), _length_heap(0)
+		{
+		}
+
+		void append(const xpath_string& o, xpath_allocator* alloc)
+		{
+			// skip empty sources
+			if (!*o._buffer) return;
+
+			// fast append for constant empty target and constant source
+			if (!*_buffer && !_uses_heap && !o._uses_heap)
+			{
+				_buffer = o._buffer;
+			}
+			else
+			{
+				// need to make heap copy
+				size_t target_length = length();
+				size_t source_length = o.length();
+				size_t result_length = target_length + source_length;
+
+				// allocate new buffer
+				char_t* result = static_cast<char_t*>(alloc->reallocate(_uses_heap ? const_cast<char_t*>(_buffer) : 0, (target_length + 1) * sizeof(char_t), (result_length + 1) * sizeof(char_t)));
+				assert(result);
+
+				// append first string to the new buffer in case there was no reallocation
+				if (!_uses_heap) memcpy(result, _buffer, target_length * sizeof(char_t));
+
+				// append second string to the new buffer
+				memcpy(result + target_length, o._buffer, source_length * sizeof(char_t));
+				result[result_length] = 0;
+
+				// finalize
+				_buffer = result;
+				_uses_heap = true;
+				_length_heap = result_length;
+			}
+		}
+
+		const char_t* c_str() const
+		{
+			return _buffer;
+		}
+
+		size_t length() const
+		{
+			return _uses_heap ? _length_heap : strlength(_buffer);
+		}
+		
+		char_t* data(xpath_allocator* alloc)
+		{
+			// make private heap copy
+			if (!_uses_heap)
+			{
+				size_t length_ = strlength(_buffer);
+
+				_buffer = duplicate_string(_buffer, length_, alloc);
+				_uses_heap = true;
+				_length_heap = length_;
+			}
+
+			return const_cast<char_t*>(_buffer);
+		}
+
+		bool empty() const
+		{
+			return *_buffer == 0;
+		}
+
+		bool operator==(const xpath_string& o) const
+		{
+			return strequal(_buffer, o._buffer);
+		}
+
+		bool operator!=(const xpath_string& o) const
+		{
+			return !strequal(_buffer, o._buffer);
+		}
+
+		bool uses_heap() const
+		{
+			return _uses_heap;
+		}
+	};
+PUGI__NS_END
+
+PUGI__NS_BEGIN
+	PUGI__FN bool starts_with(const char_t* string, const char_t* pattern)
+	{
+		while (*pattern && *string == *pattern)
+		{
+			string++;
+			pattern++;
+		}
+
+		return *pattern == 0;
+	}
+
+	PUGI__FN const char_t* find_char(const char_t* s, char_t c)
+	{
+	#ifdef PUGIXML_WCHAR_MODE
+		return wcschr(s, c);
+	#else
+		return strchr(s, c);
+	#endif
+	}
+
+	PUGI__FN const char_t* find_substring(const char_t* s, const char_t* p)
+	{
+	#ifdef PUGIXML_WCHAR_MODE
+		// MSVC6 wcsstr bug workaround (if s is empty it always returns 0)
+		return (*p == 0) ? s : wcsstr(s, p);
+	#else
+		return strstr(s, p);
+	#endif
+	}
+
+	// Converts symbol to lower case, if it is an ASCII one
+	PUGI__FN char_t tolower_ascii(char_t ch)
+	{
+		return static_cast<unsigned int>(ch - 'A') < 26 ? static_cast<char_t>(ch | ' ') : ch;
+	}
+
+	PUGI__FN xpath_string string_value(const xpath_node& na, xpath_allocator* alloc)
+	{
+		if (na.attribute())
+			return xpath_string::from_const(na.attribute().value());
+		else
+		{
+			xml_node n = na.node();
+
+			switch (n.type())
+			{
+			case node_pcdata:
+			case node_cdata:
+			case node_comment:
+			case node_pi:
+				return xpath_string::from_const(n.value());
+			
+			case node_document:
+			case node_element:
+			{
+				xpath_string result;
+
+				xml_node cur = n.first_child();
+				
+				while (cur && cur != n)
+				{
+					if (cur.type() == node_pcdata || cur.type() == node_cdata)
+						result.append(xpath_string::from_const(cur.value()), alloc);
+
+					if (cur.first_child())
+						cur = cur.first_child();
+					else if (cur.next_sibling())
+						cur = cur.next_sibling();
+					else
+					{
+						while (!cur.next_sibling() && cur != n)
+							cur = cur.parent();
+
+						if (cur != n) cur = cur.next_sibling();
+					}
+				}
+				
+				return result;
+			}
+			
+			default:
+				return xpath_string();
+			}
+		}
+	}
+	
+	PUGI__FN bool node_is_before_sibling(xml_node_struct* ln, xml_node_struct* rn)
+	{
+		assert(ln->parent == rn->parent);
+
+		// there is no common ancestor (the shared parent is null), nodes are from different documents
+		if (!ln->parent) return ln < rn;
+
+		// determine sibling order
+		xml_node_struct* ls = ln;
+		xml_node_struct* rs = rn;
+
+		while (ls && rs)
+		{
+			if (ls == rn) return true;
+			if (rs == ln) return false;
+
+			ls = ls->next_sibling;
+			rs = rs->next_sibling;
+		}
+
+		// if rn sibling chain ended ln must be before rn
+		return !rs;
+	}
+	
+	PUGI__FN bool node_is_before(xml_node_struct* ln, xml_node_struct* rn)
+	{
+		// find common ancestor at the same depth, if any
+		xml_node_struct* lp = ln;
+		xml_node_struct* rp = rn;
+
+		while (lp && rp && lp->parent != rp->parent)
+		{
+			lp = lp->parent;
+			rp = rp->parent;
+		}
+
+		// parents are the same!
+		if (lp && rp) return node_is_before_sibling(lp, rp);
+
+		// nodes are at different depths, need to normalize heights
+		bool left_higher = !lp;
+
+		while (lp)
+		{
+			lp = lp->parent;
+			ln = ln->parent;
+		}
+
+		while (rp)
+		{
+			rp = rp->parent;
+			rn = rn->parent;
+		}
+
+		// one node is the ancestor of the other
+		if (ln == rn) return left_higher;
+
+		// find common ancestor... again
+		while (ln->parent != rn->parent)
+		{
+			ln = ln->parent;
+			rn = rn->parent;
+		}
+
+		return node_is_before_sibling(ln, rn);
+	}
+
+	PUGI__FN bool node_is_ancestor(xml_node_struct* parent, xml_node_struct* node)
+	{
+		while (node && node != parent) node = node->parent;
+
+		return parent && node == parent;
+	}
+
+	PUGI__FN const void* document_buffer_order(const xpath_node& xnode)
+	{
+		xml_node_struct* node = xnode.node().internal_object();
+
+		if (node)
+		{
+			if ((get_document(node).header & xml_memory_page_contents_shared_mask) == 0)
+			{
+				if (node->name && (node->header & impl::xml_memory_page_name_allocated_or_shared_mask) == 0) return node->name;
+				if (node->value && (node->header & impl::xml_memory_page_value_allocated_or_shared_mask) == 0) return node->value;
+			}
+
+			return 0;
+		}
+
+		xml_attribute_struct* attr = xnode.attribute().internal_object();
+
+		if (attr)
+		{
+			if ((get_document(attr).header & xml_memory_page_contents_shared_mask) == 0)
+			{
+				if ((attr->header & impl::xml_memory_page_name_allocated_or_shared_mask) == 0) return attr->name;
+				if ((attr->header & impl::xml_memory_page_value_allocated_or_shared_mask) == 0) return attr->value;
+			}
+
+			return 0;
+		}
+
+		return 0;
+	}
+	
+	struct document_order_comparator
+	{
+		bool operator()(const xpath_node& lhs, const xpath_node& rhs) const
+		{
+			// optimized document order based check
+			const void* lo = document_buffer_order(lhs);
+			const void* ro = document_buffer_order(rhs);
+
+			if (lo && ro) return lo < ro;
+
+			// slow comparison
+			xml_node ln = lhs.node(), rn = rhs.node();
+
+			// compare attributes
+			if (lhs.attribute() && rhs.attribute())
+			{
+				// shared parent
+				if (lhs.parent() == rhs.parent())
+				{
+					// determine sibling order
+					for (xml_attribute a = lhs.attribute(); a; a = a.next_attribute())
+						if (a == rhs.attribute())
+							return true;
+					
+					return false;
+				}
+				
+				// compare attribute parents
+				ln = lhs.parent();
+				rn = rhs.parent();
+			}
+			else if (lhs.attribute())
+			{
+				// attributes go after the parent element
+				if (lhs.parent() == rhs.node()) return false;
+				
+				ln = lhs.parent();
+			}
+			else if (rhs.attribute())
+			{
+				// attributes go after the parent element
+				if (rhs.parent() == lhs.node()) return true;
+				
+				rn = rhs.parent();
+			}
+
+			if (ln == rn) return false;
+
+			if (!ln || !rn) return ln < rn;
+			
+			return node_is_before(ln.internal_object(), rn.internal_object());
+		}
+	};
+
+	struct duplicate_comparator
+	{
+		bool operator()(const xpath_node& lhs, const xpath_node& rhs) const
+		{
+			if (lhs.attribute()) return rhs.attribute() ? lhs.attribute() < rhs.attribute() : true;
+			else return rhs.attribute() ? false : lhs.node() < rhs.node();
+		}
+	};
+	
+	PUGI__FN double gen_nan()
+	{
+	#if defined(__STDC_IEC_559__) || ((FLT_RADIX - 0 == 2) && (FLT_MAX_EXP - 0 == 128) && (FLT_MANT_DIG - 0 == 24))
+		union { float f; uint32_t i; } u[sizeof(float) == sizeof(uint32_t) ? 1 : -1];
+		u[0].i = 0x7fc00000;
+		return u[0].f;
+	#else
+		// fallback
+		const volatile double zero = 0.0;
+		return zero / zero;
+	#endif
+	}
+	
+	PUGI__FN bool is_nan(double value)
+	{
+	#if defined(PUGI__MSVC_CRT_VERSION) || defined(__BORLANDC__)
+		return !!_isnan(value);
+	#elif defined(fpclassify) && defined(FP_NAN)
+		return fpclassify(value) == FP_NAN;
+	#else
+		// fallback
+		const volatile double v = value;
+		return v != v;
+	#endif
+	}
+	
+	PUGI__FN const char_t* convert_number_to_string_special(double value)
+	{
+	#if defined(PUGI__MSVC_CRT_VERSION) || defined(__BORLANDC__)
+		if (_finite(value)) return (value == 0) ? PUGIXML_TEXT("0") : 0;
+		if (_isnan(value)) return PUGIXML_TEXT("NaN");
+		return value > 0 ? PUGIXML_TEXT("Infinity") : PUGIXML_TEXT("-Infinity");
+	#elif defined(fpclassify) && defined(FP_NAN) && defined(FP_INFINITE) && defined(FP_ZERO)
+		switch (fpclassify(value))
+		{
+		case FP_NAN:
+			return PUGIXML_TEXT("NaN");
+
+		case FP_INFINITE:
+			return value > 0 ? PUGIXML_TEXT("Infinity") : PUGIXML_TEXT("-Infinity");
+
+		case FP_ZERO:
+			return PUGIXML_TEXT("0");
+
+		default:
+			return 0;
+		}
+	#else
+		// fallback
+		const volatile double v = value;
+
+		if (v == 0) return PUGIXML_TEXT("0");
+		if (v != v) return PUGIXML_TEXT("NaN");
+		if (v * 2 == v) return value > 0 ? PUGIXML_TEXT("Infinity") : PUGIXML_TEXT("-Infinity");
+		return 0;
+	#endif
+	}
+	
+	PUGI__FN bool convert_number_to_boolean(double value)
+	{
+		return (value != 0 && !is_nan(value));
+	}
+	
+	PUGI__FN void truncate_zeros(char* begin, char* end)
+	{
+		while (begin != end && end[-1] == '0') end--;
+
+		*end = 0;
+	}
+
+	// gets mantissa digits in the form of 0.xxxxx with 0. implied and the exponent
+#if defined(PUGI__MSVC_CRT_VERSION) && PUGI__MSVC_CRT_VERSION >= 1400 && !defined(_WIN32_WCE)
+	PUGI__FN void convert_number_to_mantissa_exponent(double value, char* buffer, size_t buffer_size, char** out_mantissa, int* out_exponent)
+	{
+		// get base values
+		int sign, exponent;
+		_ecvt_s(buffer, buffer_size, value, DBL_DIG + 1, &exponent, &sign);
+
+		// truncate redundant zeros
+		truncate_zeros(buffer, buffer + strlen(buffer));
+
+		// fill results
+		*out_mantissa = buffer;
+		*out_exponent = exponent;
+	}
+#else
+	PUGI__FN void convert_number_to_mantissa_exponent(double value, char* buffer, size_t buffer_size, char** out_mantissa, int* out_exponent)
+	{
+		// get a scientific notation value with IEEE DBL_DIG decimals
+		sprintf(buffer, "%.*e", DBL_DIG, value);
+		assert(strlen(buffer) < buffer_size);
+		(void)!buffer_size;
+
+		// get the exponent (possibly negative)
+		char* exponent_string = strchr(buffer, 'e');
+		assert(exponent_string);
+
+		int exponent = atoi(exponent_string + 1);
+
+		// extract mantissa string: skip sign
+		char* mantissa = buffer[0] == '-' ? buffer + 1 : buffer;
+		assert(mantissa[0] != '0' && mantissa[1] == '.');
+
+		// divide mantissa by 10 to eliminate integer part
+		mantissa[1] = mantissa[0];
+		mantissa++;
+		exponent++;
+
+		// remove extra mantissa digits and zero-terminate mantissa
+		truncate_zeros(mantissa, exponent_string);
+
+		// fill results
+		*out_mantissa = mantissa;
+		*out_exponent = exponent;
+	}
+#endif
+
+	PUGI__FN xpath_string convert_number_to_string(double value, xpath_allocator* alloc)
+	{
+		// try special number conversion
+		const char_t* special = convert_number_to_string_special(value);
+		if (special) return xpath_string::from_const(special);
+
+		// get mantissa + exponent form
+		char mantissa_buffer[32];
+
+		char* mantissa;
+		int exponent;
+		convert_number_to_mantissa_exponent(value, mantissa_buffer, sizeof(mantissa_buffer), &mantissa, &exponent);
+
+		// allocate a buffer of suitable length for the number
+		size_t result_size = strlen(mantissa_buffer) + (exponent > 0 ? exponent : -exponent) + 4;
+		char_t* result = static_cast<char_t*>(alloc->allocate(sizeof(char_t) * result_size));
+		assert(result);
+
+		// make the number!
+		char_t* s = result;
+
+		// sign
+		if (value < 0) *s++ = '-';
+
+		// integer part
+		if (exponent <= 0)
+		{
+			*s++ = '0';
+		}
+		else
+		{
+			while (exponent > 0)
+			{
+				assert(*mantissa == 0 || static_cast<unsigned int>(static_cast<unsigned int>(*mantissa) - '0') <= 9);
+				*s++ = *mantissa ? *mantissa++ : '0';
+				exponent--;
+			}
+		}
+
+		// fractional part
+		if (*mantissa)
+		{
+			// decimal point
+			*s++ = '.';
+
+			// extra zeroes from negative exponent
+			while (exponent < 0)
+			{
+				*s++ = '0';
+				exponent++;
+			}
+
+			// extra mantissa digits
+			while (*mantissa)
+			{
+				assert(static_cast<unsigned int>(*mantissa - '0') <= 9);
+				*s++ = *mantissa++;
+			}
+		}
+
+		// zero-terminate
+		assert(s < result + result_size);
+		*s = 0;
+
+		return xpath_string::from_heap_preallocated(result, s);
+	}
+	
+	PUGI__FN bool check_string_to_number_format(const char_t* string)
+	{
+		// parse leading whitespace
+		while (PUGI__IS_CHARTYPE(*string, ct_space)) ++string;
+
+		// parse sign
+		if (*string == '-') ++string;
+
+		if (!*string) return false;
+
+		// if there is no integer part, there should be a decimal part with at least one digit
+		if (!PUGI__IS_CHARTYPEX(string[0], ctx_digit) && (string[0] != '.' || !PUGI__IS_CHARTYPEX(string[1], ctx_digit))) return false;
+
+		// parse integer part
+		while (PUGI__IS_CHARTYPEX(*string, ctx_digit)) ++string;
+
+		// parse decimal part
+		if (*string == '.')
+		{
+			++string;
+
+			while (PUGI__IS_CHARTYPEX(*string, ctx_digit)) ++string;
+		}
+
+		// parse trailing whitespace
+		while (PUGI__IS_CHARTYPE(*string, ct_space)) ++string;
+
+		return *string == 0;
+	}
+
+	PUGI__FN double convert_string_to_number(const char_t* string)
+	{
+		// check string format
+		if (!check_string_to_number_format(string)) return gen_nan();
+
+		// parse string
+	#ifdef PUGIXML_WCHAR_MODE
+		return wcstod(string, 0);
+	#else
+		return strtod(string, 0);
+	#endif
+	}
+
+	PUGI__FN bool convert_string_to_number_scratch(char_t (&buffer)[32], const char_t* begin, const char_t* end, double* out_result)
+	{
+		size_t length = static_cast<size_t>(end - begin);
+		char_t* scratch = buffer;
+
+		if (length >= sizeof(buffer) / sizeof(buffer[0]))
+		{
+			// need to make dummy on-heap copy
+			scratch = static_cast<char_t*>(xml_memory::allocate((length + 1) * sizeof(char_t)));
+			if (!scratch) return false;
+		}
+
+		// copy string to zero-terminated buffer and perform conversion
+		memcpy(scratch, begin, length * sizeof(char_t));
+		scratch[length] = 0;
+
+		*out_result = convert_string_to_number(scratch);
+
+		// free dummy buffer
+		if (scratch != buffer) xml_memory::deallocate(scratch);
+
+		return true;
+	}
+	
+	PUGI__FN double round_nearest(double value)
+	{
+		return floor(value + 0.5);
+	}
+
+	PUGI__FN double round_nearest_nzero(double value)
+	{
+		// same as round_nearest, but returns -0 for [-0.5, -0]
+		// ceil is used to differentiate between +0 and -0 (we return -0 for [-0.5, -0] and +0 for +0)
+		return (value >= -0.5 && value <= 0) ? ceil(value) : floor(value + 0.5);
+	}
+	
+	PUGI__FN const char_t* qualified_name(const xpath_node& node)
+	{
+		return node.attribute() ? node.attribute().name() : node.node().name();
+	}
+	
+	PUGI__FN const char_t* local_name(const xpath_node& node)
+	{
+		const char_t* name = qualified_name(node);
+		const char_t* p = find_char(name, ':');
+		
+		return p ? p + 1 : name;
+	}
+
+	struct namespace_uri_predicate
+	{
+		const char_t* prefix;
+		size_t prefix_length;
+
+		namespace_uri_predicate(const char_t* name)
+		{
+			const char_t* pos = find_char(name, ':');
+
+			prefix = pos ? name : 0;
+			prefix_length = pos ? static_cast<size_t>(pos - name) : 0;
+		}
+
+		bool operator()(xml_attribute a) const
+		{
+			const char_t* name = a.name();
+
+			if (!starts_with(name, PUGIXML_TEXT("xmlns"))) return false;
+
+			return prefix ? name[5] == ':' && strequalrange(name + 6, prefix, prefix_length) : name[5] == 0;
+		}
+	};
+
+	PUGI__FN const char_t* namespace_uri(xml_node node)
+	{
+		namespace_uri_predicate pred = node.name();
+		
+		xml_node p = node;
+		
+		while (p)
+		{
+			xml_attribute a = p.find_attribute(pred);
+			
+			if (a) return a.value();
+			
+			p = p.parent();
+		}
+		
+		return PUGIXML_TEXT("");
+	}
+
+	PUGI__FN const char_t* namespace_uri(xml_attribute attr, xml_node parent)
+	{
+		namespace_uri_predicate pred = attr.name();
+		
+		// Default namespace does not apply to attributes
+		if (!pred.prefix) return PUGIXML_TEXT("");
+		
+		xml_node p = parent;
+		
+		while (p)
+		{
+			xml_attribute a = p.find_attribute(pred);
+			
+			if (a) return a.value();
+			
+			p = p.parent();
+		}
+		
+		return PUGIXML_TEXT("");
+	}
+
+	PUGI__FN const char_t* namespace_uri(const xpath_node& node)
+	{
+		return node.attribute() ? namespace_uri(node.attribute(), node.parent()) : namespace_uri(node.node());
+	}
+
+	PUGI__FN char_t* normalize_space(char_t* buffer)
+	{
+		char_t* write = buffer;
+
+		for (char_t* it = buffer; *it; )
+		{
+			char_t ch = *it++;
+
+			if (PUGI__IS_CHARTYPE(ch, ct_space))
+			{
+				// replace whitespace sequence with single space
+				while (PUGI__IS_CHARTYPE(*it, ct_space)) it++;
+
+				// avoid leading spaces
+				if (write != buffer) *write++ = ' ';
+			}
+			else *write++ = ch;
+		}
+
+		// remove trailing space
+		if (write != buffer && PUGI__IS_CHARTYPE(write[-1], ct_space)) write--;
+
+		// zero-terminate
+		*write = 0;
+
+		return write;
+	}
+
+	PUGI__FN char_t* translate(char_t* buffer, const char_t* from, const char_t* to, size_t to_length)
+	{
+		char_t* write = buffer;
+
+		while (*buffer)
+		{
+			PUGI__DMC_VOLATILE char_t ch = *buffer++;
+
+			const char_t* pos = find_char(from, ch);
+
+			if (!pos)
+				*write++ = ch; // do not process
+			else if (static_cast<size_t>(pos - from) < to_length)
+				*write++ = to[pos - from]; // replace
+		}
+
+		// zero-terminate
+		*write = 0;
+
+		return write;
+	}
+
+	PUGI__FN unsigned char* translate_table_generate(xpath_allocator* alloc, const char_t* from, const char_t* to)
+	{
+		unsigned char table[128] = {0};
+
+		while (*from)
+		{
+			unsigned int fc = static_cast<unsigned int>(*from);
+			unsigned int tc = static_cast<unsigned int>(*to);
+
+			if (fc >= 128 || tc >= 128)
+				return 0;
+
+			// code=128 means "skip character"
+			if (!table[fc])
+				table[fc] = static_cast<unsigned char>(tc ? tc : 128);
+
+			from++;
+			if (tc) to++;
+		}
+
+		for (int i = 0; i < 128; ++i)
+			if (!table[i])
+				table[i] = static_cast<unsigned char>(i);
+
+		void* result = alloc->allocate_nothrow(sizeof(table));
+
+		if (result)
+		{
+			memcpy(result, table, sizeof(table));
+		}
+
+		return static_cast<unsigned char*>(result);
+	}
+
+	PUGI__FN char_t* translate_table(char_t* buffer, const unsigned char* table)
+	{
+		char_t* write = buffer;
+
+		while (*buffer)
+		{
+			char_t ch = *buffer++;
+			unsigned int index = static_cast<unsigned int>(ch);
+
+			if (index < 128)
+			{
+				unsigned char code = table[index];
+
+				// code=128 means "skip character" (table size is 128 so 128 can be a special value)
+				// this code skips these characters without extra branches
+				*write = static_cast<char_t>(code);
+				write += 1 - (code >> 7);
+			}
+			else
+			{
+				*write++ = ch;
+			}
+		}
+
+		// zero-terminate
+		*write = 0;
+
+		return write;
+	}
+
+	inline bool is_xpath_attribute(const char_t* name)
+	{
+		return !(starts_with(name, PUGIXML_TEXT("xmlns")) && (name[5] == 0 || name[5] == ':'));
+	}
+
+	struct xpath_variable_boolean: xpath_variable
+	{
+		xpath_variable_boolean(): xpath_variable(xpath_type_boolean), value(false)
+		{
+		}
+
+		bool value;
+		char_t name[1];
+	};
+
+	struct xpath_variable_number: xpath_variable
+	{
+		xpath_variable_number(): xpath_variable(xpath_type_number), value(0)
+		{
+		}
+
+		double value;
+		char_t name[1];
+	};
+
+	struct xpath_variable_string: xpath_variable
+	{
+		xpath_variable_string(): xpath_variable(xpath_type_string), value(0)
+		{
+		}
+
+		~xpath_variable_string()
+		{
+			if (value) xml_memory::deallocate(value);
+		}
+
+		char_t* value;
+		char_t name[1];
+	};
+
+	struct xpath_variable_node_set: xpath_variable
+	{
+		xpath_variable_node_set(): xpath_variable(xpath_type_node_set)
+		{
+		}
+
+		xpath_node_set value;
+		char_t name[1];
+	};
+
+	static const xpath_node_set dummy_node_set;
+
+	PUGI__FN unsigned int hash_string(const char_t* str)
+	{
+		// Jenkins one-at-a-time hash (http://en.wikipedia.org/wiki/Jenkins_hash_function#one-at-a-time)
+		unsigned int result = 0;
+
+		while (*str)
+		{
+			result += static_cast<unsigned int>(*str++);
+			result += result << 10;
+			result ^= result >> 6;
+		}
+	
+		result += result << 3;
+		result ^= result >> 11;
+		result += result << 15;
+	
+		return result;
+	}
+
+	template <typename T> PUGI__FN T* new_xpath_variable(const char_t* name)
+	{
+		size_t length = strlength(name);
+		if (length == 0) return 0; // empty variable names are invalid
+
+		// $$ we can't use offsetof(T, name) because T is non-POD, so we just allocate additional length characters
+		void* memory = xml_memory::allocate(sizeof(T) + length * sizeof(char_t));
+		if (!memory) return 0;
+
+		T* result = new (memory) T();
+
+		memcpy(result->name, name, (length + 1) * sizeof(char_t));
+
+		return result;
+	}
+
+	PUGI__FN xpath_variable* new_xpath_variable(xpath_value_type type, const char_t* name)
+	{
+		switch (type)
+		{
+		case xpath_type_node_set:
+			return new_xpath_variable<xpath_variable_node_set>(name);
+
+		case xpath_type_number:
+			return new_xpath_variable<xpath_variable_number>(name);
+
+		case xpath_type_string:
+			return new_xpath_variable<xpath_variable_string>(name);
+
+		case xpath_type_boolean:
+			return new_xpath_variable<xpath_variable_boolean>(name);
+
+		default:
+			return 0;
+		}
+	}
+
+	template <typename T> PUGI__FN void delete_xpath_variable(T* var)
+	{
+		var->~T();
+		xml_memory::deallocate(var);
+	}
+
+	PUGI__FN void delete_xpath_variable(xpath_value_type type, xpath_variable* var)
+	{
+		switch (type)
+		{
+		case xpath_type_node_set:
+			delete_xpath_variable(static_cast<xpath_variable_node_set*>(var));
+			break;
+
+		case xpath_type_number:
+			delete_xpath_variable(static_cast<xpath_variable_number*>(var));
+			break;
+
+		case xpath_type_string:
+			delete_xpath_variable(static_cast<xpath_variable_string*>(var));
+			break;
+
+		case xpath_type_boolean:
+			delete_xpath_variable(static_cast<xpath_variable_boolean*>(var));
+			break;
+
+		default:
+			assert(!"Invalid variable type");
+		}
+	}
+
+	PUGI__FN bool copy_xpath_variable(xpath_variable* lhs, const xpath_variable* rhs)
+	{
+		switch (rhs->type())
+		{
+		case xpath_type_node_set:
+			return lhs->set(static_cast<const xpath_variable_node_set*>(rhs)->value);
+
+		case xpath_type_number:
+			return lhs->set(static_cast<const xpath_variable_number*>(rhs)->value);
+
+		case xpath_type_string:
+			return lhs->set(static_cast<const xpath_variable_string*>(rhs)->value);
+
+		case xpath_type_boolean:
+			return lhs->set(static_cast<const xpath_variable_boolean*>(rhs)->value);
+
+		default:
+			assert(!"Invalid variable type");
+			return false;
+		}
+	}
+
+	PUGI__FN bool get_variable_scratch(char_t (&buffer)[32], xpath_variable_set* set, const char_t* begin, const char_t* end, xpath_variable** out_result)
+	{
+		size_t length = static_cast<size_t>(end - begin);
+		char_t* scratch = buffer;
+
+		if (length >= sizeof(buffer) / sizeof(buffer[0]))
+		{
+			// need to make dummy on-heap copy
+			scratch = static_cast<char_t*>(xml_memory::allocate((length + 1) * sizeof(char_t)));
+			if (!scratch) return false;
+		}
+
+		// copy string to zero-terminated buffer and perform lookup
+		memcpy(scratch, begin, length * sizeof(char_t));
+		scratch[length] = 0;
+
+		*out_result = set->get(scratch);
+
+		// free dummy buffer
+		if (scratch != buffer) xml_memory::deallocate(scratch);
+
+		return true;
+	}
+PUGI__NS_END
+
+// Internal node set class
+PUGI__NS_BEGIN
+	PUGI__FN xpath_node_set::type_t xpath_get_order(const xpath_node* begin, const xpath_node* end)
+	{
+		if (end - begin < 2)
+			return xpath_node_set::type_sorted;
+
+		document_order_comparator cmp;
+
+		bool first = cmp(begin[0], begin[1]);
+
+		for (const xpath_node* it = begin + 1; it + 1 < end; ++it)
+			if (cmp(it[0], it[1]) != first)
+				return xpath_node_set::type_unsorted;
+
+		return first ? xpath_node_set::type_sorted : xpath_node_set::type_sorted_reverse;
+	}
+
+	PUGI__FN xpath_node_set::type_t xpath_sort(xpath_node* begin, xpath_node* end, xpath_node_set::type_t type, bool rev)
+	{
+		xpath_node_set::type_t order = rev ? xpath_node_set::type_sorted_reverse : xpath_node_set::type_sorted;
+
+		if (type == xpath_node_set::type_unsorted)
+		{
+			xpath_node_set::type_t sorted = xpath_get_order(begin, end);
+
+			if (sorted == xpath_node_set::type_unsorted)
+			{
+				sort(begin, end, document_order_comparator());
+
+				type = xpath_node_set::type_sorted;
+			}
+			else
+				type = sorted;
+		}
+		
+		if (type != order) reverse(begin, end);
+			
+		return order;
+	}
+
+	PUGI__FN xpath_node xpath_first(const xpath_node* begin, const xpath_node* end, xpath_node_set::type_t type)
+	{
+		if (begin == end) return xpath_node();
+
+		switch (type)
+		{
+		case xpath_node_set::type_sorted:
+			return *begin;
+
+		case xpath_node_set::type_sorted_reverse:
+			return *(end - 1);
+
+		case xpath_node_set::type_unsorted:
+			return *min_element(begin, end, document_order_comparator());
+
+		default:
+			assert(!"Invalid node set type");
+			return xpath_node();
+		}
+	}
+
+	class xpath_node_set_raw
+	{
+		xpath_node_set::type_t _type;
+
+		xpath_node* _begin;
+		xpath_node* _end;
+		xpath_node* _eos;
+
+	public:
+		xpath_node_set_raw(): _type(xpath_node_set::type_unsorted), _begin(0), _end(0), _eos(0)
+		{
+		}
+
+		xpath_node* begin() const
+		{
+			return _begin;
+		}
+
+		xpath_node* end() const
+		{
+			return _end;
+		}
+
+		bool empty() const
+		{
+			return _begin == _end;
+		}
+
+		size_t size() const
+		{
+			return static_cast<size_t>(_end - _begin);
+		}
+
+		xpath_node first() const
+		{
+			return xpath_first(_begin, _end, _type);
+		}
+
+		void push_back_grow(const xpath_node& node, xpath_allocator* alloc);
+
+		void push_back(const xpath_node& node, xpath_allocator* alloc)
+		{
+			if (_end != _eos)
+				*_end++ = node;
+			else
+				push_back_grow(node, alloc);
+		}
+
+		void append(const xpath_node* begin_, const xpath_node* end_, xpath_allocator* alloc)
+		{
+			if (begin_ == end_) return;
+
+			size_t size_ = static_cast<size_t>(_end - _begin);
+			size_t capacity = static_cast<size_t>(_eos - _begin);
+			size_t count = static_cast<size_t>(end_ - begin_);
+
+			if (size_ + count > capacity)
+			{
+				// reallocate the old array or allocate a new one
+				xpath_node* data = static_cast<xpath_node*>(alloc->reallocate(_begin, capacity * sizeof(xpath_node), (size_ + count) * sizeof(xpath_node)));
+				assert(data);
+
+				// finalize
+				_begin = data;
+				_end = data + size_;
+				_eos = data + size_ + count;
+			}
+
+			memcpy(_end, begin_, count * sizeof(xpath_node));
+			_end += count;
+		}
+
+		void sort_do()
+		{
+			_type = xpath_sort(_begin, _end, _type, false);
+		}
+
+		void truncate(xpath_node* pos)
+		{
+			assert(_begin <= pos && pos <= _end);
+
+			_end = pos;
+		}
+
+		void remove_duplicates()
+		{
+			if (_type == xpath_node_set::type_unsorted)
+				sort(_begin, _end, duplicate_comparator());
+		
+			_end = unique(_begin, _end);
+		}
+
+		xpath_node_set::type_t type() const
+		{
+			return _type;
+		}
+
+		void set_type(xpath_node_set::type_t value)
+		{
+			_type = value;
+		}
+	};
+
+	PUGI__FN_NO_INLINE void xpath_node_set_raw::push_back_grow(const xpath_node& node, xpath_allocator* alloc)
+	{
+		size_t capacity = static_cast<size_t>(_eos - _begin);
+
+		// get new capacity (1.5x rule)
+		size_t new_capacity = capacity + capacity / 2 + 1;
+
+		// reallocate the old array or allocate a new one
+		xpath_node* data = static_cast<xpath_node*>(alloc->reallocate(_begin, capacity * sizeof(xpath_node), new_capacity * sizeof(xpath_node)));
+		assert(data);
+
+		// finalize
+		_begin = data;
+		_end = data + capacity;
+		_eos = data + new_capacity;
+
+		// push
+		*_end++ = node;
+	}
+PUGI__NS_END
+
+PUGI__NS_BEGIN
+	struct xpath_context
+	{
+		xpath_node n;
+		size_t position, size;
+
+		xpath_context(const xpath_node& n_, size_t position_, size_t size_): n(n_), position(position_), size(size_)
+		{
+		}
+	};
+
+	enum lexeme_t
+	{
+		lex_none = 0,
+		lex_equal,
+		lex_not_equal,
+		lex_less,
+		lex_greater,
+		lex_less_or_equal,
+		lex_greater_or_equal,
+		lex_plus,
+		lex_minus,
+		lex_multiply,
+		lex_union,
+		lex_var_ref,
+		lex_open_brace,
+		lex_close_brace,
+		lex_quoted_string,
+		lex_number,
+		lex_slash,
+		lex_double_slash,
+		lex_open_square_brace,
+		lex_close_square_brace,
+		lex_string,
+		lex_comma,
+		lex_axis_attribute,
+		lex_dot,
+		lex_double_dot,
+		lex_double_colon,
+		lex_eof
+	};
+
+	struct xpath_lexer_string
+	{
+		const char_t* begin;
+		const char_t* end;
+
+		xpath_lexer_string(): begin(0), end(0)
+		{
+		}
+
+		bool operator==(const char_t* other) const
+		{
+			size_t length = static_cast<size_t>(end - begin);
+
+			return strequalrange(other, begin, length);
+		}
+	};
+
+	class xpath_lexer
+	{
+		const char_t* _cur;
+		const char_t* _cur_lexeme_pos;
+		xpath_lexer_string _cur_lexeme_contents;
+
+		lexeme_t _cur_lexeme;
+
+	public:
+		explicit xpath_lexer(const char_t* query): _cur(query)
+		{
+			next();
+		}
+		
+		const char_t* state() const
+		{
+			return _cur;
+		}
+		
+		void next()
+		{
+			const char_t* cur = _cur;
+
+			while (PUGI__IS_CHARTYPE(*cur, ct_space)) ++cur;
+
+			// save lexeme position for error reporting
+			_cur_lexeme_pos = cur;
+
+			switch (*cur)
+			{
+			case 0:
+				_cur_lexeme = lex_eof;
+				break;
+			
+			case '>':
+				if (*(cur+1) == '=')
+				{
+					cur += 2;
+					_cur_lexeme = lex_greater_or_equal;
+				}
+				else
+				{
+					cur += 1;
+					_cur_lexeme = lex_greater;
+				}
+				break;
+
+			case '<':
+				if (*(cur+1) == '=')
+				{
+					cur += 2;
+					_cur_lexeme = lex_less_or_equal;
+				}
+				else
+				{
+					cur += 1;
+					_cur_lexeme = lex_less;
+				}
+				break;
+
+			case '!':
+				if (*(cur+1) == '=')
+				{
+					cur += 2;
+					_cur_lexeme = lex_not_equal;
+				}
+				else
+				{
+					_cur_lexeme = lex_none;
+				}
+				break;
+
+			case '=':
+				cur += 1;
+				_cur_lexeme = lex_equal;
+
+				break;
+			
+			case '+':
+				cur += 1;
+				_cur_lexeme = lex_plus;
+
+				break;
+
+			case '-':
+				cur += 1;
+				_cur_lexeme = lex_minus;
+
+				break;
+
+			case '*':
+				cur += 1;
+				_cur_lexeme = lex_multiply;
+
+				break;
+
+			case '|':
+				cur += 1;
+				_cur_lexeme = lex_union;
+
+				break;
+			
+			case '$':
+				cur += 1;
+
+				if (PUGI__IS_CHARTYPEX(*cur, ctx_start_symbol))
+				{
+					_cur_lexeme_contents.begin = cur;
+
+					while (PUGI__IS_CHARTYPEX(*cur, ctx_symbol)) cur++;
+
+					if (cur[0] == ':' && PUGI__IS_CHARTYPEX(cur[1], ctx_symbol)) // qname
+					{
+						cur++; // :
+
+						while (PUGI__IS_CHARTYPEX(*cur, ctx_symbol)) cur++;
+					}
+
+					_cur_lexeme_contents.end = cur;
+				
+					_cur_lexeme = lex_var_ref;
+				}
+				else
+				{
+					_cur_lexeme = lex_none;
+				}
+
+				break;
+
+			case '(':
+				cur += 1;
+				_cur_lexeme = lex_open_brace;
+
+				break;
+
+			case ')':
+				cur += 1;
+				_cur_lexeme = lex_close_brace;
+
+				break;
+			
+			case '[':
+				cur += 1;
+				_cur_lexeme = lex_open_square_brace;
+
+				break;
+
+			case ']':
+				cur += 1;
+				_cur_lexeme = lex_close_square_brace;
+
+				break;
+
+			case ',':
+				cur += 1;
+				_cur_lexeme = lex_comma;
+
+				break;
+
+			case '/':
+				if (*(cur+1) == '/')
+				{
+					cur += 2;
+					_cur_lexeme = lex_double_slash;
+				}
+				else
+				{
+					cur += 1;
+					_cur_lexeme = lex_slash;
+				}
+				break;
+		
+			case '.':
+				if (*(cur+1) == '.')
+				{
+					cur += 2;
+					_cur_lexeme = lex_double_dot;
+				}
+				else if (PUGI__IS_CHARTYPEX(*(cur+1), ctx_digit))
+				{
+					_cur_lexeme_contents.begin = cur; // .
+
+					++cur;
+
+					while (PUGI__IS_CHARTYPEX(*cur, ctx_digit)) cur++;
+
+					_cur_lexeme_contents.end = cur;
+					
+					_cur_lexeme = lex_number;
+				}
+				else
+				{
+					cur += 1;
+					_cur_lexeme = lex_dot;
+				}
+				break;
+
+			case '@':
+				cur += 1;
+				_cur_lexeme = lex_axis_attribute;
+
+				break;
+
+			case '"':
+			case '\'':
+			{
+				char_t terminator = *cur;
+
+				++cur;
+
+				_cur_lexeme_contents.begin = cur;
+				while (*cur && *cur != terminator) cur++;
+				_cur_lexeme_contents.end = cur;
+				
+				if (!*cur)
+					_cur_lexeme = lex_none;
+				else
+				{
+					cur += 1;
+					_cur_lexeme = lex_quoted_string;
+				}
+
+				break;
+			}
+
+			case ':':
+				if (*(cur+1) == ':')
+				{
+					cur += 2;
+					_cur_lexeme = lex_double_colon;
+				}
+				else
+				{
+					_cur_lexeme = lex_none;
+				}
+				break;
+
+			default:
+				if (PUGI__IS_CHARTYPEX(*cur, ctx_digit))
+				{
+					_cur_lexeme_contents.begin = cur;
+
+					while (PUGI__IS_CHARTYPEX(*cur, ctx_digit)) cur++;
+				
+					if (*cur == '.')
+					{
+						cur++;
+
+						while (PUGI__IS_CHARTYPEX(*cur, ctx_digit)) cur++;
+					}
+
+					_cur_lexeme_contents.end = cur;
+
+					_cur_lexeme = lex_number;
+				}
+				else if (PUGI__IS_CHARTYPEX(*cur, ctx_start_symbol))
+				{
+					_cur_lexeme_contents.begin = cur;
+
+					while (PUGI__IS_CHARTYPEX(*cur, ctx_symbol)) cur++;
+
+					if (cur[0] == ':')
+					{
+						if (cur[1] == '*') // namespace test ncname:*
+						{
+							cur += 2; // :*
+						}
+						else if (PUGI__IS_CHARTYPEX(cur[1], ctx_symbol)) // namespace test qname
+						{
+							cur++; // :
+
+							while (PUGI__IS_CHARTYPEX(*cur, ctx_symbol)) cur++;
+						}
+					}
+
+					_cur_lexeme_contents.end = cur;
+				
+					_cur_lexeme = lex_string;
+				}
+				else
+				{
+					_cur_lexeme = lex_none;
+				}
+			}
+
+			_cur = cur;
+		}
+
+		lexeme_t current() const
+		{
+			return _cur_lexeme;
+		}
+
+		const char_t* current_pos() const
+		{
+			return _cur_lexeme_pos;
+		}
+
+		const xpath_lexer_string& contents() const
+		{
+			assert(_cur_lexeme == lex_var_ref || _cur_lexeme == lex_number || _cur_lexeme == lex_string || _cur_lexeme == lex_quoted_string);
+
+			return _cur_lexeme_contents;
+		}
+	};
+
+	enum ast_type_t
+	{
+		ast_unknown,
+		ast_op_or,						// left or right
+		ast_op_and,						// left and right
+		ast_op_equal,					// left = right
+		ast_op_not_equal,				// left != right
+		ast_op_less,					// left < right
+		ast_op_greater,					// left > right
+		ast_op_less_or_equal,			// left <= right
+		ast_op_greater_or_equal,		// left >= right
+		ast_op_add,						// left + right
+		ast_op_subtract,				// left - right
+		ast_op_multiply,				// left * right
+		ast_op_divide,					// left / right
+		ast_op_mod,						// left % right
+		ast_op_negate,					// left - right
+		ast_op_union,					// left | right
+		ast_predicate,					// apply predicate to set; next points to next predicate
+		ast_filter,						// select * from left where right
+		ast_string_constant,			// string constant
+		ast_number_constant,			// number constant
+		ast_variable,					// variable
+		ast_func_last,					// last()
+		ast_func_position,				// position()
+		ast_func_count,					// count(left)
+		ast_func_id,					// id(left)
+		ast_func_local_name_0,			// local-name()
+		ast_func_local_name_1,			// local-name(left)
+		ast_func_namespace_uri_0,		// namespace-uri()
+		ast_func_namespace_uri_1,		// namespace-uri(left)
+		ast_func_name_0,				// name()
+		ast_func_name_1,				// name(left)
+		ast_func_string_0,				// string()
+		ast_func_string_1,				// string(left)
+		ast_func_concat,				// concat(left, right, siblings)
+		ast_func_starts_with,			// starts_with(left, right)
+		ast_func_contains,				// contains(left, right)
+		ast_func_substring_before,		// substring-before(left, right)
+		ast_func_substring_after,		// substring-after(left, right)
+		ast_func_substring_2,			// substring(left, right)
+		ast_func_substring_3,			// substring(left, right, third)
+		ast_func_string_length_0,		// string-length()
+		ast_func_string_length_1,		// string-length(left)
+		ast_func_normalize_space_0,		// normalize-space()
+		ast_func_normalize_space_1,		// normalize-space(left)
+		ast_func_translate,				// translate(left, right, third)
+		ast_func_boolean,				// boolean(left)
+		ast_func_not,					// not(left)
+		ast_func_true,					// true()
+		ast_func_false,					// false()
+		ast_func_lang,					// lang(left)
+		ast_func_number_0,				// number()
+		ast_func_number_1,				// number(left)
+		ast_func_sum,					// sum(left)
+		ast_func_floor,					// floor(left)
+		ast_func_ceiling,				// ceiling(left)
+		ast_func_round,					// round(left)
+		ast_step,						// process set left with step
+		ast_step_root,					// select root node
+
+		ast_opt_translate_table,		// translate(left, right, third) where right/third are constants
+		ast_opt_compare_attribute		// @name = 'string'
+	};
+
+	enum axis_t
+	{
+		axis_ancestor,
+		axis_ancestor_or_self,
+		axis_attribute,
+		axis_child,
+		axis_descendant,
+		axis_descendant_or_self,
+		axis_following,
+		axis_following_sibling,
+		axis_namespace,
+		axis_parent,
+		axis_preceding,
+		axis_preceding_sibling,
+		axis_self
+	};
+	
+	enum nodetest_t
+	{
+		nodetest_none,
+		nodetest_name,
+		nodetest_type_node,
+		nodetest_type_comment,
+		nodetest_type_pi,
+		nodetest_type_text,
+		nodetest_pi,
+		nodetest_all,
+		nodetest_all_in_namespace
+	};
+
+	enum predicate_t
+	{
+		predicate_default,
+		predicate_posinv,
+		predicate_constant,
+		predicate_constant_one
+	};
+
+	enum nodeset_eval_t
+	{
+		nodeset_eval_all,
+		nodeset_eval_any,
+		nodeset_eval_first
+	};
+
+	template <axis_t N> struct axis_to_type
+	{
+		static const axis_t axis;
+	};
+
+	template <axis_t N> const axis_t axis_to_type<N>::axis = N;
+		
+	class xpath_ast_node
+	{
+	private:
+		// node type
+		char _type;
+		char _rettype;
+
+		// for ast_step
+		char _axis;
+
+		// for ast_step/ast_predicate/ast_filter
+		char _test;
+
+		// tree node structure
+		xpath_ast_node* _left;
+		xpath_ast_node* _right;
+		xpath_ast_node* _next;
+
+		union
+		{
+			// value for ast_string_constant
+			const char_t* string;
+			// value for ast_number_constant
+			double number;
+			// variable for ast_variable
+			xpath_variable* variable;
+			// node test for ast_step (node name/namespace/node type/pi target)
+			const char_t* nodetest;
+			// table for ast_opt_translate_table
+			const unsigned char* table;
+		} _data;
+
+		xpath_ast_node(const xpath_ast_node&);
+		xpath_ast_node& operator=(const xpath_ast_node&);
+
+		template <class Comp> static bool compare_eq(xpath_ast_node* lhs, xpath_ast_node* rhs, const xpath_context& c, const xpath_stack& stack, const Comp& comp)
+		{
+			xpath_value_type lt = lhs->rettype(), rt = rhs->rettype();
+
+			if (lt != xpath_type_node_set && rt != xpath_type_node_set)
+			{
+				if (lt == xpath_type_boolean || rt == xpath_type_boolean)
+					return comp(lhs->eval_boolean(c, stack), rhs->eval_boolean(c, stack));
+				else if (lt == xpath_type_number || rt == xpath_type_number)
+					return comp(lhs->eval_number(c, stack), rhs->eval_number(c, stack));
+				else if (lt == xpath_type_string || rt == xpath_type_string)
+				{
+					xpath_allocator_capture cr(stack.result);
+
+					xpath_string ls = lhs->eval_string(c, stack);
+					xpath_string rs = rhs->eval_string(c, stack);
+
+					return comp(ls, rs);
+				}
+			}
+			else if (lt == xpath_type_node_set && rt == xpath_type_node_set)
+			{
+				xpath_allocator_capture cr(stack.result);
+
+				xpath_node_set_raw ls = lhs->eval_node_set(c, stack, nodeset_eval_all);
+				xpath_node_set_raw rs = rhs->eval_node_set(c, stack, nodeset_eval_all);
+
+				for (const xpath_node* li = ls.begin(); li != ls.end(); ++li)
+					for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri)
+					{
+						xpath_allocator_capture cri(stack.result);
+
+						if (comp(string_value(*li, stack.result), string_value(*ri, stack.result)))
+							return true;
+					}
+
+				return false;
+			}
+			else
+			{
+				if (lt == xpath_type_node_set)
+				{
+					swap(lhs, rhs);
+					swap(lt, rt);
+				}
+
+				if (lt == xpath_type_boolean)
+					return comp(lhs->eval_boolean(c, stack), rhs->eval_boolean(c, stack));
+				else if (lt == xpath_type_number)
+				{
+					xpath_allocator_capture cr(stack.result);
+
+					double l = lhs->eval_number(c, stack);
+					xpath_node_set_raw rs = rhs->eval_node_set(c, stack, nodeset_eval_all);
+
+					for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri)
+					{
+						xpath_allocator_capture cri(stack.result);
+
+						if (comp(l, convert_string_to_number(string_value(*ri, stack.result).c_str())))
+							return true;
+					}
+
+					return false;
+				}
+				else if (lt == xpath_type_string)
+				{
+					xpath_allocator_capture cr(stack.result);
+
+					xpath_string l = lhs->eval_string(c, stack);
+					xpath_node_set_raw rs = rhs->eval_node_set(c, stack, nodeset_eval_all);
+
+					for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri)
+					{
+						xpath_allocator_capture cri(stack.result);
+
+						if (comp(l, string_value(*ri, stack.result)))
+							return true;
+					}
+
+					return false;
+				}
+			}
+
+			assert(!"Wrong types");
+			return false;
+		}
+
+		static bool eval_once(xpath_node_set::type_t type, nodeset_eval_t eval)
+		{
+			return type == xpath_node_set::type_sorted ? eval != nodeset_eval_all : eval == nodeset_eval_any;
+		}
+
+		template <class Comp> static bool compare_rel(xpath_ast_node* lhs, xpath_ast_node* rhs, const xpath_context& c, const xpath_stack& stack, const Comp& comp)
+		{
+			xpath_value_type lt = lhs->rettype(), rt = rhs->rettype();
+
+			if (lt != xpath_type_node_set && rt != xpath_type_node_set)
+				return comp(lhs->eval_number(c, stack), rhs->eval_number(c, stack));
+			else if (lt == xpath_type_node_set && rt == xpath_type_node_set)
+			{
+				xpath_allocator_capture cr(stack.result);
+
+				xpath_node_set_raw ls = lhs->eval_node_set(c, stack, nodeset_eval_all);
+				xpath_node_set_raw rs = rhs->eval_node_set(c, stack, nodeset_eval_all);
+
+				for (const xpath_node* li = ls.begin(); li != ls.end(); ++li)
+				{
+					xpath_allocator_capture cri(stack.result);
+
+					double l = convert_string_to_number(string_value(*li, stack.result).c_str());
+
+					for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri)
+					{
+						xpath_allocator_capture crii(stack.result);
+
+						if (comp(l, convert_string_to_number(string_value(*ri, stack.result).c_str())))
+							return true;
+					}
+				}
+
+				return false;
+			}
+			else if (lt != xpath_type_node_set && rt == xpath_type_node_set)
+			{
+				xpath_allocator_capture cr(stack.result);
+
+				double l = lhs->eval_number(c, stack);
+				xpath_node_set_raw rs = rhs->eval_node_set(c, stack, nodeset_eval_all);
+
+				for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri)
+				{
+					xpath_allocator_capture cri(stack.result);
+
+					if (comp(l, convert_string_to_number(string_value(*ri, stack.result).c_str())))
+						return true;
+				}
+
+				return false;
+			}
+			else if (lt == xpath_type_node_set && rt != xpath_type_node_set)
+			{
+				xpath_allocator_capture cr(stack.result);
+
+				xpath_node_set_raw ls = lhs->eval_node_set(c, stack, nodeset_eval_all);
+				double r = rhs->eval_number(c, stack);
+
+				for (const xpath_node* li = ls.begin(); li != ls.end(); ++li)
+				{
+					xpath_allocator_capture cri(stack.result);
+
+					if (comp(convert_string_to_number(string_value(*li, stack.result).c_str()), r))
+						return true;
+				}
+
+				return false;
+			}
+			else
+			{
+				assert(!"Wrong types");
+				return false;
+			}
+		}
+
+		static void apply_predicate_boolean(xpath_node_set_raw& ns, size_t first, xpath_ast_node* expr, const xpath_stack& stack, bool once)
+		{
+			assert(ns.size() >= first);
+			assert(expr->rettype() != xpath_type_number);
+
+			size_t i = 1;
+			size_t size = ns.size() - first;
+
+			xpath_node* last = ns.begin() + first;
+
+			// remove_if... or well, sort of
+			for (xpath_node* it = last; it != ns.end(); ++it, ++i)
+			{
+				xpath_context c(*it, i, size);
+
+				if (expr->eval_boolean(c, stack))
+				{
+					*last++ = *it;
+
+					if (once) break;
+				}
+			}
+
+			ns.truncate(last);
+		}
+
+		static void apply_predicate_number(xpath_node_set_raw& ns, size_t first, xpath_ast_node* expr, const xpath_stack& stack, bool once)
+		{
+			assert(ns.size() >= first);
+			assert(expr->rettype() == xpath_type_number);
+
+			size_t i = 1;
+			size_t size = ns.size() - first;
+
+			xpath_node* last = ns.begin() + first;
+
+			// remove_if... or well, sort of
+			for (xpath_node* it = last; it != ns.end(); ++it, ++i)
+			{
+				xpath_context c(*it, i, size);
+
+				if (expr->eval_number(c, stack) == i)
+				{
+					*last++ = *it;
+
+					if (once) break;
+				}
+			}
+
+			ns.truncate(last);
+		}
+
+		static void apply_predicate_number_const(xpath_node_set_raw& ns, size_t first, xpath_ast_node* expr, const xpath_stack& stack)
+		{
+			assert(ns.size() >= first);
+			assert(expr->rettype() == xpath_type_number);
+
+			size_t size = ns.size() - first;
+
+			xpath_node* last = ns.begin() + first;
+
+			xpath_context c(xpath_node(), 1, size);
+
+			double er = expr->eval_number(c, stack);
+
+			if (er >= 1.0 && er <= size)
+			{
+				size_t eri = static_cast<size_t>(er);
+
+				if (er == eri)
+				{
+					xpath_node r = last[eri - 1];
+
+					*last++ = r;
+				}
+			}
+
+			ns.truncate(last);
+		}
+
+		void apply_predicate(xpath_node_set_raw& ns, size_t first, const xpath_stack& stack, bool once)
+		{
+			if (ns.size() == first) return;
+
+			assert(_type == ast_filter || _type == ast_predicate);
+
+			if (_test == predicate_constant || _test == predicate_constant_one)
+				apply_predicate_number_const(ns, first, _right, stack);
+			else if (_right->rettype() == xpath_type_number)
+				apply_predicate_number(ns, first, _right, stack, once);
+			else
+				apply_predicate_boolean(ns, first, _right, stack, once);
+		}
+
+		void apply_predicates(xpath_node_set_raw& ns, size_t first, const xpath_stack& stack, nodeset_eval_t eval)
+		{
+			if (ns.size() == first) return;
+
+			bool last_once = eval_once(ns.type(), eval);
+
+			for (xpath_ast_node* pred = _right; pred; pred = pred->_next)
+				pred->apply_predicate(ns, first, stack, !pred->_next && last_once);
+		}
+
+		bool step_push(xpath_node_set_raw& ns, xml_attribute_struct* a, xml_node_struct* parent, xpath_allocator* alloc)
+		{
+			assert(a);
+
+			const char_t* name = a->name ? a->name + 0 : PUGIXML_TEXT("");
+
+			switch (_test)
+			{
+			case nodetest_name:
+				if (strequal(name, _data.nodetest) && is_xpath_attribute(name))
+				{
+					ns.push_back(xpath_node(xml_attribute(a), xml_node(parent)), alloc);
+					return true;
+				}
+				break;
+				
+			case nodetest_type_node:
+			case nodetest_all:
+				if (is_xpath_attribute(name))
+				{
+					ns.push_back(xpath_node(xml_attribute(a), xml_node(parent)), alloc);
+					return true;
+				}
+				break;
+				
+			case nodetest_all_in_namespace:
+				if (starts_with(name, _data.nodetest) && is_xpath_attribute(name))
+				{
+					ns.push_back(xpath_node(xml_attribute(a), xml_node(parent)), alloc);
+					return true;
+				}
+				break;
+			
+			default:
+				;
+			}
+
+			return false;
+		}
+		
+		bool step_push(xpath_node_set_raw& ns, xml_node_struct* n, xpath_allocator* alloc)
+		{
+			assert(n);
+
+			xml_node_type type = PUGI__NODETYPE(n);
+
+			switch (_test)
+			{
+			case nodetest_name:
+				if (type == node_element && n->name && strequal(n->name, _data.nodetest))
+				{
+					ns.push_back(xml_node(n), alloc);
+					return true;
+				}
+				break;
+				
+			case nodetest_type_node:
+				ns.push_back(xml_node(n), alloc);
+				return true;
+				
+			case nodetest_type_comment:
+				if (type == node_comment)
+				{
+					ns.push_back(xml_node(n), alloc);
+					return true;
+				}
+				break;
+				
+			case nodetest_type_text:
+				if (type == node_pcdata || type == node_cdata)
+				{
+					ns.push_back(xml_node(n), alloc);
+					return true;
+				}
+				break;
+				
+			case nodetest_type_pi:
+				if (type == node_pi)
+				{
+					ns.push_back(xml_node(n), alloc);
+					return true;
+				}
+				break;
+									
+			case nodetest_pi:
+				if (type == node_pi && n->name && strequal(n->name, _data.nodetest))
+				{
+					ns.push_back(xml_node(n), alloc);
+					return true;
+				}
+				break;
+				
+			case nodetest_all:
+				if (type == node_element)
+				{
+					ns.push_back(xml_node(n), alloc);
+					return true;
+				}
+				break;
+				
+			case nodetest_all_in_namespace:
+				if (type == node_element && n->name && starts_with(n->name, _data.nodetest))
+				{
+					ns.push_back(xml_node(n), alloc);
+					return true;
+				}
+				break;
+
+			default:
+				assert(!"Unknown axis");
+			}
+
+			return false;
+		}
+
+		template <class T> void step_fill(xpath_node_set_raw& ns, xml_node_struct* n, xpath_allocator* alloc, bool once, T)
+		{
+			const axis_t axis = T::axis;
+
+			switch (axis)
+			{
+			case axis_attribute:
+			{
+				for (xml_attribute_struct* a = n->first_attribute; a; a = a->next_attribute)
+					if (step_push(ns, a, n, alloc) & once)
+						return;
+				
+				break;
+			}
+			
+			case axis_child:
+			{
+				for (xml_node_struct* c = n->first_child; c; c = c->next_sibling)
+					if (step_push(ns, c, alloc) & once)
+						return;
+					
+				break;
+			}
+			
+			case axis_descendant:
+			case axis_descendant_or_self:
+			{
+				if (axis == axis_descendant_or_self)
+					if (step_push(ns, n, alloc) & once)
+						return;
+					
+				xml_node_struct* cur = n->first_child;
+				
+				while (cur)
+				{
+					if (step_push(ns, cur, alloc) & once)
+						return;
+					
+					if (cur->first_child)
+						cur = cur->first_child;
+					else
+					{
+						while (!cur->next_sibling)
+						{
+							cur = cur->parent;
+
+							if (cur == n) return;
+						}
+					
+						cur = cur->next_sibling;
+					}
+				}
+				
+				break;
+			}
+			
+			case axis_following_sibling:
+			{
+				for (xml_node_struct* c = n->next_sibling; c; c = c->next_sibling)
+					if (step_push(ns, c, alloc) & once)
+						return;
+				
+				break;
+			}
+			
+			case axis_preceding_sibling:
+			{
+				for (xml_node_struct* c = n->prev_sibling_c; c->next_sibling; c = c->prev_sibling_c)
+					if (step_push(ns, c, alloc) & once)
+						return;
+				
+				break;
+			}
+			
+			case axis_following:
+			{
+				xml_node_struct* cur = n;
+
+				// exit from this node so that we don't include descendants
+				while (!cur->next_sibling)
+				{
+					cur = cur->parent;
+
+					if (!cur) return;
+				}
+
+				cur = cur->next_sibling;
+
+				while (cur)
+				{
+					if (step_push(ns, cur, alloc) & once)
+						return;
+
+					if (cur->first_child)
+						cur = cur->first_child;
+					else
+					{
+						while (!cur->next_sibling)
+						{
+							cur = cur->parent;
+
+							if (!cur) return;
+						}
+
+						cur = cur->next_sibling;
+					}
+				}
+
+				break;
+			}
+
+			case axis_preceding:
+			{
+				xml_node_struct* cur = n;
+
+				// exit from this node so that we don't include descendants
+				while (!cur->prev_sibling_c->next_sibling)
+				{
+					cur = cur->parent;
+
+					if (!cur) return;
+				}
+
+				cur = cur->prev_sibling_c;
+
+				while (cur)
+				{
+					if (cur->first_child)
+						cur = cur->first_child->prev_sibling_c;
+					else
+					{
+						// leaf node, can't be ancestor
+						if (step_push(ns, cur, alloc) & once)
+							return;
+
+						while (!cur->prev_sibling_c->next_sibling)
+						{
+							cur = cur->parent;
+
+							if (!cur) return;
+
+							if (!node_is_ancestor(cur, n))
+								if (step_push(ns, cur, alloc) & once)
+									return;
+						}
+
+						cur = cur->prev_sibling_c;
+					}
+				}
+
+				break;
+			}
+			
+			case axis_ancestor:
+			case axis_ancestor_or_self:
+			{
+				if (axis == axis_ancestor_or_self)
+					if (step_push(ns, n, alloc) & once)
+						return;
+
+				xml_node_struct* cur = n->parent;
+				
+				while (cur)
+				{
+					if (step_push(ns, cur, alloc) & once)
+						return;
+					
+					cur = cur->parent;
+				}
+				
+				break;
+			}
+
+			case axis_self:
+			{
+				step_push(ns, n, alloc);
+
+				break;
+			}
+
+			case axis_parent:
+			{
+				if (n->parent)
+					step_push(ns, n->parent, alloc);
+
+				break;
+			}
+				
+			default:
+				assert(!"Unimplemented axis");
+			}
+		}
+		
+		template <class T> void step_fill(xpath_node_set_raw& ns, xml_attribute_struct* a, xml_node_struct* p, xpath_allocator* alloc, bool once, T v)
+		{
+			const axis_t axis = T::axis;
+
+			switch (axis)
+			{
+			case axis_ancestor:
+			case axis_ancestor_or_self:
+			{
+				if (axis == axis_ancestor_or_self && _test == nodetest_type_node) // reject attributes based on principal node type test
+					if (step_push(ns, a, p, alloc) & once)
+						return;
+
+				xml_node_struct* cur = p;
+				
+				while (cur)
+				{
+					if (step_push(ns, cur, alloc) & once)
+						return;
+					
+					cur = cur->parent;
+				}
+				
+				break;
+			}
+
+			case axis_descendant_or_self:
+			case axis_self:
+			{
+				if (_test == nodetest_type_node) // reject attributes based on principal node type test
+					step_push(ns, a, p, alloc);
+
+				break;
+			}
+
+			case axis_following:
+			{
+				xml_node_struct* cur = p;
+				
+				while (cur)
+				{
+					if (cur->first_child)
+						cur = cur->first_child;
+					else
+					{
+						while (!cur->next_sibling)
+						{
+							cur = cur->parent;
+
+							if (!cur) return;
+						}
+
+						cur = cur->next_sibling;
+					}
+
+					if (step_push(ns, cur, alloc) & once)
+						return;
+				}
+
+				break;
+			}
+
+			case axis_parent:
+			{
+				step_push(ns, p, alloc);
+
+				break;
+			}
+
+			case axis_preceding:
+			{
+				// preceding:: axis does not include attribute nodes and attribute ancestors (they are the same as parent's ancestors), so we can reuse node preceding
+				step_fill(ns, p, alloc, once, v);
+				break;
+			}
+			
+			default:
+				assert(!"Unimplemented axis");
+			}
+		}
+
+		template <class T> void step_fill(xpath_node_set_raw& ns, const xpath_node& xn, xpath_allocator* alloc, bool once, T v)
+		{
+			const axis_t axis = T::axis;
+			const bool axis_has_attributes = (axis == axis_ancestor || axis == axis_ancestor_or_self || axis == axis_descendant_or_self || axis == axis_following || axis == axis_parent || axis == axis_preceding || axis == axis_self);
+
+			if (xn.node())
+				step_fill(ns, xn.node().internal_object(), alloc, once, v);
+			else if (axis_has_attributes && xn.attribute() && xn.parent())
+				step_fill(ns, xn.attribute().internal_object(), xn.parent().internal_object(), alloc, once, v);
+		}
+
+		template <class T> xpath_node_set_raw step_do(const xpath_context& c, const xpath_stack& stack, nodeset_eval_t eval, T v)
+		{
+			const axis_t axis = T::axis;
+			const bool axis_reverse = (axis == axis_ancestor || axis == axis_ancestor_or_self || axis == axis_preceding || axis == axis_preceding_sibling);
+			const xpath_node_set::type_t axis_type = axis_reverse ? xpath_node_set::type_sorted_reverse : xpath_node_set::type_sorted;
+
+			bool once =
+				(axis == axis_attribute && _test == nodetest_name) ||
+				(!_right && eval_once(axis_type, eval)) ||
+				(_right && !_right->_next && _right->_test == predicate_constant_one);
+
+			xpath_node_set_raw ns;
+			ns.set_type(axis_type);
+
+			if (_left)
+			{
+				xpath_node_set_raw s = _left->eval_node_set(c, stack, nodeset_eval_all);
+
+				// self axis preserves the original order
+				if (axis == axis_self) ns.set_type(s.type());
+
+				for (const xpath_node* it = s.begin(); it != s.end(); ++it)
+				{
+					size_t size = ns.size();
+
+					// in general, all axes generate elements in a particular order, but there is no order guarantee if axis is applied to two nodes
+					if (axis != axis_self && size != 0) ns.set_type(xpath_node_set::type_unsorted);
+					
+					step_fill(ns, *it, stack.result, once, v);
+					if (_right) apply_predicates(ns, size, stack, eval);
+				}
+			}
+			else
+			{
+				step_fill(ns, c.n, stack.result, once, v);
+				if (_right) apply_predicates(ns, 0, stack, eval);
+			}
+
+			// child, attribute and self axes always generate unique set of nodes
+			// for other axis, if the set stayed sorted, it stayed unique because the traversal algorithms do not visit the same node twice
+			if (axis != axis_child && axis != axis_attribute && axis != axis_self && ns.type() == xpath_node_set::type_unsorted)
+				ns.remove_duplicates();
+
+			return ns;
+		}
+		
+	public:
+		xpath_ast_node(ast_type_t type, xpath_value_type rettype_, const char_t* value):
+			_type(static_cast<char>(type)), _rettype(static_cast<char>(rettype_)), _axis(0), _test(0), _left(0), _right(0), _next(0)
+		{
+			assert(type == ast_string_constant);
+			_data.string = value;
+		}
+
+		xpath_ast_node(ast_type_t type, xpath_value_type rettype_, double value):
+			_type(static_cast<char>(type)), _rettype(static_cast<char>(rettype_)), _axis(0), _test(0), _left(0), _right(0), _next(0)
+		{
+			assert(type == ast_number_constant);
+			_data.number = value;
+		}
+		
+		xpath_ast_node(ast_type_t type, xpath_value_type rettype_, xpath_variable* value):
+			_type(static_cast<char>(type)), _rettype(static_cast<char>(rettype_)), _axis(0), _test(0), _left(0), _right(0), _next(0)
+		{
+			assert(type == ast_variable);
+			_data.variable = value;
+		}
+		
+		xpath_ast_node(ast_type_t type, xpath_value_type rettype_, xpath_ast_node* left = 0, xpath_ast_node* right = 0):
+			_type(static_cast<char>(type)), _rettype(static_cast<char>(rettype_)), _axis(0), _test(0), _left(left), _right(right), _next(0)
+		{
+		}
+
+		xpath_ast_node(ast_type_t type, xpath_ast_node* left, axis_t axis, nodetest_t test, const char_t* contents):
+			_type(static_cast<char>(type)), _rettype(xpath_type_node_set), _axis(static_cast<char>(axis)), _test(static_cast<char>(test)), _left(left), _right(0), _next(0)
+		{
+			assert(type == ast_step);
+			_data.nodetest = contents;
+		}
+
+		xpath_ast_node(ast_type_t type, xpath_ast_node* left, xpath_ast_node* right, predicate_t test):
+			_type(static_cast<char>(type)), _rettype(xpath_type_node_set), _axis(0), _test(static_cast<char>(test)), _left(left), _right(right), _next(0)
+		{
+			assert(type == ast_filter || type == ast_predicate);
+		}
+
+		void set_next(xpath_ast_node* value)
+		{
+			_next = value;
+		}
+
+		void set_right(xpath_ast_node* value)
+		{
+			_right = value;
+		}
+
+		bool eval_boolean(const xpath_context& c, const xpath_stack& stack)
+		{
+			switch (_type)
+			{
+			case ast_op_or:
+				return _left->eval_boolean(c, stack) || _right->eval_boolean(c, stack);
+				
+			case ast_op_and:
+				return _left->eval_boolean(c, stack) && _right->eval_boolean(c, stack);
+				
+			case ast_op_equal:
+				return compare_eq(_left, _right, c, stack, equal_to());
+
+			case ast_op_not_equal:
+				return compare_eq(_left, _right, c, stack, not_equal_to());
+	
+			case ast_op_less:
+				return compare_rel(_left, _right, c, stack, less());
+			
+			case ast_op_greater:
+				return compare_rel(_right, _left, c, stack, less());
+
+			case ast_op_less_or_equal:
+				return compare_rel(_left, _right, c, stack, less_equal());
+			
+			case ast_op_greater_or_equal:
+				return compare_rel(_right, _left, c, stack, less_equal());
+
+			case ast_func_starts_with:
+			{
+				xpath_allocator_capture cr(stack.result);
+
+				xpath_string lr = _left->eval_string(c, stack);
+				xpath_string rr = _right->eval_string(c, stack);
+
+				return starts_with(lr.c_str(), rr.c_str());
+			}
+
+			case ast_func_contains:
+			{
+				xpath_allocator_capture cr(stack.result);
+
+				xpath_string lr = _left->eval_string(c, stack);
+				xpath_string rr = _right->eval_string(c, stack);
+
+				return find_substring(lr.c_str(), rr.c_str()) != 0;
+			}
+
+			case ast_func_boolean:
+				return _left->eval_boolean(c, stack);
+				
+			case ast_func_not:
+				return !_left->eval_boolean(c, stack);
+				
+			case ast_func_true:
+				return true;
+				
+			case ast_func_false:
+				return false;
+
+			case ast_func_lang:
+			{
+				if (c.n.attribute()) return false;
+				
+				xpath_allocator_capture cr(stack.result);
+
+				xpath_string lang = _left->eval_string(c, stack);
+				
+				for (xml_node n = c.n.node(); n; n = n.parent())
+				{
+					xml_attribute a = n.attribute(PUGIXML_TEXT("xml:lang"));
+					
+					if (a)
+					{
+						const char_t* value = a.value();
+						
+						// strnicmp / strncasecmp is not portable
+						for (const char_t* lit = lang.c_str(); *lit; ++lit)
+						{
+							if (tolower_ascii(*lit) != tolower_ascii(*value)) return false;
+							++value;
+						}
+						
+						return *value == 0 || *value == '-';
+					}
+				}
+				
+				return false;
+			}
+
+			case ast_opt_compare_attribute:
+			{
+				const char_t* value = (_right->_type == ast_string_constant) ? _right->_data.string : _right->_data.variable->get_string();
+
+				xml_attribute attr = c.n.node().attribute(_left->_data.nodetest);
+
+				return attr && strequal(attr.value(), value) && is_xpath_attribute(attr.name());
+			}
+
+			case ast_variable:
+			{
+				assert(_rettype == _data.variable->type());
+
+				if (_rettype == xpath_type_boolean)
+					return _data.variable->get_boolean();
+
+				// fallthrough to type conversion
+			}
+
+			default:
+			{
+				switch (_rettype)
+				{
+				case xpath_type_number:
+					return convert_number_to_boolean(eval_number(c, stack));
+					
+				case xpath_type_string:
+				{
+					xpath_allocator_capture cr(stack.result);
+
+					return !eval_string(c, stack).empty();
+				}
+					
+				case xpath_type_node_set:				
+				{
+					xpath_allocator_capture cr(stack.result);
+
+					return !eval_node_set(c, stack, nodeset_eval_any).empty();
+				}
+
+				default:
+					assert(!"Wrong expression for return type boolean");
+					return false;
+				}
+			}
+			}
+		}
+
+		double eval_number(const xpath_context& c, const xpath_stack& stack)
+		{
+			switch (_type)
+			{
+			case ast_op_add:
+				return _left->eval_number(c, stack) + _right->eval_number(c, stack);
+				
+			case ast_op_subtract:
+				return _left->eval_number(c, stack) - _right->eval_number(c, stack);
+
+			case ast_op_multiply:
+				return _left->eval_number(c, stack) * _right->eval_number(c, stack);
+
+			case ast_op_divide:
+				return _left->eval_number(c, stack) / _right->eval_number(c, stack);
+
+			case ast_op_mod:
+				return fmod(_left->eval_number(c, stack), _right->eval_number(c, stack));
+
+			case ast_op_negate:
+				return -_left->eval_number(c, stack);
+
+			case ast_number_constant:
+				return _data.number;
+
+			case ast_func_last:
+				return static_cast<double>(c.size);
+			
+			case ast_func_position:
+				return static_cast<double>(c.position);
+
+			case ast_func_count:
+			{
+				xpath_allocator_capture cr(stack.result);
+
+				return static_cast<double>(_left->eval_node_set(c, stack, nodeset_eval_all).size());
+			}
+			
+			case ast_func_string_length_0:
+			{
+				xpath_allocator_capture cr(stack.result);
+
+				return static_cast<double>(string_value(c.n, stack.result).length());
+			}
+			
+			case ast_func_string_length_1:
+			{
+				xpath_allocator_capture cr(stack.result);
+
+				return static_cast<double>(_left->eval_string(c, stack).length());
+			}
+			
+			case ast_func_number_0:
+			{
+				xpath_allocator_capture cr(stack.result);
+
+				return convert_string_to_number(string_value(c.n, stack.result).c_str());
+			}
+			
+			case ast_func_number_1:
+				return _left->eval_number(c, stack);
+
+			case ast_func_sum:
+			{
+				xpath_allocator_capture cr(stack.result);
+
+				double r = 0;
+				
+				xpath_node_set_raw ns = _left->eval_node_set(c, stack, nodeset_eval_all);
+				
+				for (const xpath_node* it = ns.begin(); it != ns.end(); ++it)
+				{
+					xpath_allocator_capture cri(stack.result);
+
+					r += convert_string_to_number(string_value(*it, stack.result).c_str());
+				}
+			
+				return r;
+			}
+
+			case ast_func_floor:
+			{
+				double r = _left->eval_number(c, stack);
+				
+				return r == r ? floor(r) : r;
+			}
+
+			case ast_func_ceiling:
+			{
+				double r = _left->eval_number(c, stack);
+				
+				return r == r ? ceil(r) : r;
+			}
+
+			case ast_func_round:
+				return round_nearest_nzero(_left->eval_number(c, stack));
+			
+			case ast_variable:
+			{
+				assert(_rettype == _data.variable->type());
+
+				if (_rettype == xpath_type_number)
+					return _data.variable->get_number();
+
+				// fallthrough to type conversion
+			}
+
+			default:
+			{
+				switch (_rettype)
+				{
+				case xpath_type_boolean:
+					return eval_boolean(c, stack) ? 1 : 0;
+					
+				case xpath_type_string:
+				{
+					xpath_allocator_capture cr(stack.result);
+
+					return convert_string_to_number(eval_string(c, stack).c_str());
+				}
+					
+				case xpath_type_node_set:
+				{
+					xpath_allocator_capture cr(stack.result);
+
+					return convert_string_to_number(eval_string(c, stack).c_str());
+				}
+					
+				default:
+					assert(!"Wrong expression for return type number");
+					return 0;
+				}
+				
+			}
+			}
+		}
+		
+		xpath_string eval_string_concat(const xpath_context& c, const xpath_stack& stack)
+		{
+			assert(_type == ast_func_concat);
+
+			xpath_allocator_capture ct(stack.temp);
+
+			// count the string number
+			size_t count = 1;
+			for (xpath_ast_node* nc = _right; nc; nc = nc->_next) count++;
+
+			// gather all strings
+			xpath_string static_buffer[4];
+			xpath_string* buffer = static_buffer;
+
+			// allocate on-heap for large concats
+			if (count > sizeof(static_buffer) / sizeof(static_buffer[0]))
+			{
+				buffer = static_cast<xpath_string*>(stack.temp->allocate(count * sizeof(xpath_string)));
+				assert(buffer);
+			}
+
+			// evaluate all strings to temporary stack
+			xpath_stack swapped_stack = {stack.temp, stack.result};
+
+			buffer[0] = _left->eval_string(c, swapped_stack);
+
+			size_t pos = 1;
+			for (xpath_ast_node* n = _right; n; n = n->_next, ++pos) buffer[pos] = n->eval_string(c, swapped_stack);
+			assert(pos == count);
+
+			// get total length
+			size_t length = 0;
+			for (size_t i = 0; i < count; ++i) length += buffer[i].length();
+
+			// create final string
+			char_t* result = static_cast<char_t*>(stack.result->allocate((length + 1) * sizeof(char_t)));
+			assert(result);
+
+			char_t* ri = result;
+
+			for (size_t j = 0; j < count; ++j)
+				for (const char_t* bi = buffer[j].c_str(); *bi; ++bi)
+					*ri++ = *bi;
+
+			*ri = 0;
+
+			return xpath_string::from_heap_preallocated(result, ri);
+		}
+
+		xpath_string eval_string(const xpath_context& c, const xpath_stack& stack)
+		{
+			switch (_type)
+			{
+			case ast_string_constant:
+				return xpath_string::from_const(_data.string);
+			
+			case ast_func_local_name_0:
+			{
+				xpath_node na = c.n;
+				
+				return xpath_string::from_const(local_name(na));
+			}
+
+			case ast_func_local_name_1:
+			{
+				xpath_allocator_capture cr(stack.result);
+
+				xpath_node_set_raw ns = _left->eval_node_set(c, stack, nodeset_eval_first);
+				xpath_node na = ns.first();
+				
+				return xpath_string::from_const(local_name(na));
+			}
+
+			case ast_func_name_0:
+			{
+				xpath_node na = c.n;
+				
+				return xpath_string::from_const(qualified_name(na));
+			}
+
+			case ast_func_name_1:
+			{
+				xpath_allocator_capture cr(stack.result);
+
+				xpath_node_set_raw ns = _left->eval_node_set(c, stack, nodeset_eval_first);
+				xpath_node na = ns.first();
+				
+				return xpath_string::from_const(qualified_name(na));
+			}
+
+			case ast_func_namespace_uri_0:
+			{
+				xpath_node na = c.n;
+				
+				return xpath_string::from_const(namespace_uri(na));
+			}
+
+			case ast_func_namespace_uri_1:
+			{
+				xpath_allocator_capture cr(stack.result);
+
+				xpath_node_set_raw ns = _left->eval_node_set(c, stack, nodeset_eval_first);
+				xpath_node na = ns.first();
+				
+				return xpath_string::from_const(namespace_uri(na));
+			}
+
+			case ast_func_string_0:
+				return string_value(c.n, stack.result);
+
+			case ast_func_string_1:
+				return _left->eval_string(c, stack);
+
+			case ast_func_concat:
+				return eval_string_concat(c, stack);
+
+			case ast_func_substring_before:
+			{
+				xpath_allocator_capture cr(stack.temp);
+
+				xpath_stack swapped_stack = {stack.temp, stack.result};
+
+				xpath_string s = _left->eval_string(c, swapped_stack);
+				xpath_string p = _right->eval_string(c, swapped_stack);
+
+				const char_t* pos = find_substring(s.c_str(), p.c_str());
+				
+				return pos ? xpath_string::from_heap(s.c_str(), pos, stack.result) : xpath_string();
+			}
+			
+			case ast_func_substring_after:
+			{
+				xpath_allocator_capture cr(stack.temp);
+
+				xpath_stack swapped_stack = {stack.temp, stack.result};
+
+				xpath_string s = _left->eval_string(c, swapped_stack);
+				xpath_string p = _right->eval_string(c, swapped_stack);
+				
+				const char_t* pos = find_substring(s.c_str(), p.c_str());
+				if (!pos) return xpath_string();
+
+				const char_t* rbegin = pos + p.length();
+				const char_t* rend = s.c_str() + s.length();
+
+				return s.uses_heap() ? xpath_string::from_heap(rbegin, rend, stack.result) : xpath_string::from_const(rbegin);
+			}
+
+			case ast_func_substring_2:
+			{
+				xpath_allocator_capture cr(stack.temp);
+
+				xpath_stack swapped_stack = {stack.temp, stack.result};
+
+				xpath_string s = _left->eval_string(c, swapped_stack);
+				size_t s_length = s.length();
+
+				double first = round_nearest(_right->eval_number(c, stack));
+				
+				if (is_nan(first)) return xpath_string(); // NaN
+				else if (first >= s_length + 1) return xpath_string();
+				
+				size_t pos = first < 1 ? 1 : static_cast<size_t>(first);
+				assert(1 <= pos && pos <= s_length + 1);
+
+				const char_t* rbegin = s.c_str() + (pos - 1);
+				const char_t* rend = s.c_str() + s.length();
+				
+				return s.uses_heap() ? xpath_string::from_heap(rbegin, rend, stack.result) : xpath_string::from_const(rbegin);
+			}
+			
+			case ast_func_substring_3:
+			{
+				xpath_allocator_capture cr(stack.temp);
+
+				xpath_stack swapped_stack = {stack.temp, stack.result};
+
+				xpath_string s = _left->eval_string(c, swapped_stack);
+				size_t s_length = s.length();
+
+				double first = round_nearest(_right->eval_number(c, stack));
+				double last = first + round_nearest(_right->_next->eval_number(c, stack));
+				
+				if (is_nan(first) || is_nan(last)) return xpath_string();
+				else if (first >= s_length + 1) return xpath_string();
+				else if (first >= last) return xpath_string();
+				else if (last < 1) return xpath_string();
+				
+				size_t pos = first < 1 ? 1 : static_cast<size_t>(first);
+				size_t end = last >= s_length + 1 ? s_length + 1 : static_cast<size_t>(last);
+
+				assert(1 <= pos && pos <= end && end <= s_length + 1);
+				const char_t* rbegin = s.c_str() + (pos - 1);
+				const char_t* rend = s.c_str() + (end - 1);
+
+				return (end == s_length + 1 && !s.uses_heap()) ? xpath_string::from_const(rbegin) : xpath_string::from_heap(rbegin, rend, stack.result);
+			}
+
+			case ast_func_normalize_space_0:
+			{
+				xpath_string s = string_value(c.n, stack.result);
+
+				char_t* begin = s.data(stack.result);
+				char_t* end = normalize_space(begin);
+
+				return xpath_string::from_heap_preallocated(begin, end);
+			}
+
+			case ast_func_normalize_space_1:
+			{
+				xpath_string s = _left->eval_string(c, stack);
+
+				char_t* begin = s.data(stack.result);
+				char_t* end = normalize_space(begin);
+			
+				return xpath_string::from_heap_preallocated(begin, end);
+			}
+
+			case ast_func_translate:
+			{
+				xpath_allocator_capture cr(stack.temp);
+
+				xpath_stack swapped_stack = {stack.temp, stack.result};
+
+				xpath_string s = _left->eval_string(c, stack);
+				xpath_string from = _right->eval_string(c, swapped_stack);
+				xpath_string to = _right->_next->eval_string(c, swapped_stack);
+
+				char_t* begin = s.data(stack.result);
+				char_t* end = translate(begin, from.c_str(), to.c_str(), to.length());
+
+				return xpath_string::from_heap_preallocated(begin, end);
+			}
+
+			case ast_opt_translate_table:
+			{
+				xpath_string s = _left->eval_string(c, stack);
+
+				char_t* begin = s.data(stack.result);
+				char_t* end = translate_table(begin, _data.table);
+
+				return xpath_string::from_heap_preallocated(begin, end);
+			}
+
+			case ast_variable:
+			{
+				assert(_rettype == _data.variable->type());
+
+				if (_rettype == xpath_type_string)
+					return xpath_string::from_const(_data.variable->get_string());
+
+				// fallthrough to type conversion
+			}
+
+			default:
+			{
+				switch (_rettype)
+				{
+				case xpath_type_boolean:
+					return xpath_string::from_const(eval_boolean(c, stack) ? PUGIXML_TEXT("true") : PUGIXML_TEXT("false"));
+					
+				case xpath_type_number:
+					return convert_number_to_string(eval_number(c, stack), stack.result);
+					
+				case xpath_type_node_set:
+				{
+					xpath_allocator_capture cr(stack.temp);
+
+					xpath_stack swapped_stack = {stack.temp, stack.result};
+
+					xpath_node_set_raw ns = eval_node_set(c, swapped_stack, nodeset_eval_first);
+					return ns.empty() ? xpath_string() : string_value(ns.first(), stack.result);
+				}
+				
+				default:
+					assert(!"Wrong expression for return type string");
+					return xpath_string();
+				}
+			}
+			}
+		}
+
+		xpath_node_set_raw eval_node_set(const xpath_context& c, const xpath_stack& stack, nodeset_eval_t eval)
+		{
+			switch (_type)
+			{
+			case ast_op_union:
+			{
+				xpath_allocator_capture cr(stack.temp);
+
+				xpath_stack swapped_stack = {stack.temp, stack.result};
+
+				xpath_node_set_raw ls = _left->eval_node_set(c, swapped_stack, eval);
+				xpath_node_set_raw rs = _right->eval_node_set(c, stack, eval);
+
+				// we can optimize merging two sorted sets, but this is a very rare operation, so don't bother
+				rs.set_type(xpath_node_set::type_unsorted);
+
+				rs.append(ls.begin(), ls.end(), stack.result);
+				rs.remove_duplicates();
+
+				return rs;
+			}
+
+			case ast_filter:
+			{
+				xpath_node_set_raw set = _left->eval_node_set(c, stack, _test == predicate_constant_one ? nodeset_eval_first : nodeset_eval_all);
+
+				// either expression is a number or it contains position() call; sort by document order
+				if (_test != predicate_posinv) set.sort_do();
+
+				bool once = eval_once(set.type(), eval);
+
+				apply_predicate(set, 0, stack, once);
+			
+				return set;
+			}
+			
+			case ast_func_id:
+				return xpath_node_set_raw();
+			
+			case ast_step:
+			{
+				switch (_axis)
+				{
+				case axis_ancestor:
+					return step_do(c, stack, eval, axis_to_type<axis_ancestor>());
+					
+				case axis_ancestor_or_self:
+					return step_do(c, stack, eval, axis_to_type<axis_ancestor_or_self>());
+
+				case axis_attribute:
+					return step_do(c, stack, eval, axis_to_type<axis_attribute>());
+
+				case axis_child:
+					return step_do(c, stack, eval, axis_to_type<axis_child>());
+				
+				case axis_descendant:
+					return step_do(c, stack, eval, axis_to_type<axis_descendant>());
+
+				case axis_descendant_or_self:
+					return step_do(c, stack, eval, axis_to_type<axis_descendant_or_self>());
+
+				case axis_following:
+					return step_do(c, stack, eval, axis_to_type<axis_following>());
+				
+				case axis_following_sibling:
+					return step_do(c, stack, eval, axis_to_type<axis_following_sibling>());
+				
+				case axis_namespace:
+					// namespaced axis is not supported
+					return xpath_node_set_raw();
+				
+				case axis_parent:
+					return step_do(c, stack, eval, axis_to_type<axis_parent>());
+				
+				case axis_preceding:
+					return step_do(c, stack, eval, axis_to_type<axis_preceding>());
+
+				case axis_preceding_sibling:
+					return step_do(c, stack, eval, axis_to_type<axis_preceding_sibling>());
+				
+				case axis_self:
+					return step_do(c, stack, eval, axis_to_type<axis_self>());
+
+				default:
+					assert(!"Unknown axis");
+					return xpath_node_set_raw();
+				}
+			}
+
+			case ast_step_root:
+			{
+				assert(!_right); // root step can't have any predicates
+
+				xpath_node_set_raw ns;
+
+				ns.set_type(xpath_node_set::type_sorted);
+
+				if (c.n.node()) ns.push_back(c.n.node().root(), stack.result);
+				else if (c.n.attribute()) ns.push_back(c.n.parent().root(), stack.result);
+
+				return ns;
+			}
+
+			case ast_variable:
+			{
+				assert(_rettype == _data.variable->type());
+
+				if (_rettype == xpath_type_node_set)
+				{
+					const xpath_node_set& s = _data.variable->get_node_set();
+
+					xpath_node_set_raw ns;
+
+					ns.set_type(s.type());
+					ns.append(s.begin(), s.end(), stack.result);
+
+					return ns;
+				}
+
+				// fallthrough to type conversion
+			}
+
+			default:
+				assert(!"Wrong expression for return type node set");
+				return xpath_node_set_raw();
+			}
+		}
+
+		void optimize(xpath_allocator* alloc)
+		{
+			if (_left) _left->optimize(alloc);
+			if (_right) _right->optimize(alloc);
+			if (_next) _next->optimize(alloc);
+
+			optimize_self(alloc);
+		}
+
+		void optimize_self(xpath_allocator* alloc)
+		{
+			// Rewrite [position()=expr] with [expr]
+			// Note that this step has to go before classification to recognize [position()=1]
+			if ((_type == ast_filter || _type == ast_predicate) &&
+				_right->_type == ast_op_equal && _right->_left->_type == ast_func_position && _right->_right->_rettype == xpath_type_number)
+			{
+				_right = _right->_right;
+			}
+
+			// Classify filter/predicate ops to perform various optimizations during evaluation
+			if (_type == ast_filter || _type == ast_predicate)
+			{
+				assert(_test == predicate_default);
+
+				if (_right->_type == ast_number_constant && _right->_data.number == 1.0)
+					_test = predicate_constant_one;
+				else if (_right->_rettype == xpath_type_number && (_right->_type == ast_number_constant || _right->_type == ast_variable || _right->_type == ast_func_last))
+					_test = predicate_constant;
+				else if (_right->_rettype != xpath_type_number && _right->is_posinv_expr())
+					_test = predicate_posinv;
+			}
+
+			// Rewrite descendant-or-self::node()/child::foo with descendant::foo
+			// The former is a full form of //foo, the latter is much faster since it executes the node test immediately
+			// Do a similar kind of rewrite for self/descendant/descendant-or-self axes
+			// Note that we only rewrite positionally invariant steps (//foo[1] != /descendant::foo[1])
+			if (_type == ast_step && (_axis == axis_child || _axis == axis_self || _axis == axis_descendant || _axis == axis_descendant_or_self) && _left &&
+				_left->_type == ast_step && _left->_axis == axis_descendant_or_self && _left->_test == nodetest_type_node && !_left->_right &&
+				is_posinv_step())
+			{
+				if (_axis == axis_child || _axis == axis_descendant)
+					_axis = axis_descendant;
+				else
+					_axis = axis_descendant_or_self;
+
+				_left = _left->_left;
+			}
+
+			// Use optimized lookup table implementation for translate() with constant arguments
+			if (_type == ast_func_translate && _right->_type == ast_string_constant && _right->_next->_type == ast_string_constant)
+			{
+				unsigned char* table = translate_table_generate(alloc, _right->_data.string, _right->_next->_data.string);
+
+				if (table)
+				{
+					_type = ast_opt_translate_table;
+					_data.table = table;
+				}
+			}
+
+			// Use optimized path for @attr = 'value' or @attr = $value
+			if (_type == ast_op_equal &&
+				_left->_type == ast_step && _left->_axis == axis_attribute && _left->_test == nodetest_name && !_left->_left && !_left->_right &&
+				(_right->_type == ast_string_constant || (_right->_type == ast_variable && _right->_rettype == xpath_type_string)))
+			{
+				_type = ast_opt_compare_attribute;
+			}
+		}
+		
+		bool is_posinv_expr() const
+		{
+			switch (_type)
+			{
+			case ast_func_position:
+			case ast_func_last:
+				return false;
+
+			case ast_string_constant:
+			case ast_number_constant:
+			case ast_variable:
+				return true;
+
+			case ast_step:
+			case ast_step_root:
+				return true;
+
+			case ast_predicate:
+			case ast_filter:
+				return true;
+
+			default:
+				if (_left && !_left->is_posinv_expr()) return false;
+				
+				for (xpath_ast_node* n = _right; n; n = n->_next)
+					if (!n->is_posinv_expr()) return false;
+					
+				return true;
+			}
+		}
+
+		bool is_posinv_step() const
+		{
+			assert(_type == ast_step);
+
+			for (xpath_ast_node* n = _right; n; n = n->_next)
+			{
+				assert(n->_type == ast_predicate);
+
+				if (n->_test != predicate_posinv)
+					return false;
+			}
+
+			return true;
+		}
+
+		xpath_value_type rettype() const
+		{
+			return static_cast<xpath_value_type>(_rettype);
+		}
+	};
+
+	struct xpath_parser
+	{
+		xpath_allocator* _alloc;
+		xpath_lexer _lexer;
+
+		const char_t* _query;
+		xpath_variable_set* _variables;
+
+		xpath_parse_result* _result;
+
+		char_t _scratch[32];
+
+	#ifdef PUGIXML_NO_EXCEPTIONS
+		jmp_buf _error_handler;
+	#endif
+
+		void throw_error(const char* message)
+		{
+			_result->error = message;
+			_result->offset = _lexer.current_pos() - _query;
+
+		#ifdef PUGIXML_NO_EXCEPTIONS
+			longjmp(_error_handler, 1);
+		#else
+			throw xpath_exception(*_result);
+		#endif
+		}
+
+		void throw_error_oom()
+		{
+		#ifdef PUGIXML_NO_EXCEPTIONS
+			throw_error("Out of memory");
+		#else
+			throw std::bad_alloc();
+		#endif
+		}
+
+		void* alloc_node()
+		{
+			void* result = _alloc->allocate_nothrow(sizeof(xpath_ast_node));
+
+			if (!result) throw_error_oom();
+
+			return result;
+		}
+
+		const char_t* alloc_string(const xpath_lexer_string& value)
+		{
+			if (value.begin)
+			{
+				size_t length = static_cast<size_t>(value.end - value.begin);
+
+				char_t* c = static_cast<char_t*>(_alloc->allocate_nothrow((length + 1) * sizeof(char_t)));
+				if (!c) throw_error_oom();
+				assert(c); // workaround for clang static analysis
+
+				memcpy(c, value.begin, length * sizeof(char_t));
+				c[length] = 0;
+
+				return c;
+			}
+			else return 0;
+		}
+
+		xpath_ast_node* parse_function_helper(ast_type_t type0, ast_type_t type1, size_t argc, xpath_ast_node* args[2])
+		{
+			assert(argc <= 1);
+
+			if (argc == 1 && args[0]->rettype() != xpath_type_node_set) throw_error("Function has to be applied to node set");
+
+			return new (alloc_node()) xpath_ast_node(argc == 0 ? type0 : type1, xpath_type_string, args[0]);
+		}
+
+		xpath_ast_node* parse_function(const xpath_lexer_string& name, size_t argc, xpath_ast_node* args[2])
+		{
+			switch (name.begin[0])
+			{
+			case 'b':
+				if (name == PUGIXML_TEXT("boolean") && argc == 1)
+					return new (alloc_node()) xpath_ast_node(ast_func_boolean, xpath_type_boolean, args[0]);
+					
+				break;
+			
+			case 'c':
+				if (name == PUGIXML_TEXT("count") && argc == 1)
+				{
+					if (args[0]->rettype() != xpath_type_node_set) throw_error("Function has to be applied to node set");
+					return new (alloc_node()) xpath_ast_node(ast_func_count, xpath_type_number, args[0]);
+				}
+				else if (name == PUGIXML_TEXT("contains") && argc == 2)
+					return new (alloc_node()) xpath_ast_node(ast_func_contains, xpath_type_boolean, args[0], args[1]);
+				else if (name == PUGIXML_TEXT("concat") && argc >= 2)
+					return new (alloc_node()) xpath_ast_node(ast_func_concat, xpath_type_string, args[0], args[1]);
+				else if (name == PUGIXML_TEXT("ceiling") && argc == 1)
+					return new (alloc_node()) xpath_ast_node(ast_func_ceiling, xpath_type_number, args[0]);
+					
+				break;
+			
+			case 'f':
+				if (name == PUGIXML_TEXT("false") && argc == 0)
+					return new (alloc_node()) xpath_ast_node(ast_func_false, xpath_type_boolean);
+				else if (name == PUGIXML_TEXT("floor") && argc == 1)
+					return new (alloc_node()) xpath_ast_node(ast_func_floor, xpath_type_number, args[0]);
+					
+				break;
+			
+			case 'i':
+				if (name == PUGIXML_TEXT("id") && argc == 1)
+					return new (alloc_node()) xpath_ast_node(ast_func_id, xpath_type_node_set, args[0]);
+					
+				break;
+			
+			case 'l':
+				if (name == PUGIXML_TEXT("last") && argc == 0)
+					return new (alloc_node()) xpath_ast_node(ast_func_last, xpath_type_number);
+				else if (name == PUGIXML_TEXT("lang") && argc == 1)
+					return new (alloc_node()) xpath_ast_node(ast_func_lang, xpath_type_boolean, args[0]);
+				else if (name == PUGIXML_TEXT("local-name") && argc <= 1)
+					return parse_function_helper(ast_func_local_name_0, ast_func_local_name_1, argc, args);
+			
+				break;
+			
+			case 'n':
+				if (name == PUGIXML_TEXT("name") && argc <= 1)
+					return parse_function_helper(ast_func_name_0, ast_func_name_1, argc, args);
+				else if (name == PUGIXML_TEXT("namespace-uri") && argc <= 1)
+					return parse_function_helper(ast_func_namespace_uri_0, ast_func_namespace_uri_1, argc, args);
+				else if (name == PUGIXML_TEXT("normalize-space") && argc <= 1)
+					return new (alloc_node()) xpath_ast_node(argc == 0 ? ast_func_normalize_space_0 : ast_func_normalize_space_1, xpath_type_string, args[0], args[1]);
+				else if (name == PUGIXML_TEXT("not") && argc == 1)
+					return new (alloc_node()) xpath_ast_node(ast_func_not, xpath_type_boolean, args[0]);
+				else if (name == PUGIXML_TEXT("number") && argc <= 1)
+					return new (alloc_node()) xpath_ast_node(argc == 0 ? ast_func_number_0 : ast_func_number_1, xpath_type_number, args[0]);
+			
+				break;
+			
+			case 'p':
+				if (name == PUGIXML_TEXT("position") && argc == 0)
+					return new (alloc_node()) xpath_ast_node(ast_func_position, xpath_type_number);
+				
+				break;
+			
+			case 'r':
+				if (name == PUGIXML_TEXT("round") && argc == 1)
+					return new (alloc_node()) xpath_ast_node(ast_func_round, xpath_type_number, args[0]);
+
+				break;
+			
+			case 's':
+				if (name == PUGIXML_TEXT("string") && argc <= 1)
+					return new (alloc_node()) xpath_ast_node(argc == 0 ? ast_func_string_0 : ast_func_string_1, xpath_type_string, args[0]);
+				else if (name == PUGIXML_TEXT("string-length") && argc <= 1)
+					return new (alloc_node()) xpath_ast_node(argc == 0 ? ast_func_string_length_0 : ast_func_string_length_1, xpath_type_number, args[0]);
+				else if (name == PUGIXML_TEXT("starts-with") && argc == 2)
+					return new (alloc_node()) xpath_ast_node(ast_func_starts_with, xpath_type_boolean, args[0], args[1]);
+				else if (name == PUGIXML_TEXT("substring-before") && argc == 2)
+					return new (alloc_node()) xpath_ast_node(ast_func_substring_before, xpath_type_string, args[0], args[1]);
+				else if (name == PUGIXML_TEXT("substring-after") && argc == 2)
+					return new (alloc_node()) xpath_ast_node(ast_func_substring_after, xpath_type_string, args[0], args[1]);
+				else if (name == PUGIXML_TEXT("substring") && (argc == 2 || argc == 3))
+					return new (alloc_node()) xpath_ast_node(argc == 2 ? ast_func_substring_2 : ast_func_substring_3, xpath_type_string, args[0], args[1]);
+				else if (name == PUGIXML_TEXT("sum") && argc == 1)
+				{
+					if (args[0]->rettype() != xpath_type_node_set) throw_error("Function has to be applied to node set");
+					return new (alloc_node()) xpath_ast_node(ast_func_sum, xpath_type_number, args[0]);
+				}
+
+				break;
+			
+			case 't':
+				if (name == PUGIXML_TEXT("translate") && argc == 3)
+					return new (alloc_node()) xpath_ast_node(ast_func_translate, xpath_type_string, args[0], args[1]);
+				else if (name == PUGIXML_TEXT("true") && argc == 0)
+					return new (alloc_node()) xpath_ast_node(ast_func_true, xpath_type_boolean);
+					
+				break;
+
+			default:
+				break;
+			}
+
+			throw_error("Unrecognized function or wrong parameter count");
+
+			return 0;
+		}
+
+		axis_t parse_axis_name(const xpath_lexer_string& name, bool& specified)
+		{
+			specified = true;
+
+			switch (name.begin[0])
+			{
+			case 'a':
+				if (name == PUGIXML_TEXT("ancestor"))
+					return axis_ancestor;
+				else if (name == PUGIXML_TEXT("ancestor-or-self"))
+					return axis_ancestor_or_self;
+				else if (name == PUGIXML_TEXT("attribute"))
+					return axis_attribute;
+				
+				break;
+			
+			case 'c':
+				if (name == PUGIXML_TEXT("child"))
+					return axis_child;
+				
+				break;
+			
+			case 'd':
+				if (name == PUGIXML_TEXT("descendant"))
+					return axis_descendant;
+				else if (name == PUGIXML_TEXT("descendant-or-self"))
+					return axis_descendant_or_self;
+				
+				break;
+			
+			case 'f':
+				if (name == PUGIXML_TEXT("following"))
+					return axis_following;
+				else if (name == PUGIXML_TEXT("following-sibling"))
+					return axis_following_sibling;
+				
+				break;
+			
+			case 'n':
+				if (name == PUGIXML_TEXT("namespace"))
+					return axis_namespace;
+				
+				break;
+			
+			case 'p':
+				if (name == PUGIXML_TEXT("parent"))
+					return axis_parent;
+				else if (name == PUGIXML_TEXT("preceding"))
+					return axis_preceding;
+				else if (name == PUGIXML_TEXT("preceding-sibling"))
+					return axis_preceding_sibling;
+				
+				break;
+			
+			case 's':
+				if (name == PUGIXML_TEXT("self"))
+					return axis_self;
+				
+				break;
+
+			default:
+				break;
+			}
+
+			specified = false;
+			return axis_child;
+		}
+
+		nodetest_t parse_node_test_type(const xpath_lexer_string& name)
+		{
+			switch (name.begin[0])
+			{
+			case 'c':
+				if (name == PUGIXML_TEXT("comment"))
+					return nodetest_type_comment;
+
+				break;
+
+			case 'n':
+				if (name == PUGIXML_TEXT("node"))
+					return nodetest_type_node;
+
+				break;
+
+			case 'p':
+				if (name == PUGIXML_TEXT("processing-instruction"))
+					return nodetest_type_pi;
+
+				break;
+
+			case 't':
+				if (name == PUGIXML_TEXT("text"))
+					return nodetest_type_text;
+
+				break;
+			
+			default:
+				break;
+			}
+
+			return nodetest_none;
+		}
+
+		// PrimaryExpr ::= VariableReference | '(' Expr ')' | Literal | Number | FunctionCall
+		xpath_ast_node* parse_primary_expression()
+		{
+			switch (_lexer.current())
+			{
+			case lex_var_ref:
+			{
+				xpath_lexer_string name = _lexer.contents();
+
+				if (!_variables)
+					throw_error("Unknown variable: variable set is not provided");
+
+				xpath_variable* var = 0;
+				if (!get_variable_scratch(_scratch, _variables, name.begin, name.end, &var))
+					throw_error_oom();
+
+				if (!var)
+					throw_error("Unknown variable: variable set does not contain the given name");
+
+				_lexer.next();
+
+				return new (alloc_node()) xpath_ast_node(ast_variable, var->type(), var);
+			}
+
+			case lex_open_brace:
+			{
+				_lexer.next();
+
+				xpath_ast_node* n = parse_expression();
+
+				if (_lexer.current() != lex_close_brace)
+					throw_error("Unmatched braces");
+
+				_lexer.next();
+
+				return n;
+			}
+
+			case lex_quoted_string:
+			{
+				const char_t* value = alloc_string(_lexer.contents());
+
+				xpath_ast_node* n = new (alloc_node()) xpath_ast_node(ast_string_constant, xpath_type_string, value);
+				_lexer.next();
+
+				return n;
+			}
+
+			case lex_number:
+			{
+				double value = 0;
+
+				if (!convert_string_to_number_scratch(_scratch, _lexer.contents().begin, _lexer.contents().end, &value))
+					throw_error_oom();
+
+				xpath_ast_node* n = new (alloc_node()) xpath_ast_node(ast_number_constant, xpath_type_number, value);
+				_lexer.next();
+
+				return n;
+			}
+
+			case lex_string:
+			{
+				xpath_ast_node* args[2] = {0};
+				size_t argc = 0;
+				
+				xpath_lexer_string function = _lexer.contents();
+				_lexer.next();
+				
+				xpath_ast_node* last_arg = 0;
+				
+				if (_lexer.current() != lex_open_brace)
+					throw_error("Unrecognized function call");
+				_lexer.next();
+
+				if (_lexer.current() != lex_close_brace)
+					args[argc++] = parse_expression();
+
+				while (_lexer.current() != lex_close_brace)
+				{
+					if (_lexer.current() != lex_comma)
+						throw_error("No comma between function arguments");
+					_lexer.next();
+					
+					xpath_ast_node* n = parse_expression();
+					
+					if (argc < 2) args[argc] = n;
+					else last_arg->set_next(n);
+
+					argc++;
+					last_arg = n;
+				}
+				
+				_lexer.next();
+
+				return parse_function(function, argc, args);
+			}
+
+			default:
+				throw_error("Unrecognizable primary expression");
+
+				return 0;
+			}
+		}
+		
+		// FilterExpr ::= PrimaryExpr | FilterExpr Predicate
+		// Predicate ::= '[' PredicateExpr ']'
+		// PredicateExpr ::= Expr
+		xpath_ast_node* parse_filter_expression()
+		{
+			xpath_ast_node* n = parse_primary_expression();
+
+			while (_lexer.current() == lex_open_square_brace)
+			{
+				_lexer.next();
+
+				xpath_ast_node* expr = parse_expression();
+
+				if (n->rettype() != xpath_type_node_set) throw_error("Predicate has to be applied to node set");
+
+				n = new (alloc_node()) xpath_ast_node(ast_filter, n, expr, predicate_default);
+
+				if (_lexer.current() != lex_close_square_brace)
+					throw_error("Unmatched square brace");
+			
+				_lexer.next();
+			}
+			
+			return n;
+		}
+		
+		// Step ::= AxisSpecifier NodeTest Predicate* | AbbreviatedStep
+		// AxisSpecifier ::= AxisName '::' | '@'?
+		// NodeTest ::= NameTest | NodeType '(' ')' | 'processing-instruction' '(' Literal ')'
+		// NameTest ::= '*' | NCName ':' '*' | QName
+		// AbbreviatedStep ::= '.' | '..'
+		xpath_ast_node* parse_step(xpath_ast_node* set)
+		{
+			if (set && set->rettype() != xpath_type_node_set)
+				throw_error("Step has to be applied to node set");
+
+			bool axis_specified = false;
+			axis_t axis = axis_child; // implied child axis
+
+			if (_lexer.current() == lex_axis_attribute)
+			{
+				axis = axis_attribute;
+				axis_specified = true;
+				
+				_lexer.next();
+			}
+			else if (_lexer.current() == lex_dot)
+			{
+				_lexer.next();
+				
+				return new (alloc_node()) xpath_ast_node(ast_step, set, axis_self, nodetest_type_node, 0);
+			}
+			else if (_lexer.current() == lex_double_dot)
+			{
+				_lexer.next();
+				
+				return new (alloc_node()) xpath_ast_node(ast_step, set, axis_parent, nodetest_type_node, 0);
+			}
+		
+			nodetest_t nt_type = nodetest_none;
+			xpath_lexer_string nt_name;
+			
+			if (_lexer.current() == lex_string)
+			{
+				// node name test
+				nt_name = _lexer.contents();
+				_lexer.next();
+
+				// was it an axis name?
+				if (_lexer.current() == lex_double_colon)
+				{
+					// parse axis name
+					if (axis_specified) throw_error("Two axis specifiers in one step");
+
+					axis = parse_axis_name(nt_name, axis_specified);
+
+					if (!axis_specified) throw_error("Unknown axis");
+
+					// read actual node test
+					_lexer.next();
+
+					if (_lexer.current() == lex_multiply)
+					{
+						nt_type = nodetest_all;
+						nt_name = xpath_lexer_string();
+						_lexer.next();
+					}
+					else if (_lexer.current() == lex_string)
+					{
+						nt_name = _lexer.contents();
+						_lexer.next();
+					}
+					else throw_error("Unrecognized node test");
+				}
+				
+				if (nt_type == nodetest_none)
+				{
+					// node type test or processing-instruction
+					if (_lexer.current() == lex_open_brace)
+					{
+						_lexer.next();
+						
+						if (_lexer.current() == lex_close_brace)
+						{
+							_lexer.next();
+
+							nt_type = parse_node_test_type(nt_name);
+
+							if (nt_type == nodetest_none) throw_error("Unrecognized node type");
+							
+							nt_name = xpath_lexer_string();
+						}
+						else if (nt_name == PUGIXML_TEXT("processing-instruction"))
+						{
+							if (_lexer.current() != lex_quoted_string)
+								throw_error("Only literals are allowed as arguments to processing-instruction()");
+						
+							nt_type = nodetest_pi;
+							nt_name = _lexer.contents();
+							_lexer.next();
+							
+							if (_lexer.current() != lex_close_brace)
+								throw_error("Unmatched brace near processing-instruction()");
+							_lexer.next();
+						}
+						else
+							throw_error("Unmatched brace near node type test");
+
+					}
+					// QName or NCName:*
+					else
+					{
+						if (nt_name.end - nt_name.begin > 2 && nt_name.end[-2] == ':' && nt_name.end[-1] == '*') // NCName:*
+						{
+							nt_name.end--; // erase *
+							
+							nt_type = nodetest_all_in_namespace;
+						}
+						else nt_type = nodetest_name;
+					}
+				}
+			}
+			else if (_lexer.current() == lex_multiply)
+			{
+				nt_type = nodetest_all;
+				_lexer.next();
+			}
+			else throw_error("Unrecognized node test");
+			
+			xpath_ast_node* n = new (alloc_node()) xpath_ast_node(ast_step, set, axis, nt_type, alloc_string(nt_name));
+			
+			xpath_ast_node* last = 0;
+			
+			while (_lexer.current() == lex_open_square_brace)
+			{
+				_lexer.next();
+				
+				xpath_ast_node* expr = parse_expression();
+
+				xpath_ast_node* pred = new (alloc_node()) xpath_ast_node(ast_predicate, 0, expr, predicate_default);
+				
+				if (_lexer.current() != lex_close_square_brace)
+					throw_error("Unmatched square brace");
+				_lexer.next();
+				
+				if (last) last->set_next(pred);
+				else n->set_right(pred);
+				
+				last = pred;
+			}
+
+			return n;
+		}
+		
+		// RelativeLocationPath ::= Step | RelativeLocationPath '/' Step | RelativeLocationPath '//' Step
+		xpath_ast_node* parse_relative_location_path(xpath_ast_node* set)
+		{
+			xpath_ast_node* n = parse_step(set);
+			
+			while (_lexer.current() == lex_slash || _lexer.current() == lex_double_slash)
+			{
+				lexeme_t l = _lexer.current();
+				_lexer.next();
+
+				if (l == lex_double_slash)
+					n = new (alloc_node()) xpath_ast_node(ast_step, n, axis_descendant_or_self, nodetest_type_node, 0);
+				
+				n = parse_step(n);
+			}
+			
+			return n;
+		}
+		
+		// LocationPath ::= RelativeLocationPath | AbsoluteLocationPath
+		// AbsoluteLocationPath ::= '/' RelativeLocationPath? | '//' RelativeLocationPath
+		xpath_ast_node* parse_location_path()
+		{
+			if (_lexer.current() == lex_slash)
+			{
+				_lexer.next();
+				
+				xpath_ast_node* n = new (alloc_node()) xpath_ast_node(ast_step_root, xpath_type_node_set);
+
+				// relative location path can start from axis_attribute, dot, double_dot, multiply and string lexemes; any other lexeme means standalone root path
+				lexeme_t l = _lexer.current();
+
+				if (l == lex_string || l == lex_axis_attribute || l == lex_dot || l == lex_double_dot || l == lex_multiply)
+					return parse_relative_location_path(n);
+				else
+					return n;
+			}
+			else if (_lexer.current() == lex_double_slash)
+			{
+				_lexer.next();
+				
+				xpath_ast_node* n = new (alloc_node()) xpath_ast_node(ast_step_root, xpath_type_node_set);
+				n = new (alloc_node()) xpath_ast_node(ast_step, n, axis_descendant_or_self, nodetest_type_node, 0);
+				
+				return parse_relative_location_path(n);
+			}
+
+			// else clause moved outside of if because of bogus warning 'control may reach end of non-void function being inlined' in gcc 4.0.1
+			return parse_relative_location_path(0);
+		}
+		
+		// PathExpr ::= LocationPath
+		//				| FilterExpr
+		//				| FilterExpr '/' RelativeLocationPath
+		//				| FilterExpr '//' RelativeLocationPath
+		// UnionExpr ::= PathExpr | UnionExpr '|' PathExpr
+		// UnaryExpr ::= UnionExpr | '-' UnaryExpr
+		xpath_ast_node* parse_path_or_unary_expression()
+		{
+			// Clarification.
+			// PathExpr begins with either LocationPath or FilterExpr.
+			// FilterExpr begins with PrimaryExpr
+			// PrimaryExpr begins with '$' in case of it being a variable reference,
+			// '(' in case of it being an expression, string literal, number constant or
+			// function call.
+
+			if (_lexer.current() == lex_var_ref || _lexer.current() == lex_open_brace || 
+				_lexer.current() == lex_quoted_string || _lexer.current() == lex_number ||
+				_lexer.current() == lex_string)
+			{
+				if (_lexer.current() == lex_string)
+				{
+					// This is either a function call, or not - if not, we shall proceed with location path
+					const char_t* state = _lexer.state();
+					
+					while (PUGI__IS_CHARTYPE(*state, ct_space)) ++state;
+					
+					if (*state != '(') return parse_location_path();
+
+					// This looks like a function call; however this still can be a node-test. Check it.
+					if (parse_node_test_type(_lexer.contents()) != nodetest_none) return parse_location_path();
+				}
+				
+				xpath_ast_node* n = parse_filter_expression();
+
+				if (_lexer.current() == lex_slash || _lexer.current() == lex_double_slash)
+				{
+					lexeme_t l = _lexer.current();
+					_lexer.next();
+					
+					if (l == lex_double_slash)
+					{
+						if (n->rettype() != xpath_type_node_set) throw_error("Step has to be applied to node set");
+
+						n = new (alloc_node()) xpath_ast_node(ast_step, n, axis_descendant_or_self, nodetest_type_node, 0);
+					}
+	
+					// select from location path
+					return parse_relative_location_path(n);
+				}
+
+				return n;
+			}
+			else if (_lexer.current() == lex_minus)
+			{
+				_lexer.next();
+
+				// precedence 7+ - only parses union expressions
+				xpath_ast_node* expr = parse_expression_rec(parse_path_or_unary_expression(), 7);
+
+				return new (alloc_node()) xpath_ast_node(ast_op_negate, xpath_type_number, expr);
+			}
+			else
+				return parse_location_path();
+		}
+
+		struct binary_op_t
+		{
+			ast_type_t asttype;
+			xpath_value_type rettype;
+			int precedence;
+
+			binary_op_t(): asttype(ast_unknown), rettype(xpath_type_none), precedence(0)
+			{
+			}
+
+			binary_op_t(ast_type_t asttype_, xpath_value_type rettype_, int precedence_): asttype(asttype_), rettype(rettype_), precedence(precedence_)
+			{
+			}
+
+			static binary_op_t parse(xpath_lexer& lexer)
+			{
+				switch (lexer.current())
+				{
+				case lex_string:
+					if (lexer.contents() == PUGIXML_TEXT("or"))
+						return binary_op_t(ast_op_or, xpath_type_boolean, 1);
+					else if (lexer.contents() == PUGIXML_TEXT("and"))
+						return binary_op_t(ast_op_and, xpath_type_boolean, 2);
+					else if (lexer.contents() == PUGIXML_TEXT("div"))
+						return binary_op_t(ast_op_divide, xpath_type_number, 6);
+					else if (lexer.contents() == PUGIXML_TEXT("mod"))
+						return binary_op_t(ast_op_mod, xpath_type_number, 6);
+					else
+						return binary_op_t();
+
+				case lex_equal:
+					return binary_op_t(ast_op_equal, xpath_type_boolean, 3);
+
+				case lex_not_equal:
+					return binary_op_t(ast_op_not_equal, xpath_type_boolean, 3);
+
+				case lex_less:
+					return binary_op_t(ast_op_less, xpath_type_boolean, 4);
+
+				case lex_greater:
+					return binary_op_t(ast_op_greater, xpath_type_boolean, 4);
+
+				case lex_less_or_equal:
+					return binary_op_t(ast_op_less_or_equal, xpath_type_boolean, 4);
+
+				case lex_greater_or_equal:
+					return binary_op_t(ast_op_greater_or_equal, xpath_type_boolean, 4);
+
+				case lex_plus:
+					return binary_op_t(ast_op_add, xpath_type_number, 5);
+
+				case lex_minus:
+					return binary_op_t(ast_op_subtract, xpath_type_number, 5);
+
+				case lex_multiply:
+					return binary_op_t(ast_op_multiply, xpath_type_number, 6);
+
+				case lex_union:
+					return binary_op_t(ast_op_union, xpath_type_node_set, 7);
+
+				default:
+					return binary_op_t();
+				}
+			}
+		};
+
+		xpath_ast_node* parse_expression_rec(xpath_ast_node* lhs, int limit)
+		{
+			binary_op_t op = binary_op_t::parse(_lexer);
+
+			while (op.asttype != ast_unknown && op.precedence >= limit)
+			{
+				_lexer.next();
+
+				xpath_ast_node* rhs = parse_path_or_unary_expression();
+
+				binary_op_t nextop = binary_op_t::parse(_lexer);
+
+				while (nextop.asttype != ast_unknown && nextop.precedence > op.precedence)
+				{
+					rhs = parse_expression_rec(rhs, nextop.precedence);
+
+					nextop = binary_op_t::parse(_lexer);
+				}
+
+				if (op.asttype == ast_op_union && (lhs->rettype() != xpath_type_node_set || rhs->rettype() != xpath_type_node_set))
+					throw_error("Union operator has to be applied to node sets");
+
+				lhs = new (alloc_node()) xpath_ast_node(op.asttype, op.rettype, lhs, rhs);
+
+				op = binary_op_t::parse(_lexer);
+			}
+
+			return lhs;
+		}
+
+		// Expr ::= OrExpr
+		// OrExpr ::= AndExpr | OrExpr 'or' AndExpr
+		// AndExpr ::= EqualityExpr | AndExpr 'and' EqualityExpr
+		// EqualityExpr ::= RelationalExpr
+		//					| EqualityExpr '=' RelationalExpr
+		//					| EqualityExpr '!=' RelationalExpr
+		// RelationalExpr ::= AdditiveExpr
+		//					  | RelationalExpr '<' AdditiveExpr
+		//					  | RelationalExpr '>' AdditiveExpr
+		//					  | RelationalExpr '<=' AdditiveExpr
+		//					  | RelationalExpr '>=' AdditiveExpr
+		// AdditiveExpr ::= MultiplicativeExpr
+		//					| AdditiveExpr '+' MultiplicativeExpr
+		//					| AdditiveExpr '-' MultiplicativeExpr
+		// MultiplicativeExpr ::= UnaryExpr
+		//						  | MultiplicativeExpr '*' UnaryExpr
+		//						  | MultiplicativeExpr 'div' UnaryExpr
+		//						  | MultiplicativeExpr 'mod' UnaryExpr
+		xpath_ast_node* parse_expression()
+		{
+			return parse_expression_rec(parse_path_or_unary_expression(), 0);
+		}
+
+		xpath_parser(const char_t* query, xpath_variable_set* variables, xpath_allocator* alloc, xpath_parse_result* result): _alloc(alloc), _lexer(query), _query(query), _variables(variables), _result(result)
+		{
+		}
+
+		xpath_ast_node* parse()
+		{
+			xpath_ast_node* result = parse_expression();
+			
+			if (_lexer.current() != lex_eof)
+			{
+				// there are still unparsed tokens left, error
+				throw_error("Incorrect query");
+			}
+			
+			return result;
+		}
+
+		static xpath_ast_node* parse(const char_t* query, xpath_variable_set* variables, xpath_allocator* alloc, xpath_parse_result* result)
+		{
+			xpath_parser parser(query, variables, alloc, result);
+
+		#ifdef PUGIXML_NO_EXCEPTIONS
+			int error = setjmp(parser._error_handler);
+
+			return (error == 0) ? parser.parse() : 0;
+		#else
+			return parser.parse();
+		#endif
+		}
+	};
+
+	struct xpath_query_impl
+	{
+		static xpath_query_impl* create()
+		{
+			void* memory = xml_memory::allocate(sizeof(xpath_query_impl));
+			if (!memory) return 0;
+
+			return new (memory) xpath_query_impl();
+		}
+
+		static void destroy(xpath_query_impl* impl)
+		{
+			// free all allocated pages
+			impl->alloc.release();
+
+			// free allocator memory (with the first page)
+			xml_memory::deallocate(impl);
+		}
+
+		xpath_query_impl(): root(0), alloc(&block)
+		{
+			block.next = 0;
+			block.capacity = sizeof(block.data);
+		}
+
+		xpath_ast_node* root;
+		xpath_allocator alloc;
+		xpath_memory_block block;
+	};
+
+	PUGI__FN xpath_string evaluate_string_impl(xpath_query_impl* impl, const xpath_node& n, xpath_stack_data& sd)
+	{
+		if (!impl) return xpath_string();
+
+	#ifdef PUGIXML_NO_EXCEPTIONS
+		if (setjmp(sd.error_handler)) return xpath_string();
+	#endif
+
+		xpath_context c(n, 1, 1);
+
+		return impl->root->eval_string(c, sd.stack);
+	}
+
+	PUGI__FN impl::xpath_ast_node* evaluate_node_set_prepare(xpath_query_impl* impl)
+	{
+		if (!impl) return 0;
+
+		if (impl->root->rettype() != xpath_type_node_set)
+		{
+		#ifdef PUGIXML_NO_EXCEPTIONS
+			return 0;
+		#else
+			xpath_parse_result res;
+			res.error = "Expression does not evaluate to node set";
+
+			throw xpath_exception(res);
+		#endif
+		}
+
+		return impl->root;
+	}
+PUGI__NS_END
+
+namespace pugi
+{
+#ifndef PUGIXML_NO_EXCEPTIONS
+	PUGI__FN xpath_exception::xpath_exception(const xpath_parse_result& result_): _result(result_)
+	{
+		assert(_result.error);
+	}
+	
+	PUGI__FN const char* xpath_exception::what() const throw()
+	{
+		return _result.error;
+	}
+
+	PUGI__FN const xpath_parse_result& xpath_exception::result() const
+	{
+		return _result;
+	}
+#endif
+	
+	PUGI__FN xpath_node::xpath_node()
+	{
+	}
+		
+	PUGI__FN xpath_node::xpath_node(const xml_node& node_): _node(node_)
+	{
+	}
+		
+	PUGI__FN xpath_node::xpath_node(const xml_attribute& attribute_, const xml_node& parent_): _node(attribute_ ? parent_ : xml_node()), _attribute(attribute_)
+	{
+	}
+
+	PUGI__FN xml_node xpath_node::node() const
+	{
+		return _attribute ? xml_node() : _node;
+	}
+		
+	PUGI__FN xml_attribute xpath_node::attribute() const
+	{
+		return _attribute;
+	}
+	
+	PUGI__FN xml_node xpath_node::parent() const
+	{
+		return _attribute ? _node : _node.parent();
+	}
+
+	PUGI__FN static void unspecified_bool_xpath_node(xpath_node***)
+	{
+	}
+
+	PUGI__FN xpath_node::operator xpath_node::unspecified_bool_type() const
+	{
+		return (_node || _attribute) ? unspecified_bool_xpath_node : 0;
+	}
+	
+	PUGI__FN bool xpath_node::operator!() const
+	{
+		return !(_node || _attribute);
+	}
+
+	PUGI__FN bool xpath_node::operator==(const xpath_node& n) const
+	{
+		return _node == n._node && _attribute == n._attribute;
+	}
+	
+	PUGI__FN bool xpath_node::operator!=(const xpath_node& n) const
+	{
+		return _node != n._node || _attribute != n._attribute;
+	}
+
+#ifdef __BORLANDC__
+	PUGI__FN bool operator&&(const xpath_node& lhs, bool rhs)
+	{
+		return (bool)lhs && rhs;
+	}
+
+	PUGI__FN bool operator||(const xpath_node& lhs, bool rhs)
+	{
+		return (bool)lhs || rhs;
+	}
+#endif
+
+	PUGI__FN void xpath_node_set::_assign(const_iterator begin_, const_iterator end_, type_t type_)
+	{
+		assert(begin_ <= end_);
+
+		size_t size_ = static_cast<size_t>(end_ - begin_);
+
+		if (size_ <= 1)
+		{
+			// deallocate old buffer
+			if (_begin != &_storage) impl::xml_memory::deallocate(_begin);
+
+			// use internal buffer
+			if (begin_ != end_) _storage = *begin_;
+
+			_begin = &_storage;
+			_end = &_storage + size_;
+			_type = type_;
+		}
+		else
+		{
+			// make heap copy
+			xpath_node* storage = static_cast<xpath_node*>(impl::xml_memory::allocate(size_ * sizeof(xpath_node)));
+
+			if (!storage)
+			{
+			#ifdef PUGIXML_NO_EXCEPTIONS
+				return;
+			#else
+				throw std::bad_alloc();
+			#endif
+			}
+
+			memcpy(storage, begin_, size_ * sizeof(xpath_node));
+			
+			// deallocate old buffer
+			if (_begin != &_storage) impl::xml_memory::deallocate(_begin);
+
+			// finalize
+			_begin = storage;
+			_end = storage + size_;
+			_type = type_;
+		}
+	}
+
+#if __cplusplus >= 201103
+	PUGI__FN void xpath_node_set::_move(xpath_node_set& rhs)
+	{
+		_type = rhs._type;
+		_storage = rhs._storage;
+		_begin = (rhs._begin == &rhs._storage) ? &_storage : rhs._begin;
+		_end = _begin + (rhs._end - rhs._begin);
+
+		rhs._type = type_unsorted;
+		rhs._begin = &rhs._storage;
+		rhs._end = rhs._begin;
+	}
+#endif
+
+	PUGI__FN xpath_node_set::xpath_node_set(): _type(type_unsorted), _begin(&_storage), _end(&_storage)
+	{
+	}
+
+	PUGI__FN xpath_node_set::xpath_node_set(const_iterator begin_, const_iterator end_, type_t type_): _type(type_unsorted), _begin(&_storage), _end(&_storage)
+	{
+		_assign(begin_, end_, type_);
+	}
+
+	PUGI__FN xpath_node_set::~xpath_node_set()
+	{
+		if (_begin != &_storage)
+			impl::xml_memory::deallocate(_begin);
+	}
+		
+	PUGI__FN xpath_node_set::xpath_node_set(const xpath_node_set& ns): _type(type_unsorted), _begin(&_storage), _end(&_storage)
+	{
+		_assign(ns._begin, ns._end, ns._type);
+	}
+	
+	PUGI__FN xpath_node_set& xpath_node_set::operator=(const xpath_node_set& ns)
+	{
+		if (this == &ns) return *this;
+
+		_assign(ns._begin, ns._end, ns._type);
+
+		return *this;
+	}
+
+#if __cplusplus >= 201103
+	PUGI__FN xpath_node_set::xpath_node_set(xpath_node_set&& rhs): _type(type_unsorted), _begin(&_storage), _end(&_storage)
+	{
+		_move(rhs);
+	}
+
+	PUGI__FN xpath_node_set& xpath_node_set::operator=(xpath_node_set&& rhs)
+	{
+		if (this == &rhs) return *this;
+
+		if (_begin != &_storage)
+			impl::xml_memory::deallocate(_begin);
+
+		_move(rhs);
+
+		return *this;
+	}
+#endif
+
+	PUGI__FN xpath_node_set::type_t xpath_node_set::type() const
+	{
+		return _type;
+	}
+		
+	PUGI__FN size_t xpath_node_set::size() const
+	{
+		return _end - _begin;
+	}
+		
+	PUGI__FN bool xpath_node_set::empty() const
+	{
+		return _begin == _end;
+	}
+		
+	PUGI__FN const xpath_node& xpath_node_set::operator[](size_t index) const
+	{
+		assert(index < size());
+		return _begin[index];
+	}
+
+	PUGI__FN xpath_node_set::const_iterator xpath_node_set::begin() const
+	{
+		return _begin;
+	}
+		
+	PUGI__FN xpath_node_set::const_iterator xpath_node_set::end() const
+	{
+		return _end;
+	}
+	
+	PUGI__FN void xpath_node_set::sort(bool reverse)
+	{
+		_type = impl::xpath_sort(_begin, _end, _type, reverse);
+	}
+
+	PUGI__FN xpath_node xpath_node_set::first() const
+	{
+		return impl::xpath_first(_begin, _end, _type);
+	}
+
+	PUGI__FN xpath_parse_result::xpath_parse_result(): error("Internal error"), offset(0)
+	{
+	}
+
+	PUGI__FN xpath_parse_result::operator bool() const
+	{
+		return error == 0;
+	}
+
+	PUGI__FN const char* xpath_parse_result::description() const
+	{
+		return error ? error : "No error";
+	}
+
+	PUGI__FN xpath_variable::xpath_variable(xpath_value_type type_): _type(type_), _next(0)
+	{
+	}
+
+	PUGI__FN const char_t* xpath_variable::name() const
+	{
+		switch (_type)
+		{
+		case xpath_type_node_set:
+			return static_cast<const impl::xpath_variable_node_set*>(this)->name;
+
+		case xpath_type_number:
+			return static_cast<const impl::xpath_variable_number*>(this)->name;
+
+		case xpath_type_string:
+			return static_cast<const impl::xpath_variable_string*>(this)->name;
+
+		case xpath_type_boolean:
+			return static_cast<const impl::xpath_variable_boolean*>(this)->name;
+
+		default:
+			assert(!"Invalid variable type");
+			return 0;
+		}
+	}
+
+	PUGI__FN xpath_value_type xpath_variable::type() const
+	{
+		return _type;
+	}
+
+	PUGI__FN bool xpath_variable::get_boolean() const
+	{
+		return (_type == xpath_type_boolean) ? static_cast<const impl::xpath_variable_boolean*>(this)->value : false;
+	}
+
+	PUGI__FN double xpath_variable::get_number() const
+	{
+		return (_type == xpath_type_number) ? static_cast<const impl::xpath_variable_number*>(this)->value : impl::gen_nan();
+	}
+
+	PUGI__FN const char_t* xpath_variable::get_string() const
+	{
+		const char_t* value = (_type == xpath_type_string) ? static_cast<const impl::xpath_variable_string*>(this)->value : 0;
+		return value ? value : PUGIXML_TEXT("");
+	}
+
+	PUGI__FN const xpath_node_set& xpath_variable::get_node_set() const
+	{
+		return (_type == xpath_type_node_set) ? static_cast<const impl::xpath_variable_node_set*>(this)->value : impl::dummy_node_set;
+	}
+
+	PUGI__FN bool xpath_variable::set(bool value)
+	{
+		if (_type != xpath_type_boolean) return false;
+
+		static_cast<impl::xpath_variable_boolean*>(this)->value = value;
+		return true;
+	}
+
+	PUGI__FN bool xpath_variable::set(double value)
+	{
+		if (_type != xpath_type_number) return false;
+
+		static_cast<impl::xpath_variable_number*>(this)->value = value;
+		return true;
+	}
+
+	PUGI__FN bool xpath_variable::set(const char_t* value)
+	{
+		if (_type != xpath_type_string) return false;
+
+		impl::xpath_variable_string* var = static_cast<impl::xpath_variable_string*>(this);
+
+		// duplicate string
+		size_t size = (impl::strlength(value) + 1) * sizeof(char_t);
+
+		char_t* copy = static_cast<char_t*>(impl::xml_memory::allocate(size));
+		if (!copy) return false;
+
+		memcpy(copy, value, size);
+
+		// replace old string
+		if (var->value) impl::xml_memory::deallocate(var->value);
+		var->value = copy;
+
+		return true;
+	}
+
+	PUGI__FN bool xpath_variable::set(const xpath_node_set& value)
+	{
+		if (_type != xpath_type_node_set) return false;
+
+		static_cast<impl::xpath_variable_node_set*>(this)->value = value;
+		return true;
+	}
+
+	PUGI__FN xpath_variable_set::xpath_variable_set()
+	{
+		for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i)
+			_data[i] = 0;
+	}
+
+	PUGI__FN xpath_variable_set::~xpath_variable_set()
+	{
+		for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i)
+			_destroy(_data[i]);
+	}
+
+	PUGI__FN xpath_variable_set::xpath_variable_set(const xpath_variable_set& rhs)
+	{
+		for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i)
+			_data[i] = 0;
+
+		_assign(rhs);
+	}
+
+	PUGI__FN xpath_variable_set& xpath_variable_set::operator=(const xpath_variable_set& rhs)
+	{
+		if (this == &rhs) return *this;
+
+		_assign(rhs);
+
+		return *this;
+	}
+
+#if __cplusplus >= 201103
+	PUGI__FN xpath_variable_set::xpath_variable_set(xpath_variable_set&& rhs)
+	{
+		for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i)
+		{
+			_data[i] = rhs._data[i];
+			rhs._data[i] = 0;
+		}
+	}
+
+	PUGI__FN xpath_variable_set& xpath_variable_set::operator=(xpath_variable_set&& rhs)
+	{
+		for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i)
+		{
+			_destroy(_data[i]);
+
+			_data[i] = rhs._data[i];
+			rhs._data[i] = 0;
+		}
+
+		return *this;
+	}
+#endif
+
+	PUGI__FN void xpath_variable_set::_assign(const xpath_variable_set& rhs)
+	{
+		xpath_variable_set temp;
+
+		for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i)
+			if (rhs._data[i] && !_clone(rhs._data[i], &temp._data[i]))
+				return;
+
+		_swap(temp);
+	}
+
+	PUGI__FN void xpath_variable_set::_swap(xpath_variable_set& rhs)
+	{
+		for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i)
+		{
+			xpath_variable* chain = _data[i];
+
+			_data[i] = rhs._data[i];
+			rhs._data[i] = chain;
+		}
+	}
+
+	PUGI__FN xpath_variable* xpath_variable_set::_find(const char_t* name) const
+	{
+		const size_t hash_size = sizeof(_data) / sizeof(_data[0]);
+		size_t hash = impl::hash_string(name) % hash_size;
+
+		// look for existing variable
+		for (xpath_variable* var = _data[hash]; var; var = var->_next)
+			if (impl::strequal(var->name(), name))
+				return var;
+
+		return 0;
+	}
+
+	PUGI__FN bool xpath_variable_set::_clone(xpath_variable* var, xpath_variable** out_result)
+	{
+		xpath_variable* last = 0;
+
+		while (var)
+		{
+			// allocate storage for new variable
+			xpath_variable* nvar = impl::new_xpath_variable(var->_type, var->name());
+			if (!nvar) return false;
+
+			// link the variable to the result immediately to handle failures gracefully
+			if (last)
+				last->_next = nvar;
+			else
+				*out_result = nvar;
+
+			last = nvar;
+
+			// copy the value; this can fail due to out-of-memory conditions
+			if (!impl::copy_xpath_variable(nvar, var)) return false;
+
+			var = var->_next;
+		}
+
+		return true;
+	}
+
+	PUGI__FN void xpath_variable_set::_destroy(xpath_variable* var)
+	{
+		while (var)
+		{
+			xpath_variable* next = var->_next;
+
+			impl::delete_xpath_variable(var->_type, var);
+
+			var = next;
+		}
+	}
+
+	PUGI__FN xpath_variable* xpath_variable_set::add(const char_t* name, xpath_value_type type)
+	{
+		const size_t hash_size = sizeof(_data) / sizeof(_data[0]);
+		size_t hash = impl::hash_string(name) % hash_size;
+
+		// look for existing variable
+		for (xpath_variable* var = _data[hash]; var; var = var->_next)
+			if (impl::strequal(var->name(), name))
+				return var->type() == type ? var : 0;
+
+		// add new variable
+		xpath_variable* result = impl::new_xpath_variable(type, name);
+
+		if (result)
+		{
+			result->_next = _data[hash];
+
+			_data[hash] = result;
+		}
+
+		return result;
+	}
+
+	PUGI__FN bool xpath_variable_set::set(const char_t* name, bool value)
+	{
+		xpath_variable* var = add(name, xpath_type_boolean);
+		return var ? var->set(value) : false;
+	}
+
+	PUGI__FN bool xpath_variable_set::set(const char_t* name, double value)
+	{
+		xpath_variable* var = add(name, xpath_type_number);
+		return var ? var->set(value) : false;
+	}
+
+	PUGI__FN bool xpath_variable_set::set(const char_t* name, const char_t* value)
+	{
+		xpath_variable* var = add(name, xpath_type_string);
+		return var ? var->set(value) : false;
+	}
+
+	PUGI__FN bool xpath_variable_set::set(const char_t* name, const xpath_node_set& value)
+	{
+		xpath_variable* var = add(name, xpath_type_node_set);
+		return var ? var->set(value) : false;
+	}
+
+	PUGI__FN xpath_variable* xpath_variable_set::get(const char_t* name)
+	{
+		return _find(name);
+	}
+
+	PUGI__FN const xpath_variable* xpath_variable_set::get(const char_t* name) const
+	{
+		return _find(name);
+	}
+
+	PUGI__FN xpath_query::xpath_query(const char_t* query, xpath_variable_set* variables): _impl(0)
+	{
+		impl::xpath_query_impl* qimpl = impl::xpath_query_impl::create();
+
+		if (!qimpl)
+		{
+		#ifdef PUGIXML_NO_EXCEPTIONS
+			_result.error = "Out of memory";
+		#else
+			throw std::bad_alloc();
+		#endif
+		}
+		else
+		{
+			using impl::auto_deleter; // MSVC7 workaround
+			auto_deleter<impl::xpath_query_impl> impl(qimpl, impl::xpath_query_impl::destroy);
+
+			qimpl->root = impl::xpath_parser::parse(query, variables, &qimpl->alloc, &_result);
+
+			if (qimpl->root)
+			{
+				qimpl->root->optimize(&qimpl->alloc);
+
+				_impl = impl.release();
+				_result.error = 0;
+			}
+		}
+	}
+
+	PUGI__FN xpath_query::xpath_query(): _impl(0)
+	{
+	}
+
+	PUGI__FN xpath_query::~xpath_query()
+	{
+		if (_impl)
+			impl::xpath_query_impl::destroy(static_cast<impl::xpath_query_impl*>(_impl));
+	}
+
+#if __cplusplus >= 201103
+	PUGI__FN xpath_query::xpath_query(xpath_query&& rhs)
+	{
+		_impl = rhs._impl;
+		rhs._impl = 0;
+	}
+
+	xpath_query& xpath_query::operator=(xpath_query&& rhs)
+	{
+		if (this == &rhs) return *this;
+
+		if (_impl)
+			impl::xpath_query_impl::destroy(static_cast<impl::xpath_query_impl*>(_impl));
+
+		_impl = rhs._impl;
+		rhs._impl = 0;
+
+		return *this;
+	}
+#endif
+
+	PUGI__FN xpath_value_type xpath_query::return_type() const
+	{
+		if (!_impl) return xpath_type_none;
+
+		return static_cast<impl::xpath_query_impl*>(_impl)->root->rettype();
+	}
+
+	PUGI__FN bool xpath_query::evaluate_boolean(const xpath_node& n) const
+	{
+		if (!_impl) return false;
+		
+		impl::xpath_context c(n, 1, 1);
+		impl::xpath_stack_data sd;
+
+	#ifdef PUGIXML_NO_EXCEPTIONS
+		if (setjmp(sd.error_handler)) return false;
+	#endif
+		
+		return static_cast<impl::xpath_query_impl*>(_impl)->root->eval_boolean(c, sd.stack);
+	}
+	
+	PUGI__FN double xpath_query::evaluate_number(const xpath_node& n) const
+	{
+		if (!_impl) return impl::gen_nan();
+		
+		impl::xpath_context c(n, 1, 1);
+		impl::xpath_stack_data sd;
+
+	#ifdef PUGIXML_NO_EXCEPTIONS
+		if (setjmp(sd.error_handler)) return impl::gen_nan();
+	#endif
+
+		return static_cast<impl::xpath_query_impl*>(_impl)->root->eval_number(c, sd.stack);
+	}
+
+#ifndef PUGIXML_NO_STL
+	PUGI__FN string_t xpath_query::evaluate_string(const xpath_node& n) const
+	{
+		impl::xpath_stack_data sd;
+
+		impl::xpath_string r = impl::evaluate_string_impl(static_cast<impl::xpath_query_impl*>(_impl), n, sd);
+
+		return string_t(r.c_str(), r.length());
+	}
+#endif
+
+	PUGI__FN size_t xpath_query::evaluate_string(char_t* buffer, size_t capacity, const xpath_node& n) const
+	{
+		impl::xpath_stack_data sd;
+
+		impl::xpath_string r = impl::evaluate_string_impl(static_cast<impl::xpath_query_impl*>(_impl), n, sd);
+
+		size_t full_size = r.length() + 1;
+		
+		if (capacity > 0)
+		{
+			size_t size = (full_size < capacity) ? full_size : capacity;
+			assert(size > 0);
+
+			memcpy(buffer, r.c_str(), (size - 1) * sizeof(char_t));
+			buffer[size - 1] = 0;
+		}
+		
+		return full_size;
+	}
+
+	PUGI__FN xpath_node_set xpath_query::evaluate_node_set(const xpath_node& n) const
+	{
+		impl::xpath_ast_node* root = impl::evaluate_node_set_prepare(static_cast<impl::xpath_query_impl*>(_impl));
+		if (!root) return xpath_node_set();
+
+		impl::xpath_context c(n, 1, 1);
+		impl::xpath_stack_data sd;
+
+	#ifdef PUGIXML_NO_EXCEPTIONS
+		if (setjmp(sd.error_handler)) return xpath_node_set();
+	#endif
+
+		impl::xpath_node_set_raw r = root->eval_node_set(c, sd.stack, impl::nodeset_eval_all);
+
+		return xpath_node_set(r.begin(), r.end(), r.type());
+	}
+
+	PUGI__FN xpath_node xpath_query::evaluate_node(const xpath_node& n) const
+	{
+		impl::xpath_ast_node* root = impl::evaluate_node_set_prepare(static_cast<impl::xpath_query_impl*>(_impl));
+		if (!root) return xpath_node();
+
+		impl::xpath_context c(n, 1, 1);
+		impl::xpath_stack_data sd;
+
+	#ifdef PUGIXML_NO_EXCEPTIONS
+		if (setjmp(sd.error_handler)) return xpath_node();
+	#endif
+
+		impl::xpath_node_set_raw r = root->eval_node_set(c, sd.stack, impl::nodeset_eval_first);
+
+		return r.first();
+	}
+
+	PUGI__FN const xpath_parse_result& xpath_query::result() const
+	{
+		return _result;
+	}
+
+	PUGI__FN static void unspecified_bool_xpath_query(xpath_query***)
+	{
+	}
+
+	PUGI__FN xpath_query::operator xpath_query::unspecified_bool_type() const
+	{
+		return _impl ? unspecified_bool_xpath_query : 0;
+	}
+
+	PUGI__FN bool xpath_query::operator!() const
+	{
+		return !_impl;
+	}
+
+	PUGI__FN xpath_node xml_node::select_node(const char_t* query, xpath_variable_set* variables) const
+	{
+		xpath_query q(query, variables);
+		return select_node(q);
+	}
+
+	PUGI__FN xpath_node xml_node::select_node(const xpath_query& query) const
+	{
+		return query.evaluate_node(*this);
+	}
+
+	PUGI__FN xpath_node_set xml_node::select_nodes(const char_t* query, xpath_variable_set* variables) const
+	{
+		xpath_query q(query, variables);
+		return select_nodes(q);
+	}
+
+	PUGI__FN xpath_node_set xml_node::select_nodes(const xpath_query& query) const
+	{
+		return query.evaluate_node_set(*this);
+	}
+
+	PUGI__FN xpath_node xml_node::select_single_node(const char_t* query, xpath_variable_set* variables) const
+	{
+		xpath_query q(query, variables);
+		return select_single_node(q);
+	}
+
+	PUGI__FN xpath_node xml_node::select_single_node(const xpath_query& query) const
+	{
+		return query.evaluate_node(*this);
+	}
+}
+
+#endif
+
+#ifdef __BORLANDC__
+#	pragma option pop
+#endif
+
+// Intel C++ does not properly keep warning state for function templates,
+// so popping warning state at the end of translation unit leads to warnings in the middle.
+#if defined(_MSC_VER) && !defined(__INTEL_COMPILER)
+#	pragma warning(pop)
+#endif
+
+// Undefine all local macros (makes sure we're not leaking macros in header-only mode)
+#undef PUGI__NO_INLINE
+#undef PUGI__UNLIKELY
+#undef PUGI__STATIC_ASSERT
+#undef PUGI__DMC_VOLATILE
+#undef PUGI__MSVC_CRT_VERSION
+#undef PUGI__NS_BEGIN
+#undef PUGI__NS_END
+#undef PUGI__FN
+#undef PUGI__FN_NO_INLINE
+#undef PUGI__GETPAGE_IMPL
+#undef PUGI__GETPAGE
+#undef PUGI__NODETYPE
+#undef PUGI__IS_CHARTYPE_IMPL
+#undef PUGI__IS_CHARTYPE
+#undef PUGI__IS_CHARTYPEX
+#undef PUGI__ENDSWITH
+#undef PUGI__SKIPWS
+#undef PUGI__OPTSET
+#undef PUGI__PUSHNODE
+#undef PUGI__POPNODE
+#undef PUGI__SCANFOR
+#undef PUGI__SCANWHILE
+#undef PUGI__SCANWHILE_UNROLL
+#undef PUGI__ENDSEG
+#undef PUGI__THROW_ERROR
+#undef PUGI__CHECK_ERROR
+
+#endif
+
+/**
+ * Copyright (c) 2006-2015 Arseny Kapoulkine
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use,
+ * copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following
+ * conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
+ * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
+ * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
diff --git a/lib/pugixml/pugixml.hpp b/lib/pugixml/pugixml.hpp
new file mode 100644
index 00000000..cdd24b6d
--- /dev/null
+++ b/lib/pugixml/pugixml.hpp
@@ -0,0 +1,1400 @@
+/**
+ * pugixml parser - version 1.6
+ * --------------------------------------------------------
+ * Copyright (C) 2006-2015, by Arseny Kapoulkine (arseny.kapoulkine@gmail.com)
+ * Report bugs and download new versions at http://pugixml.org/
+ *
+ * This library is distributed under the MIT License. See notice at the end
+ * of this file.
+ *
+ * This work is based on the pugxml parser, which is:
+ * Copyright (C) 2003, by Kristen Wegner (kristen@tima.net)
+ */
+
+#ifndef PUGIXML_VERSION
+// Define version macro; evaluates to major * 100 + minor so that it's safe to use in less-than comparisons
+#	define PUGIXML_VERSION 160
+#endif
+
+// Include user configuration file (this can define various configuration macros)
+#include "pugiconfig.hpp"
+
+#ifndef HEADER_PUGIXML_HPP
+#define HEADER_PUGIXML_HPP
+
+// Include stddef.h for size_t and ptrdiff_t
+#include <stddef.h>
+
+// Include exception header for XPath
+#if !defined(PUGIXML_NO_XPATH) && !defined(PUGIXML_NO_EXCEPTIONS)
+#	include <exception>
+#endif
+
+// Include STL headers
+#ifndef PUGIXML_NO_STL
+#	include <iterator>
+#	include <iosfwd>
+#	include <string>
+#endif
+
+// Macro for deprecated features
+#ifndef PUGIXML_DEPRECATED
+#	if defined(__GNUC__)
+#		define PUGIXML_DEPRECATED __attribute__((deprecated))
+#	elif defined(_MSC_VER) && _MSC_VER >= 1300
+#		define PUGIXML_DEPRECATED __declspec(deprecated)
+#	else
+#		define PUGIXML_DEPRECATED
+#	endif
+#endif
+
+// If no API is defined, assume default
+#ifndef PUGIXML_API
+#	define PUGIXML_API
+#endif
+
+// If no API for classes is defined, assume default
+#ifndef PUGIXML_CLASS
+#	define PUGIXML_CLASS PUGIXML_API
+#endif
+
+// If no API for functions is defined, assume default
+#ifndef PUGIXML_FUNCTION
+#	define PUGIXML_FUNCTION PUGIXML_API
+#endif
+
+// If the platform is known to have long long support, enable long long functions
+#ifndef PUGIXML_HAS_LONG_LONG
+#	if __cplusplus >= 201103
+#		define PUGIXML_HAS_LONG_LONG
+#	elif defined(_MSC_VER) && _MSC_VER >= 1400
+#		define PUGIXML_HAS_LONG_LONG
+#	endif
+#endif
+
+// Character interface macros
+#ifdef PUGIXML_WCHAR_MODE
+#	define PUGIXML_TEXT(t) L ## t
+#	define PUGIXML_CHAR wchar_t
+#else
+#	define PUGIXML_TEXT(t) t
+#	define PUGIXML_CHAR char
+#endif
+
+namespace pugi
+{
+	// Character type used for all internal storage and operations; depends on PUGIXML_WCHAR_MODE
+	typedef PUGIXML_CHAR char_t;
+
+#ifndef PUGIXML_NO_STL
+	// String type used for operations that work with STL string; depends on PUGIXML_WCHAR_MODE
+	typedef std::basic_string<PUGIXML_CHAR, std::char_traits<PUGIXML_CHAR>, std::allocator<PUGIXML_CHAR> > string_t;
+#endif
+}
+
+// The PugiXML namespace
+namespace pugi
+{
+	// Tree node types
+	enum xml_node_type
+	{
+		node_null,			// Empty (null) node handle
+		node_document,		// A document tree's absolute root
+		node_element,		// Element tag, i.e. '<node/>'
+		node_pcdata,		// Plain character data, i.e. 'text'
+		node_cdata,			// Character data, i.e. '<![CDATA[text]]>'
+		node_comment,		// Comment tag, i.e. '<!-- text -->'
+		node_pi,			// Processing instruction, i.e. '<?name?>'
+		node_declaration,	// Document declaration, i.e. '<?xml version="1.0"?>'
+		node_doctype		// Document type declaration, i.e. '<!DOCTYPE doc>'
+	};
+
+	// Parsing options
+
+	// Minimal parsing mode (equivalent to turning all other flags off).
+	// Only elements and PCDATA sections are added to the DOM tree, no text conversions are performed.
+	const unsigned int parse_minimal = 0x0000;
+
+	// This flag determines if processing instructions (node_pi) are added to the DOM tree. This flag is off by default.
+	const unsigned int parse_pi = 0x0001;
+
+	// This flag determines if comments (node_comment) are added to the DOM tree. This flag is off by default.
+	const unsigned int parse_comments = 0x0002;
+
+	// This flag determines if CDATA sections (node_cdata) are added to the DOM tree. This flag is on by default.
+	const unsigned int parse_cdata = 0x0004;
+
+	// This flag determines if plain character data (node_pcdata) that consist only of whitespace are added to the DOM tree.
+	// This flag is off by default; turning it on usually results in slower parsing and more memory consumption.
+	const unsigned int parse_ws_pcdata = 0x0008;
+
+	// This flag determines if character and entity references are expanded during parsing. This flag is on by default.
+	const unsigned int parse_escapes = 0x0010;
+
+	// This flag determines if EOL characters are normalized (converted to #xA) during parsing. This flag is on by default.
+	const unsigned int parse_eol = 0x0020;
+	
+	// This flag determines if attribute values are normalized using CDATA normalization rules during parsing. This flag is on by default.
+	const unsigned int parse_wconv_attribute = 0x0040;
+
+	// This flag determines if attribute values are normalized using NMTOKENS normalization rules during parsing. This flag is off by default.
+	const unsigned int parse_wnorm_attribute = 0x0080;
+	
+	// This flag determines if document declaration (node_declaration) is added to the DOM tree. This flag is off by default.
+	const unsigned int parse_declaration = 0x0100;
+
+	// This flag determines if document type declaration (node_doctype) is added to the DOM tree. This flag is off by default.
+	const unsigned int parse_doctype = 0x0200;
+
+	// This flag determines if plain character data (node_pcdata) that is the only child of the parent node and that consists only
+	// of whitespace is added to the DOM tree.
+	// This flag is off by default; turning it on may result in slower parsing and more memory consumption.
+	const unsigned int parse_ws_pcdata_single = 0x0400;
+
+	// This flag determines if leading and trailing whitespace is to be removed from plain character data. This flag is off by default.
+	const unsigned int parse_trim_pcdata = 0x0800;
+
+	// This flag determines if plain character data that does not have a parent node is added to the DOM tree, and if an empty document
+	// is a valid document. This flag is off by default.
+	const unsigned int parse_fragment = 0x1000;
+
+	// The default parsing mode.
+	// Elements, PCDATA and CDATA sections are added to the DOM tree, character/reference entities are expanded,
+	// End-of-Line characters are normalized, attribute values are normalized using CDATA normalization rules.
+	const unsigned int parse_default = parse_cdata | parse_escapes | parse_wconv_attribute | parse_eol;
+
+	// The full parsing mode.
+	// Nodes of all types are added to the DOM tree, character/reference entities are expanded,
+	// End-of-Line characters are normalized, attribute values are normalized using CDATA normalization rules.
+	const unsigned int parse_full = parse_default | parse_pi | parse_comments | parse_declaration | parse_doctype;
+
+	// These flags determine the encoding of input data for XML document
+	enum xml_encoding
+	{
+		encoding_auto,		// Auto-detect input encoding using BOM or < / <? detection; use UTF8 if BOM is not found
+		encoding_utf8,		// UTF8 encoding
+		encoding_utf16_le,	// Little-endian UTF16
+		encoding_utf16_be,	// Big-endian UTF16
+		encoding_utf16,		// UTF16 with native endianness
+		encoding_utf32_le,	// Little-endian UTF32
+		encoding_utf32_be,	// Big-endian UTF32
+		encoding_utf32,		// UTF32 with native endianness
+		encoding_wchar,		// The same encoding wchar_t has (either UTF16 or UTF32)
+		encoding_latin1
+	};
+
+	// Formatting flags
+	
+	// Indent the nodes that are written to output stream with as many indentation strings as deep the node is in DOM tree. This flag is on by default.
+	const unsigned int format_indent = 0x01;
+	
+	// Write encoding-specific BOM to the output stream. This flag is off by default.
+	const unsigned int format_write_bom = 0x02;
+
+	// Use raw output mode (no indentation and no line breaks are written). This flag is off by default.
+	const unsigned int format_raw = 0x04;
+	
+	// Omit default XML declaration even if there is no declaration in the document. This flag is off by default.
+	const unsigned int format_no_declaration = 0x08;
+
+	// Don't escape attribute values and PCDATA contents. This flag is off by default.
+	const unsigned int format_no_escapes = 0x10;
+
+	// Open file using text mode in xml_document::save_file. This enables special character (i.e. new-line) conversions on some systems. This flag is off by default.
+	const unsigned int format_save_file_text = 0x20;
+
+	// Write every attribute on a new line with appropriate indentation. This flag is off by default.
+	const unsigned int format_indent_attributes = 0x40;
+
+	// The default set of formatting flags.
+	// Nodes are indented depending on their depth in DOM tree, a default declaration is output if document has none.
+	const unsigned int format_default = format_indent;
+
+	// Forward declarations
+	struct xml_attribute_struct;
+	struct xml_node_struct;
+
+	class xml_node_iterator;
+	class xml_attribute_iterator;
+	class xml_named_node_iterator;
+
+	class xml_tree_walker;
+
+	struct xml_parse_result;
+
+	class xml_node;
+
+	class xml_text;
+	
+	#ifndef PUGIXML_NO_XPATH
+	class xpath_node;
+	class xpath_node_set;
+	class xpath_query;
+	class xpath_variable_set;
+	#endif
+
+	// Range-based for loop support
+	template <typename It> class xml_object_range
+	{
+	public:
+		typedef It const_iterator;
+		typedef It iterator;
+
+		xml_object_range(It b, It e): _begin(b), _end(e)
+		{
+		}
+
+		It begin() const { return _begin; }
+		It end() const { return _end; }
+
+	private:
+		It _begin, _end;
+	};
+
+	// Writer interface for node printing (see xml_node::print)
+	class PUGIXML_CLASS xml_writer
+	{
+	public:
+		virtual ~xml_writer() {}
+
+		// Write memory chunk into stream/file/whatever
+		virtual void write(const void* data, size_t size) = 0;
+	};
+
+	// xml_writer implementation for FILE*
+	class PUGIXML_CLASS xml_writer_file: public xml_writer
+	{
+	public:
+		// Construct writer from a FILE* object; void* is used to avoid header dependencies on stdio
+		xml_writer_file(void* file);
+
+		virtual void write(const void* data, size_t size);
+
+	private:
+		void* file;
+	};
+
+	#ifndef PUGIXML_NO_STL
+	// xml_writer implementation for streams
+	class PUGIXML_CLASS xml_writer_stream: public xml_writer
+	{
+	public:
+		// Construct writer from an output stream object
+		xml_writer_stream(std::basic_ostream<char, std::char_traits<char> >& stream);
+		xml_writer_stream(std::basic_ostream<wchar_t, std::char_traits<wchar_t> >& stream);
+
+		virtual void write(const void* data, size_t size);
+
+	private:
+		std::basic_ostream<char, std::char_traits<char> >* narrow_stream;
+		std::basic_ostream<wchar_t, std::char_traits<wchar_t> >* wide_stream;
+	};
+	#endif
+
+	// A light-weight handle for manipulating attributes in DOM tree
+	class PUGIXML_CLASS xml_attribute
+	{
+		friend class xml_attribute_iterator;
+		friend class xml_node;
+
+	private:
+		xml_attribute_struct* _attr;
+	
+		typedef void (*unspecified_bool_type)(xml_attribute***);
+
+	public:
+		// Default constructor. Constructs an empty attribute.
+		xml_attribute();
+		
+		// Constructs attribute from internal pointer
+		explicit xml_attribute(xml_attribute_struct* attr);
+
+		// Safe bool conversion operator
+		operator unspecified_bool_type() const;
+
+		// Borland C++ workaround
+		bool operator!() const;
+
+		// Comparison operators (compares wrapped attribute pointers)
+		bool operator==(const xml_attribute& r) const;
+		bool operator!=(const xml_attribute& r) const;
+		bool operator<(const xml_attribute& r) const;
+		bool operator>(const xml_attribute& r) const;
+		bool operator<=(const xml_attribute& r) const;
+		bool operator>=(const xml_attribute& r) const;
+
+		// Check if attribute is empty
+		bool empty() const;
+
+		// Get attribute name/value, or "" if attribute is empty
+		const char_t* name() const;
+		const char_t* value() const;
+
+		// Get attribute value, or the default value if attribute is empty
+		const char_t* as_string(const char_t* def = PUGIXML_TEXT("")) const;
+
+		// Get attribute value as a number, or the default value if conversion did not succeed or attribute is empty
+		int as_int(int def = 0) const;
+		unsigned int as_uint(unsigned int def = 0) const;
+		double as_double(double def = 0) const;
+		float as_float(float def = 0) const;
+
+	#ifdef PUGIXML_HAS_LONG_LONG
+		long long as_llong(long long def = 0) const;
+		unsigned long long as_ullong(unsigned long long def = 0) const;
+	#endif
+
+		// Get attribute value as bool (returns true if first character is in '1tTyY' set), or the default value if attribute is empty
+		bool as_bool(bool def = false) const;
+
+		// Set attribute name/value (returns false if attribute is empty or there is not enough memory)
+		bool set_name(const char_t* rhs);
+		bool set_value(const char_t* rhs);
+
+		// Set attribute value with type conversion (numbers are converted to strings, boolean is converted to "true"/"false")
+		bool set_value(int rhs);
+		bool set_value(unsigned int rhs);
+		bool set_value(double rhs);
+		bool set_value(float rhs);
+		bool set_value(bool rhs);
+
+	#ifdef PUGIXML_HAS_LONG_LONG
+		bool set_value(long long rhs);
+		bool set_value(unsigned long long rhs);
+	#endif
+
+		// Set attribute value (equivalent to set_value without error checking)
+		xml_attribute& operator=(const char_t* rhs);
+		xml_attribute& operator=(int rhs);
+		xml_attribute& operator=(unsigned int rhs);
+		xml_attribute& operator=(double rhs);
+		xml_attribute& operator=(float rhs);
+		xml_attribute& operator=(bool rhs);
+
+	#ifdef PUGIXML_HAS_LONG_LONG
+		xml_attribute& operator=(long long rhs);
+		xml_attribute& operator=(unsigned long long rhs);
+	#endif
+
+		// Get next/previous attribute in the attribute list of the parent node
+		xml_attribute next_attribute() const;
+		xml_attribute previous_attribute() const;
+
+		// Get hash value (unique for handles to the same object)
+		size_t hash_value() const;
+
+		// Get internal pointer
+		xml_attribute_struct* internal_object() const;
+	};
+
+#ifdef __BORLANDC__
+	// Borland C++ workaround
+	bool PUGIXML_FUNCTION operator&&(const xml_attribute& lhs, bool rhs);
+	bool PUGIXML_FUNCTION operator||(const xml_attribute& lhs, bool rhs);
+#endif
+
+	// A light-weight handle for manipulating nodes in DOM tree
+	class PUGIXML_CLASS xml_node
+	{
+		friend class xml_attribute_iterator;
+		friend class xml_node_iterator;
+		friend class xml_named_node_iterator;
+
+	protected:
+		xml_node_struct* _root;
+
+		typedef void (*unspecified_bool_type)(xml_node***);
+
+	public:
+		// Default constructor. Constructs an empty node.
+		xml_node();
+
+		// Constructs node from internal pointer
+		explicit xml_node(xml_node_struct* p);
+
+		// Safe bool conversion operator
+		operator unspecified_bool_type() const;
+
+		// Borland C++ workaround
+		bool operator!() const;
+	
+		// Comparison operators (compares wrapped node pointers)
+		bool operator==(const xml_node& r) const;
+		bool operator!=(const xml_node& r) const;
+		bool operator<(const xml_node& r) const;
+		bool operator>(const xml_node& r) const;
+		bool operator<=(const xml_node& r) const;
+		bool operator>=(const xml_node& r) const;
+
+		// Check if node is empty.
+		bool empty() const;
+
+		// Get node type
+		xml_node_type type() const;
+
+		// Get node name, or "" if node is empty or it has no name
+		const char_t* name() const;
+
+		// Get node value, or "" if node is empty or it has no value
+		// Note: For <node>text</node> node.value() does not return "text"! Use child_value() or text() methods to access text inside nodes.
+		const char_t* value() const;
+	
+		// Get attribute list
+		xml_attribute first_attribute() const;
+		xml_attribute last_attribute() const;
+
+		// Get children list
+		xml_node first_child() const;
+		xml_node last_child() const;
+
+		// Get next/previous sibling in the children list of the parent node
+		xml_node next_sibling() const;
+		xml_node previous_sibling() const;
+		
+		// Get parent node
+		xml_node parent() const;
+
+		// Get root of DOM tree this node belongs to
+		xml_node root() const;
+
+		// Get text object for the current node
+		xml_text text() const;
+
+		// Get child, attribute or next/previous sibling with the specified name
+		xml_node child(const char_t* name) const;
+		xml_attribute attribute(const char_t* name) const;
+		xml_node next_sibling(const char_t* name) const;
+		xml_node previous_sibling(const char_t* name) const;
+
+		// Get attribute, starting the search from a hint (and updating hint so that searching for a sequence of attributes is fast)
+		xml_attribute attribute(const char_t* name, xml_attribute& hint) const;
+
+		// Get child value of current node; that is, value of the first child node of type PCDATA/CDATA
+		const char_t* child_value() const;
+
+		// Get child value of child with specified name. Equivalent to child(name).child_value().
+		const char_t* child_value(const char_t* name) const;
+
+		// Set node name/value (returns false if node is empty, there is not enough memory, or node can not have name/value)
+		bool set_name(const char_t* rhs);
+		bool set_value(const char_t* rhs);
+		
+		// Add attribute with specified name. Returns added attribute, or empty attribute on errors.
+		xml_attribute append_attribute(const char_t* name);
+		xml_attribute prepend_attribute(const char_t* name);
+		xml_attribute insert_attribute_after(const char_t* name, const xml_attribute& attr);
+		xml_attribute insert_attribute_before(const char_t* name, const xml_attribute& attr);
+
+		// Add a copy of the specified attribute. Returns added attribute, or empty attribute on errors.
+		xml_attribute append_copy(const xml_attribute& proto);
+		xml_attribute prepend_copy(const xml_attribute& proto);
+		xml_attribute insert_copy_after(const xml_attribute& proto, const xml_attribute& attr);
+		xml_attribute insert_copy_before(const xml_attribute& proto, const xml_attribute& attr);
+
+		// Add child node with specified type. Returns added node, or empty node on errors.
+		xml_node append_child(xml_node_type type = node_element);
+		xml_node prepend_child(xml_node_type type = node_element);
+		xml_node insert_child_after(xml_node_type type, const xml_node& node);
+		xml_node insert_child_before(xml_node_type type, const xml_node& node);
+
+		// Add child element with specified name. Returns added node, or empty node on errors.
+		xml_node append_child(const char_t* name);
+		xml_node prepend_child(const char_t* name);
+		xml_node insert_child_after(const char_t* name, const xml_node& node);
+		xml_node insert_child_before(const char_t* name, const xml_node& node);
+
+		// Add a copy of the specified node as a child. Returns added node, or empty node on errors.
+		xml_node append_copy(const xml_node& proto);
+		xml_node prepend_copy(const xml_node& proto);
+		xml_node insert_copy_after(const xml_node& proto, const xml_node& node);
+		xml_node insert_copy_before(const xml_node& proto, const xml_node& node);
+
+		// Move the specified node to become a child of this node. Returns moved node, or empty node on errors.
+		xml_node append_move(const xml_node& moved);
+		xml_node prepend_move(const xml_node& moved);
+		xml_node insert_move_after(const xml_node& moved, const xml_node& node);
+		xml_node insert_move_before(const xml_node& moved, const xml_node& node);
+
+		// Remove specified attribute
+		bool remove_attribute(const xml_attribute& a);
+		bool remove_attribute(const char_t* name);
+
+		// Remove specified child
+		bool remove_child(const xml_node& n);
+		bool remove_child(const char_t* name);
+
+		// Parses buffer as an XML document fragment and appends all nodes as children of the current node.
+		// Copies/converts the buffer, so it may be deleted or changed after the function returns.
+		// Note: append_buffer allocates memory that has the lifetime of the owning document; removing the appended nodes does not immediately reclaim that memory.
+		xml_parse_result append_buffer(const void* contents, size_t size, unsigned int options = parse_default, xml_encoding encoding = encoding_auto);
+
+		// Find attribute using predicate. Returns first attribute for which predicate returned true.
+		template <typename Predicate> xml_attribute find_attribute(Predicate pred) const
+		{
+			if (!_root) return xml_attribute();
+			
+			for (xml_attribute attrib = first_attribute(); attrib; attrib = attrib.next_attribute())
+				if (pred(attrib))
+					return attrib;
+		
+			return xml_attribute();
+		}
+
+		// Find child node using predicate. Returns first child for which predicate returned true.
+		template <typename Predicate> xml_node find_child(Predicate pred) const
+		{
+			if (!_root) return xml_node();
+	
+			for (xml_node node = first_child(); node; node = node.next_sibling())
+				if (pred(node))
+					return node;
+		
+			return xml_node();
+		}
+
+		// Find node from subtree using predicate. Returns first node from subtree (depth-first), for which predicate returned true.
+		template <typename Predicate> xml_node find_node(Predicate pred) const
+		{
+			if (!_root) return xml_node();
+
+			xml_node cur = first_child();
+			
+			while (cur._root && cur._root != _root)
+			{
+				if (pred(cur)) return cur;
+
+				if (cur.first_child()) cur = cur.first_child();
+				else if (cur.next_sibling()) cur = cur.next_sibling();
+				else
+				{
+					while (!cur.next_sibling() && cur._root != _root) cur = cur.parent();
+
+					if (cur._root != _root) cur = cur.next_sibling();
+				}
+			}
+
+			return xml_node();
+		}
+
+		// Find child node by attribute name/value
+		xml_node find_child_by_attribute(const char_t* name, const char_t* attr_name, const char_t* attr_value) const;
+		xml_node find_child_by_attribute(const char_t* attr_name, const char_t* attr_value) const;
+
+	#ifndef PUGIXML_NO_STL
+		// Get the absolute node path from root as a text string.
+		string_t path(char_t delimiter = '/') const;
+	#endif
+
+		// Search for a node by path consisting of node names and . or .. elements.
+		xml_node first_element_by_path(const char_t* path, char_t delimiter = '/') const;
+
+		// Recursively traverse subtree with xml_tree_walker
+		bool traverse(xml_tree_walker& walker);
+	
+	#ifndef PUGIXML_NO_XPATH
+		// Select single node by evaluating XPath query. Returns first node from the resulting node set.
+		xpath_node select_node(const char_t* query, xpath_variable_set* variables = 0) const;
+		xpath_node select_node(const xpath_query& query) const;
+
+		// Select node set by evaluating XPath query
+		xpath_node_set select_nodes(const char_t* query, xpath_variable_set* variables = 0) const;
+		xpath_node_set select_nodes(const xpath_query& query) const;
+
+		// (deprecated: use select_node instead) Select single node by evaluating XPath query.
+		xpath_node select_single_node(const char_t* query, xpath_variable_set* variables = 0) const;
+		xpath_node select_single_node(const xpath_query& query) const;
+
+	#endif
+		
+		// Print subtree using a writer object
+		void print(xml_writer& writer, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, xml_encoding encoding = encoding_auto, unsigned int depth = 0) const;
+
+	#ifndef PUGIXML_NO_STL
+		// Print subtree to stream
+		void print(std::basic_ostream<char, std::char_traits<char> >& os, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, xml_encoding encoding = encoding_auto, unsigned int depth = 0) const;
+		void print(std::basic_ostream<wchar_t, std::char_traits<wchar_t> >& os, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, unsigned int depth = 0) const;
+	#endif
+
+		// Child nodes iterators
+		typedef xml_node_iterator iterator;
+
+		iterator begin() const;
+		iterator end() const;
+
+		// Attribute iterators
+		typedef xml_attribute_iterator attribute_iterator;
+
+		attribute_iterator attributes_begin() const;
+		attribute_iterator attributes_end() const;
+
+		// Range-based for support
+		xml_object_range<xml_node_iterator> children() const;
+		xml_object_range<xml_named_node_iterator> children(const char_t* name) const;
+		xml_object_range<xml_attribute_iterator> attributes() const;
+
+		// Get node offset in parsed file/string (in char_t units) for debugging purposes
+		ptrdiff_t offset_debug() const;
+
+		// Get hash value (unique for handles to the same object)
+		size_t hash_value() const;
+
+		// Get internal pointer
+		xml_node_struct* internal_object() const;
+	};
+
+#ifdef __BORLANDC__
+	// Borland C++ workaround
+	bool PUGIXML_FUNCTION operator&&(const xml_node& lhs, bool rhs);
+	bool PUGIXML_FUNCTION operator||(const xml_node& lhs, bool rhs);
+#endif
+
+	// A helper for working with text inside PCDATA nodes
+	class PUGIXML_CLASS xml_text
+	{
+		friend class xml_node;
+
+		xml_node_struct* _root;
+
+		typedef void (*unspecified_bool_type)(xml_text***);
+
+		explicit xml_text(xml_node_struct* root);
+
+		xml_node_struct* _data_new();
+		xml_node_struct* _data() const;
+
+	public:
+		// Default constructor. Constructs an empty object.
+		xml_text();
+
+		// Safe bool conversion operator
+		operator unspecified_bool_type() const;
+
+		// Borland C++ workaround
+		bool operator!() const;
+
+		// Check if text object is empty
+		bool empty() const;
+
+		// Get text, or "" if object is empty
+		const char_t* get() const;
+
+		// Get text, or the default value if object is empty
+		const char_t* as_string(const char_t* def = PUGIXML_TEXT("")) const;
+
+		// Get text as a number, or the default value if conversion did not succeed or object is empty
+		int as_int(int def = 0) const;
+		unsigned int as_uint(unsigned int def = 0) const;
+		double as_double(double def = 0) const;
+		float as_float(float def = 0) const;
+
+	#ifdef PUGIXML_HAS_LONG_LONG
+		long long as_llong(long long def = 0) const;
+		unsigned long long as_ullong(unsigned long long def = 0) const;
+	#endif
+
+		// Get text as bool (returns true if first character is in '1tTyY' set), or the default value if object is empty
+		bool as_bool(bool def = false) const;
+
+		// Set text (returns false if object is empty or there is not enough memory)
+		bool set(const char_t* rhs);
+
+		// Set text with type conversion (numbers are converted to strings, boolean is converted to "true"/"false")
+		bool set(int rhs);
+		bool set(unsigned int rhs);
+		bool set(double rhs);
+		bool set(float rhs);
+		bool set(bool rhs);
+
+	#ifdef PUGIXML_HAS_LONG_LONG
+		bool set(long long rhs);
+		bool set(unsigned long long rhs);
+	#endif
+
+		// Set text (equivalent to set without error checking)
+		xml_text& operator=(const char_t* rhs);
+		xml_text& operator=(int rhs);
+		xml_text& operator=(unsigned int rhs);
+		xml_text& operator=(double rhs);
+		xml_text& operator=(float rhs);
+		xml_text& operator=(bool rhs);
+
+	#ifdef PUGIXML_HAS_LONG_LONG
+		xml_text& operator=(long long rhs);
+		xml_text& operator=(unsigned long long rhs);
+	#endif
+
+		// Get the data node (node_pcdata or node_cdata) for this object
+		xml_node data() const;
+	};
+
+#ifdef __BORLANDC__
+	// Borland C++ workaround
+	bool PUGIXML_FUNCTION operator&&(const xml_text& lhs, bool rhs);
+	bool PUGIXML_FUNCTION operator||(const xml_text& lhs, bool rhs);
+#endif
+
+	// Child node iterator (a bidirectional iterator over a collection of xml_node)
+	class PUGIXML_CLASS xml_node_iterator
+	{
+		friend class xml_node;
+
+	private:
+		mutable xml_node _wrap;
+		xml_node _parent;
+
+		xml_node_iterator(xml_node_struct* ref, xml_node_struct* parent);
+
+	public:
+		// Iterator traits
+		typedef ptrdiff_t difference_type;
+		typedef xml_node value_type;
+		typedef xml_node* pointer;
+		typedef xml_node& reference;
+
+	#ifndef PUGIXML_NO_STL
+		typedef std::bidirectional_iterator_tag iterator_category;
+	#endif
+
+		// Default constructor
+		xml_node_iterator();
+
+		// Construct an iterator which points to the specified node
+		xml_node_iterator(const xml_node& node);
+
+		// Iterator operators
+		bool operator==(const xml_node_iterator& rhs) const;
+		bool operator!=(const xml_node_iterator& rhs) const;
+
+		xml_node& operator*() const;
+		xml_node* operator->() const;
+
+		const xml_node_iterator& operator++();
+		xml_node_iterator operator++(int);
+
+		const xml_node_iterator& operator--();
+		xml_node_iterator operator--(int);
+	};
+
+	// Attribute iterator (a bidirectional iterator over a collection of xml_attribute)
+	class PUGIXML_CLASS xml_attribute_iterator
+	{
+		friend class xml_node;
+
+	private:
+		mutable xml_attribute _wrap;
+		xml_node _parent;
+
+		xml_attribute_iterator(xml_attribute_struct* ref, xml_node_struct* parent);
+
+	public:
+		// Iterator traits
+		typedef ptrdiff_t difference_type;
+		typedef xml_attribute value_type;
+		typedef xml_attribute* pointer;
+		typedef xml_attribute& reference;
+
+	#ifndef PUGIXML_NO_STL
+		typedef std::bidirectional_iterator_tag iterator_category;
+	#endif
+
+		// Default constructor
+		xml_attribute_iterator();
+
+		// Construct an iterator which points to the specified attribute
+		xml_attribute_iterator(const xml_attribute& attr, const xml_node& parent);
+
+		// Iterator operators
+		bool operator==(const xml_attribute_iterator& rhs) const;
+		bool operator!=(const xml_attribute_iterator& rhs) const;
+
+		xml_attribute& operator*() const;
+		xml_attribute* operator->() const;
+
+		const xml_attribute_iterator& operator++();
+		xml_attribute_iterator operator++(int);
+
+		const xml_attribute_iterator& operator--();
+		xml_attribute_iterator operator--(int);
+	};
+
+	// Named node range helper
+	class PUGIXML_CLASS xml_named_node_iterator
+	{
+		friend class xml_node;
+
+	public:
+		// Iterator traits
+		typedef ptrdiff_t difference_type;
+		typedef xml_node value_type;
+		typedef xml_node* pointer;
+		typedef xml_node& reference;
+
+	#ifndef PUGIXML_NO_STL
+		typedef std::bidirectional_iterator_tag iterator_category;
+	#endif
+
+		// Default constructor
+		xml_named_node_iterator();
+
+		// Construct an iterator which points to the specified node
+		xml_named_node_iterator(const xml_node& node, const char_t* name);
+
+		// Iterator operators
+		bool operator==(const xml_named_node_iterator& rhs) const;
+		bool operator!=(const xml_named_node_iterator& rhs) const;
+
+		xml_node& operator*() const;
+		xml_node* operator->() const;
+
+		const xml_named_node_iterator& operator++();
+		xml_named_node_iterator operator++(int);
+
+		const xml_named_node_iterator& operator--();
+		xml_named_node_iterator operator--(int);
+
+	private:
+		mutable xml_node _wrap;
+		xml_node _parent;
+		const char_t* _name;
+
+		xml_named_node_iterator(xml_node_struct* ref, xml_node_struct* parent, const char_t* name);
+	};
+
+	// Abstract tree walker class (see xml_node::traverse)
+	class PUGIXML_CLASS xml_tree_walker
+	{
+		friend class xml_node;
+
+	private:
+		int _depth;
+	
+	protected:
+		// Get current traversal depth
+		int depth() const;
+	
+	public:
+		xml_tree_walker();
+		virtual ~xml_tree_walker();
+
+		// Callback that is called when traversal begins
+		virtual bool begin(xml_node& node);
+
+		// Callback that is called for each node traversed
+		virtual bool for_each(xml_node& node) = 0;
+
+		// Callback that is called when traversal ends
+		virtual bool end(xml_node& node);
+	};
+
+	// Parsing status, returned as part of xml_parse_result object
+	enum xml_parse_status
+	{
+		status_ok = 0,				// No error
+
+		status_file_not_found,		// File was not found during load_file()
+		status_io_error,			// Error reading from file/stream
+		status_out_of_memory,		// Could not allocate memory
+		status_internal_error,		// Internal error occurred
+
+		status_unrecognized_tag,	// Parser could not determine tag type
+
+		status_bad_pi,				// Parsing error occurred while parsing document declaration/processing instruction
+		status_bad_comment,			// Parsing error occurred while parsing comment
+		status_bad_cdata,			// Parsing error occurred while parsing CDATA section
+		status_bad_doctype,			// Parsing error occurred while parsing document type declaration
+		status_bad_pcdata,			// Parsing error occurred while parsing PCDATA section
+		status_bad_start_element,	// Parsing error occurred while parsing start element tag
+		status_bad_attribute,		// Parsing error occurred while parsing element attribute
+		status_bad_end_element,		// Parsing error occurred while parsing end element tag
+		status_end_element_mismatch,// There was a mismatch of start-end tags (closing tag had incorrect name, some tag was not closed or there was an excessive closing tag)
+
+		status_append_invalid_root,	// Unable to append nodes since root type is not node_element or node_document (exclusive to xml_node::append_buffer)
+
+		status_no_document_element	// Parsing resulted in a document without element nodes
+	};
+
+	// Parsing result
+	struct PUGIXML_CLASS xml_parse_result
+	{
+		// Parsing status (see xml_parse_status)
+		xml_parse_status status;
+
+		// Last parsed offset (in char_t units from start of input data)
+		ptrdiff_t offset;
+
+		// Source document encoding
+		xml_encoding encoding;
+
+		// Default constructor, initializes object to failed state
+		xml_parse_result();
+
+		// Cast to bool operator
+		operator bool() const;
+
+		// Get error description
+		const char* description() const;
+	};
+
+	// Document class (DOM tree root)
+	class PUGIXML_CLASS xml_document: public xml_node
+	{
+	private:
+		char_t* _buffer;
+
+		char _memory[192];
+		
+		// Non-copyable semantics
+		xml_document(const xml_document&);
+		const xml_document& operator=(const xml_document&);
+
+		void create();
+		void destroy();
+
+	public:
+		// Default constructor, makes empty document
+		xml_document();
+
+		// Destructor, invalidates all node/attribute handles to this document
+		~xml_document();
+
+		// Removes all nodes, leaving the empty document
+		void reset();
+
+		// Removes all nodes, then copies the entire contents of the specified document
+		void reset(const xml_document& proto);
+
+	#ifndef PUGIXML_NO_STL
+		// Load document from stream.
+		xml_parse_result load(std::basic_istream<char, std::char_traits<char> >& stream, unsigned int options = parse_default, xml_encoding encoding = encoding_auto);
+		xml_parse_result load(std::basic_istream<wchar_t, std::char_traits<wchar_t> >& stream, unsigned int options = parse_default);
+	#endif
+
+		// (deprecated: use load_string instead) Load document from zero-terminated string. No encoding conversions are applied.
+		xml_parse_result load(const char_t* contents, unsigned int options = parse_default);
+
+		// Load document from zero-terminated string. No encoding conversions are applied.
+		xml_parse_result load_string(const char_t* contents, unsigned int options = parse_default);
+
+		// Load document from file
+		xml_parse_result load_file(const char* path, unsigned int options = parse_default, xml_encoding encoding = encoding_auto);
+		xml_parse_result load_file(const wchar_t* path, unsigned int options = parse_default, xml_encoding encoding = encoding_auto);
+
+		// Load document from buffer. Copies/converts the buffer, so it may be deleted or changed after the function returns.
+		xml_parse_result load_buffer(const void* contents, size_t size, unsigned int options = parse_default, xml_encoding encoding = encoding_auto);
+
+		// Load document from buffer, using the buffer for in-place parsing (the buffer is modified and used for storage of document data).
+		// You should ensure that buffer data will persist throughout the document's lifetime, and free the buffer memory manually once document is destroyed.
+		xml_parse_result load_buffer_inplace(void* contents, size_t size, unsigned int options = parse_default, xml_encoding encoding = encoding_auto);
+
+		// Load document from buffer, using the buffer for in-place parsing (the buffer is modified and used for storage of document data).
+		// You should allocate the buffer with pugixml allocation function; document will free the buffer when it is no longer needed (you can't use it anymore).
+		xml_parse_result load_buffer_inplace_own(void* contents, size_t size, unsigned int options = parse_default, xml_encoding encoding = encoding_auto);
+
+		// Save XML document to writer (semantics is slightly different from xml_node::print, see documentation for details).
+		void save(xml_writer& writer, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, xml_encoding encoding = encoding_auto) const;
+
+	#ifndef PUGIXML_NO_STL
+		// Save XML document to stream (semantics is slightly different from xml_node::print, see documentation for details).
+		void save(std::basic_ostream<char, std::char_traits<char> >& stream, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, xml_encoding encoding = encoding_auto) const;
+		void save(std::basic_ostream<wchar_t, std::char_traits<wchar_t> >& stream, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default) const;
+	#endif
+
+		// Save XML to file
+		bool save_file(const char* path, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, xml_encoding encoding = encoding_auto) const;
+		bool save_file(const wchar_t* path, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, xml_encoding encoding = encoding_auto) const;
+
+		// Get document element
+		xml_node document_element() const;
+	};
+
+#ifndef PUGIXML_NO_XPATH
+	// XPath query return type
+	enum xpath_value_type
+	{
+		xpath_type_none,	  // Unknown type (query failed to compile)
+		xpath_type_node_set,  // Node set (xpath_node_set)
+		xpath_type_number,	  // Number
+		xpath_type_string,	  // String
+		xpath_type_boolean	  // Boolean
+	};
+
+	// XPath parsing result
+	struct PUGIXML_CLASS xpath_parse_result
+	{
+		// Error message (0 if no error)
+		const char* error;
+
+		// Last parsed offset (in char_t units from string start)
+		ptrdiff_t offset;
+
+		// Default constructor, initializes object to failed state
+		xpath_parse_result();
+
+		// Cast to bool operator
+		operator bool() const;
+
+		// Get error description
+		const char* description() const;
+	};
+
+	// A single XPath variable
+	class PUGIXML_CLASS xpath_variable
+	{
+		friend class xpath_variable_set;
+
+	protected:
+		xpath_value_type _type;
+		xpath_variable* _next;
+
+		xpath_variable(xpath_value_type type);
+
+		// Non-copyable semantics
+		xpath_variable(const xpath_variable&);
+		xpath_variable& operator=(const xpath_variable&);
+		
+	public:
+		// Get variable name
+		const char_t* name() const;
+
+		// Get variable type
+		xpath_value_type type() const;
+
+		// Get variable value; no type conversion is performed, default value (false, NaN, empty string, empty node set) is returned on type mismatch error
+		bool get_boolean() const;
+		double get_number() const;
+		const char_t* get_string() const;
+		const xpath_node_set& get_node_set() const;
+
+		// Set variable value; no type conversion is performed, false is returned on type mismatch error
+		bool set(bool value);
+		bool set(double value);
+		bool set(const char_t* value);
+		bool set(const xpath_node_set& value);
+	};
+
+	// A set of XPath variables
+	class PUGIXML_CLASS xpath_variable_set
+	{
+	private:
+		xpath_variable* _data[64];
+
+		void _assign(const xpath_variable_set& rhs);
+		void _swap(xpath_variable_set& rhs);
+
+		xpath_variable* _find(const char_t* name) const;
+
+		static bool _clone(xpath_variable* var, xpath_variable** out_result);
+		static void _destroy(xpath_variable* var);
+
+	public:
+		// Default constructor/destructor
+		xpath_variable_set();
+		~xpath_variable_set();
+
+		// Copy constructor/assignment operator
+		xpath_variable_set(const xpath_variable_set& rhs);
+		xpath_variable_set& operator=(const xpath_variable_set& rhs);
+
+	#if __cplusplus >= 201103
+		// Move semantics support
+		xpath_variable_set(xpath_variable_set&& rhs);
+		xpath_variable_set& operator=(xpath_variable_set&& rhs);
+	#endif
+
+		// Add a new variable or get the existing one, if the types match
+		xpath_variable* add(const char_t* name, xpath_value_type type);
+
+		// Set value of an existing variable; no type conversion is performed, false is returned if there is no such variable or if types mismatch
+		bool set(const char_t* name, bool value);
+		bool set(const char_t* name, double value);
+		bool set(const char_t* name, const char_t* value);
+		bool set(const char_t* name, const xpath_node_set& value);
+
+		// Get existing variable by name
+		xpath_variable* get(const char_t* name);
+		const xpath_variable* get(const char_t* name) const;
+	};
+
+	// A compiled XPath query object
+	class PUGIXML_CLASS xpath_query
+	{
+	private:
+		void* _impl;
+		xpath_parse_result _result;
+
+		typedef void (*unspecified_bool_type)(xpath_query***);
+
+		// Non-copyable semantics
+		xpath_query(const xpath_query&);
+		xpath_query& operator=(const xpath_query&);
+
+	public:
+		// Construct a compiled object from XPath expression.
+		// If PUGIXML_NO_EXCEPTIONS is not defined, throws xpath_exception on compilation errors.
+		explicit xpath_query(const char_t* query, xpath_variable_set* variables = 0);
+
+		// Constructor
+		xpath_query();
+
+		// Destructor
+		~xpath_query();
+
+	#if __cplusplus >= 201103
+		// Move semantics support
+		xpath_query(xpath_query&& rhs);
+		xpath_query& operator=(xpath_query&& rhs);
+	#endif
+
+		// Get query expression return type
+		xpath_value_type return_type() const;
+		
+		// Evaluate expression as boolean value in the specified context; performs type conversion if necessary.
+		// If PUGIXML_NO_EXCEPTIONS is not defined, throws std::bad_alloc on out of memory errors.
+		bool evaluate_boolean(const xpath_node& n) const;
+		
+		// Evaluate expression as double value in the specified context; performs type conversion if necessary.
+		// If PUGIXML_NO_EXCEPTIONS is not defined, throws std::bad_alloc on out of memory errors.
+		double evaluate_number(const xpath_node& n) const;
+		
+	#ifndef PUGIXML_NO_STL
+		// Evaluate expression as string value in the specified context; performs type conversion if necessary.
+		// If PUGIXML_NO_EXCEPTIONS is not defined, throws std::bad_alloc on out of memory errors.
+		string_t evaluate_string(const xpath_node& n) const;
+	#endif
+		
+		// Evaluate expression as string value in the specified context; performs type conversion if necessary.
+		// At most capacity characters are written to the destination buffer, full result size is returned (includes terminating zero).
+		// If PUGIXML_NO_EXCEPTIONS is not defined, throws std::bad_alloc on out of memory errors.
+		// If PUGIXML_NO_EXCEPTIONS is defined, returns empty  set instead.
+		size_t evaluate_string(char_t* buffer, size_t capacity, const xpath_node& n) const;
+
+		// Evaluate expression as node set in the specified context.
+		// If PUGIXML_NO_EXCEPTIONS is not defined, throws xpath_exception on type mismatch and std::bad_alloc on out of memory errors.
+		// If PUGIXML_NO_EXCEPTIONS is defined, returns empty node set instead.
+		xpath_node_set evaluate_node_set(const xpath_node& n) const;
+
+		// Evaluate expression as node set in the specified context.
+		// Return first node in document order, or empty node if node set is empty.
+		// If PUGIXML_NO_EXCEPTIONS is not defined, throws xpath_exception on type mismatch and std::bad_alloc on out of memory errors.
+		// If PUGIXML_NO_EXCEPTIONS is defined, returns empty node instead.
+		xpath_node evaluate_node(const xpath_node& n) const;
+
+		// Get parsing result (used to get compilation errors in PUGIXML_NO_EXCEPTIONS mode)
+		const xpath_parse_result& result() const;
+
+		// Safe bool conversion operator
+		operator unspecified_bool_type() const;
+
+		// Borland C++ workaround
+		bool operator!() const;
+	};
+	
+	#ifndef PUGIXML_NO_EXCEPTIONS
+	// XPath exception class
+	class PUGIXML_CLASS xpath_exception: public std::exception
+	{
+	private:
+		xpath_parse_result _result;
+
+	public:
+		// Construct exception from parse result
+		explicit xpath_exception(const xpath_parse_result& result);
+
+		// Get error message
+		virtual const char* what() const throw();
+
+		// Get parse result
+		const xpath_parse_result& result() const;
+	};
+	#endif
+	
+	// XPath node class (either xml_node or xml_attribute)
+	class PUGIXML_CLASS xpath_node
+	{
+	private:
+		xml_node _node;
+		xml_attribute _attribute;
+	
+		typedef void (*unspecified_bool_type)(xpath_node***);
+
+	public:
+		// Default constructor; constructs empty XPath node
+		xpath_node();
+		
+		// Construct XPath node from XML node/attribute
+		xpath_node(const xml_node& node);
+		xpath_node(const xml_attribute& attribute, const xml_node& parent);
+
+		// Get node/attribute, if any
+		xml_node node() const;
+		xml_attribute attribute() const;
+		
+		// Get parent of contained node/attribute
+		xml_node parent() const;
+
+		// Safe bool conversion operator
+		operator unspecified_bool_type() const;
+		
+		// Borland C++ workaround
+		bool operator!() const;
+
+		// Comparison operators
+		bool operator==(const xpath_node& n) const;
+		bool operator!=(const xpath_node& n) const;
+	};
+
+#ifdef __BORLANDC__
+	// Borland C++ workaround
+	bool PUGIXML_FUNCTION operator&&(const xpath_node& lhs, bool rhs);
+	bool PUGIXML_FUNCTION operator||(const xpath_node& lhs, bool rhs);
+#endif
+
+	// A fixed-size collection of XPath nodes
+	class PUGIXML_CLASS xpath_node_set
+	{
+	public:
+		// Collection type
+		enum type_t
+		{
+			type_unsorted,			// Not ordered
+			type_sorted,			// Sorted by document order (ascending)
+			type_sorted_reverse		// Sorted by document order (descending)
+		};
+		
+		// Constant iterator type
+		typedef const xpath_node* const_iterator;
+
+		// We define non-constant iterator to be the same as constant iterator so that various generic algorithms (i.e. boost foreach) work
+		typedef const xpath_node* iterator;
+	
+		// Default constructor. Constructs empty set.
+		xpath_node_set();
+
+		// Constructs a set from iterator range; data is not checked for duplicates and is not sorted according to provided type, so be careful
+		xpath_node_set(const_iterator begin, const_iterator end, type_t type = type_unsorted);
+
+		// Destructor
+		~xpath_node_set();
+		
+		// Copy constructor/assignment operator
+		xpath_node_set(const xpath_node_set& ns);
+		xpath_node_set& operator=(const xpath_node_set& ns);
+
+	#if __cplusplus >= 201103
+		// Move semantics support
+		xpath_node_set(xpath_node_set&& rhs);
+		xpath_node_set& operator=(xpath_node_set&& rhs);
+	#endif
+
+		// Get collection type
+		type_t type() const;
+		
+		// Get collection size
+		size_t size() const;
+
+		// Indexing operator
+		const xpath_node& operator[](size_t index) const;
+		
+		// Collection iterators
+		const_iterator begin() const;
+		const_iterator end() const;
+
+		// Sort the collection in ascending/descending order by document order
+		void sort(bool reverse = false);
+		
+		// Get first node in the collection by document order
+		xpath_node first() const;
+		
+		// Check if collection is empty
+		bool empty() const;
+	
+	private:
+		type_t _type;
+		
+		xpath_node _storage;
+		
+		xpath_node* _begin;
+		xpath_node* _end;
+
+		void _assign(const_iterator begin, const_iterator end, type_t type);
+		void _move(xpath_node_set& rhs);
+	};
+#endif
+
+#ifndef PUGIXML_NO_STL
+	// Convert wide string to UTF8
+	std::basic_string<char, std::char_traits<char>, std::allocator<char> > PUGIXML_FUNCTION as_utf8(const wchar_t* str);
+	std::basic_string<char, std::char_traits<char>, std::allocator<char> > PUGIXML_FUNCTION as_utf8(const std::basic_string<wchar_t, std::char_traits<wchar_t>, std::allocator<wchar_t> >& str);
+	
+	// Convert UTF8 to wide string
+	std::basic_string<wchar_t, std::char_traits<wchar_t>, std::allocator<wchar_t> > PUGIXML_FUNCTION as_wide(const char* str);
+	std::basic_string<wchar_t, std::char_traits<wchar_t>, std::allocator<wchar_t> > PUGIXML_FUNCTION as_wide(const std::basic_string<char, std::char_traits<char>, std::allocator<char> >& str);
+#endif
+
+	// Memory allocation function interface; returns pointer to allocated memory or NULL on failure
+	typedef void* (*allocation_function)(size_t size);
+	
+	// Memory deallocation function interface
+	typedef void (*deallocation_function)(void* ptr);
+
+	// Override default memory management functions. All subsequent allocations/deallocations will be performed via supplied functions.
+	void PUGIXML_FUNCTION set_memory_management_functions(allocation_function allocate, deallocation_function deallocate);
+	
+	// Get current memory management functions
+	allocation_function PUGIXML_FUNCTION get_memory_allocation_function();
+	deallocation_function PUGIXML_FUNCTION get_memory_deallocation_function();
+}
+
+#if !defined(PUGIXML_NO_STL) && (defined(_MSC_VER) || defined(__ICC))
+namespace std
+{
+	// Workarounds for (non-standard) iterator category detection for older versions (MSVC7/IC8 and earlier)
+	std::bidirectional_iterator_tag PUGIXML_FUNCTION _Iter_cat(const pugi::xml_node_iterator&);
+	std::bidirectional_iterator_tag PUGIXML_FUNCTION _Iter_cat(const pugi::xml_attribute_iterator&);
+	std::bidirectional_iterator_tag PUGIXML_FUNCTION _Iter_cat(const pugi::xml_named_node_iterator&);
+}
+#endif
+
+#if !defined(PUGIXML_NO_STL) && defined(__SUNPRO_CC)
+namespace std
+{
+	// Workarounds for (non-standard) iterator category detection
+	std::bidirectional_iterator_tag PUGIXML_FUNCTION __iterator_category(const pugi::xml_node_iterator&);
+	std::bidirectional_iterator_tag PUGIXML_FUNCTION __iterator_category(const pugi::xml_attribute_iterator&);
+	std::bidirectional_iterator_tag PUGIXML_FUNCTION __iterator_category(const pugi::xml_named_node_iterator&);
+}
+#endif
+
+#endif
+
+// Make sure implementation is included in header-only mode
+// Use macro expansion in #include to work around QMake (QTBUG-11923)
+#if defined(PUGIXML_HEADER_ONLY) && !defined(PUGIXML_SOURCE)
+#	define PUGIXML_SOURCE "pugixml.cpp"
+#	include PUGIXML_SOURCE
+#endif
+
+/**
+ * Copyright (c) 2006-2015 Arseny Kapoulkine
+ *
+ * Permission is hereby granted, free of charge, to any person
+ * obtaining a copy of this software and associated documentation
+ * files (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use,
+ * copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following
+ * conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
+ * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
+ * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+ * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
diff --git a/lib/pugixml/readme.txt b/lib/pugixml/readme.txt
new file mode 100644
index 00000000..faa41d37
--- /dev/null
+++ b/lib/pugixml/readme.txt
@@ -0,0 +1,52 @@
+pugixml 1.6 - an XML processing library
+
+Copyright (C) 2006-2015, by Arseny Kapoulkine (arseny.kapoulkine@gmail.com)
+Report bugs and download new versions at http://pugixml.org/
+
+This is the distribution of pugixml, which is a C++ XML processing library,
+which consists of a DOM-like interface with rich traversal/modification
+capabilities, an extremely fast XML parser which constructs the DOM tree from
+an XML file/buffer, and an XPath 1.0 implementation for complex data-driven
+tree queries. Full Unicode support is also available, with Unicode interface
+variants and conversions between different Unicode encodings (which happen
+automatically during parsing/saving).
+
+The distribution contains the following folders:
+
+	contrib/ - various contributions to pugixml
+
+	docs/ - documentation
+		docs/samples - pugixml usage examples
+		docs/quickstart.html - quick start guide
+		docs/manual.html - complete manual
+
+	scripts/ - project files for IDE/build systems
+
+	src/ - header and source files
+
+	readme.txt - this file.
+
+This library is distributed under the MIT License:
+
+Copyright (c) 2006-2015 Arseny Kapoulkine
+
+Permission is hereby granted, free of charge, to any person
+obtaining a copy of this software and associated documentation
+files (the "Software"), to deal in the Software without
+restriction, including without limitation the rights to use,
+copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the
+Software is furnished to do so, subject to the following
+conditions:
+
+The above copyright notice and this permission notice shall be
+included in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
+OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
+HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+OTHER DEALINGS IN THE SOFTWARE.
diff --git a/lib/qcd/action/ActionBase.h b/lib/qcd/action/ActionBase.h
index 05edce86..df5ee50b 100644
--- a/lib/qcd/action/ActionBase.h
+++ b/lib/qcd/action/ActionBase.h
@@ -7,14 +7,30 @@ template<class GaugeField>
 class Action { 
 
  public:
-  virtual void  init(const GaugeField &U, GridParallelRNG& pRNG) = 0;
-  virtual RealD S(const GaugeField &U)                           = 0;  // evaluate the action
-  virtual void  deriv(const GaugeField &U,GaugeField & dSdU )    = 0;  // evaluate the action derivative
-  //virtual void  refresh(const GaugeField & ) {}                ; 
+  virtual void  init (const GaugeField &U, GridParallelRNG& pRNG) = 0;  // 
+  virtual RealD S    (const GaugeField &U)                        = 0;  // evaluate the action
+  virtual void  deriv(const GaugeField &U,GaugeField & dSdU )     = 0;  // evaluate the action derivative
+  virtual void  refresh(const GaugeField & ) {};                        // Default to no-op for actions with no internal fields
   // Boundary conditions?
   // Heatbath?
   virtual ~Action() {};
 };
 
+// Could derive PseudoFermion action with a PF field, FermionField, and a Grid; implement refresh
+template<class GaugeField, class FermionField>
+class PseudoFermionAction : public Action<GaugeField> {
+ public:
+  FermionField Phi;
+  GridParallelRNG &pRNG;
+  GridBase &Grid;
+
+  PseudoFermionAction(GridBase &_Grid,GridParallelRNG &_pRNG) : Grid(_Grid), Phi(&_Grid), pRNG(_pRNG) {
+  };
+
+  virtual void refresh(const GaugeField &gauge) {
+    gaussian(Phi,pRNG);
+  };
+
+};
 }}
 #endif
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.cc b/lib/qcd/action/fermion/WilsonFermion5D.cc
index 523934c4..0387a26e 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.cc
+++ b/lib/qcd/action/fermion/WilsonFermion5D.cc
@@ -149,6 +149,7 @@ PARALLEL_FOR_LOOP
     ////////////////////////////
     // spin trace outer product
     ////////////////////////////
+// FIXME : need to sum over fifth direction.
     tmp = TraceIndex<SpinIndex>(outerProduct(Btilde,A)); // ordering here
     PokeIndex<LorentzIndex>(mat,tmp,mu);
 
diff --git a/lib/qcd/hmc/HMC.cc b/lib/qcd/hmc/HMC.cc
index 9db2c604..a6130224 100644
--- a/lib/qcd/hmc/HMC.cc
+++ b/lib/qcd/hmc/HMC.cc
@@ -17,8 +17,5 @@ namespace Grid{
 	  
       }
 
-
-   
-
   }
 }
diff --git a/lib/qcd/hmc/HMC.h b/lib/qcd/hmc/HMC.h
index 21c6bb0d..10947719 100644
--- a/lib/qcd/hmc/HMC.h
+++ b/lib/qcd/hmc/HMC.h
@@ -27,13 +27,22 @@ namespace Grid{
     
     template <class Algorithm> 
     class HybridMonteCarlo{
+
       const HMCparameters Params;
-      GridSerialRNG sRNG;
+
+      GridSerialRNG sRNG; // Fixme: need a RNG management strategy.
+
       Integrator<Algorithm>& MD;
-      
+
+      /////////////////////////////////////////////////////////
+      // Metropolis step
+      /////////////////////////////////////////////////////////
       bool metropolis_test(const RealD DeltaH){
+
 	RealD rn_test;
+
 	RealD prob = std::exp(-DeltaH);
+
 	random(sRNG,rn_test);
       
 	std::cout<<GridLogMessage<< "--------------------------------------------\n";
@@ -47,14 +56,19 @@ namespace Grid{
 	  std::cout<<GridLogMessage <<"-- REJECTED\n";
 	  return false;
 	}
+
       }
 
+      /////////////////////////////////////////////////////////
+      // Evolution
+      /////////////////////////////////////////////////////////
       RealD evolve_step(LatticeLorentzColourMatrix& U){
 
 	MD.init(U); // set U and initialize P and phi's 
+
 	RealD H0 = MD.S(U); // initial state action  
 	std::cout<<GridLogMessage<<"Total H before = "<< H0 << "\n";
-      
+
 	MD.integrate(U);
       
 	RealD H1 = MD.S(U); // updated state action            
@@ -64,13 +78,17 @@ namespace Grid{
       }
       
     public:
-    HybridMonteCarlo(HMCparameters Pms, 
-		     Integrator<Algorithm>& MolDyn):
-      Params(Pms),MD(MolDyn){
-	//FIXME
-	// initialize RNGs also with seed
+
+      /////////////////////////////////////////
+      // Constructor
+      /////////////////////////////////////////
+      HybridMonteCarlo(HMCparameters Pms,  Integrator<Algorithm>& MolDyn): Params(Pms),MD(MolDyn) {
+
+	//FIXME...  initialize RNGs also with seed ; RNG management strategy
 	sRNG.SeedRandomDevice();
+
       }
+
       ~HybridMonteCarlo(){};
 
 
@@ -95,7 +113,8 @@ namespace Grid{
 	  DeltaH = evolve_step(Ucopy);
 		
 	  if(metropolis_test(DeltaH)) Uin = Ucopy;
-	  //need sync?
+	  //need sync? // Query Guido on this comment
+
 	}
       }
     };
diff --git a/tests/Make.inc b/tests/Make.inc
index b78b4d16..8b313ccc 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,5 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd Test_wilson_force Test_wilson_force_phiMphi
+bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd Test_wilson_force Test_wilson_force_phiMdagMphi Test_wilson_force_phiMphi
 
 
 Test_GaugeAction_SOURCES=Test_GaugeAction.cc
@@ -146,6 +146,10 @@ Test_wilson_force_SOURCES=Test_wilson_force.cc
 Test_wilson_force_LDADD=-lGrid
 
 
+Test_wilson_force_phiMdagMphi_SOURCES=Test_wilson_force_phiMdagMphi.cc
+Test_wilson_force_phiMdagMphi_LDADD=-lGrid
+
+
 Test_wilson_force_phiMphi_SOURCES=Test_wilson_force_phiMphi.cc
 Test_wilson_force_phiMphi_LDADD=-lGrid
 

From cc4ca48d1344fa6fa651b77635a7927dcd0c4301 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 29 Jul 2015 17:53:39 +0900
Subject: [PATCH 409/429] Two flavour HMC for Wilson/Wilson is conserving
 energy. Still to check plaq and <e(-dH)>, but nevertheless this is progress

---
 lib/Make.inc                                  |  2 +-
 lib/algorithms/iterative/ConjugateGradient.h  | 20 ++--
 lib/qcd/action/Actions.h                      |  6 ++
 lib/qcd/action/fermion/FermionOperator.h      |  4 +
 lib/qcd/action/fermion/WilsonFermion.cc       |  4 +
 lib/qcd/action/fermion/WilsonFermion.h        |  1 +
 lib/qcd/action/fermion/WilsonFermion5D.cc     |  5 +-
 lib/qcd/action/fermion/WilsonFermion5D.h      |  1 +
 lib/qcd/action/gauge/WilsonGaugeAction.h      |  6 +-
 lib/qcd/action/pseudofermion/TwoFlavour.h     | 73 ++++++++++----
 lib/qcd/hmc/HMC.cc                            |  4 +-
 lib/qcd/hmc/HMC.h                             |  1 +
 lib/qcd/hmc/integrators/Integrator.h          |  9 +-
 .../hmc/integrators/Integrator_algorithm.h    | 17 ++--
 lib/qcd/hmc/integrators/Integrator_base.h     | 12 +++
 lib/qcd/utils/SUn.h                           |  2 -
 lib/tensors/Tensor_Ta.h                       | 12 ++-
 tests/Make.inc                                |  6 +-
 tests/Test_hmc_WilsonFermionGauge.cc          | 55 +++++++++++
 tests/Test_wilson_force.cc                    | 95 ++++++++++++++++---
 20 files changed, 259 insertions(+), 76 deletions(-)
 create mode 100644 tests/Test_hmc_WilsonFermionGauge.cc

diff --git a/lib/Make.inc b/lib/Make.inc
index 92699dbe..409bf754 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/BfmHDCG.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/GeneralisedMinimumResidual.h ./algorithms/iterative/gmres.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Config.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./Init.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./Log.h ./MacroMagic.h ./Old/Tensor_peek.h ./Old/Tensor_poke.h ./parallelIO/NerscIO.h ./qcd/action/ActionBase.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/pseudofermion/TwoFlavour.h ./qcd/hmc/HMC.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/hmc/integrators/Integrator_base.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_index.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h ./Timer.h
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/BfmHDCG.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/GeneralisedMinimumResidual.h ./algorithms/iterative/gmres.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Config.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./Init.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./Log.h ./MacroMagic.h ./Old/Tensor_peek.h ./Old/Tensor_poke.h ./parallelIO/NerscIO.h ./parameterIO/XMLReader.h ./qcd/action/ActionBase.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/pseudofermion/TwoFlavour.h ./qcd/hmc/HMC.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/hmc/integrators/Integrator_base.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_index.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h ./Timer.h
 
 CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./Init.cc ./Log.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/hmc/HMC.cc ./qcd/hmc/integrators/Integrator.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/algorithms/iterative/ConjugateGradient.h b/lib/algorithms/iterative/ConjugateGradient.h
index 70b89f90..b919dd91 100644
--- a/lib/algorithms/iterative/ConjugateGradient.h
+++ b/lib/algorithms/iterative/ConjugateGradient.h
@@ -13,9 +13,7 @@ namespace Grid {
 public:                                                
     RealD   Tolerance;
     Integer MaxIterations;
-    int verbose;
     ConjugateGradient(RealD tol,Integer maxit) : Tolerance(tol), MaxIterations(maxit) { 
-      verbose=0;
     };
 
 
@@ -42,14 +40,12 @@ public:
       cp =a;
       ssq=norm2(src);
 
-      if ( verbose ) {
-	std::cout<<GridLogMessage <<std::setprecision(4)<< "ConjugateGradient: guess "<<guess<<std::endl;
-	std::cout<<GridLogMessage <<std::setprecision(4)<< "ConjugateGradient:   src "<<ssq  <<std::endl;
-	std::cout<<GridLogMessage <<std::setprecision(4)<< "ConjugateGradient:    mp "<<d    <<std::endl;
-	std::cout<<GridLogMessage <<std::setprecision(4)<< "ConjugateGradient:   mmp "<<b    <<std::endl;
-	std::cout<<GridLogMessage <<std::setprecision(4)<< "ConjugateGradient:  cp,r "<<cp   <<std::endl;
-	std::cout<<GridLogMessage <<std::setprecision(4)<< "ConjugateGradient:     p "<<a    <<std::endl;
-      }
+      std::cout<<GridLogIterative <<std::setprecision(4)<< "ConjugateGradient: guess "<<guess<<std::endl;
+      std::cout<<GridLogIterative <<std::setprecision(4)<< "ConjugateGradient:   src "<<ssq  <<std::endl;
+      std::cout<<GridLogIterative <<std::setprecision(4)<< "ConjugateGradient:    mp "<<d    <<std::endl;
+      std::cout<<GridLogIterative <<std::setprecision(4)<< "ConjugateGradient:   mmp "<<b    <<std::endl;
+      std::cout<<GridLogIterative <<std::setprecision(4)<< "ConjugateGradient:  cp,r "<<cp   <<std::endl;
+      std::cout<<GridLogIterative <<std::setprecision(4)<< "ConjugateGradient:     p "<<a    <<std::endl;
 
       RealD rsq =  Tolerance* Tolerance*ssq;
       
@@ -58,7 +54,7 @@ public:
 	return;
       }
       
-      if(verbose) std::cout<<GridLogMessage << std::setprecision(4)<< "ConjugateGradient: k=0 residual "<<cp<<" rsq"<<rsq<<std::endl;
+      std::cout<<GridLogIterative << std::setprecision(4)<< "ConjugateGradient: k=0 residual "<<cp<<" rsq"<<rsq<<std::endl;
       
       int k;
       for (k=1;k<=MaxIterations;k++){
@@ -80,7 +76,7 @@ public:
 	psi= a*p+psi;
 	p  = p*b+r;
 	  
-	if (verbose) std::cout<<GridLogMessage<<"ConjugateGradient: Iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
+	std::cout<<GridLogIterative<<"ConjugateGradient: Iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
 	
 	// Stopping condition
 	if ( cp <= rsq ) { 
diff --git a/lib/qcd/action/Actions.h b/lib/qcd/action/Actions.h
index 94809723..26f26b8f 100644
--- a/lib/qcd/action/Actions.h
+++ b/lib/qcd/action/Actions.h
@@ -79,4 +79,10 @@
 ///////////////////////////////////////////////////////////////////////////////
 #include <qcd/action/fermion/g5HermitianLinop.h>
 
+////////////////////////////////////////
+// Pseudo fermion combinations
+////////////////////////////////////////
+#include <qcd/action/pseudofermion/TwoFlavour.h>
+
+
 #endif
diff --git a/lib/qcd/action/fermion/FermionOperator.h b/lib/qcd/action/fermion/FermionOperator.h
index a8f10380..b6ba2d08 100644
--- a/lib/qcd/action/fermion/FermionOperator.h
+++ b/lib/qcd/action/fermion/FermionOperator.h
@@ -56,6 +56,10 @@ namespace Grid {
       virtual void  Mdiag  (const FermionField &in, FermionField &out) { Mooee(in,out);};   // Same as Mooee applied to both CB's
       virtual void  Mdir   (const FermionField &in, FermionField &out,int dir,int disp)=0;   // case by case Wilson, Clover, Cayley, ContFrac, PartFrac
 
+      ///////////////////////////////////////////////
+      // Updates gauge field during HMC
+      ///////////////////////////////////////////////
+      virtual void ImportGauge(const GaugeField & _U);
 
     };
 
diff --git a/lib/qcd/action/fermion/WilsonFermion.cc b/lib/qcd/action/fermion/WilsonFermion.cc
index afd7edd7..2c2132d1 100644
--- a/lib/qcd/action/fermion/WilsonFermion.cc
+++ b/lib/qcd/action/fermion/WilsonFermion.cc
@@ -24,6 +24,10 @@ WilsonFermion::WilsonFermion(LatticeGaugeField &_Umu,
 {
   // Allocate the required comms buffer
   comm_buf.resize(Stencil._unified_buffer_size); // this is always big enough to contain EO
+  ImportGauge(_Umu);
+}
+void WilsonFermion::ImportGauge(const LatticeGaugeField &_Umu)
+{
   DoubleStore(Umu,_Umu);
   pickCheckerboard(Even,UmuEven,Umu);
   pickCheckerboard(Odd ,UmuOdd,Umu);
diff --git a/lib/qcd/action/fermion/WilsonFermion.h b/lib/qcd/action/fermion/WilsonFermion.h
index 5422ca52..79d3e626 100644
--- a/lib/qcd/action/fermion/WilsonFermion.h
+++ b/lib/qcd/action/fermion/WilsonFermion.h
@@ -136,6 +136,7 @@ namespace Grid {
       WilsonFermion(LatticeGaugeField &_Umu,GridCartesian &Fgrid,GridRedBlackCartesian &Hgrid,RealD _mass);
 
       // DoubleStore
+      virtual void ImportGauge(const LatticeGaugeField &_Umu);
       void DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu);
 
       ///////////////////////////////////////////////////////////////
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.cc b/lib/qcd/action/fermion/WilsonFermion5D.cc
index 0387a26e..c3329503 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.cc
+++ b/lib/qcd/action/fermion/WilsonFermion5D.cc
@@ -65,7 +65,10 @@ namespace QCD {
 
   // Allocate the required comms buffer
   comm_buf.resize(Stencil._unified_buffer_size); // this is always big enough to contain EO
-  
+  ImportGauge(_Umu);
+}  
+void WilsonFermion5D::ImportGauge(const LatticeGaugeField &_Umu)
+{
   DoubleStore(Umu,_Umu);
   pickCheckerboard(Even,UmuEven,Umu);
   pickCheckerboard(Odd ,UmuOdd,Umu);
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.h b/lib/qcd/action/fermion/WilsonFermion5D.h
index 7ff5d1c3..d5cff3e6 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.h
+++ b/lib/qcd/action/fermion/WilsonFermion5D.h
@@ -94,6 +94,7 @@ namespace Grid {
 			  double _M5);
 
       // DoubleStore
+      virtual void ImportGauge(const LatticeGaugeField &_Umu);
       void DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu);
 
       ///////////////////////////////////////////////////////////////
diff --git a/lib/qcd/action/gauge/WilsonGaugeAction.h b/lib/qcd/action/gauge/WilsonGaugeAction.h
index fc31a5c3..f5f46070 100644
--- a/lib/qcd/action/gauge/WilsonGaugeAction.h
+++ b/lib/qcd/action/gauge/WilsonGaugeAction.h
@@ -19,8 +19,10 @@ namespace Grid{
       virtual RealD S(const GaugeField &U) {
 	RealD plaq = WilsonLoops<MatrixField,GaugeField>::avgPlaquette(U);
 	std::cout<<GridLogMessage << "Plaq : "<<plaq << "\n";
-	double vol = U._grid->gSites();
-	return beta*(1.0 -plaq)*(Nd*(Nd-1.0))*vol*0.5;
+	RealD vol = U._grid->gSites();
+	RealD action=beta*(1.0 -plaq)*(Nd*(Nd-1.0))*vol*0.5;
+	std::cout << GridLogMessage << "WilsonGauge action "<<action<<std::endl;
+	return action;
       };
       virtual void deriv(const GaugeField &U,GaugeField & dSdU) {
 
diff --git a/lib/qcd/action/pseudofermion/TwoFlavour.h b/lib/qcd/action/pseudofermion/TwoFlavour.h
index a65f4ee3..67f3101c 100644
--- a/lib/qcd/action/pseudofermion/TwoFlavour.h
+++ b/lib/qcd/action/pseudofermion/TwoFlavour.h
@@ -95,18 +95,18 @@ namespace Grid{
     ////////////////////////////////////////////////////////////////////////
     // Two flavour pseudofermion action for any dop
     ////////////////////////////////////////////////////////////////////////
-    template<class GaugeField,class MatrixField,class FermionField,class FermionOperator>
+    template<class GaugeField,class MatrixField,class FermionField>
       class TwoFlavourPseudoFermionAction : public Action<GaugeField> {
 
     private:
-
+      
       FermionOperator<FermionField,GaugeField> & FermOp;// the basic operator
 
       OperatorFunction<FermionField> &DerivativeSolver;
 
       OperatorFunction<FermionField> &ActionSolver;
 
-      GridBase *Grid;
+      GridBase &Grid;
 
       FermionField Phi; // the pseudo fermion field for this trajectory
 
@@ -114,10 +114,11 @@ namespace Grid{
       /////////////////////////////////////////////////
       // Pass in required objects.
       /////////////////////////////////////////////////
-      TwoFlavourPseudoFermionAction(FermionOperator &Op, 
-				    OperatorFunction<FermionField> & DS,
-				    OperatorFunction<FermionField> & AS
-				    ) : FermOp(Op), DerivativeSolver(DS), ActionSolver(AS) {
+    TwoFlavourPseudoFermionAction(FermionOperator<FermionField,GaugeField>  &Op, 
+				  OperatorFunction<FermionField> & DS,
+				  OperatorFunction<FermionField> & AS,
+				  GridBase &_Grid
+				  ) : FermOp(Op), DerivativeSolver(DS), ActionSolver(AS), Phi(&_Grid), Grid(_Grid) {
       };
       
       //////////////////////////////////////////////////////////////////////////////////////
@@ -125,9 +126,28 @@ namespace Grid{
       //////////////////////////////////////////////////////////////////////////////////////
       virtual void init(const GaugeField &U, GridParallelRNG& pRNG) {
 
-	// width? Must check
-	gaussian(Phi,pRNG);
+	// P(phi) = e^{- phi^dag (MdagM)^-1 phi}
+	// Phi = Mdag eta 
+	// P(eta) = e^{- eta^dag eta}
+	//
+	// e^{x^2/2 sig^2} => sig^2 = 0.5.
+	// 
+	// So eta should be of width sig = 1/sqrt(2).
+	// and must multiply by 0.707....
+	//
+	// Chroma has this scale factor: two_flavor_monomial_w.h
+	// IroIro: does not use this scale. It is absorbed by a change of vars
+	//         in the Phi integral, and thus is only an irrelevant prefactor for the partition function.
+	//
+	RealD scale = std::sqrt(0.5);
+	FermionField eta(&Grid);
 
+	gaussian(pRNG,eta);
+
+	FermOp.Mdag(eta,Phi);
+
+	Phi=Phi*scale;
+	
       };
 
       //////////////////////////////////////////////////////
@@ -135,38 +155,49 @@ namespace Grid{
       //////////////////////////////////////////////////////
       virtual RealD S(const GaugeField &U) {
 
-	FermionField X(Grid);
-	FermionField Y(Grid);
-	
-	MdagMLinearOperator<FermionOperator<FermionField,GaugeField>,FermionField> MdagMOp(FermOp);
+	FermOp.ImportGauge(U);
 
-	ActionSolver(MdagMop,Phi,X);
+	FermionField X(&Grid);
+	FermionField Y(&Grid);
+	
+	MdagMLinearOperator<FermionOperator<FermionField,GaugeField> ,FermionField> MdagMOp(FermOp);
+	X=zero;
+	ActionSolver(MdagMOp,Phi,X);
 	MdagMOp.Op(X,Y);
 
 	RealD action = norm2(Y);
-	
+	std::cout << GridLogMessage << "Pseudofermion action "<<action<<std::endl;
 	return action;
       };
 
       //////////////////////////////////////////////////////
       // dS/du = - phi^dag  (Mdag M)^-1 [ Mdag dM + dMdag M ]  (Mdag M)^-1 phi
+      //       = - phi^dag M^-1 dM (MdagM)^-1 phi -  phi^dag (MdagM)^-1 dMdag dM (Mdag)^-1 phi 
+      //
+      //       = - Ydag dM X  - Xdag dMdag Y
+      //
       //////////////////////////////////////////////////////
       virtual void deriv(const GaugeField &U,GaugeField & dSdU) {
 
-	FermionField X(Grid);
-	FermionField Y(Grid);
-	GaugeField   tmp(Grid);
+	FermOp.ImportGauge(U);
 
-	MdagMLinearOperator<FermionOperator<FermionField,GaugeField>,FermionField> MdagMOp(FermOp);
+	FermionField X(&Grid);
+	FermionField Y(&Grid);
+	GaugeField   tmp(&Grid);
 
-	DerivativeSolver(MdagMop,Phi,X);
+	MdagMLinearOperator<FermionOperator<FermionField,GaugeField> ,FermionField> MdagMOp(FermOp);
+
+	X=zero;
+	DerivativeSolver(MdagMOp,Phi,X);
 	MdagMOp.Op(X,Y);
 
 	// Our conventions really make this UdSdU; We do not differentiate wrt Udag here.
 	// So must take dSdU - adj(dSdU) and left multiply by mom to get dS/dt.
 
 	FermOp.MDeriv(tmp , Y, X,DaggerNo );  dSdU=tmp;
-	FermOp.MDeriv(tmp , X, Y,DaggerYes);  dSdU=-UdSdU-tmp;
+	FermOp.MDeriv(tmp , X, Y,DaggerYes);  dSdU=dSdU+tmp;
+	
+	dSdU = Ta(dSdU);
 
       };
 
diff --git a/lib/qcd/hmc/HMC.cc b/lib/qcd/hmc/HMC.cc
index a6130224..f7a5d0b8 100644
--- a/lib/qcd/hmc/HMC.cc
+++ b/lib/qcd/hmc/HMC.cc
@@ -7,8 +7,8 @@ namespace Grid{
 	// FIXME fill this constructor  now just default values
 	  
 	////////////////////////////// Default values
-	Nsweeps             = 100;
-	TotalSweeps         = 20;
+	Nsweeps             = 200;
+	TotalSweeps         = 220;
 	ThermalizationSteps = 20;
 	StartingConfig      = 0;
 	SaveInterval        = 1;
diff --git a/lib/qcd/hmc/HMC.h b/lib/qcd/hmc/HMC.h
index 10947719..bc4f47ff 100644
--- a/lib/qcd/hmc/HMC.h
+++ b/lib/qcd/hmc/HMC.h
@@ -73,6 +73,7 @@ namespace Grid{
       
 	RealD H1 = MD.S(U); // updated state action            
 	std::cout<<GridLogMessage<<"Total H after = "<< H1 << "\n";
+	std::cout<<GridLogMessage<<"DeltaH is "<< H1-H0 << "\n";
       
 	return (H1-H0);
       }
diff --git a/lib/qcd/hmc/integrators/Integrator.h b/lib/qcd/hmc/integrators/Integrator.h
index c0ed74ff..6b8eb77f 100644
--- a/lib/qcd/hmc/integrators/Integrator.h
+++ b/lib/qcd/hmc/integrators/Integrator.h
@@ -63,7 +63,6 @@ namespace Grid{
 	}
       }
 
-
       void update_U(LatticeLorentzColourMatrix&U, double ep){
 	//rewrite exponential to deal automatically  with the lorentz index?
 	LatticeColourMatrix Umu(U._grid);
@@ -77,7 +76,6 @@ namespace Grid{
 
       }
       
-
       
       friend void IntegratorAlgorithm::step (LatticeLorentzColourMatrix& U, 
 					     int level, std::vector<int>& clock,
@@ -92,9 +90,9 @@ namespace Grid{
       
       ~Integrator(){}
 
-
       //Initialization of momenta and actions
       void init(LatticeLorentzColourMatrix& U){
+
 	std::cout<<GridLogMessage<< "Integrator init\n";
 
 	MDutils::generate_momenta(*P,pRNG);
@@ -105,7 +103,6 @@ namespace Grid{
 	}
       }
 
-      
       // Calculate action
       RealD S(LatticeLorentzColourMatrix& U){
 	LatticeComplex Hloc(U._grid);
@@ -119,12 +116,14 @@ namespace Grid{
 	
 	RealD H = Hsum.real();
 
-	std::cout<<GridLogMessage << "H_p = "<< H << "\n";
+	std::cout<<GridLogMessage << "Momentum action H_p = "<< H << "\n";
 
 	// Actions
 	for(int level=0; level<as.size(); ++level)
 	  for(int actionID=0; actionID<as.at(level).size(); ++actionID)
 	    H += as[level].at(actionID)->S(U);
+
+	std::cout<<GridLogMessage << "Total action H = "<< H << "\n";
 	
 	return H;
       }
diff --git a/lib/qcd/hmc/integrators/Integrator_algorithm.h b/lib/qcd/hmc/integrators/Integrator_algorithm.h
index 0369021f..37c1cc84 100644
--- a/lib/qcd/hmc/integrators/Integrator_algorithm.h
+++ b/lib/qcd/hmc/integrators/Integrator_algorithm.h
@@ -32,8 +32,7 @@ namespace Grid{
 	int fin = Integ->Nrel[0];
 	for(int l=1; l<=level; ++l) fin*= 2.0*Integ->Nrel[l];
 	fin = 3*Integ->Params.MDsteps*fin -1;
-	
-	
+
 	for(int e=0; e<Integ->Nrel[level]; ++e){
 	  
 	  if(clock[level] == 0){    // initial half step 
@@ -45,7 +44,6 @@ namespace Grid{
 	  
 	  if(level == fl){          // lowest level 
 	    Integ->update_U(U,0.5*eps);
-	    
 	    for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
 	    std::cout<<GridLogMessage<<"U "<< (clock[level]+1) <<std::endl;
 	  }else{                 // recursive function call 
@@ -81,9 +79,7 @@ namespace Grid{
 	    std::cout<<GridLogMessage<<"P "<< clock[level] <<std::endl;
 	  }
 	}
-	
-	
-	
+		
       }
       
     };
@@ -93,6 +89,7 @@ namespace Grid{
       void step (LatticeLorentzColourMatrix& U, 
 		 int level, std::vector<int>& clock,
 		 Integrator<LeapFrog>* Integ){
+
 	// level  : current level
 	// fl     : final level
 	// eps    : current step size
@@ -115,6 +112,7 @@ namespace Grid{
 	    for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
 	    std::cout<<GridLogMessage<<"P "<< 0.5*clock[level] <<std::endl;
 	  }
+
 	  if(level == fl){          // lowest level
 	    Integ->update_U(U, eps);
 	    for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
@@ -122,24 +120,21 @@ namespace Grid{
 	  }else{                 // recursive function call
 	    step(U, level+1,clock, Integ);
 	  }
+
 	  if(clock[level] == fin){  // final half step
 	    Integ->update_P(U, level,eps/2.0);
-	    
 	    ++clock[level];
 	    for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
 	    std::cout<<GridLogMessage<<"P "<< 0.5*clock[level] <<std::endl;
 	  }else{                  // bulk step
 	    Integ->update_P(U, level,eps);
-	    
 	    clock[level]+=2;
 	    for(int l=0; l<level;++l) std::cout<<GridLogMessage<<"   ";
 	    std::cout<<GridLogMessage<<"P "<< 0.5*clock[level] <<std::endl;
 	  }
+
 	}
 
-
-
-
       }
     };
 
diff --git a/lib/qcd/hmc/integrators/Integrator_base.h b/lib/qcd/hmc/integrators/Integrator_base.h
index ef65f694..1cc32808 100644
--- a/lib/qcd/hmc/integrators/Integrator_base.h
+++ b/lib/qcd/hmc/integrators/Integrator_base.h
@@ -65,6 +65,18 @@ namespace Grid{
 	for(int a=0; a<as[level].size(); ++a){
 	  LatticeLorentzColourMatrix force(U._grid);
 	  as[level].at(a)->deriv(U,force);
+
+	  Complex dSdt=0.0;
+	  for(int mu=0;mu<Nd;mu++){
+	    LatticeColourMatrix forcemu(U._grid);
+	    LatticeColourMatrix mommu(U._grid);
+	    forcemu=PeekIndex<LorentzIndex>(force,mu);
+	    mommu=PeekIndex<LorentzIndex>(*P,mu);
+
+	    dSdt += sum(trace(forcemu*(*P)));
+
+	  }	  
+	  std::cout << GridLogMessage << " action "<<level<<","<<a<<" dSdt "<< dSdt << " dt "<<ep  <<std::endl;
 	  *P -= force*ep;
 	}
       }
diff --git a/lib/qcd/utils/SUn.h b/lib/qcd/utils/SUn.h
index 4ae69a0e..66e5f8ab 100644
--- a/lib/qcd/utils/SUn.h
+++ b/lib/qcd/utils/SUn.h
@@ -554,9 +554,7 @@ Note that in step D setting B ~ X - A and using B in place of A in step E will g
     for(int a=0;a<generators();a++){
       gaussian(pRNG,ca); 
       generator(a,ta);
-      
       la=toComplex(ca)*ci*ta;
-   
       out += la; 
     }
 
diff --git a/lib/tensors/Tensor_Ta.h b/lib/tensors/Tensor_Ta.h
index 724cbd6e..e71e54fd 100644
--- a/lib/tensors/Tensor_Ta.h
+++ b/lib/tensors/Tensor_Ta.h
@@ -7,11 +7,12 @@ namespace Grid {
   /////////////////////////////////////////////// 
   // Ta function for scalar, vector, matrix
   /////////////////////////////////////////////// 
+  /*
   inline ComplexF Ta( const ComplexF &arg){    return arg;}
   inline ComplexD Ta( const ComplexD &arg){    return arg;}
   inline RealF Ta( const RealF &arg){    return arg;}
   inline RealD Ta( const RealD &arg){    return arg;}
-
+  */
 
   template<class vtype> inline iScalar<vtype> Ta(const iScalar<vtype>&r)
     {
@@ -29,10 +30,11 @@ namespace Grid {
     }
   template<class vtype,int N> inline iMatrix<vtype,N> Ta(const iMatrix<vtype,N> &arg)
     {
-      iMatrix<vtype,N> ret(arg);
-      double factor = (1/(double)N);
-      ret = (ret - adj(arg))*0.5;
-      ret -= trace(ret)*factor;
+      iMatrix<vtype,N> ret;
+
+      double factor = (1.0/(double)N);
+      ret= (arg - adj(arg))*0.5;
+      ret=ret - (trace(ret)*factor);
       return ret;
     }
 
diff --git a/tests/Make.inc b/tests/Make.inc
index 8b313ccc..38714c2f 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,5 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd Test_wilson_force Test_wilson_force_phiMdagMphi Test_wilson_force_phiMphi
+bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_hmc_WilsonFermionGauge Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd Test_wilson_force Test_wilson_force_phiMdagMphi Test_wilson_force_phiMphi
 
 
 Test_GaugeAction_SOURCES=Test_GaugeAction.cc
@@ -78,6 +78,10 @@ Test_gamma_SOURCES=Test_gamma.cc
 Test_gamma_LDADD=-lGrid
 
 
+Test_hmc_WilsonFermionGauge_SOURCES=Test_hmc_WilsonFermionGauge.cc
+Test_hmc_WilsonFermionGauge_LDADD=-lGrid
+
+
 Test_hmc_WilsonGauge_SOURCES=Test_hmc_WilsonGauge.cc
 Test_hmc_WilsonGauge_LDADD=-lGrid
 
diff --git a/tests/Test_hmc_WilsonFermionGauge.cc b/tests/Test_hmc_WilsonFermionGauge.cc
new file mode 100644
index 00000000..3a10b191
--- /dev/null
+++ b/tests/Test_hmc_WilsonFermionGauge.cc
@@ -0,0 +1,55 @@
+#include "Grid.h"
+
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplex::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+  
+  GridCartesian            Fine(latt_size,simd_layout,mpi_layout);
+  GridRedBlackCartesian  RBFine(latt_size,simd_layout,mpi_layout);
+  GridParallelRNG  pRNG(&Fine);
+  pRNG.SeedRandomDevice();
+  LatticeLorentzColourMatrix     U(&Fine);
+
+  SU3::HotConfiguration(pRNG, U);
+
+  // simplify template declaration? Strip the lorentz from the second template
+  WilsonGaugeAction<LatticeLorentzColourMatrix, LatticeColourMatrix> Waction(5.6);
+
+  Real mass=0.01;
+  WilsonFermion FermOp(U,Fine,RBFine,mass);
+  
+  ConjugateGradient<LatticeFermion>  CG(1.0e-8,10000);
+  
+  TwoFlavourPseudoFermionAction<LatticeLorentzColourMatrix, LatticeColourMatrix,LatticeFermion> 
+    Pseudofermion(FermOp,CG,CG,Fine);
+
+  
+  //Collect actions
+  ActionLevel Level1;
+  Level1.push_back(&Waction);
+  Level1.push_back(&Pseudofermion);
+  ActionSet FullSet;
+  FullSet.push_back(Level1);
+
+  // Create integrator
+  typedef LeapFrog  IntegratorAlgorithm;// change here to change the algorithm
+  IntegratorParameters MDpar(12,40,1.0);
+  std::vector<int> rel ={1};
+  Integrator<IntegratorAlgorithm> MDynamics(&Fine,MDpar, FullSet,rel);
+
+  // Create HMC
+  HMCparameters HMCpar;
+  HybridMonteCarlo<IntegratorAlgorithm>  HMC(HMCpar, MDynamics);
+
+  HMC.evolve(U);
+
+}
diff --git a/tests/Test_wilson_force.cc b/tests/Test_wilson_force.cc
index 6a8d7611..b08ccd45 100644
--- a/tests/Test_wilson_force.cc
+++ b/tests/Test_wilson_force.cc
@@ -48,17 +48,19 @@ int main (int argc, char ** argv)
   LatticeGaugeField tmp(&Grid);
 
   Dw.MDeriv(tmp , Mphi,  phi,DaggerNo );  UdSdU=tmp;
-  Dw.MDeriv(tmp , phi,  Mphi,DaggerYes ); UdSdU=UdSdU+tmp;
-
-
+  Dw.MDeriv(tmp , phi,  Mphi,DaggerYes ); UdSdU=(UdSdU+tmp);
+  
   LatticeFermion Ftmp      (&Grid);
 
   ////////////////////////////////////
   // Modify the gauge field a little 
   ////////////////////////////////////
-  RealD dt = 1.0e-6;
-
+  RealD dt = 0.0001;
+  RealD Hmom = 0.0;
+  RealD Hmomprime = 0.0;
+  RealD Hmompp    = 0.0;
   LatticeColourMatrix mommu(&Grid); 
+  LatticeColourMatrix forcemu(&Grid); 
   LatticeGaugeField mom(&Grid); 
   LatticeGaugeField Uprime(&Grid); 
 
@@ -66,13 +68,26 @@ int main (int argc, char ** argv)
 
     SU3::GaussianLieAlgebraMatrix(pRNG, mommu); // Traceless antihermitian momentum; gaussian in lie alg
 
+    Hmom -= real(sum(trace(mommu*mommu)));
+
     PokeIndex<LorentzIndex>(mom,mommu,mu);
+
+    // fourth order exponential approx
     parallel_for(auto i=mom.begin();i<mom.end();i++){
-      Uprime[i](mu) =U[i](mu)+ mom[i](mu)*U[i](mu)*dt;
+      Uprime[i](mu) =
+	  U[i](mu)
+	+ mom[i](mu)*U[i](mu)*dt 
+	+ mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt/2.0)
+	+ mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt/6.0)
+	+ mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt*dt/24.0)
+	+ mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt*dt*dt/120.0)
+	+ mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt*dt*dt*dt/720.0)
+	;
     }
 
   }
 
+  std::cout << GridLogMessage <<"Initial mom hamiltonian is "<< Hmom <<std::endl;
   Dw.DoubleStore(Dw.Umu,Uprime);
   Dw.M          (phi,MphiPrime);
 
@@ -81,21 +96,75 @@ int main (int argc, char ** argv)
   //////////////////////////////////////////////
   // Use derivative to estimate dS
   //////////////////////////////////////////////
-  LatticeComplex dS(&Grid); dS = zero;
 
-  parallel_for(auto i=mom.begin();i<mom.end();i++){
-    for(int mu=0;mu<Nd;mu++){
-      //      dS[i]() = dS[i]()+trace(mom[i](mu) * UdSdU[i](mu) - mom[i](mu)* adj( UdSdU[i](mu)) )*dt;
-      dS[i]() =    dS[i]()+trace(mom[i](mu) * (UdSdU[i](mu)))*dt;
-      dS[i]() =    dS[i]()-trace(mom[i](mu) * adj(UdSdU[i](mu)))*dt;
-    }
+
+  for(int mu=0;mu<Nd;mu++){
+    std::cout << "" <<std::endl;
+    mommu   = PeekIndex<LorentzIndex>(mom,mu);
+    std::cout << GridLogMessage<< " Mommu  " << norm2(mommu)<<std::endl;
+    mommu   = mommu+adj(mommu);
+    std::cout << GridLogMessage<< " Mommu + Mommudag " << norm2(mommu)<<std::endl;
+    mommu   = PeekIndex<LorentzIndex>(UdSdU,mu);
+    std::cout << GridLogMessage<< " dsdumu  " << norm2(mommu)<<std::endl;
+    mommu   = mommu+adj(mommu);
+    std::cout << GridLogMessage<< " dsdumu + dag  " << norm2(mommu)<<std::endl;
   }
+
+  LatticeComplex dS(&Grid); dS = zero;
+  LatticeComplex dSmom(&Grid); dSmom = zero;
+  LatticeComplex dSmom2(&Grid); dSmom2 = zero;
+  for(int mu=0;mu<Nd;mu++){
+    mommu   = PeekIndex<LorentzIndex>(UdSdU,mu);
+    mommu=Ta(mommu)*2.0;
+    PokeIndex<LorentzIndex>(UdSdU,mommu,mu);
+  }
+
+  for(int mu=0;mu<Nd;mu++){
+    mommu   = PeekIndex<LorentzIndex>(mom,mu);
+    std::cout << GridLogMessage<< " Mommu  " << norm2(mommu)<<std::endl;
+    mommu   = mommu+adj(mommu);
+    std::cout << GridLogMessage<< " Mommu + Mommudag " << norm2(mommu)<<std::endl;
+    mommu   = PeekIndex<LorentzIndex>(UdSdU,mu);
+    std::cout << GridLogMessage<< " dsdumu  " << norm2(mommu)<<std::endl;
+    mommu   = mommu+adj(mommu);
+    std::cout << GridLogMessage<< " dsdumu + dag  " << norm2(mommu)<<std::endl;
+  }
+
+  for(int mu=0;mu<Nd;mu++){
+    forcemu = PeekIndex<LorentzIndex>(UdSdU,mu);
+    mommu   = PeekIndex<LorentzIndex>(mom,mu);
+
+    // Update PF action density
+    dS = dS+trace(mommu*forcemu)*dt;
+
+    dSmom  = dSmom  - trace(mommu*forcemu) * dt;
+    dSmom2 = dSmom2 - trace(forcemu*forcemu) *(0.25* dt*dt);
+
+    // Update mom action density
+    mommu = mommu + forcemu*(dt*0.5);
+
+    Hmomprime -= real(sum(trace(mommu*mommu)));
+
+  }
+
   Complex dSpred    = sum(dS);
+  Complex dSm       = sum(dSmom);
+  Complex dSm2      = sum(dSmom2);
+
+
+  std::cout << GridLogMessage <<"Initial mom hamiltonian is "<< Hmom <<std::endl;
+  std::cout << GridLogMessage <<"Final   mom hamiltonian is "<< Hmomprime <<std::endl;
+  std::cout << GridLogMessage <<"Delta   mom hamiltonian is "<< Hmomprime-Hmom <<std::endl;
 
   std::cout << GridLogMessage << " S      "<<S<<std::endl;
   std::cout << GridLogMessage << " Sprime "<<Sprime<<std::endl;
   std::cout << GridLogMessage << "dS      "<<Sprime-S<<std::endl;
   std::cout << GridLogMessage << "predict dS    "<< dSpred <<std::endl;
+  std::cout << GridLogMessage <<"dSm "<< dSm<<std::endl;
+  std::cout << GridLogMessage <<"dSm2"<< dSm2<<std::endl;
+
+  std::cout << GridLogMessage << "Total dS    "<< Hmomprime - Hmom + Sprime - S <<std::endl;
+
 
   std::cout<< GridLogMessage << "Done" <<std::endl;
   Grid_finalize();

From de153b70cee85e3a4850c03f5c2fb234bc118fd0 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 29 Jul 2015 21:02:07 +0900
Subject: [PATCH 410/429] Bug in two flav pseudofermion corrected to reimport
 gauge field upon rejection. exp(-DeltaH) = 1 now, and plaquette is sensible.
 Will reproduce an old Wilson Gauge Wilson Fermion SCRI plaquette with
 precision in mass matching shortly.

---
 lib/qcd/action/pseudofermion/TwoFlavour.h | 1 +
 lib/qcd/hmc/HMC.h                         | 7 +++++--
 lib/qcd/hmc/integrators/Integrator_base.h | 2 +-
 tests/Test_hmc_WilsonFermionGauge.cc      | 5 +++--
 4 files changed, 10 insertions(+), 5 deletions(-)

diff --git a/lib/qcd/action/pseudofermion/TwoFlavour.h b/lib/qcd/action/pseudofermion/TwoFlavour.h
index 67f3101c..68de0756 100644
--- a/lib/qcd/action/pseudofermion/TwoFlavour.h
+++ b/lib/qcd/action/pseudofermion/TwoFlavour.h
@@ -144,6 +144,7 @@ namespace Grid{
 
 	gaussian(pRNG,eta);
 
+	FermOp.ImportGauge(U);
 	FermOp.Mdag(eta,Phi);
 
 	Phi=Phi*scale;
diff --git a/lib/qcd/hmc/HMC.h b/lib/qcd/hmc/HMC.h
index bc4f47ff..684d784a 100644
--- a/lib/qcd/hmc/HMC.h
+++ b/lib/qcd/hmc/HMC.h
@@ -94,6 +94,7 @@ namespace Grid{
 
 
       void evolve(LatticeLorentzColourMatrix& Uin){
+
 	Real DeltaH;
 	
 	// Thermalizations
@@ -106,14 +107,16 @@ namespace Grid{
 
 	// Actual updates (evolve a copy Ucopy then copy back eventually)
 	LatticeLorentzColourMatrix Ucopy(Uin._grid);
-	for(int iter=Params.StartingConfig; 
-	    iter < Params.Nsweeps+Params.StartingConfig; ++iter){
+	for(int iter=Params.StartingConfig; iter < Params.Nsweeps+Params.StartingConfig; ++iter){
+
 	  std::cout<<GridLogMessage << "-- # Sweep = "<< iter <<  "\n";
 	  
 	  Ucopy = Uin;
+
 	  DeltaH = evolve_step(Ucopy);
 		
 	  if(metropolis_test(DeltaH)) Uin = Ucopy;
+
 	  //need sync? // Query Guido on this comment
 
 	}
diff --git a/lib/qcd/hmc/integrators/Integrator_base.h b/lib/qcd/hmc/integrators/Integrator_base.h
index 1cc32808..9d8cd932 100644
--- a/lib/qcd/hmc/integrators/Integrator_base.h
+++ b/lib/qcd/hmc/integrators/Integrator_base.h
@@ -60,10 +60,10 @@ namespace Grid{
       
       IntegratorPolicy TheIntegrator;// contains parameters too
 
-
       void update_P(LatticeLorentzColourMatrix&U, int level,double ep){
 	for(int a=0; a<as[level].size(); ++a){
 	  LatticeLorentzColourMatrix force(U._grid);
+
 	  as[level].at(a)->deriv(U,force);
 
 	  Complex dSdt=0.0;
diff --git a/tests/Test_hmc_WilsonFermionGauge.cc b/tests/Test_hmc_WilsonFermionGauge.cc
index 3a10b191..f54503af 100644
--- a/tests/Test_hmc_WilsonFermionGauge.cc
+++ b/tests/Test_hmc_WilsonFermionGauge.cc
@@ -41,8 +41,9 @@ int main (int argc, char ** argv)
   FullSet.push_back(Level1);
 
   // Create integrator
-  typedef LeapFrog  IntegratorAlgorithm;// change here to change the algorithm
-  IntegratorParameters MDpar(12,40,1.0);
+  typedef MinimumNorm2  IntegratorAlgorithm;// change here to change the algorithm
+  //  typedef LeapFrog  IntegratorAlgorithm;// change here to change the algorithm
+  IntegratorParameters MDpar(12,20,1.0);
   std::vector<int> rel ={1};
   Integrator<IntegratorAlgorithm> MDynamics(&Fine,MDpar, FullSet,rel);
 

From f78cea58fa9d2c1c63484088af2f243d8de6b628 Mon Sep 17 00:00:00 2001
From: neo <cossu@post.kek.jp>
Date: Thu, 30 Jul 2015 17:21:42 +0900
Subject: [PATCH 411/429] Amending a merge mistake

---
 tests/Test_simd.cc | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)

diff --git a/tests/Test_simd.cc b/tests/Test_simd.cc
index 004f2f52..d311f8f0 100644
--- a/tests/Test_simd.cc
+++ b/tests/Test_simd.cc
@@ -194,9 +194,9 @@ int main (int argc, char ** argv)
   // Insist that operations on random scalars gives
   // identical results to on vectors.
 
-  std::cout << "==================================="<<  std::endl;
-  std::cout << "Testing vRealF "<<std::endl;
-  std::cout << "==================================="<<  std::endl;
+  std::cout << GridLogMessage <<"==================================="<<  std::endl;
+  std::cout << GridLogMessage <<"Testing vRealF "<<std::endl;
+  std::cout << GridLogMessage <<"==================================="<<  std::endl;
 
 
   Tester<RealF,vRealF>(funcPlus());
@@ -207,9 +207,9 @@ int main (int argc, char ** argv)
   Tester<RealF,vRealF>(funcInnerProduct());
   ReductionTester<RealF,RealF,vRealF>(funcReduce());
 
-  std::cout << "==================================="<<  std::endl;
-  std::cout << "Testing vRealD "<<std::endl;
-  std::cout << "==================================="<<  std::endl;
+  std::cout << GridLogMessage <<"==================================="<<  std::endl;
+  std::cout << GridLogMessage <<"Testing vRealD "<<std::endl;
+  std::cout << GridLogMessage <<"==================================="<<  std::endl;
 
   Tester<RealD,vRealD>(funcPlus());
   Tester<RealD,vRealD>(funcMinus());
@@ -219,9 +219,9 @@ int main (int argc, char ** argv)
   Tester<RealD,vRealD>(funcInnerProduct());
   ReductionTester<RealD,RealD,vRealD>(funcReduce());
 
-  std::cout << "==================================="<<  std::endl;
-  std::cout << "Testing vComplexF "<<std::endl;
-  std::cout << "==================================="<<  std::endl;
+  std::cout << GridLogMessage <<"==================================="<<  std::endl;
+  std::cout << GridLogMessage <<"Testing vComplexF "<<std::endl;
+  std::cout << GridLogMessage <<"==================================="<<  std::endl;
 
   Tester<ComplexF,vComplexF>(funcTimesI());
   Tester<ComplexF,vComplexF>(funcTimesMinusI());

From 1d0be956ae706fa2d8e8fc670c42f95067488d4d Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 1 Aug 2015 05:58:35 +0900
Subject: [PATCH 412/429] Jackson smoothed chebyshev and (untested) completion
 of force terms for Cayley, Partial and Cont fraction dwf and overlap. have
 even odd and unprec forces.

---
 lib/algorithms/approx/Chebyshev.h             |  31 ++++-
 lib/qcd/action/fermion/CayleyFermion5D.cc     | 108 +++++++++++++++++-
 lib/qcd/action/fermion/CayleyFermion5D.h      |   8 ++
 .../fermion/ContinuedFractionFermion5D.cc     |  47 ++++++++
 .../fermion/ContinuedFractionFermion5D.h      |   5 +
 .../fermion/PartialFractionFermion5D.cc       |  45 ++++++++
 .../action/fermion/PartialFractionFermion5D.h |   5 +
 lib/qcd/action/fermion/WilsonFermion5D.cc     |  23 +++-
 scripts/substitute                            |  17 ++-
 scripts/substitute-global                     |   6 +
 tests/Make.inc                                |   6 +-
 tests/Test_dwf_hdcr.cc                        |   4 +-
 tests/Test_hmc_WilsonFermionGauge.cc          |   5 +-
 13 files changed, 292 insertions(+), 18 deletions(-)
 create mode 100755 scripts/substitute-global

diff --git a/lib/algorithms/approx/Chebyshev.h b/lib/algorithms/approx/Chebyshev.h
index e4fa0b61..37331816 100644
--- a/lib/algorithms/approx/Chebyshev.h
+++ b/lib/algorithms/approx/Chebyshev.h
@@ -58,7 +58,9 @@ namespace Grid {
       }
     };
 
-
+    
+    // c.f. numerical recipes "chebft"/"chebev". This is sec 5.8 "Chebyshev approximation".
+    //
     Chebyshev(double _lo,double _hi,int _order, double (* func)(double) ){
       lo=_lo;
       hi=_hi;
@@ -77,7 +79,34 @@ namespace Grid {
 	Coeffs[j] = s * 2.0/order;
       }
     };
+    void JacksonSmooth(void){
+      double M=order;
+      double alpha = M_PI/(M+2);
+      double lmax = std::cos(alpha);
+      double sumUsq =0;
+      std::vector<double> U(M);
+      std::vector<double> a(M);
+      std::vector<double> g(M);
+      for(int n=0;n<=M;n++){
+	U[n] = std::sin((n+1)*std::acos(lmax))/std::sin(std::acos(lmax));
+	sumUsq += U[n]*U[n];
+      }      
+      sumUsq = std::sqrt(sumUsq);
 
+      for(int i=1;i<=M;i++){
+	a[i] = U[i]/sumUsq;
+      }
+      g[0] = 1.0;
+      for(int m=1;m<=M;m++){
+	g[m] = 0;
+	for(int i=0;i<=M-m;i++){
+	  g[m]+= a[i]*a[m+i];
+	}
+      }
+      for(int m=1;m<=M;m++){
+	Coeffs[m]*=g[m];
+      }
+    }
     double approx(double x) // Convenience for plotting the approximation
     {
       double Tn;
diff --git a/lib/qcd/action/fermion/CayleyFermion5D.cc b/lib/qcd/action/fermion/CayleyFermion5D.cc
index 7db67dd7..c18edb7b 100644
--- a/lib/qcd/action/fermion/CayleyFermion5D.cc
+++ b/lib/qcd/action/fermion/CayleyFermion5D.cc
@@ -17,11 +17,8 @@ namespace QCD {
  {
  }
 
-  // override multiply
-  RealD CayleyFermion5D::M    (const LatticeFermion &psi, LatticeFermion &chi)
+  void CayleyFermion5D::Meooe5D    (const LatticeFermion &psi, LatticeFermion &Din)
   {
-    LatticeFermion Din(psi._grid);
-
     // Assemble Din
     for(int s=0;s<Ls;s++){
       if ( s==0 ) {
@@ -37,11 +34,52 @@ namespace QCD {
 	axpby_ssp_pplus(Din,1.0,Din,cs[s],psi,s,s-1);
       }
     }
+  }
+  void CayleyFermion5D::MeooeDag5D    (const LatticeFermion &psi, LatticeFermion &Din)
+  {
+    for(int s=0;s<Ls;s++){
+      if ( s==0 ) {
+	axpby_ssp_pplus (Din,bs[s],psi,cs[s+1],psi,s,s+1);
+	axpby_ssp_pminus(Din,1.0,Din,-mass*cs[Ls-1],psi,s,Ls-1);
+      } else if ( s==(Ls-1)) { 
+	axpby_ssp_pplus (Din,bs[s],psi,-mass*cs[0],psi,s,0);
+	axpby_ssp_pminus(Din,1.0,Din,cs[s-1],psi,s,s-1);
+      } else {
+	axpby_ssp_pplus (Din,bs[s],psi,cs[s+1],psi,s,s+1);
+	axpby_ssp_pminus(Din,1.0,Din,cs[s-1],psi,s,s-1);
+      }
+    }
+  }
+
+  // override multiply
+  RealD CayleyFermion5D::M    (const LatticeFermion &psi, LatticeFermion &chi)
+  {
+    LatticeFermion Din(psi._grid);
+
+    // Assemble Din
+    /*
+    for(int s=0;s<Ls;s++){
+      if ( s==0 ) {
+	//	Din = bs psi[s] + cs[s] psi[s+1}
+	axpby_ssp_pminus(Din,bs[s],psi,cs[s],psi,s,s+1);
+	//      Din+= -mass*cs[s] psi[s+1}
+	axpby_ssp_pplus (Din,1.0,Din,-mass*cs[s],psi,s,Ls-1);
+      } else if ( s==(Ls-1)) { 
+	axpby_ssp_pminus(Din,bs[s],psi,-mass*cs[s],psi,s,0);
+	axpby_ssp_pplus (Din,1.0,Din,cs[s],psi,s,s-1);
+      } else {
+	axpby_ssp_pminus(Din,bs[s],psi,cs[s],psi,s,s+1);
+	axpby_ssp_pplus(Din,1.0,Din,cs[s],psi,s,s-1);
+      }
+    }
+    */
+    Meooe5D(psi,Din);
 
     DW(Din,chi,DaggerNo);
     // ((b D_W + D_w hop terms +1) on s-diag
     axpby(chi,1.0,1.0,chi,psi); 
 
+    // Call Mooee??
     for(int s=0;s<Ls;s++){
       if ( s==0 ){
 	axpby_ssp_pminus(chi,1.0,chi,-1.0,psi,s,s+1);
@@ -67,10 +105,14 @@ namespace QCD {
     // Apply Dw
     DW(psi,Din,DaggerYes); 
 
+    Meooe5D(Din,chi);
+
     for(int s=0;s<Ls;s++){
+
       // Collect the terms in DW
       //	Chi = bs Din[s] + cs[s] Din[s+1}
       //    Chi+= -mass*cs[s] psi[s+1}
+      /*
       if ( s==0 ) {
 	axpby_ssp_pplus (chi,bs[s],Din,cs[s+1],Din,s,s+1);
 	axpby_ssp_pminus(chi,1.0,chi,-mass*cs[Ls-1],Din,s,Ls-1);
@@ -81,6 +123,10 @@ namespace QCD {
 	axpby_ssp_pplus (chi,bs[s],Din,cs[s+1],Din,s,s+1);
 	axpby_ssp_pminus(chi,1.0,chi,cs[s-1],Din,s,s-1);
       }
+      */
+
+      // FIXME just call MooeeDag??
+
       // Collect the terms indept of DW
       if ( s==0 ){
 	axpby_ssp_pplus (chi,1.0,chi,-1.0,psi,s,s+1);
@@ -103,6 +149,9 @@ namespace QCD {
   {
     LatticeFermion tmp(psi._grid);
     // Assemble the 5d matrix
+    Meooe5D(psi,tmp); 
+    std::cout << "Meooe Test replacement norm2 tmp = " <<norm2(tmp)<<std::endl;
+
     for(int s=0;s<Ls;s++){
       if ( s==0 ) {
 	//	tmp = bs psi[s] + cs[s] psi[s+1}
@@ -117,6 +166,7 @@ namespace QCD {
 	axpby_ssp_pplus (tmp,1.0,tmp,-ceo[s],psi,s,s-1);
       }
     }
+    std::cout << "Meooe Test replacement norm2 tmp old = " <<norm2(tmp)<<std::endl;
     // Apply 4d dslash
     if ( psi.checkerboard == Odd ) {
       DhopEO(tmp,chi,DaggerNo);
@@ -134,6 +184,10 @@ namespace QCD {
     } else {
       DhopOE(psi,tmp,DaggerYes);
     }
+
+    Meooe5D(tmp,chi); 
+    std::cout << "Meooe Test replacement norm2 chi new = " <<norm2(chi)<<std::endl;
+
     // Assemble the 5d matrix
     for(int s=0;s<Ls;s++){
       if ( s==0 ) {
@@ -147,6 +201,8 @@ namespace QCD {
 	axpby_ssp_pminus(chi,1.0   ,chi,-ceo[s-1],tmp,s,s-1);
       }
     }
+    std::cout << "Meooe Test replacement norm2 chi old = " <<norm2(chi)<<std::endl;
+
   }
 
   void CayleyFermion5D::Mooee       (const LatticeFermion &psi, LatticeFermion &chi)
@@ -249,6 +305,50 @@ namespace QCD {
       axpby_ssp_pplus (chi,1.0,chi,-lee[s],chi,s,s+1);  // chi[Ls]
     }
   }
+
+  // force terms; five routines; default to Dhop on diagonal
+  void CayleyFermion5D::MDeriv  (LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+  {
+    LatticeFermion Din(V._grid);
+
+    if ( dag == DaggerNo ) {
+      //      U d/du [D_w D5] V = U d/du DW D5 V
+      Meooe5D(V,Din);
+      DhopDeriv(mat,U,Din,dag);
+    } else {
+      //      U d/du [D_w D5]^dag V = U D5^dag d/du DW^dag Y // implicit adj on U in call
+      Meooe5D(U,Din);
+      DhopDeriv(mat,Din,V,dag);
+    }
+  };
+  void CayleyFermion5D::MoeDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+  {
+    LatticeFermion Din(V._grid);
+
+    if ( dag == DaggerNo ) {
+      //      U d/du [D_w D5] V = U d/du DW D5 V
+      Meooe5D(V,Din);
+      DhopDerivOE(mat,U,Din,dag);
+    } else {
+      //      U d/du [D_w D5]^dag V = U D5^dag d/du DW^dag Y // implicit adj on U in call
+      Meooe5D(U,Din);
+      DhopDerivOE(mat,Din,V,dag);
+    }
+  };
+  void CayleyFermion5D::MeoDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+  {
+    LatticeFermion Din(V._grid);
+
+    if ( dag == DaggerNo ) {
+      //      U d/du [D_w D5] V = U d/du DW D5 V
+      Meooe5D(V,Din);
+      DhopDerivEO(mat,U,Din,dag);
+    } else {
+      //      U d/du [D_w D5]^dag V = U D5^dag d/du DW^dag Y // implicit adj on U in call
+      Meooe5D(U,Din);
+      DhopDerivEO(mat,Din,V,dag);
+    }
+  };
   
   // Tanh
   void CayleyFermion5D::SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD b,RealD c)
diff --git a/lib/qcd/action/fermion/CayleyFermion5D.h b/lib/qcd/action/fermion/CayleyFermion5D.h
index 208ab0bb..a139c32e 100644
--- a/lib/qcd/action/fermion/CayleyFermion5D.h
+++ b/lib/qcd/action/fermion/CayleyFermion5D.h
@@ -22,8 +22,16 @@ namespace Grid {
       virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
       virtual void   Instantiatable(void)=0;
 
+      // force terms; five routines; default to Dhop on diagonal
+      virtual void MDeriv  (LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+      virtual void MoeDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+      virtual void MeoDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+
       // Efficient support for multigrid coarsening
       virtual void  Mdir (const LatticeFermion &in, LatticeFermion &out,int dir,int disp);
+      
+      void   Meooe5D       (const LatticeFermion &in, LatticeFermion &out);
+      void   MeooeDag5D    (const LatticeFermion &in, LatticeFermion &out);
 
       //    protected:
       RealD mass;
diff --git a/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc b/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
index 2206bbc8..cae73dfc 100644
--- a/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
+++ b/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
@@ -169,6 +169,53 @@ namespace Grid {
     {
       MooeeInv(psi,chi);
     }
+
+  // force terms; five routines; default to Dhop on diagonal
+   void ContinuedFractionFermion5D::MDeriv  (LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+  {
+    LatticeFermion D(V._grid);
+
+    int sign=1;
+    for(int s=0;s<Ls;s++){
+      if ( s==(Ls-1) ){
+	ag5xpby_ssp(D,Beta[s]*ZoloHiInv,U,0.0,U,s,s);
+      } else {
+	ag5xpby_ssp(D,cc[s]*Beta[s]*sign*ZoloHiInv,U,0.0,U,s,s);
+      }
+      sign=-sign; 
+    }
+    DhopDeriv(mat,D,V,DaggerNo); 
+  };
+   void ContinuedFractionFermion5D::MoeDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+  {
+    LatticeFermion D(V._grid);
+
+    int sign=1;
+    for(int s=0;s<Ls;s++){
+      if ( s==(Ls-1) ){
+	ag5xpby_ssp(D,Beta[s]*ZoloHiInv,U,0.0,U,s,s);
+      } else {
+	ag5xpby_ssp(D,cc[s]*Beta[s]*sign*ZoloHiInv,U,0.0,U,s,s);
+      }
+      sign=-sign; 
+    }
+    DhopDerivOE(mat,D,V,DaggerNo); 
+  };
+   void ContinuedFractionFermion5D::MeoDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+  {
+    LatticeFermion D(V._grid);
+
+    int sign=1;
+    for(int s=0;s<Ls;s++){
+      if ( s==(Ls-1) ){
+	ag5xpby_ssp(D,Beta[s]*ZoloHiInv,U,0.0,U,s,s);
+      } else {
+	ag5xpby_ssp(D,cc[s]*Beta[s]*sign*ZoloHiInv,U,0.0,U,s,s);
+      }
+      sign=-sign; 
+    }
+    DhopDerivEO(mat,D,V,DaggerNo); 
+  };
     
     // Constructors
     ContinuedFractionFermion5D::ContinuedFractionFermion5D(
diff --git a/lib/qcd/action/fermion/ContinuedFractionFermion5D.h b/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
index 43b0f756..7d0c75bc 100644
--- a/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
+++ b/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
@@ -21,6 +21,11 @@ namespace Grid {
       virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out);
       virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
 
+      // force terms; five routines; default to Dhop on diagonal
+      virtual void MDeriv  (LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+      virtual void MoeDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+      virtual void MeoDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+
       //      virtual void   Instantiatable(void)=0;
       virtual void   Instantiatable(void) =0;
 
diff --git a/lib/qcd/action/fermion/PartialFractionFermion5D.cc b/lib/qcd/action/fermion/PartialFractionFermion5D.cc
index 958cea30..74174f51 100644
--- a/lib/qcd/action/fermion/PartialFractionFermion5D.cc
+++ b/lib/qcd/action/fermion/PartialFractionFermion5D.cc
@@ -254,6 +254,51 @@ namespace Grid {
       MooeeInv_internal(in,out,DaggerYes);
     }
 
+
+  // force terms; five routines; default to Dhop on diagonal
+   void PartialFractionFermion5D::MDeriv  (LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+  {
+    LatticeFermion D(V._grid);
+
+    int nblock=(Ls-1)/2;
+    for(int b=0;b<nblock;b++){
+      int s = 2*b;
+      ag5xpby_ssp(D,-scale,U,0.0,U,s,s); 
+      ag5xpby_ssp(D, scale,U,0.0,U,s+1,s+1); 
+    }
+    ag5xpby_ssp(D,p[nblock]*scale/amax,U,0.0,U,Ls-1,Ls-1);
+
+    DhopDeriv(mat,D,V,DaggerNo); 
+  };
+   void PartialFractionFermion5D::MoeDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+  {
+    LatticeFermion D(V._grid);
+
+    int nblock=(Ls-1)/2;
+    for(int b=0;b<nblock;b++){
+      int s = 2*b;
+      ag5xpby_ssp(D,-scale,U,0.0,U,s,s); 
+      ag5xpby_ssp(D, scale,U,0.0,U,s+1,s+1); 
+    }
+    ag5xpby_ssp(D,p[nblock]*scale/amax,U,0.0,U,Ls-1,Ls-1);
+
+    DhopDerivOE(mat,D,V,DaggerNo); 
+  };
+   void PartialFractionFermion5D::MeoDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+  {
+    LatticeFermion D(V._grid);
+
+    int nblock=(Ls-1)/2;
+    for(int b=0;b<nblock;b++){
+      int s = 2*b;
+      ag5xpby_ssp(D,-scale,U,0.0,U,s,s); 
+      ag5xpby_ssp(D, scale,U,0.0,U,s+1,s+1); 
+    }
+    ag5xpby_ssp(D,p[nblock]*scale/amax,U,0.0,U,Ls-1,Ls-1);
+
+    DhopDerivEO(mat,D,V,DaggerNo); 
+  };
+
     void  PartialFractionFermion5D::SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD scale){
       SetCoefficientsZolotarev(1.0/scale,zdata);
     }
diff --git a/lib/qcd/action/fermion/PartialFractionFermion5D.h b/lib/qcd/action/fermion/PartialFractionFermion5D.h
index e734caa0..3ba2481a 100644
--- a/lib/qcd/action/fermion/PartialFractionFermion5D.h
+++ b/lib/qcd/action/fermion/PartialFractionFermion5D.h
@@ -28,6 +28,11 @@ namespace Grid {
       virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out);
       virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
 
+      // force terms; five routines; default to Dhop on diagonal
+      virtual void MDeriv  (LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+      virtual void MoeDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+      virtual void MeoDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+
       virtual void   Instantiatable(void) =0; // ensure no make-eee
 
       // Efficient support for multigrid coarsening
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.cc b/lib/qcd/action/fermion/WilsonFermion5D.cc
index c3329503..7622205a 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.cc
+++ b/lib/qcd/action/fermion/WilsonFermion5D.cc
@@ -124,11 +124,14 @@ void WilsonFermion5D::DerivInternal(CartesianStencil & st,
 
   WilsonCompressor compressor(dag);
   
-  LatticeColourMatrix tmp(B._grid);
+  LatticeColourMatrix tmp(mat._grid);
   LatticeFermion Btilde(B._grid);
+  LatticeFermion Atilde(B._grid);
 
   st.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(B,comm_buf,compressor);
-  
+
+  Atilde=A;
+
   for(int mu=0;mu<Nd;mu++){
       
     ////////////////////////////////////////////////////////////////////////
@@ -140,20 +143,28 @@ void WilsonFermion5D::DerivInternal(CartesianStencil & st,
     ////////////////////////
     // Call the single hop
     ////////////////////////
+    tmp = zero;
+
 PARALLEL_FOR_LOOP
     for(int sss=0;sss<B._grid->oSites();sss++){
       for(int s=0;s<Ls;s++){
 	int sU=sss;
 	int sF = s+Ls*sU;
 	DiracOptDhopDir(st,U,comm_buf,sF,sU,B,Btilde,mu,gamma);
-      }
-    }
 
     ////////////////////////////
     // spin trace outer product
     ////////////////////////////
-// FIXME : need to sum over fifth direction.
-    tmp = TraceIndex<SpinIndex>(outerProduct(Btilde,A)); // ordering here
+
+	tmp[sU] = tmp[sU]+ traceIndex<SpinIndex>(
+		   outerProduct(Btilde[sF],Atilde[sF])); // ordering here
+
+      }
+
+
+    }
+
+
     PokeIndex<LorentzIndex>(mat,tmp,mu);
 
   }
diff --git a/scripts/substitute b/scripts/substitute
index 3066c4ef..fab833c7 100755
--- a/scripts/substitute
+++ b/scripts/substitute
@@ -1,6 +1,17 @@
 #!/bin/bash
 
 export LANG=C
-find . -name "*.cc" -print -exec sed -e "s/${1}/${2}/g" -i .bak {} \;
-find . -name "*.h" -print -exec sed -e "s/${1}/${2}/g" -i .bak {} \;
-find . -name "*.bak" -exec rm -f {} \;
+
+WAS=$1
+shift
+IS=$1
+shift
+
+while (( "$#" )); do
+
+echo $1
+sed -e "s/${WAS}/${IS}/g" -i .bak $1
+
+shift
+
+done
\ No newline at end of file
diff --git a/scripts/substitute-global b/scripts/substitute-global
new file mode 100755
index 00000000..3066c4ef
--- /dev/null
+++ b/scripts/substitute-global
@@ -0,0 +1,6 @@
+#!/bin/bash
+
+export LANG=C
+find . -name "*.cc" -print -exec sed -e "s/${1}/${2}/g" -i .bak {} \;
+find . -name "*.h" -print -exec sed -e "s/${1}/${2}/g" -i .bak {} \;
+find . -name "*.bak" -exec rm -f {} \;
diff --git a/tests/Make.inc b/tests/Make.inc
index 38714c2f..d510e98b 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,5 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_hmc_WilsonFermionGauge Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd Test_wilson_force Test_wilson_force_phiMdagMphi Test_wilson_force_phiMphi
+bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_cheby Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_hmc_WilsonFermionGauge Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd Test_wilson_force Test_wilson_force_phiMdagMphi Test_wilson_force_phiMphi
 
 
 Test_GaugeAction_SOURCES=Test_GaugeAction.cc
@@ -30,6 +30,10 @@ Test_cf_cr_unprec_SOURCES=Test_cf_cr_unprec.cc
 Test_cf_cr_unprec_LDADD=-lGrid
 
 
+Test_cheby_SOURCES=Test_cheby.cc
+Test_cheby_LDADD=-lGrid
+
+
 Test_contfrac_cg_SOURCES=Test_contfrac_cg.cc
 Test_contfrac_cg_LDADD=-lGrid
 
diff --git a/tests/Test_dwf_hdcr.cc b/tests/Test_dwf_hdcr.cc
index 29f6e998..6ff26425 100644
--- a/tests/Test_dwf_hdcr.cc
+++ b/tests/Test_dwf_hdcr.cc
@@ -285,6 +285,8 @@ public:
     //    Chebyshev<FineField> ChebyAccu(0.5,70.0,30,InverseApproximation);
     Chebyshev<FineField> Cheby    (2.0,70.0,10,InverseApproximation);
     Chebyshev<FineField> ChebyAccu(2.0,70.0,10,InverseApproximation);
+    Cheby.JacksonSmooth();
+    ChebyAccu.JacksonSmooth();
 
     _Aggregates.ProjectToSubspace  (Csrc,in);
     _Aggregates.PromoteFromSubspace(Csrc,out);
@@ -385,7 +387,7 @@ int main (int argc, char ** argv)
   LatticeGaugeField Umu(UGrid); 
 
   NerscField header;
-  std::string file("./ckpoint_lat.400");
+  std::string file("./ckpoint_lat.4000");
   readNerscConfiguration(Umu,header,file);
 
   //  SU3::ColdConfiguration(RNG4,Umu);
diff --git a/tests/Test_hmc_WilsonFermionGauge.cc b/tests/Test_hmc_WilsonFermionGauge.cc
index f54503af..867a9f96 100644
--- a/tests/Test_hmc_WilsonFermionGauge.cc
+++ b/tests/Test_hmc_WilsonFermionGauge.cc
@@ -24,13 +24,14 @@ int main (int argc, char ** argv)
   // simplify template declaration? Strip the lorentz from the second template
   WilsonGaugeAction<LatticeLorentzColourMatrix, LatticeColourMatrix> Waction(5.6);
 
-  Real mass=0.01;
+  Real mass=-0.77;
   WilsonFermion FermOp(U,Fine,RBFine,mass);
   
   ConjugateGradient<LatticeFermion>  CG(1.0e-8,10000);
+  ConjugateGradient<LatticeFermion>  CGmd(1.0e-6,10000);
   
   TwoFlavourPseudoFermionAction<LatticeLorentzColourMatrix, LatticeColourMatrix,LatticeFermion> 
-    Pseudofermion(FermOp,CG,CG,Fine);
+    Pseudofermion(FermOp,CGmd,CG,Fine);
 
   
   //Collect actions

From 2994274267d66f34fd6673cba10256fdfac0f4e0 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 1 Aug 2015 22:06:07 +0900
Subject: [PATCH 413/429] Changes making force term test for DWF pass.

---
 TODO                                      | 35 +++++++++++++++++++++++
 lib/Make.inc                              |  2 +-
 lib/qcd/action/fermion/FermionOperator.h  |  2 +-
 lib/qcd/action/fermion/WilsonFermion5D.cc | 12 ++++++--
 lib/qcd/action/fermion/WilsonKernels.cc   |  2 +-
 tests/Make.inc                            |  6 +++-
 6 files changed, 52 insertions(+), 7 deletions(-)

diff --git a/TODO b/TODO
index 5c7cf35d..47067c70 100644
--- a/TODO
+++ b/TODO
@@ -1,3 +1,38 @@
+- PseudoFermions
+TwoFlavour -- done. Fill in rest:
+
+  - TwoFlavourEvenOddConstEE
+  - TwoFlavourEvenOdd
+  - TwoFlavourRatioEvenOddConstEE
+  - TwoFlavourRatioEvenOdd
+  - TwoFlavourRatio
+
+  - OneFlavourRationalEvenOddConstEE
+  - OneFlavourRationalEvenOdd
+  - OneFlavourRationalRatioEvenOddConstEE
+  - OneFlavourRationalRatioEvenOdd
+  - OneFlavourRationalRatio
+
+- MacroMagic -> readers
+
+- Link smearing/boundary conds
+
+- Clean up HMC
+
+- Test Cayley, Partial , ContFrac force terms.
+
+- Test DWF HMC
+
+- Rectangle gauge actions.
+  Iwasaki,
+  Symanzik,
+  ... etc...
+
+- Prepare multigrid for HMC.
+  Alternate setup schemes.
+
+- RNG filling from sparser grid, lower dim grid.
+
 ================================================================
 *** Hacks and bug fixes to clean up and Audits
 ================================================================
diff --git a/lib/Make.inc b/lib/Make.inc
index 409bf754..e96c6fac 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/BfmHDCG.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/GeneralisedMinimumResidual.h ./algorithms/iterative/gmres.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Config.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./Init.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./Log.h ./MacroMagic.h ./Old/Tensor_peek.h ./Old/Tensor_poke.h ./parallelIO/NerscIO.h ./parameterIO/XMLReader.h ./qcd/action/ActionBase.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/pseudofermion/TwoFlavour.h ./qcd/hmc/HMC.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/hmc/integrators/Integrator_base.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_index.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h ./Timer.h
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/BfmHDCG.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/GeneralisedMinimumResidual.h ./algorithms/iterative/gmres.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Config.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./Init.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./Log.h ./MacroMagic.h ./Old/Tensor_peek.h ./Old/Tensor_poke.h ./parallelIO/NerscIO.h ./parameterIO/XMLReader.h ./qcd/action/ActionBase.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/pseudofermion/TwoFlavour.h ./qcd/action/pseudofermion/TwoFlavourEvenOdd.h ./qcd/action/pseudofermion/TwoFlavourEvenOddConstEE.h ./qcd/hmc/HMC.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/hmc/integrators/Integrator_base.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_index.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h ./Timer.h
 
 CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./Init.cc ./Log.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/hmc/HMC.cc ./qcd/hmc/integrators/Integrator.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/qcd/action/fermion/FermionOperator.h b/lib/qcd/action/fermion/FermionOperator.h
index b6ba2d08..57040037 100644
--- a/lib/qcd/action/fermion/FermionOperator.h
+++ b/lib/qcd/action/fermion/FermionOperator.h
@@ -59,7 +59,7 @@ namespace Grid {
       ///////////////////////////////////////////////
       // Updates gauge field during HMC
       ///////////////////////////////////////////////
-      virtual void ImportGauge(const GaugeField & _U);
+      virtual void ImportGauge(const GaugeField & _U)=0;
 
     };
 
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.cc b/lib/qcd/action/fermion/WilsonFermion5D.cc
index 7622205a..cbe66bb5 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.cc
+++ b/lib/qcd/action/fermion/WilsonFermion5D.cc
@@ -65,6 +65,7 @@ namespace QCD {
 
   // Allocate the required comms buffer
   comm_buf.resize(Stencil._unified_buffer_size); // this is always big enough to contain EO
+
   ImportGauge(_Umu);
 }  
 void WilsonFermion5D::ImportGauge(const LatticeGaugeField &_Umu)
@@ -122,6 +123,9 @@ void WilsonFermion5D::DerivInternal(CartesianStencil & st,
 {
   assert((dag==DaggerNo) ||(dag==DaggerYes));
 
+  conformable(st._grid,A._grid);
+  conformable(st._grid,B._grid);
+
   WilsonCompressor compressor(dag);
   
   LatticeColourMatrix tmp(mat._grid);
@@ -146,10 +150,14 @@ void WilsonFermion5D::DerivInternal(CartesianStencil & st,
     tmp = zero;
 
 PARALLEL_FOR_LOOP
-    for(int sss=0;sss<B._grid->oSites();sss++){
+    for(int sss=0;sss<U._grid->oSites();sss++){
       for(int s=0;s<Ls;s++){
 	int sU=sss;
 	int sF = s+Ls*sU;
+
+	assert ( sF< B._grid->oSites());
+	assert ( sU< U._grid->oSites());
+
 	DiracOptDhopDir(st,U,comm_buf,sF,sU,B,Btilde,mu,gamma);
 
     ////////////////////////////
@@ -161,10 +169,8 @@ PARALLEL_FOR_LOOP
 
       }
 
-
     }
 
-
     PokeIndex<LorentzIndex>(mat,tmp,mu);
 
   }
diff --git a/lib/qcd/action/fermion/WilsonKernels.cc b/lib/qcd/action/fermion/WilsonKernels.cc
index 43ccff63..2b8b7929 100644
--- a/lib/qcd/action/fermion/WilsonKernels.cc
+++ b/lib/qcd/action/fermion/WilsonKernels.cc
@@ -303,7 +303,7 @@ void DiracOptDhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
     vHalfSpinColourVector Uchi;
     int offset,local,perm, ptype;
     int ss=sF;
-    
+
     offset = st._offsets [dir][ss];
     local  = st._is_local[dir][ss];
     perm   = st._permute[dir][ss];
diff --git a/tests/Make.inc b/tests/Make.inc
index d510e98b..4367ad5d 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,5 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_cheby Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_hmc_WilsonFermionGauge Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd Test_wilson_force Test_wilson_force_phiMdagMphi Test_wilson_force_phiMphi
+bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_cheby Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_force Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_hmc_WilsonFermionGauge Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd Test_wilson_force Test_wilson_force_phiMdagMphi Test_wilson_force_phiMphi
 
 
 Test_GaugeAction_SOURCES=Test_GaugeAction.cc
@@ -70,6 +70,10 @@ Test_dwf_even_odd_SOURCES=Test_dwf_even_odd.cc
 Test_dwf_even_odd_LDADD=-lGrid
 
 
+Test_dwf_force_SOURCES=Test_dwf_force.cc
+Test_dwf_force_LDADD=-lGrid
+
+
 Test_dwf_fpgcr_SOURCES=Test_dwf_fpgcr.cc
 Test_dwf_fpgcr_LDADD=-lGrid
 

From 4866467e983b73b66b28130cd12474ab5c4932f3 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 1 Aug 2015 22:07:05 +0900
Subject: [PATCH 414/429] Test for DWF force term passes

---
 tests/Test_cheby.cc     |  61 +++++++++++++++++++
 tests/Test_dwf_force.cc | 126 ++++++++++++++++++++++++++++++++++++++++
 2 files changed, 187 insertions(+)
 create mode 100644 tests/Test_cheby.cc
 create mode 100644 tests/Test_dwf_force.cc

diff --git a/tests/Test_cheby.cc b/tests/Test_cheby.cc
new file mode 100644
index 00000000..08897c90
--- /dev/null
+++ b/tests/Test_cheby.cc
@@ -0,0 +1,61 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+RealD InverseApproximation(RealD x){
+  return 1.0/x;
+}
+RealD SqrtApproximation(RealD x){
+  return std::sqrt(x);
+}
+RealD StepFunction(RealD x){
+  if ( x<0.1 )  return 1.0;
+  else return 0.0;
+}
+
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  GridCartesian *grid = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), 
+						       GridDefaultSimd(Nd,vComplex::Nsimd()),
+						       GridDefaultMpi());
+
+  double     lo=0.1;
+  double     hi=64.0;
+
+  Chebyshev<LatticeFermion> ChebyInv(lo,hi,2000,InverseApproximation);
+
+
+  {
+    std::ofstream of("chebyinv");
+    ChebyInv.csv(of);
+  }
+
+  ChebyInv.JacksonSmooth();
+
+  {
+    std::ofstream of("chebyinvjack");
+    ChebyInv.csv(of);
+  }
+
+
+  Chebyshev<LatticeFermion> ChebyStep(lo,hi,200,StepFunction);
+  {
+    std::ofstream of("chebystep");
+    ChebyStep.csv(of);
+  }
+
+
+  ChebyStep.JacksonSmooth();
+
+  {
+    std::ofstream of("chebystepjack");
+    ChebyStep.csv(of);
+  }
+
+  Grid_finalize();
+}
diff --git a/tests/Test_dwf_force.cc b/tests/Test_dwf_force.cc
new file mode 100644
index 00000000..a3dbb0f2
--- /dev/null
+++ b/tests/Test_dwf_force.cc
@@ -0,0 +1,126 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+#define parallel_for PARALLEL_FOR_LOOP for
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+
+  const int Ls=8;
+
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);  RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);  RNG4.SeedFixedIntegers(seeds4);
+
+  int threads = GridThread::GetThreads();
+  std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+
+  LatticeFermion phi        (FGrid); gaussian(RNG5,phi);
+  LatticeFermion Mphi       (FGrid); 
+  LatticeFermion MphiPrime  (FGrid); 
+
+  LatticeGaugeField U(UGrid);
+
+  SU3::HotConfiguration(RNG4,U);
+  
+  ////////////////////////////////////
+  // Unmodified matrix element
+  ////////////////////////////////////
+  RealD mass=0.01; 
+  RealD M5=1.8; 
+  DomainWallFermion Ddwf(U,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  Ddwf.M   (phi,Mphi);
+
+  ComplexD S    = innerProduct(Mphi,Mphi); // pdag MdagM p
+
+  // get the deriv of phidag MdagM phi with respect to "U"
+  LatticeGaugeField UdSdU(UGrid);
+  LatticeGaugeField tmp(UGrid);
+  
+
+  Ddwf.MDeriv(tmp , Mphi,  phi,DaggerNo );  UdSdU=tmp;
+  Ddwf.MDeriv(tmp , phi,  Mphi,DaggerYes ); UdSdU=(UdSdU+tmp);  
+  
+  LatticeFermion Ftmp      (FGrid);
+
+  ////////////////////////////////////
+  // Modify the gauge field a little 
+  ////////////////////////////////////
+  RealD dt = 0.0001;
+
+  LatticeColourMatrix mommu(UGrid); 
+  LatticeColourMatrix forcemu(UGrid); 
+  LatticeGaugeField mom(UGrid); 
+  LatticeGaugeField Uprime(UGrid); 
+
+  for(int mu=0;mu<Nd;mu++){
+
+    SU3::GaussianLieAlgebraMatrix(RNG4, mommu); // Traceless antihermitian momentum; gaussian in lie alg
+
+    PokeIndex<LorentzIndex>(mom,mommu,mu);
+
+    // fourth order exponential approx
+    parallel_for(auto i=mom.begin();i<mom.end();i++){
+      Uprime[i](mu) =
+	  U[i](mu)
+	+ mom[i](mu)*U[i](mu)*dt 
+	+ mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt/2.0)
+	+ mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt/6.0)
+	+ mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt*dt/24.0)
+	+ mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt*dt*dt/120.0)
+	+ mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt*dt*dt*dt/720.0)
+	;
+    }
+
+  }
+  
+  Ddwf.ImportGauge(Uprime);
+  Ddwf.M          (phi,MphiPrime);
+
+  ComplexD Sprime    = innerProduct(MphiPrime   ,MphiPrime);
+
+  //////////////////////////////////////////////
+  // Use derivative to estimate dS
+  //////////////////////////////////////////////
+
+
+  LatticeComplex dS(UGrid); dS = zero;
+  for(int mu=0;mu<Nd;mu++){
+    mommu   = PeekIndex<LorentzIndex>(UdSdU,mu);
+    mommu=Ta(mommu)*2.0;
+    PokeIndex<LorentzIndex>(UdSdU,mommu,mu);
+  }
+
+  for(int mu=0;mu<Nd;mu++){
+    forcemu = PeekIndex<LorentzIndex>(UdSdU,mu);
+    mommu   = PeekIndex<LorentzIndex>(mom,mu);
+
+    // Update PF action density
+    dS = dS+trace(mommu*forcemu)*dt;
+
+  }
+
+  Complex dSpred    = sum(dS);
+
+  std::cout << GridLogMessage << " S      "<<S<<std::endl;
+  std::cout << GridLogMessage << " Sprime "<<Sprime<<std::endl;
+  std::cout << GridLogMessage << "dS      "<<Sprime-S<<std::endl;
+  std::cout << GridLogMessage << "predict dS    "<< dSpred <<std::endl;
+
+  std::cout<< GridLogMessage << "Done" <<std::endl;
+  Grid_finalize();
+}

From 742db5d8b4501d3a55f5b549e42627c288a292f5 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 1 Aug 2015 22:30:00 +0900
Subject: [PATCH 415/429] Merge problem fixed

---
 configure                            | 13 +++++++------
 tests/Test_hmc_WilsonFermionGauge.cc |  2 +-
 2 files changed, 8 insertions(+), 7 deletions(-)

diff --git a/configure b/configure
index e02b9e7e..7d7e1b0b 100755
--- a/configure
+++ b/configure
@@ -2560,7 +2560,7 @@ test -n "$target_alias" &&
     NONENONEs,x,x, &&
   program_prefix=${target_alias}-
 
-am__api_version='1.14'
+am__api_version='1.15'
 
 # Find a good install program.  We prefer a C program (faster),
 # so one script is as good as another.  But avoid the broken or
@@ -2732,8 +2732,8 @@ test "$program_suffix" != NONE &&
 ac_script='s/[\\$]/&&/g;s/;s,x,x,$//'
 program_transform_name=`$as_echo "$program_transform_name" | sed "$ac_script"`
 
-# expand $ac_aux_dir to an absolute path
-am_aux_dir=`cd $ac_aux_dir && pwd`
+# Expand $ac_aux_dir to an absolute path.
+am_aux_dir=`cd "$ac_aux_dir" && pwd`
 
 if test x"${MISSING+set}" != xset; then
   case $am_aux_dir in
@@ -2752,7 +2752,7 @@ else
 $as_echo "$as_me: WARNING: 'missing' script is too old or missing" >&2;}
 fi
 
-if test x"${install_sh}" != xset; then
+if test x"${install_sh+set}" != xset; then
   case $am_aux_dir in
   *\ * | *\	*)
     install_sh="\${SHELL} '$am_aux_dir/install-sh'" ;;
@@ -3080,8 +3080,8 @@ MAKEINFO=${MAKEINFO-"${am_missing_run}makeinfo"}
 # <http://lists.gnu.org/archive/html/automake/2012-07/msg00014.html>
 mkdir_p='$(MKDIR_P)'
 
-# We need awk for the "check" target.  The system "awk" is bad on
-# some platforms.
+# We need awk for the "check" target (and possibly the TAP driver).  The
+# system "awk" is bad on some platforms.
 # Always define AMTAR for backward compatibility.  Yes, it's still used
 # in the wild :-(  We should find a proper way to deprecate it ...
 AMTAR='$${TAR-tar}'
@@ -3140,6 +3140,7 @@ END
 fi
 
 
+
 ac_config_headers="$ac_config_headers lib/Config.h"
 
 # Check whether --enable-silent-rules was given.
diff --git a/tests/Test_hmc_WilsonFermionGauge.cc b/tests/Test_hmc_WilsonFermionGauge.cc
index e40ad32f..cb18d09c 100644
--- a/tests/Test_hmc_WilsonFermionGauge.cc
+++ b/tests/Test_hmc_WilsonFermionGauge.cc
@@ -47,7 +47,7 @@ int main (int argc, char ** argv)
   //  typedef LeapFrog  IntegratorAlgorithm;// change here to change the algorithm
   IntegratorParameters MDpar(12,20,1.0);
   std::vector<int> rel ={1};
-  Integrator<IntegratorAlgorithm> MDynamics(&Fine,MDpar, FullSet,rel);
+  Integrator<IntegratorAlgorithm> MDynamics(&Fine,MDpar, FullSet);
 
   // Create HMC
   HMCparameters HMCpar;

From d98e8366a0c6a1df988efb8ed6f141497228049e Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Sat, 1 Aug 2015 22:48:21 +0900
Subject: [PATCH 416/429] Continued fraction overlap, partial fraction overlap
 force terms have a successful test passing.

---
 tests/Make.inc               |  10 ++-
 tests/Test_contfrac_force.cc | 126 +++++++++++++++++++++++++++++++++++
 tests/Test_partfrac_force.cc | 126 +++++++++++++++++++++++++++++++++++
 3 files changed, 261 insertions(+), 1 deletion(-)
 create mode 100644 tests/Test_contfrac_force.cc
 create mode 100644 tests/Test_partfrac_force.cc

diff --git a/tests/Make.inc b/tests/Make.inc
index 4367ad5d..0c8332a7 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,5 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_cheby Test_contfrac_cg Test_contfrac_even_odd Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_force Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_hmc_WilsonFermionGauge Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd Test_wilson_force Test_wilson_force_phiMdagMphi Test_wilson_force_phiMphi
+bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_cheby Test_contfrac_cg Test_contfrac_even_odd Test_contfrac_force Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_force Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_hmc_WilsonFermionGauge Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_partfrac_force Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd Test_wilson_force Test_wilson_force_phiMdagMphi Test_wilson_force_phiMphi
 
 
 Test_GaugeAction_SOURCES=Test_GaugeAction.cc
@@ -42,6 +42,10 @@ Test_contfrac_even_odd_SOURCES=Test_contfrac_even_odd.cc
 Test_contfrac_even_odd_LDADD=-lGrid
 
 
+Test_contfrac_force_SOURCES=Test_contfrac_force.cc
+Test_contfrac_force_LDADD=-lGrid
+
+
 Test_cshift_SOURCES=Test_cshift.cc
 Test_cshift_LDADD=-lGrid
 
@@ -110,6 +114,10 @@ Test_nersc_io_SOURCES=Test_nersc_io.cc
 Test_nersc_io_LDADD=-lGrid
 
 
+Test_partfrac_force_SOURCES=Test_partfrac_force.cc
+Test_partfrac_force_LDADD=-lGrid
+
+
 Test_quenched_update_SOURCES=Test_quenched_update.cc
 Test_quenched_update_LDADD=-lGrid
 
diff --git a/tests/Test_contfrac_force.cc b/tests/Test_contfrac_force.cc
new file mode 100644
index 00000000..a39d596d
--- /dev/null
+++ b/tests/Test_contfrac_force.cc
@@ -0,0 +1,126 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+#define parallel_for PARALLEL_FOR_LOOP for
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+
+  const int Ls=9;
+
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);  RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);  RNG4.SeedFixedIntegers(seeds4);
+
+  int threads = GridThread::GetThreads();
+  std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+
+  LatticeFermion phi        (FGrid); gaussian(RNG5,phi);
+  LatticeFermion Mphi       (FGrid); 
+  LatticeFermion MphiPrime  (FGrid); 
+
+  LatticeGaugeField U(UGrid);
+
+  SU3::HotConfiguration(RNG4,U);
+  
+  ////////////////////////////////////
+  // Unmodified matrix element
+  ////////////////////////////////////
+  RealD mass=0.01; 
+  RealD M5=1.8; 
+  OverlapWilsonContFracTanhFermion Dcf(U,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+  Dcf.M   (phi,Mphi);
+
+  ComplexD S    = innerProduct(Mphi,Mphi); // pdag MdagM p
+
+  // get the deriv of phidag MdagM phi with respect to "U"
+  LatticeGaugeField UdSdU(UGrid);
+  LatticeGaugeField tmp(UGrid);
+  
+
+  Dcf.MDeriv(tmp , Mphi,  phi,DaggerNo );  UdSdU=tmp;
+  Dcf.MDeriv(tmp , phi,  Mphi,DaggerYes ); UdSdU=(UdSdU+tmp);  
+  
+  LatticeFermion Ftmp      (FGrid);
+
+  ////////////////////////////////////
+  // Modify the gauge field a little 
+  ////////////////////////////////////
+  RealD dt = 0.0001;
+
+  LatticeColourMatrix mommu(UGrid); 
+  LatticeColourMatrix forcemu(UGrid); 
+  LatticeGaugeField mom(UGrid); 
+  LatticeGaugeField Uprime(UGrid); 
+
+  for(int mu=0;mu<Nd;mu++){
+
+    SU3::GaussianLieAlgebraMatrix(RNG4, mommu); // Traceless antihermitian momentum; gaussian in lie alg
+
+    PokeIndex<LorentzIndex>(mom,mommu,mu);
+
+    // fourth order exponential approx
+    parallel_for(auto i=mom.begin();i<mom.end();i++){
+      Uprime[i](mu) =
+	  U[i](mu)
+	+ mom[i](mu)*U[i](mu)*dt 
+	+ mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt/2.0)
+	+ mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt/6.0)
+	+ mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt*dt/24.0)
+	+ mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt*dt*dt/120.0)
+	+ mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt*dt*dt*dt/720.0)
+	;
+    }
+
+  }
+  
+  Dcf.ImportGauge(Uprime);
+  Dcf.M          (phi,MphiPrime);
+
+  ComplexD Sprime    = innerProduct(MphiPrime   ,MphiPrime);
+
+  //////////////////////////////////////////////
+  // Use derivative to estimate dS
+  //////////////////////////////////////////////
+
+
+  LatticeComplex dS(UGrid); dS = zero;
+  for(int mu=0;mu<Nd;mu++){
+    mommu   = PeekIndex<LorentzIndex>(UdSdU,mu);
+    mommu=Ta(mommu)*2.0;
+    PokeIndex<LorentzIndex>(UdSdU,mommu,mu);
+  }
+
+  for(int mu=0;mu<Nd;mu++){
+    forcemu = PeekIndex<LorentzIndex>(UdSdU,mu);
+    mommu   = PeekIndex<LorentzIndex>(mom,mu);
+
+    // Update PF action density
+    dS = dS+trace(mommu*forcemu)*dt;
+
+  }
+
+  Complex dSpred    = sum(dS);
+
+  std::cout << GridLogMessage << " S      "<<S<<std::endl;
+  std::cout << GridLogMessage << " Sprime "<<Sprime<<std::endl;
+  std::cout << GridLogMessage << "dS      "<<Sprime-S<<std::endl;
+  std::cout << GridLogMessage << "predict dS    "<< dSpred <<std::endl;
+
+  std::cout<< GridLogMessage << "Done" <<std::endl;
+  Grid_finalize();
+}
diff --git a/tests/Test_partfrac_force.cc b/tests/Test_partfrac_force.cc
new file mode 100644
index 00000000..05925e55
--- /dev/null
+++ b/tests/Test_partfrac_force.cc
@@ -0,0 +1,126 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+#define parallel_for PARALLEL_FOR_LOOP for
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+
+  const int Ls=9;
+
+  GridCartesian         * UGrid   = SpaceTimeGrid::makeFourDimGrid(GridDefaultLatt(), GridDefaultSimd(Nd,vComplex::Nsimd()),GridDefaultMpi());
+  GridRedBlackCartesian * UrbGrid = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid);
+  GridCartesian         * FGrid   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid);
+  GridRedBlackCartesian * FrbGrid = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid);
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  GridParallelRNG          RNG5(FGrid);  RNG5.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4(UGrid);  RNG4.SeedFixedIntegers(seeds4);
+
+  int threads = GridThread::GetThreads();
+  std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+
+  LatticeFermion phi        (FGrid); gaussian(RNG5,phi);
+  LatticeFermion Mphi       (FGrid); 
+  LatticeFermion MphiPrime  (FGrid); 
+
+  LatticeGaugeField U(UGrid);
+
+  SU3::HotConfiguration(RNG4,U);
+  
+  ////////////////////////////////////
+  // Unmodified matrix element
+  ////////////////////////////////////
+  RealD mass=0.01; 
+  RealD M5=1.8; 
+  OverlapWilsonPartialFractionTanhFermion Dpf(U,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+  Dpf.M   (phi,Mphi);
+
+  ComplexD S    = innerProduct(Mphi,Mphi); // pdag MdagM p
+
+  // get the deriv of phidag MdagM phi with respect to "U"
+  LatticeGaugeField UdSdU(UGrid);
+  LatticeGaugeField tmp(UGrid);
+  
+
+  Dpf.MDeriv(tmp , Mphi,  phi,DaggerNo );  UdSdU=tmp;
+  Dpf.MDeriv(tmp , phi,  Mphi,DaggerYes ); UdSdU=(UdSdU+tmp);  
+  
+  LatticeFermion Ftmp      (FGrid);
+
+  ////////////////////////////////////
+  // Modify the gauge field a little 
+  ////////////////////////////////////
+  RealD dt = 0.0001;
+
+  LatticeColourMatrix mommu(UGrid); 
+  LatticeColourMatrix forcemu(UGrid); 
+  LatticeGaugeField mom(UGrid); 
+  LatticeGaugeField Uprime(UGrid); 
+
+  for(int mu=0;mu<Nd;mu++){
+
+    SU3::GaussianLieAlgebraMatrix(RNG4, mommu); // Traceless antihermitian momentum; gaussian in lie alg
+
+    PokeIndex<LorentzIndex>(mom,mommu,mu);
+
+    // fourth order exponential approx
+    parallel_for(auto i=mom.begin();i<mom.end();i++){
+      Uprime[i](mu) =
+	  U[i](mu)
+	+ mom[i](mu)*U[i](mu)*dt 
+	+ mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt/2.0)
+	+ mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt/6.0)
+	+ mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt*dt/24.0)
+	+ mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt*dt*dt/120.0)
+	+ mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *mom[i](mu) *U[i](mu)*(dt*dt*dt*dt*dt*dt/720.0)
+	;
+    }
+
+  }
+  
+  Dpf.ImportGauge(Uprime);
+  Dpf.M          (phi,MphiPrime);
+
+  ComplexD Sprime    = innerProduct(MphiPrime   ,MphiPrime);
+
+  //////////////////////////////////////////////
+  // Use derivative to estimate dS
+  //////////////////////////////////////////////
+
+
+  LatticeComplex dS(UGrid); dS = zero;
+  for(int mu=0;mu<Nd;mu++){
+    mommu   = PeekIndex<LorentzIndex>(UdSdU,mu);
+    mommu=Ta(mommu)*2.0;
+    PokeIndex<LorentzIndex>(UdSdU,mommu,mu);
+  }
+
+  for(int mu=0;mu<Nd;mu++){
+    forcemu = PeekIndex<LorentzIndex>(UdSdU,mu);
+    mommu   = PeekIndex<LorentzIndex>(mom,mu);
+
+    // Update PF action density
+    dS = dS+trace(mommu*forcemu)*dt;
+
+  }
+
+  Complex dSpred    = sum(dS);
+
+  std::cout << GridLogMessage << " S      "<<S<<std::endl;
+  std::cout << GridLogMessage << " Sprime "<<Sprime<<std::endl;
+  std::cout << GridLogMessage << "dS      "<<Sprime-S<<std::endl;
+  std::cout << GridLogMessage << "predict dS    "<< dSpred <<std::endl;
+
+  std::cout<< GridLogMessage << "Done" <<std::endl;
+  Grid_finalize();
+}

From ce34856e32f8e77fe18270653b0dd400cf060bb2 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 7 Aug 2015 08:37:15 +0100
Subject: [PATCH 417/429] Adding components for even odd decomposed determinant
 in HMC. dH not yet conserved, so something wrong in the eo force code still

---
 lib/algorithms/LinearOperator.h               |   3 +
 lib/qcd/action/Actions.h                      |   1 +
 lib/qcd/action/pseudofermion/TwoFlavour.h     |  19 +-
 .../action/pseudofermion/TwoFlavourEvenOdd.h  | 226 ++++++++++++++++++
 tests/Make.inc                                |   6 +-
 tests/Test_hmc_EOWilsonFermionGauge.cc        |  58 +++++
 tests/Test_hmc_WilsonFermionGauge.cc          |   2 +-
 7 files changed, 302 insertions(+), 13 deletions(-)
 create mode 100644 lib/qcd/action/pseudofermion/TwoFlavourEvenOdd.h
 create mode 100644 tests/Test_hmc_EOWilsonFermionGauge.cc

diff --git a/lib/algorithms/LinearOperator.h b/lib/algorithms/LinearOperator.h
index 31e086ba..c49d7c7d 100644
--- a/lib/algorithms/LinearOperator.h
+++ b/lib/algorithms/LinearOperator.h
@@ -188,6 +188,7 @@ namespace Grid {
     };
     template<class Matrix,class Field>
       class SchurDiagMooeeOperator :  public SchurOperatorBase<Field> {
+    protected:
       Matrix &_Mat;
     public:
       SchurDiagMooeeOperator (Matrix &Mat): _Mat(Mat){};
@@ -214,6 +215,7 @@ namespace Grid {
     };
     template<class Matrix,class Field>
       class SchurDiagOneOperator :  public SchurOperatorBase<Field> {
+    protected:
       Matrix &_Mat;
     public:
       SchurDiagOneOperator (Matrix &Mat): _Mat(Mat){};
@@ -240,6 +242,7 @@ namespace Grid {
       }
     };
 
+
     /////////////////////////////////////////////////////////////
     // Base classes for functions of operators
     /////////////////////////////////////////////////////////////
diff --git a/lib/qcd/action/Actions.h b/lib/qcd/action/Actions.h
index 26f26b8f..f40cd554 100644
--- a/lib/qcd/action/Actions.h
+++ b/lib/qcd/action/Actions.h
@@ -83,6 +83,7 @@
 // Pseudo fermion combinations
 ////////////////////////////////////////
 #include <qcd/action/pseudofermion/TwoFlavour.h>
+#include <qcd/action/pseudofermion/TwoFlavourEvenOdd.h>
 
 
 #endif
diff --git a/lib/qcd/action/pseudofermion/TwoFlavour.h b/lib/qcd/action/pseudofermion/TwoFlavour.h
index 68de0756..22f761fe 100644
--- a/lib/qcd/action/pseudofermion/TwoFlavour.h
+++ b/lib/qcd/action/pseudofermion/TwoFlavour.h
@@ -106,8 +106,6 @@ namespace Grid{
 
       OperatorFunction<FermionField> &ActionSolver;
 
-      GridBase &Grid;
-
       FermionField Phi; // the pseudo fermion field for this trajectory
 
     public:
@@ -116,9 +114,8 @@ namespace Grid{
       /////////////////////////////////////////////////
     TwoFlavourPseudoFermionAction(FermionOperator<FermionField,GaugeField>  &Op, 
 				  OperatorFunction<FermionField> & DS,
-				  OperatorFunction<FermionField> & AS,
-				  GridBase &_Grid
-				  ) : FermOp(Op), DerivativeSolver(DS), ActionSolver(AS), Phi(&_Grid), Grid(_Grid) {
+				  OperatorFunction<FermionField> & AS
+				  ) : FermOp(Op), DerivativeSolver(DS), ActionSolver(AS), Phi(Op.FermionGrid()) {
       };
       
       //////////////////////////////////////////////////////////////////////////////////////
@@ -140,7 +137,7 @@ namespace Grid{
 	//         in the Phi integral, and thus is only an irrelevant prefactor for the partition function.
 	//
 	RealD scale = std::sqrt(0.5);
-	FermionField eta(&Grid);
+	FermionField eta(FermOp.FermionGrid());
 
 	gaussian(pRNG,eta);
 
@@ -158,8 +155,8 @@ namespace Grid{
 
 	FermOp.ImportGauge(U);
 
-	FermionField X(&Grid);
-	FermionField Y(&Grid);
+	FermionField X(FermOp.FermionGrid());
+	FermionField Y(FermOp.FermionGrid());
 	
 	MdagMLinearOperator<FermionOperator<FermionField,GaugeField> ,FermionField> MdagMOp(FermOp);
 	X=zero;
@@ -182,9 +179,9 @@ namespace Grid{
 
 	FermOp.ImportGauge(U);
 
-	FermionField X(&Grid);
-	FermionField Y(&Grid);
-	GaugeField   tmp(&Grid);
+	FermionField X(FermOp.FermionGrid());
+	FermionField Y(FermOp.FermionGrid());
+	GaugeField   tmp(FermOp.GaugeGrid());
 
 	MdagMLinearOperator<FermionOperator<FermionField,GaugeField> ,FermionField> MdagMOp(FermOp);
 
diff --git a/lib/qcd/action/pseudofermion/TwoFlavourEvenOdd.h b/lib/qcd/action/pseudofermion/TwoFlavourEvenOdd.h
new file mode 100644
index 00000000..63dec41a
--- /dev/null
+++ b/lib/qcd/action/pseudofermion/TwoFlavourEvenOdd.h
@@ -0,0 +1,226 @@
+#ifndef QCD_PSEUDOFERMION_TWO_FLAVOUR_EVEN_ODD_H
+#define QCD_PSEUDOFERMION_TWO_FLAVOUR_EVEN_ODD_H
+
+namespace Grid{
+  namespace QCD{
+
+    template<class Matrix,class FermionField>
+      class SchurDifferentiableOperator :  public SchurDiagMooeeOperator<Matrix,FermionField> 
+      {
+    public:
+	SchurDifferentiableOperator (Matrix &Mat) : SchurDiagMooeeOperator<Matrix,FermionField>(Mat) {};
+
+	void MpcDeriv(LatticeGaugeField &Force,const FermionField &U,const FermionField &V) {
+	
+	  GridBase *fgrid   = this->_Mat.FermionGrid();
+	  GridBase *fcbgrid = this->_Mat.FermionRedBlackGrid();
+	  GridBase *ugrid = this->_Mat.GaugeGrid();
+	  GridBase *ucbgrid = this->_Mat.GaugeRedBlackGrid();
+
+	  Real coeff = 1.0;
+	  FermionField tmp1(fcbgrid);
+	  FermionField tmp2(fcbgrid);
+
+	  conformable(fcbgrid,U._grid);
+	  conformable(fcbgrid,V._grid);
+
+	  // Assert the checkerboard?? or code for either
+	  assert(U.checkerboard==Odd);
+	  assert(V.checkerboard==V.checkerboard);
+
+	  LatticeGaugeField ForceO(ucbgrid);
+	  LatticeGaugeField ForceE(ucbgrid);
+
+	  //  X^dag Der_oe MeeInv Meo Y
+	  // Use Mooee as nontrivial but gauge field indept
+	  this->_Mat.Meooe   (V,tmp1);      // odd->even -- implicit -0.5 factor to be applied
+	  this->_Mat.MooeeInv(tmp1,tmp2);   // even->even 
+	  this->_Mat.MoeDeriv(ForceO,U,tmp2,DaggerNo);
+	  
+	  //  Accumulate X^dag M_oe MeeInv Der_eo Y
+	  this->_Mat.MeooeDag   (U,tmp1);    // even->odd -- implicit -0.5 factor to be applied
+	  this->_Mat.MooeeInvDag(tmp1,tmp2); // even->even 
+	  this->_Mat.MeoDeriv(ForceE,tmp2,V,DaggerNo);
+	  
+	  setCheckerboard(Force,ForceE); 
+	  setCheckerboard(Force,ForceO);
+	  Force=-Force;
+	}
+
+
+	void MpcDagDeriv(LatticeGaugeField &Force,const FermionField &U,const FermionField &V) {
+	
+	  GridBase *fgrid   = this->_Mat.FermionGrid();
+	  GridBase *fcbgrid = this->_Mat.FermionRedBlackGrid();
+	  GridBase *ugrid   = this->_Mat.GaugeGrid();
+	  GridBase *ucbgrid = this->_Mat.GaugeRedBlackGrid();
+
+	  Real coeff = 1.0;
+	  FermionField tmp1(fcbgrid);
+	  FermionField tmp2(fcbgrid);
+
+	  conformable(fcbgrid,U._grid);
+	  conformable(fcbgrid,V._grid);
+
+	  // Assert the checkerboard?? or code for either
+	  assert(V.checkerboard==Odd);
+	  assert(V.checkerboard==V.checkerboard);
+
+	  LatticeGaugeField ForceO(ucbgrid);
+	  LatticeGaugeField ForceE(ucbgrid);
+
+	  //  X^dag Der_oe MeeInv Meo Y
+	  // Use Mooee as nontrivial but gauge field indept
+	  this->_Mat.MeooeDag   (V,tmp1);      // odd->even -- implicit -0.5 factor to be applied
+	  this->_Mat.MooeeInvDag(tmp1,tmp2);   // even->even 
+	  this->_Mat.MoeDeriv(ForceO,U,tmp2,DaggerYes);
+	  
+	  //  Accumulate X^dag M_oe MeeInv Der_eo Y
+	  this->_Mat.Meooe   (U,tmp1);    // even->odd -- implicit -0.5 factor to be applied
+	  this->_Mat.MooeeInv(tmp1,tmp2); // even->even 
+	  this->_Mat.MeoDeriv(ForceE,tmp2,V,DaggerYes);
+
+	  setCheckerboard(Force,ForceE); 
+	  setCheckerboard(Force,ForceO);
+	  Force=-Force;
+	}
+
+    };
+
+
+    ////////////////////////////////////////////////////////////////////////
+    // Two flavour pseudofermion action for any EO prec dop
+    ////////////////////////////////////////////////////////////////////////
+    template<class GaugeField,class MatrixField,class FermionField>
+      class TwoFlavourEvenOddPseudoFermionAction : public Action<GaugeField> {
+
+    private:
+      
+      FermionOperator<FermionField,GaugeField> & FermOp;// the basic operator
+
+      OperatorFunction<FermionField> &DerivativeSolver;
+
+      OperatorFunction<FermionField> &ActionSolver;
+
+      FermionField PhiOdd;   // the pseudo fermion field for this trajectory
+      FermionField PhiEven;  // the pseudo fermion field for this trajectory
+
+    public:
+      /////////////////////////////////////////////////
+      // Pass in required objects.
+      /////////////////////////////////////////////////
+      TwoFlavourEvenOddPseudoFermionAction(FermionOperator<FermionField,GaugeField>  &Op, 
+					 OperatorFunction<FermionField> & DS,
+					 OperatorFunction<FermionField> & AS
+					   ) : 
+        FermOp(Op), 
+	DerivativeSolver(DS), 
+	ActionSolver(AS), 
+        PhiEven(Op.FermionRedBlackGrid()),
+	PhiOdd(Op.FermionRedBlackGrid())
+		  {};
+      
+      //////////////////////////////////////////////////////////////////////////////////////
+      // Push the gauge field in to the dops. Assume any BC's and smearing already applied
+      //////////////////////////////////////////////////////////////////////////////////////
+      virtual void init(const GaugeField &U, GridParallelRNG& pRNG) {
+
+	// P(phi) = e^{- phi^dag (MpcdagMpc)^-1 phi}
+	// Phi = McpDag eta 
+	// P(eta) = e^{- eta^dag eta}
+	//
+	// e^{x^2/2 sig^2} => sig^2 = 0.5.
+	RealD scale = std::sqrt(0.5);
+
+	FermionField eta    (FermOp.FermionGrid());
+	FermionField etaOdd (FermOp.FermionRedBlackGrid());
+	FermionField etaEven(FermOp.FermionRedBlackGrid());
+
+	gaussian(pRNG,eta);
+	pickCheckerboard(Even,etaEven,eta);
+	pickCheckerboard(Odd,etaOdd,eta);
+
+	SchurDifferentiableOperator<FermionOperator<FermionField,GaugeField>,FermionField> PCop(FermOp);
+	
+	FermOp.ImportGauge(U);
+
+	PCop.MpcDag(etaOdd,PhiOdd);
+	FermOp.MooeeDag(etaEven,PhiEven);
+
+	PhiOdd =PhiOdd*scale;
+	PhiEven=PhiEven*scale;
+	
+      };
+
+      //////////////////////////////////////////////////////
+      // S = phi^dag (Mdag M)^-1 phi  (odd)
+      //   + phi^dag (Mdag M)^-1 phi  (even)
+      //////////////////////////////////////////////////////
+      virtual RealD S(const GaugeField &U) {
+
+	FermOp.ImportGauge(U);
+
+	FermionField X(FermOp.FermionRedBlackGrid());
+	FermionField Y(FermOp.FermionRedBlackGrid());
+	
+	SchurDifferentiableOperator<FermionOperator<FermionField,GaugeField>,FermionField> PCop(FermOp);
+
+	X=zero;
+	ActionSolver(PCop,PhiOdd,X);
+	PCop.Op(X,Y);
+	RealD action = norm2(Y);
+
+	// The EE factorised block; normally can replace with zero if det is constant (gauge field indept)
+	// Only really clover term that creates this.
+	FermOp.MooeeInvDag(PhiEven,Y);
+	action = action + norm2(Y);
+
+	std::cout << GridLogMessage << "Pseudofermion EO action "<<action<<std::endl;
+	return action;
+      };
+
+      //////////////////////////////////////////////////////
+      //
+      // dS/du = - phi^dag  (Mdag M)^-1 [ Mdag dM + dMdag M ]  (Mdag M)^-1 phi
+      //       = - phi^dag M^-1 dM (MdagM)^-1 phi -  phi^dag (MdagM)^-1 dMdag dM (Mdag)^-1 phi 
+      //
+      //       = - Ydag dM X  - Xdag dMdag Y
+      //
+      //////////////////////////////////////////////////////
+      virtual void deriv(const GaugeField &U,GaugeField & dSdU) {
+
+	FermOp.ImportGauge(U);
+
+	FermionField X(FermOp.FermionRedBlackGrid());
+	FermionField Y(FermOp.FermionRedBlackGrid());
+	GaugeField tmp(FermOp.GaugeGrid());
+
+	SchurDifferentiableOperator<FermionOperator<FermionField,GaugeField>,FermionField> PCop(FermOp);
+
+	X=zero;
+	DerivativeSolver(PCop,PhiOdd,X);
+	PCop.Op(X,Y);
+
+	// Our conventions really make this UdSdU; We do not differentiate wrt Udag here.
+	// So must take dSdU - adj(dSdU) and left multiply by mom to get dS/dt.
+
+  	PCop.MpcDeriv(tmp , Y, X );    dSdU=tmp;
+	PCop.MpcDagDeriv(tmp , X, Y);  dSdU=dSdU+tmp;
+
+	// Treat the EE case. (MdagM)^-1 = Minv Minvdag
+	// Deriv defaults to zero.
+        FermOp.MooeeInvDag(PhiOdd,Y);
+        FermOp.MooeeInv(Y,X);
+  	FermOp.MeeDeriv(tmp , Y, X,DaggerNo );    dSdU=tmp;
+	FermOp.MeeDeriv(tmp , X, Y,DaggerYes);  dSdU=dSdU+tmp;
+	
+	dSdU = Ta(dSdU);
+
+      };
+
+    };
+    
+  }
+}
+
+#endif
diff --git a/tests/Make.inc b/tests/Make.inc
index 0c8332a7..d5b926e6 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,5 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_cheby Test_contfrac_cg Test_contfrac_even_odd Test_contfrac_force Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_force Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_hmc_WilsonFermionGauge Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_partfrac_force Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd Test_wilson_force Test_wilson_force_phiMdagMphi Test_wilson_force_phiMphi
+bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_cheby Test_contfrac_cg Test_contfrac_even_odd Test_contfrac_force Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_force Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_hmc_EOWilsonFermionGauge Test_hmc_WilsonFermionGauge Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_partfrac_force Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd Test_wilson_force Test_wilson_force_phiMdagMphi Test_wilson_force_phiMphi
 
 
 Test_GaugeAction_SOURCES=Test_GaugeAction.cc
@@ -90,6 +90,10 @@ Test_gamma_SOURCES=Test_gamma.cc
 Test_gamma_LDADD=-lGrid
 
 
+Test_hmc_EOWilsonFermionGauge_SOURCES=Test_hmc_EOWilsonFermionGauge.cc
+Test_hmc_EOWilsonFermionGauge_LDADD=-lGrid
+
+
 Test_hmc_WilsonFermionGauge_SOURCES=Test_hmc_WilsonFermionGauge.cc
 Test_hmc_WilsonFermionGauge_LDADD=-lGrid
 
diff --git a/tests/Test_hmc_EOWilsonFermionGauge.cc b/tests/Test_hmc_EOWilsonFermionGauge.cc
new file mode 100644
index 00000000..1c90933e
--- /dev/null
+++ b/tests/Test_hmc_EOWilsonFermionGauge.cc
@@ -0,0 +1,58 @@
+#include "Grid.h"
+
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(4,vComplex::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+  
+  GridCartesian            Fine(latt_size,simd_layout,mpi_layout);
+  GridRedBlackCartesian  RBFine(latt_size,simd_layout,mpi_layout);
+  GridParallelRNG  pRNG(&Fine);
+  pRNG.SeedRandomDevice();
+  LatticeLorentzColourMatrix     U(&Fine);
+
+  SU3::HotConfiguration(pRNG, U);
+
+  // simplify template declaration? Strip the lorentz from the second template
+  WilsonGaugeAction<LatticeLorentzColourMatrix, LatticeColourMatrix> Waction(5.6);
+
+  Real mass=-0.77;
+  WilsonFermion FermOp(U,Fine,RBFine,mass);
+  
+  ConjugateGradient<LatticeFermion>  CG(1.0e-8,10000);
+  ConjugateGradient<LatticeFermion>  CGmd(1.0e-6,10000);
+  
+  TwoFlavourEvenOddPseudoFermionAction<LatticeLorentzColourMatrix, LatticeColourMatrix,LatticeFermion> 
+    WilsonNf2(FermOp,CG,CG);
+  
+  //Collect actions
+  ActionLevel Level1;
+  Level1.push_back(&WilsonNf2);
+  ActionLevel Level2(3);
+  Level2.push_back(&Waction);
+  ActionSet FullSet;
+  FullSet.push_back(Level1);
+  FullSet.push_back(Level2);
+
+  // Create integrator
+  typedef MinimumNorm2  IntegratorAlgorithm;// change here to change the algorithm
+  //  typedef LeapFrog  IntegratorAlgorithm;// change here to change the algorithm
+  IntegratorParameters MDpar(12,20,1.0);
+  std::vector<int> rel ={1};
+  Integrator<IntegratorAlgorithm> MDynamics(&Fine,MDpar, FullSet);
+
+  // Create HMC
+  HMCparameters HMCpar;
+  HybridMonteCarlo<IntegratorAlgorithm>  HMC(HMCpar, MDynamics);
+
+  HMC.evolve(U);
+
+}
diff --git a/tests/Test_hmc_WilsonFermionGauge.cc b/tests/Test_hmc_WilsonFermionGauge.cc
index cb18d09c..e3bbf393 100644
--- a/tests/Test_hmc_WilsonFermionGauge.cc
+++ b/tests/Test_hmc_WilsonFermionGauge.cc
@@ -31,7 +31,7 @@ int main (int argc, char ** argv)
   ConjugateGradient<LatticeFermion>  CGmd(1.0e-6,10000);
   
   TwoFlavourPseudoFermionAction<LatticeLorentzColourMatrix, LatticeColourMatrix,LatticeFermion> 
-    WilsonNf2(FermOp,CG,CG,Fine);
+    WilsonNf2(FermOp,CG,CG);
   
   //Collect actions
   ActionLevel Level1;

From aeb7442d8f831a8e655bb81350d50004c89b146f Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Mon, 10 Aug 2015 20:47:44 +0100
Subject: [PATCH 418/429] Rework/global edit to enforce type templating of
 fermion operators. Allows multi-precision work and paves the way for
 alternate BC's and such like allowing for example G-parity which is important
 for K pipi programme. In particular, can drive an extra flavour index into
 the fermion fields using template types.

---
 benchmarks/Benchmark_dwf.cc                   |   2 +-
 benchmarks/Benchmark_wilson.cc                |   2 +-
 lib/Init.cc                                   |   3 +-
 lib/qcd/QCD.h                                 |  16 +-
 lib/qcd/action/Actions.h                      | 117 ++-
 lib/qcd/action/fermion/CayleyFermion5D.cc     | 124 ++--
 lib/qcd/action/fermion/CayleyFermion5D.h      |  35 +-
 .../fermion/ContinuedFractionFermion5D.cc     | 101 ++-
 .../fermion/ContinuedFractionFermion5D.h      |  31 +-
 lib/qcd/action/fermion/DomainWallFermion.h    |  21 +-
 lib/qcd/action/fermion/FermionOperator.h      | 175 ++++-
 lib/qcd/action/fermion/MobiusFermion.h        |  13 +-
 .../action/fermion/MobiusZolotarevFermion.h   |  13 +-
 .../fermion/OverlapWilsonCayleyTanhFermion.h  |  17 +-
 .../OverlapWilsonCayleyZolotarevFermion.h     |   9 +-
 .../OverlapWilsonContfracTanhFermion.h        |  15 +-
 .../OverlapWilsonContfracZolotarevFermion.h   |  15 +-
 .../OverlapWilsonPartialFractionTanhFermion.h |  23 +-
 ...lapWilsonPartialFractionZolotarevFermion.h |  23 +-
 .../fermion/PartialFractionFermion5D.cc       | 104 ++-
 .../action/fermion/PartialFractionFermion5D.h |  49 +-
 lib/qcd/action/fermion/ScaledShamirFermion.h  |   9 +-
 .../action/fermion/ShamirZolotarevFermion.h   |   9 +-
 lib/qcd/action/fermion/WilsonCompressor.h     |  37 +-
 lib/qcd/action/fermion/WilsonFermion.cc       | 519 ++++++-------
 lib/qcd/action/fermion/WilsonFermion.h        | 186 ++---
 lib/qcd/action/fermion/WilsonFermion5D.cc     | 135 ++--
 lib/qcd/action/fermion/WilsonFermion5D.h      | 104 ++-
 lib/qcd/action/fermion/WilsonKernels.cc       | 696 +++++++++---------
 lib/qcd/action/fermion/WilsonKernels.h        |  64 +-
 lib/qcd/action/fermion/WilsonKernelsHand.cc   | 149 ++--
 lib/qcd/action/pseudofermion/TwoFlavour.h     |  13 +-
 .../action/pseudofermion/TwoFlavourEvenOdd.h  |  38 +-
 lib/qcd/spin/Dirac.h                          |   1 -
 lib/qcd/spin/TwoSpinor.h                      | 452 ++++++------
 lib/tensors/Tensor_traits.h                   |   1 -
 scripts/cleanup                               |   1 +
 tests/Test_cayley_cg.cc                       |  28 +-
 tests/Test_cayley_coarsen_support.cc          |   6 +-
 tests/Test_cayley_even_odd.cc                 |  28 +-
 tests/Test_cayley_ldop_cr.cc                  |   6 +-
 tests/Test_cf_coarsen_support.cc              |   8 +-
 tests/Test_cf_cr_unprec.cc                    |   6 +-
 tests/Test_contfrac_cg.cc                     |  16 +-
 tests/Test_contfrac_even_odd.cc               |  16 +-
 tests/Test_contfrac_force.cc                  |   2 +-
 tests/Test_dwf_cg_prec.cc                     |   4 +-
 tests/Test_dwf_cg_schur.cc                    |   2 +-
 tests/Test_dwf_cg_unprec.cc                   |   4 +-
 tests/Test_dwf_cr_unprec.cc                   |   6 +-
 tests/Test_dwf_even_odd.cc                    |   4 +-
 tests/Test_dwf_force.cc                       |   2 +-
 tests/Test_dwf_fpgcr.cc                       |   6 +-
 tests/Test_dwf_hdcr.cc                        |  10 +-
 tests/Test_hmc_EOWilsonFermionGauge.cc        |   5 +-
 tests/Test_hmc_WilsonFermionGauge.cc          |   5 +-
 tests/Test_partfrac_force.cc                  |   2 +-
 tests/Test_wilson_cg_prec.cc                  |   4 +-
 tests/Test_wilson_cg_schur.cc                 |   2 +-
 tests/Test_wilson_cg_unprec.cc                |   4 +-
 tests/Test_wilson_cr_unprec.cc                |   4 +-
 tests/Test_wilson_even_odd.cc                 |   4 +-
 tests/Test_wilson_force.cc                    |   4 +-
 tests/Test_wilson_force_phiMdagMphi.cc        |   2 +-
 tests/Test_wilson_force_phiMphi.cc            |   2 +-
 65 files changed, 1935 insertions(+), 1579 deletions(-)

diff --git a/benchmarks/Benchmark_dwf.cc b/benchmarks/Benchmark_dwf.cc
index 9d9303ae..27772fdb 100644
--- a/benchmarks/Benchmark_dwf.cc
+++ b/benchmarks/Benchmark_dwf.cc
@@ -79,7 +79,7 @@ int main (int argc, char ** argv)
 
   RealD mass=0.1;
   RealD M5  =1.8;
-  DomainWallFermion Dw(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  DomainWallFermionR Dw(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
   
   std::cout<<GridLogMessage << "Calling Dw"<<std::endl;
   int ncall=10;
diff --git a/benchmarks/Benchmark_wilson.cc b/benchmarks/Benchmark_wilson.cc
index 7d23dec3..0308df33 100644
--- a/benchmarks/Benchmark_wilson.cc
+++ b/benchmarks/Benchmark_wilson.cc
@@ -87,7 +87,7 @@ int main (int argc, char ** argv)
   }
   ref = -0.5*ref;
   RealD mass=0.1;
-  WilsonFermion Dw(Umu,Grid,RBGrid,mass);
+  WilsonFermionR Dw(Umu,Grid,RBGrid,mass);
   
   std::cout<<GridLogMessage << "Calling Dw"<<std::endl;
   int ncall=10000;
diff --git a/lib/Init.cc b/lib/Init.cc
index 194fd589..2eeee6b4 100644
--- a/lib/Init.cc
+++ b/lib/Init.cc
@@ -177,8 +177,7 @@ void Grid_init(int *argc,char ***argv)
     Grid_quiesce_nodes();
   }
   if( GridCmdOptionExists(*argv,*argv+*argc,"--dslash-opt") ){
-    WilsonFermion::HandOptDslash=1;
-    WilsonFermion5D::HandOptDslash=1;
+    WilsonFermionStatic::HandOptDslash=1;
   }
   if( GridCmdOptionExists(*argv,*argv+*argc,"--lebesgue") ){
     LebesgueOrder::UseLebesgueOrder=1;
diff --git a/lib/qcd/QCD.h b/lib/qcd/QCD.h
index 1227ad74..16920f2b 100644
--- a/lib/qcd/QCD.h
+++ b/lib/qcd/QCD.h
@@ -19,6 +19,7 @@ namespace QCD {
     static const int Nd=4;
     static const int Nhs=2; // half spinor
     static const int Nds=8; // double stored gauge field
+    static const int Ngp=2; // gparity index range
 
     //////////////////////////////////////////////////////////////////////////////
     // QCD iMatrix types
@@ -27,7 +28,16 @@ namespace QCD {
     static const int ColourIndex = 2;
     static const int SpinIndex   = 1;
     static const int LorentzIndex= 0;
-    
+
+    // Useful traits is this a spin index
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+
+    const int SpinorIndex = 2;
+    template<typename T> struct isSpinor {
+      static const bool value = (SpinorIndex==T::TensorLevel);
+    };
+    template <typename T> using IfSpinor    = Invoke<std::enable_if<isSpinor<T>::value,int> > ;
+    template <typename T> using IfNotSpinor = Invoke<std::enable_if<!isSpinor<T>::value,int> > ;
 
     // ChrisK very keen to add extra space for Gparity doubling.
     //
@@ -49,6 +59,10 @@ namespace QCD {
     template<typename vtype> using iHalfSpinVector            = iScalar<iVector<iScalar<vtype>, Nhs> >;
     template<typename vtype> using iHalfSpinColourVector      = iScalar<iVector<iVector<vtype, Nc>, Nhs> >;
 
+    template<typename vtype> using iGparitySpinColourVector       = iVector<iVector<iVector<vtype, Nc>, Nhs>, Ngp >;
+    template<typename vtype> using iGparityHalfSpinColourVector   = iVector<iVector<iVector<vtype, Nc>, Nhs>, Ngp >;
+
+
     // Spin matrix
     typedef iSpinMatrix<Complex  >          SpinMatrix;
     typedef iSpinMatrix<ComplexF >          SpinMatrixF;
diff --git a/lib/qcd/action/Actions.h b/lib/qcd/action/Actions.h
index f40cd554..9b119858 100644
--- a/lib/qcd/action/Actions.h
+++ b/lib/qcd/action/Actions.h
@@ -17,9 +17,6 @@
 ////////////////////////////////////////////
 #include <qcd/action/ActionBase.h>
 
-#include <qcd/action/fermion/FermionOperator.h>
-
-
 ////////////////////////////////////////////
 // Gauge Actions
 ////////////////////////////////////////////
@@ -32,58 +29,118 @@
 #include <qcd/action/fermion/WilsonCompressor.h>     //used by all wilson type fermions
 #include <qcd/action/fermion/WilsonKernels.h>        //used by all wilson type fermions
 
+
+////////////////////////////////////////////////////////////////////////////////////////////////////
+// Explicit explicit template instantiation is still required in the .cc files
+//
+// - CayleyFermion5D.cc
+// - PartialFractionFermion5D.cc
+// - WilsonFermion5D.cc
+// - WilsonKernelsHand.cc
+// - ContinuedFractionFermion5D.cc
+// - WilsonFermion.cc
+// - WilsonKernels.cc
+//
+// The explicit instantiation is only avoidable if we move this source to headers and end up with include/parse/recompile
+// for EVERY .cc file. This define centralises the list and restores global push of impl cases
+////////////////////////////////////////////////////////////////////////////////////////////////////
+
+#define FermOpTemplateInstantiate(A) \
+  template class A<WilsonImplF>;		\
+  template class A<WilsonImplD>;
+
 ////////////////////////////////////////////
-// 4D formulations
+// Fermion operators / actions
 ////////////////////////////////////////////
-#include <qcd/action/fermion/WilsonFermion.h>
+#include <qcd/action/fermion/FermionOperator.h>
+
+#include <qcd/action/fermion/WilsonFermion.h>       // 4d wilson like
 //#include <qcd/action/fermion/CloverFermion.h>
 
-////////////////////////////////////////////
-// 5D formulations...
-////////////////////////////////////////////
-
-#include <qcd/action/fermion/WilsonFermion5D.h> // used by all 5d overlap types
-
-//////////
-// Cayley
-//////////
-#include <qcd/action/fermion/CayleyFermion5D.h>
+#include <qcd/action/fermion/WilsonFermion5D.h>     // 5d base used by all 5d overlap types
 
+#include <qcd/action/fermion/CayleyFermion5D.h>     // Cayley types
 #include <qcd/action/fermion/DomainWallFermion.h>
 #include <qcd/action/fermion/DomainWallFermion.h>
-
 #include <qcd/action/fermion/MobiusFermion.h>
 #include <qcd/action/fermion/ScaledShamirFermion.h>
-#include <qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h>
-
 #include <qcd/action/fermion/MobiusZolotarevFermion.h>
 #include <qcd/action/fermion/ShamirZolotarevFermion.h>
+#include <qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h>
 #include <qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h>
 
-//////////////////////
-// Continued fraction
-//////////////////////
-#include <qcd/action/fermion/ContinuedFractionFermion5D.h>
-#include <qcd/action/fermion/OverlapWilsonContfracTanhFermion.h>
-#include <qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h>
+#include <qcd/action/fermion/ContinuedFractionFermion5D.h>               // Continued fraction
+#include <qcd/action/fermion/OverlapWilsonContFracTanhFermion.h>
+#include <qcd/action/fermion/OverlapWilsonContFracZolotarevFermion.h>
 
-//////////////////////
-// Partial fraction
-//////////////////////
-#include <qcd/action/fermion/PartialFractionFermion5D.h>
+#include <qcd/action/fermion/PartialFractionFermion5D.h>                 // Partial fraction
 #include <qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h>
 #include <qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h>
 
+////////////////////////////////////////////////////////////////////////////////////////////////////
+// More maintainable to maintain the following typedef list centrally, as more "impl" targets
+// are added, (e.g. extension for gparity, half precision project in comms etc..)
+////////////////////////////////////////////////////////////////////////////////////////////////////
+
+
+// Cayley 5d
+namespace Grid {
+  namespace QCD {
+
+typedef WilsonFermion<WilsonImplR> WilsonFermionR;
+typedef WilsonFermion<WilsonImplF> WilsonFermionF;
+typedef WilsonFermion<WilsonImplD> WilsonFermionD;
+
+typedef DomainWallFermion<WilsonImplR> DomainWallFermionR;
+typedef DomainWallFermion<WilsonImplF> DomainWallFermionF;
+typedef DomainWallFermion<WilsonImplD> DomainWallFermionD;
+typedef MobiusFermion<WilsonImplR> MobiusFermionR;
+typedef MobiusFermion<WilsonImplF> MobiusFermionF;
+typedef MobiusFermion<WilsonImplD> MobiusFermionD;
+typedef ScaledShamirFermion<WilsonImplR> ScaledShamirFermionR;
+typedef ScaledShamirFermion<WilsonImplF> ScaledShamirFermionF;
+typedef ScaledShamirFermion<WilsonImplD> ScaledShamirFermionD;
+
+typedef MobiusZolotarevFermion<WilsonImplR> MobiusZolotarevFermionR;
+typedef MobiusZolotarevFermion<WilsonImplF> MobiusZolotarevFermionF;
+typedef MobiusZolotarevFermion<WilsonImplD> MobiusZolotarevFermionD;
+typedef ShamirZolotarevFermion<WilsonImplR> ShamirZolotarevFermionR;
+typedef ShamirZolotarevFermion<WilsonImplF> ShamirZolotarevFermionF;
+typedef ShamirZolotarevFermion<WilsonImplD> ShamirZolotarevFermionD;
+
+typedef OverlapWilsonCayleyTanhFermion<WilsonImplR> OverlapWilsonCayleyTanhFermionR;
+typedef OverlapWilsonCayleyTanhFermion<WilsonImplF> OverlapWilsonCayleyTanhFermionF;
+typedef OverlapWilsonCayleyTanhFermion<WilsonImplD> OverlapWilsonCayleyTanhFermionD;
+typedef OverlapWilsonCayleyZolotarevFermion<WilsonImplR> OverlapWilsonCayleyZolotarevFermionR;
+typedef OverlapWilsonCayleyZolotarevFermion<WilsonImplF> OverlapWilsonCayleyZolotarevFermionF;
+typedef OverlapWilsonCayleyZolotarevFermion<WilsonImplD> OverlapWilsonCayleyZolotarevFermionD;
+
+// Continued fraction
+typedef OverlapWilsonContFracTanhFermion<WilsonImplR> OverlapWilsonContFracTanhFermionR;
+typedef OverlapWilsonContFracTanhFermion<WilsonImplF> OverlapWilsonContFracTanhFermionF;
+typedef OverlapWilsonContFracTanhFermion<WilsonImplD> OverlapWilsonContFracTanhFermionD;
+typedef OverlapWilsonContFracZolotarevFermion<WilsonImplR> OverlapWilsonContFracZolotarevFermionR;
+typedef OverlapWilsonContFracZolotarevFermion<WilsonImplF> OverlapWilsonContFracZolotarevFermionF;
+typedef OverlapWilsonContFracZolotarevFermion<WilsonImplD> OverlapWilsonContFracZolotarevFermionD;
+
+// Partial fraction
+typedef OverlapWilsonPartialFractionTanhFermion<WilsonImplR> OverlapWilsonPartialFractionTanhFermionR;
+typedef OverlapWilsonPartialFractionTanhFermion<WilsonImplF> OverlapWilsonPartialFractionTanhFermionF;
+typedef OverlapWilsonPartialFractionTanhFermion<WilsonImplD> OverlapWilsonPartialFractionTanhFermionD;
+
+typedef OverlapWilsonPartialFractionZolotarevFermion<WilsonImplR> OverlapWilsonPartialFractionZolotarevFermionR;
+typedef OverlapWilsonPartialFractionZolotarevFermion<WilsonImplF> OverlapWilsonPartialFractionZolotarevFermionF;
+typedef OverlapWilsonPartialFractionZolotarevFermion<WilsonImplD> OverlapWilsonPartialFractionZolotarevFermionD;
+  }}
 ///////////////////////////////////////////////////////////////////////////////
 // G5 herm -- this has to live in QCD since dirac matrix is not in the broader sector of code
 ///////////////////////////////////////////////////////////////////////////////
 #include <qcd/action/fermion/g5HermitianLinop.h>
 
 ////////////////////////////////////////
-// Pseudo fermion combinations
+// Pseudo fermion combinations for HMC
 ////////////////////////////////////////
 #include <qcd/action/pseudofermion/TwoFlavour.h>
 #include <qcd/action/pseudofermion/TwoFlavourEvenOdd.h>
 
-
 #endif
diff --git a/lib/qcd/action/fermion/CayleyFermion5D.cc b/lib/qcd/action/fermion/CayleyFermion5D.cc
index c18edb7b..1ec7f930 100644
--- a/lib/qcd/action/fermion/CayleyFermion5D.cc
+++ b/lib/qcd/action/fermion/CayleyFermion5D.cc
@@ -2,13 +2,14 @@
 namespace Grid {
 namespace QCD {
 
- CayleyFermion5D::CayleyFermion5D(LatticeGaugeField &_Umu,
-				  GridCartesian         &FiveDimGrid,
-				  GridRedBlackCartesian &FiveDimRedBlackGrid,
-				  GridCartesian         &FourDimGrid,
-				  GridRedBlackCartesian &FourDimRedBlackGrid,
-				  RealD _mass,RealD _M5) :
-   WilsonFermion5D(_Umu,
+ template<class Impl>
+ CayleyFermion5D<Impl>::CayleyFermion5D(GaugeField &_Umu,
+					GridCartesian         &FiveDimGrid,
+					GridRedBlackCartesian &FiveDimRedBlackGrid,
+					GridCartesian         &FourDimGrid,
+					GridRedBlackCartesian &FourDimRedBlackGrid,
+					RealD _mass,RealD _M5) :
+   WilsonFermion5D<Impl>(_Umu,
 		   FiveDimGrid,
 		   FiveDimRedBlackGrid,
 		   FourDimGrid,
@@ -17,9 +18,11 @@ namespace QCD {
  {
  }
 
-  void CayleyFermion5D::Meooe5D    (const LatticeFermion &psi, LatticeFermion &Din)
+ template<class Impl>
+  void CayleyFermion5D<Impl>::Meooe5D    (const FermionField &psi, FermionField &Din)
   {
     // Assemble Din
+    int Ls=this->Ls;
     for(int s=0;s<Ls;s++){
       if ( s==0 ) {
 	//	Din = bs psi[s] + cs[s] psi[s+1}
@@ -35,8 +38,10 @@ namespace QCD {
       }
     }
   }
-  void CayleyFermion5D::MeooeDag5D    (const LatticeFermion &psi, LatticeFermion &Din)
+ template<class Impl>
+  void CayleyFermion5D<Impl>::MeooeDag5D    (const FermionField &psi, FermionField &Din)
   {
+    int Ls=this->Ls;
     for(int s=0;s<Ls;s++){
       if ( s==0 ) {
 	axpby_ssp_pplus (Din,bs[s],psi,cs[s+1],psi,s,s+1);
@@ -52,9 +57,12 @@ namespace QCD {
   }
 
   // override multiply
-  RealD CayleyFermion5D::M    (const LatticeFermion &psi, LatticeFermion &chi)
+ template<class Impl>
+  RealD CayleyFermion5D<Impl>::M    (const FermionField &psi, FermionField &chi)
   {
-    LatticeFermion Din(psi._grid);
+    int Ls=this->Ls;
+
+    FermionField Din(psi._grid);
 
     // Assemble Din
     /*
@@ -75,7 +83,7 @@ namespace QCD {
     */
     Meooe5D(psi,Din);
 
-    DW(Din,chi,DaggerNo);
+    this->DW(Din,chi,DaggerNo);
     // ((b D_W + D_w hop terms +1) on s-diag
     axpby(chi,1.0,1.0,chi,psi); 
 
@@ -95,18 +103,20 @@ namespace QCD {
     return norm2(chi);
   }
 
-  RealD CayleyFermion5D::Mdag (const LatticeFermion &psi, LatticeFermion &chi)
+ template<class Impl>
+  RealD CayleyFermion5D<Impl>::Mdag (const FermionField &psi, FermionField &chi)
   {
     // Under adjoint
     //D1+        D1- P-    ->   D1+^dag   P+ D2-^dag
     //D2- P+     D2+            P-D1-^dag D2+dag
 
-    LatticeFermion Din(psi._grid);
+    FermionField Din(psi._grid);
     // Apply Dw
-    DW(psi,Din,DaggerYes); 
+    this->DW(psi,Din,DaggerYes); 
 
     Meooe5D(Din,chi);
 
+    int Ls=this->Ls;
     for(int s=0;s<Ls;s++){
 
       // Collect the terms in DW
@@ -145,13 +155,15 @@ namespace QCD {
   }
 
   // half checkerboard operations
-  void CayleyFermion5D::Meooe       (const LatticeFermion &psi, LatticeFermion &chi)
+ template<class Impl>
+  void CayleyFermion5D<Impl>::Meooe       (const FermionField &psi, FermionField &chi)
   {
-    LatticeFermion tmp(psi._grid);
+    FermionField tmp(psi._grid);
     // Assemble the 5d matrix
     Meooe5D(psi,tmp); 
     std::cout << "Meooe Test replacement norm2 tmp = " <<norm2(tmp)<<std::endl;
 
+    int Ls=this->Ls;
     for(int s=0;s<Ls;s++){
       if ( s==0 ) {
 	//	tmp = bs psi[s] + cs[s] psi[s+1}
@@ -169,26 +181,28 @@ namespace QCD {
     std::cout << "Meooe Test replacement norm2 tmp old = " <<norm2(tmp)<<std::endl;
     // Apply 4d dslash
     if ( psi.checkerboard == Odd ) {
-      DhopEO(tmp,chi,DaggerNo);
+      this->DhopEO(tmp,chi,DaggerNo);
     } else {
-      DhopOE(tmp,chi,DaggerNo);
+      this->DhopOE(tmp,chi,DaggerNo);
     }
   }
 
-  void CayleyFermion5D::MeooeDag    (const LatticeFermion &psi, LatticeFermion &chi)
+  template<class Impl>
+  void CayleyFermion5D<Impl>::MeooeDag    (const FermionField &psi, FermionField &chi)
   {
-    LatticeFermion tmp(psi._grid);
+    FermionField tmp(psi._grid);
     // Apply 4d dslash
     if ( psi.checkerboard == Odd ) {
-      DhopEO(psi,tmp,DaggerYes);
+      this->DhopEO(psi,tmp,DaggerYes);
     } else {
-      DhopOE(psi,tmp,DaggerYes);
+      this->DhopOE(psi,tmp,DaggerYes);
     }
 
     Meooe5D(tmp,chi); 
     std::cout << "Meooe Test replacement norm2 chi new = " <<norm2(chi)<<std::endl;
 
     // Assemble the 5d matrix
+    int Ls=this->Ls;
     for(int s=0;s<Ls;s++){
       if ( s==0 ) {
 	axpby_ssp_pplus(chi,beo[s],tmp,   -ceo[s+1]  ,tmp,s,s+1);
@@ -205,8 +219,10 @@ namespace QCD {
 
   }
 
-  void CayleyFermion5D::Mooee       (const LatticeFermion &psi, LatticeFermion &chi)
+ template<class Impl>
+  void CayleyFermion5D<Impl>::Mooee       (const FermionField &psi, FermionField &chi)
   {
+    int Ls=this->Ls;
     for (int s=0;s<Ls;s++){
       if ( s==0 ) {
 	axpby_ssp_pminus(chi,bee[s],psi ,-cee[s],psi,s,s+1);
@@ -221,8 +237,10 @@ namespace QCD {
     }
   }
 
-  void  CayleyFermion5D::Mdir (const LatticeFermion &psi, LatticeFermion &chi,int dir,int disp){
-    LatticeFermion tmp(psi._grid);
+ template<class Impl>
+  void  CayleyFermion5D<Impl>::Mdir (const FermionField &psi, FermionField &chi,int dir,int disp){
+    int Ls=this->Ls;
+    FermionField tmp(psi._grid);
     // Assemble the 5d matrix
     for(int s=0;s<Ls;s++){
       if ( s==0 ) {
@@ -239,11 +257,13 @@ namespace QCD {
       }
     }
     // Apply 4d dslash fragment
-    DhopDir(tmp,chi,dir,disp);
+    this->DhopDir(tmp,chi,dir,disp);
   }
 
-  void CayleyFermion5D::MooeeDag    (const LatticeFermion &psi, LatticeFermion &chi)
+ template<class Impl>
+  void CayleyFermion5D<Impl>::MooeeDag    (const FermionField &psi, FermionField &chi)
   {
+    int Ls=this->Ls;
     for (int s=0;s<Ls;s++){
       // Assemble the 5d matrix
       if ( s==0 ) {
@@ -259,8 +279,10 @@ namespace QCD {
     }
   }
 
-  void CayleyFermion5D::MooeeInv    (const LatticeFermion &psi, LatticeFermion &chi)
+ template<class Impl>
+  void CayleyFermion5D<Impl>::MooeeInv    (const FermionField &psi, FermionField &chi)
   {
+    int Ls=this->Ls;
     // Apply (L^{\prime})^{-1}
     axpby_ssp (chi,1.0,psi,     0.0,psi,0,0);      // chi[0]=psi[0]
     for (int s=1;s<Ls;s++){
@@ -283,8 +305,10 @@ namespace QCD {
     }
   }
 
-  void CayleyFermion5D::MooeeInvDag (const LatticeFermion &psi, LatticeFermion &chi)
+ template<class Impl>
+  void CayleyFermion5D<Impl>::MooeeInvDag (const FermionField &psi, FermionField &chi)
   {
+    int Ls=this->Ls;
     // Apply (U^{\prime})^{-dagger}
     axpby_ssp (chi,1.0,psi,     0.0,psi,0,0);      // chi[0]=psi[0]
     for (int s=1;s<Ls;s++){
@@ -307,58 +331,64 @@ namespace QCD {
   }
 
   // force terms; five routines; default to Dhop on diagonal
-  void CayleyFermion5D::MDeriv  (LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+  template<class Impl>
+  void CayleyFermion5D<Impl>::MDeriv  (GaugeField &mat,const FermionField &U,const FermionField &V,int dag)
   {
-    LatticeFermion Din(V._grid);
+    FermionField Din(V._grid);
 
     if ( dag == DaggerNo ) {
       //      U d/du [D_w D5] V = U d/du DW D5 V
       Meooe5D(V,Din);
-      DhopDeriv(mat,U,Din,dag);
+      this->DhopDeriv(mat,U,Din,dag);
     } else {
       //      U d/du [D_w D5]^dag V = U D5^dag d/du DW^dag Y // implicit adj on U in call
       Meooe5D(U,Din);
-      DhopDeriv(mat,Din,V,dag);
+      this->DhopDeriv(mat,Din,V,dag);
     }
   };
-  void CayleyFermion5D::MoeDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+ template<class Impl>
+  void CayleyFermion5D<Impl>::MoeDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag)
   {
-    LatticeFermion Din(V._grid);
+    FermionField Din(V._grid);
 
     if ( dag == DaggerNo ) {
       //      U d/du [D_w D5] V = U d/du DW D5 V
       Meooe5D(V,Din);
-      DhopDerivOE(mat,U,Din,dag);
+      this->DhopDerivOE(mat,U,Din,dag);
     } else {
       //      U d/du [D_w D5]^dag V = U D5^dag d/du DW^dag Y // implicit adj on U in call
       Meooe5D(U,Din);
-      DhopDerivOE(mat,Din,V,dag);
+      this->DhopDerivOE(mat,Din,V,dag);
     }
   };
-  void CayleyFermion5D::MeoDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+ template<class Impl>
+  void CayleyFermion5D<Impl>::MeoDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag)
   {
-    LatticeFermion Din(V._grid);
+    FermionField Din(V._grid);
 
     if ( dag == DaggerNo ) {
       //      U d/du [D_w D5] V = U d/du DW D5 V
       Meooe5D(V,Din);
-      DhopDerivEO(mat,U,Din,dag);
+      this->DhopDerivEO(mat,U,Din,dag);
     } else {
       //      U d/du [D_w D5]^dag V = U D5^dag d/du DW^dag Y // implicit adj on U in call
       Meooe5D(U,Din);
-      DhopDerivEO(mat,Din,V,dag);
+      this->DhopDerivEO(mat,Din,V,dag);
     }
   };
   
   // Tanh
-  void CayleyFermion5D::SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD b,RealD c)
+ template<class Impl>
+  void CayleyFermion5D<Impl>::SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD b,RealD c)
   {
     SetCoefficientsZolotarev(1.0,zdata,b,c);
 
   }
   //Zolo
-  void CayleyFermion5D::SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata,RealD b,RealD c)
+ template<class Impl>
+  void CayleyFermion5D<Impl>::SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata,RealD b,RealD c)
   {
+    int Ls=this->Ls;
 
     ///////////////////////////////////////////////////////////
     // The Cayley coeffs (unprec)
@@ -408,8 +438,8 @@ namespace QCD {
     ceo.resize(Ls);
     
     for(int i=0;i<Ls;i++){
-      bee[i]=as[i]*(bs[i]*(4.0-M5) +1.0);
-      cee[i]=as[i]*(1.0-cs[i]*(4.0-M5));
+      bee[i]=as[i]*(bs[i]*(4.0-this->M5) +1.0);
+      cee[i]=as[i]*(1.0-cs[i]*(4.0-this->M5));
       beo[i]=as[i]*bs[i];
       ceo[i]=-as[i]*cs[i];
     }
@@ -462,6 +492,8 @@ namespace QCD {
     }
   }
 
+  FermOpTemplateInstantiate(CayleyFermion5D);
+
 }}
 
 
diff --git a/lib/qcd/action/fermion/CayleyFermion5D.h b/lib/qcd/action/fermion/CayleyFermion5D.h
index a139c32e..c0d2a292 100644
--- a/lib/qcd/action/fermion/CayleyFermion5D.h
+++ b/lib/qcd/action/fermion/CayleyFermion5D.h
@@ -5,33 +5,36 @@ namespace Grid {
 
   namespace QCD {
 
-    class CayleyFermion5D : public WilsonFermion5D
+    template<class Impl>
+    class CayleyFermion5D : public WilsonFermion5D<Impl>
     {
     public:
+#include <qcd/action/fermion/FermionImplTypedefs.h>
+    public:
 
       // override multiply
-      virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out);
-      virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out);
+      virtual RealD  M    (const FermionField &in, FermionField &out);
+      virtual RealD  Mdag (const FermionField &in, FermionField &out);
 
       // half checkerboard operations
-      virtual void   Meooe       (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   Meooe       (const FermionField &in, FermionField &out);
+      virtual void   MeooeDag    (const FermionField &in, FermionField &out);
+      virtual void   Mooee       (const FermionField &in, FermionField &out);
+      virtual void   MooeeDag    (const FermionField &in, FermionField &out);
+      virtual void   MooeeInv    (const FermionField &in, FermionField &out);
+      virtual void   MooeeInvDag (const FermionField &in, FermionField &out);
       virtual void   Instantiatable(void)=0;
 
       // force terms; five routines; default to Dhop on diagonal
-      virtual void MDeriv  (LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
-      virtual void MoeDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
-      virtual void MeoDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+      virtual void MDeriv  (GaugeField &mat,const FermionField &U,const FermionField &V,int dag);
+      virtual void MoeDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag);
+      virtual void MeoDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag);
 
       // Efficient support for multigrid coarsening
-      virtual void  Mdir (const LatticeFermion &in, LatticeFermion &out,int dir,int disp);
+      virtual void  Mdir (const FermionField &in, FermionField &out,int dir,int disp);
       
-      void   Meooe5D       (const LatticeFermion &in, LatticeFermion &out);
-      void   MeooeDag5D    (const LatticeFermion &in, LatticeFermion &out);
+      void   Meooe5D       (const FermionField &in, FermionField &out);
+      void   MeooeDag5D    (const FermionField &in, FermionField &out);
 
       //    protected:
       RealD mass;
@@ -56,7 +59,7 @@ namespace Grid {
       std::vector<RealD> dee;    
 
       // Constructors
-      CayleyFermion5D(LatticeGaugeField &_Umu,
+      CayleyFermion5D(GaugeField &_Umu,
 		      GridCartesian         &FiveDimGrid,
 		      GridRedBlackCartesian &FiveDimRedBlackGrid,
 		      GridCartesian         &FourDimGrid,
diff --git a/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc b/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
index cae73dfc..846e6a1b 100644
--- a/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
+++ b/lib/qcd/action/fermion/ContinuedFractionFermion5D.cc
@@ -3,11 +3,13 @@
 namespace Grid {
   namespace QCD {
 
-    void ContinuedFractionFermion5D::SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD scale)
+    template<class Impl>
+    void ContinuedFractionFermion5D<Impl>::SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD scale)
     {
       SetCoefficientsZolotarev(1.0/scale,zdata);
     }
-    void ContinuedFractionFermion5D::SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata)
+    template<class Impl>
+    void ContinuedFractionFermion5D<Impl>::SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata)
     {
       // How to check Ls matches??
       //      std::cout<<GridLogMessage << Ls << " Ls"<<std::endl;
@@ -16,7 +18,7 @@ namespace Grid {
       //      std::cout<<GridLogMessage << zdata->db << " -db"<<std::endl;
       //      std::cout<<GridLogMessage << zdata->dn << " -dn"<<std::endl;
       //      std::cout<<GridLogMessage << zdata->dd << " -dd"<<std::endl;
-
+      int Ls = this->Ls;
       assert(zdata->db==Ls);// Beta has Ls coeffs
 
       R=(1+this->mass)/(1-this->mass);
@@ -39,7 +41,7 @@ namespace Grid {
 
 
       ZoloHiInv =1.0/zolo_hi;
-      dw_diag = (4.0-M5)*ZoloHiInv;
+      dw_diag = (4.0-this->M5)*ZoloHiInv;
     
       See.resize(Ls);
       Aee.resize(Ls);
@@ -61,11 +63,14 @@ namespace Grid {
 
 
 
-    RealD  ContinuedFractionFermion5D::M           (const LatticeFermion &psi, LatticeFermion &chi)
+    template<class Impl>
+    RealD  ContinuedFractionFermion5D<Impl>::M           (const FermionField &psi, FermionField &chi)
     {
-      LatticeFermion D(psi._grid);
+      int Ls = this->Ls;
 
-      DW(psi,D,DaggerNo); 
+      FermionField D(psi._grid);
+
+      this->DW(psi,D,DaggerNo); 
 
       int sign=1;
       for(int s=0;s<Ls;s++){
@@ -83,15 +88,20 @@ namespace Grid {
       }
       return norm2(chi);
     }
-    RealD  ContinuedFractionFermion5D::Mdag        (const LatticeFermion &psi, LatticeFermion &chi)
+    template<class Impl>
+    RealD  ContinuedFractionFermion5D<Impl>::Mdag        (const FermionField &psi, FermionField &chi)
     {
       // This matrix is already hermitian. (g5 Dw) = Dw dag g5 = (g5 Dw)dag
       // The rest of matrix is symmetric.
       // Can ignore "dag"
       return M(psi,chi);
     }
-    void  ContinuedFractionFermion5D::Mdir (const LatticeFermion &psi, LatticeFermion &chi,int dir,int disp){
-      DhopDir(psi,chi,dir,disp); // Dslash on diagonal. g5 Dslash is hermitian
+    template<class Impl>
+    void  ContinuedFractionFermion5D<Impl>::Mdir (const FermionField &psi, FermionField &chi,int dir,int disp){
+      int Ls = this->Ls;
+
+      this->DhopDir(psi,chi,dir,disp); // Dslash on diagonal. g5 Dslash is hermitian
+
       int sign=1;
       for(int s=0;s<Ls;s++){
 	if ( s==(Ls-1) ){
@@ -102,13 +112,16 @@ namespace Grid {
 	sign=-sign; 
       }
     }
-    void   ContinuedFractionFermion5D::Meooe       (const LatticeFermion &psi, LatticeFermion &chi)
+    template<class Impl>
+    void   ContinuedFractionFermion5D<Impl>::Meooe       (const FermionField &psi, FermionField &chi)
     {
+      int Ls = this->Ls;
+
       // Apply 4d dslash
       if ( psi.checkerboard == Odd ) {
-	DhopEO(psi,chi,DaggerNo); // Dslash on diagonal. g5 Dslash is hermitian
+	this->DhopEO(psi,chi,DaggerNo); // Dslash on diagonal. g5 Dslash is hermitian
       } else {
-	DhopOE(psi,chi,DaggerNo); // Dslash on diagonal. g5 Dslash is hermitian
+	this->DhopOE(psi,chi,DaggerNo); // Dslash on diagonal. g5 Dslash is hermitian
       }
       
       int sign=1;
@@ -121,12 +134,16 @@ namespace Grid {
 	sign=-sign; 
       }
     }
-    void   ContinuedFractionFermion5D::MeooeDag    (const LatticeFermion &psi, LatticeFermion &chi)
+    template<class Impl>
+    void   ContinuedFractionFermion5D<Impl>::MeooeDag    (const FermionField &psi, FermionField &chi)
     {
-      Meooe(psi,chi);
+      this->Meooe(psi,chi);
     }
-    void   ContinuedFractionFermion5D::Mooee       (const LatticeFermion &psi, LatticeFermion &chi)
+    template<class Impl>
+    void   ContinuedFractionFermion5D<Impl>::Mooee       (const FermionField &psi, FermionField &chi)
     {
+      int Ls = this->Ls;
+
       int sign=1;
       for(int s=0;s<Ls;s++){
 	if ( s==0 ) {
@@ -144,12 +161,16 @@ namespace Grid {
       }
     }
 
-    void   ContinuedFractionFermion5D::MooeeDag    (const LatticeFermion &psi, LatticeFermion &chi)
+    template<class Impl>
+    void   ContinuedFractionFermion5D<Impl>::MooeeDag    (const FermionField &psi, FermionField &chi)
     {
-      Mooee(psi,chi);
+      this->Mooee(psi,chi);
     }
-    void   ContinuedFractionFermion5D::MooeeInv    (const LatticeFermion &psi, LatticeFermion &chi)
+    template<class Impl>
+    void   ContinuedFractionFermion5D<Impl>::MooeeInv    (const FermionField &psi, FermionField &chi)
     {
+      int Ls = this->Ls;
+
       // Apply Linv
       axpby_ssp(chi,1.0/cc_d[0],psi,0.0,psi,0,0); 
       for(int s=1;s<Ls;s++){
@@ -165,15 +186,19 @@ namespace Grid {
 	axpbg5y_ssp(chi,1.0/cc_d[s],chi,-1.0*cc_d[s+1]/See[s]/cc_d[s],chi,s,s+1);
       }
     }
-    void   ContinuedFractionFermion5D::MooeeInvDag (const LatticeFermion &psi, LatticeFermion &chi)
+    template<class Impl>
+    void   ContinuedFractionFermion5D<Impl>::MooeeInvDag (const FermionField &psi, FermionField &chi)
     {
-      MooeeInv(psi,chi);
+      this->MooeeInv(psi,chi);
     }
 
   // force terms; five routines; default to Dhop on diagonal
-   void ContinuedFractionFermion5D::MDeriv  (LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+    template<class Impl>
+   void ContinuedFractionFermion5D<Impl>::MDeriv  (GaugeField &mat,const FermionField &U,const FermionField &V,int dag)
   {
-    LatticeFermion D(V._grid);
+    int Ls = this->Ls;
+
+    FermionField D(V._grid);
 
     int sign=1;
     for(int s=0;s<Ls;s++){
@@ -184,11 +209,14 @@ namespace Grid {
       }
       sign=-sign; 
     }
-    DhopDeriv(mat,D,V,DaggerNo); 
+    this->DhopDeriv(mat,D,V,DaggerNo); 
   };
-   void ContinuedFractionFermion5D::MoeDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+    template<class Impl>
+   void ContinuedFractionFermion5D<Impl>::MoeDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag)
   {
-    LatticeFermion D(V._grid);
+    int Ls = this->Ls;
+
+    FermionField D(V._grid);
 
     int sign=1;
     for(int s=0;s<Ls;s++){
@@ -199,11 +227,14 @@ namespace Grid {
       }
       sign=-sign; 
     }
-    DhopDerivOE(mat,D,V,DaggerNo); 
+    this->DhopDerivOE(mat,D,V,DaggerNo); 
   };
-   void ContinuedFractionFermion5D::MeoDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+  template<class Impl>
+  void ContinuedFractionFermion5D<Impl>::MeoDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag)
   {
-    LatticeFermion D(V._grid);
+    int Ls = this->Ls;
+
+    FermionField D(V._grid);
 
     int sign=1;
     for(int s=0;s<Ls;s++){
@@ -214,25 +245,29 @@ namespace Grid {
       }
       sign=-sign; 
     }
-    DhopDerivEO(mat,D,V,DaggerNo); 
+    this->DhopDerivEO(mat,D,V,DaggerNo); 
   };
     
     // Constructors
-    ContinuedFractionFermion5D::ContinuedFractionFermion5D(
-							   LatticeGaugeField &_Umu,
+    template<class Impl>
+    ContinuedFractionFermion5D<Impl>::ContinuedFractionFermion5D(
+							   GaugeField &_Umu,
 							   GridCartesian         &FiveDimGrid,
 							   GridRedBlackCartesian &FiveDimRedBlackGrid,
 							   GridCartesian         &FourDimGrid,
 							   GridRedBlackCartesian &FourDimRedBlackGrid,
 							   RealD _mass,RealD M5) :
-      WilsonFermion5D(_Umu,
+      WilsonFermion5D<Impl>(_Umu,
 		      FiveDimGrid, FiveDimRedBlackGrid,
 		      FourDimGrid, FourDimRedBlackGrid,M5),
       mass(_mass)
     {
+      int Ls = this->Ls;
       assert((Ls&0x1)==1); // Odd Ls required
     }
 
+    FermOpTemplateInstantiate(ContinuedFractionFermion5D);
+
   }
 }
 
diff --git a/lib/qcd/action/fermion/ContinuedFractionFermion5D.h b/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
index 7d0c75bc..b1a7fba2 100644
--- a/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
+++ b/lib/qcd/action/fermion/ContinuedFractionFermion5D.h
@@ -5,35 +5,38 @@ namespace Grid {
 
   namespace QCD {
 
-    class ContinuedFractionFermion5D : public WilsonFermion5D
+    template<class Impl>
+    class ContinuedFractionFermion5D : public WilsonFermion5D<Impl>
     {
     public:
+#include <qcd/action/fermion/FermionImplTypedefs.h>
+    public:
 
       // override multiply
-      virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out);
-      virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out);
+      virtual RealD  M    (const FermionField &in, FermionField &out);
+      virtual RealD  Mdag (const FermionField &in, FermionField &out);
 
       // half checkerboard operaions
-      virtual void   Meooe       (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   Meooe       (const FermionField &in, FermionField &out);
+      virtual void   MeooeDag    (const FermionField &in, FermionField &out);
+      virtual void   Mooee       (const FermionField &in, FermionField &out);
+      virtual void   MooeeDag    (const FermionField &in, FermionField &out);
+      virtual void   MooeeInv    (const FermionField &in, FermionField &out);
+      virtual void   MooeeInvDag (const FermionField &in, FermionField &out);
 
       // force terms; five routines; default to Dhop on diagonal
-      virtual void MDeriv  (LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
-      virtual void MoeDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
-      virtual void MeoDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+      virtual void MDeriv  (GaugeField &mat,const FermionField &U,const FermionField &V,int dag);
+      virtual void MoeDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag);
+      virtual void MeoDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag);
 
       //      virtual void   Instantiatable(void)=0;
       virtual void   Instantiatable(void) =0;
 
       // Efficient support for multigrid coarsening
-      virtual void  Mdir (const LatticeFermion &in, LatticeFermion &out,int dir,int disp);
+      virtual void  Mdir (const FermionField &in, FermionField &out,int dir,int disp);
 
       // Constructors
-      ContinuedFractionFermion5D(LatticeGaugeField &_Umu,
+      ContinuedFractionFermion5D(GaugeField &_Umu,
 				 GridCartesian         &FiveDimGrid,
 				 GridRedBlackCartesian &FiveDimRedBlackGrid,
 				 GridCartesian         &FourDimGrid,
diff --git a/lib/qcd/action/fermion/DomainWallFermion.h b/lib/qcd/action/fermion/DomainWallFermion.h
index 498d8b34..00a6baba 100644
--- a/lib/qcd/action/fermion/DomainWallFermion.h
+++ b/lib/qcd/action/fermion/DomainWallFermion.h
@@ -7,24 +7,27 @@ namespace Grid {
 
   namespace QCD {
 
-    class DomainWallFermion : public CayleyFermion5D
+    template<class Impl>
+    class DomainWallFermion : public CayleyFermion5D<Impl>
     {
     public:
+#include <qcd/action/fermion/FermionImplTypedefs.h>
+    public:
 
       virtual void   Instantiatable(void) {};
       // Constructors
-      DomainWallFermion(LatticeGaugeField &_Umu,
+      DomainWallFermion(GaugeField &_Umu,
 			GridCartesian         &FiveDimGrid,
 			GridRedBlackCartesian &FiveDimRedBlackGrid,
 			GridCartesian         &FourDimGrid,
 			GridRedBlackCartesian &FourDimRedBlackGrid,
 			RealD _mass,RealD _M5) : 
 
-      CayleyFermion5D(_Umu,
-		      FiveDimGrid,
-		      FiveDimRedBlackGrid,
-		      FourDimGrid,
-		      FourDimRedBlackGrid,_mass,_M5)
+      CayleyFermion5D<Impl>(_Umu,
+			    FiveDimGrid,
+			    FiveDimRedBlackGrid,
+			    FourDimGrid,
+			    FourDimRedBlackGrid,_mass,_M5)
 
       {
 	RealD eps = 1.0;
@@ -32,9 +35,9 @@ namespace Grid {
 	Approx::zolotarev_data *zdata = Approx::higham(eps,this->Ls);// eps is ignored for higham
 	assert(zdata->n==this->Ls);
 	
-	std::cout<<GridLogMessage << "DomainWallFermion with Ls="<<Ls<<std::endl;
+	std::cout<<GridLogMessage << "DomainWallFermion with Ls="<<this->Ls<<std::endl;
 	// Call base setter
-	this->CayleyFermion5D::SetCoefficientsTanh(zdata,1.0,0.0);
+	this->SetCoefficientsTanh(zdata,1.0,0.0);
 
 	Approx::zolotarev_free(zdata);
       }
diff --git a/lib/qcd/action/fermion/FermionOperator.h b/lib/qcd/action/fermion/FermionOperator.h
index 57040037..75641b83 100644
--- a/lib/qcd/action/fermion/FermionOperator.h
+++ b/lib/qcd/action/fermion/FermionOperator.h
@@ -5,15 +5,169 @@ namespace Grid {
 
   namespace QCD {
 
+    ////////////////////////////////////////////////////////////////
+    // Hardwire to four spinors, allow to select 
+    // between gauge representation rank, and gparity/flavour index,
+    // and single/double precision.
+    ////////////////////////////////////////////////////////////////
+
+    template<class S,int Nrepresentation=Nc>
+    class WilsonImpl { 
+    public:
+
+      typedef S Simd;
+
+      template<typename vtype> using iImplSpinor             = iScalar<iVector<iVector<vtype, Nrepresentation>, Ns> >;
+      template<typename vtype> using iImplHalfSpinor         = iScalar<iVector<iVector<vtype, Nrepresentation>, Nhs> >;
+
+      template<typename vtype> using iImplGaugeLink          = iScalar<iScalar<iMatrix<vtype, Nrepresentation> > >;
+      template<typename vtype> using iImplGaugeField         = iVector<iScalar<iMatrix<vtype, Nrepresentation> >, Nd  >;
+      template<typename vtype> using iImplDoubledGaugeField  = iVector<iScalar<iMatrix<vtype, Nrepresentation> >, Nds >;
+    
+      typedef iImplSpinor    <Simd>           SiteSpinor;
+      typedef iImplHalfSpinor<Simd>           SiteHalfSpinor;
+      typedef iImplGaugeLink <Simd>           SiteGaugeLink;
+      typedef iImplGaugeField<Simd>           SiteGaugeField;
+      typedef iImplDoubledGaugeField<Simd>    SiteDoubledGaugeField;
+
+      typedef Lattice<SiteSpinor>                 FermionField;
+      typedef Lattice<SiteGaugeLink>            GaugeLinkField; // bit ugly naming; polarised gauge field, lorentz... all ugly
+      typedef Lattice<SiteGaugeField>               GaugeField;
+      typedef Lattice<SiteDoubledGaugeField> DoubledGaugeField;
+
+      typedef WilsonCompressor<SiteHalfSpinor,SiteSpinor> Compressor;
+
+      // provide the multiply by link that is differentiated between Gparity (with flavour index) and 
+      // non-Gparity
+      static inline void multLink(SiteHalfSpinor &phi,const SiteDoubledGaugeField &U,const SiteHalfSpinor &chi,int mu){
+        mult(&phi(),&U(mu),&chi());
+      }
+
+      static inline void DoubleStore(GridBase *GaugeGrid,DoubledGaugeField &Uds,const GaugeField &Umu)
+      {
+        conformable(Uds._grid,GaugeGrid);
+        conformable(Umu._grid,GaugeGrid);
+        GaugeLinkField U(GaugeGrid);
+        for(int mu=0;mu<Nd;mu++){
+  	  U = PeekIndex<LorentzIndex>(Umu,mu);
+	  PokeIndex<LorentzIndex>(Uds,U,mu);
+	  U = adj(Cshift(U,mu,-1));
+	  PokeIndex<LorentzIndex>(Uds,U,mu+4);
+	}
+      }
+
+    };
+
+    typedef WilsonImpl<vComplex,Nc>  WilsonImplR; // Real.. whichever prec
+    typedef WilsonImpl<vComplexF,Nc> WilsonImplF; // Float
+    typedef WilsonImpl<vComplexD,Nc> WilsonImplD; // Double
+
+
+
+    template<class S,int Nrepresentation=Nc>
+    class GparityWilsonImpl { 
+    public:
+
+      typedef S Simd;
+
+      template<typename vtype> using iImplSpinor             = iVector<iVector<iVector<vtype, Nrepresentation>, Ns>, Ngp >;
+      template<typename vtype> using iImplHalfSpinor         = iVector<iVector<iVector<vtype, Nrepresentation>, Nhs>, Ngp >;
+      template<typename vtype> using iImplGaugeField         = iVector<iScalar<iMatrix<vtype, Nrepresentation> >, Nd  >;
+
+      template<typename vtype> using iImplGaugeLink          = iScalar<iScalar<iScalar<iMatrix<vtype, Nrepresentation> > > >;
+      template<typename vtype> using iImplDoubledGaugeField  = iVector<iVector<iScalar<iMatrix<vtype, Nrepresentation> >, Nds >, Ngp >;
+    
+      typedef iImplSpinor    <Simd>           SiteSpinor;
+      typedef iImplHalfSpinor<Simd>           SiteHalfSpinor;
+      typedef iImplGaugeLink <Simd>           SiteGaugeLink;
+      typedef iImplGaugeField<Simd>           SiteGaugeField;
+      typedef iImplDoubledGaugeField<Simd>    SiteDoubledGaugeField;
+
+      typedef Lattice<SiteSpinor>                 FermionField;
+      typedef Lattice<SiteGaugeLink>            GaugeLinkField; // bit ugly naming; polarised gauge field, lorentz... all ugly
+      typedef Lattice<SiteGaugeField>               GaugeField;
+      typedef Lattice<SiteDoubledGaugeField> DoubledGaugeField;
+
+      typedef GparityWilsonCompressor<SiteHalfSpinor,SiteSpinor> Compressor;
+
+      // provide the multiply by link that is differentiated between Gparity (with flavour index) and 
+      // non-Gparity
+      static inline void multLink(SiteHalfSpinor &phi,const SiteDoubledGaugeField &U,const SiteHalfSpinor &chi,int mu){
+	for(int f=0;f<Ngp;f++){
+	  mult(&phi(f),&U(f)(mu),&chi(f));
+	}
+      }
+
+      static inline void DoubleStore(GridBase *GaugeGrid,DoubledGaugeField &Uds,const GaugeField &Umu)
+      {
+        conformable(Uds._grid,GaugeGrid);
+        conformable(Umu._grid,GaugeGrid);
+
+        GaugeLinkField Utmp(GaugeGrid);
+        GaugeLinkField U(GaugeGrid);
+        GaugeLinkField Uconj(GaugeGrid);
+
+	Lattice<iScalar<vInteger> > coor(GaugeGrid);
+	
+	std::vector<int> gpdirs({1,0,0,0});
+
+        for(int mu=0;mu<Nd;mu++){
+
+	  LatticeCoordinate(coor,mu);
+
+  	  U     = PeekIndex<LorentzIndex>(Umu,mu);
+	  Uconj = conj(U);
+
+	  int neglink = GaugeGrid->GlobalDimensions()[mu]-1;
+
+	  if ( gpdirs[mu] ) { 
+	    Uconj = where(coor==neglink,-Uconj,Uconj);
+	  }
+PARALLEL_FOR_LOOP
+	  for(auto ss=U.begin();ss<U.end();ss++){
+	    Uds[ss](0)(mu) = U[ss]();
+	    Uds[ss](1)(mu) = Uconj[ss]();
+	  }
+
+
+	  U     = adj(Cshift(U,mu,-1));      // correct except for spanning the boundary
+	  Uconj = adj(Cshift(Uconj,mu,-1));
+
+	  Utmp = where(coor==0,U,Uconj);
+PARALLEL_FOR_LOOP
+	  for(auto ss=U.begin();ss<U.end();ss++){
+	    Uds[ss](1)(mu) = Utmp[ss]();
+	  }
+
+	  Utmp = where(coor==0,Uconj,U);
+PARALLEL_FOR_LOOP
+	  for(auto ss=U.begin();ss<U.end();ss++){
+	    Uds[ss](0)(mu) = Utmp[ss]();
+	  }
+
+
+	}
+      }
+
+    };
+
+    typedef WilsonImpl<vComplex,Nc>  WilsonImplR; // Real.. whichever prec
+    typedef WilsonImpl<vComplexF,Nc> WilsonImplF; // Float
+    typedef WilsonImpl<vComplexD,Nc> WilsonImplD; // Double
+
+
+    
     //////////////////////////////////////////////////////////////////////////////
     // Four component fermions
     // Should type template the vector and gauge types
     // Think about multiple representations
     //////////////////////////////////////////////////////////////////////////////
-    template<class FermionField,class GaugeField>
-    class FermionOperator : public CheckerBoardedSparseMatrixBase<FermionField>
+    template<class Impl>
+    class FermionOperator : public CheckerBoardedSparseMatrixBase<typename Impl::FermionField>
     {
     public:
+#include <qcd/action/fermion/FermionImplTypedefs.h>
+    public:
 
       GridBase * Grid(void)   { return FermionGrid(); };   // this is all the linalg routines need to know
       GridBase * RedBlackGrid(void) { return FermionRedBlackGrid(); };
@@ -42,15 +196,15 @@ namespace Grid {
       virtual void DhopDir(const FermionField &in, FermionField &out,int dir,int disp)=0; // implemented by WilsonFermion and WilsonFermion5D
 
       // force terms; five routines; default to Dhop on diagonal
-      virtual void MDeriv  (LatticeGaugeField &mat,const FermionField &U,const FermionField &V,int dag){DhopDeriv(mat,U,V,dag);};
-      virtual void MoeDeriv(LatticeGaugeField &mat,const FermionField &U,const FermionField &V,int dag){DhopDerivOE(mat,U,V,dag);};
-      virtual void MeoDeriv(LatticeGaugeField &mat,const FermionField &U,const FermionField &V,int dag){DhopDerivEO(mat,U,V,dag);};
-      virtual void MooDeriv(LatticeGaugeField &mat,const FermionField &U,const FermionField &V,int dag){mat=zero;};
-      virtual void MeeDeriv(LatticeGaugeField &mat,const FermionField &U,const FermionField &V,int dag){mat=zero;};
+      virtual void MDeriv  (GaugeField &mat,const FermionField &U,const FermionField &V,int dag){DhopDeriv(mat,U,V,dag);};
+      virtual void MoeDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag){DhopDerivOE(mat,U,V,dag);};
+      virtual void MeoDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag){DhopDerivEO(mat,U,V,dag);};
+      virtual void MooDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag){mat=zero;};
+      virtual void MeeDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag){mat=zero;};
 
-      virtual void DhopDeriv  (LatticeGaugeField &mat,const FermionField &U,const FermionField &V,int dag)=0;
-      virtual void DhopDerivEO(LatticeGaugeField &mat,const FermionField &U,const FermionField &V,int dag)=0;
-      virtual void DhopDerivOE(LatticeGaugeField &mat,const FermionField &U,const FermionField &V,int dag)=0;
+      virtual void DhopDeriv  (GaugeField &mat,const FermionField &U,const FermionField &V,int dag)=0;
+      virtual void DhopDerivEO(GaugeField &mat,const FermionField &U,const FermionField &V,int dag)=0;
+      virtual void DhopDerivOE(GaugeField &mat,const FermionField &U,const FermionField &V,int dag)=0;
 
 
       virtual void  Mdiag  (const FermionField &in, FermionField &out) { Mooee(in,out);};   // Same as Mooee applied to both CB's
@@ -65,4 +219,5 @@ namespace Grid {
 
   }
 }
+
 #endif
diff --git a/lib/qcd/action/fermion/MobiusFermion.h b/lib/qcd/action/fermion/MobiusFermion.h
index 5aaaa188..e84661dc 100644
--- a/lib/qcd/action/fermion/MobiusFermion.h
+++ b/lib/qcd/action/fermion/MobiusFermion.h
@@ -7,13 +7,16 @@ namespace Grid {
 
   namespace QCD {
 
-    class MobiusFermion : public CayleyFermion5D
+    template<class Impl>
+    class MobiusFermion : public CayleyFermion5D<Impl>
     {
     public:
+#include <qcd/action/fermion/FermionImplTypedefs.h>
+    public:
 
       virtual void   Instantiatable(void) {};
       // Constructors
-      MobiusFermion(LatticeGaugeField &_Umu,
+      MobiusFermion(GaugeField &_Umu,
 		    GridCartesian         &FiveDimGrid,
 		    GridRedBlackCartesian &FiveDimRedBlackGrid,
 		    GridCartesian         &FourDimGrid,
@@ -21,7 +24,7 @@ namespace Grid {
 		    RealD _mass,RealD _M5,
 		    RealD b, RealD c) : 
       
-      CayleyFermion5D(_Umu,
+      CayleyFermion5D<Impl>(_Umu,
 		      FiveDimGrid,
 		      FiveDimRedBlackGrid,
 		      FourDimGrid,
@@ -30,12 +33,12 @@ namespace Grid {
       {
 	RealD eps = 1.0;
 
-	std::cout<<GridLogMessage << "MobiusFermion (b="<<b<<",c="<<c<<") with Ls= "<<Ls<<" Tanh approx"<<std::endl;
+	std::cout<<GridLogMessage << "MobiusFermion (b="<<b<<",c="<<c<<") with Ls= "<<this->Ls<<" Tanh approx"<<std::endl;
 	Approx::zolotarev_data *zdata = Approx::higham(eps,this->Ls);// eps is ignored for higham
 	assert(zdata->n==this->Ls);
 	
 	// Call base setter
-	this->CayleyFermion5D::SetCoefficientsTanh(zdata,b,c);
+	this->SetCoefficientsTanh(zdata,b,c);
 
 	Approx::zolotarev_free(zdata);
  
diff --git a/lib/qcd/action/fermion/MobiusZolotarevFermion.h b/lib/qcd/action/fermion/MobiusZolotarevFermion.h
index 5f74702c..1c572326 100644
--- a/lib/qcd/action/fermion/MobiusZolotarevFermion.h
+++ b/lib/qcd/action/fermion/MobiusZolotarevFermion.h
@@ -7,13 +7,16 @@ namespace Grid {
 
   namespace QCD {
 
-    class MobiusZolotarevFermion : public CayleyFermion5D
+    template<class Impl>
+    class MobiusZolotarevFermion : public CayleyFermion5D<Impl>
     {
     public:
+#include <qcd/action/fermion/FermionImplTypedefs.h>
+    public:
 
       virtual void   Instantiatable(void) {};
       // Constructors
-       MobiusZolotarevFermion(LatticeGaugeField &_Umu,
+       MobiusZolotarevFermion(GaugeField &_Umu,
 			      GridCartesian         &FiveDimGrid,
 			      GridRedBlackCartesian &FiveDimRedBlackGrid,
 			      GridCartesian         &FourDimGrid,
@@ -22,7 +25,7 @@ namespace Grid {
 			      RealD b, RealD c,
 			      RealD lo, RealD hi) : 
       
-      CayleyFermion5D(_Umu,
+      CayleyFermion5D<Impl>(_Umu,
 		      FiveDimGrid,
 		      FiveDimRedBlackGrid,
 		      FourDimGrid,
@@ -34,10 +37,10 @@ namespace Grid {
 	Approx::zolotarev_data *zdata = Approx::zolotarev(eps,this->Ls,0);
 	assert(zdata->n==this->Ls);
 
-	std::cout<<GridLogMessage << "MobiusZolotarevFermion (b="<<b<<",c="<<c<<") with Ls= "<<Ls<<" Zolotarev range ["<<lo<<","<<hi<<"]"<<std::endl;
+	std::cout<<GridLogMessage << "MobiusZolotarevFermion (b="<<b<<",c="<<c<<") with Ls= "<<this->Ls<<" Zolotarev range ["<<lo<<","<<hi<<"]"<<std::endl;
 	
 	// Call base setter
-	this->CayleyFermion5D::SetCoefficientsZolotarev(hi,zdata,b,c);
+	this->SetCoefficientsZolotarev(hi,zdata,b,c);
  
 	Approx::zolotarev_free(zdata);
       }
diff --git a/lib/qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h b/lib/qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h
index e764c8ae..0e8d4e19 100644
--- a/lib/qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h
+++ b/lib/qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h
@@ -7,12 +7,15 @@ namespace Grid {
 
   namespace QCD {
 
-    class OverlapWilsonCayleyTanhFermion : public MobiusFermion
+    template<class Impl>
+    class OverlapWilsonCayleyTanhFermion : public MobiusFermion<Impl>
     {
     public:
+#include <qcd/action/fermion/FermionImplTypedefs.h>
+    public:
 
       // Constructors
-    OverlapWilsonCayleyTanhFermion(LatticeGaugeField &_Umu,
+    OverlapWilsonCayleyTanhFermion(GaugeField &_Umu,
 				   GridCartesian         &FiveDimGrid,
 				   GridRedBlackCartesian &FiveDimRedBlackGrid,
 				   GridCartesian         &FourDimGrid,
@@ -21,11 +24,11 @@ namespace Grid {
 				   RealD scale) :
       
       // b+c=scale, b-c = 0 <=> b =c = scale/2
-      MobiusFermion(_Umu,
-		    FiveDimGrid,
-		    FiveDimRedBlackGrid,
-		    FourDimGrid,
-		    FourDimRedBlackGrid,_mass,_M5,0.5*scale,0.5*scale)
+      MobiusFermion<Impl>(_Umu,
+			  FiveDimGrid,
+			  FiveDimRedBlackGrid,
+			  FourDimGrid,
+			  FourDimRedBlackGrid,_mass,_M5,0.5*scale,0.5*scale)
 	{
 	}
     };
diff --git a/lib/qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h b/lib/qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h
index 82c43fb7..549292db 100644
--- a/lib/qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h
+++ b/lib/qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h
@@ -7,13 +7,16 @@ namespace Grid {
 
   namespace QCD {
 
-    class OverlapWilsonCayleyZolotarevFermion : public MobiusZolotarevFermion
+    template<class Impl>
+    class OverlapWilsonCayleyZolotarevFermion : public MobiusZolotarevFermion<Impl>
     {
     public:
+#include <qcd/action/fermion/FermionImplTypedefs.h>
+    public:
 
       // Constructors
 
-    OverlapWilsonCayleyZolotarevFermion(LatticeGaugeField &_Umu,
+    OverlapWilsonCayleyZolotarevFermion(GaugeField &_Umu,
 					GridCartesian         &FiveDimGrid,
 					GridRedBlackCartesian &FiveDimRedBlackGrid,
 					GridCartesian         &FourDimGrid,
@@ -21,7 +24,7 @@ namespace Grid {
 					RealD _mass,RealD _M5,
 					RealD lo, RealD hi) : 
       // b+c=1.0, b-c = 0 <=> b =c = 1/2
-      MobiusZolotarevFermion(_Umu,
+      MobiusZolotarevFermion<Impl>(_Umu,
 			     FiveDimGrid,
 			     FiveDimRedBlackGrid,
 			     FourDimGrid,
diff --git a/lib/qcd/action/fermion/OverlapWilsonContfracTanhFermion.h b/lib/qcd/action/fermion/OverlapWilsonContfracTanhFermion.h
index 6ecd7822..73baffa7 100644
--- a/lib/qcd/action/fermion/OverlapWilsonContfracTanhFermion.h
+++ b/lib/qcd/action/fermion/OverlapWilsonContfracTanhFermion.h
@@ -7,13 +7,16 @@ namespace Grid {
 
   namespace QCD {
 
-    class OverlapWilsonContFracTanhFermion : public ContinuedFractionFermion5D
+    template<class Impl>
+    class OverlapWilsonContFracTanhFermion : public ContinuedFractionFermion5D<Impl>
     {
     public:
+#include <qcd/action/fermion/FermionImplTypedefs.h>
+    public:
 
       virtual void   Instantiatable(void){};
       // Constructors
-    OverlapWilsonContFracTanhFermion(LatticeGaugeField &_Umu,
+    OverlapWilsonContFracTanhFermion(GaugeField &_Umu,
 				     GridCartesian         &FiveDimGrid,
 				     GridRedBlackCartesian &FiveDimRedBlackGrid,
 				     GridCartesian         &FourDimGrid,
@@ -22,16 +25,16 @@ namespace Grid {
 				     RealD scale) :
       
       // b+c=scale, b-c = 0 <=> b =c = scale/2
-      ContinuedFractionFermion5D(_Umu,
+      ContinuedFractionFermion5D<Impl>(_Umu,
 				 FiveDimGrid,
 				 FiveDimRedBlackGrid,
 				 FourDimGrid,
 				 FourDimRedBlackGrid,_mass,_M5)
 	{
-	  assert((Ls&0x1)==1); // Odd Ls required
-	  int nrational=Ls-1;// Even rational order
+	  assert((this->Ls&0x1)==1); // Odd Ls required
+	  int nrational=this->Ls-1;// Even rational order
 	  Approx::zolotarev_data *zdata = Approx::higham(1.0,nrational);// eps is ignored for higham
-	  SetCoefficientsTanh(zdata,scale);
+	  this->SetCoefficientsTanh(zdata,scale);
 	  Approx::zolotarev_free(zdata);
 	}
     };
diff --git a/lib/qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h b/lib/qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h
index 595daef3..3c662e92 100644
--- a/lib/qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h
+++ b/lib/qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h
@@ -7,13 +7,16 @@ namespace Grid {
 
   namespace QCD {
 
-    class OverlapWilsonContFracZolotarevFermion : public ContinuedFractionFermion5D
+    template<class Impl>
+    class OverlapWilsonContFracZolotarevFermion : public ContinuedFractionFermion5D<Impl>
     {
     public:
+#include <qcd/action/fermion/FermionImplTypedefs.h>
+    public:
 
       virtual void   Instantiatable(void){};
       // Constructors
-    OverlapWilsonContFracZolotarevFermion(LatticeGaugeField &_Umu,
+    OverlapWilsonContFracZolotarevFermion(GaugeField &_Umu,
 					  GridCartesian         &FiveDimGrid,
 					  GridRedBlackCartesian &FiveDimRedBlackGrid,
 					  GridCartesian         &FourDimGrid,
@@ -22,19 +25,19 @@ namespace Grid {
 					  RealD lo,RealD hi):
       
       // b+c=scale, b-c = 0 <=> b =c = scale/2
-      ContinuedFractionFermion5D(_Umu,
+      ContinuedFractionFermion5D<Impl>(_Umu,
 				 FiveDimGrid,
 				 FiveDimRedBlackGrid,
 				 FourDimGrid,
 				 FourDimRedBlackGrid,_mass,_M5)
 	{
-	  assert((Ls&0x1)==1); // Odd Ls required
+	  assert((this->Ls&0x1)==1); // Odd Ls required
 
-	  int nrational=Ls;// Odd rational order
+	  int nrational=this->Ls;// Odd rational order
 	  RealD eps = lo/hi;
 
 	  Approx::zolotarev_data *zdata = Approx::zolotarev(eps,nrational,0);
-	  SetCoefficientsZolotarev(hi,zdata);
+	  this->SetCoefficientsZolotarev(hi,zdata);
 	  Approx::zolotarev_free(zdata);
 
 	}
diff --git a/lib/qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h b/lib/qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h
index 1fa6c099..b017a9d7 100644
--- a/lib/qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h
+++ b/lib/qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h
@@ -7,13 +7,16 @@ namespace Grid {
 
   namespace QCD {
 
-    class OverlapWilsonPartialFractionTanhFermion : public PartialFractionFermion5D
+    template<class Impl>
+    class OverlapWilsonPartialFractionTanhFermion : public PartialFractionFermion5D<Impl>
     {
     public:
+#include <qcd/action/fermion/FermionImplTypedefs.h>
+    public:
 
       virtual void   Instantiatable(void){};
       // Constructors
-    OverlapWilsonPartialFractionTanhFermion(LatticeGaugeField &_Umu,
+    OverlapWilsonPartialFractionTanhFermion(GaugeField &_Umu,
 					    GridCartesian         &FiveDimGrid,
 					    GridRedBlackCartesian &FiveDimRedBlackGrid,
 					    GridCartesian         &FourDimGrid,
@@ -22,16 +25,16 @@ namespace Grid {
 					    RealD scale) :
       
       // b+c=scale, b-c = 0 <=> b =c = scale/2
-      PartialFractionFermion5D(_Umu,
-			       FiveDimGrid,
-			       FiveDimRedBlackGrid,
-			       FourDimGrid,
-			       FourDimRedBlackGrid,_mass,_M5)
+      PartialFractionFermion5D<Impl>(_Umu,
+				     FiveDimGrid,
+				     FiveDimRedBlackGrid,
+				     FourDimGrid,
+				     FourDimRedBlackGrid,_mass,_M5)
 	{
-	  assert((Ls&0x1)==1); // Odd Ls required
-	  int nrational=Ls-1;// Even rational order
+	  assert((this->Ls&0x1)==1); // Odd Ls required
+	  int nrational=this->Ls-1;// Even rational order
 	  Approx::zolotarev_data *zdata = Approx::higham(1.0,nrational);// eps is ignored for higham
-	  SetCoefficientsTanh(zdata,scale);
+	  this->SetCoefficientsTanh(zdata,scale);
 	  Approx::zolotarev_free(zdata);
 	}
     };
diff --git a/lib/qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h b/lib/qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h
index 3997c17f..3294f079 100644
--- a/lib/qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h
+++ b/lib/qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h
@@ -7,13 +7,16 @@ namespace Grid {
 
   namespace QCD {
 
-    class OverlapWilsonPartialFractionZolotarevFermion : public PartialFractionFermion5D
+    template<class Impl>
+    class OverlapWilsonPartialFractionZolotarevFermion : public PartialFractionFermion5D<Impl>
     {
     public:
+#include <qcd/action/fermion/FermionImplTypedefs.h>
+    public:
 
       virtual void   Instantiatable(void){};
       // Constructors
-    OverlapWilsonPartialFractionZolotarevFermion(LatticeGaugeField &_Umu,
+    OverlapWilsonPartialFractionZolotarevFermion(GaugeField &_Umu,
 					  GridCartesian         &FiveDimGrid,
 					  GridRedBlackCartesian &FiveDimRedBlackGrid,
 					  GridCartesian         &FourDimGrid,
@@ -22,19 +25,19 @@ namespace Grid {
 					  RealD lo,RealD hi):
       
       // b+c=scale, b-c = 0 <=> b =c = scale/2
-      PartialFractionFermion5D(_Umu,
-			       FiveDimGrid,
-			       FiveDimRedBlackGrid,
-			       FourDimGrid,
-			       FourDimRedBlackGrid,_mass,_M5)
+      PartialFractionFermion5D<Impl>(_Umu,
+				     FiveDimGrid,
+				     FiveDimRedBlackGrid,
+				     FourDimGrid,
+				     FourDimRedBlackGrid,_mass,_M5)
 	{
-	  assert((Ls&0x1)==1); // Odd Ls required
+	  assert((this->Ls&0x1)==1); // Odd Ls required
 
-	  int nrational=Ls;// Odd rational order
+	  int nrational=this->Ls;// Odd rational order
 	  RealD eps = lo/hi;
 
 	  Approx::zolotarev_data *zdata = Approx::zolotarev(eps,nrational,0);
-	  SetCoefficientsZolotarev(hi,zdata);
+	  this->SetCoefficientsZolotarev(hi,zdata);
 	  Approx::zolotarev_free(zdata);
 
 	}
diff --git a/lib/qcd/action/fermion/PartialFractionFermion5D.cc b/lib/qcd/action/fermion/PartialFractionFermion5D.cc
index 74174f51..81afb653 100644
--- a/lib/qcd/action/fermion/PartialFractionFermion5D.cc
+++ b/lib/qcd/action/fermion/PartialFractionFermion5D.cc
@@ -2,12 +2,15 @@
 namespace Grid {
   namespace QCD {
 
-    void  PartialFractionFermion5D::Mdir (const LatticeFermion &psi, LatticeFermion &chi,int dir,int disp){
+
+    template<class Impl>
+    void  PartialFractionFermion5D<Impl>::Mdir (const FermionField &psi, FermionField &chi,int dir,int disp){
       // this does both dag and undag but is trivial; make a common helper routing
 
       int sign = 1;
+      int Ls = this->Ls;
 
-      DhopDir(psi,chi,dir,disp);
+      this->DhopDir(psi,chi,dir,disp);
 
       int nblock=(Ls-1)/2;
       for(int b=0;b<nblock;b++){
@@ -18,15 +21,16 @@ namespace Grid {
       ag5xpby_ssp(chi,p[nblock]*scale/amax,chi,0.0,chi,Ls-1,Ls-1);
 
     }
-    void   PartialFractionFermion5D::Meooe_internal(const LatticeFermion &psi, LatticeFermion &chi,int dag)
+    template<class Impl>
+    void   PartialFractionFermion5D<Impl>::Meooe_internal(const FermionField &psi, FermionField &chi,int dag)
     {
-      // this does both dag and undag but is trivial; make a common helper routing
+      int Ls = this->Ls;
       int sign = dag ? (-1) : 1;
 
       if ( psi.checkerboard == Odd ) {
-	DhopEO(psi,chi,DaggerNo);
+	this->DhopEO(psi,chi,DaggerNo);
       } else {
-	DhopOE(psi,chi,DaggerNo);
+	this->DhopOE(psi,chi,DaggerNo);
       }
 
       int nblock=(Ls-1)/2;
@@ -38,10 +42,12 @@ namespace Grid {
       ag5xpby_ssp(chi,p[nblock]*scale/amax,chi,0.0,chi,Ls-1,Ls-1);
     }
 
-    void   PartialFractionFermion5D::Mooee_internal(const LatticeFermion &psi, LatticeFermion &chi,int dag)
+    template<class Impl>
+    void   PartialFractionFermion5D<Impl>::Mooee_internal(const FermionField &psi, FermionField &chi,int dag)
     {
       // again dag and undag are trivially related
       int sign = dag ? (-1) : 1;
+      int Ls = this->Ls;
       
       int nblock=(Ls-1)/2;
       for(int b=0;b<nblock;b++){
@@ -69,11 +75,13 @@ namespace Grid {
       }
     }
 
-    void   PartialFractionFermion5D::MooeeInv_internal(const LatticeFermion &psi, LatticeFermion &chi,int dag)
+    template<class Impl>
+    void   PartialFractionFermion5D<Impl>::MooeeInv_internal(const FermionField &psi, FermionField &chi,int dag)
     {
       int sign = dag ? (-1) : 1;
+      int Ls = this->Ls;
 
-      LatticeFermion tmp(psi._grid);
+      FermionField tmp(psi._grid);
       
       ///////////////////////////////////////////////////////////////////////////////////////
       //Linv
@@ -129,10 +137,12 @@ namespace Grid {
       axpby_ssp  (chi, 1.0/scale,tmp,0.0,tmp,Ls-1,Ls-1);
     }
 
-    void   PartialFractionFermion5D::M_internal(const LatticeFermion &psi, LatticeFermion &chi,int dag)
+    template<class Impl>
+    void   PartialFractionFermion5D<Impl>::M_internal(const FermionField &psi, FermionField &chi,int dag)
     {
-      LatticeFermion D(psi._grid);
+      FermionField D(psi._grid);
   
+      int Ls = this->Ls;
       int sign = dag ? (-1) : 1;
 
       // For partial frac Hw case (b5=c5=1) chroma quirkily computes
@@ -186,7 +196,7 @@ namespace Grid {
       //           ( 0     -sqrt(p_i)*amax   |  2 R gamma_5 + p0/amax 2H
       //
 
-      DW(psi,D,DaggerNo); 
+      this->DW(psi,D,DaggerNo); 
 
       int nblock=(Ls-1)/2;
       for(int b=0;b<nblock;b++){
@@ -217,48 +227,59 @@ namespace Grid {
 
     }
 
-    RealD  PartialFractionFermion5D::M    (const LatticeFermion &in, LatticeFermion &out)
+    template<class Impl>
+    RealD  PartialFractionFermion5D<Impl>::M    (const FermionField &in, FermionField &out)
     {
       M_internal(in,out,DaggerNo);
       return norm2(out);
     }
-    RealD  PartialFractionFermion5D::Mdag (const LatticeFermion &in, LatticeFermion &out)
+    template<class Impl>
+    RealD  PartialFractionFermion5D<Impl>::Mdag (const FermionField &in, FermionField &out)
     {
       M_internal(in,out,DaggerYes);
       return norm2(out);
     }
 
-    void PartialFractionFermion5D::Meooe       (const LatticeFermion &in, LatticeFermion &out)
+    template<class Impl>
+    void PartialFractionFermion5D<Impl>::Meooe       (const FermionField &in, FermionField &out)
     {
       Meooe_internal(in,out,DaggerNo);
     }
-    void PartialFractionFermion5D::MeooeDag    (const LatticeFermion &in, LatticeFermion &out)
+    template<class Impl>
+    void PartialFractionFermion5D<Impl>::MeooeDag    (const FermionField &in, FermionField &out)
     {
       Meooe_internal(in,out,DaggerYes);
     }
-    void PartialFractionFermion5D::Mooee       (const LatticeFermion &in, LatticeFermion &out)
+    template<class Impl>
+    void PartialFractionFermion5D<Impl>::Mooee       (const FermionField &in, FermionField &out)
     {
       Mooee_internal(in,out,DaggerNo);
     }
-    void PartialFractionFermion5D::MooeeDag    (const LatticeFermion &in, LatticeFermion &out)
+    template<class Impl>
+    void PartialFractionFermion5D<Impl>::MooeeDag    (const FermionField &in, FermionField &out)
     {
       Mooee_internal(in,out,DaggerYes);
     }
 
-    void PartialFractionFermion5D::MooeeInv    (const LatticeFermion &in, LatticeFermion &out)
+    template<class Impl>
+    void PartialFractionFermion5D<Impl>::MooeeInv    (const FermionField &in, FermionField &out)
     {
       MooeeInv_internal(in,out,DaggerNo);
     }
-    void PartialFractionFermion5D::MooeeInvDag (const LatticeFermion &in, LatticeFermion &out)
+    template<class Impl>
+    void PartialFractionFermion5D<Impl>::MooeeInvDag (const FermionField &in, FermionField &out)
     {
       MooeeInv_internal(in,out,DaggerYes);
     }
 
 
   // force terms; five routines; default to Dhop on diagonal
-   void PartialFractionFermion5D::MDeriv  (LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+    template<class Impl>
+   void PartialFractionFermion5D<Impl>::MDeriv  (GaugeField &mat,const FermionField &U,const FermionField &V,int dag)
   {
-    LatticeFermion D(V._grid);
+    int Ls = this->Ls;
+
+    FermionField D(V._grid);
 
     int nblock=(Ls-1)/2;
     for(int b=0;b<nblock;b++){
@@ -268,11 +289,14 @@ namespace Grid {
     }
     ag5xpby_ssp(D,p[nblock]*scale/amax,U,0.0,U,Ls-1,Ls-1);
 
-    DhopDeriv(mat,D,V,DaggerNo); 
+    this->DhopDeriv(mat,D,V,DaggerNo); 
   };
-   void PartialFractionFermion5D::MoeDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+    template<class Impl>
+   void PartialFractionFermion5D<Impl>::MoeDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag)
   {
-    LatticeFermion D(V._grid);
+    int Ls = this->Ls;
+
+    FermionField D(V._grid);
 
     int nblock=(Ls-1)/2;
     for(int b=0;b<nblock;b++){
@@ -282,11 +306,14 @@ namespace Grid {
     }
     ag5xpby_ssp(D,p[nblock]*scale/amax,U,0.0,U,Ls-1,Ls-1);
 
-    DhopDerivOE(mat,D,V,DaggerNo); 
+    this->DhopDerivOE(mat,D,V,DaggerNo); 
   };
-   void PartialFractionFermion5D::MeoDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
+    template<class Impl>
+   void PartialFractionFermion5D<Impl>::MeoDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag)
   {
-    LatticeFermion D(V._grid);
+    int Ls = this->Ls;
+
+    FermionField D(V._grid);
 
     int nblock=(Ls-1)/2;
     for(int b=0;b<nblock;b++){
@@ -296,13 +323,15 @@ namespace Grid {
     }
     ag5xpby_ssp(D,p[nblock]*scale/amax,U,0.0,U,Ls-1,Ls-1);
 
-    DhopDerivEO(mat,D,V,DaggerNo); 
+    this->DhopDerivEO(mat,D,V,DaggerNo); 
   };
 
-    void  PartialFractionFermion5D::SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD scale){
+    template<class Impl>
+    void  PartialFractionFermion5D<Impl>::SetCoefficientsTanh(Approx::zolotarev_data *zdata,RealD scale){
       SetCoefficientsZolotarev(1.0/scale,zdata);
     }
-    void  PartialFractionFermion5D::SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata){
+    template<class Impl>
+    void  PartialFractionFermion5D<Impl>::SetCoefficientsZolotarev(RealD zolo_hi,Approx::zolotarev_data *zdata){
 
       // check on degree matching
       //      std::cout<<GridLogMessage << Ls << " Ls"<<std::endl;
@@ -311,12 +340,14 @@ namespace Grid {
       //      std::cout<<GridLogMessage << zdata->db << " -db"<<std::endl;
       //      std::cout<<GridLogMessage << zdata->dn << " -dn"<<std::endl;
       //      std::cout<<GridLogMessage << zdata->dd << " -dd"<<std::endl;
+      int Ls = this->Ls;
+
       assert(Ls == (2*zdata->da -1) );
 
       // Part frac
       //      RealD R;
       R=(1+mass)/(1-mass);
-      dw_diag = (4.0-M5);
+      dw_diag = (4.0-this->M5);
 
       //      std::vector<RealD> p; 
       //      std::vector<RealD> q;
@@ -336,18 +367,21 @@ namespace Grid {
     }
 
       // Constructors
-    PartialFractionFermion5D::PartialFractionFermion5D(LatticeGaugeField &_Umu,
+    template<class Impl>
+    PartialFractionFermion5D<Impl>::PartialFractionFermion5D(GaugeField &_Umu,
 						       GridCartesian         &FiveDimGrid,
 						       GridRedBlackCartesian &FiveDimRedBlackGrid,
 						       GridCartesian         &FourDimGrid,
 						       GridRedBlackCartesian &FourDimRedBlackGrid,
 						       RealD _mass,RealD M5) :
-      WilsonFermion5D(_Umu,
+      WilsonFermion5D<Impl>(_Umu,
 		      FiveDimGrid, FiveDimRedBlackGrid,
 		      FourDimGrid, FourDimRedBlackGrid,M5),
       mass(_mass)
 
     {
+      int Ls = this->Ls;
+
       assert((Ls&0x1)==1); // Odd Ls required
       int nrational=Ls-1;
 
@@ -366,6 +400,8 @@ namespace Grid {
 
     }
  
+    FermOpTemplateInstantiate(PartialFractionFermion5D);
+
  }
 }
 
diff --git a/lib/qcd/action/fermion/PartialFractionFermion5D.h b/lib/qcd/action/fermion/PartialFractionFermion5D.h
index 3ba2481a..acbc30e1 100644
--- a/lib/qcd/action/fermion/PartialFractionFermion5D.h
+++ b/lib/qcd/action/fermion/PartialFractionFermion5D.h
@@ -5,46 +5,49 @@ namespace Grid {
 
   namespace QCD {
 
-    class PartialFractionFermion5D : public WilsonFermion5D
+    template<class Impl>
+    class PartialFractionFermion5D : public WilsonFermion5D<Impl>
     {
     public:
+#include <qcd/action/fermion/FermionImplTypedefs.h>
+    public:
 
       const int part_frac_chroma_convention=1;
 
-      void   Meooe_internal(const LatticeFermion &in, LatticeFermion &out,int dag);
-      void   Mooee_internal(const LatticeFermion &in, LatticeFermion &out,int dag);
-      void   MooeeInv_internal(const LatticeFermion &in, LatticeFermion &out,int dag);
-      void   M_internal(const LatticeFermion &in, LatticeFermion &out,int dag);
+      void   Meooe_internal(const FermionField &in, FermionField &out,int dag);
+      void   Mooee_internal(const FermionField &in, FermionField &out,int dag);
+      void   MooeeInv_internal(const FermionField &in, FermionField &out,int dag);
+      void   M_internal(const FermionField &in, FermionField &out,int dag);
 
       // override multiply
-      virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out);
-      virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out);
+      virtual RealD  M    (const FermionField &in, FermionField &out);
+      virtual RealD  Mdag (const FermionField &in, FermionField &out);
 
       // half checkerboard operaions
-      virtual void   Meooe       (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out);
-      virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
+      virtual void   Meooe       (const FermionField &in, FermionField &out);
+      virtual void   MeooeDag    (const FermionField &in, FermionField &out);
+      virtual void   Mooee       (const FermionField &in, FermionField &out);
+      virtual void   MooeeDag    (const FermionField &in, FermionField &out);
+      virtual void   MooeeInv    (const FermionField &in, FermionField &out);
+      virtual void   MooeeInvDag (const FermionField &in, FermionField &out);
 
       // force terms; five routines; default to Dhop on diagonal
-      virtual void MDeriv  (LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
-      virtual void MoeDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
-      virtual void MeoDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+      virtual void MDeriv  (GaugeField &mat,const FermionField &U,const FermionField &V,int dag);
+      virtual void MoeDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag);
+      virtual void MeoDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag);
 
       virtual void   Instantiatable(void) =0; // ensure no make-eee
 
       // Efficient support for multigrid coarsening
-      virtual void  Mdir (const LatticeFermion &in, LatticeFermion &out,int dir,int disp);
+      virtual void  Mdir (const FermionField &in, FermionField &out,int dir,int disp);
 
       // Constructors
-      PartialFractionFermion5D(LatticeGaugeField &_Umu,
-				    GridCartesian         &FiveDimGrid,
-				    GridRedBlackCartesian &FiveDimRedBlackGrid,
-				    GridCartesian         &FourDimGrid,
-				    GridRedBlackCartesian &FourDimRedBlackGrid,
-				    RealD _mass,RealD M5);
+      PartialFractionFermion5D(GaugeField &_Umu,
+			       GridCartesian         &FiveDimGrid,
+			       GridRedBlackCartesian &FiveDimRedBlackGrid,
+			       GridCartesian         &FourDimGrid,
+			       GridRedBlackCartesian &FourDimRedBlackGrid,
+			       RealD _mass,RealD M5);
 
     protected:
 
diff --git a/lib/qcd/action/fermion/ScaledShamirFermion.h b/lib/qcd/action/fermion/ScaledShamirFermion.h
index 59fb16a8..de7ded0c 100644
--- a/lib/qcd/action/fermion/ScaledShamirFermion.h
+++ b/lib/qcd/action/fermion/ScaledShamirFermion.h
@@ -7,12 +7,15 @@ namespace Grid {
 
   namespace QCD {
 
-    class ScaledShamirFermion : public MobiusFermion
+    template<class Impl>
+    class ScaledShamirFermion : public MobiusFermion<Impl>
     {
     public:
+#include <qcd/action/fermion/FermionImplTypedefs.h>
+    public:
 
       // Constructors
-    ScaledShamirFermion(LatticeGaugeField &_Umu,
+    ScaledShamirFermion(GaugeField &_Umu,
 			GridCartesian         &FiveDimGrid,
 			GridRedBlackCartesian &FiveDimRedBlackGrid,
 			GridCartesian         &FourDimGrid,
@@ -21,7 +24,7 @@ namespace Grid {
 			RealD scale) :
       
       // b+c=scale, b-c = 1 <=> 2b = scale+1; 2c = scale-1
-      MobiusFermion(_Umu,
+      MobiusFermion<Impl>(_Umu,
 		    FiveDimGrid,
 		    FiveDimRedBlackGrid,
 		    FourDimGrid,
diff --git a/lib/qcd/action/fermion/ShamirZolotarevFermion.h b/lib/qcd/action/fermion/ShamirZolotarevFermion.h
index 6a7df439..339a37f1 100644
--- a/lib/qcd/action/fermion/ShamirZolotarevFermion.h
+++ b/lib/qcd/action/fermion/ShamirZolotarevFermion.h
@@ -7,14 +7,17 @@ namespace Grid {
 
   namespace QCD {
 
-    class ShamirZolotarevFermion : public MobiusZolotarevFermion
+    template<class Impl>
+    class ShamirZolotarevFermion : public MobiusZolotarevFermion<Impl>
     {
     public:
+#include <qcd/action/fermion/FermionImplTypedefs.h>
+    public:
 
       // Constructors
 
 
-    ShamirZolotarevFermion(LatticeGaugeField &_Umu,
+    ShamirZolotarevFermion(GaugeField &_Umu,
 			   GridCartesian         &FiveDimGrid,
 			   GridRedBlackCartesian &FiveDimRedBlackGrid,
 			   GridCartesian         &FourDimGrid,
@@ -23,7 +26,7 @@ namespace Grid {
 			   RealD lo, RealD hi) : 
       
       // b+c = 1; b-c = 1 => b=1, c=0
-      MobiusZolotarevFermion(_Umu,
+      MobiusZolotarevFermion<Impl>(_Umu,
 			     FiveDimGrid,
 			     FiveDimRedBlackGrid,
 			     FourDimGrid,
diff --git a/lib/qcd/action/fermion/WilsonCompressor.h b/lib/qcd/action/fermion/WilsonCompressor.h
index d9fe977a..c7a20c59 100644
--- a/lib/qcd/action/fermion/WilsonCompressor.h
+++ b/lib/qcd/action/fermion/WilsonCompressor.h
@@ -4,6 +4,7 @@
 namespace Grid {
 namespace QCD {
 
+  template<class SiteHalfSpinor,class SiteSpinor>
   class WilsonCompressor {
   public:
     int mu;
@@ -18,9 +19,13 @@ namespace QCD {
       mu=p;
     };
 
-    vHalfSpinColourVector operator () (const vSpinColourVector &in)
+    virtual SiteHalfSpinor operator () (const SiteSpinor &in) {
+      return spinproject(in);
+    }
+
+    SiteHalfSpinor spinproject(const SiteSpinor &in)
     {
-      vHalfSpinColourVector ret;
+      SiteHalfSpinor ret;
       int mudag=mu;
       if (dag) {
 	mudag=(mu+Nd)%(2*Nd);
@@ -57,5 +62,33 @@ namespace QCD {
       return ret;
     }
   };
+
+
+  template<class SiteHalfSpinor,class SiteSpinor>
+    class GparityWilsonCompressor : public WilsonCompressor<SiteHalfSpinor,SiteSpinor>{
+  public:
+
+    GparityWilsonCompressor(int _dag) : WilsonCompressor<SiteHalfSpinor,SiteSpinor> (_dag){};
+
+    SiteHalfSpinor operator () (const SiteSpinor &in)
+    {
+      SiteHalfSpinor tmp = spinproject(in);
+      if( 0 ) tmp = flavourflip(tmp);
+      return tmp;
+    }
+
+    SiteHalfSpinor flavourflip(const SiteHalfSpinor &in) {
+
+      SiteHalfSpinor ret;
+      for(int f=0;f<Ngp;f++){
+	ret(0) = in(1);
+	ret(1) = in(0);
+      }
+      return ret;
+    }
+
+  };
+
+
 }} // namespace close
 #endif
diff --git a/lib/qcd/action/fermion/WilsonFermion.cc b/lib/qcd/action/fermion/WilsonFermion.cc
index 2c2132d1..260e3851 100644
--- a/lib/qcd/action/fermion/WilsonFermion.cc
+++ b/lib/qcd/action/fermion/WilsonFermion.cc
@@ -3,269 +3,290 @@
 namespace Grid {
 namespace QCD {
 
-const std::vector<int> WilsonFermion::directions   ({0,1,2,3, 0, 1, 2, 3});
-const std::vector<int> WilsonFermion::displacements({1,1,1,1,-1,-1,-1,-1});
+  const std::vector<int> WilsonFermionStatic::directions   ({0,1,2,3, 0, 1, 2, 3});
+  const std::vector<int> WilsonFermionStatic::displacements({1,1,1,1,-1,-1,-1,-1});
+  int WilsonFermionStatic::HandOptDslash;
 
-int WilsonFermion::HandOptDslash;
+  /////////////////////////////////
+  // Constructor and gauge import
+  /////////////////////////////////
 
-WilsonFermion::WilsonFermion(LatticeGaugeField &_Umu,
-			     GridCartesian         &Fgrid,
-			     GridRedBlackCartesian &Hgrid, 
-			     RealD _mass) :
-  _grid(&Fgrid),
-  _cbgrid(&Hgrid),
-  Stencil    (&Fgrid,npoint,Even,directions,displacements),
-  StencilEven(&Hgrid,npoint,Even,directions,displacements), // source is Even
-  StencilOdd (&Hgrid,npoint,Odd ,directions,displacements), // source is Odd
-  mass(_mass),
-  Umu(&Fgrid),
-  UmuEven(&Hgrid),
-  UmuOdd (&Hgrid)
-{
-  // Allocate the required comms buffer
-  comm_buf.resize(Stencil._unified_buffer_size); // this is always big enough to contain EO
-  ImportGauge(_Umu);
-}
-void WilsonFermion::ImportGauge(const LatticeGaugeField &_Umu)
-{
-  DoubleStore(Umu,_Umu);
-  pickCheckerboard(Even,UmuEven,Umu);
-  pickCheckerboard(Odd ,UmuOdd,Umu);
-}
-      
-void WilsonFermion::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu)
-{
-  conformable(Uds._grid,GaugeGrid());
-  conformable(Umu._grid,GaugeGrid());
-  LatticeColourMatrix U(GaugeGrid());
-  for(int mu=0;mu<Nd;mu++){
-    U = PeekIndex<LorentzIndex>(Umu,mu);
-    PokeIndex<LorentzIndex>(Uds,U,mu);
-    U = adj(Cshift(U,mu,-1));
-    PokeIndex<LorentzIndex>(Uds,U,mu+4);
+  template<class Impl>
+  WilsonFermion<Impl>::WilsonFermion(GaugeField &_Umu,
+				     GridCartesian         &Fgrid,
+				     GridRedBlackCartesian &Hgrid, 
+				     RealD _mass) :
+        _grid(&Fgrid),
+	_cbgrid(&Hgrid),
+	Stencil    (&Fgrid,npoint,Even,directions,displacements),
+	StencilEven(&Hgrid,npoint,Even,directions,displacements), // source is Even
+	StencilOdd (&Hgrid,npoint,Odd ,directions,displacements), // source is Odd
+	mass(_mass),
+	Umu(&Fgrid),
+	UmuEven(&Hgrid),
+	UmuOdd (&Hgrid) 
+  {
+    // Allocate the required comms buffer
+    comm_buf.resize(Stencil._unified_buffer_size); // this is always big enough to contain EO
+    ImportGauge(_Umu);
   }
-}
 
-RealD WilsonFermion::M(const LatticeFermion &in, LatticeFermion &out)
-{
-  out.checkerboard=in.checkerboard;
-  Dhop(in,out,DaggerNo);
-  return axpy_norm(out,4+mass,in,out);
-}
-RealD WilsonFermion::Mdag(const LatticeFermion &in, LatticeFermion &out)
-{
-  out.checkerboard=in.checkerboard;
-  Dhop(in,out,DaggerYes);
-  return axpy_norm(out,4+mass,in,out);
-}
-
-void WilsonFermion::Meooe(const LatticeFermion &in, LatticeFermion &out)
-{
-  if ( in.checkerboard == Odd ) {
-    DhopEO(in,out,DaggerNo);
-  } else {
-    DhopOE(in,out,DaggerNo);
+  template<class Impl>
+  void WilsonFermion<Impl>::ImportGauge(const GaugeField &_Umu)
+  {
+    Impl::DoubleStore(GaugeGrid(),Umu,_Umu);
+    pickCheckerboard(Even,UmuEven,Umu);
+    pickCheckerboard(Odd ,UmuOdd,Umu);
   }
-}
-void WilsonFermion::MeooeDag(const LatticeFermion &in, LatticeFermion &out)
-{
-  if ( in.checkerboard == Odd ) {
-    DhopEO(in,out,DaggerYes);
-  } else {
-    DhopOE(in,out,DaggerYes);
-  }
-}
-void WilsonFermion::Mooee(const LatticeFermion &in, LatticeFermion &out)
-{
-  out.checkerboard = in.checkerboard;
-  out = (4.0+mass)*in;
-  return ;
-}
-void WilsonFermion::MooeeDag(const LatticeFermion &in, LatticeFermion &out)
-{
-  out.checkerboard = in.checkerboard;
-  Mooee(in,out);
-}
-void WilsonFermion::MooeeInv(const LatticeFermion &in, LatticeFermion &out)
-{
-  out.checkerboard = in.checkerboard;
-  out = (1.0/(4.0+mass))*in;
-  return ;
-}
-void WilsonFermion::MooeeInvDag(const LatticeFermion &in, LatticeFermion &out)
-{
-  out.checkerboard = in.checkerboard;
-  MooeeInv(in,out);
-}
-void WilsonFermion::Mdir (const LatticeFermion &in, LatticeFermion &out,int dir,int disp)
-{
-  DhopDir(in,out,dir,disp);
-}
-void WilsonFermion::DhopDir(const LatticeFermion &in, LatticeFermion &out,int dir,int disp){
-
-  WilsonCompressor compressor(DaggerNo);
-
-  Stencil.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
   
-  assert( (disp==1)||(disp==-1) );
-
-  int skip = (disp==1) ? 0 : 1;
-
-  int dirdisp = dir+skip*4;
-
-PARALLEL_FOR_LOOP
-  for(int sss=0;sss<in._grid->oSites();sss++){
-    DiracOptDhopDir(Stencil,Umu,comm_buf,sss,sss,in,out,dirdisp,dirdisp);
-  }
- 
-};
-void WilsonFermion::DhopDirDisp(const LatticeFermion &in, LatticeFermion &out,int dirdisp,int gamma,int dag)
-{
-
-  WilsonCompressor compressor(dag);
-
-  Stencil.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
-
-PARALLEL_FOR_LOOP
-  for(int sss=0;sss<in._grid->oSites();sss++){
-    DiracOptDhopDir(Stencil,Umu,comm_buf,sss,sss,in,out,dirdisp,gamma);
-  }
- 
-};
-
-void WilsonFermion::DhopInternal(CartesianStencil & st,LatticeDoubledGaugeField & U,
-				const LatticeFermion &in, LatticeFermion &out,int dag)
-{
-  assert((dag==DaggerNo) ||(dag==DaggerYes));
-  WilsonCompressor compressor(dag);
-  st.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
-
-  if ( dag == DaggerYes ) {
-    if( HandOptDslash ) {
-PARALLEL_FOR_LOOP
-      for(int sss=0;sss<in._grid->oSites();sss++){
-        DiracOptHandDhopSiteDag(st,U,comm_buf,sss,sss,in,out);
-      }
-    } else { 
-PARALLEL_FOR_LOOP
-      for(int sss=0;sss<in._grid->oSites();sss++){
-        DiracOptDhopSiteDag(st,U,comm_buf,sss,sss,in,out);
-      }
-    }
-  } else {
-    if( HandOptDslash ) {
-PARALLEL_FOR_LOOP
-      for(int sss=0;sss<in._grid->oSites();sss++){
-        DiracOptHandDhopSite(st,U,comm_buf,sss,sss,in,out);
-      }
-    } else { 
-PARALLEL_FOR_LOOP
-      for(int sss=0;sss<in._grid->oSites();sss++){
-        DiracOptDhopSite(st,U,comm_buf,sss,sss,in,out);
-      }
-    }
-  }
-}
-void WilsonFermion::DhopOE(const LatticeFermion &in, LatticeFermion &out,int dag)
-{
-  conformable(in._grid,_cbgrid);    // verifies half grid
-  conformable(in._grid,out._grid); // drops the cb check
-
-  assert(in.checkerboard==Even);
-  out.checkerboard = Odd;
-
-  DhopInternal(StencilEven,UmuOdd,in,out,dag);
-}
-void WilsonFermion::DhopEO(const LatticeFermion &in, LatticeFermion &out,int dag)
-{
-  conformable(in._grid,_cbgrid);    // verifies half grid
-  conformable(in._grid,out._grid); // drops the cb check
-
-  assert(in.checkerboard==Odd);
-  out.checkerboard = Even;
-
-  DhopInternal(StencilOdd,UmuEven,in,out,dag);
-}
-void WilsonFermion::Dhop(const LatticeFermion &in, LatticeFermion &out,int dag)
-{
-  conformable(in._grid,_grid); // verifies full grid
-  conformable(in._grid,out._grid);
-
-  out.checkerboard = in.checkerboard;
-
-  DhopInternal(Stencil,Umu,in,out,dag);
-}
-
-void WilsonFermion::DerivInternal(CartesianStencil & st,LatticeDoubledGaugeField & U,
-				  LatticeGaugeField &mat,const LatticeFermion &A,const LatticeFermion &B,int dag)
-{
-  assert((dag==DaggerNo) ||(dag==DaggerYes));
-
-  WilsonCompressor compressor(dag);
-
-  LatticeColourMatrix tmp(B._grid);
-  LatticeFermion Btilde(B._grid);
-
-  st.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(B,comm_buf,compressor);
-  
-  for(int mu=0;mu<Nd;mu++){
+  /////////////////////////////
+  // Implement the interface
+  /////////////////////////////
       
-    ////////////////////////////////////////////////////////////////////////
-    // Flip gamma (1+g)<->(1-g) if dag
-    ////////////////////////////////////////////////////////////////////////
-    int gamma = mu;
-    if ( dag ) gamma+= Nd;
-
-    ////////////////////////
-    // Call the single hop
-    ////////////////////////
-PARALLEL_FOR_LOOP
-    for(int sss=0;sss<B._grid->oSites();sss++){
-      DiracOptDhopDir(st,U,comm_buf,sss,sss,B,Btilde,mu,gamma);
-    }
-
-    //////////////////////////////////////////////////
-    // spin trace outer product
-    //////////////////////////////////////////////////
-    tmp = TraceIndex<SpinIndex>(outerProduct(Btilde,A)); 
-    PokeIndex<LorentzIndex>(mat,tmp,mu);
-
+  template<class Impl>
+  RealD WilsonFermion<Impl>::M(const FermionField &in, FermionField &out) 
+  {
+    out.checkerboard=in.checkerboard;
+    Dhop(in,out,DaggerNo);
+    return axpy_norm(out,4+mass,in,out);
   }
-}
-void WilsonFermion::DhopDeriv(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
-{
-  conformable(U._grid,_grid);  
-  conformable(U._grid,V._grid);
-  conformable(U._grid,mat._grid);
 
-  mat.checkerboard = U.checkerboard;
+  template<class Impl>
+  RealD WilsonFermion<Impl>::Mdag(const FermionField &in, FermionField &out) 
+  {
+    out.checkerboard=in.checkerboard;
+    Dhop(in,out,DaggerYes);
+    return axpy_norm(out,4+mass,in,out);
+  }
 
-  DerivInternal(Stencil,Umu,mat,U,V,dag);
-}
-void WilsonFermion::DhopDerivOE(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
-{
-  conformable(U._grid,_cbgrid);  
-  conformable(U._grid,V._grid);
-  conformable(U._grid,mat._grid);
+  template<class Impl>
+  void WilsonFermion<Impl>::Meooe(const FermionField &in, FermionField &out) 
+  {
+    if ( in.checkerboard == Odd ) {
+      DhopEO(in,out,DaggerNo);
+    } else {
+      DhopOE(in,out,DaggerNo);
+    }
+  }
+  template<class Impl>
+  void WilsonFermion<Impl>::MeooeDag(const FermionField &in, FermionField &out) 
+  {
+    if ( in.checkerboard == Odd ) {
+      DhopEO(in,out,DaggerYes);
+    } else {
+      DhopOE(in,out,DaggerYes);
+    }
+  }
 
-  assert(V.checkerboard==Even);
-  assert(U.checkerboard==Odd);
-  mat.checkerboard = Odd;
+  template<class Impl>
+  void WilsonFermion<Impl>::Mooee(const FermionField &in, FermionField &out) {
+    out.checkerboard = in.checkerboard;
+    out = (4.0+mass)*in;
+  }
+  
+  template<class Impl>
+  void WilsonFermion<Impl>::MooeeDag(const FermionField &in, FermionField &out) {
+    out.checkerboard = in.checkerboard;
+    Mooee(in,out);
+  }
+  
+  template<class Impl>
+  void WilsonFermion<Impl>::MooeeInv(const FermionField &in, FermionField &out) {
+    out.checkerboard = in.checkerboard;
+    out = (1.0/(4.0+mass))*in;
+  }
+  
+  template<class Impl>
+  void WilsonFermion<Impl>::MooeeInvDag(const FermionField &in, FermionField &out) {
+    out.checkerboard = in.checkerboard;
+    MooeeInv(in,out);
+  }
+  
+  ///////////////////////////////////
+  // Internal
+  ///////////////////////////////////
 
-  DerivInternal(StencilEven,UmuOdd,mat,U,V,dag);
-}
-void WilsonFermion::DhopDerivEO(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag)
-{
-  conformable(U._grid,_cbgrid);  
-  conformable(U._grid,V._grid);
-  conformable(U._grid,mat._grid);
+  template<class Impl>
+  void WilsonFermion<Impl>::DerivInternal(CartesianStencil & st,
+					  DoubledGaugeField & U,
+					  GaugeField &mat,
+					  const FermionField &A,
+					  const FermionField &B,int dag) {
+	
+    assert((dag==DaggerNo) ||(dag==DaggerYes));
+    
+    Compressor compressor(dag);
+    
+    GaugeLinkField tmp(B._grid);
+    FermionField Btilde(B._grid);
+    
+    st.HaloExchange<SiteSpinor,SiteHalfSpinor,Compressor>(B,comm_buf,compressor);
+    
+    for(int mu=0;mu<Nd;mu++){
+      
+      ////////////////////////////////////////////////////////////////////////
+      // Flip gamma (1+g)<->(1-g) if dag
+      ////////////////////////////////////////////////////////////////////////
+      int gamma = mu;
+      if ( dag ) gamma+= Nd;
+      
+      ////////////////////////
+      // Call the single hop
+      ////////////////////////
+PARALLEL_FOR_LOOP
+	for(int sss=0;sss<B._grid->oSites();sss++){
+	  Kernels::DiracOptDhopDir(st,U,comm_buf,sss,sss,B,Btilde,mu,gamma);
+	}
+      
+      //////////////////////////////////////////////////
+      // spin trace outer product
+      //////////////////////////////////////////////////
+      tmp = TraceIndex<SpinIndex>(outerProduct(Btilde,A)); 
+      PokeIndex<LorentzIndex>(mat,tmp,mu);
+      
+    }
+  }
+  
+  template<class Impl>
+  void WilsonFermion<Impl>::DhopDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag)
+  {
+    conformable(U._grid,_grid);  
+    conformable(U._grid,V._grid);
+    conformable(U._grid,mat._grid);
+    
+    mat.checkerboard = U.checkerboard;
+    
+    DerivInternal(Stencil,Umu,mat,U,V,dag);
+  }
+  
+  template<class Impl>
+  void WilsonFermion<Impl>::DhopDerivOE(GaugeField &mat,const FermionField &U,const FermionField &V,int dag)
+  {
+    conformable(U._grid,_cbgrid);  
+    conformable(U._grid,V._grid);
+    conformable(U._grid,mat._grid);
+    
+    assert(V.checkerboard==Even);
+    assert(U.checkerboard==Odd);
+    mat.checkerboard = Odd;
+    
+    DerivInternal(StencilEven,UmuOdd,mat,U,V,dag);
+  }
+  
+  template<class Impl>
+  void WilsonFermion<Impl>::DhopDerivEO(GaugeField &mat,const FermionField &U,const FermionField &V,int dag)
+  {
+    conformable(U._grid,_cbgrid);  
+    conformable(U._grid,V._grid);
+    conformable(U._grid,mat._grid);
+	
+    assert(V.checkerboard==Odd);
+    assert(U.checkerboard==Even);
+    mat.checkerboard = Even;
+	
+    DerivInternal(StencilOdd,UmuEven,mat,U,V,dag);
+  }
+  
 
-  assert(V.checkerboard==Odd);
-  assert(U.checkerboard==Even);
-  mat.checkerboard = Even;
+  template<class Impl>
+  void WilsonFermion<Impl>::Dhop(const FermionField &in, FermionField &out,int dag) {
+    conformable(in._grid,_grid); // verifies full grid
+    conformable(in._grid,out._grid);
+    
+    out.checkerboard = in.checkerboard;
+    
+    DhopInternal(Stencil,Umu,in,out,dag);
+  }
+  
+  template<class Impl>
+  void WilsonFermion<Impl>::DhopOE(const FermionField &in, FermionField &out,int dag) {
+    conformable(in._grid,_cbgrid);    // verifies half grid
+    conformable(in._grid,out._grid); // drops the cb check
+    
+    assert(in.checkerboard==Even);
+    out.checkerboard = Odd;
+    
+    DhopInternal(StencilEven,UmuOdd,in,out,dag);
+  }
+  
+  template<class Impl>
+  void WilsonFermion<Impl>::DhopEO(const FermionField &in, FermionField &out,int dag) {
+    conformable(in._grid,_cbgrid);    // verifies half grid
+    conformable(in._grid,out._grid); // drops the cb check
+    
+    assert(in.checkerboard==Odd);
+    out.checkerboard = Even;
+    
+    DhopInternal(StencilOdd,UmuEven,in,out,dag);
+  }
+  
+  template<class Impl>
+  void WilsonFermion<Impl>::Mdir (const FermionField &in, FermionField &out,int dir,int disp) {
+    DhopDir(in,out,dir,disp);
+  }
+  
+  template<class Impl>
+  void WilsonFermion<Impl>::DhopDir(const FermionField &in, FermionField &out,int dir,int disp){
+    
+    int skip = (disp==1) ? 0 : 1;
+    int dirdisp = dir+skip*4;
+    
+    DhopDirDisp(in,out,dirdisp,dirdisp,DaggerNo);
+    
+  };
+  
+  template<class Impl>
+  void WilsonFermion<Impl>::DhopDirDisp(const FermionField &in, FermionField &out,int dirdisp,int gamma,int dag) {
+    
+    Compressor compressor(dag);
+    
+    Stencil.HaloExchange<SiteSpinor,SiteHalfSpinor,Compressor>(in,comm_buf,compressor);
+    
+PARALLEL_FOR_LOOP
+      for(int sss=0;sss<in._grid->oSites();sss++){
+	Kernels::DiracOptDhopDir(Stencil,Umu,comm_buf,sss,sss,in,out,dirdisp,gamma);
+      }
+    
+  };
 
-  DerivInternal(StencilOdd,UmuEven,mat,U,V,dag);
-}
+
+  template<class Impl>
+  void WilsonFermion<Impl>::DhopInternal(CartesianStencil & st,DoubledGaugeField & U,
+					 const FermionField &in, FermionField &out,int dag) {
+
+    assert((dag==DaggerNo) ||(dag==DaggerYes));
+
+    Compressor compressor(dag);
+    st.HaloExchange<SiteSpinor,SiteHalfSpinor,Compressor>(in,comm_buf,compressor);
+    
+    if ( dag == DaggerYes ) {
+      if( HandOptDslash ) {
+PARALLEL_FOR_LOOP
+        for(int sss=0;sss<in._grid->oSites();sss++){
+	  Kernels::DiracOptHandDhopSiteDag(st,U,comm_buf,sss,sss,in,out);
+	}
+      } else { 
+PARALLEL_FOR_LOOP
+        for(int sss=0;sss<in._grid->oSites();sss++){
+	  Kernels::DiracOptDhopSiteDag(st,U,comm_buf,sss,sss,in,out);
+	}
+      }
+    } else {
+      if( HandOptDslash ) {
+PARALLEL_FOR_LOOP
+        for(int sss=0;sss<in._grid->oSites();sss++){
+	  Kernels::DiracOptHandDhopSite(st,U,comm_buf,sss,sss,in,out);
+	}
+      } else { 
+PARALLEL_FOR_LOOP
+        for(int sss=0;sss<in._grid->oSites();sss++){
+	  Kernels::DiracOptDhopSite(st,U,comm_buf,sss,sss,in,out);
+	}
+      }
+    }
+  };
+ 
+  FermOpTemplateInstantiate(WilsonFermion);
 
 }}
 
diff --git a/lib/qcd/action/fermion/WilsonFermion.h b/lib/qcd/action/fermion/WilsonFermion.h
index 79d3e626..1dc71c00 100644
--- a/lib/qcd/action/fermion/WilsonFermion.h
+++ b/lib/qcd/action/fermion/WilsonFermion.h
@@ -5,8 +5,20 @@ namespace Grid {
 
   namespace QCD {
 
-    class WilsonFermion : public FermionOperator<LatticeFermion,LatticeGaugeField>
+    class WilsonFermionStatic {
+    public:
+      static int HandOptDslash; // these are a temporary hack
+      static int MortonOrder;
+      static const std::vector<int> directions   ;
+      static const std::vector<int> displacements;
+      static const int npoint=8;
+    };
+
+    template<class Impl>
+    class WilsonFermion : public FermionOperator<Impl>, public WilsonFermionStatic
     {
+#include <qcd/action/fermion/FermionImplTypedefs.h>
+
     public:
 
       ///////////////////////////////////////////////////////////////
@@ -17,133 +29,73 @@ namespace Grid {
       GridBase *FermionGrid(void)            { return _grid;}
       GridBase *FermionRedBlackGrid(void)    { return _cbgrid;}
 
-      // override multiply
-      virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out);
-      virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out);
+      //////////////////////////////////////////////////////////////////
+      // override multiply; cut number routines if pass dagger argument
+      // and also make interface more uniformly consistent
+      //////////////////////////////////////////////////////////////////
+      RealD M(const FermionField &in, FermionField &out);
+      RealD Mdag(const FermionField &in, FermionField &out);
 
-      // half checkerboard operaions
-      void   Meooe       (const LatticeFermion &in, LatticeFermion &out);
-      void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out);
-
-      virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out); // remain virtual so we 
-      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out); // can derive Clover
-      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out); // from Wilson base
-      virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out);
+      /////////////////////////////////////////////////////////
+      // half checkerboard operations
+      // could remain virtual so we  can derive Clover from Wilson base
+      /////////////////////////////////////////////////////////
+      void Meooe(const FermionField &in, FermionField &out) ;
+      void MeooeDag(const FermionField &in, FermionField &out) ;
+      void Mooee(const FermionField &in, FermionField &out) ;
+      void MooeeDag(const FermionField &in, FermionField &out) ;
+      void MooeeInv(const FermionField &in, FermionField &out) ;
+      void MooeeInvDag(const FermionField &in, FermionField &out) ;
 
       ////////////////////////
-      // 
-      // Force term: d/dtau S = 0
-      //
-      // It is simplest to consider the two flavour force term 
-      // 
-      //       S[U,phi] = phidag (MdagM)^-1 phi
-      //
-      // But simplify even this to
-      //
-      //       S[U,phi] = phidag MdagM phi
-      //
-      // (other options exist depending on nature of action fragment.)
-      // 
-      // Require momentum be traceless anti-hermitian to move within group manifold [ P = i P^a T^a ]
-      //
-      // Define the HMC hamiltonian
-      //
-      //       H = 1/2 Tr P^2 + S(U,phi)
-      // 
-      // .
-      // U =  P U    (lorentz & color indices multiplied)
-      //
-      // Hence
-      //
-      // .c    c  c       c
-      // U =  U  P   = - U  P        (c == dagger)
-      //
-      // So, taking some liberty with implicit indices
-      //                  .      .          .c      c
-      // dH/dt = 0 = Tr P P +Tr[ U  dS/dU + U  dS/dU  ]
-      //
-      //                .                          c      c
-      //           = Tr P P + i Tr[  P U dS/dU  - U   P dS/dU  ]
-      //
-      //                   .                        c  c
-      //           = Tr P (P + i ( U dS/dU - P dS/dU  U ]
-      //
-      //              .                        c  c
-      //           => P  = -i [ U dS/dU - dS/dU  U ]      generates HMC EoM
-      //
-      // Simple case work this out using S = phi^dag MdagM phi for wilson:
-      //                               c       c
-      // dSdt     = dU_xdt  dSdUx  + dUxdt dSdUx
-      //          
-      //         = Tr i P U_x [ (\phi^\dag)_x (1+g) (M \phi)_x+\mu   +(\phi^\dag M^\dag)_x (1-g) \phi_{x+\mu} ]
-      //                 c
-      //            - i U_x P [ (\phi^\dag)_x+mu (1-g) (M \phi)_x    +(\phi^\dag M^\dag)_(x+\mu) (1+g) \phi_{x} ]
-      //
-      //         = i [(\phi^\dag)_x      ]_j P_jk  [U_x(1+g) (M \phi)_x+\mu]_k        (1)
-      //         + i [(\phi^\dagM^\dag)_x]_j P_jk  [U_x(1-g) (\phi)_x+\mu]_k          (2)
-      //         - i [(\phi^\dag)_x+mu (1-g) U^dag_x]_j P_jk  [(M \phi)_xk            (3)
-      //         - i [(\phi^\dagM^\dag)_x+mu (1+g) U^dag_x]_j P_jk  [ \phi]_xk        (4)
-      //
-      // Observe that (1)* = (4)
-      //              (2)* = (3)
-      //
-      // Write as    .
-      //             P_{kj}  = - i (  [U_x(1+g) (M \phi)_x+\mu] (x) [(\phi^\dag)_x] + [U_x(1-g) (\phi)_x+\mu] (x) [(\phi^\dagM^\dag)_x] - h.c )
-      //
-      // where (x) denotes outer product in colour and spins are traced.
-      //
-      // Need only evaluate (1) and (2)  [Chroma] or (2) and (4) [IroIro] and take the 
-      // traceless anti hermitian part (of term in brackets w/o the "i")
-      // 
-      // Generalisation to S=phi^dag (MdagM)^{-1} phi is simple:
-      //
-      // For more complicated DWF etc... apply product rule in differentiation
-      //
+      // Derivative interface
       ////////////////////////
-      void DhopDeriv  (LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
-      void DhopDerivEO(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
-      void DhopDerivOE(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
-
-      // Extra support internal
-      void DerivInternal(CartesianStencil & st,
-			 LatticeDoubledGaugeField & U,
-			 LatticeGaugeField &mat,
-			 const LatticeFermion &A,
-			 const LatticeFermion &B,
-			 int dag);
+      // Interface calls an internal routine
+      void DhopDeriv(GaugeField &mat,const FermionField &U,const FermionField &V,int dag);
+      void DhopDerivOE(GaugeField &mat,const FermionField &U,const FermionField &V,int dag);
+      void DhopDerivEO(GaugeField &mat,const FermionField &U,const FermionField &V,int dag);
 
 
+      ///////////////////////////////////////////////////////////////
       // non-hermitian hopping term; half cb or both
-      void Dhop  (const LatticeFermion &in, LatticeFermion &out,int dag);
-      void DhopOE(const LatticeFermion &in, LatticeFermion &out,int dag);
-      void DhopEO(const LatticeFermion &in, LatticeFermion &out,int dag);
+      ///////////////////////////////////////////////////////////////
+      void Dhop(const FermionField &in, FermionField &out,int dag) ;
+      void DhopOE(const FermionField &in, FermionField &out,int dag) ;
+      void DhopEO(const FermionField &in, FermionField &out,int dag) ;
 
-      // Multigrid assistance
-      void   Mdir (const LatticeFermion &in, LatticeFermion &out,int dir,int disp);
-      void DhopDir(const LatticeFermion &in, LatticeFermion &out,int dir,int disp);
-      void DhopDirDisp(const LatticeFermion &in, LatticeFermion &out,int dirdisp,int gamma,int dag);
+      ///////////////////////////////////////////////////////////////
+      // Multigrid assistance; force term uses too
+      ///////////////////////////////////////////////////////////////
+      void Mdir (const FermionField &in, FermionField &out,int dir,int disp) ;
+      void DhopDir(const FermionField &in, FermionField &out,int dir,int disp);
+      void DhopDirDisp(const FermionField &in, FermionField &out,int dirdisp,int gamma,int dag) ;
 
       ///////////////////////////////////////////////////////////////
       // Extra methods added by derived
       ///////////////////////////////////////////////////////////////
-      void DhopInternal(CartesianStencil & st,
-			LatticeDoubledGaugeField &U,
-			const LatticeFermion &in, 
-			LatticeFermion &out,
-			int dag);
+      void DerivInternal(CartesianStencil & st,
+			 DoubledGaugeField & U,
+			 GaugeField &mat,
+			 const FermionField &A,
+			 const FermionField &B,
+			 int dag);
+
+      void DhopInternal(CartesianStencil & st,DoubledGaugeField & U,
+			const FermionField &in, FermionField &out,int dag) ;
+
 
       // Constructor
-      WilsonFermion(LatticeGaugeField &_Umu,GridCartesian &Fgrid,GridRedBlackCartesian &Hgrid,RealD _mass);
+      WilsonFermion(GaugeField &_Umu,
+		    GridCartesian         &Fgrid,
+		    GridRedBlackCartesian &Hgrid, 
+		    RealD _mass) ;
 
-      // DoubleStore
-      virtual void ImportGauge(const LatticeGaugeField &_Umu);
-      void DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu);
+      // DoubleStore impl dependent
+      void ImportGauge(const GaugeField &_Umu);
 
       ///////////////////////////////////////////////////////////////
       // Data members require to support the functionality
       ///////////////////////////////////////////////////////////////
-      static int HandOptDslash; // these are a temporary hack
-      static int MortonOrder;
 
       //    protected:
     public:
@@ -153,26 +105,24 @@ namespace Grid {
       GridBase                     *    _grid; 
       GridBase                     *  _cbgrid;
 
-      static const int npoint=8;
-      static const std::vector<int> directions   ;
-      static const std::vector<int> displacements;
-
       //Defines the stencils for even and odd
       CartesianStencil Stencil; 
       CartesianStencil StencilEven; 
       CartesianStencil StencilOdd; 
 
       // Copy of the gauge field , with even and odd subsets
-      LatticeDoubledGaugeField Umu;
-      LatticeDoubledGaugeField UmuEven;
-      LatticeDoubledGaugeField UmuOdd;
+      DoubledGaugeField Umu;
+      DoubledGaugeField UmuEven;
+      DoubledGaugeField UmuOdd;
 
       // Comms buffer
-      std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  comm_buf;
-
+      std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> >  comm_buf;
       
     };
 
+    typedef WilsonFermion<WilsonImplF> WilsonFermionF;
+    typedef WilsonFermion<WilsonImplD> WilsonFermionD;
+
   }
 }
 #endif
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.cc b/lib/qcd/action/fermion/WilsonFermion5D.cc
index cbe66bb5..f324db8d 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.cc
+++ b/lib/qcd/action/fermion/WilsonFermion5D.cc
@@ -3,19 +3,19 @@
 namespace Grid {
 namespace QCD {
   
-  // S-direction is INNERMOST and takes no part in the parity.
-  const std::vector<int> WilsonFermion5D::directions   ({1,2,3,4, 1, 2, 3, 4});
-  const std::vector<int> WilsonFermion5D::displacements({1,1,1,1,-1,-1,-1,-1});
-
-  int WilsonFermion5D::HandOptDslash;
+// S-direction is INNERMOST and takes no part in the parity.
+const std::vector<int> WilsonFermion5DStatic::directions   ({1,2,3,4, 1, 2, 3, 4});
+const std::vector<int> WilsonFermion5DStatic::displacements({1,1,1,1,-1,-1,-1,-1});
+int WilsonFermion5DStatic::HandOptDslash;
 
   // 5d lattice for DWF.
-  WilsonFermion5D::WilsonFermion5D(LatticeGaugeField &_Umu,
-					   GridCartesian         &FiveDimGrid,
-					   GridRedBlackCartesian &FiveDimRedBlackGrid,
-					   GridCartesian         &FourDimGrid,
-					   GridRedBlackCartesian &FourDimRedBlackGrid,
-					   RealD _M5) :
+template<class Impl>
+WilsonFermion5D<Impl>::WilsonFermion5D(GaugeField &_Umu,
+				       GridCartesian         &FiveDimGrid,
+				       GridRedBlackCartesian &FiveDimRedBlackGrid,
+				       GridCartesian         &FourDimGrid,
+				       GridRedBlackCartesian &FourDimRedBlackGrid,
+				       RealD _M5) :
   _FiveDimGrid(&FiveDimGrid),
   _FiveDimRedBlackGrid(&FiveDimRedBlackGrid),
   _FourDimGrid(&FourDimGrid),
@@ -68,34 +68,23 @@ namespace QCD {
 
   ImportGauge(_Umu);
 }  
-void WilsonFermion5D::ImportGauge(const LatticeGaugeField &_Umu)
+template<class Impl>
+void WilsonFermion5D<Impl>::ImportGauge(const GaugeField &_Umu)
 {
-  DoubleStore(Umu,_Umu);
+  Impl::DoubleStore(GaugeGrid(),Umu,_Umu);
   pickCheckerboard(Even,UmuEven,Umu);
   pickCheckerboard(Odd ,UmuOdd,Umu);
 }
-void WilsonFermion5D::DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu)
-{
-  assert(GaugeGrid()->_ndimension==4);
-  conformable(Uds._grid,GaugeGrid());
-  conformable(Umu._grid,GaugeGrid());
-  LatticeColourMatrix U(GaugeGrid());
-  for(int mu=0;mu<Nd;mu++){
-    U = PeekIndex<LorentzIndex>(Umu,mu);
-    PokeIndex<LorentzIndex>(Uds,U,mu);
-    U = adj(Cshift(U,mu,-1));
-    PokeIndex<LorentzIndex>(Uds,U,mu+4);
-  }
-}
-void WilsonFermion5D::DhopDir(const LatticeFermion &in, LatticeFermion &out,int dir5,int disp)
+template<class Impl>
+void WilsonFermion5D<Impl>::DhopDir(const FermionField &in, FermionField &out,int dir5,int disp)
 {
   int dir = dir5-1; // Maps to the ordering above in "directions" that is passed to stencil
                     // we drop off the innermost fifth dimension
   //  assert( (disp==1)||(disp==-1) );
   //  assert( (dir>=0)&&(dir<4) ); //must do x,y,z or t;
 
-  WilsonCompressor compressor(DaggerNo);
-  Stencil.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
+  Compressor compressor(DaggerNo);
+  Stencil.HaloExchange<SiteSpinor,SiteHalfSpinor,Compressor>(in,comm_buf,compressor);
   
   int skip = (disp==1) ? 0 : 1;
 
@@ -109,16 +98,17 @@ PARALLEL_FOR_LOOP
     for(int s=0;s<Ls;s++){
       int sU=ss;
       int sF = s+Ls*sU; 
-      DiracOptDhopDir(Stencil,Umu,comm_buf,sF,sU,in,out,dirdisp,dirdisp);
+      Kernels::DiracOptDhopDir(Stencil,Umu,comm_buf,sF,sU,in,out,dirdisp,dirdisp);
     }
   }
 };
 
-void WilsonFermion5D::DerivInternal(CartesianStencil & st,
-				    LatticeDoubledGaugeField & U,
-				    LatticeGaugeField &mat,
-				    const LatticeFermion &A,
-				    const LatticeFermion &B,
+template<class Impl>
+void WilsonFermion5D<Impl>::DerivInternal(CartesianStencil & st,
+				    DoubledGaugeField & U,
+				    GaugeField &mat,
+				    const FermionField &A,
+				    const FermionField &B,
 				    int dag)
 {
   assert((dag==DaggerNo) ||(dag==DaggerYes));
@@ -126,13 +116,13 @@ void WilsonFermion5D::DerivInternal(CartesianStencil & st,
   conformable(st._grid,A._grid);
   conformable(st._grid,B._grid);
 
-  WilsonCompressor compressor(dag);
+  Compressor compressor(dag);
   
-  LatticeColourMatrix tmp(mat._grid);
-  LatticeFermion Btilde(B._grid);
-  LatticeFermion Atilde(B._grid);
+  GaugeLinkField tmp(mat._grid);
+  FermionField Btilde(B._grid);
+  FermionField Atilde(B._grid);
 
-  st.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(B,comm_buf,compressor);
+  st.HaloExchange<SiteSpinor,SiteHalfSpinor,Compressor>(B,comm_buf,compressor);
 
   Atilde=A;
 
@@ -158,7 +148,7 @@ PARALLEL_FOR_LOOP
 	assert ( sF< B._grid->oSites());
 	assert ( sU< U._grid->oSites());
 
-	DiracOptDhopDir(st,U,comm_buf,sF,sU,B,Btilde,mu,gamma);
+	Kernels::DiracOptDhopDir(st,U,comm_buf,sF,sU,B,Btilde,mu,gamma);
 
     ////////////////////////////
     // spin trace outer product
@@ -176,10 +166,11 @@ PARALLEL_FOR_LOOP
   }
 }
 
-void WilsonFermion5D::DhopDeriv(      LatticeGaugeField &mat,
-				const LatticeFermion &A,
-				const LatticeFermion &B,
-				int dag)
+template<class Impl>
+void WilsonFermion5D<Impl>::DhopDeriv(      GaugeField &mat,
+					    const FermionField &A,
+					    const FermionField &B,
+					    int dag)
 {
   conformable(A._grid,FermionGrid());  
   conformable(A._grid,B._grid);
@@ -190,10 +181,11 @@ void WilsonFermion5D::DhopDeriv(      LatticeGaugeField &mat,
   DerivInternal(Stencil,Umu,mat,A,B,dag);
 }
 
-void WilsonFermion5D::DhopDerivEO(LatticeGaugeField &mat,
-				  const LatticeFermion &A,
-				  const LatticeFermion &B,
-				  int dag)
+template<class Impl>
+void WilsonFermion5D<Impl>::DhopDerivEO(GaugeField &mat,
+					const FermionField &A,
+					const FermionField &B,
+					int dag)
 {
   conformable(A._grid,FermionRedBlackGrid());
   conformable(GaugeRedBlackGrid(),mat._grid);
@@ -206,9 +198,10 @@ void WilsonFermion5D::DhopDerivEO(LatticeGaugeField &mat,
   DerivInternal(StencilOdd,UmuEven,mat,A,B,dag);
 }
 
-void WilsonFermion5D::DhopDerivOE(LatticeGaugeField &mat,
-				  const LatticeFermion &A,
-				  const LatticeFermion &B,
+template<class Impl>
+void WilsonFermion5D<Impl>::DhopDerivOE(GaugeField &mat,
+				  const FermionField &A,
+				  const FermionField &B,
 				  int dag)
 {
   conformable(A._grid,FermionRedBlackGrid());
@@ -222,15 +215,16 @@ void WilsonFermion5D::DhopDerivOE(LatticeGaugeField &mat,
   DerivInternal(StencilEven,UmuOdd,mat,A,B,dag);
 }
 
-void WilsonFermion5D::DhopInternal(CartesianStencil & st, LebesgueOrder &lo,
-				   LatticeDoubledGaugeField & U,
-				   const LatticeFermion &in, LatticeFermion &out,int dag)
+template<class Impl>
+void WilsonFermion5D<Impl>::DhopInternal(CartesianStencil & st, LebesgueOrder &lo,
+					 DoubledGaugeField & U,
+					 const FermionField &in, FermionField &out,int dag)
 {
   //  assert((dag==DaggerNo) ||(dag==DaggerYes));
 
-  WilsonCompressor compressor(dag);
+  Compressor compressor(dag);
 
-  st.HaloExchange<vSpinColourVector,vHalfSpinColourVector,WilsonCompressor>(in,comm_buf,compressor);
+  st.HaloExchange<SiteSpinor,SiteHalfSpinor,Compressor>(in,comm_buf,compressor);
   
   // Dhop takes the 4d grid from U, and makes a 5d index for fermion
   // Not loop ordering and data layout.
@@ -238,13 +232,13 @@ void WilsonFermion5D::DhopInternal(CartesianStencil & st, LebesgueOrder &lo,
   // - per thread reuse in L1 cache for U
   // - 8 linear access unit stride streams per thread for Fermion for hw prefetchable.
   if ( dag == DaggerYes ) {
-    if( HandOptDslash ) {
+    if( this->HandOptDslash ) {
 PARALLEL_FOR_LOOP
       for(int ss=0;ss<U._grid->oSites();ss++){
 	for(int s=0;s<Ls;s++){
 	  int sU=ss;
 	  int sF = s+Ls*sU;
-	  DiracOptHandDhopSiteDag(st,U,comm_buf,sF,sU,in,out);
+	  Kernels::DiracOptHandDhopSiteDag(st,U,comm_buf,sF,sU,in,out);
 	  }
       }
     } else { 
@@ -255,20 +249,20 @@ PARALLEL_FOR_LOOP
 	  for(sd=0;sd<Ls;sd++){
 	    int sU=ss;
 	    int sF = sd+Ls*sU;
-	    DiracOptDhopSiteDag(st,U,comm_buf,sF,sU,in,out);
+	    Kernels::DiracOptDhopSiteDag(st,U,comm_buf,sF,sU,in,out);
 	  }
 	}
       }
     }
   } else {
-    if( HandOptDslash ) {
+    if( this->HandOptDslash ) {
 PARALLEL_FOR_LOOP
       for(int ss=0;ss<U._grid->oSites();ss++){
 	for(int s=0;s<Ls;s++){
 	  //	  int sU=lo.Reorder(ss);
 	  int sU=ss;
 	  int sF = s+Ls*sU;
-	  DiracOptHandDhopSite(st,U,comm_buf,sF,sU,in,out);
+	  Kernels::DiracOptHandDhopSite(st,U,comm_buf,sF,sU,in,out);
 	}
       }
 
@@ -279,13 +273,14 @@ PARALLEL_FOR_LOOP
 	  //	  int sU=lo.Reorder(ss);
 	  int sU=ss;
 	  int sF = s+Ls*sU; 
-	  DiracOptDhopSite(st,U,comm_buf,sF,sU,in,out);
+	  Kernels::DiracOptDhopSite(st,U,comm_buf,sF,sU,in,out);
 	}
       }
     }
   }
 }
-void WilsonFermion5D::DhopOE(const LatticeFermion &in, LatticeFermion &out,int dag)
+template<class Impl>
+void WilsonFermion5D<Impl>::DhopOE(const FermionField &in, FermionField &out,int dag)
 {
   conformable(in._grid,FermionRedBlackGrid());    // verifies half grid
   conformable(in._grid,out._grid); // drops the cb check
@@ -295,7 +290,8 @@ void WilsonFermion5D::DhopOE(const LatticeFermion &in, LatticeFermion &out,int d
 
   DhopInternal(StencilEven,LebesgueEvenOdd,UmuOdd,in,out,dag);
 }
-void WilsonFermion5D::DhopEO(const LatticeFermion &in, LatticeFermion &out,int dag)
+template<class Impl>
+void WilsonFermion5D<Impl>::DhopEO(const FermionField &in, FermionField &out,int dag)
 {
   conformable(in._grid,FermionRedBlackGrid());    // verifies half grid
   conformable(in._grid,out._grid); // drops the cb check
@@ -305,7 +301,8 @@ void WilsonFermion5D::DhopEO(const LatticeFermion &in, LatticeFermion &out,int d
 
   DhopInternal(StencilOdd,LebesgueEvenOdd,UmuEven,in,out,dag);
 }
-void WilsonFermion5D::Dhop(const LatticeFermion &in, LatticeFermion &out,int dag)
+template<class Impl>
+void WilsonFermion5D<Impl>::Dhop(const FermionField &in, FermionField &out,int dag)
 {
   conformable(in._grid,FermionGrid()); // verifies full grid
   conformable(in._grid,out._grid);
@@ -314,12 +311,16 @@ void WilsonFermion5D::Dhop(const LatticeFermion &in, LatticeFermion &out,int dag
 
   DhopInternal(Stencil,Lebesgue,Umu,in,out,dag);
 }
-void WilsonFermion5D::DW(const LatticeFermion &in, LatticeFermion &out,int dag)
+template<class Impl>
+void WilsonFermion5D<Impl>::DW(const FermionField &in, FermionField &out,int dag)
 {
   out.checkerboard=in.checkerboard;
   Dhop(in,out,dag); // -0.5 is included
   axpy(out,4.0-M5,in,out);
 }
+
+FermOpTemplateInstantiate(WilsonFermion5D);
+
 }}
 
 
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.h b/lib/qcd/action/fermion/WilsonFermion5D.h
index d5cff3e6..5159d9e1 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.h
+++ b/lib/qcd/action/fermion/WilsonFermion5D.h
@@ -14,21 +14,21 @@ namespace Grid {
     // i.e. even even contains fifth dim hopping term.
     //
     // [DIFFERS from original CPS red black implementation parity = (x+y+z+t+s)|2 ]
-    ////////////////////////////
-    //ContFrac:
-    //  Ls always odd. Rational poly deg is either Ls or Ls-1
-    //PartFrac 
-    //  Ls always odd. Rational poly deg is either Ls or Ls-1
-    //
-    //Cayley: Ls always even, Rational poly deg is Ls
-    // 
-    // Just set nrational as Ls. Forget about Ls-1 cases.
-    //
-    // Require odd Ls for cont and part frac
-    ////////////////////////////
     ////////////////////////////////////////////////////////////////////////////////
-    class WilsonFermion5D : public FermionOperator<LatticeFermion,LatticeGaugeField>
+
+    class WilsonFermion5DStatic { 
+    public:
+      // S-direction is INNERMOST and takes no part in the parity.
+      static int HandOptDslash; // these are a temporary hack
+      static const std::vector<int> directions;
+      static const std::vector<int> displacements;
+      const int npoint = 8;
+    };
+
+    template<class Impl>
+    class WilsonFermion5D : public FermionOperator<Impl>, public WilsonFermion5DStatic
     {
+#include <qcd/action/fermion/FermionImplTypedefs.h>
     public:
       ///////////////////////////////////////////////////////////////
       // Implement the abstract base
@@ -39,69 +39,65 @@ namespace Grid {
       GridBase *FermionRedBlackGrid(void)    { return _FiveDimRedBlackGrid;}
 
       // full checkerboard operations; leave unimplemented as abstract for now
-      virtual RealD  M    (const LatticeFermion &in, LatticeFermion &out){assert(0); return 0.0;};
-      virtual RealD  Mdag (const LatticeFermion &in, LatticeFermion &out){assert(0); return 0.0;};
+      virtual RealD  M    (const FermionField &in, FermionField &out){assert(0); return 0.0;};
+      virtual RealD  Mdag (const FermionField &in, FermionField &out){assert(0); return 0.0;};
 
       // half checkerboard operations; leave unimplemented as abstract for now
-      virtual void   Meooe       (const LatticeFermion &in, LatticeFermion &out){assert(0);};
-      virtual void   Mooee       (const LatticeFermion &in, LatticeFermion &out){assert(0);};
-      virtual void   MooeeInv    (const LatticeFermion &in, LatticeFermion &out){assert(0);};
+      virtual void   Meooe       (const FermionField &in, FermionField &out){assert(0);};
+      virtual void   Mooee       (const FermionField &in, FermionField &out){assert(0);};
+      virtual void   MooeeInv    (const FermionField &in, FermionField &out){assert(0);};
 
-      virtual void   MeooeDag    (const LatticeFermion &in, LatticeFermion &out){assert(0);};
-      virtual void   MooeeDag    (const LatticeFermion &in, LatticeFermion &out){assert(0);};
-      virtual void   MooeeInvDag (const LatticeFermion &in, LatticeFermion &out){assert(0);};
+      virtual void   MeooeDag    (const FermionField &in, FermionField &out){assert(0);};
+      virtual void   MooeeDag    (const FermionField &in, FermionField &out){assert(0);};
+      virtual void   MooeeInvDag (const FermionField &in, FermionField &out){assert(0);};
 
-      // These can be overridden by fancy 5d chiral actions
-      virtual void DhopDeriv  (LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
-      virtual void DhopDerivEO(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
-      virtual void DhopDerivOE(LatticeGaugeField &mat,const LatticeFermion &U,const LatticeFermion &V,int dag);
+      // These can be overridden by fancy 5d chiral action
+      virtual void DhopDeriv  (GaugeField &mat,const FermionField &U,const FermionField &V,int dag);
+      virtual void DhopDerivEO(GaugeField &mat,const FermionField &U,const FermionField &V,int dag);
+      virtual void DhopDerivOE(GaugeField &mat,const FermionField &U,const FermionField &V,int dag);
 
       // Implement hopping term non-hermitian hopping term; half cb or both
       // Implement s-diagonal DW
-      void DW    (const LatticeFermion &in, LatticeFermion &out,int dag);
-      void Dhop  (const LatticeFermion &in, LatticeFermion &out,int dag);
-      void DhopOE(const LatticeFermion &in, LatticeFermion &out,int dag);
-      void DhopEO(const LatticeFermion &in, LatticeFermion &out,int dag);
+      void DW    (const FermionField &in, FermionField &out,int dag);
+      void Dhop  (const FermionField &in, FermionField &out,int dag);
+      void DhopOE(const FermionField &in, FermionField &out,int dag);
+      void DhopEO(const FermionField &in, FermionField &out,int dag);
 
       // add a DhopComm
       // -- suboptimal interface will presently trigger multiple comms.
-      void DhopDir(const LatticeFermion &in, LatticeFermion &out,int dir,int disp);
+      void DhopDir(const FermionField &in, FermionField &out,int dir,int disp);
 
       ///////////////////////////////////////////////////////////////
       // New methods added 
       ///////////////////////////////////////////////////////////////
-
       void DerivInternal(CartesianStencil & st,
-			 LatticeDoubledGaugeField & U,
-			 LatticeGaugeField &mat,
-			 const LatticeFermion &A,
-			 const LatticeFermion &B,
+			 DoubledGaugeField & U,
+			 GaugeField &mat,
+			 const FermionField &A,
+			 const FermionField &B,
 			 int dag);
 
       void DhopInternal(CartesianStencil & st,
 			LebesgueOrder &lo,
-			LatticeDoubledGaugeField &U,
-			const LatticeFermion &in, 
-			LatticeFermion &out,
+			DoubledGaugeField &U,
+			const FermionField &in, 
+			FermionField &out,
 			int dag);
 
       // Constructors
-      WilsonFermion5D(LatticeGaugeField &_Umu,
-			  GridCartesian         &FiveDimGrid,
-			  GridRedBlackCartesian &FiveDimRedBlackGrid,
-			  GridCartesian         &FourDimGrid,
-			  GridRedBlackCartesian &FourDimRedBlackGrid,
-			  double _M5);
+      WilsonFermion5D(GaugeField &_Umu,
+		      GridCartesian         &FiveDimGrid,
+		      GridRedBlackCartesian &FiveDimRedBlackGrid,
+		      GridCartesian         &FourDimGrid,
+		      GridRedBlackCartesian &FourDimRedBlackGrid,
+		      double _M5);
 
       // DoubleStore
-      virtual void ImportGauge(const LatticeGaugeField &_Umu);
-      void DoubleStore(LatticeDoubledGaugeField &Uds,const LatticeGaugeField &Umu);
+      void ImportGauge(const GaugeField &_Umu);
 
       ///////////////////////////////////////////////////////////////
       // Data members require to support the functionality
       ///////////////////////////////////////////////////////////////
-      static int HandOptDslash; // these are a temporary hack
-
     protected:
 
       // Add these to the support from Wilson
@@ -110,10 +106,6 @@ namespace Grid {
       GridBase *_FiveDimGrid;
       GridBase *_FiveDimRedBlackGrid;
 
-      static const int npoint=8;
-      static const std::vector<int> directions   ;
-      static const std::vector<int> displacements;
-
       double                        M5;
       int Ls;
 
@@ -123,15 +115,15 @@ namespace Grid {
       CartesianStencil StencilOdd; 
 
       // Copy of the gauge field , with even and odd subsets
-      LatticeDoubledGaugeField Umu;
-      LatticeDoubledGaugeField UmuEven;
-      LatticeDoubledGaugeField UmuOdd;
+      DoubledGaugeField Umu;
+      DoubledGaugeField UmuEven;
+      DoubledGaugeField UmuOdd;
 
       LebesgueOrder Lebesgue;
       LebesgueOrder LebesgueEvenOdd;
 
       // Comms buffer
-      std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  comm_buf;
+      std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> >  comm_buf;
       
     };
   }
diff --git a/lib/qcd/action/fermion/WilsonKernels.cc b/lib/qcd/action/fermion/WilsonKernels.cc
index 2b8b7929..6d558e03 100644
--- a/lib/qcd/action/fermion/WilsonKernels.cc
+++ b/lib/qcd/action/fermion/WilsonKernels.cc
@@ -2,23 +2,317 @@
 namespace Grid {
 namespace QCD {
 
-void DiracOptDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
-			std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			int sF,int sU,const LatticeFermion &in, LatticeFermion &out)
+template<class Impl> 
+void WilsonKernels<Impl>::DiracOptDhopSite(CartesianStencil &st,DoubledGaugeField &U,
+						  std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> >  &buf,
+						  int sF,int sU,const FermionField &in, FermionField &out)
 {
-    vHalfSpinColourVector  tmp;    
-    vHalfSpinColourVector  chi;    
-    vSpinColourVector result;
-    vHalfSpinColourVector Uchi;
-    int offset,local,perm, ptype;
+  SiteHalfSpinor  tmp;    
+  SiteHalfSpinor  chi;    
+  SiteSpinor result;
+  SiteHalfSpinor Uchi;
+  int offset,local,perm, ptype;
 
-    // Xp
-    int ss = sF;
-    offset = st._offsets [Xp][ss];
-    local  = st._is_local[Xp][ss];
-    perm   = st._permute[Xp][ss];
+  // Xp
+  int ss = sF;
+  offset = st._offsets [Xp][ss];
+  local  = st._is_local[Xp][ss];
+  perm   = st._permute[Xp][ss];
+  
+  ptype  = st._permute_type[Xp];
+  if ( local && perm ) {
+    spProjXp(tmp,in._odata[offset]);
+    permute(chi,tmp,ptype);
+  } else if ( local ) {
+    spProjXp(chi,in._odata[offset]);
+  } else { 
+    chi=buf[offset];
+  }
+  Impl::multLink(Uchi,U._odata[sU],chi,Xp);
+  spReconXp(result,Uchi);
+    
+  // Yp
+  offset = st._offsets [Yp][ss];
+  local  = st._is_local[Yp][ss];
+  perm   = st._permute[Yp][ss];
+  ptype  = st._permute_type[Yp];
+  if ( local && perm ) {
+    spProjYp(tmp,in._odata[offset]);
+    permute(chi,tmp,ptype);
+  } else if ( local ) {
+    spProjYp(chi,in._odata[offset]);
+  } else { 
+    chi=buf[offset];
+  }
+  Impl::multLink(Uchi,U._odata[sU],chi,Yp);
+  accumReconYp(result,Uchi);
 
-    ptype  = st._permute_type[Xp];
+  // Zp
+  offset = st._offsets [Zp][ss];
+  local  = st._is_local[Zp][ss];
+  perm   = st._permute[Zp][ss];
+  ptype  = st._permute_type[Zp];
+  if ( local && perm ) {
+    spProjZp(tmp,in._odata[offset]);
+    permute(chi,tmp,ptype);
+  } else if ( local ) {
+    spProjZp(chi,in._odata[offset]);
+  } else { 
+    chi=buf[offset];
+  }
+  Impl::multLink(Uchi,U._odata[sU],chi,Zp);
+  accumReconZp(result,Uchi);
+
+  // Tp
+  offset = st._offsets [Tp][ss];
+  local  = st._is_local[Tp][ss];
+  perm   = st._permute[Tp][ss];
+  ptype  = st._permute_type[Tp];
+  if ( local && perm ) {
+    spProjTp(tmp,in._odata[offset]);
+    permute(chi,tmp,ptype);
+  } else if ( local ) {
+    spProjTp(chi,in._odata[offset]);
+  } else { 
+    chi=buf[offset];
+  }
+  Impl::multLink(Uchi,U._odata[sU],chi,Tp);
+  accumReconTp(result,Uchi);
+
+  // Xm
+  offset = st._offsets [Xm][ss];
+  local  = st._is_local[Xm][ss];
+  perm   = st._permute[Xm][ss];
+  ptype  = st._permute_type[Xm];
+  
+  if ( local && perm ) {
+    spProjXm(tmp,in._odata[offset]);
+    permute(chi,tmp,ptype);
+  } else if ( local ) {
+    spProjXm(chi,in._odata[offset]);
+  } else { 
+    chi=buf[offset];
+  }
+  Impl::multLink(Uchi,U._odata[sU],chi,Xm);
+  accumReconXm(result,Uchi);
+  
+  // Ym
+  offset = st._offsets [Ym][ss];
+  local  = st._is_local[Ym][ss];
+  perm   = st._permute[Ym][ss];
+  ptype  = st._permute_type[Ym];
+  
+  if ( local && perm ) {
+    spProjYm(tmp,in._odata[offset]);
+    permute(chi,tmp,ptype);
+  } else if ( local ) {
+    spProjYm(chi,in._odata[offset]);
+  } else { 
+    chi=buf[offset];
+  }
+  Impl::multLink(Uchi,U._odata[sU],chi,Ym);
+  accumReconYm(result,Uchi);
+  
+  // Zm
+  offset = st._offsets [Zm][ss];
+  local  = st._is_local[Zm][ss];
+  perm   = st._permute[Zm][ss];
+  ptype  = st._permute_type[Zm];
+  if ( local && perm ) {
+    spProjZm(tmp,in._odata[offset]);
+    permute(chi,tmp,ptype);
+  } else if ( local ) {
+    spProjZm(chi,in._odata[offset]);
+  } else { 
+    chi=buf[offset];
+  }
+  Impl::multLink(Uchi,U._odata[sU],chi,Zm);
+  accumReconZm(result,Uchi);
+
+  // Tm
+  offset = st._offsets [Tm][ss];
+  local  = st._is_local[Tm][ss];
+  perm   = st._permute[Tm][ss];
+  ptype  = st._permute_type[Tm];
+  if ( local && perm ) {
+    spProjTm(tmp,in._odata[offset]);
+    permute(chi,tmp,ptype);
+  } else if ( local ) {
+    spProjTm(chi,in._odata[offset]);
+  } else { 
+    chi=buf[offset];
+  }
+  Impl::multLink(Uchi,U._odata[sU],chi,Tm);
+  accumReconTm(result,Uchi);
+
+  vstream(out._odata[ss],result*(-0.5));
+};
+
+template<class Impl> 
+void WilsonKernels<Impl>::DiracOptDhopSiteDag(CartesianStencil &st,DoubledGaugeField &U,
+					      std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> >  &buf,
+					      int sF,int sU,const FermionField &in, FermionField &out)
+{
+  SiteHalfSpinor  tmp;    
+  SiteHalfSpinor  chi;    
+  SiteSpinor result;
+  SiteHalfSpinor Uchi;
+  int offset,local,perm, ptype;
+
+  // Xp
+  int ss=sF;
+  offset = st._offsets [Xm][ss];
+  local  = st._is_local[Xm][ss];
+  perm   = st._permute[Xm][ss];
+
+  ptype  = st._permute_type[Xm];
+  if ( local && perm ) {
+    spProjXp(tmp,in._odata[offset]);
+    permute(chi,tmp,ptype);
+  } else if ( local ) {
+    spProjXp(chi,in._odata[offset]);
+  } else { 
+    chi=buf[offset];
+  }
+  Impl::multLink(Uchi,U._odata[sU],chi,Xm);
+  spReconXp(result,Uchi);
+
+  // Yp
+  offset = st._offsets [Ym][ss];
+  local  = st._is_local[Ym][ss];
+  perm   = st._permute[Ym][ss];
+  ptype  = st._permute_type[Ym];
+  if ( local && perm ) {
+    spProjYp(tmp,in._odata[offset]);
+    permute(chi,tmp,ptype);
+  } else if ( local ) {
+    spProjYp(chi,in._odata[offset]);
+  } else { 
+    chi=buf[offset];
+  }
+  Impl::multLink(Uchi,U._odata[sU],chi,Ym);
+  accumReconYp(result,Uchi);
+  
+  // Zp
+  offset = st._offsets [Zm][ss];
+  local  = st._is_local[Zm][ss];
+  perm   = st._permute[Zm][ss];
+  ptype  = st._permute_type[Zm];
+  if ( local && perm ) {
+    spProjZp(tmp,in._odata[offset]);
+    permute(chi,tmp,ptype);
+  } else if ( local ) {
+    spProjZp(chi,in._odata[offset]);
+  } else { 
+    chi=buf[offset];
+  }
+  Impl::multLink(Uchi,U._odata[sU],chi,Zm);
+  accumReconZp(result,Uchi);
+  
+  // Tp
+  offset = st._offsets [Tm][ss];
+  local  = st._is_local[Tm][ss];
+  perm   = st._permute[Tm][ss];
+  ptype  = st._permute_type[Tm];
+  if ( local && perm ) {
+    spProjTp(tmp,in._odata[offset]);
+    permute(chi,tmp,ptype);
+  } else if ( local ) {
+    spProjTp(chi,in._odata[offset]);
+  } else { 
+    chi=buf[offset];
+  }
+  Impl::multLink(Uchi,U._odata[sU],chi,Tm);
+  accumReconTp(result,Uchi);
+  
+  // Xm
+  offset = st._offsets [Xp][ss];
+  local  = st._is_local[Xp][ss];
+  perm   = st._permute[Xp][ss];
+  ptype  = st._permute_type[Xp];
+
+  if ( local && perm ) {
+    spProjXm(tmp,in._odata[offset]);
+    permute(chi,tmp,ptype);
+  } else if ( local ) {
+    spProjXm(chi,in._odata[offset]);
+  } else { 
+    chi=buf[offset];
+  }
+  Impl::multLink(Uchi,U._odata[sU],chi,Xp);
+  accumReconXm(result,Uchi);
+
+  // Ym
+  offset = st._offsets [Yp][ss];
+  local  = st._is_local[Yp][ss];
+  perm   = st._permute[Yp][ss];
+  ptype  = st._permute_type[Yp];
+
+  if ( local && perm ) {
+    spProjYm(tmp,in._odata[offset]);
+    permute(chi,tmp,ptype);
+  } else if ( local ) {
+    spProjYm(chi,in._odata[offset]);
+  } else { 
+    chi=buf[offset];
+  }
+  Impl::multLink(Uchi,U._odata[sU],chi,Yp);
+  accumReconYm(result,Uchi);
+
+  // Zm
+  offset = st._offsets [Zp][ss];
+  local  = st._is_local[Zp][ss];
+  perm   = st._permute[Zp][ss];
+  ptype  = st._permute_type[Zp];
+  if ( local && perm ) {
+    spProjZm(tmp,in._odata[offset]);
+    permute(chi,tmp,ptype);
+  } else if ( local ) {
+    spProjZm(chi,in._odata[offset]);
+  } else { 
+    chi=buf[offset];
+  }
+  Impl::multLink(Uchi,U._odata[sU],chi,Zp);
+  accumReconZm(result,Uchi);
+    
+  // Tm
+  offset = st._offsets [Tp][ss];
+  local  = st._is_local[Tp][ss];
+  perm   = st._permute[Tp][ss];
+  ptype  = st._permute_type[Tp];
+  if ( local && perm ) {
+    spProjTm(tmp,in._odata[offset]);
+    permute(chi,tmp,ptype);
+  } else if ( local ) {
+    spProjTm(chi,in._odata[offset]);
+  } else { 
+    chi=buf[offset];
+  }
+  Impl::multLink(Uchi,U._odata[sU],chi,Tp);
+  accumReconTm(result,Uchi);
+  
+  vstream(out._odata[ss],result*(-0.5));
+}
+
+template<class Impl> 
+void WilsonKernels<Impl>::DiracOptDhopDir(CartesianStencil &st,DoubledGaugeField &U,
+					  std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> >  &buf,
+					  int sF,int sU,const FermionField &in, FermionField &out,int dir,int gamma)
+{
+  SiteHalfSpinor  tmp;    
+  SiteHalfSpinor  chi;    
+  SiteSpinor   result;
+  SiteHalfSpinor Uchi;
+  int offset,local,perm, ptype;
+  int ss=sF;
+
+  offset = st._offsets [dir][ss];
+  local  = st._is_local[dir][ss];
+  perm   = st._permute[dir][ss];
+  ptype  = st._permute_type[dir];
+
+  // Xp
+  if(gamma==Xp){
     if ( local && perm ) {
       spProjXp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
@@ -27,14 +321,12 @@ void DiracOptDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[sU](Xp),&chi());
+    Impl::multLink(Uchi,U._odata[sU],chi,dir);
     spReconXp(result,Uchi);
+  }
 
-    // Yp
-    offset = st._offsets [Yp][ss];
-    local  = st._is_local[Yp][ss];
-    perm   = st._permute[Yp][ss];
-    ptype  = st._permute_type[Yp];
+  // Yp
+  if ( gamma==Yp ){
     if ( local && perm ) {
       spProjYp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
@@ -43,14 +335,12 @@ void DiracOptDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[sU](Yp),&chi());
-    accumReconYp(result,Uchi);
-
-    // Zp
-    offset = st._offsets [Zp][ss];
-    local  = st._is_local[Zp][ss];
-    perm   = st._permute[Zp][ss];
-    ptype  = st._permute_type[Zp];
+    Impl::multLink(Uchi,U._odata[sU],chi,dir);
+    spReconYp(result,Uchi);
+  }
+  
+  // Zp
+  if ( gamma ==Zp ){
     if ( local && perm ) {
       spProjZp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
@@ -59,14 +349,12 @@ void DiracOptDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[sU](Zp),&chi());
-    accumReconZp(result,Uchi);
-
-    // Tp
-    offset = st._offsets [Tp][ss];
-    local  = st._is_local[Tp][ss];
-    perm   = st._permute[Tp][ss];
-    ptype  = st._permute_type[Tp];
+    Impl::multLink(Uchi,U._odata[sU],chi,dir);
+    spReconZp(result,Uchi);
+  }
+  
+  // Tp
+  if ( gamma ==Tp ){
     if ( local && perm ) {
       spProjTp(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
@@ -75,15 +363,12 @@ void DiracOptDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[sU](Tp),&chi());
-    accumReconTp(result,Uchi);
-
-    // Xm
-    offset = st._offsets [Xm][ss];
-    local  = st._is_local[Xm][ss];
-    perm   = st._permute[Xm][ss];
-    ptype  = st._permute_type[Xm];
+    Impl::multLink(Uchi,U._odata[sU],chi,dir);
+    spReconTp(result,Uchi);
+  }
 
+  // Xm
+  if ( gamma==Xm ){
     if ( local && perm ) {
       spProjXm(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
@@ -92,15 +377,12 @@ void DiracOptDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[sU](Xm),&chi());
-    accumReconXm(result,Uchi);
-
-    // Ym
-    offset = st._offsets [Ym][ss];
-    local  = st._is_local[Ym][ss];
-    perm   = st._permute[Ym][ss];
-    ptype  = st._permute_type[Ym];
+    Impl::multLink(Uchi,U._odata[sU],chi,dir);
+    spReconXm(result,Uchi);
+  }
 
+  // Ym
+  if ( gamma == Ym ){
     if ( local && perm ) {
       spProjYm(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
@@ -109,14 +391,12 @@ void DiracOptDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[sU](Ym),&chi());
-    accumReconYm(result,Uchi);
+    Impl::multLink(Uchi,U._odata[sU],chi,dir);
+    spReconYm(result,Uchi);
+  }
 
-    // Zm
-    offset = st._offsets [Zm][ss];
-    local  = st._is_local[Zm][ss];
-    perm   = st._permute[Zm][ss];
-    ptype  = st._permute_type[Zm];
+  // Zm
+  if ( gamma == Zm ){
     if ( local && perm ) {
       spProjZm(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
@@ -125,14 +405,12 @@ void DiracOptDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[sU](Zm),&chi());
-    accumReconZm(result,Uchi);
-
-    // Tm
-    offset = st._offsets [Tm][ss];
-    local  = st._is_local[Tm][ss];
-    perm   = st._permute[Tm][ss];
-    ptype  = st._permute_type[Tm];
+    Impl::multLink(Uchi,U._odata[sU],chi,dir);
+    spReconZm(result,Uchi);
+  }
+  
+  // Tm
+  if ( gamma==Tm ) {
     if ( local && perm ) {
       spProjTm(tmp,in._odata[offset]);
       permute(chi,tmp,ptype);
@@ -141,287 +419,13 @@ void DiracOptDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     } else { 
       chi=buf[offset];
     }
-    mult(&Uchi(),&U._odata[sU](Tm),&chi());
-    accumReconTm(result,Uchi);
+    Impl::multLink(Uchi,U._odata[sU],chi,dir);
+    spReconTm(result,Uchi);
+  }
 
-    vstream(out._odata[ss],result*(-0.5));
+  vstream(out._odata[ss],result*(-0.5));
 }
 
-void DiracOptDhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
-			   std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			   int sF,int sU,const LatticeFermion &in, LatticeFermion &out)
-{
-    vHalfSpinColourVector  tmp;    
-    vHalfSpinColourVector  chi;    
-    vSpinColourVector result;
-    vHalfSpinColourVector Uchi;
-    int offset,local,perm, ptype;
-
-    // Xp
-    int ss=sF;
-    offset = st._offsets [Xm][ss];
-    local  = st._is_local[Xm][ss];
-    perm   = st._permute[Xm][ss];
-
-    ptype  = st._permute_type[Xm];
-    if ( local && perm ) {
-      spProjXp(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjXp(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[sU](Xm),&chi());
-    spReconXp(result,Uchi);
-
-    // Yp
-    offset = st._offsets [Ym][ss];
-    local  = st._is_local[Ym][ss];
-    perm   = st._permute[Ym][ss];
-    ptype  = st._permute_type[Ym];
-    if ( local && perm ) {
-      spProjYp(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjYp(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[sU](Ym),&chi());
-    accumReconYp(result,Uchi);
-
-    // Zp
-    offset = st._offsets [Zm][ss];
-    local  = st._is_local[Zm][ss];
-    perm   = st._permute[Zm][ss];
-    ptype  = st._permute_type[Zm];
-    if ( local && perm ) {
-      spProjZp(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjZp(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[sU](Zm),&chi());
-    accumReconZp(result,Uchi);
-
-    // Tp
-    offset = st._offsets [Tm][ss];
-    local  = st._is_local[Tm][ss];
-    perm   = st._permute[Tm][ss];
-    ptype  = st._permute_type[Tm];
-    if ( local && perm ) {
-      spProjTp(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjTp(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[sU](Tm),&chi());
-    accumReconTp(result,Uchi);
-
-    // Xm
-    offset = st._offsets [Xp][ss];
-    local  = st._is_local[Xp][ss];
-    perm   = st._permute[Xp][ss];
-    ptype  = st._permute_type[Xp];
-
-    if ( local && perm ) 
-    {
-      spProjXm(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjXm(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[sU](Xp),&chi());
-    accumReconXm(result,Uchi);
-
-    // Ym
-    offset = st._offsets [Yp][ss];
-    local  = st._is_local[Yp][ss];
-    perm   = st._permute[Yp][ss];
-    ptype  = st._permute_type[Yp];
-
-    if ( local && perm ) {
-      spProjYm(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjYm(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[sU](Yp),&chi());
-    accumReconYm(result,Uchi);
-
-    // Zm
-    offset = st._offsets [Zp][ss];
-    local  = st._is_local[Zp][ss];
-    perm   = st._permute[Zp][ss];
-    ptype  = st._permute_type[Zp];
-    if ( local && perm ) {
-      spProjZm(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjZm(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[sU](Zp),&chi());
-    accumReconZm(result,Uchi);
-
-    // Tm
-    offset = st._offsets [Tp][ss];
-    local  = st._is_local[Tp][ss];
-    perm   = st._permute[Tp][ss];
-    ptype  = st._permute_type[Tp];
-    if ( local && perm ) {
-      spProjTm(tmp,in._odata[offset]);
-      permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjTm(chi,in._odata[offset]);
-    } else { 
-      chi=buf[offset];
-    }
-    mult(&Uchi(),&U._odata[sU](Tp),&chi());
-    accumReconTm(result,Uchi);
-
-    vstream(out._odata[ss],result*(-0.5));
-}
-
-void DiracOptDhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
-		     std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-		     int sF,int sU,const LatticeFermion &in, LatticeFermion &out,int dir,int gamma)
-{
-    vHalfSpinColourVector  tmp;    
-    vHalfSpinColourVector  chi;    
-    vSpinColourVector result;
-    vHalfSpinColourVector Uchi;
-    int offset,local,perm, ptype;
-    int ss=sF;
-
-    offset = st._offsets [dir][ss];
-    local  = st._is_local[dir][ss];
-    perm   = st._permute[dir][ss];
-    ptype  = st._permute_type[dir];
-
-    // Xp
-    if(gamma==Xp){
-      if ( local && perm ) {
-	spProjXp(tmp,in._odata[offset]);
-	permute(chi,tmp,ptype);
-      } else if ( local ) {
-	spProjXp(chi,in._odata[offset]);
-      } else { 
-	chi=buf[offset];
-      }
-      mult(&Uchi(),&U._odata[sU](dir),&chi());
-      spReconXp(result,Uchi);
-    }
-
-    // Yp
-    if ( gamma==Yp ){
-      if ( local && perm ) {
-	spProjYp(tmp,in._odata[offset]);
-	permute(chi,tmp,ptype);
-      } else if ( local ) {
-	spProjYp(chi,in._odata[offset]);
-      } else { 
-	chi=buf[offset];
-      }
-      mult(&Uchi(),&U._odata[sU](dir),&chi());
-      spReconYp(result,Uchi);
-    }
-
-    // Zp
-    if ( gamma ==Zp ){
-      if ( local && perm ) {
-	spProjZp(tmp,in._odata[offset]);
-	permute(chi,tmp,ptype);
-      } else if ( local ) {
-	spProjZp(chi,in._odata[offset]);
-      } else { 
-	chi=buf[offset];
-      }
-      mult(&Uchi(),&U._odata[sU](dir),&chi());
-      spReconZp(result,Uchi);
-    }
-
-    // Tp
-    if ( gamma ==Tp ){
-      if ( local && perm ) {
-	spProjTp(tmp,in._odata[offset]);
-	permute(chi,tmp,ptype);
-      } else if ( local ) {
-	spProjTp(chi,in._odata[offset]);
-      } else { 
-	chi=buf[offset];
-      }
-      mult(&Uchi(),&U._odata[sU](dir),&chi());
-      spReconTp(result,Uchi);
-    }
-
-    // Xm
-    if ( gamma==Xm ){
-      if ( local && perm ) {
-	spProjXm(tmp,in._odata[offset]);
-	permute(chi,tmp,ptype);
-      } else if ( local ) {
-	spProjXm(chi,in._odata[offset]);
-      } else { 
-	chi=buf[offset];
-      }
-      mult(&Uchi(),&U._odata[sU](dir),&chi());
-      spReconXm(result,Uchi);
-    }
-
-    // Ym
-    if ( gamma == Ym ){
-      if ( local && perm ) {
-	spProjYm(tmp,in._odata[offset]);
-	permute(chi,tmp,ptype);
-      } else if ( local ) {
-	spProjYm(chi,in._odata[offset]);
-      } else { 
-	chi=buf[offset];
-      }
-      mult(&Uchi(),&U._odata[sU](dir),&chi());
-      spReconYm(result,Uchi);
-    }
-
-    // Zm
-    if ( gamma == Zm ){
-      if ( local && perm ) {
-	spProjZm(tmp,in._odata[offset]);
-	permute(chi,tmp,ptype);
-      } else if ( local ) {
-	spProjZm(chi,in._odata[offset]);
-      } else { 
-	chi=buf[offset];
-      }
-      mult(&Uchi(),&U._odata[sU](dir),&chi());
-      spReconZm(result,Uchi);
-    }
-
-    // Tm
-    if ( gamma==Tm ) {
-      if ( local && perm ) {
-	spProjTm(tmp,in._odata[offset]);
-	permute(chi,tmp,ptype);
-      } else if ( local ) {
-	spProjTm(chi,in._odata[offset]);
-      } else { 
-	chi=buf[offset];
-      }
-      mult(&Uchi(),&U._odata[sU](dir),&chi());
-      spReconTm(result,Uchi);
-    }
-
-    vstream(out._odata[ss],result*(-0.5));
-}
+  FermOpTemplateInstantiate(WilsonKernels);
 
 }}
diff --git a/lib/qcd/action/fermion/WilsonKernels.h b/lib/qcd/action/fermion/WilsonKernels.h
index cb0e07bb..eb0981fc 100644
--- a/lib/qcd/action/fermion/WilsonKernels.h
+++ b/lib/qcd/action/fermion/WilsonKernels.h
@@ -6,44 +6,46 @@ namespace Grid {
   namespace QCD {
 
     ////////////////////////////////////////////////////////////////////////////////////////////////////////////////
-    // Helper classes that implement Wilson stencil for a single site.
+    // Helper routines that implement Wilson stencil for a single site.
     ////////////////////////////////////////////////////////////////////////////////////////////////////////////////
-    
-    // Generic version works for any Nc and with extra flavour indices
-    //    namespace DiracOpt {
 
-      // These ones will need to be package intelligently. WilsonType base class
-      // for use by DWF etc..
-      void DiracOptDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
-			    std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			    int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
-      void DiracOptDhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
-			       std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			       int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
-      void DiracOptDhopDir(CartesianStencil &st,LatticeDoubledGaugeField &U,
-			   std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			   int sF,int sU,const LatticeFermion &in, LatticeFermion &out,int dirdisp,int gamma);
+    template<class Impl> 
+    class WilsonKernels { 
+    public:
+
+#include <qcd/action/fermion/FermionImplTypedefs.h>
+
+    public:
+      static void 
+	DiracOptDhopSite(CartesianStencil &st,DoubledGaugeField &U,
+			 std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> >  &buf,
+			 int sF,int sU,const FermionField &in, FermionField &out);
       
-      //  };
+      static void 
+	DiracOptDhopSiteDag(CartesianStencil &st,DoubledGaugeField &U,
+			    std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> >  &buf,
+			    int sF,int sU,const FermionField &in,FermionField &out);
 
-      // Hand unrolled for Nc=3, one flavour
-      //    namespace DiracOptHand {
-      // These ones will need to be package intelligently. WilsonType base class
-      // for use by DWF etc..
+      static void 
+	DiracOptDhopDir(CartesianStencil &st,DoubledGaugeField &U,
+			std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> >  &buf,
+			int sF,int sU,const FermionField &in, FermionField &out,int dirdisp,int gamma);
 
-      void DiracOptHandDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
-				std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-				int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
-      void DiracOptHandDhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
-				   std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-				   int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
+      static void 
+	DiracOptHandDhopSite(CartesianStencil &st,DoubledGaugeField &U,
+			     std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> >  &buf,
+			     int sF,int sU,const FermionField &in, FermionField &out){
+	DiracOptDhopSite(st,U,buf,sF,sU,in,out); // will template override for Wilson Nc=3
+      }
 
-      //    };
-  
+      static void 
+	DiracOptHandDhopSiteDag(CartesianStencil &st,DoubledGaugeField &U,
+				std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> >  &buf,
+				int sF,int sU,const FermionField &in, FermionField &out){
+	DiracOptDhopSiteDag(st,U,buf,sF,sU,in,out); // will template override for Wilson Nc=3
+      }
 
-    void DiracOptHandDhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
-				 std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-				 int sF,int sU,const LatticeFermion &in, LatticeFermion &out);
+    };
 
   }
 }
diff --git a/lib/qcd/action/fermion/WilsonKernelsHand.cc b/lib/qcd/action/fermion/WilsonKernelsHand.cc
index cff700fb..ea9d5457 100644
--- a/lib/qcd/action/fermion/WilsonKernelsHand.cc
+++ b/lib/qcd/action/fermion/WilsonKernelsHand.cc
@@ -280,48 +280,50 @@
 namespace Grid {
 namespace QCD {
 
-void DiracOptHandDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
-			    std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			    int sF,int sU,const LatticeFermion &in, LatticeFermion &out)
+#if 0
+template<class Simd>
+void WilsonKernels<WilsonImpl<Simd,3> >::DiracOptHandDhopSite(CartesianStencil &st,DoubledGaugeField &U,
+					       std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> >  &buf,
+					       int sF,int sU,const FermionField &in, FermionField &out)
 {
-  REGISTER vComplex result_00; // 12 regs on knc
-  REGISTER vComplex result_01;
-  REGISTER vComplex result_02;
+  REGISTER Simd result_00; // 12 regs on knc
+  REGISTER Simd result_01;
+  REGISTER Simd result_02;
   
-  REGISTER vComplex result_10;
-  REGISTER vComplex result_11;
-  REGISTER vComplex result_12;
+  REGISTER Simd result_10;
+  REGISTER Simd result_11;
+  REGISTER Simd result_12;
 
-  REGISTER vComplex result_20;
-  REGISTER vComplex result_21;
-  REGISTER vComplex result_22;
+  REGISTER Simd result_20;
+  REGISTER Simd result_21;
+  REGISTER Simd result_22;
 
-  REGISTER vComplex result_30;
-  REGISTER vComplex result_31;
-  REGISTER vComplex result_32; // 20 left
+  REGISTER Simd result_30;
+  REGISTER Simd result_31;
+  REGISTER Simd result_32; // 20 left
 
-  REGISTER vComplex Chi_00;    // two spinor; 6 regs
-  REGISTER vComplex Chi_01;
-  REGISTER vComplex Chi_02;
+  REGISTER Simd Chi_00;    // two spinor; 6 regs
+  REGISTER Simd Chi_01;
+  REGISTER Simd Chi_02;
 
-  REGISTER vComplex Chi_10;
-  REGISTER vComplex Chi_11;
-  REGISTER vComplex Chi_12;   // 14 left
+  REGISTER Simd Chi_10;
+  REGISTER Simd Chi_11;
+  REGISTER Simd Chi_12;   // 14 left
 
-  REGISTER vComplex UChi_00;  // two spinor; 6 regs
-  REGISTER vComplex UChi_01;
-  REGISTER vComplex UChi_02;
+  REGISTER Simd UChi_00;  // two spinor; 6 regs
+  REGISTER Simd UChi_01;
+  REGISTER Simd UChi_02;
 
-  REGISTER vComplex UChi_10;
-  REGISTER vComplex UChi_11;
-  REGISTER vComplex UChi_12;  // 8 left
+  REGISTER Simd UChi_10;
+  REGISTER Simd UChi_11;
+  REGISTER Simd UChi_12;  // 8 left
 
-  REGISTER vComplex U_00;  // two rows of U matrix
-  REGISTER vComplex U_10;
-  REGISTER vComplex U_20;  
-  REGISTER vComplex U_01;
-  REGISTER vComplex U_11;
-  REGISTER vComplex U_21;  // 2 reg left.
+  REGISTER Simd U_00;  // two rows of U matrix
+  REGISTER Simd U_10;
+  REGISTER Simd U_20;  
+  REGISTER Simd U_01;
+  REGISTER Simd U_11;
+  REGISTER Simd U_21;  // 2 reg left.
 
 #define Chimu_00 Chi_00
 #define Chimu_01 Chi_01
@@ -360,11 +362,6 @@ void DiracOptHandDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     MULT_2SPIN(Xp);
   }
   XP_RECON;
-  //  std::cout<<GridLogMessage << "XP_RECON"<<std::endl;
-  //  std::cout<<GridLogMessage << result_00 <<" "<<result_01 <<" "<<result_02 <<std::endl;
-  //  std::cout<<GridLogMessage << result_10 <<" "<<result_11 <<" "<<result_12 <<std::endl;
-  //  std::cout<<GridLogMessage << result_20 <<" "<<result_21 <<" "<<result_22 <<std::endl;
-  //  std::cout<<GridLogMessage << result_30 <<" "<<result_31 <<" "<<result_32 <<std::endl;
 
   // Yp
   offset = st._offsets [Yp][ss];
@@ -446,12 +443,6 @@ void DiracOptHandDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
     MULT_2SPIN(Xm);
   }
   XM_RECON_ACCUM;
-  //  std::cout<<GridLogMessage << "XM_RECON_ACCUM"<<std::endl;
-  //  std::cout<<GridLogMessage << result_00 <<" "<<result_01 <<" "<<result_02 <<std::endl;
-  //  std::cout<<GridLogMessage << result_10 <<" "<<result_11 <<" "<<result_12 <<std::endl;
-  //  std::cout<<GridLogMessage << result_20 <<" "<<result_21 <<" "<<result_22 <<std::endl;
-  //  std::cout<<GridLogMessage << result_30 <<" "<<result_31 <<" "<<result_32 <<std::endl;
-  
   
   // Ym
   offset = st._offsets [Ym][ss];
@@ -530,48 +521,49 @@ void DiracOptHandDhopSite(CartesianStencil &st,LatticeDoubledGaugeField &U,
   }
 }
 
-void DiracOptHandDhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
-			       std::vector<vHalfSpinColourVector,alignedAllocator<vHalfSpinColourVector> >  &buf,
-			       int ss,int sU,const LatticeFermion &in, LatticeFermion &out)
+template<class Simd>
+void WilsonKernels<WilsonImpl<Simd,3> >::DiracOptHandDhopSiteDag(CartesianStencil &st,DoubledGaugeField &U,
+							      std::vector<SiteHalfSpinor,alignedAllocator<SiteHalfSpinor> >  &buf,
+							      int ss,int sU,const FermionField &in, FermionField &out)
 {
-  REGISTER vComplex result_00; // 12 regs on knc
-  REGISTER vComplex result_01;
-  REGISTER vComplex result_02;
+  REGISTER Simd result_00; // 12 regs on knc
+  REGISTER Simd result_01;
+  REGISTER Simd result_02;
 
-  REGISTER vComplex result_10;
-  REGISTER vComplex result_11;
-  REGISTER vComplex result_12;
+  REGISTER Simd result_10;
+  REGISTER Simd result_11;
+  REGISTER Simd result_12;
 
-  REGISTER vComplex result_20;
-  REGISTER vComplex result_21;
-  REGISTER vComplex result_22;
+  REGISTER Simd result_20;
+  REGISTER Simd result_21;
+  REGISTER Simd result_22;
 
-  REGISTER vComplex result_30;
-  REGISTER vComplex result_31;
-  REGISTER vComplex result_32; // 20 left
+  REGISTER Simd result_30;
+  REGISTER Simd result_31;
+  REGISTER Simd result_32; // 20 left
 
-  REGISTER vComplex Chi_00;    // two spinor; 6 regs
-  REGISTER vComplex Chi_01;
-  REGISTER vComplex Chi_02;
+  REGISTER Simd Chi_00;    // two spinor; 6 regs
+  REGISTER Simd Chi_01;
+  REGISTER Simd Chi_02;
 
-  REGISTER vComplex Chi_10;
-  REGISTER vComplex Chi_11;
-  REGISTER vComplex Chi_12;   // 14 left
+  REGISTER Simd Chi_10;
+  REGISTER Simd Chi_11;
+  REGISTER Simd Chi_12;   // 14 left
 
-  REGISTER vComplex UChi_00;  // two spinor; 6 regs
-  REGISTER vComplex UChi_01;
-  REGISTER vComplex UChi_02;
+  REGISTER Simd UChi_00;  // two spinor; 6 regs
+  REGISTER Simd UChi_01;
+  REGISTER Simd UChi_02;
 
-  REGISTER vComplex UChi_10;
-  REGISTER vComplex UChi_11;
-  REGISTER vComplex UChi_12;  // 8 left
+  REGISTER Simd UChi_10;
+  REGISTER Simd UChi_11;
+  REGISTER Simd UChi_12;  // 8 left
 
-  REGISTER vComplex U_00;  // two rows of U matrix
-  REGISTER vComplex U_10;
-  REGISTER vComplex U_20;  
-  REGISTER vComplex U_01;
-  REGISTER vComplex U_11;
-  REGISTER vComplex U_21;  // 2 reg left.
+  REGISTER Simd U_00;  // two rows of U matrix
+  REGISTER Simd U_10;
+  REGISTER Simd U_20;  
+  REGISTER Simd U_01;
+  REGISTER Simd U_11;
+  REGISTER Simd U_21;  // 2 reg left.
 
 #define Chimu_00 Chi_00
 #define Chimu_01 Chi_01
@@ -752,7 +744,7 @@ void DiracOptHandDhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
   TP_RECON_ACCUM;
 
   {
-    vSpinColourVector & ref (out._odata[ss]);
+    SiteSpinor & ref (out._odata[ss]);
     vstream(ref()(0)(0),result_00*(-0.5));
     vstream(ref()(0)(1),result_01*(-0.5));
     vstream(ref()(0)(2),result_02*(-0.5));
@@ -767,4 +759,5 @@ void DiracOptHandDhopSiteDag(CartesianStencil &st,LatticeDoubledGaugeField &U,
     vstream(ref()(3)(2),result_32*(-0.5));
   }
 }
+#endif
 }}
diff --git a/lib/qcd/action/pseudofermion/TwoFlavour.h b/lib/qcd/action/pseudofermion/TwoFlavour.h
index 22f761fe..635c1ca6 100644
--- a/lib/qcd/action/pseudofermion/TwoFlavour.h
+++ b/lib/qcd/action/pseudofermion/TwoFlavour.h
@@ -95,12 +95,13 @@ namespace Grid{
     ////////////////////////////////////////////////////////////////////////
     // Two flavour pseudofermion action for any dop
     ////////////////////////////////////////////////////////////////////////
-    template<class GaugeField,class MatrixField,class FermionField>
-      class TwoFlavourPseudoFermionAction : public Action<GaugeField> {
+    template<class Impl>
+    class TwoFlavourPseudoFermionAction : public Action<typename Impl::GaugeField> {
 
     private:
+#include <qcd/action/fermion/FermionImplTypedefs.h>
       
-      FermionOperator<FermionField,GaugeField> & FermOp;// the basic operator
+      FermionOperator<Impl> & FermOp;// the basic operator
 
       OperatorFunction<FermionField> &DerivativeSolver;
 
@@ -112,7 +113,7 @@ namespace Grid{
       /////////////////////////////////////////////////
       // Pass in required objects.
       /////////////////////////////////////////////////
-    TwoFlavourPseudoFermionAction(FermionOperator<FermionField,GaugeField>  &Op, 
+    TwoFlavourPseudoFermionAction(FermionOperator<Impl>  &Op, 
 				  OperatorFunction<FermionField> & DS,
 				  OperatorFunction<FermionField> & AS
 				  ) : FermOp(Op), DerivativeSolver(DS), ActionSolver(AS), Phi(Op.FermionGrid()) {
@@ -158,7 +159,7 @@ namespace Grid{
 	FermionField X(FermOp.FermionGrid());
 	FermionField Y(FermOp.FermionGrid());
 	
-	MdagMLinearOperator<FermionOperator<FermionField,GaugeField> ,FermionField> MdagMOp(FermOp);
+	MdagMLinearOperator<FermionOperator<Impl> ,FermionField> MdagMOp(FermOp);
 	X=zero;
 	ActionSolver(MdagMOp,Phi,X);
 	MdagMOp.Op(X,Y);
@@ -183,7 +184,7 @@ namespace Grid{
 	FermionField Y(FermOp.FermionGrid());
 	GaugeField   tmp(FermOp.GaugeGrid());
 
-	MdagMLinearOperator<FermionOperator<FermionField,GaugeField> ,FermionField> MdagMOp(FermOp);
+	MdagMLinearOperator<FermionOperator<Impl> ,FermionField> MdagMOp(FermOp);
 
 	X=zero;
 	DerivativeSolver(MdagMOp,Phi,X);
diff --git a/lib/qcd/action/pseudofermion/TwoFlavourEvenOdd.h b/lib/qcd/action/pseudofermion/TwoFlavourEvenOdd.h
index 63dec41a..35a208c8 100644
--- a/lib/qcd/action/pseudofermion/TwoFlavourEvenOdd.h
+++ b/lib/qcd/action/pseudofermion/TwoFlavourEvenOdd.h
@@ -4,13 +4,18 @@
 namespace Grid{
   namespace QCD{
 
-    template<class Matrix,class FermionField>
-      class SchurDifferentiableOperator :  public SchurDiagMooeeOperator<Matrix,FermionField> 
+    template<class Impl>
+    class SchurDifferentiableOperator :  public SchurDiagMooeeOperator<FermionOperator<Impl>,typename Impl::FermionField> 
       {
+      public:
+#include <qcd/action/fermion/FermionImplTypedefs.h>
+	
     public:
+	typedef FermionOperator<Impl> Matrix;
+
 	SchurDifferentiableOperator (Matrix &Mat) : SchurDiagMooeeOperator<Matrix,FermionField>(Mat) {};
 
-	void MpcDeriv(LatticeGaugeField &Force,const FermionField &U,const FermionField &V) {
+	void MpcDeriv(GaugeField &Force,const FermionField &U,const FermionField &V) {
 	
 	  GridBase *fgrid   = this->_Mat.FermionGrid();
 	  GridBase *fcbgrid = this->_Mat.FermionRedBlackGrid();
@@ -28,8 +33,8 @@ namespace Grid{
 	  assert(U.checkerboard==Odd);
 	  assert(V.checkerboard==V.checkerboard);
 
-	  LatticeGaugeField ForceO(ucbgrid);
-	  LatticeGaugeField ForceE(ucbgrid);
+	  GaugeField ForceO(ucbgrid);
+	  GaugeField ForceE(ucbgrid);
 
 	  //  X^dag Der_oe MeeInv Meo Y
 	  // Use Mooee as nontrivial but gauge field indept
@@ -48,7 +53,7 @@ namespace Grid{
 	}
 
 
-	void MpcDagDeriv(LatticeGaugeField &Force,const FermionField &U,const FermionField &V) {
+	void MpcDagDeriv(GaugeField &Force,const FermionField &U,const FermionField &V) {
 	
 	  GridBase *fgrid   = this->_Mat.FermionGrid();
 	  GridBase *fcbgrid = this->_Mat.FermionRedBlackGrid();
@@ -66,8 +71,8 @@ namespace Grid{
 	  assert(V.checkerboard==Odd);
 	  assert(V.checkerboard==V.checkerboard);
 
-	  LatticeGaugeField ForceO(ucbgrid);
-	  LatticeGaugeField ForceE(ucbgrid);
+	  GaugeField ForceO(ucbgrid);
+	  GaugeField ForceE(ucbgrid);
 
 	  //  X^dag Der_oe MeeInv Meo Y
 	  // Use Mooee as nontrivial but gauge field indept
@@ -91,12 +96,15 @@ namespace Grid{
     ////////////////////////////////////////////////////////////////////////
     // Two flavour pseudofermion action for any EO prec dop
     ////////////////////////////////////////////////////////////////////////
-    template<class GaugeField,class MatrixField,class FermionField>
-      class TwoFlavourEvenOddPseudoFermionAction : public Action<GaugeField> {
+    template<class Impl>
+    class TwoFlavourEvenOddPseudoFermionAction : public Action<typename Impl::GaugeField> {
+
+    public:
+#include <qcd/action/fermion/FermionImplTypedefs.h>
 
     private:
       
-      FermionOperator<FermionField,GaugeField> & FermOp;// the basic operator
+      FermionOperator<Impl> & FermOp;// the basic operator
 
       OperatorFunction<FermionField> &DerivativeSolver;
 
@@ -109,7 +117,7 @@ namespace Grid{
       /////////////////////////////////////////////////
       // Pass in required objects.
       /////////////////////////////////////////////////
-      TwoFlavourEvenOddPseudoFermionAction(FermionOperator<FermionField,GaugeField>  &Op, 
+      TwoFlavourEvenOddPseudoFermionAction(FermionOperator<Impl>  &Op, 
 					 OperatorFunction<FermionField> & DS,
 					 OperatorFunction<FermionField> & AS
 					   ) : 
@@ -140,7 +148,7 @@ namespace Grid{
 	pickCheckerboard(Even,etaEven,eta);
 	pickCheckerboard(Odd,etaOdd,eta);
 
-	SchurDifferentiableOperator<FermionOperator<FermionField,GaugeField>,FermionField> PCop(FermOp);
+	SchurDifferentiableOperator<Impl> PCop(FermOp);
 	
 	FermOp.ImportGauge(U);
 
@@ -163,7 +171,7 @@ namespace Grid{
 	FermionField X(FermOp.FermionRedBlackGrid());
 	FermionField Y(FermOp.FermionRedBlackGrid());
 	
-	SchurDifferentiableOperator<FermionOperator<FermionField,GaugeField>,FermionField> PCop(FermOp);
+	SchurDifferentiableOperator<Impl> PCop(FermOp);
 
 	X=zero;
 	ActionSolver(PCop,PhiOdd,X);
@@ -195,7 +203,7 @@ namespace Grid{
 	FermionField Y(FermOp.FermionRedBlackGrid());
 	GaugeField tmp(FermOp.GaugeGrid());
 
-	SchurDifferentiableOperator<FermionOperator<FermionField,GaugeField>,FermionField> PCop(FermOp);
+	SchurDifferentiableOperator<Impl> PCop(FermOp);
 
 	X=zero;
 	DerivativeSolver(PCop,PhiOdd,X);
diff --git a/lib/qcd/spin/Dirac.h b/lib/qcd/spin/Dirac.h
index 3208d0b5..65d8d6b5 100644
--- a/lib/qcd/spin/Dirac.h
+++ b/lib/qcd/spin/Dirac.h
@@ -4,7 +4,6 @@ namespace Grid{
 
 namespace QCD {
 
-  const int SpinorIndex = 2;
 
   class Gamma {
 
diff --git a/lib/qcd/spin/TwoSpinor.h b/lib/qcd/spin/TwoSpinor.h
index 919f0f76..94ba9978 100644
--- a/lib/qcd/spin/TwoSpinor.h
+++ b/lib/qcd/spin/TwoSpinor.h
@@ -42,18 +42,14 @@ namespace QCD {
    * -i 0  0  0
    */
 
-
-  template<class vtype> strong_inline void 
-    spProjXp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+  // To fail is not to err (Cryptic clue: suggest to Google SFINAE ;) )
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProjXp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      // To fail is not to err (Cryptic clue: suggest to Google SFINAE ;) )
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)+timesI(fspin(3));
       hspin(1)=fspin(1)+timesI(fspin(2));
     }
-  template<class vtype> strong_inline void spProjXm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProjXm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)-timesI(fspin(3));
       hspin(1)=fspin(1)-timesI(fspin(2));
     }
@@ -62,15 +58,13 @@ namespace QCD {
       //  0 0  1  0   [1] +- [2]
       //  0 1  0  0
       // -1 0  0  0
-  template<class vtype> strong_inline void spProjYp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProjYp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)-fspin(3);
       hspin(1)=fspin(1)+fspin(2);
     }
-  template<class vtype> strong_inline void spProjYm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProjYm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)+fspin(3);
       hspin(1)=fspin(1)-fspin(2);
     }
@@ -80,15 +74,14 @@ namespace QCD {
 	     * -i 0  0  0
 	     *  0 i  0  0
 	     */
-  template<class vtype> strong_inline void spProjZp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProjZp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)+timesI(fspin(2));
       hspin(1)=fspin(1)-timesI(fspin(3));
     }
-  template<class vtype> strong_inline void spProjZm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProjZm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)-timesI(fspin(2));
       hspin(1)=fspin(1)+timesI(fspin(3));
     }
@@ -98,15 +91,15 @@ namespace QCD {
 	     *  1 0  0  0
 	     *  0 1  0  0
 	     */
-  template<class vtype> strong_inline void spProjTp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProjTp (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)+fspin(2);
       hspin(1)=fspin(1)+fspin(3);
     }
-  template<class vtype> strong_inline void spProjTm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProjTm (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0)-fspin(2);
       hspin(1)=fspin(1)-fspin(3);
     }
@@ -117,32 +110,33 @@ namespace QCD {
 	     *  0 0  0 -1
 	     */
 
-  template<class vtype> strong_inline void spProj5p (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProj5p (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(0);
       hspin(1)=fspin(1);
     }
-  template<class vtype> strong_inline void spProj5m (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
+
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProj5m (iVector<vtype,Nhs> &hspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       hspin(0)=fspin(2);
       hspin(1)=fspin(3);
     }
   
   //  template<class vtype> strong_inline void fspProj5p (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
-  template<class vtype> strong_inline void spProj5p (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProj5p (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       rfspin(0)=fspin(0);
       rfspin(1)=fspin(1);
       rfspin(2)=zero;
       rfspin(3)=zero;
     }
   //  template<class vtype> strong_inline void fspProj5m (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
-  template<class vtype> strong_inline void spProj5m (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spProj5m (iVector<vtype,Ns> &rfspin,const iVector<vtype,Ns> &fspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       rfspin(0)=zero;
       rfspin(1)=zero;
       rfspin(2)=fspin(2);
@@ -158,33 +152,33 @@ namespace QCD {
    *  0 -i 0  0  -i[1]+-[2]   == -i ([0]+-i[3]) = -i (1)
    * -i 0  0  0  -i[0]+-[3]   == -i ([1]+-i[2]) = -i (0)
    */
-  template<class vtype> strong_inline void spReconXp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spReconXp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
       fspin(1)=hspin(1);
       fspin(2)=timesMinusI(hspin(1));
       fspin(3)=timesMinusI(hspin(0));
     }
-  template<class vtype> strong_inline void spReconXm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spReconXm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
       fspin(1)=hspin(1);
       fspin(2)=timesI(hspin(1));
       fspin(3)=timesI(hspin(0));
     }
-  template<class vtype> strong_inline void accumReconXp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void accumReconXp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
       fspin(1)+=hspin(1);
       fspin(2)-=timesI(hspin(1));
       fspin(3)-=timesI(hspin(0));
     }
-  template<class vtype> strong_inline void accumReconXm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void accumReconXm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
       fspin(1)+=hspin(1);
       fspin(2)+=timesI(hspin(1));
@@ -196,33 +190,33 @@ namespace QCD {
       //  0 1  0  0              == 1(1)
       // -1 0  0  0              ==-1(0)
 
-  template<class vtype> strong_inline void spReconYp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spReconYp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
       fspin(1)=hspin(1);
       fspin(2)= hspin(1);
       fspin(3)=-hspin(0);//Unary minus?
     }
-  template<class vtype> strong_inline void spReconYm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spReconYm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
       fspin(1)=hspin(1);
       fspin(2)=-hspin(1);
       fspin(3)= hspin(0);
     }
-  template<class vtype> strong_inline void accumReconYp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void accumReconYp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
       fspin(1)+=hspin(1);
       fspin(2)+=hspin(1);
       fspin(3)-=hspin(0);
     }
-  template<class vtype> strong_inline void accumReconYm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void accumReconYm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
       fspin(1)+=hspin(1);
       fspin(2)-=hspin(1);
@@ -235,33 +229,33 @@ namespace QCD {
 	     * -i 0  0  0     => -i (0)
 	     *  0 i  0  0     =>  i (1)
 	     */
-  template<class vtype> strong_inline void spReconZp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spReconZp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
       fspin(1)=hspin(1);
       fspin(2)=timesMinusI(hspin(0));
       fspin(3)=timesI(hspin(1));
     }
-  template<class vtype> strong_inline void spReconZm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spReconZm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
       fspin(1)=hspin(1);
       fspin(2)=     timesI(hspin(0));
       fspin(3)=timesMinusI(hspin(1));
     }
-  template<class vtype> strong_inline void accumReconZp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void accumReconZp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
       fspin(1)+=hspin(1);
       fspin(2)-=timesI(hspin(0));
       fspin(3)+=timesI(hspin(1));
     }
-  template<class vtype> strong_inline void accumReconZm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void accumReconZm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
       fspin(1)+=hspin(1);
       fspin(2)+=timesI(hspin(0));
@@ -273,33 +267,33 @@ namespace QCD {
 	     *  1 0  0  0    => (0)
 	     *  0 1  0  0    => (1)
 	     */
-  template<class vtype> strong_inline void spReconTp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spReconTp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
       fspin(1)=hspin(1);
       fspin(2)=hspin(0);
       fspin(3)=hspin(1);
     }
-  template<class vtype> strong_inline void spReconTm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spReconTm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0);
       fspin(1)=hspin(1);
       fspin(2)=-hspin(0);
       fspin(3)=-hspin(1);
     }
-  template<class vtype> strong_inline void accumReconTp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void accumReconTp (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
       fspin(1)+=hspin(1);
       fspin(2)+=hspin(0);
       fspin(3)+=hspin(1);
     }
-  template<class vtype> strong_inline void accumReconTm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void accumReconTm (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0);
       fspin(1)+=hspin(1);
       fspin(2)-=hspin(0);
@@ -311,31 +305,31 @@ namespace QCD {
 	     *  0 0 -1  0
 	     *  0 0  0 -1
 	     */
-  template<class vtype> strong_inline void spRecon5p (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spRecon5p (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=hspin(0)+hspin(0); // add is lower latency than mul
       fspin(1)=hspin(1)+hspin(1); // probably no measurable diffence though
       fspin(2)=zero;
       fspin(3)=zero;
     }
-  template<class vtype> strong_inline void spRecon5m (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void spRecon5m (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)=zero;
       fspin(1)=zero;
       fspin(2)=hspin(0)+hspin(0);
       fspin(3)=hspin(1)+hspin(1);
     }
-  template<class vtype> strong_inline void accumRecon5p (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void accumRecon5p (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(0)+=hspin(0)+hspin(0);
       fspin(1)+=hspin(1)+hspin(1);
     }
-  template<class vtype> strong_inline void accumRecon5m (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
+  template<class vtype,IfSpinor<iVector<vtype,Ns> > = 0> strong_inline void accumRecon5m (iVector<vtype,Ns> &fspin,const iVector<vtype,Nhs> &hspin)
     {
-      typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
+      //typename std::enable_if<matchGridTensorIndex<iVector<vtype,Ns>,SpinorIndex>::value,iVector<vtype,Ns> >::type *SFINAE;
       fspin(2)+=hspin(0)+hspin(0);
       fspin(3)+=hspin(1)+hspin(1);
     }
@@ -347,21 +341,19 @@ namespace QCD {
   //////////
   // Xp
   //////////
-  template<class rtype,class vtype> strong_inline void spProjXp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProjXp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
-    spProjXp(hspin._internal,fspin._internal);
-  }
-  template<class rtype,class vtype,int N> strong_inline void spProjXp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
-  {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjXp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spProjXp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype> strong_inline void spProjXp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  {
+    spProjXp(hspin._internal,fspin._internal);
+  }
+  template<class rtype,class vtype,int N> strong_inline void spProjXp (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProjXp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -371,19 +363,16 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spReconXp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconXp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spReconXp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spReconXp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconXp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spReconXp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconXp (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spReconXp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -392,19 +381,16 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void accumReconXp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconXp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void accumReconXp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumReconXp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconXp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void accumReconXp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconXp (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumReconXp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -418,19 +404,19 @@ namespace QCD {
   ////////
   template<class rtype,class vtype> strong_inline void spProjXm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjXm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spProjXm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProjXm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjXm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spProjXm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjXm (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProjXm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -440,19 +426,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spReconXm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconXm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spReconXm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spReconXm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconXm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spReconXm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconXm (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spReconXm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -461,19 +447,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void accumReconXm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconXm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void accumReconXm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumReconXm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconXm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void accumReconXm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconXm (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumReconXm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -487,19 +473,19 @@ namespace QCD {
   ////////
   template<class rtype,class vtype> strong_inline void spProjYp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjYp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spProjYp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProjYp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjYp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spProjYp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjYp (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProjYp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -509,19 +495,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spReconYp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconYp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spReconYp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spReconYp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconYp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spReconYp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconYp (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spReconYp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -530,19 +516,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void accumReconYp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconYp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void accumReconYp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumReconYp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconYp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void accumReconYp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconYp (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumReconYp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -555,19 +541,19 @@ namespace QCD {
   ////////
   template<class rtype,class vtype> strong_inline void spProjYm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjYm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spProjYm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProjYm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjYm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spProjYm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjYm (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProjYm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -577,19 +563,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spReconYm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconYm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spReconYm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spReconYm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,const iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconYm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spReconYm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconYm (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spReconYm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -598,19 +584,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void accumReconYm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconYm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void accumReconYm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumReconYm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconYm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void accumReconYm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconYm (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumReconYm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -623,19 +609,19 @@ namespace QCD {
   ////////
   template<class rtype,class vtype> strong_inline void spProjZp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjZp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spProjZp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProjZp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjZp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spProjZp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjZp (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProjZp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -645,19 +631,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spReconZp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconZp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spReconZp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spReconZp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconZp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spReconZp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconZp (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spReconZp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -666,19 +652,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void accumReconZp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconZp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void accumReconZp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumReconZp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconZp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void accumReconZp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconZp (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumReconZp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -691,19 +677,19 @@ namespace QCD {
   ////////
   template<class rtype,class vtype> strong_inline void spProjZm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjZm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spProjZm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProjZm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjZm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spProjZm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjZm (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProjZm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -713,19 +699,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spReconZm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconZm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spReconZm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spReconZm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconZm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spReconZm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconZm (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spReconZm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -734,19 +720,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void accumReconZm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconZm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void accumReconZm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumReconZm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconZm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void accumReconZm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconZm (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumReconZm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -759,19 +745,19 @@ namespace QCD {
   ////////
   template<class rtype,class vtype> strong_inline void spProjTp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjTp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spProjTp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProjTp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjTp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spProjTp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjTp (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProjTp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -781,40 +767,40 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spReconTp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconTp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spReconTp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spReconTp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconTp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spReconTp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconTp (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spReconTp(hspin._internal[i][j],fspin._internal[i][j]);
     }}
   }
 
-  template<class rtype,class vtype> strong_inline void accumReconTp (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void accumReconTp (iScalar<rtype> &hspin, iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconTp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void accumReconTp (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumReconTp (iVector<rtype,N> &hspin, const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconTp(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void accumReconTp (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconTp (iMatrix<rtype,N> &hspin, const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumReconTp(hspin._internal[i][j],fspin._internal[i][j]);
@@ -827,19 +813,19 @@ namespace QCD {
   ////////
   template<class rtype,class vtype> strong_inline void spProjTm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjTm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spProjTm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProjTm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProjTm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spProjTm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProjTm (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProjTm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -847,42 +833,42 @@ namespace QCD {
   }
 
 
-  template<class rtype,class vtype> strong_inline void spReconTm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void spReconTm (iScalar<rtype> &hspin, const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconTm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spReconTm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spReconTm (iVector<rtype,N> &hspin, const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spReconTm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spReconTm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spReconTm (iMatrix<rtype,N> &hspin, const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spReconTm(hspin._internal[i][j],fspin._internal[i][j]);
     }}
   }
 
-  template<class rtype,class vtype> strong_inline void accumReconTm (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
+  template<class rtype,class vtype> strong_inline void accumReconTm (iScalar<rtype> &hspin, const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconTm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void accumReconTm (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumReconTm (iVector<rtype,N> &hspin, const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumReconTm(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void accumReconTm (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumReconTm (iMatrix<rtype,N> &hspin, const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumReconTm(hspin._internal[i][j],fspin._internal[i][j]);
@@ -895,19 +881,19 @@ namespace QCD {
   ////////
   template<class rtype,class vtype> strong_inline void spProj5p (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProj5p(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spProj5p (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProj5p (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProj5p(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spProj5p (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProj5p (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProj5p(hspin._internal[i][j],fspin._internal[i][j]);
@@ -916,19 +902,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spRecon5p (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spRecon5p(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spRecon5p (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spRecon5p (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spRecon5p(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spRecon5p (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spRecon5p (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spRecon5p(hspin._internal[i][j],fspin._internal[i][j]);
@@ -937,19 +923,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void accumRecon5p (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumRecon5p(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void accumRecon5p (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumRecon5p (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumRecon5p(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void accumRecon5p (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumRecon5p (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumRecon5p(hspin._internal[i][j],fspin._internal[i][j]);
@@ -960,21 +946,21 @@ namespace QCD {
   //  template<class vtype> strong_inline void fspProj5p (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
   template<class vtype> strong_inline void spProj5p (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProj5p(hspin._internal,fspin._internal);
   }
   //  template<class vtype,int N> strong_inline void fspProj5p (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
-  template<class vtype,int N> strong_inline void spProj5p (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProj5p (iVector<vtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProj5p(hspin._internal[i],fspin._internal[i]);
     }
   }
   //  template<class vtype,int N> strong_inline void fspProj5p (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
-  template<class vtype,int N> strong_inline void spProj5p (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class vtype,int N> strong_inline void spProj5p (iMatrix<vtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProj5p(hspin._internal[i][j],fspin._internal[i][j]);
@@ -985,21 +971,19 @@ namespace QCD {
   ////////
   // 5m
   ////////
+
   template<class rtype,class vtype> strong_inline void spProj5m (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProj5m(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spProj5m (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<rtype> = 0> strong_inline void spProj5m (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProj5m(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spProj5m (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spProj5m (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProj5m(hspin._internal[i][j],fspin._internal[i][j]);
@@ -1008,19 +992,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void spRecon5m (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spRecon5m(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void spRecon5m (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spRecon5m (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spRecon5m(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void spRecon5m (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void spRecon5m (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spRecon5m(hspin._internal[i][j],fspin._internal[i][j]);
@@ -1029,19 +1013,19 @@ namespace QCD {
 
   template<class rtype,class vtype> strong_inline void accumRecon5m (iScalar<rtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumRecon5m(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N> strong_inline void accumRecon5m (iVector<rtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumRecon5m (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       accumRecon5m(hspin._internal[i],fspin._internal[i]);
     }
   }
-  template<class rtype,class vtype,int N> strong_inline void accumRecon5m (iMatrix<rtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class rtype,class vtype,int N> strong_inline void accumRecon5m (iMatrix<rtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       accumRecon5m(hspin._internal[i][j],fspin._internal[i][j]);
@@ -1053,21 +1037,21 @@ namespace QCD {
   //  template<class vtype> strong_inline void fspProj5m (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
   template<class vtype> strong_inline void spProj5m (iScalar<vtype> &hspin,const iScalar<vtype> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProj5m(hspin._internal,fspin._internal);
   }
   //  template<class vtype,int N> strong_inline void fspProj5m (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
-  template<class vtype,int N> strong_inline void spProj5m (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
+  template<class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProj5m (iVector<vtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
       spProj5m(hspin._internal[i],fspin._internal[i]);
     }
   }
   //  template<class vtype,int N> strong_inline void fspProj5m (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
-  template<class vtype,int N> strong_inline void spProj5m (iMatrix<vtype,N> &hspin,iMatrix<vtype,N> &fspin)
+  template<class vtype,int N> strong_inline void spProj5m (iMatrix<vtype,N> &hspin,const iMatrix<vtype,N> &fspin)
   {
-    typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
+    //typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type *temp;
     for(int i=0;i<N;i++){ 
     for(int j=0;j<N;j++){
       spProj5m(hspin._internal[i][j],fspin._internal[i][j]);
diff --git a/lib/tensors/Tensor_traits.h b/lib/tensors/Tensor_traits.h
index 703a2ff5..a7d51b0e 100644
--- a/lib/tensors/Tensor_traits.h
+++ b/lib/tensors/Tensor_traits.h
@@ -202,7 +202,6 @@ namespace Grid {
     static const bool value = true;
   };
 
-
 }
 
 #endif
diff --git a/scripts/cleanup b/scripts/cleanup
index 956a6884..ff6dcf39 100755
--- a/scripts/cleanup
+++ b/scripts/cleanup
@@ -1,3 +1,4 @@
 find . -name "*~" -print -exec rm -f {} \;
 find . -name "errs" -print -exec rm -f {} \;
 find . -name "log" -print -exec rm -f {} \;
+find . -name "*.bak" -print -exec rm -f {} \;
diff --git a/tests/Test_cayley_cg.cc b/tests/Test_cayley_cg.cc
index f8d1187c..0f4955fd 100644
--- a/tests/Test_cayley_cg.cc
+++ b/tests/Test_cayley_cg.cc
@@ -72,34 +72,34 @@ int main (int argc, char ** argv)
   RealD mass=0.1;
   RealD M5  =1.8;
   std::cout<<GridLogMessage <<"DomainWallFermion test"<<std::endl;
-  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
-  TestCGinversions<DomainWallFermion>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  DomainWallFermionR Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  TestCGinversions<DomainWallFermionR>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   RealD b=1.5;// Scale factor b+c=2, b-c=1
   RealD c=0.5;
   std::cout<<GridLogMessage <<"MobiusFermion test"<<std::endl;
-  MobiusFermion Dmob(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c);
-  TestCGinversions<MobiusFermion>(Dmob,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  MobiusFermionR Dmob(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c);
+  TestCGinversions<MobiusFermionR>(Dmob,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   std::cout<<GridLogMessage <<"MobiusZolotarevFermion test"<<std::endl;
-  MobiusZolotarevFermion Dzolo(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c,0.1,2.0);
-  TestCGinversions<MobiusZolotarevFermion>(Dzolo,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  MobiusZolotarevFermionR Dzolo(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c,0.1,2.0);
+  TestCGinversions<MobiusZolotarevFermionR>(Dzolo,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   std::cout<<GridLogMessage <<"ScaledShamirFermion test"<<std::endl;
-  ScaledShamirFermion Dsham(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,2.0);
-  TestCGinversions<ScaledShamirFermion>(Dsham,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  ScaledShamirFermionR Dsham(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,2.0);
+  TestCGinversions<ScaledShamirFermionR>(Dsham,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   std::cout<<GridLogMessage <<"ShamirZolotarevFermion test"<<std::endl;
-  ShamirZolotarevFermion Dshamz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,2.0);
-  TestCGinversions<ShamirZolotarevFermion>(Dshamz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  ShamirZolotarevFermionR Dshamz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,2.0);
+  TestCGinversions<ShamirZolotarevFermionR>(Dshamz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   std::cout<<GridLogMessage <<"OverlapWilsonCayleyTanhFermion test"<<std::endl;
-  OverlapWilsonCayleyTanhFermion Dov(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
-  TestCGinversions<OverlapWilsonCayleyTanhFermion>(Dov,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  OverlapWilsonCayleyTanhFermionR Dov(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+  TestCGinversions<OverlapWilsonCayleyTanhFermionR>(Dov,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   std::cout<<GridLogMessage <<"OverlapWilsonCayleyZolotarevFermion test"<<std::endl;
-  OverlapWilsonCayleyZolotarevFermion Dovz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,2.0);
-  TestCGinversions<OverlapWilsonCayleyZolotarevFermion>(Dovz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  OverlapWilsonCayleyZolotarevFermionR Dovz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,2.0);
+  TestCGinversions<OverlapWilsonCayleyZolotarevFermionR>(Dovz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   Grid_finalize();
 }
diff --git a/tests/Test_cayley_coarsen_support.cc b/tests/Test_cayley_coarsen_support.cc
index 3ddf6205..77e77307 100644
--- a/tests/Test_cayley_coarsen_support.cc
+++ b/tests/Test_cayley_coarsen_support.cc
@@ -67,8 +67,8 @@ int main (int argc, char ** argv)
   RealD mass=0.5;
   RealD M5=1.8;
 
-  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
-  Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> HermIndefOp(Ddwf);
+  DomainWallFermionR Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  Gamma5R5HermitianLinearOperator<DomainWallFermionR,LatticeFermion> HermIndefOp(Ddwf);
 
   HermIndefOp.Op(src,ref);
   HermIndefOp.OpDiag(src,result);
@@ -89,7 +89,7 @@ int main (int argc, char ** argv)
 
   std::cout<<GridLogMessage<<"Calling Aggregation class" <<std::endl;
 
-  MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
+  MdagMLinearOperator<DomainWallFermionR,LatticeFermion> HermDefOp(Ddwf);
   typedef Aggregation<vSpinColourVector,vTComplex,nbasis> Subspace;
   Subspace Aggregates(Coarse5d,FGrid);
   Aggregates.CreateSubspaceRandom(RNG5);
diff --git a/tests/Test_cayley_even_odd.cc b/tests/Test_cayley_even_odd.cc
index d0c7b74a..c36567cd 100644
--- a/tests/Test_cayley_even_odd.cc
+++ b/tests/Test_cayley_even_odd.cc
@@ -49,35 +49,35 @@ int main (int argc, char ** argv)
   RealD mass=0.1;
   RealD M5  =1.8;
   std::cout<<GridLogMessage <<"DomainWallFermion test"<<std::endl;
-  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
-  TestWhat<DomainWallFermion>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  DomainWallFermionR Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  TestWhat<DomainWallFermionR>(Ddwf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   RealD b=1.5;// Scale factor b+c=2, b-c=1
   RealD c=0.5;
   std::cout<<GridLogMessage <<"MobiusFermion test"<<std::endl;
-  MobiusFermion Dmob(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c);
-  TestWhat<MobiusFermion>(Dmob,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  MobiusFermionR Dmob(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c);
+  TestWhat<MobiusFermionR>(Dmob,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   std::cout<<GridLogMessage <<"MobiusZolotarevFermion test"<<std::endl;
-  MobiusZolotarevFermion Dzolo(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c,0.1,2.0);
-  TestWhat<MobiusZolotarevFermion>(Dzolo,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  MobiusZolotarevFermionR Dzolo(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,b,c,0.1,2.0);
+  TestWhat<MobiusZolotarevFermionR>(Dzolo,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   std::cout<<GridLogMessage <<"ScaledShamirFermion test"<<std::endl;
-  ScaledShamirFermion Dsham(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,2.0);
-  TestWhat<ScaledShamirFermion>(Dsham,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  ScaledShamirFermionR Dsham(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,2.0);
+  TestWhat<ScaledShamirFermionR>(Dsham,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
 
   std::cout<<GridLogMessage <<"ShamirZolotarevFermion test"<<std::endl;
-  ShamirZolotarevFermion Dshamz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,2.0);
-  TestWhat<ShamirZolotarevFermion>(Dshamz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  ShamirZolotarevFermionR Dshamz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,2.0);
+  TestWhat<ShamirZolotarevFermionR>(Dshamz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   std::cout<<GridLogMessage <<"OverlapWilsonCayleyTanhFermion test"<<std::endl;
-  OverlapWilsonCayleyTanhFermion Dov(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
-  TestWhat<OverlapWilsonCayleyTanhFermion>(Dov,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  OverlapWilsonCayleyTanhFermionR Dov(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+  TestWhat<OverlapWilsonCayleyTanhFermionR>(Dov,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   std::cout<<GridLogMessage <<"OverlapWilsonCayleyZolotarevFermion test"<<std::endl;
-  OverlapWilsonCayleyZolotarevFermion Dovz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,2.0);
-  TestWhat<OverlapWilsonCayleyZolotarevFermion>(Dovz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  OverlapWilsonCayleyZolotarevFermionR Dovz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,2.0);
+  TestWhat<OverlapWilsonCayleyZolotarevFermionR>(Dovz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   Grid_finalize();
 }
diff --git a/tests/Test_cayley_ldop_cr.cc b/tests/Test_cayley_ldop_cr.cc
index 48d7c9dc..2414d6f1 100644
--- a/tests/Test_cayley_ldop_cr.cc
+++ b/tests/Test_cayley_ldop_cr.cc
@@ -55,8 +55,8 @@ int main (int argc, char ** argv)
   std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   std::cout<<GridLogMessage << "Building g5R5 hermitian DWF operator" <<std::endl;
   std::cout<<GridLogMessage << "**************************************************"<< std::endl;
-  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
-  Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> HermIndefOp(Ddwf);
+  DomainWallFermionR Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  Gamma5R5HermitianLinearOperator<DomainWallFermionR,LatticeFermion> HermIndefOp(Ddwf);
 
   const int nbasis = 8;
 
@@ -67,7 +67,7 @@ int main (int argc, char ** argv)
   std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   std::cout<<GridLogMessage << "Calling Aggregation class to build subspace" <<std::endl;
   std::cout<<GridLogMessage << "**************************************************"<< std::endl;
-  MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
+  MdagMLinearOperator<DomainWallFermionR,LatticeFermion> HermDefOp(Ddwf);
   Subspace Aggregates(Coarse5d,FGrid);
   Aggregates.CreateSubspace(RNG5,HermDefOp);
 
diff --git a/tests/Test_cf_coarsen_support.cc b/tests/Test_cf_coarsen_support.cc
index caa862be..d5cb7dfc 100644
--- a/tests/Test_cf_coarsen_support.cc
+++ b/tests/Test_cf_coarsen_support.cc
@@ -48,8 +48,8 @@ int main (int argc, char ** argv)
   RealD M5=1.8;
 
   {
-    OverlapWilsonContFracTanhFermion Dcf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
-    HermitianLinearOperator<OverlapWilsonContFracTanhFermion,LatticeFermion> HermIndefOp(Dcf);
+    OverlapWilsonContFracTanhFermionR Dcf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+    HermitianLinearOperator<OverlapWilsonContFracTanhFermionR,LatticeFermion> HermIndefOp(Dcf);
 
     HermIndefOp.Op(src,ref);
     HermIndefOp.OpDiag(src,result);
@@ -65,8 +65,8 @@ int main (int argc, char ** argv)
   }
 
   {
-    OverlapWilsonPartialFractionTanhFermion Dpf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
-    HermitianLinearOperator<OverlapWilsonPartialFractionTanhFermion,LatticeFermion> HermIndefOp(Dpf);
+    OverlapWilsonPartialFractionTanhFermionR Dpf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+    HermitianLinearOperator<OverlapWilsonPartialFractionTanhFermionR,LatticeFermion> HermIndefOp(Dpf);
     
     HermIndefOp.Op(src,ref);
     HermIndefOp.OpDiag(src,result);
diff --git a/tests/Test_cf_cr_unprec.cc b/tests/Test_cf_cr_unprec.cc
index 90acb503..7cb238ea 100644
--- a/tests/Test_cf_cr_unprec.cc
+++ b/tests/Test_cf_cr_unprec.cc
@@ -44,14 +44,14 @@ int main (int argc, char ** argv)
   RealD mass=0.1;
   RealD M5=1.8;
 
-  OverlapWilsonContFracTanhFermion Dcf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+  OverlapWilsonContFracTanhFermionR Dcf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
 
   ConjugateResidual<LatticeFermion> MCR(1.0e-8,10000);
 
-  MdagMLinearOperator<OverlapWilsonContFracTanhFermion,LatticeFermion> HermPosDefOp(Dcf);
+  MdagMLinearOperator<OverlapWilsonContFracTanhFermionR,LatticeFermion> HermPosDefOp(Dcf);
   MCR(HermPosDefOp,src,result);
 
-  HermitianLinearOperator<OverlapWilsonContFracTanhFermion,LatticeFermion> HermIndefOp(Dcf);
+  HermitianLinearOperator<OverlapWilsonContFracTanhFermionR,LatticeFermion> HermIndefOp(Dcf);
   MCR(HermIndefOp,src,result);
 
   Grid_finalize();
diff --git a/tests/Test_contfrac_cg.cc b/tests/Test_contfrac_cg.cc
index da6a81b9..abf46754 100644
--- a/tests/Test_contfrac_cg.cc
+++ b/tests/Test_contfrac_cg.cc
@@ -75,21 +75,21 @@ int main (int argc, char ** argv)
 
 
   std::cout<<GridLogMessage <<"OverlapWilsonContFracTanhFermion  test"<<std::endl;
-  OverlapWilsonContFracTanhFermion Dcf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
-  TestCGinversions<OverlapWilsonContFracTanhFermion>(Dcf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  OverlapWilsonContFracTanhFermionR Dcf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+  TestCGinversions<OverlapWilsonContFracTanhFermionR>(Dcf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   std::cout<<GridLogMessage <<"OverlapWilsonContFracZolotarevFermion  test"<<std::endl;
-  OverlapWilsonContFracZolotarevFermion Dcfz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,6.0);
-  TestCGinversions<OverlapWilsonContFracZolotarevFermion>(Dcfz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  OverlapWilsonContFracZolotarevFermionR Dcfz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,6.0);
+  TestCGinversions<OverlapWilsonContFracZolotarevFermionR>(Dcfz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
 
   std::cout<<GridLogMessage <<"OverlapWilsonPartialFractionTanhFermion  test"<<std::endl;
-  OverlapWilsonPartialFractionTanhFermion Dpf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
-  TestCGinversions<OverlapWilsonPartialFractionTanhFermion>(Dpf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  OverlapWilsonPartialFractionTanhFermionR Dpf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+  TestCGinversions<OverlapWilsonPartialFractionTanhFermionR>(Dpf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   std::cout<<GridLogMessage <<"OverlapWilsonPartialFractionZolotarevFermion  test"<<std::endl;
-  OverlapWilsonPartialFractionZolotarevFermion Dpfz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,6.0);
-  TestCGinversions<OverlapWilsonPartialFractionZolotarevFermion>(Dpfz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  OverlapWilsonPartialFractionZolotarevFermionR Dpfz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,6.0);
+  TestCGinversions<OverlapWilsonPartialFractionZolotarevFermionR>(Dpfz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
 
   Grid_finalize();
diff --git a/tests/Test_contfrac_even_odd.cc b/tests/Test_contfrac_even_odd.cc
index c8e5b093..7fcfc0f4 100644
--- a/tests/Test_contfrac_even_odd.cc
+++ b/tests/Test_contfrac_even_odd.cc
@@ -50,20 +50,20 @@ int main (int argc, char ** argv)
   RealD M5  =1.8;
 
   std::cout<<GridLogMessage <<"OverlapWilsonContFracTanhFermion  test"<<std::endl;
-  OverlapWilsonContFracTanhFermion Dcf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
-  TestWhat<OverlapWilsonContFracTanhFermion>(Dcf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  OverlapWilsonContFracTanhFermionR Dcf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+  TestWhat<OverlapWilsonContFracTanhFermionR>(Dcf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   std::cout<<GridLogMessage <<"OverlapWilsonContFracZolotarevFermion  test"<<std::endl;
-  OverlapWilsonContFracZolotarevFermion Dcfz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,6.0);
-  TestWhat<OverlapWilsonContFracZolotarevFermion>(Dcfz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  OverlapWilsonContFracZolotarevFermionR Dcfz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,6.0);
+  TestWhat<OverlapWilsonContFracZolotarevFermionR>(Dcfz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   std::cout<<GridLogMessage <<"OverlapWilsonPartialFractionTanhFermion  test"<<std::endl;
-  OverlapWilsonPartialFractionTanhFermion Dpf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
-  TestWhat<OverlapWilsonPartialFractionTanhFermion>(Dpf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  OverlapWilsonPartialFractionTanhFermionR Dpf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+  TestWhat<OverlapWilsonPartialFractionTanhFermionR>(Dpf,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   std::cout<<GridLogMessage <<"OverlapWilsonPartialFractionZolotarevFermion  test"<<std::endl;
-  OverlapWilsonPartialFractionZolotarevFermion Dpfz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,6.0);
-  TestWhat<OverlapWilsonPartialFractionZolotarevFermion>(Dpfz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
+  OverlapWilsonPartialFractionZolotarevFermionR Dpfz(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,0.1,6.0);
+  TestWhat<OverlapWilsonPartialFractionZolotarevFermionR>(Dpfz,FGrid,FrbGrid,UGrid,UrbGrid,mass,M5,&RNG4,&RNG5);
 
   Grid_finalize();
 }
diff --git a/tests/Test_contfrac_force.cc b/tests/Test_contfrac_force.cc
index a39d596d..58b42693 100644
--- a/tests/Test_contfrac_force.cc
+++ b/tests/Test_contfrac_force.cc
@@ -42,7 +42,7 @@ int main (int argc, char ** argv)
   ////////////////////////////////////
   RealD mass=0.01; 
   RealD M5=1.8; 
-  OverlapWilsonContFracTanhFermion Dcf(U,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+  OverlapWilsonContFracTanhFermionR Dcf(U,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
   Dcf.M   (phi,Mphi);
 
   ComplexD S    = innerProduct(Mphi,Mphi); // pdag MdagM p
diff --git a/tests/Test_dwf_cg_prec.cc b/tests/Test_dwf_cg_prec.cc
index 633d280f..a39b6529 100644
--- a/tests/Test_dwf_cg_prec.cc
+++ b/tests/Test_dwf_cg_prec.cc
@@ -45,14 +45,14 @@ int main (int argc, char ** argv)
   
   RealD mass=0.1;
   RealD M5=1.8;
-  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  DomainWallFermionR Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
 
   LatticeFermion    src_o(FrbGrid);
   LatticeFermion result_o(FrbGrid);
   pickCheckerboard(Odd,src_o,src);
   result_o=zero;
 
-  SchurDiagMooeeOperator<DomainWallFermion,LatticeFermion> HermOpEO(Ddwf);
+  SchurDiagMooeeOperator<DomainWallFermionR,LatticeFermion> HermOpEO(Ddwf);
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   CG(HermOpEO,src_o,result_o);
 
diff --git a/tests/Test_dwf_cg_schur.cc b/tests/Test_dwf_cg_schur.cc
index ac3989d9..e9db375e 100644
--- a/tests/Test_dwf_cg_schur.cc
+++ b/tests/Test_dwf_cg_schur.cc
@@ -43,7 +43,7 @@ int main (int argc, char ** argv)
   
   RealD mass=0.1;
   RealD M5=1.8;
-  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  DomainWallFermionR Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
 
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   SchurRedBlackDiagMooeeSolve<LatticeFermion> SchurSolver(CG);
diff --git a/tests/Test_dwf_cg_unprec.cc b/tests/Test_dwf_cg_unprec.cc
index 055e1bef..55d0c8d0 100644
--- a/tests/Test_dwf_cg_unprec.cc
+++ b/tests/Test_dwf_cg_unprec.cc
@@ -43,9 +43,9 @@ int main (int argc, char ** argv)
   
   RealD mass=0.1;
   RealD M5=1.8;
-  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  DomainWallFermionR Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
 
-  MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermOp(Ddwf);
+  MdagMLinearOperator<DomainWallFermionR,LatticeFermion> HermOp(Ddwf);
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   CG(HermOp,src,result);
 
diff --git a/tests/Test_dwf_cr_unprec.cc b/tests/Test_dwf_cr_unprec.cc
index c2b75853..2d791ce2 100644
--- a/tests/Test_dwf_cr_unprec.cc
+++ b/tests/Test_dwf_cr_unprec.cc
@@ -50,12 +50,12 @@ int main (int argc, char ** argv)
   
   RealD mass=0.5;
   RealD M5=1.8;
-  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  DomainWallFermionR Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
 
-  MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermOp(Ddwf);
+  MdagMLinearOperator<DomainWallFermionR,LatticeFermion> HermOp(Ddwf);
   MCR(HermOp,src,result);
 
-  Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> g5HermOp(Ddwf);
+  Gamma5R5HermitianLinearOperator<DomainWallFermionR,LatticeFermion> g5HermOp(Ddwf);
   MCR(g5HermOp,src,result);
 
 
diff --git a/tests/Test_dwf_even_odd.cc b/tests/Test_dwf_even_odd.cc
index 4e3d9799..96853b4c 100644
--- a/tests/Test_dwf_even_odd.cc
+++ b/tests/Test_dwf_even_odd.cc
@@ -58,7 +58,7 @@ int main (int argc, char ** argv)
 
   RealD mass=0.1;
   RealD M5  =1.8;
-  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  DomainWallFermionR Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
 
   LatticeFermion src_e (FrbGrid);
   LatticeFermion src_o (FrbGrid);
@@ -187,7 +187,7 @@ int main (int argc, char ** argv)
   RealD t1,t2;
 
 
-  SchurDiagMooeeOperator<DomainWallFermion,LatticeFermion> HermOpEO(Ddwf);
+  SchurDiagMooeeOperator<DomainWallFermionR,LatticeFermion> HermOpEO(Ddwf);
   HermOpEO.MpcDagMpc(chi_e,dchi_e,t1,t2);
   HermOpEO.MpcDagMpc(chi_o,dchi_o,t1,t2);
 
diff --git a/tests/Test_dwf_force.cc b/tests/Test_dwf_force.cc
index a3dbb0f2..ac6fd26a 100644
--- a/tests/Test_dwf_force.cc
+++ b/tests/Test_dwf_force.cc
@@ -42,7 +42,7 @@ int main (int argc, char ** argv)
   ////////////////////////////////////
   RealD mass=0.01; 
   RealD M5=1.8; 
-  DomainWallFermion Ddwf(U,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  DomainWallFermionR Ddwf(U,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
   Ddwf.M   (phi,Mphi);
 
   ComplexD S    = innerProduct(Mphi,Mphi); // pdag MdagM p
diff --git a/tests/Test_dwf_fpgcr.cc b/tests/Test_dwf_fpgcr.cc
index 6c589c6c..6c9645b9 100644
--- a/tests/Test_dwf_fpgcr.cc
+++ b/tests/Test_dwf_fpgcr.cc
@@ -52,19 +52,19 @@ int main (int argc, char ** argv)
   
   RealD mass=0.5;
   RealD M5=1.8;
-  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  DomainWallFermionR Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
 
   std::cout<<GridLogMessage<<"*********************************************************"<<std::endl;
   std::cout<<GridLogMessage<<"* Solving with MdagM VPGCR "<<std::endl;
   std::cout<<GridLogMessage<<"*********************************************************"<<std::endl;
-  MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermOp(Ddwf);
+  MdagMLinearOperator<DomainWallFermionR,LatticeFermion> HermOp(Ddwf);
   result=zero;
   PGCR(HermOp,src,result);
 
   std::cout<<GridLogMessage<<"*********************************************************"<<std::endl;
   std::cout<<GridLogMessage<<"* Solving with g5-VPGCR "<<std::endl;
   std::cout<<GridLogMessage<<"*********************************************************"<<std::endl;
-  Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> g5HermOp(Ddwf);
+  Gamma5R5HermitianLinearOperator<DomainWallFermionR,LatticeFermion> g5HermOp(Ddwf);
   result=zero;
   PGCR(g5HermOp,src,result);
 
diff --git a/tests/Test_dwf_hdcr.cc b/tests/Test_dwf_hdcr.cc
index 6ff26425..45647936 100644
--- a/tests/Test_dwf_hdcr.cc
+++ b/tests/Test_dwf_hdcr.cc
@@ -401,7 +401,7 @@ int main (int argc, char ** argv)
   std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   std::cout<<GridLogMessage << "Building g5R5 hermitian DWF operator" <<std::endl;
   std::cout<<GridLogMessage << "**************************************************"<< std::endl;
-  DomainWallFermion Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
+  DomainWallFermionR Ddwf(Umu,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5);
 
   const int nbasis = 32;
   //  const int nbasis = 4;
@@ -413,7 +413,7 @@ int main (int argc, char ** argv)
   std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   std::cout<<GridLogMessage << "Calling Aggregation class to build subspace" <<std::endl;
   std::cout<<GridLogMessage << "**************************************************"<< std::endl;
-  MdagMLinearOperator<DomainWallFermion,LatticeFermion> HermDefOp(Ddwf);
+  MdagMLinearOperator<DomainWallFermionR,LatticeFermion> HermDefOp(Ddwf);
   Subspace Aggregates(Coarse5d,FGrid);
   //  Aggregates.CreateSubspace(RNG5,HermDefOp,nbasis);
   assert ( (nbasis & 0x1)==0);
@@ -437,7 +437,7 @@ int main (int argc, char ** argv)
   std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   std::cout<<GridLogMessage << "Building coarse representation of Indef operator" <<std::endl;
   std::cout<<GridLogMessage << "**************************************************"<< std::endl;
-  Gamma5R5HermitianLinearOperator<DomainWallFermion,LatticeFermion> HermIndefOp(Ddwf);
+  Gamma5R5HermitianLinearOperator<DomainWallFermionR,LatticeFermion> HermIndefOp(Ddwf);
   CoarsenedMatrix<vSpinColourVector,vTComplex,nbasis> LDOp(*Coarse5d);
   LDOp.CoarsenOperator(FGrid,HermIndefOp,Aggregates);
 
@@ -467,7 +467,7 @@ int main (int argc, char ** argv)
   std::cout<<GridLogMessage << "Building deflation preconditioner "<< std::endl;
   std::cout<<GridLogMessage << "**************************************************"<< std::endl;
 
-  MultiGridPreconditioner <vSpinColourVector,vTComplex,nbasis,DomainWallFermion> Precon(Aggregates, LDOp,HermIndefOp,Ddwf);
+  MultiGridPreconditioner <vSpinColourVector,vTComplex,nbasis,DomainWallFermionR> Precon(Aggregates, LDOp,HermIndefOp,Ddwf);
   TrivialPrecon<LatticeFermion> simple;
 
   std::cout<<GridLogMessage << "**************************************************"<< std::endl;
@@ -506,7 +506,7 @@ int main (int argc, char ** argv)
   std::cout<<GridLogMessage << "**************************************************"<< std::endl;
   std::cout<<GridLogMessage << "Red Black Prec CG "<< std::endl;
   std::cout<<GridLogMessage << "**************************************************"<< std::endl;
-  SchurDiagMooeeOperator<DomainWallFermion,LatticeFermion> HermOpEO(Ddwf);
+  SchurDiagMooeeOperator<DomainWallFermionR,LatticeFermion> HermOpEO(Ddwf);
   ConjugateGradient<LatticeFermion> pCG(1.0e-8,10000);
 
   LatticeFermion    src_o(FrbGrid);
diff --git a/tests/Test_hmc_EOWilsonFermionGauge.cc b/tests/Test_hmc_EOWilsonFermionGauge.cc
index 1c90933e..66ec118b 100644
--- a/tests/Test_hmc_EOWilsonFermionGauge.cc
+++ b/tests/Test_hmc_EOWilsonFermionGauge.cc
@@ -25,13 +25,12 @@ int main (int argc, char ** argv)
   WilsonGaugeAction<LatticeLorentzColourMatrix, LatticeColourMatrix> Waction(5.6);
 
   Real mass=-0.77;
-  WilsonFermion FermOp(U,Fine,RBFine,mass);
+  WilsonFermionR FermOp(U,Fine,RBFine,mass);
   
   ConjugateGradient<LatticeFermion>  CG(1.0e-8,10000);
   ConjugateGradient<LatticeFermion>  CGmd(1.0e-6,10000);
   
-  TwoFlavourEvenOddPseudoFermionAction<LatticeLorentzColourMatrix, LatticeColourMatrix,LatticeFermion> 
-    WilsonNf2(FermOp,CG,CG);
+  TwoFlavourEvenOddPseudoFermionAction<WilsonImplR> WilsonNf2(FermOp,CG,CG);
   
   //Collect actions
   ActionLevel Level1;
diff --git a/tests/Test_hmc_WilsonFermionGauge.cc b/tests/Test_hmc_WilsonFermionGauge.cc
index e3bbf393..ef598062 100644
--- a/tests/Test_hmc_WilsonFermionGauge.cc
+++ b/tests/Test_hmc_WilsonFermionGauge.cc
@@ -25,13 +25,12 @@ int main (int argc, char ** argv)
   WilsonGaugeAction<LatticeLorentzColourMatrix, LatticeColourMatrix> Waction(5.6);
 
   Real mass=-0.77;
-  WilsonFermion FermOp(U,Fine,RBFine,mass);
+  WilsonFermionR FermOp(U,Fine,RBFine,mass);
   
   ConjugateGradient<LatticeFermion>  CG(1.0e-8,10000);
   ConjugateGradient<LatticeFermion>  CGmd(1.0e-6,10000);
   
-  TwoFlavourPseudoFermionAction<LatticeLorentzColourMatrix, LatticeColourMatrix,LatticeFermion> 
-    WilsonNf2(FermOp,CG,CG);
+  TwoFlavourPseudoFermionAction<WilsonImplR> WilsonNf2(FermOp,CG,CG);
   
   //Collect actions
   ActionLevel Level1;
diff --git a/tests/Test_partfrac_force.cc b/tests/Test_partfrac_force.cc
index 05925e55..6e543d3c 100644
--- a/tests/Test_partfrac_force.cc
+++ b/tests/Test_partfrac_force.cc
@@ -42,7 +42,7 @@ int main (int argc, char ** argv)
   ////////////////////////////////////
   RealD mass=0.01; 
   RealD M5=1.8; 
-  OverlapWilsonPartialFractionTanhFermion Dpf(U,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
+  OverlapWilsonPartialFractionTanhFermionR Dpf(U,*FGrid,*FrbGrid,*UGrid,*UrbGrid,mass,M5,1.0);
   Dpf.M   (phi,Mphi);
 
   ComplexD S    = innerProduct(Mphi,Mphi); // pdag MdagM p
diff --git a/tests/Test_wilson_cg_prec.cc b/tests/Test_wilson_cg_prec.cc
index c59699e4..756f8b6f 100644
--- a/tests/Test_wilson_cg_prec.cc
+++ b/tests/Test_wilson_cg_prec.cc
@@ -42,7 +42,7 @@ int main (int argc, char ** argv)
   }
   
   RealD mass=0.5;
-  WilsonFermion Dw(Umu,Grid,RBGrid,mass);
+  WilsonFermionR Dw(Umu,Grid,RBGrid,mass);
 
   //  HermitianOperator<WilsonFermion,LatticeFermion> HermOp(Dw);
   //  ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
@@ -53,7 +53,7 @@ int main (int argc, char ** argv)
   pickCheckerboard(Odd,src_o,src);
   result_o=zero;
 
-  SchurDiagMooeeOperator<WilsonFermion,LatticeFermion> HermOpEO(Dw);
+  SchurDiagMooeeOperator<WilsonFermionR,LatticeFermion> HermOpEO(Dw);
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   CG(HermOpEO,src_o,result_o);
 
diff --git a/tests/Test_wilson_cg_schur.cc b/tests/Test_wilson_cg_schur.cc
index 1484a6a8..2174d644 100644
--- a/tests/Test_wilson_cg_schur.cc
+++ b/tests/Test_wilson_cg_schur.cc
@@ -37,7 +37,7 @@ int main (int argc, char ** argv)
   LatticeFermion resid(&Grid); 
 
   RealD mass=0.5;
-  WilsonFermion Dw(Umu,Grid,RBGrid,mass);
+  WilsonFermionR Dw(Umu,Grid,RBGrid,mass);
 
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   SchurRedBlackDiagMooeeSolve<LatticeFermion> SchurSolver(CG);
diff --git a/tests/Test_wilson_cg_unprec.cc b/tests/Test_wilson_cg_unprec.cc
index 5984cac9..9269aeaa 100644
--- a/tests/Test_wilson_cg_unprec.cc
+++ b/tests/Test_wilson_cg_unprec.cc
@@ -40,9 +40,9 @@ int main (int argc, char ** argv)
   }  
   
   RealD mass=0.5;
-  WilsonFermion Dw(Umu,Grid,RBGrid,mass);
+  WilsonFermionR Dw(Umu,Grid,RBGrid,mass);
 
-  MdagMLinearOperator<WilsonFermion,LatticeFermion> HermOp(Dw);
+  MdagMLinearOperator<WilsonFermionR,LatticeFermion> HermOp(Dw);
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   CG(HermOp,src,result);
 
diff --git a/tests/Test_wilson_cr_unprec.cc b/tests/Test_wilson_cr_unprec.cc
index 468f0d5a..c7e0bc33 100644
--- a/tests/Test_wilson_cr_unprec.cc
+++ b/tests/Test_wilson_cr_unprec.cc
@@ -43,9 +43,9 @@ int main (int argc, char ** argv)
   }  
 
   RealD mass=0.5;
-  WilsonFermion Dw(Umu,Grid,RBGrid,mass);
+  WilsonFermionR Dw(Umu,Grid,RBGrid,mass);
 
-  MdagMLinearOperator<WilsonFermion,LatticeFermion> HermOp(Dw);
+  MdagMLinearOperator<WilsonFermionR,LatticeFermion> HermOp(Dw);
 
   ConjugateResidual<LatticeFermion> MCR(1.0e-8,10000);
 
diff --git a/tests/Test_wilson_even_odd.cc b/tests/Test_wilson_even_odd.cc
index 66ac89dc..37ba40f1 100644
--- a/tests/Test_wilson_even_odd.cc
+++ b/tests/Test_wilson_even_odd.cc
@@ -62,7 +62,7 @@ int main (int argc, char ** argv)
 
   RealD mass=0.1;
 
-  WilsonFermion Dw(Umu,Grid,RBGrid,mass);
+  WilsonFermionR Dw(Umu,Grid,RBGrid,mass);
 
   LatticeFermion src_e   (&RBGrid);
   LatticeFermion src_o   (&RBGrid);
@@ -179,7 +179,7 @@ int main (int argc, char ** argv)
   pickCheckerboard(Odd ,phi_o,phi);
   RealD t1,t2;
 
-  SchurDiagMooeeOperator<WilsonFermion,LatticeFermion> HermOpEO(Dw);
+  SchurDiagMooeeOperator<WilsonFermionR,LatticeFermion> HermOpEO(Dw);
   HermOpEO.MpcDagMpc(chi_e,dchi_e,t1,t2);
   HermOpEO.MpcDagMpc(chi_o,dchi_o,t1,t2);
 
diff --git a/tests/Test_wilson_force.cc b/tests/Test_wilson_force.cc
index b08ccd45..1b22c2e4 100644
--- a/tests/Test_wilson_force.cc
+++ b/tests/Test_wilson_force.cc
@@ -38,7 +38,7 @@ int main (int argc, char ** argv)
   // Unmodified matrix element
   ////////////////////////////////////
   RealD mass=-4.0; //kills the diagonal term
-  WilsonFermion Dw     (U,     Grid,RBGrid,mass);
+  WilsonFermionR Dw     (U,     Grid,RBGrid,mass);
   Dw.M   (phi,Mphi);
 
   ComplexD S    = innerProduct(Mphi,Mphi); // pdag MdagM p
@@ -88,7 +88,7 @@ int main (int argc, char ** argv)
   }
 
   std::cout << GridLogMessage <<"Initial mom hamiltonian is "<< Hmom <<std::endl;
-  Dw.DoubleStore(Dw.Umu,Uprime);
+  Dw.ImportGauge(Uprime);
   Dw.M          (phi,MphiPrime);
 
   ComplexD Sprime    = innerProduct(MphiPrime   ,MphiPrime);
diff --git a/tests/Test_wilson_force_phiMdagMphi.cc b/tests/Test_wilson_force_phiMdagMphi.cc
index 33c5f6e1..88f3161f 100644
--- a/tests/Test_wilson_force_phiMdagMphi.cc
+++ b/tests/Test_wilson_force_phiMdagMphi.cc
@@ -42,7 +42,7 @@ int main (int argc, char ** argv)
   // Unmodified matrix element
   ////////////////////////////////////
   RealD mass=-4.0; //kills the diagonal term
-  WilsonFermion Dw     (U,     Grid,RBGrid,mass);
+  WilsonFermionR Dw     (U,     Grid,RBGrid,mass);
   Dw.M   (phi,Mphi);
   Dw.Mdag(phi,Mdagphi);
 
diff --git a/tests/Test_wilson_force_phiMphi.cc b/tests/Test_wilson_force_phiMphi.cc
index f1e42195..ff9e0fb3 100644
--- a/tests/Test_wilson_force_phiMphi.cc
+++ b/tests/Test_wilson_force_phiMphi.cc
@@ -39,7 +39,7 @@ int main (int argc, char ** argv)
   // Unmodified matrix element
   ////////////////////////////////////
   RealD mass=-4.0; //kills the diagonal term
-  WilsonFermion Dw     (U,     Grid,RBGrid,mass);
+  WilsonFermionR Dw     (U,     Grid,RBGrid,mass);
   Dw.M(phi,Mphi);
 
   ComplexD S = innerProduct(phi,Mphi);

From 881acaa065ad0c2570099994599d54644f4f3e61 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 11 Aug 2015 06:22:20 +0100
Subject: [PATCH 419/429] Gparity modifications in the Gparity compressor
 variant.

---
 lib/cshift/Cshift_common.h                |  8 ++--
 lib/qcd/action/fermion/WilsonCompressor.h | 47 ++++++++++++++++++++---
 2 files changed, 46 insertions(+), 9 deletions(-)

diff --git a/lib/cshift/Cshift_common.h b/lib/cshift/Cshift_common.h
index ffa3f771..822e02c9 100644
--- a/lib/cshift/Cshift_common.h
+++ b/lib/cshift/Cshift_common.h
@@ -8,7 +8,7 @@ class SimpleCompressor {
 public:
   void Point(int) {};
 
-  vobj operator() (const vobj &arg) {
+  vobj operator() (const vobj &arg,int dimension,int plane,int osite,GridBase *grid) {
     return arg;
   }
 };
@@ -24,7 +24,7 @@ Gather_plane_simple (const Lattice<vobj> &rhs,std::vector<cobj,alignedAllocator<
   if ( !rhs._grid->CheckerBoarded(dimension) ) {
     cbmask = 0x3;
   }
-
+  
   int so  = plane*rhs._grid->_ostride[dimension]; // base offset for start of plane 
   
   int e1=rhs._grid->_slice_nblock[dimension];
@@ -36,7 +36,7 @@ PARALLEL_NESTED_LOOP2
       int bo = n*rhs._grid->_slice_block[dimension];
       int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);// Could easily be a table lookup
       if ( ocb &cbmask ) {
-	buffer[bo+b]=compress(rhs._odata[so+o+b]);
+	buffer[bo+b]=compress(rhs._odata[so+o+b],dimension,plane,so+o+b,rhs._grid);
       }
     }
   }
@@ -69,7 +69,7 @@ PARALLEL_NESTED_LOOP2
       int ocb=1<<rhs._grid->CheckerBoardFromOindex(o+b);
       if ( ocb & cbmask ) {
 	cobj temp; 
-	temp =compress(rhs._odata[so+o+b]);
+	temp =compress(rhs._odata[so+o+b],dimension,plane,so+o+b,rhs._grid);
 	extract<cobj>(temp,pointers,offset);
       }
     }
diff --git a/lib/qcd/action/fermion/WilsonCompressor.h b/lib/qcd/action/fermion/WilsonCompressor.h
index c7a20c59..c879e9f1 100644
--- a/lib/qcd/action/fermion/WilsonCompressor.h
+++ b/lib/qcd/action/fermion/WilsonCompressor.h
@@ -19,7 +19,7 @@ namespace QCD {
       mu=p;
     };
 
-    virtual SiteHalfSpinor operator () (const SiteSpinor &in) {
+    virtual SiteHalfSpinor operator () (const SiteSpinor &in,int dim,int plane,int osite,GridBase *grid) {
       return spinproject(in);
     }
 
@@ -67,13 +67,50 @@ namespace QCD {
   template<class SiteHalfSpinor,class SiteSpinor>
     class GparityWilsonCompressor : public WilsonCompressor<SiteHalfSpinor,SiteSpinor>{
   public:
-
     GparityWilsonCompressor(int _dag) : WilsonCompressor<SiteHalfSpinor,SiteSpinor> (_dag){};
 
-    SiteHalfSpinor operator () (const SiteSpinor &in)
+    SiteHalfSpinor operator () (const SiteSpinor &in,int dim,int plane,int osite,GridBase *grid)
     {
-      SiteHalfSpinor tmp = spinproject(in);
-      if( 0 ) tmp = flavourflip(tmp);
+      std::vector<int> Gbcs({1,0,0,0});
+
+      typedef typename SiteHalfSpinor::scalar_object scalar_object;
+
+      const int Nsimd = grid->Nsimd();
+
+      int ocb=grid->CheckerBoardFromOindex(osite);
+
+      SiteHalfSpinor tmp = spinproject(in); // spin projected
+
+      //////////////////////////////////////////////////////////////
+      // Check whether we must flavour flip
+      // do this if source is plane 0 on processor 0 in dimension dim
+      //////////////////////////////////////////////////////////////
+      if ( (this->mu==Xp)||(this->mu==Yp)||(this->mu==Zp)||(this->mu==Tp) ) {      
+
+	if ( Gbcs[this->mu] && (grid->_processor_coor[dim] == 0) && (plane==0) && (ocb==0) ) {
+
+	  std::vector<scalar_object>  flat(Nsimd);
+
+	  extract(tmp,flat);
+
+	  for(int i=0;i<Nsimd;i++) {
+	    std::vector<int> coor;
+
+	    grid->iCoorFromIindex(coor,i);
+
+	    if ( coor[dim]==0 ) {
+	      scalar_object stmp;
+	      stmp(0) = flat[i](1);
+	      stmp(1) = flat[i](0);
+	      flat[i] = stmp;
+	    }
+	  }
+
+	  merge(tmp,flat);
+
+	} // coor & bc guard
+      } // shift guard
+
       return tmp;
     }
 

From 26f5ee06219d078ecb927b03d52466d485205850 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 11 Aug 2015 06:23:38 +0100
Subject: [PATCH 420/429] Header

---
 lib/qcd/action/fermion/FermionImplTypedefs.h | 10 ++++++++++
 1 file changed, 10 insertions(+)
 create mode 100644 lib/qcd/action/fermion/FermionImplTypedefs.h

diff --git a/lib/qcd/action/fermion/FermionImplTypedefs.h b/lib/qcd/action/fermion/FermionImplTypedefs.h
new file mode 100644
index 00000000..99657d83
--- /dev/null
+++ b/lib/qcd/action/fermion/FermionImplTypedefs.h
@@ -0,0 +1,10 @@
+    public:
+      typedef typename Impl::Simd                           Simd;
+      typedef typename Impl::FermionField           FermionField;
+      typedef typename Impl::GaugeLinkField       GaugeLinkField;
+      typedef typename Impl::GaugeField               GaugeField;
+      typedef typename Impl::DoubledGaugeField DoubledGaugeField;
+      typedef typename Impl::SiteSpinor               SiteSpinor;
+      typedef typename Impl::SiteHalfSpinor       SiteHalfSpinor;
+      typedef typename Impl::Compressor               Compressor;
+      typedef WilsonKernels<Impl> Kernels;

From 9cd7f9ecad49689d0ffb6ff35003f61212f31eca Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 11 Aug 2015 06:24:53 +0100
Subject: [PATCH 421/429] Preconditioned conjugate residual

---
 .../iterative/PrecConjugateResidual.h         | 92 +++++++++++++++++++
 1 file changed, 92 insertions(+)
 create mode 100644 lib/algorithms/iterative/PrecConjugateResidual.h

diff --git a/lib/algorithms/iterative/PrecConjugateResidual.h b/lib/algorithms/iterative/PrecConjugateResidual.h
new file mode 100644
index 00000000..19f6e5f0
--- /dev/null
+++ b/lib/algorithms/iterative/PrecConjugateResidual.h
@@ -0,0 +1,92 @@
+#ifndef GRID_PREC_CONJUGATE_RESIDUAL_H
+#define GRID_PREC_CONJUGATE_RESIDUAL_H
+
+namespace Grid {
+
+    /////////////////////////////////////////////////////////////
+    // Base classes for iterative processes based on operators
+    // single input vec, single output vec.
+    /////////////////////////////////////////////////////////////
+
+  template<class Field> 
+    class PrecConjugateResidual : public OperatorFunction<Field> {
+  public:                                                
+    RealD   Tolerance;
+    Integer MaxIterations;
+    int verbose;
+    LinearFunction<Field> &Preconditioner;
+
+    PrecConjugateResidual(RealD tol,Integer maxit,LinearFunction<Field> &Prec) : Tolerance(tol), MaxIterations(maxit),      Preconditioner(Prec)
+    { 
+      verbose=1;
+    };
+
+    void operator() (LinearOperatorBase<Field> &Linop,const Field &src, Field &psi){
+
+      RealD a, b, c, d;
+      RealD cp, ssq,rsq;
+      
+      RealD rAr, rAAr, rArp;
+      RealD pAp, pAAp;
+
+      GridBase *grid = src._grid;
+      Field r(grid),  p(grid), Ap(grid), Ar(grid), z(grid);
+      
+      psi=zero;
+      r  = src;
+      Preconditioner(r,p);
+
+      
+
+      Linop.HermOpAndNorm(p,Ap,pAp,pAAp);
+      Ar=Ap;
+      rAr=pAp;
+      rAAr=pAAp;
+
+      cp =norm2(r);
+      ssq=norm2(src);
+      rsq=Tolerance*Tolerance*ssq;
+
+      if (verbose) std::cout<<GridLogMessage<<"PrecConjugateResidual: iteration " <<0<<" residual "<<cp<< " target"<< rsq<<std::endl;
+
+      for(int k=0;k<MaxIterations;k++){
+
+
+	Preconditioner(Ap,z);
+	RealD rq= real(innerProduct(Ap,z)); 
+
+	a = rAr/rq;
+
+   	axpy(psi,a,p,psi);
+   cp = axpy_norm(r,-a,z,r);
+
+	rArp=rAr;
+
+	Linop.HermOpAndNorm(r,Ar,rAr,rAAr);
+
+	b   =rAr/rArp;
+ 
+	axpy(p,b,p,r);
+	pAAp=axpy_norm(Ap,b,Ap,Ar);
+	
+	if(verbose) std::cout<<GridLogMessage<<"PrecConjugateResidual: iteration " <<k<<" residual "<<cp<< " target"<< rsq<<std::endl;
+
+	if(cp<rsq) {
+	  Linop.HermOp(psi,Ap);
+	  axpy(r,-1.0,src,Ap);
+	  RealD true_resid = norm2(r)/ssq;
+	  std::cout<<GridLogMessage<<"PrecConjugateResidual: Converged on iteration " <<k
+		   << " computed residual "<<sqrt(cp/ssq)
+	           << " true residual "<<sqrt(true_resid)
+	           << " target "       <<Tolerance <<std::endl;
+	  return;
+	}
+
+      }
+
+      std::cout<<GridLogMessage<<"PrecConjugateResidual did NOT converge"<<std::endl;
+      assert(0);
+    }
+  };
+}
+#endif

From ded3945467880b5e0a5e6b3250ae1807b7256e61 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Tue, 11 Aug 2015 06:37:42 +0100
Subject: [PATCH 422/429] File list update.

---
 lib/Make.inc | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/Make.inc b/lib/Make.inc
index e96c6fac..8b2bcf40 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/BfmHDCG.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/GeneralisedMinimumResidual.h ./algorithms/iterative/gmres.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Config.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./Init.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./Log.h ./MacroMagic.h ./Old/Tensor_peek.h ./Old/Tensor_poke.h ./parallelIO/NerscIO.h ./parameterIO/XMLReader.h ./qcd/action/ActionBase.h ./qcd/action/Actions.h ./qcd/action/DiffAction.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/pseudofermion/TwoFlavour.h ./qcd/action/pseudofermion/TwoFlavourEvenOdd.h ./qcd/action/pseudofermion/TwoFlavourEvenOddConstEE.h ./qcd/hmc/HMC.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/hmc/integrators/Integrator_base.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_index.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h ./Timer.h
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Config.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./Init.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./Log.h ./MacroMagic.h ./Old/Tensor_peek.h ./Old/Tensor_poke.h ./parallelIO/NerscIO.h ./parameterIO/XMLReader.h ./qcd/action/ActionBase.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionImplTypedefs.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/pseudofermion/TwoFlavour.h ./qcd/action/pseudofermion/TwoFlavourEvenOdd.h ./qcd/hmc/HMC.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/hmc/integrators/Integrator_base.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_index.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h ./Timer.h
 
 CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./Init.cc ./Log.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/hmc/HMC.cc ./qcd/hmc/integrators/Integrator.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc

From 8a0be4208059f031862745783879be5009687347 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Wed, 12 Aug 2015 09:49:33 +0100
Subject: [PATCH 423/429] Gparity test added; partial implementation -- this is
 Chris K's doubled lattice only and have to regress this with the 2 flavour
 implementation.

---
 benchmarks/Benchmark_su3.cc               |  8 +--
 lib/Make.inc                              |  2 +-
 lib/qcd/action/fermion/CayleyFermion5D.cc | 11 +++-
 scripts/configure-commands                |  3 +
 tests/Make.inc                            |  6 +-
 tests/Test_gparity.cc                     | 73 +++++++++++++++++++++++
 6 files changed, 94 insertions(+), 9 deletions(-)
 create mode 100644 tests/Test_gparity.cc

diff --git a/benchmarks/Benchmark_su3.cc b/benchmarks/Benchmark_su3.cc
index 3312e3c8..d1accefd 100644
--- a/benchmarks/Benchmark_su3.cc
+++ b/benchmarks/Benchmark_su3.cc
@@ -22,7 +22,7 @@ int main (int argc, char ** argv)
   std::cout<<GridLogMessage << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
   std::cout<<GridLogMessage << "----------------------------------------------------------"<<std::endl;
 
-  for(int lat=2;lat<=24;lat+=2){
+  for(int lat=2;lat<=32;lat+=2){
 
       std::vector<int> latt_size  ({lat*mpi_layout[0],lat*mpi_layout[1],lat*mpi_layout[2],lat*mpi_layout[3]});
       int vol = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
@@ -54,7 +54,7 @@ int main (int argc, char ** argv)
   std::cout<<GridLogMessage << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
   std::cout<<GridLogMessage << "----------------------------------------------------------"<<std::endl;
 
-  for(int lat=2;lat<=24;lat+=2){
+  for(int lat=2;lat<=32;lat+=2){
 
       std::vector<int> latt_size  ({lat*mpi_layout[0],lat*mpi_layout[1],lat*mpi_layout[2],lat*mpi_layout[3]});
       int vol = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
@@ -85,7 +85,7 @@ int main (int argc, char ** argv)
   std::cout<<GridLogMessage << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
   std::cout<<GridLogMessage << "----------------------------------------------------------"<<std::endl;
 
-  for(int lat=2;lat<=24;lat+=2){
+  for(int lat=2;lat<=32;lat+=2){
 
       std::vector<int> latt_size  ({lat*mpi_layout[0],lat*mpi_layout[1],lat*mpi_layout[2],lat*mpi_layout[3]});
       int vol = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
@@ -116,7 +116,7 @@ int main (int argc, char ** argv)
   std::cout<<GridLogMessage << "  L  "<<"\t\t"<<"bytes"<<"\t\t\t"<<"GB/s\t\t GFlop/s"<<std::endl;
   std::cout<<GridLogMessage << "----------------------------------------------------------"<<std::endl;
 
-  for(int lat=2;lat<=24;lat+=2){
+  for(int lat=2;lat<=32;lat+=2){
 
       std::vector<int> latt_size  ({lat*mpi_layout[0],lat*mpi_layout[1],lat*mpi_layout[2],lat*mpi_layout[3]});
       int vol = latt_size[0]*latt_size[1]*latt_size[2]*latt_size[3];
diff --git a/lib/Make.inc b/lib/Make.inc
index 8b2bcf40..8d595ae5 100644
--- a/lib/Make.inc
+++ b/lib/Make.inc
@@ -1,4 +1,4 @@
 
-HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Config.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./Init.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./Log.h ./MacroMagic.h ./Old/Tensor_peek.h ./Old/Tensor_poke.h ./parallelIO/NerscIO.h ./parameterIO/XMLReader.h ./qcd/action/ActionBase.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionImplTypedefs.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/pseudofermion/TwoFlavour.h ./qcd/action/pseudofermion/TwoFlavourEvenOdd.h ./qcd/hmc/HMC.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/hmc/integrators/Integrator_base.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_index.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h ./Timer.h
+HFILES=./algorithms/approx/bigfloat.h ./algorithms/approx/bigfloat_double.h ./algorithms/approx/Chebyshev.h ./algorithms/approx/MultiShiftFunction.h ./algorithms/approx/Remez.h ./algorithms/approx/Zolotarev.h ./algorithms/CoarsenedMatrix.h ./algorithms/iterative/AdefGeneric.h ./algorithms/iterative/ConjugateGradient.h ./algorithms/iterative/ConjugateGradientMultiShift.h ./algorithms/iterative/ConjugateResidual.h ./algorithms/iterative/NormalEquations.h ./algorithms/iterative/PrecConjugateResidual.h ./algorithms/iterative/PrecGeneralisedConjugateResidual.h ./algorithms/iterative/SchurRedBlack.h ./algorithms/LinearOperator.h ./algorithms/Preconditioner.h ./algorithms/SparseMatrix.h ./Algorithms.h ./AlignedAllocator.h ./cartesian/Cartesian_base.h ./cartesian/Cartesian_full.h ./cartesian/Cartesian_red_black.h ./Cartesian.h ./communicator/Communicator_base.h ./Communicator.h ./Config.h ./cshift/Cshift_common.h ./cshift/Cshift_mpi.h ./cshift/Cshift_none.h ./Cshift.h ./Grid.h ./Init.h ./lattice/Lattice_arith.h ./lattice/Lattice_base.h ./lattice/Lattice_comparison.h ./lattice/Lattice_comparison_utils.h ./lattice/Lattice_conformable.h ./lattice/Lattice_coordinate.h ./lattice/Lattice_ET.h ./lattice/Lattice_local.h ./lattice/Lattice_overload.h ./lattice/Lattice_peekpoke.h ./lattice/Lattice_reality.h ./lattice/Lattice_reduction.h ./lattice/Lattice_rng.h ./lattice/Lattice_trace.h ./lattice/Lattice_transfer.h ./lattice/Lattice_transpose.h ./lattice/Lattice_unary.h ./lattice/Lattice_where.h ./Lattice.h ./Log.h ./MacroMagic.h ./Old/Tensor_peek.h ./Old/Tensor_poke.h ./parallelIO/NerscIO.h ./qcd/action/ActionBase.h ./qcd/action/Actions.h ./qcd/action/fermion/CayleyFermion5D.h ./qcd/action/fermion/ContinuedFractionFermion5D.h ./qcd/action/fermion/DomainWallFermion.h ./qcd/action/fermion/FermionImplTypedefs.h ./qcd/action/fermion/FermionOperator.h ./qcd/action/fermion/g5HermitianLinop.h ./qcd/action/fermion/MobiusFermion.h ./qcd/action/fermion/MobiusZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonCayleyTanhFermion.h ./qcd/action/fermion/OverlapWilsonCayleyZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonContfracTanhFermion.h ./qcd/action/fermion/OverlapWilsonContfracZolotarevFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionTanhFermion.h ./qcd/action/fermion/OverlapWilsonPartialFractionZolotarevFermion.h ./qcd/action/fermion/PartialFractionFermion5D.h ./qcd/action/fermion/ScaledShamirFermion.h ./qcd/action/fermion/ShamirZolotarevFermion.h ./qcd/action/fermion/WilsonCompressor.h ./qcd/action/fermion/WilsonFermion.h ./qcd/action/fermion/WilsonFermion5D.h ./qcd/action/fermion/WilsonKernels.h ./qcd/action/gauge/WilsonGaugeAction.h ./qcd/action/pseudofermion/TwoFlavour.h ./qcd/action/pseudofermion/TwoFlavourEvenOdd.h ./qcd/hmc/HMC.h ./qcd/hmc/integrators/Integrator.h ./qcd/hmc/integrators/Integrator_algorithm.h ./qcd/hmc/integrators/Integrator_base.h ./qcd/QCD.h ./qcd/spin/Dirac.h ./qcd/spin/TwoSpinor.h ./qcd/utils/CovariantCshift.h ./qcd/utils/LinalgUtils.h ./qcd/utils/SpaceTimeGrid.h ./qcd/utils/SUn.h ./qcd/utils/WilsonLoops.h ./simd/Grid_avx.h ./simd/Grid_avx512.h ./simd/Grid_empty.h ./simd/Grid_neon.h ./simd/Grid_qpx.h ./simd/Grid_sse4.h ./simd/Grid_vector_types.h ./simd/Grid_vector_unops.h ./Simd.h ./stencil/Lebesgue.h ./Stencil.h ./tensors/Tensor_arith.h ./tensors/Tensor_arith_add.h ./tensors/Tensor_arith_mac.h ./tensors/Tensor_arith_mul.h ./tensors/Tensor_arith_scalar.h ./tensors/Tensor_arith_sub.h ./tensors/Tensor_class.h ./tensors/Tensor_determinant.h ./tensors/Tensor_exp.h ./tensors/Tensor_extract_merge.h ./tensors/Tensor_index.h ./tensors/Tensor_inner.h ./tensors/Tensor_logical.h ./tensors/Tensor_outer.h ./tensors/Tensor_reality.h ./tensors/Tensor_Ta.h ./tensors/Tensor_trace.h ./tensors/Tensor_traits.h ./tensors/Tensor_transpose.h ./tensors/Tensor_unary.h ./Tensors.h ./Threads.h ./Timer.h
 
 CCFILES=./algorithms/approx/MultiShiftFunction.cc ./algorithms/approx/Remez.cc ./algorithms/approx/Zolotarev.cc ./Init.cc ./Log.cc ./qcd/action/fermion/CayleyFermion5D.cc ./qcd/action/fermion/ContinuedFractionFermion5D.cc ./qcd/action/fermion/PartialFractionFermion5D.cc ./qcd/action/fermion/WilsonFermion.cc ./qcd/action/fermion/WilsonFermion5D.cc ./qcd/action/fermion/WilsonKernels.cc ./qcd/action/fermion/WilsonKernelsHand.cc ./qcd/hmc/HMC.cc ./qcd/hmc/integrators/Integrator.cc ./qcd/spin/Dirac.cc ./qcd/utils/SpaceTimeGrid.cc ./stencil/Lebesgue.cc ./stencil/Stencil_common.cc
diff --git a/lib/qcd/action/fermion/CayleyFermion5D.cc b/lib/qcd/action/fermion/CayleyFermion5D.cc
index 1ec7f930..76066adf 100644
--- a/lib/qcd/action/fermion/CayleyFermion5D.cc
+++ b/lib/qcd/action/fermion/CayleyFermion5D.cc
@@ -158,12 +158,14 @@ namespace QCD {
  template<class Impl>
   void CayleyFermion5D<Impl>::Meooe       (const FermionField &psi, FermionField &chi)
   {
+    int Ls=this->Ls;
+
     FermionField tmp(psi._grid);
     // Assemble the 5d matrix
     Meooe5D(psi,tmp); 
-    std::cout << "Meooe Test replacement norm2 tmp = " <<norm2(tmp)<<std::endl;
 
-    int Ls=this->Ls;
+#if 0
+    std::cout << "Meooe Test replacement norm2 tmp = " <<norm2(tmp)<<std::endl;
     for(int s=0;s<Ls;s++){
       if ( s==0 ) {
 	//	tmp = bs psi[s] + cs[s] psi[s+1}
@@ -179,6 +181,8 @@ namespace QCD {
       }
     }
     std::cout << "Meooe Test replacement norm2 tmp old = " <<norm2(tmp)<<std::endl;
+#endif
+
     // Apply 4d dslash
     if ( psi.checkerboard == Odd ) {
       this->DhopEO(tmp,chi,DaggerNo);
@@ -199,8 +203,8 @@ namespace QCD {
     }
 
     Meooe5D(tmp,chi); 
+#if 0
     std::cout << "Meooe Test replacement norm2 chi new = " <<norm2(chi)<<std::endl;
-
     // Assemble the 5d matrix
     int Ls=this->Ls;
     for(int s=0;s<Ls;s++){
@@ -216,6 +220,7 @@ namespace QCD {
       }
     }
     std::cout << "Meooe Test replacement norm2 chi old = " <<norm2(chi)<<std::endl;
+#endif
 
   }
 
diff --git a/scripts/configure-commands b/scripts/configure-commands
index 4cb4b44f..d50da271 100755
--- a/scripts/configure-commands
+++ b/scripts/configure-commands
@@ -79,5 +79,8 @@ CXX=clang++ ../../configure --enable-simd=AVX CXXFLAGS="-mavx -O3 -I/opt/local/i
 clang-avx2-mpi)
 CXX=clang++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -I/opt/local/include/openmpi-mp/ -std=c++11" LDFLAGS=-L/opt/local/lib/openmpi-mp/ LIBS="-lmpi -lmpi_cxx -lgmp -lmpfr" --enable-comms=mpi
 ;;
+clang-avx2)
+CXX=clang++ ../../configure --enable-simd=AVX2 CXXFLAGS="-mavx2 -mfma -O3 -std=c++11" LDFLAGS="-L/usr/local/lib/" LIBS="-lgmp -lmpfr" --enable-comms=none
+;;
 esac
 echo -e $NORMAL
diff --git a/tests/Make.inc b/tests/Make.inc
index d5b926e6..111eed35 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,5 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_cheby Test_contfrac_cg Test_contfrac_even_odd Test_contfrac_force Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_force Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_hmc_EOWilsonFermionGauge Test_hmc_WilsonFermionGauge Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_partfrac_force Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd Test_wilson_force Test_wilson_force_phiMdagMphi Test_wilson_force_phiMphi
+bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_cheby Test_contfrac_cg Test_contfrac_even_odd Test_contfrac_force Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_force Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_gparity Test_hmc_EOWilsonFermionGauge Test_hmc_WilsonFermionGauge Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_partfrac_force Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd Test_wilson_force Test_wilson_force_phiMdagMphi Test_wilson_force_phiMphi
 
 
 Test_GaugeAction_SOURCES=Test_GaugeAction.cc
@@ -90,6 +90,10 @@ Test_gamma_SOURCES=Test_gamma.cc
 Test_gamma_LDADD=-lGrid
 
 
+Test_gparity_SOURCES=Test_gparity.cc
+Test_gparity_LDADD=-lGrid
+
+
 Test_hmc_EOWilsonFermionGauge_SOURCES=Test_hmc_EOWilsonFermionGauge.cc
 Test_hmc_EOWilsonFermionGauge_LDADD=-lGrid
 
diff --git a/tests/Test_gparity.cc b/tests/Test_gparity.cc
new file mode 100644
index 00000000..46c019c6
--- /dev/null
+++ b/tests/Test_gparity.cc
@@ -0,0 +1,73 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+template<class d>
+struct scal {
+  d internal;
+};
+
+  Gamma::GammaMatrix Gmu [] = {
+    Gamma::GammaX,
+    Gamma::GammaY,
+    Gamma::GammaZ,
+    Gamma::GammaT
+  };
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  const int Ls=4;
+  const int L=4;
+  std::vector<int> latt_2f(Nd,L);
+  std::vector<int> latt_1f(Nd,L); latt_1f[0] = 2L;
+
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi(); //node layout
+
+  GridCartesian         * UGrid_1f   = SpaceTimeGrid::makeFourDimGrid(latt_1f, simd_layout, mpi_layout);
+  GridRedBlackCartesian * UrbGrid_1f = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid_1f);
+  GridCartesian         * FGrid_1f   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid_1f);
+  GridRedBlackCartesian * FrbGrid_1f = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid_1f);
+
+  std::vector<int> seeds4({1,2,3,4});
+  std::vector<int> seeds5({5,6,7,8});
+  GridParallelRNG          RNG5_1f(FGrid_1f);  RNG5_1f.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4_1f(UGrid_1f);  RNG4_1f.SeedFixedIntegers(seeds4);
+
+  LatticeFermion    src_1f(FGrid_1f); random(RNG5_1f,src_1f);
+  LatticeFermion result_1f(FGrid_1f); result_1f=zero;
+  LatticeGaugeField Umu_1f(UGrid_1f); random(RNG4_1f,Umu_1f);
+
+  //Coordinate grid for reference
+  LatticeInteger xcoor_1f(UGrid_1f);
+  LatticeCoordinate(xcoor_1f,0);
+
+  //Copy-conjugate the gauge field
+  //First C-shift the lattice by Lx/2
+  {
+    LatticeGaugeField Umu_shift = conjugate( Cshift(Umu_1f,0,L) );
+    Umu_1f = where( xcoor_1f >= Integer(L), Umu_shift, Umu_1f );
+  }
+
+  //Make the gauge field antiperiodic in x-direction
+  Umu_1f = where(xcoor_1f == Integer(L-1), -Umu_1f, Umu_1f);
+    
+  RealD mass=0.1;
+  RealD M5=1.8;
+  DomainWallFermionR Ddwf(Umu_1f,*FGrid_1f,*FrbGrid_1f,*UGrid_1f,*UrbGrid_1f,mass,M5);
+
+  LatticeFermion    src_o_1f(FrbGrid_1f);
+  LatticeFermion result_o_1f(FrbGrid_1f);
+  pickCheckerboard(Odd,src_o_1f,src_1f);
+  result_o_1f=zero;
+
+  SchurDiagMooeeOperator<DomainWallFermionR,LatticeFermion> HermOpEO(Ddwf);
+  ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
+  CG(HermOpEO,src_o_1f,result_o_1f);
+
+  Grid_finalize();
+}

From 8d4c43327baeb8b205a3992b5a4cd625870918c0 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 13 Aug 2015 10:50:34 +0100
Subject: [PATCH 424/429] Some bug fixes for more complicated types introduced
 with gparity

---
 lib/qcd/spin/Dirac.h | 18 ++++++++++++++----
 1 file changed, 14 insertions(+), 4 deletions(-)

diff --git a/lib/qcd/spin/Dirac.h b/lib/qcd/spin/Dirac.h
index 65d8d6b5..8eb1bcf0 100644
--- a/lib/qcd/spin/Dirac.h
+++ b/lib/qcd/spin/Dirac.h
@@ -343,14 +343,19 @@ namespace QCD {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type 
 	{
 	  iVector<vtype,N> ret;
-	  ret._internal=G*arg._internal;
+	  for(int i=0;i<N;i++){
+	    ret._internal[i]=G*arg._internal[i];
+	  }
 	  return ret;
 	}
     template<class vtype,int N> inline auto operator * ( const Gamma &G,const iMatrix<vtype,N> &arg) ->
       typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type 
 	{
 	  iMatrix<vtype,N> ret;
-	  ret._internal=G*arg._internal;
+	  for(int i=0;i<N;i++){
+	  for(int j=0;j<N;j++){
+	    ret._internal[i][j]=G*arg._internal[i][j];
+	  }}
 	  return ret;
 	}
 
@@ -368,14 +373,19 @@ namespace QCD {
       typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type 
 	{
 	  iVector<vtype,N> ret;
-	  ret._internal=arg._internal*G;
+	  for(int i=0;i<N;i++){
+	    ret._internal=arg._internal[i]*G;
+	  }
 	  return ret;
 	}
     template<class vtype,int N> inline auto operator * (const iMatrix<vtype,N> &arg, const Gamma &G) ->
       typename std::enable_if<matchGridTensorIndex<iMatrix<vtype,N>,SpinorIndex>::notvalue,iMatrix<vtype,N> >::type 
 	{
 	  iMatrix<vtype,N> ret;
-	  ret._internal=arg._internal*G;
+	  for(int i=0;i<N;i++){
+	  for(int j=0;j<N;j++){
+	    ret._internal[i][j]=arg._internal[i][j]*G;
+	  }}
 	  return ret;
 	}
 

From 145b807ba2255caa3486a62f43a156649a36fabe Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 13 Aug 2015 10:51:01 +0100
Subject: [PATCH 425/429] Gparity improvements

---
 lib/qcd/spin/TwoSpinor.h | 64 ++++++++++++++++++++--------------------
 1 file changed, 32 insertions(+), 32 deletions(-)

diff --git a/lib/qcd/spin/TwoSpinor.h b/lib/qcd/spin/TwoSpinor.h
index 94ba9978..2637baba 100644
--- a/lib/qcd/spin/TwoSpinor.h
+++ b/lib/qcd/spin/TwoSpinor.h
@@ -341,7 +341,7 @@ namespace QCD {
   //////////
   // Xp
   //////////
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProjXp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spProjXp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -365,7 +365,7 @@ namespace QCD {
   {
     spReconXp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spReconXp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spReconXp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     for(int i=0;i<N;i++) {
       spReconXp(hspin._internal[i],fspin._internal[i]);
@@ -383,7 +383,7 @@ namespace QCD {
   {
     accumReconXp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumReconXp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void accumReconXp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     for(int i=0;i<N;i++) {
       accumReconXp(hspin._internal[i],fspin._internal[i]);
@@ -407,7 +407,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjXm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProjXm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spProjXm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -429,7 +429,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconXm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spReconXm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spReconXm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -450,7 +450,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconXm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumReconXm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void accumReconXm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -476,7 +476,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjYp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProjYp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spProjYp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -498,7 +498,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconYp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spReconYp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spReconYp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -519,7 +519,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconYp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumReconYp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void accumReconYp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -544,7 +544,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjYm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProjYm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spProjYm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -566,7 +566,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconYm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spReconYm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spReconYm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,const iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -587,7 +587,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconYm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumReconYm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void accumReconYm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -612,7 +612,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjZp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProjZp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spProjZp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -634,7 +634,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconZp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spReconZp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spReconZp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -655,7 +655,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconZp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumReconZp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void accumReconZp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -680,7 +680,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjZm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProjZm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spProjZm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -702,7 +702,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconZm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spReconZm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spReconZm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -723,7 +723,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconZm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumReconZm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void accumReconZm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -748,7 +748,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjTp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProjTp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spProjTp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -770,7 +770,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconTp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spReconTp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spReconTp (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -791,7 +791,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconTp(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumReconTp (iVector<rtype,N> &hspin, const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void accumReconTp (iVector<rtype,N> &hspin, const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -816,7 +816,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProjTm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProjTm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spProjTm (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -838,7 +838,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spReconTm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spReconTm (iVector<rtype,N> &hspin, const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spReconTm (iVector<rtype,N> &hspin, const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -859,7 +859,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumReconTm(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumReconTm (iVector<rtype,N> &hspin, const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void accumReconTm (iVector<rtype,N> &hspin, const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -884,7 +884,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spProj5p(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProj5p (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spProj5p (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -905,7 +905,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spRecon5p(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spRecon5p (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spRecon5p (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -926,7 +926,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumRecon5p(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumRecon5p (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void accumRecon5p (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -950,7 +950,7 @@ namespace QCD {
     spProj5p(hspin._internal,fspin._internal);
   }
   //  template<class vtype,int N> strong_inline void fspProj5p (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
-  template<class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProj5p (iVector<vtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spProj5p (iVector<vtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -976,7 +976,7 @@ namespace QCD {
   {
     spProj5m(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<rtype> = 0> strong_inline void spProj5m (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<rtype,N> > = 0> strong_inline void spProj5m (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     for(int i=0;i<N;i++) {
       spProj5m(hspin._internal[i],fspin._internal[i]);
@@ -995,7 +995,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     spRecon5m(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spRecon5m (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spRecon5m (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -1016,7 +1016,7 @@ namespace QCD {
     //typename std::enable_if<matchGridTensorIndex<iScalar<vtype>,SpinorIndex>::notvalue,iScalar<vtype> >::type *temp;
     accumRecon5m(hspin._internal,fspin._internal);
   }
-  template<class rtype,class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void accumRecon5m (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class rtype,class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void accumRecon5m (iVector<rtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {
@@ -1041,7 +1041,7 @@ namespace QCD {
     spProj5m(hspin._internal,fspin._internal);
   }
   //  template<class vtype,int N> strong_inline void fspProj5m (iVector<vtype,N> &hspin,iVector<vtype,N> &fspin)
-  template<class vtype,int N,IfNotSpinor<vtype> = 0> strong_inline void spProj5m (iVector<vtype,N> &hspin,const iVector<vtype,N> &fspin)
+  template<class vtype,int N,IfNotSpinor<iVector<vtype,N> > = 0> strong_inline void spProj5m (iVector<vtype,N> &hspin,const iVector<vtype,N> &fspin)
   {
     //typename std::enable_if<matchGridTensorIndex<iVector<vtype,N>,SpinorIndex>::notvalue,iVector<vtype,N> >::type *temp;
     for(int i=0;i<N;i++) {

From 2c216a42f93af770055159ecaba7e06012089627 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Thu, 13 Aug 2015 10:51:29 +0100
Subject: [PATCH 426/429] Gamma5 mult direct

---
 lib/qcd/utils/LinalgUtils.h | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/lib/qcd/utils/LinalgUtils.h b/lib/qcd/utils/LinalgUtils.h
index ef85e728..9ad2786e 100644
--- a/lib/qcd/utils/LinalgUtils.h
+++ b/lib/qcd/utils/LinalgUtils.h
@@ -32,10 +32,11 @@ void ag5xpby_ssp(Lattice<vobj> &z,RealD a,const Lattice<vobj> &x,RealD b,const L
   conformable(x,z);
   GridBase *grid=x._grid;
   int Ls = grid->_rdimensions[0];
+  Gamma G5(Gamma::Gamma5);
 PARALLEL_FOR_LOOP
   for(int ss=0;ss<grid->oSites();ss+=Ls){ // adds Ls
     vobj tmp;
-    multGamma5(tmp(),a*x._odata[ss+s]());
+    tmp = G5*x._odata[ss+s]*a;
     tmp = tmp + b*y._odata[ss+sp];
     vstream(z._odata[ss+s],tmp);
   }
@@ -49,10 +50,11 @@ void axpbg5y_ssp(Lattice<vobj> &z,RealD a,const Lattice<vobj> &x,RealD b,const L
   conformable(x,z);
   GridBase *grid=x._grid;
   int Ls = grid->_rdimensions[0];
+  Gamma G5(Gamma::Gamma5);
 PARALLEL_FOR_LOOP
   for(int ss=0;ss<grid->oSites();ss+=Ls){ // adds Ls
     vobj tmp;
-    multGamma5(tmp(),b*y._odata[ss+sp]());
+    tmp = G5*y._odata[ss+sp]*b;
     tmp = tmp + a*x._odata[ss+s];
     vstream(z._odata[ss+s],tmp);
   }
@@ -66,12 +68,13 @@ void ag5xpbg5y_ssp(Lattice<vobj> &z,RealD a,const Lattice<vobj> &x,RealD b,const
   conformable(x,z);
   GridBase *grid=x._grid;
   int Ls = grid->_rdimensions[0];
+  Gamma G5(Gamma::Gamma5);
 PARALLEL_FOR_LOOP
   for(int ss=0;ss<grid->oSites();ss+=Ls){ // adds Ls
     vobj tmp1;
     vobj tmp2;
     tmp1 = a*x._odata[ss+s]+b*y._odata[ss+sp];
-    multGamma5(tmp2(),tmp1());
+    tmp2 = G5*tmp1;
     vstream(z._odata[ss+s],tmp2);
   }
 }
@@ -117,12 +120,13 @@ void G5R5(Lattice<vobj> &z,const Lattice<vobj> &x)
   z.checkerboard = x.checkerboard;
   conformable(x,z);
   int Ls = grid->_rdimensions[0];
+  Gamma G5(Gamma::Gamma5);
 PARALLEL_FOR_LOOP
   for(int ss=0;ss<grid->oSites();ss+=Ls){ // adds Ls
     vobj tmp;
     for(int s=0;s<Ls;s++){
       int sp = Ls-1-s;
-      multGamma5(tmp(),x._odata[ss+s]());
+      tmp = G5*x._odata[ss+s];
       vstream(z._odata[ss+sp],tmp);
     }
   }

From 028e2061e0be9fe84b51d8823b026a4acc2d86fc Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 14 Aug 2015 00:01:04 +0100
Subject: [PATCH 427/429] Gparity valence test now working. Interface in
 FermionOperator will change a lot in future

---
 lib/Stencil.h                             |  16 +-
 lib/algorithms/CoarsenedMatrix.h          |  20 +-
 lib/lattice/Lattice_peekpoke.h            |   2 +
 lib/qcd/QCD.h                             |   2 +-
 lib/qcd/action/Actions.h                  |  11 +
 lib/qcd/action/fermion/FermionOperator.h  |  66 ++--
 lib/qcd/action/fermion/WilsonCompressor.h |  75 ++--
 lib/qcd/action/fermion/WilsonFermion.cc   |   6 +-
 lib/qcd/action/fermion/WilsonFermion5D.cc |  11 +-
 lib/qcd/action/fermion/WilsonKernels.cc   | 399 ++++++++++------------
 lib/stencil/Stencil_common.cc             |  72 ++--
 tests/Make.inc                            |   6 +-
 tests/Test_gparity.cc                     | 237 +++++++++++--
 tests/Test_stencil.cc                     |  20 +-
 14 files changed, 572 insertions(+), 371 deletions(-)

diff --git a/lib/Stencil.h b/lib/Stencil.h
index 8529e73a..c6e51059 100644
--- a/lib/Stencil.h
+++ b/lib/Stencil.h
@@ -41,6 +41,12 @@
 
 namespace Grid {
   
+  struct StencilEntry { 
+    int _offset;
+    int _is_local;
+    int _permute;
+    int _around_the_world;
+  };
 
   class CartesianStencil { // Stencil runs along coordinate axes only; NO diagonal fill in.
   public:
@@ -58,9 +64,9 @@ namespace Grid {
       std::vector<int>                  _permute_type;
 
       // npoints x Osites() of these
-      std::vector<std::vector<int>    > _offsets;
-      std::vector<std::vector<int>    > _is_local;
-      std::vector<std::vector<int> >    _permute;
+      std::vector<std::vector<StencilEntry> > _entries;
+
+      inline StencilEntry * GetEntry(int &ptype,int point,int osite) { ptype = _permute_type[point]; return & _entries[point][osite]; }
 
       int _unified_buffer_size;
       int _request_count;
@@ -77,8 +83,8 @@ namespace Grid {
       // Can this be avoided with simpler coding of comms?
       void Local     (int point, int dimension,int shift,int cbmask);
       void Comms     (int point, int dimension,int shift,int cbmask);
-      void CopyPlane(int point, int dimension,int lplane,int rplane,int cbmask,int permute);
-      void ScatterPlane (int point,int dimension,int plane,int cbmask,int offset);
+      void CopyPlane(int point, int dimension,int lplane,int rplane,int cbmask,int permute,int wrap);
+      void ScatterPlane (int point,int dimension,int plane,int cbmask,int offset,int wrap);
 
       // Could allow a functional munging of the halo to another type during the comms.
       // this could implement the 16bit/32bit/64bit compression.
diff --git a/lib/algorithms/CoarsenedMatrix.h b/lib/algorithms/CoarsenedMatrix.h
index c4b039f1..a854bba7 100644
--- a/lib/algorithms/CoarsenedMatrix.h
+++ b/lib/algorithms/CoarsenedMatrix.h
@@ -193,20 +193,18 @@ PARALLEL_FOR_LOOP
       for(int ss=0;ss<Grid()->oSites();ss++){
         siteVector res = zero;
 	siteVector nbr;
-	int offset,local,perm,ptype;
-
+	int ptype;
+	StencilEntry *SE;
 	for(int point=0;point<geom.npoint;point++){
-	  offset = Stencil._offsets [point][ss];
-	  local  = Stencil._is_local[point][ss];
-	  perm   = Stencil._permute [point][ss];
-	  ptype  = Stencil._permute_type[point];
+
+	  SE=Stencil.GetEntry(ptype,point,ss);
 	  
-	  if(local&&perm) { 
-	    permute(nbr,in._odata[offset],ptype);
-	  } else if(local) { 
-	    nbr = in._odata[offset];
+	  if(SE->_is_local&&SE->_permute) { 
+	    permute(nbr,in._odata[SE->_offset],ptype);
+	  } else if(SE->_is_local) { 
+	    nbr = in._odata[SE->_offset];
 	  } else {
-	    nbr = comm_buf[offset];
+	    nbr = comm_buf[SE->_offset];
 	  }
 	  res = res + A[point]._odata[ss]*nbr;
 	}
diff --git a/lib/lattice/Lattice_peekpoke.h b/lib/lattice/Lattice_peekpoke.h
index e60618c5..8ac952ea 100644
--- a/lib/lattice/Lattice_peekpoke.h
+++ b/lib/lattice/Lattice_peekpoke.h
@@ -14,6 +14,7 @@ namespace Grid {
        auto PeekIndex(const Lattice<vobj> &lhs,int i) -> Lattice<decltype(peekIndex<Index>(lhs._odata[0],i))>
     {
       Lattice<decltype(peekIndex<Index>(lhs._odata[0],i))> ret(lhs._grid);
+      ret.checkerboard=lhs.checkerboard;
 PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){
 	  ret._odata[ss] = peekIndex<Index>(lhs._odata[ss],i);
@@ -24,6 +25,7 @@ PARALLEL_FOR_LOOP
        auto PeekIndex(const Lattice<vobj> &lhs,int i,int j) -> Lattice<decltype(peekIndex<Index>(lhs._odata[0],i,j))>
     {
       Lattice<decltype(peekIndex<Index>(lhs._odata[0],i,j))> ret(lhs._grid);
+      ret.checkerboard=lhs.checkerboard;
 PARALLEL_FOR_LOOP
         for(int ss=0;ss<lhs._grid->oSites();ss++){
 	  ret._odata[ss] = peekIndex<Index>(lhs._odata[ss],i,j);
diff --git a/lib/qcd/QCD.h b/lib/qcd/QCD.h
index 16920f2b..1286c6bd 100644
--- a/lib/qcd/QCD.h
+++ b/lib/qcd/QCD.h
@@ -36,7 +36,7 @@ namespace QCD {
     template<typename T> struct isSpinor {
       static const bool value = (SpinorIndex==T::TensorLevel);
     };
-    template <typename T> using IfSpinor    = Invoke<std::enable_if<isSpinor<T>::value,int> > ;
+    template <typename T> using IfSpinor    = Invoke<std::enable_if< isSpinor<T>::value,int> > ;
     template <typename T> using IfNotSpinor = Invoke<std::enable_if<!isSpinor<T>::value,int> > ;
 
     // ChrisK very keen to add extra space for Gparity doubling.
diff --git a/lib/qcd/action/Actions.h b/lib/qcd/action/Actions.h
index 9b119858..9976c670 100644
--- a/lib/qcd/action/Actions.h
+++ b/lib/qcd/action/Actions.h
@@ -46,6 +46,8 @@
 ////////////////////////////////////////////////////////////////////////////////////////////////////
 
 #define FermOpTemplateInstantiate(A) \
+  template class A<GparityWilsonImplF>;		\
+  template class A<GparityWilsonImplD>;		\
   template class A<WilsonImplF>;		\
   template class A<WilsonImplD>;
 
@@ -131,6 +133,15 @@ typedef OverlapWilsonPartialFractionTanhFermion<WilsonImplD> OverlapWilsonPartia
 typedef OverlapWilsonPartialFractionZolotarevFermion<WilsonImplR> OverlapWilsonPartialFractionZolotarevFermionR;
 typedef OverlapWilsonPartialFractionZolotarevFermion<WilsonImplF> OverlapWilsonPartialFractionZolotarevFermionF;
 typedef OverlapWilsonPartialFractionZolotarevFermion<WilsonImplD> OverlapWilsonPartialFractionZolotarevFermionD;
+
+// Gparity cases; partial list until tested
+typedef WilsonFermion<GparityWilsonImplR>     GparityWilsonFermionR;
+typedef WilsonFermion<GparityWilsonImplF>     GparityWilsonFermionF;
+typedef WilsonFermion<GparityWilsonImplD>     GparityWilsonFermionD;
+typedef DomainWallFermion<GparityWilsonImplR> GparityDomainWallFermionR;
+typedef DomainWallFermion<GparityWilsonImplF> GparityDomainWallFermionF;
+typedef DomainWallFermion<GparityWilsonImplD> GparityDomainWallFermionD;
+
   }}
 ///////////////////////////////////////////////////////////////////////////////
 // G5 herm -- this has to live in QCD since dirac matrix is not in the broader sector of code
diff --git a/lib/qcd/action/fermion/FermionOperator.h b/lib/qcd/action/fermion/FermionOperator.h
index 75641b83..e1f65a27 100644
--- a/lib/qcd/action/fermion/FermionOperator.h
+++ b/lib/qcd/action/fermion/FermionOperator.h
@@ -37,12 +37,10 @@ namespace Grid {
 
       typedef WilsonCompressor<SiteHalfSpinor,SiteSpinor> Compressor;
 
-      // provide the multiply by link that is differentiated between Gparity (with flavour index) and 
-      // non-Gparity
-      static inline void multLink(SiteHalfSpinor &phi,const SiteDoubledGaugeField &U,const SiteHalfSpinor &chi,int mu){
+      // provide the multiply by link that is differentiated between Gparity (with flavour index) and non-Gparity
+      static inline void multLink(SiteHalfSpinor &phi,const SiteDoubledGaugeField &U,const SiteHalfSpinor &chi,int mu,StencilEntry *SE){
         mult(&phi(),&U(mu),&chi());
       }
-
       static inline void DoubleStore(GridBase *GaugeGrid,DoubledGaugeField &Uds,const GaugeField &Umu)
       {
         conformable(Uds._grid,GaugeGrid);
@@ -55,6 +53,11 @@ namespace Grid {
 	  PokeIndex<LorentzIndex>(Uds,U,mu+4);
 	}
       }
+      static inline void InsertForce(GaugeField &mat,const FermionField &Btilde,const FermionField &A,int mu){
+	GaugeLinkField link(mat._grid);
+	link = TraceIndex<SpinIndex>(outerProduct(Btilde,A)); 
+	PokeIndex<LorentzIndex>(mat,link,mu);
+      }   
 
     };
 
@@ -62,8 +65,6 @@ namespace Grid {
     typedef WilsonImpl<vComplexF,Nc> WilsonImplF; // Float
     typedef WilsonImpl<vComplexD,Nc> WilsonImplD; // Double
 
-
-
     template<class S,int Nrepresentation=Nc>
     class GparityWilsonImpl { 
     public:
@@ -74,7 +75,7 @@ namespace Grid {
       template<typename vtype> using iImplHalfSpinor         = iVector<iVector<iVector<vtype, Nrepresentation>, Nhs>, Ngp >;
       template<typename vtype> using iImplGaugeField         = iVector<iScalar<iMatrix<vtype, Nrepresentation> >, Nd  >;
 
-      template<typename vtype> using iImplGaugeLink          = iScalar<iScalar<iScalar<iMatrix<vtype, Nrepresentation> > > >;
+      template<typename vtype> using iImplGaugeLink          = iScalar<iScalar<iMatrix<vtype, Nrepresentation> > >;
       template<typename vtype> using iImplDoubledGaugeField  = iVector<iVector<iScalar<iMatrix<vtype, Nrepresentation> >, Nds >, Ngp >;
     
       typedef iImplSpinor    <Simd>           SiteSpinor;
@@ -88,16 +89,26 @@ namespace Grid {
       typedef Lattice<SiteGaugeField>               GaugeField;
       typedef Lattice<SiteDoubledGaugeField> DoubledGaugeField;
 
-      typedef GparityWilsonCompressor<SiteHalfSpinor,SiteSpinor> Compressor;
+    //      typedef GparityWilsonCompressor<SiteHalfSpinor,SiteSpinor> Compressor;
+    typedef WilsonCompressor<SiteHalfSpinor,SiteSpinor> Compressor;
 
       // provide the multiply by link that is differentiated between Gparity (with flavour index) and 
       // non-Gparity
-      static inline void multLink(SiteHalfSpinor &phi,const SiteDoubledGaugeField &U,const SiteHalfSpinor &chi,int mu){
-	for(int f=0;f<Ngp;f++){
-	  mult(&phi(f),&U(f)(mu),&chi(f));
-	}
+    static inline void multLink(SiteHalfSpinor &phi,const SiteDoubledGaugeField &U,const SiteHalfSpinor &chi,int mu,StencilEntry *SE){
+      // FIXME; need to be more careful. If this is a simd direction we are still stuffed
+      if ( SE->_around_the_world && ((mu==Xp)||(mu==Xm)) ) {
+	mult(&phi(0),&U(0)(mu),&chi(1));
+	mult(&phi(1),&U(1)(mu),&chi(0));
+      } else { 
+	mult(&phi(0),&U(0)(mu),&chi(0));
+	mult(&phi(1),&U(1)(mu),&chi(1));
       }
+    }
 
+      static inline void InsertForce(GaugeField &mat,const FermionField &Btilde,const FermionField &A,int mu){
+	// Fixme
+	return;
+      }
       static inline void DoubleStore(GridBase *GaugeGrid,DoubledGaugeField &Uds,const GaugeField &Umu)
       {
         conformable(Uds._grid,GaugeGrid);
@@ -116,44 +127,51 @@ namespace Grid {
 	  LatticeCoordinate(coor,mu);
 
   	  U     = PeekIndex<LorentzIndex>(Umu,mu);
-	  Uconj = conj(U);
+	  Uconj = conjugate(U);
 
 	  int neglink = GaugeGrid->GlobalDimensions()[mu]-1;
 
 	  if ( gpdirs[mu] ) { 
 	    Uconj = where(coor==neglink,-Uconj,Uconj);
 	  }
+
 PARALLEL_FOR_LOOP
 	  for(auto ss=U.begin();ss<U.end();ss++){
 	    Uds[ss](0)(mu) = U[ss]();
 	    Uds[ss](1)(mu) = Uconj[ss]();
 	  }
 
-
-	  U     = adj(Cshift(U,mu,-1));      // correct except for spanning the boundary
+	  U     = adj(Cshift(U    ,mu,-1));      // correct except for spanning the boundary
 	  Uconj = adj(Cshift(Uconj,mu,-1));
 
-	  Utmp = where(coor==0,U,Uconj);
-PARALLEL_FOR_LOOP
-	  for(auto ss=U.begin();ss<U.end();ss++){
-	    Uds[ss](1)(mu) = Utmp[ss]();
+	  Utmp = U;
+	  if ( gpdirs[mu] ) { 
+	    Utmp = where(coor==0,Uconj,Utmp);
 	  }
 
-	  Utmp = where(coor==0,Uconj,U);
 PARALLEL_FOR_LOOP
 	  for(auto ss=U.begin();ss<U.end();ss++){
-	    Uds[ss](0)(mu) = Utmp[ss]();
+	    Uds[ss](0)(mu+4) = Utmp[ss]();
 	  }
 
+	  Utmp = Uconj;
+	  if ( gpdirs[mu] ) { 
+	    Utmp = where(coor==0,U,Utmp);
+	  }
+
+PARALLEL_FOR_LOOP
+	  for(auto ss=U.begin();ss<U.end();ss++){
+	    Uds[ss](1)(mu+4) = Utmp[ss]();
+	  }
 
 	}
       }
 
     };
 
-    typedef WilsonImpl<vComplex,Nc>  WilsonImplR; // Real.. whichever prec
-    typedef WilsonImpl<vComplexF,Nc> WilsonImplF; // Float
-    typedef WilsonImpl<vComplexD,Nc> WilsonImplD; // Double
+    typedef GparityWilsonImpl<vComplex,Nc>  GparityWilsonImplR; // Real.. whichever prec
+    typedef GparityWilsonImpl<vComplexF,Nc> GparityWilsonImplF; // Float
+    typedef GparityWilsonImpl<vComplexD,Nc> GparityWilsonImplD; // Double
 
 
     
diff --git a/lib/qcd/action/fermion/WilsonCompressor.h b/lib/qcd/action/fermion/WilsonCompressor.h
index c879e9f1..f2aae8aa 100644
--- a/lib/qcd/action/fermion/WilsonCompressor.h
+++ b/lib/qcd/action/fermion/WilsonCompressor.h
@@ -63,6 +63,7 @@ namespace QCD {
     }
   };
 
+  /*
 
   template<class SiteHalfSpinor,class SiteSpinor>
     class GparityWilsonCompressor : public WilsonCompressor<SiteHalfSpinor,SiteSpinor>{
@@ -77,55 +78,65 @@ namespace QCD {
 
       const int Nsimd = grid->Nsimd();
 
-      int ocb=grid->CheckerBoardFromOindex(osite);
+      int checkered=grid->CheckerBoarded(dim);
+      int       ocb=grid->CheckerBoardFromOindex(osite);
 
-      SiteHalfSpinor tmp = spinproject(in); // spin projected
+      SiteHalfSpinor tmp = this->spinproject(in); // spin projected
 
       //////////////////////////////////////////////////////////////
       // Check whether we must flavour flip
       // do this if source is plane 0 on processor 0 in dimension dim
       //////////////////////////////////////////////////////////////
-      if ( (this->mu==Xp)||(this->mu==Yp)||(this->mu==Zp)||(this->mu==Tp) ) {      
+      int do_flip  = 0;
+      int flipicoor= 0;
+      if(Gbcs[this->mu]){
 
-	if ( Gbcs[this->mu] && (grid->_processor_coor[dim] == 0) && (plane==0) && (ocb==0) ) {
+	std::cout << "Applying Gparity BC's in direction "<<this->mu<<std::endl;
 
-	  std::vector<scalar_object>  flat(Nsimd);
-
-	  extract(tmp,flat);
-
-	  for(int i=0;i<Nsimd;i++) {
-	    std::vector<int> coor;
-
-	    grid->iCoorFromIindex(coor,i);
-
-	    if ( coor[dim]==0 ) {
-	      scalar_object stmp;
-	      stmp(0) = flat[i](1);
-	      stmp(1) = flat[i](0);
-	      flat[i] = stmp;
-	    }
+	if ( (this->mu==Xp)||(this->mu==Yp)||(this->mu==Zp)||(this->mu==Tp) ) {      
+	  if ( (grid->_processor_coor[dim] == 0) 
+	       && (plane==0) 
+	       && ( (!checkered)||(ocb==0) ) ) {
+	    do_flip=1;
+	    flipicoor=0;
 	  }
+	}
+	if ( (this->mu==Xm)||(this->mu==Ym)||(this->mu==Zm)||(this->mu==Tm) ) {      
+	  if ( (grid->_processor_coor[dim] == (grid->_processors[dim]-1) ) 
+	    && (plane==grid->_rdimensions[dim]-1) 
+            && ( (!checkered)||(ocb==1) ) ) {
+	    do_flip=1;
+	    flipicoor=grid->_simd_layout[dim]-1;
+	  }
+	}
+      }
+      if ( do_flip ) {
 
-	  merge(tmp,flat);
+	std::cout << "Applying Gparity BC's in direction "<<this->mu<< " osite " << osite << " plane "<<plane <<std::endl;
 
-	} // coor & bc guard
-      } // shift guard
+	std::vector<scalar_object>  flat(Nsimd);
+	std::vector<int> coor;
+
+	extract(tmp,flat);
+	for(int i=0;i<Nsimd;i++) {
+	  grid->iCoorFromIindex(coor,i);
+	  if ( coor[dim]==flipicoor ) {
+	    scalar_object stmp;
+	    stmp(0) = flat[i](1);
+	    stmp(1) = flat[i](0);
+	    flat[i] = stmp;
+	  }
+	}
+	merge(tmp,flat);
+
+      }
 
       return tmp;
     }
 
-    SiteHalfSpinor flavourflip(const SiteHalfSpinor &in) {
-
-      SiteHalfSpinor ret;
-      for(int f=0;f<Ngp;f++){
-	ret(0) = in(1);
-	ret(1) = in(0);
-      }
-      return ret;
-    }
-
   };
 
+  */
 
 }} // namespace close
 #endif
diff --git a/lib/qcd/action/fermion/WilsonFermion.cc b/lib/qcd/action/fermion/WilsonFermion.cc
index 260e3851..ea8c2209 100644
--- a/lib/qcd/action/fermion/WilsonFermion.cc
+++ b/lib/qcd/action/fermion/WilsonFermion.cc
@@ -117,7 +117,6 @@ namespace QCD {
     
     Compressor compressor(dag);
     
-    GaugeLinkField tmp(B._grid);
     FermionField Btilde(B._grid);
     
     st.HaloExchange<SiteSpinor,SiteHalfSpinor,Compressor>(B,comm_buf,compressor);
@@ -141,9 +140,8 @@ PARALLEL_FOR_LOOP
       //////////////////////////////////////////////////
       // spin trace outer product
       //////////////////////////////////////////////////
-      tmp = TraceIndex<SpinIndex>(outerProduct(Btilde,A)); 
-      PokeIndex<LorentzIndex>(mat,tmp,mu);
-      
+      Impl::InsertForce(mat,Btilde,A,mu);
+
     }
   }
   
diff --git a/lib/qcd/action/fermion/WilsonFermion5D.cc b/lib/qcd/action/fermion/WilsonFermion5D.cc
index f324db8d..10caad13 100644
--- a/lib/qcd/action/fermion/WilsonFermion5D.cc
+++ b/lib/qcd/action/fermion/WilsonFermion5D.cc
@@ -137,7 +137,6 @@ void WilsonFermion5D<Impl>::DerivInternal(CartesianStencil & st,
     ////////////////////////
     // Call the single hop
     ////////////////////////
-    tmp = zero;
 
 PARALLEL_FOR_LOOP
     for(int sss=0;sss<U._grid->oSites();sss++){
@@ -154,14 +153,18 @@ PARALLEL_FOR_LOOP
     // spin trace outer product
     ////////////////////////////
 
-	tmp[sU] = tmp[sU]+ traceIndex<SpinIndex>(
-		   outerProduct(Btilde[sF],Atilde[sF])); // ordering here
-
       }
 
     }
 
+    Impl::InsertForce(mat,Btilde,A,mu);
+    /*
+    tmp = zero;
+    for(int sss=0;sss<U._grid->oSites();sss++){
+      tmp[sU] = tmp[sU]+ traceIndex<SpinIndex>(outerProduct(Btilde[sF],Atilde[sF])); // ordering here
+    }
     PokeIndex<LorentzIndex>(mat,tmp,mu);
+    */
 
   }
 }
diff --git a/lib/qcd/action/fermion/WilsonKernels.cc b/lib/qcd/action/fermion/WilsonKernels.cc
index 6d558e03..d098d138 100644
--- a/lib/qcd/action/fermion/WilsonKernels.cc
+++ b/lib/qcd/action/fermion/WilsonKernels.cc
@@ -9,143 +9,116 @@ void WilsonKernels<Impl>::DiracOptDhopSite(CartesianStencil &st,DoubledGaugeFiel
 {
   SiteHalfSpinor  tmp;    
   SiteHalfSpinor  chi;    
-  SiteSpinor result;
   SiteHalfSpinor Uchi;
-  int offset,local,perm, ptype;
+  SiteSpinor result;
+  StencilEntry *SE;
+  int ptype;
 
   // Xp
-  int ss = sF;
-  offset = st._offsets [Xp][ss];
-  local  = st._is_local[Xp][ss];
-  perm   = st._permute[Xp][ss];
-  
-  ptype  = st._permute_type[Xp];
-  if ( local && perm ) {
-    spProjXp(tmp,in._odata[offset]);
+  SE=st.GetEntry(ptype,Xp,sF);
+  if ( SE->_is_local && SE->_permute ) {
+    spProjXp(tmp,in._odata[SE->_offset]);
     permute(chi,tmp,ptype);
-  } else if ( local ) {
-    spProjXp(chi,in._odata[offset]);
+  } else if ( SE->_is_local ) {
+    spProjXp(chi,in._odata[SE->_offset]);
   } else { 
-    chi=buf[offset];
+    chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Xp);
+  Impl::multLink(Uchi,U._odata[sU],chi,Xp,SE);
   spReconXp(result,Uchi);
     
   // Yp
-  offset = st._offsets [Yp][ss];
-  local  = st._is_local[Yp][ss];
-  perm   = st._permute[Yp][ss];
-  ptype  = st._permute_type[Yp];
-  if ( local && perm ) {
-    spProjYp(tmp,in._odata[offset]);
+  SE=st.GetEntry(ptype,Yp,sF);
+  if ( SE->_is_local && SE->_permute ) {
+    spProjYp(tmp,in._odata[SE->_offset]);
     permute(chi,tmp,ptype);
-  } else if ( local ) {
-    spProjYp(chi,in._odata[offset]);
+  } else if ( SE->_is_local ) {
+    spProjYp(chi,in._odata[SE->_offset]);
   } else { 
-    chi=buf[offset];
+    chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Yp);
+  Impl::multLink(Uchi,U._odata[sU],chi,Yp,SE);
   accumReconYp(result,Uchi);
 
   // Zp
-  offset = st._offsets [Zp][ss];
-  local  = st._is_local[Zp][ss];
-  perm   = st._permute[Zp][ss];
-  ptype  = st._permute_type[Zp];
-  if ( local && perm ) {
-    spProjZp(tmp,in._odata[offset]);
+  SE=st.GetEntry(ptype,Zp,sF);
+  if (  SE->_is_local && SE->_permute ) {
+    spProjZp(tmp,in._odata[SE->_offset]);
     permute(chi,tmp,ptype);
-  } else if ( local ) {
-    spProjZp(chi,in._odata[offset]);
+  } else if ( SE->_is_local ) {
+    spProjZp(chi,in._odata[SE->_offset]);
   } else { 
-    chi=buf[offset];
+    chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Zp);
+  Impl::multLink(Uchi,U._odata[sU],chi,Zp,SE);
   accumReconZp(result,Uchi);
 
   // Tp
-  offset = st._offsets [Tp][ss];
-  local  = st._is_local[Tp][ss];
-  perm   = st._permute[Tp][ss];
-  ptype  = st._permute_type[Tp];
-  if ( local && perm ) {
-    spProjTp(tmp,in._odata[offset]);
+  SE=st.GetEntry(ptype,Tp,sF);
+  if (  SE->_is_local && SE->_permute ) {
+    spProjTp(tmp,in._odata[SE->_offset]);
     permute(chi,tmp,ptype);
-  } else if ( local ) {
-    spProjTp(chi,in._odata[offset]);
+  } else if ( SE->_is_local ) {
+    spProjTp(chi,in._odata[SE->_offset]);
   } else { 
-    chi=buf[offset];
+    chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Tp);
+  Impl::multLink(Uchi,U._odata[sU],chi,Tp,SE);
   accumReconTp(result,Uchi);
 
   // Xm
-  offset = st._offsets [Xm][ss];
-  local  = st._is_local[Xm][ss];
-  perm   = st._permute[Xm][ss];
-  ptype  = st._permute_type[Xm];
-  
-  if ( local && perm ) {
-    spProjXm(tmp,in._odata[offset]);
+  SE=st.GetEntry(ptype,Xm,sF);
+  if (  SE->_is_local && SE->_permute ) {
+    spProjXm(tmp,in._odata[SE->_offset]);
     permute(chi,tmp,ptype);
-  } else if ( local ) {
-    spProjXm(chi,in._odata[offset]);
+  } else if ( SE->_is_local ) {
+    spProjXm(chi,in._odata[SE->_offset]);
   } else { 
-    chi=buf[offset];
+    chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Xm);
+  Impl::multLink(Uchi,U._odata[sU],chi,Xm,SE);
   accumReconXm(result,Uchi);
   
   // Ym
-  offset = st._offsets [Ym][ss];
-  local  = st._is_local[Ym][ss];
-  perm   = st._permute[Ym][ss];
-  ptype  = st._permute_type[Ym];
-  
-  if ( local && perm ) {
-    spProjYm(tmp,in._odata[offset]);
+  SE=st.GetEntry(ptype,Ym,sF);
+  if (  SE->_is_local && SE->_permute ) {
+    spProjYm(tmp,in._odata[SE->_offset]);
     permute(chi,tmp,ptype);
-  } else if ( local ) {
-    spProjYm(chi,in._odata[offset]);
+  } else if ( SE->_is_local ) {
+    spProjYm(chi,in._odata[SE->_offset]);
   } else { 
-    chi=buf[offset];
+    chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Ym);
+  Impl::multLink(Uchi,U._odata[sU],chi,Ym,SE);
   accumReconYm(result,Uchi);
   
   // Zm
-  offset = st._offsets [Zm][ss];
-  local  = st._is_local[Zm][ss];
-  perm   = st._permute[Zm][ss];
-  ptype  = st._permute_type[Zm];
-  if ( local && perm ) {
-    spProjZm(tmp,in._odata[offset]);
+  SE=st.GetEntry(ptype,Zm,sF);
+  if (  SE->_is_local && SE->_permute ) {
+    spProjZm(tmp,in._odata[SE->_offset]);
     permute(chi,tmp,ptype);
-  } else if ( local ) {
-    spProjZm(chi,in._odata[offset]);
+  } else if ( SE->_is_local ) {
+    spProjZm(chi,in._odata[SE->_offset]);
   } else { 
-    chi=buf[offset];
+    chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Zm);
+  Impl::multLink(Uchi,U._odata[sU],chi,Zm,SE);
   accumReconZm(result,Uchi);
 
   // Tm
-  offset = st._offsets [Tm][ss];
-  local  = st._is_local[Tm][ss];
-  perm   = st._permute[Tm][ss];
-  ptype  = st._permute_type[Tm];
-  if ( local && perm ) {
-    spProjTm(tmp,in._odata[offset]);
+  SE=st.GetEntry(ptype,Tm,sF);
+  if (  SE->_is_local && SE->_permute ) {
+    spProjTm(tmp,in._odata[SE->_offset]);
     permute(chi,tmp,ptype);
-  } else if ( local ) {
-    spProjTm(chi,in._odata[offset]);
+  } else if ( SE->_is_local ) {
+    spProjTm(chi,in._odata[SE->_offset]);
   } else { 
-    chi=buf[offset];
+    chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Tm);
+  Impl::multLink(Uchi,U._odata[sU],chi,Tm,SE);
   accumReconTm(result,Uchi);
 
-  vstream(out._odata[ss],result*(-0.5));
+  vstream(out._odata[sF],result*(-0.5));
 };
 
 template<class Impl> 
@@ -157,141 +130,114 @@ void WilsonKernels<Impl>::DiracOptDhopSiteDag(CartesianStencil &st,DoubledGaugeF
   SiteHalfSpinor  chi;    
   SiteSpinor result;
   SiteHalfSpinor Uchi;
-  int offset,local,perm, ptype;
+  StencilEntry *SE;
+  int ptype;
 
   // Xp
-  int ss=sF;
-  offset = st._offsets [Xm][ss];
-  local  = st._is_local[Xm][ss];
-  perm   = st._permute[Xm][ss];
-
-  ptype  = st._permute_type[Xm];
-  if ( local && perm ) {
-    spProjXp(tmp,in._odata[offset]);
+  SE=st.GetEntry(ptype,Xm,sF);
+  if (  SE->_is_local && SE->_permute ) {
+    spProjXp(tmp,in._odata[SE->_offset]);
     permute(chi,tmp,ptype);
-  } else if ( local ) {
-    spProjXp(chi,in._odata[offset]);
+  } else if ( SE->_is_local ) {
+    spProjXp(chi,in._odata[SE->_offset]);
   } else { 
-    chi=buf[offset];
+    chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Xm);
+  Impl::multLink(Uchi,U._odata[sU],chi,Xm,SE);
   spReconXp(result,Uchi);
 
   // Yp
-  offset = st._offsets [Ym][ss];
-  local  = st._is_local[Ym][ss];
-  perm   = st._permute[Ym][ss];
-  ptype  = st._permute_type[Ym];
-  if ( local && perm ) {
-    spProjYp(tmp,in._odata[offset]);
+  SE=st.GetEntry(ptype,Ym,sF);
+  if (  SE->_is_local && SE->_permute ) {
+    spProjYp(tmp,in._odata[SE->_offset]);
     permute(chi,tmp,ptype);
-  } else if ( local ) {
-    spProjYp(chi,in._odata[offset]);
+  } else if ( SE->_is_local ) {
+    spProjYp(chi,in._odata[SE->_offset]);
   } else { 
-    chi=buf[offset];
+    chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Ym);
+  Impl::multLink(Uchi,U._odata[sU],chi,Ym,SE);
   accumReconYp(result,Uchi);
   
   // Zp
-  offset = st._offsets [Zm][ss];
-  local  = st._is_local[Zm][ss];
-  perm   = st._permute[Zm][ss];
-  ptype  = st._permute_type[Zm];
-  if ( local && perm ) {
-    spProjZp(tmp,in._odata[offset]);
+  SE=st.GetEntry(ptype,Zm,sF);
+  if (  SE->_is_local && SE->_permute ) {
+    spProjZp(tmp,in._odata[SE->_offset]);
     permute(chi,tmp,ptype);
-  } else if ( local ) {
-    spProjZp(chi,in._odata[offset]);
+  } else if ( SE->_is_local ) {
+    spProjZp(chi,in._odata[SE->_offset]);
   } else { 
-    chi=buf[offset];
+    chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Zm);
+  Impl::multLink(Uchi,U._odata[sU],chi,Zm,SE);
   accumReconZp(result,Uchi);
   
   // Tp
-  offset = st._offsets [Tm][ss];
-  local  = st._is_local[Tm][ss];
-  perm   = st._permute[Tm][ss];
-  ptype  = st._permute_type[Tm];
-  if ( local && perm ) {
-    spProjTp(tmp,in._odata[offset]);
+  SE=st.GetEntry(ptype,Tm,sF);
+  if (  SE->_is_local && SE->_permute ) {
+    spProjTp(tmp,in._odata[SE->_offset]);
     permute(chi,tmp,ptype);
-  } else if ( local ) {
-    spProjTp(chi,in._odata[offset]);
+  } else if ( SE->_is_local ) {
+    spProjTp(chi,in._odata[SE->_offset]);
   } else { 
-    chi=buf[offset];
+    chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Tm);
+  Impl::multLink(Uchi,U._odata[sU],chi,Tm,SE);
   accumReconTp(result,Uchi);
   
   // Xm
-  offset = st._offsets [Xp][ss];
-  local  = st._is_local[Xp][ss];
-  perm   = st._permute[Xp][ss];
-  ptype  = st._permute_type[Xp];
-
-  if ( local && perm ) {
-    spProjXm(tmp,in._odata[offset]);
+  SE=st.GetEntry(ptype,Xp,sF);
+  if (  SE->_is_local && SE->_permute ) {
+    spProjXm(tmp,in._odata[SE->_offset]);
     permute(chi,tmp,ptype);
-  } else if ( local ) {
-    spProjXm(chi,in._odata[offset]);
+  } else if ( SE->_is_local ) {
+    spProjXm(chi,in._odata[SE->_offset]);
   } else { 
-    chi=buf[offset];
+    chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Xp);
+  Impl::multLink(Uchi,U._odata[sU],chi,Xp,SE);
   accumReconXm(result,Uchi);
 
   // Ym
-  offset = st._offsets [Yp][ss];
-  local  = st._is_local[Yp][ss];
-  perm   = st._permute[Yp][ss];
-  ptype  = st._permute_type[Yp];
-
-  if ( local && perm ) {
-    spProjYm(tmp,in._odata[offset]);
+  SE=st.GetEntry(ptype,Yp,sF);
+  if (  SE->_is_local && SE->_permute ) {
+    spProjYm(tmp,in._odata[SE->_offset]);
     permute(chi,tmp,ptype);
-  } else if ( local ) {
-    spProjYm(chi,in._odata[offset]);
+  } else if ( SE->_is_local ) {
+    spProjYm(chi,in._odata[SE->_offset]);
   } else { 
-    chi=buf[offset];
+    chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Yp);
+  Impl::multLink(Uchi,U._odata[sU],chi,Yp,SE);
   accumReconYm(result,Uchi);
 
   // Zm
-  offset = st._offsets [Zp][ss];
-  local  = st._is_local[Zp][ss];
-  perm   = st._permute[Zp][ss];
-  ptype  = st._permute_type[Zp];
-  if ( local && perm ) {
-    spProjZm(tmp,in._odata[offset]);
+  SE=st.GetEntry(ptype,Zp,sF);
+  if (  SE->_is_local && SE->_permute ) {
+    spProjZm(tmp,in._odata[SE->_offset]);
     permute(chi,tmp,ptype);
-  } else if ( local ) {
-    spProjZm(chi,in._odata[offset]);
+  } else if ( SE->_is_local ) {
+    spProjZm(chi,in._odata[SE->_offset]);
   } else { 
-    chi=buf[offset];
+    chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Zp);
+  Impl::multLink(Uchi,U._odata[sU],chi,Zp,SE);
   accumReconZm(result,Uchi);
     
   // Tm
-  offset = st._offsets [Tp][ss];
-  local  = st._is_local[Tp][ss];
-  perm   = st._permute[Tp][ss];
-  ptype  = st._permute_type[Tp];
-  if ( local && perm ) {
-    spProjTm(tmp,in._odata[offset]);
+  SE=st.GetEntry(ptype,Tp,sF);
+  if (  SE->_is_local && SE->_permute ) {
+    spProjTm(tmp,in._odata[SE->_offset]);
     permute(chi,tmp,ptype);
-  } else if ( local ) {
-    spProjTm(chi,in._odata[offset]);
+  } else if ( SE->_is_local ) {
+    spProjTm(chi,in._odata[SE->_offset]);
   } else { 
-    chi=buf[offset];
+    chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Tp);
+  Impl::multLink(Uchi,U._odata[sU],chi,Tp,SE);
   accumReconTm(result,Uchi);
   
-  vstream(out._odata[ss],result*(-0.5));
+  vstream(out._odata[sF],result*(-0.5));
 }
 
 template<class Impl> 
@@ -303,127 +249,124 @@ void WilsonKernels<Impl>::DiracOptDhopDir(CartesianStencil &st,DoubledGaugeField
   SiteHalfSpinor  chi;    
   SiteSpinor   result;
   SiteHalfSpinor Uchi;
-  int offset,local,perm, ptype;
-  int ss=sF;
+  StencilEntry *SE;
+  int ptype;
 
-  offset = st._offsets [dir][ss];
-  local  = st._is_local[dir][ss];
-  perm   = st._permute[dir][ss];
-  ptype  = st._permute_type[dir];
+  SE=st.GetEntry(ptype,dir,sF);
 
   // Xp
   if(gamma==Xp){
-    if ( local && perm ) {
-      spProjXp(tmp,in._odata[offset]);
+    if (  SE->_is_local && SE->_permute ) {
+      spProjXp(tmp,in._odata[SE->_offset]);
       permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjXp(chi,in._odata[offset]);
+    } else if ( SE->_is_local ) {
+      spProjXp(chi,in._odata[SE->_offset]);
     } else { 
-      chi=buf[offset];
+      chi=buf[SE->_offset];
     }
-    Impl::multLink(Uchi,U._odata[sU],chi,dir);
+    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE);
     spReconXp(result,Uchi);
   }
 
   // Yp
   if ( gamma==Yp ){
-    if ( local && perm ) {
-      spProjYp(tmp,in._odata[offset]);
+    if (  SE->_is_local && SE->_permute ) {
+      spProjYp(tmp,in._odata[SE->_offset]);
       permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjYp(chi,in._odata[offset]);
+    } else if ( SE->_is_local ) {
+      spProjYp(chi,in._odata[SE->_offset]);
     } else { 
-      chi=buf[offset];
+      chi=buf[SE->_offset];
     }
-    Impl::multLink(Uchi,U._odata[sU],chi,dir);
+    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE);
     spReconYp(result,Uchi);
   }
   
   // Zp
   if ( gamma ==Zp ){
-    if ( local && perm ) {
-      spProjZp(tmp,in._odata[offset]);
+    if (  SE->_is_local && SE->_permute ) {
+      spProjZp(tmp,in._odata[SE->_offset]);
       permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjZp(chi,in._odata[offset]);
+    } else if ( SE->_is_local ) {
+      spProjZp(chi,in._odata[SE->_offset]);
     } else { 
-      chi=buf[offset];
+      chi=buf[SE->_offset];
     }
-    Impl::multLink(Uchi,U._odata[sU],chi,dir);
+    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE);
     spReconZp(result,Uchi);
   }
   
   // Tp
   if ( gamma ==Tp ){
-    if ( local && perm ) {
-      spProjTp(tmp,in._odata[offset]);
+    if (  SE->_is_local && SE->_permute ) {
+      spProjTp(tmp,in._odata[SE->_offset]);
       permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjTp(chi,in._odata[offset]);
+    } else if ( SE->_is_local ) {
+      spProjTp(chi,in._odata[SE->_offset]);
     } else { 
-      chi=buf[offset];
+      chi=buf[SE->_offset];
     }
-    Impl::multLink(Uchi,U._odata[sU],chi,dir);
+    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE);
     spReconTp(result,Uchi);
   }
 
   // Xm
   if ( gamma==Xm ){
-    if ( local && perm ) {
-      spProjXm(tmp,in._odata[offset]);
+    if (  SE->_is_local && SE->_permute ) {
+      spProjXm(tmp,in._odata[SE->_offset]);
       permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjXm(chi,in._odata[offset]);
+    } else if ( SE->_is_local ) {
+      spProjXm(chi,in._odata[SE->_offset]);
     } else { 
-      chi=buf[offset];
+      chi=buf[SE->_offset];
     }
-    Impl::multLink(Uchi,U._odata[sU],chi,dir);
+    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE);
     spReconXm(result,Uchi);
   }
 
   // Ym
   if ( gamma == Ym ){
-    if ( local && perm ) {
-      spProjYm(tmp,in._odata[offset]);
+    if (  SE->_is_local && SE->_permute ) {
+      spProjYm(tmp,in._odata[SE->_offset]);
       permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjYm(chi,in._odata[offset]);
+    } else if ( SE->_is_local ) {
+      spProjYm(chi,in._odata[SE->_offset]);
     } else { 
-      chi=buf[offset];
+      chi=buf[SE->_offset];
     }
-    Impl::multLink(Uchi,U._odata[sU],chi,dir);
+    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE);
     spReconYm(result,Uchi);
   }
 
   // Zm
   if ( gamma == Zm ){
-    if ( local && perm ) {
-      spProjZm(tmp,in._odata[offset]);
+    if (  SE->_is_local && SE->_permute ) {
+      spProjZm(tmp,in._odata[SE->_offset]);
       permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjZm(chi,in._odata[offset]);
+    } else if ( SE->_is_local ) {
+      spProjZm(chi,in._odata[SE->_offset]);
     } else { 
-      chi=buf[offset];
+      chi=buf[SE->_offset];
     }
-    Impl::multLink(Uchi,U._odata[sU],chi,dir);
+    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE);
     spReconZm(result,Uchi);
   }
   
   // Tm
   if ( gamma==Tm ) {
-    if ( local && perm ) {
-      spProjTm(tmp,in._odata[offset]);
+    if (  SE->_is_local && SE->_permute ) {
+      spProjTm(tmp,in._odata[SE->_offset]);
       permute(chi,tmp,ptype);
-    } else if ( local ) {
-      spProjTm(chi,in._odata[offset]);
+    } else if ( SE->_is_local ) {
+      spProjTm(chi,in._odata[SE->_offset]);
     } else { 
-      chi=buf[offset];
+      chi=buf[SE->_offset];
     }
-    Impl::multLink(Uchi,U._odata[sU],chi,dir);
+    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE);
     spReconTm(result,Uchi);
   }
 
-  vstream(out._odata[ss],result*(-0.5));
+  vstream(out._odata[sF],result*(-0.5));
 }
 
   FermOpTemplateInstantiate(WilsonKernels);
diff --git a/lib/stencil/Stencil_common.cc b/lib/stencil/Stencil_common.cc
index 7f894faf..8a8b3a54 100644
--- a/lib/stencil/Stencil_common.cc
+++ b/lib/stencil/Stencil_common.cc
@@ -8,11 +8,7 @@ namespace Grid {
 				     int checkerboard,
 				     const std::vector<int> &directions,
 				     const std::vector<int> &distances) 
-    :   _offsets(npoints), 
-	_is_local(npoints), 
-	_comm_buf_size(npoints), 
-	_permute_type(npoints),
-	_permute(npoints)
+    :   _entries(npoints), _permute_type(npoints)
     {
       _npoints = npoints;
       _grid    = grid;
@@ -27,9 +23,7 @@ namespace Grid {
 
 	int point = i;
 
-	_offsets[i].resize( osites);
-	_is_local[i].resize(osites);
-	_permute[i].resize( osites);
+	_entries[i].resize( osites);
 
 	int dimension    = directions[i];
 	int displacement = distances[i];
@@ -76,7 +70,7 @@ namespace Grid {
     }
 
 
-    void CartesianStencil::Local     (int point, int dimension,int shift,int cbmask)
+    void CartesianStencil::Local     (int point, int dimension,int shiftpm,int cbmask)
     {
       int fd = _grid->_fdimensions[dimension];
       int rd = _grid->_rdimensions[dimension];
@@ -84,7 +78,7 @@ namespace Grid {
       int gd = _grid->_gdimensions[dimension];
       
       // Map to always positive shift modulo global full dimension.
-      shift = (shift+fd)%fd;
+      int shift = (shiftpm+fd)%fd;
       
       // the permute type
       int permute_dim =_grid->PermuteDim(dimension);
@@ -98,6 +92,15 @@ namespace Grid {
 	  
 	int sshift = _grid->CheckerBoardShiftForCB(_checkerboard,dimension,shift,cb);
 	int sx     = (x+sshift)%rd;
+
+	int wraparound=0;
+	if ( (shiftpm==-1) && (sx>x)  ) {
+	  wraparound = 1;
+	}
+	if ( (shiftpm== 1) && (sx<x)  ) {
+	  wraparound = 1;
+	}
+
 	  
 	int permute_slice=0;
 	if(permute_dim){
@@ -107,12 +110,12 @@ namespace Grid {
 	  else permute_slice = 1-wrap;
 	}
 
-  	CopyPlane(point,dimension,x,sx,cbmask,permute_slice);
+  	CopyPlane(point,dimension,x,sx,cbmask,permute_slice,wraparound);
   
       }
     }
 
-    void CartesianStencil::Comms     (int point,int dimension,int shift,int cbmask)
+    void CartesianStencil::Comms     (int point,int dimension,int shiftpm,int cbmask)
     {
       GridBase *grid=_grid;
       
@@ -125,7 +128,7 @@ namespace Grid {
       
       //      assert(simd_layout==1); // Why?
       assert(comm_dim==1);
-      shift = (shift + fd) %fd;
+      int shift = (shiftpm + fd) %fd;
       assert(shift>=0);
       assert(shift<fd);
       
@@ -143,10 +146,17 @@ namespace Grid {
 	//	int offnode     = (comm_proc!=0);
 	int sx          = (x+sshift)%rd;
 
+	int wraparound=0;
+	if ( (shiftpm==-1) && (sx>x) && (grid->_processor_coor[dimension]==0) ) {
+	  wraparound = 1;
+	}
+	if ( (shiftpm== 1) && (sx<x) && (grid->_processor_coor[dimension]==grid->_processors[dimension]-1) ) {
+	  wraparound = 1;
+	}
 	if (!offnode) {
 	  
 	  int permute_slice=0;
-	  CopyPlane(point,dimension,x,sx,cbmask,permute_slice); 
+	  CopyPlane(point,dimension,x,sx,cbmask,permute_slice,wraparound); 
 	  
 	} else {
 	  
@@ -161,13 +171,13 @@ namespace Grid {
 
 	  int unified_buffer_offset = _unified_buffer_size;
 	  _unified_buffer_size    += words;
-	  ScatterPlane(point,dimension,x,cbmask,unified_buffer_offset); // permute/extract/merge is done in comms phase
+	  ScatterPlane(point,dimension,x,cbmask,unified_buffer_offset,wraparound); // permute/extract/merge is done in comms phase
 	  
 	}
       }
     }
   // Routine builds up integer table for each site in _offsets, _is_local, _permute
-  void CartesianStencil::CopyPlane(int point, int dimension,int lplane,int rplane,int cbmask,int permute)
+  void CartesianStencil::CopyPlane(int point, int dimension,int lplane,int rplane,int cbmask,int permute,int wrap)
     {
       int rd = _grid->_rdimensions[dimension];
       
@@ -180,9 +190,10 @@ namespace Grid {
 	// Simple block stride gather of SIMD objects
 	for(int n=0;n<_grid->_slice_nblock[dimension];n++){
 	  for(int b=0;b<_grid->_slice_block[dimension];b++){
-	    _offsets [point][lo+o+b]=ro+o+b;
-	    _is_local[point][lo+o+b]=1;
-	    _permute [point][lo+o+b]=permute;
+	    _entries[point][lo+o+b]._offset  =ro+o+b;
+	    _entries[point][lo+o+b]._is_local=1;
+	    _entries[point][lo+o+b]._permute=permute;
+	    _entries[point][lo+o+b]._around_the_world=wrap;
 	  }
 	  o +=_grid->_slice_stride[dimension];
 	}
@@ -199,9 +210,10 @@ namespace Grid {
 	    int ocb=1<<_grid->CheckerBoardFromOindex(o+b);
 	    
 	    if ( ocb&cbmask ) {
-	      _offsets [point][lo+o+b]=ro+o+b;
-	      _is_local[point][lo+o+b]=1;
-	      _permute [point][lo+o+b]=permute;
+	      _entries[point][lo+o+b]._offset =ro+o+b;
+	      _entries[point][lo+o+b]._is_local=1;
+	      _entries[point][lo+o+b]._permute=permute;
+	      _entries[point][lo+o+b]._around_the_world=wrap;
 	    }
 	    
 	    }
@@ -211,7 +223,7 @@ namespace Grid {
       }
     }
   // Routine builds up integer table for each site in _offsets, _is_local, _permute
-    void CartesianStencil::ScatterPlane (int point,int dimension,int plane,int cbmask,int offset)
+   void CartesianStencil::ScatterPlane (int point,int dimension,int plane,int cbmask,int offset, int wrap)
     {
       int rd = _grid->_rdimensions[dimension];
       
@@ -224,9 +236,10 @@ namespace Grid {
 	// Simple block stride gather of SIMD objects
 	for(int n=0;n<_grid->_slice_nblock[dimension];n++){
 	  for(int b=0;b<_grid->_slice_block[dimension];b++){
-	    _offsets [point][so+o+b]=offset+(bo++);
-	    _is_local[point][so+o+b]=0;
-	    _permute [point][so+o+b]=0;
+	    _entries[point][so+o+b]._offset  =offset+(bo++);
+	    _entries[point][so+o+b]._is_local=0;
+	    _entries[point][so+o+b]._permute=0;
+	    _entries[point][so+o+b]._around_the_world=wrap;
 	  }
 	  o +=_grid->_slice_stride[dimension];
 	}
@@ -242,9 +255,10 @@ namespace Grid {
 
 	    int ocb=1<<_grid->CheckerBoardFromOindex(o+b);// Could easily be a table lookup
 	    if ( ocb & cbmask ) {
-	      _offsets [point][so+o+b]=offset+(bo++);
-	      _is_local[point][so+o+b]=0;
-	      _permute [point][so+o+b]=0;
+	      _entries[point][so+o+b]._offset  =offset+(bo++);
+	      _entries[point][so+o+b]._is_local=0;
+	      _entries[point][so+o+b]._permute =0;
+	      _entries[point][so+o+b]._around_the_world=wrap;
 	    }
 	  }
 	  o +=_grid->_slice_stride[dimension];
diff --git a/tests/Make.inc b/tests/Make.inc
index 111eed35..c765cdf7 100644
--- a/tests/Make.inc
+++ b/tests/Make.inc
@@ -1,5 +1,5 @@
 
-bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_cheby Test_contfrac_cg Test_contfrac_even_odd Test_contfrac_force Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_force Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_gparity Test_hmc_EOWilsonFermionGauge Test_hmc_WilsonFermionGauge Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_partfrac_force Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd Test_wilson_force Test_wilson_force_phiMdagMphi Test_wilson_force_phiMphi
+bin_PROGRAMS = Test_GaugeAction Test_cayley_cg Test_cayley_coarsen_support Test_cayley_even_odd Test_cayley_ldop_cr Test_cf_coarsen_support Test_cf_cr_unprec Test_cheby Test_contfrac_cg Test_contfrac_even_odd Test_contfrac_force Test_cshift Test_cshift_red_black Test_dwf_cg_prec Test_dwf_cg_schur Test_dwf_cg_unprec Test_dwf_cr_unprec Test_dwf_even_odd Test_dwf_force Test_dwf_fpgcr Test_dwf_hdcr Test_gamma Test_gparity Test_gpwilson_even_odd Test_hmc_EOWilsonFermionGauge Test_hmc_WilsonFermionGauge Test_hmc_WilsonGauge Test_lie_generators Test_main Test_multishift_sqrt Test_nersc_io Test_partfrac_force Test_quenched_update Test_remez Test_rng Test_rng_fixed Test_simd Test_stencil Test_wilson_cg_prec Test_wilson_cg_schur Test_wilson_cg_unprec Test_wilson_cr_unprec Test_wilson_even_odd Test_wilson_force Test_wilson_force_phiMdagMphi Test_wilson_force_phiMphi
 
 
 Test_GaugeAction_SOURCES=Test_GaugeAction.cc
@@ -94,6 +94,10 @@ Test_gparity_SOURCES=Test_gparity.cc
 Test_gparity_LDADD=-lGrid
 
 
+Test_gpwilson_even_odd_SOURCES=Test_gpwilson_even_odd.cc
+Test_gpwilson_even_odd_LDADD=-lGrid
+
+
 Test_hmc_EOWilsonFermionGauge_SOURCES=Test_hmc_EOWilsonFermionGauge.cc
 Test_hmc_EOWilsonFermionGauge_LDADD=-lGrid
 
diff --git a/tests/Test_gparity.cc b/tests/Test_gparity.cc
index 46c019c6..52bf5e27 100644
--- a/tests/Test_gparity.cc
+++ b/tests/Test_gparity.cc
@@ -4,26 +4,54 @@ using namespace std;
 using namespace Grid;
 using namespace Grid::QCD;
 
-template<class d>
-struct scal {
-  d internal;
-};
+typedef typename GparityDomainWallFermionR::FermionField FermionField;
 
-  Gamma::GammaMatrix Gmu [] = {
-    Gamma::GammaX,
-    Gamma::GammaY,
-    Gamma::GammaZ,
-    Gamma::GammaT
-  };
+template<class vobj>
+void Replicate(Lattice<vobj> &coarse,Lattice<vobj> & fine)
+{
+  typedef typename vobj::scalar_object sobj;
+
+  GridBase *cg = coarse._grid;
+  GridBase *fg =   fine._grid;
+
+  int nd = cg->_ndimension;
+
+  subdivides(cg,fg); 
+
+  assert(cg->_ndimension==fg->_ndimension);
+
+  std::vector<int> ratio(cg->_ndimension);
+
+  for(int d=0;d<cg->_ndimension;d++){
+    ratio[d] = fg->_fdimensions[d]/cg->_fdimensions[d];
+  }
+
+  std::vector<int> fcoor(nd);
+  std::vector<int> ccoor(nd);
+  for(int g=0;g<fg->gSites();g++){
+
+    fg->GlobalIndexToGlobalCoor(g,fcoor);
+    for(int d=0;d<nd;d++){
+      ccoor[d] = fcoor[d]%cg->_gdimensions[d];
+    }
+    
+    sobj tmp;
+    peekSite(tmp,coarse,ccoor);
+    pokeSite(tmp,fine,fcoor);
+  }
+
+}
 
 int main (int argc, char ** argv)
 {
+  const int nu = 0;
+
   Grid_init(&argc,&argv);
 
   const int Ls=4;
-  const int L=4;
+  const int L =4;
   std::vector<int> latt_2f(Nd,L);
-  std::vector<int> latt_1f(Nd,L); latt_1f[0] = 2L;
+  std::vector<int> latt_1f(Nd,L); latt_1f[nu] = 2*L;
 
   std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
   std::vector<int> mpi_layout  = GridDefaultMpi(); //node layout
@@ -33,30 +61,64 @@ int main (int argc, char ** argv)
   GridCartesian         * FGrid_1f   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid_1f);
   GridRedBlackCartesian * FrbGrid_1f = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid_1f);
 
+
+  GridCartesian         * UGrid_2f   = SpaceTimeGrid::makeFourDimGrid(latt_2f, simd_layout, mpi_layout);
+  GridRedBlackCartesian * UrbGrid_2f = SpaceTimeGrid::makeFourDimRedBlackGrid(UGrid_2f);
+  GridCartesian         * FGrid_2f   = SpaceTimeGrid::makeFiveDimGrid(Ls,UGrid_2f);
+  GridRedBlackCartesian * FrbGrid_2f = SpaceTimeGrid::makeFiveDimRedBlackGrid(Ls,UGrid_2f);
+
   std::vector<int> seeds4({1,2,3,4});
   std::vector<int> seeds5({5,6,7,8});
-  GridParallelRNG          RNG5_1f(FGrid_1f);  RNG5_1f.SeedFixedIntegers(seeds5);
-  GridParallelRNG          RNG4_1f(UGrid_1f);  RNG4_1f.SeedFixedIntegers(seeds4);
+  GridParallelRNG          RNG5_2f(FGrid_2f);  RNG5_2f.SeedFixedIntegers(seeds5);
+  GridParallelRNG          RNG4_2f(UGrid_2f);  RNG4_2f.SeedFixedIntegers(seeds4);
+
+  LatticeGaugeField Umu_2f(UGrid_2f);
+  SU3::HotConfiguration(RNG4_2f,Umu_2f);
+
+  LatticeFermion    src   (FGrid_2f); 
+  LatticeFermion    tmpsrc(FGrid_2f); 
+  FermionField      src_2f(FGrid_2f); 
+  LatticeFermion    src_1f(FGrid_1f); 
+
+  // Replicate fermion source
+  random(RNG5_2f,src);
+  PokeIndex<0>(src_2f,src,0);
+  tmpsrc=src*2.0;
+  PokeIndex<0>(src_2f,tmpsrc,1);
 
-  LatticeFermion    src_1f(FGrid_1f); random(RNG5_1f,src_1f);
   LatticeFermion result_1f(FGrid_1f); result_1f=zero;
-  LatticeGaugeField Umu_1f(UGrid_1f); random(RNG4_1f,Umu_1f);
+  LatticeGaugeField Umu_1f(UGrid_1f); 
+  Replicate(Umu_2f,Umu_1f);
 
   //Coordinate grid for reference
   LatticeInteger xcoor_1f(UGrid_1f);
-  LatticeCoordinate(xcoor_1f,0);
+  LatticeCoordinate(xcoor_1f,nu);
 
   //Copy-conjugate the gauge field
   //First C-shift the lattice by Lx/2
   {
-    LatticeGaugeField Umu_shift = conjugate( Cshift(Umu_1f,0,L) );
+    LatticeGaugeField Umu_shift = conjugate( Cshift(Umu_1f,nu,L) );
     Umu_1f = where( xcoor_1f >= Integer(L), Umu_shift, Umu_1f );
+
+    // hack test to check the same
+    Replicate(Umu_2f,Umu_shift);
+    Umu_shift=Umu_shift-Umu_1f;
+    cout << GridLogMessage << "Umu diff " << norm2(Umu_shift)<<std::endl;
+
+    //Make the gauge field antiperiodic in nu-direction
+    LatticeColourMatrix Unu(UGrid_1f);
+    Unu = PeekIndex<LorentzIndex>(Umu_1f,nu);
+    Unu = where(xcoor_1f == Integer(2*L-1), -Unu, Unu);
+    PokeIndex<LorentzIndex>(Umu_1f,Unu,nu);
   }
 
-  //Make the gauge field antiperiodic in x-direction
-  Umu_1f = where(xcoor_1f == Integer(L-1), -Umu_1f, Umu_1f);
-    
-  RealD mass=0.1;
+  //Coordinate grid for reference
+  LatticeInteger    xcoor_1f5(FGrid_1f);
+  LatticeCoordinate(xcoor_1f5,1+nu);
+  Replicate(src,src_1f);
+  src_1f   = where( xcoor_1f5 >= Integer(L), 2.0*src_1f,src_1f );
+
+  RealD mass=0.0;
   RealD M5=1.8;
   DomainWallFermionR Ddwf(Umu_1f,*FGrid_1f,*FrbGrid_1f,*UGrid_1f,*UrbGrid_1f,mass,M5);
 
@@ -69,5 +131,136 @@ int main (int argc, char ** argv)
   ConjugateGradient<LatticeFermion> CG(1.0e-8,10000);
   CG(HermOpEO,src_o_1f,result_o_1f);
 
+  GparityDomainWallFermionR GPDdwf(Umu_2f,*FGrid_2f,*FrbGrid_2f,*UGrid_2f,*UrbGrid_2f,mass,M5);
+
+  for(int disp=-1;disp<=1;disp+=2)
+  for(int mu=0;mu<5;mu++)
+  { 
+    FermionField Dsrc_2f(FGrid_2f);
+
+    LatticeFermion Dsrc_1f(FGrid_1f);
+    LatticeFermion Dsrc_2freplica(FGrid_1f);
+    LatticeFermion Dsrc_2freplica0(FGrid_1f);
+    LatticeFermion Dsrc_2freplica1(FGrid_1f);
+
+    if ( mu ==0 ) {
+      std::cout << GridLogMessage<< " Cross checking entire hopping term"<<std::endl;
+      GPDdwf.Dhop(src_2f,Dsrc_2f,DaggerNo);
+        Ddwf.Dhop(src_1f,Dsrc_1f,DaggerNo);
+    } else { 
+      std::cout << GridLogMessage<< " Cross checking mu="<<mu<< " disp="<< disp<<std::endl;
+      GPDdwf.DhopDir(src_2f,Dsrc_2f,mu,disp);
+        Ddwf.DhopDir(src_1f,Dsrc_1f,mu,disp);
+    }
+
+    std::cout << GridLogMessage << "S norms "<< norm2(src_2f) << " " << norm2(src_1f)  <<std::endl;
+    std::cout << GridLogMessage << "D norms "<< norm2(Dsrc_2f)<< " " << norm2(Dsrc_1f) <<std::endl;
+
+    LatticeFermion Dsrc_2f0(FGrid_2f); Dsrc_2f0 = PeekIndex<0>(Dsrc_2f,0);
+    LatticeFermion Dsrc_2f1(FGrid_2f); Dsrc_2f1 = PeekIndex<0>(Dsrc_2f,1);
+
+    //    Dsrc_2f1 = Dsrc_2f1 - Dsrc_2f0;
+    //    std::cout << GridLogMessage << " Cross check two halves " <<norm2(Dsrc_2f1)<<std::endl;
+
+    Replicate(Dsrc_2f0,Dsrc_2freplica0);
+    Replicate(Dsrc_2f1,Dsrc_2freplica1);
+
+    Dsrc_2freplica = where( xcoor_1f5 >= Integer(L), Dsrc_2freplica1,Dsrc_2freplica0 );
+    Dsrc_2freplica = Dsrc_2freplica - Dsrc_1f ;
+    std::cout << GridLogMessage << " Cross check against doubled latt " <<norm2(Dsrc_2freplica)<<std::endl;
+
+    //    std::cout << Dsrc_2f <<std::endl;
+
+  }
+
+  {
+    FermionField chi   (FGrid_2f); gaussian(RNG5_2f,chi);
+    FermionField phi   (FGrid_2f); gaussian(RNG5_2f,phi);
+  
+    FermionField chi_e   (FrbGrid_2f);
+    FermionField chi_o   (FrbGrid_2f);
+    
+    FermionField dchi_e  (FrbGrid_2f);
+    FermionField dchi_o  (FrbGrid_2f);
+    
+    FermionField phi_e   (FrbGrid_2f);
+    FermionField phi_o   (FrbGrid_2f);
+    
+    FermionField dphi_e  (FrbGrid_2f);
+    FermionField dphi_o  (FrbGrid_2f);
+
+    pickCheckerboard(Even,chi_e,chi);
+    pickCheckerboard(Odd ,chi_o,chi);
+    pickCheckerboard(Even,phi_e,phi);
+    pickCheckerboard(Odd ,phi_o,phi);
+
+    GPDdwf.Meooe(chi_e,dchi_o);
+    GPDdwf.Meooe(chi_o,dchi_e);
+    GPDdwf.MeooeDag(phi_e,dphi_o);
+    GPDdwf.MeooeDag(phi_o,dphi_e);
+    
+    ComplexD pDce = innerProduct(phi_e,dchi_e);
+    ComplexD pDco = innerProduct(phi_o,dchi_o);
+    ComplexD cDpe = innerProduct(chi_e,dphi_e);
+    ComplexD cDpo = innerProduct(chi_o,dphi_o);
+    
+    std::cout<<GridLogMessage <<"e "<<pDce<<" "<<cDpe <<std::endl;
+    std::cout<<GridLogMessage <<"o "<<pDco<<" "<<cDpo <<std::endl;
+    
+    std::cout<<GridLogMessage <<"pDce - conj(cDpo) "<< pDce-conj(cDpo) <<std::endl;
+    std::cout<<GridLogMessage <<"pDco - conj(cDpe) "<< pDco-conj(cDpe) <<std::endl;
+
+  }
+
+  FermionField result_2f(FGrid_2f); result_2f=zero;
+  FermionField    src_o_2f(FrbGrid_2f);
+  FermionField result_o_2f(FrbGrid_2f);
+  pickCheckerboard(Odd,src_o_2f,src_2f);
+  result_o_2f=zero;
+
+  ConjugateGradient<FermionField> CG2f(1.0e-8,10000);
+  SchurDiagMooeeOperator<GparityDomainWallFermionR,FermionField> HermOpEO2f(GPDdwf);
+  CG2f(HermOpEO2f,src_o_2f,result_o_2f);
+
+  std::cout << "2f cb "<<result_o_2f.checkerboard<<std::endl;
+  std::cout << "1f cb "<<result_o_1f.checkerboard<<std::endl;
+
+  std::cout << " result norms " <<norm2(result_o_2f)<<" " <<norm2(result_o_1f)<<std::endl;
+
+  LatticeFermion    res0o  (FrbGrid_2f); 
+  LatticeFermion    res1o  (FrbGrid_2f); 
+  LatticeFermion    res0  (FGrid_2f); 
+  LatticeFermion    res1  (FGrid_2f); 
+
+  res0=zero;
+  res1=zero;
+
+  res0o = PeekIndex<0>(result_o_2f,0);
+  res1o = PeekIndex<0>(result_o_2f,1);
+
+  std::cout << "res cb "<<res0o.checkerboard<<std::endl;
+  std::cout << "res cb "<<res1o.checkerboard<<std::endl;
+
+  setCheckerboard(res0,res0o);
+  setCheckerboard(res1,res1o);
+
+  LatticeFermion replica (FGrid_1f);
+  LatticeFermion replica0(FGrid_1f);
+  LatticeFermion replica1(FGrid_1f);
+  Replicate(res0,replica0);
+  Replicate(res1,replica1);
+
+  replica = where( xcoor_1f5 >= Integer(L), replica1,replica0 );
+
+  replica0 = zero;
+  setCheckerboard(replica0,result_o_1f);
+
+  std::cout << "Norm2 solutions is " <<norm2(replica)<<" "<< norm2(replica0)<<std::endl;
+
+  replica = replica - replica0;
+  
+  std::cout << "Norm2 of difference in solutions is " <<norm2(replica)<<std::endl;
+  
+
   Grid_finalize();
 }
diff --git a/tests/Test_stencil.cc b/tests/Test_stencil.cc
index d455c1c1..cc49ab22 100644
--- a/tests/Test_stencil.cc
+++ b/tests/Test_stencil.cc
@@ -56,17 +56,17 @@ int main (int argc, char ** argv)
 
 	// Implement a stencil code that should agree with cshift!
 	for(int i=0;i<Check._grid->oSites();i++){
-
-	  int offset = myStencil._offsets [0][i];
-	  int  local = myStencil._is_local[0][i];
-	  int permute_type = myStencil._permute_type[0];
-	  int perm =myStencil._permute[0][i];
-	  if ( local && perm )
-	    permute(Check._odata[i],Foo._odata[offset],permute_type);
-	  else if (local)
-	    Check._odata[i] = Foo._odata[offset];
+	  
+	  int permute_type;
+	  StencilEntry *SE;
+	  SE = myStencil.GetEntry(permute_type,0,i);
+	  
+	  if ( SE->_is_local && SE->_permute )
+	    permute(Check._odata[i],Foo._odata[SE->_offset],permute_type);
+	  else if (SE->_is_local)
+	    Check._odata[i] = Foo._odata[SE->_offset];
 	  else 
-	    Check._odata[i] = comm_buf[offset];
+	    Check._odata[i] = comm_buf[SE->_offset];
 	}
 
 	Real nrmC = norm2(Check);

From 59d66eb17a7846c910066e3a2d6c3fb38d16d606 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 14 Aug 2015 12:57:42 +0100
Subject: [PATCH 428/429] Gparity works now even if simd distributed in a
 Gparity twist direction. Tested by doubling lattice in t-direction.

---
 lib/lattice/Lattice_where.h               |  1 -
 lib/qcd/action/fermion/FermionOperator.h  | 73 +++++++++++++++++++---
 lib/qcd/action/fermion/WilsonCompressor.h | 74 -----------------------
 lib/qcd/action/fermion/WilsonKernels.cc   | 48 +++++++--------
 tests/Test_gparity.cc                     |  2 +-
 5 files changed, 90 insertions(+), 108 deletions(-)

diff --git a/lib/lattice/Lattice_where.h b/lib/lattice/Lattice_where.h
index c83d80c1..72bc936e 100644
--- a/lib/lattice/Lattice_where.h
+++ b/lib/lattice/Lattice_where.h
@@ -22,7 +22,6 @@ inline void whereWolf(Lattice<vobj> &ret,const Lattice<iobj> &predicate,Lattice<
   typedef typename iobj::vector_type mask_type;
 
   const int Nsimd = grid->Nsimd();
-  const int words = sizeof(vobj)/sizeof(vector_type);
 
   std::vector<Integer> mask(Nsimd);
   std::vector<scalar_object> truevals (Nsimd);
diff --git a/lib/qcd/action/fermion/FermionOperator.h b/lib/qcd/action/fermion/FermionOperator.h
index e1f65a27..9829d350 100644
--- a/lib/qcd/action/fermion/FermionOperator.h
+++ b/lib/qcd/action/fermion/FermionOperator.h
@@ -7,7 +7,7 @@ namespace Grid {
 
     ////////////////////////////////////////////////////////////////
     // Hardwire to four spinors, allow to select 
-    // between gauge representation rank, and gparity/flavour index,
+    // between gauge representation rank bc's, flavours etc.
     // and single/double precision.
     ////////////////////////////////////////////////////////////////
 
@@ -37,8 +37,7 @@ namespace Grid {
 
       typedef WilsonCompressor<SiteHalfSpinor,SiteSpinor> Compressor;
 
-      // provide the multiply by link that is differentiated between Gparity (with flavour index) and non-Gparity
-      static inline void multLink(SiteHalfSpinor &phi,const SiteDoubledGaugeField &U,const SiteHalfSpinor &chi,int mu,StencilEntry *SE){
+      static inline void multLink(SiteHalfSpinor &phi,const SiteDoubledGaugeField &U,const SiteHalfSpinor &chi,int mu,StencilEntry *SE,CartesianStencil &St){
         mult(&phi(),&U(mu),&chi());
       }
       static inline void DoubleStore(GridBase *GaugeGrid,DoubledGaugeField &Uds,const GaugeField &Umu)
@@ -94,15 +93,73 @@ namespace Grid {
 
       // provide the multiply by link that is differentiated between Gparity (with flavour index) and 
       // non-Gparity
-    static inline void multLink(SiteHalfSpinor &phi,const SiteDoubledGaugeField &U,const SiteHalfSpinor &chi,int mu,StencilEntry *SE){
+    static inline void multLink(SiteHalfSpinor &phi,const SiteDoubledGaugeField &U,const SiteHalfSpinor &chi,int mu,StencilEntry *SE,CartesianStencil &St){
       // FIXME; need to be more careful. If this is a simd direction we are still stuffed
-      if ( SE->_around_the_world && ((mu==Xp)||(mu==Xm)) ) {
-	mult(&phi(0),&U(0)(mu),&chi(1));
-	mult(&phi(1),&U(1)(mu),&chi(0));
+      // Need access to _simd_layout[mu]. mu is not necessarily dim.
+      typedef SiteHalfSpinor vobj;
+      typedef typename SiteHalfSpinor::scalar_object sobj;
+
+      vobj vtmp;
+      sobj stmp;
+      std::vector<int> gpbc({0,0,0,1,0,0,0,1});
+
+      GridBase *grid = St._grid;
+
+      const int Nsimd = grid->Nsimd();
+
+      int direction    = St._directions[mu];
+      int distance     = St._distances[mu];
+      int ptype     = St._permute_type[mu]; 
+      int sl        = St._grid->_simd_layout[direction];
+
+      // assert our assumptions
+      assert((distance==1)||(distance==-1)); // nearest neighbour stencil hard code
+      assert((sl==1)||(sl==2));
+      
+      std::vector<int> icoor;
+
+      if ( SE->_around_the_world && gpbc[mu] ) {
+	if ( sl == 2 ) {
+
+	  //	  std::cout << "multLink for mu= "<<mu<<" simd length "<<sl<<std::endl;
+
+	  std::vector<sobj> vals(Nsimd);
+	  extract(chi,vals);
+
+	  for(int s=0;s<Nsimd;s++){
+
+	    grid->iCoorFromIindex(icoor,s);
+
+	    assert((icoor[direction]==0)||(icoor[direction]==1));
+
+	    int permute_lane;
+	    if ( distance == 1) {
+	      permute_lane = icoor[direction]?1:0;
+	    } else {
+	      permute_lane = icoor[direction]?0:1;
+	    }
+
+            if ( permute_lane ) { 
+	      stmp(0) = vals[s](1);
+	      stmp(1) = vals[s](0);
+	      vals[s] = stmp;
+	    }
+	  }
+	  
+	  merge(vtmp,vals);
+
+	} else { 
+	  vtmp(0) = chi(1);
+	  vtmp(1) = chi(0);
+	}
+	mult(&phi(0),&U(0)(mu),&vtmp(0));
+	mult(&phi(1),&U(1)(mu),&vtmp(1));
+	
       } else { 
 	mult(&phi(0),&U(0)(mu),&chi(0));
 	mult(&phi(1),&U(1)(mu),&chi(1));
       }
+
     }
 
       static inline void InsertForce(GaugeField &mat,const FermionField &Btilde,const FermionField &A,int mu){
@@ -120,7 +177,7 @@ namespace Grid {
 
 	Lattice<iScalar<vInteger> > coor(GaugeGrid);
 	
-	std::vector<int> gpdirs({1,0,0,0});
+	std::vector<int> gpdirs({0,0,0,1});
 
         for(int mu=0;mu<Nd;mu++){
 
diff --git a/lib/qcd/action/fermion/WilsonCompressor.h b/lib/qcd/action/fermion/WilsonCompressor.h
index f2aae8aa..6c6d560b 100644
--- a/lib/qcd/action/fermion/WilsonCompressor.h
+++ b/lib/qcd/action/fermion/WilsonCompressor.h
@@ -63,80 +63,6 @@ namespace QCD {
     }
   };
 
-  /*
-
-  template<class SiteHalfSpinor,class SiteSpinor>
-    class GparityWilsonCompressor : public WilsonCompressor<SiteHalfSpinor,SiteSpinor>{
-  public:
-    GparityWilsonCompressor(int _dag) : WilsonCompressor<SiteHalfSpinor,SiteSpinor> (_dag){};
-
-    SiteHalfSpinor operator () (const SiteSpinor &in,int dim,int plane,int osite,GridBase *grid)
-    {
-      std::vector<int> Gbcs({1,0,0,0});
-
-      typedef typename SiteHalfSpinor::scalar_object scalar_object;
-
-      const int Nsimd = grid->Nsimd();
-
-      int checkered=grid->CheckerBoarded(dim);
-      int       ocb=grid->CheckerBoardFromOindex(osite);
-
-      SiteHalfSpinor tmp = this->spinproject(in); // spin projected
-
-      //////////////////////////////////////////////////////////////
-      // Check whether we must flavour flip
-      // do this if source is plane 0 on processor 0 in dimension dim
-      //////////////////////////////////////////////////////////////
-      int do_flip  = 0;
-      int flipicoor= 0;
-      if(Gbcs[this->mu]){
-
-	std::cout << "Applying Gparity BC's in direction "<<this->mu<<std::endl;
-
-	if ( (this->mu==Xp)||(this->mu==Yp)||(this->mu==Zp)||(this->mu==Tp) ) {      
-	  if ( (grid->_processor_coor[dim] == 0) 
-	       && (plane==0) 
-	       && ( (!checkered)||(ocb==0) ) ) {
-	    do_flip=1;
-	    flipicoor=0;
-	  }
-	}
-	if ( (this->mu==Xm)||(this->mu==Ym)||(this->mu==Zm)||(this->mu==Tm) ) {      
-	  if ( (grid->_processor_coor[dim] == (grid->_processors[dim]-1) ) 
-	    && (plane==grid->_rdimensions[dim]-1) 
-            && ( (!checkered)||(ocb==1) ) ) {
-	    do_flip=1;
-	    flipicoor=grid->_simd_layout[dim]-1;
-	  }
-	}
-      }
-      if ( do_flip ) {
-
-	std::cout << "Applying Gparity BC's in direction "<<this->mu<< " osite " << osite << " plane "<<plane <<std::endl;
-
-	std::vector<scalar_object>  flat(Nsimd);
-	std::vector<int> coor;
-
-	extract(tmp,flat);
-	for(int i=0;i<Nsimd;i++) {
-	  grid->iCoorFromIindex(coor,i);
-	  if ( coor[dim]==flipicoor ) {
-	    scalar_object stmp;
-	    stmp(0) = flat[i](1);
-	    stmp(1) = flat[i](0);
-	    flat[i] = stmp;
-	  }
-	}
-	merge(tmp,flat);
-
-      }
-
-      return tmp;
-    }
-
-  };
-
-  */
 
 }} // namespace close
 #endif
diff --git a/lib/qcd/action/fermion/WilsonKernels.cc b/lib/qcd/action/fermion/WilsonKernels.cc
index d098d138..77f68e4d 100644
--- a/lib/qcd/action/fermion/WilsonKernels.cc
+++ b/lib/qcd/action/fermion/WilsonKernels.cc
@@ -24,7 +24,7 @@ void WilsonKernels<Impl>::DiracOptDhopSite(CartesianStencil &st,DoubledGaugeFiel
   } else { 
     chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Xp,SE);
+  Impl::multLink(Uchi,U._odata[sU],chi,Xp,SE,st);
   spReconXp(result,Uchi);
     
   // Yp
@@ -37,7 +37,7 @@ void WilsonKernels<Impl>::DiracOptDhopSite(CartesianStencil &st,DoubledGaugeFiel
   } else { 
     chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Yp,SE);
+  Impl::multLink(Uchi,U._odata[sU],chi,Yp,SE,st);
   accumReconYp(result,Uchi);
 
   // Zp
@@ -50,7 +50,7 @@ void WilsonKernels<Impl>::DiracOptDhopSite(CartesianStencil &st,DoubledGaugeFiel
   } else { 
     chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Zp,SE);
+  Impl::multLink(Uchi,U._odata[sU],chi,Zp,SE,st);
   accumReconZp(result,Uchi);
 
   // Tp
@@ -63,7 +63,7 @@ void WilsonKernels<Impl>::DiracOptDhopSite(CartesianStencil &st,DoubledGaugeFiel
   } else { 
     chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Tp,SE);
+  Impl::multLink(Uchi,U._odata[sU],chi,Tp,SE,st);
   accumReconTp(result,Uchi);
 
   // Xm
@@ -76,7 +76,7 @@ void WilsonKernels<Impl>::DiracOptDhopSite(CartesianStencil &st,DoubledGaugeFiel
   } else { 
     chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Xm,SE);
+  Impl::multLink(Uchi,U._odata[sU],chi,Xm,SE,st);
   accumReconXm(result,Uchi);
   
   // Ym
@@ -89,7 +89,7 @@ void WilsonKernels<Impl>::DiracOptDhopSite(CartesianStencil &st,DoubledGaugeFiel
   } else { 
     chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Ym,SE);
+  Impl::multLink(Uchi,U._odata[sU],chi,Ym,SE,st);
   accumReconYm(result,Uchi);
   
   // Zm
@@ -102,7 +102,7 @@ void WilsonKernels<Impl>::DiracOptDhopSite(CartesianStencil &st,DoubledGaugeFiel
   } else { 
     chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Zm,SE);
+  Impl::multLink(Uchi,U._odata[sU],chi,Zm,SE,st);
   accumReconZm(result,Uchi);
 
   // Tm
@@ -115,7 +115,7 @@ void WilsonKernels<Impl>::DiracOptDhopSite(CartesianStencil &st,DoubledGaugeFiel
   } else { 
     chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Tm,SE);
+  Impl::multLink(Uchi,U._odata[sU],chi,Tm,SE,st);
   accumReconTm(result,Uchi);
 
   vstream(out._odata[sF],result*(-0.5));
@@ -143,7 +143,7 @@ void WilsonKernels<Impl>::DiracOptDhopSiteDag(CartesianStencil &st,DoubledGaugeF
   } else { 
     chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Xm,SE);
+  Impl::multLink(Uchi,U._odata[sU],chi,Xm,SE,st);
   spReconXp(result,Uchi);
 
   // Yp
@@ -156,7 +156,7 @@ void WilsonKernels<Impl>::DiracOptDhopSiteDag(CartesianStencil &st,DoubledGaugeF
   } else { 
     chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Ym,SE);
+  Impl::multLink(Uchi,U._odata[sU],chi,Ym,SE,st);
   accumReconYp(result,Uchi);
   
   // Zp
@@ -169,7 +169,7 @@ void WilsonKernels<Impl>::DiracOptDhopSiteDag(CartesianStencil &st,DoubledGaugeF
   } else { 
     chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Zm,SE);
+  Impl::multLink(Uchi,U._odata[sU],chi,Zm,SE,st);
   accumReconZp(result,Uchi);
   
   // Tp
@@ -182,7 +182,7 @@ void WilsonKernels<Impl>::DiracOptDhopSiteDag(CartesianStencil &st,DoubledGaugeF
   } else { 
     chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Tm,SE);
+  Impl::multLink(Uchi,U._odata[sU],chi,Tm,SE,st);
   accumReconTp(result,Uchi);
   
   // Xm
@@ -195,7 +195,7 @@ void WilsonKernels<Impl>::DiracOptDhopSiteDag(CartesianStencil &st,DoubledGaugeF
   } else { 
     chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Xp,SE);
+  Impl::multLink(Uchi,U._odata[sU],chi,Xp,SE,st);
   accumReconXm(result,Uchi);
 
   // Ym
@@ -208,7 +208,7 @@ void WilsonKernels<Impl>::DiracOptDhopSiteDag(CartesianStencil &st,DoubledGaugeF
   } else { 
     chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Yp,SE);
+  Impl::multLink(Uchi,U._odata[sU],chi,Yp,SE,st);
   accumReconYm(result,Uchi);
 
   // Zm
@@ -221,7 +221,7 @@ void WilsonKernels<Impl>::DiracOptDhopSiteDag(CartesianStencil &st,DoubledGaugeF
   } else { 
     chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Zp,SE);
+  Impl::multLink(Uchi,U._odata[sU],chi,Zp,SE,st);
   accumReconZm(result,Uchi);
     
   // Tm
@@ -234,7 +234,7 @@ void WilsonKernels<Impl>::DiracOptDhopSiteDag(CartesianStencil &st,DoubledGaugeF
   } else { 
     chi=buf[SE->_offset];
   }
-  Impl::multLink(Uchi,U._odata[sU],chi,Tp,SE);
+  Impl::multLink(Uchi,U._odata[sU],chi,Tp,SE,st);
   accumReconTm(result,Uchi);
   
   vstream(out._odata[sF],result*(-0.5));
@@ -264,7 +264,7 @@ void WilsonKernels<Impl>::DiracOptDhopDir(CartesianStencil &st,DoubledGaugeField
     } else { 
       chi=buf[SE->_offset];
     }
-    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE);
+    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE,st);
     spReconXp(result,Uchi);
   }
 
@@ -278,7 +278,7 @@ void WilsonKernels<Impl>::DiracOptDhopDir(CartesianStencil &st,DoubledGaugeField
     } else { 
       chi=buf[SE->_offset];
     }
-    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE);
+    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE,st);
     spReconYp(result,Uchi);
   }
   
@@ -292,7 +292,7 @@ void WilsonKernels<Impl>::DiracOptDhopDir(CartesianStencil &st,DoubledGaugeField
     } else { 
       chi=buf[SE->_offset];
     }
-    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE);
+    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE,st);
     spReconZp(result,Uchi);
   }
   
@@ -306,7 +306,7 @@ void WilsonKernels<Impl>::DiracOptDhopDir(CartesianStencil &st,DoubledGaugeField
     } else { 
       chi=buf[SE->_offset];
     }
-    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE);
+    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE,st);
     spReconTp(result,Uchi);
   }
 
@@ -320,7 +320,7 @@ void WilsonKernels<Impl>::DiracOptDhopDir(CartesianStencil &st,DoubledGaugeField
     } else { 
       chi=buf[SE->_offset];
     }
-    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE);
+    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE,st);
     spReconXm(result,Uchi);
   }
 
@@ -334,7 +334,7 @@ void WilsonKernels<Impl>::DiracOptDhopDir(CartesianStencil &st,DoubledGaugeField
     } else { 
       chi=buf[SE->_offset];
     }
-    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE);
+    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE,st);
     spReconYm(result,Uchi);
   }
 
@@ -348,7 +348,7 @@ void WilsonKernels<Impl>::DiracOptDhopDir(CartesianStencil &st,DoubledGaugeField
     } else { 
       chi=buf[SE->_offset];
     }
-    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE);
+    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE,st);
     spReconZm(result,Uchi);
   }
   
@@ -362,7 +362,7 @@ void WilsonKernels<Impl>::DiracOptDhopDir(CartesianStencil &st,DoubledGaugeField
     } else { 
       chi=buf[SE->_offset];
     }
-    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE);
+    Impl::multLink(Uchi,U._odata[sU],chi,dir,SE,st);
     spReconTm(result,Uchi);
   }
 
diff --git a/tests/Test_gparity.cc b/tests/Test_gparity.cc
index 52bf5e27..271c8548 100644
--- a/tests/Test_gparity.cc
+++ b/tests/Test_gparity.cc
@@ -44,7 +44,7 @@ void Replicate(Lattice<vobj> &coarse,Lattice<vobj> & fine)
 
 int main (int argc, char ** argv)
 {
-  const int nu = 0;
+  const int nu = 3;
 
   Grid_init(&argc,&argv);
 

From b3b46fd4560b03046783d47f425e397b0fb35fc4 Mon Sep 17 00:00:00 2001
From: Peter Boyle <paboyle@ph.ed.ac.uk>
Date: Fri, 14 Aug 2015 13:18:59 +0100
Subject: [PATCH 429/429] Extra test

---
 tests/Test_gpwilson_even_odd.cc | 191 ++++++++++++++++++++++++++++++++
 1 file changed, 191 insertions(+)
 create mode 100644 tests/Test_gpwilson_even_odd.cc

diff --git a/tests/Test_gpwilson_even_odd.cc b/tests/Test_gpwilson_even_odd.cc
new file mode 100644
index 00000000..4675bd7c
--- /dev/null
+++ b/tests/Test_gpwilson_even_odd.cc
@@ -0,0 +1,191 @@
+#include <Grid.h>
+
+using namespace std;
+using namespace Grid;
+using namespace Grid::QCD;
+
+int main (int argc, char ** argv)
+{
+  Grid_init(&argc,&argv);
+
+  std::vector<int> latt_size   = GridDefaultLatt();
+  std::vector<int> simd_layout = GridDefaultSimd(Nd,vComplex::Nsimd());
+  std::vector<int> mpi_layout  = GridDefaultMpi();
+  GridCartesian               Grid(latt_size,simd_layout,mpi_layout);
+  GridRedBlackCartesian     RBGrid(latt_size,simd_layout,mpi_layout);
+
+  int threads = GridThread::GetThreads();
+  std::cout<<GridLogMessage << "Grid is setup to use "<<threads<<" threads"<<std::endl;
+
+  std::vector<int> seeds({1,2,3,4});
+
+  GridParallelRNG          pRNG(&Grid);
+  //  std::vector<int> seeds({1,2,3,4});
+  //  pRNG.SeedFixedIntegers(seeds);
+  pRNG.SeedRandomDevice();
+
+  typedef typename GparityWilsonFermionR::FermionField FermionField;
+
+  FermionField src   (&Grid); random(pRNG,src);
+  FermionField phi   (&Grid); random(pRNG,phi);
+  FermionField chi   (&Grid); random(pRNG,chi);
+  FermionField result(&Grid); result=zero;
+  FermionField    ref(&Grid);    ref=zero;
+  FermionField    tmp(&Grid);    tmp=zero;
+  FermionField    err(&Grid);    tmp=zero;
+  LatticeGaugeField Umu(&Grid); random(pRNG,Umu);
+  std::vector<LatticeColourMatrix> U(4,&Grid);
+
+  double volume=1;
+  for(int mu=0;mu<Nd;mu++){
+    volume=volume*latt_size[mu];
+  }  
+
+  // Only one non-zero (y)
+  Umu=zero;
+  for(int nn=0;nn<Nd;nn++){
+    random(pRNG,U[nn]);
+    std::cout<<GridLogMessage<<"U[nn]"<<norm2(U[nn])<<std::endl;
+    PokeIndex<LorentzIndex>(Umu,U[nn],nn);
+    std::cout<<GridLogMessage<<"Umu"<<norm2(Umu)<<std::endl;
+  }
+
+  RealD mass=0.1;
+
+  GparityWilsonFermionR Dw(Umu,Grid,RBGrid,mass);
+
+  FermionField src_e   (&RBGrid);
+  FermionField src_o   (&RBGrid);
+  FermionField r_e   (&RBGrid);
+  FermionField r_o   (&RBGrid);
+  FermionField r_eo  (&Grid);
+
+  std::cout<<GridLogMessage<<"=========================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Testing that Deo + Doe = Dunprec "<<std::endl;
+  std::cout<<GridLogMessage<<"=========================================================="<<std::endl;
+
+  pickCheckerboard(Even,src_e,src);
+  pickCheckerboard(Odd,src_o,src);
+
+  Dw.Meooe(src_e,r_o);  std::cout<<GridLogMessage<<"Applied Meo"<<std::endl;
+  Dw.Meooe(src_o,r_e);  std::cout<<GridLogMessage<<"Applied Moe"<<std::endl;
+  Dw.Dhop (src,ref,DaggerNo);
+
+  setCheckerboard(r_eo,r_o);
+  setCheckerboard(r_eo,r_e);
+
+  err= ref - r_eo;
+  std::cout<<GridLogMessage << "EO norm diff   "<< norm2(err)<< " "<<norm2(ref)<< " " << norm2(r_eo) <<std::endl;
+
+  LatticeComplex cerr(&Grid);
+  cerr = localInnerProduct(err,err);
+
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test Ddagger is the dagger of D by requiring                "<<std::endl;
+  std::cout<<GridLogMessage<<"=  < phi | Deo | chi > * = < chi | Deo^dag| phi>  "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  
+  FermionField chi_e   (&RBGrid);
+  FermionField chi_o   (&RBGrid);
+
+  FermionField dchi_e  (&RBGrid);
+  FermionField dchi_o  (&RBGrid);
+
+  FermionField phi_e   (&RBGrid);
+  FermionField phi_o   (&RBGrid);
+
+  FermionField dphi_e  (&RBGrid);
+  FermionField dphi_o  (&RBGrid);
+
+
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+  pickCheckerboard(Even,phi_e,phi);
+  pickCheckerboard(Odd ,phi_o,phi);
+
+  Dw.Meooe(chi_e,dchi_o);
+  Dw.Meooe(chi_o,dchi_e);
+  Dw.MeooeDag(phi_e,dphi_o);
+  Dw.MeooeDag(phi_o,dphi_e);
+
+  ComplexD pDce = innerProduct(phi_e,dchi_e);
+  ComplexD pDco = innerProduct(phi_o,dchi_o);
+  ComplexD cDpe = innerProduct(chi_e,dphi_e);
+  ComplexD cDpo = innerProduct(chi_o,dphi_o);
+
+  std::cout<<GridLogMessage <<"e "<<pDce<<" "<<cDpe <<std::endl;
+  std::cout<<GridLogMessage <<"o "<<pDco<<" "<<cDpo <<std::endl;
+
+  std::cout<<GridLogMessage <<"pDce - conj(cDpo) "<< pDce-conj(cDpo) <<std::endl;
+  std::cout<<GridLogMessage <<"pDco - conj(cDpe) "<< pDco-conj(cDpe) <<std::endl;
+
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test MeeInv Mee = 1                                         "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+
+  Dw.Mooee(chi_e,src_e);
+  Dw.MooeeInv(src_e,phi_e);
+
+  Dw.Mooee(chi_o,src_o);
+  Dw.MooeeInv(src_o,phi_o);
+  
+  setCheckerboard(phi,phi_e);
+  setCheckerboard(phi,phi_o);
+
+  err = phi-chi;
+  std::cout<<GridLogMessage << "norm diff   "<< norm2(err)<< std::endl;
+
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test MeeInvDag MeeDag = 1                                   "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+
+  Dw.MooeeDag(chi_e,src_e);
+  Dw.MooeeInvDag(src_e,phi_e);
+
+  Dw.MooeeDag(chi_o,src_o);
+  Dw.MooeeInvDag(src_o,phi_o);
+  
+  setCheckerboard(phi,phi_e);
+  setCheckerboard(phi,phi_o);
+
+  err = phi-chi;
+  std::cout<<GridLogMessage << "norm diff   "<< norm2(err)<< std::endl;
+
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  std::cout<<GridLogMessage<<"= Test MpcDagMpc is Hermitian              "<<std::endl;
+  std::cout<<GridLogMessage<<"=============================================================="<<std::endl;
+  
+  random(pRNG,phi);
+  random(pRNG,chi);
+  pickCheckerboard(Even,chi_e,chi);
+  pickCheckerboard(Odd ,chi_o,chi);
+  pickCheckerboard(Even,phi_e,phi);
+  pickCheckerboard(Odd ,phi_o,phi);
+  RealD t1,t2;
+
+  SchurDiagMooeeOperator<GparityWilsonFermionR,FermionField> HermOpEO(Dw);
+  HermOpEO.MpcDagMpc(chi_e,dchi_e,t1,t2);
+  HermOpEO.MpcDagMpc(chi_o,dchi_o,t1,t2);
+
+  HermOpEO.MpcDagMpc(phi_e,dphi_e,t1,t2);
+  HermOpEO.MpcDagMpc(phi_o,dphi_o,t1,t2);
+
+  pDce = innerProduct(phi_e,dchi_e);
+  pDco = innerProduct(phi_o,dchi_o);
+  cDpe = innerProduct(chi_e,dphi_e);
+  cDpo = innerProduct(chi_o,dphi_o);
+
+  std::cout<<GridLogMessage <<"e "<<pDce<<" "<<cDpe <<std::endl;
+  std::cout<<GridLogMessage <<"o "<<pDco<<" "<<cDpo <<std::endl;
+
+  std::cout<<GridLogMessage <<"pDce - conj(cDpo) "<< pDco-conj(cDpo) <<std::endl;
+  std::cout<<GridLogMessage <<"pDco - conj(cDpe) "<< pDce-conj(cDpe) <<std::endl;
+  
+  Grid_finalize();
+}